From 5930594d0b523bfe996c0c95cf963354c9965804 Mon Sep 17 00:00:00 2001
From: Manuel Rigger <manuel.rigger@inf.ethz.ch>
Date: Sun, 13 Sep 2026 23:26:16 +0800
Subject: [PATCH 1/2] Add an /impact/ section: what SQLancer found, who uses
 it, who writes about it

The project's reach was folklore -- a bug count somebody remembered, a sense
that papers cite it. This builds it as data instead: 2,130 records, each one
carrying the source it came from and the reason it counts, collected by a
pipeline that can be re-run and disagreed with.

What the section says

  2,038 bugs across 41 database systems, 541 of them found by people outside
  the lab. 22 projects whose own evidence shows them running SQLancer and 9
  more that have proposed it. 209 external papers, 69 built on the tool or its
  techniques, 34 calling it state of the art. Nine recorded talks where
  somebody stood up and described it.

  The bug count is published as a floor, and the page says so. Closed-source
  systems report nothing, some open-source ones (Materialize) report elsewhere,
  and a bug report need not name the tool that found it -- six projects that
  demonstrably run SQLancer contribute no counted bug at all, and the page
  names all six rather than implying the gap away.

How a claim gets in

  Every record cites a source and an attribution rule, and the rules are
  ranked: a project saying "found by SQLancer" outranks a technique name,
  which outranks the fact that a known reporter was running a campaign. The
  weakest rule alone admits nothing.

  Where prose is ambiguous an LLM classifies, under three constraints that
  shaped most of the design: it answers about text it was given and cannot
  fetch more; it never supplies a quotation, only cites the id of a sentence
  already extracted, which makes fabricated evidence structurally impossible
  rather than caught afterwards; and it may answer "uncertain", which routes
  the candidate to a person instead of publishing a guess. Answers are cached
  by content hash and committed, so the dataset rebuilds offline and a rerun
  costs nothing.

  Nothing is published on a name match alone. An acronym like TLP or CERT
  needs corroborating context; a reproducer that generates `t0(c0 ...)` is a
  fingerprint, but 556 of the records rest on it together with a known
  reporter, never on the shape by itself.

Papers are read, not counted

  `_data/papers/` holds one file per paper: the sentences that mention
  SQLancer, the citation markers that resolve to it -- which is how a paper
  saying "existing works [9]-[11]" is found at all -- what its artifact
  contains, and the classification each sentence supports. This is what
  separates a paper that builds on SQLancer from one that runs it as a
  baseline, a distinction name-matching got backwards.

  The PDFs themselves stay out of the repository, permanently.

Talks are read four ways

  Slides can be quoted, a frame from the recording can only be shown, captions
  are verbatim with respect to the caption track and nothing more, and a
  viewer's account is a note. The four are kept apart on the page, because an
  automatic caption is not a quotation and a picture of a slide is not text.

Also here: a page per database system, per paper and per year of bugs; charts;
a review queue for what no rule settled, down to nine items; JSON Schema over
every data file; and 281 tests.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EvUdZCeqsafccQ8rt7jt3N
---
 .cache/impact/README.md                       |     23 +
 .cache/impact/classifications/.gitkeep        |      0
 ...27133acf0ece46c7310ab724e35aa023f3468.json |     22 +
 ...4f2beb7ea6fdbca92818282945fd4f8592e5c.json |     22 +
 ...e0149e1eb1994ae783a10e6bd6b5f43dceb42.json |     22 +
 ...0822cdcd50a35e7cb8c4656d70629bd6cdcaa.json |     24 +
 ...8258d7cc148dd2ef47da3f5d0c64a0482e69c.json |     22 +
 ...322ffe2afc56f509a58453dffa2bb78c8ec72.json |     22 +
 ...d0a7a7d8367f6711135e566db2ca1957f7560.json |     22 +
 ...e0450c637f97ea01b4705276e5d80d80d0283.json |     22 +
 ...2cfe94062778e8511f26a6cca1367ca4e7972.json |     22 +
 ...606930c426ae9a9e9a95ab82bf7fb20f44d83.json |     24 +
 ...c95b3b6803698deb67cffb477d348397bb473.json |     22 +
 ...55d5f0cf5cf69bc5b2c02fa475e4213c6e1cd.json |     22 +
 ...02e4fd22e06dd403e1afd1b68a3fde9bd9221.json |     22 +
 ...448c380fb849554566a3a12072fedf99555dc.json |     22 +
 ...a399acf166fbc3a6355c2bb47996c8a8ff015.json |     22 +
 ...4a46053ed5ab89252753a52eb333469ee137d.json |     22 +
 ...be8e4ac3a2f187cad26cb1c3898e126606548.json |     22 +
 ...9d9239fde75092b8f59c675b7625e9531c225.json |     22 +
 ...a3a47149f35a758a7fa18e3293631259e71f9.json |     22 +
 ...7aa7be47ab108a96072e7256be4239367d3a6.json |     22 +
 ...698b70e9797030d5e2d2a210ed6716f49535a.json |     22 +
 ...c9b48df307b204314a2cea1028db091d76844.json |     22 +
 ...93e1a9108bfd386b91480df29481cc06f4fef.json |     22 +
 ...62002e85cf1a0140dd15f6d2799531c695457.json |     22 +
 ...8e93d8e3dd1d854dba7ec10144a1bb2838b79.json |     22 +
 ...25e70924ee86d7f642f1bfaf566ca7f381882.json |     22 +
 ...2d58cf60a255211afa66ea2a50debf412561f.json |     22 +
 ...79336861b31e0b40a64aa088cb90e18673a41.json |     22 +
 ...9c07a2ffa192bdab9bbf5e2484b7908d92dcd.json |     22 +
 ...3846b6a4563d7ba313f5e52c4103115836fe8.json |     22 +
 ...acf44bbf213375d147bbbe1909b5a6bb7acdd.json |     22 +
 ...e96f38a9789e3e4c7ae7e194e376189e4d5a5.json |     22 +
 ...ea8a617b37b49f87a4f8cc3e0ac2e0829f42b.json |     22 +
 ...518a2d6b906ec4ea1a1fe2f88ef58ac66da29.json |     22 +
 ...bdcf9273f815f09dc3317b674c4c9ac724110.json |     22 +
 ...a4ce959654a1d5902fe03a05bfeae34869bd4.json |     24 +
 ...f618cc362574ba879d0152d9e00ceb5d33558.json |     22 +
 ...852af4f62c5ff1530f48774e42d5fb86a1ef8.json |     24 +
 ...0433e6c96b2c925958aa110091f55f8ba8f62.json |     22 +
 ...f9afb12a43e276d7707f05dcef6de4c8376b4.json |     22 +
 ...23dd0552b76a316c2208c5106c01d927cd7d9.json |     22 +
 ...4b1f24359d6167304aef53511eef311ab36a0.json |     22 +
 ...d2be521284cc0416c827763e5211895613642.json |     24 +
 ...fb593034b0f38ad2f630c6e5d6c3b1e011371.json |     22 +
 ...7fd22426c55a15710dbd66d41e2ee52eb8811.json |     22 +
 ...316b00843229e0b170b2029180d78f1125557.json |     22 +
 ...cc1d51ae661f54620e57a9d51f7d538831633.json |     22 +
 ...78b6276257f16cd884c11f2471396981e611e.json |     22 +
 ...94f94b37d085440267379eb4e2c9927039571.json |     24 +
 ...49b1d5a1cb85b53b1469e2f41e3e7f28cfa69.json |     24 +
 ...25d88a9db2efc0ed2872098c3d7acdc890e9e.json |     22 +
 ...12d9b04d79af4840adcb453ba99bbe5e60ac1.json |     22 +
 ...906a96d8a818c75f7a3fd86be02c5cd1e361c.json |     22 +
 ...992955dabadcfcc18e65e4681f6d0218115a6.json |     22 +
 ...ad5da31019f1066c0a1eb44c8e37b6e136c35.json |     22 +
 ...c919bee81abf5bf9b2e7b2397e6da5586f339.json |     22 +
 ...bb99830b6b5b88418faccfb09de51e77f2cf7.json |     22 +
 ...2ec613395774f0185215de8a602fd926a594b.json |     22 +
 ...2aad7b0b6fbccd83b4879df02f0b46fca5fc3.json |     22 +
 ...1da75dfa7d38b4f4bd41b3757320a29d7ea98.json |     24 +
 ...17033a5151f20ee58dac89d5ff8661542f54c.json |     22 +
 ...0848e2324bb4262df3645002566a4c0c58a76.json |     22 +
 ...dce8af25f99a6f976a9f043946c1f6b85e2ae.json |     22 +
 ...e85b68451030ecd6a196e1ce64a6c85d98cbd.json |     22 +
 ...e9d60d0432799d296898a8f2b3de36e4c6230.json |     22 +
 ...77bf621d9f70a1c346ebd2c559d37e9d6689b.json |     22 +
 ...de298f29167ff468faf09a2a12967c55413bf.json |     24 +
 ...2e478a561e643f7e0e146e85699dfafe16b9f.json |     22 +
 ...a191dcb7c2e6baaa78d23a0aaa74b182681df.json |     22 +
 ...2520c8e59e2ea2b06823904797bf32769d93a.json |     24 +
 ...5cd13765352477adb7d81864d2b6b3171d657.json |     22 +
 ...9d69f01f222d14c53b9f4b362c9f1b59a4fa1.json |     22 +
 ...8225383fa43ab7737dea8f53aaf1361785b79.json |     22 +
 ...367062f81c770d4e291235fd1aff4d64c934e.json |     24 +
 ...c9ad577b99d02d54e52d308dd89f3be3acb61.json |     24 +
 ...5f55f9d689e860cb3ea937d58e659b5f4e82a.json |     22 +
 ...6004a1abfc020020d5a2188dc4ddc9a734f19.json |     22 +
 ...41bfc34c2a474f23182f5c44110e4cd7eb95d.json |     22 +
 ...af09c0e1a559ea3939328d6df63baeeec4cc9.json |     22 +
 ...9b7721a4ff519bd44f3f87aa9a39568e6c577.json |     22 +
 ...40e99ba3489cf17decf1d65164f7a6c81a90b.json |     22 +
 ...4075434acf86ef93ae8dfd013c56e0c1e6d46.json |     22 +
 ...58668b14904ece7180035651f5875f8beb339.json |     22 +
 ...2dc40a006d6aac49ba49c723cd9e286078bbb.json |     22 +
 ...32da6854a804f39fcaebf23f18a7d56755fb8.json |     22 +
 ...4d3497a7f0e572ec5d418852682cb3874ee60.json |     22 +
 ...c8431a3843b099f5e9fc87b7a6e95abf81b19.json |     22 +
 ...f63e34e4f0f43326d2f1d7d8aae77634be74c.json |     22 +
 ...8b61f7b2f6c1348437a62b3d0ceeaa7eb5ccb.json |     22 +
 ...53572a6edc5aea2a5326109d2e9267e882c46.json |     22 +
 ...ee6700692d26f6e80ed11681c83af1355786f.json |     22 +
 ...f4bb84e386638da8687d8a672ec040872934f.json |     24 +
 ...7dc0983296a09eb47b94c778070dd4d6beb0a.json |     22 +
 ...60e131b3edcd61e1b315fb04f9ca7b1f83999.json |     22 +
 ...e087d5904b0848e2472cfb89264fc6c6b1323.json |     22 +
 ...982b044b4a62ed1f529087f888fd8099cffb7.json |     22 +
 ...6e0b345bcb74e535a05a6ecad906fe0712247.json |     24 +
 ...f3b54a466c3533b1c0cd71af045f59129201c.json |     22 +
 ...a09f461dcaab7dc097878b7feee9ef20b219c.json |     24 +
 ...9be784d89919691dfcae55cdec8aeb0140de9.json |     22 +
 ...60761e2794d024dd0580198fa8fbaedb154a9.json |     22 +
 ...b50653eb060ae545efa8804db985d35c1ddce.json |     22 +
 ...824a5990fd11dc11359bd6dc53f91b861359f.json |     22 +
 ...427f046aedf49e0e5361180e79842e51e76b1.json |     22 +
 ...57b8ee8d71a0ef0bab104e4b6ad739187858c.json |     24 +
 ...3cae106b1e3e1282898ca594959a9a9581559.json |     22 +
 ...8039eba5c773095d86e8adeb0c18383569e5b.json |     22 +
 ...8a710fcce367d8c911889e1dc045d71d2c084.json |     24 +
 ...86a6b1276ae75b43e3b2617a9623ca765a938.json |     22 +
 ...04f199cf62840af483eab81a2ec93bbc90eeb.json |     22 +
 ...23efd3603da8b6d3033597cf02ec319d51ff4.json |     22 +
 ...2d2b32c9766927dbfacfe68e68e8200fe3fbf.json |     22 +
 ...ed197ded3133f484f4fced09e20644a4efe8c.json |     22 +
 ...b8c6a3333f6237c4c51bbf513a277054a8f84.json |     22 +
 ...fecd65f0f63dda21b35d5829ace408ce6d3f5.json |     24 +
 ...c8a64297ed48a727ba96c27b2171c483040ad.json |     22 +
 ...6c02ee7c6f26c9ced31ae4d2b6531880065d3.json |     22 +
 ...b7833f4274ee9bc98dc300969e8822bb84cfa.json |     22 +
 ...95aae36b40e108c2860f13839334ca2bb4df6.json |     22 +
 ...afb7191b2b9f8f9cc5372dd8ee52cf4db6149.json |     22 +
 ...dc3ab1565301c1e694be63372626798ca0e7d.json |     24 +
 ...eecfba734fdd70ad880908bba0a4d4d15ee09.json |     24 +
 ...7090ca44711c4b5893dd639571392639c4221.json |     22 +
 ...38600b327d16c7329f7049a54b9b194b2e686.json |     24 +
 ...408e693c9b10bfc023463d64235fbf0424709.json |     22 +
 ...1ade8e223fb6225b14bb5dceecfc436afa6e6.json |     22 +
 ...e770ad376d1eea4bc6c3f1cbd7d4464e234d3.json |     22 +
 ...12eb4ad818219c934e4b38b10dcc1f9a0c0a6.json |     22 +
 ...c16b46df14cc9c71723467e0dd5cb9e894e38.json |     22 +
 ...313b50e0ecd8661427c729dcb530b82115fbf.json |     24 +
 ...78fd12ca7db457bb290b77b4a6ba002da4e6f.json |     22 +
 ...b1ec77c7cee0cf1826f383087f3ba64fa8753.json |     22 +
 ...8adcacdafc9d085aef3f3c7929c63419e2d10.json |     22 +
 ...7cb93c540436a56948f395b6f2228a12fa431.json |     22 +
 ...38b46053ed970020ba8631b747d057b9fbf56.json |     22 +
 ...9b766bbd9935efbe0aeae7d78b2e6755958be.json |     22 +
 ...5dacd8fa8755f80f984b5f844b6b066757b02.json |     22 +
 ...f6ead998f58bf1c6c402b7c1ab242dd6a61cf.json |     22 +
 ...45142473719bab8a70165dd2b917b39a19202.json |     22 +
 ...7cd1c0aad1e7364588a3d215f9db45ffd14b0.json |     22 +
 ...e4e074128f7d771f626de11bbd12a31382254.json |     22 +
 ...920737cc2dbf185350db475dd5e0e426969d4.json |     22 +
 ...f7502e0ba5f49a289e9cb8272675bc64e2532.json |     22 +
 ...3d8305d9100295434ae9d746695d3fc5ce45e.json |     22 +
 ...3cfec857997bb3aa0a5db10bb5ad7c9fa8139.json |     24 +
 ...9479876b7841c8b6baa47c524570279000a29.json |     22 +
 ...149ba8806109596e63d6c95de884ec1b3a44b.json |     22 +
 ...30497527d2309b3b3d6600f223a0df13b88d3.json |     22 +
 ...4fbdb52ff8adf8208f2c9beee873c27ec805f.json |     22 +
 ...292c86b30ff0f149ef3d33cd91b758da870e3.json |     24 +
 ...dbc52e70b60e4de6b4f4e485a188d387ca20a.json |     22 +
 ...d3e22ceb6f9544a3a38aa2abf417f40b9d029.json |     22 +
 ...9aba3aa18f3a2cb54e5c288f061f31e000a5e.json |     22 +
 ...33293cccf99fadbd0759b4c6bdb70d087dffa.json |     22 +
 ...5b5d460b52676ec568f60a1490acd2db2419c.json |     22 +
 ...c89d82c7829182a9d19ba96d26eb50030f6e8.json |     22 +
 ...1bdfe45e61af252905f368909cbc552c4b5c4.json |     22 +
 ...5b1dc0c27570c56c69b16f9818b186c92d185.json |     24 +
 ...26f281dca24ead77dca80e31773f9925ed368.json |     22 +
 ...227bab7bc70fa3de58c9565c969b15ecfd704.json |     22 +
 ...da67402d87384fecbd04a8bb754db1cf14c62.json |     22 +
 ...e7c64b03988c34d83d451bebd4f524be9fc66.json |     22 +
 ...ccdce8d098bf29efb58216a52ae304fa24f96.json |     24 +
 ...af0f25d18905f56015d0b0f1d9fee40f75f13.json |     22 +
 ...ac7f8770500df101b1e03339f11f842783b97.json |     22 +
 ...5ea657793f52cd1aa281ff5294f6941b6a7a2.json |     22 +
 ...c493f6734a28f1c4226e38d16e0593b993462.json |     24 +
 ...1313325de1a8988b0c8e0d70f8da3837a8562.json |     22 +
 ...42809991dcf7cf40a2c0712537fa8e0546dbe.json |     22 +
 ...7605d2c7612d188fdc09eb1bb168b77ad224f.json |     22 +
 ...3dc16d4fdc24acd94a6c60ffd8a737bbf6464.json |     22 +
 ...2f53db26a768332d128f0726eb921e3804304.json |     22 +
 ...8d2ac5f10c5f9f14564e9529a73e42a7f34a3.json |     22 +
 ...105d320c03eb4c23e5cf696ad249ad6a2128a.json |     22 +
 ...74893891e1aa3f992b02d65bf2b21cca702e6.json |     22 +
 ...36e3eb50b6b219e716ff19e2ee0557dfe02ed.json |     22 +
 ...8a62d716c1379b68ea32ba957532202e15a17.json |     24 +
 ...a51bb94ce0760c2a8c64d9b2c4a4025900660.json |     22 +
 ...af20b83209cf2b91f410cd8b581776d89ecb5.json |     22 +
 ...2d910d1c86a76d315f823df00daa1482bcfa2.json |     22 +
 ...0883baf956a0139cbf0d08024f68f7665f2a0.json |     24 +
 ...92425a82cb37ef49688ef9a7d8cb1bbed1b80.json |     24 +
 ...3fa45e93a34d88a7f61757085d5faf857118c.json |     24 +
 ...e0a944da40d5b19fec334adf3852f23a40a30.json |     22 +
 ...0035b6c3e0b80167ba3817438688236b45f77.json |     22 +
 ...aeea731afc4c062bde302ed7a558204b30b08.json |     22 +
 ...99a41d9f0688faadb287be6d33a3e327e4db8.json |     22 +
 ...dddca40878c70b54ea078a9967dffbcfc8658.json |     24 +
 ...fe55fee267498e747e27326881ce5bc51d21a.json |     22 +
 ...36c9289e2a53aebaf5de0a6d8991efc49d52c.json |     24 +
 ...12c3245f94c67e3f73e57fb0d69f7537e7fb0.json |     22 +
 ...08289a955adc3363c13dfa23a74eecc12a1de.json |     22 +
 ...878ec222e624203c2a07beec07f57c73c6437.json |     22 +
 ...84ecc29ddd5beab318412e915f18f89d370f5.json |     22 +
 ...87c00907a04b3c7e9e672024d9e7a0c76c5b1.json |     22 +
 ...af144e83dc7fc156b1b353c6419d5cdd9dd6f.json |     24 +
 ...9d91309ca6794ee08fd066ed7f1c69de35592.json |     24 +
 ...a201ae38a302e9ba778cab953442cc85b1d59.json |     22 +
 ...0d419465844a91c07964b232f8f34931e115e.json |     22 +
 ...96787159abeb77cfd26c5ff43b5982583b5c7.json |     22 +
 ...2bb852908789c81a8499eaf406680d029b77f.json |     22 +
 ...054fb1fda43be48b9bc5b0769fe575e3376ea.json |     22 +
 ...da469f3b34bc5bdceb471e333075a80310eeb.json |     22 +
 ...f5fd7b851eefce04c4d7a01e32377b9f2973d.json |     22 +
 ...4974e1454b93933976bd181a4b53272027588.json |     22 +
 ...0a28443642d3dc26a3f1f8027c5b344c7bea2.json |     22 +
 ...1bd9f15c08f9d05a6ebd520f59dab274c9da5.json |     22 +
 ...acb050b2f2c6e8572bd314c8dc1ea6f8f9acc.json |     22 +
 ...b6dc5828057e2108f1918f3e1078fddd0933b.json |     22 +
 ...dd56980687ff50e0a2bd1dc622fede603a2b7.json |     22 +
 ...9fc21a5158241afe6d087184f9076f97233bc.json |     22 +
 ...777726ebdbae2082ebc55ae80c9e1f0e4fae4.json |     24 +
 ...928233ce19c5044f6c3df1f260c45369e3f8e.json |     22 +
 ...eacac1e9309274de671a65553bf4eeb9c40bc.json |     22 +
 ...4bea188a4f6de04864290c784177fc02202d6.json |     22 +
 ...a0347b4beaf1fe3751df8315cd879ba619b6d.json |     22 +
 ...074fa1f8c21aefc9ab7b013a897182a9003ed.json |     22 +
 ...25d6de29249f25b5365f29dd3ec6c1541ed68.json |     24 +
 ...f70750014a42250ff436ea9649b184f59823a.json |     22 +
 ...794f748a0769b73eb8dd8a6663e2c28a77bee.json |     22 +
 ...a1ea6477c2de0748d14c5992b1342e5ed27df.json |     22 +
 ...abb01c62dc95348ca66c529eed100c31094e5.json |     22 +
 ...2784bf3e8422f8bf6004f866fe1f31de1ee39.json |     22 +
 ...fb90a98e6f4a52ca0a3d0e0a0b965eabaa8aa.json |     22 +
 ...74eb17e84ff3ac778d11f2cc6b2d18e2451a3.json |     22 +
 ...6e6494bc9d79c93bbf7b8509ad89f1d140b3c.json |     22 +
 ...956a3076269d2f5ae1d435ea291177b9d95a4.json |     22 +
 ...be296f3ea0a789066736f46d810abb83f37e1.json |     24 +
 ...04ba894d5e84837c731e5cba2ea31d732c4d1.json |     22 +
 ...ce4f7d28da3c7daaf6c69e2c038842a739c01.json |     22 +
 ...a70e5ec6dfb07090bff27b629c794f1877666.json |     22 +
 ...b3174b09de62698e484236700499e128eab50.json |     22 +
 ...3ffce68b27173b695ad21edf67dcff5f2a0ad.json |     22 +
 ...a4d35cbe2f42e7d1f77072c5c1f11dc318230.json |     24 +
 ...e97801f823c3e2e591514ecaddf021f764274.json |     24 +
 ...fa8223ee03cc99019ba2d4046b2b9cec32cc0.json |     22 +
 ...efa758494e5cce9e69cc646093bd6f2d8788c.json |     22 +
 ...32bfaaa3185895c0fb0a3b9c99d829a653221.json |     22 +
 ...8ab7a5c6d6505275ace026e9265d1dd8c41ff.json |     22 +
 ...4f977133c3b608c7c4b211242a7d7e2838b08.json |     24 +
 ...697c73b7767b98c7e2a9392e4a54ab7b6388f.json |     22 +
 ...36531315ea87144e4c018032c4a696b86c642.json |     22 +
 ...aafbfa9c0fc5e1f7dc476599d2ae4bf682cc6.json |     22 +
 ...711fbb79c8215d432afcacf647c44f2a002fc.json |     22 +
 ...0c9d46c45749849c932ce0604c8a7332f67ab.json |     22 +
 ...473c0004c87f12095a68145ed0da631c06ea7.json |     22 +
 ...64e196be6b861922f5818189d7c89e399e330.json |     22 +
 ...478fec927aca3d6edbb59dd23d8d274612367.json |     22 +
 ...f41007d48c112ac71d8f13777a0c8f0bea958.json |     24 +
 ...86e606297e158b3f90d6fd03b328c1708e74a.json |     24 +
 ...65b1ebec88301f531f6d263679ed61370ce51.json |     22 +
 ...d0260556bd8c825c9110f96a9a99cd6c6e8f8.json |     24 +
 ...3539bac44d3554c2fe704bcfa7e4282090456.json |     24 +
 ...e9c8ad2d5c773c5c0b1de6170e5a73ca71d21.json |     22 +
 ...c179137cd9b9c636cb5a2730158972e240ea5.json |     22 +
 ...ae20bf616cd1908f0757155f9ceb4c6870ef3.json |     22 +
 ...1c5272bfd3ae4751bf1745eb1625ba9560c79.json |     22 +
 ...c9b8338ccdbde6e1f982c18fa711b90b5a5cf.json |     24 +
 ...e5305c63485ab7fe28092b67d8af039c2ba81.json |     24 +
 ...6a95e82f357018871daa70c31cd3c8cf7c484.json |     22 +
 ...4b470a09b882fa0d8139591c5b4aa345d20a8.json |     22 +
 ...96c3f717d8c845f9c8d28570001b48fae97b3.json |     22 +
 ...4f7713a66a7ea4ecdcbd4ca9a49bf1735693f.json |     24 +
 ...d0bea5f83c442bccc5b2e792fe45f190ebce3.json |     22 +
 ...c40e7ea6bee39b6d49888e604414710af5fd6.json |     22 +
 ...690827b0985350fdd199e762361c198e8b63d.json |     22 +
 ...b883e7cf402b6ae7fc5fc1f5d9b708530a260.json |     22 +
 ...36fb5b15898a696974bab1c7c603c39c73bac.json |     22 +
 ...11caaab48472e0902bde71e0e0c73d31216c6.json |     22 +
 ...f341b013bb7fbfeb3fe3d893434e04fd9cb2f.json |     24 +
 ...d856f7310e97764a4d1e4f58eaaa7b5e984ed.json |     22 +
 ...be13af50cc6b8f08987803f680d4231815739.json |     22 +
 ...cc736ec2fea497f475bacbad2cae0520f1999.json |     24 +
 ...c3f9fcbfabfc4fce34ae41500dc506213c13c.json |     22 +
 ...aa608c500869dd063f929f098a014dcbe3391.json |     24 +
 ...4355d7046ca30bafc2470345a440f2fad54a2.json |     22 +
 ...b95704632109f0da24ff21b5d3ff378493fb5.json |     22 +
 ...1cb92d955e98dd57e2a8f4ba8f385e3e0859e.json |     22 +
 ...2e8af354e83dbf7865b135de8f0deb414d75a.json |     22 +
 ...18ec068e225e87ab2db3cd2f877953174ee3c.json |     22 +
 ...24c9cab3c5755a0bb181d892114926fa07726.json |     22 +
 ...d20582b6d70c743dd2e020644768ab7c2696e.json |     22 +
 ...06268fdf6e738498b2fe700a78fff4343bf3d.json |     22 +
 ...97812d17e662b1c05958526e4e45c1c6d14d3.json |     22 +
 ...057d87a3b969e3373d3e69b79db8ad2dd5d81.json |     22 +
 ...fe63b66e901d8d7e62a849a4160b7d4c33c1c.json |     22 +
 ...2a34a405abdae4bb9507487627a4265c36005.json |     22 +
 ...e30c7558430fa4c0813fc4cbbcdb1ac2effcd.json |     22 +
 ...52beeb2c98d708e52566f4bb244ae0be22edc.json |     22 +
 ...a002b8055316c9ea976467e9a09705155f4e1.json |     24 +
 ...565a89122634f0b7d4a2527ef8fba5ab7c90b.json |     22 +
 ...4b019a8d8683ba78c77d8df612ebf2c030cf8.json |     22 +
 ...1cebccf92877e5e9d53952d18265e80bfec02.json |     24 +
 ...e4add44b342faf4e52fd4e722c4cd142df8e7.json |     22 +
 ...a96bf8ef5652372a97871dbcec7e1338b600c.json |     22 +
 ...8bad99352053a321f7690d361948426866854.json |     22 +
 ...a97c2bde73046d214035f29a3726d3e36b803.json |     22 +
 ...9c53f119b8a99e1d6e53a817abd053a8e7e71.json |     22 +
 ...4675384458cd387913cd97e101615834dc097.json |     22 +
 ...f37a97b1550327cedfb2caa86815d54383ca9.json |     22 +
 ...d52462295333353ca3f266f7b01a8cce325a7.json |     22 +
 ...4584990c2429ce754b76041a6ad26e167d5b3.json |     22 +
 ...c66575ee6bfaf323400e7917204d1a9657751.json |     22 +
 ...b10cf3b87322ad41d85a4d65561a6d184b18e.json |     22 +
 ...bd52e17c0fdcf4082d5bbe6461ac052c673cb.json |     24 +
 ...b788404159aaf8a11adacee8629f75a37fcc5.json |     24 +
 ...4c38703bb30d20843bf5da8c52593d505a1fa.json |     22 +
 ...86247cd866a3edb9e4d2b9ec2e60f82bfa6a0.json |     22 +
 ...d050c205a8281946311797e3453ccad96d3d0.json |     24 +
 ...4bff0eb95a167f6e71e9dee6846887f585630.json |     22 +
 ...d64ebbc641a179d20174d2a1fe985962497eb.json |     24 +
 ...5ec1394a7d28581426c47319fd6229fbccff6.json |     24 +
 ...8d19eb5c240f63e70edec38a2914ff65fe877.json |     22 +
 ...c36a988a4a793c742dc76f07d2f2b33ccc319.json |     22 +
 ...4fe8d8279d65f99ed2cbbdefa1186927adf65.json |     22 +
 ...91ee10374ca37b1ebd1a82936d5261f558b10.json |     24 +
 ...41e48fd5dac356321481a85cc2c33177d2d28.json |     22 +
 ...f5f2863af23690e49aa0efb6eeef391f1f047.json |     22 +
 ...aab19142af3ffc94ad555f0ee6dab20f9215b.json |     24 +
 ...79cfb23563f3e34750e00afb4ab25fd1f5f23.json |     22 +
 ...9b5982c1e6d86a00241d1ddd31712e998ab66.json |     22 +
 ...806f471a0ab46dc5246cf88d128d489f3a485.json |     22 +
 ...663610f8cebafdd537d2135a1648c4e4eb69c.json |     22 +
 ...9ed95ad06d9ce3b84b0c0ac9ecb71176b48cd.json |     22 +
 ...5de90c16942b95e6d5f0122f01a4c9ed38cc6.json |     22 +
 ...a0c8508c59640fb4cd5d5075541f9c74dad23.json |     22 +
 ...6233eb9fab69d7d7444bcc38193acd373522f.json |     22 +
 ...40743083a487b559d832bc11a6cdbb84cef91.json |     22 +
 ...8ee565932e41bba450728de8d5e1c569a6dff.json |     22 +
 ...7ee67bc40ac5cade27833e506996274dc31d4.json |     22 +
 ...f583f9001b5eea5c7cc359697dedd35517d8a.json |     22 +
 ...a88b34bd2bd9138b5b1ca37a561ec655f4d5f.json |     22 +
 ...4af3af727a0419a09c2ad4bbd347927e4bb20.json |     22 +
 ...6e8872b109003f798a5d163418b5a6e41da9e.json |     24 +
 ...b8144a122607f408fb3c9f1baa3f19735f1c3.json |     22 +
 ...4cfdb28929a236c54ff451c8178e88bfa0676.json |     22 +
 ...d74429f0f3fda89cb43274531edaeedb3bc7f.json |     22 +
 ...5361dda970801bcc8e0fae74b5cfa7902b850.json |     22 +
 ...ed1e255bc9ab436a3025627fd8c6ea87c1c23.json |     22 +
 ...532e78b023b25f8a44e677aa32bb8b56e11c1.json |     22 +
 ...973324f1059d8062ba2ca3a59d6d352f2be91.json |     22 +
 ...97c884f2c4b30b26a7b757bf1da42b7d2b699.json |     22 +
 ...db29449e0e574eff01c898316e603775c78b2.json |     22 +
 ...ef9cc756019b67bfc99afa1c80b56c29698fd.json |     22 +
 ...01ed161a1e3a0c0fbb398ec3c063ee77f95b6.json |     24 +
 ...ef526487cce5cb84834e80e42c3f9fa47098d.json |     24 +
 ...a114e392bebf05058ba91f7153943d9866164.json |     22 +
 ...778d4111bd75aa96b7dcfb2ca72adbb14c452.json |     22 +
 ...2770b5f71d7d79b8487ba4db11e0e01482675.json |     22 +
 ...f61d267e424094d2d6fa73780cd6c5c45277c.json |     22 +
 ...633702040e256ecd9cb4383d031a95ce8bc8e.json |     22 +
 ...05ec9ce4d8863198cae9ffdc869fb91eb89d7.json |     24 +
 ...deada8719312d00047dd1f445efd3a12499e6.json |     22 +
 ...f87a11c9b7fdae6e306f9cc47d05afb4cfb64.json |     22 +
 ...4d2cf7d7373ebabd55db0b517b40e136e8ea0.json |     22 +
 ...cf361438a6a166a9ed1d2a2f196282119ba8d.json |     24 +
 ...1d01d6e3c4a364eec5311bd087127136ec108.json |     24 +
 ...b5eecc1e7b7c091c962354cd0173efcab7d2b.json |     22 +
 ...f562438caa6715e83d967646e029ff1273898.json |     22 +
 ...5ae054fffa3d58f6d3e8fc3f9dcce3b3012d8.json |     22 +
 ...421be4c29e7c3e5392e0e00910a4a86d342c9.json |     22 +
 ...2ec85ca3ac3f3eb4747d3138b79e725157cfa.json |     24 +
 ...851245ce3afd330483f4051c6042610298f41.json |     22 +
 ...c787e59d3152d6cad46ca6a70e22815d06a0f.json |     22 +
 ...f724f5f2554bc01414f19b8990334b24a5ff9.json |     22 +
 ...5f3a531ca80dac0176b2fcc6c864176128f69.json |     22 +
 ...7a7ee12e883355142d0dfba9e73cf49e1ad53.json |     22 +
 ...90f85f8e893720fb47b42d5cde712e3964025.json |     22 +
 ...942e9c603b3acc94eee4d3978849c6a550268.json |     22 +
 ...6eb8a1c1db7e0edc15c0943e66ebbf602705b.json |     24 +
 ...b950d503814f9aed886d66bfa88a9e7e0cb9d.json |     24 +
 ...f4dbb4cab5925cd7af52f5c148eb1f459cdfd.json |     22 +
 ...16dc5539e8a59dc1b12f107032cdc53492561.json |     24 +
 ...1382721f53278a7c285b051242ab46e3b6738.json |     22 +
 ...85083468e34163df1b7b19e861fea896d79ed.json |     22 +
 ...13e12f1bfccb485ef6f4db2f644be175d6606.json |     22 +
 ...e3cf4a09739a5829ee1c8d0b7bb59c056d756.json |     22 +
 ...811b0429e8873beab9e2c568ec0051691d2aa.json |     22 +
 ...c8ed9edfc49d8557b16b07d954d7a21082df4.json |     22 +
 ...1a6691ddb21e1aaad5acdd4dbc286f872f2f4.json |     24 +
 ...158465987d9d395af11cd640be8fe42c42734.json |     22 +
 ...2ead092999dc5c1ac8b61ab0e41fc60dff552.json |     24 +
 ...9bf6199e451f19300d9e231a2feff1f8a3af8.json |     22 +
 ...35ed5d709abad7af5c5956e5483215b33eb89.json |     22 +
 ...c6f88254480a6e8fc723cb0761429e0e82f17.json |     22 +
 ...fadcb6e7f579235727cc074594d8598b8ee7d.json |     22 +
 ...988f56777c79fe8f7e0b560f7a2dfd59599a1.json |     22 +
 ...b0642defcd119922d73681a73a3e12169d280.json |     22 +
 ...c928a9187ec0b14439604f766ffa2b504453b.json |     22 +
 ...20fd0f39ae70a52ada3fe312a082867b1eff4.json |     22 +
 ...f987c7feee54afafa416c7e522171a901cc39.json |     22 +
 ...c7fdc856eb4789e6425bb5fc1117ec58d0301.json |     22 +
 ...44cb17ee5e255d790f3e94111b34cd07e19cc.json |     22 +
 ...7a944dd6fd3a3f6adc432b058abc4197b8123.json |     22 +
 ...e6bacb2e626c3291e663473ad9007eed18b4b.json |     22 +
 ...b7f7d4a560cef3ea4893f20463251b09af809.json |     22 +
 ...036babe276c06460f0b547b2b586dfba65d21.json |     22 +
 ...346571182571d475775c3102204a41fe8fb52.json |     22 +
 ...84c4010eac45911e85654cff110f7bdadc0cd.json |     22 +
 ...38df8f780137fc0f65fab754162354f82a485.json |     22 +
 ...f3064e55b90ae349522495d673f8bc25e1e16.json |     22 +
 ...1aef0c9fcdc4200c2bc7b925246c8004e4918.json |     22 +
 ...4a685f4c36ea97ead3c529fac595dd1746abc.json |     22 +
 ...ef9d0f05c63248255d912584e680a8fffa206.json |     22 +
 ...5424dc97d8e7be9368cbd0465bf14194a86c9.json |     22 +
 ...597afc58229295f8a2662614f7ea81ab89663.json |     22 +
 ...3054b22b99ea3a337f2fa4ac7cd9a12aed487.json |     22 +
 ...7e9a71f42449c0778a062bb527337efe73b88.json |     22 +
 ...40bd35f3021a3e9c73a803c9aef52bdf2939d.json |     22 +
 ...9f4a9f0fa467c3bd1602f1f2f38dc8f0b1a9a.json |     22 +
 ...30b8e18cfe811cb1fd705ae73002402c715a2.json |     22 +
 ...5671678bfd4307407876fe15860eb90934d7c.json |     22 +
 ...f57baac3682fb852067d58e5502a6e3c73ba0.json |     24 +
 ...fc711118fb797ef6d5c7d45f6901927836478.json |     22 +
 ...3a5a68dffe69fd7aec13fe037e43ece9aff04.json |     22 +
 ...42498e8a2aeb62c50baa19120568c8fe10aa1.json |     22 +
 ...b5f63a583a4f1ffff1978ccc2942ad590573b.json |     22 +
 ...f845d79887d8a2b190792aca06c3481ad5ecc.json |     22 +
 ...ff533f63f4ce4c5ebc3d9c66b489cb732e4b7.json |     22 +
 ...965ae488ee100de93b87a9594bfb10deb32a3.json |     22 +
 ...fc7731acf0e31cabf9d28659c49f5095a1db1.json |     22 +
 ...724852ec7c2138377b6bfb48c0c92c2ff9cef.json |     22 +
 ...090fc9d6bfa6e193fdea696075aaf7640dd4a.json |     22 +
 ...83a16743a094508aabc75cdcf6173fd7fdf16.json |     22 +
 ...2079584c3df0f60d6a003b4e00fc5e7e0a995.json |     22 +
 ...756940852239107113528fe20df7a2d4dadbe.json |     22 +
 ...2bc38b5e14c3b7944b1c39a03f4955ce41dcf.json |     22 +
 .cache/impact/derived/.gitkeep                |      0
 .cache/impact/papers-pdf/README.md            |     23 +
 .cache/impact/papers-pdf/WANTED.md            |     31 +
 .cache/impact/state.json                      |     24 +
 .../impact/talk-transcripts/6YGqFRTe2D0.json  |     21 +
 .../impact/talk-transcripts/BgC79Zt2fPs.json  |     57 +
 .../impact/talk-transcripts/CW4Ntdtp7lg.json  |     37 +
 .../impact/talk-transcripts/L90MBb6NLBE.json  |     11 +
 .../impact/talk-transcripts/QRwxHGpWaUA.json  |     21 +
 .cache/impact/talk-transcripts/README.md      |     39 +
 .../impact/talk-transcripts/V_qzqY1bb7I.json  |     32 +
 .../impact/talk-transcripts/wHo-VtzTHx0.json  |     41 +
 .github/workflows/ci.yml                      |     40 +
 .github/workflows/impact.yml                  |    169 +
 .gitignore                                    |     32 +-
 Makefile                                      |     37 +
 README.md                                     |     77 +-
 _config.yml                                   |     28 +-
 _data/impact/adoption.json                    |   1883 +
 _data/impact/bugs.json                        | 114380 +++++++++++++++
 _data/impact/dbms.json                        |   1131 +
 _data/impact/needs_review.json                |   4753 +
 _data/impact/paper_decisions.json             |     46 +
 .../paper_notes/paper_arxiv_2105_10016.json   |    132 +
 .../paper_notes/paper_arxiv_2206_08530.json   |    139 +
 .../paper_notes/paper_arxiv_2304_10044.json   |     56 +
 .../paper_notes/paper_arxiv_2310_06433.json   |     91 +
 .../paper_notes/paper_arxiv_2503_03893.json   |     98 +
 .../paper_notes/paper_arxiv_2503_17322.json   |     55 +
 .../paper_notes/paper_arxiv_2506_02617.json   |     59 +
 .../paper_notes/paper_arxiv_2509_10819.json   |     56 +
 .../paper_notes/paper_arxiv_2510_06663.json   |    141 +
 .../paper_notes/paper_arxiv_2511_17377.json   |    109 +
 .../paper_notes/paper_arxiv_2601_15074.json   |     56 +
 .../paper_notes/paper_arxiv_2602_19490.json   |    148 +
 .../paper_notes/paper_arxiv_2603_00311.json   |     56 +
 .../paper_notes/paper_arxiv_2603_19434.json   |     53 +
 .../paper_notes/paper_arxiv_2603_21530.json   |     67 +
 .../paper_notes/paper_arxiv_2604_01442.json   |     54 +
 .../paper_notes/paper_arxiv_2604_03024.json   |     62 +
 .../paper_notes/paper_arxiv_2605_20473.json   |     65 +
 .../paper_notes/paper_arxiv_2605_22992.json   |    126 +
 .../paper_notes/paper_arxiv_2606_11132.json   |     81 +
 .../paper_notes/paper_arxiv_2606_14164.json   |     60 +
 .../paper_notes/paper_arxiv_2607_03741.json   |     88 +
 .../paper_notes/paper_arxiv_2607_09072.json   |     56 +
 .../paper_notes/paper_arxiv_2607_13276.json   |     58 +
 .../paper_notes/paper_arxiv_2608_15709.json   |     56 +
 .../paper_notes/paper_arxiv_2608_23402.json   |     76 +
 .../paper_notes/paper_arxiv_2608_25573.json   |     59 +
 .../paper_notes/paper_arxiv_2608_30385.json   |    147 +
 .../paper_notes/paper_arxiv_2609_00381.json   |     56 +
 .../paper_doi_10_1002_9781119880929_ch13.json |     42 +
 ...paper_doi_10_1007_978_3_030_71058_3_5.json |     40 +
 ...aper_doi_10_1007_978_3_030_88494_9_12.json |     62 +
 ...aper_doi_10_1007_978_3_031_51479_1_17.json |     55 +
 ...paper_doi_10_1007_978_3_031_94706_3_7.json |     57 +
 ...paper_doi_10_1007_978_981_95_3182_0_3.json |     40 +
 ...paper_doi_10_1007_978_981_95_4721_0_7.json |     40 +
 ...aper_doi_10_1007_978_981_96_4506_0_18.json |     43 +
 ...aper_doi_10_1007_978_981_96_6465_8_28.json |     43 +
 .../paper_doi_10_1007_s10664_025_10662_w.json |     57 +
 .../paper_doi_10_1016_j_cose_2025_104564.json |    170 +
 ...aper_doi_10_1016_j_displa_2024_102854.json |     39 +
 ...paper_doi_10_1109_ase56229_2023_00106.json |     56 +
 ...paper_doi_10_1109_ase63991_2025_00095.json |     61 +
 ...paper_doi_10_1109_ase63991_2025_00151.json |     70 +
 ...paper_doi_10_1109_ase63991_2025_00322.json |     93 +
 ...r_doi_10_1109_compsac57700_2023_00273.json |     97 +
 ...r_doi_10_1109_compsac61105_2024_00141.json |     58 +
 ...paper_doi_10_1109_dsc55868_2022_00057.json |     82 +
 ..._doi_10_1109_girst67753_2025_11382165.json |     61 +
 ...doi_10_1109_iaecst64597_2024_11117732.json |     65 +
 ...doi_10_1109_iaecst68792_2025_11415166.json |     80 +
 ..._doi_10_1109_icccs65393_2025_11069832.json |     59 +
 ...doi_10_1109_iccste65902_2025_11138310.json |     59 +
 ...aper_doi_10_1109_icde55515_2023_00057.json |    105 +
 ...aper_doi_10_1109_icde60146_2024_00011.json |     58 +
 ...aper_doi_10_1109_icde60146_2024_00441.json |     65 +
 ...aper_doi_10_1109_icde65706_2026_00180.json |     72 +
 ...aper_doi_10_1109_icde65706_2026_00224.json |    149 +
 ...aper_doi_10_1109_icde65706_2026_00240.json |    100 +
 ...oi_10_1109_icicnis64247_2024_10823213.json |     54 +
 ...aper_doi_10_1109_icpc66645_2025_00021.json |    117 +
 ...aper_doi_10_1109_icse43902_2021_00137.json |     84 +
 ...aper_doi_10_1109_icse48619_2023_00024.json |     58 +
 ...aper_doi_10_1109_icse48619_2023_00101.json |    162 +
 ...aper_doi_10_1109_icse48619_2023_00173.json |     57 +
 ...aper_doi_10_1109_icse48619_2023_00175.json |    158 +
 ...aper_doi_10_1109_icse55347_2025_00003.json |     91 +
 ...aper_doi_10_1109_icse55347_2025_00013.json |     59 +
 ...aper_doi_10_1109_icse55347_2025_00045.json |     69 +
 ...aper_doi_10_1109_icse55347_2025_00183.json |     57 +
 ...aper_doi_10_1109_icse55347_2025_00257.json |     95 +
 ...0_1109_icse_companion58688_2023_00041.json |     86 +
 ...doi_10_1109_icse_seip52600_2021_00042.json |     78 +
 ..._doi_10_1109_icsip61881_2024_10671554.json |     57 +
 ...per_doi_10_1109_icsme64153_2025_00030.json |    180 +
 ...aper_doi_10_1109_icst60714_2024_00012.json |     90 +
 ...per_doi_10_1109_ictai62512_2024_00085.json |     61 +
 ...doi_10_1109_iemcon67450_2025_11381190.json |     54 +
 ...er_doi_10_1109_issrew55968_2022_00056.json |     82 +
 ..._doi_10_1109_missf68264_2026_11521893.json |     53 +
 ...oi_10_1109_punecon67554_2025_11378586.json |     56 +
 ...per_doi_10_1109_saner60148_2024_00096.json |     39 +
 .../paper_doi_10_1109_sp54263_2024_00109.json |    110 +
 ...aper_doi_10_1109_srds69199_2025_00038.json |     55 +
 .../paper_doi_10_1109_tdsc_2024_3521591.json  |     57 +
 .../paper_doi_10_1109_tkde_2026_3656491.json  |    108 +
 .../paper_doi_10_1109_tse_2025_3574328.json   |     97 +
 .../paper_doi_10_1109_tse_2025_3625300.json   |    105 +
 .../paper_doi_10_1117_12_3006402.json         |     57 +
 .../paper_doi_10_1142_s021819402150039x.json  |     91 +
 .../paper_doi_10_1145_3372297_3417260.json    |     67 +
 .../paper_doi_10_1145_3395032_3395322.json    |     56 +
 .../paper_doi_10_1145_3428261.json            |     56 +
 .../paper_doi_10_1145_3448016_3457559.json    |     74 +
 .../paper_doi_10_1145_3468264_3468540.json    |     59 +
 .../paper_doi_10_1145_3468264_3468573.json    |     58 +
 .../paper_doi_10_1145_3471485_3471491.json    |     53 +
 .../paper_doi_10_1145_3485529.json            |     57 +
 .../paper_doi_10_1145_3510003_3510093.json    |    136 +
 .../paper_doi_10_1145_3510457_3513034.json    |     75 +
 .../paper_doi_10_1145_3531348_3532176.json    |     57 +
 .../paper_doi_10_1145_3533767_3534364.json    |     60 +
 .../paper_doi_10_1145_3533767_3534409.json    |    156 +
 .../paper_doi_10_1145_3551349_3556924.json    |    131 +
 .../paper_doi_10_1145_3551349_3560431.json    |     98 +
 .../paper_doi_10_1145_3552326_3587448.json    |     60 +
 .../paper_doi_10_1145_3582016_3582053.json    |     55 +
 .../paper_doi_10_1145_3588909.json            |    121 +
 .../paper_doi_10_1145_3597503_3608133.json    |     57 +
 .../paper_doi_10_1145_3597503_3639112.json    |     70 +
 .../paper_doi_10_1145_3597503_3639200.json    |    100 +
 .../paper_doi_10_1145_3597503_3639207.json    |     92 +
 .../paper_doi_10_1145_3597503_3639210.json    |     76 +
 .../paper_doi_10_1145_3597503_3639212.json    |     61 +
 .../paper_doi_10_1145_3597926_3598046.json    |    131 +
 .../paper_doi_10_1145_3597926_3598052.json    |     56 +
 .../paper_doi_10_1145_3597926_3598068.json    |     58 +
 .../paper_doi_10_1145_3597926_3598130.json    |     57 +
 .../paper_doi_10_1145_3605157_3605177.json    |     67 +
 .../paper_doi_10_1145_3611643_3613893.json    |     61 +
 .../paper_doi_10_1145_3611643_3616286.json    |     60 +
 .../paper_doi_10_1145_3622819.json            |    102 +
 .../paper_doi_10_1145_3643779.json            |     57 +
 .../paper_doi_10_1145_3643781.json            |     59 +
 .../paper_doi_10_1145_3650212_3680311.json    |     70 +
 .../paper_doi_10_1145_3650212_3680317.json    |     59 +
 .../paper_doi_10_1145_3650212_3680318.json    |    112 +
 .../paper_doi_10_1145_3650212_3680392.json    |     66 +
 .../paper_doi_10_1145_3689031_3696064.json    |     69 +
 .../paper_doi_10_1145_3689491_3691821.json    |     74 +
 .../paper_doi_10_1145_3689757.json            |     59 +
 .../paper_doi_10_1145_3690631.json            |     60 +
 .../paper_doi_10_1145_3704870.json            |     64 +
 .../paper_doi_10_1145_3708533.json            |     67 +
 .../paper_doi_10_1145_3720504.json            |     58 +
 .../paper_doi_10_1145_3728908.json            |    120 +
 .../paper_doi_10_1145_3728953.json            |     63 +
 .../paper_doi_10_1145_3728965.json            |     62 +
 .../paper_doi_10_1145_3728973.json            |     88 +
 .../paper_doi_10_1145_3729175.json            |     61 +
 .../paper_doi_10_1145_3729319.json            |     57 +
 .../paper_doi_10_1145_3731569_3764841.json    |     62 +
 .../paper_doi_10_1145_3744916_3773102.json    |     59 +
 .../paper_doi_10_1145_3749186.json            |     99 +
 .../paper_doi_10_1145_3757347_3759132.json    |     57 +
 .../paper_doi_10_1145_3758316_3763249.json    |     54 +
 .../paper_doi_10_1145_3764583.json            |    104 +
 .../paper_doi_10_1145_3769779.json            |     84 +
 .../paper_doi_10_1145_3769828.json            |    173 +
 .../paper_doi_10_1145_3769832.json            |     57 +
 .../paper_doi_10_1145_3779212_3790244.json    |     56 +
 .../paper_doi_10_1145_3785021_3787993.json    |     38 +
 .../paper_doi_10_1145_3786673.json            |    101 +
 .../paper_doi_10_1145_3786699.json            |     55 +
 .../paper_doi_10_1145_3798226.json            |     60 +
 .../paper_doi_10_1145_3798232.json            |     61 +
 .../paper_doi_10_1145_3798245.json            |     57 +
 .../paper_doi_10_1145_3799227.json            |    152 +
 .../paper_doi_10_1145_3802034.json            |     59 +
 .../paper_doi_10_1145_3802053.json            |     62 +
 .../paper_doi_10_1145_3802061.json            |     65 +
 .../paper_doi_10_1145_3803437_3806090.json    |     60 +
 .../paper_doi_10_1145_3808109.json            |     59 +
 .../paper_doi_10_1145_3810991_3811632.json    |    127 +
 .../paper_doi_10_1145_3810991_3811634.json    |     54 +
 .../paper_doi_10_1145_3810991_3811637.json    |     89 +
 .../paper_doi_10_1145_3828685.json            |     57 +
 .../paper_doi_10_1177_0926227x251370258.json  |    125 +
 ...oi_10_14711_thesis_991012980220103412.json |    120 +
 .../paper_doi_10_14722_ndss_2025_230530.json  |     58 +
 .../paper_doi_10_14722_ndss_2026_240198.json  |     90 +
 .../paper_doi_10_14778_3494124_3494139.json   |     73 +
 .../paper_doi_10_14778_3611540_3611584.json   |     69 +
 .../paper_doi_10_14778_3636218_3636236.json   |    180 +
 .../paper_doi_10_14778_3659437_3659445.json   |    135 +
 .../paper_doi_10_14778_3712221_3712247.json   |    211 +
 .../paper_doi_10_14778_3725688_3725698.json   |     74 +
 .../paper_doi_10_14778_3725688_3725713.json   |    157 +
 .../paper_doi_10_14778_3734839_3734861.json   |    104 +
 .../paper_doi_10_14778_3742728_3742747.json   |    113 +
 .../paper_doi_10_14778_3749646_3749661.json   |     67 +
 .../paper_doi_10_14778_3749646_3749683.json   |     57 +
 .../paper_doi_10_14778_3797919_3797928.json   |     58 +
 .../paper_doi_10_14778_3819518_3819547.json   |     61 +
 .../paper_doi_10_14778_3828612_3828639.json   |     56 +
 .../paper_doi_10_14778_3836663_3836700.json   |     50 +
 .../paper_doi_10_22399_ijcesen_5462.json      |     88 +
 .../paper_doi_10_32672_jnkti_v6i1_5830.json   |     56 +
 .../paper_doi_10_3390_app13042519.json        |     67 +
 ...paper_doi_10_3390_electronics14193910.json |     62 +
 ...rnalmantik_vol5_2021_1448_pp1065_1071.json |     37 +
 .../paper_doi_10_3929_ethz_b_000507577.json   |     72 +
 .../paper_doi_10_53799_zb6b1375.json          |     56 +
 .../paper_doi_10_7717_peerj_cs_1592.json      |     77 +
 ...4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85.json |     59 +
 ...c629691d5654def2d9d85f72c756c67c68583.json |     77 +
 ...373bd891702ae93d7e058241a7e8be25e89eb.json |     98 +
 ...f8601da9663add6a85713414f8c945cf97355.json |     57 +
 ...aa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json |    135 +
 ...637ad3297e0d4e45a4f5245d0472a82a59268.json |     54 +
 ...64dbe0130ef1bf6af12266b958a5d5396101f.json |     53 +
 ...2bcb84acde8c7fe8964aba2d6d27f220ae80b.json |    142 +
 ...db0b52f31ed5444602a67e9a275faf57e15bf.json |     57 +
 ...3042eb3c2a8f2e208515490d6b6a400e00fec.json |    118 +
 _data/impact/papers.json                      |  25820 ++++
 _data/impact/people.json                      |    883 +
 _data/impact/people_decisions.json            |     12 +
 _data/impact/policy.json                      |    204 +
 _data/impact/recognition_highlights.json      |     51 +
 _data/impact/resources.json                   |    844 +
 _data/impact/stats.json                       |   7952 +
 _data/impact/talks.json                       |    888 +
 _data/impact/techniques.json                  |    502 +
 _data/navigation.yml                          |      4 +
 _data/papers/paper_arxiv_2105_10016.json      |    883 +
 _data/papers/paper_arxiv_2206_08530.json      |    596 +
 _data/papers/paper_arxiv_2304_10044.json      |    438 +
 _data/papers/paper_arxiv_2310_06433.json      |    529 +
 _data/papers/paper_arxiv_2311_06728.json      |   2011 +
 _data/papers/paper_arxiv_2312_04941.json      |   1247 +
 _data/papers/paper_arxiv_2402_00292.json      |    518 +
 _data/papers/paper_arxiv_2406_09469.json      |   1417 +
 _data/papers/paper_arxiv_2407_04294.json      |   1366 +
 _data/papers/paper_arxiv_2503_03893.json      |    947 +
 _data/papers/paper_arxiv_2503_17322.json      |    517 +
 _data/papers/paper_arxiv_2506_02617.json      |    479 +
 _data/papers/paper_arxiv_2509_10819.json      |    374 +
 _data/papers/paper_arxiv_2510_06663.json      |   2122 +
 _data/papers/paper_arxiv_2511_17377.json      |    754 +
 _data/papers/paper_arxiv_2601_15074.json      |    578 +
 _data/papers/paper_arxiv_2602_19490.json      |    867 +
 _data/papers/paper_arxiv_2603_00311.json      |    470 +
 _data/papers/paper_arxiv_2603_19434.json      |    305 +
 _data/papers/paper_arxiv_2603_21530.json      |    666 +
 _data/papers/paper_arxiv_2604_01442.json      |    579 +
 _data/papers/paper_arxiv_2604_03024.json      |    670 +
 _data/papers/paper_arxiv_2605_20473.json      |    856 +
 _data/papers/paper_arxiv_2605_22992.json      |    832 +
 _data/papers/paper_arxiv_2606_11132.json      |    599 +
 _data/papers/paper_arxiv_2606_14164.json      |    234 +
 _data/papers/paper_arxiv_2607_03741.json      |    685 +
 _data/papers/paper_arxiv_2607_09072.json      |    362 +
 _data/papers/paper_arxiv_2607_13276.json      |    433 +
 _data/papers/paper_arxiv_2608_15709.json      |    321 +
 _data/papers/paper_arxiv_2608_23402.json      |    519 +
 _data/papers/paper_arxiv_2608_25573.json      |    607 +
 _data/papers/paper_arxiv_2608_30385.json      |    944 +
 _data/papers/paper_arxiv_2609_00381.json      |    258 +
 ...paper_doi_10_1007_978_3_030_71058_3_5.json |    314 +
 ...aper_doi_10_1007_978_3_030_88494_9_12.json |    414 +
 ...aper_doi_10_1007_978_3_031_51479_1_17.json |    474 +
 ...paper_doi_10_1007_978_3_031_94706_3_7.json |    429 +
 ...paper_doi_10_1007_978_981_95_3182_0_3.json |    484 +
 ...paper_doi_10_1007_978_981_95_4721_0_7.json |    454 +
 ...aper_doi_10_1007_978_981_96_4506_0_18.json |    831 +
 ...aper_doi_10_1007_978_981_96_6465_8_28.json |    420 +
 .../paper_doi_10_1007_s10664_025_10662_w.json |    192 +
 .../paper_doi_10_1016_j_cose_2025_104564.json |   1279 +
 ...aper_doi_10_1016_j_displa_2024_102854.json |    382 +
 ...paper_doi_10_1109_ase56229_2023_00106.json |    618 +
 ...paper_doi_10_1109_ase63991_2025_00095.json |    710 +
 ...paper_doi_10_1109_ase63991_2025_00151.json |    750 +
 ...paper_doi_10_1109_ase63991_2025_00322.json |    227 +
 ...r_doi_10_1109_compsac57700_2023_00273.json |     92 +
 ...r_doi_10_1109_compsac61105_2024_00141.json |    567 +
 ...paper_doi_10_1109_dsc55868_2022_00057.json |    438 +
 ..._doi_10_1109_girst67753_2025_11382165.json |    356 +
 ...doi_10_1109_iaecst64597_2024_11117732.json |    331 +
 ...doi_10_1109_iaecst68792_2025_11415166.json |    390 +
 ..._doi_10_1109_icccs65393_2025_11069832.json |    396 +
 ...doi_10_1109_iccste65902_2025_11138310.json |    258 +
 ...aper_doi_10_1109_icde55515_2023_00057.json |   1009 +
 ...aper_doi_10_1109_icde60146_2024_00011.json |    752 +
 ...aper_doi_10_1109_icde60146_2024_00441.json |    743 +
 ...aper_doi_10_1109_icde65706_2026_00180.json |   1015 +
 ...aper_doi_10_1109_icde65706_2026_00224.json |   1511 +
 ...aper_doi_10_1109_icde65706_2026_00240.json |    933 +
 ...oi_10_1109_icicnis64247_2024_10823213.json |    320 +
 ...aper_doi_10_1109_icpc66645_2025_00021.json |   2345 +
 ...aper_doi_10_1109_icse43902_2021_00137.json |    654 +
 ...aper_doi_10_1109_icse48619_2023_00024.json |    510 +
 ...aper_doi_10_1109_icse48619_2023_00101.json |    992 +
 ...aper_doi_10_1109_icse48619_2023_00173.json |    467 +
 ...aper_doi_10_1109_icse48619_2023_00175.json |   1124 +
 ...aper_doi_10_1109_icse55347_2025_00003.json |    715 +
 ...aper_doi_10_1109_icse55347_2025_00013.json |    493 +
 ...aper_doi_10_1109_icse55347_2025_00045.json |    665 +
 ...aper_doi_10_1109_icse55347_2025_00183.json |    541 +
 ...aper_doi_10_1109_icse55347_2025_00257.json |    891 +
 ...0_1109_icse_companion58688_2023_00041.json |    574 +
 ...doi_10_1109_icse_seip52600_2021_00042.json |    950 +
 ..._doi_10_1109_icsip61881_2024_10671554.json |    614 +
 ...per_doi_10_1109_icsme64153_2025_00030.json |    821 +
 ...aper_doi_10_1109_icst60714_2024_00012.json |   1033 +
 ...per_doi_10_1109_ictai62512_2024_00085.json |    675 +
 ...doi_10_1109_iemcon67450_2025_11381190.json |    433 +
 ...er_doi_10_1109_issrew55968_2022_00056.json |    317 +
 ..._doi_10_1109_missf68264_2026_11521893.json |    259 +
 ...oi_10_1109_punecon67554_2025_11378586.json |    194 +
 ...per_doi_10_1109_saner60148_2024_00096.json |    406 +
 .../paper_doi_10_1109_sp54263_2024_00109.json |    696 +
 ...aper_doi_10_1109_srds69199_2025_00038.json |    367 +
 .../paper_doi_10_1109_tdsc_2024_3521591.json  |    500 +
 .../paper_doi_10_1109_tkde_2026_3656491.json  |    521 +
 .../paper_doi_10_1109_tse_2025_3574328.json   |   1256 +
 .../paper_doi_10_1109_tse_2025_3625300.json   |    861 +
 .../papers/paper_doi_10_1117_12_3006402.json  |    248 +
 .../paper_doi_10_1145_3372297_3417260.json    |    553 +
 .../paper_doi_10_1145_3395032_3395322.json    |    210 +
 _data/papers/paper_doi_10_1145_3428261.json   |    165 +
 .../paper_doi_10_1145_3448016_3457559.json    |    641 +
 .../paper_doi_10_1145_3468264_3468540.json    |    523 +
 .../paper_doi_10_1145_3468264_3468573.json    |    630 +
 .../paper_doi_10_1145_3471485_3471491.json    |    173 +
 _data/papers/paper_doi_10_1145_3485529.json   |    157 +
 .../paper_doi_10_1145_3510003_3510093.json    |    730 +
 .../paper_doi_10_1145_3510457_3513034.json    |    268 +
 .../paper_doi_10_1145_3531348_3532176.json    |    304 +
 .../paper_doi_10_1145_3533767_3534364.json    |    953 +
 .../paper_doi_10_1145_3533767_3534409.json    |    729 +
 .../paper_doi_10_1145_3551349_3556924.json    |    684 +
 .../paper_doi_10_1145_3551349_3560431.json    |    786 +
 .../paper_doi_10_1145_3552326_3587448.json    |    827 +
 .../paper_doi_10_1145_3582016_3582053.json    |    437 +
 _data/papers/paper_doi_10_1145_3588909.json   |    940 +
 .../paper_doi_10_1145_3597503_3608133.json    |    404 +
 .../paper_doi_10_1145_3597503_3639112.json    |   1095 +
 .../paper_doi_10_1145_3597503_3639200.json    |    961 +
 .../paper_doi_10_1145_3597503_3639207.json    |    801 +
 .../paper_doi_10_1145_3597503_3639210.json    |    569 +
 .../paper_doi_10_1145_3597503_3639212.json    |    723 +
 .../paper_doi_10_1145_3597926_3598046.json    |    532 +
 .../paper_doi_10_1145_3597926_3598052.json    |    516 +
 .../paper_doi_10_1145_3597926_3598068.json    |    729 +
 .../paper_doi_10_1145_3597926_3598130.json    |    755 +
 .../paper_doi_10_1145_3605157_3605177.json    |    474 +
 .../paper_doi_10_1145_3611643_3613893.json    |    470 +
 .../paper_doi_10_1145_3611643_3616286.json    |    622 +
 _data/papers/paper_doi_10_1145_3622819.json   |    191 +
 _data/papers/paper_doi_10_1145_3643779.json   |    161 +
 _data/papers/paper_doi_10_1145_3643781.json   |    478 +
 .../paper_doi_10_1145_3650212_3680311.json    |    742 +
 .../paper_doi_10_1145_3650212_3680317.json    |    629 +
 .../paper_doi_10_1145_3650212_3680318.json    |   1045 +
 .../paper_doi_10_1145_3650212_3680392.json    |   1010 +
 .../paper_doi_10_1145_3689031_3696064.json    |    884 +
 .../paper_doi_10_1145_3689491_3691821.json    |    247 +
 _data/papers/paper_doi_10_1145_3689757.json   |    637 +
 _data/papers/paper_doi_10_1145_3690631.json   |    479 +
 _data/papers/paper_doi_10_1145_3704870.json   |    148 +
 _data/papers/paper_doi_10_1145_3708533.json   |   1286 +
 _data/papers/paper_doi_10_1145_3720504.json   |    587 +
 _data/papers/paper_doi_10_1145_3728908.json   |    990 +
 _data/papers/paper_doi_10_1145_3728953.json   |    920 +
 _data/papers/paper_doi_10_1145_3728965.json   |    781 +
 _data/papers/paper_doi_10_1145_3728973.json   |    778 +
 _data/papers/paper_doi_10_1145_3729175.json   |    164 +
 _data/papers/paper_doi_10_1145_3729319.json   |    545 +
 .../paper_doi_10_1145_3731569_3764841.json    |    573 +
 .../paper_doi_10_1145_3744916_3773102.json    |    781 +
 _data/papers/paper_doi_10_1145_3749186.json   |    983 +
 .../paper_doi_10_1145_3757347_3759132.json    |    414 +
 .../paper_doi_10_1145_3758316_3763249.json    |    243 +
 _data/papers/paper_doi_10_1145_3764583.json   |   1290 +
 _data/papers/paper_doi_10_1145_3769779.json   |    943 +
 _data/papers/paper_doi_10_1145_3769828.json   |   1667 +
 _data/papers/paper_doi_10_1145_3769832.json   |    404 +
 .../paper_doi_10_1145_3779212_3790244.json    |    976 +
 .../paper_doi_10_1145_3785021_3787993.json    |    431 +
 _data/papers/paper_doi_10_1145_3786673.json   |   1085 +
 _data/papers/paper_doi_10_1145_3786699.json   |    467 +
 _data/papers/paper_doi_10_1145_3798226.json   |    589 +
 _data/papers/paper_doi_10_1145_3798232.json   |    829 +
 _data/papers/paper_doi_10_1145_3798245.json   |    793 +
 _data/papers/paper_doi_10_1145_3799227.json   |   2562 +
 _data/papers/paper_doi_10_1145_3802034.json   |   1048 +
 _data/papers/paper_doi_10_1145_3802053.json   |   1064 +
 _data/papers/paper_doi_10_1145_3802061.json   |    857 +
 .../paper_doi_10_1145_3803437_3806090.json    |    286 +
 _data/papers/paper_doi_10_1145_3808109.json   |    494 +
 .../paper_doi_10_1145_3810991_3811632.json    |   1318 +
 .../paper_doi_10_1145_3810991_3811634.json    |    382 +
 .../paper_doi_10_1145_3810991_3811637.json    |    355 +
 _data/papers/paper_doi_10_1145_3828685.json   |    476 +
 .../paper_doi_10_1177_0926227x251370258.json  |   1314 +
 ...oi_10_14711_thesis_991012980220103412.json |    654 +
 .../paper_doi_10_14722_ndss_2025_230530.json  |    854 +
 .../paper_doi_10_14722_ndss_2026_240198.json  |    988 +
 .../paper_doi_10_14778_3712221_3712247.json   |   1410 +
 .../paper_doi_10_14778_3749646_3749661.json   |    424 +
 .../paper_doi_10_14778_3797919_3797928.json   |    885 +
 .../paper_doi_10_14778_3828612_3828639.json   |    705 +
 .../paper_doi_10_22399_ijcesen_5462.json      |    475 +
 .../papers/paper_doi_10_3390_app13042519.json |    522 +
 ...paper_doi_10_3390_electronics14193910.json |    401 +
 _data/papers/paper_doi_10_53799_zb6b1375.json |    352 +
 .../paper_doi_10_7717_peerj_cs_1592.json      |    148 +
 ...c629691d5654def2d9d85f72c756c67c68583.json |    821 +
 ...373bd891702ae93d7e058241a7e8be25e89eb.json |   1065 +
 ...aa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json |   2384 +
 ...637ad3297e0d4e45a4f5245d0472a82a59268.json |    583 +
 ...2bcb84acde8c7fe8964aba2d6d27f220ae80b.json |   1368 +
 ...db0b52f31ed5444602a67e9a275faf57e15bf.json |    563 +
 ...3042eb3c2a8f2e208515490d6b6a400e00fec.json |   1004 +
 _includes/head/custom.html                    |      8 +
 _includes/impact/artifact-evidence.html       |     68 +
 _includes/impact/bar-list.html                |     26 +
 _includes/impact/counts-table.html            |     34 +
 _includes/impact/dbms-detail.html             |    214 +
 _includes/impact/feature-row.html             |     80 +
 _includes/impact/headline.html                |     28 +
 _includes/impact/paper-detail.html            |    212 +
 _includes/impact/paper-list.html              |    113 +
 _includes/impact/plots/bugs-by-dbms.svg       |    227 +
 _includes/impact/plots/bugs-by-reporter.svg   |      9 +
 _includes/impact/plots/bugs-by-status.svg     |      9 +
 _includes/impact/plots/bugs-by-technique.svg  |     25 +
 _includes/impact/plots/bugs-by-year.svg       |     55 +
 _includes/impact/plots/papers-by-year.svg     |     36 +
 .../plots/papers-relationships-by-year.svg    |     54 +
 _includes/impact/stat-tiles.html              |     30 +
 _includes/impact/talk-detail.html             |    114 +
 _includes/impact/talk-frames.html             |     33 +
 _includes/impact/year-detail.html             |    130 +
 _pages/found-bugs.md                          |     20 +-
 _pages/impact.html                            |    636 +
 _pages/impact/bug-statistics.html             |     68 +
 _pages/impact/bugs/2019.html                  |     10 +
 _pages/impact/bugs/2020.html                  |     10 +
 _pages/impact/bugs/2021.html                  |     10 +
 _pages/impact/bugs/2022.html                  |     10 +
 _pages/impact/bugs/2023.html                  |     10 +
 _pages/impact/bugs/2024.html                  |     10 +
 _pages/impact/bugs/2025.html                  |     10 +
 _pages/impact/bugs/2026.html                  |     10 +
 _pages/impact/database-systems.html           |    160 +
 _pages/impact/dbms/bharatdbms.html            |     10 +
 _pages/impact/dbms/citus.html                 |     10 +
 _pages/impact/dbms/clickhouse.html            |     10 +
 _pages/impact/dbms/cloudberry.html            |     10 +
 _pages/impact/dbms/cnosdb.html                |     10 +
 _pages/impact/dbms/cockroachdb.html           |     10 +
 _pages/impact/dbms/cratedb.html               |     10 +
 _pages/impact/dbms/cubrid.html                |     10 +
 _pages/impact/dbms/databend.html              |     10 +
 _pages/impact/dbms/datafusion.html            |     10 +
 _pages/impact/dbms/dolt.html                  |     10 +
 _pages/impact/dbms/doris.html                 |     10 +
 _pages/impact/dbms/duckdb.html                |     10 +
 _pages/impact/dbms/falkordb.html              |     10 +
 _pages/impact/dbms/feldera.html               |     10 +
 _pages/impact/dbms/firebird.html              |     10 +
 _pages/impact/dbms/greptimedb.html            |     10 +
 _pages/impact/dbms/h2.html                    |     10 +
 _pages/impact/dbms/hazelcast.html             |     10 +
 _pages/impact/dbms/hive.html                  |     10 +
 _pages/impact/dbms/hsqldb.html                |     10 +
 _pages/impact/dbms/kyzo.html                  |     10 +
 _pages/impact/dbms/mariadb.html               |     10 +
 _pages/impact/dbms/materialize.html           |     10 +
 _pages/impact/dbms/matrixone.html             |     10 +
 _pages/impact/dbms/monetdb.html               |     10 +
 _pages/impact/dbms/mysql.html                 |     10 +
 _pages/impact/dbms/noisepage.html             |     10 +
 _pages/impact/dbms/oceanbase.html             |     10 +
 _pages/impact/dbms/oxla.html                  |     10 +
 _pages/impact/dbms/postgresql.html            |     10 +
 _pages/impact/dbms/presto.html                |     10 +
 _pages/impact/dbms/questdb.html               |     10 +
 _pages/impact/dbms/risingwave.html            |     10 +
 _pages/impact/dbms/seekdb.html                |     10 +
 _pages/impact/dbms/serenedb.html              |     10 +
 _pages/impact/dbms/spark.html                 |     10 +
 _pages/impact/dbms/sparq.html                 |     10 +
 _pages/impact/dbms/spiceai.html               |     10 +
 _pages/impact/dbms/sqlite.html                |     10 +
 _pages/impact/dbms/starrocks.html             |     10 +
 _pages/impact/dbms/stonedb.html               |     10 +
 _pages/impact/dbms/tarantool.html             |     10 +
 _pages/impact/dbms/tdengine.html              |     10 +
 _pages/impact/dbms/tidb.html                  |     10 +
 _pages/impact/dbms/tikv.html                  |     10 +
 _pages/impact/dbms/turso.html                 |     10 +
 _pages/impact/dbms/umbra.html                 |     10 +
 _pages/impact/dbms/virtuoso.html              |     10 +
 _pages/impact/dbms/wadjet.html                |     10 +
 _pages/impact/dbms/xugu.html                  |     10 +
 _pages/impact/dbms/ydb.html                   |     10 +
 _pages/impact/dbms/yugabytedb.html            |     10 +
 .../impact/papers/paper_arxiv_2105_10016.html |     10 +
 .../impact/papers/paper_arxiv_2206_08530.html |     10 +
 .../impact/papers/paper_arxiv_2304_10044.html |     10 +
 .../impact/papers/paper_arxiv_2310_06433.html |     10 +
 .../impact/papers/paper_arxiv_2311_06728.html |     10 +
 .../impact/papers/paper_arxiv_2312_04941.html |     10 +
 .../impact/papers/paper_arxiv_2402_00292.html |     10 +
 .../impact/papers/paper_arxiv_2406_09469.html |     10 +
 .../impact/papers/paper_arxiv_2407_04294.html |     10 +
 .../impact/papers/paper_arxiv_2503_03893.html |     10 +
 .../impact/papers/paper_arxiv_2503_17322.html |     10 +
 .../impact/papers/paper_arxiv_2506_02617.html |     10 +
 .../impact/papers/paper_arxiv_2509_10819.html |     10 +
 .../impact/papers/paper_arxiv_2510_06663.html |     10 +
 .../impact/papers/paper_arxiv_2511_17377.html |     10 +
 .../impact/papers/paper_arxiv_2601_15074.html |     10 +
 .../impact/papers/paper_arxiv_2602_19490.html |     10 +
 .../impact/papers/paper_arxiv_2603_00311.html |     10 +
 .../impact/papers/paper_arxiv_2603_19434.html |     10 +
 .../impact/papers/paper_arxiv_2603_21530.html |     10 +
 .../impact/papers/paper_arxiv_2604_01442.html |     10 +
 .../impact/papers/paper_arxiv_2604_03024.html |     10 +
 .../impact/papers/paper_arxiv_2605_20473.html |     10 +
 .../impact/papers/paper_arxiv_2605_22992.html |     10 +
 .../impact/papers/paper_arxiv_2606_11132.html |     10 +
 .../impact/papers/paper_arxiv_2606_14164.html |     10 +
 .../impact/papers/paper_arxiv_2607_03741.html |     10 +
 .../impact/papers/paper_arxiv_2607_09072.html |     10 +
 .../impact/papers/paper_arxiv_2607_13276.html |     10 +
 .../impact/papers/paper_arxiv_2608_15709.html |     10 +
 .../impact/papers/paper_arxiv_2608_23402.html |     10 +
 .../impact/papers/paper_arxiv_2608_25573.html |     10 +
 .../impact/papers/paper_arxiv_2608_30385.html |     10 +
 .../impact/papers/paper_arxiv_2609_00381.html |     10 +
 ...paper_doi_10_1007_978_3_030_71058_3_5.html |     10 +
 ...aper_doi_10_1007_978_3_030_88494_9_12.html |     10 +
 ...aper_doi_10_1007_978_3_031_51479_1_17.html |     10 +
 ...paper_doi_10_1007_978_3_031_94706_3_7.html |     10 +
 ...paper_doi_10_1007_978_981_95_3182_0_3.html |     10 +
 ...paper_doi_10_1007_978_981_95_4721_0_7.html |     10 +
 ...aper_doi_10_1007_978_981_96_4506_0_18.html |     10 +
 ...aper_doi_10_1007_978_981_96_6465_8_28.html |     10 +
 .../paper_doi_10_1007_s10664_025_10662_w.html |     10 +
 .../paper_doi_10_1016_j_cose_2025_104564.html |     10 +
 ...aper_doi_10_1016_j_displa_2024_102854.html |     10 +
 ...paper_doi_10_1109_ase56229_2023_00106.html |     10 +
 ...paper_doi_10_1109_ase63991_2025_00095.html |     10 +
 ...paper_doi_10_1109_ase63991_2025_00151.html |     10 +
 ...paper_doi_10_1109_ase63991_2025_00322.html |     10 +
 ...r_doi_10_1109_compsac57700_2023_00273.html |     10 +
 ...r_doi_10_1109_compsac61105_2024_00141.html |     10 +
 ...paper_doi_10_1109_dsc55868_2022_00057.html |     10 +
 ..._doi_10_1109_girst67753_2025_11382165.html |     10 +
 ...doi_10_1109_iaecst64597_2024_11117732.html |     10 +
 ...doi_10_1109_iaecst68792_2025_11415166.html |     10 +
 ..._doi_10_1109_icccs65393_2025_11069832.html |     10 +
 ...doi_10_1109_iccste65902_2025_11138310.html |     10 +
 ...aper_doi_10_1109_icde55515_2023_00057.html |     10 +
 ...aper_doi_10_1109_icde60146_2024_00011.html |     10 +
 ...aper_doi_10_1109_icde60146_2024_00441.html |     10 +
 ...aper_doi_10_1109_icde65706_2026_00180.html |     10 +
 ...aper_doi_10_1109_icde65706_2026_00224.html |     10 +
 ...aper_doi_10_1109_icde65706_2026_00240.html |     10 +
 ...oi_10_1109_icicnis64247_2024_10823213.html |     10 +
 ...aper_doi_10_1109_icpc66645_2025_00021.html |     10 +
 ...aper_doi_10_1109_icse43902_2021_00137.html |     10 +
 ...aper_doi_10_1109_icse48619_2023_00024.html |     10 +
 ...aper_doi_10_1109_icse48619_2023_00101.html |     10 +
 ...aper_doi_10_1109_icse48619_2023_00173.html |     10 +
 ...aper_doi_10_1109_icse48619_2023_00175.html |     10 +
 ...aper_doi_10_1109_icse55347_2025_00003.html |     10 +
 ...aper_doi_10_1109_icse55347_2025_00013.html |     10 +
 ...aper_doi_10_1109_icse55347_2025_00045.html |     10 +
 ...aper_doi_10_1109_icse55347_2025_00183.html |     10 +
 ...aper_doi_10_1109_icse55347_2025_00257.html |     10 +
 ...0_1109_icse_companion58688_2023_00041.html |     10 +
 ...doi_10_1109_icse_seip52600_2021_00042.html |     10 +
 ..._doi_10_1109_icsip61881_2024_10671554.html |     10 +
 ...per_doi_10_1109_icsme64153_2025_00030.html |     10 +
 ...aper_doi_10_1109_icst60714_2024_00012.html |     10 +
 ...per_doi_10_1109_ictai62512_2024_00085.html |     10 +
 ...doi_10_1109_iemcon67450_2025_11381190.html |     10 +
 ...er_doi_10_1109_issrew55968_2022_00056.html |     10 +
 ..._doi_10_1109_missf68264_2026_11521893.html |     10 +
 ...oi_10_1109_punecon67554_2025_11378586.html |     10 +
 ...per_doi_10_1109_saner60148_2024_00096.html |     10 +
 .../paper_doi_10_1109_sp54263_2024_00109.html |     10 +
 ...aper_doi_10_1109_srds69199_2025_00038.html |     10 +
 .../paper_doi_10_1109_tdsc_2024_3521591.html  |     10 +
 .../paper_doi_10_1109_tkde_2026_3656491.html  |     10 +
 .../paper_doi_10_1109_tse_2025_3574328.html   |     10 +
 .../paper_doi_10_1109_tse_2025_3625300.html   |     10 +
 .../papers/paper_doi_10_1117_12_3006402.html  |     10 +
 .../paper_doi_10_1145_3372297_3417260.html    |     10 +
 .../paper_doi_10_1145_3395032_3395322.html    |     10 +
 .../papers/paper_doi_10_1145_3428261.html     |     10 +
 .../paper_doi_10_1145_3448016_3457559.html    |     10 +
 .../paper_doi_10_1145_3468264_3468540.html    |     10 +
 .../paper_doi_10_1145_3468264_3468573.html    |     10 +
 .../paper_doi_10_1145_3471485_3471491.html    |     10 +
 .../papers/paper_doi_10_1145_3485529.html     |     10 +
 .../paper_doi_10_1145_3510003_3510093.html    |     10 +
 .../paper_doi_10_1145_3510457_3513034.html    |     10 +
 .../paper_doi_10_1145_3531348_3532176.html    |     10 +
 .../paper_doi_10_1145_3533767_3534364.html    |     10 +
 .../paper_doi_10_1145_3533767_3534409.html    |     10 +
 .../paper_doi_10_1145_3551349_3556924.html    |     10 +
 .../paper_doi_10_1145_3551349_3560431.html    |     10 +
 .../paper_doi_10_1145_3552326_3587448.html    |     10 +
 .../paper_doi_10_1145_3582016_3582053.html    |     10 +
 .../papers/paper_doi_10_1145_3588909.html     |     10 +
 .../paper_doi_10_1145_3597503_3608133.html    |     10 +
 .../paper_doi_10_1145_3597503_3639112.html    |     10 +
 .../paper_doi_10_1145_3597503_3639200.html    |     10 +
 .../paper_doi_10_1145_3597503_3639207.html    |     10 +
 .../paper_doi_10_1145_3597503_3639210.html    |     10 +
 .../paper_doi_10_1145_3597503_3639212.html    |     10 +
 .../paper_doi_10_1145_3597926_3598046.html    |     10 +
 .../paper_doi_10_1145_3597926_3598052.html    |     10 +
 .../paper_doi_10_1145_3597926_3598068.html    |     10 +
 .../paper_doi_10_1145_3597926_3598130.html    |     10 +
 .../paper_doi_10_1145_3605157_3605177.html    |     10 +
 .../paper_doi_10_1145_3611643_3613893.html    |     10 +
 .../paper_doi_10_1145_3611643_3616286.html    |     10 +
 .../papers/paper_doi_10_1145_3622819.html     |     10 +
 .../papers/paper_doi_10_1145_3643779.html     |     10 +
 .../papers/paper_doi_10_1145_3643781.html     |     10 +
 .../paper_doi_10_1145_3650212_3680311.html    |     10 +
 .../paper_doi_10_1145_3650212_3680317.html    |     10 +
 .../paper_doi_10_1145_3650212_3680318.html    |     10 +
 .../paper_doi_10_1145_3650212_3680392.html    |     10 +
 .../paper_doi_10_1145_3689031_3696064.html    |     10 +
 .../paper_doi_10_1145_3689491_3691821.html    |     10 +
 .../papers/paper_doi_10_1145_3689757.html     |     10 +
 .../papers/paper_doi_10_1145_3690631.html     |     10 +
 .../papers/paper_doi_10_1145_3704870.html     |     10 +
 .../papers/paper_doi_10_1145_3708533.html     |     10 +
 .../papers/paper_doi_10_1145_3720504.html     |     10 +
 .../papers/paper_doi_10_1145_3728908.html     |     10 +
 .../papers/paper_doi_10_1145_3728953.html     |     10 +
 .../papers/paper_doi_10_1145_3728965.html     |     10 +
 .../papers/paper_doi_10_1145_3728973.html     |     10 +
 .../papers/paper_doi_10_1145_3729175.html     |     10 +
 .../papers/paper_doi_10_1145_3729319.html     |     10 +
 .../paper_doi_10_1145_3731569_3764841.html    |     10 +
 .../paper_doi_10_1145_3744916_3773102.html    |     10 +
 .../papers/paper_doi_10_1145_3749186.html     |     10 +
 .../paper_doi_10_1145_3757347_3759132.html    |     10 +
 .../paper_doi_10_1145_3758316_3763249.html    |     10 +
 .../papers/paper_doi_10_1145_3764583.html     |     10 +
 .../papers/paper_doi_10_1145_3769779.html     |     10 +
 .../papers/paper_doi_10_1145_3769828.html     |     10 +
 .../papers/paper_doi_10_1145_3769832.html     |     10 +
 .../paper_doi_10_1145_3779212_3790244.html    |     10 +
 .../paper_doi_10_1145_3785021_3787993.html    |     10 +
 .../papers/paper_doi_10_1145_3786673.html     |     10 +
 .../papers/paper_doi_10_1145_3786699.html     |     10 +
 .../papers/paper_doi_10_1145_3798226.html     |     10 +
 .../papers/paper_doi_10_1145_3798232.html     |     10 +
 .../papers/paper_doi_10_1145_3798245.html     |     10 +
 .../papers/paper_doi_10_1145_3799227.html     |     10 +
 .../papers/paper_doi_10_1145_3802034.html     |     10 +
 .../papers/paper_doi_10_1145_3802053.html     |     10 +
 .../papers/paper_doi_10_1145_3802061.html     |     10 +
 .../paper_doi_10_1145_3803437_3806090.html    |     10 +
 .../papers/paper_doi_10_1145_3808109.html     |     10 +
 .../paper_doi_10_1145_3810991_3811632.html    |     10 +
 .../paper_doi_10_1145_3810991_3811634.html    |     10 +
 .../paper_doi_10_1145_3810991_3811637.html    |     10 +
 .../papers/paper_doi_10_1145_3828685.html     |     10 +
 .../paper_doi_10_1177_0926227x251370258.html  |     10 +
 ...oi_10_14711_thesis_991012980220103412.html |     10 +
 .../paper_doi_10_14722_ndss_2025_230530.html  |     10 +
 .../paper_doi_10_14722_ndss_2026_240198.html  |     10 +
 .../paper_doi_10_14778_3712221_3712247.html   |     10 +
 .../paper_doi_10_14778_3749646_3749661.html   |     10 +
 .../paper_doi_10_14778_3797919_3797928.html   |     10 +
 .../paper_doi_10_14778_3828612_3828639.html   |     10 +
 .../paper_doi_10_22399_ijcesen_5462.html      |     10 +
 .../papers/paper_doi_10_3390_app13042519.html |     10 +
 ...paper_doi_10_3390_electronics14193910.html |     10 +
 .../papers/paper_doi_10_53799_zb6b1375.html   |     10 +
 .../paper_doi_10_7717_peerj_cs_1592.html      |     10 +
 ...c629691d5654def2d9d85f72c756c67c68583.html |     10 +
 ...373bd891702ae93d7e058241a7e8be25e89eb.html |     10 +
 ...aa12e3aa52f3cb41009e9e8f65c79c9ef9f42.html |     10 +
 ...637ad3297e0d4e45a4f5245d0472a82a59268.html |     10 +
 ...2bcb84acde8c7fe8964aba2d6d27f220ae80b.html |     10 +
 ...db0b52f31ed5444602a67e9a275faf57e15bf.html |     10 +
 ...3042eb3c2a8f2e208515490d6b6a400e00fec.html |     10 +
 _pages/impact/recognition.html                |     58 +
 _pages/impact/talks/645a47ac1426.html         |     10 +
 _pages/impact/talks/6YGqFRTe2D0.html          |     10 +
 _pages/impact/talks/9d70cce25a55.html         |     10 +
 _pages/impact/talks/BgC79Zt2fPs.html          |     10 +
 _pages/impact/talks/CW4Ntdtp7lg.html          |     10 +
 _pages/impact/talks/L90MBb6NLBE.html          |     10 +
 _pages/impact/talks/QRwxHGpWaUA.html          |     10 +
 _pages/impact/talks/V_qzqY1bb7I.html          |     10 +
 _pages/impact/talks/wHo-VtzTHx0.html          |     10 +
 _pages/splash-page.md                         |     55 +-
 _pages/supported-databases.md                 |     40 +-
 assets/css/impact.css                         |    487 +
 .../images/impact/talks/6YGqFRTe2D0-1509.jpg  |    Bin 0 -> 73144 bytes
 .../images/impact/talks/6YGqFRTe2D0-295.jpg   |    Bin 0 -> 59480 bytes
 .../images/impact/talks/BgC79Zt2fPs-2144.jpg  |    Bin 0 -> 32980 bytes
 .../images/impact/talks/BgC79Zt2fPs-2170.jpg  |    Bin 0 -> 54045 bytes
 .../images/impact/talks/BgC79Zt2fPs-2455.jpg  |    Bin 0 -> 52448 bytes
 .../images/impact/talks/CW4Ntdtp7lg-110.jpg   |    Bin 0 -> 56320 bytes
 .../images/impact/talks/CW4Ntdtp7lg-827.jpg   |    Bin 0 -> 64634 bytes
 .../images/impact/talks/L90MBb6NLBE-1703.jpg  |    Bin 0 -> 45502 bytes
 .../images/impact/talks/QRwxHGpWaUA-205.jpg   |    Bin 0 -> 57444 bytes
 assets/images/impact/talks/README.md          |     59 +
 .../images/impact/talks/V_qzqY1bb7I-2836.jpg  |    Bin 0 -> 31216 bytes
 .../talks/clickhouse-2021-cpp-siberia.jpg     |    Bin 0 -> 70313 bytes
 .../talks/clickhouse-2022-release-22.3.jpg    |    Bin 0 -> 53560 bytes
 index.html                                    |      7 -
 requirements.txt                              |      5 +
 schemas/impact/adoption.schema.json           |     66 +
 schemas/impact/bugs.schema.json               |    204 +
 schemas/impact/common.schema.json             |    177 +
 schemas/impact/dbms.schema.json               |    114 +
 schemas/impact/needs_review.schema.json       |     48 +
 schemas/impact/paper-note.schema.json         |    216 +
 schemas/impact/papers.schema.json             |    293 +
 schemas/impact/people.schema.json             |     86 +
 schemas/impact/people_decisions.schema.json   |     36 +
 schemas/impact/policy.schema.json             |     46 +
 .../impact/recognition_highlights.schema.json |     34 +
 schemas/impact/resources.schema.json          |    121 +
 schemas/impact/stats.schema.json              |    507 +
 schemas/impact/talks.schema.json              |    380 +
 schemas/impact/techniques.schema.json         |    432 +
 schemas/papers/paper-analysis.schema.json     |    485 +
 tools/__init__.py                             |      0
 tools/impact/README.md                        |    158 +
 tools/impact/__init__.py                      |      8 +
 tools/impact/cache.py                         |    251 +
 tools/impact/classify/__init__.py             |    261 +
 tools/impact/classify/prompts.py              |    383 +
 tools/impact/collectors/__init__.py           |      0
 tools/impact/collectors/adoption.py           |    597 +
 tools/impact/collectors/artifact_links.py     |    196 +
 tools/impact/collectors/artifacts.py          |    557 +
 tools/impact/collectors/dbms_registry.py      |    329 +
 tools/impact/collectors/forks.py              |    185 +
 tools/impact/collectors/fulltext.py           |    375 +
 tools/impact/collectors/gitee_bugs.py         |    275 +
 tools/impact/collectors/github_bugs.py        |    853 +
 tools/impact/collectors/lab_members.py        |    103 +
 tools/impact/collectors/mariadb_jira.py       |    247 +
 tools/impact/collectors/nus_test.py           |    412 +
 tools/impact/collectors/papers.py             |    635 +
 tools/impact/collectors/people.py             |    354 +
 tools/impact/collectors/provider_bugs.py      |    350 +
 tools/impact/collectors/resources.py          |    693 +
 tools/impact/collectors/sqlancer_bugs.py      |    288 +
 tools/impact/collectors/talks.py              |    457 +
 tools/impact/config.py                        |     98 +
 tools/impact/dataset.py                       |    257 +
 tools/impact/dedupe.py                        |    325 +
 tools/impact/github.py                        |    181 +
 tools/impact/http.py                          |    253 +
 tools/impact/intake.py                        |    511 +
 tools/impact/notes.py                         |    369 +
 tools/impact/pages.py                         |    256 +
 tools/impact/plots.py                         |    423 +
 tools/impact/pr.py                            |    229 +
 tools/impact/reconcile_papers.py              |    162 +
 tools/impact/repos.py                         |     76 +
 tools/impact/review.py                        |    199 +
 tools/impact/run.py                           |   1081 +
 tools/impact/scholarly.py                     |    367 +
 tools/impact/seed/__init__.py                 |      0
 tools/impact/seed/paper_classifications.py    |    274 +
 tools/impact/seed/techniques_seed.py          |    465 +
 tools/impact/stats.py                         |    509 +
 tools/impact/talks.py                         |    679 +
 tools/impact/taxonomy.py                      |    295 +
 tools/impact/tests/__init__.py                |      0
 tools/impact/tests/test_data.py               |    446 +
 tools/impact/tests/test_pipeline.py           |   3037 +
 tools/impact/util.py                          |    195 +
 tools/impact/validate.py                      |    634 +
 tools/impact/worklist.py                      |    196 +
 tools/papers/__init__.py                      |      0
 tools/papers/analysis.py                      |    223 +
 tools/papers/checks.py                        |    119 +
 tools/papers/cli.py                           |     98 +
 tools/papers/collect.py                       |    135 +
 tools/papers/extract.py                       |    456 +
 tools/papers/mentions.py                      |    325 +
 tools/papers/pages.py                         |    126 +
 tools/papers/prompt.py                        |    255 +
 tools/papers/store.py                         |    144 +
 tools/papers/tests/__init__.py                |      0
 tools/papers/tests/test_analysis.py           |    142 +
 tools/papers/tests/test_extract.py            |    244 +
 tools/papers/tests/test_mentions.py           |    182 +
 1240 files changed, 347892 insertions(+), 110 deletions(-)
 create mode 100644 .cache/impact/README.md
 create mode 100644 .cache/impact/classifications/.gitkeep
 create mode 100644 .cache/impact/classifications/00/00ec0c4ab2395175ad39e0db6af27133acf0ece46c7310ab724e35aa023f3468.json
 create mode 100644 .cache/impact/classifications/02/022bc8ec1b7eaddf62d93509b5b4f2beb7ea6fdbca92818282945fd4f8592e5c.json
 create mode 100644 .cache/impact/classifications/02/0291964c195d72f8a24f4c54a20e0149e1eb1994ae783a10e6bd6b5f43dceb42.json
 create mode 100644 .cache/impact/classifications/02/02b1ecf20f9657dd9f6fba8da840822cdcd50a35e7cb8c4656d70629bd6cdcaa.json
 create mode 100644 .cache/impact/classifications/03/03b1a6cffb6b046f97f2ef36d208258d7cc148dd2ef47da3f5d0c64a0482e69c.json
 create mode 100644 .cache/impact/classifications/05/05a9632477e1270abb608e2426c322ffe2afc56f509a58453dffa2bb78c8ec72.json
 create mode 100644 .cache/impact/classifications/05/05e103a12eb6ae65021d96b2743d0a7a7d8367f6711135e566db2ca1957f7560.json
 create mode 100644 .cache/impact/classifications/05/05f9a6064d39f03460452b8b9d3e0450c637f97ea01b4705276e5d80d80d0283.json
 create mode 100644 .cache/impact/classifications/06/0620dd48ff05b205fb824aa7e1d2cfe94062778e8511f26a6cca1367ca4e7972.json
 create mode 100644 .cache/impact/classifications/06/0685de5c34d1689373f48f48823606930c426ae9a9e9a95ab82bf7fb20f44d83.json
 create mode 100644 .cache/impact/classifications/07/071db0fbf07b947c8aa8b304afac95b3b6803698deb67cffb477d348397bb473.json
 create mode 100644 .cache/impact/classifications/07/074e5bd7ab805341f895bd5d7c555d5f0cf5cf69bc5b2c02fa475e4213c6e1cd.json
 create mode 100644 .cache/impact/classifications/08/084e424f6707ac75bd465a026d302e4fd22e06dd403e1afd1b68a3fde9bd9221.json
 create mode 100644 .cache/impact/classifications/0a/0adea20f4f706ceb22e4ce210c6448c380fb849554566a3a12072fedf99555dc.json
 create mode 100644 .cache/impact/classifications/0a/0ae74c01ced1a4ef65d237d0567a399acf166fbc3a6355c2bb47996c8a8ff015.json
 create mode 100644 .cache/impact/classifications/0b/0b9a58e8ebf293ea894e2c86ace4a46053ed5ab89252753a52eb333469ee137d.json
 create mode 100644 .cache/impact/classifications/0b/0ba21c223b508f3b5fe69ec1d87be8e4ac3a2f187cad26cb1c3898e126606548.json
 create mode 100644 .cache/impact/classifications/0c/0cb54e815b343da317dba60d90d9d9239fde75092b8f59c675b7625e9531c225.json
 create mode 100644 .cache/impact/classifications/0c/0cb8307f2658bdd072298b663d7a3a47149f35a758a7fa18e3293631259e71f9.json
 create mode 100644 .cache/impact/classifications/0c/0ce5ae07a2ef1883288dab55d9f7aa7be47ab108a96072e7256be4239367d3a6.json
 create mode 100644 .cache/impact/classifications/0d/0d6c8c592b79d58edbd2a96dcab698b70e9797030d5e2d2a210ed6716f49535a.json
 create mode 100644 .cache/impact/classifications/0d/0de27f4f815dcf7cf3d5aa233e6c9b48df307b204314a2cea1028db091d76844.json
 create mode 100644 .cache/impact/classifications/0e/0e39f3309390381085eb5e988d993e1a9108bfd386b91480df29481cc06f4fef.json
 create mode 100644 .cache/impact/classifications/0e/0e4fb261364f3c9d0516d492b8f62002e85cf1a0140dd15f6d2799531c695457.json
 create mode 100644 .cache/impact/classifications/0e/0ecf0687a3b2c8434513fbe453d8e93d8e3dd1d854dba7ec10144a1bb2838b79.json
 create mode 100644 .cache/impact/classifications/0f/0f24025e5039fb3393fd19a213c25e70924ee86d7f642f1bfaf566ca7f381882.json
 create mode 100644 .cache/impact/classifications/0f/0f9cb7df12716cb64abc3a288862d58cf60a255211afa66ea2a50debf412561f.json
 create mode 100644 .cache/impact/classifications/0f/0fbbf0043d76d2a11217fbebaa579336861b31e0b40a64aa088cb90e18673a41.json
 create mode 100644 .cache/impact/classifications/0f/0ff924a32c49459c18614d8e7399c07a2ffa192bdab9bbf5e2484b7908d92dcd.json
 create mode 100644 .cache/impact/classifications/11/11a5a4f4d0adffe44e99f9a52b93846b6a4563d7ba313f5e52c4103115836fe8.json
 create mode 100644 .cache/impact/classifications/12/1235c9568b5f5257407888fcfb2acf44bbf213375d147bbbe1909b5a6bb7acdd.json
 create mode 100644 .cache/impact/classifications/13/13010fbc1de3f4360172e901910e96f38a9789e3e4c7ae7e194e376189e4d5a5.json
 create mode 100644 .cache/impact/classifications/13/137bb24122c4b373161e3490d14ea8a617b37b49f87a4f8cc3e0ac2e0829f42b.json
 create mode 100644 .cache/impact/classifications/15/150be7a1f97107c484a440ab68f518a2d6b906ec4ea1a1fe2f88ef58ac66da29.json
 create mode 100644 .cache/impact/classifications/15/155081c5f11ca6aa85ca9ea5c64bdcf9273f815f09dc3317b674c4c9ac724110.json
 create mode 100644 .cache/impact/classifications/15/15534187ea71903f63b427738a4a4ce959654a1d5902fe03a05bfeae34869bd4.json
 create mode 100644 .cache/impact/classifications/15/15e97369f43c59e5fa8f9b42ccef618cc362574ba879d0152d9e00ceb5d33558.json
 create mode 100644 .cache/impact/classifications/18/1855156389f387c02a49c2707dd852af4f62c5ff1530f48774e42d5fb86a1ef8.json
 create mode 100644 .cache/impact/classifications/18/1878770d70fa6c7dad0bc7dc46d0433e6c96b2c925958aa110091f55f8ba8f62.json
 create mode 100644 .cache/impact/classifications/19/199104e1cb80d36f2d60fe7a02bf9afb12a43e276d7707f05dcef6de4c8376b4.json
 create mode 100644 .cache/impact/classifications/19/19b024ccc309e35baaee5a2399e23dd0552b76a316c2208c5106c01d927cd7d9.json
 create mode 100644 .cache/impact/classifications/19/19e6fa7d054f751ff6e1b02d0c14b1f24359d6167304aef53511eef311ab36a0.json
 create mode 100644 .cache/impact/classifications/1a/1a070f77f658bec092ad1821749d2be521284cc0416c827763e5211895613642.json
 create mode 100644 .cache/impact/classifications/1a/1ae04d7f797a16fc64d3cbd05c5fb593034b0f38ad2f630c6e5d6c3b1e011371.json
 create mode 100644 .cache/impact/classifications/1a/1ae2b17a75fac71963baa528bec7fd22426c55a15710dbd66d41e2ee52eb8811.json
 create mode 100644 .cache/impact/classifications/1b/1b99ae9e948dd31c2f70aa43c6e316b00843229e0b170b2029180d78f1125557.json
 create mode 100644 .cache/impact/classifications/1c/1c146d4e84aeeff9dc22428f8c8cc1d51ae661f54620e57a9d51f7d538831633.json
 create mode 100644 .cache/impact/classifications/1c/1c60481568a9590f82b256626fa78b6276257f16cd884c11f2471396981e611e.json
 create mode 100644 .cache/impact/classifications/1c/1c9a85183a72cf9dd1d9275b79a94f94b37d085440267379eb4e2c9927039571.json
 create mode 100644 .cache/impact/classifications/1d/1d05667dd3209af965f1527938349b1d5a1cb85b53b1469e2f41e3e7f28cfa69.json
 create mode 100644 .cache/impact/classifications/1d/1d980a15bc8f7558ca99cf68cad25d88a9db2efc0ed2872098c3d7acdc890e9e.json
 create mode 100644 .cache/impact/classifications/1d/1da4fd2f9c9bf1209742bfc129712d9b04d79af4840adcb453ba99bbe5e60ac1.json
 create mode 100644 .cache/impact/classifications/1d/1da702858ce7541595df3a7f7d6906a96d8a818c75f7a3fd86be02c5cd1e361c.json
 create mode 100644 .cache/impact/classifications/1f/1fc9a19bcbd009be9d31e376883992955dabadcfcc18e65e4681f6d0218115a6.json
 create mode 100644 .cache/impact/classifications/20/203557b0bb1ffead7b93ac4c6d4ad5da31019f1066c0a1eb44c8e37b6e136c35.json
 create mode 100644 .cache/impact/classifications/23/23540ab9c04e9166e431788cee8c919bee81abf5bf9b2e7b2397e6da5586f339.json
 create mode 100644 .cache/impact/classifications/23/23d585228e1e9c99720346b9e6fbb99830b6b5b88418faccfb09de51e77f2cf7.json
 create mode 100644 .cache/impact/classifications/24/2493e2517e747bdda9a8ecc7ddf2ec613395774f0185215de8a602fd926a594b.json
 create mode 100644 .cache/impact/classifications/25/257046ba8fb14f1467d3dc8fbc22aad7b0b6fbccd83b4879df02f0b46fca5fc3.json
 create mode 100644 .cache/impact/classifications/25/25f4602b8edb4bae06c3f98e4af1da75dfa7d38b4f4bd41b3757320a29d7ea98.json
 create mode 100644 .cache/impact/classifications/27/27970ddc5e276c40fe4a19437f017033a5151f20ee58dac89d5ff8661542f54c.json
 create mode 100644 .cache/impact/classifications/27/27a4be2a418de4eaf5ffa36f91c0848e2324bb4262df3645002566a4c0c58a76.json
 create mode 100644 .cache/impact/classifications/27/27ba35f664c555f54309efad55bdce8af25f99a6f976a9f043946c1f6b85e2ae.json
 create mode 100644 .cache/impact/classifications/28/2819cd2a388dd7823a60d0d21e6e85b68451030ecd6a196e1ce64a6c85d98cbd.json
 create mode 100644 .cache/impact/classifications/28/28ce00af063fd29755acb934a9be9d60d0432799d296898a8f2b3de36e4c6230.json
 create mode 100644 .cache/impact/classifications/29/2962cd01151ad1522e01be3028477bf621d9f70a1c346ebd2c559d37e9d6689b.json
 create mode 100644 .cache/impact/classifications/29/29774f45cc50aa17c8de2b220c0de298f29167ff468faf09a2a12967c55413bf.json
 create mode 100644 .cache/impact/classifications/2a/2a639b6bdfb34a42843d6297cdb2e478a561e643f7e0e146e85699dfafe16b9f.json
 create mode 100644 .cache/impact/classifications/2a/2abdbc05b56246b5688d4431334a191dcb7c2e6baaa78d23a0aaa74b182681df.json
 create mode 100644 .cache/impact/classifications/2b/2bee8fa11643903d6598d869c6f2520c8e59e2ea2b06823904797bf32769d93a.json
 create mode 100644 .cache/impact/classifications/2c/2ca3ea002d0fa04127cd4365bf85cd13765352477adb7d81864d2b6b3171d657.json
 create mode 100644 .cache/impact/classifications/2c/2ce4528996df164df768e7e2eea9d69f01f222d14c53b9f4b362c9f1b59a4fa1.json
 create mode 100644 .cache/impact/classifications/2d/2d6bb0e6ad160496b9540dfa9e28225383fa43ab7737dea8f53aaf1361785b79.json
 create mode 100644 .cache/impact/classifications/2e/2e57140ec9a6b4f19a6850494ae367062f81c770d4e291235fd1aff4d64c934e.json
 create mode 100644 .cache/impact/classifications/2e/2eb99af6527366bc85c9d22f557c9ad577b99d02d54e52d308dd89f3be3acb61.json
 create mode 100644 .cache/impact/classifications/2f/2f0067431ddf49ac7dbb79bba275f55f9d689e860cb3ea937d58e659b5f4e82a.json
 create mode 100644 .cache/impact/classifications/2f/2f627059e953077e00b8e998ff16004a1abfc020020d5a2188dc4ddc9a734f19.json
 create mode 100644 .cache/impact/classifications/2f/2f696a216c22c769aa42dafd84441bfc34c2a474f23182f5c44110e4cd7eb95d.json
 create mode 100644 .cache/impact/classifications/30/3043c6120eddf0520b74edfd0a9af09c0e1a559ea3939328d6df63baeeec4cc9.json
 create mode 100644 .cache/impact/classifications/31/31769ce23be6f3a48c2912df04e9b7721a4ff519bd44f3f87aa9a39568e6c577.json
 create mode 100644 .cache/impact/classifications/32/32351ca3f564c548f77d8c3fb1c40e99ba3489cf17decf1d65164f7a6c81a90b.json
 create mode 100644 .cache/impact/classifications/32/325c43ee4d0297fc1892bbc12ff4075434acf86ef93ae8dfd013c56e0c1e6d46.json
 create mode 100644 .cache/impact/classifications/33/330331d5bb137b422485c0dcd3c58668b14904ece7180035651f5875f8beb339.json
 create mode 100644 .cache/impact/classifications/33/3336529e53c2ebc93e917cacd622dc40a006d6aac49ba49c723cd9e286078bbb.json
 create mode 100644 .cache/impact/classifications/33/334bb3a4c8077dd5796a3c9ca7a32da6854a804f39fcaebf23f18a7d56755fb8.json
 create mode 100644 .cache/impact/classifications/34/34598887436178b53a11b458c494d3497a7f0e572ec5d418852682cb3874ee60.json
 create mode 100644 .cache/impact/classifications/34/34b5174058c215c84c0f33d4975c8431a3843b099f5e9fc87b7a6e95abf81b19.json
 create mode 100644 .cache/impact/classifications/34/34bf3d7bdb37acf3e6be0cff646f63e34e4f0f43326d2f1d7d8aae77634be74c.json
 create mode 100644 .cache/impact/classifications/34/34e7c2630373aa6c27230387dc88b61f7b2f6c1348437a62b3d0ceeaa7eb5ccb.json
 create mode 100644 .cache/impact/classifications/35/350a9bba7f6cfcab64e3c6f18d353572a6edc5aea2a5326109d2e9267e882c46.json
 create mode 100644 .cache/impact/classifications/35/354b3ca7f084208b733d45120a4ee6700692d26f6e80ed11681c83af1355786f.json
 create mode 100644 .cache/impact/classifications/35/358240121ac85c1fc490a373133f4bb84e386638da8687d8a672ec040872934f.json
 create mode 100644 .cache/impact/classifications/36/3696329a0479dd1c62dbb91c31c7dc0983296a09eb47b94c778070dd4d6beb0a.json
 create mode 100644 .cache/impact/classifications/36/36c679c5cdfd4fc072f1cc0f39e60e131b3edcd61e1b315fb04f9ca7b1f83999.json
 create mode 100644 .cache/impact/classifications/37/37121406943b592e568d06b71d6e087d5904b0848e2472cfb89264fc6c6b1323.json
 create mode 100644 .cache/impact/classifications/37/374ab22691ed577bf78e8df5ed8982b044b4a62ed1f529087f888fd8099cffb7.json
 create mode 100644 .cache/impact/classifications/37/3769ee162a03a4e359e831d82e86e0b345bcb74e535a05a6ecad906fe0712247.json
 create mode 100644 .cache/impact/classifications/37/37a16251babed2e362f96803ac5f3b54a466c3533b1c0cd71af045f59129201c.json
 create mode 100644 .cache/impact/classifications/37/37e7d8733b11d19c49beba9b422a09f461dcaab7dc097878b7feee9ef20b219c.json
 create mode 100644 .cache/impact/classifications/38/38318f9ef674a3bf674b3fcc00d9be784d89919691dfcae55cdec8aeb0140de9.json
 create mode 100644 .cache/impact/classifications/38/38a33126a7014ac8deacb5324e160761e2794d024dd0580198fa8fbaedb154a9.json
 create mode 100644 .cache/impact/classifications/39/394719e5863c3928987760ce4a4b50653eb060ae545efa8804db985d35c1ddce.json
 create mode 100644 .cache/impact/classifications/39/394ed6d2db0422909f821a9ace9824a5990fd11dc11359bd6dc53f91b861359f.json
 create mode 100644 .cache/impact/classifications/39/399a6685b1439f37644978e3c87427f046aedf49e0e5361180e79842e51e76b1.json
 create mode 100644 .cache/impact/classifications/39/39c933270f2736392758208a8ba57b8ee8d71a0ef0bab104e4b6ad739187858c.json
 create mode 100644 .cache/impact/classifications/39/39e488e79ddb6819d4549d27cd33cae106b1e3e1282898ca594959a9a9581559.json
 create mode 100644 .cache/impact/classifications/3a/3ab3f1e9f6afd523df60713fadb8039eba5c773095d86e8adeb0c18383569e5b.json
 create mode 100644 .cache/impact/classifications/3b/3b0905ebb8f157755762ae4bc668a710fcce367d8c911889e1dc045d71d2c084.json
 create mode 100644 .cache/impact/classifications/3d/3d3e80e814bc3f6c8863de81cfe86a6b1276ae75b43e3b2617a9623ca765a938.json
 create mode 100644 .cache/impact/classifications/3d/3df58d9f4baf155d3760ce531aa04f199cf62840af483eab81a2ec93bbc90eeb.json
 create mode 100644 .cache/impact/classifications/3e/3e5b2501fecc172646dda868bc823efd3603da8b6d3033597cf02ec319d51ff4.json
 create mode 100644 .cache/impact/classifications/3e/3ed6d135ca67d5d9402164156312d2b32c9766927dbfacfe68e68e8200fe3fbf.json
 create mode 100644 .cache/impact/classifications/3e/3eecfca54d3cf7c233bc5f451ceed197ded3133f484f4fced09e20644a4efe8c.json
 create mode 100644 .cache/impact/classifications/3f/3f0a4beeb5c10faa6f3b446799db8c6a3333f6237c4c51bbf513a277054a8f84.json
 create mode 100644 .cache/impact/classifications/3f/3f873111b12e1bf3d360ef43c66fecd65f0f63dda21b35d5829ace408ce6d3f5.json
 create mode 100644 .cache/impact/classifications/3f/3f9eb764034e417513cb7d5afbec8a64297ed48a727ba96c27b2171c483040ad.json
 create mode 100644 .cache/impact/classifications/40/404131a968214973c44cb927ce36c02ee7c6f26c9ced31ae4d2b6531880065d3.json
 create mode 100644 .cache/impact/classifications/42/42e8227089f49fc990c670f71cfb7833f4274ee9bc98dc300969e8822bb84cfa.json
 create mode 100644 .cache/impact/classifications/43/43514fd76182f1615f73e1ab81e95aae36b40e108c2860f13839334ca2bb4df6.json
 create mode 100644 .cache/impact/classifications/43/43b89d2e810e5ebb83b49e62d80afb7191b2b9f8f9cc5372dd8ee52cf4db6149.json
 create mode 100644 .cache/impact/classifications/44/4455c264cd2c05e749bf9061611dc3ab1565301c1e694be63372626798ca0e7d.json
 create mode 100644 .cache/impact/classifications/44/44838e68a270ebd04d7547e84a8eecfba734fdd70ad880908bba0a4d4d15ee09.json
 create mode 100644 .cache/impact/classifications/44/44effe4caa65bd4046307e558957090ca44711c4b5893dd639571392639c4221.json
 create mode 100644 .cache/impact/classifications/44/44f0bd930ab5ee79571013d4f5038600b327d16c7329f7049a54b9b194b2e686.json
 create mode 100644 .cache/impact/classifications/45/45f01122322428a4a78238c95ab408e693c9b10bfc023463d64235fbf0424709.json
 create mode 100644 .cache/impact/classifications/46/4605a68f426f556226e8235beb21ade8e223fb6225b14bb5dceecfc436afa6e6.json
 create mode 100644 .cache/impact/classifications/46/46a48522814fdf7f0399d758483e770ad376d1eea4bc6c3f1cbd7d4464e234d3.json
 create mode 100644 .cache/impact/classifications/46/46f823d675773e0775d70c0f82412eb4ad818219c934e4b38b10dcc1f9a0c0a6.json
 create mode 100644 .cache/impact/classifications/48/489fa49a93ffd8ed27231945f28c16b46df14cc9c71723467e0dd5cb9e894e38.json
 create mode 100644 .cache/impact/classifications/48/48cc86691238b160f23f49cb409313b50e0ecd8661427c729dcb530b82115fbf.json
 create mode 100644 .cache/impact/classifications/49/4927362e10f288d1fe270f5b64278fd12ca7db457bb290b77b4a6ba002da4e6f.json
 create mode 100644 .cache/impact/classifications/49/4993e3bad1954605b119cab430fb1ec77c7cee0cf1826f383087f3ba64fa8753.json
 create mode 100644 .cache/impact/classifications/49/49aabb888967bf77623cbd85ec18adcacdafc9d085aef3f3c7929c63419e2d10.json
 create mode 100644 .cache/impact/classifications/4a/4a20c43b3dd2bc6c1de6010e5577cb93c540436a56948f395b6f2228a12fa431.json
 create mode 100644 .cache/impact/classifications/4a/4a74495a0016467441a2bb521ce38b46053ed970020ba8631b747d057b9fbf56.json
 create mode 100644 .cache/impact/classifications/4a/4a894b5f7279c0371db5717fb169b766bbd9935efbe0aeae7d78b2e6755958be.json
 create mode 100644 .cache/impact/classifications/4a/4a8eb1bf791dd027e366c7dbe315dacd8fa8755f80f984b5f844b6b066757b02.json
 create mode 100644 .cache/impact/classifications/4b/4b976502f3f2b561b6f6579cf3ef6ead998f58bf1c6c402b7c1ab242dd6a61cf.json
 create mode 100644 .cache/impact/classifications/4d/4d5bdb0b5e507ba677a2394c4c745142473719bab8a70165dd2b917b39a19202.json
 create mode 100644 .cache/impact/classifications/4d/4d8003520949393c66f72aae4bd7cd1c0aad1e7364588a3d215f9db45ffd14b0.json
 create mode 100644 .cache/impact/classifications/4d/4dc74403f733aa2009cbefeb5efe4e074128f7d771f626de11bbd12a31382254.json
 create mode 100644 .cache/impact/classifications/4e/4e66ba18301a20ca3943fd0f369920737cc2dbf185350db475dd5e0e426969d4.json
 create mode 100644 .cache/impact/classifications/4f/4f6e38b70f579e24aeaab3b58eef7502e0ba5f49a289e9cb8272675bc64e2532.json
 create mode 100644 .cache/impact/classifications/4f/4f9dba8edc092c030802352a1883d8305d9100295434ae9d746695d3fc5ce45e.json
 create mode 100644 .cache/impact/classifications/50/506a59fc6153b3f20265012fa5a3cfec857997bb3aa0a5db10bb5ad7c9fa8139.json
 create mode 100644 .cache/impact/classifications/50/50bc8bbca8434bbd3f96e6238329479876b7841c8b6baa47c524570279000a29.json
 create mode 100644 .cache/impact/classifications/51/512792b78e199ae00264fa027ac149ba8806109596e63d6c95de884ec1b3a44b.json
 create mode 100644 .cache/impact/classifications/52/5203719cd9a14c2c544bea8a73c30497527d2309b3b3d6600f223a0df13b88d3.json
 create mode 100644 .cache/impact/classifications/52/5244f576e9649a3bf27d4e6c6534fbdb52ff8adf8208f2c9beee873c27ec805f.json
 create mode 100644 .cache/impact/classifications/52/524ef0640358d622fc5455183db292c86b30ff0f149ef3d33cd91b758da870e3.json
 create mode 100644 .cache/impact/classifications/53/531bb84784800d0ac158f1a81dfdbc52e70b60e4de6b4f4e485a188d387ca20a.json
 create mode 100644 .cache/impact/classifications/53/53453a3c991e040f1677ee3c3c6d3e22ceb6f9544a3a38aa2abf417f40b9d029.json
 create mode 100644 .cache/impact/classifications/53/537a6e513973de4259a0dceac629aba3aa18f3a2cb54e5c288f061f31e000a5e.json
 create mode 100644 .cache/impact/classifications/53/53ed904452ad4a7f926863e8b0633293cccf99fadbd0759b4c6bdb70d087dffa.json
 create mode 100644 .cache/impact/classifications/54/545bbb2753d7fc1eedee53b8a775b5d460b52676ec568f60a1490acd2db2419c.json
 create mode 100644 .cache/impact/classifications/54/549c3dd260c607aea5cd538d511c89d82c7829182a9d19ba96d26eb50030f6e8.json
 create mode 100644 .cache/impact/classifications/55/5547facf2467a1dfbd64db232561bdfe45e61af252905f368909cbc552c4b5c4.json
 create mode 100644 .cache/impact/classifications/56/56d513e1fada73b3e7ae9de37b25b1dc0c27570c56c69b16f9818b186c92d185.json
 create mode 100644 .cache/impact/classifications/56/56fab7ff30971184a36d17ee3d426f281dca24ead77dca80e31773f9925ed368.json
 create mode 100644 .cache/impact/classifications/57/571e14f37f222c5e38a7a6cceef227bab7bc70fa3de58c9565c969b15ecfd704.json
 create mode 100644 .cache/impact/classifications/57/57d1d1457138578912e7a6979e9da67402d87384fecbd04a8bb754db1cf14c62.json
 create mode 100644 .cache/impact/classifications/58/58bb6c707dbb1faa38e74395db0e7c64b03988c34d83d451bebd4f524be9fc66.json
 create mode 100644 .cache/impact/classifications/58/58ed54dd184a8b2edfce15ecdb7ccdce8d098bf29efb58216a52ae304fa24f96.json
 create mode 100644 .cache/impact/classifications/59/5915c258b03b0d16bba87b3db3baf0f25d18905f56015d0b0f1d9fee40f75f13.json
 create mode 100644 .cache/impact/classifications/59/595510e769eafcf0b275b659f1dac7f8770500df101b1e03339f11f842783b97.json
 create mode 100644 .cache/impact/classifications/59/59e0883ad90f346f4f6df11b9a25ea657793f52cd1aa281ff5294f6941b6a7a2.json
 create mode 100644 .cache/impact/classifications/59/59e0efdead4ad79525a71d3fe73c493f6734a28f1c4226e38d16e0593b993462.json
 create mode 100644 .cache/impact/classifications/5b/5b4497c429e7478e70c7a7dbec31313325de1a8988b0c8e0d70f8da3837a8562.json
 create mode 100644 .cache/impact/classifications/5c/5cc81a4e394e3c17040dcfa7ea642809991dcf7cf40a2c0712537fa8e0546dbe.json
 create mode 100644 .cache/impact/classifications/5d/5d05eb14f6f3da47465d4c397f57605d2c7612d188fdc09eb1bb168b77ad224f.json
 create mode 100644 .cache/impact/classifications/5d/5d7bbaa5e4af57afde1ad3ad8cd3dc16d4fdc24acd94a6c60ffd8a737bbf6464.json
 create mode 100644 .cache/impact/classifications/5d/5d90b40170adfdbfaca2a6391232f53db26a768332d128f0726eb921e3804304.json
 create mode 100644 .cache/impact/classifications/5d/5d9e3c14dff91b1b3e5549a143c8d2ac5f10c5f9f14564e9529a73e42a7f34a3.json
 create mode 100644 .cache/impact/classifications/5e/5e25e4212fc626bf7415dd119b1105d320c03eb4c23e5cf696ad249ad6a2128a.json
 create mode 100644 .cache/impact/classifications/5e/5e4c5954e3b0fa38c57b4d551d274893891e1aa3f992b02d65bf2b21cca702e6.json
 create mode 100644 .cache/impact/classifications/5e/5e9348927654223e7a8731c714636e3eb50b6b219e716ff19e2ee0557dfe02ed.json
 create mode 100644 .cache/impact/classifications/5f/5fbfcff15755404b4373d7039268a62d716c1379b68ea32ba957532202e15a17.json
 create mode 100644 .cache/impact/classifications/60/6000ba53b7957f96e45b4d21488a51bb94ce0760c2a8c64d9b2c4a4025900660.json
 create mode 100644 .cache/impact/classifications/60/6072b4145c97bc2926ac50936a2af20b83209cf2b91f410cd8b581776d89ecb5.json
 create mode 100644 .cache/impact/classifications/60/60900ac72cef6bab9243ae516432d910d1c86a76d315f823df00daa1482bcfa2.json
 create mode 100644 .cache/impact/classifications/60/6095cccecd36e4a3dcfb9f7aef30883baf956a0139cbf0d08024f68f7665f2a0.json
 create mode 100644 .cache/impact/classifications/61/61bd15c6adeb577a3140b4cc68a92425a82cb37ef49688ef9a7d8cb1bbed1b80.json
 create mode 100644 .cache/impact/classifications/62/62845e48c5547fd5470856a612a3fa45e93a34d88a7f61757085d5faf857118c.json
 create mode 100644 .cache/impact/classifications/64/64177254330bad439b23b84086ae0a944da40d5b19fec334adf3852f23a40a30.json
 create mode 100644 .cache/impact/classifications/64/642045babfb46738d61659786bc0035b6c3e0b80167ba3817438688236b45f77.json
 create mode 100644 .cache/impact/classifications/64/64cf21577d6540c5454a80ea273aeea731afc4c062bde302ed7a558204b30b08.json
 create mode 100644 .cache/impact/classifications/67/6707c8a0ddbe3a376002a54285b99a41d9f0688faadb287be6d33a3e327e4db8.json
 create mode 100644 .cache/impact/classifications/67/67b463c4a08fe1ba60996404766dddca40878c70b54ea078a9967dffbcfc8658.json
 create mode 100644 .cache/impact/classifications/68/6852e1d7c3410ffb485fad53bd8fe55fee267498e747e27326881ce5bc51d21a.json
 create mode 100644 .cache/impact/classifications/69/69176cd33db131dcbcea2af095d36c9289e2a53aebaf5de0a6d8991efc49d52c.json
 create mode 100644 .cache/impact/classifications/69/6988bdca83c01ba81d72ac475bd12c3245f94c67e3f73e57fb0d69f7537e7fb0.json
 create mode 100644 .cache/impact/classifications/69/69b7cb07e87cf1e27dbb7626c0108289a955adc3363c13dfa23a74eecc12a1de.json
 create mode 100644 .cache/impact/classifications/6a/6abfde34d627ab8791eab0b6e69878ec222e624203c2a07beec07f57c73c6437.json
 create mode 100644 .cache/impact/classifications/6c/6c722d18071970e6194ac2aa75684ecc29ddd5beab318412e915f18f89d370f5.json
 create mode 100644 .cache/impact/classifications/6d/6d8597a86030d5f54e13da951ed87c00907a04b3c7e9e672024d9e7a0c76c5b1.json
 create mode 100644 .cache/impact/classifications/6e/6efa9cf70f3c884dbcd9256901aaf144e83dc7fc156b1b353c6419d5cdd9dd6f.json
 create mode 100644 .cache/impact/classifications/70/70669a1e477ab9f76e486c9924c9d91309ca6794ee08fd066ed7f1c69de35592.json
 create mode 100644 .cache/impact/classifications/70/70a8f65f348e8d6fe4c7cdf5bf8a201ae38a302e9ba778cab953442cc85b1d59.json
 create mode 100644 .cache/impact/classifications/72/7209979408b0b5fe565b8eeac6c0d419465844a91c07964b232f8f34931e115e.json
 create mode 100644 .cache/impact/classifications/73/7345a6326b04d0f0d4a8c05145696787159abeb77cfd26c5ff43b5982583b5c7.json
 create mode 100644 .cache/impact/classifications/73/735e01432f9c25713423d8f7d2d2bb852908789c81a8499eaf406680d029b77f.json
 create mode 100644 .cache/impact/classifications/74/7438ee65aad5351bd98e030d6ee054fb1fda43be48b9bc5b0769fe575e3376ea.json
 create mode 100644 .cache/impact/classifications/75/7508f0afcca40df3864c18de3deda469f3b34bc5bdceb471e333075a80310eeb.json
 create mode 100644 .cache/impact/classifications/75/750e0c2d07ecddbf3756bd9320cf5fd7b851eefce04c4d7a01e32377b9f2973d.json
 create mode 100644 .cache/impact/classifications/75/755534f2ed595dd03c863dfc1304974e1454b93933976bd181a4b53272027588.json
 create mode 100644 .cache/impact/classifications/75/75a72c5e43799268ab92490f6af0a28443642d3dc26a3f1f8027c5b344c7bea2.json
 create mode 100644 .cache/impact/classifications/76/764a6bb919a2f55c82f1a2118771bd9f15c08f9d05a6ebd520f59dab274c9da5.json
 create mode 100644 .cache/impact/classifications/77/773f5076e2392a3f601ba64e864acb050b2f2c6e8572bd314c8dc1ea6f8f9acc.json
 create mode 100644 .cache/impact/classifications/7b/7b846f2c31ff4f840d6d7d8a80eb6dc5828057e2108f1918f3e1078fddd0933b.json
 create mode 100644 .cache/impact/classifications/7b/7bf2bebfb1e0bd2b582dd144451dd56980687ff50e0a2bd1dc622fede603a2b7.json
 create mode 100644 .cache/impact/classifications/7d/7dd6bf46431916627e40258d7ab9fc21a5158241afe6d087184f9076f97233bc.json
 create mode 100644 .cache/impact/classifications/7d/7dd7dc5c4126a27c7402c93659c777726ebdbae2082ebc55ae80c9e1f0e4fae4.json
 create mode 100644 .cache/impact/classifications/7e/7e873b812d4827a72cd6ae34568928233ce19c5044f6c3df1f260c45369e3f8e.json
 create mode 100644 .cache/impact/classifications/7e/7ec3d0f838b6c8c603a651c0110eacac1e9309274de671a65553bf4eeb9c40bc.json
 create mode 100644 .cache/impact/classifications/7e/7ef333611a6dba21078b3cfe7084bea188a4f6de04864290c784177fc02202d6.json
 create mode 100644 .cache/impact/classifications/7f/7fc40a728d23d21f53df5bca854a0347b4beaf1fe3751df8315cd879ba619b6d.json
 create mode 100644 .cache/impact/classifications/7f/7fcac3447c153419b060df1eaa1074fa1f8c21aefc9ab7b013a897182a9003ed.json
 create mode 100644 .cache/impact/classifications/80/801af6ace6e7ee0ce4c53414a8925d6de29249f25b5365f29dd3ec6c1541ed68.json
 create mode 100644 .cache/impact/classifications/80/8022d5a14b465bca5b1b0310426f70750014a42250ff436ea9649b184f59823a.json
 create mode 100644 .cache/impact/classifications/80/803e0602843e8a5e76af7788a32794f748a0769b73eb8dd8a6663e2c28a77bee.json
 create mode 100644 .cache/impact/classifications/80/807d91847bedd5741399aa00b7fa1ea6477c2de0748d14c5992b1342e5ed27df.json
 create mode 100644 .cache/impact/classifications/81/81d5b3b459826435ff3774c6556abb01c62dc95348ca66c529eed100c31094e5.json
 create mode 100644 .cache/impact/classifications/82/82c12c179ffe86f4e7631b5e2582784bf3e8422f8bf6004f866fe1f31de1ee39.json
 create mode 100644 .cache/impact/classifications/84/8423ecab3a43b398e2a91a43d32fb90a98e6f4a52ca0a3d0e0a0b965eabaa8aa.json
 create mode 100644 .cache/impact/classifications/84/8434ff69970224b52705c886c2a74eb17e84ff3ac778d11f2cc6b2d18e2451a3.json
 create mode 100644 .cache/impact/classifications/85/850ed109b8ddac3b1692fc286526e6494bc9d79c93bbf7b8509ad89f1d140b3c.json
 create mode 100644 .cache/impact/classifications/85/8556425a369e9f40726b3c826d8956a3076269d2f5ae1d435ea291177b9d95a4.json
 create mode 100644 .cache/impact/classifications/86/8602b7af2e515160ea05a7b12aebe296f3ea0a789066736f46d810abb83f37e1.json
 create mode 100644 .cache/impact/classifications/86/86ead30bddb9719bf7d5d4b5ef804ba894d5e84837c731e5cba2ea31d732c4d1.json
 create mode 100644 .cache/impact/classifications/87/8726b52251c2dc28e8466c6c27ace4f7d28da3c7daaf6c69e2c038842a739c01.json
 create mode 100644 .cache/impact/classifications/87/879ba31e10cb142b745e3502965a70e5ec6dfb07090bff27b629c794f1877666.json
 create mode 100644 .cache/impact/classifications/89/891d5e8dbccf874724fc0cacad2b3174b09de62698e484236700499e128eab50.json
 create mode 100644 .cache/impact/classifications/89/8971a06533e5efa711b6fde9fdb3ffce68b27173b695ad21edf67dcff5f2a0ad.json
 create mode 100644 .cache/impact/classifications/8a/8a5dbb6202f5792b2e6b2b45d7ba4d35cbe2f42e7d1f77072c5c1f11dc318230.json
 create mode 100644 .cache/impact/classifications/8b/8b20d95616bce65309ccff2bcfae97801f823c3e2e591514ecaddf021f764274.json
 create mode 100644 .cache/impact/classifications/8b/8bcf0e49637f879d974525a587efa8223ee03cc99019ba2d4046b2b9cec32cc0.json
 create mode 100644 .cache/impact/classifications/8c/8c58f6be1b45691393e27ac8cb9efa758494e5cce9e69cc646093bd6f2d8788c.json
 create mode 100644 .cache/impact/classifications/8c/8c6aa66934923e5d761ebaed2ec32bfaaa3185895c0fb0a3b9c99d829a653221.json
 create mode 100644 .cache/impact/classifications/8d/8da1f7e2c9c7ba8902730bca6c28ab7a5c6d6505275ace026e9265d1dd8c41ff.json
 create mode 100644 .cache/impact/classifications/8d/8dbb4879c6236a8b4595c7a655f4f977133c3b608c7c4b211242a7d7e2838b08.json
 create mode 100644 .cache/impact/classifications/8f/8f40afd50f453231b27624beb8d697c73b7767b98c7e2a9392e4a54ab7b6388f.json
 create mode 100644 .cache/impact/classifications/8f/8fb2016b3f2e36ed1a57cd26db236531315ea87144e4c018032c4a696b86c642.json
 create mode 100644 .cache/impact/classifications/90/9046611b92aa5ef719ab092484baafbfa9c0fc5e1f7dc476599d2ae4bf682cc6.json
 create mode 100644 .cache/impact/classifications/92/927e53c502ebdfac4f0135d8bed711fbb79c8215d432afcacf647c44f2a002fc.json
 create mode 100644 .cache/impact/classifications/92/928f6156c971928f57c89ed50460c9d46c45749849c932ce0604c8a7332f67ab.json
 create mode 100644 .cache/impact/classifications/92/92b35786c9640695e0cd9f9ded6473c0004c87f12095a68145ed0da631c06ea7.json
 create mode 100644 .cache/impact/classifications/92/92ba6b3b4355d6e9d0c8595316564e196be6b861922f5818189d7c89e399e330.json
 create mode 100644 .cache/impact/classifications/94/941e17b9b3021ec7be4bfcef252478fec927aca3d6edbb59dd23d8d274612367.json
 create mode 100644 .cache/impact/classifications/94/94c10922dfb6c29ade09f182701f41007d48c112ac71d8f13777a0c8f0bea958.json
 create mode 100644 .cache/impact/classifications/94/94e385d7e369bad3192ac17a93c86e606297e158b3f90d6fd03b328c1708e74a.json
 create mode 100644 .cache/impact/classifications/95/9527ac9688bdb36a06ba8f91f4d65b1ebec88301f531f6d263679ed61370ce51.json
 create mode 100644 .cache/impact/classifications/95/95bea481babdfb180b195780b03d0260556bd8c825c9110f96a9a99cd6c6e8f8.json
 create mode 100644 .cache/impact/classifications/97/970cfcf4c748bb1fd6a08af1f473539bac44d3554c2fe704bcfa7e4282090456.json
 create mode 100644 .cache/impact/classifications/98/984628086c040a0af4b522cec3fe9c8ad2d5c773c5c0b1de6170e5a73ca71d21.json
 create mode 100644 .cache/impact/classifications/98/985e7804d1a19dd025c77c7cbbfc179137cd9b9c636cb5a2730158972e240ea5.json
 create mode 100644 .cache/impact/classifications/99/99325d7c319ae2c84e728030d8eae20bf616cd1908f0757155f9ceb4c6870ef3.json
 create mode 100644 .cache/impact/classifications/9a/9a755a3f2a71926503ed59d1c1e1c5272bfd3ae4751bf1745eb1625ba9560c79.json
 create mode 100644 .cache/impact/classifications/9b/9b84a719086edae541b45ebd1f8c9b8338ccdbde6e1f982c18fa711b90b5a5cf.json
 create mode 100644 .cache/impact/classifications/9d/9d0f3f911535c1e1e950535292ee5305c63485ab7fe28092b67d8af039c2ba81.json
 create mode 100644 .cache/impact/classifications/9d/9d24b38f02219420d59090a3af76a95e82f357018871daa70c31cd3c8cf7c484.json
 create mode 100644 .cache/impact/classifications/9d/9d923a458fce3bc9ab455c772d94b470a09b882fa0d8139591c5b4aa345d20a8.json
 create mode 100644 .cache/impact/classifications/9d/9da94908d3a84f0a3551c97bd5d96c3f717d8c845f9c8d28570001b48fae97b3.json
 create mode 100644 .cache/impact/classifications/9d/9dc659ab8ecaf1af83e3c44617a4f7713a66a7ea4ecdcbd4ca9a49bf1735693f.json
 create mode 100644 .cache/impact/classifications/9d/9dd4b6dfc1ae8081c0a8a3b2881d0bea5f83c442bccc5b2e792fe45f190ebce3.json
 create mode 100644 .cache/impact/classifications/9e/9e917e40c578750b344ff338d19c40e7ea6bee39b6d49888e604414710af5fd6.json
 create mode 100644 .cache/impact/classifications/a0/a059465b39743949e73d824476f690827b0985350fdd199e762361c198e8b63d.json
 create mode 100644 .cache/impact/classifications/a0/a05a715b3f6987098dcd7c5b26cb883e7cf402b6ae7fc5fc1f5d9b708530a260.json
 create mode 100644 .cache/impact/classifications/a0/a05ea93b289167fbf58be981ee436fb5b15898a696974bab1c7c603c39c73bac.json
 create mode 100644 .cache/impact/classifications/a0/a07a41d826a7ab3ca525d75cff511caaab48472e0902bde71e0e0c73d31216c6.json
 create mode 100644 .cache/impact/classifications/a2/a22ce4f50c68ebc89e701c30ae8f341b013bb7fbfeb3fe3d893434e04fd9cb2f.json
 create mode 100644 .cache/impact/classifications/a3/a350463d5bd989ebea0853a4d00d856f7310e97764a4d1e4f58eaaa7b5e984ed.json
 create mode 100644 .cache/impact/classifications/a4/a4376eecd04b2ec545fff70a705be13af50cc6b8f08987803f680d4231815739.json
 create mode 100644 .cache/impact/classifications/a5/a5405c511bd7bea95773b1c4d8dcc736ec2fea497f475bacbad2cae0520f1999.json
 create mode 100644 .cache/impact/classifications/a5/a542536234c761a13dec4e4b48cc3f9fcbfabfc4fce34ae41500dc506213c13c.json
 create mode 100644 .cache/impact/classifications/a5/a581f888af0f3116194eb1ea131aa608c500869dd063f929f098a014dcbe3391.json
 create mode 100644 .cache/impact/classifications/a5/a5952be05636f58793ff5c9d81e4355d7046ca30bafc2470345a440f2fad54a2.json
 create mode 100644 .cache/impact/classifications/a5/a5a68b3a609329213704fd68d94b95704632109f0da24ff21b5d3ff378493fb5.json
 create mode 100644 .cache/impact/classifications/a6/a6aa13ea8fc1822d7d38cea4e0c1cb92d955e98dd57e2a8f4ba8f385e3e0859e.json
 create mode 100644 .cache/impact/classifications/a6/a6e79380197744a9dd8c8c53cc82e8af354e83dbf7865b135de8f0deb414d75a.json
 create mode 100644 .cache/impact/classifications/a7/a79d79ac05a180c8d35d742996f18ec068e225e87ab2db3cd2f877953174ee3c.json
 create mode 100644 .cache/impact/classifications/a7/a7dce9c9c56d61fe2b2387b7d6924c9cab3c5755a0bb181d892114926fa07726.json
 create mode 100644 .cache/impact/classifications/a9/a9a7fc18600cbb8c66eab007112d20582b6d70c743dd2e020644768ab7c2696e.json
 create mode 100644 .cache/impact/classifications/aa/aa16bd441a8201b60e34e02f80606268fdf6e738498b2fe700a78fff4343bf3d.json
 create mode 100644 .cache/impact/classifications/aa/aa912c95033923570ab9eec850497812d17e662b1c05958526e4e45c1c6d14d3.json
 create mode 100644 .cache/impact/classifications/aa/aa94497921750fd2a1e03ee3c80057d87a3b969e3373d3e69b79db8ad2dd5d81.json
 create mode 100644 .cache/impact/classifications/ab/ab6839b50d6bc4d1b09c211e440fe63b66e901d8d7e62a849a4160b7d4c33c1c.json
 create mode 100644 .cache/impact/classifications/ac/ac36cabb569ba5302bf1cc025762a34a405abdae4bb9507487627a4265c36005.json
 create mode 100644 .cache/impact/classifications/ac/ac828d5bf10cbde7137174b35b1e30c7558430fa4c0813fc4cbbcdb1ac2effcd.json
 create mode 100644 .cache/impact/classifications/ac/ac9d4b9049e7857a6c28cd9dfe252beeb2c98d708e52566f4bb244ae0be22edc.json
 create mode 100644 .cache/impact/classifications/ad/ad6dfdf77ecd6f7e6648dca684da002b8055316c9ea976467e9a09705155f4e1.json
 create mode 100644 .cache/impact/classifications/ae/ae03b0a32bdb51cddd4f3c4069a565a89122634f0b7d4a2527ef8fba5ab7c90b.json
 create mode 100644 .cache/impact/classifications/ae/ae832e4dd0fb83346f546c3e3be4b019a8d8683ba78c77d8df612ebf2c030cf8.json
 create mode 100644 .cache/impact/classifications/af/afd4f2c9a74a0f55ea5146511651cebccf92877e5e9d53952d18265e80bfec02.json
 create mode 100644 .cache/impact/classifications/b0/b04d5d59ae6ddcb63615a042de1e4add44b342faf4e52fd4e722c4cd142df8e7.json
 create mode 100644 .cache/impact/classifications/b0/b0610a21773608a068a66a596bda96bf8ef5652372a97871dbcec7e1338b600c.json
 create mode 100644 .cache/impact/classifications/b2/b2e28297138158d035825af04b28bad99352053a321f7690d361948426866854.json
 create mode 100644 .cache/impact/classifications/b5/b5b7e3ede036f51f10a1ef00145a97c2bde73046d214035f29a3726d3e36b803.json
 create mode 100644 .cache/impact/classifications/b6/b645377f2d9ca00415336c401b89c53f119b8a99e1d6e53a817abd053a8e7e71.json
 create mode 100644 .cache/impact/classifications/b6/b672875a8140d583a6991ecbcb24675384458cd387913cd97e101615834dc097.json
 create mode 100644 .cache/impact/classifications/b8/b83ed71d58777c5bca1b11066eef37a97b1550327cedfb2caa86815d54383ca9.json
 create mode 100644 .cache/impact/classifications/b8/b89ee278e21530bb77a0a7fbbc8d52462295333353ca3f266f7b01a8cce325a7.json
 create mode 100644 .cache/impact/classifications/b8/b8ac1af954ede3f7bd2fcb4546e4584990c2429ce754b76041a6ad26e167d5b3.json
 create mode 100644 .cache/impact/classifications/b9/b9d8fc07eb2bc70541bc6df953dc66575ee6bfaf323400e7917204d1a9657751.json
 create mode 100644 .cache/impact/classifications/b9/b9f4a838de674b2759e19fbe3e0b10cf3b87322ad41d85a4d65561a6d184b18e.json
 create mode 100644 .cache/impact/classifications/ba/ba67374c42b448284b2bb7dc0c7bd52e17c0fdcf4082d5bbe6461ac052c673cb.json
 create mode 100644 .cache/impact/classifications/bb/bb000621030bcc35f7781f2eb1fb788404159aaf8a11adacee8629f75a37fcc5.json
 create mode 100644 .cache/impact/classifications/bb/bb3c3af68b783e63129ab8b0fb54c38703bb30d20843bf5da8c52593d505a1fa.json
 create mode 100644 .cache/impact/classifications/bb/bb617fc7a5dc2dd2652e37ead0f86247cd866a3edb9e4d2b9ec2e60f82bfa6a0.json
 create mode 100644 .cache/impact/classifications/bb/bbfb68810178e27f1c9ca0f6271d050c205a8281946311797e3453ccad96d3d0.json
 create mode 100644 .cache/impact/classifications/bc/bc02d3c99dc7962635bdebcb6f54bff0eb95a167f6e71e9dee6846887f585630.json
 create mode 100644 .cache/impact/classifications/bc/bc088255649370b98df47258da6d64ebbc641a179d20174d2a1fe985962497eb.json
 create mode 100644 .cache/impact/classifications/bd/bda0a53a913725806e7d985108a5ec1394a7d28581426c47319fd6229fbccff6.json
 create mode 100644 .cache/impact/classifications/bd/bdaa11108172265ef5adc5437548d19eb5c240f63e70edec38a2914ff65fe877.json
 create mode 100644 .cache/impact/classifications/be/be130f009ed7bfcca301b063e97c36a988a4a793c742dc76f07d2f2b33ccc319.json
 create mode 100644 .cache/impact/classifications/bf/bf0029fedbee143422dd50c9a8e4fe8d8279d65f99ed2cbbdefa1186927adf65.json
 create mode 100644 .cache/impact/classifications/bf/bfb510deda29952011218e4291d91ee10374ca37b1ebd1a82936d5261f558b10.json
 create mode 100644 .cache/impact/classifications/bf/bfb8049a4c90182e8eb77f23e7241e48fd5dac356321481a85cc2c33177d2d28.json
 create mode 100644 .cache/impact/classifications/c0/c03a3f1bff883fd7e2685bfec74f5f2863af23690e49aa0efb6eeef391f1f047.json
 create mode 100644 .cache/impact/classifications/c0/c0883bd5aab77f7389e96a9b353aab19142af3ffc94ad555f0ee6dab20f9215b.json
 create mode 100644 .cache/impact/classifications/c1/c108b4cc78a8a42c1fd038dc1e279cfb23563f3e34750e00afb4ab25fd1f5f23.json
 create mode 100644 .cache/impact/classifications/c1/c173ad3d39ee4c87fd437bcb0cb9b5982c1e6d86a00241d1ddd31712e998ab66.json
 create mode 100644 .cache/impact/classifications/c3/c33948e31ab7ce4429e6f133a0c806f471a0ab46dc5246cf88d128d489f3a485.json
 create mode 100644 .cache/impact/classifications/c3/c3403d87237f0c7a23f0793304e663610f8cebafdd537d2135a1648c4e4eb69c.json
 create mode 100644 .cache/impact/classifications/c3/c37adadab0f8a5d3a6709e332399ed95ad06d9ce3b84b0c0ac9ecb71176b48cd.json
 create mode 100644 .cache/impact/classifications/c5/c588f4bc991d287a0c47ff2690e5de90c16942b95e6d5f0122f01a4c9ed38cc6.json
 create mode 100644 .cache/impact/classifications/c6/c6559584db371a0540f93a35eb4a0c8508c59640fb4cd5d5075541f9c74dad23.json
 create mode 100644 .cache/impact/classifications/c7/c7a1a9db2e171934f5ffafe80196233eb9fab69d7d7444bcc38193acd373522f.json
 create mode 100644 .cache/impact/classifications/c7/c7b7db7a5d5109f4b9c5ba5561640743083a487b559d832bc11a6cdbb84cef91.json
 create mode 100644 .cache/impact/classifications/c7/c7c43bf0ce69038445569d0d97a8ee565932e41bba450728de8d5e1c569a6dff.json
 create mode 100644 .cache/impact/classifications/c9/c9130f3161bfeaa25e9a2a7373c7ee67bc40ac5cade27833e506996274dc31d4.json
 create mode 100644 .cache/impact/classifications/ca/ca697598155693e067dcfe46b08f583f9001b5eea5c7cc359697dedd35517d8a.json
 create mode 100644 .cache/impact/classifications/ca/cae41e660c13a64c3806b356a3aa88b34bd2bd9138b5b1ca37a561ec655f4d5f.json
 create mode 100644 .cache/impact/classifications/cc/cc6426dd2a34ae4bdd4916921864af3af727a0419a09c2ad4bbd347927e4bb20.json
 create mode 100644 .cache/impact/classifications/cc/cc9934db1d426ed7461fed921f36e8872b109003f798a5d163418b5a6e41da9e.json
 create mode 100644 .cache/impact/classifications/cc/ccccd056aec9246c9e0202ae837b8144a122607f408fb3c9f1baa3f19735f1c3.json
 create mode 100644 .cache/impact/classifications/cd/cd2da50a4fae7d02f20689f37fd4cfdb28929a236c54ff451c8178e88bfa0676.json
 create mode 100644 .cache/impact/classifications/cd/cdef68024f3eb8bbf93d520e5b9d74429f0f3fda89cb43274531edaeedb3bc7f.json
 create mode 100644 .cache/impact/classifications/ce/ce6bd7f1beecb5e25643e2c3ccb5361dda970801bcc8e0fae74b5cfa7902b850.json
 create mode 100644 .cache/impact/classifications/ce/cef34b69f413faed5b7519ce0c2ed1e255bc9ab436a3025627fd8c6ea87c1c23.json
 create mode 100644 .cache/impact/classifications/cf/cf3aa11bf98d9d82beda3a0b12e532e78b023b25f8a44e677aa32bb8b56e11c1.json
 create mode 100644 .cache/impact/classifications/cf/cf4a905cd5ee191f08dbb3dbee3973324f1059d8062ba2ca3a59d6d352f2be91.json
 create mode 100644 .cache/impact/classifications/d0/d06c32e4cf477cb8aef58e54ecc97c884f2c4b30b26a7b757bf1da42b7d2b699.json
 create mode 100644 .cache/impact/classifications/d1/d106bb699364d21a6877a0f2510db29449e0e574eff01c898316e603775c78b2.json
 create mode 100644 .cache/impact/classifications/d1/d123032068159d8880fe350e3d0ef9cc756019b67bfc99afa1c80b56c29698fd.json
 create mode 100644 .cache/impact/classifications/d2/d20b6b77f9c724c7e73c8b0609001ed161a1e3a0c0fbb398ec3c063ee77f95b6.json
 create mode 100644 .cache/impact/classifications/d2/d2a8804084aa6d2c08073f0cefcef526487cce5cb84834e80e42c3f9fa47098d.json
 create mode 100644 .cache/impact/classifications/d3/d3ba866388bb236b66c919c67eba114e392bebf05058ba91f7153943d9866164.json
 create mode 100644 .cache/impact/classifications/d3/d3c048bead4231b1587c5d75a12778d4111bd75aa96b7dcfb2ca72adbb14c452.json
 create mode 100644 .cache/impact/classifications/d3/d3fa4ab9c49204b340755bf72192770b5f71d7d79b8487ba4db11e0e01482675.json
 create mode 100644 .cache/impact/classifications/d4/d42dfbcc23013ad16e0c7c36c6ff61d267e424094d2d6fa73780cd6c5c45277c.json
 create mode 100644 .cache/impact/classifications/d5/d59aeac589a94618b84433a5210633702040e256ecd9cb4383d031a95ce8bc8e.json
 create mode 100644 .cache/impact/classifications/d5/d5bba29512401e0e97d95943d9405ec9ce4d8863198cae9ffdc869fb91eb89d7.json
 create mode 100644 .cache/impact/classifications/d6/d65929eff883f2119328a6a4aeedeada8719312d00047dd1f445efd3a12499e6.json
 create mode 100644 .cache/impact/classifications/d6/d69854f8f7db81232391c60d188f87a11c9b7fdae6e306f9cc47d05afb4cfb64.json
 create mode 100644 .cache/impact/classifications/d9/d97725254e6beb2946213a0262c4d2cf7d7373ebabd55db0b517b40e136e8ea0.json
 create mode 100644 .cache/impact/classifications/d9/d9be9dfe5c87f194fe74e154bfacf361438a6a166a9ed1d2a2f196282119ba8d.json
 create mode 100644 .cache/impact/classifications/da/da7798d2b188ccd04d00be45d121d01d6e3c4a364eec5311bd087127136ec108.json
 create mode 100644 .cache/impact/classifications/da/da99420b6151f221f441349955eb5eecc1e7b7c091c962354cd0173efcab7d2b.json
 create mode 100644 .cache/impact/classifications/dd/dd5976acf3475f78ddf9f6a4debf562438caa6715e83d967646e029ff1273898.json
 create mode 100644 .cache/impact/classifications/dd/ddbf6171a8861a1b988067807215ae054fffa3d58f6d3e8fc3f9dcce3b3012d8.json
 create mode 100644 .cache/impact/classifications/de/de61eba54f28ebcb15e3611b2d9421be4c29e7c3e5392e0e00910a4a86d342c9.json
 create mode 100644 .cache/impact/classifications/de/de7545d72dff7f63e38aa53ac4d2ec85ca3ac3f3eb4747d3138b79e725157cfa.json
 create mode 100644 .cache/impact/classifications/de/dedb1a67633e3befb399b3114fc851245ce3afd330483f4051c6042610298f41.json
 create mode 100644 .cache/impact/classifications/df/df17dba18cad9097af12174fc12c787e59d3152d6cad46ca6a70e22815d06a0f.json
 create mode 100644 .cache/impact/classifications/df/df64860d29057569e21a94cd7d3f724f5f2554bc01414f19b8990334b24a5ff9.json
 create mode 100644 .cache/impact/classifications/e0/e065dc4af509129ca6d66a7bac15f3a531ca80dac0176b2fcc6c864176128f69.json
 create mode 100644 .cache/impact/classifications/e0/e06b4a8861ad41bf641738889457a7ee12e883355142d0dfba9e73cf49e1ad53.json
 create mode 100644 .cache/impact/classifications/e0/e0812d7b44f3ef4c1042388713b90f85f8e893720fb47b42d5cde712e3964025.json
 create mode 100644 .cache/impact/classifications/e0/e08c7129e5147fe3cd78b129091942e9c603b3acc94eee4d3978849c6a550268.json
 create mode 100644 .cache/impact/classifications/e0/e0c83f7c5c4a15a3f5ce0154b0b6eb8a1c1db7e0edc15c0943e66ebbf602705b.json
 create mode 100644 .cache/impact/classifications/e0/e0fb32134077747bf777e72cd6cb950d503814f9aed886d66bfa88a9e7e0cb9d.json
 create mode 100644 .cache/impact/classifications/e1/e1eb06e050baa67037a91d3ee4bf4dbb4cab5925cd7af52f5c148eb1f459cdfd.json
 create mode 100644 .cache/impact/classifications/e2/e22f22f74ff7ffe80d062e4d31416dc5539e8a59dc1b12f107032cdc53492561.json
 create mode 100644 .cache/impact/classifications/e3/e3266a7ef15df348bf653399d3a1382721f53278a7c285b051242ab46e3b6738.json
 create mode 100644 .cache/impact/classifications/e3/e327f1ae926f9715f843b350cc485083468e34163df1b7b19e861fea896d79ed.json
 create mode 100644 .cache/impact/classifications/e3/e37dc7a5c72a2548d0860103a6f13e12f1bfccb485ef6f4db2f644be175d6606.json
 create mode 100644 .cache/impact/classifications/e3/e38e2d6e0501c9fe42b03bb3cb0e3cf4a09739a5829ee1c8d0b7bb59c056d756.json
 create mode 100644 .cache/impact/classifications/e4/e4280b6577b15b1e1e72ab9f2a9811b0429e8873beab9e2c568ec0051691d2aa.json
 create mode 100644 .cache/impact/classifications/e4/e4d8a7b759a9b17dc19be4f69afc8ed9edfc49d8557b16b07d954d7a21082df4.json
 create mode 100644 .cache/impact/classifications/e6/e6066aa9ae22acc719577e983dd1a6691ddb21e1aaad5acdd4dbc286f872f2f4.json
 create mode 100644 .cache/impact/classifications/e7/e75f46c32e9cae496c3bd58de35158465987d9d395af11cd640be8fe42c42734.json
 create mode 100644 .cache/impact/classifications/e7/e79e02e1aedb3b0a5c88c0527b42ead092999dc5c1ac8b61ab0e41fc60dff552.json
 create mode 100644 .cache/impact/classifications/e7/e7dabe197493a78ea9b140485b99bf6199e451f19300d9e231a2feff1f8a3af8.json
 create mode 100644 .cache/impact/classifications/e8/e82b0b636ffe6a1ad71637d690835ed5d709abad7af5c5956e5483215b33eb89.json
 create mode 100644 .cache/impact/classifications/e8/e86dedc3b4cdcb26e31f7ecced2c6f88254480a6e8fc723cb0761429e0e82f17.json
 create mode 100644 .cache/impact/classifications/e9/e98fb6ab65b64999ef96a241989fadcb6e7f579235727cc074594d8598b8ee7d.json
 create mode 100644 .cache/impact/classifications/e9/e9de857cb133843e69b8741ffe6988f56777c79fe8f7e0b560f7a2dfd59599a1.json
 create mode 100644 .cache/impact/classifications/e9/e9ef03c53497bcce87d3ea5a8b2b0642defcd119922d73681a73a3e12169d280.json
 create mode 100644 .cache/impact/classifications/ea/ea4f91f6f6c0cf235d9d7d79446c928a9187ec0b14439604f766ffa2b504453b.json
 create mode 100644 .cache/impact/classifications/ea/ea9b3e60fb14b0c66cbaa8d53a420fd0f39ae70a52ada3fe312a082867b1eff4.json
 create mode 100644 .cache/impact/classifications/ea/eadabe417ccb09f7a658e8060def987c7feee54afafa416c7e522171a901cc39.json
 create mode 100644 .cache/impact/classifications/ea/eae5956e7387d12b2c1768a1875c7fdc856eb4789e6425bb5fc1117ec58d0301.json
 create mode 100644 .cache/impact/classifications/eb/ebb001b3d79269e8a2c9a77033344cb17ee5e255d790f3e94111b34cd07e19cc.json
 create mode 100644 .cache/impact/classifications/eb/ebeef0e27eea46336aff68913407a944dd6fd3a3f6adc432b058abc4197b8123.json
 create mode 100644 .cache/impact/classifications/ec/ecff98d9d9b0b92d240e96ffe62e6bacb2e626c3291e663473ad9007eed18b4b.json
 create mode 100644 .cache/impact/classifications/ed/ed4b6de4191e743c2963902dc9fb7f7d4a560cef3ea4893f20463251b09af809.json
 create mode 100644 .cache/impact/classifications/ed/edd13a2f0e7fc99d8c8fd6fab8e036babe276c06460f0b547b2b586dfba65d21.json
 create mode 100644 .cache/impact/classifications/f0/f03e0224afdf007e3ccded8cae4346571182571d475775c3102204a41fe8fb52.json
 create mode 100644 .cache/impact/classifications/f0/f0908cf7625cdf49be7ec344fc784c4010eac45911e85654cff110f7bdadc0cd.json
 create mode 100644 .cache/impact/classifications/f2/f20cd8c531d9c2bb4cf49f95c0a38df8f780137fc0f65fab754162354f82a485.json
 create mode 100644 .cache/impact/classifications/f2/f2c297dcabfc92088e81dcc6d89f3064e55b90ae349522495d673f8bc25e1e16.json
 create mode 100644 .cache/impact/classifications/f2/f2f7ab462bb707be89f85d060601aef0c9fcdc4200c2bc7b925246c8004e4918.json
 create mode 100644 .cache/impact/classifications/f4/f45addada480068179c1a041dfa4a685f4c36ea97ead3c529fac595dd1746abc.json
 create mode 100644 .cache/impact/classifications/f4/f492c18fbb3982068b8845c0c8fef9d0f05c63248255d912584e680a8fffa206.json
 create mode 100644 .cache/impact/classifications/f5/f5397432722e3fbb02a8e0d78c25424dc97d8e7be9368cbd0465bf14194a86c9.json
 create mode 100644 .cache/impact/classifications/f5/f5ae654869ea91a7ca772b15323597afc58229295f8a2662614f7ea81ab89663.json
 create mode 100644 .cache/impact/classifications/f5/f5bc53a07c83087327df462738c3054b22b99ea3a337f2fa4ac7cd9a12aed487.json
 create mode 100644 .cache/impact/classifications/f6/f6558e93efc0f9ecf983d66acfe7e9a71f42449c0778a062bb527337efe73b88.json
 create mode 100644 .cache/impact/classifications/f7/f73a06ce379860c348599a463e340bd35f3021a3e9c73a803c9aef52bdf2939d.json
 create mode 100644 .cache/impact/classifications/f8/f83edbf83ee91b60290f80b535d9f4a9f0fa467c3bd1602f1f2f38dc8f0b1a9a.json
 create mode 100644 .cache/impact/classifications/f8/f8a8008932d19705f5a06574ed730b8e18cfe811cb1fd705ae73002402c715a2.json
 create mode 100644 .cache/impact/classifications/f8/f8e0170a23002eeaad3ccd7ae365671678bfd4307407876fe15860eb90934d7c.json
 create mode 100644 .cache/impact/classifications/f9/f91d20d7246b0ce09f79b4b2252f57baac3682fb852067d58e5502a6e3c73ba0.json
 create mode 100644 .cache/impact/classifications/f9/f94db7eebdeeffdf4cdfc924066fc711118fb797ef6d5c7d45f6901927836478.json
 create mode 100644 .cache/impact/classifications/f9/f978057b7d5525c04e0f35af0ef3a5a68dffe69fd7aec13fe037e43ece9aff04.json
 create mode 100644 .cache/impact/classifications/f9/f9c20c80590bb65f3a1700deca342498e8a2aeb62c50baa19120568c8fe10aa1.json
 create mode 100644 .cache/impact/classifications/f9/f9d93ad3a1e28559327a9119fd7b5f63a583a4f1ffff1978ccc2942ad590573b.json
 create mode 100644 .cache/impact/classifications/fa/fa22cbfc16a1428c54a15873d54f845d79887d8a2b190792aca06c3481ad5ecc.json
 create mode 100644 .cache/impact/classifications/fa/fa97191695ae316d864e0baefcbff533f63f4ce4c5ebc3d9c66b489cb732e4b7.json
 create mode 100644 .cache/impact/classifications/fa/fab39f76d5ed00d192fa7242a0b965ae488ee100de93b87a9594bfb10deb32a3.json
 create mode 100644 .cache/impact/classifications/fa/faf134b33f05237010572e05f76fc7731acf0e31cabf9d28659c49f5095a1db1.json
 create mode 100644 .cache/impact/classifications/fb/fbe0e665cd80a3f8572fe52bd9e724852ec7c2138377b6bfb48c0c92c2ff9cef.json
 create mode 100644 .cache/impact/classifications/fc/fc584d49f074025891e74e86a2b090fc9d6bfa6e193fdea696075aaf7640dd4a.json
 create mode 100644 .cache/impact/classifications/fc/fc981ee0e706b1c037c2a60937283a16743a094508aabc75cdcf6173fd7fdf16.json
 create mode 100644 .cache/impact/classifications/fd/fd3366f350ca660cfd702c3636f2079584c3df0f60d6a003b4e00fc5e7e0a995.json
 create mode 100644 .cache/impact/classifications/ff/ff0a85ba8fa6196dfc51b40b4af756940852239107113528fe20df7a2d4dadbe.json
 create mode 100644 .cache/impact/classifications/ff/fff00b9d80ac558922eb80c77532bc38b5e14c3b7944b1c39a03f4955ce41dcf.json
 create mode 100644 .cache/impact/derived/.gitkeep
 create mode 100644 .cache/impact/papers-pdf/README.md
 create mode 100644 .cache/impact/papers-pdf/WANTED.md
 create mode 100644 .cache/impact/state.json
 create mode 100644 .cache/impact/talk-transcripts/6YGqFRTe2D0.json
 create mode 100644 .cache/impact/talk-transcripts/BgC79Zt2fPs.json
 create mode 100644 .cache/impact/talk-transcripts/CW4Ntdtp7lg.json
 create mode 100644 .cache/impact/talk-transcripts/L90MBb6NLBE.json
 create mode 100644 .cache/impact/talk-transcripts/QRwxHGpWaUA.json
 create mode 100644 .cache/impact/talk-transcripts/README.md
 create mode 100644 .cache/impact/talk-transcripts/V_qzqY1bb7I.json
 create mode 100644 .cache/impact/talk-transcripts/wHo-VtzTHx0.json
 create mode 100644 .github/workflows/ci.yml
 create mode 100644 .github/workflows/impact.yml
 create mode 100644 Makefile
 create mode 100644 _data/impact/adoption.json
 create mode 100644 _data/impact/bugs.json
 create mode 100644 _data/impact/dbms.json
 create mode 100644 _data/impact/needs_review.json
 create mode 100644 _data/impact/paper_decisions.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2105_10016.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2206_08530.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2304_10044.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2310_06433.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2503_03893.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2503_17322.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2506_02617.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2509_10819.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2510_06663.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2511_17377.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2601_15074.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2602_19490.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2603_00311.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2603_19434.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2603_21530.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2604_01442.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2604_03024.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2605_20473.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2605_22992.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2606_11132.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2606_14164.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2607_03741.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2607_09072.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2607_13276.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2608_15709.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2608_23402.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2608_25573.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2608_30385.json
 create mode 100644 _data/impact/paper_notes/paper_arxiv_2609_00381.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1002_9781119880929_ch13.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1007_978_3_030_71058_3_5.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1007_978_3_030_88494_9_12.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1007_978_3_031_51479_1_17.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1007_978_3_031_94706_3_7.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1007_978_981_95_3182_0_3.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1007_978_981_95_4721_0_7.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1007_978_981_96_4506_0_18.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1007_978_981_96_6465_8_28.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1007_s10664_025_10662_w.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1016_j_cose_2025_104564.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1016_j_displa_2024_102854.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_ase56229_2023_00106.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00095.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00151.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00322.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_compsac57700_2023_00273.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_compsac61105_2024_00141.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_dsc55868_2022_00057.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_girst67753_2025_11382165.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_iaecst64597_2024_11117732.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_iaecst68792_2025_11415166.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icccs65393_2025_11069832.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_iccste65902_2025_11138310.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icde55515_2023_00057.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icde60146_2024_00011.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icde60146_2024_00441.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00180.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00224.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00240.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icicnis64247_2024_10823213.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icpc66645_2025_00021.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse43902_2021_00137.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00024.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00101.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00173.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00175.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00003.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00013.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00045.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00183.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00257.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse_companion58688_2023_00041.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icse_seip52600_2021_00042.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icsip61881_2024_10671554.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icsme64153_2025_00030.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_icst60714_2024_00012.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_ictai62512_2024_00085.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_iemcon67450_2025_11381190.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_issrew55968_2022_00056.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_missf68264_2026_11521893.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_punecon67554_2025_11378586.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_saner60148_2024_00096.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_sp54263_2024_00109.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_srds69199_2025_00038.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_tdsc_2024_3521591.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_tkde_2026_3656491.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_tse_2025_3574328.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1109_tse_2025_3625300.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1117_12_3006402.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1142_s021819402150039x.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3372297_3417260.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3395032_3395322.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3428261.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3448016_3457559.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3468264_3468540.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3468264_3468573.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3471485_3471491.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3485529.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3510003_3510093.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3510457_3513034.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3531348_3532176.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3533767_3534364.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3533767_3534409.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3551349_3556924.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3551349_3560431.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3552326_3587448.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3582016_3582053.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3588909.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3597503_3608133.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3597503_3639112.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3597503_3639200.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3597503_3639207.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3597503_3639210.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3597503_3639212.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3597926_3598046.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3597926_3598052.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3597926_3598068.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3597926_3598130.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3605157_3605177.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3611643_3613893.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3611643_3616286.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3622819.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3643779.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3643781.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3650212_3680311.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3650212_3680317.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3650212_3680318.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3650212_3680392.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3689031_3696064.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3689491_3691821.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3689757.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3690631.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3704870.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3708533.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3720504.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3728908.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3728953.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3728965.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3728973.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3729175.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3729319.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3731569_3764841.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3744916_3773102.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3749186.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3757347_3759132.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3758316_3763249.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3764583.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3769779.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3769828.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3769832.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3779212_3790244.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3785021_3787993.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3786673.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3786699.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3798226.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3798232.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3798245.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3799227.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3802034.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3802053.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3802061.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3803437_3806090.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3808109.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3810991_3811632.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3810991_3811634.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3810991_3811637.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1145_3828685.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_1177_0926227x251370258.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14711_thesis_991012980220103412.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14722_ndss_2025_230530.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14722_ndss_2026_240198.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3494124_3494139.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3611540_3611584.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3636218_3636236.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3659437_3659445.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3712221_3712247.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3725688_3725698.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3725688_3725713.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3734839_3734861.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3742728_3742747.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3749646_3749661.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3749646_3749683.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3797919_3797928.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3819518_3819547.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3828612_3828639.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_14778_3836663_3836700.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_22399_ijcesen_5462.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_32672_jnkti_v6i1_5830.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_3390_app13042519.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_3390_electronics14193910.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_35335_jurnalmantik_vol5_2021_1448_pp1065_1071.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_3929_ethz_b_000507577.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_53799_zb6b1375.json
 create mode 100644 _data/impact/paper_notes/paper_doi_10_7717_peerj_cs_1592.json
 create mode 100644 _data/impact/paper_notes/paper_s2_07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85.json
 create mode 100644 _data/impact/paper_notes/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json
 create mode 100644 _data/impact/paper_notes/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.json
 create mode 100644 _data/impact/paper_notes/paper_s2_5bef8601da9663add6a85713414f8c945cf97355.json
 create mode 100644 _data/impact/paper_notes/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json
 create mode 100644 _data/impact/paper_notes/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.json
 create mode 100644 _data/impact/paper_notes/paper_s2_b6c64dbe0130ef1bf6af12266b958a5d5396101f.json
 create mode 100644 _data/impact/paper_notes/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json
 create mode 100644 _data/impact/paper_notes/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.json
 create mode 100644 _data/impact/paper_notes/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json
 create mode 100644 _data/impact/papers.json
 create mode 100644 _data/impact/people.json
 create mode 100644 _data/impact/people_decisions.json
 create mode 100644 _data/impact/policy.json
 create mode 100644 _data/impact/recognition_highlights.json
 create mode 100644 _data/impact/resources.json
 create mode 100644 _data/impact/stats.json
 create mode 100644 _data/impact/talks.json
 create mode 100644 _data/impact/techniques.json
 create mode 100644 _data/papers/paper_arxiv_2105_10016.json
 create mode 100644 _data/papers/paper_arxiv_2206_08530.json
 create mode 100644 _data/papers/paper_arxiv_2304_10044.json
 create mode 100644 _data/papers/paper_arxiv_2310_06433.json
 create mode 100644 _data/papers/paper_arxiv_2311_06728.json
 create mode 100644 _data/papers/paper_arxiv_2312_04941.json
 create mode 100644 _data/papers/paper_arxiv_2402_00292.json
 create mode 100644 _data/papers/paper_arxiv_2406_09469.json
 create mode 100644 _data/papers/paper_arxiv_2407_04294.json
 create mode 100644 _data/papers/paper_arxiv_2503_03893.json
 create mode 100644 _data/papers/paper_arxiv_2503_17322.json
 create mode 100644 _data/papers/paper_arxiv_2506_02617.json
 create mode 100644 _data/papers/paper_arxiv_2509_10819.json
 create mode 100644 _data/papers/paper_arxiv_2510_06663.json
 create mode 100644 _data/papers/paper_arxiv_2511_17377.json
 create mode 100644 _data/papers/paper_arxiv_2601_15074.json
 create mode 100644 _data/papers/paper_arxiv_2602_19490.json
 create mode 100644 _data/papers/paper_arxiv_2603_00311.json
 create mode 100644 _data/papers/paper_arxiv_2603_19434.json
 create mode 100644 _data/papers/paper_arxiv_2603_21530.json
 create mode 100644 _data/papers/paper_arxiv_2604_01442.json
 create mode 100644 _data/papers/paper_arxiv_2604_03024.json
 create mode 100644 _data/papers/paper_arxiv_2605_20473.json
 create mode 100644 _data/papers/paper_arxiv_2605_22992.json
 create mode 100644 _data/papers/paper_arxiv_2606_11132.json
 create mode 100644 _data/papers/paper_arxiv_2606_14164.json
 create mode 100644 _data/papers/paper_arxiv_2607_03741.json
 create mode 100644 _data/papers/paper_arxiv_2607_09072.json
 create mode 100644 _data/papers/paper_arxiv_2607_13276.json
 create mode 100644 _data/papers/paper_arxiv_2608_15709.json
 create mode 100644 _data/papers/paper_arxiv_2608_23402.json
 create mode 100644 _data/papers/paper_arxiv_2608_25573.json
 create mode 100644 _data/papers/paper_arxiv_2608_30385.json
 create mode 100644 _data/papers/paper_arxiv_2609_00381.json
 create mode 100644 _data/papers/paper_doi_10_1007_978_3_030_71058_3_5.json
 create mode 100644 _data/papers/paper_doi_10_1007_978_3_030_88494_9_12.json
 create mode 100644 _data/papers/paper_doi_10_1007_978_3_031_51479_1_17.json
 create mode 100644 _data/papers/paper_doi_10_1007_978_3_031_94706_3_7.json
 create mode 100644 _data/papers/paper_doi_10_1007_978_981_95_3182_0_3.json
 create mode 100644 _data/papers/paper_doi_10_1007_978_981_95_4721_0_7.json
 create mode 100644 _data/papers/paper_doi_10_1007_978_981_96_4506_0_18.json
 create mode 100644 _data/papers/paper_doi_10_1007_978_981_96_6465_8_28.json
 create mode 100644 _data/papers/paper_doi_10_1007_s10664_025_10662_w.json
 create mode 100644 _data/papers/paper_doi_10_1016_j_cose_2025_104564.json
 create mode 100644 _data/papers/paper_doi_10_1016_j_displa_2024_102854.json
 create mode 100644 _data/papers/paper_doi_10_1109_ase56229_2023_00106.json
 create mode 100644 _data/papers/paper_doi_10_1109_ase63991_2025_00095.json
 create mode 100644 _data/papers/paper_doi_10_1109_ase63991_2025_00151.json
 create mode 100644 _data/papers/paper_doi_10_1109_ase63991_2025_00322.json
 create mode 100644 _data/papers/paper_doi_10_1109_compsac57700_2023_00273.json
 create mode 100644 _data/papers/paper_doi_10_1109_compsac61105_2024_00141.json
 create mode 100644 _data/papers/paper_doi_10_1109_dsc55868_2022_00057.json
 create mode 100644 _data/papers/paper_doi_10_1109_girst67753_2025_11382165.json
 create mode 100644 _data/papers/paper_doi_10_1109_iaecst64597_2024_11117732.json
 create mode 100644 _data/papers/paper_doi_10_1109_iaecst68792_2025_11415166.json
 create mode 100644 _data/papers/paper_doi_10_1109_icccs65393_2025_11069832.json
 create mode 100644 _data/papers/paper_doi_10_1109_iccste65902_2025_11138310.json
 create mode 100644 _data/papers/paper_doi_10_1109_icde55515_2023_00057.json
 create mode 100644 _data/papers/paper_doi_10_1109_icde60146_2024_00011.json
 create mode 100644 _data/papers/paper_doi_10_1109_icde60146_2024_00441.json
 create mode 100644 _data/papers/paper_doi_10_1109_icde65706_2026_00180.json
 create mode 100644 _data/papers/paper_doi_10_1109_icde65706_2026_00224.json
 create mode 100644 _data/papers/paper_doi_10_1109_icde65706_2026_00240.json
 create mode 100644 _data/papers/paper_doi_10_1109_icicnis64247_2024_10823213.json
 create mode 100644 _data/papers/paper_doi_10_1109_icpc66645_2025_00021.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse43902_2021_00137.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse48619_2023_00024.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse48619_2023_00101.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse48619_2023_00173.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse48619_2023_00175.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse55347_2025_00003.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse55347_2025_00013.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse55347_2025_00045.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse55347_2025_00183.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse55347_2025_00257.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse_companion58688_2023_00041.json
 create mode 100644 _data/papers/paper_doi_10_1109_icse_seip52600_2021_00042.json
 create mode 100644 _data/papers/paper_doi_10_1109_icsip61881_2024_10671554.json
 create mode 100644 _data/papers/paper_doi_10_1109_icsme64153_2025_00030.json
 create mode 100644 _data/papers/paper_doi_10_1109_icst60714_2024_00012.json
 create mode 100644 _data/papers/paper_doi_10_1109_ictai62512_2024_00085.json
 create mode 100644 _data/papers/paper_doi_10_1109_iemcon67450_2025_11381190.json
 create mode 100644 _data/papers/paper_doi_10_1109_issrew55968_2022_00056.json
 create mode 100644 _data/papers/paper_doi_10_1109_missf68264_2026_11521893.json
 create mode 100644 _data/papers/paper_doi_10_1109_punecon67554_2025_11378586.json
 create mode 100644 _data/papers/paper_doi_10_1109_saner60148_2024_00096.json
 create mode 100644 _data/papers/paper_doi_10_1109_sp54263_2024_00109.json
 create mode 100644 _data/papers/paper_doi_10_1109_srds69199_2025_00038.json
 create mode 100644 _data/papers/paper_doi_10_1109_tdsc_2024_3521591.json
 create mode 100644 _data/papers/paper_doi_10_1109_tkde_2026_3656491.json
 create mode 100644 _data/papers/paper_doi_10_1109_tse_2025_3574328.json
 create mode 100644 _data/papers/paper_doi_10_1109_tse_2025_3625300.json
 create mode 100644 _data/papers/paper_doi_10_1117_12_3006402.json
 create mode 100644 _data/papers/paper_doi_10_1145_3372297_3417260.json
 create mode 100644 _data/papers/paper_doi_10_1145_3395032_3395322.json
 create mode 100644 _data/papers/paper_doi_10_1145_3428261.json
 create mode 100644 _data/papers/paper_doi_10_1145_3448016_3457559.json
 create mode 100644 _data/papers/paper_doi_10_1145_3468264_3468540.json
 create mode 100644 _data/papers/paper_doi_10_1145_3468264_3468573.json
 create mode 100644 _data/papers/paper_doi_10_1145_3471485_3471491.json
 create mode 100644 _data/papers/paper_doi_10_1145_3485529.json
 create mode 100644 _data/papers/paper_doi_10_1145_3510003_3510093.json
 create mode 100644 _data/papers/paper_doi_10_1145_3510457_3513034.json
 create mode 100644 _data/papers/paper_doi_10_1145_3531348_3532176.json
 create mode 100644 _data/papers/paper_doi_10_1145_3533767_3534364.json
 create mode 100644 _data/papers/paper_doi_10_1145_3533767_3534409.json
 create mode 100644 _data/papers/paper_doi_10_1145_3551349_3556924.json
 create mode 100644 _data/papers/paper_doi_10_1145_3551349_3560431.json
 create mode 100644 _data/papers/paper_doi_10_1145_3552326_3587448.json
 create mode 100644 _data/papers/paper_doi_10_1145_3582016_3582053.json
 create mode 100644 _data/papers/paper_doi_10_1145_3588909.json
 create mode 100644 _data/papers/paper_doi_10_1145_3597503_3608133.json
 create mode 100644 _data/papers/paper_doi_10_1145_3597503_3639112.json
 create mode 100644 _data/papers/paper_doi_10_1145_3597503_3639200.json
 create mode 100644 _data/papers/paper_doi_10_1145_3597503_3639207.json
 create mode 100644 _data/papers/paper_doi_10_1145_3597503_3639210.json
 create mode 100644 _data/papers/paper_doi_10_1145_3597503_3639212.json
 create mode 100644 _data/papers/paper_doi_10_1145_3597926_3598046.json
 create mode 100644 _data/papers/paper_doi_10_1145_3597926_3598052.json
 create mode 100644 _data/papers/paper_doi_10_1145_3597926_3598068.json
 create mode 100644 _data/papers/paper_doi_10_1145_3597926_3598130.json
 create mode 100644 _data/papers/paper_doi_10_1145_3605157_3605177.json
 create mode 100644 _data/papers/paper_doi_10_1145_3611643_3613893.json
 create mode 100644 _data/papers/paper_doi_10_1145_3611643_3616286.json
 create mode 100644 _data/papers/paper_doi_10_1145_3622819.json
 create mode 100644 _data/papers/paper_doi_10_1145_3643779.json
 create mode 100644 _data/papers/paper_doi_10_1145_3643781.json
 create mode 100644 _data/papers/paper_doi_10_1145_3650212_3680311.json
 create mode 100644 _data/papers/paper_doi_10_1145_3650212_3680317.json
 create mode 100644 _data/papers/paper_doi_10_1145_3650212_3680318.json
 create mode 100644 _data/papers/paper_doi_10_1145_3650212_3680392.json
 create mode 100644 _data/papers/paper_doi_10_1145_3689031_3696064.json
 create mode 100644 _data/papers/paper_doi_10_1145_3689491_3691821.json
 create mode 100644 _data/papers/paper_doi_10_1145_3689757.json
 create mode 100644 _data/papers/paper_doi_10_1145_3690631.json
 create mode 100644 _data/papers/paper_doi_10_1145_3704870.json
 create mode 100644 _data/papers/paper_doi_10_1145_3708533.json
 create mode 100644 _data/papers/paper_doi_10_1145_3720504.json
 create mode 100644 _data/papers/paper_doi_10_1145_3728908.json
 create mode 100644 _data/papers/paper_doi_10_1145_3728953.json
 create mode 100644 _data/papers/paper_doi_10_1145_3728965.json
 create mode 100644 _data/papers/paper_doi_10_1145_3728973.json
 create mode 100644 _data/papers/paper_doi_10_1145_3729175.json
 create mode 100644 _data/papers/paper_doi_10_1145_3729319.json
 create mode 100644 _data/papers/paper_doi_10_1145_3731569_3764841.json
 create mode 100644 _data/papers/paper_doi_10_1145_3744916_3773102.json
 create mode 100644 _data/papers/paper_doi_10_1145_3749186.json
 create mode 100644 _data/papers/paper_doi_10_1145_3757347_3759132.json
 create mode 100644 _data/papers/paper_doi_10_1145_3758316_3763249.json
 create mode 100644 _data/papers/paper_doi_10_1145_3764583.json
 create mode 100644 _data/papers/paper_doi_10_1145_3769779.json
 create mode 100644 _data/papers/paper_doi_10_1145_3769828.json
 create mode 100644 _data/papers/paper_doi_10_1145_3769832.json
 create mode 100644 _data/papers/paper_doi_10_1145_3779212_3790244.json
 create mode 100644 _data/papers/paper_doi_10_1145_3785021_3787993.json
 create mode 100644 _data/papers/paper_doi_10_1145_3786673.json
 create mode 100644 _data/papers/paper_doi_10_1145_3786699.json
 create mode 100644 _data/papers/paper_doi_10_1145_3798226.json
 create mode 100644 _data/papers/paper_doi_10_1145_3798232.json
 create mode 100644 _data/papers/paper_doi_10_1145_3798245.json
 create mode 100644 _data/papers/paper_doi_10_1145_3799227.json
 create mode 100644 _data/papers/paper_doi_10_1145_3802034.json
 create mode 100644 _data/papers/paper_doi_10_1145_3802053.json
 create mode 100644 _data/papers/paper_doi_10_1145_3802061.json
 create mode 100644 _data/papers/paper_doi_10_1145_3803437_3806090.json
 create mode 100644 _data/papers/paper_doi_10_1145_3808109.json
 create mode 100644 _data/papers/paper_doi_10_1145_3810991_3811632.json
 create mode 100644 _data/papers/paper_doi_10_1145_3810991_3811634.json
 create mode 100644 _data/papers/paper_doi_10_1145_3810991_3811637.json
 create mode 100644 _data/papers/paper_doi_10_1145_3828685.json
 create mode 100644 _data/papers/paper_doi_10_1177_0926227x251370258.json
 create mode 100644 _data/papers/paper_doi_10_14711_thesis_991012980220103412.json
 create mode 100644 _data/papers/paper_doi_10_14722_ndss_2025_230530.json
 create mode 100644 _data/papers/paper_doi_10_14722_ndss_2026_240198.json
 create mode 100644 _data/papers/paper_doi_10_14778_3712221_3712247.json
 create mode 100644 _data/papers/paper_doi_10_14778_3749646_3749661.json
 create mode 100644 _data/papers/paper_doi_10_14778_3797919_3797928.json
 create mode 100644 _data/papers/paper_doi_10_14778_3828612_3828639.json
 create mode 100644 _data/papers/paper_doi_10_22399_ijcesen_5462.json
 create mode 100644 _data/papers/paper_doi_10_3390_app13042519.json
 create mode 100644 _data/papers/paper_doi_10_3390_electronics14193910.json
 create mode 100644 _data/papers/paper_doi_10_53799_zb6b1375.json
 create mode 100644 _data/papers/paper_doi_10_7717_peerj_cs_1592.json
 create mode 100644 _data/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json
 create mode 100644 _data/papers/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.json
 create mode 100644 _data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json
 create mode 100644 _data/papers/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.json
 create mode 100644 _data/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json
 create mode 100644 _data/papers/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.json
 create mode 100644 _data/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json
 create mode 100644 _includes/head/custom.html
 create mode 100644 _includes/impact/artifact-evidence.html
 create mode 100644 _includes/impact/bar-list.html
 create mode 100644 _includes/impact/counts-table.html
 create mode 100644 _includes/impact/dbms-detail.html
 create mode 100644 _includes/impact/feature-row.html
 create mode 100644 _includes/impact/headline.html
 create mode 100644 _includes/impact/paper-detail.html
 create mode 100644 _includes/impact/paper-list.html
 create mode 100644 _includes/impact/plots/bugs-by-dbms.svg
 create mode 100644 _includes/impact/plots/bugs-by-reporter.svg
 create mode 100644 _includes/impact/plots/bugs-by-status.svg
 create mode 100644 _includes/impact/plots/bugs-by-technique.svg
 create mode 100644 _includes/impact/plots/bugs-by-year.svg
 create mode 100644 _includes/impact/plots/papers-by-year.svg
 create mode 100644 _includes/impact/plots/papers-relationships-by-year.svg
 create mode 100644 _includes/impact/stat-tiles.html
 create mode 100644 _includes/impact/talk-detail.html
 create mode 100644 _includes/impact/talk-frames.html
 create mode 100644 _includes/impact/year-detail.html
 create mode 100644 _pages/impact.html
 create mode 100644 _pages/impact/bug-statistics.html
 create mode 100644 _pages/impact/bugs/2019.html
 create mode 100644 _pages/impact/bugs/2020.html
 create mode 100644 _pages/impact/bugs/2021.html
 create mode 100644 _pages/impact/bugs/2022.html
 create mode 100644 _pages/impact/bugs/2023.html
 create mode 100644 _pages/impact/bugs/2024.html
 create mode 100644 _pages/impact/bugs/2025.html
 create mode 100644 _pages/impact/bugs/2026.html
 create mode 100644 _pages/impact/database-systems.html
 create mode 100644 _pages/impact/dbms/bharatdbms.html
 create mode 100644 _pages/impact/dbms/citus.html
 create mode 100644 _pages/impact/dbms/clickhouse.html
 create mode 100644 _pages/impact/dbms/cloudberry.html
 create mode 100644 _pages/impact/dbms/cnosdb.html
 create mode 100644 _pages/impact/dbms/cockroachdb.html
 create mode 100644 _pages/impact/dbms/cratedb.html
 create mode 100644 _pages/impact/dbms/cubrid.html
 create mode 100644 _pages/impact/dbms/databend.html
 create mode 100644 _pages/impact/dbms/datafusion.html
 create mode 100644 _pages/impact/dbms/dolt.html
 create mode 100644 _pages/impact/dbms/doris.html
 create mode 100644 _pages/impact/dbms/duckdb.html
 create mode 100644 _pages/impact/dbms/falkordb.html
 create mode 100644 _pages/impact/dbms/feldera.html
 create mode 100644 _pages/impact/dbms/firebird.html
 create mode 100644 _pages/impact/dbms/greptimedb.html
 create mode 100644 _pages/impact/dbms/h2.html
 create mode 100644 _pages/impact/dbms/hazelcast.html
 create mode 100644 _pages/impact/dbms/hive.html
 create mode 100644 _pages/impact/dbms/hsqldb.html
 create mode 100644 _pages/impact/dbms/kyzo.html
 create mode 100644 _pages/impact/dbms/mariadb.html
 create mode 100644 _pages/impact/dbms/materialize.html
 create mode 100644 _pages/impact/dbms/matrixone.html
 create mode 100644 _pages/impact/dbms/monetdb.html
 create mode 100644 _pages/impact/dbms/mysql.html
 create mode 100644 _pages/impact/dbms/noisepage.html
 create mode 100644 _pages/impact/dbms/oceanbase.html
 create mode 100644 _pages/impact/dbms/oxla.html
 create mode 100644 _pages/impact/dbms/postgresql.html
 create mode 100644 _pages/impact/dbms/presto.html
 create mode 100644 _pages/impact/dbms/questdb.html
 create mode 100644 _pages/impact/dbms/risingwave.html
 create mode 100644 _pages/impact/dbms/seekdb.html
 create mode 100644 _pages/impact/dbms/serenedb.html
 create mode 100644 _pages/impact/dbms/spark.html
 create mode 100644 _pages/impact/dbms/sparq.html
 create mode 100644 _pages/impact/dbms/spiceai.html
 create mode 100644 _pages/impact/dbms/sqlite.html
 create mode 100644 _pages/impact/dbms/starrocks.html
 create mode 100644 _pages/impact/dbms/stonedb.html
 create mode 100644 _pages/impact/dbms/tarantool.html
 create mode 100644 _pages/impact/dbms/tdengine.html
 create mode 100644 _pages/impact/dbms/tidb.html
 create mode 100644 _pages/impact/dbms/tikv.html
 create mode 100644 _pages/impact/dbms/turso.html
 create mode 100644 _pages/impact/dbms/umbra.html
 create mode 100644 _pages/impact/dbms/virtuoso.html
 create mode 100644 _pages/impact/dbms/wadjet.html
 create mode 100644 _pages/impact/dbms/xugu.html
 create mode 100644 _pages/impact/dbms/ydb.html
 create mode 100644 _pages/impact/dbms/yugabytedb.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2105_10016.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2206_08530.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2304_10044.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2310_06433.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2311_06728.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2312_04941.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2402_00292.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2406_09469.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2407_04294.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2503_03893.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2503_17322.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2506_02617.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2509_10819.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2510_06663.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2511_17377.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2601_15074.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2602_19490.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2603_00311.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2603_19434.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2603_21530.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2604_01442.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2604_03024.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2605_20473.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2605_22992.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2606_11132.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2606_14164.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2607_03741.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2607_09072.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2607_13276.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2608_15709.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2608_23402.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2608_25573.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2608_30385.html
 create mode 100644 _pages/impact/papers/paper_arxiv_2609_00381.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1007_978_3_030_71058_3_5.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1007_978_3_030_88494_9_12.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1007_978_3_031_51479_1_17.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1007_978_3_031_94706_3_7.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1007_978_981_95_3182_0_3.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1007_978_981_95_4721_0_7.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1007_978_981_96_4506_0_18.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1007_978_981_96_6465_8_28.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1007_s10664_025_10662_w.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1016_j_cose_2025_104564.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1016_j_displa_2024_102854.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_ase56229_2023_00106.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_ase63991_2025_00095.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_ase63991_2025_00151.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_ase63991_2025_00322.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_compsac57700_2023_00273.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_compsac61105_2024_00141.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_dsc55868_2022_00057.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_girst67753_2025_11382165.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_iaecst64597_2024_11117732.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_iaecst68792_2025_11415166.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icccs65393_2025_11069832.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_iccste65902_2025_11138310.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icde55515_2023_00057.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icde60146_2024_00011.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icde60146_2024_00441.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icde65706_2026_00180.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icde65706_2026_00224.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icde65706_2026_00240.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icicnis64247_2024_10823213.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icpc66645_2025_00021.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse43902_2021_00137.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse48619_2023_00024.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse48619_2023_00101.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse48619_2023_00173.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse48619_2023_00175.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse55347_2025_00003.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse55347_2025_00013.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse55347_2025_00045.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse55347_2025_00183.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse55347_2025_00257.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse_companion58688_2023_00041.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icse_seip52600_2021_00042.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icsip61881_2024_10671554.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icsme64153_2025_00030.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_icst60714_2024_00012.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_ictai62512_2024_00085.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_iemcon67450_2025_11381190.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_issrew55968_2022_00056.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_missf68264_2026_11521893.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_punecon67554_2025_11378586.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_saner60148_2024_00096.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_sp54263_2024_00109.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_srds69199_2025_00038.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_tdsc_2024_3521591.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_tkde_2026_3656491.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_tse_2025_3574328.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1109_tse_2025_3625300.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1117_12_3006402.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3372297_3417260.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3395032_3395322.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3428261.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3448016_3457559.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3468264_3468540.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3468264_3468573.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3471485_3471491.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3485529.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3510003_3510093.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3510457_3513034.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3531348_3532176.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3533767_3534364.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3533767_3534409.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3551349_3556924.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3551349_3560431.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3552326_3587448.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3582016_3582053.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3588909.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3597503_3608133.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3597503_3639112.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3597503_3639200.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3597503_3639207.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3597503_3639210.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3597503_3639212.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3597926_3598046.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3597926_3598052.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3597926_3598068.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3597926_3598130.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3605157_3605177.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3611643_3613893.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3611643_3616286.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3622819.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3643779.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3643781.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3650212_3680311.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3650212_3680317.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3650212_3680318.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3650212_3680392.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3689031_3696064.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3689491_3691821.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3689757.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3690631.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3704870.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3708533.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3720504.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3728908.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3728953.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3728965.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3728973.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3729175.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3729319.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3731569_3764841.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3744916_3773102.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3749186.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3757347_3759132.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3758316_3763249.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3764583.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3769779.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3769828.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3769832.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3779212_3790244.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3785021_3787993.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3786673.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3786699.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3798226.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3798232.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3798245.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3799227.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3802034.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3802053.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3802061.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3803437_3806090.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3808109.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3810991_3811632.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3810991_3811634.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3810991_3811637.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1145_3828685.html
 create mode 100644 _pages/impact/papers/paper_doi_10_1177_0926227x251370258.html
 create mode 100644 _pages/impact/papers/paper_doi_10_14711_thesis_991012980220103412.html
 create mode 100644 _pages/impact/papers/paper_doi_10_14722_ndss_2025_230530.html
 create mode 100644 _pages/impact/papers/paper_doi_10_14722_ndss_2026_240198.html
 create mode 100644 _pages/impact/papers/paper_doi_10_14778_3712221_3712247.html
 create mode 100644 _pages/impact/papers/paper_doi_10_14778_3749646_3749661.html
 create mode 100644 _pages/impact/papers/paper_doi_10_14778_3797919_3797928.html
 create mode 100644 _pages/impact/papers/paper_doi_10_14778_3828612_3828639.html
 create mode 100644 _pages/impact/papers/paper_doi_10_22399_ijcesen_5462.html
 create mode 100644 _pages/impact/papers/paper_doi_10_3390_app13042519.html
 create mode 100644 _pages/impact/papers/paper_doi_10_3390_electronics14193910.html
 create mode 100644 _pages/impact/papers/paper_doi_10_53799_zb6b1375.html
 create mode 100644 _pages/impact/papers/paper_doi_10_7717_peerj_cs_1592.html
 create mode 100644 _pages/impact/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.html
 create mode 100644 _pages/impact/papers/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.html
 create mode 100644 _pages/impact/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.html
 create mode 100644 _pages/impact/papers/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.html
 create mode 100644 _pages/impact/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.html
 create mode 100644 _pages/impact/papers/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.html
 create mode 100644 _pages/impact/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.html
 create mode 100644 _pages/impact/recognition.html
 create mode 100644 _pages/impact/talks/645a47ac1426.html
 create mode 100644 _pages/impact/talks/6YGqFRTe2D0.html
 create mode 100644 _pages/impact/talks/9d70cce25a55.html
 create mode 100644 _pages/impact/talks/BgC79Zt2fPs.html
 create mode 100644 _pages/impact/talks/CW4Ntdtp7lg.html
 create mode 100644 _pages/impact/talks/L90MBb6NLBE.html
 create mode 100644 _pages/impact/talks/QRwxHGpWaUA.html
 create mode 100644 _pages/impact/talks/V_qzqY1bb7I.html
 create mode 100644 _pages/impact/talks/wHo-VtzTHx0.html
 create mode 100644 assets/css/impact.css
 create mode 100644 assets/images/impact/talks/6YGqFRTe2D0-1509.jpg
 create mode 100644 assets/images/impact/talks/6YGqFRTe2D0-295.jpg
 create mode 100644 assets/images/impact/talks/BgC79Zt2fPs-2144.jpg
 create mode 100644 assets/images/impact/talks/BgC79Zt2fPs-2170.jpg
 create mode 100644 assets/images/impact/talks/BgC79Zt2fPs-2455.jpg
 create mode 100644 assets/images/impact/talks/CW4Ntdtp7lg-110.jpg
 create mode 100644 assets/images/impact/talks/CW4Ntdtp7lg-827.jpg
 create mode 100644 assets/images/impact/talks/L90MBb6NLBE-1703.jpg
 create mode 100644 assets/images/impact/talks/QRwxHGpWaUA-205.jpg
 create mode 100644 assets/images/impact/talks/README.md
 create mode 100644 assets/images/impact/talks/V_qzqY1bb7I-2836.jpg
 create mode 100644 assets/images/impact/talks/clickhouse-2021-cpp-siberia.jpg
 create mode 100644 assets/images/impact/talks/clickhouse-2022-release-22.3.jpg
 delete mode 100644 index.html
 create mode 100644 requirements.txt
 create mode 100644 schemas/impact/adoption.schema.json
 create mode 100644 schemas/impact/bugs.schema.json
 create mode 100644 schemas/impact/common.schema.json
 create mode 100644 schemas/impact/dbms.schema.json
 create mode 100644 schemas/impact/needs_review.schema.json
 create mode 100644 schemas/impact/paper-note.schema.json
 create mode 100644 schemas/impact/papers.schema.json
 create mode 100644 schemas/impact/people.schema.json
 create mode 100644 schemas/impact/people_decisions.schema.json
 create mode 100644 schemas/impact/policy.schema.json
 create mode 100644 schemas/impact/recognition_highlights.schema.json
 create mode 100644 schemas/impact/resources.schema.json
 create mode 100644 schemas/impact/stats.schema.json
 create mode 100644 schemas/impact/talks.schema.json
 create mode 100644 schemas/impact/techniques.schema.json
 create mode 100644 schemas/papers/paper-analysis.schema.json
 create mode 100644 tools/__init__.py
 create mode 100644 tools/impact/README.md
 create mode 100644 tools/impact/__init__.py
 create mode 100644 tools/impact/cache.py
 create mode 100644 tools/impact/classify/__init__.py
 create mode 100644 tools/impact/classify/prompts.py
 create mode 100644 tools/impact/collectors/__init__.py
 create mode 100644 tools/impact/collectors/adoption.py
 create mode 100644 tools/impact/collectors/artifact_links.py
 create mode 100644 tools/impact/collectors/artifacts.py
 create mode 100644 tools/impact/collectors/dbms_registry.py
 create mode 100644 tools/impact/collectors/forks.py
 create mode 100644 tools/impact/collectors/fulltext.py
 create mode 100644 tools/impact/collectors/gitee_bugs.py
 create mode 100644 tools/impact/collectors/github_bugs.py
 create mode 100644 tools/impact/collectors/lab_members.py
 create mode 100644 tools/impact/collectors/mariadb_jira.py
 create mode 100644 tools/impact/collectors/nus_test.py
 create mode 100644 tools/impact/collectors/papers.py
 create mode 100644 tools/impact/collectors/people.py
 create mode 100644 tools/impact/collectors/provider_bugs.py
 create mode 100644 tools/impact/collectors/resources.py
 create mode 100644 tools/impact/collectors/sqlancer_bugs.py
 create mode 100644 tools/impact/collectors/talks.py
 create mode 100644 tools/impact/config.py
 create mode 100644 tools/impact/dataset.py
 create mode 100644 tools/impact/dedupe.py
 create mode 100644 tools/impact/github.py
 create mode 100644 tools/impact/http.py
 create mode 100644 tools/impact/intake.py
 create mode 100644 tools/impact/notes.py
 create mode 100644 tools/impact/pages.py
 create mode 100644 tools/impact/plots.py
 create mode 100644 tools/impact/pr.py
 create mode 100644 tools/impact/reconcile_papers.py
 create mode 100644 tools/impact/repos.py
 create mode 100644 tools/impact/review.py
 create mode 100644 tools/impact/run.py
 create mode 100644 tools/impact/scholarly.py
 create mode 100644 tools/impact/seed/__init__.py
 create mode 100644 tools/impact/seed/paper_classifications.py
 create mode 100644 tools/impact/seed/techniques_seed.py
 create mode 100644 tools/impact/stats.py
 create mode 100644 tools/impact/talks.py
 create mode 100644 tools/impact/taxonomy.py
 create mode 100644 tools/impact/tests/__init__.py
 create mode 100644 tools/impact/tests/test_data.py
 create mode 100644 tools/impact/tests/test_pipeline.py
 create mode 100644 tools/impact/util.py
 create mode 100644 tools/impact/validate.py
 create mode 100644 tools/impact/worklist.py
 create mode 100644 tools/papers/__init__.py
 create mode 100644 tools/papers/analysis.py
 create mode 100644 tools/papers/checks.py
 create mode 100644 tools/papers/cli.py
 create mode 100644 tools/papers/collect.py
 create mode 100644 tools/papers/extract.py
 create mode 100644 tools/papers/mentions.py
 create mode 100644 tools/papers/pages.py
 create mode 100644 tools/papers/prompt.py
 create mode 100644 tools/papers/store.py
 create mode 100644 tools/papers/tests/__init__.py
 create mode 100644 tools/papers/tests/test_analysis.py
 create mode 100644 tools/papers/tests/test_extract.py
 create mode 100644 tools/papers/tests/test_mentions.py

diff --git a/.cache/impact/README.md b/.cache/impact/README.md
new file mode 100644
index 0000000..32b3da3
--- /dev/null
+++ b/.cache/impact/README.md
@@ -0,0 +1,23 @@
+# Collector caches
+
+Everything here is an optimisation. The pipeline produces the same records with
+this directory empty as with it warm, and a test asserts that.
+
+What is committed is decided by one question: could a fresh checkout get this
+back by itself? A fetched page could, so it is not committed. An answer that
+cost tokens, a transcript a person had to read out of a browser, and the record
+of how far each source has been scanned could not.
+
+| Path | Committed | What it holds |
+| --- | --- | --- |
+| `state.json` | yes | Last successful scan and last full reconciliation per source. This is what makes weekly runs incremental. |
+| `classifications/` | yes | Every classifier answer, including the negative and uncertain ones, keyed so that a change to the evidence, policy, taxonomy, prompt version or model forces a fresh answer and nothing else does. Only the answer and a hash of the evidence are stored, never the evidence itself. |
+| `talk-transcripts/` | yes | Windows of caption text around the moments a talk mentions SQLancer. Committed because a person had to obtain them by hand -- YouTube serves caption tracks only to its own player -- so no rerun can recover them. See that directory's README. |
+| `derived/` | yes | Deterministic extraction results, keyed by source id + content hash + extractor version. Empty so far: nothing has cost enough to keep. |
+| `http/`, `github/`, `papers/`, `artifacts/` | no | Fetched bodies with their ETag / Last-Modified / upstream `updated_at` validators. Large, re-fetchable, and restored from the GitHub Actions cache instead. |
+| `papers-pdf/`, `papers-text/` | no | Supplied PDFs and the text extracted from them. Deliberately never committed; only the two Markdown files explaining the directory are. |
+
+Committing the classification cache is what keeps the weekly token cost
+proportional to genuinely new evidence: a paper that turned out to only cite
+SQLancer, or an issue that turned out to be unrelated, is never asked about
+again.
diff --git a/.cache/impact/classifications/.gitkeep b/.cache/impact/classifications/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/.cache/impact/classifications/00/00ec0c4ab2395175ad39e0db6af27133acf0ece46c7310ab724e35aa023f3468.json b/.cache/impact/classifications/00/00ec0c4ab2395175ad39e0db6af27133acf0ece46c7310ab724e35aa023f3468.json
new file mode 100644
index 0000000..17788cd
--- /dev/null
+++ b/.cache/impact/classifications/00/00ec0c4ab2395175ad39e0db6af27133acf0ece46c7310ab724e35aa023f3468.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sourcenetwork/defradb.rs#157|sha256:0159711df6fd52cd2a699b6c60acd1bc8047472e69973af3824d6a7ef28c924c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0159711df6fd52cd2a699b6c60acd1bc8047472e69973af3824d6a7ef28c924c",
+    "source_id": "github:sourcenetwork/defradb.rs#157",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/02/022bc8ec1b7eaddf62d93509b5b4f2beb7ea6fdbca92818282945fd4f8592e5c.json b/.cache/impact/classifications/02/022bc8ec1b7eaddf62d93509b5b4f2beb7ea6fdbca92818282945fd4f8592e5c.json
new file mode 100644
index 0000000..33a6a98
--- /dev/null
+++ b/.cache/impact/classifications/02/022bc8ec1b7eaddf62d93509b5b4f2beb7ea6fdbca92818282945fd4f8592e5c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:apache/datafusion#14535|sha256:9fceee6658a445ab1f52c3c607ce61416d06f32872dd4d16b38e5da1f3896984|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9fceee6658a445ab1f52c3c607ce61416d06f32872dd4d16b38e5da1f3896984",
+    "source_id": "github:apache/datafusion#14535",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/02/0291964c195d72f8a24f4c54a20e0149e1eb1994ae783a10e6bd6b5f43dceb42.json b/.cache/impact/classifications/02/0291964c195d72f8a24f4c54a20e0149e1eb1994ae783a10e6bd6b5f43dceb42.json
new file mode 100644
index 0000000..be16dc6
--- /dev/null
+++ b/.cache/impact/classifications/02/0291964c195d72f8a24f4c54a20e0149e1eb1994ae783a10e6bd6b5f43dceb42.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#171420|sha256:a3c4e88181f416bcfcc47103f659f614cea733da51110147d481b561a1de04f0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a3c4e88181f416bcfcc47103f659f614cea733da51110147d481b561a1de04f0",
+    "source_id": "github:cockroachdb/cockroach#171420",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/02/02b1ecf20f9657dd9f6fba8da840822cdcd50a35e7cb8c4656d70629bd6cdcaa.json b/.cache/impact/classifications/02/02b1ecf20f9657dd9f6fba8da840822cdcd50a35e7cb8c4656d70629bd6cdcaa.json
new file mode 100644
index 0000000..7aab086
--- /dev/null
+++ b/.cache/impact/classifications/02/02b1ecf20f9657dd9f6fba8da840822cdcd50a35e7cb8c4656d70629bd6cdcaa.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:yugabyte/yugabyte-db#10236|sha256:d009f54b290299815ca53cce87b5723188d8f2c320cae0c72c694d475c419ecf|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "SQLancer: IOException occurs on UDPATE statement"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": null
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d009f54b290299815ca53cce87b5723188d8f2c320cae0c72c694d475c419ecf",
+    "source_id": "github:yugabyte/yugabyte-db#10236",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/03/03b1a6cffb6b046f97f2ef36d208258d7cc148dd2ef47da3f5d0c64a0482e69c.json b/.cache/impact/classifications/03/03b1a6cffb6b046f97f2ef36d208258d7cc148dd2ef47da3f5d0c64a0482e69c.json
new file mode 100644
index 0000000..209893f
--- /dev/null
+++ b/.cache/impact/classifications/03/03b1a6cffb6b046f97f2ef36d208258d7cc148dd2ef47da3f5d0c64a0482e69c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:databendlabs/databend#9448|sha256:7dcd4cd0e809656806ecd0d617ff11284dd2f7b0cc9704140858aafaf661f416|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7dcd4cd0e809656806ecd0d617ff11284dd2f7b0cc9704140858aafaf661f416",
+    "source_id": "github:databendlabs/databend#9448",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/05/05a9632477e1270abb608e2426c322ffe2afc56f509a58453dffa2bb78c8ec72.json b/.cache/impact/classifications/05/05a9632477e1270abb608e2426c322ffe2afc56f509a58453dffa2bb78c8ec72.json
new file mode 100644
index 0000000..11e98a8
--- /dev/null
+++ b/.cache/impact/classifications/05/05a9632477e1270abb608e2426c322ffe2afc56f509a58453dffa2bb78c8ec72.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#10517|sha256:9e8fa41de7c8985480545d2866fc42c35d16c68a9008682756189cd329c45ef2|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9e8fa41de7c8985480545d2866fc42c35d16c68a9008682756189cd329c45ef2",
+    "source_id": "github:cockroachdb/cockroach#10517",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/05/05e103a12eb6ae65021d96b2743d0a7a7d8367f6711135e566db2ca1957f7560.json b/.cache/impact/classifications/05/05e103a12eb6ae65021d96b2743d0a7a7d8367f6711135e566db2ca1957f7560.json
new file mode 100644
index 0000000..520c717
--- /dev/null
+++ b/.cache/impact/classifications/05/05e103a12eb6ae65021d96b2743d0a7a7d8367f6711135e566db2ca1957f7560.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#84484|sha256:137209a7e0f3d138ec5dd07b57203cc45d4c97a57dd67f856d396f16cd8daacf|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:137209a7e0f3d138ec5dd07b57203cc45d4c97a57dd67f856d396f16cd8daacf",
+    "source_id": "github:cockroachdb/cockroach#84484",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/05/05f9a6064d39f03460452b8b9d3e0450c637f97ea01b4705276e5d80d80d0283.json b/.cache/impact/classifications/05/05f9a6064d39f03460452b8b9d3e0450c637f97ea01b4705276e5d80d80d0283.json
new file mode 100644
index 0000000..d215ce4
--- /dev/null
+++ b/.cache/impact/classifications/05/05f9a6064d39f03460452b8b9d3e0450c637f97ea01b4705276e5d80d80d0283.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#144923|sha256:e736395c073f9f044c9d08687afc01f28a64ead7fa9871a03d3b34f4539e8bc3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e736395c073f9f044c9d08687afc01f28a64ead7fa9871a03d3b34f4539e8bc3",
+    "source_id": "github:cockroachdb/cockroach#144923",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/06/0620dd48ff05b205fb824aa7e1d2cfe94062778e8511f26a6cca1367ca4e7972.json b/.cache/impact/classifications/06/0620dd48ff05b205fb824aa7e1d2cfe94062778e8511f26a6cca1367ca4e7972.json
new file mode 100644
index 0000000..1df8935
--- /dev/null
+++ b/.cache/impact/classifications/06/0620dd48ff05b205fb824aa7e1d2cfe94062778e8511f26a6cca1367ca4e7972.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cnosdb/cnosdb#860|sha256:c622080e7aff8357d05af2346e7ab5091ca08878fd86774eca534bed340f8ee6|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c622080e7aff8357d05af2346e7ab5091ca08878fd86774eca534bed340f8ee6",
+    "source_id": "github:cnosdb/cnosdb#860",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/06/0685de5c34d1689373f48f48823606930c426ae9a9e9a95ab82bf7fb20f44d83.json b/.cache/impact/classifications/06/0685de5c34d1689373f48f48823606930c426ae9a9e9a95ab82bf7fb20f44d83.json
new file mode 100644
index 0000000..da34cb9
--- /dev/null
+++ b/.cache/impact/classifications/06/0685de5c34d1689373f48f48823606930c426ae9a9e9a95ab82bf7fb20f44d83.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#142948|sha256:6307bf2d2a226beb2df570e307b9c1a79de312d557cb87d8f1fca5a932e8d47e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6307bf2d2a226beb2df570e307b9c1a79de312d557cb87d8f1fca5a932e8d47e",
+    "source_id": "github:cockroachdb/cockroach#142948",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/07/071db0fbf07b947c8aa8b304afac95b3b6803698deb67cffb477d348397bb473.json b/.cache/impact/classifications/07/071db0fbf07b947c8aa8b304afac95b3b6803698deb67cffb477d348397bb473.json
new file mode 100644
index 0000000..b632e42
--- /dev/null
+++ b/.cache/impact/classifications/07/071db0fbf07b947c8aa8b304afac95b3b6803698deb67cffb477d348397bb473.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8132|sha256:deff14e2a938598c3e5c486a783043a354a320423922223e16a55e0b4424a4cb|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:deff14e2a938598c3e5c486a783043a354a320423922223e16a55e0b4424a4cb",
+    "source_id": "github:cockroachdb/cockroach#8132",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/07/074e5bd7ab805341f895bd5d7c555d5f0cf5cf69bc5b2c02fa475e4213c6e1cd.json b/.cache/impact/classifications/07/074e5bd7ab805341f895bd5d7c555d5f0cf5cf69bc5b2c02fa475e4213c6e1cd.json
new file mode 100644
index 0000000..aa2adf2
--- /dev/null
+++ b/.cache/impact/classifications/07/074e5bd7ab805341f895bd5d7c555d5f0cf5cf69bc5b2c02fa475e4213c6e1cd.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#166648|sha256:706f3c7fdde3c11a114060a3ad441f3d576cdeea6aa77802ac99a1cce636c026|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:706f3c7fdde3c11a114060a3ad441f3d576cdeea6aa77802ac99a1cce636c026",
+    "source_id": "github:cockroachdb/cockroach#166648",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/08/084e424f6707ac75bd465a026d302e4fd22e06dd403e1afd1b68a3fde9bd9221.json b/.cache/impact/classifications/08/084e424f6707ac75bd465a026d302e4fd22e06dd403e1afd1b68a3fde9bd9221.json
new file mode 100644
index 0000000..ddf3d1d
--- /dev/null
+++ b/.cache/impact/classifications/08/084e424f6707ac75bd465a026d302e4fd22e06dd403e1afd1b68a3fde9bd9221.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#1674|sha256:4dbe760255839765c52142e1a6a215fd5b6fc02ea0b0ebe7be823993313a4613|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4dbe760255839765c52142e1a6a215fd5b6fc02ea0b0ebe7be823993313a4613",
+    "source_id": "github:cockroachdb/cockroach#1674",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0a/0adea20f4f706ceb22e4ce210c6448c380fb849554566a3a12072fedf99555dc.json b/.cache/impact/classifications/0a/0adea20f4f706ceb22e4ce210c6448c380fb849554566a3a12072fedf99555dc.json
new file mode 100644
index 0000000..f9c5189
--- /dev/null
+++ b/.cache/impact/classifications/0a/0adea20f4f706ceb22e4ce210c6448c380fb849554566a3a12072fedf99555dc.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8167|sha256:d3b53c6dccac57fa9e5a2bd066a0d05460c08bc77221cd1b61e8ed4cf2d68400|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d3b53c6dccac57fa9e5a2bd066a0d05460c08bc77221cd1b61e8ed4cf2d68400",
+    "source_id": "github:cockroachdb/cockroach#8167",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0a/0ae74c01ced1a4ef65d237d0567a399acf166fbc3a6355c2bb47996c8a8ff015.json b/.cache/impact/classifications/0a/0ae74c01ced1a4ef65d237d0567a399acf166fbc3a6355c2bb47996c8a8ff015.json
new file mode 100644
index 0000000..20a7b7e
--- /dev/null
+++ b/.cache/impact/classifications/0a/0ae74c01ced1a4ef65d237d0567a399acf166fbc3a6355c2bb47996c8a8ff015.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#79610|sha256:82d763ed044bcd67096ce217d7f6672aa0d1b15ae806a2f667d7bde736be70d3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:82d763ed044bcd67096ce217d7f6672aa0d1b15ae806a2f667d7bde736be70d3",
+    "source_id": "github:cockroachdb/cockroach#79610",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0b/0b9a58e8ebf293ea894e2c86ace4a46053ed5ab89252753a52eb333469ee137d.json b/.cache/impact/classifications/0b/0b9a58e8ebf293ea894e2c86ace4a46053ed5ab89252753a52eb333469ee137d.json
new file mode 100644
index 0000000..b2e736b
--- /dev/null
+++ b/.cache/impact/classifications/0b/0b9a58e8ebf293ea894e2c86ace4a46053ed5ab89252753a52eb333469ee137d.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#4917|sha256:c34abd5ad691c44639b6d5cc6461ef774cd3a6c6cdc3a65cb5d70f6affdada81|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c34abd5ad691c44639b6d5cc6461ef774cd3a6c6cdc3a65cb5d70f6affdada81",
+    "source_id": "github:cockroachdb/cockroach#4917",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0b/0ba21c223b508f3b5fe69ec1d87be8e4ac3a2f187cad26cb1c3898e126606548.json b/.cache/impact/classifications/0b/0ba21c223b508f3b5fe69ec1d87be8e4ac3a2f187cad26cb1c3898e126606548.json
new file mode 100644
index 0000000..7a2aa59
--- /dev/null
+++ b/.cache/impact/classifications/0b/0ba21c223b508f3b5fe69ec1d87be8e4ac3a2f187cad26cb1c3898e126606548.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#3452|sha256:8ebc0b8e1ce6b89f3e1e8b57657c3890ffd7d37eb4fe3e778eaf6d11bddca416|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8ebc0b8e1ce6b89f3e1e8b57657c3890ffd7d37eb4fe3e778eaf6d11bddca416",
+    "source_id": "github:sparq-org/sparq#3452",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0c/0cb54e815b343da317dba60d90d9d9239fde75092b8f59c675b7625e9531c225.json b/.cache/impact/classifications/0c/0cb54e815b343da317dba60d90d9d9239fde75092b8f59c675b7625e9531c225.json
new file mode 100644
index 0000000..b8128cb
--- /dev/null
+++ b/.cache/impact/classifications/0c/0cb54e815b343da317dba60d90d9d9239fde75092b8f59c675b7625e9531c225.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#48116|sha256:0be8f4a0b3e88494c53f7f86101c54ec5837b86eaf6974c0582a65c39bd27f17|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0be8f4a0b3e88494c53f7f86101c54ec5837b86eaf6974c0582a65c39bd27f17",
+    "source_id": "github:cockroachdb/cockroach#48116",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0c/0cb8307f2658bdd072298b663d7a3a47149f35a758a7fa18e3293631259e71f9.json b/.cache/impact/classifications/0c/0cb8307f2658bdd072298b663d7a3a47149f35a758a7fa18e3293631259e71f9.json
new file mode 100644
index 0000000..a9150ee
--- /dev/null
+++ b/.cache/impact/classifications/0c/0cb8307f2658bdd072298b663d7a3a47149f35a758a7fa18e3293631259e71f9.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#22429|sha256:5ccfaeb6ad10a21d71b749b2549587a2725878d1485c32f29a9cbaffaab6abb1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5ccfaeb6ad10a21d71b749b2549587a2725878d1485c32f29a9cbaffaab6abb1",
+    "source_id": "github:pingcap/tidb#22429",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0c/0ce5ae07a2ef1883288dab55d9f7aa7be47ab108a96072e7256be4239367d3a6.json b/.cache/impact/classifications/0c/0ce5ae07a2ef1883288dab55d9f7aa7be47ab108a96072e7256be4239367d3a6.json
new file mode 100644
index 0000000..e4d8602
--- /dev/null
+++ b/.cache/impact/classifications/0c/0ce5ae07a2ef1883288dab55d9f7aa7be47ab108a96072e7256be4239367d3a6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#68174|sha256:a2de8940a693f2ed3cb50ac92160669c58f39844af181e123513af94ce1dbb4d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a2de8940a693f2ed3cb50ac92160669c58f39844af181e123513af94ce1dbb4d",
+    "source_id": "github:cockroachdb/cockroach#68174",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0d/0d6c8c592b79d58edbd2a96dcab698b70e9797030d5e2d2a210ed6716f49535a.json b/.cache/impact/classifications/0d/0d6c8c592b79d58edbd2a96dcab698b70e9797030d5e2d2a210ed6716f49535a.json
new file mode 100644
index 0000000..4707495
--- /dev/null
+++ b/.cache/impact/classifications/0d/0d6c8c592b79d58edbd2a96dcab698b70e9797030d5e2d2a210ed6716f49535a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#70483|sha256:f5b6bd99e8fbc7f1dc323859df1fbad3a50feeb927e0d4e28f321025df39ab76|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f5b6bd99e8fbc7f1dc323859df1fbad3a50feeb927e0d4e28f321025df39ab76",
+    "source_id": "github:cockroachdb/cockroach#70483",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0d/0de27f4f815dcf7cf3d5aa233e6c9b48df307b204314a2cea1028db091d76844.json b/.cache/impact/classifications/0d/0de27f4f815dcf7cf3d5aa233e6c9b48df307b204314a2cea1028db091d76844.json
new file mode 100644
index 0000000..8de4c97
--- /dev/null
+++ b/.cache/impact/classifications/0d/0de27f4f815dcf7cf3d5aa233e6c9b48df307b204314a2cea1028db091d76844.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#34209|sha256:c7dba67f34e8d38db6d8e7899cfe749c3769af6780bb720b99732ba323499bea|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c7dba67f34e8d38db6d8e7899cfe749c3769af6780bb720b99732ba323499bea",
+    "source_id": "github:pingcap/tidb#34209",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0e/0e39f3309390381085eb5e988d993e1a9108bfd386b91480df29481cc06f4fef.json b/.cache/impact/classifications/0e/0e39f3309390381085eb5e988d993e1a9108bfd386b91480df29481cc06f4fef.json
new file mode 100644
index 0000000..7afd89f
--- /dev/null
+++ b/.cache/impact/classifications/0e/0e39f3309390381085eb5e988d993e1a9108bfd386b91480df29481cc06f4fef.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#170332|sha256:f90d574054bf8356e90ea808222cbdcd9b012e5a73da6534772205bea1f6f57a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f90d574054bf8356e90ea808222cbdcd9b012e5a73da6534772205bea1f6f57a",
+    "source_id": "github:cockroachdb/cockroach#170332",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0e/0e4fb261364f3c9d0516d492b8f62002e85cf1a0140dd15f6d2799531c695457.json b/.cache/impact/classifications/0e/0e4fb261364f3c9d0516d492b8f62002e85cf1a0140dd15f6d2799531c695457.json
new file mode 100644
index 0000000..3ee27a1
--- /dev/null
+++ b/.cache/impact/classifications/0e/0e4fb261364f3c9d0516d492b8f62002e85cf1a0140dd15f6d2799531c695457.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#13394|sha256:038f8b00e3e76f10043335d5147600b2c4ecb9507c6f88aa1778d3f6f85da001|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:038f8b00e3e76f10043335d5147600b2c4ecb9507c6f88aa1778d3f6f85da001",
+    "source_id": "github:cockroachdb/cockroach#13394",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0e/0ecf0687a3b2c8434513fbe453d8e93d8e3dd1d854dba7ec10144a1bb2838b79.json b/.cache/impact/classifications/0e/0ecf0687a3b2c8434513fbe453d8e93d8e3dd1d854dba7ec10144a1bb2838b79.json
new file mode 100644
index 0000000..4e32600
--- /dev/null
+++ b/.cache/impact/classifications/0e/0ecf0687a3b2c8434513fbe453d8e93d8e3dd1d854dba7ec10144a1bb2838b79.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#62900|sha256:f249fd6091bdf5499ef650cd43563125411760ce349981ff0c93de3fb9fb3d9b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f249fd6091bdf5499ef650cd43563125411760ce349981ff0c93de3fb9fb3d9b",
+    "source_id": "github:cockroachdb/cockroach#62900",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0f/0f24025e5039fb3393fd19a213c25e70924ee86d7f642f1bfaf566ca7f381882.json b/.cache/impact/classifications/0f/0f24025e5039fb3393fd19a213c25e70924ee86d7f642f1bfaf566ca7f381882.json
new file mode 100644
index 0000000..f7e24df
--- /dev/null
+++ b/.cache/impact/classifications/0f/0f24025e5039fb3393fd19a213c25e70924ee86d7f642f1bfaf566ca7f381882.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tikv/tikv#9234|sha256:f546da89ad441094947ab88fd26b5577364c9e4d9688b66c6a65efa5f238ac0e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f546da89ad441094947ab88fd26b5577364c9e4d9688b66c6a65efa5f238ac0e",
+    "source_id": "github:tikv/tikv#9234",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0f/0f9cb7df12716cb64abc3a288862d58cf60a255211afa66ea2a50debf412561f.json b/.cache/impact/classifications/0f/0f9cb7df12716cb64abc3a288862d58cf60a255211afa66ea2a50debf412561f.json
new file mode 100644
index 0000000..a01db2b
--- /dev/null
+++ b/.cache/impact/classifications/0f/0f9cb7df12716cb64abc3a288862d58cf60a255211afa66ea2a50debf412561f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#73558|sha256:6e6717acbc49bafa27634b798227ab1e3fb6822efe14a5decb69044a411531ca|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6e6717acbc49bafa27634b798227ab1e3fb6822efe14a5decb69044a411531ca",
+    "source_id": "github:cockroachdb/cockroach#73558",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0f/0fbbf0043d76d2a11217fbebaa579336861b31e0b40a64aa088cb90e18673a41.json b/.cache/impact/classifications/0f/0fbbf0043d76d2a11217fbebaa579336861b31e0b40a64aa088cb90e18673a41.json
new file mode 100644
index 0000000..6d1b3a1
--- /dev/null
+++ b/.cache/impact/classifications/0f/0fbbf0043d76d2a11217fbebaa579336861b31e0b40a64aa088cb90e18673a41.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#6262|sha256:d0718e937c2106c0d52d001b670646cecac3f3be4af279dee417f4afcf75bb6c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d0718e937c2106c0d52d001b670646cecac3f3be4af279dee417f4afcf75bb6c",
+    "source_id": "github:cockroachdb/cockroach#6262",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/0f/0ff924a32c49459c18614d8e7399c07a2ffa192bdab9bbf5e2484b7908d92dcd.json b/.cache/impact/classifications/0f/0ff924a32c49459c18614d8e7399c07a2ffa192bdab9bbf5e2484b7908d92dcd.json
new file mode 100644
index 0000000..01d6afe
--- /dev/null
+++ b/.cache/impact/classifications/0f/0ff924a32c49459c18614d8e7399c07a2ffa192bdab9bbf5e2484b7908d92dcd.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#61624|sha256:a7718c1a9ce04d04cb62fdb858222dffeee7d3043c58ce3de745ff9a9070c0d6|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a7718c1a9ce04d04cb62fdb858222dffeee7d3043c58ce3de745ff9a9070c0d6",
+    "source_id": "github:cockroachdb/cockroach#61624",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/11/11a5a4f4d0adffe44e99f9a52b93846b6a4563d7ba313f5e52c4103115836fe8.json b/.cache/impact/classifications/11/11a5a4f4d0adffe44e99f9a52b93846b6a4563d7ba313f5e52c4103115836fe8.json
new file mode 100644
index 0000000..f6e40a9
--- /dev/null
+++ b/.cache/impact/classifications/11/11a5a4f4d0adffe44e99f9a52b93846b6a4563d7ba313f5e52c4103115836fe8.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#128765|sha256:fcc0c3e9a1c113642a740f69bb1a4ed4ccd3feef22ea4d8720502a2676287dd7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:fcc0c3e9a1c113642a740f69bb1a4ed4ccd3feef22ea4d8720502a2676287dd7",
+    "source_id": "github:cockroachdb/cockroach#128765",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/12/1235c9568b5f5257407888fcfb2acf44bbf213375d147bbbe1909b5a6bb7acdd.json b/.cache/impact/classifications/12/1235c9568b5f5257407888fcfb2acf44bbf213375d147bbbe1909b5a6bb7acdd.json
new file mode 100644
index 0000000..a1ea417
--- /dev/null
+++ b/.cache/impact/classifications/12/1235c9568b5f5257407888fcfb2acf44bbf213375d147bbbe1909b5a6bb7acdd.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#51601|sha256:9be8c32e614ac7926e83aa29556b28a3aa1e0ca17191666738d4bf88f75b397c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9be8c32e614ac7926e83aa29556b28a3aa1e0ca17191666738d4bf88f75b397c",
+    "source_id": "github:cockroachdb/cockroach#51601",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/13/13010fbc1de3f4360172e901910e96f38a9789e3e4c7ae7e194e376189e4d5a5.json b/.cache/impact/classifications/13/13010fbc1de3f4360172e901910e96f38a9789e3e4c7ae7e194e376189e4d5a5.json
new file mode 100644
index 0000000..fd46d70
--- /dev/null
+++ b/.cache/impact/classifications/13/13010fbc1de3f4360172e901910e96f38a9789e3e4c7ae7e194e376189e4d5a5.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#4450|sha256:6c6af9c0ae935013600c1b8c35f6608cbe430ec54882038835b20923315b8c6a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6c6af9c0ae935013600c1b8c35f6608cbe430ec54882038835b20923315b8c6a",
+    "source_id": "github:cockroachdb/cockroach#4450",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/13/137bb24122c4b373161e3490d14ea8a617b37b49f87a4f8cc3e0ac2e0829f42b.json b/.cache/impact/classifications/13/137bb24122c4b373161e3490d14ea8a617b37b49f87a4f8cc3e0ac2e0829f42b.json
new file mode 100644
index 0000000..f471d11
--- /dev/null
+++ b/.cache/impact/classifications/13/137bb24122c4b373161e3490d14ea8a617b37b49f87a4f8cc3e0ac2e0829f42b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#11961|sha256:efb799e7dcdaff10b9b57a50606b8402d87342a94aada2c52f2c7c5a8447cb4d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:efb799e7dcdaff10b9b57a50606b8402d87342a94aada2c52f2c7c5a8447cb4d",
+    "source_id": "github:pingcap/tidb#11961",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/15/150be7a1f97107c484a440ab68f518a2d6b906ec4ea1a1fe2f88ef58ac66da29.json b/.cache/impact/classifications/15/150be7a1f97107c484a440ab68f518a2d6b906ec4ea1a1fe2f88ef58ac66da29.json
new file mode 100644
index 0000000..9afc6fb
--- /dev/null
+++ b/.cache/impact/classifications/15/150be7a1f97107c484a440ab68f518a2d6b906ec4ea1a1fe2f88ef58ac66da29.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#9202|sha256:5d7fd1c31c39037034612f227c319f1905da40ecda1e63f88a9c348c48bdfbdd|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5d7fd1c31c39037034612f227c319f1905da40ecda1e63f88a9c348c48bdfbdd",
+    "source_id": "github:cockroachdb/cockroach#9202",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/15/155081c5f11ca6aa85ca9ea5c64bdcf9273f815f09dc3317b674c4c9ac724110.json b/.cache/impact/classifications/15/155081c5f11ca6aa85ca9ea5c64bdcf9273f815f09dc3317b674c4c9ac724110.json
new file mode 100644
index 0000000..a90b540
--- /dev/null
+++ b/.cache/impact/classifications/15/155081c5f11ca6aa85ca9ea5c64bdcf9273f815f09dc3317b674c4c9ac724110.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#125099|sha256:4cc11f621abe5e8ac51ae6e3a5487dedb0653b11e400988b2508a160458582b6|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4cc11f621abe5e8ac51ae6e3a5487dedb0653b11e400988b2508a160458582b6",
+    "source_id": "github:cockroachdb/cockroach#125099",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/15/15534187ea71903f63b427738a4a4ce959654a1d5902fe03a05bfeae34869bd4.json b/.cache/impact/classifications/15/15534187ea71903f63b427738a4a4ce959654a1d5902fe03a05bfeae34869bd4.json
new file mode 100644
index 0000000..52c86b3
--- /dev/null
+++ b/.cache/impact/classifications/15/15534187ea71903f63b427738a4a4ce959654a1d5902fe03a05bfeae34869bd4.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#76289|sha256:26011e583fbe9afa6fbc75978121b26e37f026fc061a98a113c58908f40ff8e9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:26011e583fbe9afa6fbc75978121b26e37f026fc061a98a113c58908f40ff8e9",
+    "source_id": "github:cockroachdb/cockroach#76289",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/15/15e97369f43c59e5fa8f9b42ccef618cc362574ba879d0152d9e00ceb5d33558.json b/.cache/impact/classifications/15/15e97369f43c59e5fa8f9b42ccef618cc362574ba879d0152d9e00ceb5d33558.json
new file mode 100644
index 0000000..1898988
--- /dev/null
+++ b/.cache/impact/classifications/15/15e97369f43c59e5fa8f9b42ccef618cc362574ba879d0152d9e00ceb5d33558.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#94049|sha256:295d23b1c66abc6c527320f232fe1600f7f4586864a875df7a1c093dcf1e23b0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:295d23b1c66abc6c527320f232fe1600f7f4586864a875df7a1c093dcf1e23b0",
+    "source_id": "github:elastic/elasticsearch#94049",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/18/1855156389f387c02a49c2707dd852af4f62c5ff1530f48774e42d5fb86a1ef8.json b/.cache/impact/classifications/18/1855156389f387c02a49c2707dd852af4f62c5ff1530f48774e42d5fb86a1ef8.json
new file mode 100644
index 0000000..ebb4b97
--- /dev/null
+++ b/.cache/impact/classifications/18/1855156389f387c02a49c2707dd852af4f62c5ff1530f48774e42d5fb86a1ef8.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:yugabyte/yugabyte-db#11298|sha256:1629582753e4a79419af06f917331a9ea26796a64f16a03959ba0e4fd905e310|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "[SQLancer][YSQL] ASC Primary keys update operation may insert new value"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": null
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1629582753e4a79419af06f917331a9ea26796a64f16a03959ba0e4fd905e310",
+    "source_id": "github:yugabyte/yugabyte-db#11298",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/18/1878770d70fa6c7dad0bc7dc46d0433e6c96b2c925958aa110091f55f8ba8f62.json b/.cache/impact/classifications/18/1878770d70fa6c7dad0bc7dc46d0433e6c96b2c925958aa110091f55f8ba8f62.json
new file mode 100644
index 0000000..6aea83f
--- /dev/null
+++ b/.cache/impact/classifications/18/1878770d70fa6c7dad0bc7dc46d0433e6c96b2c925958aa110091f55f8ba8f62.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:spiceai/spiceai#2190|sha256:bce852b42ebf0423e7bebf6d0b514cef783b30122c30ff3ef99b4f3650627035|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:bce852b42ebf0423e7bebf6d0b514cef783b30122c30ff3ef99b4f3650627035",
+    "source_id": "github:spiceai/spiceai#2190",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/19/199104e1cb80d36f2d60fe7a02bf9afb12a43e276d7707f05dcef6de4c8376b4.json b/.cache/impact/classifications/19/199104e1cb80d36f2d60fe7a02bf9afb12a43e276d7707f05dcef6de4c8376b4.json
new file mode 100644
index 0000000..98c8c41
--- /dev/null
+++ b/.cache/impact/classifications/19/199104e1cb80d36f2d60fe7a02bf9afb12a43e276d7707f05dcef6de4c8376b4.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#144834|sha256:ebc0945f07b4b786ffb502002fb1904c5cc84327ef1c00662a830490e42d7d5b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ebc0945f07b4b786ffb502002fb1904c5cc84327ef1c00662a830490e42d7d5b",
+    "source_id": "github:cockroachdb/cockroach#144834",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/19/19b024ccc309e35baaee5a2399e23dd0552b76a316c2208c5106c01d927cd7d9.json b/.cache/impact/classifications/19/19b024ccc309e35baaee5a2399e23dd0552b76a316c2208c5106c01d927cd7d9.json
new file mode 100644
index 0000000..7c7187c
--- /dev/null
+++ b/.cache/impact/classifications/19/19b024ccc309e35baaee5a2399e23dd0552b76a316c2208c5106c01d927cd7d9.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#10873|sha256:ae7caa417953b617f5dbff92bae26b884eac75a161f5c91d7090350664d8c1fb|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ae7caa417953b617f5dbff92bae26b884eac75a161f5c91d7090350664d8c1fb",
+    "source_id": "github:cockroachdb/cockroach#10873",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/19/19e6fa7d054f751ff6e1b02d0c14b1f24359d6167304aef53511eef311ab36a0.json b/.cache/impact/classifications/19/19e6fa7d054f751ff6e1b02d0c14b1f24359d6167304aef53511eef311ab36a0.json
new file mode 100644
index 0000000..492203d
--- /dev/null
+++ b/.cache/impact/classifications/19/19e6fa7d054f751ff6e1b02d0c14b1f24359d6167304aef53511eef311ab36a0.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#2961|sha256:e2489dc1417c4a409989931841b77b4553fcce0a2862b077b0badd1873ecdd68|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e2489dc1417c4a409989931841b77b4553fcce0a2862b077b0badd1873ecdd68",
+    "source_id": "github:sparq-org/sparq#2961",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1a/1a070f77f658bec092ad1821749d2be521284cc0416c827763e5211895613642.json b/.cache/impact/classifications/1a/1a070f77f658bec092ad1821749d2be521284cc0416c827763e5211895613642.json
new file mode 100644
index 0000000..f26dc6f
--- /dev/null
+++ b/.cache/impact/classifications/1a/1a070f77f658bec092ad1821749d2be521284cc0416c827763e5211895613642.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#88993|sha256:a01286a70d5db439830771a74c1f444763f6276634f7884159be6a41b12e67a4|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a01286a70d5db439830771a74c1f444763f6276634f7884159be6a41b12e67a4",
+    "source_id": "github:cockroachdb/cockroach#88993",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1a/1ae04d7f797a16fc64d3cbd05c5fb593034b0f38ad2f630c6e5d6c3b1e011371.json b/.cache/impact/classifications/1a/1ae04d7f797a16fc64d3cbd05c5fb593034b0f38ad2f630c6e5d6c3b1e011371.json
new file mode 100644
index 0000000..58564c1
--- /dev/null
+++ b/.cache/impact/classifications/1a/1ae04d7f797a16fc64d3cbd05c5fb593034b0f38ad2f630c6e5d6c3b1e011371.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:spiceai/spiceai#10116|sha256:95884822ec3801afea4e83143c97cfd41c27f6e27c43d1ef1e09b0ebfdb949a7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:95884822ec3801afea4e83143c97cfd41c27f6e27c43d1ef1e09b0ebfdb949a7",
+    "source_id": "github:spiceai/spiceai#10116",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1a/1ae2b17a75fac71963baa528bec7fd22426c55a15710dbd66d41e2ee52eb8811.json b/.cache/impact/classifications/1a/1ae2b17a75fac71963baa528bec7fd22426c55a15710dbd66d41e2ee52eb8811.json
new file mode 100644
index 0000000..51783c4
--- /dev/null
+++ b/.cache/impact/classifications/1a/1ae2b17a75fac71963baa528bec7fd22426c55a15710dbd66d41e2ee52eb8811.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#90583|sha256:461d10321d56e8f6aaf7675fedfec6f33d2560d167092f0a3d9f1cdde5a5a0a9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:461d10321d56e8f6aaf7675fedfec6f33d2560d167092f0a3d9f1cdde5a5a0a9",
+    "source_id": "github:cockroachdb/cockroach#90583",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1b/1b99ae9e948dd31c2f70aa43c6e316b00843229e0b170b2029180d78f1125557.json b/.cache/impact/classifications/1b/1b99ae9e948dd31c2f70aa43c6e316b00843229e0b170b2029180d78f1125557.json
new file mode 100644
index 0000000..8160935
--- /dev/null
+++ b/.cache/impact/classifications/1b/1b99ae9e948dd31c2f70aa43c6e316b00843229e0b170b2029180d78f1125557.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:ydb-platform/ydb#49010|sha256:19c9a9cb26258a906ad1d9b3e41ec630ffa904fc01bd1ba75ffbc5fc7c3f259d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:19c9a9cb26258a906ad1d9b3e41ec630ffa904fc01bd1ba75ffbc5fc7c3f259d",
+    "source_id": "github:ydb-platform/ydb#49010",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1c/1c146d4e84aeeff9dc22428f8c8cc1d51ae661f54620e57a9d51f7d538831633.json b/.cache/impact/classifications/1c/1c146d4e84aeeff9dc22428f8c8cc1d51ae661f54620e57a9d51f7d538831633.json
new file mode 100644
index 0000000..a017995
--- /dev/null
+++ b/.cache/impact/classifications/1c/1c146d4e84aeeff9dc22428f8c8cc1d51ae661f54620e57a9d51f7d538831633.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#3321|sha256:cdf3771129daeb55c14dca6d75b6487e75dc94f6cc556293a1a54568072eb02b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:cdf3771129daeb55c14dca6d75b6487e75dc94f6cc556293a1a54568072eb02b",
+    "source_id": "github:sparq-org/sparq#3321",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1c/1c60481568a9590f82b256626fa78b6276257f16cd884c11f2471396981e611e.json b/.cache/impact/classifications/1c/1c60481568a9590f82b256626fa78b6276257f16cd884c11f2471396981e611e.json
new file mode 100644
index 0000000..1466da4
--- /dev/null
+++ b/.cache/impact/classifications/1c/1c60481568a9590f82b256626fa78b6276257f16cd884c11f2471396981e611e.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:ydb-platform/ydb#32959|sha256:9eeb1c66d86b4feb6a29dbba041f548fd16cd819ac9d2f0da9b167b87a88f036|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9eeb1c66d86b4feb6a29dbba041f548fd16cd819ac9d2f0da9b167b87a88f036",
+    "source_id": "github:ydb-platform/ydb#32959",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1c/1c9a85183a72cf9dd1d9275b79a94f94b37d085440267379eb4e2c9927039571.json b/.cache/impact/classifications/1c/1c9a85183a72cf9dd1d9275b79a94f94b37d085440267379eb4e2c9927039571.json
new file mode 100644
index 0000000..dbf2554
--- /dev/null
+++ b/.cache/impact/classifications/1c/1c9a85183a72cf9dd1d9275b79a94f94b37d085440267379eb4e2c9927039571.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:pingcap/tidb#66601|sha256:edd51789962165c5d86abebabe8b99356b1fbbeb89620667a07917feeb6418a1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "oracle: `DQP`"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "dqp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:edd51789962165c5d86abebabe8b99356b1fbbeb89620667a07917feeb6418a1",
+    "source_id": "github:pingcap/tidb#66601",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1d/1d05667dd3209af965f1527938349b1d5a1cb85b53b1469e2f41e3e7f28cfa69.json b/.cache/impact/classifications/1d/1d05667dd3209af965f1527938349b1d5a1cb85b53b1469e2f41e3e7f28cfa69.json
new file mode 100644
index 0000000..cfcd70e
--- /dev/null
+++ b/.cache/impact/classifications/1d/1d05667dd3209af965f1527938349b1d5a1cb85b53b1469e2f41e3e7f28cfa69.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#138267|sha256:c6b600aed58c34d9461c33978d3edbcc401fb54eb953305ba56f2f8b16f244dc|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c6b600aed58c34d9461c33978d3edbcc401fb54eb953305ba56f2f8b16f244dc",
+    "source_id": "github:cockroachdb/cockroach#138267",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1d/1d980a15bc8f7558ca99cf68cad25d88a9db2efc0ed2872098c3d7acdc890e9e.json b/.cache/impact/classifications/1d/1d980a15bc8f7558ca99cf68cad25d88a9db2efc0ed2872098c3d7acdc890e9e.json
new file mode 100644
index 0000000..1c33bf8
--- /dev/null
+++ b/.cache/impact/classifications/1d/1d980a15bc8f7558ca99cf68cad25d88a9db2efc0ed2872098c3d7acdc890e9e.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:ClickHouse/ClickHouse#72493|sha256:152076d29a30571d81b9dd1f676c56eb1f9abac336e877b27a3814c598c93032|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:152076d29a30571d81b9dd1f676c56eb1f9abac336e877b27a3814c598c93032",
+    "source_id": "github:ClickHouse/ClickHouse#72493",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1d/1da4fd2f9c9bf1209742bfc129712d9b04d79af4840adcb453ba99bbe5e60ac1.json b/.cache/impact/classifications/1d/1da4fd2f9c9bf1209742bfc129712d9b04d79af4840adcb453ba99bbe5e60ac1.json
new file mode 100644
index 0000000..b3c35cf
--- /dev/null
+++ b/.cache/impact/classifications/1d/1da4fd2f9c9bf1209742bfc129712d9b04d79af4840adcb453ba99bbe5e60ac1.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:ydb-platform/ydb#7728|sha256:7ba767c47ae31404497d13f6b84d1985248d14d8f5c429d6303f045070f424fa|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7ba767c47ae31404497d13f6b84d1985248d14d8f5c429d6303f045070f424fa",
+    "source_id": "github:ydb-platform/ydb#7728",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1d/1da702858ce7541595df3a7f7d6906a96d8a818c75f7a3fd86be02c5cd1e361c.json b/.cache/impact/classifications/1d/1da702858ce7541595df3a7f7d6906a96d8a818c75f7a3fd86be02c5cd1e361c.json
new file mode 100644
index 0000000..eeb3d83
--- /dev/null
+++ b/.cache/impact/classifications/1d/1da702858ce7541595df3a7f7d6906a96d8a818c75f7a3fd86be02c5cd1e361c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#144835|sha256:761768f66c33db5aa9ba72c713ff3e7139e8cf549a59a4b294ee3574c64ee5d4|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:761768f66c33db5aa9ba72c713ff3e7139e8cf549a59a4b294ee3574c64ee5d4",
+    "source_id": "github:cockroachdb/cockroach#144835",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/1f/1fc9a19bcbd009be9d31e376883992955dabadcfcc18e65e4681f6d0218115a6.json b/.cache/impact/classifications/1f/1fc9a19bcbd009be9d31e376883992955dabadcfcc18e65e4681f6d0218115a6.json
new file mode 100644
index 0000000..fd129f6
--- /dev/null
+++ b/.cache/impact/classifications/1f/1fc9a19bcbd009be9d31e376883992955dabadcfcc18e65e4681f6d0218115a6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#88486|sha256:d20dfea29bc22c09294fc8b07f575364db89806b275c1952a5ddf6736a79be38|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d20dfea29bc22c09294fc8b07f575364db89806b275c1952a5ddf6736a79be38",
+    "source_id": "github:elastic/elasticsearch#88486",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/20/203557b0bb1ffead7b93ac4c6d4ad5da31019f1066c0a1eb44c8e37b6e136c35.json b/.cache/impact/classifications/20/203557b0bb1ffead7b93ac4c6d4ad5da31019f1066c0a1eb44c8e37b6e136c35.json
new file mode 100644
index 0000000..f899a93
--- /dev/null
+++ b/.cache/impact/classifications/20/203557b0bb1ffead7b93ac4c6d4ad5da31019f1066c0a1eb44c8e37b6e136c35.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#81013|sha256:ec4da6ed9f7e3756982d9ea767b901a0f9df9fdd1de1d5fee0f1ee4df0db3ac3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ec4da6ed9f7e3756982d9ea767b901a0f9df9fdd1de1d5fee0f1ee4df0db3ac3",
+    "source_id": "github:cockroachdb/cockroach#81013",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/23/23540ab9c04e9166e431788cee8c919bee81abf5bf9b2e7b2397e6da5586f339.json b/.cache/impact/classifications/23/23540ab9c04e9166e431788cee8c919bee81abf5bf9b2e7b2397e6da5586f339.json
new file mode 100644
index 0000000..4d2a48a
--- /dev/null
+++ b/.cache/impact/classifications/23/23540ab9c04e9166e431788cee8c919bee81abf5bf9b2e7b2397e6da5586f339.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#84720|sha256:91f506e640992f19dd03911beb356a885aff7c4d4e98858ff61bb3f262c2fbc5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:91f506e640992f19dd03911beb356a885aff7c4d4e98858ff61bb3f262c2fbc5",
+    "source_id": "github:cockroachdb/cockroach#84720",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/23/23d585228e1e9c99720346b9e6fbb99830b6b5b88418faccfb09de51e77f2cf7.json b/.cache/impact/classifications/23/23d585228e1e9c99720346b9e6fbb99830b6b5b88418faccfb09de51e77f2cf7.json
new file mode 100644
index 0000000..644c007
--- /dev/null
+++ b/.cache/impact/classifications/23/23d585228e1e9c99720346b9e6fbb99830b6b5b88418faccfb09de51e77f2cf7.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8030|sha256:f97b52a643a1e82a3d2ca1beee7a9ff203e5391595637dec79711dc549380478|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f97b52a643a1e82a3d2ca1beee7a9ff203e5391595637dec79711dc549380478",
+    "source_id": "github:cockroachdb/cockroach#8030",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/24/2493e2517e747bdda9a8ecc7ddf2ec613395774f0185215de8a602fd926a594b.json b/.cache/impact/classifications/24/2493e2517e747bdda9a8ecc7ddf2ec613395774f0185215de8a602fd926a594b.json
new file mode 100644
index 0000000..f2251ec
--- /dev/null
+++ b/.cache/impact/classifications/24/2493e2517e747bdda9a8ecc7ddf2ec613395774f0185215de8a602fd926a594b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#47210|sha256:9694ed285ef8067b9294a64aa488251a121d91f8aa9b9e56a77f158212b06074|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9694ed285ef8067b9294a64aa488251a121d91f8aa9b9e56a77f158212b06074",
+    "source_id": "github:cockroachdb/cockroach#47210",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/25/257046ba8fb14f1467d3dc8fbc22aad7b0b6fbccd83b4879df02f0b46fca5fc3.json b/.cache/impact/classifications/25/257046ba8fb14f1467d3dc8fbc22aad7b0b6fbccd83b4879df02f0b46fca5fc3.json
new file mode 100644
index 0000000..bd44c52
--- /dev/null
+++ b/.cache/impact/classifications/25/257046ba8fb14f1467d3dc8fbc22aad7b0b6fbccd83b4879df02f0b46fca5fc3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tikv/tikv#4874|sha256:da7c4926182c9c8c01b2e909e81535826b3440b783ef58e230ec452809b831d0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:da7c4926182c9c8c01b2e909e81535826b3440b783ef58e230ec452809b831d0",
+    "source_id": "github:tikv/tikv#4874",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/25/25f4602b8edb4bae06c3f98e4af1da75dfa7d38b4f4bd41b3757320a29d7ea98.json b/.cache/impact/classifications/25/25f4602b8edb4bae06c3f98e4af1da75dfa7d38b4f4bd41b3757320a29d7ea98.json
new file mode 100644
index 0000000..d8d19c3
--- /dev/null
+++ b/.cache/impact/classifications/25/25f4602b8edb4bae06c3f98e4af1da75dfa7d38b4f4bd41b3757320a29d7ea98.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#69327|sha256:705d873685ebcfb9b833519dac764a11bf803fdb51a2fded89458f1730abfe20|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "TLP found a correctness bug"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:705d873685ebcfb9b833519dac764a11bf803fdb51a2fded89458f1730abfe20",
+    "source_id": "github:cockroachdb/cockroach#69327",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/27/27970ddc5e276c40fe4a19437f017033a5151f20ee58dac89d5ff8661542f54c.json b/.cache/impact/classifications/27/27970ddc5e276c40fe4a19437f017033a5151f20ee58dac89d5ff8661542f54c.json
new file mode 100644
index 0000000..32edca7
--- /dev/null
+++ b/.cache/impact/classifications/27/27970ddc5e276c40fe4a19437f017033a5151f20ee58dac89d5ff8661542f54c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#43870|sha256:204fa6f8c33494b29a330d518dbb3ca17e7c83e088db47717be98b192425a277|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:204fa6f8c33494b29a330d518dbb3ca17e7c83e088db47717be98b192425a277",
+    "source_id": "github:cockroachdb/cockroach#43870",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/27/27a4be2a418de4eaf5ffa36f91c0848e2324bb4262df3645002566a4c0c58a76.json b/.cache/impact/classifications/27/27a4be2a418de4eaf5ffa36f91c0848e2324bb4262df3645002566a4c0c58a76.json
new file mode 100644
index 0000000..7e571ca
--- /dev/null
+++ b/.cache/impact/classifications/27/27a4be2a418de4eaf5ffa36f91c0848e2324bb4262df3645002566a4c0c58a76.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#12168|sha256:fa4fed9d50d2343be1fcce104cb6b1fdcde980ab38e5c6f2b69fb77031632ce8|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:fa4fed9d50d2343be1fcce104cb6b1fdcde980ab38e5c6f2b69fb77031632ce8",
+    "source_id": "github:cockroachdb/cockroach#12168",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/27/27ba35f664c555f54309efad55bdce8af25f99a6f976a9f043946c1f6b85e2ae.json b/.cache/impact/classifications/27/27ba35f664c555f54309efad55bdce8af25f99a6f976a9f043946c1f6b85e2ae.json
new file mode 100644
index 0000000..db38dd2
--- /dev/null
+++ b/.cache/impact/classifications/27/27ba35f664c555f54309efad55bdce8af25f99a6f976a9f043946c1f6b85e2ae.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#61087|sha256:fc69ce9537dda1392bfaa7cfaea0159a98d54605658a87adba183926f1106837|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:fc69ce9537dda1392bfaa7cfaea0159a98d54605658a87adba183926f1106837",
+    "source_id": "github:elastic/elasticsearch#61087",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/28/2819cd2a388dd7823a60d0d21e6e85b68451030ecd6a196e1ce64a6c85d98cbd.json b/.cache/impact/classifications/28/2819cd2a388dd7823a60d0d21e6e85b68451030ecd6a196e1ce64a6c85d98cbd.json
new file mode 100644
index 0000000..0671beb
--- /dev/null
+++ b/.cache/impact/classifications/28/2819cd2a388dd7823a60d0d21e6e85b68451030ecd6a196e1ce64a6c85d98cbd.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#34206|sha256:1b870221e5914712df87c043095155462d82698b4be7e9168a7b5b65e534ab41|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1b870221e5914712df87c043095155462d82698b4be7e9168a7b5b65e534ab41",
+    "source_id": "github:pingcap/tidb#34206",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/28/28ce00af063fd29755acb934a9be9d60d0432799d296898a8f2b3de36e4c6230.json b/.cache/impact/classifications/28/28ce00af063fd29755acb934a9be9d60d0432799d296898a8f2b3de36e4c6230.json
new file mode 100644
index 0000000..01cc93b
--- /dev/null
+++ b/.cache/impact/classifications/28/28ce00af063fd29755acb934a9be9d60d0432799d296898a8f2b3de36e4c6230.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#14426|sha256:8a3dc7112044b3b0862377a8cb08dbfa804d4b91e8f3fae784c6c4e8ba91409a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8a3dc7112044b3b0862377a8cb08dbfa804d4b91e8f3fae784c6c4e8ba91409a",
+    "source_id": "github:yugabyte/yugabyte-db#14426",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/29/2962cd01151ad1522e01be3028477bf621d9f70a1c346ebd2c559d37e9d6689b.json b/.cache/impact/classifications/29/2962cd01151ad1522e01be3028477bf621d9f70a1c346ebd2c559d37e9d6689b.json
new file mode 100644
index 0000000..dc10c97
--- /dev/null
+++ b/.cache/impact/classifications/29/2962cd01151ad1522e01be3028477bf621d9f70a1c346ebd2c559d37e9d6689b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#126239|sha256:bc394e38d5d387fb45c78e5e6003b9465f34112a27486158476f5772ad6f4a27|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:bc394e38d5d387fb45c78e5e6003b9465f34112a27486158476f5772ad6f4a27",
+    "source_id": "github:cockroachdb/cockroach#126239",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/29/29774f45cc50aa17c8de2b220c0de298f29167ff468faf09a2a12967c55413bf.json b/.cache/impact/classifications/29/29774f45cc50aa17c8de2b220c0de298f29167ff468faf09a2a12967c55413bf.json
new file mode 100644
index 0000000..4b417ce
--- /dev/null
+++ b/.cache/impact/classifications/29/29774f45cc50aa17c8de2b220c0de298f29167ff468faf09a2a12967c55413bf.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#66706|sha256:aa431027b0ada5b7a834046007339ae61321fb02cd95cc93aeae5f773475dc96|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "colfetcher: incorrect decoding of unique secondary indexes"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The TLP roachtest found this defect; the issue was retitled to name it. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:aa431027b0ada5b7a834046007339ae61321fb02cd95cc93aeae5f773475dc96",
+    "source_id": "github:cockroachdb/cockroach#66706",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2a/2a639b6bdfb34a42843d6297cdb2e478a561e643f7e0e146e85699dfafe16b9f.json b/.cache/impact/classifications/2a/2a639b6bdfb34a42843d6297cdb2e478a561e643f7e0e146e85699dfafe16b9f.json
new file mode 100644
index 0000000..e2983cd
--- /dev/null
+++ b/.cache/impact/classifications/2a/2a639b6bdfb34a42843d6297cdb2e478a561e643f7e0e146e85699dfafe16b9f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#138218|sha256:623998cdce9378891d5a296e5bd6eb978c835c4cae16b4ec23124ff8d0e4319a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:623998cdce9378891d5a296e5bd6eb978c835c4cae16b4ec23124ff8d0e4319a",
+    "source_id": "github:cockroachdb/cockroach#138218",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2a/2abdbc05b56246b5688d4431334a191dcb7c2e6baaa78d23a0aaa74b182681df.json b/.cache/impact/classifications/2a/2abdbc05b56246b5688d4431334a191dcb7c2e6baaa78d23a0aaa74b182681df.json
new file mode 100644
index 0000000..90c2da5
--- /dev/null
+++ b/.cache/impact/classifications/2a/2abdbc05b56246b5688d4431334a191dcb7c2e6baaa78d23a0aaa74b182681df.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#1474|sha256:1e86935b8ee79145f223cbffa574bd75917873ed862324c52722aa50566e0dc0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1e86935b8ee79145f223cbffa574bd75917873ed862324c52722aa50566e0dc0",
+    "source_id": "github:sparq-org/sparq#1474",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2b/2bee8fa11643903d6598d869c6f2520c8e59e2ea2b06823904797bf32769d93a.json b/.cache/impact/classifications/2b/2bee8fa11643903d6598d869c6f2520c8e59e2ea2b06823904797bf32769d93a.json
new file mode 100644
index 0000000..7f7146b
--- /dev/null
+++ b/.cache/impact/classifications/2b/2bee8fa11643903d6598d869c6f2520c8e59e2ea2b06823904797bf32769d93a.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#102661|sha256:d2d5ada837c335c404216046c72b88508db9abe3574d18f8a83cadda56effcb1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d2d5ada837c335c404216046c72b88508db9abe3574d18f8a83cadda56effcb1",
+    "source_id": "github:cockroachdb/cockroach#102661",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2c/2ca3ea002d0fa04127cd4365bf85cd13765352477adb7d81864d2b6b3171d657.json b/.cache/impact/classifications/2c/2ca3ea002d0fa04127cd4365bf85cd13765352477adb7d81864d2b6b3171d657.json
new file mode 100644
index 0000000..a531f71
--- /dev/null
+++ b/.cache/impact/classifications/2c/2ca3ea002d0fa04127cd4365bf85cd13765352477adb7d81864d2b6b3171d657.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:questdb/questdb#433|sha256:0eebd130171ce790290cb647555381062ddbdee2f909750067e246e39bf6698d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0eebd130171ce790290cb647555381062ddbdee2f909750067e246e39bf6698d",
+    "source_id": "github:questdb/questdb#433",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2c/2ce4528996df164df768e7e2eea9d69f01f222d14c53b9f4b362c9f1b59a4fa1.json b/.cache/impact/classifications/2c/2ce4528996df164df768e7e2eea9d69f01f222d14c53b9f4b362c9f1b59a4fa1.json
new file mode 100644
index 0000000..2bf22f1
--- /dev/null
+++ b/.cache/impact/classifications/2c/2ce4528996df164df768e7e2eea9d69f01f222d14c53b9f4b362c9f1b59a4fa1.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:hazelcast/hazelcast#5560|sha256:8d124fde051cd38fd3f41c95b10796575cc04725b0791f3ad7998fc9044d9d06|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8d124fde051cd38fd3f41c95b10796575cc04725b0791f3ad7998fc9044d9d06",
+    "source_id": "github:hazelcast/hazelcast#5560",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2d/2d6bb0e6ad160496b9540dfa9e28225383fa43ab7737dea8f53aaf1361785b79.json b/.cache/impact/classifications/2d/2d6bb0e6ad160496b9540dfa9e28225383fa43ab7737dea8f53aaf1361785b79.json
new file mode 100644
index 0000000..286c893
--- /dev/null
+++ b/.cache/impact/classifications/2d/2d6bb0e6ad160496b9540dfa9e28225383fa43ab7737dea8f53aaf1361785b79.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#113089|sha256:b58eb9a68fe87143376ce43b81014985bf34001e93aabd741103b5d577fcfd41|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b58eb9a68fe87143376ce43b81014985bf34001e93aabd741103b5d577fcfd41",
+    "source_id": "github:cockroachdb/cockroach#113089",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2e/2e57140ec9a6b4f19a6850494ae367062f81c770d4e291235fd1aff4d64c934e.json b/.cache/impact/classifications/2e/2e57140ec9a6b4f19a6850494ae367062f81c770d4e291235fd1aff4d64c934e.json
new file mode 100644
index 0000000..3e24b85
--- /dev/null
+++ b/.cache/impact/classifications/2e/2e57140ec9a6b4f19a6850494ae367062f81c770d4e291235fd1aff4d64c934e.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:yugabyte/yugabyte-db#11144|sha256:c1c8d8a4f8ed51d81fbcdbf0a3b544995da57e74cae2403db30e886cb7357ec7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "[SQLancer][YSQL] It is possible to create a temp table"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": null
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c1c8d8a4f8ed51d81fbcdbf0a3b544995da57e74cae2403db30e886cb7357ec7",
+    "source_id": "github:yugabyte/yugabyte-db#11144",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2e/2eb99af6527366bc85c9d22f557c9ad577b99d02d54e52d308dd89f3be3acb61.json b/.cache/impact/classifications/2e/2eb99af6527366bc85c9d22f557c9ad577b99d02d54e52d308dd89f3be3acb61.json
new file mode 100644
index 0000000..0332aec
--- /dev/null
+++ b/.cache/impact/classifications/2e/2eb99af6527366bc85c9d22f557c9ad577b99d02d54e52d308dd89f3be3acb61.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:ClickHouse/ClickHouse#75541|sha256:9d9f126549a3dbcb4f518491b0f4a70fe9e68fd01e2ebb5fb17abfc9da93a632|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "SQLancer:  No equality condition found in JOIN ON expression"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": null
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9d9f126549a3dbcb4f518491b0f4a70fe9e68fd01e2ebb5fb17abfc9da93a632",
+    "source_id": "github:ClickHouse/ClickHouse#75541",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2f/2f0067431ddf49ac7dbb79bba275f55f9d689e860cb3ea937d58e659b5f4e82a.json b/.cache/impact/classifications/2f/2f0067431ddf49ac7dbb79bba275f55f9d689e860cb3ea937d58e659b5f4e82a.json
new file mode 100644
index 0000000..5dd3f5b
--- /dev/null
+++ b/.cache/impact/classifications/2f/2f0067431ddf49ac7dbb79bba275f55f9d689e860cb3ea937d58e659b5f4e82a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:duckdb/duckdb#5031|sha256:f25a0d37f7bb919715d4d6067e4e1cc48baac881fb3eb20e648f37f9f2d0b46d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f25a0d37f7bb919715d4d6067e4e1cc48baac881fb3eb20e648f37f9f2d0b46d",
+    "source_id": "github:duckdb/duckdb#5031",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2f/2f627059e953077e00b8e998ff16004a1abfc020020d5a2188dc4ddc9a734f19.json b/.cache/impact/classifications/2f/2f627059e953077e00b8e998ff16004a1abfc020020d5a2188dc4ddc9a734f19.json
new file mode 100644
index 0000000..0299927
--- /dev/null
+++ b/.cache/impact/classifications/2f/2f627059e953077e00b8e998ff16004a1abfc020020d5a2188dc4ddc9a734f19.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#37841|sha256:d9400bac615e96a7570e9961f59d41fc6dbf4b9f57b6c15f442ef8bbbd4cdfaa|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d9400bac615e96a7570e9961f59d41fc6dbf4b9f57b6c15f442ef8bbbd4cdfaa",
+    "source_id": "github:elastic/elasticsearch#37841",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/2f/2f696a216c22c769aa42dafd84441bfc34c2a474f23182f5c44110e4cd7eb95d.json b/.cache/impact/classifications/2f/2f696a216c22c769aa42dafd84441bfc34c2a474f23182f5c44110e4cd7eb95d.json
new file mode 100644
index 0000000..6872e1c
--- /dev/null
+++ b/.cache/impact/classifications/2f/2f696a216c22c769aa42dafd84441bfc34c2a474f23182f5c44110e4cd7eb95d.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:RedisGraph/RedisGraph#2931|sha256:c608b96377015ac725efb1c34ddee3330392faafd8d966b625e34a84a99f24b8|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c608b96377015ac725efb1c34ddee3330392faafd8d966b625e34a84a99f24b8",
+    "source_id": "github:RedisGraph/RedisGraph#2931",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/30/3043c6120eddf0520b74edfd0a9af09c0e1a559ea3939328d6df63baeeec4cc9.json b/.cache/impact/classifications/30/3043c6120eddf0520b74edfd0a9af09c0e1a559ea3939328d6df63baeeec4cc9.json
new file mode 100644
index 0000000..48e3090
--- /dev/null
+++ b/.cache/impact/classifications/30/3043c6120eddf0520b74edfd0a9af09c0e1a559ea3939328d6df63baeeec4cc9.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:spiceai/spiceai#2186|sha256:d7de5503a9fb3b89f731fe76b51e1a9eb8860afcbfd3cc0bb6df0177d54cc529|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d7de5503a9fb3b89f731fe76b51e1a9eb8860afcbfd3cc0bb6df0177d54cc529",
+    "source_id": "github:spiceai/spiceai#2186",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/31/31769ce23be6f3a48c2912df04e9b7721a4ff519bd44f3f87aa9a39568e6c577.json b/.cache/impact/classifications/31/31769ce23be6f3a48c2912df04e9b7721a4ff519bd44f3f87aa9a39568e6c577.json
new file mode 100644
index 0000000..3d3aafc
--- /dev/null
+++ b/.cache/impact/classifications/31/31769ce23be6f3a48c2912df04e9b7721a4ff519bd44f3f87aa9a39568e6c577.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#8345|sha256:819317d6324852e7ca7952b6c5aeaa7952d7b8e433038277d3732774ffd1db9b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:819317d6324852e7ca7952b6c5aeaa7952d7b8e433038277d3732774ffd1db9b",
+    "source_id": "github:yugabyte/yugabyte-db#8345",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/32/32351ca3f564c548f77d8c3fb1c40e99ba3489cf17decf1d65164f7a6c81a90b.json b/.cache/impact/classifications/32/32351ca3f564c548f77d8c3fb1c40e99ba3489cf17decf1d65164f7a6c81a90b.json
new file mode 100644
index 0000000..3c0da10
--- /dev/null
+++ b/.cache/impact/classifications/32/32351ca3f564c548f77d8c3fb1c40e99ba3489cf17decf1d65164f7a6c81a90b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#12006|sha256:de2bf1985f07ad144bd23e5f6b08e62020d864c5a044a330dda9f5d096586233|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:de2bf1985f07ad144bd23e5f6b08e62020d864c5a044a330dda9f5d096586233",
+    "source_id": "github:yugabyte/yugabyte-db#12006",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/32/325c43ee4d0297fc1892bbc12ff4075434acf86ef93ae8dfd013c56e0c1e6d46.json b/.cache/impact/classifications/32/325c43ee4d0297fc1892bbc12ff4075434acf86ef93ae8dfd013c56e0c1e6d46.json
new file mode 100644
index 0000000..e53dbd1
--- /dev/null
+++ b/.cache/impact/classifications/32/325c43ee4d0297fc1892bbc12ff4075434acf86ef93ae8dfd013c56e0c1e6d46.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#7238|sha256:09ab012685cfb2422a72d88414a9c179cd009298bb0b56f2af51d19ff55c12f9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:09ab012685cfb2422a72d88414a9c179cd009298bb0b56f2af51d19ff55c12f9",
+    "source_id": "github:cockroachdb/cockroach#7238",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/33/330331d5bb137b422485c0dcd3c58668b14904ece7180035651f5875f8beb339.json b/.cache/impact/classifications/33/330331d5bb137b422485c0dcd3c58668b14904ece7180035651f5875f8beb339.json
new file mode 100644
index 0000000..b36932f
--- /dev/null
+++ b/.cache/impact/classifications/33/330331d5bb137b422485c0dcd3c58668b14904ece7180035651f5875f8beb339.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:risingwavelabs/risingwave#7504|sha256:019062e809491610a5c23c7d71cf55a719f79de4230d5147b92201a1fe741ce2|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:019062e809491610a5c23c7d71cf55a719f79de4230d5147b92201a1fe741ce2",
+    "source_id": "github:risingwavelabs/risingwave#7504",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/33/3336529e53c2ebc93e917cacd622dc40a006d6aac49ba49c723cd9e286078bbb.json b/.cache/impact/classifications/33/3336529e53c2ebc93e917cacd622dc40a006d6aac49ba49c723cd9e286078bbb.json
new file mode 100644
index 0000000..58eb1b6
--- /dev/null
+++ b/.cache/impact/classifications/33/3336529e53c2ebc93e917cacd622dc40a006d6aac49ba49c723cd9e286078bbb.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:kyzobuild/kyzo#376|sha256:8d06c5deb926819c1e90a97f29833d955081cdc66019e0bfd703d7e293c5804d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8d06c5deb926819c1e90a97f29833d955081cdc66019e0bfd703d7e293c5804d",
+    "source_id": "github:kyzobuild/kyzo#376",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/33/334bb3a4c8077dd5796a3c9ca7a32da6854a804f39fcaebf23f18a7d56755fb8.json b/.cache/impact/classifications/33/334bb3a4c8077dd5796a3c9ca7a32da6854a804f39fcaebf23f18a7d56755fb8.json
new file mode 100644
index 0000000..b18e442
--- /dev/null
+++ b/.cache/impact/classifications/33/334bb3a4c8077dd5796a3c9ca7a32da6854a804f39fcaebf23f18a7d56755fb8.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#71675|sha256:4266373f523b06d257d7d2372ca41cb9a771ccba74d258ba1209dea7e2c88431|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4266373f523b06d257d7d2372ca41cb9a771ccba74d258ba1209dea7e2c88431",
+    "source_id": "github:cockroachdb/cockroach#71675",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/34/34598887436178b53a11b458c494d3497a7f0e572ec5d418852682cb3874ee60.json b/.cache/impact/classifications/34/34598887436178b53a11b458c494d3497a7f0e572ec5d418852682cb3874ee60.json
new file mode 100644
index 0000000..33b614c
--- /dev/null
+++ b/.cache/impact/classifications/34/34598887436178b53a11b458c494d3497a7f0e572ec5d418852682cb3874ee60.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#31376|sha256:cbb2c64e02da72f7e4d39abb803df9fc14b7d8f9d2321e43aa2bcc34095da7cf|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:cbb2c64e02da72f7e4d39abb803df9fc14b7d8f9d2321e43aa2bcc34095da7cf",
+    "source_id": "github:pingcap/tidb#31376",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/34/34b5174058c215c84c0f33d4975c8431a3843b099f5e9fc87b7a6e95abf81b19.json b/.cache/impact/classifications/34/34b5174058c215c84c0f33d4975c8431a3843b099f5e9fc87b7a6e95abf81b19.json
new file mode 100644
index 0000000..947cea0
--- /dev/null
+++ b/.cache/impact/classifications/34/34b5174058c215c84c0f33d4975c8431a3843b099f5e9fc87b7a6e95abf81b19.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#170925|sha256:529e03536149683d52f507df25f23ffbf1fee8beef54da5ced053fa28ac0b659|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:529e03536149683d52f507df25f23ffbf1fee8beef54da5ced053fa28ac0b659",
+    "source_id": "github:cockroachdb/cockroach#170925",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/34/34bf3d7bdb37acf3e6be0cff646f63e34e4f0f43326d2f1d7d8aae77634be74c.json b/.cache/impact/classifications/34/34bf3d7bdb37acf3e6be0cff646f63e34e4f0f43326d2f1d7d8aae77634be74c.json
new file mode 100644
index 0000000..a7412da
--- /dev/null
+++ b/.cache/impact/classifications/34/34bf3d7bdb37acf3e6be0cff646f63e34e4f0f43326d2f1d7d8aae77634be74c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#78905|sha256:72edfc8d88676d81a3bf3ec04f941258f35138f274a41cce070fd9cfa189eb77|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:72edfc8d88676d81a3bf3ec04f941258f35138f274a41cce070fd9cfa189eb77",
+    "source_id": "github:cockroachdb/cockroach#78905",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/34/34e7c2630373aa6c27230387dc88b61f7b2f6c1348437a62b3d0ceeaa7eb5ccb.json b/.cache/impact/classifications/34/34e7c2630373aa6c27230387dc88b61f7b2f6c1348437a62b3d0ceeaa7eb5ccb.json
new file mode 100644
index 0000000..8886cf4
--- /dev/null
+++ b/.cache/impact/classifications/34/34e7c2630373aa6c27230387dc88b61f7b2f6c1348437a62b3d0ceeaa7eb5ccb.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#92069|sha256:42406c562771390495a2cd6c30f5f46157b9ca5e6add02366383f200800e7ad8|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:42406c562771390495a2cd6c30f5f46157b9ca5e6add02366383f200800e7ad8",
+    "source_id": "github:cockroachdb/cockroach#92069",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/35/350a9bba7f6cfcab64e3c6f18d353572a6edc5aea2a5326109d2e9267e882c46.json b/.cache/impact/classifications/35/350a9bba7f6cfcab64e3c6f18d353572a6edc5aea2a5326109d2e9267e882c46.json
new file mode 100644
index 0000000..e39b73d
--- /dev/null
+++ b/.cache/impact/classifications/35/350a9bba7f6cfcab64e3c6f18d353572a6edc5aea2a5326109d2e9267e882c46.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#48011|sha256:16eedde213fecd46bcfb5b3e738da33dbc0297d07a019f017c5fdc8369d1198c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:16eedde213fecd46bcfb5b3e738da33dbc0297d07a019f017c5fdc8369d1198c",
+    "source_id": "github:cockroachdb/cockroach#48011",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/35/354b3ca7f084208b733d45120a4ee6700692d26f6e80ed11681c83af1355786f.json b/.cache/impact/classifications/35/354b3ca7f084208b733d45120a4ee6700692d26f6e80ed11681c83af1355786f.json
new file mode 100644
index 0000000..d96e145
--- /dev/null
+++ b/.cache/impact/classifications/35/354b3ca7f084208b733d45120a4ee6700692d26f6e80ed11681c83af1355786f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#42377|sha256:31b845f97fd6efe14879bb431dd40fcd4398d1d1a2c47555ca6c54267283c445|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:31b845f97fd6efe14879bb431dd40fcd4398d1d1a2c47555ca6c54267283c445",
+    "source_id": "github:cockroachdb/cockroach#42377",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/35/358240121ac85c1fc490a373133f4bb84e386638da8687d8a672ec040872934f.json b/.cache/impact/classifications/35/358240121ac85c1fc490a373133f4bb84e386638da8687d8a672ec040872934f.json
new file mode 100644
index 0000000..320d01f
--- /dev/null
+++ b/.cache/impact/classifications/35/358240121ac85c1fc490a373133f4bb84e386638da8687d8a672ec040872934f.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#136400|sha256:bc24adf473c81a1acd8ed5d17d9f5700a424dd449fb8928be9d915c31e0bc6e5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:bc24adf473c81a1acd8ed5d17d9f5700a424dd449fb8928be9d915c31e0bc6e5",
+    "source_id": "github:cockroachdb/cockroach#136400",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/36/3696329a0479dd1c62dbb91c31c7dc0983296a09eb47b94c778070dd4d6beb0a.json b/.cache/impact/classifications/36/3696329a0479dd1c62dbb91c31c7dc0983296a09eb47b94c778070dd4d6beb0a.json
new file mode 100644
index 0000000..a9b215e
--- /dev/null
+++ b/.cache/impact/classifications/36/3696329a0479dd1c62dbb91c31c7dc0983296a09eb47b94c778070dd4d6beb0a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tikv/tikv#13959|sha256:dbb2bbd80a8955bb24e004d7c0d3e718830e72f09a434da0b3a22e8d87086e6f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:dbb2bbd80a8955bb24e004d7c0d3e718830e72f09a434da0b3a22e8d87086e6f",
+    "source_id": "github:tikv/tikv#13959",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/36/36c679c5cdfd4fc072f1cc0f39e60e131b3edcd61e1b315fb04f9ca7b1f83999.json b/.cache/impact/classifications/36/36c679c5cdfd4fc072f1cc0f39e60e131b3edcd61e1b315fb04f9ca7b1f83999.json
new file mode 100644
index 0000000..3f2b221
--- /dev/null
+++ b/.cache/impact/classifications/36/36c679c5cdfd4fc072f1cc0f39e60e131b3edcd61e1b315fb04f9ca7b1f83999.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#104381|sha256:d96c83ae88bd1c08a88a90091702393024120d993458f1464a479f3ccf0ee9e6|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d96c83ae88bd1c08a88a90091702393024120d993458f1464a479f3ccf0ee9e6",
+    "source_id": "github:cockroachdb/cockroach#104381",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/37/37121406943b592e568d06b71d6e087d5904b0848e2472cfb89264fc6c6b1323.json b/.cache/impact/classifications/37/37121406943b592e568d06b71d6e087d5904b0848e2472cfb89264fc6c6b1323.json
new file mode 100644
index 0000000..2cfaa57
--- /dev/null
+++ b/.cache/impact/classifications/37/37121406943b592e568d06b71d6e087d5904b0848e2472cfb89264fc6c6b1323.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:databendlabs/databend#8238|sha256:7ff5718ca197f0da61f4ceed4bbfaa6b9f7a6bd30133d00cb78956594a6816fe|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7ff5718ca197f0da61f4ceed4bbfaa6b9f7a6bd30133d00cb78956594a6816fe",
+    "source_id": "github:databendlabs/databend#8238",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/37/374ab22691ed577bf78e8df5ed8982b044b4a62ed1f529087f888fd8099cffb7.json b/.cache/impact/classifications/37/374ab22691ed577bf78e8df5ed8982b044b4a62ed1f529087f888fd8099cffb7.json
new file mode 100644
index 0000000..e69776e
--- /dev/null
+++ b/.cache/impact/classifications/37/374ab22691ed577bf78e8df5ed8982b044b4a62ed1f529087f888fd8099cffb7.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#96215|sha256:44ff1fda583e80d6342eca42b8633e517f956d7cb21cedaad1cb613f1d257534|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:44ff1fda583e80d6342eca42b8633e517f956d7cb21cedaad1cb613f1d257534",
+    "source_id": "github:cockroachdb/cockroach#96215",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/37/3769ee162a03a4e359e831d82e86e0b345bcb74e535a05a6ecad906fe0712247.json b/.cache/impact/classifications/37/3769ee162a03a4e359e831d82e86e0b345bcb74e535a05a6ecad906fe0712247.json
new file mode 100644
index 0000000..a758a46
--- /dev/null
+++ b/.cache/impact/classifications/37/3769ee162a03a4e359e831d82e86e0b345bcb74e535a05a6ecad906fe0712247.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#84957|sha256:4b600fc361ea68147fa24ed77777469aa80e4ba60c61787a5aad751c9eb71c57|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4b600fc361ea68147fa24ed77777469aa80e4ba60c61787a5aad751c9eb71c57",
+    "source_id": "github:cockroachdb/cockroach#84957",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/37/37a16251babed2e362f96803ac5f3b54a466c3533b1c0cd71af045f59129201c.json b/.cache/impact/classifications/37/37a16251babed2e362f96803ac5f3b54a466c3533b1c0cd71af045f59129201c.json
new file mode 100644
index 0000000..66e4d4f
--- /dev/null
+++ b/.cache/impact/classifications/37/37a16251babed2e362f96803ac5f3b54a466c3533b1c0cd71af045f59129201c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#95263|sha256:dc7b22a564ba9a68a010d7a2fb302f5dbb7708d07e30e9df0b38d647292f4a39|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:dc7b22a564ba9a68a010d7a2fb302f5dbb7708d07e30e9df0b38d647292f4a39",
+    "source_id": "github:cockroachdb/cockroach#95263",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/37/37e7d8733b11d19c49beba9b422a09f461dcaab7dc097878b7feee9ef20b219c.json b/.cache/impact/classifications/37/37e7d8733b11d19c49beba9b422a09f461dcaab7dc097878b7feee9ef20b219c.json
new file mode 100644
index 0000000..b60f7b5
--- /dev/null
+++ b/.cache/impact/classifications/37/37e7d8733b11d19c49beba9b422a09f461dcaab7dc097878b7feee9ef20b219c.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#154421|sha256:f7305bbe0bcf7d90094548972eec373fe2f89d779bd7ed9499bbb8954179008d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f7305bbe0bcf7d90094548972eec373fe2f89d779bd7ed9499bbb8954179008d",
+    "source_id": "github:cockroachdb/cockroach#154421",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/38/38318f9ef674a3bf674b3fcc00d9be784d89919691dfcae55cdec8aeb0140de9.json b/.cache/impact/classifications/38/38318f9ef674a3bf674b3fcc00d9be784d89919691dfcae55cdec8aeb0140de9.json
new file mode 100644
index 0000000..75e8a00
--- /dev/null
+++ b/.cache/impact/classifications/38/38318f9ef674a3bf674b3fcc00d9be784d89919691dfcae55cdec8aeb0140de9.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:spiceai/spiceai#2188|sha256:7184ad96d0aead1371dc5c7d4a759deaaab19056fdc2400cfa238301f68f03c5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7184ad96d0aead1371dc5c7d4a759deaaab19056fdc2400cfa238301f68f03c5",
+    "source_id": "github:spiceai/spiceai#2188",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/38/38a33126a7014ac8deacb5324e160761e2794d024dd0580198fa8fbaedb154a9.json b/.cache/impact/classifications/38/38a33126a7014ac8deacb5324e160761e2794d024dd0580198fa8fbaedb154a9.json
new file mode 100644
index 0000000..6587d8f
--- /dev/null
+++ b/.cache/impact/classifications/38/38a33126a7014ac8deacb5324e160761e2794d024dd0580198fa8fbaedb154a9.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#92718|sha256:7019c6cfbfe293cd9ff2e588f961ec504855a7bcc09296e30107b2d5329b17ff|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7019c6cfbfe293cd9ff2e588f961ec504855a7bcc09296e30107b2d5329b17ff",
+    "source_id": "github:cockroachdb/cockroach#92718",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/39/394719e5863c3928987760ce4a4b50653eb060ae545efa8804db985d35c1ddce.json b/.cache/impact/classifications/39/394719e5863c3928987760ce4a4b50653eb060ae545efa8804db985d35c1ddce.json
new file mode 100644
index 0000000..cc87929
--- /dev/null
+++ b/.cache/impact/classifications/39/394719e5863c3928987760ce4a4b50653eb060ae545efa8804db985d35c1ddce.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#3054|sha256:38b0e8c9d9a507fb038371270278bedfc99ff626e7a49fa9a0ee39bf82a38a55|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:38b0e8c9d9a507fb038371270278bedfc99ff626e7a49fa9a0ee39bf82a38a55",
+    "source_id": "github:cockroachdb/cockroach#3054",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/39/394ed6d2db0422909f821a9ace9824a5990fd11dc11359bd6dc53f91b861359f.json b/.cache/impact/classifications/39/394ed6d2db0422909f821a9ace9824a5990fd11dc11359bd6dc53f91b861359f.json
new file mode 100644
index 0000000..143e388
--- /dev/null
+++ b/.cache/impact/classifications/39/394ed6d2db0422909f821a9ace9824a5990fd11dc11359bd6dc53f91b861359f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#9151|sha256:2fe98d82313c4606c2d898d4da935d172de8cecbbd1f6155b4b605a1d2093f7c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2fe98d82313c4606c2d898d4da935d172de8cecbbd1f6155b4b605a1d2093f7c",
+    "source_id": "github:cockroachdb/cockroach#9151",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/39/399a6685b1439f37644978e3c87427f046aedf49e0e5361180e79842e51e76b1.json b/.cache/impact/classifications/39/399a6685b1439f37644978e3c87427f046aedf49e0e5361180e79842e51e76b1.json
new file mode 100644
index 0000000..02521bc
--- /dev/null
+++ b/.cache/impact/classifications/39/399a6685b1439f37644978e3c87427f046aedf49e0e5361180e79842e51e76b1.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#120903|sha256:1d0f8b5cbe15ca55644d47e84036ba683e7dd79b0987bff42311900e64b79c6f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1d0f8b5cbe15ca55644d47e84036ba683e7dd79b0987bff42311900e64b79c6f",
+    "source_id": "github:elastic/elasticsearch#120903",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/39/39c933270f2736392758208a8ba57b8ee8d71a0ef0bab104e4b6ad739187858c.json b/.cache/impact/classifications/39/39c933270f2736392758208a8ba57b8ee8d71a0ef0bab104e4b6ad739187858c.json
new file mode 100644
index 0000000..b1a045b
--- /dev/null
+++ b/.cache/impact/classifications/39/39c933270f2736392758208a8ba57b8ee8d71a0ef0bab104e4b6ad739187858c.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#159137|sha256:a124f6d857ea172e043e1f50734df05a3200035e864865f577f1baf83e4eddc4|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a124f6d857ea172e043e1f50734df05a3200035e864865f577f1baf83e4eddc4",
+    "source_id": "github:cockroachdb/cockroach#159137",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/39/39e488e79ddb6819d4549d27cd33cae106b1e3e1282898ca594959a9a9581559.json b/.cache/impact/classifications/39/39e488e79ddb6819d4549d27cd33cae106b1e3e1282898ca594959a9a9581559.json
new file mode 100644
index 0000000..9c01b41
--- /dev/null
+++ b/.cache/impact/classifications/39/39e488e79ddb6819d4549d27cd33cae106b1e3e1282898ca594959a9a9581559.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#43729|sha256:b3899c0c284b68e72a8874dab2bd714bcd40c163a6e30965ac95f5c0e87d2e53|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b3899c0c284b68e72a8874dab2bd714bcd40c163a6e30965ac95f5c0e87d2e53",
+    "source_id": "github:cockroachdb/cockroach#43729",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/3a/3ab3f1e9f6afd523df60713fadb8039eba5c773095d86e8adeb0c18383569e5b.json b/.cache/impact/classifications/3a/3ab3f1e9f6afd523df60713fadb8039eba5c773095d86e8adeb0c18383569e5b.json
new file mode 100644
index 0000000..23acab5
--- /dev/null
+++ b/.cache/impact/classifications/3a/3ab3f1e9f6afd523df60713fadb8039eba5c773095d86e8adeb0c18383569e5b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:apache/datafusion#21076|sha256:e190073530e7654bc9278d5ace789e28ca2790f2f9a1452961a6d0814b9e50d1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e190073530e7654bc9278d5ace789e28ca2790f2f9a1452961a6d0814b9e50d1",
+    "source_id": "github:apache/datafusion#21076",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/3b/3b0905ebb8f157755762ae4bc668a710fcce367d8c911889e1dc045d71d2c084.json b/.cache/impact/classifications/3b/3b0905ebb8f157755762ae4bc668a710fcce367d8c911889e1dc045d71d2c084.json
new file mode 100644
index 0000000..40680ec
--- /dev/null
+++ b/.cache/impact/classifications/3b/3b0905ebb8f157755762ae4bc668a710fcce367d8c911889e1dc045d71d2c084.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:duckdb/duckdb#22025|sha256:aa76891323505377a6bcb7353a8f9f45e34d54d5073eff2d74eed22775e2bdd0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "in the form of TLP is inconsistent with the result of the original sql execution"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:aa76891323505377a6bcb7353a8f9f45e34d54d5073eff2d74eed22775e2bdd0",
+    "source_id": "github:duckdb/duckdb#22025",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/3d/3d3e80e814bc3f6c8863de81cfe86a6b1276ae75b43e3b2617a9623ca765a938.json b/.cache/impact/classifications/3d/3d3e80e814bc3f6c8863de81cfe86a6b1276ae75b43e3b2617a9623ca765a938.json
new file mode 100644
index 0000000..5e1caf5
--- /dev/null
+++ b/.cache/impact/classifications/3d/3d3e80e814bc3f6c8863de81cfe86a6b1276ae75b43e3b2617a9623ca765a938.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:crate/crate#4921|sha256:ea2cb02cdcd1a1a5c5c3a78b4ac99c2fce1e7d84040b1ce396526645a2117cdc|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ea2cb02cdcd1a1a5c5c3a78b4ac99c2fce1e7d84040b1ce396526645a2117cdc",
+    "source_id": "github:crate/crate#4921",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/3d/3df58d9f4baf155d3760ce531aa04f199cf62840af483eab81a2ec93bbc90eeb.json b/.cache/impact/classifications/3d/3df58d9f4baf155d3760ce531aa04f199cf62840af483eab81a2ec93bbc90eeb.json
new file mode 100644
index 0000000..94e2f97
--- /dev/null
+++ b/.cache/impact/classifications/3d/3df58d9f4baf155d3760ce531aa04f199cf62840af483eab81a2ec93bbc90eeb.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#13805|sha256:bba8495e6073f849251b3fc871238d99bd34bebb68f1c300113c4294e513c6c3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:bba8495e6073f849251b3fc871238d99bd34bebb68f1c300113c4294e513c6c3",
+    "source_id": "github:cockroachdb/cockroach#13805",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/3e/3e5b2501fecc172646dda868bc823efd3603da8b6d3033597cf02ec319d51ff4.json b/.cache/impact/classifications/3e/3e5b2501fecc172646dda868bc823efd3603da8b6d3033597cf02ec319d51ff4.json
new file mode 100644
index 0000000..70a9b36
--- /dev/null
+++ b/.cache/impact/classifications/3e/3e5b2501fecc172646dda868bc823efd3603da8b6d3033597cf02ec319d51ff4.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#114065|sha256:a54c878bfd0108d58914aa2a662aaf736bb11854d4ef3ee21eb5a24b5c626531|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a54c878bfd0108d58914aa2a662aaf736bb11854d4ef3ee21eb5a24b5c626531",
+    "source_id": "github:cockroachdb/cockroach#114065",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/3e/3ed6d135ca67d5d9402164156312d2b32c9766927dbfacfe68e68e8200fe3fbf.json b/.cache/impact/classifications/3e/3ed6d135ca67d5d9402164156312d2b32c9766927dbfacfe68e68e8200fe3fbf.json
new file mode 100644
index 0000000..c7ffd47
--- /dev/null
+++ b/.cache/impact/classifications/3e/3ed6d135ca67d5d9402164156312d2b32c9766927dbfacfe68e68e8200fe3fbf.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#144774|sha256:2f294ff82c444b74db92dfd1e3e12caa5cd444c40e898747af6b5a4688e3f7a8|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2f294ff82c444b74db92dfd1e3e12caa5cd444c40e898747af6b5a4688e3f7a8",
+    "source_id": "github:cockroachdb/cockroach#144774",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/3e/3eecfca54d3cf7c233bc5f451ceed197ded3133f484f4fced09e20644a4efe8c.json b/.cache/impact/classifications/3e/3eecfca54d3cf7c233bc5f451ceed197ded3133f484f4fced09e20644a4efe8c.json
new file mode 100644
index 0000000..8035a59
--- /dev/null
+++ b/.cache/impact/classifications/3e/3eecfca54d3cf7c233bc5f451ceed197ded3133f484f4fced09e20644a4efe8c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4635|sha256:d64724b99d1dcce5d74e19f89495828f96b73957575c5cd5b5cd97f7b6f02a90|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d64724b99d1dcce5d74e19f89495828f96b73957575c5cd5b5cd97f7b6f02a90",
+    "source_id": "github:sparq-org/sparq#4635",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/3f/3f0a4beeb5c10faa6f3b446799db8c6a3333f6237c4c51bbf513a277054a8f84.json b/.cache/impact/classifications/3f/3f0a4beeb5c10faa6f3b446799db8c6a3333f6237c4c51bbf513a277054a8f84.json
new file mode 100644
index 0000000..4a7a6da
--- /dev/null
+++ b/.cache/impact/classifications/3f/3f0a4beeb5c10faa6f3b446799db8c6a3333f6237c4c51bbf513a277054a8f84.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#123695|sha256:7359de1bbf204481958f5fd17b0735ad14e5a3e6a6c0930e2ad17433948dfc8b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7359de1bbf204481958f5fd17b0735ad14e5a3e6a6c0930e2ad17433948dfc8b",
+    "source_id": "github:cockroachdb/cockroach#123695",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/3f/3f873111b12e1bf3d360ef43c66fecd65f0f63dda21b35d5829ace408ce6d3f5.json b/.cache/impact/classifications/3f/3f873111b12e1bf3d360ef43c66fecd65f0f63dda21b35d5829ace408ce6d3f5.json
new file mode 100644
index 0000000..8e95f68
--- /dev/null
+++ b/.cache/impact/classifications/3f/3f873111b12e1bf3d360ef43c66fecd65f0f63dda21b35d5829ace408ce6d3f5.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#79992|sha256:927cd6ad1310aa19becea5df875ef77ba93a6b39a20bc34a92678bfcbf91a1f5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:927cd6ad1310aa19becea5df875ef77ba93a6b39a20bc34a92678bfcbf91a1f5",
+    "source_id": "github:cockroachdb/cockroach#79992",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/3f/3f9eb764034e417513cb7d5afbec8a64297ed48a727ba96c27b2171c483040ad.json b/.cache/impact/classifications/3f/3f9eb764034e417513cb7d5afbec8a64297ed48a727ba96c27b2171c483040ad.json
new file mode 100644
index 0000000..9a0caa8
--- /dev/null
+++ b/.cache/impact/classifications/3f/3f9eb764034e417513cb7d5afbec8a64297ed48a727ba96c27b2171c483040ad.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#16381|sha256:6a1982daad58a528b647d211c2d8f5eca1ac966aa9ec9b8a4fd427a826e123f3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6a1982daad58a528b647d211c2d8f5eca1ac966aa9ec9b8a4fd427a826e123f3",
+    "source_id": "github:pingcap/tidb#16381",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/40/404131a968214973c44cb927ce36c02ee7c6f26c9ced31ae4d2b6531880065d3.json b/.cache/impact/classifications/40/404131a968214973c44cb927ce36c02ee7c6f26c9ced31ae4d2b6531880065d3.json
new file mode 100644
index 0000000..378ec01
--- /dev/null
+++ b/.cache/impact/classifications/40/404131a968214973c44cb927ce36c02ee7c6f26c9ced31ae4d2b6531880065d3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:spiceai/spiceai#10832|sha256:5f9f482feadf4c9d3a050d467b450e10bbe48e3f488663ac497940f910de2ace|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5f9f482feadf4c9d3a050d467b450e10bbe48e3f488663ac497940f910de2ace",
+    "source_id": "github:spiceai/spiceai#10832",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/42/42e8227089f49fc990c670f71cfb7833f4274ee9bc98dc300969e8822bb84cfa.json b/.cache/impact/classifications/42/42e8227089f49fc990c670f71cfb7833f4274ee9bc98dc300969e8822bb84cfa.json
new file mode 100644
index 0000000..4b6ef64
--- /dev/null
+++ b/.cache/impact/classifications/42/42e8227089f49fc990c670f71cfb7833f4274ee9bc98dc300969e8822bb84cfa.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#37144|sha256:baa76d74047c24b1e1c365c176aceb1829a8efbb1ead17e005885573bed967b6|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:baa76d74047c24b1e1c365c176aceb1829a8efbb1ead17e005885573bed967b6",
+    "source_id": "github:cockroachdb/cockroach#37144",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/43/43514fd76182f1615f73e1ab81e95aae36b40e108c2860f13839334ca2bb4df6.json b/.cache/impact/classifications/43/43514fd76182f1615f73e1ab81e95aae36b40e108c2860f13839334ca2bb4df6.json
new file mode 100644
index 0000000..9ca3a50
--- /dev/null
+++ b/.cache/impact/classifications/43/43514fd76182f1615f73e1ab81e95aae36b40e108c2860f13839334ca2bb4df6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#144869|sha256:63365c4ea107abf646ab8d50d33132eccddf973109aa45d49f25a9367497d797|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:63365c4ea107abf646ab8d50d33132eccddf973109aa45d49f25a9367497d797",
+    "source_id": "github:cockroachdb/cockroach#144869",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/43/43b89d2e810e5ebb83b49e62d80afb7191b2b9f8f9cc5372dd8ee52cf4db6149.json b/.cache/impact/classifications/43/43b89d2e810e5ebb83b49e62d80afb7191b2b9f8f9cc5372dd8ee52cf4db6149.json
new file mode 100644
index 0000000..b989d36
--- /dev/null
+++ b/.cache/impact/classifications/43/43b89d2e810e5ebb83b49e62d80afb7191b2b9f8f9cc5372dd8ee52cf4db6149.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#109498|sha256:637850424559c190f571ef70b13b4c6dad67702c6edc4b82f0043a5944832d65|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:637850424559c190f571ef70b13b4c6dad67702c6edc4b82f0043a5944832d65",
+    "source_id": "github:cockroachdb/cockroach#109498",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/44/4455c264cd2c05e749bf9061611dc3ab1565301c1e694be63372626798ca0e7d.json b/.cache/impact/classifications/44/4455c264cd2c05e749bf9061611dc3ab1565301c1e694be63372626798ca0e7d.json
new file mode 100644
index 0000000..6bbddb6
--- /dev/null
+++ b/.cache/impact/classifications/44/4455c264cd2c05e749bf9061611dc3ab1565301c1e694be63372626798ca0e7d.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#80047|sha256:722a193f287bebd26f0c7058a744174353e55201f2c34a07740eaafb6e0c6991|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:722a193f287bebd26f0c7058a744174353e55201f2c34a07740eaafb6e0c6991",
+    "source_id": "github:cockroachdb/cockroach#80047",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/44/44838e68a270ebd04d7547e84a8eecfba734fdd70ad880908bba0a4d4d15ee09.json b/.cache/impact/classifications/44/44838e68a270ebd04d7547e84a8eecfba734fdd70ad880908bba0a4d4d15ee09.json
new file mode 100644
index 0000000..da700e0
--- /dev/null
+++ b/.cache/impact/classifications/44/44838e68a270ebd04d7547e84a8eecfba734fdd70ad880908bba0a4d4d15ee09.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:ClickHouse/ClickHouse#104203|sha256:45b0539a555705486efe114186076cfc46d2d352fd6f9861133742a50279c3c1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "the `UNION ALL` of three TLP partitions"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:45b0539a555705486efe114186076cfc46d2d352fd6f9861133742a50279c3c1",
+    "source_id": "github:ClickHouse/ClickHouse#104203",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/44/44effe4caa65bd4046307e558957090ca44711c4b5893dd639571392639c4221.json b/.cache/impact/classifications/44/44effe4caa65bd4046307e558957090ca44711c4b5893dd639571392639c4221.json
new file mode 100644
index 0000000..376337f
--- /dev/null
+++ b/.cache/impact/classifications/44/44effe4caa65bd4046307e558957090ca44711c4b5893dd639571392639c4221.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#19357|sha256:71be9b03d3ccc940908cf795dc64b5df1f6c08d377af6360a7a3478e668b843d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:71be9b03d3ccc940908cf795dc64b5df1f6c08d377af6360a7a3478e668b843d",
+    "source_id": "github:cockroachdb/cockroach#19357",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/44/44f0bd930ab5ee79571013d4f5038600b327d16c7329f7049a54b9b194b2e686.json b/.cache/impact/classifications/44/44f0bd930ab5ee79571013d4f5038600b327d16c7329f7049a54b9b194b2e686.json
new file mode 100644
index 0000000..9d587ce
--- /dev/null
+++ b/.cache/impact/classifications/44/44f0bd930ab5ee79571013d4f5038600b327d16c7329f7049a54b9b194b2e686.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#83584|sha256:63b65112278c9d12b0dfb37cbe2fe06d6263bac4086bf29020bde18a39386d66|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:63b65112278c9d12b0dfb37cbe2fe06d6263bac4086bf29020bde18a39386d66",
+    "source_id": "github:cockroachdb/cockroach#83584",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/45/45f01122322428a4a78238c95ab408e693c9b10bfc023463d64235fbf0424709.json b/.cache/impact/classifications/45/45f01122322428a4a78238c95ab408e693c9b10bfc023463d64235fbf0424709.json
new file mode 100644
index 0000000..5a43cc9
--- /dev/null
+++ b/.cache/impact/classifications/45/45f01122322428a4a78238c95ab408e693c9b10bfc023463d64235fbf0424709.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#17842|sha256:0c74079105ac52b2ccca50769f1528adbd8603c0170bc06eb798dd895c3019ed|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0c74079105ac52b2ccca50769f1528adbd8603c0170bc06eb798dd895c3019ed",
+    "source_id": "github:cockroachdb/cockroach#17842",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/46/4605a68f426f556226e8235beb21ade8e223fb6225b14bb5dceecfc436afa6e6.json b/.cache/impact/classifications/46/4605a68f426f556226e8235beb21ade8e223fb6225b14bb5dceecfc436afa6e6.json
new file mode 100644
index 0000000..022c9e0
--- /dev/null
+++ b/.cache/impact/classifications/46/4605a68f426f556226e8235beb21ade8e223fb6225b14bb5dceecfc436afa6e6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#2421|sha256:043245f8a75a736dfe52d55ef307371c614aedda97323572650f01654ec058a7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:043245f8a75a736dfe52d55ef307371c614aedda97323572650f01654ec058a7",
+    "source_id": "github:cockroachdb/cockroach#2421",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/46/46a48522814fdf7f0399d758483e770ad376d1eea4bc6c3f1cbd7d4464e234d3.json b/.cache/impact/classifications/46/46a48522814fdf7f0399d758483e770ad376d1eea4bc6c3f1cbd7d4464e234d3.json
new file mode 100644
index 0000000..ce73f3d
--- /dev/null
+++ b/.cache/impact/classifications/46/46a48522814fdf7f0399d758483e770ad376d1eea4bc6c3f1cbd7d4464e234d3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#2553|sha256:0ffeb78b965f82c60a87f433d11e8a2ae5edd2b3f0631c56cf21b8378f95f7ad|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0ffeb78b965f82c60a87f433d11e8a2ae5edd2b3f0631c56cf21b8378f95f7ad",
+    "source_id": "github:sparq-org/sparq#2553",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/46/46f823d675773e0775d70c0f82412eb4ad818219c934e4b38b10dcc1f9a0c0a6.json b/.cache/impact/classifications/46/46f823d675773e0775d70c0f82412eb4ad818219c934e4b38b10dcc1f9a0c0a6.json
new file mode 100644
index 0000000..9f8360a
--- /dev/null
+++ b/.cache/impact/classifications/46/46f823d675773e0775d70c0f82412eb4ad818219c934e4b38b10dcc1f9a0c0a6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:feldera/feldera#2934|sha256:457637491be10c62ffcda62eb2b0005ac30463e085a927e0ba3579e22990b9fa|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:457637491be10c62ffcda62eb2b0005ac30463e085a927e0ba3579e22990b9fa",
+    "source_id": "github:feldera/feldera#2934",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/48/489fa49a93ffd8ed27231945f28c16b46df14cc9c71723467e0dd5cb9e894e38.json b/.cache/impact/classifications/48/489fa49a93ffd8ed27231945f28c16b46df14cc9c71723467e0dd5cb9e894e38.json
new file mode 100644
index 0000000..a0f3448
--- /dev/null
+++ b/.cache/impact/classifications/48/489fa49a93ffd8ed27231945f28c16b46df14cc9c71723467e0dd5cb9e894e38.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#15026|sha256:7dd5cd1435c8d90393bf922fadf29d502b230101874dc211062d045b59fbfa1b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7dd5cd1435c8d90393bf922fadf29d502b230101874dc211062d045b59fbfa1b",
+    "source_id": "github:cockroachdb/cockroach#15026",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/48/48cc86691238b160f23f49cb409313b50e0ecd8661427c729dcb530b82115fbf.json b/.cache/impact/classifications/48/48cc86691238b160f23f49cb409313b50e0ecd8661427c729dcb530b82115fbf.json
new file mode 100644
index 0000000..ee51e33
--- /dev/null
+++ b/.cache/impact/classifications/48/48cc86691238b160f23f49cb409313b50e0ecd8661427c729dcb530b82115fbf.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:yugabyte/yugabyte-db#11077|sha256:07d4eca635da3a05c638483b4277da8e9511afe0782a090645cecdaaaecfe92f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "[SQLancer] [YSQL] Select statement may return phantom value"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": null
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:07d4eca635da3a05c638483b4277da8e9511afe0782a090645cecdaaaecfe92f",
+    "source_id": "github:yugabyte/yugabyte-db#11077",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/49/4927362e10f288d1fe270f5b64278fd12ca7db457bb290b77b4a6ba002da4e6f.json b/.cache/impact/classifications/49/4927362e10f288d1fe270f5b64278fd12ca7db457bb290b77b4a6ba002da4e6f.json
new file mode 100644
index 0000000..c1fd3cc
--- /dev/null
+++ b/.cache/impact/classifications/49/4927362e10f288d1fe270f5b64278fd12ca7db457bb290b77b4a6ba002da4e6f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#76729|sha256:18ab7071460d7a1fd2a849b491801e1bdee28ca92a26c784813ab9dd79775c6f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:18ab7071460d7a1fd2a849b491801e1bdee28ca92a26c784813ab9dd79775c6f",
+    "source_id": "github:cockroachdb/cockroach#76729",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/49/4993e3bad1954605b119cab430fb1ec77c7cee0cf1826f383087f3ba64fa8753.json b/.cache/impact/classifications/49/4993e3bad1954605b119cab430fb1ec77c7cee0cf1826f383087f3ba64fa8753.json
new file mode 100644
index 0000000..4f824d9
--- /dev/null
+++ b/.cache/impact/classifications/49/4993e3bad1954605b119cab430fb1ec77c7cee0cf1826f383087f3ba64fa8753.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#104264|sha256:2e4d778cffb26720bfd38372414c32db8856b0322e0a37bbe684112fa45f3060|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2e4d778cffb26720bfd38372414c32db8856b0322e0a37bbe684112fa45f3060",
+    "source_id": "github:elastic/elasticsearch#104264",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/49/49aabb888967bf77623cbd85ec18adcacdafc9d085aef3f3c7929c63419e2d10.json b/.cache/impact/classifications/49/49aabb888967bf77623cbd85ec18adcacdafc9d085aef3f3c7929c63419e2d10.json
new file mode 100644
index 0000000..f6fca0f
--- /dev/null
+++ b/.cache/impact/classifications/49/49aabb888967bf77623cbd85ec18adcacdafc9d085aef3f3c7929c63419e2d10.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#43733|sha256:e6a7d1ffdef1f60342a15cb8c25cb3384e590377255e1815d0c07bdc9d17d2c3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e6a7d1ffdef1f60342a15cb8c25cb3384e590377255e1815d0c07bdc9d17d2c3",
+    "source_id": "github:cockroachdb/cockroach#43733",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4a/4a20c43b3dd2bc6c1de6010e5577cb93c540436a56948f395b6f2228a12fa431.json b/.cache/impact/classifications/4a/4a20c43b3dd2bc6c1de6010e5577cb93c540436a56948f395b6f2228a12fa431.json
new file mode 100644
index 0000000..9888d20
--- /dev/null
+++ b/.cache/impact/classifications/4a/4a20c43b3dd2bc6c1de6010e5577cb93c540436a56948f395b6f2228a12fa431.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:derekmwright/wadjet#289|sha256:2a6637d062a4c540979b7e89262726c469d7129ebaf8a12af0186f7ad10856ab|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2a6637d062a4c540979b7e89262726c469d7129ebaf8a12af0186f7ad10856ab",
+    "source_id": "github:derekmwright/wadjet#289",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4a/4a74495a0016467441a2bb521ce38b46053ed970020ba8631b747d057b9fbf56.json b/.cache/impact/classifications/4a/4a74495a0016467441a2bb521ce38b46053ed970020ba8631b747d057b9fbf56.json
new file mode 100644
index 0000000..d037183
--- /dev/null
+++ b/.cache/impact/classifications/4a/4a74495a0016467441a2bb521ce38b46053ed970020ba8631b747d057b9fbf56.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8117|sha256:742b141b697d5bccef2cc3f0d1e63263db025305b5a07e316e177bac619b45bb|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:742b141b697d5bccef2cc3f0d1e63263db025305b5a07e316e177bac619b45bb",
+    "source_id": "github:cockroachdb/cockroach#8117",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4a/4a894b5f7279c0371db5717fb169b766bbd9935efbe0aeae7d78b2e6755958be.json b/.cache/impact/classifications/4a/4a894b5f7279c0371db5717fb169b766bbd9935efbe0aeae7d78b2e6755958be.json
new file mode 100644
index 0000000..41d0e3e
--- /dev/null
+++ b/.cache/impact/classifications/4a/4a894b5f7279c0371db5717fb169b766bbd9935efbe0aeae7d78b2e6755958be.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#168834|sha256:cb9f3cc0ddb814ede923288ffe72e7883f9b2152ceb9df3ae9e89b94c7787329|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:cb9f3cc0ddb814ede923288ffe72e7883f9b2152ceb9df3ae9e89b94c7787329",
+    "source_id": "github:cockroachdb/cockroach#168834",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4a/4a8eb1bf791dd027e366c7dbe315dacd8fa8755f80f984b5f844b6b066757b02.json b/.cache/impact/classifications/4a/4a8eb1bf791dd027e366c7dbe315dacd8fa8755f80f984b5f844b6b066757b02.json
new file mode 100644
index 0000000..54f74fb
--- /dev/null
+++ b/.cache/impact/classifications/4a/4a8eb1bf791dd027e366c7dbe315dacd8fa8755f80f984b5f844b6b066757b02.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#31154|sha256:271536867ffc6b7f2c841c4c34ecaf7b3cf880d541c663ddebf8da4ead50692e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:271536867ffc6b7f2c841c4c34ecaf7b3cf880d541c663ddebf8da4ead50692e",
+    "source_id": "github:elastic/elasticsearch#31154",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4b/4b976502f3f2b561b6f6579cf3ef6ead998f58bf1c6c402b7c1ab242dd6a61cf.json b/.cache/impact/classifications/4b/4b976502f3f2b561b6f6579cf3ef6ead998f58bf1c6c402b7c1ab242dd6a61cf.json
new file mode 100644
index 0000000..ca624d7
--- /dev/null
+++ b/.cache/impact/classifications/4b/4b976502f3f2b561b6f6579cf3ef6ead998f58bf1c6c402b7c1ab242dd6a61cf.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#79734|sha256:2c898e3a19041c8eca25a1f74fdddedc8f1ebce7d76aa76de4dd95f6c69fd935|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2c898e3a19041c8eca25a1f74fdddedc8f1ebce7d76aa76de4dd95f6c69fd935",
+    "source_id": "github:cockroachdb/cockroach#79734",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4d/4d5bdb0b5e507ba677a2394c4c745142473719bab8a70165dd2b917b39a19202.json b/.cache/impact/classifications/4d/4d5bdb0b5e507ba677a2394c4c745142473719bab8a70165dd2b917b39a19202.json
new file mode 100644
index 0000000..c2318dd
--- /dev/null
+++ b/.cache/impact/classifications/4d/4d5bdb0b5e507ba677a2394c4c745142473719bab8a70165dd2b917b39a19202.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#172480|sha256:904774471c6887df54c6c4b91dd8c624c3aa4916ab41cb7af3fd8d093243ba63|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:904774471c6887df54c6c4b91dd8c624c3aa4916ab41cb7af3fd8d093243ba63",
+    "source_id": "github:cockroachdb/cockroach#172480",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4d/4d8003520949393c66f72aae4bd7cd1c0aad1e7364588a3d215f9db45ffd14b0.json b/.cache/impact/classifications/4d/4d8003520949393c66f72aae4bd7cd1c0aad1e7364588a3d215f9db45ffd14b0.json
new file mode 100644
index 0000000..23cd7db
--- /dev/null
+++ b/.cache/impact/classifications/4d/4d8003520949393c66f72aae4bd7cd1c0aad1e7364588a3d215f9db45ffd14b0.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:dolthub/dolt#3636|sha256:3adbac401890b96d7215a353368689ad72f27c3cafb36be4672d24711a0bc20a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:3adbac401890b96d7215a353368689ad72f27c3cafb36be4672d24711a0bc20a",
+    "source_id": "github:dolthub/dolt#3636",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4d/4dc74403f733aa2009cbefeb5efe4e074128f7d771f626de11bbd12a31382254.json b/.cache/impact/classifications/4d/4dc74403f733aa2009cbefeb5efe4e074128f7d771f626de11bbd12a31382254.json
new file mode 100644
index 0000000..650df29
--- /dev/null
+++ b/.cache/impact/classifications/4d/4dc74403f733aa2009cbefeb5efe4e074128f7d771f626de11bbd12a31382254.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#172320|sha256:fafdf78f723aff5aa509a99a03bf45824f2be2b9cd7183859ca0b0e9979c131a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:fafdf78f723aff5aa509a99a03bf45824f2be2b9cd7183859ca0b0e9979c131a",
+    "source_id": "github:cockroachdb/cockroach#172320",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4e/4e66ba18301a20ca3943fd0f369920737cc2dbf185350db475dd5e0e426969d4.json b/.cache/impact/classifications/4e/4e66ba18301a20ca3943fd0f369920737cc2dbf185350db475dd5e0e426969d4.json
new file mode 100644
index 0000000..f5cf39d
--- /dev/null
+++ b/.cache/impact/classifications/4e/4e66ba18301a20ca3943fd0f369920737cc2dbf185350db475dd5e0e426969d4.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tikv/tikv#4816|sha256:6029dba4140534dc859888eca8bc7e1b80543bcb5c343d86921a9e9c89cdca82|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6029dba4140534dc859888eca8bc7e1b80543bcb5c343d86921a9e9c89cdca82",
+    "source_id": "github:tikv/tikv#4816",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4f/4f6e38b70f579e24aeaab3b58eef7502e0ba5f49a289e9cb8272675bc64e2532.json b/.cache/impact/classifications/4f/4f6e38b70f579e24aeaab3b58eef7502e0ba5f49a289e9cb8272675bc64e2532.json
new file mode 100644
index 0000000..d6b67b1
--- /dev/null
+++ b/.cache/impact/classifications/4f/4f6e38b70f579e24aeaab3b58eef7502e0ba5f49a289e9cb8272675bc64e2532.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:GreptimeTeam/greptimedb#4402|sha256:ec132a7609640e03b9ed89ab50e496465be58a6bca8cfb880e92de27ef87077a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T06:22:52Z",
+  "value": {
+    "classified_at": "2026-09-13T06:22:52Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "A proposal to start running SQLancer, not a defect it found. Decided by reading the issue in this session."
+    },
+    "source_hash": "sha256:ec132a7609640e03b9ed89ab50e496465be58a6bca8cfb880e92de27ef87077a",
+    "source_id": "github:GreptimeTeam/greptimedb#4402",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/4f/4f9dba8edc092c030802352a1883d8305d9100295434ae9d746695d3fc5ce45e.json b/.cache/impact/classifications/4f/4f9dba8edc092c030802352a1883d8305d9100295434ae9d746695d3fc5ce45e.json
new file mode 100644
index 0000000..4601526
--- /dev/null
+++ b/.cache/impact/classifications/4f/4f9dba8edc092c030802352a1883d8305d9100295434ae9d746695d3fc5ce45e.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#134462|sha256:ece4f488775a0e2fe5265489e2601cf281b0bb54e7d8337562236b7c5597803e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ece4f488775a0e2fe5265489e2601cf281b0bb54e7d8337562236b7c5597803e",
+    "source_id": "github:cockroachdb/cockroach#134462",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/50/506a59fc6153b3f20265012fa5a3cfec857997bb3aa0a5db10bb5ad7c9fa8139.json b/.cache/impact/classifications/50/506a59fc6153b3f20265012fa5a3cfec857997bb3aa0a5db10bb5ad7c9fa8139.json
new file mode 100644
index 0000000..5e0d210
--- /dev/null
+++ b/.cache/impact/classifications/50/506a59fc6153b3f20265012fa5a3cfec857997bb3aa0a5db10bb5ad7c9fa8139.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#89101|sha256:e3c5281693fc5c2a528f6c36457cbbafc2750dcc51d68a3100ccd0dc8b98018c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "incorrect cardinality of EXCEPT produces incorrect results"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The TLP roachtest found this defect; the issue was retitled to name it. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e3c5281693fc5c2a528f6c36457cbbafc2750dcc51d68a3100ccd0dc8b98018c",
+    "source_id": "github:cockroachdb/cockroach#89101",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/50/50bc8bbca8434bbd3f96e6238329479876b7841c8b6baa47c524570279000a29.json b/.cache/impact/classifications/50/50bc8bbca8434bbd3f96e6238329479876b7841c8b6baa47c524570279000a29.json
new file mode 100644
index 0000000..fdb93a4
--- /dev/null
+++ b/.cache/impact/classifications/50/50bc8bbca8434bbd3f96e6238329479876b7841c8b6baa47c524570279000a29.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#156254|sha256:d392891a9a97669ce2fe0ad55b97a50ead0c5d2e2bd089e403fe40ca697041d9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d392891a9a97669ce2fe0ad55b97a50ead0c5d2e2bd089e403fe40ca697041d9",
+    "source_id": "github:cockroachdb/cockroach#156254",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/51/512792b78e199ae00264fa027ac149ba8806109596e63d6c95de884ec1b3a44b.json b/.cache/impact/classifications/51/512792b78e199ae00264fa027ac149ba8806109596e63d6c95de884ec1b3a44b.json
new file mode 100644
index 0000000..d112198
--- /dev/null
+++ b/.cache/impact/classifications/51/512792b78e199ae00264fa027ac149ba8806109596e63d6c95de884ec1b3a44b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#1167|sha256:697c765f2f282322ce9bb3dda644e80af72d15aea8112612f65d1608a9d6e0c3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:697c765f2f282322ce9bb3dda644e80af72d15aea8112612f65d1608a9d6e0c3",
+    "source_id": "github:cockroachdb/cockroach#1167",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/52/5203719cd9a14c2c544bea8a73c30497527d2309b3b3d6600f223a0df13b88d3.json b/.cache/impact/classifications/52/5203719cd9a14c2c544bea8a73c30497527d2309b3b3d6600f223a0df13b88d3.json
new file mode 100644
index 0000000..b1e5532
--- /dev/null
+++ b/.cache/impact/classifications/52/5203719cd9a14c2c544bea8a73c30497527d2309b3b3d6600f223a0df13b88d3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#141329|sha256:16859243f9f09f1855c75944504ee8dc50d1ed708088c29e680238bc484728d7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:16859243f9f09f1855c75944504ee8dc50d1ed708088c29e680238bc484728d7",
+    "source_id": "github:cockroachdb/cockroach#141329",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/52/5244f576e9649a3bf27d4e6c6534fbdb52ff8adf8208f2c9beee873c27ec805f.json b/.cache/impact/classifications/52/5244f576e9649a3bf27d4e6c6534fbdb52ff8adf8208f2c9beee873c27ec805f.json
new file mode 100644
index 0000000..1225e18
--- /dev/null
+++ b/.cache/impact/classifications/52/5244f576e9649a3bf27d4e6c6534fbdb52ff8adf8208f2c9beee873c27ec805f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#174972|sha256:dd96bd17319c5346d835b89ab4673f06760e4010579e5578a640da17c94483ac|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T05:53:37Z",
+  "value": {
+    "classified_at": "2026-09-13T05:53:37Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed without its oracle firing: the report carries only the runtime-assertions boilerplate, so nothing here is a finding. Decided by reading the issue in this session rather than by an API call."
+    },
+    "source_hash": "sha256:dd96bd17319c5346d835b89ab4673f06760e4010579e5578a640da17c94483ac",
+    "source_id": "github:cockroachdb/cockroach#174972",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/52/524ef0640358d622fc5455183db292c86b30ff0f149ef3d33cd91b758da870e3.json b/.cache/impact/classifications/52/524ef0640358d622fc5455183db292c86b30ff0f149ef3d33cd91b758da870e3.json
new file mode 100644
index 0000000..92d34c7
--- /dev/null
+++ b/.cache/impact/classifications/52/524ef0640358d622fc5455183db292c86b30ff0f149ef3d33cd91b758da870e3.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#104115|sha256:136834f19f0f8ccea0cbdeb88d7d1b5347dd7b594535625958f7471293f5f315|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:136834f19f0f8ccea0cbdeb88d7d1b5347dd7b594535625958f7471293f5f315",
+    "source_id": "github:cockroachdb/cockroach#104115",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/53/531bb84784800d0ac158f1a81dfdbc52e70b60e4de6b4f4e485a188d387ca20a.json b/.cache/impact/classifications/53/531bb84784800d0ac158f1a81dfdbc52e70b60e4de6b4f4e485a188d387ca20a.json
new file mode 100644
index 0000000..5c98f6b
--- /dev/null
+++ b/.cache/impact/classifications/53/531bb84784800d0ac158f1a81dfdbc52e70b60e4de6b4f4e485a188d387ca20a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8252|sha256:221ac8b382675da1f7cff074047cfff5ee232f152f8bd4f47ed83d39f34d1324|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:221ac8b382675da1f7cff074047cfff5ee232f152f8bd4f47ed83d39f34d1324",
+    "source_id": "github:cockroachdb/cockroach#8252",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/53/53453a3c991e040f1677ee3c3c6d3e22ceb6f9544a3a38aa2abf417f40b9d029.json b/.cache/impact/classifications/53/53453a3c991e040f1677ee3c3c6d3e22ceb6f9544a3a38aa2abf417f40b9d029.json
new file mode 100644
index 0000000..34b571a
--- /dev/null
+++ b/.cache/impact/classifications/53/53453a3c991e040f1677ee3c3c6d3e22ceb6f9544a3a38aa2abf417f40b9d029.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#131532|sha256:ea03c61899cc81f1fdff5d0df50e3265f889470065cea861b07545d7c0300c8f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ea03c61899cc81f1fdff5d0df50e3265f889470065cea861b07545d7c0300c8f",
+    "source_id": "github:cockroachdb/cockroach#131532",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/53/537a6e513973de4259a0dceac629aba3aa18f3a2cb54e5c288f061f31e000a5e.json b/.cache/impact/classifications/53/537a6e513973de4259a0dceac629aba3aa18f3a2cb54e5c288f061f31e000a5e.json
new file mode 100644
index 0000000..790ef2d
--- /dev/null
+++ b/.cache/impact/classifications/53/537a6e513973de4259a0dceac629aba3aa18f3a2cb54e5c288f061f31e000a5e.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#115821|sha256:9991c1ddfef168bca930348b28e70cf042d3a7cc56af958de6bdf75712df6b97|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9991c1ddfef168bca930348b28e70cf042d3a7cc56af958de6bdf75712df6b97",
+    "source_id": "github:cockroachdb/cockroach#115821",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/53/53ed904452ad4a7f926863e8b0633293cccf99fadbd0759b4c6bdb70d087dffa.json b/.cache/impact/classifications/53/53ed904452ad4a7f926863e8b0633293cccf99fadbd0759b4c6bdb70d087dffa.json
new file mode 100644
index 0000000..cd0e29d
--- /dev/null
+++ b/.cache/impact/classifications/53/53ed904452ad4a7f926863e8b0633293cccf99fadbd0759b4c6bdb70d087dffa.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#110528|sha256:42ef12510c6a849b5610e08988b1229124f2bd0c9d3f984ed220379d3ddedd91|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:42ef12510c6a849b5610e08988b1229124f2bd0c9d3f984ed220379d3ddedd91",
+    "source_id": "github:cockroachdb/cockroach#110528",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/54/545bbb2753d7fc1eedee53b8a775b5d460b52676ec568f60a1490acd2db2419c.json b/.cache/impact/classifications/54/545bbb2753d7fc1eedee53b8a775b5d460b52676ec568f60a1490acd2db2419c.json
new file mode 100644
index 0000000..29fc09e
--- /dev/null
+++ b/.cache/impact/classifications/54/545bbb2753d7fc1eedee53b8a775b5d460b52676ec568f60a1490acd2db2419c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#94587|sha256:135c5778be7d8db7bee1b9d84fe88b9fbf0bc10d556a67cc87ee15187c1f63a9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:135c5778be7d8db7bee1b9d84fe88b9fbf0bc10d556a67cc87ee15187c1f63a9",
+    "source_id": "github:cockroachdb/cockroach#94587",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/54/549c3dd260c607aea5cd538d511c89d82c7829182a9d19ba96d26eb50030f6e8.json b/.cache/impact/classifications/54/549c3dd260c607aea5cd538d511c89d82c7829182a9d19ba96d26eb50030f6e8.json
new file mode 100644
index 0000000..f507757
--- /dev/null
+++ b/.cache/impact/classifications/54/549c3dd260c607aea5cd538d511c89d82c7829182a9d19ba96d26eb50030f6e8.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:stoneatom/stonedb#1916|sha256:21050f002cf962bd83f417a6dae4d882ee705afe5472294a3aa87f1c92af6fe3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:21050f002cf962bd83f417a6dae4d882ee705afe5472294a3aa87f1c92af6fe3",
+    "source_id": "github:stoneatom/stonedb#1916",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/55/5547facf2467a1dfbd64db232561bdfe45e61af252905f368909cbc552c4b5c4.json b/.cache/impact/classifications/55/5547facf2467a1dfbd64db232561bdfe45e61af252905f368909cbc552c4b5c4.json
new file mode 100644
index 0000000..f44486c
--- /dev/null
+++ b/.cache/impact/classifications/55/5547facf2467a1dfbd64db232561bdfe45e61af252905f368909cbc552c4b5c4.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#60632|sha256:87254223969cf1807fa7d0a92e6a56e7f1c23eeffe57c6e0fa2886699cc3fc76|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:87254223969cf1807fa7d0a92e6a56e7f1c23eeffe57c6e0fa2886699cc3fc76",
+    "source_id": "github:cockroachdb/cockroach#60632",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/56/56d513e1fada73b3e7ae9de37b25b1dc0c27570c56c69b16f9818b186c92d185.json b/.cache/impact/classifications/56/56d513e1fada73b3e7ae9de37b25b1dc0c27570c56c69b16f9818b186c92d185.json
new file mode 100644
index 0000000..6f84da4
--- /dev/null
+++ b/.cache/impact/classifications/56/56d513e1fada73b3e7ae9de37b25b1dc0c27570c56c69b16f9818b186c92d185.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:openlink/virtuoso-opensource#1426|sha256:1a9ebea79b61c7ce025baa07bd3f928e9d167559f525cf3c0248189b7288d1fc|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "TLP (tautology-based partitioning) rewritten query"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1a9ebea79b61c7ce025baa07bd3f928e9d167559f525cf3c0248189b7288d1fc",
+    "source_id": "github:openlink/virtuoso-opensource#1426",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/56/56fab7ff30971184a36d17ee3d426f281dca24ead77dca80e31773f9925ed368.json b/.cache/impact/classifications/56/56fab7ff30971184a36d17ee3d426f281dca24ead77dca80e31773f9925ed368.json
new file mode 100644
index 0000000..2728ff8
--- /dev/null
+++ b/.cache/impact/classifications/56/56fab7ff30971184a36d17ee3d426f281dca24ead77dca80e31773f9925ed368.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4103|sha256:7a95fe910fe39b1fbf100623c84dae0c248fbfc8ed7fd192b2c3d0dd6d89be3d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7a95fe910fe39b1fbf100623c84dae0c248fbfc8ed7fd192b2c3d0dd6d89be3d",
+    "source_id": "github:sparq-org/sparq#4103",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/57/571e14f37f222c5e38a7a6cceef227bab7bc70fa3de58c9565c969b15ecfd704.json b/.cache/impact/classifications/57/571e14f37f222c5e38a7a6cceef227bab7bc70fa3de58c9565c969b15ecfd704.json
new file mode 100644
index 0000000..25be7d8
--- /dev/null
+++ b/.cache/impact/classifications/57/571e14f37f222c5e38a7a6cceef227bab7bc70fa3de58c9565c969b15ecfd704.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#10877|sha256:27c7f03a9dc1b7f52f9b411f558effee7568729d88d586ea110023f04cb534b9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:27c7f03a9dc1b7f52f9b411f558effee7568729d88d586ea110023f04cb534b9",
+    "source_id": "github:cockroachdb/cockroach#10877",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/57/57d1d1457138578912e7a6979e9da67402d87384fecbd04a8bb754db1cf14c62.json b/.cache/impact/classifications/57/57d1d1457138578912e7a6979e9da67402d87384fecbd04a8bb754db1cf14c62.json
new file mode 100644
index 0000000..0d06989
--- /dev/null
+++ b/.cache/impact/classifications/57/57d1d1457138578912e7a6979e9da67402d87384fecbd04a8bb754db1cf14c62.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tikv/tikv#8883|sha256:b2d05722e41d2b691a1592f18615db18ace518f0f2d3cdf706c3a3f297b6ce26|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b2d05722e41d2b691a1592f18615db18ace518f0f2d3cdf706c3a3f297b6ce26",
+    "source_id": "github:tikv/tikv#8883",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/58/58bb6c707dbb1faa38e74395db0e7c64b03988c34d83d451bebd4f524be9fc66.json b/.cache/impact/classifications/58/58bb6c707dbb1faa38e74395db0e7c64b03988c34d83d451bebd4f524be9fc66.json
new file mode 100644
index 0000000..2a5e935
--- /dev/null
+++ b/.cache/impact/classifications/58/58bb6c707dbb1faa38e74395db0e7c64b03988c34d83d451bebd4f524be9fc66.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#36280|sha256:789e23c57dfce604218561210f59f3bb5c1f8011030b09fd1d7c50b2c2e17195|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:789e23c57dfce604218561210f59f3bb5c1f8011030b09fd1d7c50b2c2e17195",
+    "source_id": "github:cockroachdb/cockroach#36280",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/58/58ed54dd184a8b2edfce15ecdb7ccdce8d098bf29efb58216a52ae304fa24f96.json b/.cache/impact/classifications/58/58ed54dd184a8b2edfce15ecdb7ccdce8d098bf29efb58216a52ae304fa24f96.json
new file mode 100644
index 0000000..5de2c97
--- /dev/null
+++ b/.cache/impact/classifications/58/58ed54dd184a8b2edfce15ecdb7ccdce8d098bf29efb58216a52ae304fa24f96.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#79033|sha256:78d3c5cf43b3b7abd0acaad74d8a73cd716a2b2b0a067a158a34cca3f434f3d3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:78d3c5cf43b3b7abd0acaad74d8a73cd716a2b2b0a067a158a34cca3f434f3d3",
+    "source_id": "github:cockroachdb/cockroach#79033",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/59/5915c258b03b0d16bba87b3db3baf0f25d18905f56015d0b0f1d9fee40f75f13.json b/.cache/impact/classifications/59/5915c258b03b0d16bba87b3db3baf0f25d18905f56015d0b0f1d9fee40f75f13.json
new file mode 100644
index 0000000..e3773e4
--- /dev/null
+++ b/.cache/impact/classifications/59/5915c258b03b0d16bba87b3db3baf0f25d18905f56015d0b0f1d9fee40f75f13.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#83933|sha256:5a2eafc887c798dda8aa8bd040a255fdc0b899ab5a71f47c12ecfb8b267c7421|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5a2eafc887c798dda8aa8bd040a255fdc0b899ab5a71f47c12ecfb8b267c7421",
+    "source_id": "github:cockroachdb/cockroach#83933",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/59/595510e769eafcf0b275b659f1dac7f8770500df101b1e03339f11f842783b97.json b/.cache/impact/classifications/59/595510e769eafcf0b275b659f1dac7f8770500df101b1e03339f11f842783b97.json
new file mode 100644
index 0000000..bb9a340
--- /dev/null
+++ b/.cache/impact/classifications/59/595510e769eafcf0b275b659f1dac7f8770500df101b1e03339f11f842783b97.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#171381|sha256:02ab6615fd1bfde72081ed1a313c812535163dc1b194acfddbc0334a952cd759|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:02ab6615fd1bfde72081ed1a313c812535163dc1b194acfddbc0334a952cd759",
+    "source_id": "github:cockroachdb/cockroach#171381",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/59/59e0883ad90f346f4f6df11b9a25ea657793f52cd1aa281ff5294f6941b6a7a2.json b/.cache/impact/classifications/59/59e0883ad90f346f4f6df11b9a25ea657793f52cd1aa281ff5294f6941b6a7a2.json
new file mode 100644
index 0000000..d8202ac
--- /dev/null
+++ b/.cache/impact/classifications/59/59e0883ad90f346f4f6df11b9a25ea657793f52cd1aa281ff5294f6941b6a7a2.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#70260|sha256:cabb013eda46e598e3afa7fd2c714e216f5e3f99a0ca4d13f47687b4a1d52efd|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:cabb013eda46e598e3afa7fd2c714e216f5e3f99a0ca4d13f47687b4a1d52efd",
+    "source_id": "github:pingcap/tidb#70260",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/59/59e0efdead4ad79525a71d3fe73c493f6734a28f1c4226e38d16e0593b993462.json b/.cache/impact/classifications/59/59e0efdead4ad79525a71d3fe73c493f6734a28f1c4226e38d16e0593b993462.json
new file mode 100644
index 0000000..5e52e4d
--- /dev/null
+++ b/.cache/impact/classifications/59/59e0efdead4ad79525a71d3fe73c493f6734a28f1c4226e38d16e0593b993462.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#78327|sha256:64a98e9f4dec773163f0a1cbdbf7a7bb84c2eb21d85590d920f51208c440c0b5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:64a98e9f4dec773163f0a1cbdbf7a7bb84c2eb21d85590d920f51208c440c0b5",
+    "source_id": "github:cockroachdb/cockroach#78327",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/5b/5b4497c429e7478e70c7a7dbec31313325de1a8988b0c8e0d70f8da3837a8562.json b/.cache/impact/classifications/5b/5b4497c429e7478e70c7a7dbec31313325de1a8988b0c8e0d70f8da3837a8562.json
new file mode 100644
index 0000000..28f6488
--- /dev/null
+++ b/.cache/impact/classifications/5b/5b4497c429e7478e70c7a7dbec31313325de1a8988b0c8e0d70f8da3837a8562.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#101875|sha256:bf7cb974fc99e37130a763f1b3f1a1285b4d2d21dc0f5a1c4e4c67df6f295042|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:bf7cb974fc99e37130a763f1b3f1a1285b4d2d21dc0f5a1c4e4c67df6f295042",
+    "source_id": "github:cockroachdb/cockroach#101875",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/5c/5cc81a4e394e3c17040dcfa7ea642809991dcf7cf40a2c0712537fa8e0546dbe.json b/.cache/impact/classifications/5c/5cc81a4e394e3c17040dcfa7ea642809991dcf7cf40a2c0712537fa8e0546dbe.json
new file mode 100644
index 0000000..96b08bc
--- /dev/null
+++ b/.cache/impact/classifications/5c/5cc81a4e394e3c17040dcfa7ea642809991dcf7cf40a2c0712537fa8e0546dbe.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#145025|sha256:e8819cffd03ec4d1b97d4e6e02069396cab2317f28abb567cd5f5413d70274cb|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e8819cffd03ec4d1b97d4e6e02069396cab2317f28abb567cd5f5413d70274cb",
+    "source_id": "github:cockroachdb/cockroach#145025",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/5d/5d05eb14f6f3da47465d4c397f57605d2c7612d188fdc09eb1bb168b77ad224f.json b/.cache/impact/classifications/5d/5d05eb14f6f3da47465d4c397f57605d2c7612d188fdc09eb1bb168b77ad224f.json
new file mode 100644
index 0000000..83d573b
--- /dev/null
+++ b/.cache/impact/classifications/5d/5d05eb14f6f3da47465d4c397f57605d2c7612d188fdc09eb1bb168b77ad224f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#3167|sha256:4e423333e154e04766d63b89670f9c6e58b85b96c62ad7b152c428c680bc3527|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4e423333e154e04766d63b89670f9c6e58b85b96c62ad7b152c428c680bc3527",
+    "source_id": "github:sparq-org/sparq#3167",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/5d/5d7bbaa5e4af57afde1ad3ad8cd3dc16d4fdc24acd94a6c60ffd8a737bbf6464.json b/.cache/impact/classifications/5d/5d7bbaa5e4af57afde1ad3ad8cd3dc16d4fdc24acd94a6c60ffd8a737bbf6464.json
new file mode 100644
index 0000000..c3f54e3
--- /dev/null
+++ b/.cache/impact/classifications/5d/5d7bbaa5e4af57afde1ad3ad8cd3dc16d4fdc24acd94a6c60ffd8a737bbf6464.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tikv/tikv#11588|sha256:a0d73b9c14de4631386b21cb4a413f2117689ada3f68c36d230fe18631c7e610|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a0d73b9c14de4631386b21cb4a413f2117689ada3f68c36d230fe18631c7e610",
+    "source_id": "github:tikv/tikv#11588",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/5d/5d90b40170adfdbfaca2a6391232f53db26a768332d128f0726eb921e3804304.json b/.cache/impact/classifications/5d/5d90b40170adfdbfaca2a6391232f53db26a768332d128f0726eb921e3804304.json
new file mode 100644
index 0000000..9eab638
--- /dev/null
+++ b/.cache/impact/classifications/5d/5d90b40170adfdbfaca2a6391232f53db26a768332d128f0726eb921e3804304.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#47405|sha256:eec80d0d1c775330b91c40fcad499255844332b416ac9c7692630635893c7c63|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:eec80d0d1c775330b91c40fcad499255844332b416ac9c7692630635893c7c63",
+    "source_id": "github:cockroachdb/cockroach#47405",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/5d/5d9e3c14dff91b1b3e5549a143c8d2ac5f10c5f9f14564e9529a73e42a7f34a3.json b/.cache/impact/classifications/5d/5d9e3c14dff91b1b3e5549a143c8d2ac5f10c5f9f14564e9529a73e42a7f34a3.json
new file mode 100644
index 0000000..4bb8959
--- /dev/null
+++ b/.cache/impact/classifications/5d/5d9e3c14dff91b1b3e5549a143c8d2ac5f10c5f9f14564e9529a73e42a7f34a3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#76677|sha256:488378068854e4ff550650419ac556f5b068bb23a91643f3c619528a1b59dea9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:488378068854e4ff550650419ac556f5b068bb23a91643f3c619528a1b59dea9",
+    "source_id": "github:cockroachdb/cockroach#76677",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/5e/5e25e4212fc626bf7415dd119b1105d320c03eb4c23e5cf696ad249ad6a2128a.json b/.cache/impact/classifications/5e/5e25e4212fc626bf7415dd119b1105d320c03eb4c23e5cf696ad249ad6a2128a.json
new file mode 100644
index 0000000..87ef683
--- /dev/null
+++ b/.cache/impact/classifications/5e/5e25e4212fc626bf7415dd119b1105d320c03eb4c23e5cf696ad249ad6a2128a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#12038|sha256:8dbdd7391db2fab519e0ba1ea06c26953a7f740a606c6c67f2bd3e05ef7ce8d5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8dbdd7391db2fab519e0ba1ea06c26953a7f740a606c6c67f2bd3e05ef7ce8d5",
+    "source_id": "github:yugabyte/yugabyte-db#12038",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/5e/5e4c5954e3b0fa38c57b4d551d274893891e1aa3f992b02d65bf2b21cca702e6.json b/.cache/impact/classifications/5e/5e4c5954e3b0fa38c57b4d551d274893891e1aa3f992b02d65bf2b21cca702e6.json
new file mode 100644
index 0000000..5d032b5
--- /dev/null
+++ b/.cache/impact/classifications/5e/5e4c5954e3b0fa38c57b4d551d274893891e1aa3f992b02d65bf2b21cca702e6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4569|sha256:c2b4e11b7deb355617a8aa55321e050f9e04cfd061321b67e701930a45feb150|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c2b4e11b7deb355617a8aa55321e050f9e04cfd061321b67e701930a45feb150",
+    "source_id": "github:sparq-org/sparq#4569",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/5e/5e9348927654223e7a8731c714636e3eb50b6b219e716ff19e2ee0557dfe02ed.json b/.cache/impact/classifications/5e/5e9348927654223e7a8731c714636e3eb50b6b219e716ff19e2ee0557dfe02ed.json
new file mode 100644
index 0000000..d49f7ac
--- /dev/null
+++ b/.cache/impact/classifications/5e/5e9348927654223e7a8731c714636e3eb50b6b219e716ff19e2ee0557dfe02ed.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#4217|sha256:45b21c0593af2f5f387bbd800dc3c22d9051944c76c6f8729922e0c43cce99f1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:45b21c0593af2f5f387bbd800dc3c22d9051944c76c6f8729922e0c43cce99f1",
+    "source_id": "github:cockroachdb/cockroach#4217",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/5f/5fbfcff15755404b4373d7039268a62d716c1379b68ea32ba957532202e15a17.json b/.cache/impact/classifications/5f/5fbfcff15755404b4373d7039268a62d716c1379b68ea32ba957532202e15a17.json
new file mode 100644
index 0000000..3e09e81
--- /dev/null
+++ b/.cache/impact/classifications/5f/5fbfcff15755404b4373d7039268a62d716c1379b68ea32ba957532202e15a17.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#140451|sha256:ed5b1fa4196c53073769bb963c810316f3c2ba56e306dd1dedb8a0b05cb08394|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ed5b1fa4196c53073769bb963c810316f3c2ba56e306dd1dedb8a0b05cb08394",
+    "source_id": "github:cockroachdb/cockroach#140451",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/60/6000ba53b7957f96e45b4d21488a51bb94ce0760c2a8c64d9b2c4a4025900660.json b/.cache/impact/classifications/60/6000ba53b7957f96e45b4d21488a51bb94ce0760c2a8c64d9b2c4a4025900660.json
new file mode 100644
index 0000000..43c195e
--- /dev/null
+++ b/.cache/impact/classifications/60/6000ba53b7957f96e45b4d21488a51bb94ce0760c2a8c64d9b2c4a4025900660.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#174568|sha256:3cf89dab22e5a7814f65db1f36cfc2f84a4d9e6e46238a95fe0c6e15403f7978|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:3cf89dab22e5a7814f65db1f36cfc2f84a4d9e6e46238a95fe0c6e15403f7978",
+    "source_id": "github:cockroachdb/cockroach#174568",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/60/6072b4145c97bc2926ac50936a2af20b83209cf2b91f410cd8b581776d89ecb5.json b/.cache/impact/classifications/60/6072b4145c97bc2926ac50936a2af20b83209cf2b91f410cd8b581776d89ecb5.json
new file mode 100644
index 0000000..7a55bf6
--- /dev/null
+++ b/.cache/impact/classifications/60/6072b4145c97bc2926ac50936a2af20b83209cf2b91f410cd8b581776d89ecb5.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#133360|sha256:ac99aa943ff11e62d3b85a207fbb83f307886dab3888ca21c8922ccc32c67190|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ac99aa943ff11e62d3b85a207fbb83f307886dab3888ca21c8922ccc32c67190",
+    "source_id": "github:cockroachdb/cockroach#133360",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/60/60900ac72cef6bab9243ae516432d910d1c86a76d315f823df00daa1482bcfa2.json b/.cache/impact/classifications/60/60900ac72cef6bab9243ae516432d910d1c86a76d315f823df00daa1482bcfa2.json
new file mode 100644
index 0000000..fec8f8f
--- /dev/null
+++ b/.cache/impact/classifications/60/60900ac72cef6bab9243ae516432d910d1c86a76d315f823df00daa1482bcfa2.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#85211|sha256:7c1db12089f32541e6ed76b1dbb3a67704cdfa1bc1bcc2eb710f55f6eadc4e10|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7c1db12089f32541e6ed76b1dbb3a67704cdfa1bc1bcc2eb710f55f6eadc4e10",
+    "source_id": "github:cockroachdb/cockroach#85211",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/60/6095cccecd36e4a3dcfb9f7aef30883baf956a0139cbf0d08024f68f7665f2a0.json b/.cache/impact/classifications/60/6095cccecd36e4a3dcfb9f7aef30883baf956a0139cbf0d08024f68f7665f2a0.json
new file mode 100644
index 0000000..ab8bb63
--- /dev/null
+++ b/.cache/impact/classifications/60/6095cccecd36e4a3dcfb9f7aef30883baf956a0139cbf0d08024f68f7665f2a0.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#78137|sha256:d2c2da7b29ac7d0b29da3c0e7882de4191c572d8caa4deee3d55db14247935df|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d2c2da7b29ac7d0b29da3c0e7882de4191c572d8caa4deee3d55db14247935df",
+    "source_id": "github:cockroachdb/cockroach#78137",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/61/61bd15c6adeb577a3140b4cc68a92425a82cb37ef49688ef9a7d8cb1bbed1b80.json b/.cache/impact/classifications/61/61bd15c6adeb577a3140b4cc68a92425a82cb37ef49688ef9a7d8cb1bbed1b80.json
new file mode 100644
index 0000000..3a833cd
--- /dev/null
+++ b/.cache/impact/classifications/61/61bd15c6adeb577a3140b4cc68a92425a82cb37ef49688ef9a7d8cb1bbed1b80.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#76852|sha256:5e7ec0fce73e3cd91e8f92c0a96180b418a3aea26c6455ee64564e1fd8cbcb7f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5e7ec0fce73e3cd91e8f92c0a96180b418a3aea26c6455ee64564e1fd8cbcb7f",
+    "source_id": "github:cockroachdb/cockroach#76852",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/62/62845e48c5547fd5470856a612a3fa45e93a34d88a7f61757085d5faf857118c.json b/.cache/impact/classifications/62/62845e48c5547fd5470856a612a3fa45e93a34d88a7f61757085d5faf857118c.json
new file mode 100644
index 0000000..ba696b4
--- /dev/null
+++ b/.cache/impact/classifications/62/62845e48c5547fd5470856a612a3fa45e93a34d88a7f61757085d5faf857118c.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#79737|sha256:0f3d5691ea2aaeeed41f79199baf05e7b4a9888f2099b06eb4b3926c3c6930af|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0f3d5691ea2aaeeed41f79199baf05e7b4a9888f2099b06eb4b3926c3c6930af",
+    "source_id": "github:cockroachdb/cockroach#79737",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/64/64177254330bad439b23b84086ae0a944da40d5b19fec334adf3852f23a40a30.json b/.cache/impact/classifications/64/64177254330bad439b23b84086ae0a944da40d5b19fec334adf3852f23a40a30.json
new file mode 100644
index 0000000..0d9302f
--- /dev/null
+++ b/.cache/impact/classifications/64/64177254330bad439b23b84086ae0a944da40d5b19fec334adf3852f23a40a30.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4126|sha256:4df868c3cf67ba6656d7c1ce5a2ac65a3ad7cb5da194ba8a54b4c65a4fe72df0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4df868c3cf67ba6656d7c1ce5a2ac65a3ad7cb5da194ba8a54b4c65a4fe72df0",
+    "source_id": "github:sparq-org/sparq#4126",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/64/642045babfb46738d61659786bc0035b6c3e0b80167ba3817438688236b45f77.json b/.cache/impact/classifications/64/642045babfb46738d61659786bc0035b6c3e0b80167ba3817438688236b45f77.json
new file mode 100644
index 0000000..03407dd
--- /dev/null
+++ b/.cache/impact/classifications/64/642045babfb46738d61659786bc0035b6c3e0b80167ba3817438688236b45f77.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#17981|sha256:9f011259fe06d8095f6b7a4977a56064b759ee8a8b479789f226c04ff1eba94c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9f011259fe06d8095f6b7a4977a56064b759ee8a8b479789f226c04ff1eba94c",
+    "source_id": "github:yugabyte/yugabyte-db#17981",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/64/64cf21577d6540c5454a80ea273aeea731afc4c062bde302ed7a558204b30b08.json b/.cache/impact/classifications/64/64cf21577d6540c5454a80ea273aeea731afc4c062bde302ed7a558204b30b08.json
new file mode 100644
index 0000000..0c7e055
--- /dev/null
+++ b/.cache/impact/classifications/64/64cf21577d6540c5454a80ea273aeea731afc4c062bde302ed7a558204b30b08.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#115793|sha256:d62ad2266727a8f529321757d750b068e742390c63297b7db4cb8f3c1af32cd1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d62ad2266727a8f529321757d750b068e742390c63297b7db4cb8f3c1af32cd1",
+    "source_id": "github:cockroachdb/cockroach#115793",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/67/6707c8a0ddbe3a376002a54285b99a41d9f0688faadb287be6d33a3e327e4db8.json b/.cache/impact/classifications/67/6707c8a0ddbe3a376002a54285b99a41d9f0688faadb287be6d33a3e327e4db8.json
new file mode 100644
index 0000000..336b61f
--- /dev/null
+++ b/.cache/impact/classifications/67/6707c8a0ddbe3a376002a54285b99a41d9f0688faadb287be6d33a3e327e4db8.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#9153|sha256:8437f03f6ebeffaa3f3f4f2f06520e78823762fabff0e9b71ee6b1ffe052b797|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8437f03f6ebeffaa3f3f4f2f06520e78823762fabff0e9b71ee6b1ffe052b797",
+    "source_id": "github:cockroachdb/cockroach#9153",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/67/67b463c4a08fe1ba60996404766dddca40878c70b54ea078a9967dffbcfc8658.json b/.cache/impact/classifications/67/67b463c4a08fe1ba60996404766dddca40878c70b54ea078a9967dffbcfc8658.json
new file mode 100644
index 0000000..ff21fb5
--- /dev/null
+++ b/.cache/impact/classifications/67/67b463c4a08fe1ba60996404766dddca40878c70b54ea078a9967dffbcfc8658.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#89743|sha256:ca1434f65ea8bd442f306ff47c3178d5e7b758b64187b59b8d84382d1dd080c5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ca1434f65ea8bd442f306ff47c3178d5e7b758b64187b59b8d84382d1dd080c5",
+    "source_id": "github:cockroachdb/cockroach#89743",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/68/6852e1d7c3410ffb485fad53bd8fe55fee267498e747e27326881ce5bc51d21a.json b/.cache/impact/classifications/68/6852e1d7c3410ffb485fad53bd8fe55fee267498e747e27326881ce5bc51d21a.json
new file mode 100644
index 0000000..cac5a22
--- /dev/null
+++ b/.cache/impact/classifications/68/6852e1d7c3410ffb485fad53bd8fe55fee267498e747e27326881ce5bc51d21a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:ydb-platform/ydb#29532|sha256:eb6ee113b404eb6cab4a22959a5daf8580e322de7935292c915782cbe5081639|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:eb6ee113b404eb6cab4a22959a5daf8580e322de7935292c915782cbe5081639",
+    "source_id": "github:ydb-platform/ydb#29532",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/69/69176cd33db131dcbcea2af095d36c9289e2a53aebaf5de0a6d8991efc49d52c.json b/.cache/impact/classifications/69/69176cd33db131dcbcea2af095d36c9289e2a53aebaf5de0a6d8991efc49d52c.json
new file mode 100644
index 0000000..1c1df25
--- /dev/null
+++ b/.cache/impact/classifications/69/69176cd33db131dcbcea2af095d36c9289e2a53aebaf5de0a6d8991efc49d52c.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#74216|sha256:0ac689fa0f64d9b7bafad22eb2e4f7c48871d94df61d616e46dad8d37bdc645e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "TLP discovered a correctness issue"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0ac689fa0f64d9b7bafad22eb2e4f7c48871d94df61d616e46dad8d37bdc645e",
+    "source_id": "github:cockroachdb/cockroach#74216",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/69/6988bdca83c01ba81d72ac475bd12c3245f94c67e3f73e57fb0d69f7537e7fb0.json b/.cache/impact/classifications/69/6988bdca83c01ba81d72ac475bd12c3245f94c67e3f73e57fb0d69f7537e7fb0.json
new file mode 100644
index 0000000..e776497
--- /dev/null
+++ b/.cache/impact/classifications/69/6988bdca83c01ba81d72ac475bd12c3245f94c67e3f73e57fb0d69f7537e7fb0.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4645|sha256:e6b1f437cd1964a314acb10f9df410199793a6f250d30728fe07f6c1e015f57a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e6b1f437cd1964a314acb10f9df410199793a6f250d30728fe07f6c1e015f57a",
+    "source_id": "github:sparq-org/sparq#4645",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/69/69b7cb07e87cf1e27dbb7626c0108289a955adc3363c13dfa23a74eecc12a1de.json b/.cache/impact/classifications/69/69b7cb07e87cf1e27dbb7626c0108289a955adc3363c13dfa23a74eecc12a1de.json
new file mode 100644
index 0000000..605159e
--- /dev/null
+++ b/.cache/impact/classifications/69/69b7cb07e87cf1e27dbb7626c0108289a955adc3363c13dfa23a74eecc12a1de.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#42567|sha256:d5e7746a7393728b2ff2a107194fe62edb82fbe1516200c3d2d6abfa5365c2a0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d5e7746a7393728b2ff2a107194fe62edb82fbe1516200c3d2d6abfa5365c2a0",
+    "source_id": "github:cockroachdb/cockroach#42567",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/6a/6abfde34d627ab8791eab0b6e69878ec222e624203c2a07beec07f57c73c6437.json b/.cache/impact/classifications/6a/6abfde34d627ab8791eab0b6e69878ec222e624203c2a07beec07f57c73c6437.json
new file mode 100644
index 0000000..6ae78d1
--- /dev/null
+++ b/.cache/impact/classifications/6a/6abfde34d627ab8791eab0b6e69878ec222e624203c2a07beec07f57c73c6437.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#11637|sha256:9ec354313223f5bc1e08017d8776cb969d403b2bf65bb667d98793374f8c1f55|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9ec354313223f5bc1e08017d8776cb969d403b2bf65bb667d98793374f8c1f55",
+    "source_id": "github:cockroachdb/cockroach#11637",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/6c/6c722d18071970e6194ac2aa75684ecc29ddd5beab318412e915f18f89d370f5.json b/.cache/impact/classifications/6c/6c722d18071970e6194ac2aa75684ecc29ddd5beab318412e915f18f89d370f5.json
new file mode 100644
index 0000000..7689645
--- /dev/null
+++ b/.cache/impact/classifications/6c/6c722d18071970e6194ac2aa75684ecc29ddd5beab318412e915f18f89d370f5.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#130253|sha256:4ffc99a808dce15ed0f309d4ac4cae40fb30cb0e25f16d61aaa97111cd0042ff|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4ffc99a808dce15ed0f309d4ac4cae40fb30cb0e25f16d61aaa97111cd0042ff",
+    "source_id": "github:cockroachdb/cockroach#130253",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/6d/6d8597a86030d5f54e13da951ed87c00907a04b3c7e9e672024d9e7a0c76c5b1.json b/.cache/impact/classifications/6d/6d8597a86030d5f54e13da951ed87c00907a04b3c7e9e672024d9e7a0c76c5b1.json
new file mode 100644
index 0000000..02e0bfb
--- /dev/null
+++ b/.cache/impact/classifications/6d/6d8597a86030d5f54e13da951ed87c00907a04b3c7e9e672024d9e7a0c76c5b1.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#97200|sha256:5066d94c30f6d6b315ab633dbe3dad6171a4d69ed5a2225624f76d75fac6db57|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5066d94c30f6d6b315ab633dbe3dad6171a4d69ed5a2225624f76d75fac6db57",
+    "source_id": "github:cockroachdb/cockroach#97200",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/6e/6efa9cf70f3c884dbcd9256901aaf144e83dc7fc156b1b353c6419d5cdd9dd6f.json b/.cache/impact/classifications/6e/6efa9cf70f3c884dbcd9256901aaf144e83dc7fc156b1b353c6419d5cdd9dd6f.json
new file mode 100644
index 0000000..e3e0c3f
--- /dev/null
+++ b/.cache/impact/classifications/6e/6efa9cf70f3c884dbcd9256901aaf144e83dc7fc156b1b353c6419d5cdd9dd6f.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#130759|sha256:c17b16ee2d0fad8992a9ad4fe693e6a08b1dbb03e87628856d85d03d6e7dfea3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c17b16ee2d0fad8992a9ad4fe693e6a08b1dbb03e87628856d85d03d6e7dfea3",
+    "source_id": "github:cockroachdb/cockroach#130759",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/70/70669a1e477ab9f76e486c9924c9d91309ca6794ee08fd066ed7f1c69de35592.json b/.cache/impact/classifications/70/70669a1e477ab9f76e486c9924c9d91309ca6794ee08fd066ed7f1c69de35592.json
new file mode 100644
index 0000000..417addd
--- /dev/null
+++ b/.cache/impact/classifications/70/70669a1e477ab9f76e486c9924c9d91309ca6794ee08fd066ed7f1c69de35592.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:citusdata/citus#7833|sha256:0acda4d40d9d79500552743434ebfc2a772f22d068cfd7ce9e4c5df4996912b7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "SQLancer generates invalid UTF-8 byte sequences"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": null
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0acda4d40d9d79500552743434ebfc2a772f22d068cfd7ce9e4c5df4996912b7",
+    "source_id": "github:citusdata/citus#7833",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/70/70a8f65f348e8d6fe4c7cdf5bf8a201ae38a302e9ba778cab953442cc85b1d59.json b/.cache/impact/classifications/70/70a8f65f348e8d6fe4c7cdf5bf8a201ae38a302e9ba778cab953442cc85b1d59.json
new file mode 100644
index 0000000..86b5ae4
--- /dev/null
+++ b/.cache/impact/classifications/70/70a8f65f348e8d6fe4c7cdf5bf8a201ae38a302e9ba778cab953442cc85b1d59.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:questdb/questdb#4079|sha256:be76f4aea3cf9f4f577c7c28e0e16219bea85914f2e88412855d207b3b37cf0b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:be76f4aea3cf9f4f577c7c28e0e16219bea85914f2e88412855d207b3b37cf0b",
+    "source_id": "github:questdb/questdb#4079",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/72/7209979408b0b5fe565b8eeac6c0d419465844a91c07964b232f8f34931e115e.json b/.cache/impact/classifications/72/7209979408b0b5fe565b8eeac6c0d419465844a91c07964b232f8f34931e115e.json
new file mode 100644
index 0000000..5f39727
--- /dev/null
+++ b/.cache/impact/classifications/72/7209979408b0b5fe565b8eeac6c0d419465844a91c07964b232f8f34931e115e.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#168833|sha256:eb1b6692c428a68cf15c9f1b0f09ab7b5703297b33c00c4ce4f557b190ee2162|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:eb1b6692c428a68cf15c9f1b0f09ab7b5703297b33c00c4ce4f557b190ee2162",
+    "source_id": "github:cockroachdb/cockroach#168833",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/73/7345a6326b04d0f0d4a8c05145696787159abeb77cfd26c5ff43b5982583b5c7.json b/.cache/impact/classifications/73/7345a6326b04d0f0d4a8c05145696787159abeb77cfd26c5ff43b5982583b5c7.json
new file mode 100644
index 0000000..26b1def
--- /dev/null
+++ b/.cache/impact/classifications/73/7345a6326b04d0f0d4a8c05145696787159abeb77cfd26c5ff43b5982583b5c7.json
@@ -0,0 +1,22 @@
+{
+  "key": "adoption:https://github.com/StarRocks/starrocks/blob/444cb3cf593a8406c096ca79d270908411414654/docs/zh/release_notes/release-2.2.md|sha256:8d14001818916770449fc9ede060ce78ebc09b8abeadca667d38b3ac12ab5d99|adoption-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T06:22:28Z",
+  "value": {
+    "classified_at": "2026-09-13T06:22:28Z",
+    "classifier_version": "adoption-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "relationship": null
+      },
+      "reason": "A translated copy of the same release notes; the statement is recorded once, from the English file. Decided by reading the file in this session rather than by an API call."
+    },
+    "source_hash": "sha256:8d14001818916770449fc9ede060ce78ebc09b8abeadca667d38b3ac12ab5d99",
+    "source_id": "adoption:https://github.com/StarRocks/starrocks/blob/444cb3cf593a8406c096ca79d270908411414654/docs/zh/release_notes/release-2.2.md",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/73/735e01432f9c25713423d8f7d2d2bb852908789c81a8499eaf406680d029b77f.json b/.cache/impact/classifications/73/735e01432f9c25713423d8f7d2d2bb852908789c81a8499eaf406680d029b77f.json
new file mode 100644
index 0000000..fd6d35a
--- /dev/null
+++ b/.cache/impact/classifications/73/735e01432f9c25713423d8f7d2d2bb852908789c81a8499eaf406680d029b77f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4688|sha256:efdddfce48da18ec8c189e0370e58adfd330e879cd11778a05d0547f0161683c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:efdddfce48da18ec8c189e0370e58adfd330e879cd11778a05d0547f0161683c",
+    "source_id": "github:sparq-org/sparq#4688",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/74/7438ee65aad5351bd98e030d6ee054fb1fda43be48b9bc5b0769fe575e3376ea.json b/.cache/impact/classifications/74/7438ee65aad5351bd98e030d6ee054fb1fda43be48b9bc5b0769fe575e3376ea.json
new file mode 100644
index 0000000..757f5fc
--- /dev/null
+++ b/.cache/impact/classifications/74/7438ee65aad5351bd98e030d6ee054fb1fda43be48b9bc5b0769fe575e3376ea.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#4118|sha256:853e513ac9b54f097d1dc55a40453c93833e89be0e585f45557c90a5ad858212|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:853e513ac9b54f097d1dc55a40453c93833e89be0e585f45557c90a5ad858212",
+    "source_id": "github:cockroachdb/cockroach#4118",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/75/7508f0afcca40df3864c18de3deda469f3b34bc5bdceb471e333075a80310eeb.json b/.cache/impact/classifications/75/7508f0afcca40df3864c18de3deda469f3b34bc5bdceb471e333075a80310eeb.json
new file mode 100644
index 0000000..f3f415b
--- /dev/null
+++ b/.cache/impact/classifications/75/7508f0afcca40df3864c18de3deda469f3b34bc5bdceb471e333075a80310eeb.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cmu-db/noisepage#997|sha256:2cbb256256fed2f240ffeb49cc022edfd2b8678e7e78a276fb6623cd52ace275|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2cbb256256fed2f240ffeb49cc022edfd2b8678e7e78a276fb6623cd52ace275",
+    "source_id": "github:cmu-db/noisepage#997",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/75/750e0c2d07ecddbf3756bd9320cf5fd7b851eefce04c4d7a01e32377b9f2973d.json b/.cache/impact/classifications/75/750e0c2d07ecddbf3756bd9320cf5fd7b851eefce04c4d7a01e32377b9f2973d.json
new file mode 100644
index 0000000..87ded63
--- /dev/null
+++ b/.cache/impact/classifications/75/750e0c2d07ecddbf3756bd9320cf5fd7b851eefce04c4d7a01e32377b9f2973d.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#3087|sha256:cf12f4e145e197e91abdf792b8f014d5f3a543eb665f133131ca8821f1c0fc84|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:cf12f4e145e197e91abdf792b8f014d5f3a543eb665f133131ca8821f1c0fc84",
+    "source_id": "github:sparq-org/sparq#3087",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/75/755534f2ed595dd03c863dfc1304974e1454b93933976bd181a4b53272027588.json b/.cache/impact/classifications/75/755534f2ed595dd03c863dfc1304974e1454b93933976bd181a4b53272027588.json
new file mode 100644
index 0000000..1ec9b37
--- /dev/null
+++ b/.cache/impact/classifications/75/755534f2ed595dd03c863dfc1304974e1454b93933976bd181a4b53272027588.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#1350|sha256:b9d55094ae67742ed54daf6f2b9bbd25a45bf3075f4295b2453a4b27ac7fb2cc|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b9d55094ae67742ed54daf6f2b9bbd25a45bf3075f4295b2453a4b27ac7fb2cc",
+    "source_id": "github:cockroachdb/cockroach#1350",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/75/75a72c5e43799268ab92490f6af0a28443642d3dc26a3f1f8027c5b344c7bea2.json b/.cache/impact/classifications/75/75a72c5e43799268ab92490f6af0a28443642d3dc26a3f1f8027c5b344c7bea2.json
new file mode 100644
index 0000000..f9fa606
--- /dev/null
+++ b/.cache/impact/classifications/75/75a72c5e43799268ab92490f6af0a28443642d3dc26a3f1f8027c5b344c7bea2.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#72648|sha256:1beec9dcbeb7e062b1614f23506ebcef05862763e4f07ce950f8a27609630669|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1beec9dcbeb7e062b1614f23506ebcef05862763e4f07ce950f8a27609630669",
+    "source_id": "github:cockroachdb/cockroach#72648",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/76/764a6bb919a2f55c82f1a2118771bd9f15c08f9d05a6ebd520f59dab274c9da5.json b/.cache/impact/classifications/76/764a6bb919a2f55c82f1a2118771bd9f15c08f9d05a6ebd520f59dab274c9da5.json
new file mode 100644
index 0000000..375a3af
--- /dev/null
+++ b/.cache/impact/classifications/76/764a6bb919a2f55c82f1a2118771bd9f15c08f9d05a6ebd520f59dab274c9da5.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#2131|sha256:7319ca76263a55b1ddd77b2e1f2ddfe4236a76fc2a155661d3654ad1b7f32968|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7319ca76263a55b1ddd77b2e1f2ddfe4236a76fc2a155661d3654ad1b7f32968",
+    "source_id": "github:cockroachdb/cockroach#2131",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/77/773f5076e2392a3f601ba64e864acb050b2f2c6e8572bd314c8dc1ea6f8f9acc.json b/.cache/impact/classifications/77/773f5076e2392a3f601ba64e864acb050b2f2c6e8572bd314c8dc1ea6f8f9acc.json
new file mode 100644
index 0000000..d1518fc
--- /dev/null
+++ b/.cache/impact/classifications/77/773f5076e2392a3f601ba64e864acb050b2f2c6e8572bd314c8dc1ea6f8f9acc.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:apache/datafusion#11030|sha256:fb7752e8606b974929732ea482dfabaeb53d4eaacaae11c52572c4496b99d263|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:fb7752e8606b974929732ea482dfabaeb53d4eaacaae11c52572c4496b99d263",
+    "source_id": "github:apache/datafusion#11030",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/7b/7b846f2c31ff4f840d6d7d8a80eb6dc5828057e2108f1918f3e1078fddd0933b.json b/.cache/impact/classifications/7b/7b846f2c31ff4f840d6d7d8a80eb6dc5828057e2108f1918f3e1078fddd0933b.json
new file mode 100644
index 0000000..fced632
--- /dev/null
+++ b/.cache/impact/classifications/7b/7b846f2c31ff4f840d6d7d8a80eb6dc5828057e2108f1918f3e1078fddd0933b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#119849|sha256:ffd0fa42845ddc3989480093813d493289b6b1eb821f238384127c40345ce69c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ffd0fa42845ddc3989480093813d493289b6b1eb821f238384127c40345ce69c",
+    "source_id": "github:cockroachdb/cockroach#119849",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/7b/7bf2bebfb1e0bd2b582dd144451dd56980687ff50e0a2bd1dc622fede603a2b7.json b/.cache/impact/classifications/7b/7bf2bebfb1e0bd2b582dd144451dd56980687ff50e0a2bd1dc622fede603a2b7.json
new file mode 100644
index 0000000..fcd105e
--- /dev/null
+++ b/.cache/impact/classifications/7b/7bf2bebfb1e0bd2b582dd144451dd56980687ff50e0a2bd1dc622fede603a2b7.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:apache/doris#50342|sha256:5016d5e451b830288c0285372f890efccebb5693fd8c5fb4341e2c97e24cab0a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5016d5e451b830288c0285372f890efccebb5693fd8c5fb4341e2c97e24cab0a",
+    "source_id": "github:apache/doris#50342",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/7d/7dd6bf46431916627e40258d7ab9fc21a5158241afe6d087184f9076f97233bc.json b/.cache/impact/classifications/7d/7dd6bf46431916627e40258d7ab9fc21a5158241afe6d087184f9076f97233bc.json
new file mode 100644
index 0000000..fa783d2
--- /dev/null
+++ b/.cache/impact/classifications/7d/7dd6bf46431916627e40258d7ab9fc21a5158241afe6d087184f9076f97233bc.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4570|sha256:a84581172ad0fe5ba2be0f98efd927f763bc11ccd605598e34e0f057fa301152|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a84581172ad0fe5ba2be0f98efd927f763bc11ccd605598e34e0f057fa301152",
+    "source_id": "github:sparq-org/sparq#4570",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/7d/7dd7dc5c4126a27c7402c93659c777726ebdbae2082ebc55ae80c9e1f0e4fae4.json b/.cache/impact/classifications/7d/7dd7dc5c4126a27c7402c93659c777726ebdbae2082ebc55ae80c9e1f0e4fae4.json
new file mode 100644
index 0000000..d3b6cd6
--- /dev/null
+++ b/.cache/impact/classifications/7d/7dd7dc5c4126a27c7402c93659c777726ebdbae2082ebc55ae80c9e1f0e4fae4.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:hazelcast/hazelcast#19864|sha256:0355300568bf9bb2f9284d89b6267db0e237e1f4e8c2852279da6b30adfcfae2|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "approach, various wrong query results"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0355300568bf9bb2f9284d89b6267db0e237e1f4e8c2852279da6b30adfcfae2",
+    "source_id": "github:hazelcast/hazelcast#19864",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/7e/7e873b812d4827a72cd6ae34568928233ce19c5044f6c3df1f260c45369e3f8e.json b/.cache/impact/classifications/7e/7e873b812d4827a72cd6ae34568928233ce19c5044f6c3df1f260c45369e3f8e.json
new file mode 100644
index 0000000..d4176dc
--- /dev/null
+++ b/.cache/impact/classifications/7e/7e873b812d4827a72cd6ae34568928233ce19c5044f6c3df1f260c45369e3f8e.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#43974|sha256:2bb4c103d68053912fb497f567ac0938a2bce5e3f0deeeb8c5896e7ea41c38ba|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2bb4c103d68053912fb497f567ac0938a2bce5e3f0deeeb8c5896e7ea41c38ba",
+    "source_id": "github:cockroachdb/cockroach#43974",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/7e/7ec3d0f838b6c8c603a651c0110eacac1e9309274de671a65553bf4eeb9c40bc.json b/.cache/impact/classifications/7e/7ec3d0f838b6c8c603a651c0110eacac1e9309274de671a65553bf4eeb9c40bc.json
new file mode 100644
index 0000000..0c634f1
--- /dev/null
+++ b/.cache/impact/classifications/7e/7ec3d0f838b6c8c603a651c0110eacac1e9309274de671a65553bf4eeb9c40bc.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#127745|sha256:086c99ab0fe562b1b3dcd88c508972f3bcda35784f84564f6ee387c3878f7f05|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:086c99ab0fe562b1b3dcd88c508972f3bcda35784f84564f6ee387c3878f7f05",
+    "source_id": "github:cockroachdb/cockroach#127745",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/7e/7ef333611a6dba21078b3cfe7084bea188a4f6de04864290c784177fc02202d6.json b/.cache/impact/classifications/7e/7ef333611a6dba21078b3cfe7084bea188a4f6de04864290c784177fc02202d6.json
new file mode 100644
index 0000000..bca83d1
--- /dev/null
+++ b/.cache/impact/classifications/7e/7ef333611a6dba21078b3cfe7084bea188a4f6de04864290c784177fc02202d6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:risingwavelabs/risingwave#6266|sha256:05247c8868655eebbcdbd6054c37fd3286d1ff6d78fb1903a87540070639cc10|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:05247c8868655eebbcdbd6054c37fd3286d1ff6d78fb1903a87540070639cc10",
+    "source_id": "github:risingwavelabs/risingwave#6266",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/7f/7fc40a728d23d21f53df5bca854a0347b4beaf1fe3751df8315cd879ba619b6d.json b/.cache/impact/classifications/7f/7fc40a728d23d21f53df5bca854a0347b4beaf1fe3751df8315cd879ba619b6d.json
new file mode 100644
index 0000000..ccedd3c
--- /dev/null
+++ b/.cache/impact/classifications/7f/7fc40a728d23d21f53df5bca854a0347b4beaf1fe3751df8315cd879ba619b6d.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#131978|sha256:984d0eb5f3d48234ec7fd5a7781a0b521a41df90f160f419140574d3604775af|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:984d0eb5f3d48234ec7fd5a7781a0b521a41df90f160f419140574d3604775af",
+    "source_id": "github:cockroachdb/cockroach#131978",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/7f/7fcac3447c153419b060df1eaa1074fa1f8c21aefc9ab7b013a897182a9003ed.json b/.cache/impact/classifications/7f/7fcac3447c153419b060df1eaa1074fa1f8c21aefc9ab7b013a897182a9003ed.json
new file mode 100644
index 0000000..cd99d35
--- /dev/null
+++ b/.cache/impact/classifications/7f/7fcac3447c153419b060df1eaa1074fa1f8c21aefc9ab7b013a897182a9003ed.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#9178|sha256:0a4dca9a7884f858461d15fc01a322f2fb0a95d623af1f918d700f3837e35502|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0a4dca9a7884f858461d15fc01a322f2fb0a95d623af1f918d700f3837e35502",
+    "source_id": "github:cockroachdb/cockroach#9178",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/80/801af6ace6e7ee0ce4c53414a8925d6de29249f25b5365f29dd3ec6c1541ed68.json b/.cache/impact/classifications/80/801af6ace6e7ee0ce4c53414a8925d6de29249f25b5365f29dd3ec6c1541ed68.json
new file mode 100644
index 0000000..e6f5652
--- /dev/null
+++ b/.cache/impact/classifications/80/801af6ace6e7ee0ce4c53414a8925d6de29249f25b5365f29dd3ec6c1541ed68.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#79290|sha256:37c9cfd644ffba576aac6266f4248bda7334b46746e16537f2fde81fff99fdac|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:37c9cfd644ffba576aac6266f4248bda7334b46746e16537f2fde81fff99fdac",
+    "source_id": "github:cockroachdb/cockroach#79290",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/80/8022d5a14b465bca5b1b0310426f70750014a42250ff436ea9649b184f59823a.json b/.cache/impact/classifications/80/8022d5a14b465bca5b1b0310426f70750014a42250ff436ea9649b184f59823a.json
new file mode 100644
index 0000000..518ec33
--- /dev/null
+++ b/.cache/impact/classifications/80/8022d5a14b465bca5b1b0310426f70750014a42250ff436ea9649b184f59823a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#6474|sha256:f22f0453b2200c37ce21ae07359c4bbd883a786a8432a98056c3ac54c7dc11b7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f22f0453b2200c37ce21ae07359c4bbd883a786a8432a98056c3ac54c7dc11b7",
+    "source_id": "github:cockroachdb/cockroach#6474",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/80/803e0602843e8a5e76af7788a32794f748a0769b73eb8dd8a6663e2c28a77bee.json b/.cache/impact/classifications/80/803e0602843e8a5e76af7788a32794f748a0769b73eb8dd8a6663e2c28a77bee.json
new file mode 100644
index 0000000..c8eeda0
--- /dev/null
+++ b/.cache/impact/classifications/80/803e0602843e8a5e76af7788a32794f748a0769b73eb8dd8a6663e2c28a77bee.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#6619|sha256:d5cb4288e7950afbf41314f85270368c8c39728706f54ff3422f97b24e043fff|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d5cb4288e7950afbf41314f85270368c8c39728706f54ff3422f97b24e043fff",
+    "source_id": "github:yugabyte/yugabyte-db#6619",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/80/807d91847bedd5741399aa00b7fa1ea6477c2de0748d14c5992b1342e5ed27df.json b/.cache/impact/classifications/80/807d91847bedd5741399aa00b7fa1ea6477c2de0748d14c5992b1342e5ed27df.json
new file mode 100644
index 0000000..7d1c415
--- /dev/null
+++ b/.cache/impact/classifications/80/807d91847bedd5741399aa00b7fa1ea6477c2de0748d14c5992b1342e5ed27df.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4571|sha256:22574298f13364d0621cb9299b527c647d1716f6fd743af27e67f9425ea6c031|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:22574298f13364d0621cb9299b527c647d1716f6fd743af27e67f9425ea6c031",
+    "source_id": "github:sparq-org/sparq#4571",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/81/81d5b3b459826435ff3774c6556abb01c62dc95348ca66c529eed100c31094e5.json b/.cache/impact/classifications/81/81d5b3b459826435ff3774c6556abb01c62dc95348ca66c529eed100c31094e5.json
new file mode 100644
index 0000000..31c4787
--- /dev/null
+++ b/.cache/impact/classifications/81/81d5b3b459826435ff3774c6556abb01c62dc95348ca66c529eed100c31094e5.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#144775|sha256:b352954a114035f97f3c431632f30397403277d6fbd13454d85f9e278cfedbe5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b352954a114035f97f3c431632f30397403277d6fbd13454d85f9e278cfedbe5",
+    "source_id": "github:cockroachdb/cockroach#144775",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/82/82c12c179ffe86f4e7631b5e2582784bf3e8422f8bf6004f866fe1f31de1ee39.json b/.cache/impact/classifications/82/82c12c179ffe86f4e7631b5e2582784bf3e8422f8bf6004f866fe1f31de1ee39.json
new file mode 100644
index 0000000..13cd077
--- /dev/null
+++ b/.cache/impact/classifications/82/82c12c179ffe86f4e7631b5e2582784bf3e8422f8bf6004f866fe1f31de1ee39.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#122355|sha256:c59144092b504abd6b7f51c5503bbda7a658e9b9a785acca6ab4e7224c3bdc07|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c59144092b504abd6b7f51c5503bbda7a658e9b9a785acca6ab4e7224c3bdc07",
+    "source_id": "github:cockroachdb/cockroach#122355",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/84/8423ecab3a43b398e2a91a43d32fb90a98e6f4a52ca0a3d0e0a0b965eabaa8aa.json b/.cache/impact/classifications/84/8423ecab3a43b398e2a91a43d32fb90a98e6f4a52ca0a3d0e0a0b965eabaa8aa.json
new file mode 100644
index 0000000..c63ad0b
--- /dev/null
+++ b/.cache/impact/classifications/84/8423ecab3a43b398e2a91a43d32fb90a98e6f4a52ca0a3d0e0a0b965eabaa8aa.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#144366|sha256:76d1cc1c9541c6cdbdec4e0fe00885d2b349a4ce1501a448904ce6bc73446d25|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:76d1cc1c9541c6cdbdec4e0fe00885d2b349a4ce1501a448904ce6bc73446d25",
+    "source_id": "github:cockroachdb/cockroach#144366",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/84/8434ff69970224b52705c886c2a74eb17e84ff3ac778d11f2cc6b2d18e2451a3.json b/.cache/impact/classifications/84/8434ff69970224b52705c886c2a74eb17e84ff3ac778d11f2cc6b2d18e2451a3.json
new file mode 100644
index 0000000..b5250f9
--- /dev/null
+++ b/.cache/impact/classifications/84/8434ff69970224b52705c886c2a74eb17e84ff3ac778d11f2cc6b2d18e2451a3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tarantool/tarantool#4833|sha256:a60a0983d8b02407af9b2741b95196014c667191b01a0ae42834e998bbc6a990|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a60a0983d8b02407af9b2741b95196014c667191b01a0ae42834e998bbc6a990",
+    "source_id": "github:tarantool/tarantool#4833",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/85/850ed109b8ddac3b1692fc286526e6494bc9d79c93bbf7b8509ad89f1d140b3c.json b/.cache/impact/classifications/85/850ed109b8ddac3b1692fc286526e6494bc9d79c93bbf7b8509ad89f1d140b3c.json
new file mode 100644
index 0000000..999241c
--- /dev/null
+++ b/.cache/impact/classifications/85/850ed109b8ddac3b1692fc286526e6494bc9d79c93bbf7b8509ad89f1d140b3c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#93494|sha256:d98276615f8c271d1fd1073d0d4f34532a85a21058ea3ba3e1d0b33b42b3dda7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d98276615f8c271d1fd1073d0d4f34532a85a21058ea3ba3e1d0b33b42b3dda7",
+    "source_id": "github:elastic/elasticsearch#93494",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/85/8556425a369e9f40726b3c826d8956a3076269d2f5ae1d435ea291177b9d95a4.json b/.cache/impact/classifications/85/8556425a369e9f40726b3c826d8956a3076269d2f5ae1d435ea291177b9d95a4.json
new file mode 100644
index 0000000..0756e10
--- /dev/null
+++ b/.cache/impact/classifications/85/8556425a369e9f40726b3c826d8956a3076269d2f5ae1d435ea291177b9d95a4.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#126063|sha256:630307f708b04913f40b827126b5033f98194915fb9e596970f0cf6ee5f688e7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:630307f708b04913f40b827126b5033f98194915fb9e596970f0cf6ee5f688e7",
+    "source_id": "github:cockroachdb/cockroach#126063",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/86/8602b7af2e515160ea05a7b12aebe296f3ea0a789066736f46d810abb83f37e1.json b/.cache/impact/classifications/86/8602b7af2e515160ea05a7b12aebe296f3ea0a789066736f46d810abb83f37e1.json
new file mode 100644
index 0000000..c844c51
--- /dev/null
+++ b/.cache/impact/classifications/86/8602b7af2e515160ea05a7b12aebe296f3ea0a789066736f46d810abb83f37e1.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#68979|sha256:f0a39ed2f3ae5e8aecec8232a1568bd8fa52a9ae7aec41ee9c429c6574c7bff2|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "sql: vectorized IN evaluation incorrectly assumes the RHS tuple contents"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The TLP roachtest found this defect; the issue was retitled to name it. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f0a39ed2f3ae5e8aecec8232a1568bd8fa52a9ae7aec41ee9c429c6574c7bff2",
+    "source_id": "github:cockroachdb/cockroach#68979",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/86/86ead30bddb9719bf7d5d4b5ef804ba894d5e84837c731e5cba2ea31d732c4d1.json b/.cache/impact/classifications/86/86ead30bddb9719bf7d5d4b5ef804ba894d5e84837c731e5cba2ea31d732c4d1.json
new file mode 100644
index 0000000..c862212
--- /dev/null
+++ b/.cache/impact/classifications/86/86ead30bddb9719bf7d5d4b5ef804ba894d5e84837c731e5cba2ea31d732c4d1.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tursodatabase/turso#4681|sha256:e36c64ecb570ef39d39218880d14678c663e596cee65bb85ec514debc4afd87b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e36c64ecb570ef39d39218880d14678c663e596cee65bb85ec514debc4afd87b",
+    "source_id": "github:tursodatabase/turso#4681",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/87/8726b52251c2dc28e8466c6c27ace4f7d28da3c7daaf6c69e2c038842a739c01.json b/.cache/impact/classifications/87/8726b52251c2dc28e8466c6c27ace4f7d28da3c7daaf6c69e2c038842a739c01.json
new file mode 100644
index 0000000..c0b8b8b
--- /dev/null
+++ b/.cache/impact/classifications/87/8726b52251c2dc28e8466c6c27ace4f7d28da3c7daaf6c69e2c038842a739c01.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#4190|sha256:95148b13045729bd4859e44d2b701895b277987f5a16ce5e9c11e8443d9f724d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:95148b13045729bd4859e44d2b701895b277987f5a16ce5e9c11e8443d9f724d",
+    "source_id": "github:cockroachdb/cockroach#4190",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/87/879ba31e10cb142b745e3502965a70e5ec6dfb07090bff27b629c794f1877666.json b/.cache/impact/classifications/87/879ba31e10cb142b745e3502965a70e5ec6dfb07090bff27b629c794f1877666.json
new file mode 100644
index 0000000..dcfcb70
--- /dev/null
+++ b/.cache/impact/classifications/87/879ba31e10cb142b745e3502965a70e5ec6dfb07090bff27b629c794f1877666.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#29002|sha256:1a549d2d21eefd857c7460b9fcf15ee53a1b4b8beb4e0b4d05489234e580284e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1a549d2d21eefd857c7460b9fcf15ee53a1b4b8beb4e0b4d05489234e580284e",
+    "source_id": "github:pingcap/tidb#29002",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/89/891d5e8dbccf874724fc0cacad2b3174b09de62698e484236700499e128eab50.json b/.cache/impact/classifications/89/891d5e8dbccf874724fc0cacad2b3174b09de62698e484236700499e128eab50.json
new file mode 100644
index 0000000..9c197e9
--- /dev/null
+++ b/.cache/impact/classifications/89/891d5e8dbccf874724fc0cacad2b3174b09de62698e484236700499e128eab50.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4114|sha256:6f5f27ba48977730725b1b6fc5c5c68dd4c09a909c1348b1e1cfccae7578d2a5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6f5f27ba48977730725b1b6fc5c5c68dd4c09a909c1348b1e1cfccae7578d2a5",
+    "source_id": "github:sparq-org/sparq#4114",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/89/8971a06533e5efa711b6fde9fdb3ffce68b27173b695ad21edf67dcff5f2a0ad.json b/.cache/impact/classifications/89/8971a06533e5efa711b6fde9fdb3ffce68b27173b695ad21edf67dcff5f2a0ad.json
new file mode 100644
index 0000000..1d4fe78
--- /dev/null
+++ b/.cache/impact/classifications/89/8971a06533e5efa711b6fde9fdb3ffce68b27173b695ad21edf67dcff5f2a0ad.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#100353|sha256:8eb2f0fa392774dd99578275d5658580ef441c6c5250af0678368a97ef2ca81f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8eb2f0fa392774dd99578275d5658580ef441c6c5250af0678368a97ef2ca81f",
+    "source_id": "github:cockroachdb/cockroach#100353",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/8a/8a5dbb6202f5792b2e6b2b45d7ba4d35cbe2f42e7d1f77072c5c1f11dc318230.json b/.cache/impact/classifications/8a/8a5dbb6202f5792b2e6b2b45d7ba4d35cbe2f42e7d1f77072c5c1f11dc318230.json
new file mode 100644
index 0000000..acad7b3
--- /dev/null
+++ b/.cache/impact/classifications/8a/8a5dbb6202f5792b2e6b2b45d7ba4d35cbe2f42e7d1f77072c5c1f11dc318230.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#168225|sha256:27e2bb255f192061c6af4b93138d46428a814c1a2e6f1dd3262aa6bbbb8431e2|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:27e2bb255f192061c6af4b93138d46428a814c1a2e6f1dd3262aa6bbbb8431e2",
+    "source_id": "github:cockroachdb/cockroach#168225",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/8b/8b20d95616bce65309ccff2bcfae97801f823c3e2e591514ecaddf021f764274.json b/.cache/impact/classifications/8b/8b20d95616bce65309ccff2bcfae97801f823c3e2e591514ecaddf021f764274.json
new file mode 100644
index 0000000..79390b7
--- /dev/null
+++ b/.cache/impact/classifications/8b/8b20d95616bce65309ccff2bcfae97801f823c3e2e591514ecaddf021f764274.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#83756|sha256:9a527180ea8e6ae3af8eea6c08bc6f0760c892f90f246af5caee9a66486939bb|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9a527180ea8e6ae3af8eea6c08bc6f0760c892f90f246af5caee9a66486939bb",
+    "source_id": "github:cockroachdb/cockroach#83756",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/8b/8bcf0e49637f879d974525a587efa8223ee03cc99019ba2d4046b2b9cec32cc0.json b/.cache/impact/classifications/8b/8bcf0e49637f879d974525a587efa8223ee03cc99019ba2d4046b2b9cec32cc0.json
new file mode 100644
index 0000000..c032086
--- /dev/null
+++ b/.cache/impact/classifications/8b/8bcf0e49637f879d974525a587efa8223ee03cc99019ba2d4046b2b9cec32cc0.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#168832|sha256:7bfe01dd1fe0872c3d8ec4a23183b9a50ef0046b416d1b0d06f903960f64ca3f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7bfe01dd1fe0872c3d8ec4a23183b9a50ef0046b416d1b0d06f903960f64ca3f",
+    "source_id": "github:cockroachdb/cockroach#168832",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/8c/8c58f6be1b45691393e27ac8cb9efa758494e5cce9e69cc646093bd6f2d8788c.json b/.cache/impact/classifications/8c/8c58f6be1b45691393e27ac8cb9efa758494e5cce9e69cc646093bd6f2d8788c.json
new file mode 100644
index 0000000..39730c7
--- /dev/null
+++ b/.cache/impact/classifications/8c/8c58f6be1b45691393e27ac8cb9efa758494e5cce9e69cc646093bd6f2d8788c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:apache/datafusion#11106|sha256:230f1354f4fb14e4047fe9d5a2366c4ed6df7ea3342d0210b8ec97fc607ae430|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:230f1354f4fb14e4047fe9d5a2366c4ed6df7ea3342d0210b8ec97fc607ae430",
+    "source_id": "github:apache/datafusion#11106",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/8c/8c6aa66934923e5d761ebaed2ec32bfaaa3185895c0fb0a3b9c99d829a653221.json b/.cache/impact/classifications/8c/8c6aa66934923e5d761ebaed2ec32bfaaa3185895c0fb0a3b9c99d829a653221.json
new file mode 100644
index 0000000..d2b97a3
--- /dev/null
+++ b/.cache/impact/classifications/8c/8c6aa66934923e5d761ebaed2ec32bfaaa3185895c0fb0a3b9c99d829a653221.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#75853|sha256:d7bb09d9b8d18072be2df772b0e739f6c1888f8fa586d8795a379cd9035db7e0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d7bb09d9b8d18072be2df772b0e739f6c1888f8fa586d8795a379cd9035db7e0",
+    "source_id": "github:cockroachdb/cockroach#75853",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/8d/8da1f7e2c9c7ba8902730bca6c28ab7a5c6d6505275ace026e9265d1dd8c41ff.json b/.cache/impact/classifications/8d/8da1f7e2c9c7ba8902730bca6c28ab7a5c6d6505275ace026e9265d1dd8c41ff.json
new file mode 100644
index 0000000..7cc255d
--- /dev/null
+++ b/.cache/impact/classifications/8d/8da1f7e2c9c7ba8902730bca6c28ab7a5c6d6505275ace026e9265d1dd8c41ff.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#129181|sha256:38c3d40ef3e10b0149ddc3a3fd28c0c223ed56a4c31c7f15813e987f76fda7f6|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:38c3d40ef3e10b0149ddc3a3fd28c0c223ed56a4c31c7f15813e987f76fda7f6",
+    "source_id": "github:cockroachdb/cockroach#129181",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/8d/8dbb4879c6236a8b4595c7a655f4f977133c3b608c7c4b211242a7d7e2838b08.json b/.cache/impact/classifications/8d/8dbb4879c6236a8b4595c7a655f4f977133c3b608c7c4b211242a7d7e2838b08.json
new file mode 100644
index 0000000..1c0ff4f
--- /dev/null
+++ b/.cache/impact/classifications/8d/8dbb4879c6236a8b4595c7a655f4f977133c3b608c7c4b211242a7d7e2838b08.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#78492|sha256:b58278c8b3b405905e4d083a0ffa80f62779a99587d06d25bb28661b692e53a7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b58278c8b3b405905e4d083a0ffa80f62779a99587d06d25bb28661b692e53a7",
+    "source_id": "github:cockroachdb/cockroach#78492",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/8f/8f40afd50f453231b27624beb8d697c73b7767b98c7e2a9392e4a54ab7b6388f.json b/.cache/impact/classifications/8f/8f40afd50f453231b27624beb8d697c73b7767b98c7e2a9392e4a54ab7b6388f.json
new file mode 100644
index 0000000..f8aa013
--- /dev/null
+++ b/.cache/impact/classifications/8f/8f40afd50f453231b27624beb8d697c73b7767b98c7e2a9392e4a54ab7b6388f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#12432|sha256:1770be1be788262ae111cc5b177fee5d495b7ec41b7d3026366fadfdcec2ee82|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1770be1be788262ae111cc5b177fee5d495b7ec41b7d3026366fadfdcec2ee82",
+    "source_id": "github:cockroachdb/cockroach#12432",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/8f/8fb2016b3f2e36ed1a57cd26db236531315ea87144e4c018032c4a696b86c642.json b/.cache/impact/classifications/8f/8fb2016b3f2e36ed1a57cd26db236531315ea87144e4c018032c4a696b86c642.json
new file mode 100644
index 0000000..606f5e1
--- /dev/null
+++ b/.cache/impact/classifications/8f/8fb2016b3f2e36ed1a57cd26db236531315ea87144e4c018032c4a696b86c642.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tikv/tikv#4291|sha256:d2ac69c026555f93fda26173329a2b0e647b72742c7aba4877012903635465d4|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d2ac69c026555f93fda26173329a2b0e647b72742c7aba4877012903635465d4",
+    "source_id": "github:tikv/tikv#4291",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/90/9046611b92aa5ef719ab092484baafbfa9c0fc5e1f7dc476599d2ae4bf682cc6.json b/.cache/impact/classifications/90/9046611b92aa5ef719ab092484baafbfa9c0fc5e1f7dc476599d2ae4bf682cc6.json
new file mode 100644
index 0000000..be1a244
--- /dev/null
+++ b/.cache/impact/classifications/90/9046611b92aa5ef719ab092484baafbfa9c0fc5e1f7dc476599d2ae4bf682cc6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#171414|sha256:862fd5b4a7daea4898e3885a99cadb0baa023491d8cee3e7cbbf3a964bf2aeb8|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:862fd5b4a7daea4898e3885a99cadb0baa023491d8cee3e7cbbf3a964bf2aeb8",
+    "source_id": "github:cockroachdb/cockroach#171414",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/92/927e53c502ebdfac4f0135d8bed711fbb79c8215d432afcacf647c44f2a002fc.json b/.cache/impact/classifications/92/927e53c502ebdfac4f0135d8bed711fbb79c8215d432afcacf647c44f2a002fc.json
new file mode 100644
index 0000000..fd0b296
--- /dev/null
+++ b/.cache/impact/classifications/92/927e53c502ebdfac4f0135d8bed711fbb79c8215d432afcacf647c44f2a002fc.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:apache/datafusion#11190|sha256:c06b5fc402b058c4e6037766b93a81cbf45bb20545bfd30e1b8260c92b317b67|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c06b5fc402b058c4e6037766b93a81cbf45bb20545bfd30e1b8260c92b317b67",
+    "source_id": "github:apache/datafusion#11190",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/92/928f6156c971928f57c89ed50460c9d46c45749849c932ce0604c8a7332f67ab.json b/.cache/impact/classifications/92/928f6156c971928f57c89ed50460c9d46c45749849c932ce0604c8a7332f67ab.json
new file mode 100644
index 0000000..05234ce
--- /dev/null
+++ b/.cache/impact/classifications/92/928f6156c971928f57c89ed50460c9d46c45749849c932ce0604c8a7332f67ab.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4783|sha256:6189a266c13622f4df7e4b1797dc723c09c0be7174456e9612d4edbfc7af372a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6189a266c13622f4df7e4b1797dc723c09c0be7174456e9612d4edbfc7af372a",
+    "source_id": "github:sparq-org/sparq#4783",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/92/92b35786c9640695e0cd9f9ded6473c0004c87f12095a68145ed0da631c06ea7.json b/.cache/impact/classifications/92/92b35786c9640695e0cd9f9ded6473c0004c87f12095a68145ed0da631c06ea7.json
new file mode 100644
index 0000000..86e0a52
--- /dev/null
+++ b/.cache/impact/classifications/92/92b35786c9640695e0cd9f9ded6473c0004c87f12095a68145ed0da631c06ea7.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:neo4j/neo4j#13276|sha256:69f2b8ccd1daa63277955883528d24d9fb523b7762c7ed5399730d415e15fe78|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T05:53:37Z",
+  "value": {
+    "classified_at": "2026-09-13T05:53:37Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Cypher pattern partitioning found during the reporter's own testing, with no tool or SQLancer-originated oracle named. SQLancer has no provider for a graph system. Decided by reading the issue in this session rather than by an API call."
+    },
+    "source_hash": "sha256:69f2b8ccd1daa63277955883528d24d9fb523b7762c7ed5399730d415e15fe78",
+    "source_id": "github:neo4j/neo4j#13276",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/92/92ba6b3b4355d6e9d0c8595316564e196be6b861922f5818189d7c89e399e330.json b/.cache/impact/classifications/92/92ba6b3b4355d6e9d0c8595316564e196be6b861922f5818189d7c89e399e330.json
new file mode 100644
index 0000000..5241cf3
--- /dev/null
+++ b/.cache/impact/classifications/92/92ba6b3b4355d6e9d0c8595316564e196be6b861922f5818189d7c89e399e330.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#7419|sha256:a1e924e08c46e68dbe7ad76113d25ec1b9bbbe12135ac06a517ea97453fc1f82|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a1e924e08c46e68dbe7ad76113d25ec1b9bbbe12135ac06a517ea97453fc1f82",
+    "source_id": "github:cockroachdb/cockroach#7419",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/94/941e17b9b3021ec7be4bfcef252478fec927aca3d6edbb59dd23d8d274612367.json b/.cache/impact/classifications/94/941e17b9b3021ec7be4bfcef252478fec927aca3d6edbb59dd23d8d274612367.json
new file mode 100644
index 0000000..1afec86
--- /dev/null
+++ b/.cache/impact/classifications/94/941e17b9b3021ec7be4bfcef252478fec927aca3d6edbb59dd23d8d274612367.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#114038|sha256:14b9a0a109731ec7d417e62f4af1d8e94bf3a64d6b19ca4665487fe402113b81|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:14b9a0a109731ec7d417e62f4af1d8e94bf3a64d6b19ca4665487fe402113b81",
+    "source_id": "github:cockroachdb/cockroach#114038",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/94/94c10922dfb6c29ade09f182701f41007d48c112ac71d8f13777a0c8f0bea958.json b/.cache/impact/classifications/94/94c10922dfb6c29ade09f182701f41007d48c112ac71d8f13777a0c8f0bea958.json
new file mode 100644
index 0000000..ee4fffc
--- /dev/null
+++ b/.cache/impact/classifications/94/94c10922dfb6c29ade09f182701f41007d48c112ac71d8f13777a0c8f0bea958.json
@@ -0,0 +1,24 @@
+{
+  "key": "adoption:https://github.com/cnosdb/cnosdb/blob/c7609432edfdefd5a442269af42a99d61d5e87c1/README.md|sha256:ae59d9bae5e65c2169189673a5e1a6c462a6d668ef09cc63b4eebac980bae692|adoption-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T07:17:28Z",
+  "value": {
+    "classified_at": "2026-09-13T07:17:28Z",
+    "classifier_version": "adoption-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "CnosDB 2.0's bug detection is powered by [SQLancer](https://github.com/sqlancer/sqlancer)."
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "relationship": "official_testing"
+      },
+      "reason": "The project's own README says its bug detection is powered by SQLancer. Decided by reading the file in this session rather than by an API call."
+    },
+    "source_hash": "sha256:ae59d9bae5e65c2169189673a5e1a6c462a6d668ef09cc63b4eebac980bae692",
+    "source_id": "adoption:https://github.com/cnosdb/cnosdb/blob/c7609432edfdefd5a442269af42a99d61d5e87c1/README.md",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/94/94e385d7e369bad3192ac17a93c86e606297e158b3f90d6fd03b328c1708e74a.json b/.cache/impact/classifications/94/94e385d7e369bad3192ac17a93c86e606297e158b3f90d6fd03b328c1708e74a.json
new file mode 100644
index 0000000..c006d92
--- /dev/null
+++ b/.cache/impact/classifications/94/94e385d7e369bad3192ac17a93c86e606297e158b3f90d6fd03b328c1708e74a.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#147257|sha256:b71c3960b5f2a02c08f10199c966929f38ebf52244c2be68d074bce654ec8ad1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b71c3960b5f2a02c08f10199c966929f38ebf52244c2be68d074bce654ec8ad1",
+    "source_id": "github:cockroachdb/cockroach#147257",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/95/9527ac9688bdb36a06ba8f91f4d65b1ebec88301f531f6d263679ed61370ce51.json b/.cache/impact/classifications/95/9527ac9688bdb36a06ba8f91f4d65b1ebec88301f531f6d263679ed61370ce51.json
new file mode 100644
index 0000000..1fc9cc1
--- /dev/null
+++ b/.cache/impact/classifications/95/9527ac9688bdb36a06ba8f91f4d65b1ebec88301f531f6d263679ed61370ce51.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8249|sha256:ae112f0c5f402c9ca592d0a3eddf9b3223449b41c840ebd5f602a25e7ba1940f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ae112f0c5f402c9ca592d0a3eddf9b3223449b41c840ebd5f602a25e7ba1940f",
+    "source_id": "github:cockroachdb/cockroach#8249",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/95/95bea481babdfb180b195780b03d0260556bd8c825c9110f96a9a99cd6c6e8f8.json b/.cache/impact/classifications/95/95bea481babdfb180b195780b03d0260556bd8c825c9110f96a9a99cd6c6e8f8.json
new file mode 100644
index 0000000..81ad012
--- /dev/null
+++ b/.cache/impact/classifications/95/95bea481babdfb180b195780b03d0260556bd8c825c9110f96a9a99cd6c6e8f8.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#81885|sha256:4b39431e8b5500043ddfdac3cd6e394ef0d135528677fb03796cf93062542786|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4b39431e8b5500043ddfdac3cd6e394ef0d135528677fb03796cf93062542786",
+    "source_id": "github:cockroachdb/cockroach#81885",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/97/970cfcf4c748bb1fd6a08af1f473539bac44d3554c2fe704bcfa7e4282090456.json b/.cache/impact/classifications/97/970cfcf4c748bb1fd6a08af1f473539bac44d3554c2fe704bcfa7e4282090456.json
new file mode 100644
index 0000000..f043b1d
--- /dev/null
+++ b/.cache/impact/classifications/97/970cfcf4c748bb1fd6a08af1f473539bac44d3554c2fe704bcfa7e4282090456.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#83272|sha256:736ae22e5d332b64e0582d533e46d57b03aa49db961ea2a564287462e7b05542|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:736ae22e5d332b64e0582d533e46d57b03aa49db961ea2a564287462e7b05542",
+    "source_id": "github:cockroachdb/cockroach#83272",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/98/984628086c040a0af4b522cec3fe9c8ad2d5c773c5c0b1de6170e5a73ca71d21.json b/.cache/impact/classifications/98/984628086c040a0af4b522cec3fe9c8ad2d5c773c5c0b1de6170e5a73ca71d21.json
new file mode 100644
index 0000000..6c1e308
--- /dev/null
+++ b/.cache/impact/classifications/98/984628086c040a0af4b522cec3fe9c8ad2d5c773c5c0b1de6170e5a73ca71d21.json
@@ -0,0 +1,22 @@
+{
+  "key": "adoption:https://github.com/spiceai/spiceai/blob/64fc04b9466ce7010a57c067fe7df47b015d62c6/crates/cayenne/tests/correctness/support/sqllancer.rs|sha256:8821f2c781ccb3c7edf90e9061820248e1f93a69ad7d61d4f9bfc2c873c3e275|adoption-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T07:17:28Z",
+  "value": {
+    "classified_at": "2026-09-13T07:17:28Z",
+    "classifier_version": "adoption-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "relationship": null
+      },
+      "reason": "A corpus written in SQLancer's style, not a run of SQLancer: the file vendors its own deterministic queries and says so. Being inspired by the idea is not using the tool. Decided by reading the file in this session rather than by an API call."
+    },
+    "source_hash": "sha256:8821f2c781ccb3c7edf90e9061820248e1f93a69ad7d61d4f9bfc2c873c3e275",
+    "source_id": "adoption:https://github.com/spiceai/spiceai/blob/64fc04b9466ce7010a57c067fe7df47b015d62c6/crates/cayenne/tests/correctness/support/sqllancer.rs",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/98/985e7804d1a19dd025c77c7cbbfc179137cd9b9c636cb5a2730158972e240ea5.json b/.cache/impact/classifications/98/985e7804d1a19dd025c77c7cbbfc179137cd9b9c636cb5a2730158972e240ea5.json
new file mode 100644
index 0000000..0c9f7fb
--- /dev/null
+++ b/.cache/impact/classifications/98/985e7804d1a19dd025c77c7cbbfc179137cd9b9c636cb5a2730158972e240ea5.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#71216|sha256:03cf0b14e1883f0c8b92d4c79a3559988c829ad12236451bb23a2e16edfd4c28|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:03cf0b14e1883f0c8b92d4c79a3559988c829ad12236451bb23a2e16edfd4c28",
+    "source_id": "github:cockroachdb/cockroach#71216",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/99/99325d7c319ae2c84e728030d8eae20bf616cd1908f0757155f9ceb4c6870ef3.json b/.cache/impact/classifications/99/99325d7c319ae2c84e728030d8eae20bf616cd1908f0757155f9ceb4c6870ef3.json
new file mode 100644
index 0000000..8d0bd29
--- /dev/null
+++ b/.cache/impact/classifications/99/99325d7c319ae2c84e728030d8eae20bf616cd1908f0757155f9ceb4c6870ef3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#9318|sha256:2d9b75867794849a28d74444fae01460a6f8d481a652eb394b5fa0b060e955fc|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2d9b75867794849a28d74444fae01460a6f8d481a652eb394b5fa0b060e955fc",
+    "source_id": "github:cockroachdb/cockroach#9318",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/9a/9a755a3f2a71926503ed59d1c1e1c5272bfd3ae4751bf1745eb1625ba9560c79.json b/.cache/impact/classifications/9a/9a755a3f2a71926503ed59d1c1e1c5272bfd3ae4751bf1745eb1625ba9560c79.json
new file mode 100644
index 0000000..21f334d
--- /dev/null
+++ b/.cache/impact/classifications/9a/9a755a3f2a71926503ed59d1c1e1c5272bfd3ae4751bf1745eb1625ba9560c79.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#73938|sha256:9a6ba24378744b2fc4d2b66f0b1e4710609b731ee8dd44e0c0c35e43ea09e1bb|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9a6ba24378744b2fc4d2b66f0b1e4710609b731ee8dd44e0c0c35e43ea09e1bb",
+    "source_id": "github:cockroachdb/cockroach#73938",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/9b/9b84a719086edae541b45ebd1f8c9b8338ccdbde6e1f982c18fa711b90b5a5cf.json b/.cache/impact/classifications/9b/9b84a719086edae541b45ebd1f8c9b8338ccdbde6e1f982c18fa711b90b5a5cf.json
new file mode 100644
index 0000000..9d76ed7
--- /dev/null
+++ b/.cache/impact/classifications/9b/9b84a719086edae541b45ebd1f8c9b8338ccdbde6e1f982c18fa711b90b5a5cf.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:openlink/virtuoso-opensource#1427|sha256:5ce9e949816acb54d815dd1b23a46772182060f610c591976f905954dd1ba49c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "TLP query rewrite over `EXISTS` subquery"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5ce9e949816acb54d815dd1b23a46772182060f610c591976f905954dd1ba49c",
+    "source_id": "github:openlink/virtuoso-opensource#1427",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/9d/9d0f3f911535c1e1e950535292ee5305c63485ab7fe28092b67d8af039c2ba81.json b/.cache/impact/classifications/9d/9d0f3f911535c1e1e950535292ee5305c63485ab7fe28092b67d8af039c2ba81.json
new file mode 100644
index 0000000..6aaf149
--- /dev/null
+++ b/.cache/impact/classifications/9d/9d0f3f911535c1e1e950535292ee5305c63485ab7fe28092b67d8af039c2ba81.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#69665|sha256:77baddafaf1eeac3e70159b16f06e1e2ed6fe2dfd0ba4c8d9aa21ef4901a7a61|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "A TLP failure was reported"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:77baddafaf1eeac3e70159b16f06e1e2ed6fe2dfd0ba4c8d9aa21ef4901a7a61",
+    "source_id": "github:cockroachdb/cockroach#69665",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/9d/9d24b38f02219420d59090a3af76a95e82f357018871daa70c31cd3c8cf7c484.json b/.cache/impact/classifications/9d/9d24b38f02219420d59090a3af76a95e82f357018871daa70c31cd3c8cf7c484.json
new file mode 100644
index 0000000..908b6a8
--- /dev/null
+++ b/.cache/impact/classifications/9d/9d24b38f02219420d59090a3af76a95e82f357018871daa70c31cd3c8cf7c484.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#31665|sha256:b161da8720c76fb4208591f2381927eac402c5b6351fc63b1bb0068eeb6fc43d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b161da8720c76fb4208591f2381927eac402c5b6351fc63b1bb0068eeb6fc43d",
+    "source_id": "github:yugabyte/yugabyte-db#31665",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/9d/9d923a458fce3bc9ab455c772d94b470a09b882fa0d8139591c5b4aa345d20a8.json b/.cache/impact/classifications/9d/9d923a458fce3bc9ab455c772d94b470a09b882fa0d8139591c5b4aa345d20a8.json
new file mode 100644
index 0000000..2de207e
--- /dev/null
+++ b/.cache/impact/classifications/9d/9d923a458fce3bc9ab455c772d94b470a09b882fa0d8139591c5b4aa345d20a8.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#71530|sha256:0c2d7f77b5b75dd08f1c561ad4f4750929d8e1f97ab4294b0c19720dbcd0108d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0c2d7f77b5b75dd08f1c561ad4f4750929d8e1f97ab4294b0c19720dbcd0108d",
+    "source_id": "github:cockroachdb/cockroach#71530",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/9d/9da94908d3a84f0a3551c97bd5d96c3f717d8c845f9c8d28570001b48fae97b3.json b/.cache/impact/classifications/9d/9da94908d3a84f0a3551c97bd5d96c3f717d8c845f9c8d28570001b48fae97b3.json
new file mode 100644
index 0000000..aee7c45
--- /dev/null
+++ b/.cache/impact/classifications/9d/9da94908d3a84f0a3551c97bd5d96c3f717d8c845f9c8d28570001b48fae97b3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:readysettech/readyset#1534|sha256:94bffd580a395ad0884e96564c4c9433e822869a4287c32bf1f7ef1805adcf2e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:94bffd580a395ad0884e96564c4c9433e822869a4287c32bf1f7ef1805adcf2e",
+    "source_id": "github:readysettech/readyset#1534",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/9d/9dc659ab8ecaf1af83e3c44617a4f7713a66a7ea4ecdcbd4ca9a49bf1735693f.json b/.cache/impact/classifications/9d/9dc659ab8ecaf1af83e3c44617a4f7713a66a7ea4ecdcbd4ca9a49bf1735693f.json
new file mode 100644
index 0000000..e7be1f9
--- /dev/null
+++ b/.cache/impact/classifications/9d/9dc659ab8ecaf1af83e3c44617a4f7713a66a7ea4ecdcbd4ca9a49bf1735693f.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#156218|sha256:6969bd5ea06806450a3e5da575eb1bd66e5060e30077ab0eb70399c6912906ed|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6969bd5ea06806450a3e5da575eb1bd66e5060e30077ab0eb70399c6912906ed",
+    "source_id": "github:cockroachdb/cockroach#156218",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/9d/9dd4b6dfc1ae8081c0a8a3b2881d0bea5f83c442bccc5b2e792fe45f190ebce3.json b/.cache/impact/classifications/9d/9dd4b6dfc1ae8081c0a8a3b2881d0bea5f83c442bccc5b2e792fe45f190ebce3.json
new file mode 100644
index 0000000..8811b42
--- /dev/null
+++ b/.cache/impact/classifications/9d/9dd4b6dfc1ae8081c0a8a3b2881d0bea5f83c442bccc5b2e792fe45f190ebce3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#11800|sha256:589a468a770d104c79b2a236c25ce78cb2117cdb3f874abbf774ceb30c650add|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:589a468a770d104c79b2a236c25ce78cb2117cdb3f874abbf774ceb30c650add",
+    "source_id": "github:cockroachdb/cockroach#11800",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/9e/9e917e40c578750b344ff338d19c40e7ea6bee39b6d49888e604414710af5fd6.json b/.cache/impact/classifications/9e/9e917e40c578750b344ff338d19c40e7ea6bee39b6d49888e604414710af5fd6.json
new file mode 100644
index 0000000..de8480d
--- /dev/null
+++ b/.cache/impact/classifications/9e/9e917e40c578750b344ff338d19c40e7ea6bee39b6d49888e604414710af5fd6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#52951|sha256:7ed4f4e22c5bf19befba89fbdca36f8b7db9dae143044be485a23e4f191b631b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7ed4f4e22c5bf19befba89fbdca36f8b7db9dae143044be485a23e4f191b631b",
+    "source_id": "github:cockroachdb/cockroach#52951",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a0/a059465b39743949e73d824476f690827b0985350fdd199e762361c198e8b63d.json b/.cache/impact/classifications/a0/a059465b39743949e73d824476f690827b0985350fdd199e762361c198e8b63d.json
new file mode 100644
index 0000000..027a8db
--- /dev/null
+++ b/.cache/impact/classifications/a0/a059465b39743949e73d824476f690827b0985350fdd199e762361c198e8b63d.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#10473|sha256:dc040c1358255971648289d7ccee534f63a52a4fc32f615cd1b96f2b59f2ec19|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:dc040c1358255971648289d7ccee534f63a52a4fc32f615cd1b96f2b59f2ec19",
+    "source_id": "github:pingcap/tidb#10473",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a0/a05a715b3f6987098dcd7c5b26cb883e7cf402b6ae7fc5fc1f5d9b708530a260.json b/.cache/impact/classifications/a0/a05a715b3f6987098dcd7c5b26cb883e7cf402b6ae7fc5fc1f5d9b708530a260.json
new file mode 100644
index 0000000..f20b393
--- /dev/null
+++ b/.cache/impact/classifications/a0/a05a715b3f6987098dcd7c5b26cb883e7cf402b6ae7fc5fc1f5d9b708530a260.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#4269|sha256:0aede37d51e8f59be6604c9731143f9cd7f38299767e8c2c42513c71b2d0bbc0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0aede37d51e8f59be6604c9731143f9cd7f38299767e8c2c42513c71b2d0bbc0",
+    "source_id": "github:cockroachdb/cockroach#4269",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a0/a05ea93b289167fbf58be981ee436fb5b15898a696974bab1c7c603c39c73bac.json b/.cache/impact/classifications/a0/a05ea93b289167fbf58be981ee436fb5b15898a696974bab1c7c603c39c73bac.json
new file mode 100644
index 0000000..85972f8
--- /dev/null
+++ b/.cache/impact/classifications/a0/a05ea93b289167fbf58be981ee436fb5b15898a696974bab1c7c603c39c73bac.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#43847|sha256:2ed49af7b28537b4dae2cd90313987e1ac84a180fe50c4a10e3f7910c8ef33dd|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2ed49af7b28537b4dae2cd90313987e1ac84a180fe50c4a10e3f7910c8ef33dd",
+    "source_id": "github:cockroachdb/cockroach#43847",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a0/a07a41d826a7ab3ca525d75cff511caaab48472e0902bde71e0e0c73d31216c6.json b/.cache/impact/classifications/a0/a07a41d826a7ab3ca525d75cff511caaab48472e0902bde71e0e0c73d31216c6.json
new file mode 100644
index 0000000..4ff0ced
--- /dev/null
+++ b/.cache/impact/classifications/a0/a07a41d826a7ab3ca525d75cff511caaab48472e0902bde71e0e0c73d31216c6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:apache/datafusion#10403|sha256:cad3d5a2ad9c4993b553a76edd27f2cb9588ba491b3731a7c6ce072bd7d54ea4|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:cad3d5a2ad9c4993b553a76edd27f2cb9588ba491b3731a7c6ce072bd7d54ea4",
+    "source_id": "github:apache/datafusion#10403",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a2/a22ce4f50c68ebc89e701c30ae8f341b013bb7fbfeb3fe3d893434e04fd9cb2f.json b/.cache/impact/classifications/a2/a22ce4f50c68ebc89e701c30ae8f341b013bb7fbfeb3fe3d893434e04fd9cb2f.json
new file mode 100644
index 0000000..0212e1f
--- /dev/null
+++ b/.cache/impact/classifications/a2/a22ce4f50c68ebc89e701c30ae8f341b013bb7fbfeb3fe3d893434e04fd9cb2f.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:openlink/virtuoso-opensource#1428|sha256:1f77ed9c5bf77856723738584f5800d0d8fb8fc9f6c271caf25c2c74b7925d9b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "TLP (tautology-based partitioning) rewritten query"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1f77ed9c5bf77856723738584f5800d0d8fb8fc9f6c271caf25c2c74b7925d9b",
+    "source_id": "github:openlink/virtuoso-opensource#1428",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a3/a350463d5bd989ebea0853a4d00d856f7310e97764a4d1e4f58eaaa7b5e984ed.json b/.cache/impact/classifications/a3/a350463d5bd989ebea0853a4d00d856f7310e97764a4d1e4f58eaaa7b5e984ed.json
new file mode 100644
index 0000000..51de725
--- /dev/null
+++ b/.cache/impact/classifications/a3/a350463d5bd989ebea0853a4d00d856f7310e97764a4d1e4f58eaaa7b5e984ed.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#127468|sha256:b1df85fc10db87c7f26312450b5ad4ec668f5f2897b71af6e39d8142226881f3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b1df85fc10db87c7f26312450b5ad4ec668f5f2897b71af6e39d8142226881f3",
+    "source_id": "github:cockroachdb/cockroach#127468",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a4/a4376eecd04b2ec545fff70a705be13af50cc6b8f08987803f680d4231815739.json b/.cache/impact/classifications/a4/a4376eecd04b2ec545fff70a705be13af50cc6b8f08987803f680d4231815739.json
new file mode 100644
index 0000000..3a87fe0
--- /dev/null
+++ b/.cache/impact/classifications/a4/a4376eecd04b2ec545fff70a705be13af50cc6b8f08987803f680d4231815739.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#54155|sha256:19f4a78d5fbac19a7e2376b163eea0f4e6a61579cbd7c4470fb1de8abe048e7c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:19f4a78d5fbac19a7e2376b163eea0f4e6a61579cbd7c4470fb1de8abe048e7c",
+    "source_id": "github:cockroachdb/cockroach#54155",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a5/a5405c511bd7bea95773b1c4d8dcc736ec2fea497f475bacbad2cae0520f1999.json b/.cache/impact/classifications/a5/a5405c511bd7bea95773b1c4d8dcc736ec2fea497f475bacbad2cae0520f1999.json
new file mode 100644
index 0000000..c5df9e4
--- /dev/null
+++ b/.cache/impact/classifications/a5/a5405c511bd7bea95773b1c4d8dcc736ec2fea497f475bacbad2cae0520f1999.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:yugabyte/yugabyte-db#11090|sha256:e8667ed0f379bbfab6342f8bbabb89d5e710e7ed12b559cab9412fecd3afb150|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "[SQLancer] [YSQL] Autocast to real doesn't work for asc PK"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": null
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e8667ed0f379bbfab6342f8bbabb89d5e710e7ed12b559cab9412fecd3afb150",
+    "source_id": "github:yugabyte/yugabyte-db#11090",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a5/a542536234c761a13dec4e4b48cc3f9fcbfabfc4fce34ae41500dc506213c13c.json b/.cache/impact/classifications/a5/a542536234c761a13dec4e4b48cc3f9fcbfabfc4fce34ae41500dc506213c13c.json
new file mode 100644
index 0000000..b460020
--- /dev/null
+++ b/.cache/impact/classifications/a5/a542536234c761a13dec4e4b48cc3f9fcbfabfc4fce34ae41500dc506213c13c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#106781|sha256:0d733dacb28e4caaa45ec245a26be6146821c6f268ea8bd0cbbad2e87abbdb79|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0d733dacb28e4caaa45ec245a26be6146821c6f268ea8bd0cbbad2e87abbdb79",
+    "source_id": "github:cockroachdb/cockroach#106781",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a5/a581f888af0f3116194eb1ea131aa608c500869dd063f929f098a014dcbe3391.json b/.cache/impact/classifications/a5/a581f888af0f3116194eb1ea131aa608c500869dd063f929f098a014dcbe3391.json
new file mode 100644
index 0000000..02cd75d
--- /dev/null
+++ b/.cache/impact/classifications/a5/a581f888af0f3116194eb1ea131aa608c500869dd063f929f098a014dcbe3391.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:openlink/virtuoso-opensource#1424|sha256:c892c4dbbe00cba0e98a5bfaf9ca3033a2673b7cf48a7adcf81b8b944cfd7b9f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "TLP (tautology-based partitioning) rewritten query"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c892c4dbbe00cba0e98a5bfaf9ca3033a2673b7cf48a7adcf81b8b944cfd7b9f",
+    "source_id": "github:openlink/virtuoso-opensource#1424",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a5/a5952be05636f58793ff5c9d81e4355d7046ca30bafc2470345a440f2fad54a2.json b/.cache/impact/classifications/a5/a5952be05636f58793ff5c9d81e4355d7046ca30bafc2470345a440f2fad54a2.json
new file mode 100644
index 0000000..eacd956
--- /dev/null
+++ b/.cache/impact/classifications/a5/a5952be05636f58793ff5c9d81e4355d7046ca30bafc2470345a440f2fad54a2.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#94612|sha256:47ba2801204118301abdc9f2563af66e177f823340e40d7364fa469b07e90705|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:47ba2801204118301abdc9f2563af66e177f823340e40d7364fa469b07e90705",
+    "source_id": "github:cockroachdb/cockroach#94612",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a5/a5a68b3a609329213704fd68d94b95704632109f0da24ff21b5d3ff378493fb5.json b/.cache/impact/classifications/a5/a5a68b3a609329213704fd68d94b95704632109f0da24ff21b5d3ff378493fb5.json
new file mode 100644
index 0000000..27b7395
--- /dev/null
+++ b/.cache/impact/classifications/a5/a5a68b3a609329213704fd68d94b95704632109f0da24ff21b5d3ff378493fb5.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:ClickHouse/ClickHouse#116422|sha256:f8c6145f618daed2a3c175b630f17b362ae4229b91a5e8f72ff7ddab6e36b6a2|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f8c6145f618daed2a3c175b630f17b362ae4229b91a5e8f72ff7ddab6e36b6a2",
+    "source_id": "github:ClickHouse/ClickHouse#116422",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a6/a6aa13ea8fc1822d7d38cea4e0c1cb92d955e98dd57e2a8f4ba8f385e3e0859e.json b/.cache/impact/classifications/a6/a6aa13ea8fc1822d7d38cea4e0c1cb92d955e98dd57e2a8f4ba8f385e3e0859e.json
new file mode 100644
index 0000000..d7549b9
--- /dev/null
+++ b/.cache/impact/classifications/a6/a6aa13ea8fc1822d7d38cea4e0c1cb92d955e98dd57e2a8f4ba8f385e3e0859e.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:duckdblabs/duckdb-fuzzer-ci#7|sha256:31d6d8a942babb1621dbab10f751894e27704d52fe61be728b930e388d2aa5c4|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T06:22:55Z",
+  "value": {
+    "classified_at": "2026-09-13T06:22:55Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "A proposal to start running SQLancer, not a defect it found. Decided by reading the issue in this session."
+    },
+    "source_hash": "sha256:31d6d8a942babb1621dbab10f751894e27704d52fe61be728b930e388d2aa5c4",
+    "source_id": "github:duckdblabs/duckdb-fuzzer-ci#7",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a6/a6e79380197744a9dd8c8c53cc82e8af354e83dbf7865b135de8f0deb414d75a.json b/.cache/impact/classifications/a6/a6e79380197744a9dd8c8c53cc82e8af354e83dbf7865b135de8f0deb414d75a.json
new file mode 100644
index 0000000..d16a10f
--- /dev/null
+++ b/.cache/impact/classifications/a6/a6e79380197744a9dd8c8c53cc82e8af354e83dbf7865b135de8f0deb414d75a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:risingwavelabs/risingwave#6267|sha256:6a7a9452704511a4635c5ccb99cacb5b15dc94cbaf78d8f6583d4e21a64b8c32|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6a7a9452704511a4635c5ccb99cacb5b15dc94cbaf78d8f6583d4e21a64b8c32",
+    "source_id": "github:risingwavelabs/risingwave#6267",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a7/a79d79ac05a180c8d35d742996f18ec068e225e87ab2db3cd2f877953174ee3c.json b/.cache/impact/classifications/a7/a79d79ac05a180c8d35d742996f18ec068e225e87ab2db3cd2f877953174ee3c.json
new file mode 100644
index 0000000..2b58b72
--- /dev/null
+++ b/.cache/impact/classifications/a7/a79d79ac05a180c8d35d742996f18ec068e225e87ab2db3cd2f877953174ee3c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#94555|sha256:d7dabfca907b1bc87ebe8d927fe8389fe787fb508c2c91b01b5299ef96e6e406|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d7dabfca907b1bc87ebe8d927fe8389fe787fb508c2c91b01b5299ef96e6e406",
+    "source_id": "github:cockroachdb/cockroach#94555",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a7/a7dce9c9c56d61fe2b2387b7d6924c9cab3c5755a0bb181d892114926fa07726.json b/.cache/impact/classifications/a7/a7dce9c9c56d61fe2b2387b7d6924c9cab3c5755a0bb181d892114926fa07726.json
new file mode 100644
index 0000000..4c3ebeb
--- /dev/null
+++ b/.cache/impact/classifications/a7/a7dce9c9c56d61fe2b2387b7d6924c9cab3c5755a0bb181d892114926fa07726.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#69856|sha256:d23a3b8aa87330392985194be3afdf8c7595dab4af8cf47c94a1cd32c09b65df|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d23a3b8aa87330392985194be3afdf8c7595dab4af8cf47c94a1cd32c09b65df",
+    "source_id": "github:cockroachdb/cockroach#69856",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/a9/a9a7fc18600cbb8c66eab007112d20582b6d70c743dd2e020644768ab7c2696e.json b/.cache/impact/classifications/a9/a9a7fc18600cbb8c66eab007112d20582b6d70c743dd2e020644768ab7c2696e.json
new file mode 100644
index 0000000..9537755
--- /dev/null
+++ b/.cache/impact/classifications/a9/a9a7fc18600cbb8c66eab007112d20582b6d70c743dd2e020644768ab7c2696e.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#14026|sha256:ce0d87d101f8b4e26d3fcb84018e486ff0554d9ebe378dce62ad8426040d317d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ce0d87d101f8b4e26d3fcb84018e486ff0554d9ebe378dce62ad8426040d317d",
+    "source_id": "github:cockroachdb/cockroach#14026",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/aa/aa16bd441a8201b60e34e02f80606268fdf6e738498b2fe700a78fff4343bf3d.json b/.cache/impact/classifications/aa/aa16bd441a8201b60e34e02f80606268fdf6e738498b2fe700a78fff4343bf3d.json
new file mode 100644
index 0000000..60b4080
--- /dev/null
+++ b/.cache/impact/classifications/aa/aa16bd441a8201b60e34e02f80606268fdf6e738498b2fe700a78fff4343bf3d.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#5764|sha256:66e32b7b26c3e3b47fa5b323fcba6281ca76d39cb21b93681c24662fadfff2d8|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:66e32b7b26c3e3b47fa5b323fcba6281ca76d39cb21b93681c24662fadfff2d8",
+    "source_id": "github:sparq-org/sparq#5764",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/aa/aa912c95033923570ab9eec850497812d17e662b1c05958526e4e45c1c6d14d3.json b/.cache/impact/classifications/aa/aa912c95033923570ab9eec850497812d17e662b1c05958526e4e45c1c6d14d3.json
new file mode 100644
index 0000000..71d16e0
--- /dev/null
+++ b/.cache/impact/classifications/aa/aa912c95033923570ab9eec850497812d17e662b1c05958526e4e45c1c6d14d3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#173450|sha256:fb160760074e57879759fafd1fd7a7f4b7b29ce2deb7966aca56290533e5e614|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:fb160760074e57879759fafd1fd7a7f4b7b29ce2deb7966aca56290533e5e614",
+    "source_id": "github:cockroachdb/cockroach#173450",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/aa/aa94497921750fd2a1e03ee3c80057d87a3b969e3373d3e69b79db8ad2dd5d81.json b/.cache/impact/classifications/aa/aa94497921750fd2a1e03ee3c80057d87a3b969e3373d3e69b79db8ad2dd5d81.json
new file mode 100644
index 0000000..9af8821
--- /dev/null
+++ b/.cache/impact/classifications/aa/aa94497921750fd2a1e03ee3c80057d87a3b969e3373d3e69b79db8ad2dd5d81.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#811|sha256:baaf56f3967abc67e2dbde7660573958cc0cc54fb2f5b30e898f5aaac0664afe|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:baaf56f3967abc67e2dbde7660573958cc0cc54fb2f5b30e898f5aaac0664afe",
+    "source_id": "github:sparq-org/sparq#811",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ab/ab6839b50d6bc4d1b09c211e440fe63b66e901d8d7e62a849a4160b7d4c33c1c.json b/.cache/impact/classifications/ab/ab6839b50d6bc4d1b09c211e440fe63b66e901d8d7e62a849a4160b7d4c33c1c.json
new file mode 100644
index 0000000..b524333
--- /dev/null
+++ b/.cache/impact/classifications/ab/ab6839b50d6bc4d1b09c211e440fe63b66e901d8d7e62a849a4160b7d4c33c1c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#25165|sha256:381cdf379cc0c0d35e786fd119a2b3f93636e331d23ffaa0612c07d3bba1c407|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:381cdf379cc0c0d35e786fd119a2b3f93636e331d23ffaa0612c07d3bba1c407",
+    "source_id": "github:cockroachdb/cockroach#25165",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ac/ac36cabb569ba5302bf1cc025762a34a405abdae4bb9507487627a4265c36005.json b/.cache/impact/classifications/ac/ac36cabb569ba5302bf1cc025762a34a405abdae4bb9507487627a4265c36005.json
new file mode 100644
index 0000000..e955f9e
--- /dev/null
+++ b/.cache/impact/classifications/ac/ac36cabb569ba5302bf1cc025762a34a405abdae4bb9507487627a4265c36005.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#13234|sha256:9910103cf88773de78fdbeac4fd4c542a51820323dcdc6886c8dba2adb3e8463|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9910103cf88773de78fdbeac4fd4c542a51820323dcdc6886c8dba2adb3e8463",
+    "source_id": "github:yugabyte/yugabyte-db#13234",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ac/ac828d5bf10cbde7137174b35b1e30c7558430fa4c0813fc4cbbcdb1ac2effcd.json b/.cache/impact/classifications/ac/ac828d5bf10cbde7137174b35b1e30c7558430fa4c0813fc4cbbcdb1ac2effcd.json
new file mode 100644
index 0000000..574b741
--- /dev/null
+++ b/.cache/impact/classifications/ac/ac828d5bf10cbde7137174b35b1e30c7558430fa4c0813fc4cbbcdb1ac2effcd.json
@@ -0,0 +1,22 @@
+{
+  "key": "adoption:https://github.com/StarRocks/starrocks/blob/444cb3cf593a8406c096ca79d270908411414654/docs/ja/release_notes/release-2.2.md|sha256:2ba9546aee48fd763de073362357c2081f60594b36f6bc147d1e5a42d29c728f|adoption-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T06:22:25Z",
+  "value": {
+    "classified_at": "2026-09-13T06:22:25Z",
+    "classifier_version": "adoption-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "relationship": null
+      },
+      "reason": "A translated copy of the same release notes; the statement is recorded once, from the English file. Decided by reading the file in this session rather than by an API call."
+    },
+    "source_hash": "sha256:2ba9546aee48fd763de073362357c2081f60594b36f6bc147d1e5a42d29c728f",
+    "source_id": "adoption:https://github.com/StarRocks/starrocks/blob/444cb3cf593a8406c096ca79d270908411414654/docs/ja/release_notes/release-2.2.md",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ac/ac9d4b9049e7857a6c28cd9dfe252beeb2c98d708e52566f4bb244ae0be22edc.json b/.cache/impact/classifications/ac/ac9d4b9049e7857a6c28cd9dfe252beeb2c98d708e52566f4bb244ae0be22edc.json
new file mode 100644
index 0000000..6952ae6
--- /dev/null
+++ b/.cache/impact/classifications/ac/ac9d4b9049e7857a6c28cd9dfe252beeb2c98d708e52566f4bb244ae0be22edc.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#73920|sha256:193778d6f21514b9e1f8bbd53ca2ccf710d577d4aa84829de5cbe48ea3d48332|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:193778d6f21514b9e1f8bbd53ca2ccf710d577d4aa84829de5cbe48ea3d48332",
+    "source_id": "github:cockroachdb/cockroach#73920",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ad/ad6dfdf77ecd6f7e6648dca684da002b8055316c9ea976467e9a09705155f4e1.json b/.cache/impact/classifications/ad/ad6dfdf77ecd6f7e6648dca684da002b8055316c9ea976467e9a09705155f4e1.json
new file mode 100644
index 0000000..444779a
--- /dev/null
+++ b/.cache/impact/classifications/ad/ad6dfdf77ecd6f7e6648dca684da002b8055316c9ea976467e9a09705155f4e1.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#127814|sha256:4197a208281511af9d18ca06611e315b816d1b4721f3139febd4830eaf201284|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "tlp failure: false vs NULL"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The TLP roachtest found this defect; the issue was retitled to name it. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4197a208281511af9d18ca06611e315b816d1b4721f3139febd4830eaf201284",
+    "source_id": "github:cockroachdb/cockroach#127814",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ae/ae03b0a32bdb51cddd4f3c4069a565a89122634f0b7d4a2527ef8fba5ab7c90b.json b/.cache/impact/classifications/ae/ae03b0a32bdb51cddd4f3c4069a565a89122634f0b7d4a2527ef8fba5ab7c90b.json
new file mode 100644
index 0000000..6ae5853
--- /dev/null
+++ b/.cache/impact/classifications/ae/ae03b0a32bdb51cddd4f3c4069a565a89122634f0b7d4a2527ef8fba5ab7c90b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:crate/crate#19669|sha256:d73fd7979770a47b01a655e1dec92a3ea245decd9812985aca5723a3f141a612|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d73fd7979770a47b01a655e1dec92a3ea245decd9812985aca5723a3f141a612",
+    "source_id": "github:crate/crate#19669",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ae/ae832e4dd0fb83346f546c3e3be4b019a8d8683ba78c77d8df612ebf2c030cf8.json b/.cache/impact/classifications/ae/ae832e4dd0fb83346f546c3e3be4b019a8d8683ba78c77d8df612ebf2c030cf8.json
new file mode 100644
index 0000000..5872192
--- /dev/null
+++ b/.cache/impact/classifications/ae/ae832e4dd0fb83346f546c3e3be4b019a8d8683ba78c77d8df612ebf2c030cf8.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#172379|sha256:7556aa3776193af4ba33b12bfcfb2ca813e18c63476e7eb1ee940a98ce595eda|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7556aa3776193af4ba33b12bfcfb2ca813e18c63476e7eb1ee940a98ce595eda",
+    "source_id": "github:cockroachdb/cockroach#172379",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/af/afd4f2c9a74a0f55ea5146511651cebccf92877e5e9d53952d18265e80bfec02.json b/.cache/impact/classifications/af/afd4f2c9a74a0f55ea5146511651cebccf92877e5e9d53952d18265e80bfec02.json
new file mode 100644
index 0000000..372de51
--- /dev/null
+++ b/.cache/impact/classifications/af/afd4f2c9a74a0f55ea5146511651cebccf92877e5e9d53952d18265e80bfec02.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:openlink/virtuoso-opensource#1425|sha256:f8b813d668800b10a4b53880a45ef6a0fcb355a5d6caa4e4dfa5545ff046535b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "TLP (tautology-based partitioning) rewritten query"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f8b813d668800b10a4b53880a45ef6a0fcb355a5d6caa4e4dfa5545ff046535b",
+    "source_id": "github:openlink/virtuoso-opensource#1425",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b0/b04d5d59ae6ddcb63615a042de1e4add44b342faf4e52fd4e722c4cd142df8e7.json b/.cache/impact/classifications/b0/b04d5d59ae6ddcb63615a042de1e4add44b342faf4e52fd4e722c4cd142df8e7.json
new file mode 100644
index 0000000..bfe230b
--- /dev/null
+++ b/.cache/impact/classifications/b0/b04d5d59ae6ddcb63615a042de1e4add44b342faf4e52fd4e722c4cd142df8e7.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:risingwavelabs/risingwave#3364|sha256:9ac2b828efc529c0c1771e2ff22008cd992575ff77ad63216748aafec35c9042|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9ac2b828efc529c0c1771e2ff22008cd992575ff77ad63216748aafec35c9042",
+    "source_id": "github:risingwavelabs/risingwave#3364",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b0/b0610a21773608a068a66a596bda96bf8ef5652372a97871dbcec7e1338b600c.json b/.cache/impact/classifications/b0/b0610a21773608a068a66a596bda96bf8ef5652372a97871dbcec7e1338b600c.json
new file mode 100644
index 0000000..bb7773d
--- /dev/null
+++ b/.cache/impact/classifications/b0/b0610a21773608a068a66a596bda96bf8ef5652372a97871dbcec7e1338b600c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#96149|sha256:3420beef7c499c8122727357bdd66763731b248c701412d3b380d0a17c9fc30a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:3420beef7c499c8122727357bdd66763731b248c701412d3b380d0a17c9fc30a",
+    "source_id": "github:cockroachdb/cockroach#96149",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b2/b2e28297138158d035825af04b28bad99352053a321f7690d361948426866854.json b/.cache/impact/classifications/b2/b2e28297138158d035825af04b28bad99352053a321f7690d361948426866854.json
new file mode 100644
index 0000000..69d89ff
--- /dev/null
+++ b/.cache/impact/classifications/b2/b2e28297138158d035825af04b28bad99352053a321f7690d361948426866854.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#138012|sha256:f5cd57a5f5c4da2d5f1e6ff4e7b1f6dd3cc4fa7f18418a09f52242d6a8fedf4b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f5cd57a5f5c4da2d5f1e6ff4e7b1f6dd3cc4fa7f18418a09f52242d6a8fedf4b",
+    "source_id": "github:cockroachdb/cockroach#138012",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b5/b5b7e3ede036f51f10a1ef00145a97c2bde73046d214035f29a3726d3e36b803.json b/.cache/impact/classifications/b5/b5b7e3ede036f51f10a1ef00145a97c2bde73046d214035f29a3726d3e36b803.json
new file mode 100644
index 0000000..a48bc3f
--- /dev/null
+++ b/.cache/impact/classifications/b5/b5b7e3ede036f51f10a1ef00145a97c2bde73046d214035f29a3726d3e36b803.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#69754|sha256:51fd1b3616b5664175ba352b4a74377c3f9ba7f897bf84eeb11a812671f4fd0d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:51fd1b3616b5664175ba352b4a74377c3f9ba7f897bf84eeb11a812671f4fd0d",
+    "source_id": "github:cockroachdb/cockroach#69754",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b6/b645377f2d9ca00415336c401b89c53f119b8a99e1d6e53a817abd053a8e7e71.json b/.cache/impact/classifications/b6/b645377f2d9ca00415336c401b89c53f119b8a99e1d6e53a817abd053a8e7e71.json
new file mode 100644
index 0000000..3d87afb
--- /dev/null
+++ b/.cache/impact/classifications/b6/b645377f2d9ca00415336c401b89c53f119b8a99e1d6e53a817abd053a8e7e71.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#135252|sha256:d59b5805e4bc9ab60f4a30a2adfe8269142dddee924e1de2e0dca71b64d416cc|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d59b5805e4bc9ab60f4a30a2adfe8269142dddee924e1de2e0dca71b64d416cc",
+    "source_id": "github:elastic/elasticsearch#135252",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b6/b672875a8140d583a6991ecbcb24675384458cd387913cd97e101615834dc097.json b/.cache/impact/classifications/b6/b672875a8140d583a6991ecbcb24675384458cd387913cd97e101615834dc097.json
new file mode 100644
index 0000000..f7a5704
--- /dev/null
+++ b/.cache/impact/classifications/b6/b672875a8140d583a6991ecbcb24675384458cd387913cd97e101615834dc097.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#168835|sha256:7e4d21fa49da8d0cc0b936422afbcde2354dbf0b7c02f43533e15f57ade86003|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7e4d21fa49da8d0cc0b936422afbcde2354dbf0b7c02f43533e15f57ade86003",
+    "source_id": "github:cockroachdb/cockroach#168835",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b8/b83ed71d58777c5bca1b11066eef37a97b1550327cedfb2caa86815d54383ca9.json b/.cache/impact/classifications/b8/b83ed71d58777c5bca1b11066eef37a97b1550327cedfb2caa86815d54383ca9.json
new file mode 100644
index 0000000..73ecc5a
--- /dev/null
+++ b/.cache/impact/classifications/b8/b83ed71d58777c5bca1b11066eef37a97b1550327cedfb2caa86815d54383ca9.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#144832|sha256:f510f8571a6a76c5a399bea654d28a4005f4ab544356a67530a8284901b7e672|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f510f8571a6a76c5a399bea654d28a4005f4ab544356a67530a8284901b7e672",
+    "source_id": "github:cockroachdb/cockroach#144832",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b8/b89ee278e21530bb77a0a7fbbc8d52462295333353ca3f266f7b01a8cce325a7.json b/.cache/impact/classifications/b8/b89ee278e21530bb77a0a7fbbc8d52462295333353ca3f266f7b01a8cce325a7.json
new file mode 100644
index 0000000..075ba46
--- /dev/null
+++ b/.cache/impact/classifications/b8/b89ee278e21530bb77a0a7fbbc8d52462295333353ca3f266f7b01a8cce325a7.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#171549|sha256:d73607ddc96cb5b7cf631bcd37728d0d5d563f88ae4de7b25f68993ddcd856c2|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d73607ddc96cb5b7cf631bcd37728d0d5d563f88ae4de7b25f68993ddcd856c2",
+    "source_id": "github:cockroachdb/cockroach#171549",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b8/b8ac1af954ede3f7bd2fcb4546e4584990c2429ce754b76041a6ad26e167d5b3.json b/.cache/impact/classifications/b8/b8ac1af954ede3f7bd2fcb4546e4584990c2429ce754b76041a6ad26e167d5b3.json
new file mode 100644
index 0000000..09cbfd2
--- /dev/null
+++ b/.cache/impact/classifications/b8/b8ac1af954ede3f7bd2fcb4546e4584990c2429ce754b76041a6ad26e167d5b3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#29968|sha256:131413aec4383663a9c7f907543bf66e00dd67049579ff788ce3aa2b7ccf2acd|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:131413aec4383663a9c7f907543bf66e00dd67049579ff788ce3aa2b7ccf2acd",
+    "source_id": "github:cockroachdb/cockroach#29968",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b9/b9d8fc07eb2bc70541bc6df953dc66575ee6bfaf323400e7917204d1a9657751.json b/.cache/impact/classifications/b9/b9d8fc07eb2bc70541bc6df953dc66575ee6bfaf323400e7917204d1a9657751.json
new file mode 100644
index 0000000..0276a6d
--- /dev/null
+++ b/.cache/impact/classifications/b9/b9d8fc07eb2bc70541bc6df953dc66575ee6bfaf323400e7917204d1a9657751.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:duckdb/duckdb#498|sha256:0228a32507065c9918596529e0b653fa99d0752715c84206fc7cff000589cd59|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0228a32507065c9918596529e0b653fa99d0752715c84206fc7cff000589cd59",
+    "source_id": "github:duckdb/duckdb#498",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/b9/b9f4a838de674b2759e19fbe3e0b10cf3b87322ad41d85a4d65561a6d184b18e.json b/.cache/impact/classifications/b9/b9f4a838de674b2759e19fbe3e0b10cf3b87322ad41d85a4d65561a6d184b18e.json
new file mode 100644
index 0000000..e8ffab5
--- /dev/null
+++ b/.cache/impact/classifications/b9/b9f4a838de674b2759e19fbe3e0b10cf3b87322ad41d85a4d65561a6d184b18e.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:apache/datafusion#12114|sha256:b3402022b8fbf95573b8252e0ec14ecd40163528be6ccf2ff5cf8154c92a89e3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b3402022b8fbf95573b8252e0ec14ecd40163528be6ccf2ff5cf8154c92a89e3",
+    "source_id": "github:apache/datafusion#12114",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ba/ba67374c42b448284b2bb7dc0c7bd52e17c0fdcf4082d5bbe6461ac052c673cb.json b/.cache/impact/classifications/ba/ba67374c42b448284b2bb7dc0c7bd52e17c0fdcf4082d5bbe6461ac052c673cb.json
new file mode 100644
index 0000000..ca74320
--- /dev/null
+++ b/.cache/impact/classifications/ba/ba67374c42b448284b2bb7dc0c7bd52e17c0fdcf4082d5bbe6461ac052c673cb.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:ClickHouse/ClickHouse#105355|sha256:af67dff7d004359a932f04b81c64f1df2bea803a5e9d9aefcd048eed6521c9a9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "Surfaced by sqlancer fuzzing (TLP/PQS oracles) and reduced manually."
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:af67dff7d004359a932f04b81c64f1df2bea803a5e9d9aefcd048eed6521c9a9",
+    "source_id": "github:ClickHouse/ClickHouse#105355",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bb/bb000621030bcc35f7781f2eb1fb788404159aaf8a11adacee8629f75a37fcc5.json b/.cache/impact/classifications/bb/bb000621030bcc35f7781f2eb1fb788404159aaf8a11adacee8629f75a37fcc5.json
new file mode 100644
index 0000000..bd2b4ec
--- /dev/null
+++ b/.cache/impact/classifications/bb/bb000621030bcc35f7781f2eb1fb788404159aaf8a11adacee8629f75a37fcc5.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:pingcap/tidb#66272|sha256:9bb5bd41974e9354a874bce0d35275ce4ee1de1a44b7d8378d5b8a6ab6d25216|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "Oracle: PQS"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "pqs"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9bb5bd41974e9354a874bce0d35275ce4ee1de1a44b7d8378d5b8a6ab6d25216",
+    "source_id": "github:pingcap/tidb#66272",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bb/bb3c3af68b783e63129ab8b0fb54c38703bb30d20843bf5da8c52593d505a1fa.json b/.cache/impact/classifications/bb/bb3c3af68b783e63129ab8b0fb54c38703bb30d20843bf5da8c52593d505a1fa.json
new file mode 100644
index 0000000..834b740
--- /dev/null
+++ b/.cache/impact/classifications/bb/bb3c3af68b783e63129ab8b0fb54c38703bb30d20843bf5da8c52593d505a1fa.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tarantool/tarantool#4826|sha256:4501d6e4a3cee89da3a5d13c1d5647363657b727f0817515ee5b426d57376e07|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4501d6e4a3cee89da3a5d13c1d5647363657b727f0817515ee5b426d57376e07",
+    "source_id": "github:tarantool/tarantool#4826",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bb/bb617fc7a5dc2dd2652e37ead0f86247cd866a3edb9e4d2b9ec2e60f82bfa6a0.json b/.cache/impact/classifications/bb/bb617fc7a5dc2dd2652e37ead0f86247cd866a3edb9e4d2b9ec2e60f82bfa6a0.json
new file mode 100644
index 0000000..3fcb9a2
--- /dev/null
+++ b/.cache/impact/classifications/bb/bb617fc7a5dc2dd2652e37ead0f86247cd866a3edb9e4d2b9ec2e60f82bfa6a0.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8236|sha256:7f0a048f88064890122103d8745a2667d23c8a83a8531e909265a92e324e6cb3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7f0a048f88064890122103d8745a2667d23c8a83a8531e909265a92e324e6cb3",
+    "source_id": "github:cockroachdb/cockroach#8236",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bb/bbfb68810178e27f1c9ca0f6271d050c205a8281946311797e3453ccad96d3d0.json b/.cache/impact/classifications/bb/bbfb68810178e27f1c9ca0f6271d050c205a8281946311797e3453ccad96d3d0.json
new file mode 100644
index 0000000..75b44d3
--- /dev/null
+++ b/.cache/impact/classifications/bb/bbfb68810178e27f1c9ca0f6271d050c205a8281946311797e3453ccad96d3d0.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:ClickHouse/ClickHouse#106560|sha256:6a14025509cb4371aba58728de90089ca197517e199a46830131343da0246972|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "TLP partitioning over a mixed"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6a14025509cb4371aba58728de90089ca197517e199a46830131343da0246972",
+    "source_id": "github:ClickHouse/ClickHouse#106560",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bc/bc02d3c99dc7962635bdebcb6f54bff0eb95a167f6e71e9dee6846887f585630.json b/.cache/impact/classifications/bc/bc02d3c99dc7962635bdebcb6f54bff0eb95a167f6e71e9dee6846887f585630.json
new file mode 100644
index 0000000..ef3b78e
--- /dev/null
+++ b/.cache/impact/classifications/bc/bc02d3c99dc7962635bdebcb6f54bff0eb95a167f6e71e9dee6846887f585630.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#35311|sha256:ca13f36c9575c8386c6e4b9fbde3a9455c7a54da2222030ffc36ef937678cb5b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ca13f36c9575c8386c6e4b9fbde3a9455c7a54da2222030ffc36ef937678cb5b",
+    "source_id": "github:elastic/elasticsearch#35311",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bc/bc088255649370b98df47258da6d64ebbc641a179d20174d2a1fe985962497eb.json b/.cache/impact/classifications/bc/bc088255649370b98df47258da6d64ebbc641a179d20174d2a1fe985962497eb.json
new file mode 100644
index 0000000..43ff651
--- /dev/null
+++ b/.cache/impact/classifications/bc/bc088255649370b98df47258da6d64ebbc641a179d20174d2a1fe985962497eb.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#88110|sha256:8d2afcbae219857ff24b306911e9de484664f977e6faa135fc4a6fcaf269fda5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8d2afcbae219857ff24b306911e9de484664f977e6faa135fc4a6fcaf269fda5",
+    "source_id": "github:cockroachdb/cockroach#88110",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bd/bda0a53a913725806e7d985108a5ec1394a7d28581426c47319fd6229fbccff6.json b/.cache/impact/classifications/bd/bda0a53a913725806e7d985108a5ec1394a7d28581426c47319fd6229fbccff6.json
new file mode 100644
index 0000000..f0187b7
--- /dev/null
+++ b/.cache/impact/classifications/bd/bda0a53a913725806e7d985108a5ec1394a7d28581426c47319fd6229fbccff6.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#77279|sha256:c590517920e633a415b1e61ae39f92f342b0a74ed87cd9adec998fe21a6df66a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c590517920e633a415b1e61ae39f92f342b0a74ed87cd9adec998fe21a6df66a",
+    "source_id": "github:cockroachdb/cockroach#77279",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bd/bdaa11108172265ef5adc5437548d19eb5c240f63e70edec38a2914ff65fe877.json b/.cache/impact/classifications/bd/bdaa11108172265ef5adc5437548d19eb5c240f63e70edec38a2914ff65fe877.json
new file mode 100644
index 0000000..61bd5e2
--- /dev/null
+++ b/.cache/impact/classifications/bd/bdaa11108172265ef5adc5437548d19eb5c240f63e70edec38a2914ff65fe877.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#36812|sha256:32861b333553d684072b2d2f15273ff1d4ba09179a32eee92b0550694a60e8ec|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:32861b333553d684072b2d2f15273ff1d4ba09179a32eee92b0550694a60e8ec",
+    "source_id": "github:elastic/elasticsearch#36812",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/be/be130f009ed7bfcca301b063e97c36a988a4a793c742dc76f07d2f2b33ccc319.json b/.cache/impact/classifications/be/be130f009ed7bfcca301b063e97c36a988a4a793c742dc76f07d2f2b33ccc319.json
new file mode 100644
index 0000000..aefa0c0
--- /dev/null
+++ b/.cache/impact/classifications/be/be130f009ed7bfcca301b063e97c36a988a4a793c742dc76f07d2f2b33ccc319.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#12196|sha256:9f20cf04c6a11321ff7552932333d1167790ca70526fe597c66a3b31b14ba691|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9f20cf04c6a11321ff7552932333d1167790ca70526fe597c66a3b31b14ba691",
+    "source_id": "github:pingcap/tidb#12196",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bf/bf0029fedbee143422dd50c9a8e4fe8d8279d65f99ed2cbbdefa1186927adf65.json b/.cache/impact/classifications/bf/bf0029fedbee143422dd50c9a8e4fe8d8279d65f99ed2cbbdefa1186927adf65.json
new file mode 100644
index 0000000..7811aa7
--- /dev/null
+++ b/.cache/impact/classifications/bf/bf0029fedbee143422dd50c9a8e4fe8d8279d65f99ed2cbbdefa1186927adf65.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#3550|sha256:c819b6f18c051ca641ac5aba3c76a9f0b360a6f167c4adeec3b4510a145d127e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c819b6f18c051ca641ac5aba3c76a9f0b360a6f167c4adeec3b4510a145d127e",
+    "source_id": "github:cockroachdb/cockroach#3550",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bf/bfb510deda29952011218e4291d91ee10374ca37b1ebd1a82936d5261f558b10.json b/.cache/impact/classifications/bf/bfb510deda29952011218e4291d91ee10374ca37b1ebd1a82936d5261f558b10.json
new file mode 100644
index 0000000..1832cc6
--- /dev/null
+++ b/.cache/impact/classifications/bf/bfb510deda29952011218e4291d91ee10374ca37b1ebd1a82936d5261f558b10.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#80374|sha256:b2022ec4a7a0f287e4d045587bb3e0f09484ad6504d8c79bc6c02fd9d39ddf46|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b2022ec4a7a0f287e4d045587bb3e0f09484ad6504d8c79bc6c02fd9d39ddf46",
+    "source_id": "github:cockroachdb/cockroach#80374",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/bf/bfb8049a4c90182e8eb77f23e7241e48fd5dac356321481a85cc2c33177d2d28.json b/.cache/impact/classifications/bf/bfb8049a4c90182e8eb77f23e7241e48fd5dac356321481a85cc2c33177d2d28.json
new file mode 100644
index 0000000..a87cd9a
--- /dev/null
+++ b/.cache/impact/classifications/bf/bfb8049a4c90182e8eb77f23e7241e48fd5dac356321481a85cc2c33177d2d28.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:questdb/questdb#595|sha256:77ebc08c1edca953be0f660303cd6116725b5efa469eb282b6d55c762158369d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:77ebc08c1edca953be0f660303cd6116725b5efa469eb282b6d55c762158369d",
+    "source_id": "github:questdb/questdb#595",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c0/c03a3f1bff883fd7e2685bfec74f5f2863af23690e49aa0efb6eeef391f1f047.json b/.cache/impact/classifications/c0/c03a3f1bff883fd7e2685bfec74f5f2863af23690e49aa0efb6eeef391f1f047.json
new file mode 100644
index 0000000..52cd05c
--- /dev/null
+++ b/.cache/impact/classifications/c0/c03a3f1bff883fd7e2685bfec74f5f2863af23690e49aa0efb6eeef391f1f047.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#53950|sha256:25c9c639c9ea7ad1b448f14edac79b803183018a6d95dd2d2c12a031f0098adc|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:25c9c639c9ea7ad1b448f14edac79b803183018a6d95dd2d2c12a031f0098adc",
+    "source_id": "github:cockroachdb/cockroach#53950",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c0/c0883bd5aab77f7389e96a9b353aab19142af3ffc94ad555f0ee6dab20f9215b.json b/.cache/impact/classifications/c0/c0883bd5aab77f7389e96a9b353aab19142af3ffc94ad555f0ee6dab20f9215b.json
new file mode 100644
index 0000000..101ddde
--- /dev/null
+++ b/.cache/impact/classifications/c0/c0883bd5aab77f7389e96a9b353aab19142af3ffc94ad555f0ee6dab20f9215b.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#79947|sha256:60f7219dfec8788f0187107266dafb86063786bf39a14b21bc59419a6b57336b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:60f7219dfec8788f0187107266dafb86063786bf39a14b21bc59419a6b57336b",
+    "source_id": "github:cockroachdb/cockroach#79947",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c1/c108b4cc78a8a42c1fd038dc1e279cfb23563f3e34750e00afb4ab25fd1f5f23.json b/.cache/impact/classifications/c1/c108b4cc78a8a42c1fd038dc1e279cfb23563f3e34750e00afb4ab25fd1f5f23.json
new file mode 100644
index 0000000..44c2f30
--- /dev/null
+++ b/.cache/impact/classifications/c1/c108b4cc78a8a42c1fd038dc1e279cfb23563f3e34750e00afb4ab25fd1f5f23.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4102|sha256:8eb749fc46319b518098bfb79f0b6b083ed291e049a1396d9d867b27bd8b2efb|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8eb749fc46319b518098bfb79f0b6b083ed291e049a1396d9d867b27bd8b2efb",
+    "source_id": "github:sparq-org/sparq#4102",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c1/c173ad3d39ee4c87fd437bcb0cb9b5982c1e6d86a00241d1ddd31712e998ab66.json b/.cache/impact/classifications/c1/c173ad3d39ee4c87fd437bcb0cb9b5982c1e6d86a00241d1ddd31712e998ab66.json
new file mode 100644
index 0000000..af6d44c
--- /dev/null
+++ b/.cache/impact/classifications/c1/c173ad3d39ee4c87fd437bcb0cb9b5982c1e6d86a00241d1ddd31712e998ab66.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8872|sha256:8a80dc011e69310a2024eab50ac28c08e3ca58d7936d82646a76ce1cacb2d891|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8a80dc011e69310a2024eab50ac28c08e3ca58d7936d82646a76ce1cacb2d891",
+    "source_id": "github:cockroachdb/cockroach#8872",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c3/c33948e31ab7ce4429e6f133a0c806f471a0ab46dc5246cf88d128d489f3a485.json b/.cache/impact/classifications/c3/c33948e31ab7ce4429e6f133a0c806f471a0ab46dc5246cf88d128d489f3a485.json
new file mode 100644
index 0000000..975d61f
--- /dev/null
+++ b/.cache/impact/classifications/c3/c33948e31ab7ce4429e6f133a0c806f471a0ab46dc5246cf88d128d489f3a485.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#83746|sha256:d2318bf900fcd8cbad4044712f641b8cceea945da666ab260f6219865c9018c6|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:d2318bf900fcd8cbad4044712f641b8cceea945da666ab260f6219865c9018c6",
+    "source_id": "github:cockroachdb/cockroach#83746",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c3/c3403d87237f0c7a23f0793304e663610f8cebafdd537d2135a1648c4e4eb69c.json b/.cache/impact/classifications/c3/c3403d87237f0c7a23f0793304e663610f8cebafdd537d2135a1648c4e4eb69c.json
new file mode 100644
index 0000000..39bcc1a
--- /dev/null
+++ b/.cache/impact/classifications/c3/c3403d87237f0c7a23f0793304e663610f8cebafdd537d2135a1648c4e4eb69c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4791|sha256:83325b246cf7c19c93a034918a55f50552a5a13d2b8831d6724179b8b3d74b1d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:83325b246cf7c19c93a034918a55f50552a5a13d2b8831d6724179b8b3d74b1d",
+    "source_id": "github:sparq-org/sparq#4791",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c3/c37adadab0f8a5d3a6709e332399ed95ad06d9ce3b84b0c0ac9ecb71176b48cd.json b/.cache/impact/classifications/c3/c37adadab0f8a5d3a6709e332399ed95ad06d9ce3b84b0c0ac9ecb71176b48cd.json
new file mode 100644
index 0000000..b34d9d9
--- /dev/null
+++ b/.cache/impact/classifications/c3/c37adadab0f8a5d3a6709e332399ed95ad06d9ce3b84b0c0ac9ecb71176b48cd.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#3311|sha256:40fd995a48fae15793d9a7560041fd18f420903227031eff307911ec7706c1b9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:40fd995a48fae15793d9a7560041fd18f420903227031eff307911ec7706c1b9",
+    "source_id": "github:sparq-org/sparq#3311",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c5/c588f4bc991d287a0c47ff2690e5de90c16942b95e6d5f0122f01a4c9ed38cc6.json b/.cache/impact/classifications/c5/c588f4bc991d287a0c47ff2690e5de90c16942b95e6d5f0122f01a4c9ed38cc6.json
new file mode 100644
index 0000000..4e0ee18
--- /dev/null
+++ b/.cache/impact/classifications/c5/c588f4bc991d287a0c47ff2690e5de90c16942b95e6d5f0122f01a4c9ed38cc6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#8434|sha256:c2a4c851f71847a594804eb5c6c8efa85841818f6440ac62004a2a4bebfb8e01|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c2a4c851f71847a594804eb5c6c8efa85841818f6440ac62004a2a4bebfb8e01",
+    "source_id": "github:yugabyte/yugabyte-db#8434",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c6/c6559584db371a0540f93a35eb4a0c8508c59640fb4cd5d5075541f9c74dad23.json b/.cache/impact/classifications/c6/c6559584db371a0540f93a35eb4a0c8508c59640fb4cd5d5075541f9c74dad23.json
new file mode 100644
index 0000000..4710bfe
--- /dev/null
+++ b/.cache/impact/classifications/c6/c6559584db371a0540f93a35eb4a0c8508c59640fb4cd5d5075541f9c74dad23.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:opensearch-project/sql#3266|sha256:edf70ec806ada6640063ffbe1f465272533603b420622f871191bbe052a2213e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:edf70ec806ada6640063ffbe1f465272533603b420622f871191bbe052a2213e",
+    "source_id": "github:opensearch-project/sql#3266",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c7/c7a1a9db2e171934f5ffafe80196233eb9fab69d7d7444bcc38193acd373522f.json b/.cache/impact/classifications/c7/c7a1a9db2e171934f5ffafe80196233eb9fab69d7d7444bcc38193acd373522f.json
new file mode 100644
index 0000000..fd72e6c
--- /dev/null
+++ b/.cache/impact/classifications/c7/c7a1a9db2e171934f5ffafe80196233eb9fab69d7d7444bcc38193acd373522f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#9513|sha256:7c6117503b4262249dbb6fd9b134856966de79755c880f43c68b4dc192fcf205|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7c6117503b4262249dbb6fd9b134856966de79755c880f43c68b4dc192fcf205",
+    "source_id": "github:cockroachdb/cockroach#9513",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c7/c7b7db7a5d5109f4b9c5ba5561640743083a487b559d832bc11a6cdbb84cef91.json b/.cache/impact/classifications/c7/c7b7db7a5d5109f4b9c5ba5561640743083a487b559d832bc11a6cdbb84cef91.json
new file mode 100644
index 0000000..eda4fab
--- /dev/null
+++ b/.cache/impact/classifications/c7/c7b7db7a5d5109f4b9c5ba5561640743083a487b559d832bc11a6cdbb84cef91.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#88869|sha256:0a9cc301df5a309c48e7bf4eb51a998f4194cf086574bacc17e5737dba2ac122|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0a9cc301df5a309c48e7bf4eb51a998f4194cf086574bacc17e5737dba2ac122",
+    "source_id": "github:cockroachdb/cockroach#88869",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c7/c7c43bf0ce69038445569d0d97a8ee565932e41bba450728de8d5e1c569a6dff.json b/.cache/impact/classifications/c7/c7c43bf0ce69038445569d0d97a8ee565932e41bba450728de8d5e1c569a6dff.json
new file mode 100644
index 0000000..4400fc1
--- /dev/null
+++ b/.cache/impact/classifications/c7/c7c43bf0ce69038445569d0d97a8ee565932e41bba450728de8d5e1c569a6dff.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#144784|sha256:eddcf9d2e1fa2f51d111f61a61ee2823c8e1e6ee7eb6d5e9cca6d3153421fe2d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:eddcf9d2e1fa2f51d111f61a61ee2823c8e1e6ee7eb6d5e9cca6d3153421fe2d",
+    "source_id": "github:cockroachdb/cockroach#144784",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/c9/c9130f3161bfeaa25e9a2a7373c7ee67bc40ac5cade27833e506996274dc31d4.json b/.cache/impact/classifications/c9/c9130f3161bfeaa25e9a2a7373c7ee67bc40ac5cade27833e506996274dc31d4.json
new file mode 100644
index 0000000..a48a03f
--- /dev/null
+++ b/.cache/impact/classifications/c9/c9130f3161bfeaa25e9a2a7373c7ee67bc40ac5cade27833e506996274dc31d4.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#38332|sha256:1dfaaa6447c8952e7713bee1c83cc58bd1de271216a79ffc761db3f3058a1fba|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1dfaaa6447c8952e7713bee1c83cc58bd1de271216a79ffc761db3f3058a1fba",
+    "source_id": "github:pingcap/tidb#38332",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ca/ca697598155693e067dcfe46b08f583f9001b5eea5c7cc359697dedd35517d8a.json b/.cache/impact/classifications/ca/ca697598155693e067dcfe46b08f583f9001b5eea5c7cc359697dedd35517d8a.json
new file mode 100644
index 0000000..edd74e9
--- /dev/null
+++ b/.cache/impact/classifications/ca/ca697598155693e067dcfe46b08f583f9001b5eea5c7cc359697dedd35517d8a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:ydb-platform/ydb#15971|sha256:75b2d765d9ac5b53c297b12558b3e044f561d41c6dfaf3f0feeeb59d3c50e9e5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:75b2d765d9ac5b53c297b12558b3e044f561d41c6dfaf3f0feeeb59d3c50e9e5",
+    "source_id": "github:ydb-platform/ydb#15971",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ca/cae41e660c13a64c3806b356a3aa88b34bd2bd9138b5b1ca37a561ec655f4d5f.json b/.cache/impact/classifications/ca/cae41e660c13a64c3806b356a3aa88b34bd2bd9138b5b1ca37a561ec655f4d5f.json
new file mode 100644
index 0000000..d1bf14a
--- /dev/null
+++ b/.cache/impact/classifications/ca/cae41e660c13a64c3806b356a3aa88b34bd2bd9138b5b1ca37a561ec655f4d5f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#171389|sha256:f067ad6aefe90231e7bb68564d29cd661facc6c74dafb389c1f495f605a25ef5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f067ad6aefe90231e7bb68564d29cd661facc6c74dafb389c1f495f605a25ef5",
+    "source_id": "github:cockroachdb/cockroach#171389",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/cc/cc6426dd2a34ae4bdd4916921864af3af727a0419a09c2ad4bbd347927e4bb20.json b/.cache/impact/classifications/cc/cc6426dd2a34ae4bdd4916921864af3af727a0419a09c2ad4bbd347927e4bb20.json
new file mode 100644
index 0000000..75d414f
--- /dev/null
+++ b/.cache/impact/classifications/cc/cc6426dd2a34ae4bdd4916921864af3af727a0419a09c2ad4bbd347927e4bb20.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#171383|sha256:90edb92b3fe9abef619ff8087d52de27479d0da8a8d8d01cf20e0ce75d6791c5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:90edb92b3fe9abef619ff8087d52de27479d0da8a8d8d01cf20e0ce75d6791c5",
+    "source_id": "github:cockroachdb/cockroach#171383",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/cc/cc9934db1d426ed7461fed921f36e8872b109003f798a5d163418b5a6e41da9e.json b/.cache/impact/classifications/cc/cc9934db1d426ed7461fed921f36e8872b109003f798a5d163418b5a6e41da9e.json
new file mode 100644
index 0000000..50aadd6
--- /dev/null
+++ b/.cache/impact/classifications/cc/cc9934db1d426ed7461fed921f36e8872b109003f798a5d163418b5a6e41da9e.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#74675|sha256:80e1ffc6f0394e97c7eaea6f473148d6a6547cabd4d823f199bc04886fbaf244|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:80e1ffc6f0394e97c7eaea6f473148d6a6547cabd4d823f199bc04886fbaf244",
+    "source_id": "github:cockroachdb/cockroach#74675",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/cc/ccccd056aec9246c9e0202ae837b8144a122607f408fb3c9f1baa3f19735f1c3.json b/.cache/impact/classifications/cc/ccccd056aec9246c9e0202ae837b8144a122607f408fb3c9f1baa3f19735f1c3.json
new file mode 100644
index 0000000..58ba177
--- /dev/null
+++ b/.cache/impact/classifications/cc/ccccd056aec9246c9e0202ae837b8144a122607f408fb3c9f1baa3f19735f1c3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#25177|sha256:8fc170cd4bedf4dbfd5c92bb034c369f200f43c2490bb9f24c7679570fb5e66f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8fc170cd4bedf4dbfd5c92bb034c369f200f43c2490bb9f24c7679570fb5e66f",
+    "source_id": "github:cockroachdb/cockroach#25177",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/cd/cd2da50a4fae7d02f20689f37fd4cfdb28929a236c54ff451c8178e88bfa0676.json b/.cache/impact/classifications/cd/cd2da50a4fae7d02f20689f37fd4cfdb28929a236c54ff451c8178e88bfa0676.json
new file mode 100644
index 0000000..552e789
--- /dev/null
+++ b/.cache/impact/classifications/cd/cd2da50a4fae7d02f20689f37fd4cfdb28929a236c54ff451c8178e88bfa0676.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#26180|sha256:dda2ab149b4194b128824eff6d4c29e3d70fa2105af27418a80d3a11ff434823|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:dda2ab149b4194b128824eff6d4c29e3d70fa2105af27418a80d3a11ff434823",
+    "source_id": "github:cockroachdb/cockroach#26180",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/cd/cdef68024f3eb8bbf93d520e5b9d74429f0f3fda89cb43274531edaeedb3bc7f.json b/.cache/impact/classifications/cd/cdef68024f3eb8bbf93d520e5b9d74429f0f3fda89cb43274531edaeedb3bc7f.json
new file mode 100644
index 0000000..08a1efa
--- /dev/null
+++ b/.cache/impact/classifications/cd/cdef68024f3eb8bbf93d520e5b9d74429f0f3fda89cb43274531edaeedb3bc7f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8215|sha256:8ceda49f0b10cae9fd3035cce91805b30010bda4219be7e7a626b80508d72151|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8ceda49f0b10cae9fd3035cce91805b30010bda4219be7e7a626b80508d72151",
+    "source_id": "github:cockroachdb/cockroach#8215",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ce/ce6bd7f1beecb5e25643e2c3ccb5361dda970801bcc8e0fae74b5cfa7902b850.json b/.cache/impact/classifications/ce/ce6bd7f1beecb5e25643e2c3ccb5361dda970801bcc8e0fae74b5cfa7902b850.json
new file mode 100644
index 0000000..e75a92b
--- /dev/null
+++ b/.cache/impact/classifications/ce/ce6bd7f1beecb5e25643e2c3ccb5361dda970801bcc8e0fae74b5cfa7902b850.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#28276|sha256:91559c4172fa796cfcdf27bf424286da8169853299a19693cec0c88aafc5e28f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:91559c4172fa796cfcdf27bf424286da8169853299a19693cec0c88aafc5e28f",
+    "source_id": "github:pingcap/tidb#28276",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ce/cef34b69f413faed5b7519ce0c2ed1e255bc9ab436a3025627fd8c6ea87c1c23.json b/.cache/impact/classifications/ce/cef34b69f413faed5b7519ce0c2ed1e255bc9ab436a3025627fd8c6ea87c1c23.json
new file mode 100644
index 0000000..3cc2861
--- /dev/null
+++ b/.cache/impact/classifications/ce/cef34b69f413faed5b7519ce0c2ed1e255bc9ab436a3025627fd8c6ea87c1c23.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:hazelcast/hazelcast#22577|sha256:e6f45ea7790571d5f729a4d1a21718dba0d1209402b0962739d69dfa6e2671f9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e6f45ea7790571d5f729a4d1a21718dba0d1209402b0962739d69dfa6e2671f9",
+    "source_id": "github:hazelcast/hazelcast#22577",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/cf/cf3aa11bf98d9d82beda3a0b12e532e78b023b25f8a44e677aa32bb8b56e11c1.json b/.cache/impact/classifications/cf/cf3aa11bf98d9d82beda3a0b12e532e78b023b25f8a44e677aa32bb8b56e11c1.json
new file mode 100644
index 0000000..5196d56
--- /dev/null
+++ b/.cache/impact/classifications/cf/cf3aa11bf98d9d82beda3a0b12e532e78b023b25f8a44e677aa32bb8b56e11c1.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#6155|sha256:cc0813fd658da559cdc8d92db4cfc1b325de8912ccb53f29d0d58f46752b8685|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:cc0813fd658da559cdc8d92db4cfc1b325de8912ccb53f29d0d58f46752b8685",
+    "source_id": "github:yugabyte/yugabyte-db#6155",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/cf/cf4a905cd5ee191f08dbb3dbee3973324f1059d8062ba2ca3a59d6d352f2be91.json b/.cache/impact/classifications/cf/cf4a905cd5ee191f08dbb3dbee3973324f1059d8062ba2ca3a59d6d352f2be91.json
new file mode 100644
index 0000000..b7690fd
--- /dev/null
+++ b/.cache/impact/classifications/cf/cf4a905cd5ee191f08dbb3dbee3973324f1059d8062ba2ca3a59d6d352f2be91.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#87418|sha256:690a8d727ea015a356cebb2cf6963637d23c8c007254a3ffb7a6a44db3ffb89d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:690a8d727ea015a356cebb2cf6963637d23c8c007254a3ffb7a6a44db3ffb89d",
+    "source_id": "github:cockroachdb/cockroach#87418",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d0/d06c32e4cf477cb8aef58e54ecc97c884f2c4b30b26a7b757bf1da42b7d2b699.json b/.cache/impact/classifications/d0/d06c32e4cf477cb8aef58e54ecc97c884f2c4b30b26a7b757bf1da42b7d2b699.json
new file mode 100644
index 0000000..4a2f50c
--- /dev/null
+++ b/.cache/impact/classifications/d0/d06c32e4cf477cb8aef58e54ecc97c884f2c4b30b26a7b757bf1da42b7d2b699.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#132769|sha256:861619cf5656c5c98ffaa5451a258caa1e289f6975c925af406300bf5fec7ca1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:861619cf5656c5c98ffaa5451a258caa1e289f6975c925af406300bf5fec7ca1",
+    "source_id": "github:cockroachdb/cockroach#132769",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d1/d106bb699364d21a6877a0f2510db29449e0e574eff01c898316e603775c78b2.json b/.cache/impact/classifications/d1/d106bb699364d21a6877a0f2510db29449e0e574eff01c898316e603775c78b2.json
new file mode 100644
index 0000000..7c4dfac
--- /dev/null
+++ b/.cache/impact/classifications/d1/d106bb699364d21a6877a0f2510db29449e0e574eff01c898316e603775c78b2.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#143716|sha256:83ceabaa4a029fa8d4e3f082f288d72b7f6e49819431b168fbd05c28d1cd897d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:83ceabaa4a029fa8d4e3f082f288d72b7f6e49819431b168fbd05c28d1cd897d",
+    "source_id": "github:cockroachdb/cockroach#143716",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d1/d123032068159d8880fe350e3d0ef9cc756019b67bfc99afa1c80b56c29698fd.json b/.cache/impact/classifications/d1/d123032068159d8880fe350e3d0ef9cc756019b67bfc99afa1c80b56c29698fd.json
new file mode 100644
index 0000000..91642a8
--- /dev/null
+++ b/.cache/impact/classifications/d1/d123032068159d8880fe350e3d0ef9cc756019b67bfc99afa1c80b56c29698fd.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8223|sha256:f02edcadc1291f92499f35c9ae67f5e970fe897756c4deccbff29b48772b209e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f02edcadc1291f92499f35c9ae67f5e970fe897756c4deccbff29b48772b209e",
+    "source_id": "github:cockroachdb/cockroach#8223",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d2/d20b6b77f9c724c7e73c8b0609001ed161a1e3a0c0fbb398ec3c063ee77f95b6.json b/.cache/impact/classifications/d2/d20b6b77f9c724c7e73c8b0609001ed161a1e3a0c0fbb398ec3c063ee77f95b6.json
new file mode 100644
index 0000000..bde7769
--- /dev/null
+++ b/.cache/impact/classifications/d2/d20b6b77f9c724c7e73c8b0609001ed161a1e3a0c0fbb398ec3c063ee77f95b6.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:yugabyte/yugabyte-db#11076|sha256:be5bd7f12f8898eacb0698ab1fb12b48c14a866a0c4e4b06cb958fa5f32ecae1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "[SQLancer] [YSQL] Results mismatch with postgres"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": null
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:be5bd7f12f8898eacb0698ab1fb12b48c14a866a0c4e4b06cb958fa5f32ecae1",
+    "source_id": "github:yugabyte/yugabyte-db#11076",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d2/d2a8804084aa6d2c08073f0cefcef526487cce5cb84834e80e42c3f9fa47098d.json b/.cache/impact/classifications/d2/d2a8804084aa6d2c08073f0cefcef526487cce5cb84834e80e42c3f9fa47098d.json
new file mode 100644
index 0000000..260cbbf
--- /dev/null
+++ b/.cache/impact/classifications/d2/d2a8804084aa6d2c08073f0cefcef526487cce5cb84834e80e42c3f9fa47098d.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#81574|sha256:b37bf1abe6437e5fca06246b77471858abafd5c4ae63567e8e361c5f389e91f0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b37bf1abe6437e5fca06246b77471858abafd5c4ae63567e8e361c5f389e91f0",
+    "source_id": "github:cockroachdb/cockroach#81574",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d3/d3ba866388bb236b66c919c67eba114e392bebf05058ba91f7153943d9866164.json b/.cache/impact/classifications/d3/d3ba866388bb236b66c919c67eba114e392bebf05058ba91f7153943d9866164.json
new file mode 100644
index 0000000..264849c
--- /dev/null
+++ b/.cache/impact/classifications/d3/d3ba866388bb236b66c919c67eba114e392bebf05058ba91f7153943d9866164.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tikv/tikv#3679|sha256:86668abadea116a4e60171783aa38eb258f2d236866f145fcc8e092e6c7f9f85|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:86668abadea116a4e60171783aa38eb258f2d236866f145fcc8e092e6c7f9f85",
+    "source_id": "github:tikv/tikv#3679",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d3/d3c048bead4231b1587c5d75a12778d4111bd75aa96b7dcfb2ca72adbb14c452.json b/.cache/impact/classifications/d3/d3c048bead4231b1587c5d75a12778d4111bd75aa96b7dcfb2ca72adbb14c452.json
new file mode 100644
index 0000000..07b7dce
--- /dev/null
+++ b/.cache/impact/classifications/d3/d3c048bead4231b1587c5d75a12778d4111bd75aa96b7dcfb2ca72adbb14c452.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#124629|sha256:193b8324904db3b7f92d6f483c2c66fd7bb6d5f98b4683fc2097d4810ae69303|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:193b8324904db3b7f92d6f483c2c66fd7bb6d5f98b4683fc2097d4810ae69303",
+    "source_id": "github:cockroachdb/cockroach#124629",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d3/d3fa4ab9c49204b340755bf72192770b5f71d7d79b8487ba4db11e0e01482675.json b/.cache/impact/classifications/d3/d3fa4ab9c49204b340755bf72192770b5f71d7d79b8487ba4db11e0e01482675.json
new file mode 100644
index 0000000..b61dfc9
--- /dev/null
+++ b/.cache/impact/classifications/d3/d3fa4ab9c49204b340755bf72192770b5f71d7d79b8487ba4db11e0e01482675.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#13085|sha256:5f78575b202c4c033170d33cbc2459691c47e1fcd108b08f57dfca1effb1cce1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5f78575b202c4c033170d33cbc2459691c47e1fcd108b08f57dfca1effb1cce1",
+    "source_id": "github:cockroachdb/cockroach#13085",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d4/d42dfbcc23013ad16e0c7c36c6ff61d267e424094d2d6fa73780cd6c5c45277c.json b/.cache/impact/classifications/d4/d42dfbcc23013ad16e0c7c36c6ff61d267e424094d2d6fa73780cd6c5c45277c.json
new file mode 100644
index 0000000..661df63
--- /dev/null
+++ b/.cache/impact/classifications/d4/d42dfbcc23013ad16e0c7c36c6ff61d267e424094d2d6fa73780cd6c5c45277c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#37476|sha256:83509010aef4d159e4b7bcf87e26bd107c931d3414e65835a82c8cf1cda6be13|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:83509010aef4d159e4b7bcf87e26bd107c931d3414e65835a82c8cf1cda6be13",
+    "source_id": "github:cockroachdb/cockroach#37476",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d5/d59aeac589a94618b84433a5210633702040e256ecd9cb4383d031a95ce8bc8e.json b/.cache/impact/classifications/d5/d59aeac589a94618b84433a5210633702040e256ecd9cb4383d031a95ce8bc8e.json
new file mode 100644
index 0000000..9eb24ed
--- /dev/null
+++ b/.cache/impact/classifications/d5/d59aeac589a94618b84433a5210633702040e256ecd9cb4383d031a95ce8bc8e.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#14856|sha256:eb5d916718227cc14c6202621d27a8a0feee3444a4c9d14a379a1da3cd46958f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:eb5d916718227cc14c6202621d27a8a0feee3444a4c9d14a379a1da3cd46958f",
+    "source_id": "github:yugabyte/yugabyte-db#14856",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d5/d5bba29512401e0e97d95943d9405ec9ce4d8863198cae9ffdc869fb91eb89d7.json b/.cache/impact/classifications/d5/d5bba29512401e0e97d95943d9405ec9ce4d8863198cae9ffdc869fb91eb89d7.json
new file mode 100644
index 0000000..282d0ee
--- /dev/null
+++ b/.cache/impact/classifications/d5/d5bba29512401e0e97d95943d9405ec9ce4d8863198cae9ffdc869fb91eb89d7.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:ClickHouse/ClickHouse#106956|sha256:e6a6a3261b37e670d26a3a607653de170fbda94eda7e391fcc149c3b057fc9a3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "Triggered by the oracle's own TLP"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e6a6a3261b37e670d26a3a607653de170fbda94eda7e391fcc149c3b057fc9a3",
+    "source_id": "github:ClickHouse/ClickHouse#106956",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d6/d65929eff883f2119328a6a4aeedeada8719312d00047dd1f445efd3a12499e6.json b/.cache/impact/classifications/d6/d65929eff883f2119328a6a4aeedeada8719312d00047dd1f445efd3a12499e6.json
new file mode 100644
index 0000000..d4fd02f
--- /dev/null
+++ b/.cache/impact/classifications/d6/d65929eff883f2119328a6a4aeedeada8719312d00047dd1f445efd3a12499e6.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:apache/doris#19786|sha256:3709d5989d64e87771ae7e618231d856e995369c3e959d6817983ac2b56d6ae5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:3709d5989d64e87771ae7e618231d856e995369c3e959d6817983ac2b56d6ae5",
+    "source_id": "github:apache/doris#19786",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d6/d69854f8f7db81232391c60d188f87a11c9b7fdae6e306f9cc47d05afb4cfb64.json b/.cache/impact/classifications/d6/d69854f8f7db81232391c60d188f87a11c9b7fdae6e306f9cc47d05afb4cfb64.json
new file mode 100644
index 0000000..99ac9fb
--- /dev/null
+++ b/.cache/impact/classifications/d6/d69854f8f7db81232391c60d188f87a11c9b7fdae6e306f9cc47d05afb4cfb64.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#4464|sha256:2a9e776cefacad217e21edc7b53e61589249426894c09793353847e0f85d9237|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2a9e776cefacad217e21edc7b53e61589249426894c09793353847e0f85d9237",
+    "source_id": "github:cockroachdb/cockroach#4464",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d9/d97725254e6beb2946213a0262c4d2cf7d7373ebabd55db0b517b40e136e8ea0.json b/.cache/impact/classifications/d9/d97725254e6beb2946213a0262c4d2cf7d7373ebabd55db0b517b40e136e8ea0.json
new file mode 100644
index 0000000..1c16a84
--- /dev/null
+++ b/.cache/impact/classifications/d9/d97725254e6beb2946213a0262c4d2cf7d7373ebabd55db0b517b40e136e8ea0.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#19969|sha256:c550a4a6d7ffbfe3cca7edf485e1994ac8774df1048cecdf7f7cd173d655faf2|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c550a4a6d7ffbfe3cca7edf485e1994ac8774df1048cecdf7f7cd173d655faf2",
+    "source_id": "github:cockroachdb/cockroach#19969",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/d9/d9be9dfe5c87f194fe74e154bfacf361438a6a166a9ed1d2a2f196282119ba8d.json b/.cache/impact/classifications/d9/d9be9dfe5c87f194fe74e154bfacf361438a6a166a9ed1d2a2f196282119ba8d.json
new file mode 100644
index 0000000..ad30004
--- /dev/null
+++ b/.cache/impact/classifications/d9/d9be9dfe5c87f194fe74e154bfacf361438a6a166a9ed1d2a2f196282119ba8d.json
@@ -0,0 +1,24 @@
+{
+  "key": "adoption:https://github.com/StarRocks/starrocks/blob/444cb3cf593a8406c096ca79d270908411414654/docs/en/release_notes/release-2.2.md|sha256:90a93e293c86c36d21d2ba3cb7be89bbf881798a97f8eb29ad178266f146d6dc|adoption-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T06:22:22Z",
+  "value": {
+    "classified_at": "2026-09-13T06:22:22Z",
+    "classifier_version": "adoption-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "A few bugs are detected by Synthesized Query Lancer (SQLancer)."
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "relationship": "developer_use"
+      },
+      "reason": "StarRocks' own release notes credit SQLancer with finding bugs in the release, and point at the sqlancer label on their issue tracker. Decided by reading the file in this session rather than by an API call."
+    },
+    "source_hash": "sha256:90a93e293c86c36d21d2ba3cb7be89bbf881798a97f8eb29ad178266f146d6dc",
+    "source_id": "adoption:https://github.com/StarRocks/starrocks/blob/444cb3cf593a8406c096ca79d270908411414654/docs/en/release_notes/release-2.2.md",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/da/da7798d2b188ccd04d00be45d121d01d6e3c4a364eec5311bd087127136ec108.json b/.cache/impact/classifications/da/da7798d2b188ccd04d00be45d121d01d6e3c4a364eec5311bd087127136ec108.json
new file mode 100644
index 0000000..53ac163
--- /dev/null
+++ b/.cache/impact/classifications/da/da7798d2b188ccd04d00be45d121d01d6e3c4a364eec5311bd087127136ec108.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:yugabyte/yugabyte-db#11110|sha256:14bb09108f55d834bc0d1b94a48b2dc75165e7e1dc1cf7abff3da38c66022b1d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "[SQLancer] [YSQL] Trying to drop serial PK column leads to failed inserts"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": null
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:14bb09108f55d834bc0d1b94a48b2dc75165e7e1dc1cf7abff3da38c66022b1d",
+    "source_id": "github:yugabyte/yugabyte-db#11110",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/da/da99420b6151f221f441349955eb5eecc1e7b7c091c962354cd0173efcab7d2b.json b/.cache/impact/classifications/da/da99420b6151f221f441349955eb5eecc1e7b7c091c962354cd0173efcab7d2b.json
new file mode 100644
index 0000000..e6ea0c7
--- /dev/null
+++ b/.cache/impact/classifications/da/da99420b6151f221f441349955eb5eecc1e7b7c091c962354cd0173efcab7d2b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4637|sha256:2cb2c098a19f0c2feca6b49e0f9bbc13da659b0ba1b39dc6e2c2dbba9b7bc72c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2cb2c098a19f0c2feca6b49e0f9bbc13da659b0ba1b39dc6e2c2dbba9b7bc72c",
+    "source_id": "github:sparq-org/sparq#4637",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/dd/dd5976acf3475f78ddf9f6a4debf562438caa6715e83d967646e029ff1273898.json b/.cache/impact/classifications/dd/dd5976acf3475f78ddf9f6a4debf562438caa6715e83d967646e029ff1273898.json
new file mode 100644
index 0000000..4a1bd87
--- /dev/null
+++ b/.cache/impact/classifications/dd/dd5976acf3475f78ddf9f6a4debf562438caa6715e83d967646e029ff1273898.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:spiceai/spiceai#2187|sha256:54f77f8f5426f8515b4493c6d341994f56051fb8b58de826095eafd49b094a38|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:54f77f8f5426f8515b4493c6d341994f56051fb8b58de826095eafd49b094a38",
+    "source_id": "github:spiceai/spiceai#2187",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/dd/ddbf6171a8861a1b988067807215ae054fffa3d58f6d3e8fc3f9dcce3b3012d8.json b/.cache/impact/classifications/dd/ddbf6171a8861a1b988067807215ae054fffa3d58f6d3e8fc3f9dcce3b3012d8.json
new file mode 100644
index 0000000..11f8e7c
--- /dev/null
+++ b/.cache/impact/classifications/dd/ddbf6171a8861a1b988067807215ae054fffa3d58f6d3e8fc3f9dcce3b3012d8.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#105103|sha256:fc224961619d614ccff1c25f648a17a690cdd6818d1f93370831c7716c0a2d46|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:fc224961619d614ccff1c25f648a17a690cdd6818d1f93370831c7716c0a2d46",
+    "source_id": "github:cockroachdb/cockroach#105103",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/de/de61eba54f28ebcb15e3611b2d9421be4c29e7c3e5392e0e00910a4a86d342c9.json b/.cache/impact/classifications/de/de61eba54f28ebcb15e3611b2d9421be4c29e7c3e5392e0e00910a4a86d342c9.json
new file mode 100644
index 0000000..e0adb34
--- /dev/null
+++ b/.cache/impact/classifications/de/de61eba54f28ebcb15e3611b2d9421be4c29e7c3e5392e0e00910a4a86d342c9.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#89588|sha256:b7242ac131cdfdbf3eca7772282dd00fd404f4241003b5dea3d160846c4ddfd0|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:b7242ac131cdfdbf3eca7772282dd00fd404f4241003b5dea3d160846c4ddfd0",
+    "source_id": "github:cockroachdb/cockroach#89588",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/de/de7545d72dff7f63e38aa53ac4d2ec85ca3ac3f3eb4747d3138b79e725157cfa.json b/.cache/impact/classifications/de/de7545d72dff7f63e38aa53ac4d2ec85ca3ac3f3eb4747d3138b79e725157cfa.json
new file mode 100644
index 0000000..033af0e
--- /dev/null
+++ b/.cache/impact/classifications/de/de7545d72dff7f63e38aa53ac4d2ec85ca3ac3f3eb4747d3138b79e725157cfa.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:openlink/virtuoso-opensource#1429|sha256:73c9a90c6111501de92243677ea11e8db45c954dfb4429177b44309c680136a2|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "TLP (tautology-based partitioning) rewritten query"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:73c9a90c6111501de92243677ea11e8db45c954dfb4429177b44309c680136a2",
+    "source_id": "github:openlink/virtuoso-opensource#1429",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/de/dedb1a67633e3befb399b3114fc851245ce3afd330483f4051c6042610298f41.json b/.cache/impact/classifications/de/dedb1a67633e3befb399b3114fc851245ce3afd330483f4051c6042610298f41.json
new file mode 100644
index 0000000..71d454a
--- /dev/null
+++ b/.cache/impact/classifications/de/dedb1a67633e3befb399b3114fc851245ce3afd330483f4051c6042610298f41.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#7907|sha256:32fa20a7c2d0263d50b3839bf9a6696ffe661d744501b46501327325092a7abf|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:32fa20a7c2d0263d50b3839bf9a6696ffe661d744501b46501327325092a7abf",
+    "source_id": "github:cockroachdb/cockroach#7907",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/df/df17dba18cad9097af12174fc12c787e59d3152d6cad46ca6a70e22815d06a0f.json b/.cache/impact/classifications/df/df17dba18cad9097af12174fc12c787e59d3152d6cad46ca6a70e22815d06a0f.json
new file mode 100644
index 0000000..1bc953b
--- /dev/null
+++ b/.cache/impact/classifications/df/df17dba18cad9097af12174fc12c787e59d3152d6cad46ca6a70e22815d06a0f.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#53404|sha256:619a10ce3f2ab9c126531b59e95adb315707ed6137ac484ec57fcd34c9a88aa5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:619a10ce3f2ab9c126531b59e95adb315707ed6137ac484ec57fcd34c9a88aa5",
+    "source_id": "github:cockroachdb/cockroach#53404",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/df/df64860d29057569e21a94cd7d3f724f5f2554bc01414f19b8990334b24a5ff9.json b/.cache/impact/classifications/df/df64860d29057569e21a94cd7d3f724f5f2554bc01414f19b8990334b24a5ff9.json
new file mode 100644
index 0000000..c31e968
--- /dev/null
+++ b/.cache/impact/classifications/df/df64860d29057569e21a94cd7d3f724f5f2554bc01414f19b8990334b24a5ff9.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#94615|sha256:98831d231c805b319454a0b2dc074ba0158c7cdfb0c321e20035964079eb45f3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:98831d231c805b319454a0b2dc074ba0158c7cdfb0c321e20035964079eb45f3",
+    "source_id": "github:cockroachdb/cockroach#94615",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e0/e065dc4af509129ca6d66a7bac15f3a531ca80dac0176b2fcc6c864176128f69.json b/.cache/impact/classifications/e0/e065dc4af509129ca6d66a7bac15f3a531ca80dac0176b2fcc6c864176128f69.json
new file mode 100644
index 0000000..9f63f77
--- /dev/null
+++ b/.cache/impact/classifications/e0/e065dc4af509129ca6d66a7bac15f3a531ca80dac0176b2fcc6c864176128f69.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#13307|sha256:fb3fae43479201bfe6507050e29630807a3c5851e93cb2a8f15976c7fb0e6227|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:fb3fae43479201bfe6507050e29630807a3c5851e93cb2a8f15976c7fb0e6227",
+    "source_id": "github:cockroachdb/cockroach#13307",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e0/e06b4a8861ad41bf641738889457a7ee12e883355142d0dfba9e73cf49e1ad53.json b/.cache/impact/classifications/e0/e06b4a8861ad41bf641738889457a7ee12e883355142d0dfba9e73cf49e1ad53.json
new file mode 100644
index 0000000..51ad6d5
--- /dev/null
+++ b/.cache/impact/classifications/e0/e06b4a8861ad41bf641738889457a7ee12e883355142d0dfba9e73cf49e1ad53.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#128094|sha256:da0b05ee6df189a4d4d32e91c890f3b5f05a3b4424c9cc02de9c531cfb5c445b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:da0b05ee6df189a4d4d32e91c890f3b5f05a3b4424c9cc02de9c531cfb5c445b",
+    "source_id": "github:cockroachdb/cockroach#128094",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e0/e0812d7b44f3ef4c1042388713b90f85f8e893720fb47b42d5cde712e3964025.json b/.cache/impact/classifications/e0/e0812d7b44f3ef4c1042388713b90f85f8e893720fb47b42d5cde712e3964025.json
new file mode 100644
index 0000000..1fe96b9
--- /dev/null
+++ b/.cache/impact/classifications/e0/e0812d7b44f3ef4c1042388713b90f85f8e893720fb47b42d5cde712e3964025.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#173330|sha256:38fbabf75f5a742c43bef322ca7ad761afd694ea3b73392b42982e25459157c4|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:38fbabf75f5a742c43bef322ca7ad761afd694ea3b73392b42982e25459157c4",
+    "source_id": "github:cockroachdb/cockroach#173330",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e0/e08c7129e5147fe3cd78b129091942e9c603b3acc94eee4d3978849c6a550268.json b/.cache/impact/classifications/e0/e08c7129e5147fe3cd78b129091942e9c603b3acc94eee4d3978849c6a550268.json
new file mode 100644
index 0000000..e725576
--- /dev/null
+++ b/.cache/impact/classifications/e0/e08c7129e5147fe3cd78b129091942e9c603b3acc94eee4d3978849c6a550268.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#95031|sha256:1ff0790c98cd60118b7f3d602bfec35c4234eab6b647c8a8a3a786bfae8d9311|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:1ff0790c98cd60118b7f3d602bfec35c4234eab6b647c8a8a3a786bfae8d9311",
+    "source_id": "github:cockroachdb/cockroach#95031",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e0/e0c83f7c5c4a15a3f5ce0154b0b6eb8a1c1db7e0edc15c0943e66ebbf602705b.json b/.cache/impact/classifications/e0/e0c83f7c5c4a15a3f5ce0154b0b6eb8a1c1db7e0edc15c0943e66ebbf602705b.json
new file mode 100644
index 0000000..187a74b
--- /dev/null
+++ b/.cache/impact/classifications/e0/e0c83f7c5c4a15a3f5ce0154b0b6eb8a1c1db7e0edc15c0943e66ebbf602705b.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#74295|sha256:4b3449725cb5c5ca189d48c7e0d504802aa63965201dd639a622bedb21fd5268|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4b3449725cb5c5ca189d48c7e0d504802aa63965201dd639a622bedb21fd5268",
+    "source_id": "github:cockroachdb/cockroach#74295",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e0/e0fb32134077747bf777e72cd6cb950d503814f9aed886d66bfa88a9e7e0cb9d.json b/.cache/impact/classifications/e0/e0fb32134077747bf777e72cd6cb950d503814f9aed886d66bfa88a9e7e0cb9d.json
new file mode 100644
index 0000000..c327df1
--- /dev/null
+++ b/.cache/impact/classifications/e0/e0fb32134077747bf777e72cd6cb950d503814f9aed886d66bfa88a9e7e0cb9d.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#151457|sha256:2991f9f2daf253953cbaae778eee4c79bbee8af70afce7bdfa6cfb59f40f9550|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2991f9f2daf253953cbaae778eee4c79bbee8af70afce7bdfa6cfb59f40f9550",
+    "source_id": "github:cockroachdb/cockroach#151457",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e1/e1eb06e050baa67037a91d3ee4bf4dbb4cab5925cd7af52f5c148eb1f459cdfd.json b/.cache/impact/classifications/e1/e1eb06e050baa67037a91d3ee4bf4dbb4cab5925cd7af52f5c148eb1f459cdfd.json
new file mode 100644
index 0000000..f230d04
--- /dev/null
+++ b/.cache/impact/classifications/e1/e1eb06e050baa67037a91d3ee4bf4dbb4cab5925cd7af52f5c148eb1f459cdfd.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#89017|sha256:5cf8aa02564941c0ce65a16c2f8543ff901c55a015196c21c893642d36250b59|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5cf8aa02564941c0ce65a16c2f8543ff901c55a015196c21c893642d36250b59",
+    "source_id": "github:elastic/elasticsearch#89017",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e2/e22f22f74ff7ffe80d062e4d31416dc5539e8a59dc1b12f107032cdc53492561.json b/.cache/impact/classifications/e2/e22f22f74ff7ffe80d062e4d31416dc5539e8a59dc1b12f107032cdc53492561.json
new file mode 100644
index 0000000..92ad017
--- /dev/null
+++ b/.cache/impact/classifications/e2/e22f22f74ff7ffe80d062e4d31416dc5539e8a59dc1b12f107032cdc53492561.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#88910|sha256:78328a24b2c1673cf4758160ab241e39deac6de37a0d680c793c9f69c11c2d99|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "expected unpartitioned and partitioned results to be equal"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:78328a24b2c1673cf4758160ab241e39deac6de37a0d680c793c9f69c11c2d99",
+    "source_id": "github:cockroachdb/cockroach#88910",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e3/e3266a7ef15df348bf653399d3a1382721f53278a7c285b051242ab46e3b6738.json b/.cache/impact/classifications/e3/e3266a7ef15df348bf653399d3a1382721f53278a7c285b051242ab46e3b6738.json
new file mode 100644
index 0000000..6a29da1
--- /dev/null
+++ b/.cache/impact/classifications/e3/e3266a7ef15df348bf653399d3a1382721f53278a7c285b051242ab46e3b6738.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#94270|sha256:369b83930e2f4206d75b28f8efbb3b59b4bf6e16829c879b01f0c0c788655ff5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:369b83930e2f4206d75b28f8efbb3b59b4bf6e16829c879b01f0c0c788655ff5",
+    "source_id": "github:elastic/elasticsearch#94270",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e3/e327f1ae926f9715f843b350cc485083468e34163df1b7b19e861fea896d79ed.json b/.cache/impact/classifications/e3/e327f1ae926f9715f843b350cc485083468e34163df1b7b19e861fea896d79ed.json
new file mode 100644
index 0000000..7707ed3
--- /dev/null
+++ b/.cache/impact/classifications/e3/e327f1ae926f9715f843b350cc485083468e34163df1b7b19e861fea896d79ed.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:spiceai/spiceai#2119|sha256:6a89c8cc1f3cbf7e187afa755dfcad0cadec20a1972ce8b72536cf4668dce72a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6a89c8cc1f3cbf7e187afa755dfcad0cadec20a1972ce8b72536cf4668dce72a",
+    "source_id": "github:spiceai/spiceai#2119",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e3/e37dc7a5c72a2548d0860103a6f13e12f1bfccb485ef6f4db2f644be175d6606.json b/.cache/impact/classifications/e3/e37dc7a5c72a2548d0860103a6f13e12f1bfccb485ef6f4db2f644be175d6606.json
new file mode 100644
index 0000000..588cc04
--- /dev/null
+++ b/.cache/impact/classifications/e3/e37dc7a5c72a2548d0860103a6f13e12f1bfccb485ef6f4db2f644be175d6606.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:opensearch-project/sql#3220|sha256:9f05df40dc9c68de46eb8a2a07362dcdb2542ac992737c5918c1b88adb808bd8|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9f05df40dc9c68de46eb8a2a07362dcdb2542ac992737c5918c1b88adb808bd8",
+    "source_id": "github:opensearch-project/sql#3220",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e3/e38e2d6e0501c9fe42b03bb3cb0e3cf4a09739a5829ee1c8d0b7bb59c056d756.json b/.cache/impact/classifications/e3/e38e2d6e0501c9fe42b03bb3cb0e3cf4a09739a5829ee1c8d0b7bb59c056d756.json
new file mode 100644
index 0000000..2c0359d
--- /dev/null
+++ b/.cache/impact/classifications/e3/e38e2d6e0501c9fe42b03bb3cb0e3cf4a09739a5829ee1c8d0b7bb59c056d756.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#104988|sha256:839e5a275e4a28d32e6e37a48a80497d2e36c29709e6510aeedf67e80a03e60a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:839e5a275e4a28d32e6e37a48a80497d2e36c29709e6510aeedf67e80a03e60a",
+    "source_id": "github:cockroachdb/cockroach#104988",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e4/e4280b6577b15b1e1e72ab9f2a9811b0429e8873beab9e2c568ec0051691d2aa.json b/.cache/impact/classifications/e4/e4280b6577b15b1e1e72ab9f2a9811b0429e8873beab9e2c568ec0051691d2aa.json
new file mode 100644
index 0000000..3c2ae69
--- /dev/null
+++ b/.cache/impact/classifications/e4/e4280b6577b15b1e1e72ab9f2a9811b0429e8873beab9e2c568ec0051691d2aa.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#14864|sha256:6f652a73480a7e21349ff4f728a8ea603424ef90b2655c9bcb2d1b1d8b505a16|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6f652a73480a7e21349ff4f728a8ea603424ef90b2655c9bcb2d1b1d8b505a16",
+    "source_id": "github:yugabyte/yugabyte-db#14864",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e4/e4d8a7b759a9b17dc19be4f69afc8ed9edfc49d8557b16b07d954d7a21082df4.json b/.cache/impact/classifications/e4/e4d8a7b759a9b17dc19be4f69afc8ed9edfc49d8557b16b07d954d7a21082df4.json
new file mode 100644
index 0000000..8013f8f
--- /dev/null
+++ b/.cache/impact/classifications/e4/e4d8a7b759a9b17dc19be4f69afc8ed9edfc49d8557b16b07d954d7a21082df4.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#123062|sha256:af3332c4cfd91d1f28174c1d92025541f7d3e6f8ef789bf7008e4d02252ec639|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:af3332c4cfd91d1f28174c1d92025541f7d3e6f8ef789bf7008e4d02252ec639",
+    "source_id": "github:cockroachdb/cockroach#123062",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e6/e6066aa9ae22acc719577e983dd1a6691ddb21e1aaad5acdd4dbc286f872f2f4.json b/.cache/impact/classifications/e6/e6066aa9ae22acc719577e983dd1a6691ddb21e1aaad5acdd4dbc286f872f2f4.json
new file mode 100644
index 0000000..f9dfe90
--- /dev/null
+++ b/.cache/impact/classifications/e6/e6066aa9ae22acc719577e983dd1a6691ddb21e1aaad5acdd4dbc286f872f2f4.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:cockroachdb/cockroach#70587|sha256:59a10e27b89b1ed4d9198b5a5b40bba5495dc840bea25eca1ff1a82219410d16|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "The count is incorrectly computed for the unpartitioned"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": "tlp"
+      },
+      "reason": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:59a10e27b89b1ed4d9198b5a5b40bba5495dc840bea25eca1ff1a82219410d16",
+    "source_id": "github:cockroachdb/cockroach#70587",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e7/e75f46c32e9cae496c3bd58de35158465987d9d395af11cd640be8fe42c42734.json b/.cache/impact/classifications/e7/e75f46c32e9cae496c3bd58de35158465987d9d395af11cd640be8fe42c42734.json
new file mode 100644
index 0000000..df78f34
--- /dev/null
+++ b/.cache/impact/classifications/e7/e75f46c32e9cae496c3bd58de35158465987d9d395af11cd640be8fe42c42734.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#10543|sha256:58febba290e4414297628d093f9f73be0fbf9ea0fe4da0dfbcd441281f608ad8|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:58febba290e4414297628d093f9f73be0fbf9ea0fe4da0dfbcd441281f608ad8",
+    "source_id": "github:cockroachdb/cockroach#10543",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e7/e79e02e1aedb3b0a5c88c0527b42ead092999dc5c1ac8b61ab0e41fc60dff552.json b/.cache/impact/classifications/e7/e79e02e1aedb3b0a5c88c0527b42ead092999dc5c1ac8b61ab0e41fc60dff552.json
new file mode 100644
index 0000000..74e5c81
--- /dev/null
+++ b/.cache/impact/classifications/e7/e79e02e1aedb3b0a5c88c0527b42ead092999dc5c1ac8b61ab0e41fc60dff552.json
@@ -0,0 +1,24 @@
+{
+  "key": "adoption:https://github.com/cnosdb/cnosdb/blob/c7609432edfdefd5a442269af42a99d61d5e87c1/README_CN.md|sha256:0f27bfeb42face52bdf31ebb42c2ff166c3ff148d090cbd2dcae7c2e88e94071|adoption-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T07:17:28Z",
+  "value": {
+    "classified_at": "2026-09-13T07:17:28Z",
+    "classifier_version": "adoption-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "SQLancer](https://github.com/sqlancer/sqlancer)作为支持。"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "relationship": "official_testing"
+      },
+      "reason": "The Chinese README carries the same statement as the English one. Decided by reading the file in this session rather than by an API call."
+    },
+    "source_hash": "sha256:0f27bfeb42face52bdf31ebb42c2ff166c3ff148d090cbd2dcae7c2e88e94071",
+    "source_id": "adoption:https://github.com/cnosdb/cnosdb/blob/c7609432edfdefd5a442269af42a99d61d5e87c1/README_CN.md",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e7/e7dabe197493a78ea9b140485b99bf6199e451f19300d9e231a2feff1f8a3af8.json b/.cache/impact/classifications/e7/e7dabe197493a78ea9b140485b99bf6199e451f19300d9e231a2feff1f8a3af8.json
new file mode 100644
index 0000000..01f5edd
--- /dev/null
+++ b/.cache/impact/classifications/e7/e7dabe197493a78ea9b140485b99bf6199e451f19300d9e231a2feff1f8a3af8.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4100|sha256:a19ec3b849815af03a1a99082db6df390d18c30c6ab969d0ff708685bb4684a3|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The acronym is not the oracle here: an Apache Top-Level Project, a transport name, a generated identifier, or a proposal to improve the test suite rather than something it found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a19ec3b849815af03a1a99082db6df390d18c30c6ab969d0ff708685bb4684a3",
+    "source_id": "github:sparq-org/sparq#4100",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e8/e82b0b636ffe6a1ad71637d690835ed5d709abad7af5c5956e5483215b33eb89.json b/.cache/impact/classifications/e8/e82b0b636ffe6a1ad71637d690835ed5d709abad7af5c5956e5483215b33eb89.json
new file mode 100644
index 0000000..107d829
--- /dev/null
+++ b/.cache/impact/classifications/e8/e82b0b636ffe6a1ad71637d690835ed5d709abad7af5c5956e5483215b33eb89.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#144813|sha256:cd55b12c6ba474bdf11aff7568cdfd7f347588dbb66e839ce35d64883c37dd72|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:cd55b12c6ba474bdf11aff7568cdfd7f347588dbb66e839ce35d64883c37dd72",
+    "source_id": "github:cockroachdb/cockroach#144813",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e8/e86dedc3b4cdcb26e31f7ecced2c6f88254480a6e8fc723cb0761429e0e82f17.json b/.cache/impact/classifications/e8/e86dedc3b4cdcb26e31f7ecced2c6f88254480a6e8fc723cb0761429e0e82f17.json
new file mode 100644
index 0000000..3b55234
--- /dev/null
+++ b/.cache/impact/classifications/e8/e86dedc3b4cdcb26e31f7ecced2c6f88254480a6e8fc723cb0761429e0e82f17.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:neo4j/neo4j#3740|sha256:986fa8b02efa82c02f9ae5309b4389a776f0f9d80e2f6d398c93007247c2770b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:986fa8b02efa82c02f9ae5309b4389a776f0f9d80e2f6d398c93007247c2770b",
+    "source_id": "github:neo4j/neo4j#3740",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e9/e98fb6ab65b64999ef96a241989fadcb6e7f579235727cc074594d8598b8ee7d.json b/.cache/impact/classifications/e9/e98fb6ab65b64999ef96a241989fadcb6e7f579235727cc074594d8598b8ee7d.json
new file mode 100644
index 0000000..fd19d40
--- /dev/null
+++ b/.cache/impact/classifications/e9/e98fb6ab65b64999ef96a241989fadcb6e7f579235727cc074594d8598b8ee7d.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:serenedb/serenedb#892|sha256:894cae20e25b2e32d63cfca47c018d06dd16c006d038731e6aa337e9a40686bb|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:894cae20e25b2e32d63cfca47c018d06dd16c006d038731e6aa337e9a40686bb",
+    "source_id": "github:serenedb/serenedb#892",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e9/e9de857cb133843e69b8741ffe6988f56777c79fe8f7e0b560f7a2dfd59599a1.json b/.cache/impact/classifications/e9/e9de857cb133843e69b8741ffe6988f56777c79fe8f7e0b560f7a2dfd59599a1.json
new file mode 100644
index 0000000..51c0621
--- /dev/null
+++ b/.cache/impact/classifications/e9/e9de857cb133843e69b8741ffe6988f56777c79fe8f7e0b560f7a2dfd59599a1.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#65087|sha256:7df1024c3c9f5d23cc235a2a649f7808ca1443d9a9319aa7aa98063e98bfb2cd|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7df1024c3c9f5d23cc235a2a649f7808ca1443d9a9319aa7aa98063e98bfb2cd",
+    "source_id": "github:cockroachdb/cockroach#65087",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/e9/e9ef03c53497bcce87d3ea5a8b2b0642defcd119922d73681a73a3e12169d280.json b/.cache/impact/classifications/e9/e9ef03c53497bcce87d3ea5a8b2b0642defcd119922d73681a73a3e12169d280.json
new file mode 100644
index 0000000..da0f713
--- /dev/null
+++ b/.cache/impact/classifications/e9/e9ef03c53497bcce87d3ea5a8b2b0642defcd119922d73681a73a3e12169d280.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:sparq-org/sparq#4101|sha256:9c19ab6c8f7a6f875f1ee69c0c3b1ea69bb46bf325063a0331047bcf6ecfae6a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9c19ab6c8f7a6f875f1ee69c0c3b1ea69bb46bf325063a0331047bcf6ecfae6a",
+    "source_id": "github:sparq-org/sparq#4101",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ea/ea4f91f6f6c0cf235d9d7d79446c928a9187ec0b14439604f766ffa2b504453b.json b/.cache/impact/classifications/ea/ea4f91f6f6c0cf235d9d7d79446c928a9187ec0b14439604f766ffa2b504453b.json
new file mode 100644
index 0000000..3b2b536
--- /dev/null
+++ b/.cache/impact/classifications/ea/ea4f91f6f6c0cf235d9d7d79446c928a9187ec0b14439604f766ffa2b504453b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#23517|sha256:aa17037c2cfa125125bb67dc3b636ec05e83c82aed6d0f7b63b3270b61fb532e|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:aa17037c2cfa125125bb67dc3b636ec05e83c82aed6d0f7b63b3270b61fb532e",
+    "source_id": "github:yugabyte/yugabyte-db#23517",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ea/ea9b3e60fb14b0c66cbaa8d53a420fd0f39ae70a52ada3fe312a082867b1eff4.json b/.cache/impact/classifications/ea/ea9b3e60fb14b0c66cbaa8d53a420fd0f39ae70a52ada3fe312a082867b1eff4.json
new file mode 100644
index 0000000..0c9963a
--- /dev/null
+++ b/.cache/impact/classifications/ea/ea9b3e60fb14b0c66cbaa8d53a420fd0f39ae70a52ada3fe312a082867b1eff4.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#94551|sha256:76d159e5616cb7c884e3dbb9aaa2e073c38991e24e2cb6905e10566c127f42c8|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:76d159e5616cb7c884e3dbb9aaa2e073c38991e24e2cb6905e10566c127f42c8",
+    "source_id": "github:cockroachdb/cockroach#94551",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ea/eadabe417ccb09f7a658e8060def987c7feee54afafa416c7e522171a901cc39.json b/.cache/impact/classifications/ea/eadabe417ccb09f7a658e8060def987c7feee54afafa416c7e522171a901cc39.json
new file mode 100644
index 0000000..57fd56d
--- /dev/null
+++ b/.cache/impact/classifications/ea/eadabe417ccb09f7a658e8060def987c7feee54afafa416c7e522171a901cc39.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#131110|sha256:dbf53bf6a6b4ef36c6ea99e891abaa3f221bfeebbe8d02f5e5949465242c7f94|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:dbf53bf6a6b4ef36c6ea99e891abaa3f221bfeebbe8d02f5e5949465242c7f94",
+    "source_id": "github:cockroachdb/cockroach#131110",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ea/eae5956e7387d12b2c1768a1875c7fdc856eb4789e6425bb5fc1117ec58d0301.json b/.cache/impact/classifications/ea/eae5956e7387d12b2c1768a1875c7fdc856eb4789e6425bb5fc1117ec58d0301.json
new file mode 100644
index 0000000..6245f0a
--- /dev/null
+++ b/.cache/impact/classifications/ea/eae5956e7387d12b2c1768a1875c7fdc856eb4789e6425bb5fc1117ec58d0301.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:derekmwright/wadjet#626|sha256:ac337d8053110d8df8d6ed3d30e9c59de0733719fc78dda22f8e7dd80705024a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ac337d8053110d8df8d6ed3d30e9c59de0733719fc78dda22f8e7dd80705024a",
+    "source_id": "github:derekmwright/wadjet#626",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/eb/ebb001b3d79269e8a2c9a77033344cb17ee5e255d790f3e94111b34cd07e19cc.json b/.cache/impact/classifications/eb/ebb001b3d79269e8a2c9a77033344cb17ee5e255d790f3e94111b34cd07e19cc.json
new file mode 100644
index 0000000..e8be358
--- /dev/null
+++ b/.cache/impact/classifications/eb/ebb001b3d79269e8a2c9a77033344cb17ee5e255d790f3e94111b34cd07e19cc.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:matrixorigin/matrixone#1670|sha256:f8ada5ccd2321b029eaad95cf78d9ab0e9704e88b7dce2b3823ec860016a2fe4|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Mentions an oracle in passing while proposing work on the test suite itself, rather than reporting something the oracle found. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f8ada5ccd2321b029eaad95cf78d9ab0e9704e88b7dce2b3823ec860016a2fe4",
+    "source_id": "github:matrixorigin/matrixone#1670",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/eb/ebeef0e27eea46336aff68913407a944dd6fd3a3f6adc432b058abc4197b8123.json b/.cache/impact/classifications/eb/ebeef0e27eea46336aff68913407a944dd6fd3a3f6adc432b058abc4197b8123.json
new file mode 100644
index 0000000..fdd3322
--- /dev/null
+++ b/.cache/impact/classifications/eb/ebeef0e27eea46336aff68913407a944dd6fd3a3f6adc432b058abc4197b8123.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:derekmwright/wadjet#588|sha256:f50aa0ae57396539b9ea6bafe24fb0995455f72566eebd79d5ec08478733ee01|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f50aa0ae57396539b9ea6bafe24fb0995455f72566eebd79d5ec08478733ee01",
+    "source_id": "github:derekmwright/wadjet#588",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ec/ecff98d9d9b0b92d240e96ffe62e6bacb2e626c3291e663473ad9007eed18b4b.json b/.cache/impact/classifications/ec/ecff98d9d9b0b92d240e96ffe62e6bacb2e626c3291e663473ad9007eed18b4b.json
new file mode 100644
index 0000000..c7808b8
--- /dev/null
+++ b/.cache/impact/classifications/ec/ecff98d9d9b0b92d240e96ffe62e6bacb2e626c3291e663473ad9007eed18b4b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#66435|sha256:e814dbc64375807e83e5bba6bc58a2faf800412bb7ae9ef558438a3644749ead|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e814dbc64375807e83e5bba6bc58a2faf800412bb7ae9ef558438a3644749ead",
+    "source_id": "github:cockroachdb/cockroach#66435",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ed/ed4b6de4191e743c2963902dc9fb7f7d4a560cef3ea4893f20463251b09af809.json b/.cache/impact/classifications/ed/ed4b6de4191e743c2963902dc9fb7f7d4a560cef3ea4893f20463251b09af809.json
new file mode 100644
index 0000000..1d37483
--- /dev/null
+++ b/.cache/impact/classifications/ed/ed4b6de4191e743c2963902dc9fb7f7d4a560cef3ea4893f20463251b09af809.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#171981|sha256:6ffad4282acdf028206cc4ab944964a1506e03da8ef66c833c2948af6f2d5831|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6ffad4282acdf028206cc4ab944964a1506e03da8ef66c833c2948af6f2d5831",
+    "source_id": "github:cockroachdb/cockroach#171981",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ed/edd13a2f0e7fc99d8c8fd6fab8e036babe276c06460f0b547b2b586dfba65d21.json b/.cache/impact/classifications/ed/edd13a2f0e7fc99d8c8fd6fab8e036babe276c06460f0b547b2b586dfba65d21.json
new file mode 100644
index 0000000..e9a6c84
--- /dev/null
+++ b/.cache/impact/classifications/ed/edd13a2f0e7fc99d8c8fd6fab8e036babe276c06460f0b547b2b586dfba65d21.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#31338|sha256:4024f3ebb289893bfd48caa5501da64c6c40427d189b36dfd150b5851ad0dba1|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4024f3ebb289893bfd48caa5501da64c6c40427d189b36dfd150b5851ad0dba1",
+    "source_id": "github:cockroachdb/cockroach#31338",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f0/f03e0224afdf007e3ccded8cae4346571182571d475775c3102204a41fe8fb52.json b/.cache/impact/classifications/f0/f03e0224afdf007e3ccded8cae4346571182571d475775c3102204a41fe8fb52.json
new file mode 100644
index 0000000..738d078
--- /dev/null
+++ b/.cache/impact/classifications/f0/f03e0224afdf007e3ccded8cae4346571182571d475775c3102204a41fe8fb52.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#14985|sha256:8475019161b9d2b7ed3ede0b94add48117dcc1522a5df7e8c625776c46913a11|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:8475019161b9d2b7ed3ede0b94add48117dcc1522a5df7e8c625776c46913a11",
+    "source_id": "github:cockroachdb/cockroach#14985",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f0/f0908cf7625cdf49be7ec344fc784c4010eac45911e85654cff110f7bdadc0cd.json b/.cache/impact/classifications/f0/f0908cf7625cdf49be7ec344fc784c4010eac45911e85654cff110f7bdadc0cd.json
new file mode 100644
index 0000000..c4ef828
--- /dev/null
+++ b/.cache/impact/classifications/f0/f0908cf7625cdf49be7ec344fc784c4010eac45911e85654cff110f7bdadc0cd.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:stoneatom/stonedb#987|sha256:9f03e0b62ad2d17359eb6d664d7ee8a1adb0edd61cb34b170dc63f9ac193b4d9|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:9f03e0b62ad2d17359eb6d664d7ee8a1adb0edd61cb34b170dc63f9ac193b4d9",
+    "source_id": "github:stoneatom/stonedb#987",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f2/f20cd8c531d9c2bb4cf49f95c0a38df8f780137fc0f65fab754162354f82a485.json b/.cache/impact/classifications/f2/f20cd8c531d9c2bb4cf49f95c0a38df8f780137fc0f65fab754162354f82a485.json
new file mode 100644
index 0000000..bf9faa1
--- /dev/null
+++ b/.cache/impact/classifications/f2/f20cd8c531d9c2bb4cf49f95c0a38df8f780137fc0f65fab754162354f82a485.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#175282|sha256:42e8b27b724ee636fa38cd2f93993b239b2c44658f4076d811ef9e4c9b6af070|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:42e8b27b724ee636fa38cd2f93993b239b2c44658f4076d811ef9e4c9b6af070",
+    "source_id": "github:cockroachdb/cockroach#175282",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f2/f2c297dcabfc92088e81dcc6d89f3064e55b90ae349522495d673f8bc25e1e16.json b/.cache/impact/classifications/f2/f2c297dcabfc92088e81dcc6d89f3064e55b90ae349522495d673f8bc25e1e16.json
new file mode 100644
index 0000000..bdc77d5
--- /dev/null
+++ b/.cache/impact/classifications/f2/f2c297dcabfc92088e81dcc6d89f3064e55b90ae349522495d673f8bc25e1e16.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#151995|sha256:bbb7d17f92a2c9cc2a8d26a350c664ed0f784472a513294fd6fd7a936f275104|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:bbb7d17f92a2c9cc2a8d26a350c664ed0f784472a513294fd6fd7a936f275104",
+    "source_id": "github:cockroachdb/cockroach#151995",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f2/f2f7ab462bb707be89f85d060601aef0c9fcdc4200c2bc7b925246c8004e4918.json b/.cache/impact/classifications/f2/f2f7ab462bb707be89f85d060601aef0c9fcdc4200c2bc7b925246c8004e4918.json
new file mode 100644
index 0000000..6c3b938
--- /dev/null
+++ b/.cache/impact/classifications/f2/f2f7ab462bb707be89f85d060601aef0c9fcdc4200c2bc7b925246c8004e4918.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cmu-db/noisepage#1041|sha256:0c38f2c967b6b7b11553673e9169c4ee3a4121909bb2cbf1cb66765093acd5c6|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0c38f2c967b6b7b11553673e9169c4ee3a4121909bb2cbf1cb66765093acd5c6",
+    "source_id": "github:cmu-db/noisepage#1041",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f4/f45addada480068179c1a041dfa4a685f4c36ea97ead3c529fac595dd1746abc.json b/.cache/impact/classifications/f4/f45addada480068179c1a041dfa4a685f4c36ea97ead3c529fac595dd1746abc.json
new file mode 100644
index 0000000..4071374
--- /dev/null
+++ b/.cache/impact/classifications/f4/f45addada480068179c1a041dfa4a685f4c36ea97ead3c529fac595dd1746abc.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#77439|sha256:031aa17ec3ae96963622e6de68bafa71d80e42520af9a726f82e5b099ab99666|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:031aa17ec3ae96963622e6de68bafa71d80e42520af9a726f82e5b099ab99666",
+    "source_id": "github:cockroachdb/cockroach#77439",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f4/f492c18fbb3982068b8845c0c8fef9d0f05c63248255d912584e680a8fffa206.json b/.cache/impact/classifications/f4/f492c18fbb3982068b8845c0c8fef9d0f05c63248255d912584e680a8fffa206.json
new file mode 100644
index 0000000..6d074b5
--- /dev/null
+++ b/.cache/impact/classifications/f4/f492c18fbb3982068b8845c0c8fef9d0f05c63248255d912584e680a8fffa206.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:ClickHouse/ClickHouse#17623|sha256:5040bf38421b020a4343bec068cada88ba0b3c19e208a25181ced8fd910dc19d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:5040bf38421b020a4343bec068cada88ba0b3c19e208a25181ced8fd910dc19d",
+    "source_id": "github:ClickHouse/ClickHouse#17623",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f5/f5397432722e3fbb02a8e0d78c25424dc97d8e7be9368cbd0465bf14194a86c9.json b/.cache/impact/classifications/f5/f5397432722e3fbb02a8e0d78c25424dc97d8e7be9368cbd0465bf14194a86c9.json
new file mode 100644
index 0000000..9607fd3
--- /dev/null
+++ b/.cache/impact/classifications/f5/f5397432722e3fbb02a8e0d78c25424dc97d8e7be9368cbd0465bf14194a86c9.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#5594|sha256:959d7fbbe3eb5b33170373cabc2d6dc8e30598796f37f78962adc50266874db2|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:959d7fbbe3eb5b33170373cabc2d6dc8e30598796f37f78962adc50266874db2",
+    "source_id": "github:yugabyte/yugabyte-db#5594",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f5/f5ae654869ea91a7ca772b15323597afc58229295f8a2662614f7ea81ab89663.json b/.cache/impact/classifications/f5/f5ae654869ea91a7ca772b15323597afc58229295f8a2662614f7ea81ab89663.json
new file mode 100644
index 0000000..1ce0347
--- /dev/null
+++ b/.cache/impact/classifications/f5/f5ae654869ea91a7ca772b15323597afc58229295f8a2662614f7ea81ab89663.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#62902|sha256:4bcd5e32e45ac2932a1f49a59009006ea5ae236dbe4134a3eb71c25fd5911135|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4bcd5e32e45ac2932a1f49a59009006ea5ae236dbe4134a3eb71c25fd5911135",
+    "source_id": "github:cockroachdb/cockroach#62902",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f5/f5bc53a07c83087327df462738c3054b22b99ea3a337f2fa4ac7cd9a12aed487.json b/.cache/impact/classifications/f5/f5bc53a07c83087327df462738c3054b22b99ea3a337f2fa4ac7cd9a12aed487.json
new file mode 100644
index 0000000..a5e973f
--- /dev/null
+++ b/.cache/impact/classifications/f5/f5bc53a07c83087327df462738c3054b22b99ea3a337f2fa4ac7cd9a12aed487.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8153|sha256:2794bf7dc1f1d9983ad3f62828ec505a80da0973902b39a2768bc77223ac499d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2794bf7dc1f1d9983ad3f62828ec505a80da0973902b39a2768bc77223ac499d",
+    "source_id": "github:cockroachdb/cockroach#8153",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f6/f6558e93efc0f9ecf983d66acfe7e9a71f42449c0778a062bb527337efe73b88.json b/.cache/impact/classifications/f6/f6558e93efc0f9ecf983d66acfe7e9a71f42449c0778a062bb527337efe73b88.json
new file mode 100644
index 0000000..480910a
--- /dev/null
+++ b/.cache/impact/classifications/f6/f6558e93efc0f9ecf983d66acfe7e9a71f42449c0778a062bb527337efe73b88.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:elastic/elasticsearch#150128|sha256:4334ff7a7cafad097f8c93f14ea72c0e6b66df63dc6905902d10a716bd06058d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4334ff7a7cafad097f8c93f14ea72c0e6b66df63dc6905902d10a716bd06058d",
+    "source_id": "github:elastic/elasticsearch#150128",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f7/f73a06ce379860c348599a463e340bd35f3021a3e9c73a803c9aef52bdf2939d.json b/.cache/impact/classifications/f7/f73a06ce379860c348599a463e340bd35f3021a3e9c73a803c9aef52bdf2939d.json
new file mode 100644
index 0000000..7f83069
--- /dev/null
+++ b/.cache/impact/classifications/f7/f73a06ce379860c348599a463e340bd35f3021a3e9c73a803c9aef52bdf2939d.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#3060|sha256:cd22a87fae6ff9ee1c316a3fad440914933777913c490d995e82098d764dcc19|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:cd22a87fae6ff9ee1c316a3fad440914933777913c490d995e82098d764dcc19",
+    "source_id": "github:cockroachdb/cockroach#3060",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f8/f83edbf83ee91b60290f80b535d9f4a9f0fa467c3bd1602f1f2f38dc8f0b1a9a.json b/.cache/impact/classifications/f8/f83edbf83ee91b60290f80b535d9f4a9f0fa467c3bd1602f1f2f38dc8f0b1a9a.json
new file mode 100644
index 0000000..07fb560
--- /dev/null
+++ b/.cache/impact/classifications/f8/f83edbf83ee91b60290f80b535d9f4a9f0fa467c3bd1602f1f2f38dc8f0b1a9a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:databendlabs/databend#1768|sha256:76b55f9e450d058030fb8c26eff1761c29e3ba6f34d9753311f7d7c6bbd2c1c4|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:76b55f9e450d058030fb8c26eff1761c29e3ba6f34d9753311f7d7c6bbd2c1c4",
+    "source_id": "github:databendlabs/databend#1768",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f8/f8a8008932d19705f5a06574ed730b8e18cfe811cb1fd705ae73002402c715a2.json b/.cache/impact/classifications/f8/f8a8008932d19705f5a06574ed730b8e18cfe811cb1fd705ae73002402c715a2.json
new file mode 100644
index 0000000..fa88e3d
--- /dev/null
+++ b/.cache/impact/classifications/f8/f8a8008932d19705f5a06574ed730b8e18cfe811cb1fd705ae73002402c715a2.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#94920|sha256:f52a5eb269a455b4c470179dd1559cb17a2b0630216b1403d91dd7a5a7d947ae|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:f52a5eb269a455b4c470179dd1559cb17a2b0630216b1403d91dd7a5a7d947ae",
+    "source_id": "github:cockroachdb/cockroach#94920",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f8/f8e0170a23002eeaad3ccd7ae365671678bfd4307407876fe15860eb90934d7c.json b/.cache/impact/classifications/f8/f8e0170a23002eeaad3ccd7ae365671678bfd4307407876fe15860eb90934d7c.json
new file mode 100644
index 0000000..4c0aec7
--- /dev/null
+++ b/.cache/impact/classifications/f8/f8e0170a23002eeaad3ccd7ae365671678bfd4307407876fe15860eb90934d7c.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#126758|sha256:910a7f198591d79852648e09ea80387b114fb8a714928f9beda85a5f549deaf7|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:910a7f198591d79852648e09ea80387b114fb8a714928f9beda85a5f549deaf7",
+    "source_id": "github:cockroachdb/cockroach#126758",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f9/f91d20d7246b0ce09f79b4b2252f57baac3682fb852067d58e5502a6e3c73ba0.json b/.cache/impact/classifications/f9/f91d20d7246b0ce09f79b4b2252f57baac3682fb852067d58e5502a6e3c73ba0.json
new file mode 100644
index 0000000..4fef42d
--- /dev/null
+++ b/.cache/impact/classifications/f9/f91d20d7246b0ce09f79b4b2252f57baac3682fb852067d58e5502a6e3c73ba0.json
@@ -0,0 +1,24 @@
+{
+  "key": "github:yugabyte/yugabyte-db#19321|sha256:0de6cb55e846a4d8b68068f36316520cf5091293a8c57ad647541aca5939d1b8|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "yes",
+      "excerpts": [
+        "sqlancer.common.query.SQLQueryAdapter"
+      ],
+      "extra": {
+        "finder": "sqlancer",
+        "technique": null
+      },
+      "reason": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0de6cb55e846a4d8b68068f36316520cf5091293a8c57ad647541aca5939d1b8",
+    "source_id": "github:yugabyte/yugabyte-db#19321",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f9/f94db7eebdeeffdf4cdfc924066fc711118fb797ef6d5c7d45f6901927836478.json b/.cache/impact/classifications/f9/f94db7eebdeeffdf4cdfc924066fc711118fb797ef6d5c7d45f6901927836478.json
new file mode 100644
index 0000000..4c02944
--- /dev/null
+++ b/.cache/impact/classifications/f9/f94db7eebdeeffdf4cdfc924066fc711118fb797ef6d5c7d45f6901927836478.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#14427|sha256:7e50d43839e57b5d0614cb2981c3b27394ddb84ec0329410c56f9bb9512463bf|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7e50d43839e57b5d0614cb2981c3b27394ddb84ec0329410c56f9bb9512463bf",
+    "source_id": "github:yugabyte/yugabyte-db#14427",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f9/f978057b7d5525c04e0f35af0ef3a5a68dffe69fd7aec13fe037e43ece9aff04.json b/.cache/impact/classifications/f9/f978057b7d5525c04e0f35af0ef3a5a68dffe69fd7aec13fe037e43ece9aff04.json
new file mode 100644
index 0000000..361e436
--- /dev/null
+++ b/.cache/impact/classifications/f9/f978057b7d5525c04e0f35af0ef3a5a68dffe69fd7aec13fe037e43ece9aff04.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#6786|sha256:a7e39e7e8e29e2aabcd56f9a6e02f093d7c7ce6216e021078a87b5af4b35901f|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:a7e39e7e8e29e2aabcd56f9a6e02f093d7c7ce6216e021078a87b5af4b35901f",
+    "source_id": "github:cockroachdb/cockroach#6786",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f9/f9c20c80590bb65f3a1700deca342498e8a2aeb62c50baa19120568c8fe10aa1.json b/.cache/impact/classifications/f9/f9c20c80590bb65f3a1700deca342498e8a2aeb62c50baa19120568c8fe10aa1.json
new file mode 100644
index 0000000..dd14d2d
--- /dev/null
+++ b/.cache/impact/classifications/f9/f9c20c80590bb65f3a1700deca342498e8a2aeb62c50baa19120568c8fe10aa1.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#88660|sha256:ff3f6341c5983af29d82a6d72e883d4a7a45c210a1319d6a18c96754b2e0b96d|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "The TLP roachtest failed for a reason that is not a finding: an expired licence, a cluster that would not start, a library that would not load, a timeout, or no stated cause at all. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:ff3f6341c5983af29d82a6d72e883d4a7a45c210a1319d6a18c96754b2e0b96d",
+    "source_id": "github:cockroachdb/cockroach#88660",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/f9/f9d93ad3a1e28559327a9119fd7b5f63a583a4f1ffff1978ccc2942ad590573b.json b/.cache/impact/classifications/f9/f9d93ad3a1e28559327a9119fd7b5f63a583a4f1ffff1978ccc2942ad590573b.json
new file mode 100644
index 0000000..ae36210
--- /dev/null
+++ b/.cache/impact/classifications/f9/f9d93ad3a1e28559327a9119fd7b5f63a583a4f1ffff1978ccc2942ad590573b.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#97692|sha256:6aa4fc2ba36692eba64d239d40899fbab27c6b8edbe95dd3151c0496ec5951f5|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:6aa4fc2ba36692eba64d239d40899fbab27c6b8edbe95dd3151c0496ec5951f5",
+    "source_id": "github:cockroachdb/cockroach#97692",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/fa/fa22cbfc16a1428c54a15873d54f845d79887d8a2b190792aca06c3481ad5ecc.json b/.cache/impact/classifications/fa/fa22cbfc16a1428c54a15873d54f845d79887d8a2b190792aca06c3481ad5ecc.json
new file mode 100644
index 0000000..c92b7eb
--- /dev/null
+++ b/.cache/impact/classifications/fa/fa22cbfc16a1428c54a15873d54f845d79887d8a2b190792aca06c3481ad5ecc.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cmu-db/noisepage#1012|sha256:c03e79cfc86ac83b33bec5039d770e634da464f1c11eacb86d5d19575b48809b|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:c03e79cfc86ac83b33bec5039d770e634da464f1c11eacb86d5d19575b48809b",
+    "source_id": "github:cmu-db/noisepage#1012",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/fa/fa97191695ae316d864e0baefcbff533f63f4ce4c5ebc3d9c66b489cb732e4b7.json b/.cache/impact/classifications/fa/fa97191695ae316d864e0baefcbff533f63f4ce4c5ebc3d9c66b489cb732e4b7.json
new file mode 100644
index 0000000..51989c9
--- /dev/null
+++ b/.cache/impact/classifications/fa/fa97191695ae316d864e0baefcbff533f63f4ce4c5ebc3d9c66b489cb732e4b7.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:spiceai/spiceai#2189|sha256:67b24a41ce0489ba4093d4ba8e8e852f8c0942f214b450f84efe45ee6dbcec76|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "Names SQLancer, but as work to be done rather than as the source of a defect: adoption, CI plumbing, a feature SQLancer needs, a release checklist or a roadmap. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:67b24a41ce0489ba4093d4ba8e8e852f8c0942f214b450f84efe45ee6dbcec76",
+    "source_id": "github:spiceai/spiceai#2189",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/fa/fab39f76d5ed00d192fa7242a0b965ae488ee100de93b87a9594bfb10deb32a3.json b/.cache/impact/classifications/fa/fab39f76d5ed00d192fa7242a0b965ae488ee100de93b87a9594bfb10deb32a3.json
new file mode 100644
index 0000000..31b3c3b
--- /dev/null
+++ b/.cache/impact/classifications/fa/fab39f76d5ed00d192fa7242a0b965ae488ee100de93b87a9594bfb10deb32a3.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#47306|sha256:4db7612661a5d98eac50635e76274755c171c87fe4f5bdb65c4dca97b07a8229|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:4db7612661a5d98eac50635e76274755c171c87fe4f5bdb65c4dca97b07a8229",
+    "source_id": "github:cockroachdb/cockroach#47306",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/fa/faf134b33f05237010572e05f76fc7731acf0e31cabf9d28659c49f5095a1db1.json b/.cache/impact/classifications/fa/faf134b33f05237010572e05f76fc7731acf0e31cabf9d28659c49f5095a1db1.json
new file mode 100644
index 0000000..3c14aa6
--- /dev/null
+++ b/.cache/impact/classifications/fa/faf134b33f05237010572e05f76fc7731acf0e31cabf9d28659c49f5095a1db1.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#132631|sha256:2d02ead8db0fefd2f7cb9c04d712a98f422546f1e909f12e68556c3e87c383ad|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:2d02ead8db0fefd2f7cb9c04d712a98f422546f1e909f12e68556c3e87c383ad",
+    "source_id": "github:cockroachdb/cockroach#132631",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/fb/fbe0e665cd80a3f8572fe52bd9e724852ec7c2138377b6bfb48c0c92c2ff9cef.json b/.cache/impact/classifications/fb/fbe0e665cd80a3f8572fe52bd9e724852ec7c2138377b6bfb48c0c92c2ff9cef.json
new file mode 100644
index 0000000..9e6e7ad
--- /dev/null
+++ b/.cache/impact/classifications/fb/fbe0e665cd80a3f8572fe52bd9e724852ec7c2138377b6bfb48c0c92c2ff9cef.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:yugabyte/yugabyte-db#14866|sha256:e75b3d329a9323706f53734b17dc00159678f79f5df1303d1859203818d33f72|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:e75b3d329a9323706f53734b17dc00159678f79f5df1303d1859203818d33f72",
+    "source_id": "github:yugabyte/yugabyte-db#14866",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/fc/fc584d49f074025891e74e86a2b090fc9d6bfa6e193fdea696075aaf7640dd4a.json b/.cache/impact/classifications/fc/fc584d49f074025891e74e86a2b090fc9d6bfa6e193fdea696075aaf7640dd4a.json
new file mode 100644
index 0000000..49f3cd8
--- /dev/null
+++ b/.cache/impact/classifications/fc/fc584d49f074025891e74e86a2b090fc9d6bfa6e193fdea696075aaf7640dd4a.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#49381|sha256:66cad6605ce84438860bdc06925bb075377b96bb8fb9ead995d453d0da298fad|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:66cad6605ce84438860bdc06925bb075377b96bb8fb9ead995d453d0da298fad",
+    "source_id": "github:cockroachdb/cockroach#49381",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/fc/fc981ee0e706b1c037c2a60937283a16743a094508aabc75cdcf6173fd7fdf16.json b/.cache/impact/classifications/fc/fc981ee0e706b1c037c2a60937283a16743a094508aabc75cdcf6173fd7fdf16.json
new file mode 100644
index 0000000..350011d
--- /dev/null
+++ b/.cache/impact/classifications/fc/fc981ee0e706b1c037c2a60937283a16743a094508aabc75cdcf6173fd7fdf16.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:tikv/tikv#18100|sha256:fd8873e0ba4180fe9b56adeb5858ea135fb6eb1dbc146760ed5301d9e934347c|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:fd8873e0ba4180fe9b56adeb5858ea135fb6eb1dbc146760ed5301d9e934347c",
+    "source_id": "github:tikv/tikv#18100",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/fd/fd3366f350ca660cfd702c3636f2079584c3df0f60d6a003b4e00fc5e7e0a995.json b/.cache/impact/classifications/fd/fd3366f350ca660cfd702c3636f2079584c3df0f60d6a003b4e00fc5e7e0a995.json
new file mode 100644
index 0000000..c969753
--- /dev/null
+++ b/.cache/impact/classifications/fd/fd3366f350ca660cfd702c3636f2079584c3df0f60d6a003b4e00fc5e7e0a995.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#8242|sha256:336ee84edf360e0b8b42fcc1f05eec1f9de7f2cae22e5361e5c8d9de7d476806|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:336ee84edf360e0b8b42fcc1f05eec1f9de7f2cae22e5361e5c8d9de7d476806",
+    "source_id": "github:cockroachdb/cockroach#8242",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ff/ff0a85ba8fa6196dfc51b40b4af756940852239107113528fe20df7a2d4dadbe.json b/.cache/impact/classifications/ff/ff0a85ba8fa6196dfc51b40b4af756940852239107113528fe20df7a2d4dadbe.json
new file mode 100644
index 0000000..0ef751d
--- /dev/null
+++ b/.cache/impact/classifications/ff/ff0a85ba8fa6196dfc51b40b4af756940852239107113528fe20df7a2d4dadbe.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:pingcap/tidb#59336|sha256:7d53709f91dce5421d0f4ce2401869cba2096c3f0e98bd558e92937d06a66474|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:7d53709f91dce5421d0f4ce2401869cba2096c3f0e98bd558e92937d06a66474",
+    "source_id": "github:pingcap/tidb#59336",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/classifications/ff/fff00b9d80ac558922eb80c77532bc38b5e14c3b7944b1c39a03f4955ce41dcf.json b/.cache/impact/classifications/ff/fff00b9d80ac558922eb80c77532bc38b5e14c3b7944b1c39a03f4955ce41dcf.json
new file mode 100644
index 0000000..5c72c4b
--- /dev/null
+++ b/.cache/impact/classifications/ff/fff00b9d80ac558922eb80c77532bc38b5e14c3b7944b1c39a03f4955ce41dcf.json
@@ -0,0 +1,22 @@
+{
+  "key": "github:cockroachdb/cockroach#150885|sha256:0a6111bdb443da015ecd5ae92db9ef57a630d70b31df466aeaae252606b1b39a|bug-attribution-v1|impact-policy-v1|taxonomy-v2|claude-opus-5",
+  "stored_at": "2026-09-13T03:09:23Z",
+  "value": {
+    "classified_at": "2026-09-13T03:09:23Z",
+    "classifier_version": "bug-attribution-v1",
+    "model": "claude-opus-5",
+    "policy_version": "impact-policy-v1",
+    "result": {
+      "answer": "no",
+      "excerpts": [],
+      "extra": {
+        "finder": null,
+        "technique": null
+      },
+      "reason": "\"cert\" here is a TLS certificate -- --ca-cert, cert-password, a cert file or a certificate warning -- not Cardinality Estimation Restriction Testing. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+    },
+    "source_hash": "sha256:0a6111bdb443da015ecd5ae92db9ef57a630d70b31df466aeaae252606b1b39a",
+    "source_id": "github:cockroachdb/cockroach#150885",
+    "taxonomy_version": "taxonomy-v2"
+  }
+}
diff --git a/.cache/impact/derived/.gitkeep b/.cache/impact/derived/.gitkeep
new file mode 100644
index 0000000..e69de29
diff --git a/.cache/impact/papers-pdf/README.md b/.cache/impact/papers-pdf/README.md
new file mode 100644
index 0000000..2e9b255
--- /dev/null
+++ b/.cache/impact/papers-pdf/README.md
@@ -0,0 +1,23 @@
+# Hand-supplied paper PDFs
+
+Two thirds of the papers in this corpus are published by ACM, IEEE or Springer,
+all of which refuse automated requests. For those, the only evidence the
+pipeline can gather on its own is a citation index's two-sentence contexts.
+
+Dropping a PDF here lets the same extraction run over the paper itself. Name the
+file for the paper's identifier:
+
+- by DOI, with `/` replaced by `_` — `10.1145_3764583.pdf`
+- or by arXiv id — `2604.16373.pdf`
+- or, for a paper the indexes give neither, by its Semantic Scholar id
+  with an `s2_` prefix — `s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.pdf`
+
+The PDFs stay on your machine: everything in this folder is gitignored except
+this file and WANTED.md, so a publisher copy cannot be committed by arriving
+with an unexpected name. Nothing here is required: a missing PDF costs coverage
+for that paper, never correctness. Claims drawn from a supplied PDF quote it
+verbatim exactly as they would from a preprint, so they remain checkable by
+anyone holding the same paper.
+
+The routes that need no help: arXiv preprints, and PVLDB, which serves every
+paper freely from vldb.org.
diff --git a/.cache/impact/papers-pdf/WANTED.md b/.cache/impact/papers-pdf/WANTED.md
new file mode 100644
index 0000000..950c25c
--- /dev/null
+++ b/.cache/impact/papers-pdf/WANTED.md
@@ -0,0 +1,31 @@
+# Papers to download
+
+13 papers whose publisher will not serve a PDF to an automated client. Downloading one is entirely optional: a missing paper costs coverage for that paper, never correctness.
+
+Save each file into this directory under the **Save as** name — that is how the pipeline finds it — then re-run:
+
+```sh
+python3 -m tools.impact.run collect --only papers --full
+```
+
+Claims drawn from a supplied PDF quote it verbatim, exactly as they would from a preprint, so they stay checkable by anyone holding the same paper.
+
+Papers on arXiv, in PVLDB, or at a USENIX venue are not listed: the pipeline reads those itself.
+
+IEEE links go through the NUS library proxy, which is what makes them resolve; ACM and Springer links point straight at the PDF.
+
+| # | Paper | Download | Known so far | Save as |
+| ---: | --- | --- | --- | --- |
+| 1 | [A Formal Framework for Typing and Cast Semantics in SQL Engines](https://doi.org/10.1145/3834861) (2026) | [ACM PDF](https://libproxy1.nus.edu.sg/login?url=https://dl.acm.org/doi/pdf/10.1145/3834861) | cites only (3 sentences hint at more) | `10.1145_3834861.pdf` |
+| 2 | [MTKeras: An Automated Metamorphic Testing Platform](https://doi.org/10.1142/s021819402150039x) (2021) | [other PDF](https://libproxy1.nus.edu.sg/login?url=https://doi.org/10.1142/s021819402150039x) | cites only (1 sentences hint at more) | `10.1142_s021819402150039x.pdf` |
+| 3 | [A Framework for Systematic Analysis and Automated Detection of Dark Patterns on E-Commerce Websites](https://doi.org/10.4018/979-8-2600-0747-1.ch001) (2026) | [other PDF](https://libproxy1.nus.edu.sg/login?url=https://doi.org/10.4018/979-8-2600-0747-1.ch001) | cites only | `10.4018_979-8-2600-0747-1.ch001.pdf` |
+| 4 | [Evaluating ERD Models and RAID-Based Storage for Query Performance Optimization in Relational Databases](https://doi.org/10.35970/jinita.v7i1.2707) (2025) | [other PDF](https://libproxy1.nus.edu.sg/login?url=https://doi.org/10.35970/jinita.v7i1.2707) | cites only | `10.35970_jinita.v7i1.2707.pdf` |
+| 5 | [Language-Based Testing for Knowledge Graphs](https://www.semanticscholar.org/paper/5bef8601da9663add6a85713414f8c945cf97355) (2025) | [other PDF](https://libproxy1.nus.edu.sg/login?url=https://www.semanticscholar.org/paper/5bef8601da9663add6a85713414f8c945cf97355) | cites only | `s2_5bef8601da9663add6a85713414f8c945cf97355.pdf` |
+| 6 | [Real-World Scalability of PostgreSQL: Practical Techniques Use Case Study](https://doi.org/10.1109/icbds67396.2025.11377486) (2025) | [IEEE PDF](https://libproxy1.nus.edu.sg/login?url=https://doi.org/10.1109/icbds67396.2025.11377486) | cites only | `10.1109_icbds67396.2025.11377486.pdf` |
+| 7 | [Siloso: Finding Logic Bugs in RDBMS via Dialect-Adaptable Reference Engine Construction](https://doi.org/10.1145/3758316.3763253) (2025) | [ACM PDF](https://libproxy1.nus.edu.sg/login?url=https://dl.acm.org/doi/pdf/10.1145/3758316.3763253) | cites only | `10.1145_3758316.3763253.pdf` |
+| 8 | [Leopard: A General Test Suite for Isolation Level Verification](https://www.semanticscholar.org/paper/07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85) (2024) | [other PDF](https://libproxy1.nus.edu.sg/login?url=https://www.semanticscholar.org/paper/07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85) | cites only | `s2_07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85.pdf` |
+| 9 | [Emerging Aspects of Software Fault Localization](https://doi.org/10.1002/9781119880929.ch13) (2023) | [other PDF](https://libproxy1.nus.edu.sg/login?url=https://doi.org/10.1002/9781119880929.ch13) | cites only | `10.1002_9781119880929.ch13.pdf` |
+| 10 | [Siklus Hidup Pengembangan Sistem Basis Data Pada Sistem Informasi Buku Tamu di Badan Pusat Statistik Kabupaten Kediri Menggunakan MySQL](https://doi.org/10.32672/jnkti.v6i1.5830) (2023) | [other PDF](https://libproxy1.nus.edu.sg/login?url=https://doi.org/10.32672/jnkti.v6i1.5830) | cites only | `10.32672_jnkti.v6i1.5830.pdf` |
+| 11 | [Database System Development Life Cycle (DSDLC) on System Libraries for Data Manipulation Language (DML) Using SQL Server 2008](https://doi.org/10.35335/jurnalmantik.vol5.2021.1448.pp1065-1071) (2021) | [other PDF](https://libproxy1.nus.edu.sg/login?url=https://doi.org/10.35335/jurnalmantik.vol5.2021.1448.pp1065-1071) | cites only | `10.35335_jurnalmantik.vol5.2021.1448.pp1065-1071.pdf` |
+| 12 | [Differential Monitoring - Technical Report ⋆](https://www.semanticscholar.org/paper/b6c64dbe0130ef1bf6af12266b958a5d5396101f) (2021) | [other PDF](https://libproxy1.nus.edu.sg/login?url=https://www.semanticscholar.org/paper/b6c64dbe0130ef1bf6af12266b958a5d5396101f) | cites only | `s2_b6c64dbe0130ef1bf6af12266b958a5d5396101f.pdf` |
+| 13 | [Verifying Serializability Protocols With Version Order Recovery](https://doi.org/10.3929/ethz-b-000507577) (2021) | [other PDF](https://libproxy1.nus.edu.sg/login?url=https://doi.org/10.3929/ethz-b-000507577) | cites only | `10.3929_ethz-b-000507577.pdf` |
diff --git a/.cache/impact/state.json b/.cache/impact/state.json
new file mode 100644
index 0000000..f89962b
--- /dev/null
+++ b/.cache/impact/state.json
@@ -0,0 +1,24 @@
+{
+  "adoption": {
+    "last_full_reconciliation": "2026-09-13T06:26:24Z",
+    "last_successful_scan": "2026-09-13T07:18:13Z"
+  },
+  "github": {
+    "last_full_reconciliation": "2026-09-13T06:24:33Z",
+    "last_successful_scan": "2026-09-13T06:24:33Z"
+  },
+  "nus_test": {
+    "last_full_reconciliation": "2026-09-13T06:24:33Z",
+    "last_successful_scan": "2026-09-13T06:24:33Z"
+  },
+  "papers": {
+    "last_full_reconciliation": "2026-09-10T15:21:36Z",
+    "last_successful_scan": "2026-09-10T15:21:36Z"
+  },
+  "resources": {
+    "last_successful_scan": "2026-09-12T14:19:11Z"
+  },
+  "talks": {
+    "last_successful_scan": "2026-09-12T02:04:11Z"
+  }
+}
diff --git a/.cache/impact/talk-transcripts/6YGqFRTe2D0.json b/.cache/impact/talk-transcripts/6YGqFRTe2D0.json
new file mode 100644
index 0000000..bbae4d8
--- /dev/null
+++ b/.cache/impact/talk-transcripts/6YGqFRTe2D0.json
@@ -0,0 +1,21 @@
+{
+  "video_id": "6YGqFRTe2D0",
+  "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0",
+  "title": "[FUZZING'23] \"Three Colours of Fuzzing: Reflections and Open Challenges\" Keynote by Cristian Cadar",
+  "source": "youtube_transcript_panel",
+  "kind": "auto_captions",
+  "retrieved_at": "2026-09-11T00:35:00Z",
+  "total_segments": 456,
+  "total_characters": 53044,
+  "note": "Windows around candidate moments, read out of YouTube's transcript panel in a browser. Not the whole talk: only the segments a loose prefilter flagged, so the matcher decides on text a person can check.",
+  "segments": [
+    {
+      "start_ms": 295000,
+      "text": "as another example SQL lenser which is a fer for database Management Systems again it has found hundreds of bugs in popular database Management Systems like sqlite and postgress SQL"
+    },
+    {
+      "start_ms": 1509000,
+      "text": "so I have this quote here and I think that you know if Manuel would be here he would certainly blush uh it says one fing researcher of particular noce Manuel riger and I completely agree with that this is actually from the sqlite web page"
+    }
+  ]
+}
\ No newline at end of file
diff --git a/.cache/impact/talk-transcripts/BgC79Zt2fPs.json b/.cache/impact/talk-transcripts/BgC79Zt2fPs.json
new file mode 100644
index 0000000..a8b4e72
--- /dev/null
+++ b/.cache/impact/talk-transcripts/BgC79Zt2fPs.json
@@ -0,0 +1,57 @@
+{
+  "video_id": "BgC79Zt2fPs",
+  "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs",
+  "title": "Keynote 1: DuckDB Testing - Present and Future",
+  "source": "youtube_transcript_panel",
+  "kind": "auto_captions",
+  "retrieved_at": "2026-09-12T02:20:00Z",
+  "total_segments": 481,
+  "total_characters": 57721,
+  "note": "Windows around candidate moments, read out of YouTube's transcript panel in a browser. Not the whole talk: only the segments a loose prefilter flagged, so the matcher decides on text a person can check.",
+  "segments": [
+    {
+      "start_ms": 2135000,
+      "text": "thought we had like a pretty robust system right we had like um thousands of tests from various systems like sqlite postgres hundreds of our own tests and"
+    },
+    {
+      "start_ms": 2144000,
+      "text": "then this guy called dr rieger came along and he uh unleashed his creation upon us and started furiously opening"
+    },
+    {
+      "start_ms": 2154000,
+      "text": "bug reports so uh using sql answer manual found around 80 bugs inductive and those were"
+    },
+    {
+      "start_ms": 2162000,
+      "text": "bugs that were not found using the test suites of the other systems right so we ran the sql light tests they did not find these bugs we ran the postgres test"
+    },
+    {
+      "start_ms": 2171000,
+      "text": "they didn't find these bugs and it turns out this kind of thing where database systems are complex surprise surprise and each system has their own"
+    },
+    {
+      "start_ms": 2260000,
+      "text": "important not to just run one fuzzer but to run many different types of buzzers because these fuzzers they all they have like different domains so if you run you"
+    },
+    {
+      "start_ms": 2268000,
+      "text": "shouldn't just run sql lancer you should also run sql smith because sql smith will find bug sequel answer will not and sql answer will find bugs sql smith"
+    },
+    {
+      "start_ms": 2275000,
+      "text": "so uh it's like it's very powerful you should run as many as possible i think um also kind of a lesson we learned is that"
+    },
+    {
+      "start_ms": 2443000,
+      "text": "just file github issues and so the the fuzzer of docs robot was born and this is what runs now in our ci"
+    },
+    {
+      "start_ms": 2451000,
+      "text": "it currently runs sql answer and sql smith we also run oss fuzz but that's managed by google that's not by fuzzer ducks"
+    },
+    {
+      "start_ms": 2459000,
+      "text": "and what it does is if a bug is found it creates a reproducible test case based on the output of the fuzzer it does test case reduction and then it files an"
+    }
+  ]
+}
\ No newline at end of file
diff --git a/.cache/impact/talk-transcripts/CW4Ntdtp7lg.json b/.cache/impact/talk-transcripts/CW4Ntdtp7lg.json
new file mode 100644
index 0000000..00cdbce
--- /dev/null
+++ b/.cache/impact/talk-transcripts/CW4Ntdtp7lg.json
@@ -0,0 +1,37 @@
+{
+  "video_id": "CW4Ntdtp7lg",
+  "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg",
+  "title": "Fuzzing databases is difficult",
+  "source": "youtube_transcript_panel",
+  "kind": "auto_captions",
+  "retrieved_at": "2026-09-11T03:10:00Z",
+  "total_segments": 430,
+  "total_characters": 22738,
+  "note": "Windows around candidate moments, read out of YouTube's transcript panel in a browser. Not the whole talk: only the segments a loose prefilter flagged, so the matcher decides on text a person can check.",
+  "segments": [
+    {
+      "start_ms": 107180,
+      "text": "Uh, there are a few here, uh, a few of them probably already know, uh,"
+    },
+    {
+      "start_ms": 110940,
+      "text": "like, uh, SQL Answer, which was, uh, Pioneer to find the wrong results."
+    },
+    {
+      "start_ms": 115915,
+      "text": "Uh, there are others like FAL and WebFuzzer that are known to do"
+    },
+    {
+      "start_ms": 823675,
+      "text": "And there are things like running a query oracle."
+    },
+    {
+      "start_ms": 827265,
+      "text": "This was, SQLanswer actually started this a few years ago."
+    },
+    {
+      "start_ms": 831864,
+      "text": "We can do something like, for example, select count from a query with a predicate"
+    }
+  ]
+}
\ No newline at end of file
diff --git a/.cache/impact/talk-transcripts/L90MBb6NLBE.json b/.cache/impact/talk-transcripts/L90MBb6NLBE.json
new file mode 100644
index 0000000..e5ec324
--- /dev/null
+++ b/.cache/impact/talk-transcripts/L90MBb6NLBE.json
@@ -0,0 +1,11 @@
+{
+  "kind": "auto_captions",
+  "note": "Windows around the moments SQLancer is mentioned, cut from a transcript supplied by hand. Not the whole talk.",
+  "retrieved_at": "2026-09-11T16:20:15Z",
+  "segments": [],
+  "source": "supplied_transcript",
+  "total_characters": 38388,
+  "total_segments": 439,
+  "url": "https://www.youtube.com/watch?v=L90MBb6NLBE",
+  "video_id": "L90MBb6NLBE"
+}
diff --git a/.cache/impact/talk-transcripts/QRwxHGpWaUA.json b/.cache/impact/talk-transcripts/QRwxHGpWaUA.json
new file mode 100644
index 0000000..8c536d4
--- /dev/null
+++ b/.cache/impact/talk-transcripts/QRwxHGpWaUA.json
@@ -0,0 +1,21 @@
+{
+  "video_id": "QRwxHGpWaUA",
+  "url": "https://www.youtube.com/watch?v=QRwxHGpWaUA",
+  "title": "The Art of Database Testing by Alperen Keles | DC Systems 011",
+  "source": "youtube_transcript_panel",
+  "kind": "auto_captions",
+  "retrieved_at": "2026-09-11T00:20:00Z",
+  "total_segments": 371,
+  "total_characters": 32044,
+  "note": "Windows around candidate moments, read out of YouTube's transcript panel in a browser. Not the whole talk: only the segments a loose prefilter flagged, so the matcher decides on text a person can check.",
+  "segments": [
+    {
+      "start_ms": 205760,
+      "text": "So the first one is called pivoted query synthesis and the idea here is containment."
+    },
+    {
+      "start_ms": 2087760,
+      "text": "I think uh in SQL answer when running you pick uh you pick the oracle so they are for this spe more complex heristics based on the oracle I I don't think I have seen that in the codebase but maybe they have it"
+    }
+  ]
+}
\ No newline at end of file
diff --git a/.cache/impact/talk-transcripts/README.md b/.cache/impact/talk-transcripts/README.md
new file mode 100644
index 0000000..ad1f375
--- /dev/null
+++ b/.cache/impact/talk-transcripts/README.md
@@ -0,0 +1,39 @@
+# Captured talk transcripts
+
+The input behind the caption mentions in `_data/impact/talks.json`. Committed,
+small and few on purpose: keeping them makes that file reproducible, and keeping
+only windows around the moments SQLancer comes up avoids storing a transcription
+of somebody else's whole talk.
+
+## Getting one
+
+YouTube serves its caption tracks only to its own player, and for some talks it
+will not even fill its own transcript panel. So this step is done by a person.
+Any of these works — the importer takes whichever shape the text arrives in:
+
+1. **The transcript panel.** Open the talk, expand the description, click
+   *Show transcript*, then use the panel's ⋮ menu (*Toggle timestamps* off is
+   not needed) — select the transcript and copy it. Save it to a file.
+2. **A caption file.** `yt-dlp --write-auto-subs --sub-lang en --skip-download
+   --sub-format vtt <url>` leaves a `.vtt` beside itself. A `.srt` works too.
+3. **The JSON shape this directory holds**, if you already have segments.
+
+Then:
+
+    python3 -m tools.impact.talks import <file> <talk url>
+    python3 -m tools.impact.run collect --only talks
+
+The first command prints every mention it found, with its timestamp and how the
+transcriber heard the name, and writes the windows here. The second folds them
+into the dataset. Rolling repetition in auto-captions — each cue restating the
+tail of the one before — is dropped on the way in.
+
+## What the text is worth
+
+A machine transcription, wrong often enough to matter: it renders "SQLancer" as
+"SQL answer", "SQL lenser" and "SQLanswer" in the talks collected so far. The
+records built from it carry the misheard form and a link to the second it was
+said, and the page says plainly that the words are the transcriber's rather than
+the speaker's. Where a mention is on a slide and never spoken, no transcript
+reaches it at all — that is what the frames in `assets/images/impact/talks/` are
+for.
diff --git a/.cache/impact/talk-transcripts/V_qzqY1bb7I.json b/.cache/impact/talk-transcripts/V_qzqY1bb7I.json
new file mode 100644
index 0000000..299b41f
--- /dev/null
+++ b/.cache/impact/talk-transcripts/V_qzqY1bb7I.json
@@ -0,0 +1,32 @@
+{
+  "kind": "auto_captions",
+  "note": "Windows around the moments SQLancer is mentioned, cut from a transcript supplied by hand. Not the whole talk.",
+  "retrieved_at": "2026-09-11T16:23:20Z",
+  "segments": [
+    {
+      "start_ms": 2823000,
+      "text": "was great. Uh in other words, we were finding the bugs ourselves before the external bug"
+    },
+    {
+      "start_ms": 2829000,
+      "text": "finders were. And that was great. And then a few years later, um, Manual"
+    },
+    {
+      "start_ms": 2836000,
+      "text": "Rigger came up with this idea of we, you know, the the original fuzzers"
+    },
+    {
+      "start_ms": 2843000,
+      "text": "were just looking for memory errors or searching faults or something like that. He came up with the idea we can we can"
+    },
+    {
+      "start_ms": 2848000,
+      "text": "do fuzzing ideas to test for inconsistencies in SQL. Like if you have"
+    }
+  ],
+  "source": "supplied_transcript",
+  "total_characters": 46496,
+  "total_segments": 515,
+  "url": "https://www.youtube.com/watch?v=V_qzqY1bb7I",
+  "video_id": "V_qzqY1bb7I"
+}
diff --git a/.cache/impact/talk-transcripts/wHo-VtzTHx0.json b/.cache/impact/talk-transcripts/wHo-VtzTHx0.json
new file mode 100644
index 0000000..0f04de3
--- /dev/null
+++ b/.cache/impact/talk-transcripts/wHo-VtzTHx0.json
@@ -0,0 +1,41 @@
+{
+  "video_id": "wHo-VtzTHx0",
+  "url": "https://www.youtube.com/watch?v=wHo-VtzTHx0",
+  "title": "CockroachDB's Query Optimizer (Rebecca Taft, Cockroach Labs)",
+  "source": "youtube_transcript_panel",
+  "kind": "auto_captions",
+  "retrieved_at": "2026-09-12T02:00:00Z",
+  "total_segments": 626,
+  "total_characters": 60535,
+  "note": "Windows around candidate moments, read out of YouTube's transcript panel in a browser. Not the whole talk: only the segments a loose prefilter flagged, so the matcher decides on text a person can check.",
+  "segments": [
+    {
+      "start_ms": 3488960,
+      "text": "like something like sql smith yeah yeah we have we have sql smith so but you know that that is more kind"
+    },
+    {
+      "start_ms": 3494960,
+      "text": "of just catching errors like if if it's gonna cause an internal error or crash"
+    },
+    {
+      "start_ms": 3500720,
+      "text": "or something like that lasting correctness but um yeah actually uh manuel rigger uh he's the guy that"
+    },
+    {
+      "start_ms": 3507280,
+      "text": "was doing more kind of logically yeah exactly he he did a bunch"
+    },
+    {
+      "start_ms": 3513359,
+      "text": "of experiments with cockroaches and he opened a bunch of github issues for us which is which is pretty great sql lanza is awesome yeah i'm trying to"
+    },
+    {
+      "start_ms": 3519920,
+      "text": "get trying to get it running in our system um all right so then it's the"
+    },
+    {
+      "start_ms": 3526079,
+      "text": "in general what like what's the complexity of the queries that you're seeing i i understand that like that might be"
+    }
+  ]
+}
\ No newline at end of file
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
new file mode 100644
index 0000000..519057a
--- /dev/null
+++ b/.github/workflows/ci.yml
@@ -0,0 +1,40 @@
+name: CI
+
+# Runs on every push and pull request: validates the impact dataset against its
+# schemas, checks the derived statistics and charts are in sync with the
+# records, and builds the site.
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+
+permissions:
+  contents: read
+
+jobs:
+  impact-data:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+      - run: pip install jsonschema
+      - name: Validate the impact dataset
+        run: python -m tools.impact.run validate
+      - name: Check statistics and charts match the records
+        # The tests fail if stats.json or any chart is stale, so a data change
+        # cannot be merged without its derived outputs.
+        run: python -m unittest discover -s tools/impact/tests -t . -v
+
+  site:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: ruby/setup-ruby@v1
+        with:
+          ruby-version: "3.2"
+          bundler-cache: true
+      - name: Build the site
+        run: bundle exec jekyll build --trace
diff --git a/.github/workflows/impact.yml b/.github/workflows/impact.yml
new file mode 100644
index 0000000..909aa89
--- /dev/null
+++ b/.github/workflows/impact.yml
@@ -0,0 +1,169 @@
+name: Impact data
+
+# Collects new impact evidence once a week and proposes it as a pull request.
+#
+# Nothing here writes to the default branch. The job restores its caches,
+# collects only what changed since the last successful scan, validates the
+# result against the JSON Schemas, rebuilds the derived statistics and charts,
+# checks the site still builds, and opens or updates a pull request -- but only
+# when the authoritative records actually changed. A run that merely warms the
+# cache exits without touching anything a human would have to review.
+
+on:
+  schedule:
+    # Weekly incremental run, Monday 04:17 UTC.
+    - cron: "17 4 * * 1"
+    # Monthly full reconciliation, 1st of the month. Deterministic discovery is
+    # repeated across every source; cached evidence and cached classifications
+    # are still reused, so this is far cheaper than it sounds.
+    - cron: "43 3 1 * *"
+  workflow_dispatch:
+    inputs:
+      full:
+        description: "Run a full reconciliation instead of an incremental scan"
+        type: boolean
+        default: false
+      only:
+        description: "Restrict to these sources (space separated)"
+        type: string
+        default: ""
+
+permissions:
+  contents: write
+  pull-requests: write
+
+concurrency:
+  group: impact-data
+  cancel-in-progress: false
+
+jobs:
+  collect:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Install dependencies
+        run: pip install -r requirements.txt
+
+      - name: Restore the bulky, disposable part of the cache
+        # Committed cache files under .cache/impact carry the collector state
+        # that is worth reproducing; this restores the rest (fetched bodies)
+        # so a run does not re-download unchanged pages. The pipeline is
+        # correct with this cache empty -- it is only ever an optimisation.
+        uses: actions/cache@v4
+        with:
+          path: |
+            .cache/impact/http
+            .cache/impact/papers
+            .cache/impact/artifacts
+          key: impact-http-${{ github.run_id }}
+          restore-keys: |
+            impact-http-
+
+      - name: Decide the run mode
+        id: mode
+        run: |
+          if [ "${{ github.event.schedule }}" = "43 3 1 * *" ] || \
+             [ "${{ inputs.full }}" = "true" ]; then
+            echo "full=--full" >> "$GITHUB_OUTPUT"
+            echo "label=full reconciliation" >> "$GITHUB_OUTPUT"
+          else
+            echo "full=" >> "$GITHUB_OUTPUT"
+            echo "label=incremental scan" >> "$GITHUB_OUTPUT"
+          fi
+          if [ -n "${{ inputs.only }}" ]; then
+            echo "only=--only ${{ inputs.only }}" >> "$GITHUB_OUTPUT"
+          else
+            echo "only=" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Collect
+        env:
+          # Deterministic discovery. The default token is enough for the REST
+          # API; code search needs it too.
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          # Optional: lifts the Semantic Scholar rate limit considerably.
+          SEMANTIC_SCHOLAR_API_KEY: ${{ secrets.SEMANTIC_SCHOLAR_API_KEY }}
+          # Optional: without it, ambiguous candidates are reported as needing
+          # human judgement instead of being classified. The run still succeeds.
+          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
+          IMPACT_CONTACT_EMAIL: ${{ secrets.IMPACT_CONTACT_EMAIL }}
+        run: |
+          python -m tools.impact.run collect \
+            ${{ steps.mode.outputs.full }} ${{ steps.mode.outputs.only }} \
+            --report .cache/impact/last-run.json
+
+      - name: Validate and rebuild derived outputs
+        run: python -m tools.impact.run build
+
+      - name: Run the test suite
+        run: python -m unittest discover -s tools/impact/tests -t . -v
+
+      - name: Decide whether there is anything to review
+        id: changes
+        run: |
+          python - <<'PY' >> "$GITHUB_OUTPUT"
+          import json, pathlib
+          from tools.impact.pr import has_meaningful_changes
+          path = pathlib.Path(".cache/impact/last-run.json")
+          report = json.loads(path.read_text()) if path.exists() else {}
+          print(f"meaningful={'true' if has_meaningful_changes(report) else 'false'}")
+          PY
+          if git diff --quiet -- _data/impact _includes/impact; then
+            echo "dirty=false" >> "$GITHUB_OUTPUT"
+          else
+            echo "dirty=true" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Set up Ruby
+        if: steps.changes.outputs.dirty == 'true'
+        uses: ruby/setup-ruby@v1
+        with:
+          ruby-version: "3.2"
+          bundler-cache: true
+
+      - name: Build the website
+        if: steps.changes.outputs.dirty == 'true'
+        run: bundle exec jekyll build --trace
+
+      - name: Render the pull-request body
+        if: steps.changes.outputs.dirty == 'true'
+        run: |
+          python -m tools.impact.run report \
+            --input .cache/impact/last-run.json \
+            --output .cache/impact/pr-body.md
+          cat .cache/impact/pr-body.md
+
+      - name: Open or update the pull request
+        if: steps.changes.outputs.dirty == 'true'
+        uses: peter-evans/create-pull-request@v6
+        with:
+          branch: impact/automated-update
+          base: main
+          title: "Impact data update (${{ steps.mode.outputs.label }})"
+          body-path: .cache/impact/pr-body.md
+          commit-message: |
+            Update impact data (${{ steps.mode.outputs.label }})
+
+            Automated collection run. Every new claim carries the primary
+            source it rests on; see the pull-request body for the evidence.
+          labels: |
+            impact-data
+            automated
+          add-paths: |
+            _data/impact/**
+            _includes/impact/**
+            .cache/impact/state.json
+            .cache/impact/derived/**
+            .cache/impact/classifications/**
+          delete-branch: true
+
+      - name: Report a no-op run
+        if: steps.changes.outputs.dirty != 'true'
+        run: |
+          echo "No authoritative data changed; no pull request opened." \
+            >> "$GITHUB_STEP_SUMMARY"
diff --git a/.gitignore b/.gitignore
index 2038447..cdf24ae 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,4 +1,34 @@
 _site
 .sass-cache
 .jekyll-metadata
-Gemfile.lock
\ No newline at end of file
+Gemfile.lock
+__pycache__/
+*.pyc
+
+# Impact pipeline caches. The bulky, disposable layers -- fetched HTTP bodies,
+# GitHub payloads, scholarly API responses -- are re-fetchable and restored from
+# the GitHub Actions cache instead. The compact layers below them are committed
+# because they are the collector's reproducible state: what has been scanned,
+# what was deterministically extracted, and every classification answer.
+.cache/impact/http/
+.cache/impact/github/
+.cache/impact/papers/
+.cache/impact/artifacts/
+.cache/impact/last-run.json
+.cache/impact/pr-body.md
+.jekyll-cache
+
+# Hand-supplied paper PDFs. These stay on the machine that downloaded them:
+# they are publisher copies, often licensed to one reader, and the pipeline
+# only ever needs them locally. Everything in the folder is ignored except the
+# two files that explain it, so a PDF cannot be committed by arriving with an
+# unexpected name -- an uppercase .PDF, which the loader also reads, or a
+# subfolder of a bulk download.
+.cache/impact/papers-pdf/*
+!.cache/impact/papers-pdf/README.md
+!.cache/impact/papers-pdf/WANTED.md
+
+# Extracted paper text, cached locally. The dossiers in _data/papers keep the
+# mentions and their context; the whole text of a publisher's PDF does not
+# belong in the repository.
+.cache/impact/papers-text/
diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..1252cdb
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,37 @@
+# Convenience targets for the impact pipeline. See tools/impact/README.md.
+
+.PHONY: help validate stats plots build test collect collect-full serve
+
+help:
+	@echo "validate     - validate _data/impact against the JSON Schemas"
+	@echo "stats        - regenerate _data/impact/stats.json from the records"
+	@echo "plots        - regenerate the charts in _includes/impact/plots"
+	@echo "build        - stats + plots + validate"
+	@echo "test         - run the impact test suite"
+	@echo "collect      - incremental collection run (writes _data/impact)"
+	@echo "collect-full - periodic full reconciliation"
+	@echo "serve        - build and serve the website locally"
+
+validate:
+	python3 -m tools.impact.run validate
+
+stats:
+	python3 -m tools.impact.run stats
+
+plots:
+	python3 -m tools.impact.run plots
+
+build:
+	python3 -m tools.impact.run build
+
+test:
+	python3 -m unittest discover -s tools/impact/tests -t . -v
+
+collect:
+	python3 -m tools.impact.run collect --report .cache/impact/last-run.json
+
+collect-full:
+	python3 -m tools.impact.run collect --full --report .cache/impact/last-run.json
+
+serve:
+	bundle exec jekyll serve
diff --git a/README.md b/README.md
index c58cc88..f3f5e1c 100644
--- a/README.md
+++ b/README.md
@@ -1,27 +1,50 @@
-# Minimal Mistakes remote theme starter
-
-Click [**Use this template**](https://github.com/mmistakes/mm-github-pages-starter/generate) button above for the quickest method of getting started with the [Minimal Mistakes Jekyll theme](https://github.com/mmistakes/minimal-mistakes).
-
-Contains basic configuration to get you a site with:
-
-- Sample posts.
-- Sample top navigation.
-- Sample author sidebar with social links.
-- Sample footer links.
-- Paginated home page.
-- Archive pages for posts grouped by year, category, and tag.
-- Sample about page.
-- Sample 404 page.
-- Site wide search.
-
-Replace sample content with your own and [configure as necessary](https://mmistakes.github.io/minimal-mistakes/docs/configuration/).
-
----
-
-## Troubleshooting
-
-If you have a question about using Jekyll, start a discussion on the [Jekyll Forum](https://talk.jekyllrb.com/) or [StackOverflow](https://stackoverflow.com/questions/tagged/jekyll). Other resources:
-
-- [Ruby 101](https://jekyllrb.com/docs/ruby-101/)
-- [Setting up a Jekyll site with GitHub Pages](https://jekyllrb.com/docs/github-pages/)
-- [Configuring GitHub Metadata](https://github.com/jekyll/github-metadata/blob/master/docs/configuration.md#configuration) to work properly when developing locally and avoid `No GitHub API authentication could be found. Some fields may be missing or have incorrect data.` warnings.
+# sqlancer.github.io
+
+The website for [SQLancer](https://github.com/sqlancer/sqlancer), built with
+Jekyll and the [Minimal Mistakes](https://github.com/mmistakes/minimal-mistakes)
+remote theme, and published through GitHub Pages.
+
+## Running it locally
+
+```sh
+bundle install
+bundle exec jekyll serve
+```
+
+## The impact section
+
+[`/impact/`](_pages/impact.html) presents what SQLancer has found and what has
+been built on it: bugs by database system and over time, which database system
+projects use SQLancer, and the research that reuses, extends or compares against
+it. Every figure on that page -- and the statistics on the homepage -- is
+computed from structured, evidence-backed records rather than written by hand.
+
+- **The records** live in [`_data/impact/`](_data/impact/), one JSON file per
+  kind of claim, each validated against a schema in
+  [`schemas/impact/`](schemas/impact/). Every record carries the primary source
+  that justifies it.
+- **The policy** that governs what is admitted is in
+  [`_data/impact/policy.json`](_data/impact/policy.json) and is rendered on the
+  impact page, so the rules the collectors enforce and the rules the page
+  describes cannot drift apart.
+- **The pipeline** that proposes changes to the records is in
+  [`tools/impact/`](tools/impact/) -- see
+  [its README](tools/impact/README.md) for how it works, how to run it, and what
+  it enforces. It runs weekly through
+  [`.github/workflows/impact.yml`](.github/workflows/impact.yml) and opens a pull
+  request; it never writes to `main`.
+
+```sh
+make validate   # schemas and cross-file integrity
+make build      # regenerate derived statistics and charts, then validate
+make test       # the impact test suite
+```
+
+`stats.json` and the charts are generated: CI fails if they do not match the
+records they are derived from.
+
+## Corrections
+
+If a bug, paper, integration or resource is missing or wrongly attributed,
+please [open an issue](https://github.com/sqlancer/sqlancer.github.io/issues/new).
+The dataset is meant to be corrected in public.
diff --git a/_config.yml b/_config.yml
index 79d9c92..009546c 100644
--- a/_config.yml
+++ b/_config.yml
@@ -27,24 +27,26 @@ markdown: kramdown
 remote_theme: mmistakes/minimal-mistakes
 # Outputting
 permalink: /:categories/:title/
-paginate: 5 # amount of posts to show
-paginate_path: /page:num/
+# No paginated index page: / is the splash page and posts are listed at /posts/.
 timezone: # https://en.wikipedia.org/wiki/List_of_tz_database_time_zones
 
 include:
   - _pages
 
-# Exclude from processing.
-# The following items will not be processed, by default. Create a custom list
-# to override the default setting.
-# exclude:
-#   - Gemfile
-#   - Gemfile.lock
-#   - node_modules
-#   - vendor/bundle/
-#   - vendor/cache/
-#   - vendor/gems/
-#   - vendor/ruby/
+# Exclude from processing. The impact pipeline and its schemas live in the
+# repository but are not part of the published site.
+exclude:
+  - Gemfile
+  - Gemfile.lock
+  - node_modules
+  - vendor/bundle/
+  - vendor/cache/
+  - vendor/gems/
+  - vendor/ruby/
+  - tools/
+  - schemas/
+  - requirements.txt
+  - Makefile
 
 # Plugins (previously gems:)
 plugins:
diff --git a/_data/impact/adoption.json b/_data/impact/adoption.json
new file mode 100644
index 0000000..7aa27f3
--- /dev/null
+++ b/_data/impact/adoption.json
@@ -0,0 +1,1883 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "adoption": [
+    {
+      "id": "adoption:bharatdbms:planned_adoption",
+      "dbms": "bharatdbms",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "BharatDBPG, owner of the BharatDBMS project, proposed adopting SQLancer; the issue is open.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/1712",
+          "source_type": "github_issue",
+          "excerpt": "Using sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the BharatDBMS tracker proposing SQLancer, opened by BharatDBPG (owner).",
+          "content_sha256": "sha256:e2ac652368926a1660d704792b573d4ea392c9667d0ff02180659bd165156c4b",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        },
+        {
+          "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2605",
+          "source_type": "github_issue",
+          "excerpt": "running sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the BharatDBMS tracker proposing SQLancer, opened by BharatDBPG (owner).",
+          "content_sha256": "sha256:efc626f13f54267a8ec3546511e7e6a4f2f2fb558cc6d0a54d21541a582d1272",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-06T16:54:23Z",
+        "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/1712",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/BharatDBPG/BharatDBMS-PG"
+    },
+    {
+      "id": "adoption:citus:official_testing",
+      "dbms": "citus",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The Citus project maintains its own SQLancer fork, last updated 2020-07-19.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/citusdata/citus/blob/f1728591d33bd13d113d2a8d66bb99c377c31116/src/test/regress/expected/issue_8468.out",
+          "source_type": "github_code",
+          "excerpt": "--\n-- ISSUE_8468\n--\n-- Test for GitHub issue #8468: TLP test fail on sqlancer due to\n-- num_nulls interpreted differently on workers.",
+          "excerpt_is_verbatim": true,
+          "note": "src/test/regress/expected/issue_8468.out in the Citus repository states that the project uses SQLancer.",
+          "content_sha256": "sha256:709a56df51ae28f8cc78b0c2991873305aa21a3d1d4d5ab3fa7240fa3d53e149",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/citusdata/citus/blob/f1728591d33bd13d113d2a8d66bb99c377c31116/src/test/regress/sql/issue_8468.sql",
+          "source_type": "github_code",
+          "excerpt": "--\n-- ISSUE_8468\n--\n-- Test for GitHub issue #8468: TLP test fail on sqlancer due to\n-- num_nulls interpreted differently on workers.",
+          "excerpt_is_verbatim": true,
+          "note": "src/test/regress/sql/issue_8468.sql in the Citus repository states that the project uses SQLancer.",
+          "content_sha256": "sha256:699578d3a41999e3c6466b124e00fa254fe43509afbc071298a1d2eff1a4b6fa",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/citusdata/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "citusdata/sqlancer is a fork of sqlancer/sqlancer under the Citus project's own GitHub organisation.",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:08:58Z",
+        "last_verified": "2026-09-13T05:57:42Z",
+        "source_url": "https://github.com/citusdata/citus/blob/f1728591d33bd13d113d2a8d66bb99c377c31116/src/test/regress/expected/issue_8468.out",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/citusdata/citus"
+    },
+    {
+      "id": "adoption:clickhouse:official_ci",
+      "dbms": "clickhouse",
+      "relationship": "official_ci",
+      "finder": "sqlancer",
+      "summary": "The ClickHouse project runs SQLancer from its own repository (.github/workflows/nightly_sqlancer.yml).",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/ClickHouse/ClickHouse/blob/5f46b053c26229829ac704bc2f54045ffb6e81a1/.github/workflows/nightly_sqlancer.yml",
+          "source_type": "github_workflow",
+          "excerpt": "PYTHONUNBUFFERED=1 python3 -m praktika run 'Dockers Build (amd)' --workflow \"NightlySQLancer\" --ci --timestamp",
+          "excerpt_is_verbatim": true,
+          "note": ".github/workflows/nightly_sqlancer.yml in the ClickHouse repository invokes SQLancer.",
+          "content_sha256": "sha256:445c23419010837ca4a08fd3c9e2c31639fa210c33c23af523281ce2f26acfed",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:08:58Z",
+        "last_verified": "2026-09-13T05:57:42Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/blob/5f46b053c26229829ac704bc2f54045ffb6e81a1/.github/workflows/nightly_sqlancer.yml",
+        "source_type": "github_workflow"
+      },
+      "project_repository": "https://github.com/ClickHouse/ClickHouse"
+    },
+    {
+      "id": "adoption:clickhouse:official_testing",
+      "dbms": "clickhouse",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The ClickHouse project maintains its own SQLancer fork, 209 commit(s) ahead of upstream, last updated 2026-09-02.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/ClickHouse/ClickHouse/blob/5f46b053c26229829ac704bc2f54045ffb6e81a1/ci/jobs/sqlancer_pp_job.sh",
+          "source_type": "github_code",
+          "excerpt": "# `clickhouse/sqlancer-test` image (ci/docker/sqlancer-test/Dockerfile) installs",
+          "excerpt_is_verbatim": true,
+          "note": "ci/jobs/sqlancer_pp_job.sh in the ClickHouse repository invokes SQLancer.",
+          "content_sha256": "sha256:b0b245da0f0ca7b3e853e6ffe790d7f23ce7a4d48f384822691d4604118a6e4f",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/ClickHouse/ClickHouse/blob/5f46b053c26229829ac704bc2f54045ffb6e81a1/ci/jobs/sqlancer_job.sh",
+          "source_type": "github_code",
+          "excerpt": "# `ci/docker/sqlancer-test/Dockerfile`), but that image is only rebuilt when its",
+          "excerpt_is_verbatim": true,
+          "note": "ci/jobs/sqlancer_job.sh in the ClickHouse repository invokes SQLancer.",
+          "content_sha256": "sha256:3f17c6f358fb65bb3f74e50d3a240d394151ab1ad5e87a0faabdf1dae4fed7e8",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/ClickHouse/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "ClickHouse/sqlancer is a fork of sqlancer/sqlancer under the ClickHouse project's own GitHub organisation, 209 commit(s) ahead of upstream.",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:08:58Z",
+        "last_verified": "2026-09-13T05:57:42Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/blob/5f46b053c26229829ac704bc2f54045ffb6e81a1/ci/jobs/sqlancer_pp_job.sh",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/ClickHouse/ClickHouse"
+    },
+    {
+      "id": "adoption:clickhouse:planned_adoption",
+      "dbms": "clickhouse",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "alexey-milovidov, member of the ClickHouse project, proposed adopting SQLancer; the issue is open.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/ClickHouse/ClickHouse/issues/115033",
+          "source_type": "github_issue",
+          "excerpt": "run has no sanitizer reports and no logical errors (\"exception\" class); the SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the ClickHouse tracker proposing SQLancer, opened by alexey-milovidov (member).",
+          "content_sha256": "sha256:86360362055b123fc0dc3f0399d14f2c8d2b9e9a0fd6f2d3485615b088c20455",
+          "retrieved_at": "2026-09-06T16:25:52Z",
+          "first_seen": "2026-09-06T16:25:52Z",
+          "last_verified": "2026-09-06T16:25:52Z"
+        },
+        {
+          "source_url": "https://github.com/ClickHouse/ClickHouse/issues/43101",
+          "source_type": "github_issue",
+          "excerpt": "run queries with SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the ClickHouse tracker proposing SQLancer, opened by qoega (member).",
+          "content_sha256": "sha256:70d651df078e5111357531755d518ff582858098ca4777119ad11d8d9e116e93",
+          "retrieved_at": "2026-09-06T16:25:52Z",
+          "first_seen": "2026-09-06T16:25:52Z",
+          "last_verified": "2026-09-06T16:25:52Z"
+        },
+        {
+          "source_url": "https://github.com/ClickHouse/ClickHouse/issues/17623",
+          "source_type": "github_issue",
+          "excerpt": "Integrate SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the ClickHouse tracker proposing SQLancer, opened by alexey-milovidov (member).",
+          "content_sha256": "sha256:222637ee24f2162d625adbf7dfa99d66483e0f3b054f54092f67efc7105975de",
+          "retrieved_at": "2026-09-06T16:25:52Z",
+          "first_seen": "2026-09-06T16:25:52Z",
+          "last_verified": "2026-09-06T16:25:52Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:25:52Z",
+        "last_verified": "2026-09-06T16:25:52Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/115033",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/ClickHouse/ClickHouse"
+    },
+    {
+      "id": "adoption:cloudberry:planned_adoption",
+      "dbms": "cloudberry",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "my-ship-it, contributor of the Apache Cloudberry project, proposed adopting SQLancer; the issue is open.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/apache/cloudberry/issues/1948",
+          "source_type": "github_issue",
+          "excerpt": "running SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Cloudberry tracker proposing SQLancer, opened by my-ship-it (contributor).",
+          "content_sha256": "sha256:b03c252301ac3bd8fe41b4f02d784ad3da9e529a99eb8e33e7d212f99d791822",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/cloudberry/issues/1950",
+          "source_type": "github_issue",
+          "excerpt": "running SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Cloudberry tracker proposing SQLancer, opened by my-ship-it (contributor).",
+          "content_sha256": "sha256:4edf946c645b77268cae82dfd8f11600827318c316f82b278d18389c944a428b",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/cloudberry/issues/1949",
+          "source_type": "github_issue",
+          "excerpt": "running SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Cloudberry tracker proposing SQLancer, opened by my-ship-it (contributor).",
+          "content_sha256": "sha256:be2b373953800df3b3c8bb6e123678e8afd6a7e012ed80920ce85f8b9cb804c0",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:44:53Z",
+        "last_verified": "2026-09-06T16:44:53Z",
+        "source_url": "https://github.com/apache/cloudberry/issues/1948",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/apache/cloudberry"
+    },
+    {
+      "id": "adoption:cnosdb:official_testing",
+      "dbms": "cnosdb",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The CnosDB project maintains its own SQLancer fork, 3 commit(s) ahead of upstream, last updated 2024-06-28.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/cnosdb/cnosdb/blob/c7609432edfdefd5a442269af42a99d61d5e87c1/README_CN.md",
+          "source_type": "github_code",
+          "excerpt": "SQLancer](https://github.com/sqlancer/sqlancer)作为支持。",
+          "excerpt_is_verbatim": true,
+          "note": "README_CN.md in the CnosDB repository.",
+          "content_sha256": "sha256:c67eb660217788915b5563473e837cba72bc624fc3f7aaa52036a765a1455e60",
+          "retrieved_at": "2026-09-13T07:17:40Z",
+          "first_seen": "2026-09-13T07:17:40Z",
+          "last_verified": "2026-09-13T07:17:40Z"
+        },
+        {
+          "source_url": "https://github.com/cnosdb/cnosdb/blob/c7609432edfdefd5a442269af42a99d61d5e87c1/README.md",
+          "source_type": "github_code",
+          "excerpt": "CnosDB 2.0's bug detection is powered by [SQLancer](https://github.com/sqlancer/sqlancer).",
+          "excerpt_is_verbatim": true,
+          "note": "README.md in the CnosDB repository.",
+          "content_sha256": "sha256:ff1e508aa28b9ed04dce2a2af6bb394df43ec210345f1622a6b2316c200b48ab",
+          "retrieved_at": "2026-09-13T07:17:40Z",
+          "first_seen": "2026-09-13T07:17:40Z",
+          "last_verified": "2026-09-13T07:17:40Z"
+        },
+        {
+          "source_url": "https://github.com/cnosdb/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "cnosdb/sqlancer is a fork of sqlancer/sqlancer under the CnosDB project's own GitHub organisation, 3 commit(s) ahead of upstream.",
+          "retrieved_at": "2026-09-13T07:17:40Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-13T07:17:40Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T15:17:57Z",
+        "last_verified": "2026-09-13T07:17:40Z",
+        "source_url": "https://github.com/cnosdb/cnosdb/blob/c7609432edfdefd5a442269af42a99d61d5e87c1/README_CN.md",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/cnosdb/cnosdb",
+      "classifier": {
+        "method": "llm_classification",
+        "classifier_version": "adoption-v1",
+        "policy_version": "impact-policy-v1",
+        "taxonomy_version": "taxonomy-v2",
+        "evidence_sha256": "sha256:0f27bfeb42face52bdf31ebb42c2ff166c3ff148d090cbd2dcae7c2e88e94071",
+        "classified_at": "2026-09-13T07:17:41Z",
+        "model": "claude-opus-5",
+        "rationale": "The Chinese README carries the same statement as the English one. Decided by reading the file in this session rather than by an API call."
+      }
+    },
+    {
+      "id": "adoption:cnosdb:planned_adoption",
+      "dbms": "cnosdb",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "Benxiaohai001, contributor of the CnosDB project, proposed adopting SQLancer; the issue is closed.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/cnosdb/cnosdb/issues/1335",
+          "source_type": "github_issue",
+          "excerpt": "use cases are 'select at /__w/cnosdb/cnosdb/query_server/test/cases/sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the CnosDB tracker proposing SQLancer, opened by Benxiaohai001 (contributor).",
+          "content_sha256": "sha256:bf891b8292ed1ed890b31cec2e3c60738743ab1194af7e094862ac116ac579b5",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/cnosdb/cnosdb/issues/860",
+          "source_type": "github_issue",
+          "excerpt": "SQLancer Support",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the CnosDB tracker proposing SQLancer, opened by ZuoTiJia (contributor).",
+          "content_sha256": "sha256:c32fb94eed74bca4f9c827f9979ebe3defecf86994ca80141caf88e9536dae3a",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:44:53Z",
+        "last_verified": "2026-09-06T16:44:53Z",
+        "source_url": "https://github.com/cnosdb/cnosdb/issues/1335",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/cnosdb/cnosdb"
+    },
+    {
+      "id": "adoption:cockroachdb:planned_adoption",
+      "dbms": "cockroachdb",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "srosenberg, contributor of the CockroachDB project, proposed adopting SQLancer; the issue is open.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/cockroachdb/cockroach/issues/110322",
+          "source_type": "github_issue",
+          "excerpt": "running `sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the CockroachDB tracker proposing SQLancer, opened by srosenberg (contributor).",
+          "content_sha256": "sha256:b4dc61113bdb90d968a27ebcb6bc34ff9f256147908fd71be6679bc9ad60e35b",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/cockroachdb/cockroach/issues/110528",
+          "source_type": "github_issue",
+          "excerpt": "add a _long-running_ workload utilizing `sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the CockroachDB tracker proposing SQLancer, opened by srosenberg (contributor).",
+          "content_sha256": "sha256:675320bfb18841b2a575593a089ca930f2bf3670e397aa8d72479445817fcf9d",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/cockroachdb/cockroach/issues/110321",
+          "source_type": "github_issue",
+          "excerpt": "running `sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the CockroachDB tracker proposing SQLancer, opened by srosenberg (contributor).",
+          "content_sha256": "sha256:b28c501447bbf29c6dbb81f318b7c45bcbe0ec92fd765b6ff1fd16e068067b8f",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/cockroachdb/cockroach/issues/101964",
+          "source_type": "github_issue",
+          "excerpt": "running sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the CockroachDB tracker proposing SQLancer, opened by rharding6373 (collaborator).",
+          "content_sha256": "sha256:cecdbbf0d394b6e975003d3911547f1d71cc9787762943909f73713a0aad834c",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:25:52Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/cockroachdb/cockroach/issues/62902",
+          "source_type": "github_issue",
+          "excerpt": "Add additional features to SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the CockroachDB tracker proposing SQLancer, opened by mgartner (contributor).",
+          "content_sha256": "sha256:1e5dc15f64eac5a40fb32882be1c0c237a94004897dd8817719834c327d28d05",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:25:52Z",
+        "last_verified": "2026-09-06T16:44:53Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/110322",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/cockroachdb/cockroach"
+    },
+    {
+      "id": "adoption:databend:official_testing",
+      "dbms": "databend",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The Databend project maintains its own SQLancer fork, last updated 2022-09-28.",
+      "since_year": null,
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/datafuse-extras/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "datafuse-extras/sqlancer is a fork of sqlancer/sqlancer under the Databend project's own GitHub organisation.",
+          "retrieved_at": "2026-09-06T15:17:57Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-06T15:17:57Z"
+        }
+      ],
+      "provenance": {
+        "collector": "forks",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T15:17:57Z",
+        "last_verified": "2026-09-06T15:17:57Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/forks",
+        "source_type": "github_repository"
+      },
+      "project_repository": "https://github.com/datafuselabs/databend"
+    },
+    {
+      "id": "adoption:databend:planned_adoption",
+      "dbms": "databend",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "bohutang, member of the Databend project, proposed adopting SQLancer; the issue is closed.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/databendlabs/databend/issues/1768",
+          "source_type": "github_issue",
+          "excerpt": "add sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Databend tracker proposing SQLancer, opened by bohutang (member).",
+          "content_sha256": "sha256:db694c37b84eaa2394bbdbb56ac1ffeca937ee6867640e9b17f9792aba01f230",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        },
+        {
+          "source_url": "https://github.com/databendlabs/databend/issues/8238",
+          "source_type": "github_issue",
+          "excerpt": "set up databend：[fix: unstable deployment databend · sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Databend tracker proposing SQLancer, opened by hanyisong (contributor).",
+          "content_sha256": "sha256:a80cb2883a5e93c37ff70e11115e1ee3b4a0a6baeb6f120e65341cc63fdaf30a",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-06T16:54:23Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/1768",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/datafuselabs/databend"
+    },
+    {
+      "id": "adoption:datafusion:official_testing",
+      "dbms": "datafusion",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The Apache DataFusion project maintains SQLancer testing setup in its own repository (docs/source/contributor-guide/testing.md).",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/apache/datafusion/blob/9082d6b10c29b72d56bede3d8e353d9d61fde542/docs/source/contributor-guide/testing.md",
+          "source_type": "github_code",
+          "excerpt": "DataFusion uses the [SQLancer] for \"fuzz\" testing: it generates random SQL\nqueries and execute them against DataFusion to find bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "docs/source/contributor-guide/testing.md in the Apache DataFusion repository states that the project uses SQLancer.",
+          "content_sha256": "sha256:36944b630c9f0bfe9d21869257a7b243b2f7e2bff36943a52553719dd5ba0229",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:08:58Z",
+        "last_verified": "2026-09-13T05:57:42Z",
+        "source_url": "https://github.com/apache/datafusion/blob/9082d6b10c29b72d56bede3d8e353d9d61fde542/docs/source/contributor-guide/testing.md",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/apache/datafusion"
+    },
+    {
+      "id": "adoption:datafusion:planned_adoption",
+      "dbms": "datafusion",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "2010YOUY01, contributor of the Apache DataFusion project, proposed adopting SQLancer; the issue is open.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/apache/datafusion/issues/11030",
+          "source_type": "github_issue",
+          "excerpt": "using randomly generated SQLs, `SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache DataFusion tracker proposing SQLancer, opened by 2010YOUY01 (contributor).",
+          "content_sha256": "sha256:131ec048980ed6149fd0417ea6e03e0a2f838a48c87683a3772be22825818186",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/datafusion/issues/16689",
+          "source_type": "github_issue",
+          "excerpt": "adding a decimal256 to a float (SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache DataFusion tracker proposing SQLancer, opened by 2010YOUY01 (contributor).",
+          "content_sha256": "sha256:21e414ff3591ea474b586a4ccfcdf23aff0092e67879360eb350746f71f96609",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/datafusion/issues/12814",
+          "source_type": "github_issue",
+          "excerpt": "running an aggregate query (SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache DataFusion tracker proposing SQLancer, opened by 2010YOUY01 (contributor).",
+          "content_sha256": "sha256:554561a05d58cb405765c99593f15a295d4d1d55a0ff4a0bce94e989083eb32c",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:44:53Z",
+        "last_verified": "2026-09-06T16:44:53Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11030",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/apache/datafusion"
+    },
+    {
+      "id": "adoption:doris:planned_adoption",
+      "dbms": "doris",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "ChaseHuangxu, contributor of the Apache Doris project, proposed adopting SQLancer; the issue is closed.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/apache/doris/issues/19786",
+          "source_type": "github_issue",
+          "excerpt": "Add support for Doris in SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Doris tracker proposing SQLancer, opened by ChaseHuangxu (contributor).",
+          "content_sha256": "sha256:6c853ac969e9b2e12fe87008dcb9b2eed008c7e4b3ae6e47e49ca89144c7912d",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/doris/issues/17701",
+          "source_type": "github_issue",
+          "excerpt": "add Apache Doris support to SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Doris tracker proposing SQLancer, opened by ChaseHuangxu (contributor).",
+          "content_sha256": "sha256:cb02cffbc73d8e9927718230768178ec2741aec25bf82ced0e4e15e163978065",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/doris/issues/19614",
+          "source_type": "github_issue",
+          "excerpt": "add Apache Doris support to SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Doris tracker proposing SQLancer, opened by ChaseHuangxu (contributor).",
+          "content_sha256": "sha256:612a1c1c413c869c3753fc21b15bac785a1b23141bd4a4b43e3573745ccf72cd",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/doris/issues/3874",
+          "source_type": "github_issue",
+          "excerpt": "running the [sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Doris tracker proposing SQLancer, opened by sduzh (contributor).",
+          "content_sha256": "sha256:8aaf67c0a761be4bc02267cda6b27e94dabbda8c2beca75eb5ebca6eab091d3c",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/doris/issues/19611",
+          "source_type": "github_issue",
+          "excerpt": "add Apache Doris support to SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Doris tracker proposing SQLancer, opened by ChaseHuangxu (contributor).",
+          "content_sha256": "sha256:4ec66bf9ee407d73683c1516ac3e65e01144ec08494061b2ad74e115f2ec7b59",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/doris/issues/17697",
+          "source_type": "github_issue",
+          "excerpt": "add Apache Doris support to SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Doris tracker proposing SQLancer, opened by ChaseHuangxu (contributor).",
+          "content_sha256": "sha256:75848211c0ba241550c0e77ef955964fa5f23d398fdc9fbe3f80fb102d6ffa35",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/doris/issues/17700",
+          "source_type": "github_issue",
+          "excerpt": "add Apache Doris support to SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Doris tracker proposing SQLancer, opened by ChaseHuangxu (contributor).",
+          "content_sha256": "sha256:9676511a9eacf40563abc038e59aeba7a0d7ea0d668226d01ab0c3a01e7de03d",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/doris/issues/19613",
+          "source_type": "github_issue",
+          "excerpt": "add Apache Doris support to SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Doris tracker proposing SQLancer, opened by ChaseHuangxu (contributor).",
+          "content_sha256": "sha256:159914eced72436bcb75fcd79b88032fb071ba6b91b729ea5d35e50d989f8120",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/apache/doris/issues/17705",
+          "source_type": "github_issue",
+          "excerpt": "add Apache Doris support to SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Apache Doris tracker proposing SQLancer, opened by ChaseHuangxu (contributor).",
+          "content_sha256": "sha256:8c1c68c6cc5b46f14e9f074688fa688b2df23231ccd3d6a8dca3ab8f56983ca5",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:44:53Z",
+        "last_verified": "2026-09-06T16:44:53Z",
+        "source_url": "https://github.com/apache/doris/issues/19786",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/apache/doris"
+    },
+    {
+      "id": "adoption:duckdb:official_testing",
+      "dbms": "duckdb",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The DuckDB project keeps a regression test for each SQLancer finding under test/issues/rigger/, so the bugs it found stay fixed.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/duckdb/duckdb/blob/HEAD/test/issues/rigger/instr_crash.test",
+          "source_type": "github_code",
+          "excerpt": "# description: SQLancer bug that found a crash in the instr implementation",
+          "excerpt_is_verbatim": true,
+          "note": "test/issues/rigger/instr_crash.test in the DuckDB repository, one of the files the project keeps for this.",
+          "content_sha256": "sha256:aba5e901063851acd87e522db37d8ab9237f6c93c8d9e16d270671b4a71de232",
+          "retrieved_at": "2026-09-13T07:10:22Z",
+          "first_seen": "2026-09-13T07:10:22Z",
+          "last_verified": "2026-09-13T07:10:22Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T07:10:22Z",
+        "last_verified": "2026-09-13T07:10:22Z",
+        "source_url": "https://github.com/duckdb/duckdb/blob/HEAD/test/issues/rigger/instr_crash.test",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/duckdb/duckdb"
+    },
+    {
+      "id": "adoption:duckdb:planned_adoption",
+      "dbms": "duckdb",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "Mytherin, collaborator of the DuckDB project, proposed adopting SQLancer; the issue is closed.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/duckdb/duckdb/issues/5031",
+          "source_type": "github_issue",
+          "excerpt": "RUNNING SQLANCER",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the DuckDB tracker proposing SQLancer, opened by Mytherin (collaborator).",
+          "content_sha256": "sha256:49042991f9bf13672f056327695e14c94b9ec4111bbff5a3a60da8828b0ecf10",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-06T16:25:52Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/duckdblabs/duckdb-fuzzer-ci/issues/7",
+          "source_type": "github_issue",
+          "excerpt": "Add SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the DuckDB tracker proposing SQLancer, opened by Mytherin (contributor).",
+          "content_sha256": "sha256:7abb50093ab8aa9e3ed9cf25538e6f4a323ba561831ef8890c4a983a4a1b211a",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:25:52Z",
+        "last_verified": "2026-09-13T05:57:42Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/5031",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/duckdb/duckdb"
+    },
+    {
+      "id": "adoption:feldera:official_testing",
+      "dbms": "feldera",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The Feldera project maintains its own SQLancer fork, 8 commit(s) ahead of upstream, last updated 2025-01-08.",
+      "since_year": null,
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/feldera/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "feldera/sqlancer is a fork of sqlancer/sqlancer under the Feldera project's own GitHub organisation, 8 commit(s) ahead of upstream.",
+          "retrieved_at": "2026-09-06T15:17:57Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-06T15:17:57Z"
+        }
+      ],
+      "provenance": {
+        "collector": "forks",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T15:17:57Z",
+        "last_verified": "2026-09-06T15:17:57Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/forks",
+        "source_type": "github_repository"
+      },
+      "project_repository": "https://github.com/feldera/feldera"
+    },
+    {
+      "id": "adoption:feldera:planned_adoption",
+      "dbms": "feldera",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "abhizer, contributor of the Feldera project, proposed adopting SQLancer; the issue is closed.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/feldera/feldera/issues/2934",
+          "source_type": "github_issue",
+          "excerpt": "support for feldera in SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Feldera tracker proposing SQLancer, opened by abhizer (contributor).",
+          "content_sha256": "sha256:f6c601a0134bfd3fddab4398c46147dc356f98eef2f09a86ed01baf8f763b92b",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:44:53Z",
+        "last_verified": "2026-09-06T16:44:53Z",
+        "source_url": "https://github.com/feldera/feldera/issues/2934",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/feldera/feldera"
+    },
+    {
+      "id": "adoption:greptimedb:planned_adoption",
+      "dbms": "greptimedb",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "J0HN50N133, contributor of the GreptimeDB project, proposed adopting SQLancer; the issue is closed.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/GreptimeTeam/greptimedb/issues/4402",
+          "source_type": "github_issue",
+          "excerpt": "Introduce sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the GreptimeDB tracker proposing SQLancer, opened by J0HN50N133 (contributor).",
+          "content_sha256": "sha256:cfb4a604d33087113e07cac43b0549d6b5fafeebf847721c4ce33ba522a698b2",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T05:57:42Z",
+        "last_verified": "2026-09-13T05:57:42Z",
+        "source_url": "https://github.com/GreptimeTeam/greptimedb/issues/4402",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/GreptimeTeam/greptimedb"
+    },
+    {
+      "id": "adoption:hazelcast:official_testing",
+      "dbms": "hazelcast",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The Hazelcast project maintains its own SQLancer fork, last updated 2023-10-12.",
+      "since_year": null,
+      "active": false,
+      "evidence": [
+        {
+          "source_url": "https://github.com/hazelcast/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "hazelcast/sqlancer is a fork of sqlancer/sqlancer under the Hazelcast project's own GitHub organisation.",
+          "retrieved_at": "2026-09-06T15:17:57Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-06T15:17:57Z"
+        }
+      ],
+      "provenance": {
+        "collector": "forks",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T15:17:57Z",
+        "last_verified": "2026-09-06T15:17:57Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/forks",
+        "source_type": "github_repository"
+      },
+      "project_repository": "https://github.com/hazelcast/hazelcast"
+    },
+    {
+      "id": "adoption:materialize:official_testing",
+      "dbms": "materialize",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The Materialize project maintains its own SQLancer fork, 28 commit(s) ahead of upstream, last updated 2026-07-24.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/MaterializeInc/materialize/blob/43467ac56efe6a6e844548d0d37449532aeb63be/test/sqlancer/Dockerfile",
+          "source_type": "github_code",
+          "excerpt": "RUN git clone https://github.com/MaterializeInc/sqlancer \\",
+          "excerpt_is_verbatim": true,
+          "note": "test/sqlancer/Dockerfile in the Materialize repository invokes SQLancer.",
+          "content_sha256": "sha256:4650f1ed952ca3a9d3c46fd49069cc78b1aecbfcd277a03e26c0166cff533285",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/MaterializeInc/materialize/blob/43467ac56efe6a6e844548d0d37449532aeb63be/test/sqlancer/mzcompose.py",
+          "source_type": "github_code",
+          "excerpt": "from materialize.sqlancer import create_services, run_sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "test/sqlancer/mzcompose.py in the Materialize repository invokes SQLancer.",
+          "content_sha256": "sha256:8a8c5dda3151a4ed94d38ad5d6b01d100f252b35cee45cbe21f7b75f4f502b0d",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/MaterializeInc/materialize/blob/43467ac56efe6a6e844548d0d37449532aeb63be/test/sqlancer/mzbuild.yml",
+          "source_type": "github_code",
+          "excerpt": "name: sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "test/sqlancer/mzbuild.yml in the Materialize repository is part of a SQLancer setup the project maintains.",
+          "content_sha256": "sha256:371e6d0acfe0b9ffcbfec484e364cde4a8f65adb7f542d72f3b5d7990e1b68ba",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/MaterializeInc/materialize/blob/43467ac56efe6a6e844548d0d37449532aeb63be/test/sqlancerplusplus/Dockerfile",
+          "source_type": "github_code",
+          "excerpt": "RUN git clone --single-branch --depth=1 https://github.com/MaterializeInc/sqlancerplusplus \\",
+          "excerpt_is_verbatim": true,
+          "note": "test/sqlancerplusplus/Dockerfile in the Materialize repository invokes SQLancer.",
+          "content_sha256": "sha256:9f1ea4c46bf15245ce3b14c59c19d3b7bf0d720ecce7a7bf5a5db389b4c9091e",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/MaterializeInc/materialize/blob/43467ac56efe6a6e844548d0d37449532aeb63be/test/sqlancerplusplus/mzcompose.py",
+          "source_type": "github_code",
+          "excerpt": "from materialize.sqlancer import create_services, run_sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "test/sqlancerplusplus/mzcompose.py in the Materialize repository invokes SQLancer.",
+          "content_sha256": "sha256:3239e0c2fe8560cd4c72360be9edcbcacf26fb5afb00c118a73e672507fa4441",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/MaterializeInc/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "MaterializeInc/sqlancer is a fork of sqlancer/sqlancer under the Materialize project's own GitHub organisation, 28 commit(s) ahead of upstream.",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:08:58Z",
+        "last_verified": "2026-09-13T05:57:42Z",
+        "source_url": "https://github.com/MaterializeInc/materialize/blob/43467ac56efe6a6e844548d0d37449532aeb63be/test/sqlancer/Dockerfile",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/MaterializeInc/materialize"
+    },
+    {
+      "id": "adoption:monetdb:official_testing",
+      "dbms": "monetdb",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The MonetDB project maintains its own SQLancer fork, 279 commit(s) ahead of upstream, last updated 2023-02-27.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/MonetDB/MonetDB/blob/08da5de637aea8f702f5ccbda0a6c53abc4bba05/sql/test/SQLancer/Tests/All",
+          "source_type": "github_code",
+          "excerpt": "sqlancer01",
+          "excerpt_is_verbatim": true,
+          "note": "sql/test/SQLancer/Tests/All in the MonetDB repository is part of a SQLancer setup the project maintains.",
+          "content_sha256": "sha256:d38e03147dd77e47ace167836653e4235648f3d210a0bfeffc0a608374730cd3",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/MonetDB/MonetDB/blob/08da5de637aea8f702f5ccbda0a6c53abc4bba05/sql/test/Tests/orderby-nulls-first-last.test",
+          "source_type": "github_code",
+          "excerpt": "test\n-- sql/test/SQLancer/Tests/sqlancer05.",
+          "excerpt_is_verbatim": true,
+          "note": "sql/test/Tests/orderby-nulls-first-last.test in the MonetDB repository states that the project uses SQLancer.",
+          "content_sha256": "sha256:ad5e1ec49fdd01d42038a28cd3702d3888db27861115e8695cc2d9d601cdd175",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        },
+        {
+          "source_url": "https://github.com/MonetDB/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "MonetDB/sqlancer is a fork of sqlancer/sqlancer under the MonetDB project's own GitHub organisation, 279 commit(s) ahead of upstream.",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:08:58Z",
+        "last_verified": "2026-09-13T05:57:42Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/blob/08da5de637aea8f702f5ccbda0a6c53abc4bba05/sql/test/SQLancer/Tests/All",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/MonetDB/MonetDB"
+    },
+    {
+      "id": "adoption:noisepage:official_testing",
+      "dbms": "noisepage",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The NoisePage project maintains its own SQLancer fork, 14 commit(s) ahead of upstream, last updated 2023-02-28.",
+      "since_year": null,
+      "active": false,
+      "evidence": [
+        {
+          "source_url": "https://github.com/cmu-db/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "cmu-db/sqlancer is a fork of sqlancer/sqlancer under the NoisePage project's own GitHub organisation, 14 commit(s) ahead of upstream.",
+          "retrieved_at": "2026-09-06T16:25:52Z",
+          "first_seen": "2026-09-06T16:25:52Z",
+          "last_verified": "2026-09-06T16:25:52Z"
+        }
+      ],
+      "provenance": {
+        "collector": "forks",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:25:52Z",
+        "last_verified": "2026-09-06T16:25:52Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/forks",
+        "source_type": "github_repository"
+      },
+      "project_repository": "https://github.com/cmu-db/noisepage"
+    },
+    {
+      "id": "adoption:noisepage:planned_adoption",
+      "dbms": "noisepage",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "jazzplato, member of the NoisePage project, proposed adopting SQLancer; the issue is open.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/cmu-db/noisepage/issues/1012",
+          "source_type": "github_issue",
+          "excerpt": "running the sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the NoisePage tracker proposing SQLancer, opened by jazzplato (member).",
+          "content_sha256": "sha256:c8c25b83b190114bba223e61f642a3254e16bf01724c75589c9c06068756e57a",
+          "retrieved_at": "2026-09-06T16:25:52Z",
+          "first_seen": "2026-09-06T16:25:52Z",
+          "last_verified": "2026-09-06T16:25:52Z"
+        },
+        {
+          "source_url": "https://github.com/cmu-db/noisepage/issues/997",
+          "source_type": "github_issue",
+          "excerpt": "support [SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the NoisePage tracker proposing SQLancer, opened by apavlo (member).",
+          "content_sha256": "sha256:2e037c0e123a3be1b1d12484f54958e521a3dea9f8f9b5fa65a56e7cd82c45af",
+          "retrieved_at": "2026-09-06T16:25:52Z",
+          "first_seen": "2026-09-06T16:25:52Z",
+          "last_verified": "2026-09-06T16:25:52Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:25:52Z",
+        "last_verified": "2026-09-06T16:25:52Z",
+        "source_url": "https://github.com/cmu-db/noisepage/issues/1012",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/cmu-db/noisepage"
+    },
+    {
+      "id": "adoption:oceanbase:official_testing",
+      "dbms": "oceanbase",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The OceanBase project maintains its own SQLancer fork, last updated 2022-12-05.",
+      "since_year": null,
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/oceanbase/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "oceanbase/sqlancer is a fork of sqlancer/sqlancer under the OceanBase project's own GitHub organisation.",
+          "retrieved_at": "2026-09-06T15:17:57Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-06T15:17:57Z"
+        }
+      ],
+      "provenance": {
+        "collector": "forks",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T15:17:57Z",
+        "last_verified": "2026-09-06T15:17:57Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/forks",
+        "source_type": "github_repository"
+      },
+      "project_repository": "https://github.com/oceanbase/oceanbase"
+    },
+    {
+      "id": "adoption:oxla:official_testing",
+      "dbms": "oxla",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The Oxla project maintains its own SQLancer fork, 15 commit(s) ahead of upstream, last updated 2026-03-10.",
+      "since_year": null,
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/redpanda-data/oxla-sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "redpanda-data/oxla-sqlancer is a fork of sqlancer/sqlancer under the Oxla project's own GitHub organisation, 15 commit(s) ahead of upstream.",
+          "retrieved_at": "2026-09-13T05:57:42Z",
+          "first_seen": "2026-09-13T05:57:42Z",
+          "last_verified": "2026-09-13T05:57:42Z"
+        }
+      ],
+      "provenance": {
+        "collector": "forks",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T05:57:42Z",
+        "last_verified": "2026-09-13T05:57:42Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/forks",
+        "source_type": "github_repository"
+      }
+    },
+    {
+      "id": "adoption:presto:planned_adoption",
+      "dbms": "presto",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "tdcmeehan, contributor of the Presto project, proposed adopting SQLancer; the issue is open.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/prestodb/presto/issues/23324",
+          "source_type": "github_issue",
+          "excerpt": "using [SQLAncer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Presto tracker proposing SQLancer, opened by tdcmeehan (contributor).",
+          "content_sha256": "sha256:171ab55c618537a7dddfa8b237d3d7f3792e4351d4e089cc932425c85ab42abd",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:44:53Z",
+        "last_verified": "2026-09-06T16:44:53Z",
+        "source_url": "https://github.com/prestodb/presto/issues/23324",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/prestodb/presto"
+    },
+    {
+      "id": "adoption:risingwave:planned_adoption",
+      "dbms": "risingwave",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "lmatz, contributor of the RisingWave project, proposed adopting SQLancer; the issue is open.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/risingwavelabs/risingwave/issues/3364",
+          "source_type": "github_issue",
+          "excerpt": "Introduce Sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the RisingWave tracker proposing SQLancer, opened by lmatz (contributor).",
+          "content_sha256": "sha256:eb6b29ba6a61e5eb0dcf9c2959ef678acd401b541f4883e8df2bc46777392d7d",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:44:53Z",
+        "last_verified": "2026-09-06T16:44:53Z",
+        "source_url": "https://github.com/risingwavelabs/risingwave/issues/3364",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/risingwavelabs/risingwave"
+    },
+    {
+      "id": "adoption:seekdb:official_testing",
+      "dbms": "seekdb",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The SeekDB project maintains its own SQLancer fork, last updated 2022-12-05.",
+      "since_year": null,
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/oceanbase/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "oceanbase/sqlancer is a fork of sqlancer/sqlancer under the SeekDB project's own GitHub organisation.",
+          "retrieved_at": "2026-09-06T16:25:52Z",
+          "first_seen": "2026-09-06T16:25:52Z",
+          "last_verified": "2026-09-06T16:25:52Z"
+        }
+      ],
+      "provenance": {
+        "collector": "forks",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:25:52Z",
+        "last_verified": "2026-09-06T16:25:52Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/forks",
+        "source_type": "github_repository"
+      },
+      "project_repository": "https://github.com/oceanbase/seekdb"
+    },
+    {
+      "id": "adoption:serenedb:official_testing",
+      "dbms": "serenedb",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The SereneDB project maintains SQLancer testing setup in its own repository (tests/drivers/sqlsmith/README.md).",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/serenedb/serenedb/blob/3084b33f190134c3d6a9c33feeb44914e56de46c/tests/drivers/sqlsmith/README.md",
+          "source_type": "github_code",
+          "excerpt": "DDL/DML fuzzing belongs to later phases (SQLancer).",
+          "excerpt_is_verbatim": true,
+          "note": "tests/drivers/sqlsmith/README.md in the SereneDB repository states that the project uses SQLancer.",
+          "content_sha256": "sha256:d5aa844b4c020a7f2e2f8c68787097ee5c712f61114a82cb0d057f1f0f9cedf5",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-06T16:54:23Z",
+        "source_url": "https://github.com/serenedb/serenedb/blob/3084b33f190134c3d6a9c33feeb44914e56de46c/tests/drivers/sqlsmith/README.md",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/serenedb/serenedb"
+    },
+    {
+      "id": "adoption:sparq:official_testing",
+      "dbms": "sparq",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The sparq project maintains SQLancer testing setup in its own repository (research/paper-selection.md).",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/sparq-org/sparq/blob/60e3f84c1e9ef2dc6a9352a131e825d3e60894c1/research/paper-selection.md",
+          "source_type": "github_code",
+          "excerpt": "**Field bar (search-verified).** SQLancer (Rigger & Su): ~196 previously-unknown",
+          "excerpt_is_verbatim": true,
+          "note": "research/paper-selection.md in the sparq repository invokes SQLancer.",
+          "content_sha256": "sha256:d5af603a926ec1c250fdab4b999e46d8f02a8137f56bd5444d20741f858158c4",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        },
+        {
+          "source_url": "https://github.com/sparq-org/sparq/blob/60e3f84c1e9ef2dc6a9352a131e825d3e60894c1/skills/academic-paper/SKILL.md",
+          "source_type": "github_code",
+          "excerpt": "7`) is the first dedicated\nlogic-bug testing paper for SPARQL engines (SQL/Datalog/Cypher are covered by\nSQLancer/queryFuzz/GDsmith; SPARQL is not).",
+          "excerpt_is_verbatim": true,
+          "note": "skills/academic-paper/SKILL.md in the sparq repository states that the project uses SQLancer.",
+          "content_sha256": "sha256:8d794b7223a5ef891f542b5776f11ac1c0bca336fe519b0ad0bf6d86aa295a45",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-06T16:54:23Z",
+        "source_url": "https://github.com/sparq-org/sparq/blob/60e3f84c1e9ef2dc6a9352a131e825d3e60894c1/research/paper-selection.md",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/sparq-org/sparq"
+    },
+    {
+      "id": "adoption:spiceai:planned_adoption",
+      "dbms": "spiceai",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "digadeesh proposed that the Spice.ai OSS project adopt SQLancer for SQL fuzz testing; the issue is open. This is an intention, not evidence that the project runs SQLancer.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/spiceai/spiceai/issues/2119",
+          "source_type": "github_issue",
+          "excerpt": "Enhancement: SQL Fuzz testing with SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Spice.ai OSS tracker proposing SQLancer, opened by digadeesh.",
+          "content_sha256": "sha256:e3b1453a80c37223ded0b6d0c218e92270fafdb2497abe19c12f811da69d903b",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:19:32Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/spiceai/spiceai/issues/2119",
+          "source_type": "github_issue",
+          "excerpt": "using [SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Spice.ai OSS tracker proposing SQLancer, opened by digadeesh (contributor).",
+          "content_sha256": "sha256:e3b1453a80c37223ded0b6d0c218e92270fafdb2497abe19c12f811da69d903b",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:19:32Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/spiceai/spiceai/issues/2186",
+          "source_type": "github_issue",
+          "excerpt": "Add parameters for `spiceai-sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Spice.ai OSS tracker proposing SQLancer, opened by y-f-u (contributor).",
+          "content_sha256": "sha256:2c9a6228d59c059154fb231a0f3d6472673118de7fa385f7191f0fe28341b3f9",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/spiceai/spiceai/issues/2190",
+          "source_type": "github_issue",
+          "excerpt": "Add a CI task to run `spiceai-sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Spice.ai OSS tracker proposing SQLancer, opened by y-f-u (contributor).",
+          "content_sha256": "sha256:8be652522af8ac00c86195fdb3264a57a9e1524f7a601291c3a31c2cd587679b",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/spiceai/spiceai/issues/2187",
+          "source_type": "github_issue",
+          "excerpt": "Add parameters for `spiceai-sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Spice.ai OSS tracker proposing SQLancer, opened by y-f-u (contributor).",
+          "content_sha256": "sha256:e9167018c3f598377e22082640116bb09ba93e22fdc19f59365180fe9347ba10",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/spiceai/spiceai/issues/2188",
+          "source_type": "github_issue",
+          "excerpt": "Add a process in `spiceai-sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Spice.ai OSS tracker proposing SQLancer, opened by y-f-u (contributor).",
+          "content_sha256": "sha256:e3a621d5a2450925842764b8ad7d382504ceaaced2a1478b0a7d77fcddd1ddb5",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-06T16:44:53Z",
+        "source_url": "https://github.com/spiceai/spiceai/issues/2119",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/spiceai/spiceai"
+    },
+    {
+      "id": "adoption:starrocks:developer_use",
+      "dbms": "starrocks",
+      "relationship": "developer_use",
+      "finder": "sqlancer",
+      "summary": "StarRocks' own release notes credit SQLancer with finding bugs in the release, and point at the sqlancer label on their issue tracker. Decided by reading the file in this session rather than by an API call.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/StarRocks/starrocks/blob/444cb3cf593a8406c096ca79d270908411414654/docs/en/release_notes/release-2.2.md",
+          "source_type": "github_code",
+          "excerpt": "A few bugs are detected by Synthesized Query Lancer (SQLancer).",
+          "excerpt_is_verbatim": true,
+          "note": "docs/en/release_notes/release-2.2.md in the StarRocks repository.",
+          "content_sha256": "sha256:595a2b870d2d9bbd9fc9eb3f09844d1c9f6a98f32d45c135c75507e45e3e6257",
+          "retrieved_at": "2026-09-13T06:25:57Z",
+          "first_seen": "2026-09-13T06:25:57Z",
+          "last_verified": "2026-09-13T06:25:57Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T06:25:57Z",
+        "last_verified": "2026-09-13T06:25:57Z",
+        "source_url": "https://github.com/StarRocks/starrocks/blob/444cb3cf593a8406c096ca79d270908411414654/docs/en/release_notes/release-2.2.md",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/StarRocks/starrocks",
+      "classifier": {
+        "method": "llm_classification",
+        "classifier_version": "adoption-v1",
+        "policy_version": "impact-policy-v1",
+        "taxonomy_version": "taxonomy-v2",
+        "evidence_sha256": "sha256:90a93e293c86c36d21d2ba3cb7be89bbf881798a97f8eb29ad178266f146d6dc",
+        "classified_at": "2026-09-13T06:25:59Z",
+        "model": "claude-opus-5",
+        "rationale": "StarRocks' own release notes credit SQLancer with finding bugs in the release, and point at the sqlancer label on their issue tracker. Decided by reading the file in this session rather than by an API call."
+      }
+    },
+    {
+      "id": "adoption:tarantool:planned_adoption",
+      "dbms": "tarantool",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "ligurio, member of the Tarantool project, proposed adopting SQLancer; the issue is open.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/tarantool/tarantool/issues/4833",
+          "source_type": "github_issue",
+          "excerpt": "support tarantool in sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Tarantool tracker proposing SQLancer, opened by ligurio (member).",
+          "content_sha256": "sha256:14af7c9268f1ae55832457bb2833ae8bb30038d0c172c90589f9338f00d7f33f",
+          "retrieved_at": "2026-09-06T16:25:52Z",
+          "first_seen": "2026-09-06T16:25:52Z",
+          "last_verified": "2026-09-06T16:25:52Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:25:52Z",
+        "last_verified": "2026-09-06T16:25:52Z",
+        "source_url": "https://github.com/tarantool/tarantool/issues/4833",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/tarantool/tarantool"
+    },
+    {
+      "id": "adoption:turso:official_testing",
+      "dbms": "turso",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The Turso project maintains SQLancer testing setup in its own repository (testing/sqlancer/sqlancer-runner/Dockerfile.sqlancer).",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/tursodatabase/turso/blob/7e2fc39de6ddce1100c5c014ee4b801db49e3ac2/testing/sqlancer/sqlancer-runner/Dockerfile.sqlancer",
+          "source_type": "github_code",
+          "excerpt": "# Dockerfile for SQLancer ECS Runner",
+          "excerpt_is_verbatim": true,
+          "note": "testing/sqlancer/sqlancer-runner/Dockerfile.sqlancer in the Turso repository invokes SQLancer.",
+          "content_sha256": "sha256:223114516fd39ee42f1c1afc2c215a312b0ba110a4c3374d62049972618baaa3",
+          "retrieved_at": "2026-09-06T16:33:38Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:33:38Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/blob/7e2fc39de6ddce1100c5c014ee4b801db49e3ac2/testing/sqlancer/sqlancer-runner/docker-entrypoint.sqlancer.ts",
+          "source_type": "github_code",
+          "excerpt": "import { parseFailure, isCorruptionError, type SqlancerFailure } from \"./logParse.ts\";",
+          "excerpt_is_verbatim": true,
+          "note": "testing/sqlancer/sqlancer-runner/docker-entrypoint.sqlancer.ts in the Turso repository invokes SQLancer.",
+          "content_sha256": "sha256:3f636cbb3e1794340d4fe7691fc02d4289b310eaf8fa5f97c86ab350aa2c3d36",
+          "retrieved_at": "2026-09-06T16:33:38Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:33:38Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/blob/7e2fc39de6ddce1100c5c014ee4b801db49e3ac2/testing/sqlancer/README.md",
+          "source_type": "github_code",
+          "excerpt": "Run [SQLancer](https://github.",
+          "excerpt_is_verbatim": true,
+          "note": "testing/sqlancer/README.md in the Turso repository states that the project uses SQLancer.",
+          "content_sha256": "sha256:2d35b552fc5779ccaf6531c6eb4bf02b54777093b3de2c7f99b81d37d8ffc505",
+          "retrieved_at": "2026-09-06T16:33:38Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:33:38Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/blob/7e2fc39de6ddce1100c5c014ee4b801db49e3ac2/testing/sqlancer/sqlancer-runner/README.md",
+          "source_type": "github_code",
+          "excerpt": "TIME_LIMIT_MINUTES=5 LOG_TO_STDOUT=true bun docker-entrypoint.sqlancer.ts",
+          "excerpt_is_verbatim": true,
+          "note": "testing/sqlancer/sqlancer-runner/README.md in the Turso repository invokes SQLancer.",
+          "content_sha256": "sha256:d65a0bade1946e5fdc9a1e8555fe9628446915aa2e0f03457c78508e80f095bd",
+          "retrieved_at": "2026-09-06T16:33:38Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:33:38Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/blob/7e2fc39de6ddce1100c5c014ee4b801db49e3ac2/testing/sqlancer/sqlancer-runner/slack.ts",
+          "source_type": "github_code",
+          "excerpt": "* Slack client for posting SQLancer run summaries.",
+          "excerpt_is_verbatim": true,
+          "note": "testing/sqlancer/sqlancer-runner/slack.ts in the Turso repository is part of a SQLancer setup the project maintains.",
+          "content_sha256": "sha256:4847bb1d2997ef358979e6ff286073e3ec5ca335ea0966c065833ebea4ae67ec",
+          "retrieved_at": "2026-09-06T16:33:38Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:33:38Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/blob/7e2fc39de6ddce1100c5c014ee4b801db49e3ac2/testing/sqlancer/sqlancer-runner/github.ts",
+          "source_type": "github_code",
+          "excerpt": "import type { SqlancerFailure, CorruptionAnalysis } from \"./logParse.ts\";",
+          "excerpt_is_verbatim": true,
+          "note": "testing/sqlancer/sqlancer-runner/github.ts in the Turso repository invokes SQLancer.",
+          "content_sha256": "sha256:ebe977969968f4ec376bf919cc793f4e23cce1b8618d9b4657a4430ff3f17c1f",
+          "retrieved_at": "2026-09-06T16:33:38Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:33:38Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-06T16:33:38Z",
+        "source_url": "https://github.com/tursodatabase/turso/blob/7e2fc39de6ddce1100c5c014ee4b801db49e3ac2/testing/sqlancer/sqlancer-runner/Dockerfile.sqlancer",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/tursodatabase/turso"
+    },
+    {
+      "id": "adoption:turso:planned_adoption",
+      "dbms": "turso",
+      "relationship": "planned_adoption",
+      "finder": "sqlancer",
+      "summary": "LeMikaelF, collaborator of the Turso project, proposed adopting SQLancer; the issue is closed.",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/tursodatabase/turso/issues/4489",
+          "source_type": "github_issue",
+          "excerpt": "using SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Turso tracker proposing SQLancer, opened by LeMikaelF (collaborator).",
+          "content_sha256": "sha256:ed5d1bb3b3c7ba2c43617cf835e478936d9ccfc2ae276c0154222472c0327cda",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/issues/4488",
+          "source_type": "github_issue",
+          "excerpt": "using SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Turso tracker proposing SQLancer, opened by LeMikaelF (collaborator).",
+          "content_sha256": "sha256:fd59a559e1e16131e3645ea7696b2e7102cb8ce427bcde2775b027435d625b12",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/issues/4606",
+          "source_type": "github_issue",
+          "excerpt": "running `run-sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Turso tracker proposing SQLancer, opened by LeMikaelF (collaborator).",
+          "content_sha256": "sha256:7d473976e906d0172f6badfd7ee7e40897916b03659add22049db164031738b9",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/issues/4538",
+          "source_type": "github_issue",
+          "excerpt": "run-sqlancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Turso tracker proposing SQLancer, opened by LeMikaelF (collaborator).",
+          "content_sha256": "sha256:5eb0ae880692be920e3be775189b23eaa3ce68353482b325e108dfb2ef4e871f",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/issues/4603",
+          "source_type": "github_issue",
+          "excerpt": "running SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Turso tracker proposing SQLancer, opened by LeMikaelF (collaborator).",
+          "content_sha256": "sha256:0657a8140539d238f6a5e2d581f632cbfbc203a93162fca8ef6b807a8d20597c",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/issues/4602",
+          "source_type": "github_issue",
+          "excerpt": "running SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Turso tracker proposing SQLancer, opened by LeMikaelF (collaborator).",
+          "content_sha256": "sha256:e043c7cbe3e1594d37afd730c4b3b497c6e9bfa52198734211552e8b91e2829e",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        },
+        {
+          "source_url": "https://github.com/tursodatabase/turso/issues/4681",
+          "source_type": "github_issue",
+          "excerpt": "running all the SQLancer",
+          "excerpt_is_verbatim": true,
+          "note": "Issue in the Turso tracker proposing SQLancer, opened by PThorpe92 (contributor).",
+          "content_sha256": "sha256:212f95c07749ac39457be8959a0da5df6a2d67272b85bc6a6a1322e04157f00f",
+          "retrieved_at": "2026-09-06T16:44:53Z",
+          "first_seen": "2026-09-06T16:44:53Z",
+          "last_verified": "2026-09-06T16:44:53Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-06T16:44:53Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4489",
+        "source_type": "github_issue"
+      },
+      "project_repository": "https://github.com/tursodatabase/turso"
+    },
+    {
+      "id": "adoption:wadjet:official_testing",
+      "dbms": "wadjet",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The Wadjet project maintains SQLancer testing setup in its own repository (tools/sqlancer/run.sh).",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/derekmwright/wadjet/blob/672bb5e13151d255bd633e7efbc85c42e5642c15/tools/sqlancer/run.sh",
+          "source_type": "github_code",
+          "excerpt": "# Usage: tools/sqlancer/run.sh <args to pass to \"java -jar $SQLANCER_JAR\">",
+          "excerpt_is_verbatim": true,
+          "note": "tools/sqlancer/run.sh in the Wadjet repository invokes SQLancer.",
+          "content_sha256": "sha256:5d40f9528e7d1fe509f8775ecbd3b7d31d6153d1fbf1bb7945d4249a82910956",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        },
+        {
+          "source_url": "https://github.com/derekmwright/wadjet/blob/672bb5e13151d255bd633e7efbc85c42e5642c15/tools/sqlancer/build.sh",
+          "source_type": "github_code",
+          "excerpt": "git clone --depth 1 https://github.com/sqlancer/sqlancer.git \"$TARGET_DIR\"",
+          "excerpt_is_verbatim": true,
+          "note": "tools/sqlancer/build.sh in the Wadjet repository invokes SQLancer.",
+          "content_sha256": "sha256:6431eb4ea5695a5e139520531d1ec2e45f4a1d8d1d9cd9d6e630bd840a509085",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        },
+        {
+          "source_url": "https://github.com/derekmwright/wadjet/blob/672bb5e13151d255bd633e7efbc85c42e5642c15/tools/sqlancer/README.md",
+          "source_type": "github_code",
+          "excerpt": "Runbook for wadjet issue #289 (\"evaluate SQLancer (TLP/NoREC) against the",
+          "excerpt_is_verbatim": true,
+          "note": "tools/sqlancer/README.md in the Wadjet repository invokes SQLancer.",
+          "content_sha256": "sha256:766f2779b41accf65a2035434059df8d529975a6ace75494764c6496e1106987",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        },
+        {
+          "source_url": "https://github.com/derekmwright/wadjet/blob/672bb5e13151d255bd633e7efbc85c42e5642c15/tools/sqlancer/triage/report.go",
+          "source_type": "github_code",
+          "excerpt": "fmt.Fprintln(w, \"SQLancer triage report\")",
+          "excerpt_is_verbatim": true,
+          "note": "tools/sqlancer/triage/report.go in the Wadjet repository is part of a SQLancer setup the project maintains.",
+          "content_sha256": "sha256:593728d42d148d779b6a07c28aa6d9bd1792d3fcc6d732be528cc10f5b238cdc",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        },
+        {
+          "source_url": "https://github.com/derekmwright/wadjet/blob/672bb5e13151d255bd633e7efbc85c42e5642c15/tools/sqlancer/triage/classify.go",
+          "source_type": "github_code",
+          "excerpt": "// Package triage classifies SQLancer soak-log output (and the wadjet",
+          "excerpt_is_verbatim": true,
+          "note": "tools/sqlancer/triage/classify.go in the Wadjet repository is part of a SQLancer setup the project maintains.",
+          "content_sha256": "sha256:875d96f50d11053d3c644c563f76b538b3f8ae0b934e569678a9889890ccc92b",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        },
+        {
+          "source_url": "https://github.com/derekmwright/wadjet/blob/672bb5e13151d255bd633e7efbc85c42e5642c15/tools/sqlancer/adapter-src/sqlancer/wadjet/WadjetProvider.java",
+          "source_type": "github_code",
+          "excerpt": "import sqlancer.AbstractAction;",
+          "excerpt_is_verbatim": true,
+          "note": "tools/sqlancer/adapter-src/sqlancer/wadjet/WadjetProvider.java in the Wadjet repository invokes SQLancer.",
+          "content_sha256": "sha256:929d6102d9acd43524cabcd979b0bc6651c83cfa1910ff738782b74c3be41bc4",
+          "retrieved_at": "2026-09-06T16:54:23Z",
+          "first_seen": "2026-09-06T16:54:23Z",
+          "last_verified": "2026-09-06T16:54:23Z"
+        }
+      ],
+      "provenance": {
+        "collector": "adoption",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-06T16:54:23Z",
+        "source_url": "https://github.com/derekmwright/wadjet/blob/672bb5e13151d255bd633e7efbc85c42e5642c15/tools/sqlancer/run.sh",
+        "source_type": "github_code"
+      },
+      "project_repository": "https://github.com/derekmwright/wadjet"
+    },
+    {
+      "id": "adoption:xugu:official_testing",
+      "dbms": "xugu",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The XuGu project maintains its own SQLancer fork, 1 commit(s) ahead of upstream, last updated 2025-05-20.",
+      "since_year": null,
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/Xugu-Open-Source/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "Xugu-Open-Source/sqlancer is a fork of sqlancer/sqlancer under the XuGu project's own GitHub organisation, 1 commit(s) ahead of upstream.",
+          "retrieved_at": "2026-09-06T15:17:57Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-06T15:17:57Z"
+        }
+      ],
+      "provenance": {
+        "collector": "forks",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T15:17:57Z",
+        "last_verified": "2026-09-06T15:17:57Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/forks",
+        "source_type": "github_repository"
+      },
+      "project_repository": "https://github.com/Xugu-Open-Source/xugu"
+    },
+    {
+      "id": "adoption:ydb:official_testing",
+      "dbms": "ydb",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The YDB project maintains its own SQLancer fork, last updated 2022-09-30.",
+      "since_year": null,
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/ydb-platform/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "ydb-platform/sqlancer is a fork of sqlancer/sqlancer under the YDB project's own GitHub organisation.",
+          "retrieved_at": "2026-09-06T15:17:57Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-06T15:17:57Z"
+        }
+      ],
+      "provenance": {
+        "collector": "forks",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T15:17:57Z",
+        "last_verified": "2026-09-06T15:17:57Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/forks",
+        "source_type": "github_repository"
+      },
+      "project_repository": "https://github.com/ydb-platform/ydb"
+    },
+    {
+      "id": "adoption:yugabytedb:official_testing",
+      "dbms": "yugabytedb",
+      "relationship": "official_testing",
+      "finder": "sqlancer",
+      "summary": "The YugabyteDB project maintains its own SQLancer fork, 149 commit(s) ahead of upstream, last updated 2026-09-12.",
+      "since_year": null,
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/yugabyte/sqlancer",
+          "source_type": "github_repository",
+          "excerpt": null,
+          "note": "yugabyte/sqlancer is a fork of sqlancer/sqlancer under the YugabyteDB project's own GitHub organisation, 149 commit(s) ahead of upstream.",
+          "retrieved_at": "2026-09-13T07:10:22Z",
+          "first_seen": "2026-09-06T15:17:57Z",
+          "last_verified": "2026-09-13T07:10:22Z"
+        }
+      ],
+      "provenance": {
+        "collector": "forks",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T15:17:57Z",
+        "last_verified": "2026-09-13T07:10:22Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/forks",
+        "source_type": "github_repository"
+      },
+      "project_repository": "https://github.com/yugabyte/yugabyte-db"
+    }
+  ]
+}
diff --git a/_data/impact/bugs.json b/_data/impact/bugs.json
new file mode 100644
index 0000000..5264c8c
--- /dev/null
+++ b/_data/impact/bugs.json
@@ -0,0 +1,114380 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "bugs": [
+    {
+      "id": "bug:bharatdbms:9211d9aa05b2",
+      "dbms": "bharatdbms",
+      "title": "negative value of numgroups",
+      "reported_date": "2025-09-22",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "BharatDBPG",
+      "links": {
+        "report": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/1712"
+      },
+      "primary_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/1712",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/1712",
+            "source_type": "github_issue",
+            "excerpt": "Using sqlancer I've found curious vulnerability. In some places of the   \ncode we convert LONG_MAX to double. After value of 2^53 double doesn't",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0c9b49add6cd02a55d0402d9980b600cb3d3337fe30d1b2a710e4b4bdb7d44b8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/1712",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0c9b49add6cd02a55d0402d9980b600cb3d3337fe30d1b2a710e4b4bdb7d44b8"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:bharatdbms:321ac8605fd1",
+      "dbms": "bharatdbms",
+      "title": "clause accidentally pushed down ( possible bug in making vars outer-join aware)",
+      "reported_date": "2025-09-23",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "BharatDBPG",
+      "links": {
+        "report": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2155"
+      },
+      "primary_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2155",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2155",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer benchmark raised an issue with pushing down clauses in LEFT   \nJOINs. Example:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:121c33763a7e3888027dfa115a37dd3a2c430642590feefdf981ad2637fe3411",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2155",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:121c33763a7e3888027dfa115a37dd3a2c430642590feefdf981ad2637fe3411"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:bharatdbms:cc3ac2ee5735",
+      "dbms": "bharatdbms",
+      "title": "assert failure when create temp table",
+      "reported_date": "2025-09-24",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "BharatDBPG",
+      "links": {
+        "report": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2605"
+      },
+      "primary_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2605",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2605",
+            "source_type": "github_issue",
+            "excerpt": "I came across an Assert failure while running sqlancer. The repro query  \nis",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c81afcf9af24db8a33221445f6940fda0a67960b601762370c3303c6ac369925",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2605",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c81afcf9af24db8a33221445f6940fda0a67960b601762370c3303c6ac369925"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:bharatdbms:331efb2cbb89",
+      "dbms": "bharatdbms",
+      "title": "\"type with xxxx does not exist\" when doing execmemoize()",
+      "reported_date": "2025-09-25",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "BharatDBPG",
+      "links": {
+        "report": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2851"
+      },
+      "primary_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2851",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2851",
+            "source_type": "github_issue",
+            "excerpt": "I met Memoize node failed When I used sqlancer test postgres.  \ndatabase0=# explain select t0.c0 from t0 join t5 on t0.c0 = (t5.c0 - t5.c0);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3eccae42d0876408a7f4308b93d6446fcb3c3edebba566a61ea22ef163dca243",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/2851",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3eccae42d0876408a7f4308b93d6446fcb3c3edebba566a61ea22ef163dca243"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:bharatdbms:1d624c2642e4",
+      "dbms": "bharatdbms",
+      "title": "bug #19350: short circuit optimization missed when running sql scriptes in jdbc",
+      "reported_date": "2025-12-11",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "BharatDBPG",
+      "links": {
+        "report": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/4145"
+      },
+      "primary_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/4145",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/4145",
+            "source_type": "github_issue",
+            "excerpt": "String user = \"sqlancer\";  \nString password = \"sqlancer\";",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:17c39e39005b81792657743de323c62532a47e28cb0d9e2c4be1ec8eb45171da",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/4145",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:17c39e39005b81792657743de323c62532a47e28cb0d9e2c4be1ec8eb45171da"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:bharatdbms:ff0d40eb56c4",
+      "dbms": "bharatdbms",
+      "title": "bug #19350: short circuit optimization missed when running sql scriptes in jdbc",
+      "reported_date": "2026-02-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "BharatDBPG",
+      "links": {
+        "report": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/4237"
+      },
+      "primary_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/4237",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/4237",
+            "source_type": "github_issue",
+            "excerpt": "String user = \"sqlancer\";  \nString password = \"sqlancer\";",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:17c39e39005b81792657743de323c62532a47e28cb0d9e2c4be1ec8eb45171da",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/4237",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:17c39e39005b81792657743de323c62532a47e28cb0d9e2c4be1ec8eb45171da"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:bharatdbms:6e2b38496139",
+      "dbms": "bharatdbms",
+      "title": "bug #19491: segmentation fault triggered by is null",
+      "reported_date": "2026-05-26",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "BharatDBPG",
+      "links": {
+        "report": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/6154"
+      },
+      "primary_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/6154",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/6154",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer=# CREATE SCHEMA IF NOT EXISTS poc;  \nCREATE SCHEMA",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:fb413a26aa3ecd013d85e3d8a14769a27a24f2b198cbbe52264ae60e01e49078",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/BharatDBPG/BharatDBMS-PG/issues/6154",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fb413a26aa3ecd013d85e3d8a14769a27a24f2b198cbbe52264ae60e01e49078"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:citus:044ffca710c2",
+      "dbms": "citus",
+      "title": "Citus unnecessarily errors out for some expressions on WHERE clause",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "report": "https://github.com/citusdata/citus/issues/3981"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/citusdata/citus/issues/3981",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its citus provider, as the field bug3981 that works around it.",
+            "content_sha256": "sha256:044ffca710c2c57a50d7aee3b0775e31e6a3e20e3752b8a21ede3a7b20637282",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:044ffca710c2c57a50d7aee3b0775e31e6a3e20e3752b8a21ede3a7b20637282"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/3981",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:citus:69a3075c66e4",
+      "dbms": "citus",
+      "title": "Citus unnecessarily errors out for some expressions with COLLATION on target list",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "report": "https://github.com/citusdata/citus/issues/3982"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/citusdata/citus/issues/3982",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its citus provider, as the field bug3982 that works around it.",
+            "content_sha256": "sha256:69a3075c66e4437f5cda5aa2553d3667b2e83f3d5cf0228b975e78aeaa3ae41c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:69a3075c66e4437f5cda5aa2553d3667b2e83f3d5cf0228b975e78aeaa3ae41c"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/3982",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:citus:412a9ba2a4ef",
+      "dbms": "citus",
+      "title": "Columnar ON CONFLICT DO NOTHING fails on tables with PK",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "report": "https://github.com/citusdata/citus/issues/6298"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/citusdata/citus/issues/6298",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its citus provider, as the field bug6298 that works around it.",
+            "content_sha256": "sha256:412a9ba2a4ef364202112dd29d8f893857e1dc67e90a8aa0fc6f4a521a6afc6c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:412a9ba2a4ef364202112dd29d8f893857e1dc67e90a8aa0fc6f4a521a6afc6c"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/6298",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:citus:52d7f7c4e9c8",
+      "dbms": "citus",
+      "title": "Error in propagating commands to workers in repartition joins",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "report": "https://github.com/citusdata/citus/issues/4079"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/citusdata/citus/issues/4079",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its citus provider, as the field bug4079 that works around it.",
+            "content_sha256": "sha256:52d7f7c4e9c897ec72c7d814a54e8a092c181c345adc81e2ff5cdecfa6934f1e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:52d7f7c4e9c897ec72c7d814a54e8a092c181c345adc81e2ff5cdecfa6934f1e"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/4079",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:citus:8677c1f0aa31",
+      "dbms": "citus",
+      "title": "Wrong results: WHERE on inheritance parent column drops all rows when cross-joined with a distributed table through LEFT JOIN ... ON FALSE",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "report": "https://github.com/citusdata/citus/issues/8553"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/citusdata/citus/issues/8553",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its citus provider, as the field bug8553 that works around it.",
+            "content_sha256": "sha256:8677c1f0aa31390230fddb4dcec56993fecc32ce1d46e7b54519a2866e02b450",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:8677c1f0aa31390230fddb4dcec56993fecc32ce1d46e7b54519a2866e02b450"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/8553",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:citus:9eb5a7f42741",
+      "dbms": "citus",
+      "title": "error in propogating select queries with casted columns to workers",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "report": "https://github.com/citusdata/citus/issues/4014"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/citusdata/citus/issues/4014",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its citus provider, as the field bug4014 that works around it.",
+            "content_sha256": "sha256:9eb5a7f427413903790ce17204b87a698a817d5a00f982f39a2110eb38e415b4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:9eb5a7f427413903790ce17204b87a698a817d5a00f982f39a2110eb38e415b4"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/4014",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:citus:a697b1e2ecd9",
+      "dbms": "citus",
+      "title": "negative LIMIT error",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "report": "https://github.com/citusdata/citus/issues/4013"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/citusdata/citus/issues/4013",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its citus provider, as the field bug4013 that works around it.",
+            "content_sha256": "sha256:a697b1e2ecd9c923c22271f6d36f14c195a131d4ef61d99652d496ac58350ff0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:a697b1e2ecd9c923c22271f6d36f14c195a131d4ef61d99652d496ac58350ff0"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/4013",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:citus:e8298a0a6090",
+      "dbms": "citus",
+      "title": "wrongly raised syntax error with INSERT ... VALUES(DEFAULT)",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "report": "https://github.com/citusdata/citus/issues/4019"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/citusdata/citus/issues/4019",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its citus provider, as the field bug4019 that works around it.",
+            "content_sha256": "sha256:e8298a0a60907676ea832f9d3e499ec5de5d50908dd91cb196c2e379b97c0444",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/citus/CitusBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:e8298a0a60907676ea832f9d3e499ec5de5d50908dd91cb196c2e379b97c0444"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/4019",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:citus:9307626ac25e",
+      "dbms": "citus",
+      "title": "ERROR: unrecognized node type: 127 with queries involving COLLATE",
+      "reported_date": "2020-06-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "nukoyluoglu",
+      "links": {
+        "report": "https://github.com/citusdata/citus/issues/3957"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/3957",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/citusdata/citus/issues/3957",
+            "source_type": "github_issue",
+            "excerpt": "The error \"PSQLException: ERROR: unrecognized node type: 127\" shows up with queries involving COLLATE. This error was hit many times by the SQLancer tool, and logs for 2 sample executions are attached including steps to reproduce.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b9a806322fbe2fad72907d2a25db766ab091b6a9430bf2e10cd8608626912e26",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/citusdata/citus/issues/3957",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b9a806322fbe2fad72907d2a25db766ab091b6a9430bf2e10cd8608626912e26"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:citus:920a762e1d58",
+      "dbms": "citus",
+      "title": "Some check constraints on distributed tables fails",
+      "reported_date": "2020-07-06",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "onderkalaci",
+      "links": {
+        "report": "https://github.com/citusdata/citus/issues/3980"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/3980",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/citusdata/citus/issues/3980",
+            "source_type": "github_issue",
+            "excerpt": "Generated by SQLancer:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d30008d44a3228395ef7b55206122662d1051dfb776a76ff5ee3b10236ae1c88",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/citusdata/citus/issues/3980",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d30008d44a3228395ef7b55206122662d1051dfb776a76ff5ee3b10236ae1c88"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:citus:b1466805b08f",
+      "dbms": "citus",
+      "title": "SQLancer \"database is being accessed by other users\"",
+      "reported_date": "2023-01-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/citusdata/citus/issues/6670"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/6670",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/citusdata/citus/issues/6670",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer \"database is being accessed by other users\"",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f531ccea764c8a0345fc060c00569eece5f3a51f75644e7853df58621d7c289f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T17:06:22Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T17:06:22Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/citusdata/citus/issues/6670",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f531ccea764c8a0345fc060c00569eece5f3a51f75644e7853df58621d7c289f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:citus:fe620100ca91",
+      "dbms": "citus",
+      "title": "Release 13.0 Error when using unsupported binary operators in queries",
+      "reported_date": "2025-01-06",
+      "reported_year": 2025,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "m3hm3t",
+      "links": {
+        "report": "https://github.com/citusdata/citus/issues/7835"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/7835",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/citusdata/citus/issues/7835",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer generates queries with unsupported operators like #>>, resulting in syntax errors:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6344cf4eaba5fb0d42d01d62535409fdd5d3ae53906ed6bd601242be73694450",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/citusdata/citus/issues/7835",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6344cf4eaba5fb0d42d01d62535409fdd5d3ae53906ed6bd601242be73694450"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:citus:5d1fe916dabc",
+      "dbms": "citus",
+      "title": "Release 13.0 Improve handling of invalid UTF-8 byte sequences",
+      "reported_date": "2025-01-06",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "m3hm3t",
+      "links": {
+        "report": "https://github.com/citusdata/citus/issues/7833"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/7833",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/citusdata/citus/issues/7833",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer generates invalid UTF-8 byte sequences",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:d9f1a987af351842c41b2242f3f915b1840d643432ea8a251b25aed77e86a29e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:0acda4d40d9d79500552743434ebfc2a772f22d068cfd7ce9e4c5df4996912b7",
+          "classified_at": "2026-09-13T06:23:26Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/citusdata/citus/issues/7833",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d9f1a987af351842c41b2242f3f915b1840d643432ea8a251b25aed77e86a29e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:citus:d4e270bfab8d",
+      "dbms": "citus",
+      "title": "TLP test fail on sqlancer due to num_nulls interpreted differently on workers",
+      "reported_date": "2026-02-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "alperkocatas",
+      "links": {
+        "report": "https://github.com/citusdata/citus/issues/8468"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/8468",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/citusdata/citus/issues/8468",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer found the following error, where a query rewritten based on TLP (Ternary Logic Partitioning) returns different cardinalities. It looks like the problem is caused by the condition ` (num_nulls(t1.c0) IS DISTINCT FROM t1.c0)` being interpreted differently when they are pushed down to workers.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:d62cb22b604dd20641dead29f8f98255e9163fb21c1d95ee92b4a653401d4dcf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/citusdata/citus/issues/8468",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d62cb22b604dd20641dead29f8f98255e9163fb21c1d95ee92b4a653401d4dcf"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:citus:dc875a867428",
+      "dbms": "citus",
+      "title": "sqlancer test fail with COLLATE used in string concatenation",
+      "reported_date": "2026-02-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "alperkocatas",
+      "links": {
+        "report": "https://github.com/citusdata/citus/issues/8469"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/8469",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/citusdata/citus/issues/8469",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer test fail with COLLATE used in string concatenation",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:db01bcc540c65a135e3f86ca9e7560ca5c27983670da2b583d7f6ec0df23c704",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/citusdata/citus/issues/8469",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:db01bcc540c65a135e3f86ca9e7560ca5c27983670da2b583d7f6ec0df23c704"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:citus:668f8a22d2f4",
+      "dbms": "citus",
+      "title": "Bug: DISTINCT over a JSON_EXISTS predicate returns different row counts when the predicate is in WHERE vs. relocated into a derived-table projection (distributed tables)",
+      "reported_date": "2026-08-24",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Manuel-Neuer1",
+      "links": {
+        "report": "https://github.com/citusdata/citus/issues/8792"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/8792",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/citusdata/citus/issues/8792",
+            "source_type": "github_issue",
+            "excerpt": "- Found by an automated equivalence-testing fuzzer (SQLancer-derived DQR oracle).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:58d07194d4c3c1035273aeff5d998a330948ee05ba2136c776f41bdaa93b9246",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/citusdata/citus/issues/8792",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:58d07194d4c3c1035273aeff5d998a330948ee05ba2136c776f41bdaa93b9246"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:citus:ef639c5af26e",
+      "dbms": "citus",
+      "title": "Bug: HAVING `count(*) FILTER (WHERE ref.bool)` is rewritten to `FILTER (WHERE dist.int)` on a two-table distributed join",
+      "reported_date": "2026-09-03",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Manuel-Neuer1",
+      "links": {
+        "report": "https://github.com/citusdata/citus/issues/8823"
+      },
+      "primary_url": "https://github.com/citusdata/citus/issues/8823",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/citusdata/citus/issues/8823",
+            "source_type": "github_issue",
+            "excerpt": "- Found by an automated equivalence-testing fuzzer (SQLancer-derived DQR oracle): relocating a `HAVING` predicate into a derived-table projection must not change the result.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:888aa7c9f49e103b2551baf38e6814edd11593a41d7edffa2e3f1f856b899f27",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/citusdata/citus/issues/8823",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:888aa7c9f49e103b2551baf38e6814edd11593a41d7edffa2e3f1f856b899f27"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:8ce4058114aa",
+      "dbms": "clickhouse",
+      "title": "Unexpected result when comparing IN expression and integer",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/65316"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/65316",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/65316",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER, PRIMARY KEY(c0)) ENGINE=MergeTree();\r\nCREATE TABLE t1(c0 INTEGER, PRIMARY KEY(c0)) ENGINE=MergeTree();\r\nINSERT INTO t0(c0) VALUES (1);\r\nINSERT INTO t1(c0) VALUES (0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3970c664c57e8d53baaba4352530fa1d841795a865f14c7841dc691235064ff2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3970c664c57e8d53baaba4352530fa1d841795a865f14c7841dc691235064ff2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:clickhouse:5aa340b3c49c",
+      "dbms": "clickhouse",
+      "title": "Non-trivial error if Int is used in WHERE expression[DB::Exception: Bad get: has Int64, requested UInt64]",
+      "reported_date": "2020-06-24",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/11905"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/11905",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/11905",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer TLPWhere test run.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e96ead12b19a68c0fd16dc98fb7e5efe1cc3f3a71feb7e72e0c28eed7997efed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/11905",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e96ead12b19a68c0fd16dc98fb7e5efe1cc3f3a71feb7e72e0c28eed7997efed"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:40ccd668466e",
+      "dbms": "clickhouse",
+      "title": "Not obvious error message [Attempt to read after eof: while converting '' to Int32]",
+      "reported_date": "2020-06-24",
+      "reported_year": 2020,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/11906"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/11906",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/11906",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer TLPWhere",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:edaa55c12893c5ea92a491abae339b2ca30942268f9f6699922040464096e259",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/11906",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:edaa55c12893c5ea92a491abae339b2ca30942268f9f6699922040464096e259"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:1749f7f8e037",
+      "dbms": "clickhouse",
+      "title": "HAVING clause can change result of aggregation",
+      "reported_date": "2020-07-07",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/12264"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/12264",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/12264",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer TLP Having oracle https://github.com/sqlancer/sqlancer/pull/39",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:951fcb791c7fad9295f2df774b0cd785d7e59dc24d1268851a76fe56314bca7f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/12264",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:951fcb791c7fad9295f2df774b0cd785d7e59dc24d1268851a76fe56314bca7f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:d84ca4f23ba1",
+      "dbms": "clickhouse",
+      "title": "Invalid number of rows in Chunk column Int32: expected 99, got 5: While executing TinyLog",
+      "reported_date": "2020-07-10",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/12402"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/12402",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/12402",
+            "source_type": "github_issue",
+            "excerpt": "Bug found with SQLancer TLPDistinct oracle.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:31e83625fd252bb0bfcafc32d31da6bbaa51a0b03f4394906cda71b6b99eaafe",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/12402",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:31e83625fd252bb0bfcafc32d31da6bbaa51a0b03f4394906cda71b6b99eaafe"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:2b15888a1a89",
+      "dbms": "clickhouse",
+      "title": "MAX over empty set should return NULL according to standard SQL",
+      "reported_date": "2020-08-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/13894"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/13894",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/13894",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer by ClickHouseTLPAggregateOracle",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f7dbff5236f25deb081469be1788a634be8d4dd0ff1266d1d436d5b71e6cf87c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/13894",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f7dbff5236f25deb081469be1788a634be8d4dd0ff1266d1d436d5b71e6cf87c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:0b676ac904ec",
+      "dbms": "clickhouse",
+      "title": "Logical error: 'Constraint `x0` does not return a value of type UInt8",
+      "reported_date": "2021-01-20",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/19334"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/19334",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/19334",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:69f322b05d12b8599adf1a8fad5ad58a8bb9fc6456a6162dd38ff3fcbafd8766",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/19334",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:69f322b05d12b8599adf1a8fad5ad58a8bb9fc6456a6162dd38ff3fcbafd8766"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:6d4a1a1ff524",
+      "dbms": "clickhouse",
+      "title": "UndefinedBehaviorSanitizer AggregateFunctionAvg.h:153: runtime error: signed integer overflow: ... cannot be represented in type 'long'",
+      "reported_date": "2021-01-20",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/19302"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/19302",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/19302",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:0bcae800f2096c8b463d0aedda385b2c3b036c837b9edd9f0e3caad436c343c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/19302",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0bcae800f2096c8b463d0aedda385b2c3b036c837b9edd9f0e3caad436c343c2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:9bc9d60ce6b7",
+      "dbms": "clickhouse",
+      "title": "Logical error: Block structure mismatch in QueryPipeline::unitePipelines stream: different columns",
+      "reported_date": "2021-04-13",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "songenjie",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/23029"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/23029",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/23029",
+            "source_type": "github_issue",
+            "excerpt": "TLPHaving of Sqlancer test",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4f3ac2084c7edf9b9288622878be92d056bfc35bdee99110d87b1ba5f3ece5db",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/23029",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4f3ac2084c7edf9b9288622878be92d056bfc35bdee99110d87b1ba5f3ece5db"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:c5c004ee768a",
+      "dbms": "clickhouse",
+      "title": "Cannot convert column `less(c0, c1)` because it is non constant in source stream but must be constant in result",
+      "reported_date": "2021-05-11",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/24020"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/24020",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/24020",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://clickhouse-test-reports.s3.yandex.net/20393/6aa7c0fafcca9c1493e6923ef68592ab68b286da/sqlancer_test.html#fail1",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e9f21d7710ff32fcd0e36d6ee0a00997db540cab8832caa9f66d9a397e1e8139",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/24020",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e9f21d7710ff32fcd0e36d6ee0a00997db540cab8832caa9f66d9a397e1e8139"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:4a9bd6be5322",
+      "dbms": "clickhouse",
+      "title": "Sigsegv Interpreters/Aggregator.cpp in memory sanitiser build",
+      "reported_date": "2022-11-09",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/43101"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/43101",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/43101",
+            "source_type": "github_issue",
+            "excerpt": "I run queries with SQLancer and server crashed. Same stack traces are reported from 3 threads but they are the same.\r\nI could not reproduce error by running the same query after server restart. Probably some data race?",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:70d651df078e5111357531755d518ff582858098ca4777119ad11d8d9e116e93",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/43101",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:70d651df078e5111357531755d518ff582858098ca4777119ad11d8d9e116e93"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:e172df59f57a",
+      "dbms": "clickhouse",
+      "title": "[sqlancer] \"Cannot convert NULL value to non-Nullable type\" or corrupted sum() value.",
+      "reported_date": "2022-11-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/43821"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/43821",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/43821",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] \"Cannot convert NULL value to non-Nullable type\" or corrupted sum() value.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3119328902dbca116ff3c3584cea56b1d84a53f69df12b8bddbb99a47f3e17d6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/43821",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3119328902dbca116ff3c3584cea56b1d84a53f69df12b8bddbb99a47f3e17d6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:4e56dd907097",
+      "dbms": "clickhouse",
+      "title": "LOGICAL_ERROR: Arguments of 'lcm' have incorrect data types",
+      "reported_date": "2023-07-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Algunenano",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/52075"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/52075",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/52075",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer -> https://s3.amazonaws.com/clickhouse-test-reports/51692/42a7ec77895c3b3def77192ab7ecb8b3e7692beb/sqlancer__release_.html",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:7a8b87c70235f9f0672c20e0d0b97549ced5007eafefe69b7616f636c0476944",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/52075",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7a8b87c70235f9f0672c20e0d0b97549ced5007eafefe69b7616f636c0476944"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:f6a6d2a65675",
+      "dbms": "clickhouse",
+      "title": "Flaky result of the expression",
+      "reported_date": "2023-12-28",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/58279"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/58279",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/58279",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:1a3dde27dc8af13fb62a8ed582b26624a0edd0edd6deacabae7c6e9b9ec824aa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/58279",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1a3dde27dc8af13fb62a8ed582b26624a0edd0edd6deacabae7c6e9b9ec824aa"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:3247e0a7224e",
+      "dbms": "clickhouse",
+      "title": "Adding condition that is true we get no result from a table",
+      "reported_date": "2024-07-25",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/67156"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/67156",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/67156",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer:\r\nhttps://fiddle.clickhouse.com/407ee2ad-f36f-4ea3-bd52-9e941fd4b537",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:123327e97695992f2f11964ad9cd217ceff78e392f2c21e58d12894ecc934df4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/67156",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:123327e97695992f2f11964ad9cd217ceff78e392f2c21e58d12894ecc934df4"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:78f82ccd7978",
+      "dbms": "clickhouse",
+      "title": "Inconsistent representation of bool values",
+      "reported_date": "2024-09-13",
+      "reported_year": 2024,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/69577"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/69577",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/69577",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 VARCHAR ,PRIMARY KEY(c0)) ENGINE=MergeTree();\r\nCREATE  TABLE  t1(c0 BOOL, c1 INT, PRIMARY KEY(c1)) ENGINE=MergeTree();\r\nINSERT INTO t1(c0, c1) VALUES (FALSE, 1);\r\nINSERT INTO t0(c0) VALUES ('a');",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2357e8c3e602d919ea9617833ab586c96969f4683b177bc9a35e2090d4f54e3d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/69577",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2357e8c3e602d919ea9617833ab586c96969f4683b177bc9a35e2090d4f54e3d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:clickhouse:0dbe5406ecf2",
+      "dbms": "clickhouse",
+      "title": "SQLancer:  No equality condition found in JOIN ON expression",
+      "reported_date": "2025-02-04",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Algunenano",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/75541"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/75541",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/75541",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer:  No equality condition found in JOIN ON expression",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b56218e993e84cbc470d541da10817b0fb8017987de820d6f6e92c55bcbfe717",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:9d9f126549a3dbcb4f518491b0f4a70fe9e68fd01e2ebb5fb17abfc9da93a632",
+          "classified_at": "2026-09-13T06:23:26Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/75541",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b56218e993e84cbc470d541da10817b0fb8017987de820d6f6e92c55bcbfe717"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:d7743a86d1f3",
+      "dbms": "clickhouse",
+      "title": "Crash after Logical error: 'sign's argument does not match the expected data type'.",
+      "reported_date": "2025-02-25",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/76709"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/76709",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/76709",
+            "source_type": "github_issue",
+            "excerpt": "https://s3.amazonaws.com/clickhouse-test-reports/json.html?REF=master&sha=2c1648152aeddb96fc07aa1d3d349ede0a61364f&name_0=MasterCI&name_1=SQLancer%20%28debug%29",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b71ec905872e20e9bd2bc9c5285a2a06ee3e40c6815b2a07c3255fa9e1cc5f83",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/76709",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b71ec905872e20e9bd2bc9c5285a2a06ee3e40c6815b2a07c3255fa9e1cc5f83"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:8a8fd68d6fc8",
+      "dbms": "clickhouse",
+      "title": "Equivalent semi join but the second form produces Cannot clone ReadFromStorage plan step",
+      "reported_date": "2025-05-25",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/80775"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/80775",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/80775",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0 (c0 String) ENGINE = Log() ;\nCREATE TABLE t2 (c0 String, c1 String, c2 String) ENGINE = Log() ;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:182dac35293f97e2b50d687cf9720e503c8851d81c8ba66821978a86c46d7f0c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:182dac35293f97e2b50d687cf9720e503c8851d81c8ba66821978a86c46d7f0c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:clickhouse:01b49330b3e5",
+      "dbms": "clickhouse",
+      "title": "LEFT ANTI JOIN right-table columns contain left key values instead of default values",
+      "reported_date": "2026-03-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "zlareb1",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/99959"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/99959",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/99959",
+            "source_type": "github_issue",
+            "excerpt": "Discovered by SQLancer (TLPDistinct oracle) in PR #98390 - https://s3.amazonaws.com/clickhouse-test-reports/json.html?PR=98390&sha=9eadadb4b32eb05176260d7aec5c82de740bf474&name_0=PR&name_1=SQLancer%20%28amd_debug%29.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:15f34e7a09acf3b671bbf69154a7f8efa1f73c0a2d4125c1a9453337bd72a8ff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/99959",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:15f34e7a09acf3b671bbf69154a7f8efa1f73c0a2d4125c1a9453337bd72a8ff"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:6c4b5523d5be",
+      "dbms": "clickhouse",
+      "title": "MIN/MAX produces non-deterministic results when input contains NaN",
+      "reported_date": "2026-03-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "zlareb1",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/99974"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/99974",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/99974",
+            "source_type": "github_issue",
+            "excerpt": "Discovered by SQLancer (TLPAggregate oracle) in PR https://github.com/ClickHouse/ClickHouse/pull/98390 - [s3.amazonaws.com/clickhouse-test-reports/json.html?PR=98390&sha=9eadadb4b32eb05176260d7aec5c82de740bf474&name_0=PR&name_1=SQLancer%20%28amd_debug%29](https://s3.amazonaws.com/clickhouse-test-reports/json.html?PR=98390&sha=9eadadb4b32eb05176260d7aec5c82de740bf474&name_0=PR&name_1=SQLancer%20%28amd_debug%29).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:6deb1cc598c17db414aa2296ed6851540aa4418b5253b3160732cb4709a7008e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/99974",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6deb1cc598c17db414aa2296ed6851540aa4418b5253b3160732cb4709a7008e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:8cd341ca1160",
+      "dbms": "clickhouse",
+      "title": "`(2147483648 > b) AND 2147483648` is not the same result as `(2147483648 > b)`",
+      "reported_date": "2026-03-30",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/101269"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/101269",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/101269",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:d065016f9e1d9d966403a05333a66a14a196f0081ea3a43cfdb6ba28358af9ee",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/101269",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d065016f9e1d9d966403a05333a66a14a196f0081ea3a43cfdb6ba28358af9ee"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:966a6c5f8508",
+      "dbms": "clickhouse",
+      "title": "WHERE x AND toNullable(N) returns 0 rows for any N >= 256",
+      "reported_date": "2026-04-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/103049"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/103049",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/103049",
+            "source_type": "github_issue",
+            "excerpt": "-- Found by AST fuzzer oracle (TLP WHERE) on master",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:cb2e2c5dc116b37aff0cfeabd643aeb8d5b97e5fd95a75119f287d01315e1060",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/103049",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cb2e2c5dc116b37aff0cfeabd643aeb8d5b97e5fd95a75119f287d01315e1060"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:4748c51bada5",
+      "dbms": "clickhouse",
+      "title": "`QueryConditionCache` is poisoned by AST fuzzer queries — subsequent normal `WHERE` queries return zero rows",
+      "reported_date": "2026-05-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/104203"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/104203",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/104203",
+            "source_type": "github_issue",
+            "excerpt": "the `UNION ALL` of three TLP partitions",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:d08e5d658808901de1661f0269029c88cbb2627b2dc8c5bca31e7c6a42415b6e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:45b0539a555705486efe114186076cfc46d2d352fd6f9861133742a50279c3c1",
+          "classified_at": "2026-09-13T06:23:26Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/104203",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d08e5d658808901de1661f0269029c88cbb2627b2dc8c5bca31e7c6a42415b6e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:1d0d30925e2e",
+      "dbms": "clickhouse",
+      "title": "`WHERE p AND <LowCardinality(Nullable(UInt8)) constant>` returns zero rows on `MergeTree`",
+      "reported_date": "2026-05-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/104393"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/104393",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/104393",
+            "source_type": "github_issue",
+            "excerpt": "## Found by\n\nPR #99980 AST fuzzer SQLancer oracle (Identity WHERE oracle, the `p ≡ p AND 1` invariant): a fuzzer-mutated query of the form",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:ac92c460654d2a3c13a445557354aa4e81dee1afd01050c2e1fa07389046b9fb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/104393",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ac92c460654d2a3c13a445557354aa4e81dee1afd01050c2e1fa07389046b9fb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:62991767d421",
+      "dbms": "clickhouse",
+      "title": "`WHERE x OR x` silently drops rows whose low byte is 0 on non-UInt8 columns",
+      "reported_date": "2026-05-15",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer_pp",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/105009"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/105009",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/105009",
+            "source_type": "github_issue",
+            "excerpt": "## Found by\n\nSQLancer++ (https://github.com/suyZhong/SQLancerPlusPlus) NoREC oracle, 6-hour soak against ClickHouse 26.5.1.658 on branch `nik/sqlancer-pp-check` / PR #104984.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:4b3a61415aa4bc4d5360cd56d971a84c480ab590c4393bf073454f2034125759",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/105009",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4b3a61415aa4bc4d5360cd56d971a84c480ab590c4393bf073454f2034125759"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:6e31e733a60e",
+      "dbms": "clickhouse",
+      "title": "`__bitWrapperFunc` internal assertion (\"It's a bug!\") on Float WHERE predicate against a `TYPE set(N)` skip index",
+      "reported_date": "2026-05-19",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/105355"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/105355",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/105355",
+            "source_type": "github_issue",
+            "excerpt": "Surfaced by sqlancer fuzzing (TLP/PQS oracles) and reduced manually.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c80c751a36b9ceb9c5271c8951508dfbbd9f288dbbe91f01bfa94a55cb4ee9cd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:af67dff7d004359a932f04b81c64f1df2bea803a5e9d9aefcd048eed6521c9a9",
+          "classified_at": "2026-09-13T06:23:26Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/105355",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c80c751a36b9ceb9c5271c8951508dfbbd9f288dbbe91f01bfa94a55cb4ee9cd"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:ba45bf866f7b",
+      "dbms": "clickhouse",
+      "title": "Row policy with a bare-column USING expression: LOGICAL_ERROR \"Duplicate column name ... in row policy actions output\" (26.x regression)",
+      "reported_date": "2026-05-29",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/106099"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/106099",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106099",
+            "source_type": "github_issue",
+            "excerpt": "Found by a SQLancer-style row-policy oracle (`CREATE ROW POLICY ... USING <generated predicate>`), where the generated predicate was occasionally a bare column used as a truthy filter.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:65525df0ba0d6d05e1850c2bfdf4a0846bf0607663acbebd09ecf068a0dd612d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106099",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:65525df0ba0d6d05e1850c2bfdf4a0846bf0607663acbebd09ecf068a0dd612d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:0ecc548d9917",
+      "dbms": "clickhouse",
+      "title": "Wrong results: SummingMergeTree FINAL drops a present row when reading only a summation column that is 0 for it (read-in-order + column pruning)",
+      "reported_date": "2026-05-29",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/106125"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/106125",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106125",
+            "source_type": "github_issue",
+            "excerpt": "Found via SQLancer (SortedUnionLimitBy oracle).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b7b9ca22263dea11b5d26e7663bba908c9622445723c8704c1dec01a47f71b76",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106125",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b7b9ca22263dea11b5d26e7663bba908c9622445723c8704c1dec01a47f71b76"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:88f460cebaf9",
+      "dbms": "clickhouse",
+      "title": "Wrong results: partition pruning with intDiv/divide by a negative constant drops rows for range predicates",
+      "reported_date": "2026-05-29",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "coddtest",
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/106124"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/106124",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106124",
+            "source_type": "github_issue",
+            "excerpt": "Found via SQLancer (CODDTest oracle).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:83653365c57c0ac6319104094098e7636b658015e2d996b477c5eaac48ef7a90",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106124",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:83653365c57c0ac6319104094098e7636b658015e2d996b477c5eaac48ef7a90"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:93689ffd5584",
+      "dbms": "clickhouse",
+      "title": "Wrong results: equality predicate (col = const) drops rows when ORDER BY key is a NaN-producing function (sqrt/log of negatives)",
+      "reported_date": "2026-06-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/106262"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/106262",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106262",
+            "source_type": "github_issue",
+            "excerpt": "Found via SQLancer.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:74e2b9823c13d3e70ea7f7777d50c51b434b818cf8ea82da4c543d9a0e9be0ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106262",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:74e2b9823c13d3e70ea7f7777d50c51b434b818cf8ea82da4c543d9a0e9be0ae"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:efa4d9c31750",
+      "dbms": "clickhouse",
+      "title": "LOGICAL_ERROR \"Join restriction violated\" in JoinOrderOptimizer on comma-join + LEFT JOIN with IS NULL in ON (still reproduces after #89409/#89834)",
+      "reported_date": "2026-06-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/106426"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/106426",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106426",
+            "source_type": "github_issue",
+            "excerpt": "`SELECT count()` instead of `SELECT *` masks it (the columns get pruned before the optimizer reaches the bad path). All-`MergeTree ORDER BY tuple()`, no special settings needed. Found by SQLancer (TLPWhere/NoREC).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:7b5bab626c13d04726b998f738ec17f9214372f53947de9f1d029e831fee6933",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106426",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7b5bab626c13d04726b998f738ec17f9214372f53947de9f1d029e831fee6933"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:f7e659da1e93",
+      "dbms": "clickhouse",
+      "title": "Wrong result: WHERE toStartOfYear(Date32) < const returns 0 rows after merge when column has pre-1970 values",
+      "reported_date": "2026-06-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/106419"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/106419",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106419",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer (TLPGroupBy oracle: a `GROUP BY` partitioned by this predicate lost rows).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:3f9c12b802eebed8c13c4ecd2d5481761a2b51d56bafc804eba4e2bae037b29e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106419",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3f9c12b802eebed8c13c4ecd2d5481761a2b51d56bafc804eba4e2bae037b29e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:4b861f132830",
+      "dbms": "clickhouse",
+      "title": "TLP partitioning over a mixed `RIGHT JOIN` + comma join tree can produce rows from an empty result set",
+      "reported_date": "2026-06-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "123lpygithub",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/106560"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/106560",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106560",
+            "source_type": "github_issue",
+            "excerpt": "TLP partitioning over a mixed `RIGHT JOIN` + comma join tree can produce rows from an empty result set",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f2c42793105a41d6625d545c54acb1f3a3db576a20b3fe5616e0046efaf553d0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106560",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f2c42793105a41d6625d545c54acb1f3a3db576a20b3fe5616e0046efaf553d0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:16dc339ef774",
+      "dbms": "clickhouse",
+      "title": "Wrong results: GROUP BY on the partition key with a WHERE filter is served from _minmax_count_projection and collapses all groups into one",
+      "reported_date": "2026-06-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/106573"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/106573",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106573",
+            "source_type": "github_issue",
+            "excerpt": "Found by [SQLancer](https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e05a3efdbff85bb5d3df15c2e0042e4fe3b3aa970b2b0c9e7ff9937d7619b93c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106573",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e05a3efdbff85bb5d3df15c2e0042e4fe3b3aa970b2b0c9e7ff9937d7619b93c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:e076775f71e6",
+      "dbms": "clickhouse",
+      "title": "Block structure mismatch in `UnionStep` stream when `UNION ALL` branches differ only in `WHERE` and one branch's predicate constant-folds (server abort `LOGICAL_ERROR`)",
+      "reported_date": "2026-06-10",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/106956"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/106956",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106956",
+            "source_type": "github_issue",
+            "excerpt": "Triggered by the oracle's own TLP",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:6e0e2da879847e775a8ded4f29495b9774b42d97cc69c1b2bb10d70b55789b6c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:e6a6a3261b37e670d26a3a607653de170fbda94eda7e391fcc149c3b057fc9a3",
+          "classified_at": "2026-09-13T06:23:26Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106956",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6e0e2da879847e775a8ded4f29495b9774b42d97cc69c1b2bb10d70b55789b6c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:9f3f7ba92b03",
+      "dbms": "clickhouse",
+      "title": "Join reordering returns a different result for a LEFT ANTI / RIGHT SEMI / INNER join chain",
+      "reported_date": "2026-06-10",
+      "reported_year": 2026,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/107073"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/107073",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/107073",
+            "source_type": "github_issue",
+            "excerpt": "Found by [SQLancer](https://github.com/sqlancer/sqlancer) (a join-reorder differential oracle: same chain under `query_plan_optimize_join_order_limit` 10 vs 0 vs `randomize=1`, compared as multisets).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f2ca56e4aa1815095f788eb3059645916e76e8b4d5732471511a01831e2aec8a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/107073",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f2ca56e4aa1815095f788eb3059645916e76e8b4d5732471511a01831e2aec8a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:79a1b9672abe",
+      "dbms": "clickhouse",
+      "title": "`DISTINCT` over `GROUP BY ... WITH CUBE/ROLLUP/GROUPING SETS` returns duplicates (`query_plan_remove_redundant_distinct` over-removal)",
+      "reported_date": "2026-06-10",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "dqp",
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/106921"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/106921",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106921",
+            "source_type": "github_issue",
+            "excerpt": "<https://github.com/ClickHouse/ClickHouse/pull/99980> (DQP oracle, toggling\n`query_plan_enable_optimizations`).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:9a2a8a888e42d7b81cdc901ec01cce4cd8c8c48c022fdd4f3c22a6de08a1ff1c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/106921",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9a2a8a888e42d7b81cdc901ec01cce4cd8c8c48c022fdd4f3c22a6de08a1ff1c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:bfb07b0b26c5",
+      "dbms": "clickhouse",
+      "title": "Wrong results: ORDER BY col DESC LIMIT 1 returns 0 rows — TopK granule skip-index (use_skip_indexes_for_top_k) ignores lightweight-delete mask and prunes the part with live top rows",
+      "reported_date": "2026-06-12",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/107309"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/107309",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/107309",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer (TopK differential oracle). The only setting that matters is `use_skip_indexes_for_top_k`; `use_top_k_dynamic_filtering` and `query_plan_top_k_through_join` are not involved. The trigger is a part whose `minmax` index bounds dominate the `ORDER BY` direction but whose rows are all removed by a lightweight delete — the TopK granule pruning trusts the stale `minmax` instead of the live row set.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b0b19c76fc3510b01e9a8d58f52f13badd8232eb4df784aacec57f8972594d92",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/107309",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b0b19c76fc3510b01e9a8d58f52f13badd8232eb4df784aacec57f8972594d92"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:5063a9512b38",
+      "dbms": "clickhouse",
+      "title": "LOGICAL_ERROR \"Not-ready Set ... for function 'in'\": IN-subquery inside an expression on a PARTITION BY table (still reproduces after #100375)",
+      "reported_date": "2026-06-15",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/107503"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/107503",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/107503",
+            "source_type": "github_issue",
+            "excerpt": "* Found by [SQLancer](https://github.com/sqlancer/sqlancer) (ClickHouse PrewhereEquivalence oracle).\n* Related: #97235 (closed, fixed by #100375) — same \"Not-ready Set\" via the MergeTree read path; #101671 / #102981 / #103029 are the distinct `convertAnyJoinToSemiOrAntiJoin` (join-optimization) path; #102192 is the distinct `buildOrderedSetInplace`-on-subquery-timeout path. The case here is none of those — it is a deterministic partition-key-analysis failure for an `IN` subquery used inside an expression.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:807f216ac96bcc592b467be25bcf842561704b0478dc45ddc4b3c5c0743d5b71",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/107503",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:807f216ac96bcc592b467be25bcf842561704b0478dc45ddc4b3c5c0743d5b71"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:befa0ba651b0",
+      "dbms": "clickhouse",
+      "title": "`RANK()` / `DENSE_RANK()` with an explicit `ROWS` frame silently degenerate into `ROW_NUMBER()`",
+      "reported_date": "2026-07-30",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/112665"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/112665",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/112665",
+            "source_type": "github_issue",
+            "excerpt": "`RANK()` / `DENSE_RANK()` with an explicit `ROWS` frame silently degenerate into `ROW_NUMBER()`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a2312ed2e702daa1fb1fac618d0bc2f560f699f561d16ecbb775047cddcd2221",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/112665",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a2312ed2e702daa1fb1fac618d0bc2f560f699f561d16ecbb775047cddcd2221"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:clickhouse:9690b1fcd6a3",
+      "dbms": "clickhouse",
+      "title": "Unexpected results with a minmax index and a NULL-containing `SOME` comparison",
+      "reported_date": "2026-08-01",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/112905"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/112905",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/112905",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected results with a minmax index and a NULL-containing `SOME` comparison",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7095ced78a9f2c5d01933a0313901798959b5f034849bc27758878c0080fd7ce",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/112905",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7095ced78a9f2c5d01933a0313901798959b5f034849bc27758878c0080fd7ce"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:clickhouse:d33733902bc8",
+      "dbms": "clickhouse",
+      "title": "Code 10 `NOT_FOUND_COLUMN_IN_BLOCK` for a right-side join column inside a lambda in `PREWHERE`",
+      "reported_date": "2026-08-10",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/114206"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/114206",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/114206",
+            "source_type": "github_issue",
+            "excerpt": "Code 10 `NOT_FOUND_COLUMN_IN_BLOCK` for a right-side join column inside a lambda in `PREWHERE`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ba20614cb1f3ab534d3901d7061c7e40e9d9b2f0f90183068320867044947025",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/114206",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ba20614cb1f3ab534d3901d7061c7e40e9d9b2f0f90183068320867044947025"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:clickhouse:c4867713b4ee",
+      "dbms": "clickhouse",
+      "title": "LOGICAL_ERROR (Code 49) in a three-way join with a VIEW",
+      "reported_date": "2026-08-10",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/114113"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/114113",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/114113",
+            "source_type": "github_issue",
+            "excerpt": "LOGICAL_ERROR (Code 49) in a three-way join with a VIEW",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a15b3634e2d76a39d28fee0d6e4a1a8f4a6be418c90778105eb2006ea4ad9f26",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/114113",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a15b3634e2d76a39d28fee0d6e4a1a8f4a6be418c90778105eb2006ea4ad9f26"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:clickhouse:22eb557ac570",
+      "dbms": "clickhouse",
+      "title": "NightlySQLancer: LIMIT 1 BY returned the same key twice under ORDER BY (not reproduced)",
+      "reported_date": "2026-08-16",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "alexey-milovidov",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/115033"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/115033",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/115033",
+            "source_type": "github_issue",
+            "excerpt": "Report: https://s3.amazonaws.com/clickhouse-test-reports/json.html?REF=master&sha=5c9f28ed5a2c585ad46bf74d6600cdcf3aa07a9e&name_0=NightlySQLancer&name_1=SQLancer%20%28arm_asan_ubsan%29&name_1=SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ea9d0c12c509be587bd5b19bea6e0748b1788e70efb6b1cd7edc4460f6319f7f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/115033",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ea9d0c12c509be587bd5b19bea6e0748b1788e70efb6b1cd7edc4460f6319f7f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:1971ad7e050c",
+      "dbms": "clickhouse",
+      "title": "Wrong results: per-partition `GROUP BY` / `DISTINCT` / window skip the cross-partition merge for `date_col - INTERVAL n YEAR`, because `isInjectiveFunction` claims `minus` is injective",
+      "reported_date": "2026-09-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fm4v",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/118729"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/118729",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/118729",
+            "source_type": "github_issue",
+            "excerpt": "`NightlySQLancer` run of 2026-09-04, https://github.com/ClickHouse/ClickHouse/actions/runs/33845081523, job `SQLancer (arm_asan_ubsan)`, `TLPDistinct` oracle:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f637e236893ca73108c73689f0cf07cf6b537e720e7a06e53ee240cd61097f9f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/118729",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f637e236893ca73108c73689f0cf07cf6b537e720e7a06e53ee240cd61097f9f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:720d56041a6f",
+      "dbms": "clickhouse",
+      "title": "AST fuzzer: TLP Aggregate oracle mismatch with `STREAM BOUNDED UNORDERED CURSOR`",
+      "reported_date": "2026-09-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "JasonLi-cn",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/119420"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/119420",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/119420",
+            "source_type": "github_issue",
+            "excerpt": "Since `STREAM ... CURSOR` is experimental (`enable_streaming_queries`), the TLP oracle likely needs to account for cursor non-determinism, or the cursor implementation mishandles the decomposed scans.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:459e72ca0cc438fb6df33453de64525286e1da6409cdd9c1c0633d41e06e7c2b",
+            "retrieved_at": "2026-09-12T17:02:58Z",
+            "first_seen": "2026-09-12T17:02:58Z",
+            "last_verified": "2026-09-12T17:02:58Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-12T17:02:58Z",
+        "last_verified": "2026-09-12T17:02:58Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/119420",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:459e72ca0cc438fb6df33453de64525286e1da6409cdd9c1c0633d41e06e7c2b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:d419f31d1a17",
+      "dbms": "clickhouse",
+      "title": "Correlated scalar subquery `(SELECT nan_col) IS NULL` returns wrong result under default `query_plan_enable_optimizations` — silent wrong result on plain `MergeTree`",
+      "reported_date": "2026-09-12",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "dqp",
+      "symptom": "unknown",
+      "reporter": "qoega",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/119679"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/119679",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/119679",
+            "source_type": "github_issue",
+            "excerpt": "- Found by the [AST fuzzer + server-side correctness oracles](https://github.com/ClickHouse/ClickHouse/pull/99980) fleet on `complete52-m0911`, DQP oracle (setting `query_plan_enable_optimizations` flipped), inst 02 run 36 at 2026-09-12 19:08 UTC. Reduced to the standalone form above and re-confirmed on master release (19200) and 19220.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f7b319d1499b7a08e0ae13201d730a3f6bd4b66f66c69207674e59a22f9ee5c7",
+            "retrieved_at": "2026-09-13T03:19:40Z",
+            "first_seen": "2026-09-13T03:09:36Z",
+            "last_verified": "2026-09-13T03:19:40Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:09:36Z",
+        "last_verified": "2026-09-13T03:19:40Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/119679",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f7b319d1499b7a08e0ae13201d730a3f6bd4b66f66c69207674e59a22f9ee5c7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:clickhouse:c87024fa68ca",
+      "dbms": "clickhouse",
+      "title": "`STREAM BOUNDED CURSOR` read reports rows to `count()` but returns no column values (`sum`, `GROUP BY` and aggregate states answer as if the table were empty)",
+      "reported_date": "2026-09-12",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "alexey-milovidov",
+      "links": {
+        "report": "https://github.com/ClickHouse/ClickHouse/issues/119599"
+      },
+      "primary_url": "https://github.com/ClickHouse/ClickHouse/issues/119599",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/ClickHouse/ClickHouse/issues/119599",
+            "source_type": "github_issue",
+            "excerpt": "Found by the AST fuzzer's TLP-aggregate oracle, which compares `count()` with `countMerge(countState())`: it reported the mismatch on https://github.com/ClickHouse/ClickHouse/pull/119582, a pull request that only changes how a part-metadata file is loaded, and the mismatch reproduces on a build without that change.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:7cd35dfb85e758cdf176aaf5e3c508f224dadd94846904c033f4d6c98b34490c",
+            "retrieved_at": "2026-09-13T03:09:36Z",
+            "first_seen": "2026-09-12T17:02:58Z",
+            "last_verified": "2026-09-13T03:09:36Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-12T17:02:58Z",
+        "last_verified": "2026-09-13T03:09:36Z",
+        "source_url": "https://github.com/ClickHouse/ClickHouse/issues/119599",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7cd35dfb85e758cdf176aaf5e3c508f224dadd94846904c033f4d6c98b34490c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cloudberry:5015573eb200",
+      "dbms": "cloudberry",
+      "title": "[Bug] abscissa type not supported, from sqlancer test",
+      "reported_date": "2023-11-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "congxuebin",
+      "links": {
+        "report": "https://github.com/apache/cloudberry/issues/317"
+      },
+      "primary_url": "https://github.com/apache/cloudberry/issues/317",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/cloudberry/issues/317",
+            "source_type": "github_issue",
+            "excerpt": "[Bug] abscissa type not supported, from sqlancer test",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bf802aa196ce5354e481e9a93899222acb0f33c9fee7ac0456508aee710d139b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/cloudberry/issues/317",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bf802aa196ce5354e481e9a93899222acb0f33c9fee7ac0456508aee710d139b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cloudberry:22f985f4b33b",
+      "dbms": "cloudberry",
+      "title": "[Bug] Error detected by sqlancer",
+      "reported_date": "2024-08-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "shmiwy",
+      "links": {
+        "report": "https://github.com/apache/cloudberry/issues/594"
+      },
+      "primary_url": "https://github.com/apache/cloudberry/issues/594",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/cloudberry/issues/594",
+            "source_type": "github_issue",
+            "excerpt": "[Bug] Error detected by sqlancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c85f0d50623a96ff251625e38e60705db6e1bc49906031e2762c8780502f5228",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/cloudberry/issues/594",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c85f0d50623a96ff251625e38e60705db6e1bc49906031e2762c8780502f5228"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cloudberry:6f693a1f7f48",
+      "dbms": "cloudberry",
+      "title": "[Bug] Error found by sqlancer",
+      "reported_date": "2024-08-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "shmiwy",
+      "links": {
+        "report": "https://github.com/apache/cloudberry/issues/596"
+      },
+      "primary_url": "https://github.com/apache/cloudberry/issues/596",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/cloudberry/issues/596",
+            "source_type": "github_issue",
+            "excerpt": "[Bug] Error found by sqlancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:bede5ef874bddf6b0bb4defadc477111608c260a7e1b1a431949559a8fe3acde",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/cloudberry/issues/596",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bede5ef874bddf6b0bb4defadc477111608c260a7e1b1a431949559a8fe3acde"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cloudberry:724b37c977e2",
+      "dbms": "cloudberry",
+      "title": "[Bug] ORCA: \"bogus index qualification\" when a filter on an INCLUDE-only index column is pushed into the index qual",
+      "reported_date": "2026-09-03",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "my-ship-it",
+      "links": {
+        "report": "https://github.com/apache/cloudberry/issues/1948"
+      },
+      "primary_url": "https://github.com/apache/cloudberry/issues/1948",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/cloudberry/issues/1948",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:bfa9503eb66aade6d3bd427b0958e142e9b3340519182894e4ae674707ed3840",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/cloudberry/issues/1948",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bfa9503eb66aade6d3bd427b0958e142e9b3340519182894e4ae674707ed3840"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cloudberry:877cb8aab7f1",
+      "dbms": "cloudberry",
+      "title": "[Bug] ORCA: QD segfault in CExtendedStatsProcessor when extended statistics (dependencies) do not cover all filtered columns",
+      "reported_date": "2026-09-03",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "my-ship-it",
+      "links": {
+        "report": "https://github.com/apache/cloudberry/issues/1949"
+      },
+      "primary_url": "https://github.com/apache/cloudberry/issues/1949",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/cloudberry/issues/1949",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:002fdcda044bcce47be467a11540d455e11df1f307282df6b81718de4c8101ff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/cloudberry/issues/1949",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:002fdcda044bcce47be467a11540d455e11df1f307282df6b81718de4c8101ff"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cloudberry:9a173c6e7882",
+      "dbms": "cloudberry",
+      "title": "[Bug] Planner: FailedAssertion \"pselec >= 0.0 && pselec <= 1.0\" in adjust_selectivity_for_nulltest() for OR clause under LEFT JOIN",
+      "reported_date": "2026-09-03",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "my-ship-it",
+      "links": {
+        "report": "https://github.com/apache/cloudberry/issues/1950"
+      },
+      "primary_url": "https://github.com/apache/cloudberry/issues/1950",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/cloudberry/issues/1950",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e2aeed8b96dd1a248962e01f00af737e29cc95988ab6fba49306b37598251f5d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/cloudberry/issues/1950",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e2aeed8b96dd1a248962e01f00af737e29cc95988ab6fba49306b37598251f5d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cnosdb:1be31c6a537c",
+      "dbms": "cnosdb",
+      "title": "Arrow CSV writer should not fail when cannot cast the value",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/cnosdb/CnosDBBugs.java",
+        "report": "https://github.com/apache/arrow-rs/issues/3547"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/cnosdb/CnosDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/apache/arrow-rs/issues/3547",
+            "excerpt_is_verbatim": true,
+            "note": "cnosdb/sqlancer records this bug in its cnosdb provider, as the field BUG3547 that works around it.",
+            "content_sha256": "sha256:1be31c6a537cebf8968ac43f0e2134cc0f3c3a348c00f63f9c266b19a9669bd4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-09T01:27:20Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/cnosdb/CnosDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:1be31c6a537cebf8968ac43f0e2134cc0f3c3a348c00f63f9c266b19a9669bd4"
+      },
+      "primary_url": "https://github.com/apache/arrow-rs/issues/3547",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cnosdb:b7306cf502ea",
+      "dbms": "cnosdb",
+      "title": "Planning to upgrade datafusion version",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/cnosdb/CnosDBBugs.java",
+        "report": "https://github.com/cnosdb/cnosdb/issues/786"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/cnosdb/CnosDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cnosdb/cnosdb/issues/786",
+            "excerpt_is_verbatim": true,
+            "note": "cnosdb/sqlancer records this bug in its cnosdb provider, as the field BUG786 that works around it.",
+            "content_sha256": "sha256:b7306cf502ea8e79ccce48d2dc0f905f61ce4668b42e887094f5db00a7ac9a49",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-09T01:27:20Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/cnosdb/CnosDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:b7306cf502ea8e79ccce48d2dc0f905f61ce4668b42e887094f5db00a7ac9a49"
+      },
+      "primary_url": "https://github.com/cnosdb/cnosdb/issues/786",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cnosdb:fa664d9284ad",
+      "dbms": "cnosdb",
+      "title": "[BUG] Execute sql query SELECT SUM(1025561311408145024) FROM m3; , db returns error.",
+      "reported_date": "2023-04-10",
+      "reported_year": 2023,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Benxiaohai001",
+      "links": {
+        "report": "https://github.com/cnosdb/cnosdb/issues/1085"
+      },
+      "primary_url": "https://github.com/cnosdb/cnosdb/issues/1085",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/cnosdb/issues/1085",
+            "source_type": "github_issue",
+            "excerpt": "https://github.com/cnosdb/cnosdb/blob/main/query_server/test/cases/sqlancer/function.sql\r\nAutomated Execution Records：",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0e29670f9df35e48f8724ed25b53bd0bb5bcf04a07c26f4f142587f450199113",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/cnosdb/issues/1085",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0e29670f9df35e48f8724ed25b53bd0bb5bcf04a07c26f4f142587f450199113"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cnosdb:572cbca2c791",
+      "dbms": "cnosdb",
+      "title": "[BUG]Inconsistent information returned by distributed and single nodes",
+      "reported_date": "2023-06-14",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Benxiaohai001",
+      "links": {
+        "report": "https://github.com/cnosdb/cnosdb/issues/1270"
+      },
+      "primary_url": "https://github.com/cnosdb/cnosdb/issues/1270",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/cnosdb/issues/1270",
+            "source_type": "github_issue",
+            "excerpt": "Case: function at /data/ce/cnosdb/query_server/test/cases/sqlancer begin.\r\n\tfunction: 100/311",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d7413b086ab3883867f0c36e01423edc453a87f2a8ad2b63517054e26ee61239",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/cnosdb/issues/1270",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d7413b086ab3883867f0c36e01423edc453a87f2a8ad2b63517054e26ee61239"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cnosdb:474ddc25f3ca",
+      "dbms": "cnosdb",
+      "title": "[BUG]In the deployment mode where storage and computing are separated, CI of some use cases cannot pass",
+      "reported_date": "2023-07-07",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Benxiaohai001",
+      "links": {
+        "report": "https://github.com/cnosdb/cnosdb/issues/1335"
+      },
+      "primary_url": "https://github.com/cnosdb/cnosdb/issues/1335",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/cnosdb/issues/1335",
+            "source_type": "github_issue",
+            "excerpt": "Deployment form of separation of storage and calculation, execution of the use case, executed three times before and after, the abnormal use cases are 'select at /__w/cnosdb/cnosdb/query_server/test/cases/sqlancer'",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bf891b8292ed1ed890b31cec2e3c60738743ab1194af7e094862ac116ac579b5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/cnosdb/issues/1335",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bf891b8292ed1ed890b31cec2e3c60738743ab1194af7e094862ac116ac579b5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cnosdb:516da34d353e",
+      "dbms": "cnosdb",
+      "title": "[BUG]During data writing, the insert statement deadlocks",
+      "reported_date": "2024-06-26",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "Benxiaohai001",
+      "links": {
+        "report": "https://github.com/cnosdb/cnosdb/issues/2205"
+      },
+      "primary_url": "https://github.com/cnosdb/cnosdb/issues/2205",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/cnosdb/issues/2205",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer:\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6e4012dd9dd3da1bda0d614d3bcd52c372616456258ffbf3a39e055ed82e6e52",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/cnosdb/issues/2205",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6e4012dd9dd3da1bda0d614d3bcd52c372616456258ffbf3a39e055ed82e6e52"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cnosdb:45b883cd9690",
+      "dbms": "cnosdb",
+      "title": "[BUG][sqlancer]Query results are inconsistent",
+      "reported_date": "2024-06-27",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Benxiaohai001",
+      "links": {
+        "report": "https://github.com/cnosdb/cnosdb/issues/2210"
+      },
+      "primary_url": "https://github.com/cnosdb/cnosdb/issues/2210",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/cnosdb/issues/2210",
+            "source_type": "github_issue",
+            "excerpt": "[BUG][sqlancer]Query results are inconsistent",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:11ce401ea2e69d4aa6b11aab7d822a3ac1b1cbace893361f633a7b6a85bfbbda",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/cnosdb/issues/2210",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:11ce401ea2e69d4aa6b11aab7d822a3ac1b1cbace893361f633a7b6a85bfbbda"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cnosdb:f5de78d50b0a",
+      "dbms": "cnosdb",
+      "title": "DROP DATABASE returns EAGAIN (\"Tskv: Index: index storage error: Resource temporarily unavailable (os error 11)\") under sustained DDL load",
+      "reported_date": "2026-04-26",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/cnosdb/cnosdb/issues/2435"
+      },
+      "primary_url": "https://github.com/cnosdb/cnosdb/issues/2435",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/cnosdb/issues/2435",
+            "source_type": "github_issue",
+            "excerpt": "git clone https://github.com/sqlancer/sqlancer && cd sqlancer\nmvn -B package -DskipTests=true",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:aba10c0bfad140955fe909e4b2ab60a616e963f61965d35b03096c7b906b9ce6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/cnosdb/issues/2435",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:aba10c0bfad140955fe909e4b2ab60a616e963f61965d35b03096c7b906b9ce6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:d164f6c1f377",
+      "dbms": "cockroachdb",
+      "title": "ERROR: no builtin aggregate for SUM_INT on [unknown]",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/83874"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/83874",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its cockroachdb provider, as the field bug83874 that works around it.",
+            "content_sha256": "sha256:d164f6c1f377bb0d68cd1423cf9aa21f925bb45009cbab17283c52c489e334db",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:d164f6c1f377bb0d68cd1423cf9aa21f925bb45009cbab17283c52c489e334db"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/83874",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:a1e3834f01d7",
+      "dbms": "cockroachdb",
+      "title": "Invalid Memory Address Error of Specific SQL Query",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/83973"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/83973",
+            "excerpt_is_verbatim": true,
+            "note": "oceanbase/sqlancer records this bug in its cockroachdb provider, as the field bug83973 that works around it.",
+            "content_sha256": "sha256:a1e3834f01d7154b928d22146752eab00d61a524b505b60edd2c5c86eabf18e3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:a1e3834f01d7154b928d22146752eab00d61a524b505b60edd2c5c86eabf18e3"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/83973",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:b956c79f798f",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Crash in Multiple Table Join",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/85371"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/85371",
+            "excerpt_is_verbatim": true,
+            "note": "oceanbase/sqlancer records this bug in its cockroachdb provider, as the field bug85371 that works around it.",
+            "content_sha256": "sha256:b956c79f798f9c58eddcf5178773dbd00b5d746e04fe481be6459f77ccf10f91",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:b956c79f798f9c58eddcf5178773dbd00b5d746e04fe481be6459f77ccf10f91"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85371",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:ba718d984a57",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Error of Unique Index",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/83976"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/83976",
+            "excerpt_is_verbatim": true,
+            "note": "oceanbase/sqlancer records this bug in its cockroachdb provider, as the field bug83976 that works around it.",
+            "content_sha256": "sha256:ba718d984a57cef1eb55b8bd57eefbada7cdb9d9c6e2890f0f4dfcb6191adcbf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:ba718d984a57cef1eb55b8bd57eefbada7cdb9d9c6e2890f0f4dfcb6191adcbf"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/83976",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:cb8ea9a90e94",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Overflow Error by Huge Interval Value",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/84154"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "The following bug is closed, but leave it enabled until\nthe underlying interval issue is resolved.\nhttps://github.com/cockroachdb/cockroach/issues/84078\nhttps://github.com/cockroachdb/cockroach/issues/84154",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its cockroachdb provider, as the field bug84154 that works around it.",
+            "content_sha256": "sha256:ff59d08c2b519364c64619ec4e35ebfb7d1befefe07b4a406ee033aca5003e05",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:ff59d08c2b519364c64619ec4e35ebfb7d1befefe07b4a406ee033aca5003e05"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/84154",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:23e00e995d45",
+      "dbms": "cockroachdb",
+      "title": "cannot cast jsonb numeric to type bool",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/85441"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/85441",
+            "excerpt_is_verbatim": true,
+            "note": "oceanbase/sqlancer records this bug in its cockroachdb provider, as the field bug85441 that works around it.",
+            "content_sha256": "sha256:23e00e995d457db57b62d88ebbdd0c78d7a6aecd737cd9780b947ec3dedf5a92",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:23e00e995d457db57b62d88ebbdd0c78d7a6aecd737cd9780b947ec3dedf5a92"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85441",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:1688a0adbfc2",
+      "dbms": "cockroachdb",
+      "title": "costfuzz: expected required columns to be a subset of output columns in projectBuildProvided",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/88037"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/88037\nTODO: This should be fixed in v22.2.1.",
+            "excerpt_is_verbatim": true,
+            "note": "oceanbase/sqlancer records this bug in its cockroachdb provider, as the field bug88037 that works around it.",
+            "content_sha256": "sha256:55ed33f4dce5dd57ce6b321f4af2fdc538dcedfefc34435698e2bc5bd18ea045",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-09T01:27:20Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-09T01:27:20Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:55ed33f4dce5dd57ce6b321f4af2fdc538dcedfefc34435698e2bc5bd18ea045"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/88037",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:ab84dbb1120b",
+      "dbms": "cockroachdb",
+      "title": "estimated rows from `EXPLAIN SELECT` incorrectly reduced by DISTINCT",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/131875"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/131875",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its cockroachdb provider, as the field bug131875 that works around it.",
+            "content_sha256": "sha256:ab84dbb1120b3920d54b798e8d4b8d3841bd8b99abeaf6a28b334fd2929d3fcf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:ab84dbb1120b3920d54b798e8d4b8d3841bd8b99abeaf6a28b334fd2929d3fcf"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/131875",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:b81e18b92db6",
+      "dbms": "cockroachdb",
+      "title": "internal error: comparison overload not found (ge, unknown, unknown)",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/128889"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/128889",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/128889",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c1 INTEGER); \r\nCREATE VIEW v0(c0) AS SELECT ('a'||NULL) FROM t1;\r\nSELECT * FROM v0 WHERE (v0.c0 BETWEEN (CASE v0.c0 WHEN v0.c0 THEN v0.c0 END ) AND ('a'));\r\n-- internal error: comparison overload not found (ge, unknown, unknown)",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3fea06209e29ebc23c623c9a0905c490ef8bf579eb0ade9c1b79437bc0906dd8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3fea06209e29ebc23c623c9a0905c490ef8bf579eb0ade9c1b79437bc0906dd8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:677ca7d124ab",
+      "dbms": "cockroachdb",
+      "title": "opt: Inconsistent Case Return Types Decimal Int",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/85356"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/85356",
+            "excerpt_is_verbatim": true,
+            "note": "oceanbase/sqlancer records this bug in its cockroachdb provider, as the field bug85356 that works around it.",
+            "content_sha256": "sha256:677ca7d124abe15a898b0787d9f100e117c25d990bf1c60cad72124df1b037e8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:677ca7d124abe15a898b0787d9f100e117c25d990bf1c60cad72124df1b037e8"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85356",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:c0e7e3d82500",
+      "dbms": "cockroachdb",
+      "title": "opt: Internal Error: Comparison Overload not Found",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/83792"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/83792",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its cockroachdb provider, as the field bug83792 that works around it.",
+            "content_sha256": "sha256:c0e7e3d825003b46013fb03d46e29064aff184a253788da49386140860b449d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:c0e7e3d825003b46013fb03d46e29064aff184a253788da49386140860b449d8"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/83792",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:33c7e7daac74",
+      "dbms": "cockroachdb",
+      "title": "opt: internal error: estimated row count must be non-zero",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/85499"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/85499",
+            "excerpt_is_verbatim": true,
+            "note": "oceanbase/sqlancer records this bug in its cockroachdb provider, as the field bug85499 that works around it.",
+            "content_sha256": "sha256:33c7e7daac74159b359102b5e3720b22a8343b9e04da595d60e9295d8e261ff7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:33c7e7daac74159b359102b5e3720b22a8343b9e04da595d60e9295d8e261ff7"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85499",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:cf9c96ef5229",
+      "dbms": "cockroachdb",
+      "title": "opt: internal error: lookup for ComparisonExpr",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/85390"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/85390",
+            "excerpt_is_verbatim": true,
+            "note": "oceanbase/sqlancer records this bug in its cockroachdb provider, as the field bug85390 that works around it.",
+            "content_sha256": "sha256:cf9c96ef5229dcf770302d665f178e432971ef3bf66a7d42add10bccd212d6a4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:cf9c96ef5229dcf770302d665f178e432971ef3bf66a7d42add10bccd212d6a4"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85390",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:e64cba00bd6c",
+      "dbms": "cockroachdb",
+      "title": "opt: internal error: no output column equivalent to 2",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/85393"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/85393",
+            "excerpt_is_verbatim": true,
+            "note": "oceanbase/sqlancer records this bug in its cockroachdb provider, as the field bug85393 that works around it.",
+            "content_sha256": "sha256:e64cba00bd6c4edb8816cad9aa41b1d2e30f34790b9667835a1c5bd8e2e3813d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:e64cba00bd6c4edb8816cad9aa41b1d2e30f34790b9667835a1c5bd8e2e3813d"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85393",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:5c430d62bd50",
+      "dbms": "cockroachdb",
+      "title": "opt: internal error: no volatility for cast decimal::timestamp",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/85389"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/85389",
+            "excerpt_is_verbatim": true,
+            "note": "oceanbase/sqlancer records this bug in its cockroachdb provider, as the field bug85389 that works around it.",
+            "content_sha256": "sha256:5c430d62bd50e12a814a34e46865127febeffadc588d8c2ccaff8f313c06da70",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:5c430d62bd50e12a814a34e46865127febeffadc588d8c2ccaff8f313c06da70"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85389",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:a69ffcc7159f",
+      "dbms": "cockroachdb",
+      "title": "sql: Inconsistent estimated rows from WHERE clause",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/131640"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/131640",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its cockroachdb provider, as the field bug131640 that works around it.",
+            "content_sha256": "sha256:a69ffcc7159f2e2d01b72b4a3bbcadf2b89747a8b8d0dc0d14053a527257aefc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:a69ffcc7159f2e2d01b72b4a3bbcadf2b89747a8b8d0dc0d14053a527257aefc"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/131640",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:46b5a0205f7e",
+      "dbms": "cockroachdb",
+      "title": "sql: Unexpected Error in SHOW COLUMNS with collated string",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/85394"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/85394",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its cockroachdb provider, as the field bug85394 that works around it.",
+            "content_sha256": "sha256:46b5a0205f7eb912b6ef788d894736d7caee10ab98eb27e4bcd54ed57ea4d75a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:46b5a0205f7eb912b6ef788d894736d7caee10ab98eb27e4bcd54ed57ea4d75a"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85394",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:af1a6ab11ed8",
+      "dbms": "cockroachdb",
+      "title": "sql: inconsistent estimated rows with JOINs",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/131647"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/131647",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its cockroachdb provider, as the field bug131647 that works around it.",
+            "content_sha256": "sha256:af1a6ab11ed8c4fd75cde0cf2e7cf55360f241f200b9cab47491f8c98296becb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:af1a6ab11ed8c4fd75cde0cf2e7cf55360f241f200b9cab47491f8c98296becb"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/131647",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:d0c084590770",
+      "dbms": "cockroachdb",
+      "title": "sql: internal error on collated string array",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/MonetDB/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/45703"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/45703",
+            "excerpt_is_verbatim": true,
+            "note": "MonetDB/sqlancer records this bug in its cockroachdb provider, as the field bug45703 that works around it.",
+            "content_sha256": "sha256:d0c0845907706153be481794d733d581c4787d113ebd3426b56af0143ec88fc0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-09T01:27:20Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-09T01:27:20Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/sqlancer/blob/master/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:d0c0845907706153be481794d733d581c4787d113ebd3426b56af0143ec88fc0"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/45703",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:c3786f943fe4",
+      "dbms": "cockroachdb",
+      "title": "sql: truncate fails when an index is referenced by a view",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "report": "https://github.com/cockroachdb/cockroach/issues/85230"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/cockroachdb/cockroach/issues/85230",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its cockroachdb provider, as the field bug85230 that works around it.",
+            "content_sha256": "sha256:c3786f943fe41e81e9b72e44abe66b2c048aeb77e7b225ab3c98944973cf4d37",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/cockroachdb/CockroachDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:c3786f943fe41e81e9b72e44abe66b2c048aeb77e7b225ab3c98944973cf4d37"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85230",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:cockroachdb:2a169d47fd23",
+      "dbms": "cockroachdb",
+      "title": "Complex ORDER BY clause results in internal error with VECTORIZE=experimental_on",
+      "reported_date": "2020-01-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44624",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44660"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44624",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44624\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44660\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:fe1905b2580d5df84b03a6a3a2d028a0045887d740718d8e26b5e29f07d3b372",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:fe1905b2580d5df84b03a6a3a2d028a0045887d740718d8e26b5e29f07d3b372"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44624",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:c16d77ec1342",
+      "dbms": "cockroachdb",
+      "title": "Internal error for ILIKE_ESCAPE and special characters",
+      "reported_date": "2020-01-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44621",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44633"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44621",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44621\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44633\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:28931191ad177124d331ac8fc718ece52a2ca5c848c3567771bd1e7e89d85c8a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:28931191ad177124d331ac8fc718ece52a2ca5c848c3567771bd1e7e89d85c8a"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44621",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:8c8970251d05",
+      "dbms": "cockroachdb",
+      "title": "Internal error for SUBSTRING with negative length and VECTORIZE=experimental_on",
+      "reported_date": "2020-01-02",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44625"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44625",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44625\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:872891f925e54b0ed55036f707f68cd08b46b459daa1603c86affa25b0c576ea",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:872891f925e54b0ed55036f707f68cd08b46b459daa1603c86affa25b0c576ea"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44625",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:d98f9dba6297",
+      "dbms": "cockroachdb",
+      "title": "Internal error for arithmetic operators and NULLIF",
+      "reported_date": "2020-01-03",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44632",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44718"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44632",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"03/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44632\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44718\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:4c82930a6c1cd4a960f117ab4a61f8bd11e1e69e5b85cb31619dcd6cb45cba6a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4c82930a6c1cd4a960f117ab4a61f8bd11e1e69e5b85cb31619dcd6cb45cba6a"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44632",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:b595dfa92346",
+      "dbms": "cockroachdb",
+      "title": "Internal error for case expression involving NULLIF and VECTORIZE=experimental_on",
+      "reported_date": "2020-01-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44726",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44756"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44726",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44726\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44756\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:b3abb287d34c1a07e289b1fb1a7a5d1c33977bdd67f196f10c6b9733868bec66",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b3abb287d34c1a07e289b1fb1a7a5d1c33977bdd67f196f10c6b9733868bec66"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44726",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:0b3858185513",
+      "dbms": "cockroachdb",
+      "title": "Internal error for query with negative LIMIT on view with negative LIMIT",
+      "reported_date": "2020-01-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44683",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/45009"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44683",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44683\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/45009\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a8cf076425cb82ab8c5e3ecaaa9aca2e9eaeaa1affad59d917b48af1b2dd080c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a8cf076425cb82ab8c5e3ecaaa9aca2e9eaeaa1affad59d917b48af1b2dd080c"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44683",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:4b699f8a8769",
+      "dbms": "cockroachdb",
+      "title": "FULL JOIN with an SUBSTRING ON clause results in an internal error",
+      "reported_date": "2020-01-05",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44746",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44788"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44746",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44746\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44788\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f127e22d5820aee006a4b3acbdbb0a7135925b1e9529dfdbe3f82ba8e4877455",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f127e22d5820aee006a4b3acbdbb0a7135925b1e9529dfdbe3f82ba8e4877455"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44746",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:b8ad1d607926",
+      "dbms": "cockroachdb",
+      "title": "Incorrect result for LIKE query",
+      "reported_date": "2020-01-18",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44123"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44123",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"18/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44123\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:f66fa6a204b08a20ae9ac0c031d919a37e51c512dc283c63eb032c086de73174",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f66fa6a204b08a20ae9ac0c031d919a37e51c512dc283c63eb032c086de73174"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44123",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:917e270ae274",
+      "dbms": "cockroachdb",
+      "title": "Crash when using VECTORIZE=experimental_on",
+      "reported_date": "2020-01-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44133",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44144"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44133",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44133\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44144\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:3f9d86c7a2866442440a72685b6c8d39ca7c1cf2199205b1ce70d553fd53edfc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3f9d86c7a2866442440a72685b6c8d39ca7c1cf2199205b1ce70d553fd53edfc"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44133",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:a75649b620ce",
+      "dbms": "cockroachdb",
+      "title": "Generated column causes query to omit a record in the result set",
+      "reported_date": "2020-01-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44132",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44728"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44132",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44132\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44728\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:bc4e18a69c206e635e25ac090a3b0697226a6d839f1f46505b9e833b0f375355",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:bc4e18a69c206e635e25ac090a3b0697226a6d839f1f46505b9e833b0f375355"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44132",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:9ac0ef300b51",
+      "dbms": "cockroachdb",
+      "title": "DEFAULT value causes unexpected syntax error when executing INSERT",
+      "reported_date": "2020-01-20",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44148",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44775"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44148",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44148\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44775\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:d01423b66ff856aa1add8fc7f27ca0cd139609d7fb3a8b86123c3313231be155",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d01423b66ff856aa1add8fc7f27ca0cd139609d7fb3a8b86123c3313231be155"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44148",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:cf226e86dd2e",
+      "dbms": "cockroachdb",
+      "title": "Internal error in expression that uses a CAST and COLLATE",
+      "reported_date": "2020-01-20",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44137"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44137",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44137\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ff278573bd473343e75fdf04f6ccd75e0bb686b2eeb4e4362e9b1b6308ef5c8c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ff278573bd473343e75fdf04f6ccd75e0bb686b2eeb4e4362e9b1b6308ef5c8c"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44137",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:298741d2c9fb",
+      "dbms": "cockroachdb",
+      "title": "Unexpected error when using EXPERIMENTAL SCRUB",
+      "reported_date": "2020-01-20",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44142",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/45260"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44142",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44142\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/45260\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:2deb698ca36c3f5bf1f2d41d936937aa740140f47c8db01258b59bad1c881789",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2deb698ca36c3f5bf1f2d41d936937aa740140f47c8db01258b59bad1c881789"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44142",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:6bb73dd6759d",
+      "dbms": "cockroachdb",
+      "title": "Incorrect result for BETWEEN SYMMETRIC query",
+      "reported_date": "2020-01-21",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44154",
+        "fix1": "https://github.com/cockroachdb/cockroach/pull/44500",
+        "fix2": "https://github.com/cockroachdb/cockroach/pull/44668"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44154",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44154\",\n            \"fix1\": \"https://github.com/cockroachdb/cockroach/pull/44500\",\n            \"fix2\": \"https://github.com/cockroachdb/cockroach/pull/44668\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:fac7d42b9f94f2bd8813121b95a5267671d0cc019d220e176b4a7e3d0c4dc674",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:fac7d42b9f94f2bd8813121b95a5267671d0cc019d220e176b4a7e3d0c4dc674"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44154",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:3bd57b005ed0",
+      "dbms": "cockroachdb",
+      "title": "Internal error for BETWEEN operator and CAST to BYTES",
+      "reported_date": "2020-01-21",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44181",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44216"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44181",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44181\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44216\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:703897dcc4dd99bdf03e88eb80e1168faf80b83c0da35973e533564b9400c78c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:703897dcc4dd99bdf03e88eb80e1168faf80b83c0da35973e533564b9400c78c"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44181",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:5361afb7e2a2",
+      "dbms": "cockroachdb",
+      "title": "TO_ENGLISH(-9223372036854775808) results in an internal error",
+      "reported_date": "2020-01-21",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44152",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44251"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44152",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44152\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44251\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5e3f8f94b31b4e7e2bc6387789e0c51d6ce01deb0b2cca9b0942e7cbc24cdc0b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5e3f8f94b31b4e7e2bc6387789e0c51d6ce01deb0b2cca9b0942e7cbc24cdc0b"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44152",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:c3f0fb7c27d9",
+      "dbms": "cockroachdb",
+      "title": "Incorrect result for query with IS NULL condition on UNIQUE column and VECTORIZE=experimental_on",
+      "reported_date": "2020-01-22",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44207",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44219"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44207",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44207\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44219\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:d315abe7f6d80d9102a706ac5f92259ed479da333cd04812eab67b9b64d17d95",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d315abe7f6d80d9102a706ac5f92259ed479da333cd04812eab67b9b64d17d95"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44207",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:05f0a1945696",
+      "dbms": "cockroachdb",
+      "title": "Query on VIEW with OFFSET NULL and WHERE condition involving CURRENT_USER() unexpectedly fetches a row",
+      "reported_date": "2020-01-22",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44203",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44307"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44203",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44203\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44307\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:e6796fe297a56608cb7e79a4afcdcd2914d5e878ca6aa7d5a10f1599300e0648",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e6796fe297a56608cb7e79a4afcdcd2914d5e878ca6aa7d5a10f1599300e0648"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44203",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:9e429fd4de3e",
+      "dbms": "cockroachdb",
+      "title": "SHOW EXPERIMENTAL_FINGERPRINTS error for STRING columns and character escapes",
+      "reported_date": "2020-01-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44237"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44237",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44237\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:22bda50b41edb25778cec9d872693154603f37cedea7a1e914944b29bce9d267",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:22bda50b41edb25778cec9d872693154603f37cedea7a1e914944b29bce9d267"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44237",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:9dbc25141602",
+      "dbms": "cockroachdb",
+      "title": "Incorrect result for IS NULL query on VIEW using SELECT DISTINCT",
+      "reported_date": "2020-01-23",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44296",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44386"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44296",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44296\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44386\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:1ac534419c132e18fa871dd174663d3e4bc9b5a311e5f0ff8b5efa82e7e21d62",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1ac534419c132e18fa871dd174663d3e4bc9b5a311e5f0ff8b5efa82e7e21d62"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44296",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:70c137001554",
+      "dbms": "cockroachdb",
+      "title": "Internal error for CASE expression and VECTORIZE=experimental_on",
+      "reported_date": "2020-01-23",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44304",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44346"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44304",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44304\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44346\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:874277507c5e4a00f72651ae2d6dddb49f576bbb2cfe568eb602abfb64a76601",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:874277507c5e4a00f72651ae2d6dddb49f576bbb2cfe568eb602abfb64a76601"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44304",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:58a94dc3069a",
+      "dbms": "cockroachdb",
+      "title": "Internal error setting tracing=true",
+      "reported_date": "2020-01-23",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44244",
+        "fix": "https://github.com/cockroachdb/cockroach/commit/6f7239ba84b5988951d6148f8feba6344cd40c9d"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44244",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44244\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/commit/6f7239ba84b5988951d6148f8feba6344cd40c9d\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:2d7e2e7eacc16bf4b65d3fdb470840ab12a1f86be54f278002f48726181c205d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2d7e2e7eacc16bf4b65d3fdb470840ab12a1f86be54f278002f48726181c205d"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44244",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:a494d404b2ff",
+      "dbms": "cockroachdb",
+      "title": "Deadlock in SHOW TABLES",
+      "reported_date": "2020-01-26",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "hang",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44385",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46384"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44385",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44385\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46384\"\n        },\n        \"oracle\": \"hang\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5c628679f803ca221d9fe16e0a8a76087ad8b86888171b34f55a68bb2525579d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5c628679f803ca221d9fe16e0a8a76087ad8b86888171b34f55a68bb2525579d"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44385",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:fa6977218fe8",
+      "dbms": "cockroachdb",
+      "title": "Internal error \"estimated distinct count must be non-zero\"",
+      "reported_date": "2020-01-27",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44418",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44430"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44418",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"27/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44418\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44430\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:26ca9d07af9e02e3df57b94fb6ac63f5d5c1e8f57a1acb1227c5cb8ce1603358",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:26ca9d07af9e02e3df57b94fb6ac63f5d5c1e8f57a1acb1227c5cb8ce1603358"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44418",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:e575fea7548f",
+      "dbms": "cockroachdb",
+      "title": "Multi-record UPSERT inserts duplicate values in PRIMARY KEY, resulting in inconsistent results",
+      "reported_date": "2020-01-29",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44466",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/45372"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44466",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44466\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/45372\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:12ada2f1c85f43c44d7dcfcfac103513e9b9cfeb4ac0dff36a09aed87458e57a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:12ada2f1c85f43c44d7dcfcfac103513e9b9cfeb4ac0dff36a09aed87458e57a"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44466",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:c210521d3328",
+      "dbms": "cockroachdb",
+      "title": "Internal error for NATURAL JOIN and VECTORIZE='experimental_on'",
+      "reported_date": "2020-01-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44547",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44796"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44547",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44547\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44796\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:28b7b5ab7c60a38fd46ecb45828abd19777a2dd180f83aa34c387dcc2ae8e40b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:28b7b5ab7c60a38fd46ecb45828abd19777a2dd180f83aa34c387dcc2ae8e40b"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44547",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:55299924eb63",
+      "dbms": "cockroachdb",
+      "title": "Server exits on query with LEFT JOIN",
+      "reported_date": "2020-01-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44565",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44590"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44565",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/01/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44565\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44590\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:67a3ddfa10f479f9593411855a489e5ea6dfb38cc571848cbb220fc282c45b1d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:67a3ddfa10f479f9593411855a489e5ea6dfb38cc571848cbb220fc282c45b1d"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44565",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:2a6cc3dfdb20",
+      "dbms": "cockroachdb",
+      "title": "Unexpected error for aggregate functions",
+      "reported_date": "2020-02-05",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44757"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44757",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/02/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44757\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f8725880790b28876af48c968d36d7ffcbd45fbd162d81fcac5263a0298d080a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f8725880790b28876af48c968d36d7ffcbd45fbd162d81fcac5263a0298d080a"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44757",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:0f46cf5671d0",
+      "dbms": "cockroachdb",
+      "title": "NOT BETWEEN query on VIEW results in an internal error with VECTORIZE=experimental_on",
+      "reported_date": "2020-02-06",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44822",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44829"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44822",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"06/02/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44822\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44829\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:0e2d6f3a190b44bb54f097dc547c7adb8886fe6ba05cdf3c1d07b4d58f239581",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0e2d6f3a190b44bb54f097dc547c7adb8886fe6ba05cdf3c1d07b4d58f239581"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44822",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:aff47828e2ba",
+      "dbms": "cockroachdb",
+      "title": "Internal error for SUBSTRING, INT4 cast, and VECTORIZE=experimental_on",
+      "reported_date": "2020-02-07",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44881",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44887"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44881",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/02/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44881\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44887\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5314a8728f8d0a8b27d21dc0064d94883192e6c5dfa840b488790647c88498b1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5314a8728f8d0a8b27d21dc0064d94883192e6c5dfa840b488790647c88498b1"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44881",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:96c1152b79d1",
+      "dbms": "cockroachdb",
+      "title": "EXPERIMENTAL SCRUB TABLE results in an \"overflow during Encode\" error",
+      "reported_date": "2020-02-08",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44891"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44891",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"08/02/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44891\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:940df6d8e8d341b0babecb674a2699052fec0f17cf3b7a7133604b99a8b2db27",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:940df6d8e8d341b0babecb674a2699052fec0f17cf3b7a7133604b99a8b2db27"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44891",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:0d6b46749dcb",
+      "dbms": "cockroachdb",
+      "title": "Internal error when setting VECTORIZE = experimental_on and DEFAULT_INT_SIZE = 4",
+      "reported_date": "2020-02-09",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44904",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44930"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44904",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"09/02/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44904\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44930\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:4e47158857626dcc0c87e4b48f42a498f35f206ed95e1b2de2490f24ac95b0de",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4e47158857626dcc0c87e4b48f42a498f35f206ed95e1b2de2490f24ac95b0de"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44904",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:4a0631a906f5",
+      "dbms": "cockroachdb",
+      "title": "Internal error for NATURAL JOIN on INT and INT4 column for VECTORIZE=experimental_on",
+      "reported_date": "2020-02-11",
+      "reported_year": 2020,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44982",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/44942"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44982",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/02/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44982\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/44942\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:da153a41bcd8ce64f13e239b344adc3e948c6b7a6f8b11a49b675e95c03bc8d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:da153a41bcd8ce64f13e239b344adc3e948c6b7a6f8b11a49b675e95c03bc8d8"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44982",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:fdecf4be5a84",
+      "dbms": "cockroachdb",
+      "title": "EXPERIMENTAL SCRUB TABLE crashes server for table with TIMESTAMP column",
+      "reported_date": "2020-02-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/44992",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/45410"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/44992",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/02/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/44992\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/45410\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:0a1a7ca441d6a0156ee5da79f9909615644ab8ea5e258de4cdd796c94a0b48ab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0a1a7ca441d6a0156ee5da79f9909615644ab8ea5e258de4cdd796c94a0b48ab"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/44992",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:32e00f5b236b",
+      "dbms": "cockroachdb",
+      "title": "Internal error for VECTORIZE=experimental_on, DEFAULT_INT_SIZE=4, and aggregate query",
+      "reported_date": "2020-02-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/45038",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/45042"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/45038",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/02/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/45038\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/45042\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:0b0401bac850e0bef613d209863e7afa0536785a762ab33c5471ef85c61670ee",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0b0401bac850e0bef613d209863e7afa0536785a762ab33c5471ef85c61670ee"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/45038",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:e69ad90289a1",
+      "dbms": "cockroachdb",
+      "title": "Syntax error for multi-valued comparison and COLLATE",
+      "reported_date": "2020-02-17",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/45142"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/45142",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"17/02/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/45142\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1538c0ac8d7d731c2ab34760e905b365c01e48f27a8ef2f97c3a92a6f4c6f293",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1538c0ac8d7d731c2ab34760e905b365c01e48f27a8ef2f97c3a92a6f4c6f293"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/45142",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:cc33fa88ae19",
+      "dbms": "cockroachdb",
+      "title": "COUNT_ROWS() malfunctions for GROUP BY",
+      "reported_date": "2020-02-26",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/45453",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46879"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/45453",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/02/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/45453\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46879\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:5b4ad0b9db14f324d4751ea14a3e1a4c3dcdf51fbcf306f0031710b389add766",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5b4ad0b9db14f324d4751ea14a3e1a4c3dcdf51fbcf306f0031710b389add766"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/45453",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:04320e7d6ce1",
+      "dbms": "cockroachdb",
+      "title": "EXPLAIN (VEC) SELECT results in an internal error \"input to aggregatorBase is not an execinfra.OpNode\"",
+      "reported_date": "2020-03-14",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46123",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46200"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46123",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46123\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46200\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c6b38f5bb1f99c328aec9b622fed713b474d0505059ecc42bf15aceb34449e50",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c6b38f5bb1f99c328aec9b622fed713b474d0505059ecc42bf15aceb34449e50"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46123",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:6f3d07a0c111",
+      "dbms": "cockroachdb",
+      "title": "EXPLAIN (VEC) SELECT results in an internal error \"zero length schema unsupported\"",
+      "reported_date": "2020-03-14",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46122",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46204"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46122",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46122\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46204\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c719996ee52d0a1d49ddb75041155e2a4c02015bb559b55a1b4a4f56fd4c83b5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c719996ee52d0a1d49ddb75041155e2a4c02015bb559b55a1b4a4f56fd4c83b5"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46122",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:8b05bc84350f",
+      "dbms": "cockroachdb",
+      "title": "Incorrect result for MAX, INTERLEAVE IN PARENT, and vectorize=experimental_on",
+      "reported_date": "2020-03-16",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46140",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46494"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46140",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"16/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46140\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46494\"\n        },\n        \"oracle\": \"TLP (aggregate)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:e6864c6c4ced5aa83f9f53974d0ac723cbe651c7eeaf6fc00c93d2e93865dd6b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e6864c6c4ced5aa83f9f53974d0ac723cbe651c7eeaf6fc00c93d2e93865dd6b"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46140",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:355b8a5159c4",
+      "dbms": "cockroachdb",
+      "title": "Query with negative LIMIT results in an internal error \"node lookup-join with MaxCost added to the memo\"",
+      "reported_date": "2020-03-17",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46187",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46440"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46187",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"17/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46187\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46440\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:400f58ddec95296b1abbf1489833fddc61054a254f3e7f7b5973b87de765032e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:400f58ddec95296b1abbf1489833fddc61054a254f3e7f7b5973b87de765032e"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46187",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:71b220e016c9",
+      "dbms": "cockroachdb",
+      "title": "VALUES table expression malfunctions with MAX",
+      "reported_date": "2020-03-17",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/46196"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46196",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46196",
+            "source_type": "github_issue",
+            "excerpt": "VALUES table expression malfunctions with MAX",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:bf46bcc84bf5adc08550cac3002871116e02d5635d47bf5463930ba71a87544b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/46196",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bf46bcc84bf5adc08550cac3002871116e02d5635d47bf5463930ba71a87544b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:469894265e2c",
+      "dbms": "cockroachdb",
+      "title": "DROP DATABASE issue with TEMP tables",
+      "reported_date": "2020-03-21",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46393",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46422"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46393",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46393\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46422\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:0add0cd6f4a0022d0f3726e8cfaf918303ba59e668de074f1234671120379ead",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0add0cd6f4a0022d0f3726e8cfaf918303ba59e668de074f1234671120379ead"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46393",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:46021eca95f5",
+      "dbms": "cockroachdb",
+      "title": "EXPLAIN (VEC) results in an internal error \"input to aggregatorBase is not an execinfra.OpNode\"",
+      "reported_date": "2020-03-21",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46404",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46439"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46404",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46404\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46439\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a934de2fb0c28daa55e767af409e2417d0265e6452b68a717a4a05cc57a8dd8a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a934de2fb0c28daa55e767af409e2417d0265e6452b68a717a4a05cc57a8dd8a"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46404",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:d9e8367b452e",
+      "dbms": "cockroachdb",
+      "title": "INSERT ... DO NOTHING results in an error \"UPSERT or INSERT...ON CONFLICT command cannot affect row a second time\"",
+      "reported_date": "2020-03-21",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46395",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46408"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46395",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46395\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46408\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e0bd31bc5cdfca75dae389fa939e5ebcd02be827d6d4de482b8015d66ee7aee8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e0bd31bc5cdfca75dae389fa939e5ebcd02be827d6d4de482b8015d66ee7aee8"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46395",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:b777b5f3eb1d",
+      "dbms": "cockroachdb",
+      "title": "NATURAL JOIN fails with \"duplicate column name\" on view",
+      "reported_date": "2020-03-21",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46403",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/47099"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46403",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46403\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/47099\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:9831c1aa22d4e3f5acfec99ef8427ac415b8c00a1a221c5cdd9ca7e1b1c71d07",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9831c1aa22d4e3f5acfec99ef8427ac415b8c00a1a221c5cdd9ca7e1b1c71d07"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46403",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:533991b0e719",
+      "dbms": "cockroachdb",
+      "title": "SCRUB fails for table that has a FOREIGN KEY constraint on a rowid column",
+      "reported_date": "2020-03-21",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46401"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46401",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46401\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:08cce119d149c06a683961f992747d30fd442c309f99405ec36ac5ff6c37f551",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:08cce119d149c06a683961f992747d30fd442c309f99405ec36ac5ff6c37f551"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46401",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:a27f9279ade0",
+      "dbms": "cockroachdb",
+      "title": "UPSERT causes a decoding error in collated string column",
+      "reported_date": "2020-03-21",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46394",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46570"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46394",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46394\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46570\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e3e6b85921185fcc4b828c3c64bc8b93af7ba973f441f75a7b59c4e9a9ca0883",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e3e6b85921185fcc4b828c3c64bc8b93af7ba973f441f75a7b59c4e9a9ca0883"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46394",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:a21a6cc66c87",
+      "dbms": "cockroachdb",
+      "title": "UPSERT on table with FOREIGN KEY constraint results in an internal error \"cannot convert int to type bool\"",
+      "reported_date": "2020-03-21",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46397",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46409"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46397",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46397\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46409\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:16aa518e6a348621fae6d90cfa982c487a21634796783bbb058306a5bf073b5a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:16aa518e6a348621fae6d90cfa982c487a21634796783bbb058306a5bf073b5a"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46397",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:9978f0f2d945",
+      "dbms": "cockroachdb",
+      "title": "HAVING clause incorrectly evaluates to TRUE for VARIANCE(0) IS NOT NULL predicate",
+      "reported_date": "2020-03-22",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46412",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46436"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46412",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46412\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46436\"\n        },\n        \"oracle\": \"TLP (HAVING)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:c5661bc299e9e41bce53f3c8a9dcac27b693512158708c5595256893bee8aba8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c5661bc299e9e41bce53f3c8a9dcac27b693512158708c5595256893bee8aba8"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46412",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:5c71eb88b53b",
+      "dbms": "cockroachdb",
+      "title": "Query with HAVING clause, DISTSQL=off, and vectorize=on results in an internal error",
+      "reported_date": "2020-03-24",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46503"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46503",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46503\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ff274280231103306f26e03b3fa41ce638dd44a6245f9698188c39c033e79037",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ff274280231103306f26e03b3fa41ce638dd44a6245f9698188c39c033e79037"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46503",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:88c2aa4d5260",
+      "dbms": "cockroachdb",
+      "title": "Internal error for vectorize=on and multiplication of INT4 values",
+      "reported_date": "2020-03-29",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46714",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/46712"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46714",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/03/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46714\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/46712\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:43bab3c0ae1c27f92e074ef44221f6372111c4f9cc5b324ec4522f35ff3d917e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:43bab3c0ae1c27f92e074ef44221f6372111c4f9cc5b324ec4522f35ff3d917e"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46714",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:775f0a6c8e24",
+      "dbms": "cockroachdb",
+      "title": "Internal error: lookup for ComparisonExpr ((@2)[float] != ('NaN')[string])[bool]'s CmpOp failed",
+      "reported_date": "2020-04-02",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46914",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/50815"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46914",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46914\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/50815\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:dd4e36fea10ec245462255795823f135311fc88ba48a6ccd56da01d63cce8d28",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:dd4e36fea10ec245462255795823f135311fc88ba48a6ccd56da01d63cce8d28"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46914",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:25c111f9a9c8",
+      "dbms": "cockroachdb",
+      "title": "Unexpected syntax error for expression involving ARRAY[NULL]",
+      "reported_date": "2020-04-02",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46915"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46915",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46915\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:38fced19ec2ac02afb532b38e21838180badfcd82b103524087f6a3ad3f2d146",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:38fced19ec2ac02afb532b38e21838180badfcd82b103524087f6a3ad3f2d146"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46915",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:c29d5a1c0770",
+      "dbms": "cockroachdb",
+      "title": "CAST to INT2 yields an incorrect/inconsistent result",
+      "reported_date": "2020-04-03",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46970"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46970",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"03/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46970\"\n        },\n        \"oracle\": \"TLP (aggregate)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:4dc12b63d4c27d5ec17be29df42694b43783f3b35941fb9af13cc8356eb2bc7a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4dc12b63d4c27d5ec17be29df42694b43783f3b35941fb9af13cc8356eb2bc7a"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46970",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:a7e3a7604c29",
+      "dbms": "cockroachdb",
+      "title": "CREATE STATISTICS on table with ARRAY column",
+      "reported_date": "2020-04-03",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46964",
+        "fixed": "https://github.com/cockroachdb/cockroach/pull/47281"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46964",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"03/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46964\",\n            \"fixed\": \"https://github.com/cockroachdb/cockroach/pull/47281\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c5d5daabcad4dae941556fe9963b91b6311762835e05c7844578ebb89d57510d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c5d5daabcad4dae941556fe9963b91b6311762835e05c7844578ebb89d57510d"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46964",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:a4b3fd275513",
+      "dbms": "cockroachdb",
+      "title": "Comparison with (-9223372036854775808)::TIMESTAMP results in an unexpected error",
+      "reported_date": "2020-04-03",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46973",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/47077"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46973",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"03/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46973\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/47077\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:8896b6ce894621a0eb4330b5bcdb497cdcf9c60c3545c146e8235022c28b8027",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8896b6ce894621a0eb4330b5bcdb497cdcf9c60c3545c146e8235022c28b8027"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46973",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:bb021b4e9b81",
+      "dbms": "cockroachdb",
+      "title": "Query with an invalid regular expression unexpectedly does not fail",
+      "reported_date": "2020-04-03",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/46981",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/47036"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/46981",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"03/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/46981\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/47036\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:835f70b5a54726b778755343c234c4aafd6d7a5201d903748867008124ce3068",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:835f70b5a54726b778755343c234c4aafd6d7a5201d903748867008124ce3068"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/46981",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:042edf29e01c",
+      "dbms": "cockroachdb",
+      "title": "COLLATE in array causes an internal error \"ARRAY[]:::STRING[] COLLATE en: incompatible type for COLLATE: string[]\"",
+      "reported_date": "2020-04-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/47026"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47026",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47026",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nSELECT IF(false, ARRAY['' COLLATE en], ARRAY[]) FROM t0; -- ERROR: internal error: ARRAY[]:::STRING[] COLLATE en: incompatible type for COLLATE: string[]\r\n```\r\nUnexpectedly, the `SELECT` results in an internal error with the following stack trace:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d2edc31c56a5c7734d0260cb2ffd061c69c22966df133e5570f980728d74f264",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/47026",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d2edc31c56a5c7734d0260cb2ffd061c69c22966df133e5570f980728d74f264"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:ed36645169fe",
+      "dbms": "cockroachdb",
+      "title": "Incorrect result for query on interleaved index when vectorize=on",
+      "reported_date": "2020-04-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47029",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/47035"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47029",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47029\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/47035\"\n        },\n        \"oracle\": \"TLP (aggregate)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:66bf1753614bad975560266893a3e317088439f814776befaa86190a4c5af72b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:66bf1753614bad975560266893a3e317088439f814776befaa86190a4c5af72b"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47029",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:bb119c57369e",
+      "dbms": "cockroachdb",
+      "title": "Internal error for an invalid CREATE INDEX statement",
+      "reported_date": "2020-04-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47037",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/47090"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47037",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47037\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/47090\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:6abb9c04b1b954eb9b73d34600e17c1659fc3f859d5f62ad141334b05293c720",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6abb9c04b1b954eb9b73d34600e17c1659fc3f859d5f62ad141334b05293c720"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47037",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:62a664d9f21d",
+      "dbms": "cockroachdb",
+      "title": "SCRUB on a temporary table that has a hash-sharded index results in an error",
+      "reported_date": "2020-04-04",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47031"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47031",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47031\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e4ee0ca4b3c4f836a6815550d140017fec4ec6f4853995b05f0dd2833c670889",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e4ee0ca4b3c4f836a6815550d140017fec4ec6f4853995b05f0dd2833c670889"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47031",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:3f8884268107",
+      "dbms": "cockroachdb",
+      "title": "SELECT statement on table with a hash-sharded index results in an internal error",
+      "reported_date": "2020-04-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47041",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/47159"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47041",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47041\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/47159\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:76ff957bd56e6f8cce8baff466d54a0d858c4e325117175b0dcf89559296cd1f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:76ff957bd56e6f8cce8baff466d54a0d858c4e325117175b0dcf89559296cd1f"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47041",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:25f63c8b385f",
+      "dbms": "cockroachdb",
+      "title": "TRUNCATE on temporary table results in \"unexpected value: <nil>\"",
+      "reported_date": "2020-04-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47030",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/47482"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47030",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47030\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/47482\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:4e232fec9389f063fb4e857ad234a6015f94aedded0a5bc36bbd87df1fe6d133",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4e232fec9389f063fb4e857ad234a6015f94aedded0a5bc36bbd87df1fe6d133"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47030",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:dbbf4216d9fc",
+      "dbms": "cockroachdb",
+      "title": "CREATE INDEX results in \"invalid interleave backreference\" and prevents database from being dropped",
+      "reported_date": "2020-04-07",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47122"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47122",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47122\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:23d830f8042841bdfa1eb31125cafe9f2728b226c78a63d65a10c5abd7435c5a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:23d830f8042841bdfa1eb31125cafe9f2728b226c78a63d65a10c5abd7435c5a"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47122",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:40f8ce726651",
+      "dbms": "cockroachdb",
+      "title": "Column that is used both as an INT2 and INT4 results in an internal error",
+      "reported_date": "2020-04-07",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47131",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/47174"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47131",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47131\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/47174\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e3e5049cd56f52ca2ea27a833d068056d46c32c382b3ae700cc209a98655ac09",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e3e5049cd56f52ca2ea27a833d068056d46c32c382b3ae700cc209a98655ac09"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47131",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:3c7e202d37dd",
+      "dbms": "cockroachdb",
+      "title": "Disconnecting from server results in crash with \"unexpected leftover bytes\"",
+      "reported_date": "2020-04-07",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47114"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47114",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47114\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:93b22568ed1033aa8a0a7cd7169a1857944a131dcea370863f71b51392293300",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:93b22568ed1033aa8a0a7cd7169a1857944a131dcea370863f71b51392293300"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47114",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:bae3bae296a4",
+      "dbms": "cockroachdb",
+      "title": "ORDER BY unexpectedly causes a value to be represented in E notation when VECTORIZE=on",
+      "reported_date": "2020-04-07",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47115",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/48052"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47115",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47115\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/48052\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:209535f48798141be4bb65ca2093b680b4f949f2b4b105a0f03c17bceacbbac8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:209535f48798141be4bb65ca2093b680b4f949f2b4b105a0f03c17bceacbbac8"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47115",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:b12edb3cd6ae",
+      "dbms": "cockroachdb",
+      "title": "SCRUB on a table with FLOAT PRIMARY KEY and INTERLEAVE IN PARENT index causes server to exit",
+      "reported_date": "2020-04-07",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47116"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47116",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47116\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:10cb153a77f9749ea61fefdbc73d6a44a6df6b342ec77c101eea6a30f70c8bab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:10cb153a77f9749ea61fefdbc73d6a44a6df6b342ec77c101eea6a30f70c8bab"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47116",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:4d37b75e379e",
+      "dbms": "cockroachdb",
+      "title": "Arithmetic expression results in \"internal error: could not find type for binary expression mult\"",
+      "reported_date": "2020-04-09",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/47299"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47299",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47299",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nSELECT * FROM t0 ORDER BY (CASE WHEN NULL THEN 0 WHEN 0 < t0.c0 THEN NULL END) * (CASE WHEN false THEN 0 WHEN t0.rowid > 0 THEN NULL END); --  could not find type for binary expression mult\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:37eb1e8f973f724c138d7bfc4d8736996fb184407812ccfb7826bc39e8cd5121",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/47299",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:37eb1e8f973f724c138d7bfc4d8736996fb184407812ccfb7826bc39e8cd5121"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:80aeca17d546",
+      "dbms": "cockroachdb",
+      "title": "BETWEEN expression with an INTERVAL and DECIMAL cast results in an incorrect result",
+      "reported_date": "2020-04-10",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cockroachdb/cockroach/issues/47327",
+        "fix": "https://github.com/cockroachdb/cockroach/pull/47483"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/47327",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/04/2020\",\n        \"dbms\": \"CockroachDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cockroachdb/cockroach/issues/47327\",\n            \"fix\": \"https://github.com/cockroachdb/cockroach/pull/47483\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:dd958259839bae6f4ece0c635c20cd2dfeaefd604d068923862c391222c4d512",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:dd958259839bae6f4ece0c635c20cd2dfeaefd604d068923862c391222c4d512"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/47327",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:3be41e7a8307",
+      "dbms": "cockroachdb",
+      "title": "opt: internal error for comparison overload not found",
+      "reported_date": "2020-12-15",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "jordanlewis",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/57959"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/57959",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/57959",
+            "source_type": "github_issue",
+            "excerpt": "Thanks, SQLancer!",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:82cccc493aee2ab7a95da49f84edc4919216733d27eda940f38901d6ca8999f7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/57959",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:82cccc493aee2ab7a95da49f84edc4919216733d27eda940f38901d6ca8999f7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:945ee7525743",
+      "dbms": "cockroachdb",
+      "title": "release-20.2: DROP DATABASE causes ERROR: descriptor not found",
+      "reported_date": "2021-03-19",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mgartner",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/62281"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/62281",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/62281",
+            "source_type": "github_issue",
+            "excerpt": "release-20.2: DROP DATABASE causes ERROR: descriptor not found\n\nWhile [updating SQLancer to run against Cockroach v20.2.6](https://github.com/sqlancer/sqlancer/pull/316#issuecomment-803167858), SQLancer revealed a bug. Oddly, I'm unable to reproduce the error by manually running the statements that caused it:",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:55c9d65c8da7c3901cac059d7e98d508a3dec980c60116dc650a4e7e8c6df62f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/62281",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:55c9d65c8da7c3901cac059d7e98d508a3dec980c60116dc650a4e7e8c6df62f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:6b1c173c3292",
+      "dbms": "cockroachdb",
+      "title": "colfetcher: incorrect decoding of unique secondary indexes with multiple column families",
+      "reported_date": "2021-06-22",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/66706"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/66706",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/66706",
+            "source_type": "github_issue",
+            "excerpt": "colfetcher: incorrect decoding of unique secondary indexes",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:5905db6b095e68021652809a4f8b01680b3400eff4898eea2e4467f7b3518cc1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:aa431027b0ada5b7a834046007339ae61321fb02cd95cc93aeae5f773475dc96",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "The TLP roachtest found this defect; the issue was retitled to name it. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/66706",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5905db6b095e68021652809a4f8b01680b3400eff4898eea2e4467f7b3518cc1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:12a604c7caf2",
+      "dbms": "cockroachdb",
+      "title": "potential correctness bug in 21.1 found by SQLancer",
+      "reported_date": "2021-07-01",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mgartner",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/67102"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/67102",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/67102",
+            "source_type": "github_issue",
+            "excerpt": "potential correctness bug in 21.1 found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:9f5106a013dca1976a4fb40e93d69b28ee61ddacc5cdf54c179cac5db5e4d451",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/67102",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9f5106a013dca1976a4fb40e93d69b28ee61ddacc5cdf54c179cac5db5e4d451"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:7b6891b72d0b",
+      "dbms": "cockroachdb",
+      "title": "table with a `CHECK (false)` constraint allows inserts",
+      "reported_date": "2021-07-01",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mgartner",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/67100"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/67100",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/67100",
+            "source_type": "github_issue",
+            "excerpt": "This bug was found while investigating a correctness bug found by SQLancer when upgrading it to test v21.1.3 (https://github.com/sqlancer/sqlancer/pull/369).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:2e6c7bfa802e09ec6559fdce158e0e595ea5567d555e9c1ad5922777b053f474",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/67100",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2e6c7bfa802e09ec6559fdce158e0e595ea5567d555e9c1ad5922777b053f474"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:109e9bc3c474",
+      "dbms": "cockroachdb",
+      "title": "sql: vectorized IN evaluation incorrectly assumes the RHS tuple contents are sorted",
+      "reported_date": "2021-08-16",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/68979"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/68979",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/68979",
+            "source_type": "github_issue",
+            "excerpt": "sql: vectorized IN evaluation incorrectly assumes the RHS tuple contents",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e60989da8424d337024f8e17cec5ba13a698af9518b857497934c249e891d31e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:f0a39ed2f3ae5e8aecec8232a1568bd8fa52a9ae7aec41ee9c429c6574c7bff2",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "The TLP roachtest found this defect; the issue was retitled to name it. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/68979",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e60989da8424d337024f8e17cec5ba13a698af9518b857497934c249e891d31e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:f5e89c1718a4",
+      "dbms": "cockroachdb",
+      "title": "sql: computed column expressions incorrectly evaluated when dependent on \"char\" type",
+      "reported_date": "2021-08-24",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "mgartner",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/69327"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/69327",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/69327",
+            "source_type": "github_issue",
+            "excerpt": "TLP found a correctness bug",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:09f6dee104921ed9db7bb65c330e2227c8a7dab54c1594f3aeb7751542b4beb0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:705d873685ebcfb9b833519dac764a11bf803fdb51a2fded89458f1730abfe20",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/69327",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:09f6dee104921ed9db7bb65c330e2227c8a7dab54c1594f3aeb7751542b4beb0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:19a960fac558",
+      "dbms": "cockroachdb",
+      "title": "sql: computed column expressions incorrectly evaluated when dependent on certain types",
+      "reported_date": "2021-08-31",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "mgartner",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/69665"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/69665",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/69665",
+            "source_type": "github_issue",
+            "excerpt": "A TLP failure was reported",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:ee95ffe0f43f16677b38f35ba75deced55660073ce8746df580cc77c5ac21bfe",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:77baddafaf1eeac3e70159b16f06e1e2ed6fe2dfd0ba4c8d9aa21ef4901a7a61",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/69665",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ee95ffe0f43f16677b38f35ba75deced55660073ce8746df580cc77c5ac21bfe"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:2465e911296c",
+      "dbms": "cockroachdb",
+      "title": "TLP Failure with COUNT",
+      "reported_date": "2021-09-22",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "nehageorge",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/70587"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/70587",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/70587",
+            "source_type": "github_issue",
+            "excerpt": "The count is incorrectly computed for the unpartitioned",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c698a9ad18a0ae3eea2700837a40b5d32935d240d996b09c66d3458efee2980e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:59a10e27b89b1ed4d9198b5a5b40bba5495dc840bea25eca1ff1a82219410d16",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/70587",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c698a9ad18a0ae3eea2700837a40b5d32935d240d996b09c66d3458efee2980e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:985db74cbdf7",
+      "dbms": "cockroachdb",
+      "title": "sql: corruption of unique index with virtual computed column (or expression column)",
+      "reported_date": "2021-12-22",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "mgartner",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/74216"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/74216",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/74216",
+            "source_type": "github_issue",
+            "excerpt": "TLP discovered a correctness issue",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:729210ee933f2cfbca218a27161c9325e8a9a37569a9fda77baeb7383645242f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:0ac689fa0f64d9b7bafad22eb2e4f7c48871d94df61d616e46dad8d37bdc645e",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/74216",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:729210ee933f2cfbca218a27161c9325e8a9a37569a9fda77baeb7383645242f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:8101cfed979b",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2021-12-28",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/74295"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/74295",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/74295",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:10488042f36baa1c8bb7c0881aafad0b1933a4ff7d4f8bad763a2f5c01f3e26d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:4b3449725cb5c5ca189d48c7e0d504802aa63965201dd639a622bedb21fd5268",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/74295",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:10488042f36baa1c8bb7c0881aafad0b1933a4ff7d4f8bad763a2f5c01f3e26d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:bf5b19d66a7e",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-01-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/74675"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/74675",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/74675",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:38d7f3624c0424731797e7c92eee042580923d7a900a63879163cbab9f534920",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:80e1ffc6f0394e97c7eaea6f473148d6a6547cabd4d823f199bc04886fbaf244",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/74675",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:38d7f3624c0424731797e7c92eee042580923d7a900a63879163cbab9f534920"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:3a8594c93050",
+      "dbms": "cockroachdb",
+      "title": "sql: incorrect splitting lookup span into family spans",
+      "reported_date": "2022-02-09",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/76289"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/76289",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/76289",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:3c40b992b0c15034a57b976fe934f310e7607e95de21b6c36f21283c7a9556f7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:26011e583fbe9afa6fbc75978121b26e37f026fc061a98a113c58908f40ff8e9",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/76289",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3c40b992b0c15034a57b976fe934f310e7607e95de21b6c36f21283c7a9556f7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:f1b189e969dd",
+      "dbms": "cockroachdb",
+      "title": "opt: cascading deletes do not correctly remove KVs of indexes on virtual columns",
+      "reported_date": "2022-02-21",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/76852"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/76852",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/76852",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:ef4c31699288872a9946b2d8d8d4b4276c19f465663bdb496bfe8a3214e3579c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:5e7ec0fce73e3cd91e8f92c0a96180b418a3aea26c6455ee64564e1fd8cbcb7f",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/76852",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ef4c31699288872a9946b2d8d8d4b4276c19f465663bdb496bfe8a3214e3579c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:20f1e4a80773",
+      "dbms": "cockroachdb",
+      "title": "roachtest: -0 is not converted to 0, causing spurious tlp failure",
+      "reported_date": "2022-03-02",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/77279"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/77279",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/77279",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:dc714556f4680ef0a3c13cc63b4879a4febe32a1e19a20eb1e96d08c026e77e1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:c590517920e633a415b1e61ae39f92f342b0a74ed87cd9adec998fe21a6df66a",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/77279",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dc714556f4680ef0a3c13cc63b4879a4febe32a1e19a20eb1e96d08c026e77e1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:2074ced27a88",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-03-19",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/78137"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/78137",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/78137",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:9b4c1a05c20383a9633e84f776382eada5875486cea114f12703fa7f823ec3ec",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:d2c2da7b29ac7d0b29da3c0e7882de4191c572d8caa4deee3d55db14247935df",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/78137",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9b4c1a05c20383a9633e84f776382eada5875486cea114f12703fa7f823ec3ec"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:0dab3a81353f",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-03-23",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/78327"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/78327",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/78327",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:9bf4c8a2f6d5593f126e809b648c6dd697911adf29bd8c70d365791da2ef058d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:64a98e9f4dec773163f0a1cbdbf7a7bb84c2eb21d85590d920f51208c440c0b5",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/78327",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9bf4c8a2f6d5593f126e809b648c6dd697911adf29bd8c70d365791da2ef058d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:0c12adec2acc",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-03-25",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/78492"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/78492",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/78492",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b87897880822e010806d3a6ce36d6f5c12ad881b7028b4cf7962c7fe1db21146",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:b58278c8b3b405905e4d083a0ffa80f62779a99587d06d25bb28661b692e53a7",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/78492",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b87897880822e010806d3a6ce36d6f5c12ad881b7028b4cf7962c7fe1db21146"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:a61a7b162796",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-03-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/79033"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/79033",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/79033",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:331f0fb4fc2a6c4d908970752e443ba4e9f62e2a083ca7844802206aeb24aff0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:78d3c5cf43b3b7abd0acaad74d8a73cd716a2b2b0a067a158a34cca3f434f3d3",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/79033",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:331f0fb4fc2a6c4d908970752e443ba4e9f62e2a083ca7844802206aeb24aff0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:3fda46644c28",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-04-03",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/79290"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/79290",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/79290",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b41dca1b89970828fddfea5251352493eb7f9163a1c4c4e865c3d15f3a4ebf2b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:37c9cfd644ffba576aac6266f4248bda7334b46746e16537f2fde81fff99fdac",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/79290",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b41dca1b89970828fddfea5251352493eb7f9163a1c4c4e865c3d15f3a4ebf2b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:26f8d688e1a5",
+      "dbms": "cockroachdb",
+      "title": "tlp: do not generate mutations in partitioned/unpartitioned queries",
+      "reported_date": "2022-04-10",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/79737"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/79737",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/79737",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:8f66e17795ec7ee7cee2903eadb430e50cc71bd236c408137717383c25ee660c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:0f3d5691ea2aaeeed41f79199baf05e7b4a9888f2099b06eb4b3926c3c6930af",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/79737",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8f66e17795ec7ee7cee2903eadb430e50cc71bd236c408137717383c25ee660c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:e1c6575c6cf5",
+      "dbms": "cockroachdb",
+      "title": "sql: v22.1: spurious tlp failure due to string comparison",
+      "reported_date": "2022-04-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/79947"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/79947",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/79947",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:8af74170ed29d1c4ee89865c5d9b92cdedcac857c6fc7586714e2b5abf2ab7bb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:60f7219dfec8788f0187107266dafb86063786bf39a14b21bc59419a6b57336b",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/79947",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8af74170ed29d1c4ee89865c5d9b92cdedcac857c6fc7586714e2b5abf2ab7bb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:7b366936cff7",
+      "dbms": "cockroachdb",
+      "title": "sql: v21.2, v22.1 incorrect results for st_intersects filter",
+      "reported_date": "2022-04-15",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/79992"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/79992",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/79992",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:8b2d133156b4b8da1d2f03ef103066d09300391d2ef0853e0f1f8bc11fce60f0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:927cd6ad1310aa19becea5df875ef77ba93a6b39a20bc34a92678bfcbf91a1f5",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/79992",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8b2d133156b4b8da1d2f03ef103066d09300391d2ef0853e0f1f8bc11fce60f0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:d82242b77079",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-04-16",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/80047"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/80047",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/80047",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f6bf6de3b9570599ba8348cd5f91c1f3fa7df8f4266dcb56a121e049509fe0bb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:722a193f287bebd26f0c7058a744174353e55201f2c34a07740eaafb6e0c6991",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/80047",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f6bf6de3b9570599ba8348cd5f91c1f3fa7df8f4266dcb56a121e049509fe0bb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:0ae26dc38bd6",
+      "dbms": "cockroachdb",
+      "title": "geoindex: false negatives producing incorrect results for geography intersection",
+      "reported_date": "2022-04-22",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/80374"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/80374",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/80374",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:dd7e20edfdc9105c99f342b0be929805b8d969cde167c139aa01f0772df3ba55",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:b2022ec4a7a0f287e4d045587bb3e0f09484ad6504d8c79bc6c02fd9d39ddf46",
+          "classified_at": "2026-09-13T06:23:27Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/80374",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dd7e20edfdc9105c99f342b0be929805b8d969cde167c139aa01f0772df3ba55"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:fac0372ccee3",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-05-20",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/81574"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/81574",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/81574",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:54e6667efb349d795f03ddc0d7b9d997579c87f75269e6956302d44f05cd38b5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:b37bf1abe6437e5fca06246b77471858abafd5c4ae63567e8e361c5f389e91f0",
+          "classified_at": "2026-09-13T06:23:29Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/81574",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:54e6667efb349d795f03ddc0d7b9d997579c87f75269e6956302d44f05cd38b5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:4b76230b8b4e",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-05-26",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/81885"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/81885",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/81885",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:0f60fd8eaaacd6c32fdef99610c57d407f9cfe77d20c18cb67cf82a5c61769c1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:4b39431e8b5500043ddfdac3cd6e394ef0d135528677fb03796cf93062542786",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/81885",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0f60fd8eaaacd6c32fdef99610c57d407f9cfe77d20c18cb67cf82a5c61769c1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:76bb3c09b576",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-06-23",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/83272"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/83272",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/83272",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:08b4c3dc6925b3508f28f85a034fd063af885d93b7005087c7bdb25057f120df",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:736ae22e5d332b64e0582d533e46d57b03aa49db961ea2a564287462e7b05542",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/83272",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:08b4c3dc6925b3508f28f85a034fd063af885d93b7005087c7bdb25057f120df"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:5f9f5508ba9d",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-06-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/83584"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/83584",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/83584",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:7203c13743f9bf10ce3bf6ea78339743089dd4e548bfd76f7d8f0c8b1a5f10a5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:63b65112278c9d12b0dfb37cbe2fe06d6263bac4086bf29020bde18a39386d66",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/83584",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7203c13743f9bf10ce3bf6ea78339743089dd4e548bfd76f7d8f0c8b1a5f10a5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:b35b92a12df8",
+      "dbms": "cockroachdb",
+      "title": "sql: Duration.Encode overflows causing incorrect results",
+      "reported_date": "2022-07-03",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/83756"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/83756",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/83756",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:23786bf5cc8e8cce4320e0b6700f10076f79fc6dba31d5ef749343a7feb6b36b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:9a527180ea8e6ae3af8eea6c08bc6f0760c892f90f246af5caee9a66486939bb",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/83756",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:23786bf5cc8e8cce4320e0b6700f10076f79fc6dba31d5ef749343a7feb6b36b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:255256183580",
+      "dbms": "cockroachdb",
+      "title": "sql: support SCRUB on temp tables",
+      "reported_date": "2022-07-04",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/83770"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/83770",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/83770",
+            "source_type": "github_issue",
+            "excerpt": "sql: support SCRUB on temp tables",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d0c45750dec8d68fe27fba1270d1636235576e0f02d723d2ab0fbec1a26425a1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/83770",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d0c45750dec8d68fe27fba1270d1636235576e0f02d723d2ab0fbec1a26425a1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:14726d223a91",
+      "dbms": "cockroachdb",
+      "title": "Crashing by EXPLAIN Statement",
+      "reported_date": "2022-07-07",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/83965"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/83965",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/83965",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nCREATE TABLE t1 (c0 INT);\r\nCREATE TABLE t2 (c0 INT);\r\nCREATE TABLE t3 (c0 INT);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cb61c0d1da09b258b972ad729ee4c3a2f7baf8156143b9afda3a625c68658a82",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cb61c0d1da09b258b972ad729ee4c3a2f7baf8156143b9afda3a625c68658a82"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:6e97a0ca79fe",
+      "dbms": "cockroachdb",
+      "title": "Crash: panic: RecordingStructured has 30 recordings; expected 1",
+      "reported_date": "2022-07-08",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/84056"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/84056",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/84056",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t2 ();\r\nSET SESSION TRACING=true;\r\nEXPLAIN SELECT count0 FROM t1;\r\nEXPLAIN  SELECT count FROM t1, t2, t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:7977432dc587f9e44de76a3814bd1a482384ab057b164df2241087f5c08c1546",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:7977432dc587f9e44de76a3814bd1a482384ab057b164df2241087f5c08c1546"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:d79fd3aec969",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed on complex join condition involving geometry",
+      "reported_date": "2022-07-23",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/84957"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/84957",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/84957",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:94795172c1c1456433df7cd86869f0fa0c829c83ed6c94a409c0f85949384bf9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:4b600fc361ea68147fa24ed77777469aa80e4ba60c61787a5aad751c9eb71c57",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/84957",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:94795172c1c1456433df7cd86869f0fa0c829c83ed6c94a409c0f85949384bf9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:c89d4bddaa8c",
+      "dbms": "cockroachdb",
+      "title": "sql: No Result Returned by SHOW COLUMN on a temporary table",
+      "reported_date": "2022-08-01",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/85388"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85388",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/85388",
+            "source_type": "github_issue",
+            "excerpt": "sql: No Result Returned by SHOW COLUMN on a temporary table",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:87b683ab1ecb89db3d1907135fad6293a298ca40b2cb9dcbeddaacf46bfc65e7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/85388",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:87b683ab1ecb89db3d1907135fad6293a298ca40b2cb9dcbeddaacf46bfc65e7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:7725eaeff9ea",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Result by UNION",
+      "reported_date": "2022-08-03",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/85502"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/85502",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/85502",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 BOOL AS (1 IS NULL) STORED, CONSTRAINT \"primary\" PRIMARY KEY(c0));\r\nCREATE TABLE t2 (c0 INT);\r\nINSERT INTO t2 (c0) VALUES(1);\r\nINSERT INTO t1 DEFAULT VALUES;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3f11a66fe928da698cdd394df9d4c632ce311a9f62c8ebbfd9ad4ddddddf4d59",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3f11a66fe928da698cdd394df9d4c632ce311a9f62c8ebbfd9ad4ddddddf4d59"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:fa4415ad2689",
+      "dbms": "cockroachdb",
+      "title": "An Unexpected Error in `CROSS MERGE JOIN`",
+      "reported_date": "2022-09-18",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/88104"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/88104",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/88104",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nCREATE TABLE t1 (c0 INT);\r\nSELECT * FROM t0 CROSS MERGE JOIN t1; -- ERROR: could not produce a query plan conforming to the MERGE JOIN hint\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:265e99f87ba777a753d95e1193c4033489d13da10ccdeead874bba54fa6338e6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:265e99f87ba777a753d95e1193c4033489d13da10ccdeead874bba54fa6338e6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:579337095268",
+      "dbms": "cockroachdb",
+      "title": "tlp: row with null regproc missing from partitioned query",
+      "reported_date": "2022-09-18",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/88110"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/88110",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/88110",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:abd0a83014335d6c4af6dd572c217f3af15ecb8524c93c54f0a0f86bae822480",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:8d2afcbae219857ff24b306911e9de484664f977e6faa135fc4a6fcaf269fda5",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/88110",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:abd0a83014335d6c4af6dd572c217f3af15ecb8524c93c54f0a0f86bae822480"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:0d3fe6b4d845",
+      "dbms": "cockroachdb",
+      "title": "An unexpected error in `LEFT LOOKUP JOIN`",
+      "reported_date": "2022-09-19",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/88120"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/88120",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/88120",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nCREATE TABLE t1 (c0 INT);\r\nSELECT * FROM t0 LEFT LOOKUP JOIN t1 ON t0.c0=t1.c0; --ERROR: could not produce a query plan conforming to the LOOKUP JOIN hint\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e62dca55d325ad7fa4cac019a350f1023d26d948ecc7795010a64ee7a5de9b20",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/88120",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e62dca55d325ad7fa4cac019a350f1023d26d948ecc7795010a64ee7a5de9b20"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:055be6507c2a",
+      "dbms": "cockroachdb",
+      "title": "internal error: index found in depended-on-by references, no such index in this relation",
+      "reported_date": "2022-09-21",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/88344"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/88344",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/88344",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 BOOL);\r\nCREATE VIEW v0(c0) AS SELECT c0 FROM t0 @{FORCE_INDEX=t0_pkey};\r\n\r\nTRUNCATE TABLE t0; -- ERROR: internal error: relation \"t0\" (112): index ID 1 found in depended-on-by references, no such index in this relation",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c6d503e4ecd4713d0fa1f008282b4b022f93eb11f07d8c273e4ccc2a751f0f47",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/88344",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c6d503e4ecd4713d0fa1f008282b4b022f93eb11f07d8c273e4ccc2a751f0f47"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:83e3134a04bb",
+      "dbms": "cockroachdb",
+      "title": "Potential Issue for Estimated Rows",
+      "reported_date": "2022-09-22",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/88455"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/88455",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/88455",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nCREATE TABLE t1 (c0 INT);\r\nINSERT INTO t0 VALUES (1),(2),(3),(4),(5),(6),(7),(8),(9),(10),(11),(12),(13);\r\nINSERT INTO t1 VALUES (21),(22),(23),(24),(25);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:91cdbb25bc8534865474c1dd722c78ec5f964e0239b088ca02010ec3d3f7a5af",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:91cdbb25bc8534865474c1dd722c78ec5f964e0239b088ca02010ec3d3f7a5af"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:d10e3006bc76",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-09-28",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/88910"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/88910",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/88910",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e3bf03c8b75b5574e236008daf97c8314e58885a92d3ea7b4a11e94ba3d8a67f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:78328a24b2c1673cf4758160ab241e39deac6de37a0d680c793c9f69c11c2d99",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/88910",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e3bf03c8b75b5574e236008daf97c8314e58885a92d3ea7b4a11e94ba3d8a67f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:784f4afa2fb7",
+      "dbms": "cockroachdb",
+      "title": "Suspicious Estimated Rows of `HAVING` Clause",
+      "reported_date": "2022-09-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/88982"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/88982",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/88982",
+            "source_type": "github_issue",
+            "excerpt": "create table t0 (c1 INT);\r\ncreate table t1 (c1 INT);\r\ninsert into t0 values(1),(2);\r\ninsert into t1 values(3),(4),(5);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e1b735b19811f7360ce9e00dc4d519d0f45308fedba8160c7b60d92de3205d45",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/88982",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e1b735b19811f7360ce9e00dc4d519d0f45308fedba8160c7b60d92de3205d45"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:f3467c22c285",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed: SplitLimitedSelectIntoUnionSelects makes incorrect transformation",
+      "reported_date": "2022-09-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/88993"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/88993",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/88993",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e5104d109a630703a3ce815a38e837e4d3e40b8c60ee7090ef78d59c740f1310",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:a01286a70d5db439830771a74c1f444763f6276634f7884159be6a41b12e67a4",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/88993",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e5104d109a630703a3ce815a38e837e4d3e40b8c60ee7090ef78d59c740f1310"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:4c7fb4231a1a",
+      "dbms": "cockroachdb",
+      "title": "Estimated Rows of `RIGHT JOIN` and `FULL OUTER JOIN`",
+      "reported_date": "2022-09-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/89060"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/89060",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/89060",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nCREATE TABLE t2 (c0 INT, c1 INT);\r\nINSERT INTO t2 VALUES(1,2), (3,4), (5,6), (null, null);\r\nINSERT INTO t0 VALUES(1), (2);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:873c69891a35cba808c065d171a72d45b61cbbbd9807abd90dddea13f178041d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/89060",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:873c69891a35cba808c065d171a72d45b61cbbbd9807abd90dddea13f178041d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:61bcbed58323",
+      "dbms": "cockroachdb",
+      "title": "roachtest.tlp failed: incorrect cardinality of EXCEPT produces incorrect results",
+      "reported_date": "2022-09-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "rytaft",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/89101"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/89101",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/89101",
+            "source_type": "github_issue",
+            "excerpt": "incorrect cardinality of EXCEPT produces incorrect results",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:2614fcd19e96911cf73a4a13dceacd19ed6dee4d402fdbde0ea1a68f8eee70bc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:e3c5281693fc5c2a528f6c36457cbbafc2750dcc51d68a3100ccd0dc8b98018c",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "The TLP roachtest found this defect; the issue was retitled to name it. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/89101",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2614fcd19e96911cf73a4a13dceacd19ed6dee4d402fdbde0ea1a68f8eee70bc"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:092e692cb336",
+      "dbms": "cockroachdb",
+      "title": "An Issue of Estimated Rows",
+      "reported_date": "2022-10-02",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/89161"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/89161",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/89161",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nINSERT INTO t0 VALUES (1), (2), (3), (4), (5), (6), (7), (8), (9), (10);\r\nANALYZE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:626a7aadb2de66ef1bc6ad353c83703c6fa259be1c9db5419752f25f652a4baa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:626a7aadb2de66ef1bc6ad353c83703c6fa259be1c9db5419752f25f652a4baa"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:8fc8e8a48b47",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Estimated Rows in `HAVING` clause",
+      "reported_date": "2022-10-06",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/89462"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/89462",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/89462",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 SMALLSERIAL, c1 INT, PRIMARY KEY(c1, c0));\r\nCREATE TABLE t1 (c0 INT);\r\nINSERT INTO t0 (c1) VALUES(1), (2), (3), (4), (5);\r\nANALYZE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d22942a8e09fb04aa8a9f30fbc1f5609759e4911ecb89902938e865f969ed6b7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d22942a8e09fb04aa8a9f30fbc1f5609759e4911ecb89902938e865f969ed6b7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:1c9291664fd3",
+      "dbms": "cockroachdb",
+      "title": "opt: internal error in FoldBinaryCheckOverflow",
+      "reported_date": "2022-10-10",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mgartner",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/89692"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/89692",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/89692",
+            "source_type": "github_issue",
+            "excerpt": "I found a bug while upgrading SQLancer to 22.2.0-beta.2.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8c03aa0c91f7cd2d1862a12efc31c1c2dc1f163fd5909c020bdd95ce7a925d1d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/89692",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8c03aa0c91f7cd2d1862a12efc31c1c2dc1f163fd5909c020bdd95ce7a925d1d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:26e815fc760c",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2022-10-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/89743"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/89743",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/89743",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:bd0fae327ab13d764f6d651f29e788460f227a45e3b77d661b6cceda1a31645a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:ca1434f65ea8bd442f306ff47c3178d5e7b758b64187b59b8d84382d1dd080c5",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/89743",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bd0fae327ab13d764f6d651f29e788460f227a45e3b77d661b6cceda1a31645a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:6d933050fc23",
+      "dbms": "cockroachdb",
+      "title": "Suspicious Estimated Rows by `DISTINCT`",
+      "reported_date": "2022-10-18",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/90113"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/90113",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/90113",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nCREATE TABLE t1 (c0 INT, c1 TIMESTAMPTZ);\r\nCREATE TABLE t2 (c0 TIMESTAMPTZ[]);\r\nINSERT INTO t0 (rowid) VALUES(438922823), (1547647092);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:43d90c8f9a837688fa81e5b9e4e3a320ab9be04b810740975d608b60edd05907",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:43d90c8f9a837688fa81e5b9e4e3a320ab9be04b810740975d608b60edd05907"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:a6b248c92459",
+      "dbms": "cockroachdb",
+      "title": "Suspicious Estimated Rows by `OR`",
+      "reported_date": "2022-10-18",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/90112"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/90112",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/90112",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 INT);\r\nINSERT INTO t1 VALUES(1),(2),(3),(4),(5),(6),(7),(8),(9),(10);\r\nANALYZE t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:258207504528ea46b1ff3bea93f5d1b50036428ec04987f885f1db4b1f7b2123",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:258207504528ea46b1ff3bea93f5d1b50036428ec04987f885f1db4b1f7b2123"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:ffb8e8929374",
+      "dbms": "cockroachdb",
+      "title": "ERROR: internal error: expected *DInt, found tree.dNull",
+      "reported_date": "2022-12-24",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/94264"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/94264",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/94264",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nCREATE TABLE t1 (c0 INT);\r\nSELECT * FROM t0 INNER JOIN t1 ON 2-(9223372036854775807+436256318)<(CASE  WHEN false THEN -1 END);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:bfc4ebecf069499c678543008a3855cdbd97383e3d67e333cc97357b5d578410",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:bfc4ebecf069499c678543008a3855cdbd97383e3d67e333cc97357b5d578410"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:deee60bf2a27",
+      "dbms": "cockroachdb",
+      "title": "internal error about index",
+      "reported_date": "2023-04-05",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/100681"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/100681",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/100681",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 BIT(98), c1 BYTES[]);\r\nCREATE TABLE t1 (c0 BOOL);\r\nCREATE INDEX ON t1(c0);\r\nCREATE INDEX ON t1(c0);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4b8fca72436a3c9a176714eb450fcafe557c39c0f92d923bda909dab236865a0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/100681",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4b8fca72436a3c9a176714eb450fcafe557c39c0f92d923bda909dab236865a0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:b92b988d3f15",
+      "dbms": "cockroachdb",
+      "title": "`SQRDIFF` has different results when the order of arguments changed",
+      "reported_date": "2023-04-15",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/101588"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/101588",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/101588",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 STRING);\r\nCREATE TABLE t2 (c0 INT);\r\n\r\nINSERT INTO t1 (c0) VALUES('');",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2125b220ce0b8474e1bf34c87248051cfbd7b8663e47868a56e65b9581f1508b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/101588",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2125b220ce0b8474e1bf34c87248051cfbd7b8663e47868a56e65b9581f1508b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:b47d982498a0",
+      "dbms": "cockroachdb",
+      "title": "sql: create unique index error: failed to construct index entries during backfill: overflow during Encode",
+      "reported_date": "2023-04-20",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "rharding6373",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/101964"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/101964",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/101964",
+            "source_type": "github_issue",
+            "excerpt": "Found this issue running sqlancer locally.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:cecdbbf0d394b6e975003d3911547f1d71cc9787762943909f73713a0aad834c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/101964",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cecdbbf0d394b6e975003d3911547f1d71cc9787762943909f73713a0aad834c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:317a97b7874d",
+      "dbms": "cockroachdb",
+      "title": "sql: inconsistent ASC/DESC ordering of BOX2D",
+      "reported_date": "2023-04-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/102661"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/102661",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/102661",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:29beaabc3e4c3a224e5b4f76b4a23f7b8af1df4a5b2221d2234de831ba5ba749",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:d2d5ada837c335c404216046c72b88508db9abe3574d18f8a83cadda56effcb1",
+          "classified_at": "2026-09-13T06:23:27Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/102661",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:29beaabc3e4c3a224e5b4f76b4a23f7b8af1df4a5b2221d2234de831ba5ba749"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:88c6581b0e9c",
+      "dbms": "cockroachdb",
+      "title": "roachtest/tlp: 'expected unpartitioned and partitioned results to be equal' assertion",
+      "reported_date": "2023-05-30",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "irfansharif",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/104115"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/104115",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/104115",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:ec980e78ec8691ec0d0f543905b56dda746c4eac423f09ab4703e55ab3172acb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:136834f19f0f8ccea0cbdeb88d7d1b5347dd7b594535625958f7471293f5f315",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/104115",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ec980e78ec8691ec0d0f543905b56dda746c4eac423f09ab4703e55ab3172acb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:45554390f0fc",
+      "dbms": "cockroachdb",
+      "title": "sql: unexpected internal error message from new schema changer during CREATE INDEX",
+      "reported_date": "2023-06-07",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/104484"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/104484",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/104484",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 VARBIT(10) AS (B'1') STORED);\r\nCREATE INDEX ON t1(c0 DESC) USING HASH;\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8d07cfd90b4e33703029bb8ec6e13f978cb79e68311880dd9a4c143f5ae06978",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/104484",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8d07cfd90b4e33703029bb8ec6e13f978cb79e68311880dd9a4c143f5ae06978"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:ccc9b1b61e65",
+      "dbms": "cockroachdb",
+      "title": "sql: data race in NewDCollatedString",
+      "reported_date": "2023-09-10",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "srosenberg",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/110322"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/110322",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/110322",
+            "source_type": "github_issue",
+            "excerpt": "Given its non-deterministic nature, it first popped up on my radar in June [4], after running `sqlancer` against a 3-node cluster. However, it took a while to obtain the root cause, as well as, a semi-deterministic reproduction. The root cause is the reuse of `evalCtx` on a gateway in the case where a pair (or more) outboxes are executed _asynchronously_, each using `COLLATE` in some form. A detailed summary of the data race and a reproduction is given below.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b4dc61113bdb90d968a27ebcb6bc34ff9f256147908fd71be6679bc9ad60e35b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/110322",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b4dc61113bdb90d968a27ebcb6bc34ff9f256147908fd71be6679bc9ad60e35b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:5cd62696841a",
+      "dbms": "cockroachdb",
+      "title": "upgrades: auto-repair of catalog corruptions during upgrade may extend cluster finalization indefinitely",
+      "reported_date": "2023-09-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "srosenberg",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/110321"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/110321",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/110321",
+            "source_type": "github_issue",
+            "excerpt": "I'm able to reproduce this behavior in a 5-node cluster. Originally the cluster is on `22.2.13` running `sqlancer` which continuously generates random sql statements, modifying the schema in the process; it creates ~10k small databases spanning ~20k ranges. I haven't been able to confirm what causes the corrupted descriptors [4], but the number continued to steadily climb from ~100 to ~1000 over the 24 hour window,",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b28c501447bbf29c6dbb81f318b7c45bcbe0ec92fd765b6ff1fd16e068067b8f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/110321",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b28c501447bbf29c6dbb81f318b7c45bcbe0ec92fd765b6ff1fd16e068067b8f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:383c2f2a8604",
+      "dbms": "cockroachdb",
+      "title": "Integer Overflow with Index Hints",
+      "reported_date": "2023-09-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/110409"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/110409",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/110409",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 VARBIT(134), c1 DECIMAL);\r\nCREATE TABLE t3 (c0 TIMESTAMP);\r\nCREATE TABLE t5 (c0 BIT(184));",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:92f8e7eb016bc9b33d39b1b7ffa5ecfbf10c620f6134bb972dc10f1951c431a5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/110409",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:92f8e7eb016bc9b33d39b1b7ffa5ecfbf10c620f6134bb972dc10f1951c431a5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:0c4417e50b6d",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Error for ascii()",
+      "reported_date": "2023-09-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/111474"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/111474",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/111474",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nCREATE TABLE t1 (c0 INT);\r\nINSERT INTO t0 (c0) VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bd49b4e806b9d1318b38b074a8ea4b96417a15117110126c8458bfce41c4536b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/111474",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bd49b4e806b9d1318b38b074a8ea4b96417a15117110126c8458bfce41c4536b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:7ba35d009a6e",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Error for timestamp overflow",
+      "reported_date": "2023-09-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/111476"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/111476",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/111476",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 INT, c1 TIMESTAMP);\r\nCREATE INDEX ON t1(c1 ASC, c0);\r\nINSERT INTO t1 (c0) VALUES(1), (2);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1b5b18f09c16b5594b769cf3c9f114ebfadfdef1b4fb599541f1a84b62b224c6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/111476",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1b5b18f09c16b5594b769cf3c9f114ebfadfdef1b4fb599541f1a84b62b224c6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:9e9e04c53448",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Error: overflow during Encode",
+      "reported_date": "2023-09-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/111473"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/111473",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/111473",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t4 (c0 CHAR, c1 INTERVAL UNIQUE);\r\nINSERT INTO t4 (c0, c1) VALUES('1', NULL);\r\n\r\nSELECT * FROM t4 WHERE (INTERVAL '100000000 year')<t4.c1; -- {}",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:384ec4f055c817ec47bd5a9e5d0e27ea02bb6c35aa2c3668d6a2c6ffcba1d830",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/111473",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:384ec4f055c817ec47bd5a9e5d0e27ea02bb6c35aa2c3668d6a2c6ffcba1d830"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:fb98f8ba4059",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failure: false vs NULL",
+      "reported_date": "2024-07-28",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/127814"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/127814",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/127814",
+            "source_type": "github_issue",
+            "excerpt": "tlp failure: false vs NULL",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:84182521bc64a46188c73201e058ff4a1b9e46a23704636eba1b48297cb93ead",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:4197a208281511af9d18ca06611e315b816d1b4721f3139febd4830eaf201284",
+          "classified_at": "2026-09-13T06:23:27Z",
+          "model": "claude-opus-5",
+          "rationale": "The TLP roachtest found this defect; the issue was retitled to name it. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/127814",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:84182521bc64a46188c73201e058ff4a1b9e46a23704636eba1b48297cb93ead"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:c0250ea47fd0",
+      "dbms": "cockroachdb",
+      "title": "Unexpected result when using a record in WHERE filter",
+      "reported_date": "2024-09-13",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/130649"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/130649",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/130649",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 INT, c1 INT);\r\nINSERT INTO t0(c0) VALUES (1);\r\n\r\nSELECT * FROM t0; -- 1 NULL",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:368b2679e77a3883fdcf0c68783bc7edc82aa0b3205da6583a99372f8d397e07",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/130649",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:368b2679e77a3883fdcf0c68783bc7edc82aa0b3205da6583a99372f8d397e07"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:44ef0fdb0861",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2024-09-15",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/130759"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/130759",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/130759",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:9a4cdf565967896e89aeb99d61c9b0efda2b83ba196db2fc08efcd22b935b349",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:c17b16ee2d0fad8992a9ad4fe693e6a08b1dbb03e87628856d85d03d6e7dfea3",
+          "classified_at": "2026-09-13T06:23:27Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/130759",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9a4cdf565967896e89aeb99d61c9b0efda2b83ba196db2fc08efcd22b935b349"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:4dd2eb345894",
+      "dbms": "cockroachdb",
+      "title": "Performance of Query 7 on TPC-H Benchmark",
+      "reported_date": "2024-11-12",
+      "reported_year": 2024,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/135001"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/135001",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/135001",
+            "source_type": "github_issue",
+            "excerpt": "Performance of Query 7 on TPC-H Benchmark",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:39423d4a4e42ba35c2b1e9caa79c0a05a40e873512782a50066a0f3d772808e1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/135001",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:39423d4a4e42ba35c2b1e9caa79c0a05a40e873512782a50066a0f3d772808e1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:e0899462eb33",
+      "dbms": "cockroachdb",
+      "title": "Performance of Query 63 in TPC-DS Benchmark",
+      "reported_date": "2024-11-28",
+      "reported_year": 2024,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/136350"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/136350",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/136350",
+            "source_type": "github_issue",
+            "excerpt": "Performance of Query 63 in TPC-DS Benchmark",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1377d1212d35e2aa72b1a868322e6e0b03f0cf50f59d6863e4fab00674355787",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/136350",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1377d1212d35e2aa72b1a868322e6e0b03f0cf50f59d6863e4fab00674355787"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:5c6bde3694d7",
+      "dbms": "cockroachdb",
+      "title": "roachtest: incorrect geo results with tlp",
+      "reported_date": "2024-11-29",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/136400"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/136400",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/136400",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:386297ce6e7d0af01627c175610feced9663c97f4849cb68cc123dc61a1be220",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:bc24adf473c81a1acd8ed5d17d9f5700a424dd449fb8928be9d915c31e0bc6e5",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/136400",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:386297ce6e7d0af01627c175610feced9663c97f4849cb68cc123dc61a1be220"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:8382c8d367fa",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2025-01-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/138267"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/138267",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/138267",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:a8143e4836a7d111fd3c97c36153b7af52b479b1ecf1e44002c2ddb2c3f44676",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:c6b600aed58c34d9461c33978d3edbcc401fb54eb953305ba56f2f8b16f244dc",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/138267",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a8143e4836a7d111fd3c97c36153b7af52b479b1ecf1e44002c2ddb2c3f44676"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:4066ecb83333",
+      "dbms": "cockroachdb",
+      "title": "ERROR: internal error: inconsistent Case return types bool unknown",
+      "reported_date": "2025-01-12",
+      "reported_year": 2025,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/138890"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/138890",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/138890",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 BOOL);\r\nWITH temp_table(c0)  AS ((VALUES (NULL))) SELECT t0.c0 FROM t0 FULL OUTER MERGE JOIN temp_table ON false WHERE NULLIF((temp_table.c0) IN (true), (true));\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:87d9199bc36f05e06e88092747dc899d96bc6e617f28345f5615d9a981be6315",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/138890",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:87d9199bc36f05e06e88092747dc899d96bc6e617f28345f5615d9a981be6315"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:35e86560bfbe",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2025-02-04",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/140451"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/140451",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/140451",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:cf4ddde0197ceb9858ce9e74f3305c6f93f405bb3bd9254964aaad0777d0f178",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:ed5b1fa4196c53073769bb963c810316f3c2ba56e306dd1dedb8a0b05cb08394",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/140451",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cf4ddde0197ceb9858ce9e74f3305c6f93f405bb3bd9254964aaad0777d0f178"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:26d4e7122aa1",
+      "dbms": "cockroachdb",
+      "title": "Join Crash with ERROR: internal error: invalid datum type given",
+      "reported_date": "2025-02-06",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/147585"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/147585",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/147585",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 TIMESTAMP);\nCREATE TABLE t3 (c0 TIMESTAMPTZ);\nUPSERT INTO t3 (c0) VALUES(TIMESTAMPTZ '1970-01-05T10:12:17');\nSELECT ALL * FROM  t1 NATURAL FULL JOIN t3;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a70b1c1a7221b986c8edf36f32bf1dd8d0384220de52b70766508ae6b0f0f552",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a70b1c1a7221b986c8edf36f32bf1dd8d0384220de52b70766508ae6b0f0f552"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:54e8d4fa06a3",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2025-03-15",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/142948"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/142948",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/142948",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:0a0a2223bae153284298e23c3bf942952c40ad1dc4f78e92e8a80698160ba32d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:6307bf2d2a226beb2df570e307b9c1a79de312d557cb87d8f1fca5a932e8d47e",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/142948",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0a0a2223bae153284298e23c3bf942952c40ad1dc4f78e92e8a80698160ba32d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:1a747aacb8ab",
+      "dbms": "cockroachdb",
+      "title": "Unexpected -0.0",
+      "reported_date": "2025-04-24",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/145108"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/145108",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/145108",
+            "source_type": "github_issue",
+            "excerpt": "Some mathematical calculations return negative zeros, which seems unexpected.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:594a9db723c5fbb77f6d657dbcd6f31b9ddb4284ef06b424462a8f1302f1dfc4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/145108",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:594a9db723c5fbb77f6d657dbcd6f31b9ddb4284ef06b424462a8f1302f1dfc4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:de24a1fa8f66",
+      "dbms": "cockroachdb",
+      "title": "runtime error: index out of range in SPLIT_PART()",
+      "reported_date": "2025-05-01",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/145888"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/145888",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/145888",
+            "source_type": "github_issue",
+            "excerpt": "runtime error: index out of range in SPLIT_PART()",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7aa6aeebef22120936d95bbab56dc788a6d4c0c3039720efb8a834aeb4a37846",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/145888",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7aa6aeebef22120936d95bbab56dc788a6d4c0c3039720efb8a834aeb4a37846"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:730690cc0927",
+      "dbms": "cockroachdb",
+      "title": "Crash of Truncate",
+      "reported_date": "2025-05-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/146065"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/146065",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/146065",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT UNIQUE CHECK(true));\nCREATE VIEW v0 AS SELECT c0 FROM t0 @{FORCE_INDEX = t0_c0_key};\nTRUNCATE TABLE t0;\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:defef30add536f53597c1b0a4c50d191e723c0dbd8b524d9c5a8040dcf0ef1bb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/146065",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:defef30add536f53597c1b0a4c50d191e723c0dbd8b524d9c5a8040dcf0ef1bb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:917f37e870f2",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Result of to_regclass()",
+      "reported_date": "2025-05-17",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/146898"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/146898",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/146898",
+            "source_type": "github_issue",
+            "excerpt": "create table t6(c0 text);\ncreate index i1 on t6(c0);\ninsert into t6 values('i1');\nselect to_regclass(c0) from t6;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f9d20c3cb81f716ef2c819a3dcb60173e84d8397a87f87709050001541882ca6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/146898",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f9d20c3cb81f716ef2c819a3dcb60173e84d8397a87f87709050001541882ca6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:3077fbd515dc",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2025-05-25",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/147257"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/147257",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/147257",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:707aec6b902916a4fdf283a368e190dbba6d9024709568f766361a22b41ecee6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:b71c3960b5f2a02c08f10199c966929f38ebf52244c2be68d074bce654ec8ad1",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/147257",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:707aec6b902916a4fdf283a368e190dbba6d9024709568f766361a22b41ecee6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:e5b3d45911ea",
+      "dbms": "cockroachdb",
+      "title": "Join panic with connection lost:  ERROR: a panic has occurred invalid datum type given: DECIMAL, expected INT4",
+      "reported_date": "2025-06-02",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/147591"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/147591",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/147591",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT4);\nCREATE TABLE t2 (c0 DECIMAL);\nINSERT INTO t2 (c0) VALUES(-1);\nSET SESSION VECTORIZE=off;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9f71e3ad4d15675eba1b174fedc1d7feeaabe0e33ffdb62b8d32ad9c6c62550d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/147591",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9f71e3ad4d15675eba1b174fedc1d7feeaabe0e33ffdb62b8d32ad9c6c62550d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:c5fd05d675cd",
+      "dbms": "cockroachdb",
+      "title": "Join with internal error: unexpected error from the vectorized engine: runtime error",
+      "reported_date": "2025-06-02",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/147593"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/147593",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/147593",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t2 (c0 STRING);\nCREATE TABLE t60 (c0 INT);\n\nINSERT INTO t60 values (-9223372036854775808);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:80e152a9f600d1fdd82fd8e2a47ad919dc7da718d4e893eaf23893f2b8ab6bd6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/147593",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:80e152a9f600d1fdd82fd8e2a47ad919dc7da718d4e893eaf23893f2b8ab6bd6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:870883dfca2f",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Result of a complex expression",
+      "reported_date": "2025-06-13",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/148279"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/148279",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/148279",
+            "source_type": "github_issue",
+            "excerpt": "create table t0(c0 text);\ninsert into t0 values('/a');\nselect * from t0 join (select 1) on not(-61.100 // 79 || c0 <= c0); -- {/a|1}, Expected: {}\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:fddd3b22d3ec2ed77f060aa3469485b64781b0f8470da2f91c53b3b4f15334a1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/148279",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fddd3b22d3ec2ed77f060aa3469485b64781b0f8470da2f91c53b3b4f15334a1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:d62f71cae1f9",
+      "dbms": "cockroachdb",
+      "title": "a single float8 constant was parsered to different float value in a single query",
+      "reported_date": "2025-07-07",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/149579"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/149579",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/149579",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 FLOAT UNIQUE , c1 FLOAT);\nINSERT INTO t1 (c1) VALUES(0.22602838822570093);\nUPSERT INTO t1 (rowid, c1) VALUES(1283693985, 0.116859849449506);\nINSERT INTO t1 (c1) VALUES(0.6838727571067713), (0.20784120532437889);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:704bc87f7075a4b39526b272c040f64bb80df8fb143011381786e5718f3706fa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/149579",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:704bc87f7075a4b39526b272c040f64bb80df8fb143011381786e5718f3706fa"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:cda7f8a42ec6",
+      "dbms": "cockroachdb",
+      "title": "sql: secondary index corruption for `0::REGCLASS::STRING` when handling Oid column",
+      "reported_date": "2025-08-06",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/151457"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/151457",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/151457",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:003765f521f805b62de5b8e7a08166fae21d4582f4363abff7ee1e2d5a945999",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:2991f9f2daf253953cbaae778eee4c79bbee8af70afce7bdfa6cfb59f40f9550",
+          "classified_at": "2026-09-13T06:23:28Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/151457",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:003765f521f805b62de5b8e7a08166fae21d4582f4363abff7ee1e2d5a945999"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:d0ff88f1bab0",
+      "dbms": "cockroachdb",
+      "title": "remote wall time is too far ahead to be trustworthy",
+      "reported_date": "2025-08-21",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/152247"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/152247",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/152247",
+            "source_type": "github_issue",
+            "excerpt": "I found that sometimes when I run a sql script in CockroachDB, it will trigger a crash `remote wall time is too far ahead to be trustworthy`, but when I rerun this script, I can not reproduce this crash. This is the log and the script, I generated this script with SQLancer (https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:04a1f36d3dc378a0a03e7d9e0a5204568ec5278d49baef5ddca5c82fe7ce09f1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/152247",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:04a1f36d3dc378a0a03e7d9e0a5204568ec5278d49baef5ddca5c82fe7ce09f1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:98d777d37592",
+      "dbms": "cockroachdb",
+      "title": "IF operator has wrong result in prepared statement",
+      "reported_date": "2025-08-28",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/152664"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/152664",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/152664",
+            "source_type": "github_issue",
+            "excerpt": "IF operator has wrong result in prepared statement",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b7b313eef304394994df6b611662a326ddd93c295835243d580a33460d7803ad",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/152664",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b7b313eef304394994df6b611662a326ddd93c295835243d580a33460d7803ad"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:fca08beff6e6",
+      "dbms": "cockroachdb",
+      "title": "Integer overflow with prepared statement",
+      "reported_date": "2025-08-28",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/152657"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/152657",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/152657",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 SMALLINT);\nPREPARE prepare_query (unknown, unknown, int8) AS UPSERT INTO t1 (c0) VALUES($1), ($2), ($3);\nEXECUTE prepare_query(NULL, NULL, 2010947502);\nDEALLOCATE prepare_query;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:01a9f5787dffbf818408d257dcb939e51613bb8fecaae9e4476b03fd7b65b077",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/152657",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:01a9f5787dffbf818408d257dcb939e51613bb8fecaae9e4476b03fd7b65b077"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:dbbcb90f3b94",
+      "dbms": "cockroachdb",
+      "title": "Unexpected error when insert a collatedstring{es}[] value into column with type string[]",
+      "reported_date": "2025-08-28",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/152659"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/152659",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/152659",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c1 STRING[]);\nINSERT INTO t1 (c1) VALUES(ARRAY[('' COLLATE es)]), (ARRAY['']);\n\nERROR: VALUES types string[] and collatedstring{es}[] cannot be matched",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bfab5be9863bdd28b7af74287ca660f93a6f6cdb184c68f9874e491dd9fc6b72",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/152659",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bfab5be9863bdd28b7af74287ca660f93a6f6cdb184c68f9874e491dd9fc6b72"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:bb4db056a907",
+      "dbms": "cockroachdb",
+      "title": "Unexpected error in prepared statement",
+      "reported_date": "2025-08-30",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/152771"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/152771",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/152771",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 TIMESTAMP[] DEFAULT (ARRAY[]::TIMESTAMP[]));\nINSERT INTO t0 (rowid) VALUES(-893615486);\nPREPARE prepare_query (INT64, int8, int8, INT4) AS UPDATE t0@{FORCE_INDEX=t0_pkey} SET rowid=(((($1)-($2)))+((($3)-((($4) :::INT4)))));\nEXECUTE prepare_query(NULL, -35279814, 116703597, NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f8b6d8c202270e9579584d254e39c401cfde59a1c8fd4ac81cf1659ac6b0500b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/152771",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f8b6d8c202270e9579584d254e39c401cfde59a1c8fd4ac81cf1659ac6b0500b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:f8e8b354a371",
+      "dbms": "cockroachdb",
+      "title": "Unexpected syntex error when use COLLATE in the definition of the default value of a column",
+      "reported_date": "2025-08-30",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/152776"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/152776",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/152776",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN DEFAULT (((('') COLLATE ca)) BETWEEN ('') AND ('')));\n\nERROR: relation \"t0\" (1502): at or near \"collate\": syntax error\nSQLSTATE: 42601",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f468be0e7b7c652c9b2cb822046271f524ddd35c0e9c372592a411f84435a30b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/152776",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f468be0e7b7c652c9b2cb822046271f524ddd35c0e9c372592a411f84435a30b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:282ced20905e",
+      "dbms": "cockroachdb",
+      "title": "Unexpected internal error triggered by prepared statement:`trying to add a column of UNKNOWN type at index 6 but BOOL vector already present `",
+      "reported_date": "2025-09-03",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/152914"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/152914",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/152914",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c1 STRING UNIQUE);\nPREPARE prepare_query (text, int8) AS INSERT INTO t1 (c1, rowid) VALUES($1, $2);\nEXECUTE prepare_query('', 1);\nDEALLOCATE prepare_query;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8ad6769a0444b6d32d964142d04e814abc39408e33afd9c931fa793f0ed7e2d2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/152914",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8ad6769a0444b6d32d964142d04e814abc39408e33afd9c931fa793f0ed7e2d2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:bb3979349da0",
+      "dbms": "cockroachdb",
+      "title": "Incorrect plan time of prepared statement",
+      "reported_date": "2025-09-04",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/152945"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/152945",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/152945",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect plan time of prepared statement",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:16d769856e87f2a76227e5ebb03578553d3096cb214078d1bcda972ffcf8c2e9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/152945",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:16d769856e87f2a76227e5ebb03578553d3096cb214078d1bcda972ffcf8c2e9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:c90a933981f7",
+      "dbms": "cockroachdb",
+      "title": "EXECUTE statement triggers unexpected error with JDBC: `Received resultset tuples, but no field structure for them`",
+      "reported_date": "2025-09-15",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/153505"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/153505",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/153505",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT8);\nINSERT INTO t0 VALUES (1);\nPREPARE prepare_query (int8) AS SELECT * FROM t0 WHERE c0 = $1;\nEXECUTE prepare_query(1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5c364011a8d2cf584c5f4af6117aca206940c0db9edddec19855ee1d8475472e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/153505",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5c364011a8d2cf584c5f4af6117aca206940c0db9edddec19855ee1d8475472e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:933ea0af31c6",
+      "dbms": "cockroachdb",
+      "title": "EXPLAIN statement triggers unexpected error with JDBC: `A result was returned when none was expected.`",
+      "reported_date": "2025-09-15",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/153506"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/153506",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/153506",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT8);\nINSERT INTO t0 VALUES (1);\nEXPLAIN SELECT * FROM t0 WHERE c0 = 1;\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b7a16990eb584143ff39d2e8c871341f01f3582bc383ae7ae7aba2659b9aa89f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/153506",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b7a16990eb584143ff39d2e8c871341f01f3582bc383ae7ae7aba2659b9aa89f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:5bbd41f411c6",
+      "dbms": "cockroachdb",
+      "title": "EXECUTE a select statement triggers unexpected error `No results were returned by the query`",
+      "reported_date": "2025-09-17",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/153645"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/153645",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/153645",
+            "source_type": "github_issue",
+            "excerpt": "EXECUTE a select statement triggers unexpected error `No results were returned by the query`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:34460a546884a613fb9ce6ced5ee6bf62eee5c3e9a1f609a30c542032cf312c5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/153645",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:34460a546884a613fb9ce6ced5ee6bf62eee5c3e9a1f609a30c542032cf312c5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:dbb61d5b0117",
+      "dbms": "cockroachdb",
+      "title": "Unexpected error of MAX(NULL) in prepared statement `ERROR: no builtin aggregate for MAX on [unknown]`",
+      "reported_date": "2025-09-17",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/153664"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/153664",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/153664",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t33 (c0 FLOAT);\nINSERT INTO t33 (c0) VALUES(1);\nSET SESSION VECTORIZE=off;\nPREPARE prepare_query (FLOAT) AS SELECT VARIANCE(9.4834975E7), MAX($1) FROM t33;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c6aae10c700e37a4bc72e689dcb608cba14192fe9bf8e5fe13ff8ed0231c95d3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/153664",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c6aae10c700e37a4bc72e689dcb608cba14192fe9bf8e5fe13ff8ed0231c95d3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:873b012e2cd0",
+      "dbms": "cockroachdb",
+      "title": "Unexpected error when bind a value in GROUP BY",
+      "reported_date": "2025-09-17",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/153647"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/153647",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/153647",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t2 (c0 DECIMAL);\nPREPARE prepare_query (DECIMAL) AS SELECT 1 FROM t2 GROUP BY $1;\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b399f04b56b5d26ced1348ba5bec8de44756bba43b6290b27f90092dfcf59fbf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/153647",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b399f04b56b5d26ced1348ba5bec8de44756bba43b6290b27f90092dfcf59fbf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:f0f224088563",
+      "dbms": "cockroachdb",
+      "title": "Unexpected error `ERROR: internal error: non-int start argument type unknown`",
+      "reported_date": "2025-09-18",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/153687"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/153687",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/153687",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 STRING);\nPREPARE prepare_query (BIGINT) AS SELECT 1 FROM t1 WHERE (('a')!~*(SUBSTRING('k', $1, 1)));\nEXECUTE prepare_query(NULL);\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ac8b3003a8fbe0f29721f8271e9c4e19d6732bab5daa57f503d6ea763b97004d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/153687",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ac8b3003a8fbe0f29721f8271e9c4e19d6732bab5daa57f503d6ea763b97004d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:b2e4516deb4b",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed due to unordered limit",
+      "reported_date": "2025-09-30",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/154421"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/154421",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/154421",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:32f55c7be65cb0fe0dc3c9ecce731e5e41aebd30abf1a9e89ceb00751bed17a7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:f7305bbe0bcf7d90094548972eec373fe2f89d779bd7ed9499bbb8954179008d",
+          "classified_at": "2026-09-13T06:23:27Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/154421",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:32f55c7be65cb0fe0dc3c9ecce731e5e41aebd30abf1a9e89ceb00751bed17a7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:82c10c2d83cf",
+      "dbms": "cockroachdb",
+      "title": "Unexpected error \"ERROR: overflow during Encode\"",
+      "reported_date": "2025-10-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/155158"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/155158",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/155158",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 FLOAT);\nINSERT INTO t1 (c0) VALUES(1);\nSET SESSION VECTORIZE=off;\nSELECT DISTINCT MAX((INTERVAL '2027 year')) FROM t1 GROUP BY t1.c0; -- ERROR: overflow during Encode",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:51b3649d8e4844999448fcbdbdb64be8678d6494245d18a941f65b1d88960bed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/155158",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:51b3649d8e4844999448fcbdbdb64be8678d6494245d18a941f65b1d88960bed"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:205c8682a1af",
+      "dbms": "cockroachdb",
+      "title": "Malformed SPLIT_PART error is silently ignored when used in ORDER BY",
+      "reported_date": "2025-10-15",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/155428"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/155428",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/155428",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 int4);\nINSERT INTO t0 VALUES (1);\nSELECT * FROM t0 ORDER BY (SPLIT_PART('', 'A', -1)); \n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:520f36feb8ce44ccc44275d087e615f812b37ff0f937cc19a600eb34c1d006f3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/155428",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:520f36feb8ce44ccc44275d087e615f812b37ff0f937cc19a600eb34c1d006f3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:4e158a36b106",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2025-10-26",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/156218"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/156218",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/156218",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f2f2f5031204904ef10996da83877752764c97a6c41be84f56b721680c157c6f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:6969bd5ea06806450a3e5da575eb1bd66e5060e30077ab0eb70399c6912906ed",
+          "classified_at": "2026-09-13T06:23:27Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/156218",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f2f2f5031204904ef10996da83877752764c97a6c41be84f56b721680c157c6f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:3a7057cb0474",
+      "dbms": "cockroachdb",
+      "title": "Unexpected result of prepared statement with JOINing subquery",
+      "reported_date": "2025-12-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/159043"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/159043",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/159043",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 TIMESTAMPTZ);\nCREATE TABLE t1 (c0 BIT(1) UNIQUE);\nINSERT INTO t1 (c0) VALUES(B'1');\nSELECT 1 FROM t1 CROSS JOIN (SELECT MIN(1)  as subq_1_c0 FROM t0 OFFSET (801969055)::INT8) AS subq_1 GROUP BY t1.c0;   -- empty result",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:33306aa47b7af7be70e335d60fd0ffe3b05f02647b0a77728901ef3bd30ca895",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/159043",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:33306aa47b7af7be70e335d60fd0ffe3b05f02647b0a77728901ef3bd30ca895"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:ea62ce7ef6b8",
+      "dbms": "cockroachdb",
+      "title": "sql: type annotation does not work correctly in prepared statement",
+      "reported_date": "2025-12-09",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/159021"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/159021",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/159021",
+            "source_type": "github_issue",
+            "excerpt": "sql: type annotation does not work correctly in prepared statement",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a9f94c751699a9ec4a8f282714ca34526d3104634e114931cb879bb64f1e8b52",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/159021",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a9f94c751699a9ec4a8f282714ca34526d3104634e114931cb879bb64f1e8b52"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:5a309dce21ef",
+      "dbms": "cockroachdb",
+      "title": "Unexpected result of BIT_AND in prepared statement",
+      "reported_date": "2025-12-10",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/159127"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/159127",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/159127",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 BYTES);\nUPSERT INTO t0 (c0) VALUES('a'::BYTES);\nPREPARE prepare_query (JSONB, JSONB) AS SELECT BIT_AND(ASCII(''::text)) FROM t0 HAVING ($1) IN (((MIN($2)) :::JSONB)); \nEXECUTE prepare_query(NULL::JSONB, NULL::JSONB); -- empty result",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3af4ef1017007ae2fe1226f57c48fab9c206b62a4d6ba598ffb481280d0b735c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/159127",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3af4ef1017007ae2fe1226f57c48fab9c206b62a4d6ba598ffb481280d0b735c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:d633aa73ddb4",
+      "dbms": "cockroachdb",
+      "title": "internal error: non-const expression: function: current_user [type=string]",
+      "reported_date": "2025-12-10",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/159124"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/159124",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/159124",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t15 (c0 TIMESTAMP, c1 STRING);\nCREATE UNIQUE INDEX ON t15(c1) COVERING (c0);\nPREPARE prepare_query (text) AS SELECT DISTINCT STRING_AGG('', '') FROM t15 WHERE (CURRENT_USER()) IN (((t15.c1) :::STRING));\nEXECUTE prepare_query(''::text);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c90bb95e2422d7920470f1ac225416ad5712ab434b8673f89bfd0a853e259724",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/159124",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c90bb95e2422d7920470f1ac225416ad5712ab434b8673f89bfd0a853e259724"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:99e597bb3d61",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed [flake due to lack of `ORDER BY` with `LIMIT` clause]",
+      "reported_date": "2025-12-10",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/159137"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/159137",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/159137",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b993a443c96a99e6b185a195232c805d623fb1adf98fc76c1b86494e8da41dc2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:a124f6d857ea172e043e1f50734df05a3200035e864865f577f1baf83e4eddc4",
+          "classified_at": "2026-09-13T06:23:27Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/159137",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b993a443c96a99e6b185a195232c805d623fb1adf98fc76c1b86494e8da41dc2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:d8275b379311",
+      "dbms": "cockroachdb",
+      "title": "BYTEA to text to BYTEA generate inconsystent value from the original one",
+      "reported_date": "2026-01-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/160754"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/160754",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/160754",
+            "source_type": "github_issue",
+            "excerpt": "BYTEA to text to BYTEA generate inconsystent value from the original one",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ee3d0a4b0309b3fca45ca2578286d2a535274405612726b1e286e60a7a89ba23",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/160754",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ee3d0a4b0309b3fca45ca2578286d2a535274405612726b1e286e60a7a89ba23"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:f2d27bcbbc11",
+      "dbms": "cockroachdb",
+      "title": "ERROR: internal error: could not find format code for column 1 in [FormatText]",
+      "reported_date": "2026-01-19",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/161382"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/161382",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/161382",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 FLOAT);\nPREPARE prepare_query (float8, float8, float8) AS SELECT ALL SUM($1), STDDEV($2), AVG($3) FROM t0 GROUP BY t0.c0;\nEXPLAIN ANALYZE EXECUTE prepare_query(1.95573964E8::float8, 0.03965984596601002::float8, 0.4293795941585613::float8);\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7b13b2173b1189e57d0946fc138dd44e0f39626ab905f10b1026ce4871bf8bb1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/161382",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7b13b2173b1189e57d0946fc138dd44e0f39626ab905f10b1026ce4871bf8bb1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:0efda875cc3e",
+      "dbms": "cockroachdb",
+      "title": "Less efficient query plan generated by normal SELECT than the equivalent prepared SELECT",
+      "reported_date": "2026-02-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/162300"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/162300",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/162300",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 FLOAT DEFAULT (0.03983949807118381::float8) CHECK (((1)=(2))), c1 STRING PRIMARY KEY, c2 TIMESTAMP[]);\nCREATE INDEX ON t0(c0 ASC);\nUPDATE t0@{FORCE_INDEX=t0_pkey} SET c1=LEFT(''::text, (1785117369)::INT8), c2=t0.c2, c0=t0.c0 WHERE true;\nSET SESSION SQL_SAFE_UPDATES=off;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0959b40eac4a16ac04cf37e51d2cbe5cf2e9cbfa2e7a1f0b560fb73fbe6892ac",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/162300",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0959b40eac4a16ac04cf37e51d2cbe5cf2e9cbfa2e7a1f0b560fb73fbe6892ac"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:4bde9a4d105b",
+      "dbms": "cockroachdb",
+      "title": "Less efficient query plan generated by normal SELECT than the equivalent prepared SELECT related to group and limit node in query plan",
+      "reported_date": "2026-02-04",
+      "reported_year": 2026,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/162367"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/162367",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/162367",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 FLOAT UNIQUE , c1 TIMETZ[], c2 BIT(142) DEFAULT (B'0110110000110000001010001101010101000100100001110111101001111100011110110111110000110001010111011010011011001000011101100111000101000011100100'));\nCREATE TABLE t1 (c0 TIMESTAMPTZ CHECK ((c0 < ANY (c0, c0, c0, c0))), FAMILY \"primary\" (c0));\nCREATE VIEW v0(c0) AS SELECT ALL MAX(((0.3714927092448269::float8) :::FLOAT)) FROM t0 GROUP BY t0.c2 HAVING MAX(false);\nINSERT INTO t1 (rowid) VALUES((-1)::INT8), ((66854211)::INT8);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:436c74a432a091c42e354c70ddd077e9e3e31a687d42ef70ba2e41fbbc23358f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/162367",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:436c74a432a091c42e354c70ddd077e9e3e31a687d42ef70ba2e41fbbc23358f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:7bce52426e03",
+      "dbms": "cockroachdb",
+      "title": "redundant filter and render in the query plan of a normal SELECT with empty table",
+      "reported_date": "2026-02-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/162382"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/162382",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/162382",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INTERVAL);\nCREATE TABLE t2 (c0 TIMETZ );\nUPSERT INTO t2 (c0, rowid) VALUES(TIMETZ '1970-01-26T01:04:38', (-1988383072)::INT8);\nUPSERT INTO t2 (rowid, c0) VALUES((-27858679)::INT8, TIMETZ '1969-12-26T16:31:58');",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:773f2ff89f807086d1084b9379d872aaed26dcbe5dd802a2f19eea937287ed45",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/162382",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:773f2ff89f807086d1084b9379d872aaed26dcbe5dd802a2f19eea937287ed45"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:f44112d166cf",
+      "dbms": "cockroachdb",
+      "title": "Inefficient query plan of a normal SELECT with merge join",
+      "reported_date": "2026-02-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/162510"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/162510",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/162510",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 SERIAL8 PRIMARY KEY);\nINSERT INTO t1 (c0) SELECT (random() * 9223372036854775807)::INT8 FROM generate_series(1, 10000) ON CONFLICT DO NOTHING;\nEXPLAIN ANALYZE SELECT DISTINCT SUM((-592761556)::INT8), XOR_AGG((19169261)::INT8), AVG((-1296868711)::INT8) FROM t1  FULL OUTER JOIN t1 AS t1_0 ON true  CROSS JOIN t1 AS t1_1  NATURAL JOIN t1 AS t1_2 WHERE NULLIF(false, ((t1_1.c0) IS NULL));\n                                               info",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b870d5fbe0cb716e7a007a2e5c5e357743fef97881e28cfb86fed5de532a5626",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/162510",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b870d5fbe0cb716e7a007a2e5c5e357743fef97881e28cfb86fed5de532a5626"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:9556e60f8f46",
+      "dbms": "cockroachdb",
+      "title": "redundant join operation in the query plan of a normal SELECT",
+      "reported_date": "2026-02-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/162506"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/162506",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/162506",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 BYTES, CONSTRAINT \"primary\" PRIMARY KEY(c0 ASC), FAMILY \"primary\" (c0));\nINSERT INTO t0 (c0) SELECT (gen_random_uuid()::TEXT)::BYTES FROM generate_series(1, 100000);\nEXPLAIN ANALYZE SELECT CONCAT_AGG(((''::text))) FROM t0  NATURAL JOIN t0 AS t0_0  FULL OUTER JOIN t0 AS t0_1 ON IF(((true)AND(false)), (t0.c0) IN (t0_1.c0, t0_1.c0, t0.c0, t0.c0, t0_1.c0), ((t0.c0)IS DISTINCT FROM(t0.c0))) WHERE (t0.c0) BETWEEN SYMMETRIC (t0_1.c0) AND (TO_IP(''::text)) ORDER BY TIME '1969-12-12T20:00:15' DESC;\n--------------------------------------------------------------------------------------------",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:680931fb48a22811965e6aa4fbef490fd7ea17b616b6df894fb5f7d2e9f30262",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/162506",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:680931fb48a22811965e6aa4fbef490fd7ea17b616b6df894fb5f7d2e9f30262"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:892d920bd16f",
+      "dbms": "cockroachdb",
+      "title": "Index makes the normal SELECT query slow down",
+      "reported_date": "2026-03-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/164810"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/164810",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/164810",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 DECIMAL DEFAULT ((-145324376)::INT8));\nCREATE INDEX ON t0(c0 DESC, rowid);\nINSERT INTO t0 (c0, rowid) VALUES((-1690356295)::INT8, (-1471155344)::INT8);\nEXPLAIN ANALYZE SELECT MAX((-1119468776)::INT8), MIN((-1843237911)::INT8), SUM((-1118064885)::INT8) FROM t0 GROUP BY IF(false, (-656410759)::INT8, (-1139961941)::INT8), NULLIF((((0)::INT8)-((-969398383)::INT8)), (((393604092)::INT8)-((-564669343)::INT8))), t0.c0 ORDER BY t0.c0 ASC LIMIT (1846914734)::INT8;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c7c6da6bf9f13bcc9733fa9e01f79745a432d9e3a1d69926e0372222f301d890",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/164810",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c7c6da6bf9f13bcc9733fa9e01f79745a432d9e3a1d69926e0372222f301d890"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:1789346aac55",
+      "dbms": "cockroachdb",
+      "title": "Unexpected Fault error",
+      "reported_date": "2026-03-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/164887"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/164887",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/164887",
+            "source_type": "github_issue",
+            "excerpt": "I am running SQLancer (https://github.com/sqlancer/sqlancer) on CockroachDB and found it quit while running. I checked the log and found there is a fault error but not sure which test case triggered it. This is the last lines of the log and I attached the whole log. I can provide the whole test case if you can tell me how to find that which one triggered the error.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:60ea99a98556d4fb6d1159df487bfa6c7acb19fbabe951a275198cfbaef6a574",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/164887",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:60ea99a98556d4fb6d1159df487bfa6c7acb19fbabe951a275198cfbaef6a574"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cockroachdb:739bc2ab221f",
+      "dbms": "cockroachdb",
+      "title": "roachtest: tlp failed",
+      "reported_date": "2026-04-12",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "cockroach-teamcity",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/168225"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/168225",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/168225",
+            "source_type": "github_issue",
+            "excerpt": "expected unpartitioned and partitioned results to be equal",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b81ec5932f62962a267c66806098793b60eb477441566a0a8b250ee17a61cb80",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:27e2bb255f192061c6af4b93138d46428a814c1a2e6f1dd3262aa6bbbb8431e2",
+          "classified_at": "2026-09-13T06:23:27Z",
+          "model": "claude-opus-5",
+          "rationale": "CockroachDB's own TLP roachtest reports a partitioned/unpartitioned mismatch: the oracle fired. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/168225",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b81ec5932f62962a267c66806098793b60eb477441566a0a8b250ee17a61cb80"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:18db6ee72af6",
+      "dbms": "cockroachdb",
+      "title": "sql/flowinfra: row stream can emit unset EncDatum row after INTERVAL encode overflow",
+      "reported_date": "2026-07-02",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "srosenberg",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/172185"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/172185",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/172185",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer found an intermittent internal error:",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:4722025ac1479ef881d5124c79963ca095c42f99e04b0293a536a95fd1cfc305",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/172185",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4722025ac1479ef881d5124c79963ca095c42f99e04b0293a536a95fd1cfc305"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:6db16b0728a8",
+      "dbms": "cockroachdb",
+      "title": "sql: Internal error \"expected inverted index scan to have a constraint\" for TIMETZ[] inverted index with NULL-containing containment query",
+      "reported_date": "2026-08-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Annie191",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/173046"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/173046",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/173046",
+            "source_type": "github_issue",
+            "excerpt": "The query does not rely on data, SQLancer, or any specific schema beyond a TIMETZ[] column with an inverted index.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e57e471a9bc6205b980de893dea26a89c0bb49aaeae501f09b178ab7906d6e5e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/173046",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e57e471a9bc6205b980de893dea26a89c0bb49aaeae501f09b178ab7906d6e5e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:cockroachdb:587b81106b59",
+      "dbms": "cockroachdb",
+      "title": "sql: ST_DWithin inverted join fails for POINT EMPTY geography input",
+      "reported_date": "2026-09-13",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/175342"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/175342",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/175342",
+            "source_type": "github_issue",
+            "excerpt": "sql: ST_DWithin inverted join fails for POINT EMPTY geography input",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e8e5bad04d6d352fce75600f31fe3dbeab5c70ce1aad53571e3a9b1e3ac0958e",
+            "retrieved_at": "2026-09-13T03:19:40Z",
+            "first_seen": "2026-09-13T03:09:36Z",
+            "last_verified": "2026-09-13T03:19:40Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T03:19:40Z",
+            "first_seen": "2026-09-13T03:09:36Z",
+            "last_verified": "2026-09-13T03:19:40Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:09:36Z",
+        "last_verified": "2026-09-13T03:19:40Z",
+        "source_url": "https://github.com/cockroachdb/cockroach/issues/175342",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e8e5bad04d6d352fce75600f31fe3dbeab5c70ce1aad53571e3a9b1e3ac0958e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:f9388dbf1229",
+      "dbms": "cratedb",
+      "title": "Unexpected result when querying partitioned table with null values",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16577"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16577",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16577",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 VARCHAR, c1 STRING) PARTITIONED BY (c0);\r\nINSERT INTO t0(c0) VALUES ('');\r\nINSERT INTO t0(c0, c1) VALUES (NULL, NULL);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:212d4b3d404577e661c7af5079964308e0dfb470eac38d37a82ed7fa83521c8f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:212d4b3d404577e661c7af5079964308e0dfb470eac38d37a82ed7fa83521c8f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:52dcfada259e",
+      "dbms": "cratedb",
+      "title": "Unexpected result when querying with boolean comparisons",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16227"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16227",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16227",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0 (c0 INTEGER DEFAULT 1);\r\nCREATE TABLE IF NOT EXISTS t1 (c0 INTEGER);\r\nINSERT INTO t0( c0) VALUES (1);\r\nINSERT INTO t1(c0) VALUES (0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:93c2c6a78a8ef7bacf25fc92e37eba2f684746a1583b2bf7d2363484df864ee2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:93c2c6a78a8ef7bacf25fc92e37eba2f684746a1583b2bf7d2363484df864ee2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:d0fdb8f0fc33",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using CURRENT_SETTING",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16062"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16062",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16062",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nINSERT INTO t0(c0) VALUES ('a');\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:453e8b0d5e9f9812ccf82fbc54ec4f93b29cd62830b8de9e11eba72c27ea43e9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:453e8b0d5e9f9812ccf82fbc54ec4f93b29cd62830b8de9e11eba72c27ea43e9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:6826de65c056",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using DEFAULT during creating table",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16173"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16173",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16173",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t0 (c1 INT NOT NULL DEFAULT 1);\r\nCREATE TABLE  t1 (c1 INTEGER);\r\nINSERT INTO t0(c1) VALUES (1);\r\nINSERT INTO t1(c1) VALUES (2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:290092e88590710472b4dd6ce62f293fd8aa0932bada9a93a77253d0a1248796",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:290092e88590710472b4dd6ce62f293fd8aa0932bada9a93a77253d0a1248796"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:adf5ac077bdb",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using FORMAT_TYPE",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16225"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16225",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16225",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t0 ( c1 INTEGER) ;\r\nINSERT INTO t0(c1) VALUES (NULL);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a448f766049d88517fc339befba9375e177917e018ee7eb002b8f046b4d2dd1e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a448f766049d88517fc339befba9375e177917e018ee7eb002b8f046b4d2dd1e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:a375dd8bd5d8",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using PG_GET_PARTKEYDEF",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16172"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16172",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16172",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INTEGER) ;\r\nINSERT INTO t0(c0) VALUES (1);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b5da9db44b97913da02083134627778c675f0c42ac32ea245c4e076836639ca0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b5da9db44b97913da02083134627778c675f0c42ac32ea245c4e076836639ca0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:7d43bdd3413b",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using empty OBJECT literal",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16476"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16476",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16476",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 OBJECT);\r\nINSERT INTO t0(c0) VALUES (NULL);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:56031501b9e8a92ace864a50633829738d259d51112605e74a1233edab0c16c9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:56031501b9e8a92ace864a50633829738d259d51112605e74a1233edab0c16c9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:1187b036204e",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using escape character in LIKE",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16350"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16350",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16350",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c2 VARCHAR(500)) ;\r\nINSERT INTO t0(c2) VALUES  ('');\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:34605974213a73794d82f4e950719953bbbbf5cff80d65dfea8be1738a06f3f1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:34605974213a73794d82f4e950719953bbbbf5cff80d65dfea8be1738a06f3f1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:85a0141fd217",
+      "dbms": "cratedb",
+      "title": "Unexpected results when matching empty string with indexing disabled",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16567"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16567",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16567",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 VARCHAR(500) INDEX OFF, PRIMARY KEY(c0)); -- INDEX OFF is important\r\nINSERT INTO t0(c0) VALUES ('');\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:42614f3c1b0621186d80f51085fc529f58bf82475e265dcf9a05a9e17634037d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:42614f3c1b0621186d80f51085fc529f58bf82475e265dcf9a05a9e17634037d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:dd77d8b999f1",
+      "dbms": "cratedb",
+      "title": "Unexpected results when using OBJECT nulls",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16556"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16556",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16556",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t1(c0 OBJECT NULL);\r\nINSERT INTO t1(c0) VALUES (NULL);\r\nREFRESH TABLE t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:8d40437d6d16f3d01bba13fd36b4e561918b359eccf3f04128e420d76f29eaf8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:8d40437d6d16f3d01bba13fd36b4e561918b359eccf3f04128e420d76f29eaf8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:13530983298f",
+      "dbms": "cratedb",
+      "title": "Unexpected results when using system information functions",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16026"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16026",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16026",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INTEGER);\r\nINSERT INTO t0(c1) VALUES (-1);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:8769d3222d1eb2783cdcd92934b4be6f5db443155e7d54e8f106e3ec416b413f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:8769d3222d1eb2783cdcd92934b4be6f5db443155e7d54e8f106e3ec416b413f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:9aafc3f52fe8",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using CASE statement",
+      "reported_date": "2023-09-11",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15029"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15029",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15029",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 INT);\r\nINSERT INTO t1 VALUES (NULL);\r\n\r\nSELECT * FROM t1; -- NULL",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:22034d8b4a9ff07b24f9d4d2dd6841d57e0d79267b21d0a9b68355046d250753",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:22034d8b4a9ff07b24f9d4d2dd6841d57e0d79267b21d0a9b68355046d250753"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:4ed90a7606e6",
+      "dbms": "cratedb",
+      "title": "Couldn't create execution plan from logical plan because of: Index 2 out of bounds for length 2",
+      "reported_date": "2023-10-05",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/14805"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/14805",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/14805",
+            "source_type": "github_issue",
+            "excerpt": "Couldn't create execution plan from logical plan because of: Index 2 out of bounds for length 2",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3902351168580eb217788472dc4d8181c66d8013ed348df0aa0584366426086f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/14805",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3902351168580eb217788472dc4d8181c66d8013ed348df0aa0584366426086f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:94004a3ade60",
+      "dbms": "cratedb",
+      "title": "Incorrect result returned by `UNION`",
+      "reported_date": "2023-10-05",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/14807"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/14807",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/14807",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result returned by `UNION`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:234a464eea52db126c0b6c3fcd526d44471200631f72810ed5cba927a53c1ffd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/14807",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:234a464eea52db126c0b6c3fcd526d44471200631f72810ed5cba927a53c1ffd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:9dab78dce7d8",
+      "dbms": "cratedb",
+      "title": "Unexpected Exception with `ORDER BY` clause with `UNION`",
+      "reported_date": "2023-10-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/14815"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/14815",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/14815",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Exception with `ORDER BY` clause with `UNION`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:be9df1ca4e59a0c752c4fca411dd2c79c717cf6c22dab53c61baa94095eaf0bc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/14815",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:be9df1ca4e59a0c752c4fca411dd2c79c717cf6c22dab53c61baa94095eaf0bc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:0ffe5c5c66f1",
+      "dbms": "cratedb",
+      "title": "Unexpected Results when Using AND Operator",
+      "reported_date": "2023-11-14",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15050"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15050",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15050",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT);\r\nINSERT INTO t0(c1) VALUES (-1);\r\n\r\nSELECT * FROM t0; -- c0: NULL, c1: -1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:2566410400c28ea6c7598eed35b1375982342359b0c32f1b0f0eee7ef42dd2d4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:2566410400c28ea6c7598eed35b1375982342359b0c32f1b0f0eee7ef42dd2d4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:06e43e037cca",
+      "dbms": "cratedb",
+      "title": "Unexpected Results when Using Concat Operator",
+      "reported_date": "2023-11-22",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15083"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15083",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15083",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0(c0) VALUES (NULL);\r\n\r\nSELECT * FROM t0; -- NULL",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:0625f6daf2e5b9e4aed15ee01f4fe549a36c5a2b08703e8848a08e9cb7f8ab2b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:0625f6daf2e5b9e4aed15ee01f4fe549a36c5a2b08703e8848a08e9cb7f8ab2b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:f431725eea57",
+      "dbms": "cratedb",
+      "title": "Unexpected Results Using Boolean",
+      "reported_date": "2023-11-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15120"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15120",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15120",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT,c2 INT);\r\nINSERT INTO t0(c1) VALUES (1);\r\n\r\nSELECT * FROM t0; -- 1 NULL",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:1acd135622017981e6ca49d4260ae76c5cba018939941b11ea6c4bc6db523539",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:1acd135622017981e6ca49d4260ae76c5cba018939941b11ea6c4bc6db523539"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:d334bedc1086",
+      "dbms": "cratedb",
+      "title": "Unexpected Results Using NULLIF",
+      "reported_date": "2023-11-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15121"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15121",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15121",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\nINSERT INTO t0(c0) VALUES (NULL);\r\n\r\nSELECT * FROM t0; -- NULL",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a6688ea38b6f05804159966c8e8ee9e5ba4525134c0e6f6ea80ed5506c49937a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a6688ea38b6f05804159966c8e8ee9e5ba4525134c0e6f6ea80ed5506c49937a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:14230453c73b",
+      "dbms": "cratedb",
+      "title": "Inconsistent Results executing multiple statements at once",
+      "reported_date": "2023-12-15",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15199"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15199",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15199",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1( c1 INT, PRIMARY KEY(c1));\r\n    INSERT INTO t1 (c1) VALUES (0);\r\n\r\n    SELECT * FROM t1;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8f6f75de5ffe2ae91a66e5795e46cccb0fb50d82c495d90a551522627234d750",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/15199",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8f6f75de5ffe2ae91a66e5795e46cccb0fb50d82c495d90a551522627234d750"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:d2bfb567db90",
+      "dbms": "cratedb",
+      "title": "Unexpected result when Concatenating NULL",
+      "reported_date": "2023-12-16",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15202"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15202",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15202",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INTEGER);\r\nINSERT INTO t1 (c0) VALUES (NULL);\r\nREFRESH TABLE t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ca9d77be777a1d2edff911e4fdd88ea2da7ef7a7d7df69bb6afc368e010abc0a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ca9d77be777a1d2edff911e4fdd88ea2da7ef7a7d7df69bb6afc368e010abc0a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:e7aafc16eee8",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using BETWEEN AND",
+      "reported_date": "2023-12-16",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15201"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15201",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15201",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN);\r\nINSERT INTO t0 (c0) VALUES (true);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5f2b437d7568fb97ac30e6620848565a12800659e0d48023e117c802bbe6cac8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5f2b437d7568fb97ac30e6620848565a12800659e0d48023e117c802bbe6cac8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:41d221d40056",
+      "dbms": "cratedb",
+      "title": "Unexpected Result when using IS NOT NULL",
+      "reported_date": "2023-12-24",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15265"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15265",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15265",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1( c1 BOOLEAN);\r\nINSERT INTO t1 (c1) VALUES (NULL);\r\nREFRESH TABLE t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:40a0680fb60b65169d9c1f6a82c57f1135af35828a4955c3a64d76a3b0525de0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:40a0680fb60b65169d9c1f6a82c57f1135af35828a4955c3a64d76a3b0525de0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:398db4e0cf2a",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using string comparison AND IS NULL",
+      "reported_date": "2024-01-18",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15395"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15395",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15395",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 VARCHAR, PRIMARY KEY(c0));\r\nINSERT INTO t1 (c0) VALUES ('a');\r\nREFRESH TABLE t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:571113edefe24897be699e80eb64d7d99253c5c6fd22f77e6d24c6f7d0b91572",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:571113edefe24897be699e80eb64d7d99253c5c6fd22f77e6d24c6f7d0b91572"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:9d9edba8769e",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using CONCAT_WS containing NULL",
+      "reported_date": "2024-01-26",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15459"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15459",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15459",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c1 BOOLEAN);\r\nINSERT INTO t1 (c1) VALUES (null);\r\nREFRESH TABLE t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b4752003e6f48311f5d37b8a483c5bf1dacc78dc08b8705e77a8d4f35d599295",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b4752003e6f48311f5d37b8a483c5bf1dacc78dc08b8705e77a8d4f35d599295"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:5f255ee3cab3",
+      "dbms": "cratedb",
+      "title": "Unexpected results when using AND",
+      "reported_date": "2024-01-26",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15458"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15458",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15458",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR, c1 INT, PRIMARY KEY(c0));\r\nINSERT INTO t0 (c0, c1) VALUES ('a', 1);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3aff78bc8a08c51768ac8ee45951ce08a272634a300bfaa07a2bfce49316a42c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3aff78bc8a08c51768ac8ee45951ce08a272634a300bfaa07a2bfce49316a42c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:8f1fc2e7710e",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using % with NULL",
+      "reported_date": "2024-02-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15500"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15500",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15500",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0( c0 INT);\r\nINSERT INTO t0 (c0) VALUES (2);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d3295ed36e9fbb773687b4098655e259235fa68d0f435118cdd51e94f417db2e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d3295ed36e9fbb773687b4098655e259235fa68d0f435118cdd51e94f417db2e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:6d12997cc8fd",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using `LTRIM`",
+      "reported_date": "2024-02-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15498"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15498",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15498",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT, PRIMARY KEY(c0));\r\nINSERT INTO t1 (c0) VALUES (0);\r\nREFRESH TABLE t1;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c0c5850aa97dbd405135d0fbce4be7c8a1b4c8e9dac6a9ad7da3d6983324026a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/15498",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c0c5850aa97dbd405135d0fbce4be7c8a1b4c8e9dac6a9ad7da3d6983324026a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:56983b7ab537",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using `REPLACE`",
+      "reported_date": "2024-02-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15499"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15499",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15499",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500));\r\nINSERT INTO t0 (c0) VALUES ('a');\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0ea062c7dc9c28e5379abd20e40e654f2b5975388e2e2955c9ee1a26674e66d2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/15499",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0ea062c7dc9c28e5379abd20e40e654f2b5975388e2e2955c9ee1a26674e66d2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:e4bac65cc956",
+      "dbms": "cratedb",
+      "title": "Weird string substitution when using JDBC",
+      "reported_date": "2024-02-16",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15566"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15566",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15566",
+            "source_type": "github_issue",
+            "excerpt": "Weird string substitution when using JDBC",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:dad154d808066a77b697d090bf8e05c5048fbeda570207181f1ce9645ac12e0c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/15566",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dad154d808066a77b697d090bf8e05c5048fbeda570207181f1ce9645ac12e0c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:b25732b3b99d",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using OR",
+      "reported_date": "2024-02-22",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15592"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15592",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15592",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR, c1 BOOLEAN, PRIMARY KEY(c0)); -- PK is needed\r\nINSERT INTO t0 (c0, c1) VALUES ('a', true);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:73719505f320ee0302434f3db3d689e3f5c0bfaaa3f377dbc6a851f22b9c18eb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:73719505f320ee0302434f3db3d689e3f5c0bfaaa3f377dbc6a851f22b9c18eb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:f9efeaebfc0d",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using INNER JOIN and CASE in filter",
+      "reported_date": "2024-02-27",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15613"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15613",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15613",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\nCREATE TABLE t1(c0 VARCHAR, c1 BOOLEAN);\r\nINSERT INTO t0 (c0) VALUES (0);\r\nINSERT INTO t1 (c1, c0) VALUES (true, 'a');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:56ef4a4b76908d9e6af741f107b355577feb23f9331c4f9318266a78a5fd8f8b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:56ef4a4b76908d9e6af741f107b355577feb23f9331c4f9318266a78a5fd8f8b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:447664d9693b",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using LIKE",
+      "reported_date": "2024-03-22",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15743"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15743",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15743",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 VARCHAR);\r\nINSERT INTO t0 (c1) VALUES ('');\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:87cceed7043707c48e47848a69544ae21df5e7b57ddd3696216f76111b5d0814",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:87cceed7043707c48e47848a69544ae21df5e7b57ddd3696216f76111b5d0814"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:7609ddbb65bb",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using LTRIM with NULL as second argument",
+      "reported_date": "2024-08-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15527"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15527",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15527",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c1 INT);\r\nINSERT INTO t1 (c1) VALUES (null);\r\nREFRESH TABLE t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:4ea2cd9c75f58f200be853ce327ca1b7dce76d259865b923fc17552bd39a0c90",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:4ea2cd9c75f58f200be853ce327ca1b7dce76d259865b923fc17552bd39a0c90"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:2b90e92f7ff6",
+      "dbms": "cratedb",
+      "title": "Unexpected result when casting integer as boolean",
+      "reported_date": "2024-09-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15537"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15537",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15537",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER, PRIMARY KEY(c0)); -- PK is needed\r\nINSERT INTO t0 (c0) VALUES (2);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:64323547293782aea160b657a37c3a6a2f340c6cd35a5926686a9b60df8d2dd0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:64323547293782aea160b657a37c3a6a2f340c6cd35a5926686a9b60df8d2dd0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:3b9ba7212c70",
+      "dbms": "cratedb",
+      "title": "Unexpected result when querying partitioned table with `CASE WHEN`",
+      "reported_date": "2024-09-12",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16614"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16614",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16614",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t1(c0 TEXT, c1 TEXT) PARTITIONED BY (c0);\r\nINSERT INTO t1(c1, c0) VALUES ('B', 'B'), ('C', 'C'), ('D', ''), ('A', 'A');\r\nREFRESH TABLE t1;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ad2fcc2a8354a0a0db29d58e0e19249ea19e0ea88e691a397e7bf1891a8773ef",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/16614",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ad2fcc2a8354a0a0db29d58e0e19249ea19e0ea88e691a397e7bf1891a8773ef"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:48d2d39bc7ba",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using bigint in `ORDER BY`",
+      "reported_date": "2024-09-12",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16615"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16615",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16615",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c1 LONG );\r\nINSERT INTO t0(c1) VALUES (9223372036854775807);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b329f9dc137527ff51630d0fbebd02d164ef51cbee5b23a37ba2d8757d2de1bf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/16615",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b329f9dc137527ff51630d0fbebd02d164ef51cbee5b23a37ba2d8757d2de1bf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:a37824e136f1",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using `ARRAY(TEXT)` in `IN`",
+      "reported_date": "2024-09-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16652"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16652",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16652",
+            "source_type": "github_issue",
+            "excerpt": "CREATE   TABLE  t0(c1 ARRAY(TEXT));\r\nINSERT INTO t0(c1) VALUES (ARRAY['hello','world']);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:53a945b91ee96c2288d034d661e16d039c1626471f4f00b624e4dd2b9e3da301",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/16652",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:53a945b91ee96c2288d034d661e16d039c1626471f4f00b624e4dd2b9e3da301"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:eb65fede5a5a",
+      "dbms": "cratedb",
+      "title": "OutOfMemory when creating table with large replicas",
+      "reported_date": "2024-09-25",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16668"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16668",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16668",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 INT)WITH (number_of_replicas = 1000);\r\n-- SQLParseException[Validation Failed: 1: this action would add [4004] total shards, but this cluster currently has [0]/[1000] maximum shards open;]\r\nCREATE  TABLE  t0(c0 INT)WITH (number_of_replicas = 1925152226);\r\n-- hang and crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ac29d14d8c284bffa1f55dd862bc308f613335b49d77343b949620798a7357a7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/16668",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ac29d14d8c284bffa1f55dd862bc308f613335b49d77343b949620798a7357a7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:d524c9677716",
+      "dbms": "cratedb",
+      "title": "Unexpected result when comparing `CHAR` with `''`",
+      "reported_date": "2024-09-29",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16689"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16689",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16689",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 CHAR(3));\r\nCREATE  TABLE  t1(c0 CHAR(3));\r\nINSERT INTO t0(c0) VALUES (' ');\r\nINSERT INTO t1(c0) VALUES ('a');",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8c58a414322b9202305eef31795af6109408b5a5371f03b918fcbda993b294ab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/16689",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8c58a414322b9202305eef31795af6109408b5a5371f03b918fcbda993b294ab"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:5882a47e50e9",
+      "dbms": "cratedb",
+      "title": "Unexpected result when comparing `IP`",
+      "reported_date": "2024-09-29",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16690"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16690",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16690",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 IP);\r\nINSERT INTO t0(c0) VALUES ('192.168.0.1');\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ffcc6e4293514580ca47eb64208215460697fc19ba5f6309ee31666d21ff29f7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/16690",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ffcc6e4293514580ca47eb64208215460697fc19ba5f6309ee31666d21ff29f7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:641dbd7bcbb2",
+      "dbms": "cratedb",
+      "title": "Unexpected result when comparing `FLOAT` in a partitioned table",
+      "reported_date": "2024-10-03",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16705"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16705",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16705",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 FLOAT, PRIMARY KEY(c0)) PARTITIONED BY (c0);\r\nINSERT INTO t0(c0) VALUES (1.2);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:91a5e9c3a4de72076b76142c089bd03c0dd1f6d2ce6fa0a6335e4977513439d0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/16705",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:91a5e9c3a4de72076b76142c089bd03c0dd1f6d2ce6fa0a6335e4977513439d0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:a58fcfa7f37b",
+      "dbms": "cratedb",
+      "title": "Unexpected Result when using IN",
+      "reported_date": "2024-11-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15326"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15326",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15326",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR, PRIMARY KEY(c0));\r\nINSERT INTO t0 (c0) VALUES (true);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:1588c2e0143d1a3bc5ae579a18e400a43176b0d6d396e10d16126a647d688734",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:1588c2e0143d1a3bc5ae579a18e400a43176b0d6d396e10d16126a647d688734"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:e440f597ae79",
+      "dbms": "cratedb",
+      "title": "Unexpected Result when using LEFT JOIN",
+      "reported_date": "2024-11-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15324"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15324",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15324",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\nCREATE TABLE t1(c1 INTEGER);\r\nINSERT INTO t0 (c0) VALUES (0);\r\nINSERT INTO t1 (c1) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:84456f6d4ed55f4af37da19575ec4b583278ec0a24d76f643e84b8342116eb08",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:84456f6d4ed55f4af37da19575ec4b583278ec0a24d76f643e84b8342116eb08"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:3024b78be876",
+      "dbms": "cratedb",
+      "title": "Unexpected Results Using NULLIF with VARCHAR",
+      "reported_date": "2024-11-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/15325"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/15325",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/15325",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0( c1 VARCHAR(500));\r\nINSERT INTO t0 (c1) VALUES (NULL);\r\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:1d9212dffab6c6fcada2fd8aa4330931e84b91f2e8ef6754bdec8f6859ad08b3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:1d9212dffab6c6fcada2fd8aa4330931e84b91f2e8ef6754bdec8f6859ad08b3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:275baf1b0da7",
+      "dbms": "cratedb",
+      "title": "Potential hang when `LEFT JOIN` a subquery with `EXP`",
+      "reported_date": "2024-11-11",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16952"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16952",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16952",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0( c1 FLOAT ); \n\n-- ok\nSELECT * FROM t0 LEFT JOIN (SELECT * FROM t0 WHERE ((EXP(-1110102730)) <> t0.c1)) AS sub0  ON true;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5cdb99081ea0892fb80686d9eec810718988d8932d49f0f3275539152807375a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/16952",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5cdb99081ea0892fb80686d9eec810718988d8932d49f0f3275539152807375a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:0ffb180bf545",
+      "dbms": "cratedb",
+      "title": "Unexpected result when `RIGHT JOIN` a subquery",
+      "reported_date": "2024-11-11",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/16951"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/16951",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/16951",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c1 VARCHAR(500));\nCREATE  TABLE  t1(c0 VARCHAR(500));\nINSERT INTO t0(c1) VALUES ('');\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9c3019f56a2f435568abb697df952cc98d94c12185f0c516389b7cabce38097d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/16951",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9c3019f56a2f435568abb697df952cc98d94c12185f0c516389b7cabce38097d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:a245c380c8c8",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using multiple `RIGHT JOIN`",
+      "reported_date": "2024-11-30",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/17089"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/17089",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/17089",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 BOOLEAN , c1 INT );\nCREATE  TABLE  t1(c0 BOOLEAN);\nINSERT INTO t0(c0, c1) VALUES (false, NULL);\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2472cff53577f3f88731fa44ab77bac0058d34aab778665521d845d8ae0b9ecd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/17089",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2472cff53577f3f88731fa44ab77bac0058d34aab778665521d845d8ae0b9ecd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:03ace015868a",
+      "dbms": "cratedb",
+      "title": "Unexpected results when querying `BOOLEAN` values",
+      "reported_date": "2025-02-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/17379"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/17379",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/17379",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE t0(c0 BOOLEAN , PRIMARY KEY(c0));\nINSERT INTO t0(c0) VALUES (true), (1=1); -- not sure if expected\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0475e5583e7e917ad67b65278cb302bea34d4e79217ecc8e8a44b71739aacd43",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/17379",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0475e5583e7e917ad67b65278cb302bea34d4e79217ecc8e8a44b71739aacd43"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:9bd444750e4b",
+      "dbms": "cratedb",
+      "title": "Unexpected results when using `ARRAY_POSITION`",
+      "reported_date": "2025-02-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/17378"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/17378",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/17378",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 INT , c1 ARRAY(STRING));\nINSERT INTO t0(c0, c1) VALUES (1, ['alpha', 'beta', 'gamma']);\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5d14b8559103636310f71815564e15e5f41dd572e60d9a320bd59a10f1dd9800",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/17378",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5d14b8559103636310f71815564e15e5f41dd572e60d9a320bd59a10f1dd9800"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:ebe17adcf424",
+      "dbms": "cratedb",
+      "title": "Unexpected results when using `LEFT JOIN` with a subquery",
+      "reported_date": "2025-02-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/17380"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/17380",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/17380",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 INT , c1 BOOLEAN);\nINSERT INTO t0(c0, c1) VALUES (1, TRUE);\nINSERT INTO t0(c0) VALUES (2);\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1db6d453f758cac3e318ac101370c50f3ced8dce3d689c46765131d40b3f2a34",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/17380",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1db6d453f758cac3e318ac101370c50f3ced8dce3d689c46765131d40b3f2a34"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:5f11379517aa",
+      "dbms": "cratedb",
+      "title": "Unexpected result when comparing `CHAR` values with newline",
+      "reported_date": "2025-04-10",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/17765"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/17765",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/17765",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 CHAR(10));\nINSERT INTO t0(c0) VALUES  ('');\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0bc16f460d3fc2d454efc527c26e336d9c5528c524558692443949cd1b2cb85c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/17765",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0bc16f460d3fc2d454efc527c26e336d9c5528c524558692443949cd1b2cb85c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:39d77a005e81",
+      "dbms": "cratedb",
+      "title": "Unexpected result when querying a partitioned table by a `BOOLEAN` column",
+      "reported_date": "2025-04-10",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/17766"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/17766",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/17766",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t1(c0 BOOLEAN ,PRIMARY KEY(c0))PARTITIONED BY (c0);\nINSERT INTO t1(c0) VALUES (FALSE);\nREFRESH TABLE t1;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c57a72962ba45cd4940b2b62d517fb5aa6e8f5c519278858dbc4308a82657e47",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/17766",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c57a72962ba45cd4940b2b62d517fb5aa6e8f5c519278858dbc4308a82657e47"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:09cd73948d4f",
+      "dbms": "cratedb",
+      "title": "Unexpected result when using `IP` addresses",
+      "reported_date": "2025-04-10",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/17764"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/17764",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/17764",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 IP , PRIMARY KEY(c0));\nINSERT INTO t0(c0) VALUES ('::ffff:192.168.1.1');\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0be52ec055b4b55a38e317193a9deea0558c331bac20c8c3a04ac40a67089a63",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/17764",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0be52ec055b4b55a38e317193a9deea0558c331bac20c8c3a04ac40a67089a63"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:7b05447f10b6",
+      "dbms": "cratedb",
+      "title": "Unexpected Left Join Result",
+      "reported_date": "2025-10-11",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/18680"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/18680",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/18680",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 BOOLEAN, c1 VARCHAR);\nINSERT INTO t0(c0) VALUES (NULL);\nINSERT INTO t0(c1, c0) VALUES ('', true);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:20daa7f3e16f62f1cf7adab34dfa26ff67fd56d351560f5c9c2c464241754167",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:20daa7f3e16f62f1cf7adab34dfa26ff67fd56d351560f5c9c2c464241754167"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:dedbe3b1e39e",
+      "dbms": "cratedb",
+      "title": "Unexpected Inner Join Result",
+      "reported_date": "2025-12-19",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/18860"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/18860",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/18860",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN);\nCREATE TABLE t1(c0 VARCHAR);\n\nINSERT INTO t1(c0) VALUES ('1');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3202a55a19e56aa33e8f3cd12fdec3a35ce9a798bc05a11638b0d930737bb80a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3202a55a19e56aa33e8f3cd12fdec3a35ce9a798bc05a11638b0d930737bb80a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:f1cc5544f539",
+      "dbms": "cratedb",
+      "title": "Unexpected ANTI JOIN Result",
+      "reported_date": "2026-05-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/18873"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/18873",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/18873",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 BOOLEAN , c1 INT , c2 VARCHAR, PRIMARY KEY(c0));\nINSERT INTO t0(c1, c0, c2) VALUES (-8, true, '1');\nINSERT INTO t0(c1, c0) VALUES (1, false);\nREFRESH TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:2e3dc2a8924920897191717430247f23ad451c68790fb90aa6bd2d30b798b7c8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:2e3dc2a8924920897191717430247f23ad451c68790fb90aa6bd2d30b798b7c8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:1d0133de195b",
+      "dbms": "cratedb",
+      "title": "`WHERE <bare boolean column>` silently dropped when the column is projected with an alias through a join + filtered subquery",
+      "reported_date": "2026-07-31",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19837"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19837",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19837",
+            "source_type": "github_issue",
+            "excerpt": "`WHERE <bare boolean column>` silently dropped when the column is projected with an alias through a join + filtered subquery",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2e2fb011459fe4383f3994c144162a0e2ccccc0849bec079a17d34d3ee2a76db",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19837",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2e2fb011459fe4383f3994c144162a0e2ccccc0849bec079a17d34d3ee2a76db"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:5710c4eca637",
+      "dbms": "cratedb",
+      "title": "UNION ALL + ORDER BY on an unselected column fails on `XX000`",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19852"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19852",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19852",
+            "source_type": "github_issue",
+            "excerpt": "UNION ALL + ORDER BY on an unselected column fails on `XX000`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b7912c5480df77af6a8377169d186ccff809a35778edb3fa65e64249260dfc32",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19852",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b7912c5480df77af6a8377169d186ccff809a35778edb3fa65e64249260dfc32"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:3e8d06e2c205",
+      "dbms": "cratedb",
+      "title": "`RANK()` / `DENSE_RANK()` with explicit frame returns wrong answer",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19853"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19853",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19853",
+            "source_type": "github_issue",
+            "excerpt": "`RANK()` / `DENSE_RANK()` with explicit frame returns wrong answer",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:775b971fc6585f2d23b3579d65a876dbb31f76418480b26e2730f53d6fa1b0e5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19853",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:775b971fc6585f2d23b3579d65a876dbb31f76418480b26e2730f53d6fa1b0e5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:1fe55afff407",
+      "dbms": "cratedb",
+      "title": "LAG ignores a row-dependent offset expression",
+      "reported_date": "2026-08-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19870"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19870",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19870",
+            "source_type": "github_issue",
+            "excerpt": "LAG ignores a row-dependent offset expression",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c2a1dfac0c9f382a747416cf68344bab2ebc019a8873dc140bf8a8a1d1f5cbf1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19870",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c2a1dfac0c9f382a747416cf68344bab2ebc019a8873dc140bf8a8a1d1f5cbf1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:8704b7e7afb2",
+      "dbms": "cratedb",
+      "title": "Merge filter and move filter on a `PARTITIONED` table drop rows",
+      "reported_date": "2026-08-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19855"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19855",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19855",
+            "source_type": "github_issue",
+            "excerpt": "Merge filter and move filter on a `PARTITIONED` table drop rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:10d5e822843ceada75a9353f5e6acd871644b1e27071bfdff573aa6962ff97c5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19855",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:10d5e822843ceada75a9353f5e6acd871644b1e27071bfdff573aa6962ff97c5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:a395d40f374f",
+      "dbms": "cratedb",
+      "title": "Window `sum` returns `0` for a frame containing only `NULL`",
+      "reported_date": "2026-08-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19871"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19871",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19871",
+            "source_type": "github_issue",
+            "excerpt": "Window `sum` returns `0` for a frame containing only `NULL`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:68c413a5b1dad691c59ddac5ab4a39f51b4ac9559251ec527fadeaf896cb4c77",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19871",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:68c413a5b1dad691c59ddac5ab4a39f51b4ac9559251ec527fadeaf896cb4c77"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:fbe119c032ac",
+      "dbms": "cratedb",
+      "title": "Window aggregate `FILTER` crashes when the filter predicate is `NULL`",
+      "reported_date": "2026-08-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19869"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19869",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19869",
+            "source_type": "github_issue",
+            "excerpt": "Window aggregate `FILTER` crashes when the filter predicate is `NULL`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:41805b12aa0913250791a30d514dd839a250fccac08b00e955d14dafaca82353",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19869",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:41805b12aa0913250791a30d514dd839a250fccac08b00e955d14dafaca82353"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:414fee8eb6e9",
+      "dbms": "cratedb",
+      "title": "`DESC RANGE` window frames can produce negative counts",
+      "reported_date": "2026-08-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19872"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19872",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19872",
+            "source_type": "github_issue",
+            "excerpt": "`DESC RANGE` window frames can produce negative counts",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1d37c7537bee26baad07596ef000e65e7f43c18a7729077a898725679e35e7e0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19872",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1d37c7537bee26baad07596ef000e65e7f43c18a7729077a898725679e35e7e0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:2e3a88d2f1af",
+      "dbms": "cratedb",
+      "title": "`FULL OUTER JOIN` returns wrong answer due to `optimizer_rewrite_filter_on_outer_join_to_inner_join`",
+      "reported_date": "2026-08-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19854"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19854",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19854",
+            "source_type": "github_issue",
+            "excerpt": "`FULL OUTER JOIN` returns wrong answer due to `optimizer_rewrite_filter_on_outer_join_to_inner_join`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d25d140d9dab050e6a4ff76795d8f9f1bd9eb0492bac1c58dcd5341da3f71809",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19854",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d25d140d9dab050e6a4ff76795d8f9f1bd9eb0492bac1c58dcd5341da3f71809"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:a8097909314c",
+      "dbms": "cratedb",
+      "title": "`UNION ALL` branch reading a relation with an extra column shifts column position",
+      "reported_date": "2026-08-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19856"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19856",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19856",
+            "source_type": "github_issue",
+            "excerpt": "`UNION ALL` branch reading a relation with an extra column shifts column position",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1883cd58bed42f7e4814afa9d1a4532077f1094f3bf272837b051734addeabfa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19856",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1883cd58bed42f7e4814afa9d1a4532077f1094f3bf272837b051734addeabfa"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:a2fd5767bfdd",
+      "dbms": "cratedb",
+      "title": "`string_agg` over a single-row moving frame raises an internal error",
+      "reported_date": "2026-08-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19873"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19873",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19873",
+            "source_type": "github_issue",
+            "excerpt": "`string_agg` over a single-row moving frame raises an internal error",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a923264bfebdd73222670d6b70adc520f5635d986fc62125cce35a919b94b82e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19873",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a923264bfebdd73222670d6b70adc520f5635d986fc62125cce35a919b94b82e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:fc508c8c05bb",
+      "dbms": "cratedb",
+      "title": "Range predicates on a `NUMERIC`/DECIMAL column with `INDEX OFF` silently return zero rows",
+      "reported_date": "2026-08-13",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19921"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19921",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19921",
+            "source_type": "github_issue",
+            "excerpt": "Range predicates on a `NUMERIC`/DECIMAL column with `INDEX OFF` silently return zero rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e907513e3a0064b40695e171882d4efff46b20bd5c1eb0d256ab824fca694eb5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19921",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e907513e3a0064b40695e171882d4efff46b20bd5c1eb0d256ab824fca694eb5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:4aacaa43f13b",
+      "dbms": "cratedb",
+      "title": "UNION ALL with an empty equi-join branch makes ORDER BY leak rows",
+      "reported_date": "2026-08-13",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19913"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19913",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19913",
+            "source_type": "github_issue",
+            "excerpt": "UNION ALL with an empty equi-join branch makes ORDER BY leak rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:30eaa6ea1c1ffd334d8a63a3aacbc597c913a07c6d359e72526927bfce3bbae9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19913",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:30eaa6ea1c1ffd334d8a63a3aacbc597c913a07c6d359e72526927bfce3bbae9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:fd1aac3e4636",
+      "dbms": "cratedb",
+      "title": "`ROUND(x, N)` with a large-magnitude negative `N` hangs",
+      "reported_date": "2026-08-13",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19918"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19918",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19918",
+            "source_type": "github_issue",
+            "excerpt": "`ROUND(x, N)` with a large-magnitude negative `N` hangs",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9d8c7fd0c318357d305440c8ba31e7bbd9f5a10354aa1e7c9b5cd1037729a562",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19918",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9d8c7fd0c318357d305440c8ba31e7bbd9f5a10354aa1e7c9b5cd1037729a562"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:ad4f0caf67b1",
+      "dbms": "cratedb",
+      "title": "Inconsistent behavior of `starts_with(col, '')` on a TEXT column",
+      "reported_date": "2026-08-14",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19923"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19923",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19923",
+            "source_type": "github_issue",
+            "excerpt": "Inconsistent behavior of `starts_with(col, '')` on a TEXT column",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6dfb616aef731ae3df1bb65805f821ba8b6ebf0d1e1cfa3c16a6a3a22262fb4a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19923",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6dfb616aef731ae3df1bb65805f821ba8b6ebf0d1e1cfa3c16a6a3a22262fb4a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:6509f500eb29",
+      "dbms": "cratedb",
+      "title": "`NULL = ALL (empty)` drops the row when the compared column is a window output",
+      "reported_date": "2026-08-14",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19922"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19922",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19922",
+            "source_type": "github_issue",
+            "excerpt": "`NULL = ALL (empty)` drops the row when the compared column is a window output",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:beb6669c83e049c1b330bc96551571adc94375a69c38364deceddaa2b754ea9b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19922",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:beb6669c83e049c1b330bc96551571adc94375a69c38364deceddaa2b754ea9b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:c3a32c4147fe",
+      "dbms": "cratedb",
+      "title": "Outer filter pushdown changes window results",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19982"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19982",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19982",
+            "source_type": "github_issue",
+            "excerpt": "Outer filter pushdown changes window results",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:eb19322c25803e84d29f7843799be5d815fde74b41eaaea20b9d53a26b840b0a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19982",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:eb19322c25803e84d29f7843799be5d815fde74b41eaaea20b9d53a26b840b0a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:ec2c78b85840",
+      "dbms": "cratedb",
+      "title": "`SMALLINT` `RANGE` window boundaries raise `ClassCastException`",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19981"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19981",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19981",
+            "source_type": "github_issue",
+            "excerpt": "`SMALLINT` `RANGE` window boundaries raise `ClassCastException`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3c82b798abc62c143d8d331c5f51818f5bdc65d979fbc64d4ad9c510e5f899c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19981",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3c82b798abc62c143d8d331c5f51818f5bdc65d979fbc64d4ad9c510e5f899c2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:3a9fdbbb0003",
+      "dbms": "cratedb",
+      "title": "`array_agg` ignores a running window frame",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/19980"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/19980",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/19980",
+            "source_type": "github_issue",
+            "excerpt": "`array_agg` ignores a running window frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:95fff3eed1fe9a954a5a660d6c1d24bcec08b13092db2acb4944918fc1bfddaa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/19980",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:95fff3eed1fe9a954a5a660d6c1d24bcec08b13092db2acb4944918fc1bfddaa"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:644fa0d51472",
+      "dbms": "cratedb",
+      "title": "REGEXP filter merged into `Collect` over an `INDEX OFF` text column matches no rows",
+      "reported_date": "2026-08-19",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/20003"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/20003",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/20003",
+            "source_type": "github_issue",
+            "excerpt": "REGEXP filter merged into `Collect` over an `INDEX OFF` text column matches no rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:34c675d6484eacc73f4a145b4dac58b64c1949e0f2e0c7259c4c268637b10f95",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/20003",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:34c675d6484eacc73f4a145b4dac58b64c1949e0f2e0c7259c4c268637b10f95"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:eb22942b4236",
+      "dbms": "cratedb",
+      "title": "`CROSS JOIN … WHERE l.x = r.x` drops a correlated `Filter` after rewrite to HashJoin",
+      "reported_date": "2026-08-19",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/20002"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/20002",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/20002",
+            "source_type": "github_issue",
+            "excerpt": "`CROSS JOIN … WHERE l.x = r.x` drops a correlated `Filter` after rewrite to HashJoin",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ef40f9ed34589952bc08bda1d41918d53c5a8cb99dfe1c525f1c542489f0ff8a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/20002",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ef40f9ed34589952bc08bda1d41918d53c5a8cb99dfe1c525f1c542489f0ff8a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:859ee4af436a",
+      "dbms": "cratedb",
+      "title": "HashJoin keeps an intra-side predicate",
+      "reported_date": "2026-08-21",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/20023"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/20023",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/20023",
+            "source_type": "github_issue",
+            "excerpt": "HashJoin keeps an intra-side predicate",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:bb81f516c40996cd793aab1051c4b94896b0f91bf74b08c1b0928a79b385fd09",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/20023",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bb81f516c40996cd793aab1051c4b94896b0f91bf74b08c1b0928a79b385fd09"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:f61ca7f801b2",
+      "dbms": "cratedb",
+      "title": "UNION filter pushdown breaks branch types",
+      "reported_date": "2026-08-21",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/20022"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/20022",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/20022",
+            "source_type": "github_issue",
+            "excerpt": "UNION filter pushdown breaks branch types",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7f7c948123d16250e298b52e660340d7ab203cbcfde84a661aab4aed53e904d4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/20022",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7f7c948123d16250e298b52e660340d7ab203cbcfde84a661aab4aed53e904d4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cratedb:23db16adf974",
+      "dbms": "cratedb",
+      "title": "`string_agg` leaves stale delimiters with row-dependent window delimiters",
+      "reported_date": "2026-08-21",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/crate/crate/issues/20021"
+      },
+      "primary_url": "https://github.com/crate/crate/issues/20021",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/crate/crate/issues/20021",
+            "source_type": "github_issue",
+            "excerpt": "`string_agg` leaves stale delimiters with row-dependent window delimiters",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:cc41684bf5e3b5b3f21dd6f8c9af833a2a6caf10a5822bbde2bcce7c45d0cc61",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/crate/crate/issues/20021",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cc41684bf5e3b5b3f21dd6f8c9af833a2a6caf10a5822bbde2bcce7c45d0cc61"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:cubrid:d86a4653ced7",
+      "dbms": "cubrid",
+      "title": "ORDER BY causes row to be omitted with index and case expression",
+      "reported_date": "2023-12-11",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "report": "http://jira.cubrid.org/browse/CBRD-25067"
+      },
+      "primary_url": "https://jira.cubrid.org/browse/CBRD-25067",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "http://jira.cubrid.org/browse/CBRD-25067",
+            "source_type": "issue_tracker",
+            "excerpt": "CREATE TABLE t0(c0 INT UNIQUE);\r\nINSERT INTO t0(c0) VALUES (NULL);\r\nSELECT * FROM t0 WHERE (CASE c0 WHEN 0 THEN 1 ELSE 0 END = 0) ORDER BY c0; -- {}\r\nSELECT * FROM t0 WHERE (CASE c0 WHEN 0 THEN 1 ELSE 0 END = 0); -- {NULL}",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Manuel Rigger of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e19ac93470c0b8ef82d38205a8d25a6781c9cc62b75e2bb4c19501189ed4f63e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Manuel Rigger.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e19ac93470c0b8ef82d38205a8d25a6781c9cc62b75e2bb4c19501189ed4f63e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:databend:b9f95ecb25d3",
+      "dbms": "databend",
+      "title": "sqlancer: expression expansion error",
+      "reported_date": "2022-08-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "hanyisong",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/7360"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/7360",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/7360",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer: expression expansion error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:adf39bba4f5a63b2d9f2cba58ee9b524558c22ac3a6595f677ca223c0a596c3b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/7360",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:adf39bba4f5a63b2d9f2cba58ee9b524558c22ac3a6595f677ca223c0a596c3b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:databend:49223bb4cbe2",
+      "dbms": "databend",
+      "title": "bug: cannot convert NULL to a non-nullable type",
+      "reported_date": "2022-09-06",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "hanyisong",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/7498"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/7498",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/7498",
+            "source_type": "github_issue",
+            "excerpt": "MySQL [sqlancer]> select c0::float32 from t0;\r\n+-------------+",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:902272c68c05738661b6926c75e2921dc1048b334860bf1746e6445aeda1de91",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/7498",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:902272c68c05738661b6926c75e2921dc1048b334860bf1746e6445aeda1de91"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:databend:a3848221587d",
+      "dbms": "databend",
+      "title": "bug: index out of bounds",
+      "reported_date": "2022-11-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "hanyisong",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/9018"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/9018",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/9018",
+            "source_type": "github_issue",
+            "excerpt": "The case is from sqlancer.\r\n```sql",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:94eb51ec9d87a15b65d249820ef0a08e427cfa657ad2ad874a4a8ec9ac32b1f3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/9018",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:94eb51ec9d87a15b65d249820ef0a08e427cfa657ad2ad874a4a8ec9ac32b1f3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:databend:a543a3fdd032",
+      "dbms": "databend",
+      "title": "bug: The logic bug obtained by sqlancer",
+      "reported_date": "2022-12-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "hanyisong",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/9196"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/9196",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/9196",
+            "source_type": "github_issue",
+            "excerpt": "bug: The logic bug obtained by sqlancer",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:84fd32bb0356dd17fa1eb2f1776b3087f82c123cd341c956dab75b61455fd7e3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/9196",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:84fd32bb0356dd17fa1eb2f1776b3087f82c123cd341c956dab75b61455fd7e3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:databend:b4dd8e04a173",
+      "dbms": "databend",
+      "title": "bug: Connection error using mysql-connector-java:8.0.30",
+      "reported_date": "2023-01-31",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "hanyisong",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/9800"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/9800",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/9800",
+            "source_type": "github_issue",
+            "excerpt": "https://github.com/sqlancer/sqlancer/issues/707#issue-1561132047 I tested it manually and found that there is no problem with databend version 0.8, but this problem will occur in version 0.9.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e883c406caa787fd656beee1b45fb0126842cae1d9389243aaa2371af2fef6b0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/9800",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e883c406caa787fd656beee1b45fb0126842cae1d9389243aaa2371af2fef6b0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:databend:80f0ca22fe2a",
+      "dbms": "databend",
+      "title": "Decimal overflow conversion error",
+      "reported_date": "2024-05-18",
+      "reported_year": 2024,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Ming Wei Tan",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/15568"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/15568",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/15568",
+            "source_type": "github_issue",
+            "excerpt": "The bug discovered using SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:589c9a9ec1e2422b410175c44718e801324f93d88b976865af8a45dcfbf464fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T12:38:22Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Ming Wei Tan.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T12:38:22Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:589c9a9ec1e2422b410175c44718e801324f93d88b976865af8a45dcfbf464fc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:databend:cc19287ae25e",
+      "dbms": "databend",
+      "title": "WHERE clause not working as expected with NOT",
+      "reported_date": "2024-05-18",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Ming Wei Tan",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/15572"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/15572",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/15572",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN);\r\nINSERT INTO t0(c0) VALUES (false);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Ming Wei Tan of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5c44a1c235268450622f0495acd53e2991397f60ee1629b74b8a93da568f9328",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Ming Wei Tan.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5c44a1c235268450622f0495acd53e2991397f60ee1629b74b8a93da568f9328"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:databend:296e7a0e3133",
+      "dbms": "databend",
+      "title": "`FALSE AND exp1 OR exp1` in WHERE clause results in Lost connection from server",
+      "reported_date": "2024-05-18",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Ming Wei Tan",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/15569"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/15569",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/15569",
+            "source_type": "github_issue",
+            "excerpt": "Bug discovered with SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:9969a64136cb69b1dc4956f0c7b4fd3a806a0f688abe8391aa059493b270f611",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T12:38:22Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Ming Wei Tan.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T12:38:22Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9969a64136cb69b1dc4956f0c7b4fd3a806a0f688abe8391aa059493b270f611"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:databend:64e27abcaa02",
+      "dbms": "databend",
+      "title": "bug: WHERE clause not working as expected with null column values and string functions (`like`, `in`, etc)",
+      "reported_date": "2024-05-18",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "malwaregarry",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/15570"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/15570",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/15570",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c1 int, c2 VARCHAR NULL);\r\nINSERT INTO t1(c1) VALUES (5);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:831a9b36af122676af66c33bf6563cf74aed01285f1314e6a86baed3d4a57e56",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/15570",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:831a9b36af122676af66c33bf6563cf74aed01285f1314e6a86baed3d4a57e56"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:databend:49960d82bdab",
+      "dbms": "databend",
+      "title": "Mismatch between SUM and AVERAGE due to optimisation",
+      "reported_date": "2024-06-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Ming Wei Tan",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/15867"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/15867",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/15867",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\r\nINSERT INTO t0(c1) VALUES (1), (null);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Ming Wei Tan of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:140d3aa24d1da2e185eb7976be7aac2158a8db4ba0049c678dc23b9240904ed8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Ming Wei Tan.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:140d3aa24d1da2e185eb7976be7aac2158a8db4ba0049c678dc23b9240904ed8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:databend:0e8de5fe0c76",
+      "dbms": "databend",
+      "title": "bug: Overflow / divide by zero exceptions not detected in where clause",
+      "reported_date": "2024-06-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "malwaregarry",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/15866"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/15866",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/15866",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT);\r\nINSERT INTO t1(c0) VALUES (1);\r\nSELECT t1.c0 FROM t1 WHERE ( -9223372036854775809); \r\nSELECT t1.c0 FROM t1 WHERE (NOT  (-9223372036854775809) );",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6ee7ec55e66963c14ec7ba80fd268860c63fac9e0f8316fd3ee12689d32a93b1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/15866",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6ee7ec55e66963c14ec7ba80fd268860c63fac9e0f8316fd3ee12689d32a93b1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:databend:0afc053a7261",
+      "dbms": "databend",
+      "title": "bug: SELECT AVG(constant) over cross join panics with Decimal precision mismatch",
+      "reported_date": "2026-04-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/19738"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/19738",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/19738",
+            "source_type": "github_issue",
+            "excerpt": "Found by [SQLancer](https://github.com/sqlancer/sqlancer) and analyzed using Claude.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:794d35f98321685875d8bc8c49e9185fe5cde70da95b6731d80b4f536db8893b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/19738",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:794d35f98321685875d8bc8c49e9185fe5cde70da95b6731d80b4f536db8893b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:databend:dfe1829fce4c",
+      "dbms": "databend",
+      "title": "bug: SUM(constant) over cross join inside UNION ALL panics with Decimal precision mismatch (regression after #19740)",
+      "reported_date": "2026-04-26",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/19773"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/19773",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/19773",
+            "source_type": "github_issue",
+            "excerpt": "Found by [SQLancer](https://github.com/sqlancer/sqlancer)'s TLP query-partitioning oracle and analyzed using Claude.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:4c7dd2e6ece44ddeb8aa96f1980561e9cae59f7d1f54680288087fc5965136ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/19773",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4c7dd2e6ece44ddeb8aa96f1980561e9cae59f7d1f54680288087fc5965136ed"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:databend:c5d14d847564",
+      "dbms": "databend",
+      "title": "bug: CAST(FLOAT AS VARCHAR) returns different strings for the same value depending on whether EXISTS is in WHERE vs projection",
+      "reported_date": "2026-08-21",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Manuel-Neuer1",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/20350"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/20350",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/20350",
+            "source_type": "github_issue",
+            "excerpt": "This was discovered by SQLancer's DQR (Derived Query Relocation) equivalence oracle: it rewrites a query by relocating the `EXISTS` predicate from `WHERE` into the projection list of a derived table, and asserts the two forms return the same row signatures. They do not, because the `CAST(t0.c0float AS VARCHAR)` signature string differs.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:6cccb7c1630a33a9943d74d60a38aaec9f3f0ed2106c9816c03060578fc8d8b5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/20350",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6cccb7c1630a33a9943d74d60a38aaec9f3f0ed2106c9816c03060578fc8d8b5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:databend:fe732cbbd899",
+      "dbms": "databend",
+      "title": "bug: EXISTS correlated subquery treats NULL probe key as the type's default value in LEFT SEMI hash join",
+      "reported_date": "2026-08-21",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Manuel-Neuer1",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/20351"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/20351",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/20351",
+            "source_type": "github_issue",
+            "excerpt": "This was discovered by SQLancer's DQR (Derived Query Relocation) equivalence oracle (rule `EXISTS_TO_SUBQUERY`): it rewrites a query by relocating the `EXISTS` predicate from `WHERE` into the projection list of a derived table, and asserts both forms return the same rows. They do not — the base form wrongly retains NULL-keyed rows. 9 distinct occurrences were hit in one fuzzing campaign.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:7e71191e6a9ae1567156332c9e75a5d33a375eeefbb14dd7e5ea9ad4ecbdc26a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/20351",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7e71191e6a9ae1567156332c9e75a5d33a375eeefbb14dd7e5ea9ad4ecbdc26a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:databend:f6ef94677415",
+      "dbms": "databend",
+      "title": "bug: LIKE with an empty-string pattern from a column reference matches any subject",
+      "reported_date": "2026-08-21",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Manuel-Neuer1",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/20358"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/20358",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/20358",
+            "source_type": "github_issue",
+            "excerpt": "This was discovered by SQLancer's DQR (Derived Query Relocation) equivalence oracle: relocating an `EXISTS` predicate from `WHERE` into a derived-table projection changes the plan, and the two forms evaluate the correlated `LIKE` through different paths (semi-join filter keeps the pattern as a column reference; the materialized scalar path constant-folds it), producing different row counts for logically-equivalent queries.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b69608635bdd545868e9ecf8b3a2bd07dd47edf37ebe4bf0409067fe99b72cf6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/20358",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b69608635bdd545868e9ecf8b3a2bd07dd47edf37ebe4bf0409067fe99b72cf6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:databend:f20363571998",
+      "dbms": "databend",
+      "title": "bug: correlated `EXISTS` with cross-type join keys (`INT = DECIMAL`) wrongly matches NULL probe keys — and returns an unstable number of rows",
+      "reported_date": "2026-08-24",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Manuel-Neuer1",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/20364"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/20364",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/20364",
+            "source_type": "github_issue",
+            "excerpt": "**Cluster information**: single node, local `datafuselabs/databend` Docker container (MySQL protocol). No special configuration. Found by an automated equivalence-testing fuzzer (SQLancer-derived) that compares a query against a logically-equivalent rewrite.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:54bcf30d8e537415c30dd8a13f5db2b07bd9dc561460af7b98b1c17d07b56b34",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/20364",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:54bcf30d8e537415c30dd8a13f5db2b07bd9dc561460af7b98b1c17d07b56b34"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:databend:81a8f0121c24",
+      "dbms": "databend",
+      "title": "bug: Correlated EXISTS relocated into a projection fails decorrelation with `Internal error 1001: Invalid scalar for flattening subquery`",
+      "reported_date": "2026-08-25",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Manuel-Neuer1",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/20380"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/20380",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/20380",
+            "source_type": "github_issue",
+            "excerpt": "This was discovered by SQLancer's DQR (Derived Query Relocation) equivalence oracle: it relocates the `EXISTS` predicate from `WHERE` into a derived-table projection and asserts both forms behave the same. Here one form succeeds and the other errors out.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:727f22f11efdc0227afd1c999458e8e2686425e7da383dbcb51cd11741fbe910",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/20380",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:727f22f11efdc0227afd1c999458e8e2686425e7da383dbcb51cd11741fbe910"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:databend:d845b1c04264",
+      "dbms": "databend",
+      "title": "bug: `HAVING (AGG(col) = AGG(col))` keeps groups whose aggregate is NULL — the same self-comparison in a projection column filters them correctly",
+      "reported_date": "2026-08-25",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Manuel-Neuer1",
+      "links": {
+        "report": "https://github.com/databendlabs/databend/issues/20379"
+      },
+      "primary_url": "https://github.com/databendlabs/databend/issues/20379",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/databendlabs/databend/issues/20379",
+            "source_type": "github_issue",
+            "excerpt": "This was discovered by SQLancer's DQR (Derived Query Relocation) equivalence oracle: it relocates a `HAVING` predicate into a derived-table projection and asserts both forms return the same rows. They do not — the `HAVING` form retains the NULL-aggregate groups.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:4f410575bef6a51ab1ee2fee8a71ef612d137ee6b45710d9cb396ee2db3c6eeb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/databendlabs/databend/issues/20379",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4f410575bef6a51ab1ee2fee8a71ef612d137ee6b45710d9cb396ee2db3c6eeb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:17c878c86f5d",
+      "dbms": "datafusion",
+      "title": "Bugs in LCM/GCD scalar functions (found by SQLancer)",
+      "reported_date": "2024-06-20",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11031"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11031",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11031",
+            "source_type": "github_issue",
+            "excerpt": "Bugs in LCM/GCD scalar functions (found by SQLancer)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:41202c894ff87245fd59ab672dbbbab460f040790686e1a893bc49f6fa94fb49",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11031",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:41202c894ff87245fd59ab672dbbbab460f040790686e1a893bc49f6fa94fb49"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:dfcb843d72e1",
+      "dbms": "datafusion",
+      "title": "Overflow bug in FACTORIAL scalar function (found by SQLancer)",
+      "reported_date": "2024-06-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11074"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11074",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11074",
+            "source_type": "github_issue",
+            "excerpt": "Overflow bug in FACTORIAL scalar function (found by SQLancer)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:816922e66434af79befa6fd3593bf9e2b88a82a1fbb0f3fea8fe236dd73efdba",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11074",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:816922e66434af79befa6fd3593bf9e2b88a82a1fbb0f3fea8fe236dd73efdba"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:061e60183677",
+      "dbms": "datafusion",
+      "title": "Overflow bug in POW scalar function (found by SQLancer)",
+      "reported_date": "2024-06-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11075"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11075",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11075",
+            "source_type": "github_issue",
+            "excerpt": "Overflow bug in POW scalar function (found by SQLancer)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:8527268a5568e163b42f4d10adbc363a3d20777c7b709c75f044bdf69b9523f7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11075",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8527268a5568e163b42f4d10adbc363a3d20777c7b709c75f044bdf69b9523f7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:adc56066ca08",
+      "dbms": "datafusion",
+      "title": "Overflow bug in negate arithmetic operator (found by SQLancer)",
+      "reported_date": "2024-06-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11076"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11076",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11076",
+            "source_type": "github_issue",
+            "excerpt": "Overflow bug in negate arithmetic operator (found by SQLancer)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:331471d86a3112a8d8783c38d4ead3bf6208bf25b6b19a927843971d63965538",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11076",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:331471d86a3112a8d8783c38d4ead3bf6208bf25b6b19a927843971d63965538"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:52e04d79d205",
+      "dbms": "datafusion",
+      "title": "Tracking issue: Overflow bugs in scalar math functions found by SQLancer",
+      "reported_date": "2024-06-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LorrensP-2158466",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11078"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11078",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11078",
+            "source_type": "github_issue",
+            "excerpt": "Tracking issue: Overflow bugs in scalar math functions found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:60beeeaef7639233449c664d31bf9e8988721cf545bac3afe1242ae46b4154e1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11078",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:60beeeaef7639233449c664d31bf9e8988721cf545bac3afe1242ae46b4154e1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:c336d4f0538a",
+      "dbms": "datafusion",
+      "title": "Error evaluating clause `where COL_BIGINT < 1e100` (Found by SQLancer-NoREC)",
+      "reported_date": "2024-07-03",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11252"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11252",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11252",
+            "source_type": "github_issue",
+            "excerpt": "Error evaluating clause `where COL_BIGINT < 1e100` (Found by SQLancer-NoREC)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:97933f273701eb1467450e9726ea57bf1aa7b40d01217eeef73d02adf0c89720",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11252",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:97933f273701eb1467450e9726ea57bf1aa7b40d01217eeef73d02adf0c89720"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:414f0ede5878",
+      "dbms": "datafusion",
+      "title": "`where` clause incorrectly reject `NULL` literal (by SQLancer-NoREC)",
+      "reported_date": "2024-07-03",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11248"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11248",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11248",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer's NoREC oracle https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:4869049f5d52c066de2da92f5a5ff1fe2148860f7666d9eb53223f4935d8615f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11248",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4869049f5d52c066de2da92f5a5ff1fe2148860f7666d9eb53223f4935d8615f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:700bb10ef7ed",
+      "dbms": "datafusion",
+      "title": "Bug in Nested Loop Join (Found by SQLancer-NoREC)",
+      "reported_date": "2024-07-04",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11269"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11269",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11269",
+            "source_type": "github_issue",
+            "excerpt": "Bug in Nested Loop Join (Found by SQLancer-NoREC)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:6fe7ad1d9215b7f82f241eea2817720642dc163b9e41d1312178fb9481aa9ea7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11269",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6fe7ad1d9215b7f82f241eea2817720642dc163b9e41d1312178fb9481aa9ea7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:01fb69cdcaf7",
+      "dbms": "datafusion",
+      "title": "Non-deterministic `SUM()` aggregate function result (found by SQLancer-TLP)",
+      "reported_date": "2024-07-04",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11264"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11264",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11264",
+            "source_type": "github_issue",
+            "excerpt": "Non-deterministic `SUM()` aggregate function result (found by SQLancer-TLP)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:80f2b7944aab86c871e04d8ddbe7d6150bbccee4483a6f5757db0a7bae231105",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11264",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:80f2b7944aab86c871e04d8ddbe7d6150bbccee4483a6f5757db0a7bae231105"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:6e0136e869f0",
+      "dbms": "datafusion",
+      "title": "`bitwise*` operators won't check arguments type when short circuited (Found by SQLancer-NoREC)",
+      "reported_date": "2024-07-04",
+      "reported_year": 2024,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11260"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11260",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11260",
+            "source_type": "github_issue",
+            "excerpt": "`bitwise*` operators won't check arguments type when short circuited (Found by SQLancer-NoREC)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:fec51c032ab74a98b4da206e79ed25601cbc7fe49007b7e3d21a240177571f86",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11260",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fec51c032ab74a98b4da206e79ed25601cbc7fe49007b7e3d21a240177571f86"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:7272c5949439",
+      "dbms": "datafusion",
+      "title": "Incorrect result for a Nested Loop Join query (Found by SQLancer-NoREC)",
+      "reported_date": "2024-07-05",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11275"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11275",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11275",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result for a Nested Loop Join query (Found by SQLancer-NoREC)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:d544424491f4e024ca0f444f2a2fc9e43d1e80f636112beab4fcaef51d84163c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11275",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d544424491f4e024ca0f444f2a2fc9e43d1e80f636112beab4fcaef51d84163c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:e420d0fee9da",
+      "dbms": "datafusion",
+      "title": "A projection pushdown related bug (SQLancer-TLP)",
+      "reported_date": "2024-07-11",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11409"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11409",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11409",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:0c207cf96aca8d14cb1455092e763e46da44752aa1e8ce4bbafc5e3560c92162",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11409",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0c207cf96aca8d14cb1455092e763e46da44752aa1e8ce4bbafc5e3560c92162"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:6460cb20ec50",
+      "dbms": "datafusion",
+      "title": "Crash bug from an inner join query (SQLancer)",
+      "reported_date": "2024-07-11",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11414"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11414",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11414",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f2c704620245b3564bbcb287e1347b56986149e25504c27532288dc57ba470f2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11414",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f2c704620245b3564bbcb287e1347b56986149e25504c27532288dc57ba470f2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:a953f2f7dc1e",
+      "dbms": "datafusion",
+      "title": "Internal Error for an INNER JOIN query (SQLancer)",
+      "reported_date": "2024-07-11",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11412"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11412",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11412",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b999052e4d97d893ca6142457a8985c2da36b64f4058dc3d87147ee0c948941d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11412",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b999052e4d97d893ca6142457a8985c2da36b64f4058dc3d87147ee0c948941d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:21ea13bee7b0",
+      "dbms": "datafusion",
+      "title": "LEFT/RIGHT OUTER JOIN failed to detect ambiguous column reference (SQLancer-NoREC)",
+      "reported_date": "2024-07-11",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11408"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11408",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11408",
+            "source_type": "github_issue",
+            "excerpt": "found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:bf5b8a843e8801315890fdcf914cdf77026ff8ee5e342f29118dd557059ebf4d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11408",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bf5b8a843e8801315890fdcf914cdf77026ff8ee5e342f29118dd557059ebf4d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:91785155c3b8",
+      "dbms": "datafusion",
+      "title": "Crash bug when `log()` is used in `order by` clause (SQLancer)",
+      "reported_date": "2024-07-19",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11549"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11549",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11549",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c45fc1d78312d0c9bd9c54bf0ce4b2a37d285628f7b0682749632faa60d26a96",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11549",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c45fc1d78312d0c9bd9c54bf0ce4b2a37d285628f7b0682749632faa60d26a96"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:e0af10d9de2c",
+      "dbms": "datafusion",
+      "title": "Query with `order by acos(sin(v1))` panic (SQLancer)",
+      "reported_date": "2024-07-19",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11552"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11552",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11552",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:13f170ecb60117e2ee326d010f2c1a0d2a30f546d2845379f456583a4db20db6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11552",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:13f170ecb60117e2ee326d010f2c1a0d2a30f546d2845379f456583a4db20db6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:4d55f635a746",
+      "dbms": "datafusion",
+      "title": "Internal error when there is a bitwise operation in `order by` clause (SQLancer)",
+      "reported_date": "2024-07-20",
+      "reported_year": 2024,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11561"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11561",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11561",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:ec4eb1467032cfbdf7b00f7e3cf4f281cf011e0c1caa0232717f002c075ce148",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11561",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ec4eb1467032cfbdf7b00f7e3cf4f281cf011e0c1caa0232717f002c075ce148"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:ce2374db6acb",
+      "dbms": "datafusion",
+      "title": "Incorrect `NULL` handling for regex match `~` (SQLancer)",
+      "reported_date": "2024-07-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11623"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11623",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11623",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b3257363bf2ada401cac0f77eb4a1a1d199a4f876a530b7d43dac345d1603a1d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11623",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b3257363bf2ada401cac0f77eb4a1a1d199a4f876a530b7d43dac345d1603a1d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:e4f1d308e264",
+      "dbms": "datafusion",
+      "title": "Incorrect predicate evaluation result in a query (SQLancer-NoREC)",
+      "reported_date": "2024-07-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11621"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11621",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11621",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f79d514d03feb911aa09a5202e64b369a5de6bd57ef8811c15039569ef9a4896",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11621",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f79d514d03feb911aa09a5202e64b369a5de6bd57ef8811c15039569ef9a4896"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:b1a2264a8f5f",
+      "dbms": "datafusion",
+      "title": "Internal error when regex operator `~` is used with `List`s (SQLancer)",
+      "reported_date": "2024-07-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11622"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11622",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11622",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f30ef9bb547ba91ae962c1915a9b1c6744d8f9762f2707e6ac83305d427ccaa2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11622",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f30ef9bb547ba91ae962c1915a9b1c6744d8f9762f2707e6ac83305d427ccaa2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:4ab34d55fc54",
+      "dbms": "datafusion",
+      "title": "A valid SQL query returned 'Schema error: ...' (SQLancer)",
+      "reported_date": "2024-07-24",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11635"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11635",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11635",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b289cdc7ff8216c218404566685805a3496bb389799e7c3e204f3dff5ddf4a1c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11635",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b289cdc7ff8216c218404566685805a3496bb389799e7c3e204f3dff5ddf4a1c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:8e02e4f1a96e",
+      "dbms": "datafusion",
+      "title": "Incorrect result returned by a NLJ query with filter (SQLancer-NoREC)",
+      "reported_date": "2024-07-29",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11704"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11704",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11704",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer(https://github.com/apache/datafusion/issues/11030)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:2b6f99b0c5d5fc6e1abce96510a0d835e74f07a6a19b5fb9f354fdfc8f1d2efb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11704",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2b6f99b0c5d5fc6e1abce96510a0d835e74f07a6a19b5fb9f354fdfc8f1d2efb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:53783a09100d",
+      "dbms": "datafusion",
+      "title": "Incorrect result returned by `UNION ALL` (SQLancer-TLP)",
+      "reported_date": "2024-07-31",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11742"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11742",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11742",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:731eb01404ff87b92ad7d1428eec55d4a9663d980258661f7970767284f1f52b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11742",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:731eb01404ff87b92ad7d1428eec55d4a9663d980258661f7970767284f1f52b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:94576b5d12a9",
+      "dbms": "datafusion",
+      "title": "Incorrect result returned by a group by query (SQLancer-TLP)",
+      "reported_date": "2024-07-31",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11748"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11748",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11748",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:816f374d2285d12bf2112300646b2ead20b3d3a050ded2c63d3d061c07306187",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11748",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:816f374d2285d12bf2112300646b2ead20b3d3a050ded2c63d3d061c07306187"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:7c9b07b8fad8",
+      "dbms": "datafusion",
+      "title": "Scalar NULL literal in aggregate functions are not supported (SQLancer)",
+      "reported_date": "2024-07-31",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11749"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11749",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11749",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:4df72d474046be11d92ec118df476aa8582f26b42c05bb4d414052dce3a5f795",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11749",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4df72d474046be11d92ec118df476aa8582f26b42c05bb4d414052dce3a5f795"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:b03183999940",
+      "dbms": "datafusion",
+      "title": "Crash bug in `APPROX_MEDIAN` aggregate function (SQLancer)",
+      "reported_date": "2024-08-07",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11871"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11871",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11871",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:d673eb480a9dd7bfc391e04e489f60e527ff5221c3243c473516c7eaa4dcb225",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11871",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d673eb480a9dd7bfc391e04e489f60e527ff5221c3243c473516c7eaa4dcb225"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:2958718f4597",
+      "dbms": "datafusion",
+      "title": "Crash bug in `APPROX_PERCENTILE_CONT_WITH_WEIGHT` aggregate function (SQLancer)",
+      "reported_date": "2024-08-07",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11869"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11869",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11869",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:78a5e14ef214151c488fd0d4e48bca4908aa15bf001e3c9a694c1e19b3cf84bf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11869",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:78a5e14ef214151c488fd0d4e48bca4908aa15bf001e3c9a694c1e19b3cf84bf"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:cc2eaeedf353",
+      "dbms": "datafusion",
+      "title": "Crash bug in `APPROX_PERCENTILE_CONT` aggregate function (SQLancer)",
+      "reported_date": "2024-08-07",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11870"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11870",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11870",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:40d1acaebc0e3fc44e458fe94575e26afa9563bf1f962f008547a10d23704ad8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11870",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:40d1acaebc0e3fc44e458fe94575e26afa9563bf1f962f008547a10d23704ad8"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:2099a9c3d89c",
+      "dbms": "datafusion",
+      "title": "`ILIKE` string operator not working for `NULL` literal (SQLancer)",
+      "reported_date": "2024-08-07",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/11872"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/11872",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/11872",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e9fd9a0adad21f109f1f8241f3c39b502b6e39a676a73ec8b9dbe339c24b44ff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/11872",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e9fd9a0adad21f109f1f8241f3c39b502b6e39a676a73ec8b9dbe339c24b44ff"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:0ce8543600d5",
+      "dbms": "datafusion",
+      "title": "Internal error in `approx_percentile_cont()` aggregate function (SQLancer)",
+      "reported_date": "2024-08-15",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12012"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12012",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12012",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c47088e9014c54d66d96d92018f2f016666be4fd3fd9b0e14fbe9d2eb251efa9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12012",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c47088e9014c54d66d96d92018f2f016666be4fd3fd9b0e14fbe9d2eb251efa9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:e5d15b024844",
+      "dbms": "datafusion",
+      "title": "Invalid aggregate SQL query with `HAVING` can be executed without error (SQLancer-TLP)",
+      "reported_date": "2024-08-15",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12013"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12013",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12013",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:6677f74bc753da6168b70e856f7cde33c0184e8f1ae3119a57dfb9cf4002e4e3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12013",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6677f74bc753da6168b70e856f7cde33c0184e8f1ae3119a57dfb9cf4002e4e3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:66525eef393a",
+      "dbms": "datafusion",
+      "title": "Panics in `MIN()/MAX()` aggregate functions (SQLancer)",
+      "reported_date": "2024-08-15",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12011"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12011",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12011",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e0a171e0aa724537258daa778b39d6d68ba7c399dea6f14888a65b6fdb3080a6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12011",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e0a171e0aa724537258daa778b39d6d68ba7c399dea6f14888a65b6fdb3080a6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:b45777ebee1a",
+      "dbms": "datafusion",
+      "title": "Internal Error in `APPROX_PERCENTILE_CONT` window function (SQLancer)",
+      "reported_date": "2024-08-19",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12058"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12058",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12058",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:172beee601120fd9fedd25b2474312ec941f4aabf7c984aa5b4c10d64c83d0a3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12058",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:172beee601120fd9fedd25b2474312ec941f4aabf7c984aa5b4c10d64c83d0a3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:c6deb5305b9d",
+      "dbms": "datafusion",
+      "title": "SQL query with window function `PARTITION BY` caused panic in 'tokio-runtime-worker' (SQLancer)",
+      "reported_date": "2024-08-19",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12057"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12057",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12057",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e51542c894d351f1ce380d098c32778acf844a4b62a7bd24502a1115fc6e03c3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12057",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e51542c894d351f1ce380d098c32778acf844a4b62a7bd24502a1115fc6e03c3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:c5a259d9029f",
+      "dbms": "datafusion",
+      "title": "A simple count() query caused Internal Error in PhysicalOptimizer (SQLancer)",
+      "reported_date": "2024-08-20",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12077"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12077",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12077",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b7d03e33e45a20c70fd2dd178e5eeeda13d1a2c81082e5d7307b128543dbc708",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12077",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b7d03e33e45a20c70fd2dd178e5eeeda13d1a2c81082e5d7307b128543dbc708"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:6d17695ca4a2",
+      "dbms": "datafusion",
+      "title": "Panic in `NTH_VALUE()` window function (SQLancer)",
+      "reported_date": "2024-08-20",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12073"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12073",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12073",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:fe2af4349da2658a97fb9fd4a129ac4946c313304bfcb67756dddf3f6d8e4493",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12073",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fe2af4349da2658a97fb9fd4a129ac4946c313304bfcb67756dddf3f6d8e4493"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:256d19635d59",
+      "dbms": "datafusion",
+      "title": "Panic in `substring()` scalar function (SQLancer)",
+      "reported_date": "2024-08-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12129"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12129",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12129",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:2e6aa4b377c13f0e47f897672b92c4a381bd62aaea271e2238c45bdfc57a3665",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12129",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2e6aa4b377c13f0e47f897672b92c4a381bd62aaea271e2238c45bdfc57a3665"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:23fac4b8b93a",
+      "dbms": "datafusion",
+      "title": "Internal error in `regexp_replace()` for some StringView input (SQLancer)",
+      "reported_date": "2024-08-24",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12150"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12150",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12150",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:bce71eb08462f2d3f05146a97ce383231ca534a316f83ce6eb0cb2c5b430efb7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12150",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bce71eb08462f2d3f05146a97ce383231ca534a316f83ce6eb0cb2c5b430efb7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:5d4cc2a2c57b",
+      "dbms": "datafusion",
+      "title": "`octet_length()` function not working for StringView columns (SQLancer)",
+      "reported_date": "2024-08-24",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12149"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12149",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12149",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c6f3d485e2b895693ee96ed69655b55c22c4d21333e2e7732106a6b56f66bd98",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12149",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c6f3d485e2b895693ee96ed69655b55c22c4d21333e2e7732106a6b56f66bd98"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:c72ebb7684ce",
+      "dbms": "datafusion",
+      "title": "Bug in `nth_value()` window function for `NULL` input (SQLancer)",
+      "reported_date": "2024-09-04",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12320"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12320",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12320",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e44f9e946edddb97cb8a0117ffd9cfd08974c9bb78593f8c632f67fd89d9767e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12320",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e44f9e946edddb97cb8a0117ffd9cfd08974c9bb78593f8c632f67fd89d9767e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:dcd3d7be38dd",
+      "dbms": "datafusion",
+      "title": "A join SQL query with ambiguous table reference executes without error (SQLancer-NoREC)",
+      "reported_date": "2024-09-05",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12337"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12337",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12337",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:86dc0c3c6313b78f696df5515437f8700b53e89db197da1f0dbd3bbcc0d0f270",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12337",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:86dc0c3c6313b78f696df5515437f8700b53e89db197da1f0dbd3bbcc0d0f270"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:ea5a5cfe6cdf",
+      "dbms": "datafusion",
+      "title": "`Utf8View` column produced incorrect result in a natural join query (SQLancer-NoREC)",
+      "reported_date": "2024-09-15",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12468"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12468",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12468",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c48ac79cbe69bd9b09700ace81e2a18165a141b2db8ea3307167b2a7c8cc112e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12468",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c48ac79cbe69bd9b09700ace81e2a18165a141b2db8ea3307167b2a7c8cc112e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:b42410a6ef45",
+      "dbms": "datafusion",
+      "title": "Panic when an invalid expression in GROUP BY clause (SQLancer)",
+      "reported_date": "2024-10-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12699"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12699",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12699",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b81856fd9317641f0051066cb4961e1e100ff17230cc9ad5b8f19ba56c8b3892",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12699",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b81856fd9317641f0051066cb4961e1e100ff17230cc9ad5b8f19ba56c8b3892"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:95c7c0e9c8b3",
+      "dbms": "datafusion",
+      "title": "Incorrect NULL handling in `lead` window function (SQLancer)",
+      "reported_date": "2024-10-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12717"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12717",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12717",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:bacb6d81bf46a5e2a3885e523e55040f3fae5429ca7e6a196c5d79e701de85f5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12717",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bacb6d81bf46a5e2a3885e523e55040f3fae5429ca7e6a196c5d79e701de85f5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:9e3831f414ab",
+      "dbms": "datafusion",
+      "title": "Panic in scalar function `approx_percentile_cont_with_weight` (SQLancer)",
+      "reported_date": "2024-10-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12716"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12716",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12716",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f0380206d475e281523cc431236b54ab5ffeb01edd40e04bf0462fa08a15e3a4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12716",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f0380206d475e281523cc431236b54ab5ffeb01edd40e04bf0462fa08a15e3a4"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:0107cf6b2fe4",
+      "dbms": "datafusion",
+      "title": "Panic in `nth_value` window function (SQLancer)",
+      "reported_date": "2024-10-08",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12815"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12815",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12815",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:7f1f2ae076fde8f2c568450f2c5932b04a9d6d189aa7911d35ffe6f6993fa602",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12815",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7f1f2ae076fde8f2c568450f2c5932b04a9d6d189aa7911d35ffe6f6993fa602"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:5f821c309a01",
+      "dbms": "datafusion",
+      "title": "Panic in `simplify_expressions` optimizer rules when running an aggregate query (SQLancer)",
+      "reported_date": "2024-10-08",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/12814"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/12814",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/12814",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:554561a05d58cb405765c99593f15a295d4d1d55a0ff4a0bce94e989083eb32c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/12814",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:554561a05d58cb405765c99593f15a295d4d1d55a0ff4a0bce94e989083eb32c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:ac90f910aac9",
+      "dbms": "datafusion",
+      "title": "Panic in a query with NATURAL JOIN (SQLancer)",
+      "reported_date": "2025-01-06",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/14015"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/14015",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/14015",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c10fa0e984c6d049b866c609caadd15ae1cc09de57646831e6f64104e98e81a4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/14015",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c10fa0e984c6d049b866c609caadd15ae1cc09de57646831e6f64104e98e81a4"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:458e8a5da8e7",
+      "dbms": "datafusion",
+      "title": "Panic happens when adding a decimal256 to a float (SQLancer)",
+      "reported_date": "2025-07-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/16689"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/16689",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/16689",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer (https://github.com/apache/datafusion/issues/11030)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:21e414ff3591ea474b586a4ccfcdf23aff0092e67879360eb350746f71f96609",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/16689",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:21e414ff3591ea474b586a4ccfcdf23aff0092e67879360eb350746f71f96609"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:93dbb91621ba",
+      "dbms": "datafusion",
+      "title": "Arithmetic expression on `Date` type with `Null` returns planning error (SQLancer)",
+      "reported_date": "2025-07-13",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/16760"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/16760",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/16760",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b41efb62533ee29a9ee87440498912e50cfa1600ac56b1d19dcb8f74fd62eeca",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/16760",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b41efb62533ee29a9ee87440498912e50cfa1600ac56b1d19dcb8f74fd62eeca"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:50481ab52291",
+      "dbms": "datafusion",
+      "title": "Incorrect null handling for `%` operator (SQLancer)",
+      "reported_date": "2025-09-03",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/17387"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/17387",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/17387",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:dc9eca3f0439424f5b9507d5b36c012ef4db62ec0133bfecc9c803d246ef78d6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/17387",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dc9eca3f0439424f5b9507d5b36c012ef4db62ec0133bfecc9c803d246ef78d6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:8db6d2aec835",
+      "dbms": "datafusion",
+      "title": "Schema error in a query with `RIGHT ANTI JOIN` (SQLancer)",
+      "reported_date": "2025-09-03",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/17390"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/17390",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/17390",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:06a7de4b19ee02ed3e4010a70dce01cf5cb866734f0f2ca3d2def978c1701bc9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/17390",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:06a7de4b19ee02ed3e4010a70dce01cf5cb866734f0f2ca3d2def978c1701bc9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:datafusion:3ac27f063aa8",
+      "dbms": "datafusion",
+      "title": "Incorrect null literal handling for `to_local_time()` function (SQLancer)",
+      "reported_date": "2025-09-08",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "2010YOUY01",
+      "links": {
+        "report": "https://github.com/apache/datafusion/issues/17472"
+      },
+      "primary_url": "https://github.com/apache/datafusion/issues/17472",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/datafusion/issues/17472",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer https://github.com/apache/datafusion/issues/11030",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:ea044019965880a52a772f5525cc503f67ac63163e64539689e55294d9f79891",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/datafusion/issues/17472",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ea044019965880a52a772f5525cc503f67ac63163e64539689e55294d9f79891"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:dolt:b6ebac58b9bf",
+      "dbms": "dolt",
+      "title": "Unexpected Results when Using IN for Floating-Point",
+      "reported_date": "2023-07-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7120"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7120",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7120",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INTEGER, PRIMARY KEY(c0));\r\nINSERT INTO t1 (c0) VALUES (1);\r\n\r\nSELECT * FROM t1; -- 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5754e7336534affa780d9fad970a12d3d7301fc8c4891c3f155645923f5c5acd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5754e7336534affa780d9fad970a12d3d7301fc8c4891c3f155645923f5c5acd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:8b93037089f9",
+      "dbms": "dolt",
+      "title": "Unexpected Results when Using '%' operator",
+      "reported_date": "2023-11-16",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7006"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7006",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7006",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\r\nINSERT INTO t0(c1) VALUES (1);\r\n\r\nSELECT * FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:62327d3d9e348de2139933e81b69ae26d2544d8f528772f8bef13f024b74e10a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:62327d3d9e348de2139933e81b69ae26d2544d8f528772f8bef13f024b74e10a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:07f9e669be1b",
+      "dbms": "dolt",
+      "title": "Unexpected Results about Floating-point Type Casting",
+      "reported_date": "2023-11-17",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7018"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7018",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7018",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c1 INT);\r\nINSERT INTO t1(c1) VALUES (1);\r\n\r\nSELECT * FROM t1 WHERE (1 AND true); -- 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cf2113e89a96266d6f9bbe110478d776488977c6cc1c8d7699d2ecf5e6f7b708",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cf2113e89a96266d6f9bbe110478d776488977c6cc1c8d7699d2ecf5e6f7b708"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:c7ded2497b5c",
+      "dbms": "dolt",
+      "title": "Crashing by Division Operators",
+      "reported_date": "2023-11-19",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7026"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7026",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7026",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0( c1 INT);\r\nINSERT INTO t0(c1) VALUES (1);\r\nSELECT * FROM t0 WHERE ((c1/-364240480)/(880447354))%1;\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ef0948ba353034da34635142eab7cdc09db388889cc6da37c9933a8557134adc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ef0948ba353034da34635142eab7cdc09db388889cc6da37c9933a8557134adc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:7525dd8ab0af",
+      "dbms": "dolt",
+      "title": "Unexpected Results when Querying with NULL values",
+      "reported_date": "2023-11-19",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7025"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7025",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7025",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t2(c0 INT, PRIMARY KEY(c0))\r\nINSERT INTO t2(c0) VALUES (1);\r\n\r\nSELECT * FROM t2; -- 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:8e319d90ea0f9904f32718f95aed267c9af6e1828a61552ae794740f951d7243",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:8e319d90ea0f9904f32718f95aed267c9af6e1828a61552ae794740f951d7243"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:8196e36f97f8",
+      "dbms": "dolt",
+      "title": "Crash by Function DAYNAME",
+      "reported_date": "2023-11-22",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7039"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7039",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7039",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c2 INT, c3 INT, PRIMARY KEY(c2));\r\nINSERT INTO t0(c2) VALUES (1);\r\nINSERT INTO t0(c2) VALUES (2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:02f56ac3b58c06ec73bfeceb6db36a95f467411672a69a3c61c69792865acf83",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:02f56ac3b58c06ec73bfeceb6db36a95f467411672a69a3c61c69792865acf83"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:b56a3eb87b61",
+      "dbms": "dolt",
+      "title": "Unexpected Results about Decimal-Boolean Casting in Filters",
+      "reported_date": "2023-11-22",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7038"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7038",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7038",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0 VALUES (1);\r\n\r\nSELECT * FROM t0; -- 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:fc22531c58339fcb1d7a1b983d125043663e534b9cd5d2def3624d9766803cd9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:fc22531c58339fcb1d7a1b983d125043663e534b9cd5d2def3624d9766803cd9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e29e32ff1f29",
+      "dbms": "dolt",
+      "title": "Crash by ACOS",
+      "reported_date": "2023-11-23",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7046"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7046",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7046",
+            "source_type": "github_issue",
+            "excerpt": "The following test case crashed go-mysql-server.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c6dc5e347ce67386873363beac96d8ece2dceca9e563004ae217fe528ba2e632",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/7046",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c6dc5e347ce67386873363beac96d8ece2dceca9e563004ae217fe528ba2e632"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:6b5e0be4c650",
+      "dbms": "dolt",
+      "title": "Crash by Time Functions",
+      "reported_date": "2023-11-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7056"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7056",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7056",
+            "source_type": "github_issue",
+            "excerpt": "The following test case crashed go-mysql-server. Seems related to https://github.com/dolthub/dolt/issues/7039 ?",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:bef87de48ba521e27f8492e7d686b374fdfaf10c6fdbaedb4731e40771c4ad83",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/7056",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bef87de48ba521e27f8492e7d686b374fdfaf10c6fdbaedb4731e40771c4ad83"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:9bde7feda05b",
+      "dbms": "dolt",
+      "title": "Crash by SQRT",
+      "reported_date": "2023-11-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7060"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7060",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7060",
+            "source_type": "github_issue",
+            "excerpt": "The following test case crashed go-mysql-server.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:77ac8bbce50cbba87cb0406bd8e4207ff281e1a94861139b6c93987d8fda730b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/7060",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:77ac8bbce50cbba87cb0406bd8e4207ff281e1a94861139b6c93987d8fda730b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:ad772da841b3",
+      "dbms": "dolt",
+      "title": "Potential Issue Using ROUND",
+      "reported_date": "2023-11-30",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7073"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7073",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7073",
+            "source_type": "github_issue",
+            "excerpt": "Potential Issue Using ROUND",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fa019c41cc9030c8b890fde72c49702077c30f63a36953e855ef077f5f926ca3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/7073",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fa019c41cc9030c8b890fde72c49702077c30f63a36953e855ef077f5f926ca3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:4810ad38424f",
+      "dbms": "dolt",
+      "title": "Unexpected Results when Querying with COT",
+      "reported_date": "2023-11-30",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7072"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7072",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7072",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\r\nCREATE INDEX i0 ON t0(c1 );\r\nINSERT INTO t0 (c1) VALUES (-1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:fc10d4d185d081aad9a7f2d793335452d8f405486eeb3edd6de2b59f0b3ab90a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:fc10d4d185d081aad9a7f2d793335452d8f405486eeb3edd6de2b59f0b3ab90a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:ee823a303c03",
+      "dbms": "dolt",
+      "title": "Unexpected Results of In expressions",
+      "reported_date": "2023-12-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7147"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7147",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7147",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INTEGER);\r\nINSERT INTO t1 (c0) VALUES (1);\r\n\r\nSELECT * FROM t1; -- 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ad252e0ff7dc0ef7ad0b2ae4ebd883cbb0f0065399cff9d44a34d56525fc78b9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ad252e0ff7dc0ef7ad0b2ae4ebd883cbb0f0065399cff9d44a34d56525fc78b9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:1257e9b0ec45",
+      "dbms": "dolt",
+      "title": "Panic when Using BETWEEN and CASE WHEN",
+      "reported_date": "2023-12-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7154"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7154",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7154",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\n\r\nSELECT t0.c0 FROM t0 WHERE (t0.c0 BETWEEN (CASE 1 WHEN 2 THEN 1.0 ELSE (1||2) END ) AND t0.c0);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cca3160b3775c536de3125cbdfda20127c36248475c7984f12fcc913cf3815d2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cca3160b3775c536de3125cbdfda20127c36248475c7984f12fcc913cf3815d2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:815ad7c59ec9",
+      "dbms": "dolt",
+      "title": "Unexpected Results when Using IN after Creating Index",
+      "reported_date": "2023-12-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7155"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7155",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7155",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0 (c0) VALUES (0);\r\nCREATE INDEX i0 ON t0(c0 );",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:4333f90f09c676faf25d8627235e0ca4cab8e892e691cbd79da6febb026f6309",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:4333f90f09c676faf25d8627235e0ca4cab8e892e691cbd79da6febb026f6309"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:88abc04188c2",
+      "dbms": "dolt",
+      "title": "Previous panic concerning CASE WHEN",
+      "reported_date": "2023-12-17",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7182"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7182",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7182",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\n\r\nSELECT t0.c0 FROM t0 WHERE (t0.c0 BETWEEN (CASE 1 WHEN 2 THEN 1.0 ELSE (1||2) END ) AND t0.c0);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:dd6a98380fea2bf114b5cbb6abe6a10a38bff708d8acbe7021bdb79cdaebd6cc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/7182",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dd6a98380fea2bf114b5cbb6abe6a10a38bff708d8acbe7021bdb79cdaebd6cc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:fb84d00a7d77",
+      "dbms": "dolt",
+      "title": "Panic when Using BETWEEN AND",
+      "reported_date": "2023-12-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7216"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7216",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7216",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\n\r\nSELECT * FROM t0 WHERE (t0.c0 BETWEEN ('' BETWEEN '' AND (('')||('#'))) AND t0.c0); \r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:4bd15f2b9c64786aea3a1a5cb757b83f5fc47baaffa5ec0ad64c06d4f153417a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:4bd15f2b9c64786aea3a1a5cb757b83f5fc47baaffa5ec0ad64c06d4f153417a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:a88335fcf9f5",
+      "dbms": "dolt",
+      "title": "Unexpected Results when Using ROUND()",
+      "reported_date": "2023-12-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7222"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7222",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7222",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Results when Using ROUND()",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2766c49c2c0e478479d62e9e1f5f4cd17bfe2ea40e9f5664cc1d84b134eafc53",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/7222",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2766c49c2c0e478479d62e9e1f5f4cd17bfe2ea40e9f5664cc1d84b134eafc53"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:b7909e69c551",
+      "dbms": "dolt",
+      "title": "Unexpected Results when Using BETWEEN and LEFT JOIN",
+      "reported_date": "2024-01-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7229"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7229",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7229",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0( c0 BOOL);\r\nCREATE TABLE t1( c1 BOOL);\r\nINSERT INTO t0 (c0) VALUES (1);\r\nINSERT INTO t1 (c1) VALUES (false), (true);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:565841e9f9f50e80b3ac5e90d78f00a56cc29a258681ebfbb15931c9252f768d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:565841e9f9f50e80b3ac5e90d78f00a56cc29a258681ebfbb15931c9252f768d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:4e186d3e2ce3",
+      "dbms": "dolt",
+      "title": "Unexpected Result when Using `-''` in `IN`",
+      "reported_date": "2024-01-04",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7246"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7246",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7246",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 BOOL);\r\nINSERT INTO t1 (c0) VALUES (0);\r\n\r\nSELECT * FROM t1; -- 0",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0be122d851a71632ec18cda8fc72a8d24d50c8ff5d69f06c1485bba81d819f70",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/7246",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0be122d851a71632ec18cda8fc72a8d24d50c8ff5d69f06c1485bba81d819f70"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:dfb9c4c8d984",
+      "dbms": "dolt",
+      "title": "Unexpected results when comparing with empty string",
+      "reported_date": "2024-01-16",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7323"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7323",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7323",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL, PRIMARY KEY(c0));\r\nCREATE TABLE t1(c1 VARCHAR(500));\r\nINSERT INTO t0 (c0) VALUES (true);\r\nINSERT INTO t0 (c0) VALUES ('');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:29e986035bd2a8d7aa8b7a286476d8b17959f2aee522b4bc50f8090dfb277932",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:29e986035bd2a8d7aa8b7a286476d8b17959f2aee522b4bc50f8090dfb277932"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:b01475d25e06",
+      "dbms": "dolt",
+      "title": "Unexpected result when using Boolean in IN",
+      "reported_date": "2024-01-17",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7338"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7338",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7338",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0( c0 VARCHAR(500));\r\nCREATE INDEX t0i0 ON t0( c0 );\r\nINSERT INTO t0 (c0) VALUES (false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:36a55c2b94364beb4a3036c3f7bbc7b708879cf2a50e6025ba5eb3356cf02110",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:36a55c2b94364beb4a3036c3f7bbc7b708879cf2a50e6025ba5eb3356cf02110"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:f9350019bda0",
+      "dbms": "dolt",
+      "title": "Crash by INNER JOIN",
+      "reported_date": "2024-01-19",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7371"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7371",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7371",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500));\r\nCREATE TABLE t1(c0 INTEGER, PRIMARY KEY(c0));\r\nINSERT INTO t1(c0) VALUES (0);\r\nINSERT INTO t0(c0) VALUES ('a');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:f1136424c9624270b1d54956f3e4b4b553af11591ec0a6496943a529d6c97e84",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:f1136424c9624270b1d54956f3e4b4b553af11591ec0a6496943a529d6c97e84"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:a2e970fc1e74",
+      "dbms": "dolt",
+      "title": "Unexpected results when comparing string with type conversion",
+      "reported_date": "2024-01-19",
+      "reported_year": 2024,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7372"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7372",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7372",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 VARCHAR(500), PRIMARY KEY(c1)); -- PK is needed to reproduce the issue\r\nINSERT INTO t0 (c1) VALUES ('a');\r\n\r\nSELECT * FROM t0; -- 'a'",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a61dd954cb4112b7007e1dbb578d996333738f861b2764370fae3aea374c4d03",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a61dd954cb4112b7007e1dbb578d996333738f861b2764370fae3aea374c4d03"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:5464bdfa1c42",
+      "dbms": "dolt",
+      "title": "Crash by IF and CHAR",
+      "reported_date": "2024-02-18",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7515"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7515",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7515",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER); \r\n\r\nINSERT INTO t0(c0) VALUES ('');\r\nSELECT * FROM t0 WHERE t0.c0 = IF(t0.c0, 1, CHAR(t0.c0));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ec67b219e94d93a177095412e17dc6421f6fac83eb0e953b53d51306d3d1add9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ec67b219e94d93a177095412e17dc6421f6fac83eb0e953b53d51306d3d1add9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:a2b22693b373",
+      "dbms": "dolt",
+      "title": "Panic when Using INNER JOIN",
+      "reported_date": "2024-03-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7235"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7235",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7235",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\r\nCREATE TABLE t1(c1 BOOL);\r\nSELECT * FROM t0 INNER  JOIN t1 ON (t1.c1 IN (false IN ('(')));\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:27d034254bfba0ed5f7d05b7c91066af37382d499e5127c26bb10c6601a7b7af",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:27d034254bfba0ed5f7d05b7c91066af37382d499e5127c26bb10c6601a7b7af"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:63023f03c858",
+      "dbms": "dolt",
+      "title": "Unexpected Result when Using -'' in IN",
+      "reported_date": "2024-04-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Result when Using -'' in IN",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Dolt bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:63023f03c8585bdf4f3da60748e8072f94befcd9451566c854782c75c467723a",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:63023f03c8585bdf4f3da60748e8072f94befcd9451566c854782c75c467723a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:658898ea0b1e",
+      "dbms": "dolt",
+      "title": "Panic when Using INNER JOIN with String in IN",
+      "reported_date": "2024-05-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7262"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7262",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7262",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c1 INT);\r\n\r\nSELECT * FROM t0 INNER  JOIN t1 ON (t1.c1 IN (((true)%('')))); -- panic",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:c9e7a3dc98566f1da6b560f71a09ba881a21c598afd84c2028dbfffab6892c1a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:c9e7a3dc98566f1da6b560f71a09ba881a21c598afd84c2028dbfffab6892c1a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:1d8c2d37c8af",
+      "dbms": "dolt",
+      "title": "Unexpected Result when Querying with CONCAT",
+      "reported_date": "2024-05-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7261"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7261",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7261",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, PRIMARY KEY(c0));\r\nINSERT INTO t0 (c0) VALUES (0);\r\n\r\nSELECT c0 FROM t0; -- 0",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:8e91eeb8520336d62dde0596543381bb9fd9294c9156e66f4cc1de3927298d48",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:8e91eeb8520336d62dde0596543381bb9fd9294c9156e66f4cc1de3927298d48"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:1360db522ecc",
+      "dbms": "dolt",
+      "title": "Unexpected Results when Using BETWEEN AND after CREATE INDEX",
+      "reported_date": "2024-05-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/7260"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/7260",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/7260",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT, c1 INT, PRIMARY KEY(c0));\r\nCREATE INDEX t1i1 ON t1(c1 , c0 ); -- necessary to reproduce the issue\r\nINSERT INTO t1 (c1, c0) VALUES (1, 1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:6c8d5d50491de53ea788fa8182d6fe59585ce369e2bb1db08755181b6c59c359",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:6c8d5d50491de53ea788fa8182d6fe59585ce369e2bb1db08755181b6c59c359"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:0096f436378a",
+      "dbms": "dolt",
+      "title": "Unexpected ANTI JOIN Result",
+      "reported_date": "2025-03-12",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10157"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10157",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10157",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN, PRIMARY KEY(c0));\nINSERT INTO t0(c0) VALUES (false);\n\nSELECT * FROM t0 WHERE NOT EXISTS (SELECT 1 FROM (SELECT 1) AS sub0 WHERE LOG2(t0.c0));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:f6a137dbe8d6c26b5a355c4f7d1895c735f824e0f6efa054db7ad0c1effc4d38",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:f6a137dbe8d6c26b5a355c4f7d1895c735f824e0f6efa054db7ad0c1effc4d38"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e82624b7c260",
+      "dbms": "dolt",
+      "title": "NULL in GROUP BY column is interpreted as 1",
+      "reported_date": "2025-03-26",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9035"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9035",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9035",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT);\nINSERT INTO t0(c0, c1) VALUES(NULL, 1), (1, NULL);\nSELECT\n    t0.c0 = t0.c1 AS ref0,",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d1ad7c9cb8c2c4a2f30b1108442c66e5c89ae449df3f460bc48bb53e409d5000",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9035",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d1ad7c9cb8c2c4a2f30b1108442c66e5c89ae449df3f460bc48bb53e409d5000"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:de9d69d6add7",
+      "dbms": "dolt",
+      "title": "Unexpected crash when using GROUP BY with non-column position",
+      "reported_date": "2025-03-26",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9037"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9037",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9037",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 INT);\nINSERT INTO t0(c0) VALUES(1);\nSELECT 1 FROM t0 GROUP BY 2;\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b58dea8c989a72eb16dd684b03ec9f3a357ee2243d144a0dfe72af41b9254879",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9037",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b58dea8c989a72eb16dd684b03ec9f3a357ee2243d144a0dfe72af41b9254879"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:4d9cd85ab770",
+      "dbms": "dolt",
+      "title": "Unexpected invalid type error on boolean",
+      "reported_date": "2025-03-26",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9036"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9036",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9036",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected invalid type error on boolean",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b5c43524b59f05a009cb536586f8ae5f86bcad6fc44e173973f24ba8f4bb3cc9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "EmilyOng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9036",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b5c43524b59f05a009cb536586f8ae5f86bcad6fc44e173973f24ba8f4bb3cc9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:17d93ae97973",
+      "dbms": "dolt",
+      "title": "Double negation is treated as original value in WHERE clause",
+      "reported_date": "2025-03-31",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9054"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9054",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9054",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nINSERT INTO t0(c0) VALUES(123);\nSELECT t0.c0 AS ref0 FROM t0 WHERE 1 = (NOT(NOT(t0.c0)));\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4b7cf55dd503b089c371dd03aa442fb1d1c871071d6c864d2ae58dda57a44bf2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9054",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4b7cf55dd503b089c371dd03aa442fb1d1c871071d6c864d2ae58dda57a44bf2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:d37a387775b4",
+      "dbms": "dolt",
+      "title": "Incorrect negation of minimum signed integer",
+      "reported_date": "2025-03-31",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9053"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9053",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9053",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nINSERT INTO t0(c0) VALUES(-2147483648);\nSELECT -t0.c0 AS ref0 FROM t0;\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:324b69e7b1b2a81a6fecd4fb6f60bd57b7a84c7b6c88e4d10b130c8508a29093",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9053",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:324b69e7b1b2a81a6fecd4fb6f60bd57b7a84c7b6c88e4d10b130c8508a29093"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:d87902854293",
+      "dbms": "dolt",
+      "title": "Incorrect optimization of OR operation in expression in WHERE clause",
+      "reported_date": "2025-03-31",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9052"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9052",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9052",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nINSERT INTO t0(c0) VALUES(2);\nSELECT t0.c0 AS ref0 FROM t0 WHERE (FALSE OR t0.c0) != t0.c0;\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:140f96963145b5b314aabf167dd321b27d8eea686b461e6c29e54c33cfdfabef",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9052",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:140f96963145b5b314aabf167dd321b27d8eea686b461e6c29e54c33cfdfabef"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:081bb212f92c",
+      "dbms": "dolt",
+      "title": "Incorrect optimization of AND operation in expression in WHERE clause",
+      "reported_date": "2025-04-04",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9074"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9074",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9074",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nINSERT INTO t0(c0) VALUES(1);\nSELECT * FROM t0 WHERE ((TRUE AND -1) >= 0);\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4b5397ecb2dd96f9e0020e438e9f5ab6e5c5dfa77d812ad895213de979f5cc78",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9074",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4b5397ecb2dd96f9e0020e438e9f5ab6e5c5dfa77d812ad895213de979f5cc78"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:8187284df71d",
+      "dbms": "dolt",
+      "title": "Unexpected type conversion in IFNULL",
+      "reported_date": "2025-04-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9076"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9076",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9076",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected type conversion in IFNULL",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f3387c65f038b9476739ea8a9446caf23be9aeb9864a3438abe94529ed3a4ce1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "EmilyOng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9076",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f3387c65f038b9476739ea8a9446caf23be9aeb9864a3438abe94529ed3a4ce1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:77b7a84a9ee1",
+      "dbms": "dolt",
+      "title": "Group-By with `IF` function fails to distinguish the integer \"1\" and the boolean value \"true\"",
+      "reported_date": "2025-06-08",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9320"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9320",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9320",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nINSERT INTO t0(c0) VALUES(1);\nINSERT INTO t0(c0) VALUES(123);\nSELECT",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:778f80774673bc5dec5ae826460a1f4fbac5ac98513915768c765cc459df084d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9320",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:778f80774673bc5dec5ae826460a1f4fbac5ac98513915768c765cc459df084d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:ccd50dbec062",
+      "dbms": "dolt",
+      "title": "Using an IFNULL with TINYINT param restricts the output as TINYINT",
+      "reported_date": "2025-06-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9321"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9321",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9321",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 TINYINT);\nINSERT INTO t0(c0) VALUES(NULL);\n-- expected: 128; actual: 127\nSELECT IFNULL(t0.c0, 128) AS ref0 FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9329e036c377685d31c946d0fe68861b451ee1ac6909b27f124f289577cd8255",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9321",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9329e036c377685d31c946d0fe68861b451ee1ac6909b27f124f289577cd8255"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:7053b19e7e00",
+      "dbms": "dolt",
+      "title": "Unexpected Result Under Hint",
+      "reported_date": "2025-09-12",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10186"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10186",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10186",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 VARCHAR(500) , c1 BOOLEAN , PRIMARY KEY(c0));\nCREATE  TABLE  t1(c0 BOOLEAN , c1 BOOLEAN);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5534030b08a6dad773a1d6e9b1afaf932983a105e6adf1dbe75d176e62fd0787",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5534030b08a6dad773a1d6e9b1afaf932983a105e6adf1dbe75d176e62fd0787"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:39fe0e73a106",
+      "dbms": "dolt",
+      "title": "Incorrect double-negation of integer",
+      "reported_date": "2025-10-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9927"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9927",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9927",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect double-negation of integer",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:453c568206bc6000fc3bf7146c36dd625157a9bc3552b2800fe27dd2d5cb1b0a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "EmilyOng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9927",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:453c568206bc6000fc3bf7146c36dd625157a9bc3552b2800fe27dd2d5cb1b0a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:b3c27986921a",
+      "dbms": "dolt",
+      "title": "Incorrect use of negation in AntiJoinIncludingNulls",
+      "reported_date": "2025-10-08",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9935"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9935",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9935",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nINSERT INTO t0(c0) VALUES(1);\nSELECT * FROM t0 WHERE (! (1 || (EXISTS (SELECT 1))));\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1b5956e4ccf7d3cad255c8a4f4b2ce10ca6e428eb6861e27284a2a9cb78e9010",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9935",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1b5956e4ccf7d3cad255c8a4f4b2ce10ca6e428eb6861e27284a2a9cb78e9010"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:f0fdf9e08874",
+      "dbms": "dolt",
+      "title": "Use of invisible hash index under-fetches rows",
+      "reported_date": "2025-10-08",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/9936"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/9936",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/9936",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500));\nINSERT INTO t0(c0) VALUES(77367106);\nCREATE INDEX i0 USING HASH ON t0(c0) INVISIBLE;\nSELECT c0 FROM t0 WHERE (1620944823 >= t0.c0);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:300ced697a68a1c88ede08840cfee71c10d70b216151c7f9dc0efa61dc2d6fca",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/9936",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:300ced697a68a1c88ede08840cfee71c10d70b216151c7f9dc0efa61dc2d6fca"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:9d23c824d9a4",
+      "dbms": "dolt",
+      "title": "Unexpected UNION ALL Result of INNER JOIN and ANTI JOIN",
+      "reported_date": "2025-11-11",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10064"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10064",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10064",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 VARCHAR(500));\nINSERT INTO t0(c1, c0) VALUES ('-1', 1), ('-1', 1);\n\nSELECT * FROM t0 INNER JOIN (SELECT NULL WHERE FALSE) AS sub0 ON ((((t0.c0)*(t0.c0))))",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b91b020ff60f04f0ef43c00478fadabca96905b6e206ce0e495400c7a8ca981d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b91b020ff60f04f0ef43c00478fadabca96905b6e206ce0e495400c7a8ca981d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:56fc5b17305b",
+      "dbms": "dolt",
+      "title": "Unexpected `DAYOFMONTH` Result",
+      "reported_date": "2025-11-13",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10075"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10075",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10075",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 BOOLEAN);\n\nINSERT INTO t0(c1) VALUES (true);\nINSERT INTO t0(c1) VALUES (false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e0d326c2fdad6969927d74d82463780ab7dc23ab1f70abc86ea237ac7fc6d0f7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e0d326c2fdad6969927d74d82463780ab7dc23ab1f70abc86ea237ac7fc6d0f7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:967f1dd080d2",
+      "dbms": "dolt",
+      "title": "Unexpected ANTI JOIN Result",
+      "reported_date": "2025-11-18",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10092"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10092",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10092",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT, PRIMARY KEY(c0));-- Primary Key is necessary for triggering this bug\n\nINSERT INTO t1(c0) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:bda3c291cb0071e0dfdd4ba7de95e6b472bab190e6f775f34b41ce806c374d42",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:bda3c291cb0071e0dfdd4ba7de95e6b472bab190e6f775f34b41ce806c374d42"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:8145832b3bb6",
+      "dbms": "dolt",
+      "title": "Unexpected ANTI JOIN Result",
+      "reported_date": "2025-11-19",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10102"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10102",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10102",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 VARCHAR(500), PRIMARY KEY(c0));\nINSERT INTO t1(c0) VALUES ('9');\nSELECT * FROM t1 WHERE NOT EXISTS (SELECT 1 FROM (SELECT 1) AS sub0 WHERE ASIN(t1.c0));\n-- Wrong Execution Result: Empty Result",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:c9cf014179de6d6b1eb3e56e986fce11261a66bfe8e394822b51f1b11b9d45c6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:c9cf014179de6d6b1eb3e56e986fce11261a66bfe8e394822b51f1b11b9d45c6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:29b0f2e973ff",
+      "dbms": "dolt",
+      "title": "Unexpected Result in ANTI-JOIN",
+      "reported_date": "2025-12-11",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10070"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10070",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10070",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT , c1 INT);\n\nINSERT INTO t0(c0, c1) VALUES (1, -2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9722bd454874003f5091ca64559cb520145e3a8b8223fda3388757eeacadd8ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9722bd454874003f5091ca64559cb520145e3a8b8223fda3388757eeacadd8ed"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:b42a2bda4573",
+      "dbms": "dolt",
+      "title": "Lookup Join returns incorrect result due to out-of-range key conversion",
+      "reported_date": "2025-12-20",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10233"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10233",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10233",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 BOOLEAN , c1 INT , c2 VARCHAR(500) , c3 BOOLEAN );\nCREATE  TABLE  t1(c0 INT , c1 BOOLEAN , c2 BOOLEAN);\nCREATE INDEX t0i0 ON t0 (c3 , c2 ) ;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b6ed55695f355fd42d405f93ebf14f5749bbf5ee0cea8fad6e5c60db2b0b628a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b6ed55695f355fd42d405f93ebf14f5749bbf5ee0cea8fad6e5c60db2b0b628a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:4b520d159cda",
+      "dbms": "dolt",
+      "title": "Unexpected Anti Join Result",
+      "reported_date": "2025-12-21",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10234"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10234",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10234",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN);\nCREATE TABLE t1(c0 BOOLEAN);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:743640a033e294a8efea9db0763c23d0d035dea1433955332b168f2be4c885d0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:743640a033e294a8efea9db0763c23d0d035dea1433955332b168f2be4c885d0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:f20474989ee3",
+      "dbms": "dolt",
+      "title": "Unexpected Anti Join Result",
+      "reported_date": "2025-12-25",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10243"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10243",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10243",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c1 BOOLEAN);\nCREATE  TABLE  t1(c1 INT);\n\nINSERT INTO t0(c1) VALUES (true);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3cb72a3e13b0ed8f29997250897b6d0acd94a1ffdf9e9dd036b9deeb925bd413",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3cb72a3e13b0ed8f29997250897b6d0acd94a1ffdf9e9dd036b9deeb925bd413"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:afc7f5353bf0",
+      "dbms": "dolt",
+      "title": "Booleans should not be used as keys for varchar index lookups",
+      "reported_date": "2025-12-27",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10246"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10246",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10246",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN , c1 VARCHAR(500));\nCREATE TABLE t1(c0 VARCHAR(500) , c1 BOOLEAN, PRIMARY KEY(c0));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:0a69fd5fe48ae81e36d23abbbf786547bacbb5f1fd82f6e84a40bd060824b69b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:0a69fd5fe48ae81e36d23abbbf786547bacbb5f1fd82f6e84a40bd060824b69b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:72d8de4688cd",
+      "dbms": "dolt",
+      "title": "`WHERE NOT EXISTS` from an empty view does not return correct results",
+      "reported_date": "2026-02-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10258"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10258",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10258",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 BOOLEAN , c1 BOOLEAN);\nINSERT INTO t1(c0) VALUES (true), (false);\nCREATE VIEW v0(c0) AS SELECT true HAVING false;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:884f7079279e5837b040aa9ea9c4c5019ca9131dff693971e0170d8609ddb629",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:884f7079279e5837b040aa9ea9c4c5019ca9131dff693971e0170d8609ddb629"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:eec6358f1ff0",
+      "dbms": "dolt",
+      "title": "`NATURAL FULL JOIN` incorrectly parsed as `NATURAL RIGHT JOIN`",
+      "reported_date": "2026-06-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10268"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10268",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10268",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN, c1 INT, PRIMARY KEY(c0));\nCREATE TABLE t1(c0 BOOLEAN, c1 VARCHAR(500), PRIMARY KEY(c0));\nINSERT INTO t1(c1, c0) VALUES (NULL, true);\nINSERT INTO t0(c0, c1) VALUES (true, 4);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cbafb7ae7cbb1dec0861ef336c77ad3b84570ab89ce28e0ddaaf8dc3024a4128",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cbafb7ae7cbb1dec0861ef336c77ad3b84570ab89ce28e0ddaaf8dc3024a4128"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e622d7f2fad8",
+      "dbms": "dolt",
+      "title": "`abs` returns `null` when inputs are boolean values",
+      "reported_date": "2026-06-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10270"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10270",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10270",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT, PRIMARY KEY(c1));\nCREATE TABLE t1(c0 VARCHAR(50), c1 VARCHAR(50), PRIMARY KEY(c0, c1));\n\nINSERT INTO t0(c1) VALUES (true);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3ce3c214ce55948cd08320cdbe1a2d417b5f94a019c6ece6b12572008b92a9d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3ce3c214ce55948cd08320cdbe1a2d417b5f94a019c6ece6b12572008b92a9d8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:a4d666c92271",
+      "dbms": "dolt",
+      "title": "Disable RangeHeapJoin with string and number columns",
+      "reported_date": "2026-06-02",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10435"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10435",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10435",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500) , c1 INT);\nCREATE TABLE t1(c0 VARCHAR(500) , c1 INT);\n\nINSERT INTO t1(c0) VALUES (-1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:23dc2894094932a23ba4a5ce95cdbd298113503588b0e3e3fdb29bcdedb50468",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:23dc2894094932a23ba4a5ce95cdbd298113503588b0e3e3fdb29bcdedb50468"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:426be05f4680",
+      "dbms": "dolt",
+      "title": "`ColumnId`'s not added for `EmptyTable` during `assignExecIndexes`, causing offset when getting column index",
+      "reported_date": "2026-06-02",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10434"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10434",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10434",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t4(c1 BOOLEAN, PRIMARY KEY(c1));\nCREATE  TABLE  t0(c0 INT);\n\nCREATE VIEW v0(c0) AS SELECT 1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:beed0a422567ca6e27a767b3a9a4943dc6b42869f577c936dfd6001cdc5c5200",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:beed0a422567ca6e27a767b3a9a4943dc6b42869f577c936dfd6001cdc5c5200"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:0e75b798154c",
+      "dbms": "dolt",
+      "title": "Avoid RangeHeapJoin when lower and upper bounds are the same field",
+      "reported_date": "2026-07-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10284"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10284",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10284",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500) , c1 VARCHAR(500) , c2 VARCHAR(500));\nCREATE TABLE t1(c0 INT, c1 VARCHAR(500));\nINSERT INTO t0(c0, c1) VALUES (1, 5);\nINSERT INTO t0(c2) VALUES ('KZ');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:89c413bd95783cde0bdb48ee9f6ff5bec528825f2a45545a2d327734f5af8ce7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:89c413bd95783cde0bdb48ee9f6ff5bec528825f2a45545a2d327734f5af8ce7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:efc15ea0f327",
+      "dbms": "dolt",
+      "title": "Incorrect indexes in left outer lookup join",
+      "reported_date": "2026-07-02",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10451"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10451",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10451",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 BOOLEAN , c1 INT);\nCREATE TABLE t0(c0 INT , c1 VARCHAR(500) , PRIMARY KEY(c0)); -- Primary key constraint is necessary to trigger this bug\nINSERT INTO t1(c0, c1) VALUES (true, 2063453753);\nINSERT INTO t1(c0) VALUES (false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:811793644fb93d1942fe20830ce729a61951eac1fbb0588d3bfab3c83aea3032",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:811793644fb93d1942fe20830ce729a61951eac1fbb0588d3bfab3c83aea3032"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:044f53082935",
+      "dbms": "dolt",
+      "title": "Correlated subquery over a column-split-rejoin view raises internal \"unable to find field with index\" planner error",
+      "reported_date": "2026-07-31",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11378"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11378",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11378",
+            "source_type": "github_issue",
+            "excerpt": "Correlated subquery over a column-split-rejoin view raises internal \"unable to find field with index\" planner error",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:0b40c29ddd3c8bf2bbbc5bc0c2e189a8a8fbb5591b5ef389ba145738956ad1c4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11378",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0b40c29ddd3c8bf2bbbc5bc0c2e189a8a8fbb5591b5ef389ba145738956ad1c4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:83c3929a4a84",
+      "dbms": "dolt",
+      "title": "Aggregate window function with `OVER (ORDER BY <col>)` has inconsistent behavior",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11381"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11381",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11381",
+            "source_type": "github_issue",
+            "excerpt": "Aggregate window function with `OVER (ORDER BY <col>)` has inconsistent behavior",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:8095dbdf3df707dc8052cc7e2e3434a459208da8342e5ef747c20a101053a905",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11381",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8095dbdf3df707dc8052cc7e2e3434a459208da8342e5ef747c20a101053a905"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:598d0c62ea97",
+      "dbms": "dolt",
+      "title": "`RPAD()/LPAD()` has unexpected behvior",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11380"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11380",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11380",
+            "source_type": "github_issue",
+            "excerpt": "`RPAD()/LPAD()` has unexpected behvior",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d6d865feb0ff533043f788ded5a87b6a41c2f02519e30289b8135bd764a96124",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11380",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d6d865feb0ff533043f788ded5a87b6a41c2f02519e30289b8135bd764a96124"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:4a141f85fd82",
+      "dbms": "dolt",
+      "title": "Dolt `LAST_VALUE` with an explicit `RANGE` peer frame mishandles a `NULL` order key.",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11398"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11398",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11398",
+            "source_type": "github_issue",
+            "excerpt": "Dolt `LAST_VALUE` with an explicit `RANGE` peer frame mishandles a `NULL` order key.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b7097bd617c561fe7a1129d65e6e5195171e9d78f74c25fed23cd151e489e7be",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11398",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b7097bd617c561fe7a1129d65e6e5195171e9d78f74c25fed23cd151e489e7be"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:166c05749f62",
+      "dbms": "dolt",
+      "title": "Dolt `RANGE BETWEEN 1 PRECEDING AND CURRENT ROW` over `BIGINT UNSIGNED` loses the current row at the lower boundary.",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11396"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11396",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11396",
+            "source_type": "github_issue",
+            "excerpt": "Dolt `RANGE BETWEEN 1 PRECEDING AND CURRENT ROW` over `BIGINT UNSIGNED` loses the current row at the lower boundary.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ea93a9fa39e07b8327b1ba2043bb3c011ac6ce40fcb4e0aedbfa66f98ded6a6d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11396",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ea93a9fa39e07b8327b1ba2043bb3c011ac6ce40fcb4e0aedbfa66f98ded6a6d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:bf24227188a2",
+      "dbms": "dolt",
+      "title": "Dolt computes population statistical aggregates incorrectly for `INT` input.",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11391"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11391",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11391",
+            "source_type": "github_issue",
+            "excerpt": "Dolt computes population statistical aggregates incorrectly for `INT` input.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:443d8ae2ba7d1ad1dc962f27b122e33ffcff8c137d4e6c9a7e32d6b71dd14493",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11391",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:443d8ae2ba7d1ad1dc962f27b122e33ffcff8c137d4e6c9a7e32d6b71dd14493"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:ff2fdf42ea23",
+      "dbms": "dolt",
+      "title": "Dolt panics for a `RANGE BETWEEN CURRENT ROW AND 1 FOLLOWING` window over a `SET` order key.",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11397"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11397",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11397",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics for a `RANGE BETWEEN CURRENT ROW AND 1 FOLLOWING` window over a `SET` order key.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:90176e09b9856d743bce75ecdc6f801ee1416744434cb52f496eb115c33def63",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11397",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:90176e09b9856d743bce75ecdc6f801ee1416744434cb52f496eb115c33def63"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:f35b539fa9af",
+      "dbms": "dolt",
+      "title": "Dolt panics when `DEFAULT` is used in an `ON DUPLICATE KEY UPDATE` assignment.",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11389"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11389",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11389",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics when `DEFAULT` is used in an `ON DUPLICATE KEY UPDATE` assignment.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3ad5c430a2d527b7e36db56d323fb510c0d3bbe4fcd8e82d4a550f24cb4ca346",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11389",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3ad5c430a2d527b7e36db56d323fb510c0d3bbe4fcd8e82d4a550f24cb4ca346"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:1e9fe373f27d",
+      "dbms": "dolt",
+      "title": "Dolt panics when `JSON_TABLE` is evaluated over JSON `null`.",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11394"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11394",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11394",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics when `JSON_TABLE` is evaluated over JSON `null`.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2b9f28ec60509eeaaa1148c982617e9750968bb549beacc799f1716a74215eca",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11394",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2b9f28ec60509eeaaa1148c982617e9750968bb549beacc799f1716a74215eca"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:0021f6362795",
+      "dbms": "dolt",
+      "title": "Dolt panics when evaluating `LOCATE` with a negative start position.",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11393"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11393",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11393",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics when evaluating `LOCATE` with a negative start position.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7650ad1b3bacac71069febd4a827641a4c4f8f25ee3b33f232b70b1d53472ef6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11393",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7650ad1b3bacac71069febd4a827641a4c4f8f25ee3b33f232b70b1d53472ef6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:d23981971acc",
+      "dbms": "dolt",
+      "title": "Dolt raises an internal field-index error for sibling `SUM`/`AVG` windows",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11399"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11399",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11399",
+            "source_type": "github_issue",
+            "excerpt": "Dolt raises an internal field-index error for sibling `SUM`/`AVG` windows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b1070c210cd2d08d966e80e6cad0344df65a369bae90218a764661765378d564",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11399",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b1070c210cd2d08d966e80e6cad0344df65a369bae90218a764661765378d564"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:d9dd9bb9560a",
+      "dbms": "dolt",
+      "title": "Dolt raises an internal row-index error for  empty`INSERT ()`",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11388"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11388",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11388",
+            "source_type": "github_issue",
+            "excerpt": "Dolt raises an internal row-index error for  empty`INSERT ()`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:bd522213ba094c79a1d8fb1d710ab40547403488e103001613fd4441ff0c065b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11388",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bd522213ba094c79a1d8fb1d710ab40547403488e103001613fd4441ff0c065b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:8c09a51263ae",
+      "dbms": "dolt",
+      "title": "Dolt sibling window aggregates with different explicit frames reuse the wrong frame.",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11395"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11395",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11395",
+            "source_type": "github_issue",
+            "excerpt": "Dolt sibling window aggregates with different explicit frames reuse the wrong frame.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:abdf60c54ac6d4a5eb0e81ae8d58c60bdbb6e68b32defe652e0755b28ee3eeb8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11395",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:abdf60c54ac6d4a5eb0e81ae8d58c60bdbb6e68b32defe652e0755b28ee3eeb8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:030fdb61b45c",
+      "dbms": "dolt",
+      "title": "Dolt silently ignores `DISTINCT` for `COUNT` and `SUM` window aggregates.",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11392"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11392",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11392",
+            "source_type": "github_issue",
+            "excerpt": "Dolt silently ignores `DISTINCT` for `COUNT` and `SUM` window aggregates.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7851d0526bcd416df2763d6b2f84746c1b2ab409b7e1129459b114074d2febd7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11392",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7851d0526bcd416df2763d6b2f84746c1b2ab409b7e1129459b114074d2febd7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:3e6f8a77996d",
+      "dbms": "dolt",
+      "title": "Dolt supports `BIT_AND`, `BIT_OR`, and `BIT_XOR` as ordinary aggregates, but the same functions fail when used as window aggregates.",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11390"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11390",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11390",
+            "source_type": "github_issue",
+            "excerpt": "Dolt supports `BIT_AND`, `BIT_OR`, and `BIT_XOR` as ordinary aggregates, but the same functions fail when used as window aggregates.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:02235079a038d7071e9c0b152d33e74c4af9cbd440cf432a18b8b221864ccc37",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11390",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:02235079a038d7071e9c0b152d33e74c4af9cbd440cf432a18b8b221864ccc37"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:476f51455989",
+      "dbms": "dolt",
+      "title": "Dolt `GROUP_CONCAT(BIT)` concatenates numeric text instead of BIT bytes.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11422"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11422",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11422",
+            "source_type": "github_issue",
+            "excerpt": "Dolt `GROUP_CONCAT(BIT)` concatenates numeric text instead of BIT bytes.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2ab4936fb543f1a393a99e2d7cd050b925243f14517c01e3190539d7a309dccc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11422",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2ab4936fb543f1a393a99e2d7cd050b925243f14517c01e3190539d7a309dccc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:db58c55d2b7f",
+      "dbms": "dolt",
+      "title": "Dolt `GROUP_CONCAT(expr1, expr2, ...)` ignores later expressions.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11427"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11427",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11427",
+            "source_type": "github_issue",
+            "excerpt": "Dolt `GROUP_CONCAT(expr1, expr2, ...)` ignores later expressions.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:79f8017b3d3063e1a7866e13e8536d40b29c3ace193bb72bafa7f515d967f977",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11427",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:79f8017b3d3063e1a7866e13e8536d40b29c3ace193bb72bafa7f515d967f977"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:846300351e9d",
+      "dbms": "dolt",
+      "title": "Dolt `JSON_EXTRACT` does not preserve large JSON integer values.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11416"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11416",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11416",
+            "source_type": "github_issue",
+            "excerpt": "Dolt `JSON_EXTRACT` does not preserve large JSON integer values.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a2f6657a06232ef81762be0c6a06d57f6a0aa722b341688fddf2b075a86ffd65",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11416",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a2f6657a06232ef81762be0c6a06d57f6a0aa722b341688fddf2b075a86ffd65"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:51f10cd61436",
+      "dbms": "dolt",
+      "title": "Dolt `ONLY_FULL_GROUP_BY` recognizes functional dependency through a primary key but not through a `UNIQUE NOT NULL` key.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11415"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11415",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11415",
+            "source_type": "github_issue",
+            "excerpt": "Dolt `ONLY_FULL_GROUP_BY` recognizes functional dependency through a primary key but not through a `UNIQUE NOT NULL` key.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b5344a06ee5d6ff7777f55bed2935aa320d4f6531572128734443850dc0adc98",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11415",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b5344a06ee5d6ff7777f55bed2935aa320d4f6531572128734443850dc0adc98"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:ff088623e506",
+      "dbms": "dolt",
+      "title": "Dolt `TIME_FORMAT` mishandles hour format specifiers for `TIME` values.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11413"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11413",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11413",
+            "source_type": "github_issue",
+            "excerpt": "Dolt `TIME_FORMAT` mishandles hour format specifiers for `TIME` values.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e898b5ddebba7e093e6c71293f17a688f64b43b07648b58c562bc68ea540949e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11413",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e898b5ddebba7e093e6c71293f17a688f64b43b07648b58c562bc68ea540949e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:1577ef62a17e",
+      "dbms": "dolt",
+      "title": "Dolt aggregate numeric coercion over `TIME` returns zero.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11429"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11429",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11429",
+            "source_type": "github_issue",
+            "excerpt": "Dolt aggregate numeric coercion over `TIME` returns zero.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5fd67cc532892c2665e8777625e2b34920bfa15f277aabee236a34deb4824d6d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11429",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5fd67cc532892c2665e8777625e2b34920bfa15f277aabee236a34deb4824d6d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:9d81a3b4b2be",
+      "dbms": "dolt",
+      "title": "Dolt arithmetic on `BIGINT UNSIGNED` values wraps into signed negative values.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11411"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11411",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11411",
+            "source_type": "github_issue",
+            "excerpt": "Dolt arithmetic on `BIGINT UNSIGNED` values wraps into signed negative values.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:63f13a94d8c9f8c72816fecd96b8de4afbd715e7533406c9352a08b7d08acdc3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11411",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:63f13a94d8c9f8c72816fecd96b8de4afbd715e7533406c9352a08b7d08acdc3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:4126c6ff588a",
+      "dbms": "dolt",
+      "title": "Dolt converts valid date strings with leading or trailing whitespace to `NULL`",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11414"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11414",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11414",
+            "source_type": "github_issue",
+            "excerpt": "Dolt converts valid date strings with leading or trailing whitespace to `NULL`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:422e75bf682f3ac02b93d3c43883fb22b3adb00cf0f5b522b6362bf3e35ac8ef",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11414",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:422e75bf682f3ac02b93d3c43883fb22b3adb00cf0f5b522b6362bf3e35ac8ef"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:074358bb84ba",
+      "dbms": "dolt",
+      "title": "Dolt ignores `LIMIT` inside a correlated `EXISTS` subquery.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11424"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11424",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11424",
+            "source_type": "github_issue",
+            "excerpt": "Dolt ignores `LIMIT` inside a correlated `EXISTS` subquery.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7d2402dcfe5b3b70ad57ce064e8c5e25e4e7b5dcc6345ca2ca27862f8307e34e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11424",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7d2402dcfe5b3b70ad57ce064e8c5e25e4e7b5dcc6345ca2ca27862f8307e34e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:c8bee7463c60",
+      "dbms": "dolt",
+      "title": "Dolt ignores secondary `ORDER BY` keys for `RANGE CURRENT ROW` peer groups.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11423"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11423",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11423",
+            "source_type": "github_issue",
+            "excerpt": "Dolt ignores secondary `ORDER BY` keys for `RANGE CURRENT ROW` peer groups.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7c1b082e9911873d2d908e5dbf08b8ca14eb011eed886210c5a711d0fd7e07bf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11423",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7c1b082e9911873d2d908e5dbf08b8ca14eb011eed886210c5a711d0fd7e07bf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e4cac4917e36",
+      "dbms": "dolt",
+      "title": "Dolt misorders a window when `ORDER BY` contains a correlated scalar subquery.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11419"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11419",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11419",
+            "source_type": "github_issue",
+            "excerpt": "Dolt misorders a window when `ORDER BY` contains a correlated scalar subquery.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:13f6b91384cb8bc52576b5b60add78ba3d7a9bcbfbeb9be9e2529a5594202d8d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11419",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:13f6b91384cb8bc52576b5b60add78ba3d7a9bcbfbeb9be9e2529a5594202d8d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:39965a50a529",
+      "dbms": "dolt",
+      "title": "Dolt panics for `FIRST_VALUE` with an empty `OVER ()` spec.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11428"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11428",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11428",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics for `FIRST_VALUE` with an empty `OVER ()` spec.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d88de4298e392e22988c5d18934783e66de581744032bfd53535a37f612a95b3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11428",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d88de4298e392e22988c5d18934783e66de581744032bfd53535a37f612a95b3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:4c2fc4ec6452",
+      "dbms": "dolt",
+      "title": "Dolt panics when a named window inherits from an undefined window.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11426"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11426",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11426",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics when a named window inherits from an undefined window.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d1e027f3065012c61d7e326e1bd208c06ac8199aeaf047b87c59c5f39f76beed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11426",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d1e027f3065012c61d7e326e1bd208c06ac8199aeaf047b87c59c5f39f76beed"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e845b37ebc70",
+      "dbms": "dolt",
+      "title": "Dolt panics when preparing/executing `CREATE EVENT`.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11417"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11417",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11417",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics when preparing/executing `CREATE EVENT`.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5e34e30a5441bf76a732d65702148491e21ecf46b5fadd94e576fe32402b42e2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11417",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5e34e30a5441bf76a732d65702148491e21ecf46b5fadd94e576fe32402b42e2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:be8301645ffb",
+      "dbms": "dolt",
+      "title": "Dolt raises an internal row-index error for `EXISTS` with a window expression.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11421"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11421",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11421",
+            "source_type": "github_issue",
+            "excerpt": "Dolt raises an internal row-index error for `EXISTS` with a window expression.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b60041c2bedafc1070749b6855320ec83bd5effdda63fcd65cb33839f33849fe",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11421",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b60041c2bedafc1070749b6855320ec83bd5effdda63fcd65cb33839f33849fe"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:40efc4a1c043",
+      "dbms": "dolt",
+      "title": "Dolt rejects a repeated window expression in `ORDER BY` as ambiguous.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11418"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11418",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11418",
+            "source_type": "github_issue",
+            "excerpt": "Dolt rejects a repeated window expression in `ORDER BY` as ambiguous.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:93e08237d4e8affa6fc64de89890923415b8c1bf847ecd68adb700f30d3040bf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11418",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:93e08237d4e8affa6fc64de89890923415b8c1bf847ecd68adb700f30d3040bf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:3f745ba0d710",
+      "dbms": "dolt",
+      "title": "Dolt rejects unary negation of a `ROW_NUMBER()` window result.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11420"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11420",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11420",
+            "source_type": "github_issue",
+            "excerpt": "Dolt rejects unary negation of a `ROW_NUMBER()` window result.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:abdb0dda246371fdc50265a9d8edcefcbd5ded79e360f4fc634d14a59cbc63f5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11420",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:abdb0dda246371fdc50265a9d8edcefcbd5ded79e360f4fc634d14a59cbc63f5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:11aab7fdf860",
+      "dbms": "dolt",
+      "title": "Dolt reports an internal row-index bug for invalid `NTILE(column)`.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11425"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11425",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11425",
+            "source_type": "github_issue",
+            "excerpt": "Dolt reports an internal row-index bug for invalid `NTILE(column)`.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:947b8cfccf09ccde241cad554080bba778f3b0d0aee90db123cd207e14b78de9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11425",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:947b8cfccf09ccde241cad554080bba778f3b0d0aee90db123cd207e14b78de9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:85f471322201",
+      "dbms": "dolt",
+      "title": "Dolt returns a negative value for `ABS(-9223372036854775808)`.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11412"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11412",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11412",
+            "source_type": "github_issue",
+            "excerpt": "Dolt returns a negative value for `ABS(-9223372036854775808)`.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ed6c93829bb2ba0c1d08ffe1c276b36f3a3c6809437c10b10c12cc5316615bd6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11412",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ed6c93829bb2ba0c1d08ffe1c276b36f3a3c6809437c10b10c12cc5316615bd6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:d211696d9e25",
+      "dbms": "dolt",
+      "title": "Dolt window `SUM` returns `0` for an all-NULL frame.",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11410"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11410",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11410",
+            "source_type": "github_issue",
+            "excerpt": "Dolt window `SUM` returns `0` for an all-NULL frame.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:84b64e032e8a5f9c24de95e7020c06d7112af6f6f12db2e73a28eef70ee47000",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11410",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:84b64e032e8a5f9c24de95e7020c06d7112af6f6f12db2e73a28eef70ee47000"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:a15d77517b11",
+      "dbms": "dolt",
+      "title": "` SPACE(n)` results in a hang when `n` is large",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11408"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11408",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11408",
+            "source_type": "github_issue",
+            "excerpt": "` SPACE(n)` results in a hang when `n` is large",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:67da74b50ef04a3ca706094a9588561d16eea0ba48722012c32aec6dd4b23f71",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11408",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:67da74b50ef04a3ca706094a9588561d16eea0ba48722012c32aec6dd4b23f71"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e084b06da02b",
+      "dbms": "dolt",
+      "title": "`COUNT(*) OVER ()` returns 1 instead of the partition size",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11409"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11409",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11409",
+            "source_type": "github_issue",
+            "excerpt": "`COUNT(*) OVER ()` returns 1 instead of the partition size",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:039c3a330e5f23e8aec974d61c3d78d97414e43993e73f74079686335995b38b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11409",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:039c3a330e5f23e8aec974d61c3d78d97414e43993e73f74079686335995b38b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:3fe6c18b4f0a",
+      "dbms": "dolt",
+      "title": "`if(<cond>, <literal>, <column>)` in a WHERE over a view raises internal error",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11407"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11407",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11407",
+            "source_type": "github_issue",
+            "excerpt": "`if(<cond>, <literal>, <column>)` in a WHERE over a view raises internal error",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1c93962326c28f7a2a890e3c733b931af0f699110e127f51a0297895b3c66539",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11407",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1c93962326c28f7a2a890e3c733b931af0f699110e127f51a0297895b3c66539"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:5e6cada54889",
+      "dbms": "dolt",
+      "title": "Distinct window partition/order shapes collide in `PartitionId`",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11461"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11461",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11461",
+            "source_type": "github_issue",
+            "excerpt": "Distinct window partition/order shapes collide in `PartitionId`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a2d097cdca63bda1d8d3f6c18f561714f75d097ac7d1a835b8caee7ff9bc5250",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11461",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a2d097cdca63bda1d8d3f6c18f561714f75d097ac7d1a835b8caee7ff9bc5250"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:a3959e9771ae",
+      "dbms": "dolt",
+      "title": "Dolt forms incorrect `RANGE` boundaries for a correlated computed order key",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11460"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11460",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11460",
+            "source_type": "github_issue",
+            "excerpt": "Dolt forms incorrect `RANGE` boundaries for a correlated computed order key",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ad9cb5cd62ff153fa846fbfbf31ee18ff2363766647ea0f7634e7cd99c1529c7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11460",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ad9cb5cd62ff153fa846fbfbf31ee18ff2363766647ea0f7634e7cd99c1529c7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:d31dd3457255",
+      "dbms": "dolt",
+      "title": "Dolt loses an explicit full frame when a named window inherits a base window",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11458"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11458",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11458",
+            "source_type": "github_issue",
+            "excerpt": "Dolt loses an explicit full frame when a named window inherits a base window",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f8d51376c873ceeca8b09166f93fd659ce47b30945c6a0defc024a4a50d96594",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11458",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f8d51376c873ceeca8b09166f93fd659ce47b30945c6a0defc024a4a50d96594"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:08fffc5d7faa",
+      "dbms": "dolt",
+      "title": "Dolt panics on a non-numeric NTILE bucket expression",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11467"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11467",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11467",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics on a non-numeric NTILE bucket expression",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3f88201e0b6d018f574a4998f2a6c1a1ae81110c7bee2a3eab089f54c79b3e55",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11467",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3f88201e0b6d018f574a4998f2a6c1a1ae81110c7bee2a3eab089f54c79b3e55"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e42d207e6c5e",
+      "dbms": "dolt",
+      "title": "Dolt panics on an unchecked AST cast within prepared `CREATE PROCEDURE`",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11451"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11451",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11451",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics on an unchecked AST cast within prepared `CREATE PROCEDURE`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:26db1aa3c9fe17cae69ef075b33fce13c4317ebea5b03f34cd5c217855860148",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11451",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:26db1aa3c9fe17cae69ef075b33fce13c4317ebea5b03f34cd5c217855860148"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e53e5d05e091",
+      "dbms": "dolt",
+      "title": "Dolt panics when `FIRST_VALUE` receives the star placeholder",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11468"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11468",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11468",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics when `FIRST_VALUE` receives the star placeholder",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1aa62c72197a779388457b8f90be77729dd7588c9de4d2cc79d71879f49616cc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11468",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1aa62c72197a779388457b8f90be77729dd7588c9de4d2cc79d71879f49616cc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:b279aeeec3ed",
+      "dbms": "dolt",
+      "title": "Dolt panics with `Interval.Eval`within  standalone `INTERVAL` expression",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11452"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11452",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11452",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics with `Interval.Eval`within  standalone `INTERVAL` expression",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:767c7182e29b462362c7f121eaf49281db3358f84c51a1c16da3454e0de91f90",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11452",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:767c7182e29b462362c7f121eaf49281db3358f84c51a1c16da3454e0de91f90"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:2ff1936b4818",
+      "dbms": "dolt",
+      "title": "Dolt reverses the direction of a numeric `DESC RANGE` offset frame",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11459"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11459",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11459",
+            "source_type": "github_issue",
+            "excerpt": "Dolt reverses the direction of a numeric `DESC RANGE` offset frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f244481d1908b335991ec26f5df04cdb9442ee110dd3ad5130019209867b3a08",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11459",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f244481d1908b335991ec26f5df04cdb9442ee110dd3ad5130019209867b3a08"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:b463dd29a444",
+      "dbms": "dolt",
+      "title": "Dolt sibling NTILE expressions reuse the first bucket count",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11466"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11466",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11466",
+            "source_type": "github_issue",
+            "excerpt": "Dolt sibling NTILE expressions reuse the first bucket count",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9e8050370e23018d5e2edf6567f73d5c92288ed192582c72c0e7e6d2a9876a56",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11466",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9e8050370e23018d5e2edf6567f73d5c92288ed192582c72c0e7e6d2a9876a56"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:88cb9d6c3979",
+      "dbms": "dolt",
+      "title": "Empty value-window frame is incorrectly non-nullable in CTAS",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11471"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11471",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11471",
+            "source_type": "github_issue",
+            "excerpt": "Empty value-window frame is incorrectly non-nullable in CTAS",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1cbbe767225c3b7d65a1db9f412a753d93fcd73381ca268c56fa31ab4a394057",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11471",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1cbbe767225c3b7d65a1db9f412a753d93fcd73381ca268c56fa31ab4a394057"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:f51ac314813c",
+      "dbms": "dolt",
+      "title": "Final window `ORDER BY` on a set operation panics",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11462"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11462",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11462",
+            "source_type": "github_issue",
+            "excerpt": "Final window `ORDER BY` on a set operation panics",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2292411ca422396dca9f4981a2c56bb10ea9994c870bd5e325197b976a15e99b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11462",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2292411ca422396dca9f4981a2c56bb10ea9994c870bd5e325197b976a15e99b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:75b9fffab898",
+      "dbms": "dolt",
+      "title": "Legal `BIT(2)` RANGE following frame panics",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11469"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11469",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11469",
+            "source_type": "github_issue",
+            "excerpt": "Legal `BIT(2)` RANGE following frame panics",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b15b00fba9b9411f63671114abd019a00d9b64ce06dd32d8b5522d81f4fadeb2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11469",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b15b00fba9b9411f63671114abd019a00d9b64ce06dd32d8b5522d81f4fadeb2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:2deb45a5f6a6",
+      "dbms": "dolt",
+      "title": "Temporal `RANGE INTERVAL` endpoints use incompatible types",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11463"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11463",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11463",
+            "source_type": "github_issue",
+            "excerpt": "Temporal `RANGE INTERVAL` endpoints use incompatible types",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:14d6d1b7e9195699016a7c14b0d794f0306e85ff827e8e1aafd812aee93981f8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11463",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:14d6d1b7e9195699016a7c14b0d794f0306e85ff827e8e1aafd812aee93981f8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:aa3770c65229",
+      "dbms": "dolt",
+      "title": "Window `AVG` drops a non-NULL nonnumeric `VARCHAR`",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11470"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11470",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11470",
+            "source_type": "github_issue",
+            "excerpt": "Window `AVG` drops a non-NULL nonnumeric `VARCHAR`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:07571539168308d85216fb369a9d65d2feadbaa110ff73d9360688426a0049fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11470",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:07571539168308d85216fb369a9d65d2feadbaa110ff73d9360688426a0049fc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:bbd3bc9d274f",
+      "dbms": "dolt",
+      "title": "`BIN_TO_UUID` panics for a NULL swap flag",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11457"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11457",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11457",
+            "source_type": "github_issue",
+            "excerpt": "`BIN_TO_UUID` panics for a NULL swap flag",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e1d1529bda8b9dde52b73702a779fed6331dfd37c673eac9af771286f24ac9ee",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11457",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e1d1529bda8b9dde52b73702a779fed6331dfd37c673eac9af771286f24ac9ee"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:01a4a02a9621",
+      "dbms": "dolt",
+      "title": "`CHAR` PAD SPACE values split a window partition",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11464"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11464",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11464",
+            "source_type": "github_issue",
+            "excerpt": "`CHAR` PAD SPACE values split a window partition",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:eb4fbb87177e00a0716cf0e7bf67c974a2f6caacf27e2a1b10e61420fa09309a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11464",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:eb4fbb87177e00a0716cf0e7bf67c974a2f6caacf27e2a1b10e61420fa09309a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:52f282723b64",
+      "dbms": "dolt",
+      "title": "`DEFAULT(column)` reaches an unresolved placeholder during SELECT analysis",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11453"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11453",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11453",
+            "source_type": "github_issue",
+            "excerpt": "`DEFAULT(column)` reaches an unresolved placeholder during SELECT analysis",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:933535e24d6ca5f9f1f64b0be4f205b932ecd9c235eaa13da218dffbaed68b8f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11453",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:933535e24d6ca5f9f1f64b0be4f205b932ecd9c235eaa13da218dffbaed68b8f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:244a6d9f035f",
+      "dbms": "dolt",
+      "title": "`LAST_INSERT_ID(NULL)` panics on a nil result",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11454"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11454",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11454",
+            "source_type": "github_issue",
+            "excerpt": "`LAST_INSERT_ID(NULL)` panics on a nil result",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2f52b3ec026c137d89f4377550e64157fae0b88766eb1ba5abe354668ca06039",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11454",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2f52b3ec026c137d89f4377550e64157fae0b88766eb1ba5abe354668ca06039"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:a648fb7aabf3",
+      "dbms": "dolt",
+      "title": "`MIN` panics on a legal empty leading `ROWS` frame",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11465"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11465",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11465",
+            "source_type": "github_issue",
+            "excerpt": "`MIN` panics on a legal empty leading `ROWS` frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:52e21300fc40b084f7c56a1733e5364440762dd0b1ff5db192155aafd2bcd3a1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11465",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:52e21300fc40b084f7c56a1733e5364440762dd0b1ff5db192155aafd2bcd3a1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:d4d05a3aead6",
+      "dbms": "dolt",
+      "title": "`TRIM` panics on a native `TIME` value",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11455"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11455",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11455",
+            "source_type": "github_issue",
+            "excerpt": "`TRIM` panics on a native `TIME` value",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:4964cf04556ff41d5421a5567811f85c236d1038bba13f03287b93767f9c34fb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11455",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4964cf04556ff41d5421a5567811f85c236d1038bba13f03287b93767f9c34fb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:826080a2a071",
+      "dbms": "dolt",
+      "title": "`WHERE <col> <= <expr>` and `>=` return rows where <col> IS NULL on `dolt sql-server`",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11473"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11473",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11473",
+            "source_type": "github_issue",
+            "excerpt": "`WHERE <col> <= <expr>` and `>=` return rows where <col> IS NULL on `dolt sql-server`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b22d9a45621f805ede6334f622b75fd17945f3cb2c8f48e5742211e39ff6c45e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11473",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b22d9a45621f805ede6334f622b75fd17945f3cb2c8f48e5742211e39ff6c45e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:6e30cb2bf1ce",
+      "dbms": "dolt",
+      "title": "CRC32 rejects a legal binary value produced by a window",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11507"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11507",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11507",
+            "source_type": "github_issue",
+            "excerpt": "CRC32 rejects a legal binary value produced by a window",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9c7add3d3be7cb403fde8c5b6256859db20695a2d20ada9da27655f2ecebffd9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11507",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9c7add3d3be7cb403fde8c5b6256859db20695a2d20ada9da27655f2ecebffd9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:998354966616",
+      "dbms": "dolt",
+      "title": "Correlated `EXISTS` loses an empty aggregate row",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11508"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11508",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11508",
+            "source_type": "github_issue",
+            "excerpt": "Correlated `EXISTS` loses an empty aggregate row",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c7adb28a0f92aceb890c981f06c139f104f1ed20304343088ea70eaff0a513b1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11508",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c7adb28a0f92aceb890c981f06c139f104f1ed20304343088ea70eaff0a513b1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:9bf9ee901807",
+      "dbms": "dolt",
+      "title": "Date-valued window arithmetic appends a zero-time suffix",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11505"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11505",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11505",
+            "source_type": "github_issue",
+            "excerpt": "Date-valued window arithmetic appends a zero-time suffix",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:dc35954b66940ccd9c7ee4c4f785d821ad10be4759e2f71e4e88ebb3ca377b29",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11505",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dc35954b66940ccd9c7ee4c4f785d821ad10be4759e2f71e4e88ebb3ca377b29"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:1caf3c1a896e",
+      "dbms": "dolt",
+      "title": "Decimal `RANGE` offsets are rejected",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11496"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11496",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11496",
+            "source_type": "github_issue",
+            "excerpt": "Decimal `RANGE` offsets are rejected",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:69e8a7b2a827b90d3d9f4a24f076fb9b75056687f0097c085cbc5ec6353bea3d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11496",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:69e8a7b2a827b90d3d9f4a24f076fb9b75056687f0097c085cbc5ec6353bea3d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:65df63a6b85c",
+      "dbms": "dolt",
+      "title": "Dolt panics on `INET6_ATON` on a non-string SQL argument",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11493"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11493",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11493",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics on `INET6_ATON` on a non-string SQL argument",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:567db102f27d56cc4a912d1fd37f5c2c24ece32af855b3beb071a4b998faf5ee",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11493",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:567db102f27d56cc4a912d1fd37f5c2c24ece32af855b3beb071a4b998faf5ee"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:c9e968876fb5",
+      "dbms": "dolt",
+      "title": "Dolt panics on `RAND(NULL)` on a nil seed.",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11494"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11494",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11494",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics on `RAND(NULL)` on a nil seed.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fd6c2edd140a20bf057d19647a229cfa44bbd2c60b3404a22fff9d6f2301b0b9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11494",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fd6c2edd140a20bf057d19647a229cfa44bbd2c60b3404a22fff9d6f2301b0b9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e69858a48024",
+      "dbms": "dolt",
+      "title": "Dolt panics on `ROUND` on a date-valued window result",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11495"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11495",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11495",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics on `ROUND` on a date-valued window result",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c75d1c7d262b5f17e4877da1fb6fbb921649e2d8175e3a3b12af96719671c53c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11495",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c75d1c7d262b5f17e4877da1fb6fbb921649e2d8175e3a3b12af96719671c53c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:b83a94cac374",
+      "dbms": "dolt",
+      "title": "Dolt panics on a numeric `FORMAT` locale from a window expression",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11517"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11517",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11517",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics on a numeric `FORMAT` locale from a window expression",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c379987bf921e53d4dece556f2e6906c2e2126bbb8029e50914faa9a73ed611e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11517",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c379987bf921e53d4dece556f2e6906c2e2126bbb8029e50914faa9a73ed611e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:1a93a5e84510",
+      "dbms": "dolt",
+      "title": "Dolt panics on empty `LIKE ... ESCAPE ''` within window expressions",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11518"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11518",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11518",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics on empty `LIKE ... ESCAPE ''` within window expressions",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1153029362c1c6c2ea234e586621c834af79cf9635bed3a96857c6c7fb8c4bf2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11518",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1153029362c1c6c2ea234e586621c834af79cf9635bed3a96857c6c7fb8c4bf2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:9a14d49bb464",
+      "dbms": "dolt",
+      "title": "Dolt panics on empty `POLYGON` WKT",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11492"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11492",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11492",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics on empty `POLYGON` WKT",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b0e9f5afd8089ff1e97a98d7a43ecaae34f0defdb008b483e93e2ac9f688913b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11492",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b0e9f5afd8089ff1e97a98d7a43ecaae34f0defdb008b483e93e2ac9f688913b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:29c17ba27898",
+      "dbms": "dolt",
+      "title": "Dolt panics on recursive CTE column-arity error.",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11491"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11491",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11491",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics on recursive CTE column-arity error.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1c32e02376f30131a239261c12bf06da8c4b83fc248cfca9de6ac33d600610c1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11491",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1c32e02376f30131a239261c12bf06da8c4b83fc248cfca9de6ac33d600610c1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:cce6b8479ffe",
+      "dbms": "dolt",
+      "title": "Dolt reaches an internal error on grouped one-column input.",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11500"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11500",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11500",
+            "source_type": "github_issue",
+            "excerpt": "Dolt reaches an internal error on grouped one-column input.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e0a87f6d54352771437d9181ff91379eda32aaff8291d693a74da8ba05efed1a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11500",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e0a87f6d54352771437d9181ff91379eda32aaff8291d693a74da8ba05efed1a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:7a7e4f80d71c",
+      "dbms": "dolt",
+      "title": "Dolt reaches an internal error on invalid window `ORDER BY` alias.",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11502"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11502",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11502",
+            "source_type": "github_issue",
+            "excerpt": "Dolt reaches an internal error on invalid window `ORDER BY` alias.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3c0c9f3bfe7b76faf756737a333999c87cfeb9c9bb307d422cb441e74914c085",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11502",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3c0c9f3bfe7b76faf756737a333999c87cfeb9c9bb307d422cb441e74914c085"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:007d208d3824",
+      "dbms": "dolt",
+      "title": "Dolt window expression omits `LIKE ESCAPE`",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11498"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11498",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11498",
+            "source_type": "github_issue",
+            "excerpt": "Dolt window expression omits `LIKE ESCAPE`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:548954e9b68a98702a839123140ea4bdc94bd54459d0a452d606cad009a74f55",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11498",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:548954e9b68a98702a839123140ea4bdc94bd54459d0a452d606cad009a74f55"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:92ee9cc8f76f",
+      "dbms": "dolt",
+      "title": "Dolt: `DISTINCT` window output with a composite outer sort reaches a stale field index",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11504"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11504",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11504",
+            "source_type": "github_issue",
+            "excerpt": "Dolt: `DISTINCT` window output with a composite outer sort reaches a stale field index",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:863fed4954074ca3c3ea09e79663877c85cc5f20ce21d115adcbeca39888747d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11504",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:863fed4954074ca3c3ea09e79663877c85cc5f20ce21d115adcbeca39888747d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:f5adb340216b",
+      "dbms": "dolt",
+      "title": "Dolt: explicit Unicode-collation RANGE peers are split",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11506"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11506",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11506",
+            "source_type": "github_issue",
+            "excerpt": "Dolt: explicit Unicode-collation RANGE peers are split",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f87e453b3d4e4788aae9e87529905ca7a9f39dac95df58b8086a265947a6a76e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11506",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f87e453b3d4e4788aae9e87529905ca7a9f39dac95df58b8086a265947a6a76e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:90d0566b7028",
+      "dbms": "dolt",
+      "title": "Identical nondeterministic window expressions are incorrectly merged",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11499"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11499",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11499",
+            "source_type": "github_issue",
+            "excerpt": "Identical nondeterministic window expressions are incorrectly merged",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e42158c5c88fbaa9ba12070ef518416276af99ef257327f0a44380b3a99e9d19",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11499",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e42158c5c88fbaa9ba12070ef518416276af99ef257327f0a44380b3a99e9d19"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:54149009d7ba",
+      "dbms": "dolt",
+      "title": "Nested CTE scope disappears inside a recursive CTE with a window query",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11514"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11514",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11514",
+            "source_type": "github_issue",
+            "excerpt": "Nested CTE scope disappears inside a recursive CTE with a window query",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7a2701ad1a9731ce513ae9a04c552ca595545ae4e7d2cd6aaaaede628406aa5c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11514",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7a2701ad1a9731ce513ae9a04c552ca595545ae4e7d2cd6aaaaede628406aa5c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:7dc8ee65a312",
+      "dbms": "dolt",
+      "title": "Nested multi-column VIEW + self `EXISTS` → ERROR 1105 column-count mismatch",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11501"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11501",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11501",
+            "source_type": "github_issue",
+            "excerpt": "Nested multi-column VIEW + self `EXISTS` → ERROR 1105 column-count mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2e25deeb6de6df02c7fc9a340e45bcc29cebf481a5ed4e48364aea09d2c9ca85",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11501",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2e25deeb6de6df02c7fc9a340e45bcc29cebf481a5ed4e48364aea09d2c9ca85"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e125731c1217",
+      "dbms": "dolt",
+      "title": "UPDATE incorrectly permits a windowed target-table subquery",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11516"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11516",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11516",
+            "source_type": "github_issue",
+            "excerpt": "UPDATE incorrectly permits a windowed target-table subquery",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:339df5adc551c37ff9a179630d97dc277c22eee7becfa06166315927208bc5ea",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11516",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:339df5adc551c37ff9a179630d97dc277c22eee7becfa06166315927208bc5ea"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:e83f324eabd4",
+      "dbms": "dolt",
+      "title": "Window expression de-duplication folds case-sensitive literals",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11497"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11497",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11497",
+            "source_type": "github_issue",
+            "excerpt": "Window expression de-duplication folds case-sensitive literals",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:86229d3d7a0ec955fc4e8befd07cd80b4198bcfce78679110d81ab515a5fe11c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11497",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:86229d3d7a0ec955fc4e8befd07cd80b4198bcfce78679110d81ab515a5fe11c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:638a23c313a1",
+      "dbms": "dolt",
+      "title": "Windowed `HOUR()` returns `NULL` for a valid time-only value",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11512"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11512",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11512",
+            "source_type": "github_issue",
+            "excerpt": "Windowed `HOUR()` returns `NULL` for a valid time-only value",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1f961958572b2bfecd395b804306a005332b1c21d5c01cb6e6080911b2034b50",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11512",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1f961958572b2bfecd395b804306a005332b1c21d5c01cb6e6080911b2034b50"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:cce8517384f6",
+      "dbms": "dolt",
+      "title": "Windowed `TRIM(LEADING FROM str)` is rejected by the parser",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11511"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11511",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11511",
+            "source_type": "github_issue",
+            "excerpt": "Windowed `TRIM(LEADING FROM str)` is rejected by the parser",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d23ee3de5702e3bc6d1a486da4437ed55530430710dab1f7bd8cfa1eab394c95",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11511",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d23ee3de5702e3bc6d1a486da4437ed55530430710dab1f7bd8cfa1eab394c95"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:0ddcbc2d11d5",
+      "dbms": "dolt",
+      "title": "Windowed recursive CTE names collide with a differently cased base table",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11513"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11513",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11513",
+            "source_type": "github_issue",
+            "excerpt": "Windowed recursive CTE names collide with a differently cased base table",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:bdc24f6f8125a73b7b1b09ee6aeaf5bee43b519f69f1ae628e7695898feee905",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11513",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bdc24f6f8125a73b7b1b09ee6aeaf5bee43b519f69f1ae628e7695898feee905"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:758493850a19",
+      "dbms": "dolt",
+      "title": "`COUNT(<column>)` counts the wrong column on `dolt sql-server`",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11521"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11521",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11521",
+            "source_type": "github_issue",
+            "excerpt": "`COUNT(<column>)` counts the wrong column on `dolt sql-server`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:45243f7fc2883f47726e84bdaf3e2629f5b51663ff6dc74d4511d8844ba6ec68",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11521",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:45243f7fc2883f47726e84bdaf3e2629f5b51663ff6dc74d4511d8844ba6ec68"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:1f81ba292c11",
+      "dbms": "dolt",
+      "title": "`INET_NTOA` corrupts valid IPv4 values above signed 32-bit",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11510"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11510",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11510",
+            "source_type": "github_issue",
+            "excerpt": "`INET_NTOA` corrupts valid IPv4 values above signed 32-bit",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f95963bed4d076d616e97cf51f4eb666c56ef83ea4a6bd9b555b0655b2c21c8a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11510",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f95963bed4d076d616e97cf51f4eb666c56ef83ea4a6bd9b555b0655b2c21c8a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:dea50d0c1f7f",
+      "dbms": "dolt",
+      "title": "`ORDER BY` + huge `LIMIT` (≥ 2^62) panics",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11503"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11503",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11503",
+            "source_type": "github_issue",
+            "excerpt": "`ORDER BY` + huge `LIMIT` (≥ 2^62) panics",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:12d4ca6adef6cb7163e73a561147871c73890f8b5c7c3421d7a119be97ffbe59",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11503",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:12d4ca6adef6cb7163e73a561147871c73890f8b5c7c3421d7a119be97ffbe59"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:a9df8d79f8f5",
+      "dbms": "dolt",
+      "title": "`STR_TO_DATE()` errors instead of propagating `NULL`",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11509"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11509",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11509",
+            "source_type": "github_issue",
+            "excerpt": "`STR_TO_DATE()` errors instead of propagating `NULL`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3c9bf2e3a41df626d5df99c49d964cf4d747bc2de754ffa2e264eaea0d51a78b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11509",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3c9bf2e3a41df626d5df99c49d964cf4d747bc2de754ffa2e264eaea0d51a78b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:cbf9c2ff3a51",
+      "dbms": "dolt",
+      "title": "`CEIL`/`FLOOR` of a windowed `DECIMAL`  returns unexpected results",
+      "reported_date": "2026-08-16",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11542"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11542",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11542",
+            "source_type": "github_issue",
+            "excerpt": "`CEIL`/`FLOOR` of a windowed `DECIMAL`  returns unexpected results",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:4da2496beaf865b7580b2c072e949780d8f39fab6565841a209cbacf1080c291",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11542",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4da2496beaf865b7580b2c072e949780d8f39fab6565841a209cbacf1080c291"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:07debea4dcb4",
+      "dbms": "dolt",
+      "title": "Dolt drops sql_select_limit",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11549"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11549",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11549",
+            "source_type": "github_issue",
+            "excerpt": "Dolt drops sql_select_limit",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ddb944aa91ee9d913caea1ecf0eb3ff1198063f0ae3342cb02e9fb8e1abbd86c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11549",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ddb944aa91ee9d913caea1ecf0eb3ff1198063f0ae3342cb02e9fb8e1abbd86c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:9c0f525baeb3",
+      "dbms": "dolt",
+      "title": "Dolt exposes a user INVISIBLE column through SELECT *",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11551"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11551",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11551",
+            "source_type": "github_issue",
+            "excerpt": "Dolt exposes a user INVISIBLE column through SELECT *",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:554371f8d79fbdf717a752b1a7e9194c85aa868df6282adc0f4c371ec8b0a0bc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11551",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:554371f8d79fbdf717a752b1a7e9194c85aa868df6282adc0f4c371ec8b0a0bc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:bf6dee0891a3",
+      "dbms": "dolt",
+      "title": "Dolt ignores the PM marker in STR_TO_DATE",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11544"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11544",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11544",
+            "source_type": "github_issue",
+            "excerpt": "Dolt ignores the PM marker in STR_TO_DATE",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2d163ded93b1ff4662d232e0ceb0ddd004d8567cb912fc670e94af8944434d2e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11544",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2d163ded93b1ff4662d232e0ceb0ddd004d8567cb912fc670e94af8944434d2e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:4fd122264b33",
+      "dbms": "dolt",
+      "title": "Dolt mis-evaluates a correlated anti-join over a derived ROW_NUMBER() result",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11548"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11548",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11548",
+            "source_type": "github_issue",
+            "excerpt": "Dolt mis-evaluates a correlated anti-join over a derived ROW_NUMBER() result",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f54467c2342199f50e42157fe042f8483ecd76c962ed203aaea900dc964b3b80",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11548",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f54467c2342199f50e42157fe042f8483ecd76c962ed203aaea900dc964b3b80"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:b1eed4a8975b",
+      "dbms": "dolt",
+      "title": "Dolt raises an internal child-count error for CURRENT_TIME(precision)",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11543"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11543",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11543",
+            "source_type": "github_issue",
+            "excerpt": "Dolt raises an internal child-count error for CURRENT_TIME(precision)",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5bc24bce081c15fe27d15d1d19d6ec13bebf8af71695ec4bbc552a98c4350b92",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11543",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5bc24bce081c15fe27d15d1d19d6ec13bebf8af71695ec4bbc552a98c4350b92"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:c6f9b178b318",
+      "dbms": "dolt",
+      "title": "Dolt reaches an analyzer internal error for a legal correlated LEFT JOIN LATERAL",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11552"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11552",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11552",
+            "source_type": "github_issue",
+            "excerpt": "Dolt reaches an analyzer internal error for a legal correlated LEFT JOIN LATERAL",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1e32dd467c34f2fb39797c862a7993f905a14d91f5042ac8e62bec8407d5285b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11552",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1e32dd467c34f2fb39797c862a7993f905a14d91f5042ac8e62bec8407d5285b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:1d336af0fdb4",
+      "dbms": "dolt",
+      "title": "Dolt rejects a legal REGEXP_REPLACE position after the source",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11547"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11547",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11547",
+            "source_type": "github_issue",
+            "excerpt": "Dolt rejects a legal REGEXP_REPLACE position after the source",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:0b04f79c3b897415326e83bdf21e47974d9e2861dd4757bbed7ea8064c30f692",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11547",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0b04f79c3b897415326e83bdf21e47974d9e2861dd4757bbed7ea8064c30f692"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:ed523babade4",
+      "dbms": "dolt",
+      "title": "Dolt returns a wrong result for REGEXP_LIKE across partitions",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11550"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11550",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11550",
+            "source_type": "github_issue",
+            "excerpt": "Dolt returns a wrong result for REGEXP_LIKE across partitions",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:8e7222fdea49944b3c56f560379400d20b43ec143104b0613c9a1a999e1bc88e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11550",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8e7222fdea49944b3c56f560379400d20b43ec143104b0613c9a1a999e1bc88e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:7cc7392ad4b1",
+      "dbms": "dolt",
+      "title": "Dolt swallows a child-expression evaluation error in SUM",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11545"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11545",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11545",
+            "source_type": "github_issue",
+            "excerpt": "Dolt swallows a child-expression evaluation error in SUM",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:511aa243c21585ef2b1cd742da25131327dfedd1bdbd28e4efb8c2d71ef5c737",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11545",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:511aa243c21585ef2b1cd742da25131327dfedd1bdbd28e4efb8c2d71ef5c737"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:c1b96b31ff2d",
+      "dbms": "dolt",
+      "title": "Dolt uppercases a lowercase non-ASCII initial in SOUNDEX",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11546"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11546",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11546",
+            "source_type": "github_issue",
+            "excerpt": "Dolt uppercases a lowercase non-ASCII initial in SOUNDEX",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c2d6c65c1fa628749e4ed0c7a064cb0cb0b95a576fb14c3abbf9603a76082e93",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11546",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c2d6c65c1fa628749e4ed0c7a064cb0cb0b95a576fb14c3abbf9603a76082e93"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:71f789e8b69f",
+      "dbms": "dolt",
+      "title": "Dolt collides distinct text tuples in COUNT(DISTINCT ...) before a window wrapper",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11562"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11562",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11562",
+            "source_type": "github_issue",
+            "excerpt": "Dolt collides distinct text tuples in COUNT(DISTINCT ...) before a window wrapper",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:eb920945176e7c568527740961bea8c1dbf618020edbd5a7cab4be5a2b2b87ad",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11562",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:eb920945176e7c568527740961bea8c1dbf618020edbd5a7cab4be5a2b2b87ad"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:13432cf8d46c",
+      "dbms": "dolt",
+      "title": "Dolt ignores default backslash escaping for wildcard characters in a window child",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11558"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11558",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11558",
+            "source_type": "github_issue",
+            "excerpt": "Dolt ignores default backslash escaping for wildcard characters in a window child",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b10ae0ec7fbc4bcb6b69f0b4c7354a539e7e0dcb6c20b43692b4be999d787e7e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11558",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b10ae0ec7fbc4bcb6b69f0b4c7354a539e7e0dcb6c20b43692b4be999d787e7e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:9c7f63f33610",
+      "dbms": "dolt",
+      "title": "Dolt keeps boolean true separate from integer 1 in recursive UNION DISTINCT",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11563"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11563",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11563",
+            "source_type": "github_issue",
+            "excerpt": "Dolt keeps boolean true separate from integer 1 in recursive UNION DISTINCT",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b6660abf1d8c984c6550de17ca5418c14e53bb52f3ef218c64b068174c7b0f9d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11563",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b6660abf1d8c984c6550de17ca5418c14e53bb52f3ef218c64b068174c7b0f9d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:3b24b7710594",
+      "dbms": "dolt",
+      "title": "Dolt loses expression-index metadata in a windowed information_schema projection",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11559"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11559",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11559",
+            "source_type": "github_issue",
+            "excerpt": "Dolt loses expression-index metadata in a windowed information_schema projection",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5eb7dab9211fec7b7b86e3421cfd1745f5c0e5cef4ae54cad83dcd3c81511d2c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11559",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5eb7dab9211fec7b7b86e3421cfd1745f5c0e5cef4ae54cad83dcd3c81511d2c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:ca077bc80ef7",
+      "dbms": "dolt",
+      "title": "Dolt panics when LOAD_FILE receives a numeric argument below a window",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11564"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11564",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11564",
+            "source_type": "github_issue",
+            "excerpt": "Dolt panics when LOAD_FILE receives a numeric argument below a window",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:deadf1801002dc0d2b1c75065466f9e32ad239626c6068ce22ba21f426a840e0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11564",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:deadf1801002dc0d2b1c75065466f9e32ad239626c6068ce22ba21f426a840e0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:6f85d63564b4",
+      "dbms": "dolt",
+      "title": "Dolt returns NULL for a 25-hour TIME in a windowed TIME_TO_SEC call",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11560"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11560",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11560",
+            "source_type": "github_issue",
+            "excerpt": "Dolt returns NULL for a 25-hour TIME in a windowed TIME_TO_SEC call",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e303356668be303a28bf523454ae12533b57362d19b5dd6c3dd9f7333414a927",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11560",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e303356668be303a28bf523454ae12533b57362d19b5dd6c3dd9f7333414a927"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:2e21d1c58a31",
+      "dbms": "dolt",
+      "title": "Dolt skips empty strings in GROUP_CONCAT before a window wrapper",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/11561"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/11561",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/11561",
+            "source_type": "github_issue",
+            "excerpt": "Dolt skips empty strings in GROUP_CONCAT before a window wrapper",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:25f6f3f5f44b50196352fe854c3215a0f0121ecafb87d7b4f5871591834f2496",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/dolthub/dolt/issues/11561",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:25f6f3f5f44b50196352fe854c3215a0f0121ecafb87d7b4f5871591834f2496"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:dolt:943469b752ca",
+      "dbms": "dolt",
+      "title": "Invalid CrossJoin in query plan",
+      "reported_date": "2026-11-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/dolthub/dolt/issues/10304"
+      },
+      "primary_url": "https://github.com/dolthub/dolt/issues/10304",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/dolthub/dolt/issues/10304",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500), c1 INT);\nCREATE TABLE t6(t6c0 VARCHAR(500), t6c1 INT, PRIMARY KEY(t6c0));\nCREATE VIEW v0(c0) AS SELECT t0.c1 FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:50d22db1d8ac0bb2db9f37cc93c25e4b8575109ada965cf18b14b26c6aaf86f5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:50d22db1d8ac0bb2db9f37cc93c25e4b8575109ada965cf18b14b26c6aaf86f5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:doris:2dfa8a3d9d09",
+      "dbms": "doris",
+      "title": "[Bug]  SELECT DISTINCT returns duplicate column with aggregate key",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "report": "https://github.com/apache/doris/issues/36072"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/apache/doris/issues/36072\nSELECT DISTINCT does not work with aggregate key column",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its doris provider, as the field bug36072 that works around it.",
+            "content_sha256": "sha256:6bac99bd788eeef7afaccaf5e1a7213cbda555470ea12bbaaed7d160f530fc43",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:6bac99bd788eeef7afaccaf5e1a7213cbda555470ea12bbaaed7d160f530fc43"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/36072",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:doris:d7743c7147b3",
+      "dbms": "doris",
+      "title": "[Bug] ERROR occur in nested subqueries with same column name and union",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "report": "https://github.com/apache/doris/issues/19611"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/apache/doris/issues/19611\nERROR occur in nested subqueries with same column name and union",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its doris provider, as the field bug19611 that works around it.",
+            "content_sha256": "sha256:0cca170487c1199ba8ee10b8089b2ee1c171731bb8e697b0e33d5a6aab273584",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:0cca170487c1199ba8ee10b8089b2ee1c171731bb8e697b0e33d5a6aab273584"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/19611",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:doris:1838c6f96b0e",
+      "dbms": "doris",
+      "title": "[Bug] NULL expression from function not recognised with BETWEEN and floats",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "report": "https://github.com/apache/doris/issues/36070"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/apache/doris/issues/36070\nExpression evaluate to NULL but is treated as FALSE in where clause",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its doris provider, as the field bug36070 that works around it.",
+            "content_sha256": "sha256:26d7f5ad1961c46463fcd8f051e7a32d869d1c2a4f5329c3e8a460b4ae97a1a5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:26d7f5ad1961c46463fcd8f051e7a32d869d1c2a4f5329c3e8a460b4ae97a1a5"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/36070",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:doris:9d7798bd5098",
+      "dbms": "doris",
+      "title": "[Bug] SELECT DISTINCT returns wrong value for UNIQUE model",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "report": "https://github.com/apache/doris/issues/36343"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/apache/doris/issues/36343\nWrong result with SELECT DISTINCT and UNIQUE model",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its doris provider, as the field bug36343 that works around it.",
+            "content_sha256": "sha256:87fdae95bfc4e88249d1d1cfac9c7f61e218d64f48fb76fd1e330223921d8110",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:87fdae95bfc4e88249d1d1cfac9c7f61e218d64f48fb76fd1e330223921d8110"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/36343",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:doris:b6d376dad478",
+      "dbms": "doris",
+      "title": "[Bug] WHERE / HAVING `TIMESTAMP 'CURRENT_TIMESTAMP' IS NULL` retrieves unwanted rows",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "report": "https://github.com/apache/doris/issues/36342"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/apache/doris/issues/36342\nWrong result with INNER JOIN and CURRENT_TIMESTAMP",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its doris provider, as the field bug36342 that works around it.",
+            "content_sha256": "sha256:75c67c0b082a9b6f3e926c3818ee3e5c48492794298fb24d5b95a4b24eeb1271",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:75c67c0b082a9b6f3e926c3818ee3e5c48492794298fb24d5b95a4b24eeb1271"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/36342",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:doris:0da83985c9ba",
+      "dbms": "doris",
+      "title": "[Bug] Wrong result with TINYINT column with value -1049190528",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "report": "https://github.com/apache/doris/issues/36351"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/apache/doris/issues/36351\nWrong result with TINYINT column with value -1049190528",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its doris provider, as the field bug36351 that works around it.",
+            "content_sha256": "sha256:03cdb3c81cade0f56957b324fe859cc6045cc39c788c02ab07ff66907d2323a8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:03cdb3c81cade0f56957b324fe859cc6045cc39c788c02ab07ff66907d2323a8"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/36351",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:doris:3afc5d1695ad",
+      "dbms": "doris",
+      "title": "[Bug] Wrong result with `LEFT JOIN` and filtering with `IN` operation",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "report": "https://github.com/apache/doris/issues/36346"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/apache/doris/issues/36346\nWrong result with LEFT JOIN SELECT DISTINCT and IN operation",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its doris provider, as the field bug36346 that works around it.",
+            "content_sha256": "sha256:237ae3273c62eff0ac1916f6e3974ecc0841a9cb5ffdf059a7a25c588b32e568",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/doris/DorisBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:237ae3273c62eff0ac1916f6e3974ecc0841a9cb5ffdf059a7a25c588b32e568"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/36346",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:doris:c89c98bf971d",
+      "dbms": "doris",
+      "title": "BE crashed",
+      "reported_date": "2020-06-15",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "sduzh",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/3874"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/3874",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/3874",
+            "source_type": "github_issue",
+            "excerpt": "BE crashed when running the [sqlancer](https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8aaf67c0a761be4bc02267cda6b27e94dabbda8c2beca75eb5ebca6eab091d3c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/3874",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8aaf67c0a761be4bc02267cda6b27e94dabbda8c2beca75eb5ebca6eab091d3c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:doris:024e8627f61f",
+      "dbms": "doris",
+      "title": "[Bug] Logical bug about `where true not in (columns)`",
+      "reported_date": "2023-03-11",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "ChaseHuangxu",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/17697"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/17697",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/17697",
+            "source_type": "github_issue",
+            "excerpt": "I'm trying to add Apache Doris support to SQLancer, this may be a logic bug found by @[SQLancer](https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:091f9499785cb6b6b7c870cf5fe6bc35ecd91053e8917b996d432f28bc95ee67",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/17697",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:091f9499785cb6b6b7c870cf5fe6bc35ecd91053e8917b996d432f28bc95ee67"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:doris:2e3c1a37ec9b",
+      "dbms": "doris",
+      "title": "[Bug] Cannot use `between and` in boolean column and doris throws Unexpected exception",
+      "reported_date": "2023-03-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "ChaseHuangxu",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/17700"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/17700",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/17700",
+            "source_type": "github_issue",
+            "excerpt": "I'm trying to add Apache Doris support to SQLancer, this may be a bug found by @[SQLancer](https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:98b0da3c0c3d5664e8cce1a01be0f6d027c1873d6867a36fcb84d1ade3790e0d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/17700",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:98b0da3c0c3d5664e8cce1a01be0f6d027c1873d6867a36fcb84d1ade3790e0d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:doris:fddd33e6c867",
+      "dbms": "doris",
+      "title": "[Bug] Two similar QUERYs return different results",
+      "reported_date": "2023-03-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "ChaseHuangxu",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/17705"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/17705",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/17705",
+            "source_type": "github_issue",
+            "excerpt": "I'm trying to add Apache Doris support to SQLancer, this may be a logic bug found by @[SQLancer](https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:8c6236c66e89c4b957b0d27e6b565e5adc30392b9e2f1dc98977ffb4577e55c6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/17705",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8c6236c66e89c4b957b0d27e6b565e5adc30392b9e2f1dc98977ffb4577e55c6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:doris:78ee608f022b",
+      "dbms": "doris",
+      "title": "[Bug] Wrong result when SQL contain `where column not in (values)`",
+      "reported_date": "2023-03-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "ChaseHuangxu",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/17701"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/17701",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/17701",
+            "source_type": "github_issue",
+            "excerpt": "I'm trying to add Apache Doris support to SQLancer, this may be a logic bug found by @[SQLancer](https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:208fc456e93d55479d9d034fe06956bee9184f64df136006cb147845e66939b8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/17701",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:208fc456e93d55479d9d034fe06956bee9184f64df136006cb147845e66939b8"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:doris:54ec9a176365",
+      "dbms": "doris",
+      "title": "[Bug] Different result of `having not ($value in column)` and `having ($value not in column)`",
+      "reported_date": "2023-05-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "codenohup",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/19374"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/19374",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/19374",
+            "source_type": "github_issue",
+            "excerpt": "I'm trying to add Apache Doris support to SQLancer, this may be a logic bug found by @[SQLancer](https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c1b04b36ba2d119824833e3a9d9894ab5abfcf06d922f20f01e42ca12621214d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/19374",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c1b04b36ba2d119824833e3a9d9894ab5abfcf06d922f20f01e42ca12621214d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:doris:600cbabf22ad",
+      "dbms": "doris",
+      "title": "[Bug] Internal Error occur in GroupBy&Having sql",
+      "reported_date": "2023-05-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "codenohup",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/19370"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/19370",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/19370",
+            "source_type": "github_issue",
+            "excerpt": "I'm trying to add Apache Doris support to SQLancer, this may be a logic bug found by @[SQLancer](https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:4172f22de7f4e44b989999467c4d779a11a3674121973bbd548e7e6d949c1dd0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/19370",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4172f22de7f4e44b989999467c4d779a11a3674121973bbd548e7e6d949c1dd0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:doris:7cd946c3b63c",
+      "dbms": "doris",
+      "title": "[Bug] Wrong result when `right outer join and where false`",
+      "reported_date": "2023-05-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "ChaseHuangxu",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/19613"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/19613",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/19613",
+            "source_type": "github_issue",
+            "excerpt": "I'm trying to add Apache Doris support to SQLancer, this may be a logic bug found by @[SQLancer](https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:669fd6ca7b786b03b8e1762b8e63c68f2c0b5fbb820b50b58f84221047443ea3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/19613",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:669fd6ca7b786b03b8e1762b8e63c68f2c0b5fbb820b50b58f84221047443ea3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:doris:7ec49e7edfea",
+      "dbms": "doris",
+      "title": "[Bug] Wrong result when `value like column from table_join`",
+      "reported_date": "2023-05-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "ChaseHuangxu",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/19614"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/19614",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/19614",
+            "source_type": "github_issue",
+            "excerpt": "I'm trying to add Apache Doris support to SQLancer, this may be a logic bug found by @[SQLancer](https://github.com/sqlancer/sqlancer).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:8547441ade8199af695d742fa48f1b765a961463b9bf91b8234be72e974ff0e9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/19614",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8547441ade8199af695d742fa48f1b765a961463b9bf91b8234be72e974ff0e9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:doris:662c693a397f",
+      "dbms": "doris",
+      "title": "[Bug] Expression evaluate to NULL but is treated as FALSE in where clause",
+      "reported_date": "2024-05-26",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "malwaregarry",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/35404"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/35404",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/35404",
+            "source_type": "github_issue",
+            "excerpt": "[Bug] Expression evaluate to NULL but is treated as FALSE in where clause",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:767ec16ce9601e5717014fa2be3abe60b2895dcf5995a5e2adb8996333062cb4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "malwaregarry is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/35404",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:767ec16ce9601e5717014fa2be3abe60b2895dcf5995a5e2adb8996333062cb4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:doris:6924b702015f",
+      "dbms": "doris",
+      "title": "[Bug] Bug with BIGINT boundary values",
+      "reported_date": "2024-06-09",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "malwaregarry",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/36077"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/36077",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/36077",
+            "source_type": "github_issue",
+            "excerpt": "[Bug] Bug with BIGINT boundary values",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:35d96f69f9626ee3035a84c294843a82210ed14f1d35d35b4a1c7c68823aed8f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "malwaregarry is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/36077",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:35d96f69f9626ee3035a84c294843a82210ed14f1d35d35b4a1c7c68823aed8f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:doris:c45040b4a2f2",
+      "dbms": "doris",
+      "title": "[Bug] Error: num_columns!=merged_columns.size(), num_columns=3, merged_columns.size()=2",
+      "reported_date": "2024-06-09",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "malwaregarry",
+      "links": {
+        "report": "https://github.com/apache/doris/issues/36079"
+      },
+      "primary_url": "https://github.com/apache/doris/issues/36079",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/apache/doris/issues/36079",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BIGINT NOT NULL) DISTRIBUTED BY HASH (c0) PROPERTIES (\"replication_num\" = \"1\");\r\nCREATE TABLE t1(c0 INT) UNIQUE KEY(c0) DISTRIBUTED BY HASH (c0) PROPERTIES (\"replication_num\" = \"1\");\r\nINSERT INTO t1 (c0) VALUES (1), (2), (3);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bb8f026acf221572ab94eb55a335427ef1306103bb31c04aae9efd52417ef9e7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/apache/doris/issues/36079",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bb8f026acf221572ab94eb55a335427ef1306103bb31c04aae9efd52417ef9e7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:9b79a5a4a152",
+      "dbms": "duckdb",
+      "title": "Unexpected Result When Using `IS DISTINCT FROM` and `CASE WHEN` in a JOIN clause",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/12181"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/12181",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/12181",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT, c1 INT);\r\nCREATE TABLE t0(c0 INT);\r\nINSERT INTO t1 (c0, c1) VALUES (0, 1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:848e61a227c3b13f7771e056b56c0ce5aac9a88ad24364c0a00214e9fb892e28",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:848e61a227c3b13f7771e056b56c0ce5aac9a88ad24364c0a00214e9fb892e28"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:591028c0ebc6",
+      "dbms": "duckdb",
+      "title": "Unexpected result after creating index on varchar column",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/13785"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/13785",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/13785",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 VARCHAR);\r\nINSERT INTO t0(c0) VALUES ('a');\r\nCREATE INDEX t0i0 ON t0(c0 );\r\nINSERT INTO t0(c0) VALUES ('a');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:390279622fd6adb9481bff700d12fd9e142d024754a3f8cdb327ffd4c70dba4a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:390279622fd6adb9481bff700d12fd9e142d024754a3f8cdb327ffd4c70dba4a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:7d4c5b445d0f",
+      "dbms": "duckdb",
+      "title": "Unexpected result when casting negative integer to BIT",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/13506"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/13506",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/13506",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t1( c1 INTEGER);\r\nINSERT INTO t1(c1) VALUES (-1);\r\n\r\nSELECT t1.c1 FROM t1; -- -1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:311c1229661b53aebd1a30f9ed5ecd3d55cd48121b60e4e8d33e29bc40637d1d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:311c1229661b53aebd1a30f9ed5ecd3d55cd48121b60e4e8d33e29bc40637d1d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8a56ac6cb5e4",
+      "dbms": "duckdb",
+      "title": "Unexpected result when using IN with DATE values",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/13380"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/13380",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/13380",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 TIMESTAMP) ;\r\nINSERT INTO t1(c0) VALUES ('2024-08-09 14:48:00');\r\n\r\nSELECT * FROM t1; -- 2024-08-09 14:48:00",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:c2256e4c9b9e1bafea5e4962e80946fc3a31551a455f044aa982c4a4835eea5e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:c2256e4c9b9e1bafea5e4962e80946fc3a31551a455f044aa982c4a4835eea5e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c3617570d2d5",
+      "dbms": "duckdb",
+      "title": "Unexpected result when using IN with TIME values",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/13813"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/13813",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/13813",
+            "source_type": "github_issue",
+            "excerpt": "CREATE   TABLE  t1(c0 TIME WITH TIME ZONE);\r\nINSERT INTO t1(c0) VALUES ('12:34:56');\r\n\r\nSELECT t1.c0 FROM t1; -- 12:34:56+08",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b301478ada4e80812f82222e198ddd4a06abf4a1c0a66579e1e5f21f8374cae1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b301478ada4e80812f82222e198ddd4a06abf4a1c0a66579e1e5f21f8374cae1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:0c887e1de0ec",
+      "dbms": "duckdb",
+      "title": "`INTERNAL Error: Unimplemented type for normalify` in select query",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/feldera/sqlancer/blob/main/src/sqlancer/duckdb/DuckDBBugs.java",
+        "report": "https://github.com/duckdb/duckdb/issues/13933"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/feldera/sqlancer/blob/main/src/sqlancer/duckdb/DuckDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/duckdb/duckdb/issues/13933",
+            "excerpt_is_verbatim": true,
+            "note": "feldera/sqlancer records this bug in its duckdb provider, as the field bug13933 that works around it.",
+            "content_sha256": "sha256:0c887e1de0ec013ed65d2c4bff846f7eb688647f8ae733a0332e0bcfdaa6ad85",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/feldera/sqlancer/blob/main/src/sqlancer/duckdb/DuckDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:0c887e1de0ec013ed65d2c4bff846f7eb688647f8ae733a0332e0bcfdaa6ad85"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/13933",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:duckdb:4f9932da6dfe",
+      "dbms": "duckdb",
+      "title": "A comparison column=column unexpectedly evaluates to TRUE for column=NULL",
+      "reported_date": "2020-04-07",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/490",
+        "fix": "https://github.com/cwida/duckdb/commit/f18ef427cfbe75f76abc40a00937dc1bea675012"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/490",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/490\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/f18ef427cfbe75f76abc40a00937dc1bea675012\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:66936363ad7cafe5d098b179f09f9aec1457415edfc316ece13ffef428744ae5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/490",
+            "source_type": "github_issue",
+            "excerpt": "A comparison column=column unexpectedly evaluates to TRUE for column=NULL",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:21d983f3e59570aef579461bb55f94e0a841699cfe4cb544f3f1e5655061c138",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:66936363ad7cafe5d098b179f09f9aec1457415edfc316ece13ffef428744ae5"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/490",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:134e429091ea",
+      "dbms": "duckdb",
+      "title": "A predicate NOT(NULL OR TRUE) unexpectedly evaluates to TRUE",
+      "reported_date": "2020-04-07",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/489",
+        "fix1": "https://github.com/cwida/duckdb/commit/b838c589e26c47c342dd29b890438f5fb9dce4d4",
+        "fix2": "https://github.com/cwida/duckdb/commit/ef2286faef94a75d1ed02de0e86b804610da281e"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/489",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/489\",\n            \"fix1\": \"https://github.com/cwida/duckdb/commit/b838c589e26c47c342dd29b890438f5fb9dce4d4\",\n            \"fix2\": \"https://github.com/cwida/duckdb/commit/ef2286faef94a75d1ed02de0e86b804610da281e\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:c537a847f81d8af00bb48e2a6c399a019bb87f256cf653d8e35ae10b7b8f5e0c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/489",
+            "source_type": "github_issue",
+            "excerpt": "A predicate NOT(NULL OR TRUE) unexpectedly evaluates to TRUE",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:32aa29cf261e67be809149a09a0ecc87438b4f819ff6263225ff478bdf52426a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c537a847f81d8af00bb48e2a6c399a019bb87f256cf653d8e35ae10b7b8f5e0c"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/489",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:36afafdc40e6",
+      "dbms": "duckdb",
+      "title": "PRAGMA table_info provides no output",
+      "reported_date": "2020-04-07",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/491"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/491",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/491",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nPRAGMA table_info('t0'); -- expected: {0|c0|INTEGER|false|NULL|false}, actual: {}\r\n```\r\nUnexpectedly, the `PRAGMA` provides no output. According to the documentation, it should be equivalent to the following, which actually provides an output:",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by mrigger of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5a37032a0c63d30b59b4f3ec79c3699ea9298bc57e08b4d1a87b2aa26fbd8e9a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by mrigger.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/491",
+            "source_type": "github_issue",
+            "excerpt": "PRAGMA table_info provides no output",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5a37032a0c63d30b59b4f3ec79c3699ea9298bc57e08b4d1a87b2aa26fbd8e9a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5a37032a0c63d30b59b4f3ec79c3699ea9298bc57e08b4d1a87b2aa26fbd8e9a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:20c2ece233d5",
+      "dbms": "duckdb",
+      "title": "Comparison of two boolean columns in different tables results in an error \"Not implemented: Unimplemented type for sort\"",
+      "reported_date": "2020-04-08",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/497",
+        "fix": "https://github.com/cwida/duckdb/commit/a90e05ae7d4995ff094c3f0551c4564006f095f2"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/497",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"08/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/497\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/a90e05ae7d4995ff094c3f0551c4564006f095f2\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7446e0164918dcd652501aa34b96352296ede73380a3bb1c0534f98de467d973",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/497",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\r\nCREATE TABLE t1(c0 BOOL);\r\nINSERT INTO t1(c0) VALUES (0);\r\nINSERT INTO t0(c0) VALUES (0);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:587dd40dd97b1603b6ad15f6acbff8318bdc8c66c117ebf940fd659a2dfc84d7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7446e0164918dcd652501aa34b96352296ede73380a3bb1c0534f98de467d973"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/497",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:abf7ad3c5914",
+      "dbms": "duckdb",
+      "title": "Comparison on UNIQUE NUMERIC column causes a query to omit a row in the result set",
+      "reported_date": "2020-04-08",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/495",
+        "fix": "https://github.com/cwida/duckdb/commit/7156b358369203ab45fc55eac8605e475c7fdce4"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/495",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"08/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/495\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/7156b358369203ab45fc55eac8605e475c7fdce4\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:28a02555c763960ad00c4328b510a57d9c13eba6eeeda7d5f8de86206c6ed598",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/495",
+            "source_type": "github_issue",
+            "excerpt": "Comparison on UNIQUE NUMERIC column causes a query to omit a row in the result set",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:8da2abf4f474cbcd2beb95d77f8ce48df1a36ab881b39867e980d26a0f8881e9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:28a02555c763960ad00c4328b510a57d9c13eba6eeeda7d5f8de86206c6ed598"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/495",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:06eb99a32458",
+      "dbms": "duckdb",
+      "title": "INSERT causes an abort with \"terminate called after throwing an instance of 'duckdb::InvalidTypeException'\"",
+      "reported_date": "2020-04-08",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/496",
+        "fix": "https://github.com/cwida/duckdb/commit/f11d74a04e5326bde44058b0724ab33ac33547f8"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/496",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"08/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/496\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/f11d74a04e5326bde44058b0724ab33ac33547f8\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:0676c2ffeb5c1ab12d489be38a06f5eeeec3e18283f780fef53ecb3ca226d11f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/496",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN, c1 INT);\r\nCREATE INDEX i0 ON t0(c1, c0);\r\nINSERT INTO t0(c1) VALUES (0); -- terminate called after throwing an instance of 'duckdb::InvalidTypeException'\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:dc5c122f9437891ce7f905e9ee1f7b1cf1ebd9edd36589920a9c13bf81e575d9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0676c2ffeb5c1ab12d489be38a06f5eeeec3e18283f780fef53ecb3ca226d11f"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/496",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:9a5f7ce3208f",
+      "dbms": "duckdb",
+      "title": "SIMILAR TO results in an \"Unknown error -1\"",
+      "reported_date": "2020-04-08",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/493",
+        "fix": "https://github.com/cwida/duckdb/commit/403219728c302ad1eba1cf9e30934b85539a8945"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/493",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"08/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/493\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/403219728c302ad1eba1cf9e30934b85539a8945\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:9ddce4a8b6396cb510ea1ff4598c00519afa621fd61a80bde18f1f130265a4b2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/493",
+            "source_type": "github_issue",
+            "excerpt": "SIMILAR TO results in an \"Unknown error -1\"",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a24e3859d1f0455bfa0c63e6b6ce97938989c4f244cf07807fdee1941dde70a7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9ddce4a8b6396cb510ea1ff4598c00519afa621fd61a80bde18f1f130265a4b2"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/493",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:2eb9d25bc751",
+      "dbms": "duckdb",
+      "title": "SELECT causes JDBC driver to crash",
+      "reported_date": "2020-04-09",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/500",
+        "fix": "https://github.com/cwida/duckdb/commit/57c3f002e6017a11964c598b377f849cc7722438"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/500",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"09/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/500\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/57c3f002e6017a11964c598b377f849cc7722438\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:94e579f0675f895b22dfcc2d838bff2c353aa2fbca26223c7fce72702c2f02a1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/500",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR, c1 DOUBLE);\r\nCREATE TABLE t1(c0 DOUBLE, PRIMARY KEY(c0));\r\nINSERT INTO t0(c0) VALUES (0), (0), (0), (0);\r\nINSERT INTO t0(c0) VALUES (NULL), (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3b175bbbaa234523cd7b0afa83457dc4863cbac41a601b6d1f67bb8bdf56f172",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:94e579f0675f895b22dfcc2d838bff2c353aa2fbca26223c7fce72702c2f02a1"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/500",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:903dc8e5a71b",
+      "dbms": "duckdb",
+      "title": "A RIGHT JOIN unexpectedly fetches rows",
+      "reported_date": "2020-04-10",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/505",
+        "fix": "https://github.com/cwida/duckdb/commit/1014b562dcd8228e7bdb3f984323034d3f0ae3ca"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/505",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/505\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/1014b562dcd8228e7bdb3f984323034d3f0ae3ca\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:81e77ba06617e7ca7a01a8049ad70ffef643624a4e321d7c7e6e268d4c812959",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/505",
+            "source_type": "github_issue",
+            "excerpt": "A RIGHT JOIN unexpectedly fetches rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:997239d1905b543e0cc8f05acb9cca8dd96e28606c438c1fd73c0cc60ee8e941",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:81e77ba06617e7ca7a01a8049ad70ffef643624a4e321d7c7e6e268d4c812959"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/505",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:419929e37452",
+      "dbms": "duckdb",
+      "title": "Creating an empty table results in a crash",
+      "reported_date": "2020-04-10",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/507",
+        "fix": "https://github.com/cwida/duckdb/commit/a56a90108d8e0e012abbbd0b97bfc224555c6a11"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/507",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:49:31Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/507\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/a56a90108d8e0e012abbbd0b97bfc224555c6a11\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ca634998cd3a9f64ec31f518895be239e3a82b217f1133d0af882aa2a1a2bfa6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:49:31Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/507",
+            "source_type": "github_issue",
+            "excerpt": "Creating an empty table results in a crash",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7fd80f36daddf83a18f6489c68c56544cd074600b3a97df48c2a33314aa870c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ca634998cd3a9f64ec31f518895be239e3a82b217f1133d0af882aa2a1a2bfa6"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/507",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:e64bd304aad1",
+      "dbms": "duckdb",
+      "title": "INSERT results in an error \"Not implemented: Cannot create data from this type\"",
+      "reported_date": "2020-04-10",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/504",
+        "fix": "https://github.com/cwida/duckdb/commit/3a536f71d9ef1ee9feab8cb0870e425c0cdae9d6"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/504",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/504\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/3a536f71d9ef1ee9feab8cb0870e425c0cdae9d6\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:eaa4a6a5e65d2e49b0abe91c6c3bac8e1322f407284910e68810a666c672b531",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/504",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN, c1 INT, PRIMARY KEY(c0, c1));\r\nINSERT INTO t0(c1, c0) VALUES (0, 0); -- Error: Not implemented: Cannot create data from this type\r\n```\r\nUnexpectedly, the `INSERT` results in the error above. When omitting `c0` from the `PRIMARY KEY`, the `INSERT` works without errors:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:28571ec79c7a9680f48b356c1a40ac3147a386e37214c7eef7d427f642984137",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:eaa4a6a5e65d2e49b0abe91c6c3bac8e1322f407284910e68810a666c672b531"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/504",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5ffda6a31d86",
+      "dbms": "duckdb",
+      "title": "LEFT JOIN on column with NULL value results in a segmentation fault",
+      "reported_date": "2020-04-10",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/508",
+        "fix": "https://github.com/cwida/duckdb/commit/c40e4ca4b08c65213d74f6f4349d46cf5c25f430"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/508",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/508\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/c40e4ca4b08c65213d74f6f4349d46cf5c25f430\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f5b9e1da4dd88ba3dc5b75a8bd3ebeda28680a87546f9f0e68e274c24995dd09",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/508",
+            "source_type": "github_issue",
+            "excerpt": "LEFT JOIN on column with NULL value results in a segmentation fault",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2d4d69216fc8164c2123c4daab6e2412d94c525a611e1a59953642302bdd9b93",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f5b9e1da4dd88ba3dc5b75a8bd3ebeda28680a87546f9f0e68e274c24995dd09"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/508",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5a3e8a49fcd4",
+      "dbms": "duckdb",
+      "title": "Query results in an error \"INTERNAL: Failed to bind column reference \"c0\" [5.0] (bindings: [6.0])\"",
+      "reported_date": "2020-04-10",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/506",
+        "fix": "https://github.com/cwida/duckdb/commit/801101000f874aac259bb8f5af30983a03475fba"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/506",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/506\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/801101000f874aac259bb8f5af30983a03475fba\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:b370c6dde0fd7170ee354cc7115fee70a0ad7c0d205345ddbc3ad89cbaadd7ba",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/506",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT);\r\nSELECT * FROM t1 JOIN t0 ON t1.c0 < t1.c0 - t0.c0 WHERE t0.c0 <= t1.c0; -- Error: INTERNAL: Failed to bind column reference \"c0\" [5.0] (bindings: [6.0])\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7db120d0b3c544635b6e8d8f96a04106138f172d7bdf2f560c69e04e6bda5af9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b370c6dde0fd7170ee354cc7115fee70a0ad7c0d205345ddbc3ad89cbaadd7ba"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/506",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:35d02f31b4cd",
+      "dbms": "duckdb",
+      "title": "RIGHT JOIN with a predicate that compares two integer columns results in an \"Unhandled type\" error",
+      "reported_date": "2020-04-10",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/503",
+        "fix": "https://github.com/cwida/duckdb/commit/1014b562dcd8228e7bdb3f984323034d3f0ae3ca"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/503",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/503\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/1014b562dcd8228e7bdb3f984323034d3f0ae3ca\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:11d0bc8bc59127840594cbd732658195f1b85aee026aee497bd9452f62b1713a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/503",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT);\r\nSELECT * FROM t0 RIGHT JOIN t1 ON t0.c0!=t1.c0; -- Error: Unhandled type for empty NL join\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a8dcf1c5aa7e7b04d560e8c243b1ec78ed75e04d247a61591977f7869bdfb9d4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:11d0bc8bc59127840594cbd732658195f1b85aee026aee497bd9452f62b1713a"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/503",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:a14de3671424",
+      "dbms": "duckdb",
+      "title": "SIMILAR TO results in an incorrect result",
+      "reported_date": "2020-04-10",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/510",
+        "fix": "https://github.com/cwida/duckdb/commit/0ba277225e9748b805dd9bf0270da75ec8db286b"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/510",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/510\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/0ba277225e9748b805dd9bf0270da75ec8db286b\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:ffd3178acc5500b9b965313ba645f9a92855c623f552535a1389fd3c93193238",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/510",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0(c0) VALUES (-10);\r\nSELECT * FROM t0 WHERE t0.c0 NOT SIMILAR TO 0; -- expected: {-10}, actual: {}\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6effd3cb743eeb68170dcae7888cba30d14fa66a1838e68fa0b06246716c9b21",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ffd3178acc5500b9b965313ba645f9a92855c623f552535a1389fd3c93193238"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/510",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:ca0d1e1b5cfc",
+      "dbms": "duckdb",
+      "title": "Casting a large number to REAL and multiplying it with zero results in -nan",
+      "reported_date": "2020-04-11",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/522",
+        "fix": "https://github.com/cwida/duckdb/pull/541"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/522",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/522\",\n            \"fix\": \"https://github.com/cwida/duckdb/pull/541\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:155ea7835e0ee2265e5e2775e3b93b8b092558ba063fc0611632889170cfe6e7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/522",
+            "source_type": "github_issue",
+            "excerpt": "Casting a large number to REAL and multiplying it with zero results in -nan",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6c7f2866d754974fedc2683bb135d082cfa31510c70a7d1558b95cc11d8846fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:155ea7835e0ee2265e5e2775e3b93b8b092558ba063fc0611632889170cfe6e7"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/522",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:34a5531799e7",
+      "dbms": "duckdb",
+      "title": "Incorrect result after an INSERT violates a UNIQUE constraint",
+      "reported_date": "2020-04-11",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/514",
+        "fix": "https://github.com/cwida/duckdb/commit/0d29535fe6aefdf2fbe8a7de0775c253f8038a1a"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/514",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/514\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/0d29535fe6aefdf2fbe8a7de0775c253f8038a1a\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:0b73374ce3bcec9647701ac6301f9b179396b2dae776f69adf7d5980fc539387",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/514",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE UNIQUE INDEX i0 ON t0(c0);\r\nINSERT INTO t0(c0) VALUES (1);\r\nINSERT INTO t0(c0) VALUES (1); -- Failed to commit: Constraint: PRIMARY KEY or UNIQUE constraint violated: duplicated key",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b5b20ba38e24571c9368f77086d2f8e862aed61415a97da4d2e38326cbd230fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0b73374ce3bcec9647701ac6301f9b179396b2dae776f69adf7d5980fc539387"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/514",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:00f40fd0d151",
+      "dbms": "duckdb",
+      "title": "LEFT JOIN with comparison on integer columns results in \"Not implemented: Unimplemented type for nested loop join!\"",
+      "reported_date": "2020-04-11",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/513",
+        "fix": "https://github.com/cwida/duckdb/commit/215e40cd5cdb7eee68797df4982219c63f04e2a9"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/513",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/513\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/215e40cd5cdb7eee68797df4982219c63f04e2a9\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c8a09b176cbde1a87a9ecc22d3e368b0f8cb5c1667a82b00a9d7713abf04dfbc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/513",
+            "source_type": "github_issue",
+            "excerpt": "LEFT JOIN with comparison on integer columns results in \"Not implemented: Unimplemented type for nested loop join!\"",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2ad33dbdd1f9e37cf17d1ec7e0709b3a164ea8ac7ccd24cf99a5ef918641713a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c8a09b176cbde1a87a9ecc22d3e368b0f8cb5c1667a82b00a9d7713abf04dfbc"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/513",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:cc0e6ca8c87f",
+      "dbms": "duckdb",
+      "title": "Query using the LN() function does not terminate",
+      "reported_date": "2020-04-11",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/518",
+        "fix": "https://github.com/cwida/duckdb/commit/230785f2e6dceac410cfd894db214d21e356b127"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/518",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/518\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/230785f2e6dceac410cfd894db214d21e356b127\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:85dddaa956a63a7e46d3f1e27616444371888dd505d4e9770845fc0b3f0d6d9b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/518",
+            "source_type": "github_issue",
+            "excerpt": "Query using the LN() function does not terminate",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6e942d78d620f3937c1164d9c46d5e5634f5e4c98a11402d7a7c8058a4713569",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:85dddaa956a63a7e46d3f1e27616444371888dd505d4e9770845fc0b3f0d6d9b"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/518",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:3ea90f424c3f",
+      "dbms": "duckdb",
+      "title": "Query with a negative shift predicate yields an incorrect result",
+      "reported_date": "2020-04-11",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/515",
+        "fix": "https://github.com/cwida/duckdb/commit/7e1234d8f414f1e1de4e15a57939594036d74f37"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/515",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/515\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/7e1234d8f414f1e1de4e15a57939594036d74f37\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:67e16dda53097050edeb8f82896eae23b6233b0b4e8ccc3a9bbb5a8075d511f3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/515",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT8, c1 DOUBLE);\r\nINSERT INTO t1(c0) VALUES (0);\r\nINSERT INTO t1(c1, c0) VALUES (1, 1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1ef947b7698e5b8134989f568e34ce9d9000f972b269f8ab7a279264fbfbb52f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:67e16dda53097050edeb8f82896eae23b6233b0b4e8ccc3a9bbb5a8075d511f3"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/515",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:ace0a1be6e3c",
+      "dbms": "duckdb",
+      "title": "Query with an AND predicate, NOT and comparison yields an incorrect result",
+      "reported_date": "2020-04-11",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/517",
+        "fix": "https://github.com/cwida/duckdb/commit/d88e21d926e732096081ee8afda03d70310bee2d"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/517",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/517\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/d88e21d926e732096081ee8afda03d70310bee2d\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:c88ae4668154a41b4dd23c4643815043745446d8c0ce33285407a91f55c3e1d5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/517",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT);\r\nINSERT INTO t0(c0) VALUES (0);\r\nINSERT INTO t1(c0) VALUES (0);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1657aeef12777f71a6068c6cfe26b91b2e0adcc7be6f695940b51cf6dc0417e3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c88ae4668154a41b4dd23c4643815043745446d8c0ce33285407a91f55c3e1d5"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/517",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:95b4899b4243",
+      "dbms": "duckdb",
+      "title": "Query with comparison on boolean column results in \"Invalid type: Invalid Type [BOOL]: Invalid type for index\"",
+      "reported_date": "2020-04-11",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/516",
+        "fix": "https://github.com/cwida/duckdb/commit/331cce4a0c88ee1b477d5859de7211c455664349"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/516",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/516\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/331cce4a0c88ee1b477d5859de7211c455664349\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:d2e54fb37edcdbda5e4f9db311c28723f532c37be8518294309b1c1bccbe64a7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/516",
+            "source_type": "github_issue",
+            "excerpt": "Query with comparison on boolean column results in \"Invalid type: Invalid Type [BOOL]: Invalid type for index\"",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7ae09e0165d0b5e3252a71d5042ad66ab501611f947d7896aaab014ebea40554",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d2e54fb37edcdbda5e4f9db311c28723f532c37be8518294309b1c1bccbe64a7"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/516",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:06df04c89c67",
+      "dbms": "duckdb",
+      "title": "ROUND() evaluates to -nan",
+      "reported_date": "2020-04-11",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/521",
+        "fix": "https://github.com/cwida/duckdb/commit/e399d40eb76f2fbe97c9e139551171862ab7848d"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/521",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/521\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/e399d40eb76f2fbe97c9e139551171862ab7848d\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:5c9f549b4ee589dd84050c53015445e6398cc558abc810ea532edbbaa4c2891c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/521",
+            "source_type": "github_issue",
+            "excerpt": "ROUND() evaluates to -nan",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9ff124f2aa3a39cf99e5a4a9f6f0743fb6365dc71a661b2de9733a2bfd727a5d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5c9f549b4ee589dd84050c53015445e6398cc558abc810ea532edbbaa4c2891c"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/521",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c5a2568f4c94",
+      "dbms": "duckdb",
+      "title": "The trigonometric functions can result in -nan",
+      "reported_date": "2020-04-11",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/523",
+        "fix": "https://github.com/cwida/duckdb/pull/541"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/523",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/523\",\n            \"fix\": \"https://github.com/cwida/duckdb/pull/541\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:e0c72945bbda4b900c0c934881126c27c715f53a78e2d4b95c1c4a51dc44cc05",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/523",
+            "source_type": "github_issue",
+            "excerpt": "The trigonometric functions can result in -nan",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:bf7a739001057109bf45aef80e03ed7b45036b523c5915f52504714d4cab74e0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e0c72945bbda4b900c0c934881126c27c715f53a78e2d4b95c1c4a51dc44cc05"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/523",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f07db8e2cb3f",
+      "dbms": "duckdb",
+      "title": "Fetching a TIMESTAMP column results in an error",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/532"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/532",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/532",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DATETIME);\r\nINSERT INTO t0 VALUES(DATE '1-1-1');\r\nSELECT t0.c0 FROM t0; -- Not implemented type: TIMESTAMP\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8eaa93b23dc0684b65e0a99ba1cd669600730adce8ce62b088afb1f8f15e01ec",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/532",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8eaa93b23dc0684b65e0a99ba1cd669600730adce8ce62b088afb1f8f15e01ec"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:898d645a183e",
+      "dbms": "duckdb",
+      "title": "Fetching from table and view results in a crash",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/537",
+        "fix": "https://github.com/cwida/duckdb/commit/504eb1a3e14c5db4ef46fa8c7dff09702e32d9d4"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/537",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/537\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/504eb1a3e14c5db4ef46fa8c7dff09702e32d9d4\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:42ad3db657beb7ede6183e33aacf7cad0440f0ead617e171c5451b9034a1b2a1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/537",
+            "source_type": "github_issue",
+            "excerpt": "The original test case crashed SQLancer. I reduced the crash based on the debug build, where the current test case causes an ASan error:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6f80f1dd7a97c5864e79527e23cbeb29f89d99448cfd02287d9903024ce5a64c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:42ad3db657beb7ede6183e33aacf7cad0440f0ead617e171c5451b9034a1b2a1"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/537",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b8a1540c8182",
+      "dbms": "duckdb",
+      "title": "GROUP BY clause results in non-deterministic result",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/535",
+        "fix": "https://github.com/cwida/duckdb/commit/03eafe5d77900714f56f0c6e3d6fb3f0204926cf"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/535",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/535\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/03eafe5d77900714f56f0c6e3d6fb3f0204926cf\"\n        },\n        \"oracle\": \"TLP (HAVING)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:489bbbe816a2561d51fd06c1d330a8b7c38374a8a64b1dd685a980a82307553e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/535",
+            "source_type": "github_issue",
+            "excerpt": "GROUP BY clause results in non-deterministic result",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e86081d2af80d41f7b058bbef33fd623372ae8e0a64624708dec6b3f2e27ecae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:489bbbe816a2561d51fd06c1d330a8b7c38374a8a64b1dd685a980a82307553e"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/535",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:4a14b6bbe78b",
+      "dbms": "duckdb",
+      "title": "Incorrect result for a JOIN predicate t1.c0 IN (t0.c0) and WHERE predicate t1.c0<=t0.c0",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/525",
+        "fix": "https://github.com/cwida/duckdb/commit/0e5a3c47eb28c8b6b9797c0c7f64e8ed5c322cb6"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/525",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/525\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/0e5a3c47eb28c8b6b9797c0c7f64e8ed5c322cb6\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:8980da9aa8ad1a936c79e9d092c9eacf553620f65fc38b55dcf7464f5d75a5fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/525",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 FLOAT);\r\nINSERT INTO t0(c0) VALUES (1), (0);\r\nINSERT INTO t1(c0) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:054bedc0e6b28008f0041723e1589e84e6031f54b2c361883cb05452e996ffbd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8980da9aa8ad1a936c79e9d092c9eacf553620f65fc38b55dcf7464f5d75a5fc"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/525",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:6208c9447d0c",
+      "dbms": "duckdb",
+      "title": "Incorrect result for predicate with shift on a BIGINT column",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/538",
+        "fix": "https://github.com/cwida/duckdb/commit/faf41956b6363882ef217b81e0c0db9fb7f39970"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/538",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/538\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/faf41956b6363882ef217b81e0c0db9fb7f39970\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:8839de7d0d40c94cdc0337ce42ec0b340300a7d9fbc91010db3965066c05190d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/538",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BIGINT);\r\nINSERT INTO t0(c0) VALUES (-1);\r\nINSERT INTO t0(c0) VALUES (0);\r\nSELECT * FROM t0 WHERE t0.c0 AND (t0.c0<<64);-- expected: {}, actual: {-1}",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:60502215aea88d5f381cab9837d1b19b33e32ec91f281a3104a5cdc47e0e8929",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8839de7d0d40c94cdc0337ce42ec0b340300a7d9fbc91010db3965066c05190d"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/538",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c33ee9b10954",
+      "dbms": "duckdb",
+      "title": "Incorrect result for query that uses REGEXP_MATCHES()",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/528",
+        "fix": "https://github.com/cwida/duckdb/commit/1035a091f6feba4cc1c37bea5dda3d3ccfcacd17"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/528",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/528\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/1035a091f6feba4cc1c37bea5dda3d3ccfcacd17\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:fc35f56a6c2c12543faad64fdb63fd769529e19b2c9b712d26606dc84611c8c8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/528",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nINSERT INTO t0(c0) VALUES (0.1);\r\nSELECT * FROM t0 WHERE REGEXP_MATCHES(t0.c0, '1'); -- expected: {0.1}, actual: {}\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6f0314cc94227fe05607268e6a2e15b856b251bbbcc3c4a683362ffdedeaa051",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:fc35f56a6c2c12543faad64fdb63fd769529e19b2c9b712d26606dc84611c8c8"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/528",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f3610313041f",
+      "dbms": "duckdb",
+      "title": "Nondeterministic clause when using an UNION query and HAVING clause",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/536",
+        "fix": "https://github.com/cwida/duckdb/commit/7548d0244bd24cd5da20ec1fdb8f722b39367016"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/536",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/536\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/7548d0244bd24cd5da20ec1fdb8f722b39367016\"\n        },\n        \"oracle\": \"TLP (HAVING)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:2cd4fcf352d9d3480292fb71f1c00d7ee7fe51876b355c7810496e1fe251cced",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/536",
+            "source_type": "github_issue",
+            "excerpt": "Nondeterministic clause when using an UNION query and HAVING clause",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3f1ef94778931df53a37b02a9f8617f31684231601498d8043b747960687659f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2cd4fcf352d9d3480292fb71f1c00d7ee7fe51876b355c7810496e1fe251cced"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/536",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:90dd57d39e13",
+      "dbms": "duckdb",
+      "title": "Overflow when casting from REAL to INT results in \"Invalid TypeId -1\"",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/533",
+        "fix": "https://github.com/cwida/duckdb/commit/7905e332b02cd069c238c0ea878afdf4cf8c0419"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/533",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/533\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/7905e332b02cd069c238c0ea878afdf4cf8c0419\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:4347b3b45a0e0b96352e3b062da781f07f09df237805ce7095a2dee9db906972",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/533",
+            "source_type": "github_issue",
+            "excerpt": "Overflow when casting from REAL to INT results in \"Invalid TypeId -1\"",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d81c77e8d44dc644698dcebec4ba01ba6226d19746b2a0e37facffe616f268a9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4347b3b45a0e0b96352e3b062da781f07f09df237805ce7095a2dee9db906972"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/533",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:6dc81fbecbc7",
+      "dbms": "duckdb",
+      "title": "Query that uses the CONCAT() function and OR expression crashes",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/526",
+        "fix": "https://github.com/cwida/duckdb/commit/a5a88b41f8346a582d6cecb80a12e7287d300bf1"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/526",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/526\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/a5a88b41f8346a582d6cecb80a12e7287d300bf1\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c9fcfeb02f535a9e206beceb9a17d3070de22153f90a3f3a21d5d6a3fe9e95d6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/526",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 REAL);\r\nCREATE TABLE t1(c0 INT2);\r\nCREATE TABLE t2(c0 INT);\r\nINSERT INTO t0 VALUES (-1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d4130e31cb1092b7aff02b4a429cf8ee16f595aefb4e4b53147e0f25346819de",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c9fcfeb02f535a9e206beceb9a17d3070de22153f90a3f3a21d5d6a3fe9e95d6"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/526",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c40793f7660b",
+      "dbms": "duckdb",
+      "title": "Query with JOIN and WHERE condition unexpectedly fetches a value not present in the table",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/527"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/527",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/527\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:9e1a0cd556ba024593f07e5945f4921f0cf6503b586744b192397a11926e29cf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/527",
+            "source_type": "github_issue",
+            "excerpt": "Query with JOIN and WHERE condition unexpectedly fetches a value not present in the table",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3be706738f8f31c08e5c63184473e4f698145efc632c7926fedbe2342f9e9882",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9e1a0cd556ba024593f07e5945f4921f0cf6503b586744b192397a11926e29cf"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/527",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:52c4e3b6d846",
+      "dbms": "duckdb",
+      "title": "SELECT on DATE column with a large negative value results in a \"double free or corruption\"",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/531",
+        "fix": "https://github.com/cwida/duckdb/commit/409b26b96fd9daeb9117ec620f9046f6f657b8eb"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/531",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/531\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/409b26b96fd9daeb9117ec620f9046f6f657b8eb\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:90e2cbff89e2a1b76e5e0c0fe569ca2c82e6c9cfa87b660e6860766b6f2f34eb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/531",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DATE);\r\nINSERT INTO t0 VALUES (-10000000);\r\nSELECT c0 FROM t0; -- double free or corruption (out)\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2c7dcf12fd69accf6d38aa7d0df01d10ac6fd5919ca854c94a1380b7ae442e1e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:90e2cbff89e2a1b76e5e0c0fe569ca2c82e6c9cfa87b660e6860766b6f2f34eb"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/531",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f9681afb5010",
+      "dbms": "duckdb",
+      "title": "UPDATE causes subsequent query to segfault",
+      "reported_date": "2020-04-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/534",
+        "fix": "https://github.com/cwida/duckdb/commit/1866e91805ddcc6ac0e0236d44209325334524e4"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/534",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/534\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/1866e91805ddcc6ac0e0236d44209325334524e4\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7928b0d8b7965453f7b768dcc3c78742f3a2fb95aebfa1793a4c660f7243b0bf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/534",
+            "source_type": "github_issue",
+            "excerpt": "UPDATE causes subsequent query to segfault",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5185553b0b0bd6acc0f91b354db2eef9046dac6a1f30beed2afa94c85746855e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7928b0d8b7965453f7b768dcc3c78742f3a2fb95aebfa1793a4c660f7243b0bf"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/534",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:a13d80b0b668",
+      "dbms": "duckdb",
+      "title": "Assertions in the JDBC build of DuckDB?",
+      "reported_date": "2020-04-13",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/545"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/545",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/545",
+            "source_type": "github_issue",
+            "excerpt": "Assertions in the JDBC build of DuckDB?",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fe1838123146815fabf7f0b7d74d8274d6fec922cc9054be975d3ef483257d68",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/545",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fe1838123146815fabf7f0b7d74d8274d6fec922cc9054be975d3ef483257d68"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c81ddf3a76fc",
+      "dbms": "duckdb",
+      "title": "Nested CASE expression results in Assertion `other.auxiliary->type == VectorBufferType::STRING_BUFFER' failed",
+      "reported_date": "2020-04-13",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/549",
+        "fix": "https://github.com/cwida/duckdb/pull/558/commits/13c37227d777be60e5653830d12c894944bcd549"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/549",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"13/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/549\",\n            \"fix\": \"https://github.com/cwida/duckdb/pull/558/commits/13c37227d777be60e5653830d12c894944bcd549\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:fa29aea5e46b7452ee39ff8506d08cb4e7375d4ea5c5307d94c5fa460dab62b5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/549",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0 VALUES (NULL), (0), (1);\r\nSELECT * FROM t0 WHERE CASE WHEN c0 THEN 0 ELSE CASE '0.1' WHEN c0 THEN '' END END; -- Assertion `other.auxiliary->type == VectorBufferType::STRING_BUFFER' failed.\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:cfc59f0eed61ac0daf5d14c48415c911f27aa0ac0bcf735816dd27f097d4aa49",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:fa29aea5e46b7452ee39ff8506d08cb4e7375d4ea5c5307d94c5fa460dab62b5"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/549",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c8feb0e23b81",
+      "dbms": "duckdb",
+      "title": "Nested MAX() results in nondeterministic result or double free",
+      "reported_date": "2020-04-13",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/540",
+        "fix": "https://github.com/cwida/duckdb/commit/7945aa5c9e591b715cf16d8ef0dc4c9612f08219"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/540",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"13/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/540\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/7945aa5c9e591b715cf16d8ef0dc4c9612f08219\"\n        },\n        \"oracle\": \"TLP (aggregate)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:ee2ceb72d4bc69790de2e657c3c68ff233b9f128fd9d05b6e7365f8c6323c49e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/540",
+            "source_type": "github_issue",
+            "excerpt": "Nested MAX() results in nondeterministic result or double free",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ff44c941ff3a259a182838aa4d0a76d98c5ab7d0705e55c3dd4fab6c65c2fdd0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ee2ceb72d4bc69790de2e657c3c68ff233b9f128fd9d05b6e7365f8c6323c49e"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/540",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:264c71df8be7",
+      "dbms": "duckdb",
+      "title": "Query with SIMILAR TO results in \"Assertion `strlen(dataptr) == length' failed\"",
+      "reported_date": "2020-04-13",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/547",
+        "fix": "https://github.com/cwida/duckdb/commit/09565d0bf0df4968defedc23dc00dacbd5afbd8e"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/547",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"13/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/547\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/09565d0bf0df4968defedc23dc00dacbd5afbd8e\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:2c5f121430dcb5cecd60e96cfcbc1b26fae52cf53c26c7ebb31bb51b344b4e14",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/547",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0 VALUES (0);\r\nSELECT * FROM t0 WHERE t0.c0 SIMILAR TO '.'; -- Assertion `strlen(dataptr) == length' failed\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:23f128ec6d6fb325a36b024dc170743b23b673ef1ebafd2611fd4800bef09b17",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2c5f121430dcb5cecd60e96cfcbc1b26fae52cf53c26c7ebb31bb51b344b4e14"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/547",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:adae59a2068f",
+      "dbms": "duckdb",
+      "title": "SELECT on view with text constant in ORDER BY crashes",
+      "reported_date": "2020-04-13",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/544",
+        "fix": "https://github.com/cwida/duckdb/commit/72491b0046cd45cd0702fc60d5f42d4234f82d54"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/544",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"13/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/544\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/72491b0046cd45cd0702fc60d5f42d4234f82d54\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5cffb568c2a368b3d70e0b60169a1467b9b9e92060867b35be29a12d5641d25c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/544",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0(c0) VALUES (0);\r\nCREATE VIEW v0(c0) AS SELECT 1 FROM t0;\r\nSELECT * FROM v0 ORDER BY 'a'; --  Assertion `types.size() > 0' failed (original test case crashed)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7c80b10bdb3cb5014903f2e2c21cc1bf1ecea85477be597207fc9639c31bafb0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5cffb568c2a368b3d70e0b60169a1467b9b9e92060867b35be29a12d5641d25c"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/544",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:223a4667cb7b",
+      "dbms": "duckdb",
+      "title": "Unexpected result for SUM() upon overflow",
+      "reported_date": "2020-04-13",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/543",
+        "fix": "https://github.com/cwida/duckdb/pull/558/commits/fda8813bb208886a310d927f35566a2f74862aeb"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/543",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"13/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/543\",\n            \"fix\": \"https://github.com/cwida/duckdb/pull/558/commits/fda8813bb208886a310d927f35566a2f74862aeb\"\n        },\n        \"oracle\": \"TLP (aggregate)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:b3d4b907e034ebdf964008c0275c045dc015e2eca77f8c1ebc74470def4ed5bd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/543",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 BIGINT);\r\nINSERT INTO t0(c1) VALUES (2);\r\nINSERT INTO t0(c1) VALUES (9223372036854775807);\r\nSELECT SUM(t0.c1) FROM t0; -- expected: {9223372036854776000}, actual: {-9223372036854775807}",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:908fb9ebce45593f046f88649488460cf1714d21e9432bd7ca4427852f6c56f3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b3d4b907e034ebdf964008c0275c045dc015e2eca77f8c1ebc74470def4ed5bd"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/543",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b6d050203807",
+      "dbms": "duckdb",
+      "title": "[JDBC] Special character causes the JVM to terminate with \"basic_string::_M_construct null not valid\"",
+      "reported_date": "2020-04-13",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/539",
+        "fix": "https://github.com/cwida/duckdb/pull/550/commits/7d5ea9f7eb3b249b57f3e42afca6eeac885ae151"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/539",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"13/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/539\",\n            \"fix\": \"https://github.com/cwida/duckdb/pull/550/commits/7d5ea9f7eb3b249b57f3e42afca6eeac885ae151\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f529ff05ac1c7a33ac44443b06c19ca35b5fb62469e7231ebe200bbfba822aa9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/539",
+            "source_type": "github_issue",
+            "excerpt": "[JDBC] Special character causes the JVM to terminate with \"basic_string::_M_construct null not valid\"",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:97cda4f7dc3411f9f9029ffebf57cdc57f8d8916a9c7d4c34cc1bc36a9d1e5ef",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f529ff05ac1c7a33ac44443b06c19ca35b5fb62469e7231ebe200bbfba822aa9"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/539",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:039147c666e8",
+      "dbms": "duckdb",
+      "title": "RIGHT JOIN results in Assertion `filter->expressions.size() == 1' failed",
+      "reported_date": "2020-04-14",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/552"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/552",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/552\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:9dec14a29082d452c7892511f1d3d8743c0ed0357b70337eab5e6662d98f0782",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/552",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT);\r\nSELECT * FROM t0 RIGHT JOIN t1 ON 0 WHERE t0.c0 OR t1.c0 BETWEEN t0.c0 AND 1; -- Assertion `filter->expressions.size() == 1' failed.\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:afb5b9becdab94343147d48e022f1421f7e54ea4f36a6699407a77782fefe84e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9dec14a29082d452c7892511f1d3d8743c0ed0357b70337eab5e6662d98f0782"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/552",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b902f8d5804a",
+      "dbms": "duckdb",
+      "title": "Incorrect result for SUM() and negative number",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/560",
+        "fix": "https://github.com/cwida/duckdb/commit/db0e6131397d4a09cd2c997fb86a9974b6c356e3"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/560",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/560\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/db0e6131397d4a09cd2c997fb86a9974b6c356e3\"\n        },\n        \"oracle\": \"TLP (aggregate)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:d3ac49d5b7a340080948d108ad78db6150bc4a37f0a3e473824334487ecc3d30",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/560",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nINSERT INTO t0 VALUES (0);\r\nSELECT SUM(-1) FROM t0; -- expected: {-1}, actual: {1.8446744073709552e+19}\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:dda8077a26454c333f04e3c77a41cc56199129692e17a439b4c8d1fa4f178734",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d3ac49d5b7a340080948d108ad78db6150bc4a37f0a3e473824334487ecc3d30"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/560",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:2dbbd9dde136",
+      "dbms": "duckdb",
+      "title": "SELECT with CASE expression causes an assertion failure \"Assertion `!entry.first->Equals(&expr)' failed\"",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/562",
+        "fix": "https://github.com/cwida/duckdb/commit/ae863586dc73d4113ad508a04a8b2aa892038eec"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/562",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/562\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/ae863586dc73d4113ad508a04a8b2aa892038eec\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:9d25d8230791bee758228cfdff1683ed374d93b52e61392fbe64224d4ba246b0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/562",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nSELECT * FROM t0 GROUP BY -4.40304405E8 ORDER BY (CASE 1 WHEN 0 THEN 0 ELSE -440304405 END); -- Assertion `!entry.first->Equals(&expr)' failed.\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e49a574d38cfb9593126f730307958593b9e7a1e534f5ea23e60602498de8a2e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9d25d8230791bee758228cfdff1683ed374d93b52e61392fbe64224d4ba246b0"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/562",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:65a5bb644f3b",
+      "dbms": "duckdb",
+      "title": "REVERSE() on special character results in \"Assertion `strcmp(dataptr, normalized) == 0' failed.\"",
+      "reported_date": "2020-04-16",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/567",
+        "fix": "https://github.com/cwida/duckdb/commit/2927f7835bdee40695b8cf946890cca86c42623a"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/567",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"16/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/567\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/2927f7835bdee40695b8cf946890cca86c42623a\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:4a3fdd216d6f1bcadaadaa44e9946e18769621cef112e879eede81fdcf1821be",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/567",
+            "source_type": "github_issue",
+            "excerpt": "REVERSE() on special character results in \"Assertion `strcmp(dataptr, normalized) == 0' failed.\"",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:44b46f2fa4401bdb31837d1dc9e4cd7bf0ccce6d30e4af53d7fcf3780c873e92",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4a3fdd216d6f1bcadaadaa44e9946e18769621cef112e879eede81fdcf1821be"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/567",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:75424118fd65",
+      "dbms": "duckdb",
+      "title": "Incorrect result for BETWEEN query that casts column to boolean",
+      "reported_date": "2020-04-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/571",
+        "fix": "https://github.com/cwida/duckdb/commit/e913dacbdfc00fa6686a5c470e3deeb1a5893ee0"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/571",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/571\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/e913dacbdfc00fa6686a5c470e3deeb1a5893ee0\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:71e8e87b73e70d7e3c4fea042dcafe9259454f4f23120c65e78e5b24b954ba9d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/571",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0 VALUES (-1);\r\nSELECT t0.c0 FROM t0 WHERE NOT (0 BETWEEN 0 AND t0.c0::BOOL); -- expected: {}, actual: {-1}\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d4d7db2215eac100bc92e57502675cf81cea45ef27c580b320710615dec0b5dd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:71e8e87b73e70d7e3c4fea042dcafe9259454f4f23120c65e78e5b24b954ba9d"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/571",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b37db909a2b8",
+      "dbms": "duckdb",
+      "title": "Query using PREFIX() results in an error std::bad_alloc",
+      "reported_date": "2020-04-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/572",
+        "fix": "https://github.com/cwida/duckdb/pull/573"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/572",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/572\",\n            \"fix\": \"https://github.com/cwida/duckdb/pull/573\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1148495510697e0b30e64d96858430ff4a494ed97295d6d247f5fc371091f2aa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/572",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nSELECT * FROM t0 WHERE PREFIX(t0.c0, ''); -- Error: std::bad_alloc\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5899f08d413e60dc7f6509ae4ba693363b7a13526fef735058918e507d24f264",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1148495510697e0b30e64d96858430ff4a494ed97295d6d247f5fc371091f2aa"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/572",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f1762b423466",
+      "dbms": "duckdb",
+      "title": "DISTINCT malfunctions for BOOLEAN",
+      "reported_date": "2020-04-24",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/581",
+        "fix": "https://github.com/cwida/duckdb/commit/0cb8b09273b37ee9ab0ccfaa677926a0e970c51b"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/581",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/581\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/0cb8b09273b37ee9ab0ccfaa677926a0e970c51b\"\n        },\n        \"oracle\": \"TLP (DISTINCT)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:6f7af325d1442738685178c4892fe583fefe40d2bb49e148ae93266e325e057e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/581",
+            "source_type": "github_issue",
+            "excerpt": "DISTINCT malfunctions for BOOLEAN",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6632231e3b15e774999059f0529376893fe3259e5ae6327fa2c14b82eb8a54bf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6f7af325d1442738685178c4892fe583fefe40d2bb49e148ae93266e325e057e"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/581",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:383ad4788ec6",
+      "dbms": "duckdb",
+      "title": "SUBSTRING with an invalid start position causes a segmentation fault",
+      "reported_date": "2020-04-24",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/580",
+        "fix": "https://github.com/cwida/duckdb/commit/184f51539e3df106c148f5816dccfb1927e85c5d"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/580",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/580\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/184f51539e3df106c148f5816dccfb1927e85c5d\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:39b72921fe82ef0c8fd822fc1f1d632f8791f949a0024809b605f511ff4ffb30",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/580",
+            "source_type": "github_issue",
+            "excerpt": "SUBSTRING with an invalid start position causes a segmentation fault",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2ceaf4db6a06f851add87f5d82c8bfe02243c633cb5a590557bf00bc54afead0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:39b72921fe82ef0c8fd822fc1f1d632f8791f949a0024809b605f511ff4ffb30"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/580",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8b24673f262d",
+      "dbms": "duckdb",
+      "title": "Updated value in column is not visible in a SELECT",
+      "reported_date": "2020-04-24",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/579",
+        "fix": "https://github.com/cwida/duckdb/commit/f77519c6fb26fc1df1f503f182525b7af68c59fb"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/579",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/579\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/f77519c6fb26fc1df1f503f182525b7af68c59fb\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:3c3b0fa0955eb0d059ebd815d07cc09fc533ca8214884e74b5ba37a68be312c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/579",
+            "source_type": "github_issue",
+            "excerpt": "Updated value in column is not visible in a SELECT",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:395b607bc303af6c5b50b1052fab6be330999517a90de786ec5df5d7c38ee0f5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3c3b0fa0955eb0d059ebd815d07cc09fc533ca8214884e74b5ba37a68be312c2"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/579",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:6f02f577c1e2",
+      "dbms": "duckdb",
+      "title": "A negative DATE results in a \"double free or corruption\" crash",
+      "reported_date": "2020-04-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/587",
+        "fix": "https://github.com/cwida/duckdb/commit/a750473a7e4979a5237471268a752e46dee36e24"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/587",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/587\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/a750473a7e4979a5237471268a752e46dee36e24\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:9a95c96e57c327a881d8ee14ae3b3ea75a9776079d101d1a39fee4df8c70edd6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/587",
+            "source_type": "github_issue",
+            "excerpt": "A negative DATE results in a \"double free or corruption\" crash",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e35f4b6c3788f641bbefac0e6bd8fd7da8e852277da2883d48e7b7c5b9512298",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9a95c96e57c327a881d8ee14ae3b3ea75a9776079d101d1a39fee4df8c70edd6"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/587",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:00cffb6962bf",
+      "dbms": "duckdb",
+      "title": "A select with BETWEEN and VARCHAR cast results in an incorrect result",
+      "reported_date": "2020-04-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/584",
+        "fix": "https://github.com/cwida/duckdb/commit/e8ec15172fab1dd5c0547f4259ed3dc8ce31f8e9"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/584",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/584\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/e8ec15172fab1dd5c0547f4259ed3dc8ce31f8e9\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:5c9145766797dc33e46958a0e75d137c5f7ae66fadd96cfe987d4699f60485dd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/584",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\nINSERT INTO t0(c0) VALUES (-2);\r\nSELECT t0.c0 FROM t0 WHERE -1 BETWEEN t0.c0::VARCHAR AND 1; -- expected: {-2}, actual: {}\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:294bfab6031d957bf1fc14bc0edaa4fb700bbf06a813ae7650dccfbab5badf88",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5c9145766797dc33e46958a0e75d137c5f7ae66fadd96cfe987d4699f60485dd"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/584",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b5b9f8a6326f",
+      "dbms": "duckdb",
+      "title": "Comparing a string with a boolean yields an incorrect result after UPDATE",
+      "reported_date": "2020-04-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/583",
+        "fix": "https://github.com/cwida/duckdb/commit/bc96c1ef70867fc402906739b5133c1674f054cf"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/583",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/583\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/bc96c1ef70867fc402906739b5133c1674f054cf\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:762b78357e791323aec570758b727057d08185403e57c590e034942beac2cd5e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/583",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nINSERT INTO t0(c0) VALUES (0);\r\nUPDATE t0 SET c0=0;\r\nUPDATE t0 SET c0=true;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:922d9477bf9c16f5acc8c8462ce778beeb8cf271eeecc9d35478376a74b3b6e4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:762b78357e791323aec570758b727057d08185403e57c590e034942beac2cd5e"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/583",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:acbef89a3058",
+      "dbms": "duckdb",
+      "title": "Comparison with a DATE yields an incorrect result",
+      "reported_date": "2020-04-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/590",
+        "fix": "https://github.com/cwida/duckdb/commit/63f420535e327a874a46b77dc75c4518c606439b"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/590",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/590\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/63f420535e327a874a46b77dc75c4518c606439b\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:8034f73b0927e94bfdb18653d64fb001c1619b6def2f34bd6146b7cc19d6961c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/590",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nINSERT INTO t0(c0) VALUES (DATE '2000-01-02');\r\nSELECT * FROM t0 WHERE DATE '2000-01-01' < t0.c0; -- expected: {2000-01-01}, actual: {}\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0807ff4d625b0e857a7802ef03ca3832b04a9d81ff50a7b4af1f012b1c9bae79",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8034f73b0927e94bfdb18653d64fb001c1619b6def2f34bd6146b7cc19d6961c"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/590",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:e7f9c4711bab",
+      "dbms": "duckdb",
+      "title": "Creating an index on rowid results in an internal error \"Failed to bind column reference\"",
+      "reported_date": "2020-04-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/589",
+        "fix": "https://github.com/cwida/duckdb/commit/0bfdbf39638c8b623056698850389349ec463ccc"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/589",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/589\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/0bfdbf39638c8b623056698850389349ec463ccc\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e3b1c8e448d4f88d46992a7c6fdda95ba8d2c0d968b4769c520e43b90beb1727",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/589",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE INDEX i0 ON t0(rowid, c0); -- INTERNAL: Failed to bind column reference \"c0\" [0.1] (bindings: [0.0])\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:94a22f206e56e4d7de71a9f9ad47fe0915677abc1c2711f73b997b909ff87248",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e3b1c8e448d4f88d46992a7c6fdda95ba8d2c0d968b4769c520e43b90beb1727"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/589",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:48744f141665",
+      "dbms": "duckdb",
+      "title": "NOACCENT.NOCASE comparison with a special character results in a segmentation fault",
+      "reported_date": "2020-04-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/586",
+        "fix": "https://github.com/cwida/duckdb/commit/77f3151ecfa85dfa894f3a23feb2271a4cac1ddc"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/586",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/586\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/77f3151ecfa85dfa894f3a23feb2271a4cac1ddc\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:79100461c207d287992534ec3bb701a3deefbbb94929210d10ccab35b5ab6db2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/586",
+            "source_type": "github_issue",
+            "excerpt": "NOACCENT.NOCASE comparison with a special character results in a segmentation fault",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:91535daaa4acbcc17840cf0cf687c9bcda56dbdc8a15fe95b45afcc5f84fec30",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:79100461c207d287992534ec3bb701a3deefbbb94929210d10ccab35b5ab6db2"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/586",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:918549ed67f4",
+      "dbms": "duckdb",
+      "title": "Predicate checking for an empty string yields an incorrect result",
+      "reported_date": "2020-04-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/585",
+        "fix": "https://github.com/cwida/duckdb/commit/dc352a3729f57252a1f6bfa18077fed9c4598a54"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/585",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/585\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/dc352a3729f57252a1f6bfa18077fed9c4598a54\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:9ecfcd3105a9cf92e8bdac6d9773ef1f46f2d0fb7278b07c8450547a890203d6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/585",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nINSERT INTO t0(c0) VALUES (''), (0);\r\nSELECT * FROM t0 WHERE t0.c0 = ''; -- expected: {''}, actual: {}\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3b65f74332eb37a603c389a9c28c2cbbcfb3bbfd95d1d2c7191c6d356c0d6629",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9ecfcd3105a9cf92e8bdac6d9773ef1f46f2d0fb7278b07c8450547a890203d6"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/585",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:04f44f3d0b40",
+      "dbms": "duckdb",
+      "title": "Query with complex ORDER BY causes an incorrect rowid value",
+      "reported_date": "2020-04-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/588",
+        "fix": "https://github.com/cwida/duckdb/commit/8a7cf974f7372a66e837b9cfa5d5c3929e1f625f"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/588",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/588\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/8a7cf974f7372a66e837b9cfa5d5c3929e1f625f\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:04943f6c94628caa73c46980715b368322a3765544da8191fa562ce2b30c00f7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/588",
+            "source_type": "github_issue",
+            "excerpt": "Query with complex ORDER BY causes an incorrect rowid value",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a44ebb4c2ea43ba87adc7f7484eea546309e1cde6c7184b7b264d6496550e3d4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:04943f6c94628caa73c46980715b368322a3765544da8191fa562ce2b30c00f7"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/588",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:063c4715cc77",
+      "dbms": "duckdb",
+      "title": "Subtracting a large integer from a DATE results in a \"double free or corruption\"",
+      "reported_date": "2020-04-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/591"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/591",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/591\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:869924bce5c6a68bf8d13df9f60a7582044e681d6603b226485a2aa98621c3b1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/591",
+            "source_type": "github_issue",
+            "excerpt": "Subtracting a large integer from a DATE results in a \"double free or corruption\"",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9e65f0778c2f66e23345a1a902beb681411c9e041738b3736e5fadeafea68842",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:869924bce5c6a68bf8d13df9f60a7582044e681d6603b226485a2aa98621c3b1"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/591",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8b21082755b9",
+      "dbms": "duckdb",
+      "title": "Expression with LIKE and comparison causes an assertion failure",
+      "reported_date": "2020-04-26",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/592",
+        "fix": "https://github.com/cwida/duckdb/commit/96d5ae2e3a4b3aa52eab3878b78c5bdb8f6946b9"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/592",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/592\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/96d5ae2e3a4b3aa52eab3878b78c5bdb8f6946b9\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:55d64c2282d86d600d3c95d8facc893bcdf6e2994715ad83982ef79d21bfe2c5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/592",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nINSERT INTO t0 VALUES (0);\r\nSELECT * FROM t0 WHERE c0 LIKE '' AND c0 < true; -- Assertion `tableFilter[0].comparison_type == ExpressionType::COMPARE_GREATERTHAN || tableFilter[0].comparison_type == ExpressionType::COMPARE_GREATERTHANOREQUALTO' failed.\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bed998d4aa74c4edcd8efffae27112d1eb6e2780cd915423918404bdf3a33a2d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:55d64c2282d86d600d3c95d8facc893bcdf6e2994715ad83982ef79d21bfe2c5"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/592",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5d58433cd090",
+      "dbms": "duckdb",
+      "title": "STDDEV_POP unexpectedly does not fetch any rows",
+      "reported_date": "2020-04-27",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/596",
+        "fix": "https://github.com/cwida/duckdb/commit/2e03516e04ae94c33bff66201cb4290c1dc86662"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/596",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"27/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/596\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/2e03516e04ae94c33bff66201cb4290c1dc86662\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:85c7ca0cec4bbf8c71e9d49ae3bb334a611d4fa0e29b738ce77b74667256ae7b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/596",
+            "source_type": "github_issue",
+            "excerpt": "STDDEV_POP unexpectedly does not fetch any rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:973b0026a5225562c30e877ee5f581835059a2f827af22a39d9958df7c05d9bd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:85c7ca0cec4bbf8c71e9d49ae3bb334a611d4fa0e29b738ce77b74667256ae7b"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/596",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:4d6d86fe61c2",
+      "dbms": "duckdb",
+      "title": "UPDATE results in crash or assertion failure",
+      "reported_date": "2020-04-28",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/599",
+        "fix": "https://github.com/cwida/duckdb/commit/29e801a7a8c1d219f5633b35dca5c88ced20abdd"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/599",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"28/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/599\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/29e801a7a8c1d219f5633b35dca5c88ced20abdd\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7f7f98d08426b34cba4e3cd99c232795a210dd97c686f1d9d96f27c5d6553ee4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/599",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 VARCHAR);\r\nINSERT INTO t0 VALUES (0, 0), (NULL, 0);\r\nUPDATE t0 SET c1 = c0; -- SEGV on unknown address 0x000000000000\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7b9fa532bc7328208a9548cb06bf555d18c4f31df396d7ba90256b65ee1a74d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7f7f98d08426b34cba4e3cd99c232795a210dd97c686f1d9d96f27c5d6553ee4"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/599",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:fffe8655e979",
+      "dbms": "duckdb",
+      "title": "BETWEEN with COLLATE NOACCENT.NOCASE expression results in a segfault/ASan failure",
+      "reported_date": "2020-04-29",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/603",
+        "fix": "https://github.com/cwida/duckdb/commit/67368aa3793590fe503269118c1312b73097be40"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/603",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/603\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/67368aa3793590fe503269118c1312b73097be40\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:06b378306331c4127a6c7b0dd94410d500c5c1334a0f29d6c6931a52f5718754",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/603",
+            "source_type": "github_issue",
+            "excerpt": "BETWEEN with COLLATE NOACCENT.NOCASE expression results in a segfault/ASan failure",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:669848e5f30f895896faa4e892815dffba9d14ddaed0f6953e0bc35381e5c512",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:06b378306331c4127a6c7b0dd94410d500c5c1334a0f29d6c6931a52f5718754"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/603",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:1ffabce469d0",
+      "dbms": "duckdb",
+      "title": "GROUP BY does not take COLLATE into account",
+      "reported_date": "2020-04-29",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/602",
+        "fix": "https://github.com/cwida/duckdb/pull/601/commits/ef292aa079f4a1c6b477d9eb052220d0d1f30c46"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/602",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/04/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/602\",\n            \"fix\": \"https://github.com/cwida/duckdb/pull/601/commits/ef292aa079f4a1c6b477d9eb052220d0d1f30c46\"\n        },\n        \"oracle\": \"TLP (GROUP BY)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:b8331eea1b339457e5b26fbc7714b19781215f022deb330ad9600b9e16f1d469",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/602",
+            "source_type": "github_issue",
+            "excerpt": "GROUP BY does not take COLLATE into account",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ba2d5aab33c78e3e8b846967995bd4a0a35ecf9581f85602e3ca5ce083601311",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b8331eea1b339457e5b26fbc7714b19781215f022deb330ad9600b9e16f1d469"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/602",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:7cb0cc5b502c",
+      "dbms": "duckdb",
+      "title": "Incorrect result for MIN() on expression involving rowid",
+      "reported_date": "2020-05-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/609",
+        "fix": "https://github.com/cwida/duckdb/commit/7cf32b2983e6b579e4e3ec17cc62673a13eae9ee"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/609",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/609\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/7cf32b2983e6b579e4e3ec17cc62673a13eae9ee\"\n        },\n        \"oracle\": \"TLP (aggregate)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:3d07d9d248be3838d86eff32bf8e037255b4577cef26ea9683146fdd9778acab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/609",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT);\r\nINSERT INTO t0(c0) VALUES (0), (0), (0), (0), (0), (0), (0), (0), (0), (0), (0), (0), (0), (0), (0), (0),  (0), (0), (0), (0), (0), (0), (NULL), (NULL);\r\nCREATE INDEX b ON t0(c1);\r\nUPDATE t0 SET c1 = NULL;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:65a9781a75c2c196af762decdb0bf2db162ffa0b43c4f7fe5e061eff7065e710",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3d07d9d248be3838d86eff32bf8e037255b4577cef26ea9683146fdd9778acab"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/609",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:96abee33fad4",
+      "dbms": "duckdb",
+      "title": "Large argument to RPAD results in std::bad_alloc",
+      "reported_date": "2020-05-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/610",
+        "fix": "https://github.com/cwida/duckdb/pull/611/commits/fa0e59bef20c9ed8fd64eaca59405b1f2254903b"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/610",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/610\",\n            \"fix\": \"https://github.com/cwida/duckdb/pull/611/commits/fa0e59bef20c9ed8fd64eaca59405b1f2254903b\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7d7e975b6bebc85270648c6c273c6df2a367679c903971e633165c3fdd211c4d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/610",
+            "source_type": "github_issue",
+            "excerpt": "Large argument to RPAD results in std::bad_alloc",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1a49ad7bc35601be5e80e57975203019587da05e9e6d1d3561b6636a52f0161f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7d7e975b6bebc85270648c6c273c6df2a367679c903971e633165c3fdd211c4d"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/610",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:655d2ffb9804",
+      "dbms": "duckdb",
+      "title": "Building duckdb_java.cpp fails with an include error",
+      "reported_date": "2020-05-02",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/613"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/613",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/613",
+            "source_type": "github_issue",
+            "excerpt": "Building duckdb_java.cpp fails with an include error",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2e1fb2685c2449180031cf45fa591676944b8447cc77370cd81ad62e6fdcd614",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/613",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2e1fb2685c2449180031cf45fa591676944b8447cc77370cd81ad62e6fdcd614"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:4ba1e7291293",
+      "dbms": "duckdb",
+      "title": "Failed ALTER COLUMN results in a \"Transaction conflict\" error that cannot be aborted",
+      "reported_date": "2020-05-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/618",
+        "fix": "https://github.com/cwida/duckdb/commit/1a1990c786bfd5b3a327e97d8c8028319b2ebef8"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/618",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/618\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/1a1990c786bfd5b3a327e97d8c8028319b2ebef8\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:8440e7fce360c21b5b4e24333456f45a197fe8d330dfbe25cd90bf059f1243be",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/618",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DATE);\r\nINSERT INTO t0 VALUES (DATE '2000-01-01');\r\nALTER TABLE t0 ALTER COLUMN c0 SET DATA TYPE INT;\r\nINSERT INTO t0 VALUES (DEFAULT); -- TransactionContext: Transaction conflict: adding entries to a table that has been altered!",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:28c4fd53b2ca0791cc88ce8c2698f9328f3eb21294715f86e63fd87076176538",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8440e7fce360c21b5b4e24333456f45a197fe8d330dfbe25cd90bf059f1243be"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/618",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f325c96c3fcb",
+      "dbms": "duckdb",
+      "title": "Query on altered table results in a segmentation fault",
+      "reported_date": "2020-05-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/619",
+        "fix": "https://github.com/cwida/duckdb/commit/6ef7f2f0202d9d6a8556de022dbe3c9d02851eb5"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/619",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/619\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/6ef7f2f0202d9d6a8556de022dbe3c9d02851eb5\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ce7c81431947fc5507c6971a2fce873a80dda6c0d9398bcdcd7e929a4eb863f3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/619",
+            "source_type": "github_issue",
+            "excerpt": "Query on altered table results in a segmentation fault",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:485e7baa840a3696f558d25526392721925f9ac6f9332461fdd4ef94a4068524",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ce7c81431947fc5507c6971a2fce873a80dda6c0d9398bcdcd7e929a4eb863f3"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/619",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:078c056cd206",
+      "dbms": "duckdb",
+      "title": "ALTER TABLE results in an assertion failure \"Assertion `expr.return_type == vector.type' failed\"",
+      "reported_date": "2020-05-05",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/624",
+        "fix": "https://github.com/cwida/duckdb/commit/b6a065d6ebbab25ae8e023cf1105c5237f26de46"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/624",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/624\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/b6a065d6ebbab25ae8e023cf1105c5237f26de46\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5c63a104d227c09afc7e48a010ed97992a7e8d28c5fb33d0cc137404453fc591",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/624",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 VARCHAR);\r\nINSERT INTO t0(c1) VALUES(NULL);\r\nALTER TABLE t0 ALTER c1 TYPE TIMESTAMP;\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:cad4468ee1ec2ea5ce3a6d05145a5b926b599208b321711c83a542bf722ce5f1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5c63a104d227c09afc7e48a010ed97992a7e8d28c5fb33d0cc137404453fc591"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/624",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:42698c7cc53e",
+      "dbms": "duckdb",
+      "title": "ALTER TYPE with USING results in an assertion failure \"types.size() > 0\"",
+      "reported_date": "2020-05-05",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/629",
+        "fix": "https://github.com/cwida/duckdb/commit/ea6cd8245fd192cd181bdd05fe5ecce5d9a48aa8"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/629",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/629\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/ea6cd8245fd192cd181bdd05fe5ecce5d9a48aa8\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:42610a8765ea27f3d21014ab73485a952174076fa9b323d4355c76665ce82e28",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/629",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nALTER TABLE t0 ALTER c0 TYPE VARCHAR USING ''; -- Assertion `types.size() > 0' failed.\r\n```\r\nUnexpectedly, the `ALTER TABLE` results in an assertion failure:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d99583792310a0af8d623247e15ab9bb7c4f85f096abeb183b8ff59d82bb66b4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:42610a8765ea27f3d21014ab73485a952174076fa9b323d4355c76665ce82e28"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/629",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:718a6dfdc85d",
+      "dbms": "duckdb",
+      "title": "DROP column results in an assertion failure unique.index < base.columns.size() 2",
+      "reported_date": "2020-05-05",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/628",
+        "fix": "https://github.com/cwida/duckdb/commit/5624a825ce7de0844b22356a76c1894bd90f9093"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/628",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/628\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/5624a825ce7de0844b22356a76c1894bd90f9093\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:81991c7a3b390516719fab32617935601031eb996e0cfd10d2e3a580c0f799bb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/628",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT UNIQUE);\r\nALTER TABLE t0 DROP c1; -- Assertion `unique.index < base.columns.size()' failed.\r\n```\r\nUnexpectedly, the `ALTER TABLE` results in an assertion failure:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a96f557be8756728f44c278092c366f174d5efa89ba958711719c5df3077030f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:81991c7a3b390516719fab32617935601031eb996e0cfd10d2e3a580c0f799bb"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/628",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:817bcae5604c",
+      "dbms": "duckdb",
+      "title": "DROP column results in an assertion failure unique.index < base.columns.size().",
+      "reported_date": "2020-05-05",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/625",
+        "fix": "https://github.com/cwida/duckdb/commit/b6665e723c5f3d9cbd7b0018ce8c242260ac6dc3"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/625",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/625\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/b6665e723c5f3d9cbd7b0018ce8c242260ac6dc3\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:8bb725c3c7947a016791721334dd3896a292b028a055d0483137b02cf24ccd60",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/625",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c INT UNIQUE);\r\nALTER TABLE t0 DROP c0; -- Assertion `unique.index < base.columns.size()' failed.\r\n```\r\nUnexpectedly, the `ALTER TABLE` results in an assertion failure:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:38b311cafb5397def1095175f248f4575a622b46f2c7305f89dbefd1702274b2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8bb725c3c7947a016791721334dd3896a292b028a055d0483137b02cf24ccd60"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/625",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:9369543e29ab",
+      "dbms": "duckdb",
+      "title": "UPDATE on altered table results in an error \"Could not find node in column segment tree\"",
+      "reported_date": "2020-05-05",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/622",
+        "fix": "https://github.com/cwida/duckdb/commit/289b2eaa836d22d7994154a912f8b786be48984b"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/622",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/622\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/289b2eaa836d22d7994154a912f8b786be48984b\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:3983dce10b371514c243555beaf47f4d4e43bba06d82e9f650fcbe68bd423806",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/622",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 TIMESTAMP);\r\nINSERT INTO t0 VALUES(NULL);\r\nDELETE FROM t0;\r\nALTER TABLE t0 ALTER c0 TYPE DATE;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:417b3590b99e9f488b15c80f90013b65bb2feea74b8450f2f36fc832621cb0cb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3983dce10b371514c243555beaf47f4d4e43bba06d82e9f650fcbe68bd423806"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/622",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:60d35ebc98f0",
+      "dbms": "duckdb",
+      "title": "Query using LEFT() results in a segmentation fault",
+      "reported_date": "2020-05-06",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/633",
+        "fix": "https://github.com/cwida/duckdb/commit/e05902abc4358bcbaacf41c4075290a0bf6432ee"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/633",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"06/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/633\",\n            \"fix\": \"https://github.com/cwida/duckdb/commit/e05902abc4358bcbaacf41c4075290a0bf6432ee\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:870de7c24c4ae9417762fd09db97cf272ad56fc9bfad6230958a15f203f2c122",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/633",
+            "source_type": "github_issue",
+            "excerpt": "Query using LEFT() results in a segmentation fault",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:397d6d7b9a402976c526d727fca753cfe2f5dc137ab4e7da9ea95736c190e4c3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:870de7c24c4ae9417762fd09db97cf272ad56fc9bfad6230958a15f203f2c122"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/633",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:015594841299",
+      "dbms": "duckdb",
+      "title": "SELECT with RIGHT JOIN causes an assertion failure \"Assertion `!finalized' failed\"",
+      "reported_date": "2020-05-07",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/637",
+        "fix": "https://github.com/cwida/duckdb/pull/690"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/637",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/05/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/637\",\n            \"fix\": \"https://github.com/cwida/duckdb/pull/690\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:09e9e9104f8d041a4e831b31e3cac0732b080782a90591c2c7963f87836f87b6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/637",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nCREATE TABLE t1(c0 VARCHAR);\r\nINSERT INTO t0 VALUES('');\r\nINSERT INTO t1 VALUES(0);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d0aa3d242e10e262e33e1685a28930834dcda1c9353574f71c445e35b0cc524f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:09e9e9104f8d041a4e831b31e3cac0732b080782a90591c2c7963f87836f87b6"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/637",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:2e69cafbc02d",
+      "dbms": "duckdb",
+      "title": "Maven artifact & JDBC driver fat JAR",
+      "reported_date": "2020-05-26",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/649"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/649",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/649",
+            "source_type": "github_issue",
+            "excerpt": "Maven artifact & JDBC driver fat JAR",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9918eb5063265dccad3f123b5b7015261a1c034ba4ccce211a973ddb1c045478",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/649",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9918eb5063265dccad3f123b5b7015261a1c034ba4ccce211a973ddb1c045478"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:278f234170ff",
+      "dbms": "duckdb",
+      "title": "Column ID mismatch in adaptive filters",
+      "reported_date": "2020-06-24",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "hannes",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/709"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/709",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/709",
+            "source_type": "github_issue",
+            "excerpt": "This was found by @mrigger's SQLancer (somewhat minified by us)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:8db462e921dacf216f5b6990f84edd6d8094e54d4514b89a8c4b84ec332a60ca",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/709",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8db462e921dacf216f5b6990f84edd6d8094e54d4514b89a8c4b84ec332a60ca"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:774a0a09fcb9",
+      "dbms": "duckdb",
+      "title": "DuckDB JDBC driver issue in the presence of other JDBC drivers",
+      "reported_date": "2020-08-18",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/848"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/848",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/848",
+            "source_type": "github_issue",
+            "excerpt": "DuckDB JDBC driver issue in the presence of other JDBC drivers",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9804b5e3174a81107130e6a0e33938e4cabf0805d5edffe8e8469df5386c42f9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/848",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9804b5e3174a81107130e6a0e33938e4cabf0805d5edffe8e8469df5386c42f9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f8ab0e25da21",
+      "dbms": "duckdb",
+      "title": "Buffer overflow in duckdb::ART::IteratorNext",
+      "reported_date": "2020-09-24",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/cwida/duckdb/issues/956"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cwida/duckdb/issues/956",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/09/2020\",\n        \"dbms\": \"DuckDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/cwida/duckdb/issues/956\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f2288d7e0e997e4040b96840eecec7f676c6fa6d6509b444f2dfb31788de974b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/956",
+            "source_type": "github_issue",
+            "excerpt": "Buffer overflow in duckdb::ART::IteratorNext",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:0567f4fa5bbefab2ce4f0e954cd693cd7482671af8cc6caea9546605f52ef5cc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f2288d7e0e997e4040b96840eecec7f676c6fa6d6509b444f2dfb31788de974b"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/956",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:0577cd7a1a36",
+      "dbms": "duckdb",
+      "title": "Getting a query plan using JDBC",
+      "reported_date": "2020-09-24",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/958"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/958",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/958",
+            "source_type": "github_issue",
+            "excerpt": "Getting a query plan using JDBC",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2421b5874ba51a4a0f3d49681dcf9aefeabfcf0f990953e5fcd6798942006578",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/958",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2421b5874ba51a4a0f3d49681dcf9aefeabfcf0f990953e5fcd6798942006578"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:1fe19a23147e",
+      "dbms": "duckdb",
+      "title": "Typeof does not work when using JDBC",
+      "reported_date": "2020-09-28",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/966"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/966",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/966",
+            "source_type": "github_issue",
+            "excerpt": "Typeof does not work when using JDBC",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:670a0fb4d715deffe1ed7b9fb0cf2be5ccb8972bf8954c8191cd5c6f1aff56bf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/966",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:670a0fb4d715deffe1ed7b9fb0cf2be5ccb8972bf8954c8191cd5c6f1aff56bf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:51ceeab2e3f6",
+      "dbms": "duckdb",
+      "title": "Comparison with boolean unexpectedly evaluates to false",
+      "reported_date": "2020-12-13",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/1224"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/1224",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/1224",
+            "source_type": "github_issue",
+            "excerpt": "Unexpectedly, no row is fetched. I think this is a regression bug. SQLancer found it using the TLP WHERE oracle, because the following two queries unexpectedly did not compute the same result set:",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:79dad11609467b5f1108d2a343c30c4c55379fb4ee9aed601f1ad0d9b25f6ad5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/1224",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:79dad11609467b5f1108d2a343c30c4c55379fb4ee9aed601f1ad0d9b25f6ad5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:25f3e4de951a",
+      "dbms": "duckdb",
+      "title": "Filter issue with `IN` and `NULL`",
+      "reported_date": "2021-01-27",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "hannes",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/1334"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/1334",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/1334",
+            "source_type": "github_issue",
+            "excerpt": "This was found by SQLancer:",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:06e58053ec13d1e58ff788b2d0fc266fc5ff1145134713bca38bc43dffe204f3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/1334",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:06e58053ec13d1e58ff788b2d0fc266fc5ff1145134713bca38bc43dffe204f3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:9edcd9e39b20",
+      "dbms": "duckdb",
+      "title": "BETWEEN with subquery and column results in unexpected error",
+      "reported_date": "2022-05-06",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "kokrui",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/3588"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/3588",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/3588",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nSELECT * FROM t0 WHERE (SELECT 0) BETWEEN 0 AND t0.c0; \r\n-- Error: INTERNAL Error: Attempting to initialize state of expression of unknown type!\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by kokrui of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:da9c3bfd1d5d410f30d91037bc98c4e38fd16f1d2f60e62ff2b208bc9827453a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by kokrui.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:da9c3bfd1d5d410f30d91037bc98c4e38fd16f1d2f60e62ff2b208bc9827453a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c6f925d0e31f",
+      "dbms": "duckdb",
+      "title": "Unexpected Error for RTRIM Function in Where Clause",
+      "reported_date": "2022-05-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "kokrui",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/3616"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/3616",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/3616",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nSELECT * FROM t0 WHERE RTRIM(false) > c0 OR 0 <= c0; \r\n-- Error: INTERNAL Error: Attempting to initialize state of expression of unknown type!\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by kokrui of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:8d52f565eeb9d1f07d00792e98f7d2f43ca02214120874ed292aa6ca8a198c4c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by kokrui.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:8d52f565eeb9d1f07d00792e98f7d2f43ca02214120874ed292aa6ca8a198c4c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:64fecb70f667",
+      "dbms": "duckdb",
+      "title": "Catalog Error: Table with name t0 does not exist!\nDid you mean \"t1\"?\nCatalog Error: Table with name t0 does not exist!\nDid you mean \"t1\"?\nLINE 1: UPDATE t0 SET c0=DATE '1970-01-22';\n               ^",
+      "reported_date": "2022-08-12",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fuzzerofducks",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb-fuzzer/issues/71"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb-fuzzer/issues/71",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/71",
+            "source_type": "github_issue",
+            "excerpt": "Issue found by SQLancer on git commit hash [ebe45](https://github.com/duckdb/duckdb/commit/ebe45abbb552d184dff9173c3412f63c173a0918) using seed 943318896.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:447afddf52d8ff0dbf4a215622303910d67b43d86728fb2c5b72e9401830f40f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T05:57:42Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T05:57:42Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/71",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:447afddf52d8ff0dbf4a215622303910d67b43d86728fb2c5b72e9401830f40f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:efbe1afc62a1",
+      "dbms": "duckdb",
+      "title": "Catalog Error: Table with name t25 does not exist!\nDid you mean \"t0\"?\nCatalog Error: Table with name t25 does not exist!\nDid you mean \"t0\"?\nCatalog Error: Table with name t25 does not exist!\nDid you mean \"t0\"?\nCatalog Error: Table with name...",
+      "reported_date": "2022-09-07",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fuzzerofducks",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb-fuzzer/issues/78"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb-fuzzer/issues/78",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/78",
+            "source_type": "github_issue",
+            "excerpt": "Issue found by SQLancer on git commit hash [9bf9c](https://github.com/duckdb/duckdb/commit/9bf9c4b8f69bc445cbbd175a2d8d408b201c6bbe) using seed 993149434.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:07194516390e03e907ac02bdcae6ca4c8032bc6a8c64377bfa7d11c406af45e9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T05:57:42Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T05:57:42Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/78",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:07194516390e03e907ac02bdcae6ca4c8032bc6a8c64377bfa7d11c406af45e9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:5c6550b5d846",
+      "dbms": "duckdb",
+      "title": "Catalog Error: Table with name t0 does not exist!\nDid you mean \"t44\"?\nCatalog Error: Table with name t0 does not exist!\nDid you mean \"t44\"?\nCatalog Error: Table with name t0 does not exist!\nDid you mean \"t44\"?\nCatalog Error: Table with name...",
+      "reported_date": "2022-09-08",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fuzzerofducks",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb-fuzzer/issues/79"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb-fuzzer/issues/79",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/79",
+            "source_type": "github_issue",
+            "excerpt": "Issue found by SQLancer on git commit hash [1199e](https://github.com/duckdb/duckdb/commit/1199e3dbff28fdd8124d5b255241250b6a6b9bc3) using seed 881305653.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:6765ef9c9964f305ec92e4a7a3803e96c30da9d01925d488e5a30739daa69757",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T05:57:42Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T05:57:42Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/79",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6765ef9c9964f305ec92e4a7a3803e96c30da9d01925d488e5a30739daa69757"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:7c7d9355c0d5",
+      "dbms": "duckdb",
+      "title": "Catalog Error: Table with name t4 does not exist!\nDid you mean \"t0\"?\nCatalog Error: Table with name t4 does not exist!\nDid you mean \"t0\"?\nLINE 1: UPDATE t4 SET c1=TIMESTAMP '1969-12-16 01:34:5...\n               ^\nCatalog Error: Table with n...",
+      "reported_date": "2022-09-23",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fuzzerofducks",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb-fuzzer/issues/82"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb-fuzzer/issues/82",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/82",
+            "source_type": "github_issue",
+            "excerpt": "Issue found by SQLancer on git commit hash [201c1](https://github.com/duckdb/duckdb/commit/201c1305f4ae7edad74ea5366b9666d93c227d7d) using seed 569502376.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:abd443d045e78917daae50fd913e55c67dd841beeeded1b635db8e49d34feee5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T05:57:42Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T05:57:42Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/82",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:abd443d045e78917daae50fd913e55c67dd841beeeded1b635db8e49d34feee5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:e3ae71582d84",
+      "dbms": "duckdb",
+      "title": "/usr/include/c++/9/bits/unique_ptr.h:347:9: runtime error: reference binding to null pointer of type 'struct Key'",
+      "reported_date": "2022-09-24",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fuzzerofducks",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb-fuzzer/issues/85"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb-fuzzer/issues/85",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/85",
+            "source_type": "github_issue",
+            "excerpt": "Issue found by SQLancer on git commit hash [a0fc1](https://github.com/duckdb/duckdb/commit/a0fc1e9cd0c9b03020b4ed9c7ba61fcd67d54681) using seed 149410154.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:7c1c42b732a30e66caea720fafb33b9da044a41eb35e471a127c818f374a4cf7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T05:57:42Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T05:57:42Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/85",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7c1c42b732a30e66caea720fafb33b9da044a41eb35e471a127c818f374a4cf7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:85bb8ce1c4fd",
+      "dbms": "duckdb",
+      "title": "Crash When Creating Index",
+      "reported_date": "2022-10-13",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/4976"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/4976",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/4976",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DOUBLE, c1 TIMESTAMP DEFAULT(TIMESTAMP '1970-01-04 12:58:32'));\r\nINSERT INTO t0(c1, c0) VALUES (TIMESTAMP '1969-12-28 23:02:08', 1);\r\nINSERT INTO t0(c0) VALUES (DEFAULT);\r\nCREATE INDEX i2 ON t0(c1, c0); -- segment fault",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:31a014de2c7da537e992aa5b5386177d736037c9bf322d0318f86dc84930c66b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:31a014de2c7da537e992aa5b5386177d736037c9bf322d0318f86dc84930c66b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:cac734c75695",
+      "dbms": "duckdb",
+      "title": "Results are different between Python Interface and CLI when executing queries with UNION and LIMIT",
+      "reported_date": "2022-11-19",
+      "reported_year": 2022,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/5413"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/5413",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/5413",
+            "source_type": "github_issue",
+            "excerpt": "Results are different between Python Interface and CLI when executing queries with UNION and LIMIT",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:bea29a15d3d6ce035336ac619e1f958370a91dc1969e3123d15ce3c8312d30fa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/5413",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bea29a15d3d6ce035336ac619e1f958370a91dc1969e3123d15ce3c8312d30fa"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:6cff6b713d48",
+      "dbms": "duckdb",
+      "title": "TestDuckDBJDBC::test_lots_of_decimals is broken",
+      "reported_date": "2023-02-02",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "SuriZhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/6073"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/6073",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/6073",
+            "source_type": "github_issue",
+            "excerpt": "TestDuckDBJDBC::test_lots_of_decimals is broken",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:672d0df5eae385cce58e5567d9d5b7dae34e25b8c12f63ea5319083a48c2552b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "SuriZhang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/6073",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:672d0df5eae385cce58e5567d9d5b7dae34e25b8c12f63ea5319083a48c2552b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:0b4d51f2a7f9",
+      "dbms": "duckdb",
+      "title": "Internal Error: Calling StringValue::Get on a NULL value",
+      "reported_date": "2023-02-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/9870"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/9870",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/9870",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nSELECT REGEXP_MATCHES(c0, NULL) FROM t0;\r\n-- Error: INTERNAL Error: Calling StringValue::Get on a NULL value\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cf6138bf270724a099560f9bc6e6b85559f2f4c75ba3f7af6e98be321bdef708",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cf6138bf270724a099560f9bc6e6b85559f2f4c75ba3f7af6e98be321bdef708"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:605a3a032917",
+      "dbms": "duckdb",
+      "title": "Crash When Altering Schema",
+      "reported_date": "2023-03-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/6640"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/6640",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/6640",
+            "source_type": "github_issue",
+            "excerpt": "Crash When Altering Schema",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6b48fbab778b4486a671bda8b58c5dedad3c39be8f0afe8f5c3393fa2a69ed79",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/6640",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6b48fbab778b4486a671bda8b58c5dedad3c39be8f0afe8f5c3393fa2a69ed79"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:3d164bc2d1db",
+      "dbms": "duckdb",
+      "title": "Crash When Updating after Commit",
+      "reported_date": "2023-03-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/6651"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/6651",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/6651",
+            "source_type": "github_issue",
+            "excerpt": "Crash When Updating after Commit",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:abeb391428a3c5ca9bf80cc4392ed325cb70d5a766a3fed432bdb62bb3f45b5c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/6651",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:abeb391428a3c5ca9bf80cc4392ed325cb70d5a766a3fed432bdb62bb3f45b5c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:ce223785fc99",
+      "dbms": "duckdb",
+      "title": "Inconsistencies Found When Comparing Null Values within Rows",
+      "reported_date": "2023-03-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/6664"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/6664",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/6664",
+            "source_type": "github_issue",
+            "excerpt": "Inconsistencies Found When Comparing Null Values within Rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:8523020318f330dca2ed01c29f5275ab56ad1f536edb9cc1b8946651b38455f5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/6664",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8523020318f330dca2ed01c29f5275ab56ad1f536edb9cc1b8946651b38455f5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:2deb503f8823",
+      "dbms": "duckdb",
+      "title": "Segmentation Fault on Select query on table with column of array type.",
+      "reported_date": "2023-03-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "nish-d",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/6656"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/6656",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/6656",
+            "source_type": "github_issue",
+            "excerpt": "Segmentation Fault on Select query on table with column of array type.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:46d3875a9e0b3b1b53a785ce5d35efd40b919dbdeb1a053bfff51ed4afe311be",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "nish-d is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/6656",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:46d3875a9e0b3b1b53a785ce5d35efd40b919dbdeb1a053bfff51ed4afe311be"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f2289a87aa6c",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Failed to bind column reference \"\" [18.2] (bindings: [18.1])",
+      "reported_date": "2023-03-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fuzzerofducks",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb-fuzzer/issues/129"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb-fuzzer/issues/129",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/129",
+            "source_type": "github_issue",
+            "excerpt": "Issue found by SQLancer on git commit hash [65257](https://github.com/duckdb/duckdb/commit/6525767cf115f7996eaab67641a5eae3a41ab2fd) using seed 303089133.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:5d6bdd8d10ec30650d119d71f20bcfa4f777cfbdba4f973af35d987011d12f8a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T05:57:42Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T05:57:42Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb-fuzzer/issues/129",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5d6bdd8d10ec30650d119d71f20bcfa4f777cfbdba4f973af35d987011d12f8a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:c7e729b6aa8e",
+      "dbms": "duckdb",
+      "title": "Incorrect results might caused by `INDEX`",
+      "reported_date": "2023-03-25",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/6860"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/6860",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/6860",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\nINSERT INTO t0(c0) VALUES (1), (1), (1), (1), (1), (1), (1), (1), (1), (1), (1), (1), (1), (1), (1), (1);\r\nCREATE INDEX i3 ON t0(c0);\r\nUPDATE t0 SET c0=(0.1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:066e17b6240019503905dbf629dfefe928a4ae2fa5938c0aae7aa9a28e2a234f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/6860",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:066e17b6240019503905dbf629dfefe928a4ae2fa5938c0aae7aa9a28e2a234f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:bd547f9895dc",
+      "dbms": "duckdb",
+      "title": "Segmentation fault on `BETWEEN`",
+      "reported_date": "2023-03-25",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/6861"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/6861",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/6861",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 TIMESTAMP);\r\n\r\nINSERT INTO t0(c0) VALUES ((DATE '1969-12-10')), ((DATE '1969-12-16')), ((DATE '1969-12-07')), ((TIMESTAMP '1969-12-09 10:08:32')), ((DATE '1969-12-30')), ((TIMESTAMP '1969-12-21 00:06:38'));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:caff3ddb2a9106df0ca5bdece86e51ad6f7fae54378b65b4f04ce57afe124a84",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:caff3ddb2a9106df0ca5bdece86e51ad6f7fae54378b65b4f04ce57afe124a84"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:cf56ef699311",
+      "dbms": "duckdb",
+      "title": "INSERT statement returns unexpected exception",
+      "reported_date": "2023-03-27",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "SuriZhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/6876"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/6876",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/6876",
+            "source_type": "github_issue",
+            "excerpt": "INSERT statement returns unexpected exception",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9d822c4e53fbf3bace213f6fd337a947fc545963f8ed7941fb3d069131c08ba6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "SuriZhang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/6876",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9d822c4e53fbf3bace213f6fd337a947fc545963f8ed7941fb3d069131c08ba6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:1f0042ca26d0",
+      "dbms": "duckdb",
+      "title": "Incorrect result by subquery and join",
+      "reported_date": "2023-04-04",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/100561"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/100561",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/100561",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 SERIAL2);\r\nCREATE TABLE t1 (c0 BYTES);\r\n\r\nUPSERT INTO t1 (rowid) VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:27535dfcda82825ba088a4aa3ce3f8be7e7520b8cb16e02e7c34e28bd33a7f4f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:27535dfcda82825ba088a4aa3ce3f8be7e7520b8cb16e02e7c34e28bd33a7f4f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8a204dd665e2",
+      "dbms": "duckdb",
+      "title": "internal error: null rejection requested on non-null column",
+      "reported_date": "2023-04-04",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/100559"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/100559",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/100559",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INTERVAL);\r\nCREATE TABLE t1 (c0 INTERVAL);\r\nCREATE TABLE t2 (c0 FLOAT);\r\nSELECT t1.c0 AS c0 FROM t0, t2 FULL OUTER JOIN t1 ON true WHERE (((t0.c0) IN (t1.c0)) AND ((t2.c0) IN (SELECT STDDEV(t2.c0) FROM t2)));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:0bee93ee65bfa74c75221357769585ac7b590d618b67a830a66395ab198437f8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:0bee93ee65bfa74c75221357769585ac7b590d618b67a830a66395ab198437f8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:08f3ae8fe8a7",
+      "dbms": "duckdb",
+      "title": "kvcoord: setting `kv.range_descriptor_cache.size=0` results in range iterator to retry endlessly",
+      "reported_date": "2023-04-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/101011"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/101011",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/101011",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 TIMESTAMP);\r\nSHOW RANGES FROM TABLE t0;\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cb5664a44e302d269f27ee4fde1b80fa4d0f33e5ef361a8579cd1a193de9043f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cb5664a44e302d269f27ee4fde1b80fa4d0f33e5ef361a8579cd1a193de9043f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5ef5b1e2840d",
+      "dbms": "duckdb",
+      "title": "The `AVG` function modify the value of `REAL` in some query",
+      "reported_date": "2023-04-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7045"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7045",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7045",
+            "source_type": "github_issue",
+            "excerpt": "The `AVG` function modify the value of `REAL` in some query",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c678d716b24a843618dab31c1dc2761f7b6b2b955b6c62758f642e69ccd53c3a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/7045",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c678d716b24a843618dab31c1dc2761f7b6b2b955b6c62758f642e69ccd53c3a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:bef7f029b3aa",
+      "dbms": "duckdb",
+      "title": "Unexpected result when take subquery as operand",
+      "reported_date": "2023-04-12",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7044"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7044",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7044",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT2);\r\nCREATE TABLE t1(c0 BOOL);\r\n\r\nINSERT INTO t1(c0) VALUES ((true));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:de8ec9e9306f4dc79914711794ecc01109c00d7cc525b71caccd5d1831cae8fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:de8ec9e9306f4dc79914711794ecc01109c00d7cc525b71caccd5d1831cae8fc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:90dd43a37169",
+      "dbms": "duckdb",
+      "title": "Different returned datatype of a `CASE` statement in different query.",
+      "reported_date": "2023-04-13",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7064"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7064",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7064",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DATE);\r\nINSERT INTO t0(c0) VALUES ((DATE '1969-12-10'));\r\n\r\nSELECT t0.c0 FROM t0 WHERE (((CASE (1) WHEN (2) THEN (TIMESTAMP '1969-12-21 06:13:06') ELSE t0.c0 END )) NOT LIKE((DATE '1969-12-10')));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:2feb4116b868232c45b08b7733b0d005e731ee69b6d42bee7425f527244ca1ea",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:2feb4116b868232c45b08b7733b0d005e731ee69b6d42bee7425f527244ca1ea"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:9e0dbda03bf4",
+      "dbms": "duckdb",
+      "title": "DuckDB JDBC driver segfault at duplicate method call",
+      "reported_date": "2023-04-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "SuriZhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7062"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7062",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7062",
+            "source_type": "github_issue",
+            "excerpt": "DuckDB JDBC driver segfault at duplicate method call",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d2edb43ff87d6e01f5bc22be764c778addfb413cad6bc23f031c0179eb28178c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "SuriZhang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/7062",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d2edb43ff87d6e01f5bc22be764c778addfb413cad6bc23f031c0179eb28178c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:25e365b45cd4",
+      "dbms": "duckdb",
+      "title": "internal error: failed to parse locale \"\": language: tag is not well-formed",
+      "reported_date": "2023-04-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/101418"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/101418",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/101418",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 SMALLINT, c1 STRING COLLATE cmn);\r\nCREATE TABLE t1 (c0 BOOL);\r\nUPSERT INTO t1 (c0) VALUES(true);\r\nSELECT t1.rowid FROM t1 WHERE ((CASE  WHEN false THEN CASE  WHEN (NOT (false)) THEN NULL END ELSE ((((''))) COLLATE es) END)>=(SELECT MAX(t0.c1) AS c0 FROM t0));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a254d185b6e84a704b5e198bc3405c6d1fbd56fd7324115f6d0f7af7e29b62f7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a254d185b6e84a704b5e198bc3405c6d1fbd56fd7324115f6d0f7af7e29b62f7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:77192a3cd096",
+      "dbms": "duckdb",
+      "title": "Unexpected results of `TIME` value under JDBC",
+      "reported_date": "2023-04-15",
+      "reported_year": 2023,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/101591"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/101591",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/101591",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 TIMETZ);\r\n\r\nINSERT INTO t0 (c0) VALUES (TIMETZ '1969-12-17T16:27:27');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9ed99e6122291353b0bb3811b416727dfdd0ee35a0496e8ca76997396a4583e5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9ed99e6122291353b0bb3811b416727dfdd0ee35a0496e8ca76997396a4583e5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:d651b43a0cc3",
+      "dbms": "duckdb",
+      "title": "Unexpected results when integer overflow",
+      "reported_date": "2023-04-15",
+      "reported_year": 2023,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7094"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7094",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7094",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT8);\r\nINSERT INTO t0(c1) VALUES ((1));\r\n\r\nSELECT t0.c1 FROM t0 WHERE (((-1314689763))+((-1947665992))<=((false)));  -- 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9d030621308fba13e0b11a571c17c24caadcda615615d16826a17c109ccd1df2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9d030621308fba13e0b11a571c17c24caadcda615615d16826a17c109ccd1df2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:db302206d1d2",
+      "dbms": "duckdb",
+      "title": "Hang on `SELECT` query",
+      "reported_date": "2023-04-19",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7143"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7143",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7143",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\r\nCREATE TABLE t1(c0 DATE);\r\n\r\nINSERT INTO t0(c0) VALUES ((true));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:876e52aefe3803c35c52e2eaf6cda9f6518a9b1e52d9da5f99f80701a2d7c6e9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:876e52aefe3803c35c52e2eaf6cda9f6518a9b1e52d9da5f99f80701a2d7c6e9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:efb5fb651d3e",
+      "dbms": "duckdb",
+      "title": "Unexpected results on specific values",
+      "reported_date": "2023-04-19",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7142"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7142",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7142",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 SMALLINT);\r\n\r\nINSERT INTO t0(c0) VALUES ((0.8486559842619935));\r\nINSERT INTO t0(c0) VALUES ((false));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5a0c70d04e151670f173390a4c180cf69b006a501ade39ba587f51221eca8fff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5a0c70d04e151670f173390a4c180cf69b006a501ade39ba587f51221eca8fff"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:e30d9da8092e",
+      "dbms": "duckdb",
+      "title": "`internal error: expected subquery to be lazily planned as a routine`",
+      "reported_date": "2023-04-21",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/101980"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/101980",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/101980",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 FLOAT);\r\nCREATE TABLE t1 (c0 SERIAL4);\r\nCREATE TABLE t2 (c0 SERIAL2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:299227cb0b006cb5e2e79f320d53a25e187f5c15d0ffb9a80e8cf60685f50167",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:299227cb0b006cb5e2e79f320d53a25e187f5c15d0ffb9a80e8cf60685f50167"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8bcda4cb1f3c",
+      "dbms": "duckdb",
+      "title": "Unexpected results of `CASE` and `BETWEEN`",
+      "reported_date": "2023-04-24",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/102110"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/102110",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/102110",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 TIMESTAMPTZ, c1 VARBIT(101) DEFAULT (B'111111111'));\r\nUPSERT INTO t1 (c0) VALUES(TIMESTAMPTZ '1969-12-10T06:25:44');\r\nINSERT INTO t1 (c1) VALUES(NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:01238c04da749ae2a80c92c110d491e77b805f5c4828c3d1024bb92c5a8d8b73",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:01238c04da749ae2a80c92c110d491e77b805f5c4828c3d1024bb92c5a8d8b73"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f0928e08f760",
+      "dbms": "duckdb",
+      "title": "`INTERNAL Error: Logical column index 1 out of range`",
+      "reported_date": "2023-04-26",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7250"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7250",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7250",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t21(c0 INT64);\r\nCREATE TABLE t0(c0 INT64, c1 INT64);\r\n\r\nCREATE VIEW v0(c0) AS SELECT t0.c1 FROM t21, t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:baae0236fcd31898d9d5bca95220e95ffa029f13f90f44620cd33f8d5a0165f4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:baae0236fcd31898d9d5bca95220e95ffa029f13f90f44620cd33f8d5a0165f4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5155a63ebc3f",
+      "dbms": "duckdb",
+      "title": "Unexpected results might be related to `BIT` and `INDEX`",
+      "reported_date": "2023-04-27",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/102412"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/102412",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/102412",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 VARBIT(55)[] AS (ARRAY[B'1111111111111111111111111111111111110010101010100100111000111001', B'1']) STORED);\r\n\r\nCREATE UNIQUE INDEX ON t0(rowid DESC, c0 DESC);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:91116d9ed51bccfa75a598387b8b9bddefa45e7d1a9998f54d829652a850e8c1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:91116d9ed51bccfa75a598387b8b9bddefa45e7d1a9998f54d829652a850e8c1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b15aa9cc6e1d",
+      "dbms": "duckdb",
+      "title": "Unexpected error `unknown signature: bit_length(string) (desired <int[]>)`",
+      "reported_date": "2023-05-02",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/102716"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/102716",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/102716",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nSELECT 0 FROM t0 HAVING (1 > ANY (BIT_LENGTH('')));\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:1617e30c7d17dad5586b4b7a80015a6966c5d8f5f1a2821477bd5d069b8b9686",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:1617e30c7d17dad5586b4b7a80015a6966c5d8f5f1a2821477bd5d069b8b9686"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:1de9177500e3",
+      "dbms": "duckdb",
+      "title": "Unexpected results of `IN` with redundant values list",
+      "reported_date": "2023-05-08",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/102864"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/102864",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/102864",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT4);\r\nINSERT INTO t0 (rowid, c0) VALUES(0, 0);\r\nINSERT INTO t0 (c0) VALUES(5);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ddf3fa44af97b7dae44b94f05913df71d7ba6543770915c9f2a8855da4275bea",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ddf3fa44af97b7dae44b94f05913df71d7ba6543770915c9f2a8855da4275bea"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b684385ce4dd",
+      "dbms": "duckdb",
+      "title": "DuckDB JDBC throws `Resource temporarily unavailable`",
+      "reported_date": "2023-05-11",
+      "reported_year": 2023,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "SuriZhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7462"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7462",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7462",
+            "source_type": "github_issue",
+            "excerpt": "DuckDB JDBC throws `Resource temporarily unavailable`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c08be9f8b58814fb951d9a85d769d39062d5f91dc8a895f3cc88500d4fa5f569",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "SuriZhang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/7462",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c08be9f8b58814fb951d9a85d769d39062d5f91dc8a895f3cc88500d4fa5f569"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f267312b72a0",
+      "dbms": "duckdb",
+      "title": "Unexpected error `Invalid Error: Argument index \"0\" out of range`",
+      "reported_date": "2023-05-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7498"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7498",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7498",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nSELECT (t0.c0) FROM t0 WHERE (1 IN (2, PRINTF(('S%m'))));\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6ca68cfc82cb77926e6a5a7d0825d4a307c13183a4105ac84b65caa8deaf7d4d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/7498",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6ca68cfc82cb77926e6a5a7d0825d4a307c13183a4105ac84b65caa8deaf7d4d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:9570eaa745e2",
+      "dbms": "duckdb",
+      "title": "DuckDB hang on `DELETE`",
+      "reported_date": "2023-05-16",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7530"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7530",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7530",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t14(c0 BIGINT);\r\nINSERT INTO t14(c0) VALUES ((1)), ((1)), ((1));\r\nCREATE INDEX i1 ON t14(c0 );\r\nDELETE FROM t14 WHERE t14.rowid;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ad33ce8abcd70aae0925e50c586ebd93b0cdc5da4ad2e608e960e9f92cdb25d5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ad33ce8abcd70aae0925e50c586ebd93b0cdc5da4ad2e608e960e9f92cdb25d5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8819927c8645",
+      "dbms": "duckdb",
+      "title": "Segmentation fault on `SELECT`",
+      "reported_date": "2023-05-17",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7551"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7551",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7551",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nCREATE TABLE t1(c0 DATE);\r\n\r\nINSERT INTO t1(c0) VALUES ((TIMESTAMP '1970-01-25 15:59:18'));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:846b945ecbd657be57cbad81e54c0cd96fc9453ac04cb9d6949dc3d48344790d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:846b945ecbd657be57cbad81e54c0cd96fc9453ac04cb9d6949dc3d48344790d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:6192e806124b",
+      "dbms": "duckdb",
+      "title": "Inconsistent results of subquery in `INSERT` and `SELECT`, and an unexpected error",
+      "reported_date": "2023-05-20",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7601"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7601",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7601",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BIGINT);\r\n\r\nINSERT INTO t0(c0) VALUES ((-1089286905)), ((315540959)), ((1.591828186E9)), (DEFAULT), ((true)), ((false));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9b211a1fcfafa2f1cb65e25cd8f72e6869ba9154f62f0a6a0c7d02ec94902650",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9b211a1fcfafa2f1cb65e25cd8f72e6869ba9154f62f0a6a0c7d02ec94902650"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b5b7698b1bdb",
+      "dbms": "duckdb",
+      "title": "DuckDB hang on `DELETE`",
+      "reported_date": "2023-05-21",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7610"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7610",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7610",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\n\r\nINSERT INTO t0(c0) VALUES ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a'), ('a');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b770a5684e89bbe094ad06f14ccc5df2ed4206297924c7767c2bac784def79a6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b770a5684e89bbe094ad06f14ccc5df2ed4206297924c7767c2bac784def79a6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:254a07849fb8",
+      "dbms": "duckdb",
+      "title": "Unexpected results related to `INDEX`",
+      "reported_date": "2023-05-23",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/103755"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/103755",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/103755",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 INT4);\r\nCREATE INDEX ON t1(c0) USING HASH WITH BUCKET_COUNT=794;\r\nINSERT INTO t1 (c0) VALUES(-10), (-20);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:77d186ec5690315882a4e80d901181f7cb985318c1f4ddf86981e67da5ee60a7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:77d186ec5690315882a4e80d901181f7cb985318c1f4ddf86981e67da5ee60a7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:6a0d86fe6390",
+      "dbms": "duckdb",
+      "title": "Inconsistent result of `AVG` on big float number",
+      "reported_date": "2023-06-05",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/104319"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/104319",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/104319",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t86 (c0 FLOAT);\r\nCREATE TABLE t0 (c1 STRING);\r\n\r\nINSERT INTO t86 (c0) VALUES(-1.7976931348623157E308), (1.7976931348623157E308);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:1719c7c75bf00c67f2c3e761ad3b2e98dba2866021a83bd10ed107dddbf2d5a9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:1719c7c75bf00c67f2c3e761ad3b2e98dba2866021a83bd10ed107dddbf2d5a9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:bbb0f6abccfd",
+      "dbms": "duckdb",
+      "title": "Unexpected results on common table expressions",
+      "reported_date": "2023-06-16",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/7960"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/7960",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/7960",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 SIGNED);\r\nCREATE TABLE t1(c0 INT2);\r\nINSERT INTO t0(c0) VALUES ((0.5)), ((0.5));\r\nINSERT INTO t1(c0) VALUES ((0.3)), ((NULL));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:38b8f22d8fe886ec2b907b6e14c86fca9b1e3bf5507c2385a1970091284371f8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:38b8f22d8fe886ec2b907b6e14c86fca9b1e3bf5507c2385a1970091284371f8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:fee84c2b0a1f",
+      "dbms": "duckdb",
+      "title": "Unexpected hang on subquery",
+      "reported_date": "2023-06-30",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/cockroachdb/cockroach/issues/105882"
+      },
+      "primary_url": "https://github.com/cockroachdb/cockroach/issues/105882",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cockroachdb/cockroach/issues/105882",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c0 INT);\r\n\r\nUPSERT INTO t1 (c0) VALUES(1);\r\nSELECT (SELECT COUNT(t1a.rowid) FROM t1 AS t1a WHERE ((t1.rowid) IN (SELECT MAX(t1.rowid) FROM t1))) FROM t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5b0e226edd61146b11319c34ba9a5f53364d63b0419c9d84c07000a298579174",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5b0e226edd61146b11319c34ba9a5f53364d63b0419c9d84c07000a298579174"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:a4b6385ad168",
+      "dbms": "duckdb",
+      "title": "Recursive SQL Query Leads to Infinite Loop",
+      "reported_date": "2023-08-23",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/8667"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/8667",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/8667",
+            "source_type": "github_issue",
+            "excerpt": "Recursive SQL Query Leads to Infinite Loop",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:917f2bedf798b1d94136f4d1ade3325d1c7433bd3f74e91fdf99654b6dab1ddf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/8667",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:917f2bedf798b1d94136f4d1ade3325d1c7433bd3f74e91fdf99654b6dab1ddf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:e0817ec4eff7",
+      "dbms": "duckdb",
+      "title": "Unexpected Results when using IS NOT NULL",
+      "reported_date": "2023-11-26",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/9806"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/9806",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/9806",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0 VALUES (1);\r\n\r\n-- Here 2147483647 + 1 could be any two integers adding up larger than INT32 and causing an overflow",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:00db8a3da6cbf155953164cca244a8b9f9081658d4937e47d46788dd0162bff8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:00db8a3da6cbf155953164cca244a8b9f9081658d4937e47d46788dd0162bff8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:852a70604824",
+      "dbms": "duckdb",
+      "title": "Missing Results when using Floating-point in BETWEEN",
+      "reported_date": "2023-11-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/9825"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/9825",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/9825",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\r\nINSERT INTO t0(c1) VALUES (-1);\r\nINSERT INTO t0(c1) VALUES (5);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:26d70dc69fdc237951df0eaf4d8612a7446ab8e1cffd845efcbc900fcb768037",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:26d70dc69fdc237951df0eaf4d8612a7446ab8e1cffd845efcbc900fcb768037"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b2759e614df4",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Failed to bind column reference \"\" [1.2] (bindings: [1.1])",
+      "reported_date": "2023-12-30",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/10087"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/10087",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/10087",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\r\nINSERT INTO t0(c1) VALUES (1);\r\nCREATE VIEW v0(c0, c1, c2) AS SELECT '1', true, t0.c1 FROM t0 ORDER BY -1-2 LIMIT 2;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:f59fdf3d902c63f9ee02f3efe2e73a7a7b84577e063fda893420198675a63426",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:f59fdf3d902c63f9ee02f3efe2e73a7a7b84577e063fda893420198675a63426"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:24f183c320e4",
+      "dbms": "duckdb",
+      "title": "Unexpected result after creating index on `DATE` column",
+      "reported_date": "2024-09-10",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/13842"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/13842",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/13842",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t0(c0 DATE);\r\nINSERT INTO t0(c0) VALUES  ('1970-01-02');\r\nINSERT INTO t0(c0) VALUES ('1970-01-02');\r\nCREATE INDEX t0i0 ON t0(c0 DESC);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2db6fe78c64f4e9cc222adc378bff83223810ee5afb554507fd766a159901c0d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/13842",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2db6fe78c64f4e9cc222adc378bff83223810ee5afb554507fd766a159901c0d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:81469f903aef",
+      "dbms": "duckdb",
+      "title": "Inconsistent results when casting `BIT` and `VARBINARY`",
+      "reported_date": "2024-09-12",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/13905"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/13905",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/13905",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t1(c0 VARBINARY NOT NULL);\r\nINSERT INTO t1(c0) VALUES ('ab'), ('cd');\r\n-- INSERT INTO t1(c0) VALUES ('🦆'); -- error\r\n-- INSERT INTO t1(c0) VALUES (NULL); -- error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:42cc3b4965ac1a6c660f6606e92b558c7d8ec5a65fdf3d2082d1b964e1126260",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/13905",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:42cc3b4965ac1a6c660f6606e92b558c7d8ec5a65fdf3d2082d1b964e1126260"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8419fc38d488",
+      "dbms": "duckdb",
+      "title": "Unexpected result when comparing `STRUCT` with `INET`",
+      "reported_date": "2024-09-13",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/13924"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/13924",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/13924",
+            "source_type": "github_issue",
+            "excerpt": "-- original test case found by TLP\r\nSELECT * FROM t1 WHERE ((NULL, t1.c0, NULL)<>(t1.c1)) UNION ALL SELECT * FROM t1 WHERE (NOT ((NULL, t1.c0, NULL)<>(t1.c1))) UNION ALL SELECT * FROM t1 WHERE ((((NULL, t1.c0, NULL)<>(t1.c1))) IS NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:8c5cebf55a39c122c806ddfe8e481d9800e5b6157640a5ad72afa44d9a871466",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/13924",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8c5cebf55a39c122c806ddfe8e481d9800e5b6157640a5ad72afa44d9a871466"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:17e454259566",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Attempted to dereference unique_ptr that is NULL",
+      "reported_date": "2024-09-14",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/13938"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/13938",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/13938",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 TIMESTAMP, c1 VARCHAR[]);\r\nCREATE TABLE t1(c0 FLOAT, c1 TIMESTAMP, c2 FLOAT);\r\nINSERT INTO t0 VALUES('2023-10-10 00:00:00+00:00', NULL);\r\nINSERT INTO t0 VALUES('2025-12-25 12:00:00+02:00', []), ('2004-07-27 10:00:00+02', []);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:afc3bf62875249bf736bb6db0402d9099087c73fe17baead6e3395264d59684e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/13938",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:afc3bf62875249bf736bb6db0402d9099087c73fe17baead6e3395264d59684e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:663591ccc5b4",
+      "dbms": "duckdb",
+      "title": "Unexpected result when casting `TIMESTAMP_S` to `TIME`",
+      "reported_date": "2024-09-19",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/14026"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/14026",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/14026",
+            "source_type": "github_issue",
+            "excerpt": "-- original query found by TLP\r\nSELECT * FROM t0 WHERE (CAST(t0.c0 AS TIME)>=('12:34:56')) UNION ALL SELECT * FROM t0 WHERE (NOT (CAST(t0.c0 AS TIME)>=('12:34:56'))) UNION ALL SELECT * FROM t0 WHERE (((CAST(t0.c0 AS TIME)>=('12:34:56'))) IS NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:fd202a6a1099b1ca842c955d216f3c30f5d824e2eaea194c8a494b00ed4bc1c7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/14026",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fd202a6a1099b1ca842c955d216f3c30f5d824e2eaea194c8a494b00ed4bc1c7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:63bf0c52e6de",
+      "dbms": "duckdb",
+      "title": "Unexpected result of comparison of nested types containing `NULL`",
+      "reported_date": "2024-10-03",
+      "reported_year": 2024,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/14206"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/14206",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/14206",
+            "source_type": "github_issue",
+            "excerpt": "-- original test case found by TLP\r\nCREATE TABLE  t0(c0 VARCHAR);",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:20961af53b04872c36d3ae1fd54253dbce045fee785f860f6f9199b429d8c143",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/14206",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:20961af53b04872c36d3ae1fd54253dbce045fee785f860f6f9199b429d8c143"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:987c4043b60e",
+      "dbms": "duckdb",
+      "title": "Unexpected result when comparing `BLOB`",
+      "reported_date": "2024-10-26",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/14567"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/14567",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/14567",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t1(c0 BLOB);\r\nCREATE TABLE  t0(c0 BIT);\r\nCREATE VIEW v0(c0) AS SELECT 1 FROM t1, t0 GROUP BY t0.c0;\r\nINSERT INTO t0(c0) VALUES ( NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3ff76437ed0ade78cefd0df365fc14f5ca4773d65aef7d7e8c5e8fa2ef042bff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/14567",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3ff76437ed0ade78cefd0df365fc14f5ca4773d65aef7d7e8c5e8fa2ef042bff"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:d4f2cdd3e63f",
+      "dbms": "duckdb",
+      "title": "Unexpected result when using `INTERVAL` and `INNER JOIN` subquery",
+      "reported_date": "2024-11-14",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/14834"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/14834",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/14834",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c1 INTERVAL);\r\nINSERT INTO t0(c1) VALUES ('2 years 3 months');\r\nINSERT INTO t0(c1) VALUES ('-1734799452 DAYS'), ('2 DAYS');\r\nINSERT INTO t0(c1) VALUES ('13 days');",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:eaad4f2e548e9a0583375689ddd62ca5fb92a92345b51f97f8cfe0408668b4c5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/14834",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:eaad4f2e548e9a0583375689ddd62ca5fb92a92345b51f97f8cfe0408668b4c5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5dad36aac2d8",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Attempted to access index 0 within vector of size 0",
+      "reported_date": "2024-11-27",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/15005"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/15005",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/15005",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t1(c1 INT );\r\n\r\nSELECT * FROM t1 ORDER BY LIST_ZIP(FALSE);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e1dd297b86f61d1a030f642d771601296299ae88a80de913f45669f3f0d0900f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/15005",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e1dd297b86f61d1a030f642d771601296299ae88a80de913f45669f3f0d0900f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:bfbbc65f6068",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: ConstantFilter constant cannot be NULL - use IsNullFilter instead",
+      "reported_date": "2024-12-28",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/15479"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/15479",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/15479",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t0(c1 INT);\r\nSELECT * FROM t0 WHERE ((t0.c1 IS DISTINCT FROM NULL) OR (NOT NULL));\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e9b51647d98abde69cf29f0ad1dfc7e082cf848279f3b21f3ae6a64899d34707",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/15479",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e9b51647d98abde69cf29f0ad1dfc7e082cf848279f3b21f3ae6a64899d34707"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:74900a5aeac7",
+      "dbms": "duckdb",
+      "title": "\"Binder Error: Referenced table not found!\" occurred in the HAVING clause of subquery",
+      "reported_date": "2025-01-05",
+      "reported_year": 2025,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/15554"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/15554",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/15554",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 REAL, c1 REAL);\r\nSELECT (SELECT t0.c0 FROM t0 WHERE 1 GROUP BY t0.c0 HAVING t0.c1);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f0c879a453819aee19a809b74caeec96a15f66d30c3cc2acc70de8ffc956269b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/15554",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f0c879a453819aee19a809b74caeec96a15f66d30c3cc2acc70de8ffc956269b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:0e94c32c8c67",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Failed to bind column reference",
+      "reported_date": "2025-01-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/15586"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/15586",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/15586",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t5(c0 INT, c2 DATE);\r\nCREATE  TABLE  t0(c0 VARCHAR  );\r\nCREATE VIEW v0 AS SELECT CAST(0 AS DATE) ;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6ccff9827aca45deba69c0441ae4f36f0f4ce78df0e398c0039851b906636581",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/15586",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6ccff9827aca45deba69c0441ae4f36f0f4ce78df0e398c0039851b906636581"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:59993f383116",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Vector::Reference used on vector of different type",
+      "reported_date": "2025-01-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/15584"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/15584",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/15584",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT , c1 VARCHAR);\r\nCREATE TABLE t1( c1 INT);\r\nINSERT INTO t0 VALUES(4, 3);\r\nINSERT INTO t1 VALUES(2);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3b4ec2cda78000d4b1f531243fea601a175b115099d74746403604a984a42f12",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/15584",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3b4ec2cda78000d4b1f531243fea601a175b115099d74746403604a984a42f12"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:a6a1fe61b7a0",
+      "dbms": "duckdb",
+      "title": "Unexpected result when using `BETWEEN` and `CASE WHEN`",
+      "reported_date": "2025-01-08",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/15602"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/15602",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/15602",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t0(c0 INT);\r\nINSERT INTO t0( c0) VALUES ( -1);\r\n\r\nSELECT * FROM t0; -- -1",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b62cecfab95ad226ea735cc28f712032c66327091a5b45317d7aad715cc42a4a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/15602",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b62cecfab95ad226ea735cc28f712032c66327091a5b45317d7aad715cc42a4a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:d93812407a23",
+      "dbms": "duckdb",
+      "title": "Multiple Join Crash with Vector::Reference used on vector of different type",
+      "reported_date": "2025-02-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17335"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17335",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17335",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 BOOLEAN);\nCREATE TABLE t2(c0 DOUBLE);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:60f5ce7d2f40829d1fed354fcda7f584880cf49e8e7d77571cfda675bb132fa5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:60f5ce7d2f40829d1fed354fcda7f584880cf49e8e7d77571cfda675bb132fa5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f4fdfe564375",
+      "dbms": "duckdb",
+      "title": "Unexpected results after indexing a column with multiple rows",
+      "reported_date": "2025-02-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/16074"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/16074",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/16074",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c1 TIMESTAMP);\nINSERT INTO t0( c1) VALUES ('2020-02-29 12:00:00'), ('1969-12-09 09:26:38'), ('2020-02-29 12:00:00');\nCREATE INDEX i0 ON t0(c1 );",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4df82a04c19a8ebecb2045615df0431812654c87b19d8d2f46ef86142d0ca675",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/16074",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4df82a04c19a8ebecb2045615df0431812654c87b19d8d2f46ef86142d0ca675"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:fe95f5088e92",
+      "dbms": "duckdb",
+      "title": "Wrong join result",
+      "reported_date": "2025-02-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17338"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17338",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17338",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT);\nCREATE TABLE t2(c0 INT);\n\nINSERT INTO t2(c0) VALUES (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:234676592d54d72d4e03d04c8ebd99ec0700394edebcaa31d6cbbfa8b73d104f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:234676592d54d72d4e03d04c8ebd99ec0700394edebcaa31d6cbbfa8b73d104f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f73bbfc423a7",
+      "dbms": "duckdb",
+      "title": "Unexpected Join Internal Error",
+      "reported_date": "2025-03-08",
+      "reported_year": 2025,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18487"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18487",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18487",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 INT);\nCREATE TABLE t2(c2 DOUBLE);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d4036747a58e1b670c194bfcc68f15116fe26434f2b4da41549605ee78947b4e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d4036747a58e1b670c194bfcc68f15116fe26434f2b4da41549605ee78947b4e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c6729715e514",
+      "dbms": "duckdb",
+      "title": "Anti-join meets INTERNAL Error: Attempted to dereference unique_ptr that is NULL",
+      "reported_date": "2025-03-22",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/16783"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/16783",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/16783",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 FLOAT);\nCREATE TABLE t1(c0 FLOAT);\nselect * from t0\nwhere not exists(",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:fc0c294b7b0ced9cfc07a058e41506ccfa7e97c7dbcb242c96d47722a614f005",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:fc0c294b7b0ced9cfc07a058e41506ccfa7e97c7dbcb242c96d47722a614f005"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:29de8dd9731f",
+      "dbms": "duckdb",
+      "title": "An Anti-Join produces wrong result",
+      "reported_date": "2025-03-24",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/16803"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/16803",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/16803",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\nCREATE TABLE t1(c0 VARCHAR);\nINSERT INTO t1(c0) VALUES (NULL);\nINSERT INTO t0(c0) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d22dc68b4e01eb856ac0333b427b26efabb9219810f00c7c41caeca730de24f3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d22dc68b4e01eb856ac0333b427b26efabb9219810f00c7c41caeca730de24f3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f1bed071976d",
+      "dbms": "duckdb",
+      "title": "Unexpected result when `RIGHT JOIN` with a subquery",
+      "reported_date": "2025-03-27",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/16863"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/16863",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/16863",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c1 DATE);\nINSERT INTO t1 (c1) VALUES ('2023-10-31');\n\nSELECT t1.c1, (t1.c1 IS NULL)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:21014c3b8994bf12edd672ce82017e7b9458f73be8b658c7b3deb279921abbf0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/16863",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:21014c3b8994bf12edd672ce82017e7b9458f73be8b658c7b3deb279921abbf0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:9a0096121c70",
+      "dbms": "duckdb",
+      "title": "multiple right joins produce wrong result",
+      "reported_date": "2025-03-31",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/16901"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/16901",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/16901",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DOUBLE);\nCREATE TABLE t1(c0 DOUBLE);\nCREATE TABLE t2(c0 DOUBLE);\nINSERT INTO t0(c0) VALUES (0.0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b36d919b73d8a1ba3f297b8891fdb8a71ec7fdcdc461bd5627836aa41de8af05",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b36d919b73d8a1ba3f297b8891fdb8a71ec7fdcdc461bd5627836aa41de8af05"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5ebc4e2719ab",
+      "dbms": "duckdb",
+      "title": "Unexpected Left Join Result",
+      "reported_date": "2025-04-06",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17792"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17792",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17792",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 INT);\nCREATE TABLE t2(c0 INT);\nINSERT INTO t2(c0) VALUES (2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:bde70b3ac0185ed2d9b17cda9a040be3ba8030953d8496e90a050b68d8ef0812",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:bde70b3ac0185ed2d9b17cda9a040be3ba8030953d8496e90a050b68d8ef0812"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f23d833372f3",
+      "dbms": "duckdb",
+      "title": "Unexpected result when using `LEFT JOIN`",
+      "reported_date": "2025-04-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17042"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17042",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17042",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t2(c1 INTEGER);\nCREATE  TABLE  t0(c1 DOUBLE);\nINSERT INTO t0(c1) VALUES ( 0.1);\nINSERT INTO t2(c1) VALUES (2);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8aa85139ddaf0c87b52f571ae119578a4535480b47ab8d96cbd46c06b36cf60b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/17042",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8aa85139ddaf0c87b52f571ae119578a4535480b47ab8d96cbd46c06b36cf60b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:ae25323e37d4",
+      "dbms": "duckdb",
+      "title": "Wrong inner Join result",
+      "reported_date": "2025-04-23",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17217"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17217",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17217",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT);\nCREATE TABLE t2(c0 INT);\n\nINSERT INTO t1(c0) VALUES (-18), (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:4921ee0751c06fd38cfb6aac81dde6fb8affa4a590e3189989a1394f6e65d252",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:4921ee0751c06fd38cfb6aac81dde6fb8affa4a590e3189989a1394f6e65d252"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:244600df83c1",
+      "dbms": "duckdb",
+      "title": "Unexpected results when comparing `NULL` values",
+      "reported_date": "2025-04-25",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17257"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17257",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17257",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0( c1 INT );\nINSERT INTO t0(c1) VALUES (NULL);\n\nSELECT * FROM t0; -- NULL",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:95ddebcd0b2bd782e71e978cc3b33872fd1f5abb99686a4a15fdf68d84602e51",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/17257",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:95ddebcd0b2bd782e71e978cc3b33872fd1f5abb99686a4a15fdf68d84602e51"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8375ae66b629",
+      "dbms": "duckdb",
+      "title": "Wrong inner join result",
+      "reported_date": "2025-05-20",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17561"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17561",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17561",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 BOOL);\n\nINSERT INTO t1(c0) VALUES (false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:8e6af3dd5c26c68eb9ef60f46ad03d82779ae1a5719f72e03b4cf372a3aa6d54",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:8e6af3dd5c26c68eb9ef60f46ad03d82779ae1a5719f72e03b4cf372a3aa6d54"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:1438fccf9803",
+      "dbms": "duckdb",
+      "title": "A inner join meets INTERNAL Error: Failed to bind column reference",
+      "reported_date": "2025-05-28",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17701"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17701",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17701",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 TIMESTAMP);\nCREATE TABLE t3(c0 INT);\n\nINSERT INTO t0(c0) VALUES (DATE '1969-12-29');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cad1cf2dfc2d5b98f55926d3d05a2c5f84cafe484df24640e7b9f77be56b8303",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cad1cf2dfc2d5b98f55926d3d05a2c5f84cafe484df24640e7b9f77be56b8303"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:83866d25399e",
+      "dbms": "duckdb",
+      "title": "Unexpected internal error in multi-join query",
+      "reported_date": "2025-05-28",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17700"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17700",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17700",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN);\nCREATE TABLE t2(c1 INT);\nCREATE TABLE t3(c0 INT);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:4139beb56cfefbe247aa651b338de11ad93a861189166da9499c6f0168bd970c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:4139beb56cfefbe247aa651b338de11ad93a861189166da9499c6f0168bd970c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:e84327e83656",
+      "dbms": "duckdb",
+      "title": "Wrong multiple join result",
+      "reported_date": "2025-06-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17372"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17372",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17372",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 INT);\nCREATE TABLE t2(c0 INT);\nINSERT INTO t2(c0) VALUES (3);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:dd56e0f1e58cafd67f322f2cc7ba47b69b78e17e0c3c63423e885eff7c80a8d4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:dd56e0f1e58cafd67f322f2cc7ba47b69b78e17e0c3c63423e885eff7c80a8d4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:6f27e6fbba90",
+      "dbms": "duckdb",
+      "title": "Assertion Failure in INNER JOIN",
+      "reported_date": "2025-06-11",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/19666"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/19666",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/19666",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 BOOL);\nPRAGMA old_implicit_casting='true';",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e7223f42f9d80f3d7978a4d5e0c5fd204ee1c9f34e93bea3d33db5d4291e3a8f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e7223f42f9d80f3d7978a4d5e0c5fd204ee1c9f34e93bea3d33db5d4291e3a8f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:3ee71a742cd6",
+      "dbms": "duckdb",
+      "title": "Unexpected Execution Result",
+      "reported_date": "2025-06-11",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/19680"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/19680",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/19680",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c2 INT);\n\nINSERT INTO t0(c2) VALUES (NULL);\nINSERT INTO t0(c2) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5298bd1491ec9871e7e25955c3db489c3ca22c7cd8ae168731b5e5cdab7f4b1b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5298bd1491ec9871e7e25955c3db489c3ca22c7cd8ae168731b5e5cdab7f4b1b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5207c2b32d63",
+      "dbms": "duckdb",
+      "title": "Unexpected Join Crash",
+      "reported_date": "2025-06-21",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18000"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18000",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18000",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DATE);\nCREATE TABLE t1(c0 DATE, c1 BOOLEAN);\nSELECT * FROM t0, t1 CROSS JOIN LATERAL (SELECT t0.c0 AS col_0 WHERE t1.c1) as subQuery0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d992a5555d4b2e4c2722e275b673b3b51369b3548b787f2c00c0ddb2147e5114",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d992a5555d4b2e4c2722e275b673b3b51369b3548b787f2c00c0ddb2147e5114"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:d737ea52f354",
+      "dbms": "duckdb",
+      "title": "Strange Estimated Cardinality and Suboptimal Plan in TPC-DS Q31 Caused by Insufficient Filter Pushdown",
+      "reported_date": "2025-06-25",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "joyemang33",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18060"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18060",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18060",
+            "source_type": "github_issue",
+            "excerpt": "Strange Estimated Cardinality and Suboptimal Plan in TPC-DS Q31 Caused by Insufficient Filter Pushdown",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:40c89d9cfa93e267b19f9e1e4aeb157e4af22f936e7687c9e4092280d3f92e2d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "joyemang33 is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18060",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:40c89d9cfa93e267b19f9e1e4aeb157e4af22f936e7687c9e4092280d3f92e2d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:78aa2fe63fd7",
+      "dbms": "duckdb",
+      "title": "Unexpected Internal Error with Delim join - binding index out of range",
+      "reported_date": "2025-06-25",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18049"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18049",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18049",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 INT);\nCREATE TABLE t2(c0 INT);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a16cc9a48506d3ca8c16a92ca73dcffd7a0cd32071a8f742b338f48b49d19c9d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a16cc9a48506d3ca8c16a92ca73dcffd7a0cd32071a8f742b338f48b49d19c9d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:85d40347873c",
+      "dbms": "duckdb",
+      "title": "Unexpected result of prepared statement",
+      "reported_date": "2025-07-01",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18108"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18108",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18108",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\nINSERT INTO t0(c0) VALUES ('0.2');\nPREPARE prepare_query AS SELECT ((t0.c0)AND(?)) FROM t0;\nEXECUTE prepare_query(false);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:920252e33315a359dd9e7b985c8cc49b3dfd527323402b4bd395a840be40ae8b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18108",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:920252e33315a359dd9e7b985c8cc49b3dfd527323402b4bd395a840be40ae8b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:56f9a48818b3",
+      "dbms": "duckdb",
+      "title": "Binder Error about a non-existent table name",
+      "reported_date": "2025-07-05",
+      "reported_year": 2025,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17378"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17378",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17378",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\nCREATE TABLE t2(c0 DOUBLE);\nCREATE TABLE t3(c0 INT);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5c833451e9502f82531b00fb0b9f55bdcb5fa3c7dd8b1996848ba71a4b0c2c41",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5c833451e9502f82531b00fb0b9f55bdcb5fa3c7dd8b1996848ba71a4b0c2c41"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b43d7f800479",
+      "dbms": "duckdb",
+      "title": "Query on TPC-H Benchmark Crashes Server Due to Abnormal Memory and CPU Usage",
+      "reported_date": "2025-07-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "joyemang33",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18157"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18157",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18157",
+            "source_type": "github_issue",
+            "excerpt": "Query on TPC-H Benchmark Crashes Server Due to Abnormal Memory and CPU Usage",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1fb3fece020026807f31095d5b71264c5064e633faa70ae87c74953cf42febe2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "joyemang33 is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18157",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1fb3fece020026807f31095d5b71264c5064e633faa70ae87c74953cf42febe2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:63560f502e5e",
+      "dbms": "duckdb",
+      "title": "Wrong inner join result when handling join condition related to primary key",
+      "reported_date": "2025-07-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17380"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17380",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17380",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t2(c0 DOUBLE, PRIMARY KEY(c0));\nCREATE TABLE t1(c0 DOUBLE, PRIMARY KEY(c0));\n-- two primary key is necessary",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:dea7e92886d141dd4a848d2dd13b4ba01c89914fb615445662cb4650c9d3a1fd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:dea7e92886d141dd4a848d2dd13b4ba01c89914fb615445662cb4650c9d3a1fd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b30205570d8f",
+      "dbms": "duckdb",
+      "title": "Deterministic query return undeterministic results",
+      "reported_date": "2025-07-06",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18163"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18163",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18163",
+            "source_type": "github_issue",
+            "excerpt": "Deterministic query return undeterministic results",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e946f2f27b6cd34905c696f6a8b3a012433662a9029b116db6b8f9d5c9d756ff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18163",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e946f2f27b6cd34905c696f6a8b3a012433662a9029b116db6b8f9d5c9d756ff"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:28e0a797dc54",
+      "dbms": "duckdb",
+      "title": "Segmentation fault when JOIN with a VIEW",
+      "reported_date": "2025-07-10",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18202"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18202",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18202",
+            "source_type": "github_issue",
+            "excerpt": "Segmentation fault when JOIN with a VIEW",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:abd68144aaac484b8399eba67a9c0b63105603fec9fc5ebb80ebcf775de198c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18202",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:abd68144aaac484b8399eba67a9c0b63105603fec9fc5ebb80ebcf775de198c2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f969f973c36f",
+      "dbms": "duckdb",
+      "title": "Unexpected internal error in joins",
+      "reported_date": "2025-07-16",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18267"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18267",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18267",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 DOUBLE, c1 INT8);\nCREATE TABLE t3(c0 VARCHAR);\nINSERT INTO t1(c1) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ab125193e93117cb45d555ab272749c13bc878ab061cd82645707eb4fc0688b8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ab125193e93117cb45d555ab272749c13bc878ab061cd82645707eb4fc0688b8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:bfbbd766238e",
+      "dbms": "duckdb",
+      "title": "Inconsistent behavior between CROSS JOIN and implicit JOIN in SELECT",
+      "reported_date": "2025-07-19",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "joyemang33",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18328"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18328",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18328",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 BOOLEAN , c1 INT , PRIMARY KEY(c0));\nCREATE  TABLE  t2(c0 INT , c1 BOOLEAN , c2 BOOLEAN , c3 BOOLEAN , PRIMARY KEY(c0, c2, c1));\nCREATE  TABLE  t3(c0 INT , c1 INT , PRIMARY KEY(c0));\nINSERT INTO t3(c0, c1) VALUES (1, 1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ec1855cb453452916f2cfa4632631772ee70e864b0aef497a39d037e8895b15b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18328",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ec1855cb453452916f2cfa4632631772ee70e864b0aef497a39d037e8895b15b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:fdc75db407f6",
+      "dbms": "duckdb",
+      "title": "`HAVING FIRST` returns undeterministic result",
+      "reported_date": "2025-07-31",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18469"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18469",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18469",
+            "source_type": "github_issue",
+            "excerpt": "`HAVING FIRST` returns undeterministic result",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7c74b568a1f4c877b409cacc982a66345f166413be6602532054ddda36e7acff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18469",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7c74b568a1f4c877b409cacc982a66345f166413be6602532054ddda36e7acff"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:26995f1fbc90",
+      "dbms": "duckdb",
+      "title": "A wrong anti-join result when joining with an empty table",
+      "reported_date": "2025-08-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17417"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17417",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17417",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t2(c0 INT);\nINSERT INTO t0(c0) VALUES (5);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d1495072898c5e2f0a3262c75c058dc9b9f5babbdb98861d8708849a8664e1f4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d1495072898c5e2f0a3262c75c058dc9b9f5babbdb98861d8708849a8664e1f4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:d8c4805e07e5",
+      "dbms": "duckdb",
+      "title": "[nightly build] Incorrect handling with CTE and EXIST",
+      "reported_date": "2025-08-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "joyemang33",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18543"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18543",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18543",
+            "source_type": "github_issue",
+            "excerpt": "[nightly build] Incorrect handling with CTE and EXIST",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6c29dc0de323be6e5d1d0222e932d2fb34ff96566aacab69bbce29fd0d9141d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "joyemang33 is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18543",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6c29dc0de323be6e5d1d0222e932d2fb34ff96566aacab69bbce29fd0d9141d8"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:ef596e5d2a81",
+      "dbms": "duckdb",
+      "title": "CHECK Constraints trigger unexpected type cast",
+      "reported_date": "2025-08-11",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18569"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18569",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18569",
+            "source_type": "github_issue",
+            "excerpt": "CHECK Constraints trigger unexpected type cast",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1917f1884fb9df27bd7d73046603c516e95b0979ea030054cfe553e971feb953",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18569",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1917f1884fb9df27bd7d73046603c516e95b0979ea030054cfe553e971feb953"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f0b46fb323c4",
+      "dbms": "duckdb",
+      "title": "Inconsistenct behaviour between prepared statement and normal statement",
+      "reported_date": "2025-08-11",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18570"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18570",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18570",
+            "source_type": "github_issue",
+            "excerpt": "Inconsistenct behaviour between prepared statement and normal statement",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b72d1f62644161004a7b9f437e8b8fd5c4f19d5c4e02a627b403e18f88f4ad2d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18570",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b72d1f62644161004a7b9f437e8b8fd5c4f19d5c4e02a627b403e18f88f4ad2d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:7fa7cc466e08",
+      "dbms": "duckdb",
+      "title": "Inconsistenct behaviour on ORDER BY",
+      "reported_date": "2025-08-11",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18571"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18571",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18571",
+            "source_type": "github_issue",
+            "excerpt": "Inconsistenct behaviour on ORDER BY",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:79aecbcb2f2f068f69468b66d2580d146d678c8148060a0d033341a5cd619352",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18571",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:79aecbcb2f2f068f69468b66d2580d146d678c8148060a0d033341a5cd619352"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:e320d553ba79",
+      "dbms": "duckdb",
+      "title": "Unexpected Inner Join Result",
+      "reported_date": "2025-08-14",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18603"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18603",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18603",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 BOOLEAN);\nCREATE TABLE t1(c0 INT);\nINSERT INTO t0(c0, c1) VALUES (0, 0);\nINSERT INTO t1(c0) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:422efbca85893a7b0d476a0566ea877e66f12e1acd8173154213a3b307fd895b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:422efbca85893a7b0d476a0566ea877e66f12e1acd8173154213a3b307fd895b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:a929ea0a47f8",
+      "dbms": "duckdb",
+      "title": "[DEV BUILD] DuckDB Crash with Correlated Aggregation inside a WITH clause",
+      "reported_date": "2025-08-22",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "joyemang33",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18703"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18703",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18703",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 BOOLEAN);\n\nSELECT * FROM t0",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d0afd574d02df89f4350ea8c9fb635208e1a37fb69e9a90948206d202fe6b652",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18703",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d0afd574d02df89f4350ea8c9fb635208e1a37fb69e9a90948206d202fe6b652"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:7b929cafcee7",
+      "dbms": "duckdb",
+      "title": "Incorrect results for predicate with TRY_CAST",
+      "reported_date": "2025-09-11",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "joyemang33",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18967"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18967",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18967",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 INT , c1 BOOLEAN , PRIMARY KEY(c0));\n\nINSERT INTO t0(c0, c1) VALUES (890608529, false);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f842bc8d57f88b8b16781ec6edf4e11b044616295bf4a7c7bab3a22d2dda41cd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18967",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f842bc8d57f88b8b16781ec6edf4e11b044616295bf4a7c7bab3a22d2dda41cd"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:1610d8536c9f",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Attempted to dereference unique_ptr that is NULL",
+      "reported_date": "2025-09-12",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18971"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18971",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18971",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\n\nSELECT * FROM t0 WHERE(NOT(false = ANY([]))) ORDER\n    BY(~((SUM(true) OVER() - SUM(true) OVER())::INT)) ASC;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d08f8a091ec9aa7e24becfc6f6068d61f8a4857e13547c10f75033d637f08bd2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18971",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d08f8a091ec9aa7e24becfc6f6068d61f8a4857e13547c10f75033d637f08bd2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:70bd346af109",
+      "dbms": "duckdb",
+      "title": "Unexpected error in prepared statement",
+      "reported_date": "2025-09-13",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/18979"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/18979",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/18979",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 REAL, PRIMARY KEY(c0));\nINSERT INTO t0(c0) VALUES (1);\nPREPARE prepare_query AS SELECT (? NOT BETWEEN ((NOT t0.c0) ::BOOL) AND t0.c0) FROM t0;\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:717da22885e610df36ab62daedbddf1fd58043f3532e1908c4baf3f9b53e9782",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/18979",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:717da22885e610df36ab62daedbddf1fd58043f3532e1908c4baf3f9b53e9782"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b18c826102ec",
+      "dbms": "duckdb",
+      "title": "Inconsistency between prepared statement and normal query when negation a big integer",
+      "reported_date": "2025-09-23",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/19095"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/19095",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/19095",
+            "source_type": "github_issue",
+            "excerpt": "Inconsistency between prepared statement and normal query when negation a big integer",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2515434154549389b3fa914157fff401cab2a0558c90ccae393a5a06fae36cfb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/19095",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2515434154549389b3fa914157fff401cab2a0558c90ccae393a5a06fae36cfb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:7df271dc129c",
+      "dbms": "duckdb",
+      "title": "Unexpected error triggered by LENGTH(NULL)",
+      "reported_date": "2025-09-24",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/19128"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/19128",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/19128",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\nINSERT INTO t0(c0) VALUES (1);\nSELECT LENGTH(NULL) FROM t0 GROUP BY NULL;\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4971d776c18eaa2b4c26b3712597ce78c04215f6c3c830d3dc50dd4962ed30a7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/19128",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4971d776c18eaa2b4c26b3712597ce78c04215f6c3c830d3dc50dd4962ed30a7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:2efc0df6494b",
+      "dbms": "duckdb",
+      "title": "Should-be rejected conversions unexpectedly execute successfully in the join and cause bugs",
+      "reported_date": "2025-11-05",
+      "reported_year": 2025,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17440"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17440",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17440",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 INT);\nCREATE TABLE t2(c0 INT);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:36e6a0c9f8f8a30469e8a84b84523f2a8ed6fa2b06bd9b7edcc4dda8511fc5c9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:36e6a0c9f8f8a30469e8a84b84523f2a8ed6fa2b06bd9b7edcc4dda8511fc5c9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:14ec4fca6cce",
+      "dbms": "duckdb",
+      "title": "Unexpected Execution Result",
+      "reported_date": "2025-11-29",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/19984"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/19984",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/19984",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\n\nINSERT INTO t0(c0) VALUES (-1), (1), (1);\nCREATE VIEW v0(c0) AS SELECT t0.c0 FROM t0 GROUP BY (t0.c0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:45a4593fee523ae1e4d860243ed140455798d79ff529eb2ad4c8cb1434196034",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:45a4593fee523ae1e4d860243ed140455798d79ff529eb2ad4c8cb1434196034"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:cc2cf667f3ca",
+      "dbms": "duckdb",
+      "title": "Potential Unexpected Result when Using `TRY` Expression",
+      "reported_date": "2025-12-01",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20006"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20006",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20006",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 INT , c1 INT);\nINSERT INTO t0(c0, c1) VALUES (1819832341, -2039202523);\nINSERT INTO t0(c0) VALUES (517438920);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b6578e3fd38a636947ece89d29928e5a7a320aebc4c5d61b7d2e464b39ffe38b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20006",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b6578e3fd38a636947ece89d29928e5a7a320aebc4c5d61b7d2e464b39ffe38b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:3f0247d2cb02",
+      "dbms": "duckdb",
+      "title": "Unexpected Result when Using Utility Function `ALIAS`",
+      "reported_date": "2025-12-01",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20008"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20008",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20008",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 VARCHAR(500));\nINSERT INTO t0(c0) VALUES ('2');\n\nSELECT * FROM t0; -- 2",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5682ec2b4953f34fdcd2d0cfe87ba49dc34d78f6653b35829e378cc746232839",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20008",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5682ec2b4953f34fdcd2d0cfe87ba49dc34d78f6653b35829e378cc746232839"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:920936cd664f",
+      "dbms": "duckdb",
+      "title": "multiple joins crash with 'Execute called with a result vector of type INTEGER that does not match expression type VARCHAR'",
+      "reported_date": "2025-12-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/17446"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/17446",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/17446",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 varchar);\nCREATE TABLE t1(c0 INT);\nCREATE TABLE t2(c1 INT);\nCREATE TABLE t40(c0 varchar);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:80a16b787311081dce2b350632bfeca4c63eb47466095c694a2196423f3872e6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:80a16b787311081dce2b350632bfeca4c63eb47466095c694a2196423f3872e6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:ab15aa042664",
+      "dbms": "duckdb",
+      "title": "Unexpected result when using `POSITIONAL JOIN`",
+      "reported_date": "2025-12-08",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20086"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20086",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20086",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 INT , c1 INT);\nCREATE  TABLE  t1(c0 INT);\nINSERT INTO t1(c0) VALUES (0);\nCREATE UNIQUE INDEX t0i0 ON t0 (c0 , c1 ) ;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:634771aa91f833d7545bd48fadc816af4acd95f2c310941df3015c7fc273abf1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20086",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:634771aa91f833d7545bd48fadc816af4acd95f2c310941df3015c7fc273abf1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c35d9aa1611e",
+      "dbms": "duckdb",
+      "title": "Potential Unexpected Results of `CONCAT(NULL)`",
+      "reported_date": "2025-12-25",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20306"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20306",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20306",
+            "source_type": "github_issue",
+            "excerpt": "Potential Unexpected Results of `CONCAT(NULL)`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9bb6fedab0267d79af15af2f41016f13f9a1beb85af14cc53cd5a5daa7ba47d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20306",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9bb6fedab0267d79af15af2f41016f13f9a1beb85af14cc53cd5a5daa7ba47d8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:1640d41f04da",
+      "dbms": "duckdb",
+      "title": "Inconsistent behaviour of normal and prepared statement when use values with different types in CASE WHEN",
+      "reported_date": "2026-01-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20378"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20378",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20378",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t97(c0 DOUBLE);\nINSERT INTO t97(c0) VALUES (0.7430236362015866);\nSELECT (CASE t97.c0 WHEN t97.c0 THEN t97.c0 ELSE '' END ) FROM t97 GROUP BY '', t97.c0; -- Binder Error: Cannot mix values of type VARCHAR and DOUBLE in CASE expression - an explicit cast is required\nPREPARE prepare_query AS SELECT (CASE t97.c0 WHEN t97.c0 THEN t97.c0 ELSE ? END ) FROM t97 GROUP BY '', t97.c0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0b488a4b255196b8d8b7fa0a8b3f44f580db4c129ab06e36c01800de9b42d2a4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20378",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0b488a4b255196b8d8b7fa0a8b3f44f580db4c129ab06e36c01800de9b42d2a4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:77dab6096fd5",
+      "dbms": "duckdb",
+      "title": "Unexpected Floating point exception",
+      "reported_date": "2026-01-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20375"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20375",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20375",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Floating point exception",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:06e0d686ef3bad5f32b2d303cb8b12f3f98d824025995c141ff9a85360f4204d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20375",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:06e0d686ef3bad5f32b2d303cb8b12f3f98d824025995c141ff9a85360f4204d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:fbd985a30376",
+      "dbms": "duckdb",
+      "title": "Unexpected error in the HAVING clause of subquery",
+      "reported_date": "2026-01-27",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20688"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20688",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20688",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DOUBLE, c2 INT4);\nCREATE TABLE t1(c0 VARCHAR);\nSELECT (SELECT 1 FROM t0 GROUP BY t0.c2 HAVING t0.c0 OFFSET 1) FROM t1 GROUP BY t1.c0 OFFSET 1; \n-- Binder Error:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bd64081a8a26bba5a5c9420f8f330e29e5a929366da62453a6784f8a3981a393",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20688",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bd64081a8a26bba5a5c9420f8f330e29e5a929366da62453a6784f8a3981a393"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:7e175e13e4aa",
+      "dbms": "duckdb",
+      "title": "Less efficient query plan generated by normal SELECT than the equivalent prepared SELECT",
+      "reported_date": "2026-01-28",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20709"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20709",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20709",
+            "source_type": "github_issue",
+            "excerpt": "Less efficient query plan generated by normal SELECT than the equivalent prepared SELECT",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a404d8adc647d51fc74d68416b4dfd3625cb115d06b24869de46ae3caf3e3cf6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20709",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a404d8adc647d51fc74d68416b4dfd3625cb115d06b24869de46ae3caf3e3cf6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:a0d08b1a29a1",
+      "dbms": "duckdb",
+      "title": "inefficient query plan generated by normal SELECT",
+      "reported_date": "2026-01-28",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20713"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20713",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20713",
+            "source_type": "github_issue",
+            "excerpt": "inefficient query plan generated by normal SELECT",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9da18603d2fc512e415b464b0383abb21a9a083a3d7d27b567c5a94c0dca6c15",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20713",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9da18603d2fc512e415b464b0383abb21a9a083a3d7d27b567c5a94c0dca6c15"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:3bfa86e69d17",
+      "dbms": "duckdb",
+      "title": "Inconsistent results between equivalent normal and prepared SELECT",
+      "reported_date": "2026-02-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20792"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20792",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20792",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN);\nCREATE TABLE t1(c0 VARCHAR, c1 REAL, c2 DATE);\nINSERT INTO t1(c1, c2) VALUES (-330443717, TIMESTAMP '1969-12-28 12:12:36');\nINSERT INTO t0(c0) VALUES (682835472);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0c28f5b869aa5ea642b0ec97babb30e8a08fba3b39516a94df22f3d02e217902",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20792",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0c28f5b869aa5ea642b0ec97babb30e8a08fba3b39516a94df22f3d02e217902"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:dddb6954fc41",
+      "dbms": "duckdb",
+      "title": "Unexpected error `LIMIT/OFFSET cannot be negative` in the subquery of prepared statement",
+      "reported_date": "2026-02-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/20793"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/20793",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/20793",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT2);\nPREPARE prepare_query AS SELECT (? NOT IN (SELECT t1.c0 FROM t1 LIMIT ?));\nEXECUTE prepare_query(-675986880, 1); -- LIMIT/OFFSET cannot be negative\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c4967d1a9de21f2bb6d547110060f832303f612500455eff7b991deb880c2065",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/20793",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c4967d1a9de21f2bb6d547110060f832303f612500455eff7b991deb880c2065"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:552e02e12da2",
+      "dbms": "duckdb",
+      "title": "Floating point exception (core dumped)",
+      "reported_date": "2026-03-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/21429"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/21429",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/21429",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 BOOL);\nSELECT (((((CASE t1.rowid WHEN 0.8275004431515318 THEN t1.c0 ELSE '0.2377015493079887' END ))))) FROM t1;\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f5dbdfb1954fbc07e250e42c8b84dd182e6d3e3c274c59fb341a15b7a55738ef",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/21429",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f5dbdfb1954fbc07e250e42c8b84dd182e6d3e3c274c59fb341a15b7a55738ef"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:276ae78e572b",
+      "dbms": "duckdb",
+      "title": "Logical inconsistency: The SQL we geenenrated in the form of TLP is inconsistent with the result of the original sql execution",
+      "reported_date": "2026-04-13",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "Jasper0209",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22025"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22025",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22025",
+            "source_type": "github_issue",
+            "excerpt": "in the form of TLP is inconsistent with the result of the original sql execution",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:9bf65ef09fffbc766e69e90a9f88e73f6a07fa32990404cbabee201f7a5958d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:aa76891323505377a6bcb7353a8f9f45e34d54d5073eff2d74eed22775e2bdd0",
+          "classified_at": "2026-09-13T06:23:40Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22025",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9bf65ef09fffbc766e69e90a9f88e73f6a07fa32990404cbabee201f7a5958d8"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:duckdb:5b6c930d6eba",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Attempted to dereference unique_ptr that is NULL!",
+      "reported_date": "2026-05-19",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22764"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22764",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22764",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error: Attempted to dereference unique_ptr that is NULL!",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:aaa32dae4ec9a258c2fa1505e4ac3dbce1d14e96f46711e9ce70d530647c607a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22764",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:aaa32dae4ec9a258c2fa1505e4ac3dbce1d14e96f46711e9ce70d530647c607a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:4ad514413e00",
+      "dbms": "duckdb",
+      "title": "Internal Error: Unsupported type arrived at JSON create function",
+      "reported_date": "2026-05-19",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22762"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22762",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22762",
+            "source_type": "github_issue",
+            "excerpt": "Internal Error: Unsupported type arrived at JSON create function",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:eb81593a28f62f8e124fe5ac2077095671c10b3183598afbfb74716cd1ad4bf5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22762",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:eb81593a28f62f8e124fe5ac2077095671c10b3183598afbfb74716cd1ad4bf5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:9f166bcda7bf",
+      "dbms": "duckdb",
+      "title": "Internal error: Expected vector of type VARCHAR, but found vector of type LIST triggered by json_keys",
+      "reported_date": "2026-05-19",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22749"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22749",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22749",
+            "source_type": "github_issue",
+            "excerpt": "Internal error: Expected vector of type VARCHAR, but found vector of type LIST triggered by json_keys",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:213c0b18c71c4c0bca45e865b75926ef29bd40af6ea78eefcdb9fb3318462661",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22749",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:213c0b18c71c4c0bca45e865b75926ef29bd40af6ea78eefcdb9fb3318462661"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:45dabba861fb",
+      "dbms": "duckdb",
+      "title": "Optimizer triggers parser error on JSON",
+      "reported_date": "2026-05-19",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22761"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22761",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22761",
+            "source_type": "github_issue",
+            "excerpt": "Optimizer triggers parser error on JSON",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d273378bc6a63dd6821dce0898f0250e2b03f332fca8a78bc6191c1c2e0dace4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22761",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d273378bc6a63dd6821dce0898f0250e2b03f332fca8a78bc6191c1c2e0dace4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:3f7859c2fabe",
+      "dbms": "duckdb",
+      "title": "Unexpected Parser Error: Expected but did not find property 'cte_map' in json object:",
+      "reported_date": "2026-05-19",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22766"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22766",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22766",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Parser Error: Expected but did not find property 'cte_map' in json object:",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fcb4e6ac67ad1b7858cb254e61687638b5d580aeb74bcf4947aa8748d5aa8e87",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22766",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fcb4e6ac67ad1b7858cb254e61687638b5d580aeb74bcf4947aa8748d5aa8e87"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:3f4197d2a89a",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: 'back' called on an empty vector!",
+      "reported_date": "2026-05-20",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22774"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22774",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22774",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error: 'back' called on an empty vector!",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:241057f03e242eb3240a46ce1f1fb9aab4f2bea4d54c9d25b062e032666aacc8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22774",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:241057f03e242eb3240a46ce1f1fb9aab4f2bea4d54c9d25b062e032666aacc8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:3b1da81abb45",
+      "dbms": "duckdb",
+      "title": "Incorrect result of json_serialize_sql",
+      "reported_date": "2026-05-20",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22775"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22775",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22775",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result of json_serialize_sql",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e6468b2ccf5e950916e141e6e0b739a0f1d67b7bb9ab55b702c1b49611665365",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22775",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e6468b2ccf5e950916e141e6e0b739a0f1d67b7bb9ab55b702c1b49611665365"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f5ed17816f96",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Attempted to access index 0 within vector of size 0",
+      "reported_date": "2026-05-22",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22826"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22826",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22826",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error: Attempted to access index 0 within vector of size 0",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6eb7ee651cb79e041876b9db0a99f30bd93a477a66f48dbbfd10cbbdaed781ce",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22826",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6eb7ee651cb79e041876b9db0a99f30bd93a477a66f48dbbfd10cbbdaed781ce"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:15d69ad4695b",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Attempting to dereference an optional pointer that is not set",
+      "reported_date": "2026-05-22",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22827"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22827",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22827",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error: Attempting to dereference an optional pointer that is not set",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:32d49e6e5123aeccd85213469d0e53ff8c7def827abe7f1d9068111a43090643",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22827",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:32d49e6e5123aeccd85213469d0e53ff8c7def827abe7f1d9068111a43090643"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:880c8e4a76a4",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Calling StringValue::Get on a NULL value",
+      "reported_date": "2026-05-22",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22829"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22829",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22829",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error: Calling StringValue::Get on a NULL value",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:febd67324b602fc11a246d9ec85759330f301c598f81a673fb979326139333dd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22829",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:febd67324b602fc11a246d9ec85759330f301c598f81a673fb979326139333dd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:7529881c78d6",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Unsupported type arrived at JSON create function",
+      "reported_date": "2026-05-22",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22828"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22828",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22828",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error: Unsupported type arrived at JSON create function",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1a5c439f1611c1e3c55cce095729d9c5ceaa459b0d91d9451560fd0b352b952d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22828",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1a5c439f1611c1e3c55cce095729d9c5ceaa459b0d91d9451560fd0b352b952d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:357fb2bfcd03",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Attempted to dereference unique_ptr that is NULL!",
+      "reported_date": "2026-05-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/22938"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/22938",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/22938",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error: Attempted to dereference unique_ptr that is NULL!",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c29b51e374dbf10d5d1ef7432120def165d825127379ccc07031a096af881d61",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/22938",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c29b51e374dbf10d5d1ef7432120def165d825127379ccc07031a096af881d61"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:036abc753b1b",
+      "dbms": "duckdb",
+      "title": "COPY FROM has different behaviors with different option order",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23121"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23121",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23121",
+            "source_type": "github_issue",
+            "excerpt": "COPY FROM has different behaviors with different option order",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:45ab93d5cc37e326c63ee10d4ff4aeeb4c39622f7ec227a42969264ccf8f44ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23121",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:45ab93d5cc37e326c63ee10d4ff4aeeb4c39622f7ec227a42969264ccf8f44ae"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:381f2aa4b400",
+      "dbms": "duckdb",
+      "title": "Data type lost in PARQUET file",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23122"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23122",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23122",
+            "source_type": "github_issue",
+            "excerpt": "Data type lost in PARQUET file",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d53e0405feff758c0a4ec1f47b5620a6f1612e183d2d3fad12e3372c292d4fdf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23122",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d53e0405feff758c0a4ec1f47b5620a6f1612e183d2d3fad12e3372c292d4fdf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:22d2bf55eb0c",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Unsupported type arrived at JSON create function",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23126"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23126",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23126",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error: Unsupported type arrived at JSON create function",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3aa8adb621015320835da72b660a6d5088e0e183ec7cd91465c6e0ed76da172a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23126",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3aa8adb621015320835da72b660a6d5088e0e183ec7cd91465c6e0ed76da172a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c6b317025614",
+      "dbms": "duckdb",
+      "title": "JSON keys are case-sensitive at the top level but case-insensitive at nested levels in read_json",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23143"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23143",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23143",
+            "source_type": "github_issue",
+            "excerpt": "JSON keys are case-sensitive at the top level but case-insensitive at nested levels in read_json",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:de801b4d124acc1ed37ac3b7b87b331eaaae1ccaf4470c8186b3d69158f8adde",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23143",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:de801b4d124acc1ed37ac3b7b87b331eaaae1ccaf4470c8186b3d69158f8adde"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:093ffa3aa44a",
+      "dbms": "duckdb",
+      "title": "Unexpected error `Invalid Input Error` for COPY FROM",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23118"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23118",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23118",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected error `Invalid Input Error` for COPY FROM",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9409483458c6f63d44971dca4b97017c7868fd11698be26e721536e7fb55dc30",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23118",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9409483458c6f63d44971dca4b97017c7868fd11698be26e721536e7fb55dc30"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8c53cf5ee568",
+      "dbms": "duckdb",
+      "title": "Unexpected error `No function matches the given name and argument types 'json_type(JSON, INTEGER_LITERAL)'`",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23115"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23115",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23115",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected error `No function matches the given name and argument types 'json_type(JSON, INTEGER_LITERAL)'`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:72b7a6b72b76ac172198a568310a97c9638c342e08df31ea8ae0ac8e62cef836",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23115",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:72b7a6b72b76ac172198a568310a97c9638c342e08df31ea8ae0ac8e62cef836"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:656e22ee65a5",
+      "dbms": "duckdb",
+      "title": "Unexpected error `Unrecognized option \"ROW_GROUP_SIZE\" for csv` and for json",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23117"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23117",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23117",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected error `Unrecognized option \"ROW_GROUP_SIZE\" for csv` and for json",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fcdfd2fb7315e8b7bde0ed872ef2f74ca63966924d9faffe3706738c0395d4eb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23117",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fcdfd2fb7315e8b7bde0ed872ef2f74ca63966924d9faffe3706738c0395d4eb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:27df9b4ffccf",
+      "dbms": "duckdb",
+      "title": "Unexpected error triggered by array_to_json",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23124"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23124",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23124",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected error triggered by array_to_json",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5f4cc8e9156d06dd1c70cbfd8239a874797ffb517397d7dda0aa550b79203462",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23124",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5f4cc8e9156d06dd1c70cbfd8239a874797ffb517397d7dda0aa550b79203462"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:ce83d51f1e5c",
+      "dbms": "duckdb",
+      "title": "Unexpected result of json_deserialize_sql",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23141"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23141",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23141",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected result of json_deserialize_sql",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c82c64b5caee3cc6f622aa5e449b0ab248ccaacc581f6f1bafbf38637b2e2950",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23141",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c82c64b5caee3cc6f622aa5e449b0ab248ccaacc581f6f1bafbf38637b2e2950"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b3c439757c3a",
+      "dbms": "duckdb",
+      "title": "ignore_errors makes read_ndjson_objects accept invalid json object",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23120"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23120",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23120",
+            "source_type": "github_issue",
+            "excerpt": "ignore_errors makes read_ndjson_objects accept invalid json object",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:80cd7b3ada8be48235554d10bb570106280039827b4372fbed242fc926451706",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23120",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:80cd7b3ada8be48235554d10bb570106280039827b4372fbed242fc926451706"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:1972fb41728c",
+      "dbms": "duckdb",
+      "title": "json_object and json_group_object silently drop the NULL-key entry",
+      "reported_date": "2026-06-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23114"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23114",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23114",
+            "source_type": "github_issue",
+            "excerpt": "json_object and json_group_object silently drop the NULL-key entry",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:49665e9004c869e4b3001e84f678375024c1864b54b36c19cfff6ccdb2fe1d4e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23114",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:49665e9004c869e4b3001e84f678375024c1864b54b36c19cfff6ccdb2fe1d4e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c1f28bd64509",
+      "dbms": "duckdb",
+      "title": "Inconsistency between type cast to fixed array and unfixed array",
+      "reported_date": "2026-06-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23151"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23151",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23151",
+            "source_type": "github_issue",
+            "excerpt": "Inconsistency between type cast to fixed array and unfixed array",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:eeab5d258e33486bdaef249cf280a55d0b080cc80bf8c1de444e99647389f123",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23151",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:eeab5d258e33486bdaef249cf280a55d0b080cc80bf8c1de444e99647389f123"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:0181f69ed508",
+      "dbms": "duckdb",
+      "title": "The fullkey field incorrectly escapes keys containing special characters",
+      "reported_date": "2026-06-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23148"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23148",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23148",
+            "source_type": "github_issue",
+            "excerpt": "The fullkey field incorrectly escapes keys containing special characters",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b664c1b33c70a1897aba8595ea1fb279895b9f8c2223fa343d77a9594c10fd4d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23148",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b664c1b33c70a1897aba8595ea1fb279895b9f8c2223fa343d77a9594c10fd4d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:4b49dc7581aa",
+      "dbms": "duckdb",
+      "title": "VARIANT changes the json_type of value",
+      "reported_date": "2026-06-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23149"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23149",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23149",
+            "source_type": "github_issue",
+            "excerpt": "VARIANT changes the json_type of value",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:23332ef1237282eccfb92d69722dffee34830821bdadb2120a658c68f8051c95",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23149",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:23332ef1237282eccfb92d69722dffee34830821bdadb2120a658c68f8051c95"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:ce85a19f5234",
+      "dbms": "duckdb",
+      "title": "dateformat does not work for the top layer date value",
+      "reported_date": "2026-06-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23152"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23152",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23152",
+            "source_type": "github_issue",
+            "excerpt": "dateformat does not work for the top layer date value",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:46491ec0e6a2dd3041c980e366e8bec9e7da02ad5b8ef3856b4a97ad8c0fca90",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23152",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:46491ec0e6a2dd3041c980e366e8bec9e7da02ad5b8ef3856b4a97ad8c0fca90"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:7e30441dd725",
+      "dbms": "duckdb",
+      "title": "json_transform generate incorrect results on NUL symbol",
+      "reported_date": "2026-06-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23150"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23150",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23150",
+            "source_type": "github_issue",
+            "excerpt": "json_transform generate incorrect results on NUL symbol",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:86dbaf8e3081e6d0bd98adcc0661cc9ce43d38bb43513105c9c3223d960ced96",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23150",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:86dbaf8e3081e6d0bd98adcc0661cc9ce43d38bb43513105c9c3223d960ced96"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:d977336efb66",
+      "dbms": "duckdb",
+      "title": "Inherited named window fails to resolve earlier base window",
+      "reported_date": "2026-06-24",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23443"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23443",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23443",
+            "source_type": "github_issue",
+            "excerpt": "Inherited named window fails to resolve earlier base window",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f4392cc2878d393a7d50c7296ec3250ba682d05a2675992ea142c5e9e95af795",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23443",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f4392cc2878d393a7d50c7296ec3250ba682d05a2675992ea142c5e9e95af795"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:7c84f9f065f5",
+      "dbms": "duckdb",
+      "title": "Wrong results for DISTINCT window aggregates",
+      "reported_date": "2026-06-24",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23441"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23441",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23441",
+            "source_type": "github_issue",
+            "excerpt": "Wrong results for DISTINCT window aggregates",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fe80fbdc6fffd0ed7b2567ddeb80bac56f9bc1c51753f02a2c147472020a1677",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23441",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fe80fbdc6fffd0ed7b2567ddeb80bac56f9bc1c51753f02a2c147472020a1677"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8aa79aa75667",
+      "dbms": "duckdb",
+      "title": "Wrong result: FILL ignores ROWS frame",
+      "reported_date": "2026-06-25",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23463"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23463",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23463",
+            "source_type": "github_issue",
+            "excerpt": "Wrong result: FILL ignores ROWS frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9234b15417ab7779eea1896111a61392f4df71658d9a573b83ad552116b39db6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23463",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9234b15417ab7779eea1896111a61392f4df71658d9a573b83ad552116b39db6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5187e3b260ed",
+      "dbms": "duckdb",
+      "title": "Wrong result: FIRST_VALUE crosses partition",
+      "reported_date": "2026-06-25",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23457"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23457",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23457",
+            "source_type": "github_issue",
+            "excerpt": "Wrong result: FIRST_VALUE crosses partition",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:edb71a6a70ae864e06400cfb32d385f081e9737065338e06df9e02aad186f5ab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23457",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:edb71a6a70ae864e06400cfb32d385f081e9737065338e06df9e02aad186f5ab"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:362682cdb491",
+      "dbms": "duckdb",
+      "title": "Wrong results for LIST(DISTINCT ... ORDER BY ...) window aggregate",
+      "reported_date": "2026-06-25",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23448"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23448",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23448",
+            "source_type": "github_issue",
+            "excerpt": "Wrong results for LIST(DISTINCT ... ORDER BY ...) window aggregate",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:8c6c4bd4835e0dd546c96bc24fde1c05101f7f4a727a8f7c4e7ecca35addf26d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23448",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8c6c4bd4835e0dd546c96bc24fde1c05101f7f4a727a8f7c4e7ecca35addf26d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:be28148095a8",
+      "dbms": "duckdb",
+      "title": "COUNT(*) over a reversed ROWS frame can return a negative count",
+      "reported_date": "2026-07-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23589"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23589",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23589",
+            "source_type": "github_issue",
+            "excerpt": "COUNT(*) over a reversed ROWS frame can return a negative count",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f1d213c7ad2adc4ee3f394ae30a763092f676d16e7c106c476923bc64c67b170",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23589",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f1d213c7ad2adc4ee3f394ae30a763092f676d16e7c106c476923bc64c67b170"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:615527f8bff4",
+      "dbms": "duckdb",
+      "title": "Internal error in common_subplan for UNION of QUALIFY ROW_NUMBER EXCEPT branches",
+      "reported_date": "2026-07-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23585"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23585",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23585",
+            "source_type": "github_issue",
+            "excerpt": "Internal error in common_subplan for UNION of QUALIFY ROW_NUMBER EXCEPT branches",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e1d4e818b30d16d163fe4df12d0c50b9c82281402518ed1ce92f10acf435785e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23585",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e1d4e818b30d16d163fe4df12d0c50b9c82281402518ed1ce92f10acf435785e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:429080ef255e",
+      "dbms": "duckdb",
+      "title": "ROW_NUMBER over read_duckdb can return 0-based file-local row numbers",
+      "reported_date": "2026-07-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23586"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23586",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23586",
+            "source_type": "github_issue",
+            "excerpt": "ROW_NUMBER over read_duckdb can return 0-based file-local row numbers",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:393f746a6035a10a4534ab2b7f00c5a38f3386dbacdd4bb4623d7efa2cb2932e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23586",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:393f746a6035a10a4534ab2b7f00c5a38f3386dbacdd4bb4623d7efa2cb2932e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:46c2c8cb23a9",
+      "dbms": "duckdb",
+      "title": "top_n_window_elimination can collapse duplicate payload rows",
+      "reported_date": "2026-07-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23588"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23588",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23588",
+            "source_type": "github_issue",
+            "excerpt": "top_n_window_elimination can collapse duplicate payload rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d4606d63a43379f58b5425628b7c6764d480caeed5c38aad0fdb718f3bc764ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23588",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d4606d63a43379f58b5425628b7c6764d480caeed5c38aad0fdb718f3bc764ed"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f276296e1355",
+      "dbms": "duckdb",
+      "title": "CUME_DIST with argument ORDER BY can underflow on future ROWS frames",
+      "reported_date": "2026-07-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23641"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23641",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23641",
+            "source_type": "github_issue",
+            "excerpt": "CUME_DIST with argument ORDER BY can underflow on future ROWS frames",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:cddcdda4557074d2a46401b0704c15f413be4059dccdad422cccdcf1f60fb62c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23641",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cddcdda4557074d2a46401b0704c15f413be4059dccdad422cccdcf1f60fb62c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:2aa481827913",
+      "dbms": "duckdb",
+      "title": "Value functions reject argument ORDER BY with EXCLUDE",
+      "reported_date": "2026-07-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23640"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23640",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23640",
+            "source_type": "github_issue",
+            "excerpt": "Value functions reject argument ORDER BY with EXCLUDE",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:8f8e0602df17a229778142bb52b72f1b564aeaef99622a10117422ddcaf4b1c4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23640",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8f8e0602df17a229778142bb52b72f1b564aeaef99622a10117422ddcaf4b1c4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:a0740328e737",
+      "dbms": "duckdb",
+      "title": "filter_pushdown can turn a correlated aggregate comparison below a window into NULL rows",
+      "reported_date": "2026-07-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23639"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23639",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23639",
+            "source_type": "github_issue",
+            "excerpt": "filter_pushdown can turn a correlated aggregate comparison below a window into NULL rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:128add90f9989ee98f797b0ee656daccd02b37b6b16763354982d3198c70cd99",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23639",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:128add90f9989ee98f797b0ee656daccd02b37b6b16763354982d3198c70cd99"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:4ca05b9e8203",
+      "dbms": "duckdb",
+      "title": "BITSTRING_AGG without explicit min/max fails as a window aggregate",
+      "reported_date": "2026-07-07",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23663"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23663",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23663",
+            "source_type": "github_issue",
+            "excerpt": "BITSTRING_AGG without explicit min/max fails as a window aggregate",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:bd33e9ee3de043bdd80c06fcfec4fba26c95d25f38d80a0d2ebd3522464728c8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23663",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bd33e9ee3de043bdd80c06fcfec4fba26c95d25f38d80a0d2ebd3522464728c8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:9fbcfdd9e740",
+      "dbms": "duckdb",
+      "title": "RANGE offset 0 frames can split duplicate ORDER BY peers",
+      "reported_date": "2026-07-07",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23664"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23664",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23664",
+            "source_type": "github_issue",
+            "excerpt": "RANGE offset 0 frames can split duplicate ORDER BY peers",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a3827cf7caa67a60040c562453e26d822003b090cd5bfaa791a01208c048313a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23664",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a3827cf7caa67a60040c562453e26d822003b090cd5bfaa791a01208c048313a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:932dc9cec1fd",
+      "dbms": "duckdb",
+      "title": "top_n_window_elimination can drop rows when ROW_NUMBER orders by a nullable expression",
+      "reported_date": "2026-07-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23677"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23677",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23677",
+            "source_type": "github_issue",
+            "excerpt": "top_n_window_elimination can drop rows when ROW_NUMBER orders by a nullable expression",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:91a81e4c2c16c53883835693ab4282269fc793154130b6d4ccb375a0350b9b67",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23677",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:91a81e4c2c16c53883835693ab4282269fc793154130b6d4ccb375a0350b9b67"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:4f21c2150b48",
+      "dbms": "duckdb",
+      "title": "top_n_window_elimination can produce a NULL row for ROW_NUMBER on empty input",
+      "reported_date": "2026-07-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23675"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23675",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23675",
+            "source_type": "github_issue",
+            "excerpt": "top_n_window_elimination can produce a NULL row for ROW_NUMBER on empty input",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a82e4a8bf242830547bd71c301dda453df79cca0fa4d49b4113e37bfbee28fcd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23675",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a82e4a8bf242830547bd71c301dda453df79cca0fa4d49b4113e37bfbee28fcd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:965380b734b0",
+      "dbms": "duckdb",
+      "title": "top_n_window_elimination can rank NaN before finite values",
+      "reported_date": "2026-07-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23676"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23676",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23676",
+            "source_type": "github_issue",
+            "excerpt": "top_n_window_elimination can rank NaN before finite values",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9ed37cebe3cd347a0cbd0bb3f5006249327f0413eb96b9d6e87613481fc00c37",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23676",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9ed37cebe3cd347a0cbd0bb3f5006249327f0413eb96b9d6e87613481fc00c37"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:303827646925",
+      "dbms": "duckdb",
+      "title": "top_n_window_elimination ignores ROW_NUMBER argument ORDER BY",
+      "reported_date": "2026-07-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23678"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23678",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23678",
+            "source_type": "github_issue",
+            "excerpt": "top_n_window_elimination ignores ROW_NUMBER argument ORDER BY",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:59bdfa495f65ae3a8f374a4b0ea50d6618b22ef7cb8cb7a4e5e950ef678c7096",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23678",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:59bdfa495f65ae3a8f374a4b0ea50d6618b22ef7cb8cb7a4e5e950ef678c7096"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:eb13fe80bbe4",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Attempted to access index 0 within vector of size 0",
+      "reported_date": "2026-07-18",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23918"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23918",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23918",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t8(c0 INT);\nINSERT INTO t0 VALUES (1);\nINSERT INTO t8 VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b8ac5af2d3a859f8b129693a1662342a007c2d543462c355f259c5cfd63f429b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23918",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b8ac5af2d3a859f8b129693a1662342a007c2d543462c355f259c5cfd63f429b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:91849bff20eb",
+      "dbms": "duckdb",
+      "title": "Unexpected result when comparing `STRUCT` values in a `JOIN`",
+      "reported_date": "2026-07-19",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23927"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23927",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23927",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c2 STRUCT(a INT));\nINSERT INTO t0 VALUES ({'a': 1}), ({'a': 2}), (NULL), (NULL), ({'a': NULL});\n\nSELECT t0.c2 FROM t0 INNER JOIN t0 AS sub0 ON (t0.c2 <= sub0.c2); -- 6 rows",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a86013ded18f09b433dc0554569085b237ca71677c1e8fa033d9fd8eeb4f7cc3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23927",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a86013ded18f09b433dc0554569085b237ca71677c1e8fa033d9fd8eeb4f7cc3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:a0676a0bd101",
+      "dbms": "duckdb",
+      "title": "Unexpected result when `NOT` applied to a LIST `<` comparison",
+      "reported_date": "2026-07-20",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/23955"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/23955",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/23955",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT[]);\nCREATE TABLE t1(c1 INTERVAL);\nINSERT INTO t1(c1) VALUES (INTERVAL '23 months 5 days 40 seconds'), (INTERVAL '16 months 25 days 29 seconds'), (INTERVAL '31' SECOND);\nINSERT INTO t1(c1) VALUES (NULL), (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e664ee7cfb6ce2f5ec6db7baa7fd92e1771b0559140cfb32dcdcac78b6b158bb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/23955",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e664ee7cfb6ce2f5ec6db7baa7fd92e1771b0559140cfb32dcdcac78b6b158bb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:acee3538e135",
+      "dbms": "duckdb",
+      "title": "Crashes when `DISTINCT` over a `VARIANT` at greater than 7 rows.",
+      "reported_date": "2026-07-27",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24205"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24205",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24205",
+            "source_type": "github_issue",
+            "excerpt": "Crashes when `DISTINCT` over a `VARIANT` at greater than 7 rows.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1dfbce7308372026f3585b203b195a4a4342f7294671c734ed4d88e5e5e1e9c6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24205",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1dfbce7308372026f3585b203b195a4a4342f7294671c734ed4d88e5e5e1e9c6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:0bd36f521b63",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Failed to bind column reference -- quantified ALL/ANY subquery in a non-inner join condition over UNION ALL",
+      "reported_date": "2026-07-29",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24287"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24287",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24287",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error: Failed to bind column reference -- quantified ALL/ANY subquery in a non-inner join condition over UNION ALL",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1a8e7c3a7c258c4df40e87a8a05eda50437369c94ebcb64074cf65f676379036",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24287",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1a8e7c3a7c258c4df40e87a8a05eda50437369c94ebcb64074cf65f676379036"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:4094d33d0ec8",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: Vector::Reference used on vector of different type (BIGINT/VARCHAR) -- COALESCE over a self-joined SEMI/ANTI JOIN relation (v2.0 regression, clean on 1.5.5))",
+      "reported_date": "2026-07-29",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24288"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24288",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24288",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error: Vector::Reference used on vector of different type (BIGINT/VARCHAR) -- COALESCE over a self-joined SEMI/ANTI JOIN relation (v2.0 regression, clean on 1.5.5))",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2bd62dbed8d31128b80e6bde8592e2c2efbcb6e4a4f0df6cc29f6fb75316e1dc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24288",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2bd62dbed8d31128b80e6bde8592e2c2efbcb6e4a4f0df6cc29f6fb75316e1dc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:29ed21cd8c85",
+      "dbms": "duckdb",
+      "title": "FIRST_VALUE without window ORDER BY ignores a singleton ROWS frame",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24458"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24458",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24458",
+            "source_type": "github_issue",
+            "excerpt": "FIRST_VALUE without window ORDER BY ignores a singleton ROWS frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:23c8bd43995ff94b0c747bbc4f6064aa98d3e798ae2bd9f77f03b3e8e54a8b61",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24458",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:23c8bd43995ff94b0c747bbc4f6064aa98d3e798ae2bd9f77f03b3e8e54a8b61"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f3d620866acd",
+      "dbms": "duckdb",
+      "title": "HISTOGRAM window combine double-counts NaN map keys",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24439"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24439",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24439",
+            "source_type": "github_issue",
+            "excerpt": "HISTOGRAM window combine double-counts NaN map keys",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d968405faa72e415e66a1a8f7248e0f37e553d66b83a6ad4cbac2919dba5cf1d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24439",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d968405faa72e415e66a1a8f7248e0f37e553d66b83a6ad4cbac2919dba5cf1d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:e5ced5a74380",
+      "dbms": "duckdb",
+      "title": "HISTOGRAM_EXACT with row-dependent bins fails in a singleton window frame",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24451"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24451",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24451",
+            "source_type": "github_issue",
+            "excerpt": "HISTOGRAM_EXACT with row-dependent bins fails in a singleton window frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:67563621295c296f2d69e21f1252376695093a372ff9f28d148fd70cc1479bcd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24451",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:67563621295c296f2d69e21f1252376695093a372ff9f28d148fd70cc1479bcd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:6b71a53b8cc9",
+      "dbms": "duckdb",
+      "title": "MAD FILTER over an excluded ROWS frame can include filtered/excluded rows",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24442"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24442",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24442",
+            "source_type": "github_issue",
+            "excerpt": "MAD FILTER over an excluded ROWS frame can include filtered/excluded rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5eeb079b1eab741be1065b753057268165d94b2795aaec96c46ba48266371d2f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24442",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5eeb079b1eab741be1065b753057268165d94b2795aaec96c46ba48266371d2f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:b90ae284b74a",
+      "dbms": "duckdb",
+      "title": "MODE(DISTINCT ...) can ignore the window ORDER BY input order",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24452"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24452",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24452",
+            "source_type": "github_issue",
+            "excerpt": "MODE(DISTINCT ...) can ignore the window ORDER BY input order",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2d88c23e1b79077f7f04f20516fabaac2aa62d95846cb1ca8bec27badd373ec1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24452",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2d88c23e1b79077f7f04f20516fabaac2aa62d95846cb1ca8bec27badd373ec1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:7b4bcf48ca0c",
+      "dbms": "duckdb",
+      "title": "Row-dependent `APPROX_TOP_K` capacity is rejected by the combine window path",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24455"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24455",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24455",
+            "source_type": "github_issue",
+            "excerpt": "Row-dependent `APPROX_TOP_K` capacity is rejected by the combine window path",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a50671fa64b8ba3d5d94d747a4e2ee65575e0ff53dc5091031cb622d00f9beff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24455",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a50671fa64b8ba3d5d94d747a4e2ee65575e0ff53dc5091031cb622d00f9beff"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:411c2a573445",
+      "dbms": "duckdb",
+      "title": "Top-N MIN with row-dependent n fails in a singleton window frame",
+      "reported_date": "2026-08-03",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24449"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24449",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24449",
+            "source_type": "github_issue",
+            "excerpt": "Top-N MIN with row-dependent n fails in a singleton window frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2d069ae82212907f3f148b23d3898c71264cf64f59d35143d06f4c5aa863a7f5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24449",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2d069ae82212907f3f148b23d3898c71264cf64f59d35143d06f4c5aa863a7f5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:bd8467dd5c3a",
+      "dbms": "duckdb",
+      "title": "Segmentation fault on `RIGHT_SEMI` hash join over a window-aggregate view",
+      "reported_date": "2026-08-04",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24485"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24485",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24485",
+            "source_type": "github_issue",
+            "excerpt": "Segmentation fault on `RIGHT_SEMI` hash join over a window-aggregate view",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:88856635e13b4c1ae53bb58b7b238d9d5c1ce449ebde87852b6e45cbb7f5fe69",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24485",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:88856635e13b4c1ae53bb58b7b238d9d5c1ce449ebde87852b6e45cbb7f5fe69"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:dcd63bfe3b4a",
+      "dbms": "duckdb",
+      "title": "`INTERNAL Error: Operator occurrence N was reconstructed more than once` on `INNER JOIN` with a single-sided `ON` predicate",
+      "reported_date": "2026-08-08",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24615"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24615",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24615",
+            "source_type": "github_issue",
+            "excerpt": "`INTERNAL Error: Operator occurrence N was reconstructed more than once` on `INNER JOIN` with a single-sided `ON` predicate",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:292f6d0a19568e626a0d9175da467204a872c3efb898591c87810707409afc96",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24615",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:292f6d0a19568e626a0d9175da467204a872c3efb898591c87810707409afc96"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5276abb51a2b",
+      "dbms": "duckdb",
+      "title": "Duplicate PARTITION BY expressions can make different window partitions share results",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24629"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24629",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24629",
+            "source_type": "github_issue",
+            "excerpt": "Duplicate PARTITION BY expressions can make different window partitions share results",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2a25db840308bf7c756cab65ed7efa708e369d13ca97aac71ea53c3541818cf5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24629",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2a25db840308bf7c756cab65ed7efa708e369d13ca97aac71ea53c3541818cf5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:475f3d19ffb6",
+      "dbms": "duckdb",
+      "title": "Dynamic ROWS NULL PRECEDING bound can reuse a previous frame start",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24632"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24632",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24632",
+            "source_type": "github_issue",
+            "excerpt": "Dynamic ROWS NULL PRECEDING bound can reuse a previous frame start",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:51da6125590f37a13927a8e9591f64e12dfca07df6421d528b6ad5eb1260c70e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24632",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:51da6125590f37a13927a8e9591f64e12dfca07df6421d528b6ad5eb1260c70e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:92bb58fa9c30",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: regexp_full_match() on a non-constant pattern is not marked fallible",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24622"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24622",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24622",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 VARCHAR);\nINSERT INTO t0 VALUES ('(');\n\nSELECT regexp_full_match('abc', c1) FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:eb20d05121b9054194cf11b04f2c58c5511dd9d9c70d0a5d08467b2bb46870b4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24622",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:eb20d05121b9054194cf11b04f2c58c5511dd9d9c70d0a5d08467b2bb46870b4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:44d482056836",
+      "dbms": "duckdb",
+      "title": "LEAD/LAG ignores EXCLUDE GROUP",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24630"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24630",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24630",
+            "source_type": "github_issue",
+            "excerpt": "LEAD/LAG ignores EXCLUDE GROUP",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:8ce6538111c7217305b2fbd78e09db348741ce024554a2bc2d12c1f171c48668",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24630",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8ce6538111c7217305b2fbd78e09db348741ce024554a2bc2d12c1f171c48668"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:574a57cbd76e",
+      "dbms": "duckdb",
+      "title": "PERCENT_RANK with argument ORDER BY can split peers when the argument order is a prefix of the window order",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24628"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24628",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24628",
+            "source_type": "github_issue",
+            "excerpt": "PERCENT_RANK with argument ORDER BY can split peers when the argument order is a prefix of the window order",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6d7f2f4eb8a4b76a2c4980b6de926836f23a561e1d512ea10352b0a343b530ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24628",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6d7f2f4eb8a4b76a2c4980b6de926836f23a561e1d512ea10352b0a343b530ae"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:385fb2a10726",
+      "dbms": "duckdb",
+      "title": "RANGE PRECEDING over an INTERVAL order key can exclude the lower endpoint",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24631"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24631",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24631",
+            "source_type": "github_issue",
+            "excerpt": "RANGE PRECEDING over an INTERVAL order key can exclude the lower endpoint",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d04d705ba0d2d7caa43b864ad0f65f15ec70a859ea1a8bb16459eef78906fa3d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24631",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d04d705ba0d2d7caa43b864ad0f65f15ec70a859ea1a8bb16459eef78906fa3d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:e9b69d3bb58a",
+      "dbms": "duckdb",
+      "title": "Unexpected result of `IS NULL` filter on a RIGHT JOIN",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24623"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24623",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24623",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 BOOLEAN, c2 BOOLEAN);\nINSERT INTO t0 VALUES (false, true);\n\n-- (A) the join never matches (col0 = NOT c2 = false, and false BETWEEN",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:fcbb501b708f4b6459af52b651b501eee99a6c0f4bdd100c50a93d52a7a30404",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24623",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fcbb501b708f4b6459af52b651b501eee99a6c0f4bdd100c50a93d52a7a30404"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:3e96edf09717",
+      "dbms": "duckdb",
+      "title": "Unexpected result of `UNION ALL` over filtered LEFT JOIN branches",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24624"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24624",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24624",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t2(c0 BOOLEAN);\nCREATE TABLE t0(c1 VARCHAR);\nINSERT INTO t0 VALUES ('a');\nINSERT INTO t2 VALUES (false);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1831aa40c96a2719e92a5118288830ec76097cb96716afb55254502292e0c0a0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24624",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1831aa40c96a2719e92a5118288830ec76097cb96716afb55254502292e0c0a0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f0784d93a7f7",
+      "dbms": "duckdb",
+      "title": "`aggregate_function_rewriter` leaves a stale `GROUPING()` binding for ordered `list()`",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24627"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24627",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24627",
+            "source_type": "github_issue",
+            "excerpt": "`aggregate_function_rewriter` leaves a stale `GROUPING()` binding for ordered `list()`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3cf8c047c7b7bf5d02b62240774d970d5e535a9121fc2aa94005d64dc6ea9364",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24627",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3cf8c047c7b7bf5d02b62240774d970d5e535a9121fc2aa94005d64dc6ea9364"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:1441f6047f6a",
+      "dbms": "duckdb",
+      "title": "`last(DISTINCT x)` can return the non-distinct last value",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24625"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24625",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24625",
+            "source_type": "github_issue",
+            "excerpt": "`last(DISTINCT x)` can return the non-distinct last value",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b8ea730b8dbae3058dec8e06a397964dfe119bcd4eef47e53a093049de88b629",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24625",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b8ea730b8dbae3058dec8e06a397964dfe119bcd4eef47e53a093049de88b629"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:8d4b6021365e",
+      "dbms": "duckdb",
+      "title": "`partial_aggregate_pushdown` returns wrong holistic aggregate results over joins",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24626"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24626",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24626",
+            "source_type": "github_issue",
+            "excerpt": "`partial_aggregate_pushdown` returns wrong holistic aggregate results over joins",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:dee2573975672b16a1e8179949a3fa9a579d96dfcfd3e1fac24c48e56d8cd740",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24626",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dee2573975672b16a1e8179949a3fa9a579d96dfcfd3e1fac24c48e56d8cd740"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:9f9ad08d4284",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error when ASOF JOIN's left child is EXCEPT ALL / INTERSECT ALL",
+      "reported_date": "2026-08-10",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24672"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24672",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24672",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error when ASOF JOIN's left child is EXCEPT ALL / INTERSECT ALL",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:23b649a09faa484f61ae6c0aa6fb75f348da1ddc5ed09da70489203b8390d9c7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24672",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:23b649a09faa484f61ae6c0aa6fb75f348da1ddc5ed09da70489203b8390d9c7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:a6449e2798fc",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error on a flag-joined view and `QUALIFY`",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24711"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24711",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24711",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error on a flag-joined view and `QUALIFY`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:44e2aea0dd49a80fbfa33afb0a0179c35707646132237e43c2df7ee2298728de",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24711",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:44e2aea0dd49a80fbfa33afb0a0179c35707646132237e43c2df7ee2298728de"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:ebce7bd88a16",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error with `QUALIFY` over a view and a type cast",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24710"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24710",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24710",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error with `QUALIFY` over a view and a type cast",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9815af7afb698fdaf5f3a4e20e4c125ae006b129404364c2074118018ed82283",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24710",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9815af7afb698fdaf5f3a4e20e4c125ae006b129404364c2074118018ed82283"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5c9a3098eb0a",
+      "dbms": "duckdb",
+      "title": "`ROW_NUMBER() <= k` triggers an `arg_min`/`arg_max` error",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24708"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24708",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24708",
+            "source_type": "github_issue",
+            "excerpt": "`ROW_NUMBER() <= k` triggers an `arg_min`/`arg_max` error",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:872ffd49036a346e816a7682785283d06fcc312756152254cfc93a64f803944f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24708",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:872ffd49036a346e816a7682785283d06fcc312756152254cfc93a64f803944f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5468921479ab",
+      "dbms": "duckdb",
+      "title": "Segmentation fault on `AGE(...)` call in `GROUP BY`",
+      "reported_date": "2026-08-12",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24715"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24715",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24715",
+            "source_type": "github_issue",
+            "excerpt": "Segmentation fault on `AGE(...)` call in `GROUP BY`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:06880c9ce451f2b2e11c728e47f370fa179bcda97a95c93237dd36b5743651a8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24715",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:06880c9ce451f2b2e11c728e47f370fa179bcda97a95c93237dd36b5743651a8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:6a300d08b371",
+      "dbms": "duckdb",
+      "title": "WindowSelfJoinOptimizer returns wrong results when PARTITION BY contains duplicate keys",
+      "reported_date": "2026-08-14",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24780"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24780",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24780",
+            "source_type": "github_issue",
+            "excerpt": "WindowSelfJoinOptimizer returns wrong results when PARTITION BY contains duplicate keys",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1f0f927c8c317f272829bed9d13539a9b5c9fb9b3346d785a505d3ba1f7afca6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24780",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1f0f927c8c317f272829bed9d13539a9b5c9fb9b3346d785a505d3ba1f7afca6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:95f7f0e77f5f",
+      "dbms": "duckdb",
+      "title": "`SUBSTR` ASCII fast path clamps a hugely negative start; Unicode path returns empty",
+      "reported_date": "2026-08-14",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24766"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24766",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24766",
+            "source_type": "github_issue",
+            "excerpt": "`SUBSTR` ASCII fast path clamps a hugely negative start; Unicode path returns empty",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:844bdc3994e4066f76b75e2fefbc2d51917b8bfb3ab88d3fe3a374923d23a94a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24766",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:844bdc3994e4066f76b75e2fefbc2d51917b8bfb3ab88d3fe3a374923d23a94a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:4f2de796cdfc",
+      "dbms": "duckdb",
+      "title": "Dict-surviving hash join InternalException on a RIGHT join whose build is a nested-loop of two identity hash joins",
+      "reported_date": "2026-08-16",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24807"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24807",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24807",
+            "source_type": "github_issue",
+            "excerpt": "Dict-surviving hash join InternalException on a RIGHT join whose build is a nested-loop of two identity hash joins",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d7511d364acd082e8ded647171045893139891195b2293e701bb0509797f11fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24807",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d7511d364acd082e8ded647171045893139891195b2293e701bb0509797f11fc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:c1a5f9f4319f",
+      "dbms": "duckdb",
+      "title": "Wrong `MEDIAN` / `QUANTILE_CONT` / `LIST` over an inner join",
+      "reported_date": "2026-08-16",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24806"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24806",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24806",
+            "source_type": "github_issue",
+            "excerpt": "Wrong `MEDIAN` / `QUANTILE_CONT` / `LIST` over an inner join",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ebbf17e42398c1c3d50db7a79efae20d2e6a161d1e89d331d9818b906c83bb0b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24806",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ebbf17e42398c1c3d50db7a79efae20d2e6a161d1e89d331d9818b906c83bb0b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:5935079ca7e1",
+      "dbms": "duckdb",
+      "title": "Optimized window ordering mis-associates a volatile key with `LIST` input rows",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24830"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24830",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24830",
+            "source_type": "github_issue",
+            "excerpt": "Optimized window ordering mis-associates a volatile key with `LIST` input rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1f6b8519a1e23c393c31a141528afac727e2e38b54eba1f01f468ccb81cef979",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24830",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1f6b8519a1e23c393c31a141528afac727e2e38b54eba1f01f468ccb81cef979"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:49a606b7fc45",
+      "dbms": "duckdb",
+      "title": "Streaming `NTH_VALUE IGNORE NULLS` loses a found value",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24827"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24827",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24827",
+            "source_type": "github_issue",
+            "excerpt": "Streaming `NTH_VALUE IGNORE NULLS` loses a found value",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:bd6f1444e55efa7220a349cb059108b15d4cfd47d3c5c89c1755e6c8937b36ba",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24827",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bd6f1444e55efa7220a349cb059108b15d4cfd47d3c5c89c1755e6c8937b36ba"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f5db0f8a22a5",
+      "dbms": "duckdb",
+      "title": "`MODE(... ORDER BY ...)` can use a stale tie-break value in a sliding ROWS frame",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24826"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24826",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24826",
+            "source_type": "github_issue",
+            "excerpt": "`MODE(... ORDER BY ...)` can use a stale tie-break value in a sliding ROWS frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:55cb3453eaddd0b855cebfd3aaa7f217387de826e9e45af2e68f3e72b533b500",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24826",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:55cb3453eaddd0b855cebfd3aaa7f217387de826e9e45af2e68f3e72b533b500"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:20cd4d9f8838",
+      "dbms": "duckdb",
+      "title": "`NTILE` with argument `ORDER BY` fails on a future-only `ROWS` frame",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24831"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24831",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24831",
+            "source_type": "github_issue",
+            "excerpt": "`NTILE` with argument `ORDER BY` fails on a future-only `ROWS` frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e9952ca7a7f54c4df7c3b1d31d7980cbe9789b2d0eef9fbb418f025b04ee4b57",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24831",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e9952ca7a7f54c4df7c3b1d31d7980cbe9789b2d0eef9fbb418f025b04ee4b57"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:f6d217066b19",
+      "dbms": "duckdb",
+      "title": "`QUALIFY` volatile predicates are pushed below a window aggregate",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24829"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24829",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24829",
+            "source_type": "github_issue",
+            "excerpt": "`QUALIFY` volatile predicates are pushed below a window aggregate",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d96ffe3a2bdace612fe2a4fc536740337a1903386d876fd506bf964e2b693632",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24829",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d96ffe3a2bdace612fe2a4fc536740337a1903386d876fd506bf964e2b693632"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:ed153730c641",
+      "dbms": "duckdb",
+      "title": "`ROW_NUMBER` argument order ignores a singleton following frame",
+      "reported_date": "2026-08-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/24828"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/24828",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/24828",
+            "source_type": "github_issue",
+            "excerpt": "`ROW_NUMBER` argument order ignores a singleton following frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5ede2396d6779ecbe33a8be38cfce574d717e9778b265baf579611b8af63308a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/24828",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5ede2396d6779ecbe33a8be38cfce574d717e9778b265baf579611b8af63308a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:dfbbd931b207",
+      "dbms": "duckdb",
+      "title": "Unexpected Left Join Result",
+      "reported_date": "2026-08-27",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/25046"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/25046",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/25046",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Left Join Result",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:03d2262b24aa6a970ef0b09502859a0f0fb351a0c48f78365ae1e16a1ceb7771",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "TheoristCoder is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/25046",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:03d2262b24aa6a970ef0b09502859a0f0fb351a0c48f78365ae1e16a1ceb7771"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:fd89c673889f",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error and SIGSEGV in string functions applied to `printf('%c', 255)`",
+      "reported_date": "2026-08-31",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/25166"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/25166",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/25166",
+            "source_type": "github_issue",
+            "excerpt": "INTERNAL Error and SIGSEGV in string functions applied to `printf('%c', 255)`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9f24041b250259d7f280fa81cd144573c501491f0f2cb3d1c11d0216eb7e5d6d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/25166",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9f24041b250259d7f280fa81cd144573c501491f0f2cb3d1c11d0216eb7e5d6d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:7f1ff986357a",
+      "dbms": "duckdb",
+      "title": "INTERNAL Error: \"Failed to bind column reference\" for `unnest()` in a comma cross join",
+      "reported_date": "2026-08-31",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/25167"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/25167",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/25167",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INTEGER[]);\nINSERT INTO t0 VALUES ([1]);\n\nSELECT t0.c1 FROM t0, unnest(t0.c1) AS x(c0) WHERE CAST('x' AS BOOLEAN);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f41ee016dc4e1beaa424d74e66b212847db08c39cf89b17febbb15742c876f32",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/25167",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f41ee016dc4e1beaa424d74e66b212847db08c39cf89b17febbb15742c876f32"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:cf14da2e7c2d",
+      "dbms": "duckdb",
+      "title": "Unexpected results of `UNION ALL` over two identical RIGHT JOINs with an `EXISTS` in the ON clause",
+      "reported_date": "2026-08-31",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/25170"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/25170",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/25170",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN, c1 INT, c2 INT, PRIMARY KEY(c1));\n\nCREATE UNIQUE INDEX i0 ON t0 (c1);\nCREATE UNIQUE INDEX i1 ON t0 (c1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8dfd59f38943cf4bfbc784f15f946cce26b1a4e4574a7a3c6344f6efe74d8fe6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/25170",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8dfd59f38943cf4bfbc784f15f946cce26b1a4e4574a7a3c6344f6efe74d8fe6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:d9395f8d617b",
+      "dbms": "duckdb",
+      "title": "Unexpected results when a sampled scan of an indexed column filtered by `IN (VALUES ...)`",
+      "reported_date": "2026-08-31",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/25168"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/25168",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/25168",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 BOOLEAN);\nCREATE INDEX i0 ON t0 (c1);\nINSERT INTO t0(c1) SELECT (i % 8 = 1) FROM range(16384) g(i);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:02036e6ea397897f9c293ef82c5c07ef9dd9e77a8c7fa848ef74605ca0b49ebc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/25168",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:02036e6ea397897f9c293ef82c5c07ef9dd9e77a8c7fa848ef74605ca0b49ebc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:39e7077642e8",
+      "dbms": "duckdb",
+      "title": "Unexpected results when comparing a LIST column",
+      "reported_date": "2026-08-31",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/25169"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/25169",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/25169",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER[], c1 INT);\nINSERT INTO t0 VALUES ([], 1), ([-2, -1], 2), ([-2, -1], 3);\n\nSELECT c1 FROM t0 WHERE NOT (c0 > [0]) ORDER BY c1;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6b76d15e5c0825fadbff76a7cd45b4cfc275322d7fea37a9a4df83dc9c3cb6b0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/25169",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6b76d15e5c0825fadbff76a7cd45b4cfc275322d7fea37a9a4df83dc9c3cb6b0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:duckdb:1b02482c6e40",
+      "dbms": "duckdb",
+      "title": "Expression::HasParameter crashes planning a correlated subquery with a STRUCT-access join predicate",
+      "reported_date": "2026-09-07",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/duckdb/duckdb/issues/25430"
+      },
+      "primary_url": "https://github.com/duckdb/duckdb/issues/25430",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/duckdb/duckdb/issues/25430",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 BOOLEAN, c2 STRUCT(k0 BOOLEAN, k1 VARCHAR));\nCREATE TABLE t1(c0 INT, c1 VARCHAR(500));\nSELECT t1.* FROM t1, t0 INNER JOIN (SELECT t0.c2 AS col0 FROM t0) AS sub0 ON t0.c2['k0'] WHERE ((true)IS NOT DISTINCT FROM(t0.c1));\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:47ab45d35de1abe1aea24ab29730f73fa28e863263ef25f2c58f8b102f1519ae",
+            "retrieved_at": "2026-09-12T17:02:58Z",
+            "first_seen": "2026-09-12T17:02:58Z",
+            "last_verified": "2026-09-12T17:02:58Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-12T17:02:58Z",
+        "last_verified": "2026-09-12T17:02:58Z",
+        "source_url": "https://github.com/duckdb/duckdb/issues/25430",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:47ab45d35de1abe1aea24ab29730f73fa28e863263ef25f2c58f8b102f1519ae"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:falkordb:dda38a476313",
+      "dbms": "falkordb",
+      "title": "[Bug] Incorrect row generation with grouped aggregation after an empty Cartesian product match",
+      "reported_date": "2026-06-18",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "trrrrr617",
+      "links": {
+        "report": "https://github.com/FalkorDB/FalkorDB/issues/2136"
+      },
+      "primary_url": "https://github.com/FalkorDB/FalkorDB/issues/2136",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FalkorDB/FalkorDB/issues/2136",
+            "source_type": "github_issue",
+            "excerpt": "Note: Found via automated fuzzing with SQLancer.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b79970183be4106336f32da99da223a11847ba7523fd9bfc9c6940ae1423761f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/FalkorDB/FalkorDB/issues/2136",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b79970183be4106336f32da99da223a11847ba7523fd9bfc9c6940ae1423761f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:firebird:a9ae9542ac57",
+      "dbms": "firebird",
+      "title": "Unexpected results when using LIKE with boolean values",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/8132"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/8132",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/8132",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\r\nCREATE TABLE t1(c1 BOOLEAN);\r\nINSERT INTO t1(c1) VALUES ( false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d0f6c77dd437918e76103d6f8732c6dd8ceeed1eb33c9103fc28044b1649d40c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d0f6c77dd437918e76103d6f8732c6dd8ceeed1eb33c9103fc28044b1649d40c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:firebird:8ac2cffb4eaf",
+      "dbms": "firebird",
+      "title": "Unexpected Results when Using CASE-WHEN with LEFT JOIN",
+      "reported_date": "2023-03-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/7903"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/7903",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/7903",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\nCREATE TABLE t1(c0 INTEGER);\r\nINSERT INTO t0(c0) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9eb2e4a8f93f2cdd5e031632cf1dc70fc439e1aa66143a4ca7c66827e86e8af3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9eb2e4a8f93f2cdd5e031632cf1dc70fc439e1aa66143a4ca7c66827e86e8af3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:firebird:7a76edb6a889",
+      "dbms": "firebird",
+      "title": "Unexpected results when the join condition contains the OR predicate",
+      "reported_date": "2023-06-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/7908"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/7908",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/7908",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 BOOLEAN);\r\nCREATE TABLE t0(c0 BOOLEAN, c1 BOOLEAN);\r\nCREATE UNIQUE INDEX i0 ON t0(c1 , c0 );\r\nINSERT INTO t0 (c0) VALUES (false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:7718aecb9e9a3a8de4dc2ee17d9e9916db72c010aaf632be41f5cd16179aa354",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:7718aecb9e9a3a8de4dc2ee17d9e9916db72c010aaf632be41f5cd16179aa354"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:firebird:13a9ddce41ff",
+      "dbms": "firebird",
+      "title": "Potential bug in BETWEEN operator",
+      "reported_date": "2023-11-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/7839"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/7839",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/7839",
+            "source_type": "github_issue",
+            "excerpt": "Potential bug in BETWEEN operator",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:65b131c7f1374fa6dcb58c4608a41c5ebeba58dd064ed50466baf2828a622571",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/FirebirdSQL/firebird/issues/7839",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:65b131c7f1374fa6dcb58c4608a41c5ebeba58dd064ed50466baf2828a622571"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:firebird:ab646c451142",
+      "dbms": "firebird",
+      "title": "Crash potentially caused by BETWEEN Operator",
+      "reported_date": "2023-11-20",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/7860"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/7860",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/7860",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, PRIMARY KEY(c0)); \r\nSELECT c0 FROM t0 WHERE (1 NOT BETWEEN false AND c0);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:f9befb4c7d636770f1d0b80d700ef62a0e98cc2888968c78e7edccb6ef8fe9a6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:f9befb4c7d636770f1d0b80d700ef62a0e98cc2888968c78e7edccb6ef8fe9a6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:firebird:d947ae5eca7f",
+      "dbms": "firebird",
+      "title": "Unexpected Results when Using Natural Right Join",
+      "reported_date": "2023-11-27",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/7879"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/7879",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/7879",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT);\r\nCREATE TABLE t1(c0 INT);\r\n\r\nINSERT INTO t0(c0, c1) VALUES (1, 2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:11e9e0103326dd780323b929f8827716063873c2d8c535afd2cdba92f3a4011e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:11e9e0103326dd780323b929f8827716063873c2d8c535afd2cdba92f3a4011e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:firebird:ba45f17149c7",
+      "dbms": "firebird",
+      "title": "Unexpected results when using newline in string",
+      "reported_date": "2024-02-21",
+      "reported_year": 2024,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/8014"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/8014",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/8014",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500), PRIMARY KEY(c0)); -- PK is needed\r\nINSERT INTO t0 (c0) VALUES ('a\r\n');",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:31a367ff26e8cb3ae9e9868b2bc2228f531730c9d6e6d0981dd791d8f954db6d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/FirebirdSQL/firebird/issues/8014",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:31a367ff26e8cb3ae9e9868b2bc2228f531730c9d6e6d0981dd791d8f954db6d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:firebird:c10b2ef32702",
+      "dbms": "firebird",
+      "title": "Unexpected results when using CASE WHEN with RIGHT JOIN",
+      "reported_date": "2024-05-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/7993"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/7993",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/7993",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN);\r\nCREATE TABLE t1(c1 BOOLEAN);\r\nINSERT INTO t0 (c0) VALUES (true);\r\nINSERT INTO t1 (c1) VALUES (false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:436c60849c00c3908c0382fa6d298f816f1a8006d1b7e63aa006e2e32ef4f216",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:436c60849c00c3908c0382fa6d298f816f1a8006d1b7e63aa006e2e32ef4f216"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:firebird:8a24467d6b17",
+      "dbms": "firebird",
+      "title": "Unexpected results after creating partial index",
+      "reported_date": "2024-07-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/7995"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/7995",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/7995",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN);\r\nCREATE TABLE t1(c0 INT, c1 INTEGER);\r\nCREATE UNIQUE INDEX t1i0 ON t1(c0 ) WHERE ((t1.c0) IS NOT NULL);\r\nINSERT INTO t0 (c0) VALUES (true);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:0d9a4ae6ddf399bcd653b0d118e5f07e96be3374c27a6e5c7a03c254b6e47fa5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:0d9a4ae6ddf399bcd653b0d118e5f07e96be3374c27a6e5c7a03c254b6e47fa5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:firebird:cf2f094bb27d",
+      "dbms": "firebird",
+      "title": "Unexpected error and crash after creating a partial index",
+      "reported_date": "2024-09-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/7998"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/7998",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/7998",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500), c1 INT); \r\nCREATE UNIQUE INDEX t0i0 ON t0(c0 , c1 ) WHERE (t0.c1 BETWEEN false AND true);\r\nINSERT INTO t0(c0, c1) VALUES (1, 2);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:fa25a609de95b0ade215cd25fcadcfa26fc12eacdf6a05012c499ad0b01a52b9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:fa25a609de95b0ade215cd25fcadcfa26fc12eacdf6a05012c499ad0b01a52b9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:firebird:624b83ee3200",
+      "dbms": "firebird",
+      "title": "Unexpected results when using string concatenation for INTEGER",
+      "reported_date": "2024-09-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/FirebirdSQL/firebird/issues/7997"
+      },
+      "primary_url": "https://github.com/FirebirdSQL/firebird/issues/7997",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/FirebirdSQL/firebird/issues/7997",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER, PRIMARY KEY(c0)); -- PK is needed\r\nINSERT INTO t0 (c0) VALUES (-766027665);\r\n\r\nSELECT t0.c0 FROM t0; -- -766027665",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:34c466e3eb92b808dc394af56d2b81b2b58e8aa99316eb791c533e1420cb3246",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:34c466e3eb92b808dc394af56d2b81b2b58e8aa99316eb791c533e1420cb3246"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:6ca4f731be0e",
+      "dbms": "h2",
+      "title": "Expression NOT(c0 AND c0) unexpectedly evaluates to FALSE",
+      "reported_date": "2020-08-12",
+      "reported_year": 2020,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2807"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2807",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2807\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:7004e8ccc32a06893c5669ef4c71d418246b44de434fc767e2e8e96370ae88f5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7004e8ccc32a06893c5669ef4c71d418246b44de434fc767e2e8e96370ae88f5"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2807",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:9452399bc082",
+      "dbms": "h2",
+      "title": "Incorrect result for query and NOT (c0 != -1 AND c0) predicate",
+      "reported_date": "2020-08-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2803",
+        "fix": "https://github.com/h2database/h2database/pull/2808"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2803",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2803\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2808\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:9fd3eb158addb25a2cc48cac5d82869326c17d9e856e8dce5c2c369eba340e63",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9fd3eb158addb25a2cc48cac5d82869326c17d9e856e8dce5c2c369eba340e63"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2803",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:9ae2c5653fd3",
+      "dbms": "h2",
+      "title": "Query with an expression \"NOT NOT - C0\" causes a NullPointerException",
+      "reported_date": "2020-08-12",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2804",
+        "fix": "https://github.com/h2database/h2database/pull/2805/commits/e07127455669a116b25a099ccc8c898ea8f0700b"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2804",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2804\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2805/commits/e07127455669a116b25a099ccc8c898ea8f0700b\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a4917ff198498c52ab1f313c11eec32fb3734d9c23f5901c8e7fc1b0c5865c5a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a4917ff198498c52ab1f313c11eec32fb3734d9c23f5901c8e7fc1b0c5865c5a"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2804",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:a4ae1edbaa1a",
+      "dbms": "h2",
+      "title": "Query with CASE operator unexpectedly results in \"Column must be in the GROUP BY list\" error",
+      "reported_date": "2020-08-14",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2813",
+        "fix": "https://github.com/h2database/h2database/pull/2814/commits/47996fd10483512806f83d81281294c15865d0f3"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2813",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2813\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2814/commits/47996fd10483512806f83d81281294c15865d0f3\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:035f0a4ab36ea3360b68b3abde3a54b227efca8e664a7d4fc38d54fd24c68deb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:035f0a4ab36ea3360b68b3abde3a54b227efca8e664a7d4fc38d54fd24c68deb"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2813",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:b8580a10099f",
+      "dbms": "h2",
+      "title": "Unexpected result for query that compares an integer with a string",
+      "reported_date": "2020-08-14",
+      "reported_year": 2020,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2812"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2812",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2812\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:42deab60a97a3042d319d9cecc670b517c01459b336e0949415b08bdca08022d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:42deab60a97a3042d319d9cecc670b517c01459b336e0949415b08bdca08022d"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2812",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:37aff275b2d2",
+      "dbms": "h2",
+      "title": "Query on view that uses the BETWEEN operator results in an unexpected syntax error",
+      "reported_date": "2020-08-15",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2816",
+        "fix": "https://github.com/h2database/h2database/pull/2817/commits/909d130d662fd3be8845d6b105a936aa19c87c51"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2816",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2816\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2817/commits/909d130d662fd3be8845d6b105a936aa19c87c51\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:4d4cc5b4ed30d45d4aacea864f6a4a935dce76a3ea5f9e359e7863ea7395b833",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4d4cc5b4ed30d45d4aacea864f6a4a935dce76a3ea5f9e359e7863ea7395b833"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2816",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:157fbbdecfd6",
+      "dbms": "h2",
+      "title": "Query on view with BINARY column results in an unexpected syntax error",
+      "reported_date": "2020-08-18",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2828",
+        "fix": "https://github.com/h2database/h2database/pull/2830/commits/24c4240ff685f9d8b4f2ee33c40a9184f734e6df"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2828",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"18/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2828\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2830/commits/24c4240ff685f9d8b4f2ee33c40a9184f734e6df\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:2f8a69b407f77f645d575abc8164dc2cfebd5b365fe0f9ce42ae86eb5e03458a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2f8a69b407f77f645d575abc8164dc2cfebd5b365fe0f9ce42ae86eb5e03458a"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2828",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:9d1c10035b3d",
+      "dbms": "h2",
+      "title": "Query with % operator results in a ClassCastException",
+      "reported_date": "2020-08-18",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2825",
+        "fix": "https://github.com/h2database/h2database/pull/2827/commits/938ae0ded2a0f3b72dca4d9f7fae57f8e12206a3"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2825",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"18/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2825\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2827/commits/938ae0ded2a0f3b72dca4d9f7fae57f8e12206a3\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c6b28242b5653938646851fecbb02c96f083ab9625fec3db3ed5b261988eb51e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c6b28242b5653938646851fecbb02c96f083ab9625fec3db3ed5b261988eb51e"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2825",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:f82e2803927f",
+      "dbms": "h2",
+      "title": "Table with a generated column cycle results in a NullPointerException",
+      "reported_date": "2020-08-18",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2826",
+        "fix": "https://github.com/h2database/h2database/pull/2827/commits/8361cc902df325960ac9afaa70096cbbe1b4cf63"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2826",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"18/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2826\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2827/commits/8361cc902df325960ac9afaa70096cbbe1b4cf63\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:b38c25d799e3964956fcd085c7f394c33f69e3052f56f0ab954754c4e40e9aab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b38c25d799e3964956fcd085c7f394c33f69e3052f56f0ab954754c4e40e9aab"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2826",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:8586fcc455cf",
+      "dbms": "h2",
+      "title": "Calling math functions with a string argument results in a NullPointerException",
+      "reported_date": "2020-08-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2835",
+        "fix": "https://github.com/h2database/h2database/pull/2836/commits/49e3559ad79204b321f726db13a925d541e49985"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2835",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2835\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2836/commits/49e3559ad79204b321f726db13a925d541e49985\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:22f3863728727c359ab9dea719fd40842334a019433e1f52efc11a83db9cfcf9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:22f3863728727c359ab9dea719fd40842334a019433e1f52efc11a83db9cfcf9"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2835",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:b81432b7c614",
+      "dbms": "h2",
+      "title": "INSERT() with NULL arguments for the original string and string to be added results in NPE",
+      "reported_date": "2020-08-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2838",
+        "fix": "https://github.com/h2database/h2database/pull/2840/commits/23ca1f8d587e39cbc9e2173f9c558f3a29c9beb6"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2838",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2838\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2840/commits/23ca1f8d587e39cbc9e2173f9c558f3a29c9beb6\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:75e70dd99b85f5e606b02fd4c8451ea8d0f5bb9834fd055b91fe15075ceff9da",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:75e70dd99b85f5e606b02fd4c8451ea8d0f5bb9834fd055b91fe15075ceff9da"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2838",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:5214e8dc2810",
+      "dbms": "h2",
+      "title": "MERGE INTO causes an unexpected syntax error",
+      "reported_date": "2020-08-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2833",
+        "fix": "https://github.com/h2database/h2database/pull/2836/commits/4b2441e0629b8fea2bafca453128b7f7194625b8"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2833",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2833\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2836/commits/4b2441e0629b8fea2bafca453128b7f7194625b8\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:573b845243b4c7ff2ce3ad9cbf5480f974ca52682412735d430d68500ea58615",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:573b845243b4c7ff2ce3ad9cbf5480f974ca52682412735d430d68500ea58615"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2833",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:52bd87997b26",
+      "dbms": "h2",
+      "title": "MERGE INTO fails with an error \"Timeout trying to lock table\"",
+      "reported_date": "2020-08-19",
+      "reported_year": 2020,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2834"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2834",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2834\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:b146be12242cdb69adcf45205e7e63642d60becff1c5b559c49750d773844a78",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b146be12242cdb69adcf45205e7e63642d60becff1c5b559c49750d773844a78"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2834",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:e57ddbaf7253",
+      "dbms": "h2",
+      "title": "Querying a view that uses the POSITION() function results in an unexpected syntax error",
+      "reported_date": "2020-08-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2839",
+        "fix": "https://github.com/h2database/h2database/pull/2840/commits/40d925d33cfacc74d4f407aa947ca4e1dc43cbce"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2839",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2839\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2840/commits/40d925d33cfacc74d4f407aa947ca4e1dc43cbce\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:38022097392d6c40c595c622220166bb35885dcfb79a801b33d3f993c09821d5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:38022097392d6c40c595c622220166bb35885dcfb79a801b33d3f993c09821d5"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2839",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:2466dcf2cde0",
+      "dbms": "h2",
+      "title": "ROUND() function runs very long",
+      "reported_date": "2020-08-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2837",
+        "fix": "https://github.com/h2database/h2database/pull/2840/commits/21c86dd18621adbfdc534c6f77ed73f54ef26e68"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2837",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2837\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2840/commits/21c86dd18621adbfdc534c6f77ed73f54ef26e68\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c5530cd2098bc587ea13c47b033b9f06e145f14b05e3c05ed47c62351dc8a342",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c5530cd2098bc587ea13c47b033b9f06e145f14b05e3c05ed47c62351dc8a342"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2837",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:99f33f69bd9c",
+      "dbms": "h2",
+      "title": "Restore YEAR data type for MySQL compatibility mode",
+      "reported_date": "2020-08-19",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2831",
+        "fix": "https://github.com/h2database/h2database/pull/2836/commits/61514ae8309115bf69a34d60b6be92fd443eaab2"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2831",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2831\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2836/commits/61514ae8309115bf69a34d60b6be92fd443eaab2\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7bdc31341878a4b54604bc7e12fd9d2199a9608dfb9297469dda286b505203d7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7bdc31341878a4b54604bc7e12fd9d2199a9608dfb9297469dda286b505203d7"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2831",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:89ce7fe50cde",
+      "dbms": "h2",
+      "title": "Call to STRINGDECODE results in StringIndexOutOfBoundsException",
+      "reported_date": "2020-08-20",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2841",
+        "fix": "https://github.com/h2database/h2database/pull/2844"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2841",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2841\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2844\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5820922f0ac6d005a19a8b1cf5228915186f2a0e1c13059d45401c3ba2cf1a09",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5820922f0ac6d005a19a8b1cf5228915186f2a0e1c13059d45401c3ba2cf1a09"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2841",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:c9bd8c7a6dd9",
+      "dbms": "h2",
+      "title": "Querying view that uses LTRIM/RTRIM results in a syntax error",
+      "reported_date": "2020-08-20",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2842",
+        "fix": "https://github.com/h2database/h2database/pull/2844"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2842",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2842\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2844\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:b75178eb8410a9d536f54eef0de6a10238e7e92ca233b89423ccdb9934ead478",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b75178eb8410a9d536f54eef0de6a10238e7e92ca233b89423ccdb9934ead478"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2842",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:14213880e853",
+      "dbms": "h2",
+      "title": "CREATE TABLE with a BINARY column and large size specification results in a NegativeArraySizeException",
+      "reported_date": "2020-08-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/h2database/h2database/issues/2832",
+        "fix": "https://github.com/h2database/h2database/pull/2850"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/2832",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/08/2020\",\n        \"dbms\": \"H2\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/h2database/h2database/issues/2832\",\n            \"fix\": \"https://github.com/h2database/h2database/pull/2850\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ee8457204f90a038643b52cd995f235fbfe1affc3065f704eb341926e40b13c8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ee8457204f90a038643b52cd995f235fbfe1affc3065f704eb341926e40b13c8"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/2832",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:6b402636c298",
+      "dbms": "h2",
+      "title": "Potential issue when using `ROUND`",
+      "reported_date": "2024-01-25",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/h2database/h2database/issues/3982"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/3982",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/3982",
+            "source_type": "github_issue",
+            "excerpt": "Potential issue when using `ROUND`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7ffda49bc8a199038faf48338842a5f955b9567655d929f07905a2bed40ed5a1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/h2database/h2database/issues/3982",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7ffda49bc8a199038faf48338842a5f955b9567655d929f07905a2bed40ed5a1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:h2:089e62bb5bdb",
+      "dbms": "h2",
+      "title": "Unexpected result when using trigonomeric functions",
+      "reported_date": "2024-01-25",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/h2database/h2database/issues/3981"
+      },
+      "primary_url": "https://github.com/h2database/h2database/issues/3981",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/h2database/h2database/issues/3981",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT, PRIMARY KEY(c0));\r\nINSERT INTO t1 (c0) VALUES (1);\r\n\r\nSELECT * FROM t1; -- 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3e232d71fc58e70ca5756a63a8c7fc97cf4dce6ef986d1496c0fe0557f9abba3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3e232d71fc58e70ca5756a63a8c7fc97cf4dce6ef986d1496c0fe0557f9abba3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:hazelcast:184ceffedc48",
+      "dbms": "hazelcast",
+      "title": "SQL: Various wrong query results with string functions [HZ-1229]",
+      "reported_date": "2021-10-29",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "Fly-Style",
+      "links": {
+        "report": "https://github.com/hazelcast/hazelcast/issues/19864"
+      },
+      "primary_url": "https://github.com/hazelcast/hazelcast/issues/19864",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/hazelcast/hazelcast/issues/19864",
+            "source_type": "github_issue",
+            "excerpt": "approach, various wrong query results",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:ace59f091a14bac6a4d79dc595e0ccf2f92b0aaec0d485a67cb925771200052f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:0355300568bf9bb2f9284d89b6267db0e237e1f4e8c2852279da6b30adfcfae2",
+          "classified_at": "2026-09-13T06:23:45Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits TLP with finding the defect. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/hazelcast/hazelcast/issues/19864",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ace59f091a14bac6a4d79dc595e0ccf2f92b0aaec0d485a67cb925771200052f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:kyzo:ef7cdccc9f21",
+      "dbms": "kyzo",
+      "title": "Trial: SQLancer-style logic-bug gauntlet for Datalog (TLP/NoREC)",
+      "reported_date": "2026-07-02",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "kylejtobin",
+      "links": {
+        "report": "https://github.com/kyzobuild/kyzo/issues/29"
+      },
+      "primary_url": "https://github.com/kyzobuild/kyzo/issues/29",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/kyzobuild/kyzo/issues/29",
+            "source_type": "github_issue",
+            "excerpt": "Adapt SQLancer's logic-bug techniques (ternary logic partitioning, NoREC) to Datalog and run for published CPU-months. SQLancer found thousands of logic bugs in SQLite, DuckDB, and every major engine; the community trusts this instrument. The claim: N CPU-months, zero wrong-answer bugs surviving, plus the full ledger of defects found during development with their fixes. Can begin once eval seals.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:aad5bb2d496c7006247bf06d6b70c190910b90581962f845426b6b1994da4552",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/kyzobuild/kyzo/issues/29",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:aad5bb2d496c7006247bf06d6b70c190910b90581962f845426b6b1994da4552"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:mariadb:d727ed32eeab",
+      "dbms": "mariadb",
+      "title": "https://jira.mariadb.org/browse/MDEV-32076",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "report": "https://jira.mariadb.org/browse/MDEV-32076"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://jira.mariadb.org/browse/MDEV-32076",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mariadb provider, as the field bug32076 that works around it.",
+            "content_sha256": "sha256:d727ed32eeab48214e447f5cf46163dff79c32ff166ea671905c00b41d1d467c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:d727ed32eeab48214e447f5cf46163dff79c32ff166ea671905c00b41d1d467c"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-32076",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mariadb:737602e5eaa0",
+      "dbms": "mariadb",
+      "title": "https://jira.mariadb.org/browse/MDEV-32099",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "report": "https://jira.mariadb.org/browse/MDEV-32099"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://jira.mariadb.org/browse/MDEV-32099",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mariadb provider, as the field bug32099 that works around it.",
+            "content_sha256": "sha256:737602e5eaa09714f56078d6635f3eed63f0ce7967668c18d4ba7dfe80a5d178",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:737602e5eaa09714f56078d6635f3eed63f0ce7967668c18d4ba7dfe80a5d178"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-32099",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mariadb:651c7bfaefa9",
+      "dbms": "mariadb",
+      "title": "https://jira.mariadb.org/browse/MDEV-32105",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "report": "https://jira.mariadb.org/browse/MDEV-32105"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://jira.mariadb.org/browse/MDEV-32105",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mariadb provider, as the field bug32105 that works around it.",
+            "content_sha256": "sha256:651c7bfaefa9b7876611415135872ddfa3fa5f09ab4027cc3c09875c166ea5f8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:651c7bfaefa9b7876611415135872ddfa3fa5f09ab4027cc3c09875c166ea5f8"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-32105",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mariadb:ccdb8bbb9ee5",
+      "dbms": "mariadb",
+      "title": "https://jira.mariadb.org/browse/MDEV-32106",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "report": "https://jira.mariadb.org/browse/MDEV-32106"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://jira.mariadb.org/browse/MDEV-32106",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mariadb provider, as the field bug32106 that works around it.",
+            "content_sha256": "sha256:ccdb8bbb9ee577cfef1b71671134536fc6931fb6d8d5a81dabc88482383caac1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:ccdb8bbb9ee577cfef1b71671134536fc6931fb6d8d5a81dabc88482383caac1"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-32106",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mariadb:17087e1961a6",
+      "dbms": "mariadb",
+      "title": "https://jira.mariadb.org/browse/MDEV-32107",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "report": "https://jira.mariadb.org/browse/MDEV-32107"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://jira.mariadb.org/browse/MDEV-32107",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mariadb provider, as the field bug32107 that works around it.",
+            "content_sha256": "sha256:17087e1961a631592c6a549fafcc620c3645adb61d084bb679e5b7941fedf592",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:17087e1961a631592c6a549fafcc620c3645adb61d084bb679e5b7941fedf592"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-32107",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mariadb:04f17acd1633",
+      "dbms": "mariadb",
+      "title": "https://jira.mariadb.org/browse/MDEV-32108",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "report": "https://jira.mariadb.org/browse/MDEV-32108"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://jira.mariadb.org/browse/MDEV-32108",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mariadb provider, as the field bug32108 that works around it.",
+            "content_sha256": "sha256:04f17acd1633187784c1197d1680c6e1083b0b6d46225d7d327d4a8a6aa54a5d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:04f17acd1633187784c1197d1680c6e1083b0b6d46225d7d327d4a8a6aa54a5d"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-32108",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mariadb:74b33236fcf4",
+      "dbms": "mariadb",
+      "title": "https://jira.mariadb.org/browse/MDEV-32143",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "report": "https://jira.mariadb.org/browse/MDEV-32143"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://jira.mariadb.org/browse/MDEV-32143",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mariadb provider, as the field bug32143 that works around it.",
+            "content_sha256": "sha256:74b33236fcf40f8a386a8284541a8f85313f4fdc6831bf2ac91c6975b0db508c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:74b33236fcf40f8a386a8284541a8f85313f4fdc6831bf2ac91c6975b0db508c"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-32143",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mariadb:f0b119d2aa10",
+      "dbms": "mariadb",
+      "title": "https://jira.mariadb.org/browse/MDEV-33893",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "report": "https://jira.mariadb.org/browse/MDEV-33893"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://jira.mariadb.org/browse/MDEV-33893",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mariadb provider, as the field bug33893 that works around it.",
+            "content_sha256": "sha256:f0b119d2aa10d02357e76c9d39019ede75c51bf0fde98e30f8718840b82d308b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T12:38:22Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mariadb/MariaDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:f0b119d2aa10d02357e76c9d39019ede75c51bf0fde98e30f8718840b82d308b"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-33893",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mariadb:86acc66c9560",
+      "dbms": "mariadb",
+      "title": "GREATEST() and LEAST() malfunction for NULL",
+      "reported_date": "2019-11-11",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://jira.mariadb.org/browse/MDEV-21034"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-21034",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/11/2019\",\n        \"dbms\": \"MariaDB\",\n        \"links\": {\n            \"bugtracker\": \"https://jira.mariadb.org/browse/MDEV-21034\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:a891b2f702a2459ed59a6d645aaf4f27836313cb2dda1a3405cf7eb7c5683b55",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a891b2f702a2459ed59a6d645aaf4f27836313cb2dda1a3405cf7eb7c5683b55"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-21034",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mariadb:f0de2fde3d5a",
+      "dbms": "mariadb",
+      "title": "Incorrect result for expression with the <=> operator and IS NULL",
+      "reported_date": "2019-11-11",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://jira.mariadb.org/browse/MDEV-21029"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-21029",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/11/2019\",\n        \"dbms\": \"MariaDB\",\n        \"links\": {\n            \"bugtracker\": \"https://jira.mariadb.org/browse/MDEV-21029\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:a9fcc184b89a9c8b515d73f5a3b60bc94b43ec9723a486b2d697cbe7e0a863ad",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a9fcc184b89a9c8b515d73f5a3b60bc94b43ec9723a486b2d697cbe7e0a863ad"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-21029",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mariadb:3cbe71c1da9f",
+      "dbms": "mariadb",
+      "title": "Index causes incorrect result when comparing float-point number with INT",
+      "reported_date": "2019-11-11",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://jira.mariadb.org/browse/MDEV-21032"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-21032",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/11/2019\",\n        \"dbms\": \"MariaDB\",\n        \"links\": {\n            \"bugtracker\": \"https://jira.mariadb.org/browse/MDEV-21032\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:37865ad37e3e9f1eaffaf9575fe12dc98cc072307a7f1f8c99705c75a8117aac",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:37865ad37e3e9f1eaffaf9575fe12dc98cc072307a7f1f8c99705c75a8117aac"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-21032",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mariadb:f4622310a47e",
+      "dbms": "mariadb",
+      "title": "CREATE TABLE with generated column and RLIKE results in segfault",
+      "reported_date": "2019-11-14",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://jira.mariadb.org/browse/MDEV-21058"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-21058",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/11/2019\",\n        \"dbms\": \"MariaDB\",\n        \"links\": {\n            \"bugtracker\": \"https://jira.mariadb.org/browse/MDEV-21058\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:fd27f43a5b0e5331d5e2c8950b4364bf695bcd365fd4c2c48bb1b8f7151725b7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:fd27f43a5b0e5331d5e2c8950b4364bf695bcd365fd4c2c48bb1b8f7151725b7"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-21058",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mariadb:5c6029ff67f8",
+      "dbms": "mariadb",
+      "title": "CREATE UNIQUE INDEX USING HASH malfunctions for engine=Aria table",
+      "reported_date": "2019-11-14",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://jira.mariadb.org/browse/MDEV-21057"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-21057",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/11/2019\",\n        \"dbms\": \"MariaDB\",\n        \"links\": {\n            \"bugtracker\": \"https://jira.mariadb.org/browse/MDEV-21057\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a1c93c160384e0dd4c9535f782f70011f79864753de9f9db31b58889d7ba0639",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a1c93c160384e0dd4c9535f782f70011f79864753de9f9db31b58889d7ba0639"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-21057",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mariadb:4e86540aad43",
+      "dbms": "mariadb",
+      "title": "UNIQUE constraint causes a query with string comparison to omit a row in the result set",
+      "reported_date": "2019-11-15",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://jira.mariadb.org/browse/MDEV-21065",
+        "fix": "https://github.com/mariadb/server/commit/fc860d3fa3bc854fbe6aab9179d7a4aaf6eb9edf"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-21065",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/11/2019\",\n        \"dbms\": \"MariaDB\",\n        \"links\": {\n            \"bugtracker\": \"https://jira.mariadb.org/browse/MDEV-21065\",\n            \"fix\": \"https://github.com/mariadb/server/commit/fc860d3fa3bc854fbe6aab9179d7a4aaf6eb9edf\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:cafbe3a80b25e1e2b882ceea3bd4cfc4bf79248c8027e9c119f22897ec0e461e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:cafbe3a80b25e1e2b882ceea3bd4cfc4bf79248c8027e9c119f22897ec0e461e"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-21065",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mariadb:76537c10f456",
+      "dbms": "mariadb",
+      "title": "NOT NULL and UNIQUE constraints cause SUM() to yield an incorrect result",
+      "reported_date": "2019-11-18",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://jira.mariadb.org/browse/MDEV-21076"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-21076",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"18/11/2019\",\n        \"dbms\": \"MariaDB\",\n        \"links\": {\n            \"bugtracker\": \"https://jira.mariadb.org/browse/MDEV-21076\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:75b5ba8de821775c1d59f6cbc14da8e41fa1ca2b87ef746a8eef92b848689098",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:75b5ba8de821775c1d59f6cbc14da8e41fa1ca2b87ef746a8eef92b848689098"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-21076",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mariadb:644b88f0dd0b",
+      "dbms": "mariadb",
+      "title": "The phenomenon of inconsistent query results caused by null data values",
+      "reported_date": "2023-04-07",
+      "reported_year": 2023,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "Chenglin Tian",
+      "links": {
+        "report": "https://jira.mariadb.org/browse/MDEV-31023"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-31023",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-31023",
+            "source_type": "issue_tracker",
+            "excerpt": "Subsequently, we used SQLancer's TLP method for validation and found that the two queries that were supposed to be equivalent had inconsistent results：",
+            "excerpt_is_verbatim": true,
+            "note": "MDEV-31023 in MariaDB's own Jira, filed as a Bug.",
+            "content_sha256": "sha256:1ea5c599dc6c40ac55451d3714e0f96251e5cb6a97519c92be9278f59a435a6a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-09T01:23:11Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "mariadb_jira",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-09T01:23:11Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://jira.mariadb.org/browse/MDEV-31023",
+        "source_type": "issue_tracker",
+        "content_sha256": "sha256:1ea5c599dc6c40ac55451d3714e0f96251e5cb6a97519c92be9278f59a435a6a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:mariadb:d655feda8bb4",
+      "dbms": "mariadb",
+      "title": "OPTIMIZE TABLE t0 NOWAIT results in Got error -2 \"Internal error < 0 (Not system error)\" from storage engine InnoDB",
+      "reported_date": "2024-04-11",
+      "reported_year": 2024,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Tan Ming Wei",
+      "links": {
+        "report": "https://jira.mariadb.org/browse/MDEV-33882"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-33882",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-33882",
+            "source_type": "issue_tracker",
+            "excerpt": "Bug found with SQLancer.",
+            "excerpt_is_verbatim": true,
+            "note": "MDEV-33882 in MariaDB's own Jira, filed as a Bug.",
+            "content_sha256": "sha256:1360c185bb4e356cd125c501ec97b409769fb8dffd57c07967998734782176a5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-09T01:23:11Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "mariadb_jira",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-09T01:23:11Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://jira.mariadb.org/browse/MDEV-33882",
+        "source_type": "issue_tracker",
+        "content_sha256": "sha256:1360c185bb4e356cd125c501ec97b409769fb8dffd57c07967998734782176a5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:mariadb:eb3dfd17345b",
+      "dbms": "mariadb",
+      "title": "NoREC logical bug: unoptimized evaluation of REGEXP_REPLACE in the WHERE clause produces inconsistent counts when compared with transformed boolean aggregation.",
+      "reported_date": "2026-04-01",
+      "reported_year": 2026,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "Jasper Andrew",
+      "links": {
+        "report": "https://jira.mariadb.org/browse/MDEV-39237"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-39237",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-39237",
+            "source_type": "issue_tracker",
+            "excerpt": "NoREC logical bug: unoptimized evaluation of REGEXP_REPLACE in the WHERE clause produces inconsistent counts when compared with transformed boolean aggregation.",
+            "excerpt_is_verbatim": true,
+            "note": "MDEV-39237 in MariaDB's own Jira, filed as a Bug.",
+            "content_sha256": "sha256:957db88c8c396f3ce5d2e103304993915c62ffa460e2076566561f0b64c0072a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-09T01:23:11Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "mariadb_jira",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-09T01:23:11Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://jira.mariadb.org/browse/MDEV-39237",
+        "source_type": "issue_tracker",
+        "content_sha256": "sha256:957db88c8c396f3ce5d2e103304993915c62ffa460e2076566561f0b64c0072a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:mariadb:8eea440b8035",
+      "dbms": "mariadb",
+      "title": "NoREC logical bug: unoptimized evaluation of REGEXP_SUBSTR in the WHERE clause produces inconsistent counts when compared with transformed boolean aggregation.",
+      "reported_date": "2026-04-01",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "Jasper Andrew",
+      "links": {
+        "report": "https://jira.mariadb.org/browse/MDEV-39238"
+      },
+      "primary_url": "https://jira.mariadb.org/browse/MDEV-39238",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://jira.mariadb.org/browse/MDEV-39238",
+            "source_type": "issue_tracker",
+            "excerpt": "NoREC logical bug: unoptimized evaluation of REGEXP_SUBSTR in the WHERE clause produces inconsistent counts when compared with transformed boolean aggregation.",
+            "excerpt_is_verbatim": true,
+            "note": "MDEV-39238 in MariaDB's own Jira, filed as a Bug.",
+            "content_sha256": "sha256:a44c61b2b3c01cd822c6a47cc566eab9cbd0a7fc8e677e133327b3562862846c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-09T01:23:11Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "mariadb_jira",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-09T01:23:11Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://jira.mariadb.org/browse/MDEV-39238",
+        "source_type": "issue_tracker",
+        "content_sha256": "sha256:a44c61b2b3c01cd822c6a47cc566eab9cbd0a7fc8e677e133327b3562862846c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:matrixone:c825f4e7a359",
+      "dbms": "matrixone",
+      "title": "[Bug]: [Sqlancer] show tables from database failed when do not use database",
+      "reported_date": "2022-01-10",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "iamlinjunhong",
+      "links": {
+        "report": "https://github.com/matrixorigin/matrixone/issues/1589"
+      },
+      "primary_url": "https://github.com/matrixorigin/matrixone/issues/1589",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/matrixorigin/matrixone/issues/1589",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]: [Sqlancer] show tables from database failed when do not use database",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:222326b36f5636c243ed770c7855e6fbb730250b95ff7981a78374b2d62aa6ea",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T15:56:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T15:56:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/matrixorigin/matrixone/issues/1589",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:222326b36f5636c243ed770c7855e6fbb730250b95ff7981a78374b2d62aa6ea"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:matrixone:c6f9a9ae60b2",
+      "dbms": "matrixone",
+      "title": "[Bug]: [Sqlancer] execute the query of select failed",
+      "reported_date": "2022-01-17",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "iamlinjunhong",
+      "links": {
+        "report": "https://github.com/matrixorigin/matrixone/issues/1617"
+      },
+      "primary_url": "https://github.com/matrixorigin/matrixone/issues/1617",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/matrixorigin/matrixone/issues/1617",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]: [Sqlancer] execute the query of select failed",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b746284f5348378263610ebd1285200e566f68bd3f59e8e74eaae9d86017ff03",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T15:56:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T15:56:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/matrixorigin/matrixone/issues/1617",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b746284f5348378263610ebd1285200e566f68bd3f59e8e74eaae9d86017ff03"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:matrixone:cd9907015e77",
+      "dbms": "matrixone",
+      "title": "[Bug]: [sqlancer] error expression: not (not value)",
+      "reported_date": "2022-01-26",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "iamlinjunhong",
+      "links": {
+        "report": "https://github.com/matrixorigin/matrixone/issues/1641"
+      },
+      "primary_url": "https://github.com/matrixorigin/matrixone/issues/1641",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/matrixorigin/matrixone/issues/1641",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]: [sqlancer] error expression: not (not value)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bff07e619b1984772da6bfdb01a363b00db42cd5296345189d12c9a8aae86bdd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T15:56:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T15:56:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/matrixorigin/matrixone/issues/1641",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bff07e619b1984772da6bfdb01a363b00db42cd5296345189d12c9a8aae86bdd"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:matrixone:e47cfb33c39d",
+      "dbms": "matrixone",
+      "title": "[Bug]: [sqlancer] expression: - column",
+      "reported_date": "2022-01-27",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "iamlinjunhong",
+      "links": {
+        "report": "https://github.com/matrixorigin/matrixone/issues/1651"
+      },
+      "primary_url": "https://github.com/matrixorigin/matrixone/issues/1651",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/matrixorigin/matrixone/issues/1651",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]: [sqlancer] expression: - column",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:26dbde004c7d7154e7aa14e654d5d81f61142f7b221e9030f9a07ea1e0d1772f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T15:56:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T15:56:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/matrixorigin/matrixone/issues/1651",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:26dbde004c7d7154e7aa14e654d5d81f61142f7b221e9030f9a07ea1e0d1772f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:matrixone:d5d540d9e979",
+      "dbms": "matrixone",
+      "title": "[Bug]: [sqlancer] expression: not (-0)",
+      "reported_date": "2022-01-28",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "iamlinjunhong",
+      "links": {
+        "report": "https://github.com/matrixorigin/matrixone/issues/1653"
+      },
+      "primary_url": "https://github.com/matrixorigin/matrixone/issues/1653",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/matrixorigin/matrixone/issues/1653",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]: [sqlancer] expression: not (-0)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1bfcf56d1a4eded783e73c7c25723054a3b8a893e950b618285e5d35246b6f5a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T15:56:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T15:56:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/matrixorigin/matrixone/issues/1653",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1bfcf56d1a4eded783e73c7c25723054a3b8a893e950b618285e5d35246b6f5a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:matrixone:6272c2696d9d",
+      "dbms": "matrixone",
+      "title": "[Bug]:[sqlancer] expression cause mo-server crash",
+      "reported_date": "2022-01-28",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "iamlinjunhong",
+      "links": {
+        "report": "https://github.com/matrixorigin/matrixone/issues/1654"
+      },
+      "primary_url": "https://github.com/matrixorigin/matrixone/issues/1654",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/matrixorigin/matrixone/issues/1654",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]:[sqlancer] expression cause mo-server crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5151d68dea50a2322a441191fb186bc2637b15a3625c697a635ec6df5a276b9a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T15:56:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T15:56:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/matrixorigin/matrixone/issues/1654",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5151d68dea50a2322a441191fb186bc2637b15a3625c697a635ec6df5a276b9a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:matrixone:6af272273558",
+      "dbms": "matrixone",
+      "title": "[Bug]: [sqlancer] expression crash: -column",
+      "reported_date": "2022-02-07",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "iamlinjunhong",
+      "links": {
+        "report": "https://github.com/matrixorigin/matrixone/issues/1660"
+      },
+      "primary_url": "https://github.com/matrixorigin/matrixone/issues/1660",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/matrixorigin/matrixone/issues/1660",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]: [sqlancer] expression crash: -column",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bf8a929e1ff45fa8a80423aa3f6a97b988a63771eec005284a854a1ecea41ca4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T15:56:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T15:56:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/matrixorigin/matrixone/issues/1660",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bf8a929e1ff45fa8a80423aa3f6a97b988a63771eec005284a854a1ecea41ca4"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:matrixone:d00b9b39d674",
+      "dbms": "matrixone",
+      "title": "[Bug]: [sqlancer] expression: (- -number)",
+      "reported_date": "2022-02-07",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "iamlinjunhong",
+      "links": {
+        "report": "https://github.com/matrixorigin/matrixone/issues/1659"
+      },
+      "primary_url": "https://github.com/matrixorigin/matrixone/issues/1659",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/matrixorigin/matrixone/issues/1659",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]: [sqlancer] expression: (- -number)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:03bee2fc874e468ad36b1ae9bf7522650ce0339f172dce4f83e394422c16f080",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T15:56:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T15:56:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/matrixorigin/matrixone/issues/1659",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:03bee2fc874e468ad36b1ae9bf7522650ce0339f172dce4f83e394422c16f080"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:matrixone:a4c8e481e290",
+      "dbms": "matrixone",
+      "title": "[Bug]: [sqlancer] join crash",
+      "reported_date": "2022-02-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "iamlinjunhong",
+      "links": {
+        "report": "https://github.com/matrixorigin/matrixone/issues/1697"
+      },
+      "primary_url": "https://github.com/matrixorigin/matrixone/issues/1697",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/matrixorigin/matrixone/issues/1697",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]: [sqlancer] join crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8879dfeaac48705a0796d8c986bf91f8c706661d15181c62eb15c2a9745409e6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T15:56:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T15:56:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/matrixorigin/matrixone/issues/1697",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8879dfeaac48705a0796d8c986bf91f8c706661d15181c62eb15c2a9745409e6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:e4e9318e74e2",
+      "dbms": "monetdb",
+      "title": "Assertion failure when comparing INTERVAL value",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7562"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7562",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7562",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 INTERVAL YEAR TO MONTH);\r\nINSERT INTO t0( c0) VALUES ('2-5'); \r\n\r\nSELECT * FROM t0 WHERE (NOT ((t0.c0)>=('--18945599805')));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:558bd65c26b2b22c7789e390b581a3bc1797f78a47035881876871df89b57f83",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:558bd65c26b2b22c7789e390b581a3bc1797f78a47035881876871df89b57f83"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:f68fde3196a8",
+      "dbms": "monetdb",
+      "title": "Assertion failure when using CONTAINS",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7523"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7523",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7523",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INTEGER); \r\nINSERT INTO t1(c0) VALUES (1);\r\n\r\nSELECT * FROM t1 WHERE CONTAINS(1, NULL); -- unexpected end of file",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:30f493cb97f2876e0ede18d3c9270a0827e554c7996301a7db2fa70138fbaab2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:30f493cb97f2876e0ede18d3c9270a0827e554c7996301a7db2fa70138fbaab2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:5e3cbe65b7e8",
+      "dbms": "monetdb",
+      "title": "Assertion failure when using GROUP BY when CREATE VIEW",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7535"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7535",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7535",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR, c1 INTEGER, PRIMARY KEY(c0));\r\nCREATE VIEW v0(c0) AS SELECT ('a'||t0.c0) FROM t0 GROUP BY (CASE t0.c1 WHEN t0.c1 THEN 'a' END ); -- unexpected end of file\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:6e755b38757bf550fe19c61b1b7d307deeab4634e52633520ac88bbde99fb989",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:6e755b38757bf550fe19c61b1b7d307deeab4634e52633520ac88bbde99fb989"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:1ead504431c6",
+      "dbms": "monetdb",
+      "title": "Assertion failure when using INNER JOIN on STARTSWITH",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7553"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7553",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7553",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t0 (c0 VARCHAR(500) , c1 INT ) ; \r\nCREATE TABLE  t1 (c0 BOOLEAN ) ;\r\n\r\nSELECT * FROM t1 INNER  JOIN t0 ON STARTSWITH(((t0.c1)>=(t0.c0)), t1.c0, ('a' NOT LIKE(t0.c1))); -- unexpected end of file",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:97a19c4c52c903fd43e37675c43bdc7b3175b27289c9d47fcd1c98951c63dafb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:97a19c4c52c903fd43e37675c43bdc7b3175b27289c9d47fcd1c98951c63dafb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:79b40d52a58b",
+      "dbms": "monetdb",
+      "title": "Assertion failure when using JAROWINKLER in ORDER BY clause",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7528"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7528",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7528",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 VARCHAR);\r\nSELECT * FROM t0 WHERE NOT NULL ORDER BY JAROWINKLER('a', NULL) DESC; -- unexpected end of file\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e36c4f22360617257ebe972a8754c72551c4609b8e331a1cb77664d6cdd539c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e36c4f22360617257ebe972a8754c72551c4609b8e331a1cb77664d6cdd539c2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:727f97ef0a03",
+      "dbms": "monetdb",
+      "title": "Assertion failure when using JAROWINKLER with empty string",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7530"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7530",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7530",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0( c1 VARCHAR(500), PRIMARY KEY(c1)); \r\nINSERT INTO t0(c1) VALUES (true);\r\n\r\nSELECT t0.c1 FROM t0 WHERE JAROWINKLER(1, ''); -- unexpected end of file",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:09a472b9037e376839c7942c1d55ed15b515ebf4d1b2e66e65a0792e5bac7976",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:09a472b9037e376839c7942c1d55ed15b515ebf4d1b2e66e65a0792e5bac7976"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:318ada3caeb2",
+      "dbms": "monetdb",
+      "title": "Assertion failure when using STARTSWITH",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7540"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7540",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7540",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t1 (c0 BOOLEAN , c1 INT);\r\nCREATE TABLE  t0 (c0 BOOLEAN , c1 INT);\r\nINSERT INTO t1(c0) VALUES (true);\r\nINSERT INTO t1(c0) VALUES (false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:049dda189f4f3ef6f6670e16ccb73b7078f78947207e79363b5333649307627e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:049dda189f4f3ef6f6670e16ccb73b7078f78947207e79363b5333649307627e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:2570348e235e",
+      "dbms": "monetdb",
+      "title": "Assertion failure when using STARTSWITH with view",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7556"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7556",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7556",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t0 ( c1 INTEGER  ) ;\r\nCREATE VIEW v0(c0) AS SELECT NOT('a') FROM t0 ORDER BY (1); \r\n\r\nSELECT v0.c0 FROM v0, t0 WHERE (STARTSWITH(t0.c1, v0.c0, v0.c0)) ;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e310f12792394f3e62fd15407454a975e3202aa214d66f5f476714a9ceacee57",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e310f12792394f3e62fd15407454a975e3202aa214d66f5f476714a9ceacee57"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:1eac4daee3c6",
+      "dbms": "monetdb",
+      "title": "Crash at strcpy_len ()",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7573"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7573",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7573",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c1 INTEGER);\r\n\r\nINSERT INTO t0(c1) VALUES (-296623006);\r\nINSERT INTO t0(c1) VALUES (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:c8787bcf9f86168ecde94bac043be9043293ac466059e4faf93ceb8b682626a0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:c8787bcf9f86168ecde94bac043be9043293ac466059e4faf93ceb8b682626a0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:77c164d81f0c",
+      "dbms": "monetdb",
+      "title": "Crash when creating view with GROUP BY",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7545"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7545",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7545",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 VARCHAR, c1 VARCHAR ) ;\r\nCREATE UNIQUE INDEX i0 ON t0(c0 ); \r\nCREATE VIEW v0(c0) AS SELECT t0.c1 FROM t0 GROUP BY (CASE (NULL) WHEN t0.c1 THEN t0.c0 END ); -- unexpected end of file\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a6277cf48ffbddee5194911f41ebfc488e955a61b20953e3bbcd9204f46a3160",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a6277cf48ffbddee5194911f41ebfc488e955a61b20953e3bbcd9204f46a3160"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:f2abc9a642d6",
+      "dbms": "monetdb",
+      "title": "Crash when creating view with HAVING",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7522"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7522",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7522",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, PRIMARY KEY(c0)); \r\nCREATE VIEW v0(c0) AS SELECT '' FROM t0 HAVING ((t0.c0 BETWEEN t0.c0 AND t0.c0)); -- unexpected end of file\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9d8c3e3a36b3e7c8db2bd3a19b542071f96531384dcc8f45943cfb39a12c9ee4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9d8c3e3a36b3e7c8db2bd3a19b542071f96531384dcc8f45943cfb39a12c9ee4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:729daf927e7d",
+      "dbms": "monetdb",
+      "title": "Crash when integer overflow in ORDER BY",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7571"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7571",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7571",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t1(c0 BOOLEAN, c2 VARCHAR(500));\r\n\r\nSELECT t1.c2, t1.c0 FROM t1  ORDER BY ((IFNULL('', t1.c0) IS NOT DISTINCT FROM LEFT('1', t1.c0))LIKE ('1e500' >> 2)) ASC;\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:db3186ad6656044657be5338e4e508c5e7342ba59ca0f29e8e798b5d624cbf01",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:db3186ad6656044657be5338e4e508c5e7342ba59ca0f29e8e798b5d624cbf01"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:23f4deeb474d",
+      "dbms": "monetdb",
+      "title": "Crash when using IS DISTINCT FROM with SIN",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7539"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7539",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7539",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\r\nSELECT * FROM t0 WHERE (SIN(1) IS DISTINCT FROM NOT 1); -- unexpected end of file\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:42a57870b270c67b7a4e502812abbfdb3148776660be582d6924942d02cf98a6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:42a57870b270c67b7a4e502812abbfdb3148776660be582d6924942d02cf98a6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:70e75f7ea6b5",
+      "dbms": "monetdb",
+      "title": "Unexpected error when using NATURAL RIGHT JOIN",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7524"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7524",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7524",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT); \r\nCREATE TABLE t1(c0 VARCHAR);\r\n\r\nSELECT * FROM t1 RIGHT JOIN t0 ON t1.c0 = t0.c0; -- empty table",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:48f3c8fd99d5712f6e8a853bdfb2bb20fa928834c423015902076e20d7db013e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:48f3c8fd99d5712f6e8a853bdfb2bb20fa928834c423015902076e20d7db013e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:9e1184886e52",
+      "dbms": "monetdb",
+      "title": "Unexpected result when casting integer to boolean in comparison",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7555"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7555",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7555",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1 (c1 BOOLEAN) ;\r\nCREATE TABLE t0 (c0 INTEGER, PRIMARY KEY(c0) ) ;\r\nINSERT INTO t1(c1) VALUES (true);\r\nINSERT INTO t0(c0) VALUES (2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:985b8bed45ed8ed29eb5bc59085abdaa4d66cf76ae30e4ed7e2a6ea2e84709af",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:985b8bed45ed8ed29eb5bc59085abdaa4d66cf76ae30e4ed7e2a6ea2e84709af"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:600387d0f6d4",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using IS DISTINCT FROM in VIEW",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7563"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7563",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7563",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t1(c1 VARCHAR(500));\r\nINSERT INTO t1(c1) VALUES ('a');\r\nCREATE VIEW v0(c0) AS SELECT ('a'||NULL) FROM t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3c8b304b50eaa209a43f3164d9d9239e02d6545778f8a3606c1f2a4b47bdee76",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3c8b304b50eaa209a43f3164d9d9239e02d6545778f8a3606c1f2a4b47bdee76"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:4175128b3e77",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using IS DISTINCT FROM with AND",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7534"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7534",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7534",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nINSERT INTO t0 (c0) VALUES ('a');\r\n\r\nSELECT t0.c0 FROM t0; -- a",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d0f5587713ab6319978d66d09188a2465005b53cf4988f5a4f09622b3a882853",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d0f5587713ab6319978d66d09188a2465005b53cf4988f5a4f09622b3a882853"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:c50bd351b33e",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using IS DISTINCT FROM with RIGHT JOIN",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7527"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7527",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7527",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c1 INT);\r\nINSERT INTO t0 (c0) VALUES (0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a29c718febfce9be04890c41dc4eb35f79019eb95318474c62cd9bbf9246f644",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a29c718febfce9be04890c41dc4eb35f79019eb95318474c62cd9bbf9246f644"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:d6bdb7c12f24",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using LEVENSHTEIN",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7541"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7541",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7541",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 BOOLEAN);\r\nCREATE TABLE t1 (c1 INT) ;\r\nINSERT INTO t1(c1) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:80c215557d76c563c78fb4e3f9fa3325241eb7f556e6b574f53952d2610a1d3c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:80c215557d76c563c78fb4e3f9fa3325241eb7f556e6b574f53952d2610a1d3c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:5e2c098e67db",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using NULL constant in comparison",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7552"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7552",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7552",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  IF NOT EXISTS t0 ( c1 INT );\r\nINSERT INTO t0(c1) VALUES (1);\r\n\r\nSELECT * FROM t0; -- 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:2676934db109342faf61cfe9b42de17bf48317fdeb8d4d389c2d92a557c766d7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:2676934db109342faf61cfe9b42de17bf48317fdeb8d4d389c2d92a557c766d7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:551dcd057b26",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using STARTSWITH",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7544"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7544",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7544",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE  t0 (c0 VARCHAR) ;\r\nCREATE TABLE  t1 (c1 BOOLEAN) ;\r\nINSERT INTO t0(c0) VALUES ('a');\r\nINSERT INTO t1(c1) VALUES (true), (true);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:48db5b19cffc5597b146be0220e09d88040b472e8bb85b0874df8c52472b89a8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:48db5b19cffc5597b146be0220e09d88040b472e8bb85b0874df8c52472b89a8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:920b8b8685bd",
+      "dbms": "monetdb",
+      "title": "Query randomly failing with Object not found",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6936"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6936",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6936",
+            "source_type": "github_issue",
+            "excerpt": "I found this error with SQLancer. Maybe it already got resolved for SP1. It had to do with MAL functions missing conditional execution/candidate lists.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7a1f21aeec6aa7f4476cfad6dafebe818d5d29a59b0c037eb1b59ab2df98dd7b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6936",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7a1f21aeec6aa7f4476cfad6dafebe818d5d29a59b0c037eb1b59ab2df98dd7b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:611ebb2637fd",
+      "dbms": "monetdb",
+      "title": "SQLancer TLP query with wrong results",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6899"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6899",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6899",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer TLP query with wrong results",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:68c67c32890ef0c572fdb01d2849efdda9571f4f89dc8c00c46f49b72da5445b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6899",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:68c67c32890ef0c572fdb01d2849efdda9571f4f89dc8c00c46f49b72da5445b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:88b69a436945",
+      "dbms": "monetdb",
+      "title": "SQLancer TLP query with wrong results 2",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6901"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6901",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6901",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer TLP query with wrong results 2",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2284ff38ce40832b480acfee04819dc9eefbff82082d4462b2749b3a53f5a47d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6901",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2284ff38ce40832b480acfee04819dc9eefbff82082d4462b2749b3a53f5a47d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:dc4021a53ee9",
+      "dbms": "monetdb",
+      "title": "SQLancer TLP query with wrong results 3",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6905"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6905",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6905",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer TLP query with wrong results 3",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7fe2b3f06b48de1ac1743a7af3a762e5035886adea99cde7fcb55aa80a0a668e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6905",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7fe2b3f06b48de1ac1743a7af3a762e5035886adea99cde7fcb55aa80a0a668e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:fdba4c509f47",
+      "dbms": "monetdb",
+      "title": "SQLancer aggr.subavg undefined",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6904"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6904",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6904",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer aggr.subavg undefined",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8ced8ebdecc90d68b6150c1b3f2fd02d0f61189d3d040239ff4031ffc4f39497",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6904",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8ced8ebdecc90d68b6150c1b3f2fd02d0f61189d3d040239ff4031ffc4f39497"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:27df03cdbe15",
+      "dbms": "monetdb",
+      "title": "SQLancer algebra.select undefined",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6907"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6907",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6907",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer algebra.select undefined",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c656b3e36f9cd2a55705d502a07ed0bc9d89b66349ee39e42d2be348b98920d5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6907",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c656b3e36f9cd2a55705d502a07ed0bc9d89b66349ee39e42d2be348b98920d5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:709dad874336",
+      "dbms": "monetdb",
+      "title": "SQLancer algebra.select' undefined 2",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6896"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6896",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6896",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer algebra.select' undefined 2",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4d7b36f5e095f97f7608604251548b63e04b4b4c929dc4aa724e810a5638ddf3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6896",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4d7b36f5e095f97f7608604251548b63e04b4b4c929dc4aa724e810a5638ddf3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:ceba5609141e",
+      "dbms": "monetdb",
+      "title": "SQLancer alter table add unique gives strange error message",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6886"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6886",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6886",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer alter table add unique gives strange error message",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5a69c843850cb5330979742e727003e2058f6d3e67a569c021c1014101261c74",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6886",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5a69c843850cb5330979742e727003e2058f6d3e67a569c021c1014101261c74"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:0482fcfa32d7",
+      "dbms": "monetdb",
+      "title": "SQLancer calc.abs undefined",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6903"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6903",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6903",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer calc.abs undefined",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e8c5221ad3b84fcfb6f9bff7fa3f04904336f11319bff0fb5b50b681dda0e1c8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6903",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e8c5221ad3b84fcfb6f9bff7fa3f04904336f11319bff0fb5b50b681dda0e1c8"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:3c0d571d372f",
+      "dbms": "monetdb",
+      "title": "SQLancer calc.date undefined",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6929"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6929",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6929",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer calc.date undefined",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:09b1aa27cccc7ed46efc57dd00a8c39d7b95c2e2c61e90a86c1f6c9fd8c0f8f0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6929",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:09b1aa27cccc7ed46efc57dd00a8c39d7b95c2e2c61e90a86c1f6c9fd8c0f8f0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:902aa2d4775a",
+      "dbms": "monetdb",
+      "title": "SQLancer causes assertion error on UTF8_strlen",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6885"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6885",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6885",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer causes assertion error on UTF8_strlen",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:564203e4490c7084964649683a4b5cae0a4fd1b6d521d63bce399c568c2d79c9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6885",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:564203e4490c7084964649683a4b5cae0a4fd1b6d521d63bce399c568c2d79c9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:efae398c2f58",
+      "dbms": "monetdb",
+      "title": "SQLancer causing 'algebra.select' undefined error",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6895"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6895",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6895",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer causing 'algebra.select' undefined error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a175f4cbdcd1938c3ef1f270f3e5f899359af9325ea9a70b999470e0e6ab13df",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6895",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a175f4cbdcd1938c3ef1f270f3e5f899359af9325ea9a70b999470e0e6ab13df"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:193dcd3163fb",
+      "dbms": "monetdb",
+      "title": "SQLancer crash on coalesce",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6928"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6928",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6928",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer crash on coalesce",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d351b07aa1634a589ae5be525a16641a6b660059b4e6eca51c4bfc4861495191",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6928",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d351b07aa1634a589ae5be525a16641a6b660059b4e6eca51c4bfc4861495191"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:614890c4ad74",
+      "dbms": "monetdb",
+      "title": "SQLancer crash on complex join",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6906"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6906",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6906",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer crash on complex join",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:dc9c99973291081b862d61207b90679834aa1b86294dd26757489bbc48fc0fbf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6906",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dc9c99973291081b862d61207b90679834aa1b86294dd26757489bbc48fc0fbf"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:cf7de085bd06",
+      "dbms": "monetdb",
+      "title": "SQLancer crash on complex query",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6887"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6887",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6887",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer crash on complex query",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:49a5c3b106b245db8ee39e8792c281cab09984497cab8b7ef6d38d66aca2b9ac",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6887",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:49a5c3b106b245db8ee39e8792c281cab09984497cab8b7ef6d38d66aca2b9ac"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:043c89102760",
+      "dbms": "monetdb",
+      "title": "SQLancer crash on cross join on view",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6888"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6888",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6888",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer crash on cross join on view",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7595fd79d8711640cef3783782e531f49207e35d936ad28be8ad8e56df04ba75",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6888",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7595fd79d8711640cef3783782e531f49207e35d936ad28be8ad8e56df04ba75"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:fa529d06f0e5",
+      "dbms": "monetdb",
+      "title": "SQLancer crash on delete query",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6883"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6883",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6883",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer crash on delete query",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:dec76d483dad7e1b6e7c72803c858a93a2990f954eaf68be3f125892f5d90ad6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6883",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dec76d483dad7e1b6e7c72803c858a93a2990f954eaf68be3f125892f5d90ad6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:c1d93836eeee",
+      "dbms": "monetdb",
+      "title": "SQLancer crash on join query",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6898"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6898",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6898",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer crash on join query",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f6ea3d26d3e2d6b6d5fcae7901a78ff69c6f9f657e49bc0651e232c789fdd0bb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6898",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f6ea3d26d3e2d6b6d5fcae7901a78ff69c6f9f657e49bc0651e232c789fdd0bb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:2e149e29a529",
+      "dbms": "monetdb",
+      "title": "SQLancer crash on join with coalesce",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6930"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6930",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6930",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer crash on join with coalesce",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:45781fe77855be36b84a422e5530c0494873bbb94d68e211bb9dcffbf5027216",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6930",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:45781fe77855be36b84a422e5530c0494873bbb94d68e211bb9dcffbf5027216"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:3a2ea85a7ac5",
+      "dbms": "monetdb",
+      "title": "SQLancer crash on long query",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6889"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6889",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6889",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer crash on long query",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:67486a3ca2f881cc8373773aabb5ea750372aa427693afd7ccd5f45c06169fe0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6889",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:67486a3ca2f881cc8373773aabb5ea750372aa427693afd7ccd5f45c06169fe0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:0d203d63709b",
+      "dbms": "monetdb",
+      "title": "SQLancer crash on query with HAVING",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6892"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6892",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6892",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer crash on query with HAVING",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9dceeaae14f384b2fb9f6154e17ec68c847fcf582eb0dd11f769e5fa05e84a5a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6892",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9dceeaae14f384b2fb9f6154e17ec68c847fcf582eb0dd11f769e5fa05e84a5a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:ccd1462f82b2",
+      "dbms": "monetdb",
+      "title": "SQLancer crash on rtrim function",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6894"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6894",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6894",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer crash on rtrim function",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0b007a0a9ac68cf2650bc11c273fc100c3dcfcb6cd087a6cd1af09572c715b36",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6894",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0b007a0a9ac68cf2650bc11c273fc100c3dcfcb6cd087a6cd1af09572c715b36"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:4ee2e4d32d21",
+      "dbms": "monetdb",
+      "title": "SQLancer distinct aggregate with error on group by constant",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6897"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6897",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6897",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer distinct aggregate with error on group by constant",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0fac1ff0286b30f22c09ba8fba39390b2de574492479018c74807fe295f7996e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6897",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0fac1ff0286b30f22c09ba8fba39390b2de574492479018c74807fe295f7996e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:71bcc5cbf585",
+      "dbms": "monetdb",
+      "title": "SQLancer generated SIGFPE",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6900"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6900",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6900",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer generated SIGFPE",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e9510d8da90e965e2f96dc2e8552b251a0c1c64da9787b08264e530d1c414840",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6900",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e9510d8da90e965e2f96dc2e8552b251a0c1c64da9787b08264e530d1c414840"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:9f0faee8e3e6",
+      "dbms": "monetdb",
+      "title": "SQLancer generates query with unclear error message",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6884"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6884",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6884",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer generates query with unclear error message",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:acd1cc6c4633020ce583ebfff8628d8e1635e26c0044b0d7bca7f556ac4b0249",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6884",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:acd1cc6c4633020ce583ebfff8628d8e1635e26c0044b0d7bca7f556ac4b0249"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:ba56f9653479",
+      "dbms": "monetdb",
+      "title": "SQLancer inner join reporting GDK error",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6893"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6893",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6893",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer inner join reporting GDK error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c00aa87e98e6ee1991ded974d3ac6da0c1699841110a3c9c3db0f5bac5814473",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6893",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c00aa87e98e6ee1991ded974d3ac6da0c1699841110a3c9c3db0f5bac5814473"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:08c0b83efd8e",
+      "dbms": "monetdb",
+      "title": "SQLancer inputs not the same size",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6908"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6908",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6908",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer inputs not the same size",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:edd860931c48568c96497aaf1bfe3286bed8ecaa5648a4defc14f52cd1d1a8c0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6908",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:edd860931c48568c96497aaf1bfe3286bed8ecaa5648a4defc14f52cd1d1a8c0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:5bae89ae82b5",
+      "dbms": "monetdb",
+      "title": "SQLancer inputs not the same size",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6927"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6927",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6927",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer inputs not the same size",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:84259ad065482b6c276c2d7f9db36b4d8ed0b30715a89ce862b737093870845b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6927",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:84259ad065482b6c276c2d7f9db36b4d8ed0b30715a89ce862b737093870845b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:3485290c9877",
+      "dbms": "monetdb",
+      "title": "SQLancer insert function doesn't handle utf-8 strings",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6919"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6919",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6919",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer insert function doesn't handle utf-8 strings",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a4a001749ac189f0e7569ce9b109a72fde95f41dfd887c6762b35e8f9225a7b5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6919",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a4a001749ac189f0e7569ce9b109a72fde95f41dfd887c6762b35e8f9225a7b5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:3595d4d1f0b0",
+      "dbms": "monetdb",
+      "title": "SQLancer project_bte: does not match always",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6920"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6920",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6920",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer project_bte: does not match always",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7d697d4565311e37f30ebc5de8d0f1e4b901429974438c8dd32b5bf010ceece0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6920",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7d697d4565311e37f30ebc5de8d0f1e4b901429974438c8dd32b5bf010ceece0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:37408065a5a3",
+      "dbms": "monetdb",
+      "title": "SQLancer query compilation error",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6918"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6918",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6918",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer query compilation error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c9da07b3039b569a7ef8247fc9ea1f445e70765d8ad6a99d9c09ec35f41fb3cb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6918",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c9da07b3039b569a7ef8247fc9ea1f445e70765d8ad6a99d9c09ec35f41fb3cb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:9a57d76fffbe",
+      "dbms": "monetdb",
+      "title": "SQLancer query copy on unique pair of columns fails and complex query with GDK error",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6924"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6924",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6924",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer query copy on unique pair of columns fails and complex query with GDK error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6ceec69a204561bc508dbada1ba0f5c815c5de8ed172af87d92df6ad1e1135a3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6924",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6ceec69a204561bc508dbada1ba0f5c815c5de8ed172af87d92df6ad1e1135a3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:87fe3e71972c",
+      "dbms": "monetdb",
+      "title": "SQLancer query with wrong results",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6916"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6916",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6916",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer query with wrong results",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a7c8d6cf7f94869c0e18f4410b52d08a1be57fd336cac3614e2dcb7bddb5a80f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6916",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a7c8d6cf7f94869c0e18f4410b52d08a1be57fd336cac3614e2dcb7bddb5a80f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:d1b9a1794c6a",
+      "dbms": "monetdb",
+      "title": "SQLancer query with wrong results",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6909"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6909",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6909",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer query with wrong results",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b5cf416aca6a167296f0eb8f4b5e3ce073065842af36b7576afbd19d3c356118",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6909",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b5cf416aca6a167296f0eb8f4b5e3ce073065842af36b7576afbd19d3c356118"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:e820071e222c",
+      "dbms": "monetdb",
+      "title": "SQLancer query with wrong results",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6926"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6926",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6926",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer query with wrong results",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c2b8cfe1c79ec6bab4182523e2d99fef9404541fc3290f8bc9b556b076cde9c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6926",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c2b8cfe1c79ec6bab4182523e2d99fef9404541fc3290f8bc9b556b076cde9c2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:98f21a5da4c4",
+      "dbms": "monetdb",
+      "title": "SQLancer query: 'bat.append' undefined",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6910"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6910",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6910",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer query: 'bat.append' undefined",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a310046b5157ee923a047103a1c313be873cafeaf0a8626dd677344bb93e9a19",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6910",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a310046b5157ee923a047103a1c313be873cafeaf0a8626dd677344bb93e9a19"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:4ee0ce7cbf1b",
+      "dbms": "monetdb",
+      "title": "SQLancer query: 'calc.bit' undefined",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6911"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6911",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6911",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer query: 'calc.bit' undefined",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:61e5facade0d8629d1f4e3b7f2a5b8783a4e1576dcbf31b16de333b7b04e929a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6911",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:61e5facade0d8629d1f4e3b7f2a5b8783a4e1576dcbf31b16de333b7b04e929a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:fcc9a79fc60b",
+      "dbms": "monetdb",
+      "title": "SQLancer query: batcalc.between undefined",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6902"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6902",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6902",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer query: batcalc.between undefined",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8064bf52d3c64783c98de1af1e2e9f11b4efb2ec9a66991d23ad28f69fdb8ae3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6902",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8064bf52d3c64783c98de1af1e2e9f11b4efb2ec9a66991d23ad28f69fdb8ae3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:56d03c096532",
+      "dbms": "monetdb",
+      "title": "Segfault on large chain of constant decimal multiplication",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7003"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7003",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7003",
+            "source_type": "github_issue",
+            "excerpt": "While running SQLancer, I found there was an issue on the output scales for large numbers (it caused an assertion error). On the Oct2020 release commit I get wrong results. Meanwhile this got fixed and it will be available on the SP1 release soon.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4734d3b54520d26543474709603b1efe580f5d145cd6ee0e225db24ed31ef868",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7003",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4734d3b54520d26543474709603b1efe580f5d145cd6ee0e225db24ed31ef868"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:dea11e25004d",
+      "dbms": "monetdb",
+      "title": "Some MAL function are mapped incorrectly",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6973"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6973",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6973",
+            "source_type": "github_issue",
+            "excerpt": "This Summer we ran a fuzzer called SQLancer which found many bugs all around MonetDB. With many fixes including ABI changes, it was difficult to keep them on Jun2020, so that's why we are making the next feature release earlier than usual.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4661151b5040c2fb2a9c1ed68cc0ebb1965c6f861ffcb94ed54e25f57b3b66d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6973",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4661151b5040c2fb2a9c1ed68cc0ebb1965c6f861ffcb94ed54e25f57b3b66d8"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:d3814ddd4118",
+      "dbms": "monetdb",
+      "title": "Wrong result when dividing interval by literal float",
+      "reported_date": "2020-11-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "monetdb-team",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/6935"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/6935",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/6935",
+            "source_type": "github_issue",
+            "excerpt": "The second issue was found by SQLancer and it is fixed on default by disabling addition between intervals and other numeric types.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:40bbcb521c2294c859b23f553b61858904c54e26d0ceec1532bac941ac1a8e78",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/6935",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:40bbcb521c2294c859b23f553b61858904c54e26d0ceec1532bac941ac1a8e78"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:62a4e546196a",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using AND/OR chain",
+      "reported_date": "2024-01-24",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7448"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7448",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7448",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN,  c1 VARCHAR);\r\nINSERT INTO t0 (c0, c1) VALUES (true, true);\r\nCREATE INDEX i0 ON t0(c1 , c0 );",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:f2796d89e69f6f3ae00c4d76c3511829a55ae2b79088b71c9381c49efda22881",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:f2796d89e69f6f3ae00c4d76c3511829a55ae2b79088b71c9381c49efda22881"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:926ad4aadae6",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using BETWEEN in INNER JOIN",
+      "reported_date": "2024-01-24",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7447"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7447",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7447",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER, c1 BOOLEAN, PRIMARY KEY(c0));\r\nCREATE TABLE t1(c1 INTEGER);\r\nINSERT INTO t1 (c1) VALUES (null);\r\nINSERT INTO t0 (c0, c1) VALUES (-1, false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:aef06c151b92755b3741eea89c84c319bafead7f266d25a2c7e9ffc3f37712ec",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:aef06c151b92755b3741eea89c84c319bafead7f266d25a2c7e9ffc3f37712ec"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:6516ba9092b3",
+      "dbms": "monetdb",
+      "title": "Unexpected result when CREATE VIEW with WHERE NULL",
+      "reported_date": "2024-01-31",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7450"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7450",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7450",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0 ( c0) VALUES (false);\r\nCREATE VIEW v0(c0) AS SELECT true FROM t0 WHERE NULL GROUP BY t0.c0, true;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9de0c1b07e002b297529b17c40472f839e70d2f935077887f4db150c3e78326d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9de0c1b07e002b297529b17c40472f839e70d2f935077887f4db150c3e78326d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:d2a119fff922",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using BETWEEN and CAST",
+      "reported_date": "2024-01-31",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7451"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7451",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7451",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR, PRIMARY KEY(c0)); -- PK is needed\r\nCREATE TABLE t1(c1 INTEGER, c2 VARCHAR(500));\r\nINSERT INTO t0 ( c0) VALUES ('a');\r\nINSERT INTO t0 (c0) VALUES (false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9726c52fc2e5b03632abe4d1ec70afb49a9ecdb8a8a6246ed38705b2b005860a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9726c52fc2e5b03632abe4d1ec70afb49a9ecdb8a8a6246ed38705b2b005860a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:8d5fe4dd6f14",
+      "dbms": "monetdb",
+      "title": "Unexpected result for INNER JOIN with IS NOT NULL",
+      "reported_date": "2024-02-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7426"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7426",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7426",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c1 INTEGER);\r\nCREATE TABLE t0(c0 BOOL, c1 INTEGER);\r\nINSERT INTO t1 (c1) VALUES (0);\r\nCREATE UNIQUE INDEX i0 ON t0(c1 , c0 );",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ae45cbb1c088d2e219537e2e5b2441c6069aa437657303403a5949a4c1917833",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ae45cbb1c088d2e219537e2e5b2441c6069aa437657303403a5949a4c1917833"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:8bf6edb3d8c3",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using AND with INTEGER",
+      "reported_date": "2024-02-18",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7457"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7457",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7457",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500), c1 BOOLEAN);\r\nINSERT INTO t0 (c0, c1) VALUES ('a', false);\r\nINSERT INTO t0 (c0) VALUES ('b');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:7e40888aa322739acfa8fef3e7d06ce77332fb5b31037057805acc5ce2c2e6b4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:7e40888aa322739acfa8fef3e7d06ce77332fb5b31037057805acc5ce2c2e6b4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:e37ea62d42ba",
+      "dbms": "monetdb",
+      "title": "Crash when using CONTAINS in ORDER BY clause",
+      "reported_date": "2024-02-19",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7459"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7459",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7459",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER, c1 VARCHAR(500));\r\n\r\nSELECT * FROM t0 ORDER BY CONTAINS(((CASE '1' WHEN t0.c0 THEN '' ELSE t0.c0 END ) NOT BETWEEN 0 AND t0.c0>=t0.c1), t0.c1) DESC;  -- unexpected end of file\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:2d9c644ad18a804ab3870bf205d46ce885c24bf5f6562602d3adcfad3372ab97",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:2d9c644ad18a804ab3870bf205d46ce885c24bf5f6562602d3adcfad3372ab97"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:4b6379a6d710",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using SIGN",
+      "reported_date": "2024-02-19",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7458"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7458",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7458",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\nINSERT INTO t0 (c0) VALUES (0);\r\n\r\nSELECT * FROM t0; -- 0",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d0cb4388a4a1eb1ebeab8e4c4e87f35f6c347023a66c4a2c94525f1d75ea01b7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d0cb4388a4a1eb1ebeab8e4c4e87f35f6c347023a66c4a2c94525f1d75ea01b7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:397ec1817c73",
+      "dbms": "monetdb",
+      "title": "Crash by potentially use of bad escape characters",
+      "reported_date": "2024-02-20",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7461"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7461",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7461",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN, c1 BOOLEAN, PRIMARY KEY(c0));\r\nCREATE TABLE t1(c0 INTEGER, c1 INT, PRIMARY KEY(c0));\r\nINSERT INTO t0(c1) VALUES (-2041868105), (true);\r\nINSERT INTO t0(c1, c0) VALUES (-1820721628, -337528041);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:23df1708226b0c98e366729a929b25f1d1857081d74ab38b0dada5c3dc0649ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:23df1708226b0c98e366729a929b25f1d1857081d74ab38b0dada5c3dc0649ed"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:d513bda302c8",
+      "dbms": "monetdb",
+      "title": "Crash when using CAST and BETWEEN AND",
+      "reported_date": "2024-02-20",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7460"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7460",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7460",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0( c2 BOOLEAN);\r\nCREATE VIEW v0(c0) AS SELECT 1 FROM t0;\r\n\r\nINSERT INTO t0(c2) VALUES (true);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5fdb21e24d85ab7a83c75bf5aa853e132e78aa3c9de4728aff1db3c27971ff07",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5fdb21e24d85ab7a83c75bf5aa853e132e78aa3c9de4728aff1db3c27971ff07"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:53ba41a4956b",
+      "dbms": "monetdb",
+      "title": "Crash when using BETWEEN AND",
+      "reported_date": "2024-02-22",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7462"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7462",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7462",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500), c1 VARCHAR(500));\r\nINSERT INTO t0(c1) VALUES ('a');\r\n\r\nSELECT t0.c1, t0.c0 FROM t0 WHERE (((t0.c1 BETWEEN -1 AND 1))OR((t0.c1 BETWEEN -1 AND ('b' IN (t0.c1, t0.c0)))));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9b912c109583ccb9317bd96b62f31a908b8299d152823681bbcd19d5ba37cd13",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9b912c109583ccb9317bd96b62f31a908b8299d152823681bbcd19d5ba37cd13"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:383a17b410fa",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using CONTAINS and type casting",
+      "reported_date": "2024-02-22",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7463"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7463",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7463",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT);\r\nINSERT INTO t0 (c0) VALUES (0);\r\nINSERT INTO t0 (c0, c1) VALUES (1, 1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:4fd1207eac7e30db1bb891bd15a952aa9cbf642cbff685afba42b1beddd509ee",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:4fd1207eac7e30db1bb891bd15a952aa9cbf642cbff685afba42b1beddd509ee"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:7add6d9acf2e",
+      "dbms": "monetdb",
+      "title": "Crash when INNER JOIN with CONTAINS",
+      "reported_date": "2024-02-27",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7466"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7466",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7466",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\r\nCREATE TABLE t1(c1 VARCHAR(500));\r\nCREATE VIEW v0(c0) AS SELECT (('a')||('b')) FROM t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:932e1639944ee22ba66fc5d24c0fbd938a6af037c8f3150568f792b5d4d9d73e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:932e1639944ee22ba66fc5d24c0fbd938a6af037c8f3150568f792b5d4d9d73e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:dfeed4624dc8",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using NULL in BETWEEN",
+      "reported_date": "2024-02-27",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7465"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7465",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7465",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 BOOLEAN);\r\nINSERT INTO t1 (c0) VALUES (true);\r\n\r\nSELECT * FROM t1; -- true",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:58302c28f585c427b6426b73f414000f518d0087450c8c9fc100160b07582456",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:58302c28f585c427b6426b73f414000f518d0087450c8c9fc100160b07582456"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:e8044da3a263",
+      "dbms": "monetdb",
+      "title": "Crash when using CONTAINS",
+      "reported_date": "2024-03-13",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7469"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7469",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7469",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN, c1 VARCHAR);\r\nINSERT INTO t0(c0, c1) VALUES (true, 'a');\r\nINSERT INTO t0(c0, c1) VALUES (true, 'b');\r\nINSERT INTO t0(c0, c1) VALUES (false, 'c');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:7775ff2d924db95729d166d5e8743129885a9635887c3a7e3711456745e4dcd3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:7775ff2d924db95729d166d5e8743129885a9635887c3a7e3711456745e4dcd3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:d6b2db2759f0",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using `IS DISTINCT FROM`",
+      "reported_date": "2024-05-24",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7521"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7521",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7521",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected result when using `IS DISTINCT FROM`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:8f9bac5fa632f6b7fb7bb88ee4c9e71b83f1d6e068209ef1ec215511506ea177",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7521",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8f9bac5fa632f6b7fb7bb88ee4c9e71b83f1d6e068209ef1ec215511506ea177"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:199a1f7eea28",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using `IS DISTINCT FROM` with constants",
+      "reported_date": "2024-06-26",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7543"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7543",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7543",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected result when using `IS DISTINCT FROM` with constants",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d3a4f823dc9e8e4ddc8342bcbaa3fce1c407fbba85bfdc557e0dda007df86559",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7543",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d3a4f823dc9e8e4ddc8342bcbaa3fce1c407fbba85bfdc557e0dda007df86559"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:a70fb9c3b063",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using range comparison with `NULL`",
+      "reported_date": "2024-07-21",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7554"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7554",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7554",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected result when using range comparison with `NULL`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6e52c9dbb905ac4641f3338f503b791f08b49002302e4d392e6c4818c31e7388",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7554",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6e52c9dbb905ac4641f3338f503b791f08b49002302e4d392e6c4818c31e7388"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:62b43599802c",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using IFNULL with large numbers",
+      "reported_date": "2024-08-03",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7468"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7468",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7468",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INTEGER);\r\nINSERT INTO t0 (c1) VALUES (0);\r\n\r\nSELECT * FROM t0; -- 0",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:0057ed9abc35bda47123e2bc018229317359b78f6b7ce664a3ac6e99c0980f9f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:0057ed9abc35bda47123e2bc018229317359b78f6b7ce664a3ac6e99c0980f9f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:88212b134189",
+      "dbms": "monetdb",
+      "title": "Assertion failure at `rel2bin_select` when using `STARTSWITH`",
+      "reported_date": "2024-09-07",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7574"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7574",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7574",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t1(c0 INTEGER, c1 VARCHAR);\r\nCREATE VIEW v0(c0) AS SELECT 'a' FROM t1;\r\n\r\nSELECT * FROM t1, v0 WHERE STARTSWITH(t1.c0, v0.c0, t1.c1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9aca57f091d338a7b045370b1112d9aa8ae81a451c970d16d39ccfb6cbb9a80c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7574",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9aca57f091d338a7b045370b1112d9aa8ae81a451c970d16d39ccfb6cbb9a80c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:e20cb7223ea5",
+      "dbms": "monetdb",
+      "title": "Crash when using `CHECK` constraint",
+      "reported_date": "2024-09-23",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7577"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7577",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7577",
+            "source_type": "github_issue",
+            "excerpt": "CREATE  TABLE  t0(c0 INT, c1 INT CHECK (c0 < 10)) ;\r\nINSERT INTO t0(c0) VALUES (11);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f3be8abbed4fbfb7f607d9a380c82408a0c8fc13e14ddbce130d87af4c4dd53b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7577",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f3be8abbed4fbfb7f607d9a380c82408a0c8fc13e14ddbce130d87af4c4dd53b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:cddda88ea330",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using AND and IS NOT NULL",
+      "reported_date": "2024-10-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7430"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7430",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7430",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN, c1 VARCHAR);\r\nINSERT INTO t0 (c0, c1) VALUES (1, 1);\r\nCREATE UNIQUE INDEX i0t0 ON t0(c0 , c1 );",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:2b468b847eece2f31bbffeb351f7702fa63b709aca7f2724112326acff09fb53",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:2b468b847eece2f31bbffeb351f7702fa63b709aca7f2724112326acff09fb53"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:b44730998333",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using BETWEEN operator",
+      "reported_date": "2024-10-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7428"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7428",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7428",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER, c1 BOOL, PRIMARY KEY(c0));\r\nCREATE TABLE t1(c0 INTEGER, PRIMARY KEY(c0));\r\nINSERT INTO t0 (c0) VALUES (0);\r\nINSERT INTO t1 (c0) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:2be96e2c6faf84d9ec600d03d6101b6a6641e7a7ee27a88b566cf574c525a5ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:2be96e2c6faf84d9ec600d03d6101b6a6641e7a7ee27a88b566cf574c525a5ed"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:bb8dcca9df5b",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using CASE WHEN",
+      "reported_date": "2024-10-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7429"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7429",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7429",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nINSERT INTO t0 (c0) VALUES ('a');\r\n\r\nSELECT * FROM t0; -- a",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:34bc4a39694f6e248bbf058f454237f21b8fbed8bf7f043ee892aeec06704718",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:34bc4a39694f6e248bbf058f454237f21b8fbed8bf7f043ee892aeec06704718"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:e5f7fb812870",
+      "dbms": "monetdb",
+      "title": "Unexpected result when using BETWEEN with BOOLEAN values",
+      "reported_date": "2024-10-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7455"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7455",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7455",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR, c1 VARCHAR(500));\r\nCREATE VIEW v0(c0) AS SELECT true FROM t0;\r\nINSERT INTO t0 (c0, c1) VALUES ('', 0);\r\nINSERT INTO t0 (c0) VALUES ('a');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:1d34a8d31b82252642516965de2f8b8d607ac16e8b46cd31ecb50c27816a9271",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:1d34a8d31b82252642516965de2f8b8d607ac16e8b46cd31ecb50c27816a9271"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:d735d22f32cb",
+      "dbms": "monetdb",
+      "title": "Crash when INNER JOIN with VIEW",
+      "reported_date": "2024-12-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7456"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7456",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7456",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN);\r\nCREATE VIEW v0(c0) AS SELECT 1 FROM t0 HAVING true ORDER BY ((false)OR(false));\r\n\r\nSELECT * FROM t0 INNER  JOIN v0 ON v0.c0; -- unexpected end of file",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:51b59563d1ec4ce9be4eba8238941bc6783fd1703c97aaedc1897e60c7f34406",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:51b59563d1ec4ce9be4eba8238941bc6783fd1703c97aaedc1897e60c7f34406"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:530817b611b2",
+      "dbms": "monetdb",
+      "title": "Unexpected execution result",
+      "reported_date": "2025-01-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7695"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7695",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7695",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\n\nINSERT INTO t0(c0) VALUES(82);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:81560ce048eef6e2f0e7824f5c5fff1491ae7d66ff88b86046851b2356c41f95",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:81560ce048eef6e2f0e7824f5c5fff1491ae7d66ff88b86046851b2356c41f95"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:ec75a46dd714",
+      "dbms": "monetdb",
+      "title": "Unexpected execution result",
+      "reported_date": "2025-02-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7696"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7696",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7696",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT);\n\nINSERT INTO t0(c0, c1) VALUES(-2, 1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:dec6d8f14a480b66e5f7dab5369451fb320372c89c93b2dfae6915d544992d10",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:dec6d8f14a480b66e5f7dab5369451fb320372c89c93b2dfae6915d544992d10"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:c8402d573757",
+      "dbms": "monetdb",
+      "title": "Unexpected execution result",
+      "reported_date": "2025-03-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7697"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7697",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7697",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\n\nINSERT INTO t0(c0) VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:545914bf4ff13c765e745cd0d5bca011c8f1141dcfa947680b46871bb54a5d13",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:545914bf4ff13c765e745cd0d5bca011c8f1141dcfa947680b46871bb54a5d13"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:20428a7e38dd",
+      "dbms": "monetdb",
+      "title": "Unexpected Right Join Result",
+      "reported_date": "2025-04-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7698"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7698",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7698",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 INT );\nSET \"sql_optimizer\"='1'; --necessary\n\nINSERT INTO t0(c0) VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:7d0612275c92c7d777fac205ea14e41e8b43b4ffd20501c6953b07a38f378dc0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:7d0612275c92c7d777fac205ea14e41e8b43b4ffd20501c6953b07a38f378dc0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:94bbe1d7a72f",
+      "dbms": "monetdb",
+      "title": "Unexpected Left Join Result",
+      "reported_date": "2025-05-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7703"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7703",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7703",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 INT);\nCREATE TABLE IF NOT EXISTS t1(c0 INT);\n\nINSERT INTO t1(c0) VALUES(NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:2afbaee5270a7159ada3af7c89ad5e0036bc9efc8c3a4afbf73b5bdb9a698a32",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:2afbaee5270a7159ada3af7c89ad5e0036bc9efc8c3a4afbf73b5bdb9a698a32"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:04b9e56e9ede",
+      "dbms": "monetdb",
+      "title": "Unexpected Left Join Crash",
+      "reported_date": "2025-05-18",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7632"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7632",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7632",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\n\n\nSELECT * FROM t0 LEFT JOIN (VALUES (1)) ON EXISTS (SELECT ALL * FROM t0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cf3966e922d301c4cb58413503f80288ef5e02ffbb8d3625b6888eaccd98a201",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cf3966e922d301c4cb58413503f80288ef5e02ffbb8d3625b6888eaccd98a201"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:e383f9d8a341",
+      "dbms": "monetdb",
+      "title": "Unexpected Out of Memory of Inner Join",
+      "reported_date": "2025-05-18",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7633"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7633",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7633",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 INT);\nCREATE TABLE t1(LIKE t0);\nINSERT INTO t0 VALUES(1);\nINSERT INTO t1 VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cb6b3d0496a0d8378db8a15aea5ba0aa6222f86e612ec34a6ba90d154c672f1c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cb6b3d0496a0d8378db8a15aea5ba0aa6222f86e612ec34a6ba90d154c672f1c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:9a21645cdbd9",
+      "dbms": "monetdb",
+      "title": "Join with subquery crash",
+      "reported_date": "2025-05-19",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7634"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7634",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7634",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nINSERT INTO t0 VALUES (1);\n\nSELECT * FROM t0,(SELECT 1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:0405918ae5b8607a7cab222854abcc029c9f81df855b0a02616e1a54edc8babc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:0405918ae5b8607a7cab222854abcc029c9f81df855b0a02616e1a54edc8babc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:c48d32d6393f",
+      "dbms": "monetdb",
+      "title": "Unexpected Anti Join Crash",
+      "reported_date": "2025-05-19",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7636"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7636",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7636",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 INT);\nINSERT INTO t0 VALUES(0);\nINSERT INTO t1 VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:eeb46241e1d4d46a79865d6983375042922eb62218276b0c4c42168130aea1fe",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:eeb46241e1d4d46a79865d6983375042922eb62218276b0c4c42168130aea1fe"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:f7f5900dba87",
+      "dbms": "monetdb",
+      "title": "Unexpected Inner Join Crash",
+      "reported_date": "2025-05-19",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7635"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7635",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7635",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nINSERT INTO t0 VALUES (1);\nSELECT * FROM t0 INNER JOIN (SELECT (1 = ANY(VALUES (1)))) AS sub ON TRUE;\n-- 1, true",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:35f56f3ba0a3cfbf52b73bc44441ee6cbc3e7318cedf76d2ca73bb6b3ed51841",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:35f56f3ba0a3cfbf52b73bc44441ee6cbc3e7318cedf76d2ca73bb6b3ed51841"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:151d4c0f9474",
+      "dbms": "monetdb",
+      "title": "Unexpected Execution Results",
+      "reported_date": "2025-06-11",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7745"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7745",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7745",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\nCREATE TABLE t1(c1 INT);\n\nINSERT INTO t1(c1) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:75a7d4ef956225f1ee4896ff459071c09acbf7b1f9187b89e947d47a0503095f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:75a7d4ef956225f1ee4896ff459071c09acbf7b1f9187b89e947d47a0503095f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:ef36a8175e34",
+      "dbms": "monetdb",
+      "title": "Unexpected Internal Error in Inner Join",
+      "reported_date": "2025-06-13",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7645"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7645",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7645",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Internal Error in Inner Join",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e0262737c0b09d450461ef526f5d3f2fb1fcf578126dfaaa7ae3ff7c5e158a49",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "TheoristCoder is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7645",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e0262737c0b09d450461ef526f5d3f2fb1fcf578126dfaaa7ae3ff7c5e158a49"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:f7f41e912916",
+      "dbms": "monetdb",
+      "title": "Unexpected Left Join Crash",
+      "reported_date": "2025-06-13",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7646"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7646",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7646",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT);\nSELECT * FROM t1 LEFT JOIN (SELECT 1) ON EXISTS (SELECT (VALUES (1)));\n-- unexpected end of file\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3d019b85f3af0ef9550c660256cc805323b87c2e89bbb36d05fa911d93d368ea",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3d019b85f3af0ef9550c660256cc805323b87c2e89bbb36d05fa911d93d368ea"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:7d2fcd688a15",
+      "dbms": "monetdb",
+      "title": "Unexpected anti join crash",
+      "reported_date": "2025-06-13",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7644"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7644",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7644",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected anti join crash",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:415ec3ce297ec15d46f4b19f70f83e919abb46b2b3a108218a11132da3875c65",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "TheoristCoder is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7644",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:415ec3ce297ec15d46f4b19f70f83e919abb46b2b3a108218a11132da3875c65"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:09cebc9d04b2",
+      "dbms": "monetdb",
+      "title": "Unexpected Inner Join Crash",
+      "reported_date": "2025-06-19",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7649"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7649",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7649",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t1(c0 DOUBLE );\nCREATE TABLE IF NOT EXISTS t2(c0 DOUBLE );\nSELECT * FROM t1 INNER JOIN t2 ON (NOT EXISTS (VALUES (((t1.c0)+(t2.c0))))) IS NULL;\n-- unexpected end of file",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ff02fcfd44e9a2a56dd92b21eaf4ec58d351a586a6d732d74ee3d3bc49ac8e77",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ff02fcfd44e9a2a56dd92b21eaf4ec58d351a586a6d732d74ee3d3bc49ac8e77"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:98272ce86ac9",
+      "dbms": "monetdb",
+      "title": "Unexpected Right Join Assertion Error",
+      "reported_date": "2025-06-19",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7648"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7648",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7648",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 INT);\nSELECT *\nFROM t0\n         RIGHT JOIN (VALUES (1)) AS subQuery1(col_1) ON EXISTS (VALUES (CASE subQuery1.col_1 WHEN subQuery1.col_1 THEN 1 END));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:901886c991eb41636e68e4277960b3c9974c3ff6dddad942081bb223ece6587f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:901886c991eb41636e68e4277960b3c9974c3ff6dddad942081bb223ece6587f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:4b43e3b97574",
+      "dbms": "monetdb",
+      "title": "Unexpected Right Join Crash",
+      "reported_date": "2025-06-19",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7650"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7650",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7650",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t1(c0 INTERVAL DAY,c2 DOUBLE);\nSELECT ALL * FROM t1 RIGHT JOIN (VALUES (0)) AS subQuery1(col_1) ON ( t1.c0 < ANY(VALUES (t1.c0+t1.c0)) ) AND (VALUES (subQuery1.col_1), (EXISTS (VALUES (t1.c2))) );\n-- unexpected end of file\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:7e1721e22a374569bb606fb96f8f68416e017c9345d628a783fbb88f62120906",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:7e1721e22a374569bb606fb96f8f68416e017c9345d628a783fbb88f62120906"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:de3b7c2e7543",
+      "dbms": "monetdb",
+      "title": "Incorrect Anti Join Result",
+      "reported_date": "2025-06-24",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7651"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7651",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7651",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT);\nCREATE TABLE t1(c0 INT);\n\nINSERT INTO t0(c0) VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:6715a82546333d147e6fbbafc0ccf37a408cdb0afd61ba478862607921ca6ca5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:6715a82546333d147e6fbbafc0ccf37a408cdb0afd61ba478862607921ca6ca5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:ae86a190518c",
+      "dbms": "monetdb",
+      "title": "Incorrect Anti Join Result related to optimization",
+      "reported_date": "2025-06-24",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7652"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7652",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7652",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT , c1 INT , c2 INT);\nSET \"sql_optimizer\"='0'; -- necessary\nINSERT INTO t0(c2, c0, c1) VALUES(45, 1, 1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:61e31f77ff44ab8f7bd18db16ed33e7c7ad52be26c313034004fce7af9fed384",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:61e31f77ff44ab8f7bd18db16ed33e7c7ad52be26c313034004fce7af9fed384"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:b04946697bef",
+      "dbms": "monetdb",
+      "title": "Incorrect Inner Join Result",
+      "reported_date": "2025-06-24",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7653"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7653",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7653",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 INT , c1 INT);\n\nINSERT INTO t0(c1) VALUES(1), (2), (3);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:668029958f68a761b83f82954edcb026c2913e7432fb347befb7e760bf282777",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:668029958f68a761b83f82954edcb026c2913e7432fb347befb7e760bf282777"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:eb50e72797cd",
+      "dbms": "monetdb",
+      "title": "Unexpected execution result",
+      "reported_date": "2025-08-30",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7694"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7694",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7694",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nINSERT INTO t0(c0) VALUES(60);\nCREATE TABLE t1(c0 INT);\nINSERT INTO t1(c0) VALUES(0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:71f194c169a73243597278ca345cdcbb144f2a72cafb784d395dc31b0fee2c71",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:71f194c169a73243597278ca345cdcbb144f2a72cafb784d395dc31b0fee2c71"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:44ce829f02f4",
+      "dbms": "monetdb",
+      "title": "Unexpected Right Join Result",
+      "reported_date": "2025-09-13",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7707"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7707",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7707",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 TIMESTAMP );\nINSERT INTO t0(c0) VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ed6b5ad3bb7ec33acb31b8696e681fabf1188811072fff1551f94d84f0044f6a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ed6b5ad3bb7ec33acb31b8696e681fabf1188811072fff1551f94d84f0044f6a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:57aa6535f3fe",
+      "dbms": "monetdb",
+      "title": "Unexpected Join Result",
+      "reported_date": "2025-09-17",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7708"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7708",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7708",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c2 INT );\nINSERT INTO t0(c2) VALUES(NULL);\n\nSELECT * FROM t0 INNER JOIN LATERAL (SELECT t0.c2) AS subQuery0(c0) ON NOT EXISTS (VALUES (1), (subQuery0.c0));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:8915e3d2e167e3ffb3d81c4f5c73982e7090732fe5abb323ae1ac612e6412e4b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:8915e3d2e167e3ffb3d81c4f5c73982e7090732fe5abb323ae1ac612e6412e4b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:0f95259a273f",
+      "dbms": "monetdb",
+      "title": "Unexpected Crash",
+      "reported_date": "2025-09-18",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7711"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7711",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7711",
+            "source_type": "github_issue",
+            "excerpt": "I came across an unexpected execution result in MonetDB. Whenever convenient, it would mean a lot if you could take a look. I sincerely appreciate your efforts in continuously improving MonetDB’s robustness.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f71932ae9b314df97ebd9b281ff65ca514a140ed65c93383b26219f547fab70d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "TheoristCoder is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7711",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f71932ae9b314df97ebd9b281ff65ca514a140ed65c93383b26219f547fab70d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:6dd9e869e13b",
+      "dbms": "monetdb",
+      "title": "Unexpected Out of Memory",
+      "reported_date": "2025-09-18",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7712"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7712",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7712",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 boolean);\nSELECT * FROM t1 WHERE (((SELECT listagg('1') WHERE t1.c0)));\n-- out of memory\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e5eee396b2b3655f02c7be4e58e268dd6c2bf56440cb5a2c78acc0838a674526",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e5eee396b2b3655f02c7be4e58e268dd6c2bf56440cb5a2c78acc0838a674526"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:97c7140e03ee",
+      "dbms": "monetdb",
+      "title": "Unexpected Anti Join Result",
+      "reported_date": "2025-09-20",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7715"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7715",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7715",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 INT);\nINSERT INTO t0(c0) VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:c3a9cee5548fe9cd6a2b306e7a5c21d31f91bfda17ab6b9f820f8ad2b2d8e0d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:c3a9cee5548fe9cd6a2b306e7a5c21d31f91bfda17ab6b9f820f8ad2b2d8e0d8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:3621a90391d4",
+      "dbms": "monetdb",
+      "title": "Unexpected Anti Join Result",
+      "reported_date": "2025-09-22",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7716"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7716",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7716",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\n\nINSERT INTO t0(c0) VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:57a76f7d233dd3e6c6667c164a95bb7f3e96bf64322569370c88cfc44b6cfc1a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:57a76f7d233dd3e6c6667c164a95bb7f3e96bf64322569370c88cfc44b6cfc1a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:dfbee791e0ae",
+      "dbms": "monetdb",
+      "title": "Unexpected Crash in Left Join",
+      "reported_date": "2025-10-23",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7739"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7739",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7739",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DOUBLE);\nCREATE TABLE t1(c0 DOUBLE);\n\nINSERT INTO t0(c0) VALUES(0.9);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9f44f4e08549962a611cab6ccc4decbd37aa2d75b563dc578a73dfcff0cc7ad8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9f44f4e08549962a611cab6ccc4decbd37aa2d75b563dc578a73dfcff0cc7ad8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:37b8bfbf7c80",
+      "dbms": "monetdb",
+      "title": "Unexpected Error in Join",
+      "reported_date": "2025-10-24",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7744"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7744",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7744",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 INT);\n\nSELECT * FROM t0 JOIN (VALUES (1)) AS subQuery0(c0) ON NOT EXISTS (SELECT 1 WHERE subQuery0.c0 > NULL);\n-- Error in optimizer defaultpipe: TypeException:user.main[13]:'calc.>' undefined in:     X_82:any := calc.>(X_80:bte, nil:bte, true:bit);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:fb7fa034ba8a29b25a1e3ae7ff1976a54088c48209cac5f634b64335fd54da3c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:fb7fa034ba8a29b25a1e3ae7ff1976a54088c48209cac5f634b64335fd54da3c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:f3741b40443c",
+      "dbms": "monetdb",
+      "title": "Bug Report: Incorrect Boolean Expression Evaluation in WHERE Clause",
+      "reported_date": "2025-11-07",
+      "reported_year": 2025,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "AldonahZero",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7746"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7746",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7746",
+            "source_type": "github_issue",
+            "excerpt": "-- Test 4: TLP Oracle validation (combined query)\nSELECT * FROM t0 WHERE TRUE AND (TRUE AND TRUE)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:3aed6aa47f448c23e978f2f7968a3712f2117598b90e0a831c8d61a7b8eb241f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7746",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3aed6aa47f448c23e978f2f7968a3712f2117598b90e0a831c8d61a7b8eb241f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:b5c4c974f936",
+      "dbms": "monetdb",
+      "title": "Unexpected ANTI JOIN Result",
+      "reported_date": "2025-11-26",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7756"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7756",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7756",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOLEAN);\n\nINSERT INTO t0(c0) VALUES (true);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:0512f8b0fc78b84b136390880d7a393dadb32db198b8c4b6ee573b09f2688e8b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:0512f8b0fc78b84b136390880d7a393dadb32db198b8c4b6ee573b09f2688e8b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:fd7301c08366",
+      "dbms": "monetdb",
+      "title": "Unexpected SEMI Join Result",
+      "reported_date": "2025-11-29",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7761"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7761",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7761",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER, c1 INTEGER);\n\nINSERT INTO t0 (c0, c1) VALUES (-9, -1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:acb628994a6b87efdf143a2a50a729837af5c3a8996810cda9afecd1bb704b35",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:acb628994a6b87efdf143a2a50a729837af5c3a8996810cda9afecd1bb704b35"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:884e1bddd7be",
+      "dbms": "monetdb",
+      "title": "Unexpected Anti Join Result",
+      "reported_date": "2025-12-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7705"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7705",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7705",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 INT);\nINSERT INTO t0(c0) VALUES(1);\n\nSELECT * FROM t0 WHERE NOT EXISTS (SELECT 1 FROM (VALUES (t0.c0)) WHERE (((SELECT FALSE FROM t0)) = FALSE) = TRUE);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d075ef64e34d30e4a094726e9ba9ebf9737cda393a851475cc2f7c1b7d80f6b1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d075ef64e34d30e4a094726e9ba9ebf9737cda393a851475cc2f7c1b7d80f6b1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:0d0acab11ecf",
+      "dbms": "monetdb",
+      "title": "DECIMAL column values incorrectly returned as 0 in large JOIN result sets when connecting via pymonetdb",
+      "reported_date": "2026-05-23",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "fyr03",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7939"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7939",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7939",
+            "source_type": "github_issue",
+            "excerpt": "cur.execute('SET SCHEMA \"sqlancer\"')\nconn.commit()",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5f600f4c6057eb1a622f32791eefa5e0d6572905ae660047f0f28104615fa2ee",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7939",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5f600f4c6057eb1a622f32791eefa5e0d6572905ae660047f0f28104615fa2ee"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:monetdb:44ce019ba6bc",
+      "dbms": "monetdb",
+      "title": "Dynamic `LAG` offset is ignored and behaves like offset 0",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7980"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7980",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7980",
+            "source_type": "github_issue",
+            "excerpt": "Dynamic `LAG` offset is ignored and behaves like offset 0",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9bdcf12ad3e7359f019cc37e888e6798ab79a179e60cd6dc644f2814dc54b3cc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7980",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9bdcf12ad3e7359f019cc37e888e6798ab79a179e60cd6dc644f2814dc54b3cc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:e8e295719044",
+      "dbms": "monetdb",
+      "title": "`NOT` over a false self-comparison returns `false`",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7981"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7981",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7981",
+            "source_type": "github_issue",
+            "excerpt": "`NOT` over a false self-comparison returns `false`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5a77a5bb88ef0c4f148fffe7d18e76e54d511cca5089ffde904bf66c3d93a2be",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7981",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5a77a5bb88ef0c4f148fffe7d18e76e54d511cca5089ffde904bf66c3d93a2be"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:5c284047991c",
+      "dbms": "monetdb",
+      "title": "`RANGE CURRENT ROW` ignores peer rows",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7982"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7982",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7982",
+            "source_type": "github_issue",
+            "excerpt": "`RANGE CURRENT ROW` ignores peer rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:92dfe0658e983a899bda9d05bca93e818c5f321033dfe3eb11025074c5c9fefb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7982",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:92dfe0658e983a899bda9d05bca93e818c5f321033dfe3eb11025074c5c9fefb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:4178c7f33c52",
+      "dbms": "monetdb",
+      "title": "`QUALIFY` with inline `ROW_NUMBER` leaks an `ORDER BY` column and filters the wrong rows",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7985"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7985",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7985",
+            "source_type": "github_issue",
+            "excerpt": "`QUALIFY` with inline `ROW_NUMBER` leaks an `ORDER BY` column and filters the wrong rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:46a34233aaf3004197be803833c389b74b0fb4b15dfc40ed8ecd3376174591cd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7985",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:46a34233aaf3004197be803833c389b74b0fb4b15dfc40ed8ecd3376174591cd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:2e42b89e74db",
+      "dbms": "monetdb",
+      "title": "`RIGHT` returns the empty string for a one-character UTF-8 input",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7986"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7986",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7986",
+            "source_type": "github_issue",
+            "excerpt": "`RIGHT` returns the empty string for a one-character UTF-8 input",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b1be46403b032dd7cca269f24394201c57bfffaf638132887e751785550a64b0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7986",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b1be46403b032dd7cca269f24394201c57bfffaf638132887e751785550a64b0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:29efa05a30cc",
+      "dbms": "monetdb",
+      "title": "`WHERE` treats an UNKNOWN boolean expression as true",
+      "reported_date": "2026-08-06",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7987"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7987",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7987",
+            "source_type": "github_issue",
+            "excerpt": "`WHERE` treats an UNKNOWN boolean expression as true",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5181eed2ce687680649a73d92d3093818778197ca4ca2728f5fc16b1fe254a41",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7987",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5181eed2ce687680649a73d92d3093818778197ca4ca2728f5fc16b1fe254a41"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:db3ab080055b",
+      "dbms": "monetdb",
+      "title": "ASC temporal `RANGE` frames mishandle NULL order keys",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7992"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7992",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7992",
+            "source_type": "github_issue",
+            "excerpt": "ASC temporal `RANGE` frames mishandle NULL order keys",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7be2ee5d5a38273d1c2b91aaf2583a31e425201efc5bf628c8291686eb621d26",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7992",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7be2ee5d5a38273d1c2b91aaf2583a31e425201efc5bf628c8291686eb621d26"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:4856e1017164",
+      "dbms": "monetdb",
+      "title": "Window `GROUP_CONCAT` mishandles a NULL separator",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7991"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7991",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7991",
+            "source_type": "github_issue",
+            "excerpt": "Window `GROUP_CONCAT` mishandles a NULL separator",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:21c0f8727306d0f3743839573fb01cb9f2998e5632b8b932183010d78a817ee0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7991",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:21c0f8727306d0f3743839573fb01cb9f2998e5632b8b932183010d78a817ee0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:c92b5b6d3267",
+      "dbms": "monetdb",
+      "title": "Windowed `AVG` mis-merges a wide bounded frame",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7989"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7989",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7989",
+            "source_type": "github_issue",
+            "excerpt": "Windowed `AVG` mis-merges a wide bounded frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a1e12c8cab207a19faf27e4a763f3ebbbd3fcced869accef80e30724fd808f60",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7989",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a1e12c8cab207a19faf27e4a763f3ebbbd3fcced869accef80e30724fd808f60"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:feca1d2ef695",
+      "dbms": "monetdb",
+      "title": "`RANK() <= constant` drops rows tied at the cutoff rank",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/7990"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/7990",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/7990",
+            "source_type": "github_issue",
+            "excerpt": "`RANK() <= constant` drops rows tied at the cutoff rank",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d6bbaed27b5511fabc507926f53311bf8cab214d55845ed12a0bd3d681051f5e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/7990",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d6bbaed27b5511fabc507926f53311bf8cab214d55845ed12a0bd3d681051f5e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:181b3a96c209",
+      "dbms": "monetdb",
+      "title": "Numeric `RANGE NULLS LAST` includes the wrong rows",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/8009"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/8009",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/8009",
+            "source_type": "github_issue",
+            "excerpt": "Numeric `RANGE NULLS LAST` includes the wrong rows",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3c09cab6504a0c35ea271f52853cc308a7cf63d22251ce5b82c6b716f971f3ad",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/8009",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3c09cab6504a0c35ea271f52853cc308a7cf63d22251ce5b82c6b716f971f3ad"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:6555eed371c2",
+      "dbms": "monetdb",
+      "title": "Numeric `RANGE` frames cross interleaved partitions",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/8008"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/8008",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/8008",
+            "source_type": "github_issue",
+            "excerpt": "Numeric `RANGE` frames cross interleaved partitions",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b4ba4f39af36be769341eb624e701a80c98325412f5491efa014f64641247615",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/8008",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b4ba4f39af36be769341eb624e701a80c98325412f5491efa014f64641247615"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:161e04e7f47e",
+      "dbms": "monetdb",
+      "title": "Value functions lose an unordered `RANGE CURRENT ROW` frame",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/8010"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/8010",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/8010",
+            "source_type": "github_issue",
+            "excerpt": "Value functions lose an unordered `RANGE CURRENT ROW` frame",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:cc7c4fc62d99d2aafb9a4223292be939ab6d92b062affe721ad377f8607853d3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/8010",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cc7c4fc62d99d2aafb9a4223292be939ab6d92b062affe721ad377f8607853d3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:2a40b3534dd3",
+      "dbms": "monetdb",
+      "title": "Window `GROUP_CONCAT` ignores aggregate `ORDER BY`",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/8003"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/8003",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/8003",
+            "source_type": "github_issue",
+            "excerpt": "Window `GROUP_CONCAT` ignores aggregate `ORDER BY`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:204cbce19d45a26e320b55d26a97c7a816b5fda282f28cf6adba4245aaba145f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/8003",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:204cbce19d45a26e320b55d26a97c7a816b5fda282f28cf6adba4245aaba145f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:05b5f31f6272",
+      "dbms": "monetdb",
+      "title": "`LAG`/`LEAD` reuse the first default expression value at partition boundaries",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/8007"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/8007",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/8007",
+            "source_type": "github_issue",
+            "excerpt": "`LAG`/`LEAD` reuse the first default expression value at partition boundaries",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6a5e939bc851ccfbf9e1c232c8df54e14e7ba69329816249ae67148514138d87",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/8007",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6a5e939bc851ccfbf9e1c232c8df54e14e7ba69329816249ae67148514138d87"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:807c808e2e85",
+      "dbms": "monetdb",
+      "title": "`NTH_VALUE` with a row-dependent `nth` returns a non-NULL value when the requested row is missing",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/8004"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/8004",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/8004",
+            "source_type": "github_issue",
+            "excerpt": "`NTH_VALUE` with a row-dependent `nth` returns a non-NULL value when the requested row is missing",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:abdbb5e0dcc38e9098a47190a486554dd3692088d59217a590dd2f37d6276b8f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/8004",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:abdbb5e0dcc38e9098a47190a486554dd3692088d59217a590dd2f37d6276b8f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:35eb3dfbecd4",
+      "dbms": "monetdb",
+      "title": "`field()` function returns multiple rows when the searched value appears more than once",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/8006"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/8006",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/8006",
+            "source_type": "github_issue",
+            "excerpt": "`field()` function returns multiple rows when the searched value appears more than once",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:808deae0d8483d5665667487fb828963894fd828009ba079dbba61c52c9e0e4d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/8006",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:808deae0d8483d5665667487fb828963894fd828009ba079dbba61c52c9e0e4d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:monetdb:cbe44e3611d2",
+      "dbms": "monetdb",
+      "title": "`prod` treats an all-NULL input differently from an empty input",
+      "reported_date": "2026-08-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Yibo-Dong",
+      "links": {
+        "report": "https://github.com/MonetDB/MonetDB/issues/8005"
+      },
+      "primary_url": "https://github.com/MonetDB/MonetDB/issues/8005",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/MonetDB/MonetDB/issues/8005",
+            "source_type": "github_issue",
+            "excerpt": "`prod` treats an all-NULL input differently from an empty input",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5bf75033954b189a94ea2efb674787934edcf56526a8f176c57aa8d56eb181bb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "Yibo-Dong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/MonetDB/MonetDB/issues/8005",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5bf75033954b189a94ea2efb674787934edcf56526a8f176c57aa8d56eb181bb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:42f3dfe0b3a1",
+      "dbms": "mysql",
+      "title": "BETWEEN malfunctions for DECIMAL and TEXT",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/99182"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/99182 BETWEEN malfunctions for DECIMAL and TEXT",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug99182 that works around it.",
+            "content_sha256": "sha256:eda7d013a57312ba97ff6597c486865611d1b2a93aedbed10ac537b359331ce5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:eda7d013a57312ba97ff6597c486865611d1b2a93aedbed10ac537b359331ce5"
+      },
+      "primary_url": "https://bugs.mysql.com/99182",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:b7f9f83f2e9e",
+      "dbms": "mysql",
+      "title": "Creating an index in between two NULL inserts causes inconsistent CERT result",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/bug.php?id=120712"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/bug.php?id=120712\nCreating an index in between two NULL inserts causes inconsistent CERT result.",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug120712 that works around it.",
+            "content_sha256": "sha256:d24b1c3817c13b37fcdedf07e8d05a45d9f628d7f89b07249efa07bf8e841391",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:d24b1c3817c13b37fcdedf07e8d05a45d9f628d7f89b07249efa07bf8e841391"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=120712",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:8c403b77bbc1",
+      "dbms": "mysql",
+      "title": "Creating an index on an integer-type column, then inserting a value which rounds to 1, causes result set mismatch",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/bug.php?id=120711"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/bug.php?id=120711\nCreating an index on an integer-type column, then inserting a value which rounds to 1, causes result set\nmismatch.",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug120711 that works around it.",
+            "content_sha256": "sha256:1d3f73f5b4c4d6780570f4bdb8dfe5ef3fcc9ce675d48614adbb5fa859f8c555",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:1d3f73f5b4c4d6780570f4bdb8dfe5ef3fcc9ce675d48614adbb5fa859f8c555"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=120711",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:384a8d48c71e",
+      "dbms": "mysql",
+      "title": "Inserting a NULL and a value which rounds to 0 into a DECIMAL column causes result set mismatch",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/bug.php?id=120710"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/bug.php?id=120710\nInserting a NULL and a value which rounds to 0 into a DECIMAL column causes result set mismatch.",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug120710 that works around it.",
+            "content_sha256": "sha256:23160e4ee251dbf1e621b2dac444a5c23067024126ba3854b9be6669a5870d4b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:23160e4ee251dbf1e621b2dac444a5c23067024126ba3854b9be6669a5870d4b"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=120710",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:3eb425674cbe",
+      "dbms": "mysql",
+      "title": "https://bugs.mysql.com/bug.php?id=111471",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/bug.php?id=111471"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/bug.php?id=111471",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug111471 that works around it.",
+            "content_sha256": "sha256:3eb425674cbeec5e9086425739d508b07bbb6d312d71ef9fd4c4cba101a92445",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:3eb425674cbeec5e9086425739d508b07bbb6d312d71ef9fd4c4cba101a92445"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=111471",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:a4b90033492e",
+      "dbms": "mysql",
+      "title": "https://bugs.mysql.com/bug.php?id=112242",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/bug.php?id=112242"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/bug.php?id=112242",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug112242 that works around it.",
+            "content_sha256": "sha256:a4b90033492e411fa62c01914ee11b7c8102a0542b02950bcbf2f0dee9544d13",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:a4b90033492e411fa62c01914ee11b7c8102a0542b02950bcbf2f0dee9544d13"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=112242",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:52cbee77f7f2",
+      "dbms": "mysql",
+      "title": "https://bugs.mysql.com/bug.php?id=112243",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/bug.php?id=112243"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/bug.php?id=112243",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug112243 that works around it.",
+            "content_sha256": "sha256:52cbee77f7f27eafc35e305f6ac9803022dd6030ea3890edac0db9b6e15ae8ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:52cbee77f7f27eafc35e305f6ac9803022dd6030ea3890edac0db9b6e15ae8ed"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=112243",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:ed4990ccd0f2",
+      "dbms": "mysql",
+      "title": "https://bugs.mysql.com/bug.php?id=112264",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/bug.php?id=112264"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/bug.php?id=112264",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug112264 that works around it.",
+            "content_sha256": "sha256:ed4990ccd0f21314e4a77dc302af8dab999a1a626bf5fee537931e7b85a94f7c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:ed4990ccd0f21314e4a77dc302af8dab999a1a626bf5fee537931e7b85a94f7c"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=112264",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:8623ad6e273f",
+      "dbms": "mysql",
+      "title": "https://bugs.mysql.com/bug.php?id=114533",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/bug.php?id=114533"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/bug.php?id=114533",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug114533 that works around it.",
+            "content_sha256": "sha256:8623ad6e273f3648f9f2df3f1a419f957193d2877e43a67be507208c0e7d6a3e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:8623ad6e273f3648f9f2df3f1a419f957193d2877e43a67be507208c0e7d6a3e"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=114533",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:67183b688f9c",
+      "dbms": "mysql",
+      "title": "https://bugs.mysql.com/bug.php?id=114534",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/bug.php?id=114534"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/bug.php?id=114534",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug114534 that works around it.",
+            "content_sha256": "sha256:67183b688f9ca6e079b4c581c1a9ddc65fe6822cd117c8068dc680a91cac2867",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:67183b688f9ca6e079b4c581c1a9ddc65fe6822cd117c8068dc680a91cac2867"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=114534",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:dcdf8573694d",
+      "dbms": "mysql",
+      "title": "https://bugs.mysql.com/bug.php?id=99183",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "report": "https://bugs.mysql.com/bug.php?id=99183"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://bugs.mysql.com/bug.php?id=99183",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its mysql provider, as the field bug99183 that works around it.",
+            "content_sha256": "sha256:dcdf8573694d1385bfb12a75f83607b5f4e15f0e77e90adc3531151dff36c8aa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/mysql/MySQLBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:dcdf8573694d1385bfb12a75f83607b5f4e15f0e77e90adc3531151dff36c8aa"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99183",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:mysql:3b0b4227e3d9",
+      "dbms": "mysql",
+      "title": "REPAIR TABLE on a functional table marks it as crashed",
+      "reported_date": "2019-06-15",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95820"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95820",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95820\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:3097ef3a4d517bf778af3ba123c968dd5548ab8ea037dc304ba5cacfc01e4f8d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3097ef3a4d517bf778af3ba123c968dd5548ab8ea037dc304ba5cacfc01e4f8d"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95820",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:ed0cab0ebb7d",
+      "dbms": "mysql",
+      "title": "\"Can't find record\" error in SELECT statement 1 (MyISAM engine)",
+      "reported_date": "2019-06-18",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95856"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95856",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"18/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95856\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:67e421b73f431b367de6e9d7b56a118456849db625360381be61219882bf285a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:67e421b73f431b367de6e9d7b56a118456849db625360381be61219882bf285a"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95856",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:678464cdef72",
+      "dbms": "mysql",
+      "title": "\"Can't find record\" error in SELECT statement 2 (HEAP engine)",
+      "reported_date": "2019-06-18",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95866"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95866",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"18/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95866\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5f2dddf81329bae04a4efa76aaee867ba8633fee9bd76f11aa923400eb7d2cec",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5f2dddf81329bae04a4efa76aaee867ba8633fee9bd76f11aa923400eb7d2cec"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95866",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:9a1766bd10fc",
+      "dbms": "mysql",
+      "title": "Applying NOT twice on an integer results in wrong result in WHERE condition",
+      "reported_date": "2019-06-20",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95900"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95900",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.17\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95900\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:b0b5a30bea0c8f02ccf83c2a2e92115f7ea6a87b860ffae25618a070350d68d9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b0b5a30bea0c8f02ccf83c2a2e92115f7ea6a87b860ffae25618a070350d68d9"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95900",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:aa8c7c8656bd",
+      "dbms": "mysql",
+      "title": "DROP COLUMN error on INVISIBLE UNIQUE INDEX that refers to constant expression",
+      "reported_date": "2019-06-20",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95897"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95897",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"false_positive\": \"yes\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95897\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:b4d5e2501c0eea73d762d06cf4b26cdea64ed1c278c9d6cd38e1eb4aa15d047e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b4d5e2501c0eea73d762d06cf4b26cdea64ed1c278c9d6cd38e1eb4aa15d047e"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95897",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:449615d58c08",
+      "dbms": "mysql",
+      "title": "DROP PRIMARY KEY on UNIQUE PRIMARY KEY does not update information_schema.column",
+      "reported_date": "2019-06-20",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95894"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95894",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95894\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1ea576b12e480bb4cc6868db428dab08f1cf5dfac0b01fd0ccba588894ef7c0c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1ea576b12e480bb4cc6868db428dab08f1cf5dfac0b01fd0ccba588894ef7c0c"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95894",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:9907dcd0623f",
+      "dbms": "mysql",
+      "title": "Functional index seems to malfunction with UNSIGNED column",
+      "reported_date": "2019-06-20",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95889"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95889",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95889\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:976396dc7aee3593bcb8358920e12cc03541f0ab7c766c3225cb40e7d9bb460b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:976396dc7aee3593bcb8358920e12cc03541f0ab7c766c3225cb40e7d9bb460b"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95889",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:5a3d877a9178",
+      "dbms": "mysql",
+      "title": "The negation of a \"<=>\" comparison malfunctions depending on the column's type",
+      "reported_date": "2019-06-21",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95908"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95908",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.17\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95908\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:46cbfa3fc63279be8d3b5e4f11803d5c6736f0435f781fcb9142fe6e469d6bb0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:46cbfa3fc63279be8d3b5e4f11803d5c6736f0435f781fcb9142fe6e469d6bb0"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95908",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:d4079ffa771c",
+      "dbms": "mysql",
+      "title": "Duplicate entry for key 'PRIMARY' when querying information_schema.TABLES",
+      "reported_date": "2019-06-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95929"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95929",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.20\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95929\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:2ede6713fc615092e8a105512f43f17f162a406663e3988310db8cd893aaedaf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2ede6713fc615092e8a105512f43f17f162a406663e3988310db8cd893aaedaf"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95929",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:f6e786362b1e",
+      "dbms": "mysql",
+      "title": "Row is not fetched when using a function expression that should yield TRUE",
+      "reported_date": "2019-06-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95926"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95926",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.17\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95926\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:21fd9d8c69fa6c38942d37f9f7091e37eff7676554fe05139436bd73d4b06311",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:21fd9d8c69fa6c38942d37f9f7091e37eff7676554fe05139436bd73d4b06311"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95926",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:44f73b8dd1fc",
+      "dbms": "mysql",
+      "title": "XOR operator returns incorrect result for strings with a floating-point number",
+      "reported_date": "2019-06-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95927"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95927",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95927\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:36397164ceb417f0ce8f4e1a26aef87b8983feb8dfa69b8da9b5cd16b9b1e432",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:36397164ceb417f0ce8f4e1a26aef87b8983feb8dfa69b8da9b5cd16b9b1e432"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95927",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:0419ec1c7efb",
+      "dbms": "mysql",
+      "title": "CHECK TABLE FOR UPGRADE crashes server (segfault)",
+      "reported_date": "2019-06-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "CHECK TABLE FOR UPGRADE crashes server (segfault)",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a MySQL bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:0419ec1c7efb1a3c31e66ee373db3ec9bdabe819e1f35d72f75f1728d9804c4d",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:0419ec1c7efb1a3c31e66ee373db3ec9bdabe819e1f35d72f75f1728d9804c4d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:19c9842e444a",
+      "dbms": "mysql",
+      "title": "CHECK TABLE FOR UPGRADE crashes server (segfault)",
+      "reported_date": "2019-06-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "cve": "CVE-2019-2879",
+      "links": {
+        "curated_entry": "https://github.com/sqlancer/bugs/blob/master/bugs.json"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.17\",\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1b8da6210abf14d3f92a222e0939e1b9c79bf91ec2c3feee942b422dba31bc06",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1b8da6210abf14d3f92a222e0939e1b9c79bf91ec2c3feee942b422dba31bc06"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:46b4c1170f3b",
+      "dbms": "mysql",
+      "title": "AND/OR/XOR compute wrong result for small floating-point numbers in TEXT columns",
+      "reported_date": "2019-06-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95958"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95958",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.17\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95958\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:5fb286ffc2e0f875f86205257d8f88ac15b3b4c61a16c8c8bc4381b62ee56b94",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5fb286ffc2e0f875f86205257d8f88ac15b3b4c61a16c8c8bc4381b62ee56b94"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95958",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:0b2668278b32",
+      "dbms": "mysql",
+      "title": "BENCHMARK() returns NULL in some cases but is documented to always return 0",
+      "reported_date": "2019-06-24",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95937"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95937",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95937\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:c2f48b910d24c4901020349e167143307c44afea5e4c5c12269a780e51aa2d1b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c2f48b910d24c4901020349e167143307c44afea5e4c5c12269a780e51aa2d1b"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95937",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:06143a10034b",
+      "dbms": "mysql",
+      "title": "CAST of negative function return value to UNSIGNED malfunctions with BIGINT",
+      "reported_date": "2019-06-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95954"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95954",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.18\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95954\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:1c4e417a87cc06586dc1478be9586a90525412feb4fe9eb6982545c44bb7ed54",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1c4e417a87cc06586dc1478be9586a90525412feb4fe9eb6982545c44bb7ed54"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95954",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:597b4c29d2ff",
+      "dbms": "mysql",
+      "title": "Function return value with newline yields wrong result when used as a boolean",
+      "reported_date": "2019-06-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95938"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95938",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.17\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95938\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:19f17124e446f280d53e9f1a54773c38b73e251889cd48f970bdf85c5bcf6b82",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:19f17124e446f280d53e9f1a54773c38b73e251889cd48f970bdf85c5bcf6b82"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95938",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:3ecd11026d45",
+      "dbms": "mysql",
+      "title": "IN operator issue when comparing signed column and the column cast to unsigned",
+      "reported_date": "2019-06-24",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95957"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95957",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95957\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:90b8950a19529dd5c2e096e8af67c0d45740ab88d7b59c0afd9fabb7fedae4dc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:90b8950a19529dd5c2e096e8af67c0d45740ab88d7b59c0afd9fabb7fedae4dc"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95957",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:08497d5e93b6",
+      "dbms": "mysql",
+      "title": "TEXT column used as boolean incorrectly evaluates to false",
+      "reported_date": "2019-06-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95942"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95942",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.17\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95942\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:e98de315a2941157671c903985e5d879d3e6115c3a603c980416871a3172e1cb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e98de315a2941157671c903985e5d879d3e6115c3a603c980416871a3172e1cb"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95942",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:3d61965b76b2",
+      "dbms": "mysql",
+      "title": "The logical operators &, |, and ^ do not ignore newlines in TEXT",
+      "reported_date": "2019-06-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95960"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95960",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95960\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:9e718abba0ce5b30afb2392747996c13159004642dd3a2c88dcdd9432c00787b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9e718abba0ce5b30afb2392747996c13159004642dd3a2c88dcdd9432c00787b"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95960",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:d75c0d21a790",
+      "dbms": "mysql",
+      "title": "Compare that uses UNSIGNED cast and function malfunctions in the MEMORY engine",
+      "reported_date": "2019-06-25",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95964"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95964",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95964\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:86e229ca43e47a366132a1e4101fa3c8502fde6e23e3c6cdbb781a15ec135907",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:86e229ca43e47a366132a1e4101fa3c8502fde6e23e3c6cdbb781a15ec135907"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95964",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:d2a820bb3666",
+      "dbms": "mysql",
+      "title": "Query involving &, <, and AND operators computes incorrect result",
+      "reported_date": "2019-06-25",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95983"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95983",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95983\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:2d5834a5854658fffbfa076c226a0c96ab26b70fc6d301fdf68ea8b2eccb967c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2d5834a5854658fffbfa076c226a0c96ab26b70fc6d301fdf68ea8b2eccb967c"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95983",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:54fb7a2ebc3d",
+      "dbms": "mysql",
+      "title": "SET GLOBAL on rbr_exec_mode fails",
+      "reported_date": "2019-06-25",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95985"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95985",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95985\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:192734942896a8cb8dd0e0e43d605df00410e8ab37236ad189df53a5641d18b8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:192734942896a8cb8dd0e0e43d605df00410e8ab37236ad189df53a5641d18b8"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95985",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:61fac4a93c90",
+      "dbms": "mysql",
+      "title": "Setting sort_buffer_size to a large value causes query to go out of memory",
+      "reported_date": "2019-06-25",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95969"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95969",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.18\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95969\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:19aab54b343ad28c534bd93dba0669ca1e1849d7cc11c6900218c1b42035bd98",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:19aab54b343ad28c534bd93dba0669ca1e1849d7cc11c6900218c1b42035bd98"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95969",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:092da925a649",
+      "dbms": "mysql",
+      "title": "Unexpected result for IN operator and constants",
+      "reported_date": "2019-06-25",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95975"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95975",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95975\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:16e472da87c51b083787f0ea8bf7edec9300fa59193a0d62d0051122a078cc29",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:16e472da87c51b083787f0ea8bf7edec9300fa59193a0d62d0051122a078cc29"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95975",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:c6cab8c51e78",
+      "dbms": "mysql",
+      "title": "Index makes DELETE fail with \"Truncated incorrect DOUBLE value\"",
+      "reported_date": "2019-06-26",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95997"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95997",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95997\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:60638c8339d600637ccc74ce18dce96dd5a4a73f828dd5c4beb2e2e518b5e8a4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:60638c8339d600637ccc74ce18dce96dd5a4a73f828dd5c4beb2e2e518b5e8a4"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95997",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:dd6d64eeed3c",
+      "dbms": "mysql",
+      "title": "Non-unique functional index prevents PRIMARY KEY from being dropped",
+      "reported_date": "2019-06-26",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=96010"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=96010",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=96010\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:d6db9124920e9e1ec3adfc0b0e927293fadfecce34242c34af30d5f0c44e92cc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d6db9124920e9e1ec3adfc0b0e927293fadfecce34242c34af30d5f0c44e92cc"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=96010",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:e02ab80bbfaf",
+      "dbms": "mysql",
+      "title": "Query with GREATEST function malfunctions",
+      "reported_date": "2019-06-26",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://bugs.mysql.com/bug.php?id=96012"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=96012",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"fix_version\": \"8.0.18\",\n        \"links\": {\n            \"email\": \"https://bugs.mysql.com/bug.php?id=96012\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:89e53ced2c7301ca014892084d932c7a0197863d096418414b92784932786267",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:89e53ced2c7301ca014892084d932c7a0197863d096418414b92784932786267"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=96012",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:f8582eebcc4a",
+      "dbms": "mysql",
+      "title": "SET key_cache_* and key_buffer_* variables fails nondeterministically",
+      "reported_date": "2019-06-26",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=95987"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=95987",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/06/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=95987\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:bb8760322447ce57382cccd7d5c5b4de74f341ed3b9d6a08fd67e0d0ef0da580",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:bb8760322447ce57382cccd7d5c5b4de74f341ed3b9d6a08fd67e0d0ef0da580"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=95987",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:0bc23de11b13",
+      "dbms": "mysql",
+      "title": "CAST of STRING with newlines to SIGNED/UNSIGNED returns unexpected result",
+      "reported_date": "2019-07-23",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=96294"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=96294",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/07/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=96294\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:490814d6ca35934f7d46f536e2ae37dd655eddadba6d266310fc20e2e23913f2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:490814d6ca35934f7d46f536e2ae37dd655eddadba6d266310fc20e2e23913f2"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=96294",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:bacc45960e72",
+      "dbms": "mysql",
+      "title": "DISABLE KEYS, DELAY_KEY_WRITE results in \"Data truncated for functional index\"",
+      "reported_date": "2019-07-23",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://bugs.mysql.com/bug.php?id=96295"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=96295",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/07/2019\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugtracker\": \"https://bugs.mysql.com/bug.php?id=96295\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5eb74752c8dbaf97f9076f29e39a8d5a153380d3b0731726e5d9acd5bdc77f4e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5eb74752c8dbaf97f9076f29e39a8d5a153380d3b0731726e5d9acd5bdc77f4e"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=96295",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:7bbc92855342",
+      "dbms": "mysql",
+      "title": "A predicate that compares 0 with -0 yields an incorrect result",
+      "reported_date": "2020-03-31",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://bugs.mysql.com/bug.php?id=99122"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=99122",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"31/03/2020\",\n        \"dbms\": \"MySQL\",\n        \"fixed_version\": \"8.0.21\",\n        \"links\": {\n            \"bugreport\": \"https://bugs.mysql.com/bug.php?id=99122\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:c53c0d314d28a9a989385812bd73561000c6b7df63f38c1a85b48ae637fbb66a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c53c0d314d28a9a989385812bd73561000c6b7df63f38c1a85b48ae637fbb66a"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99122",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:f1c3e9f69460",
+      "dbms": "mysql",
+      "title": "BETWEEN malfunctions when comparing a string containing a newline",
+      "reported_date": "2020-03-31",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://bugs.mysql.com/bug.php?id=99130"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=99130",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"31/03/2020\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugreport\": \"https://bugs.mysql.com/bug.php?id=99130\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:f5a9851e55a7a1b647c3a6011ffce1e995267a1c6d3470b2a983f8f33e6d0304",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f5a9851e55a7a1b647c3a6011ffce1e995267a1c6d3470b2a983f8f33e6d0304"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99130",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:91207cd8b51d",
+      "dbms": "mysql",
+      "title": "BETWEEN malfunctions when comparing large numbers",
+      "reported_date": "2020-03-31",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://bugs.mysql.com/bug.php?id=99135"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=99135",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"31/03/2020\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugreport\": \"https://bugs.mysql.com/bug.php?id=99135\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:31431ffbc279f626b715462372366b616a5b7b1d20757a4e11b2e536c794a2dc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:31431ffbc279f626b715462372366b616a5b7b1d20757a4e11b2e536c794a2dc"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99135",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:ac5666456793",
+      "dbms": "mysql",
+      "title": "Incorrect result for query that uses an AND operator on floats",
+      "reported_date": "2020-03-31",
+      "reported_year": 2020,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://bugs.mysql.com/bug.php?id=99120"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=99120",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"31/03/2020\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugreport\": \"https://bugs.mysql.com/bug.php?id=99120\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:3fd1891d80862bfab333561b26828173fc2baf37a72f50acf311a4e4613d248a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3fd1891d80862bfab333561b26828173fc2baf37a72f50acf311a4e4613d248a"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99120",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:57fc45fa2326",
+      "dbms": "mysql",
+      "title": "Incorrect result when comparing an UNSIGNED INT with a floating-point number",
+      "reported_date": "2020-03-31",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://bugs.mysql.com/bug.php?id=99127"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=99127",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"31/03/2020\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugreport\": \"https://bugs.mysql.com/bug.php?id=99127\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:b89a32f8d7f21b7b81e55b78dbaa54a632ca4cf59d6349eae5856380bca3b560",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b89a32f8d7f21b7b81e55b78dbaa54a632ca4cf59d6349eae5856380bca3b560"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99127",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:4a4800e277b5",
+      "dbms": "mysql",
+      "title": "BETWEEN query malfunctions for special character and TEXT index",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://bugs.mysql.com/bug.php?id=99149"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=99149",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugreport\": \"https://bugs.mysql.com/bug.php?id=99149\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:5d221f8205aa33e36721f8a6b918b4a9a6d7b8d0e969128206e1dcd6b6e268dd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5d221f8205aa33e36721f8a6b918b4a9a6d7b8d0e969128206e1dcd6b6e268dd"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99149",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:ea4be2d1e39f",
+      "dbms": "mysql",
+      "title": "Comparison on FLOAT column and large value malfunctions",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://bugs.mysql.com/bug.php?id=99146"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=99146",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugreport\": \"https://bugs.mysql.com/bug.php?id=99146\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:bfa3ecbc53da2594f29d3589fb15007e099920e84a421c37d7eceea3a4b9539a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:bfa3ecbc53da2594f29d3589fb15007e099920e84a421c37d7eceea3a4b9539a"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99146",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:71e358888ddb",
+      "dbms": "mysql",
+      "title": "Incorrect result when comparing a floating-point number with an integer",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://bugs.mysql.com/bug.php?id=99145"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=99145",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugreport\": \"https://bugs.mysql.com/bug.php?id=99145\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:a08f793461749f114774e8ea2c2a3d0295744059bd685356f615f24c5376671a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a08f793461749f114774e8ea2c2a3d0295744059bd685356f615f24c5376671a"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99145",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:5f8ab2ad1873",
+      "dbms": "mysql",
+      "title": "The IN operator malfunctions for floating-poing numbers",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://bugs.mysql.com/bug.php?id=99150"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=99150",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugreport\": \"https://bugs.mysql.com/bug.php?id=99150\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:0824cf96fb850bd7e5d51496497f5f781c0c9261da060fb224c910b5a8658b99",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0824cf96fb850bd7e5d51496497f5f781c0c9261da060fb224c910b5a8658b99"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99150",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:mysql:1e1cab3a6ac3",
+      "dbms": "mysql",
+      "title": "BETWEEN computes incorrect result when comparing DECIMAL with a string",
+      "reported_date": "2020-04-04",
+      "reported_year": 2020,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://bugs.mysql.com/bug.php?id=99182"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://bugs.mysql.com/bug.php?id=99182",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/04/2020\",\n        \"dbms\": \"MySQL\",\n        \"links\": {\n            \"bugreport\": \"https://bugs.mysql.com/bug.php?id=99182\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:c469023f5ca162be287cc7c3ebf75f852b6fc2269f2bdb2080f2d212e2b97459",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c469023f5ca162be287cc7c3ebf75f852b6fc2269f2bdb2080f2d212e2b97459"
+      },
+      "primary_url": "https://bugs.mysql.com/bug.php?id=99182",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:oceanbase:f543406639c2",
+      "dbms": "oceanbase",
+      "title": "[Bug]: Unexpected result when `RIGHT JOIN` with subqueries",
+      "reported_date": "2025-03-26",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/oceanbase/oceanbase/issues/2247"
+      },
+      "primary_url": "https://github.com/oceanbase/oceanbase/issues/2247",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/oceanbase/issues/2247",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c2 INT, PRIMARY KEY (c2));\nCREATE TABLE t1 (c0 INT, c1 INT, PRIMARY KEY (c0));\nINSERT INTO t0 (c2) VALUES (1);\nINSERT INTO t1 (c0, c1) VALUES (1, 0);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d1ccd6a0b427a1ad89662be08f45a576829373e094a4dbccd76755520e8f87fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/oceanbase/issues/2247",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d1ccd6a0b427a1ad89662be08f45a576829373e094a4dbccd76755520e8f87fc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:oceanbase:d3a093eb702d",
+      "dbms": "oceanbase",
+      "title": "[Bug]: Internal Error",
+      "reported_date": "2025-05-12",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/oceanbase/oceanbase/issues/2272"
+      },
+      "primary_url": "https://github.com/oceanbase/oceanbase/issues/2272",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/oceanbase/issues/2272",
+            "source_type": "github_issue",
+            "excerpt": "create table t2(a varchar(0));",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:231e92ce65b12b1817f56fc5b0148ddb6842ae0046714ffa970ce257bb21f2fd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/oceanbase/issues/2272",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:231e92ce65b12b1817f56fc5b0148ddb6842ae0046714ffa970ce257bb21f2fd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:oceanbase:3cf906fe84ac",
+      "dbms": "oceanbase",
+      "title": "[Bug]: Internal Error of Column Names",
+      "reported_date": "2025-05-12",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/oceanbase/oceanbase/issues/2274"
+      },
+      "primary_url": "https://github.com/oceanbase/oceanbase/issues/2274",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/oceanbase/issues/2274",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]: Internal Error of Column Names",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f344ea7fd2f6c46af83c6ef99d87b7d3ece6e588080a4d72f6971edc0fb1d1fa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/oceanbase/issues/2274",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f344ea7fd2f6c46af83c6ef99d87b7d3ece6e588080a4d72f6971edc0fb1d1fa"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:oceanbase:a31f5b1008c5",
+      "dbms": "oceanbase",
+      "title": "[Bug]: Internal Error on CRC32",
+      "reported_date": "2025-05-12",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/oceanbase/oceanbase/issues/2273"
+      },
+      "primary_url": "https://github.com/oceanbase/oceanbase/issues/2273",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/oceanbase/issues/2273",
+            "source_type": "github_issue",
+            "excerpt": "[Bug]: Internal Error on CRC32",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:dd12f3240a1cf10ea9004ba859908f07d2af5b90e81f77b05777ceb42308e21d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/oceanbase/issues/2273",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dd12f3240a1cf10ea9004ba859908f07d2af5b90e81f77b05777ceb42308e21d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:oxla:0de289a2255e",
+      "dbms": "oxla",
+      "title": "'pg_*' functions that accept INT4 do not work with INT8",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8350"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8350\n'pg_*' functions that accept INT4 do not work with INT8.",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8350 that works around it.",
+            "content_sha256": "sha256:a4584ccdb06f3cf97789356cf4fc4f8092c2208e7fdda8f5631c52d47d092dd4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:a4584ccdb06f3cf97789356cf4fc4f8092c2208e7fdda8f5631c52d47d092dd4"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8350",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:2669851dc54a",
+      "dbms": "oxla",
+      "title": "Adding/Subtracting large integers to/from a date will crash Oxla in Debug builds, and fail silently in Release",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8328"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8328\nAdding/Subtracting large integers to/from a date will crash Oxla in Debug builds, and fail silently in Release.",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8328 that works around it.",
+            "content_sha256": "sha256:46672069488ce60281a7907c8964cc269e88f651ee12df45de452c3026cb528b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:46672069488ce60281a7907c8964cc269e88f651ee12df45de452c3026cb528b"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8328",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:ffc5426591fa",
+      "dbms": "oxla",
+      "title": "Errors caused in JOIN's WHERE condition return internal error non-deterministically",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8323"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8323\nErrors caused in JOIN's WHERE condition return internal error non-deterministically.",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8323 that works around it.",
+            "content_sha256": "sha256:313d9a7fa782c577e384db0a54e8fc49164442ad5b36ca6a38d16a8bb84a4c18",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:313d9a7fa782c577e384db0a54e8fc49164442ad5b36ca6a38d16a8bb84a4c18"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8323",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:6e140deb0018",
+      "dbms": "oxla",
+      "title": "Oxla instantly crashes for REGEX patterns containing invalid symbols",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8329"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8329\nOxla instantly crashes for REGEX patterns containing invalid symbols.",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8329 that works around it.",
+            "content_sha256": "sha256:ff7532cd2555e576c7f3967edd6c9b4a36e646a731d617e0ff42ebe74ab08f2b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:ff7532cd2555e576c7f3967edd6c9b4a36e646a731d617e0ff42ebe74ab08f2b"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8329",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:93bb2ee28e91",
+      "dbms": "oxla",
+      "title": "Oxla parses ~~, !~~, ~~*, !~~* operators incorrectly",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8330"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8330\nOxla parses ~~, !~~, ~~*, !~~* operators incorrectly.",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8330 that works around it.",
+            "content_sha256": "sha256:4b75364a6ccb826ea70a3dad59945ca414c68d47735db9014602a0ef2e34715a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:4b75364a6ccb826ea70a3dad59945ca414c68d47735db9014602a0ef2e34715a"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8330",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:386e4b11905c",
+      "dbms": "oxla",
+      "title": "Oxla returns Internal Compiler Error for NULL literal JSON extract(s)",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8332"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8332\nOxla returns Internal Compiler Error for NULL literal JSON extract(s).",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8332 that works around it.",
+            "content_sha256": "sha256:11b071b5b455b6b39323cdf243d0084dfb9448c0af1225bf00c1c30be391d537",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:11b071b5b455b6b39323cdf243d0084dfb9448c0af1225bf00c1c30be391d537"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8332",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:6fc3c78a6c21",
+      "dbms": "oxla",
+      "title": "PG_TYPEOF function resolves its type into the expression's type instead of text",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8347"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8347\nPG_TYPEOF function resolves its type into the expression's type instead of text.",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8347 that works around it.",
+            "content_sha256": "sha256:3526bd4d70788a0b77ca2824926abd634e68f2b27d62bc2d0a64942140a90b69",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:3526bd4d70788a0b77ca2824926abd634e68f2b27d62bc2d0a64942140a90b69"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8347",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:a1448ce4e314",
+      "dbms": "oxla",
+      "title": "Some FOR_MIN/FOR_MAX queries cause the Oxla to crash",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8349"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8349\nSome FOR_MIN/FOR_MAX queries cause the Oxla to crash.",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8349 that works around it.",
+            "content_sha256": "sha256:8b8aa01ab767626dbf2f01880f8478e673d2a78e943b42223b0cef2c25efbe91",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:8b8aa01ab767626dbf2f01880f8478e673d2a78e943b42223b0cef2c25efbe91"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8349",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:021a2cbd4172",
+      "dbms": "oxla",
+      "title": "Some `ORDER BY` statements in `SELECT DISTINCT` result in incorrect error messages that contain internal  intrinsic functions",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8546"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8546\nSome `ORDER BY` statements in `SELECT DISTINCT` result in incorrect error messages that contain internal\n intrinsic functions.",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8546 that works around it.",
+            "content_sha256": "sha256:7a321649dcf93e2f8583152cae9819f07191ee975efbdc8da53ea488665854f3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:7a321649dcf93e2f8583152cae9819f07191ee975efbdc8da53ea488665854f3"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8546",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:10b751c46f06",
+      "dbms": "oxla",
+      "title": "Valid INT_MIN value results in an \"Integer literal error. Value of literal exceeds range.\" parsing error",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-3376"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-3376\nValid INT_MIN value results in an \"Integer literal error. Value of literal exceeds range.\" parsing error.",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla3376 that works around it.",
+            "content_sha256": "sha256:e35851bb6198cab427ccba67980e6f0c841b9f842c3eeda2866366f3fb4c21b7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:e35851bb6198cab427ccba67980e6f0c841b9f842c3eeda2866366f3fb4c21b7"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-3376",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:71b551437071",
+      "dbms": "oxla",
+      "title": "`WHERE false` queries with multiple FROM tables return _Map_base::at error(s)",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8408"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8408\n`WHERE false` queries with multiple FROM tables return _Map_base::at error(s).",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8408 that works around it.",
+            "content_sha256": "sha256:ba74d6b06b50a1c3d9e923142712a28275fe82c3c3a90ce2fe7b258dd0714732",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:ba74d6b06b50a1c3d9e923142712a28275fe82c3c3a90ce2fe7b258dd0714732"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8408",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:oxla:4fd0e08e166b",
+      "dbms": "oxla",
+      "title": "`timestamp_trunc` returns invalid parsing errors for numeric params",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "unknown",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "report": "https://oxla.atlassian.net/browse/OXLA-8364"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "See: https://oxla.atlassian.net/browse/OXLA-8364\n`timestamp_trunc` returns invalid parsing errors for numeric params.",
+            "excerpt_is_verbatim": true,
+            "note": "redpanda-data/oxla-sqlancer records this bug in its oxla provider, as the field bugOxla8364 that works around it.",
+            "content_sha256": "sha256:507949604a9bcb1eff5f2d08a6ae8241ed36fb35e36472f638854eb3cf1d3454",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/redpanda-data/oxla-sqlancer/blob/main/src/sqlancer/oxla/OxlaBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:507949604a9bcb1eff5f2d08a6ae8241ed36fb35e36472f638854eb3cf1d3454"
+      },
+      "primary_url": "https://oxla.atlassian.net/browse/OXLA-8364",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:postgresql:8849e9de09a8",
+      "dbms": "postgresql",
+      "title": "ALTER TABLE SET WITH OIDS fails after failed CONCURRENTLY index creation",
+      "reported_date": "2019-06-02",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA4XYhc-qyCgJqwwgMGZDWAyeH821oa5oMzm_HEifZ4BeA%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA4XYhc-qyCgJqwwgMGZDWAyeH821oa5oMzm_HEifZ4BeA%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/06/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA4XYhc-qyCgJqwwgMGZDWAyeH821oa5oMzm_HEifZ4BeA%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:2532f63e5bafcca134251aa40cca7636c995b3ffae888e214504db723ec899b2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2532f63e5bafcca134251aa40cca7636c995b3ffae888e214504db723ec899b2"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA4XYhc-qyCgJqwwgMGZDWAyeH821oa5oMzm_HEifZ4BeA%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:168617eb9822",
+      "dbms": "postgresql",
+      "title": "VACUUM FULL results in deadlock",
+      "reported_date": "2019-06-29",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA6pL%2B7Xm_NXHLenxffe3tCr3gTamVdr7zPjcWqW0RFM-A%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA6pL%2B7Xm_NXHLenxffe3tCr3gTamVdr7zPjcWqW0RFM-A%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/06/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA6pL%2B7Xm_NXHLenxffe3tCr3gTamVdr7zPjcWqW0RFM-A%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5c3ddfac602efc73d9d8904e51fbfab37efbb03da7a0616c97992987b7eebabc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5c3ddfac602efc73d9d8904e51fbfab37efbb03da7a0616c97992987b7eebabc"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA6pL%2B7Xm_NXHLenxffe3tCr3gTamVdr7zPjcWqW0RFM-A%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:5983b2f1ea14",
+      "dbms": "postgresql",
+      "title": "GROUP BY and inheritance issue",
+      "reported_date": "2019-07-02",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA7VLKf_vEr6kLF3MnWSA9LToJYncgpNX2tQ-oWzYCBQAw%40mail.gmail.com",
+        "fix": "https://github.com/postgres/postgres/commit/a5be4062f7bf2ae9487c5a31ee337a56425cdc84"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA7VLKf_vEr6kLF3MnWSA9LToJYncgpNX2tQ-oWzYCBQAw%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA7VLKf_vEr6kLF3MnWSA9LToJYncgpNX2tQ-oWzYCBQAw%40mail.gmail.com\",\n            \"fix\": \"https://github.com/postgres/postgres/commit/a5be4062f7bf2ae9487c5a31ee337a56425cdc84\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:997ec70b2acc548ec8ad645f9921feeb66324abf924f48066043bad05858713e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:997ec70b2acc548ec8ad645f9921feeb66324abf924f48066043bad05858713e"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA7VLKf_vEr6kLF3MnWSA9LToJYncgpNX2tQ-oWzYCBQAw%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:50ace8a7643b",
+      "dbms": "postgresql",
+      "title": "SELECT results in \"ERROR: index key does not match expected index column\"",
+      "reported_date": "2019-07-02",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA4CV21mUfwjQj3va7MWvqB2EAdJT31%3Duf_yArBRVzQjYw%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA4CV21mUfwjQj3va7MWvqB2EAdJT31%3Duf_yArBRVzQjYw%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA4CV21mUfwjQj3va7MWvqB2EAdJT31%3Duf_yArBRVzQjYw%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c1f9e5126a74f6f2a060a07e38cee361e83d059eec0c7f3ad3adc9fc70ea7459",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c1f9e5126a74f6f2a060a07e38cee361e83d059eec0c7f3ad3adc9fc70ea7459"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA4CV21mUfwjQj3va7MWvqB2EAdJT31%3Duf_yArBRVzQjYw%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:9636cb5d2a96",
+      "dbms": "postgresql",
+      "title": "ALTER TABLE results in \"ERROR: could not open relation with OID 43707388\"",
+      "reported_date": "2019-07-04",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA4hkFSV_Y%3DsW_vNcYgKFEoq0WL5GtrBWEHUZnCqSqjhAA%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA4hkFSV_Y%3DsW_vNcYgKFEoq0WL5GtrBWEHUZnCqSqjhAA%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"4/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA4hkFSV_Y%3DsW_vNcYgKFEoq0WL5GtrBWEHUZnCqSqjhAA%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:2b028ffd69a9444b937379a8d945de95bcc8436f12c828f79a6eb6cf3a52bbf7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2b028ffd69a9444b937379a8d945de95bcc8436f12c828f79a6eb6cf3a52bbf7"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA4hkFSV_Y%3DsW_vNcYgKFEoq0WL5GtrBWEHUZnCqSqjhAA%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:1da841a6f6d1",
+      "dbms": "postgresql",
+      "title": "DISCARD TEMP results in \"ERROR: cache lookup failed for type 0\"",
+      "reported_date": "2019-07-04",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA4JKCPFrdrAbOs7XBiCyD61XJxeNav4LefkSmBLQ-Vobg%40mail.gmail.com",
+        "email 2": "https://www.postgresql.org/message-id/31920.1562526703%40sss.pgh.pa.us",
+        "fix": "https://github.com/postgres/postgres/commit/a0555ddab9b672a04681ce0d9f6c94104c01b15f"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA4JKCPFrdrAbOs7XBiCyD61XJxeNav4LefkSmBLQ-Vobg%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"4/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA4JKCPFrdrAbOs7XBiCyD61XJxeNav4LefkSmBLQ-Vobg%40mail.gmail.com\",\n            \"email 2\": \"https://www.postgresql.org/message-id/31920.1562526703%40sss.pgh.pa.us\",\n            \"fix\": \"https://github.com/postgres/postgres/commit/a0555ddab9b672a04681ce0d9f6c94104c01b15f\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:69543cf20b87557df2cb0467c7e78a34cec0eae68037218e665a6ba788c1732b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:69543cf20b87557df2cb0467c7e78a34cec0eae68037218e665a6ba788c1732b"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA4JKCPFrdrAbOs7XBiCyD61XJxeNav4LefkSmBLQ-Vobg%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:90c0a623c6e1",
+      "dbms": "postgresql",
+      "title": "Issue with CHAR column and \"column LIKE column\" condition",
+      "reported_date": "2019-07-07",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA7mwhq7Xu9ZsaYOZ6aeodMiomKX0aMm8bbtYE%3DSyqSs_Q%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA7mwhq7Xu9ZsaYOZ6aeodMiomKX0aMm8bbtYE%3DSyqSs_Q%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA7mwhq7Xu9ZsaYOZ6aeodMiomKX0aMm8bbtYE%3DSyqSs_Q%40mail.gmail.com\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:91ff414c77a4a0e38f58f6520dccc34a6009e00341da1019e18d6f6b2d119cd4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:91ff414c77a4a0e38f58f6520dccc34a6009e00341da1019e18d6f6b2d119cd4"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA7mwhq7Xu9ZsaYOZ6aeodMiomKX0aMm8bbtYE%3DSyqSs_Q%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:5c9e5d2cde90",
+      "dbms": "postgresql",
+      "title": "VACUUM FULL results in ERROR: integer out of range",
+      "reported_date": "2019-07-07",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA7VT2YvyRz1FWXtr%2Bp9ayUxGEqB8xJOsJT%2BsywXVKVvGw%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA7VT2YvyRz1FWXtr%2Bp9ayUxGEqB8xJOsJT%2BsywXVKVvGw%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA7VT2YvyRz1FWXtr%2Bp9ayUxGEqB8xJOsJT%2BsywXVKVvGw%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7c37ac95635c6fcb33824652bf5ce6e1a8d1d704475ac9452610fe013db14ce2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7c37ac95635c6fcb33824652bf5ce6e1a8d1d704475ac9452610fe013db14ce2"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA7VT2YvyRz1FWXtr%2Bp9ayUxGEqB8xJOsJT%2BsywXVKVvGw%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:731f6d8dff14",
+      "dbms": "postgresql",
+      "title": "ERROR: found unexpected null value in index",
+      "reported_date": "2019-07-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA7W4NWEhCvftdV6_8bbm2vgypi5nuxfnSEJQqVKFSUoMg%40mail.gmail.com",
+        "fix": "https://github.com/postgres/postgres/commit/d3751adcf14d3baacc9738ee9ce869dc1c31d7ad"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA7W4NWEhCvftdV6_8bbm2vgypi5nuxfnSEJQqVKFSUoMg%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA7W4NWEhCvftdV6_8bbm2vgypi5nuxfnSEJQqVKFSUoMg%40mail.gmail.com\",\n            \"fix\": \"https://github.com/postgres/postgres/commit/d3751adcf14d3baacc9738ee9ce869dc1c31d7ad\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e36de6389576b79f622d52afedf222bb9165c5388290207550f784d32f5745b7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e36de6389576b79f622d52afedf222bb9165c5388290207550f784d32f5745b7"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA7W4NWEhCvftdV6_8bbm2vgypi5nuxfnSEJQqVKFSUoMg%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:f0e0a3c75966",
+      "dbms": "postgresql",
+      "title": "ERROR: negative bitmapset member not allowed in SELECT",
+      "reported_date": "2019-07-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA7fmtm3kR%2BY6Mgr9djU3WeOsMzQLWtNPMvoEVSm%3Dr9XaQ%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA7fmtm3kR%2BY6Mgr9djU3WeOsMzQLWtNPMvoEVSm%3Dr9XaQ%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA7fmtm3kR%2BY6Mgr9djU3WeOsMzQLWtNPMvoEVSm%3Dr9XaQ%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:67986483b93a73d3c1e4d890ed840b8f6cb9c88f6758f509ee019a3b5e0e6795",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:67986483b93a73d3c1e4d890ed840b8f6cb9c88f6758f509ee019a3b5e0e6795"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA7fmtm3kR%2BY6Mgr9djU3WeOsMzQLWtNPMvoEVSm%3Dr9XaQ%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:8406590bdadd",
+      "dbms": "postgresql",
+      "title": "Generated column and string concatenation issue",
+      "reported_date": "2019-07-10",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA6AD9OLziFW%2Bio1HgN8q_XH0o1Y5RyufXYO5%3D0fnhG5zQ%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA6AD9OLziFW%2Bio1HgN8q_XH0o1Y5RyufXYO5%3D0fnhG5zQ%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA6AD9OLziFW%2Bio1HgN8q_XH0o1Y5RyufXYO5%3D0fnhG5zQ%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5f43fcf060ed7cf1c73c63ac77ad536bc26a6a79a45afdd2e8c5d05a18126874",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5f43fcf060ed7cf1c73c63ac77ad536bc26a6a79a45afdd2e8c5d05a18126874"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA6AD9OLziFW%2Bio1HgN8q_XH0o1Y5RyufXYO5%3D0fnhG5zQ%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:6e37a2988a09",
+      "dbms": "postgresql",
+      "title": "SEGFAULT in 12beta2 release",
+      "reported_date": "2019-07-10",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "SEGFAULT in 12beta2 release",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a PostgreSQL bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:6e37a2988a094b9300133bc2fa803c44e68a91324df32316b2fcb49b0ce86cb8",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:6e37a2988a094b9300133bc2fa803c44e68a91324df32316b2fcb49b0ce86cb8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:f66a8084aa0d",
+      "dbms": "postgresql",
+      "title": "SEGFAULT in 12beta2 release",
+      "reported_date": "2019-07-10",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "curated_entry": "https://github.com/sqlancer/bugs/blob/master/bugs.json"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:318eb8c32ebb0152fd2210192a78de3add913babed620428580423d2298fd564",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:318eb8c32ebb0152fd2210192a78de3add913babed620428580423d2298fd564"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:028dab7f56ed",
+      "dbms": "postgresql",
+      "title": "Stack buffer overflow in 12beta2 release",
+      "reported_date": "2019-07-11",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "curated_entry": "https://github.com/sqlancer/bugs/blob/master/bugs.json"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:54e968709d0bfa9e1a1f167d00f11ccfc69c8b98347db286d698fc5337ace89d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:54e968709d0bfa9e1a1f167d00f11ccfc69c8b98347db286d698fc5337ace89d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:dcec171cc4d8",
+      "dbms": "postgresql",
+      "title": "Stack buffer overflow in 12beta2 release",
+      "reported_date": "2019-07-11",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Stack buffer overflow in 12beta2 release",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a PostgreSQL bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:dcec171cc4d809f469eba9601098e1c5dcb4c94313eee80853b2cd7e4936d82f",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:dcec171cc4d809f469eba9601098e1c5dcb4c94313eee80853b2cd7e4936d82f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:be51597ffb8b",
+      "dbms": "postgresql",
+      "title": "SELECT with COLLATE results in segfault on trunk and 12 Beta 2",
+      "reported_date": "2019-07-14",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA40Fe9%3DA3wQ2PHWy8VZg8%3DGpD6dxQXeXVDx6HAhRSPeRA%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA40Fe9%3DA3wQ2PHWy8VZg8%3DGpD6dxQXeXVDx6HAhRSPeRA%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA40Fe9%3DA3wQ2PHWy8VZg8%3DGpD6dxQXeXVDx6HAhRSPeRA%40mail.gmail.com\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:59770361a1fc8a4a8a711128109f392887989678d60b0f7017f500f35fa1e5a6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:59770361a1fc8a4a8a711128109f392887989678d60b0f7017f500f35fa1e5a6"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA40Fe9%3DA3wQ2PHWy8VZg8%3DGpD6dxQXeXVDx6HAhRSPeRA%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:062753c781d0",
+      "dbms": "postgresql",
+      "title": "UPDATE causes segfault on trunk",
+      "reported_date": "2019-07-15",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA4XKjVv86MTUHZYzzdQAgPOPJDSuA5HrD8%2BxkNQsnAsAQ%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA4XKjVv86MTUHZYzzdQAgPOPJDSuA5HrD8%2BxkNQsnAsAQ%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA4XKjVv86MTUHZYzzdQAgPOPJDSuA5HrD8%2BxkNQsnAsAQ%40mail.gmail.com\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:152456dfcc7aa0a3fa0887e5396b40c4faae6aa4f8b7ba35dca5eff44c2a48ff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:152456dfcc7aa0a3fa0887e5396b40c4faae6aa4f8b7ba35dca5eff44c2a48ff"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA4XKjVv86MTUHZYzzdQAgPOPJDSuA5HrD8%2BxkNQsnAsAQ%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:dec7b504c9b2",
+      "dbms": "postgresql",
+      "title": "BETWEEN SYMMETRIC condition results in \"row is too big: ..., maximum size 8160\"",
+      "reported_date": "2019-07-16",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA6zYX7qfkGGDfqeFwuh6HbD4B-0fmOfx13Jm4xsp3s%3D-Q%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA6zYX7qfkGGDfqeFwuh6HbD4B-0fmOfx13Jm4xsp3s%3D-Q%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"16/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA6zYX7qfkGGDfqeFwuh6HbD4B-0fmOfx13Jm4xsp3s%3D-Q%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:463f0d6112d4f514bc995bc9482378a64298c2dad999533e51a494e49daa3939",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:463f0d6112d4f514bc995bc9482378a64298c2dad999533e51a494e49daa3939"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA6zYX7qfkGGDfqeFwuh6HbD4B-0fmOfx13Jm4xsp3s%3D-Q%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:e84cbf441a7e",
+      "dbms": "postgresql",
+      "title": "REINDEX CONCURRENTLY causes ALTER TABLE to fail",
+      "reported_date": "2019-07-17",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/CA%2Bu7OA5Hp0ra235F3czPom_FyAd-3%2BXwSJmX95r1%2BsRPOJc9VQ%40mail.gmail.com",
+        "fix": "https://github.com/postgres/postgres/commit/a904abe2e284f570168839e52e18ef0b7f26179d"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA5Hp0ra235F3czPom_FyAd-3%2BXwSJmX95r1%2BsRPOJc9VQ%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"17/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/CA%2Bu7OA5Hp0ra235F3czPom_FyAd-3%2BXwSJmX95r1%2BsRPOJc9VQ%40mail.gmail.com\",\n            \"fix\": \"https://github.com/postgres/postgres/commit/a904abe2e284f570168839e52e18ef0b7f26179d\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:b66d0e308001d10051aa9bb5ce1bc8925c5ab3b636a185dd4ff8089072cf9f6e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b66d0e308001d10051aa9bb5ce1bc8925c5ab3b636a185dd4ff8089072cf9f6e"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA5Hp0ra235F3czPom_FyAd-3%2BXwSJmX95r1%2BsRPOJc9VQ%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:7d0ff702c2b7",
+      "dbms": "postgresql",
+      "title": "ADD CHECK fails for parent table if column used in CHECK is fully-qualified",
+      "reported_date": "2019-07-24",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA6Zs6bZmuakbLv4%3D7VBOYpBsyswpb7A3hN%3Dn%2BKO0_z0pQ%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA6Zs6bZmuakbLv4%3D7VBOYpBsyswpb7A3hN%3Dn%2BKO0_z0pQ%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/flat/CA%2Bu7OA6Zs6bZmuakbLv4%3D7VBOYpBsyswpb7A3hN%3Dn%2BKO0_z0pQ%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a539831280abadd586f73447a9d9c1d9b1a212a4954213e97c39f03827cd2753",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a539831280abadd586f73447a9d9c1d9b1a212a4954213e97c39f03827cd2753"
+      },
+      "primary_url": "https://postgresql.org/message-id/flat/CA%2Bu7OA6Zs6bZmuakbLv4%3D7VBOYpBsyswpb7A3hN%3Dn%2BKO0_z0pQ%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:258e70a92df0",
+      "dbms": "postgresql",
+      "title": "ANALYZE on parent table results in an error \"tuple already updated by self\"",
+      "reported_date": "2019-07-24",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://www.postgresql.org/message-id/CA%2Bu7OA69sgyCE0VhEpgZLBkR4X0frA2%3Dar8brntoH1vcvi-%2BWA%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/CA%2Bu7OA69sgyCE0VhEpgZLBkR4X0frA2%3Dar8brntoH1vcvi-%2BWA%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"bugtracker\": \"https://www.postgresql.org/message-id/CA%2Bu7OA69sgyCE0VhEpgZLBkR4X0frA2%3Dar8brntoH1vcvi-%2BWA%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:9e317a5d04c051028f3379d1a22ea970888cd9a0592abfae3d1d4df176c90235",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9e317a5d04c051028f3379d1a22ea970888cd9a0592abfae3d1d4df176c90235"
+      },
+      "primary_url": "https://postgresql.org/message-id/CA%2Bu7OA69sgyCE0VhEpgZLBkR4X0frA2%3Dar8brntoH1vcvi-%2BWA%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:92b29e8fb93f",
+      "dbms": "postgresql",
+      "title": "Attribute has wrong type in ALTER TABLE",
+      "reported_date": "2019-07-24",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA6RmQLwvXGxx0nGt3T79Ka5MkOJs4_qcdx6X5c-rR_yNg%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA6RmQLwvXGxx0nGt3T79Ka5MkOJs4_qcdx6X5c-rR_yNg%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"bugtracker\": \"https://www.postgresql.org/message-id/flat/CA%2Bu7OA6RmQLwvXGxx0nGt3T79Ka5MkOJs4_qcdx6X5c-rR_yNg%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:09a364608c4858584af5442b28bc30167de00adc29739bbc7437e9fb24a6634f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:09a364608c4858584af5442b28bc30167de00adc29739bbc7437e9fb24a6634f"
+      },
+      "primary_url": "https://postgresql.org/message-id/flat/CA%2Bu7OA6RmQLwvXGxx0nGt3T79Ka5MkOJs4_qcdx6X5c-rR_yNg%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:e98047c1f2e4",
+      "dbms": "postgresql",
+      "title": "Multiple inheritance and ALTER TABLE issue",
+      "reported_date": "2019-07-27",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA4qogDv9rz1HAb-ADxttXYPqQdUdPY_yd4kCzywNxRQXA%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA4qogDv9rz1HAb-ADxttXYPqQdUdPY_yd4kCzywNxRQXA%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"27/07/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/flat/CA%2Bu7OA4qogDv9rz1HAb-ADxttXYPqQdUdPY_yd4kCzywNxRQXA%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ff8e1600d1e411e8431c50afe062a6c9c8dad05f8b8034bdb85c3c3724cbfa04",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ff8e1600d1e411e8431c50afe062a6c9c8dad05f8b8034bdb85c3c3724cbfa04"
+      },
+      "primary_url": "https://postgresql.org/message-id/flat/CA%2Bu7OA4qogDv9rz1HAb-ADxttXYPqQdUdPY_yd4kCzywNxRQXA%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:066f9849749d",
+      "dbms": "postgresql",
+      "title": "INSERT INTO causes segfault on trunk",
+      "reported_date": "2019-09-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "INSERT INTO causes segfault on trunk",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a PostgreSQL bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:066f9849749d588e2e5fa2eb9b3f4765dcb13e5c3af33580a9853acc42d5a420",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:066f9849749d588e2e5fa2eb9b3f4765dcb13e5c3af33580a9853acc42d5a420"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:db1efdea095f",
+      "dbms": "postgresql",
+      "title": "INSERT INTO causes segfault on trunk",
+      "reported_date": "2019-09-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "curated_entry": "https://github.com/sqlancer/bugs/blob/master/bugs.json"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/09/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:b4ad8ea3c62a6495833b3196072d21474a598651a0849a88c0a5b602b1f14959",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b4ad8ea3c62a6495833b3196072d21474a598651a0849a88c0a5b602b1f14959"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:5288cc7e3e1c",
+      "dbms": "postgresql",
+      "title": "REINDEX CONCURRENTLY unexpectedly fails",
+      "reported_date": "2019-11-13",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA6gP7YAeCguyseusYcc%3DuR8%2BypjCcgDDCTzjQ%2Bk6S9ksQ%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA6gP7YAeCguyseusYcc%3DuR8%2BypjCcgDDCTzjQ%2Bk6S9ksQ%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"13/11/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/flat/CA%2Bu7OA6gP7YAeCguyseusYcc%3DuR8%2BypjCcgDDCTzjQ%2Bk6S9ksQ%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f6cf3b6610a26253d1cb51ca57f1fe931073887c6702276b0bc3108622c90647",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f6cf3b6610a26253d1cb51ca57f1fe931073887c6702276b0bc3108622c90647"
+      },
+      "primary_url": "https://postgresql.org/message-id/flat/CA%2Bu7OA6gP7YAeCguyseusYcc%3DuR8%2BypjCcgDDCTzjQ%2Bk6S9ksQ%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:5b2b8468b534",
+      "dbms": "postgresql",
+      "title": "Unexpected \"cache lookup failed for collation 0\" failure",
+      "reported_date": "2019-11-13",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA4HOjtymxAbuGNh4-X_2R0Lw5n01tzvP8E5-i-2gQXYWA%40mail.gmail.com",
+        "fix": "https://github.com/postgres/postgres/commit/d57d61533a2b5b27b60cc9024c54688390871bf6"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA4HOjtymxAbuGNh4-X_2R0Lw5n01tzvP8E5-i-2gQXYWA%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"13/11/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/flat/CA%2Bu7OA4HOjtymxAbuGNh4-X_2R0Lw5n01tzvP8E5-i-2gQXYWA%40mail.gmail.com\",\n            \"fix\": \"https://github.com/postgres/postgres/commit/d57d61533a2b5b27b60cc9024c54688390871bf6\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:539df179cda4b7f1e2da7a203de0492400aef5fffcf63ddf836cfbe78ddd6987",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:539df179cda4b7f1e2da7a203de0492400aef5fffcf63ddf836cfbe78ddd6987"
+      },
+      "primary_url": "https://postgresql.org/message-id/flat/CA%2Bu7OA4HOjtymxAbuGNh4-X_2R0Lw5n01tzvP8E5-i-2gQXYWA%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:d9a6e4e7afaf",
+      "dbms": "postgresql",
+      "title": "ALTER TABLE results in \"could not find cast from 3904 to 3831\"",
+      "reported_date": "2019-11-16",
+      "reported_year": 2019,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA4n2Q8G1DD5MEp%2Bmu08ayB%2BULRFrG4s_v3LhOj8f4bX-g%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA4n2Q8G1DD5MEp%2Bmu08ayB%2BULRFrG4s_v3LhOj8f4bX-g%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"16/11/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/flat/CA%2Bu7OA4n2Q8G1DD5MEp%2Bmu08ayB%2BULRFrG4s_v3LhOj8f4bX-g%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5c7f30bbe9dec52736bac780d3a5f5b2eda1eedfd02387662e5c58ca6d235d8a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5c7f30bbe9dec52736bac780d3a5f5b2eda1eedfd02387662e5c58ca6d235d8a"
+      },
+      "primary_url": "https://postgresql.org/message-id/flat/CA%2Bu7OA4n2Q8G1DD5MEp%2Bmu08ayB%2BULRFrG4s_v3LhOj8f4bX-g%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:0d1c395fb3f1",
+      "dbms": "postgresql",
+      "title": "Failed assertion clauses != NIL",
+      "reported_date": "2019-11-19",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA7H5rcE2%3D8f263w4NZD6ipO_XOrYB816nuLXbmSTH9pQQ%40mail.gmail.com",
+        "fix": "https://github.com/postgres/postgres/commit/c676e659b246f94d571b57b559f80cb2dc03e73b"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA7H5rcE2%3D8f263w4NZD6ipO_XOrYB816nuLXbmSTH9pQQ%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/11/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/flat/CA%2Bu7OA7H5rcE2%3D8f263w4NZD6ipO_XOrYB816nuLXbmSTH9pQQ%40mail.gmail.com\",\n            \"fix\": \"https://github.com/postgres/postgres/commit/c676e659b246f94d571b57b559f80cb2dc03e73b\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:227e760f464cf5dc2c66b8ba215dbd417d2af0c39664c28a19239849e8f9f9b4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:227e760f464cf5dc2c66b8ba215dbd417d2af0c39664c28a19239849e8f9f9b4"
+      },
+      "primary_url": "https://postgresql.org/message-id/flat/CA%2Bu7OA7H5rcE2%3D8f263w4NZD6ipO_XOrYB816nuLXbmSTH9pQQ%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:a40a6052b2da",
+      "dbms": "postgresql",
+      "title": "No = operator for opfamily 426",
+      "reported_date": "2019-11-19",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA7nnGYy8rY0vdTe811NuA%2BFrr9nbcBO9u2Z%2BJxqNaud%2Bg%40mail.gmail.com",
+        "fix": "https://github.com/postgres/postgres/commit/b3c265d7be42484bd0ab4a9c0a920289e8f5c995",
+        "improvement 1": "https://github.com/postgres/postgres/commit/2ddedcafca116c99e08c777ab2ab3a4de6f00c7e",
+        "improvement 2": "https://github.com/postgres/postgres/commit/9ff5b699ed3e2d922ff6f5660e53b51bb5db983c"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA7nnGYy8rY0vdTe811NuA%2BFrr9nbcBO9u2Z%2BJxqNaud%2Bg%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/11/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/flat/CA%2Bu7OA7nnGYy8rY0vdTe811NuA%2BFrr9nbcBO9u2Z%2BJxqNaud%2Bg%40mail.gmail.com\",\n            \"fix\": \"https://github.com/postgres/postgres/commit/b3c265d7be42484bd0ab4a9c0a920289e8f5c995\",\n            \"improvement 1\": \"https://github.com/postgres/postgres/commit/2ddedcafca116c99e08c777ab2ab3a4de6f00c7e\",\n            \"improvement 2\": \"https://github.com/postgres/postgres/commit/9ff5b699ed3e2d922ff6f5660e53b51bb5db983c\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:42b6cd74d1da3ccd392b35d31a855e9bc83a1bff91abbad3bdf4c7e8a80a1d32",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:42b6cd74d1da3ccd392b35d31a855e9bc83a1bff91abbad3bdf4c7e8a80a1d32"
+      },
+      "primary_url": "https://postgresql.org/message-id/flat/CA%2Bu7OA7nnGYy8rY0vdTe811NuA%2BFrr9nbcBO9u2Z%2BJxqNaud%2Bg%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:beef6b2d6101",
+      "dbms": "postgresql",
+      "title": "ALTER TABLE fails when changing column type due to index with bit_ops opclass",
+      "reported_date": "2019-11-20",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA4%3Dvs9LjjP_643kgLRENv2eW%2B8v%2B5M8QGsY3Sg1K_18fg%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA4%3Dvs9LjjP_643kgLRENv2eW%2B8v%2B5M8QGsY3Sg1K_18fg%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/11/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/flat/CA%2Bu7OA4%3Dvs9LjjP_643kgLRENv2eW%2B8v%2B5M8QGsY3Sg1K_18fg%40mail.gmail.com\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:d4a513860f4febfff2493318522e7c463c9e8686d94d32bd4faf35c3e520363a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d4a513860f4febfff2493318522e7c463c9e8686d94d32bd4faf35c3e520363a"
+      },
+      "primary_url": "https://postgresql.org/message-id/flat/CA%2Bu7OA4%3Dvs9LjjP_643kgLRENv2eW%2B8v%2B5M8QGsY3Sg1K_18fg%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:4bc2b573648f",
+      "dbms": "postgresql",
+      "title": "Fetching from information_schema.tables results in segfault",
+      "reported_date": "2019-11-20",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "fix": "https://github.com/postgres/postgres/commit/4a0aab14dcb35550b55e623a3c194442c5666084",
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/postgres/postgres/commit/4a0aab14dcb35550b55e623a3c194442c5666084",
+            "source_type": "commit",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/11/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"fix\": \"https://github.com/postgres/postgres/commit/4a0aab14dcb35550b55e623a3c194442c5666084\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:54dae6acdc61dc1b979516f8b17c8a06b3e60ea6b907e6018728e9c1aef59893",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Fetching from information_schema.tables results in segfault",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a PostgreSQL bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:1dc5497a4d620a813666068decd60b7ba14754dfb849f9bc603ca620e6a04e32",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:54dae6acdc61dc1b979516f8b17c8a06b3e60ea6b907e6018728e9c1aef59893"
+      },
+      "primary_url": "https://github.com/postgres/postgres/commit/4a0aab14dcb35550b55e623a3c194442c5666084",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:1d7d3e0d57d5",
+      "dbms": "postgresql",
+      "title": "Segfault when creating constant bit_length() index on TEMP table",
+      "reported_date": "2019-11-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "fix": "https://github.com/postgres/postgres/commit/c35b714caff008c875b484656de7d168a7bc45f9",
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/postgres/postgres/commit/c35b714caff008c875b484656de7d168a7bc45f9",
+            "source_type": "commit",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/11/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"fix\": \"https://github.com/postgres/postgres/commit/c35b714caff008c875b484656de7d168a7bc45f9\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e569bc40feb159604a396c1dccefb51fc16b8d1295b938c573de4cd3ff41fb36",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Segfault when creating constant bit_length() index on TEMP table",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a PostgreSQL bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:5081a4db9420a45acf7e0e9913ef570dac86b5c9bef09f605082c0962c200e95",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e569bc40feb159604a396c1dccefb51fc16b8d1295b938c573de4cd3ff41fb36"
+      },
+      "primary_url": "https://github.com/postgres/postgres/commit/c35b714caff008c875b484656de7d168a7bc45f9",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:postgresql:36de0b677072",
+      "dbms": "postgresql",
+      "title": "FailedAssertion(\"!OidIsValid(def->collOid)\", File: \"view.c\", Line: 89)",
+      "reported_date": "2019-12-02",
+      "reported_year": 2019,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA5TwqWTrvRoFOF1Vipf5WpfcbQYwtLGzNVXGwa6Ptor9w%40mail.gmail.com"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.postgresql.org/message-id/flat/CA%2Bu7OA5TwqWTrvRoFOF1Vipf5WpfcbQYwtLGzNVXGwa6Ptor9w%40mail.gmail.com",
+            "source_type": "other",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/12/2019\",\n        \"dbms\": \"PostgreSQL\",\n        \"links\": {\n            \"email\": \"https://www.postgresql.org/message-id/flat/CA%2Bu7OA5TwqWTrvRoFOF1Vipf5WpfcbQYwtLGzNVXGwa6Ptor9w%40mail.gmail.com\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:2d740b8a2a86a17b007e44ee3e5576bb82e110390d8db1927bf95e976cc0571a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2d740b8a2a86a17b007e44ee3e5576bb82e110390d8db1927bf95e976cc0571a"
+      },
+      "primary_url": "https://postgresql.org/message-id/flat/CA%2Bu7OA5TwqWTrvRoFOF1Vipf5WpfcbQYwtLGzNVXGwa6Ptor9w%40mail.gmail.com",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:presto:3e0a7dccd473",
+      "dbms": "presto",
+      "title": "Compiler failed with complex CASE expression using SOME comparison",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+        "report": "https://github.com/prestodb/presto/issues/27609"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/prestodb/presto/issues/27609",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its presto provider, as the field bugCompilerFailed that works around it.",
+            "content_sha256": "sha256:3e0a7dccd4735440b23bda8c526fe96f60568078254e9fd2c3ab19b096f38149",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:3e0a7dccd4735440b23bda8c526fe96f60568078254e9fd2c3ab19b096f38149"
+      },
+      "primary_url": "https://github.com/prestodb/presto/issues/27609",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:presto:39c1b818faa5",
+      "dbms": "presto",
+      "title": "Constant folding fails during case statement with mixed long decimal and integral types",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+        "report": "https://github.com/prestodb/presto/issues/23324"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/prestodb/presto/issues/23324",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its presto provider, as the field bug23324 that works around it.",
+            "content_sha256": "sha256:39c1b818faa55d3156605fe7a542c7c30d18b002ae9138ac50989c66d156d890",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:39c1b818faa55d3156605fe7a542c7c30d18b002ae9138ac50989c66d156d890"
+      },
+      "primary_url": "https://github.com/prestodb/presto/issues/23324",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:presto:974b2cfe9f3e",
+      "dbms": "presto",
+      "title": "Query fail with error `failed: at index 1`",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+        "report": "https://github.com/prestodb/presto/issues/23613"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/prestodb/presto/issues/23613",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its presto provider, as the field bug23613 that works around it.",
+            "content_sha256": "sha256:974b2cfe9f3e3befde096ae5ef6c56a81e7e0690882529e53f4469cb82b08c0a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:974b2cfe9f3e3befde096ae5ef6c56a81e7e0690882529e53f4469cb82b08c0a"
+      },
+      "primary_url": "https://github.com/prestodb/presto/issues/23613",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:presto:27fa460e9dd4",
+      "dbms": "presto",
+      "title": "VerifyError: Bad type on operand stack in CASE with mixed numeric types",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+        "report": "https://github.com/prestodb/presto/issues/27608"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/prestodb/presto/issues/27608",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its presto provider, as the field bugVerifyError that works around it.",
+            "content_sha256": "sha256:27fa460e9dd4ca5fe65bd7d4ea12ae555b5d805071da6f4a7df251a51acd31be",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/presto/PrestoBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:27fa460e9dd4ca5fe65bd7d4ea12ae555b5d805071da6f4a7df251a51acd31be"
+      },
+      "primary_url": "https://github.com/prestodb/presto/issues/27608",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:questdb:7f5313d57ded",
+      "dbms": "questdb",
+      "title": "CREATE TABLE with INDEX causing error",
+      "reported_date": "2022-10-26",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "SuriZhang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/2689"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/2689",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/2689",
+            "source_type": "github_issue",
+            "excerpt": "I was implementing sqlancer support for questdb, and have found the bug below:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2d0bc69aacffe3e0634acd6b7e79805d69f988578c2917418a432d07a5012fd7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/2689",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2d0bc69aacffe3e0634acd6b7e79805d69f988578c2917418a432d07a5012fd7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:276732e5fa2b",
+      "dbms": "questdb",
+      "title": "SQL engine issues found by SQLancer",
+      "reported_date": "2022-11-03",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "puzpuzpuz",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/2721"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/2721",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/2721",
+            "source_type": "github_issue",
+            "excerpt": "SQL engine issues found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f6d872b19dfe161822c11b338609428368e9cc35413c3974bb8ccbbfb143b1ef",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/2721",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f6d872b19dfe161822c11b338609428368e9cc35413c3974bb8ccbbfb143b1ef"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:questdb:fad106955f31",
+      "dbms": "questdb",
+      "title": "Unexpected result set with a no-op numeric filter",
+      "reported_date": "2022-11-03",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "puzpuzpuz",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/2714"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/2714",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/2714",
+            "source_type": "github_issue",
+            "excerpt": "Reported by @SuriZhang (found by SQLancer)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b4832e7cb82d8b9795839d0c4c37adec8592d22a4433b338bda46df8b3be4fcd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/2714",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b4832e7cb82d8b9795839d0c4c37adec8592d22a4433b338bda46df8b3be4fcd"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:questdb:640f3cd939f1",
+      "dbms": "questdb",
+      "title": "Unexpected error when using `count()`",
+      "reported_date": "2023-05-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3313"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3313",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3313",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected error when using `count()`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:85d9de54f67e39ecfebcafffb5e14d6169d404d51499c8a59ac094ac4f6ca144",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3313",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:85d9de54f67e39ecfebcafffb5e14d6169d404d51499c8a59ac094ac4f6ca144"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:23a382dd6867",
+      "dbms": "questdb",
+      "title": "Unexpected Internal Error: Index -1 out of bounds for length 16",
+      "reported_date": "2023-05-11",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3322"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3322",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3322",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Internal Error: Index -1 out of bounds for length 16",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:25b55a0cef8e603ee99821a55e329e07bec4e0b48b0a0d61d8fde992c7c4623e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3322",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:25b55a0cef8e603ee99821a55e329e07bec4e0b48b0a0d61d8fde992c7c4623e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:61a64658c260",
+      "dbms": "questdb",
+      "title": "Unexpected Internal Error: NullPointerException",
+      "reported_date": "2023-05-11",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3324"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3324",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3324",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Internal Error: NullPointerException",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b3ce7dbe9152aaeb5eef491a99542b4d69ba55e1eb9326a538bfa84911970379",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3324",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b3ce7dbe9152aaeb5eef491a99542b4d69ba55e1eb9326a538bfa84911970379"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:b0f4d024f8d2",
+      "dbms": "questdb",
+      "title": "Unexpected Internal Error: UnsupportedOperationException",
+      "reported_date": "2023-05-11",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3323"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3323",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3323",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Internal Error: UnsupportedOperationException",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:69e49a813f8c523716496c6d4a5214f16b130e6acc36e8c8e576fa4629cb075e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3323",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:69e49a813f8c523716496c6d4a5214f16b130e6acc36e8c8e576fa4629cb075e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:81b3d0570657",
+      "dbms": "questdb",
+      "title": "Unexpected ClassCastException with `null` query",
+      "reported_date": "2023-05-17",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3357"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3357",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3357",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected ClassCastException with `null` query",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:201280f17b69ba6b6fdd68a31cc48fc0e008fbfe0f248190123b97e33bddbe27",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3357",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:201280f17b69ba6b6fdd68a31cc48fc0e008fbfe0f248190123b97e33bddbe27"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:b76149830924",
+      "dbms": "questdb",
+      "title": "Unexpected NullPointerException with `null` query",
+      "reported_date": "2023-05-17",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3356"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3356",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3356",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected NullPointerException with `null` query",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9ebd4b277b9697450ef16c8898d908736e8af55ac01f26a5365f6874877b7cd1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3356",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9ebd4b277b9697450ef16c8898d908736e8af55ac01f26a5365f6874877b7cd1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:48548f64a899",
+      "dbms": "questdb",
+      "title": "`Null` in equality operator",
+      "reported_date": "2023-05-17",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3358"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3358",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3358",
+            "source_type": "github_issue",
+            "excerpt": "`Null` in equality operator",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:886caa48cd5fb7c109d3935d17bd65771017485ca337a556476f498294628c4b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3358",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:886caa48cd5fb7c109d3935d17bd65771017485ca337a556476f498294628c4b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:e6c0e2dd0c51",
+      "dbms": "questdb",
+      "title": "Consuming input failed: server closed the connection unexpectedly",
+      "reported_date": "2023-05-22",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3376"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3376",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3376",
+            "source_type": "github_issue",
+            "excerpt": "Consuming input failed: server closed the connection unexpectedly",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1f8a261cec73fc59bc9c901faa942dd4e0b14e229a63ff4b9a38b6a7b15dabff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3376",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1f8a261cec73fc59bc9c901faa942dd4e0b14e229a63ff4b9a38b6a7b15dabff"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:33cfbd816c4b",
+      "dbms": "questdb",
+      "title": "Unexpected UnsupportedOperationException with `join` query on non-boolean expression",
+      "reported_date": "2023-05-23",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3386"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3386",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3386",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected UnsupportedOperationException with `join` query on non-boolean expression",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e8f0fdbe7ab6f145761d9bcf4cf5458dd609f85d10b7906a70954c8dbace3e9d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3386",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e8f0fdbe7ab6f145761d9bcf4cf5458dd609f85d10b7906a70954c8dbace3e9d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:b86934782ff8",
+      "dbms": "questdb",
+      "title": "NullPointerException Cannot invoke \".getAst()\" because \"queryColumn\" is null with `order by` clause",
+      "reported_date": "2023-05-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3419"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3419",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3419",
+            "source_type": "github_issue",
+            "excerpt": "NullPointerException Cannot invoke \".getAst()\" because \"queryColumn\" is null with `order by` clause",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fa18d34ef24619aadc5935e2463951efe49fbbdbb17560e1064c42a806cb1a0d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3419",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fa18d34ef24619aadc5935e2463951efe49fbbdbb17560e1064c42a806cb1a0d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:99a77cd2dc5a",
+      "dbms": "questdb",
+      "title": "NullPointerException: Cannot read field \"type\" because \"node\" is null with constant expression",
+      "reported_date": "2023-05-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3420"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3420",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3420",
+            "source_type": "github_issue",
+            "excerpt": "NullPointerException: Cannot read field \"type\" because \"node\" is null with constant expression",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:95f1d7a6d6adfcf4acd82da3b68d43448834f5d25689bd0c419307b0596a102a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3420",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:95f1d7a6d6adfcf4acd82da3b68d43448834f5d25689bd0c419307b0596a102a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:5bf5257eadd3",
+      "dbms": "questdb",
+      "title": "Unexpected StringIndexOutOfBoundsException with empty string",
+      "reported_date": "2023-05-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3418"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3418",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3418",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected StringIndexOutOfBoundsException with empty string",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:cdc3e91b8781c5c1acbf1e965ce2e3606bd3599b995de7e22e66ab8dcfeb2afc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3418",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cdc3e91b8781c5c1acbf1e965ce2e3606bd3599b995de7e22e66ab8dcfeb2afc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:d24c6f684c0d",
+      "dbms": "questdb",
+      "title": "ArrayIndexOutOfBoundsException: Index -1 out of bounds for length 16",
+      "reported_date": "2023-06-01",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3433"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3433",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3433",
+            "source_type": "github_issue",
+            "excerpt": "ArrayIndexOutOfBoundsException: Index -1 out of bounds for length 16",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6876615a520c46d285112f7e4c74837d1307e53deb62d046bfd37e450a2350c7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3433",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6876615a520c46d285112f7e4c74837d1307e53deb62d046bfd37e450a2350c7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:14723d3c9d36",
+      "dbms": "questdb",
+      "title": "NullPointerException Cannot invoke \".isUndefined()\" because the return value of \".getQuick(int)\" is null",
+      "reported_date": "2023-06-05",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3445"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3445",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3445",
+            "source_type": "github_issue",
+            "excerpt": "NullPointerException Cannot invoke \".isUndefined()\" because the return value of \".getQuick(int)\" is null",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:77466073e034e5521433a5200a4f81778c5afaee40d7a35651ad198c8d95da67",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3445",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:77466073e034e5521433a5200a4f81778c5afaee40d7a35651ad198c8d95da67"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:c4a91cdb2f0a",
+      "dbms": "questdb",
+      "title": "ClassCastException: NullConstant cannot be cast to class SymbolFunction",
+      "reported_date": "2023-06-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3455"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3455",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3455",
+            "source_type": "github_issue",
+            "excerpt": "ClassCastException: NullConstant cannot be cast to class SymbolFunction",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:856709b20a620eb4bdc6edf47ded66acf7861835c806d2b280ee9ca1ddb81802",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3455",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:856709b20a620eb4bdc6edf47ded66acf7861835c806d2b280ee9ca1ddb81802"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:b9068ba57aa4",
+      "dbms": "questdb",
+      "title": "ImplicitCastException and stuck when comparing `STRING` with `NULL`",
+      "reported_date": "2023-06-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3454"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3454",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3454",
+            "source_type": "github_issue",
+            "excerpt": "ImplicitCastException and stuck when comparing `STRING` with `NULL`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c8f10e354dccbe9b54161f47b93fea72bfa98cb33975073f2faf439d2c9200b8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3454",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c8f10e354dccbe9b54161f47b93fea72bfa98cb33975073f2faf439d2c9200b8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:b4bef73470fe",
+      "dbms": "questdb",
+      "title": "ArrayIndexOutOfBoundsException: Index -1 out of bounds for length 0 at io.questdb.std.LongList.extendAndSet",
+      "reported_date": "2023-06-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3469"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3469",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3469",
+            "source_type": "github_issue",
+            "excerpt": "ArrayIndexOutOfBoundsException: Index -1 out of bounds for length 0 at io.questdb.std.LongList.extendAndSet",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fb5d7f9855c219ad278a46750e73bcfa33f6e4a721c3ddcc92b7b3f3f780f3a9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3469",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fb5d7f9855c219ad278a46750e73bcfa33f6e4a721c3ddcc92b7b3f3f780f3a9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:4f9ffafa2101",
+      "dbms": "questdb",
+      "title": "CairoException when casting string to timestamp",
+      "reported_date": "2023-06-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3470"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3470",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3470",
+            "source_type": "github_issue",
+            "excerpt": "CairoException when casting string to timestamp",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:3f0e518654364d09ee93c25438a2813b3f7ce468b2376da8fa7f7bbc7ecd8753",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3470",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3f0e518654364d09ee93c25438a2813b3f7ce468b2376da8fa7f7bbc7ecd8753"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:100dfc80d57a",
+      "dbms": "questdb",
+      "title": "NullPointerException: Cannot read field \"type\" because \"column\" is null",
+      "reported_date": "2023-06-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3468"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3468",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3468",
+            "source_type": "github_issue",
+            "excerpt": "NullPointerException: Cannot read field \"type\" because \"column\" is null",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:4afe6a8f102398ebca3aad088047c2d415a53a53b70104170b53d733614a8cc2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3468",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4afe6a8f102398ebca3aad088047c2d415a53a53b70104170b53d733614a8cc2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:ea0d901bd0f4",
+      "dbms": "questdb",
+      "title": "StringIndexOutOfBoundsException: String index out of range: 0",
+      "reported_date": "2023-06-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3467"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3467",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3467",
+            "source_type": "github_issue",
+            "excerpt": "StringIndexOutOfBoundsException: String index out of range: 0",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:cc5671f1c1a5b2b794ff7b03733099e9bbb04641449ea9b1b2211ae2c02ec2d3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3467",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cc5671f1c1a5b2b794ff7b03733099e9bbb04641449ea9b1b2211ae2c02ec2d3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:125189547929",
+      "dbms": "questdb",
+      "title": "java.lang.NullPointerException: Cannot read field \"type\" because \"op\" is null in rewriteAggregate(SqlOptimiser.java:3300)",
+      "reported_date": "2023-06-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3526"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3526",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3526",
+            "source_type": "github_issue",
+            "excerpt": "java.lang.NullPointerException: Cannot read field \"type\" because \"op\" is null in rewriteAggregate(SqlOptimiser.java:3300)",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9195ca803591936bd3c560524ccac6e88c4588b157601e06ad7f7f5f45bfd40f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3526",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9195ca803591936bd3c560524ccac6e88c4588b157601e06ad7f7f5f45bfd40f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:031382cf74fe",
+      "dbms": "questdb",
+      "title": "Suspicious incorrect result when left join with an empty table",
+      "reported_date": "2023-07-18",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3575"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3575",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3575",
+            "source_type": "github_issue",
+            "excerpt": "Suspicious incorrect result when left join with an empty table",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:0ba27d13e6621b6291fd192a4aec67abb0353f1ff51955ddb2097c20c65163da",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3575",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0ba27d13e6621b6291fd192a4aec67abb0353f1ff51955ddb2097c20c65163da"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:01ffb8dc20d8",
+      "dbms": "questdb",
+      "title": "Unrobust comparison between string and char",
+      "reported_date": "2023-07-18",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3576"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3576",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3576",
+            "source_type": "github_issue",
+            "excerpt": "Unrobust comparison between string and char",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7ae0301760fd6059db9cf781a0b4e282877be86f6ad3ccd18e07948b8dd0b685",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3576",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7ae0301760fd6059db9cf781a0b4e282877be86f6ad3ccd18e07948b8dd0b685"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:50857d39530e",
+      "dbms": "questdb",
+      "title": "Inaccurate results with `EXCEPT`/`INTERSECT`",
+      "reported_date": "2023-07-19",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3580"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3580",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3580",
+            "source_type": "github_issue",
+            "excerpt": "Inaccurate results with `EXCEPT`/`INTERSECT`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e3de4c22b1cf1851b91fc4e5deebd0c3c94aacac7755e10398ee7651fc8b2155",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3580",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e3de4c22b1cf1851b91fc4e5deebd0c3c94aacac7755e10398ee7651fc8b2155"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:c13cb7d5cb2c",
+      "dbms": "questdb",
+      "title": "Incorrect result returned by `INTERSECT`",
+      "reported_date": "2023-07-19",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3581"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3581",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3581",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result returned by `INTERSECT`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:45916f6d83c3ba98f0292cd74df6ff75ed4828739fd56d5006049518764366d1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3581",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:45916f6d83c3ba98f0292cd74df6ff75ed4828739fd56d5006049518764366d1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:9984a15b1bec",
+      "dbms": "questdb",
+      "title": "ClassCastException: GenericRecordMetadata cannot be cast to JoinRecordMetadata in generateJoins()",
+      "reported_date": "2023-07-21",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3590"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3590",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3590",
+            "source_type": "github_issue",
+            "excerpt": "ClassCastException: GenericRecordMetadata cannot be cast to JoinRecordMetadata in generateJoins()",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:8cda17118f389243648b84694598244b1e8ddd1b409ac494fdd9e22e8deade00",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3590",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8cda17118f389243648b84694598244b1e8ddd1b409ac494fdd9e22e8deade00"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:2af45b9cef4e",
+      "dbms": "questdb",
+      "title": "Unexpected Error Invalid Column with multiple JOIN",
+      "reported_date": "2023-07-25",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3595"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3595",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3595",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Error Invalid Column with multiple JOIN",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:95653420bf15fab8c0a3e764bfb9018a2b5e20880a51b9dd61467c9cb4a8f0e2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3595",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:95653420bf15fab8c0a3e764bfb9018a2b5e20880a51b9dd61467c9cb4a8f0e2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:1557e6fc0daa",
+      "dbms": "questdb",
+      "title": "Invalid Column error with multiple JOIN",
+      "reported_date": "2023-08-04",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3619"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3619",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3619",
+            "source_type": "github_issue",
+            "excerpt": "Invalid Column error with multiple JOIN",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:55966b1753adf1c389d21b05a8d330af38d80e4b89eb95dd42172fb30a2d63d5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3619",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:55966b1753adf1c389d21b05a8d330af38d80e4b89eb95dd42172fb30a2d63d5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:b7ae1755473e",
+      "dbms": "questdb",
+      "title": "StringIndexOutOfBoundsException: String index out of range: 0 with ORDER BY empty string",
+      "reported_date": "2023-08-07",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3623"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3623",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3623",
+            "source_type": "github_issue",
+            "excerpt": "StringIndexOutOfBoundsException: String index out of range: 0 with ORDER BY empty string",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9e131dd5b0f7a2c167b3f7d0852505702b56de90bac6dcd423b7628d1eabe9f6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3623",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9e131dd5b0f7a2c167b3f7d0852505702b56de90bac6dcd423b7628d1eabe9f6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:cf77bfedfa5b",
+      "dbms": "questdb",
+      "title": "UnsupportedOperationException in IntFunction.getStr(IntFunction.java:145)",
+      "reported_date": "2023-08-07",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3622"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3622",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3622",
+            "source_type": "github_issue",
+            "excerpt": "UnsupportedOperationException in IntFunction.getStr(IntFunction.java:145)",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9cff301aab275858558534c86b7b861409ae0e16d27713ff275336e46c59eb26",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3622",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9cff301aab275858558534c86b7b861409ae0e16d27713ff275336e46c59eb26"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:131b64233ae9",
+      "dbms": "questdb",
+      "title": "Incorrect results returned from `UNION`",
+      "reported_date": "2023-08-19",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3669"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3669",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3669",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect results returned from `UNION`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6d6632311eef46dc1fc29b4b6fb9ff86c5ea41f1a87575cfd549e753e529d4b6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3669",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6d6632311eef46dc1fc29b4b6fb9ff86c5ea41f1a87575cfd549e753e529d4b6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:4b962fccdee9",
+      "dbms": "questdb",
+      "title": "Invalid column error with LEFT OUTER JOIN",
+      "reported_date": "2023-08-19",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3670"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3670",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3670",
+            "source_type": "github_issue",
+            "excerpt": "Invalid column error with LEFT OUTER JOIN",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fa07806530ce32f4e90e41373871479111d4a706b0ce4ac5f5bffd3073a8af65",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3670",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fa07806530ce32f4e90e41373871479111d4a706b0ce4ac5f5bffd3073a8af65"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:4d67e80cc2be",
+      "dbms": "questdb",
+      "title": "Invalid column error not completely fixed",
+      "reported_date": "2023-09-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3733"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3733",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3733",
+            "source_type": "github_issue",
+            "excerpt": "Invalid column error not completely fixed",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:80c3d6724b32a34735c223ea119ba1a5a2220ea6dbe83dcad17616807ff2e800",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3733",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:80c3d6724b32a34735c223ea119ba1a5a2220ea6dbe83dcad17616807ff2e800"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:a15655d9034e",
+      "dbms": "questdb",
+      "title": "Incorrect results with symbol comparison",
+      "reported_date": "2023-10-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3828"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3828",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3828",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect results with symbol comparison",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c490c39edbb79821b463f7fb71a980c968978d95f5acd4a651bd045fd6098153",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3828",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c490c39edbb79821b463f7fb71a980c968978d95f5acd4a651bd045fd6098153"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:87482aa53714",
+      "dbms": "questdb",
+      "title": "ArrayIndexOutOfBoundsException: Index -1 out of bounds for length 16",
+      "reported_date": "2023-10-10",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3833"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3833",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3833",
+            "source_type": "github_issue",
+            "excerpt": "ArrayIndexOutOfBoundsException: Index -1 out of bounds for length 16",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a66f927e6f740ccf407ec65cc9f2492b6d8c84ea6e87d463429353fe7e4f6e8c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3833",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a66f927e6f740ccf407ec65cc9f2492b6d8c84ea6e87d463429353fe7e4f6e8c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:658d6a05c3c6",
+      "dbms": "questdb",
+      "title": "Inaccurate result with sum float data",
+      "reported_date": "2023-11-09",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3932"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3932",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3932",
+            "source_type": "github_issue",
+            "excerpt": "Inaccurate result with sum float data",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ddb5e270d0f53e8160955cdde6c2cde92886cc786d9780b0e2556ac83f549667",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3932",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ddb5e270d0f53e8160955cdde6c2cde92886cc786d9780b0e2556ac83f549667"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:45f244b3f3cd",
+      "dbms": "questdb",
+      "title": "Incorrect result with window function",
+      "reported_date": "2023-11-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3936"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3936",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3936",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result with window function",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b40705785b394d93c23e8c55eab8db8f154060f1c797d9f875db44a1db86d376",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3936",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b40705785b394d93c23e8c55eab8db8f154060f1c797d9f875db44a1db86d376"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:a94a61c3e250",
+      "dbms": "questdb",
+      "title": "NullPointerException: Cannot invoke \"io.questdb.cairo.sql.SymbolTable.valueOf(int)\" because \"this.symbolTable\" is null",
+      "reported_date": "2023-11-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3935"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3935",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3935",
+            "source_type": "github_issue",
+            "excerpt": "NullPointerException: Cannot invoke \"io.questdb.cairo.sql.SymbolTable.valueOf(int)\" because \"this.symbolTable\" is null",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:374cac11c5154c8eb148fccfb77b8e8a1f98a87e105066b2d5742c6815e00a40",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3935",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:374cac11c5154c8eb148fccfb77b8e8a1f98a87e105066b2d5742c6815e00a40"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:8378e4746768",
+      "dbms": "questdb",
+      "title": "Unexpected DatabaseError: queries have different number of columns in window function with union",
+      "reported_date": "2023-11-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3934"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3934",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3934",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected DatabaseError: queries have different number of columns in window function with union",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:63d96f0995285aa95186c6466bbf131fac17ebcc6007326b6ca82730d18aa573",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3934",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:63d96f0995285aa95186c6466bbf131fac17ebcc6007326b6ca82730d18aa573"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:bf5f023bdb63",
+      "dbms": "questdb",
+      "title": "Unexpected invalid column not completely fixed",
+      "reported_date": "2023-11-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3933"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3933",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3933",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected invalid column not completely fixed",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:4721d9447371b4112fe4c3d26fb5dbb86839aa1c31eb7e2310959739f5e5ac2d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3933",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4721d9447371b4112fe4c3d26fb5dbb86839aa1c31eb7e2310959739f5e5ac2d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:97a109ef6e80",
+      "dbms": "questdb",
+      "title": "Unexpected results when having brackets with window query",
+      "reported_date": "2023-11-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3938"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3938",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3938",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected results when having brackets with window query",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:6d74eeb030be59e26188d81940fbc2959f670e2c753b285dcc44ff0f3a60851e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3938",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6d74eeb030be59e26188d81940fbc2959f670e2c753b285dcc44ff0f3a60851e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:592c7677ad12",
+      "dbms": "questdb",
+      "title": "NullPointerException: Cannot read field \"token\" because \"first\" is null",
+      "reported_date": "2023-11-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/3949"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/3949",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/3949",
+            "source_type": "github_issue",
+            "excerpt": "NullPointerException: Cannot read field \"token\" because \"first\" is null",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7523f9f40344d277581112afd156ac01dd1f03af6351e76ebac0f2d0a8815b98",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/3949",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7523f9f40344d277581112afd156ac01dd1f03af6351e76ebac0f2d0a8815b98"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:082bc3d48997",
+      "dbms": "questdb",
+      "title": "Incorrect results with nested joins",
+      "reported_date": "2023-11-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/4010"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/4010",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/4010",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect results with nested joins",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:d5a6d4f461ae04e9ea8cc5e5a91af6b2950ca32623c30b034711b4dc47bae9c4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/4010",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d5a6d4f461ae04e9ea8cc5e5a91af6b2950ca32623c30b034711b4dc47bae9c4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:63ed9f0f9a22",
+      "dbms": "questdb",
+      "title": "Unexpected results when comparing NULL value in WHERE clause",
+      "reported_date": "2023-12-11",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/4051"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/4051",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/4051",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\r\nINSERT INTO t0 (c1) VALUES (1);\r\n\r\nSELECT * FROM t0; -- 1",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e2c82f4cf8ae690449ffc74fd3df4216fd9aff9249b76bdb503f4237ba75e448",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/4051",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e2c82f4cf8ae690449ffc74fd3df4216fd9aff9249b76bdb503f4237ba75e448"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:1dcab02f352b",
+      "dbms": "questdb",
+      "title": "Incorrect result when having NULL with IN operation",
+      "reported_date": "2024-05-13",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/4496"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/4496",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/4496",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result when having NULL with IN operation",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:4286a1dfcfeefc4674f3ddb2ab63c20907732b368ee76108f5ddc99569512238",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/4496",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4286a1dfcfeefc4674f3ddb2ab63c20907732b368ee76108f5ddc99569512238"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:067551b9838b",
+      "dbms": "questdb",
+      "title": "Incorrect result when having null in case clause",
+      "reported_date": "2024-05-15",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/4505"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/4505",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/4505",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result when having null in case clause",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a12fa3a50b132a075a3cae9d6795c84cb7212c99d9986838982142267be72a32",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/4505",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a12fa3a50b132a075a3cae9d6795c84cb7212c99d9986838982142267be72a32"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:05d3c14662cd",
+      "dbms": "questdb",
+      "title": "Incorrect result when having NULL within the CASE...WHEN clause",
+      "reported_date": "2024-05-21",
+      "reported_year": 2024,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/4528"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/4528",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/4528",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result when having NULL within the CASE...WHEN clause",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fed43c7954f04717a7dc7e230c531b0b16efd01901d60e4e8befaf148e2c7238",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/4528",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fed43c7954f04717a7dc7e230c531b0b16efd01901d60e4e8befaf148e2c7238"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:5e0de6cbf0ef",
+      "dbms": "questdb",
+      "title": "Incorrect result with window function and table joins",
+      "reported_date": "2024-05-21",
+      "reported_year": 2024,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/4531"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/4531",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/4531",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result with window function and table joins",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:755d6263c9d6dccc9bad8330b0496f172c66bcb259eca656469e50753c6886c7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/4531",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:755d6263c9d6dccc9bad8330b0496f172c66bcb259eca656469e50753c6886c7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:920149204fcb",
+      "dbms": "questdb",
+      "title": "NULL record not retrieved if NULL is in WHERE clause",
+      "reported_date": "2024-09-18",
+      "reported_year": 2024,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Ming Wei Tan",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/4965"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/4965",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/4965",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE IF NOT EXISTS t0(c0 BOOLEAN);\r\nINSERT INTO t0(c0) VALUES (NULL);\r\n\r\nSELECT t0.c0 FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Ming Wei Tan of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3cf818db65744343434e9ea5d480739a73b82cd4e4215dad17a675fb1cd5fd10",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Ming Wei Tan.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3cf818db65744343434e9ea5d480739a73b82cd4e4215dad17a675fb1cd5fd10"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:77045e27bb45",
+      "dbms": "questdb",
+      "title": "Internal error thrown in select query",
+      "reported_date": "2024-09-25",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Ming Wei Tan",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/4981"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/4981",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/4981",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT, c1 SYMBOL);\r\nINSERT INTO t1(c0) VALUES (1);\r\nSELECT t1.c0, t1.c1 FROM t1 WHERE (t1.c1 >= t1.c0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Ming Wei Tan of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:c9d92574e80f260f2cec37b42fc285b8e86c311ee184d69b6993b3f89857f7cc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Ming Wei Tan.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:c9d92574e80f260f2cec37b42fc285b8e86c311ee184d69b6993b3f89857f7cc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:3aa25b8a39c6",
+      "dbms": "questdb",
+      "title": "AssertionError at FilteredRecordCursorFactory.java",
+      "reported_date": "2026-02-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/6762"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/6762",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/6762",
+            "source_type": "github_issue",
+            "excerpt": "AssertionError at FilteredRecordCursorFactory.java",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9419eec534abc815b7a4005f3ea54d255db442e8ee18f4631372cd28eedd70b1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/6762",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9419eec534abc815b7a4005f3ea54d255db442e8ee18f4631372cd28eedd70b1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:questdb:cc8642572ac9",
+      "dbms": "questdb",
+      "title": "AssertionError: index 1 out of bounds for list size 1",
+      "reported_date": "2026-02-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/questdb/questdb/issues/6761"
+      },
+      "primary_url": "https://github.com/questdb/questdb/issues/6761",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/questdb/questdb/issues/6761",
+            "source_type": "github_issue",
+            "excerpt": "AssertionError: index 1 out of bounds for list size 1",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:87712b9b5a2800df2eedc306069e45c880cd1e56ab86ece929419d4c4ce6a9bc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/questdb/questdb/issues/6761",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:87712b9b5a2800df2eedc306069e45c880cd1e56ab86ece929419d4c4ce6a9bc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:risingwave:fe5be1bc0416",
+      "dbms": "risingwave",
+      "title": "Support `pg_opclass`",
+      "reported_date": "2022-06-23",
+      "reported_year": 2022,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "lmatz",
+      "links": {
+        "report": "https://github.com/risingwavelabs/risingwave/issues/3431"
+      },
+      "primary_url": "https://github.com/risingwavelabs/risingwave/issues/3431",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/risingwavelabs/risingwave/issues/3431",
+            "source_type": "github_issue",
+            "excerpt": "Try to make Sqlancer make. See #3364",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:55b33967618331bafd50c09369feae9cfaf3e0322c2d73a9b2995b8461f2ae61",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/risingwavelabs/risingwave/issues/3431",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:55b33967618331bafd50c09369feae9cfaf3e0322c2d73a9b2995b8461f2ae61"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:risingwave:f1b0fd02e258",
+      "dbms": "risingwave",
+      "title": "Incorrect result with `ORDER BY` in `OVER(PARTITION BY ...)`",
+      "reported_date": "2023-10-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/risingwavelabs/risingwave/issues/12822"
+      },
+      "primary_url": "https://github.com/risingwavelabs/risingwave/issues/12822",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/risingwavelabs/risingwave/issues/12822",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result with `ORDER BY` in `OVER(PARTITION BY ...)`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:36a941bd6615452d6ef0532993cd527eb080f2b72780c4b79af8820de9377a8c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/risingwavelabs/risingwave/issues/12822",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:36a941bd6615452d6ef0532993cd527eb080f2b72780c4b79af8820de9377a8c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:risingwave:603b76d37a66",
+      "dbms": "risingwave",
+      "title": "Panicked when processing: assertion failed: `(left == right)`",
+      "reported_date": "2023-10-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/risingwavelabs/risingwave/issues/12818"
+      },
+      "primary_url": "https://github.com/risingwavelabs/risingwave/issues/12818",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/risingwavelabs/risingwave/issues/12818",
+            "source_type": "github_issue",
+            "excerpt": "Panicked when processing: assertion failed: `(left == right)`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fee686cc15ba6e94cc5329cf8f075cb684ef1a1ee2916b923fbcdd9a8f0e9fc1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/risingwavelabs/risingwave/issues/12818",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fee686cc15ba6e94cc5329cf8f075cb684ef1a1ee2916b923fbcdd9a8f0e9fc1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:risingwave:f24282f3fb0f",
+      "dbms": "risingwave",
+      "title": "Panicked when processing: insert at index 4 exceeds fixbitset size 2",
+      "reported_date": "2023-10-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/risingwavelabs/risingwave/issues/12819"
+      },
+      "primary_url": "https://github.com/risingwavelabs/risingwave/issues/12819",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/risingwavelabs/risingwave/issues/12819",
+            "source_type": "github_issue",
+            "excerpt": "Panicked when processing: insert at index 4 exceeds fixbitset size 2",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b4c8da559e45762eaafbb0354687e8fe34b77738ec4095a0a4be6c9ecb2ac946",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/risingwavelabs/risingwave/issues/12819",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b4c8da559e45762eaafbb0354687e8fe34b77738ec4095a0a4be6c9ecb2ac946"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:risingwave:b8fd9e72e318",
+      "dbms": "risingwave",
+      "title": "InternalError_: Panicked when processing: internal error: entered unreachable code: Subquery Subquery",
+      "reported_date": "2023-10-14",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/risingwavelabs/risingwave/issues/12847"
+      },
+      "primary_url": "https://github.com/risingwavelabs/risingwave/issues/12847",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/risingwavelabs/risingwave/issues/12847",
+            "source_type": "github_issue",
+            "excerpt": "InternalError_: Panicked when processing: internal error: entered unreachable code: Subquery Subquery",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:0762ee8d850aeaa1346a0cf3be2dfafff9fb337ed607a10742ba50eb5b740721",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/risingwavelabs/risingwave/issues/12847",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0762ee8d850aeaa1346a0cf3be2dfafff9fb337ed607a10742ba50eb5b740721"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:risingwave:1b90c62c8b10",
+      "dbms": "risingwave",
+      "title": "Unexpected Results when Inserting NULL to PK column",
+      "reported_date": "2023-11-17",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/risingwavelabs/risingwave/issues/13497"
+      },
+      "primary_url": "https://github.com/risingwavelabs/risingwave/issues/13497",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/risingwavelabs/risingwave/issues/13497",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT, PRIMARY KEY(c1, c0));\r\nINSERT INTO t0(c0) VALUES (1);\r\n\r\nSELECT * FROM t0 WHERE (c1 > 1) IS NULL; -- 1 NULL",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e9b76677703c3f7c4701654577c1a8fb8228b3371407b217a860b1193fb1fb13",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e9b76677703c3f7c4701654577c1a8fb8228b3371407b217a860b1193fb1fb13"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:risingwave:2975e8584e1e",
+      "dbms": "risingwave",
+      "title": "Unexpected Results when Using NATURAL LEFT JOIN",
+      "reported_date": "2023-11-21",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/risingwavelabs/risingwave/issues/13572"
+      },
+      "primary_url": "https://github.com/risingwavelabs/risingwave/issues/13572",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/risingwavelabs/risingwave/issues/13572",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT, c2 INT, PRIMARY KEY(c0));\r\nCREATE TABLE t1(c0 INT, c1 INT, PRIMARY KEY(c0));\r\n\r\nINSERT INTO t0(c0, c1, c2) VALUES (1, 1,  1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:1d2421f3a22038d97ac82297701a17d1c78c8508ae39d0b301917197bb02fec0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:1d2421f3a22038d97ac82297701a17d1c78c8508ae39d0b301917197bb02fec0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:risingwave:ae600418baed",
+      "dbms": "risingwave",
+      "title": "Unexpected Results when Comparing with Bigint",
+      "reported_date": "2023-11-22",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/risingwavelabs/risingwave/issues/13601"
+      },
+      "primary_url": "https://github.com/risingwavelabs/risingwave/issues/13601",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/risingwavelabs/risingwave/issues/13601",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT, PRIMARY KEY(c0));\r\nINSERT INTO t0(c1) VALUES (1);\r\n\r\nSELECT c1 FROM t0 WHERE -9223372036854775808<=c0;  -- 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:75cdd46ef0b4102eff2c3f0189d330406b1572f4b7194901de4cf30bb47776b0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:75cdd46ef0b4102eff2c3f0189d330406b1572f4b7194901de4cf30bb47776b0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:risingwave:7093d608bb2c",
+      "dbms": "risingwave",
+      "title": "bug(pgwire): byte index 1024 is not a char boundary",
+      "reported_date": "2023-12-30",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/risingwavelabs/risingwave/issues/14283"
+      },
+      "primary_url": "https://github.com/risingwavelabs/risingwave/issues/14283",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/risingwavelabs/risingwave/issues/14283",
+            "source_type": "github_issue",
+            "excerpt": "bug(pgwire): byte index 1024 is not a char boundary",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:af30d8575203dadc63913d14bed340928fd77d7846a2e56fa65811c8efe9215d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/risingwavelabs/risingwave/issues/14283",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:af30d8575203dadc63913d14bed340928fd77d7846a2e56fa65811c8efe9215d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:seekdb:2d0a2268df24",
+      "dbms": "seekdb",
+      "title": "Fix memory allocation failure (errcode=-4013) during vector benchmark with async index",
+      "reported_date": "2026-04-16",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "cms-cms",
+      "links": {
+        "report": "https://github.com/oceanbase/seekdb/issues/563"
+      },
+      "primary_url": "https://github.com/oceanbase/seekdb/issues/563",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/seekdb/issues/563",
+            "source_type": "github_issue",
+            "excerpt": "1.  `[sqlancer|mysql] 报错[errcode=-4016] total_alloc_size_ is less than 0后环境持续4013且clog盘爆` (Similarity: 68.15%)\n2.  `【混搜】insert报错4013 No memory or reach tenant memory limit` (Similarity: 67.08%)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a35f2f797737fc9aac32e80dd938b64731d21e06af37bd0c6bf81f12da0d8020",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/seekdb/issues/563",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a35f2f797737fc9aac32e80dd938b64731d21e06af37bd0c6bf81f12da0d8020"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:seekdb:9e46d4af804a",
+      "dbms": "seekdb",
+      "title": "seekdb: Result size mismatch for GROUP BY ... WITH ROLLUP + HAVING partitioned TLP (RIGHT JOIN null-extended columns)",
+      "reported_date": "2026-08-04",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "footka",
+      "links": {
+        "report": "https://github.com/oceanbase/seekdb/issues/1254"
+      },
+      "primary_url": "https://github.com/oceanbase/seekdb/issues/1254",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/seekdb/issues/1254",
+            "source_type": "github_issue",
+            "excerpt": "A differential test (`MySQLQueryPartitioningHavingTester` TLP) in sqlancer reports a mismatch in result set size on seekdb.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:65f2498058d8ae957afc539ba04be30961d8dd73554efbb5468d13a38bd4458c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/seekdb/issues/1254",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:65f2498058d8ae957afc539ba04be30961d8dd73554efbb5468d13a38bd4458c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:seekdb:bbe6b2353119",
+      "dbms": "seekdb",
+      "title": "Plan cache memory usage far exceeds limit and memory continues to grow after SQLancer stress test",
+      "reported_date": "2026-08-11",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "chyujing",
+      "links": {
+        "report": "https://github.com/oceanbase/seekdb/issues/1281"
+      },
+      "primary_url": "https://github.com/oceanbase/seekdb/issues/1281",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/oceanbase/seekdb/issues/1281",
+            "source_type": "github_issue",
+            "excerpt": "Plan cache memory usage far exceeds limit and memory continues to grow after SQLancer stress test",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:06e04b4683f775b61a0d122e8775087c1fddcd45cba6002e788567fcf3701c1d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/oceanbase/seekdb/issues/1281",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:06e04b4683f775b61a0d122e8775087c1fddcd45cba6002e788567fcf3701c1d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:sparq:5c0013598eb7",
+      "dbms": "sparq",
+      "title": "[metamorph] shard=demo: 20 TLP/NoREC oracle failure(s) (first seed=0)",
+      "reported_date": "2026-07-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "jeswr",
+      "links": {
+        "report": "https://github.com/sparq-org/sparq/issues/1938"
+      },
+      "primary_url": "https://github.com/sparq-org/sparq/issues/1938",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sparq-org/sparq/issues/1938",
+            "source_type": "github_issue",
+            "excerpt": "[metamorph] shard=demo: 20 TLP/NoREC oracle failure(s) (first seed=0)",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:a55d64b89d193e8f58fcc074a3b59936d0895c81896bc1c445556f632ef2badf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sparq-org/sparq/issues/1938",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a55d64b89d193e8f58fcc074a3b59936d0895c81896bc1c445556f632ef2badf"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:sqlite:7b06126e0517",
+      "dbms": "sqlite",
+      "title": "COLLATE NOCASE index on REAL column malfunctions",
+      "reported_date": "2019-05-01",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084334.html",
+        "fix": "https://www.sqlite.org/src/info/b043a54c3de54b28"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084334.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"1/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084334.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/b043a54c3de54b28\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:6d46db6996daa92cffe9515ad559e98616d7c48f306410e15d6db80aaafc1035",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6d46db6996daa92cffe9515ad559e98616d7c48f306410e15d6db80aaafc1035"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084334.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:861e742b8e25",
+      "dbms": "sqlite",
+      "title": "Multi-row insert circumvents index check",
+      "reported_date": "2019-05-01",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/info/3be1295b264be2fa",
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084326.html",
+        "fix": "https://www.sqlite.org/src/info/713caa382cf7ddef"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/info/3be1295b264be2fa",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"1/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/info/3be1295b264be2fa\",\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084326.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/713caa382cf7ddef\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e40f6bf111b8279ffd8c730002886097bd309c2e4c0d1f3fbf718f968e514cb4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e40f6bf111b8279ffd8c730002886097bd309c2e4c0d1f3fbf718f968e514cb4"
+      },
+      "primary_url": "https://sqlite.org/src/info/3be1295b264be2fa",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:ac21f87c71fd",
+      "dbms": "sqlite",
+      "title": "Index on REAL column malfunctions when multiplying with a string",
+      "reported_date": "2019-05-02",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084355.html",
+        "fix": "https://www.sqlite.org/src/info/5a8a23ee5f60a31d"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084355.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"2/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084355.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/5a8a23ee5f60a31d\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:6cd72bdbe2c086b3fcd6addcdf541edd51ca05e4a6a478e6e3a9f4e29ddd1226",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6cd72bdbe2c086b3fcd6addcdf541edd51ca05e4a6a478e6e3a9f4e29ddd1226"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084355.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:8989af0af214",
+      "dbms": "sqlite",
+      "title": "TYPEOF index on REAL column malfunctions",
+      "reported_date": "2019-05-02",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084350.html",
+        "fix": "https://www.sqlite.org/src/info/48889530a9de22fe"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084350.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"2/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084350.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/48889530a9de22fe\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:305b6aef396a53a928113f4ed07498b5ed27ec6f02bf0c898d6e0d1fec4abf25",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:305b6aef396a53a928113f4ed07498b5ed27ec6f02bf0c898d6e0d1fec4abf25"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084350.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:c160a0e7bc81",
+      "dbms": "sqlite",
+      "title": "UPSERT documentation issue",
+      "reported_date": "2019-05-02",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084347.html"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084347.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"2/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084347.html\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:aaab854656782103aa7910b0f5e20ada81042fd78ba46f8a87b874a9cfeefea2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:aaab854656782103aa7910b0f5e20ada81042fd78ba46f8a87b874a9cfeefea2"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084347.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d4fea923278d",
+      "dbms": "sqlite",
+      "title": "GLOB and minus in index and real column",
+      "reported_date": "2019-05-03",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084367.html",
+        "fix": "https://www.sqlite.org/src/info/5a8a23ee5f60a31d"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084367.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"3/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084367.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/5a8a23ee5f60a31d\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:310e91b8dd033615e4e91fd6d841d0219229592abe1fe179851744953db5ee0d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:310e91b8dd033615e4e91fd6d841d0219229592abe1fe179851744953db5ee0d"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084367.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:6fb8b12aa80c",
+      "dbms": "sqlite",
+      "title": "Problem with REAL values and string functions used in indexes or on expressions",
+      "reported_date": "2019-05-03",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084366.html",
+        "fix": "https://www.sqlite.org/src/info/5a8a23ee5f60a31d"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084366.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"3/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084366.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/5a8a23ee5f60a31d\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:561b40eb2fd15b409ea5ff7d94860ac41823d56fe79c6ec6f524b4ca55dea26a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:561b40eb2fd15b409ea5ff7d94860ac41823d56fe79c6ec6f524b4ca55dea26a"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084366.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:1f339a183897",
+      "dbms": "sqlite",
+      "title": "Incorrect result on a table scan of a partial index",
+      "reported_date": "2019-05-04",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=5c6955204c",
+        "email": "https://www.mail-archive.com/sqlite-users@mailinglists.sqlite.org/msg115136.html",
+        "fix": "https://www.sqlite.org/src/info/c2e439bccc40825e211bf"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=5c6955204c",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"4/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=5c6955204c\",\n            \"email\": \"https://www.mail-archive.com/sqlite-users@mailinglists.sqlite.org/msg115136.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/c2e439bccc40825e211bf\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:e64be50116b482de1e3edf7cf94b448189a2b42a0e22fb02ff4c1c0441e96a7b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e64be50116b482de1e3edf7cf94b448189a2b42a0e22fb02ff4c1c0441e96a7b"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=5c6955204c",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:36a57dceeb06",
+      "dbms": "sqlite",
+      "title": "REAL PRIMARY KEY and floating point comparison does not work",
+      "reported_date": "2019-05-05",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084409.html"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084409.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"5/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084409.html\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:c4e806e859d4a0b8a55b48b04fbeb8520b899e696fafc76b6ad1bcd263ca6fc0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c4e806e859d4a0b8a55b48b04fbeb8520b899e696fafc76b6ad1bcd263ca6fc0"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084409.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:7ea02445d3dd",
+      "dbms": "sqlite",
+      "title": "ALTER TABLE fails when renaming an INTEGER PRIMARY KEY column in a WITHOUT ROWID table",
+      "reported_date": "2019-05-06",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084426.html",
+        "fix": "https://sqlite.org/src/info/91f701d39852ef1ddb29"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084426.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"6/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084426.html\",\n            \"fix\": \"https://sqlite.org/src/info/91f701d39852ef1ddb29\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:30c378022c97c064b465b6f3b3899c4679c7424cc8678018c7dc9ff9dc7286b3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:30c378022c97c064b465b6f3b3899c4679c7424cc8678018c7dc9ff9dc7286b3"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084426.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:87482e99a7c1",
+      "dbms": "sqlite",
+      "title": "INSERT OR FAIL inserts row although it violates a table constraint",
+      "reported_date": "2019-05-07",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084439.html",
+        "fix1": "https://www.sqlite.org/src/info/659c551dcc374a0d",
+        "fix2": "https://www.sqlite.org/src/info/3f1c8051648a341d"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084439.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"7/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084439.html\",\n            \"fix1\": \"https://www.sqlite.org/src/info/659c551dcc374a0d\",\n            \"fix2\": \"https://www.sqlite.org/src/info/3f1c8051648a341d\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:eb84bac43e9a0a82a757854606acdff768202e7a43f99850be7924fcee56dd4e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:eb84bac43e9a0a82a757854606acdff768202e7a43f99850be7924fcee56dd4e"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084439.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:77f35ee0fb9e",
+      "dbms": "sqlite",
+      "title": "Incorrect result for \"<\" and \"<=\" comparison of rowid and non-numeric text value",
+      "reported_date": "2019-05-07",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=9cf6c9bb51",
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084460.html",
+        "fix": "https://www.sqlite.org/src/info/658b84d7"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=9cf6c9bb51",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"7/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=9cf6c9bb51\",\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084460.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/658b84d7\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:59bd6256a9b47b3ac9d74adfefcb8d7c1228be6f1d15c9c1714f9331f6a08ea7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:59bd6256a9b47b3ac9d74adfefcb8d7c1228be6f1d15c9c1714f9331f6a08ea7"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=9cf6c9bb51",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:525c705ead46",
+      "dbms": "sqlite",
+      "title": "'./' LIKE './' does not match",
+      "reported_date": "2019-05-08",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084478.html",
+        "fix": "https://www.sqlite.org/src/info/740201107ae802c1"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084478.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"8/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084478.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/740201107ae802c1\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:034ceed55717809d6578aa8860eb447823c33d346ad4618e4bfb380e830f3d13",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:034ceed55717809d6578aa8860eb447823c33d346ad4618e4bfb380e830f3d13"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084478.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:96fc6095fc56",
+      "dbms": "sqlite",
+      "title": "Malformed database image when using a REAL PRIMARY KEY",
+      "reported_date": "2019-05-09",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview/6c1d3febc00b22d457c7",
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084483.html",
+        "fix": "https://www.sqlite.org/src/info/9b0915272f4d4052"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview/6c1d3febc00b22d457c7",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"9/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview/6c1d3febc00b22d457c7\",\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084483.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/9b0915272f4d4052\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:de469b95f5f6e3d27cc1b76c21c63be03d88e0eeaa4072fdd63dbbda170a6737",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:de469b95f5f6e3d27cc1b76c21c63be03d88e0eeaa4072fdd63dbbda170a6737"
+      },
+      "primary_url": "https://sqlite.org/src/tktview/6c1d3febc00b22d457c7",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:7d4af008fb9b",
+      "dbms": "sqlite",
+      "title": "Row is not fetched with PRAGMA reverse_unordered_selects=true",
+      "reported_date": "2019-05-09",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084480.html",
+        "fix": "https://www.sqlite.org/src/info/ebe4845cd0ffb96b"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084480.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"9/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084480.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/ebe4845cd0ffb96b\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:db800b18059348c0e8303426cd645d6051f6bb1789f89c7243ff42b09303261c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:db800b18059348c0e8303426cd645d6051f6bb1789f89c7243ff42b09303261c"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084480.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:0c4e576379b7",
+      "dbms": "sqlite",
+      "title": "Incorrect handling of Infinity by the ROUND function",
+      "reported_date": "2019-05-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084497.html",
+        "fix": "https://www.sqlite.org/src/info/db9acef14d492121"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084497.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084497.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/db9acef14d492121\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:ddc0800ace19ff54b666e2a68e455f10bcf2737244619695d6db1fa30eb7f49f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ddc0800ace19ff54b666e2a68e455f10bcf2737244619695d6db1fa30eb7f49f"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084497.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:04e4030664f4",
+      "dbms": "sqlite",
+      "title": "Partial NOT NULL index malfunctions with IS NOT/!=",
+      "reported_date": "2019-05-11",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview/80256748471a01",
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084499.html",
+        "fix": "https://sqlite.org/src/info/0ba6d709"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview/80256748471a01",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview/80256748471a01\",\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084499.html\",\n            \"fix\": \"https://sqlite.org/src/info/0ba6d709\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:e5a9c67a2d7799c79c84f05aeef564923147713393cd927a31afbb4241bddd13",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e5a9c67a2d7799c79c84f05aeef564923147713393cd927a31afbb4241bddd13"
+      },
+      "primary_url": "https://sqlite.org/src/tktview/80256748471a01",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:a0a78410bfbf",
+      "dbms": "sqlite",
+      "title": "Mixing main and temp databases in foreign keys is not supported",
+      "reported_date": "2019-05-13",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084515.html"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084515.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"13/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084515.html\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:41b722dfabaf2cd6304b16d860c09a553e2d58850a502bc6b57da9c82e299e8c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:41b722dfabaf2cd6304b16d860c09a553e2d58850a502bc6b57da9c82e299e8c"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084515.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:4bf0383745e8",
+      "dbms": "sqlite",
+      "title": "REINDEX causes rows not to be fetched in a WITHOUT ROWIDs table and PRIMARY KEY DESC",
+      "reported_date": "2019-05-13",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=bba7b69f98",
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084528.html",
+        "fix": "https://www.sqlite.org/src/info/f7aadfab3bb8eb8e"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=bba7b69f98",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"13/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=bba7b69f98\",\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084528.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/f7aadfab3bb8eb8e\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:ed492b50d06a658af055f2ecfdb5ed74feb373c58e7d46095cb4694816271133",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ed492b50d06a658af055f2ecfdb5ed74feb373c58e7d46095cb4694816271133"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=bba7b69f98",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:eedb98c127aa",
+      "dbms": "sqlite",
+      "title": "PRAGMA reverse_unordered_selects=true results in row not being fetched",
+      "reported_date": "2019-05-14",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084539.html",
+        "fix": "https://www.sqlite.org/src/info/bc7d2c1656396bb4"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084539.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084539.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/bc7d2c1656396bb4\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:06525653fe731b4adea5e13aab6621e545aae3d623fac8799623e9fab6fcec1e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:06525653fe731b4adea5e13aab6621e545aae3d623fac8799623e9fab6fcec1e"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-May/084539.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:38658ef9621d",
+      "dbms": "sqlite",
+      "title": "Malformed image when using no journal mode, zero cache size, and failing when creating an index",
+      "reported_date": "2019-05-16",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=f4ec250930",
+        "fix": "https://www.sqlite.org/src/info/a0f5eb5c79cc33b7"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=f4ec250930",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"16/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=f4ec250930\",\n            \"fix\": \"https://www.sqlite.org/src/info/a0f5eb5c79cc33b7\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c3075712d68e6f9974d894c9ebf3534b1cf53891e36ff2e2af88adbd08e6f9d5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c3075712d68e6f9974d894c9ebf3534b1cf53891e36ff2e2af88adbd08e6f9d5"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=f4ec250930",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:ecaec8dc9bba",
+      "dbms": "sqlite",
+      "title": "REAL rounding seems to depend on FROM clause",
+      "reported_date": "2019-05-16",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=3c27b97e31",
+        "fix": "https://www.sqlite.org/src/info/14c00b1016ba53ab"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=3c27b97e31",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"16/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=3c27b97e31\",\n            \"fix\": \"https://www.sqlite.org/src/info/14c00b1016ba53ab\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:72dd30208df79ebbdccdcc660a7c355e0b62ca2ae429181fbeee944230ecb705",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:72dd30208df79ebbdccdcc660a7c355e0b62ca2ae429181fbeee944230ecb705"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=3c27b97e31",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:43e643366ed7",
+      "dbms": "sqlite",
+      "title": "Query results in a SEGFAULT",
+      "reported_date": "2019-05-18",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/info/787fa716be3a7f650c",
+        "fix": "https://sqlite.org/src/info/778b1224a318d013"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/info/787fa716be3a7f650c",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"18/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/info/787fa716be3a7f650c\",\n            \"fix\": \"https://sqlite.org/src/info/778b1224a318d013\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7cb191eecc02a26ec25371cbf8eb74786ae6b6f01774b8b9cd50df0c1cfd9259",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7cb191eecc02a26ec25371cbf8eb74786ae6b6f01774b8b9cd50df0c1cfd9259"
+      },
+      "primary_url": "https://sqlite.org/src/info/787fa716be3a7f650c",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:f9761d627e85",
+      "dbms": "sqlite",
+      "title": "Index on non-existing column results in a fabricated value being fetched",
+      "reported_date": "2019-05-19",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Cosmetic",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview/9b78184be266fd7084e9e8038ad631a21b37eb9e",
+        "fix1": "https://www.sqlite.org/src/info/1685610e",
+        "fix2": "https://www.sqlite.org/src/info/3e1b55f3ab85710e"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview/9b78184be266fd7084e9e8038ad631a21b37eb9e",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview/9b78184be266fd7084e9e8038ad631a21b37eb9e\",\n            \"fix1\": \"https://www.sqlite.org/src/info/1685610e\",\n            \"fix2\": \"https://www.sqlite.org/src/info/3e1b55f3ab85710e\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:31fdfb219c14db2e6c9673dbc401ba997c29199122bb9aa061b7a139f43e8ba9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:31fdfb219c14db2e6c9673dbc401ba997c29199122bb9aa061b7a139f43e8ba9"
+      },
+      "primary_url": "https://sqlite.org/src/tktview/9b78184be266fd7084e9e8038ad631a21b37eb9e",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:ed259ce831c6",
+      "dbms": "sqlite",
+      "title": "Nested boolean formula with IN operator computes an incorrect result",
+      "reported_date": "2019-05-19",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=d3e7f2ba5b",
+        "fix": "https://www.sqlite.org/src/info/99eba69b3a64741c"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=d3e7f2ba5b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=d3e7f2ba5b\",\n            \"fix\": \"https://www.sqlite.org/src/info/99eba69b3a64741c\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:a9f2fa25c13e3cb8d21e069712acdce11ca56b7b9e2c4ddf0cbe8aa3f111b212",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a9f2fa25c13e3cb8d21e069712acdce11ca56b7b9e2c4ddf0cbe8aa3f111b212"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=d3e7f2ba5b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:7881933929c6",
+      "dbms": "sqlite",
+      "title": "\"Malformed database schema\" when creating a failing index within a transaction",
+      "reported_date": "2019-05-21",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Cosmetic",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview/b5ca442af9fadf5eff5b2bf64839516ab82cfc3d",
+        "fix": "https://www.sqlite.org/src/info/b8071d10cba8f6c1"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview/b5ca442af9fadf5eff5b2bf64839516ab82cfc3d",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview/b5ca442af9fadf5eff5b2bf64839516ab82cfc3d\",\n            \"fix\": \"https://www.sqlite.org/src/info/b8071d10cba8f6c1\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1e8acf9565a148e4de33dc80eef709b28404bb789f2a90d30fb5afa97fd8eaab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1e8acf9565a148e4de33dc80eef709b28404bb789f2a90d30fb5afa97fd8eaab"
+      },
+      "primary_url": "https://sqlite.org/src/tktview/b5ca442af9fadf5eff5b2bf64839516ab82cfc3d",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:67d4f82fed68",
+      "dbms": "sqlite",
+      "title": "CAST('-' AS NUMERIC) computes 0.0",
+      "reported_date": "2019-05-25",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=4c2d7639f0",
+        "fix": "https://www.sqlite.org/src/info/67a68af5578f08d2"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=4c2d7639f0",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=4c2d7639f0\",\n            \"fix\": \"https://www.sqlite.org/src/info/67a68af5578f08d2\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:923be56f3e61b6151bf7800c4b54ff3254780402201c5ec81cf02c65ebf2efce",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:923be56f3e61b6151bf7800c4b54ff3254780402201c5ec81cf02c65ebf2efce"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=4c2d7639f0",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:103d7b0fbb3a",
+      "dbms": "sqlite",
+      "title": "Incorrect result when subtracting a large integer number from a TEXT value",
+      "reported_date": "2019-05-25",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=e8bedb2a18",
+        "fix": "https://www.sqlite.org/src/info/67a68af5578f08d2"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=e8bedb2a18",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=e8bedb2a18\",\n            \"fix\": \"https://www.sqlite.org/src/info/67a68af5578f08d2\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:1b2cfe4f383f8980ddd6c7987450c220d1d042547d1003a2935ddd18867eef72",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1b2cfe4f383f8980ddd6c7987450c220d1d042547d1003a2935ddd18867eef72"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=e8bedb2a18",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:ad27adb8eefd",
+      "dbms": "sqlite",
+      "title": "COLLATE nocase index on a WITHOUT ROWID table malfunctions",
+      "reported_date": "2019-05-28",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview/3182d3879020ef3b2e6db56be2470a0266d3c773",
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084293.html",
+        "fix": "https://www.sqlite.org/src/info/1b1dd4d48cd79a58"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview/3182d3879020ef3b2e6db56be2470a0266d3c773",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"28/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview/3182d3879020ef3b2e6db56be2470a0266d3c773\",\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084293.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/1b1dd4d48cd79a58\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:2585578b6bc2975e0919989103a69892149645fcb6f4ad07b3262646a9aeee72",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2585578b6bc2975e0919989103a69892149645fcb6f4ad07b3262646a9aeee72"
+      },
+      "primary_url": "https://sqlite.org/src/tktview/3182d3879020ef3b2e6db56be2470a0266d3c773",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:67bfef10412f",
+      "dbms": "sqlite",
+      "title": "PRAGMA case_sensitive_like can corrupt some databases",
+      "reported_date": "2019-05-28",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/info/a340eef47b0cad5",
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084322.html",
+        "fix": "https://www.sqlite.org/src/info/eabe7f2d4ff0e0dd"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/info/a340eef47b0cad5",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"28/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/info/a340eef47b0cad5\",\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084322.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/eabe7f2d4ff0e0dd\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ff784fe1a14948c65e69279c725348afc8e69f9f57e425e3b8829a3a37f17863",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ff784fe1a14948c65e69279c725348afc8e69f9f57e425e3b8829a3a37f17863"
+      },
+      "primary_url": "https://sqlite.org/src/info/a340eef47b0cad5",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:6b3f8f21f866",
+      "dbms": "sqlite",
+      "title": "PRAGMA case_sensitive_like conflicts with LIKE operator when creating an index",
+      "reported_date": "2019-05-28",
+      "reported_year": 2019,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=a340eef47b",
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084299.html"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=a340eef47b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"28/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=a340eef47b\",\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084299.html\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:117197fc87bc5f27e41b0458cc8cc3972431b99b213a7122dcdf905b9539ad76",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:117197fc87bc5f27e41b0458cc8cc3972431b99b213a7122dcdf905b9539ad76"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=a340eef47b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:450197b0b75e",
+      "dbms": "sqlite",
+      "title": "String interpreted as a column name when creating an index",
+      "reported_date": "2019-05-28",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084283.html"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084283.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"28/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084283.html\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:9e9a91967f2fa1d60fe042d6f1cec44adf85e20a7431b62b37b0d0e20fff921c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9e9a91967f2fa1d60fe042d6f1cec44adf85e20a7431b62b37b0d0e20fff921c"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084283.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:8c2aa141872b",
+      "dbms": "sqlite",
+      "title": "Unique index that uses GLOB does not detect duplicate due to REAL conversion",
+      "reported_date": "2019-05-30",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "email": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084324.html",
+        "fix": "https://www.sqlite.org/src/info/5997d075665faca6"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084324.html",
+            "source_type": "mailing_list",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/5/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"email\": \"http://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084324.html\",\n            \"fix\": \"https://www.sqlite.org/src/info/5997d075665faca6\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f78d9428c9bc10f5a2a0e7a5372069137196d8c0087d0e7e38b87773194f9a87",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f78d9428c9bc10f5a2a0e7a5372069137196d8c0087d0e7e38b87773194f9a87"
+      },
+      "primary_url": "https://mailinglists.sqlite.org/cgi-bin/mailman/private/sqlite-users/2019-April/084324.html",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:2950c7291e42",
+      "dbms": "sqlite",
+      "title": "CAST to NUMERIC no longer converts to INTEGER",
+      "reported_date": "2019-06-08",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview/dd6bffbfb6e61db9ecc9ea833d586427961ccc9d",
+        "fix": "https://www.sqlite.org/src/info/c0c90961b4fa1c11"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview/dd6bffbfb6e61db9ecc9ea833d586427961ccc9d",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"08/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview/dd6bffbfb6e61db9ecc9ea833d586427961ccc9d\",\n            \"fix\": \"https://www.sqlite.org/src/info/c0c90961b4fa1c11\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:af542060007ddafa48fd2e2a4c94dc24a259f03dd7d40952b5c00f5ec8906c5d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:af542060007ddafa48fd2e2a4c94dc24a259f03dd7d40952b5c00f5ec8906c5d"
+      },
+      "primary_url": "https://sqlite.org/src/tktview/dd6bffbfb6e61db9ecc9ea833d586427961ccc9d",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:bdd228ba5bb1",
+      "dbms": "sqlite",
+      "title": "Another case of Illegal argument to LIKELIHOOD() does not result in error when combined with \"IN ()\"",
+      "reported_date": "2019-06-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview/533010b8cacebe82533a8cd4e230fbb819565115",
+        "fix": "https://www.sqlite.org/src/info/04bd5cb73287f926"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview/533010b8cacebe82533a8cd4e230fbb819565115",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview/533010b8cacebe82533a8cd4e230fbb819565115\",\n            \"fix\": \"https://www.sqlite.org/src/info/04bd5cb73287f926\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a95d2a69a3bd731a8f8942ad1c3287d0f2cd3e84c45be9c9ffcf25c9066cb10c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a95d2a69a3bd731a8f8942ad1c3287d0f2cd3e84c45be9c9ffcf25c9066cb10c"
+      },
+      "primary_url": "https://sqlite.org/src/tktview/533010b8cacebe82533a8cd4e230fbb819565115",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5ceedf5f3e43",
+      "dbms": "sqlite",
+      "title": "CAST('.' AS NUMERIC) computes 0.0 rather than 0",
+      "reported_date": "2019-06-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=412bba9b22",
+        "fix": "https://www.sqlite.org/src/info/57050162294efec9"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=412bba9b22",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=412bba9b22\",\n            \"fix\": \"https://www.sqlite.org/src/info/57050162294efec9\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:cf6711ef5f114dae4749a4e476bdac976a8241f6e6a38d411adc90ad3293f1a3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:cf6711ef5f114dae4749a4e476bdac976a8241f6e6a38d411adc90ad3293f1a3"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=412bba9b22",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:1023fb37ff21",
+      "dbms": "sqlite",
+      "title": "COLLATE expression has an affinity",
+      "reported_date": "2019-06-10",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview/d60b3cd7cb0bdff8ff39f76c9faf16ba2efa442f",
+        "fix": "https://www.sqlite.org/docsrc/info/722f0828b3074e89"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview/d60b3cd7cb0bdff8ff39f76c9faf16ba2efa442f",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview/d60b3cd7cb0bdff8ff39f76c9faf16ba2efa442f\",\n            \"fix\": \"https://www.sqlite.org/docsrc/info/722f0828b3074e89\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:0373f64771ad061e4849b3791847b4dba62d67024f876ec27384af2b2847f858",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0373f64771ad061e4849b3791847b4dba62d67024f876ec27384af2b2847f858"
+      },
+      "primary_url": "https://sqlite.org/src/tktview/d60b3cd7cb0bdff8ff39f76c9faf16ba2efa442f",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:4c68ba98b8b8",
+      "dbms": "sqlite",
+      "title": "Illegal argument to LIKELIHOOD() does not result in error when combined with \"IN ()\"",
+      "reported_date": "2019-06-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=533010b8ca",
+        "fix": "https://www.sqlite.org/src/info/71643deb"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=533010b8ca",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=533010b8ca\",\n            \"fix\": \"https://www.sqlite.org/src/info/71643deb\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:fc23172600d4dcd80c2988f7d0ae33478cb325f7101a5371e55b394af7c10d7e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:fc23172600d4dcd80c2988f7d0ae33478cb325f7101a5371e55b394af7c10d7e"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=533010b8ca",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:8188339e4646",
+      "dbms": "sqlite",
+      "title": "LIKE malfunctions for INT PRIMARY KEY COLLATE NOCASE column",
+      "reported_date": "2019-06-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=b1d8c79314",
+        "fix": "https://sqlite.org/src/info/94b58ab059cba977"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=b1d8c79314",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=b1d8c79314\",\n            \"fix\": \"https://sqlite.org/src/info/94b58ab059cba977\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:a65189eb9708a6b9aaf53f88ea83a3b38b78f5062ad8dd84e65ad91cb5e4d045",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a65189eb9708a6b9aaf53f88ea83a3b38b78f5062ad8dd84e65ad91cb5e4d045"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=b1d8c79314",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:a3a233cb5acc",
+      "dbms": "sqlite",
+      "title": "TEXT value interpreted as column name in an index with empty list in an IN expression",
+      "reported_date": "2019-06-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview/fd76310a5e843e074a30ed98b859dd0be11d0276",
+        "fix": "https://sqlite.org/src/info/567b13093956185b"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview/fd76310a5e843e074a30ed98b859dd0be11d0276",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview/fd76310a5e843e074a30ed98b859dd0be11d0276\",\n            \"fix\": \"https://sqlite.org/src/info/567b13093956185b\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e379c3b0806e2bf85f13df2cdf0d0f424894bbd21b57014c1f2b577b187c030f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e379c3b0806e2bf85f13df2cdf0d0f424894bbd21b57014c1f2b577b187c030f"
+      },
+      "primary_url": "https://sqlite.org/src/tktview/fd76310a5e843e074a30ed98b859dd0be11d0276",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:157f6c8e7e51",
+      "dbms": "sqlite",
+      "title": "-'1.0' computes -1.0 rather than -1",
+      "reported_date": "2019-06-11",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=1819598c09",
+        "fix": "https://www.sqlite.org/docsrc/info/e0f700bb6c8142c5"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=1819598c09",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=1819598c09\",\n            \"fix\": \"https://www.sqlite.org/docsrc/info/e0f700bb6c8142c5\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:ccb635d8e9cda8e7add4a592f0d05ce0307738f11963d4a76338916e9fa979f0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ccb635d8e9cda8e7add4a592f0d05ce0307738f11963d4a76338916e9fa979f0"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=1819598c09",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:ad16e94fb3bb",
+      "dbms": "sqlite",
+      "title": "COLLATE expression in the right side of an IN operator results in an affinity conversion",
+      "reported_date": "2019-06-11",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=57353f8243",
+        "fix": "https://www.sqlite.org/src/info/0f748fe58bbbb7ce"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=57353f8243",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=57353f8243\",\n            \"fix\": \"https://www.sqlite.org/src/info/0f748fe58bbbb7ce\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:58b4d19b845ec2616ba4eac0901c06e1fd18625135be0e3c57c72ccfd317a9f0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:58b4d19b845ec2616ba4eac0901c06e1fd18625135be0e3c57c72ccfd317a9f0"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=57353f8243",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:deed70017f76",
+      "dbms": "sqlite",
+      "title": "LIKELY(), UNLIKELY() and LIKELIHOOD() have affinities",
+      "reported_date": "2019-06-11",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=0c620df60b",
+        "fix": "https://www.sqlite.org/src/info/614ecb0af4703884"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=0c620df60b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=0c620df60b\",\n            \"fix\": \"https://www.sqlite.org/src/info/614ecb0af4703884\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:dacd0ee6b8a0bb50d933bd8c8d66e5a86b4cc2b25dd7d91be1c69274b6b66014",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:dacd0ee6b8a0bb50d933bd8c8d66e5a86b4cc2b25dd7d91be1c69274b6b66014"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=0c620df60b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:78fd03ee9677",
+      "dbms": "sqlite",
+      "title": "Lossless conversion when casting a large TEXT number to NUMERIC is not performed",
+      "reported_date": "2019-06-11",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=afdc5a29dc",
+        "fix": "https://www.sqlite.org/docsrc/info/7a51b32537ac7e95"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=afdc5a29dc",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=afdc5a29dc\",\n            \"fix\": \"https://www.sqlite.org/docsrc/info/7a51b32537ac7e95\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:8882f54ba10cf4eab08e2e7a5fcdf7611a1e25df6f5f9613d7c0d749d74348d6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8882f54ba10cf4eab08e2e7a5fcdf7611a1e25df6f5f9613d7c0d749d74348d6"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=afdc5a29dc",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:01e8bfec9c0b",
+      "dbms": "sqlite",
+      "title": "CAST takes implicit COLLATE of its operand",
+      "reported_date": "2019-06-12",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=b148fa6105",
+        "fix": "https://www.sqlite.org/docsrc/info/9f887f15e57978df"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=b148fa6105",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=b148fa6105\",\n            \"fix\": \"https://www.sqlite.org/docsrc/info/9f887f15e57978df\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:702dc2e58c39deb548728373704e01a8a1ec2375218e22df8064892dafaa8bd3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:702dc2e58c39deb548728373704e01a8a1ec2375218e22df8064892dafaa8bd3"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=b148fa6105",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:fd388b71c5ed",
+      "dbms": "sqlite",
+      "title": "CAST('-0.0' AS NUMERIC) computes 0.0 rather than 0",
+      "reported_date": "2019-06-12",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview/674385aeba91c774d47736f1aefd259b074dc5d3",
+        "fix": "https://www.sqlite.org/src/info/491f0f9bbddb6302"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview/674385aeba91c774d47736f1aefd259b074dc5d3",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview/674385aeba91c774d47736f1aefd259b074dc5d3\",\n            \"fix\": \"https://www.sqlite.org/src/info/491f0f9bbddb6302\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:638c944f662ab7fbacf984882b1d2dc3dac378e1c3fad03941d19017797c686a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:638c944f662ab7fbacf984882b1d2dc3dac378e1c3fad03941d19017797c686a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview/674385aeba91c774d47736f1aefd259b074dc5d3",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:ee99dae2aeff",
+      "dbms": "sqlite",
+      "title": "IS TRUE operator malfunctions with COLLATE and REAL value",
+      "reported_date": "2019-06-12",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=4d01eda811",
+        "fix": "https://www.sqlite.org/src/info/5c6146b56a75a94f"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=4d01eda811",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"12/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=4d01eda811\",\n            \"fix\": \"https://www.sqlite.org/src/info/5c6146b56a75a94f\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:6c269a0fde7f6879f8e314971ff0ec28a2b640ff1f8b9eac61070fdf2c307a6c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6c269a0fde7f6879f8e314971ff0ec28a2b640ff1f8b9eac61070fdf2c307a6c"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=4d01eda811",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:373312296acc",
+      "dbms": "sqlite",
+      "title": "Built-in RTRIM collating sequence yields incorrect comparisons",
+      "reported_date": "2019-06-14",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=f1580ba1b5",
+        "fix": "https://www.sqlite.org/src/info/86fa0087cd1f5c79"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=f1580ba1b5",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=f1580ba1b5\",\n            \"fix\": \"https://www.sqlite.org/src/info/86fa0087cd1f5c79\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:35c5aac530038ee1c1a7d5462b16cebc81008677c9e1f21c6b05c05e2fa3ee90",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:35c5aac530038ee1c1a7d5462b16cebc81008677c9e1f21c6b05c05e2fa3ee90"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=f1580ba1b5",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5f2dee7b4451",
+      "dbms": "sqlite",
+      "title": "LIKE malfunctions for UNIQUE COLLATE NOCASE column",
+      "reported_date": "2019-06-14",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=ce8717f088",
+        "fix": "https://www.sqlite.org/src/info/b4a9e09e60213ccf"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=ce8717f088",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=ce8717f088\",\n            \"fix\": \"https://www.sqlite.org/src/info/b4a9e09e60213ccf\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:4bae4d4d41160c0bffb0cd8ba23b26233722f4d57b8ad8690855094fff941293",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4bae4d4d41160c0bffb0cd8ba23b26233722f4d57b8ad8690855094fff941293"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=ce8717f088",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:23bb5026b499",
+      "dbms": "sqlite",
+      "title": "COLLATE in BETWEEN expression is ignored",
+      "reported_date": "2019-06-16",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=e1e07ef202",
+        "fix": "https://www.sqlite.org/src/info/54110870487f7801"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=e1e07ef202",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"16/06/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=e1e07ef202\",\n            \"fix\": \"https://www.sqlite.org/src/info/54110870487f7801\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:b6fc05d5777a6a20c7e42207ec14cb8c0feeea93bf47cfa6cd1934666953fdb6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b6fc05d5777a6a20c7e42207ec14cb8c0feeea93bf47cfa6cd1934666953fdb6"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=e1e07ef202",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d20af4158007",
+      "dbms": "sqlite",
+      "title": "MIN() malfunctions for UNIQUE column",
+      "reported_date": "2019-07-02",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=41866dc373",
+        "fix": "https://www.sqlite.org/src/info/faaaae4940b5f4f7"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=41866dc373",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/07/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=41866dc373\",\n            \"fix\": \"https://www.sqlite.org/src/info/faaaae4940b5f4f7\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:a7b67853a08f726ee3f13cd7f1cfce1e6d617475e2073a320dfd6cd17a95bfaa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a7b67853a08f726ee3f13cd7f1cfce1e6d617475e2073a320dfd6cd17a95bfaa"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=41866dc373",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:2fa3e396ee87",
+      "dbms": "sqlite",
+      "title": "ANALYZE causes DISTINCT to malfunction in CROSS JOIN",
+      "reported_date": "2019-07-29",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=ccbe5759fb",
+        "fix": "https://www.sqlite.org/src/info/6ac0f822450b26c7"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=ccbe5759fb",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/07/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=ccbe5759fb\",\n            \"fix\": \"https://www.sqlite.org/src/info/6ac0f822450b26c7\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:157af5a397905245adc587c9c9c7a58ed254d569980e431ed077297d85664472",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:157af5a397905245adc587c9c9c7a58ed254d569980e431ed077297d85664472"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=ccbe5759fb",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d73f3443ae0c",
+      "dbms": "sqlite",
+      "title": "Issue with DISTINCT and COLLATE",
+      "reported_date": "2019-07-29",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=18ab5da2c0"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=18ab5da2c0",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/07/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=18ab5da2c0\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:634872ea7f92085eabab8d06d4687ae4f3cf54619e7fb4104ccce9c105cce001",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:634872ea7f92085eabab8d06d4687ae4f3cf54619e7fb4104ccce9c105cce001"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=18ab5da2c0",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:19beb8f4e964",
+      "dbms": "sqlite",
+      "title": "Query with ORDER BY results in \"database disk image is malformed\" error",
+      "reported_date": "2019-07-29",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=ba2f4585cf",
+        "fix": "https://www.sqlite.org/src/info/2b221bb15fd2b9f6"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=ba2f4585cf",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/07/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=ba2f4585cf\",\n            \"fix\": \"https://www.sqlite.org/src/info/2b221bb15fd2b9f6\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:8e091089a525578ae67985dd47196b1ca098e0e03c111101f9f4cc51ee15204c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8e091089a525578ae67985dd47196b1ca098e0e03c111101f9f4cc51ee15204c"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=ba2f4585cf",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:af9067579f68",
+      "dbms": "sqlite",
+      "title": "Query with DISTINCT does not fetch all distinct rows",
+      "reported_date": "2019-07-30",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=ced41c7c7d",
+        "fix": "https://www.sqlite.org/src/info/a871d69c6de65038"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=ced41c7c7d",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/07/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=ced41c7c7d\",\n            \"fix\": \"https://www.sqlite.org/src/info/a871d69c6de65038\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:4b9df8e85f8bc28556aab058c8e778584347fa27700ac326f4e46498cbfbf175",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4b9df8e85f8bc28556aab058c8e778584347fa27700ac326f4e46498cbfbf175"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=ced41c7c7d",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:2c916d4dfce7",
+      "dbms": "sqlite",
+      "title": "MIN() malfunctions for a query with ISNULL condition",
+      "reported_date": "2019-08-03",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=71e183cab6",
+        "fix": "https://www.sqlite.org/src/info/d465c3ee"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=71e183cab6",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"03/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=71e183cab6\",\n            \"fix\": \"https://www.sqlite.org/src/info/d465c3ee\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:ccd8daf6465e9ae25a697c8acf8d561438222da98e99d92e50c13e0423236bac",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ccd8daf6465e9ae25a697c8acf8d561438222da98e99d92e50c13e0423236bac"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=71e183cab6",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:bbd341192650",
+      "dbms": "sqlite",
+      "title": "Row is not fetched in SELECT from VIEW",
+      "reported_date": "2019-08-05",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=61c853857f",
+        "fix1": "https://www.sqlite.org/src/info/09cd0c0c6e6c963e",
+        "fix2": "https://www.sqlite.org/src/info/470ac8d50ce2f7cc",
+        "fix3": "https://www.sqlite.org/src/info/e15a0977ddfad3d0"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=61c853857f",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=61c853857f\",\n            \"fix1\": \"https://www.sqlite.org/src/info/09cd0c0c6e6c963e\",\n            \"fix2\": \"https://www.sqlite.org/src/info/470ac8d50ce2f7cc\",\n            \"fix3\": \"https://www.sqlite.org/src/info/e15a0977ddfad3d0\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:a4404500f3d4c79207214a88c02fb0e7f78e89eb0c66b2296f8a653b4ea77907",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a4404500f3d4c79207214a88c02fb0e7f78e89eb0c66b2296f8a653b4ea77907"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=61c853857f",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:0b6887f2d37d",
+      "dbms": "sqlite",
+      "title": "Unexpected affinity conversion in view",
+      "reported_date": "2019-08-05",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=d52a29a9e6",
+        "fix": "https://www.sqlite.org/src/info/9c8c1092a8ce80e1"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=d52a29a9e6",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=d52a29a9e6\",\n            \"fix\": \"https://www.sqlite.org/src/info/9c8c1092a8ce80e1\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:b780fe4133b08c403e1e329099fa57909800693eb600dbe9dd1dc7b2c0103a5a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b780fe4133b08c403e1e329099fa57909800693eb600dbe9dd1dc7b2c0103a5a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=d52a29a9e6",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:b9c0b763c730",
+      "dbms": "sqlite",
+      "title": "Unexpected affinity conversion for view column in IN operator",
+      "reported_date": "2019-08-06",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=0a5e2c1dcb",
+        "fix": "https://www.sqlite.org/src/info/17b3d2218c02a400"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=0a5e2c1dcb",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"06/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=0a5e2c1dcb\",\n            \"fix\": \"https://www.sqlite.org/src/info/17b3d2218c02a400\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:b6d0826fdd3384afbb0a50017cca57c4137d0fcfea0edfcb245bfe3c22b83088",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b6d0826fdd3384afbb0a50017cca57c4137d0fcfea0edfcb245bfe3c22b83088"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=0a5e2c1dcb",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:bbe43099ac00",
+      "dbms": "sqlite",
+      "title": "Incorrect result for query that uses MIN() and a CAST on rowid",
+      "reported_date": "2019-08-07",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=f8a7060ece",
+        "fix": "https://www.sqlite.org/src/info/94085fb3"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=f8a7060ece",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=f8a7060ece\",\n            \"fix\": \"https://www.sqlite.org/src/info/94085fb3\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:2a86678946caa81aa41a6014cb3c73c8a1b01686231244aa7c196bc0daffe68a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2a86678946caa81aa41a6014cb3c73c8a1b01686231244aa7c196bc0daffe68a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=f8a7060ece",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d0620ca2a569",
+      "dbms": "sqlite",
+      "title": "Constant expression in partial index results in row not being fetched",
+      "reported_date": "2019-08-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=9080b6227f",
+        "fix": "https://www.sqlite.org/src/info/da01ba4fa47c6508"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=9080b6227f",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=9080b6227f\",\n            \"fix\": \"https://www.sqlite.org/src/info/da01ba4fa47c6508\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:947efd1d18cf0d7e53ee224b6f38d4b6b0dfef2f607f43997dc87aac138c84c7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:947efd1d18cf0d7e53ee224b6f38d4b6b0dfef2f607f43997dc87aac138c84c7"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=9080b6227f",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:a6cd19fb0b0a",
+      "dbms": "sqlite",
+      "title": "Unexpected error in DELETE with existing trigger",
+      "reported_date": "2019-08-14",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=17db54d744"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=17db54d744",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=17db54d744\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:2093760d56403ee65c7b4a0f3ff4c770a25122cfc0ec932971324fe70f76094e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2093760d56403ee65c7b4a0f3ff4c770a25122cfc0ec932971324fe70f76094e"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=17db54d744",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:017f92933b81",
+      "dbms": "sqlite",
+      "title": "Null pointer dereference caused by window functions in result-set of EXISTS(SELECT ...)",
+      "reported_date": "2019-08-15",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=256741a16b",
+        "fix": "https://www.sqlite.org/src/info/4f5b2d938194fab7"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=256741a16b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=256741a16b\",\n            \"fix\": \"https://www.sqlite.org/src/info/4f5b2d938194fab7\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:d0bb7a8965ced811bf5a51340c770e8cd46b47035c2ea694a861e484bc55a3f6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d0bb7a8965ced811bf5a51340c770e8cd46b47035c2ea694a861e484bc55a3f6"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=256741a16b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:2740bed3b4f4",
+      "dbms": "sqlite",
+      "title": "LEFT JOIN fails to fetch row",
+      "reported_date": "2019-08-17",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=6710d2f7a1",
+        "fix": "https://www.sqlite.org/src/info/500c9152daaf11cf"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=6710d2f7a1",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"17/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=6710d2f7a1\",\n            \"fix\": \"https://www.sqlite.org/src/info/500c9152daaf11cf\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:7ba841eba716c66a5fcaa9e3af8f92552c5ad918507f46acb1c71628026fbda7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7ba841eba716c66a5fcaa9e3af8f92552c5ad918507f46acb1c71628026fbda7"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=6710d2f7a1",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:86ae80e6fd24",
+      "dbms": "sqlite",
+      "title": "WHERE clause erroneously influences value of fetched column from view",
+      "reported_date": "2019-08-17",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=c7a1171907",
+        "fix": "https://www.sqlite.org/src/info/43e8b143"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=c7a1171907",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"17/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=c7a1171907\",\n            \"fix\": \"https://www.sqlite.org/src/info/43e8b143\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:adb3c39a069433adbe2bca3dc7f25f8064f5c43d0b993efe9ecdf05842af4b9e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:adb3c39a069433adbe2bca3dc7f25f8064f5c43d0b993efe9ecdf05842af4b9e"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=c7a1171907",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:248d3a9c57fa",
+      "dbms": "sqlite",
+      "title": "INDEXED BY results in row not being fetched",
+      "reported_date": "2019-08-20",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=f043b1130b",
+        "fix": "https://www.sqlite.org/src/info/511da081"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=f043b1130b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=f043b1130b\",\n            \"fix\": \"https://www.sqlite.org/src/info/511da081\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:60671ee428d2adf5028979d7ac7eed349417610bd6a63fc5e03b86522ff41053",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:60671ee428d2adf5028979d7ac7eed349417610bd6a63fc5e03b86522ff41053"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=f043b1130b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:fa5c1f9f171d",
+      "dbms": "sqlite",
+      "title": "DISTINCT malfunctions for IS NULL",
+      "reported_date": "2019-08-21",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=b86894020e",
+        "fix": "https://www.sqlite.org/src/info/d02490a2"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=b86894020e",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=b86894020e\",\n            \"fix\": \"https://www.sqlite.org/src/info/d02490a2\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:6215fb167bdf9b64834a76b89f93bfe113957658099413e35ff795b46481baf2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6215fb167bdf9b64834a76b89f93bfe113957658099413e35ff795b46481baf2"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=b86894020e",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:7bb7967d51f3",
+      "dbms": "sqlite",
+      "title": "LIKELY() seems to cause unexpected affinity conversion for rowid",
+      "reported_date": "2019-08-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=7e07a3dbf5",
+        "fix": "https://www.sqlite.org/src/info/44578865fa7baf97"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=7e07a3dbf5",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=7e07a3dbf5\",\n            \"fix\": \"https://www.sqlite.org/src/info/44578865fa7baf97\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:a4e613a5df395143dba8c4942dbdb58b579d12783ee2b69831a5963a24472cdb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a4e613a5df395143dba8c4942dbdb58b579d12783ee2b69831a5963a24472cdb"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=7e07a3dbf5",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:204e9ccab06b",
+      "dbms": "sqlite",
+      "title": "Row is not fetched in table with INTEGER PRIMARY KEY",
+      "reported_date": "2019-08-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=d9f584e936",
+        "fix": "https://www.sqlite.org/src/info/81b9f0f55042777b"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=d9f584e936",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=d9f584e936\",\n            \"fix\": \"https://www.sqlite.org/src/info/81b9f0f55042777b\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:6e77f3d4fee269983cd8e44534225c1fe0e29239975b7b32e2077232d297605a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6e77f3d4fee269983cd8e44534225c1fe0e29239975b7b32e2077232d297605a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=d9f584e936",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:cc77eae22575",
+      "dbms": "sqlite",
+      "title": "Row with comparison on TEXT UNIQUE column is not fetched",
+      "reported_date": "2019-08-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=ac184eb571",
+        "fix": "https://www.sqlite.org/src/info/e62eddbb048cbc2c"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=ac184eb571",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=ac184eb571\",\n            \"fix\": \"https://www.sqlite.org/src/info/e62eddbb048cbc2c\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:90ef1ca370c5244ff466ed4e6e94d8b1063927b86cc1b90f5f2a7c3a9e676355",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:90ef1ca370c5244ff466ed4e6e94d8b1063927b86cc1b90f5f2a7c3a9e676355"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=ac184eb571",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:f4941d184ba2",
+      "dbms": "sqlite",
+      "title": "Unexpected affinity conversion is performed for the IN operator",
+      "reported_date": "2019-08-27",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=dbaf8a6820",
+        "fix": "https://www.sqlite.org/src/info/7f5168a76a400fc2"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=dbaf8a6820",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"27/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=dbaf8a6820\",\n            \"fix\": \"https://www.sqlite.org/src/info/7f5168a76a400fc2\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:5e5f574124ce3e1e7e514bdad52e247fb52c6e0f6011bd6fddcea5cb2c98b2b7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5e5f574124ce3e1e7e514bdad52e247fb52c6e0f6011bd6fddcea5cb2c98b2b7"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=dbaf8a6820",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:c51816e58380",
+      "dbms": "sqlite",
+      "title": "Short-circuit evaluation issue",
+      "reported_date": "2019-08-28",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=51f5b6e427"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=51f5b6e427",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"28/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=51f5b6e427\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:37fe08b3ab1752bbbaec0db734f3ce4fced6e5a927e83b4bd6de74fbd535feb2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:37fe08b3ab1752bbbaec0db734f3ce4fced6e5a927e83b4bd6de74fbd535feb2"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=51f5b6e427",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:b8fe06f50611",
+      "dbms": "sqlite",
+      "title": "Partial index and BETWEEN issue",
+      "reported_date": "2019-08-30",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=f8f472cbc7",
+        "fix": "https://www.sqlite.org/src/info/057fb8b1809b8b9c"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=f8f472cbc7",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=f8f472cbc7\",\n            \"fix\": \"https://www.sqlite.org/src/info/057fb8b1809b8b9c\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:cdbda9d8f8559bb3a1a709e9fbfdb4aca7678ea33fab0fcfd5dd72b135af46f4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:cdbda9d8f8559bb3a1a709e9fbfdb4aca7678ea33fab0fcfd5dd72b135af46f4"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=f8f472cbc7",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:43080d24fa41",
+      "dbms": "sqlite",
+      "title": "Partial index causes row to not be fetched",
+      "reported_date": "2019-08-30",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=a6408d42b9",
+        "fix": "https://www.sqlite.org/src/info/45ff2b1f2693bb02"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=a6408d42b9",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=a6408d42b9\",\n            \"fix\": \"https://www.sqlite.org/src/info/45ff2b1f2693bb02\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:62287ce9e4b3b719fdfb885b43ee3c2fd2ec2c9ff86509d27e48751e85a535d1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:62287ce9e4b3b719fdfb885b43ee3c2fd2ec2c9ff86509d27e48751e85a535d1"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=a6408d42b9",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:b70af5d67ca7",
+      "dbms": "sqlite",
+      "title": "Partial index causes row to not be fetched in BETWEEN expression",
+      "reported_date": "2019-08-30",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=fba33c8b1d",
+        "fix": "https://www.sqlite.org/src/info/057fb8b1809b8b9c"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=fba33c8b1d",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=fba33c8b1d\",\n            \"fix\": \"https://www.sqlite.org/src/info/057fb8b1809b8b9c\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:0c51960137a99436153c3e602586d6bc3f11f50ffc3e38879a090e1161acdf2a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0c51960137a99436153c3e602586d6bc3f11f50ffc3e38879a090e1161acdf2a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=fba33c8b1d",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:146ab8355743",
+      "dbms": "sqlite",
+      "title": "REINDEX causes \"UNIQUE constraint failed\" error",
+      "reported_date": "2019-08-30",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=5a3dba8104",
+        "fix": "https://www.sqlite.org/src/info/67381dadede98a55"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=5a3dba8104",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/08/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=5a3dba8104\",\n            \"fix\": \"https://www.sqlite.org/src/info/67381dadede98a55\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:dd3540fc3d03d1cac90356bb73a772e63abd54cfaf4cd5909935053cc3f09746",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:dd3540fc3d03d1cac90356bb73a772e63abd54cfaf4cd5909935053cc3f09746"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=5a3dba8104",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:e04eff9f4dba",
+      "dbms": "sqlite",
+      "title": "Different rounding when converting TEXT to REAL",
+      "reported_date": "2019-09-02",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=2841e99d10",
+        "fix": "https://www.sqlite.org/src/info/88833a9c2849c959"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=2841e99d10",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/09/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=2841e99d10\",\n            \"fix\": \"https://www.sqlite.org/src/info/88833a9c2849c959\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:3633481ab6cb79aec078f2adbbe841d67dfd272fd0a74995d79506fe1dbb9f50",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3633481ab6cb79aec078f2adbbe841d67dfd272fd0a74995d79506fe1dbb9f50"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=2841e99d10",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:9e6c456d95aa",
+      "dbms": "sqlite",
+      "title": "Expression computed on row yields incorrect result",
+      "reported_date": "2019-09-02",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=57af00b664",
+        "fix": "https://www.sqlite.org/src/info/0658c16e311393c8"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=57af00b664",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/09/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=57af00b664\",\n            \"fix\": \"https://www.sqlite.org/src/info/0658c16e311393c8\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:3edd2b6edf0b57945b1e4358181480f89cb89827b92d9a16c2998fa9fd824c30",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3edd2b6edf0b57945b1e4358181480f89cb89827b92d9a16c2998fa9fd824c30"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=57af00b664",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:ffd30f457be2",
+      "dbms": "sqlite",
+      "title": "COLLATE NOCASE string comparison yields incorrect result",
+      "reported_date": "2019-09-03",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=767a8cbc6d",
+        "fix": "https://www.sqlite.org/src/info/5351e920f489562f"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=767a8cbc6d",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"03/09/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=767a8cbc6d\",\n            \"fix\": \"https://www.sqlite.org/src/info/5351e920f489562f\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:08e801fda3adda108901a162d781223678f9f74e0da6666be694806280762016",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:08e801fda3adda108901a162d781223678f9f74e0da6666be694806280762016"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=767a8cbc6d",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:008ff94c9de8",
+      "dbms": "sqlite",
+      "title": "IS NULL unexpectedly evaluates to TRUE",
+      "reported_date": "2019-09-03",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=29f635e0af",
+        "fix": "https://www.sqlite.org/src/info/6e7b4527d32cc1be"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=29f635e0af",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"03/09/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=29f635e0af\",\n            \"fix\": \"https://www.sqlite.org/src/info/6e7b4527d32cc1be\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:f2e94fb30f62e5817e556044b84ab6788dd7d5faa57a819c42fd9f043c1f120d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f2e94fb30f62e5817e556044b84ab6788dd7d5faa57a819c42fd9f043c1f120d"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=29f635e0af",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:9ebd47ea8673",
+      "dbms": "sqlite",
+      "title": "BETWEEN issue in view",
+      "reported_date": "2019-09-09",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=a7debbe0ad",
+        "fix": "https://www.sqlite.org/src/info/b9ec72203c19c2b9"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=a7debbe0ad",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"09/09/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=a7debbe0ad\",\n            \"fix\": \"https://www.sqlite.org/src/info/b9ec72203c19c2b9\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:e9306071cc20fff19b276ed952810d28a26c679f10f61aaf4f3485329f5b29c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e9306071cc20fff19b276ed952810d28a26c679f10f61aaf4f3485329f5b29c2"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=a7debbe0ad",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:bca0aced5075",
+      "dbms": "sqlite",
+      "title": "DISTINCT malfunctions for VIEW",
+      "reported_date": "2019-09-10",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=c4130c33be"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=c4130c33be",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/09/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=c4130c33be\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:09baf6565ec2c2a33e09285b6e609de4a88908a8b7ff787c62e4656606e0e2cf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:09baf6565ec2c2a33e09285b6e609de4a88908a8b7ff787c62e4656606e0e2cf"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=c4130c33be",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:101331cf59f0",
+      "dbms": "sqlite",
+      "title": "COLLATE issue in view",
+      "reported_date": "2019-09-11",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=18458b1ad6",
+        "fix": "https://www.sqlite.org/src/info/36997c4ade2ef3a2"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=18458b1ad6",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/09/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=18458b1ad6\",\n            \"fix\": \"https://www.sqlite.org/src/info/36997c4ade2ef3a2\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:d0b08bbbb84296f9366ec2673f1d8ba14112e7f1091503959ca46c2de92a5072",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d0b08bbbb84296f9366ec2673f1d8ba14112e7f1091503959ca46c2de92a5072"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=18458b1ad6",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:57813ca43f75",
+      "dbms": "sqlite",
+      "title": "GLOB unexpectedly does not match",
+      "reported_date": "2019-09-16",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=0f0428096f",
+        "fix": "https://www.sqlite.org/src/info/6fe0367f9a337b7c"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=0f0428096f",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"16/09/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=0f0428096f\",\n            \"fix\": \"https://www.sqlite.org/src/info/6fe0367f9a337b7c\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:2a86c73111b4254a5f65f2a2036522cf0fbfac548421d0e4678f28e9f8653af7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2a86c73111b4254a5f65f2a2036522cf0fbfac548421d0e4678f28e9f8653af7"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=0f0428096f",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:31417196d76b",
+      "dbms": "sqlite",
+      "title": "Row is not fetched when using WHERE clause with INSTR()",
+      "reported_date": "2019-09-17",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=587791f926",
+        "fix": "https://www.sqlite.org/src/info/3fb40f518086c1e8"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=587791f926",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"17/09/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=587791f926\",\n            \"fix\": \"https://www.sqlite.org/src/info/3fb40f518086c1e8\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:b441fc8a849e7b2b5fbb5b30f4c352e1b0bc8c8eeee11c1b4af00e0f170ee54f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b441fc8a849e7b2b5fbb5b30f4c352e1b0bc8c8eeee11c1b4af00e0f170ee54f"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=587791f926",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:bb05ebc54841",
+      "dbms": "sqlite",
+      "title": "Comparison on view malfunctions",
+      "reported_date": "2019-10-07",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=b2d4edaffd",
+        "fix": "https://www.sqlite.org/src/info/3cde82c86b963fa7"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=b2d4edaffd",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=b2d4edaffd\",\n            \"fix\": \"https://www.sqlite.org/src/info/3cde82c86b963fa7\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:154edbd986f392a3846017c5a888f07bb89f343af66b832b4312dfa8478ad561",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:154edbd986f392a3846017c5a888f07bb89f343af66b832b4312dfa8478ad561"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=b2d4edaffd",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:520a98c5b13e",
+      "dbms": "sqlite",
+      "title": "FTS integrity-check malfunctions",
+      "reported_date": "2019-10-07",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=8fe768e9c9",
+        "fix": "https://www.sqlite.org/src/info/31e85fbbc4cfd09a"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=8fe768e9c9",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=8fe768e9c9\",\n            \"fix\": \"https://www.sqlite.org/src/info/31e85fbbc4cfd09a\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:56b6f8aad6dfe1a51efefa5a8fdfdb1eb37a1d10e448001c4bbb95116d940c7a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:56b6f8aad6dfe1a51efefa5a8fdfdb1eb37a1d10e448001c4bbb95116d940c7a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=8fe768e9c9",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5fd91490914d",
+      "dbms": "sqlite",
+      "title": "FTS pgsz option results in \"database disk image is malformed\" error",
+      "reported_date": "2019-10-07",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=265e935b26",
+        "fix": "https://www.sqlite.org/src/info/8ab0aebdb3c2d6fb"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=265e935b26",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=265e935b26\",\n            \"fix\": \"https://www.sqlite.org/src/info/8ab0aebdb3c2d6fb\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:6eb81c70f7e97bd6b122992d7a40463be57ed40caead6a0ba9d121bb3b1b752b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6eb81c70f7e97bd6b122992d7a40463be57ed40caead6a0ba9d121bb3b1b752b"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=265e935b26",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:f0e1df4f0d2d",
+      "dbms": "sqlite",
+      "title": "FTS rebuild in transaction causes integrity-check to fail",
+      "reported_date": "2019-10-07",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=e258f008ce",
+        "fix": "https://www.sqlite.org/src/info/238e0835714696ab"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=e258f008ce",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"07/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=e258f008ce\",\n            \"fix\": \"https://www.sqlite.org/src/info/238e0835714696ab\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:cf1789e6b3b775d418ee4e7bff56dbcd506fdecbc3b889d63bf289f9b68d8d03",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:cf1789e6b3b775d418ee4e7bff56dbcd506fdecbc3b889d63bf289f9b68d8d03"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=e258f008ce",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d91633beacb6",
+      "dbms": "sqlite",
+      "title": "FTS integrity-check indicates that the database disk image is malformed",
+      "reported_date": "2019-10-09",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=81a7f7b9e2",
+        "fix": "https://www.sqlite.org/src/info/75775c5ab44e497c"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=81a7f7b9e2",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"09/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=81a7f7b9e2\",\n            \"fix\": \"https://www.sqlite.org/src/info/75775c5ab44e497c\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:3bf109d3a5efe09c2a74967bed2bc68b031ccf286d2b01253e26403ca8e7cc47",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3bf109d3a5efe09c2a74967bed2bc68b031ccf286d2b01253e26403ca8e7cc47"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=81a7f7b9e2",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d4668aee9ca3",
+      "dbms": "sqlite",
+      "title": "FTS rebuild in combination with crisismerge results in error \"database or disk is full\"",
+      "reported_date": "2019-10-09",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=d392017c8e",
+        "fix": "https://www.sqlite.org/src/info/86e497209217abb7"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=d392017c8e",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"09/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=d392017c8e\",\n            \"fix\": \"https://www.sqlite.org/src/info/86e497209217abb7\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a1540868f1589b308660a836c8e5ccb613a886e6bcfa3998a140066d18185cc4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a1540868f1589b308660a836c8e5ccb613a886e6bcfa3998a140066d18185cc4"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=d392017c8e",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:75f6320c8f38",
+      "dbms": "sqlite",
+      "title": "LEFT JOIN in view malfunctions",
+      "reported_date": "2019-10-09",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=a976c487d1",
+        "fix": "https://www.sqlite.org/src/info/eb7ed90b8a65748f"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=a976c487d1",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"09/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=a976c487d1\",\n            \"fix\": \"https://www.sqlite.org/src/info/eb7ed90b8a65748f\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:47018999e2099fddadb47b9a90521cb44d70cdae04012a4f7e9287b511412165",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:47018999e2099fddadb47b9a90521cb44d70cdae04012a4f7e9287b511412165"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=a976c487d1",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:721c3d141f0f",
+      "dbms": "sqlite",
+      "title": "LEFT JOIN in view malfunctions with NOTNULL",
+      "reported_date": "2019-10-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=c31034044b",
+        "fix": "https://www.sqlite.org/src/info/7833feecfe745e23"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=c31034044b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=c31034044b\",\n            \"fix\": \"https://www.sqlite.org/src/info/7833feecfe745e23\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:953b26ad81b9daee5e93787fe79ce277b6acfbfb2b663eec66ca78c2732a07d6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:953b26ad81b9daee5e93787fe79ce277b6acfbfb2b663eec66ca78c2732a07d6"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=c31034044b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d038ae314c6e",
+      "dbms": "sqlite",
+      "title": "FTS integrity-check malfunctions for transaction and the prefix option",
+      "reported_date": "2019-10-11",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=745f1abcdc",
+        "fix": "https://www.sqlite.org/src/info/4ed905b18847d4db"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=745f1abcdc",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=745f1abcdc\",\n            \"fix\": \"https://www.sqlite.org/src/info/4ed905b18847d4db\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:daf843852c24cb6d139ff42faf0a2a68f92e8e64f83c7e9377a641344a20d89c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:daf843852c24cb6d139ff42faf0a2a68f92e8e64f83c7e9377a641344a20d89c"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=745f1abcdc",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:55e3a688e659",
+      "dbms": "sqlite",
+      "title": "FTS integrity_check fails when inserting x'00'",
+      "reported_date": "2019-10-11",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=278ac14262",
+        "fix": "https://www.sqlite.org/src/info/629e20c9880acc2c"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=278ac14262",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=278ac14262\",\n            \"fix\": \"https://www.sqlite.org/src/info/629e20c9880acc2c\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:0c2d4f8ab1e6c60ee816577b3013a15dd3d32fa1e309751b813e579fe61471c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0c2d4f8ab1e6c60ee816577b3013a15dd3d32fa1e309751b813e579fe61471c2"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=278ac14262",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:e3037e6c08e6",
+      "dbms": "sqlite",
+      "title": "FTS order=DESC results into integrity-check failing",
+      "reported_date": "2019-10-11",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=8a6fa2bb22",
+        "fix": "https://www.sqlite.org/src/info/5863546df99abd1a"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=8a6fa2bb22",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=8a6fa2bb22\",\n            \"fix\": \"https://www.sqlite.org/src/info/5863546df99abd1a\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ac25d616f69fe8e676cce6a22d919aa110a9a86df8dfe26532fa8c68b84eb700",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ac25d616f69fe8e676cce6a22d919aa110a9a86df8dfe26532fa8c68b84eb700"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=8a6fa2bb22",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:e05794cf64c4",
+      "dbms": "sqlite",
+      "title": "Trigger inserts duplicate value in UNIQUE column",
+      "reported_date": "2019-10-16",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=a8a4847a2d",
+        "fix": "https://www.sqlite.org/src/info/eea1e7aa57e74c43"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=a8a4847a2d",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"16/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=a8a4847a2d\",\n            \"fix\": \"https://www.sqlite.org/src/info/eea1e7aa57e74c43\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:748587b47867043f56a14185c3ce5c448f128e7a386052932485d9f92ca7727f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:748587b47867043f56a14185c3ce5c448f128e7a386052932485d9f92ca7727f"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=a8a4847a2d",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:e58b3b1eb830",
+      "dbms": "sqlite",
+      "title": "FTS merge does not terminate",
+      "reported_date": "2019-10-17",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "hang",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=bf1aab8978",
+        "fix": "https://www.sqlite.org/src/info/35beaee059a6ccce"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=bf1aab8978",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"17/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=bf1aab8978\",\n            \"fix\": \"https://www.sqlite.org/src/info/35beaee059a6ccce\"\n        },\n        \"oracle\": \"hang\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:526a5ad414ada08d6b1162d469916e2f18400ebf32e0b521556e8a06b8dbb8c9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:526a5ad414ada08d6b1162d469916e2f18400ebf32e0b521556e8a06b8dbb8c9"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=bf1aab8978",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:2b0e69f3dab0",
+      "dbms": "sqlite",
+      "title": "INSERT into table with two triggers does not terminate",
+      "reported_date": "2019-10-21",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "hang",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=7fc8e5ff25"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=7fc8e5ff25",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=7fc8e5ff25\"\n        },\n        \"oracle\": \"hang\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:96655b870f1c92aedb071bd179ccbaf4f9d90856c7ad4ca09228053b2ccfb7ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:96655b870f1c92aedb071bd179ccbaf4f9d90856c7ad4ca09228053b2ccfb7ae"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=7fc8e5ff25",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:2d53754c5c22",
+      "dbms": "sqlite",
+      "title": "Comparison of row values results in incorrect result",
+      "reported_date": "2019-10-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=b47e3627ec",
+        "fix": "https://www.sqlite.org/src/info/90f7c477354d67d2"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=b47e3627ec",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=b47e3627ec\",\n            \"fix\": \"https://www.sqlite.org/src/info/90f7c477354d67d2\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:5f8544328f81ca803b4234d76011bc932d9dc5b75de339fd6afe8d949475385f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=b47e3627ec\",\n            \"fix\": \"https://www.sqlite.org/src/info/c7da1c01f1f239e6\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:2bfc62899683f7a2d7b1d0ddd4e4f27375e2fe2b0ac3bfffa77acd0672e2c195",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5f8544328f81ca803b4234d76011bc932d9dc5b75de339fd6afe8d949475385f"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=b47e3627ec",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:8456e6f3a25d",
+      "dbms": "sqlite",
+      "title": "Row value comparison yields incorrect result",
+      "reported_date": "2019-10-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=6ef984af89",
+        "fix": "https://www.sqlite.org/src/info/5c118617cf08e17a"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=6ef984af89",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=6ef984af89\",\n            \"fix\": \"https://www.sqlite.org/src/info/5c118617cf08e17a\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:f3758fc525b8a3aec2398acc42c9a54d8837de5265b01ff11c6a06fc7ac8da7d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f3758fc525b8a3aec2398acc42c9a54d8837de5265b01ff11c6a06fc7ac8da7d"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=6ef984af89",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:0e98334c1bcb",
+      "dbms": "sqlite",
+      "title": "Comparison of row values with COLLATE NOCASE yields incorrect result",
+      "reported_date": "2019-10-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=135c9da751",
+        "fix": "https://www.sqlite.org/src/info/978b2d20cf95d0b7"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=135c9da751",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=135c9da751\",\n            \"fix\": \"https://www.sqlite.org/src/info/978b2d20cf95d0b7\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:3e0ef5d14bc369dc34dcc204fb52fc0b82f74b84c6febb3f83089f3190b3125a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3e0ef5d14bc369dc34dcc204fb52fc0b82f74b84c6febb3f83089f3190b3125a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=135c9da751",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:3f0fe10caafe",
+      "dbms": "sqlite",
+      "title": "Crash on REPLACE INTO of a table with an AFTER DELETE trigger",
+      "reported_date": "2019-10-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=50c09fc2cf",
+        "fix": "https://www.sqlite.org/src/info/521f1d3628254948"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=50c09fc2cf",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=50c09fc2cf\",\n            \"fix\": \"https://www.sqlite.org/src/info/521f1d3628254948\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7fed0450347c4ab759835baa68a87a80b852797490581c09710901f96d4af4d9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7fed0450347c4ab759835baa68a87a80b852797490581c09710901f96d4af4d9"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=50c09fc2cf",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:a7e7e2b4c9c0",
+      "dbms": "sqlite",
+      "title": "FTS integrity-check malfunctions nondeterministically with tokenize=\"ascii\"",
+      "reported_date": "2019-10-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "minor",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=dd1f67bf25",
+        "fix": "https://www.sqlite.org/src/info/8d964e1c21d4cea6"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=dd1f67bf25",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=dd1f67bf25\",\n            \"fix\": \"https://www.sqlite.org/src/info/8d964e1c21d4cea6\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:bab46e92a70890c7952ffecf624fbfaf95a5343a9966927e4f0189e4ec186ccd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:bab46e92a70890c7952ffecf624fbfaf95a5343a9966927e4f0189e4ec186ccd"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=dd1f67bf25",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:6d3e54470102",
+      "dbms": "sqlite",
+      "title": "Trigger causes query to compute incorrect result",
+      "reported_date": "2019-10-25",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=c1e19e1204",
+        "fix": "https://www.sqlite.org/src/info/fbac0c65d8464b12"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=c1e19e1204",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=c1e19e1204\",\n            \"fix\": \"https://www.sqlite.org/src/info/fbac0c65d8464b12\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:c55a7639c9c2e5637da5066b1edd17a55656aac2446038cb53653eedc3bd9eba",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c55a7639c9c2e5637da5066b1edd17a55656aac2446038cb53653eedc3bd9eba"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=c1e19e1204",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:bf784708f029",
+      "dbms": "sqlite",
+      "title": "REINDEX causes \"UNIQUE constraint failed\" error for generated column",
+      "reported_date": "2019-10-26",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=3ea1755124",
+        "fix": "https://www.sqlite.org/src/info/5b4c0f2ddc6f324e"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=3ea1755124",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=3ea1755124\",\n            \"fix\": \"https://www.sqlite.org/src/info/5b4c0f2ddc6f324e\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:56d9dda8ad3935bcbbe21d2675abd35e0085361f932453947a82941107722e4b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:56d9dda8ad3935bcbbe21d2675abd35e0085361f932453947a82941107722e4b"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=3ea1755124",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:77d9fdf0af3d",
+      "dbms": "sqlite",
+      "title": "Segfault in table with generated columns",
+      "reported_date": "2019-10-27",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=91e8695101",
+        "fix": "https://www.sqlite.org/src/info/6d1bbba9a004a249"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=91e8695101",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"27/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=91e8695101\",\n            \"fix\": \"https://www.sqlite.org/src/info/6d1bbba9a004a249\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e23cb422638fc4d64a6fc1cfc61c44e5be66dd73d6c1fab9539ca5b11e3ac69b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e23cb422638fc4d64a6fc1cfc61c44e5be66dd73d6c1fab9539ca5b11e3ac69b"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=91e8695101",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:b23ce4764326",
+      "dbms": "sqlite",
+      "title": "Segfault when updating table with generated columns",
+      "reported_date": "2019-10-29",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=9621dd78a0",
+        "fix": "https://sqlite.org/src/info/361ea81ae8a13e7d"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=9621dd78a0",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=9621dd78a0\",\n            \"fix\": \"https://sqlite.org/src/info/361ea81ae8a13e7d\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5e8ddad737f2f41b8cf6d5b3c12932c0cd144fb866d58c528483385d0eff6035",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5e8ddad737f2f41b8cf6d5b3c12932c0cd144fb866d58c528483385d0eff6035"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=9621dd78a0",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:4b7fcd50eba2",
+      "dbms": "sqlite",
+      "title": "VACUUM issue on table with generated column",
+      "reported_date": "2019-10-29",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=166347c6fc",
+        "fix": "https://sqlite.org/src/info/4fba090e678ef184"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=166347c6fc",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=166347c6fc\",\n            \"fix\": \"https://sqlite.org/src/info/4fba090e678ef184\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:feb87f11665453d13aa7a519ecce3eb931fb486504bcfd50386d4fc33903b02a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:feb87f11665453d13aa7a519ecce3eb931fb486504bcfd50386d4fc33903b02a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=166347c6fc",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:3361d384c76a",
+      "dbms": "sqlite",
+      "title": "VACUUM on table with generated column results in an error",
+      "reported_date": "2019-10-29",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=6484e6ce67",
+        "fix": "https://sqlite.org/src/info/4d424f3047b48fc4"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=6484e6ce67",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=6484e6ce67\",\n            \"fix\": \"https://sqlite.org/src/info/4d424f3047b48fc4\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f846ba80274a95678220262a921f0db8ba1ef527c4ba53c19cdf0a50d29ed199",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f846ba80274a95678220262a921f0db8ba1ef527c4ba53c19cdf0a50d29ed199"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=6484e6ce67",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:c7d4bef37b63",
+      "dbms": "sqlite",
+      "title": "Segfault in table with generated column and foreign key",
+      "reported_date": "2019-10-31",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=b9befa4b83",
+        "fix": "https://sqlite.org/src/info/40d3282ec285d9f7"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=b9befa4b83",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"31/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=b9befa4b83\",\n            \"fix\": \"https://sqlite.org/src/info/40d3282ec285d9f7\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c4ffd86e8e01cca3879b3c476ab4214550d61308ab8a30c5bad6b2956571f969",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c4ffd86e8e01cca3879b3c476ab4214550d61308ab8a30c5bad6b2956571f969"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=b9befa4b83",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:6ff1cb9a50fd",
+      "dbms": "sqlite",
+      "title": "VACUUM on table with generated column that uses TYPEOF results in an error",
+      "reported_date": "2019-10-31",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=1d2a8efc6c",
+        "fix": "https://sqlite.org/src/info/329820673a12ff6a"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=1d2a8efc6c",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"31/10/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=1d2a8efc6c\",\n            \"fix\": \"https://sqlite.org/src/info/329820673a12ff6a\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e5e0ca6ae3cfab04a6384458c82d245612957e2b8cbd9b62c51dfb4b55b1d984",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e5e0ca6ae3cfab04a6384458c82d245612957e2b8cbd9b62c51dfb4b55b1d984"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=1d2a8efc6c",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:258e2e6740d3",
+      "dbms": "sqlite",
+      "title": "REPLACE causes segfault in table with generated column and foreign key",
+      "reported_date": "2019-11-01",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=c28a01da72",
+        "fix": "https://sqlite.org/src/info/bc6a43e7ee6353b9"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=c28a01da72",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"1/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=c28a01da72\",\n            \"fix\": \"https://sqlite.org/src/info/bc6a43e7ee6353b9\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a3d15a88e14f0b5f65a549e96dc9250c3974a631b07a3f6833f16ec19683ae45",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a3d15a88e14f0b5f65a549e96dc9250c3974a631b07a3f6833f16ec19683ae45"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=c28a01da72",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:8be2f3e8ca90",
+      "dbms": "sqlite",
+      "title": "Incorrect result for GLOB operator",
+      "reported_date": "2019-11-02",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=a9efb42811",
+        "fix": "https://sqlite.org/src/info/17e9f65814264de9"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=a9efb42811",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"2/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=a9efb42811\",\n            \"fix\": \"https://sqlite.org/src/info/17e9f65814264de9\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:1e38b1c404e30d8a40fa64904e05c2ecdc4a1e915f401c38764f90aa2c422bc7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1e38b1c404e30d8a40fa64904e05c2ecdc4a1e915f401c38764f90aa2c422bc7"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=a9efb42811",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:f868d1f440e2",
+      "dbms": "sqlite",
+      "title": "LEFT JOIN in view malfunctions with partial index on table",
+      "reported_date": "2019-11-03",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=623eff57e7",
+        "fix": "https://sqlite.org/src/info/3be19e1151af1850"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=623eff57e7",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"3/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=623eff57e7\",\n            \"fix\": \"https://sqlite.org/src/info/3be19e1151af1850\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:3dc6f1ad92d5eb21ee727a86c443b848c80bb5fff87d9a3f63af03e0e9da5df5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3dc6f1ad92d5eb21ee727a86c443b848c80bb5fff87d9a3f63af03e0e9da5df5"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=623eff57e7",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:0c96ebd36a9c",
+      "dbms": "sqlite",
+      "title": "PRAGMA integrity_check fails due to CHECK constraint even without records",
+      "reported_date": "2019-11-03",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=3c9eadd2a6",
+        "fix": "https://www.sqlite.org/src/info/c5f96a085db9688a"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=3c9eadd2a6",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"3/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=3c9eadd2a6\",\n            \"fix\": \"https://www.sqlite.org/src/info/c5f96a085db9688a\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ef6865e1119401acb6a3513f584686d6065429bc036a4f99522ca410e0375dc9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ef6865e1119401acb6a3513f584686d6065429bc036a4f99522ca410e0375dc9"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=3c9eadd2a6",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:38252adf8c92",
+      "dbms": "sqlite",
+      "title": "Row value comparison malfunctions on view with left join",
+      "reported_date": "2019-11-04",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=02aa2bd02f",
+        "fix": "https://sqlite.org/src/info/ea20068e6d97c934"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=02aa2bd02f",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"4/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=02aa2bd02f\",\n            \"fix\": \"https://sqlite.org/src/info/ea20068e6d97c934\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:d42daeb535cfd659b73dcb9b9e80f7ff9353a2d0885786b22c814a4142a16bfd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d42daeb535cfd659b73dcb9b9e80f7ff9353a2d0885786b22c814a4142a16bfd"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=02aa2bd02f",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:f87b057feb8c",
+      "dbms": "sqlite",
+      "title": "UNLIKELY in query causes row to not be fetched",
+      "reported_date": "2019-11-05",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=1cda8f4b54"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=1cda8f4b54",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"5/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=1cda8f4b54\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:79443489645a8ae2fc48dea0e60ddbb8a6ea234ecf7fed7414d655e731bbea48",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:79443489645a8ae2fc48dea0e60ddbb8a6ea234ecf7fed7414d655e731bbea48"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=1cda8f4b54",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:8abc2455f533",
+      "dbms": "sqlite",
+      "title": "NULL WHERE condition unexpectedly results in row being fetched",
+      "reported_date": "2019-11-06",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=ce22a07731",
+        "fix": "https://sqlite.org/src/info/36c11ad51fe9ab1b"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=ce22a07731",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"6/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=ce22a07731\",\n            \"fix\": \"https://sqlite.org/src/info/36c11ad51fe9ab1b\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:4732684446e0be0896248ad2787ced40a212239341a17bea37200b0ff10a6b4a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4732684446e0be0896248ad2787ced40a212239341a17bea37200b0ff10a6b4a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=ce22a07731",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:0829ac3280b6",
+      "dbms": "sqlite",
+      "title": "REPLACE on table with generated NOT NULL column results in segfault",
+      "reported_date": "2019-11-06",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=2399f59861",
+        "fix": "https://sqlite.org/src/info/77b1c90add514050"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=2399f59861",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"6/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=2399f59861\",\n            \"fix\": \"https://sqlite.org/src/info/77b1c90add514050\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f2f375e1564c32bcff6430e9193c902b8a167c025c1cb698024b4c77eeffc543",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f2f375e1564c32bcff6430e9193c902b8a167c025c1cb698024b4c77eeffc543"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=2399f59861",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:1d4d5da2ca86",
+      "dbms": "sqlite",
+      "title": "Segfault when inserting into table with generated columns",
+      "reported_date": "2019-11-06",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://sqlite.org/src/tktview?name=4fc08501f4",
+        "fix": "https://www.sqlite.org/src/info/9e07b48934e9a972"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=4fc08501f4",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"6/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://sqlite.org/src/tktview?name=4fc08501f4\",\n            \"fix\": \"https://www.sqlite.org/src/info/9e07b48934e9a972\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ce9059372f600549ed1eb7ca1f3e65ac8c31d6a0dd1988e19e6b23409289db54",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ce9059372f600549ed1eb7ca1f3e65ac8c31d6a0dd1988e19e6b23409289db54"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=4fc08501f4",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:c719492cd6b2",
+      "dbms": "sqlite",
+      "title": "DISTINCT malfunctions for VIEW with virtual table",
+      "reported_date": "2019-11-07",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Severe",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=7e59041f9c",
+        "fix": "https://www.sqlite.org/src/info/b59f94e4da9b1653"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=7e59041f9c",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"7/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=7e59041f9c\",\n            \"fix\": \"https://www.sqlite.org/src/info/b59f94e4da9b1653\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:8d0aa7a4ccec4f1f3ce8452f6cf8080f9cd947737582b3b5f376d8cbc64af0c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8d0aa7a4ccec4f1f3ce8452f6cf8080f9cd947737582b3b5f376d8cbc64af0c2"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=7e59041f9c",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:23167d547a92",
+      "dbms": "sqlite",
+      "title": "UPDATE on table with two generated columns and CHECK clause results in segfault",
+      "reported_date": "2019-11-07",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=299b50ba81",
+        "fix": "https://www.sqlite.org/src/info/104a2beb57037f93"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=299b50ba81",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"7/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=299b50ba81\",\n            \"fix\": \"https://www.sqlite.org/src/info/104a2beb57037f93\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:74b1660fb10cf9c54bfdb9db0c0df8ce3299ccd02a71173a3eee8e37a0b701c5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:74b1660fb10cf9c54bfdb9db0c0df8ce3299ccd02a71173a3eee8e37a0b701c5"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=299b50ba81",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5468736b775c",
+      "dbms": "sqlite",
+      "title": "VACUUM results in \"database disk image is malformed\" for PRIMARY KEY with duplicate column",
+      "reported_date": "2019-11-07",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugtracker": "https://www.sqlite.org/src/tktview?name=302027baf1",
+        "fix": "https://www.sqlite.org/src/info/34f64f11ca481996"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=302027baf1",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"7/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugtracker\": \"https://www.sqlite.org/src/tktview?name=302027baf1\",\n            \"fix\": \"https://www.sqlite.org/src/info/34f64f11ca481996\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:81c31d9f49aa04e06614aac1fb11368a05bfef1dd61237686a0e987a23ea1135",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:81c31d9f49aa04e06614aac1fb11368a05bfef1dd61237686a0e987a23ea1135"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=302027baf1",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:9661c31be724",
+      "dbms": "sqlite",
+      "title": "LEFT JOIN malfunctions with partial ISNULL index",
+      "reported_date": "2019-11-30",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=7f39060a24",
+        "fix": "https://www.sqlite.org/src/info/4066a34da7bcdcec"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=7f39060a24",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/11/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=7f39060a24\",\n            \"fix\": \"https://www.sqlite.org/src/info/4066a34da7bcdcec\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:1190c2744dadc856fff04c9c22dd83045751c27cc7a35892de4a1a47ad2bf14e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1190c2744dadc856fff04c9c22dd83045751c27cc7a35892de4a1a47ad2bf14e"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=7f39060a24",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:dd950d738b2f",
+      "dbms": "sqlite",
+      "title": "Incorrect result for TEXT comparison on rtree table",
+      "reported_date": "2019-12-04",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=a55ab6d97d",
+        "fix": "https://www.sqlite.org/src/info/f898d04cf272ef01"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=a55ab6d97d",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=a55ab6d97d\",\n            \"fix\": \"https://www.sqlite.org/src/info/f898d04cf272ef01\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:d913e4658237df14063440442501aa905862e727e25225ea40254fd1acc1252f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d913e4658237df14063440442501aa905862e727e25225ea40254fd1acc1252f"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=a55ab6d97d",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:4b7f2dadbf47",
+      "dbms": "sqlite",
+      "title": "Join on two rtree tables malfunctions",
+      "reported_date": "2019-12-05",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=9fe487ba3c",
+        "fix": "https://www.sqlite.org/src/info/7ae8c0d52f6aa7f2"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=9fe487ba3c",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=9fe487ba3c\",\n            \"fix\": \"https://www.sqlite.org/src/info/7ae8c0d52f6aa7f2\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:09cf8b45918ef28957c830f1e550241a7e05f153c90bead9c79f31b77c1380c7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:09cf8b45918ef28957c830f1e550241a7e05f153c90bead9c79f31b77c1380c7"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=9fe487ba3c",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:20e1a36f897c",
+      "dbms": "sqlite",
+      "title": "Row value comparison malfunctions with rtree table",
+      "reported_date": "2019-12-05",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=f096d191b6",
+        "fix": "https://www.sqlite.org/src/info/b7810062ec2489e1"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=f096d191b6",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=f096d191b6\",\n            \"fix\": \"https://www.sqlite.org/src/info/b7810062ec2489e1\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:a6574e74b97974b94b62b03fc66d01c31bc9b5efb8be4e2250e5cacd861ddd3c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a6574e74b97974b94b62b03fc66d01c31bc9b5efb8be4e2250e5cacd861ddd3c"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=f096d191b6",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:a0d879335369",
+      "dbms": "sqlite",
+      "title": "column = NULL predicate evaluates to TRUE for rtree table",
+      "reported_date": "2019-12-05",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=30e2c183b6",
+        "fix": "https://www.sqlite.org/src/info/d43e0efb9642037d"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=30e2c183b6",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=30e2c183b6\",\n            \"fix\": \"https://www.sqlite.org/src/info/d43e0efb9642037d\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:547161225e2b34dab30f402db8a5b85e78a239a5f4d316f0e21d5aa0ce8ac30b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:547161225e2b34dab30f402db8a5b85e78a239a5f4d316f0e21d5aa0ce8ac30b"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=30e2c183b6",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:93c1276395e5",
+      "dbms": "sqlite",
+      "title": "CREATE VIRTUAL TABLE causes segfault",
+      "reported_date": "2019-12-06",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "fix": "https://www.sqlite.org/src/info/ef73107f475e40e6",
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/info/ef73107f475e40e6",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"06/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"fix\": \"https://www.sqlite.org/src/info/ef73107f475e40e6\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:09256070b420896c4fa0b08570c698b26d04de3b2bd46975df646b93fa6a9457",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "CREATE VIRTUAL TABLE causes segfault",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a SQLite bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:04c64af15a226967036b0c1b239d95ec8a069caf6c0d6aed91ad27c159fd31ce",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:09256070b420896c4fa0b08570c698b26d04de3b2bd46975df646b93fa6a9457"
+      },
+      "primary_url": "https://sqlite.org/src/info/ef73107f475e40e6",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:b4b24eb0b4be",
+      "dbms": "sqlite",
+      "title": "Comparison on INT column in rtree table malfunctions",
+      "reported_date": "2019-12-06",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=aa573a787b",
+        "fix": "https://www.sqlite.org/src/info/32772dfd50b602c0"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=aa573a787b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"06/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=aa573a787b\",\n            \"fix\": \"https://www.sqlite.org/src/info/32772dfd50b602c0\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:efd1dd8233914d86a11ca22dd2fdc50b187064122bb82710789a20f239fe3bf0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:efd1dd8233914d86a11ca22dd2fdc50b187064122bb82710789a20f239fe3bf0"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=aa573a787b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:451cf96488c8",
+      "dbms": "sqlite",
+      "title": "Generated column and foreign key causes a segfault",
+      "reported_date": "2019-12-06",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "fix": "https://www.sqlite.org/src/info/27c0fdab1ba4d499",
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/info/27c0fdab1ba4d499",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"06/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"fix\": \"https://www.sqlite.org/src/info/27c0fdab1ba4d499\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5eabe2c6ba7afdd6705eb8fda8fb9977e7666cd196181ed8b3c5f03019d4fb99",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Generated column and foreign key causes a segfault",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a SQLite bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:2ed1f7407c1d930f8a3a660d2b49aa32673c99b19c1d630c2375ee3a382c3b79",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5eabe2c6ba7afdd6705eb8fda8fb9977e7666cd196181ed8b3c5f03019d4fb99"
+      },
+      "primary_url": "https://sqlite.org/src/info/27c0fdab1ba4d499",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:aa93d28a36af",
+      "dbms": "sqlite",
+      "title": "Incorrect result for predicate on rtree table",
+      "reported_date": "2019-12-06",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=367a86e288",
+        "docs fix": "https://www.sqlite.org/docsrc/info/8199b59a39c017c7",
+        "fix": "https://www.sqlite.org/src/info/97fb5a72f91a44d5"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=367a86e288",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"06/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=367a86e288\",\n            \"docs fix\": \"https://www.sqlite.org/docsrc/info/8199b59a39c017c7\",\n            \"fix\": \"https://www.sqlite.org/src/info/97fb5a72f91a44d5\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:5113d1ccfcc1e6e810fc6d0f2023b9e957e2078aff7c974bfff93ed13f486ed0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5113d1ccfcc1e6e810fc6d0f2023b9e957e2078aff7c974bfff93ed13f486ed0"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=367a86e288",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:892e446bf7d0",
+      "dbms": "sqlite",
+      "title": "NOT NULL auxiliary column in rtree table malfunctions",
+      "reported_date": "2019-12-06",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Cosmetic",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=8bf76328ac",
+        "docs fix": "https://www.sqlite.org/docsrc/info/c3ab325994a8f495"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=8bf76328ac",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"06/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=8bf76328ac\",\n            \"docs fix\": \"https://www.sqlite.org/docsrc/info/c3ab325994a8f495\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:b52c9be892a0fd7a97be191072c167c5308635c4e0c56c58389d11725d207c66",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b52c9be892a0fd7a97be191072c167c5308635c4e0c56c58389d11725d207c66"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=8bf76328ac",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:cf9ba8f1b163",
+      "dbms": "sqlite",
+      "title": "Query on table without rows and generated column results in \"out of memory\" error",
+      "reported_date": "2019-12-08",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=b92e5e8ec2",
+        "fix": "https://www.sqlite.org/src/info/9d75e1ccc72e9f53"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=b92e5e8ec2",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"08/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=b92e5e8ec2\",\n            \"fix\": \"https://www.sqlite.org/src/info/9d75e1ccc72e9f53\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1c625f94e4fd84f60e30d353ad9ed0b29c02edb4bfbffeb31b1ef18ad0b2cc23",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1c625f94e4fd84f60e30d353ad9ed0b29c02edb4bfbffeb31b1ef18ad0b2cc23"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=b92e5e8ec2",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:e67ce9f773b6",
+      "dbms": "sqlite",
+      "title": "PRAGMA integrity_check does not terminate on table with generated column",
+      "reported_date": "2019-12-09",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "hang",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=bd8c280671",
+        "fix": "https://www.sqlite.org/src/info/f3b39c71b88cb672"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=bd8c280671",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"09/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=bd8c280671\",\n            \"fix\": \"https://www.sqlite.org/src/info/f3b39c71b88cb672\"\n        },\n        \"oracle\": \"hang\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f07efe766d1c5059c88cb9b5bbb403b0added09caa4474efa5b77384a281a6a0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f07efe766d1c5059c88cb9b5bbb403b0added09caa4474efa5b77384a281a6a0"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=bd8c280671",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:1813f8664b9b",
+      "dbms": "sqlite",
+      "title": "REINDEX results in \"UNIQUE constraint failed\" for generated column",
+      "reported_date": "2019-12-10",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=d7c3f125c9",
+        "fix": "https://www.sqlite.org/src/info/d47d66e3d360d8aa"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=d7c3f125c9",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=d7c3f125c9\",\n            \"fix\": \"https://www.sqlite.org/src/info/d47d66e3d360d8aa\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:fe7aab24f19a168ebb5be830bc37c78e6c3af572fd35487950cc59f702fdf581",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:fe7aab24f19a168ebb5be830bc37c78e6c3af572fd35487950cc59f702fdf581"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=d7c3f125c9",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:f8916549c330",
+      "dbms": "sqlite",
+      "title": "FTS database disk image is malformed for UTF-16 encoding after update",
+      "reported_date": "2019-12-11",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=cf36cb4e02",
+        "commit": "https://www.sqlite.org/src/info/c16305eba0eb1436"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=cf36cb4e02",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=cf36cb4e02\",\n            \"commit\": \"https://www.sqlite.org/src/info/c16305eba0eb1436\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e4a0504710677af1774e056e63ae6f5581eeab192a127135e4d528d3db25414a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e4a0504710677af1774e056e63ae6f5581eeab192a127135e4d528d3db25414a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=cf36cb4e02",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:6a7d2a329cce",
+      "dbms": "sqlite",
+      "title": "LEFT JOIN segfault on rtree table",
+      "reported_date": "2019-12-11",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "fix": "https://www.sqlite.org/src/info/840de36df1aaeb4b",
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/info/840de36df1aaeb4b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"11/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"fix\": \"https://www.sqlite.org/src/info/840de36df1aaeb4b\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:0f2ff2823a5b0af7716b6baac05d18ccd5829009dffe806c353ba298495fbb98",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "LEFT JOIN segfault on rtree table",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a SQLite bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:1e8b70895a99d846f22abd953afa9891eda5e4b65f5ba6278f395c8cc0faa070",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0f2ff2823a5b0af7716b6baac05d18ccd5829009dffe806c353ba298495fbb98"
+      },
+      "primary_url": "https://sqlite.org/src/info/840de36df1aaeb4b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:a561c5f595ad",
+      "dbms": "sqlite",
+      "title": "REINDEX segfaults on table with generated columns",
+      "reported_date": "2019-12-14",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=b439bfcfb7",
+        "fix": "https://www.sqlite.org/src/info/2401e04730a156aa"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=b439bfcfb7",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"14/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=b439bfcfb7\",\n            \"fix\": \"https://www.sqlite.org/src/info/2401e04730a156aa\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:3862b5e89388be00b9d4779016e0aaf4ff5f7f0220233037efb2cf47ad2fe3a7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3862b5e89388be00b9d4779016e0aaf4ff5f7f0220233037efb2cf47ad2fe3a7"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=b439bfcfb7",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:452ee4c6321a",
+      "dbms": "sqlite",
+      "title": "LEFT JOIN malfunctions with generated column",
+      "reported_date": "2019-12-16",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=3b84b42943",
+        "fix": "https://www.sqlite.org/src/info/0271491438ad2a98"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=3b84b42943",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"16/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=3b84b42943\",\n            \"fix\": \"https://www.sqlite.org/src/info/0271491438ad2a98\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:07e45d96accc4a750be25ee7f130eb30abb46df1e52076ed2233973a0ddc38f5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:07e45d96accc4a750be25ee7f130eb30abb46df1e52076ed2233973a0ddc38f5"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=3b84b42943",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:34c035c31f58",
+      "dbms": "sqlite",
+      "title": "UPDATE causes \"database table is locked\" for rtree table",
+      "reported_date": "2019-12-19",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=56a74875be",
+        "fix": "https://www.sqlite.org/src/info/eb95dac7f6482c36"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=56a74875be",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=56a74875be\",\n            \"fix\": \"https://www.sqlite.org/src/info/eb95dac7f6482c36\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:fbecba8a619d3ba09fad5b358866701895d0562ca99bfad29d2002734e4f4ee7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:fbecba8a619d3ba09fad5b358866701895d0562ca99bfad29d2002734e4f4ee7"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=56a74875be",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d902752860c4",
+      "dbms": "sqlite",
+      "title": "FILTER clause in window function causes a segfault",
+      "reported_date": "2019-12-20",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "fix": "https://www.sqlite.org/src/info/3cc2b5709e66ef60",
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/info/3cc2b5709e66ef60",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"fix\": \"https://www.sqlite.org/src/info/3cc2b5709e66ef60\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:3db543430267e498798b0c1fac6f95133dc78d603e11050073c7d20e3863a280",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "FILTER clause in window function causes a segfault",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a SQLite bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:80dd1bd3118a5eec64541460aee00304861d308e7e04929da75134e9f0121d1d",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3db543430267e498798b0c1fac6f95133dc78d603e11050073c7d20e3863a280"
+      },
+      "primary_url": "https://sqlite.org/src/info/3cc2b5709e66ef60",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:17155d55e644",
+      "dbms": "sqlite",
+      "title": "FTS database disk image is malformed for UTF-16 encoding",
+      "reported_date": "2019-12-20",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=e431c35542",
+        "commit": "https://sqlite.org/src/info/a1ba9a37d7a68a6d"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=e431c35542",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=e431c35542\",\n            \"commit\": \"https://sqlite.org/src/info/a1ba9a37d7a68a6d\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1340949aed283e46ee0fdb41734c48569e240c88c2caaa456c631bc09a9b6cde",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1340949aed283e46ee0fdb41734c48569e240c88c2caaa456c631bc09a9b6cde"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=e431c35542",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:2515296df302",
+      "dbms": "sqlite",
+      "title": "Incorrect result for BETWEEN and generated column",
+      "reported_date": "2019-12-20",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=e0a8120553",
+        "fix": "https://sqlite.org/src/info/728ad39e3bd07a25"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=e0a8120553",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=e0a8120553\",\n            \"fix\": \"https://sqlite.org/src/info/728ad39e3bd07a25\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:c47adac6e7825d5e235c7e180fed8b86e8161d9bd4a1a1ddad65ff1898b2e3ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c47adac6e7825d5e235c7e180fed8b86e8161d9bd4a1a1ddad65ff1898b2e3ae"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=e0a8120553",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:b27bf1f5b865",
+      "dbms": "sqlite",
+      "title": "FTS database disk image is malformed for update on languageid",
+      "reported_date": "2019-12-21",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=cd3311e323",
+        "commit": "https://sqlite.org/src/info/70815e273f511481"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=cd3311e323",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"21/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=cd3311e323\",\n            \"commit\": \"https://sqlite.org/src/info/70815e273f511481\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1568896e2e248d71030277e2ba54f040d202272afcf567ae192ca12d79652145",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1568896e2e248d71030277e2ba54f040d202272afcf567ae192ca12d79652145"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=cd3311e323",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d950d2648cef",
+      "dbms": "sqlite",
+      "title": "Debug assertion constructAutomaticIndex: Assertion `!ExprHasProperty(pExpr, EP_FromJoin) || pExpr->iRightJoinTable!=pSrc->iCursor || pLoop->prereq!=0' failed",
+      "reported_date": "2019-12-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=7929c1efb2",
+        "fix": "https://sqlite.org/src/info/ef604882a275d3d5"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=7929c1efb2",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=7929c1efb2\",\n            \"fix\": \"https://sqlite.org/src/info/ef604882a275d3d5\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:4e413b56141fa72483ceabc2b4de2264258b027984646e9c95304bbd40002598",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4e413b56141fa72483ceabc2b4de2264258b027984646e9c95304bbd40002598"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=7929c1efb2",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:da2107f58458",
+      "dbms": "sqlite",
+      "title": "Debug assertion fts5StructureRead: Assertion `p->iStructVersion!=0' failed",
+      "reported_date": "2019-12-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=9cb7d0ef44",
+        "fix": "https://sqlite.org/src/info/45748e2db028ffbd"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=9cb7d0ef44",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=9cb7d0ef44\",\n            \"fix\": \"https://sqlite.org/src/info/45748e2db028ffbd\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:d73d2ca6c0b71f3ca4d0b88f05f8af45f2dcbd8acc7fdd50cf011d53dabe56ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d73d2ca6c0b71f3ca4d0b88f05f8af45f2dcbd8acc7fdd50cf011d53dabe56ed"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=9cb7d0ef44",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:ec4e87996227",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3ExprSkipCollateAndLikely: Assertion `pExpr->op==TK_COLLATE' failed",
+      "reported_date": "2019-12-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=b0cb8aff56",
+        "fix": "https://sqlite.org/src/info/56539e1c132632c0"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=b0cb8aff56",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=b0cb8aff56\",\n            \"fix\": \"https://sqlite.org/src/info/56539e1c132632c0\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:49c870ecc4066e415d83c23ccddfde50121e2d5819c903f39bc3e150197903fb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:49c870ecc4066e415d83c23ccddfde50121e2d5819c903f39bc3e150197903fb"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=b0cb8aff56",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:cbaea0fa2a72",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeExec: Assertion `flags3==pIn3->flags' failed",
+      "reported_date": "2019-12-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=1b06916e01",
+        "fix": "https://sqlite.org/src/info/ddb17d92df194337"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=1b06916e01",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=1b06916e01\",\n            \"fix\": \"https://sqlite.org/src/info/ddb17d92df194337\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:3c246d2d1b3c1eaff6ab7cd40121c7f1b84ed30e009003724b91d225e9b70899",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3c246d2d1b3c1eaff6ab7cd40121c7f1b84ed30e009003724b91d225e9b70899"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=1b06916e01",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:f4e0fdca8e32",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeExec: Assertion `memIsValid(pRec)' failed",
+      "reported_date": "2019-12-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=f00d096cae",
+        "fix": "https://sqlite.org/src/info/8c856404b4e98d29"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=f00d096cae",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=f00d096cae\",\n            \"fix\": \"https://sqlite.org/src/info/8c856404b4e98d29\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:8b91dad6923aa39ff7dc22fcb124523afe2f7561cc81b53eea19db28c92f9d50",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8b91dad6923aa39ff7dc22fcb124523afe2f7561cc81b53eea19db28c92f9d50"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=f00d096cae",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:b1f0fc10cd04",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeExec: Assertion `pIn1!=pIn3' failed",
+      "reported_date": "2019-12-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=188f912b51",
+        "fix": "https://sqlite.org/src/info/9ab985a9c8160b90"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=188f912b51",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=188f912b51\",\n            \"fix\": \"https://sqlite.org/src/info/9ab985a9c8160b90\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:0e1e346578999104159cb175ddb1d38b7654eedafcf215ec17c91bd1f35f7a1e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0e1e346578999104159cb175ddb1d38b7654eedafcf215ec17c91bd1f35f7a1e"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=188f912b51",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:76eec4df7239",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeMemAboutToChange: Assertion `(mFlags&MEM_Str)==0 || (pMem->n==pX->n && pMem->z==pX->z)' failed",
+      "reported_date": "2019-12-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=5ad2aa6921",
+        "fix": "https://www.sqlite.org/src/info/89a9dad6330270a4"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=5ad2aa6921",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=5ad2aa6921\",\n            \"fix\": \"https://www.sqlite.org/src/info/89a9dad6330270a4\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:57c96b8509b79069db68b635c2f8deb734fd3ee0bfed198b95c69c99ec165bbe",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:57c96b8509b79069db68b635c2f8deb734fd3ee0bfed198b95c69c99ec165bbe"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=5ad2aa6921",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:a2b2bfd34766",
+      "dbms": "sqlite",
+      "title": "SELECT on window function causes a segfault",
+      "reported_date": "2019-12-22",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=d9ed4ebef1",
+        "fix": "https://www.sqlite.org/src/info/0b1dbd60f5db3abe"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=d9ed4ebef1",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=d9ed4ebef1\",\n            \"fix\": \"https://www.sqlite.org/src/info/0b1dbd60f5db3abe\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:afe20a3a8325f09f224c7a95c13457b31481f9fb360bfcdeb687dc7825b3ad1d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:afe20a3a8325f09f224c7a95c13457b31481f9fb360bfcdeb687dc7825b3ad1d"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=d9ed4ebef1",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:694c764afd19",
+      "dbms": "sqlite",
+      "title": "Debug assertion impliesNotNullRow: Assertion `pWalker->eCode==0' failed",
+      "reported_date": "2019-12-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=dfd66334cf",
+        "fix": "https://sqlite.org/src/info/2f17974912ec5e99"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=dfd66334cf",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=dfd66334cf\",\n            \"fix\": \"https://sqlite.org/src/info/2f17974912ec5e99\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:4607d4543518cbd85efba57e0f96bd9a3c1479a8a22cafda0b4ee038d2b59c0a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4607d4543518cbd85efba57e0f96bd9a3c1479a8a22cafda0b4ee038d2b59c0a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=dfd66334cf",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:7c4742860233",
+      "dbms": "sqlite",
+      "title": "Debug assertion rtreeRelease: Assertion `pRtree->nNodeRef==0 || pRtree->bCorrupt' failed",
+      "reported_date": "2019-12-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=5eadca17c4",
+        "fix": "https://sqlite.org/src/info/4c50afafce841636"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=5eadca17c4",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=5eadca17c4\",\n            \"fix\": \"https://sqlite.org/src/info/4c50afafce841636\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:dc9a4b4b9c94c04e9a526798303f95b8e7b08c658256c8a740cfadc56a58446e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:dc9a4b4b9c94c04e9a526798303f95b8e7b08c658256c8a740cfadc56a58446e"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=5eadca17c4",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:0577c4d352f7",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3MemCompare: Assertion `pMem1->enc==pMem2->enc || pMem1->db->mallocFailed' failed",
+      "reported_date": "2019-12-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=0911b5d161",
+        "fix": "https://www.sqlite.org/src/info/f347744e0d576f02"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=0911b5d161",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=0911b5d161\",\n            \"fix\": \"https://www.sqlite.org/src/info/f347744e0d576f02\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:31fe0cf817b751e77c5261a8eab27488725d71ebc279b7b8f4f8c2521d90fba7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:31fe0cf817b751e77c5261a8eab27488725d71ebc279b7b8f4f8c2521d90fba7"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=0911b5d161",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:8272c0b3a013",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeExec: Assertion `flags3==pIn3->flags' failed (2)",
+      "reported_date": "2019-12-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=9d708e4742",
+        "fix": "https://www.sqlite.org/src/info/2c44c73499154bc5"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=9d708e4742",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=9d708e4742\",\n            \"fix\": \"https://www.sqlite.org/src/info/2c44c73499154bc5\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:4938377b691d868fce63455759326dd9503b1273e7b1c779b8efa4b7a272e043",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:4938377b691d868fce63455759326dd9503b1273e7b1c779b8efa4b7a272e043"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=9d708e4742",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:85f5685f0280",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeExec: Assertion `pC!=0' failed",
+      "reported_date": "2019-12-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=3ab4a9f02c",
+        "fix": "https://www.sqlite.org/src/info/0b1dbd60"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=3ab4a9f02c",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=3ab4a9f02c\",\n            \"fix\": \"https://www.sqlite.org/src/info/0b1dbd60\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:8bd5efddc33c2ef6fc9ce452aff81654e78cc41fea0b706a7557b98588805d64",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8bd5efddc33c2ef6fc9ce452aff81654e78cc41fea0b706a7557b98588805d64"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=3ab4a9f02c",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:1946cf3ce89b",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeExec: Assertion `pIn1!=pIn3' failed (2)",
+      "reported_date": "2019-12-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=b383b90278",
+        "fix": "https://sqlite.org/src/info/401c9d30e06191d9"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=b383b90278",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=b383b90278\",\n            \"fix\": \"https://sqlite.org/src/info/401c9d30e06191d9\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:14c83cce06edc06166dee04d7782b236809f5462fc5189406cb3a5e01d620a6c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:14c83cce06edc06166dee04d7782b236809f5462fc5189406cb3a5e01d620a6c"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=b383b90278",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:001249361741",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeMemAboutToChange: Assertion `(mFlags&MEM_Str)==0 || (pMem->n==pX->n && pMem->z==pX->z)' failed (2)",
+      "reported_date": "2019-12-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=c62c5e5852",
+        "fix": "https://www.sqlite.org/src/info/36fdeb4f0a66970a"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=c62c5e5852",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=c62c5e5852\",\n            \"fix\": \"https://www.sqlite.org/src/info/36fdeb4f0a66970a\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c41270c6b5cfc484b775ae041b4508056fccc51242574603c83979581ad6d358",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c41270c6b5cfc484b775ae041b4508056fccc51242574603c83979581ad6d358"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=c62c5e5852",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:8320f9cda53a",
+      "dbms": "sqlite",
+      "title": "FTS database disk image is malformed for special characters in table",
+      "reported_date": "2019-12-23",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=d62981b76d",
+        "fix": "https://www.sqlite.org/src/info/a11b393dc2c882cf"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=d62981b76d",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=d62981b76d\",\n            \"fix\": \"https://www.sqlite.org/src/info/a11b393dc2c882cf\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7b7ed91d24788ea5589c42156b77f9f689c06976501a74a5eae7ce12fa3d5825",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7b7ed91d24788ea5589c42156b77f9f689c06976501a74a5eae7ce12fa3d5825"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=d62981b76d",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:e62d39848364",
+      "dbms": "sqlite",
+      "title": "UPDATE with complex WHERE condition on rtree results in \"database table is locked\" error",
+      "reported_date": "2019-12-23",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=cafeafe605"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=cafeafe605",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=cafeafe605\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:96f73494f8bb80da100a64ff809a8a4b5e52fdce10369606fafcac9095c902e5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:96f73494f8bb80da100a64ff809a8a4b5e52fdce10369606fafcac9095c902e5"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=cafeafe605",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:622884ec40f3",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3Fts5HashScanNext: Assertion `!sqlite3Fts5HashScanEof(p)' failed",
+      "reported_date": "2019-12-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=df46a6f38e",
+        "fix": "https://www.sqlite.org/src/info/1c0a05b09a97e6e2"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=df46a6f38e",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=df46a6f38e\",\n            \"fix\": \"https://www.sqlite.org/src/info/1c0a05b09a97e6e2\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:507dcd5777404df5118c02de37325bbd4d73e9f9472041f67d4d85428f9eae41",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:507dcd5777404df5118c02de37325bbd4d73e9f9472041f67d4d85428f9eae41"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=df46a6f38e",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:cc7dfb976d44",
+      "dbms": "sqlite",
+      "title": "FTS database disk image is malformed for UTF-16 encoding and integrity check",
+      "reported_date": "2019-12-24",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=752fdbf6e7",
+        "fix": "https://sqlite.org/src/info/bae060f382e4386c"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=752fdbf6e7",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=752fdbf6e7\",\n            \"fix\": \"https://sqlite.org/src/info/bae060f382e4386c\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5cc232eda37789bac74d7b521c842a3582b20c4a39c839c5cc9cafaf344df4fe",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5cc232eda37789bac74d7b521c842a3582b20c4a39c839c5cc9cafaf344df4fe"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=752fdbf6e7",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:e89ded3a0f20",
+      "dbms": "sqlite",
+      "title": "Debug assertion fts5CheckTransactionState: Assertion `iSavepoint<=p->ts.iSavepoint' failed",
+      "reported_date": "2019-12-25",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=167b2aac34",
+        "fix": "https://www.sqlite.org/src/info/a5d7f5d24a239f72"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=167b2aac34",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=167b2aac34\",\n            \"fix\": \"https://www.sqlite.org/src/info/a5d7f5d24a239f72\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:291af5818d563c5539cf764220fb220fee7f6eda979d66d58c9f88f734863250",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:291af5818d563c5539cf764220fb220fee7f6eda979d66d58c9f88f734863250"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=167b2aac34",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:de0505186cec",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3TableColumnAffinity: Assertion `iCol<pTab->nCol' failed.",
+      "reported_date": "2019-12-25",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=d87336c81c",
+        "fix": "https://www.sqlite.org/src/info/fa58aad48a788802"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=d87336c81c",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=d87336c81c\",\n            \"fix\": \"https://www.sqlite.org/src/info/fa58aad48a788802\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:30aa18ac33df2ddacd7191c5241cc99ad7444ca631b3a07ef89a3c4c83d37830",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:30aa18ac33df2ddacd7191c5241cc99ad7444ca631b3a07ef89a3c4c83d37830"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=d87336c81c",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5cccb13a46fc",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3BtreeInsert: Assertion `pCur->curFlags & BTCF_ValidNKey' failed",
+      "reported_date": "2019-12-26",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=ec8abb025e",
+        "fix": "https://www.sqlite.org/src/info/e54560495926fbb8"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=ec8abb025e",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=ec8abb025e\",\n            \"fix\": \"https://www.sqlite.org/src/info/e54560495926fbb8\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7cb35694580e0ade272567715840a9b0f34a4487e75d139ddcf4416a67c417ad",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7cb35694580e0ade272567715840a9b0f34a4487e75d139ddcf4416a67c417ad"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=ec8abb025e",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:c4f08655cea6",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3FinishCoding: Assertion `!pParse->isMultiWrite || sqlite3VdbeAssertMayAbort(v, pParse->mayAbort)' failed",
+      "reported_date": "2019-12-26",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=7c13db5c3b",
+        "fix": "https://www.sqlite.org/src/info/f14ce948662f3445"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=7c13db5c3b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=7c13db5c3b\",\n            \"fix\": \"https://www.sqlite.org/src/info/f14ce948662f3445\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:3319daf7fe5517c682f5532efd2950686b908141765569e3d15206bbe2fd3435",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3319daf7fe5517c682f5532efd2950686b908141765569e3d15206bbe2fd3435"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=7c13db5c3b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:7e337a218927",
+      "dbms": "sqlite",
+      "title": "Debug assertion assert_pager_state: Assertion `pPager->changeCountDone==0 || pPager->eLock>=RESERVED_LOCK' failed",
+      "reported_date": "2019-12-27",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=fb3b3024ea",
+        "fix": "https://sqlite.org/src/info/846b1de6e5a9e418"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=fb3b3024ea",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"27/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=fb3b3024ea\",\n            \"fix\": \"https://sqlite.org/src/info/846b1de6e5a9e418\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:46983470572510aab65445b26746de52ac96ee7ee6cb958f13c856f02fa7d8c4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:46983470572510aab65445b26746de52ac96ee7ee6cb958f13c856f02fa7d8c4"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=fb3b3024ea",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:84960197085a",
+      "dbms": "sqlite",
+      "title": "FTS4 integrity-check results in \"database disk image is malformed\" for UTF-16 encoding",
+      "reported_date": "2019-12-27",
+      "reported_year": 2019,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=1dc432db3b",
+        "documentation": "https://www.sqlite.org/fts3.html#_utf_16_byte_order_mark_problem_"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=1dc432db3b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"27/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=1dc432db3b\",\n            \"documentation\": \"https://www.sqlite.org/fts3.html#_utf_16_byte_order_mark_problem_\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:69df1cc8b63cd944de590464f91e153f32bb3bb157cac7c66b3e2036b89f873a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:69df1cc8b63cd944de590464f91e153f32bb3bb157cac7c66b3e2036b89f873a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=1dc432db3b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:24605c1e30b4",
+      "dbms": "sqlite",
+      "title": "Debug assertion exprSrcCount: Assertion `0' failed",
+      "reported_date": "2019-12-28",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=36ffedcb9a",
+        "fix": "https://sqlite.org/src/info/597896ed0ae9e296"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=36ffedcb9a",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"28/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=36ffedcb9a\",\n            \"fix\": \"https://sqlite.org/src/info/597896ed0ae9e296\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c8cb8149497d72c85ef3222f54eb1f90778cb930c3631deab90b2054482b77ab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c8cb8149497d72c85ef3222f54eb1f90778cb930c3631deab90b2054482b77ab"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=36ffedcb9a",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:60645ce55a09",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeExec: Assertion `memIsValid(&aMem[pOp->p1])' failed.",
+      "reported_date": "2019-12-28",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=37823501c6",
+        "fix": "https://www.sqlite.org/src/info/4cc12c18860bc480"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=37823501c6",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"28/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=37823501c6\",\n            \"fix\": \"https://www.sqlite.org/src/info/4cc12c18860bc480\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5a03ba7f25bb27da63e3dc4b9ac57b393d57c4ce75ee2464053268892d1d9c03",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5a03ba7f25bb27da63e3dc4b9ac57b393d57c4ce75ee2464053268892d1d9c03"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=37823501c6",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:33797f08f66e",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeExec: Assertion `memIsValid(pRec)' failed",
+      "reported_date": "2019-12-28",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=5fbc159eeb",
+        "fix": "https://sqlite.org/src/info/4cc12c18860bc480"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=5fbc159eeb",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"28/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=5fbc159eeb\",\n            \"fix\": \"https://sqlite.org/src/info/4cc12c18860bc480\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7d29aa7089e6c4a5cec3ab37250bff7ec90deb2ee25faca9b65a69ca53cb5418",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7d29aa7089e6c4a5cec3ab37250bff7ec90deb2ee25faca9b65a69ca53cb5418"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=5fbc159eeb",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5d35f77a11ab",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeMemAboutToChange: Assertion `(mFlags&MEM_Real)==0 || pMem->u.r==pX->u.r' failed",
+      "reported_date": "2019-12-28",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://sqlite.org/src/tktview?name=de4b04149b",
+        "fix": "https://sqlite.org/src/info/6afadd3b3a40b0ef"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/src/tktview?name=de4b04149b",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"28/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://sqlite.org/src/tktview?name=de4b04149b\",\n            \"fix\": \"https://sqlite.org/src/info/6afadd3b3a40b0ef\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:42902c70899b8b064b30b58964a3b63c8c92fe80c07de795ad2e17a0eb147794",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:42902c70899b8b064b30b58964a3b63c8c92fe80c07de795ad2e17a0eb147794"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=de4b04149b",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:87d226e2b8a0",
+      "dbms": "sqlite",
+      "title": "Trigger on normal table causes the database disk image to become malformed",
+      "reported_date": "2019-12-29",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=314cc133e5",
+        "fix": "https://www.sqlite.org/src/info/db4b7e1dc399c1f1"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=314cc133e5",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=314cc133e5\",\n            \"fix\": \"https://www.sqlite.org/src/info/db4b7e1dc399c1f1\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:64115c47c22a113e22dd63150c63f881b0a16e1b294a0c2d92084408b333ea41",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:64115c47c22a113e22dd63150c63f881b0a16e1b294a0c2d92084408b333ea41"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=314cc133e5",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:822137648ece",
+      "dbms": "sqlite",
+      "title": "Debug assertion codeVectorCompare: Assertion `0' failed",
+      "reported_date": "2019-12-30",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=892575cdba",
+        "fix": "https://www.sqlite.org/src/info/f481636f1a0333c6"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=892575cdba",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=892575cdba\",\n            \"fix\": \"https://www.sqlite.org/src/info/f481636f1a0333c6\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:3e34322156597dca81efe2d5feaad2c4c4c192887905691c68f695c1bbd1a3f9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3e34322156597dca81efe2d5feaad2c4c4c192887905691c68f695c1bbd1a3f9"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=892575cdba",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:fc126a16895b",
+      "dbms": "sqlite",
+      "title": "Inconsistent handling of subqueries in index expressions",
+      "reported_date": "2019-12-30",
+      "reported_year": 2019,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=a0e88d8d7a"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=a0e88d8d7a",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=a0e88d8d7a\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:d00241c6dd2082fbf0993eb5b41bf99ab83fc48f1aa82fed30e0c74ad44b2d29",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d00241c6dd2082fbf0993eb5b41bf99ab83fc48f1aa82fed30e0c74ad44b2d29"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=a0e88d8d7a",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d30d85ba2cbf",
+      "dbms": "sqlite",
+      "title": "NATURAL JOIN on virtual table results in \"parse error in rank function\"",
+      "reported_date": "2019-12-30",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "Critical",
+      "links": {
+        "bugreport": "https://www3.sqlite.org/src/tktview?name=7c0e06b162",
+        "fix": "https://www3.sqlite.org/src/info/ab09ef427181130b"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www3.sqlite.org/src/tktview?name=7c0e06b162",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www3.sqlite.org/src/tktview?name=7c0e06b162\",\n            \"fix\": \"https://www3.sqlite.org/src/info/ab09ef427181130b\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1cf021960701b1ed993482d084cceba6a2a2bc0c7f267f0b0ae17ce3bc34cd19",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1cf021960701b1ed993482d084cceba6a2a2bc0c7f267f0b0ae17ce3bc34cd19"
+      },
+      "primary_url": "https://www3.sqlite.org/src/tktview?name=7c0e06b162",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:66439845f6a1",
+      "dbms": "sqlite",
+      "title": "Debug assertion sqlite3VdbeMemAboutToChange: Assertion `(mFlags&MEM_Real)==0 || pMem->u.r==pX->u.r' failed (2)",
+      "reported_date": "2019-12-31",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=46fcd700b8",
+        "fix": "https://www.sqlite.org/src/info/eca7ec9cda4606c4"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=46fcd700b8",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"31/12/2019\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=46fcd700b8\",\n            \"fix\": \"https://www.sqlite.org/src/info/eca7ec9cda4606c4\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f78e434547cc4b5d309757fe8d5dc8d18d08e1c22ab3e35a82b67591dc83f3b6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f78e434547cc4b5d309757fe8d5dc8d18d08e1c22ab3e35a82b67591dc83f3b6"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=46fcd700b8",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:0a49e018da1b",
+      "dbms": "sqlite",
+      "title": "FTS5 integrity-check results in \"database disk image is malformed\" for UTF-16 encoding and SUBSTR",
+      "reported_date": "2020-01-02",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=771fe61761",
+        "fix": "https://www.sqlite.org/src/info/e782096aa06fcf41"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=771fe61761",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/01/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=771fe61761\",\n            \"fix\": \"https://www.sqlite.org/src/info/e782096aa06fcf41\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:b4e850243f306b758bf78ebc0093c3ecda49f83e45961833ae877da610a3b95f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b4e850243f306b758bf78ebc0093c3ecda49f83e45961833ae877da610a3b95f"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=771fe61761",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d48c206e2239",
+      "dbms": "sqlite",
+      "title": "DBSTAT query computes incorrect result for aggregate column",
+      "reported_date": "2020-01-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=727074e2ae",
+        "fix": "https://www.sqlite.org/src/info/74ef6f2b6d9f50ff"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=727074e2ae",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/01/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=727074e2ae\",\n            \"fix\": \"https://www.sqlite.org/src/info/74ef6f2b6d9f50ff\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:1cc2c12a3e78c09a2152ea16d0874f9b65de72fe028a25950ac77464041e0095",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1cc2c12a3e78c09a2152ea16d0874f9b65de72fe028a25950ac77464041e0095"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=727074e2ae",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:4b4c2acd27f8",
+      "dbms": "sqlite",
+      "title": "DBSTAT query computes incorrect result for name column",
+      "reported_date": "2020-01-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=74a4c3860c",
+        "fix": "https://www.sqlite.org/src/info/cfff5cb2279088aa"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=74a4c3860c",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/01/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=74a4c3860c\",\n            \"fix\": \"https://www.sqlite.org/src/info/cfff5cb2279088aa\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:d3bb519f38a5876a7e2d7def1eba8e195793127e17cc0cc07164b13f77881d6a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d3bb519f38a5876a7e2d7def1eba8e195793127e17cc0cc07164b13f77881d6a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=74a4c3860c",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5c5c6a60e104",
+      "dbms": "sqlite",
+      "title": "DBSTAT query computes incorrect result for stat.aggregate = 1 condition",
+      "reported_date": "2020-01-04",
+      "reported_year": 2020,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=a3713a5fca"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=a3713a5fca",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/01/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=a3713a5fca\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:abbe4f00d76a08b8355fbc33de8d79566f6fbc3e966b311841a4f2e2d746e47c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:abbe4f00d76a08b8355fbc33de8d79566f6fbc3e966b311841a4f2e2d746e47c"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=a3713a5fca",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:44b0b3f6a7bd",
+      "dbms": "sqlite",
+      "title": "Incorrect result for COUNT(), UTF16be encoding and SUBSTR",
+      "reported_date": "2020-01-08",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Minor",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=9eda2697f5",
+        "fix": "https://www.sqlite.org/src/info/1c76f1d8ec0937a2"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=9eda2697f5",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"08/01/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=9eda2697f5\",\n            \"fix\": \"https://www.sqlite.org/src/info/1c76f1d8ec0937a2\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:1b182560cb10e3d9a5e7d80ff619b0cccf2285f6fc420cd97b4553c08679472a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1b182560cb10e3d9a5e7d80ff619b0cccf2285f6fc420cd97b4553c08679472a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=9eda2697f5",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:6c33bafe63e6",
+      "dbms": "sqlite",
+      "title": "Incorrect result for query with 0 >= t0.c0 AND t0.c0 = v0.c0 condition",
+      "reported_date": "2020-01-08",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "Important",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=82ac75ba00",
+        "fix": "https://www.sqlite.org/src/info/6db1c3498f6bfa01"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=82ac75ba00",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"08/01/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=82ac75ba00\",\n            \"fix\": \"https://www.sqlite.org/src/info/6db1c3498f6bfa01\"\n        },\n        \"oracle\": \"NoREC\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Non-optimizing Reference Engine Construction (NoREC) oracle found this bug.",
+            "content_sha256": "sha256:19badf67abfa847cfa6a14e02636d2646164be8ab48de9c433b56861cc08a886",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:19badf67abfa847cfa6a14e02636d2646164be8ab48de9c433b56861cc08a886"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=82ac75ba00",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:df2a407de816",
+      "dbms": "sqlite",
+      "title": "MAX yields unexpected result for UTF-16",
+      "reported_date": "2020-03-04",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=1b8d726456",
+        "fix": "https://www.sqlite.org/src/info/4a5851893c3d71cc"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=1b8d726456",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"04/03/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=1b8d726456\",\n            \"fix\": \"https://www.sqlite.org/src/info/4a5851893c3d71cc\"\n        },\n        \"oracle\": \"TLP (aggregate)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:cde8ba4ca94ed01828828e1b9b83ebebd619a62b61c5b50a2637b6cc8e0c7906",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:cde8ba4ca94ed01828828e1b9b83ebebd619a62b61c5b50a2637b6cc8e0c7906"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=1b8d726456",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:6f4fdbb6a7cc",
+      "dbms": "sqlite",
+      "title": "Unexpected result for MIN on string that contains a null character",
+      "reported_date": "2020-03-05",
+      "reported_year": 2020,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=fa146dcfe6",
+        "fix": "https://www.sqlite.org/docsrc/info/e751b0c91a80dd31"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=fa146dcfe6",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"05/03/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=fa146dcfe6\",\n            \"fix\": \"https://www.sqlite.org/docsrc/info/e751b0c91a80dd31\"\n        },\n        \"oracle\": \"TLP (aggregate)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:05dff676b923b486bc9ce7f3d27c6e831b8982430b1bdab5ae6dfd767935559e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:05dff676b923b486bc9ce7f3d27c6e831b8982430b1bdab5ae6dfd767935559e"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=fa146dcfe6",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:e2debd9a20ff",
+      "dbms": "sqlite",
+      "title": "GROUP BY causes unexpected conversion",
+      "reported_date": "2020-03-10",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=e0c2ad1aa8",
+        "fix": "https://www.sqlite.org/src/info/101f7dea75a203f1"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=e0c2ad1aa8",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"10/03/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=e0c2ad1aa8\",\n            \"fix\": \"https://www.sqlite.org/src/info/101f7dea75a203f1\"\n        },\n        \"oracle\": \"TLP (aggregate)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:0759f9d2ae6f3c9f68bdb1c36d3c83abe116fb06fc7dc8c676161f6d30f22bb3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0759f9d2ae6f3c9f68bdb1c36d3c83abe116fb06fc7dc8c676161f6d30f22bb3"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=e0c2ad1aa8",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:33957f94deec",
+      "dbms": "sqlite",
+      "title": "UNION operator malfunctions in LEFT JOIN on view",
+      "reported_date": "2020-04-25",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=45f4bf4eb4",
+        "fix": "https://www.sqlite.org/src/info/ac31edd3eeafcef4"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=45f4bf4eb4",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"25/04/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=45f4bf4eb4\",\n            \"fix\": \"https://www.sqlite.org/src/info/ac31edd3eeafcef4\"\n        },\n        \"oracle\": \"TLP (DISTINCT)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:93be513f1e2f52e4a6c16bdc30315ddcd45c0df4ebb01ca7bb9fdee753ebe7ac",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:93be513f1e2f52e4a6c16bdc30315ddcd45c0df4ebb01ca7bb9fdee753ebe7ac"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=45f4bf4eb4",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5edc65d93502",
+      "dbms": "sqlite",
+      "title": "Incorrect result for IN expression with right-hand IS TRUE sub-expression",
+      "reported_date": "2020-08-24",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreports": "https://www.sqlite.org/src/tktview?name=f3ff147288",
+        "fix": "https://www.sqlite.org/src/info/4236103379df0b3d",
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=f3ff147288",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"24/08/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreports\": \"https://www.sqlite.org/src/tktview?name=f3ff147288\",\n            \"fix\": \"https://www.sqlite.org/src/info/4236103379df0b3d\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:14562c8418e0ad028e507acad7b73b44f01c3ad8b0d263c54184bcba02d0fb6b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Incorrect result for IN expression with right-hand IS TRUE sub-expression",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a SQLite bug reported by mrigger. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:b885f735a39e49ec137934b6ea1040428af8c67cec90052472c263b87aa37c22",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:14562c8418e0ad028e507acad7b73b44f01c3ad8b0d263c54184bcba02d0fb6b"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=f3ff147288",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:9e7abd4ccbcf",
+      "dbms": "sqlite",
+      "title": "Unexpected result for % and '1E1'",
+      "reported_date": "2020-08-27",
+      "reported_year": 2020,
+      "status": "fixed_in_documentation",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://www.sqlite.org/src/tktview?name=be31cf009c",
+        "fix": "https://www.sqlite.org/docsrc/info/153857859d220dbd"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/src/tktview?name=be31cf009c",
+            "source_type": "issue_tracker",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"27/08/2020\",\n        \"dbms\": \"SQLite\",\n        \"links\": {\n            \"bugreport\": \"https://www.sqlite.org/src/tktview?name=be31cf009c\",\n            \"fix\": \"https://www.sqlite.org/docsrc/info/153857859d220dbd\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:e3f0bf4dbd6c6b01762aa8a14b7db5e313b04a9ca74bc6dc673ab4085e85e86a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e3f0bf4dbd6c6b01762aa8a14b7db5e313b04a9ca74bc6dc673ab4085e85e86a"
+      },
+      "primary_url": "https://sqlite.org/src/tktview?name=be31cf009c",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:4fdf3f0c24b6",
+      "dbms": "sqlite",
+      "title": "An Inconsistent Result Depending on Parenthesization",
+      "reported_date": "2022-05-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/af3d07f908"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/af3d07f908",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/af3d07f908",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0);\nCREATE TABLE t1(c0);\nCREATE VIEW v0(c0) AS SELECT 0.4 FROM t0;\nINSERT INTO t1(c0) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:38af53534a7223b9b84510e2d8a5c7c4eb927bb9e072bbabb3350a8d8a93188e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:38af53534a7223b9b84510e2d8a5c7c4eb927bb9e072bbabb3350a8d8a93188e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:b82be49728aa",
+      "dbms": "sqlite",
+      "title": "An Unexpected NULL Column Caused by Where Clause in RIGHT JOIN",
+      "reported_date": "2022-05-13",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/41cc3851d8"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/41cc3851d8",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/41cc3851d8",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\nCREATE TABLE t1 (c0 INT);\ninsert into t0 values(2);\ninsert into t1 values(NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e93ae435b5dad4229b4ead9d2b8e7078d5e7acb6b009ae02fe21c757f6152dc2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e93ae435b5dad4229b4ead9d2b8e7078d5e7acb6b009ae02fe21c757f6152dc2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5cf10b9548fc",
+      "dbms": "sqlite",
+      "title": "Rows are Unexpectedly Filtered Out by DISTINCT in RIGHT JOIN",
+      "reported_date": "2022-05-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/c06b10ad7e"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/c06b10ad7e",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/c06b10ad7e",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t1 (c0 INTEGER UNIQUE);\nCREATE TABLE t2 (c0);\nCREATE TABLE t3 (c0);\nINSERT INTO t1 VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:1b3f93e7e7377310951c6d92a3e1cda559d01d610dfa17d4115bb5493df538ab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:1b3f93e7e7377310951c6d92a3e1cda559d01d610dfa17d4115bb5493df538ab"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:71e58a2bd76b",
+      "dbms": "sqlite",
+      "title": "Null value returned by TOTAL",
+      "reported_date": "2022-05-15",
+      "reported_year": 2022,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/c4c4fcff3d"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/c4c4fcff3d",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/c4c4fcff3d",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0);\nINSERT INTO t0 VALUES(1e900);\nINSERT INTO t0 VALUES(-1e900);\nINSERT INTO t0 VALUES(10);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:938ea50825e8c2cf0d61b8581d67afac26bad07bc8f4a7393911ad0526643c35",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:938ea50825e8c2cf0d61b8581d67afac26bad07bc8f4a7393911ad0526643c35"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:f387c94cd7f6",
+      "dbms": "sqlite",
+      "title": "Expression or Constant in GroupBy Clause",
+      "reported_date": "2022-05-17",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/2458c5dea2"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/2458c5dea2",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/2458c5dea2",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0);\nSELECT COUNT(*) FROM t0 ORDER BY (t0.c0 IN ()); \n-- Parse error: 1st ORDER BY term out of range - should be between 1 and 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9e86d4f16597fbe4c33839e71456b1f372a86820a2e88a2a047236f8f26d4bd9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9e86d4f16597fbe4c33839e71456b1f372a86820a2e88a2a047236f8f26d4bd9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:7d2580bf52e8",
+      "dbms": "sqlite",
+      "title": "Ambiguous Reference Error for Right Join",
+      "reported_date": "2022-05-23",
+      "reported_year": 2022,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/e90a8e6e6f"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/e90a8e6e6f",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/e90a8e6e6f",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t2 (c0);\nCREATE TABLE t0 (c0);\nCREATE TABLE t1 (c0);\nSELECT * FROM t2 RIGHT OUTER JOIN t1 ON t1.c0 NATURAL JOIN t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:080ef858f22f86a625a1d775f05c1f0f5076aaa8adfd405fae71496f5a582267",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:080ef858f22f86a625a1d775f05c1f0f5076aaa8adfd405fae71496f5a582267"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5f7f2f314d27",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by WHERE when Joining Tables",
+      "reported_date": "2022-05-24",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/687b0bf563"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/687b0bf563",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/687b0bf563",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0, c1);\nCREATE TABLE t1(c0);\nINSERT INTO t0 VALUES(1, 1);\nINSERT INTO t1 VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:f1f22eca31a1de0e4e824891187907e22cd743fcb8e0f9713ccbab188a093213",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:f1f22eca31a1de0e4e824891187907e22cd743fcb8e0f9713ccbab188a093213"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:c9b7bf37e343",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by WHERE/RIGHT JOIN",
+      "reported_date": "2022-05-25",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/5cfe08eed6"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/5cfe08eed6",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/5cfe08eed6",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0, c1, c2);\nCREATE TABLE t1(c0);\nINSERT INTO t1 VALUES('1');\nCREATE VIEW v0 AS SELECT 0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9a55ec8765bc86f5998b53f789590519aeec9f28fe569a90749c85b954d0723e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9a55ec8765bc86f5998b53f789590519aeec9f28fe569a90749c85b954d0723e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:ac5409e7464d",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by Joining",
+      "reported_date": "2022-05-31",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/c2554d560b"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/c2554d560b",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/c2554d560b",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0);\nCREATE TABLE t1 (c0);\nCREATE TABLE t2 (c0);\nINSERT INTO t1(c0) VALUES ('x');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:c87bddc05c6098cd787c8a411492c7a140233e1588a9028218a44753c2fb3236",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:c87bddc05c6098cd787c8a411492c7a140233e1588a9028218a44753c2fb3236"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:7083cb013801",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by RIGHT JOIN on RTree Tables",
+      "reported_date": "2022-06-01",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/087de2d9ec"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/087de2d9ec",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/087de2d9ec",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t1(c0);\nINSERT INTO rt0(c1, c2) VALUES (x'01', x'02');\n\nSELECT * FROM t1 RIGHT OUTER JOIN rt0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:c05259b4eff322d19f0feb3bb7f4448c49773a5ff7e20886e8ab94e5ffa796b4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:c05259b4eff322d19f0feb3bb7f4448c49773a5ff7e20886e8ab94e5ffa796b4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:559be64b0410",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by WHERE Again",
+      "reported_date": "2022-06-02",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/de16c4abe2"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/de16c4abe2",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/de16c4abe2",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0);\nCREATE VIEW v0(c0, c1) AS SELECT CUME_DIST() OVER (PARTITION BY t0.c0), TRUE FROM t0;\nINSERT INTO t0 VALUES ('x');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:2ca8321c43faab71a7aefad3bb0131e50c2490d4984a94b2b4cfe3157307ae1f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:2ca8321c43faab71a7aefad3bb0131e50c2490d4984a94b2b4cfe3157307ae1f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:1d6f4d5f5140",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by RIGHT JOIN",
+      "reported_date": "2022-06-06",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/206d99a16d"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/206d99a16d",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/206d99a16d",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t1(a BOOLEAN); INSERT INTO t1 VALUES (false);\nCREATE TABLE t2(x INT); INSERT INTO t2 VALUES (0);\nSELECT *, x NOTNULL, (x NOTNULL)=a FROM t2 RIGHT JOIN t1 ON true WHERE (x NOTNULL)=a;\nCREATE INDEX t1a ON t1(a);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:bdf903412c789dbeb2b5d6f290e2dbfca3a4590f03c271920f2e05793dbb29f9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:bdf903412c789dbeb2b5d6f290e2dbfca3a4590f03c271920f2e05793dbb29f9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:cb1660477f5d",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by ORDER BY",
+      "reported_date": "2022-06-07",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/323f86cc30"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/323f86cc30",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/323f86cc30",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0);\nCREATE VIEW v0(c0) AS SELECT t0.c0 FROM t0;\nCREATE TABLE t1 (c0);\nINSERT INTO t0 VALUES ('1');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3d7a15d9105ea5dc07f52e731822a77f3b7da7532d3987c0ff5583ebc76663a9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3d7a15d9105ea5dc07f52e731822a77f3b7da7532d3987c0ff5583ebc76663a9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:89029b5b5f3c",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by RIGHT JOIN with INDEX",
+      "reported_date": "2022-06-08",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/c4676c4956"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/c4676c4956",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/c4676c4956",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t1 (c0, c1);\nINSERT INTO t1(c0) VALUES (2);\nCREATE TABLE t2 (c0);\nCREATE INDEX i0 ON t2 (c0) WHERE c0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cf093f7ef3ee6c7ea120c5c26fa0278e4e43ce2e0256beddc11af3ca1874ba84",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cf093f7ef3ee6c7ea120c5c26fa0278e4e43ce2e0256beddc11af3ca1874ba84"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:8bf9d79158d8",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by JSON",
+      "reported_date": "2022-06-09",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/3d9caa45cb"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/3d9caa45cb",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/3d9caa45cb",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t1 (c0);\nCREATE VIEW v0(c0) AS SELECT json(TRUE);\nINSERT INTO t1 VALUES ('x');\nSELECT * FROM v0, t1; -- 1|x",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:92cb4d87d267c615c7adb5e26993fcce56043498689ca385d27cf8c3f298aa67",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:92cb4d87d267c615c7adb5e26993fcce56043498689ca385d27cf8c3f298aa67"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:1da8434d98e2",
+      "dbms": "sqlite",
+      "title": "Assertion `pCur->eCurType==CURTYPE_VTAB' failed",
+      "reported_date": "2022-06-10",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/dafe0500b0"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/dafe0500b0",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/dafe0500b0",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t2(c0 TEXT);\nCREATE TABLE t1(c0 INTEGER PRIMARY KEY AUTOINCREMENT, c1 INT, c2 INTEGER, c48 TEXT);\nCREATE TABLE t3(c0 REAL);\nCREATE TABLE t4(c0 TEXT);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:4b8e654711a61ad7a5007aa967d5ac0947830f3098ecd7d082f2ccf5afafa76b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:4b8e654711a61ad7a5007aa967d5ac0947830f3098ecd7d082f2ccf5afafa76b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:0561ebe9658b",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by Complicated JOINING",
+      "reported_date": "2022-06-10",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/eeb8173cf8"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/eeb8173cf8",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/eeb8173cf8",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0);\nCREATE TABLE t1(c0);\nALTER TABLE t0 ADD c1 c2;\nINSERT INTO t1 VALUES(NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:75718ad8fc43b5fc7d3ec960bdf9ba91d408cf90e16d3f11b55f15d994475f9b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:75718ad8fc43b5fc7d3ec960bdf9ba91d408cf90e16d3f11b55f15d994475f9b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:9a18a6c2f132",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by RIGHT JOIN Again",
+      "reported_date": "2022-06-10",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/51e6959f61"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/51e6959f61",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/51e6959f61",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0 INTEGER);\nINSERT INTO t0 VALUES ('x');\nCREATE TABLE t1(c0 INTEGER);\nINSERT INTO t1 VALUES ('y');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ca98152093a5760e23a42c81e3cf1a9859196da477c65339d54d478f38be8289",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ca98152093a5760e23a42c81e3cf1a9859196da477c65339d54d478f38be8289"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:cb55cdd582a1",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by Complicated JOINING Again",
+      "reported_date": "2022-06-13",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/b40696f501"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/b40696f501",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/b40696f501",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\nINSERT INTO t0 VALUES('x');\nCREATE TABLE t1(c0 INTEGER);\nINSERT INTO t1 VALUES('y');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:7a3c31f6d8d3cd77188aa39b4ad475750a921c70ec8ee9f80bf84e1ced8eee00",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:7a3c31f6d8d3cd77188aa39b4ad475750a921c70ec8ee9f80bf84e1ced8eee00"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:10982ba2bd82",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by FULL OUTER JOIN",
+      "reported_date": "2022-06-17",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/5610c17c3d"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/5610c17c3d",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/5610c17c3d",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t1 (c0 );\nCREATE TABLE t0 (c0);\nINSERT INTO t0(c0) VALUES (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:f0d9226acd1776568e6c168283ae0a040aa3a33502c5287bcff892acb4625d7b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:f0d9226acd1776568e6c168283ae0a040aa3a33502c5287bcff892acb4625d7b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:2b4040b18191",
+      "dbms": "sqlite",
+      "title": "Unexpected Expression Result by FULL OUTER JOIN",
+      "reported_date": "2022-06-20",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/6650cd40b5"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/6650cd40b5",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/6650cd40b5",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0, c1 UNIQUE);\nCREATE TABLE t1 (c0);\nCREATE TABLE t2 (c0, c1);\nCREATE TABLE t3 (c1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:7f9b26aa4a843d35427eac5d24e546dcf731ffbe7da4214c4edb49d0751b496b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:7f9b26aa4a843d35427eac5d24e546dcf731ffbe7da4214c4edb49d0751b496b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:3d38dc1bb808",
+      "dbms": "sqlite",
+      "title": "Unexpected Expression on ON clause",
+      "reported_date": "2022-06-20",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/57bdf2217d"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/57bdf2217d",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/57bdf2217d",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0);\nCREATE TABLE t1 (c0);\nINSERT INTO rt0 VALUES (0, 0, 0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:0ceacee5d333bdf51c3b84d3b08f8e99ffd8086bc8fc281a6089bb16a78933d6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:0ceacee5d333bdf51c3b84d3b08f8e99ffd8086bc8fc281a6089bb16a78933d6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5f0a405475ee",
+      "dbms": "sqlite",
+      "title": "Unexpected Parse Error",
+      "reported_date": "2022-06-21",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/1a7fea4651"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/1a7fea4651",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/1a7fea4651",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t1(a INT);\nINSERT INTO t1 VALUES(1),(2),(3);\nSELECT sum(a) FROM t1 HAVING sum(a)>0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e87c4a6c315a9873234d01e1ca402307a3bf654ae59db1a9e6a75616f998a470",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e87c4a6c315a9873234d01e1ca402307a3bf654ae59db1a9e6a75616f998a470"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:dc3d101fc7cc",
+      "dbms": "sqlite",
+      "title": "Unexpected Assertion Error in whereRangeScanEst",
+      "reported_date": "2022-06-24",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/c3496cf6b1"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/c3496cf6b1",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/c3496cf6b1",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0);\nCREATE TABLE t1 (c0);\nCREATE TABLE t2 (c0 , c1 , c2 , UNIQUE (c0), UNIQUE (c2 DESC));\nINSERT INTO t2 VALUES ('x', 'y', 'z');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:244f1bb8c4e776a4ee4a5864948e30fd48edcc381ffc5ed75806f8cafa8076db",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:244f1bb8c4e776a4ee4a5864948e30fd48edcc381ffc5ed75806f8cafa8076db"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:e45cf0f52f09",
+      "dbms": "sqlite",
+      "title": "Unexpected Assertion Error in valueFromFunction",
+      "reported_date": "2022-06-25",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/e3243e07e8"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/e3243e07e8",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/e3243e07e8",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0 PRIMARY KEY);\nINSERT INTO t0 VALUES(1);\nANALYZE;\nSELECT *FROM t0 WHERE(c0) BETWEEN(json('')) AND 0; --sqlite3: sqlite3.c:80168: valueFromFunction: Assertion `pCtx->pParse->rc==SQLITE_OK' failed.",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b3c12b1aeb2f9c01a8e1b8907b607d368f7a9703dfd6534f6a18fab733595f76",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b3c12b1aeb2f9c01a8e1b8907b607d368f7a9703dfd6534f6a18fab733595f76"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:be52a4e37d4f",
+      "dbms": "sqlite",
+      "title": "Unexpected Result by Union",
+      "reported_date": "2022-07-13",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/174afeae57"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/174afeae57",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/174afeae57",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0);\nCREATE TABLE t1 (c0, c1);\nCREATE VIEW v0(c0) AS SELECT NULL FROM t1 FULL OUTER JOIN t0 ON t1.c1 UNION ALL SELECT t1.c0 FROM t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:37cdbd780e39a99104c7a2ddaa353ac2b59f697f248d4ad2c5182bbeb078c20e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:37cdbd780e39a99104c7a2ddaa353ac2b59f697f248d4ad2c5182bbeb078c20e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:d23502a660a5",
+      "dbms": "sqlite",
+      "title": "Assertion `pRec->nField>0 && pRec->nField<=pIdx->nSampleCol' failed.",
+      "reported_date": "2022-07-15",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/3607259d3c"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/3607259d3c",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/3607259d3c",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0 INT, c1 INT PRIMARY KEY ) WITHOUT ROWID;\nINSERT INTO t0(c0, c1) VALUES (NULL, 1);\nANALYZE;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:1e577c947f566035b2a26dc08566721078909c94b06ea7169a7641ceeb653f2b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:1e577c947f566035b2a26dc08566721078909c94b06ea7169a7641ceeb653f2b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:8b3ee4f9962d",
+      "dbms": "sqlite",
+      "title": "Incorrect results with the latest trunk version",
+      "reported_date": "2023-02-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://sqlite.org/forum/info/0846211821513f50"
+      },
+      "primary_url": "https://sqlite.org/forum/info/0846211821513f50",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/info/0846211821513f50",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0);\nINSERT INTO t0 VALUES (0), (0), (0), (' '), (0), (0), (0), (0), (1);\nCREATE VIEW v0(c0) AS SELECT DISTINCT t0.c0 FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a75f62b180cc0a1688bf96c9cfa43a0a1696dc0e13e9e857aa9324a67e3df0c5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a75f62b180cc0a1688bf96c9cfa43a0a1696dc0e13e9e857aa9324a67e3df0c5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:42ba67a746da",
+      "dbms": "sqlite",
+      "title": "Incorrect result that might be caused by json function",
+      "reported_date": "2023-03-01",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://sqlite.org/forum/info/37dd14a538ea84e0"
+      },
+      "primary_url": "https://sqlite.org/forum/info/37dd14a538ea84e0",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/info/37dd14a538ea84e0",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0);\nINSERT INTO t0 VALUES ('1');\nCREATE VIEW v0(c0) AS SELECT CASE WHEN 1 THEN json_patch('1', '1') END FROM t0 GROUP BY t0.c0;\nSELECT * FROM v0 WHERE json_quote(v0.c0) != '1';",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:fd808b72db5e89d8be208b906544c69d680d7793f133a4071cec04e2e902a565",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:fd808b72db5e89d8be208b906544c69d680d7793f133a4071cec04e2e902a565"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:5d0a44d335e3",
+      "dbms": "sqlite",
+      "title": "Incorrect result for COUNT on view with window function",
+      "reported_date": "2023-03-03",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://sqlite.org/forum/info/2cd11c2d37696553"
+      },
+      "primary_url": "https://sqlite.org/forum/info/2cd11c2d37696553",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/info/2cd11c2d37696553",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0);\nINSERT INTO t0(c0) VALUES (0);\nCREATE VIEW v0(c0) AS SELECT TOTAL(0) OVER (PARTITION BY t0.c0) FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:c8c2c1d5c3e04bca6426ce0c05d62bc3812d18bb0a2dba36d43e4490518d01f0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:c8c2c1d5c3e04bca6426ce0c05d62bc3812d18bb0a2dba36d43e4490518d01f0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:f3487413aeef",
+      "dbms": "sqlite",
+      "title": "Incorrect result might caused by INDEX",
+      "reported_date": "2023-03-04",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://sqlite.org/forum/info/a68313d0545273c8"
+      },
+      "primary_url": "https://sqlite.org/forum/info/a68313d0545273c8",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/info/a68313d0545273c8",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0);\nINSERT INTO t0(c0) VALUES (1);\nCREATE INDEX i0 ON t0(c0 > 0);\nCREATE VIEW v0(c0) AS SELECT AVG(t0.c0) FROM t0 GROUP BY 1>t0.c0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:962d7f51f6417c51fc3778abebb5c16fd9cee376cf8a3de115a38001728ce1fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:962d7f51f6417c51fc3778abebb5c16fd9cee376cf8a3de115a38001728ce1fc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:866e23f870e7",
+      "dbms": "sqlite",
+      "title": "An assertion failure only triggers in JDBC driver with -DSQLITE_DEBUG=1 compile option",
+      "reported_date": "2023-03-23",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://sqlite.org/forum/info/dc4854437bda462c"
+      },
+      "primary_url": "https://sqlite.org/forum/info/dc4854437bda462c",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/info/dc4854437bda462c",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c1 TEXT );\nINSERT INTO t0 VALUES ('');\nCREATE INDEX i0 ON t0(c1);\nANALYZE;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3423ff4f31a1f53a6fc82454cb0a2caf97d4f5fb74f4c90af2eab70dbd2816bb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3423ff4f31a1f53a6fc82454cb0a2caf97d4f5fb74f4c90af2eab70dbd2816bb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:c5b261a8d525",
+      "dbms": "sqlite",
+      "title": "Unexpected result of `JOIN`",
+      "reported_date": "2023-05-01",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://sqlite.org/forum/forumpost/96cd4a7e9e"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/96cd4a7e9e",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/forumpost/96cd4a7e9e",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t1 (c0 TEXT );\n\nINSERT INTO t1(c0) VALUES (1);\nINSERT INTO vt0(c2) VALUES (-1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:827eae878ef23bf9209fca10ca2e57a9576979cf4a1084bfa1eabeaf56f7862e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:827eae878ef23bf9209fca10ca2e57a9576979cf4a1084bfa1eabeaf56f7862e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:24e4c54d5266",
+      "dbms": "sqlite",
+      "title": "Unexpected multiple JOIN result involving FULL JOIN and INNER JOIN",
+      "reported_date": "2025-05-28",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/forumpost/a8704b30f3"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/a8704b30f3",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/forumpost/a8704b30f3",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t1(c0 INT);\n\nCREATE UNIQUE INDEX IF NOT EXISTS index1 ON t1((c0 NOT BETWEEN (((c0 AND c0))) AND c0)) WHERE c0 IN ();",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5ef5715927082ed74357ac6871f22b2ce4b6f68195ff553ee1de6e1a81822a17",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5ef5715927082ed74357ac6871f22b2ce4b6f68195ff553ee1de6e1a81822a17"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:6aa51dfe0bc9",
+      "dbms": "sqlite",
+      "title": "Unexpected multiple JOIN result involving FULL JOIN, INNER JOIN and LEFT JOIN",
+      "reported_date": "2025-05-30",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/forumpost/7dee41d32506c4ae"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/7dee41d32506c4ae",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/forumpost/7dee41d32506c4ae",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0);\nCREATE TABLE t1(c0);\nCREATE TABLE t2(c0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d6f370744708b7bcfc2169f47828795d82fd3a63c7f86a36aa9cbfd2f38440a7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d6f370744708b7bcfc2169f47828795d82fd3a63c7f86a36aa9cbfd2f38440a7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:2461d3a41e27",
+      "dbms": "sqlite",
+      "title": "Unexpected multiple JOIN result involving NATURAL JOIN",
+      "reported_date": "2025-05-30",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/forumpost/4fc70203b61c7e12"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/4fc70203b61c7e12",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/forumpost/4fc70203b61c7e12",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t1 (c0 INT , c1 INT );\nCREATE TABLE t2 (c0 INT NOT NULL );\n\nINSERT INTO t1(c1) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:8a19df2f80ca6b174db05141c6862846406168ef69ca0404e07bf286047a761f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:8a19df2f80ca6b174db05141c6862846406168ef69ca0404e07bf286047a761f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:aee00e53295e",
+      "dbms": "sqlite",
+      "title": "Wrong FULL JOIN result",
+      "reported_date": "2025-05-31",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/forumpost/5028c785b6"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/5028c785b6",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/forumpost/5028c785b6",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0, c1, c2);\nCREATE TABLE t1 (c0);\nCREATE VIRTUAL TABLE t2 USING rtree(c0, c1, c2);\nCREATE TABLE v0(c0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e90a21e12ec245b2f25b31ed688ed75482a5556e27b576608b05a404b906b776",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e90a21e12ec245b2f25b31ed688ed75482a5556e27b576608b05a404b906b776"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:82b43daa9724",
+      "dbms": "sqlite",
+      "title": "Unexpected multiple JOIN involving subquery containing CAST",
+      "reported_date": "2025-06-01",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/forumpost/829306db47"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/829306db47",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/forumpost/829306db47",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0);\nCREATE TABLE t1(c0);\nCREATE TABLE t2(c0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:613fc4aea1e60f74f9b769d57296f78679e26f29f03e418ecef6efa9c7608d6a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:613fc4aea1e60f74f9b769d57296f78679e26f29f03e418ecef6efa9c7608d6a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:2fc109ba7118",
+      "dbms": "sqlite",
+      "title": "Multiple JOIN involving subquery containing TOTAL function produces wrong result",
+      "reported_date": "2025-06-09",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/forumpost/35bf50ea011cba84"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/35bf50ea011cba84",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/forumpost/35bf50ea011cba84",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\nCREATE TABLE t1 (c0 INT);\nINSERT INTO t1 VALUES (1);\nINSERT INTO t0 VALUES (0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:23fd834db2d48b76e120717cfa80ddee511992afe17aa9824b6157de6202ca9f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:23fd834db2d48b76e120717cfa80ddee511992afe17aa9824b6157de6202ca9f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:fe1e6076ac93",
+      "dbms": "sqlite",
+      "title": "Unexpected multiple JOIN result",
+      "reported_date": "2025-06-16",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/forumpost/68f29a2005"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/68f29a2005",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/forumpost/68f29a2005",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0, c1, c2);\nCREATE TABLE t1(c0);\nCREATE TABLE t2 (c0 REAL, PRIMARY KEY (c0));\nCREATE TEMP VIEW IF NOT EXISTS v0(c0) AS SELECT CAST(t1.c0 AS NUMERIC) AS col_0 FROM t1 NATURAL LEFT JOIN t2;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3f3fc9efd83b2e89a03ea5c10b84a22837896e31b4e6b8e370a779f4590ca5a7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3f3fc9efd83b2e89a03ea5c10b84a22837896e31b4e6b8e370a779f4590ca5a7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:64d6b017cb5c",
+      "dbms": "sqlite",
+      "title": "Unexpected ANTI JOIN result",
+      "reported_date": "2025-07-13",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/forumpost/3e45ed31d8"
+      },
+      "primary_url": "https://sqlite.org/forum/forumpost/3e45ed31d8",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/forumpost/3e45ed31d8",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0);\nCREATE VIRTUAL TABLE t1 USING rtree(c0, c1, c2, +c3 BLOB );\n\nINSERT INTO t0(c0) VALUES (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:80537ec93592efc8165f684dbbdff9cf05d94aa52c5070d49ac194d4c118e15e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:80537ec93592efc8165f684dbbdff9cf05d94aa52c5070d49ac194d4c118e15e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:92b2064cd418",
+      "dbms": "sqlite",
+      "title": "Unexpected execution result in the new EXISTS-to-JOIN optimization",
+      "reported_date": "2025-07-23",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/info/6e1d387c7e50a06f"
+      },
+      "primary_url": "https://sqlite.org/forum/info/6e1d387c7e50a06f",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/info/6e1d387c7e50a06f",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\nINSERT INTO t0(c0) VALUES (1);\nCREATE TABLE t1(c0 INT);\nINSERT INTO t1(c0) VALUES (2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ea894d8acdf613b0f1fde5a2c168bd2588a9240d195e8610db37422af48dc0cc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ea894d8acdf613b0f1fde5a2c168bd2588a9240d195e8610db37422af48dc0cc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:e354f6b0836b",
+      "dbms": "sqlite",
+      "title": "ON clause references tables to its right in INNER JOIN",
+      "reported_date": "2025-08-21",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/info/ccfb3b5052"
+      },
+      "primary_url": "https://sqlite.org/forum/info/ccfb3b5052",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/info/ccfb3b5052",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT);\nCREATE TABLE t1 (c0 INT);\n\nCREATE VIEW v1(c0) AS SELECT t0.c0 FROM t0 NATURAL RIGHT JOIN t1;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ba01d1197b2489f9261cb03f81ed16df8603af4c0ea163a9721f490c11ee28bf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ba01d1197b2489f9261cb03f81ed16df8603af4c0ea163a9721f490c11ee28bf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:37f85a546d40",
+      "dbms": "sqlite",
+      "title": "Unexpected execution result in LEFT JOIN with subquery",
+      "reported_date": "2025-10-29",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/info/d5f32040c5d50d2d"
+      },
+      "primary_url": "https://sqlite.org/forum/info/d5f32040c5d50d2d",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/info/d5f32040c5d50d2d",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0(c0);\nINSERT INTO t0 VALUES (1);\n\nSELECT * FROM (SELECT 0 AS col_0) LEFT JOIN (SELECT 1 AS col_1, (0 OR 2) AS col_2 FROM t0) as subQuery ON (subQuery.col_1 % subQuery.col_2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:4d2ef4207d97f40d9124c966d6bfda4d868a515b9215e45e61a016c9fa919527",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:4d2ef4207d97f40d9124c966d6bfda4d868a515b9215e45e61a016c9fa919527"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:7c6f78b68d5d",
+      "dbms": "sqlite",
+      "title": "Unexpected NATURAL FULL JOIN result",
+      "reported_date": "2025-10-30",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://www.sqlite.org/forum/info/e730f81d062f65d1"
+      },
+      "primary_url": "https://sqlite.org/forum/info/e730f81d062f65d1",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://www.sqlite.org/forum/info/e730f81d062f65d1",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0 INT);\nCREATE TABLE t1 (c0 INT);\nCREATE TABLE t2 (c0 INT);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:10069dc16131bb93a3cf0badf8d5f983c5f158ad2c0e7044c6a6937b19cb8207",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:10069dc16131bb93a3cf0badf8d5f983c5f158ad2c0e7044c6a6937b19cb8207"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:91e5e84c2c04",
+      "dbms": "sqlite",
+      "title": "Incorrect comparison optimization on collation sequences",
+      "reported_date": "2026-01-16",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Axel Teo",
+      "links": {
+        "report": "https://sqlite.org/forum/info/0b299630cbe94a2c"
+      },
+      "primary_url": "https://sqlite.org/forum/info/0b299630cbe94a2c",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/info/0b299630cbe94a2c",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c1 COLLATE NOCASE);\nCREATE INDEX i0 ON t0 (c1);\nCREATE TABLE t1 (c1);\nINSERT INTO t0 VALUES ('a');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Axel Teo of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:3cb8ea8c28bc128087df4763e0848b18b316049cc7847fca8b143dbb0cf3d787",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Axel Teo.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:3cb8ea8c28bc128087df4763e0848b18b316049cc7847fca8b143dbb0cf3d787"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:c241c1692845",
+      "dbms": "sqlite",
+      "title": "Wrong collation used as part of a row-value comparison between columns with different collation sequences.",
+      "reported_date": "2026-01-26",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Axel Teo",
+      "links": {
+        "report": "https://sqlite.org/forum/info/6ceca07fc3ef868c"
+      },
+      "primary_url": "https://sqlite.org/forum/info/6ceca07fc3ef868c",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/info/6ceca07fc3ef868c",
+            "source_type": "forum_post",
+            "excerpt": "Case 1: Without UNIQUE constraint CREATE TABLE t0 (c0 PRIMARY KEY COLLATE NOCASE, c1) WITHOUT ROWID;\nINSERT INTO t0 VALUES ('X', 'a');\n\nCREATE TABLE t1 (c0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Axel Teo of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b99cb503b217850b7914d07392400cf0a2507388b4c1c0814a96e80cf973cf0d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Axel Teo.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b99cb503b217850b7914d07392400cf0a2507388b4c1c0814a96e80cf973cf0d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:96018fbfe280",
+      "dbms": "sqlite",
+      "title": "Incorrect skip-ahead optimization after modification to internal statistics table",
+      "reported_date": "2026-03-05",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Axel Teo",
+      "links": {
+        "report": "https://sqlite.org/forum/info/5c0ce7ae01f92c99"
+      },
+      "primary_url": "https://sqlite.org/forum/info/5c0ce7ae01f92c99",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/info/5c0ce7ae01f92c99",
+            "source_type": "forum_post",
+            "excerpt": "Case 1: Without modified statistics CREATE TABLE t0 (c1 UNIQUE NOT NULL);\nINSERT INTO t0 VALUES (1);\n\nCREATE TABLE t1 (c0 UNIQUE);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Axel Teo of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:40b062b859ef71e7e7786ba3086a4944c59f61eafd199b2ef059b1a172538239",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Axel Teo.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:40b062b859ef71e7e7786ba3086a4944c59f61eafd199b2ef059b1a172538239"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:sqlite:839c00b85f4f",
+      "dbms": "sqlite",
+      "title": "Incorrect use of collation sequences on non-PK columns of a WITHOUT ROWID table for row value comparison",
+      "reported_date": "2026-03-20",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Axel Teo",
+      "links": {
+        "report": "https://sqlite.org/forum/info/7a308e933db9b702"
+      },
+      "primary_url": "https://sqlite.org/forum/info/7a308e933db9b702",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://sqlite.org/forum/info/7a308e933db9b702",
+            "source_type": "forum_post",
+            "excerpt": "CREATE TABLE t0 (c0);\nINSERT INTO t0 VALUES ('True');\n\nCREATE TABLE t1 (c0 COLLATE NOCASE, c1 PRIMARY KEY) WITHOUT ROWID;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Axel Teo of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:831b403ed62c0f10d80a771809cc01213590d6314bf2a6bbe1718fe70a9045d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Axel Teo.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:831b403ed62c0f10d80a771809cc01213590d6314bf2a6bbe1718fe70a9045d8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:starrocks:3130c716b8ad",
+      "dbms": "starrocks",
+      "title": "Sqlancer crashes in ASAN Mode",
+      "reported_date": "2021-10-16",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dirtysalt",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/712"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/712",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/712",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer crashes in ASAN Mode",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:746057c174aea999c7ba6ab17ce6da763a7e39a5953e08ef38e8ee440a8c2cfe",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/712",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:746057c174aea999c7ba6ab17ce6da763a7e39a5953e08ef38e8ee440a8c2cfe"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:8742ac08c494",
+      "dbms": "starrocks",
+      "title": "Constant-cmp-Column not supported here, should be deal earlier",
+      "reported_date": "2021-11-29",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/1689"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/1689",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/1689",
+            "source_type": "github_issue",
+            "excerpt": "--\tat sqlancer.common.query.QueryAdapter.checkException(QueryAdapter.java:98)\r\n--\tat sqlancer.common.query.QueryAdapter.executeAndGet(QueryAdapter.java:128)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4774c46a26d853d9db2d58ba8e98fe2efbe4fcfe8f2d60d3c067834bbbf596a5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/1689",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4774c46a26d853d9db2d58ba8e98fe2efbe4fcfe8f2d60d3c067834bbbf596a5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:f8b2d40a44fd",
+      "dbms": "starrocks",
+      "title": "Query return error: No equal on predicate in LEFT SEMI JOIN is not supported",
+      "reported_date": "2021-12-22",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/2350"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/2350",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/2350",
+            "source_type": "github_issue",
+            "excerpt": "--\tat sqlancer.common.query.QueryAdapter.checkException(QueryAdapter.java:98)\r\n--\tat sqlancer.common.query.QueryAdapter.executeAndGet(QueryAdapter.java:128)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:94cb5a15b008814ce3c11a0db338c66b7ef0dbe05ce4411fd421daf059dfe789",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/2350",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:94cb5a15b008814ce3c11a0db338c66b7ef0dbe05ce4411fd421daf059dfe789"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:abd9c2332c36",
+      "dbms": "starrocks",
+      "title": "BE core with SQLancer fuzzy mode",
+      "reported_date": "2022-04-08",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "kangkaisen",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/4912"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/4912",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/4912",
+            "source_type": "github_issue",
+            "excerpt": "BE core with SQLancer fuzzy mode",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4056eb507a0f0b2f979a8fe20ed3a978c9ae6a8c035365a6c3b7cddea397e8ef",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/4912",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4056eb507a0f0b2f979a8fe20ed3a978c9ae6a8c035365a6c3b7cddea397e8ef"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b9d88f978260",
+      "dbms": "starrocks",
+      "title": "[SQLancer] Result error",
+      "reported_date": "2022-04-28",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "murphyatwork",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/5657"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/5657",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/5657",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer] Result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5777c8589fb45c7b8736bc0f1be220d4153675338eed39daa34bb949c30839b7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/5657",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5777c8589fb45c7b8736bc0f1be220d4153675338eed39daa34bb949c30839b7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:19fd17e333ba",
+      "dbms": "starrocks",
+      "title": "[sqlancer] in is null Wrong result.",
+      "reported_date": "2022-05-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "yongbingwang",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/5955"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/5955",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/5955",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] in is null Wrong result.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a78b332e7e3d24cd8af39996abc8b42ee903fa7bd99c9fce1e1143efac9d82eb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/5955",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a78b332e7e3d24cd8af39996abc8b42ee903fa7bd99c9fce1e1143efac9d82eb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b75eaa62754f",
+      "dbms": "starrocks",
+      "title": "[sqlancer]Wrong result  in norec mode.",
+      "reported_date": "2022-05-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "yongbingwang",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/5940"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/5940",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/5940",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer]Wrong result  in norec mode.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:feea4944fca34a3019a0aea6fa5b4e93d5e85696c98ce02e8faaa538b014552c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/5940",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:feea4944fca34a3019a0aea6fa5b4e93d5e85696c98ce02e8faaa538b014552c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:3e1bcb621196",
+      "dbms": "starrocks",
+      "title": "[sqlancer]Wrong result in norec mode.",
+      "reported_date": "2022-05-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "yongbingwang",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/5950"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/5950",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/5950",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer]Wrong result in norec mode.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e3ea560171b12ee2ba274dbabb943993af5aee8e2b731f4c5b7c143c2604001d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/5950",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e3ea560171b12ee2ba274dbabb943993af5aee8e2b731f4c5b7c143c2604001d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:5dc916ece761",
+      "dbms": "starrocks",
+      "title": "[sqlancer]Wrong result in norec mode.",
+      "reported_date": "2022-05-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "yongbingwang",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/5949"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/5949",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/5949",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer]Wrong result in norec mode.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:150ff5476183d615e69d8b6bc74689b81515141e7b8634d2dd9b859a1db68ee4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/5949",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:150ff5476183d615e69d8b6bc74689b81515141e7b8634d2dd9b859a1db68ee4"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:5ec6b9577998",
+      "dbms": "starrocks",
+      "title": "[sqlancer]Wrong result in norec mode.",
+      "reported_date": "2022-05-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "yongbingwang",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/5941"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/5941",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/5941",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer]Wrong result in norec mode.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:aa569682061a7368115ec433de9b7d1927c2a8c9d8c87ffb7cf733d1e5c9f9ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/5941",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:aa569682061a7368115ec433de9b7d1927c2a8c9d8c87ffb7cf733d1e5c9f9ae"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:647853bbdb4b",
+      "dbms": "starrocks",
+      "title": "[sqlancer]Wrong result in norec mode.",
+      "reported_date": "2022-05-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "yongbingwang",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/5947"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/5947",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/5947",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer]Wrong result in norec mode.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a722efda1d3e32ef93fa4323c9de7ab91a496a2efc1524aee814dcf5ed19b45a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/5947",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a722efda1d3e32ef93fa4323c9de7ab91a496a2efc1524aee814dcf5ed19b45a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:fe1f1fcae801",
+      "dbms": "starrocks",
+      "title": "[sqlancer]Wrong result in norec mode.",
+      "reported_date": "2022-05-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "yongbingwang",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/5942"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/5942",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/5942",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer]Wrong result in norec mode.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e828a4ab945f3f9df6b3ebbb18d665ac1d33d84cab5a28716c98ce6a37bee9ee",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/5942",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e828a4ab945f3f9df6b3ebbb18d665ac1d33d84cab5a28716c98ce6a37bee9ee"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:60f0262a1a05",
+      "dbms": "starrocks",
+      "title": "[sqlancer]Wrong result.",
+      "reported_date": "2022-05-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "yongbingwang",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/5957"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/5957",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/5957",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer]Wrong result.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:09c365a3d1e60d284fd9581f92a31a6e8c81fed17cce6441265b09fde1318043",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/5957",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:09c365a3d1e60d284fd9581f92a31a6e8c81fed17cce6441265b09fde1318043"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a1bd23f836c8",
+      "dbms": "starrocks",
+      "title": "[sqlancer]Unknown error.",
+      "reported_date": "2022-05-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "yongbingwang",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6089"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6089",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6089",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer]Unknown error.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:94a0eee8aa81679c075d01ab44ec87e965adcddbf990eda9cc2bd9b2a40d7491",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6089",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:94a0eee8aa81679c075d01ab44ec87e965adcddbf990eda9cc2bd9b2a40d7491"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a8b905270568",
+      "dbms": "starrocks",
+      "title": "[sqlancer]Unknown error.",
+      "reported_date": "2022-05-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "yongbingwang",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6093"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6093",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6093",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer]Unknown error.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f9afa9faba76bad467f435b9e6cb140f7ef6f773b46f556cdcca23237fc6317c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6093",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f9afa9faba76bad467f435b9e6cb140f7ef6f773b46f556cdcca23237fc6317c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:7b7bd7b5747f",
+      "dbms": "starrocks",
+      "title": "[sqlancer] planner use long time 3000 ms",
+      "reported_date": "2022-05-18",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wangruin",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6262"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6262",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6262",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] planner use long time 3000 ms",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6cfee8012398e69726d350cc5ab54b8ad84474e058c42abf00d6c8e5765d1a8e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6262",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6cfee8012398e69726d350cc5ab54b8ad84474e058c42abf00d6c8e5765d1a8e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:76cb83cd9376",
+      "dbms": "starrocks",
+      "title": "[sqlancer] wrong result",
+      "reported_date": "2022-05-18",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wangruin",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6261"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6261",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6261",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] wrong result",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:12ddcc63060cd0ec68c32cbe09146e0701f0c3e4bac7906e383d77bc1c1faa00",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6261",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:12ddcc63060cd0ec68c32cbe09146e0701f0c3e4bac7906e383d77bc1c1faa00"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:0e4ebd18bd82",
+      "dbms": "starrocks",
+      "title": "[sqlancer] BE crash",
+      "reported_date": "2022-05-19",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wangruin",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6316"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6316",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6316",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] BE crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e2dc2a1ff5a9f63fed35f2c5f156609eddf2a58f5432ce3278f1497e6e76159b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6316",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e2dc2a1ff5a9f63fed35f2c5f156609eddf2a58f5432ce3278f1497e6e76159b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:451dfaf1c0a5",
+      "dbms": "starrocks",
+      "title": "[sqlancer] BE crash",
+      "reported_date": "2022-05-19",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wangruin",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6319"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6319",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6319",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] BE crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:901adf55f324837083b03766b5bf7bceb52b3c7f571a4ba54449d1761ce56875",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6319",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:901adf55f324837083b03766b5bf7bceb52b3c7f571a4ba54449d1761ce56875"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d2fd5fa23580",
+      "dbms": "starrocks",
+      "title": "[sqlancer] BE crash at implicit join on unsupported type",
+      "reported_date": "2022-05-19",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wangruin",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6291"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6291",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6291",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] BE crash at implicit join on unsupported type",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b4bfe931b3fd7f4e09b19e1189f29066e788d0425c137afdb61ec05046d80d49",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6291",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b4bfe931b3fd7f4e09b19e1189f29066e788d0425c137afdb61ec05046d80d49"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:8abe702b7664",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query sql crash",
+      "reported_date": "2022-05-23",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6444"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6444",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6444",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query sql crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:811bf4e197fe0b3f95bf8bf9efb1ed92c74bc06ca027baa142caaa32684266c5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6444",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:811bf4e197fe0b3f95bf8bf9efb1ed92c74bc06ca027baa142caaa32684266c5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:6c0b49a03b91",
+      "dbms": "starrocks",
+      "title": "[sqlancer] delete out of partitions data error",
+      "reported_date": "2022-05-24",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6461"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6461",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6461",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] delete out of partitions data error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:48bb28c37314d514504337256a06c9e3ab533b782420442905d624bae8b06032",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6461",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:48bb28c37314d514504337256a06c9e3ab533b782420442905d624bae8b06032"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:ba72ad530702",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query result error",
+      "reported_date": "2022-05-24",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6460"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6460",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6460",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c6f1ade9bd7bc695588b272b40d1d16d33939ba35624321cfe34b78fbe5be57e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6460",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c6f1ade9bd7bc695588b272b40d1d16d33939ba35624321cfe34b78fbe5be57e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:c1538777b848",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query result error",
+      "reported_date": "2022-05-24",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6474"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6474",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6474",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:31d12aa9096e8aa8146fb25724c4f658132158195cb7f1bbec716e7a431aec63",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6474",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:31d12aa9096e8aa8146fb25724c4f658132158195cb7f1bbec716e7a431aec63"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:0f168f69312b",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query result error with having tlp",
+      "reported_date": "2022-05-24",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6475"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6475",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6475",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query result error with having tlp",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6549548fa3d4505f855c69a011ef75461a6c0861328b23818b51e83461a948b1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6475",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6549548fa3d4505f855c69a011ef75461a6c0861328b23818b51e83461a948b1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:8a9448568ca6",
+      "dbms": "starrocks",
+      "title": "[sqlancer] unknow error",
+      "reported_date": "2022-05-24",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6473"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6473",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6473",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:98c2868189b5c68aa6362c5ccdd810b3f4d324323677eaaaea958e86812cc237",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6473",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:98c2868189b5c68aa6362c5ccdd810b3f4d324323677eaaaea958e86812cc237"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b18f724f3224",
+      "dbms": "starrocks",
+      "title": "[sqlancer] unknow error",
+      "reported_date": "2022-05-24",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6463"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6463",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6463",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f909da6ed6509bd4e801047cc83df8aa1ef4278aba56b866ad721e7b4e88d180",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6463",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f909da6ed6509bd4e801047cc83df8aa1ef4278aba56b866ad721e7b4e88d180"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:545ce25797b3",
+      "dbms": "starrocks",
+      "title": "[sqlancer] JoinReorderGreedy  unknow error",
+      "reported_date": "2022-05-25",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6515"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6515",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6515",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] JoinReorderGreedy  unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:da6615964e021d39daa05cfa588e74a96d3a0a6dc2fcb87371bb9ef9c2f8f86d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6515",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:da6615964e021d39daa05cfa588e74a96d3a0a6dc2fcb87371bb9ef9c2f8f86d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:2b7162522b15",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query result error",
+      "reported_date": "2022-05-25",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6512"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6512",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6512",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e447911db000c5f0e463684929677aa60a9518aaf4cb5de2297caaa6c600cdc7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6512",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e447911db000c5f0e463684929677aa60a9518aaf4cb5de2297caaa6c600cdc7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:7da3c77f0eee",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Cross join RF BE crash",
+      "reported_date": "2022-05-26",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6556"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6556",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6556",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Cross join RF BE crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a5fb9d51b92d374a7659a9f93d72a8dfb5c2cb80a31952440614edc0829989a4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6556",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a5fb9d51b92d374a7659a9f93d72a8dfb5c2cb80a31952440614edc0829989a4"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:9c829021d6f2",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query result error",
+      "reported_date": "2022-05-26",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6552"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6552",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6552",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:64f0039b740f08aeb9f2e04d762c59a8d860e3d4079bef13f036956b2b31d978",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6552",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:64f0039b740f08aeb9f2e04d762c59a8d860e3d4079bef13f036956b2b31d978"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a7710db6a11e",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query result error in norec mode",
+      "reported_date": "2022-05-26",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6553"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6553",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6553",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query result error in norec mode",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d6872c20af0e9d0152d1b8ceb718e996fa501c77fdd47458aa6a8accab554005",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6553",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d6872c20af0e9d0152d1b8ceb718e996fa501c77fdd47458aa6a8accab554005"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:782a862b7afe",
+      "dbms": "starrocks",
+      "title": "[sqlancer] create view failed",
+      "reported_date": "2022-05-27",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6616"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6616",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6616",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] create view failed",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:89e22d10450abd2271aa4c34dfd1504e61b188ebdc1d95148ec5e005a8253029",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6616",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:89e22d10450abd2271aa4c34dfd1504e61b188ebdc1d95148ec5e005a8253029"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a45292574b49",
+      "dbms": "starrocks",
+      "title": "[sqlancer] improve the error message when using json type in sql having clause",
+      "reported_date": "2022-05-27",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6617"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6617",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6617",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] improve the error message when using json type in sql having clause",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:77e8a1ce02648d4223172a89aa41617663c5aca69fabf20d319e7ffd1b58a1ab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6617",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:77e8a1ce02648d4223172a89aa41617663c5aca69fabf20d319e7ffd1b58a1ab"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:54d1aa6bf127",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query crash",
+      "reported_date": "2022-05-27",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6615"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6615",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6615",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:040087d14a2aa541936344e5d3160e07fd42955dd0a40fca5d997d16b9f44e4b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6615",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:040087d14a2aa541936344e5d3160e07fd42955dd0a40fca5d997d16b9f44e4b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d6bd91829ec4",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query crash in asan mode",
+      "reported_date": "2022-05-28",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6663"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6663",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6663",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query crash in asan mode",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b83ce9334eefaa691d7dc66bf1c1677828359debb5c648cdb28723491c82bfbd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6663",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b83ce9334eefaa691d7dc66bf1c1677828359debb5c648cdb28723491c82bfbd"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:c71d69dfdc8e",
+      "dbms": "starrocks",
+      "title": "result error from sqlancer query",
+      "reported_date": "2022-05-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "tiannan-sr",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6704"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6704",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6704",
+            "source_type": "github_issue",
+            "excerpt": "result error from sqlancer query",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9e63af82cdde79e51884782c70870e9a165664ab95a34151862f7083e2c7fc57",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6704",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9e63af82cdde79e51884782c70870e9a165664ab95a34151862f7083e2c7fc57"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:7bfad11ff6c8",
+      "dbms": "starrocks",
+      "title": "result error from sqlancer querys",
+      "reported_date": "2022-05-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "tiannan-sr",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6700"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6700",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6700",
+            "source_type": "github_issue",
+            "excerpt": "result error from sqlancer querys",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8c10852b45bbb30fcfbc08460bcd016e3066f651523a33371586dd5928dce6e8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6700",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8c10852b45bbb30fcfbc08460bcd016e3066f651523a33371586dd5928dce6e8"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:7e79143aa2d2",
+      "dbms": "starrocks",
+      "title": "result error from sqlancer querys",
+      "reported_date": "2022-05-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "tiannan-sr",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6703"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6703",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6703",
+            "source_type": "github_issue",
+            "excerpt": "result error from sqlancer querys",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:40ae56ab308f633c0aadd47dbe2f29761099477aaf772421cc34854ec66fa0f1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6703",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:40ae56ab308f633c0aadd47dbe2f29761099477aaf772421cc34854ec66fa0f1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:2dc7afedd3d1",
+      "dbms": "starrocks",
+      "title": "BE crash when sqlancer querys",
+      "reported_date": "2022-06-01",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "tiannan-sr",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6805"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6805",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6805",
+            "source_type": "github_issue",
+            "excerpt": "BE crash when sqlancer querys",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f5900d863db4f01ad969a01d6da1e3ee29bf72de518ac8873f9cd4858ba180f6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6805",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f5900d863db4f01ad969a01d6da1e3ee29bf72de518ac8873f9cd4858ba180f6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:cdadf6a48bbf",
+      "dbms": "starrocks",
+      "title": "error message - Function 'last_value(`t1_86`.`c_1_2`)' not supported with OVER clause",
+      "reported_date": "2022-06-02",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6863"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6863",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6863",
+            "source_type": "github_issue",
+            "excerpt": "--\tat sqlancer.common.query.QueryAdapter.checkException(QueryAdapter.java:98)\r\n--\tat sqlancer.common.query.QueryAdapter.executeAndGet(QueryAdapter.java:128)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2e1748a0c11a275a3b9ca19e44f8620ca67d9c545b2cb0ce38a719b05c94fae6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6863",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2e1748a0c11a275a3b9ca19e44f8620ca67d9c545b2cb0ce38a719b05c94fae6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:1e723b587489",
+      "dbms": "starrocks",
+      "title": "result error from sqlancer query",
+      "reported_date": "2022-06-08",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "colorfulu",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/6999"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/6999",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/6999",
+            "source_type": "github_issue",
+            "excerpt": "result error from sqlancer query",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7f93324efc956f73f89910e916680ab81ed61b17b45541e651fe93bf907ab37e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/6999",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7f93324efc956f73f89910e916680ab81ed61b17b45541e651fe93bf907ab37e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:515ea4fe07a5",
+      "dbms": "starrocks",
+      "title": "result error from sqlancer query",
+      "reported_date": "2022-06-08",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "colorfulu",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/7017"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/7017",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/7017",
+            "source_type": "github_issue",
+            "excerpt": "result error from sqlancer query",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d806fe706abcda7dc770a2a9efca18031af46c1cdd1ba28c14c069046d6d8e63",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/7017",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d806fe706abcda7dc770a2a9efca18031af46c1cdd1ba28c14c069046d6d8e63"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:4a2049cb091a",
+      "dbms": "starrocks",
+      "title": "sqlancer query result error",
+      "reported_date": "2022-06-08",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "colorfulu",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/7020"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/7020",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/7020",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer query result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1df93c142db29d3c92b15363e65a0c92c20784f6712461b4da286803c0e01ffd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/7020",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1df93c142db29d3c92b15363e65a0c92c20784f6712461b4da286803c0e01ffd"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:e1efa83f1983",
+      "dbms": "starrocks",
+      "title": "sqlancer query result error",
+      "reported_date": "2022-06-08",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "colorfulu",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/7019"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/7019",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/7019",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer query result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:45456bcbd9e7c604d838d11f5bd215aabbd51e35bf2ffe39697f198ec70968ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/7019",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:45456bcbd9e7c604d838d11f5bd215aabbd51e35bf2ffe39697f198ec70968ed"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b0c356e75ba9",
+      "dbms": "starrocks",
+      "title": "sqlancer query result error",
+      "reported_date": "2022-06-09",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "colorfulu",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/7042"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/7042",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/7042",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer query result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f10f1fbc17375c382483963cecb33299be451b36002feac22df94f7696bffb7e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/7042",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f10f1fbc17375c382483963cecb33299be451b36002feac22df94f7696bffb7e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:c8411223c734",
+      "dbms": "starrocks",
+      "title": "sqlancer query result error",
+      "reported_date": "2022-06-09",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "colorfulu",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/7039"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/7039",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/7039",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer query result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c43d584b921c7274e492263c1efb02aae8575e9a6130aad3738ab22c3ab75954",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/7039",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c43d584b921c7274e492263c1efb02aae8575e9a6130aad3738ab22c3ab75954"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d6bd930180eb",
+      "dbms": "starrocks",
+      "title": "sqlancer query result error",
+      "reported_date": "2022-06-09",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "colorfulu",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/7048"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/7048",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/7048",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer query result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5477092988dd807e297a1ff7400924860f8b591c3fad65aec238b81f1c80d03d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/7048",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5477092988dd807e297a1ff7400924860f8b591c3fad65aec238b81f1c80d03d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:f898fb61d4e2",
+      "dbms": "starrocks",
+      "title": "sqlancer query result error",
+      "reported_date": "2022-06-09",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "colorfulu",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/7043"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/7043",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/7043",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer query result error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:646f307fdb75625a451952531f386ed9729ad86aa04090ce97d59fa24c5a21ac",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/7043",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:646f307fdb75625a451952531f386ed9729ad86aa04090ce97d59fa24c5a21ac"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:4a535e00470f",
+      "dbms": "starrocks",
+      "title": "[sqlancer] empty json path",
+      "reported_date": "2022-06-17",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/7422"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/7422",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/7422",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] empty json path",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:cbe6fe8d78dfc8dade05fd74c05c1e2a84780fb262b7483e1212cd32159aaf40",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/7422",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cbe6fe8d78dfc8dade05fd74c05c1e2a84780fb262b7483e1212cd32159aaf40"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:9564ddb0efc9",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be crash",
+      "reported_date": "2022-07-12",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/8588"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/8588",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/8588",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9b9840e9e635692f78900b928dc8d3d45a4803c223945920f574885e775db0e9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/8588",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9b9840e9e635692f78900b928dc8d3d45a4803c223945920f574885e775db0e9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:1f11c2ab57df",
+      "dbms": "starrocks",
+      "title": "[sqlancer] create view error",
+      "reported_date": "2022-07-12",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/8590"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/8590",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/8590",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] create view error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f0b19a8d8e6e44758b646e64a57ad42ca493f03a4127cd4491ba1516c4da9554",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/8590",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f0b19a8d8e6e44758b646e64a57ad42ca493f03a4127cd4491ba1516c4da9554"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:0d262f33c0ae",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query unknow error",
+      "reported_date": "2022-07-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9351"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9351",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9351",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a00cbd8c0985143cdb3e67b4977f4507e615b5e9b4b5b5a2da9679c69f33c2c6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9351",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a00cbd8c0985143cdb3e67b4977f4507e615b5e9b4b5b5a2da9679c69f33c2c6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:6a7c5cd6c4ad",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query unknow error",
+      "reported_date": "2022-07-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9347"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9347",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9347",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ffef54d81cafad6dd967a3c1665e66e076e3bcc0ab1fce15955bb780d912ae2b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9347",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ffef54d81cafad6dd967a3c1665e66e076e3bcc0ab1fce15955bb780d912ae2b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a387fcecc944",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query unknow error",
+      "reported_date": "2022-07-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9349"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9349",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9349",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ba0be71dce7c18ac156357a3bcdff7c642b67a68ac43f1c740497d87cf67b119",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9349",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ba0be71dce7c18ac156357a3bcdff7c642b67a68ac43f1c740497d87cf67b119"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:06b093d36165",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query sql crash",
+      "reported_date": "2022-08-01",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9419"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9419",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9419",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query sql crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c923c7fee79ac5dc758ec75599ca1017603c8e9dc317079bff87484b09e82048",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9419",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c923c7fee79ac5dc758ec75599ca1017603c8e9dc317079bff87484b09e82048"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a1d377f36e78",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query sql crash",
+      "reported_date": "2022-08-01",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9421"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9421",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9421",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query sql crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8fc9a706753346c6be3ef5f115aeabd147517e2d79c3b384bd916145a44e7801",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9421",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8fc9a706753346c6be3ef5f115aeabd147517e2d79c3b384bd916145a44e7801"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:0b80f9134bf0",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query cast error",
+      "reported_date": "2022-08-05",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9618"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9618",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9618",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query cast error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4abbf2aaf61a32fa1bfd80decba7b332e837a5af3de44dc0c8e2f6700a178590",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9618",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4abbf2aaf61a32fa1bfd80decba7b332e837a5af3de44dc0c8e2f6700a178590"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b578c8174feb",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query return unknow error",
+      "reported_date": "2022-08-05",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9616"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9616",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9616",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query return unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:fb330d4aa7a233aecc861d79ddd0fcb9e05d4cd087214af7d9e1a7e6c375d8a5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9616",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fb330d4aa7a233aecc861d79ddd0fcb9e05d4cd087214af7d9e1a7e6c375d8a5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b72a0e6d93ab",
+      "dbms": "starrocks",
+      "title": "unknown error from sqlancer",
+      "reported_date": "2022-08-06",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "tiannan-sr",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9665"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9665",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9665",
+            "source_type": "github_issue",
+            "excerpt": "unknown error from sqlancer",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0c095fe1efcbf3ba505f3ec24aac3eba1cdb484c3a7491a02b584a8f30821d4e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9665",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0c095fe1efcbf3ba505f3ec24aac3eba1cdb484c3a7491a02b584a8f30821d4e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:9fbf7bea35d9",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be crash",
+      "reported_date": "2022-08-08",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9721"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9721",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9721",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:649aa69fd8445f6565712ebfea09f3bc67b43fb1ab3d24ee83f3b9645ecbd87e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9721",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:649aa69fd8445f6565712ebfea09f3bc67b43fb1ab3d24ee83f3b9645ecbd87e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:7f07f4421422",
+      "dbms": "starrocks",
+      "title": "[sqlancer] client exit caused by query",
+      "reported_date": "2022-08-08",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9722"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9722",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9722",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] client exit caused by query",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ed114a48bead6c6f8214314fc7e4fea054b584486f82c8bc47807dac759c3d15",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9722",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ed114a48bead6c6f8214314fc7e4fea054b584486f82c8bc47807dac759c3d15"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:512972bdd3b8",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be crash",
+      "reported_date": "2022-08-12",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9947"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9947",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9947",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e5ee115b9e0d92d912c0615917e97817af37df3b47d0d6b6708966208d3484d1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9947",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e5ee115b9e0d92d912c0615917e97817af37df3b47d0d6b6708966208d3484d1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:80d0e46297c7",
+      "dbms": "starrocks",
+      "title": "[sqlancer] execute sql error",
+      "reported_date": "2022-08-12",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9943"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9943",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9943",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] execute sql error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4552fc349f5f1fc35e1cac2101b936a21aa9ab6d4736d8287bde07c21ee1a52d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9943",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4552fc349f5f1fc35e1cac2101b936a21aa9ab6d4736d8287bde07c21ee1a52d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:8e4033965991",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query crash",
+      "reported_date": "2022-08-12",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9929"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9929",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9929",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8ae53971184795c3f470a46db54c060b6c68257eac80c4371de746db4f4d5f1c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9929",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8ae53971184795c3f470a46db54c060b6c68257eac80c4371de746db4f4d5f1c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:bfc2cb5f882c",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query error which contains UNKNOWN_TYPE",
+      "reported_date": "2022-08-12",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9941"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9941",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9941",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query error which contains UNKNOWN_TYPE",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:dfffd533ba455ebd9bc0aa182b4269a1aa0e2d3b98cf0bf35efe5acb4be62411",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9941",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:dfffd533ba455ebd9bc0aa182b4269a1aa0e2d3b98cf0bf35efe5acb4be62411"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:8860d18815e1",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query unknow error",
+      "reported_date": "2022-08-12",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9933"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9933",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9933",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a85d8bce6fe6828b014e7c4752ff7142673089d2e57bd4d9ccd998f5ca45840e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9933",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a85d8bce6fe6828b014e7c4752ff7142673089d2e57bd4d9ccd998f5ca45840e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a8081415e75d",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query unknow error",
+      "reported_date": "2022-08-12",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/9934"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/9934",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/9934",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f154df494dd23064d31091f1e29dea83335d111b99ac4740fd717341aca8f791",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/9934",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f154df494dd23064d31091f1e29dea83335d111b99ac4740fd717341aca8f791"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:4c90514a6e72",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query crash after set enable_filter_unused_columns_in_scan_stage = true",
+      "reported_date": "2022-08-19",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/10205"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/10205",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/10205",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query crash after set enable_filter_unused_columns_in_scan_stage = true",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b3351d0dc134c307032af1c9ae7709d5ebe400578c49947af4d861e6d84dea74",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/10205",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b3351d0dc134c307032af1c9ae7709d5ebe400578c49947af4d861e6d84dea74"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:8d59893c019e",
+      "dbms": "starrocks",
+      "title": "[sqlancer] unknow error",
+      "reported_date": "2022-08-19",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/10208"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/10208",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/10208",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:db8ac02cb18cac4adb179998ec1d236006a14f693aa46524a980b2ce039f30a2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/10208",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:db8ac02cb18cac4adb179998ec1d236006a14f693aa46524a980b2ce039f30a2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:2ddf69ebf9fe",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query error",
+      "reported_date": "2022-08-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/10525"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/10525",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/10525",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c77f1190c3a2b5bb43d1f40046c6062b63501ba41865e03991d16da66aa535bc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/10525",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c77f1190c3a2b5bb43d1f40046c6062b63501ba41865e03991d16da66aa535bc"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:aca2529a6bc5",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query return no plan sql",
+      "reported_date": "2022-08-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/10524"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/10524",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/10524",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query return no plan sql",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1734117479cbf00b63e21965e6117eab01ca8e8a1a32747de06cb2fbc5466df5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/10524",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1734117479cbf00b63e21965e6117eab01ca8e8a1a32747de06cb2fbc5466df5"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:bbff53232c9e",
+      "dbms": "starrocks",
+      "title": "sqlancer query crash",
+      "reported_date": "2022-08-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/10529"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/10529",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/10529",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer query crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:338c7f4da026c43b699f0fc6d78152865451051d65c7b0ae99ea0d1bbc5b9ca7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/10529",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:338c7f4da026c43b699f0fc6d78152865451051d65c7b0ae99ea0d1bbc5b9ca7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:8215d368424f",
+      "dbms": "starrocks",
+      "title": "BE crash in sqlancer test env",
+      "reported_date": "2022-09-03",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "tiannan-sr",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/10814"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/10814",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/10814",
+            "source_type": "github_issue",
+            "excerpt": "BE crash in sqlancer test env",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3ca06e4d9e79c5ad979467e9295b07d6038b9f11dd87defd38bec89f17d97785",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/10814",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3ca06e4d9e79c5ad979467e9295b07d6038b9f11dd87defd38bec89f17d97785"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:02515592a23d",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query crash",
+      "reported_date": "2022-09-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/11895"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/11895",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/11895",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c37a4285d2f7a2df50eb310a61fb3af6daeeb21ea7030711127563345d7c1427",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/11895",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c37a4285d2f7a2df50eb310a61fb3af6daeeb21ea7030711127563345d7c1427"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:cbd4857e5d9b",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query crash in asan mode",
+      "reported_date": "2022-09-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/11911"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/11911",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/11911",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query crash in asan mode",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a3f5649f010d14525cf26996faeee5539d097e2f808744033c6d76adab25565e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/11911",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a3f5649f010d14525cf26996faeee5539d097e2f808744033c6d76adab25565e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:091a8baa94fb",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query unknown error",
+      "reported_date": "2022-09-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/11910"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/11910",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/11910",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:83f01c042660b35dbda7b443b2e8068f5ab851fb9b0f8579fe224f55565b8e72",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/11910",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:83f01c042660b35dbda7b443b2e8068f5ab851fb9b0f8579fe224f55565b8e72"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b00fa973d211",
+      "dbms": "starrocks",
+      "title": "[sqlancer] result sets mismatch",
+      "reported_date": "2022-10-09",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/11969"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/11969",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/11969",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] result sets mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a2322fdea65cb66fe0f28c3589c369fafb09314ff6db1d2f808bbd9243af2fdb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/11969",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a2322fdea65cb66fe0f28c3589c369fafb09314ff6db1d2f808bbd9243af2fdb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:bae2e9fbf198",
+      "dbms": "starrocks",
+      "title": "[sqlancer] unkown error for query",
+      "reported_date": "2022-10-09",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/11978"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/11978",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/11978",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] unkown error for query",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6f8897bc33c2bdb8f0794a922bfa5f5295ba04b26323cd805318d100bcfacf1a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/11978",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6f8897bc33c2bdb8f0794a922bfa5f5295ba04b26323cd805318d100bcfacf1a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:cb6850b1689d",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Invalid regex expression",
+      "reported_date": "2022-10-10",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/12010"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/12010",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/12010",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Invalid regex expression",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3ba3c2dd118bfea73417bda3a1a81ec0af463b3dcb9c6fd6c5d50ed3980587d2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/12010",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3ba3c2dd118bfea73417bda3a1a81ec0af463b3dcb9c6fd6c5d50ed3980587d2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b12de18c6377",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query unkown error",
+      "reported_date": "2022-11-03",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/12901"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/12901",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/12901",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query unkown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a8ef436c236bc8b7bf0adf056bef46699c06527f7fe230dbe855465ec7b00330",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/12901",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a8ef436c236bc8b7bf0adf056bef46699c06527f7fe230dbe855465ec7b00330"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:9aa119c497f3",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error for NullPointerException",
+      "reported_date": "2022-11-04",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/12956"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/12956",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/12956",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error for NullPointerException",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d6d313e669c7914bf1152a4a116a6710bd578335b629a56558697dc1fd4c7016",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/12956",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d6d313e669c7914bf1152a4a116a6710bd578335b629a56558697dc1fd4c7016"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:6da0f134a17b",
+      "dbms": "starrocks",
+      "title": "[sqlancer] no executable plan for this sql",
+      "reported_date": "2022-11-04",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/12951"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/12951",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/12951",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] no executable plan for this sql",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2c6d5a772e696ac726485324689afa9312e48c09bd50cc362103976025a2016a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/12951",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2c6d5a772e696ac726485324689afa9312e48c09bd50cc362103976025a2016a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:5d6f7599c16e",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of left anti join, the size of the result sets mismatch",
+      "reported_date": "2022-11-04",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/12949"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/12949",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/12949",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of left anti join, the size of the result sets mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:51677ac536690b5a923c0292799cecdd3107832d94480c00635054b9694f508b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/12949",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:51677ac536690b5a923c0292799cecdd3107832d94480c00635054b9694f508b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:cf0a5a373260",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2022-11-08",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13084"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13084",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13084",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:076837bf17435b969065de9db2882f3cf3ef47f8b2b1095793341cd7498a6ccd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13084",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:076837bf17435b969065de9db2882f3cf3ef47f8b2b1095793341cd7498a6ccd"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:2bdb3cd9f4fd",
+      "dbms": "starrocks",
+      "title": "[sqlancer] crash when executing query",
+      "reported_date": "2022-11-10",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13205"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13205",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13205",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] crash when executing query",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bcec41c25637cdb69550aa7226f06bd7ce70cbc64cb1d56df813b615c9986906",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13205",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bcec41c25637cdb69550aa7226f06bd7ce70cbc64cb1d56df813b615c9986906"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:722db04f4e06",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Expected LE 1 to be returned by expression",
+      "reported_date": "2022-11-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13250"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13250",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13250",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Expected LE 1 to be returned by expression",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1ace5bcb4bb7ceeb8fb01518cb14824082aba0aceffe45a6538303df736ddac1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13250",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1ace5bcb4bb7ceeb8fb01518cb14824082aba0aceffe45a6538303df736ddac1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:ac3172300144",
+      "dbms": "starrocks",
+      "title": "[sqlancer] NullPointerException - no executable plan for this sql",
+      "reported_date": "2022-11-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13248"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13248",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13248",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] NullPointerException - no executable plan for this sql",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ec23f15983c99db4f7d0b71708b3b59b10aa7901b751f594bb6c1e3a2e227286",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13248",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ec23f15983c99db4f7d0b71708b3b59b10aa7901b751f594bb6c1e3a2e227286"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:df8b564d8f3c",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be crash",
+      "reported_date": "2022-11-11",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13260"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13260",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13260",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3e3bad60810fa6f9c745d80d9496291c871e2c9693cd1edb3909a7aae33edc65",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13260",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3e3bad60810fa6f9c745d80d9496291c871e2c9693cd1edb3909a7aae33edc65"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:9129c6670516",
+      "dbms": "starrocks",
+      "title": "[sqlancer] the size of the result sets mismatch",
+      "reported_date": "2022-11-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13328"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13328",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13328",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] the size of the result sets mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e6c7843f967e5d3b4c2ade97108df9552773f64b606c05137ef71538829c13c6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13328",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e6c7843f967e5d3b4c2ade97108df9552773f64b606c05137ef71538829c13c6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:caa781c56b2b",
+      "dbms": "starrocks",
+      "title": "[sqlancer] result set correctness",
+      "reported_date": "2022-11-17",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13550"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13550",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13550",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] result set correctness",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4d34b4417b7ef465ce48d5ede40636d198a0f71dbb2b68fa7623a00dd26ad454",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13550",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4d34b4417b7ef465ce48d5ede40636d198a0f71dbb2b68fa7623a00dd26ad454"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:afc3d5314a97",
+      "dbms": "starrocks",
+      "title": "[sqlancer] asan crash",
+      "reported_date": "2022-11-18",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13649"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13649",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13649",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] asan crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1601fd721fbcc923a46c74cb4d08440854fd52ba62c04869d266ca19fc70e376",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13649",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1601fd721fbcc923a46c74cb4d08440854fd52ba62c04869d266ca19fc70e376"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a98ab5d345c1",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be crash",
+      "reported_date": "2022-11-18",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13643"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13643",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13643",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:82d108a857c7d532f123e8f690fdc300fb3fa315134159331f40a4ea5af646dc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13643",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:82d108a857c7d532f123e8f690fdc300fb3fa315134159331f40a4ea5af646dc"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:290e3a95c6e6",
+      "dbms": "starrocks",
+      "title": "[sqlancer] unknown error of NPE",
+      "reported_date": "2022-11-21",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13733"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13733",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13733",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] unknown error of NPE",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:36d4c7a1bcdeb45f9c3152e7218cd96385e0d19af7b441438e660f1bbe6644f4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13733",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:36d4c7a1bcdeb45f9c3152e7218cd96385e0d19af7b441438e660f1bbe6644f4"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:962cdf315b9e",
+      "dbms": "starrocks",
+      "title": "[sqlancer] NPE of physical plan translator",
+      "reported_date": "2022-11-22",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13798"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13798",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13798",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] NPE of physical plan translator",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:163ada43172c4c86a99be9ab0b827905e7b8db4ccb500b90c55bbe1450ece2b9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13798",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:163ada43172c4c86a99be9ab0b827905e7b8db4ccb500b90c55bbe1450ece2b9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:f6fb79fbea0d",
+      "dbms": "starrocks",
+      "title": "[sqlancer] crash",
+      "reported_date": "2022-11-23",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13902"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13902",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13902",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c5d6251dd1d550ff45ff802ce250fe9c3634e6dc513d62bfdbf9cd5af710db1c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13902",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c5d6251dd1d550ff45ff802ce250fe9c3634e6dc513d62bfdbf9cd5af710db1c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:69082eb41666",
+      "dbms": "starrocks",
+      "title": "[sqlancer] result set mismatch",
+      "reported_date": "2022-11-23",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/13892"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/13892",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/13892",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:438941a1aea553449aa77dc9f7aa4a38563395113e94237a386d87918212e437",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/13892",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:438941a1aea553449aa77dc9f7aa4a38563395113e94237a386d87918212e437"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:fa8872e5fc5d",
+      "dbms": "starrocks",
+      "title": "[sqlancer] result set mismatch",
+      "reported_date": "2022-11-25",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/14100"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/14100",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/14100",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:00581c09bbea1bc80aba7a48f5917409dcfbff3107bdb19f8edf6d7f75d3877a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/14100",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:00581c09bbea1bc80aba7a48f5917409dcfbff3107bdb19f8edf6d7f75d3877a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:c616b211d42d",
+      "dbms": "starrocks",
+      "title": "[sqlancer] crash of _M_range_insert",
+      "reported_date": "2022-11-29",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/14239"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/14239",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/14239",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] crash of _M_range_insert",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:16b1660a0945dc179c8a4e1f5ef399f2b2bcd927aa74a54ef50a5221ad3ee625",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/14239",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:16b1660a0945dc179c8a4e1f5ef399f2b2bcd927aa74a54ef50a5221ad3ee625"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:bba8eae371a8",
+      "dbms": "starrocks",
+      "title": "[sqlancer] crash in NestloopJoin",
+      "reported_date": "2022-11-30",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/14330"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/14330",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/14330",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] crash in NestloopJoin",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:11e75205e2cb335da5bc55b621857ba209f04791f533f605b28cca0da59e06cf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/14330",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:11e75205e2cb335da5bc55b621857ba209f04791f533f605b28cca0da59e06cf"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:35104b6268f1",
+      "dbms": "starrocks",
+      "title": "[sqlancer] crash when start be",
+      "reported_date": "2022-12-05",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/14602"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/14602",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/14602",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] crash when start be",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:fcc697021abf4ff147b975eeb337f7b6415c1d66934c1e954f20307f39bff02b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/14602",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fcc697021abf4ff147b975eeb337f7b6415c1d66934c1e954f20307f39bff02b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d4951ca0bc5f",
+      "dbms": "starrocks",
+      "title": "[sqlancer] unknown error of npe when creating materialized view",
+      "reported_date": "2022-12-05",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/14588"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/14588",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/14588",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] unknown error of npe when creating materialized view",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d3dab5081859e56891803c351db6ac24928b1f8e03a28170c19eae34b18828da",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/14588",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d3dab5081859e56891803c351db6ac24928b1f8e03a28170c19eae34b18828da"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:132305c93095",
+      "dbms": "starrocks",
+      "title": "[sqlancer] coredump when compacting",
+      "reported_date": "2022-12-06",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/14707"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/14707",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/14707",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] coredump when compacting",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6289dc6dc34c4cd33e613399934a8d93963c52d62bdba6f636f7e38166a273b0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/14707",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6289dc6dc34c4cd33e613399934a8d93963c52d62bdba6f636f7e38166a273b0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:eaa7a2c5b4c8",
+      "dbms": "starrocks",
+      "title": "[sqlancer] crash when compacting data",
+      "reported_date": "2022-12-07",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "dulong41",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/14755"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/14755",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/14755",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] crash when compacting data",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:910c33b264f31961d2e119b89cf35040650243c9e1ac243713457ad12d5e4950",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/14755",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:910c33b264f31961d2e119b89cf35040650243c9e1ac243713457ad12d5e4950"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:4213ccf00d7d",
+      "dbms": "starrocks",
+      "title": "[sqlancer] delete table with mv error, Column is not key column in index",
+      "reported_date": "2022-12-13",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15111"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15111",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15111",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] delete table with mv error, Column is not key column in index",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:daa5113617928234d50fbeb013926a94b6b311c8742a2038bf6936a348d27b7d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15111",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:daa5113617928234d50fbeb013926a94b6b311c8742a2038bf6936a348d27b7d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:ae4965e3f597",
+      "dbms": "starrocks",
+      "title": "[sqlancer] NullPointerException: no executable plan for this sql",
+      "reported_date": "2022-12-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15185"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15185",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15185",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] NullPointerException: no executable plan for this sql",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:32c4a88a97ce7e355b39cabbd969a0f234042137ea4ce5207f846955730c80f0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15185",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:32c4a88a97ce7e355b39cabbd969a0f234042137ea4ce5207f846955730c80f0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:4a28f358e340",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error of \"java.lang.IllegalStateException: null\"",
+      "reported_date": "2022-12-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15183"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15183",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15183",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error of \"java.lang.IllegalStateException: null\"",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1e69dfda11be696090154c0a762429a133060910ec153fa3a85b5250ee47e84f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15183",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1e69dfda11be696090154c0a762429a133060910ec153fa3a85b5250ee47e84f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:452e4555f644",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be core",
+      "reported_date": "2022-12-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15186"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15186",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15186",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be core",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d6457e4785c281854bedcc35c85490fedffe651e482af6f25f4bcb2126917127",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15186",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d6457e4785c281854bedcc35c85490fedffe651e482af6f25f4bcb2126917127"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:61a758946b61",
+      "dbms": "starrocks",
+      "title": "[sqlancer] heap-use-after-free",
+      "reported_date": "2022-12-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15184"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15184",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15184",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] heap-use-after-free",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:62be5db9532563495ec4742acde2b4aad207d89adcf63a054e8ac862c4bbd647",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15184",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:62be5db9532563495ec4742acde2b4aad207d89adcf63a054e8ac862c4bbd647"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:c7d91c2e0d66",
+      "dbms": "starrocks",
+      "title": "[sqlancer] start be crash",
+      "reported_date": "2022-12-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15202"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15202",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15202",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] start be crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d5852b1465052494b4d15271dc282638926150782f79356cc9594b32c61aacab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15202",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d5852b1465052494b4d15271dc282638926150782f79356cc9594b32c61aacab"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:0818db8c6c43",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be crash in asan mode",
+      "reported_date": "2022-12-15",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15255"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15255",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15255",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be crash in asan mode",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9d149660b9b01b0cac24339f3aeb9f57003caa28fd3574254aedb4e0580d5b6f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15255",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9d149660b9b01b0cac24339f3aeb9f57003caa28fd3574254aedb4e0580d5b6f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b13d45c18a32",
+      "dbms": "starrocks",
+      "title": "[sqlancer] execute sql error",
+      "reported_date": "2022-12-15",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15241"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15241",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15241",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] execute sql error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:aff54429cada77231c2a9e7bf1974814e8e3cffa06d65a52738000cfff1e67fe",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15241",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:aff54429cada77231c2a9e7bf1974814e8e3cffa06d65a52738000cfff1e67fe"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:fc161f61a043",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query error",
+      "reported_date": "2022-12-15",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15237"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15237",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15237",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1faac751eed6438057a2f34d71df122d9466ffcf48854efe7e942f6701cbec3e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15237",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1faac751eed6438057a2f34d71df122d9466ffcf48854efe7e942f6701cbec3e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:5f237326f79d",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query result  error",
+      "reported_date": "2022-12-15",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15260"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15260",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15260",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query result  error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:34bb40e2d3568aa2983ea1bde0ba4d8637ba4d0866b63239095ed0c9e7aae189",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15260",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:34bb40e2d3568aa2983ea1bde0ba4d8637ba4d0866b63239095ed0c9e7aae189"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:5d1ebdfe0c35",
+      "dbms": "starrocks",
+      "title": "[sqlancer] unknow error",
+      "reported_date": "2022-12-15",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15243"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15243",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15243",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7eb50282ccce5492ef4047b0834dddb6eb5aa2dead2158b2d63b2b2081aaaa5d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15243",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7eb50282ccce5492ef4047b0834dddb6eb5aa2dead2158b2d63b2b2081aaaa5d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:e34bbe16e5c6",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be memleak",
+      "reported_date": "2022-12-20",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15473"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15473",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15473",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be memleak",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:246831f81b91e018c9282f706fe37e7574c0edb62224ddb267f3df6874e70c96",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15473",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:246831f81b91e018c9282f706fe37e7574c0edb62224ddb267f3df6874e70c96"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:1fef40bbe42e",
+      "dbms": "starrocks",
+      "title": "[sqlancer] unknow error",
+      "reported_date": "2022-12-20",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15475"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15475",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15475",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] unknow error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:36c58086e2a56a5553322418ed75a009283943e8b9d94b3e5cb27fa9a705cacc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15475",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:36c58086e2a56a5553322418ed75a009283943e8b9d94b3e5cb27fa9a705cacc"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:5fe9bde91057",
+      "dbms": "starrocks",
+      "title": "[sqlancer] memory leak after be graceful exit",
+      "reported_date": "2022-12-22",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15629"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15629",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15629",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] memory leak after be graceful exit",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5ee02f9dd0379c4dace6fe5143b32a37f9099fb45dbaa42f11b28da4ccc5730e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15629",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5ee02f9dd0379c4dace6fe5143b32a37f9099fb45dbaa42f11b28da4ccc5730e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a190ea829113",
+      "dbms": "starrocks",
+      "title": "[sqlancer] unknown error",
+      "reported_date": "2022-12-22",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15631"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15631",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15631",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5d9b62e8ad1550b19f509f85102cbffc9d6f1a2fa8151b5bd1127ac92a0f6d92",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15631",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5d9b62e8ad1550b19f509f85102cbffc9d6f1a2fa8151b5bd1127ac92a0f6d92"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:cbdff8ed7087",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be has overloaded with pipeline drivers",
+      "reported_date": "2022-12-28",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15881"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15881",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15881",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be has overloaded with pipeline drivers",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7b1216694340757625644191a6625e97ee2c709cc4e18b33fa9e852c5a956b19",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15881",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7b1216694340757625644191a6625e97ee2c709cc4e18b33fa9e852c5a956b19"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:88c1a0184167",
+      "dbms": "starrocks",
+      "title": "[sqlancer] distinct query crash",
+      "reported_date": "2022-12-28",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15878"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15878",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15878",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] distinct query crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:69be432eb73a9073a88a18a633fd595b2a717e86d92e5a177654e9891de2d658",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15878",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:69be432eb73a9073a88a18a633fd595b2a717e86d92e5a177654e9891de2d658"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:e3389e01b26a",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query sql error",
+      "reported_date": "2022-12-28",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanpengfei-git",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/15884"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/15884",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/15884",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query sql error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c87f561e309c3c68baf86235abcf29c708a59beb00641e6f66cb9c76a6318d5a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/15884",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c87f561e309c3c68baf86235abcf29c708a59beb00641e6f66cb9c76a6318d5a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a230014bce54",
+      "dbms": "starrocks",
+      "title": "[sqlancer] NullPointerException: no executable plan for this sql",
+      "reported_date": "2023-01-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16910"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16910",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16910",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] NullPointerException: no executable plan for this sql",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:02543c226361ccbdb014c218e6428ca340554fa01246025d0e3a39c313fda8d6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16910",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:02543c226361ccbdb014c218e6428ca340554fa01246025d0e3a39c313fda8d6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:0ca1c3a6e0ce",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-01-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16911"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16911",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16911",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:61e4cb3833faccf67418abfee97c00dfd40b5f7600081854db55dd6fa61b863d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16911",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:61e4cb3833faccf67418abfee97c00dfd40b5f7600081854db55dd6fa61b863d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:24e493a9a049",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-01-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16917"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16917",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16917",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:61063df20e1d60247bc8e2173c6ce48ffa5983bc331dd96be4b56d07d422aceb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16917",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:61063df20e1d60247bc8e2173c6ce48ffa5983bc331dd96be4b56d07d422aceb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:96defe1623b8",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-01-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16912"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16912",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16912",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:287ab11c968be700fd9c9a855338bb84162247cea29757dab0bef38af02d6b9a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16912",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:287ab11c968be700fd9c9a855338bb84162247cea29757dab0bef38af02d6b9a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:e4aabd2b5528",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-01-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16908"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16908",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16908",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3604c22cc78d6b11b7f4de705e2b1803890933241fdf538a99f280433298268c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16908",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3604c22cc78d6b11b7f4de705e2b1803890933241fdf538a99f280433298268c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:9f87ee9284f9",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be crash",
+      "reported_date": "2023-01-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16915"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16915",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16915",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2e43a15e0b19799bae2e3786be575b4efd4a7c6c70e0bc1c1aefdc539f8cd471",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16915",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2e43a15e0b19799bae2e3786be575b4efd4a7c6c70e0bc1c1aefdc539f8cd471"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:1e2b7af14a1b",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-01-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16903"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16903",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16903",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:905a2d9870e6629db6f38f59c6cc12aa664ac30f2d693af82b9ba734a9d52548",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16903",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:905a2d9870e6629db6f38f59c6cc12aa664ac30f2d693af82b9ba734a9d52548"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:38c233b6302d",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-01-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16926"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16926",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16926",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:89425fa07cb8a30e43319e7a6ce31b551beca944b7808967fe7355f3fb6b17d9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16926",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:89425fa07cb8a30e43319e7a6ce31b551beca944b7808967fe7355f3fb6b17d9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:0fe5d063dacd",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-01-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16981"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16981",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16981",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:926dfeb2dc405630df157171ac6457a7d864bda3b1a5a9c87fe04ff756e2c1ca",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16981",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:926dfeb2dc405630df157171ac6457a7d864bda3b1a5a9c87fe04ff756e2c1ca"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:52935f77fa4f",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-01-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16984"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16984",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16984",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:cf3669681e51a50a017b488039d295b506e612333579ae8a17fde7dfd85c88c6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16984",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cf3669681e51a50a017b488039d295b506e612333579ae8a17fde7dfd85c88c6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d8560053546f",
+      "dbms": "starrocks",
+      "title": "[sqlancer] be crash in asan mode",
+      "reported_date": "2023-01-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16963"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16963",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16963",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] be crash in asan mode",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f6a1e046d7407cf8ff1677bd18cae211f49d54ee9e9976ff572ddfbe2ea9687a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16963",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f6a1e046d7407cf8ff1677bd18cae211f49d54ee9e9976ff572ddfbe2ea9687a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:510fff06deac",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-01-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/16975"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/16975",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/16975",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:91e4a0c1f38cc9f37cef87075463aabab56c5479f596cb21801b9b486c7aecb8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/16975",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:91e4a0c1f38cc9f37cef87075463aabab56c5479f596cb21801b9b486c7aecb8"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:20a426f3e940",
+      "dbms": "starrocks",
+      "title": "[sqlancer] asan be crash",
+      "reported_date": "2023-01-30",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17035"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17035",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17035",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] asan be crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3c88392dbf6317704fe864c61a0c87b8f70d24f0fd6ee59c3c61e4bed9aa2f44",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17035",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3c88392dbf6317704fe864c61a0c87b8f70d24f0fd6ee59c3c61e4bed9aa2f44"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:084bb788c71f",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-01-30",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17047"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17047",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17047",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ff9ff1f602aa52ecb0425024a06c245b76b1e93207072ca87710b03e813c3d0f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17047",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ff9ff1f602aa52ecb0425024a06c245b76b1e93207072ca87710b03e813c3d0f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b27531641546",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-01-30",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17040"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17040",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17040",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1142edef278983524f100f7a14e1b68032a4ccedebce40e055cebe289716a797",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17040",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1142edef278983524f100f7a14e1b68032a4ccedebce40e055cebe289716a797"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:ae599cd98152",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-01-31",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17085"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17085",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17085",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4d155bf88e5a7da743174788a98d79d6e9d959187fb78eb14a24ef4d33db0417",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17085",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4d155bf88e5a7da743174788a98d79d6e9d959187fb78eb14a24ef4d33db0417"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d6fa5f062fc8",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-02-01",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17173"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17173",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17173",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e134522c03de17db9581a423fc283288438d23cfbce439806b1e0c333cd9775d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17173",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e134522c03de17db9581a423fc283288438d23cfbce439806b1e0c333cd9775d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:0bd9f174309d",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-02-01",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17177"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17177",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17177",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f213bce08384efe0d4259b661456ff8148e93280ec58c70996b64fd7d1f818b7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17177",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f213bce08384efe0d4259b661456ff8148e93280ec58c70996b64fd7d1f818b7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d15146b3e758",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-02-01",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17176"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17176",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17176",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e1900abec1576110b695f487ed990a9d9d7377418d9f2e02428141a01f7b88a0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17176",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e1900abec1576110b695f487ed990a9d9d7377418d9f2e02428141a01f7b88a0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:e6baf9a94fbd",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-02-01",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17175"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17175",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17175",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1225bff2a50ee48b7b05dc6044e870716b66abf35a92fa293c302b461b33cc8c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17175",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1225bff2a50ee48b7b05dc6044e870716b66abf35a92fa293c302b461b33cc8c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:ab14a6d7d777",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch in asan mode",
+      "reported_date": "2023-02-01",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17179"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17179",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17179",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch in asan mode",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7171bf94eba1f698b20ddc67a343f3bebe580d73b6dcf625772988e5ef6bb6fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17179",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7171bf94eba1f698b20ddc67a343f3bebe580d73b6dcf625772988e5ef6bb6fc"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:33c5b4f5cfbf",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-02-02",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17206"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17206",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17206",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d076316110c67843fec9477b0572118a20d56f2c220f6f037d8c527cc17f1102",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17206",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d076316110c67843fec9477b0572118a20d56f2c220f6f037d8c527cc17f1102"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:702194d60895",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-02-02",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17207"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17207",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17207",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a75883ff212c4223a3d12a67803c7fc8fc879c7ec4c1b743d62a05a283b33ce7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17207",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a75883ff212c4223a3d12a67803c7fc8fc879c7ec4c1b743d62a05a283b33ce7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d368d62edbd0",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-02-02",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17204"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17204",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17204",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2f8e405d22b5dd9e8809f25b1f74d8c882a981bff62cef18c9c4bc19b2e0023d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17204",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2f8e405d22b5dd9e8809f25b1f74d8c882a981bff62cef18c9c4bc19b2e0023d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:1b690c331a0e",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-02-03",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17286"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17286",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17286",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:25147d28917b3838be472062e58c4d2418d78bbf78680aab03387109a35b8f80",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17286",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:25147d28917b3838be472062e58c4d2418d78bbf78680aab03387109a35b8f80"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:f6f60055b405",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-02-06",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17384"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17384",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17384",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:52dbe8898b6a614021d9a48348de50f301aa31b5c90285f0511bfac2cb82e02e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17384",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:52dbe8898b6a614021d9a48348de50f301aa31b5c90285f0511bfac2cb82e02e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a596f777a8ae",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-02-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17595"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17595",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17595",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d80230013451c79c7e033e025a10047e57466ec90d07d1adc9ec32a01a398589",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17595",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d80230013451c79c7e033e025a10047e57466ec90d07d1adc9ec32a01a398589"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:c9ce42d19aa0",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unknown error",
+      "reported_date": "2023-02-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17585"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17585",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17585",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unknown error",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8d1a33d49d41b2b63fa54cb56bea6b6a182ecfe4c59f55733cf2ee7d94d94789",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17585",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8d1a33d49d41b2b63fa54cb56bea6b6a182ecfe4c59f55733cf2ee7d94d94789"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:38ac5a37bd73",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-02-13",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17744"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17744",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17744",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ce17a92f1760ae91b54625ea476f5b0cabe4ae5d4928f6de3edc2fd24fc5e313",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17744",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ce17a92f1760ae91b54625ea476f5b0cabe4ae5d4928f6de3edc2fd24fc5e313"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d173dcb88275",
+      "dbms": "starrocks",
+      "title": "[sqlancer] query of result set mismatch",
+      "reported_date": "2023-02-17",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "andyziye",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/17972"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/17972",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/17972",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] query of result set mismatch",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ed462bdc6d7aad2b1a339c1694572f9234e8700322098593f9be7a07feae312e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/17972",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ed462bdc6d7aad2b1a339c1694572f9234e8700322098593f9be7a07feae312e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:0784a8c1cf14",
+      "dbms": "starrocks",
+      "title": "[StarOS] result mismatch from sqlancer querys",
+      "reported_date": "2023-02-21",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "tiannan-sr",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/18187"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/18187",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/18187",
+            "source_type": "github_issue",
+            "excerpt": "[StarOS] result mismatch from sqlancer querys",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:abe7962c1de816457ce71e74189cae0d17bd55f860ef680b2367859585719589",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/18187",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:abe7962c1de816457ce71e74189cae0d17bd55f860ef680b2367859585719589"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b0ce730446aa",
+      "dbms": "starrocks",
+      "title": "Sqlancer error \"Cancelled\"",
+      "reported_date": "2023-04-20",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22004"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22004",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22004",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer error \"Cancelled\"",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2d1b49cc222b4060c3e9763095494899679b29a3372589073e734af3321f55bd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22004",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2d1b49cc222b4060c3e9763095494899679b29a3372589073e734af3321f55bd"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:ccff53ef30f0",
+      "dbms": "starrocks",
+      "title": "Sqlancer error \"Query exceeded time limit of 1800 seconds\"",
+      "reported_date": "2023-04-20",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22003"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22003",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22003",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer error \"Query exceeded time limit of 1800 seconds\"",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:95ed710b8dedb0472c95e0d405b46ca82c6baf52c091e9eb5b84e39a0c91f6e3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22003",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:95ed710b8dedb0472c95e0d405b46ca82c6baf52c091e9eb5b84e39a0c91f6e3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:006e6ff201ba",
+      "dbms": "starrocks",
+      "title": "Sqlancer error \"Unsupported nest window function inside aggregation\"",
+      "reported_date": "2023-04-20",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/21976"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/21976",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/21976",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer error \"Unsupported nest window function inside aggregation\"",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:430ab6c57e4fd242cbcacf232c45d9ff206eafd5de60f2be4282f07a4e45303b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/21976",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:430ab6c57e4fd242cbcacf232c45d9ff206eafd5de60f2be4282f07a4e45303b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d5e332549f68",
+      "dbms": "starrocks",
+      "title": "Sqlancer error \"Unsupported nest window function inside aggregation.\"",
+      "reported_date": "2023-04-20",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/21973"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/21973",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/21973",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer error \"Unsupported nest window function inside aggregation.\"",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ee62038c09449e2ac0a15989db950d87c4049f0d308f0e367fcd34b992e01816",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/21973",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ee62038c09449e2ac0a15989db950d87c4049f0d308f0e367fcd34b992e01816"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:25ffb23e46cf",
+      "dbms": "starrocks",
+      "title": "Sqlaner Error \"NULL_TYPE is illegal in thrift stage\"",
+      "reported_date": "2023-04-20",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/21972"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/21972",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/21972",
+            "source_type": "github_issue",
+            "excerpt": "--\tat sqlancer.common.query.QueryAdapter.checkException(QueryAdapter.java:98)\r\n--\tat sqlancer.common.query.QueryAdapter.executeAndGet(QueryAdapter.java:128)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d19ae361994677e565d9c17b5ceaba600e03af4ed320181274f28e136743931e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/21972",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d19ae361994677e565d9c17b5ceaba600e03af4ed320181274f28e136743931e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b240ea0abe36",
+      "dbms": "starrocks",
+      "title": "Sqlaner error \"In Predicate only support literal expression list\"",
+      "reported_date": "2023-04-20",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/21975"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/21975",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/21975",
+            "source_type": "github_issue",
+            "excerpt": "--\tat sqlancer.common.query.QueryAdapter.checkException(QueryAdapter.java:98)\r\n--\tat sqlancer.common.query.QueryAdapter.executeAndGet(QueryAdapter.java:128)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:48abf2d2ccc8a443f3ac3a6c0bb327d9a8ed490ffbc8bf0bce411d0542de1b2f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/21975",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:48abf2d2ccc8a443f3ac3a6c0bb327d9a8ed490ffbc8bf0bce411d0542de1b2f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:567245d6bc83",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: Getting syntax error at line 1, column 193. Detail message: Input 'AS' is not valid at this position.",
+      "reported_date": "2023-04-21",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22194"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22194",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22194",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: Getting syntax error at line 1, column 193. Detail message: Input 'AS' is not valid at this position.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8e881a0fbe497aa303e611f92086478183daebe8fc9e88d08e9cc69e4ab88839",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22194",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8e881a0fbe497aa303e611f92086478183daebe8fc9e88d08e9cc69e4ab88839"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:d0d14ba87386",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: Getting syntax error at line 1, column 193. Detail message: Input 'AS' is not valid at this position.",
+      "reported_date": "2023-04-21",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22187"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22187",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22187",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: Getting syntax error at line 1, column 193. Detail message: Input 'AS' is not valid at this position.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8e881a0fbe497aa303e611f92086478183daebe8fc9e88d08e9cc69e4ab88839",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22187",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8e881a0fbe497aa303e611f92086478183daebe8fc9e88d08e9cc69e4ab88839"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:dc5f2949d5d8",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: Getting analyzing error from line 1, column 377 to line 1, column 393. Detail message: Column type ARRAY<int(11)> does not support binary predicate operation.",
+      "reported_date": "2023-04-24",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22340"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22340",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22340",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: Getting analyzing error from line 1, column 377 to line 1, column 393. Detail message: Column type ARRAY<int(11)> does not support binary predicate operation.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f3425f84f4b8f3a95c0a7134935a53faa58ad69a10f6a73fe035ce0b04e4da27",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22340",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f3425f84f4b8f3a95c0a7134935a53faa58ad69a10f6a73fe035ce0b04e4da27"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:0901392da6c7",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: Getting analyzing error from line 1, column 591 to line 1, column 632. Detail message: Column type ARRAY<varchar(65533)> does not support binary predicate operation.",
+      "reported_date": "2023-04-24",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22341"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22341",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22341",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: Getting analyzing error from line 1, column 591 to line 1, column 632. Detail message: Column type ARRAY<varchar(65533)> does not support binary predicate operation.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0f3a5d2d3ac863c0de20e46916b794693d540024eceb6717cbc2abfef034d94c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22341",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0f3a5d2d3ac863c0de20e46916b794693d540024eceb6717cbc2abfef034d94c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:1ab5559e7306",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: Getting analyzing error. Detail message: Column 'c_1_0' is ambiguous.",
+      "reported_date": "2023-04-24",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22338"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22338",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22338",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: Getting analyzing error. Detail message: Column 'c_1_0' is ambiguous.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1bc004d04f5075584db5ec68dec7c5897e0e020103dadf4ffc9da0e11f60d54b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22338",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1bc004d04f5075584db5ec68dec7c5897e0e020103dadf4ffc9da0e11f60d54b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a391dffbfdec",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: StarRocks planner use long time 3000 ms in logical phase, This probably because 1. FE Full GC, 2. Hive external table fetch metadata took a long time, 3. The SQL is very complex. You could 1. adjust FE JVM config, 2. try query again, 3. enlarge new_planner_optimize_timeout session variable",
+      "reported_date": "2023-04-24",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22335"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22335",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22335",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: StarRocks planner use long time 3000 ms in logical phase, This probably because 1. FE Full GC, 2. Hive external table fetch metadata took a long time, 3. The SQL is very complex. You could 1. adjust FE JVM config, 2. try query again, 3. enlarge new_planner_optimize_timeout session variable",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e4b05c2235a274608bff87934714296b605c0947d84909632146af9fae82f9ab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22335",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e4b05c2235a274608bff87934714296b605c0947d84909632146af9fae82f9ab"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:1f2a5654e1a3",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: StarRocks process failed",
+      "reported_date": "2023-04-24",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22349"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22349",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22349",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: StarRocks process failed",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:44a91dba8104e0e40cbaf8595a9f2f29f53a974b26020adc9382ca4ac20fb067",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22349",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:44a91dba8104e0e40cbaf8595a9f2f29f53a974b26020adc9382ca4ac20fb067"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:9560f7267910",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: Unexpected exception: Failed to acquire globalStateMgr lock. Try again",
+      "reported_date": "2023-04-24",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22334"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22334",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22334",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: Unexpected exception: Failed to acquire globalStateMgr lock. Try again",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3189a5619db0ffaea639792bd04766dcaf95ad7d7d0967132065ad24731a6f6f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22334",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3189a5619db0ffaea639792bd04766dcaf95ad7d7d0967132065ad24731a6f6f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:bf460d527b4a",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: Unexpected exception: fail to create tablet: timed out. unfinished replicas(1/1): 8468311(172.26.92.195)  timeout=2s",
+      "reported_date": "2023-04-24",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/22337"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/22337",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/22337",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: Unexpected exception: fail to create tablet: timed out. unfinished replicas(1/1): 8468311(172.26.92.195)  timeout=2s",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:cf19f978fe02ba6c29adb7ff2c3d9cc2dc630faf724123e3e315250e7993c805",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/22337",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cf19f978fe02ba6c29adb7ff2c3d9cc2dc630faf724123e3e315250e7993c805"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:a5338623ea59",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: cannot find statistics of col: 13: c_0_2",
+      "reported_date": "2023-07-27",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28087"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28087",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28087",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: cannot find statistics of col: 13: c_0_2",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8d6c1e4c3d41a70a3faae428904c0b638d48bbdcbde63f8ddb77e1e48a8f1c31",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28087",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8d6c1e4c3d41a70a3faae428904c0b638d48bbdcbde63f8ddb77e1e48a8f1c31"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:07e29a5f89d7",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: BE Crashed",
+      "reported_date": "2023-08-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28885"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28885",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28885",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: BE Crashed",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:959dc22681403414bb1f0774acd85b79370f2327f27741c0367fe753434ad836",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28885",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:959dc22681403414bb1f0774acd85b79370f2327f27741c0367fe753434ad836"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:4c0949b74429",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: BE Crashed",
+      "reported_date": "2023-08-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28887"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28887",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28887",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: BE Crashed",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:959dc22681403414bb1f0774acd85b79370f2327f27741c0367fe753434ad836",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28887",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:959dc22681403414bb1f0774acd85b79370f2327f27741c0367fe753434ad836"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:6b9bf4f1669d",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: BE Crashed",
+      "reported_date": "2023-08-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28889"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28889",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28889",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: BE Crashed",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:959dc22681403414bb1f0774acd85b79370f2327f27741c0367fe753434ad836",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28889",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:959dc22681403414bb1f0774acd85b79370f2327f27741c0367fe753434ad836"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:88508850a6f7",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: BE Crashed",
+      "reported_date": "2023-08-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28886"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28886",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28886",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: BE Crashed",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:959dc22681403414bb1f0774acd85b79370f2327f27741c0367fe753434ad836",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28886",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:959dc22681403414bb1f0774acd85b79370f2327f27741c0367fe753434ad836"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:9886467a299a",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: BE Crashed",
+      "reported_date": "2023-08-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28888"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28888",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28888",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: BE Crashed",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:959dc22681403414bb1f0774acd85b79370f2327f27741c0367fe753434ad836",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28888",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:959dc22681403414bb1f0774acd85b79370f2327f27741c0367fe753434ad836"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:be2220768ebb",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: Getting analyzing error from line 1, column 511 to line 1, column 554. Detail message: Unsupported nest window function inside aggregation.",
+      "reported_date": "2023-08-08",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28825"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28825",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28825",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: Getting analyzing error from line 1, column 511 to line 1, column 554. Detail message: Unsupported nest window function inside aggregation.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7d4330bb11cdf7da782f711e64c787f6a8faeb7b1bd73b82509d50b197ccb1a7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28825",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7d4330bb11cdf7da782f711e64c787f6a8faeb7b1bd73b82509d50b197ccb1a7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:b19548a676a4",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: Cancelled",
+      "reported_date": "2023-08-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28897"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28897",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28897",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: Cancelled",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c673e05cec9fc8b81c8552a1fc0723c5840a6f43ba2e5e2d036bb8603a576f4a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28897",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c673e05cec9fc8b81c8552a1fc0723c5840a6f43ba2e5e2d036bb8603a576f4a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:33da2508e3ac",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: CardinalityOfResultSetsMismatchError",
+      "reported_date": "2023-08-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28901"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28901",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28901",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: CardinalityOfResultSetsMismatchError",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:11d7748f81bc76670c5c2360adbfd3a3ab662e19261da1235c808d83e225a94c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28901",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:11d7748f81bc76670c5c2360adbfd3a3ab662e19261da1235c808d83e225a94c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:4c1c9e2d69bd",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: CardinalityOfResultSetsMismatchError",
+      "reported_date": "2023-08-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28902"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28902",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28902",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: CardinalityOfResultSetsMismatchError",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8f177696e6bfa015b46109f28656f7bdf4bf9696263b67b59e13bf4503945b27",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28902",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8f177696e6bfa015b46109f28656f7bdf4bf9696263b67b59e13bf4503945b27"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:603018adb44c",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: CardinalityOfResultSetsMismatchError",
+      "reported_date": "2023-08-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28899"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28899",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28899",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: CardinalityOfResultSetsMismatchError",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:12d8a8809553f5f2d467072a985a673f812bc50bdb38bbcaa746e157550d6cc2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28899",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:12d8a8809553f5f2d467072a985a673f812bc50bdb38bbcaa746e157550d6cc2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:dcaec61e2e81",
+      "dbms": "starrocks",
+      "title": "Sqlancer report error: CardinalityOfResultSetsMismatchError",
+      "reported_date": "2023-08-09",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "satanson",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/28898"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/28898",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/28898",
+            "source_type": "github_issue",
+            "excerpt": "Sqlancer report error: CardinalityOfResultSetsMismatchError",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f89c71f4d38dbe6b3e7843be3d68ae8311966ab7711c804ba545a78f49b81aaf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/28898",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f89c71f4d38dbe6b3e7843be3d68ae8311966ab7711c804ba545a78f49b81aaf"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:4bff5f19792b",
+      "dbms": "starrocks",
+      "title": "[sqlancer] Unexpected exception: failed to init view stmt",
+      "reported_date": "2023-10-27",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "packy92",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/33841"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/33841",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/33841",
+            "source_type": "github_issue",
+            "excerpt": "[sqlancer] Unexpected exception: failed to init view stmt",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5eef7c18912346bc4b8542d86ee18f518e08b0d6013ee9afb2293ee4f68798f9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/33841",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5eef7c18912346bc4b8542d86ee18f518e08b0d6013ee9afb2293ee4f68798f9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:cbc61179d558",
+      "dbms": "starrocks",
+      "title": "[BUG][SQLancer] array column append crash",
+      "reported_date": "2025-06-16",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Seaven",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/59921"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/59921",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/59921",
+            "source_type": "github_issue",
+            "excerpt": "[BUG][SQLancer] array column append crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0cc4615a1fd755481c0b417df80d99b6afc3583d3a5a6bbe8b79ff231ba9e296",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/59921",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0cc4615a1fd755481c0b417df80d99b6afc3583d3a5a6bbe8b79ff231ba9e296"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:starrocks:3c7873de7a1c",
+      "dbms": "starrocks",
+      "title": "[BUG][SQLancer] late materialized crash",
+      "reported_date": "2025-06-18",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Seaven",
+      "links": {
+        "report": "https://github.com/StarRocks/starrocks/issues/60021"
+      },
+      "primary_url": "https://github.com/StarRocks/starrocks/issues/60021",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/StarRocks/starrocks/issues/60021",
+            "source_type": "github_issue",
+            "excerpt": "[BUG][SQLancer] late materialized crash",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:be9bf74360990f97e406560a010e6bf8e1eec0d8f278a536311832d24c7cd9de",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/StarRocks/starrocks/issues/60021",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:be9bf74360990f97e406560a010e6bf8e1eec0d8f278a536311832d24c7cd9de"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:stonedb:7fcaba58f4b7",
+      "dbms": "stonedb",
+      "title": "bug: after altering operation, `NULL` values is evlauated to be `TRUE` in StoneDB",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+        "report": "https://github.com/stoneatom/stonedb/issues/1945"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/stoneatom/stonedb/issues/1945",
+            "excerpt_is_verbatim": true,
+            "note": "cnosdb/sqlancer records this bug in its stonedb provider, as the field bug1945 that works around it.",
+            "content_sha256": "sha256:7fcaba58f4b730cfac4e102779dcb644479017c758d97f03eea4ab1c1391fd9e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:7fcaba58f4b730cfac4e102779dcb644479017c758d97f03eea4ab1c1391fd9e"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1945",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:stonedb:189b7a9d783e",
+      "dbms": "stonedb",
+      "title": "bug: bad dpn index when deleting rows",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+        "report": "https://github.com/stoneatom/stonedb/issues/1933"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/stoneatom/stonedb/issues/1933",
+            "excerpt_is_verbatim": true,
+            "note": "cnosdb/sqlancer records this bug in its stonedb provider, as the field bug1933 that works around it.",
+            "content_sha256": "sha256:189b7a9d783e0a613b69406fd8151482af916a3972a250cadcaeb4e506899fe1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:189b7a9d783e0a613b69406fd8151482af916a3972a250cadcaeb4e506899fe1"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1933",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:stonedb:f8ec2f13c7ba",
+      "dbms": "stonedb",
+      "title": "bug: query result not correct",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+        "report": "https://github.com/stoneatom/stonedb/issues/1942"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/stoneatom/stonedb/issues/1942",
+            "excerpt_is_verbatim": true,
+            "note": "cnosdb/sqlancer records this bug in its stonedb provider, as the field bug1942 that works around it.",
+            "content_sha256": "sha256:f8ec2f13c7ba033069ccf9b430dd90811fff77092e9926b32a9976e40e38c141",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:f8ec2f13c7ba033069ccf9b430dd90811fff77092e9926b32a9976e40e38c141"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1942",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:stonedb:83c532bc99dc",
+      "dbms": "stonedb",
+      "title": "crash: StoneDB crash: SUM",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+        "report": "https://github.com/stoneatom/stonedb/issues/1953"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/stoneatom/stonedb/issues/1953",
+            "excerpt_is_verbatim": true,
+            "note": "cnosdb/sqlancer records this bug in its stonedb provider, as the field bug1953 that works around it.",
+            "content_sha256": "sha256:83c532bc99dc352470d49fcc9ca2d4df1c54dc92b52df42785d14f346a9360f4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/stonedb/StoneDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:83c532bc99dc352470d49fcc9ca2d4df1c54dc92b52df42785d14f346a9360f4"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1953",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:stonedb:8930de56edea",
+      "dbms": "stonedb",
+      "title": "bug: docker image stoneatom/stonedb:v1.0.4 is too large, is this expected?",
+      "reported_date": "2023-07-03",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1923"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1923",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1923",
+            "source_type": "github_issue",
+            "excerpt": "bug: docker image stoneatom/stonedb:v1.0.4 is too large, is this expected?",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:e8be9e53d09aa11666d60f438da49551cf99b40cec01712c7cea49a4461679e6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "zhenglin-charlie-li is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1923",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e8be9e53d09aa11666d60f438da49551cf99b40cec01712c7cea49a4461679e6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:d59d49d6fc6e",
+      "dbms": "stonedb",
+      "title": "crash: StoneDB crash when executing the command (SELECT, HAING, GROUP BY, IS NULL)",
+      "reported_date": "2023-07-14",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1941"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1941",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1941",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer find that StoneDB will crash when executing the command:\r\n```sql",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d1e6b4d27d47bc6e8e50f32a1b01e35fd115b2455bfcc1bb9897f00e91a7b439",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1941",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d1e6b4d27d47bc6e8e50f32a1b01e35fd115b2455bfcc1bb9897f00e91a7b439"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:c7f5d6444fda",
+      "dbms": "stonedb",
+      "title": "bug: expected supported SQL sytax not support (UNION ALL, LIKE)",
+      "reported_date": "2023-07-16",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1943"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1943",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1943",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0\r\n(\r\n    c0 INT\r\n);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2f915e2dd693de117239ca5f70c23b0d2e366d7c546958c3a8b9b100703ed533",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1943",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2f915e2dd693de117239ca5f70c23b0d2e366d7c546958c3a8b9b100703ed533"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:9ae389c20d18",
+      "dbms": "stonedb",
+      "title": "crash: StoneDB crash when executing the right shift operator",
+      "reported_date": "2023-07-18",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1947"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1947",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1947",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT); \r\nINSERT INTO t0(c0) VALUE (DEFAULT);\r\nDELETE FROM t0 WHERE ((t0.c0)>>(t0.c0));  \r\n-- note: to reproduce the crash, need to run all commands quickly. Copy this line to let all commands run quickly",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:15dd31c447b32444caf7249a3cdb472fcb5f535d619b585b8a8168c3ada770ac",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1947",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:15dd31c447b32444caf7249a3cdb472fcb5f535d619b585b8a8168c3ada770ac"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:ef1f94719845",
+      "dbms": "stonedb",
+      "title": "bug: query result wrong: ALTER, DEFAULT",
+      "reported_date": "2023-07-19",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1955"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1955",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1955",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT) ;\r\nINSERT INTO t0(c0) VALUES (DEFAULT);\r\nALTER TABLE t0 ADD COLUMN c1 INT DEFAULT 1;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2a44cc7b0d4ef23194b2d49958e42999be0852ea8e5a339ad9cabdda5bd0a7fd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1955",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2a44cc7b0d4ef23194b2d49958e42999be0852ea8e5a339ad9cabdda5bd0a7fd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:881e34b500ba",
+      "dbms": "stonedb",
+      "title": "bug: query result wrong: CASE WHEN THEN ELSE",
+      "reported_date": "2023-07-19",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1950"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1950",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1950",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT NULL);\r\nINSERT INTO t0(c0) VALUES (DEFAULT);\r\nSELECT * FROM t0 WHERE (CASE (t0.c0 IN (t0.c0)) WHEN TRUE THEN 'TRUE' ELSE 'FALSE' END );\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1cd8ac550ab125c9b013de2e70d9e94696ead0433835eec9872810efa8a5396d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1950",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1cd8ac550ab125c9b013de2e70d9e94696ead0433835eec9872810efa8a5396d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:4bc3105d5f13",
+      "dbms": "stonedb",
+      "title": "bug: query result wrong: INSERT, DELETE, UNION ALL",
+      "reported_date": "2023-07-19",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1956"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1956",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1956",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 CHAR) ;\r\nINSERT INTO t0(c1) VALUE (1);\r\nDELETE  FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a392ace4d2979e6b5ad897c16e964e47cc2455c48ea2cd92f4a01df06b81dcf4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1956",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a392ace4d2979e6b5ad897c16e964e47cc2455c48ea2cd92f4a01df06b81dcf4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:fd99ef970307",
+      "dbms": "stonedb",
+      "title": "bug: query result wrong: NATURAL LEFT JOIN",
+      "reported_date": "2023-07-19",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1957"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1957",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1957",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer find that StoneDB's query result does not meet expectation:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b869014fde5d862ac6d2eb96b7cd4698d10ee58a5d716f2a70690a6635aa90c9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1957",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b869014fde5d862ac6d2eb96b7cd4698d10ee58a5d716f2a70690a6635aa90c9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:29daf6f622a8",
+      "dbms": "stonedb",
+      "title": "bug: query result wrong: SELECT",
+      "reported_date": "2023-07-19",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1949"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1949",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1949",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 TEXT);\r\nINSERT INTO t0(c0) VALUES (DEFAULT);\r\nDELETE FROM t0;\r\nINSERT INTO t0(c0) VALUES (DEFAULT);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d7b6e96ee79377ec8c214536f34a6e8f44fb8fa048c3b10c786311e604df0fd5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1949",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d7b6e96ee79377ec8c214536f34a6e8f44fb8fa048c3b10c786311e604df0fd5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:00adfdb120d5",
+      "dbms": "stonedb",
+      "title": "crash: StoneDB crash: HAVING, IS NULL",
+      "reported_date": "2023-07-19",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1954"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1954",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1954",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nSELECT * FROM t0 HAVING (1 IS NULL);\r\n```\r\n### Environment",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:eb655967243a60854ad3d0dc2ee8c6bf21226df9f2891ba497c149dc735f12cb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1954",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:eb655967243a60854ad3d0dc2ee8c6bf21226df9f2891ba497c149dc735f12cb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:c32bdb8089e7",
+      "dbms": "stonedb",
+      "title": "crash: StoneDB crash: PRIMARY KEY, HAVING, IS NULL",
+      "reported_date": "2023-07-19",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1952"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1952",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1952",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT PRIMARY KEY);\r\n\r\nSELECT * FROM t0 HAVING (t0.c0 IS NULL);\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c0ab218b8cec5a2362e7da691e3fea21d503d2fc30b210aa6e3f7db803228f95",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1952",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c0ab218b8cec5a2362e7da691e3fea21d503d2fc30b210aa6e3f7db803228f95"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:stonedb:0f881f02bc4a",
+      "dbms": "stonedb",
+      "title": "bug: can not build from source code when following the doc",
+      "reported_date": "2023-07-29",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "zhenglin-charlie-li",
+      "links": {
+        "report": "https://github.com/stoneatom/stonedb/issues/1960"
+      },
+      "primary_url": "https://github.com/stoneatom/stonedb/issues/1960",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/stoneatom/stonedb/issues/1960",
+            "source_type": "github_issue",
+            "excerpt": "bug: can not build from source code when following the doc",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:ebb93c50442815fd0cc6a8c9b923919671c7ef1323651e98289e82c52bb16a4f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "zhenglin-charlie-li is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/stoneatom/stonedb/issues/1960",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ebb93c50442815fd0cc6a8c9b923919671c7ef1323651e98289e82c52bb16a4f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:115bfcc4e9a9",
+      "dbms": "tdengine",
+      "title": "DROP DATABASE seems to leave behind a table when using the \"tables\" configuration option",
+      "reported_date": "2019-10-01",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://github.com/taosdata/TDengine/issues/586"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/586",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/10/2019\",\n        \"dbms\": \"TDEngine\",\n        \"links\": {\n            \"bugtracker\": \"https://github.com/taosdata/TDengine/issues/586\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:81bdba1708c7e5c888fcc0bd14337b018d47535e664cc5dc7b17373ab1b285c0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:81bdba1708c7e5c888fcc0bd14337b018d47535e664cc5dc7b17373ab1b285c0"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/586",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:999ca0b09ee1",
+      "dbms": "tdengine",
+      "title": "Likely typo in configuration option \"abloks\"",
+      "reported_date": "2019-10-01",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "mrigger",
+      "links": {
+        "report": "https://github.com/taosdata/TDengine/issues/587"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/587",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/587",
+            "source_type": "github_issue",
+            "excerpt": "Likely typo in configuration option \"abloks\"",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:98a44fa5b2034a57fc67ab01db756fc738118878736d55fce410693c52a3e990",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "mrigger is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/taosdata/TDengine/issues/587",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:98a44fa5b2034a57fc67ab01db756fc738118878736d55fce410693c52a3e990"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:d4a77e414093",
+      "dbms": "tdengine",
+      "title": "Inserting the \"+\" and \"-\" strings fails",
+      "reported_date": "2019-10-02",
+      "reported_year": 2019,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://github.com/taosdata/TDengine/issues/589"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/589",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/10/2019\",\n        \"dbms\": \"TDEngine\",\n        \"links\": {\n            \"bugtracker\": \"https://github.com/taosdata/TDengine/issues/589\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:975b8ff0a0e7201dc140550267bf382dea7e176f2babb855fe15be7c554d2489",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:975b8ff0a0e7201dc140550267bf382dea7e176f2babb855fe15be7c554d2489"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/589",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:dc74b47c89de",
+      "dbms": "tdengine",
+      "title": "The \"<>\" and \"=\" operators do not work for NCHAR columns",
+      "reported_date": "2019-10-03",
+      "reported_year": 2019,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://github.com/taosdata/TDengine/issues/590"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/590",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"03/10/2019\",\n        \"dbms\": \"TDEngine\",\n        \"links\": {\n            \"bugtracker\": \"https://github.com/taosdata/TDengine/issues/590\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5e70493d987be1d5bc652a95e47a827b0050830452a9f31812c8418f2d5da414",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5e70493d987be1d5bc652a95e47a827b0050830452a9f31812c8418f2d5da414"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/590",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:2fca7bba2017",
+      "dbms": "tdengine",
+      "title": "Unexpected result when using arithmetic expressions in WHERE clause",
+      "reported_date": "2019-10-03",
+      "reported_year": 2019,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugtracker": "https://github.com/taosdata/TDengine/issues/591"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/591",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"03/10/2019\",\n        \"dbms\": \"TDEngine\",\n        \"links\": {\n            \"bugtracker\": \"https://github.com/taosdata/TDengine/issues/591\"\n        },\n        \"oracle\": \"PQS\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Pivoted Query Synthesis (PQS) oracle found this bug.",
+            "content_sha256": "sha256:a5a93d9174093c19fd8c82f81826553d513cf080044e01c5bf66604fd6a8d644",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a5a93d9174093c19fd8c82f81826553d513cf080044e01c5bf66604fd6a8d644"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/591",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:0b6b92cc7e2c",
+      "dbms": "tdengine",
+      "title": "Segmentation fault with certain queries",
+      "reported_date": "2023-05-16",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/taosdata/TDengine/issues/21323"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/21323",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/21323",
+            "source_type": "github_issue",
+            "excerpt": "Segmentation fault with certain queries",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2d10f1175ac1cb5dac1ecedcfc0a87ca0d11a765899b996caa82d635db32b57a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/taosdata/TDengine/issues/21323",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2d10f1175ac1cb5dac1ecedcfc0a87ca0d11a765899b996caa82d635db32b57a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:dd4c725273e8",
+      "dbms": "tdengine",
+      "title": "Unexpected Result with `null` in equality operator.",
+      "reported_date": "2023-05-17",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/taosdata/TDengine/issues/21351"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/21351",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/21351",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Result with `null` in equality operator.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:571312446886b2f46ca7cf559dfb76569f9dcb34f306cae15aaa47bd13937531",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/taosdata/TDengine/issues/21351",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:571312446886b2f46ca7cf559dfb76569f9dcb34f306cae15aaa47bd13937531"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:9797d6103544",
+      "dbms": "tdengine",
+      "title": "Unexpected TDengine Exception [0x2604] Column ambiguously defined",
+      "reported_date": "2023-05-17",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/taosdata/TDengine/issues/21349"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/21349",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/21349",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected TDengine Exception [0x2604] Column ambiguously defined",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:bad8e760116cd5decb2c4cf427f9de93eac8af76feae63c12f564728d4f3c39a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/taosdata/TDengine/issues/21349",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bad8e760116cd5decb2c4cf427f9de93eac8af76feae63c12f564728d4f3c39a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:a80399cad4d0",
+      "dbms": "tdengine",
+      "title": "Unexpected generic error with `LIKE` query",
+      "reported_date": "2023-05-18",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/taosdata/TDengine/issues/21368"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/21368",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/21368",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected generic error with `LIKE` query",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:63e779c1554fde9c2422e75dae8283c96e7e0c134075e654a71f29552dbea93e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/taosdata/TDengine/issues/21368",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:63e779c1554fde9c2422e75dae8283c96e7e0c134075e654a71f29552dbea93e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:0373a135fabc",
+      "dbms": "tdengine",
+      "title": "Unexpected Syntax Error with `Null`",
+      "reported_date": "2023-05-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/taosdata/TDengine/issues/21520"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/21520",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/21520",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected Syntax Error with `Null`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5a5af7a493a2a52c6e6d6dfcc777cffede8a29b4fe1acef51a60eac8fb85c300",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/taosdata/TDengine/issues/21520",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5a5af7a493a2a52c6e6d6dfcc777cffede8a29b4fe1acef51a60eac8fb85c300"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:2eebb16c15b9",
+      "dbms": "tdengine",
+      "title": "Type DOUBLE variable in predicate produces unexpected results",
+      "reported_date": "2023-06-05",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/taosdata/TDengine/issues/21603"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/21603",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/21603",
+            "source_type": "github_issue",
+            "excerpt": "Type DOUBLE variable in predicate produces unexpected results",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c5e154e6b4c82517f1a0e161218e41394ac3e71a8c870f19d25e34eff0fad1c5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/taosdata/TDengine/issues/21603",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c5e154e6b4c82517f1a0e161218e41394ac3e71a8c870f19d25e34eff0fad1c5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tdengine:ea3e0ec4f6be",
+      "dbms": "tdengine",
+      "title": "Segmentation fault when having `IS NULL` in `JOIN` condition",
+      "reported_date": "2023-09-12",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "YuanchengJiang",
+      "links": {
+        "report": "https://github.com/taosdata/TDengine/issues/22857"
+      },
+      "primary_url": "https://github.com/taosdata/TDengine/issues/22857",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/taosdata/TDengine/issues/22857",
+            "source_type": "github_issue",
+            "excerpt": "Segmentation fault when having `IS NULL` in `JOIN` condition",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:9b928c560e47f5963659a0c8f89afa075bfb0d37015ab32f3b4b92b90de05550",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "YuanchengJiang is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/taosdata/TDengine/issues/22857",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9b928c560e47f5963659a0c8f89afa075bfb0d37015ab32f3b4b92b90de05550"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:ed0e8cab8165",
+      "dbms": "tidb",
+      "title": "ERROR 1105 (HY000): interface conversion",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/46556"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/46556",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug46556 that works around it.",
+            "content_sha256": "sha256:ed0e8cab81656816cf052729ff63a865b542c1f3212eaa50d3dcd3bd60bd4e61",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:ed0e8cab81656816cf052729ff63a865b542c1f3212eaa50d3dcd3bd60bd4e61"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/46556",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:706fa3ed0bd7",
+      "dbms": "tidb",
+      "title": "ERROR 1105 encoding failed",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/47346"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/47346",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug47346 that works around it.",
+            "content_sha256": "sha256:706fa3ed0bd790f72e1e2d4fb43c827bf50a93134b1b5a207f0651e6b1985704",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:706fa3ed0bd790f72e1e2d4fb43c827bf50a93134b1b5a207f0651e6b1985704"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/47346",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:2bc10feb56ae",
+      "dbms": "tidb",
+      "title": "ERROR 8141 (HY000): assertion failed",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/38295"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/38295",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug38295 that works around it.",
+            "content_sha256": "sha256:2bc10feb56aebd738011aa55f50b50d78b3442fbd055f6728e685ebeb00eabfb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:2bc10feb56aebd738011aa55f50b50d78b3442fbd055f6728e685ebeb00eabfb"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38295",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:9e23d5e1e034",
+      "dbms": "tidb",
+      "title": "Error For MPP Stream",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/46598"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/46598",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug46598 that works around it.",
+            "content_sha256": "sha256:9e23d5e1e0347688257ad73187a2a70d27676e41ce8ed957fd5d93bb9da3fc53",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:9e23d5e1e0347688257ad73187a2a70d27676e41ce8ed957fd5d93bb9da3fc53"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/46598",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:9f1777ac97fa",
+      "dbms": "tidb",
+      "title": "Internal Error: Cannot found physical plan",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/46591"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/46591",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug46591 that works around it.",
+            "content_sha256": "sha256:9f1777ac97fab5721acab19d5473d39de549c6c74e8732d0d16c713985ac15c2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:9f1777ac97fab5721acab19d5473d39de549c6c74e8732d0d16c713985ac15c2"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/46591",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:6c0d55cc30df",
+      "dbms": "tidb",
+      "title": "Unexpected Error Overflow",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/47348"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/47348",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug47348 that works around it.",
+            "content_sha256": "sha256:6c0d55cc30df2ebefaa4193ff617194e118b66aba192f70587e82c3c488db47e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:6c0d55cc30df2ebefaa4193ff617194e118b66aba192f70587e82c3c488db47e"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/47348",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:b35d0c5791be",
+      "dbms": "tidb",
+      "title": "Unexpected Error by CAST and CHAR functions",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/35652"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/35652",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug35652 that works around it.",
+            "content_sha256": "sha256:b35d0c5791be3ac4e1ab98636f261d9cec63e5260fb78533d174e70f9348b030",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:b35d0c5791be3ac4e1ab98636f261d9cec63e5260fb78533d174e70f9348b030"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/35652",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:0bfbd6f267aa",
+      "dbms": "tidb",
+      "title": "Unexpected Error for Function INET_ATON",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/35677"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/35677",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug35677 that works around it.",
+            "content_sha256": "sha256:0bfbd6f267aaa5ea12cb817451d1ff5e2c4caf2ce774ee7fe56a359fc60a52fc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:0bfbd6f267aaa5ea12cb817451d1ff5e2c4caf2ce774ee7fe56a359fc60a52fc"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/35677",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:e37bd1efd566",
+      "dbms": "tidb",
+      "title": "Unexpected Estimated Rows by GROUP BY",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/51525"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/51525",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug51525 that works around it.",
+            "content_sha256": "sha256:e37bd1efd56694eb99ced2ba3ca7a09370908fb98ce4ab2b54c92992c20dcf24",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-08T16:31:08Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:31:08Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:e37bd1efd56694eb99ced2ba3ca7a09370908fb98ce4ab2b54c92992c20dcf24"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/51525",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:6c79eb8aeb55",
+      "dbms": "tidb",
+      "title": "Unexpected Estimated Rows of `OR`",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/38319"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/38319",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug38319 that works around it.",
+            "content_sha256": "sha256:6c79eb8aeb553b4aa7a61e396ffb2f53b0ef3c65a6ca6c32b7ce4e596a623e39",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:6c79eb8aeb553b4aa7a61e396ffb2f53b0ef3c65a6ca6c32b7ce4e596a623e39"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38319",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:661dd62d1457",
+      "dbms": "tidb",
+      "title": "Unexpected Result with NATURAL RIGHT JOIN and Bitwise NOT",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/53506"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/53506",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/53506",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE VIEW v0(c0) AS SELECT (0) FROM t0;\r\nINSERT INTO t0 (c0) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:92fbb04cf7a53bf61a22e5c7bd1b881e186b4c446cad1632475aa8fcdce63cf4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:92fbb04cf7a53bf61a22e5c7bd1b881e186b4c446cad1632475aa8fcdce63cf4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:8e3afd789a3b",
+      "dbms": "tidb",
+      "title": "incorrect unresolved column when using natural join",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/35522"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/35522",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug35522 that works around it.",
+            "content_sha256": "sha256:8e3afd789a3b1b6671e56eec58568ce02e194ab6aa78d3e37e30546ea2a429aa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:8e3afd789a3b1b6671e56eec58568ce02e194ab6aa78d3e37e30546ea2a429aa"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/35522",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:b1db3a832efe",
+      "dbms": "tidb",
+      "title": "runtime error: index out of range [7] with length 4",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "report": "https://github.com/pingcap/tidb/issues/44747"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/pingcap/tidb/issues/44747",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its tidb provider, as the field bug44747 that works around it.",
+            "content_sha256": "sha256:b1db3a832efe825e3f88ce0bcadb0585d75266fd7d10755cff3a842b0dbf819c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/tidb/TiDBBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:b1db3a832efe825e3f88ce0bcadb0585d75266fd7d10755cff3a842b0dbf819c"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/44747",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:tidb:24d479578ac2",
+      "dbms": "tidb",
+      "title": "ANALYZE TABLE causes SIGSEGV on latest trunk",
+      "reported_date": "2020-03-26",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15746",
+        "fix": "https://github.com/pingcap/tidb/pull/15765"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15746",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15746\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/15765\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f5f8dc5d10e53432be8b86531a52d05f97635cc04879ce2a6424605a9f448814",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f5f8dc5d10e53432be8b86531a52d05f97635cc04879ce2a6424605a9f448814"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15746",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:af9c06af74ca",
+      "dbms": "tidb",
+      "title": "ANALYZE TABLE results in \"analyze worker panic\" with tidb_enable_fast_analyze=1",
+      "reported_date": "2020-03-26",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15751",
+        "fix": "https://github.com/pingcap/tidb/pull/15845"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15751",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15751\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/15845\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1108c4a226b78c4ce90fa6ea6c56d9a131eb474752bcdc4479e0e50499102959",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1108c4a226b78c4ce90fa6ea6c56d9a131eb474752bcdc4479e0e50499102959"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15751",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:71de68daa004",
+      "dbms": "tidb",
+      "title": "ANALYZE TABLE results in \"invalid encoded key\" with tidb_enable_fast_analyze=1",
+      "reported_date": "2020-03-26",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15752",
+        "fix": "https://github.com/pingcap/tidb/pull/15889"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15752",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15752\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/15889\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c5f3bafbc08f850a3643e9635fe12a2419ac29b7a3ba1415e965cd17dd90ec5d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c5f3bafbc08f850a3643e9635fe12a2419ac29b7a3ba1415e965cd17dd90ec5d"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15752",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:74051d463671",
+      "dbms": "tidb",
+      "title": "Double negation causes incorrect result",
+      "reported_date": "2020-03-26",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15725",
+        "fix": "https://github.com/pingcap/tidb/pull/16108"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15725",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15725\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16108\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:9b5feb9116174c003f5d17ea89e96593e1b6e7ecdd401ad15e9a7b18b698b78a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9b5feb9116174c003f5d17ea89e96593e1b6e7ecdd401ad15e9a7b18b698b78a"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15725",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:bac331e34721",
+      "dbms": "tidb",
+      "title": "Incorrect result for an UNION query and a generated column",
+      "reported_date": "2020-03-26",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15733"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15733",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15733\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:db9a9d5d00fb50d1fc561222a2361c6e4710dacf9022a286446cda1546b4c2a6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:db9a9d5d00fb50d1fc561222a2361c6e4710dacf9022a286446cda1546b4c2a6"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15733",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:cf8471953433",
+      "dbms": "tidb",
+      "title": "Incorrect result for query that uses an AND operator on floats",
+      "reported_date": "2020-03-26",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15743"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15743",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"26/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15743\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:bb0290d9be366dd6db4aad98a8288197a9b3e32f0f1c24f406efc3cedd9820ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:bb0290d9be366dd6db4aad98a8288197a9b3e32f0f1c24f406efc3cedd9820ed"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15743",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:1db419770558",
+      "dbms": "tidb",
+      "title": "COLLATE causes an incorrect result in the presence of an index",
+      "reported_date": "2020-03-27",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15789",
+        "fix": "https://github.com/pingcap/tidb/pull/15971"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15789",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"27/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15789\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/15971\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:ae871d229a0ff59a08cb72fb731e9e8247170f63f4e4e8655612ff5e3ebb6d43",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ae871d229a0ff59a08cb72fb731e9e8247170f63f4e4e8655612ff5e3ebb6d43"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15789",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:8383c5e280b2",
+      "dbms": "tidb",
+      "title": "Unexpected \"Data truncation: %s value is out of range in '%s'\" error in UNION query",
+      "reported_date": "2020-03-27",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15790",
+        "fix": "https://github.com/pingcap/tidb/pull/16073"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15790",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"27/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15790\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16073\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a309da88d2368d41cf4561ee8d286c2c122b04d4455f91a8346a52c1b97c8ea0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a309da88d2368d41cf4561ee8d286c2c122b04d4455f91a8346a52c1b97c8ea0"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15790",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:31de5c1dc286",
+      "dbms": "tidb",
+      "title": "A predicate column1 = -column2 incorrectly evaluates to false for 0 values",
+      "reported_date": "2020-03-29",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15812",
+        "fix": "https://github.com/pingcap/tidb/pull/15837"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15812",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15812\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/15837\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:b3ecaaa61ab805b33b490806dd0185a78728b37b89ec1f0dbae190b570c4562c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:b3ecaaa61ab805b33b490806dd0185a78728b37b89ec1f0dbae190b570c4562c"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15812",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:8cc26f07b530",
+      "dbms": "tidb",
+      "title": "GROUP BY clause nondeterministically results in an incorrect result or error",
+      "reported_date": "2020-03-29",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15810",
+        "fix": "https://github.com/pingcap/tidb/pull/16600"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15810",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15810\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16600\"\n        },\n        \"oracle\": \"TLP (GROUP BY)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:d419b0ba5a7dc08f495b20f71e5a6872d1697cab0bb68d4fd24add23c8af37d2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d419b0ba5a7dc08f495b20f71e5a6872d1697cab0bb68d4fd24add23c8af37d2"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15810",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:55cdbcfc2e9d",
+      "dbms": "tidb",
+      "title": "Join on tables with redundant indexes causes a server panic",
+      "reported_date": "2020-03-29",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15813",
+        "fix": "https://github.com/pingcap/tidb/pull/15840"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15813",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"29/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15813\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/15840\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7f5c69fa9195e2aa603f0d35579f566f6b0f2542c6fe06aeee43c223b8468742",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7f5c69fa9195e2aa603f0d35579f566f6b0f2542c6fe06aeee43c223b8468742"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15813",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:735d618642d3",
+      "dbms": "tidb",
+      "title": "Incorrect result for LEFT JOIN and NULL values",
+      "reported_date": "2020-03-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15846",
+        "fix": "https://github.com/pingcap/tidb/pull/15894"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15846",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15846\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/15894\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:49c79a8dbd4ad1708ee92b0397fd4fa73b8cf101b50b9cc3cdc2f24b8b021d9f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:49c79a8dbd4ad1708ee92b0397fd4fa73b8cf101b50b9cc3cdc2f24b8b021d9f"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15846",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:79dadaad3ed4",
+      "dbms": "tidb",
+      "title": "NATURAL RIGHT JOIN results in an unexpected \"Unknown column\" error",
+      "reported_date": "2020-03-30",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15844"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15844",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15844\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ed5065f6fa5733469d283f4512a41b31329a344f0e6f4d5f0150a59e709244ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ed5065f6fa5733469d283f4512a41b31329a344f0e6f4d5f0150a59e709244ed"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15844",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:da227834d3fe",
+      "dbms": "tidb",
+      "title": "Query results in \"baseBuiltinFunc.vecEvalString() should never be called\" error",
+      "reported_date": "2020-03-30",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15847"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15847",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15847\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:6e45e04be8872c06bc93916a7fa342729d6f8bcb32ca84a18c354b8b65903cc1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6e45e04be8872c06bc93916a7fa342729d6f8bcb32ca84a18c354b8b65903cc1"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15847",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:17c27e5484a4",
+      "dbms": "tidb",
+      "title": "Query with RIGHT JOIN causes a server panic",
+      "reported_date": "2020-03-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15858",
+        "fix": "https://github.com/pingcap/tidb/pull/15947"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15858",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15858\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/15947\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f21e20175c61ed70fe28c8763d5b749a4497418770afa8c33b0008c12f8f77d4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f21e20175c61ed70fe28c8763d5b749a4497418770afa8c33b0008c12f8f77d4"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15858",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:5c42b224022a",
+      "dbms": "tidb",
+      "title": "Query with a NATURAL LEFT JOIN unexpectedly results in an error",
+      "reported_date": "2020-03-30",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15850",
+        "fix": "https://github.com/pingcap/tidb/pull/15984"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15850",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"30/03/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15850\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/15984\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a4afa6fa5245e59b57a182989088368e3fa74a5d942ea5d4ce02b872e01b763f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a4afa6fa5245e59b57a182989088368e3fa74a5d942ea5d4ce02b872e01b763f"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15850",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:b2f7bd6ac33e",
+      "dbms": "tidb",
+      "title": "A USE_INDEX_MERGE hint causes a server panic",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15991",
+        "fix": "https://github.com/pingcap/tidb/pull/16001"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15991",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15991\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16001\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:76129d665483e5f9c06d69c372bfe2c2c4665806a4850b6ef999dab190f4987e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:76129d665483e5f9c06d69c372bfe2c2c4665806a4850b6ef999dab190f4987e"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15991",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:362fe8b75c41",
+      "dbms": "tidb",
+      "title": "Incorrect result for a predicate that uses the CHAR() function",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15986",
+        "fix": "https://github.com/pingcap/tidb/pull/16014"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15986",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15986\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16014\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:5c269311ccf87e920e1c58a8cc542a36e953f5c2944ba0e8778bb090cb51e293",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5c269311ccf87e920e1c58a8cc542a36e953f5c2944ba0e8778bb090cb51e293"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15986",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:e506d933962a",
+      "dbms": "tidb",
+      "title": "LIKE operator malfunctions for COLLATE 'latin1_bin'",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15988",
+        "fix": "https://github.com/pingcap/tidb/pull/16380"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15988",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15988\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16380\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:918c6b799b001fa64638f17dda5418cbaa66d36ece92b67015451b9915ceef9a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:918c6b799b001fa64638f17dda5418cbaa66d36ece92b67015451b9915ceef9a"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15988",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:30442098b746",
+      "dbms": "tidb",
+      "title": "OR clause on FLOAT/DOUBLE column unexpectedly evaluates to TRUE",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15987"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15987",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15987\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:054a920c9c2aa62e544b5ff53ed6d41bd9e87232b393101a5736cf5946dca670",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:054a920c9c2aa62e544b5ff53ed6d41bd9e87232b393101a5736cf5946dca670"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15987",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:051da4ca2346",
+      "dbms": "tidb",
+      "title": "USE_INDEX_MERGE on table with generated column causes a server crash",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "crash",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15992",
+        "fix": "https://github.com/pingcap/tidb/pull/16376"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15992",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15992\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16376\"\n        },\n        \"oracle\": \"crash\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:2e12f438634a5cd9ff46b12376a1a2ab792b79d853d4e33dbd8ee0d7e040d803",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2e12f438634a5cd9ff46b12376a1a2ab792b79d853d4e33dbd8ee0d7e040d803"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15992",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:c130d4db1ca6",
+      "dbms": "tidb",
+      "title": "USE_INDEX_MERGE results in an incorrect result for a generated column",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15994",
+        "fix": "https://github.com/pingcap/tidb/pull/16002"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15994",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15994\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16002\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:2ca8dd304454c0c7926a8a2b543989f7c8acefad0817e90f3e287df591cf124a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2ca8dd304454c0c7926a8a2b543989f7c8acefad0817e90f3e287df591cf124a"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15994",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:0841489764fd",
+      "dbms": "tidb",
+      "title": "Using a column both in a string comparison and as a boolean yields an incorrect result",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15990",
+        "fix": "https://github.com/pingcap/tidb/pull/16135"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15990",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15990\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16135\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:90ee45353e8c9c541944a80c6280b36a1b1b04bc08845727efc7866db2f285c1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:90ee45353e8c9c541944a80c6280b36a1b1b04bc08845727efc7866db2f285c1"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15990",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:cfd7ce1c9e2a",
+      "dbms": "tidb",
+      "title": "fast ANALYZE TABLE on INDEX PRIMARY causes an error \"analyze worker panic\"",
+      "reported_date": "2020-04-01",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/15993",
+        "fix": "https://github.com/pingcap/tidb/pull/16005"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/15993",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"01/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/15993\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16005\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e9bd957cb617f076571b6359302e4412f0b24603d50affda95badd062fb7e87c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e9bd957cb617f076571b6359302e4412f0b24603d50affda95badd062fb7e87c"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/15993",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:d04d50ac3340",
+      "dbms": "tidb",
+      "title": "INL_MERGE_JOIN hint results in an error \"Internal : Can't find a proper physical plan for this query\"",
+      "reported_date": "2020-04-02",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/16017"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/16017",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/16017\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:23f46c4b33aa7f049f606d0a74feda820b799f91347dffd13f20737817124b83",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:23f46c4b33aa7f049f606d0a74feda820b799f91347dffd13f20737817124b83"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/16017",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:a8e23f356d70",
+      "dbms": "tidb",
+      "title": "INSERT IGNORE causes an incorrect result for a query on a DECIMAL column",
+      "reported_date": "2020-04-02",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/16025",
+        "fix": "https://github.com/pingcap/tidb/pull/16518"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/16025",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/16025\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16518\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:63acc48ea9f63184dea93651ba581e2b3ea2f06eb81aa57ca5668cf5f2d9dd53",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:63acc48ea9f63184dea93651ba581e2b3ea2f06eb81aa57ca5668cf5f2d9dd53"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/16025",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:2381a231a50b",
+      "dbms": "tidb",
+      "title": "Incorrect result when comparing a FLOAT/DOUBLE UNSIGNED with a negative number",
+      "reported_date": "2020-04-02",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/16028"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/16028",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/16028\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:3a6be6be30e6d43fc27a92469938d895ebae3e50f699c38897ff86b95d429b8f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3a6be6be30e6d43fc27a92469938d895ebae3e50f699c38897ff86b95d429b8f"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/16028",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:c978cd4be7db",
+      "dbms": "tidb",
+      "title": "LEFT JOIN on a view results in \"runtime error: slice bounds out of range [:264] with capacity 256\"",
+      "reported_date": "2020-04-02",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/16027"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/16027",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/16027\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:7b21a34674a4ad6880b50f6710c2ac4c121f7ff52797c0f32920358a285cab2e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7b21a34674a4ad6880b50f6710c2ac4c121f7ff52797c0f32920358a285cab2e"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/16027",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:33f640878217",
+      "dbms": "tidb",
+      "title": "REGEXP predicate unexpectedly results in an error \"No valid regexp pattern found\"",
+      "reported_date": "2020-04-02",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/16029",
+        "fix": "https://github.com/pingcap/tidb/pull/16405"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/16029",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/16029\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16405\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:1f3f37db35ed05d25164e75aed79fa04ceb2be8211b98c325c15efc160dbfb39",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:1f3f37db35ed05d25164e75aed79fa04ceb2be8211b98c325c15efc160dbfb39"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/16029",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:4daa1540db06",
+      "dbms": "tidb",
+      "title": "RIGHT JOIN with CONCAT_WS fails to fetch a row",
+      "reported_date": "2020-04-02",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/16023",
+        "fix": "https://github.com/pingcap/tidb/pull/16444"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/16023",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/16023\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16444\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:d0bdf8bd680df1d55ed75999efa81bca1936a8884b16e073f6672fb368589155",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:d0bdf8bd680df1d55ed75999efa81bca1936a8884b16e073f6672fb368589155"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/16023",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:01b16267ebcb",
+      "dbms": "tidb",
+      "title": "SELECT on table with generated column causes a server panic",
+      "reported_date": "2020-04-02",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/16020",
+        "fix": "https://github.com/pingcap/tidb/pull/16316"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/16020",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"02/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/16020\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16316\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:21bd15ee2669ef60f9f0165b7423164e6a49d9882690d46a2855dff2362dac64",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:21bd15ee2669ef60f9f0165b7423164e6a49d9882690d46a2855dff2362dac64"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/16020",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:419a26f52836",
+      "dbms": "tidb",
+      "title": "A NOT NULL predicate unexpectedly evaluates to TRUE",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "closed_duplicate",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/16440"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/16440",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/16440\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:7aaf58700a4f10ca301b5c592ae0dc914a83ea4fc6c8b853f8b00e40301c056c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:7aaf58700a4f10ca301b5c592ae0dc914a83ea4fc6c8b853f8b00e40301c056c"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/16440",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:bbd8e9948fab",
+      "dbms": "tidb",
+      "title": "Incorrect result for CAST to DATETIME",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/13"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/13",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/13\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:9babc945555f1bed05a7eebddc39d3c5a93adc3e24138873f5c30e8bc948a1c4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9babc945555f1bed05a7eebddc39d3c5a93adc3e24138873f5c30e8bc948a1c4"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/13",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:76d2c612d88f",
+      "dbms": "tidb",
+      "title": "Incorrect result for LEFT JOIN and CASE operator",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/19"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/19",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/19\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:2e11c000e8ad94b8d02b7bdc3950ee504b51d355d5dcdde7fff5d3781491cd95",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:2e11c000e8ad94b8d02b7bdc3950ee504b51d355d5dcdde7fff5d3781491cd95"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/19",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:9a5e1365dd4a",
+      "dbms": "tidb",
+      "title": "Incorrect result or run-time error after changing column type",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/10"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/10",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/10\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:0a000be1b748e7ff89f291206c55fdbe229e2f871f72bfdefd7f36f5763ce399",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:0a000be1b748e7ff89f291206c55fdbe229e2f871f72bfdefd7f36f5763ce399"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/10",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:6d205c5f0d09",
+      "dbms": "tidb",
+      "title": "Incorrect result when fetching from a view",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/8"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/8",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/8\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:8f8437611e9f91436b68b96278426a24a5df40f78f63f7a0b4874ce3f56b60ca",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:8f8437611e9f91436b68b96278426a24a5df40f78f63f7a0b4874ce3f56b60ca"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/8",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:f476a6bc0e47",
+      "dbms": "tidb",
+      "title": "Incorrect result when using SPACE() in a predicate",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/6"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/6",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/6\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:3e1de7cfafd579c9f0b437e1053f526dc4bb145830fa372e3fd8a6e044ba6f3e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3e1de7cfafd579c9f0b437e1053f526dc4bb145830fa372e3fd8a6e044ba6f3e"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/6",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:8b17ad48d027",
+      "dbms": "tidb",
+      "title": "Incorrect result when using the empty string as a predicate in a RIGHT JOIN",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/7"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/7",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/7\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:6e1aacdddafc0036230837f2eb3791f19a38abe2aac8da001b68bfa64ebf9870",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6e1aacdddafc0036230837f2eb3791f19a38abe2aac8da001b68bfa64ebf9870"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/7",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:f9d26b9d949d",
+      "dbms": "tidb",
+      "title": "Inserting into a partitioned table results in an \"Missing session variable when eval builtin\" error",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P2",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/18"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/18",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/18\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:73a27b9a710914f480b931ca5eccf2300d80dbe8afc1a956f668a397fae7107f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:73a27b9a710914f480b931ca5eccf2300d80dbe8afc1a956f668a397fae7107f"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/18",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:53c527d9e939",
+      "dbms": "tidb",
+      "title": "Internal error message when using CASE in partitioned table",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P2",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/16"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/16",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/16\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:c360a45b10c15e63a64af57bebd449ef1d6f2d669b3dd793ce6f84b6f20a177b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:c360a45b10c15e63a64af57bebd449ef1d6f2d669b3dd793ce6f84b6f20a177b"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/16",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:d288f02d8892",
+      "dbms": "tidb",
+      "title": "NATURAL LEFT JOIN results in incorrect result for <=> operator",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/5"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/5",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/5\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:eb26b18ffa2fb5d867316f1f7ec6df47b1991543674f897e79e8f8c4e0ce28bd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:eb26b18ffa2fb5d867316f1f7ec6df47b1991543674f897e79e8f8c4e0ce28bd"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/5",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:f9fa192ba9da",
+      "dbms": "tidb",
+      "title": "NATURAL LEFT JOIN with a NOT predicate results in a server panic",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/16419",
+        "fix": "https://github.com/pingcap/tidb/pull/16430"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/16419",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/16419\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16430\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:f0b82000b7a49e437252f8623f3a977c87f169830e6a52cf8f21dc510e7b91ba",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:f0b82000b7a49e437252f8623f3a977c87f169830e6a52cf8f21dc510e7b91ba"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/16419",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:c376c14fcbae",
+      "dbms": "tidb",
+      "title": "RIGHT JOIN with ELT() predicate returns an incorrect result",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/4"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/4",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/4\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:606a06001806b34854d70c832d997b0409c46672e867e921a71f638b4f1264cd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:606a06001806b34854d70c832d997b0409c46672e867e921a71f638b4f1264cd"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/4",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:3f7e35219fa4",
+      "dbms": "tidb",
+      "title": "Server panic for NATURAL LEFT JOIN on partitioned table",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/14"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/14",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/14\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:11352f3681611703d06c35aae8ca8a33d948d975a8ac2cbf5650026822422295",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:11352f3681611703d06c35aae8ca8a33d948d975a8ac2cbf5650026822422295"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/14",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:55cfbb83e302",
+      "dbms": "tidb",
+      "title": "Server panic for predicate t0.c0=9223372036854775808 when using partitions",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/9"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/9",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/9\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:e9f6a34af0798d95c67698bab5fa0796a9146a3f8b54b821c5eb353d20dbd604",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:e9f6a34af0798d95c67698bab5fa0796a9146a3f8b54b821c5eb353d20dbd604"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/9",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:584712171e27",
+      "dbms": "tidb",
+      "title": "Using the DEFAULT() function results in an ambiguous column error",
+      "reported_date": "2020-04-15",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/15"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/15",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"15/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/15\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:39d77c9819711c2ae350ee4b49ed0ede35af6e7ae6ad8016f1a27aa0d4ca6edc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:39d77c9819711c2ae350ee4b49ed0ede35af6e7ae6ad8016f1a27aa0d4ca6edc"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/15",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:2f0f7fb39555",
+      "dbms": "tidb",
+      "title": "Unexpected error \"Data Too Long\" when using a WHERE clause in an UPDATE",
+      "reported_date": "2020-04-19",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/43"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/43",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"19/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/43\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:397acfa5407640809824e3cd3d4a8701918fd729d3330cf4c3b9ef0b21be805e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:397acfa5407640809824e3cd3d4a8701918fd729d3330cf4c3b9ef0b21be805e"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/43",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:29409d2362f4",
+      "dbms": "tidb",
+      "title": "Incorrect result for LEFT JOIN AND NULLIF",
+      "reported_date": "2020-04-20",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/45"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/45",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"20/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/45\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:ef0df7960ce77ec75a8df15130db7324f9d05cf9c1ec11775622e6577674c244",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ef0df7960ce77ec75a8df15130db7324f9d05cf9c1ec11775622e6577674c244"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/45",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:c9f2377672e4",
+      "dbms": "tidb",
+      "title": "CREATE TABLE with generated column and escaped backslash causes a syntax error",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P2",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/53"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/53",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/53\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:da02859617317366ba8b76aef8853fae628e5ef146104e1a5de4a5f56e17002b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:da02859617317366ba8b76aef8853fae628e5ef146104e1a5de4a5f56e17002b"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/53",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:31489ca2a252",
+      "dbms": "tidb",
+      "title": "CREATE TABLE with generated column unexpectedly causes a syntax error",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P2",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/52"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/52",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/52\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:bab72b14de1831463fc699d0354057b38699f8b4c041b3d5272be70918fd953d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:bab72b14de1831463fc699d0354057b38699f8b4c041b3d5272be70918fd953d"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/52",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:5ed6c4ccb906",
+      "dbms": "tidb",
+      "title": "INL_HASH_JOIN hint causes an incorrect result for a table with a generated column",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/50"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/50",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/50\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:3d36812d78d6f1630200b29624025dacb9d1534fe5bae3765daab060d60f08e0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:3d36812d78d6f1630200b29624025dacb9d1534fe5bae3765daab060d60f08e0"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/50",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:4990a9ac58e8",
+      "dbms": "tidb",
+      "title": "INSERT IGNORE allows NULL value in a NOT NULL generated column",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/56"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/56",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/56\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:a06596f8860f079254e99467f3c4d08a6d5ce87fd50a2704318c481181a91151",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a06596f8860f079254e99467f3c4d08a6d5ce87fd50a2704318c481181a91151"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/56",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:1356c32e50ae",
+      "dbms": "tidb",
+      "title": "INSERT INTO with ON DUPLICATE clause results in an unexpected \"Miss column\" error",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/54"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/54",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/54\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:5f8a70fcce2145bf7145fc683da54301c1746b74b88cded5f9dc3123031ef3ff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5f8a70fcce2145bf7145fc683da54301c1746b74b88cded5f9dc3123031ef3ff"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/54",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:a032db6d559d",
+      "dbms": "tidb",
+      "title": "INSERT INTO with ON DUPLICATE clause results in an unexpected \"key not exist\" error",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/55"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/55",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/55\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:ed942a508c3aebc5e8343c483237bb4f805f6ad2b3086944f9745bffb8621633",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:ed942a508c3aebc5e8343c483237bb4f805f6ad2b3086944f9745bffb8621633"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/55",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:0ab99fea6d6b",
+      "dbms": "tidb",
+      "title": "Inf value in FLOAT column causes JDBC driver error",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/57"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/57",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/57\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:46ef0e1f8e935e5b680c62f0f880fa1a9b4e6081898a2a36dc6559c6d5c77e31",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:46ef0e1f8e935e5b680c62f0f880fa1a9b4e6081898a2a36dc6559c6d5c77e31"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/57",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:f6d5f2aa0bc4",
+      "dbms": "tidb",
+      "title": "SELECT with ORDER BY results in an \"inconsistent index\" error",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/58"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/58",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/58\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:6b70e7ea07e86d3ee42cdb934da37f2a58eaf27cc7a0bfc3d9be7bc80f5c632d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:6b70e7ea07e86d3ee42cdb934da37f2a58eaf27cc7a0bfc3d9be7bc80f5c632d"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/58",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:6fadc45aad8d",
+      "dbms": "tidb",
+      "title": "UNIQUE constraint on DECIMAL/floating-point columns causes incorrect result for NULL in AND",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P2",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/48"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/48",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/48\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:5135f7c569df2d5f999c0c55862f9ccacaf8b374b55dd66712c4e427a8c2d2f4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:5135f7c569df2d5f999c0c55862f9ccacaf8b374b55dd66712c4e427a8c2d2f4"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/48",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:1f5116d70a20",
+      "dbms": "tidb",
+      "title": "UNIQUE constraint on boolean column results in an incorrect result in a comparison",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/49"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/49",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/49\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:9a40f8e81f2fe85a0857e5529a7d4c0424c33d7f622a163c1e4f991bcb2aa8e6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:9a40f8e81f2fe85a0857e5529a7d4c0424c33d7f622a163c1e4f991bcb2aa8e6"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/49",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:213d27702d07",
+      "dbms": "tidb",
+      "title": "Using an index twice in an index hint results in an incorrect result",
+      "reported_date": "2020-04-22",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P2",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/47"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/47",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"22/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/47\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:94f95524eb86c9f28a39477bf89877ba083c9deb7ab2ea2d35d03fc0d8523fb8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:94f95524eb86c9f28a39477bf89877ba083c9deb7ab2ea2d35d03fc0d8523fb8"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/47",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:6ca052dd3be5",
+      "dbms": "tidb",
+      "title": "A predicate with COLLATE 'binary' results in a server panic \"invalid memory address or nil pointer dereference\"",
+      "reported_date": "2020-04-23",
+      "reported_year": 2020,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "links": {
+        "bugreport": "https://github.com/pingcap/tidb/issues/16779",
+        "fix": "https://github.com/pingcap/tidb/pull/16866"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/16779",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/pingcap/tidb/issues/16779\",\n            \"fix\": \"https://github.com/pingcap/tidb/pull/16866\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:269a01bdedecbeb29d01a7ac314c676838bbe0a668a7ec9e5366f091259232bd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:269a01bdedecbeb29d01a7ac314c676838bbe0a668a7ec9e5366f091259232bd"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/16779",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:f886f42a25bd",
+      "dbms": "tidb",
+      "title": "Fetching from a view with an escaped backslash results in an unexpected syntax error",
+      "reported_date": "2020-04-23",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "error",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/63"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/63",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/63\"\n        },\n        \"oracle\": \"error\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry for this report in the curated bugs.json.",
+            "content_sha256": "sha256:a161ee9e9f9799ac9bfcd59f3e153ce8f9814dd8b900c0b38060513b41a79524",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:a161ee9e9f9799ac9bfcd59f3e153ce8f9814dd8b900c0b38060513b41a79524"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/63",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:9a335af8d117",
+      "dbms": "tidb",
+      "title": "Value in generated column depends on a WHERE clause",
+      "reported_date": "2020-04-23",
+      "reported_year": 2020,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "logic",
+      "reporter": "Manuel Rigger",
+      "severity": "P1",
+      "links": {
+        "bugreport": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/62"
+      },
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/62",
+            "source_type": "github_issue",
+            "excerpt": null,
+            "note": "Upstream bug report or fix for this issue.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T06:21:44Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/README.md",
+            "source_type": "curated_bug_repository",
+            "excerpt": "the repository stores a list of bugs found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "SQLancer's own curated bug repository, which is the primary record of what the tool found.",
+            "content_sha256": "sha256:46a6a01477ee7a0b1e62878685526118673a2ef044d19b3a09f46f0eca8601ae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+            "source_type": "curated_bug_repository",
+            "excerpt": "\"date\": \"23/04/2020\",\n        \"dbms\": \"TiDB\",\n        \"links\": {\n            \"bugreport\": \"https://github.com/tidb-challenge-program/bug-hunting-issue/issues/62\"\n        },\n        \"oracle\": \"TLP (WHERE)\"",
+            "excerpt_is_verbatim": true,
+            "note": "Entry in bugs.json recording that the Ternary Logic Partitioning (TLP) oracle found this bug.",
+            "content_sha256": "sha256:fcdab38f8bb4a80f06de9f9611a20cbf086b0813dcc1c20253a93c9016faa8a5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "sqlancer_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T06:21:44Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/bugs/blob/master/bugs.json",
+        "source_type": "curated_bug_repository",
+        "content_sha256": "sha256:fcdab38f8bb4a80f06de9f9611a20cbf086b0813dcc1c20253a93c9016faa8a5"
+      },
+      "primary_url": "https://github.com/tidb-challenge-program/bug-hunting-issue/issues/62",
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:fc0624e53218",
+      "dbms": "tidb",
+      "title": "unexpected unresolved column error when the view refers to dual table",
+      "reported_date": "2022-06-20",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/35527"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/35527",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/35527",
+            "source_type": "github_issue",
+            "excerpt": "unexpected unresolved column error when the view refers to dual table",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:65953733482a40e73efe7d89cae7c045381d03a77146228eba881509fd589a6a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/35527",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:65953733482a40e73efe7d89cae7c045381d03a77146228eba881509fd589a6a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:3384412070ef",
+      "dbms": "tidb",
+      "title": "Runtime error: invalid memory address",
+      "reported_date": "2022-06-21",
+      "reported_year": 2022,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/35623"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/35623",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/35623",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT UNIQUE);\r\nCREATE VIEW v1(c0) AS SELECT 1 FROM t1;\r\n\r\nSELECT v1.c0 FROM v1 WHERE (true)LIKE(v1.c0); --runtime error: invalid memory address or nil pointer dereference",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:cb5cd01d6f3d23dc5f64a5d1d34893bbd78cbed607c0be01db653f1a714aff7a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:cb5cd01d6f3d23dc5f64a5d1d34893bbd78cbed607c0be01db653f1a714aff7a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:47c37d69e972",
+      "dbms": "tidb",
+      "title": "Unexpected Result with a FALSE Expression in WHERE",
+      "reported_date": "2022-06-22",
+      "reported_year": 2022,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/35645"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/35645",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/35645",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0 (c0 NUMERIC);\r\nCREATE INDEX i0 ON t0(c0);\r\nINSERT INTO t0(c0) VALUES (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a1c4ee680fadd4e8e1975cd8ee5c61f6d73b2406593dff2d087b74cfbca2b44a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a1c4ee680fadd4e8e1975cd8ee5c61f6d73b2406593dff2d087b74cfbca2b44a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:57d0a97e84e0",
+      "dbms": "tidb",
+      "title": "Inconsistent Results in SELECT",
+      "reported_date": "2022-08-03",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/36853"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/36853",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/36853",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT);\r\nCREATE VIEW v0(c0) AS SELECT 1 FROM t0, t1 WHERE 1;\r\nINSERT INTO t0 VALUES (2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9ccf66db586864a4566742062e1291feca701b7779e920fd64749362595935a1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9ccf66db586864a4566742062e1291feca701b7779e920fd64749362595935a1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:163c8623b838",
+      "dbms": "tidb",
+      "title": "Unexpected Result by CONCAT_WS",
+      "reported_date": "2022-08-04",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/36888"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/36888",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/36888",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT);\r\nINSERT INTO t0 VALUES (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:920d537e1a832b2c9bcde3208678862e9712d01ffbf461cb6de76c5c4a8f9703",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:920d537e1a832b2c9bcde3208678862e9712d01ffbf461cb6de76c5c4a8f9703"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:ced4b9c2ec71",
+      "dbms": "tidb",
+      "title": "Incorrect Result by `LEFT JOIN`",
+      "reported_date": "2022-10-06",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38304"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38304",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38304",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BLOB(5), c1 BLOB(5));\r\nCREATE TABLE t1 (c0 BOOL);\r\nINSERT INTO t1 VALUES(false);\r\nINSERT INTO t0(c0, c1) VALUES ('>', true);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:10426a734403de41f463f5f5243aa846c5f452852f226a773ba0ebb968876b7d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:10426a734403de41f463f5f5243aa846c5f452852f226a773ba0ebb968876b7d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:6e9614ab8a42",
+      "dbms": "tidb",
+      "title": "Incorrect Results by `REGEXP`",
+      "reported_date": "2022-10-06",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38303"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38303",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38303",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0(c0) VALUES (0.01);\r\nCREATE VIEW v0(c0) AS SELECT t0.c0 FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d02e0de69247ee279ecd6a8aafc3f6cb9debb027ba4f2ed7bb7ae2c23e66a2ba",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d02e0de69247ee279ecd6a8aafc3f6cb9debb027ba4f2ed7bb7ae2c23e66a2ba"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:b087c696ad64",
+      "dbms": "tidb",
+      "title": "runtime error: invalid memory address or nil pointer dereference",
+      "reported_date": "2022-10-06",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38305"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38305",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38305",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nDELETE FROM t0 LIMIT 122;\r\nALTER TABLE t0 ADD PRIMARY KEY(c0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ce06b2e32cc1379b93890e66299a5144b3e6165e9f641c67d6527d3f834c3b8a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ce06b2e32cc1379b93890e66299a5144b3e6165e9f641c67d6527d3f834c3b8a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:474ec1979140",
+      "dbms": "tidb",
+      "title": "Unexpected Results",
+      "reported_date": "2022-10-07",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38310"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38310",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38310",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL, PRIMARY KEY(c0));\r\nCREATE TABLE t1(c0 INT);\r\nINSERT INTO t0 VALUES (FALSE);\r\nINSERT INTO t1 VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:7da6c78ec442185ceb80a1cb1899f883bfa06aa8e53363045ea12f4352c975e1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:7da6c78ec442185ceb80a1cb1899f883bfa06aa8e53363045ea12f4352c975e1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:b0b9cb067eb7",
+      "dbms": "tidb",
+      "title": "Error [types:1690]%s value is out of range in '%s'",
+      "reported_date": "2022-10-10",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38352"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38352",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38352",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL, c1 BLOB(146) NOT NULL , c2 TEXT(438) );\r\nINSERT INTO t0 VALUES (1337129865, '\\\\', 1248373951);\r\nINSERT INTO t0(c0, c1) VALUES (1838451691, 498566332), (-1238701268, 1901594276) ON DUPLICATE KEY UPDATE c2=(CASE ((t0.c0)NOT REGEXP((('4') IS NULL))) WHEN '' THEN ((DEFAULT(t0.c0))>((CASE t0.c0 WHEN true THEN t0.c2 ELSE t0.c1 END ))) ELSE t0.c1 END );\r\nINSERT IGNORE  INTO t0 VALUES ('e', -1064475704, NULL), (-9223372036854775808, -1404202394, false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:e84eb7ec302afcc039aa6877c36c107c961d117f6e15d38ffb76c9e80e8cee5e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:e84eb7ec302afcc039aa6877c36c107c961d117f6e15d38ffb76c9e80e8cee5e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:d2b5ca354ab4",
+      "dbms": "tidb",
+      "title": "Question About the Estimated Rows in `GROUP BY`",
+      "reported_date": "2022-10-14",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38474"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38474",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38474",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\r\nINSERT INTO t0 VALUES(TRUE);\r\nANALYZE TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:88e3ab848e10e5836a7dee5f73cc6736c0729beea3b136bbff7ade8a124bbc8e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:88e3ab848e10e5836a7dee5f73cc6736c0729beea3b136bbff7ade8a124bbc8e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:99e7875a73df",
+      "dbms": "tidb",
+      "title": "Suspicious Estimated Rows by `JOIN`",
+      "reported_date": "2022-10-14",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38479"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38479",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38479",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nINSERT INTO t0 VALUES(1),(2),(3),(4),(5),(6),(7),(8),(9),(10);\r\nCREATE VIEW v0(c0, c1, c2) AS SELECT 'a', NULL, t0.c0 FROM t0 GROUP BY DEFAULT(t0.c0) HAVING SUM(1);\r\nANALYZE TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:71968a360688cbf3c218b81e9ad34f8f2c0dcfae951d62485e5690814d63f6b7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:71968a360688cbf3c218b81e9ad34f8f2c0dcfae951d62485e5690814d63f6b7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:239543bec582",
+      "dbms": "tidb",
+      "title": "Suspicious Estimated Rows by HAVING",
+      "reported_date": "2022-10-15",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38482"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38482",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38482",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR);\r\nCREATE TABLE t1(c0 BLOB(10));\r\nINSERT INTO t1(c0) VALUES ('a'),('b'),('c'),('d'),('e'),('f'),('g'),('h'),('i'),('j');\r\nINSERT INTO t0 VALUES ('k'),('l'),('m'),('n'),('o'),('p'),('q'),('r'),('s'),('t');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d037a149f94a6a7006651ce6c72b18c18fd6451b5601760374f6d1311456dcec",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d037a149f94a6a7006651ce6c72b18c18fd6451b5601760374f6d1311456dcec"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:fe91d9658f33",
+      "dbms": "tidb",
+      "title": "Unexpected Error: Failed to read auto-increment value from storage engine",
+      "reported_date": "2022-10-15",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38483"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38483",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38483",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 SMALLINT AUTO_INCREMENT PRIMARY KEY);\r\nINSERT IGNORE  INTO t0(c0) VALUES (194626268);\r\n\r\nINSERT IGNORE  INTO t0(c0) VALUES ('*'); -- Error: Failed to read auto-increment value from storage engine",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a9aa48cc32292cad0bfc78eb1a14ec4436442793539a1c514ab5f5e717e688ff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a9aa48cc32292cad0bfc78eb1a14ec4436442793539a1c514ab5f5e717e688ff"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:23976f964e9d",
+      "dbms": "tidb",
+      "title": "Unexpected Estimated Rows by INNER JOIN",
+      "reported_date": "2022-10-26",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38665"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38665",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38665",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR);\r\nCREATE TABLE t1(c0 INT);\r\nINSERT INTO t0 VALUES ('a'), ('b'), ('c'), ('d'), ('e'), ('f'), ('g'), ('h'), ('i'), ('j');\r\nINSERT INTO t1 VALUES (1), (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ab88649e05956d6497573f56e136825f54e92cf30c9eb198bcb317fdffe71651",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ab88649e05956d6497573f56e136825f54e92cf30c9eb198bcb317fdffe71651"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:28a31625d27e",
+      "dbms": "tidb",
+      "title": "Unexpected Results by RIGHT JOIN",
+      "reported_date": "2022-10-26",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38654"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38654",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38654",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT);\r\nINSERT INTO t1 VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:08411ef2619b8cd4de8ea991109dca6e6af285a77db9bdd9c5ff7cc68361984c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:08411ef2619b8cd4de8ea991109dca6e6af285a77db9bdd9c5ff7cc68361984c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:a08d1db354bd",
+      "dbms": "tidb",
+      "title": "Unexpected Estimated Rows by WHERE clause",
+      "reported_date": "2022-10-28",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38721"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38721",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38721",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE VIEW v0(c1, c2) AS SELECT t0.c0, DATE_FORMAT('2022-10-28 16:23:00', '%W %M %Y') FROM t0 WHERE t0.c0;\r\nINSERT INTO t0 VALUES (0), (1), (2), (3), (4), (5), (6), (7), (8), (9);\r\nANALYZE TABLE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ea799e2e8b0227566292ac600022bd6d5f18ad014ba89b629d18fc0d2cb41be0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ea799e2e8b0227566292ac600022bd6d5f18ad014ba89b629d18fc0d2cb41be0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:fb52edab58ef",
+      "dbms": "tidb",
+      "title": "rule PredicatePushDown pushes wrong filter across projection",
+      "reported_date": "2022-10-29",
+      "reported_year": 2022,
+      "status": "unknown",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/38736"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/38736",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/38736",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL, c1 INT);\r\nCREATE TABLE t1 LIKE t0;\r\nCREATE VIEW v0(c0) AS SELECT IS_IPV4(t0.c1) FROM t0, t1;\r\nINSERT INTO t0(c0, c1) VALUES (true, 0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by bajinsheng of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:027e400efbc58fb628b1a1bbcbedd76e264e7c38cd1972c91812b75ad525bdd5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by bajinsheng.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:027e400efbc58fb628b1a1bbcbedd76e264e7c38cd1972c91812b75ad525bdd5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:a799255c99b2",
+      "dbms": "tidb",
+      "title": "Unexpected error in `ORDER BY`",
+      "reported_date": "2023-04-11",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/42941"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/42941",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/42941",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL, c1 FLOAT, c2 DOUBLE);\r\nCREATE TABLE t1 LIKE t0;\r\nSELECT t1.c1 AS c0 FROM  t0 NATURAL LEFT JOIN t1  WHERE t0.c2 ORDER BY (CASE (CASE NULL WHEN t1.c1 THEN t0.c1 ELSE DEFAULT(t1.c0) END ) WHEN t1.c0 THEN 1 ELSE t0.c2 END );\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:7930a5e8fc13f97f197b713c44ffafa92ebcde5cac10d3694ab5d4b527e5d0eb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:7930a5e8fc13f97f197b713c44ffafa92ebcde5cac10d3694ab5d4b527e5d0eb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:e556205bcca9",
+      "dbms": "tidb",
+      "title": "Unexpected error message on constant",
+      "reported_date": "2023-04-11",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/42942"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/42942",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/42942",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 FLOAT );\r\nINSERT INTO t0(c0) VALUES (1);\r\nUPDATE t0 SET c0=1;\r\nALTER TABLE t0 MODIFY c0 TINYINT;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:477c8d45ac4407134f027fbf8a7e21dd2c5f336bbe59266fa05024be5ab7d076",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:477c8d45ac4407134f027fbf8a7e21dd2c5f336bbe59266fa05024be5ab7d076"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:e383c6235419",
+      "dbms": "tidb",
+      "title": "Unexpected result of subquery",
+      "reported_date": "2023-04-11",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/42912"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/42912",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/42912",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 TEXT(328) );\r\n\r\nCREATE VIEW v0(c0) AS SELECT 'c' FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:0a2f1372f361c9dd9e849ca098e6a91148150c79ef5dc1b91e8c88b558d031f8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:0a2f1372f361c9dd9e849ca098e6a91148150c79ef5dc1b91e8c88b558d031f8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:e80c33fca582",
+      "dbms": "tidb",
+      "title": "An expression has two different value in two queries",
+      "reported_date": "2023-04-13",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/43026"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/43026",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/43026",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL );\r\n\r\nINSERT INTO t0(c0) VALUES (true), (0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9755fcb31f9a45a3bff3de9ebbbe3025b62a9d2197087ba94f7a4a57b47dbdf9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9755fcb31f9a45a3bff3de9ebbbe3025b62a9d2197087ba94f7a4a57b47dbdf9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:b9d90ab7c7b8",
+      "dbms": "tidb",
+      "title": "`runtime error: index out of range [1] with length 1` on `SELECT`",
+      "reported_date": "2023-04-20",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/43256"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/43256",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/43256",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL );\r\nREPLACE INTO t0 VALUES (0.4117160754744159);\r\nCREATE VIEW v0(c0) AS SELECT NULL FROM t0 WHERE t0.c0;",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b932571dc76bfdbf7c4eeab7c2f70dc2142f2892083177b35aeaaaa8d5b5aef8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b932571dc76bfdbf7c4eeab7c2f70dc2142f2892083177b35aeaaaa8d5b5aef8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:ac5d21901fa3",
+      "dbms": "tidb",
+      "title": "Unexpected results of subquery in `INSERT`",
+      "reported_date": "2023-04-24",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/43373"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/43373",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/43373",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL, c1 BIGINT NOT NULL , c2 INTEGER  AUTO_INCREMENT );\r\n\r\nINSERT INTO t0(c0, c1) VALUES (1, 1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:bfacd50fcdf80f2d8f7881589e8a08d3cb3f0cacddc147f16a9c131ea9adadf3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:bfacd50fcdf80f2d8f7881589e8a08d3cb3f0cacddc147f16a9c131ea9adadf3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:325b345f8cf8",
+      "dbms": "tidb",
+      "title": "Unexpected error `Unknown column 't1.c0' in 'where clause'`",
+      "reported_date": "2023-05-06",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/43569"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/43569",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/43569",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR);\r\nCREATE TABLE t1(c0 CHAR);\r\nSELECT COUNT(t1.c0) AS c0 FROM  t0 NATURAL JOIN t1  WHERE 'Q';  -- Unknown column 't1.c0' in 'where clause'\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ce4bade6c6ba188c838d8392f5c9c21633816cc3547cf1cc17383d4c9a40ce16",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ce4bade6c6ba188c838d8392f5c9c21633816cc3547cf1cc17383d4c9a40ce16"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:4cf5f9d084b3",
+      "dbms": "tidb",
+      "title": "Unexpected results when `CASE` and `IN` use together",
+      "reported_date": "2023-05-08",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/43624"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/43624",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/43624",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\r\n\r\nINSERT INTO t0 VALUES (true);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:363f4cfc3da9a3804b6c64cc9b7f8c24bd81325ee92a66b669d4181b4165d83d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:363f4cfc3da9a3804b6c64cc9b7f8c24bd81325ee92a66b669d4181b4165d83d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:a63ac9f53a0b",
+      "dbms": "tidb",
+      "title": "Unexpected error message `Data truncated for column '%s' at row %d`",
+      "reported_date": "2023-05-17",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/43905"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/43905",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/43905",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL);\r\nINSERT INTO t0(c0) VALUES (1);\r\nDELETE FROM t0 WHERE (t0.c0)=('1p');\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:79daca74b7aff12283b42d5b2ff3034ac033e360a42a7598ffbeb25fde67f178",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/43905",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:79daca74b7aff12283b42d5b2ff3034ac033e360a42a7598ffbeb25fde67f178"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:d817cfe41f74",
+      "dbms": "tidb",
+      "title": "Unexpected error when `CHECK` a predicate that is always true.",
+      "reported_date": "2023-06-15",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/44689"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/44689",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/44689",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 NUMERIC CHECK (true));\r\n```\r\n\r\nThis is the error message:",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ac4ed73e771c8cae22b3e7f08b0a4a5d8914ceab012d768a077f659fed78f042",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ac4ed73e771c8cae22b3e7f08b0a4a5d8914ceab012d768a077f659fed78f042"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:2323019d840c",
+      "dbms": "tidb",
+      "title": "Unexpected results of `ANY` operator",
+      "reported_date": "2023-06-15",
+      "reported_year": 2023,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Chi Zhang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/44706"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/44706",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/44706",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c2 BIGINT );\r\nINSERT INTO t0(c2) VALUES (1);\r\n\r\nSELECT MIN(t0.c2) FROM t0 WHERE false; -- NULL",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Chi Zhang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:449ccce42f0f70fd5afdec72bbdaf10c786f0606cb4fc7125f8d3dc1a4723cff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Chi Zhang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:449ccce42f0f70fd5afdec72bbdaf10c786f0606cb4fc7125f8d3dc1a4723cff"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:c1fa2f05f6d3",
+      "dbms": "tidb",
+      "title": "Unexpected Results of IN expression With NATURAL RIGHT JOIN",
+      "reported_date": "2023-12-14",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/49476"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/49476",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/49476",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 BOOLEAN);\r\nINSERT INTO t0 (c0) VALUES (0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:ba319adab8122e3c6271d5cd501b5be00a86345415685acfe8bc3f2410b3db8a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:ba319adab8122e3c6271d5cd501b5be00a86345415685acfe8bc3f2410b3db8a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:fdc4b2cc9ac2",
+      "dbms": "tidb",
+      "title": "Runtime error when using overflow integers",
+      "reported_date": "2024-01-17",
+      "reported_year": 2024,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/50489"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/50489",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/50489",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 INT);\r\nCREATE VIEW v0(c0) AS SELECT 142805906 FROM t0 ;\r\n\r\nSELECT * FROM v0 WHERE (v0.c0 NOT IN (v0.c0, (CASE ((-9223372036854775808)+(-1582704113)) WHEN v0.c0 THEN v0.c0 END )));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b2bfcc588d2bc3db12bfebc5ac70d6e6179cc330cb57254f20096b08b2d76a35",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b2bfcc588d2bc3db12bfebc5ac70d6e6179cc330cb57254f20096b08b2d76a35"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:7a8f4a188d22",
+      "dbms": "tidb",
+      "title": "Where clause on right outer join might lead to rows not being fetched",
+      "reported_date": "2025-01-23",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/59162"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/59162",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/59162",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nINSERT INTO t0 VALUES (1);\nCREATE TABLE t1(c0 INT);\nINSERT INTO t1(c0) VALUES (0);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:026d10bb0555deb8108f523db26cfe4d1e50738766e78a9b8596831f5ee2082e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/59162",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:026d10bb0555deb8108f523db26cfe4d1e50738766e78a9b8596831f5ee2082e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:f12e4ea8c9a1",
+      "dbms": "tidb",
+      "title": "Null-safe equals operator in WHERE clause might under-fetch rows",
+      "reported_date": "2025-01-27",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/59220"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/59220",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/59220",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\nINSERT INTO t0 (c0) VALUES (NULL);\nCREATE TABLE t1(c0 BOOL, c1 BOOL);\nINSERT INTO t1 (c0, c1) VALUES (NULL, true);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4a520e54a7b0a2a330e58c1e225667aed41955ea5b8e27e8d43aad98b91ca108",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/59220",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4a520e54a7b0a2a330e58c1e225667aed41955ea5b8e27e8d43aad98b91ca108"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:2d6f1c9d3ff3",
+      "dbms": "tidb",
+      "title": "Cast double as datetime bug and adding expression causes different result",
+      "reported_date": "2025-03-14",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60095"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60095",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60095",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 double, c2 BOOL , PRIMARY KEY(c2));\n\nCREATE TABLE t1 LIKE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a7462e7c8f17126b1861ffb288d68b39a30fb296cfdfa7d68ce9ef6fa916a840",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/60095",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a7462e7c8f17126b1861ffb288d68b39a30fb296cfdfa7d68ce9ef6fa916a840"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:5b60708f77fe",
+      "dbms": "tidb",
+      "title": "Incorrect Join Result After adding an additional true expression",
+      "reported_date": "2025-03-14",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60087"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60087",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60087",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 TEXT(234) );\nCREATE TABLE t1 LIKE t0;\n\nINSERT INTO t1(c0) VALUES ('P');",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d40933bb0e8b21ebba62bafae2db8c05c6b904df0566c384f5a1c55f53209bf6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/60087",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d40933bb0e8b21ebba62bafae2db8c05c6b904df0566c384f5a1c55f53209bf6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:861c136af16c",
+      "dbms": "tidb",
+      "title": "Join Involving View and ATAN2 function produce wrong result",
+      "reported_date": "2025-03-14",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60093"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60093",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60093",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 int);\n\nCREATE TABLE t1(c0 DECIMAL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:5617013ffed047eb8e4c3d1773c8d5100b6eed0496052340f098d558a8ff215d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:5617013ffed047eb8e4c3d1773c8d5100b6eed0496052340f098d558a8ff215d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:94167caf403f",
+      "dbms": "tidb",
+      "title": "Behaviour of expression index is incompatible with MySQL",
+      "reported_date": "2025-03-17",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "EmilyOng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60101"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60101",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60101",
+            "source_type": "github_issue",
+            "excerpt": "Behaviour of expression index is incompatible with MySQL",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:c5229f36a22bb50de5d29b626686e69e4564f82a7c0c69910fd1c28c177db38c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "EmilyOng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/60101",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c5229f36a22bb50de5d29b626686e69e4564f82a7c0c69910fd1c28c177db38c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:a9c565691256",
+      "dbms": "tidb",
+      "title": "Discrepant Join Result on the same-value join condition",
+      "reported_date": "2025-03-17",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60127"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60127",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60127",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 TEXT(5) );\nCREATE TABLE t1(c0 TEXT(5) );",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ee162d952a804c99edd7d62df291fb2f55b9dc5067f97cda0bf4bc1cefb36f4c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/60127",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ee162d952a804c99edd7d62df291fb2f55b9dc5067f97cda0bf4bc1cefb36f4c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:00214f5dfa28",
+      "dbms": "tidb",
+      "title": "Incorrect Join Result under the condition containing date_format, REGEXP and additional true expression",
+      "reported_date": "2025-03-17",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60126"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60126",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60126",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\nCREATE TABLE t1 LIKE t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:436d00b7a8dfd0e6ac622a168254ca3fcd8888f8986693a3bef20fd34066add8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/60126",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:436d00b7a8dfd0e6ac622a168254ca3fcd8888f8986693a3bef20fd34066add8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:48cdf738c7e3",
+      "dbms": "tidb",
+      "title": "Incorrect Join result in the join condition containing Cast operator and additional true expression",
+      "reported_date": "2025-03-17",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60111"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60111",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60111",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\nCREATE TABLE t1(c0 DOUBLE);\n\nINSERT INTO t1 VALUES (0.5);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5340f478d0f37323c968e6f2dd301af809cb32e61a57601cb548717c8146dc5e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/60111",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5340f478d0f37323c968e6f2dd301af809cb32e61a57601cb548717c8146dc5e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:8c0f31ae5147",
+      "dbms": "tidb",
+      "title": "Wrong Join Result under the condition of <=> and NULL",
+      "reported_date": "2025-03-17",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60110"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60110",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60110",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 int);\nCREATE TABLE t1 LIKE t0;\nINSERT INTO t1 VALUES (NULL);\nREPLACE INTO t1 VALUES (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:07e150fefa53887e32bcdc7d72c3accb5eb4c42765a1317ba144c72ce8af06f6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/60110",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:07e150fefa53887e32bcdc7d72c3accb5eb4c42765a1317ba144c72ce8af06f6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:36671c319d45",
+      "dbms": "tidb",
+      "title": "Wrong Inner Join Result",
+      "reported_date": "2025-03-21",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60217"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60217",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60217",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c1 DOUBLE , c2 BOOL);\n\n\nINSERT INTO t1 VALUES (0.5, false);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:415fae08ce3a44a9385ec7adfbd29772e9e2a03d999eef9afb8326d5c09da6d3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:415fae08ce3a44a9385ec7adfbd29772e9e2a03d999eef9afb8326d5c09da6d3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:7497c299155f",
+      "dbms": "tidb",
+      "title": "Unexpected INNER JOIN Result",
+      "reported_date": "2025-03-28",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60322"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60322",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60322",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL );\nCREATE TABLE t1(c0 CHAR );\n\nINSERT INTO t1 VALUES (NULL);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:a328f304a279ad54c416b8ea537db7b8c24fb807b7d986fbc34167ffc51db84a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:a328f304a279ad54c416b8ea537db7b8c24fb807b7d986fbc34167ffc51db84a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:f96181ff3067",
+      "dbms": "tidb",
+      "title": "join with view produces 'interface conversion' with a small change on join condition",
+      "reported_date": "2025-04-10",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60489"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60489",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60489",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR);\nCREATE VIEW v0(c0) AS SELECT NULL;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7a74c71ee4b7a0f6e81f66b8aa3a6b755e9697a0c5e09cfdd94fef2d5764bc18",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/60489",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7a74c71ee4b7a0f6e81f66b8aa3a6b755e9697a0c5e09cfdd94fef2d5764bc18"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:fd87a8bad7c8",
+      "dbms": "tidb",
+      "title": "Internal Error of PI() DIV FLOOR() Expression",
+      "reported_date": "2025-04-16",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "TheoristCoder",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60610"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60610",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60610",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 BLOB(24));\n\nINSERT IGNORE  INTO t0 VALUES ('-1e500');",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:fdbabc36dc913173444b078cc48422ca5862089160e1ae2f68b88174a7c6e4aa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/60610",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fdbabc36dc913173444b078cc48422ca5862089160e1ae2f68b88174a7c6e4aa"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:f2b7274180a0",
+      "dbms": "tidb",
+      "title": "Join Crash When dealing with inner join involving subquery and like operator",
+      "reported_date": "2025-04-17",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60625"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60625",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60625",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT );\n\nCREATE  VIEW v0(c0) AS SELECT NULL AS col_0 FROM t1 WHERE ((t1.c0));",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:6eaf9b92e18fe5eee35a0db9450e93e7e2e44f04de86e721ecfd8fad546515b5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:6eaf9b92e18fe5eee35a0db9450e93e7e2e44f04de86e721ecfd8fad546515b5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:e6a981374c34",
+      "dbms": "tidb",
+      "title": "Serious Wrong Result of Right Join involving subquery",
+      "reported_date": "2025-04-18",
+      "reported_year": 2025,
+      "status": "verified",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60659"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60659",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60659",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DOUBLE);\nCREATE TABLE t1(c1 DOUBLE);\n\nINSERT INTO t0(c0) VALUES (0.6);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:51c5a24b6f2dcb1049e1cc3c057b0979a82aaa06b77f54fe4632c6ef2d177b38",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:51c5a24b6f2dcb1049e1cc3c057b0979a82aaa06b77f54fe4632c6ef2d177b38"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:e57bd8cfa020",
+      "dbms": "tidb",
+      "title": "Unexpected Result for NATURAL JOIN in TiFlash",
+      "reported_date": "2025-05-01",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/60958"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/60958",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/60958",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL ZEROFILL);\nCREATE TABLE t1 (c0 BLOB);\nINSERT IGNORE INTO t0 VALUES(-1749300172);\nINSERT INTO t1 VALUES('aa'), ('-0.0');",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:03243a1362a97d6dbc4367a1e5601bee6dc2bc1cc9e90589f6f65ccfdfe25074",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/60958",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:03243a1362a97d6dbc4367a1e5601bee6dc2bc1cc9e90589f6f65ccfdfe25074"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:c5d2598287f6",
+      "dbms": "tidb",
+      "title": "Unexpected Inner Join result",
+      "reported_date": "2025-05-25",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/61306"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/61306",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/61306",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT);\nCREATE TABLE t3(c0 INT );",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:020bb8e474a2790be95a84d6f6f6432d74411174a13a70ef60c23d5d23863272",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:020bb8e474a2790be95a84d6f6f6432d74411174a13a70ef60c23d5d23863272"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:ede2f857eee1",
+      "dbms": "tidb",
+      "title": "Unexpected Multiple Join Result",
+      "reported_date": "2025-05-26",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/61327"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/61327",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/61327",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t2(c0 INT);\nCREATE TABLE t3(c0 INT);\nINSERT INTO t0 VALUES(0);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Zhaokun Xiang of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:4318d5d216070feea4c01afc145fdd8451d5f99afa09b0bf6754a1d3a379e475",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Zhaokun Xiang.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:4318d5d216070feea4c01afc145fdd8451d5f99afa09b0bf6754a1d3a379e475"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:6b5bd9018072",
+      "dbms": "tidb",
+      "title": "Unexpected result",
+      "reported_date": "2025-07-08",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/62269"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/62269",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/62269",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c2 BOOL);\nINSERT INTO t0(c2) VALUES (false), (true);\nSELECT t0.c2 FROM t0 GROUP BY t0.c2 HAVING BIT_AND(BINARY (- (-1)));   -- 0, 1\nSET @a = -1;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0e152d8841ed807e96ad650b787598ca840cfb1dc5f86800b4be1c3e233e2741",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/62269",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0e152d8841ed807e96ad650b787598ca840cfb1dc5f86800b4be1c3e233e2741"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:69720e49a0b7",
+      "dbms": "tidb",
+      "title": "`CASE WHEN` returns different result in prepared statement and normal query",
+      "reported_date": "2025-07-23",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/62564"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/62564",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/62564",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL);\nINSERT INTO t0 VALUES (0);\nSELECT * FROM t0 WHERE CAST((CASE t0.c0 WHEN t0.c0 THEN CAST(t0.c0 AS TIME) ELSE NULL END ) AS DATE);\nSET @b = NULL;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6b8995d9291063a070ad327645a0aacbc46e692d3799a254f100ea2bbf50e653",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/62564",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6b8995d9291063a070ad327645a0aacbc46e692d3799a254f100ea2bbf50e653"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:41ce36873c36",
+      "dbms": "tidb",
+      "title": "`GROUP BY` returns different results in prepared statement and normal SELECT query",
+      "reported_date": "2025-07-24",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/62606"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/62606",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/62606",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL PRIMARY KEY);\nINSERT INTO t0 VALUES (true), (false);\nSELECT count(*) FROM t0 WHERE 1 GROUP BY DATE_FORMAT(t0.c0, t0.c0);\nSET @b = 1;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:be8bcabf0061ee614ade0c9a4bb56ca1b764c9880434dfe9be9e7da1be3f21fe",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/62606",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:be8bcabf0061ee614ade0c9a4bb56ca1b764c9880434dfe9be9e7da1be3f21fe"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:cee2c84bd46a",
+      "dbms": "tidb",
+      "title": "`CASE WHEN` in prepared statement return wrong result on FLOAT value",
+      "reported_date": "2025-07-30",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/62717"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/62717",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/62717",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 FLOAT);\nREPLACE INTO t0(c0) VALUES (100000030);\nSET @c = false;\nSELECT t0.c0 FROM t0 WHERE ((t0.c0)NOT LIKE(CASE WHEN false THEN false ELSE t0.c0 END ));",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:31519f5bfeaa1ce7e07397e54b49c792c59f4cf0a8a98c3b3e5b3a94951622b4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/62717",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:31519f5bfeaa1ce7e07397e54b49c792c59f4cf0a8a98c3b3e5b3a94951622b4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:0d72f5610c00",
+      "dbms": "tidb",
+      "title": "Integer was cast to wrong char value in prepared statement.",
+      "reported_date": "2025-08-01",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/62772"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/62772",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/62772",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR );\nINSERT IGNORE  INTO t0 VALUES ('7');\nSET @b = 'M';\nSELECT CAST((CASE false WHEN CAST('M' AS CHAR) THEN -1 ELSE ( (t0.c0)) END ) AS CHAR) FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9b01dc5dd89a5e40e8dd64c2b44fa3be63d73ae412ae00598c4d6122cf003c7d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/62772",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9b01dc5dd89a5e40e8dd64c2b44fa3be63d73ae412ae00598c4d6122cf003c7d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:8ef0b1454a04",
+      "dbms": "tidb",
+      "title": "Inconsistency between normal select and prepared statement query",
+      "reported_date": "2025-09-21",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63640"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63640",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63640",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL);\nINSERT IGNORE  INTO t0 VALUES (1);\nSELECT 2 FROM t0 WHERE t0.c0 GROUP BY 1; -- 2\nSET @a = 2;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e72efddbc1f5e5fe4f3e952e9d5d503e2b80e3e1a8f36e574233babd405f5120",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63640",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e72efddbc1f5e5fe4f3e952e9d5d503e2b80e3e1a8f36e574233babd405f5120"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:f121a6c78b67",
+      "dbms": "tidb",
+      "title": "Inconsistent behavior of `~` in normal query and prepared statement",
+      "reported_date": "2025-09-21",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63646"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63646",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63646",
+            "source_type": "github_issue",
+            "excerpt": "Inconsistent behavior of `~` in normal query and prepared statement",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:7486c27aac5326d4862213025629b4cd223f88e3211b7864934efed5ac265e5e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63646",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7486c27aac5326d4862213025629b4cd223f88e3211b7864934efed5ac265e5e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:f1cfd192ec8d",
+      "dbms": "tidb",
+      "title": "Incorrect type cast in prepared statement",
+      "reported_date": "2025-09-21",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63644"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63644",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63644",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DOUBLE);\nINSERT INTO t0(c0) VALUES (0);\nSELECT ((INET6_ATON((t0.c0)))OR('a')) FROM t0;  -- NULL\nSET @a = 'a';",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8b703dcea84e5c2c01aa6ea0f3339f9ccd536bef37f6d932aeb6e345f3deceff",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63644",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8b703dcea84e5c2c01aa6ea0f3339f9ccd536bef37f6d932aeb6e345f3deceff"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:2f7359459b40",
+      "dbms": "tidb",
+      "title": "Not well handle \\ in REGEXP function",
+      "reported_date": "2025-09-21",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63641"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63641",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63641",
+            "source_type": "github_issue",
+            "excerpt": "Not well handle \\ in REGEXP function",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:007a806e77284f9824948a3e48187a5600d09a74aec612400a40f3d11eeb8eb0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63641",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:007a806e77284f9824948a3e48187a5600d09a74aec612400a40f3d11eeb8eb0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:fa10df0c2778",
+      "dbms": "tidb",
+      "title": "Unexpected error `baseBuiltinFunc.vecEvalInt() should never be called, please contact the TiDB team for help`",
+      "reported_date": "2025-09-21",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63647"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63647",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63647",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\nINSERT IGNORE  INTO t0 VALUES (false);\nSET @b = NULL;\nSET @c = NULL;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ed3bc2601832ad591052712bc016dea6491fc5c6e0114dbaaa503060dabd9d63",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63647",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ed3bc2601832ad591052712bc016dea6491fc5c6e0114dbaaa503060dabd9d63"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:0d98e6d97b37",
+      "dbms": "tidb",
+      "title": "Unexpected result of IFNULL and LIKE functions",
+      "reported_date": "2025-09-21",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63642"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63642",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63642",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL NOT NULL);\nINSERT INTO t0 VALUES (-1324907974);\nSELECT ((IFNULL(t0.c0, 'k'))LIKE(t0.c0)) FROM t0;  -- 0\nSET @c = 'k';",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:687f86c6b9255e6ade6b7d69bd35ed27441ac9e52fe5b29031d2cdb7600512b0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63642",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:687f86c6b9255e6ade6b7d69bd35ed27441ac9e52fe5b29031d2cdb7600512b0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:528bd12f39f3",
+      "dbms": "tidb",
+      "title": "Unexpected result when use HEX as the WHERE condition of prepared statement",
+      "reported_date": "2025-09-22",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63649"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63649",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63649",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL);\nINSERT t0 VALUES (1);\nSET @b = '-1';",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ef4b5d7389f37ef62de5cd0fc57d8c7760b0771fefe4169236fa036c1d3272ea",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63649",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ef4b5d7389f37ef62de5cd0fc57d8c7760b0771fefe4169236fa036c1d3272ea"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:471446fa8bb3",
+      "dbms": "tidb",
+      "title": "Inconsistent behaviour of `\\\\` in normal query and prepared statement",
+      "reported_date": "2025-09-23",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63680"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63680",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63680",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR(100));\nINSERT INTO t0 VALUES ('A');\nSELECT (('\\\\6' IS TRUE) NOT IN (t0.c0, CAST(449007834 AS DECIMAL))) FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:857faae8ce2308a0000afc8e4a7cb07fad248d5086f0b632921a84e131fbc518",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63680",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:857faae8ce2308a0000afc8e4a7cb07fad248d5086f0b632921a84e131fbc518"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:0b65a6b3bc68",
+      "dbms": "tidb",
+      "title": "Incorrect type cast in prepared statement",
+      "reported_date": "2025-09-23",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63678"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63678",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63678",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR);\nINSERT  INTO t0 VALUES ('6');\nSELECT CAST(SUBSTRING_INDEX(0.4, 'a', t0.c0) AS BINARY) FROM t0;  -- 0.4\nSET @c = 0.4;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5ca443f2f0ca11c0ce596350db8c4e031bf5969753d9d51d80f9ebbb72b394bb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63678",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5ca443f2f0ca11c0ce596350db8c4e031bf5969753d9d51d80f9ebbb72b394bb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:18f9f427b942",
+      "dbms": "tidb",
+      "title": "Unexpected result of `<=>` in prepared statement",
+      "reported_date": "2025-09-23",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63684"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63684",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63684",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected result of `<=>` in prepared statement",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:5284db2de42fbd6c593554e15681cf314b3f53e4bdac0ef1e336acacb9beb807",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63684",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5284db2de42fbd6c593554e15681cf314b3f53e4bdac0ef1e336acacb9beb807"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:e2ee0a6af871",
+      "dbms": "tidb",
+      "title": "Incorrect result of ELT when used in prepared statement.",
+      "reported_date": "2025-10-10",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63900"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63900",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63900",
+            "source_type": "github_issue",
+            "excerpt": "Incorrect result of ELT when used in prepared statement.",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:96fc71596fac90e16c203e7a5638a02c9ddb5ccea2f1fd844167748d781c3478",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63900",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:96fc71596fac90e16c203e7a5638a02c9ddb5ccea2f1fd844167748d781c3478"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:17c98e5829d4",
+      "dbms": "tidb",
+      "title": "Unexpected error `%s value is out of range in '%s'`",
+      "reported_date": "2025-10-10",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63899"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63899",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63899",
+            "source_type": "github_issue",
+            "excerpt": "Unexpected error `%s value is out of range in '%s'`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:b1ff9136aee7f89a0cc5bbc64b66caea8646be9a818e251797e2905b855ce2ed",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63899",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b1ff9136aee7f89a0cc5bbc64b66caea8646be9a818e251797e2905b855ce2ed"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:21b22ab5f877",
+      "dbms": "tidb",
+      "title": "Unexpected error `Can't find column Column#3 in schema Column` when construct prepared statement for EXPLAIN",
+      "reported_date": "2025-10-10",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63898"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63898",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63898",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR );\nCREATE VIEW v0(c0) AS SELECT ((FIELD(t0.c0, 1))!=(false)) FROM t0;\nPREPARE prepare_query FROM 'EXPLAIN SELECT t0.c0 FROM v0, t0 WHERE ((t0.c0)=(REPLACE(?, v0.c0, ?)))';\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0766853e42d447572d980f2fa44b91b58a6e9704653aefe18224f7a824e55ff6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63898",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0766853e42d447572d980f2fa44b91b58a6e9704653aefe18224f7a824e55ff6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:58df23af5e0b",
+      "dbms": "tidb",
+      "title": "Unexpected error `interface conversion: expression.Expression is *expression.Constant, not *expression.ScalarFunction`",
+      "reported_date": "2025-10-12",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/63914"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/63914",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/63914",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DOUBLE );\nCREATE TABLE t1 LIKE t0;\nCREATE INDEX i0 ON t1(c0 ASC);\nALTER TABLE t1  CHANGE c0 c0 DOUBLE NOT NULL ;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c8a52115f0dda51c1800b590d6f5a67bb16827074b2cb3827e6bcf57a34576cc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/63914",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c8a52115f0dda51c1800b590d6f5a67bb16827074b2cb3827e6bcf57a34576cc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:60d9aa210ad3",
+      "dbms": "tidb",
+      "title": "runtime error: index out of range [7] with length 4",
+      "reported_date": "2025-12-05",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/64886"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/64886",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/64886",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 FLOAT UNSIGNED ZEROFILL , PRIMARY KEY(c0));\nINSERT INTO t1(c0) VALUES (0.3);\nSET @a = 602065819;\nPREPARE prepare_query FROM 'SELECT IF(?, t1.c0, t1.c0) FROM t1 WHERE t1.c0 GROUP BY true';",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:419a1c27f4e56c42cf2bb366ddb5482d6beb996a60887aeafb4d118678819e30",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/64886",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:419a1c27f4e56c42cf2bb366ddb5482d6beb996a60887aeafb4d118678819e30"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:750080ce631f",
+      "dbms": "tidb",
+      "title": "Unexpected error \"Column 't1.c0' in field list is ambiguous\"",
+      "reported_date": "2025-12-29",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65325"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65325",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65325",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\nCREATE TABLE t1(c0 DOUBLE);\nSELECT t1.c0, t1.c0 FROM  t0 NATURAL JOIN t1 ORDER BY CASE DEFAULT(t1.c0) WHEN t1.c0 THEN 397344251 ELSE t0.c0 END;\n-- ERROR 1052 (23000) at line 7: Column 't1.c0' in field list is ambiguous",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0523b4d0ddfc4d5ab42046791ef7bb581e66ce41f1d6141856349ac6ee23d711",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65325",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0523b4d0ddfc4d5ab42046791ef7bb581e66ce41f1d6141856349ac6ee23d711"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:6d53002da9f9",
+      "dbms": "tidb",
+      "title": "Unexpected error \"Incorrect string value: '0.8949238218722565' for function inet_aton\"",
+      "reported_date": "2025-12-29",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65324"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65324",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65324",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t67(c0 DOUBLE);\nREPLACE INTO t67(c0) VALUES (-1.930236983E9);\nCREATE VIEW v0(c4) AS SELECT ((t67.c0)<(BIT_COUNT(INET_ATON(0.8949238218722565)))) FROM t67;\nSET @a = 'a';",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3eea9eb37f911f397fa92f60b104f2ac9ef2a0f29880adffd3cf26b41a1ff279",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65324",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3eea9eb37f911f397fa92f60b104f2ac9ef2a0f29880adffd3cf26b41a1ff279"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:13a025fd17cc",
+      "dbms": "tidb",
+      "title": "Unexpected error on prepared statement \"ERROR 1105 (HY000) at line 16: Data Too Long, field len 11, data len 29\"",
+      "reported_date": "2025-12-29",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65323"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65323",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65323",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL );\nREPLACE INTO t0 VALUES (-1448876223);\nUPDATE t0 SET c0=-1.184245842E9 WHERE CAST((CASE (- (0.6007861749962686)) WHEN -1247397029 THEN -2006287455 WHEN t0.c0 THEN t0.c0 ELSE t0.c0 END ) AS BINARY); -- no error\nSET @a = -1.184245842E9;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:32008529a4539ebfc09a0eb17bec9879ced88a4f2e671be8a7f19006ef0f8903",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65323",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:32008529a4539ebfc09a0eb17bec9879ced88a4f2e671be8a7f19006ef0f8903"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:2496f6f6ef18",
+      "dbms": "tidb",
+      "title": "Undeterministic result of deterministic query",
+      "reported_date": "2026-01-03",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65387"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65387",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65387",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 FLOAT);\nCREATE TABLE t1 LIKE t0;\nREPLACE INTO t1 VALUES (0.26201291039277286);\nINSERT IGNORE  INTO t0(c0) VALUES (0.47821491788303083);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ccde42a2071cd7e6e1752b19d5be6cfaf81c8d06633461899f5d23fda3d9eafc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65387",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ccde42a2071cd7e6e1752b19d5be6cfaf81c8d06633461899f5d23fda3d9eafc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:8d9384ef7cee",
+      "dbms": "tidb",
+      "title": "Unexpected error \"Cannot convert string '\\x15t\\x03\\x10\\x93' from binary to utf8mb4\"",
+      "reported_date": "2026-01-03",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65386"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65386",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65386",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 TEXT(458) );\nINSERT INTO t0 VALUES ('-1574031093');\nSELECT SUBSTRING_INDEX(\"a\", UNHEX(CAST((- (CAST(t0.c0 AS BINARY))) AS CHAR)), 1) FROM t0; -- Cannot convert string '\\x15t\\x03\\x10\\x93' from binary to utf8mb4\nSELECT UNHEX(CAST((- (CAST(t0.c0 AS BINARY))) AS CHAR)) FROM t0; -- t",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ec933713ff6f3970d67c90f2077280e06c5f1aebc9d59e8ce5d4088fc928da4d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65386",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ec933713ff6f3970d67c90f2077280e06c5f1aebc9d59e8ce5d4088fc928da4d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:35c9425e08a0",
+      "dbms": "tidb",
+      "title": "BIT_AND returns incorrect result in prepared statement",
+      "reported_date": "2026-01-22",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65716"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65716",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65716",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 NUMERIC UNSIGNED );\nINSERT INTO t0(c0) VALUES (355074420);\nCREATE VIEW v0(c0) AS SELECT CAST(((((t0.c0)NOT LIKE(t0.c0))) IS NULL) AS DATE) FROM t0;\nSELECT * FROM v0 WHERE true GROUP BY v0.c0 HAVING BIT_AND(((v0.c0)>(CAST(v0.c0 AS CHAR)))); -- empty",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:469f6aa8e55d86a6705dd2aeae84423f2f104b60b910e9f0e2b5de4391f157da",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65716",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:469f6aa8e55d86a6705dd2aeae84423f2f104b60b910e9f0e2b5de4391f157da"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:b03830c3b4f8",
+      "dbms": "tidb",
+      "title": "Deterministic query returns indeterministic results",
+      "reported_date": "2026-01-22",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65721"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65721",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65721",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL UNSIGNED , c1 TEXT(228), c2 BOOL UNSIGNED );\nCREATE TABLE t16(c0 NUMERIC UNSIGNED , PRIMARY KEY(c0));\nINSERT INTO t0(c0) VALUES (false);\nREPLACE LOW_PRIORITY INTO t0(c0, c2) VALUES (true, false);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d6a578c21fb287b14c2ce20746c06266d2df70ffb3939a2828ed60c77f11dee4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65721",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d6a578c21fb287b14c2ce20746c06266d2df70ffb3939a2828ed60c77f11dee4"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:724ba8cc5fa4",
+      "dbms": "tidb",
+      "title": "Inconsistent results of DATA_FORMAT when used in normal SELECT and prepared SELECT",
+      "reported_date": "2026-01-22",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65717"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65717",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65717",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t29(c0 DECIMAL );\nREPLACE INTO t29(c0) VALUES (1);\nSELECT t29.c0 FROM t29 WHERE DATE_FORMAT(422123206, (BINARY (-81775))); -- 1\nSET @c=-81775;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:fd80a9dacc1348ba0dce24872e2908c617928d7bf38d29ca0861d93572503b54",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65717",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fd80a9dacc1348ba0dce24872e2908c617928d7bf38d29ca0861d93572503b54"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:ab10d859b04c",
+      "dbms": "tidb",
+      "title": "Incorrect result of ELT function when used in WHERE clause of prepared statement",
+      "reported_date": "2026-01-26",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65804"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65804",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65804",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\nINSERT INTO t0 VALUES (true);\nSELECT t0.c0 FROM t0 WHERE ELT(0.6706395853005295, 758832383, '&', NULL); -- 1\nSET @c = '&';",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6117fd41ba2c24b9972c26bfd12cade4f8f71bb5ba1755ecaa190603dc4361dd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65804",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6117fd41ba2c24b9972c26bfd12cade4f8f71bb5ba1755ecaa190603dc4361dd"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:b15b6ef09117",
+      "dbms": "tidb",
+      "title": "cast an invalid string to DATETIME returns NULL in normal SELECT, but triggers an error in prepared SELECT",
+      "reported_date": "2026-01-26",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65807"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65807",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65807",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BLOB);\nINSERT INTO t0(c0) VALUES ('5{8**<f');\nSET @a = '-';\nPREPARE prepare_query FROM 'SELECT ((CAST(CAST(t0.c0 AS DATETIME) AS DECIMAL))OR(?)) FROM  t0';",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:44cd621d722e73b09399a5cde3d2646ad71e15c5ff7e5936b66dc0731d3b9705",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65807",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:44cd621d722e73b09399a5cde3d2646ad71e15c5ff7e5936b66dc0731d3b9705"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:e526bf75bc54",
+      "dbms": "tidb",
+      "title": "Inconsistent result between normal SELECT and prepared SELECT, related to GROUP BY",
+      "reported_date": "2026-01-27",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65835"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65835",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65835",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 NUMERIC UNSIGNED , c2 DOUBLE ZEROFILL NOT NULL );\nINSERT IGNORE  INTO t0(c2) VALUES (-1.997930299E9);\nREPLACE INTO t0 VALUES (1403506458, 0.20756409900353212);\nCREATE OR REPLACE ALGORITHM=MERGE VIEW v0(c0) AS SELECT CAST(t0.c2 AS BINARY) FROM t0;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a6048610c9d898230bcfac871081ce71d4c961e28d949bd094d9286affcda9dc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65835",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a6048610c9d898230bcfac871081ce71d4c961e28d949bd094d9286affcda9dc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:3675a119d686",
+      "dbms": "tidb",
+      "title": "index out of bound only triggered when query run twice",
+      "reported_date": "2026-01-27",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65839"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65839",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65839",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c1 FLOAT);\nCREATE INDEX i0 ON t0(c1 ASC);\nset @@tidb_enable_chunk_rpc=0;\nREPLACE INTO t0 VALUES (0.5223332881233415), (-0.0);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:244b9a11cf38f2741164ef54ab3c7ea9dae1ee27bb6e9aef7366683ff3c185a8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65839",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:244b9a11cf38f2741164ef54ab3c7ea9dae1ee27bb6e9aef7366683ff3c185a8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:d46d6c998a23",
+      "dbms": "tidb",
+      "title": "nil pointer dereference triggered by INSERT",
+      "reported_date": "2026-01-27",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65836"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65836",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65836",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR , c1 NUMERIC ZEROFILL , c2 BOOL DEFAULT true );\nCREATE ALGORITHM=MERGE VIEW v0(c0, c1, c2) AS SELECT 0.5744235779532181, CAST((SELECT (NOT (t0.c1)) FROM  t0 CROSS JOIN t0 AS t0_0 ON '듑' WHERE t0.c2) AS SIGNED), ((true)LIKE(-353269119)) FROM  t0 LEFT  OUTER JOIN t0 AS t0_0 ON t0_0.c0;\nINSERT INTO t0(c2, c1) VALUES (false, 1420172923) ON DUPLICATE KEY UPDATE c2=(SELECT (SELECT v0.c2 FROM v0 WHERE v0.c2) FROM t0 WHERE 973867198 ORDER BY ((((t0.c1)&(((t0.c1)NOT REGEXP('-1015633450')))))NOT REGEXP((CASE t0.c2 WHEN NULL THEN '#\\r' ELSE '' END ))) ASC);\n-- runtime error: invalid memory address or nil pointer dereference",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:bbb9eb6725dec3b5e5971fc8f07eb7514ab1b3444768850ee786d1680d1ed826",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65836",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bbb9eb6725dec3b5e5971fc8f07eb7514ab1b3444768850ee786d1680d1ed826"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:fa28ee1e8886",
+      "dbms": "tidb",
+      "title": "Index out of bound when calling DEALLOCATE statement in JDBC",
+      "reported_date": "2026-01-28",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65872"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65872",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65872",
+            "source_type": "github_issue",
+            "excerpt": "Index out of bound when calling DEALLOCATE statement in JDBC",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:4f0336e32ca82b7b3077288ccfce06fbae22da425a2821faa484956f2bd92776",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "DerZc is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65872",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4f0336e32ca82b7b3077288ccfce06fbae22da425a2821faa484956f2bd92776"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:3e3c6c858b78",
+      "dbms": "tidb",
+      "title": "Inconsistent behaviour between normal and prepared SELECT when use subquery in ON condition",
+      "reported_date": "2026-02-02",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/65975"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/65975",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/65975",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR, c1 BOOL, c2 BOOL);\nREPLACE INTO t0 VALUES (NULL, false, false);\nSELECT t0.c0 FROM  t0 LEFT JOIN t0 AS t0_0 ON (('')!=((SELECT t0.c1 FROM t0 WHERE t0.c1 GROUP BY t0.c1 HAVING ((t0.c2)AND(NULL))))); -- NULL\nSET @d = '';",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:cd6c92b332b8e937ee5636c1afeaf16cbc0a458c69ade98171d0b8db9c1a4003",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/65975",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cd6c92b332b8e937ee5636c1afeaf16cbc0a458c69ade98171d0b8db9c1a4003"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:67e3cd6cf736",
+      "dbms": "tidb",
+      "title": "runtime error: nil pointer dereference with RIGHT JOINs and views (TLP)",
+      "reported_date": "2026-02-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "hawkingrei",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66170"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66170",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66170",
+            "source_type": "github_issue",
+            "excerpt": "- Found by Shiro (oracle: TLP).\n- Error reason: `tlp:base_signature_error`, bug hint: `tidb:runtime_error`.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b4b344b8cf8afbdbb96ca1fd99978d3dc2d94cc70a99a9bfaf0e4e3e5ee7192a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66170",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b4b344b8cf8afbdbb96ca1fd99978d3dc2d94cc70a99a9bfaf0e4e3e5ee7192a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:tidb:781ec56db5d4",
+      "dbms": "tidb",
+      "title": "runtime error: index out of range on ANY subquery predicate",
+      "reported_date": "2026-02-13",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "unknown",
+      "reporter": "hawkingrei",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66259"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66259",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66259",
+            "source_type": "github_issue",
+            "excerpt": "Observed from Shiro fuzz case `019c555f-1ff2-7186-9167-87b1f656db43` (PQS oracle).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:ea836cc7088424869acc8be85d4994b74ed0c8791449ad75f246b3785ff2569e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:10:43Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:10:43Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66259",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ea836cc7088424869acc8be85d4994b74ed0c8791449ad75f246b3785ff2569e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:tidb:0a14db5a351b",
+      "dbms": "tidb",
+      "title": "planner/executor: internal missing-column error with JOIN ... USING + qualified column predicate",
+      "reported_date": "2026-02-14",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "pqs",
+      "symptom": "unknown",
+      "reporter": "hawkingrei",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66272"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66272",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66272",
+            "source_type": "github_issue",
+            "excerpt": "- Oracle: PQS\n- Error classification from fuzz runner: `pqs:missing_column` / `tidb:schema_column_missing`",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:68ea61074523d5a7631c61b5062c12138782953830930cbfcd4dfc7942ad5815",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66272",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:68ea61074523d5a7631c61b5062c12138782953830930cbfcd4dfc7942ad5815"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:tidb:cae522daff38",
+      "dbms": "tidb",
+      "title": "The simple prepared query run out of 1 GiB memory with only 70 rows",
+      "reported_date": "2026-02-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66301"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66301",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66301",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t72(c0 TEXT(417) );\nINSERT INTO t72(c0)\nWITH RECURSIVE cte AS (\n    SELECT 1 AS n",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:097ccdbf9f9a5fe419bfc78aa17ec542a5e1692acab9f5478f0733398ae19216",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66301",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:097ccdbf9f9a5fe419bfc78aa17ec542a5e1692acab9f5478f0733398ae19216"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:2830676e61f3",
+      "dbms": "tidb",
+      "title": "planner: missing-column 1105 with RIGHT JOIN + ORDER BY subtraction expression",
+      "reported_date": "2026-02-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "dqp",
+      "symptom": "unknown",
+      "reporter": "hawkingrei",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66601"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66601",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66601",
+            "source_type": "github_issue",
+            "excerpt": "- oracle: `DQP`\n- minimize status: `success`",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5db07e091d564034d0ad0c6f09f64612d8d4a42d8ac25a1e2cf023a04d66fcb2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66601",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5db07e091d564034d0ad0c6f09f64612d8d4a42d8ac25a1e2cf023a04d66fcb2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:tidb:50a8351eb79e",
+      "dbms": "tidb",
+      "title": "Index with StreamAgg used in the prepared SELECT but not in the normal SELECT",
+      "reported_date": "2026-03-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66607"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66607",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66607",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t5(c0 NUMERIC UNSIGNED ZEROFILL );\nINSERT INTO t5 (c0)\nWITH RECURSIVE seq AS (\n    SELECT 1 AS n",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a39f0fca00fd346012e7d887cc0f6b3217e3e14ef4142a10f0fb4f065d2635fe",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66607",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a39f0fca00fd346012e7d887cc0f6b3217e3e14ef4142a10f0fb4f065d2635fe"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:90336de93f77",
+      "dbms": "tidb",
+      "title": "Inefficient query plan generated by normal SELECT compare to the prepared SELECT with false condition",
+      "reported_date": "2026-03-01",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66604"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66604",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66604",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR , c2 DOUBLE UNSIGNED );\nINSERT INTO t0 (c0, c2)\nWITH RECURSIVE nr AS (\n    SELECT 1 AS n",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e2a42663132c443c22288229045ef1aeeca439bd895e7121bb20d4fdeded4b1e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66604",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e2a42663132c443c22288229045ef1aeeca439bd895e7121bb20d4fdeded4b1e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:bc19e81ffeef",
+      "dbms": "tidb",
+      "title": "Inefficient query plan generated by the normal SELECT than the prepared SELECT when use subquery in the SELECT clause",
+      "reported_date": "2026-03-01",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66606"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66606",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66606",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 FLOAT);\nINSERT INTO t0 (c0)\nWITH RECURSIVE nr AS (\n    SELECT 1 AS n",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:fe8bef7e8cfab5eaba321305a161f86c8a952aeede50af3109b4abd9c9d9b34b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66606",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fe8bef7e8cfab5eaba321305a161f86c8a952aeede50af3109b4abd9c9d9b34b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:30633e7d7778",
+      "dbms": "tidb",
+      "title": "Inefficient query plan generated by the normal SELECT than the prepared SELECT with false HAVING condition",
+      "reported_date": "2026-03-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66605"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66605",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66605",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 CHAR );\nINSERT INTO t0 (c0)\nWITH RECURSIVE seq AS (\n    SELECT 1 AS n",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:77c08a9d262df78dc2c37935f04ffe4cdfca44e8bd9b68be07ce8ac097c4455d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66605",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:77c08a9d262df78dc2c37935f04ffe4cdfca44e8bd9b68be07ce8ac097c4455d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:285502f4a17a",
+      "dbms": "tidb",
+      "title": "Redundant HashAgg in the query plan of normal SELECT, causes it slower than the prepared SELECT",
+      "reported_date": "2026-03-01",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66610"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66610",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66610",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 FLOAT UNSIGNED );\nINSERT INTO t0 (c0)\nWITH RECURSIVE seq AS (\n    SELECT 1 AS n",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3bf958d1458fc6279fa1c16947e9e640ee7a63c33c2a28896de2953bb7555d8e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66610",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3bf958d1458fc6279fa1c16947e9e640ee7a63c33c2a28896de2953bb7555d8e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:42d7464f24e2",
+      "dbms": "tidb",
+      "title": "Redundant Selection and Projection in the query plan of normal SELECT, causes it slower than the prepared SELECT",
+      "reported_date": "2026-03-01",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66609"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66609",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66609",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 NUMERIC , c1 BOOL, c2 CHAR);\nINSERT INTO t1 (c0, c1, c2)\nWITH RECURSIVE seq AS (\n    SELECT 1 AS n",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2fc6c7887c36f7312628e35cfccdba692f0590dfed1eddc7ba04bd37e5311c6f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66609",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2fc6c7887c36f7312628e35cfccdba692f0590dfed1eddc7ba04bd37e5311c6f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:38850471cdcd",
+      "dbms": "tidb",
+      "title": "Redundant sort in the query plan of normal SELECT, causes it slower than the prepared SELECT",
+      "reported_date": "2026-03-01",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "DerZc",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/66608"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/66608",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/66608",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 BOOL);\nINSERT INTO t0 (c0)\nWITH RECURSIVE seq AS (\n    SELECT 1 AS n",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0483c99b9a91104d4a2cd1b72b799daf4ee33839c19f877367b5e79b58388618",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/66608",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0483c99b9a91104d4a2cd1b72b799daf4ee33839c19f877367b5e79b58388618"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:3905bb22aa4a",
+      "dbms": "tidb",
+      "title": "TiKV panic in impl_like::like with invalid regexp pattern (slice out of bounds)",
+      "reported_date": "2026-03-17",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LLMTesting-Ojiken",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/67082"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/67082",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/67082",
+            "source_type": "github_issue",
+            "excerpt": "This issue was discovered via SQLancer-based randomized testing.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:8375e58e211687ea01c883fffa0958e17f526c37d19a5a74af2698df8203440e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/67082",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8375e58e211687ea01c883fffa0958e17f526c37d19a5a74af2698df8203440e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:tidb:994c3ea721b6",
+      "dbms": "tidb",
+      "title": "TiDB nightly panic in mockstore/unistore badger Table.HasOverlap during transaction rollback cleanup",
+      "reported_date": "2026-06-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Annie191",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/69276"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/69276",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/69276",
+            "source_type": "github_issue",
+            "excerpt": "Run SQLancer against this TiDB instance:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3560ab6344960f4b0dab6ba551b7a18181b0e368ba44b0a127e46fd5caa52db1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/69276",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3560ab6344960f4b0dab6ba551b7a18181b0e368ba44b0a127e46fd5caa52db1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:tidb:3fbb9055acad",
+      "dbms": "tidb",
+      "title": "`UNION ALL` relation + `IN (correlated subquery)` over an `ALL` subquery panics",
+      "reported_date": "2026-07-30",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/70248"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/70248",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/70248",
+            "source_type": "github_issue",
+            "excerpt": "`UNION ALL` relation + `IN (correlated subquery)` over an `ALL` subquery panics",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:edc6bc7eed0e08046db48f1dba9866a780ce2a87bf52e29181d9f802770590d3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/70248",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:edc6bc7eed0e08046db48f1dba9866a780ce2a87bf52e29181d9f802770590d3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:fdf4d188aef1",
+      "dbms": "tidb",
+      "title": "An aggregate window function inside a view breaks correlated quantified subqueries (`< ANY` / `= ALL` / `IN (SELECT …)`)",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/70362"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/70362",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/70362",
+            "source_type": "github_issue",
+            "excerpt": "An aggregate window function inside a view breaks correlated quantified subqueries (`< ANY` / `= ALL` / `IN (SELECT …)`)",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1f3b3d85464d8590ea30c8d9be79529c9fbe7cbe30ef4d2901b4e369676d9898",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/70362",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1f3b3d85464d8590ea30c8d9be79529c9fbe7cbe30ef4d2901b4e369676d9898"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:e32a45334066",
+      "dbms": "tidb",
+      "title": "`ANY_VALUE()` in a view raises planner error 1105 `Can't find column Column#N in schema`",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/70361"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/70361",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/70361",
+            "source_type": "github_issue",
+            "excerpt": "`ANY_VALUE()` in a view raises planner error 1105 `Can't find column Column#N in schema`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:df7c33e2495991a416e1ff6e2b095d889f8482ec8ab3fcc6dec0ed9db04b2600",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/70361",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:df7c33e2495991a416e1ff6e2b095d889f8482ec8ab3fcc6dec0ed9db04b2600"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:007812a4df43",
+      "dbms": "tidb",
+      "title": "`predicate_push_down` returns a nil plan",
+      "reported_date": "2026-08-05",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/70360"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/70360",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/70360",
+            "source_type": "github_issue",
+            "excerpt": "`predicate_push_down` returns a nil plan",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:2d347d84993329347d458fc039a70e8e05bce87d895ee278447bb2bc57774b25",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/70360",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2d347d84993329347d458fc039a70e8e05bce87d895ee278447bb2bc57774b25"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:27b81ea165d1",
+      "dbms": "tidb",
+      "title": "`STDDEV_POP` / `VAR_*` over `BIGINT UNSIGNED` near `2^64` wrong on `UNION ALL`",
+      "reported_date": "2026-08-09",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/70400"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/70400",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/70400",
+            "source_type": "github_issue",
+            "excerpt": "`STDDEV_POP` / `VAR_*` over `BIGINT UNSIGNED` near `2^64` wrong on `UNION ALL`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:f41dbc158d52838ec63299bfa806d568dff598769b8ff3cd6734bb668e14e34b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/70400",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f41dbc158d52838ec63299bfa806d568dff598769b8ff3cd6734bb668e14e34b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:8500fd92a4b7",
+      "dbms": "tidb",
+      "title": "Inconsistent results from `REGEXP_REPLACE` on a `CACHE` table",
+      "reported_date": "2026-08-12",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/70458"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/70458",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/70458",
+            "source_type": "github_issue",
+            "excerpt": "Inconsistent results from `REGEXP_REPLACE` on a `CACHE` table",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:678cf6f56176d92e80f8f1053495970ebc7f728d4f6e031be362b904bf0a7f35",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/70458",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:678cf6f56176d92e80f8f1053495970ebc7f728d4f6e031be362b904bf0a7f35"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:e37fcde348c1",
+      "dbms": "tidb",
+      "title": "`ENUM` numeric predicates wrong through `UNION ALL`",
+      "reported_date": "2026-08-12",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/70457"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/70457",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/70457",
+            "source_type": "github_issue",
+            "excerpt": "`ENUM` numeric predicates wrong through `UNION ALL`",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:27c283349c43595793cbc2a042aa9aeced7cc3277bd0991009746da34a82b833",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/70457",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:27c283349c43595793cbc2a042aa9aeced7cc3277bd0991009746da34a82b833"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:2426fe718ca0",
+      "dbms": "tidb",
+      "title": "Grouped `REPEAT` over a `WITH` CTE returns a string; the same query on the table returns NULL",
+      "reported_date": "2026-08-14",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/70493"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/70493",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/70493",
+            "source_type": "github_issue",
+            "excerpt": "Grouped `REPEAT` over a `WITH` CTE returns a string; the same query on the table returns NULL",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fc625c2a2e445604981de202dd50708812dc2c40abf0b99aef3035110a44ffc9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/70493",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fc625c2a2e445604981de202dd50708812dc2c40abf0b99aef3035110a44ffc9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:276fe0292cf4",
+      "dbms": "tidb",
+      "title": "`MIN(GREATEST(<datetime string>, DATE)) GROUP BY` over a `(ROW_NUMBER() = 1)` filter view returns DATETIME instead of DATE",
+      "reported_date": "2026-08-16",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "wanteatfruit",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/70503"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/70503",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/70503",
+            "source_type": "github_issue",
+            "excerpt": "`MIN(GREATEST(<datetime string>, DATE)) GROUP BY` over a `(ROW_NUMBER() = 1)` filter view returns DATETIME instead of DATE",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:a2b938657ff43d7bdf476d5a06a9e00577f3ead600b49395ae589f5248a65b7f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "wanteatfruit is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/70503",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a2b938657ff43d7bdf476d5a06a9e00577f3ead600b49395ae589f5248a65b7f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:tidb:cf4894d3b458",
+      "dbms": "tidb",
+      "title": "With pseudo statistics, GROUP BY / DISTINCT over COALESCE(float_col) collapses rows into wrong groups and mis-renders the keys (two-level aggregate re-encodes the pushed-down key)",
+      "reported_date": "2026-08-31",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "Manuel-Neuer1",
+      "links": {
+        "report": "https://github.com/pingcap/tidb/issues/70754"
+      },
+      "primary_url": "https://github.com/pingcap/tidb/issues/70754",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/pingcap/tidb/issues/70754",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer's DQR oracle (query-relocation differential testing) on TiDB nightly:\na 44-statement generated test case produced the 7-vs-6 disagreement above; minimized by",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:978a49e15908da3853ad6bdf799b53f6bbb9e3c3df7071d40f9590334cc9fbfb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/pingcap/tidb/issues/70754",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:978a49e15908da3853ad6bdf799b53f6bbb9e3c3df7071d40f9590334cc9fbfb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:tikv:74cb692a7e3a",
+      "dbms": "tikv",
+      "title": "TiKV panic in `tidb_query_expr::impl_like::LikeFn::eval` during SQLancer workload",
+      "reported_date": "2025-09-28",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "qpg",
+      "symptom": "unknown",
+      "reporter": "hbisheng",
+      "links": {
+        "report": "https://github.com/tikv/tikv/issues/19010"
+      },
+      "primary_url": "https://github.com/tikv/tikv/issues/19010",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tikv/tikv/issues/19010",
+            "source_type": "github_issue",
+            "excerpt": "TiKV panic in `tidb_query_expr::impl_like::LikeFn::eval` during SQLancer workload",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:748dfc646cd713b461ebdb30b230e7145bfdbe9bc15525a101236154dcc66038",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tikv/tikv/issues/19010",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:748dfc646cd713b461ebdb30b230e7145bfdbe9bc15525a101236154dcc66038"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:991db28c32ff",
+      "dbms": "turso",
+      "title": "\"Corrupt database: Unresolved label: Placeholder\" when creating index with NOT IN",
+      "reported_date": "2026-01-07",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4484"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4484",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4484",
+            "source_type": "github_issue",
+            "excerpt": "Found with SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:e78a3c1cbe70cc7cce0881036d33d6864d9a079752542237642d4833169dd01c",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4484",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e78a3c1cbe70cc7cce0881036d33d6864d9a079752542237642d4833169dd01c"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:c2b61e99b6e0",
+      "dbms": "turso",
+      "title": "Corrupt database: Reference to undefined or unresolved label in Found: 8",
+      "reported_date": "2026-01-07",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4512"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4512",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4512",
+            "source_type": "github_issue",
+            "excerpt": "Found by Claude while investigating a panic found by SQLancer.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:3658e761c2c4b9e4f103f9f63e8c3fde1179bae2c175a9db6ab6b5ed887ca011",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4512",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3658e761c2c4b9e4f103f9f63e8c3fde1179bae2c175a9db6ab6b5ed887ca011"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:b504653d0d7b",
+      "dbms": "turso",
+      "title": "Panic: \"Should be valid f64: ParseFloatError { kind: Invalid }\"",
+      "reported_date": "2026-01-07",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4507"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4507",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4507",
+            "source_type": "github_issue",
+            "excerpt": "Found with SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:a1e2db24c37b3ed206a9b069196a59a7255be2ff3c852a8a0c369cde79e769eb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4507",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a1e2db24c37b3ed206a9b069196a59a7255be2ff3c852a8a0c369cde79e769eb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:c156b53c72a9",
+      "dbms": "turso",
+      "title": "Panic: \"column name is None\" when NATURAL JOIN has unnamed columns",
+      "reported_date": "2026-01-07",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4489"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4489",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4489",
+            "source_type": "github_issue",
+            "excerpt": "Found using SQLancer + Claude",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:ed5d1bb3b3c7ba2c43617cf835e478936d9ccfc2ae276c0154222472c0327cda",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4489",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ed5d1bb3b3c7ba2c43617cf835e478936d9ccfc2ae276c0154222472c0327cda"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:086acb723c6c",
+      "dbms": "turso",
+      "title": "Panic: cursor id 0 is None",
+      "reported_date": "2026-01-07",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4515"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4515",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4515",
+            "source_type": "github_issue",
+            "excerpt": "Found with SQLancer + Claude (many iterations with planning and ULTRATHINK).",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:7c0f6e46fb12ce1e6375fc01bd5379c1e2f2632301e83fb0990b9e9314663a73",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4515",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7c0f6e46fb12ce1e6375fc01bd5379c1e2f2632301e83fb0990b9e9314663a73"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:653a246def0c",
+      "dbms": "turso",
+      "title": "Panic: entered unreachable code in jsonb.rs",
+      "reported_date": "2026-01-07",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4488"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4488",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4488",
+            "source_type": "github_issue",
+            "excerpt": "Found using SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:fd59a559e1e16131e3645ea7696b2e7102cb8ce427bcde2775b027435d625b12",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4488",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fd59a559e1e16131e3645ea7696b2e7102cb8ce427bcde2775b027435d625b12"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:9a0a1ee2091c",
+      "dbms": "turso",
+      "title": "expression should have been rewritten in optimizer",
+      "reported_date": "2026-01-07",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4486"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4486",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4486",
+            "source_type": "github_issue",
+            "excerpt": "found with SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:cee6c17a81e02f027ecb1296427b594138ed78fb2e37370798efb7c29ee24d68",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4486",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cee6c17a81e02f027ecb1296427b594138ed78fb2e37370798efb7c29ee24d68"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:3c9ae5699d73",
+      "dbms": "turso",
+      "title": "Panic when comparing NaN",
+      "reported_date": "2026-01-09",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4538"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4538",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4538",
+            "source_type": "github_issue",
+            "excerpt": "Found with SQLancer and reproduced with Claude. And this ralph-wiggum-esque loop:",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:5eb0ae880692be920e3be775189b23eaa3ce68353482b325e108dfb2ef4e871f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4538",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5eb0ae880692be920e3be775189b23eaa3ce68353482b325e108dfb2ef4e871f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:6190fd35651d",
+      "dbms": "turso",
+      "title": "Turso hangs on \"select 1 as a group by a order by a;\"",
+      "reported_date": "2026-01-10",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4588"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4588",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4588",
+            "source_type": "github_issue",
+            "excerpt": "Found by either AI or SQLancer, something that was running on my computer and left a few hanging copies of this query.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:83f7211f77b8f20d490c5bab08c99134f7fb74935631cfa818953550049801da",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4588",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:83f7211f77b8f20d490c5bab08c99134f7fb74935631cfa818953550049801da"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:712608cf97e4",
+      "dbms": "turso",
+      "title": "Panic in b-tree balancing",
+      "reported_date": "2026-01-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4603"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4603",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4603",
+            "source_type": "github_issue",
+            "excerpt": "I ran into this while running SQLancer. Unfortunately I don't have the failing seed or SQL. But it seems important, because it means that parent page was somehow actually a leaf page.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a142240958f3d8d02d4e0ca1580c9944c5b189550734f077368ccf7e05f32863",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4603",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a142240958f3d8d02d4e0ca1580c9944c5b189550734f077368ccf7e05f32863"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:dabdc2bab61a",
+      "dbms": "turso",
+      "title": "Panic: yield_reg 1 contains non-integer value: Value(Null)",
+      "reported_date": "2026-01-11",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4602"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4602",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4602",
+            "source_type": "github_issue",
+            "excerpt": "I ran into this error while running SQLancer. Unfortunately, I don't have the failing seed, or the failing SQL, and Claude seems unable to figure it out.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e043c7cbe3e1594d37afd730c4b3b497c6e9bfa52198734211552e8b91e2829e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4602",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e043c7cbe3e1594d37afd730c4b3b497c6e9bfa52198734211552e8b91e2829e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:6ad4b49ec394",
+      "dbms": "turso",
+      "title": "Logic bug found by SQLancer",
+      "reported_date": "2026-01-12",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4606"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4606",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4606",
+            "source_type": "github_issue",
+            "excerpt": "Logic bug found by SQLancer",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:7d473976e906d0172f6badfd7ee7e40897916b03659add22049db164031738b9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4606",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7d473976e906d0172f6badfd7ee7e40897916b03659add22049db164031738b9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:43f5eac44c64",
+      "dbms": "turso",
+      "title": "ORDER BY with float literal causes parse error",
+      "reported_date": "2026-01-12",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4608"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4608",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4608",
+            "source_type": "github_issue",
+            "excerpt": "**Mikaël: found with SQLancer, explained/reduced by Claude Code + Claude**",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:76b54367da9ff981f39a159f23c818dff7664826afc2ed2c25e26b25d2ca97e9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4608",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:76b54367da9ff981f39a159f23c818dff7664826afc2ed2c25e26b25d2ca97e9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:2c75950db942",
+      "dbms": "turso",
+      "title": "Panic: \"attempt to negate with overflow\"",
+      "reported_date": "2026-01-12",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4621"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4621",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4621",
+            "source_type": "github_issue",
+            "excerpt": "Found by SQLancer, unfortunately I don't have the seed or the SQL that triggered it.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:1eedaea612568c4c9563e813368431ae399790dc9618f99e3c613934e21e2ae2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4621",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1eedaea612568c4c9563e813368431ae399790dc9618f99e3c613934e21e2ae2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:2ec5a19889c9",
+      "dbms": "turso",
+      "title": "Reference to undefined or unresolved label in Goto: 18",
+      "reported_date": "2026-01-12",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4607"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4607",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4607",
+            "source_type": "github_issue",
+            "excerpt": "Found with SQLancer and the WHERE oracle:",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:a0dc834f60ba21b524f76cbc0600056b2c3cb115f92fa573ba31cdad77a266ab",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4607",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a0dc834f60ba21b524f76cbc0600056b2c3cb115f92fa573ba31cdad77a266ab"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:505ddb72f67a",
+      "dbms": "turso",
+      "title": "Set up SQLRight fuzzing",
+      "reported_date": "2026-01-12",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "LeMikaelF",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4604"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4604",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4604",
+            "source_type": "github_issue",
+            "excerpt": "[SQLRight](https://github.com/PSU-Security-Universe/sqlright/tree/main) is a coverage-guided SQL fuzzer that supports SQLite. It supports the NoREC and TLP oracles, and 2 more custom oracles. They found 27 bugs in SQLite.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:aae83116d8e5739fd1e7a8dccaa86d716c8557fede9adc47a03bac25832dea74",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4604",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:aae83116d8e5739fd1e7a8dccaa86d716c8557fede9adc47a03bac25832dea74"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:d7a9d035d4df",
+      "dbms": "turso",
+      "title": "SQLancer NoREC: timeout after 300s",
+      "reported_date": "2026-01-13",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "turso-github-handyman[bot]",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4628"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4628",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4628",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer NoREC: timeout after 300s",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f953a74c4ae7fd89e5273b12e6baee2de63d069a867efba07d4e40f8be9d822f",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4628",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f953a74c4ae7fd89e5273b12e6baee2de63d069a867efba07d4e40f8be9d822f"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:turso:702b4bffd803",
+      "dbms": "turso",
+      "title": "SQLancer run failures - 2026-01-14 (789b96e)",
+      "reported_date": "2026-01-14",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "turso-github-handyman[bot]",
+      "links": {
+        "report": "https://github.com/tursodatabase/turso/issues/4656"
+      },
+      "primary_url": "https://github.com/tursodatabase/turso/issues/4656",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/tursodatabase/turso/issues/4656",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer run failures - 2026-01-14 (789b96e)",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:773f8ee4848803a071edecbe0e100d910c453a73c5a07a402336ab056da2047d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:33:38Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/tursodatabase/turso/issues/4656",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:773f8ee4848803a071edecbe0e100d910c453a73c5a07a402336ab056da2047d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:umbra:713d7ac0fe96",
+      "dbms": "umbra",
+      "title": "Potential issue handling Boolean values after INDEX",
+      "reported_date": "2023-05-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Potential issue handling Boolean values after INDEX",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:713d7ac0fe96140bf71a3612afa8fb7e9b6b7545c1e611dc780db26167e5b791",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:713d7ac0fe96140bf71a3612afa8fb7e9b6b7545c1e611dc780db26167e5b791"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:0b5c125bed17",
+      "dbms": "umbra",
+      "title": "Potential issue when Using Boolean values and BETWEEN Operator",
+      "reported_date": "2023-05-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Potential issue when Using Boolean values and BETWEEN Operator",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:0b5c125bed17755dedcb6fe04796465ccb3e083656ea0af53951fca62b7ee03d",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:0b5c125bed17755dedcb6fe04796465ccb3e083656ea0af53951fca62b7ee03d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:42ea6031d67a",
+      "dbms": "umbra",
+      "title": "Segmentation fault when Using NATURAL LEFT JOIN",
+      "reported_date": "2023-05-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Segmentation fault when Using NATURAL LEFT JOIN",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:42ea6031d67a2f8a2640a4930bf492d8d8e0850d9e58bf6525f672700a635fea",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:42ea6031d67a2f8a2640a4930bf492d8d8e0850d9e58bf6525f672700a635fea"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:2152dc1f1fa4",
+      "dbms": "umbra",
+      "title": "Unexpected results after triggering an index scan",
+      "reported_date": "2023-06-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results after triggering an index scan",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:2152dc1f1fa4e01ae9fe54825f105f4ec67b1370f2fdbbca47d20dd4d108a5d3",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:2152dc1f1fa4e01ae9fe54825f105f4ec67b1370f2fdbbca47d20dd4d108a5d3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:1f617afd36e3",
+      "dbms": "umbra",
+      "title": "Potential Issue when Using CONCAT with overflow integers",
+      "reported_date": "2023-09-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Potential Issue when Using CONCAT with overflow integers",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:1f617afd36e3934f108522c5717e3259b3f2401dae014aafb0e010a7d23d9c6b",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:1f617afd36e3934f108522c5717e3259b3f2401dae014aafb0e010a7d23d9c6b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:7359003391b0",
+      "dbms": "umbra",
+      "title": "Segmentation fault when Using Between Operator",
+      "reported_date": "2023-09-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Segmentation fault when Using Between Operator",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:7359003391b0654ec8a89cf1c2e94046402dac4c6369ea61dbd6002f7c82bf07",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:7359003391b0654ec8a89cf1c2e94046402dac4c6369ea61dbd6002f7c82bf07"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:a976d28f8dfd",
+      "dbms": "umbra",
+      "title": "Unexpected Results when Comparing Boolean Values",
+      "reported_date": "2023-11-27",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Results when Comparing Boolean Values",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:a976d28f8dfd8b959427fe23335c99a71274fea19b043885743bf541376584ae",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:a976d28f8dfd8b959427fe23335c99a71274fea19b043885743bf541376584ae"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:a9b823881391",
+      "dbms": "umbra",
+      "title": "Potential Issue in BETWEEN operator",
+      "reported_date": "2023-11-28",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Potential Issue in BETWEEN operator",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:a9b8238813910475e51b97801632ffdbf2852d5fb4f3b3b8c3b836be08084a88",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:a9b8238813910475e51b97801632ffdbf2852d5fb4f3b3b8c3b836be08084a88"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:50dc01ed2916",
+      "dbms": "umbra",
+      "title": "Potential Issue about UNIQUE INDEX",
+      "reported_date": "2023-11-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Potential Issue about UNIQUE INDEX",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:50dc01ed291654b8f3937974310efba7c2c0732a047f097e6a4dc9092f3d8918",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:50dc01ed291654b8f3937974310efba7c2c0732a047f097e6a4dc9092f3d8918"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:9bb2e76505cb",
+      "dbms": "umbra",
+      "title": "Potential Issue about boolean pruning in filter",
+      "reported_date": "2023-11-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Potential Issue about boolean pruning in filter",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:9bb2e76505cb31e2a6d8a1265c55cddb4fcbc396946abbe86eafe65b6edf7c60",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:9bb2e76505cb31e2a6d8a1265c55cddb4fcbc396946abbe86eafe65b6edf7c60"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:a176c3cccb4f",
+      "dbms": "umbra",
+      "title": "Potential Issue in NULLIF function",
+      "reported_date": "2023-11-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Potential Issue in NULLIF function",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:a176c3cccb4f27211970dc8dc0f0fa0c6a6bdc8c1826e5af00d74d6e7825639a",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:a176c3cccb4f27211970dc8dc0f0fa0c6a6bdc8c1826e5af00d74d6e7825639a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:662eaecfd2ea",
+      "dbms": "umbra",
+      "title": "Potential Issue when Dropping table",
+      "reported_date": "2023-11-29",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Potential Issue when Dropping table",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:662eaecfd2ea0d95ce523876f3b2243b2c882eff799fa57c064da56995eafeb7",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:662eaecfd2ea0d95ce523876f3b2243b2c882eff799fa57c064da56995eafeb7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:8fd797194d08",
+      "dbms": "umbra",
+      "title": "Unexpected Results When Using Between Operator and String Concatenation",
+      "reported_date": "2023-12-12",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Results When Using Between Operator and String Concatenation",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:8fd797194d08496829876c244e5ec56df0314df5ad9078b596b8f082986be163",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:8fd797194d08496829876c244e5ec56df0314df5ad9078b596b8f082986be163"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:1bd89cd569d0",
+      "dbms": "umbra",
+      "title": "Segmentation fault when Using Between Operator and integer",
+      "reported_date": "2023-12-14",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Segmentation fault when Using Between Operator and integer",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:1bd89cd569d0349da13b683305f79a42e4b6782e936a4e4fddce871e600f8477",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:1bd89cd569d0349da13b683305f79a42e4b6782e936a4e4fddce871e600f8477"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:041bbfc7864e",
+      "dbms": "umbra",
+      "title": "Unexpected results when Using BETWEEN Operator and Comparison",
+      "reported_date": "2023-12-16",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results when Using BETWEEN Operator and Comparison",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:041bbfc7864eac13eedec724bb75dfe17f2333bd0e88173b44337eac3c281fcf",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:041bbfc7864eac13eedec724bb75dfe17f2333bd0e88173b44337eac3c281fcf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:ef1da11dba90",
+      "dbms": "umbra",
+      "title": "Unexpected results when LEFT JOIN a view",
+      "reported_date": "2023-12-19",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results when LEFT JOIN a view",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:ef1da11dba903baaf633d0978eb1ef2f74cf454181115fd7da1b07012b94de51",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:ef1da11dba903baaf633d0978eb1ef2f74cf454181115fd7da1b07012b94de51"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:6a8189f0e5bb",
+      "dbms": "umbra",
+      "title": "Unexpected results when Mod negative number with View",
+      "reported_date": "2023-12-23",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results when Mod negative number with View",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:6a8189f0e5bb23335286d65775805b44a7c2046395a7a1ba2cb9537408e1a4a0",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:6a8189f0e5bb23335286d65775805b44a7c2046395a7a1ba2cb9537408e1a4a0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:013740fc9aa6",
+      "dbms": "umbra",
+      "title": "Unexpected results when Using UNION ALL for JOIN",
+      "reported_date": "2023-12-23",
+      "reported_year": 2023,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results when Using UNION ALL for JOIN",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:013740fc9aa627588144fcd7d2a94b9de475a354720b071816360e176a91c9c8",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:013740fc9aa627588144fcd7d2a94b9de475a354720b071816360e176a91c9c8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:d991734bc5a1",
+      "dbms": "umbra",
+      "title": "Segmentation fault when Using NATURAL FULL JOIN",
+      "reported_date": "2024-01-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Segmentation fault when Using NATURAL FULL JOIN",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:d991734bc5a1ee4df606e20828eabba4c3e0f4393646f2de69d8e436a1193127",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:d991734bc5a1ee4df606e20828eabba4c3e0f4393646f2de69d8e436a1193127"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:90e82a16d65a",
+      "dbms": "umbra",
+      "title": "Assertion failure when using NATURAL JOIN",
+      "reported_date": "2024-01-13",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Assertion failure when using NATURAL JOIN",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:90e82a16d65a91c5de7b6b0df5db685a7f685178e0ceaeb062b21fb381f5f904",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:90e82a16d65a91c5de7b6b0df5db685a7f685178e0ceaeb062b21fb381f5f904"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:a4e999318d9f",
+      "dbms": "umbra",
+      "title": "Unexpected results when Using COS and IN expression",
+      "reported_date": "2024-01-13",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results when Using COS and IN expression",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:a4e999318d9febbbd558919b7ca89ef3ff93071f113064b6545d583044c732a7",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:a4e999318d9febbbd558919b7ca89ef3ff93071f113064b6545d583044c732a7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:dcf1c4fc3dc4",
+      "dbms": "umbra",
+      "title": "Unexpected results when Casting to VARCHAR",
+      "reported_date": "2024-01-14",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results when Casting to VARCHAR",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:dcf1c4fc3dc4077855aa908263ed4eec06110848eede1eff521ae1f33618f67d",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:dcf1c4fc3dc4077855aa908263ed4eec06110848eede1eff521ae1f33618f67d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:ce5361c5a2b5",
+      "dbms": "umbra",
+      "title": "Assertion failure when using COT and BETWEEN AND",
+      "reported_date": "2024-01-15",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Assertion failure when using COT and BETWEEN AND",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:ce5361c5a2b58edf3c1f1e1b97511c1ab4a88c65e6b1645a35c565cc27528c5f",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:ce5361c5a2b58edf3c1f1e1b97511c1ab4a88c65e6b1645a35c565cc27528c5f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:69afba1e817f",
+      "dbms": "umbra",
+      "title": "ERROR: AddressSanitizer: stack-buffer-overflow",
+      "reported_date": "2024-01-15",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "ERROR: AddressSanitizer: stack-buffer-overflow",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:69afba1e817fe9897c14d8ec1cae593db5767db35c64130264bd90e6d0feabb5",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:69afba1e817fe9897c14d8ec1cae593db5767db35c64130264bd90e6d0feabb5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:884a86570e37",
+      "dbms": "umbra",
+      "title": "Unexpected results when using NATURAL RIGHT JOIN with INDEX",
+      "reported_date": "2024-01-15",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results when using NATURAL RIGHT JOIN with INDEX",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:884a86570e37e62f28d6395399f77cd8a26880814863b5ab3b358736eb0c662b",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:884a86570e37e62f28d6395399f77cd8a26880814863b5ab3b358736eb0c662b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:08e1120bb37c",
+      "dbms": "umbra",
+      "title": "Crash when using BETWEEN operator and large integer",
+      "reported_date": "2024-01-16",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Crash when using BETWEEN operator and large integer",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:08e1120bb37c91a2589396679208dbbc7f7b037daee03624cecf7c0a0007c712",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:08e1120bb37c91a2589396679208dbbc7f7b037daee03624cecf7c0a0007c712"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:6b2c43c15bb4",
+      "dbms": "umbra",
+      "title": "Assertion failure when using COALESCE and CASE WHEN",
+      "reported_date": "2024-01-18",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Assertion failure when using COALESCE and CASE WHEN",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:6b2c43c15bb47385439350d0053f9d7f881555e1f6e7bea8e10c721066fab3b0",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:6b2c43c15bb47385439350d0053f9d7f881555e1f6e7bea8e10c721066fab3b0"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:6bd9ed8b6ae0",
+      "dbms": "umbra",
+      "title": "ERROR: AddressSanitizer: use-after-poison",
+      "reported_date": "2024-01-18",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "ERROR: AddressSanitizer: use-after-poison",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:6bd9ed8b6ae0ea94ccd0ba60ebe7f0ce1c30d551db7eb6a5cec13dc09726518a",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:6bd9ed8b6ae0ea94ccd0ba60ebe7f0ce1c30d551db7eb6a5cec13dc09726518a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:8c4dec4dd7a2",
+      "dbms": "umbra",
+      "title": "Unexpected results when Using BETWEEN With VIEW",
+      "reported_date": "2024-04-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results when Using BETWEEN With VIEW",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:8c4dec4dd7a2819ca6f3827451da64c77f12170e2274b2ff580365e1e2f2988a",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:8c4dec4dd7a2819ca6f3827451da64c77f12170e2274b2ff580365e1e2f2988a"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:681fecb85251",
+      "dbms": "umbra",
+      "title": "Unexpected results when Using IN expression With INDEX",
+      "reported_date": "2024-04-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results when Using IN expression With INDEX",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:681fecb85251c73abf96e6c7df873ebae15e72e39d18d789f915f26e2bd9dda1",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:681fecb85251c73abf96e6c7df873ebae15e72e39d18d789f915f26e2bd9dda1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:aea49cb44d31",
+      "dbms": "umbra",
+      "title": "Crash when using INNER JOIN",
+      "reported_date": "2024-04-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Crash when using INNER JOIN",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:aea49cb44d3107d0b53eee13a016add3e2897de2f6460e964a1de8e6bfb4f0a9",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:aea49cb44d3107d0b53eee13a016add3e2897de2f6460e964a1de8e6bfb4f0a9"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:b7bf257c32e2",
+      "dbms": "umbra",
+      "title": "Unexpected result when using LEFT JOIN and VIEW",
+      "reported_date": "2024-04-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected result when using LEFT JOIN and VIEW",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:b7bf257c32e28ae79cf45fff314f97b8a573794b239824aeeb3a71f23a04ed61",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:b7bf257c32e28ae79cf45fff314f97b8a573794b239824aeeb3a71f23a04ed61"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:b708468ecabb",
+      "dbms": "umbra",
+      "title": "Unexpected Results when Using IS NULL with empty VIEW",
+      "reported_date": "2024-06-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Results when Using IS NULL with empty VIEW",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:b708468ecabb31a4d54237b8828d6883cd093822ba0a7f8ab756ed7a06d64102",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:b708468ecabb31a4d54237b8828d6883cd093822ba0a7f8ab756ed7a06d64102"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:1f821e4ee093",
+      "dbms": "umbra",
+      "title": "Unexpected Results when Using NOT IN expression",
+      "reported_date": "2024-06-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Results when Using NOT IN expression",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:1f821e4ee093ed4c599e163c9c319d4e071da9ac957e24b492a0bc387006e980",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:1f821e4ee093ed4c599e163c9c319d4e071da9ac957e24b492a0bc387006e980"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:1e8a03afc46e",
+      "dbms": "umbra",
+      "title": "Unexpected result when using LEFT JOIN and NULL values",
+      "reported_date": "2024-06-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected result when using LEFT JOIN and NULL values",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:1e8a03afc46e567d08d9585d8344fb666b83957fc03e3bc21a0bbf673f57dfb2",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:1e8a03afc46e567d08d9585d8344fb666b83957fc03e3bc21a0bbf673f57dfb2"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:4d7fe2536c0c",
+      "dbms": "umbra",
+      "title": "Unexpected result when using NULLIF and CAST in VIEW",
+      "reported_date": "2024-06-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected result when using NULLIF and CAST in VIEW",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:4d7fe2536c0c499ba84fc458ed4c1dda24213cf49e61b6689b8b31865e9c6dad",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:4d7fe2536c0c499ba84fc458ed4c1dda24213cf49e61b6689b8b31865e9c6dad"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:0bc84ae3a875",
+      "dbms": "umbra",
+      "title": "Unexpected Results when Using CAST",
+      "reported_date": "2024-07-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Results when Using CAST",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:0bc84ae3a875e17b138f53fac420522977cbd5e59520947f8c68a5a8f9b55b3c",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:0bc84ae3a875e17b138f53fac420522977cbd5e59520947f8c68a5a8f9b55b3c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:5b199cc16583",
+      "dbms": "umbra",
+      "title": "Unexpected Results when Using IN and BETWEEN AND for VIEW",
+      "reported_date": "2024-07-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Results when Using IN and BETWEEN AND for VIEW",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:5b199cc165835c3d88dc4729fa042dbb230ba43ac1a451b4a21b01ede00fe7f7",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:5b199cc165835c3d88dc4729fa042dbb230ba43ac1a451b4a21b01ede00fe7f7"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:5c82967626cb",
+      "dbms": "umbra",
+      "title": "FATAL ERROR when Using NATURAL FULL JOIN",
+      "reported_date": "2024-08-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "FATAL ERROR when Using NATURAL FULL JOIN",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:5c82967626cbf410c634d185d2ea63cf5967097ceb15d82f027c7b4d5d502c7e",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:5c82967626cbf410c634d185d2ea63cf5967097ceb15d82f027c7b4d5d502c7e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:28437371ed70",
+      "dbms": "umbra",
+      "title": "Unexpected Results when Using IN and BETWEEN AND for VIEW",
+      "reported_date": "2024-08-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Results when Using IN and BETWEEN AND for VIEW",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:28437371ed7094e200f681f943fd7085fe85ffe04570233588dae786bc0db2aa",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:28437371ed7094e200f681f943fd7085fe85ffe04570233588dae786bc0db2aa"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:8f4de0f223c7",
+      "dbms": "umbra",
+      "title": "Unexpected results when using logical operator in VIEW",
+      "reported_date": "2024-08-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected results when using logical operator in VIEW",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:8f4de0f223c7e099c575d5376218a61c7efba71f4c6cfb48a63de0607fd811af",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:8f4de0f223c7e099c575d5376218a61c7efba71f4c6cfb48a63de0607fd811af"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:740d3dcc922d",
+      "dbms": "umbra",
+      "title": "Segmentation fault when Using NATURAL FULL JOIN",
+      "reported_date": "2024-10-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Segmentation fault when Using NATURAL FULL JOIN",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:740d3dcc922d6c89fccef6f2bca3abd2bf1de3e99b5faef29151f4418e115d68",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:740d3dcc922d6c89fccef6f2bca3abd2bf1de3e99b5faef29151f4418e115d68"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:f94977b0fb2d",
+      "dbms": "umbra",
+      "title": "Unexpected Results when Using Multiple Comparison",
+      "reported_date": "2024-10-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Results when Using Multiple Comparison",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:f94977b0fb2df375dc2826531af0441c383daf1f3136503ad0dcd6e834fc047b",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:f94977b0fb2df375dc2826531af0441c383daf1f3136503ad0dcd6e834fc047b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:964b628ada1b",
+      "dbms": "umbra",
+      "title": "Unexpected Results when Using LEFT JOIN with VIEW on IS NOT NULL",
+      "reported_date": "2024-11-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Results when Using LEFT JOIN with VIEW on IS NOT NULL",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:964b628ada1b5a6b8b62fbccb7d5b16bde6c6501a95f11954f582fe290abb28f",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:964b628ada1b5a6b8b62fbccb7d5b16bde6c6501a95f11954f582fe290abb28f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:42996e568f2c",
+      "dbms": "umbra",
+      "title": "Assertion failure when using UNION ALL",
+      "reported_date": "2024-12-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Assertion failure when using UNION ALL",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:42996e568f2c686f54ba488a95be9f2544213695cc7efb5cc5bbe9a72f93c4c1",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:42996e568f2c686f54ba488a95be9f2544213695cc7efb5cc5bbe9a72f93c4c1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:78c2a07f9508",
+      "dbms": "umbra",
+      "title": "Unexpected Abort",
+      "reported_date": "2024-12-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected Abort",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Suyang Zhong. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:78c2a07f9508198b5392a97ccfd776a1c4a16e394a190d98f64488d53697b61d",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:78c2a07f9508198b5392a97ccfd776a1c4a16e394a190d98f64488d53697b61d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:95ae121f79a7",
+      "dbms": "umbra",
+      "title": "Unexpected lateral join Result",
+      "reported_date": "2025-06-30",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "Unexpected lateral join Result",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Zhaokun Xiang. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:95ae121f79a75e217a286104ae7fb0d8d2e78abbae3b67f8fb0fd0b2bfa207b8",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:95ae121f79a75e217a286104ae7fb0d8d2e78abbae3b67f8fb0fd0b2bfa207b8"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:f960db0a16dc",
+      "dbms": "umbra",
+      "title": "An unexpected core dumped in index creation",
+      "reported_date": "2025-07-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "An unexpected core dumped in index creation",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Zhaokun Xiang. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:f960db0a16dcbb1dbe2f69aad15762bb54292985026e1c7699d565a6c027245c",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:f960db0a16dcbb1dbe2f69aad15762bb54292985026e1c7699d565a6c027245c"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:e0c96f830ea0",
+      "dbms": "umbra",
+      "title": "An unexpected internal error",
+      "reported_date": "2025-07-16",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "An unexpected internal error",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Zhaokun Xiang. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:e0c96f830ea012ae45b3d69bbfcfb15076523fc38059bf1873cd9cef41783fbf",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:e0c96f830ea012ae45b3d69bbfcfb15076523fc38059bf1873cd9cef41783fbf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:61160359adf3",
+      "dbms": "umbra",
+      "title": "An unexpected core dump in multiple joins",
+      "reported_date": "2025-07-23",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "An unexpected core dump in multiple joins",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Zhaokun Xiang. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:61160359adf3a6183d7f58b208a78fa93de2c0f372aa6987ad9956cf2451a197",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:61160359adf3a6183d7f58b208a78fa93de2c0f372aa6987ad9956cf2451a197"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:1fa073d1d868",
+      "dbms": "umbra",
+      "title": "An unexpected core dumped in multiple joins",
+      "reported_date": "2025-08-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "An unexpected core dumped in multiple joins",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Zhaokun Xiang. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:1fa073d1d868abc6a6ccbb825adb8babc990e3739d6724ef1596a0a752bc28e6",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:1fa073d1d868abc6a6ccbb825adb8babc990e3739d6724ef1596a0a752bc28e6"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:7a2cae0c77d9",
+      "dbms": "umbra",
+      "title": "An unexpected anti-join logic bug",
+      "reported_date": "2025-09-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "An unexpected anti-join logic bug",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Zhaokun Xiang. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:7a2cae0c77d94fd1f84b89128600e47bb0b014b489b33a716956153c758a665f",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:7a2cae0c77d94fd1f84b89128600e47bb0b014b489b33a716956153c758a665f"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:324e4dd9782e",
+      "dbms": "umbra",
+      "title": "An unexpected internal error in joins",
+      "reported_date": "2025-09-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "An unexpected internal error in joins",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Zhaokun Xiang. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:324e4dd9782e260645ee49726e510fe8648fbf4393f9aec89642dbabf764aa5b",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:324e4dd9782e260645ee49726e510fe8648fbf4393f9aec89642dbabf764aa5b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:0016ae2a2af7",
+      "dbms": "umbra",
+      "title": "An unexpected internal error in an equivalent join",
+      "reported_date": "2025-10-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "An unexpected internal error in an equivalent join",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Zhaokun Xiang. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:0016ae2a2af725c207f178ddbc49d0544ab087ecdb7a339e40ddad62a6af8171",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:0016ae2a2af725c207f178ddbc49d0544ab087ecdb7a339e40ddad62a6af8171"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:umbra:5e3c6889396c",
+      "dbms": "umbra",
+      "title": "An unexpected internal error in a left join",
+      "reported_date": "2025-11-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Zhaokun Xiang",
+      "links": {
+        "record": "https://nus-test.github.io/bugs/"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": "An unexpected internal error in a left join",
+            "excerpt_is_verbatim": true,
+            "note": "Listed on the NUS TEST lab bug page as a Umbra bug reported by Zhaokun Xiang. The report was sent to the developers directly, so the list is its only public record.",
+            "content_sha256": "sha256:5e3c6889396ceb36b3c4ca88b4cb4d1095028e78162803c144bffde7b618e93b",
+            "retrieved_at": "2026-09-08T16:41:32Z",
+            "first_seen": "2026-09-08T16:41:32Z",
+            "last_verified": "2026-09-08T16:41:32Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-08T16:41:32Z",
+        "last_verified": "2026-09-08T16:41:32Z",
+        "source_url": "https://nus-test.github.io/bugs/",
+        "source_type": "website",
+        "content_sha256": "sha256:5e3c6889396ceb36b3c4ca88b4cb4d1095028e78162803c144bffde7b618e93b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:a1822b7ec0b1",
+      "dbms": "virtuoso",
+      "title": "Unexpected Results when using LEFT JOIN with NULL",
+      "reported_date": "2024-01-14",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1236"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1236",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1236",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR);\r\nCREATE TABLE t1(c0 INTEGER);\r\nINSERT INTO t1 (c0) VALUES (2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9566b317146b7d801db34c857b51e722f37ebe490991124ec110859a4c2253c5",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9566b317146b7d801db34c857b51e722f37ebe490991124ec110859a4c2253c5"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:387fb4ec5c4f",
+      "dbms": "virtuoso",
+      "title": "Unexpected results when using math functions",
+      "reported_date": "2024-01-14",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1235"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1235",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1235",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 INT, PRIMARY KEY(c0));\r\nINSERT INTO t0 (c0, c1) VALUES (1, 1);\r\nINSERT INTO t0 (c0) VALUES (-1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3e4dfb73856704189e00ec2a03a404cd183c32ce111e67f1fab23af7eef1303d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1235",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3e4dfb73856704189e00ec2a03a404cd183c32ce111e67f1fab23af7eef1303d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:304c5d9523b4",
+      "dbms": "virtuoso",
+      "title": "Unexpected results when using trigonometric functions",
+      "reported_date": "2024-01-14",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1213"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1213",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1213",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INTEGER);\r\nINSERT INTO t0 (c0) VALUES (1);\r\n\r\n-- At least one of the query should return empty result",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:b243983398e426cd4332760bfd32ee8eabcb43923fe584a06826da81efabdabf",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:b243983398e426cd4332760bfd32ee8eabcb43923fe584a06826da81efabdabf"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:3c619e4c29f9",
+      "dbms": "virtuoso",
+      "title": "Crash by CASE WHEN and unknown identifier",
+      "reported_date": "2024-01-19",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1239"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1239",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1239",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 VARCHAR(500), c2 INTEGER, PRIMARY KEY(c1));\r\nCREATE TABLE t1(c0 INTEGER);\r\nINSERT INTO t0(c0, c1) VALUES ('eኧ', '');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:afeac2f20219bc24f5770933099db04ff15e37b00c0c05c663027ba14997e1fa",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:afeac2f20219bc24f5770933099db04ff15e37b00c0c05c663027ba14997e1fa"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:0aa0d100bf9a",
+      "dbms": "virtuoso",
+      "title": "Unexpected results when using LEFT JOIN with NULL as predicate",
+      "reported_date": "2024-01-19",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1238"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1238",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1238",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500));\r\nCREATE TABLE t1(c0 INTEGER, c1 INTEGER);\r\nINSERT INTO t1 (c0) VALUES (1);\r\nINSERT INTO t1 (c1) VALUES (2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:8f173e9404f1aee576a209b42a418cfcc56a5c73b971c3fac4070d1324839337",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:8f173e9404f1aee576a209b42a418cfcc56a5c73b971c3fac4070d1324839337"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:e078ac653e28",
+      "dbms": "virtuoso",
+      "title": "Unexpected results when Using RIGHT JOIN",
+      "reported_date": "2024-02-01",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1214"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1214",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1214",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c1 INT);\r\nINSERT INTO t1 (c1) VALUES (1);\r\nSELECT * FROM t0 RIGHT JOIN t1 ON 1; -- NULL 1",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:1729145003a70ec6a74ef93fc61d881596ed6d31d1781ff2906c6a7bb735aadc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:1729145003a70ec6a74ef93fc61d881596ed6d31d1781ff2906c6a7bb735aadc"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:8ce2aae1cff1",
+      "dbms": "virtuoso",
+      "title": "Potential issue when multi-threading testing using virtuoso jdbc driver",
+      "reported_date": "2024-02-07",
+      "reported_year": 2024,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "suyZhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1242"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1242",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1242",
+            "source_type": "github_issue",
+            "excerpt": "Potential issue when multi-threading testing using virtuoso jdbc driver",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:1b61a9dabd27fdf975938b9a07107118addb9cd6529b59aad0be266653b87942",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "suyZhong is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:10:48Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:10:48Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1242",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1b61a9dabd27fdf975938b9a07107118addb9cd6529b59aad0be266653b87942"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:7a7199f6ebdb",
+      "dbms": "virtuoso",
+      "title": "Crash by CASE WHEN",
+      "reported_date": "2024-03-14",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1249"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1249",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1249",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 VARCHAR(500), PRIMARY KEY(c0));\r\nCREATE TABLE t1(c0 INTEGER, PRIMARY KEY(c0));\r\nINSERT INTO t1(c0) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:fc9b4e543ffcbfe18a59acf5efa232c17989a197c8beb97f87b66218adbbafdb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:fc9b4e543ffcbfe18a59acf5efa232c17989a197c8beb97f87b66218adbbafdb"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:38a7f6bc0209",
+      "dbms": "virtuoso",
+      "title": "Unexpected results when using LEFT JOIN and UNION",
+      "reported_date": "2024-03-14",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1250"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1250",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1250",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 VARCHAR(500));\r\nINSERT INTO t0 (c0) VALUES (1);\r\nINSERT INTO t1 (c0) VALUES ('a');",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:9a24b0d49d6b0ffd486cc80addc275b1e9f935913a6ccced95fae985167d6299",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:9a24b0d49d6b0ffd486cc80addc275b1e9f935913a6ccced95fae985167d6299"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:f999cb97bbb7",
+      "dbms": "virtuoso",
+      "title": "Unexpected results when using INNER JOIN and UNION",
+      "reported_date": "2024-03-15",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1251"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1251",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1251",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT);\r\nINSERT INTO t1 ( c0) VALUES (1);\r\nINSERT INTO t0 ( c0) VALUES (2);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:fc1ef36a3998a8dea008c727d3e167c5bfdececbeb9bb2b44f55a65493cc8d77",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:fc1ef36a3998a8dea008c727d3e167c5bfdececbeb9bb2b44f55a65493cc8d77"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:3a9f038df009",
+      "dbms": "virtuoso",
+      "title": "Unexpected results when using ORDER BY",
+      "reported_date": "2024-07-02",
+      "reported_year": 2024,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "Suyang Zhong",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1241"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1241",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1241",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\r\nCREATE TABLE t1(c0 INT);\r\nINSERT INTO t1 (c0) VALUES (1);",
+            "excerpt_is_verbatim": true,
+            "note": "Reported by Suyang Zhong of the TEST lab, with a reproducer using SQLancer's generated-schema convention (tables t0, t1, ...; columns c0, c1, ...).",
+            "content_sha256": "sha256:d3c14602bdcf1025b0b528600f62673bed157db0b0ca7b818e3dd5cb02d90316",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T01:44:41Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/bugs/",
+            "source_type": "website",
+            "excerpt": null,
+            "note": "Listed on the NUS TEST lab bug page, reported by Suyang Zhong.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "nus_test",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T01:44:41Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+        "source_type": "website",
+        "content_sha256": "sha256:d3c14602bdcf1025b0b528600f62673bed157db0b0ca7b818e3dd5cb02d90316"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:virtuoso:de4519128218",
+      "dbms": "virtuoso",
+      "title": "Logic Error: Incorrect `COUNT(*)` results in TLP query rewrite involving `NULL` predicate and `CASE` in Virtuoso",
+      "reported_date": "2026-05-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "Jasper0209",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1424"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1424",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1424",
+            "source_type": "github_issue",
+            "excerpt": "TLP (tautology-based partitioning) rewritten query",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:844c2e9357fa0994cb891700dc7c28e24bf4ad902faa67c95eef9c2c6a171103",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:c892c4dbbe00cba0e98a5bfaf9ca3033a2673b7cf48a7adcf81b8b944cfd7b9f",
+          "classified_at": "2026-09-13T06:23:45Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1424",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:844c2e9357fa0994cb891700dc7c28e24bf4ad902faa67c95eef9c2c6a171103"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:virtuoso:a70235f75799",
+      "dbms": "virtuoso",
+      "title": "Logic Error: TLP query rewrite over `EXISTS` subquery produces incorrect result due to predicate rewriting and column duplication in Virtuoso",
+      "reported_date": "2026-05-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "Jasper0209",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1427"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1427",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1427",
+            "source_type": "github_issue",
+            "excerpt": "TLP query rewrite over `EXISTS` subquery",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:50e887ca248a51506b8744f62e6a0f898560c3869a39a28d6ca2b6fc3bb9f05e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:5ce9e949816acb54d815dd1b23a46772182060f610c591976f905954dd1ba49c",
+          "classified_at": "2026-09-13T06:23:45Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1427",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:50e887ca248a51506b8744f62e6a0f898560c3869a39a28d6ca2b6fc3bb9f05e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:virtuoso:c1365401c662",
+      "dbms": "virtuoso",
+      "title": "Logic Error: TLP query rewrite over `LEFT JOIN` with `NULL` columns causes double-counting in Virtuoso",
+      "reported_date": "2026-05-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "Jasper0209",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1426"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1426",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1426",
+            "source_type": "github_issue",
+            "excerpt": "TLP (tautology-based partitioning) rewritten query",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:216b989f53722182fbb34c15248575f645332dc30ca4ca1a46bcd8df973109d1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:1a9ebea79b61c7ce025baa07bd3f928e9d167559f525cf3c0248189b7288d1fc",
+          "classified_at": "2026-09-13T06:23:45Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1426",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:216b989f53722182fbb34c15248575f645332dc30ca4ca1a46bcd8df973109d1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:virtuoso:e57d688c4d05",
+      "dbms": "virtuoso",
+      "title": "Logic Error: TLP query rewrite with `NULL` filter produces incorrect count due to expression injection and predicate distortion in Virtuoso",
+      "reported_date": "2026-05-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "Jasper0209",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1428"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1428",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1428",
+            "source_type": "github_issue",
+            "excerpt": "TLP (tautology-based partitioning) rewritten query",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:d9dbfe77c1b510c9df57b5529a31f66bf522114be485ac013302488704208514",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:1f77ed9c5bf77856723738584f5800d0d8fb8fc9f6c271caf25c2c74b7925d9b",
+          "classified_at": "2026-09-13T06:23:45Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1428",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d9dbfe77c1b510c9df57b5529a31f66bf522114be485ac013302488704208514"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:virtuoso:7213bf310616",
+      "dbms": "virtuoso",
+      "title": "Logic Error: TLP query rewrite with `OR` and `NULL` semantics undercounts rows in Virtuoso",
+      "reported_date": "2026-05-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "Jasper0209",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1425"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1425",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1425",
+            "source_type": "github_issue",
+            "excerpt": "TLP (tautology-based partitioning) rewritten query",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:8f01e57e454930a285aa470d9c5c0ed881991e3fd3b45f2d7272669a67a58a31",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:f8b813d668800b10a4b53880a45ef6a0fcb355a5d6caa4e4dfa5545ff046535b",
+          "classified_at": "2026-09-13T06:23:45Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1425",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8f01e57e454930a285aa470d9c5c0ed881991e3fd3b45f2d7272669a67a58a31"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:virtuoso:44a713a98e0f",
+      "dbms": "virtuoso",
+      "title": "Logic Error: TLP query rewrite with tautological predicates and redundant `OR`-`AND` expansion causes undercounting in Virtuoso",
+      "reported_date": "2026-05-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "Jasper0209",
+      "links": {
+        "report": "https://github.com/openlink/virtuoso-opensource/issues/1429"
+      },
+      "primary_url": "https://github.com/openlink/virtuoso-opensource/issues/1429",
+      "attribution": {
+        "rule": "technique_attribution",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1429",
+            "source_type": "github_issue",
+            "excerpt": "TLP (tautology-based partitioning) rewritten query",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c3c6af99b720f78e6e16f77e23330177db8ea24b431b27c336ac5de1de67a9c9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:73c9a90c6111501de92243677ea11e8db45c954dfb4429177b44309c680136a2",
+          "classified_at": "2026-09-13T06:23:45Z",
+          "model": "claude-opus-5",
+          "rationale": "The report credits a SQLancer-originated oracle with the find. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/openlink/virtuoso-opensource/issues/1429",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c3c6af99b720f78e6e16f77e23330177db8ea24b431b27c336ac5de1de67a9c9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:a61c07624b77",
+      "dbms": "wadjet",
+      "title": "CONCAT()/|| with a non-text argument after position 0 crashes the entire server (vecConcat indexes empty Offsets)",
+      "reported_date": "2026-08-24",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/509"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/509",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/509",
+            "source_type": "github_issue",
+            "excerpt": "Found by the full SQLancer soak (wadjet#289), `--oracle WHERE`, multiple\nseeds, wadjet main @ `1cf758ba`.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:baf9401711f42056ee7b924c7d0e9b14d1b613bf32dbb4e3012cf05fd903fd1e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/509",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:baf9401711f42056ee7b924c7d0e9b14d1b613bf32dbb4e3012cf05fd903fd1e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:c5a4587bd037",
+      "dbms": "wadjet",
+      "title": "HashJoin build-side panic (fatalEval, e.g. invalid int cast) crashes the entire server — unrecovered goroutine in buildJoin",
+      "reported_date": "2026-08-24",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/508"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/508",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/508",
+            "source_type": "github_issue",
+            "excerpt": "Found by the full SQLancer soak (wadjet#289), `--oracle NOREC`, `--random-seed 777`\n(SQLancer database index 0 — the first task in a `--num-threads 1` run), wadjet",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:bb7b05f5922273149da052a1775c7e955c0a16aadb64021c890b9e41e7ad3a03",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/508",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bb7b05f5922273149da052a1775c7e955c0a16aadb64021c890b9e41e7ad3a03"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:5fb704e59cc3",
+      "dbms": "wadjet",
+      "title": "Only the designed FatalEvalPanic class survives a query error — any other panic (index-out-of-range, nil deref, ...) crashes the whole server",
+      "reported_date": "2026-08-24",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/511"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/511",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/511",
+            "source_type": "github_issue",
+            "excerpt": "Found while triaging the full SQLancer soak (wadjet#289); filed after\n#508, #509, #510 made the pattern across them clear.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:b4927c273b6d089567a51c86fad468fa299cb67850ed2841de1f9ee4329232a0",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/511",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b4927c273b6d089567a51c86fad468fa299cb67850ed2841de1f9ee4329232a0"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:b18f39daa3e6",
+      "dbms": "wadjet",
+      "title": "Pipeline.runParallel's firstErr atomic.Value panics on a concrete-type race between two workers' errors",
+      "reported_date": "2026-08-24",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/512"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/512",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/512",
+            "source_type": "github_issue",
+            "excerpt": "Found by the full SQLancer soak (wadjet#289), `--oracle HAVING`, wadjet main\n@ `1cf758ba`.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:9e81d7add0b4344a6c0da753c22c7e3da21362c6426288d4bd4dfc1ca5421556",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/512",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9e81d7add0b4344a6c0da753c22c7e3da21362c6426288d4bd4dfc1ca5421556"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:3ea4a0417ae4",
+      "dbms": "wadjet",
+      "title": "Table manifest cached at first read is never invalidated: stale writes, and DROP+CREATE of the same name silently resurrects old data",
+      "reported_date": "2026-08-24",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/483"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/483",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/483",
+            "source_type": "github_issue",
+            "excerpt": "through the real pgwire path (`internal/server/pgwire`), found while\nstanding up the SQLancer harness for #289 (before any generator ran —",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:a17e27c244cd85f0693bc4004c6075260900f87e98d8e7324ff4548c8aff6c62",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/483",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a17e27c244cd85f0693bc4004c6075260900f87e98d8e7324ff4548c8aff6c62"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:e60110cbd08e",
+      "dbms": "wadjet",
+      "title": "joinFlushSource.Close() dereferences a nil pipeline when Init() never ran, crashing the entire server",
+      "reported_date": "2026-08-24",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/510"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/510",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/510",
+            "source_type": "github_issue",
+            "excerpt": "Found by the full SQLancer soak (wadjet#289), `--oracle WHERE`, wadjet main\n@ `1cf758ba`.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:9f000267caf901d3da515c7d1663cbe74fbd5b651f25bfb5d13a69d13c6e8a78",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/510",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9f000267caf901d3da515c7d1663cbe74fbd5b651f25bfb5d13a69d13c6e8a78"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:af0ba3fcf48c",
+      "dbms": "wadjet",
+      "title": "CAST(int AS BOOLEAN) used as a WHERE filter always excludes every row, though the identical expression projects correctly in SELECT",
+      "reported_date": "2026-08-25",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/592"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/592",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/592",
+            "source_type": "github_issue",
+            "excerpt": "Found by the standing SQLancer soak (#289), `--oracle QUERY_PARTITIONING`\n(TLP-WHERE arm, `sqlancer.common.oracle.TLPWhereOracle`), wadjet main @",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:37f9df00e128ff8e32b7d03ff0787ebf90884ebb00fec060594e86aebba5a2e6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/592",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:37f9df00e128ff8e32b7d03ff0787ebf90884ebb00fec060594e86aebba5a2e6"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:6f57a5c4e7e2",
+      "dbms": "wadjet",
+      "title": "GROUP BY present: checkUngrouped's own scope note is wrong — an ungrouped SELECT/HAVING column is silently corrupted, not just unchecked",
+      "reported_date": "2026-08-25",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/590"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/590",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/590",
+            "source_type": "github_issue",
+            "excerpt": "Found by the standing SQLancer soak (#289) — TLP-HAVING (`--oracle HAVING`) and\nTLP-Aggregate (via `--oracle QUERY_PARTITIONING`), wadjet main @ 6c40c829.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:70534c4c0e30dd0ff4f1c7bb23ecd520d21ffdead684a7f7c2c0de357ea6dc15",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/590",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:70534c4c0e30dd0ff4f1c7bb23ecd520d21ffdead684a7f7c2c0de357ea6dc15"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:4fbb3e1e72ff",
+      "dbms": "wadjet",
+      "title": "HAVING with an aggregate predicate: an internal __having_0 alias leaks into results, and \"agg(...) IS [NOT] NULL\" ignores the aggregate's value",
+      "reported_date": "2026-08-25",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/591"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/591",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/591",
+            "source_type": "github_issue",
+            "excerpt": "Found by the standing SQLancer soak (#289), `--oracle HAVING` (TLP-HAVING),\nwadjet main @ 6c40c829. Distinct from #590 (which is about a bare,",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:a51a9f7aae737b741d78e61af6830ca480baa7d05dd7d2c91a67a1e067156b01",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/591",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a51a9f7aae737b741d78e61af6830ca480baa7d05dd7d2c91a67a1e067156b01"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:2619ab70e598",
+      "dbms": "wadjet",
+      "title": "FULL OUTER JOIN + aggregate: filtering by an always-true predicate changes the aggregate's answer (BOOL_OR NULL -> FALSE) — the multi-table-join TLP-Aggregate lead from the prior soak, now reproduced",
+      "reported_date": "2026-08-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/622"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/622",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/622",
+            "source_type": "github_issue",
+            "excerpt": "Found by the standing pre-release SQLancer soak (#289), TLP-Aggregate\n(via `--oracle QUERY_PARTITIONING`), wadjet main @ a8be040d.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:a7a90a3ddd57b1eae6bb0431c391b64dba37d29b0463f107847e3ae5a2f686a3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/622",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a7a90a3ddd57b1eae6bb0431c391b64dba37d29b0463f107847e3ae5a2f686a3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:c6c8011e668e",
+      "dbms": "wadjet",
+      "title": "GROUP BY + explicit JOIN: an ungrouped column from the joined table is not checked (SELECT silently wrong, HAVING silently excludes everything) — #590's fix doesn't cover this FROM shape",
+      "reported_date": "2026-08-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/620"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/620",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/620",
+            "source_type": "github_issue",
+            "excerpt": "Found by the standing pre-release SQLancer soak (#289), TLP-WHERE and\nTLP-HAVING (`--oracle WHERE`/`HAVING`/`QUERY_PARTITIONING`), wadjet",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:98d4400749b1844958c515a72e8de97d54b0cf640dc0428e8a5e6e1a285e93d8",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/620",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:98d4400749b1844958c515a72e8de97d54b0cf640dc0428e8a5e6e1a285e93d8"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:23c43e49ae18",
+      "dbms": "wadjet",
+      "title": "HAVING <agg>(<constant>) always evaluates false: MIN(1) > 0 and BOOL_AND(TRUE) exclude every group, though the same aggregate over a real column works",
+      "reported_date": "2026-08-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/621"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/621",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/621",
+            "source_type": "github_issue",
+            "excerpt": "Found by the standing pre-release SQLancer soak (#289), TLP-HAVING\n(`--oracle HAVING`/`QUERY_PARTITIONING`), wadjet main @ a8be040d.",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:4340ce479f9700a069567176851a68a00b2d4984d620a5a45f91404fcb2954eb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/621",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4340ce479f9700a069567176851a68a00b2d4984d620a5a45f91404fcb2954eb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:5479bcab560e",
+      "dbms": "wadjet",
+      "title": "exec: vectorized OrFilter reads an all-rows branch (nil Sel) as zero rows",
+      "reported_date": "2026-08-28",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/623"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/623",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/623",
+            "source_type": "github_issue",
+            "excerpt": "UnionsAnAllRowsBranch` (0 rows pre-fix). Filing separately because it is a distinct root cause from #622's aggregate-qualifier bug and lives in shared OR-filter code. Found by the v0.18.3 pre-release SQLancer soak.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:e544bcea803394efae3b0b387b70d5926016bdfb49608fe7b220b0315dec66f1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/623",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:e544bcea803394efae3b0b387b70d5926016bdfb49608fe7b220b0315dec66f1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:3b24c69d0f48",
+      "dbms": "wadjet",
+      "title": "pgwire: SELECT * returning zero rows sends no RowDescription — psql prints nothing, JDBC throws \"No results were returned\"",
+      "reported_date": "2026-09-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/846"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/846",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/846",
+            "source_type": "github_issue",
+            "excerpt": "`SELECT *` that returns ZERO rows emits no RowDescription at all on the wire: psql prints nothing, and JDBC's `executeQuery` throws `No results were returned by the query`. PostgreSQL always sends a RowDescription. Measured on 9ac9ab23 (pre-existing) by the flake arc's review — it is what broke three of five SQLancer fixture dumps.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:99e3edf81cbe05a3fc6309801bc9f34265dd68a8556ba532b2afb2f97ad2c044",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/846",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:99e3edf81cbe05a3fc6309801bc9f34265dd68a8556ba532b2afb2f97ad2c044"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:feb559d09063",
+      "dbms": "wadjet",
+      "title": "planner: UNION ALL with two output columns of the same bare name in a branch binds the first to the last's value (silent; #556 family)",
+      "reported_date": "2026-09-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/844"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/844",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/844",
+            "source_type": "github_issue",
+            "excerpt": "`UNION ALL` where a branch projects two columns with the SAME bare name binds the first to the LAST's value — the first `c0` takes the value of the second `c0`. Silent wrong answer, single-process route; measured on 9ac9ab23 against PostgreSQL 17 by the flake arc's review (three of five SQLancer TLP violations reduce to it).",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:725992a4645afad92f362c19586fcbad592c86801770de1a143ba66ee06f71c1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/844",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:725992a4645afad92f362c19586fcbad592c86801770de1a143ba66ee06f71c1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:wadjet:316bd44ee5ef",
+      "dbms": "wadjet",
+      "title": "planner: inside a derived table with two or more unaliased relations every base scan is re-aliased to the derived table's alias — qualified references bind to the wrong relation, silently",
+      "reported_date": "2026-09-03",
+      "reported_year": 2026,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "derekmwright",
+      "links": {
+        "report": "https://github.com/derekmwright/wadjet/issues/843"
+      },
+      "primary_url": "https://github.com/derekmwright/wadjet/issues/843",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/derekmwright/wadjet/issues/843",
+            "source_type": "github_issue",
+            "excerpt": "Inside a derived table (a subquery in FROM) with two or more UNALIASED base relations, every base scan is re-aliased to the DERIVED TABLE's alias, so a qualified column reference binds to whichever relation was planned last. Silent wrong answer on the default single-process route; measured on 9ac9ab23 (v0.18.22 + docs) against PostgreSQL 17 by the flake arc's review, reducing five SQLancer TLP violations to this one mechanism.",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:55aec6fb23a27810bc83a870968bb4165fb40363bf3e2e28cc7340549d563b27",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:54:23Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:54:23Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/derekmwright/wadjet/issues/843",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:55aec6fb23a27810bc83a870968bb4165fb40363bf3e2e28cc7340549d563b27"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:97d540f070a4",
+      "dbms": "yugabytedb",
+      "title": "[YCQL][SQLancer] Select NULL from table leads to dropped connection",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/yugabyte/YugabyteBugs.java",
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/14330"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/yugabyte/YugabyteBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/yugabyte/yugabyte-db/issues/14330",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its yugabytedb provider, as the field bug14330 that works around it.",
+            "content_sha256": "sha256:97d540f070a494f67fa2069baa3ed51f9aa3f3ff0bc56f5d8fce07d4d7b05060",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/yugabyte/YugabyteBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:97d540f070a494f67fa2069baa3ed51f9aa3f3ff0bc56f5d8fce07d4d7b05060"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/14330",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:yugabytedb:7d1318a3a889",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] Colocation - Table not found in Raft group exception occurs during testing",
+      "reported_date": null,
+      "reported_year": null,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": null,
+      "links": {
+        "record": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/yugabyte/YugabyteBugs.java",
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11357"
+      },
+      "attribution": {
+        "rule": "curated_primary_source",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/yugabyte/YugabyteBugs.java",
+            "source_type": "github_repository",
+            "excerpt": "https://github.com/yugabyte/yugabyte-db/issues/11357",
+            "excerpt_is_verbatim": true,
+            "note": "sqlancer/sqlancer records this bug in its yugabytedb provider, as the field bug11357 that works around it.",
+            "content_sha256": "sha256:7d1318a3a88916ad7068b56180db7cdf167d88cd941178cf4f156891f73c71b6",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T06:23:12Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "provider_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/yugabyte/YugabyteBugs.java",
+        "source_type": "github_repository",
+        "content_sha256": "sha256:7d1318a3a88916ad7068b56180db7cdf167d88cd941178cf4f156891f73c71b6"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11357",
+      "reporter_affiliation": "unknown"
+    },
+    {
+      "id": "bug:yugabytedb:303833c779a6",
+      "dbms": "yugabytedb",
+      "title": "sqlancer: Hitting issue when creating temporary table with PK and ON COMMIT DELETE in parallel workload",
+      "reported_date": "2021-09-29",
+      "reported_year": 2021,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/10140"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/10140",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/10140",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer: Hitting issue when creating temporary table with PK and ON COMMIT DELETE in parallel workload",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:28a5a87a32c0de69d2dcd7899b597d0a1fea0e52b167e4a1d625305aa4d919bd",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/10140",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:28a5a87a32c0de69d2dcd7899b597d0a1fea0e52b167e4a1d625305aa4d919bd"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:2a791d251d68",
+      "dbms": "yugabytedb",
+      "title": "SQLancer: Server restarts during evaluating workload",
+      "reported_date": "2021-10-01",
+      "reported_year": 2021,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/10160"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/10160",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/10160",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer: Server restarts during evaluating workload",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5e844799d31a0a7586b19d87b3e37aff0448834b0f61c302c1829a86fc819ae1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/10160",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5e844799d31a0a7586b19d87b3e37aff0448834b0f61c302c1829a86fc819ae1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:ab23dc9d4dec",
+      "dbms": "yugabytedb",
+      "title": "SQLancer: Wrong result in non-optimised query if current_query() function call used",
+      "reported_date": "2021-10-01",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/10156"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/10156",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/10156",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer: Wrong result in non-optimised query if current_query() function call used",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:daf8d6e4661d74327144d593273bcfcff0453ee2a588e4e7a0d21e73ace41ac2",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/10156",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:daf8d6e4661d74327144d593273bcfcff0453ee2a588e4e7a0d21e73ace41ac2"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:9b2bdc2dd8bf",
+      "dbms": "yugabytedb",
+      "title": "SQLancer: IOException occurs on UDPATE statement (local and cluster modes are affected)",
+      "reported_date": "2021-10-08",
+      "reported_year": 2021,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/10236"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/10236",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/10236",
+            "source_type": "github_issue",
+            "excerpt": "SQLancer: IOException occurs on UDPATE statement",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:bf4abf23119e9e8ffe9e1236da0d7263fca28c8924e6056bdadbffdbc58bd38a",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:d009f54b290299815ca53cce87b5723188d8f2c320cae0c72c694d475c419ecf",
+          "classified_at": "2026-09-13T06:23:54Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/10236",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:bf4abf23119e9e8ffe9e1236da0d7263fca28c8924e6056bdadbffdbc58bd38a"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:ede83c5864f7",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] [SQLancer] A lot of catalog consistency related exceptions during evaluating test in parallel",
+      "reported_date": "2021-10-28",
+      "reported_year": 2021,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/10434"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/10434",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/10434",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL] [SQLancer] A lot of catalog consistency related exceptions during evaluating test in parallel",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:637328b5bdc8a6b4816f7fa5b3e570b1456255d9c6fc1e6fc6fefc6fbe07b0af",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/10434",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:637328b5bdc8a6b4816f7fa5b3e570b1456255d9c6fc1e6fc6fefc6fbe07b0af"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:d7f40dfa3c35",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer] [YSQL] Results mismatch with postgres for query with CROSS JOIN WHERE TRUE if use temporary tables",
+      "reported_date": "2022-01-13",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11076"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11076",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11076",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer] [YSQL] Results mismatch with postgres",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:18aae1e4d9e9321d1406760cb85e0041f471449c8a4b05241b7b695a4558472e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:be5bd7f12f8898eacb0698ab1fb12b48c14a866a0c4e4b06cb958fa5f32ecae1",
+          "classified_at": "2026-09-13T06:23:54Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11076",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:18aae1e4d9e9321d1406760cb85e0041f471449c8a4b05241b7b695a4558472e"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:f3a02e2ac942",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer] [YSQL] Select statement may return phantom value from failed insert statement if ORDER BY used in temporary tables",
+      "reported_date": "2022-01-13",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11077"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11077",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11077",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer] [YSQL] Select statement may return phantom value",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:904d2307121283877ec9a822309b756f28ecf8145630447fc2a779090de1bb54",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:07d4eca635da3a05c638483b4277da8e9511afe0782a090645cecdaaaecfe92f",
+          "classified_at": "2026-09-13T06:23:54Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11077",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:904d2307121283877ec9a822309b756f28ecf8145630447fc2a779090de1bb54"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:78cb3753dea3",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer] [YSQL] Autocast to real doesn't work for asc PK in where clause",
+      "reported_date": "2022-01-14",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11090"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11090",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11090",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer] [YSQL] Autocast to real doesn't work for asc PK",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:ad109234393aea9e43fc9e8ce9e3983096c13d545e445b72ad1eceefcf32a402",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:e8667ed0f379bbfab6342f8bbabb89d5e710e7ed12b559cab9412fecd3afb150",
+          "classified_at": "2026-09-13T06:23:54Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11090",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ad109234393aea9e43fc9e8ce9e3983096c13d545e445b72ad1eceefcf32a402"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:3082696b409d",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer] [YSQL] Trying to drop serial PK column leads to failed inserts",
+      "reported_date": "2022-01-17",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11110"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11110",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11110",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer] [YSQL] Trying to drop serial PK column leads to failed inserts",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:873ff160520ac07d5ddd6e5a4e5ec3eb0994d4f64ae1695ed7c4155f8d94326d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:14bb09108f55d834bc0d1b94a48b2dc75165e7e1dc1cf7abff3da38c66022b1d",
+          "classified_at": "2026-09-13T06:23:54Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11110",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:873ff160520ac07d5ddd6e5a4e5ec3eb0994d4f64ae1695ed7c4155f8d94326d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:7f5393f5965f",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer][YSQL] It is possible to create a temp table with usually forbidden primary key",
+      "reported_date": "2022-01-20",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11144"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11144",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11144",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer][YSQL] It is possible to create a temp table",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:5473e94ef38206983369f69aa084308c3950183a1b4b3f2790e6b891adf4db70",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:c1c8d8a4f8ed51d81fbcdbf0a3b544995da57e74cae2403db30e886cb7357ec7",
+          "classified_at": "2026-09-13T06:23:54Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11144",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5473e94ef38206983369f69aa084308c3950183a1b4b3f2790e6b891adf4db70"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:d889a2b9835e",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer][YSQL] Create temporary error may be continuously thrown in following statements",
+      "reported_date": "2022-01-21",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11166"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11166",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11166",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer][YSQL] Create temporary error may be continuously thrown in following statements",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:524b896d4872092b93c07866b12aa3b17328afba6d54e744d0868019ce4112d7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11166",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:524b896d4872092b93c07866b12aa3b17328afba6d54e744d0868019ce4112d7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:6d87a81ab2f1",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer][YSQL] \"Infinite loop detected at\" occurred on select statement against colocated database if MONEY PK used",
+      "reported_date": "2022-01-31",
+      "reported_year": 2022,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "tlp",
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11293"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11293",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11293",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer][YSQL] \"Infinite loop detected at\" occurred on select statement against colocated database if MONEY PK used",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d895e4c56777a50c9d70240ee84388652f9fca9f295060599f09b8d567805bf1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11293",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d895e4c56777a50c9d70240ee84388652f9fca9f295060599f09b8d567805bf1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:c3c6bbc13c4f",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer][YSQL] ASC Primary keys update operation may insert new value into a table if complex WHERE statement used",
+      "reported_date": "2022-02-01",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11298"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11298",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11298",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer][YSQL] ASC Primary keys update operation may insert new value",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:f22681be99bdf767142179e3e16432449493c79df69db711b55ae7cc2c15dfa1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:1629582753e4a79419af06f917331a9ea26796a64f16a03959ba0e4fd905e310",
+          "classified_at": "2026-09-13T06:23:54Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11298",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f22681be99bdf767142179e3e16432449493c79df69db711b55ae7cc2c15dfa1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:1fee941cb3b7",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] Attribute value type does not match exception occurs on DELETE statement with WHERE clause with MONEY PK",
+      "reported_date": "2022-02-03",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11346"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11346",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11346",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] Attribute value type does not match exception occurs on DELETE statement with WHERE clause with MONEY PK",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:524e6c219a2a16173dce83ddf975cdeaf1a41fdfef8a86a9d3cfd9eccb34aadb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11346",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:524e6c219a2a16173dce83ddf975cdeaf1a41fdfef8a86a9d3cfd9eccb34aadb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:e2b90a8d1a36",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] Could not open relation error when trying yo ALTER table with secutiry and PK with MONEY datatype",
+      "reported_date": "2022-02-03",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11347"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11347",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11347",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] Could not open relation error when trying yo ALTER table with secutiry and PK with MONEY datatype",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:375d144bf81c8bdf427f672d11b481aa0af19f7c972128d1bc0aa48379683fd1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11347",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:375d144bf81c8bdf427f672d11b481aa0af19f7c972128d1bc0aa48379683fd1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:22d807fc36f9",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] Getting inconsistent results during parallel test run",
+      "reported_date": "2022-02-17",
+      "reported_year": 2022,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11519"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11519",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11519",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] Getting inconsistent results during parallel test run",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2896bb507cf23bf4305bb744b47be0657bc75fb0185929eae066fcd0f242bfd1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11519",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2896bb507cf23bf4305bb744b47be0657bc75fb0185929eae066fcd0f242bfd1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:0d4c5af6a14d",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer][YSQL] Condition on SPLIT AT may cause Segmentation fault in PG",
+      "reported_date": "2022-03-15",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/11757"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/11757",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11757",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer][YSQL] Condition on SPLIT AT may cause Segmentation fault in PG",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:2e4b3c5a84840f7a084e1623830ca5cb501187b84bbfe1ea1a7a0106469090a3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/11757",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:2e4b3c5a84840f7a084e1623830ca5cb501187b84bbfe1ea1a7a0106469090a3"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:31d968469645",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][LST] Colocation - FATAL/ERROR:  Not found: Table <unknown_table_name> not found in Raft group",
+      "reported_date": "2022-05-09",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "def-",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/12421"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/12421",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/12421",
+            "source_type": "github_issue",
+            "excerpt": "There was a potentially related bug with Sqlancer: https://github.com/yugabyte/yugabyte-db/issues/11357",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:76c7136af11121cd06ecc257f55a8b5fe2033959f557b41aa15b55a53cca2a8d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/12421",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:76c7136af11121cd06ecc257f55a8b5fe2033959f557b41aa15b55a53cca2a8d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:8583a9fd9ab3",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] Namespace deletion not allowed error in client logs in parallel test on database creation",
+      "reported_date": "2022-06-28",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/13078"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/13078",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13078",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] Namespace deletion not allowed error in client logs in parallel test on database creation",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:3c1884cafbfe036bbeb7841217730129c2409d4757910e971fa9ea7b879ed7eb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13078",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:3c1884cafbfe036bbeb7841217730129c2409d4757910e971fa9ea7b879ed7eb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:46c715751a25",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][LST] ERROR:  catalog is missing ... attribute(s) for relid ...",
+      "reported_date": "2022-07-12",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "def-",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/13269"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/13269",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13269",
+            "source_type": "github_issue",
+            "excerpt": "This also happened with SQLancer: https://github.com/yugabyte/yugabyte-db/issues/10434",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:524150bb3e57b5498bc402e62d48485ba3c17fe349babc18bd2f0dbd43f9fa24",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13269",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:524150bb3e57b5498bc402e62d48485ba3c17fe349babc18bd2f0dbd43f9fa24"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:454f06ead50b",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] Invalid column number on ANALYZE after partially committed ALTER table transaction",
+      "reported_date": "2022-07-28",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/13466"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/13466",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13466",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] Invalid column number on ANALYZE after partially committed ALTER table transaction",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:4bbf14a6e66eda1a9c525f2a49c0b41cf1331714a076b372f459f3f7e50bb671",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13466",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:4bbf14a6e66eda1a9c525f2a49c0b41cf1331714a076b372f459f3f7e50bb671"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:94d9e9f966a4",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] cache lookup failed for type on ANALYZE table after dropping column",
+      "reported_date": "2022-07-28",
+      "reported_year": 2022,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/13467"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/13467",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13467",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] cache lookup failed for type on ANALYZE table after dropping column",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c7c99b784b3aa357816edb64dd0a3362da71b16d19510e78f41c7565f3a9f7a1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13467",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c7c99b784b3aa357816edb64dd0a3362da71b16d19510e78f41c7565f3a9f7a1"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:d50a2ca87db6",
+      "dbms": "yugabytedb",
+      "title": "[YCQL][SQLancer] FATAL: Check failed: operand.expr_case() == QLExpressionPB::ExprCase::kCondition during YCQL fuzzer workload",
+      "reported_date": "2022-08-28",
+      "reported_year": 2022,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/13787"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/13787",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13787",
+            "source_type": "github_issue",
+            "excerpt": "[YCQL][SQLancer] FATAL: Check failed: operand.expr_case() == QLExpressionPB::ExprCase::kCondition during YCQL fuzzer workload",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:332d0d70da9182333c1e9ab0c8745b2950180e9a4f724a1d8cbfcf2e671a99e7",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13787",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:332d0d70da9182333c1e9ab0c8745b2950180e9a4f724a1d8cbfcf2e671a99e7"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:54658d69ece0",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] FATAL: Failed while visiting tablets in sys catalog: System catalog snapshot is corrupted or built using different build type: Tablegroup 0000555e00003000800000000000584f already exists",
+      "reported_date": "2022-08-30",
+      "reported_year": 2022,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": "norec",
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/13810"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/13810",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13810",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] FATAL: Failed while visiting tablets in sys catalog: System catalog snapshot is corrupted or built using different build type: Tablegroup 0000555e00003000800000000000584f already exists",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:1746b2b2298375b47d9de3f06d9293d3719d0f552132e3fd8e1d91e95d517eae",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/13810",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:1746b2b2298375b47d9de3f06d9293d3719d0f552132e3fd8e1d91e95d517eae"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:2752fa0f89c4",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] ANALYZE will fail if create index statement where failed before for the table",
+      "reported_date": "2023-02-13",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/16080"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/16080",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/16080",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] ANALYZE will fail if create index statement where failed before for the table",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:de94656fc66ef88dd740624043538b7117625a5b474480f887f414bb08ffd029",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/16080",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:de94656fc66ef88dd740624043538b7117625a5b474480f887f414bb08ffd029"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:22de79188ab5",
+      "dbms": "yugabytedb",
+      "title": "Permanent FK table should not be allowed to reference a temp table",
+      "reported_date": "2023-09-27",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/19321"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/19321",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/19321",
+            "source_type": "github_issue",
+            "excerpt": "sqlancer.common.query.SQLQueryAdapter",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:403cc31b362eec6924e40efbd0c8a55b74b09a45f612a9710eb90be56b3e8c38",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-13T03:28:53Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ],
+        "classifier": {
+          "method": "llm_classification",
+          "classifier_version": "bug-attribution-v1",
+          "policy_version": "impact-policy-v1",
+          "taxonomy_version": "taxonomy-v2",
+          "evidence_sha256": "sha256:0de6cb55e846a4d8b68068f36316520cf5091293a8c57ad647541aca5939d1b8",
+          "classified_at": "2026-09-13T06:23:54Z",
+          "model": "claude-opus-5",
+          "rationale": "The report attributes the find to SQLancer. Decided by reading the issue in this session rather than by an API call; the excerpt was checked to be literally present in the text the classifier would have been given."
+        }
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-13T03:28:53Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/19321",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:403cc31b362eec6924e40efbd0c8a55b74b09a45f612a9710eb90be56b3e8c38"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:04ae1dfaf3b6",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] FATAL: the database system is in recovery mode client error may occur in process kill test",
+      "reported_date": "2023-12-05",
+      "reported_year": 2023,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/20169"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/20169",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/20169",
+            "source_type": "github_issue",
+            "excerpt": "Start SQLancer and sample-apps from portal VM e.g.\r\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:cb2f88575afec0a2b45e8e20c27b16f68d8b196b23f20ba9f16f5610206ebe22",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/20169",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:cb2f88575afec0a2b45e8e20c27b16f68d8b196b23f20ba9f16f5610206ebe22"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:1e4733ace533",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] OOM handle related code in PG core dumps during test execution",
+      "reported_date": "2024-03-06",
+      "reported_year": 2024,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/21332"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/21332",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/21332",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] OOM handle related code in PG core dumps during test execution",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:a10a0bd99da2c644b157836bab32cce963dabe606340211d47b520dd4c740832",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/21332",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:a10a0bd99da2c644b157836bab32cce963dabe606340211d47b520dd4c740832"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:d83b4bdf7fcc",
+      "dbms": "yugabytedb",
+      "title": "[DocDB] tserver ResourceArrayEnlarge core dump occurred in SQLancer runs multiple times in master",
+      "reported_date": "2025-02-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/25889"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/25889",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/25889",
+            "source_type": "github_issue",
+            "excerpt": "[DocDB] tserver ResourceArrayEnlarge core dump occurred in SQLancer runs multiple times in master",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:65bafc7bf15b73cd5d993734818c20ba691f3d5069466437154ad21b9acfa8ef",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/25889",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:65bafc7bf15b73cd5d993734818c20ba691f3d5069466437154ad21b9acfa8ef"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:27ad5423d0d8",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer][YSQL] ORDER BY DESC may read deleted values from the table with UNIQUE DEFAULT columns, when fast backward scans is used",
+      "reported_date": "2025-02-15",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/26060"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/26060",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26060",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer][YSQL] ORDER BY DESC may read deleted values from the table with UNIQUE DEFAULT columns, when fast backward scans is used",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:87f9d7e8e3e5b4ca59c60763affa2dbd776037fb99aaaf3323c1c7d2784412dc",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26060",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:87f9d7e8e3e5b4ca59c60763affa2dbd776037fb99aaaf3323c1c7d2784412dc"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:9d276e4fc3ba",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Lost connection with fault RPC response traffic",
+      "reported_date": "2025-03-06",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/26308"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/26308",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26308",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL] Lost connection with fault RPC response traffic",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:fdcc02e0cd417e51026cba3d10904e27d43b2eb47110be3e98961bf80f2e8dec",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26308",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:fdcc02e0cd417e51026cba3d10904e27d43b2eb47110be3e98961bf80f2e8dec"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:526596b46237",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer][YSQL] Inconsistent results for queries with predicate pushdown if enabled yb_enable_expression_pushdown",
+      "reported_date": "2025-03-13",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/26385"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/26385",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26385",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer][YSQL] Inconsistent results for queries with predicate pushdown if enabled yb_enable_expression_pushdown",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b76ddaf626b9d8e00bcbbf7c9987bfe318674573d5a634d365b7a8fe8d1dd228",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26385",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b76ddaf626b9d8e00bcbbf7c9987bfe318674573d5a634d365b7a8fe8d1dd228"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:4d845855d3dd",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected truncated data returned by MAX()",
+      "reported_date": "2025-04-01",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/26620"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/26620",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26620",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 CHAR(100));\nINSERT INTO t1(c0) VALUES('a');\nSELECT MAX(t1.c0) FROM t1;\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:087d1be3ba659d504994a2c2311a84cfbc17ec3f40ee175852849b241dc6c3d3",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26620",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:087d1be3ba659d504994a2c2311a84cfbc17ec3f40ee175852849b241dc6c3d3"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:d98b663e7351",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected Results by yb_enable_optimizer_statistics",
+      "reported_date": "2025-04-06",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/26705"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/26705",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26705",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE TABLE t1(c0 DECIMAL);\nINSERT INTO t0(c0) VALUES(1);\nINSERT INTO t1(c0) VALUES(0.1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:8cd2573f34368069b4f26873262e51f9e77173e587961d76e716de7f452bf879",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26705",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:8cd2573f34368069b4f26873262e51f9e77173e587961d76e716de7f452bf879"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:64711c8776f0",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected result by yb_enable_distinct_pushdown",
+      "reported_date": "2025-04-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/26717"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/26717",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26717",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT);\nCREATE INDEX i0 ON t0(c0, c0 ASC);\nINSERT INTO t0(c0) VALUES(1);\nset yb_enable_base_scans_cost_model=on;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5d7360a52767780fc3b97080b3610079aa7c07257a279601d5cfd20447ae8125",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26717",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5d7360a52767780fc3b97080b3610079aa7c07257a279601d5cfd20447ae8125"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:bb422a7462e4",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected results caused by yb_enable_base_scans_cost_model",
+      "reported_date": "2025-04-07",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/26713"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/26713",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26713",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 DECIMAL);\nINSERT INTO t0(c0) VALUES(0.5);\nCREATE INDEX i0 ON t0 USING LSM(c0, c0 NULLS FIRST);\nset yb_enable_base_scans_cost_model=on;",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:ef138a2fa5696a2006888ecd85c8284d19cecd3720489b61f2f2ca1e924f4734",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26713",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:ef138a2fa5696a2006888ecd85c8284d19cecd3720489b61f2f2ca1e924f4734"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:c08bcfeb47ed",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected error: OBJECT_NOT_FOUND when trying to reference temporary table in permanent table",
+      "reported_date": "2025-04-09",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/26753"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/26753",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26753",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t1(c0 INT, FOREIGN KEY (c0) REFERENCES t0(c0)); --ERROR:  Table with identifier 00004000000030008000000000004029 not found: OBJECT_NOT_FOUND\n```\n\nThe error seems unexpected. I would assume it should return something like: \"Cannot add foreign key on temp tables\"",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:11148a42b27aca7b34cf263edea8c46892ede697498d3b4798656ff0199e21df",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26753",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:11148a42b27aca7b34cf263edea8c46892ede697498d3b4798656ff0199e21df"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:7184f2e61866",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected result by unique constraint",
+      "reported_date": "2025-04-09",
+      "reported_year": 2025,
+      "status": "closed_not_a_bug",
+      "status_is_true_positive": false,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/26752"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/26752",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26752",
+            "source_type": "github_issue",
+            "excerpt": "CREATE TABLE t0(c0 INT, c1 BOOL, UNIQUE(c1));\nCREATE TABLE t1(c0 INT);\nINSERT INTO t0(c1) VALUES(TRUE),(TRUE); -- unique constraint violation but no error\nINSERT INTO t1(c0) VALUES(1);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9e76c6e6455f361cf055ea12d0ce65c3e56f8e669b2faecdf8bb36363215092e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26752",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9e76c6e6455f361cf055ea12d0ce65c3e56f8e669b2faecdf8bb36363215092e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:dc077fc15e4e",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected Error: timed out waiting for postgres backends to catch up",
+      "reported_date": "2025-04-29",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/26983"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/26983",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26983",
+            "source_type": "github_issue",
+            "excerpt": "create table t0 (c0 int4, c1 timestamp);\ncreate index i0 on t0 (c0, c1 desc);\n```\nThe last statement of creating an index returns an unexpected error:",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5b98b7a1e2cec2762b7e18267a753cd94170b8b1a78ad76fdc48c4fbae5aa21b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/26983",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5b98b7a1e2cec2762b7e18267a753cd94170b8b1a78ad76fdc48c4fbae5aa21b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:aef2286e02ac",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected Result of bttextcmp()",
+      "reported_date": "2025-04-30",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/27005"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/27005",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27005",
+            "source_type": "github_issue",
+            "excerpt": "create table t1(c1 text, c2 numeric, c3 text);\ninsert into t1 values('a#', null, 'b&');\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:55b0b84d2e6bcbdd7d1a81de813caf9732e15162b0d8e271d4ceb0c438d4cd5d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27005",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:55b0b84d2e6bcbdd7d1a81de813caf9732e15162b0d8e271d4ceb0c438d4cd5d"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:5049b0b71729",
+      "dbms": "yugabytedb",
+      "title": "[DocDB] ysql_ddl_verification_task FATAL on master in SQLancer test",
+      "reported_date": "2025-05-01",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/27018"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/27018",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27018",
+            "source_type": "github_issue",
+            "excerpt": "[DocDB] ysql_ddl_verification_task FATAL on master in SQLancer test",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:6aef7efc94d9815ccaecc3eb426fe145d46fb25629f412794786973616b24cfb",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27018",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:6aef7efc94d9815ccaecc3eb426fe145d46fb25629f412794786973616b24cfb"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:78dc8c5af2fd",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected Crash on regexp_count()",
+      "reported_date": "2025-05-01",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/27009"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/27009",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27009",
+            "source_type": "github_issue",
+            "excerpt": "create table t0 (c0 text);\ninsert into t0 values ('aaaa');\nselect * from t0 where regexp_count(c0, lpad(c0, 10000, c0)) > 0; -- server closed the connection unexpectedly\n```",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:5b45fc39b706b2d0857eccfd2996738a787f8e70d19428b0708c9a7d06ec4f2e",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27009",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:5b45fc39b706b2d0857eccfd2996738a787f8e70d19428b0708c9a7d06ec4f2e"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:cee3acbd166b",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Crash due to multiple-table joining",
+      "reported_date": "2025-05-02",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/27029"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/27029",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27029",
+            "source_type": "github_issue",
+            "excerpt": "create table t0 (c0 numeric, c1 text, c2 int4);\ncreate table t1 (c3 numeric, c4 int4, c5 timestamp, c6 int4, c7 text);\ncreate table t2 (c11 int4, c14 text, c15 numeric);\ncreate index i0 on t2 (c11, c15, c14);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0c0e48ed1095b216e929b98ceb3a22989cd2848918a75ab997f02be15014101b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27029",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0c0e48ed1095b216e929b98ceb3a22989cd2848918a75ab997f02be15014101b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:5bdd4ecdd427",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected Result in distributed execution",
+      "reported_date": "2025-05-02",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/27031"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/27031",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27031",
+            "source_type": "github_issue",
+            "excerpt": "create table t0 (c0 int4);\ncreate table t1 (c1 int4, c2 int4, c3 int4, c4 int4, c5 int4, c6 int4);\ncreate table t2 (c7 int4);\ninsert into t2 values (14000);",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:d9d82e5503e1b9767b6df3631023c361df69cdbb30510297dc53823b943f6b8b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27031",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:d9d82e5503e1b9767b6df3631023c361df69cdbb30510297dc53823b943f6b8b"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:a053bae84114",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Bitmap scan core dump in SQLancer test",
+      "reported_date": "2025-05-03",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/27039"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/27039",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27039",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL] Bitmap scan core dump in SQLancer test",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c1ef35dc5f1ce3562f219cec580da0c6a0e92e04ed1601e560d23e118c80c8df",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27039",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c1ef35dc5f1ce3562f219cec580da0c6a0e92e04ed1601e560d23e118c80c8df"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:001eb5a84189",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Unexpected error under disabling db_catalog_version_mode",
+      "reported_date": "2025-05-05",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/27059"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/27059",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27059",
+            "source_type": "github_issue",
+            "excerpt": "create table t0 (c0 INT);\ncreate index i0 ON t0 (c0);\nNOTICE:  Retrying wait for backends catalog version: Requested catalog version is too high: req version 2, master version 0\nNOTICE:  Retrying wait for backends catalog version: Requested catalog version is too high: req version 2, master version 0",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:0ded8f1cb7752dac6517a61a0de58213d6dad215e56972c0bdf7c2eef40deee1",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T02:25:35Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T02:25:35Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27059",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:0ded8f1cb7752dac6517a61a0de58213d6dad215e56972c0bdf7c2eef40deee1"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:69bdc7cc5464",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] lost connection to parallel worker error in SQLancer tests with PG Parity flags enabled",
+      "reported_date": "2025-05-22",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/27333"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/27333",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27333",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL] lost connection to parallel worker error in SQLancer tests with PG Parity flags enabled",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b740082d9c974e76a757e90d4e3ccbd22dcbdddca7f544d48124395edddd33ec",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27333",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b740082d9c974e76a757e90d4e3ccbd22dcbdddca7f544d48124395edddd33ec"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:3ef5dae364d8",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Hang in Subquery",
+      "reported_date": "2025-06-07",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "bajinsheng",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/27540"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/27540",
+      "attribution": {
+        "rule": "campaign_reporter",
+        "confidence": "low",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27540",
+            "source_type": "github_issue",
+            "excerpt": "Jira Link: [DB-17101](https://yugabyte.atlassian.net/browse/DB-17101)",
+            "excerpt_is_verbatim": true,
+            "note": "The reporter runs SQLancer campaigns; this is their description of the defect.",
+            "content_sha256": "sha256:24b1adeaf6bd08c11e32b81d53ff4fec72b86029b8503ad96a8bfba2d66c4739",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          },
+          {
+            "source_url": "https://nus-test.github.io/people/",
+            "source_type": "website",
+            "note": "bajinsheng is listed as a member of the TEST lab, which runs SQLancer campaigns against database systems. This record rests on that membership, not on anything the report says.",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-07T12:06:34Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-07T12:06:34Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27540",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:24b1adeaf6bd08c11e32b81d53ff4fec72b86029b8503ad96a8bfba2d66c4739"
+      },
+      "reporter_affiliation": "project"
+    },
+    {
+      "id": "bug:yugabytedb:d5a6d72389c1",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] Postgres fmgr_sql  core dump occurred during evaluate_expr",
+      "reported_date": "2025-07-01",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/27834"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/27834",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27834",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] Postgres fmgr_sql  core dump occurred during evaluate_expr",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7231db5bbd2582e5b7a2a9de5b13459eea7583123622456773c0bbc454b28016",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/27834",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7231db5bbd2582e5b7a2a9de5b13459eea7583123622456773c0bbc454b28016"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:3b322e3829b0",
+      "dbms": "yugabytedb",
+      "title": "[SQLancer][YSQL] ALTER TABLE ... SET DATA TYPE on a UNIQUE column fails in a colocated database",
+      "reported_date": "2025-08-05",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/28177"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/28177",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/28177",
+            "source_type": "github_issue",
+            "excerpt": "[SQLancer][YSQL] ALTER TABLE ... SET DATA TYPE on a UNIQUE column fails in a colocated database",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:694a5b98dbd0b805b34347a6dbce720b0de3fd10fbe29782ea909193a2a8ce17",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/28177",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:694a5b98dbd0b805b34347a6dbce720b0de3fd10fbe29782ea909193a2a8ce17"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:6f2e3eb01ed5",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] TRUNCATE table might hit relation does not exist error while table exists",
+      "reported_date": "2025-08-14",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/28276"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/28276",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/28276",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] TRUNCATE table might hit relation does not exist error while table exists",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:16f628a5ca9b1d378af6f0c88e2a93dc0d7cabb03fcd0c24a9f74b0029bfa630",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/28276",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:16f628a5ca9b1d378af6f0c88e2a93dc0d7cabb03fcd0c24a9f74b0029bfa630"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:4409e19b8cb4",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Tserver spam \"not found in ysql_db_invalidation_messages_map_\" if database was droppped",
+      "reported_date": "2025-08-21",
+      "reported_year": 2025,
+      "status": "fixed",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/28348"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/28348",
+      "attribution": {
+        "rule": "explicit_tool_statement",
+        "confidence": "high",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/28348",
+            "source_type": "github_issue",
+            "excerpt": "Reproduced in SQLancer test",
+            "excerpt_is_verbatim": true,
+            "note": "The report states which tool or oracle found the bug.",
+            "content_sha256": "sha256:c3e4925feb9b4c7a0f8251228441115ec8fb53870e52867d5b7139dbe452dc43",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T07:11:04Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:11:04Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/28348",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c3e4925feb9b4c7a0f8251228441115ec8fb53870e52867d5b7139dbe452dc43"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:bb940804447e",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] Read time should NOT be specified for serializable isolation level during ANALYZE",
+      "reported_date": "2025-11-20",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/29438"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/29438",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/29438",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] Read time should NOT be specified for serializable isolation level during ANALYZE",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:7f9f9ff881339994ee140e4cfeeda033e05eb2d9fcfff914dfe99da42a54b49b",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/29438",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:7f9f9ff881339994ee140e4cfeeda033e05eb2d9fcfff914dfe99da42a54b49b"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:3daf13c7119e",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] Silent FATAL during create table statement execution: Valid ddl metadata is required",
+      "reported_date": "2025-12-01",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/29546"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/29546",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/29546",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] Silent FATAL during create table statement execution: Valid ddl metadata is required",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:187b0123b697c3559a1d3fa3c3bc42c9212c7bdeb61e3890e2aea02827143d3d",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/29546",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:187b0123b697c3559a1d3fa3c3bc42c9212c7bdeb61e3890e2aea02827143d3d"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:9eb0240d84e7",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] Workloads with relaxed concurrent DDL currently failing in master",
+      "reported_date": "2025-12-01",
+      "reported_year": 2025,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/29548"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/29548",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/29548",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] Workloads with relaxed concurrent DDL currently failing in master",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:b147e80a12cf5008e2cd473567f1a567edfa948a2286b8a2c226044edd3ad4a9",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/29548",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:b147e80a12cf5008e2cd473567f1a567edfa948a2286b8a2c226044edd3ad4a9"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:e20a6e055af2",
+      "dbms": "yugabytedb",
+      "title": "[YSQL][SQLancer] SIGSEGV in YBCTriggerRelcacheInitConnection due to NULL dbname",
+      "reported_date": "2026-01-22",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "qvad",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/30074"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/30074",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/30074",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL][SQLancer] SIGSEGV in YBCTriggerRelcacheInitConnection due to NULL dbname",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:c46bde5d434de5585a0ba9e02fc7a7d6febcfbe99cd89ebbc902886bb9e17388",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/30074",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:c46bde5d434de5585a0ba9e02fc7a7d6febcfbe99cd89ebbc902886bb9e17388"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:c0ef606ceb01",
+      "dbms": "yugabytedb",
+      "title": "[YSQL] Core dumps noticed causing multiple connection manager and sqlancer tests to fail",
+      "reported_date": "2026-05-07",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "rojasbinny-yb",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/31489"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/31489",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/31489",
+            "source_type": "github_issue",
+            "excerpt": "[YSQL] Core dumps noticed causing multiple connection manager and sqlancer tests to fail",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:9185ea20b958240e71ad9303dc41d38b2b289b477dc9accb200c3a97f35a27d4",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:49:05Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:49:05Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/31489",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:9185ea20b958240e71ad9303dc41d38b2b289b477dc9accb200c3a97f35a27d4"
+      },
+      "reporter_affiliation": "external"
+    },
+    {
+      "id": "bug:yugabytedb:216c104fde8d",
+      "dbms": "yugabytedb",
+      "title": "[DDL Atomicity] Master FATAL / crash-loop in ysql_ddl_verification_task PgSchemaChecker — table-rewrite ALTER leaves old table with alter-op but no drop-op (no PITR, no txn-DDL)",
+      "reported_date": "2026-07-17",
+      "reported_year": 2026,
+      "status": "open",
+      "status_is_true_positive": true,
+      "finder": "sqlancer",
+      "technique": null,
+      "symptom": "unknown",
+      "reporter": "yugabyte-ci",
+      "links": {
+        "report": "https://github.com/yugabyte/yugabyte-db/issues/32699"
+      },
+      "primary_url": "https://github.com/yugabyte/yugabyte-db/issues/32699",
+      "attribution": {
+        "rule": "campaign_evidence",
+        "confidence": "medium",
+        "evidence": [
+          {
+            "source_url": "https://github.com/yugabyte/yugabyte-db/issues/32699",
+            "source_type": "github_issue",
+            "excerpt": "- #27018 — same assertion seen in a SQLancer test",
+            "excerpt_is_verbatim": true,
+            "note": "The project marks this report as coming from a SQLancer run.",
+            "content_sha256": "sha256:f7d4196aa7bd7d452e93dfe28144a5117556b1b15536bfb90361826bc4bdcb29",
+            "retrieved_at": "2026-09-13T06:23:12Z",
+            "first_seen": "2026-09-06T16:19:32Z",
+            "last_verified": "2026-09-13T06:23:12Z"
+          }
+        ]
+      },
+      "provenance": {
+        "collector": "github_bugs",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T16:19:32Z",
+        "last_verified": "2026-09-13T06:23:12Z",
+        "source_url": "https://github.com/yugabyte/yugabyte-db/issues/32699",
+        "source_type": "github_issue",
+        "content_sha256": "sha256:f7d4196aa7bd7d452e93dfe28144a5117556b1b15536bfb90361826bc4bdcb29"
+      },
+      "reporter_affiliation": "external"
+    }
+  ]
+}
diff --git a/_data/impact/dbms.json b/_data/impact/dbms.json
new file mode 100644
index 0000000..39b5a8d
--- /dev/null
+++ b/_data/impact/dbms.json
@@ -0,0 +1,1131 @@
+{
+  "schema_version": "1.0.0",
+  "dbms": [
+    {
+      "id": "agensgraph",
+      "name": "AgensGraph",
+      "aliases": [
+        "agensgraph",
+        "AgensGraph"
+      ],
+      "url": "https://bitnine.net/",
+      "repository": "https://github.com/bitnine-oss/agensgraph",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "arangodb",
+      "name": "ArangoDB",
+      "aliases": [
+        "ArangoDB"
+      ],
+      "url": "https://arangodb.com/",
+      "repository": "https://github.com/arangodb/arangodb",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "bharatdbms",
+      "name": "BharatDBMS",
+      "aliases": [
+        "BharatDBMS"
+      ],
+      "url": "https://github.com/BharatDBPG/BharatDBMS-PG",
+      "repository": "https://github.com/BharatDBPG/BharatDBMS-PG",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "citus",
+      "name": "Citus",
+      "aliases": [
+        "Citus"
+      ],
+      "url": "https://www.citusdata.com/",
+      "repository": "https://github.com/citusdata/citus",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/citus",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/citus",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/citus.",
+            "content_sha256": "sha256:bfe241f60456759c6773b9db1dbfca0edfba29b7178b1d52ed3b3159bb2bc8aa",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "clickhouse",
+      "name": "ClickHouse",
+      "aliases": [
+        "ClickHouse"
+      ],
+      "url": "https://clickhouse.com/",
+      "repository": "https://github.com/ClickHouse/ClickHouse",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/clickhouse",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/clickhouse",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/clickhouse.",
+            "content_sha256": "sha256:df18e04294c9b0f2cf2b937d7e873402e75e477e45b06cb336825def779a90dd",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "cloudberry",
+      "name": "Apache Cloudberry",
+      "aliases": [
+        "Cloudberry",
+        "Apache Cloudberry",
+        "cloudberrydb"
+      ],
+      "url": "https://cloudberry.apache.org/",
+      "repository": "https://github.com/apache/cloudberry",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "cnosdb",
+      "name": "CnosDB",
+      "aliases": [
+        "CnosDB"
+      ],
+      "url": "https://www.cnosdb.com/",
+      "repository": "https://github.com/cnosdb/cnosdb",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "cockroachdb",
+      "name": "CockroachDB",
+      "aliases": [
+        "CockroachDB"
+      ],
+      "url": "https://www.cockroachlabs.com/",
+      "repository": "https://github.com/cockroachdb/cockroach",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/cockroachdb",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/cockroachdb",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/cockroachdb.",
+            "content_sha256": "sha256:55d565cea4a9dbefcc3bfa74b500c7bcd2784eb7de564a9c30060166a6d261e0",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "cratedb",
+      "name": "CrateDB",
+      "aliases": [
+        "cratedb",
+        "Crate",
+        "CrateDB"
+      ],
+      "url": "https://cratedb.com/",
+      "repository": "https://github.com/crate/crate",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "cubrid",
+      "name": "CUBRID",
+      "aliases": [
+        "CUBRID"
+      ],
+      "url": "https://www.cubrid.org/",
+      "repository": "https://github.com/CUBRID/cubrid",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "databend",
+      "name": "Databend",
+      "aliases": [
+        "Databend"
+      ],
+      "url": "https://www.databend.com/",
+      "repository": "https://github.com/datafuselabs/databend",
+      "github_owners": [
+        "datafuse-extras",
+        "databendlabs",
+        "datafuselabs"
+      ],
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/databend",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/databend",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/databend.",
+            "content_sha256": "sha256:0ceb6460a56eb6fa40a471139b541365f1cc720dc117c0adb2da29ad6eaf28fe",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "datafusion",
+      "name": "Apache DataFusion",
+      "aliases": [
+        "Apache DataFusion",
+        "datafusion",
+        "ApacheDataFusion"
+      ],
+      "url": "https://datafusion.apache.org/",
+      "repository": "https://github.com/apache/datafusion",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/datafusion",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/datafusion",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/datafusion.",
+            "content_sha256": "sha256:75299eeda624b097cd68ef858b1a0d3ebc6c1b892c2e64f093915376066fe18c",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "defradb",
+      "name": "DefraDB",
+      "aliases": [
+        "DefraDB"
+      ],
+      "url": "https://docs.source.network/",
+      "repository": "https://github.com/sourcenetwork/defradb.rs",
+      "github_owners": [
+        "sourcenetwork"
+      ],
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "dolt",
+      "name": "Dolt",
+      "aliases": [
+        "Dolt"
+      ],
+      "url": "https://www.dolthub.com/",
+      "repository": "https://github.com/dolthub/dolt",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "doris",
+      "name": "Apache Doris",
+      "aliases": [
+        "Apache Doris",
+        "doris",
+        "ApacheDoris"
+      ],
+      "url": "https://doris.apache.org/",
+      "repository": "https://github.com/apache/doris",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/doris",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/doris",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/doris.",
+            "content_sha256": "sha256:1ded854c34a0ae4271c422a10a43eef0e7195cf5d8930fe9158a23d1063d98e7",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "duckdb",
+      "name": "DuckDB",
+      "aliases": [
+        "DuckDB"
+      ],
+      "url": "https://duckdb.org/",
+      "repository": "https://github.com/duckdb/duckdb",
+      "github_owners": [
+        "cwida"
+      ],
+      "github_repositories": [
+        "duckdb/duckdb-fuzzer",
+        "duckdblabs/duckdb-fuzzer-ci"
+      ],
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/duckdb",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/duckdb",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/duckdb.",
+            "content_sha256": "sha256:de1a6926ed78991e4b4c4cdacdd67a9023df81e7e8d8287bfd861fab2a878b45",
+            "retrieved_at": "2026-09-13T06:37:03Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-13T06:37:03Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "elasticsearch",
+      "name": "Elasticsearch",
+      "aliases": [
+        "Elasticsearch"
+      ],
+      "url": "https://www.elastic.co/",
+      "repository": "https://github.com/elastic/elasticsearch",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "falkordb",
+      "name": "FalkorDB",
+      "aliases": [
+        "FalkorDB"
+      ],
+      "url": "https://www.falkordb.com/",
+      "repository": "https://github.com/FalkorDB/FalkorDB",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "feldera",
+      "name": "Feldera",
+      "aliases": [
+        "Feldera"
+      ],
+      "url": "https://feldera.com/",
+      "repository": "https://github.com/feldera/feldera",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "firebird",
+      "name": "Firebird",
+      "aliases": [
+        "Firebird"
+      ],
+      "url": "https://firebirdsql.org/",
+      "repository": "https://github.com/FirebirdSQL/firebird",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "greenplum",
+      "name": "Greenplum",
+      "aliases": [
+        "Greenplum",
+        "gpdb"
+      ],
+      "url": "https://greenplum.org/",
+      "repository": "https://github.com/greenplum-db/gpdb",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "greptimedb",
+      "name": "GreptimeDB",
+      "aliases": [
+        "GreptimeDB"
+      ],
+      "url": "https://greptime.com/",
+      "repository": "https://github.com/GreptimeTeam/greptimedb",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "h2",
+      "name": "H2",
+      "aliases": [
+        "H2"
+      ],
+      "url": "https://h2database.com/",
+      "repository": "https://github.com/h2database/h2database",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/h2",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/h2",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/h2.",
+            "content_sha256": "sha256:fad26aa037e944625b76fc3b50aa7087a02ee58101cb6d143ed70449790d8c51",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "hazelcast",
+      "name": "Hazelcast",
+      "aliases": [
+        "Hazelcast"
+      ],
+      "url": "https://hazelcast.com/",
+      "repository": "https://github.com/hazelcast/hazelcast",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "hive",
+      "name": "Apache Hive",
+      "aliases": [
+        "Apache Hive",
+        "hive",
+        "ApacheHive"
+      ],
+      "url": "https://hive.apache.org/",
+      "repository": "https://github.com/apache/hive",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/hive",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/hive",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/hive.",
+            "content_sha256": "sha256:048aaf8977f3b60007fcd766cb495362d778f155a95645e1dccae3f8f1c9ab33",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "hsqldb",
+      "name": "HSQLDB",
+      "aliases": [
+        "HSQLDB"
+      ],
+      "url": "https://hsqldb.org/",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/hsqldb",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/hsqldb",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/hsqldb.",
+            "content_sha256": "sha256:11a3e61079058fe78c9669fdf18587a974027b822c30b9d48df111a610335791",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "kyzo",
+      "name": "Kyzo",
+      "aliases": [
+        "Kyzo"
+      ],
+      "url": "https://github.com/kyzobuild/kyzo",
+      "repository": "https://github.com/kyzobuild/kyzo",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "mariadb",
+      "name": "MariaDB",
+      "aliases": [
+        "MariaDB"
+      ],
+      "url": "https://mariadb.org/",
+      "repository": "https://github.com/MariaDB/server",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/mariadb",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/mariadb",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/mariadb.",
+            "content_sha256": "sha256:454ddb0bbc398299ab910ee132c1efab4ac5bc22ced0a8002dc4733ee3d9e9d3",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "materialize",
+      "name": "Materialize",
+      "aliases": [
+        "Materialize"
+      ],
+      "url": "https://materialize.com/",
+      "repository": "https://github.com/MaterializeInc/materialize",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/materialize",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/materialize",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/materialize.",
+            "content_sha256": "sha256:40248fb99bfe091bb1dc9f35ce567551f72a1655287c4167298d1c3eb921206d",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "matrixone",
+      "name": "MatrixOne",
+      "aliases": [
+        "matrixone",
+        "MatrixOne",
+        "MO"
+      ],
+      "url": "https://www.matrixorigin.io/",
+      "repository": "https://github.com/matrixorigin/matrixone",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Registered because its own tracker carries SQLancer bug reports, not because the main SQLancer repository ships a provider for it. Without a registry entry the collector never searches the repository, and the global catch-all query cannot reach it: 'SQLancer' is far past GitHub's 1000-result search cap."
+    },
+    {
+      "id": "monetdb",
+      "name": "MonetDB",
+      "aliases": [
+        "MonetDB"
+      ],
+      "url": "https://www.monetdb.org/",
+      "repository": "https://github.com/MonetDB/MonetDB",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "mongodb",
+      "name": "MongoDB",
+      "aliases": [
+        "MongoDB"
+      ],
+      "url": "https://www.mongodb.com/",
+      "repository": "https://github.com/mongodb/mongo",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "mysql",
+      "name": "MySQL",
+      "aliases": [
+        "MySQL"
+      ],
+      "url": "https://www.mysql.com/",
+      "repository": "https://github.com/mysql/mysql-server",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/mysql",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/mysql",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/mysql.",
+            "content_sha256": "sha256:7d4456d69dfda1520c455201bfb9538d1fddebaf50019ae10a27704c60f19e43",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "neo4j",
+      "name": "Neo4j",
+      "aliases": [
+        "neo4j",
+        "Neo4j"
+      ],
+      "url": "https://neo4j.com/",
+      "repository": "https://github.com/neo4j/neo4j",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "noisepage",
+      "name": "NoisePage",
+      "aliases": [
+        "NoisePage"
+      ],
+      "url": "https://noise.page/",
+      "repository": "https://github.com/cmu-db/noisepage",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "oceanbase",
+      "name": "OceanBase",
+      "aliases": [
+        "OceanBase"
+      ],
+      "url": "https://www.oceanbase.com/",
+      "repository": "https://github.com/oceanbase/oceanbase",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/oceanbase",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/oceanbase",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/oceanbase.",
+            "content_sha256": "sha256:7cd4be3a75c004b63e19761357e67c11735d643df54e6595eb5484870ab78bd0",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "opengauss",
+      "name": "openGauss",
+      "aliases": [
+        "openGauss",
+        "opengauss"
+      ],
+      "url": "https://opengauss.org/",
+      "repository": "https://github.com/opengauss-mirror/openGauss-server",
+      "gitee_repositories": [
+        "opengauss/openGauss-server"
+      ],
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "opensearch",
+      "name": "OpenSearch SQL",
+      "aliases": [
+        "OpenSearch",
+        "OpenSearch SQL"
+      ],
+      "url": "https://opensearch.org/",
+      "repository": "https://github.com/opensearch-project/sql",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "oxla",
+      "name": "Oxla",
+      "aliases": [
+        "oxla",
+        "Oxla"
+      ],
+      "url": "https://oxla.com/",
+      "github_owners": [
+        "redpanda-data"
+      ],
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Registered because Redpanda Data maintains a SQLancer fork carrying a complete Oxla provider that upstream does not have. supported_by_sqlancer is false precisely because the provider lives only downstream."
+    },
+    {
+      "id": "percona",
+      "name": "Percona Server",
+      "aliases": [
+        "Percona"
+      ],
+      "url": "https://www.percona.com/",
+      "repository": "https://github.com/percona/percona-server",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "postgresql",
+      "name": "PostgreSQL",
+      "aliases": [
+        "PostgreSQL",
+        "postgres"
+      ],
+      "url": "https://www.postgresql.org/",
+      "repository": "https://github.com/postgres/postgres",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/postgres",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/postgres",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/postgres.",
+            "content_sha256": "sha256:385c1c056b3a82c9fbbd4447419373f4833b3414ca900a089f79a00d828466ca",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "presto",
+      "name": "Presto",
+      "aliases": [
+        "Presto"
+      ],
+      "url": "https://prestodb.io/",
+      "repository": "https://github.com/prestodb/presto",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/presto",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/presto",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/presto.",
+            "content_sha256": "sha256:b5501e5274df2bf99df04adac8d41472f8e2b8d6eb361366ca677502791590f5",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "questdb",
+      "name": "QuestDB",
+      "aliases": [
+        "QuestDB"
+      ],
+      "url": "https://questdb.io/",
+      "repository": "https://github.com/questdb/questdb",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/questdb",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/questdb",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/questdb.",
+            "content_sha256": "sha256:3a0c978b589780fa6e83279c0f5f0cdf6da187739bb3d345b8552ac3bfcf3a47",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "readyset",
+      "name": "ReadySet",
+      "aliases": [
+        "ReadySet",
+        "readyset"
+      ],
+      "url": "https://readyset.io/",
+      "repository": "https://github.com/readysettech/readyset",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "redisgraph",
+      "name": "RedisGraph",
+      "aliases": [
+        "redisgraph",
+        "RedisGraph"
+      ],
+      "url": "https://redis.io/",
+      "repository": "https://github.com/RedisGraph/RedisGraph",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "risingwave",
+      "name": "RisingWave",
+      "aliases": [
+        "risingwave",
+        "RisingWave"
+      ],
+      "url": "https://risingwave.com/",
+      "repository": "https://github.com/risingwavelabs/risingwave",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "seekdb",
+      "name": "SeekDB",
+      "aliases": [
+        "SeekDB"
+      ],
+      "url": "https://www.oceanbase.com/",
+      "repository": "https://github.com/oceanbase/seekdb",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "serenedb",
+      "name": "SereneDB",
+      "aliases": [
+        "SereneDB"
+      ],
+      "url": "https://github.com/serenedb/serenedb",
+      "repository": "https://github.com/serenedb/serenedb",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "spark",
+      "name": "Apache Spark",
+      "aliases": [
+        "Apache Spark",
+        "spark",
+        "ApacheSpark"
+      ],
+      "url": "https://spark.apache.org/",
+      "repository": "https://github.com/apache/spark",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/spark",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/spark",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/spark.",
+            "content_sha256": "sha256:ba18bb81cb28cfe4a69b2631e0325659b302a8f0f2d6c3e084a3a24678b4cfdb",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "sparq",
+      "name": "sparq",
+      "aliases": [
+        "sparq"
+      ],
+      "url": "https://github.com/sparq-org/sparq",
+      "repository": "https://github.com/sparq-org/sparq",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "spiceai",
+      "name": "Spice.ai OSS",
+      "aliases": [
+        "Spice.ai",
+        "spiceai"
+      ],
+      "url": "https://spice.ai/",
+      "repository": "https://github.com/spiceai/spiceai",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "sqlite",
+      "name": "SQLite",
+      "aliases": [
+        "SQLite",
+        "sqlite3"
+      ],
+      "url": "https://sqlite.org/",
+      "repository": "https://github.com/sqlite/sqlite",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/sqlite3",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/sqlite3",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/sqlite3.",
+            "content_sha256": "sha256:b5cb537719daa5359818b290ec49e3ac316b31715828e1ef3c5b130836313011",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "starrocks",
+      "name": "StarRocks",
+      "aliases": [
+        "StarRocks"
+      ],
+      "url": "https://www.starrocks.io/",
+      "repository": "https://github.com/StarRocks/starrocks",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "stonedb",
+      "name": "StoneDB",
+      "aliases": [
+        "StoneDB"
+      ],
+      "url": "https://stonedb.io/",
+      "repository": "https://github.com/stoneatom/stonedb",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "tarantool",
+      "name": "Tarantool",
+      "aliases": [
+        "Tarantool"
+      ],
+      "url": "https://www.tarantool.io/",
+      "repository": "https://github.com/tarantool/tarantool",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "tdengine",
+      "name": "TDengine",
+      "aliases": [
+        "TDEngine",
+        "tdengine",
+        "TDengine"
+      ],
+      "url": "https://tdengine.com/",
+      "repository": "https://github.com/taosdata/TDengine",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "tidb",
+      "name": "TiDB",
+      "aliases": [
+        "TiDB"
+      ],
+      "url": "https://www.pingcap.com/tidb/",
+      "repository": "https://github.com/pingcap/tidb",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/tidb",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/tidb",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/tidb.",
+            "content_sha256": "sha256:9964ff1fb4109f35e08881c227d16b498a91520cc53a951a86eb5c4c6dcafa73",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    },
+    {
+      "id": "tikv",
+      "name": "TiKV",
+      "aliases": [
+        "TiKV"
+      ],
+      "url": "https://tikv.org/",
+      "repository": "https://github.com/tikv/tikv",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "turso",
+      "name": "Turso",
+      "aliases": [
+        "Turso",
+        "limbo"
+      ],
+      "url": "https://turso.tech/",
+      "repository": "https://github.com/tursodatabase/turso",
+      "github_owners": [
+        "tursodatabase"
+      ],
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "umbra",
+      "name": "Umbra",
+      "aliases": [
+        "Umbra"
+      ],
+      "url": "https://umbra-db.com/",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "virtuoso",
+      "name": "Virtuoso",
+      "aliases": [
+        "Virtuoso"
+      ],
+      "url": "https://virtuoso.openlinksw.com/",
+      "repository": "https://github.com/openlink/virtuoso-opensource",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "wadjet",
+      "name": "Wadjet",
+      "aliases": [
+        "Wadjet"
+      ],
+      "url": "https://github.com/derekmwright/wadjet",
+      "repository": "https://github.com/derekmwright/wadjet",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "xugu",
+      "name": "XuGu",
+      "aliases": [
+        "XuGu",
+        "Xugu"
+      ],
+      "url": "https://www.xugudb.com/",
+      "repository": "https://github.com/Xugu-Open-Source/xugu",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "ydb",
+      "name": "YDB",
+      "aliases": [
+        "YDB"
+      ],
+      "url": "https://ydb.tech/",
+      "repository": "https://github.com/ydb-platform/ydb",
+      "supported_by_sqlancer": false,
+      "support": null,
+      "notes": "Referenced by impact records without a provider package in the current main SQLancer repository."
+    },
+    {
+      "id": "yugabytedb",
+      "name": "YugabyteDB",
+      "aliases": [
+        "YugabyteDB",
+        "yugabyte"
+      ],
+      "url": "https://www.yugabyte.com/",
+      "repository": "https://github.com/yugabyte/yugabyte-db",
+      "supported_by_sqlancer": true,
+      "support": {
+        "provider_path": "src/sqlancer/yugabyte",
+        "umbrella_tool": "sqlancer",
+        "evidence": [
+          {
+            "source_url": "https://github.com/sqlancer/sqlancer/tree/main/src/sqlancer/yugabyte",
+            "source_type": "github_repository",
+            "excerpt": null,
+            "note": "The SQLancer repository contains a testing implementation at src/sqlancer/yugabyte.",
+            "content_sha256": "sha256:e6410ab1506a8ce89d0e41be80a67c45a61e5e3b98ccb0e124e1833dda2f85a4",
+            "retrieved_at": "2026-09-06T16:54:11Z",
+            "first_seen": "2026-09-06T16:54:11Z",
+            "last_verified": "2026-09-06T16:54:11Z"
+          }
+        ]
+      }
+    }
+  ]
+}
diff --git a/_data/impact/needs_review.json b/_data/impact/needs_review.json
new file mode 100644
index 0000000..31d9240
--- /dev/null
+++ b/_data/impact/needs_review.json
@@ -0,0 +1,4753 @@
+{
+  "schema_version": "1.0.0",
+  "description": "Candidates a collector could not place, kept between runs so they can be worked through. Each entry names what was seen and why it could not be admitted. An entry leaves when the record it points at enters the dataset, or when a person dismisses it with a reason.",
+  "open": [
+    {
+      "id": "review:eff32f7ae3cc",
+      "kind": "adoption",
+      "source": "adoption",
+      "url": null,
+      "title": null,
+      "reason": "repository search failed",
+      "first_seen": "2026-09-13T05:57:42Z",
+      "last_seen": "2026-09-13T07:17:40Z"
+    },
+    {
+      "id": "review:09190abce971",
+      "kind": "bug",
+      "source": "bugs",
+      "url": "https://sqlite.org/forum/forumpost/3776b48e71, https://www.sqlite.org/src/info/01868ebcd25fadb2",
+      "title": null,
+      "reason": "not inspected yet; run budget reached",
+      "first_seen": "2026-09-12T17:02:58Z",
+      "last_seen": "2026-09-13T03:19:40Z"
+    },
+    {
+      "id": "review:1fa04ff1a7c4",
+      "kind": "bug",
+      "source": "bugs",
+      "url": "https://github.com/cockroachdb/cockroach/issues/174972",
+      "title": "roachtest: tlp failed",
+      "reason": "needs semantic classification; classifier unavailable",
+      "first_seen": "2026-09-12T17:02:58Z",
+      "last_seen": "2026-09-12T17:02:58Z"
+    },
+    {
+      "id": "review:7382795bf7ed",
+      "kind": "bug",
+      "source": "bugs",
+      "url": "https://bugs.mysql.com/bug.php?id=113180",
+      "title": null,
+      "reason": "not inspected yet; run budget reached",
+      "first_seen": "2026-09-12T17:02:58Z",
+      "last_seen": "2026-09-13T03:19:40Z"
+    },
+    {
+      "id": "review:99251563509c",
+      "kind": "bug",
+      "source": "bugs",
+      "url": "http://bugs.mysql.com/113298",
+      "title": null,
+      "reason": "not inspected yet; run budget reached",
+      "first_seen": "2026-09-12T17:02:58Z",
+      "last_seen": "2026-09-13T03:19:40Z"
+    },
+    {
+      "id": "review:0d720605e5ce",
+      "kind": "paper",
+      "source": "papers",
+      "url": "https://github.com/joyemang33/Argus",
+      "title": "Automated Discovery of Test Oracles for Database Management Systems Using LLMs",
+      "reason": "artifact shows weak SQLancer markers (sqlancer_source_in_nested_artifact); needs a human decision",
+      "first_seen": "2026-09-10T14:40:47Z",
+      "last_seen": "2026-09-10T15:15:32Z"
+    },
+    {
+      "id": "review:185db999ebe4",
+      "kind": "paper",
+      "source": "papers",
+      "url": null,
+      "title": "Automated Performance Bug Detection in Database Systems",
+      "reason": "needs semantic classification; classifier unavailable",
+      "first_seen": "2026-09-10T14:40:47Z",
+      "last_seen": "2026-09-10T15:15:32Z"
+    },
+    {
+      "id": "review:ae13741f6de5",
+      "kind": "paper",
+      "source": "papers",
+      "url": null,
+      "title": "A Controlled Evaluation of Relational Database Models for Query Performance across DAS, RAID-0, and RAID-1 Storage Configurations",
+      "reason": "no publication year from the scholarly index",
+      "first_seen": "2026-09-10T14:40:47Z",
+      "last_seen": "2026-09-10T15:15:32Z"
+    },
+    {
+      "id": "review:cb2139f133ca",
+      "kind": "paper",
+      "source": "papers",
+      "url": null,
+      "title": "This paper is included in the Proceedings of the 33rd USENIX Security Symposium.",
+      "reason": "index returned proceedings boilerplate instead of a title",
+      "first_seen": "2026-09-10T14:40:47Z",
+      "last_seen": "2026-09-10T15:15:32Z"
+    }
+  ],
+  "dismissed": [
+    {
+      "id": "review:003f03c523fa",
+      "url": "https://github.com/iamfork/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:0043cd0503db",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/13234",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:00d03ba70acc",
+      "url": "https://github.com/sqlancer/sqlancer/issues/570",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:00db3f712cc1",
+      "url": "https://github.com/gg-big-org/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:01f0a7fdf311",
+      "url": "https://github.com/duckdblabs/duckdb-fuzzer-ci/issues/7",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:022ab49474b1",
+      "url": "https://github.com/cockroachdb/cockroach/issues/90583",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:02397fe9822c",
+      "url": "https://github.com/spiceai/spiceai/issues/2190",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:02ef983c05e3",
+      "url": "https://github.com/cockroachdb/cockroach/issues/47405",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:030a56426ce7",
+      "url": "https://github.com/derekmwright/wadjet/issues/626",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:03161e767de7",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/3",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:0353f142c0ea",
+      "url": "https://github.com/serenedb/serenedb/issues/892",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:0370faa46fae",
+      "url": "https://github.com/cockroachdb/cockroach/issues/134462",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:03ec60f1affb",
+      "url": "https://github.com/cockroachdb/cockroach/issues/43733",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:0495d8f7af1b",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-24/issues/46",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:05232c843123",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/9",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:060fb8ca58e4",
+      "url": "https://github.com/cockroachdb/cockroach/issues/4917",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:06901c573c3d",
+      "url": "https://github.com/cockroachdb/cockroach/issues/170332",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:06f67ceb4bbf",
+      "url": "https://github.com/sqlancer/sqlancer/issues/533",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:06fdda1f0d66",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/87",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:084efa8b0ff6",
+      "url": "https://github.com/sqlancer/sqlancer/issues/105",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:0937dc2b020b",
+      "url": "https://github.com/cockroachdb/cockroach/issues/124629",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:0ac323fabd52",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/7",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:0b05ecdada07",
+      "url": "https://github.com/sqlancer/sqlancer/issues/732",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:0c1229b02897",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/14",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:0c5e3cbcea8f",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/6",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:0c864199d074",
+      "url": "https://github.com/cockroachdb/cockroach/issues/13805",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:0d2cf865229a",
+      "url": "https://github.com/cockroachdb/cockroach/issues/9513",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:0d7a04c01002",
+      "url": "https://github.com/elastic/elasticsearch/issues/104264",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:0e7c84396188",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-24/issues/48",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:0e9b2791b09b",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/64",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:0f0b2e5ea83a",
+      "url": "https://github.com/pingcap/tidb/issues/34209",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:0f34c3346a09",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-14/issues/39",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:0f883899341d",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-19/issues/28",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:102394041118",
+      "url": "https://github.com/spiceai/spiceai/blob/64fc04b9466ce7010a57c067fe7df47b015d62c6/crates/cayenne/tests/correctness/support/sqllancer.rs",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1059806eeffe",
+      "url": "https://github.com/neo4j/neo4j/issues/13276",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1084d1a84e3a",
+      "url": "https://github.com/kyzobuild/kyzo/issues/376",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:10b72e2aaf8e",
+      "url": "https://github.com/elastic/elasticsearch/issues/150128",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:11880b4bfa41",
+      "url": "https://github.com/cockroachdb/cockroach/issues/25177",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:11cafdf743b7",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/37",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:123eb9ae1d50",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-5/issues/21",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:124016cd5367",
+      "url": "https://github.com/cockroachdb/cockroach/issues/6262",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:12ab5fc6557b",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/24",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:12f7c904d11a",
+      "url": "https://github.com/assert-lab/CoCoMUT/issues/30",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1326b80ab976",
+      "url": "https://github.com/sparq-org/sparq/issues/5764",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:138ab15a64f5",
+      "url": "https://github.com/cockroachdb/cockroach/issues/26180",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:13b89ac96af3",
+      "url": "https://github.com/cockroachdb/cockroach/issues/84484",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:13d53b5457d0",
+      "url": "https://github.com/cockroachdb/cockroach/issues/168835",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:143e9040d14c",
+      "url": "https://github.com/satanson/starrocks/issues/8",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:151456b1a2cc",
+      "url": "https://github.com/duckdb/duckdb/issues/498",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:158694939ea2",
+      "url": "https://github.com/elastic/elasticsearch/issues/89017",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:160946f38539",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1221",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:162e85587847",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-19/issues/5",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:170467a13123",
+      "url": "https://github.com/cockroachdb/cockroach/issues/170925",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:17427ffd090f",
+      "url": "https://github.com/cockroachdb/cockroach/issues/13394",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:177f1635de5b",
+      "url": "https://github.com/cockroachdb/cockroach/issues/131978",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1853ca7b6db4",
+      "url": "https://github.com/cockroachdb/cockroach/issues/54155",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:18708ca957cf",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/23517",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:18b72de8a82b",
+      "url": "https://github.com/sqlancer/sqlancer/issues/503",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:18d003984105",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1048",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:191b3e433f85",
+      "url": "https://github.com/cockroachdb/cockroach/issues/51601",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1a25be8ffc79",
+      "url": "https://github.com/cockroachdb/cockroach/issues/95263",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1a603ee6106c",
+      "url": "https://github.com/cockroachdb/cockroach/issues/48116",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1a97b6621d34",
+      "url": "https://github.com/cockroachdb/cockroach/issues/31338",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1aee365a03a5",
+      "url": "https://github.com/jOOQ/jOOQ/issues/12840",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1b0acafedd3a",
+      "url": "https://github.com/cockroachdb/cockroach/issues/104988",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1b3bdc7142e9",
+      "url": "https://github.com/questdb/questdb/issues/433",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1bfd8c0dc908",
+      "url": "https://github.com/adamziel/sqlite-database-integration/issues/2",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1c134e3835fd",
+      "url": "https://github.com/cockroachdb/cockroach/issues/79610",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1c2d6931f01e",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-11/issues/94",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:1c82f73942c7",
+      "url": "https://github.com/sqlancer/sqlancer/issues/374",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:1d926f5e43a4",
+      "url": "https://github.com/cockroachdb/cockroach/issues/4190",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:1efc04e3e148",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/85",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:2050c76ebcf1",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/19",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:20d5c234176f",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/29",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:213f0ca25307",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/1",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:218c6bbbefa2",
+      "url": "https://github.com/cockroachdb/cockroach/issues/125099",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2276b05f6811",
+      "url": "https://github.com/CS3213-T6-1/sqlancer/issues/87",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:229fb16c49a1",
+      "url": "https://github.com/elastic/elasticsearch/issues/35311",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:22d581521d5b",
+      "url": "https://github.com/databendlabs/databend/issues/1768",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:23339767b8fa",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/6155",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:233aa0f7cb86",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/6",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:234c9e00d26f",
+      "url": "https://github.com/elastic/elasticsearch/issues/135252",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:23c3688c3d0b",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/75",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:23e543af8831",
+      "url": "https://github.com/pingcap/tidb/issues/34206",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:242a71f197d3",
+      "url": "https://github.com/pingcap/tidb/issues/31376",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:242ce56b0196",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/104",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:246a1af72aaa",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/14",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:252afab174b2",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-11/issues/42",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:25b70c6fd49a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/1",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:25ce8e9d0874",
+      "url": "https://github.com/cockroachdb/cockroach/issues/144923",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:25ffb3aa2be8",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-19/issues/17",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:263e41ee61bf",
+      "url": "https://github.com/cockroachdb/cockroach/issues/113089",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:27d48df6a14e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/53404",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:27dfb6163cca",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1138",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:28b1fcca98db",
+      "url": "https://github.com/cockroachdb/cockroach/issues/15026",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:28f5fab69913",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/23",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:2949a0523281",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/26",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:29510367b0cc",
+      "url": "https://github.com/tikv/tikv/issues/4291",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2966feb8bdae",
+      "url": "https://github.com/cockroachdb/cockroach/issues/97692",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:298e98ca1ba4",
+      "url": "https://github.com/sparq-org/sparq/issues/4645",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:29d03125cc1b",
+      "url": "https://github.com/cockroachdb/cockroach/issues/9151",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:29d3e183d00a",
+      "url": "https://github.com/apache/datafusion/issues/11030",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2accfdda0e9f",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/156",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:2adf97f310fc",
+      "url": "https://github.com/ydb-platform/ydb/issues/29532",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2b0fabb9264e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8236",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2b19279335c9",
+      "url": "https://github.com/sparq-org/sparq/issues/4635",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2b2c406562dc",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/17981",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2bdc912c3a89",
+      "url": "https://github.com/cockroachdb/cockroach/issues/42377",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2beb465ea41c",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-5/issues/59",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:2c1b716e83e8",
+      "url": "https://github.com/cockroachdb/cockroach/issues/115821",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2c52d7103252",
+      "url": "https://github.com/cockroachdb/cockroach/issues/69754",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2c58fd28774e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/3060",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2c5a1097ebdc",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1241",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:2cbe5d42bf89",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/bug-analysis-project-group_12/issues/5",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:2cf2c497e2b6",
+      "url": "https://github.com/cockroachdb/cockroach/issues/131110",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2d7797f49430",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/77",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:2d815fb025ee",
+      "url": "https://github.com/cockroachdb/cockroach/issues/168834",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2dd864dd007c",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/44",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:2ddc77dd0bb3",
+      "url": "https://github.com/cockroachdb/cockroach/issues/62900",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2e4ab40fd6ab",
+      "url": "https://github.com/cockroachdb/cockroach/issues/104381",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2e84f5d61666",
+      "url": "https://github.com/cockroachdb/cockroach/issues/101875",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2f95cb86f5a6",
+      "url": "https://github.com/cockroachdb/cockroach/issues/105103",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:2fa3de21f182",
+      "url": "https://github.com/apache/arrow-rs/issues/7886",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:30a9cd411ccf",
+      "url": "https://github.com/sqlancer/sqlancer/issues/645",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:30ff229a1d83",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1090",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:3118e518bcc1",
+      "url": "https://github.com/cockroachdb/cockroach/issues/110528",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:316b50af253f",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-11/issues/90",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:31a3df60836a",
+      "url": "https://github.com/yorklim/CS3213_Team_17/issues/40",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:31a5abedffc0",
+      "url": "https://github.com/ClickHouse/ClickHouse/issues/72493",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:31b91776164f",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/54",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:325920e48c78",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/160",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:32ed30317cb6",
+      "url": "https://github.com/cockroachdb/cockroach/issues/71675",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3363b30eddf1",
+      "url": "https://github.com/risingwavelabs/risingwave/issues/6267",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:33a476dd4eab",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-11/issues/24",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:33dcd757ff0f",
+      "url": "https://github.com/cockroachdb/cockroach/issues/168832",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:33e8e445c4ad",
+      "url": "https://github.com/cockroachdb/cockroach/issues/166648",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3476d6c90606",
+      "url": "https://github.com/risingwavelabs/risingwave/issues/3364",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:34a1537c3dce",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-6/issues/39",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:3524eb836fea",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/13",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:35369f2fb77f",
+      "url": "https://github.com/cockroachdb/cockroach/issues/73938",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3557065fb70d",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/33",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:3691c4496cb2",
+      "url": "https://github.com/CS3213-T6-1/sqlancer/issues/65",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:36a5669ac314",
+      "url": "https://github.com/Homebrew/brew/issues/11155",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:36bac3a48d62",
+      "url": "https://github.com/feldera/feldera/issues/2934",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:370c44b43eaf",
+      "url": "https://github.com/cockroachdb/cockroach/issues/168833",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:38c191f77c21",
+      "url": "https://github.com/crate/crate/issues/19669",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:38fdf1550e5d",
+      "url": "https://github.com/cockroachdb/cockroach/issues/171549",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:394295938da0",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1263",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:394f2560df15",
+      "url": "https://github.com/elastic/elasticsearch/issues/93494",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:39853cc2af59",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/8434",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3b5da484afd5",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-21/issues/17",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:3b64c063cf04",
+      "url": "https://github.com/cockroachdb/cockroach/issues/77439",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3c04a74139eb",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/14426",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3cbfa1a074e3",
+      "url": "https://github.com/sparq-org/sparq/issues/3452",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3d6fa0b8ee4d",
+      "url": "https://github.com/ydb-platform/ydb/issues/15971",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3e37aea9ecde",
+      "url": "https://github.com/caretdev/sqlancer-iris/issues/1",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3e71660d1d7c",
+      "url": "https://github.com/cockroachdb/cockroach/issues/114065",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3e72885a44eb",
+      "url": "https://github.com/cockroachdb/cockroach/issues/71216",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3f632fdbd360",
+      "url": "https://github.com/cockroachdb/cockroach/issues/36280",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:3fbdf6062136",
+      "url": "https://github.com/databaseService/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:402d78311973",
+      "url": "https://github.com/datalad/datalad-installer/issues/50",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:41b3f8c296fa",
+      "url": "https://github.com/ClickHouse/ClickHouse/issues/116422",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:41e35e095be9",
+      "url": "https://github.com/sqlancer/sqlancer/issues/909",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:41fac92f9e0c",
+      "url": "https://github.com/sqlancer/sqlancer/issues/440",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:426af117adb0",
+      "url": "https://github.com/tursodatabase/turso/issues/4681",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:434779c4ac28",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-24/issues/35",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:43c859a100fd",
+      "url": "https://github.com/tikv/tikv/issues/13959",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:453f75fa0453",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/6",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:456d0eadb138",
+      "url": "https://github.com/CS3213-T6-1/sqlancer/issues/62",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:4574de41f27f",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/15",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:457faf255b02",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8215",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:45a3b080d442",
+      "url": "https://github.com/cockroachdb/cockroach/issues/171414",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:4619c7992da0",
+      "url": "https://github.com/unified-systems-com/tap/issues/438",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:466793f4a5e4",
+      "url": "https://github.com/sqlancer/sqlancer/issues/234",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:46ab263688aa",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-24/issues/53",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:46ce61c586da",
+      "url": "https://github.com/cockroachdb/cockroach/issues/94615",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:475e8931ba2a",
+      "url": "https://github.com/pingcap/tidb/issues/59336",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:47a3a579b4fa",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/46",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:4816a9578557",
+      "url": "https://github.com/apache/doris/issues/19786",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:48af18b62d72",
+      "url": "https://github.com/sqlancer/sqlancer/issues/299",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:48bf6e7f6517",
+      "url": "https://github.com/spiceai/spiceai/issues/2189",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:491c461e0404",
+      "url": "https://github.com/cockroachdb/cockroach/issues/126063",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:49656dfeb5f0",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8252",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:49a88a09cb50",
+      "url": "https://github.com/xlab-uiuc/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:49c99c45067d",
+      "url": "https://github.com/cockroachdb/cockroach/issues/42567",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:4a699b52d0b9",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/26",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:4b5754cb1e9e",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/53",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:4c4597094178",
+      "url": "https://github.com/cockroachdb/cockroach/issues/47210",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:4d230b8aec32",
+      "url": "https://github.com/cockroachdb/cockroach/issues/92069",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:4d777f1b4044",
+      "url": "https://github.com/cockroachdb/cockroach/issues/144832",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:4dbed35d68f2",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-24/issues/31",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:4ea267bf3689",
+      "url": "https://github.com/sparq-org/sparq/issues/3087",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:50009d6d9193",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/34",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:508bed9ff84a",
+      "url": "https://github.com/elastic/elasticsearch/issues/36812",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:50e1828ccc00",
+      "url": "https://github.com/cockroachdb/cockroach/issues/61624",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:51ed3d4a5f9c",
+      "url": "https://bugs.mysql.com/bug.php?id=111421",
+      "reason": "the report's own tracker no longer serves it; nothing can be quoted or re-checked from here",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:51ed6391540c",
+      "url": "https://github.com/cockroachdb/cockroach/issues/62902",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:52316bfd3ce9",
+      "url": "https://github.com/GreptimeTeam/greptimedb/issues/4402",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:538affd00288",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/4",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:5405399ec0bd",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/3",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:54100a371d37",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-19/issues/19",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:542d21249f46",
+      "url": "https://github.com/hldgs/txn-check/issues/10",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:54313a89931a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/16",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:548190cb98fd",
+      "url": "https://github.com/cockroachdb/cockroach/issues/52951",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:54be9852fe1f",
+      "url": "https://github.com/tikv/tikv/issues/9234",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:55047cdd5503",
+      "url": "https://github.com/sqlancer/sqlancer/issues/725",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:5555f7ff6ed9",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/35",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:55d09fe8a413",
+      "url": "https://github.com/satanson/cpp_etudes/issues/13",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:56637ebaaf94",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8117",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:56d044380630",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/33",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:57a8e5fc032d",
+      "url": "https://github.com/tikv/tikv/issues/4816",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:57dd32004535",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-7/issues/8",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:5857df8fa101",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/25",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:586811f22fec",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/18",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:587a22b1aded",
+      "url": "https://github.com/cockroachdb/cockroach/issues/70483",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:587a65c52357",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/4",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:588d36509ab9",
+      "url": "https://github.com/cmu-db/noisepage/issues/997",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:58abd87c8c6a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-7/issues/14",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:58c40195a265",
+      "url": "https://github.com/cockroachdb/cockroach/issues/145025",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:5913fed9b13b",
+      "url": "https://github.com/cockroachdb/cockroach/issues/96149",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:59d1e4f2e945",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/20",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:5a3fe29e4a13",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-11/issues/8",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:5a73e34f443c",
+      "url": "https://github.com/cockroachdb/cockroach/issues/11800",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:5b7c504480ce",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/25",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:5bc3e1ac1428",
+      "url": "https://github.com/cockroachdb/cockroach/issues/14026",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:5bc3f8275986",
+      "url": "https://github.com/cockroachdb/cockroach/issues/144775",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:5bf1ebd0b45a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/25",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:5c3961bf602f",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-21/issues/19",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:5d9849932021",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/11",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:5dabbaafb025",
+      "url": "https://github.com/readysettech/readyset/issues/1534",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:5f02d0da6063",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1154",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:5f508b0d9c25",
+      "url": "https://github.com/zhenglin-charlie-li/Learning_Repo/issues/1",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:5f8e16be4b85",
+      "url": "https://github.com/cockroachdb/cockroach/issues/73558",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:60b86a39e2f6",
+      "url": "https://github.com/ClickHouse/ClickHouse/issues/17623",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:613ac7fe4565",
+      "url": "https://github.com/cockroachdb/cockroach/issues/3054",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:616624c6e8bb",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-19/issues/11",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:617cfeaa044d",
+      "url": "https://github.com/sqlancer/sqlancer/issues/762",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:6189680d4239",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/17",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:6208c78ecc98",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/151",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:62735bc3c561",
+      "url": "https://github.com/cockroachdb/cockroach/issues/48011",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:62a746b05015",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/14427",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6353bcfeb592",
+      "url": "https://github.com/cockroachdb/cockroach/issues/94612",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:64a006609ab6",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/66",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:64bc142a9ee7",
+      "url": "https://github.com/elastic/elasticsearch/issues/94049",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:64dc5bec662b",
+      "url": "https://github.com/sparq-org/sparq/issues/811",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:656ee35f5eb6",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/86",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:665af99f58a3",
+      "url": "https://github.com/cockroachdb/cockroach/issues/83746",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:66a8b312e4ef",
+      "url": "https://github.com/ruanyf/weekly/issues/2953",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:670e0e754bef",
+      "url": "https://github.com/cockroachdb/cockroach/issues/1674",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6718e2ed7859",
+      "url": "https://github.com/cockroachdb/cockroach/issues/69856",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:67a74ddec2e5",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/22",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:683a37674a2d",
+      "url": "https://github.com/cockroachdb/cockroach/issues/123695",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:683acac36ee7",
+      "url": "https://github.com/stoneatom/stonedb/issues/1916",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:68cf9f40eca1",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/2",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:692570e2ab6e",
+      "url": "https://github.com/spiceai/spiceai/issues/2187",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:692e550b2797",
+      "url": "https://github.com/Baymine/OLAP-radar/issues/206",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6967790853ce",
+      "url": "https://github.com/stoneatom/stonedb/issues/1916",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:69706a9ef2f8",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/17",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:69b874c4e402",
+      "url": "https://github.com/cockroachdb/cockroach/issues/13085",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6a8568493e04",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1025",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:6b56e6a58171",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/15",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:6b6febe81111",
+      "url": "https://github.com/cockroachdb/cockroach/issues/173330",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6bcc4f0bf240",
+      "url": "https://github.com/sqlancer/sqlancer/issues/270",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:6c5f6f2ec5dd",
+      "url": "https://bugs.mysql.com/bug.php?id=108851",
+      "reason": "the report's own tracker no longer serves it; nothing can be quoted or re-checked from here",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6c7990df7b24",
+      "url": "https://github.com/cockroachdb/cockroach/issues/6474",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6cd9a6910e86",
+      "url": "https://github.com/pingcap/tidb/issues/28276",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6ceacf953e64",
+      "url": "https://github.com/sparq-org/sparq/issues/4688",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6e8a556eaf59",
+      "url": "https://github.com/sparq-org/sparq/issues/3167",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6f1359e88f83",
+      "url": "https://github.com/cockroachdb/cockroach/issues/2131",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6f142cfbac9c",
+      "url": "https://github.com/yorklim/CS3213_Team_17/issues/79",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:6f39f7d129b6",
+      "url": "https://github.com/cockroachdb/cockroach/issues/94551",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:6f97dfc15c31",
+      "url": "https://github.com/cockroachdb/cockroach/issues/143716",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:704b6ed33ce6",
+      "url": "https://github.com/CS3213-T6-1/sqlancer/issues/64",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:706f0b52fb02",
+      "url": "https://github.com/cockroachdb/cockroach/issues/78905",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:70f99e564ddd",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/155",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:7126a9ef4a9b",
+      "url": "https://github.com/apache/datafusion/issues/21076",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:71550719883b",
+      "url": "https://github.com/questdb/questdb/issues/4079",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:71c425e65de0",
+      "url": "https://github.com/cockroachdb/cockroach/issues/94587",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:71cc62614e36",
+      "url": "https://github.com/StarRocks/starrocks/blob/444cb3cf593a8406c096ca79d270908411414654/docs/ja/release_notes/release-2.2.md",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:72a3fd57ce70",
+      "url": "https://github.com/dolthub/dolt/issues/3636",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:72f2b6a92682",
+      "url": "https://github.com/cockroachdb/cockroach/issues/76729",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7319878606a8",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/158",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:734fe55cfdba",
+      "url": "https://github.com/sparq-org/sparq/issues/4100",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:735b01b9febd",
+      "url": "https://github.com/RyanL1997/sql/issues/10",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:737f09401d16",
+      "url": "https://github.com/cockroachdb/cockroach/issues/138012",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:73a5cedd40eb",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1174",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:75088f0b6615",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8030",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:75363123d18d",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1045",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:75b495e66385",
+      "url": "https://github.com/cockroachdb/cockroach/issues/7907",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:75bd40360dd9",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-19/issues/104",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:75c491de6bb5",
+      "url": "https://github.com/pingcap/tidb/issues/70260",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:76785a338ebd",
+      "url": "https://github.com/Zahgon/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7683b9810094",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/7",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:768843475466",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/15",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:76b9d75482cf",
+      "url": "https://github.com/tensorflow/tensorflow/issues/51403",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7733fb41078c",
+      "url": "https://github.com/spiceai/spiceai/issues/2188",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:77ddde5f2b5c",
+      "url": "https://github.com/cockroachdb/cockroach/issues/9178",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:77fe36e528b4",
+      "url": "https://github.com/sparq-org/sparq/issues/4783",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:788c9e845a7e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/144834",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:79db674d359e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/128094",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7b790395eb86",
+      "url": "https://github.com/cockroachdb/cockroach/issues/94920",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7b88d86e97de",
+      "url": "https://github.com/DimaMend/ClickHouse/issues/16",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7bb45451ae05",
+      "url": "https://github.com/cockroachdb/cockroach/issues/65087",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7c3515b65c6d",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8872",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7c671ad750e1",
+      "url": "https://github.com/satanson/starrocks/issues/7",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7cb1494023ca",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/12",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:7d087ab00aeb",
+      "url": "https://github.com/CS3213-T6-1/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7d10dc9a9346",
+      "url": "http://bugs.mysql.com/113298",
+      "reason": "the report's own tracker no longer serves it; nothing can be quoted or re-checked from here",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7dc36a7863d7",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/87",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:7e5537b42cd6",
+      "url": "https://github.com/cockroachdb/cockroach/issues/144835",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:7eed81eecf8a",
+      "url": "https://github.com/spiceai/spiceai/issues/2186",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8051086f39e3",
+      "url": "https://github.com/cockroachdb/cockroach/issues/29968",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8113f9aa0734",
+      "url": "https://github.com/pingcap/tidb/issues/11961",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:811b89d490dd",
+      "url": "https://github.com/elastic/elasticsearch/issues/94270",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8160ff532667",
+      "url": "https://github.com/cockroachdb/cockroach/issues/106781",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:82f21e34d1c2",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/14866",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:833b4096aad3",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/2",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:839aafa0cda2",
+      "url": "https://github.com/cockroachdb/cockroach/issues/49381",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:83bf502f8443",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/4",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:84c56b296333",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/bug-analysis-project-group-10/issues/5",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:8512cdc43fe1",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-24/issues/42",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:85137ce01c20",
+      "url": "https://github.com/sqlancer/sqlancer/issues/288",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:8514f3fe6fea",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/110",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:856f88780cb4",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-5/issues/11",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:859ff7f50981",
+      "url": "https://github.com/cockroachdb/cockroach/issues/144774",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:85af1613c487",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1056",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:86290af2f716",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/14864",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:86a95e2c6c08",
+      "url": "https://github.com/cockroachdb/cockroach/issues/4269",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:86af4565bb9b",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1042",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:86c1bc63aa72",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1001",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:87156df703c3",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/58",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:8726b92dbfad",
+      "url": "https://github.com/cockroachdb/cockroach/issues/11637",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8768b847d141",
+      "url": "https://github.com/cockroachdb/cockroach/issues/19357",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:87e7211488a4",
+      "url": "https://github.com/apache/datafusion/issues/11106",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:886b46cd76e5",
+      "url": "https://bugs.mysql.com/bug.php?id=108852",
+      "reason": "the report's own tracker no longer serves it; nothing can be quoted or re-checked from here",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:886be086ba8c",
+      "url": "https://github.com/cockroachdb/cockroach/issues/10543",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:88cbaf029745",
+      "url": "https://bugs.mysql.com/bug.php?id=111241",
+      "reason": "the report's own tracker no longer serves it; nothing can be quoted or re-checked from here",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8905ca9fb823",
+      "url": "https://github.com/cockroachdb/cockroach/issues/151995",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:89129d0a0f6e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/19969",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:89175b812e50",
+      "url": "https://github.com/cockroachdb/cockroach/issues/95031",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:89e6bfa338f8",
+      "url": "https://github.com/clemenza/tinytable-evals/issues/4",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8a02638bb07a",
+      "url": "https://github.com/elastic/elasticsearch/issues/31154",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8bea9867bb25",
+      "url": "https://github.com/Homebrew/brew/issues/11123",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:8c0478f99064",
+      "url": "https://github.com/grafana/homebrew-pyroscope/issues/1",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8cf888c7e111",
+      "url": "https://github.com/cockroachdb/cockroach/issues/126239",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8d2928f7560f",
+      "url": "https://github.com/cockroachdb/cockroach/issues/144869",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8d2c0b71ef30",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/54",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:8d33e8c32a3e",
+      "url": "https://github.com/duckdb/duckdb/issues/5031",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8e03f6a6af9b",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1186",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:8e3a688e40f0",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/164",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:8e3db4670570",
+      "url": "https://github.com/elastic/elasticsearch/issues/88486",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8e6b2040f964",
+      "url": "https://github.com/cockroachdb/cockroach/issues/37476",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8e92e9a67dda",
+      "url": "https://github.com/cockroachdb/cockroach/issues/144813",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8e970e226d6a",
+      "url": "https://github.com/cockroachdb/cockroach/issues/131532",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8f3ed5133b66",
+      "url": "https://github.com/cockroachdb/cockroach/issues/85211",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:8f7922428b2b",
+      "url": "https://github.com/CS3213-T6-1/sqlancer/issues/27",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:8f8769a630db",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/19",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:8fb963d4c4c1",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-5/issues/61",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:906c65f83a3e",
+      "url": "https://sqlite.org/forum/forumpost/3776b48e71, https://www.sqlite.org/src/info/01868ebcd25fadb2",
+      "reason": "the report's own tracker no longer serves it; nothing can be quoted or re-checked from here",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9097736fc0fb",
+      "url": "https://github.com/sqlancer/sqlancer/issues/423",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:910a693b2be7",
+      "url": "https://github.com/ClickHouse/praktika/issues/106",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9138d34943be",
+      "url": "https://github.com/cockroachdb/cockroach/issues/4118",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:919f1306f86f",
+      "url": "https://github.com/cockroachdb/cockroach/issues/122355",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:91c590dd447e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/171389",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:91d125966f60",
+      "url": "https://github.com/tikv/tikv/issues/11588",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:91fce83a9a56",
+      "url": "https://github.com/cockroachdb/cockroach/issues/132631",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:924bf09a603b",
+      "url": "https://github.com/sqlancer/sqlancer/issues/271",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:92cd0b440dcf",
+      "url": "https://github.com/sparq-org/sparq/issues/4571",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:93573ce67602",
+      "url": "https://github.com/cockroachdb/cockroach/issues/79734",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:935bb101d9d3",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1179",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:9372f2402084",
+      "url": "https://github.com/tikv/tikv/issues/18100",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9385425d29d7",
+      "url": "https://github.com/cockroachdb/cockroach/issues/9153",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:93bcfd591f3e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8167",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:93c85dd29e4f",
+      "url": "https://github.com/cockroachdb/cockroach/issues/141329",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:943b20631b8a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/157",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:95070ac7117a",
+      "url": "https://github.com/cockroachdb/cockroach/issues/100353",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:956d11f50b06",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1222",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:96333f4d6476",
+      "url": "https://github.com/sqlancer/sqlancer/issues/824",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:9636e4c08887",
+      "url": "https://github.com/cockroachdb/cockroach/issues/53950",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:96802bf05936",
+      "url": "https://github.com/cockroachdb/cockroach/issues/174568",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:96b9c7f11782",
+      "url": "https://github.com/cockroachdb/cockroach/issues/144366",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:975540c73b91",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/18",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:97f717dc1abb",
+      "url": "https://github.com/cockroachdb/cockroach/issues/138218",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:99128facc06c",
+      "url": "https://github.com/sqlancer/sqlancer/issues/677",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:999c8c04903e",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-19/issues/18",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:99aec5e99fe0",
+      "url": "https://github.com/cockroachdb/cockroach/issues/1350",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:99b19bb46def",
+      "url": "https://github.com/cnosdb/sqlancer/blob/master/src/sqlancer/timescaledb/TimescaleDBBugs.java",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:99fc66139c3e",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/8345",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9a1c78860b06",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/24",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:9a4e3d786b7e",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/55",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:9a570474c650",
+      "url": "https://perconadev.atlassian.net/browse/PS-9049",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9a7a2267540a",
+      "url": "https://github.com/crate/crate/issues/4921",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9a9c21c0a4a4",
+      "url": "https://github.com/cockroachdb/cockroach/issues/96215",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9ba057718ad2",
+      "url": "https://github.com/cockroachdb/cockroach/issues/126758",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9bf2b35d0ee1",
+      "url": "https://github.com/Baymine/OLAP-radar/issues/216",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9c46b7aeff22",
+      "url": "https://github.com/ntdp/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9cb78d3668ce",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-24/issues/4",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:9cb814df067b",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/sqlancer-template",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9cd0d07b7042",
+      "url": "https://github.com/spiceai/spiceai/blob/64fc04b9466ce7010a57c067fe7df47b015d62c6/crates/cayenne/tests/correctness/support/sqllancer.rs",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9cd6a1dfcccb",
+      "url": "https://github.com/cockroachdb/cockroach/issues/173450",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9d7f27289b27",
+      "url": "https://github.com/cockroachdb/cockroach/issues/13307",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9d9d58f4b89c",
+      "url": "https://github.com/cockroachdb/cockroach/issues/66435",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9daa3e1dd236",
+      "url": "https://github.com/cockroachdb/cockroach/issues/75853",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9dc439289ac5",
+      "url": "https://github.com/cockroachdb/cockroach/issues/129181",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:9f653843f282",
+      "url": "https://github.com/sqlancer/sqlancer/issues/563",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:9ff5d6fc2483",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/10",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a0247ea2fb2b",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/58",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a02a380119e8",
+      "url": "https://github.com/cockroachdb/cockroach/issues/128765",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a0582f622776",
+      "url": "https://github.com/cockroachdb/cockroach/issues/172480",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a09aff7f2568",
+      "url": "https://github.com/awesomeDataTool/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a10fe251d2df",
+      "url": "https://github.com/cockroachdb/cockroach/issues/76677",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a1f20205b873",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/21",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a1feae5f28a8",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/8",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a21d591c9d86",
+      "url": "https://github.com/sqlancer/sqlancer/issues/517",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a26a98ac4277",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/16",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a2806a46e05f",
+      "url": "https://github.com/opensearch-project/sql/issues/3220",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a2a22ab3abdb",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/38",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a2f0ba674700",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/1",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a317174683fc",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/20",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a359fc416d9e",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/10",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a3a7717dca1d",
+      "url": "https://github.com/spiceai/spiceai/issues/10116",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a3ae5a59a3f9",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-19/issues/39",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a4b22af457c6",
+      "url": "https://github.com/RedisGraph/RedisGraph/issues/2931",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a4eac7de6b77",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/16",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a5042e00f358",
+      "url": "https://github.com/tamnd/rudb/issues/104",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a51d2c9d9fbd",
+      "url": "https://github.com/Homebrew/homebrew-core/issues/108964",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a5b0c731ae00",
+      "url": "https://github.com/CS3213-G18/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a5e56dcaebd0",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-7/issues/12",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a654588aeed6",
+      "url": "https://github.com/sparq-org/sparq/issues/2553",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a6617b63125b",
+      "url": "https://github.com/elastic/elasticsearch/issues/120903",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a6a7a5cb67d5",
+      "url": "https://github.com/sqlancer/sqlancer/issues/307",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a7d4b96ad237",
+      "url": "https://github.com/cockroachdb/cockroach/issues/130253",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a7f1a0a1dbe8",
+      "url": "https://github.com/cockroachdb/cockroach/issues/25165",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a82eb6440545",
+      "url": "https://github.com/cockroachdb/cockroach/issues/84720",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a8e4a7009c26",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1044",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a967ad55c4a9",
+      "url": "https://github.com/cockroachdb/cockroach/issues/47306",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a97a62eed5c7",
+      "url": "https://github.com/cockroachdb/cockroach/issues/1167",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a9e7019de80e",
+      "url": "https://github.com/taiga-programming/lireddit/issues/83",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:a9f1ebcb5f7c",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/35",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:a9f2cf35f5ef",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/18",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:aab751728baf",
+      "url": "https://github.com/elastic/elasticsearch/issues/37841",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:aab7e02ffe55",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/27",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:aaff0f5408e4",
+      "url": "https://github.com/cockroachdb/cockroach/issues/97200",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ab0c55ef4cdc",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1049",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:ac309293e7e4",
+      "url": "https://github.com/questdb/questdb/issues/595",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ac5d3d10af9d",
+      "url": "https://github.com/elastic/elasticsearch/issues/61087",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ac737ff32205",
+      "url": "https://github.com/spiceai/spiceai/issues/2119",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:aca85082cf84",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/153",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:acb286a4feff",
+      "url": "https://github.com/cockroachdb/cockroach/issues/115793",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ad0c8d21e2ee",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/bug-analysis-project-group-13/issues/18",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:adcadbef6157",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/38",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:adde353715c9",
+      "url": "https://github.com/sqlancer/sqlancer/issues/569",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:adea8aaa32e2",
+      "url": "https://github.com/CS3213-T6-1/sqlancer/issues/61",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:ae243d04a661",
+      "url": "https://bugs.mysql.com/bug.php?id=108833",
+      "reason": "the report's own tracker no longer serves it; nothing can be quoted or re-checked from here",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ae751c6df84d",
+      "url": "https://github.com/risingwavelabs/risingwave/issues/7504",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:aeb2d04dac9e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/10873",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:af88b2c5a515",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/33",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:afc9c02161a1",
+      "url": "https://github.com/cockroachdb/cockroach/issues/171981",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:affc84bb5717",
+      "url": "https://github.com/sparq-org/sparq/issues/1474",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b03060da964a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/5",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b030be67fdd1",
+      "url": "https://github.com/cockroachdb/cockroach/issues/72648",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b0869828492e",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-19/issues/48",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b08d97adee21",
+      "url": "https://github.com/cockroachdb/cockroach/issues/10877",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b0e0b3580058",
+      "url": "https://github.com/sqlancer/sqlancer/issues/764",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b1621a80f2bc",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1104",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b1da9673a26d",
+      "url": "https://github.com/cockroachdb/cockroach/issues/4450",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b2ad3aa790f3",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/12006",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b327e558cd1e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/92718",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b39a66f02404",
+      "url": "https://github.com/sparq-org/sparq/issues/4103",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b4d2e7a6a6c4",
+      "url": "https://github.com/CS3213-T6-1/sqlancer/issues/63",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b51636fdc47d",
+      "url": "https://github.com/Baymine/OLAP-radar/issues/99",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b54b55c7684c",
+      "url": "https://github.com/cockroachdb/cockroach/issues/171383",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b550454c8b87",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1057",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b5b6adae924f",
+      "url": "https://github.com/timb-machine-mirrors/sqlancer-sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b5dd8aae519c",
+      "url": "https://github.com/HSLdevcom/transitlog-ui/issues/104",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b5ea4c630d1a",
+      "url": "https://github.com/datafusion-contrib/datafusion-distributed/issues/239",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b62ece438590",
+      "url": "https://github.com/sqlancer/sqlancer/issues/704",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b66e59dd8cc4",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-19/issues/27",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b6e5ddaab245",
+      "url": "https://github.com/sparq-org/sparq/issues/3321",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b6eaa5ebf158",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/8",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b74530942d86",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1043",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b77b50424510",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1050",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b7974797f705",
+      "url": "https://github.com/hazelcast/hazelcast/issues/5560",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b7e4ad6dd486",
+      "url": "https://github.com/Homebrew/brew/issues/11109",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:b96625a0a283",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/31665",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b99afc93f133",
+      "url": "https://github.com/GreptimeTeam/greptimedb/issues/4402",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:b9dc0a55d1b5",
+      "url": "https://github.com/cockroachdb/cockroach/issues/175282",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:bae8b2ad9dd0",
+      "url": "https://github.com/ydb-platform/ydb/issues/7728",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:bb5ea4644994",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/14856",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:bb81a6e437a2",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-24/issues/69",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:bbb316cb4b43",
+      "url": "https://github.com/hazelcast/hazelcast/issues/22577",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:bc08ff09d460",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/27",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:bc701af17dff",
+      "url": "https://github.com/cockroachdb/cockroach/issues/156254",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:bca7d32ea47e",
+      "url": "https://github.com/neo4j/neo4j/issues/3740",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:bd49f82815fc",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8242",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:bd8ae76dd825",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-14/issues/49",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:bda549c80b16",
+      "url": "https://github.com/cockroachdb/cockroach/issues/9318",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:be0ea23401f7",
+      "url": "https://github.com/cockroachdb/cockroach/issues/43870",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:beab77913113",
+      "url": "https://github.com/CS3213-group-14/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:bfe0353a0fc5",
+      "url": "https://github.com/cockroachdb/cockroach/issues/43847",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c02924cd9a3b",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/1",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:c02b069f14f9",
+      "url": "https://github.com/StarRocks/starrocks/blob/444cb3cf593a8406c096ca79d270908411414654/docs/zh/release_notes/release-2.2.md",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c0979304d4ea",
+      "url": "https://github.com/cockroachdb/cockroach/issues/4217",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c14d358ad938",
+      "url": "https://github.com/cockroachdb/cockroach/issues/123062",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c17c0d7377d9",
+      "url": "https://github.com/cockroachdb/cockroach/issues/37144",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c1c33098f180",
+      "url": "https://github.com/databendlabs/databend/issues/9448",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c24f3ff794bd",
+      "url": "https://github.com/apache/datafusion/issues/10403",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c25daa71474e",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/bug-analysis-project-group-13/issues/20",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:c29757938d3c",
+      "url": "https://perconadev.atlassian.net/jira/software/c/projects/PS/issues/PS-9085",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c2d2c712a8cd",
+      "url": "https://github.com/spiceai/spiceai/issues/10832",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c3d2fa5ef973",
+      "url": "https://github.com/databendlabs/databend/issues/8238",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c41b5dd40e3f",
+      "url": "https://github.com/cockroachdb/cockroach/issues/43974",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c429ec03cab3",
+      "url": "https://github.com/pingcap/tidb/issues/10473",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c48bfbe2f92e",
+      "url": "https://github.com/sparq-org/sparq/issues/4101",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c4c17cdfe8f4",
+      "url": "https://github.com/cockroachdb/cockroach/issues/127745",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c524580ab623",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/24",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:c5270eb19fb4",
+      "url": "https://github.com/cmu-mse-cmudb/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c52bb4e2d8a5",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1055",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:c52d804bbf95",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/58",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:c5f679418b71",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-14/issues/41",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:c6490a26697a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/39",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:c699916a9307",
+      "url": "https://bugs.mysql.com/bug.php?id=113180",
+      "reason": "the report's own tracker no longer serves it; nothing can be quoted or re-checked from here",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c6aaae3249a0",
+      "url": "https://github.com/cockroachdb/cockroach/issues/172320",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c6b6f050df11",
+      "url": "https://github.com/tarantool/tarantool/issues/4826",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c70b7243dd83",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8249",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c751326737cf",
+      "url": "https://github.com/cockroachdb/cockroach/issues/60632",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c7e52999adb7",
+      "url": "https://github.com/cockroachdb/cockroach/issues/68174",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c9767c38f457",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/5594",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:c99b71ceb2bf",
+      "url": "https://github.com/cmu-db/noisepage/issues/1041",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ca2bf470eebb",
+      "url": "https://github.com/tikv/tikv/issues/3679",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ca4e9dc62dca",
+      "url": "https://github.com/cockroachdb/cockroach/issues/43729",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:cae4760a1884",
+      "url": "https://github.com/Homebrew/homebrew-core/issues/75028",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:cbdfd1214afa",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8223",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:cc868ad3bfec",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/48",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:ccfa59b63f20",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/46",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:cd2778e5ca88",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-11/issues/65",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:cd61cff99dfd",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/2",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:cda489b58e74",
+      "url": "https://github.com/antithesishq/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:cda9bb437720",
+      "url": "https://github.com/sourcenetwork/defradb.rs/issues/157",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:cdbd7dd56c5a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/9",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:cdd4bcb8e7af",
+      "url": "https://github.com/cockroachdb/cockroach/issues/132769",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ce0822888f5d",
+      "url": "https://github.com/sparq-org/sparq/issues/4114",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ce6b70e2eb6b",
+      "url": "https://github.com/jasonmhatfield/york-njrotc-ui/issues/3",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:cea793ccf3bc",
+      "url": "https://github.com/matrixorigin/matrixone/issues/1670",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:cfe514c474ae",
+      "url": "https://github.com/sparq-org/sparq/issues/4791",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:d009b40e529a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/30",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d02b52cb1f7b",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-11/issues/45",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d0c4e64e251a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/12",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d1530c85872c",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/13",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d1e64f4416dd",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-24/issues/47",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d26a5aa90913",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-11/issues/69",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d2fbc7eb2ec7",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-21/issues/7",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d353b0eb1d87",
+      "url": "https://github.com/opensearch-project/sql/issues/3266",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:d390c693f1c8",
+      "url": "https://github.com/satanson/starrocks/issues/6",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:d4a5ec5c25e9",
+      "url": "https://github.com/HSLdevcom/transitlog-ui/issues/39",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:d4ad0f93af2f",
+      "url": "https://github.com/sparq-org/sparq/issues/4102",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:d5003979a306",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-5/issues/56",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d5a8d7cbdb9f",
+      "url": "https://github.com/cockroachdb/cockroach/issues/94555",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:d5d7938148a1",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-15/issues/8",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d6148da918f9",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1151",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d62f9d90fd99",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/37",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d68d47fff958",
+      "url": "https://github.com/snowdensb/ClickHouse/issues/4",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:d729ed4000f3",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-11/issues/84",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d74375622225",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/bug-analysis-project-group_12/issues/1",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d7edf47fbeb6",
+      "url": "https://github.com/cockroachdb/cockroach/issues/7419",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:d83052e26b9a",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8132",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:d88f311dddd3",
+      "url": "https://github.com/sparq-org/sparq/issues/4126",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:d9af50ac14b5",
+      "url": "https://github.com/sqlancer/sqlancer/issues/183",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:d9cfc8f8c95a",
+      "url": "https://github.com/derekmwright/wadjet/issues/289",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:da24a1fd4ba3",
+      "url": "https://github.com/sqlancer/sqlancer/issues/236",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:da5f8939cb8f",
+      "url": "https://github.com/cockroachdb/cockroach/issues/89588",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:dadd6ec33e1b",
+      "url": "https://github.com/sqlancer/sqlancer/issues/90",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:db5113bfbd44",
+      "url": "https://github.com/cockroachdb/cockroach/issues/4464",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:db8e21779717",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/23",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:db8ef8e84e0b",
+      "url": "https://github.com/cockroachdb/cockroach/issues/150885",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:dbe8e869cb7c",
+      "url": "https://github.com/cockroachdb/cockroach/issues/133360",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:dbf3ffc9ffcd",
+      "url": "https://github.com/pingcap/tidb/issues/38332",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:dc5c3dc9f61a",
+      "url": "https://github.com/yorklim/CS3213_Team_17/issues/53",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:dcac0c16ba9a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/bug-analysis-project-group_12/issues/4",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:dcf81eb3e483",
+      "url": "https://github.com/pingcap/tidb/issues/22429",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:dd677130d2b5",
+      "url": "https://github.com/pingcap/tidb/issues/16381",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:dde04c35d4a5",
+      "url": "https://github.com/sqlancer/sqlancer/issues/314",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:dde382107b00",
+      "url": "https://github.com/tarantool/tarantool/issues/4833",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ddfa25af4e44",
+      "url": "https://github.com/cockroachdb/cockroach/issues/2421",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:de3173c8fd92",
+      "url": "https://github.com/sqlancer/sqlancer/issues/761",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:dec802cdde8d",
+      "url": "https://github.com/sqlancer/sqlancer/issues/971",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:dedc96a1f6ac",
+      "url": "https://github.com/apache/datafusion/issues/14535",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:def09ad1ca82",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/22",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:e00bf55aa826",
+      "url": "https://github.com/tikv/tikv/issues/8883",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e0c6f12d242c",
+      "url": "https://github.com/CS3213-T6-1/sqlancer/issues/25",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:e210556c351b",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/150",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:e29c1fe35bc8",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/24",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:e29dc74e9d66",
+      "url": "https://github.com/sqlancer/sqlancer/issues/630",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:e3ed2b92c4ec",
+      "url": "https://github.com/cockroachdb/cockroach/issues/81013",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e511a48d51c2",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1024",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:e65f44e84219",
+      "url": "https://github.com/cockroachdb/cockroach/issues/114038",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e6f1b55ace77",
+      "url": "https://github.com/sparq-org/sparq/issues/4569",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e724ea94ed14",
+      "url": "https://github.com/cockroachdb/cockroach/issues/144784",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e7bd9393cacf",
+      "url": "https://github.com/duckdblabs/duckdb-fuzzer-ci/issues/7",
+      "reason": "classified and turned down on a later run",
+      "decided_by": "pipeline",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e7c2588fde97",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/34",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:e7db32ff44b4",
+      "url": "https://github.com/cockroachdb/cockroach/issues/88869",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e83c9f141440",
+      "url": "https://github.com/risingwavelabs/risingwave/issues/6266",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e8745f5e2860",
+      "url": "https://github.com/cockroachdb/cockroach/issues/87418",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e88c3573e0e4",
+      "url": "https://github.com/realm/SwiftLint/issues/3596",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e8e76a59918a",
+      "url": "https://github.com/cockroachdb/cockroach/issues/109498",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:e9ceaf026f0c",
+      "url": "https://github.com/derekmwright/wadjet/issues/588",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ea669dc4998a",
+      "url": "https://github.com/cockroachdb/cockroach/issues/8153",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ea7d35ac72a8",
+      "url": "https://github.com/sparq-org/sparq/issues/4570",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:eae86265ae96",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-24/issues/13",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:eb78aa7c5f2e",
+      "url": "https://github.com/cockroachdb/cockroach/issues/12432",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:eba73291749e",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/benchmark-group-16/issues/36",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:ebd2c425c875",
+      "url": "https://github.com/cnosdb/cnosdb/issues/860",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ebe4babf7929",
+      "url": "https://github.com/cockroachdb/cockroach/issues/71530",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ec8742bf18d9",
+      "url": "https://github.com/pingcap/tidb/issues/12196",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ec8e1621b711",
+      "url": "https://github.com/itsharex/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ec97e94f2f6b",
+      "url": "https://github.com/apache/datafusion/issues/11190",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ed11f1e8cb83",
+      "url": "https://github.com/cockroachdb/cockroach/issues/10517",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ed18e1fb2b74",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/12038",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ed92ea235aad",
+      "url": "https://github.com/ydb-platform/ydb/issues/49010",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:edd2c323da67",
+      "url": "https://github.com/cockroachdb/cockroach/issues/12168",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ee416e1e3ec3",
+      "url": "https://github.com/yugabyte/yugabyte-db/issues/6619",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ee8a6fdeaf74",
+      "url": "https://github.com/cockroachdb/cockroach/issues/127468",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ee96c58851a1",
+      "url": "https://github.com/cockroachdb/cockroach/issues/119849",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:ef0ab18bcb44",
+      "url": "https://github.com/Baymine/OLAP-radar/issues/207",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:efc54bf1b983",
+      "url": "https://github.com/stoneatom/stonedb/issues/987",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f084f4b85d33",
+      "url": "https://github.com/cmu-db/noisepage/issues/1012",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f0ce124ae109",
+      "url": "https://github.com/apache/doris/issues/50342",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f110dac351e1",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/28",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:f194e3bbd32a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-7/issues/28",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:f1b3668610b9",
+      "url": "https://github.com/sqlancer/sqlancer/issues/61",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:f1fd1f1f953f",
+      "url": "https://github.com/sqlancer/sqlancer/issues/518",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:f2474aa1eefa",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1158",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:f27a7453f6a7",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-5/issues/32",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:f2aaff8556ad",
+      "url": "https://github.com/mgramin/awesome-db-tools/issues/175",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f2deb7eaa746",
+      "url": "https://github.com/cockroachdb/cockroach/issues/14985",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f30892dd11b0",
+      "url": "https://github.com/shaneclarke-whitesource/ClickHouse/issues/44",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f33077bd5cf4",
+      "url": "https://github.com/sweenu/setup/issues/16",
+      "reason": "not a database system's own tracker: a personal fork, an unrelated project, a tooling repository or a news digest",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f330c11a4817",
+      "url": "https://github.com/apache/datafusion/issues/12114",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f36926ff6f6c",
+      "url": "https://github.com/sparq-org/sparq/issues/4637",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f3732a8e5a22",
+      "url": "https://github.com/cockroachdb/cockroach/issues/171420",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f38b5a251256",
+      "url": "https://github.com/cockroachdb/cockroach/issues/171381",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f3ee6a37402a",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-11/issues/12",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:f46875c8dae4",
+      "url": "https://github.com/sparq-org/sparq/issues/2961",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f49f962b190d",
+      "url": "https://github.com/tikv/tikv/issues/4874",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f623cb51cb12",
+      "url": "https://github.com/cockroachdb/cockroach/issues/88660",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f66613c878d4",
+      "url": "https://github.com/pingcap/tidb/issues/29002",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f707a18dcfb0",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/33",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:f7352494bb66",
+      "url": "https://github.com/ydb-platform/ydb/issues/32959",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f76f28cd0ef3",
+      "url": "https://github.com/cockroachdb/cockroach/issues/9202",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f7ae0ab5e932",
+      "url": "https://github.com/cockroachdb/cockroach/issues/7238",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f7b027d408b6",
+      "url": "https://github.com/cockroachdb/cockroach/issues/6786",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f83652e70360",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/7",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:f88f6d757efc",
+      "url": "https://github.com/cockroachdb/cockroach/issues/3550",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:f890af63c106",
+      "url": "https://github.com/sqlancer/sqlancer/issues/1211",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:f8e98f5f9822",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-2/issues/28",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:fa049ddcc48d",
+      "url": "https://github.com/Homebrew/homebrew-core/issues/75243",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:fa94c82c139a",
+      "url": "https://github.com/javalibrary/sqlancer",
+      "reason": "a fork of SQLancer by a course group, a mirror or an individual -- not a database system project adopting it",
+      "decided_by": "Claude (reviewed in session)",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:fb25b2c8191f",
+      "url": "https://github.com/cockroachdb/cockroach/issues/83933",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:fc2c48779bf1",
+      "url": "https://github.com/cockroachdb/cockroach/issues/73920",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:fc7fd313ae73",
+      "url": "https://github.com/sparq-org/sparq/issues/3311",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:fc8addc71f79",
+      "url": "https://github.com/cockroachdb/cockroach/issues/172379",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    },
+    {
+      "id": "review:fd299e5e5a29",
+      "url": "https://github.com/sqlancer/sqlancer/issues/273",
+      "reason": "Not a database system: SQLancer's own tracker and a package manager, neither of which can carry a DBMS bug.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:fdc21c4b26be",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-22/issues/76",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:fefaabbf2dd1",
+      "url": "https://github.com/NUS-CS3213-AY2425S2/cs3213-project-group-4/issues/56",
+      "reason": "Student project repositories from the NUS CS3213 course. These are coursework forks of SQLancer, not database systems, and the issues in them are exercises.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-09"
+    },
+    {
+      "id": "review:ff45cf236654",
+      "url": "https://github.com/cockroachdb/cockroach/issues/17842",
+      "reason": "classified and turned down: the report does not attribute the find to SQLancer or one of its techniques",
+      "decided_by": "classifier",
+      "decided_on": "2026-09-13"
+    }
+  ]
+}
diff --git a/_data/impact/paper_decisions.json b/_data/impact/paper_decisions.json
new file mode 100644
index 0000000..51487af
--- /dev/null
+++ b/_data/impact/paper_decisions.json
@@ -0,0 +1,46 @@
+{
+  "schema_version": "1.0.0",
+  "description": "Decisions about paper records that automatic deduplication could not settle. Same authors and similar titles do not distinguish one work indexed twice from two works by one group, so each of these was read and decided, and the reason is recorded so the pair is not raised again.",
+  "same_work": [
+    {
+      "ids": [
+        "paper:doi:10.1145/3729175",
+        "paper:doi:10.1145/3712057"
+      ],
+      "canonical": "paper:doi:10.1145/3729175",
+      "reason": "One article published twice by ACM: 'Systems Correctness Practices at AWS' in Queue (2024) and 'at Amazon Web Services' in CACM (2025), same authors and same content. Automatic merging did not fire because both are published versions rather than a preprint and its paper."
+    },
+    {
+      "ids": [
+        "paper:doi:10.14778/3712221.3712247",
+        "paper:arxiv:2406.09469"
+      ],
+      "canonical": "paper:doi:10.14778/3712221.3712247",
+      "reason": "The PVLDB paper and its own preprint: the same Prolog implementation of SQL's denotational semantics used as a reference for differential testing, reporting the same 19 bugs. The published version rewrote the abstract and grew from four systems to six, so the abstracts score 0.04 alike and the automatic rule correctly declined to decide."
+    },
+    {
+      "ids": [
+        "paper:doi:10.1145/3650212.3680318",
+        "paper:s2:1b209b2da737fd26c50a2cb9116c73c9d1db904e"
+      ],
+      "canonical": "paper:doi:10.1145/3650212.3680318",
+      "reason": "One work indexed twice: the 2022 record is an earlier version of the 2024 ISSTA paper. Both are DBStorm, both generate workloads for transaction testing, and both measure validity and code coverage against the same three baselines -- SQLsmith and SQLancer as generation-based, Squirrel as mutation-based. The related-work sentence survives the rewrite: 'SQLancer [37] generates queries to fetch a specific target row' became 'SQLancer [60] creates a query that would fetch a target data to detect logical bug', renumbered for a new bibliography. An earlier note here claimed they were different work on the grounds that the 2022 sentences do not appear in the 2024 PDF, which shows only that the paper was rewritten. The automatic check could not compare them because the 2022 record carries no authors."
+    }
+  ],
+  "different_work": [
+    {
+      "ids": [
+        "paper:arxiv:2606.11132",
+        "paper:doi:10.1145/3803437.3806090"
+      ],
+      "reason": "Same authors and year, but two papers: the arXiv record is DiscPBT, a property-based testing engine for Spark with eight meta-properties and an evaluation against CometFuzz, while the ACM record is a short position paper arguing that data-intensive systems need property-based testing at all."
+    },
+    {
+      "ids": [
+        "paper:doi:10.1145/3597926.3598052",
+        "paper:doi:10.1145/3468264.3468573"
+      ],
+      "reason": "Two papers by one group on Datalog engines: the 2021 paper introduced metamorphic testing for them, and the 2023 paper builds on it using precedence information among relations, describing the earlier work as what it improves upon."
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2105_10016.json b/_data/impact/paper_notes/paper_arxiv_2105_10016.json
new file mode 100644
index 0000000..c8a2ebe
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2105_10016.json
@@ -0,0 +1,132 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2105.10016",
+    "title": "Testing DBMS Performance with Mutations",
+    "authors": [
+      "Xinyu Liu",
+      "Qi Zhou",
+      "Joy Arulraj",
+      "A. Orso"
+    ],
+    "year": 2021,
+    "venue": "arXiv (Cornell University)",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2105.10016"
+  },
+  "summary": {
+    "text": "AMOEBA detects performance bugs in database systems, a class the authors note has received far less attention than functional bugs despite affecting response time and end-user experience. It constructs pairs of semantically equivalent queries and compares their response times on the same system; a significant difference points to a performance bug. The paper contributes structure and predicate mutation rules for building such pairs and feedback mechanisms to improve efficacy and efficiency. On PostgreSQL and CockroachDB it found 20 previously unknown performance bugs, 14 confirmed.",
+    "relationship_to_sqlancer": "Applies equivalent-query construction, SQLancer's oracle idea, to performance rather than correctness, and compares against that work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2105.10016",
+    "source_sha256": "sha256:fc962449da18fbc7da21632ea0c574ce72282025e6df2756dece35c141010ef4",
+    "supporting_excerpts": [
+      "Despite their impact on end-user experience, performance bugs have received considerably less attention than functional bugs.",
+      "In this paper, we present AMOEBA, a system for automatically detecting performance bugs in database systems.",
+      "The core idea behind AMOEBA is to construct query pairs that are semantically equivalent to each other and then compare their response time on the same database system."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:53Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2105.10016",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [35–37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2105.10016",
+          "source_type": "paper_citation_context",
+          "excerpt": "They leverage tools such as S QLSMITH [4] and SQLancer [35– 37] to discover crash-inducing or logic bugs in DBMSs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2105.10016",
+          "source_type": "paper_citation_context",
+          "excerpt": "8.6.1 Query Pairs from SQLancer SQLancer is the state-of-art tool for discovering logic bugs [35–37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2105.10016",
+          "source_type": "paper",
+          "excerpt": "We defer a comparative analysis of AMOEBA against SQLancer to §8.6.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2105.10016",
+          "source_type": "paper",
+          "excerpt": "We provide the total number of bugs discovered by A MOEBA in the fourth row for comparison. a wide range of SQL operators, and (2) they use the same set of transformation rules as AMOEBA does that are effective at discovering performance bugs (§8.3). 8.6.3 Results As shown in Table 4, we compare AMOEBA against three baselines: (1) We use SQLancer to randomly generate 2000 pairs of equivalent queries.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2105.10016",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [35–37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Pivoted Query Synthesis (PQS) as state of the art.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2105.10016",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [35–37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Non-optimizing Reference Engine Construction (NoREC) as state of the art.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2206_08530.json b/_data/impact/paper_notes/paper_arxiv_2206_08530.json
new file mode 100644
index 0000000..66cc765
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2206_08530.json
@@ -0,0 +1,139 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2206.08530",
+    "title": "GDsmith: Detecting Bugs in Graph Database Engines",
+    "authors": [
+      "Weisheng Lin",
+      "Ziyue Hua",
+      "Luyao Ren",
+      "Z. Li",
+      "Lu Zhang",
+      "Tao Xie"
+    ],
+    "year": 2022,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2206.08530"
+  },
+  "summary": {
+    "text": "The preprint of GDsmith, a black-box approach for testing graph database engines. Random test generation, the most practical route to automation, faces three obstacles here — semantic validity, non-empty results, and behaviour diversity. GDsmith ensures each generated Cypher query is semantically valid through skeleton generation and completion, raises the chance of non-empty results with three kinds of structural mutation, and improves behavioural diversity by selecting property keys according to their previous frequencies. It detected 27 previously unknown bugs across three popular open-source graph engines.",
+    "relationship_to_sqlancer": "Its artifact reuses the SQLancer codebase, carrying random query generation for bug detection to graph engines.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2206.08530",
+    "source_sha256": "sha256:e9550e6c8d66a5a8e89de65a27a31adf725aa63e2acb32b2db37735d31f1d7ac",
+    "supporting_excerpts": [
+      "However, random testing, the most practical way of automated test generation, faces the challenges of semantic validity, non-empty result, and behavior diversity to detect bugs in graph database engines.",
+      "To address these challenges, in this paper, we propose GDsmith, the first black-box approach for testing graph database engines.",
+      "GDsmith successfully detects 27 previously unknown bugs on the released versions of three popular open-source graph database engines and receive positive feedback from their developers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:38:27Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2206.08530",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoREC [24] detects bugs in relational database engines by applying a semantics-preserving transformation to a given SQL query to disable the engine’s optimizations and addresses PQS’ high implementation effort.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2206.08530",
+          "source_type": "paper_citation_context",
+          "excerpt": "PQS [26] detects wrong-result bugs by checking whether a specific record is fetched correctly.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2206.08530",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [25] derives multiple SQL queries that compute a partial result of the initial query.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2206.08530",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [25] derives multiple SQL queries that",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/ddaa2000/GDsmith",
+          "source_type": "github_repository",
+          "excerpt": "# GDsmith",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after GDsmith, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/ddaa2000/GDsmith/blob/master/src/main/java/org/example/gdsmith/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.ALPHANUMERIC;",
+          "excerpt_is_verbatim": true,
+          "note": "src/main/java/org/example/gdsmith/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.example.gdsmith (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/ddaa2000/GDsmith",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "renamed_sqlancer_package",
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/ddaa2000/GDsmith",
+          "source_type": "github_repository",
+          "excerpt": "# GDsmith",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after GDsmith, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/ddaa2000/GDsmith/blob/master/src/main/java/org/example/gdsmith/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.ALPHANUMERIC;",
+          "excerpt_is_verbatim": true,
+          "note": "src/main/java/org/example/gdsmith/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.example.gdsmith (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2304_10044.json b/_data/impact/paper_notes/paper_arxiv_2304_10044.json
new file mode 100644
index 0000000..93a258b
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2304_10044.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2304.10044",
+    "title": "Finding Bug-Inducing Program Environments",
+    "authors": [
+      "Z. Mirzamomen",
+      "Marcel Böhme"
+    ],
+    "year": 2023,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2304.10044"
+  },
+  "summary": {
+    "text": "This paper synthesises bug-inducing program environments — the databases, files and devices a program accesses — for bugs that no program input alone can expose. The approach is coverage-guided and mutation-based: it intercepts the system calls through which a program interacts with its environment to capture the resources touched on a first execution as an initial environment, then mutates them, keeping any environment that increases coverage as a seed for further fuzzing. The prototype, AFLChaos, found bugs in the resource-handling code of five of seven open-source projects including OpenSSL.",
+    "relationship_to_sqlancer": "Fuzzes the environment rather than the input, including databases used for storage; SQLancer is cited among testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2304.10044",
+    "source_sha256": "sha256:82ca15780a542cc8529516989169f89ebc037c9d64cddb3705ac79929a14994d",
+    "supporting_excerpts": [
+      "Some bugs cannot be exposed by program inputs, but only by certain program environments.",
+      "In this paper, we present a coverage-guided, mutation-based environment synthesis approach of bug-inducing program environments.",
+      "We implemented a prototype called AFLChaos which found bugs in the resource-handling code of five (5) of the seven (7) open source projects in our benchmark set (incl."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2304.10044",
+          "source_type": "paper_citation_context",
+          "excerpt": "Our experimental results on seven widely used protocol implementations showed outstanding increase in code coverage for AFLChaos in comparison with AFLNet as the base and resulted in discovering 13 bugs in these widely-used and well-fuzzed programs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2310_06433.json b/_data/impact/paper_notes/paper_arxiv_2310_06433.json
new file mode 100644
index 0000000..d99242c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2310_06433.json
@@ -0,0 +1,91 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2310.06433",
+    "title": "Retromorphic Testing: A New Approach to the Test Oracle Problem",
+    "authors": [
+      "Boxi Yu",
+      "Qiuyang Mang",
+      "Qingshuo Guo",
+      "Pinjia He"
+    ],
+    "year": 2023,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2310.06433"
+  },
+  "summary": {
+    "text": "Retromorphic Testing proposes a black-box test oracle inspired by inverse functions. It pairs the program under test with an auxiliary program to form a forward and a backward program: input data is processed by the forward program and its output converted back into the original input format by the backward program, after which the relation between the initial input and the transformed output is checked in the input domain. Either program can play either role, giving different testing modes, which the paper illustrates across algorithms, traditional software and AI applications.",
+    "relationship_to_sqlancer": "Proposes a new oracle alongside differential and metamorphic testing, the family SQLancer's oracles belong to.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2310.06433",
+    "source_sha256": "sha256:e68319b0bca30b91f04e0608fb3af2875c55a2be84ac341c8fd4d2ebbd74cbc3",
+    "supporting_excerpts": [
+      "In automated testing, black-box techniques, known for their non-intrusive nature in test oracle construction, are widely used, including notable methodologies like differential testing and metamorphic testing.",
+      "Inspired by the mathematical concept of inverse function, we present Retromorphic Testing, a novel black-box testing methodology.",
+      "It leverages an auxiliary program in conjunction with the program under test, which establishes a dual-program structure consisting of a forward program and a backward program."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2310.06433",
+          "source_type": "paper_citation_context",
+          "excerpt": "Metamorphic testing is also used for testing various systems such as compilers [14, 15], database engines [13, 19, 20], SMT solvers [34], Android apps [28, 29], quantum computing platforms [1, 17], and AI systems [30, 32, 33, 36, 37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2310.06433",
+          "source_type": "paper_citation_context",
+          "excerpt": "In the context of Database Management Systems (DBMS), Rigger and Su introduced the concept of Pivoted Query Synthesis (PQS) [21] as an effective testing approach to uncover logic bugs in these systems.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2310.06433",
+          "source_type": "paper_citation_context",
+          "excerpt": "Although the original paper describes PQS as a testing technique for a specific system, we think it can be regarded as an instance of the general Retromorphic Testing methodology.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2310.06433",
+          "source_type": "paper_citation_context",
+          "excerpt": "The PQS testing approach initiates with the generation of a pivot row ( e.g., [t0.c0: 3, t0.c1: TRUE, t1.c0: -5] in Fig.2), which is in the modality of rows.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2310.06433",
+          "source_type": "paper_citation_context",
+          "excerpt": "We also discussed with one author of PQS and he agrees that PQS should be categorized as a Retromorphic Testing technique.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2503_03893.json b/_data/impact/paper_notes/paper_arxiv_2503_03893.json
new file mode 100644
index 0000000..4d636cc
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2503_03893.json
@@ -0,0 +1,98 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2503.03893",
+    "title": "Parser Knows Best: Testing DBMS with Coverage-Guided Grammar-Rule Traversal",
+    "authors": [
+      "Yu Liang",
+      "Hong Hu"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2503.03893"
+  },
+  "summary": {
+    "text": "ParserFuzz generates SQL by extracting grammar rules directly from a DBMS's own built-in syntax definition files. The motivation is that existing tools exercise only a small subset of the syntax elements in DBMS-specific SQL dialects, leaving many features untested. Without any input corpus, ParserFuzz saturates the grammar features of the system under test, and uses code coverage as feedback to guide mutation and combine features drawn from different rules. It found 81 previously unknown bugs across five popular DBMSs, all confirmed and 34 fixed.",
+    "relationship_to_sqlancer": "Argues that existing DBMS testing tools, SQLancer among those cited, cover only a small part of a dialect's syntax, and reports outperforming them on bug finding and coverage.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2503.03893",
+    "source_sha256": "sha256:d33af64c3d397209aac795620862004b6abe910d989f94156be1870a353db5e2",
+    "supporting_excerpts": [
+      "However, these tools only cover a small subset of diverse syntax elements defined in DBMS-specific SQL dialects, leaving a large number of features unexplored.",
+      "In this paper, we propose ParserFuzz, a novel fuzzing framework that automatically extracts grammar rules from DBMSs' built-in syntax definition files for SQL query generation.",
+      "In our evaluation, ParserFuzz outperforms all state-of-the-art existing DBMS testing tools in terms of bug finding, grammar rule coverage and code coverage."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2503.03893",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer +QPG supports testing with SQLite , CockroachDB and TiDB , and outperforms all other logic bug detectors including SQLRight [2, 21].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2503.03893",
+          "source_type": "paper_citation_context",
+          "excerpt": "Because NoREC oracle is claimed to be a better performer over-all compared to TLP oracle [2].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2503.03893",
+          "source_type": "paper_citation_context",
+          "excerpt": "With its latest configuration SQLancer +QPG [2], it uses the DBMS query plan to guide its query generation in order to stress test the DBMS query optimization logic.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2503.03893",
+          "source_type": "paper_citation_context",
+          "excerpt": "Recent efforts on DBMS testing [2, 14, 47, 67] can be classified into two categories: generation-based testing and mutation-based grey-box fuzzing.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2503.03893",
+          "source_type": "paper_citation_context",
+          "excerpt": "There is another generation-based DBMS testing tool called SQLancer [24], that focuses on detecting DBMS logic errors from DBMS systems [32–34].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2503.03893",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer introduces a few SQL oracles for this purpose such as NoREC , TLP and PQS , where each shares a distinct SQL pattern to match [32–34].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2503_17322.json b/_data/impact/paper_notes/paper_arxiv_2503_17322.json
new file mode 100644
index 0000000..5524f95
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2503_17322.json
@@ -0,0 +1,55 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2503.17322",
+    "title": "QITE: Assembly-Level, Cross-Platform Testing of Quantum Computing Platforms",
+    "authors": [
+      "Matteo Paltenghi",
+      "Michael Pradel"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2503.17322"
+  },
+  "summary": {
+    "text": "QITE tests quantum computing platforms across platform boundaries using QASM, an assembly-level representation, as the common ground. Its ITE process generates equivalent quantum programs by importing assembly into a platform's representation, transforming it through that platform's optimisation and gate conversion, and exporting it back to assembly. A crash oracle catches failures during these transformations and an equivalence oracle checks that the resulting assembly programs, equivalent by construction, really are. Across Qiskit, PennyLane, Pytket and BQSKit it revealed 17 bugs.",
+    "relationship_to_sqlancer": "Builds an equivalence oracle by round-tripping programs through transformations, the same oracle pattern SQLancer uses for queries.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2503.17322",
+    "source_sha256": "sha256:aa9892d5cbf10e35416943a813cd476a81d8c864670ee9e686f3a95a160414b3",
+    "supporting_excerpts": [
+      "To address these challenges, we present QITE, the first cross-platform testing framework for quantum computing platforms, which leverages QASM, an assembly-level representation, to ensure consistency across different platforms.",
+      "QITE introduces the novel ITE process to generate equivalent quantum programs by iteratively (I)mporting assembly into platform representations, (T)ransforming via platform optimization and gate conversion, and (E)xporting back to assembly.",
+      "It uses a crash oracle to detect failures during cross-platform transformations and an equivalence oracle to validate the semantic consistency of the final sets of assembly programs, which are expected to be equivalent by construction."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2503.17322",
+          "source_type": "paper_citation_context",
+          "excerpt": "Grammar-based generators, similar to our approach, have been successfully applied to test different compilers and similar software, such as the Java Virtual Machine [40], C compilers [13, 50], SMT solvers [47], database engines [38], and deep learning software infrastructure [23, 26, 27, 45].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2506_02617.json b/_data/impact/paper_notes/paper_arxiv_2506_02617.json
new file mode 100644
index 0000000..e0b95d8
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2506_02617.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2506.02617",
+    "title": "Toward Understanding Bugs in Vector Database Management Systems",
+    "authors": [
+      "Yinglin Xie",
+      "Xinyi Hou",
+      "Yanjie Zhao",
+      "Shenao Wang",
+      "Kai Chen",
+      "Haoyu Wang"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2506.02617"
+  },
+  "summary": {
+    "text": "The first large-scale empirical study of defects in vector database management systems, which the authors argue traditional database reliability models do not fit because data representation, query mechanisms and architecture all differ. They manually analysed 1,671 bug-fix pull requests across 15 open-source vector databases and built a taxonomy by symptom, root cause and fix strategy. They report five symptom categories with more than half being functional failures, 31 recurring fault patterns including failure modes unique to vector search, and 12 common fix strategies.",
+    "relationship_to_sqlancer": "An empirical bug study for a database class SQLancer does not cover, citing the DBMS reliability work it builds on.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2506.02617",
+    "source_sha256": "sha256:72bfd057e94dd1f50fad5b7fd2e123061ee4908e66c9568b10db0f2fd282275a",
+    "supporting_excerpts": [
+      "Traditional database reliability models cannot be directly applied to VDBMSs because of fundamental differences in data representation, query mechanisms, and system architecture.",
+      "To address this gap, we present the first large-scale empirical study of software defects in VDBMSs.",
+      "Our study identifies five categories of bug symptoms, with more than half manifesting as functional failures."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:32:20Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2506.02617",
+          "source_type": "paper_citation_context",
+          "excerpt": "Rigger et al. [29] further introduced the Non-Optimizing Reference Engine Construction (NoREC) method to detect optimization bugs in query engines.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2509_10819.json b/_data/impact/paper_notes/paper_arxiv_2509_10819.json
new file mode 100644
index 0000000..162845d
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2509_10819.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2509.10819",
+    "title": "Arguzz: Testing zk VMs for Soundness and Completeness Bugs",
+    "authors": [
+      "Christoph Hochrainer",
+      "Valentin Wüstholz",
+      "Maria Christakis"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2509.10819"
+  },
+  "summary": {
+    "text": "Arguzz tests zero-knowledge virtual machines for soundness bugs, where an invalid execution is accepted, and completeness bugs, where a valid one is rejected. It combines a variant of metamorphic testing with fault injection: it generates semantically equivalent program pairs, merges them into a single Rust program with a known output, runs it inside the zk VM, and injects faults to imitate a malicious or buggy prover and so expose overly weak constraints. Across six real zk VMs it found eleven bugs in three, one of which earned a $50,000 bounty despite prior audits.",
+    "relationship_to_sqlancer": "Applies equivalence-based metamorphic testing to zk VMs rather than DBMSs; SQLancer is cited as prior work using that oracle style.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2509.10819",
+    "source_sha256": "sha256:0d463510f8be93090968c06da163bae26514d77104a20751a4f680fdf9a20fea",
+    "supporting_excerpts": [
+      "We present Arguzz, the first automated tool for testing zkVMs for soundness and completeness bugs.",
+      "To detect such bugs, Arguzz combines a novel variant of metamorphic testing with fault injection.",
+      "We used Arguzz to test six real-world zkVMs (RISC Zero, Nexus, Jolt, SP1, OpenVM, and Pico) and found eleven bugs in three of them."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2509.10819",
+          "source_type": "paper_citation_context",
+          "excerpt": "Blackbox fuzzing remains a widely used and effective strategy when testing complex systems, e.g., compilers [47], database systems [39], or SMT solvers [34].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2510_06663.json b/_data/impact/paper_notes/paper_arxiv_2510_06663.json
new file mode 100644
index 0000000..61511bb
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2510_06663.json
@@ -0,0 +1,141 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2510.06663",
+    "title": "Automated Discovery of Test Oracles for Database Management Systems Using LLMs",
+    "authors": [
+      "Qiuyang Mang",
+      "Runyuan He",
+      "Suyang Zhong",
+      "Xiaoxuan Liu",
+      "Huanchen Zhang",
+      "Alvin Cheung"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2510.06663"
+  },
+  "summary": {
+    "text": "Argus uses large language models to automate what has been the manual part of DBMS test oracle design: inventing mechanisms that generate equivalent query pairs. To keep model creativity from producing false positives, it builds on the Constrained Abstract Query — a SQL skeleton with placeholders and conditions on how they may be filled. The model proposes pairs of skeletons, their equivalence is formally proven with a SQL equivalence solver, and only then are the placeholders instantiated with synthesised SQL snippets. On five extensively tested DBMSs it found 41 previously unknown bugs, 36 of them logic bugs.",
+    "relationship_to_sqlancer": "Automates the design of oracles of the kind SQLancer introduced, and its artifact reuses the SQLancer codebase.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2510.06663",
+    "source_sha256": "sha256:11d0434b4c04a54c5478538005a7206504ea321b9be47039c6de2965adb729a2",
+    "supporting_excerpts": [
+      "A cornerstone of these techniques is\n test oracle\n , which typically implements a mechanism to generate equivalent query pairs, and subsequently runs the pair and identifies bugs by checking the consistency of their results.",
+      "While running these oracles can be automated, designing the mechanism to generate equivalent queries remains a fundamentally manual endeavor.",
+      "We have implemented Argus and evaluated it on five extensively tested DBMSs, discovering 41 previously unknown bugs, 36 of which are logic bugs, with 36 confirmed and 27 already fixed by the developers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2510.06663",
+          "source_type": "paper_citation_context",
+          "excerpt": "Listing 2: An example of representing and instantiating TLP [45] oracle in CAQ.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/joyemang33/Argus",
+          "source_type": "github_repository",
+          "excerpt": "# Argus: Automated Discovery of Test Oracles for Database Management Systems Using LLMs\n\nArgus is a novel framework for automatically discovering and instantiating test oracles to find logic bugs in Database Management Systems (DBMSs) using Large Language Models (LLMs).",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/joyemang33/Argus/blob/main/MetamorphicCoverageArtifact/Code/mc-guided_fuzzing/src/sqlancer/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "MetamorphicCoverageArtifact/Code/mc-guided_fuzzing/src/sqlancer/Randomly.java is SQLancer's Randomly.java (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2510.06663",
+          "source_type": "paper",
+          "excerpt": "LLM-powered SQL snippet generation.We evaluate the effectiveness of LLM-powered SQL snippet generation by comparingArgus’s method against a variant that only uses SQLancer’s grammarbased generator to build the snippet corpus during the CAQ instantiation phase.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/joyemang33/Argus",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/joyemang33/Argus",
+          "source_type": "github_repository",
+          "excerpt": "# Argus: Automated Discovery of Test Oracles for Database Management Systems Using LLMs\n\nArgus is a novel framework for automatically discovering and instantiating test oracles to find logic bugs in Database Management Systems (DBMSs) using Large Language Models (LLMs).",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/joyemang33/Argus/blob/main/MetamorphicCoverageArtifact/Code/mc-guided_fuzzing/src/sqlancer/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "MetamorphicCoverageArtifact/Code/mc-guided_fuzzing/src/sqlancer/Randomly.java is SQLancer's Randomly.java (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2511_17377.json b/_data/impact/paper_notes/paper_arxiv_2511_17377.json
new file mode 100644
index 0000000..07661c8
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2511_17377.json
@@ -0,0 +1,109 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2511.17377",
+    "title": "Anomaly Pattern-guided Transaction Bug Testing in Relational Databases",
+    "authors": [
+      "Hui-Rong Xu",
+      "Shuang Liu",
+      "Xianyu Zhu",
+      "Qiyu Zhuang",
+      "Wei Lu",
+      "Xiaoyong Du"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2511.17377"
+  },
+  "summary": {
+    "text": "APTrans tests transaction handling in relational DBMSs under different isolation levels. The authors identify two obstacles: generating transactions that actually expose bugs, since those bugs need specific transactional constraints, and deciding whether an outcome is wrong, since the correct result of randomly generated transactions is usually unknown. They address the first with generation guided by predefined anomaly patterns and the second with a two-phase explicit and implicit error detection process. On MySQL, MariaDB and OceanBase, APTrans found 13 previously unknown transaction bugs, 11 confirmed.",
+    "relationship_to_sqlancer": "Extends automated DBMS testing from single queries to transactions, and the record's artifact evidence shows the implementation reusing the SQLancer codebase.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2511.17377",
+    "source_sha256": "sha256:427c195fcd255330c2fe1447f7b4e526cf894e95dbc62ee840e0dc4134f505a2",
+    "supporting_excerpts": [
+      "To address these challenges, we propose an anomaly pattern-guided testing approach for uncovering transaction bugs in RDBMSs.",
+      "Our solution tackles the first challenge by introducing a test case generation technique guided by predefined anomaly patterns, which increases the likelihood of exposing transactional bugs.",
+      "APTrans successfully identified 13 previously unknown transaction-related bugs, 11 of which have been confirmed by the respective development teams."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2511.17377",
+          "source_type": "paper_citation_context",
+          "excerpt": "Recent innovations have introduced paradigm shifts in testing strategies: QPG [7] uses query plan-guided database state mutation,while DQE[28]appliesdifferentialexecutionanalysisacross SELECT,UPDATE,and DELETEstatements.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2511.17377",
+          "source_type": "paper_citation_context",
+          "excerpt": "Its successor, SQLancer [25–27], introduces three metamorphic testing oracles: PQS [27], NoREC [25], and TLP [26], significantly improving its ability to detect logical bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2511.17377",
+          "source_type": "paper",
+          "excerpt": "We develop our database and SQL statement generation method based on SQLancer [ 4] and extend it to support table join operations for both database generation and SQL statement generation. 3.2.1 Database Generation.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that its implementation is built on SQLancer.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "extends_technique",
+      "title": "Extends a SQLancer technique",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2511.17377",
+          "source_type": "paper",
+          "excerpt": "To simulate real-world scenarios, which usually involve multiple tables connected with join keys, we extend SQLancer to support table join operations for both database generation and SQL statement generation.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it extends or adapts a SQLancer technique.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2601_15074.json b/_data/impact/paper_notes/paper_arxiv_2601_15074.json
new file mode 100644
index 0000000..cfb083b
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2601_15074.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2601.15074",
+    "title": "SmartOracle - An Agentic Approach to Mitigate Noise in Differential Oracles",
+    "authors": [
+      "Srinath Srinivasan",
+      "Tim Menzies",
+      "Marcelo d’Amorim"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2601.15074"
+  },
+  "summary": {
+    "text": "SmartOracle addresses the cost of validating differential fuzzing results for JavaScript engines, where oracles are built by hand, are expensive and false-positive prone, and must be redone as the specification evolves. It decomposes the manual triage workflow into specialised LLM sub-agents that gather evidence from terminal runs and targeted specification queries and combine it into a verdict. On historical benchmarks it reaches 0.84 recall at an 18% false positive rate, and in live campaigns it found specification-level issues in V8, JavaScriptCore and GraalJS.",
+    "relationship_to_sqlancer": "Differential-testing triage for JavaScript engines rather than DBMSs; SQLancer is cited as related differential-testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2601.15074",
+    "source_sha256": "sha256:5eb6adee56199250627e41eb50bb68c004b29be62760c9718805e532b27a113e",
+    "supporting_excerpts": [
+      "Inspired by the success of agentic systems in other SE domains, this paper introduces SmartOracle.",
+      "SmartOracle decomposes the manual triage workflow into specialized Large Language Model (LLM) sub-agents.",
+      "For historical benchmarks, SmartOracle achieves 0.84 recall with an 18% false positive rate."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2601.15074",
+          "source_type": "paper_citation_context",
+          "excerpt": "Research in this domain demonstrates that by endowing agents with persistent memory, explicit objectives, and access to external tools or APIs, these systems can manage significantly more complex and dynamic workflows than prompt chaining alone [19, 54].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2602_19490.json b/_data/impact/paper_notes/paper_arxiv_2602_19490.json
new file mode 100644
index 0000000..71afa16
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2602_19490.json
@@ -0,0 +1,148 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2602.19490",
+    "title": "FuzzySQL: Uncovering Hidden Vulnerabilities in DBMS Special Features with LLM-Driven Fuzzing",
+    "authors": [
+      "Yongxin Chen",
+      "Zhiyuan Jiang",
+      "Chao Zhang",
+      "Haoran Xu",
+      "Shenglin Xu",
+      "Jianping Tang",
+      "Zheming Li",
+      "Peidai Xie",
+      "Yongjun Wang"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2602.19490"
+  },
+  "summary": {
+    "text": "FuzzySQL is an LLM-driven fuzzing framework aimed at DBMS features that ordinary fuzzers rarely reach, such as system-level modes, procedural constructs and other obscure functionality. It pairs grammar-guided generation with a logic-shifting progressive mutation that negates conditions and restructures execution logic to explore alternative control paths, and repairs failed statements with a hybrid of rule-based patching and LLM-driven semantic repair. Across MySQL, MariaDB, SQLite, PostgreSQL and ClickHouse it uncovered 64 vulnerabilities, 60 confirmed and 9 assigned CVEs.",
+    "relationship_to_sqlancer": "Compares against conventional DBMS fuzzers and argues their semantic feature coverage is limited, placing SQLancer's line of work as the state of the art it measures against.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2602.19490",
+    "source_sha256": "sha256:f422fe00ab6dd6cecf0823dd72579528151a4d7aa1449d06da27787bd4871c0e",
+    "supporting_excerpts": [
+      "In this paper, we present FuzzySQL, a novel LLM-powered adaptive fuzzing framework designed to uncover subtle vulnerabilities in DBMS special features.",
+      "We evaluate FuzzySQL across multiple DBMSs, including MySQL, MariaDB, SQLite, PostgreSQL and Clickhouse, uncovering 64 vulnerabilities, 27 of which are tied to under-tested DBMS special features.",
+      "Our results highlight the limitations of conventional fuzzers in semantic feature coverage and demonstrate the potential of LLM-based fuzzing to discover deeply hidden bugs in complex database systems."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2602.19490",
+          "source_type": "paper_citation_context",
+          "excerpt": "To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art open-source fuzzing baselines: Squir-rel [56], EET [19], and SQLancer [35].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2602.19490",
+          "source_type": "paper_citation_context",
+          "excerpt": "Depending on the bug types targeted—ranging from crash vulnerabilities [1, 12, 13, 18, 22, 27, 45, 56] to logic inconsistencies [4, 19, 23, 35–37, Generation-based approaches rely on hand-crafted grammars or learned rules to synthesize SQL sequences from scratch.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2602.19490",
+          "source_type": "paper_citation_context",
+          "excerpt": "Existing DBMS fuzzers have explored SQL generation and mutation from multiple perspectives, including structural complexity [18, 56], sequence-level interactions [13, 22], and oracle-guided logic testing [19, 23, 35].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2602.19490",
+          "source_type": "paper_citation_context",
+          "excerpt": "However, most existing DBMS fuzzers [1, 19, 23, 37, 56] predominantly focus on general-purpose statement structures (e.g., SELECT ), covering only a limited portion of the DBMS grammar.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2602.19490",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer [35–37 , 52] and EET Mutation-based fuzzers, in contrast, transform existing SQL statements through changes at the syntax or intermediate representation (IR) level.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2602.19490",
+          "source_type": "paper_citation_context",
+          "excerpt": "For SQLancer, we use TLP [36] as the test oracle for MySQL, and NoREC [35] for the other DBMS targets.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2602.19490",
+          "source_type": "paper",
+          "excerpt": "To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art opensource fuzzing baselines: Squirrel [ 56], EET [ 19], and SQLancer [ 35].",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2602.19490",
+          "source_type": "paper_citation_context",
+          "excerpt": "To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art open-source fuzzing baselines: Squir-rel [56], EET [19], and SQLancer [35].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Non-optimizing Reference Engine Construction (NoREC) as state of the art.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2603_00311.json b/_data/impact/paper_notes/paper_arxiv_2603_00311.json
new file mode 100644
index 0000000..c18b70a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2603_00311.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2603.00311",
+    "title": "Towards the Systematic Testing of Regular Expression Engines",
+    "authors": [
+      "Berk Çakar",
+      "Dongyoon Lee",
+      "James C. Davis"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2603.00311"
+  },
+  "summary": {
+    "text": "ReTest is a framework for testing regular expression engines. The authors argue that differential testing across implementations is unreliable because regex syntax and semantics differ substantially between dialects, and that naive byte-level fuzzing produces invalid inputs exercising only the parser. ReTest combines grammar-aware fuzzing for coverage with metamorphic testing to obtain dialect-independent oracles. So far they have surveyed 22 engines, analysed 1,007 bugs and 156 CVEs, and curated 16 metamorphic relations derived from Kleene algebra; on PCRE it reaches 3x the edge coverage of existing fuzzing.",
+    "relationship_to_sqlancer": "Applies metamorphic oracles outside the database setting, citing SQLancer's line of work on oracles that avoid needing a reference implementation.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2603.00311",
+    "source_sha256": "sha256:5be430ebf3e2813056716de8d7e3152b1cb7b940a5e23791a7e9f5297ca69d15",
+    "supporting_excerpts": [
+      "Fuzzing is also utilized to ease testing of feature-rich regex implementations to expose defects, but naive byte-level mutations generate syntactically invalid inputs that exercise only parsing logic, not matching internals.",
+      "In this work, we describe our progress towards ReTest, a framework that systematically tests regular expression engines by combining grammar-aware fuzzing for high code coverage with metamorphic testing to generate dialect-independent test oracles.",
+      "Our preliminary evaluation on PCRE shows that ReTest achieves 3x higher edge coverage than existing fuzzing approaches and has identified three new memory safety defects."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2603.00311",
+          "source_type": "paper_citation_context",
+          "excerpt": "For DBMS, Rigger and Su [48, 49, 50] used metamorphic oracles to detect logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2603_19434.json b/_data/impact/paper_notes/paper_arxiv_2603_19434.json
new file mode 100644
index 0000000..4695cf5
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2603_19434.json
@@ -0,0 +1,53 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2603.19434",
+    "title": "Computer-Orchestrated Design of Algorithms: From Join Specification to Implementation",
+    "authors": [
+      "Zeyuan Hu"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2603.19434"
+  },
+  "summary": {
+    "text": "An experience paper on CODA, a testing framework used while co-designing the logical specification and physical implementation of TreeTracker Join. The authors argue that existing database testing frameworks cannot supply the algorithm-specific inputs such as join trees that this needs, and that macro-benchmark queries are too noisy to isolate the defects involved. By synthesising minimal reproducible examples, CODA isolated translation defects such as state mismanagement and mapping conflicts, and one boundary condition it exposed refined the algorithm's formal precondition.",
+    "relationship_to_sqlancer": "Positions itself against existing database testing frameworks, which is where SQLancer is cited; the framework itself is separate.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2603.19434",
+    "source_sha256": "sha256:58bdee8bd69b4381a4b3832dff74f6d8f6e826b510993f2abd18a84dfe8fa2ca",
+    "supporting_excerpts": [
+      "In this experience paper, we present a retrospective analysis of $\\mathsf{CODA}$, a computer-orchestrated testing framework utilized during the logical-physical co-design of TreeTracker Join ($\\mathsf{TTJ}$), a theoretically optimal yet practical join algorithm recently published in ACM TODS.",
+      "By synthesizing minimal reproducible examples, $\\mathsf{CODA}$ successfully isolates subtle translation defects, such as state mismanagement and mapping conflicts between join trees and bushy plans."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/7fa57423ffb12f9cd6dace5f26042a6289837bf4",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2603_21530.json b/_data/impact/paper_notes/paper_arxiv_2603_21530.json
new file mode 100644
index 0000000..c16dc34
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2603_21530.json
@@ -0,0 +1,67 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2603.21530",
+    "title": "LLM-Based Test Case Generation in DBMS through Monte Carlo Tree Search",
+    "authors": [
+      "Yujia Chen",
+      "Yingli Zhou",
+      "Fangyuan Zhang",
+      "Cuiyun Gao"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2603.21530"
+  },
+  "summary": {
+    "text": "MIST generates SQL test cases for DBMSs with lightweight LLMs guided by Monte Carlo tree search. It addresses two industrial constraints: organisations often must use small models for security and privacy reasons, and those models struggle with proprietary SQL dialects and tend to produce semantically similar queries that plateau in coverage. MIST builds a hierarchical feature tree and uses error feedback to steer generation. Across three DBMSs and four lightweight models it improved line coverage by 43.3%, function coverage by 32.3% and branch coverage by 46.4% over the strongest baseline.",
+    "relationship_to_sqlancer": "Coverage-driven SQL generation compared against existing automated DBMS testing approaches, among which SQLancer is cited.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2603.21530",
+    "source_sha256": "sha256:c9ad5b676ec668679d17a488d73e449a2c75e34daff2c567dc4c7fd21b579d91",
+    "supporting_excerpts": [
+      "Second, LLM-generated queries are often semantically similar and exercise only shallow execution paths, thereby quickly reaching a coverage plateau.",
+      "To address these challenges, we propose MIST, an LLM-based test case generatIon framework for DBMS through Monte Carlo Tree search.",
+      "Experiments on three widely-used DBMSs with four lightweight LLMs show that MIST achieves average improvements of 43.3% in line coverage, 32.3% in function coverage, and 46.4% in branch coverage compared to the baseline approach with the highest line coverage of 69.3% in the Optimizer module."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2603.21530",
+          "source_type": "paper_citation_context",
+          "excerpt": "Traditional approaches for generating test cases, such as BuzzBee [46], SQLsmith [34], SQLancer [32], can be effective for a specific DBMS; however, there exist many DBMSs designed for different application scenarios, such as DuckDB [29] for analytical workloads, PostgreSQL [36] for general-purpose…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2603.21530",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer [30, 31, 33] employs hand-crafted dialect-specific generators with sophisticated test oracles to detect logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2604_01442.json b/_data/impact/paper_notes/paper_arxiv_2604_01442.json
new file mode 100644
index 0000000..8c63ac9
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2604_01442.json
@@ -0,0 +1,54 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2604.01442",
+    "title": "Fuzzing with Agents? Generators Are All You Need",
+    "authors": [
+      "Vasudev Vikram",
+      "Rohan Padhye"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2604.01442"
+  },
+  "summary": {
+    "text": "This paper asks whether an AI coding agent can synthesise the domain-specific input generators that generator-based fuzzing relies on, and whether such generators are good enough to make coverage guidance and mutation unnecessary. Gentoo gives an LLM agent terminal access and the source of the fuzz target and library, and has it iteratively synthesise and refine a generator. Across seven real-world Java libraries, agent-synthesised generators beat human-written ones on branch coverage in four, and unlike the human-written ones did not benefit significantly from coverage guidance and mutation.",
+    "relationship_to_sqlancer": "Generator-based fuzzing outside the DBMS setting; SQLancer is cited as background rather than used.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2604.01442",
+    "source_sha256": "sha256:b6fd235d75f70a3da1539e959a23997df1d8a1907c8072817ec42d321655dafa",
+    "supporting_excerpts": [
+      "We investigate whether AI coding agents can automatically synthesize such target-specific generators, and whether the resulting generators are strong enough to obviate the need for coverage guidance and mutation entirely.",
+      "Our findings show that agent-synthesized generators achieve statistically significantly higher branch coverage than human-written baseline generators on 4 of 7 benchmarks."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2604.01442",
+          "source_type": "paper_citation_context",
+          "excerpt": "Tools such as CSmith [52, 65] for C compilers and SQLancer [8] for database engines generate inputs that are not merely syntactically valid but semantically rich by construction: CSmith emits C programs free of undefined behavior, while SQLancer produces queries that satisfy the relational schemas…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2604_03024.json b/_data/impact/paper_notes/paper_arxiv_2604_03024.json
new file mode 100644
index 0000000..de82df1
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2604_03024.json
@@ -0,0 +1,62 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2604.03024",
+    "title": "BugForge: Constructing and Utilizing DBMS Bug Repository to Enhance DBMS Testing",
+    "authors": [
+      "Dawei Li",
+      "Qifan Liu",
+      "Yuxiao Guo",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Chi Zhang",
+      "Jingzhou Fu",
+      "H. Mao",
+      "Zhenyu Guan",
+      "Yu Jiang"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2604.03024"
+  },
+  "summary": {
+    "text": "BugForge builds standardised DBMS bug repositories from bug reports and turns them into test cases. It collects reports, applies syntax-aware processing and input-adaptive extraction to recover raw proofs of concept, and stores structured bug metadata alongside them; semantic-guided adaptation then refines these into test cases for fuzzing, regression testing and cross-DBMS bug discovery. The authors integrated 37,632 reports spanning up to 28 years for PostgreSQL, MySQL, MariaDB and MonetDB, and found 35 previously unknown bugs, 22 confirmed.",
+    "relationship_to_sqlancer": "Reuses bug reports, including those produced by DBMS testing tools, as test material; the abstract does not state that it builds on SQLancer.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2604.03024",
+    "source_sha256": "sha256:148db53bffb075dbeec3639039d6db9c998b974d3f668b3db2968185905883b3",
+    "supporting_excerpts": [
+      "In this paper, we propose BugForge, a framework that constructs standardized DBMS bug repositories and leverages them to generate high-quality test cases to enhance DBMS testing.",
+      "Based on the repository, BugForge uncovered 35 previously unknown bugs with 22 confirmed by developers, demonstrating the value of constructing and utilizing bug repositories for DBMS testing."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/597ba93345ffe08175897ddb2019c676914992af",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2605_20473.json b/_data/impact/paper_notes/paper_arxiv_2605_20473.json
new file mode 100644
index 0000000..86448bf
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2605_20473.json
@@ -0,0 +1,65 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2605.20473",
+    "title": "Code Generation by Differential Test Time Scaling",
+    "authors": [
+      "Yifeng He",
+      "Ethan Wang",
+      "Jicheng Wang",
+      "Xuanxin Ouyang",
+      "Hao Chen"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2605.20473"
+  },
+  "summary": {
+    "text": "DiffCodeGen is a test-time scaling method for code generation that selects among candidate programs without extra model calls. It samples diverse candidates, uses coverage-guided fuzzing to synthesise inputs without needing existing tests, executes every candidate on those inputs, clusters them by behavioural similarity, and returns the medoid of the largest cluster. The authors evaluate it across four models and report competitive or better results than other test-time scaling methods at a fraction of the time and tokens.",
+    "relationship_to_sqlancer": "Applies differential testing to generated code rather than to DBMSs; SQLancer is cited as background rather than used.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2605.20473",
+    "source_sha256": "sha256:10b07319812bc452d81f8fc2122fd74cd91d833e1aacd91685f318abfc3f31a9",
+    "supporting_excerpts": [
+      "We present DiffCodeGen, a novel test-time scaling method for code generation based on coverage-guided differential analysis.",
+      "DiffCodeGen generates diverse code candidates using various sampling and prompting strategies, then applies coverage-guided fuzzing to synthesize inputs without requiring any existing tests or large language models."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2605.20473",
+          "source_type": "paper_citation_context",
+          "excerpt": "This technique has proven highly effective for testing compilers [16–18], database systems [19], and other systems where formal specifications or programmer-provided test oracles are either unavailable or impractical to use.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2605.20473",
+          "source_type": "paper_citation_context",
+          "excerpt": "Much research has applied differential testing with such assumptions to complex software systems like compilers and databases [13, 16, 19, 25], 6 Related work",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2605_22992.json b/_data/impact/paper_notes/paper_arxiv_2605_22992.json
new file mode 100644
index 0000000..3facb3e
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2605_22992.json
@@ -0,0 +1,126 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2605.22992",
+    "title": "Finding Performance Issues in Database Systems by Exploiting Dormant Code Paths",
+    "authors": [
+      "Jinsheng Ba",
+      "Zhendong Su"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2605.22992"
+  },
+  "summary": {
+    "text": "Branch Flip Analysis is a white-box method for finding DBMS performance issues. Rather than comparing execution times across black-box variants, it flips code branches to force an optimisation on or off; if disabling an optimisation makes the system significantly faster, an issue exists. The prototype, QueryZen, was evaluated on PostgreSQL, MySQL, CockroachDB and MariaDB with TPC-H and TPC-DS workloads and found 21 previously unknown performance issues.",
+    "relationship_to_sqlancer": "Argues against the black-box consistency methods it compares with, the category SQLancer's performance-oriented work belongs to.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2605.22992",
+    "source_sha256": "sha256:a9627e97f40185cc497371189d6d9a0cd83d14d0db90c827b60eb805453cbfa3",
+    "supporting_excerpts": [
+      "Existing work adopts black-box methods to examine performance consistency across executions, but cannot systematically test optimizations.",
+      "In this work, we propose a novel, general white-box methodology, Branch Flip Analysis (BFA), to systematically and effectively uncover performance issues.",
+      "We realized BFA in a prototype system QueryZen, and evaluated it on four widely-used and mature DBMSs: PostgreSQL, MySQL, CockroachDB, and MariaDB."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2605.22992",
+          "source_type": "paper_citation_context",
+          "excerpt": "Other approaches focus on discovering logic bugs in DBMSs. Oracles such as PQS [29], NoREC [27], and TLP [28] detect logic bugs in SELECT statement implementations, while DQE [31] targets logic errors in UPDATE and INSERT statements.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2605.22992",
+          "source_type": "paper_citation_context",
+          "excerpt": "CERT [1] finds performance issues by finding inconsistent cardinality estimation, which is typically deemed as the most critical component for query optimization [17].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2605.22992",
+          "source_type": "paper_citation_context",
+          "excerpt": "We considered the performance issues found by the four existing performance-testing methods: APOLLO [16], AMOEBA [21], CERT [1], and PUPPY [40].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2605.22992",
+          "source_type": "paper_citation_context",
+          "excerpt": "We evaluated QueryZen on PostgreSQL, MySQL, Mari-aDB, and CockroachDB, which are widely used in previous performance testing works [1, 16, 40].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2605.22992",
+          "source_type": "paper_citation_context",
+          "excerpt": "In this paper, we collected 10 thousand randomly generated small workloads from SQLancer [27,28], which is a popular DBMS testing tool.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2605.22992",
+          "source_type": "paper_citation_context",
+          "excerpt": "Unlike existing approaches [1,21,40], which focus on manipulating external inputs, BFA operates on the program code itself.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2605.22992",
+          "source_type": "paper",
+          "excerpt": "Issue reproducing.We evaluated whether the performance issues found byQueryZencan be potentially found by prior performance-testing methods APOLLO,AMOEBA,CERT, and PUPPY.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2605.22992",
+          "source_type": "paper",
+          "excerpt": "Issue existence date.Additionally, we evaluated whether these performance issues existed when prior performancetesting methods APOLLO,AMOEBA,CERT, and PUPPY were proposed.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2606_11132.json b/_data/impact/paper_notes/paper_arxiv_2606_11132.json
new file mode 100644
index 0000000..d6c6f34
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2606_11132.json
@@ -0,0 +1,81 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2606.11132",
+    "title": "Operationalizing Property-Based Testing for Data-Intensive Scalable Computing Systems",
+    "authors": [
+      "Yaoxuan Wu",
+      "I. Lee",
+      "Ahmad Humayun",
+      "Muhammad Ali Gulzar",
+      "Miryung Kim"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2606.11132"
+  },
+  "summary": {
+    "text": "DiscPBT is a property-based testing engine for Apache Spark. The authors argue that fuzzing's shallow oracles catch crashes but miss semantic drift and optimisation errors in data-intensive scalable computing frameworks, and that operationalising property-based testing there needs both reusable property definitions and a way to instantiate them into valid workloads and data. DiscPBT supplies eight reusable meta-properties covering equivalence rewriting, data and computation decomposition, and operator-local relations, along with generators for workload skeletons and input data. It reached 1.2x the branch coverage and 1153x the plan diversity of CometFuzz.",
+    "relationship_to_sqlancer": "Semantic-invariant testing for a different class of data system, evaluated against a fuzzer baseline rather than against SQLancer.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2606.11132",
+    "source_sha256": "sha256:1aa2eece2b83e4331f9149d1073673df590331430d0d559eb161246fb0622adf",
+    "supporting_excerpts": [
+      "We present DiscPBT, a property-based testing engine for Apache Spark.",
+      "DiscPBT introduces eight reusable meta-properties for DISC semantic testing, spanning equivalence rewriting, data decomposition, computation decomposition, and operator-local semantic relations.",
+      "Our evaluation on PySpark shows that DiscPBT achieves 1.2$\\times$ higher branch coverage and 1153$\\times$ greater plan diversity than CometFuzz."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/09e050c29b5f2592ad6d0b2a90121bb1530ead84",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2606.11132",
+          "source_type": "paper",
+          "excerpt": "We also do not compare against SQLancer because using it for Spark would require an additional adapter.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2606_14164.json b/_data/impact/paper_notes/paper_arxiv_2606_14164.json
new file mode 100644
index 0000000..c5ffc7f
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2606_14164.json
@@ -0,0 +1,60 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2606.14164",
+    "title": "Investigating Metamorphic Fuzz Oracle Enhancement via Large Language Models",
+    "authors": [
+      "Ruixiang Qian",
+      "Ding Yang",
+      "Zengxu Chen",
+      "Yue Gao",
+      "Chunrong Fang",
+      "Chao Zhang",
+      "Zhenyu Chen"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "url": "https://arxiv.org/abs/2606.14164"
+  },
+  "summary": {
+    "text": "A study of augmenting fuzz drivers with metamorphic oracles rather than relying on crash-based ones. The authors observe that crash oracles ignore library functionality and so limit what can be found, and that building metamorphic oracles by hand needs substantial domain knowledge. Their framework, MetaFOE, uses an LLM to generate metamorphic relations and integrate them into drivers automatically. On OSS-Fuzz drivers it generated 3,475 relations, 77.3% applicable, and its valid meta drivers raised edge coverage by 18.7% on average.",
+    "relationship_to_sqlancer": "Applies metamorphic oracles to general greybox fuzzing rather than to DBMSs; SQLancer is cited as prior use of that oracle style.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2606.14164",
+    "source_sha256": "sha256:e29d1a98b30c4aea1a2f7c86984b17339667a885a6c2a112d9a53754f7c033b0",
+    "supporting_excerpts": [
+      "In this paper, we present the first study on metamorphic-based fuzz oracle enhancement (MFOE), which augments existing fuzz drivers with metamorphic-based oracles derived from metamorphic relations (MRs).",
+      "To address this challenge, we propose MetaFOE, an LLM-based framework that automatically generates and integrates metamorphic-based oracles.",
+      "After three hours of fuzzing, the valid meta drivers improve edge coverage by an average of 18.7% and trigger 1,528 unique crashes."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2606.14164",
+          "source_type": "paper_citation_context",
+          "excerpt": "As a powerful technique for alleviating the oracle problem [Guo et al. 2024; Liu et al. 2013; Mu et al. 2025; Rigger and Su 2020; Segura et al. 2018], metamorphic testing has naturally attracted the attention of the fuzzing community and has been adopted to enhance fuzz oracles.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2607_03741.json b/_data/impact/paper_notes/paper_arxiv_2607_03741.json
new file mode 100644
index 0000000..82bba70
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2607_03741.json
@@ -0,0 +1,88 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2607.03741",
+    "title": "Graph-Aware Fuzzing for Graph Database Management Systems",
+    "authors": [
+      "Yu Li",
+      "Qiang Hu",
+      "Yao Zhang",
+      "Junjie Wang",
+      "Haoxuan Liu",
+      "Rui Wang",
+      "Yongqiang Lyu"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "url": "https://arxiv.org/abs/2607.03741"
+  },
+  "summary": {
+    "text": "GRAF is a black-box fuzzer for graph database query engines. It argues that existing GDBMS testing relies on differential and metamorphic testing whose consistency oracles restrict inputs to queries comparable across engines or transformations, leaving single-engine runtime failures under-explored. GRAF instantiates LLM-generated parameterised Cypher skeletons against the active graph state to produce valid, diverse queries, and applies five graph-specific mutation operators guided by execution feedback. It improved line coverage by 31.6-41.1% over the strongest baseline and found 34 previously unknown bugs, 23 with CVEs.",
+    "relationship_to_sqlancer": "Positioned against the differential and metamorphic testing methods for graph databases that SQLancer's oracles inspired.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2607.03741",
+    "source_sha256": "sha256:21025c6fe0e2b244c8ac39c488dfa310f5010829e4b38ffbc3336577c3328060",
+    "supporting_excerpts": [
+      "The result consistency oracles of these methods constrain inputs to queries that are comparable across engines or transformations, leaving single engine runtime failures, such as crashes and memory errors, insufficiently explored.",
+      "To address these challenges, we propose GRAF, a black box fuzzing framework for GDBMS query engines.",
+      "Overall, GRAF discovered 34 previously unknown bugs, with 32 confirmed by developers and 23 assigned CVEs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2607.03741",
+          "source_type": "paper_citation_context",
+          "excerpt": "Tools such as SQLsmith [7] and SQLancer [8]–[10] generate SQL queries based on abstract syntax tree models or formal semantics, successfully exposing bugs in mature systems such as PostgreSQL [11] and SQLite [12].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2607.03741",
+          "source_type": "paper_citation_context",
+          "excerpt": "To systematically detect logic bugs, SQLancer [8]–[10] automatically synthesizes queries and validates their correctness using advanced logical oracles like Ternary Logic Partitioning (TLP [8]).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2607.03741",
+          "source_type": "paper_citation_context",
+          "excerpt": "Existing GDBMS testing methods [15]–[20], [22], [25] mainly rely on differential or metamorphic testing and typically lack feedback driven exploration.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2607.03741",
+          "source_type": "paper_citation_context",
+          "excerpt": "Existing DBMS fuzzing techniques [7]–[10], [13], [25], [40]–[48] include generation-based and mutation-based approaches.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2607_09072.json b/_data/impact/paper_notes/paper_arxiv_2607_09072.json
new file mode 100644
index 0000000..1b7e48e
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2607_09072.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2607.09072",
+    "title": "Agentic Proof and Property-Based Testing via Property-Templates in Data-Intensive Computing",
+    "authors": [
+      "Seongmin Lee",
+      "Yaoxuan Wu",
+      "Miryung Kim"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "url": "https://arxiv.org/abs/2607.09072"
+  },
+  "summary": {
+    "text": "A study of recurring correctness properties in Apache Spark, expressed as parameterised property templates with holes. The authors use the templates in a dual-track framework that both proves properties in the Lean 4 theorem prover and instantiates them as executable PySpark property-based tests. They report that templates raise agentic proof success by up to 2.6x, cut proof hallucinations by 59%, and reduce intent misalignments in synthesised tests from 22 to 1.",
+    "relationship_to_sqlancer": "Testing for data-intensive systems rather than for DBMSs, and the work is not built on SQLancer; the abstract compares against a Spark fuzzer, not against SQLancer.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2607.09072",
+    "source_sha256": "sha256:efc9523e6e62c6d769b4d26f8bd6fbb346d09d82a81884bbccdaba41fc62d9b7",
+    "supporting_excerpts": [
+      "We design an agentic, dual-track validation framework that uses property templates to formally verify correctness in the Lean 4 theorem prover and instantiate PBT templates as executable PySpark tests.",
+      "Template-guided synthesis further exceeds a state-of-the-art Spark fuzzer and approaches unguided LLM-based PBT on code coverage."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2607.09072",
+          "source_type": "paper_citation_context",
+          "excerpt": "On the testing side, SQLancer detects logic and optimization bugs in database engines through constructed oracles such as query partitioning and a non-optimizing reference engine [28], [29].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2607_13276.json b/_data/impact/paper_notes/paper_arxiv_2607_13276.json
new file mode 100644
index 0000000..d39424c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2607_13276.json
@@ -0,0 +1,58 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2607.13276",
+    "title": "Aurora DSQL: Scalable, Multi-Region OLTP",
+    "authors": [
+      "Marc Brooker",
+      "Mark Bowes",
+      "Mike Hershey",
+      "Zak van der Merwe",
+      "James Morle",
+      "Matthys Strydom"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "url": "https://arxiv.org/abs/2607.13276"
+  },
+  "summary": {
+    "text": "A system description of Aurora DSQL, a serverless, multi-region active-active SQL database. Compute, storage and transaction coordination are separated into independently scalable services, with PostgreSQL-compatible query processors running statelessly in Firecracker MicroVMs. It uses multiversion concurrency control with precision timestamps for coordination-free reads and optimistic concurrency control for writes, deferring coordination to commit time so that cross-region latency is paid only on commit.",
+    "relationship_to_sqlancer": "A database system paper rather than a testing paper. It cites SQLancer among the work on database correctness; the abstract does not describe using it.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2607.13276",
+    "source_sha256": "sha256:75c240c237b457cd7cd352fa6a7a54bbea1152984b0ce3206f52d70606dbc99e",
+    "supporting_excerpts": [
+      "Aurora DSQL is a serverless SQL database designed for cloud-scale transaction processing with multi-region active-active capabilities.",
+      "The system uses multiversion concurrency control with precision timestamps for coordination-free reads and optimistic concurrency control for writes, deferring coordination to commit time through distributed adjudicators and the Journal replication system."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2607.13276",
+          "source_type": "paper_citation_context",
+          "excerpt": "The fuzz-testing approach, building on the approach of SQLancer [4], generates millions of example SQL statements and runs them both on DSQL and on Aurora PostgreSQL.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2608_15709.json b/_data/impact/paper_notes/paper_arxiv_2608_15709.json
new file mode 100644
index 0000000..eecd117
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2608_15709.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2608.15709",
+    "title": "Logos: Certified Order-Sensitive SQL Rewrites with Mechanized Semantics and LLM Guidance",
+    "authors": [
+      "Jing-Yu Ke",
+      "Jingyang Li",
+      "Guoqiang Li"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "url": "https://arxiv.org/abs/2608.15709"
+  },
+  "summary": {
+    "text": "Logos verifies SQL rewrite equivalence in the Rocq proof assistant. The authors first mechanise a compositional logical semantics for a typed SQL core with order-sensitive operators, capturing all possible ordered lists and observable failures, which they describe as the first mechanised SQL semantics to combine nested tie-sensitive top-k with a lifting from bag equivalence to ordered-list equivalence. Logos then uses an LLM-guided agent with a verified lemma library to build query-specific proofs, solving 86.9% of 389 query pairs against 64.0% for the strongest baseline.",
+    "relationship_to_sqlancer": "Proves query equivalence rather than testing for it; the equivalences it verifies are the kind SQLancer's oracles assume.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2608.15709",
+    "source_sha256": "sha256:87e2310a1ec3ebade62c918a8db8e576b6cc9b81dbf0bf4a6450904ba0947bbd",
+    "supporting_excerpts": [
+      "In Rocq, we mechanize a compositional logical semantics for a typed SQL core with order-sensitive operators, capturing all possible ordered lists and observable SQL failures in the supported fragment.",
+      "Building on this semantics, we present Logos, an LLM-guided Rocq verifier for unbounded SQL rewrite equivalence.",
+      "Logos solves 86.9% of them, compared with 64.0% for SQLSolver, the strongest baseline."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/513b2356cc9a799751b6c96ce838f98e67b8696e",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2608_23402.json b/_data/impact/paper_notes/paper_arxiv_2608_23402.json
new file mode 100644
index 0000000..ada9b05
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2608_23402.json
@@ -0,0 +1,76 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2608.23402",
+    "title": "EXPLAIN Yourself! Finding Query Planner Stalls Across DBMSes",
+    "authors": [
+      "Geoffrey X. Yu",
+      "Ryan Marcus",
+      "Tim Kraska"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "url": "https://arxiv.org/abs/2608.23402"
+  },
+  "summary": {
+    "text": "An empirical study of queries that make a DBMS spend an unreasonable amount of time planning. Using a lightweight agentic search, the authors find at least one query per system taking more than three minutes to plan across seven DBMSs, four of them commercial, and note that such queries tie up resources without doing useful work and so form a denial-of-service vector. Although the triggering queries are largely system-specific, recurring pathologies around correlated subqueries, CTE expansion, repeated subquery expressions, disjunctive joins and constant folding affect several systems.",
+    "relationship_to_sqlancer": "Targets planner latency rather than result correctness; SQLancer is cited among DBMS testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2608.23402",
+    "source_sha256": "sha256:cfe5d68b6412be0d625fd5be37dc3dae77f07b34e241c3c1406ace602268a9fc",
+    "supporting_excerpts": [
+      "Across seven DBMSes, including four commercial systems, we find at least one query per system that takes more than three minutes to plan.",
+      "We find that although the queries triggering slow planning are largely DBMS-specific, recurring pathologies involving correlated subqueries, CTE expansion, repeated subquery expressions, disjunctive joins, and constant folding affect multiple systems."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2608.23402",
+          "source_type": "paper_citation_context",
+          "excerpt": "The high-level idea is to identify a general property about a pair of queries (e.g., making a query more restrictive should result in an output cardinality estimate no larger than the original query [4]).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2608.23402",
+          "source_type": "paper_citation_context",
+          "excerpt": "A complementary set of papers propose techniques to find correctness bugs in database systems [3, 5, 16, 27–29, 40] , applying a similar approach grounded in query",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2608.23402",
+          "source_type": "paper_citation_context",
+          "excerpt": "Prior works propose techniques for finding query execution performance bugs in database systems [4, 11, 12, 15, 35, 36].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2608_25573.json b/_data/impact/paper_notes/paper_arxiv_2608_25573.json
new file mode 100644
index 0000000..21265bb
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2608_25573.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2608.25573",
+    "title": "DBcover: A White-box SQL Test Generation Framework for Coverage Improvement",
+    "authors": [
+      "Yan-Kai Rong",
+      "Shuang Liu",
+      "Jinhao Dong",
+      "Qiang Yin",
+      "Wei Lu",
+      "Jian-Hua Wang",
+      "Xiaoyong Du"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "url": "https://arxiv.org/abs/2608.25573"
+  },
+  "summary": {
+    "text": "DBcover generates SQL tests for RDBMSs with white-box knowledge of the engine. Lightweight dynamic analysis extracts the correspondence between SQL inputs and execution paths along with call graphs as global context, and source-level information around target functions as local context, all organised in a knowledge graph. Generation then runs in two phases: pick a seed whose execution path is near the uncovered target, then prompt an LLM with both contexts to produce queries reaching uncovered code. The authors report 80.1% and 82.3% coverage on PostgreSQL and MySQL.",
+    "relationship_to_sqlancer": "Coverage-oriented SQL generation, contrasted with random SQL generation of the kind SQLancer performs.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2608.25573",
+    "source_sha256": "sha256:89933da1e4b782914b287391b7b8e3565625332cc4798347ad77188d5769b9e4",
+    "supporting_excerpts": [
+      "We propose DBcover, an LLM-driven white-box SQL test generation framework based on contextual reasoning.",
+      "DBcover then performs two-phase test generation: it first selects a semantically relevant seed whose execution path is close to the uncovered target, and then guides the LLM with global and local context to generate SQL test cases that trigger previously uncovered code regions."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/2b1f7251af13eb3114cc0c8812ae6a07cf970db2",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2608_30385.json b/_data/impact/paper_notes/paper_arxiv_2608_30385.json
new file mode 100644
index 0000000..0408eba
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2608_30385.json
@@ -0,0 +1,147 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2608.30385",
+    "title": "Detecting DBMS Bugs by Constructing Equivalent Representations of Intermediate Query Results",
+    "authors": [
+      "Xiao-Xu Niu",
+      "Gong Chen",
+      "Jin-Fu Chen",
+      "Xiaoyuan Xie"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "url": "https://arxiv.org/abs/2608.30385"
+  },
+  "summary": {
+    "text": "ERIQ tests DBMSs by checking that different SQL mechanisms for representing the same intermediate result agree. It builds variants of a query using a VIEW, a common table expression, or a temporary table for the same intermediate result, runs them, and compares the outputs, treating disagreement as a logic bug. On MySQL, MariaDB, Percona and OceanBase it found 64 bugs, 63 confirmed by developers, of which 54 were unique and previously unknown.",
+    "relationship_to_sqlancer": "A logic-bug oracle built on equivalent query representations, evaluated against existing logic-bug detection approaches and describing them as the state of the art.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2608.30385",
+    "source_sha256": "sha256:e77e2f9989c4704f84ea08735bcc20af42ccd2bc7e7e4b793b2d4a2fd0599583",
+    "supporting_excerpts": [
+      "Existing approaches for detecting DBMS logic bugs have never explored result consistency across such equivalent representations.",
+      "In this paper, we propose ERIQ, a novel testing approach for detecting DBMS logic bugs from the perspective of checking result consistency across Equivalent Representations of Intermediate Query Results.",
+      "In total, ERIQ detected 64 bugs, 63 of which were confirmed by developers, and two have been fixed."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2608.30385",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, NoREC [29] compares the results of optimized and unoptimized queries; TLP [30] partitions queries based on ternary logic; PQS [31] checks whether a selected pivot row appears in the result of a generated query; Pinolo [8] checks result containment between synthesized queries and a seed…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2608.30385",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoREC [29] rewrites a query into a form that inhibits DBMS optimizations and compares the results of the original and rewritten queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2608.30385",
+          "source_type": "paper_citation_context",
+          "excerpt": "We further compared ERIQ with four state-of-the-art DBMS logic bug detection approaches: EDC [4], Radar [34], EET [11], and TLP [30].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2608.30385",
+          "source_type": "paper_citation_context",
+          "excerpt": "We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [4], Radar [34], EET [11], and TLP [30].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2608.30385",
+          "source_type": "paper_citation_context",
+          "excerpt": "To answer RQ2, we compared ERIQ with the 4 baselines introduced in Section 4.1: EDC [4], Radar [34], EET [11], and TLP [30].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2608.30385",
+          "source_type": "paper_citation_context",
+          "excerpt": "QPG [1] uses query-plan diversity to guide database-state mutations, and MIST [3] combines hierarchical SQL-feature guidance, error feedback, and coverage-guided Monte Carlo tree search for LLM-based DBMS test-case generation.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "manual_curation",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2608.30385",
+          "source_type": "paper_citation_context",
+          "excerpt": "We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [4], Radar [34], EET [11], and TLP [30].",
+          "excerpt_is_verbatim": true,
+          "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+          "retrieved_at": "2026-09-06T07:08:48Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2608.30385",
+          "source_type": "paper_citation_context",
+          "excerpt": "We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [4], Radar [34], EET [11], and TLP [30].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Ternary Logic Partitioning (TLP) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2608.30385",
+          "source_type": "paper_citation_context",
+          "excerpt": "We further compared ERIQ with four state-of-the-art DBMS logic bug detection approaches: EDC [4], Radar [34], EET [11], and TLP [30].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Ternary Logic Partitioning (TLP) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_arxiv_2609_00381.json b/_data/impact/paper_notes/paper_arxiv_2609_00381.json
new file mode 100644
index 0000000..65f7bd5
--- /dev/null
+++ b/_data/impact/paper_notes/paper_arxiv_2609_00381.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:arxiv:2609.00381",
+    "title": "Bounded, Indeterminate, or a Bug: A Condition-Aware Oracle for Differential Testing of SQL Aggregates",
+    "authors": [
+      "Madhulatha Mandarapu",
+      "Sandeep Kunkunuru"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "url": "https://arxiv.org/abs/2609.00381"
+  },
+  "summary": {
+    "text": "An oracle for differential testing of floating-point SQL aggregates. The paper argues that treating any cross-engine discrepancy as a bug is unsound for floating point, because non-associativity means engines legitimately disagree, and that the deciding factor is the engine's summation algorithm rather than the query. It takes ground truth to be the exact rational value of the stored doubles, classifies each discrepancy as exact, bounded or indeterminate, and derives a testability boundary beyond which rounding cannot be separated from a bug.",
+    "relationship_to_sqlancer": "Addresses a case SQLancer's oracles avoid, noting that the leading oracles sidestep floating point entirely.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2609.00381",
+    "source_sha256": "sha256:1bdd0892994617eaf3ee86c3a2c2f987cebea43baea4f0e14f85c347dfe8f200",
+    "supporting_excerpts": [
+      "We give the oracle this practice lacks, and show its decisive quantity is not the query but the engine's algorithm.",
+      "Ground truth is the exact rational value of the stored doubles -- arithmetic, not another engine -- and each discrepancy is classified exact, bounded, or indeterminate."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2609.00381",
+          "source_type": "paper_citation_context",
+          "excerpt": "The oracles of Rigger and Su [2020b,a,c] either restrict aggregates to a single pivot row, target predicates rather than aggregation, or compare an engine against itself by query partitioning, so a consistent rounding error cancels and is never observed.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1002_9781119880929_ch13.json b/_data/impact/paper_notes/paper_doi_10_1002_9781119880929_ch13.json
new file mode 100644
index 0000000..753adc2
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1002_9781119880929_ch13.json
@@ -0,0 +1,42 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1002/9781119880929.ch13",
+    "title": "Emerging Aspects of Software Fault Localization",
+    "authors": [
+      "T.H. Tse",
+      "David Lo",
+      "Alex Gorce",
+      "Michael Perscheid",
+      "Robert Hirschfeld",
+      "W. Eric Wong"
+    ],
+    "year": 2023,
+    "venue": null,
+    "doi": "10.1002/9781119880929.ch13",
+    "url": "https://doi.org/10.1002/9781119880929.ch13"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://openalex.org/W4366606136",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "OpenAlex records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1007_978_3_030_71058_3_5.json b/_data/impact/paper_notes/paper_doi_10_1007_978_3_030_71058_3_5.json
new file mode 100644
index 0000000..3c8ced1
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1007_978_3_030_71058_3_5.json
@@ -0,0 +1,40 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1007/978-3-030-71058-3_5",
+    "title": "Artemis: An Automatic Test Suite Generator for Large Scale OLAP Database",
+    "authors": [
+      "Kaiming Mi",
+      "Chunxi Zhang",
+      "Weining Qian",
+      "Rong Zhang"
+    ],
+    "year": 2021,
+    "venue": "Lecture notes in computer science",
+    "doi": "10.1007/978-3-030-71058-3_5",
+    "url": "https://doi.org/10.1007/978-3-030-71058-3_5"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://openalex.org/W3134710759",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "OpenAlex records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1007_978_3_030_88494_9_12.json b/_data/impact/paper_notes/paper_doi_10_1007_978_3_030_88494_9_12.json
new file mode 100644
index 0000000..385db50
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1007_978_3_030_88494_9_12.json
@@ -0,0 +1,62 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1007/978-3-030-88494-9_12",
+    "title": "Differential Monitoring",
+    "authors": [
+      "Fabian Mühlböck",
+      "T. Henzinger"
+    ],
+    "year": 2021,
+    "venue": "Runtime Verification",
+    "doi": "10.1007/978-3-030-88494-9_12",
+    "url": "https://doi.org/10.1007/978-3-030-88494-9_12"
+  },
+  "summary": {
+    "text": "A paper on differential monitoring. Only the sentence citing SQLancer is available here, so this summary is limited to it: the authors note that the technique has been applied fruitfully to finding bugs in JavaScript debuggers, C compilers and SQL databases.",
+    "relationship_to_sqlancer": "Cites SQLancer's work among the successful applications of differential techniques to SQL databases.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://doi.org/10.1007/978-3-030-88494-9_12",
+    "source_sha256": "sha256:d1560e96cc95af8e42bcf9a0fad845415d1672e1cedd4b076b821a6ee7a412b0",
+    "supporting_excerpts": [
+      "This technique has been fruitfully applied to finding bugs in Javascript debuggers [31], C compilers [41], and SQL databases [39, 37, 36]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:12Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1007/978-3-030-88494-9_12",
+          "source_type": "paper_citation_context",
+          "excerpt": "This technique has been fruitfully applied to finding bugs in Javascript debuggers [31], C compilers [41], and SQL databases [39, 37, 36].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1007/978-3-030-88494-9_12",
+          "source_type": "paper_citation_context",
+          "excerpt": "This technique has been fruitfully applied to finding bugs in Javascript debuggers [31], C compilers [42], and SQL databases [40,38,37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1007_978_3_031_51479_1_17.json b/_data/impact/paper_notes/paper_doi_10_1007_978_3_031_51479_1_17.json
new file mode 100644
index 0000000..a377e93
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1007_978_3_031_51479_1_17.json
@@ -0,0 +1,55 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1007/978-3-031-51479-1_17",
+    "title": "Syntax-Aware Mutation for Testing the Solidity Compiler",
+    "authors": [
+      "Charalambos Mitropoulos",
+      "T. Sotiropoulos",
+      "S. Ioannidis",
+      "Dimitris Mitropoulos"
+    ],
+    "year": 2023,
+    "venue": "European Symposium on Research in Computer Security",
+    "doi": "10.1007/978-3-031-51479-1_17",
+    "url": "https://doi.org/10.1007/978-3-031-51479-1_17"
+  },
+  "summary": {
+    "text": "A paper on syntax-aware mutation for testing the Solidity compiler. Only the sentence citing SQLancer is available here, so this summary is limited to it: the authors situate their work by noting that fuzzing has been used to find bugs in a range of targets including system libraries, web and cloud applications, data-oriented systems, and compilers.",
+    "relationship_to_sqlancer": "Cites SQLancer's work among the fuzzing of data-oriented systems when surveying where fuzzing has been applied.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://doi.org/10.1007/978-3-031-51479-1_17",
+    "source_sha256": "sha256:9210060d51c18407538f848db1cd3833b121bf05f7e34c0ab161c03e0b28239c",
+    "supporting_excerpts": [
+      "Fuzzing has been used to identify bugs in miscellaneous entities such as system libraries [35], web and cloud applications [10], data-oriented systems [39,40], and compilers [48,34,19]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1007/978-3-031-51479-1_17",
+          "source_type": "paper_citation_context",
+          "excerpt": "Fuzzing has been used to identify bugs in miscellaneous entities such as system libraries [35], web and cloud applications [10], data-oriented systems [39,40], and compilers [48,34,19].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1007_978_3_031_94706_3_7.json b/_data/impact/paper_notes/paper_doi_10_1007_978_3_031_94706_3_7.json
new file mode 100644
index 0000000..38b42fa
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1007_978_3_031_94706_3_7.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1007/978-3-031-94706-3_7",
+    "title": "Fuzzing Graph Database Applications with Graph Transformations",
+    "authors": [
+      "Stefania Dumbrava",
+      "Melchior W. M. Oudemans",
+      "Burcu Kulahcioglu Ozkan"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Graph Transformation",
+    "doi": "10.1007/978-3-031-94706-3_7",
+    "url": "https://doi.org/10.1007/978-3-031-94706-3_7"
+  },
+  "summary": {
+    "text": "PGFuzz is a greybox fuzzer for applications backed by graph databases, which the authors describe as the first to target that setting. The difficulty is the size of the graph schema state space: naive random graph instances cover little of an application. PGFuzz builds on existing graph generators and applies coverage-guided graph transformations that are schema-aware and honour graph schema, key and cardinality constraints. On graph database applications curated from open-source repositories it substantially improves test coverage over the state of the art.",
+    "relationship_to_sqlancer": "Tests applications built on graph databases rather than the database engines themselves; SQLancer is cited among database testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1007/978-3-031-94706-3_7",
+    "source_sha256": "sha256:1672c7808f4f2ef81ef896a7761023695b68977e8cd721dc9e60b273dd934725",
+    "supporting_excerpts": [
+      "Due to the sheer dimension of the graph schema state space, testing applications using naive random graph instances is unlikely to cover a large portion of an application program.",
+      "We present PGFuzz , a graph transformation-based greybox fuzzer for testing graph database-backed applications, that is, to the best of our knowledge, the first fuzzer to specifically target graph database applications.",
+      "PGFuzz builds on top of state-of-the-art graph generators and utilizes graph transformations guided by code coverage to produce application test inputs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "qpg",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1007/978-3-031-94706-3_7",
+          "source_type": "paper_citation_context",
+          "excerpt": "Several recent works focus on testing database management systems [9,17,45] and graph databases [47,49,83].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1007_978_981_95_3182_0_3.json b/_data/impact/paper_notes/paper_doi_10_1007_978_981_95_3182_0_3.json
new file mode 100644
index 0000000..a180770
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1007_978_981_95_3182_0_3.json
@@ -0,0 +1,40 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1007/978-981-95-3182-0_3",
+    "title": "Ghosts in DBMS: Revealing the Security Impacts of Silent Fixes",
+    "authors": [
+      "Jialiang Dong",
+      "Zihan Ni",
+      "Willy Susilo",
+      "Siqi Ma"
+    ],
+    "year": 2025,
+    "venue": "Lecture notes in computer science",
+    "doi": "10.1007/978-981-95-3182-0_3",
+    "url": "https://doi.org/10.1007/978-981-95-3182-0_3"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://openalex.org/W4415219679",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "OpenAlex records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1007_978_981_95_4721_0_7.json b/_data/impact/paper_notes/paper_doi_10_1007_978_981_95_4721_0_7.json
new file mode 100644
index 0000000..a421328
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1007_978_981_95_4721_0_7.json
@@ -0,0 +1,40 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1007/978-981-95-4721-0_7",
+    "title": "CypherFuzzer: A Tool for Testing Access Control in Graph Databases",
+    "authors": [
+      "Philipp Reisinger",
+      "Daniel Hofer",
+      "Bahara Muradi",
+      "Josef Küng"
+    ],
+    "year": 2025,
+    "venue": "Communications in computer and information science",
+    "doi": "10.1007/978-981-95-4721-0_7",
+    "url": "https://doi.org/10.1007/978-981-95-4721-0_7"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://openalex.org/W7106652132",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "OpenAlex records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1007_978_981_96_4506_0_18.json b/_data/impact/paper_notes/paper_doi_10_1007_978_981_96_4506_0_18.json
new file mode 100644
index 0000000..5cf975c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1007_978_981_96_4506_0_18.json
@@ -0,0 +1,43 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1007/978-981-96-4506-0_18",
+    "title": "Review of Fuzz Testing Techniques for Database Management Systems",
+    "authors": [
+      "Yuheng Zhang",
+      "Hui Lu",
+      "Zhourui Zhang",
+      "Guo–Cheng Wu",
+      "Houlin Zhou",
+      "Zhenghao Li"
+    ],
+    "year": 2025,
+    "venue": "Communications in computer and information science",
+    "doi": "10.1007/978-981-96-4506-0_18",
+    "url": "https://doi.org/10.1007/978-981-96-4506-0_18"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://openalex.org/W4410099864",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "OpenAlex records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1007_978_981_96_6465_8_28.json b/_data/impact/paper_notes/paper_doi_10_1007_978_981_96_6465_8_28.json
new file mode 100644
index 0000000..66d6bba
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1007_978_981_96_6465_8_28.json
@@ -0,0 +1,43 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1007/978-981-96-6465-8_28",
+    "title": "Fuzz Testing for Database Management System Configuration Errors",
+    "authors": [
+      "Haoran Zhu",
+      "Menglin Li",
+      "Bo Wang",
+      "Tengfei Li",
+      "Jingtian Liu",
+      "Zan Zhou"
+    ],
+    "year": 2025,
+    "venue": "Communications in computer and information science",
+    "doi": "10.1007/978-981-96-6465-8_28",
+    "url": "https://doi.org/10.1007/978-981-96-6465-8_28"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://openalex.org/W4413060152",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "OpenAlex records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1007_s10664_025_10662_w.json b/_data/impact/paper_notes/paper_doi_10_1007_s10664_025_10662_w.json
new file mode 100644
index 0000000..f37cf20
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1007_s10664_025_10662_w.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1007/s10664-025-10662-w",
+    "title": "Detecting data manipulation errors in android applications using scene-guided exploration",
+    "authors": [
+      "Shuqi Liu",
+      "Yu Zhou",
+      "Wenhua Yang",
+      "Taolue Chen",
+      "Harald C. Gall"
+    ],
+    "year": 2025,
+    "venue": "Empirical Software Engineering",
+    "doi": "10.1007/s10664-025-10662-w",
+    "url": "https://doi.org/10.1007/s10664-025-10662-w"
+  },
+  "summary": {
+    "text": "This paper addresses data manipulation errors in Android applications using scene-guided testing. The available text is limited to the sentence in which it cites SQLancer's work, so the summary here is confined to that: the authors position their work against existing studies that detected CRUD errors in database management systems, arguing those approaches are not tailored to Android apps.",
+    "relationship_to_sqlancer": "Cites Rigger and Su's work on detecting errors in database management systems as the closest prior work, while arguing it does not transfer to Android applications.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://doi.org/10.1007/s10664-025-10662-w",
+    "source_sha256": "sha256:e4e04254ad64c3d93e44e6063b081b4d39bca7b602d60cd2e0f4a2ad64974ee8",
+    "supporting_excerpts": [
+      "Existing studies (Rigger and Su 2020a,b have detected CRUD errors in database management systems, but these approaches are not specifically tailored for Android apps."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1007/s10664-025-10662-w",
+          "source_type": "paper_citation_context",
+          "excerpt": "Existing studies (Rigger and Su 2020a,b have detected CRUD errors in database management systems, but these approaches are not specifically tailored for Android apps.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1016_j_cose_2025_104564.json b/_data/impact/paper_notes/paper_doi_10_1016_j_cose_2025_104564.json
new file mode 100644
index 0000000..da54d06
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1016_j_cose_2025_104564.json
@@ -0,0 +1,170 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1016/j.cose.2025.104564",
+    "title": "Detecting DBMS bugs with context-sensitive instantiation and multi-plan execution",
+    "authors": [
+      "Jiaqi Li",
+      "Ke Wang",
+      "Yaoguang Chen",
+      "Yajin Zhou",
+      "Lei Wu",
+      "Jiashui Wang"
+    ],
+    "year": 2025,
+    "venue": "Computers & Security",
+    "doi": "10.1016/j.cose.2025.104564",
+    "url": "https://doi.org/10.1016/j.cose.2025.104564"
+  },
+  "summary": {
+    "text": "Kangaroo targets both memory and logic bugs in DBMSs with two techniques. Context-sensitive instantiation takes all static semantic requirements into account, not just identifier types as existing systems do, so generated queries are semantically valid. Multi-plan execution provides the oracle: rather than running only the optimizer's chosen plan, the DBMS executes all query plans for a test case and the results are compared, with any difference indicating a logic bug. Applied to SQLite, PostgreSQL and MySQL it detected 50 new bugs.",
+    "relationship_to_sqlancer": "Compares against SQLancer as a state-of-the-art system, on valid query generation, code paths explored and bugs found.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://arxiv.org/abs/2312.04941",
+    "source_sha256": "sha256:0c70c78509344e9db83c0232b8e9f130d3219a6bf4fa0d6a74a9e8a0c5f86553",
+    "supporting_excerpts": [
+      "The first key technique is called context-sensitive instantiation, which considers all static semantic requirements (including but not limited to the identifier type used by existing systems) to generate semantically valid SQL queries.",
+      "Given a test case, multi-plan execution makes the DBMS execute all query plans instead of the default optimal one, and compares the results.",
+      "The comparison between our system with the state-of-the-art systems shows that our system outperforms them in terms of the number of generated semantically valid SQL queries, the explored code paths during testing, and the detected bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/941500a2f13307898e0f0152e4bde173d51dc7bb",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Differential Query Plans (DQP).",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941",
+          "source_type": "paper_citation_context",
+          "excerpt": "The generation-based approach [7, 14–17, 27, 37, 38] is effective in generating syntax-correct test cases since it typically follows a grammar model that describes the format of the input.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, NoREC [15] requires that the effective SQL query in a test case has WHERE clauses, thus it can only detect logic bugs due to the optimization of WHERE clauses.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941",
+          "source_type": "paper_citation_context",
+          "excerpt": "Although Rigger et al. proposed three oracles for DBMS logic bug detection, including PQS [14], NoREC [15], and TLP [16], all of them put limitations on the SQL queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941",
+          "source_type": "paper_citation_context",
+          "excerpt": "The rule-based ones [14–17 , 24 The mutation-based method [6, 13, 25] generates new test cases by mutating seed queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941",
+          "source_type": "paper_citation_context",
+          "excerpt": "To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoREC [15] generates equivalent queries by shifting the conditions in the WHERE clause to the SELECT expression.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941v1",
+          "source_type": "paper",
+          "excerpt": "We also compared Kangaroo with leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941v1",
+          "source_type": "paper",
+          "excerpt": "Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo , and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941v1",
+          "source_type": "paper",
+          "excerpt": "We use Squirrel as a baseline since it performs better than SQLancer in detecting memory bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941",
+          "source_type": "paper",
+          "excerpt": "We also compared Kangaroo with leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941",
+          "source_type": "paper",
+          "excerpt": "Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo , and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2312.04941",
+          "source_type": "paper",
+          "excerpt": "We use Squirrel as a baseline since it performs better than SQLancer in detecting memory bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1016_j_displa_2024_102854.json b/_data/impact/paper_notes/paper_doi_10_1016_j_displa_2024_102854.json
new file mode 100644
index 0000000..b79efdd
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1016_j_displa_2024_102854.json
@@ -0,0 +1,39 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1016/j.displa.2024.102854",
+    "title": "Using query semantic and feature transfer fusion to enhance cardinality estimating of property graph queries",
+    "authors": [
+      "Zhenzhen He",
+      "Tiquan Gu",
+      "Jiong Yu"
+    ],
+    "year": 2024,
+    "venue": "Displays (Guildford)",
+    "doi": "10.1016/j.displa.2024.102854",
+    "url": "https://doi.org/10.1016/j.displa.2024.102854"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/1faf7f08681013961cb385344a600bd5f08c69ec",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Cardinality Estimation Restriction Testing (CERT).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_ase56229_2023_00106.json b/_data/impact/paper_notes/paper_doi_10_1109_ase56229_2023_00106.json
new file mode 100644
index 0000000..4edfae5
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_ase56229_2023_00106.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/ase56229.2023.00106",
+    "title": "Perfce: Performance Debugging on Databases with Chaos Engineering-Enhanced Causality Analysis",
+    "authors": [
+      "Zhenlan Ji",
+      "Pingchuan Ma",
+      "Shuai Wang"
+    ],
+    "year": 2022,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1109/ase56229.2023.00106",
+    "url": "https://doi.org/10.1109/ase56229.2023.00106"
+  },
+  "summary": {
+    "text": "PerfCE diagnoses performance anomalies in databases by combining causal inference with chaos engineering. Causal analysis of performance downgrades is hampered by limited observability, so PerfCE uses chaos experiments — injecting events such as network slowdowns — to gather the observations it needs. Offline, it learns statistical models of a database from both passive observation and proactive chaos experiments; online, it diagnoses root causes qualitatively and quantitatively as anomalies occur. It outperformed prior work on synthetic data and was accurate on MySQL and TiDB.",
+    "relationship_to_sqlancer": "Notes that systems under chaos stress are checked with differential testing for correct SQL results, the setting SQLancer's oracles serve.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/ase56229.2023.00106",
+    "source_sha256": "sha256:c1c8a45a099e29955c76fbe67afda1895e26bc4396572d88c1d38eda5d92aebb",
+    "supporting_excerpts": [
+      "Nevertheless, causality analysis is challenging in practice, particularly due to limited observability.",
+      "The systems under chaos stress are then tested (e.g., via differential testing) to check if they retain normal functionality, such as returning correct SQL query outputs even under stress.",
+      "This paper identifies the novel usage of CE in diagnosing performance anomalies in databases."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:38:27Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/ase56229.2023.00106",
+          "source_type": "paper_citation_context",
+          "excerpt": "This setup expands the standard “differential testing” procedure [37–39], requiring consistency between the experimental group and reference even under CE stress.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00095.json b/_data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00095.json
new file mode 100644
index 0000000..43a4b25
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00095.json
@@ -0,0 +1,61 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/ase63991.2025.00095",
+    "title": "ZendDiff: Differential Testing of PHP Interpreter",
+    "authors": [
+      "Yuancheng Jiang",
+      "Jianing Wang",
+      "Qiange Liu",
+      "Yeqi Fu",
+      "Jian Mao",
+      "Roland H. C. Yap",
+      "Zhenkai Liang"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1109/ase63991.2025.00095",
+    "url": "https://doi.org/10.1109/ase63991.2025.00095"
+  },
+  "summary": {
+    "text": "ZendDiff finds logic bugs in the PHP interpreter by differential testing. The authors observe that existing PHP bug-finding targets crashes or sanitizer-based oracles and misses silent wrong results, and that PHP's JIT compilation mode provides a second implementation of the same specification to compare against. ZendDiff compares JIT and non-JIT execution with program state probing for fine-grained comparison, JIT-aware mutation, and dual verification for non-determinism. It has found 51 previously unknown logic bugs, 37 already fixed.",
+    "relationship_to_sqlancer": "Applies differential testing for silent wrong results to a language interpreter; SQLancer is cited as prior work on logic-bug oracles.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/ase63991.2025.00095",
+    "source_sha256": "sha256:934670f385c169127e3fa2501886d05d34435f7869d342efa4408d6d0b2d4fbc",
+    "supporting_excerpts": [
+      "Existing approaches to finding bugs in PHP primarily focus on detecting explicit security issues through crashes or sanitizer-based oracles, but fail to identify logic bugs that can silently lead to incorrect results.",
+      "We observe that the introduction of Just-In-Time (JIT) compilation mode in PHP presents an opportunity for differential testing, as it provides an alternative implementation of the same language specification.",
+      "To date, ZendDiffhas identified 51 previously unknown logic bugs in the PHP interpreter, with 37 already fixed and 3 confirmed by the PHP maintainers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/ase63991.2025.00095",
+          "source_type": "paper_citation_context",
+          "excerpt": "Code coverage is a widely adopted metric in evaluating fuzzing and testing approaches [40, 41], as it provides a clear and quantifiable way to evaluate which parts of the code have been executed during tests.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00151.json b/_data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00151.json
new file mode 100644
index 0000000..33976b9
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00151.json
@@ -0,0 +1,70 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/ase63991.2025.00151",
+    "title": "ARG: Testing Query Rewriters via Abstract Rule Guided Fuzzing",
+    "authors": [
+      "Dawei Li",
+      "Yuxiao Guo",
+      "Qifan Liu",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Chi Zhang",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1109/ase63991.2025.00151",
+    "url": "https://doi.org/10.1109/ase63991.2025.00151"
+  },
+  "summary": {
+    "text": "ARG fuzzes query rewriters, the components that transform a query into a faster but semantically equivalent form. The authors argue that general DBMS testing approaches cover only a limited subset of rewrite scenarios given the diversity of rewrite rules. ARG uses abstract rules — a unified representation of the AST patterns and constraints that trigger a rewrite, plus the resulting transformation — as coverage feedback, steering generation toward patterns not yet exercised. Testing Apache Calcite, WeTune, SQLSolver and LearnedRewrite it found 38 previously unknown bugs.",
+    "relationship_to_sqlancer": "Measures itself directly against SQLancer, reporting more triggered rewrite rules and 15 more bugs than SQLancer in 24 hours.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/ase63991.2025.00151",
+    "source_sha256": "sha256:375b5fb5baba2302778222f91e22f866f201a17d803ac42e5a3c8490d8ff2f24",
+    "supporting_excerpts": [
+      "However, due to the diversity of rewrite rules, they cover only a limited subset of rewrite scenarios, potentially overlooking critical bugs.In this paper, we propose Abstract Rule Guided (ARG) fuzzing to detect bugs in query rewrites.",
+      "The key idea is to use feedback from abstract rules to guide query generation, thereby activating more rewriting logic and enhancing bug detection.",
+      "In 24 hours, ARG triggered 76% and 1017% more written rules, triggered 13 and 15 more bugs than SQLsmith and SQLancer, respectively."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/ase63991.2025.00151",
+          "source_type": "paper_citation_context",
+          "excerpt": "Recent works have also adopted metamorphic testing to construct semantically equivalent queries to test DBMSs. NoREC [18] converts an optimizable query into a non-optimizable form of the query, then identifies logic bugs by comparing their execution results.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/ase63991.2025.00151",
+          "source_type": "paper_citation_context",
+          "excerpt": "Similarly, TLP [19] employs ternary logic to generate three equivalent queries combined via UNION operations.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00322.json b/_data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00322.json
new file mode 100644
index 0000000..c279242
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_ase63991_2025_00322.json
@@ -0,0 +1,93 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/ase63991.2025.00322",
+    "title": "LLM-based Dynamic Differential Testing for Database Connectors with Reinforcement Learning-Guided Prompt Selection",
+    "authors": [
+      "C. Lyu",
+      "Minghao Zhao",
+      "Yanhao Wang",
+      "Liang Jie"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1109/ase63991.2025.00322",
+    "url": "https://doi.org/10.1109/ase63991.2025.00322"
+  },
+  "summary": {
+    "text": "This paper applies LLM-generated test cases and reinforcement learning to differential testing of database connectors, whose vulnerabilities the authors argue are neglected, subtle, and made worse by non-standardised implementations. A parameterised template supplies the domain knowledge the model otherwise lacks; generated cases are run across multiple connectors and compared, and reinforcement learning selects the best prompt each round from behavioural feedback. On MySQL Connector/J and OceanBase Connector/J it reported 16 bugs, 10 officially confirmed and the rest acknowledged as unsafe implementations.",
+    "relationship_to_sqlancer": "Its implementation reuses the SQLancer codebase, and it treats SQLancer's line of work as the state of the art whose fuzzing it finds ineffective for connectors.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/ase63991.2025.00322",
+    "source_sha256": "sha256:d2028b070edcbc38ccc6cb6bcd0da9302ec65dae98f4dfe43389383e60840f39",
+    "supporting_excerpts": [
+      "Moreover, non-standardized implementation of connectors leaves potential risks (i.e., unsafe implementations) but is more elusive.",
+      "As a result, existing fuzzing methods are ineffective in finding such vulnerabilities.",
+      "The LLM then generates test cases instructed by the constructed prompts, which are dynamically evaluated through differential testing across multiple connectors."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/ase63991.2025.00322",
+          "source_type": "paper_citation_context",
+          "excerpt": "We adapted SQLancer [2], a state-of-the-art JDBC-based database testing tool, to support multiple database connectors and evaluated both approaches over 100 rounds.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "manual_curation",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/ase63991.2025.00322",
+          "source_type": "paper_citation_context",
+          "excerpt": "We adapted SQLancer [2], a state-of-the-art JDBC-based database testing tool, to support multiple database connectors and evaluated both approaches over 100 rounds.",
+          "excerpt_is_verbatim": true,
+          "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+          "retrieved_at": "2026-09-06T07:08:48Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/ase63991.2025.00322",
+          "source_type": "paper_citation_context",
+          "excerpt": "We adapted SQLancer [2], a state-of-the-art JDBC-based database testing tool, to support multiple database connectors and evaluated both approaches over 100 rounds.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Pivoted Query Synthesis (PQS) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_compsac57700_2023_00273.json b/_data/impact/paper_notes/paper_doi_10_1109_compsac57700_2023_00273.json
new file mode 100644
index 0000000..290c1b9
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_compsac57700_2023_00273.json
@@ -0,0 +1,97 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/compsac57700.2023.00273",
+    "title": "Detecting Hidden Failures of DBMS: A Comprehensive Metamorphic Relation Output Patterns Approach",
+    "authors": [
+      "M. Tang",
+      "T. Tse",
+      "Z. Zhou"
+    ],
+    "year": 2023,
+    "venue": "Annual International Computer Software and Applications Conference",
+    "doi": "10.1109/compsac57700.2023.00273",
+    "url": "https://doi.org/10.1109/compsac57700.2023.00273"
+  },
+  "summary": {
+    "text": "This paper works on the test oracle problem for large databases in three parts. First, it investigates query partitioning and ternary logic partitioning, identifies a gap between the two techniques, and proposes a disjoint partitioning approach to close it. Second, it compares disjoint partitioning with metamorphic relation output patterns and proposes an exhaustive collection of such patterns for DBMSs. Third, it applies both to OceanBase, detecting 12 hidden failures and 8 new crashes in a system already extensively tested and widely deployed.",
+    "relationship_to_sqlancer": "Builds directly on Rigger and Su's query partitioning and ternary logic partitioning, identifying a gap between them and proposing disjoint partitioning to fill it.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+    "source_sha256": "sha256:2dc755a23e1eb8d50cbacddc7e1ecaa741bf682554f31b995a19c1502dae87e8",
+    "supporting_excerpts": [
+      "Rigger and Su applied metamorphic testing through query partitioning and ternary logic partitioning techniques to alleviate the challenge.",
+      "In Part (A) of our project, we conduct an in-depth investigation and have identified a gap between the two techniques.",
+      "We propose a disjoint partitioning approach to address it."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+          "source_type": "paper_citation_context",
+          "excerpt": "Since SQL is based on ternary Boolean logic [14], we know that name = 'Alice' can be TRUE, FALSE, or NULL, falling into the resultant lists of the three partitioning queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+          "source_type": "paper_citation_context",
+          "excerpt": "D. Extension of QP and TLP to Disjoint Partitioning1 Our thorough investigation reveals that the high-level QP is too general for practical MR construction, while the low-level TLP is too specific.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+          "source_type": "paper_citation_context",
+          "excerpt": "In particular, we find that QP and TLP [8] as well as our proposed DP cover only two MROPs, namely list equality and bag equality of complete disjoint partitioning (as explained in Subsection II.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+          "source_type": "paper_citation_context",
+          "excerpt": "Rigger and Su [8] proposed query partitioning (QP) and ternary logic partitioning (TLP) for metamorphic testing of DBMS. QP is a general strategic concept that describes an MR among DBMS queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+          "source_type": "paper_citation_context",
+          "excerpt": "In 2020, Rigger and Su [8] applied MT to address the issue in DBMS testing through the query partitioning (QP) and ternary logic partitioning (TLP) techniques.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+          "source_type": "paper_citation_context",
+          "excerpt": "Revisit of QP and TLP by Rigger and Su Rigger and Su [8] proposed query partitioning (QP) and ternary logic partitioning (TLP) for metamorphic testing of DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_compsac61105_2024_00141.json b/_data/impact/paper_notes/paper_doi_10_1109_compsac61105_2024_00141.json
new file mode 100644
index 0000000..90dd9f7
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_compsac61105_2024_00141.json
@@ -0,0 +1,58 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/compsac61105.2024.00141",
+    "title": "SQLPass: A Semantic Effective Fuzzing Method for DBMS",
+    "authors": [
+      "Yu Li",
+      "Yixiao Yang",
+      "Yong Guan",
+      "Zhiping Shi",
+      "Rui Wang"
+    ],
+    "year": 2024,
+    "venue": "Annual International Computer Software and Applications Conference",
+    "doi": "10.1109/compsac61105.2024.00141",
+    "url": "https://doi.org/10.1109/compsac61105.2024.00141"
+  },
+  "summary": {
+    "text": "SQLPass addresses the difficulty that mutation-based fuzzing has generating SQL that passes a DBMS's strict syntax and semantic checks. It introduces weak semantic correlation nodes and semantic relationship tables, designs mutation operators over those nodes in the syntax tree, and selects the best operator each time to replace, delete or insert subtrees; semantic errors introduced by mutation are then corrected against a pre-extracted relationship table. On SQLite3, MySQL, MariaDB and PostgreSQL it achieved higher semantic correctness and more code coverage than the tools compared, and found four unknown bugs.",
+    "relationship_to_sqlancer": "Mutation-based SQL generation for DBMS fuzzing; SQLancer is among the cited state-of-the-art tools.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/compsac61105.2024.00141",
+    "source_sha256": "sha256:456ff14a88cf1c0ae6185103bcf639770a51c2076e862d9abc5c727d54352bbc",
+    "supporting_excerpts": [
+      "However, due to the strict syntax and semantic checks in DBMSs, existing mutation-based testing methods are difficult to generate test cases with correct syntax and semantics.",
+      "To ensure the syntax and semantic correctness of the test case generated by mutation, this paper proposes the concepts of weak semantic correlation nodes and semantic relationship tables.",
+      "In our experiment, SQLPass achieves 5.7%-94.2% higher semantic correctness than state-of-the-art tools, and explores 1.3%-52% more code coverage than other tools."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/83c63c4935d91200817bfc19905800431df70230",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_dsc55868_2022_00057.json b/_data/impact/paper_notes/paper_doi_10_1109_dsc55868_2022_00057.json
new file mode 100644
index 0000000..99b6b11
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_dsc55868_2022_00057.json
@@ -0,0 +1,82 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/dsc55868.2022.00057",
+    "title": "Fuzzing DBMS via NNLM",
+    "authors": [
+      "Yabin Li",
+      "Yuanping Nie",
+      "Xiaohui Kuang"
+    ],
+    "year": 2022,
+    "venue": "International Conference on Data Science in Cyberspace",
+    "doi": "10.1109/dsc55868.2022.00057",
+    "url": "https://doi.org/10.1109/dsc55868.2022.00057"
+  },
+  "summary": {
+    "text": "NNFuzz applies a neural network language model to DBMS fuzzing, using a sequence model to generate test cases automatically. The authors note that few studies had taken this route, and highlight that the approach works in a black-box setting. Evaluated on SQLite, the tool generated valid test cases and achieved higher code coverage than its initial training set.",
+    "relationship_to_sqlancer": "Language-model-based SQL generation for DBMS fuzzing; SQLancer is cited among existing testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/dsc55868.2022.00057",
+    "source_sha256": "sha256:899a7f48a1dc1efe496d2221c52449819c70e57e9ce9f32f008a2b51b9689799",
+    "supporting_excerpts": [
+      "Fuzzing is an effective vulnerability mining technology, but few studies utilize neural network language model (NNLM) to fuzz DBMS.",
+      "In this paper, we explore this technical roadmap and use the sequence model to automatically generate test cases to fuzz DBMS.",
+      "One of the advantages of the method is that it can effectively test the DBMS in a black-box situation."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:38:27Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/dsc55868.2022.00057",
+          "source_type": "paper_citation_context",
+          "excerpt": "Query Partitioning [21] uses ternary logical partitioning to convert the original query into three partitioned queries, and compares whether the results of the original query are consistent with the union of the partitioned queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/dsc55868.2022.00057",
+          "source_type": "paper_citation_context",
+          "excerpt": "PQS[19] selects the pivot row in advance, and then constructs the SQL query based on the pivot row.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/dsc55868.2022.00057",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoREC [20] transforms the SQL query into an equivalent form that will not be optimized by the query optimizer, and then executes the original query and the equivalent form separately.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/dsc55868.2022.00057",
+          "source_type": "paper_citation_context",
+          "excerpt": "When a correctness error is triggered, the DBMS does not crash, but the query returns incorrect query results [18, 19, 20, 21].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_girst67753_2025_11382165.json b/_data/impact/paper_notes/paper_doi_10_1109_girst67753_2025_11382165.json
new file mode 100644
index 0000000..1d2174f
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_girst67753_2025_11382165.json
@@ -0,0 +1,61 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/girst67753.2025.11382165",
+    "title": "Testing Spatial Database Engines via Spatial Equivalent Transformation",
+    "authors": [
+      "Shijie Li",
+      "Dongping Wang",
+      "Liang Liu",
+      "Keyue Yang",
+      "Lingwei Kuang"
+    ],
+    "year": 2025,
+    "venue": "2025 4th International Conference on Geographic Information and Remote Sensing Technology (GIRST)",
+    "doi": "10.1109/girst67753.2025.11382165",
+    "url": "https://doi.org/10.1109/girst67753.2025.11382165"
+  },
+  "summary": {
+    "text": "Spatial Equivalent Transformation builds a test oracle for spatial database systems, which serve GIS and navigation applications with spatial queries such as KNN and range queries and with topological analysis. SET generates equivalent variants of a seed query covering diverse spatial operations and checks that their result sets agree, and pairs this with a geometric generator that produces representative spatial test cases through geometric transformations. Implemented as Spader and evaluated on PostGIS, MySQL and MariaDB, it detected 16 logic bugs, 15 confirmed.",
+    "relationship_to_sqlancer": "Carries equivalence-preserving query transformation, the basis of SQLancer's oracles, into spatial databases.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/girst67753.2025.11382165",
+    "source_sha256": "sha256:54910d8ab2657ea334ded71b956e1780b3956728c6917209979c359bebb17117",
+    "supporting_excerpts": [
+      "Therefore, the automated detection of SDBMS logic bugs is an urgent issue that requires resolution.We propose a testing technique called Spatial Equivalent Transformation (SET).",
+      "The technique generates equivalent variants of a seed query to test diverse and complex spatial operations by checking the consistency of their result sets.",
+      "The experimental results show that the method detected a total of 16 logic bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:32:20Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/girst67753.2025.11382165",
+          "source_type": "paper_citation_context",
+          "excerpt": "Although crash bug detection [2] [3] and logic bug detection [4] [5] [6] [7] [8] are both mature for RDBMSs, SDBMSs suffer from a relative paucity of studies.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_iaecst64597_2024_11117732.json b/_data/impact/paper_notes/paper_doi_10_1109_iaecst64597_2024_11117732.json
new file mode 100644
index 0000000..f634f86
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_iaecst64597_2024_11117732.json
@@ -0,0 +1,65 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/iaecst64597.2024.11117732",
+    "title": "Detecting Logical Bugs in DBMS via Isomerism Fuzz System",
+    "authors": [
+      "Zhe Wang",
+      "Liang Liu",
+      "Ning Wang"
+    ],
+    "year": 2024,
+    "venue": "2024 6th International Academic Exchange Conference on Science and Technology Innovation (IAECST)",
+    "doi": "10.1109/iaecst64597.2024.11117732",
+    "url": "https://doi.org/10.1109/iaecst64597.2024.11117732"
+  },
+  "summary": {
+    "text": "This paper fuzzes a DBMS against itself by running multiple instances of the same system with different components and configurations — an 'isomerism system' — so that differential testing becomes possible without a second, separate DBMS. The authors argue that growing architectural complexity limits the effectiveness and coverage of conventional DBMS fuzzing. Testing popular DBMSs this way, they discovered 10 previously unknown types of bug.",
+    "relationship_to_sqlancer": "States that the system was implemented within SQLancer, extending its differential testing to configuration-varied instances of one DBMS.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/iaecst64597.2024.11117732",
+    "source_sha256": "sha256:d1fb62e0970ce0b66a4d73cdcabb746b63dffcb8dae65b4c5e8c9b6823dc7b0f",
+    "supporting_excerpts": [
+      "This paper introduces the idea of fuzzing DBMS via Isomerism System, which includes multiple instances of one DBMS using different components and configurations.",
+      "By fuzzing these isomorphic systems, the scope of classic differential testing is expanded, increasing its applicability and efficiency.",
+      "We implemented this system within SQLancer and tested it on popular DBMSs, ultimately discovering 10 previously unknown types of bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/iaecst64597.2024.11117732",
+          "source_type": "paper_citation_context",
+          "excerpt": "Rigger M. et al. [13] proposed the Non-Optimized Reference Engine Construction (NoREC), aimed at detecting optimization bugs in DBMSs by comparing optimized and non-optimized query s.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/iaecst64597.2024.11117732",
+          "source_type": "paper_citation_context",
+          "excerpt": "[14] they proposed a query partitioning differential detection method.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_iaecst68792_2025_11415166.json b/_data/impact/paper_notes/paper_doi_10_1109_iaecst68792_2025_11415166.json
new file mode 100644
index 0000000..d4345ae
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_iaecst68792_2025_11415166.json
@@ -0,0 +1,80 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/iaecst68792.2025.11415166",
+    "title": "Semantic Hint-Based Fuzzing for Time-Series Databases",
+    "authors": [
+      "Panta Kittisatra",
+      "Liang Liu"
+    ],
+    "year": 2025,
+    "venue": "2025 7th International Academic Exchange Conference on Science and Technology Innovation (IAECST)",
+    "doi": "10.1109/iaecst68792.2025.11415166",
+    "url": "https://doi.org/10.1109/iaecst68792.2025.11415166"
+  },
+  "summary": {
+    "text": "This paper adds a semantic hint-based oracle to TSGuard for testing time-series databases, arguing that existing fuzzing frameworks are built for relational DBMSs and ignore the temporal semantics and query processing peculiar to TSDBs. The oracle injects logically neutral predicates such as tautologies and redundant conditions into a baseline query and compares results differentially, so query intent is unchanged but robustness is probed. On Apache IoTDB, InfluxDB and TDengine it surfaced several optimizer-related crashes and robustness issues.",
+    "relationship_to_sqlancer": "Carries the idea of semantics-preserving query perturbation, which SQLancer's oracles rest on, into time-series databases.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+    "source_sha256": "sha256:4ea0967b957743fe11e8605681f15ddfacb1a7272b107cce6d90a69b57157171",
+    "supporting_excerpts": [
+      "However, most existing fuzzing frameworks are designed for relational DBMSs and overlook temporal semantics and query processing behaviors unique to TSDBs.",
+      "To address this gap, we present a Semantic Hint-Based Oracle, a lightweight extension to TSGuard that evaluates TSDB robustness under semantic-preserving transformations.",
+      "The oracle injects logically neutral predicates—such as tautologies and redundant conditions—into baseline queries and performs differential comparison to identify unexpected errors or behavioral divergences without altering query intent."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:31:04Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+          "source_type": "paper_citation_context",
+          "excerpt": "However, these frameworks largely assume relational semantics[6], static schemas, and Boolean logic evaluation, making them ineffective for dealing with temporal operators, continuous data ingestion, window functions, and time-aware aggregations that are fundamental in TSDBs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer pioneered systematic bug detection using oracles such as Non-Optimizing Reference Engine Construction (NoREC) and Ternary Logic Partitioning (TLP), enabling reliable detection of logic bugs even when databases do not crash [6].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+          "source_type": "paper_citation_context",
+          "excerpt": "Tools such as SQLancer[6], SQLaser[7], and mutation-based engines like AFL leverage metamorphic testing to uncover logic bugs and optimizer inconsistencies[8], [9].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+          "source_type": "paper_citation_context",
+          "excerpt": "Research on testing relational DBMS, which supports a standardized query language SQL, has evolved considerably over the past decades[13], [14] .",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icccs65393_2025_11069832.json b/_data/impact/paper_notes/paper_doi_10_1109_icccs65393_2025_11069832.json
new file mode 100644
index 0000000..763d95a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icccs65393_2025_11069832.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icccs65393.2025.11069832",
+    "title": "OptionFuzz: The Fuzzer with Coverage-Guided Dynamic Configuration Scheduling",
+    "authors": [
+      "Lang Chu",
+      "Minhuan Huang",
+      "Xiang Li",
+      "Yuanping Nie",
+      "Wenyu Zhen",
+      "Qian Yan"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Communication, Computing & Security",
+    "doi": "10.1109/icccs65393.2025.11069832",
+    "url": "https://doi.org/10.1109/icccs65393.2025.11069832"
+  },
+  "summary": {
+    "text": "OptionFuzz explores a target program's configuration space during fuzzing, on the grounds that configuration often determines which execution paths are reachable. The authors note that generation-based configuration exploration has been successful but depends on formal documentation and lacks scheduling strategies. OptionFuzz generates configurations through a Markov decision process driven by edge coverage, then schedules them using function coverage. Across nine real-world programs it explored 46.9% more execution paths than traditional fuzzers and found 17 configuration-triggered crashes.",
+    "relationship_to_sqlancer": "General configuration-aware fuzzing; SQLancer appears among the cited testing work rather than being used.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icccs65393.2025.11069832",
+    "source_sha256": "sha256:0aea14880eeb9aa06576e7f4606020e4d64b094041978dd86554d5810838837c",
+    "supporting_excerpts": [
+      "However, these approaches also face challenges, such as dependence on formal documentation and a lack of configuration scheduling strategies.",
+      "In this paper, we propose OptionFuzz, a configuration-aware fuzzing framework based on generation, designed to address some of the limitations of previous approaches.",
+      "The framework first generates configurations through a Markov decision process based on edge coverage analysis, followed by configuration scheduling guided by function coverage analysis."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icccs65393.2025.11069832",
+          "source_type": "paper_citation_context",
+          "excerpt": "The second aspect aims to expand the application of fuzzing to more targets, such as operating systems[9–12], virtual machine managers[13], network protocols[13, 14], database systems[15, 16], and even autonomous vehicles[17].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_iccste65902_2025_11138310.json b/_data/impact/paper_notes/paper_doi_10_1109_iccste65902_2025_11138310.json
new file mode 100644
index 0000000..2b31fde
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_iccste65902_2025_11138310.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/iccste65902.2025.11138310",
+    "title": "Research on Coverage Statistics in Fuzz Testing of Closed-Source DBMS",
+    "authors": [
+      "Zhongjie Li",
+      "Hao-Tian Liang",
+      "Haoyang Jia",
+      "Qingxian Wang",
+      "Yan Cao"
+    ],
+    "year": 2025,
+    "venue": "2025 International Conference on Computer Science, Technology and Engineering (ICCSTE)",
+    "doi": "10.1109/iccste65902.2025.11138310",
+    "url": "https://doi.org/10.1109/iccste65902.2025.11138310"
+  },
+  "summary": {
+    "text": "This paper addresses fuzzing closed-source DBMSs, where source instrumentation is unavailable and published vulnerability research is scarce. It proposes collecting real-time coverage by monitoring execution paths, and uses that coverage as feedback for seed scheduling during fuzzing. The resulting coverage tracker, TrCov, was applied to Oracle and SQL Server, with experiments the authors report as validating its effectiveness.",
+    "relationship_to_sqlancer": "Coverage collection for DBMS fuzzing where the source is unavailable; SQLancer is cited among DBMS testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/iccste65902.2025.11138310",
+    "source_sha256": "sha256:b6f246599b36c7ae8c71423c58a9dc9eff8b2ddbe1ef3171de55d068fd45460e",
+    "supporting_excerpts": [
+      "Currently, publicly available research on vulnerability detection for closed-source DBMSs remains relatively limited.",
+      "To achieve effective testing of closed-source DBMSs, this paper proposes a novel coverage analysis method based on execution path monitoring to collect real-time coverage data of closed-source DBMSs.",
+      "Based on this approach, a coverage tracker named TrCov for closed-source DBMSs was implemented."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:31:04Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/iccste65902.2025.11138310",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer [4] and Troc [5] employ pre-established oracles to detect logical and transaction isolation errors in DBMSs. APOLLO [6] identifies performance regression issues by comparing different DBMS versions.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icde55515_2023_00057.json b/_data/impact/paper_notes/paper_doi_10_1109_icde55515_2023_00057.json
new file mode 100644
index 0000000..ce9096d
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icde55515_2023_00057.json
@@ -0,0 +1,105 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icde55515.2023.00057",
+    "title": "Sequence-Oriented DBMS Fuzzing",
+    "authors": [
+      "Jie Liang",
+      "Yaoguang Chen",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Mingzhe Wang",
+      "Yu Jiang",
+      "Xiangdong Huang",
+      "Ting Chen",
+      "Jiashui Wang",
+      "Jiajia Li"
+    ],
+    "year": 2023,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde55515.2023.00057",
+    "url": "https://doi.org/10.1109/icde55515.2023.00057"
+  },
+  "summary": {
+    "text": "Lego generates SQL sequences with a wide range of statement types, addressing the problem that existing DBMS fuzzers reuse a limited set of predefined types and so cannot cover the hundreds of statement types in the SQL specification. Its key notion is type-affinity, the meaningful co-occurrence of statement type pairs such as INSERT and SELECT: each iteration explores statements of different types, analyses affinities using coverage feedback, and synthesises new sequences as affinities are discovered. Lego found 102 new vendor-confirmed vulnerabilities across four DBMSs, 22 with CVEs.",
+    "relationship_to_sqlancer": "Evaluated directly against SQLancer among the DBMS fuzzers it compares with, reporting 44-198% higher branch coverage.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icde55515.2023.00057",
+    "source_sha256": "sha256:65c06ba936e4ea47075e98a1d0c97bb7c51581b5abb5fb3f7452b5aad4b44834",
+    "supporting_excerpts": [
+      "The SQL specification consists of hundreds of statement types, which leads to difficulties in DBMS fuzzing: state-of-the-art works generally reuse the statements of predefined types; the limited types cannot cover the full input space and test the corresponding logic consequently.",
+      "The key idea of sequence generation is type-affinity, which indicates the meaningful occurrence of SQL type pairs (e.g., INSERT and SELECT).",
+      "The sequence-oriented fuzzing helps Lego outperform other fuzzers on branch coverage by 44%–198%."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+          "source_type": "paper_citation_context",
+          "excerpt": "Security vulnerabilities, especially memory bugs such as buffer overflow are particularly dangerous for DBMS because they might allow attackers to steal information, tamper data, crash systems, and bring heavy losses [4, 10, 32, 35, 51, 54].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+          "source_type": "paper_citation_context",
+          "excerpt": "The two metrics are used as the standard in fuzzing evaluation [8, 17, 44], and have been widely used in fuzzing works [35, 42, 54].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+          "source_type": "paper_citation_context",
+          "excerpt": "To test logic and performance bugs, many representative schemes utilize differential testing [16, 35, 39].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+          "source_type": "paper_citation_context",
+          "excerpt": "Its following works [34, 33] also apply similar strategies by building functionally equivalent queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+          "source_type": "paper_citation_context",
+          "excerpt": "In general, fuzzers could be divided into generation-based [35, 37] and mutation-based [15, 51, 54].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer [35] synthesizes queries to fetch a random row from existing tables in the target DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icde60146_2024_00011.json b/_data/impact/paper_notes/paper_doi_10_1109_icde60146_2024_00011.json
new file mode 100644
index 0000000..c2e17dd
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icde60146_2024_00011.json
@@ -0,0 +1,58 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icde60146.2024.00011",
+    "title": "TRAP: Tailored Robustness Assessment for Index Advisors via Adversarial Perturbation",
+    "authors": [
+      "Wei Zhou",
+      "Chen Lin",
+      "Xuanhe Zhou",
+      "Guoliang Li",
+      "Tianqing Wang"
+    ],
+    "year": 2024,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde60146.2024.00011",
+    "url": "https://doi.org/10.1109/icde60146.2024.00011"
+  },
+  "summary": {
+    "text": "TRAP assesses how robust index advisors are to change, arguing that these tools are evaluated mainly on static scenarios while their stability under minor workload shifts is not well studied. The authors introduce perturbation-based workloads with three perturbation constraints drawn from real scenarios, formulate generating perturbed queries as a sequence-to-sequence problem, and train TRAP in an opaque-box setting with a two-phase paradigm so it generalises across advisors. Assessing ten existing index advisors on standard benchmarks and real workloads, they find all are vulnerable to TRAP's workloads.",
+    "relationship_to_sqlancer": "Adversarially perturbs workloads to expose weaknesses in database components, citing the DBMS testing work that perturbs queries for correctness.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icde60146.2024.00011",
+    "source_sha256": "sha256:b29ee6d3a87ba31444f903bf8e1785544977f4073e92f71461806d0f09c05f0d",
+    "supporting_excerpts": [
+      "Their robustness, i.e., stable performance in dynamic scenarios (e.g., with minor workload changes), has not been well investigated.",
+      "First, we introduce perturbation-based workloads for robustness assessment and identify three typical perturbation constraints that occur in real scenarios.",
+      "Our findings reveal that these index advisors are vulnerable to the workloads generated by TRAP."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:56Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icde60146.2024.00011",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [36] derives multiple queries by partitioning the results from the original query.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icde60146_2024_00441.json b/_data/impact/paper_notes/paper_doi_10_1109_icde60146_2024_00441.json
new file mode 100644
index 0000000..a5c70ad
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icde60146_2024_00441.json
@@ -0,0 +1,65 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icde60146.2024.00441",
+    "title": "Applications and Challenges for Large Language Models: From Data Management Perspective",
+    "authors": [
+      "Meihui Zhang",
+      "Zhaoxuan Ji",
+      "Zhaojing Luo",
+      "Yuncheng Wu",
+      "Chengliang Chai"
+    ],
+    "year": 2024,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde60146.2024.00441",
+    "url": "https://doi.org/10.1109/icde60146.2024.00441"
+  },
+  "summary": {
+    "text": "A vision paper on what large language models can do for data management. The authors identify promising application categories — data generation, data transformation, data integration and data exploration — arguing that emergent abilities such as in-context learning and improved reasoning could change how these tasks are done. They then set out the challenges of adapting language models to each and sketch possible solutions.",
+    "relationship_to_sqlancer": "Discusses data generation among the applications, where SQLancer's generation work is cited.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icde60146.2024.00441",
+    "source_sha256": "sha256:cc7b8b04657a054a6880c51209f8eb28247a6683fbce9086e29252c80ad158a8",
+    "supporting_excerpts": [
+      "The emergent abilities of LLMs, e.g., in-context learning and advanced reasoning ability, have great potential to revolutionize data management.",
+      "In this paper, we first present some promising categories of data management applications where LLMs can be adapted, including data generation, data transformation, data integration, and data exploration."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icde60146.2024.00441",
+          "source_type": "paper_citation_context",
+          "excerpt": "Meanwhile, to detect the logic bugs of DBMS, we need to generate some SQL queries with semantic equivalence, which produce the same results [20].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde60146.2024.00441",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, to comprehensively detect the bugs of DBMS, it is important to feed the database with a huge number of SQL queries [20].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00180.json b/_data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00180.json
new file mode 100644
index 0000000..90d52bb
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00180.json
@@ -0,0 +1,72 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icde65706.2026.00180",
+    "title": "VIREO: Human-in-the-Loop DBMS Fuzzing with Visualization and LLM Support",
+    "authors": [
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Chi Zhang",
+      "Runpei Miao",
+      "Zhuo Su",
+      "Yu Jiang",
+      "Shuai Ma"
+    ],
+    "year": 2026,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde65706.2026.00180",
+    "url": "https://doi.org/10.1109/icde65706.2026.00180"
+  },
+  "summary": {
+    "text": "VIREO puts a human in the loop of DBMS fuzzing. The authors observe that fuzzers explore a limited state space because many DBMS features need particular sequences of statements and configurations, and that most approaches run with a single fixed configuration and no sense of the testing state, so coverage growth stalls. VIREO first uses LLMs over source and documentation to map function modules to grammar rules and configurations, then visualises module-level coverage and bug distribution so engineers can direct exploration. It found 25 previously unknown bugs across five well-tested DBMSs, all confirmed.",
+    "relationship_to_sqlancer": "DBMS fuzzing that reacts to the coverage plateau of automated approaches, SQLancer among the cited work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icde65706.2026.00180",
+    "source_sha256": "sha256:276add52247d94027a77ff198897d4244ca40c052c68a8a120bacba752dbe59c",
+    "supporting_excerpts": [
+      "However, in practice, many DBMS fuzzers explore only a limited state space, as many functionalities involve complex constraints that require specific sequences of SQL statements and configurations, which automated methods struggle to satisfy.",
+      "In this paper, we propose VIREO, a human-in-the-loop DBMS fuzzing framework with visualization and LLM support with two stages.",
+      "VIREO discovered a total of 25 previously unknown bugs, all of which were confirmed, and 10 have been fixed."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00180",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer [57] utilizes multiple test oracles [7, 58, 59, 60] to detect logic bugs, generating queries based on these oracles.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00180",
+          "source_type": "paper_citation_context",
+          "excerpt": "Although SQLancer was designed for logic bugs, it can also expose crashes and is evaluated using its default PQS oracle [60].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00224.json b/_data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00224.json
new file mode 100644
index 0000000..6066212
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00224.json
@@ -0,0 +1,149 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icde65706.2026.00224",
+    "title": "A Set-Theoretic Approach to Detecting Logic Bugs in DBMS Inner Join Optimizations",
+    "authors": [
+      "Ce Lyu",
+      "Changzheng Wei",
+      "Yanhao Wang",
+      "Jie Liang",
+      "Li Lin",
+      "Hanghang Wu",
+      "Minghao Zhao",
+      "Ying Yan",
+      "Aoying Zhou"
+    ],
+    "year": 2026,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde65706.2026.00224",
+    "url": "https://doi.org/10.1109/icde65706.2026.00224"
+  },
+  "summary": {
+    "text": "A metamorphic testing approach for logic bugs in INNER JOIN optimisation, framed in terms of set theory. For a given query it generates equivalent forms using intersection together with three semantics-preserving rewrite rules (symmetric join, asymmetric difference, and symmetric difference transformations), then compares the results of the original and rewritten queries. The implementation, JoinEquiv, found 29 previously unknown issues in MySQL, TiDB, DuckDB and Percona, of which 27 were officially confirmed.",
+    "relationship_to_sqlancer": "Query-partitioning in the same family as TLP, applied to inner joins: the paper builds a partitioning oracle over semantics-preserving rewrites and evaluates it against SQLancer's oracles.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icde65706.2026.00224",
+    "source_sha256": "sha256:b5db668ed0011dec546bb26d997341d360ef0f5bfa5368f6c9f2e6a1c938be8b",
+    "supporting_excerpts": [
+      "In this paper, we propose a metamorphic testing approach to detect DBMS bugs related to INNER JOIN optimization through the lens of set theory.",
+      "We implement this design in JoinEquiv, which serves as a testing oracle to systematically uncover logical inconsistencies in DBMS query processing by comparing the results of original and transformed queries.",
+      "Using JoinEquiv, we uncovered 29 previously unknown issues in mainstream DBMSs (MySQL, TiDB, DuckDB, and Percona), and 27 of them were officially confirmed."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+          "source_type": "paper_citation_context",
+          "excerpt": "In the SQL standard, equality comparison predicates involving NULL are evaluated to UNKNOWN and therefore excluded from INNER JOIN results, whereas the set operator treats NULL as equal at the set level and preserves them.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+          "source_type": "paper_citation_context",
+          "excerpt": "Together, these findings demonstrate that JoinEquiv can uncover a broad, heterogeneous spectrum of logic bugs originating from multiple layers of the system, including the optimizer, executor, and set-operator implementations, thereby effectively answering RQ2 .",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+          "source_type": "paper_citation_context",
+          "excerpt": "CERT [37] aims to identify performance issues caused by unexpected cardinality estimations, that is, the cases where the estimated number of result tuples significantly deviates from the actual number.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+          "source_type": "paper_citation_context",
+          "excerpt": "Recently, differential query planning (DQP) [25] adopts a similar hint-based strategy but focuses on detecting inconsistencies between different physical plans for the same query.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+          "source_type": "paper_citation_context",
+          "excerpt": "Another approach is pivoted query synthesis (PQS) [19], which validates query results by constructing auxiliary queries centered on a specific pivot row.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+          "source_type": "paper_citation_context",
+          "excerpt": "Representative frameworks include ternary logic partitioning (TLP) [20] and non-optimizing reference engine construction (NoREC) [22].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "extends_technique",
+      "title": "Extends a SQLancer technique",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2606.23294",
+          "source_type": "paper",
+          "excerpt": "We extend SQLancer to satisfy an additional precondition that all columns are NOT NULL.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it extends or adapts a SQLancer technique.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "dqp",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2606.23294",
+          "source_type": "paper",
+          "excerpt": "Comparison to Existing DBMS Testing Approaches In order to evaluate JoinEquiv’s unique ability to detect logical bugs related to INNER JOIN, we compare it with two representative metamorphic testing approaches: Ternary Logic Partitioning (TLP) and Differential Query Plans (DQP).",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00240.json b/_data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00240.json
new file mode 100644
index 0000000..450ae04
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icde65706_2026_00240.json
@@ -0,0 +1,100 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icde65706.2026.00240",
+    "title": "LLMSQLMUTATOR: LLM-Powered Test Case Generation for Database Using Bug Reports",
+    "authors": [
+      "Chenglin Tian",
+      "Chaofan Li",
+      "Yawen Li",
+      "Yingxia Shao"
+    ],
+    "year": 2026,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde65706.2026.00240",
+    "url": "https://doi.org/10.1109/icde65706.2026.00240"
+  },
+  "summary": {
+    "text": "LLM-SQLMutator mines existing bug reports to drive mutation-based DBMS testing. It takes SQL statements from reports as mutation seeds and extracts bug patterns and root causes from the same reports as guidance, then has an LLM perform syntax-aware directed mutations under that guidance. A further LLM-based validation step uses database metadata as a starting point and expands validation rules dynamically. Across six popular relational DBMSs it detected 27 confirmed bugs.",
+    "relationship_to_sqlancer": "Reuses DBMS bug reports, including those from automated testing tools, as testing material; SQLancer is cited among existing automated approaches.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icde65706.2026.00240",
+    "source_sha256": "sha256:6eb2197e1dc1c8784d0fe56a44c6c7530ca3f59f432b0a81cb46e5ae6d5affb2",
+    "supporting_excerpts": [
+      "To address these limitations, we propose LLM-SQLMutator, an innovative mutation-based automated DBMS testing tool that leverages large language models (LLMs) and bug reports.",
+      "LLMSQLMUTATOR extracts the SQL statements from the bug report as mutation seeds, and extracts bug patterns and root causes from the bug reports as mutation guidance.",
+      "Extensive evaluations across six popular relational DBMSs demonstrate the advantages of LLMSQLMutator, and it detects 27 confirmed bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+          "source_type": "paper_citation_context",
+          "excerpt": "They also may yield false positives due to the varied implementation choices of RDBMSs. Metamorphic testing is another mainstream approaches for RDB-MS testing [51], [54], [14], [12], [13], [55], [56], [15].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+          "source_type": "paper_citation_context",
+          "excerpt": "In recent years, SQLancer [11] has emerged as the most effective black-box fuzzing tool, distinguished by its adoption of three complementary oracles [14], [12], [13].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+          "source_type": "paper_citation_context",
+          "excerpt": "2) For test cases containing SELECT queries, we employ result inconsistency detection by randomly selecting one test oracle from TLP [13], NoREC [12], or CERT [44].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+          "source_type": "paper_citation_context",
+          "excerpt": "Generally, the automated DBMS testing process comprises two phases: test case generation [6], [11] and test oracle construction [12], [13], [14], [15].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+          "source_type": "paper_citation_context",
+          "excerpt": "Existing fuzzers typically restrict recursion depth to maintain syntactic correctness [15], [13], thereby failing to reach the parser’s threshold.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+          "source_type": "paper_citation_context",
+          "excerpt": "We selected the three most effective existing oracles for detecting logic bugs through result inconsistency: NoREC [12], TLP [13], and CERT [44].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icicnis64247_2024_10823213.json b/_data/impact/paper_notes/paper_doi_10_1109_icicnis64247_2024_10823213.json
new file mode 100644
index 0000000..b1f2168
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icicnis64247_2024_10823213.json
@@ -0,0 +1,54 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icicnis64247.2024.10823213",
+    "title": "Dynamic Adjustment Paradigm based on Genetic Algorithm in Database Index Optimization",
+    "authors": [
+      "Jinsong Wang"
+    ],
+    "year": 2024,
+    "venue": "2024 International Conference on IoT Based Control Networks and Intelligent Systems (ICICNIS)",
+    "doi": "10.1109/icicnis64247.2024.10823213",
+    "url": "https://doi.org/10.1109/icicnis64247.2024.10823213"
+  },
+  "summary": {
+    "text": "This paper proposes a genetic-algorithm-based model for dynamically adjusting database indexes. It argues that heuristic-rule and enumeration-based index recommendation does not cope with large-scale, high-concurrency workloads under limited computing resources and frequent load changes, and instead adjusts the index configuration in real time with an optimised convergence function. The work builds on the OtterTune workflow and is evaluated on Oracle 20c and SQL Server 2019 through disaster recovery and comparative query performance tests.",
+    "relationship_to_sqlancer": "Index tuning for performance rather than correctness testing; the link is the citation recorded in the citation graph.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icicnis64247.2024.10823213",
+    "source_sha256": "sha256:ae1f460e1a57c3352bf6d5657140b81797db5c26e773feb68c56c37e90fa8ea1",
+    "supporting_excerpts": [
+      "Based on this, this study proposes a dynamic database index optimization model based on a new genetic algorithm, aiming to solve shortcomings of existing database index recommendation methods in large-scale, high-concurrency data processing.",
+      "First, this study analyzes the limitations of traditional index recommendation methods based on heuristic rules and enumeration algorithms in environments with limited computing resources and frequent load changes, thus leading to the core problem.",
+      "By introducing a dynamic adjustment paradigm based on genetic algorithms and optimizing the convergence function, this model can adjust the index configuration in real time to improve query efficiency and database response speed."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/c30c758ad17b50b6de00ac4878065a4436e07de0",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icpc66645_2025_00021.json b/_data/impact/paper_notes/paper_doi_10_1109_icpc66645_2025_00021.json
new file mode 100644
index 0000000..4e1dba0
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icpc66645_2025_00021.json
@@ -0,0 +1,117 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icpc66645.2025.00021",
+    "title": "Sembug: Detecting Logic Bugs in Dbms Through Generating Semantic-Aware Non-Optimizing Query",
+    "authors": [
+      "Shiyang Ye",
+      "Chao Ni",
+      "Jue Wang",
+      "Qianqian Pang",
+      "Xinrui Li",
+      "Xiaodan Xu"
+    ],
+    "year": 2025,
+    "venue": "IEEE International Conference on Program Comprehension",
+    "doi": "10.1109/icpc66645.2025.00021",
+    "url": "https://doi.org/10.1109/icpc66645.2025.00021"
+  },
+  "summary": {
+    "text": "SemBug detects logic bugs by transforming queries a DBMS can optimise heavily into equivalent but less optimised forms, and comparing the results. The authors divide prior work into two groups: approaches that target optimization bugs but check only result cardinality, missing semantic errors and advanced features, and approaches that check both cardinality and semantics but handle optimization bugs poorly. SemBug aims to cover both by adding semantic analysis to non-optimizing query construction. Over 24 hours on five widely used DBMSs it found 34 unique logic bugs, 13 to 19 more than each of three state-of-the-art approaches, and 37 in total by submission, 29 verified.",
+    "relationship_to_sqlancer": "A direct descendant of NoREC: it extends non-optimizing query construction with semantic awareness, and measures itself against NoREC, TLP and Pinolo as the state of the art. The paper also observes that over 70% of the logic bugs TLP and Pinolo find are optimization-related, which is its argument for concentrating there.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "supplied PDF",
+    "written_from_url": "https://doi.org/10.1109/icpc66645.2025.00021",
+    "source_sha256": "sha256:48388f7b6d5af8eb9e55a6acb27932af6f4f41365bd475734adb78a441ad159e",
+    "supporting_excerpts": [
+      "To investigate the effectiveness of SemBug, we conduct a large-scale experiment on five widelyused DBMS systems (i.e., MySQL, TiDB, MariaDB, SQLite, and PostgreSQL) and compare it with three state-of-the-art (SOTA) approaches (i.e., Pinolo, TLP, and NoREC).",
+      "More precisely, we find that over 70% of the logic bugs found by those approaches (i.e., TLP and Pinolo) in the second group are optimization-related bugs, which indicates that we should pay more attention to these types of logic bugs.",
+      "This particular bug cannot be detected by NoREC, as it only identifies bugs resulting from inconsistent cardinality."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T15:22:55Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/aeea391e73991ef6f033917dba0f577c948eedb4",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Cardinality Estimation Restriction Testing (CERT), Differential Query Plans (DQP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/Syang111/SemBug",
+          "source_type": "github_repository",
+          "excerpt": "# SemBug",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after Sembug, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/Syang111/SemBug/blob/master/src/Sonar/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/Sonar/Randomly.java is SQLancer's Randomly.java, with the package renamed to Sonar (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/Syang111/SemBug",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "renamed_sqlancer_package",
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/Syang111/SemBug",
+          "source_type": "github_repository",
+          "excerpt": "# SemBug",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after Sembug, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/Syang111/SemBug/blob/master/src/Sonar/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/Sonar/Randomly.java is SQLancer's Randomly.java, with the package renamed to Sonar (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse43902_2021_00137.json b/_data/impact/paper_notes/paper_doi_10_1109_icse43902_2021_00137.json
new file mode 100644
index 0000000..44916b4
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse43902_2021_00137.json
@@ -0,0 +1,84 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse43902.2021.00137",
+    "title": "Data-Oriented Differential Testing of Object-Relational Mapping Systems",
+    "authors": [
+      "T. Sotiropoulos",
+      "Stefanos Chaliasos",
+      "Vaggelis Atlidakis",
+      "Dimitris Mitropoulos",
+      "D. Spinellis"
+    ],
+    "year": 2021,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse43902.2021.00137",
+    "url": "https://doi.org/10.1109/icse43902.2021.00137"
+  },
+  "summary": {
+    "text": "Cynthia is, as far as the authors know, the first systematic approach to testing Object-Relational Mapping systems. It generates random relational schemas, sets up the corresponding databases, and queries them through the APIs of the ORMs under test. Because ORMs share no common input language, queries are written in an abstract query language and translated into concrete executable ORM queries for differential comparison. Since the results depend heavily on the stored data, a solver-based approach produces records targeted at the constraints of each generated query. Cynthia found 28 bugs in five popular ORMs.",
+    "relationship_to_sqlancer": "Differential testing with generated schemas and queries, applied one layer above the DBMS; SQLancer is cited among that work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icse43902.2021.00137",
+    "source_sha256": "sha256:e23ed5d81417d3cbe2840cf42f38c430fd584524eb71c1e3371583d50c5d1645",
+    "supporting_excerpts": [
+      "We introduce, what is to the best of our knowledge, the first approach for systematically testing Object-Relational Mapping (ORM) systems.",
+      "Our approach leverages differential testing to establish a test oracle for ORM-specific bugs.",
+      "We implement our approach as a tool, called CYNTHIA, which found 28 bugs in five popular ORM systems."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:12Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse43902.2021.00137",
+          "source_type": "paper_citation_context",
+          "excerpt": "…data generation approach and its suitability for differential testing, we compare it against a simplistic approach that populates the database with random records a-priori [21], [27], i.e., it inserts data while setting up the tables, without considering the constraints of the generated queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse43902.2021.00137",
+          "source_type": "paper_citation_context",
+          "excerpt": "More recently, Rigger et al. [21] proposed the Pivoted Query Synthesis (PQS) technique for testing database engines.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse43902.2021.00137",
+          "source_type": "paper_citation_context",
+          "excerpt": "Although DBMSs share common functionality, they differ significantly from each other [21].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse43902.2021.00137",
+          "source_type": "paper_citation_context",
+          "excerpt": "In their most recent work, they propose Ternary Logic Partitioning (TLP) [38].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00024.json b/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00024.json
new file mode 100644
index 0000000..3c1b658
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00024.json
@@ -0,0 +1,58 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse48619.2023.00024",
+    "title": "A Comprehensive Study of Real-World Bugs in Machine Learning Model Optimization",
+    "authors": [
+      "Hao Guan",
+      "Ying Xiao",
+      "Jiaying Li",
+      "Yepang Liu",
+      "Guangdong Bai"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse48619.2023.00024",
+    "url": "https://doi.org/10.1109/icse48619.2023.00024"
+  },
+  "summary": {
+    "text": "The first empirical study of model optimization bugs — defects in the pruning and quantization steps that compact pre-trained machine learning models for resource-constrained platforms. The authors note that bugs in training, compiling and deployment have been studied but optimization has not, even though optimization in other complex systems such as compilers and databases is bug-prone. They collect 371 such bugs from TensorFlow and PyTorch spanning May 2019 to August 2022 and analyse their symptoms, root causes, life cycles, detection and fixes.",
+    "relationship_to_sqlancer": "Draws the analogy to optimization bugs in databases, where SQLancer's optimizer-focused oracles are cited.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icse48619.2023.00024",
+    "source_sha256": "sha256:e335aa0eff9711d52d28cb2e86c584549f81a98f415e948d2bc07be7827aa1f3",
+    "supporting_excerpts": [
+      "Similar to the optimization process in other complex systems, e.g., program compilers and databases, optimizations for ML models can contain bugs, leading to severe consequences such as system crashes and financial loss.",
+      "While bugs in training, compiling and deployment stages have been extensively studied, there is still a lack of systematic understanding and characterization of model optimization bugs (MOBs).",
+      "We collect a comprehensive dataset containing 371 MOBs from TensorFlow and PyTorch, the most extensively used open-source ML frameworks, covering the entire development time span of their optimizers (May 2019 to August 2022)."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00024",
+          "source_type": "paper_citation_context",
+          "excerpt": "Similar to the optimization task in other software that handles complex objects, such as in program compilers [20]– [22] and databases [23], ML model optimization is an error-prone process.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00101.json b/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00101.json
new file mode 100644
index 0000000..dca4015
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00101.json
@@ -0,0 +1,162 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse48619.2023.00101",
+    "title": "Detecting Isolation Bugs via Transaction Oracle Construction",
+    "authors": [
+      "Wensheng Dou",
+      "Ziyu Cui",
+      "Qianwang Dai",
+      "Jiansen Song",
+      "Dong Wang",
+      "Yu Gao",
+      "Wei Wang",
+      "Jun Wei",
+      "Lei Chen",
+      "Han Wang",
+      "Hua Zhong",
+      "Tao Huang"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse48619.2023.00101",
+    "url": "https://doi.org/10.1109/icse48619.2023.00101"
+  },
+  "summary": {
+    "text": "Troc detects isolation bugs, where a DBMS violates the transaction isolation level it claims to provide. Its idea is to decouple a transaction into independent statements and execute each against its own database view, constructed according to the claimed isolation level; any divergence between the real transaction execution and the independent statement execution reveals an isolation bug. Implemented for MySQL, MariaDB and TiDB, it found 5 previously unknown isolation bugs in their latest versions.",
+    "relationship_to_sqlancer": "Builds on SQLancer's infrastructure — the paper states that Troc's database and SQL statement generation is mainly based on SQLancer, and that the data structures and statements it supports are those SQLancer supports. It argues that SQLancer's oracles work on single SELECT statements without considering concurrent transactions, which is the gap Troc fills.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "supplied PDF",
+    "written_from_url": "https://doi.org/10.1109/icse48619.2023.00101",
+    "source_sha256": "sha256:c0391c612ef8d8c43f4f267e301e2cb659d7cec392d47d0faf0cfccdde856a49",
+    "supporting_excerpts": [
+      "Troc’s database and SQL statement generation mainly bases on SQLancer [49].",
+      "Automatic database testing approaches like SQLancer [26]–[28] detect logic bugs for single queries (i.e., SELECT statements) without considering concurrent transactions, and cannot be applied on other statements, e.g., UPDA TE statements in s12ands22.",
+      "For now, Troc can support many complex data structures and SQL statements, e.g., primary keys, indexes, various data types, and conditions that are supported by SQLancer [49]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T15:21:51Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+          "source_type": "paper_citation_context",
+          "excerpt": "Automatic database testing approaches like SQLancer [26]–[28] detect logic bugs for single queries (i.e., SELECT statements) without considering concurrent transactions, and cannot be applied on other statements, e.g., UPDATE statements in s 12 and s 22 .",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+          "source_type": "paper_citation_context",
+          "excerpt": "Existing testing techniques for DBMSs, e.g., SQLsmith [24] and SQLancer [26]–[28], cannot generate transaction test cases, and do not have a test oracle for transaction test cases.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+          "source_type": "paper_citation_context",
+          "excerpt": "For now, Troc can support many complex data structures and SQL statements, e.g., primary keys, indexes, various data types, and conditions that are supported by SQLancer [49].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+          "source_type": "paper_citation_context",
+          "excerpt": "Automatic database testing approaches [24]–[28] can support these complex features in modern DBMSs, and have been proved as an effective technique to detect bugs in DBMSs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+          "source_type": "paper_citation_context",
+          "excerpt": "We further investigate whether these 12 unique bugs can be revealed by existing approaches, e.g., SQLsmith [24], SQLancer [26]–[28], Elle [23] and Cobra [22].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+          "source_type": "paper_citation_context",
+          "excerpt": "QPG [69] utilizes query plans to guide database state mutation for detecting bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/criszy/Troc",
+          "source_type": "github_repository",
+          "excerpt": "Artifact for \"Detecting Isolation Bugs via Transaction Oracle Construction\"\n# Troc\r\n\r\nThis is the artifact for the paper \"Detecting Isolation Bugs via Transaction Oracle Construction\". \r\nSee [paper](http://www.tcse.cn/~cuiziyu20/papers/2023-icse-troc.pdf) to learn more details.",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/criszy/Troc/blob/main/src/main/java/troc/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/main/java/troc/Randomly.java is SQLancer's Randomly.java, with the package renamed to troc (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/criszy/Troc",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "renamed_sqlancer_package",
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/criszy/Troc",
+          "source_type": "github_repository",
+          "excerpt": "Artifact for \"Detecting Isolation Bugs via Transaction Oracle Construction\"\n# Troc\r\n\r\nThis is the artifact for the paper \"Detecting Isolation Bugs via Transaction Oracle Construction\". \r\nSee [paper](http://www.tcse.cn/~cuiziyu20/papers/2023-icse-troc.pdf) to learn more details.",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/criszy/Troc/blob/main/src/main/java/troc/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/main/java/troc/Randomly.java is SQLancer's Randomly.java, with the package renamed to troc (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00173.json b/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00173.json
new file mode 100644
index 0000000..565fc9e
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00173.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse48619.2023.00173",
+    "title": "Generating Test Databases for Database-Backed Applications",
+    "authors": [
+      "Cong Yan",
+      "Suman Nath",
+      "Shan Lu"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse48619.2023.00173",
+    "url": "https://doi.org/10.1109/icse48619.2023.00173"
+  },
+  "summary": {
+    "text": "DBGriller generates database states for testing database-backed applications, where tests need not just user inputs but valid database states. Two obstacles are that valid states must satisfy complicated application-determined constraints and that the state space is enormous, while each database test is slow to run. DBGriller injects minor mutations into existing states and turns part of the application under test into a stand-alone validity checker, and uses program analysis to derive a branch-projected database view that filters out states unlikely to increase branch coverage. On nine open-source applications it increased branch coverage and exposed unknown bugs.",
+    "relationship_to_sqlancer": "Generates database states for application testing rather than engine testing; SQLancer is cited among database testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icse48619.2023.00173",
+    "source_sha256": "sha256:347bb6ea44629ff0182ed536daa084a96de2ce85a77d56dbd73249c5cc37e9b5",
+    "supporting_excerpts": [
+      "To effectively test these applications, one needs to design not only user inputs but also database states, which imposes unique challenges.",
+      "We propose DBGRILLER, a tool that generates database states to facilitate thorough testing of database-backed applications.",
+      "To effectively generate valid database states, DBGRILLER strategically injects minor mutation into existing database states and transforms part of the application-under-test into a stand-alone validity checker."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00173",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer [21], [42] includes a series of techniques, including query partitioning and pivot query synthesis, to generate SQL queries paired with certain properties of query results.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00175.json b/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00175.json
new file mode 100644
index 0000000..8f9021f
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse48619_2023_00175.json
@@ -0,0 +1,158 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse48619.2023.00175",
+    "title": "Testing Database Systems via Differential Query Execution",
+    "authors": [
+      "Jiansen Song",
+      "Wensheng Dou",
+      "Ziyu Cui",
+      "Qianwang Dai",
+      "Wei Wang",
+      "Jun Wei",
+      "Hua Zhong",
+      "Tao Huang"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse48619.2023.00175",
+    "url": "https://doi.org/10.1109/icse48619.2023.00175"
+  },
+  "summary": {
+    "text": "Differential Query Execution detects logic bugs in UPDATE and DELETE queries as well as SELECT, which the authors note existing approaches leave untouched. Its core observation is that different SQL statements sharing a predicate should access the same rows: a row updated by an UPDATE with predicate φ should also be returned by a SELECT with the same φ, and if it is not, the DBMS has a logic bug. Applied to MySQL, MariaDB, TiDB, CockroachDB and SQLite it detected 50 unique bugs, 41 confirmed and 11 fixed.",
+    "relationship_to_sqlancer": "Implemented on SQLancer — the paper states DQE is implemented based on it and that its database generation is mainly adopted from it. It argues that PQS, NoREC and TLP cannot be applied to UPDATE and DELETE statements, and reports that none of its detected bugs could be found by those oracles.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "supplied PDF",
+    "written_from_url": "https://doi.org/10.1109/icse48619.2023.00175",
+    "source_sha256": "sha256:4db2bd5e6ae326da3ef4390a74390d0859ad8cc0e1a349b115ae98c5b1d26ddf",
+    "supporting_excerpts": [
+      "E V ALUATION We implement DQE based on SQLancer [27], which is implemented in Java.",
+      "Our database generation is mainly adopted from SQLancer [27].",
+      "Existing approaches to construct oracles forSELECT queries, e.g., PQS [9], NoREC [10] and TLP [11], cannot be adopted on UPDATE andDELETE queries."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T15:21:53Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+          "source_type": "paper_citation_context",
+          "excerpt": "To answer RQ2, we perform a qualitative comparison with existing approaches (i.e., PQS [9], NoREC [10] and TLP [11]) that aim to detect logic bugs in DBMSs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+          "source_type": "paper_citation_context",
+          "excerpt": "To demonstrate the sufficiency of our testing, we compare code coverage with existing works, i.e., PQS [9], NoREC [10] and TLP [11].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoREC [10] rewrites a SELECT query as an equivalent one that the DBMS cannot optimize, and compares their results.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [11] leverages the ternary property of predicate evaluation, where the evaluation result is one of TRUE , FALSE and NULL , to partition a SELECT query into three partitioning queries, whose combined query results are equal to the original query’s query result.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [11] decomposes a SELECT query into three partitioning queries, and merges these partitioning queries’ results into a combined result, which is expected to be the same as the original query’s result.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+          "source_type": "paper_citation_context",
+          "excerpt": "Database and SQL query generation have been widely explored by existing works [21]–[26], [29], [50]–[57].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/JensonSung/dqetool",
+          "source_type": "github_repository",
+          "excerpt": "Replication package for \"Testing Database Systems via Differential Query Execution\", accepted at ICSE 2023\n\n# DQETool\n\nDQETool is the implementation of differential query execution in paper.\n\n# Getting Started\n\nRequirements:\n* Java 11 or above",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/JensonSung/dqetool/blob/main/src/dqetool/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/dqetool/Randomly.java is SQLancer's Randomly.java, with the package renamed to dqetool (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/JensonSung/dqetool",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "renamed_sqlancer_package",
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/JensonSung/dqetool",
+          "source_type": "github_repository",
+          "excerpt": "Replication package for \"Testing Database Systems via Differential Query Execution\", accepted at ICSE 2023\n\n# DQETool\n\nDQETool is the implementation of differential query execution in paper.\n\n# Getting Started\n\nRequirements:\n* Java 11 or above",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/JensonSung/dqetool/blob/main/src/dqetool/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/dqetool/Randomly.java is SQLancer's Randomly.java, with the package renamed to dqetool (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00003.json b/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00003.json
new file mode 100644
index 0000000..2d0b013
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00003.json
@@ -0,0 +1,91 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse55347.2025.00003",
+    "title": "Coni: Detecting Database Connector Bugs via State-Aware Test Case Generation",
+    "authors": [
+      "Wenqian Deng",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Mingzhe Wang",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse55347.2025.00003",
+    "url": "https://doi.org/10.1109/icse55347.2025.00003"
+  },
+  "summary": {
+    "text": "CONI targets database connectors, the client libraries applications use to talk to a DBMS. The authors note that existing DBMS fuzzing focuses on SQL generation and exercises only a small subset of connector interfaces, so it does not transfer, and that the hard part is generating semantically correct cases that drive the connector through its state transitions. CONI derives a connector state model from the standard specification, generates interface call sequences covering those transitions, and compares results against a reference connector. Across five JDBC connectors it reported 44 previously unknown bugs, 34 confirmed.",
+    "relationship_to_sqlancer": "Extends differential testing to the connector layer, arguing that DBMS fuzzers such as SQLancer do not reach it.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icse55347.2025.00003",
+    "source_sha256": "sha256:577bb9050362a889533833c42a3f3c2e18878e8047d3724a1659f3e39ac8447d",
+    "supporting_excerpts": [
+      "However, existing DBMS fuzzing works cannot be directly applied to testing database connectors as they mainly focus on SQL generation and use a small subset of connector interfaces.",
+      "To address that, we propose CONI, a framework designed for detecting logic bugs of database connectors with state-aware test case generation.",
+      "In total, Coni reported 44 previously unknown bugs, of which 34 have been confirmed."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00003",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, SQL ANCER [11] generates valid SQL queries based on AST.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00003",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, SQL ANCER [11] relies on JDBC solely to execute SQL queries and retrieve results, without exploring additional functionalities like modifying configuration properties or batch execution.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00003",
+          "source_type": "paper_citation_context",
+          "excerpt": "On the one hand, most fuzzers primarily concentrate on generating effective SQL queries [10, 11, 12, 13, 14], whereas database connectors are not directly involved in the execution of these queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00003",
+          "source_type": "paper_citation_context",
+          "excerpt": "When applying fuzzing techniques to test DBMSs, the main challenge is to generate correct and effective SQL queries [11, 12, 13, 35, 36, 37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00003",
+          "source_type": "paper_citation_context",
+          "excerpt": "In addition, SQL ANCER [11] is a popular open-source tool for testing databases using JDBC.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00013.json b/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00013.json
new file mode 100644
index 0000000..ae7134a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00013.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse55347.2025.00013",
+    "title": "Janus: Detecting Rendering Bugs in Web Browsers via Visual Delta Consistency",
+    "authors": [
+      "Chijin Zhou",
+      "Quan Zhang",
+      "Bingzhou Qian",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse55347.2025.00013",
+    "url": "https://doi.org/10.1109/icse55347.2025.00013"
+  },
+  "summary": {
+    "text": "Janus detects rendering bugs in web browsers. Plain differential testing fails here because the same HTML legitimately renders differently in different browsers. The authors' oracle, Visual Delta Consistency, compares reactions rather than outputs: given two HTML files differing only by a small modification, all browsers should either render them identically or all render them differently. Janus builds such pairs and watches the change status across browsers. On Chrome, Safari and Firefox it found 31 non-crash rendering bugs, 24 confirmed.",
+    "relationship_to_sqlancer": "Constructs a metamorphic oracle where direct differential comparison does not work, the same problem SQLancer's oracles address for DBMSs.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icse55347.2025.00013",
+    "source_sha256": "sha256:a80c6307f0da626c756773099c6f8257a20b5ba679db4bd9a821519b1a3ed2c2",
+    "supporting_excerpts": [
+      "Traditional differential testing, while successful in various domains, falls short when applied to rendering bug detection because an HTML file is likely yield different rendered outcomes across different browsers.",
+      "This paper introduces Visual Delta Consistency, a test oracle to detect rendering bugs in web browsers, aiming to make rendered pages across browsers comparable.",
+      "Our key insight is that any modifications made to an HTML file should uniformly influence rendering outcomes across browsers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00013",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, in web development, efforts [24], [25] have been made to formalize layout guidance of web development into a formal language for correctness verification; and in database management systems, tools like SQLancer [30], [29], [28] validate query results using predefined equivalence rules.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00045.json b/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00045.json
new file mode 100644
index 0000000..d1caa66
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00045.json
@@ -0,0 +1,69 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse55347.2025.00045",
+    "title": "PUPPY: Finding Performance Degradation Bugs in DBMSs via Limited-Optimization Plan Construction",
+    "authors": [
+      "Zhiyong Wu",
+      "Jie Liang",
+      "Jingzhou Fu",
+      "Mingzhe Wang",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse55347.2025.00045",
+    "url": "https://doi.org/10.1109/icse55347.2025.00045"
+  },
+  "summary": {
+    "text": "PUPPY finds performance degradation bugs, where a DBMS's fully optimised plan runs slower than a plan built with only some optimisations enabled, because interactions between optimisations are complex and some cases are overlooked. PUPPY generates queries covering sequences of optimisation operations, then analyses the plan and selectively disables optimisations to build a limited-optimisation plan; if the restricted plan is faster, that indicates a bug. Across MySQL, Percona, TiDB, PolarDB and PostgreSQL it reported 62 such bugs, 54 confirmed as previously unknown.",
+    "relationship_to_sqlancer": "A performance oracle built from within a single DBMS, evaluated against the state-of-the-art DBMS performance testing tools.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icse55347.2025.00045",
+    "source_sha256": "sha256:7b27d58d388833232db5cf5f7c86efedef0a193e3638296e28abfe0df766e488",
+    "supporting_excerpts": [
+      "However, the resulting plan may sometimes perform worse than even if no optimizations were applied.",
+      "In this paper, we present PUPPY, an automated approach for detecting PDBs in DBMSs using limited-optimization plan construction.",
+      "We evaluate PUPPY on five widely-used DBMSs, namely MySQL, Percona, TiDB, PolarDB, and PostgreSQL against the state-of-the-art DBMS performance testing tools APOLLO and AMOEBA."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00045",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLsmith [41] generates queries based on built-in code that embeds the AST generation rules for the target DBMS. SQLancer [40, 39, 38] aims to find logic bugs and it generates queries based on the test oracle it builds.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00045",
+          "source_type": "paper_citation_context",
+          "excerpt": "QPG [5] gradually mutates DDL and DML statements to change database states, aiming to cover more unique query plans to cover more DBMS logic.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00183.json b/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00183.json
new file mode 100644
index 0000000..a86e243
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00183.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse55347.2025.00183",
+    "title": "ROSA: Finding Backdoors with Fuzzing",
+    "authors": [
+      "Dimitrios Kokkonis",
+      "M. Marcozzi",
+      "Emilien Decoux",
+      "Stefano Zacchiroli"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse55347.2025.00183",
+    "url": "https://doi.org/10.1109/icse55347.2025.00183"
+  },
+  "summary": {
+    "text": "ROSA detects code-level backdoors — hidden access such as hard-coded credentials planted in a program — by fuzzing. The authors argue manual auditing is hard, existing semi-automated approaches cover a limited range of programs and backdoors and need manual reverse-engineering, and that no existing fuzzing technique can recognise a backdoor being triggered at runtime. ROSA pairs AFL++ with a new metamorphic test oracle able to detect such triggers, and comes with ROSARUM, the first open benchmark for backdoor detection. It finds all 17 benchmark backdoors in about 1.5 hours.",
+    "relationship_to_sqlancer": "Builds a metamorphic oracle for a fuzzer in a security setting; SQLancer is cited as prior work on oracles for fuzzing.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icse55347.2025.00183",
+    "source_sha256": "sha256:292601365a9c14515d661b5c122721c895b0a918f43355d3a37a1f8b107bf037",
+    "supporting_excerpts": [
+      "However, current fuzzing knowledge does not offer any means to detect the triggering of a backdoor at runtime.",
+      "In this work we introduce ROSA, a novel approach (and tool) which combines a state-of-the-art fuzzer (AFL++) with a new metamorphic test oracle, capable of detecting runtime backdoor triggers.",
+      "It finds all 17 authentic or synthetic backdooors from ROSARUM in 1 h 30 on average."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:31:04Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/e13b3c2397adc3c3047750da2263b0678cf048b6",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00257.json b/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00257.json
new file mode 100644
index 0000000..eac63e9
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse55347_2025_00257.json
@@ -0,0 +1,95 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse55347.2025.00257",
+    "title": "Thanos: DBMS Bug Detection via Storage Engine Rotation Based Differential Testing",
+    "authors": [
+      "Ying Fu",
+      "Zhiyong Wu",
+      "Yuanliang Zhang",
+      "Jie Liang",
+      "Jingzhou Fu",
+      "Yu Jiang",
+      "Shanshan Li",
+      "Xiangke Liao"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse55347.2025.00257",
+    "url": "https://doi.org/10.1109/icse55347.2025.00257"
+  },
+  "summary": {
+    "text": "Thanos builds differential test oracles without needing two separate DBMSs. The authors note that finding equivalent systems with different implementations and compatible syntax takes considerable manual effort. Their insight is that one DBMS running different storage engines must still provide consistent basic storage functionality, so rotating the storage engine produces equivalent systems that should return the same results for the same SQL. The framework selects engines, extracts their equivalence information, synthesises feature-oriented test cases and compares results. Vendors confirmed 32 previously unknown bugs, 29 rated critical.",
+    "relationship_to_sqlancer": "Evaluated directly against SQLancer among the state-of-the-art fuzzers, reporting 24-116% higher branch coverage and bugs those tools missed.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icse55347.2025.00257",
+    "source_sha256": "sha256:a70c58b0e5df4d60e9b12f16078af0f55355c26f096696d3d4c6cce2fa6b6479",
+    "supporting_excerpts": [
+      "However, meticulously selecting equivalent DBMSs with diverse implementations and compatible input syntax requires huge manual efforts.",
+      "Our key insight is that a DBMS with different storage engines must provide consistent basic storage functionalities.",
+      "We evaluate Thanos on three widely used and extensively tested DBMSs, namely MySQL, MariaDB, and Percona against state-of-the-art fuzzers SQLancer, SQLsmith, and SQUIRREL."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:32:20Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00257",
+          "source_type": "paper_citation_context",
+          "excerpt": "These outcomes align with our expectations, as the test case synthesis algorithm was specifically designed to trigger a broader range of storage engine features and behaviors within the DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00257",
+          "source_type": "paper_citation_context",
+          "excerpt": "To improve the effectiveness of one specific storage engine, one could also gather other supported features from the DBMS’s official documentation and add them to the feature list.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00257",
+          "source_type": "paper_citation_context",
+          "excerpt": "Metamorphic testing in DBMS involves transforming SQL queries and verifying if the resulting output changes align with expected behavior [28, 29, 30, 31].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00257",
+          "source_type": "paper_citation_context",
+          "excerpt": "DBMS fuzzers [22, 25, 33, 36, 37, 38, 39, 40, 41, 42], automate this process, focusing on creating complex SQL queries to uncover memory safety issues.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse55347.2025.00257",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer proposes constructing functionally equivalent queries to test one DBMS [30, 31, 33].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse_companion58688_2023_00041.json b/_data/impact/paper_notes/paper_doi_10_1109_icse_companion58688_2023_00041.json
new file mode 100644
index 0000000..94ec3f5
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse_companion58688_2023_00041.json
@@ -0,0 +1,86 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse-companion58688.2023.00041",
+    "title": "Randomized Differential Testing of RDF Stores",
+    "authors": [
+      "Rui Yang",
+      "Yingying Zheng",
+      "Leile Tang",
+      "Wensheng Dou",
+      "Wei Wang",
+      "Jun Wei"
+    ],
+    "year": 2023,
+    "venue": "2023 IEEE/ACM 45th International Conference on Software Engineering: Companion Proceedings (ICSE-Companion)",
+    "doi": "10.1109/icse-companion58688.2023.00041",
+    "url": "https://doi.org/10.1109/icse-companion58688.2023.00041"
+  },
+  "summary": {
+    "text": "RD2 applies randomized differential testing to RDF stores, which serve knowledge graph and semantic web applications and answer SPARQL queries, and for which the authors say no logic bug detection tool existed. It builds an equivalent RDF graph across several stores and checks that they return the same result for a given SPARQL query. Queries are generated automatically from the SPARQL syntax and the generated graph so that they are syntactically valid and likely to return non-empty results, and result formats are unified before comparison. Across three widely used RDF stores it found 5 logic bugs.",
+    "relationship_to_sqlancer": "Differential logic-bug detection for RDF stores; SQLancer is cited among the database testing work it follows.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+    "source_sha256": "sha256:173007f8f1145a04bde603f7f72794fd998a8adb187423adb97d66c048a29f25",
+    "supporting_excerpts": [
+      "However, no available tools can detect logic bugs in RDF stores.",
+      "In this paper, we propose RD2, a Randomized Differential testing approach of RDF stores, to reveal discrepancies among RDF stores, which indicate potential logic bugs in RDF stores.",
+      "The core idea of RD2 is to build an equivalent RDF graph for multiple RDF stores, and verify whether they can return the same query result for a given SPARQL query."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+          "source_type": "paper_citation_context",
+          "excerpt": "RAGS [14], APOLLO [31] and TAQO [32] utilize differential testing for detecting bugs in RDBMSs. SQLsmith [33] is used to detect bugs causing exceptions or crashes in RDBMSs. TLP [15], NoREC [16], PQS [17] and DQE [18] develop various test oracles to detect logic bugs and optimization bugs, and have found many bugs in popular RDBMSs. DT2 [19] and Troc [20] detect transaction bugs in RDBMSs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+          "source_type": "paper_citation_context",
+          "excerpt": "…[32] utilize differential testing for detecting bugs in RDBMSs. SQLsmith [33] is used to detect bugs causing exceptions or crashes in RDBMSs. TLP [15], NoREC [16], PQS [17] and DQE [18] develop various test oracles to detect logic bugs and optimization bugs, and have found many bugs in popular…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [15], NoREC [16], PQS [17] and DQE [18] develop various test oracles to detect logic bugs and optimization bugs, and have found many bugs in popular RDBMSs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+          "source_type": "paper_citation_context",
+          "excerpt": "Many approaches [14]–[20], [25]–[33] are proposed to find bugs in RDBMSs that use SQL as a standardized query language.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icse_seip52600_2021_00042.json b/_data/impact/paper_notes/paper_doi_10_1109_icse_seip52600_2021_00042.json
new file mode 100644
index 0000000..a5d00f3
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icse_seip52600_2021_00042.json
@@ -0,0 +1,78 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icse-seip52600.2021.00042",
+    "title": "Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing",
+    "authors": [
+      "Mingzhe Wang",
+      "Zhiyong Wu",
+      "Xinyi Xu",
+      "Jie Liang",
+      "Chijin Zhou",
+      "Huafeng Zhang",
+      "Yu Jiang"
+    ],
+    "year": 2021,
+    "venue": "2021 IEEE/ACM 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP)",
+    "doi": "10.1109/icse-seip52600.2021.00042",
+    "url": "https://doi.org/10.1109/icse-seip52600.2021.00042"
+  },
+  "summary": {
+    "text": "An industry report on bringing coverage-guided fuzzing to enterprise DBMSs at Huawei and Bloomberg. The authors note that industry has used blackbox fuzzing for decades while coverage guidance has produced impressive research gains that rarely reach enterprise systems, given their complexity and distributed nature. Testing GaussDB and Comdb2 they hit challenges at all three stages — collecting precise coverage, optimising fuzzing performance, and analysing root causes — and built Ratel to address them, discovering 32, 42 and 5 unknown bugs in GaussDB, Comdb2 and PostgreSQL.",
+    "relationship_to_sqlancer": "Compares against SQLancer as one of the industrial black-box fuzzers, reporting substantially higher basic block coverage.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icse-seip52600.2021.00042",
+    "source_sha256": "sha256:5ed15fa57ac56fd367ca90ccbdcf96ee4c61680013e4cceef718115892dab470",
+    "supporting_excerpts": [
+      "However, due to the complexity and distributed nature of enterprise-level DBMSs, seldom are these researches applied to the industry.",
+      "In search of a general method to overcome these challenges, we propose Ratel, a coverage-guided fuzzer for enterprise-level DBMSs.",
+      "Compared to industrial black box fuzzers SQLsmith and SQLancer, as well as coverage-guided academic fuzzer Squirrel, Ratel covered 38.38%, 106.14%, 583.05% more basic blocks than the best results of other three fuzzers in GaussDB, PostgreSQL, and Comdb2, respectively."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:12Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icse-seip52600.2021.00042",
+          "source_type": "paper_citation_context",
+          "excerpt": "…SQLsmith [1] triggers system bugs by continuously generating random SQL queries; RAGS [2] detects logic bugs by comparing the results of a query on multiple DBMSs; SQLancer [3] detects logic bugs by constructing an invariant oracle from different angles [4], Yu Jiang is the correspondence author.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse-seip52600.2021.00042",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, its pivoted query synthesis strategy [5] generates queries of which corresponding result table is supposed to include a specific row, and if the DBMS fails to fetch the row, a logic bug is discovered.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icse-seip52600.2021.00042",
+          "source_type": "paper_citation_context",
+          "excerpt": "In addition, SQLancer [3] integrates three different strategies [4], [5], [14] to construct invirant oracle to detect logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icsip61881_2024_10671554.json b/_data/impact/paper_notes/paper_doi_10_1109_icsip61881_2024_10671554.json
new file mode 100644
index 0000000..5b1c9a3
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icsip61881_2024_10671554.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icsip61881.2024.10671554",
+    "title": "A Review of Fuzz Testing for Configuration-Sensitive Software",
+    "authors": [
+      "Lang Chu",
+      "Minhuan Huang",
+      "Xiang Li",
+      "Yuanping Nie"
+    ],
+    "year": 2024,
+    "venue": "IEEE International Conference on Signal and Image Processing",
+    "doi": "10.1109/icsip61881.2024.10671554",
+    "url": "https://doi.org/10.1109/icsip61881.2024.10671554"
+  },
+  "summary": {
+    "text": "A review of how fuzzing research handles software configuration. The premise is that configuration options both give software flexibility and set its security boundaries at runtime, and that many vulnerabilities appear only under particular configurations, so covering configuration diversity matters for reaching deep bugs. The paper surveys work on software configuration and security testing, summarises representative tools and frameworks that fuzz configurations and inputs together with their strengths and limitations, and discusses open challenges.",
+    "relationship_to_sqlancer": "A survey of configuration-aware fuzzing that cites SQLancer among the testing work reviewed.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icsip61881.2024.10671554",
+    "source_sha256": "sha256:91af643f257ae2b576b05f463197dbcd4d1e5658afa957b100133926eba5632f",
+    "supporting_excerpts": [
+      "In the practice of fuzz testing, comprehensive consideration of the diversity of software configurations is essential for expanding test coverage and uncovering deep-seated vulnerabilities triggered only under specific configurations.",
+      "This paper reviews the techniques for handling software configurations in fuzz testing research.",
+      "Then, we summarize several representative tools and frameworks capable of fuzzing both configurations and inputs simultaneously, analyzing their strengths and limitations."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icsip61881.2024.10671554",
+          "source_type": "paper_citation_context",
+          "excerpt": "…specialized fuzz testing tools, covering various critical domains such as operating systems[11– 14], virtual machine managers (e.g., HyperCube[15]), network protocols[15, 16], database systems[17, 18], and autonomous vehicles (e.g., RVFUZZER[19]), to meet the security testing needs in these areas.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icsme64153_2025_00030.json b/_data/impact/paper_notes/paper_doi_10_1109_icsme64153_2025_00030.json
new file mode 100644
index 0000000..e9776d6
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icsme64153_2025_00030.json
@@ -0,0 +1,180 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icsme64153.2025.00030",
+    "title": "TSGuard: Detecting Logic Bugs in Time Series Management Systems Via Time Series Algebra",
+    "authors": [
+      "Lingwei Kuang",
+      "Liang Liu",
+      "Wen-Jing Wang",
+      "Ning Cao",
+      "Shijie Li",
+      "Fan Liu",
+      "Haolong Chen",
+      "WenJian Liao"
+    ],
+    "year": 2025,
+    "venue": "IEEE International Conference on Software Maintenance and Evolution",
+    "doi": "10.1109/icsme64153.2025.00030",
+    "url": "https://doi.org/10.1109/icsme64153.2025.00030"
+  },
+  "summary": {
+    "text": "TSGuard detects logic bugs in time series management systems, where the authors say the problem was open because time series SQL differs enough from relational SQL, syntactically and semantically, that existing tools do not apply. It converts a time series SQL query into an equivalent time series algebra expression, evaluates that expression to obtain the expected result set, and compares it against what the system returned. A feedback mechanism for query generation and per-system syntax validators improve efficiency. TSGuard found 48 previously unknown bugs, 45 of them logic bugs.",
+    "relationship_to_sqlancer": "SQLancer is its baseline, adapted by the authors to run against time series systems: over 7,200 seconds TSGuard found 38 bugs across three systems where SQLancer found 7, and generated 4.92 times more syntax node sequences and 3.91 times more valid queries. The paper positions its oracle alongside NoREC, TLP and PQS as the established ways of building one.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "supplied PDF",
+    "written_from_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+    "source_sha256": "sha256:07ec57ded0bcddf8449c55cffd3400eaa5b375f3b574beeeabe116e2ee9bb5df",
+    "supporting_excerpts": [
+      "Comparison with Other Techniques To evaluate the effectiveness of TSGuard in detecting logic bugs in TSMSs, we adapted the open-source relational database testing tool SQLancer as a baseline for comparison.",
+      "The results show that TSGuard detected 38 bugs across the three TSMSs, whereas SQLancer detected only 7.",
+      "Additionally, TSGuard generated 4.92 times more syntax node sequences and 3.91 times more valid queries than SQLancer, while generating only 38.36% as many invalid queries."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T15:21:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+          "source_type": "paper_citation_context",
+          "excerpt": "For instance, PQS [12] synthesizes a query Q based on an expected result set (i.e., the pivot row) and verifies whether the actual result set is a superset of the expected result set to detect logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-07T15:10:19Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+          "source_type": "paper_citation_context",
+          "excerpt": "NOREC [10] transforms an original query into a non-optimized but semantically equivalent SQL query, detecting logic bugs by comparing their execution results.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-07T15:10:19Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, NoREC [10] detects logic bugs in database optimizers by mutating the original query into a non-optimized query and comparing their result sets.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-07T15:10:19Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+          "source_type": "paper_citation_context",
+          "excerpt": "This technique has been widely applied to both RDBMSs and GDBMSs. PQS [12] randomly selects specific rows from the database and synthesizes test cases, verifying that the results of these test cases contain the selected rows to detect logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-07T15:10:19Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+          "source_type": "paper_citation_context",
+          "excerpt": "Similarly, TLP [11] mutates the original query into multiple partitioned queries, reassembles their result sets, and verifies consistency with the original result set to detect logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-07T15:10:19Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+          "source_type": "paper_citation_context",
+          "excerpt": "QPG [28] guides database state mutation using query plans and applies logic bug oracles to identify logic bugs across various database states.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-07T15:10:19Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/LingweiKuang/TSGuard",
+          "source_type": "github_repository",
+          "excerpt": "TSGuard: Detecting Logic Bugs in Time Series Management Systems via Time Series Algebra\n# TSGuard-Detecting-Logic-Bugs-in-Time-Series-Management-Systems-via-Time-Series-Algebra\nTSGuard: Detecting Logic Bugs in Time Series Management Systems via Time Series Algebra",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/LingweiKuang/TSGuard/blob/main/TSGuard/tsFuzzy/src/main/java/com/fuzzy/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "TSGuard/tsFuzzy/src/main/java/com/fuzzy/Randomly.java is SQLancer's Randomly.java, with the package renamed to com.fuzzy (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+          "source_type": "paper",
+          "excerpt": "Comparison with Other Techniques To evaluate the effectiveness of TSGuard in detecting logic bugs in TSMSs, we adapted the open-source relational database testing tool SQLancer as a baseline for comparison.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-07T15:10:19Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/LingweiKuang/TSGuard",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "renamed_sqlancer_package",
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/LingweiKuang/TSGuard",
+          "source_type": "github_repository",
+          "excerpt": "TSGuard: Detecting Logic Bugs in Time Series Management Systems via Time Series Algebra\n# TSGuard-Detecting-Logic-Bugs-in-Time-Series-Management-Systems-via-Time-Series-Algebra\nTSGuard: Detecting Logic Bugs in Time Series Management Systems via Time Series Algebra",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/LingweiKuang/TSGuard/blob/main/TSGuard/tsFuzzy/src/main/java/com/fuzzy/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "TSGuard/tsFuzzy/src/main/java/com/fuzzy/Randomly.java is SQLancer's Randomly.java, with the package renamed to com.fuzzy (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_icst60714_2024_00012.json b/_data/impact/paper_notes/paper_doi_10_1109_icst60714_2024_00012.json
new file mode 100644
index 0000000..9e5a4bd
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_icst60714_2024_00012.json
@@ -0,0 +1,90 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/icst60714.2024.00012",
+    "title": "Differential Optimization Testing of Gremlin-Based Graph Database Systems",
+    "authors": [
+      "Yingying Zheng",
+      "Wensheng Dou",
+      "Leile Tang",
+      "Ziyu Cui",
+      "Jiansen Song",
+      "Ziyue Cheng",
+      "Wei Wang",
+      "Jun Wei",
+      "Hua Zhong",
+      "Tao Huang"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Information Control Systems & Technologies",
+    "doi": "10.1109/icst60714.2024.00012",
+    "url": "https://doi.org/10.1109/icst60714.2024.00012"
+  },
+  "summary": {
+    "text": "Differential Optimization Testing finds optimization bugs in Gremlin-based graph database systems by running the same Gremlin query under two different optimization configurations on the same system and checking that both produce the same results; a difference indicates an optimization bug. An optimization-guided approach steers testing toward more optimization strategies and more graph features. Across Neo4j, OrientDB, JanusGraph, HugeGraph, TinkerGraph and ArcadeDB it found 28 unique optimization bugs, 16 confirmed as previously unknown.",
+    "relationship_to_sqlancer": "Varies optimization settings to build an oracle within one system, the same strategy SQLancer's optimizer-focused oracles use, applied to graph databases.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/icst60714.2024.00012",
+    "source_sha256": "sha256:f43f0cf0f0d6b57f7c2980fb322c86e17e9db4c3471d6c6f50313bc1e2a8ab83",
+    "supporting_excerpts": [
+      "In this paper, we propose Differential Optimization Testing (DOT), an effective and automated approach to detect optimization bugs in GDBs that adopt Gremlin as their query language.",
+      "The main idea of DOT is that, given a Gremlin query $Q$, we execute it on the target GDB with two different optimization configurations and then verify whether they can compute the same query results for query $Q$.",
+      "In total, we have found 28 unique optimization bugs, 16 of which have been confirmed as previously-unknown bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/icst60714.2024.00012",
+          "source_type": "paper_citation_context",
+          "excerpt": "Some approaches [20], [21], [25] (e.g., NoREC [21]) utilizes metamorphic testing for finding logic bugs in relational DBMSs. Query generation [17], [54] (e.g., SQLsmith [17]) and generic fuzzing approaches (e.g., AFL [53]) can also be used to detect bugs in relational DBMSs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icst60714.2024.00012",
+          "source_type": "paper_citation_context",
+          "excerpt": "GDBMeter applies query partitioning [20] to test GDBs. Specifically, it partitions a query into three disjoint queries in which a predicate is evaluated to true , false and null , respectively.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icst60714.2024.00012",
+          "source_type": "paper_citation_context",
+          "excerpt": "Among these approaches, NoREC [21] can detect optimization bugs in relational database systems by rewriting an optimized SQL query into a non-optimizing SQL query.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/icst60714.2024.00012",
+          "source_type": "paper_citation_context",
+          "excerpt": "Many testing approaches [17]–[26] (e.g., TLP [20], NoREC [21], and DQE [25]) have been proposed to test relational database systems.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_ictai62512_2024_00085.json b/_data/impact/paper_notes/paper_doi_10_1109_ictai62512_2024_00085.json
new file mode 100644
index 0000000..f8c774f
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_ictai62512_2024_00085.json
@@ -0,0 +1,61 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/ictai62512.2024.00085",
+    "title": "NNTailor: A Neural Network-Driven Fuzzer for DataBase Management Systems",
+    "authors": [
+      "Shutao Chu",
+      "Yongjun Wang",
+      "Haoran Xu",
+      "Zhiyuan Jiang",
+      "Yongxin Chen"
+    ],
+    "year": 2024,
+    "venue": "IEEE International Conference on Tools with Artificial Intelligence",
+    "doi": "10.1109/ictai62512.2024.00085",
+    "url": "https://doi.org/10.1109/ictai62512.2024.00085"
+  },
+  "summary": {
+    "text": "NNTailor applies neural network language models over AST code fragments to DBMS fuzzing, an approach the authors say has seen limited research. They highlight its suitability for black-box testing scenarios, and evaluate it on SQLite and PostgreSQL, showing it can generate effective SQL test cases.",
+    "relationship_to_sqlancer": "Language-model-driven SQL generation for DBMS fuzzing; SQLancer is cited among existing testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/ictai62512.2024.00085",
+    "source_sha256": "sha256:df2c09488540e9c402c96d7eea8e05a5690856eb2881ae6680daea0eaefaf88d",
+    "supporting_excerpts": [
+      "While fuzz testing(fuzzing) is a prevalent technique for uncovering DBMS vulnerabilities, there is limited research on employing Neural Network Language Models (NNLMs) for this purpose.",
+      "This paper introduces NNTailor, a novel approach using NNLMs based on AST code fragments for DBMS fuzzing.",
+      "A key advantage of this method is its effectiveness in black-box testing scenarios."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/ictai62512.2024.00085",
+          "source_type": "paper_citation_context",
+          "excerpt": "We conducted evaluations of NNTailor on the latest versions of PostgreSQL and SQLite, comparing its performance with SQLsmith[5] and SQLancer[6, 25, 26, 27].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_iemcon67450_2025_11381190.json b/_data/impact/paper_notes/paper_doi_10_1109_iemcon67450_2025_11381190.json
new file mode 100644
index 0000000..e44f525
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_iemcon67450_2025_11381190.json
@@ -0,0 +1,54 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/iemcon67450.2025.11381190",
+    "title": "A Comparative Survey of Mutation Testing Across Large Language Models, REST APIs, and Database Engines",
+    "authors": [
+      "B. Falah",
+      "Hayat Routaib"
+    ],
+    "year": 2025,
+    "venue": "IEEE Annual Information Technology, Electronics and Mobile Communication Conference",
+    "doi": "10.1109/iemcon67450.2025.11381190",
+    "url": "https://doi.org/10.1109/iemcon67450.2025.11381190"
+  },
+  "summary": {
+    "text": "A survey comparing how mutation testing is applied across three domains — large language models, REST APIs and database engines — based on 30 peer-reviewed studies. It compares mutation operator design, tool support and evaluation methods in each. The authors report that mutation testing for LLMs is still early-stage and focused on semantic and fairness faults, REST API work benefits from adaptive schema-based tools, and database engine work uses grammar- and query-plan-guided mutation.",
+    "relationship_to_sqlancer": "A survey covering database engine testing, where SQLancer's query-plan-guided work is among the cited studies.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/iemcon67450.2025.11381190",
+    "source_sha256": "sha256:f9c382de405534487b7225016f6726d1ae31cb97dffc75236a7ecd0f540203da",
+    "supporting_excerpts": [
+      "This paper surveys its application across three domains, Large Language Models (LLMs), REST APIs, and Database Engines, through analysis of 30 peer-reviewed studies.",
+      "Results show that LLM-focused mutation testing re-mains early-stage with emphasis on semantic and fairness-aware faults, REST APIs benefit from adaptive schema-based tools, and database engines exploit grammar-and query plan–guided mutations to reveal semantic and performance issues."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/iemcon67450.2025.11381190",
+          "source_type": "paper_citation_context",
+          "excerpt": "Ba et al. [16] proposed query plan–guided mutation strategies, while Schafer et al. developed Parser-Knows-Best [17], using grammar-rule traversal for test input mutation.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_issrew55968_2022_00056.json b/_data/impact/paper_notes/paper_doi_10_1109_issrew55968_2022_00056.json
new file mode 100644
index 0000000..34cdf84
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_issrew55968_2022_00056.json
@@ -0,0 +1,82 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/issrew55968.2022.00056",
+    "title": "A Disjoint-Partitioning Approach to Enhancing Metamorphic Testing of DBMS",
+    "authors": [
+      "M. Tang",
+      "T. Tse",
+      "Z. Zhou"
+    ],
+    "year": 2022,
+    "venue": "2022 IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW)",
+    "doi": "10.1109/issrew55968.2022.00056",
+    "url": "https://doi.org/10.1109/issrew55968.2022.00056"
+  },
+  "summary": {
+    "text": "A short paper proposing disjoint partitioning as an extension of existing metamorphic testing for DBMSs, motivated by the oracle problem that makes execution results hard to verify against expected outcomes. The authors apply it in an empirical case study on OceanBase and report uncovering various hidden failures and crashes in a system already extensively tested and widely used in industry.",
+    "relationship_to_sqlancer": "States directly that it extends Rigger and Su's metamorphic testing work, adding a disjoint partitioning approach.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/issrew55968.2022.00056",
+    "source_sha256": "sha256:e9205d901e69622b77e6eee40631d6ee8739e67dafe6b6d2945575ff495f5697",
+    "supporting_excerpts": [
+      "Owing to big data, DBMS testing faces the oracle problem, that is, it is difficult to verify execution results against expected outcomes.",
+      "Rigger and Su applied metamorphic testing to alleviate the challenge.",
+      "We propose a disjoint-partitioning approach to extend their work."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/issrew55968.2022.00056",
+          "source_type": "paper_citation_context",
+          "excerpt": "They proposed a general concept called query partitioning (QP) [5] for revealing DBMS failures.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/issrew55968.2022.00056",
+          "source_type": "paper_citation_context",
+          "excerpt": "Rigger and Su further proposed ternary logic partitioning (TLP) [5] as a specific case of QP.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/issrew55968.2022.00056",
+          "source_type": "paper_citation_context",
+          "excerpt": "In 2020, Rigger and Su applied MT to alleviate the oracle problem in DBMS testing [4][5][6].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/issrew55968.2022.00056",
+          "source_type": "paper_citation_context",
+          "excerpt": "In 2020, Rigger and Su [4][5][6] applied MT to tackle the issue in DBMS testing.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_missf68264_2026_11521893.json b/_data/impact/paper_notes/paper_doi_10_1109_missf68264_2026_11521893.json
new file mode 100644
index 0000000..0f9fbf3
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_missf68264_2026_11521893.json
@@ -0,0 +1,53 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/missf68264.2026.11521893",
+    "title": "An Automated Risk Scoring Framework for SQL Execution Plan Analysis and Performance Regression Detection in Oracle Database Systems",
+    "authors": [
+      "Raghu Gollapudi"
+    ],
+    "year": 2026,
+    "venue": "2026 International Conference on Multidisciplinary Innovations For Smart & Sustainable Future (MISSF)",
+    "doi": "10.1109/missf68264.2026.11521893",
+    "url": "https://doi.org/10.1109/missf68264.2026.11521893"
+  },
+  "summary": {
+    "text": "A framework for scoring the risk of SQL execution plans in Oracle Database installations. It combines resource consumption, execution efficiency, plan regression, bind variable health and plan quality into a weighted score on a 0-100 scale, and runs as a PL/SQL agent that watches dynamic performance views and workload history to catch silent plan changes as they happen. The authors report validation in enterprise Oracle environments.",
+    "relationship_to_sqlancer": "Performance regression monitoring in production rather than logic-bug testing, and not built on SQLancer.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/missf68264.2026.11521893",
+    "source_sha256": "sha256:45783ff9464963f6b1516aabee70f024f8954953133ca9f4ab47ad6f2d76b955",
+    "supporting_excerpts": [
+      "This paper suggests a proactive and automated risk scoring system in systematic evaluation of SQL execution plans and performance regression of Oracle Database systems in early stages of implementation.",
+      "The framework proposes a new multi-factor weighted scoring model, which measures SQL risk on a normalized 0 100 scale using a combination of resource consumption data, execution efficiency data, plan regression data, bind variable health, and execution plan quality data."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/missf68264.2026.11521893",
+          "source_type": "paper_citation_context",
+          "excerpt": "These systems however tend to show raw metrics without merging them into a risk model [14].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_punecon67554_2025_11378586.json b/_data/impact/paper_notes/paper_doi_10_1109_punecon67554_2025_11378586.json
new file mode 100644
index 0000000..222801c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_punecon67554_2025_11378586.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/punecon67554.2025.11378586",
+    "title": "AskDB: AI-Assisted Natural Language Interface for Database Querying and Visualization",
+    "authors": [
+      "S. Shilaskar",
+      "Prapti Dongaonkar",
+      "Pruthavik Gavali",
+      "Suhani Gunje"
+    ],
+    "year": 2025,
+    "venue": "2025 IEEE Pune Section International Conference (PuneCon)",
+    "doi": "10.1109/punecon67554.2025.11378586",
+    "url": "https://doi.org/10.1109/punecon67554.2025.11378586"
+  },
+  "summary": {
+    "text": "AskDB is a natural-language interface to databases. It translates a user's question into SQL with a large language model, runs it against the connected database, and returns results as a table and a chart. The system adds schema-driven query correction and prompt-engineering optimisation to improve translation accuracy, with a React frontend and a Python backend.",
+    "relationship_to_sqlancer": "A text-to-SQL application rather than DBMS testing work; the connection is the citation recorded in the citation graph.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/punecon67554.2025.11378586",
+    "source_sha256": "sha256:53edba626ea197bcb859d3443db658bb68d0613e8e37d29762589798e793a53a",
+    "supporting_excerpts": [
+      "This system proposed AskDB, an intelligent neural text-to-SQL system that enables users to use natural language queries as input in order to interact with a database.",
+      "The system automatically translates user queries into valid SQL and executes them against the connected database, returning results to the user in tabular form and easily interpreted graphical visualization."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/5c51d8d0d0446261f16acc8503d8cf2c7db0b75e",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_saner60148_2024_00096.json b/_data/impact/paper_notes/paper_doi_10_1109_saner60148_2024_00096.json
new file mode 100644
index 0000000..6f31812
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_saner60148_2024_00096.json
@@ -0,0 +1,39 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/saner60148.2024.00096",
+    "title": "Testing Constraint Checking Implementations via Principled Metamorphic Transformations",
+    "authors": [
+      "Mingchen Gao",
+      "Huiyan Wang",
+      "Chang Xu"
+    ],
+    "year": 2024,
+    "venue": null,
+    "doi": "10.1109/saner60148.2024.00096",
+    "url": "https://doi.org/10.1109/saner60148.2024.00096"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://openalex.org/W4400680628",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "OpenAlex records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_sp54263_2024_00109.json b/_data/impact/paper_notes/paper_doi_10_1109_sp54263_2024_00109.json
new file mode 100644
index 0000000..c0af955
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_sp54263_2024_00109.json
@@ -0,0 +1,110 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/sp54263.2024.00109",
+    "title": "Chronos: Finding Timeout Bugs in Practical Distributed Systems by Deep-Priority Fuzzing with Transient Delay",
+    "authors": [
+      "Yuanliang Chen",
+      "Fuchen Ma",
+      "Yuanhang Zhou",
+      "Ming Gu",
+      "Qing Liao",
+      "Yu Jiang"
+    ],
+    "year": 2024,
+    "venue": "IEEE Symposium on Security and Privacy",
+    "doi": "10.1109/sp54263.2024.00109",
+    "url": "https://doi.org/10.1109/sp54263.2024.00109"
+  },
+  "summary": {
+    "text": "Chronos finds timeout bugs in distributed systems, where incorrect timeout handling can hang or crash a system. It injects fine-grained delays through general runtime delayed libraries, uses deep-priority guided fuzzing to generate delay sequences reaching bugs on deep paths, and applies transient delays so the test does not pay the wall-clock cost of really waiting. On ZooKeeper, MySQL-Cluster, HDFS and Go-Ethereum it covered 15-26% more timeout mechanism logic than random, brute-force and coverage-guided fault injection, and found 27 timeout bugs, all repaired.",
+    "relationship_to_sqlancer": "Uses SQLancer as a workload generator rather than as an oracle: the paper states that the MySQL-Cluster workload is generated by SQLancer, describing it as one of the widely used SQL generators for testing database systems. The bugs it reports are timeout bugs found by its own delay injection, not logic bugs found by a SQLancer oracle.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "supplied PDF",
+    "written_from_url": "https://doi.org/10.1109/sp54263.2024.00109",
+    "source_sha256": "sha256:b3afdfed7cf6e16968e8900c8ba15313169791bc151112fb71a064b5b6bb5914",
+    "supporting_excerpts": [
+      "For MySQL-Cluster, the workload is generated by SQLancer [57], one of the widely used SQL generators for testing database systems."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T15:21:50Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/sp54263.2024.00109",
+          "source_type": "paper_citation_context",
+          "excerpt": "For MySQL-Cluster, the workload is generated by SQLancer [57], one of the widely used SQL generators for testing database systems.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/SecTechTool/Chronos",
+          "source_type": "github_repository",
+          "excerpt": "# Chronos: An Automatical Testing Framework for Finding Timeout Bugs in Distributed Systems by Self-Adaptive Delay Model.",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after Chronos, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/SecTechTool/Chronos/blob/main/workload/mysql-cluster/sqlancer/src/sqlancer/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "workload/mysql-cluster/sqlancer/src/sqlancer/Randomly.java is SQLancer's Randomly.java (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/SecTechTool/Chronos",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/SecTechTool/Chronos",
+          "source_type": "github_repository",
+          "excerpt": "# Chronos: An Automatical Testing Framework for Finding Timeout Bugs in Distributed Systems by Self-Adaptive Delay Model.",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after Chronos, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/SecTechTool/Chronos/blob/main/workload/mysql-cluster/sqlancer/src/sqlancer/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "workload/mysql-cluster/sqlancer/src/sqlancer/Randomly.java is SQLancer's Randomly.java (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_srds69199_2025_00038.json b/_data/impact/paper_notes/paper_doi_10_1109_srds69199_2025_00038.json
new file mode 100644
index 0000000..451713b
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_srds69199_2025_00038.json
@@ -0,0 +1,55 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/srds69199.2025.00038",
+    "title": "Diverse Database Replication Based on Snapshot Isolation – Performance Implications of Improved Dependability",
+    "authors": [
+      "Peter Popov",
+      "Vladimir Stankovic"
+    ],
+    "year": 2025,
+    "venue": "IEEE International Symposium on Reliable Distributed Systems",
+    "doi": "10.1109/srds69199.2025.00038",
+    "url": "https://doi.org/10.1109/srds69199.2025.00038"
+  },
+  "summary": {
+    "text": "This paper evaluates the performance cost of replicating a database across diverse DBMS implementations. Its premise is that most replication schemes assume crash failures, whereas studies have shown many faults in relational DBMSs cause systematic non-crash failures, which same-DBMS replication cannot mask. The authors built a middleware replication protocol, DivRep, deployed over diverse database servers as DivSQL, providing strict snapshot isolation under an incorrect-results failure model, and measure the overhead across three diverse DBMSs.",
+    "relationship_to_sqlancer": "Concerns fault tolerance against incorrect results rather than finding their causes; SQLancer is cited among work on DBMS faults.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/srds69199.2025.00038",
+    "source_sha256": "sha256:dc1a1cb214d67ec574de4d951630c341403d2babce6ec4918087f080379bad1f",
+    "supporting_excerpts": [
+      "Conversely, using diverse DBMSs is a suitable way of protecting against non-crash failures.",
+      "We have built a middleware-based database replication protocol, DivRep, and deployed it with diverse database servers (DivSQL), for improved fault tolerance.",
+      "DivSQL provides strict Snapshot Isolation (SI) guarantees, and assumes “incorrect results” failure model (IRFM) – the most realistic one based on the extensive experimental analyses of DBMS faults ([1], [2])."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/srds69199.2025.00038",
+          "source_type": "paper_citation_context",
+          "excerpt": "In addition to crash failures, it guards against failures of SELECTs (e.g., erroneous omission of a row) – referred to by some as “logic bugs” [19], and DELETEs, INSERTs and UPDATEs wrongly changing the database state.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_tdsc_2024_3521591.json b/_data/impact/paper_notes/paper_doi_10_1109_tdsc_2024_3521591.json
new file mode 100644
index 0000000..ea9628a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_tdsc_2024_3521591.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/tdsc.2024.3521591",
+    "title": "Improving Multitasking DBMS Fuzzing With More Accurate Coverage and Testcase Trimming",
+    "authors": [
+      "Jiaqi Li",
+      "Yajin Zhou",
+      "Lei Wu"
+    ],
+    "year": 2025,
+    "venue": "IEEE Transactions on Dependable and Secure Computing",
+    "doi": "10.1109/tdsc.2024.3521591",
+    "url": "https://doi.org/10.1109/tdsc.2024.3521591"
+  },
+  "summary": {
+    "text": "Tuzz improves coverage-guided DBMS fuzzing on two fronts. First, coverage feedback in existing DBMS fuzzers is imprecise, so the fuzzer cannot choose strategies well; work-task coverage tracking and unstable edge filtering make it more accurate at low instrumentation cost. Second, DBMS fuzzers lack testcase trimming, so inputs grow, execution slows and mutations are less likely to reach interesting structures. With more accurate coverage, trimming becomes practical. Tuzz explored 16-27% more edges than the state-of-the-art fuzzer and found 14 previously unknown bugs in MySQL and MariaDB.",
+    "relationship_to_sqlancer": "Coverage-guided DBMS fuzzing, a complementary line to SQLancer's oracle-based approach, which it cites.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/tdsc.2024.3521591",
+    "source_sha256": "sha256:a8ac48bb8ae04fb4f7703d238d60d0a6e85ca276a4a18bcbdb40c89695558650",
+    "supporting_excerpts": [
+      "First, the coverage feedback is imprecise which prevents fuzzers from making optimal decisions on fuzzing strategies.",
+      "Specifically, the work-task coverage tracking and unstable edge filtering improve the coverage accuracy with low instrumentation overhead.",
+      "More importantly, Tuzz has discovered 10 and 4 previously unknown bugs in MySQL and MariaDB."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/tdsc.2024.3521591",
+          "source_type": "paper_citation_context",
+          "excerpt": "Thesecondtypeproposesnewtestoraclestoidentifysemantic bugs in DBMSs. NoREC [35], TLP [36], DQE [37], Pinolo [38] leverage the idea of metamorphic testing to identify semantic bugs that lead to the incorrect result of SQL statement.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_tkde_2026_3656491.json b/_data/impact/paper_notes/paper_doi_10_1109_tkde_2026_3656491.json
new file mode 100644
index 0000000..57c503a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_tkde_2026_3656491.json
@@ -0,0 +1,108 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/tkde.2026.3656491",
+    "title": "Effective Bug Detection in Graph Database Engines: An LLM-Based Approach",
+    "authors": [
+      "Jiayi Wu",
+      "Zhengyu Wu",
+      "Xunkai Li",
+      "Ronghua Li",
+      "Hongchao Qin",
+      "Guoren Wang"
+    ],
+    "year": 2026,
+    "venue": "IEEE Transactions on Knowledge and Data Engineering",
+    "doi": "10.1109/tkde.2026.3656491",
+    "url": "https://doi.org/10.1109/tkde.2026.3656491"
+  },
+  "summary": {
+    "text": "DGDB uses a large language model to generate queries for testing graph database engines, then applies differential testing to find bugs. The motivation is that existing approaches are tied to a single graph query language and need substantial prior knowledge to generate queries. Applied to engines using Cypher, Gremlin and SPARQL it found 23 previously unknown wrong-result bugs, and the authors report at least a 20.41% higher non-empty-result query ratio and more than three times as many bugs as prior methods on Cypher engines.",
+    "relationship_to_sqlancer": "Differential testing for graph databases, compared against prior bug-detection methods in a line of work that cites SQLancer.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/tkde.2026.3656491",
+    "source_sha256": "sha256:2880095e9f0243a7c83e28ff3f8605778f6d06391557191e39da98c5152846ba",
+    "supporting_excerpts": [
+      "To address these challenges, we introduce DGDB, a novel paradigm harnessing large language models (LLM), such as ChatGPT, for comprehensive bug detection in graph database engines.",
+      "DGDB achieves at least 20.41% improvement in the non-empty-result query ratio and detects more than three times as many bugs as existing state-of-the-art methods on Cypher-based graph database engines, with further significant gains when employing more advanced LLM."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/tkde.2026.3656491",
+          "source_type": "paper_citation_context",
+          "excerpt": "PQS [35] selects a target data from randomly generated tables, generates conditional expressions based on the target data, constructs an SQL query with a where or join clause, and determines the presence of bugs by checking if the result is included in the result set.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tkde.2026.3656491",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [37] transforms randomly generated original SQL queries into three different logical queries based on the true, false, and null ternary logic.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tkde.2026.3656491",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoRec [36] converts the original SQL query into a non-optimized SQL query and compares the results of these two SQL queries for consistency.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2402.00292",
+          "source_type": "paper_citation_context",
+          "excerpt": "PQS[38] selects a target data from randomly generated tables, generates conditional expressions based on the target data, constructs an SQL query with a where or join clause, and determines the presence of bugs by checking if the result is included in the result set.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2402.00292",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP[37] transforms randomly generated original SQL queries into three different logical queries based on the true, false, and null ternary logic.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2402.00292",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoRec[36] converts the original SQL query into a non-optimized SQL query and compares the results of these two SQL queries for consistency.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2402.00292",
+          "source_type": "paper_citation_context",
+          "excerpt": "The former typically involves detecting bugs in graph database engines by generating equivalent queries[5] or utilizing predicate partitioning[23, 24, 37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_tse_2025_3574328.json b/_data/impact/paper_notes/paper_doi_10_1109_tse_2025_3574328.json
new file mode 100644
index 0000000..a0c2d9b
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_tse_2025_3574328.json
@@ -0,0 +1,97 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/tse.2025.3574328",
+    "title": "SQLaw: Detecting Bugs in GPU Database Management Systems via Rule-Based Differential Execution",
+    "authors": [
+      "Jiaxin Hu",
+      "Rongxin Wu"
+    ],
+    "year": 2025,
+    "venue": "IEEE Transactions on Software Engineering",
+    "doi": "10.1109/tse.2025.3574328",
+    "url": "https://doi.org/10.1109/tse.2025.3574328"
+  },
+  "summary": {
+    "text": "SQLaw tests GPU-based DBMSs, which the authors say had received no systematic bug-detection effort despite growing interest. It combines offline rule learning, which automatically extracts differential execution rules used to guide synthesis of configurations and queries, with an online interpreter that mutates those statements to generalise them. Evaluated on three major GPU DBMSs it outperformed state-of-the-art approaches by up to 2.22x and detected 51 previously unknown GPU-related bugs, 37 confirmed or fixed.",
+    "relationship_to_sqlancer": "Differential DBMS testing specialised to GPU engines; SQLancer is among the approaches it compares against.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/tse.2025.3574328",
+    "source_sha256": "sha256:19bf988eb19d697cbd28937510d6122c84e8a0a437bdc9bd1c40a360ba5e31c0",
+    "supporting_excerpts": [
+      "Despite the growing interest in GPU DBMSs and the inherent presence of bugs, there has been no systematic effort, to our knowledge, to detect bugs in GPU DBMSs.",
+      "To this end, we design SQLaw, an innovative and comprehensive framework that combines offline rule learning with an online interpreter incorporating mutation for efficient and general GPU-related bug detection.",
+      "Additionally, SQLaw detected 51 previously unknown GPU-related bugs, of which 37 have been confirmed or fixed by developers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:31:04Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+          "source_type": "paper_citation_context",
+          "excerpt": "Inspired by the widely adopted differential testing approach for bug detection in non-GPU DBMSs [13], [14], [15], [16] and GPU-related software [17], [18], we borrow its core concept of comparing the results of identical test cases executed on different software systems to establish a test oracle…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+          "source_type": "paper_citation_context",
+          "excerpt": "To detect correctness bugs, Grand [13] and RD2 [67] employ randomized differential testing for multiple graph DBMSs and RDF stores, respectively; THANOS [14] compares test cases on differential storage engines; and DQP [16] leverages differential query plans for simple yet effective bug detection.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+          "source_type": "paper_citation_context",
+          "excerpt": "• NoREC [52] translates query statements into non-optimized queries in the optimizer and compares the results to identify logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+          "source_type": "paper_citation_context",
+          "excerpt": "To uncover logic bugs in relational DBMSs, NoREC [52] verifies the equivalence between original queries and their non-optimized translations; PINOLO [62] employs the approximation relation; TQS [61] utilizes join optimization relations; and EET [64] constructs equivalent transformations through…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+          "source_type": "paper_citation_context",
+          "excerpt": "• TLP [53] partitions an original query into three separated queries by decomposing its predicate, and triggers a logic bug when the union result of these separated queries mis-matches the original one.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+          "source_type": "paper_citation_context",
+          "excerpt": "Many studies [52], [53], [61], [62], [64], [65], [66] employ metamorphic testing to evaluate DBMSs. Metamorphic testing verifies SQL statement results against predefined rules to detect bugs [69].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1109_tse_2025_3625300.json b/_data/impact/paper_notes/paper_doi_10_1109_tse_2025_3625300.json
new file mode 100644
index 0000000..e70023d
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1109_tse_2025_3625300.json
@@ -0,0 +1,105 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1109/tse.2025.3625300",
+    "title": "A Comprehensive Study of Bugs in Relational DBMS",
+    "authors": [
+      "Shuang Liu",
+      "Ruifeng Wang",
+      "Yuanfeng Xie",
+      "Junjie Chen",
+      "Wei Lu",
+      "Xiao Zhang",
+      "Quanqing Xu",
+      "Chuanhui Yang",
+      "Xiaoyong Du"
+    ],
+    "year": 2025,
+    "venue": "IEEE Transactions on Software Engineering",
+    "doi": "10.1109/tse.2025.3625300",
+    "url": "https://doi.org/10.1109/tse.2025.3625300"
+  },
+  "summary": {
+    "text": "An empirical study of 777 manually examined bugs in MySQL, SQLite and openGauss, analysed along four dimensions: root causes, symptoms, distribution across modules, and the correlations between these. The authors also study the SQL statements that trigger the bugs in order to identify test cases existing tools cannot generate, and report 12 findings, singling out data types and complex features such as triggers, procedures and parameter settings as areas where detection could improve. Their resulting tool, SQLT, found eight bugs, all verified by developers.",
+    "relationship_to_sqlancer": "Studies bugs that existing DBMS testing tools cannot generate test cases for, which is where SQLancer is cited.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1109/tse.2025.3625300",
+    "source_sha256": "sha256:70e3c3ca6920ed8445b38280536221b684665950e19d433c9154cfa93b1b48a1",
+    "supporting_excerpts": [
+      "This study presents the first comprehensive analysis of bugs in three popular open-source RDBMSs—MySQL, SQLite, and openGauss.",
+      "We also analyzed the bug-triggering SQL statements to uncover test cases that cannot be generated by existing tools.",
+      "Leveraging these insights, we developed a tool, SQLT, which effectively identified eight RDBMS bugs (five type-related), all verified by developers, with four subsequently fixed."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+          "source_type": "paper_citation_context",
+          "excerpt": "We’ve also compared SQLT with SQLancer on the latest version of MySQL and SQLite (MySQL 8.0.34 and SQLite 3.43.0) for 24 hours and adopt NoREC [14] as the default oracle to detect bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+          "source_type": "paper_citation_context",
+          "excerpt": "…related to SQL types often result in result inconsistencies—silent bugs without explicit error messages— we incorporated two existing metamorphic testing approaches, NOREC [14] and TLP [19], as oracles within SQLT. SQLT is publicly available and can be accessed at https://github.com/ sqlttest/SQLT.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+          "source_type": "paper_citation_context",
+          "excerpt": "Bug Type Duration # Bugs Fonseca [22] concurrency bugs 2003-2009 80 Cui [23] transaction bugs 2018-2022 140 ours general bugs 2021-2023 777 [12], [13], [14], [15], [16], [17], [18], [19], [20], [21] aiming at detecting bugs in RDBMSs, and they successfully detected new bugs in RDBMSs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+          "source_type": "paper_citation_context",
+          "excerpt": "We first consulted the DB-Engines Ranking [24] to identify ranked and widely adopted RDBMS, and surveyed existing testing approaches [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [25], [26] as well as empirical studies on RDBMS bugs [23], [27].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+          "source_type": "paper_citation_context",
+          "excerpt": "Rigger et al. [12], [14], [19] proposed SQLancer, which is one of the most popular generation-based RDBMS testing approach, and it incorporates three metamorphic methods, namely PQS [12], NoREC [14], and TLP [19], as oracles to detect logical bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLT is an extension of the widely-used database testing tool, SQLancer [12].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1117_12_3006402.json b/_data/impact/paper_notes/paper_doi_10_1117_12_3006402.json
new file mode 100644
index 0000000..a295e5a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1117_12_3006402.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1117/12.3006402",
+    "title": "An empirical study on configuration-related branch statement in database management system",
+    "authors": [
+      "Xie Li",
+      "Huiping Zhou",
+      "Haifang Zhou",
+      "Jingying Zhang"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Modelling, Identification and Control",
+    "doi": "10.1117/12.3006402",
+    "url": "https://doi.org/10.1117/12.3006402"
+  },
+  "summary": {
+    "text": "An empirical study of configuration-related branch statements in DBMSs, motivated by the observation that these systems expose thousands of configuration options whose complexity and interdependencies are a major cause of failure, and that prior work neither detects configuration-related bugs nor characterises the code involved. Using taint analysis the authors build a dataset of 347 such branch statements across MySQL, SQLite and PostgreSQL, finding that 53.6% involve configuration variables alone (93.5% of those a single variable), 46.4% mix configuration and program variables, and 17.9% are complex enough to contain function calls.",
+    "relationship_to_sqlancer": "Studies configuration-dependent DBMS code; SQLancer is cited among DBMS testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1117/12.3006402",
+    "source_sha256": "sha256:bf2752ffe0e063a63f141e4396e6827831aa967afca6a2fa9e1ddbae2b29b06a",
+    "supporting_excerpts": [
+      "Previous research neither specifically detects configuration-related bugs nor studies the characteristics of configuration-related code in DBMSs.",
+      "In this paper, we undertake one of the first attempts to conduct a real-world configuration-related branch statements characteristic study in DBMSs.",
+      "We first use a taint analysis tool to construct a data set of 347 real world configuration-related branch statements including 100 in MySQL, 98 in SQLite and 149 in PostgreSQL, which are very popular."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1117/12.3006402",
+          "source_type": "paper_citation_context",
+          "excerpt": "Malicious attackers may use these vulnerabilities to steal user information, embezzle data, crash the system, and bring unpredictable economic losses[7, 8, 9, 10].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1142_s021819402150039x.json b/_data/impact/paper_notes/paper_doi_10_1142_s021819402150039x.json
new file mode 100644
index 0000000..4e27e1a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1142_s021819402150039x.json
@@ -0,0 +1,91 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1142/s021819402150039x",
+    "title": "MTKeras: An Automated Metamorphic Testing Platform",
+    "authors": [
+      "Ye-Lin Liu",
+      "Z. Zhou",
+      "T. Chen",
+      "Yang Liu",
+      "Dave Towey"
+    ],
+    "year": 2021,
+    "venue": "International journal of software engineering and knowledge engineering",
+    "doi": "10.1142/s021819402150039x",
+    "url": "https://doi.org/10.1142/s021819402150039x"
+  },
+  "summary": {
+    "text": "MTKeras is an automated, domain-independent platform for metamorphic testing. The authors demonstrate it through five case studies across four domains — image classification, sentiment analysis, search engines and database management systems — and additionally study whether combining metamorphic relation input patterns within individual relations improves failure finding. Their experiments support both the combination of patterns and the use of the platform, and the work introduces metamorphic relation patterns as a practical tool for the research community.",
+    "relationship_to_sqlancer": "Applies its metamorphic testing platform to database management systems among four domains, citing the DBMS metamorphic testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1142/s021819402150039x",
+    "source_sha256": "sha256:1b1ed1b3a228314cdfdb862717dbec165c06de13ca33208b54008fd595811510",
+    "supporting_excerpts": [
+      "This paper presents an automated, domain-independent, metamorphic testing platform called MTKeras.",
+      "In this paper, we report on an investigation demonstrating the effectiveness and usability of MTKeras through five case studies in the four domains of image classification, sentiment analysis, search engines and database management systems.",
+      "We also report on the effectiveness of combining metamorphic relation (input) patterns in individual metamorphic relations, enhancing the failure-finding abilities of the individual relations."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:53Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1142/s021819402150039x",
+          "source_type": "paper_citation_context",
+          "excerpt": "Researchers developed a novel method called NonOptimizing Reference Engine Construction (NoREC) [6], which compares the results of the optimized and non-optimized versions of a SQL to evaluate the DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1142/s021819402150039x",
+          "source_type": "paper_citation_context",
+          "excerpt": "It has become recognized as an important quality assurance paradigm for complex systems, such as for machine learning (ML) [4], [5] and database management systems (DBMSs) [6].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1142/s021819402150039x",
+          "source_type": "paper_citation_context",
+          "excerpt": "MTKeras was also configured for database management system (DBMS) testing by using the concept of Non-Optimizing Reference Engine Construction (NoREC) [6].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1142/s021819402150039x",
+          "source_type": "paper_citation_context",
+          "excerpt": "There are multiple methods for detecting bugs in a DBMS, such as Pivoted Query Synthesis (PQS) [47] and Random Generation of SQL (RAGS) [48].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1142/s021819402150039x",
+          "source_type": "paper_citation_context",
+          "excerpt": "Database management system testing is also challenged by the test oracle problem [6].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3372297_3417260.json b/_data/impact/paper_notes/paper_doi_10_1145_3372297_3417260.json
new file mode 100644
index 0000000..911067c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3372297_3417260.json
@@ -0,0 +1,67 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3372297.3417260",
+    "title": "SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback",
+    "authors": [
+      "Rui Zhong",
+      "Yongheng Chen",
+      "Hong Hu",
+      "Hangfan Zhang",
+      "Wenke Lee",
+      "Dinghao Wu"
+    ],
+    "year": 2020,
+    "venue": "Conference on Computer and Communications Security",
+    "doi": "10.1145/3372297.3417260",
+    "url": "https://doi.org/10.1145/3372297.3417260"
+  },
+  "summary": {
+    "text": "Squirrel combines language validity with coverage feedback for DBMS fuzzing. Mutation-based fuzzers fail against DBMSs because inputs are checked strictly for syntax and semantics, while generation-based testing guarantees syntax but ignores feedback such as code coverage. Squirrel keeps SQL in a structural intermediate representation, performs type-based mutations — statement insertion, deletion and replacement — on it, and analyses logical dependencies between arguments to avoid semantic errors. It found 63 bugs across SQLite, MySQL and MariaDB, 52 fixed with 12 CVEs.",
+    "relationship_to_sqlancer": "A coverage-guided DBMS fuzzer frequently used as a baseline alongside SQLancer, which it cites.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3372297.3417260",
+    "source_sha256": "sha256:64c1eeb231ae521a1f0630853e465cf52b4659bc7d8e692b7d67ebbe867fff0a",
+    "supporting_excerpts": [
+      "However, current mutation-based fuzzers cannot effectively test database management systems (DBMSs), which strictly check inputs for valid syntax and semantics.",
+      "Generation-based testing can guarantee the syntax correctness of the inputs, but it does not utilize any feedback, like code coverage, to guide the path exploration.",
+      "In our experiment, Squirrel achieves 2.4×-243.9× higher semantic correctness than state-of-the-art fuzzers, and explores 2.0×-10.9× more new edges than mutation-based tools."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:53Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3372297.3417260",
+          "source_type": "paper_citation_context",
+          "excerpt": "DBMSs have been heavily tested for logic and performance defects [53, 56, 61, 64].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3372297.3417260",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer constructs queries to fetch a randomly selected row from a table [53].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3395032_3395322.json b/_data/impact/paper_notes/paper_doi_10_1145_3395032_3395322.json
new file mode 100644
index 0000000..a6ca67d
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3395032_3395322.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3395032.3395322",
+    "title": "Testing query execution engines with mutations",
+    "authors": [
+      "Xinyue Chen",
+      "Chenglong Wang",
+      "Alvin Cheung"
+    ],
+    "year": 2020,
+    "venue": "DBTest@SIGMOD",
+    "doi": "10.1145/3395032.3395322",
+    "url": "https://doi.org/10.1145/3395032.3395322"
+  },
+  "summary": {
+    "text": "MutaSQL finds correctness bugs in SQL execution engines by mutating a query Q over a database D into a query Q' that should return the same result on D, then comparing what the engine returns for each. The motivation is that validating a query execution engine is inherently hard given the complexity of optimizers, and that the cost of testing slows development iteration and lets bugs reach production. Evaluated on SQLite, it reproduced 34 known bugs from earlier versions and found a new one in the then-current release.",
+    "relationship_to_sqlancer": "An equivalent-query mutation oracle for SQL engines, the same idea as SQLancer's transformation-based oracles, which it cites.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3395032.3395322",
+    "source_sha256": "sha256:1d10d3b844cda046f7732af2066bf6a14ad5c578f35b048a3ab6ffd743c7c080",
+    "supporting_excerpts": [
+      "However, due to the complex nature of query optimizers, validating the correctness of a query execution engine is inherently challenging.",
+      "To address this challenge, we propose a tool, MutaSQL, that can quickly discover correctness bugs in SQL execution engines.",
+      "MutaSQL generates test cases by mutating a query Q over database D into a query Q′ that should evaluate to the same result as Q on D."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:53Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3395032.3395322",
+          "source_type": "paper_citation_context",
+          "excerpt": "Meanwhile, pivot query synthesis (PQS) [3] is a new testing algorithm that also utilizes equivalence mutation so that the mutated query contains a specific row in the original output.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3428261.json b/_data/impact/paper_notes/paper_doi_10_1145_3428261.json
new file mode 100644
index 0000000..d9d46a6
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3428261.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3428261",
+    "title": "On the unusual effectiveness of type-aware operator mutations for testing SMT solvers",
+    "authors": [
+      "Dominik Winterer",
+      "Chengyu Zhang",
+      "Zhendong Su"
+    ],
+    "year": 2020,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3428261",
+    "url": "https://doi.org/10.1145/3428261"
+  },
+  "summary": {
+    "text": "Type-aware operator mutation tests SMT solvers by mutating operators of conforming types within seed formulas to produce well-typed mutants used as test cases. Realised in OpFuzz and applied to Z3 and CVC4, it proved unusually effective: over a year the authors reported 1,092 bugs, of which 819 were confirmed and 685 fixed, spanning soundness bugs, invalid model bugs and crashes across many logics and configurations. A follow-up study found the bugs were high quality, often affecting core components; 184 of the confirmed bugs were soundness bugs and 489 occurred in the solvers' default modes.",
+    "relationship_to_sqlancer": "Mutation-based testing for solvers; SQLancer is cited among the testing work it relates to.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3428261",
+    "source_sha256": "sha256:c2ba0ca274e41fe2cdf7d4111aa0dcfcd999763342ffbadb9dcb077a8b1c97d8",
+    "supporting_excerpts": [
+      "We propose type-aware operator mutation, a simple, but unusually effective approach for testing SMT solvers.",
+      "Type-aware operator mutations are unusually effective: During one year of extensive testing with OpFuzz, we reported 1092 bugs on Z3’s and CVC4’s respective GitHub issue trackers, out of which 819 unique bugs were confirmed and 685 of the confirmed bugs were fixed by the developers.",
+      "Among the 819 confirmed bugs found by OpFuzz,184 were soundness bugs, the most critical bugs in SMT solvers,and 489 were in the default modes of the solvers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:53Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/ab6addb4b50a1af3f78fea402980648135d3b8c9",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3448016_3457559.json b/_data/impact/paper_notes/paper_doi_10_1145_3448016_3457559.json
new file mode 100644
index 0000000..0004397
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3448016_3457559.json
@@ -0,0 +1,74 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3448016.3457559",
+    "title": "FoundationDB: A Distributed Unbundled Transactional Key Value Store",
+    "authors": [
+      "Jingyu Zhou",
+      "Meng Xu",
+      "A. Shraer",
+      "B. Namasivayam",
+      "Alex Miller",
+      "Evan Tschannen",
+      "Steve Atherton",
+      "Andrew J. Beamon",
+      "Rusty Sears",
+      "J. Leach",
+      "D. Rosenthal",
+      "X. Dong",
+      "Willie B. Wilson",
+      "Ben Collins",
+      "David Scherer",
+      "Alec Grieser",
+      "Young Liu",
+      "Alvin Moore",
+      "Bhaskar Muppana",
+      "Xi-sheng Su",
+      "Vishesh Yadav"
+    ],
+    "year": 2021,
+    "venue": "SIGMOD Conference",
+    "doi": "10.1145/3448016.3457559",
+    "url": "https://doi.org/10.1145/3448016.3457559"
+  },
+  "summary": {
+    "text": "A system paper on FoundationDB, an open-source transactional key-value store combining NoSQL flexibility and scalability with ACID transactions. It uses an unbundled architecture separating an in-memory transaction management system, a distributed storage system and a distributed configuration system, each provisioned independently. Notably it integrates a deterministic simulation framework used to test every new feature under a wide range of faults, which the authors credit for the system's stability and rapid release cadence. FoundationDB underpins cloud infrastructure at Apple, Snowflake and elsewhere.",
+    "relationship_to_sqlancer": "A database system built around its own deterministic testing framework; SQLancer is cited among database testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3448016.3457559",
+    "source_sha256": "sha256:485de73e5454945303c8596808959263b123f67ba328621f3e410bae9037ece4",
+    "supporting_excerpts": [
+      "FoundationDB adopts an unbundled architecture that decouples an in-memory transaction management system, a distributed storage system, and a built-in distributed configuration system.",
+      "FoundationDB uniquely integrates a deterministic simulation framework, used to test every new feature of the system under a myriad of possible faults.",
+      "This rigorous testing makes FoundationDB extremely stable and allows developers to introduce and release new features in a rapid cadence."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:12Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3448016.3457559",
+          "source_type": "paper_citation_context",
+          "excerpt": "Finally there are numerous approaches to testing the correctness of database subsystems in the absence of faults, including the query engine [17, 57, 60] and concurrency control mechanism [63].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3468264_3468540.json b/_data/impact/paper_notes/paper_doi_10_1145_3468264_3468540.json
new file mode 100644
index 0000000..1d57a7b
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3468264_3468540.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3468264.3468540",
+    "title": "Skeletal approximation enumeration for SMT solver testing",
+    "authors": [
+      "Peisen Yao",
+      "Heqing Huang",
+      "Wensheng Tang",
+      "Qingkai Shi",
+      "Rongxin Wu",
+      "Charles Zhang"
+    ],
+    "year": 2021,
+    "venue": "ESEC/SIGSOFT FSE",
+    "doi": "10.1145/3468264.3468540",
+    "url": "https://doi.org/10.1145/3468264.3468540"
+  },
+  "summary": {
+    "text": "Skeletal approximation enumeration is a lightweight, general technique for testing SMT solvers across all first-order theories. The authors argue existing approaches are either too costly or hard to generalise across solvers and theories because of the test oracle problem. Applying the technique to Z3 and CVC4, two comprehensively tested state-of-the-art solvers, they had found 71 confirmed bugs by the time of writing, 55 of them already fixed.",
+    "relationship_to_sqlancer": "Addresses the test oracle problem for solvers; SQLancer is cited among work solving it in other domains.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3468264.3468540",
+    "source_sha256": "sha256:ed511a41da770e77a14f675ba3984336338347404b7c7844861fe0bcd4a2da71",
+    "supporting_excerpts": [
+      "Existing approaches to testing SMT solvers are either too costly or find difficulties generalizing to different solvers and theories, due to the test oracle problem.",
+      "To complement existing approaches and overcome their weaknesses, this paper introduces skeletal approximation enumeration (SAE), a novel lightweight and general testing technique for all first-order theories.",
+      "By the time of writing, our approach had found 71 confirmed bugs in Z3 and CVC4,55 of which had already been fixed."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:53Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3468264.3468540",
+          "source_type": "paper_citation_context",
+          "excerpt": "in many application domains such as bioinformatics [26], web services [24], compilers [36, 51], debuggers [53], databases [47], machine learning-based systems [31, 38], model counters [55], and SMT solvers [60].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3468264_3468573.json b/_data/impact/paper_notes/paper_doi_10_1145_3468264_3468573.json
new file mode 100644
index 0000000..ad4d7e9
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3468264_3468573.json
@@ -0,0 +1,58 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3468264.3468573",
+    "title": "Metamorphic testing of Datalog engines",
+    "authors": [
+      "Muhammad Numair Mansur",
+      "M. Christakis",
+      "Valentin Wüstholz"
+    ],
+    "year": 2021,
+    "venue": "ESEC/SIGSOFT FSE",
+    "doi": "10.1145/3468264.3468573",
+    "url": "https://doi.org/10.1145/3468264.3468573"
+  },
+  "summary": {
+    "text": "The first metamorphic testing approach for query bugs in Datalog engines — cases where evaluating a query returns incorrect results. The authors stress that because Datalog underpins program analyses, such bugs can compromise the soundness of an analysis implemented and formalised in it. Running their tool on three mature engines they found 13 previously unknown query bugs, some deep and revealing critical semantic issues.",
+    "relationship_to_sqlancer": "Carries metamorphic testing, established for SQL by SQLancer, into Datalog engines.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3468264.3468573",
+    "source_sha256": "sha256:14770d4af29f4b7ba996f7e124a8996121c600b516bacf87d3e00886686cba2f",
+    "supporting_excerpts": [
+      "The most critical ones manifest as incorrect results when evaluating queries—we refer to these as query bugs.",
+      "Given the wide applicability of the language, query bugs may have detrimental consequences, for instance, by compromising the soundness of a program analysis that is implemented and formalized in Datalog.",
+      "In this paper, we present the first metamorphic-testing approach for detecting query bugs in Datalog engines."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:12Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/13cf48695d23f7e019d670a1bf059b7f28396992",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3471485_3471491.json b/_data/impact/paper_notes/paper_doi_10_1145_3471485_3471491.json
new file mode 100644
index 0000000..de5fa47
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3471485_3471491.json
@@ -0,0 +1,53 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3471485.3471491",
+    "title": "Technical Perspective DIAMetrics",
+    "authors": [
+      "P. Boncz"
+    ],
+    "year": 2021,
+    "venue": "SIGMOD record",
+    "doi": "10.1145/3471485.3471491",
+    "url": "https://doi.org/10.1145/3471485.3471491"
+  },
+  "summary": {
+    "text": "A technical perspective on DIAMetrics. It observes that benchmarking database systems has a long and successful history, both in making industrial systems comparable and as a cornerstone of quantifiable experimental research, and that creating good benchmarks has been described as something of an art — one can draw dataset and workload design from representative use cases informed by domain experts, and also from database architects' insight into which features, operations and data distributions matter.",
+    "relationship_to_sqlancer": "A perspective on database benchmarking; SQLancer is cited among the workload generation work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3471485.3471491",
+    "source_sha256": "sha256:6e063d593bcd9003cf7c54aa19a0c97174bbac0321637702bfbdd88ab48c06c0",
+    "supporting_excerpts": [
+      "Benchmarking database systems has a long and successful history in making industrial database systems comparable, and is also a cornerstone of quantifiable experimental data systems research.",
+      "Creating good benchmarks has been described as something of an art [3]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:53Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3471485.3471491",
+          "source_type": "paper_citation_context",
+          "excerpt": "It inspired me to think about novel directions, possibly taking automatic benchmark extraction from performance monitoring accountability also towards correctness testing: one could envision enriching workload summarization with new dimensions such as code coverage [5] and automatic generation of query correctness oracles [4].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3485529.json b/_data/impact/paper_notes/paper_doi_10_1145_3485529.json
new file mode 100644
index 0000000..30ee253
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3485529.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3485529",
+    "title": "Generative type-aware mutation for testing SMT solvers",
+    "authors": [
+      "Jiwon Park",
+      "Dominik Winterer",
+      "Chengyu Zhang",
+      "Zhendong Su"
+    ],
+    "year": 2021,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3485529",
+    "url": "https://doi.org/10.1145/3485529"
+  },
+  "summary": {
+    "text": "Generative Type-Aware Mutation tests SMT solvers by mutating expressions rooted with parametric operators from the SMT-LIB specification, making generation a form of mutation. It is a hybrid of mutation-based and grammar-based fuzzing with an infinite mutation space, which the authors present as overcoming a major limitation of OpFuzz, the previous state of the art for SMT solvers. Their tool TypeFuzz reported over 237 bugs in Z3 and CVC4, 189 confirmed and 176 fixed, including 18 soundness bugs in CVC4's default mode, seven of them two years latent.",
+    "relationship_to_sqlancer": "Mutation-based testing for solvers rather than DBMSs; SQLancer is cited among the testing work it draws on.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3485529",
+    "source_sha256": "sha256:a10b35959f6e19ec6fa1cffb94b4be5e61b573a15e97f19d96956934fe8dde44",
+    "supporting_excerpts": [
+      "We propose Generative Type-Aware Mutation, an effective approach for testing SMT solvers.",
+      "Generative Type-Aware Mutation is a hybrid of mutation-based and grammar-based fuzzing and features an infinite mutation space—overcoming a major limitation of OpFuzz, the state-of-the-art fuzzer for SMT solvers.",
+      "During our testing period with TypeFuzz, we reported over 237 bugs in the state-of-the-art SMT solvers Z3 and CVC4."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:12Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/3f650c0d0a046c1fb6c068d636b138e14ded77b9",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3510003_3510093.json b/_data/impact/paper_notes/paper_doi_10_1145_3510003_3510093.json
new file mode 100644
index 0000000..7765976
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3510003_3510093.json
@@ -0,0 +1,136 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3510003.3510093",
+    "title": "Automatic Detection of Performance Bugs in Database Systems using Equivalent Queries",
+    "authors": [
+      "Xinyu Liu",
+      "Qi Zhou",
+      "Joy Arulraj",
+      "A. Orso"
+    ],
+    "year": 2022,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3510003.3510093",
+    "url": "https://doi.org/10.1145/3510003.3510093"
+  },
+  "summary": {
+    "text": "This paper detects performance bugs in database systems using equivalent queries. Only the sentence citing SQLancer is available here, so this summary is limited to it: the authors note that existing work leverages tools such as SQLsmith and SQLancer to discover crash-inducing or logic bugs in DBMSs, positioning their own focus on performance against that.",
+    "relationship_to_sqlancer": "Cites SQLancer as the existing means of finding crash-inducing and logic bugs, from which it distinguishes its performance focus.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://doi.org/10.1145/3510003.3510093",
+    "source_sha256": "sha256:6478b7fd717627382bde2de9a55279b66a35fed969a68656b1a5a0accaad79d9",
+    "supporting_excerpts": [
+      "For instance, they leverage tools such as S QLSMITH [4] and SQLancer [37–39] to discover crash-inducing or logic bugs in DBMSs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3510003.3510093",
+          "source_type": "paper_citation_context",
+          "excerpt": "For instance, they leverage tools such as S QLSMITH [4] and SQLancer [37–39] to discover crash-inducing or logic bugs in DBMSs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+          "source_type": "paper_citation_context",
+          "excerpt": "To this end, we seek to compare against the TLP technique in SQLancer, which is the closest related effort [40–42].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [40–42].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+          "source_type": "paper_citation_context",
+          "excerpt": "In particular, it generates equivalent queries based on Ternary Logic Partitioning (TLP) [41].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+          "source_type": "paper_citation_context",
+          "excerpt": "Query equivalence is a well-studied topic and is used in many applications: (1) testing correctness of DBMS and SQL queries [41, 43], (2) educating developers [30], and (3) automatically grading student assignments [19].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, they leverage tools such as S QLSMITH [4] and SQLancer [40–42], which effectively discover crash or logic bugs in DBMS, respectively.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "manual_curation",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+          "source_type": "paper_citation_context",
+          "excerpt": "To this end, we seek to compare against the TLP technique in SQLancer, which is the closest related effort [40–42].",
+          "excerpt_is_verbatim": true,
+          "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+          "retrieved_at": "2026-09-06T07:08:48Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [40–42].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Pivoted Query Synthesis (PQS) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3510457_3513034.json b/_data/impact/paper_notes/paper_doi_10_1145_3510457_3513034.json
new file mode 100644
index 0000000..2ca9799
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3510457_3513034.json
@@ -0,0 +1,75 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3510457.3513034",
+    "title": "An Empirical Study on Quality Issues of eBay's Big Data SQL Analytics Platform",
+    "authors": [
+      "Feng Zhu",
+      "Lijie Xu",
+      "Gang Ma",
+      "Shuping Ji",
+      "Jie Wang",
+      "Gang Wang",
+      "Hongyi Zhang",
+      "K. Wan",
+      "Mingming Wang",
+      "Xingchao Zhang",
+      "Yuming Wang",
+      "Jingping Li"
+    ],
+    "year": 2022,
+    "venue": "2022 IEEE/ACM 44th International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP)",
+    "doi": "10.1145/3510457.3513034",
+    "url": "https://doi.org/10.1145/3510457.3513034"
+  },
+  "summary": {
+    "text": "An empirical study of service quality issues on Carmel, eBay's company-wide interactive SQL analytics platform built on Apache Spark, which has served thousands of customers across hundreds of teams for more than three years. Noting that few empirical studies examine service quality on open-source big data SQL platforms, the authors analyse 1,884 real service quality issues, summarise common symptoms, identify root causes through typical cases, and draw lessons and directions for automated tool support.",
+    "relationship_to_sqlancer": "An industrial study of failures in a SQL analytics platform; SQLancer is cited among database testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3510457.3513034",
+    "source_sha256": "sha256:f0a745ab316b263104a7e4f33832765942b03858a0730c925db63d58deedecb8",
+    "supporting_excerpts": [
+      "Meanwhile, despite the popularity of open-source based big data SQL analytics platforms, few empirical studies on service quality issues (e.g., job failure) were carried out for them.",
+      "To fill this gap, we conduct a comprehensive empirical study on 1,884 real-word service quality issues from Carmel.",
+      "We summa-rize the common symptoms and identify the root causes with typical cases."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3510457.3513034",
+          "source_type": "paper_citation_context",
+          "excerpt": "After investigation, we found the root cause is Spark’s endless retry logic 13 : \" we should ideally differentiate these task statuses so that they don’t count towards the failure limit \".",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3510457.3513034",
+          "source_type": "paper_citation_context",
+          "excerpt": "To detect DBMS bugs, Rigger and Su devised a series of novel approaches, including PQS [15], NoRec [13] and TLP [14].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3531348_3532176.json b/_data/impact/paper_notes/paper_doi_10_1145_3531348_3532176.json
new file mode 100644
index 0000000..2106459
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3531348_3532176.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3531348.3532176",
+    "title": "DeepBench: Benchmarking JSON Document Stores",
+    "authors": [
+      "S. Belloni",
+      "D. Ritter",
+      "Marco Schröder",
+      "Nils Rörup"
+    ],
+    "year": 2022,
+    "venue": "DBTest@SIGMOD",
+    "doi": "10.1145/3531348.3532176",
+    "url": "https://doi.org/10.1145/3531348.3532176"
+  },
+  "summary": {
+    "text": "DeepBench is a benchmark for JSON document stores, filling a gap the authors identify: JSON has spread rapidly as an exchange and storage format and specialised document stores are widely used, but no JSON-specific benchmark existed. DeepBench covers nested JSON data and queries over JSON documents, with configurable domain-independent scale levels such as document size and concurrent users alongside JSON-specific ones such as object and array nesting. Evaluating well-known document stores with a prototype revealed weaknesses that existing non-JSON benchmarks had not found.",
+    "relationship_to_sqlancer": "Benchmarking document stores rather than correctness testing; SQLancer is cited among the database evaluation work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3531348.3532176",
+    "source_sha256": "sha256:6d94957d123c714f49cd848b45b34b00303c12eda5e65e3c2e9d832dce8dc727",
+    "supporting_excerpts": [
+      "Consequently, specialized JSON document stores are ubiquitously used for diverse domain-specific workloads, while a JSON-specific benchmark is missing.",
+      "In this work, we specify DeepBench, an extensible, scalable benchmark that addresses nested JSON data, as well as queries over JSON documents.",
+      "The evaluation of well-known document stores with a prototypical DeepBench implementation shows its versatility and gives new insights into potential weaknesses that were not found by existing, non-JSON benchmarks"
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:38:27Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/a48f9e74f5f0f60e033c8c290156243c16ece5c0",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3533767_3534364.json b/_data/impact/paper_notes/paper_doi_10_1145_3533767_3534364.json
new file mode 100644
index 0000000..ace72b2
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3533767_3534364.json
@@ -0,0 +1,60 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3533767.3534364",
+    "title": "Unicorn: detect runtime errors in time-series databases with hybrid input synthesis",
+    "authors": [
+      "Zhiyong Wu",
+      "Jie Liang",
+      "Mingzhe Wang",
+      "Chijin Zhou",
+      "Yu Jiang"
+    ],
+    "year": 2022,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3533767.3534364",
+    "url": "https://doi.org/10.1145/3533767.3534364"
+  },
+  "summary": {
+    "text": "Unicorn fuzzes time-series databases, which are widely used in safety-critical IoT settings. The authors argue that fuzzing successful on relational databases does not transfer: mismatched query specifications leave most time-series logic unreachable, and implicitly handled exceptions hide serious bugs. Unicorn adds hybrid input synthesis to generate queries that cover time-series features while remaining grammatically correct, and proactive exception detection to surface bugs with minuscule symptoms. It has discovered 42 previously unknown bugs.",
+    "relationship_to_sqlancer": "Compares against SQLancer on six time-series databases, reporting 34-693% more basic blocks covered.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3533767.3534364",
+    "source_sha256": "sha256:b3cd2a15c24bf1f675dd0bb9c4aeb354b232d54887c1abe71c635ca15b8f93c4",
+    "supporting_excerpts": [
+      "However, it cannot be easily applied to time-series databases: the bulk of time-series logic is unreachable because of mismatched query specifications, and serious bugs are undetectable because of implicitly handled exceptions.",
+      "In this paper, we propose Unicorn to secure time-series databases with automated fuzzing.",
+      "Specifically, Unicorn outperforms SQLsmith and SQLancer on widely used time-series databases IoTDB, KairosDB, TimescaleDB, TDEngine, QuestDB, and GridDB in the number of basic blocks by 21%-199% and 34%-693%, respectively."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:38:27Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/1ac6b3a3904b6c9022206be7c0b9d180b84b1fb4",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3533767_3534409.json b/_data/impact/paper_notes/paper_doi_10_1145_3533767_3534409.json
new file mode 100644
index 0000000..1efc2b5
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3533767_3534409.json
@@ -0,0 +1,156 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3533767.3534409",
+    "title": "Finding bugs in Gremlin-based graph database systems via Randomized differential testing",
+    "authors": [
+      "Yingying Zheng",
+      "Wensheng Dou",
+      "Yicheng Wang",
+      "Zheng Qin",
+      "Leile Tang",
+      "Yu Gao",
+      "Dong Wang",
+      "Wei Wang",
+      "Jun Wei"
+    ],
+    "year": 2022,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3533767.3534409",
+    "url": "https://doi.org/10.1145/3533767.3534409"
+  },
+  "summary": {
+    "text": "Grand finds logic bugs in Gremlin-based graph database systems by constructing semantically equivalent databases across several systems and comparing the results of the same Gremlin query on each; differing results point to a logic bug. To make this work it adds model-based query generation, producing valid queries likely to return non-empty results, and a data mapping approach that unifies result formats across systems. Across six widely used graph systems including Neo4j and HugeGraph it found 21 previously unknown logic bugs, 18 confirmed.",
+    "relationship_to_sqlancer": "Takes SQLancer's oracles as the relational precedent it carries into graph databases: the paper names it among the tools researchers developed to find logic bugs in relational systems, and describes NoREC as comparing an optimized query against its non-optimized version.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "supplied PDF",
+    "written_from_url": "https://doi.org/10.1145/3533767.3534409",
+    "source_sha256": "sha256:7dcf00fce4e84d549efaa8fc72a0d7896dd82df8a174cd110f20734ed1bc33d5",
+    "supporting_excerpts": [
+      "NoREC [ 48] compares the execution results of a given optimized query with its non-optimized version, to detect optimization bugs in DBMS."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T15:22:55Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3533767.3534409",
+          "source_type": "paper_citation_context",
+          "excerpt": "Rigger et al. provide two metamorphic testing approaches [48, 49], namely Ternary Logic Partitioning (TLP) and Non-optimizing Reference Engine Construction (NoREC), to test DBMS. TLP [49] partitions a query into three sub-queries, and detects bugs by comparing the combination of results of three sub-queries with the result of the original query.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3533767.3534409",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [49] partitions a query into three sub-queries, and detects bugs by comparing the combination of results of three sub-queries with the result of the original query.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3533767.3534409",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoREC [48] compares the execution results of a given optimized query with its non-optimized version, to detect optimization bugs in DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3533767.3534409",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer offers Pivoted Query Synthesis (PQS) [50] approach to find logic bugs by randomly selecting a pivot row as oracle and generating random queries containing the selected row to test DBMS. ADUSA [39] translates SQL query to Alloy specification, generates Alloy instance satisfying query conditions in Alloy specification, and further obtains the expected result from Alloy instance.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3533767.3534409",
+          "source_type": "paper_citation_context",
+          "excerpt": "RAGS uses differential testing for detecting bugs in relational database systems, while SQLancer offers three oracles, i.e., Pivoted Query Synthesis (PQS) [50], Ternary Logic Partitioning (TLP) [49], and Non-Optimizing Reference Engine Construction (NoREC) [48] to find logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3533767.3534409",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer offers Pivoted Query Synthesis (PQS) [50] approach to find logic bugs by randomly selecting a pivot row as oracle and generating random queries containing the selected row to test DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/choeoe/Grand",
+          "source_type": "github_repository",
+          "excerpt": "# Grand",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after Grand, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/choeoe/Grand/blob/main/src/main/java/org/gdbtesting/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/main/java/org/gdbtesting/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.gdbtesting (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/choeoe/Grand",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "renamed_sqlancer_package",
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/choeoe/Grand",
+          "source_type": "github_repository",
+          "excerpt": "# Grand",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after Grand, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/choeoe/Grand/blob/main/src/main/java/org/gdbtesting/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/main/java/org/gdbtesting/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.gdbtesting (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3551349_3556924.json b/_data/impact/paper_notes/paper_doi_10_1145_3551349_3556924.json
new file mode 100644
index 0000000..2ff60a0
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3551349_3556924.json
@@ -0,0 +1,131 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3551349.3556924",
+    "title": "Differentially Testing Database Transactions for Fun and Profit",
+    "authors": [
+      "Ziyu Cui",
+      "Wensheng Dou",
+      "Qianwang Dai",
+      "Jiansen Song",
+      "Wei Wang",
+      "Jun Wei",
+      "Dan Ye"
+    ],
+    "year": 2022,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1145/3551349.3556924",
+    "url": "https://doi.org/10.1145/3551349.3556924"
+  },
+  "summary": {
+    "text": "DT2 tests transaction implementations by differential testing across DBMSs. It randomly generates a database and a group of concurrent transactions over it, supporting complex features such as varied schemas and cross-table queries, then compares execution results on several systems to find discrepancies. Because concurrency makes results non-deterministic, the authors add a transaction test protocol that forces deterministic execution. Across MySQL, MariaDB and TiDB it found 10 unique transaction bugs and 88 transaction-related compatibility issues, showing that systems claiming compatibility diverge in practice.",
+    "relationship_to_sqlancer": "Generates its databases and transactions mainly on the basis of SQLancer, revised for the transaction setting, and places PQS, NoREC and TLP among the approaches that detect bugs in single SELECT statements — the limitation DT2 sets out to pass.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "supplied PDF",
+    "written_from_url": "https://doi.org/10.1145/3551349.3556924",
+    "source_sha256": "sha256:85a042bce4d472eb92132d1149b40f5599d006524c2c28b5ba74c69ff03547ad",
+    "supporting_excerpts": [
+      "We generate them mainly based on SQLancer [ 16], and slightly revise the approach for our target.",
+      "Some approaches, e.g., PQS [ 60], NoREC [ 58], TLP [ 59] and RAGS [61], can detect DBMS bugs that involve these complex features in single SELECT statements."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T15:21:54Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3551349.3556924",
+          "source_type": "paper_citation_context",
+          "excerpt": "Some approaches, e.g., PQS [60], NoREC [58], TLP [59] and RAGS [61], can detect DBMS bugs that involve these complex features in single SELECT statements.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3551349.3556924",
+          "source_type": "paper_citation_context",
+          "excerpt": ", PQS [60], NoREC [58], TLP [59] and RAGS [61], can detect DBMS bugs that involve these complex features in single SELECT statements.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3551349.3556924",
+          "source_type": "paper_citation_context",
+          "excerpt": "More recently, Rigger et al. [58–60] has proposed a series of works to find logical bugs by generating single SQL queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/criszy/DT2",
+          "source_type": "github_repository",
+          "excerpt": "# DT2",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after DT2, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/criszy/DT2/blob/main/src/DT2/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/DT2/Randomly.java is SQLancer's Randomly.java, with the package renamed to DT2 (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/criszy/DT2",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "renamed_sqlancer_package",
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/criszy/DT2",
+          "source_type": "github_repository",
+          "excerpt": "# DT2",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after DT2, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/criszy/DT2/blob/main/src/DT2/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/DT2/Randomly.java is SQLancer's Randomly.java, with the package renamed to DT2 (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3551349_3560431.json b/_data/impact/paper_notes/paper_doi_10_1145_3551349_3560431.json
new file mode 100644
index 0000000..01d4fd2
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3551349_3560431.json
@@ -0,0 +1,98 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3551349.3560431",
+    "title": "Griffin : Grammar-Free DBMS Fuzzing",
+    "authors": [
+      "Jingzhou Fu",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Mingzhe Wang",
+      "Yu Jiang"
+    ],
+    "year": 2022,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1145/3551349.3560431",
+    "url": "https://doi.org/10.1145/3551349.3560431"
+  },
+  "summary": {
+    "text": "Griffin fuzzes DBMSs without depending on a grammar. Grammars make fuzzing efficient because DBMSs validate inputs strictly, but the grammars differ so much between systems that adapting a fuzzer to each is laborious, leaving many systems untested. Instead Griffin summarises the DBMS's state into a metadata graph: it tracks the metadata of statements in built-in test cases as they execute, builds a graph of dependencies between metadata and statements, then reshuffles statements and applies metadata-guided substitution to fix semantic errors. It found 55 previously unknown bugs, 13 with CVEs.",
+    "relationship_to_sqlancer": "Compares directly against SQLancer, reporting substantially more branch coverage and 27 more bugs in 12 hours.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3551349.3560431",
+    "source_sha256": "sha256:4e230aa168277a508670e9fcdc44fcbfcf1f50f0ab21c80c216de995a744df02",
+    "supporting_excerpts": [
+      "However, due to the vast differences in the complex grammar of various DBMSs, it is painstaking to adapt these fuzzers to them.",
+      "In this paper, we propose Griffin, a grammar-free mutation based DBMS fuzzer.",
+      "Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Squirrel, respectively."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:38:27Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3551349.3560431",
+          "source_type": "paper_citation_context",
+          "excerpt": "Compared with SQLancer, SQLsmith, and Sqirrel, Griffin covered 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches and found 27, 27, and 22 more bugs in 12 hours on SQLite, DuckDB, MariaDB, and PostgreSQL, respectively.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3551349.3560431",
+          "source_type": "paper_citation_context",
+          "excerpt": "The two generation-based fuzzers, SQLancer and SQLsmith, can only generate SQL statements based on their predefined models, which cover only a portion of the entire SQL grammar of the DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3551349.3560431",
+          "source_type": "paper_citation_context",
+          "excerpt": "Griffin covers 73.43%-274.70%, 80.47%- 312.89%, 43.80%-199.11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Sqirrel, respectively.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3551349.3560431",
+          "source_type": "paper_citation_context",
+          "excerpt": "Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Sqirrel, respectively.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3551349.3560431",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLsmith can only generate SELECT statements, while SQLancer can not generate DELETE statements, because their generation models are based on the limited SQL grammar.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3551349.3560431",
+          "source_type": "paper_citation_context",
+          "excerpt": "Specifically, Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches than SQLancer, SQLsmith, and Sqirrel after fuzzing 12 hours, respectively.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3552326_3587448.json b/_data/impact/paper_notes/paper_doi_10_1145_3552326_3587448.json
new file mode 100644
index 0000000..f54aef6
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3552326_3587448.json
@@ -0,0 +1,60 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3552326.3587448",
+    "title": "Fail through the Cracks: Cross-System Interaction Failures in Modern Cloud Systems",
+    "authors": [
+      "Lilia Tang",
+      "Chaitanya Bhandari",
+      "Yongle Zhang",
+      "Anna Karanika",
+      "Shuyang Ji",
+      "Indranil Gupta",
+      "Tianyi Xu"
+    ],
+    "year": 2023,
+    "venue": "European Conference on Computer Systems",
+    "doi": "10.1145/3552326.3587448",
+    "url": "https://doi.org/10.1145/3552326.3587448"
+  },
+  "summary": {
+    "text": "This paper studies cross-system interaction failures in cloud systems, where reliability depends not only on each subsystem but on how independently developed systems interact. Analysing 11 such incidents at Google, Azure and AWS plus 120 failure cases across seven widely co-deployed open-source systems, the authors focus on discrepancies between interacting systems as the root cause, arguing these failures cannot be understood by studying any one system alone. They advocate cross-system testing and verification, demonstrating it by cross-testing the Spark-Hive data plane and exposing 15 new discrepancies.",
+    "relationship_to_sqlancer": "Argues for differential testing across system boundaries, citing the database testing work that compares systems against each other.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3552326.3587448",
+    "source_sha256": "sha256:31d98a107a09037da396401ca74390ff54b5788858f22432f2e0c832a2dc4431",
+    "supporting_excerpts": [
+      "However, there is a lack of systematic understanding of this emerging mode of failures, which we term as cross-system interaction failures (or CSI failures).",
+      "We focus on understanding discrepancies between interacting systems as the root causes of CSI failures---CSI failures cannot be understood by analyzing one single system in isolation.",
+      "We advocate for cross-system testing and verification and demonstrate its potential by cross-testing the Spark-Hive data plane and exposing 15 new discrepancies."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3552326.3587448",
+          "source_type": "paper_citation_context",
+          "excerpt": ", [1, 22, 33]), which also exist in traditional systems that follow POSIX or SQL standards [94, 98], but are magnified by the heterogeneity and composability of cloud APIs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3582016_3582053.json b/_data/impact/paper_notes/paper_doi_10_1145_3582016_3582053.json
new file mode 100644
index 0000000..a4dbe0a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3582016_3582053.json
@@ -0,0 +1,55 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3582016.3582053",
+    "title": "Finding Unstable Code via Compiler-Driven Differential Testing",
+    "authors": [
+      "Shaohua Li",
+      "Zhendong Su"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Architectural Support for Programming Languages and Operating Systems",
+    "doi": "10.1145/3582016.3582053",
+    "url": "https://doi.org/10.1145/3582016.3582053"
+  },
+  "summary": {
+    "text": "CompDiff finds unstable code — code whose run-time semantics vary because of undefined behaviour, which compilers exploit by assuming it never happens. Sanitizers cover frequently occurring undefined behaviours, but detecting the rest remains hard. CompDiff rests on the observation that different compilers may produce semantically inconsistent binaries from unstable code, so it compares the outputs of differently compiled binaries on the same input. Integrated into AFL++, it uniquely detected 1,409 bugs on the Juliet benchmark compared with sanitizers, and found 78 new bugs across 23 open-source projects.",
+    "relationship_to_sqlancer": "Differential testing across implementations of the same specification; SQLancer is cited among that body of work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3582016.3582053",
+    "source_sha256": "sha256:8e967d4445638b7900b957f151a75c10ee8c6075866a8d1a22d96553c55627af",
+    "supporting_excerpts": [
+      "However, it remains a big challenge how to detect UBs that are beyond the reach of current techniques.",
+      "In this paper, we introduce compiler-driven differential testing (CompDiff), a simple yet effective approach for finding unstable code in C/C++ programs.",
+      "CompDiff relies on the fact that when compiling unstable code, different compiler implementations may produce semantically inconsistent binaries."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3582016.3582053",
+          "source_type": "paper_citation_context",
+          "excerpt": "It aims at improving fault tolerance of software by having N independent individuals or groups implementing the same specification.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3588909.json b/_data/impact/paper_notes/paper_doi_10_1145_3588909.json
new file mode 100644
index 0000000..dae37a6
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3588909.json
@@ -0,0 +1,121 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3588909",
+    "title": "Detecting Logic Bugs of Join Optimizations in DBMS",
+    "authors": [
+      "Xiu Tang",
+      "Sai Wu",
+      "Dongxiang Zhang",
+      "Fei Li",
+      "Gang Chen"
+    ],
+    "year": 2023,
+    "venue": "Proc. ACM Manag. Data",
+    "doi": "10.1145/3588909",
+    "url": "https://doi.org/10.1145/3588909"
+  },
+  "summary": {
+    "text": "TQS extends generation-based logic bug detection from single-table queries to multi-table joins, which the authors identify as a substantial gap in existing tools. Its Data-guided Schema and Query Generation component solves the ground-truth problem by using database normalization to build a testing schema and a bitmap index to track expected results, and inserts artificial noise into the data to improve debugging efficiency. Knowledge-guided Query Space Exploration treats the search as isomorphic graph set discovery, combining graph embedding with weighted random walks. It found 115 bugs across four DBMSs in 24 hours.",
+    "relationship_to_sqlancer": "Extends the ground-truth-based approach of Pivoted Query Synthesis to multi-table joins and evaluates against that line of work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3588909",
+    "source_sha256": "sha256:89b9ff1f33a7f1cc44624c7fb0383a42d7b81a43f1ad10ad83fef445e4e4524c",
+    "supporting_excerpts": [
+      "Nonetheless, existing generation-based debug tools are limited to single-table queries and there is a substantial research gap regarding multi-table queries with join operators.",
+      "DSG addresses the key challenge of multi-table query debugging: how to generate ground-truth (query, result) pairs for verification.",
+      "It successfully detected 115 bugs within 24 hours, including 31 bugs in MySQL, 30 in MariaDB, 31 in TiDB, and 23 in PolarDB respectively."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3588909",
+          "source_type": "paper_citation_context",
+          "excerpt": "Based on the randomly generated database, it adopts testing approaches such as Pivoted Query Synthesis (PQS) [50] to detect logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3588909",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [47].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3588909",
+          "source_type": "paper_citation_context",
+          "excerpt": "The second one is TLP [50], which decomposes a query into three partitioning queries, each of which computes its result on that tuple.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3588909",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [48].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3588909",
+          "source_type": "paper_citation_context",
+          "excerpt": "Pivoted Query Synthesis (PQS) has recently emerged as a promising way to detect logic bugs in DBMS [50].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3588909",
+          "source_type": "paper_citation_context",
+          "excerpt": "The third one is NoRec [47], which targets at logic bugs generated by the optimization process in DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2602.21955",
+          "source_type": "paper",
+          "excerpt": "We use three methods in SQLancer as our baselines.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3597503_3608133.json b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3608133.json
new file mode 100644
index 0000000..d8b0cb8
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3608133.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3608133",
+    "title": "EDEFuzz: A Web API Fuzzer for Excessive Data Exposures",
+    "authors": [
+      "Lianglu Pan",
+      "Shaanan Cohney",
+      "Toby C. Murray",
+      "Van-Thuan Pham"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3608133",
+    "url": "https://doi.org/10.1145/3597503.3608133"
+  },
+  "summary": {
+    "text": "EDEFuzz detects Excessive Data Exposure, where a web API sends clients far more data than they need, often over public channels — OWASP's third most significant API vulnerability of 2019. The authors note there are few automated tools for it, unsurprisingly since the problem has no explicit oracle: nothing crashes or violates memory. They develop a metamorphic relation to supply that oracle and build the first fuzzer for the problem. Across 69 targets from the Alexa Top-200 it found 33,365 potential leaks, and on eight Australian websites it achieved a 98.65% true positive rate.",
+    "relationship_to_sqlancer": "Constructs a metamorphic relation where no explicit oracle exists, the same problem framing as SQLancer's oracles, applied to web APIs.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3597503.3608133",
+    "source_sha256": "sha256:00df17b886de8df6aed088a8a3b6a2c6e7cc87b90ab715a7f41ad87b00a43af1",
+    "supporting_excerpts": [
+      "This is unsurprising as the problem lacks an explicit test oracle: the vulnerability does not manifest through explicit abnormal behaviours (e.g., program crashes or memory access violations).",
+      "In this work, we develop a metamorphic relation to tackle that challenge and build the first fuzzing tool-that we call EDEFUZZ-to systematically detect EDEs.",
+      "In a more-tightly controlled experiment of eight popular websites in Australia, EDEFuzz achieved a high true positive rate of 98.65% with minimal configuration, illustrating our tool's accuracy and efficiency."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/d2bf26b7ec662c29c30e6ad4548ce53d360718f2",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639112.json b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639112.json
new file mode 100644
index 0000000..14083bd
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639112.json
@@ -0,0 +1,70 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3639112",
+    "title": "Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation",
+    "authors": [
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Mingzhe Wang",
+      "Chengnian Sun",
+      "Yu Jiang"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3639112",
+    "url": "https://doi.org/10.1145/3597503.3639112"
+  },
+  "summary": {
+    "text": "Mozi finds DBMS bugs by comparing one system against itself under different configurations rather than by comparing semantically equivalent queries. Its motivation is that metamorphic testing needs a precise grasp of the SQL specification to build equivalent inputs, and that the specification is vague and intricate enough to make modelling query semantics hard. Mozi analyses the query plan, changes configuration to produce an equivalent system, re-executes and compares: different results indicate correctness bugs, and faster execution with optimisations closed indicates performance bugs. It found 101 previously unknown bugs across four DBMSs.",
+    "relationship_to_sqlancer": "Positions itself against metamorphic approaches such as SQLancer's, replacing equivalent queries with equivalent system configurations.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3597503.3639112",
+    "source_sha256": "sha256:10654c09f956164627af1e0901b007a1753bc23ccfb90e455bd28ce677ca4cd8",
+    "supporting_excerpts": [
+      "Traditional approaches, such as metamorphic testing, need a precise comprehension of the SQL specification to create diverse inputs with equivalent semantics.",
+      "To address this, we propose Mozi, a framework that finds DBMS bugs via configuration-based equivalent transformation.",
+      "In the continuous testing, Mozi found a total of 101 previously unknown bugs, including 49 correctness and 52 performance bugs in four DBMSs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639112",
+          "source_type": "paper_citation_context",
+          "excerpt": "In 24-hour experiments, Mozi covers 64,973, 54,464, 43,236, 28,499, and 14,084 more branches, detects 25, 22, 21, 24, and 26 more bugs than PQS [48], NoREC [46], TLP 1 [47], Apollo [25], and Amoeba [33], respectively.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639112",
+          "source_type": "paper_citation_context",
+          "excerpt": "Metamorphic testing is a prevalent approach to testing DBMS, which focuses on building metamorphic relations [5, 6, 32, 46, 47, 49].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639200.json b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639200.json
new file mode 100644
index 0000000..1f91b9c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639200.json
@@ -0,0 +1,100 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3639200",
+    "title": "Testing Graph Database Systems via Equivalent Query Rewriting",
+    "authors": [
+      "Qiuyang Mang",
+      "Aoyang Fang",
+      "Boxi Yu",
+      "Hanfei Chen",
+      "Pinjia He"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3639200",
+    "url": "https://doi.org/10.1145/3597503.3639200"
+  },
+  "summary": {
+    "text": "GRev adapts Equivalent Query Rewriting to graph database systems: it rewrites a query into equivalent forms that trigger distinct query plans and checks for discrepancies in system behaviour, covering logic errors, performance bugs and unexpected exceptions. To generate the rewrites it introduces the Abstract Syntax Graph, which embeds a base query's semantics into the paths of a graph so that an equivalent query can be produced by finding paths that together carry the complete semantics, with a Random Walk Covering algorithm to select them. GRev found 22 previously unknown bugs across five graph systems, 15 confirmed.",
+    "relationship_to_sqlancer": "Applies equivalent query rewriting, the basis of SQLancer's oracles, to graph databases.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3597503.3639200",
+    "source_sha256": "sha256:1f59e917fb48459dc4193035b8b2ab237033c3985302a1d517c84699fb23dd15",
+    "supporting_excerpts": [
+      "This paper adapts Equivalent Query Rewriting (EQR) to GDBMS testing.",
+      "EQR rewrites a GDBMS query into equivalent ones that trigger distinct query plans, and checks whether they exhibit discrepancies in system behaviors.",
+      "As a practical implementation of these ideas, we develop a tool GRev, which has successfully detected 22 previously unknown bugs across 5 popular GDBMS, with 15 of them being confirmed."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:56Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639200",
+          "source_type": "paper_citation_context",
+          "excerpt": "Similarly, GDBMeter [15], a tool based on Query Partitioning [32], requires different validation rules for different queries and may struggle to handle some widely-used clauses such as DISTINCT .",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639200",
+          "source_type": "paper_citation_context",
+          "excerpt": "Specifically, GDsmith and Grand are based on Differential Testing [38] while GDBMeter is based on Query Partitioning [32] ( i.e. , partitioning query into multiple equivalent sub-queries).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639200",
+          "source_type": "paper_citation_context",
+          "excerpt": "Based on Query Partitioning, GDBMeter [15] leverages the test oracle called Ternary Logic Partitioning (TLP) to detect bugs in GDBMS. TLP [32] was initially introduced in testing RDBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639200",
+          "source_type": "paper_citation_context",
+          "excerpt": "Notably, we can not compare GRev with RDBMS testing tools, such as SQLancer [33].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639200",
+          "source_type": "paper_citation_context",
+          "excerpt": "Notably, these bugs were rarely detected by the existing techniques, especially for GDBMeter [15], which reused the test oracles designed for RDBMS ( i.e. , TLP [32]).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639200",
+          "source_type": "paper_citation_context",
+          "excerpt": "However, none of the graph-related bugs have been detected by it due to reusing TLP[32], a metamorphic relation for RDBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639207.json b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639207.json
new file mode 100644
index 0000000..cd9190c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639207.json
@@ -0,0 +1,92 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3639207",
+    "title": "Understanding Transaction Bugs in Database Systems",
+    "authors": [
+      "Ziyu Cui",
+      "Wensheng Dou",
+      "Yu Gao",
+      "Dong Wang",
+      "Jiansen Song",
+      "Yingying Zheng",
+      "Tao Wang",
+      "Rui Yang",
+      "Kang Xu",
+      "Yixin Hu",
+      "Jun Wei",
+      "Tao Huang"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3639207",
+    "url": "https://doi.org/10.1145/3597503.3639207"
+  },
+  "summary": {
+    "text": "The first comprehensive study of transaction bugs in DBMSs, based on 140 bugs collected from MySQL, PostgreSQL, SQLite, MariaDB, CockroachDB and TiDB. The authors examine how the bugs manifest, their root causes, their impact — incorrect database states and crashes among them — and how they were fixed, arguing that understanding real transaction bugs is what effective detection techniques must be built on. The study reports findings intended as guidance for transaction bug detection, testing and verification.",
+    "relationship_to_sqlancer": "Characterises the class of bug that transaction-oriented extensions of SQLancer target, citing that work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3597503.3639207",
+    "source_sha256": "sha256:6af97e9163a3704b74d0a933c631dafdad42945634075572a14e21bab074c857",
+    "supporting_excerpts": [
+      "An in-depth understanding of real-world transaction bugs can significantly promote effective techniques in combating transaction bugs in DBMSs.",
+      "In this paper, we conduct the first comprehensive study on 140 transaction bugs collected from six widely-used DBMSs, i.e., MySQL, PostgreSQL, SQLite, MariaDB, CockroachDB, and TiDB.",
+      "We investigate these bugs from their bug manifestations, root causes, bug impacts and bug fixing."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639207",
+          "source_type": "paper_citation_context",
+          "excerpt": "If a bug can be triggered by only one auto transaction, we do not consider it as a TXBug, since it usually belongs to transaction-unrelated bugs, e.g., logic bugs in single SELECT statements [64–66].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639207",
+          "source_type": "paper_citation_context",
+          "excerpt": "Although many approaches have been proposed for DBMS testing [17, 34, 49, 51, 54, 55, 59–61, 64–68, 73, 74, 77, 78], there are only few works for transaction testing in DBMSs [44, 45].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639207",
+          "source_type": "paper_citation_context",
+          "excerpt": "Some techniques have been proposed to combat logic and crash bugs in DBMSs [17, 34, 49, 51, 54, 55, 59– 61, 64–68, 73, 74, 76–78].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639207",
+          "source_type": "paper_citation_context",
+          "excerpt": "RAGS [67] utilizes differential testing to find bugs in DBMSs. SQLancer [64–66] detects logic bugs in single SELECT statements.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639210.json b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639210.json
new file mode 100644
index 0000000..7334486
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639210.json
@@ -0,0 +1,76 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3639210",
+    "title": "Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer",
+    "authors": [
+      "Jingzhou Fu",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Yu Jiang"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3639210",
+    "url": "https://doi.org/10.1145/3597503.3639210"
+  },
+  "summary": {
+    "text": "Sedar produces high-quality initial seeds for DBMS fuzzing by transferring test cases from other DBMSs, addressing the problem that many systems lack comprehensive built-in test suites. The insight is that DBMSs share many features, so a seed reaching deep paths in one can be adapted for another; the difficulty is converting it to the target's grammar. Sedar executes existing test cases in their original DBMS while capturing schema information, uses an LLM guided by that information to generate new cases, and temporarily comments out unparsable sections so fuzzers can mutate them. It found 70 new vulnerabilities, 19 with CVEs.",
+    "relationship_to_sqlancer": "Improves the seed corpus for DBMS fuzzers; SQLancer is cited among DBMS testing tools.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3597503.3639210",
+    "source_sha256": "sha256:ab52e66269a6d007f79c28ebfd5e6e23bbf1f96f153ac7231e1ced775e460d99",
+    "supporting_excerpts": [
+      "While built-in test cases are typically used as initial seeds, many DBMSs lack comprehensive test cases, making it difficult to apply state-of-the-art fuzzing techniques directly.",
+      "To address this, we propose Sedar which produces initial seeds for a target DBMS by transferring test cases from other DBMSs.",
+      "Moreover, Sedar discovered 70 new vulnerabilities, with 60 out of them being uniquely found by Sedar with transferred seeds, and 19 of them have been assigned with CVEs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639210",
+          "source_type": "paper_citation_context",
+          "excerpt": "Typically, DBMSs adhere to basic features of ANSI SQL standard [1] for common uses, while they support more advanced features with their unique SQL dialects.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639210",
+          "source_type": "paper_citation_context",
+          "excerpt": "PQS [32] generates queries that should fetch a specific row, and indicates a bug triggered when the row is not included in result sets.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597503.3639210",
+          "source_type": "paper_citation_context",
+          "excerpt": "Some generation-based approaches [15, 22, 30–32, 34, 37] are proposed to find specific types of DBMS bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639212.json b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639212.json
new file mode 100644
index 0000000..2a06bfd
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3597503_3639212.json
@@ -0,0 +1,61 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3639212",
+    "title": "Combining Structured Static Code Information and Dynamic Symbolic Traces for Software Vulnerability Prediction",
+    "authors": [
+      "Huanting Wang",
+      "Zhanyong Tang",
+      "Shin Hwei Tan",
+      "Jie Wang",
+      "Yuzhe Liu",
+      "Hejun Fang",
+      "Chunwei Xia",
+      "Zheng Wang"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3639212",
+    "url": "https://doi.org/10.1145/3597503.3639212"
+  },
+  "summary": {
+    "text": "Concoction learns program representations for bug detection by combining static source code information with dynamic symbolic execution traces, addressing the limitation that existing deep-learning representations either miss precise semantics or do not scale. Unsupervised active learning picks a subset of important paths for trace collection, and a focused symbolic execution implementation keeps the cost down. Integrated with fuzzing and applied to C programs from 20 open-source projects, it found 54 unique vulnerabilities and 37 new CVE IDs over 200 hours.",
+    "relationship_to_sqlancer": "Learning-based vulnerability detection for C programs; SQLancer is cited as background rather than used.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3597503.3639212",
+    "source_sha256": "sha256:a94422111f098beb7379201a6719e9610fc38813b13646a20951810fcd6a67aa",
+    "supporting_excerpts": [
+      "However, existing DL-based solutions for learning program representations have limitations - they either cannot capture the deep, precise program semantics or suffer from poor scalability.",
+      "We present Con-coction, the first DL system to learn program presentations by combining static source code information and dynamic program execution traces.",
+      "In 200 hours of automated concurrent test runs, Concoction has successfully uncovered vul-nerabilities in all tested projects, identifying 54 unique vulnera-bilities and yielding 37 new, unique CVE IDs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/a3f32348f15845519abf3b862fce1682001da1cd",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598046.json b/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598046.json
new file mode 100644
index 0000000..93e62f1
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598046.json
@@ -0,0 +1,131 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3597926.3598046",
+    "title": "GDsmith: Detecting Bugs in Cypher Graph Database Engines",
+    "authors": [
+      "Ziyue Hua",
+      "Weisheng Lin",
+      "Luyao Ren",
+      "Zongyang Li",
+      "Lu Zhang",
+      "Wenpin Jiao",
+      "Tao Xie"
+    ],
+    "year": 2023,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3597926.3598046",
+    "url": "https://doi.org/10.1145/3597926.3598046"
+  },
+  "summary": {
+    "text": "GDsmith is the first approach for testing Cypher graph database engines. Its central difficulty is that random test generation struggles to produce property graphs and queries complex enough to return non-empty results, which is what wrong-result bugs require. GDsmith ensures every randomly generated query meets the semantic requirements, and raises the chance of complex non-empty results with graph-guided generation of pattern combinations and data-guided generation of conditions. It detected 28 bugs across three popular open-source graph engines.",
+    "relationship_to_sqlancer": "The paper states plainly that its framework is derived from SQLancer, carrying that tool's approach from relational engines to Cypher graph engines, and summarises PQS, NoREC and TLP as the oracles that came before it.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "supplied PDF",
+    "written_from_url": "https://doi.org/10.1145/3597926.3598046",
+    "source_sha256": "sha256:a94ab880b5b91a3cf9b6747835f1feacf6424cde30a39935d4ed49da7bc7cfe2",
+    "supporting_excerpts": [
+      "Its framework is derived from SQLancer [ 18] (which is a tool to automatically test relational database engines).",
+      "PQS [ 21] detects wrong-result bugs by checking whether a specific record is fetched correctly."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T15:21:52Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3597926.3598046",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoREC [19] detects bugs in a relational database engine by applying a semantic-preserving transformation to a given SQL query to disable the engine’s optimizations and addresses PQS’ high implementation effort.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597926.3598046",
+          "source_type": "paper_citation_context",
+          "excerpt": "PQS [21] detects wrong-result bugs by checking whether a specific record is fetched correctly.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3597926.3598046",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [20] derives multiple SQL queries that compute a partial result of the initial query.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/ddaa2000/GDsmith",
+          "source_type": "github_repository",
+          "excerpt": "# GDsmith",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after GDsmith, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/ddaa2000/GDsmith/blob/master/src/main/java/org/example/gdsmith/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.ALPHANUMERIC;",
+          "excerpt_is_verbatim": true,
+          "note": "src/main/java/org/example/gdsmith/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.example.gdsmith (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/ddaa2000/GDsmith",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "renamed_sqlancer_package",
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/ddaa2000/GDsmith",
+          "source_type": "github_repository",
+          "excerpt": "# GDsmith",
+          "excerpt_is_verbatim": true,
+          "note": "Repository is named after GDsmith, the tool this paper says it built, and is about database testing.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/ddaa2000/GDsmith/blob/master/src/main/java/org/example/gdsmith/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.ALPHANUMERIC;",
+          "excerpt_is_verbatim": true,
+          "note": "src/main/java/org/example/gdsmith/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.example.gdsmith (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598052.json b/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598052.json
new file mode 100644
index 0000000..4e0781d
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598052.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3597926.3598052",
+    "title": "Dependency-Aware Metamorphic Testing of Datalog Engines",
+    "authors": [
+      "Muhammad Numair Mansur",
+      "Valentin Wüstholz",
+      "M. Christakis"
+    ],
+    "year": 2023,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3597926.3598052",
+    "url": "https://doi.org/10.1145/3597926.3598052"
+  },
+  "summary": {
+    "text": "DLSmith applies metamorphic testing to Datalog engines, whose query bugs can undermine the soundness of the program analysers built on them, with potentially serious consequences in safety-critical settings. Compared with existing work, the approach uses rich precedence information capturing dependencies among relations in the program, which the authors say permits considerably more general and effective metamorphic transformations. DLSmith detected 16 previously unknown query bugs across four Datalog engines.",
+    "relationship_to_sqlancer": "Carries metamorphic query transformation, the basis of SQLancer's oracles, into Datalog engines.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3597926.3598052",
+    "source_sha256": "sha256:8f240188d6efc064827c5585b3418f238315805955272e486ede598ed8ede76a",
+    "supporting_excerpts": [
+      "Such bugs, called query bugs, may compromise the soundness of upstream program analyzers, having potentially detrimental consequences in safety-critical settings.",
+      "To address this issue, we develop a metamorphic testing approach for detecting query bugs in Datalog engines.",
+      "We implement our approach in DLSmith, which detected 16 previously unknown query bugs in four Datalog engines."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3597926.3598052",
+          "source_type": "paper_citation_context",
+          "excerpt": "Metamorphic testing has also been successfully used to test a variety of software applications [12, 27, 47, 57], including other query-based systems [43, 46, 63].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598068.json b/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598068.json
new file mode 100644
index 0000000..2ee0a59
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598068.json
@@ -0,0 +1,58 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3597926.3598068",
+    "title": "Exploring Missed Optimizations in WebAssembly Optimizers",
+    "authors": [
+      "Zhibo Liu",
+      "Dongwei Xiao",
+      "Zongjie Li",
+      "Shuai Wang",
+      "W. Meng"
+    ],
+    "year": 2023,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3597926.3598068",
+    "url": "https://doi.org/10.1145/3597926.3598068"
+  },
+  "summary": {
+    "text": "DITWO looks for optimisations that WebAssembly optimizers miss, motivated by findings that real-world wasm applications run slower than expected. It compiles a C program to both a native x86 executable and a wasm executable, then compares optimization indication traces — sequences of global variable writes and function calls that reflect how far each was optimised — and treats divergence as a missed optimisation. Analysing the official optimizer wasm-opt, it identified 1,293 inputs triggering missed optimisations and nine root causes, and estimates fixing them would yield at least a 17.15% performance improvement.",
+    "relationship_to_sqlancer": "Differential detection of missed optimisations, the same goal as SQLancer's performance-oriented work, in a compiler setting.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3597926.3598068",
+    "source_sha256": "sha256:ef92fcb2f8641fa49d53b91c0fdf550875a55176ac3a7feda7241b2da6168dff",
+    "supporting_excerpts": [
+      "Despite the prosperous use of wasm executables, recent research has indicated that real-world wasm applications are slower than anticipated, suggesting deficiencies in wasm optimizations.",
+      "To do so, we present DITWO, a differential testing framework to uncover missed optimizations (MO) of wasm optimizers.",
+      "Our analysis of the official wasm optimizer, wasm-opt, successfully identifies 1,293 inputs triggering MO of wasm-opt."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3597926.3598068",
+          "source_type": "paper_citation_context",
+          "excerpt": "Differential testing (DT) is used in different software domains, including databases [66, 72], Java Virtual Machines (JVMs) [21, 22], symbolic execution engines [45], disassemblers [63], decompilers [54], and deep learning systems [37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598130.json b/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598130.json
new file mode 100644
index 0000000..93fc682
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3597926_3598130.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3597926.3598130",
+    "title": "GrayC: Greybox Fuzzing of Compilers and Analysers for C",
+    "authors": [
+      "Karine Even-Mendoza",
+      "Arindam Sharma",
+      "Alastair F. Donaldson",
+      "Cristian Cadar"
+    ],
+    "year": 2023,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3597926.3598130",
+    "url": "https://doi.org/10.1145/3597926.3598130"
+  },
+  "summary": {
+    "text": "GrayC brings coverage-directed, mutation-based greybox fuzzing to C compilers and analysers, which have become relatively resistant to the blackbox fuzzers used until now. The central difficulty is that naive mutations produce programs that do not compile and so cannot reach deep bugs in optimisation, analysis and code generation. GrayC introduces mutations targeting common C constructs and transforms fuzzed programs so they produce meaningful output, enabling differential testing. It found 30 confirmed compiler and analyser bugs, 25 previously unknown, and contributed 24 test cases to the Clang/LLVM suite.",
+    "relationship_to_sqlancer": "Greybox fuzzing with a differential oracle for compilers; SQLancer is cited among testing work in that tradition.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3597926.3598130",
+    "source_sha256": "sha256:52a9cf6508ad400116fc523458f8bb395e1b290e8c98f4185b71d36088208ded",
+    "supporting_excerpts": [
+      "Most such fuzzing techniques have taken a blackbox approach, with compilers and code analysers starting to become relatively immune to such fuzzers.",
+      "The main challenge of applying mutation-based fuzzing in this context is that naive mutations are likely to generate programs that do not compile.",
+      "We have used GrayC to identify 30 confirmed compiler and code analyser bugs: 25 previously unknown bugs (with 22 of them already fixed in response to our reports) and 5 confirmed bugs reported independently shortly before we found them."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/86a60acfe93752af1159dea65d01808134754467",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3605157_3605177.json b/_data/impact/paper_notes/paper_doi_10_1145_3605157_3605177.json
new file mode 100644
index 0000000..a1c8be4
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3605157_3605177.json
@@ -0,0 +1,67 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3605157.3605177",
+    "title": "Beyond the Coverage Plateau: A Comprehensive Study of Fuzz Blockers (Registered Report)",
+    "authors": [
+      "Wentao Gao",
+      "Van-Thuan Pham",
+      "Dongge Liu",
+      "Oliver Chang",
+      "Toby C. Murray",
+      "Benjamin I. P. Rubinstein"
+    ],
+    "year": 2023,
+    "venue": "FUZZING",
+    "doi": "10.1145/3605157.3605177",
+    "url": "https://doi.org/10.1145/3605157.3605177"
+  },
+  "summary": {
+    "text": "A registered report studying fuzz blockers — the obstacles that make coverage-guided greybox fuzzers plateau, which benchmarks such as FuzzBench show typically happens after around 12 hours. Using the FuzzIntrospector platform, the authors analyse and categorise these blockers to indicate where future fuzzing research should focus. Their preliminary finding is that most top blockers are not directly related to the program input, which points to fuzz driver generation and modification as the area needing better techniques.",
+    "relationship_to_sqlancer": "Studies why fuzzers stop making progress; SQLancer is cited among the testing work discussed.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3605157.3605177",
+    "source_sha256": "sha256:1e17dd050e08717f2cb30290cbb650ed65cc304e9b643e4fdf95c3e447ed4c5f",
+    "supporting_excerpts": [
+      "However, results on large benchmarks such as FuzzBench indicate that the state-of-the-art fuzzers often reach a plateau after a certain period, typically around 12 hours.",
+      "With the aid of the newly introduced FuzzIntrospector platform, this study aims to analyze and categorize the fuzz blockers that impede the progress of fuzzers.",
+      "Our preliminary findings reveal that the majority of top fuzz blockers are not directly related to the program input, emphasizing the need for enhanced techniques in automated fuzz driver generation and modification."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3605157.3605177",
+          "source_type": "paper_citation_context",
+          "excerpt": "Additionally, researchers have attempted to extend the applicability of fuzzing to challenging targets such as network protocols [18, 43], database systems [45, 50], SMT solvers [40], compilers [26], device drivers [41], and heterogeneous applications [48].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3605157.3605177",
+          "source_type": "paper_citation_context",
+          "excerpt": "These efforts have focused on enhancing fuzzing in areas such as feedback collection [16, 25, 27, 30], corpus management [29], seed selection algorithms [22, 23], input generation algorithms [15, 20, 31, 44, 47], and novel test oracle designs [39, 45].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3611643_3613893.json b/_data/impact/paper_notes/paper_doi_10_1145_3611643_3613893.json
new file mode 100644
index 0000000..305c0c4
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3611643_3613893.json
@@ -0,0 +1,61 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3611643.3613893",
+    "title": "Adapting Performance Analytic Techniques in a Real-World Database-Centric System: An Industrial Experience Report",
+    "authors": [
+      "Lizhi Liao",
+      "Heng Li",
+      "Weiyi Shang",
+      "Catalin Sporea",
+      "Andrei Toma",
+      "S. Sajedi"
+    ],
+    "year": 2023,
+    "venue": "ESEC/SIGSOFT FSE",
+    "doi": "10.1145/3611643.3613893",
+    "url": "https://doi.org/10.1145/3611643.3613893"
+  },
+  "summary": {
+    "text": "An industrial experience report on adapting performance analysis to a large-scale database-centric system. The authors argue that in such systems most business logic and calculation live in the database rather than the application, and that because those calculations are written in SQL, both the performance issues and their diagnosis differ from systems dominated by conventional programming languages. Their adapted analysis focuses on the database and on the interactions between database and application while minimising reliance on expert knowledge and manual effort, and they document the challenges met and lessons learned.",
+    "relationship_to_sqlancer": "Performance diagnosis in production database-centric systems; SQLancer is cited among database analysis work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3611643.3613893",
+    "source_sha256": "sha256:a4f45ae7fa351d85325e40424bdc723668a740f0aa190547caeae236e593c071",
+    "supporting_excerpts": [
+      "However, directly applying these existing techniques to large-scale database-centric systems can be challenging and may not perform well due to the unique nature of such systems.",
+      "In particular, compared to typical database-based systems like online shopping systems, in database-centric systems, a majority of the business logic and calculations reside in the database instead of the application.",
+      "Our adapted performance analysis pays special attention to the database and the interactions between the database and the application with minimal reliance on expert knowledge and manual effort."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3611643.3613893",
+          "source_type": "paper_citation_context",
+          "excerpt": "Rigger and Su utilize multiple analysis techniques, such as query partitioning [43], pivoted query synthesis [44], or non-optimizing reference engine construction [42] to detect database logic bugs that are critical in database-centric system quality but often go unnoticed by developers.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3611643_3616286.json b/_data/impact/paper_notes/paper_doi_10_1145_3611643_3616286.json
new file mode 100644
index 0000000..ad86a38
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3611643_3616286.json
@@ -0,0 +1,60 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3611643.3616286",
+    "title": "Property-Based Fuzzing for Finding Data Manipulation Errors in Android Apps",
+    "authors": [
+      "Jingling Sun",
+      "Ting Su",
+      "Jiayi Jiang",
+      "Jue Wang",
+      "G. Pu",
+      "Zhendong Su"
+    ],
+    "year": 2023,
+    "venue": "ESEC/SIGSOFT FSE",
+    "doi": "10.1145/3611643.3616286",
+    "url": "https://doi.org/10.1145/3611643.3616286"
+  },
+  "summary": {
+    "text": "PBFDroid applies property-based fuzzing to data manipulation errors in Android apps — bugs in the create, read, update and delete operations that handle app-specific data, especially the non-crashing logic ones. Given a type of app data, it randomly interleaves the relevant data manipulation functions with other events to explore diverse app states, characterising those functions as data-model-based properties and using consistency between the data model and the UI layout to check them. Across 20 real apps it found 30 previously unknown bugs, 29 of them data manipulation errors, 22 non-crashing.",
+    "relationship_to_sqlancer": "Applies property-based checking to CRUD correctness in applications; SQLancer's work on CRUD errors in DBMSs is the cited prior work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3611643.3616286",
+    "source_sha256": "sha256:90282dcb6e3624fad848db2b86e6bdf46ea4bba22f9c16640f03bd12bbcc9c8e",
+    "supporting_excerpts": [
+      "However, the bugs related to DMFs (named as data manipulation errors, DMEs ), especially those non-crashing logic ones, are prevalent but difficult to find.",
+      "To this end, inspired by property-based testing, we introduce a property-based fuzzing approach to effectively finding DMEs in Android apps.",
+      "Our further evaluation confirms that none of the 22 non-crashing DMEs can be found by the state-of-the-art techniques."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3611643.3616286",
+          "source_type": "paper_citation_context",
+          "excerpt": "Some work [51, 52] uses metamorphic testing to find the CRUD errors in database management systems.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3622819.json b/_data/impact/paper_notes/paper_doi_10_1145_3622819.json
new file mode 100644
index 0000000..4be052b
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3622819.json
@@ -0,0 +1,102 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3622819",
+    "title": "Towards Better Semantics Exploration for Browser Fuzzing",
+    "authors": [
+      "Chijin Zhou",
+      "Quan Zhang",
+      "Lihua Guo",
+      "Mingzhe Wang",
+      "Yu Jiang",
+      "Qing Liao",
+      "Zhiyong Wu",
+      "Shanshan Li",
+      "Bin Gu"
+    ],
+    "year": 2023,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3622819",
+    "url": "https://doi.org/10.1145/3622819"
+  },
+  "summary": {
+    "text": "SaGe improves browser fuzzing by replacing handwritten context-free grammars, which the authors say model browser semantics too crudely and yield inputs that cover only part of the semantics and are often semantically wrong. It extracts a rudimentary grammar from W3C standards and iteratively enriches it into a production-context sensitive grammar carrying semantic information, so generated inputs reach more semantics and are more often valid. Over 24-hour campaigns on Chrome, Safari and Firefox it improved edge coverage by 6-278% and found 62 bugs, 10 with CVEs.",
+    "relationship_to_sqlancer": "Grammar-based generation for browsers; SQLancer is cited among the testing work it draws on.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3622819",
+    "source_sha256": "sha256:3c4b7976169233322b0d4f6b918cdc021f144927fc8fb95ea43ffdc5aad8bbef",
+    "supporting_excerpts": [
+      "However, these CFGs fall short in adequately modeling the complex semantics of browsers, resulting in generated inputs that cover only a portion of the semantics and are prone to semantic errors.",
+      "In this paper, we present SaGe, an automated method that enhances browser fuzzing through the use of production-context sensitive grammars (PCSGs) incorporating semantic information.",
+      "Our approach demonstrated better performance compared to existing browser fuzzers, with a 6.03%-277.80% improvement in edge coverage, a 3.56%-161.71% boost in semantic correctness rate, twice the number of bugs discovered."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3622819",
+          "source_type": "paper_citation_context",
+          "excerpt": "This monitor is built on top of the JavaScript interpreter without intrusively instrumenting the browsers, providing flexibility in detecting semantic errors.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3622819",
+          "source_type": "paper_citation_context",
+          "excerpt": "An HTML document consists of three parts: (1) an HTML part to define the initial Document Object Model (DOM) tree, (2) a CSS part to specify in which style the elements of DOM tree are rendered; and (3) a JavaScript part to programmatically manipulate objects in DOM tree or enable other functionalities.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3622819",
+          "source_type": "paper_citation_context",
+          "excerpt": "…fuzzers, there are also many custom fuzzers, enhanced with domain knowledge, for fuzzing specific domain such as C/C++ compilers [Livinskii et al. 2020; Yang et al. 2011], databases [Rigger and Su 2020; Wang et al. 2021b; Zhong et al. 2020], and deep learning frameworks [Liu et al. 2023].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3622819",
+          "source_type": "paper_citation_context",
+          "excerpt": "As shown in Listing 1, the minimized code snippet is quite simple: the HTML part creates an audio element, and the JavaScript part accesses this element, and calls the setSinkId member function of this element with a random String input.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3622819",
+          "source_type": "paper_citation_context",
+          "excerpt": "Their generation strategies focus more on generating nested JavaScript code, such as code with complex variable lifetimes or multiple loops, in order to explore the interpretation/optimization logic of JavaScript engines.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3622819",
+          "source_type": "paper_citation_context",
+          "excerpt": "However, they utilize the standards to perform differential testing in order to uncover standard conformance bugs, i.e., inconsistencies of the implementation of JavaScript engine and JavaScript standards.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3643779.json b/_data/impact/paper_notes/paper_doi_10_1145_3643779.json
new file mode 100644
index 0000000..cbdfdea
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3643779.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3643779",
+    "title": "DTD: Comprehensive and Scalable Testing for Debuggers",
+    "authors": [
+      "Hongyi Lu",
+      "Zhibo Liu",
+      "Shuai Wang",
+      "Fengwei Zhang"
+    ],
+    "year": 2024,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3643779",
+    "url": "https://doi.org/10.1145/3643779"
+  },
+  "summary": {
+    "text": "DTD tests interactive debuggers, which can misretrieve or misinterpret program state and so mislead developers, and for which no scalable comprehensive correctness measurement existed. The authors propose testing criteria covering both comprehensiveness of debug information and scalability, then build a differential testing framework that runs the same C executable through two mainstream debuggers and treats discrepancies as bugs. A heuristic avoids repetitive structures such as loops, and temporal differential filtering removes false positives from uninitialised variables. DTD found 13 bugs in the LLVM toolchain and 5 in the GNU toolchain.",
+    "relationship_to_sqlancer": "Differential testing applied to debuggers; SQLancer is cited among the testing work it draws on.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3643779",
+    "source_sha256": "sha256:903a6009dca169e55788721052f1016b32d2e76b31a7c6ea6f1b86f2d8ac9298",
+    "supporting_excerpts": [
+      "Despite the wide usage of interactive debuggers, a scalable and comprehensive measurement of their functionality correctness does not exist yet.",
+      "Furthermore, based on these criteria, we present DTD, a differential testing (DT) framework for detecting bugs in interactive debuggers.",
+      "DTD compares the behaviors of two mainstream debuggers when processing an identical C executable — discrepancies indicate bugs in one of the two debuggers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/ebb7a876107f751507a0074cc8dbc28ded7115cf",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3643781.json b/_data/impact/paper_notes/paper_doi_10_1145_3643781.json
new file mode 100644
index 0000000..71e3b26
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3643781.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3643781",
+    "title": "Metamorphic Testing of Secure Multi-party Computation (MPC) Compilers",
+    "authors": [
+      "Yichen Li",
+      "Dongwei Xiao",
+      "Zhibo Liu",
+      "Qi Pang",
+      "Shuai Wang"
+    ],
+    "year": 2024,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3643781",
+    "url": "https://doi.org/10.1145/3643781"
+  },
+  "summary": {
+    "text": "MT-MPC applies metamorphic testing to compilers for secure multi-party computation, which translate high-level descriptions of MPC procedures into low-level executables through several intermediate representations and many optimisations. The authors note no systematic understanding of these compilers' correctness existed. They propose three metamorphic relations tailored to MPC programs, mutate high-level programs accordingly, and check whether the compiler produces semantically equivalent executables. Across three popular compilers they found 4,772 error-triggering inputs, which do not crash the compiler but yield incorrect executables, and traced thirteen bugs.",
+    "relationship_to_sqlancer": "Metamorphic testing for compilers in a privacy setting; SQLancer is cited as prior work in that tradition.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3643781",
+    "source_sha256": "sha256:f757bc960066379cbacf9ca83330b32129f5a955710df4e3147e5a67e42b9d80",
+    "supporting_excerpts": [
+      "Despite the prosperous adoption of MPC compilers by industrial vendors and academia, a principled and systematic understanding of the correctness of MPC compilers does not yet exist.",
+      "To fill this critical gap, this paper introduces MT-MPC, a metamorphic testing (MT) framework specifically designed for MPC compilers to effectively uncover erroneous compilations.",
+      "Nevertheless, we detected 4,772 inputs that can result in erroneous compilations in three popular MPC compilers available on the market."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3643781",
+          "source_type": "paper_citation_context",
+          "excerpt": "MT has been applied to a large variety of software systems, including data-base [37, 38], AI models [29, 44, 53, 54], and Cyber Physical Systems [13].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680311.json b/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680311.json
new file mode 100644
index 0000000..a52b9aa
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680311.json
@@ -0,0 +1,70 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3650212.3680311",
+    "title": "Testing Graph Database Systems with Graph-State Persistence Oracle",
+    "authors": [
+      "Shuang Liu",
+      "Junhao Lan",
+      "Xiaoning Du",
+      "Jiyuan Li",
+      "Wei Lu",
+      "Jiajun Jiang",
+      "Xiaoyong Du"
+    ],
+    "year": 2024,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3650212.3680311",
+    "url": "https://doi.org/10.1145/3650212.3680311"
+  },
+  "summary": {
+    "text": "GraspDB is a metamorphic testing approach aimed specifically at write operations in graph database systems, which the authors say existing testing approaches overlook even though adding nodes, creating relationships and modifying data are frequent in applications like social networks. Its Graph-State Persistence oracle is built on labeled property graph isomorphism and labeled property subgraph isomorphism relations, supported by three classes of mutation rules that reach more write-related code. It found 77 unique previously unknown bugs across four open-source graph engines, 58 confirmed and 31 write-related.",
+    "relationship_to_sqlancer": "Extends metamorphic oracles from reads to writes in graph databases, a direction opened by the query-oriented oracles it cites.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3650212.3680311",
+    "source_sha256": "sha256:603cbae7ab8f1578414d18f6fb5de9072a17c7369ef38cfb5dfa49285de0a4ef",
+    "supporting_excerpts": [
+      "However, existing GDBMS testing approaches tend to overlook these writing functionalities, failing to detect bugs arising from such operations.",
+      "In this paper we present GraspDB, the first metamorphic testing approach specifically designed to identify bugs related to writing operations in graph database systems.",
+      "GraspDB has successfully detected 77 unique, previously unknown bugs across four popular open source graph database engines, among which 58 bugs are confirmed by developers, 43 bugs have been fixed and 31 are related to writing operations."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:56Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3650212.3680311",
+          "source_type": "paper_citation_context",
+          "excerpt": "Another reason is due to the low duplicate bug report rate of GraspDB compared to baselines.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3650212.3680311",
+          "source_type": "paper_citation_context",
+          "excerpt": "If the DBMS fails to fetch the pivot row, it likely causes a bug in the RDBMS. Non-optimizing Reference Engine Construction (NoREC) [29] is another widely-known metamorphic testing approach to test RDBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680317.json b/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680317.json
new file mode 100644
index 0000000..0d70e4c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680317.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3650212.3680317",
+    "title": "SQLess: Dialect-Agnostic SQL Query Simplification",
+    "authors": [
+      "Li Lin",
+      "Zongyin Hao",
+      "Chengpeng Wang",
+      "Zhuangda Wang",
+      "Rongxin Wu",
+      "Gang Fan"
+    ],
+    "year": 2024,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3650212.3680317",
+    "url": "https://doi.org/10.1145/3650212.3680317"
+  },
+  "summary": {
+    "text": "SQLess simplifies bug-triggering SQL queries. The authors note that to reach deep bugs most testing techniques generate long, complex queries, leaving developers to debug them. Unlike previous simplifiers that depend on DBMS-specific grammar, SQLess uses an adaptive parser with error recovery and grammar expansion to handle different dialects, and performs semantics-sensitive trimming using alias and dependency analysis so the query still triggers the bug. On over 32,000 complex queries from state-of-the-art bug detection studies across six DBMSs it achieved an average simplification rate of 72.45%.",
+    "relationship_to_sqlancer": "Reduces the queries produced by DBMS testing tools, drawing its evaluation data from studies including SQLancer's.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3650212.3680317",
+    "source_sha256": "sha256:bc46f1de9027f9aaa54bb68ab646178cb5dd18a54cccd1829f56aa8c8e3c6cac",
+    "supporting_excerpts": [
+      "However, to trigger deep bugs, most of the existing techniques focus on generating lengthy and complex queries which burdens developers with the difficult of debugging.",
+      "Therefore, SQL query simplification, which aims to reduce lengthy SQL queries without compromising their ability to detect bugs, is highly demanded.",
+      "The results demonstrate SQLess’s superior performance: it achieves an average simplification rate of 72.45%, which significantly outperforms the stateof-the-art approaches by 84.91%."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3650212.3680317",
+          "source_type": "paper_citation_context",
+          "excerpt": "They typically suffer from a dilemma that the malformed, complex, and lengthy queries are more likely to trigger bugs but are less likely to be accepted by DBMS developers, which is highlighted by many prior studies [7, 9, 30].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680318.json b/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680318.json
new file mode 100644
index 0000000..af4204a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680318.json
@@ -0,0 +1,112 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3650212.3680318",
+    "title": "DBStorm: Generating Various Effective Workloads for Testing Isolation Levels",
+    "authors": [
+      "Keqiang Li",
+      "Siyang Weng",
+      "Lyu Ni",
+      "Chengcheng Yang",
+      "Rong Zhang",
+      "Xuan Zhou",
+      "Aoying Zhou"
+    ],
+    "year": 2024,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3650212.3680318",
+    "url": "https://doi.org/10.1145/3650212.3680318"
+  },
+  "summary": {
+    "text": "DBStorm generates workloads for testing isolation levels, which act as the correctness contract between applications and DBMSs. The authors identify three requirements — black-box testing of closed-source systems, avoiding redundant and ineffective tests, and distinguishing between isolation levels — and address them by studying 15 popular DBMSs, finding they share a generic conflict-graph framework. DBStorm adds lightweight data state mirroring so generated operations touch exactly the intended records, a history-independent method for producing dissimilar conflict graphs, and implantation-based orchestration of conflicting accesses. It found 33 bugs.",
+    "relationship_to_sqlancer": "Isolation-level testing alongside the query-result testing SQLancer performs, which it cites.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3650212.3680318",
+    "source_sha256": "sha256:f21476e19d09b0b2ea524ed75552576bd1d6c2c197365e4ab759061911199aaf",
+    "supporting_excerpts": [
+      "Isolation level (IL) acts as a correctness contract between applications and DBMSs.",
+      "For black-box testing, we investigate the IL implementations of 15 popular DBMSs and discover that they follow a generic framework that utilizes conflict graphs to manage all conflicts of a workload, and performs a verification policy to prevent non-serializable anomalies.",
+      "Practically, we have successfully found 33 bugs in popular DBMSs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/e54a5920715b453038a64e65a8c06fa4ce1199db",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Query Plan Guidance (QPG).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/1b209b2da737fd26c50a2cb9116c73c9d1db904e",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compare DBStorm with three state-of-the-art DBMS fuzzers on workload validity and code coverage, including two generationbased fuzzers SQLSmith [12] and SQLancer [37], and one mutationbased fuzzer Squirrel [50].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/1b209b2da737fd26c50a2cb9116c73c9d1db904e",
+          "source_type": "paper_citation_context",
+          "excerpt": ", primary-foreign key constraints, so the generated workload should comply with these constraints, which is a tough thing considering the generation efficiency [18, 37, 41, 50].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/1b209b2da737fd26c50a2cb9116c73c9d1db904e",
+          "source_type": "paper_citation_context",
+          "excerpt": "Though an automatic workload generation is imperative, current work is mainly designed for query processing instead of transaction processing [12, 28, 37, 41, 50].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/1b209b2da737fd26c50a2cb9116c73c9d1db904e",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer [37] generates queries to fetch a specific target row.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/1b209b2da737fd26c50a2cb9116c73c9d1db904e",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compare DBStorm with three state-of-the-art DBMS fuzzers on workload validity and code coverage, including two generationbased fuzzers SQLSmith [12] and SQLancer [37], and one mutationbased fuzzer Squirrel [50].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Pivoted Query Synthesis (PQS) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680392.json b/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680392.json
new file mode 100644
index 0000000..eccf22f
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3650212_3680392.json
@@ -0,0 +1,66 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3650212.3680392",
+    "title": "Testing Gremlin-Based Graph Database Systems via Query Disassembling",
+    "authors": [
+      "Yingying Zheng",
+      "Wensheng Dou",
+      "Leile Tang",
+      "Ziyu Cui",
+      "Yu Gao",
+      "Jiansen Song",
+      "Liangying Xu",
+      "Jiaxin Zhu",
+      "Wei Wang",
+      "Jun Wei",
+      "Hua Zhong",
+      "Tao Huang"
+    ],
+    "year": 2024,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3650212.3680392",
+    "url": "https://doi.org/10.1145/3650212.3680392"
+  },
+  "summary": {
+    "text": "QuDi detects logic bugs in Gremlin-based graph database systems by disassembling a query. Given a Gremlin query, it breaks the query into a sequence of atomic graph traversals with equivalent execution semantics; if the whole query and the disassembled sequence return different results, the system has a logic bug. Evaluated on six popular graph systems it found 25 logic bugs, 10 confirmed as previously unknown.",
+    "relationship_to_sqlancer": "A semantics-preserving decomposition oracle for graph databases, in the family of SQLancer's query-transformation oracles.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3650212.3680392",
+    "source_sha256": "sha256:4f7f3333f09380035c2a4bd80e3d68f59d74d93701ded83249ec9e4f4dcbab96",
+    "supporting_excerpts": [
+      "However, incorrect implementations and optimizations of GDBs can introduce logic bugs, which can cause Gremlin queries to return incorrect query results, e.g., omitting vertices in a graph database.",
+      "In this paper, we propose Query Di sassembling (QuDi), an effective testing technique to automatically detect logic bugs in Gremlin-based GDBs.",
+      "Given a Gremlin query Q, QuDi disassembles Q into a sequence of atomic graph traversals TList, which shares the equivalent execution semantics with Q."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:56Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/dce7daf3d485e83ae503cfa60e7a0ccce994a764",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3689031_3696064.json b/_data/impact/paper_notes/paper_doi_10_1145_3689031_3696064.json
new file mode 100644
index 0000000..7618342
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3689031_3696064.json
@@ -0,0 +1,69 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3689031.3696064",
+    "title": "Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs",
+    "authors": [
+      "Jingzhou Fu",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Yanyang Zhao",
+      "Shanshan Li",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "European Conference on Computer Systems",
+    "doi": "10.1145/3689031.3696064",
+    "url": "https://doi.org/10.1145/3689031.3696064"
+  },
+  "summary": {
+    "text": "A study of 318 bugs in built-in SQL functions, followed by a testing tool derived from it. The authors find that 87.4% of these bugs come from improper handling of boundary values of arguments, arriving from three sources — literal values, type castings and nested functions — and distil ten SQL patterns of bug-inducing queries. Their tool, Soft, applies those patterns to seven widely used DBMSs and found 132 previously unknown function bugs; vendors fixed 97 within three days.",
+    "relationship_to_sqlancer": "Targets built-in functions, which the authors argue general DBMS testing efforts struggle to exercise, citing that work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3689031.3696064",
+    "source_sha256": "sha256:c1af51e30f1c22bb64cd95bc0cdc15c87e902aeb26920bf97b060597fa0b4864",
+    "supporting_excerpts": [
+      "More importantly, conventional function testing methods struggle to generate semantically correct SQL test cases, while DBMS testing efforts are hard to measure built-in SQL functions.",
+      "Our investigation reveals that 87.4% of these bugs were caused by improper handling of boundary values of arguments.",
+      "Soft discovered and confirmed 132 previously unknown SQL function bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:32:20Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3689031.3696064",
+          "source_type": "paper_citation_context",
+          "excerpt": "To address these issues, we developed Soft, a tool that leverages the patterns identified in our study to generate SQL test cases that effectively target these boundary conditions.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3689031.3696064",
+          "source_type": "paper_citation_context",
+          "excerpt": "-- Case 1. global buffer overflow in MySQL SELECT AVG (1.2999999999999999999999999999999999999 999999999999999999999999999999999999); Case 1: A global buffer overflow in MySQL.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3689491_3691821.json b/_data/impact/paper_notes/paper_doi_10_1145_3689491_3691821.json
new file mode 100644
index 0000000..83f55aa
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3689491_3691821.json
@@ -0,0 +1,74 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3689491.3691821",
+    "title": "Step-wise Execution of Data-Centric Systems",
+    "authors": [
+      "Chi Zhang"
+    ],
+    "year": 2024,
+    "venue": "SPLASH Companion",
+    "doi": "10.1145/3689491.3691821",
+    "url": "https://doi.org/10.1145/3689491.3691821"
+  },
+  "summary": {
+    "text": "This paper proposes a black-box methodology for testing data-centric systems, motivated by the observation that existing methods suffer from ineffective oracles, insufficient coverage of target functionality and inefficient test generation. Its key idea is to build the test case incrementally and use the intermediate results to construct a reference test case, whose result must match the original's. The authors applied it to both Datalog engines and DBMSs, uncovering 75 unique bugs, and suggest the idea may transfer to deep learning libraries.",
+    "relationship_to_sqlancer": "Constructs a reference from a query's own intermediate results, treating SQLancer's oracles as the state of the art it builds on.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3689491.3691821",
+    "source_sha256": "sha256:04ecc3d312bf3045adb38166e8fc1e7ae64fe37e77cda98349468e8f62623f57",
+    "supporting_excerpts": [
+      "Existing methods for testing data-centric systems face challenges such as ineffective test oracles, insufficient coverage of target functionalities, and low efficiency in test case generation.",
+      "Our key insight is that we can incrementally generate the test case and use the intermediate results to construct the reference test case.",
+      "We have applied this methodology to test both Datalog engines and DBMSs, successfully uncovering 75 unique bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3689491.3691821",
+          "source_type": "paper_citation_context",
+          "excerpt": "DBMS. NoREC [7], TLP [8], and TQS [9] are all state-of-the-art approaches for detecting logic bugs in DBMSs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3689491.3691821",
+          "source_type": "paper_citation_context",
+          "excerpt": "DBMS. NoREC [7], TLP [8], and TQS [9] are all state-of-the-art approaches for detecting logic bugs in DBMSs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Non-optimizing Reference Engine Construction (NoREC) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3689757.json b/_data/impact/paper_notes/paper_doi_10_1145_3689757.json
new file mode 100644
index 0000000..3f9e1d2
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3689757.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3689757",
+    "title": "PolyJuice: Detecting Mis-compilation Bugs in Tensor Compilers with Equality Saturation Based Rewriting",
+    "authors": [
+      "Chijin Zhou",
+      "Bingzhou Qian",
+      "Gwihwan Go",
+      "Quan Zhang",
+      "Shanshan Li",
+      "Yu Jiang"
+    ],
+    "year": 2024,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3689757",
+    "url": "https://doi.org/10.1145/3689757"
+  },
+  "summary": {
+    "text": "PolyJuice detects mis-compilation bugs in tensor compilers by constructing semantically equivalent computation graphs and checking the compiler treats them alike. The difficulty is producing equivalent graphs that actually exercise different optimisation logic, which the authors address in two ways: arithmetic and structural rewrite rules that change a tensor program's dataflow, and an equality saturation based framework that finds the most simplified and most complex equivalent graphs for an input. Across five industrial and two academic tensor compilers it found 84 non-crash mis-compilation bugs, 49 confirmed.",
+    "relationship_to_sqlancer": "Equivalence-based testing for silent wrong results in tensor compilers, the oracle pattern SQLancer established for DBMSs.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3689757",
+    "source_sha256": "sha256:b552336c2698b3af16ce9f2ac3d357b0e62b120d6e0d933097dd3e276d40766c",
+    "supporting_excerpts": [
+      "This paper introduces PolyJuice, an automatic detection tool for identifying mis-compilation bugs in tensor compilers.",
+      "Its basic idea is to construct semantically-equivalent computation graphs to validate the correctness of tensor compilers.",
+      "In total, PolyJuice detected 84 non-crash mis-compilation bugs, out of which 49 were confirmed with 20 fixed."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/9889059fce1930bbb29d7cbcc33e5fc1ed168f2a",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3690631.json b/_data/impact/paper_notes/paper_doi_10_1145_3690631.json
new file mode 100644
index 0000000..20d34b2
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3690631.json
@@ -0,0 +1,60 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3690631",
+    "title": "T-Rec: Fine-Grained Language-Agnostic Program Reduction Guided by Lexical Syntax",
+    "authors": [
+      "Zhenyang Xu",
+      "Yongqiang Tian",
+      "Mengxiao Zhang",
+      "Jiarui Zhang",
+      "Puzhuo Liu",
+      "Yu Jiang",
+      "Chengnian Sun"
+    ],
+    "year": 2024,
+    "venue": "ACM Transactions on Software Engineering and Methodology",
+    "doi": "10.1145/3690631",
+    "url": "https://doi.org/10.1145/3690631"
+  },
+  "summary": {
+    "text": "T-Rec is a language-agnostic program reduction technique guided by lexical syntax. Program reduction removes bug-irrelevant code from a bug-triggering program, giving a smaller reproducer and making duplicates easier to tell apart, which matters most for language toolchains such as compilers, interpreters and debuggers. The authors observe that existing language-agnostic reducers treat tokens as atomic and so miss reduction opportunities inside them; T-Rec uses the language's lexical syntax to reduce within tokens, removing up to 65.52% and 53.73% more bytes than Perses and Vulcan respectively on a multi-lingual benchmark.",
+    "relationship_to_sqlancer": "Reduces bug-triggering programs, the step that follows automated bug finding of the kind SQLancer performs, which it cites.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3690631",
+    "source_sha256": "sha256:3cc611c82f3ddbf1ae6133551dfb7b921b051668f78bfc144f924afee0b99ad7",
+    "supporting_excerpts": [
+      "With such reduction and canonicalization functionality, program reduction facilitates debugging for software, especially language toolchains, such as compilers, interpreters, and debuggers.",
+      "While many program reduction techniques have been proposed, most of them (especially the language-agnostic ones) overlooked the potential reduction opportunities hidden within tokens.",
+      "To fill this gap, we propose \\(\\mathsf{T}\\) - \\(\\mathsf{Rec}\\) , a fine-grained language-agnostic program reduction technique guided by lexical syntax."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:56Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3690631",
+          "source_type": "paper_citation_context",
+          "excerpt": "Concretely, T - Rec Perses takes 24.22%, 31.22%, and 56.46% more time than Perses on C, Rust,and SMT-LIBv2 benchmarks, respectively, and T - Rec Vulcan takes 1.59%, 5.04%, and 5.01% more time than Vulcan on each benchmark.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3704870.json b/_data/impact/paper_notes/paper_doi_10_1145_3704870.json
new file mode 100644
index 0000000..a29ce1c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3704870.json
@@ -0,0 +1,64 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3704870",
+    "title": "The Decision Problem for Regular First Order Theories",
+    "authors": [
+      "Umang Mathur",
+      "David Mestel",
+      "Mahesh Viswanathan"
+    ],
+    "year": 2024,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3704870",
+    "url": "https://doi.org/10.1145/3704870"
+  },
+  "summary": {
+    "text": "A theory paper extending the classical decision problem from a single first-order formula to regular first-order theories, that is, infinite regular sets of formulae. Building on the established classification of syntactic classes as decidable or undecidable, the authors show that some classes decidable in the classical setting — the EPR and Gurevich classes — become undecidable here, and identify a decidable subclass of each, leaving a complete classification open. They also observe that the problem generalises prior automata-theoretic verification of uninterpreted programs.",
+    "relationship_to_sqlancer": "A decidability result rather than testing work; the connection is the citation recorded in the citation graph.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3704870",
+    "source_sha256": "sha256:ea8ba263a84b8bc2a773821e090912ecfde9c979f19432d63132b836a54eb8c3",
+    "supporting_excerpts": [
+      "In this work, we consider an extension of this problem to regular first-order theories, i.e., (infinite) regular sets of formulae.",
+      "Building on the elegant classification of syntactic classes as decidable or undecidable for the classical decision problem, we show that some classes (specifically, the EPR and Gurevich classes), which are decidable in the classical setting, become undecidable for regular theories.",
+      "Finally, we observe that our problem generalises prior work on automata-theoretic verification of uninterpreted programs and propose a semantic class of existential formulae for which the problem is decidable."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:56Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3704870",
+          "source_type": "paper_citation_context",
+          "excerpt": "Likewise, an effective test suite for testing database management systems must contain SQL queries that return different types of answers [Rigger and Su 2020].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3704870",
+          "source_type": "paper_citation_context",
+          "excerpt": "Finally, in software testing, the goal is often to design a test suite that is diverse and exercises different paths of the program under test.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3708533.json b/_data/impact/paper_notes/paper_doi_10_1145_3708533.json
new file mode 100644
index 0000000..81bf96d
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3708533.json
@@ -0,0 +1,67 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3708533",
+    "title": "Software Security Analysis in 2030 and Beyond: A Research Roadmap",
+    "authors": [
+      "Marcel Böhme",
+      "Eric Bodden",
+      "T. Bultan",
+      "Cristian Cadar",
+      "Yang Liu",
+      "Giuseppe Scanniello"
+    ],
+    "year": 2024,
+    "venue": "ACM Transactions on Software Engineering and Methodology",
+    "doi": "10.1145/3708533",
+    "url": "https://doi.org/10.1145/3708533"
+  },
+  "summary": {
+    "text": "A roadmap article on software security analysis for the systems of the coming decade. It asks what vulnerabilities future systems will have and how to detect them, how to find bugs once the shallow ones are gone, and how to protect systems when not all flaws can be found — in a setting where code is increasingly co-written by generative AI, systems are heterogeneous with automatically generated components, and dependencies span whole software supply chains. The paper surveys recent advances, discusses open challenges, and closes with a long-term perspective.",
+    "relationship_to_sqlancer": "A research roadmap citing SQLancer among the recent advances in automated testing it surveys.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3708533",
+    "source_sha256": "sha256:b16a7c2b73a7855e62f0469c70b08eb35436d8924fd1b7d746b6b6acbe8d1cc5",
+    "supporting_excerpts": [
+      "In this roadmap article, we outline our vision of software security analysis for the systems of the future.",
+      "When all the shallow bugs are found, how do we discover vulnerabilities hidden deeply in the system?",
+      "To answer these questions, we start our roadmap with a survey of recent advances in software security, then discuss open challenges and opportunities, and conclude with a long-term perspective for the field."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3708533",
+          "source_type": "paper_citation_context",
+          "excerpt": "Specifications have been for long proposed as a way to write correct software, but adoption has been limited, particularly in mainstream languages.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3708533",
+          "source_type": "paper_citation_context",
+          "excerpt": "For easy reference, each challenge or opportunity is identified uniquely using a counter and a name.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3720504.json b/_data/impact/paper_notes/paper_doi_10_1145_3720504.json
new file mode 100644
index 0000000..8785188
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3720504.json
@@ -0,0 +1,58 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3720504",
+    "title": "Checking Observational Correctness of Database Systems",
+    "authors": [
+      "Lauren Pick",
+      "Amanda Xu",
+      "Ankush Desai",
+      "S. Seshia",
+      "Aws Albarghouthi"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3720504",
+    "url": "https://doi.org/10.1145/3720504"
+  },
+  "summary": {
+    "text": "This paper checks database systems as a client would, verifying both transaction semantics and isolation guarantees from observations of running transactions against a black-box system. Observational correctness holds if some correct execution under the given isolation level could have produced the observations. The technique rests on symbolic encodings of transaction semantics under weak isolation and of isolation-level guarantees, which are discharged with an SMT solver. The tool, Troubadour, was applied to PostgreSQL and an industrial system under development, finding two new bugs.",
+    "relationship_to_sqlancer": "Verifies isolation and transaction semantics rather than single-query results, citing SQLancer's work on DBMS correctness.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3720504",
+    "source_sha256": "sha256:fdfee740df88193de53b0a39359b63ed39d8e6be49cb9bc813a757a4b5e5506b",
+    "supporting_excerpts": [
+      "This paper presents a clientcentric technique for checking both semantic correctness and isolation-level guarantees for black-box database systems based on observations collected from running transactions on these systems.",
+      "Our technique verifies observational correctness with respect to a given set of transactions and observations for them, which holds iff there exists a possible correct execution of the transactions under a given isolation level that could result in these observations.",
+      "We applied our tool Troubadour to verify observational correctness of several database systems, including PostgreSQL and an industrial system under development, in which the tool helped detect two new bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3720504",
+          "source_type": "paper_citation_context",
+          "excerpt": "There has been much work on DBMS fuzzing for SQL using a variety of techniques [3, 15, 31, 42], such as ternary logic partitioning and mutation-based fuzzing.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3728908.json b/_data/impact/paper_notes/paper_doi_10_1145_3728908.json
new file mode 100644
index 0000000..c29f90d
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3728908.json
@@ -0,0 +1,120 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3728908",
+    "title": "QTRAN: Extending Metamorphic-Oracle Based Logical Bug Detection Techniques for Multiple-DBMS Dialect Support",
+    "authors": [
+      "Li Lin",
+      "Qinglin Zhu",
+      "Hongqiao Chen",
+      "Zhuangda Wang",
+      "Rongxin Wu",
+      "Xiaoheng Xie"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3728908",
+    "url": "https://doi.org/10.1145/3728908"
+  },
+  "summary": {
+    "text": "QTRAN extends existing metamorphic-oracle logic bug detection techniques to DBMSs they were never written for. The obstacle is that these techniques depend on a specific DBMS's grammar to build valid statement pairs, so only a few systems are supported and extending them takes considerable manual work. QTRAN uses an LLM in two phases: a transfer phase that identifies dialect differences and uses SQL documentation to translate the original queries, and a mutation phase using a fine-tuned model to mutate them while preserving the metamorphic relation. Applied to four techniques across eight DBMSs, over 99% of transferred pairs satisfied the relations, and it found 24 logic bugs.",
+    "relationship_to_sqlancer": "Takes SQLancer's metamorphic oracles as the state-of-the-art techniques it extends, translating their statement pairs to DBMSs those oracles do not support.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3728908",
+    "source_sha256": "sha256:02ae086398b897368d46b2681f8f7bc02214a8425527c4549485e03f01ec4e59",
+    "supporting_excerpts": [
+      "However, current MOLTs rely heavily on specific DBMS grammar to generate valid SQL statement pairs, which makes it challenging to adapt these techniques to various DBMSs with different grammatical structures.",
+      "In this paper, we propose QTRAN, a novel LLM-powered approach that automatically extends existing MOLTs to various DBMSs.",
+      "We implement our approach as a tool and apply it to extend four state-of-the-art MOLTs for eight DBMSs: MySQL, MariaDB, TiDB, PostgreSQL, SQLite, MonetDB, DuckDB, and ClickHouse."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3728908",
+          "source_type": "paper_citation_context",
+          "excerpt": "…Metamorphic Testing in DBMSs. Recent years have witnessed tremendous efforts in resolving the test oracle for logical bug detection in the DBMSs. Notably, the metamorphic testing based approach MOLT has been recognized to be state-of-the-art in DBMS testing for logical bug detection [20, 38, 40].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3728908",
+          "source_type": "paper_citation_context",
+          "excerpt": "In our evaluation, we selected four state-of-the-art MOLTs for extension: NoRec [37], TLP [38], Pinolo [20], and DQE [44].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3728908",
+          "source_type": "paper_citation_context",
+          "excerpt": "Notably, metamorphic oracle, a widely-used method that constructs SQL statements maintaining either exact [25, 28, 29, 37–39, 44 ] or approximate To generate SQL statement pairs suitable for MOLTs, including original queries and mutated queries , a general idea is to model the SQL grammar as an…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3728908",
+          "source_type": "paper_citation_context",
+          "excerpt": "The evaluation results show that over 99% of the SQL statement pairs transferred by QTRAN satisfy the metamorphic relations required for testing, and have detected 24 logical bugs across several DBMSs, with 16 confirmed as unique and previously unknown.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3728908",
+          "source_type": "paper_citation_context",
+          "excerpt": "However, the complex code logic and diverse functionalities of DBMSs often make them susceptible to bugs, especially logical bugs that result in incorrect result sets being returned without obvious symptoms [8, 20, 24, 26, 37–39].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3728908",
+          "source_type": "paper_citation_context",
+          "excerpt": "Below, we detail each of these tools: (1) NoRec [37]: This technique involves transferring predicates from the WHERE clause to the SELECT clause.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3728908",
+          "source_type": "paper_citation_context",
+          "excerpt": "In our evaluation, we selected four state-of-the-art MOLTs for extension: NoRec [37], TLP [38], Pinolo [20], and DQE [44].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Non-optimizing Reference Engine Construction (NoREC) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3728953.json b/_data/impact/paper_notes/paper_doi_10_1145_3728953.json
new file mode 100644
index 0000000..6560766
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3728953.json
@@ -0,0 +1,63 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3728953",
+    "title": "Detecting Isolation Anomalies in Relational DBMSs",
+    "authors": [
+      "Rui Yang",
+      "Ziyu Cui",
+      "Wensheng Dou",
+      "Yu Gao",
+      "Jiansen Song",
+      "Xudong Xie",
+      "Jun Wei"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3728953",
+    "url": "https://doi.org/10.1145/3728953"
+  },
+  "summary": {
+    "text": "IsoRel is a black-box isolation checker for relational DBMSs. Existing checkers work only on key-value data models with simple read and write operations, so they cannot be applied to relational systems with complex SQL. IsoRel instruments SQL statements in an isolation-agnostic way, using two auxiliary columns per table to record which rows each statement touched, then builds a transaction dependency graph from those records and identifies anomalies by pattern. Across MySQL, PostgreSQL, MariaDB, CockroachDB and TiDB it found 48 unique isolation anomalies violating the levels defined by Adya.",
+    "relationship_to_sqlancer": "Checks isolation-level correctness, a dimension alongside the query-result correctness SQLancer targets.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3728953",
+    "source_sha256": "sha256:aee03a55e2c6b17beebd4f255af0659c1218cd3b9f9897fedbd20af2dfce7b17",
+    "supporting_excerpts": [
+      "Existing isolation checkers can only work on simple key-υalue-like data models and the associated read (key) and write (key, υalue) operations.",
+      "In this paper, we propose a novel black-box Isolation checker for Relational DBMSs, IsoRel, which can support relational data models and complex SQL operations.",
+      "Our evaluation reveals a total of 48 unique isolation anomalies that violate the isolation levels defined by Adya."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3728953",
+          "source_type": "paper_citation_context",
+          "excerpt": "Rigger et al. propose SQLancer [12] and several approaches, e.g., PQS [53], TLP [52] and NoREC [51], to detect logic bugs in SELECT statements.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3728965.json b/_data/impact/paper_notes/paper_doi_10_1145_3728965.json
new file mode 100644
index 0000000..de701b7
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3728965.json
@@ -0,0 +1,62 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3728965",
+    "title": "DepState: Detecting Synchronization Failure Bugs in Distributed Database Management Systems",
+    "authors": [
+      "Cundi Fang",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Zhouyang Jia",
+      "Chun Huang",
+      "Yu Jiang",
+      "Shanshan Li"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3728965",
+    "url": "https://doi.org/10.1145/3728965"
+  },
+  "summary": {
+    "text": "DepState tests the synchronization process in distributed DBMSs, which keeps data consistent as data and cluster state change and whose bugs can cause inconsistency, transaction errors or cluster crashes. It simulates the complexity of data sharding and dynamic cluster conditions, establishing dependencies between tables across nodes and introducing controlled variation in cluster state. On MySQL NDB Cluster, MySQL InnoDB Cluster, MariaDB Galera Cluster and TiDB Cluster it found 25 new bugs, 13 confirmed.",
+    "relationship_to_sqlancer": "Compares directly against SQLancer among its baselines, reporting more synchronization failure bugs and higher coverage of synchronization-related functions.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3728965",
+    "source_sha256": "sha256:06c9e57c2ed512d1dc2b8b71c289360efadbeb4efcd80a14086a9d7f7d8cd253",
+    "supporting_excerpts": [
+      "In this paper, we propose DepState, a framework to detect synchronization failure bugs.",
+      "We utilize DepState on four DDBMSs: MySQL NDB Cluster, MySQL InnoDB Cluster, MariaDB Galera Cluster, and TiDB Cluster, discovering 25 new bugs, with 13 confirmed.",
+      "DepState finds 14 more synchronization failure bugs and covers 6.13%-66.51%, 5.82%-57.28%, 14.12%-83.30%, 36.81%-83.88%, and 43.24%-54.28% more lines in synchronization-related functions than Jepsen, Mallory, SQLsmith, SQLancer, and Mozi in 24 hours, respectively."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/810786dd896fe2eef0a129db6e3b115c95375130",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3728973.json b/_data/impact/paper_notes/paper_doi_10_1145_3728973.json
new file mode 100644
index 0000000..f1826ab
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3728973.json
@@ -0,0 +1,88 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3728973",
+    "title": "Hulk: Exploring Data-Sensitive Performance Anomalies in DBMSs via Data-Driven Analysis",
+    "authors": [
+      "Zhiyong Wu",
+      "Jie Liang",
+      "Jingzhou Fu",
+      "Mingzhe Wang",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3728973",
+    "url": "https://doi.org/10.1145/3728973"
+  },
+  "summary": {
+    "text": "Hulk looks for performance anomalies that depend on the data rather than the query. The authors argue that developers diagnose performance issues by intuition or by comparing execution time against a baseline DBMS, and that both overlook how the dataset affects performance. Hulk instead watches performance as the dataset grows, estimating a reasonable response-time range for each data volume to locate performance cliffs, then checks whether a cliff deviates from expectation by finding a plan consistent with that expectation. Across six DBMSs it reported 135 anomalies, 129 confirmed as new bugs including 14 CVEs.",
+    "relationship_to_sqlancer": "Performance-anomaly detection for DBMSs, citing SQLancer's work as the state of the art in automated DBMS testing.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3728973",
+    "source_sha256": "sha256:99ae27394eea5f82f89917560967da64414b24a0d23aa41eaacb1f6ac24478ac",
+    "supporting_excerpts": [
+      "To diagnose performance issues, DBMS developers often rely on intuitions or compare execution times to a baseline DBMS.",
+      "In this paper, we propose Hulk to automatically explore these data-sensitive performance anomalies via data-driven analysis.",
+      "Hulk totally reports 135 anomalies, with 129 have been confirmed as new bugs, including 14 CVEs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3728973",
+          "source_type": "paper_citation_context",
+          "excerpt": "In addition, we compare Hulk with the state-of-the-art DBMS validation tools in industry, including both DBMS performance testing tool APOLLO [30] and SQLancer [11], as well as DBMS fuzzing tools Sqirrel [60].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3728973",
+          "source_type": "paper_citation_context",
+          "excerpt": "Additionally, CERT 1 [11] tests cardinality estimation to find performance anomalies.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3728973",
+          "source_type": "paper_citation_context",
+          "excerpt": "In addition, we compare Hulk with the state-of-the-art DBMS validation tools in industry, including both DBMS performance testing tool APOLLO [30] and SQLancer [11], as well as DBMS fuzzing tools Sqirrel [60].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Cardinality Estimation Restriction Testing (CERT) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3729175.json b/_data/impact/paper_notes/paper_doi_10_1145_3729175.json
new file mode 100644
index 0000000..84fa00b
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3729175.json
@@ -0,0 +1,61 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3729175",
+    "title": "Systems Correctness Practices at Amazon Web Services",
+    "authors": [
+      "Marc Brooker",
+      "A. Desai"
+    ],
+    "year": 2025,
+    "venue": "Communications of the ACM",
+    "doi": "10.1145/3729175",
+    "url": "https://doi.org/10.1145/3729175"
+  },
+  "summary": {
+    "text": "An account of how Amazon Web Services applies formal and semi-formal methods to systems correctness. Only the sentence citing SQLancer is available here, so this summary is limited to what it shows: for the database engines described, large volumes of random SQL schemas, datasets and queries are synthesised and executed against the engines under test, with results checked against an oracle derived from the non-sharded version of the engine alongside other validation approaches.",
+    "relationship_to_sqlancer": "Describes SQLancer as having pioneered validation approaches that AWS uses alongside its own non-sharded reference oracle — an industrial account of the technique in production use.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://doi.org/10.1145/3729175",
+    "source_sha256": "sha256:8a8cad3f062195c0bed07ea971aa7a2583c85309ccc592f24b1db6f0a6ab3c10",
+    "supporting_excerpts": [
+      "Large volumes of random SQL schemas, datasets, and queries are synthesized and run through the engines under test, and the results compared with an oracle based on the non-sharded version of the engine (as well as other approaches to validation, like those pioneered by SQLancer 23 )."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:32:20Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3729175",
+          "source_type": "paper_citation_context",
+          "excerpt": "Large volumes of random SQL schemas, datasets, and queries are synthesized and run through the engines under test, and the results compared with an oracle based on the non-sharded version of the engine (as well as other approaches to validation, like those pioneered by SQLancer 23 ).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://openalex.org/W4407131558",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "OpenAlex records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3729319.json b/_data/impact/paper_notes/paper_doi_10_1145_3729319.json
new file mode 100644
index 0000000..31cc290
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3729319.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3729319",
+    "title": "Graphiti: Bridging Graph and Relational Database Queries",
+    "authors": [
+      "Yang He",
+      "Ruijie Fang",
+      "Işıl Dillig",
+      "Yuepeng Wang"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3729319",
+    "url": "https://doi.org/10.1145/3729319"
+  },
+  "summary": {
+    "text": "Graphiti checks whether a Cypher graph query and a SQL relational query are equivalent. To make that question well-posed across two data models the authors introduce database transformers, which map database instances between the graph and relational representations, and define equivalence modulo a given transformer. The verification reduces to checking equivalence of two SQL queries by translating a subset of Cypher into SQL syntax-directedly, so existing SQL reasoning applies. Graphiti proved useful for both verification and refutation, uncovering subtle bugs including some in Cypher tutorials and academic papers.",
+    "relationship_to_sqlancer": "Proves query equivalence across data models, the property that SQLancer's oracles test for empirically.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3729319",
+    "source_sha256": "sha256:a9337c7c0ab58813fd5a84a6f95b45bab59722ba54cb93793fb4e846baafbcfb",
+    "supporting_excerpts": [
+      "This paper presents an automated reasoning technique for checking equivalence between graph database queries written in Cypher and relational queries in SQL.",
+      "To formalize a suitable notion of equivalence in this setting, we introduce the concept of database transformers, which transform database instances between graph and relational models.",
+      "Our experiments demonstrate that Graphiti is useful both for verification and refutation and that it can uncover subtle bugs, including those found in Cypher tutorials and academic papers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3729319",
+          "source_type": "paper_citation_context",
+          "excerpt": "Future work can further extend the transpilation rules and backend equivalence verifiers to support additional features.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3731569_3764841.json b/_data/impact/paper_notes/paper_doi_10_1145_3731569_3764841.json
new file mode 100644
index 0000000..cccc77d
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3731569_3764841.json
@@ -0,0 +1,62 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3731569.3764841",
+    "title": "Fawkes: Finding Data Durability Bugs in DBMSs via Recovered Data State Verification",
+    "authors": [
+      "Zhiyong Wu",
+      "Jie Liang",
+      "Jingzhou Fu",
+      "Wenqian Deng",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "Symposium on Operating Systems Principles",
+    "doi": "10.1145/3731569.3764841",
+    "url": "https://doi.org/10.1145/3731569.3764841"
+  },
+  "summary": {
+    "text": "Fawkes studies and detects data durability bugs, where committed data does not survive faults such as power failures. The authors first examine 43 such bugs across four DBMSs, finding they surface as data loss, inconsistency, log corruption or unavailability, usually stem from flawed durability and recovery mechanisms, and are triggered by faults during filesystem or kernel-level operations. Fawkes then combines context-aware fault injection, functionality-guided fault triggering and checkpoint-based data graph verification, finding 48 previously unknown durability bugs across eight DBMSs, eight with CVEs.",
+    "relationship_to_sqlancer": "Targets durability and recovery rather than query semantics; SQLancer is cited among DBMS testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3731569.3764841",
+    "source_sha256": "sha256:9c144a92188ad14d5d0aae559e97fbaa7937505509b5e27649e13acc495ee6aa",
+    "supporting_excerpts": [
+      "Furthermore, existing testing methods(e.g., Mallory) are often inadequate for detecting DDBs, particularly those that cause data loss or data inconsistency following DBMS failures.",
+      "Based on these findings, we developed Fawkes, a testing framework to detect DDBs with recovered data state verification.",
+      "We applied Fawkes to eight popular DBMSs and discovered 48 previously unknown DDBs, of which 16 have been fixed and 8 have been assigned CVE identifiers due to the severity."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp",
+        "coddtest"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/e8506c9f0e513b48630e0d8329746e4fc73a99af",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Differential Query Plans (DQP), Constant-Optimization-Driven Testing (CODDTest).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3744916_3773102.json b/_data/impact/paper_notes/paper_doi_10_1145_3744916_3773102.json
new file mode 100644
index 0000000..598a4e6
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3744916_3773102.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3744916.3773102",
+    "title": "Locus: Agentic Predicate Synthesis for Directed Fuzzing",
+    "authors": [
+      "Jie Zhu",
+      "Chihao Shen",
+      "Ziyang Li",
+      "Jiahao Yu",
+      "Yizheng Chen",
+      "Kexin Pei"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": "10.1145/3744916.3773102",
+    "url": "https://doi.org/10.1145/3744916.3773102"
+  },
+  "summary": {
+    "text": "Locus improves directed fuzzing, where the goal is an input reaching a specified program state. The authors argue branch distances do not precisely capture progress toward a deeply nested target and manual constraints do not generalise. Locus instead synthesises predicates that mark semantically meaningful intermediate states as milestones; instrumented into the program, they reject executions unlikely to reach the target and add coverage guidance. An agentic framework with program analysis tools synthesises and refines the predicates while keeping them strict relaxations of the target. It sped up eight fuzzers by 41.6x on average and found nine unpatched bugs.",
+    "relationship_to_sqlancer": "Directed fuzzing for general programs; SQLancer is cited as background rather than used.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3744916.3773102",
+    "source_sha256": "sha256:f14ad4ddb3f03a7d164cee64a06b613bc3fe0bae01dda2afb673f06fface46d4",
+    "supporting_excerpts": [
+      "We present Locus, a novel framework to improve the efficiency of directed fuzzing.",
+      "Our key insight is to synthesize predicates to capture fuzzing progress as semantically meaningful intermediate states, serving as milestones towards reaching the target states.",
+      "Our evaluation shows that Locus substantially improves the efficiency of eight state-of-the-art fuzzers in discovering real-world vulnerabilities, achieving an average speedup of 41.6x."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3744916.3773102",
+          "source_type": "paper_citation_context",
+          "excerpt": "However, such feedback is sometimes too sparse or indirect to reliably measure the progress, especially when there is a long chain of implicit preconditions guarding the target program states [2, 26, 36, 44, 45, 66, 73, 80, 106].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3749186.json b/_data/impact/paper_notes/paper_doi_10_1145_3749186.json
new file mode 100644
index 0000000..8f76695
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3749186.json
@@ -0,0 +1,99 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3749186",
+    "title": "Testing Graph Databases with Synthesized Queries",
+    "authors": [
+      "Zijing Yin",
+      "Si Liu",
+      "David A. Basin"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Manag. Data",
+    "doi": "10.1145/3749186",
+    "url": "https://doi.org/10.1145/3749186"
+  },
+  "summary": {
+    "text": "GQS tests graph databases against an established ground truth, which the authors say existing approaches lack, leading both to false alarms and to missed bugs. It randomly generates a graph, picks a set of properties on its elements whose key-value pairs form the expected result set, and then synthesises a query that should retrieve exactly those values; any difference between the actual result and that ground truth is a logic bug. The tool includes the first Cypher query synthesizer built for testing graph databases, and found 36 previously unknown bugs across four production systems, some undetected for five years.",
+    "relationship_to_sqlancer": "Constructs a query whose result is known in advance, the same strategy as Pivoted Query Synthesis, applied to graph databases.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3749186",
+    "source_sha256": "sha256:1514b27498fffce26efda3a06bc82a52ba044215aa9e5d0d7df0bd07680c711d",
+    "supporting_excerpts": [
+      "Despite advances in GDB testing, a common limitation of current approaches is the lack of ground truth for their test oracles.",
+      "We introduce GQS (Graph Query Synthesis), the first automated testing approach for detecting logic bugs in graph databases (GDBs) based on an established ground truth.",
+      "Our tool finds 36 previously unknown bugs across four production GDBs, of which 26 are logic bugs, with some remaining undetected for up to five years."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:32:20Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3749186",
+          "source_type": "paper_citation_context",
+          "excerpt": "…as n’.id = 4 , are sufficient to filter out undesired graph elements, more complex expressions help to extensively test GDBs. Previous work [16, 47] provides a good basis for generating complex expressions in general, like char_length(‘abc’)+sqrt(round(1.2)) , that satisfy specific value…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3749186",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, key constructs in GDBs, such as paths, neighborhoods, and recursive traversals, do not map directly to the relational structures and operators in RDBs. Similar to our approach, PQS [47], Pinolo [15], and TQS [54] test RDBs by synthesizing SQL queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3749186",
+          "source_type": "paper_citation_context",
+          "excerpt": "In line with the growing trend of using randomized testing to effectively uncover system-level bugs in general [23, 31] and relational databases in particular [20, 47, 54], these tools all generate random graph data and queries to exercise the GDB under test.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3749186",
+          "source_type": "paper_citation_context",
+          "excerpt": "The challenge lies in the current lack of ground truth on the expected execution time, where the cardinality estimation approach [45] designed for RDBs may provide a direction for future research.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3749186",
+          "source_type": "paper_citation_context",
+          "excerpt": "GDBMeter illustrates this by adopting the metamorphic oracles from TLP [46], originally designed for SQL, which partitions queries using three-valued logic.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3749186",
+          "source_type": "paper_citation_context",
+          "excerpt": "Such logic bugs are widely recognized as being more challenging to detect than database crashes or exceptions [19, 47, 54, 62].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3757347_3759132.json b/_data/impact/paper_notes/paper_doi_10_1145_3757347_3759132.json
new file mode 100644
index 0000000..fd95a45
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3757347_3759132.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3757347.3759132",
+    "title": "PANGOLIN: a Comprehensive Testing Framework for Configuration-Rich Key-Value Stores",
+    "authors": [
+      "Shaohua Duan",
+      "Sudarsun Kannan",
+      "Andrea C. Arpaci-Dusseau",
+      "Remzi H. Arpaci-Dusseau"
+    ],
+    "year": 2025,
+    "venue": "Annual Haifa Experimental Systems Conference",
+    "doi": "10.1145/3757347.3759132",
+    "url": "https://doi.org/10.1145/3757347.3759132"
+  },
+  "summary": {
+    "text": "Pangolin tests configuration-rich key-value stores. The authors first study historical bugs in five mature key-value stores over eight years, and find that most could be caught by systematically testing a small sequence of operations and configurations. Pangolin turns that into a bounded testing strategy applied across black-box and fuzzing procedures. It found 20 bugs and reproduced 443 historical ones in RocksDB, LevelDB, HyperlevelDB, BadgerDB and Redis.",
+    "relationship_to_sqlancer": "Systematic testing of key-value stores rather than SQL DBMSs; SQLancer is cited among database testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3757347.3759132",
+    "source_sha256": "sha256:a4c9a25d48801b2dddeb52fcf49f68bf28e5c8c161c40fec750e03abe32a9045",
+    "supporting_excerpts": [
+      "To better understand bugs in modern key-value stores and explore domain knowledge for efficiently identifying new ones, we first comprehensively study historical bugs in five mature key-value stores during the last eight years.",
+      "Then, we design and implement Pangolin, which is motivated by insights from our bug study, which indicated most bugs could be identified by systematically testing a small sequence of operations and configurations.",
+      "Finally, we utilize Pangolin to find 20 bugs and reproduce 443 historical bugs in five mature key-value stores (RocksDB, LevelDB, HyperlevelDB, BadgerDB, and Redis), making it an attractive supplement to handwritten test suites."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3757347.3759132",
+          "source_type": "paper_citation_context",
+          "excerpt": "Black-box testing and fuzzing have been used to identify various types of bugs in File systems and DBMSs, including crash inconsistency bugs [21, 25], logic bugs [31], and correctness bugs [3].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3758316_3763249.json b/_data/impact/paper_notes/paper_doi_10_1145_3758316_3763249.json
new file mode 100644
index 0000000..44e06ff
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3758316_3763249.json
@@ -0,0 +1,54 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3758316.3763249",
+    "title": "LLM-Assisted Dialect-Agnostic SQL Query Parsing",
+    "authors": [
+      "Junwen An"
+    ],
+    "year": 2025,
+    "venue": "SPLASH Companion",
+    "doi": "10.1145/3758316.3763249",
+    "url": "https://doi.org/10.1145/3758316.3763249"
+  },
+  "summary": {
+    "text": "SQLFlex parses SQL across dialects. Grammar-based parsers, on which query analysis and rewriting tools depend, fail on dialect-specific syntax, while large language models understand SQL but do not reliably produce accurate abstract syntax trees. SQLFlex combines the two: it tries a grammar-based parser first and, when that fails, uses an LLM to split the query into smaller parsable pieces. It parsed 96.37% of queries across eight dialects on average, and the authors demonstrate it on SQL linting and test case reduction.",
+    "relationship_to_sqlancer": "Test case reduction is one of its applications, connecting it to the DBMS testing work it cites.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3758316.3763249",
+    "source_sha256": "sha256:0433204512906224fe61564c68e26d51a663d5827c24099f9126a9931373c4b7",
+    "supporting_excerpts": [
+      "Although Large Language Models (LLMs) show promise in understanding SQL queries, they struggle in accurately generating ASTs.",
+      "To address this, we propose SQLFlex, a hybrid approach that iteratively uses a grammar-based parser and, upon failure, employs an LLM to segment the query into smaller, parsable parts.",
+      "SQLFlex successfully parsed 96.37% of queries across eight dialects on average, and demonstrated its practicality in SQL linting and test case reduction."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/541b4674b5c71530b124a319901697eb713f6746",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3764583.json b/_data/impact/paper_notes/paper_doi_10_1145_3764583.json
new file mode 100644
index 0000000..e9e8a37
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3764583.json
@@ -0,0 +1,104 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3764583",
+    "title": "Unveiling Logic Bugs in SPJG Query Optimizations within DBMS",
+    "authors": [
+      "Xiu Tang",
+      "Shijie Yang",
+      "Sai Wu",
+      "Dongxiang Zhang",
+      "Wenchao Zhou",
+      "Feifei Li",
+      "Gang Chen"
+    ],
+    "year": 2025,
+    "venue": "ACM Transactions on Database Systems",
+    "doi": "10.1145/3764583",
+    "url": "https://doi.org/10.1145/3764583"
+  },
+  "summary": {
+    "text": "TQS targets logic bugs in Select-Project-Join-Group By query optimisation, arguing that existing generation-based tools rely on random testing and so under-exercise error-prone areas such as multi-table joins and grouped aggregation. Its Data-guided Schema and Query Generation component solves the ground-truth problem for multi-table queries by deriving data from dimensionally aggregated data cubes, mapping it into a wide table, normalising that into a testing schema, and maintaining a bitmap index to track expected results. Knowledge-guided Query Space Exploration avoids repetition. Across four DBMSs it found 226 bugs in 24 hours.",
+    "relationship_to_sqlancer": "Generates queries whose correct results are known in advance, in the same tradition as Pivoted Query Synthesis, and cites SQLancer's work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3764583",
+    "source_sha256": "sha256:3ec5236450cf2c402c66817dc796eb4586d0c220722e9ddb79b00737b60bc972",
+    "supporting_excerpts": [
+      "However, existing generation-based debugging tools predominantly rely on random testing, which tends to overlook critical error-prone areas such as multi-table joining and grouped aggregation.",
+      "DSG addresses the key challenge of multi-table query debugging: how to generate ground-truth (query, result) pairs for verification.",
+      "It successfully detected 226 bugs within 24 hours, including 63 bugs in MySQL, 52 in MariaDB, 68 in TiDB, and 43 in PolarDB, respectively."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3764583",
+          "source_type": "paper_citation_context",
+          "excerpt": "Table 8 compares the join bugs detected by DQP with those TQS(join), revealing that only 7 of the bugs are truly distinct.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3764583",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [49].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3764583",
+          "source_type": "paper_citation_context",
+          "excerpt": "MySQL incorrectly retrieves the const table without finding a matching row. to the compatibility problem, SQLancer implements different approaches on different databases (PQS, TLP and DQP on MySQL and PolarDB; NoRec and DQP on MariaDB; TLP and DQP on TiDB).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3764583",
+          "source_type": "paper_citation_context",
+          "excerpt": "While DQP claims to have uncovered 26 previously unidentified bugs, our analysis shows that most of these bugs closely resemble those already found by TQS(join).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3764583",
+          "source_type": "paper_citation_context",
+          "excerpt": "DQP adopts differential testing, a strategy aimed at replicating bugs identified by TQS(join) [56], which involves ground truth verification for join queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3764583",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [50].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3769779.json b/_data/impact/paper_notes/paper_doi_10_1145_3769779.json
new file mode 100644
index 0000000..e454f96
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3769779.json
@@ -0,0 +1,84 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3769779",
+    "title": "Detecting Logic Bugs in DBMSs via Equivalent Data Construction",
+    "authors": [
+      "Wenqian Deng",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Manag. Data",
+    "doi": "10.1145/3769779",
+    "url": "https://doi.org/10.1145/3769779"
+  },
+  "summary": {
+    "text": "Equivalent Data Construction detects logic bugs in how a DBMS implements data operations — arithmetic, string manipulation, type coercion — rather than in query optimisation, which the authors say existing methods concentrate on. The insight is that replacing an operation expression with its precomputed result should not change a query's outcome, so EDC computes the operation into a derived equivalent table, rewrites the query to read the precomputed values, and treats a difference as a bug. Across six widely used DBMSs it found 52 previously unknown bugs, 38 confirmed.",
+    "relationship_to_sqlancer": "A metamorphic oracle for DBMS logic bugs, positioned against query-level strategies of the kind SQLancer introduced.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3769779",
+    "source_sha256": "sha256:294eb76c668c9758825c8b2f4cce9a17a478fd212dcdb312700d6064b28ea6eb",
+    "supporting_excerpts": [
+      "Existing logic bug detection methods primarily focus on issues introduced during query optimization by adapting query-level strategies.",
+      "To address this, we propose equivalent data construction (EDC), a novel approach to detect logic bugs in data operation implementations within DBMSs.",
+      "Our evaluation revealed 52 previously unknown bugs, of which 38 have been confirmed by developers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:28:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3769779",
+          "source_type": "paper_citation_context",
+          "excerpt": "For instance, EET [27] targets expression-level transformations, TLP [37] partitions queries based on predicate logic, and NoREC [36] rewrites queries to disable DBMS optimizations.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3769779",
+          "source_type": "paper_citation_context",
+          "excerpt": "Specifically, some approaches split or transform queries to expose discrepancies in the results (e.g., TLP [37], NoREC [36]), while others construct logically crafted SQL queries designed to return specific results in a controlled manner (e.g., PQS [38], DQE [44]).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3769779",
+          "source_type": "paper_citation_context",
+          "excerpt": "Several approaches have been proposed to detect logic bugs in DBMSs [27, 36–38, 44, 45], many of which operate at the query level and rely on various strategies to identify inconsistencies in query results.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3769779",
+          "source_type": "paper_citation_context",
+          "excerpt": "We also evaluate EDC against 3 state-of-art testing techniques, i.e., TLP [37], Radar [45], and EET [27].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3769828.json b/_data/impact/paper_notes/paper_doi_10_1145_3769828.json
new file mode 100644
index 0000000..8b5d9c1
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3769828.json
@@ -0,0 +1,173 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3769828",
+    "title": "SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis",
+    "authors": [
+      "Jinhui Lai",
+      "Chi Zhang",
+      "Bingyan Li",
+      "Chenglin Liang",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Yu Jiang",
+      "Zichen Xu"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Manag. Data",
+    "doi": "10.1145/3769828",
+    "url": "https://doi.org/10.1145/3769828"
+  },
+  "summary": {
+    "text": "SRS detects logic bugs in how a DBMS implements joins, rather than in how it optimises them. The authors argue that existing work perturbs query hints and system variables to change the optimizer's plan choice, which fails when those knobs do not influence the optimizer or when the fault lies in join implementation code unrelated to optimisation. SRS instead transforms a join query — changing join types, join orders and join conditions — so that the results of the original and transformed queries must stand in a known set relation, and treats a violated relation as a bug. Across five extensively tested DBMSs it found 33 previously unknown bugs, all confirmed.",
+    "relationship_to_sqlancer": "Built directly on SQLancer: the authors state they implemented SRS on top of it, generate the database state with SQLancer's syntax-rule-based random generation, and keep SQLancer's query generation strategy for the NoREC oracle where joins are not involved. SQLancer's oracles are also its baselines — PQS, TLP and NoREC among the six state-of-the-art approaches it is measured against.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "supplied PDF",
+    "written_from_url": "https://doi.org/10.1145/3769828",
+    "source_sha256": "sha256:93e8784edda698f9d28ebd809eddaa9a24f89300a7cd8ca7e407c030759ab02b",
+    "supporting_excerpts": [
+      "Finally, we identify potential logic bugs by detecting violations of the expected set relations among the results of the original and transformed queries. 4 Implementation We implemented SRS on top of SQLancer1, a DBMS testing framework designed for the random generation of database states and SQL queries, which also supports multiple test oracles [ 23,24].",
+      "SRS applies the automated, syntax-rule-based random generation approach of SQLancer to ensure that the database state exhibits sufficient diversity, thereby enabling thorough and effective bug detection.",
+      "Baseline.We selected six state-of-the-art approaches for comparison: PQS [ 25], TLP [ 24], NoREC [ 23], Pinolo [ 11], EET [ 13], and DQP [ 3]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T15:21:47Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3769828",
+          "source_type": "paper_citation_context",
+          "excerpt": "We implemented SRS on top of SQLancer 1 , a DBMS testing framework designed for the random generation of database states and SQL queries, which also supports multiple test oracles [23, 24].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3769828",
+          "source_type": "paper_citation_context",
+          "excerpt": "We selected six state-of-the-art approaches for comparison: PQS [25], TLP [24], NoREC [23], Pinolo [11], EET [13], and DQP [3].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3769828",
+          "source_type": "paper_citation_context",
+          "excerpt": "TQS [28] and DQP [3] are two state-of-the-art testing approaches that target logic bugs in the optimization of join operations.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3769828",
+          "source_type": "paper_citation_context",
+          "excerpt": "And this is a widely used method for comparing different test oracles [23].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3769828",
+          "source_type": "paper_citation_context",
+          "excerpt": "Non-optimizing Reference Engine (NoREC) [23] assumes that a predicate should evaluate to the same value in both the WHERE and SELECT clauses, and leverages this assumption to construct a non-optimizable form of the query.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3769828",
+          "source_type": "paper_citation_context",
+          "excerpt": "This aligns with observations from previous works [3, 24], which limited our reporting of additional potential bugs to avoid duplication.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "manual_curation",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3769828",
+          "source_type": "paper_citation_context",
+          "excerpt": "We implemented SRS on top of SQLancer 1 , a DBMS testing framework designed for the random generation of database states and SQL queries, which also supports multiple test oracles [23, 24].",
+          "excerpt_is_verbatim": true,
+          "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+          "retrieved_at": "2026-09-06T07:08:48Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "manual_curation",
+      "techniques": [
+        "pqs",
+        "tlp",
+        "norec",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3769828",
+          "source_type": "paper_citation_context",
+          "excerpt": "We selected six state-of-the-art approaches for comparison: PQS [25], TLP [24], NoREC [23], Pinolo [11], EET [13], and DQP [3].",
+          "excerpt_is_verbatim": true,
+          "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+          "retrieved_at": "2026-09-06T07:08:48Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "norec",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3769828",
+          "source_type": "paper_citation_context",
+          "excerpt": "We selected six state-of-the-art approaches for comparison: PQS [25], TLP [24], NoREC [23], Pinolo [11], EET [13], and DQP [3].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Non-optimizing Reference Engine Construction (NoREC) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3769828",
+          "source_type": "paper_citation_context",
+          "excerpt": "TQS [28] and DQP [3] are two state-of-the-art testing approaches that target logic bugs in the optimization of join operations.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Differential Query Plans (DQP) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3769832.json b/_data/impact/paper_notes/paper_doi_10_1145_3769832.json
new file mode 100644
index 0000000..e5befb9
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3769832.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3769832",
+    "title": "Test Data Generation for Complex SQL Queries",
+    "authors": [
+      "Sunanda Somwase",
+      "Parismita Das",
+      "S. Sudarshan"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Manag. Data",
+    "doi": "10.1145/3769832",
+    "url": "https://doi.org/10.1145/3769832"
+  },
+  "summary": {
+    "text": "This paper generates sample data for testing SQL queries, a task with applications in testing analytics and application queries, grading student queries, and more recently validating the output of text-to-SQL systems. Earlier work handled single-block queries and single-level nested queries but not more complex ones. The authors present an architecture and techniques aimed at complex queries, reporting that they substantially outperform prior test data generation work there, and also beat the state of the art on the narrower problem of demonstrating that two queries are not equivalent.",
+    "relationship_to_sqlancer": "Generates data that distinguishes non-equivalent queries, the complement of the equivalence-based oracles SQLancer uses.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3769832",
+    "source_sha256": "sha256:ea80c73eccf083fc9b2aac3ad6bc55c83ee65b326286e99ba0058585240cd0bd",
+    "supporting_excerpts": [
+      "Earlier work on test data generation handled basic single-block SQL queries, as well as single-level nested SQL queries, but could not handle more complex queries.",
+      "In this paper, we present a novel architecture and associated techniques for test generation that are designed to handle complex queries.",
+      "We also show that our approach outperforms the state-of-the-art for the more restricted problem of showing non-equivalence of query pairs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:32:20Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3769832",
+          "source_type": "paper_citation_context",
+          "excerpt": "Another potential area for future work is to explore the use of data generation to improve the detection of bugs in database system implementations, following the approach of Rigger et al. [19, 20].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3779212_3790244.json b/_data/impact/paper_notes/paper_doi_10_1145_3779212_3790244.json
new file mode 100644
index 0000000..9aaa334
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3779212_3790244.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3779212.3790244",
+    "title": "Understanding Query Optimization Bugs in Graph Database Systems",
+    "authors": [
+      "Yuyu Chen",
+      "Zhongxing Yu"
+    ],
+    "year": 2026,
+    "venue": "International Conference on Architectural Support for Programming Languages and Operating Systems",
+    "doi": "10.1145/3779212.3790244",
+    "url": "https://doi.org/10.1145/3779212.3790244"
+  },
+  "summary": {
+    "text": "A characteristic study of query optimization bugs in graph database systems, covering root causes, how the bugs manifest, and how they are fixed, from which the authors draw ten findings. Building on that study they developed a testing tool aimed specifically at graph query optimization, which found 20 unique bugs, 10 of them optimization bugs.",
+    "relationship_to_sqlancer": "Studies the class of bug SQLancer's optimizer-oriented oracles target, in the graph database setting.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3779212.3790244",
+    "source_sha256": "sha256:7327e37308379f659176ab266fd4db08ef2a0fa2234598489c212b36e84644a1",
+    "supporting_excerpts": [
+      "This paper conducts the first characteristic study of query optimization bugs in GDBMSs, including the root causes, manifestation methods, and fix strategies, and delivers 10 novel and important findings about them.",
+      "Based on the characteristic study, we also developed a testing tool tailored to uncover GDBMS query optimization bugs, and the tool found 20 unique GDBMS bugs, 10 of which are query optimization bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3779212.3790244",
+          "source_type": "paper_citation_context",
+          "excerpt": "For optimization bugs due to inaccurate cost estimations or defective plan space exploration algorithms, previous works on DBMS testing [11, 21, 45, 46] also frequently report them.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3785021_3787993.json b/_data/impact/paper_notes/paper_doi_10_1145_3785021_3787993.json
new file mode 100644
index 0000000..2c61b45
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3785021_3787993.json
@@ -0,0 +1,38 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3785021.3787993",
+    "title": "Reproducibility Report for ACM SIGMOD 2025 Paper: 'Constant Optimization Driven Database System Testing'",
+    "authors": [
+      "Yuvaraj Chesetti",
+      "Chi Zhang"
+    ],
+    "year": 2025,
+    "venue": null,
+    "doi": "10.1145/3785021.3787993",
+    "url": "https://doi.org/10.1145/3785021.3787993"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "coddtest"
+      ],
+      "sources": [
+        {
+          "source_url": "https://openalex.org/W7131085132",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "OpenAlex records this paper as citing the publication that introduced Constant-Optimization-Driven Testing (CODDTest).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3786673.json b/_data/impact/paper_notes/paper_doi_10_1145_3786673.json
new file mode 100644
index 0000000..4c39934
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3786673.json
@@ -0,0 +1,101 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3786673",
+    "title": "One DBMS, Two Modes, and a Bunch of Bugs: Catching Logic Bugs in Distributed DBMSs via Differential Testing",
+    "authors": [
+      "Zikun Fu",
+      "Jiaju Bai",
+      "Hong-Bo Feng",
+      "Kang Chen"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": "10.1145/3786673",
+    "url": "https://doi.org/10.1145/3786673"
+  },
+  "summary": {
+    "text": "DistSQL finds logic bugs specific to distributed query execution by running identical SQL against the same DBMS configured in centralized and in distributed mode and comparing results. The insight is that centralized execution is simpler and better tested, so it can serve as a reference for the distributed path. It adds distributed diversity-oriented state mutation to expose distribution-specific behaviour and needs no code instrumentation. Across TiDB, CockroachDB, YugabyteDB, ClickHouse, OceanBase and one commercial DBMS it found 65 previously unknown logic bugs, 38 of them distributed-specific.",
+    "relationship_to_sqlancer": "A differential logic-bug oracle for DBMSs, building on the observation that existing techniques do not target distributed execution.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3786673",
+    "source_sha256": "sha256:93638e3b97bc9be64bbf4e277776f8532102db8f98a422ffc9bedc23a68b8eb0",
+    "supporting_excerpts": [
+      "Existing testing techniques focus mainly on system level failures or general DBMS and do not target distributed execution, leaving many distributed logic bugs undiscovered.",
+      "We present DistSQL, a differential testing framework that compares the results of identical SQL queries executed on the same DBMS configured in centralized and distributed modes.",
+      "DistSQL identified 65 previously unknown logic bugs, including 38 specific to distributed execution."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3786673",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compare DistSQL against SOTA database and distributed system testing tools, including Squirrel [65], QPG [2], and EET [24], as well as distributed system testing tools, including Jepsen [20] and Mallory [36].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3786673",
+          "source_type": "paper_citation_context",
+          "excerpt": "We evaluate five open-source distributed DBMSs (Table 3) using four existing tools: Jepsen [20], Mallory [36], QPG [2], and EET [24], and compare their performance with DistSQL.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3786673",
+          "source_type": "paper_citation_context",
+          "excerpt": "DQP [3] compares executions with different query hints applied, where query plans act as local heuristics rather than guiding test case generation.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3786673",
+          "source_type": "paper_citation_context",
+          "excerpt": "This shows that DistSQL is effective and applicable in real-world production settings. feedback signals, such as formal rule coverage in SemConT [31] and graph similarity of test cases in TQS [53], while others adopt more generic runtime feedback, such as branch coverage [30] or query plan text [2].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3786673",
+          "source_type": "paper_citation_context",
+          "excerpt": "…[49] DBMS Logic None Yes Differential Generic SemConT [31] DBMS Logic Keyword and Rule Yes Verification Generic TQS [53] DBMS Logic Graph Similarity Yes Ground Truth Generic QPG [2] centralized execution as a reference, our approach finds distributed logic bugs often missed by existing techniques.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3786673",
+          "source_type": "paper_citation_context",
+          "excerpt": "We implement an abstract syntax tree (AST) based SQL generator [1, 22, 45, 47] that constructs queries in a top down manner by first selecting the SQL statement type (e.g., SELECT , CREATE TABLE ) and then recursively generating its components (e.g., ORDER BY clauses, expressions).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3786699.json b/_data/impact/paper_notes/paper_doi_10_1145_3786699.json
new file mode 100644
index 0000000..ec407e1
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3786699.json
@@ -0,0 +1,55 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3786699",
+    "title": "SQLBarber: A System Leveraging Large Language Models to Generate Customized and Realistic SQL Workloads",
+    "authors": [
+      "Jiale Lao",
+      "Immanuel Trummer"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": "10.1145/3786699",
+    "url": "https://doi.org/10.1145/3786699"
+  },
+  "summary": {
+    "text": "SQLBarber generates customised, realistic SQL workloads with large language models. The motivation is that real queries are hard to obtain for privacy reasons and existing generators offer little customisation or ability to meet realistic constraints. SQLBarber provides a declarative interface, removes the need to write SQL templates by hand while accepting natural-language constraints on them, and adds a self-correction pipeline that profiles, refines and prunes templates by query cost. The authors open-source ten benchmarks built from Snowflake and Amazon Redshift statistics.",
+    "relationship_to_sqlancer": "Workload generation for benchmarking rather than correctness oracles; SQLancer is cited among SQL generation work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3786699",
+    "source_sha256": "sha256:caaf8201dbefb3ffd1c91d390923e3d80418211af68f1a4789e409b34fdfbfa0",
+    "supporting_excerpts": [
+      "However, acquiring real-world SQL queries is challenging due to privacy concerns, and existing generation methods offer limited options for customization and for satisfying realistic constraints.",
+      "To address this issue, we present SQLBarber, a system based on Large Language Models (LLMs) to generate customized and realistic SQL workloads.",
+      "It reduces query generation time by one to two orders of magnitude and significantly improves alignment with the target cost distribution, compared with existing methods."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:31:04Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/995979576a9a43c33d87711cc7ec1757a5f1a280",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3798226.json b/_data/impact/paper_notes/paper_doi_10_1145_3798226.json
new file mode 100644
index 0000000..0cc34fe
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3798226.json
@@ -0,0 +1,60 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3798226",
+    "title": "Metamorphic Testing for Infrastructure-as-Code Engines",
+    "authors": [
+      "David Spielmann",
+      "George Zakhour",
+      "Dominik Arnold",
+      "Matteo Biagiola",
+      "Roland Meier",
+      "Guido Salvaneschi"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Programming Languages",
+    "doi": "10.1145/3798226",
+    "url": "https://doi.org/10.1145/3798226"
+  },
+  "summary": {
+    "text": "EMIaC applies metamorphic testing to Infrastructure-as-Code engines such as Terraform, OpenTofu and Pulumi, which the authors note have received little attention compared with the correctness of IaC programs themselves. It defines metamorphic relations as graph-based transformations of IaC programs and checks invariants across executions of original and transformed versions, using e-graphs both as an input generator and as an equivalence oracle. It covered 98 previously untested statements in Terraform and 1,313 in Pulumi, and three of its test cases were merged into Terraform.",
+    "relationship_to_sqlancer": "Carries metamorphic testing into a different class of system; SQLancer is cited as prior work in that testing tradition.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3798226",
+    "source_sha256": "sha256:c5f04a9b314e33fefe52fba3d7badd4df33b46dc9168d81a17fe72aa655c4d95",
+    "supporting_excerpts": [
+      "We present EMIaC, a metamorphic testing framework for IaC engines.",
+      "EMIaC defines metamorphic relations as graph-based transformations of IaC programs and checks invariants across executions of the original and transformed programs.",
+      "A central novelty is our use of e-graphs in software testing, as both a test-input generator and an equivalence oracle."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3798226",
+          "source_type": "paper_citation_context",
+          "excerpt": "Mutation analysis [17] addresses this problem by systematically injecting syntactic changes into the program under test (i.e., mutants ) to mimic programmers’ mistakes, and by assessing whether the test suite is able to catch them.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3798232.json b/_data/impact/paper_notes/paper_doi_10_1145_3798232.json
new file mode 100644
index 0000000..bd6bc16
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3798232.json
@@ -0,0 +1,61 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3798232",
+    "title": "Beacon: Detecting Broken Access Control Vulnerabilities in DBMSs via System Catalog Consistency Validation",
+    "authors": [
+      "Zongrui Peng",
+      "Jingzhou Fu",
+      "Zhiyong Wu",
+      "Jie Liang",
+      "Xiangdong Huang",
+      "Dalong Shi",
+      "Yu Jiang"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Programming Languages",
+    "doi": "10.1145/3798232",
+    "url": "https://doi.org/10.1145/3798232"
+  },
+  "summary": {
+    "text": "Beacon detects broken access control vulnerabilities in DBMSs by checking SQL operations against the system catalogs. Its premise is that catalog visibility should agree with a user's privileges: if an object is invisible to a user in the catalogs, that user should have no access to it, so any inconsistency indicates a privilege the user should not have. Applied to eight DBMSs it uncovered 39 previously unknown vulnerabilities, 19 giving privilege escalation and 20 unauthorised disclosure, all confirmed by vendors.",
+    "relationship_to_sqlancer": "A consistency oracle in the same spirit as SQLancer's, aimed at access control rather than query results. The abstract does not state that it builds on SQLancer.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3798232",
+    "source_sha256": "sha256:a4fe32fc2567f5833ff65778d52165be47125ca18ef6c2d12d78f8eecc4ac489",
+    "supporting_excerpts": [
+      "In this paper, we propose Beacon, which detects BAC vulnerabilities by validating the consistency between SQL operations and system catalogs.",
+      "Any inconsistency suggests that a user is exceeding their privileges, indicating a potential BAC vulnerability.We used Beacon to test eight popular DBMSs (e.g., MySQL and MariaDB), uncovering 39 previously unknown BAC vulnerabilities."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp",
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/804bc6d1f6c1acde45f1fabcab8b0d3ed5f647c9",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Cardinality Estimation Restriction Testing (CERT).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3798245.json b/_data/impact/paper_notes/paper_doi_10_1145_3798245.json
new file mode 100644
index 0000000..0571432
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3798245.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3798245",
+    "title": "Understanding and Finding JIT Compiler Performance Bugs",
+    "authors": [
+      "Zijian Yi",
+      "Chen Ding",
+      "August Shi",
+      "Milos Gligoric"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Programming Languages",
+    "doi": "10.1145/3798245",
+    "url": "https://doi.org/10.1145/3798245"
+  },
+  "summary": {
+    "text": "The first study of performance bugs in just-in-time compilers, which prior automated work had not targeted, having focused on functional bugs. The authors manually analyse 191 bug reports across four Java and JavaScript JIT compilers to characterise triggers, manifestation patterns and root causes, then propose layered differential performance testing and implement it as Jittery. With optimisations such as test prioritisation, which cuts testing time by 92.40% without losing bugs, Jittery found 12 previously unknown performance bugs in Oracle HotSpot and Graal, 11 confirmed.",
+    "relationship_to_sqlancer": "Differential performance testing for compilers rather than DBMSs; SQLancer is cited among the testing work it builds on.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3798245",
+    "source_sha256": "sha256:134f41a0e15af4117fd744a60f7c03b164cf96325677649bdab72e5131d08606",
+    "supporting_excerpts": [
+      "However, no prior work has targeted JIT compiler performance bugs, which can cause significant performance degradation while an application is running.",
+      "In this paper, we present the first work on demystifying JIT performance bugs.",
+      "Using Jittery, we discovered 12 previously unknown performance bugs in the Oracle HotSpot and Graal JIT compilers, with 11 confirmed and 6 fixed by developers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/c55ef302233eb0c1ccc6c9677eda03b16089b148",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Cardinality Estimation Restriction Testing (CERT).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3799227.json b/_data/impact/paper_notes/paper_doi_10_1145_3799227.json
new file mode 100644
index 0000000..c7476da
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3799227.json
@@ -0,0 +1,152 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3799227",
+    "title": "A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Evaluation",
+    "authors": [
+      "Xiyue Gao",
+      "Zhuang Liu",
+      "Jiangtao Cui",
+      "Hui Li",
+      "Hui Zhang",
+      "Kewei Wei",
+      "Kankan Zhao"
+    ],
+    "year": 2026,
+    "venue": "ACM Computing Surveys",
+    "doi": "10.1145/3799227",
+    "url": "https://doi.org/10.1145/3799227"
+  },
+  "summary": {
+    "text": "A survey of fuzzing for database management systems. It argues that DBMS fuzzing needs treatment separate from general-purpose software fuzzing because of differences in internal structure, input and output, and testing objectives, and it organises existing work into a general fuzzing procedure and a taxonomy by testing objective. It also covers non-relational systems and newer techniques, and accompanies the survey with an open-source toolkit, OpenDBFuzz, used to compare methods experimentally.",
+    "relationship_to_sqlancer": "A survey that catalogues the field SQLancer belongs to. The abstract does not describe reusing or extending SQLancer; the record's link is that the survey cites the SQLancer papers among the work it reviews.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3799227",
+    "source_sha256": "sha256:fe08ae3b4e754322677411216eb3656bf7f8c8b8d93768cd9d1da86ffb6a77ee",
+    "supporting_excerpts": [
+      "Therefore, this article focuses on DBMS fuzzing and provides a comprehensive review and comparison of the methods in this field.",
+      "To objectively evaluate the performance of each method, we present an open-source DBMS fuzzing toolkit, OpenDBFuzz."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3799227",
+          "source_type": "paper_citation_context",
+          "excerpt": "While traditional LEGO[67] SQLRight[69] QPG[13] AMOEBA[73] GARan[16] DynSQL[54] Squill[106] approaches utilize generic SAT solvers for this task solver [40, 60], recent methods such as TQS [102] also use a logical external solver as a substitute for the SAT solver to derive the ground truth results based on the original test cases through logical operations or theoretical deduction [29, 37, 102].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3799227",
+          "source_type": "paper_citation_context",
+          "excerpt": "Without this limitation, numerous complex SQL statements would be generated, which may expand the search space but reduce the overall efficiency of bug detection [28, 32, 91–93, However, static configurations require manual adjustments based on specific requirements.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3799227",
+          "source_type": "paper_citation_context",
+          "excerpt": "On the other hand, fuzzers based on random databases [12, 14, 15, 28, 32, 34, 55, 81, 88, 92, 93, 97, 102, 106, 107, 117] create random database instances from scratch.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3799227",
+          "source_type": "paper_citation_context",
+          "excerpt": "Specifically, we compare them in terms of database instance, generation type, strategy, and feedback.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3799227",
+          "source_type": "paper_citation_context",
+          "excerpt": "Database Crashes Crash Bugs Squill[106]Griffin[34] LEGO[67]DynSQL[54] SOFT[35] RAGS[96] Differential Different DBMSs Logic Bugs SQLsmith[4]Go-Randgen[88] GARan[16]DT2[28] Radar[98] Different Database Instances DDLCheck[99] APOLLO[57] Different Versions of the Same DBMS Performance Bugs AMOEBA[73] Metamorphic Statement Rewriting CERT[14]MutaSQL[26] Logic Bugs Eqsql[114]NoREC[91] SQLRight[69]PINOLO[45] EET[56]CODDTest[113] PUPPY[107] Execution Path Manipulation Performance Bugs Mozi[68] Performance & Logic Bugs Kangaroo[64] Logic & Crash Bugs DQP[15] Logic Bugs DQE[97] Statement Type Transformation TLP[92] Query Partition Troc[32] Transaction Splitting TxCheck[55] ADUSA[12] Constraint-solving Forward Solving(SAT Solver) Artemis[81] PQS[93] Backward Solving TQS[102] Forward Solving(Logical Solver) WriteCheck[29]Fucci[37] (a) Constraint Rewriting • Statement Rewriting : Statement Rewriting refers to the process of modifying a statement according to specified rules, so that the final semantics remain unchanged or change as expected.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3799227",
+          "source_type": "paper_citation_context",
+          "excerpt": "…Performance Bugs Mozi[68] Performance & Logic Bugs Kangaroo[64] Logic & Crash Bugs DQP[15] Logic Bugs DQE[97] Statement Type Transformation TLP[92] Query Partition Troc[32] Transaction Splitting TxCheck[55] ADUSA[12] Constraint-solving Forward Solving(SAT Solver) Artemis[81] PQS[93] Backward…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2311.06728",
+          "source_type": "paper_citation_context",
+          "excerpt": "As can be seen in Figure 20, QPG detected bugs more efficiently than TLP within 240 minutes, and the number of bugs gradually equalized in the later period.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2311.06728",
+          "source_type": "paper_citation_context",
+          "excerpt": "On the other hand, fuzzers based on random databases [16, 31– 33, 62, 71, 81, 85, 86, 90, 94, 101, 109] create random database instances from scratch.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2311.06728",
+          "source_type": "paper_citation_context",
+          "excerpt": "QPG implements a generation-based generator based on PQS, TLP, and NoREC.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2311.06728",
+          "source_type": "paper_citation_context",
+          "excerpt": "Supported DBMS Link SQLancer[86] (PQS[86], NoREC[84], TLP[85], QPG[18]) SQLite, MySQL, TiDB MariaDB, CockroachDB, OceanBase https://github.com/sqlancer/sqlancer DQE[90] SQLite, MySQL, MariaDB TiDB, CockroachDB https://github.com/tcse-iscas/dqetool SQLRight[60] SQLite, PostgreSQL, MySQL https: //github.com/psu-security-universe/sqlright SQLsmith[5] SQLite, MonetDB, PostgreSQL https://github.com/anse1/sqlsmith Go-Randgen[81] MySQL, TiDB https://github.com/pingcap/go-randgen Squirrel[109] SQLite, PostgreSQL, MySQL, MariaDB https://github.com/s3team/Squirrel DT2[31] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc Troc [32] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc APOLLO[51] SQLite, PostgreSQL https://github.com/sslab-gatech/apollo AMOEBA[62] PostgreSQL, CockroachDB https://bit.ly/3I995jL confirmed bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2311.06728",
+          "source_type": "paper_citation_context",
+          "excerpt": "Database Crashes Crash Bugs Squill[101] Griffin[33] LEGO[59] DynSQL[49] RAGS[89] Differential Different Databases Logic Bugs SQLsmith[5] Go-Randgen[81] GARan[19] DT2[31] APOLLO[51] Different Versions of A Same Database Performance Bugs AMOEBA[62] Metamorphic Statement Rewriting MutaSQL[28] Logic Bugs Eqsql[107] NoREC[84] SQLRight[60] DQE[90] Statement Type Transformation TLP[85] Query Partition Troc[32] Transaction Splitting ADUSA[16] Constraint-solving Forward Solving(SAT Solver) Artemis[71] PQS[86] Backward Solving TQS[94] Forward Solving(Logical Solver) Some generators of black-box fuzzers [18] obtain feedback such as the validity of query plans or test cases from the query or query plan interface provided by the DBMS, guiding the subsequent generation process.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2311.06728",
+          "source_type": "paper_citation_context",
+          "excerpt": "…Bugs AMOEBA[62] Metamorphic Statement Rewriting MutaSQL[28] Logic Bugs Eqsql[107] NoREC[84] SQLRight[60] DQE[90] Statement Type Transformation TLP[85] Query Partition Troc[32] Transaction Splitting ADUSA[16] Constraint-solving Forward Solving(SAT Solver) Artemis[71] PQS[86] Backward Solving…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3802034.json b/_data/impact/paper_notes/paper_doi_10_1145_3802034.json
new file mode 100644
index 0000000..383019c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3802034.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3802034",
+    "title": "DBugScribe: Automatic Database Bug Reproduction from Community Reports",
+    "authors": [
+      "Suyang Zhong",
+      "Mo Sha",
+      "Sheng Wang",
+      "Fangyuan Zhou",
+      "Feifei Li",
+      "Kian-Lee Tan"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": "10.1145/3802034",
+    "url": "https://doi.org/10.1145/3802034"
+  },
+  "summary": {
+    "text": "DBugScribe automatically turns natural-language database bug reports into executable reproduction scenarios. It introduces a domain-specific language with formal semantics for representing a bug scenario as a composable specification covering configuration, schema, data, queries and oracle, and synthesises those specifications with a hybrid of LLM-based extraction, rule-based validation and self-refinement. On 218 confirmed reports drawn from eight recent DBMS testing tools covering MySQL, TiDB and MariaDB it reproduces 72.9%.",
+    "relationship_to_sqlancer": "Its evaluation set is drawn from reports produced by recent DBMS testing tools, the category SQLancer belongs to.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3802034",
+    "source_sha256": "sha256:1278ac4e3c974f7f3d536e7c48fe206512647429bfc4e1e397cfc7a1cc80f252",
+    "supporting_excerpts": [
+      "We present\n DBugScribe,\n the first framework specifically designed to automatically synthesize validated, executable, and structured reproduction scenarios for database bugs directly from users' natural language bug reports.",
+      "Evaluated on 218 confirmed bug reports from eight recent DBMS testing tools covering MySQL, TiDB, and MariaDB,\n DBugScribe\n achieves 72.9% reproduction success."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/8f02d1ed0dfacfff322cccdddf971240b3c8136d",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP), Differential Query Plans (DQP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3802053.json b/_data/impact/paper_notes/paper_doi_10_1145_3802053.json
new file mode 100644
index 0000000..70bd6d3
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3802053.json
@@ -0,0 +1,62 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3802053",
+    "title": "EPSC: Testing Database Management Systems via Equivalent Prepared Statement Construction",
+    "authors": [
+      "Chi Zhang",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Dalong Shi",
+      "Linzhang Wang",
+      "Yu Jiang"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": "10.1145/3802053",
+    "url": "https://doi.org/10.1145/3802053"
+  },
+  "summary": {
+    "text": "EPSC tests DBMSs by exploiting the fact that a statement can be written both as an ordinary SQL statement and as a prepared statement with literals bound as parameters. The two forms should behave identically, so any difference between them indicates a bug. The authors note that prepared statements, though widely used in production for performance and against SQL injection, have received far less testing attention than ordinary statements. Applied to seven mature DBMSs, EPSC found 49 unique bugs, 31 confirmed and 10 fixed.",
+    "relationship_to_sqlancer": "A metamorphic logic-bug oracle for DBMSs, in the same family as SQLancer's oracles and citing that line of work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3802053",
+    "source_sha256": "sha256:164e960cbec7c9987b32581ca4e061f44450ef49d62509f52097b27a26d65d1d",
+    "supporting_excerpts": [
+      "The key insight of EPSC is that Data Manipulation Language (DML) and Query Definition Language (QDL) statements can be executed in two equivalent forms—ordinary statements and prepared statements— which should exhibit consistent behavior and produce identical results.",
+      "Any inconsistency between them indicates the presence of a bug in the target DBMS.",
+      "In total, EPSC uncovered 49 unique bugs, of which 31 have been confirmed and 10 have already been fixed."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp",
+        "dqp",
+        "coddtest"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/f42307b2aae0cdc8c3fae609e844622539b26baa",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Differential Query Plans (DQP), Constant-Optimization-Driven Testing (CODDTest).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3802061.json b/_data/impact/paper_notes/paper_doi_10_1145_3802061.json
new file mode 100644
index 0000000..25893c2
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3802061.json
@@ -0,0 +1,65 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3802061",
+    "title": "Finding Missed Optimizations in DBMSs through Unbalanced Short-circuit Query Construction",
+    "authors": [
+      "Jinhui Lai",
+      "Chi Zhang",
+      "Jie Liang",
+      "Zihao Zeng",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Chijin Zhou",
+      "Shuai Ma",
+      "Yu Jiang",
+      "Zichen Xu"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": "10.1145/3802061",
+    "url": "https://doi.org/10.1145/3802061"
+  },
+  "summary": {
+    "text": "SCor looks for optimisations a DBMS could have applied but did not. It constructs short-circuit queries whose result is obtainable from low-cost operations alone; if the system nonetheless executes the full query with expensive operations, that is a missed optimisation. The authors argue existing work finds cases where an implemented optimisation performs poorly but rarely reveals ones entirely absent. Across 11 widely used DBMSs SCor found 153 previously undetected performance bugs, 125 confirmed and 33 fixed.",
+    "relationship_to_sqlancer": "A performance oracle for DBMSs, related to SQLancer's line of work on automatically deciding whether a system behaved correctly.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3802061",
+    "source_sha256": "sha256:b04640cf62e85fceb24c153802aa076fc6058e56d381425f113f97d800da028f",
+    "supporting_excerpts": [
+      "In this paper, we present SCor, a black-box approach for identifying missed optimizations in DBMSs through unbalanced short-circuit query construction.",
+      "SCor realizes it by constructing short-circuit queries, where the result can be obtained from low-cost operations alone.",
+      "Our evaluation of SCor across 11 widely-used DBMSs, revealed 153 previously undetected performance bugs resulting from missed optimizations, including 2 bugs in Oracle and 3 bugs in PostgreSQL."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:26:39Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp",
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/96bb90109a01a144448d9bf9b7b58f3e4c440eff",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Cardinality Estimation Restriction Testing (CERT).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3803437_3806090.json b/_data/impact/paper_notes/paper_doi_10_1145_3803437_3806090.json
new file mode 100644
index 0000000..265b83c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3803437_3806090.json
@@ -0,0 +1,60 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3803437.3806090",
+    "title": "Why Property-Based Testing is Necessary for Data Intensive Scalable Computing",
+    "authors": [
+      "Yaoxuan Wu",
+      "I. Lee",
+      "Ahmad Humayun",
+      "Muhammad Ali Gulzar",
+      "Miryung Kim"
+    ],
+    "year": 2026,
+    "venue": "SIGSOFT FSE Companion",
+    "doi": "10.1145/3803437.3806090",
+    "url": "https://doi.org/10.1145/3803437.3806090"
+  },
+  "summary": {
+    "text": "A short position paper arguing that data-intensive scalable computing frameworks such as Spark, Flink, Beam and Dask need property-based testing. Its premise is that framework bugs are common and frequently show up as silent wrong results rather than crashes, and that property-based testing has succeeded elsewhere by using semantic contracts as oracles where ground truth is hard to obtain. The authors call for a general, extensible framework with reusable property templates that can be instantiated across different DISC systems.",
+    "relationship_to_sqlancer": "Makes the case for oracle-based testing in a neighbouring domain, citing the DBMS work where that approach succeeded.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3803437.3806090",
+    "source_sha256": "sha256:730e24a1d1ff88847f815bd398c4227ca595c369bbd5cf7465ae893ecf30759f",
+    "supporting_excerpts": [
+      "Despite their widespread adoption, framework bugs remain common, and many manifest as silent wrong results rather than crashes.",
+      "Property-based testing (PBT) has been successful across a range of domains by using semantic contracts as executable oracles when ground-truth outputs are difficult to obtain.",
+      "We argue that DISC calls for a general and extensible PBT framework that provides reusable property templates and supports systematic instantiation across different DISC systems."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/ba02b6d2cf64ddf05e51681950d9094399769829",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3808109.json b/_data/impact/paper_notes/paper_doi_10_1145_3808109.json
new file mode 100644
index 0000000..51f9a8c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3808109.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3808109",
+    "title": "Eidolon: Perform Noise-Aware Fuzzing on FHE Libraries via Equivalence Expression Transformation",
+    "authors": [
+      "Zhensheng Xian",
+      "Zhen Yan",
+      "Yuanliang Chen",
+      "Xuelian Cao",
+      "Fuchen Ma",
+      "Dalong Shi",
+      "Yu Jiang"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Software Engineering",
+    "doi": "10.1145/3808109",
+    "url": "https://doi.org/10.1145/3808109"
+  },
+  "summary": {
+    "text": "Eidolon is a fuzzer for fully homomorphic encryption libraries. The authors argue that existing cryptographic testing tools generate structured inputs suited to cryptographic algorithms and are unaware of FHE noise management, so they fail to exercise computation on encrypted data. Eidolon directs mutations at arithmetic expressions within the noise budget and uses Equivalence Expression Transformation as its oracle, rewriting an expression into mathematically identical but structurally different forms and comparing outputs. It found 20 previously unknown bugs in SEAL, OpenFHE, HElib and TFHE, 12 assigned CVEs.",
+    "relationship_to_sqlancer": "Carries the equivalence-transformation oracle idea into FHE libraries rather than DBMSs; SQLancer is cited as prior work using that style of oracle.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3808109",
+    "source_sha256": "sha256:ee8e6340dbcf1d8da22e28313b3b2e9c8ea25ada942eb5311cbe574d1cf4b3f8",
+    "supporting_excerpts": [
+      "To address this gap, we propose Eidolon, a noise-aware fuzzer.",
+      "As its test oracle, Eidolon leverages Equivalence Expression Transformation, which transforms a standard arithmetic expression into two mathematically identical but structurally different forms (e.g., Factored, Horner) to detect inconsistencies in their outputs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3808109",
+          "source_type": "paper_citation_context",
+          "excerpt": "For instance, metamorphic fuzz testing [28] combines MT with fuzzing to detect faults in autonomous driving systems, QFuzz [36] uses execution costs for side-channel leakage quantification, and SQLancer [41] employs a containment oracle for DBMS logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3810991_3811632.json b/_data/impact/paper_notes/paper_doi_10_1145_3810991_3811632.json
new file mode 100644
index 0000000..de98bf1
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3810991_3811632.json
@@ -0,0 +1,127 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3810991.3811632",
+    "title": "DIRT: Database-Integrated Random Testing",
+    "authors": [
+      "A. Keles",
+      "Ethan Chou",
+      "Harrison Goldstein",
+      "Leonidas Lampropoulos"
+    ],
+    "year": 2026,
+    "venue": "DBTest@SIGMOD",
+    "doi": "10.1145/3810991.3811632",
+    "url": "https://doi.org/10.1145/3810991.3811632"
+  },
+  "summary": {
+    "text": "DIRT argues that testing tools built for mature databases fit early-stage systems badly, and integrates the testing framework into the DBMS itself so that random testing evolves alongside the system and false positives are reduced by construction. It introduces generation actions, an abstraction letting database developers rather than testing experts state correctness properties. Evaluated on Turso, an actively developed SQLite-compatible OLTP engine, it found 23 unique confirmed bugs.",
+    "relationship_to_sqlancer": "Names SQLancer directly as a tool it is measured against, reporting a better true positive rate and more useful bug reports than off-the-shelf SQLancer variants on a system under active development.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3810991.3811632",
+    "source_sha256": "sha256:039b3a9c669a4f9e29a380a47b5fa84d2d1f4357c2f426e782520418d68f0f60",
+    "supporting_excerpts": [
+      "Traditional testing tools like SQLancer and SQLSmith are highly effective for mature databases, but they struggle with high false positive rates and low actionability when applied to evolving systems.",
+      "We evaluate DIRT on Turso, an actively developed SQLite-compatible OLTP engine, and show that it finds 23 unique, confirmed bugs–significantly outperforming off-the-shelf SQLancer variants in terms of true positive rate and usefulness of bug reports."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp",
+        "coddtest"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811632",
+          "source_type": "paper_citation_context",
+          "excerpt": "2b presents the GA for WHERE Extended case of Ternary Logic Partitioning (TLP) [21] oracle from SQLancer.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811632",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer currently supports Query Plan Guidance (QPG) [1] for feedback-guided generation, Cardinality Estimation Restriction Testing (CERT) [2] for finding performance bugs in DBMSs, Differential Query Plans (DQP) [3] for detecting bugs in join optimizations, and Constant Optimization Driven…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811632",
+          "source_type": "paper_citation_context",
+          "excerpt": "…generation, Cardinality Estimation Restriction Testing (CERT) [2] for finding performance bugs in DBMSs, Differential Query Plans (DQP) [3] for detecting bugs in join optimizations, and Constant Optimization Driven Database System Testing (CODDTest) [26] for finding logic bugs in…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811632",
+          "source_type": "paper_citation_context",
+          "excerpt": "We opted for the shadow state because it allows for complex reasoning over the database state for constructing arbitrary queries and properties, and because it gives us a canonical property over the database state: the shadow state is identical to the database at any given moment.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811632",
+          "source_type": "paper_citation_context",
+          "excerpt": "Two metamorphic oracles followed: Non-Optimizing Reference Engine Construction (NoREC) [20], which found 51 optimization bugs, and Ternary Logic Partitioning (TLP) [21], which discovered 77 novel logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811632",
+          "source_type": "paper_citation_context",
+          "excerpt": "It started with Pivoted Query Synthesis (PQS) [23], a rather \"simple\" containment property over databases that has found at least 121 unique logic bugs in production databases.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp",
+        "coddtest"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2604.16373",
+          "source_type": "paper",
+          "excerpt": "We compare the performance of DIRT for Turso against SQLancer with minimal modifications, which begets some questions worth discussing.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3810991_3811634.json b/_data/impact/paper_notes/paper_doi_10_1145_3810991_3811634.json
new file mode 100644
index 0000000..671ddaa
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3810991_3811634.json
@@ -0,0 +1,54 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3810991.3811634",
+    "title": "Cloud Analytics Benchmarking: Where We Are, What Is Missing, and Where We Should Go",
+    "authors": [
+      "Michail Georgoulakis Misegiannis",
+      "Till Steinert"
+    ],
+    "year": 2026,
+    "venue": "DBTest@SIGMOD",
+    "doi": "10.1145/3810991.3811634",
+    "url": "https://doi.org/10.1145/3810991.3811634"
+  },
+  "summary": {
+    "text": "A position paper on benchmarking cloud analytics. It observes that vendor workload traces carry execution telemetry such as CPU time, scanned bytes and operator counts, but omit SQL text for privacy, and asks whether realistic workloads can be synthesised from that telemetry alone. The authors argue this works at the workload level but not the query level, since telemetry varies across systems and deployments and does not uniquely determine query structure, and propose enriching traces with transferable features and building use-case-centric benchmarks instead.",
+    "relationship_to_sqlancer": "Concerns benchmarking rather than correctness testing; SQLancer appears among the cited work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3810991.3811634",
+    "source_sha256": "sha256:71ec15243e585717022fff706296160569baddd32c9f9a48657bd1625b9be8ef",
+    "supporting_excerpts": [
+      "This raises a fundamental question: Can realistic workloads be synthesized from anonymized telemetry?",
+      "To bridge this gap, we outline two complementary paths: enriching traces with engine-agnostic transferable workload features, and building use-case-centric benchmarks directly from open application scenarios such as ELT pipelines and BI dashboards."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811634",
+          "source_type": "paper_citation_context",
+          "excerpt": "Related work also considers database testing [21, 37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3810991_3811637.json b/_data/impact/paper_notes/paper_doi_10_1145_3810991_3811637.json
new file mode 100644
index 0000000..99f809a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3810991_3811637.json
@@ -0,0 +1,89 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3810991.3811637",
+    "title": "Boosting DBMS Test Coverage via LLM-Driven SQL Generation",
+    "authors": [
+      "Eslam Abdelkarim",
+      "Carsten Binnig",
+      "Anupam Sanghi"
+    ],
+    "year": 2026,
+    "venue": "DBTest@SIGMOD",
+    "doi": "10.1145/3810991.3811637",
+    "url": "https://doi.org/10.1145/3810991.3811637"
+  },
+  "summary": {
+    "text": "Quover uses a large language model to generate SQL aimed at raising code coverage in DBMS testing. It works iteratively: it identifies functions in the DBMS source that are not yet covered, prompts the model with context including function descriptions and code snippets, generates targeted queries, and validates their effect. The authors report over 57% coverage within a few hours, a 14% improvement over the automated systems they compare with, across several schemas and DBMSs.",
+    "relationship_to_sqlancer": "Coverage-directed query generation for DBMSs, positioned against existing automated testing systems in the field SQLancer defined.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3810991.3811637",
+    "source_sha256": "sha256:5b5a3b5b4b5022013385c049e2768c3df00a28008b12c776528e3c51d565f671",
+    "supporting_excerpts": [
+      "This paper presents Quover, an automated approach for improving DBMS test coverage through large language model (LLM)-generated SQL queries.",
+      "Specifically, Quover implements an iterative coverage-guided methodology that identifies uncovered functions in the target DBMS source code and generates targeted SQL queries."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811637",
+          "source_type": "paper_citation_context",
+          "excerpt": "…and pair them with metamorphic or partitioning-based oracles (e.g., Pivoted Query Synthesis (PQS) [6], NoREC [4], and Ternary Logic Partitioning (TLP) [5]) to uncover logic bugs and exercise diverse execution paths in DBMSs, but they do not explicitly optimize for exercising uncovered code regions.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811637",
+          "source_type": "paper_citation_context",
+          "excerpt": "Logic-testing tools such as SQLancer [3, 6] generate queries and pair them with metamorphic or partitioning-based oracles (e.g., Pivoted Query Synthesis (PQS) [6], NoREC [4], and Ternary Logic Partitioning (TLP) [5]) to uncover logic bugs and exercise diverse execution paths in DBMSs, but they do…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811637",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQL fuzzing tools such as SQLancer [3, 6] automatically generate syntactically valid queries and use logic-testing oracles (e.g., Pivoted Query Synthesis) to uncover logic bugs in DBMSs. Coverage-guided fuzzers such as RATEL [11] and SQLRight [1] incorporate code coverage feedback to iteratively…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811637",
+          "source_type": "paper_citation_context",
+          "excerpt": "…tools such as SQLancer [3, 6] generate queries and pair them with metamorphic or partitioning-based oracles (e.g., Pivoted Query Synthesis (PQS) [6], NoREC [4], and Ternary Logic Partitioning (TLP) [5]) to uncover logic bugs and exercise diverse execution paths in DBMSs, but they do not…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1145/3810991.3811637",
+          "source_type": "paper_citation_context",
+          "excerpt": "…as SQLancer [3, 6] generate queries and pair them with metamorphic or partitioning-based oracles (e.g., Pivoted Query Synthesis (PQS) [6], NoREC [4], and Ternary Logic Partitioning (TLP) [5]) to uncover logic bugs and exercise diverse execution paths in DBMSs, but they do not explicitly…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1145_3828685.json b/_data/impact/paper_notes/paper_doi_10_1145_3828685.json
new file mode 100644
index 0000000..7828145
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1145_3828685.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1145/3828685",
+    "title": "Programmable Property-Based Testing",
+    "authors": [
+      "A. Keles",
+      "Justine Frank",
+      "Ceren Mert",
+      "Harrison Goldstein",
+      "Leonidas Lampropoulos"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Programming Languages",
+    "doi": "10.1145/3828685",
+    "url": "https://doi.org/10.1145/3828685"
+  },
+  "summary": {
+    "text": "This paper proposes a deeper language for expressing properties in property-based testing. It observes that properties in modern frameworks are written in shallowly embedded DSLs and tightly coupled to how they are run, so users are limited to whatever configuration the framework authors anticipated. The alternative, a mixed embedding the authors call deferred binding abstract syntax, reifies a property as a data structure and separates it from the runner that executes it. They implement it in Rocq and Racket and prototype a variety of property runners on top.",
+    "relationship_to_sqlancer": "Concerns the expression of properties in general-purpose testing frameworks; SQLancer is cited as an application of property-style oracles.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1145/3828685",
+    "source_sha256": "sha256:4cd94f490b22a53f0bd069e40f96e512a4253b606693af55d0ee71aa8e26794e",
+    "supporting_excerpts": [
+      "We propose a new, deeper language for properties based on a mixed embedding that we call deferred binding abstract syntax, which reifies properties as a data structure and decouples them from the property runners that execute them.",
+      "We implement this language in Rocq and Racket, leveraging the power of dependent and dynamic types, respectively."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/1b21dbf4dfd10358f77bf5ba3b9ba39c81e18b25",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_1177_0926227x251370258.json b/_data/impact/paper_notes/paper_doi_10_1177_0926227x251370258.json
new file mode 100644
index 0000000..0961992
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_1177_0926227x251370258.json
@@ -0,0 +1,125 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.1177/0926227x251370258",
+    "title": "SQLaser: Detecting database management system (DBMS) logic bugs with clause-guided fuzzing",
+    "authors": [
+      "Jin Wei",
+      "Ping Chen",
+      "Kangjie Lu",
+      "Jun Dai",
+      "Xiaoyan Sun"
+    ],
+    "year": 2025,
+    "venue": "Journal of computing and security",
+    "doi": "10.1177/0926227x251370258",
+    "url": "https://doi.org/10.1177/0926227x251370258"
+  },
+  "summary": {
+    "text": "SQLaser is a directed fuzzer guided by SQL clauses. The authors argue rule-based detection is limited by how hard rules are to specify, while coverage-guided fuzzing explores paths that are unlikely to hold logic bugs. From examining existing non-crashing logic bugs in four DBMSs they derive 35 logic-bug patterns, expressed as clause combinations backed by sequences of functions, and model these as error-prone function chains targeted by a directed fuzzer with a new path-to-path distance mechanism. On SQLite, MySQL, PostgreSQL and TiDB it cut detection time by about 60% and found 22 bugs, four zero-day.",
+    "relationship_to_sqlancer": "Directed fuzzing for DBMS logic bugs, contrasted with the rule-based oracles SQLancer represents.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.1177/0926227x251370258",
+    "source_sha256": "sha256:6dd2cbed989856185e836e7462014d32d17d70e61c4b3f7fbabf268ded82c613",
+    "supporting_excerpts": [
+      "Existing detection employs two strategies: rule-based bug detection and coverage-guided fuzzing.",
+      "In this paper, we design SQLaser, a SQL-clause-guided fuzzer for detecting logic bugs in DBMSs.",
+      "As a standalone fuzzer, SQLaser identified 22 bugs spanning 18 of the 35 logic-bug patterns, outperforming contemporary fuzzers such as SQLRight, which only uncovered two logic bugs across two patterns within the same testing period (i.e., 60 days) when testing SQLite."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:31:04Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.1177/0926227x251370258",
+          "source_type": "paper_citation_context",
+          "excerpt": "This analysis covers nearly all logic bugs in research studies, 3–6 ensuring that the selection is comprehensive and not arbitrary.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1177/0926227x251370258",
+          "source_type": "paper_citation_context",
+          "excerpt": "3 However, since March 2023, new research studies 88–91 on DBMS logic bugs have uncovered additional bug patterns.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.1177/0926227x251370258",
+          "source_type": "paper_citation_context",
+          "excerpt": "Additionally, new bug patterns have been discovered in studies 88–91 published after March 2023.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2407.04294",
+          "source_type": "paper_citation_context",
+          "excerpt": "It is worth noting that the Pivoted Query Synthesis (PQS) oracles [40] proposed by Manuel Rigger et al. does not use the differential testing method; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2407.04294",
+          "source_type": "paper_citation_context",
+          "excerpt": "These bugs, which are due to implementation errors in the DBMSs’ code and are particularly common, can cause a variety of serious issues, such as incorrect query results, exposure of sensitive data, unauthorized access, and data corruption [28, 38–40].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2407.04294",
+          "source_type": "paper_citation_context",
+          "excerpt": "Pivoted Query Synthesis (PQS) oracle [40] does not employ the concept of differential testing; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2407.04294",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] and the Ternary Logic Partitioning (TLP) oracle [39] are two oracles employing the concept of differential testing [32].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2407.04294",
+          "source_type": "paper_citation_context",
+          "excerpt": "Currently, SQLaser only supports differential testing-based oracles and does not accommodate other types of oracles, such as PQS oracle [40], which detects logic bugs through non-differential testing.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2407.04294",
+          "source_type": "paper_citation_context",
+          "excerpt": "Ternary Logic Partitioning (TLP) oracle [39] composes several sub-queries to collectively achieve the semantics of the original query.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14711_thesis_991012980220103412.json b/_data/impact/paper_notes/paper_doi_10_14711_thesis_991012980220103412.json
new file mode 100644
index 0000000..5bd2aef
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14711_thesis_991012980220103412.json
@@ -0,0 +1,120 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14711/thesis-991012980220103412",
+    "title": "Duplicate-sensitivity Guided Transformation Synthesis for DBMS Correctness Bug Detection",
+    "authors": [
+      "Yushan Zhang",
+      "Peisen Yao",
+      "Rongxin Wu",
+      "Charles Zhang"
+    ],
+    "year": 2021,
+    "venue": "arXiv.org",
+    "doi": "10.14711/thesis-991012980220103412",
+    "url": "https://doi.org/10.14711/thesis-991012980220103412"
+  },
+  "summary": {
+    "text": "This thesis automates the discovery of query transformations for DBMS correctness testing. Existing transformation-based approaches generate an equivalent query pair and check that the system returns the same results, but rely on transformations written by hand, which limits the input space they explore. The proposed method synthesises many candidate transformations by mutating a query while preserving its duplicate sensitivity — a necessary condition for equivalence — then filters out the non-equivalent ones with a query equivalence checker. The tool, Eqsql, found 30 confirmed unique bugs in MySQL, TiDB and CynosDB in two months.",
+    "relationship_to_sqlancer": "Automates the transformation design that SQLancer's oracles specify by hand, and evaluates against that work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14711/thesis-991012980220103412",
+    "source_sha256": "sha256:e4a3173b9a9c21aaf555cc8b231192f2d8a1cbbd91b155fd4526f9abb8f48a3a",
+    "supporting_excerpts": [
+      "Recently several works proposed to automatically generate many test cases with query transformation, a process of generating an equivalent query pair and testing a DBMS by checking whether the system returns the same result set for both queries.",
+      "However, all of them still heavily rely on manual work to provide a transformation which largely confines their exploration of the valid input query space.",
+      "This paper introduces duplicate-sensitivity guided transformation synthesis which automatically finds new transformations by first synthesizing many candidates then filtering the nonequivalent ones."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:12Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+          "source_type": "paper_citation_context",
+          "excerpt": "Similarly, NoREC [8] constructs a non-optimizing version of a query and compares the result.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+          "source_type": "paper_citation_context",
+          "excerpt": "We omitted comparison with SQLancer for three reasons.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+          "source_type": "paper_citation_context",
+          "excerpt": "Though PQS can generate the test oracle more practically than the previous work, it can only reveal bugs with a symptom of missing the picked row.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+          "source_type": "paper_citation_context",
+          "excerpt": "Previous work NoREC [8] can only detect such bugs because it only manipulates the “where” predicate to create an unoptimized query mutant.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+          "source_type": "paper_citation_context",
+          "excerpt": "of the existing studies [1], [8], [10] presumes that the generated query pair should explicitly include a “where” predicate (e.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+          "source_type": "paper_citation_context",
+          "excerpt": "DBMSs return wrong result sets [1], though these systems have been extensively tested during their development [2], [3].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2107.03660",
+          "source_type": "paper",
+          "excerpt": "We omitted comparison with SQLancer for three reasons.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14722_ndss_2025_230530.json b/_data/impact/paper_notes/paper_doi_10_14722_ndss_2025_230530.json
new file mode 100644
index 0000000..02366d4
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14722_ndss_2025_230530.json
@@ -0,0 +1,58 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14722/ndss.2025.230530",
+    "title": "MTZK: Testing and Exploring Bugs in Zero-Knowledge (ZK) Compilers",
+    "authors": [
+      "Dongwei Xiao",
+      "Zhibo Liu",
+      "Yiteng Peng",
+      "Shuai Wang"
+    ],
+    "year": 2025,
+    "venue": "Network and Distributed System Security Symposium",
+    "doi": "10.14722/ndss.2025.230530",
+    "url": "https://doi.org/10.14722/ndss.2025.230530"
+  },
+  "summary": {
+    "text": "MTZK is a metamorphic testing framework for zero-knowledge compilers, which turn a program in a ZK domain-specific language into a circuit for proving and verification. The authors note that the correctness of these compilers is not well studied and that buggy ones let malicious users produce invalid proofs the verifier accepts. MTZK mutates compiler inputs according to designed metamorphic relations, so compilation correctness can be checked automatically from inputs and their variants. Across four industrial ZK compilers it uncovered 21 bugs, 15 promptly patched.",
+    "relationship_to_sqlancer": "Metamorphic testing applied to ZK compilers; SQLancer is cited as prior work in that testing tradition.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14722/ndss.2025.230530",
+    "source_sha256": "sha256:d20bf8ebb890281054a46f345940aeca2e6cde7e940d3010adbe3626913806b7",
+    "supporting_excerpts": [
+      "However, the correctness of ZK compilers is not well studied, and recent works have shown that de facto ZK compilers are buggy, which can allow malicious users to generate invalid proofs that are accepted by the verifier, causing security breaches and financial losses in cryptocurrency.",
+      "In this paper, we propose MTZK, a metamorphic testing framework to test ZK compilers and uncover incorrect compila-tions.",
+      "In the evaluation of four industrial ZK compilers, we successfully uncovered 21 bugs, out of which the developers have promptly patched 15."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14722/ndss.2025.230530",
+          "source_type": "paper_citation_context",
+          "excerpt": "MTZK shares a similar testing-based approach with quality assurance tools for critical systems, such as CPUs [56], [46], databases [89], [59], [60], [90], and operating systems [65], [16], to detect errors instead of proving their absence.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14722_ndss_2026_240198.json b/_data/impact/paper_notes/paper_doi_10_14722_ndss_2026_240198.json
new file mode 100644
index 0000000..eacc619
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14722_ndss_2026_240198.json
@@ -0,0 +1,90 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14722/ndss.2026.240198",
+    "title": "Efficiently Detecting DBMS Bugs through Bottom-up Syntax-based SQL Generation",
+    "authors": [
+      "Yu Liang",
+      "Peng Liu"
+    ],
+    "year": 2026,
+    "venue": "Network and Distributed System Security Symposium",
+    "doi": "10.14722/ndss.2026.240198",
+    "url": "https://doi.org/10.14722/ndss.2026.240198"
+  },
+  "summary": {
+    "text": "This paper proposes generating SQL for fuzzing bottom-up rather than top-down. Existing syntax-based generators start at the grammar root and expand downwards, which concentrates effort on shallow grammar and neglects the feature-rich rules deeper in the grammar space. The bottom-up method instead starts from an interesting grammar rule and backtracks to the root to form a syntax path, then expands and merges several such paths into diverse queries. The prototype, SQLBull, found 63 zero-day bugs across MySQL, MariaDB, CockroachDB, DuckDB and PostgreSQL.",
+    "relationship_to_sqlancer": "A SQL generation technique for DBMS fuzzing that positions itself against existing generation tools, SQLancer among the cited work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14722/ndss.2026.240198",
+    "source_sha256": "sha256:a7928c8ecc4584be469acb709daa33da1c5ca6f9a08be0c3d188efa9604b19d7",
+    "supporting_excerpts": [
+      "This paper proposes a new Bottom-up syntax-based SQL generation technique that puts more testing resources into exploring the feature-rich grammar rules.",
+      "A prototype tool, SQLBull , adopts the Bottom-up generation technique for fuzzing.",
+      "In the evaluation, SQLBull found 63 zero-day bugs from 5 well-tested DBMSs: MySQL , MariaDB , CockroachDB , DuckDB , and PostgreSQL ."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:24:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "qpg",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14722/ndss.2026.240198",
+          "source_type": "paper_citation_context",
+          "excerpt": "In this evaluation, we include two of the latest and most advanced testing techniques implemented in SQLancer which are capable of detecting DBMS memory errors, i.e., SQLancer +QPG [52] and SQLancer +DQP [53].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14722/ndss.2026.240198",
+          "source_type": "paper_citation_context",
+          "excerpt": "Although SQLancer was first introduced in 2022, it has become the most popular platform for implementing the latest SQL testing techniques [52]–[56].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14722/ndss.2026.240198",
+          "source_type": "paper_citation_context",
+          "excerpt": "Its latest improvement, SQLancer +QPG [52] and SQLancer +DQP [53], leverage the DBMS query plan and optimization hints to guide the query generation.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14722/ndss.2026.240198",
+          "source_type": "paper_citation_context",
+          "excerpt": "We first discuss the existing syntax-based testing tools for the DBMS testing community [30], [39], [52] in §VI-A and §VI-B.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14722/ndss.2026.240198",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer [30] is another template-based DBMS testing tool that detects DBMS memory and logic bugs [54]–[56].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3494124_3494139.json b/_data/impact/paper_notes/paper_doi_10_14778_3494124_3494139.json
new file mode 100644
index 0000000..314fa09
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3494124_3494139.json
@@ -0,0 +1,73 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3494124.3494139",
+    "title": "MT-Teql: Evaluating and Augmenting Neural NLIDB on Real-world Linguistic and Schema Variations",
+    "authors": [
+      "Pingchuan Ma",
+      "Shuai Wang"
+    ],
+    "year": 2021,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3494124.3494139",
+    "url": "https://doi.org/10.14778/3494124.3494139"
+  },
+  "summary": {
+    "text": "MT-Teql benchmarks the robustness of neural natural-language interfaces to databases against real-world variation in language and schema design. The authors argue that hold-out benchmark datasets cannot show how these systems handle such variation, and that annotating SQL for varied inputs by hand is prohibitive. MT-Teql instead applies semantics-preserving transformations to utterances and schemas to generate variants automatically. Across nine neural systems and 62,430 inputs it identified 15,433 defects, and using the error-triggering inputs for augmentation removed 46.5% of their errors.",
+    "relationship_to_sqlancer": "States it was inspired by recent advances in DBMS metamorphic testing, applying semantics-preserving transformation to text-to-SQL systems.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3494124.3494139",
+    "source_sha256": "sha256:12902cd48fdb692e00f44cc0f7301f37182add097c0ccc941e2c2c04de377ab1",
+    "supporting_excerpts": [
+      "However, existing benchmarks, using hold-out datasets, cannot provide thorough understanding of how good neural NLIDBs really are in real-world situations and its robustness against such variations.",
+      "To systematically assess the robustness of neural NLIDBs without extensive manual effort, we propose MT-Teql, a unified framework to benchmark NLIDBs against real-world language and schema variations.",
+      "Inspired by recent advances in DBMS metamorphic testing, MT-Teql implements semantics-preserving transformations on utterances and database schemas to generate their variants."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:12Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3494124.3494139",
+          "source_type": "paper_citation_context",
+          "excerpt": "To test DBMS with metamorphic relations, for instance, NoREC transforms a query into a non-optimized form and compares whether the optimized query and non-optimized query induce identical outputs [28].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3494124.3494139",
+          "source_type": "paper_citation_context",
+          "excerpt": "With recent advances in testing DBMS [28–30], software metamorphic testing-based approaches have become popular in assessing database systems.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3494124.3494139",
+          "source_type": "paper_citation_context",
+          "excerpt": "To date, MT has achieved major success in detecting bugs in DBMS [28–30] and NLP-related models [19, 27, 37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3611540_3611584.json b/_data/impact/paper_notes/paper_doi_10_14778_3611540_3611584.json
new file mode 100644
index 0000000..592ec05
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3611540_3611584.json
@@ -0,0 +1,69 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3611540.3611584",
+    "title": "A Demonstration of DLBD: Database Logic Bug Detection System",
+    "authors": [
+      "Xiu Tang",
+      "Sai Wu",
+      "Dongxiang Zhang",
+      "Ziyue Wang",
+      "Gongsheng Yuan",
+      "Gang Chen"
+    ],
+    "year": 2023,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3611540.3611584",
+    "url": "https://doi.org/10.14778/3611540.3611584"
+  },
+  "summary": {
+    "text": "DLBD is a demonstration of a system for automatically detecting database logic bugs. The authors note that current tools are limited to single-table queries and struggle with the absence of ground-truth results and with repeatedly exploring the same query space. DLBD generates schemas and queries automatically and retrieves ground-truth results, produces minimal test cases and root cause analysis for each bug, prunes the search space with heuristics and domain knowledge, avoids repetitive search, and runs on a distributed processing framework for scale.",
+    "relationship_to_sqlancer": "Addresses the ground-truth problem that Pivoted Query Synthesis introduced, extending it beyond single-table queries.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3611540.3611584",
+    "source_sha256": "sha256:1bf462047731518c0985f9ac9496670f20dbe72a28dfb61a46d782a4959e0915",
+    "supporting_excerpts": [
+      "Current debugging tools are limited to single table queries and struggle with issues like lack of ground-truth results and repetitive query space exploration.",
+      "DLBD offers holistic logic bug detection by providing automatic schema and query generation and ground-truth query result retrieval.",
+      "DLBD incorporates heuristics and domain-specific knowledge to efficiently prune the search space and employs query space exploration mechanisms to avoid the repetitive search."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3611540.3611584",
+          "source_type": "paper_citation_context",
+          "excerpt": "PQS [6] constructs queries to fetch a randomly selected tuple from a table, while NoREC [4] compares the results of randomly generated optimized queries and rewritten queries that DBMS cannot optimize.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3611540.3611584",
+          "source_type": "paper_citation_context",
+          "excerpt": "The tool employs several approaches to detect logic bugs, such as Pivoted Query Synthesis (PQS), Ternary Logic Partitioning (TLP), and Non-optimizing Reference Engine Construction (NoREC).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3636218_3636236.json b/_data/impact/paper_notes/paper_doi_10_14778_3636218_3636236.json
new file mode 100644
index 0000000..0e5e1cf
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3636218_3636236.json
@@ -0,0 +1,180 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3636218.3636236",
+    "title": "Testing Graph Database Systems via Graph-Aware Metamorphic Relations",
+    "authors": [
+      "Zeyang Zhuang",
+      "Penghui Li",
+      "Pingchuan Ma",
+      "Wei Meng",
+      "Shuai Wang"
+    ],
+    "year": 2023,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3636218.3636236",
+    "url": "https://doi.org/10.14778/3636218.3636236"
+  },
+  "summary": {
+    "text": "Gamera tests graph database systems with metamorphic relations designed around graph native structures, which the authors argue prior solutions were unaware of and therefore could not use to reveal many logic bugs. It defines three classes of graph-aware metamorphic relation, generates queries according to them to exercise diverse and complex graph operations, and checks whether results conform to the chosen relation. Across seven widely used graph systems including Neo4j and OrientDB it detected 39 logic bugs, and outperformed the prior tools Grand, GDsmith and GDBMeter.",
+    "relationship_to_sqlancer": "Extends metamorphic query transformation, the basis of SQLancer's oracles, with graph-specific relations, and its artifact reuses the SQLancer codebase.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3636218.3636236",
+    "source_sha256": "sha256:cbab56b22188e4e55923effd9da38e5ae26d2c21b6c3504111afe5d9bd35748b",
+    "supporting_excerpts": [
+      "However, the logic bugs largely cannot be revealed by prior solutions which are unaware of the graph native structures of the graph data.",
+      "In this paper, we propose Gamera (Graph-aware metamorphic relations), a novel metamorphic testing approach to uncover unknown logic bugs in GDBs.",
+      "Our experiments also demonstrated that Gamera significantly outperformed prior solutions including Grand, GD-smith and GDBMeter."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3636218.3636236",
+          "source_type": "paper_citation_context",
+          "excerpt": "We next propose the pattern partitioning technique to generate compound MRs by extending the query partitioning technique in the literature [34, 42].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3636218.3636236",
+          "source_type": "paper_citation_context",
+          "excerpt": "In relational database systems, SQLancer [18] leverages ternary logic partitioning (TLP) [42] and non-optimizing reference engine construction (NoREC) [41] to form MRs and detect logic bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3636218.3636236",
+          "source_type": "paper_citation_context",
+          "excerpt": "Similar to other software systems [41, 42], GDBs contain logic bugs, which could cause unexpected behaviors and lead to severe consequences.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3636218.3636236",
+          "source_type": "paper_citation_context",
+          "excerpt": "To date, MT methods have found hundreds of bugs in database systems, demonstrating its effectiveness [18, 41, 42].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3636218.3636236",
+          "source_type": "paper_citation_context",
+          "excerpt": "It ports the idea of ternary query partitioning [42] used in relational databases to GDBs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3636218.3636236",
+          "source_type": "paper_citation_context",
+          "excerpt": "However, most prior database MT works focus on relational database systems [18, 41, 42].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/cuhk-seclab/Gamera",
+          "source_type": "github_repository",
+          "excerpt": "Gamera in total detected 39 bugs and 5 new bugs from FalkorDB (, which is related to RedisGraph). All the results are listed here: [results.md](results.md).\n\n\n## Publication\nYou can find more details in our VLDB 2024 paper:\n[Testing Graph Database Systems via Graph-Aware Metamorphic Relations](https://www.vldb.org/pvldb/vol17/p836-zhuang.pdf)\n```\n@article{zhuang2024gamera,\n  title   = {Testing Graph Database Systems via Graph-aware Metamorphic Relations},",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/cuhk-seclab/Gamera/blob/main/GremlinChecker/src/main/java/org/gdbtesting/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "GremlinChecker/src/main/java/org/gdbtesting/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.gdbtesting (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "extends_technique",
+      "title": "Extends a SQLancer technique",
+      "value": "yes",
+      "method": "manual_curation",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3636218.3636236",
+          "source_type": "paper_citation_context",
+          "excerpt": "We next propose the pattern partitioning technique to generate compound MRs by extending the query partitioning technique in the literature [34, 42].",
+          "excerpt_is_verbatim": true,
+          "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+          "retrieved_at": "2026-09-06T07:08:48Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3636218.3636236",
+          "source_type": "paper_citation_context",
+          "excerpt": "It ports the idea of ternary query partitioning [42] used in relational databases to GDBs.",
+          "excerpt_is_verbatim": true,
+          "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+          "retrieved_at": "2026-09-06T07:08:48Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/cuhk-seclab/Gamera",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "renamed_sqlancer_package",
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/cuhk-seclab/Gamera",
+          "source_type": "github_repository",
+          "excerpt": "Gamera in total detected 39 bugs and 5 new bugs from FalkorDB (, which is related to RedisGraph). All the results are listed here: [results.md](results.md).\n\n\n## Publication\nYou can find more details in our VLDB 2024 paper:\n[Testing Graph Database Systems via Graph-Aware Metamorphic Relations](https://www.vldb.org/pvldb/vol17/p836-zhuang.pdf)\n```\n@article{zhuang2024gamera,\n  title   = {Testing Graph Database Systems via Graph-aware Metamorphic Relations},",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/cuhk-seclab/Gamera/blob/main/GremlinChecker/src/main/java/org/gdbtesting/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "GremlinChecker/src/main/java/org/gdbtesting/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.gdbtesting (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3659437_3659445.json b/_data/impact/paper_notes/paper_doi_10_14778_3659437_3659445.json
new file mode 100644
index 0000000..84b7b19
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3659437_3659445.json
@@ -0,0 +1,135 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3659437.3659445",
+    "title": "Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction",
+    "authors": [
+      "Jiansen Song",
+      "Wensheng Dou",
+      "Yu Gao",
+      "Ziyu Cui",
+      "Yingying Zheng",
+      "Dong Wang",
+      "Wei Wang",
+      "Jun Wei",
+      "Tao Huang"
+    ],
+    "year": 2024,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3659437.3659445",
+    "url": "https://doi.org/10.14778/3659437.3659445"
+  },
+  "summary": {
+    "text": "Radar detects logic bugs caused by metadata-related optimisations. DBMSs keep metadata such as integrity constraints and indexes and use it to optimise queries, so an incorrect metadata-driven optimisation yields wrong results. Radar builds a raw database that strips the metadata but keeps exactly the same data; since the two databases hold the same data, they must return the same result for a given query, and any inconsistency is a bug. A metadata-oriented optimisation strategy prioritises previously unseen metadata. Across five widely used DBMSs it detected 42 bugs, 38 confirmed as new.",
+    "relationship_to_sqlancer": "A metamorphic oracle for DBMS logic bugs whose artifact reuses the SQLancer codebase.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3659437.3659445",
+    "source_sha256": "sha256:ded1145218883c9cf6f7edeaeac75a678f6c28995748ccb226ad7a4d1427dd93",
+    "supporting_excerpts": [
+      "However, incorrect metadata-related optimizations can introduce metadata-related logic bugs, which can cause a DBMS to return an incorrect query result for a given query.",
+      "In this paper, we propose a general and effective testing approach,\n Raw database construction\n (Radar), to detect metadata-related logic bugs in DBMSs.",
+      "We implement and evaluate Radar on five widely-used DBMSs, and have detected 42 bugs, of which 38 have been confirmed as new bugs and 16 have been fixed by DBMS developers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/2212c954adfea8bd0c35fdf691c6ad50904cb5ab",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/JensonSung/radar",
+          "source_type": "github_repository",
+          "excerpt": "Replication package for \"Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction\", accepted at VLDB 2024\n  \n# Radar\n\nRadar is the implementation of raw database construction in this paper.\n\n# Getting Started",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T14:28:33Z"
+        },
+        {
+          "source_url": "https://github.com/JensonSung/radar/blob/main/src/sqlancer/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Repository retains SQLancer source files: src/sqlancer/Randomly.java, src/sqlancer/Main.java, src/sqlancer/common/oracle/TestOracle.java",
+          "retrieved_at": "2026-09-06T14:28:33Z"
+        },
+        {
+          "source_url": "https://github.com/JensonSung/radar/blob/main/pom.xml",
+          "source_type": "github_code",
+          "excerpt": "<groupId>com.sqlancer</groupId>",
+          "excerpt_is_verbatim": true,
+          "note": "pom.xml identifies the project as SQLancer.",
+          "retrieved_at": "2026-09-06T14:28:33Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/JensonSung/radar",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "randomly_java_present",
+        "retained_sqlancer_source_files",
+        "sqlancer_build_file_reference",
+        "sqlancer_package_structure",
+        "sqlancer_provider_directory"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/JensonSung/radar",
+          "source_type": "github_repository",
+          "excerpt": "Replication package for \"Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction\", accepted at VLDB 2024\n  \n# Radar\n\nRadar is the implementation of raw database construction in this paper.\n\n# Getting Started",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T14:28:33Z"
+        },
+        {
+          "source_url": "https://github.com/JensonSung/radar/blob/main/src/sqlancer/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Repository retains SQLancer source files: src/sqlancer/Randomly.java, src/sqlancer/Main.java, src/sqlancer/common/oracle/TestOracle.java",
+          "retrieved_at": "2026-09-06T14:28:33Z"
+        },
+        {
+          "source_url": "https://github.com/JensonSung/radar/blob/main/pom.xml",
+          "source_type": "github_code",
+          "excerpt": "<groupId>com.sqlancer</groupId>",
+          "excerpt_is_verbatim": true,
+          "note": "pom.xml identifies the project as SQLancer.",
+          "retrieved_at": "2026-09-06T14:28:33Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3712221_3712247.json b/_data/impact/paper_notes/paper_doi_10_14778_3712221_3712247.json
new file mode 100644
index 0000000..0be0c75
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3712221_3712247.json
@@ -0,0 +1,211 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3712221.3712247",
+    "title": "Semantic Conformance Testing of Relational DBMS",
+    "authors": [
+      "Shuang Liu",
+      "Chenglin Tian",
+      "Jun Sun",
+      "Ruifeng Wang",
+      "Wei Lu",
+      "Yongxin Zhao",
+      "Yinxing Xue",
+      "Junjie Wang",
+      "Xiaoyong Du"
+    ],
+    "year": 2024,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3712221.3712247",
+    "url": "https://doi.org/10.14778/3712221.3712247"
+  },
+  "summary": {
+    "text": "This paper introduces semantic conformance testing for relational DBMSs — checking a system against the SQL standard rather than against another system. Two obstacles are identified: the standard is written in ambiguous natural language and is not executable, and it is hard to generate queries covering all its keywords and parameters. The authors formally define SQL's denotational semantics and implement them in Prolog as an executable reference system, then add three coverage criteria over those semantics and a coverage-guided generation algorithm. Across six widely used RDBMSs it uncovered 19 bugs and 13 inconsistencies, all confirmed.",
+    "relationship_to_sqlancer": "Builds an executable reference to test what metamorphic oracles cannot, comparing against SQLancer's approach, and its artifact reuses the SQLancer codebase.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3712221.3712247",
+    "source_sha256": "sha256:47f15594c034f920cacd26456264449c565bb595e5d2fd834385435de305c92b",
+    "supporting_excerpts": [
+      "First, the SQL standard specification, documented in natural language, tends to be ambiguous and is not directly executable.",
+      "Firstly, we formally define the denotational semantics of SQL and implement them in Prolog, creating an executable reference RDBMS for differential testing against existing RDBMSs.",
+      "Lastly, we apply our approach to six widely-used and thoroughly tested RDBMSs, e.g., MySQL, PostgreSQL and OceanBase, uncovering 19 bugs and 13 inconsistencies, all of which are confirmed by RDBMS developers."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/bb1da01b4752d31f2f96f3abcc5502b58bedbc4e",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+          "retrieved_at": "2026-09-06T17:02:28Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2406.09469",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compared S EM C ON T with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS. NoREC constructs two semantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2406.09469",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compare S EM C ON T with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2406.09469",
+          "source_type": "paper_citation_context",
+          "excerpt": "PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2406.09469",
+          "source_type": "paper_citation_context",
+          "excerpt": "It is noteworthy that all four databases have been extensively tested by existing methods [40]–[42], yet S EM C ON T is still able to detect bugs that were not detected by those approaches.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2406.09469",
+          "source_type": "paper_citation_context",
+          "excerpt": "In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2406.09469",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.vldb.org/pvldb/vol18/p850-liu.pdf",
+          "source_type": "paper",
+          "excerpt": "Both NoREC and TLP are implemented in SQLancer [ 38] and SQLRight [ 22].",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that its implementation is built on SQLancer.",
+          "retrieved_at": "2026-09-06T17:02:28Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2406.09469",
+          "source_type": "paper",
+          "excerpt": "Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34].",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that its implementation is built on SQLancer.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.vldb.org/pvldb/vol18/p850-liu.pdf",
+          "source_type": "paper",
+          "excerpt": "We compared SemConT with TLP [ 32] and NoREC [ 31], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T17:02:28Z"
+        },
+        {
+          "source_url": "https://www.vldb.org/pvldb/vol18/p850-liu.pdf",
+          "source_type": "paper",
+          "excerpt": "We compare SemConT with two state-of-the-art approaches TLP [ 31] and NoREC [ 32], which are metamorphic testing approaches for relational DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T17:02:28Z"
+        },
+        {
+          "source_url": "https://www.vldb.org/pvldb/vol18/p850-liu.pdf",
+          "source_type": "paper",
+          "excerpt": "We also compared the memory usage of SemConT and SQLancer during a 6-hour test on six databases.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it evaluates against SQLancer or one of its techniques.",
+          "retrieved_at": "2026-09-06T17:02:28Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2406.09469",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compared S EM C ON T with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+          "retrieved_at": "2026-09-06T07:08:48Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://arxiv.org/abs/2406.09469",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compared S EM C ON T with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS. NoREC constructs two semantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Ternary Logic Partitioning (TLP) as state of the art.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        },
+        {
+          "source_url": "https://arxiv.org/abs/2406.09469",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compare S EM C ON T with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Ternary Logic Partitioning (TLP) as state of the art.",
+          "retrieved_at": "2026-09-06T15:33:21Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3725688_3725698.json b/_data/impact/paper_notes/paper_doi_10_14778_3725688_3725698.json
new file mode 100644
index 0000000..80c75be
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3725688_3725698.json
@@ -0,0 +1,74 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3725688.3725698",
+    "title": "QOVIS: Understanding and Diagnosing Query Optimizer via a Visualization-assisted Approach (Revision)",
+    "authors": [
+      "Zhengxin You",
+      "Qiaomu Shen",
+      "M. Yiu",
+      "Bo Tang"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3725688.3725698",
+    "url": "https://doi.org/10.14778/3725688.3725698"
+  },
+  "summary": {
+    "text": "QOVIS helps developers understand and diagnose query optimizers through visualization. The authors identify the obstacles: an optimizer generates hundreds or thousands of plans per query, and the transformation logic relating them is hard to follow. QOVIS is built in three layers — data preprocessing, transformation logic computation, and visual analysis — and was evaluated through user, case and performance studies. A user study with 24 database developers and researchers found it significantly reduced the time to investigate optimizer bugs, and it was applied to real reported bugs in Apache Spark, Apache Hive and DuckDB.",
+    "relationship_to_sqlancer": "Diagnoses optimizer bugs after they are found, complementing the automated detection SQLancer performs.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3725688.3725698",
+    "source_sha256": "sha256:b45a2efe5a274973d8a8e121e80979af933d2b756f04849c3863072be8ce76b7",
+    "supporting_excerpts": [
+      "Understanding and diagnosing query optimizers is crucial to guarantee the correctness and efficiency of query processing in database systems.",
+      "In this work, we propose QOVIS to overcome these challenges, which identifies the query optimization bugs/issues and investigates their root causes via a visualization-assisted approach.",
+      "In particular, our user study (on 24 database developers and researchers) confirms that QOVIS significantly reduces the time required to investigate the bugs/errors in the query optimizer."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3725688.3725698",
+          "source_type": "paper_citation_context",
+          "excerpt": "In addition, automatic logic bug detection tools [27, 50, 59, 60] are proposed to identify the logic bugs during the query processing.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3725688.3725698",
+          "source_type": "paper_citation_context",
+          "excerpt": "Many studies have been proposed to automatically detect the issues in the relational data-base management system (RDBMS) [18, 50, 59].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3725688.3725698",
+          "source_type": "paper_citation_context",
+          "excerpt": "Rigger et al. [50] focus on detecting logical bugs when the generated plans produce incorrect results w.r.t the ground truth.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3725688_3725713.json b/_data/impact/paper_notes/paper_doi_10_14778_3725688_3725713.json
new file mode 100644
index 0000000..ba87331
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3725688_3725713.json
@@ -0,0 +1,157 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3725688.3725713",
+    "title": "Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving",
+    "authors": [
+      "Xiyue Gao",
+      "Zhuang Liu",
+      "Yiran Shen",
+      "Hui Li",
+      "Yingfan Liu",
+      "H. Xiao",
+      "Yanguo Peng",
+      "Jiangtao Cui"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3725688.3725713",
+    "url": "https://doi.org/10.14778/3725688.3725713"
+  },
+  "summary": {
+    "text": "Fucci is a fuzzing framework for transaction management. The authors found that transaction implementations in mainstream systems including MySQL, MariaDB and TiDB can diverge from what their own documentation claims. Fucci combines Random Conflict Construction, which guarantees read-write or write-write conflicts between generated transactions so cases are valid; Multilevel Constraint Solving, which uses an external multi-version control system to resolve data visibility for the oracle; and Experience-driven Automatic Simplification for readable reports. It found 6 previously unknown and 14 known duplicate transaction bugs.",
+    "relationship_to_sqlancer": "Its artifact reuses the SQLancer codebase, carrying the approach into transaction testing.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3725688.3725713",
+    "source_sha256": "sha256:87425d08dcea8ff7e436697e38215a779b42ca10c76ffcd44d723db97afb6adb",
+    "supporting_excerpts": [
+      "However, through our study on existing solutions on transaction management, we found that transaction implementations in some mainstream databases, such as MySQL, MariaDB and TiDB, may violate what they claim in their documentation, in the form of incorrect database state or query results.",
+      "Since there is still a lack of efficient and comprehensive testing methods to detect bugs within transaction management implementation for off-the-shelf DBMSs at present, we propose Fucci, a fuzzing framework, to solve the problem.",
+      "Accordingly, 6 previously unknown transaction bugs and 14 known duplicate transaction bugs have been newly discovered, most of which have been officially acknowledged."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3725688.3725713",
+          "source_type": "paper_citation_context",
+          "excerpt": "TLP [41] ensures predicate evaluations always fall within TRUE , FALSE , or NULL , enabling queries to be decomposed into three partitioned queries, also applicable for detecting transaction isolation bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3725688.3725713",
+          "source_type": "paper_citation_context",
+          "excerpt": "DT2 and Troc were selected as base-lines because other database fuzzing tools, such as SQLsmith [31], Squirrel [49], and NoREC [41], focus primarily on non-transactional modes.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3725688.3725713",
+          "source_type": "paper_citation_context",
+          "excerpt": "Existing oracles can be divided into four types: crash oracles [46, 49], differential oracles [25], metamorphic oracles [43] and constraint-solving oracles [41, 45].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3725688.3725713",
+          "source_type": "paper_citation_context",
+          "excerpt": "However, existing reducers [23, 41] are limited to simplifying single SQL statements, not entire transactions.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3725688.3725713",
+          "source_type": "paper_citation_context",
+          "excerpt": "Test cases can be created using either generation-based [26, 41, 45] or mutation-based [19, 23, 35] methods.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3725688.3725713",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoREC [40] rewrites statements to bypass database optimization, detecting incorrect optimizations.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/Reverie4u/Fucci",
+          "source_type": "github_repository",
+          "excerpt": "Artifact for the paper \"Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving\".\n# Fucci\r\n\r\nThis is the artifact for the paper \"Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving\".",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/Reverie4u/Fucci/blob/main/src/main/java/fucci/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/main/java/fucci/Randomly.java is SQLancer's Randomly.java, with the package renamed to fucci (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/Reverie4u/Fucci",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "renamed_sqlancer_package",
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/Reverie4u/Fucci",
+          "source_type": "github_repository",
+          "excerpt": "Artifact for the paper \"Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving\".\n# Fucci\r\n\r\nThis is the artifact for the paper \"Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving\".",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/Reverie4u/Fucci/blob/main/src/main/java/fucci/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "src/main/java/fucci/Randomly.java is SQLancer's Randomly.java, with the package renamed to fucci (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3734839_3734861.json b/_data/impact/paper_notes/paper_doi_10_14778_3734839_3734861.json
new file mode 100644
index 0000000..e5a8f3e
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3734839_3734861.json
@@ -0,0 +1,104 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3734839.3734861",
+    "title": "Detecting Schema-Related Logic Bugs in Relational DBMSs via Equivalent Database Construction",
+    "authors": [
+      "Jiansen Song",
+      "Wensheng Dou",
+      "Yingying Zheng",
+      "Yu Gao",
+      "Ziyu Cui",
+      "Wei Wang",
+      "Jun Wei"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3734839.3734861",
+    "url": "https://doi.org/10.14778/3734839.3734861"
+  },
+  "summary": {
+    "text": "DDLCheck targets logic bugs tied to schema handling rather than to queries. Its insight is that a complex sequence of DDL statements and a simple sequence of CREATE statements that produce the same schema should behave identically, so running the same SQL against both databases and comparing results exposes bugs in how schema-related information — schema changes, tablespace allocation, block-level layout — is maintained and used. A DDL-sequence-oriented optimisation strategy broadens the schema states explored. On six widely used DBMSs it detected 34 bugs, 29 previously unknown.",
+    "relationship_to_sqlancer": "Extends differential logic-bug detection from DQL statements, where the authors say existing approaches concentrate, to schema definition.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3734839.3734861",
+    "source_sha256": "sha256:32dab9f88b2287f962bac37bc952f96e59b604818e934b5852e6cbdcc6759fe4",
+    "supporting_excerpts": [
+      "Existing approaches mainly focus on detecting logic bugs in DQL statements, but are ineffective in detecting schema-related logic bugs.",
+      "In this paper, we propose a novel and general testing approach, DDLCheck, to effectively detect schema-related logic bugs in relational DBMSs.",
+      "Any discrepancy between their execution results indicates a schema-related logic bug."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3734839.3734861",
+          "source_type": "paper_citation_context",
+          "excerpt": "However, differential testing fails to test SQL features specific to individual DBMSs. Metamorphic testing [15, 31, 43, 47, 48] detects logic bugs in individual DBMSs by constructing equivalent SELECT statements and observing differences among their outputs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3734839.3734861",
+          "source_type": "paper_citation_context",
+          "excerpt": "Existing testing approaches for relational DBMSs mainly focus on detecting logic bugs in the SELECT statements [15, 24, 31, 42, 43, 47, 48, 50, 51, 54].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3734839.3734861",
+          "source_type": "paper_citation_context",
+          "excerpt": "We select MariaDB, CockroachDB, and TiDB, because they have been thoroughly tested by existing approaches [31, 38, 43, 44, 47, 48, 51, 52, 58, 63].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3734839.3734861",
+          "source_type": "paper_citation_context",
+          "excerpt": "Other approaches construct equivalent SELECT statements on the same database [14, 15, 27, 31, 41, 46– 48, 56, 61].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3734839.3734861",
+          "source_type": "paper_citation_context",
+          "excerpt": "Randomly generating statements is a commonly-used and effective approach in DBMS testing [31, 47–49, 51, 52].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.14778/3734839.3734861",
+          "source_type": "paper_citation_context",
+          "excerpt": "Second, the DBMS should have been thoroughly tested by existing approaches [15, 47, 48, 51, 52].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3742728_3742747.json b/_data/impact/paper_notes/paper_doi_10_14778_3742728_3742747.json
new file mode 100644
index 0000000..8b8f96b
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3742728_3742747.json
@@ -0,0 +1,113 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3742728.3742747",
+    "title": "Simple Testing Can Expose Most Critical Transaction Bugs: Understanding and Detecting Write-Specific Serializability Violations in Database Systems",
+    "authors": [
+      "Ziyu Cui",
+      "Wensheng Dou",
+      "Yu Gao",
+      "Rui Yang",
+      "Yingying Zheng",
+      "Jiansen Song",
+      "Yuan Feng",
+      "Jun Wei"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3742728.3742747",
+    "url": "https://doi.org/10.14778/3742728.3742747"
+  },
+  "summary": {
+    "text": "This paper proposes write-specific serializability as a test oracle for concurrent transactions: a schedule should leave the database in the same state as a corresponding serial schedule of the same writes. An empirical study of 35 critical transaction bugs from six widely used DBMSs shows the property is effective at exposing them. WriteCheck implements it by comparing the final database states produced by the original and serial schedules, and found 22 violations across six production-grade DBMSs, 11 confirmed as new critical transaction bugs.",
+    "relationship_to_sqlancer": "Extends the idea of a general correctness property checkable without a reference implementation from queries to transactions, and its artifact reuses the SQLancer codebase.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3742728.3742747",
+    "source_sha256": "sha256:c3fb05a763684f6fa31aec6c494d113ebcaa2d2bd121157f4675a21ef2db3a5b",
+    "supporting_excerpts": [
+      "However, we lack an effective test oracle to determine whether a DBMS produces a correct database state for a given concurrent transaction schedule.",
+      "In this paper, we propose a general property for concurrent transaction schedules,\n write-specific serializability\n , in which a schedule of concurrent transactions should produce the same database state as a corresponding serial schedule of the same transactions.",
+      "We evaluate WriteCheck on the latest versions of six production-grade DBMSs, and have found 22 write-specific serializability violations, 11 of which have been confirmed as new critical transaction bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:31:04Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/3f929c9c702538e09f4f88cf69415ceb25d3ba77",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Query Plan Guidance (QPG), Differential Query Plans (DQP).",
+          "retrieved_at": "2026-09-06T17:02:28Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://www.vldb.org/pvldb/vol18/p2547-cui.pdf",
+          "source_type": "paper",
+          "excerpt": "We implement WriteCheck based on SQLancer [ 14], and extend SQLancer to detect WSSviolations.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that its implementation is built on SQLancer.",
+          "retrieved_at": "2026-09-06T17:02:28Z"
+        }
+      ]
+    },
+    {
+      "relationship": "extends_technique",
+      "title": "Extends a SQLancer technique",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.vldb.org/pvldb/vol18/p2547-cui.pdf",
+          "source_type": "paper",
+          "excerpt": "We implement WriteCheck based on SQLancer [ 14], and extend SQLancer to detect WSSviolations.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it extends or adapts a SQLancer technique.",
+          "retrieved_at": "2026-09-06T17:02:28Z"
+        },
+        {
+          "source_url": "https://www.vldb.org/pvldb/vol18/p2547-cui.pdf",
+          "source_type": "paper",
+          "excerpt": "We utilize SQLancer [ 58–60] to generate initial databases and individual SQL statements, and extend SQLancer to generate transactions and transaction test cases.",
+          "excerpt_is_verbatim": true,
+          "note": "The paper states that it extends or adapts a SQLancer technique.",
+          "retrieved_at": "2026-09-06T17:02:28Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3749646_3749661.json b/_data/impact/paper_notes/paper_doi_10_14778_3749646_3749661.json
new file mode 100644
index 0000000..6d5aabc
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3749646_3749661.json
@@ -0,0 +1,67 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3749646.3749661",
+    "title": "PBench: Workload Synthesizer with Real Statistics for Cloud Analytics Benchmarking",
+    "authors": [
+      "Yan Zhou",
+      "Chunwei Liu",
+      "B. Urgaonkar",
+      "Zhengle Wang",
+      "M. Mueller",
+      "Chao Zhang",
+      "Songyue Zhang",
+      "Pascal Pfeil",
+      "Dominik Horn",
+      "Zhengchun Liu",
+      "Davide Pagano",
+      "Tim Kraska",
+      "Samuel Madden",
+      "Ju Fan"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3749646.3749661",
+    "url": "https://doi.org/10.14778/3749646.3749661"
+  },
+  "summary": {
+    "text": "PBench synthesises analytics workloads that reproduce the execution statistics of real cloud workloads. Standard benchmarks use fixed query patterns and miss real production behaviour, while released workload traces lack the queries and databases needed to be benchmarks. PBench selects and combines components from existing benchmarks and augments them with new ones, using multi-objective optimisation to balance performance metrics against operator distributions, a timestamp assignment method for temporal dynamics, and LLM-generated components to close remaining gaps. It cuts approximation error by up to 6x.",
+    "relationship_to_sqlancer": "Benchmark synthesis rather than correctness testing; SQLancer is among the cited work on generating database workloads.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3749646.3749661",
+    "source_sha256": "sha256:63124ed2cce71f6ece3a4f693aff4923c313d7b797635a68dc681082d09dd718",
+    "supporting_excerpts": [
+      "To overcome this limitation, this paper studies a new problem of\n workload synthesis with real statistics\n , which generates\n synthetic workloads\n that closely approximate real execution statistics, including key performance metrics and operator distributions.",
+      "To address this problem, we propose PBench, a novel workload synthesizer that constructs synthetic workloads by (1) selecting and combining workload components from existing benchmarks and (2) augmenting new workload components.",
+      "Experimental results show that PBench reduces approximation error by up to 6X compared to state-of-the-art methods."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3749646.3749661",
+          "source_type": "paper_citation_context",
+          "excerpt": "Then, the generated synthetic workload o \" ers an e \" ective so-lution for downstream tasks , like database benchmarking [9, 39, 40], performance tuning [30], and bug detection [3], allowing database developers to perform realistic evaluations while preserving privacy and avoiding exposure of…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3749646_3749683.json b/_data/impact/paper_notes/paper_doi_10_14778_3749646_3749683.json
new file mode 100644
index 0000000..080b14c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3749646_3749683.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3749646.3749683",
+    "title": "SQLStorm: Taking Database Benchmarking into the LLM Era",
+    "authors": [
+      "Tobias Schmidt",
+      "Viktor Leis",
+      "Peter A. Boncz",
+      "Thomas Neumann"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3749646.3749683",
+    "url": "https://doi.org/10.14778/3749646.3749683"
+  },
+  "summary": {
+    "text": "SQLStorm builds database benchmarks with large language models, and releases a concrete benchmark over a real-world dataset at 1 GB, 12 GB and 220 GB scales with more than 18,000 queries. The authors highlight that generating workloads this way is cheap — about $15 for 22 MB of queries — and, more importantly, covers far more SQL functionality than hand-built benchmarks. They see the main uses as improving SQL compatibility between systems, finding and fixing crashes and errors, and improving cardinality estimators and query optimizers.",
+    "relationship_to_sqlancer": "Generates queries at scale to expose crashes and errors, a goal it shares with the DBMS testing work it cites.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3749646.3749683",
+    "source_sha256": "sha256:1bf4f5ad5c0112f8f3512ca53ef5fa4d3a1f5a293e6ad5910ae14ae440358490",
+    "supporting_excerpts": [
+      "In this paper, we introduce a new methodology for constructing database benchmarks using Large Language Models (LLMs), as well as SQLStorm v1.0, a concrete benchmark on a real-world dataset of three sizes (1 GB, 12 GB, 220 GB) consisting of over 18 K queries.",
+      "This methodology of using AI to generate query workloads breaks new ground, not only in its ability to cheaply ($15) generate huge volumes (22 MB) of realistic queries but especially because it greatly expands the amount of SQL functionality and query constructions that is covered, compared to human-written SQL benchmarks such as TPC-H, TPC-DS, and JOB.",
+      "The use cases of SQLStorm that we think will advance data systems most are: (i) improving SQL compatibility between systems, (ii) increasing system quality by identifying crashes/errors and fixing those, (iii) improving cardinality estimators and query optimizers, by identifying trends and opportunities (queries where other systems do much better), as well as (iv) overall system performance, both in terms of speed and robustness."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:31:04Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3749646.3749683",
+          "source_type": "paper_citation_context",
+          "excerpt": "A final relevant field is database testing and database fuzzing [44].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3797919_3797928.json b/_data/impact/paper_notes/paper_doi_10_14778_3797919_3797928.json
new file mode 100644
index 0000000..54e2a22
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3797919_3797928.json
@@ -0,0 +1,58 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3797919.3797928",
+    "title": "Dinkel: State-Aware and Granular Framework for Validating Graph Databases",
+    "authors": [
+      "C. Wüst",
+      "Zu-Ming Jiang",
+      "Zhendong Su"
+    ],
+    "year": 2024,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3797919.3797928",
+    "url": "https://doi.org/10.14778/3797919.3797928"
+  },
+  "summary": {
+    "text": "Dinkel tests graph database systems, addressing two limitations the authors see in existing approaches: they cannot generate complex valid queries that reach deep code, and they lack general oracles for arbitrary queries. For generation, Dinkel models query context and graph schema as graph state and builds queries clause by clause, updating the state so each clause references correct information. For validation, it applies clause-level and expression-level transformations that preserve semantics on arbitrary queries and checks the results match. It found 127 bugs across three systems, 113 confirmed and 33 logic bugs.",
+    "relationship_to_sqlancer": "Builds semantics-preserving query transformations as a general oracle for graph databases, the approach SQLancer established for SQL.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3797919.3797928",
+    "source_sha256": "sha256:9504111c70a66b75d859bd135068d53484a7a73d0f0bb9a65f9a68cea8db507d",
+    "supporting_excerpts": [
+      "However, they all suffer from two key limitations: (1) insufficient support for generating complex and valid queries to exercise deep GDBMS code, and (2) lack of general oracles to validate the execution correctness of arbitrary queries.",
+      "Second, to generally validate query results, we introduce two fine-grained query transformations: clause-level and expression-level transformations.",
+      "In total, we found 127 bugs, among which 113 were confirmed, 84 were fixed, and 33 were logic bugs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:33:22Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3797919.3797928",
+          "source_type": "paper_citation_context",
+          "excerpt": "Moreover, we propose two fine-grained query transformations: clause-level transformations and expression-level transformations, which can operate on arbitrary Cypher queries to validate their correctness.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3819518_3819547.json b/_data/impact/paper_notes/paper_doi_10_14778_3819518_3819547.json
new file mode 100644
index 0000000..9587ae8
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3819518_3819547.json
@@ -0,0 +1,61 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3819518.3819547",
+    "title": "Testing Graph Databases via Transformations Between Fixed-Length and Variable-Length Queries",
+    "authors": [
+      "Jin-Xin Gui",
+      "Yuanhong Lan",
+      "Longlong Lu",
+      "Yifei Lu",
+      "Minxue Pan"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3819518.3819547",
+    "url": "https://doi.org/10.14778/3819518.3819547"
+  },
+  "summary": {
+    "text": "PATHTest is a metamorphic testing approach for graph database systems built on result-equivalent transformations between fixed-length and variable-length queries. It pairs an iterative generator that produces diverse, non-empty variable-length queries with three transformation rules that capture equivalence between the two query forms, so both logic bugs and unexpected errors surface. Across seven widely used graph systems it revealed 41 previously unknown bugs, 24 of them logic bugs, and reports that none were within reach of the seven existing approaches it compares against.",
+    "relationship_to_sqlancer": "A metamorphic logic-bug oracle for graph databases, in the tradition SQLancer's oracles established and compared against the state-of-the-art approaches in that line.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3819518.3819547",
+    "source_sha256": "sha256:a27bb14bf57405d93becc296ee00ecfc46e8848dc68949d25b22f0961525dbd4",
+    "supporting_excerpts": [
+      "This paper presents a novel metamorphic testing approach named PATHTest that exploits result-equivalent transformations between fixed-length and variable-length queries.",
+      "Extensive evaluation on PATHTest across seven real-world, widely-used GDBMSs demonstrates the superiority of PATHTest, with 41 previously unknown bugs revealed, among which 24 are logic bugs, and 17 correspond to unexpected errors.",
+      "To note, all 41 bugs are beyond the reach of the seven existing state-of-the-art testing approaches."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp",
+        "cert",
+        "coddtest"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/64f9059bdd2cdc2dd0b46ca064ac5e59a2ce361b",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Cardinality Estimation Restriction Testing (CERT), Constant-Optimization-Driven Testing (CODDTest).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3828612_3828639.json b/_data/impact/paper_notes/paper_doi_10_14778_3828612_3828639.json
new file mode 100644
index 0000000..39afc13
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3828612_3828639.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3828612.3828639",
+    "title": "ReSequel: Robust LLM-assisted Query Rewriting and Optimization using Templatization and Sampling",
+    "authors": [
+      "Saeed Fathollahzadeh",
+      "Essam Mansour",
+      "Matthias Boehm"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": "10.14778/3828612.3828639",
+    "url": "https://doi.org/10.14778/3828612.3828639"
+  },
+  "summary": {
+    "text": "ReSequel is an outer optimisation layer that rewrites SQL with a large language model on top of an existing DBMS. It uses catalog and statistical metadata to infer template-specific rules that steer the model toward useful transformations, then generates, verifies and ranks candidate rewrites on sampled data to keep results correct while improving runtime. Across eight benchmarks and PostgreSQL, MySQL and DuckDB it reports workload-level speedups of up to 16x over the native systems and 22x over LLM-based rewriting systems.",
+    "relationship_to_sqlancer": "Query rewriting for performance; the semantic equivalence it must preserve is the same property SQLancer's oracles test for, and SQLancer appears among its citations.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.14778/3828612.3828639",
+    "source_sha256": "sha256:10845eca065424de510171dc9f03aaa9070441bd6a00fd2dd12444559016edd6",
+    "supporting_excerpts": [
+      "We present ReSequel, an outer optimization layer on top of existing DBMSs to rewrite SQL queries using LLMs.",
+      "ReSequel leverages catalog and statistical metadata to infer template-specific rules that guide the LLM toward effective query transformations.",
+      "We generate, verify, and rank rewritten query variants on sampled data to ensure result correctness and runtime improvements."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.14778/3828612.3828639",
+          "source_type": "paper_citation_context",
+          "excerpt": "Beyond these methods, prior work on SQL testing [78] and text-to-SQL generation [74] compare query results directly.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_14778_3836663_3836700.json b/_data/impact/paper_notes/paper_doi_10_14778_3836663_3836700.json
new file mode 100644
index 0000000..c64ddf8
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_14778_3836663_3836700.json
@@ -0,0 +1,50 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.14778/3836663.3836700",
+    "title": "Detecting Data-Type-Related Logic Bugs in Relational DBMSs via Compatible Database Construction",
+    "authors": [
+      "Jiansen Song",
+      "Wensheng Dou",
+      "Yingying Zheng",
+      "Yu Gao",
+      "Quanqing Xu",
+      "Ziyu Cui",
+      "Xudong Xie",
+      "Hongtao Zhou",
+      "Jiaying Zhou",
+      "Jun Wei",
+      "Wei Wang"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3836663.3836700",
+    "url": "https://doi.org/10.14778/3836663.3836700"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp",
+        "dqp",
+        "coddtest"
+      ],
+      "sources": [
+        {
+          "source_url": "https://openalex.org/W7208713041",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "OpenAlex records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Differential Query Plans (DQP), Constant-Optimization-Driven Testing (CODDTest).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_22399_ijcesen_5462.json b/_data/impact/paper_notes/paper_doi_10_22399_ijcesen_5462.json
new file mode 100644
index 0000000..43b1520
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_22399_ijcesen_5462.json
@@ -0,0 +1,88 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.22399/ijcesen.5462",
+    "title": "Benchmarking Autonomy: A Taxonomy of Human-in-the-Loop Checkpoints for AI-Generated Transformation Code",
+    "authors": [
+      "Jagan Ankam"
+    ],
+    "year": 2022,
+    "venue": "International Journal of Computational and Experimental Science and Engineering",
+    "doi": "10.22399/ijcesen.5462",
+    "url": "https://doi.org/10.22399/ijcesen.5462"
+  },
+  "summary": {
+    "text": "A benchmark and taxonomy for human review of AI-generated data transformations. The premise is that generated SQL, Spark and dbt code can compile, run and preserve the expected schema while silently changing what downstream data means. Across 12 transformation tasks, nine with seeded defects, the authors evaluate three automated checkpoint families: schema checks and data-quality rules each catch one of nine defects, while comparison against a trusted baseline catches seven, and all combined catch eight. The one residual defect preserves schema, distributions, row counts and tolerance-bounded aggregates, and is found only by reviewing logic and business intent together.",
+    "relationship_to_sqlancer": "Argues that semantic oracles outperform structural checks for silently wrong results, the same premise as SQLancer's oracles.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.22399/ijcesen.5462",
+    "source_sha256": "sha256:598c58c7e8ecd8ecb32293c97395777536912fa2b24d8990f020c3741d381d49",
+    "supporting_excerpts": [
+      "A transformation may compile, complete successfully, and preserve an expected schema while silently changing the meaning of downstream data.",
+      "Trusted-baseline comparison detects seven defects (77.8%).",
+      "First, validation depth matters more than check quantity: semantic oracles materially outperform purely structural controls."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:38:27Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.22399/ijcesen.5462",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoREC compares an optimizable query with an equivalent form intended to suppress optimization [18].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.22399/ijcesen.5462",
+          "source_type": "paper_citation_context",
+          "excerpt": "Database testing research develops oracles for defects in database engines rather than defects in application-level transformation intent [17]–[21].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.22399/ijcesen.5462",
+          "source_type": "paper_citation_context",
+          "excerpt": "Pivoted Query Synthesis constructs queries expected to retrieve a selected row [17].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.22399/ijcesen.5462",
+          "source_type": "paper_citation_context",
+          "excerpt": "Ternary Logic Partitioning checks relations among partitioned query results [19].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.22399/ijcesen.5462",
+          "source_type": "paper_citation_context",
+          "excerpt": "[17], [18], [19] Equivalent or partitioned query results",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_32672_jnkti_v6i1_5830.json b/_data/impact/paper_notes/paper_doi_10_32672_jnkti_v6i1_5830.json
new file mode 100644
index 0000000..77a250d
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_32672_jnkti_v6i1_5830.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.32672/jnkti.v6i1.5830",
+    "title": "Siklus Hidup Pengembangan Sistem Basis Data Pada Sistem Informasi Buku Tamu di Badan Pusat Statistik Kabupaten Kediri Menggunakan MySQL",
+    "authors": [
+      "Ahmad Niamilah",
+      "A. Alfin",
+      "Iin Kurniasari"
+    ],
+    "year": 2023,
+    "venue": "Jurnal Nasional Komputasi dan Teknologi Informasi (JNKTI)",
+    "doi": "10.32672/jnkti.v6i1.5830",
+    "url": "https://doi.org/10.32672/jnkti.v6i1.5830"
+  },
+  "summary": {
+    "text": "An Indonesian-language paper applying the Database System Development Lifecycle to a guest book information system at the Kediri Central Agency of Statistics. The system uses MySQL, and the study implements and tests four data manipulation statements — Select, Insert, Update and Delete — for retrieving data from one or more tables, inserting rows, modifying rows and deleting them. The result is a database design model intended for a web-based guest book information system.",
+    "relationship_to_sqlancer": "An applied database design study; the connection is the citation recorded in the citation graph.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.32672/jnkti.v6i1.5830",
+    "source_sha256": "sha256:e5fcb40c7d64b1e270fa04ad2878c2235fe6fddbc4a4a95ab054aabf77e87363",
+    "supporting_excerpts": [
+      "Database system development Lifecycle (DSDLC) is a method for designing and developing databases.",
+      "The guest book information system at Kediri Central Agency of Statistics uses MySQL Relational Database Management System (RDBMS).",
+      "In this research, there are 4 (four) commands was implemented in Guestbook Information System Database, namely Select, Insert, Update and Delete."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.32672/jnkti.v6i1.5830",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQL memiliki lima bagian dalam pemrosesan data, yaitu data retrieving, data definition, data manipulation, data control, dan data transaction language[9].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_3390_app13042519.json b/_data/impact/paper_notes/paper_doi_10_3390_app13042519.json
new file mode 100644
index 0000000..ebcde54
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_3390_app13042519.json
@@ -0,0 +1,67 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.3390/app13042519",
+    "title": "Squill: Testing DBMS with Correctness Feedback and Accurate Instantiation",
+    "authors": [
+      "Shihao Wen",
+      "Peng Jia",
+      "Pin Yang",
+      "Chi Hu"
+    ],
+    "year": 2023,
+    "venue": "Applied Sciences",
+    "doi": "10.3390/app13042519",
+    "url": "https://doi.org/10.3390/app13042519"
+  },
+  "summary": {
+    "text": "Squill improves greybox fuzzing for DBMSs on two fronts. Existing seed scheduling ignores whether seeds are correct, which the authors argue wastes effort, so Squill uses seed correctness as feedback to guide mutation. And current tools cannot correctly generate SQL with nested structures, so Squill adds semantics-aware instantiation that fills in semantics using context information collected from AST nodes. Built on Squirrel and evaluated on MySQL, MariaDB and OceanBase, it explored 29% more paths and found 3.4 times more bugs than the existing tool, with 9 CVEs assigned.",
+    "relationship_to_sqlancer": "Greybox DBMS fuzzing; SQLancer is cited among the DBMS testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.3390/app13042519",
+    "source_sha256": "sha256:7b3fa0806d61464f46fcb7304cb84aa2216ab3dc6d8366d25579f16d0e7414f7",
+    "supporting_excerpts": [
+      "However, the seed scheduling strategy of existing fuzzing techniques does not consider the seeds’ correctness, which is inefficient in finding vulnerabilities in DBMSs.",
+      "First, we propose correctness-guided mutation to utilize the correctness of seeds as feedback to guide fuzzing.",
+      "In our experiment, Squill explored 29% more paths and found 3.4× more bugs than the existing tool."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:37:35Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.3390/app13042519",
+          "source_type": "paper_citation_context",
+          "excerpt": "Sqlancer [2–4] constructs different SQL statements of functionally equivalent through several different patterns and inputs them into the same DBMS.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.3390/app13042519",
+          "source_type": "paper_citation_context",
+          "excerpt": "Black-box fuzzing, or generation-based fuzzing, has been extensively used in finding DBMS bugs, such as SQLsmith [1] and SQLancer [2–4].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_3390_electronics14193910.json b/_data/impact/paper_notes/paper_doi_10_3390_electronics14193910.json
new file mode 100644
index 0000000..22c6e6a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_3390_electronics14193910.json
@@ -0,0 +1,62 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.3390/electronics14193910",
+    "title": "Nabil: A Text-to-SQL Model Based on Brain-Inspired Computing Techniques and Large Language Modeling",
+    "authors": [
+      "Feng Zhou",
+      "Shijing Hu",
+      "Xiaozheng Du",
+      "Nan Li",
+      "Tongming Zhou",
+      "Yanni Zhao",
+      "Sitong Shang",
+      "Xufeng Ling",
+      "Huaizhong Zhu"
+    ],
+    "year": 2025,
+    "venue": "Electronics",
+    "doi": "10.3390/electronics14193910",
+    "url": "https://doi.org/10.3390/electronics14193910"
+  },
+  "summary": {
+    "text": "Nabil is a text-to-SQL model combining spiking neural networks with a large language model. It uses the spatiotemporal encoding of the spiking network to capture semantic features of the natural language input, fuses those with features from the language model, and then applies a champion model to pick the best of several candidate SQL queries. It was evaluated on DuckDB, MySQL and PostgreSQL against benchmarks including BIRD, with normalization and syntax tree abstraction algorithms improving the champion model's discrimination.",
+    "relationship_to_sqlancer": "A text-to-SQL model rather than DBMS testing work; the link is the citation recorded in the citation graph.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.3390/electronics14193910",
+    "source_sha256": "sha256:eac4e4acc65216a69f05c09f422ee0c854361498b84461bf2c557102121b0ae7",
+    "supporting_excerpts": [
+      "To improve the accuracy, generalization, and robustness of text-to-SQL, we propose Nabil (a model for natural language conversion query language based on brain-inspired computing technology and a large language model).",
+      "This model first leverages the spatiotemporal encoding capabilities of spiking neural networks to capture semantic features of natural language, then fuses these features with those generated by a large language model.",
+      "Finally, a champion model is designed to select the optimal query from multiple candidate SQLs."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:30:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "qpg"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.3390/electronics14193910",
+          "source_type": "paper_citation_context",
+          "excerpt": "Ba, J. et al. proposed query plan guidance (QPG) to fully automatically test errors in database systems and applied it to three database systems: SQLite, TiDB, and CockroachDB [20].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_35335_jurnalmantik_vol5_2021_1448_pp1065_1071.json b/_data/impact/paper_notes/paper_doi_10_35335_jurnalmantik_vol5_2021_1448_pp1065_1071.json
new file mode 100644
index 0000000..938f928
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_35335_jurnalmantik_vol5_2021_1448_pp1065_1071.json
@@ -0,0 +1,37 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.35335/jurnalmantik.vol5.2021.1448.pp1065-1071",
+    "title": "Database System Development Life Cycle (DSDLC) on System Libraries for Data Manipulation Language (DML) Using SQL Server 2008",
+    "authors": [
+      "Didik Setiyadi"
+    ],
+    "year": 2021,
+    "venue": "Jurnal Mantik",
+    "doi": "10.35335/jurnalmantik.vol5.2021.1448.pp1065-1071",
+    "url": "https://doi.org/10.35335/jurnalmantik.vol5.2021.1448.pp1065-1071"
+  },
+  "summary": null,
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/1438f9b541946409662ebf91407f849aa42fe856",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_3929_ethz_b_000507577.json b/_data/impact/paper_notes/paper_doi_10_3929_ethz_b_000507577.json
new file mode 100644
index 0000000..2793610
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_3929_ethz_b_000507577.json
@@ -0,0 +1,72 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.3929/ethz-b-000507577",
+    "title": "Verifying Serializability Protocols With Version Order Recovery",
+    "authors": [
+      "Jack Clark"
+    ],
+    "year": 2021,
+    "venue": null,
+    "doi": "10.3929/ethz-b-000507577",
+    "url": "https://doi.org/10.3929/ethz-b-000507577"
+  },
+  "summary": {
+    "text": "A thesis on verifying serializability protocols through version order recovery. Only the sentences citing SQLancer are available here, so this summary is limited to them: the author notes that automated randomized testing has proven effective for database systems, names SQLancer's PQS, TLP and NoREC strategies, and describes part of their own approach as similar to pivoted query synthesis.",
+    "relationship_to_sqlancer": "Names all three SQLancer oracles as effective randomized testing strategies and draws on pivoted query synthesis for its own approach.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://doi.org/10.3929/ethz-b-000507577",
+    "source_sha256": "sha256:65b3404d8bf155a5a5ceb6b32feaa1c464d565d1a5d7795d858833eba3fb259c",
+    "supporting_excerpts": [
+      "Automated randomized testing has proven to be an effective method of testing database systems [3, 43, 44, 45, 49].",
+      "For example, SQLancer support the PQS [45], TLP [44] and NoREC [43] strategies.",
+      "This is similar to the approach used in pivoted query synthesis [45]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:53Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.3929/ethz-b-000507577",
+          "source_type": "paper_citation_context",
+          "excerpt": "Automated randomized testing has proven to be an effective method of testing database systems [3, 43, 44, 45, 49].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.3929/ethz-b-000507577",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, SQLancer support the PQS [45], TLP [44] and NoREC [43] strategies.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.3929/ethz-b-000507577",
+          "source_type": "paper_citation_context",
+          "excerpt": "This is similar to the approach used in pivoted query synthesis [45].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_53799_zb6b1375.json b/_data/impact/paper_notes/paper_doi_10_53799_zb6b1375.json
new file mode 100644
index 0000000..0a1b54a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_53799_zb6b1375.json
@@ -0,0 +1,56 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.53799/zb6b1375",
+    "title": "Optimizing Database System Performance: Design and Query Optimization Strategies",
+    "authors": [
+      "Juanda Hakim Lubis",
+      "I. Jaya",
+      "Fajrul Malik Aminullah Napitupulu",
+      "H. Mawengkang"
+    ],
+    "year": 2026,
+    "venue": "AIUB Journal of Science and Engineering (AJSE)",
+    "doi": "10.53799/zb6b1375",
+    "url": "https://doi.org/10.53799/zb6b1375"
+  },
+  "summary": {
+    "text": "A study of relational database design and query optimisation for practitioners. It evaluates different relational models against varying data volumes, analysing query cost with a cost-based optimizer and access-time measurements. The conclusions are practical guidance for database administrators: separate entities by specialised usage, and take account of record count, attribute size, query type, key usage, order-by clauses and index sequences when structuring queries.",
+    "relationship_to_sqlancer": "A database performance study, not a testing paper; its connection to SQLancer is only the citation recorded in the citation graph.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.53799/zb6b1375",
+    "source_sha256": "sha256:c42c0cd317d876a36f2612d863cefe2d515a7613ddedeb2ed4e41c546a1ed827",
+    "supporting_excerpts": [
+      "This research evaluates different relational database models using varying amounts of data.",
+      "Query costs are analyzed using the Cost-Based Optimizer method and access time measurements."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:27:34Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "cert"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/bdaced332423f73da24f03a45785321328579190",
+          "source_type": "paper",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "Semantic Scholar records this paper as citing the publication that introduced Cardinality Estimation Restriction Testing (CERT).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_doi_10_7717_peerj_cs_1592.json b/_data/impact/paper_notes/paper_doi_10_7717_peerj_cs_1592.json
new file mode 100644
index 0000000..2540e59
--- /dev/null
+++ b/_data/impact/paper_notes/paper_doi_10_7717_peerj_cs_1592.json
@@ -0,0 +1,77 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:doi:10.7717/peerj-cs.1592",
+    "title": "DAFuzz: data-aware fuzzing of in-memory data stores",
+    "authors": [
+      "Yingpei Zeng",
+      "Fengmin Zhu",
+      "Siyi Zhang",
+      "Yu Yang",
+      "Siyu Yi",
+      "Yufan Pan",
+      "Guojie Xie",
+      "Ting Wu"
+    ],
+    "year": 2023,
+    "venue": "PeerJ Computer Science",
+    "doi": "10.7717/peerj-cs.1592",
+    "url": "https://doi.org/10.7717/peerj-cs.1592"
+  },
+  "summary": {
+    "text": "DAFuzz fuzzes in-memory data stores with attention to the data they hold, on the grounds that syntax- and semantics-aware fuzzing still under-tests these systems because some code paths run only when the right data is present. DAFuzz loads different kinds of data into the store before feeding inputs, and generates inputs that are not only syntactically and semantically valid but also use that data correctly. Built on Superion and applied to Redis and Memcached, it covered 13-95% more edges than AFL, Superion, AFL++ and AFLNet and found four new vulnerabilities, all fixed.",
+    "relationship_to_sqlancer": "Data-aware fuzzing for in-memory stores; SQLancer is cited among database testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "abstract",
+    "written_from_url": "https://doi.org/10.7717/peerj-cs.1592",
+    "source_sha256": "sha256:b32432144c94c0c7ca2a13c6fa881ec152d0a8537b505023f697fd83a7f71d06",
+    "supporting_excerpts": [
+      "However, they still cannot fuzz in-memory data stores sufficiently, since some code paths are only executed when the required data are available.",
+      "In this article, we propose a data-aware fuzzing method, DAFuzz, which is designed by considering the data used during fuzzing.",
+      "Experiments show that DAFuzz covers 13~95% more edges than AFL, Superion, AFL++, and AFLNet, and discovers vulnerabilities over 2.7× faster."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://doi.org/10.7717/peerj-cs.1592",
+          "source_type": "paper_citation_context",
+          "excerpt": "Another kind of data-Zeng related program is SQL database, and different special black-box fuzzers ( Seltenreich, Tang & Mullender, 2022 ; Guo, 2017 ; Rigger, 2023 ; Rigger & Su, 2020 ) and grey-box fuzzers ( Zhong et al., 2020 ; Wang et al., 2021 ; Liang, Liu & Hu, 2022 ) have been developed.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.7717/peerj-cs.1592",
+          "source_type": "paper_citation_context",
+          "excerpt": "Recently, fuzzers also try to ensure the statements are semantically valid ( Rigger, 2023 ; Rigger & Su, 2020 ; Zhong et al., 2020 ; Wang et al., 2021 ; Liang, Liu & Hu, 2022 ).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://doi.org/10.7717/peerj-cs.1592",
+          "source_type": "paper_citation_context",
+          "excerpt": "Several recent pieces of research focus on detecting logic bugs but not traditional crashes or assert failures ( Rigger, 2023 ; Rigger & Su, 2020 ; Liang, Liu & Hu, 2022 ).",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_s2_07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85.json b/_data/impact/paper_notes/paper_s2_07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85.json
new file mode 100644
index 0000000..e815416
--- /dev/null
+++ b/_data/impact/paper_notes/paper_s2_07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85.json
@@ -0,0 +1,59 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:s2:07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85",
+    "title": "Leopard: A General Test Suite for Isolation Level Verification",
+    "authors": [
+      "Peiyuan Liu",
+      "Siyang Weng",
+      "Keqiang Li",
+      "Lyu Ni",
+      "Chengcheng Yang",
+      "Rong Zhang",
+      "Weining Qian",
+      "Dian Qiao"
+    ],
+    "year": 2024,
+    "venue": "Conference on Innovative Data Systems Research",
+    "doi": null,
+    "url": "https://www.semanticscholar.org/paper/07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85"
+  },
+  "summary": {
+    "text": "A general test suite for verifying isolation levels. Only the sentence citing SQLancer is available here, so this summary is limited to it: the authors classify existing work into categories and place Pivoted Query Synthesis and TQS in one of them.",
+    "relationship_to_sqlancer": "Cites Pivoted Query Synthesis when classifying the categories of existing database testing work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://www.semanticscholar.org/paper/07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85",
+    "source_sha256": "sha256:546a9e1100c59fccde6075a11110558508ecdb4c69036645c70c5437e8e51368",
+    "supporting_excerpts": [
+      "PQS [15] and TQS [20] belong to this class of work."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:18Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85",
+          "source_type": "paper_citation_context",
+          "excerpt": "PQS [15] and TQS [20] belong to this class of work.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json b/_data/impact/paper_notes/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json
new file mode 100644
index 0000000..eeeefaf
--- /dev/null
+++ b/_data/impact/paper_notes/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json
@@ -0,0 +1,77 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:s2:295c629691d5654def2d9d85f72c756c67c68583",
+    "title": "WingFuzz: Implementing Continuous Fuzzing for DBMSs",
+    "authors": [
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Yiyuan Bai",
+      "Qiang Zhang",
+      "Yu Jiang"
+    ],
+    "year": 2024,
+    "venue": "USENIX Annual Technical Conference",
+    "doi": null,
+    "url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583"
+  },
+  "summary": {
+    "text": "WingFuzz implements continuous fuzzing for DBMSs. Only the sentences citing SQLancer are available here, so this summary is confined to them: the authors describe SQLancer as designing three test oracles for detecting logic errors and generating queries that follow those oracles, and single out its NoREC oracle as requiring a SQL query with WHERE and JOIN clauses.",
+    "relationship_to_sqlancer": "Describes SQLancer's three test oracles, NoREC among them, as part of the existing methods for detecting DBMS issues.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583",
+    "source_sha256": "sha256:ae3fbb2823c58279e95dded4ebc73635da061d63f68d0884650cdeb2e8a51953",
+    "supporting_excerpts": [
+      "SQLancer [40–42] designs three test oracles to detect logic errors and generates queries following the oracles.",
+      "Both the industry and academia have developed many meth-ods for detecting issues in DBMSs [24,32,42,46].",
+      "For example, its NOREC [40] oracle requires constructing a SQL query with WHERE and JOIN clauses."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:35:49Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer [40–42] designs three test oracles to detect logic errors and generates queries following the oracles.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583",
+          "source_type": "paper_citation_context",
+          "excerpt": "Both the industry and academia have developed many meth-ods for detecting issues in DBMSs [24,32,42,46].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, its NOREC [40] oracle requires constructing a SQL query with WHERE and JOIN clauses.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.json b/_data/impact/paper_notes/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.json
new file mode 100644
index 0000000..fc94a21
--- /dev/null
+++ b/_data/impact/paper_notes/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.json
@@ -0,0 +1,98 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:s2:3a7373bd891702ae93d7e058241a7e8be25e89eb",
+    "title": "DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation",
+    "authors": [
+      "Zu-Ming Jiang",
+      "Jia-Ju Bai",
+      "Zhendong Su"
+    ],
+    "year": 2023,
+    "venue": "USENIX Security Symposium",
+    "doi": null,
+    "url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb"
+  },
+  "summary": {
+    "text": "DynSQL performs stateful fuzzing of DBMSs, aiming at complex and valid queries. Only the sentences citing SQLancer are available here, so this summary is limited to them: the authors distinguish their goal from SQLancer's, describing SQLancer as focused on test oracles that need test cases with specific patterns to find logic bugs, whereas DynSQL generates complex valid queries to find common bugs. They also note the difficulty of one grammar template covering all DBMSs, since the common core of SQL features across dialects is small.",
+    "relationship_to_sqlancer": "Contrasts its goal with SQLancer's oracle-driven approach, and describes both PQS and NoREC when positioning the work.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+    "source_sha256": "sha256:1b4f37657e3f5c21ad5c8d6ecef6ed0f139a949a6bb4fca735ed60e35f8b214f",
+    "supporting_excerpts": [
+      "SQLancer [35–37] is also a well-known DBMS testing tool, but it mainly focuses on test oracles, which require test cases with specific patterns to find logic bugs in DBMSs, while DynSQL aims at generating complex and valid queries to detect common bugs, especially memory bugs.",
+      "PQS [37] can generate queries that require the target DBMS to return a result set where a specific row should be included.",
+      "NoREC [35] is a metamorphic testing approach."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+          "source_type": "paper_citation_context",
+          "excerpt": "SQLancer [35–37] is also a well-known DBMS testing tool, but it mainly focuses on test oracles, which require test cases with specific patterns to find logic bugs in DBMSs, while DynSQL aims at generating complex and valid queries to detect common bugs, especially memory bugs.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+          "source_type": "paper_citation_context",
+          "excerpt": "Because many DBMSs use their own SQL dialects and the common core of their SQL features is small [37,39], it is difficult to use one grammar template to test all DBMSs effectively.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+          "source_type": "paper_citation_context",
+          "excerpt": "PQS [37] can generate queries that require the target DBMS to return a result set where a specific row should be included.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+          "source_type": "paper_citation_context",
+          "excerpt": "However, this approach is limited because the common part of supported SQL features in different DBMSs is small [37,39].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+          "source_type": "paper_citation_context",
+          "excerpt": "They either discover specific kinds of bugs [18,23,25,34–37], or detect common bugs using general techniques [19,20,39].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+          "source_type": "paper_citation_context",
+          "excerpt": "NoREC [35] is a metamorphic testing approach.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_s2_5bef8601da9663add6a85713414f8c945cf97355.json b/_data/impact/paper_notes/paper_s2_5bef8601da9663add6a85713414f8c945cf97355.json
new file mode 100644
index 0000000..604104a
--- /dev/null
+++ b/_data/impact/paper_notes/paper_s2_5bef8601da9663add6a85713414f8c945cf97355.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:s2:5bef8601da9663add6a85713414f8c945cf97355",
+    "title": "Language-Based Testing for Knowledge Graphs",
+    "authors": [
+      "Tobias John",
+      "E. Johnsen",
+      "Eduard Kamburjan",
+      "Dominic Steinhöfel"
+    ],
+    "year": 2025,
+    "venue": "Extended Semantic Web Conference",
+    "doi": null,
+    "url": "https://www.semanticscholar.org/paper/5bef8601da9663add6a85713414f8c945cf97355"
+  },
+  "summary": {
+    "text": "A paper on language-based testing for knowledge graphs. Only the sentence citing SQLancer is available here, so the summary is limited to it: the authors note that metamorphic testing has become a popular alternative in recent years, citing SQLancer's work among the examples.",
+    "relationship_to_sqlancer": "Cites SQLancer's oracles as examples of metamorphic testing gaining ground as an alternative.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://www.semanticscholar.org/paper/5bef8601da9663add6a85713414f8c945cf97355",
+    "source_sha256": "sha256:c1f5029713981912082a85867f107a89be6de241c6493fe434b59feb255bddf6",
+    "supporting_excerpts": [
+      "In recent years, metamorphic testing has gained popularity as an alternative [82, 83, 91]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:29:23Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "tlp",
+        "dqp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/5bef8601da9663add6a85713414f8c945cf97355",
+          "source_type": "paper_citation_context",
+          "excerpt": "In recent years, metamorphic testing has gained popularity as an alternative [82, 83, 91].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json b/_data/impact/paper_notes/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json
new file mode 100644
index 0000000..e118a73
--- /dev/null
+++ b/_data/impact/paper_notes/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json
@@ -0,0 +1,135 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:s2:77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+    "title": "Detecting Logical Bugs of DBMS with Coverage-based Guidance",
+    "authors": [
+      "Yu Liang",
+      "Song Liu",
+      "Hong Hu"
+    ],
+    "year": 2022,
+    "venue": "USENIX Security Symposium",
+    "doi": null,
+    "url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42"
+  },
+  "summary": {
+    "text": "SQLRight adds coverage-based guidance to logic bug detection in DBMSs. Only the sentences citing SQLancer are available here, so this summary is limited to them: the authors describe NoREC as shifting all conditions from WHERE clauses into SELECT expressions so that most optimisations on the original query are disabled, and TLP as combining the results of three subqueries and checking equivalence with the original. SQLRight supports four oracles, two of them — NoREC and TLP — ported from SQLancer, plus Index and Rowid introduced in the paper.",
+    "relationship_to_sqlancer": "States that two of its four oracles, NoREC and TLP, are ported from SQLancer, and its artifact reuses that code.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+    "source_sha256": "sha256:8d1fbd88cc1cf8306d8b173f869dc6033ed6775f53ef0b09ab48616684651f0f",
+    "supporting_excerpts": [
+      "For example, for a given query, oracle NoREC shifts all conditions from WHERE clauses to SELECT expressions, which effectively disables most optimizations applied to the original query [43].",
+      "Currently, SQLRight supports four oracles, including NoREC and TLP ported from SQLancer [43, 44], and Index and Rowid we propose in this paper.",
+      "It combines the results from three subqueries and checks the equivalence with the original one [44]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:38:27Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, for a given query, oracle NoREC shifts all conditions from WHERE clauses to SELECT expressions, which effectively disables most optimizations applied to the original query [43].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+          "source_type": "paper_citation_context",
+          "excerpt": "Currently, SQLRight supports four oracles, including NoREC and TLP ported from SQLancer [43, 44], and Index and Rowid we propose in this paper.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+          "source_type": "paper_citation_context",
+          "excerpt": "We implemented four oracles, including two proposed in previous works [43, 44] and two proposed in this paper.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+          "source_type": "paper_citation_context",
+          "excerpt": "It combines the results from three subqueries and checks the equivalence with the original one [44].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "uses_infrastructure",
+      "title": "Uses the SQLancer codebase",
+      "value": "yes",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "sources": [
+        {
+          "source_url": "https://github.com/PSU-Security-Universe/sqlright-artifact",
+          "source_type": "github_repository",
+          "excerpt": "Artifact Evaluation code for USENIX 2022 paper: Detecting Logical Bugs of DBMS with Coverage-based Guidance\n# sqlright-artifact: The code, analysis scripts and results for USENIX 2022 Artifact Evaluation\n\n<a href=\"Paper/paper_no_names.pdf\"><img src=\"Paper/paper_no_names.png\" align=\"right\" width=\"250\"></a>",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/PSU-Security-Universe/sqlright-artifact/blob/main/MySQL/docker/sqlancer/sqlancer/src/sqlancer/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "MySQL/docker/sqlancer/sqlancer/src/sqlancer/Randomly.java is SQLancer's Randomly.java (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "artifact",
+      "title": "Artifact",
+      "value": "https://github.com/PSU-Security-Universe/sqlright-artifact",
+      "method": "artifact_inspection",
+      "techniques": [],
+      "markers": [
+        "sqlancer_source_content_match"
+      ],
+      "sources": [
+        {
+          "source_url": "https://github.com/PSU-Security-Universe/sqlright-artifact",
+          "source_type": "github_repository",
+          "excerpt": "Artifact Evaluation code for USENIX 2022 paper: Detecting Logical Bugs of DBMS with Coverage-based Guidance\n# sqlright-artifact: The code, analysis scripts and results for USENIX 2022 Artifact Evaluation\n\n<a href=\"Paper/paper_no_names.pdf\"><img src=\"Paper/paper_no_names.png\" align=\"right\" width=\"250\"></a>",
+          "excerpt_is_verbatim": true,
+          "note": "Repository names this paper.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://github.com/PSU-Security-Universe/sqlright-artifact/blob/main/MySQL/docker/sqlancer/sqlancer/src/sqlancer/Randomly.java",
+          "source_type": "github_code",
+          "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+          "excerpt_is_verbatim": true,
+          "note": "MySQL/docker/sqlancer/sqlancer/src/sqlancer/Randomly.java is SQLancer's Randomly.java (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.json b/_data/impact/paper_notes/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.json
new file mode 100644
index 0000000..766c422
--- /dev/null
+++ b/_data/impact/paper_notes/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.json
@@ -0,0 +1,54 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:s2:84c637ad3297e0d4e45a4f5245d0472a82a59268",
+    "title": "Demystifying and Checking Silent Semantic Violations in Large Distributed Systems",
+    "authors": [
+      "Chang Lou",
+      "Yuzhuo Jing",
+      "Peng Huang"
+    ],
+    "year": 2022,
+    "venue": "USENIX Symposium on Operating Systems Design and Implementation",
+    "doi": null,
+    "url": "https://www.semanticscholar.org/paper/84c637ad3297e0d4e45a4f5245d0472a82a59268"
+  },
+  "summary": {
+    "text": "A paper on silent semantic violations in large distributed systems. Only the sentence citing SQLancer is available here, so this summary is limited to it: the authors survey approaches proposed for detecting semantic bugs in file systems and DBMSs, naming cross-checking of multiple file system implementations, fuzzing, and testing using pivoted queries.",
+    "relationship_to_sqlancer": "Cites Pivoted Query Synthesis among the existing approaches to detecting silent semantic bugs.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://www.semanticscholar.org/paper/84c637ad3297e0d4e45a4f5245d0472a82a59268",
+    "source_sha256": "sha256:169953b50a74da6832ed99491335135dfaa8d0099dda2bb610655b583d543dee",
+    "supporting_excerpts": [
+      "Several solutions are proposed to detect semantic bugs in file systems and DBMS, including cross-checking multiple file system implementations [50], fuzzing [39], and testing using pivoted query [54]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:38:27Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/84c637ad3297e0d4e45a4f5245d0472a82a59268",
+          "source_type": "paper_citation_context",
+          "excerpt": "Several solutions are proposed to detect semantic bugs in file systems and DBMS, including cross-checking multiple file system implementations [50], fuzzing [39], and testing using pivoted query [54].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_s2_b6c64dbe0130ef1bf6af12266b958a5d5396101f.json b/_data/impact/paper_notes/paper_s2_b6c64dbe0130ef1bf6af12266b958a5d5396101f.json
new file mode 100644
index 0000000..bda808c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_s2_b6c64dbe0130ef1bf6af12266b958a5d5396101f.json
@@ -0,0 +1,53 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:s2:b6c64dbe0130ef1bf6af12266b958a5d5396101f",
+    "title": "Differential Monitoring - Technical Report ⋆",
+    "authors": [
+      "Fabian Muehlboeck"
+    ],
+    "year": 2021,
+    "venue": null,
+    "doi": null,
+    "url": "https://www.semanticscholar.org/paper/b6c64dbe0130ef1bf6af12266b958a5d5396101f"
+  },
+  "summary": {
+    "text": "The technical report accompanying the differential monitoring paper. Only the sentence citing SQLancer is available here, so this summary is limited to it: the authors note that the technique has been applied fruitfully to finding bugs in JavaScript debuggers, C compilers and SQL databases.",
+    "relationship_to_sqlancer": "Cites SQLancer's work among the successful applications of differential techniques to SQL databases.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://www.semanticscholar.org/paper/b6c64dbe0130ef1bf6af12266b958a5d5396101f",
+    "source_sha256": "sha256:ba89b2245189a9a4329b17f62018c69a864e040924b56ca32093bcc54158aa49",
+    "supporting_excerpts": [
+      "This technique has been fruitfully applied to finding bugs in Javascript debuggers [31], C compilers [41], and SQL databases [39,37,36]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:12Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/b6c64dbe0130ef1bf6af12266b958a5d5396101f",
+          "source_type": "paper_citation_context",
+          "excerpt": "This technique has been fruitfully applied to finding bugs in Javascript debuggers [31], C compilers [41], and SQL databases [39,37,36].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json b/_data/impact/paper_notes/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json
new file mode 100644
index 0000000..89c91ca
--- /dev/null
+++ b/_data/impact/paper_notes/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json
@@ -0,0 +1,142 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:s2:c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+    "title": "Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis",
+    "authors": [
+      "Zongyin Hao",
+      "Quanfeng Huang",
+      "Chengpeng Wang",
+      "Jianfeng Wang",
+      "Yushan Zhang",
+      "Rongxin Wu",
+      "Charles Zhang"
+    ],
+    "year": 2023,
+    "venue": "USENIX Annual Technical Conference",
+    "doi": null,
+    "url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b"
+  },
+  "summary": {
+    "text": "Pinolo detects logical bugs in DBMSs using approximate query synthesis. Only the sentences citing SQLancer are available here, so this summary is limited to them: the authors survey the test oracles for logical bug detection, describing how NoREC rewrites an optimised query into a non-optimised one and how TLP decomposes a query into TRUE, FALSE and NULL partitions before recombining them, and state that Pinolo covers more SQL features such as set operators, arithmetic expressions and sub-queries than these techniques.",
+    "relationship_to_sqlancer": "Compares directly against PQS, NoREC and TLP as the three state-of-the-art logical bug detection techniques, and calls metamorphic testing the state of the art in DBMS logic bug detection.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+    "source_sha256": "sha256:742c8aa3fd6995bff4d18be998ca689b1c065ae0455a7927af450b7f9209a6b2",
+    "supporting_excerpts": [
+      "Notably, the metamorphic testing based approach has been recognized to be state-of-the-art in DBMS testing for logical bug detection [35, 37].",
+      "We compared P INOLO with the three state-of-the-art logical bug detection techniques, namely PQS [36], N O REC [34], and TLP [35], respectively, which correspond to three kinds of test oracles.",
+      "Compared with the existing techniques [34–36], P INOLO considers more SQL features, such as set operators, arithmetic expressions, sub-queries, etc."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:36:46Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+          "source_type": "paper_citation_context",
+          "excerpt": "N O REC [34] transforms an optimized version of a query into a non-optimized one by the customized rule, e.g., changing “ SELECT * FROM t WHERE p” into “ SELECT (p IS TRUE ) FROM t.” Compared with the aforementioned two categories of approaches, metamorphic testing based approaches are much more…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+          "source_type": "paper_citation_context",
+          "excerpt": "…e.g., changing “ SELECT * FROM t WHERE p” into “ SELECT (p IS TRUE ) FROM t.” Compared with the aforementioned two categories of approaches, metamorphic testing based approaches are much more lightweight to implement and have been proven to be more effective in detecting logical bugs [34, 35].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+          "source_type": "paper_citation_context",
+          "excerpt": "Recent years have witnessed tremendous efforts in resolving the test oracle for logical bug detection in the DBMSs. Notably, the metamorphic testing based approach has been recognized to be state-of-the-art in DBMS testing for logical bug detection [35, 37].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compared P INOLO with the three state-of-the-art logical bug detection techniques, namely PQS [36], N O REC [34], and TLP [35], respectively, which correspond to three kinds of test oracles.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+          "source_type": "paper_citation_context",
+          "excerpt": "Compared with the existing techniques [34–36], P INOLO considers more SQL features, such as set operators, arithmetic expressions, sub-queries, etc.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+          "source_type": "paper_citation_context",
+          "excerpt": "For example, TLP [35] decomposes a query q into three partitioning sub-queries, each of which computes the result sets for a boolean predicate to be evaluated as TRUE , FALSE , and NULL , respectively, and then constructs an equivalent query q ′ by performing the union operation on these three…",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "manual_curation",
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compared P INOLO with the three state-of-the-art logical bug detection techniques, namely PQS [36], N O REC [34], and TLP [35], respectively, which correspond to three kinds of test oracles.",
+          "excerpt_is_verbatim": true,
+          "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+          "retrieved_at": "2026-09-06T07:08:48Z"
+        }
+      ]
+    },
+    {
+      "relationship": "describes_as_state_of_the_art",
+      "title": "Calls SQLancer state of the art",
+      "value": "yes",
+      "method": "deterministic",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+          "source_type": "paper_citation_context",
+          "excerpt": "We compared P INOLO with the three state-of-the-art logical bug detection techniques, namely PQS [36], N O REC [34], and TLP [35], respectively, which correspond to three kinds of test oracles.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper describing Pivoted Query Synthesis (PQS) as state of the art.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.json b/_data/impact/paper_notes/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.json
new file mode 100644
index 0000000..21b018c
--- /dev/null
+++ b/_data/impact/paper_notes/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.json
@@ -0,0 +1,57 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:s2:d88db0b52f31ed5444602a67e9a275faf57e15bf",
+    "title": "TCP-Fuzz: Detecting Memory and Semantic Bugs in TCP Stacks with Fuzzing",
+    "authors": [
+      "Yong-Hao Zou",
+      "Jia-Ju Bai",
+      "Jielong Zhou",
+      "Jianfeng Tan",
+      "Chenggang Qin",
+      "Shih-Min Hu"
+    ],
+    "year": 2021,
+    "venue": "USENIX Annual Technical Conference",
+    "doi": null,
+    "url": "https://www.semanticscholar.org/paper/d88db0b52f31ed5444602a67e9a275faf57e15bf"
+  },
+  "summary": {
+    "text": "TCP-Fuzz detects memory and semantic bugs in TCP stacks by fuzzing. Only the sentence citing SQLancer is available here, so this summary is limited to it: the authors describe following the same practice as SQLancer and libFuzzer when handling inconsistencies identified as semantic bugs, manually fixing them using developers' patches or their own so that related inconsistencies are reduced.",
+    "relationship_to_sqlancer": "Adopts SQLancer's practice for handling semantic-bug inconsistencies during a fuzzing campaign.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://www.semanticscholar.org/paper/d88db0b52f31ed5444602a67e9a275faf57e15bf",
+    "source_sha256": "sha256:c2ce4ba41adc3dfa8b35ba167827789543b42d5cb591f8535731df30f0aa8a5a",
+    "supporting_excerpts": [
+      "Similar to SQLancer [51] and libFuzzer [32], for in-consistencies that we identify as semantic bugs, we manually fix them using the developers’ patches or by ourselves, to reduce related inconsistencies."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:39:53Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/d88db0b52f31ed5444602a67e9a275faf57e15bf",
+          "source_type": "paper_citation_context",
+          "excerpt": "Similar to SQLancer [51] and libFuzzer [32], for in-consistencies that we identify as semantic bugs, we manually fix them using the developers’ patches or by ourselves, to reduce related inconsistencies.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/paper_notes/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json b/_data/impact/paper_notes/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json
new file mode 100644
index 0000000..0711db1
--- /dev/null
+++ b/_data/impact/paper_notes/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json
@@ -0,0 +1,118 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "paper": {
+    "id": "paper:s2:d933042eb3c2a8f2e208515490d6b6a400e00fec",
+    "title": "Towards Generic Database Management System Fuzzing",
+    "authors": [
+      "Yupeng Yang",
+      "Yongheng Chen",
+      "Rui Zhong",
+      "Jizhou Chen",
+      "Wenke Lee"
+    ],
+    "year": 2024,
+    "venue": "USENIX Security Symposium",
+    "doi": null,
+    "url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec"
+  },
+  "summary": {
+    "text": "BuzzBee is a fuzzer aimed at database management systems generically rather than at relational systems alone. Only the sentences citing SQLancer are available here, so this summary is limited to them: the authors situate their work against a substantial body of relational DBMS fuzzing research, naming SQLancer, SQLsmith and Squirrel as the tools that have emerged to test relational systems, and evaluate BuzzBee against Squirrel and SQLancer's Pivoted Query Synthesis for the relational case.",
+    "relationship_to_sqlancer": "Compares directly against SQLancer, using its Pivoted Query Synthesis as one of two specialised relational DBMS baselines.",
+    "is_model_generated": true,
+    "grounded_in_source": true,
+    "written_from": "citation contexts",
+    "written_from_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+    "source_sha256": "sha256:d015eabc4004b8c858ce5a1bd9e9e414c89fafdd5eecfc4cac2b2fcb43a74007",
+    "supporting_excerpts": [
+      "For relational DBMSs, we compare B UZZ B EE with S QUIRREL [56] and SQL ANCER (PQS [43]), two DBMS fuzzers specialized in SQL DBMS fuzzing.",
+      "Tools like SQLancer [42], SQLsmith [44], and Squirrel [56] have emerged to test relational DBMSs.",
+      "G LOT [9], Grammarinator [22], and SQL-specialized S QUIR - REL [56] and SQL ANCER [43]."
+    ],
+    "model": "claude-opus-5",
+    "generated_by": "claude_code_session",
+    "classifier_version": null,
+    "generated_at": "2026-09-07T02:34:56Z"
+  },
+  "evidence": [
+    {
+      "relationship": "references",
+      "title": "Cites SQLancer",
+      "value": "yes",
+      "method": "citation_graph",
+      "techniques": [
+        "pqs",
+        "norec"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+          "source_type": "paper_citation_context",
+          "excerpt": "For relational DBMSs, we compare B UZZ B EE with S QUIRREL [56] and SQL ANCER (PQS [43]), two DBMS fuzzers specialized in SQL DBMS fuzzing.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+          "source_type": "paper_citation_context",
+          "excerpt": "Fuzzing frameworks and research related to relational DBMSs [27, 28, 43, 44, 49, 56] have been developed and advanced extensively over the years, contributing to more secure and trustworthy systems in the relational DBMS venue.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+          "source_type": "paper_citation_context",
+          "excerpt": "Tools like SQLancer [42], SQLsmith [44], and Squirrel [56] have emerged to test relational DBMSs. SQLsmith generates random SQL queries.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+          "source_type": "paper_citation_context",
+          "excerpt": "Current research has made significant advancements in relational DBMS testing [27, 28, 43, 44, 49, 56].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+          "source_type": "paper_citation_context",
+          "excerpt": "Fuzzing DBMSs has been an active research area in recent years [15, 25–27, 43, 44, 49, 56].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        },
+        {
+          "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+          "source_type": "paper_citation_context",
+          "excerpt": "G LOT [9], Grammarinator [22], and SQL-specialized S QUIR - REL [56] and SQL ANCER [43].",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+          "retrieved_at": "2026-09-06T15:09:10Z"
+        }
+      ]
+    },
+    {
+      "relationship": "compares_with",
+      "title": "Compares against SQLancer",
+      "value": "yes",
+      "method": "manual_curation",
+      "techniques": [
+        "pqs"
+      ],
+      "sources": [
+        {
+          "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+          "source_type": "paper_citation_context",
+          "excerpt": "For relational DBMSs, we compare B UZZ B EE with S QUIRREL [56] and SQL ANCER (PQS [43]), two DBMS fuzzers specialized in SQL DBMS fuzzing.",
+          "excerpt_is_verbatim": true,
+          "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+          "retrieved_at": "2026-09-06T07:08:48Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/papers.json b/_data/impact/papers.json
new file mode 100644
index 0000000..34e79e3
--- /dev/null
+++ b/_data/impact/papers.json
@@ -0,0 +1,25820 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "papers": [
+    {
+      "id": "paper:doi:10.1145/3799227",
+      "title": "A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Evaluation",
+      "authors": [
+        "Xiyue Gao",
+        "Zhuang Liu",
+        "Jiangtao Cui",
+        "Hui Li",
+        "Hui Zhang",
+        "Kewei Wei",
+        "Kankan Zhao"
+      ],
+      "year": 2026,
+      "venue": "ACM Computing Surveys",
+      "doi": "10.1145/3799227",
+      "arxiv_id": "2311.06728",
+      "s2_paper_id": "de2dbf2b041d2deaef7756f69ef5b3bdc5e599cc",
+      "url": "https://doi.org/10.1145/3799227",
+      "open_access_pdf": "https://doi.org/10.1145/3799227",
+      "abstract": "Database Management System (DBMS) fuzzing is an automated testing technique aimed at detecting errors and vulnerabilities in DBMSs by generating, mutating, and executing test cases. It not only reduces the time and cost of manual testing but also enhances detection coverage, providing valuable assistance in developing commercial DBMSs. Existing fuzzing surveys mainly focus on general-purpose software. However, DBMSs are different from them in terms of internal structure, input/output, and test objectives, requiring specialized fuzzing strategies. Therefore, this article focuses on DBMS fuzzing and provides a comprehensive review and comparison of the methods in this field. We first introduce the fundamental concepts. Then, we systematically define a general fuzzing procedure and decompose and categorize existing methods. Furthermore, we classify existing methods from the testing objective perspective, covering various components in DBMSs. For representative works, more detailed descriptions are provided to analyze their strengths and limitations. Additionally, we review recent fuzzing methods for non-relational DBMSs as well as emerging techniques. To objectively evaluate the performance of each method, we present an open-source DBMS fuzzing toolkit, OpenDBFuzz. Based on this toolkit, we conduct a detailed experimental comparative analysis of existing methods and finally discuss future research directions.",
+      "cites_seed_techniques": [
+        "cert",
+        "dqp",
+        "norec",
+        "pqs",
+        "qpg",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3799227",
+              "source_type": "paper_citation_context",
+              "excerpt": "While traditional LEGO[67] SQLRight[69] QPG[13] AMOEBA[73] GARan[16] DynSQL[54] Squill[106] approaches utilize generic SAT solvers for this task solver [40, 60], recent methods such as TQS [102] also use a logical external solver as a substitute for the SAT solver to derive the ground truth results based on the original test cases through logical operations or theoretical deduction [29, 37, 102].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ec834f49daad0bd6cb812b0e6f6561e4549e6c7232e1a6175522d5d22374ec4e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3799227",
+              "source_type": "paper_citation_context",
+              "excerpt": "Without this limitation, numerous complex SQL statements would be generated, which may expand the search space but reduce the overall efficiency of bug detection [28, 32, 91–93, However, static configurations require manual adjustments based on specific requirements.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:694467a4db2eda765ed0d1c23338673607aadc287adda4b281a926f08d9f760c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3799227",
+              "source_type": "paper_citation_context",
+              "excerpt": "On the other hand, fuzzers based on random databases [12, 14, 15, 28, 32, 34, 55, 81, 88, 92, 93, 97, 102, 106, 107, 117] create random database instances from scratch.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4b977b15597c16a2c8998dee2e1fd89faa45580a6c6f6f953648e5bf3b949409",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3799227",
+              "source_type": "paper_citation_context",
+              "excerpt": "Specifically, we compare them in terms of database instance, generation type, strategy, and feedback.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e0d1ce7ac76b6b9a2321c71851557116e96d07089020c0021f431219dd53bb8e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3799227",
+              "source_type": "paper_citation_context",
+              "excerpt": "Database Crashes Crash Bugs Squill[106]Griffin[34] LEGO[67]DynSQL[54] SOFT[35] RAGS[96] Differential Different DBMSs Logic Bugs SQLsmith[4]Go-Randgen[88] GARan[16]DT2[28] Radar[98] Different Database Instances DDLCheck[99] APOLLO[57] Different Versions of the Same DBMS Performance Bugs AMOEBA[73] Metamorphic Statement Rewriting CERT[14]MutaSQL[26] Logic Bugs Eqsql[114]NoREC[91] SQLRight[69]PINOLO[45] EET[56]CODDTest[113] PUPPY[107] Execution Path Manipulation Performance Bugs Mozi[68] Performance & Logic Bugs Kangaroo[64] Logic & Crash Bugs DQP[15] Logic Bugs DQE[97] Statement Type Transformation TLP[92] Query Partition Troc[32] Transaction Splitting TxCheck[55] ADUSA[12] Constraint-solving Forward Solving(SAT Solver) Artemis[81] PQS[93] Backward Solving TQS[102] Forward Solving(Logical Solver) WriteCheck[29]Fucci[37] (a) Constraint Rewriting • Statement Rewriting : Statement Rewriting refers to the process of modifying a statement according to specified rules, so that the final semantics remain unchanged or change as expected.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:561938afb80b5446dd5b0cae95fea020b5cfadb5990436ff6c74af662428f443",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3799227",
+              "source_type": "paper_citation_context",
+              "excerpt": "…Performance Bugs Mozi[68] Performance & Logic Bugs Kangaroo[64] Logic & Crash Bugs DQP[15] Logic Bugs DQE[97] Statement Type Transformation TLP[92] Query Partition Troc[32] Transaction Splitting TxCheck[55] ADUSA[12] Constraint-solving Forward Solving(SAT Solver) Artemis[81] PQS[93] Backward…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:03fabca1772e5047bd7a8c215a797e1ea711edbd8db3cbabf3847cd0453d80d5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2311.06728",
+              "source_type": "paper_citation_context",
+              "excerpt": "As can be seen in Figure 20, QPG detected bugs more efficiently than TLP within 240 minutes, and the number of bugs gradually equalized in the later period.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:01122524597b63bac88041ee5d1a44a03f6bfc53b93743b2a65e97a53e6623df",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2311.06728",
+              "source_type": "paper_citation_context",
+              "excerpt": "On the other hand, fuzzers based on random databases [16, 31– 33, 62, 71, 81, 85, 86, 90, 94, 101, 109] create random database instances from scratch.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1cd121db83ca4c60b2509eda0b67ceddfb9064fe1b16636dadfaf5c8f65af45a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2311.06728",
+              "source_type": "paper_citation_context",
+              "excerpt": "QPG implements a generation-based generator based on PQS, TLP, and NoREC.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7a57dc26f34eed49aa6b067d28720f81b3b25a2930d59de970cb34e284d2fd97",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2311.06728",
+              "source_type": "paper_citation_context",
+              "excerpt": "Supported DBMS Link SQLancer[86] (PQS[86], NoREC[84], TLP[85], QPG[18]) SQLite, MySQL, TiDB MariaDB, CockroachDB, OceanBase https://github.com/sqlancer/sqlancer DQE[90] SQLite, MySQL, MariaDB TiDB, CockroachDB https://github.com/tcse-iscas/dqetool SQLRight[60] SQLite, PostgreSQL, MySQL https: //github.com/psu-security-universe/sqlright SQLsmith[5] SQLite, MonetDB, PostgreSQL https://github.com/anse1/sqlsmith Go-Randgen[81] MySQL, TiDB https://github.com/pingcap/go-randgen Squirrel[109] SQLite, PostgreSQL, MySQL, MariaDB https://github.com/s3team/Squirrel DT2[31] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc Troc [32] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc APOLLO[51] SQLite, PostgreSQL https://github.com/sslab-gatech/apollo AMOEBA[62] PostgreSQL, CockroachDB https://bit.ly/3I995jL confirmed bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2aca4ae32bb822edd0f5bb4159746455367c431be28e1c6f8963825b9def77c1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2311.06728",
+              "source_type": "paper_citation_context",
+              "excerpt": "Database Crashes Crash Bugs Squill[101] Griffin[33] LEGO[59] DynSQL[49] RAGS[89] Differential Different Databases Logic Bugs SQLsmith[5] Go-Randgen[81] GARan[19] DT2[31] APOLLO[51] Different Versions of A Same Database Performance Bugs AMOEBA[62] Metamorphic Statement Rewriting MutaSQL[28] Logic Bugs Eqsql[107] NoREC[84] SQLRight[60] DQE[90] Statement Type Transformation TLP[85] Query Partition Troc[32] Transaction Splitting ADUSA[16] Constraint-solving Forward Solving(SAT Solver) Artemis[71] PQS[86] Backward Solving TQS[94] Forward Solving(Logical Solver) Some generators of black-box fuzzers [18] obtain feedback such as the validity of query plans or test cases from the query or query plan interface provided by the DBMS, guiding the subsequent generation process.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:cb10ac999d3806cea10e5d3fda4d508a6e50ab4b9913d401da653978a550696b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2311.06728",
+              "source_type": "paper_citation_context",
+              "excerpt": "…Bugs AMOEBA[62] Metamorphic Statement Rewriting MutaSQL[28] Logic Bugs Eqsql[107] NoREC[84] SQLRight[60] DQE[90] Statement Type Transformation TLP[85] Query Partition Troc[32] Transaction Splitting ADUSA[16] Constraint-solving Forward Solving(SAT Solver) Artemis[71] PQS[86] Backward Solving…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:eb4e6f1686f06f77705ec8cdc24575e104aa6fb7d658f7318f14d08b3984162f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3799227.json",
+              "source_type": "paper",
+              "excerpt": "Within 240 minutes, QPG detects bugs more efficiently than TLP, and the number of detected bugs gradually converges in the later stage, demonstrating that query plan feedback effectively improves the efficiency of logic bug detection.",
+              "excerpt_is_verbatim": true,
+              "note": "M71 in the paper's extracted text, 6.5 Logic Bugs Detection Comparison, page 29.",
+              "content_sha256": "sha256:b0a2aad471b834840faa8cebaa65373f2d53002965b04c6d6e61fb606e668464",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/de2dbf2b041d2deaef7756f69ef5b3bdc5e599cc",
+        "source_type": "paper",
+        "content_sha256": "sha256:e3b1af3c18aae85fea23f18b70c7c4a3222c8d29f49a962f59dd753f94eba965"
+      },
+      "also_indexed_as": [
+        "paper:arxiv:2311.06728"
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3834861",
+      "title": "A Formal Framework for Typing and Cast Semantics in SQL Engines",
+      "authors": [
+        "Wenjia Ye",
+        "Matías Toro",
+        "Claudio Gutierrez",
+        "Bruno C. d. S. Oliveira",
+        "Éric Tanter"
+      ],
+      "year": 2026,
+      "venue": "ACM Transactions on Programming Languages and Systems",
+      "doi": "10.1145/3834861",
+      "arxiv_id": null,
+      "s2_paper_id": "1c4469f58aec757c9ba0c3e7e9e6b15e42f86951",
+      "url": "https://doi.org/10.1145/3834861",
+      "open_access_pdf": null,
+      "abstract": "Practical SQL engines differ in subtle ways in their handling of typing constraints and implicit type casts. These issues, usually not considered in formal accounts of SQL, directly affect the portability of queries between engines. To address this problem, we present a formal typing semantics for SQL, named TRAF, that explicitly captures both static and dynamic type behavior. The system TRAF is expressed in terms of abstract operators that provide the necessary leeway to systematically model the type and cast behavior of different SQL engines (PostgreSQL, MS SQL Server, MySQL, SQLite, and Oracle). We show that this formalism provides formal guarantees for the handling of types. We identify practical conditions under which engine instantiations satisfy type safety and type soundness. In this regard, TRAF can serve as an explicit and testable account of typing in existing engines, potentially guide their evolution, and provide a formal basis to study type-aware query optimizations and design provably-correct query translators. We also test the adequacy of the formalism by implementing TRAF in Python for these five engines and testing it with synthetic queries, SQLancer++-generated queries, and adapted Spider and Calcite benchmark queries.",
+      "cites_seed_techniques": [
+        "sqlancer_pp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "sqlancer_pp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3834861",
+              "source_type": "paper_citation_context",
+              "excerpt": "…2013], we developed multiple database interpreters and tested them with synthetic queries generated by our grammar, SQLancer++-generated queries [Zhong and Rigger 2026], and adapted Spider [Yu et al. 2018] and Calcite [Begoli et al. 2018] benchmark queries, comparing their results with those…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing sqlancer_pp, as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:de381e69f0b145c90f13dbf90b040f4be31b9387f9be43955013ce7787cf247c",
+              "retrieved_at": "2026-09-10T14:52:29Z",
+              "first_seen": "2026-09-10T14:52:29Z",
+              "last_verified": "2026-09-10T14:52:29Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3834861",
+              "source_type": "paper_citation_context",
+              "excerpt": "Third, we use SQLancer++ [Zhong and Rigger 2026] as a SQL fuzzer with an adaptive statement generator that constructs database states while maintaining an internal schema model, and then generates random queries whose feature choices are refined through feedback from the target DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing sqlancer_pp, as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d0fe792455c4170d61cc4a909c4385e24ae600eb4529850855ef5c0a851ec88c",
+              "retrieved_at": "2026-09-10T14:52:29Z",
+              "first_seen": "2026-09-10T14:52:29Z",
+              "last_verified": "2026-09-10T14:52:29Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3834861",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer++ [Zhong and Rigger 2026] constructs database states and adapts each generated query to feedback from the target DBMS, in order to exercise code paths that our fixed-schema generator (Section 6.2) cannot reach.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing sqlancer_pp, as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0afefca2f9e7b761a7eb38aff94c9bbd80299e529d16a7b4d56ae46b66a785ec",
+              "retrieved_at": "2026-09-10T14:52:29Z",
+              "first_seen": "2026-09-10T14:52:29Z",
+              "last_verified": "2026-09-10T14:52:29Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-10T14:52:29Z",
+        "last_verified": "2026-09-10T14:52:29Z",
+        "source_url": "https://www.semanticscholar.org/paper/1c4469f58aec757c9ba0c3e7e9e6b15e42f86951",
+        "source_type": "paper",
+        "content_sha256": "sha256:00265e0053aa14cc06cf812271eb50d28b815d46b9e2989d1d654913a4907eab"
+      }
+    },
+    {
+      "id": "paper:doi:10.4018/979-8-2600-0747-1.ch001",
+      "title": "A Framework for Systematic Analysis and Automated Detection of Dark Patterns on E-Commerce Websites",
+      "authors": [
+        "L. Mary Shamala",
+        "Karthika Veeramani",
+        "K. Ayshwarya"
+      ],
+      "year": 2026,
+      "venue": null,
+      "doi": "10.4018/979-8-2600-0747-1.ch001",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.4018/979-8-2600-0747-1.ch001",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/W7140414167",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-10T14:40:47Z",
+              "first_seen": "2026-09-10T14:40:47Z",
+              "last_verified": "2026-09-10T14:40:47Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-10T14:40:47Z",
+        "last_verified": "2026-09-10T14:40:47Z",
+        "source_url": "https://openalex.org/W7140414167",
+        "source_type": "paper",
+        "content_sha256": "sha256:3c5bf7a141f12bb0c0696f6f83ca4c01d0fbba03803594bafecc8eef4b89202e"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icde65706.2026.00224",
+      "title": "A Set-Theoretic Approach to Detecting Logic Bugs in DBMS Inner Join Optimizations",
+      "authors": [
+        "Ce Lyu",
+        "Changzheng Wei",
+        "Yanhao Wang",
+        "Jie Liang",
+        "Li Lin",
+        "Hanghang Wu",
+        "Minghao Zhao",
+        "Ying Yan",
+        "Aoying Zhou"
+      ],
+      "year": 2026,
+      "venue": "IEEE International Conference on Data Engineering",
+      "doi": "10.1109/icde65706.2026.00224",
+      "arxiv_id": "2606.23294",
+      "s2_paper_id": "f959eadc36d89bdfc3add245defa2b3b50c40954",
+      "url": "https://doi.org/10.1109/icde65706.2026.00224",
+      "open_access_pdf": "https://doi.org/10.48550/arxiv.2606.23294",
+      "abstract": "The query optimizer is a fundamental component of database management systems that determines the most efficient execution strategy for a given query by evaluating alternative query plans. Among its tasks, join optimization plays a central role, as the order of joins in multi-table queries can significantly affect execution performance. However, due to the inherent complexity of join optimization, logical bugs are inevitable and often difficult to detect. While existing fuzzing tools have shown notable success in uncovering crash- and performance-related errors, effectively identifying logical bugs-cases in which the system produces incorrect query results-remains largely unresolved. In this paper, we propose a metamorphic testing approach to detect DBMS bugs related to INNER JOIN optimization through the lens of set theory. For each testing case, equivalent queries are generated based on a basic set operation-intersection-and three semantics-preserving transformation rules, i.e., symmetric join transformation, asymmetric difference transformation, and symmetric difference transformation, are introduced. These rules rewrite a simple NATURAL/INNER JOIN query into a more complex, yet semantically equivalent, form. We implement this design in JoinEquiv, which serves as a testing oracle to systematically uncover logical inconsistencies in DBMS query processing by comparing the results of original and transformed queries. Using JoinEquiv, we uncovered 29 previously unknown issues in mainstream DBMSs (MySQL, TiDB, DuckDB, and Percona), and 27 of them were officially confirmed. JoinEquiv reveals deep logical flaws in DBMS optimizers and executors, underscoring its value in enhancing DBMS robustness.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "cert",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+              "source_type": "paper_citation_context",
+              "excerpt": "In the SQL standard, equality comparison predicates involving NULL are evaluated to UNKNOWN and therefore excluded from INNER JOIN results, whereas the set operator treats NULL as equal at the set level and preserves them.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:751ef46141e9036890ee704152cacb9653a7ed85568e0b749a2704a4a6560f6b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+              "source_type": "paper_citation_context",
+              "excerpt": "Together, these findings demonstrate that JoinEquiv can uncover a broad, heterogeneous spectrum of logic bugs originating from multiple layers of the system, including the optimizer, executor, and set-operator implementations, thereby effectively answering RQ2 .",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:77cbfe5c1ea92f8975cce3a845c87aef33b58941ae2827f50b9a586593f7e258",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+              "source_type": "paper_citation_context",
+              "excerpt": "CERT [37] aims to identify performance issues caused by unexpected cardinality estimations, that is, the cases where the estimated number of result tuples significantly deviates from the actual number.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b4e2f1d3a615b50af56bcba7b36cc105710303b241effebb0b4879f06e63e902",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+              "source_type": "paper_citation_context",
+              "excerpt": "Recently, differential query planning (DQP) [25] adopts a similar hint-based strategy but focuses on detecting inconsistencies between different physical plans for the same query.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:650926d114d3bebd60b23a8ad0511ec3dc3b8bebbead187dd09ab2c88f6d90cc",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+              "source_type": "paper_citation_context",
+              "excerpt": "Another approach is pivoted query synthesis (PQS) [19], which validates query results by constructing auxiliary queries centered on a specific pivot row.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:67004126aec2c1a0f7b45ff0c470f44b17593c3f2a0f890abbe617e76addd77a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00224",
+              "source_type": "paper_citation_context",
+              "excerpt": "Representative frameworks include ternary logic partitioning (TLP) [20] and non-optimizing reference engine construction (NoREC) [22].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c8cee60fa4d2bcaf98cd60508007249df54a3f19d7ea101a4917e316868ddfcb",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00224.json",
+              "source_type": "paper",
+              "excerpt": "The test case generation builds upon the SQLancer framework, which provides a grammar-aware SQL query generation engine.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, E Database and Query Generation, page 6.",
+              "content_sha256": "sha256:0ae308333d82fd05acaf64b14b544b97a2212e11fde17cf2eb7c9dbb072d2e7c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00224.json",
+              "source_type": "paper",
+              "excerpt": "We extend SQLancer to satisfy an additional precondition that all columns are NOT NULL.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, E Database and Query Generation, page 6.",
+              "content_sha256": "sha256:368ca8e8eb850b853983b021796114b1654d7b09a073bb04d0aa448e5e463f03",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00224.json",
+              "source_type": "paper",
+              "excerpt": "Such a table permutation, combined with SQLancer’s randomized population of ONpredicates,WHEREclauses, and SELECT list expressions, ensures that the framework explores a diverse range of symmetric and asymmetric execution plans.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, E Database and Query Generation, page 6.",
+              "content_sha256": "sha256:e0eb69e5b7ed9dddd18a3177dd2011fc91ce25e024f8c4a1a72d2b03a8e90ca6",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00224.json",
+              "source_type": "paper",
+              "excerpt": "Within each thread, for every INNER JOINquery, JoinEquiv, TLP, and DQP are applied sequentially under the same system TABLE IISUMMARY OFLOGICBUGREPORTS ANDVERIFICATIONSTATUS INDIFFERENTRDBMSS RDBMS Reported Verified Fixed Intended Component Severity Identifier MySQL 10 8 1 2 Optimizer (8) Critical (8)Bug#118544, Bug#118684, Bug#118710, Bug#118857, Bug#118858, Bug#118949, Bug#119032, Bug#119059 TiDB 13 13 4 0Planner (11) Execution (2)Critical (3)#62380, #62444, #62456, #62459, #62460, #62644, #...",
+              "excerpt_is_verbatim": true,
+              "note": "M13 in the paper's extracted text, A Evaluation Setup, page 7.",
+              "content_sha256": "sha256:35c8cb28efc76419b5d31061a39d4767750843c285e1c8de4e191dded59aa662",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00224.json",
+              "source_type": "paper",
+              "excerpt": "Comparison to Existing DBMS Testing Approaches In order to evaluate JoinEquiv’s unique ability to detect logical bugs related to INNER JOIN, we compare it with two representative metamorphic testing approaches: Ternary Logic Partitioning (TLP) and Differential Query Plans (DQP).",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, E Comparison to Existing DBMS Testing Approaches, page 9.",
+              "content_sha256": "sha256:751ef489d3b7632a9b6c19d1e66e60d87df93c41cedc205498a4de62ab38dedf",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00224.json",
+              "source_type": "paper",
+              "excerpt": "For each reported bug, we reconstruct equivalent test cases following TLP and DQP formulations to examine whether these approaches can reproduce the same erroneous behavior.",
+              "excerpt_is_verbatim": true,
+              "note": "M16 in the paper's extracted text, E Comparison to Existing DBMS Testing Approaches, page 10.",
+              "content_sha256": "sha256:84b77ef89f773b1a9289293989cba95ebdde42de740754c87966c699c5698007",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00224.json",
+              "source_type": "paper",
+              "excerpt": "Crucially, for every generated original INNER JOINquery that can be correctly executed, we apply all three transformation strategies (TLP, DQP, and JoinEquiv) within the same iteration, ensuring a fair comparison under an identical execution context.",
+              "excerpt_is_verbatim": true,
+              "note": "M17 in the paper's extracted text, E Comparison to Existing DBMS Testing Approaches, page 10.",
+              "content_sha256": "sha256:7d742e3fb5882a05a25b3c32cede3f8696030316a6f66e3bcdef2a39f665298f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/f959eadc36d89bdfc3add245defa2b3b50c40954",
+        "source_type": "paper",
+        "content_sha256": "sha256:a88913300728f71d0002d935b160c0dbcd17aed0e9f07babe88eb3c061966c0d"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3797134",
+      "title": "ACME: Automated Clause Mapping Engine for Testing Emerging Database Systems",
+      "authors": [
+        "Yuancheng Jiang",
+        "Jianing Wang",
+        "Chuqi Zhang",
+        "Roland H. C. Yap",
+        "Zhenkai Liang",
+        "Manuel Rigger"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Software Engineering",
+      "doi": "10.1145/3797134",
+      "arxiv_id": null,
+      "s2_paper_id": "590a51bd832d5edd4ab164b6311c86a66e69319d",
+      "url": "https://doi.org/10.1145/3797134",
+      "open_access_pdf": "https://doi.org/10.1145/3797134",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/590a51bd832d5edd4ab164b6311c86a66e69319d",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Query Plan Guidance (QPG), Differential Query Plans (DQP).",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/590a51bd832d5edd4ab164b6311c86a66e69319d",
+        "source_type": "paper",
+        "content_sha256": "sha256:45433ac6e15ef27d38d18c7c5086340a85abad9f06e4c77f435df81a16802717"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "A growing number of emerging database management systems, such as time-series and streaming database systems, have been developed to support specialized workloads with enhanced performance and functionality. However, these systems are often less mature than traditional relational database systems, making them more prone to logic bugs and internal errors affecting correctness and reliability. To address this, we propose an enhanced differential testing framework designed for emerging SQL-like database systems. Our key insight is that many of these systems are conceptually extensions of relational database systems, allowing us to uncover bugs by comparing query results with those from more robust and mature relational database systems. To bridge the differences in syntax and semantics between emerging and relational database systems, we leverage Large Language Models (LLMs) to make differential testing more effective by discovering clause mappings that translate system-specific features in emerging database systems into equivalent SQL expressions. Our approach proceeds in three steps: (i) analyzing the syntax and semantics of queries with runtime errors to reason on clause mappings using LLMs; (ii) validating the generated clause mappings by executing test queries and re-prompting upon validation failures; and (iii) generating semantically equivalent, yet syntactically diverse queries to broaden the coverage of differential testing. We implemented this approach in a tool called ACME and applied it to four widely used emerging database systems, uncovering 59 previously unknown bugs, including 17 logic bugs and 42 internal errors. Of these, 52 have been fixed and 5 confirmed by vendors. Our evaluation demonstrates that ACME enhances LLM reliability through query validation. Furthermore, the evaluation shows the effectiveness of differential testing even when using local models or a limited online token budget. Our results demonstrate the practicality and effectiveness of ACME in improving the robustness and accuracy of emerging database systems through scalable, LLM-assisted differential testing."
+    },
+    {
+      "id": "paper:arxiv:2607.09072",
+      "title": "Agentic Proof and Property-Based Testing via Property-Templates in Data-Intensive Computing",
+      "authors": [
+        "Seongmin Lee",
+        "Yaoxuan Wu",
+        "Miryung Kim"
+      ],
+      "year": 2026,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": "2607.09072",
+      "s2_paper_id": "3b5bde6d80922a14c37ac576e2ace45d800f99ca",
+      "url": "https://arxiv.org/abs/2607.09072",
+      "open_access_pdf": null,
+      "abstract": "As the cost of code generation becomes cheaper with AI, the new bottleneck in software engineering has shifted to intent specification and validation. Overcoming this durability crisis of AI-driven coding requires more than traditional fuzzing: each candidate property must be proven correct over a model and shown to hold on the real implementation, making formal proof and systematic property-based testing (PBT) complementary. However, validating properties this way at scale requires solving two subproblems: verifying candidate properties and operationalizing PBT without AI hallucination. We hypothesize that recurring property patterns, cast as property templates--abstract, parameterized forms with holes--address both at once. This paper investigates recurring property patterns in Apache Spark. In data-intensive scalable computing systems, correctness properties arise from the principles of data partition, computation decomposition, and dataflow computation. For instance, aggregation decomposition relates a global function executed on the entire dataset to a local function followed by a recombiner. We design an agentic, dual-track validation framework that uses property templates to formally verify correctness in the Lean 4 theorem prover and instantiate PBT templates as executable PySpark tests. Our evaluation shows that property templates increase agentic proof engineering success by up to 2.6x (1.6x on average) and reduce proof hallucinations by 59%. Template-guided PBT synthesis reduces intent misalignments from 22 to 1 and cuts synthesis cost by up to 5.7x (3.8x on average). Template-guided synthesis further exceeds a state-of-the-art Spark fuzzer and approaches unguided LLM-based PBT on code coverage. Finally, comparing the two tracks is informative: when a proof succeeds yet a PBT finds a counterexample, the mismatch identifies a gap between the formal model and implementation.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2607.09072",
+              "source_type": "paper_citation_context",
+              "excerpt": "On the testing side, SQLancer detects logic and optimization bugs in database engines through constructed oracles such as query partitioning and a non-optimizing reference engine [28], [29].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0f1399c6bcd7c061cc40c2ada8cfc72ac9797f30bcf62350dceb32394d1a02a8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/3b5bde6d80922a14c37ac576e2ace45d800f99ca",
+        "source_type": "paper",
+        "content_sha256": "sha256:979789e83f0281827bd7ba14be33843c0e762cb7bd796e4fb8c4196950950a62"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/missf68264.2026.11521893",
+      "title": "An Automated Risk Scoring Framework for SQL Execution Plan Analysis and Performance Regression Detection in Oracle Database Systems",
+      "authors": [
+        "Raghu Gollapudi"
+      ],
+      "year": 2026,
+      "venue": "2026 International Conference on Multidisciplinary Innovations For Smart & Sustainable Future (MISSF)",
+      "doi": "10.1109/missf68264.2026.11521893",
+      "arxiv_id": null,
+      "s2_paper_id": "cc4a8303c7da6ce70e1fb471c23442d89ada3b7e",
+      "url": "https://doi.org/10.1109/missf68264.2026.11521893",
+      "open_access_pdf": null,
+      "abstract": "This paper suggests a proactive and automated risk scoring system in systematic evaluation of SQL execution plans and performance regression of Oracle Database systems in early stages of implementation. The framework proposes a new multi-factor weighted scoring model, which measures SQL risk on a normalized 0 100 scale using a combination of resource consumption data, execution efficiency data, plan regression data, bind variable health, and execution plan quality data. In contrast to conventional reactive techniques of monitoring, the suggested PL/SQL-based agent continuously analyzes dynamic performance views, historical workload archives, and variations in the actions of execution plans to identify silent plan modifications and degradation patterns in near real time. Formalized scoring is a scoring behavior that averages the normalized factors through pre-established weights to create tiers-based risk formations, which allow prioritized intervention. It uses the built-in anti-pattern detection, wait event classification and remediation prescriptive guidance in the architecture. The framework has been experimentally validated in enterprise Oracle environments with high detection and lower mean time to detection rates alongside insignificant system overhead, placing it as an offering that is a scalable and production-ready system to provide proactive SQL performance governance.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/missf68264.2026.11521893",
+              "source_type": "paper_citation_context",
+              "excerpt": "These systems however tend to show raw metrics without merging them into a risk model [14].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ef504f6e4d608f4e948002fcaeef07521ef62dc4e72ef43049cb9b6f76710a66",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/cc4a8303c7da6ce70e1fb471c23442d89ada3b7e",
+        "source_type": "paper",
+        "content_sha256": "sha256:c1708e38d65862a332214480a507f114ca3b19cb02ba6dd48846e615667df44f"
+      }
+    },
+    {
+      "id": "paper:arxiv:2607.13276",
+      "title": "Aurora DSQL: Scalable, Multi-Region OLTP",
+      "authors": [
+        "Marc Brooker",
+        "Mark Bowes",
+        "Mike Hershey",
+        "Zak van der Merwe",
+        "James Morle",
+        "Matthys Strydom"
+      ],
+      "year": 2026,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": "2607.13276",
+      "s2_paper_id": "07b3167f54375c3db6c83026a77b0aa8bc8f93ce",
+      "url": "https://arxiv.org/abs/2607.13276",
+      "open_access_pdf": null,
+      "abstract": "Aurora DSQL is a serverless SQL database designed for cloud-scale transaction processing with multi-region active-active capabilities. Built on a disaggregated architecture, DSQL separates compute, storage, and transaction coordination into independent, horizontally scalable services. Query processors run in Firecracker MicroVMs executing PostgreSQL-compatible SQL without local state. The system uses multiversion concurrency control with precision timestamps for coordination-free reads and optimistic concurrency control for writes, deferring coordination to commit time through distributed adjudicators and the Journal replication system. This minimizes cross-region latency by requiring coordination only during commits, not individual statements. DSQL enables elastic scaling from zero to millions of transactions per second while providing strong consistency, ACID transactions, and continuous availability during availability zone or region failures.",
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2607.13276",
+              "source_type": "paper_citation_context",
+              "excerpt": "The fuzz-testing approach, building on the approach of SQLancer [4], generates millions of example SQL statements and runs them both on DSQL and on Aurora PostgreSQL.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7e72c871a2b5104b503e245b1eb572878d92bc3ca4d42b99de79852d99fd7579",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "uncertain",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2607_13276.json",
+              "source_type": "paper",
+              "excerpt": "The fuzz-testing approach, building on the approach of SQLancer [ 4], generates millions of example SQL statements and runs them both on DSQL and on AuroraPostgreSQL.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, 5.6 Effects of Clock Skew, page 10.",
+              "content_sha256": "sha256:505d40d60af6b9f49d2d41a6fcd93750e074bda4885938800f54fb400407a19f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/07b3167f54375c3db6c83026a77b0aa8bc8f93ce",
+        "source_type": "paper",
+        "content_sha256": "sha256:6b05139dbb4b03bdd47aeeb3253c5d96ee77cc375ab0c9e54d0faa28d16fca36"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3798232",
+      "title": "Beacon: Detecting Broken Access Control Vulnerabilities in DBMSs via System Catalog Consistency Validation",
+      "authors": [
+        "Zongrui Peng",
+        "Jingzhou Fu",
+        "Zhiyong Wu",
+        "Jie Liang",
+        "Xiangdong Huang",
+        "Dalong Shi",
+        "Yu Jiang"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Programming Languages",
+      "doi": "10.1145/3798232",
+      "arxiv_id": null,
+      "s2_paper_id": "804bc6d1f6c1acde45f1fabcab8b0d3ed5f647c9",
+      "url": "https://doi.org/10.1145/3798232",
+      "open_access_pdf": "https://doi.org/10.1145/3798232",
+      "abstract": "Access control in DBMSs is critical for ensuring data security and integrity. However, the increasing complexity of its implementation often introduces broken access control (BAC) vulnerabilities. These vulnerabilities can lead to severe consequences, including privilege escalation, unauthorized data access, or even full compromise of the DBMS. Existing manual testing for BAC vulnerabilities is time-consuming and incomplete. Automated methods like static analysis also struggle in DBMSs, as static rules are difficult to apply to multi-level and dynamically changing privileges. In this paper, we propose Beacon, which detects BAC vulnerabilities by validating the consistency between SQL operations and system catalogs. Our key insight is that the visibility of objects in the system catalogs is consistent with the user’s access control: if an object is invisible to a user in the system catalogs, the user should not have any access privileges on that. Any inconsistency suggests that a user is exceeding their privileges, indicating a potential BAC vulnerability.We used Beacon to test eight popular DBMSs (e.g., MySQL and MariaDB), uncovering 39 previously unknown BAC vulnerabilities. Among them, 19 result in privilege escalation, and 20 lead to unauthorized information disclosure. Moreover, 7 of them have existed in DBMSs for more than 6 years, with the longest-persisting one lasting 13 years. DBMS vendors took these issues seriously and have already confirmed all of these vulnerabilities. Many vendors provided positive feedback, recognizing the importance of addressing these vulnerabilities. For instance, OceanBase awarded bounties for reported vulnerabilities, underscoring Beacon’s role in improving DBMS access control.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/804bc6d1f6c1acde45f1fabcab8b0d3ed5f647c9",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Cardinality Estimation Restriction Testing (CERT).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/804bc6d1f6c1acde45f1fabcab8b0d3ed5f647c9",
+        "source_type": "paper",
+        "content_sha256": "sha256:cc78994758be76ebca02c6be697ebd03cc6bd7f2cbcc992b6f4fa5ef8f4db8d9"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3810991.3811637",
+      "title": "Boosting DBMS Test Coverage via LLM-Driven SQL Generation",
+      "authors": [
+        "Eslam Abdelkarim",
+        "Carsten Binnig",
+        "Anupam Sanghi"
+      ],
+      "year": 2026,
+      "venue": "DBTest@SIGMOD",
+      "doi": "10.1145/3810991.3811637",
+      "arxiv_id": null,
+      "s2_paper_id": "bc956d15e20b65ccf568bf75cf6b3c5689d8f9fa",
+      "url": "https://doi.org/10.1145/3810991.3811637",
+      "open_access_pdf": "https://doi.org/10.1145/3810991.3811637",
+      "abstract": "Database management systems require comprehensive testing, but achieving high code coverage in complex DBMS implementations remains challenging. Manual test writing is time-consuming, while automated query generation tools struggle with schema flexibility and coverage guidance. This paper presents Quover, an automated approach for improving DBMS test coverage through large language model (LLM)-generated SQL queries. Specifically, Quover implements an iterative coverage-guided methodology that identifies uncovered functions in the target DBMS source code and generates targeted SQL queries. In each iteration, an LLM call is made with a rich context, including function descriptions and code snippets, and their effectiveness is validated. Our experiments show that over 57% coverage can be achieved within a few hours–already representing a 14% improvement over state-of-the-art automated systems. Furthermore, we evaluate Quover across multiple database schemas and DBMSs, demonstrating its effectiveness as an automated approach alongside handcrafted test suites.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811637",
+              "source_type": "paper_citation_context",
+              "excerpt": "…and pair them with metamorphic or partitioning-based oracles (e.g., Pivoted Query Synthesis (PQS) [6], NoREC [4], and Ternary Logic Partitioning (TLP) [5]) to uncover logic bugs and exercise diverse execution paths in DBMSs, but they do not explicitly optimize for exercising uncovered code regions.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:cac4d3ab7f67af0ca5dc9f8782ab09a2792b4ca893c43f429c783f40748ea84a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811637",
+              "source_type": "paper_citation_context",
+              "excerpt": "Logic-testing tools such as SQLancer [3, 6] generate queries and pair them with metamorphic or partitioning-based oracles (e.g., Pivoted Query Synthesis (PQS) [6], NoREC [4], and Ternary Logic Partitioning (TLP) [5]) to uncover logic bugs and exercise diverse execution paths in DBMSs, but they do…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3a3fe909dee17d83b750c92bd4153a774cff5f376db12182760d6222adef29ba",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811637",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQL fuzzing tools such as SQLancer [3, 6] automatically generate syntactically valid queries and use logic-testing oracles (e.g., Pivoted Query Synthesis) to uncover logic bugs in DBMSs. Coverage-guided fuzzers such as RATEL [11] and SQLRight [1] incorporate code coverage feedback to iteratively…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:de8fe10e85ec5668d4c054c2da06eaac50ac17549c90d5bc43596aa5a3cf81c2",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811637",
+              "source_type": "paper_citation_context",
+              "excerpt": "…tools such as SQLancer [3, 6] generate queries and pair them with metamorphic or partitioning-based oracles (e.g., Pivoted Query Synthesis (PQS) [6], NoREC [4], and Ternary Logic Partitioning (TLP) [5]) to uncover logic bugs and exercise diverse execution paths in DBMSs, but they do not…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b39d9e4bc8932b9c5a9719b23148df881b2d3b1e9ed9f312890e237ee4d628e0",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811637",
+              "source_type": "paper_citation_context",
+              "excerpt": "…as SQLancer [3, 6] generate queries and pair them with metamorphic or partitioning-based oracles (e.g., Pivoted Query Synthesis (PQS) [6], NoREC [4], and Ternary Logic Partitioning (TLP) [5]) to uncover logic bugs and exercise diverse execution paths in DBMSs, but they do not explicitly…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:21ef2b79bc7c2527806d10cb7efefab0e27277e1590d71af9f0ff4cf2e879e12",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/bc956d15e20b65ccf568bf75cf6b3c5689d8f9fa",
+        "source_type": "paper",
+        "content_sha256": "sha256:99850d49dcc441322e670512499c47550bdf0689077276d9f8070f03152539f3"
+      }
+    },
+    {
+      "id": "paper:arxiv:2609.00381",
+      "title": "Bounded, Indeterminate, or a Bug: A Condition-Aware Oracle for Differential Testing of SQL Aggregates",
+      "authors": [
+        "Madhulatha Mandarapu",
+        "Sandeep Kunkunuru"
+      ],
+      "year": 2026,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": "2609.00381",
+      "s2_paper_id": "cc8f54203ea326fd6d9d05e6856bf8b297829375",
+      "url": "https://arxiv.org/abs/2609.00381",
+      "open_access_pdf": null,
+      "abstract": "Differential database testing compares results across engines and calls a discrepancy a bug. For floating-point aggregates this is unsound: engines legitimately disagree because floating-point arithmetic is not associative. Practice patches this with an epsilon; the leading oracles avoid floating point entirely. We give the oracle this practice lacks, and show its decisive quantity is not the query but the engine's algorithm. Ground truth is the exact rational value of the stored doubles -- arithmetic, not another engine -- and each discrepancy is classified exact, bounded, or indeterminate. The relative error of an aggregate f under an algorithm A obeys rel_err<= C_A(n,u) * kappa_f^p, so the testability boundary, beyond which no oracle can separate a bug from rounding, is kappa*_{f,A} = (1/C_A)^{1/p}. SUM and AVG are the linear case p=1; variance is p=2 for the one-pass algorithm and p=1 for Welford. Across eight engines in four classes the measured exponent recovers each algorithm, and ClickHouse is the lone one-pass engine (p=2.05); engine-wide, it returns zero standard deviation, NaN correlation and wrong-sign regression, while every other engine stays exact and the vendor ships the Welford fix. Its variance is untestable at a condition number 10^6 below SUM's, which ordinary storage conventions (epoch-nanosecond timestamps, tight sensors) cross -- there ClickHouse errs by 2100%. A randomised hunt of 360 tests finds zero anomalies, evidence the oracle is sound. Code and data are public.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2609.00381",
+              "source_type": "paper_citation_context",
+              "excerpt": "The oracles of Rigger and Su [2020b,a,c] either restrict aggregates to a single pivot row, target predicates rather than aggregation, or compare an engine against itself by query partitioning, so a consistent rounding error cancels and is never observed.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3b3846f328f41c079df56e64b6ad2f3c5622b5e1ec06e176b5b243fa0e528f1a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "uncertain",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2609_00381.json",
+              "source_type": "paper",
+              "excerpt": "The only route tosum’s indeterminate regime is fuzzer data: replicating SQLancer’s generator, exactly cancelling ±MAX drives κ≈10306, and the fraction of undecidable columns is non-monotone in n(peaking near20%at n=1000).",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 5 Results I: SUM is the linear baseline, page 5.",
+              "content_sha256": "sha256:4dd739aca447c115f76fe33227ff5acf104f098b2b5d10ebf1b59d9edc1589f3",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/cc8f54203ea326fd6d9d05e6856bf8b297829375",
+        "source_type": "paper",
+        "content_sha256": "sha256:0b312cfc16f679ad8b0418c3f629bded824f5ceddb7408f8ede02abec51a1624"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3797871",
+      "title": "Bringing Order to Practical Validation of Database Isolation Levels",
+      "authors": [
+        "Jack Clark",
+        "Manuel Rigger",
+        "John Wickerson",
+        "Alastair F. Donaldson"
+      ],
+      "year": 2026,
+      "venue": "ACM Transactions on Computer Systems",
+      "doi": "10.1145/3797871",
+      "arxiv_id": null,
+      "s2_paper_id": "16781c41bad0d986624b27ac62b35446df724bd9",
+      "url": "https://doi.org/10.1145/3797871",
+      "open_access_pdf": "https://doi.org/10.1145/3797871",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3797871",
+              "source_type": "paper_citation_context",
+              "excerpt": "This approach is inspired by pivoted query synthesis [47], which uses a database system specific SQL interpreter to execute a predicate condition to determine if a given row would match the predicate.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:347bc606e379f70a7278c8d1334feac2aa697ebe8cf45f491ccae260579a9d48",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3797871",
+              "source_type": "paper_citation_context",
+              "excerpt": "We believe this is reasonable as there are existing techniques [1, 6, 45–49] to validate the correctness of the query evaluator/optimizer and the focus of this work is on finding bugs in the implementation of concurrency control protocols.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:246d042b150233012b18ddb2285d705075a0fa68c2f06947d1c9dafcda647964",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "manual_curation",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3797871",
+              "source_type": "paper_citation_context",
+              "excerpt": "This approach is inspired by pivoted query synthesis [47], which uses a database system specific SQL interpreter to execute a predicate condition to determine if a given row would match the predicate.",
+              "excerpt_is_verbatim": true,
+              "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+              "content_sha256": "sha256:347bc606e379f70a7278c8d1334feac2aa697ebe8cf45f491ccae260579a9d48",
+              "retrieved_at": "2026-09-06T07:08:48Z",
+              "first_seen": "2026-09-06T07:08:48Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ],
+          "classifier": {
+            "method": "manual_curation",
+            "classifier_version": "manual-curation-v1",
+            "policy_version": "impact-policy-v1",
+            "taxonomy_version": "taxonomy-v1",
+            "classified_at": "2026-09-06T07:08:48Z",
+            "model": null,
+            "rationale": "Reviewed by a maintainer against the citation contexts collected for this paper."
+          }
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/16781c41bad0d986624b27ac62b35446df724bd9",
+        "source_type": "paper",
+        "content_sha256": "sha256:0a2120a32adbb103b55d494f7736d4c98beab679c84044fe038b739a99f39eae"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Transactions are a key feature of database systems and isolation levels specify the behavior of concurrently executing transactions. Ensuring the correctness of isolation levels is crucial. Recently, many isolation anomalies have been found in production database systems. Checkers can be used to validate that a particular execution history conforms to a desired isolation level. However, state-of-the-art checkers cannot handle predicate operations, which are both common in real-world workloads and essential for distinguishing between the repeatable read and serializable isolation levels. In this work, we address this issue by proposing two techniques for efficient white-box checking. Our key idea is to use information, that is, easily provided by database systems to efficiently check the isolation level of a given execution history. We present the version certificate recovery technique and its associated checker Emme. Version certificate recovery is a method of recovering the version order and each operation’s version set from the database system under test. To minimise execution time, we also propose the expected serialization order technique, along with its associated checker Enne, which obviates the need to define and recover a version certificate for many serializable concurrency control protocols. We have implemented version certificate recovery for three widely used database systems—PostgreSQL, CockroachDB, and TiDB. We demonstrate that Emme is 1.2–3.6× faster than Elle, a state-of-the-art checker. When paired with the expected serialization order technique, Enne obtains a further speedup of 34–430× when checking execution histories containing predicate operations. We show that our approaches can identify invalid execution histories that cannot be detected by Elle and also show that they can find realistic bugs purposely introduced by an engineer. Finally, we create a new checker, King Cobra, by modifying Cobra, an existing black-box checker, in order to conduct an ablation study to evaluate how the performance of a black-box checker changes when provided with varying degrees of ordering information. By doing so, we highlight the fundamental role that ordering information plays in the execution time of an isolation level checker and explore the limits that this places on a truly black-box checker."
+    },
+    {
+      "id": "paper:arxiv:2604.03024",
+      "title": "BugForge: Constructing and Utilizing DBMS Bug Repository to Enhance DBMS Testing",
+      "authors": [
+        "Dawei Li",
+        "Qifan Liu",
+        "Yuxiao Guo",
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Chi Zhang",
+        "Jingzhou Fu",
+        "H. Mao",
+        "Zhenyu Guan",
+        "Yu Jiang"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2604.03024",
+      "s2_paper_id": "597ba93345ffe08175897ddb2019c676914992af",
+      "url": "https://arxiv.org/abs/2604.03024",
+      "open_access_pdf": null,
+      "abstract": "DBMSs are complex systems prone to bugs that may lead to system failures or compromise data integrity. Establishing unified DBMS bug repositories is crucial for systematically organizing bug-related data, enabling code improvement, and supporting automated testing. In particular, bug reports often contain valuable test inputs and bug-triggering clues that help explore rare execution paths and expose critical buggy behavior, thereby guiding automated DBMS testing. However, the heterogeneity of bug reports, along with their incomplete or inaccurate content, makes it challenging to build unified repositories and convert them into high-quality test cases. In this paper, we propose BugForge, a framework that constructs standardized DBMS bug repositories and leverages them to generate high-quality test cases to enhance DBMS testing. Specifically, BugForge progressively collects bug reports, then employs syntax-aware processing and input-adaptive raw PoC extraction to construct a DBMS bug repository. The repository stores structured bug-related data, including bug metadata and raw PoCs that entail potential bug-triggering semantics. These data are further refined into high-quality test cases through semantic-guided adaptation, thereby enabling enhanced DBMS testing methods, including DBMS fuzzing, regression testing, and cross-DBMS bug discovery. We implemented BugForge for PostgreSQL, MySQL, MariaDB, and MonetDB, totally integrated 37,632 bug reports spanning up to 28 years. Based on the repository, BugForge uncovered 35 previously unknown bugs with 22 confirmed by developers, demonstrating the value of constructing and utilizing bug repositories for DBMS testing.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/597ba93345ffe08175897ddb2019c676914992af",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2604_03024.json",
+              "source_type": "paper",
+              "excerpt": "For SQLancer, we enable its FUZZ mode for testing.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, D Comparison of State-of-the-art DBMS Testing Tools., page 10.",
+              "content_sha256": "sha256:fe54da5824bd37c1f00d45b848481d8f24f04a20806d242cea658cc712688ca7",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2604_03024.json",
+              "source_type": "paper",
+              "excerpt": "Overall,BugForge outperforms both SQLancer and SQLsmith in branch coverage across all evaluated DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, D Comparison of State-of-the-art DBMS Testing Tools., page 10.",
+              "content_sha256": "sha256:63e1af18e8835d25b9a34a6d152232a9ac4bb258c7c3d3bfb558fb445f810dda",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2604_03024.json",
+              "source_type": "paper",
+              "excerpt": "Over the three commonly supported DBMSs,BugForge covers 441,160 branches in total, compared with 201,299 for SQLancer and 177,492 for SQLsmith.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, D Comparison of State-of-the-art DBMS Testing Tools., page 10.",
+              "content_sha256": "sha256:fc61bb7cb4544331749fba8119b9379bf5a86b35a33dc743deb831e7949d424f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2604_03024.json",
+              "source_type": "paper",
+              "excerpt": "Over the three jointly supported DBMSs except MonetDB,BugForgedetected 18 bugs in total, compared with 11 for SQLancer and 6 for SQLsmith.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, D Comparison of State-of-the-art DBMS Testing Tools., page 10.",
+              "content_sha256": "sha256:c4410a92ca47f6855e1f34a524c356515d5b798a69ec0327972d8144e92efcc2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2604_03024.json",
+              "source_type": "paper",
+              "excerpt": "For SQLancer, we enable its FUZZ mode for testing.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, D Comparison of State-of-the-art DBMS Testing Tools., page 10.",
+              "content_sha256": "sha256:fe54da5824bd37c1f00d45b848481d8f24f04a20806d242cea658cc712688ca7",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/597ba93345ffe08175897ddb2019c676914992af",
+        "source_type": "paper",
+        "content_sha256": "sha256:30802b630262ef8363b8e70f4baa4af1e8f546bc0ed45fc5e8f8079a17b845b0"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3810991.3811634",
+      "title": "Cloud Analytics Benchmarking: Where We Are, What Is Missing, and Where We Should Go",
+      "authors": [
+        "Michail Georgoulakis Misegiannis",
+        "Till Steinert"
+      ],
+      "year": 2026,
+      "venue": "DBTest@SIGMOD",
+      "doi": "10.1145/3810991.3811634",
+      "arxiv_id": null,
+      "s2_paper_id": "1ee6ae0a6a863da15be2b876cc1493dd2c6acde5",
+      "url": "https://doi.org/10.1145/3810991.3811634",
+      "open_access_pdf": "https://doi.org/10.1145/3810991.3811634",
+      "abstract": "Workload traces from cloud vendors have significantly advanced our understanding of real-world analytical workloads. These traces contain execution telemetry (CPU time, scanned bytes, operator counts) but omit SQL text for privacy reasons, preventing their direct use as benchmarks. This raises a fundamental question: Can realistic workloads be synthesized from anonymized telemetry? We argue that current telemetry-based synthesis can capture workload-level properties well (e.g., concurrency, repetition) but remains challenging at the query level. We show that telemetry varies across systems, deployments, and execution contexts, and does not uniquely identify the query structure. To bridge this gap, we outline two complementary paths: enriching traces with engine-agnostic transferable workload features, and building use-case-centric benchmarks directly from open application scenarios such as ELT pipelines and BI dashboards.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811634",
+              "source_type": "paper_citation_context",
+              "excerpt": "Related work also considers database testing [21, 37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1f06fe67ef811d95280cdd9fdbba892f5775e9d27c9d6b3b1a1971ac80ea1da7",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/1ee6ae0a6a863da15be2b876cc1493dd2c6acde5",
+        "source_type": "paper",
+        "content_sha256": "sha256:b9c06f877e5febdc7e2e48cb504fa096fcb3d8ff7a4712b6c45740c89c322655"
+      }
+    },
+    {
+      "id": "paper:arxiv:2605.20473",
+      "title": "Code Generation by Differential Test Time Scaling",
+      "authors": [
+        "Yifeng He",
+        "Ethan Wang",
+        "Jicheng Wang",
+        "Xuanxin Ouyang",
+        "Hao Chen"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2605.20473",
+      "s2_paper_id": "bea732652ae1cab45f2ba3678e48cfc22cf4b72a",
+      "url": "https://arxiv.org/abs/2605.20473",
+      "open_access_pdf": null,
+      "abstract": "Test-time scaling has emerged as a promising approach for improving code generation by exploring large solution spaces at inference time. However, existing methods often rely on public test cases that are unavailable in practice, or require extensive LLM inference for candidate selection, leading to significant token consumption and time overhead. We present DiffCodeGen, a novel test-time scaling method for code generation based on coverage-guided differential analysis. DiffCodeGen generates diverse code candidates using various sampling and prompting strategies, then applies coverage-guided fuzzing to synthesize inputs without requiring any existing tests or large language models. By executing all candidates on these inputs, DiffCodeGen captures their dynamic behavior and clusters candidates based on behavioral similarity. DiffCodeGen selects the medoid of the largest cluster as the final output. Unlike prior test-time scaling methods that invoke additional LLM inference for candidate selection, DiffCodeGen performs selection without any extra model calls, incurring little to no additional token consumption. DiffCodeGen is fully asynchronous, naturally suited to the current trend of agentic coding, and is thus efficient and highly scalable. We evaluate DiffCodeGen across 4 large language models, demonstrating consistent improvements over baselines. Compared to state-of-the-art test-time scaling methods, DiffCodeGen achieves competitive or superior performance while using only a fraction of time and tokens. DiffCodeGen is model-agnostic and can be combined with reasoning models to further boost performance.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2605.20473",
+              "source_type": "paper_citation_context",
+              "excerpt": "This technique has proven highly effective for testing compilers [16–18], database systems [19], and other systems where formal specifications or programmer-provided test oracles are either unavailable or impractical to use.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c4ea51f878ab81b15e54e922dddcdd5873acaf45a0700ccd2d5712fead2c239e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2605.20473",
+              "source_type": "paper_citation_context",
+              "excerpt": "Much research has applied differential testing with such assumptions to complex software systems like compilers and databases [13, 16, 19, 25], 6 Related work",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d37a345c524d53c098dafb66bdcde003e60e099ceb05bd62e213889e5c0cfc3e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/bea732652ae1cab45f2ba3678e48cfc22cf4b72a",
+        "source_type": "paper",
+        "content_sha256": "sha256:903d94c8e4ef5c76cac58cae4e2384f15c3b18eddbd118ccfb6be5fc13f900fc"
+      }
+    },
+    {
+      "id": "paper:arxiv:2603.19434",
+      "title": "Computer-Orchestrated Design of Algorithms: From Join Specification to Implementation",
+      "authors": [
+        "Zeyuan Hu"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2603.19434",
+      "s2_paper_id": "7fa57423ffb12f9cd6dace5f26042a6289837bf4",
+      "url": "https://arxiv.org/abs/2603.19434",
+      "open_access_pdf": null,
+      "abstract": "Equipping query processing systems with provable theoretical guarantees has been a central focus at the intersection of database theory and systems in recent years. However, the divergence between theoretical abstractions and system assumptions creates a gap between an algorithm's high-level logical specification and its low-level physical implementation. Ensuring the correctness of this logical-to-physical translation is crucial for realizing theoretical optimality as practical performance gains. Existing database testing frameworks struggle to address this need because necessary algorithm-specific inputs such as join trees are absent from standard test case generation, and integrating complex algorithms into these frameworks imposes prohibitive engineering overhead. Fallback solutions, such as using macro-benchmark queries, are inherently too noisy for isolating intricate defects during this translation. In this experience paper, we present a retrospective analysis of $\\mathsf{CODA}$, a computer-orchestrated testing framework utilized during the logical-physical co-design of TreeTracker Join ($\\mathsf{TTJ}$), a theoretically optimal yet practical join algorithm recently published in ACM TODS. By synthesizing minimal reproducible examples, $\\mathsf{CODA}$ successfully isolates subtle translation defects, such as state mismanagement and mapping conflicts between join trees and bushy plans. We demonstrate that this logical-to-physical translation process is a bidirectional feedback loop: early structural testing not only hardened $\\mathsf{TTJ}$'s physical implementation but also exposed a boundary condition that directly refined the formal precondition of $\\mathsf{TTJ}$ itself. Finally, we detail how confronting these translation challenges drove the architectural evolution of $\\mathsf{CODA}$ into a robust, structure-aware test generation pipeline for join-tree-dependent algorithms.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/7fa57423ffb12f9cd6dace5f26042a6289837bf4",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/7fa57423ffb12f9cd6dace5f26042a6289837bf4",
+        "source_type": "paper",
+        "content_sha256": "sha256:5de8ee88cfcc88fe326e0413398889bf9ee845d50b6318e74485e56967f4f9d5"
+      }
+    },
+    {
+      "id": "paper:arxiv:2608.25573",
+      "title": "DBcover: A White-box SQL Test Generation Framework for Coverage Improvement",
+      "authors": [
+        "Yan-Kai Rong",
+        "Shuang Liu",
+        "Jinhao Dong",
+        "Qiang Yin",
+        "Wei Lu",
+        "Jian-Hua Wang",
+        "Xiaoyong Du"
+      ],
+      "year": 2026,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": "2608.25573",
+      "s2_paper_id": "2b1f7251af13eb3114cc0c8812ae6a07cf970db2",
+      "url": "https://arxiv.org/abs/2608.25573",
+      "open_access_pdf": null,
+      "abstract": "Relational Database Management Systems (RDBMSs) are the backbone of modern data-intensive applications, making reliability and robustness critical. However, achieving high coverage in RDBMS testing remains challenging because of large codebases and complex execution logic. Traditional fuzzing relies on random SQL generation and cannot capture the correspondence between SQL inputs and internal execution paths, while symbolic execution suffers from prohibitive cost and scalability limitations. We propose DBcover, an LLM-driven white-box SQL test generation framework based on contextual reasoning. DBcover uses lightweight dynamic analysis to extract SQL-to-path correspondence and call graphs as global context, and collects source-level information around target functions as local context. These contexts are organized in a unified knowledge graph for efficient retrieval and reuse. DBcover then performs two-phase test generation: it first selects a semantically relevant seed whose execution path is close to the uncovered target, and then guides the LLM with global and local context to generate SQL test cases that trigger previously uncovered code regions. Experiments show that DBcover achieves 80.1% and 82.3% coverage on PostgreSQL and MySQL, and is also effective on the enterprise RDBMS KingbaseES, demonstrating its practical applicability to closed-source systems.",
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/2b1f7251af13eb3114cc0c8812ae6a07cf970db2",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2608_25573.json",
+              "source_type": "paper",
+              "excerpt": "•ShQveL [ 34] enhances existing SQL test case generator (SQLancer++ [ 33]) by leveraging LLM to synthesize SQL fragments.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 4.1.4 Baselines:, page 8.",
+              "content_sha256": "sha256:5d7abe408cc3b5a7a4e621950ed60b43a8b9f3c3872b1224ad2769a1e1962a6d",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2608_25573.json",
+              "source_type": "paper",
+              "excerpt": "ShQveL—a standalone LLM-augmented generator that does not leverage the seed corpus—achieves substantially lower coverage (31.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, 4.2 Comparison with Baselines, page 8.",
+              "content_sha256": "sha256:a40149253324ed39f43c7e53bdf9706da05eeefc9066552efe003881fb5c429c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2608_25573.json",
+              "source_type": "paper",
+              "excerpt": "Both SQUIRREL and shQveL lack such context information, and thus cannot effectively cover the target code region.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 4.2 Comparison with Baselines, page 8.",
+              "content_sha256": "sha256:fe67d203ec47d8b03b33711fc65000c7d8b73655e7f2b88928a8b1b93eaf25c8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/2b1f7251af13eb3114cc0c8812ae6a07cf970db2",
+        "source_type": "paper",
+        "content_sha256": "sha256:7e75b199c30355e6c3ee00bd8d060e2cd372a7521b0185d5a562fc44be5f7081"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3802034",
+      "title": "DBugScribe: Automatic Database Bug Reproduction from Community Reports",
+      "authors": [
+        "Suyang Zhong",
+        "Mo Sha",
+        "Sheng Wang",
+        "Fangyuan Zhou",
+        "Feifei Li",
+        "Kian-Lee Tan"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Management of Data",
+      "doi": "10.1145/3802034",
+      "arxiv_id": null,
+      "s2_paper_id": "8f02d1ed0dfacfff322cccdddf971240b3c8136d",
+      "url": "https://doi.org/10.1145/3802034",
+      "open_access_pdf": null,
+      "abstract": "Major Database Management Systems (DBMSs) like MySQL field a high volume of bug reports daily. Yet, reproducing database bugs from natural language reports demands substantial developer effort due to manual interpretation and trial-and-error reconstruction. Database bugs are particularly challenging because they involve reconstructing multi-dimensional states that encompass configurations, schemas, data, queries, and validation oracles, while coping with natural language ambiguity. Existing automated bug reproduction techniques, designed for stateless functional testing, cannot address the stateful, multi-dimensional nature of database bugs. We present\n DBugScribe,\n the first framework specifically designed to automatically synthesize validated, executable, and structured reproduction scenarios for database bugs directly from users' natural language bug reports. At its core,\n DBugScribe\n introduces a domain-specific language (DSL) with formal semantics to represent bug scenarios as composable specifications. A novel hybrid synthesis approach integrates LLM-based information extraction with rule-based validation and self-refinement. Evaluated on 218 confirmed bug reports from eight recent DBMS testing tools covering MySQL, TiDB, and MariaDB,\n DBugScribe\n achieves 72.9% reproduction success. In our setup,\n DBugScribe\n can synthesize and execute a reproduction scenario within minutes per report. Beyond immediate reproduction,\n DBugScribe\n aggregates validated scenarios into a structured knowledge base that enables systematic bug analysis and cross-DBMS bug detection. This work transforms database bug reproduction from a manual, ad-hoc process into an automated and principled discipline, advancing both software engineering practice and database reliability research.",
+      "cites_seed_techniques": [
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp",
+            "qpg",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/8f02d1ed0dfacfff322cccdddf971240b3c8136d",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP), Query Plan Guidance (QPG), Differential Query Plans (DQP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/8f02d1ed0dfacfff322cccdddf971240b3c8136d",
+        "source_type": "paper",
+        "content_sha256": "sha256:9d5e45d00caa6d3dde69535bb6b6211f46b0df578df4a1aa216e4023bc46490f"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3836663.3836700",
+      "title": "Detecting Data-Type-Related Logic Bugs in Relational DBMSs via Compatible Database Construction",
+      "authors": [
+        "Jiansen Song",
+        "Wensheng Dou",
+        "Yingying Zheng",
+        "Yu Gao",
+        "Quanqing Xu",
+        "Ziyu Cui",
+        "Xudong Xie",
+        "Hongtao Zhou",
+        "Jiaying Zhou",
+        "Jun Wei",
+        "Wei Wang"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3836663.3836700",
+      "arxiv_id": null,
+      "s2_paper_id": "ee292fb6ddb61da67b9a266019226ffd7a8a79d0",
+      "url": "https://doi.org/10.14778/3836663.3836700",
+      "open_access_pdf": null,
+      "abstract": "Relational Database Management Systems (DBMSs) serve as foundational systems for data storage and management, supporting a rich variety of data types to specify storage formats and value ranges. These data types play a critical role in both data storage and computation. However, complex data computation operations (e.g., explicit and implicit data type conversions) can introduce data-type-related logic bugs (TypeBugs for brevity). Specifically, TypeBugs can cause SELECT statements to return incorrect query results, which can easily be overlooked by DBMS developers. Unfortunately, existing DBMS testing approaches do not examine data type conversions, rendering them ineffective at detecting TypeBugs.\n \n We observe that database columns with different data types (e.g., INT and BIGINT) can store identical data values within specific value ranges, and executing identical SQL operations on these database columns should produce consistent results. Inspired by this observation, we propose TypeCheck, a novel testing approach for effectively detecting TypeBugs in relational DBMSs. We first identify storage-compatible data types\n T\n that can preserve identical data values and operation-oriented type compatibility rules\n opRules\n that define SQL operations producing consistent results when applied to different data types. Using\n T\n and\n opRules\n , we construct type-compatible databases and execute identical type-compatible SELECT statements on them. When these SELECT statements yield divergent execution results, we detect a TypeBug. We evaluate TypeCheck on six widely used relational DBMSs, and have detected 37 TypeBugs, of which 34 have been confirmed as previously unknown bugs.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp",
+        "coddtest",
+        "sqlancer_pp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp",
+            "qpg",
+            "cert",
+            "dqp",
+            "coddtest",
+            "sqlancer_pp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/ee292fb6ddb61da67b9a266019226ffd7a8a79d0",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Query Plan Guidance (QPG), Cardinality Estimation Restriction Testing (CERT), Differential Query Plans (DQP), Constant-Optimization-Driven Testing (CODDTest), sqlancer_pp.",
+              "retrieved_at": "2026-09-10T14:52:29Z",
+              "first_seen": "2026-09-10T14:52:29Z",
+              "last_verified": "2026-09-10T14:52:29Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T14:52:29Z",
+        "source_url": "https://www.semanticscholar.org/paper/ee292fb6ddb61da67b9a266019226ffd7a8a79d0",
+        "source_type": "paper",
+        "content_sha256": "sha256:628881eaf44057ab1260625e9f987f1f8a06b5743f8824ab42b44df41317e1cf"
+      }
+    },
+    {
+      "id": "paper:arxiv:2608.30385",
+      "title": "Detecting DBMS Bugs by Constructing Equivalent Representations of Intermediate Query Results",
+      "authors": [
+        "Xiao-Xu Niu",
+        "Gong Chen",
+        "Jin-Fu Chen",
+        "Xiaoyuan Xie"
+      ],
+      "year": 2026,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": "2608.30385",
+      "s2_paper_id": "e239393bf8f261ed8f721e3d78859f15d52d239b",
+      "url": "https://arxiv.org/abs/2608.30385",
+      "open_access_pdf": null,
+      "abstract": "Database Management Systems (DBMSs) support multiple SQL mechanisms for representing intermediate query results, including VIEWs, Common Table Expressions (CTEs), and Temporary Tables (TEMPTs). When these mechanisms are used to represent the same intermediate query result, the corresponding queries are expected to produce consistent results. However, we observe that such queries can return inconsistent results, indicating potential DBMS logic bugs. Existing approaches for detecting DBMS logic bugs have never explored result consistency across such equivalent representations. In this paper, we propose ERIQ, a novel testing approach for detecting DBMS logic bugs from the perspective of checking result consistency across Equivalent Representations of Intermediate Query Results. ERIQ constructs SQL variants using a VIEW, a CTE, or a TEMPT to represent the same intermediate query result, executes these variants, and compares their returned results. We evaluated ERIQ on four widely used open-source DBMSs: MySQL, MariaDB, Percona, and OceanBase. In total, ERIQ detected 64 bugs, 63 of which were confirmed by developers, and two have been fixed. Among the confirmed bugs, 54 were unique and previously unknown logic bugs, and one was a documentation issue.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2608.30385",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, NoREC [29] compares the results of optimized and unoptimized queries; TLP [30] partitions queries based on ternary logic; PQS [31] checks whether a selected pivot row appears in the result of a generated query; Pinolo [8] checks result containment between synthesized queries and a seed…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:40c82f12ff6ce6bbf4d72b582899a470474c031d8d37304894282afe5e43e3a7",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2608.30385",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [29] rewrites a query into a form that inhibits DBMS optimizations and compares the results of the original and rewritten queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:38fce85ce1560a78943a38db23f12a1247827678a85b8d2e3bd13f366f27ae05",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2608.30385",
+              "source_type": "paper_citation_context",
+              "excerpt": "We further compared ERIQ with four state-of-the-art DBMS logic bug detection approaches: EDC [4], Radar [34], EET [11], and TLP [30].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3d9f8ac620590154761d7cf3ae41b71b29bb3c68eb2061f5323335741e8a371f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2608.30385",
+              "source_type": "paper_citation_context",
+              "excerpt": "We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [4], Radar [34], EET [11], and TLP [30].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:46d7c1bda7c182134b7a1cd3de60f4e0a7d826addf82b294c6f2bb3d72ae7942",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2608.30385",
+              "source_type": "paper_citation_context",
+              "excerpt": "To answer RQ2, we compared ERIQ with the 4 baselines introduced in Section 4.1: EDC [4], Radar [34], EET [11], and TLP [30].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c52070b072fb2134abefe44405320e4ca0ea409eb5793d2e3bf4cc721fd0abca",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2608.30385",
+              "source_type": "paper_citation_context",
+              "excerpt": "QPG [1] uses query-plan diversity to guide database-state mutations, and MIST [3] combines hierarchical SQL-feature guidance, error feedback, and coverage-guided Monte Carlo tree search for LLM-based DBMS test-case generation.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5113b899a4baf91f5304dd3186d3e33c073c7d96235f4bd4d383e5acb23a8c5c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2608_30385.json",
+              "source_type": "paper",
+              "excerpt": "We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30].",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 4.1 Experimental Setup, page 6.",
+              "content_sha256": "sha256:c2ef76ed3d2b3dcac26996afb3ac12bdcfc0c602cedcdcdf6d5daed5bfdba40f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2608_30385.json",
+              "source_type": "paper",
+              "excerpt": "DBMSERIQ EDC Radar EETTLP MySQL 26 4 4 5 6 MariaDB 11 2 1 1 3 Percona 13 2 1 0 0 OceanBase 4 0 0 0 0 Total 54 8 6 6 9 Among the 54 unique logic bugs detected by ERIQ, EDC, Radar, EET, and TLP could detect 8, 6, 6, and 9 bugs, respectively, corresponding to 14.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 4.3 RQ2: Comparison with Existing Approaches, page 7.",
+              "content_sha256": "sha256:53f0ba851ded01ecd2fb1db596794e413a68b3a82982c952f3ad573fd436af87",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2608_30385.json",
+              "source_type": "paper",
+              "excerpt": "TLP detected the largest number of ERIQ-detected bugs in this analysis, but only 9 of the 54 bugs; the other baselines detected even fewer.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, 4.3 RQ2: Comparison with Existing Approaches, page 7.",
+              "content_sha256": "sha256:aeff1e56c2d9c502c68319969e811aefab9e0835da9288d988a527634e230b71",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2608_30385.json",
+              "source_type": "paper",
+              "excerpt": "Across the 4 target DBMSs, ERIQ detected 54 logic bugs, whereas EDC, Radar, EET, and TLP detected 16, 2, 1, and 6 bugs, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, 4.3 RQ2: Comparison with Existing Approaches, page 7.",
+              "content_sha256": "sha256:00d8168e1b77bb9b3b6e15a382f5abd85ab8ecaab6660fc7e903394cf51ee6c1",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2608_30385.json",
+              "source_type": "paper",
+              "excerpt": "We further compared ERIQ with four state-of-the-art DBMS logic bug detection approaches: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30].",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, 1 Introduction, page 1.",
+              "content_sha256": "sha256:1cbd6e670f0c695f1a222eed63d365d01460c34fc905124b404120b9cdfea1b8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2608_30385.json",
+              "source_type": "paper",
+              "excerpt": "We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30].",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 4.1 Experimental Setup, page 6.",
+              "content_sha256": "sha256:c2ef76ed3d2b3dcac26996afb3ac12bdcfc0c602cedcdcdf6d5daed5bfdba40f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/e239393bf8f261ed8f721e3d78859f15d52d239b",
+        "source_type": "paper",
+        "content_sha256": "sha256:136af9895442ac45eb6d5e03f442622dc720aec5da34cb0bd76c11b1b3508cc1"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3802036",
+      "title": "Detecting Join Bugs in Database Engines via Join Implication Reasoning",
+      "authors": [
+        "Zhaokun Xiang",
+        "Suyang Zhong",
+        "Manuel Rigger"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Management of Data",
+      "doi": "10.1145/3802036",
+      "arxiv_id": null,
+      "s2_paper_id": "875f8fff2ecc34bfe8532ad93c0c4a536f6e9e45",
+      "url": "https://doi.org/10.1145/3802036",
+      "open_access_pdf": "https://doi.org/10.1145/3802036",
+      "abstract": "Join\n is a fundamental operation in relational database management systems (RDBMSs), as it embodies core relational expressiveness of SQL, enabling users to query and analyze data across multiple tables. While approaches have been proposed specifically to find join bugs in RDBMSs, they are affected by scalability limitations. In this paper, we propose a novel, general, and effective technique for finding join logic bugs across all common join types, under arbitrary join predicates, and on join-supported RDBMSs, called\n Join Implication Reasoning\n (JIR). Our core insight is that the execution results of one or more join types allow us to infer the oracle of a semantic-related target join type. As an illustration, the execution results of\n Inner Join\n and\n Anti Join\n can be combined to infer the expected result of\n Left Join\n performed over the same left-hand side and right-hand side under the same join predicate. JIR validates joins by exploiting the join semantics of each individual DBMS itself for reasoning, and finds join logic bugs if the execution result of the target join fails to match the inferred oracle. We realized our approach and evaluated it on 11 extensively tested DBMSs: SQLite, MySQL, CockroachDB, ClickHouse, DuckDB, TiDB, MonetDB, Umbra, Dolt, CrateDB and PostgreSQL. Overall, JIR found 100 unique, previously unknown join bugs, of which 91 were fixed and 9 were verified by DBMS vendors, with 69 being join logic bugs. We expect that the generality and practicality of our approach will make it widely adoptable for understanding and testing\n Join.",
+      "cites_seed_techniques": [
+        "norec",
+        "qpg",
+        "dqp",
+        "coddtest",
+        "sqlancer_pp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "qpg",
+            "dqp",
+            "coddtest",
+            "sqlancer_pp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/875f8fff2ecc34bfe8532ad93c0c4a536f6e9e45",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Query Plan Guidance (QPG), Differential Query Plans (DQP), Constant-Optimization-Driven Testing (CODDTest), sqlancer_pp.",
+              "retrieved_at": "2026-09-10T14:52:29Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T14:52:29Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T14:52:29Z",
+        "source_url": "https://www.semanticscholar.org/paper/875f8fff2ecc34bfe8532ad93c0c4a536f6e9e45",
+        "source_type": "paper",
+        "content_sha256": "sha256:338389a8be5b057f4c590304efcf77f2d531d63f9d04d6b71aae08cccd6f8928"
+      },
+      "is_sqlancer_publication": true
+    },
+    {
+      "id": "paper:doi:10.1145/3802038",
+      "title": "Dialect-Agnostic SQL Parsing via LLM-Based Segmentation",
+      "authors": [
+        "Junwen An",
+        "Kabilan Mahathevan",
+        "Manuel Rigger"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Management of Data",
+      "doi": "10.1145/3802038",
+      "arxiv_id": "2603.16155",
+      "s2_paper_id": "355d5c571beff1edaec3a92f48d3060339b07694",
+      "url": "https://doi.org/10.1145/3802038",
+      "open_access_pdf": "https://arxiv.org/pdf/2603.16155",
+      "abstract": "SQL is a widely adopted language for querying data, which has led to the development of various SQL analysis and rewriting tools. However, due to the diversity of SQL dialects, such tools often fail when encountering unrecognized dialect-specific syntax. While Large Language Models (LLMs) have shown promise in understanding SQL queries, their inherent limitations in handling hierarchical structures and hallucination risks limit their direct applicability in parsing. To address these limitations, we propose SQLFlex, a novel query rewriting framework that integrates grammar-based parsing with LLM-based segmentation to parse diverse SQL dialects robustly. Our core idea is to decompose hierarchical parsing to sequential segmentation tasks, which better aligns with the strength of LLMs and improves output reliability through validation checks. Specifically, SQLFlex uses clause-level segmentation and expression-level segmentation as two strategies that decompose elements on different levels of a query. We extensively evaluated SQLFlex on both real-world use cases and in a standalone evaluation. In SQL linting, SQLFlex outperforms SQLFluff in ANSI mode by 63.68% in F1 score while matching its dialect-specific mode performance. In test-case reduction, SQLFlex outperforms SQLess by up to 10 times in simplification rate. In the standalone evaluation, it parses 91.55% to 100% of queries across eight distinct dialects, outperforming all baseline parsers. We believe SQLFlex can serve as a foundation for many query analysis and rewriting use cases.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp",
+        "sqlancer_pp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp",
+            "sqlancer_pp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3802038",
+              "source_type": "paper_citation_context",
+              "excerpt": "Under this formulation, a wide range of applications beyond linting involve query rewriting, such as query reduction [50], DBMS testing [35, 53, 71], and SQL grading [9, 10]. dialect-agnostic parsing effectiveness, using queries in eight different SQL dialects extracted from their respective DBMS…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5eada14bac587bf38a04e696c2f33af521add020bca432f47be3165d96316b08",
+              "retrieved_at": "2026-09-10T14:52:29Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T14:52:29Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3802038",
+              "source_type": "paper_citation_context",
+              "excerpt": "The resulting bug reports typically contain complex expressions [49, 70, 96] that make it difficult for developers to fix potential bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing sqlancer_pp, as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:08748e43b75553d7d614da120579cf95b658b8568c12214788a93042a48c8d8b",
+              "retrieved_at": "2026-09-10T14:52:29Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T14:52:29Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3802038",
+              "source_type": "paper_citation_context",
+              "excerpt": "Some approaches, like EET [35], AMOEBA [53], and SQLancer [70, 71], generate equivalent queries to check for consistent results.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4f680a72389ccb1aae734cdf433866b6e83955771fdf81c522229da5a7bbddaa",
+              "retrieved_at": "2026-09-10T14:52:29Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T14:52:29Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T14:52:29Z",
+        "source_url": "https://www.semanticscholar.org/paper/355d5c571beff1edaec3a92f48d3060339b07694",
+        "source_type": "paper",
+        "content_sha256": "sha256:c1b06cc41d542f52c15e2d7a756d66d92e64ff809c3bf28741237bb1b7034777"
+      },
+      "is_sqlancer_publication": true
+    },
+    {
+      "id": "paper:doi:10.1145/3810991.3811632",
+      "title": "DIRT: Database-Integrated Random Testing",
+      "authors": [
+        "A. Keles",
+        "Ethan Chou",
+        "Harrison Goldstein",
+        "Leonidas Lampropoulos"
+      ],
+      "year": 2026,
+      "venue": "DBTest@SIGMOD",
+      "doi": "10.1145/3810991.3811632",
+      "arxiv_id": "2604.16373",
+      "s2_paper_id": "7f0ecde1aadab94fe2464b9a8e4183e68159d931",
+      "url": "https://doi.org/10.1145/3810991.3811632",
+      "open_access_pdf": "https://doi.org/10.1145/3810991.3811632",
+      "abstract": "Database management systems (DBMSs) are notoriously complex, making them difficult to test effectively, especially during early development when many features are incomplete. Traditional testing tools like SQLancer and SQLSmith are highly effective for mature databases, but they struggle with high false positive rates and low actionability when applied to evolving systems. We present DIRT, a paradigm designed specifically for testing databases during development, which integrates a testing framework directly into the DBMS, enabling the random testing process to evolve in tandem with the system and reducing false positives by construction. We introduce generation actions, an abstraction for allowing database developers rather than testing experts to specify correctness properties. We evaluate DIRT on Turso, an actively developed SQLite-compatible OLTP engine, and show that it finds 23 unique, confirmed bugs–significantly outperforming off-the-shelf SQLancer variants in terms of true positive rate and usefulness of bug reports. Our results demonstrate that embedding testing infrastructure within the DBMS can dramatically improve its effectiveness and usability during development.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp",
+        "coddtest"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert",
+            "dqp",
+            "coddtest"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811632",
+              "source_type": "paper_citation_context",
+              "excerpt": "2b presents the GA for WHERE Extended case of Ternary Logic Partitioning (TLP) [21] oracle from SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c86f0b4aeef7cdae079ee0856729167966a501439ee1d875390a45232a4b9423",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811632",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer currently supports Query Plan Guidance (QPG) [1] for feedback-guided generation, Cardinality Estimation Restriction Testing (CERT) [2] for finding performance bugs in DBMSs, Differential Query Plans (DQP) [3] for detecting bugs in join optimizations, and Constant Optimization Driven…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:997f44339efc1dee0948b1f7ba6dabe955ab3d2b9f300234d228d673485faf7e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811632",
+              "source_type": "paper_citation_context",
+              "excerpt": "…generation, Cardinality Estimation Restriction Testing (CERT) [2] for finding performance bugs in DBMSs, Differential Query Plans (DQP) [3] for detecting bugs in join optimizations, and Constant Optimization Driven Database System Testing (CODDTest) [26] for finding logic bugs in…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e4eef6960372f873227112cd69dca64fdc7de14a9ae75d4e9e78a6f6ed6318ee",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811632",
+              "source_type": "paper_citation_context",
+              "excerpt": "We opted for the shadow state because it allows for complex reasoning over the database state for constructing arbitrary queries and properties, and because it gives us a canonical property over the database state: the shadow state is identical to the database at any given moment.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:30ffcd7acd0c0bcc466068342f066af478d58e24f84fd1858fe20faec5ede8ce",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811632",
+              "source_type": "paper_citation_context",
+              "excerpt": "Two metamorphic oracles followed: Non-Optimizing Reference Engine Construction (NoREC) [20], which found 51 optimization bugs, and Ternary Logic Partitioning (TLP) [21], which discovered 77 novel logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:48bbce6cd82013f32106efc0cb0fba812f20af041d4aa366a238d0a601affc85",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3810991.3811632",
+              "source_type": "paper_citation_context",
+              "excerpt": "It started with Pivoted Query Synthesis (PQS) [23], a rather \"simple\" containment property over databases that has found at least 121 unique logic bugs in production databases.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e6ddd2ebeb2e51f051ca39f04fdc75f4285f0f51b63e98bb3200dfa56d27da86",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3810991_3811632.json",
+              "source_type": "paper",
+              "excerpt": "1 Definitions of Oracles in DIRT In this subsection, we go over definitions of five different oracles written as sequences of generation actions, three of which are reimplementations of existing oracles in SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M19 in the paper's extracted text, 3.1 Definitions of Oracles in DIRT, page 3.",
+              "content_sha256": "sha256:36ae1b65c8ce2cb33b41d1327372855f96c500e3a607402a53ef77ee9e7e64da",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3810991_3811632.json",
+              "source_type": "paper",
+              "excerpt": "We start by defining the first of the three SQLancer oracles we implemented, Pivoted Query Synthesis (PQS) [ 23], as a universally quantified proposition in Fig.",
+              "excerpt_is_verbatim": true,
+              "note": "M20 in the paper's extracted text, 3.1 Definitions of Oracles in DIRT, page 3.",
+              "content_sha256": "sha256:c0beb8d6e34740c8392014457013d780858496e718e91ee14cc1e5892a9a4494",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3810991_3811632.json",
+              "source_type": "paper",
+              "excerpt": "2b presents the GA for WHERE Extended case of Ternary Logic Partitioning (TLP) [ 21] oracle from SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M25 in the paper's extracted text, 3.1 Definitions of Oracles in DIRT, page 3.",
+              "content_sha256": "sha256:aed86577a5dc64c2a5d69dda0b365f891f5c1410347f83b7a8f8df1b627932cd",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3810991_3811632.json",
+              "source_type": "paper",
+              "excerpt": "We can express other properties that are not present in SQLancer, including those fundamental to key-value stores such asDeleted rows should not be in the tablepresented in Fig.",
+              "excerpt_is_verbatim": true,
+              "note": "M26 in the paper's extracted text, 3.1 Definitions of Oracles in DIRT, page 3.",
+              "content_sha256": "sha256:a7fe737dafc6d7f37a5b71cd6e31cc1345dc9aae2251fee07908b1daf7c92c7c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3810991_3811632.json",
+              "source_type": "paper",
+              "excerpt": "We evaluate DIRTon Turso, an actively developed SQLite-compatible OLTP engine, and show that it finds 23 unique, confirmed bugs–significantly outperforming off-the-shelf SQLancer variants in terms of true positive rate and usefulness of bug reports.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, page 1.",
+              "content_sha256": "sha256:75c258a606df619de55ac9a8476208f59bdfd0c383b05db12ba0f25ced536770",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3810991_3811632.json",
+              "source_type": "paper",
+              "excerpt": "This does not mean that DIRThas better generators or oracles than SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M39 in the paper's extracted text, 4.2 RQ2: How does DIRTcompare to SQLancer?, page 5.",
+              "content_sha256": "sha256:4d89c4e223b5f2dc17b2bb8d989058b4a1ff1e62fbc6540558d7564119b75eb8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3810991_3811632.json",
+              "source_type": "paper",
+              "excerpt": "We compare the performance of DIRT for Turso against SQLancer with minimal modifications, which begets some questions worth discussing.",
+              "excerpt_is_verbatim": true,
+              "note": "M40 in the paper's extracted text, 5 Discussion, page 5.",
+              "content_sha256": "sha256:bed8078b0579c8f6a07828669f69c04f16e99b042f1db3d65e5aa92d462d3db2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3810991_3811632.json",
+              "source_type": "paper",
+              "excerpt": "Within prior work, SQLancer is distinguished by using specialized test oracles that reveal logic bugs in DBMSs, and it has proven extremely successful at finding bugs in mature databases.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, 1 Introduction, page 1.",
+              "content_sha256": "sha256:3001664e7d35907bd76ee088e24308017fb06cfdab8c5fa2483fd7b2dea0c4d2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/7f0ecde1aadab94fe2464b9a8e4183e68159d931",
+        "source_type": "paper",
+        "content_sha256": "sha256:8fbe33d5f9ea0dcafef0050e0f0e4c4ed6ca71a416a15675a81c2b3ac6da616a"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/tkde.2026.3656491",
+      "title": "Effective Bug Detection in Graph Database Engines: An LLM-Based Approach",
+      "authors": [
+        "Jiayi Wu",
+        "Zhengyu Wu",
+        "Xunkai Li",
+        "Ronghua Li",
+        "Hongchao Qin",
+        "Guoren Wang"
+      ],
+      "year": 2026,
+      "venue": "IEEE Transactions on Knowledge and Data Engineering",
+      "doi": "10.1109/tkde.2026.3656491",
+      "arxiv_id": "2402.00292v1",
+      "s2_paper_id": "51c1741064427aadb230a3ce760368f62577dbae",
+      "url": "https://doi.org/10.1109/tkde.2026.3656491",
+      "open_access_pdf": null,
+      "abstract": "Graph database engines play a pivotal role in efficiently storing and managing graph data across various domains, including bioinformatics, knowledge graphs, and recommender systems. Graph databases must be accurate because errors lead to faulty analysis. Current bug-detection approaches are confined to specific graph query languages, limiting their applicabilities when handling graph database engines that use various graph query languages across various domains. Moreover, they require extensive prior knowledge to generate queries for detecting bugs. To address these challenges, we introduce DGDB, a novel paradigm harnessing large language models (LLM), such as ChatGPT, for comprehensive bug detection in graph database engines. DGDB leverages ChatGPT to generate high-quality queries for different graph query languages. It subsequently employs differential testing to identify bugs in graph database engines. We applied this paradigm to graph database engines based on Cypher, Gremlin, and SPARQL, and detected a total of 23 previously unknown wrong-result bugs. DGDB achieves at least 20.41% improvement in the non-empty-result query ratio and detects more than three times as many bugs as existing state-of-the-art methods on Cypher-based graph database engines, with further significant gains when employing more advanced LLM.",
+      "cites_seed_techniques": [
+        "norec",
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/tkde.2026.3656491",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS [35] selects a target data from randomly generated tables, generates conditional expressions based on the target data, constructs an SQL query with a where or join clause, and determines the presence of bugs by checking if the result is included in the result set.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e72b9a97ee8d20d4fcddd55f96ca56d3ad1a4ee12d1b7a6c544bd8d143206396",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tkde.2026.3656491",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [37] transforms randomly generated original SQL queries into three different logical queries based on the true, false, and null ternary logic.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bfe36558db1d83becb4b1acbd71600c204437bed6de5e916b5222e96f6646fcd",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tkde.2026.3656491",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoRec [36] converts the original SQL query into a non-optimized SQL query and compares the results of these two SQL queries for consistency.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7eeb71fb1c18ef118a45fb19c376de086f48f63d489e338aa640e8655a11ae09",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2402.00292",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS[38] selects a target data from randomly generated tables, generates conditional expressions based on the target data, constructs an SQL query with a where or join clause, and determines the presence of bugs by checking if the result is included in the result set.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:43718a6939b1bf0235c99d44abfeab357d5a772a45ce3d7abea2539154d48eef",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2402.00292",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP[37] transforms randomly generated original SQL queries into three different logical queries based on the true, false, and null ternary logic.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:13e42c426f74c952c1b584b0843dc8a75a821b45d3b8804943d1a8345215e17e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2402.00292",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoRec[36] converts the original SQL query into a non-optimized SQL query and compares the results of these two SQL queries for consistency.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:47fb80cc32532a6bcbc598f0c4b432baadf94031d145c4654433d8b72d40db50",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2402.00292",
+              "source_type": "paper_citation_context",
+              "excerpt": "The former typically involves detecting bugs in graph database engines by generating equivalent queries[5] or utilizing predicate partitioning[23, 24, 37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3339cae174f31cf935c475140c9a4ae5fb20e3e35961aa0aef5a25d767112224",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/51c1741064427aadb230a3ce760368f62577dbae",
+        "source_type": "paper",
+        "content_sha256": "sha256:e4afde4939b27589bbd3045d56c909f23ec03469e0b27e2d92bbc512db634fe2"
+      },
+      "also_indexed_as": [
+        "paper:arxiv:2402.00292"
+      ]
+    },
+    {
+      "id": "paper:doi:10.14722/ndss.2026.240198",
+      "title": "Efficiently Detecting DBMS Bugs through Bottom-up Syntax-based SQL Generation",
+      "authors": [
+        "Yu Liang",
+        "Peng Liu"
+      ],
+      "year": 2026,
+      "venue": "Network and Distributed System Security Symposium",
+      "doi": "10.14722/ndss.2026.240198",
+      "arxiv_id": null,
+      "s2_paper_id": "dcc3bf3a8219e17b5990057d6185f7007da89f88",
+      "url": "https://doi.org/10.14722/ndss.2026.240198",
+      "open_access_pdf": "https://doi.org/10.14722/ndss.2026.240198",
+      "abstract": "—Syntax-based testing is a promising technique for finding bugs in Database Management Systems (DBMSs). All existing syntax-based SQL generation tools apply a Top-down generation method. To construct a SQL query (syntax tree), the generator forward explores the SQL grammar starting from the root node, and it stops when no further grammar rule can be applied to the leaves of the syntax tree. However, the Top-down generation method tends to put more effort into exploring the shallow grammar close to the root and neglects the feature-rich grammar deeper in the grammar space. Therefore, it is not efficient in finding DBMS bugs. This paper proposes a new Bottom-up syntax-based SQL generation technique that puts more testing resources into exploring the feature-rich grammar rules. The exploration of SQL grammar begins with one interesting grammar rule that outlines the syntax of feature-rich SQL functionalities. The generator then backtracks (Bottom-up) this grammar rule to the root to create a syntax path that unveils this interesting grammar. Multiple Bottom-up generated syntax paths are then expanded and merged to create diverse SQL queries for fuzzing. A prototype tool, SQLBull , adopts the Bottom-up generation technique for fuzzing. In the evaluation, SQLBull found 63 zero-day bugs from 5 well-tested DBMSs: MySQL , MariaDB , CockroachDB , DuckDB , and PostgreSQL . It outperforms all existing tools in both bug-finding and code coverage. The evaluation results verify the effectiveness of the Bottom-up generation technique.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "qpg",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "qpg",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14722/ndss.2026.240198",
+              "source_type": "paper_citation_context",
+              "excerpt": "In this evaluation, we include two of the latest and most advanced testing techniques implemented in SQLancer which are capable of detecting DBMS memory errors, i.e., SQLancer +QPG [52] and SQLancer +DQP [53].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:382e4e31077c5a7a43d3e6cb897cac8bcb8e13b5fca8fd417fc6a0e787720d01",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14722/ndss.2026.240198",
+              "source_type": "paper_citation_context",
+              "excerpt": "Although SQLancer was first introduced in 2022, it has become the most popular platform for implementing the latest SQL testing techniques [52]–[56].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:703c5fa0a19b11959b4c30716f72b4f4d31faf7b53f0377ebab2ef86f715b65f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14722/ndss.2026.240198",
+              "source_type": "paper_citation_context",
+              "excerpt": "Its latest improvement, SQLancer +QPG [52] and SQLancer +DQP [53], leverage the DBMS query plan and optimization hints to guide the query generation.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:518699946d75e762b0affe7aafc9a292042851f60f85ddcb55a0e0edf5124e95",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14722/ndss.2026.240198",
+              "source_type": "paper_citation_context",
+              "excerpt": "We first discuss the existing syntax-based testing tools for the DBMS testing community [30], [39], [52] in §VI-A and §VI-B.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e78acb369c35f4cbc9af746377bc2ccc961a8a259c288db8297d3635ed1e415b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14722/ndss.2026.240198",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [30] is another template-based DBMS testing tool that detects DBMS memory and logic bugs [54]–[56].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9d2e14643d82fa44f090a813091b54cdb50a77884ace834a5af6379848d2c135",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "qpg",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14722_ndss_2026_240198.json",
+              "source_type": "paper",
+              "excerpt": "In this evaluation, we include two of the latest and most advanced testing techniques implemented in SQLancer which are capable of detecting DBMS memory errors, i.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, page 8.",
+              "content_sha256": "sha256:dd0b57559335ed9850e1c2cd7f1f122817f73a5ec9276e6f73ed4dd9f2a77074",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14722_ndss_2026_240198.json",
+              "source_type": "paper",
+              "excerpt": ", SQLancer +QPG[52] and SQLancer +DQP[53].",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, page 8.",
+              "content_sha256": "sha256:9eabac5163d9d95ecd9b2effeff9b79a444af48d6553e1dc6f80df7ce1859a30",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14722_ndss_2026_240198.json",
+              "source_type": "paper",
+              "excerpt": "We compare SQLBull with SQLancer +QPG onCockroachDB, and SQLancer +DQPonMySQL and MariaDB.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, page 8.",
+              "content_sha256": "sha256:4e14dc37a5462882cfa10d02232269472a6ebe7c4db5c4dbcb1615a2d530a6e2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14722_ndss_2026_240198.json",
+              "source_type": "paper",
+              "excerpt": "For completeness, we also include the statistics of these SQL features from the queries generated by other testing tools, including mutation-based fuzzer Squirrel [39] and templatebased SQL generator SQLancer [30].",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, page 8.",
+              "content_sha256": "sha256:f7ae285c78fd834fc1110d911b8c25d20cc8090a104c1f5953a9d71c55d7d337",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14722_ndss_2026_240198.json",
+              "source_type": "paper",
+              "excerpt": "The most advanced tool in this category is SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, page 8.",
+              "content_sha256": "sha256:a7d840530e40d4eed5d0e0128e756b167797d73670cb8c3efe81100a7026f897",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14722_ndss_2026_240198.json",
+              "source_type": "paper",
+              "excerpt": "Although SQLancer was first introduced in 2022, it has become the most popular platform for implementing the latest SQL testing techniques [52]–[56].",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, page 8.",
+              "content_sha256": "sha256:703c5fa0a19b11959b4c30716f72b4f4d31faf7b53f0377ebab2ef86f715b65f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14722_ndss_2026_240198.json",
+              "source_type": "paper",
+              "excerpt": "SQLancer +QPG and SQLancer +DQP are the latest and most advanced DBMS testing techniques implemented on top of SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, page 9.",
+              "content_sha256": "sha256:92476ea16f4dbedab5dbbc26bede07f8bbba06556e9a0dce796db6d09366e2aa",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/dcc3bf3a8219e17b5990057d6185f7007da89f88",
+        "source_type": "paper",
+        "content_sha256": "sha256:72cacd6a79bde1670b5d81027fe86d863845cc47f11978b9f7c8dea274a8ea2b"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3808109",
+      "title": "Eidolon: Perform Noise-Aware Fuzzing on FHE Libraries via Equivalence Expression Transformation",
+      "authors": [
+        "Zhensheng Xian",
+        "Zhen Yan",
+        "Yuanliang Chen",
+        "Xuelian Cao",
+        "Fuchen Ma",
+        "Dalong Shi",
+        "Yu Jiang"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Software Engineering",
+      "doi": "10.1145/3808109",
+      "arxiv_id": null,
+      "s2_paper_id": "2dd21a6a9998e76f0247f28e7d30938c66112b8a",
+      "url": "https://doi.org/10.1145/3808109",
+      "open_access_pdf": "https://doi.org/10.1145/3808109",
+      "abstract": "Ensuring data privacy during computation is a critical challenge in many security systems. Fully Homomorphic Encryption (FHE) addresses this gap by enabling multiple operations on encrypted data without decryption, thus ensuring privacy is preserved throughout computation. However, existing cryptographic testing tools are unable to test the core functionality of FHE, which is the execution of computations on encrypted data. They are expertly designed to generate structured data for testing cryptographic algorithms. This structural mismatch, combined with a lack of awareness of FHE-specific noise management, leads them to generate invalid test inputs that fail to probe FHE libraries’ core logic. To address this gap, we propose Eidolon, a noise-aware fuzzer. It directs mutations toward arithmetic expressions that explore the computational space defined by the noise budget. As its test oracle, Eidolon leverages Equivalence Expression Transformation, which transforms a standard arithmetic expression into two mathematically identical but structurally different forms (e.g., Factored, Horner) to detect inconsistencies in their outputs. We evaluated Eidolon on SEAL, OpenFHE, HElib, and TFHE. Compared with existing cryptographic and grammar-based fuzzers, Eidolon achieves 28.7%, 45.5%, 75.6%, and 37.6% higher final code coverage than CLFuzz, Cryptofuzz, CDF, and Peach, respectively. In total, Eidolon uncovered 20 previously unknown bugs, 13 of which have been fixed and 12 assigned CVEs.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3808109",
+              "source_type": "paper_citation_context",
+              "excerpt": "For instance, metamorphic fuzz testing [28] combines MT with fuzzing to detect faults in autonomous driving systems, QFuzz [36] uses execution costs for side-channel leakage quantification, and SQLancer [41] employs a containment oracle for DBMS logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b66c7643f753c2919f1ea86f686af3653386f4eed9c1912c04aabb96369d039a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/2dd21a6a9998e76f0247f28e7d30938c66112b8a",
+        "source_type": "paper",
+        "content_sha256": "sha256:482b485a4594d5bb33ccedc71a980c0ffbc623ac3a17149751f260a74f6e8115"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3802053",
+      "title": "EPSC: Testing Database Management Systems via Equivalent Prepared Statement Construction",
+      "authors": [
+        "Chi Zhang",
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Dalong Shi",
+        "Linzhang Wang",
+        "Yu Jiang"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Management of Data",
+      "doi": "10.1145/3802053",
+      "arxiv_id": null,
+      "s2_paper_id": "f42307b2aae0cdc8c3fae609e844622539b26baa",
+      "url": "https://doi.org/10.1145/3802053",
+      "open_access_pdf": "https://doi.org/10.1145/3802053",
+      "abstract": "Database Management Systems (DBMSs) serve as the backbone for efficient data access and modification through the Structured Query Language (SQL). Bugs in DBMS implementations may prevent otherwise valid statements from executing or even cause statements to return incorrect results. Prior research has primarily focused on detecting bugs in ordinary SQL statements. In contrast, prepared statements—a language feature widely used in production environments to improve the performance of repeated queries and to guard against SQL injection—have received far less attention, and the potential bugs within them remain insufficiently explored.\n \n In this paper, we present a general black-box approach, termed\n E\n quivalent\n P\n repared\n S\n tatement\n C\n onstruction (EPSC), to detect logic bugs in both ordinary and prepared SQL statements. The key insight of EPSC is that Data Manipulation Language (DML) and Query Definition Language (QDL) statements can be executed in two equivalent forms—ordinary statements and prepared statements— which should exhibit consistent behavior and produce identical results. For instance, a SELECT statement can be transformed into its prepared statement form by extracting literal values as bound parameters; both forms are expected to yield the same output. Any inconsistency between them indicates the presence of a bug in the target DBMS. To evaluate the effectiveness of EPSC, we applied it to seven mature DBMSs: MySQL, MariaDB, TiDB, PostgreSQL, CockroachDB, SQLite3, and DuckDB. In total, EPSC uncovered 49 unique bugs, of which 31 have been confirmed and 10 have already been fixed. Moreover, our experimental results demonstrate that EPSC effectively detects logic bugs that existing approaches fail to identify. We believe that the simplicity and broad applicability of EPSC can significantly enhance the reliability of DBMS implementations.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp",
+        "coddtest"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp",
+            "qpg",
+            "dqp",
+            "coddtest"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/f42307b2aae0cdc8c3fae609e844622539b26baa",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Query Plan Guidance (QPG), Differential Query Plans (DQP), Constant-Optimization-Driven Testing (CODDTest).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3802053.json",
+              "source_type": "paper",
+              "excerpt": "We reused the statement generators from SQLancer [ 27] for statement generation, which are manually written and rule-based.",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, 4 Evaluation, page 11.",
+              "content_sha256": "sha256:93021061f944e50cc1d8d80953b874f2479d3980728fbbbc9f45759142407954",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "norec",
+            "tlp",
+            "coddtest"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3802053.json",
+              "source_type": "paper",
+              "excerpt": "For our evaluation, we chose NoREC [ 25], TLP [ 26], DQE [ 31], and CODDTest [ 35] as baseline approaches.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, 4.3 Compared with Other Approaches, page 18.",
+              "content_sha256": "sha256:7781d89c1db6b667cb3b6d3527adaf5dd7c0732ed2d96605bd75b72f3f7fc155",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3802053.json",
+              "source_type": "paper",
+              "excerpt": "Specifically, NoREC and DQE represent the class of approaches that detect logic bugs by leveraging clause equivalence; TLP represents the set-relation–based oracle design.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, 4.3 Compared with Other Approaches, page 18.",
+              "content_sha256": "sha256:af4ea1dfbc503c1e76ae248efbb1677ee805a2f6e8a75d628396a758b8a04781",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/f42307b2aae0cdc8c3fae609e844622539b26baa",
+        "source_type": "paper",
+        "content_sha256": "sha256:fd1221e8203dde301db08e3bfdea0ab91cfbf153ed27fd20d36a97457f38fc08"
+      }
+    },
+    {
+      "id": "paper:arxiv:2608.23402",
+      "title": "EXPLAIN Yourself! Finding Query Planner Stalls Across DBMSes",
+      "authors": [
+        "Geoffrey X. Yu",
+        "Ryan Marcus",
+        "Tim Kraska"
+      ],
+      "year": 2026,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": "2608.23402",
+      "s2_paper_id": "5def00bebb723d615e84104a65800e1aa21f3743",
+      "url": "https://arxiv.org/abs/2608.23402",
+      "open_access_pdf": null,
+      "abstract": "Query planners are typically expected to produce optimized plans quickly, leading many researchers (including the authors of this paper) and practitioners to design systems that assume query planning is a low-cost operation. Using a lightweight agentic search, we show that this assumption does not always hold. Across seven DBMSes, including four commercial systems, we find at least one query per system that takes more than three minutes to plan. In addition to being slow to plan, such queries risk tying up database resources without performing useful work, creating a potential denial-of-service vector. We analyze the queries our search uncovers and compare how the seven systems respond to each pattern. We find that although the queries triggering slow planning are largely DBMS-specific, recurring pathologies involving correlated subqueries, CTE expansion, repeated subquery expressions, disjunctive joins, and constant folding affect multiple systems. We release our uncovered queries along with a curated suite of parameterized query pathologies that researchers and database engineers can use to test planner robustness. Overall, our results show that query planning cannot always be treated as a predictably inexpensive operation and that its latency and robustness deserve further attention from both database researchers and engineers.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2608.23402",
+              "source_type": "paper_citation_context",
+              "excerpt": "The high-level idea is to identify a general property about a pair of queries (e.g., making a query more restrictive should result in an output cardinality estimate no larger than the original query [4]).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f8244d45c3764538932e07ea5013d8de0bcf6bb1e0d47203e51d4431db0a1ca2",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2608.23402",
+              "source_type": "paper_citation_context",
+              "excerpt": "A complementary set of papers propose techniques to find correctness bugs in database systems [3, 5, 16, 27–29, 40] , applying a similar approach grounded in query",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f3bc5c5c142150eb5643bedeb60082b5991675249d9a6ee2214bf5bb584ed50f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2608.23402",
+              "source_type": "paper_citation_context",
+              "excerpt": "Prior works propose techniques for finding query execution performance bugs in database systems [4, 11, 12, 15, 35, 36].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:82e62f5294883ad279e166653df3ff87e2b28b50c4b20d45ce7d98030dcd504f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/5def00bebb723d615e84104a65800e1aa21f3743",
+        "source_type": "paper",
+        "content_sha256": "sha256:410a7f7f66292352188792a5ae02a3cd9a826274edf907b6579fbc961c66d6e7"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3802061",
+      "title": "Finding Missed Optimizations in DBMSs through Unbalanced Short-circuit Query Construction",
+      "authors": [
+        "Jinhui Lai",
+        "Chi Zhang",
+        "Jie Liang",
+        "Zihao Zeng",
+        "Zhiyong Wu",
+        "Jingzhou Fu",
+        "Chijin Zhou",
+        "Shuai Ma",
+        "Yu Jiang",
+        "Zichen Xu"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Management of Data",
+      "doi": "10.1145/3802061",
+      "arxiv_id": null,
+      "s2_paper_id": "96bb90109a01a144448d9bf9b7b58f3e4c440eff",
+      "url": "https://doi.org/10.1145/3802061",
+      "open_access_pdf": "https://doi.org/10.1145/3802061",
+      "abstract": "DBMSs underpin modern data-intensive applications, where performance directly impacts system responsiveness and user experience. To improve efficiency, DBMSs employ many sophisticated optimization techniques for SQL queries. However, the complexity of SQL queries and DBMS architectures often causes implementations to miss potential optimization opportunities, resulting in performance degradation, inefficient resource utilization, and diminished user experience. Identifying these missed optimizations is challenging due to the intricate interactions among query semantics, optimizer decisions, and execution behaviors. Existing approaches have detected many performance issues when implemented optimizations perform poorly. However, they rarely reveal optimizations that were entirely missed, leaving many performance gaps unaddressed.\n In this paper, we present SCor, a black-box approach for identifying missed optimizations in DBMSs through unbalanced short-circuit query construction. Our key insight is that the results of many queries can be determined without full execution, yet DBMSs still execute the entire query, revealing missed optimizations. SCor realizes it by constructing short-circuit queries, where the result can be obtained from low-cost operations alone. If the DBMS still executes the full query with high-cost operations, it indicates missed optimizations. Since short-circuit patterns are prevalent in SQL queries and can be flexibly embedded into diverse query structures, SCor can be systematically applied to expose missed optimizations across a wide range of queries. Our evaluation of SCor across 11 widely-used DBMSs, revealed 153 previously undetected performance bugs resulting from missed optimizations, including 2 bugs in Oracle and 3 bugs in PostgreSQL. Among all reported bugs, 125 have been confirmed by developers, and 33 have already been fixed.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/96bb90109a01a144448d9bf9b7b58f3e4c440eff",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Cardinality Estimation Restriction Testing (CERT).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3802061.json",
+              "source_type": "paper",
+              "excerpt": "We compared SCor with five state-of-the-art DBMS performance bug detection tools, including: APOLLO [ 18], AMOEBA [ 24], and CERT [ 1], Puppy [ 54], and HULK [ 55].",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, 5.4 Comparative Evaluation, page 19.",
+              "content_sha256": "sha256:39189f4eff669a1f39a42a2fb173be783c0ce3a2086f2296d79c7e8b589c03da",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3802061.json",
+              "source_type": "paper",
+              "excerpt": "The results show that SCor found 24, 8, 36, 8, and 4 more bugs than APOLLO, AMOEBA, CERT, Puppy, and HULK, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, 5.4 Comparative Evaluation, page 19.",
+              "content_sha256": "sha256:ebd366b9d4331b448177fa856e5fa8ff0075bc790a71265027edaa19a61c0bfc",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3802061.json",
+              "source_type": "paper",
+              "excerpt": "DBMSAPOLLO AMOEBA CERT Puppy HULK SCor PostgreSQL 0 0 - 2 2 3 MySQL - - 3 6 10 13 CockroachDB-1 2 - - 6 TiDB - - 4 - - 26 SQLite 0 - - - - 21 Total 0 1 9 8 12 69 Increment 24↑8↑36↑8↑4↑data-sensitive performance anomalies through data changes, while SCor employs unbalanced shortcircuit queries.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 5.4 Comparative Evaluation, page 20.",
+              "content_sha256": "sha256:659a0d1b673da3055f3595012d83f8ab43239abc7250d25e9b18acd05c4541b6",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3802061.json",
+              "source_type": "paper",
+              "excerpt": "We compared SCor with five state-of-the-art DBMS performance bug detection tools, including: APOLLO [ 18], AMOEBA [ 24], and CERT [ 1], Puppy [ 54], and HULK [ 55].",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, 5.4 Comparative Evaluation, page 19.",
+              "content_sha256": "sha256:39189f4eff669a1f39a42a2fb173be783c0ce3a2086f2296d79c7e8b589c03da",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/96bb90109a01a144448d9bf9b7b58f3e4c440eff",
+        "source_type": "paper",
+        "content_sha256": "sha256:3f341cf89ab96e82016f26c4e819cd2a8bff1c246e5795d50ad0f2335fdf15f2"
+      }
+    },
+    {
+      "id": "paper:arxiv:2605.22992",
+      "title": "Finding Performance Issues in Database Systems by Exploiting Dormant Code Paths",
+      "authors": [
+        "Jinsheng Ba",
+        "Zhendong Su"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2605.22992",
+      "s2_paper_id": "652ecd8215728ede7bbecba4ac5049ac96af9e91",
+      "url": "https://arxiv.org/abs/2605.22992",
+      "open_access_pdf": null,
+      "abstract": "Performance is a critical characteristic of fundamental systems, such as Database Management Systems (DBMSs). Both academia and industry have invested decades in exploring efficient optimization algorithms. Despite these efforts, DBMSs are prone to performance issues, which incur suboptimal performance. Finding such issues is a longstanding challenge as no ground-truth performance is available. Existing work adopts black-box methods to examine performance consistency across executions, but cannot systematically test optimizations. In this work, we propose a novel, general white-box methodology, Branch Flip Analysis (BFA), to systematically and effectively uncover performance issues. BFA flips code branches to enforce or disable an optimization, and the performance is expected to be not significantly better. Otherwise, a performance issue exists. BFA provides a new perspective to finding performance issues and testing optimization logics in a fine-grained manner. We realized BFA in a prototype system QueryZen, and evaluated it on four widely-used and mature DBMSs: PostgreSQL, MySQL, CockroachDB, and MariaDB. QueryZen found 21 previously unknown and unique performance issues with the workload of the extensively used benchmarks TPC-H and TPC-DS. The core concept of BFA is simple and broadly applicable, and can be adapted to analyze the performance of other software systems.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "cert",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2605.22992",
+              "source_type": "paper_citation_context",
+              "excerpt": "Other approaches focus on discovering logic bugs in DBMSs. Oracles such as PQS [29], NoREC [27], and TLP [28] detect logic bugs in SELECT statement implementations, while DQE [31] targets logic errors in UPDATE and INSERT statements.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9744db965b70a1f706f432bd229030881568181d7fc21e6c8bb4331ee3ed3b31",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2605.22992",
+              "source_type": "paper_citation_context",
+              "excerpt": "CERT [1] finds performance issues by finding inconsistent cardinality estimation, which is typically deemed as the most critical component for query optimization [17].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7ba0fc695029b3d555a6f2ddd9f549e3d3713d133a7c1dde0eb0ec92566f9e0d",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2605.22992",
+              "source_type": "paper_citation_context",
+              "excerpt": "We considered the performance issues found by the four existing performance-testing methods: APOLLO [16], AMOEBA [21], CERT [1], and PUPPY [40].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6e54dac01dd8c4281405eae803aa9d1f9fff0cc92260221ae16523956c394e02",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2605.22992",
+              "source_type": "paper_citation_context",
+              "excerpt": "We evaluated QueryZen on PostgreSQL, MySQL, Mari-aDB, and CockroachDB, which are widely used in previous performance testing works [1, 16, 40].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:13ca3f2df2d41f9de9a16ecbe970ca24a954dcbfb78ee5a96950c7fce2c5b9d9",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2605.22992",
+              "source_type": "paper_citation_context",
+              "excerpt": "In this paper, we collected 10 thousand randomly generated small workloads from SQLancer [27,28], which is a popular DBMS testing tool.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0e9d438585b11ceb04db32b5617889b53f3ff79d991759f3cad30d36f5d6ac19",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2605.22992",
+              "source_type": "paper_citation_context",
+              "excerpt": "Unlike existing approaches [1,21,40], which focus on manipulating external inputs, BFA operates on the program code itself.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5453f198f1e6a8a8d149f133f1485a9f219ab3af929399a54a17d0cdceb05972",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2605_22992.json",
+              "source_type": "paper",
+              "excerpt": "In this paper, we collected 10 thousand randomly generated small workloads from SQLancer[27, 28], which is a popular DBMS testing tool.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 4 Approach, page 5.",
+              "content_sha256": "sha256:29b3422bba516946dce337dca26adaab16192a51f418c8202ece1c5be89bf0c3",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2605_22992.json",
+              "source_type": "paper",
+              "excerpt": "We considered the performance issues found by the four existing performance-testing methods:APOLLO[16],AMOEBA[21],CERT[1], and PUPPY[40].",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 3 Motivating Study, page 3.",
+              "content_sha256": "sha256:98075e8db1a30896363b690b2735d56c88cfaf3cdfce59ed6973ef3f5094f093",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2605_22992.json",
+              "source_type": "paper",
+              "excerpt": "We considered the same performancetesting methods in Section 3:APOLLO[16],AMOEBA[21], CERT[1], and PUPPY[40].",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 6 Evaluation, page 8.",
+              "content_sha256": "sha256:f37e98c4b9fb28be772dffddc56231a824d6cf8659f5a1e95bb80c5403cbab3a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2605_22992.json",
+              "source_type": "paper",
+              "excerpt": "Overall, 8, 0, 4, and 1 issues have the potential to be found by APOLLO,AMOEBA, CERT, and PUPPY, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, 6 Evaluation, page 9.",
+              "content_sha256": "sha256:35a1ab7fb292437bc96916a39b43d448a1050e69fdd29d25a29ddbd9845f6ae2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/652ecd8215728ede7bbecba4ac5049ac96af9e91",
+        "source_type": "paper",
+        "content_sha256": "sha256:925719a7bc15589ab51f0df82311ad22263ede50293d33829fd1038dbccf98da"
+      }
+    },
+    {
+      "id": "paper:arxiv:2604.01442",
+      "title": "Fuzzing with Agents? Generators Are All You Need",
+      "authors": [
+        "Vasudev Vikram",
+        "Rohan Padhye"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2604.01442",
+      "s2_paper_id": "cbba1fcd8795f5727dbe286aeb9af1f5e5d4ba09",
+      "url": "https://arxiv.org/abs/2604.01442",
+      "open_access_pdf": null,
+      "abstract": "Modern generator-based fuzzing techniques combine lightweight input generators with coverage-guided mutation as a method of exploring deep execution paths in a target program. A complimentary approach in prior research focuses on creating highly customized, domain-specific generators that encode structural and semantic logic sufficient enough to reach deep program states; the challenge comes from the overhead of writing and testing these complex generators. We investigate whether AI coding agents can automatically synthesize such target-specific generators, and whether the resulting generators are strong enough to obviate the need for coverage guidance and mutation entirely. Our approach, Gentoo, is comprised of an LLM coding agent (provided terminal access and source code of the fuzz target and its library) instructed to iteratively synthesize and refine an input generator, and optionally provided fine-grained predicate-level coverage feedback. We evaluate three configurations of Gentoo against human-written generators on fuzz targets for 7 real-world Java libraries. Our findings show that agent-synthesized generators achieve statistically significantly higher branch coverage than human-written baseline generators on 4 of 7 benchmarks. Critically, the use of coverage guidance and mutation strategies is not statistically significantly beneficial for agent-synthesized generators, but is significant for all human-written generators, suggesting that structural and semantic logic encoded in the agent generators makes coverage guidance largely unnecessary.",
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2604.01442",
+              "source_type": "paper_citation_context",
+              "excerpt": "Tools such as CSmith [52, 65] for C compilers and SQLancer [8] for database engines generate inputs that are not merely syntactically valid but semantically rich by construction: CSmith emits C programs free of undefined behavior, while SQLancer produces queries that satisfy the relational schemas…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:017e1e8cad83f6627ca61a4a3be2304aced601eea359b0c9b25e07f208cd08f0",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/cbba1fcd8795f5727dbe286aeb9af1f5e5d4ba09",
+        "source_type": "paper",
+        "content_sha256": "sha256:5329ca3101427107386b00215c85a10d4c8de7fe122f12857409c82f5cf95ba1"
+      }
+    },
+    {
+      "id": "paper:arxiv:2602.19490",
+      "title": "FuzzySQL: Uncovering Hidden Vulnerabilities in DBMS Special Features with LLM-Driven Fuzzing",
+      "authors": [
+        "Yongxin Chen",
+        "Zhiyuan Jiang",
+        "Chao Zhang",
+        "Haoran Xu",
+        "Shenglin Xu",
+        "Jianping Tang",
+        "Zheming Li",
+        "Peidai Xie",
+        "Yongjun Wang"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2602.19490",
+      "s2_paper_id": "aad1150ad6b87608cbd6c02f7792a28e785746ff",
+      "url": "https://arxiv.org/abs/2602.19490",
+      "open_access_pdf": null,
+      "abstract": "Traditional database fuzzing techniques primarily focus on syntactic correctness and general SQL structures, leaving critical yet obscure DBMS features, such as system-level modes (e.g., GTID), programmatic constructs (e.g., PROCEDURE), advanced process commands (e.g., KILL), largely underexplored. Although rarely triggered by typical inputs, these features can lead to severe crashes or security issues when executed under edge-case conditions. In this paper, we present FuzzySQL, a novel LLM-powered adaptive fuzzing framework designed to uncover subtle vulnerabilities in DBMS special features. FuzzySQL combines grammar-guided SQL generation with logic-shifting progressive mutation, a novel technique that explores alternative control paths by negating conditions and restructuring execution logic, synthesizing structurally and semantically diverse test cases. To further ensure deeper execution coverage of the back end, FuzzySQL employs a hybrid error repair pipeline that unifies rule-based patching with LLM-driven semantic repair, enabling automatic correction of syntactic and context-sensitive failures. We evaluate FuzzySQL across multiple DBMSs, including MySQL, MariaDB, SQLite, PostgreSQL and Clickhouse, uncovering 64 vulnerabilities, 27 of which are tied to under-tested DBMS special features. As of this writing, 60 cases have been confirmed with 9 assigned CVE identifiers, 31 already fixed by vendors, and additional vulnerabilities scheduled to be patched in upcoming releases. Our results highlight the limitations of conventional fuzzers in semantic feature coverage and demonstrate the potential of LLM-based fuzzing to discover deeply hidden bugs in complex database systems.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2602.19490",
+              "source_type": "paper_citation_context",
+              "excerpt": "To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art open-source fuzzing baselines: Squir-rel [56], EET [19], and SQLancer [35].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9560cad6f6e8b6134756b1fca79081fc05b85032021d259162f17b575a6d45d1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2602.19490",
+              "source_type": "paper_citation_context",
+              "excerpt": "Depending on the bug types targeted—ranging from crash vulnerabilities [1, 12, 13, 18, 22, 27, 45, 56] to logic inconsistencies [4, 19, 23, 35–37, Generation-based approaches rely on hand-crafted grammars or learned rules to synthesize SQL sequences from scratch.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:955e2c63d8172802eaab2eb2adc59b247346a9c37838fbe9997a99c70e9374aa",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2602.19490",
+              "source_type": "paper_citation_context",
+              "excerpt": "Existing DBMS fuzzers have explored SQL generation and mutation from multiple perspectives, including structural complexity [18, 56], sequence-level interactions [13, 22], and oracle-guided logic testing [19, 23, 35].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:72e16f07d9b13f6e228a23ac63af7fd3d83536e90adcea56a048ffd903bca74a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2602.19490",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, most existing DBMS fuzzers [1, 19, 23, 37, 56] predominantly focus on general-purpose statement structures (e.g., SELECT ), covering only a limited portion of the DBMS grammar.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:de1fd49611a86f66a9e0f8c94f7f99ddf849c1cc13e1fb0084fe9909b45977ce",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2602.19490",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [35–37 , 52] and EET Mutation-based fuzzers, in contrast, transform existing SQL statements through changes at the syntax or intermediate representation (IR) level.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1baca451002970a850d476ef936c6d698d61fc1c2a89c79325f593f66fc58929",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2602.19490",
+              "source_type": "paper_citation_context",
+              "excerpt": "For SQLancer, we use TLP [36] as the test oracle for MySQL, and NoREC [35] for the other DBMS targets.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0fd05fb9593d4abc569e1a74eaf66e355125f865b1c348a258c9be70270db4fb",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp",
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2602_19490.json",
+              "source_type": "paper",
+              "excerpt": "To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art opensource fuzzing baselines: Squirrel [ 56], EET [ 19], and SQLancer [ 35].",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 5.1 Experimental Setup, page 15.",
+              "content_sha256": "sha256:4c6ae5b799f62c5acce04d367dcae4951234401ddba68d7ec420558c691cceb8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2602_19490.json",
+              "source_type": "paper",
+              "excerpt": "Since EET and SQLancer are not inherently a grey-box fuzzing tool, we first collect their generated test cases and then utilize FuzzySQL’s replay mechanism to measure its coverage.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 5.1 Experimental Setup, page 15.",
+              "content_sha256": "sha256:90dbc7be66b59161f68908d25edab42bec31a58c31164f5a731ccab3a311c816",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2602_19490.json",
+              "source_type": "paper",
+              "excerpt": "For SQLancer, we use TLP [ 36] as the test oracle for MySQL, and NoREC [ 35] for the other DBMS targets.",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, 5.1 Experimental Setup, page 15.",
+              "content_sha256": "sha256:61df97066f3ac7b3bfb4cd90a1d5402ba6a61e4a9f0da3a04bc3f1f4094ad4cd",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2602_19490.json",
+              "source_type": "paper",
+              "excerpt": "Deduplicated MySQL bugs discovered within 24 hours Target ID†FR Type FuzzySQL Squirrel EET SQLancer FuzzySQL!",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, 5.2 Bug Discovery Effectiveness, page 16.",
+              "content_sha256": "sha256:4fc578cffdc951584d6d9714c2ee4ea364dc8eaf537495261deb60017fa17f96",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2602_19490.json",
+              "source_type": "paper",
+              "excerpt": "To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art opensource fuzzing baselines: Squirrel [ 56], EET [ 19], and SQLancer [ 35].",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 5.1 Experimental Setup, page 15.",
+              "content_sha256": "sha256:4c6ae5b799f62c5acce04d367dcae4951234401ddba68d7ec420558c691cceb8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/aad1150ad6b87608cbd6c02f7792a28e785746ff",
+        "source_type": "paper",
+        "content_sha256": "sha256:d3952f18b9f47693c96df1a2f97ab7df4e697f7eb53a571cfa7f2b0cacbb3489"
+      }
+    },
+    {
+      "id": "paper:arxiv:2607.03741",
+      "title": "Graph-Aware Fuzzing for Graph Database Management Systems",
+      "authors": [
+        "Yu Li",
+        "Qiang Hu",
+        "Yao Zhang",
+        "Junjie Wang",
+        "Haoxuan Liu",
+        "Rui Wang",
+        "Yongqiang Lyu"
+      ],
+      "year": 2026,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": "2607.03741",
+      "s2_paper_id": "d6433b58411aed7605dd789072997fc0884bd610",
+      "url": "https://arxiv.org/abs/2607.03741",
+      "open_access_pdf": null,
+      "abstract": "Graph Database Management Systems (GDBMSs) are essential infrastructure for managing interconnected data. Existing GDBMS testing methods primarily rely on differential and metamorphic testing. The result consistency oracles of these methods constrain inputs to queries that are comparable across engines or transformations, leaving single engine runtime failures, such as crashes and memory errors, insufficiently explored. Developing dedicated fuzzers for GDBMSs faces two key challenges: (1) generating valid and structurally diverse queries under complex graph constraints, and (2) guiding exploration to capture topology dependent execution behavior. To address these challenges, we propose GRAF, a black box fuzzing framework for GDBMS query engines. First, GRAF introduces graph context aware query generation based on cascading dependency resolution. It instantiates parameterized Cypher skeletons generated by a Large Language Model (LLM) by jointly resolving labels, relationship types, properties, values, and variable scopes against the active graph state. This process produces structurally diverse queries while eliminating syntactic and semantic violations. Second, GRAF applies five graph specific mutation operators guided by execution state feedback, including execution time, result size, and system status. This feedback steers exploration away from unproductive queries and expensive traversals, while prioritizing local mutations around abnormal executions. We evaluated GRAF against three existing approaches on six widely used GDBMSs. GRAF consistently improves line coverage by 31.6% to 41.1% over the strongest baseline on each target. In 12 hour fuzzing, it triggered 25 unique bugs, compared to six from all baselines combined. Overall, GRAF discovered 34 previously unknown bugs, with 32 confirmed by developers and 23 assigned CVEs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2607.03741",
+              "source_type": "paper_citation_context",
+              "excerpt": "Tools such as SQLsmith [7] and SQLancer [8]–[10] generate SQL queries based on abstract syntax tree models or formal semantics, successfully exposing bugs in mature systems such as PostgreSQL [11] and SQLite [12].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e7c54647458bc43bcba0e54ace9545b98716fbbf6fed192d0490592b4f6d785c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2607.03741",
+              "source_type": "paper_citation_context",
+              "excerpt": "To systematically detect logic bugs, SQLancer [8]–[10] automatically synthesizes queries and validates their correctness using advanced logical oracles like Ternary Logic Partitioning (TLP [8]).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bece67f5684ea246c2c9bbe86e68713d3cdcd115f1d8941388c2515cbc992e00",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2607.03741",
+              "source_type": "paper_citation_context",
+              "excerpt": "Existing GDBMS testing methods [15]–[20], [22], [25] mainly rely on differential or metamorphic testing and typically lack feedback driven exploration.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:678c71fe30e35275045aab3ed49440f36d8f548f323ff18b722aac5bed88a498",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2607.03741",
+              "source_type": "paper_citation_context",
+              "excerpt": "Existing DBMS fuzzing techniques [7]–[10], [13], [25], [40]–[48] include generation-based and mutation-based approaches.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e0781a75fbede4854c52f3adc9559619008bb5565324238efbf1f96994995eb7",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/d6433b58411aed7605dd789072997fc0884bd610",
+        "source_type": "paper",
+        "content_sha256": "sha256:3e23e4871f3bb1fc472d89a6891dc09c05c041efddaebec16e614fd210dd1ac9"
+      }
+    },
+    {
+      "id": "paper:arxiv:2606.14164",
+      "title": "Investigating Metamorphic Fuzz Oracle Enhancement via Large Language Models",
+      "authors": [
+        "Ruixiang Qian",
+        "Ding Yang",
+        "Zengxu Chen",
+        "Yue Gao",
+        "Chunrong Fang",
+        "Chao Zhang",
+        "Zhenyu Chen"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2606.14164",
+      "s2_paper_id": "101f3528945d4a6501c451e0c481481b440a3c91",
+      "url": "https://arxiv.org/abs/2606.14164",
+      "open_access_pdf": null,
+      "abstract": "Fuzz drivers are essential components of greybox fuzzing, as they encapsulate target interfaces, define test spaces, and largely determine fuzzing effectiveness. Existing fuzz drivers typically rely on crash-based oracles for security testing, overlooking library functionality and limiting bug detection capability. In this paper, we present the first study on metamorphic-based fuzz oracle enhancement (MFOE), which augments existing fuzz drivers with metamorphic-based oracles derived from metamorphic relations (MRs). Since constructing and integrating such oracles requires substantial domain knowledge, automating MFOE is challenging. To address this challenge, we propose MetaFOE, an LLM-based framework that automatically generates and integrates metamorphic-based oracles. We evaluate MetaFOE on OSS-Fuzz drivers using three modern LLMs and five prompt strategies. MetaFOE generates 3,475 MRs, of which 77.3% are applicable, and implements 12,351 meta drivers, with 6,228 being valid. After three hours of fuzzing, the valid meta drivers improve edge coverage by an average of 18.7% and trigger 1,528 unique crashes. Our results demonstrate both the effectiveness of metamorphic-based oracle enhancement and the feasibility of using LLMs to automate MFOE, providing valuable insights for advancing greybox fuzzing.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2606.14164",
+              "source_type": "paper_citation_context",
+              "excerpt": "As a powerful technique for alleviating the oracle problem [Guo et al. 2024; Liu et al. 2013; Mu et al. 2025; Rigger and Su 2020; Segura et al. 2018], metamorphic testing has naturally attracted the attention of the fuzzing community and has been adopted to enhance fuzz oracles.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e99dcd93816fb138e0dcf73e33774cd9a6c45bbe58b93582075a30bc145e3e56",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/101f3528945d4a6501c451e0c481481b440a3c91",
+        "source_type": "paper",
+        "content_sha256": "sha256:25841e4f2c099de1b0ab9c12c31353afa84b740488b2080369596dac235d68fa"
+      }
+    },
+    {
+      "id": "paper:arxiv:2603.21530",
+      "title": "LLM-Based Test Case Generation in DBMS through Monte Carlo Tree Search",
+      "authors": [
+        "Yujia Chen",
+        "Yingli Zhou",
+        "Fangyuan Zhang",
+        "Cuiyun Gao"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2603.21530",
+      "s2_paper_id": "c1254b7ff833135a0a9dc60220434ef2716ff35d",
+      "url": "https://arxiv.org/abs/2603.21530",
+      "open_access_pdf": null,
+      "abstract": "Database Management Systems (DBMSs) are fundamental infrastructure for modern data-driven applications, where thorough testing with high-quality SQL test cases is essential for ensuring system reliability. Traditional approaches such as fuzzing can be effective for specific DBMSs, but adapting them to different proprietary dialects requires substantial manual effort. Large Language Models (LLMs) present promising opportunities for automated SQL test generation, but face critical challenges in industrial environments. First, lightweight models are widely used in organizations due to security and privacy constraints, but they struggle to generate syntactically valid queries for proprietary SQL dialects. Second, LLM-generated queries are often semantically similar and exercise only shallow execution paths, thereby quickly reaching a coverage plateau. To address these challenges, we propose MIST, an LLM-based test case generatIon framework for DBMS through Monte Carlo Tree search. MIST consists of two stages: Feature-Guided Error-Driven Test Case Synthetization, which constructs a hierarchical feature tree and uses error feedback to guide LLM generation, aiming to produce syntactically valid and semantically diverse queries for different DBMS dialects, and Monte Carlo Tree Search-Based Test Case Mutation, which jointly optimizes seed query selection and mutation rule application guided by coverage feedback, aiming at boosting code coverage by exploring deeper execution paths. Experiments on three widely-used DBMSs with four lightweight LLMs show that MIST achieves average improvements of 43.3% in line coverage, 32.3% in function coverage, and 46.4% in branch coverage compared to the baseline approach with the highest line coverage of 69.3% in the Optimizer module.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2603.21530",
+              "source_type": "paper_citation_context",
+              "excerpt": "Traditional approaches for generating test cases, such as BuzzBee [46], SQLsmith [34], SQLancer [32], can be effective for a specific DBMS; however, there exist many DBMSs designed for different application scenarios, such as DuckDB [29] for analytical workloads, PostgreSQL [36] for general-purpose…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:cf27356f3259c6b68ff41b6f90057e71083998eef85f9dc7144e6abec4b6aa1d",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2603.21530",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [30, 31, 33] employs hand-crafted dialect-specific generators with sophisticated test oracles to detect logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:07837a212bc4b4d660941f35e6aa22c87d42335120624a2325e428ba7aaee72b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/c1254b7ff833135a0a9dc60220434ef2716ff35d",
+        "source_type": "paper",
+        "content_sha256": "sha256:0422e6dc203559919bdb0f0ef3e6de89782c7dd6e4871fff121769b138c05173"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icde65706.2026.00240",
+      "title": "LLMSQLMUTATOR: LLM-Powered Test Case Generation for Database Using Bug Reports",
+      "authors": [
+        "Chenglin Tian",
+        "Chaofan Li",
+        "Yawen Li",
+        "Yingxia Shao"
+      ],
+      "year": 2026,
+      "venue": "IEEE International Conference on Data Engineering",
+      "doi": "10.1109/icde65706.2026.00240",
+      "arxiv_id": null,
+      "s2_paper_id": "76ce4c5143cca31617dc53fe0c37b77ad8d5150b",
+      "url": "https://doi.org/10.1109/icde65706.2026.00240",
+      "open_access_pdf": null,
+      "abstract": "Relational Database Management Systems (DBMSs) are fundamental to modern software infrastructure, yet their inherent complexity and continuous evolution introduce critical defects that compromise data integrity and system reliability. While existing syntax-based and mutation-based automated testing approaches have demonstrated utility in DBMS validation, they suffer from three significant limitations, that are low quality seed SQLs, lack of bug knowledge, and ignorance of fine-grained constraints. To address these limitations, we propose LLM-SQLMutator, an innovative mutation-based automated DBMS testing tool that leverages large language models (LLMs) and bug reports. LLMSQLMUTATOR extracts the SQL statements from the bug report as mutation seeds, and extracts bug patterns and root causes from the bug reports as mutation guidance. Considering the semantic understanding capability of LLMs, we introduce bug-driven SQL mutation which uses LLMs to perform syntax-aware directed mutations on seed SQLs under the guidance of the bug knowledge. Finally, we propose the LLMbased semantic validation that uses database metadata as the basic validation knowledge and dynamically expand validation rules via LLMs, thereby helping LLMSQLMutator achieve comprehensive semantic validation. Extensive evaluations across six popular relational DBMSs demonstrate the advantages of LLMSQLMutator, and it detects 27 confirmed bugs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+              "source_type": "paper_citation_context",
+              "excerpt": "They also may yield false positives due to the varied implementation choices of RDBMSs. Metamorphic testing is another mainstream approaches for RDB-MS testing [51], [54], [14], [12], [13], [55], [56], [15].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9d5c2174cf382f876f32a46e546854d015f9521a05f0164bd75ee30b3b4d780c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+              "source_type": "paper_citation_context",
+              "excerpt": "In recent years, SQLancer [11] has emerged as the most effective black-box fuzzing tool, distinguished by its adoption of three complementary oracles [14], [12], [13].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3673fd74545a1c2505567b97b995f437ab8cf227ca45d57bfb460d56d4291c29",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+              "source_type": "paper_citation_context",
+              "excerpt": "2) For test cases containing SELECT queries, we employ result inconsistency detection by randomly selecting one test oracle from TLP [13], NoREC [12], or CERT [44].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1c5dcd1e507ffedd5b346af70c73582ed2244e008f0cb2ed61394bca034104f0",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+              "source_type": "paper_citation_context",
+              "excerpt": "Generally, the automated DBMS testing process comprises two phases: test case generation [6], [11] and test oracle construction [12], [13], [14], [15].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8596d6dc155c5347ec3c215a793cdb476fa0b04583712e481291d780929d77b3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+              "source_type": "paper_citation_context",
+              "excerpt": "Existing fuzzers typically restrict recursion depth to maintain syntactic correctness [15], [13], thereby failing to reach the parser’s threshold.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f0fbd8b41362389d8402618275419009910dd10a73024bf03d588d9f170009c6",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00240",
+              "source_type": "paper_citation_context",
+              "excerpt": "We selected the three most effective existing oracles for detecting logic bugs through result inconsistency: NoREC [12], TLP [13], and CERT [44].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:fbf02a450589ed7f5e29f34d10e161d1721eeccc3de64bde0e4b231b4e475cbc",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00240.json",
+              "source_type": "paper",
+              "excerpt": "2) For test cases containing SELECT queries, we employ result inconsistency detection by randomly selecting one test oracle from TLP [13], NoREC [12], or CERT [44].",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, D DBMS Testing, page 8.",
+              "content_sha256": "sha256:1c5dcd1e507ffedd5b346af70c73582ed2244e008f0cb2ed61394bca034104f0",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00240.json",
+              "source_type": "paper",
+              "excerpt": "We selected the three most effective existing oracles for detecting logic bugs through result inconsistency: NoREC [12], TLP [13], and CERT [44].",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, IV EVALUATION, page 8.",
+              "content_sha256": "sha256:fbf02a450589ed7f5e29f34d10e161d1721eeccc3de64bde0e4b231b4e475cbc",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00240.json",
+              "source_type": "paper",
+              "excerpt": "Of these 14 result inconsistency bugs, 12 bugs are detected by the TLP oracle, and the other two are found by the CERT oracle.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, B Experiment results, page 9.",
+              "content_sha256": "sha256:ac213469705622dfaba816e37ca8832aa96fe28fe968f9039eb545c8ce701bbb",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00240.json",
+              "source_type": "paper",
+              "excerpt": "2) Comparsion of baselines We compare LLMSQLM UTATOR with state-of-the-art DBMS testing approaches, including mutation-based methods SQLRight [15] and Squirrel [16], as well as grammar-based SQLancer [11] and random-based SQLsmith [6] to provide a comprehensive empirical comparison.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, B Experiment results, page 9.",
+              "content_sha256": "sha256:2ee0a39a296e3a9c8f2e19c4b466e8c46818769f607f2586cb7f6eaeac9a2604",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00240.json",
+              "source_type": "paper",
+              "excerpt": "TABLE VI: The bug number and bug pattern covered by SQLsmith, SQLancer, SQLRight, Squirrel and LLMSQLM UTATOR in 48h DBMSSQLsmith SQLancer Squirrel SQLRight LLMSQLM UTATOR #Bug Pattern #Bug Pattern #Bug Pattern #Bug Pattern #Bug Pattern MySQL - - 1 23.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, B Experiment results, page 10.",
+              "content_sha256": "sha256:4081d7246b0833b69fa2d2ef4a6a363169264b769c17267c1c7ae3a438886c00",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00240.json",
+              "source_type": "paper",
+              "excerpt": "SQLancer attains only modest coverage (peaking at 52.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, B Experiment results, page 11.",
+              "content_sha256": "sha256:294a4afdfaab1101ff33d63c8b7e5367c2a38c2e4dbdcc1e691f7e1e269710b5",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00240.json",
+              "source_type": "paper",
+              "excerpt": "In recent years, SQLancer [11] has emerged as the most effective black-box fuzzing tool, distinguished by its adoption of three complementary oracles [14], [12], [13].",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, V RELATED WORK, page 12.",
+              "content_sha256": "sha256:3673fd74545a1c2505567b97b995f437ab8cf227ca45d57bfb460d56d4291c29",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/76ce4c5143cca31617dc53fe0c37b77ad8d5150b",
+        "source_type": "paper",
+        "content_sha256": "sha256:1e5e3620dfaa0a559405fa5b1ef0777c52c0d36e2b8b6fccb11efde704479cac"
+      }
+    },
+    {
+      "id": "paper:arxiv:2608.15709",
+      "title": "Logos: Certified Order-Sensitive SQL Rewrites with Mechanized Semantics and LLM Guidance",
+      "authors": [
+        "Jing-Yu Ke",
+        "Jingyang Li",
+        "Guoqiang Li"
+      ],
+      "year": 2026,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": "2608.15709",
+      "s2_paper_id": "513b2356cc9a799751b6c96ce838f98e67b8696e",
+      "url": "https://arxiv.org/abs/2608.15709",
+      "open_access_pdf": null,
+      "abstract": "SQL rewrite verification must account for duplicate rows, observable row order, and typed value semantics. Existing verifiers have yet to combine proofs over database instances of arbitrary finite cardinality with an ordered-list semantics for nested, tie-sensitive top-k. Unbounded systems reason primarily over bags or handle ordering through syntax-directed restrictions, whereas bounded systems either support only restricted top-k forms or impose a deterministic ordering rather than retain all legal tie-induced outcomes. Support for typed expression and aggregate semantics, observable runtime errors, and integrity constraints also remains partial. In Rocq, we mechanize a compositional logical semantics for a typed SQL core with order-sensitive operators, capturing all possible ordered lists and observable SQL failures in the supported fragment. To our knowledge, this is the first mechanized SQL semantics to combine nested, tie-sensitive top-k with a closure-based lifting from bag equivalence to ordered-list equivalence, enabling sound reuse of bag-theoretic reasoning while preserving compositionality across order-sensitive and correlated contexts. The formalization further provides executable semantics for PostgreSQL-oriented scalar and aggregate evaluation and an explicit account of integrity constraints. Building on this semantics, we present Logos, an LLM-guided Rocq verifier for unbounded SQL rewrite equivalence. Its agent uses a verified SQL-specific lemma library to construct query-specific Rocq proofs. Our evaluation covers 389 query pairs from Apache Calcite optimizer tests, TPC-H and TPC-DS rewrites, and WeTune's real-application workloads. Logos solves 86.9% of them, compared with 64.0% for SQLSolver, the strongest baseline.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/513b2356cc9a799751b6c96ce838f98e67b8696e",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/513b2356cc9a799751b6c96ce838f98e67b8696e",
+        "source_type": "paper",
+        "content_sha256": "sha256:9bfc945c9cf982f3b986a05cf19f984777ee03499eb29eb8336d0fe9219300f4"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3798226",
+      "title": "Metamorphic Testing for Infrastructure-as-Code Engines",
+      "authors": [
+        "David Spielmann",
+        "George Zakhour",
+        "Dominik Arnold",
+        "Matteo Biagiola",
+        "Roland Meier",
+        "Guido Salvaneschi"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Programming Languages",
+      "doi": "10.1145/3798226",
+      "arxiv_id": null,
+      "s2_paper_id": "3a205ccef9fb1acd547d866fa36a98c549d88344",
+      "url": "https://doi.org/10.1145/3798226",
+      "open_access_pdf": "https://doi.org/10.1145/3798226",
+      "abstract": "Infrastructure-as-Code (IaC) engines, such as Terraform, OpenTofu, and Pulumi, automate the provisioning and management of cloud resources. They parse IaC specifications and orchestrate the required actions, making them the backbone of modern clouds, and critical to the reliability of both the underlying infrastructure and the software that depends on it. Despite this importance, this class of systems has received little attention: prior work largely targets the correctness of IaC programs rather than the IaC engines themselves. Existing test suites rely on manually written oracles and struggle to expose faults that manifest across multiple executions, leaving a significant reliability gap. We present EMIaC, a metamorphic testing framework for IaC engines. EMIaC defines metamorphic relations as graph-based transformations of IaC programs and checks invariants across executions of the original and transformed programs. A central novelty is our use of e-graphs in software testing, as both a test-input generator and an equivalence oracle. E-graphs compactly represent program equivalences, enabling the systematic generation of large spaces of equivalent IaC programs. To ground these relations, we analyze 43,593 real-world Terraform programs and show that IaC dependency graphs are typically small and sparse, making e-graphs a natural fit. Evaluating EMIaC on Pulumi, Terraform, and OpenTofu, we show that it complements existing test suites by exercising engine-critical code paths and covering 98 previously untested statements in Terraform and 1,313 in Pulumi. EMIaC also uncovers previously unknown issues in all three test suites, improving their adequacy. Three test cases have been merged into Terraform’s main branch, and Pulumi has merged a specification fix.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3798226",
+              "source_type": "paper_citation_context",
+              "excerpt": "Mutation analysis [17] addresses this problem by systematically injecting syntactic changes into the program under test (i.e., mutants ) to mimic programmers’ mistakes, and by assessing whether the test suite is able to catch them.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5152a8d2bba904b323c02ba92af37a4846a6dc068259359ef174cde44a7a79a5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/3a205ccef9fb1acd547d866fa36a98c549d88344",
+        "source_type": "paper",
+        "content_sha256": "sha256:be1b148c040a8638e465a02661a31fa36bf0f392beebf6b4fa8922f8e3f882d1"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3786673",
+      "title": "One DBMS, Two Modes, and a Bunch of Bugs: Catching Logic Bugs in Distributed DBMSs via Differential Testing",
+      "authors": [
+        "Zikun Fu",
+        "Jiaju Bai",
+        "Hong-Bo Feng",
+        "Kang Chen"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Management of Data",
+      "doi": "10.1145/3786673",
+      "arxiv_id": null,
+      "s2_paper_id": "dc60906c576d85944e5ba7612dd4ce32d34de9af",
+      "url": "https://doi.org/10.1145/3786673",
+      "open_access_pdf": "https://doi.org/10.1145/3786673",
+      "abstract": "Logic bugs in distributed database management systems silently yield incorrect results and are difficult to detect, yet they threaten correctness in critical deployments. Existing testing techniques focus mainly on system level failures or general DBMS and do not target distributed execution, leaving many distributed logic bugs undiscovered. We present DistSQL, a differential testing framework that compares the results of identical SQL queries executed on the same DBMS configured in centralized and distributed modes. The key insight is that centralized execution is simpler and typically better tested, and thus can serve as a practical reference for distributed execution. DistSQL addresses two difficulties that hinder effective bug finding in distributed settings: it performs distributed diversity oriented database state mutation to expose distribution specific behaviors, and it conducts distributed interaction guided exploration of the execution space using query plan features to prioritize novel behaviors over redundant tests. DistSQL requires no intrusive code instrumentation. The evaluation on five popular open source distributed DBMSs, including TiDB, CockroachDB, YugabyteDB, ClickHouse, and OceanBase, as well as one widely deployed commercial DBMS, demonstrates the efficacy of DistSQL. DistSQL identified 65 previously unknown logic bugs, including 38 specific to distributed execution. Of these, 61 have been confirmed and 50 have been fixed.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3786673",
+              "source_type": "paper_citation_context",
+              "excerpt": "We compare DistSQL against SOTA database and distributed system testing tools, including Squirrel [65], QPG [2], and EET [24], as well as distributed system testing tools, including Jepsen [20] and Mallory [36].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:80e616b758ae1428e63a859d29f20581524b3d40535883855b0e8718cbc14499",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3786673",
+              "source_type": "paper_citation_context",
+              "excerpt": "We evaluate five open-source distributed DBMSs (Table 3) using four existing tools: Jepsen [20], Mallory [36], QPG [2], and EET [24], and compare their performance with DistSQL.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:06e07318655a5b1372a20336bdde704c782868b8c77d58c503eb1aae9e3310f2",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3786673",
+              "source_type": "paper_citation_context",
+              "excerpt": "DQP [3] compares executions with different query hints applied, where query plans act as local heuristics rather than guiding test case generation.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:01a17381630d8c4e6af595d74ae00142f0b4fc887c04da38dd780fb8ed25d966",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3786673",
+              "source_type": "paper_citation_context",
+              "excerpt": "This shows that DistSQL is effective and applicable in real-world production settings. feedback signals, such as formal rule coverage in SemConT [31] and graph similarity of test cases in TQS [53], while others adopt more generic runtime feedback, such as branch coverage [30] or query plan text [2].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d3474e867636c5828bcb852b61ba2f3fd6fd4c376a94464b0d431e2f8cdc2f8e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3786673",
+              "source_type": "paper_citation_context",
+              "excerpt": "…[49] DBMS Logic None Yes Differential Generic SemConT [31] DBMS Logic Keyword and Rule Yes Verification Generic TQS [53] DBMS Logic Graph Similarity Yes Ground Truth Generic QPG [2] centralized execution as a reference, our approach finds distributed logic bugs often missed by existing techniques.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6748d09cee0c91e2a4499f3d1ac469ccfb3386cdae29e851da867ad61a4d9a75",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3786673",
+              "source_type": "paper_citation_context",
+              "excerpt": "We implement an abstract syntax tree (AST) based SQL generator [1, 22, 45, 47] that constructs queries in a top down manner by first selecting the SQL statement type (e.g., SELECT , CREATE TABLE ) and then recursively generating its components (e.g., ORDER BY clauses, expressions).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1f87e7476f14e53d6d19c16fe1d33283ad2794b03d0cbe7eb81daa92abab13ca",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3786673.json",
+              "source_type": "paper",
+              "excerpt": "We compare DistSQL against SOTA database and distributed system testing tools, including Squirrel [ 65], QPG [ 2], and EET [ 24], as well as distributed system testing tools, including Jepsen [20] and Mallory [36].",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, 7.1 Experimental Setup, page 14.",
+              "content_sha256": "sha256:58734eae48c9029213d9fd5136ea2465a617549ca95ea6d3a8a4e20f0d3cc5d6",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3786673.json",
+              "source_type": "paper",
+              "excerpt": "4 Comparison We evaluate five open-source distributed DBMSs (Table 3) using four existing tools: Jepsen [ 20], Mallory [36], QPG [2], and EET [24], and compare their performance with DistSQL.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, 7.4 Comparison, page 17.",
+              "content_sha256": "sha256:ddedbec8b645c70dff6b34047cacb2991d8768199e6db274e760aad372a83aa2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3786673.json",
+              "source_type": "paper",
+              "excerpt": "DistSQL identifies 28 more bugs than QPG and 31 more than EET.",
+              "excerpt_is_verbatim": true,
+              "note": "M15 in the paper's extracted text, 7.4 Comparison, page 17.",
+              "content_sha256": "sha256:c3cbbf08451194ac0e36578f4f7d03ba26e4fbe2f65272e8637ead1d28b683a9",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3786673.json",
+              "source_type": "paper",
+              "excerpt": "We compare DistSQL against SOTA database and distributed system testing tools, including Squirrel [ 65], QPG [ 2], and EET [ 24], as well as distributed system testing tools, including Jepsen [20] and Mallory [36].",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, 7.1 Experimental Setup, page 14.",
+              "content_sha256": "sha256:58734eae48c9029213d9fd5136ea2465a617549ca95ea6d3a8a4e20f0d3cc5d6",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/dc60906c576d85944e5ba7612dd4ce32d34de9af",
+        "source_type": "paper",
+        "content_sha256": "sha256:5c0efb25755604917b60c411818ad2bc004d2b02876873254e188bff9059dd5a"
+      }
+    },
+    {
+      "id": "paper:arxiv:2606.11132",
+      "title": "Operationalizing Property-Based Testing for Data-Intensive Scalable Computing Systems",
+      "authors": [
+        "Yaoxuan Wu",
+        "I. Lee",
+        "Ahmad Humayun",
+        "Muhammad Ali Gulzar",
+        "Miryung Kim"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2606.11132",
+      "s2_paper_id": "09e050c29b5f2592ad6d0b2a90121bb1530ead84",
+      "url": "https://arxiv.org/abs/2606.11132",
+      "open_access_pdf": null,
+      "abstract": "While fuzzing effectively catches crashes, its shallow oracles often miss semantic drifts and optimization-related errors in data-intensive scalable computing (DISC) frameworks. Property-based testing (PBT) addresses this limitation by checking general semantic invariants across diverse workloads and inputs, rather than relying on specific expected outputs. However, systematically operationalizing PBT for DISC systems remains difficult because it requires both reusable property definitions and effective instantiation into valid workloads and data. We present DiscPBT, a property-based testing engine for Apache Spark. DiscPBT introduces eight reusable meta-properties for DISC semantic testing, spanning equivalence rewriting, data decomposition, computation decomposition, and operator-local semantic relations. To operationalize these meta-properties, DiscPBT provides reusable generators for synthesizing valid workload skeletons and input data, together with an instantiation framework that realizes each meta-property in schema-compatible contexts through compatible operators, expressions, and UDFs. Our evaluation on PySpark shows that DiscPBT achieves 1.2$\\times$ higher branch coverage and 1153$\\times$ greater plan diversity than CometFuzz. Across 66 concrete properties, DiscPBT reveals cross-version semantic drift as well as subtle corner-case pitfalls involving NaN and empty inputs, that are not captured by crash-based fuzzing alone. These results demonstrate the value of systematic PBT for uncovering semantic issues in DISC frameworks.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/09e050c29b5f2592ad6d0b2a90121bb1530ead84",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2606_11132.json",
+              "source_type": "paper",
+              "excerpt": "In relational DBMS testing, SQLancer [ 32] demonstrated the effectiveness of query oracles, including TLP [ 2], which checks result preservation under predicate-based partitioning, and NoREC [ 3], which validates queries through equivalence-preserving rewrites that disable optimizer-dependent execution paths.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, 6 Related Work, page 11.",
+              "content_sha256": "sha256:a9dd02247a7593df29de7a073dafd069db39325183436638b8bb6619f07b83ad",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/09e050c29b5f2592ad6d0b2a90121bb1530ead84",
+        "source_type": "paper",
+        "content_sha256": "sha256:97e4c25a6aecc6641214237fe8702e68e79ed7464f94ae1571ec0ebaf3a1cac5"
+      }
+    },
+    {
+      "id": "paper:doi:10.53799/zb6b1375",
+      "title": "Optimizing Database System Performance: Design and Query Optimization Strategies",
+      "authors": [
+        "Juanda Hakim Lubis",
+        "I. Jaya",
+        "Fajrul Malik Aminullah Napitupulu",
+        "H. Mawengkang"
+      ],
+      "year": 2026,
+      "venue": "AIUB Journal of Science and Engineering (AJSE)",
+      "doi": "10.53799/zb6b1375",
+      "arxiv_id": null,
+      "s2_paper_id": "bdaced332423f73da24f03a45785321328579190",
+      "url": "https://doi.org/10.53799/zb6b1375",
+      "open_access_pdf": "https://ajse.aiub.edu/index.php/ajse/article/download/1615/247",
+      "abstract": "The amount of data stored in magnetic disks (e.g., floppy disks) increases by 100% each year for each department in a company, necessitating efforts to maintain an optimal database system. Designing a database is the initial step in creating a system with optimal performance. However, database design alone is not sufficient to enhance performance. One approach to improving data transaction speed is by optimizing query processing. This research evaluates different relational database models using varying amounts of data. Query costs are analyzed using the Cost-Based Optimizer method and access time measurements. The results of this study provide insights for database administrators in designing relational database models effectively and selecting appropriate query structures to optimize database performance. The findings indicate that: (1) database design can be optimized by separating entities based on specialized usage, and (2) factors such as record count, attribute size, query type, use of unique or primary keys, order-by clauses, index sequences, and SQL function usage significantly impact query cost and overall performance.",
+      "cites_seed_techniques": [
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/bdaced332423f73da24f03a45785321328579190",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Cardinality Estimation Restriction Testing (CERT).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/bdaced332423f73da24f03a45785321328579190",
+        "source_type": "paper",
+        "content_sha256": "sha256:d097cabed5eff4cd188f696370a71c8c46f9df8e8dea359ff279eeec6c66b293"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3828685",
+      "title": "Programmable Property-Based Testing",
+      "authors": [
+        "A. Keles",
+        "Justine Frank",
+        "Ceren Mert",
+        "Harrison Goldstein",
+        "Leonidas Lampropoulos"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Programming Languages",
+      "doi": "10.1145/3828685",
+      "arxiv_id": "2602.18545",
+      "s2_paper_id": "1b21dbf4dfd10358f77bf5ba3b9ba39c81e18b25",
+      "url": "https://doi.org/10.1145/3828685",
+      "open_access_pdf": "https://doi.org/10.1145/3828685",
+      "abstract": "Property-based testing (PBT) is a popular technique for establishing confidence in software, where users write properties---i.e. executable specifications--that can be checked many times in a loop by a testing framework. In modern PBT frameworks, properties are usually written in shallowly embedded domain-specific languages, and their definition is tightly coupled to the way they are tested. Such frameworks often provide convenient configuration options to customize aspects of the testing process, but users are limited to precisely what library authors had the prescience to allow for when developing the framework; if they want more flexibility, they may need to write a new framework from scratch. We propose a new, deeper language for properties based on a mixed embedding that we call deferred binding abstract syntax, which reifies properties as a data structure and decouples them from the property runners that execute them. We implement this language in Rocq and Racket, leveraging the power of dependent and dynamic types, respectively. Finally, we showcase the flexibility of this new approach by rapidly prototyping a variety of property runners, highlighting domain-specific testing improvements that can be unlocked by more programmable testing.",
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/1b21dbf4dfd10358f77bf5ba3b9ba39c81e18b25",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/1b21dbf4dfd10358f77bf5ba3b9ba39c81e18b25",
+        "source_type": "paper",
+        "content_sha256": "sha256:be555bb7b9e2128b863936f24f52eb9312bd9067fed91c1b1bf2474ee17e83b9"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3828612.3828639",
+      "title": "ReSequel: Robust LLM-assisted Query Rewriting and Optimization using Templatization and Sampling",
+      "authors": [
+        "Saeed Fathollahzadeh",
+        "Essam Mansour",
+        "Matthias Boehm"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": "10.14778/3828612.3828639",
+      "arxiv_id": "2606.20853",
+      "s2_paper_id": "b368cff4b76de570c592a1d84bf021214e3b3831",
+      "url": "https://doi.org/10.14778/3828612.3828639",
+      "open_access_pdf": "https://arxiv.org/pdf/2606.20853",
+      "abstract": "Heuristic query rewriting has long complemented cost-based optimization to improve performance. Such rewrites transform SQL queries into semantically equivalent forms that are easier or faster to execute. Examples are standardizing expressions, eliminating redundancy, propagating constants, pushing down selections and projections, unnesting queries, and utilizing constraints. Modern DBMSs implement hundreds to thousands of such rules, but maintaining them is notoriously difficult. The interactions among rules are complex, and their static nature and application order prevent adaptation to specific query and database characteristics. Recent approaches that use large language models (LLMs) for query rewriting show promise but face challenges regarding the large search space, reliable query verification, and exploitation of metadata. We present ReSequel, an outer optimization layer on top of existing DBMSs to rewrite SQL queries using LLMs. ReSequel leverages catalog and statistical metadata to infer template-specific rules that guide the LLM toward effective query transformations. We generate, verify, and rank rewritten query variants on sampled data to ensure result correctness and runtime improvements. Our experiments cover eight benchmarks: JOB, TPC-H, Stats(-CEB), Public BI, IMDB, DSB, and StackOverflow; multiple DBMSs: PostgreSQL, MySQL, and DuckDB; as well as LLM-based query rewriting baselines. ReSequel yields workload-level speedups of up to 16x over native DBMSs and 22x over LLM-based systems, with individual queries exceeding 600x, across eight benchmarks and three DBMSs.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3828612.3828639",
+              "source_type": "paper_citation_context",
+              "excerpt": "Beyond these methods, prior work on SQL testing [78] and text-to-SQL generation [74] compare query results directly.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:900c92c01ef7e9c9703696dd445f051794ed1b524133f0aa1d5549b4483bb3ab",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/b368cff4b76de570c592a1d84bf021214e3b3831",
+        "source_type": "paper",
+        "content_sha256": "sha256:73d8c84d73d8260354eaf8022f81cd96a082e1f8046f73523d84bc88134d9c2b"
+      }
+    },
+    {
+      "id": "paper:arxiv:2601.15074",
+      "title": "SmartOracle - An Agentic Approach to Mitigate Noise in Differential Oracles",
+      "authors": [
+        "Srinath Srinivasan",
+        "Tim Menzies",
+        "Marcelo d’Amorim"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2601.15074",
+      "s2_paper_id": "ceb1c8e11371f1b7b021fd9de6588741863607c2",
+      "url": "https://arxiv.org/abs/2601.15074",
+      "open_access_pdf": null,
+      "abstract": "Differential fuzzers detect bugs by executing identical inputs across distinct implementations of the same specification, such as JavaScript interpreters. Validating the outputs requires an oracle and for differential testing of JavaScript, these are constructed manually, making them expensive, time-consuming, and prone to false positives. Worse, when the specification evolves, this manual effort must be repeated. Inspired by the success of agentic systems in other SE domains, this paper introduces SmartOracle. SmartOracle decomposes the manual triage workflow into specialized Large Language Model (LLM) sub-agents. These agents synthesize independently gathered evidence from terminal runs and targeted specification queries to reach a final verdict. For historical benchmarks, SmartOracle achieves 0.84 recall with an 18% false positive rate. Compared to a sequential Gemini 2.5 Pro baseline, it improves triage accuracy while reducing analysis time by 4$\\times$ and API costs by 10$\\times$. In active fuzzing campaigns, SmartOracle successfully identified and reported previously unknown specification-level issues across major engines, including bugs in V8, JavaScriptCore, and GraalJS. The success of SmartOracle's agentic architecture on Javascript suggests it might be useful other software systems- a research direction we will explore in future work.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2601.15074",
+              "source_type": "paper_citation_context",
+              "excerpt": "Research in this domain demonstrates that by endowing agents with persistent memory, explicit objectives, and access to external tools or APIs, these systems can manage significantly more complex and dynamic workflows than prompt chaining alone [19, 54].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e663470125f83e273fad41528143a6c2e3b7162057a0f35c2785fbd3500eba22",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/ceb1c8e11371f1b7b021fd9de6588741863607c2",
+        "source_type": "paper",
+        "content_sha256": "sha256:d1c0277f7bbfb17e520d7bbd4d4d957bd2288d5bd87cbfa5b33fe9323a7d1f55"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3819518.3819547",
+      "title": "Testing Graph Databases via Transformations Between Fixed-Length and Variable-Length Queries",
+      "authors": [
+        "Jin-Xin Gui",
+        "Yuanhong Lan",
+        "Longlong Lu",
+        "Yifei Lu",
+        "Minxue Pan"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3819518.3819547",
+      "arxiv_id": null,
+      "s2_paper_id": "64f9059bdd2cdc2dd0b46ca064ac5e59a2ce361b",
+      "url": "https://doi.org/10.14778/3819518.3819547",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "norec",
+        "tlp",
+        "cert",
+        "coddtest"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp",
+            "cert",
+            "coddtest"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/64f9059bdd2cdc2dd0b46ca064ac5e59a2ce361b",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Cardinality Estimation Restriction Testing (CERT), Constant-Optimization-Driven Testing (CODDTest).",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/64f9059bdd2cdc2dd0b46ca064ac5e59a2ce361b",
+        "source_type": "paper",
+        "content_sha256": "sha256:5fa9b78c854ea67591b23aea76747d95d3ee63a0b9e52f9142e18f06078ded0f"
+      },
+      "abstract": "The ability of Graph Database Management Systems (GDBMSs) to efficiently store and query graph data has led to their widespread success. Unlike relational databases, GDBMSs model data as graphs and support expressive queries through graph traversal. Among the core functionalities, fixed-length and variable-length queries are particularly critical, as they underscore fundamental differences from traditional relational query execution. However, the correctness of such queries is notoriously difficult to ensure due to the intricate query semantics and the complexity of underlying optimizations like worst-case optimal joins. This paper presents a novel metamorphic testing approach named PATHTest that exploits result-equivalent transformations between fixed-length and variable-length queries. Specifically, PATHTest incorporates an iterative query generator that supports the generation of diverse and non-empty variable-length queries. During the mutation process, three transformation rules embedded within PATHTest help capture result-equivalent patterns between fixed- and variable-length queries, enhancing its capability to uncover both logical bugs and unexpected errors. Extensive evaluation on PATHTest across seven real-world, widely-used GDBMSs demonstrates the superiority of PATHTest, with 41 previously unknown bugs revealed, among which 24 are logic bugs, and 17 correspond to unexpected errors. To note, all 41 bugs are beyond the reach of the seven existing state-of-the-art testing approaches. By now, 29 of the 41 bugs have been confirmed, with 11 already fixed. Such evaluation results demonstrate the effectiveness and uniqueness of PATHTest in detecting bugs missed by existing testing approaches, contributing to the reliability of modern GDBMSs."
+    },
+    {
+      "id": "paper:arxiv:2603.00311",
+      "title": "Towards the Systematic Testing of Regular Expression Engines",
+      "authors": [
+        "Berk Çakar",
+        "Dongyoon Lee",
+        "James C. Davis"
+      ],
+      "year": 2026,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2603.00311",
+      "s2_paper_id": "ff9b5fa596fbff921f259f1c42d94dff78dc9a9e",
+      "url": "https://arxiv.org/abs/2603.00311",
+      "open_access_pdf": null,
+      "abstract": "Software engineers use regular expressions (regexes) across a wide range of domains and tasks. To support regexes, software projects must integrate a regex engine, whether provided natively by the language runtime (e.g., Python's re) or included as an external dependency (e.g., PCRE). However, these engines may contain bugs and introduce vulnerabilities. A common strategy for testing regex engines involves differential testing -- comparing outputs across different implementations. However, this approach is concerning because regex syntax and semantics vary significantly between dialects (e.g., POSIX vs. PCRE). Fuzzing is also utilized to ease testing of feature-rich regex implementations to expose defects, but naive byte-level mutations generate syntactically invalid inputs that exercise only parsing logic, not matching internals. In this work, we describe our progress towards ReTest, a framework that systematically tests regular expression engines by combining grammar-aware fuzzing for high code coverage with metamorphic testing to generate dialect-independent test oracles. So far, we have surveyed testing practices across 22 regex engines, analyzed 1,007 regex engine bugs and 156 CVEs to characterize failure modes, and curated 16 metamorphic relations for regexes derived from Kleene algebra. Our preliminary evaluation on PCRE shows that ReTest achieves 3x higher edge coverage than existing fuzzing approaches and has identified three new memory safety defects. We conclude by describing our next steps toward our ultimate goal: helping regex engine developers identify bugs without depending on a consistent cross-implementation standard.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2603.00311",
+              "source_type": "paper_citation_context",
+              "excerpt": "For DBMS, Rigger and Su [48, 49, 50] used metamorphic oracles to detect logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5aa9148a36fae05217faa538ff153f9db7d3c6335868719cfdbee72ca62e4760",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/ff9b5fa596fbff921f259f1c42d94dff78dc9a9e",
+        "source_type": "paper",
+        "content_sha256": "sha256:5583cecc545b7777c5df9db76e7093ee283f1497ed70d4c6a8b99a18a1d706d1"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3798245",
+      "title": "Understanding and Finding JIT Compiler Performance Bugs",
+      "authors": [
+        "Zijian Yi",
+        "Chen Ding",
+        "August Shi",
+        "Milos Gligoric"
+      ],
+      "year": 2026,
+      "venue": "Proceedings of the ACM on Programming Languages",
+      "doi": "10.1145/3798245",
+      "arxiv_id": "2603.06551",
+      "s2_paper_id": "c55ef302233eb0c1ccc6c9677eda03b16089b148",
+      "url": "https://doi.org/10.1145/3798245",
+      "open_access_pdf": "https://doi.org/10.1145/3798245",
+      "abstract": "Just-in-time (JIT) compilers are key components for many popular programming languages with managed runtimes (e.g., Java and JavaScript). JIT compilers perform optimizations and generate native code at runtime based on dynamic profiling data, to improve the execution performance of the running application. Like other software systems, JIT compilers might have software bugs, and prior work has developed a number of automated techniques for detecting functional bugs (i.e., generated native code does not semantically match that of the original code). However, no prior work has targeted JIT compiler performance bugs, which can cause significant performance degradation while an application is running. These performance bugs are challenging to detect due to the complexity and dynamic nature of JIT compilers. In this paper, we present the first work on demystifying JIT performance bugs. First, we perform an empirical study across four popular JIT compilers for Java and JavaScript. Our manual analysis of 191 bug reports uncovers common triggers of performance bugs, patterns in which these bugs manifest, and their root causes. Second, informed by these insights, we propose layered differential performance testing, a lightweight technique to automatically detect JIT compiler performance bugs, and implement it in a tool called Jittery. We incorporate practical optimizations into Jittery such as test prioritization, which reduces testing time by 92.40% without compromising bug-detection capability, and automatic filtering of false-positives and duplicates, which substantially reduces manual inspection effort. Using Jittery, we discovered 12 previously unknown performance bugs in the Oracle HotSpot and Graal JIT compilers, with 11 confirmed and 6 fixed by developers.",
+      "cites_seed_techniques": [
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/c55ef302233eb0c1ccc6c9677eda03b16089b148",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Cardinality Estimation Restriction Testing (CERT).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/c55ef302233eb0c1ccc6c9677eda03b16089b148",
+        "source_type": "paper",
+        "content_sha256": "sha256:c53536793f6999f5ab81c15d6d1385bf392a9fd77983646991d382053fcd18cf"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3779212.3790244",
+      "title": "Understanding Query Optimization Bugs in Graph Database Systems",
+      "authors": [
+        "Yuyu Chen",
+        "Zhongxing Yu"
+      ],
+      "year": 2026,
+      "venue": "International Conference on Architectural Support for Programming Languages and Operating Systems",
+      "doi": "10.1145/3779212.3790244",
+      "arxiv_id": null,
+      "s2_paper_id": "fd1257a7345b8ddc8fe8ed0344f300577679daab",
+      "url": "https://doi.org/10.1145/3779212.3790244",
+      "open_access_pdf": "https://doi.org/10.1145/3779212.3790244",
+      "abstract": "Recent years have witnessed an ever-growing usage of graph database management systems (GDBMSs) in various data-driven applications. Query optimization aims to improve the performance of database queries by identifying the most efficient way to execute them, and is an important stage of GDBMS workflow. Like other sophisticated systems, such as compilers, the query optimization process is complex and its implementation is prone to bugs. This paper conducts the first characteristic study of query optimization bugs in GDBMSs, including the root causes, manifestation methods, and fix strategies, and delivers 10 novel and important findings about them. Based on the characteristic study, we also developed a testing tool tailored to uncover GDBMS query optimization bugs, and the tool found 20 unique GDBMS bugs, 10 of which are query optimization bugs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790244",
+              "source_type": "paper_citation_context",
+              "excerpt": "For optimization bugs due to inaccurate cost estimations or defective plan space exploration algorithms, previous works on DBMS testing [11, 21, 45, 46] also frequently report them.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4932fe614a9c85f3aa11b66ba65384500bf0ea184175401123c18d85ae8c3bb3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/fd1257a7345b8ddc8fe8ed0344f300577679daab",
+        "source_type": "paper",
+        "content_sha256": "sha256:0381fa7c700e737b2a2f2bcbca9566556c9b607a425fae8b0d0de70a1b6b34f8"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icde65706.2026.00180",
+      "title": "VIREO: Human-in-the-Loop DBMS Fuzzing with Visualization and LLM Support",
+      "authors": [
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Jingzhou Fu",
+        "Chi Zhang",
+        "Runpei Miao",
+        "Zhuo Su",
+        "Yu Jiang",
+        "Shuai Ma"
+      ],
+      "year": 2026,
+      "venue": "IEEE International Conference on Data Engineering",
+      "doi": "10.1109/icde65706.2026.00180",
+      "arxiv_id": null,
+      "s2_paper_id": "c487d2eac342b4a4a4991d00b5bc5f3aabf3c36e",
+      "url": "https://doi.org/10.1109/icde65706.2026.00180",
+      "open_access_pdf": null,
+      "abstract": "DBMSs are critical infrastructures that directly affect the security and correctness of many dependent applications. Recent advances in DBMS fuzzing have improved coverage and efficiency, discovering many bugs. However, in practice, many DBMS fuzzers explore only a limited state space, as many functionalities involve complex constraints that require specific sequences of SQL statements and configurations, which automated methods struggle to satisfy. Moreover, many testing approaches lack an understanding of the DBMS testing state and typically run under a single configuration with fixed fuzzing strategies. Consequently, as testing progresses, coverage growth slows, and potential bugs may remain undetected. In this paper, we propose VIREO, a human-in-the-loop DBMS fuzzing framework with visualization and LLM support with two stages. In the preparation stage, VIREO analyzes source code and documentation, leveraging LLMs to construct mappings between each function module and its associated grammar rules and configurations. During the fuzzing stage, VIREO visualizes module-level coverage and bug distributions to reveal potential test boundaries. Test engineers use their domain knowledge to select modules for focused exploration. VIREO then provides the corresponding grammar rules and configuration adjustments, which engineers review to guide the fuzzer, and may also use to construct SQL queries to further enhance testing of the target modules. We have implemented VIREO and evaluated it on 5 welltested DBMSs like PostgreSQL and MySQL. VIREO discovered a total of 25 previously unknown bugs, all of which were confirmed, and 10 have been fixed. Developers provided positive feedback. For example, PostgreSQL developers acknowledged that one identified bug exposed a planner assumption “previously believed unreachable by crafted input.”",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00180",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [57] utilizes multiple test oracles [7, 58, 59, 60] to detect logic bugs, generating queries based on these oracles.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2d746afca80bfaa8f8dddebc18a88870bc773432aff26fe3c647e4baf7db3fda",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65706.2026.00180",
+              "source_type": "paper_citation_context",
+              "excerpt": "Although SQLancer was designed for logic bugs, it can also expose crashes and is evaluated using its default PQS oracle [60].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bd51c6f888951f4fb7d0f07d8551d28130da3ae2353ca8277c809dc7427baccd",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00180.json",
+              "source_type": "paper",
+              "excerpt": "We tried our best to compare VIREO against state-of-the-art DBMS fuzzers, including SQLsmith [61], SQLancer [57], SQUIRREL [70], and LEGO [34].",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, A Evaluation Setup, page 8.",
+              "content_sha256": "sha256:810b88b0ecb6367aea7e4fcaafceaf841cdca934e3697dbe1bd3a91ed0998189",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00180.json",
+              "source_type": "paper",
+              "excerpt": "Although SQLancer was designed for logic bugs, it can also expose crashes and is evaluated using its default PQS oracle [60].",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, A Evaluation Setup, page 8.",
+              "content_sha256": "sha256:bd51c6f888951f4fb7d0f07d8551d28130da3ae2353ca8277c809dc7427baccd",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00180.json",
+              "source_type": "paper",
+              "excerpt": "Table II presents the branch coverage results, showing that VIREO achieved in 3 hours results that surpassed the 24-hour runs of SQLsmith, SQLancer, SQUIRREL, and LEGO by 98%, 59%, 38%, and 15%, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M13 in the paper's extracted text, D Compared to Existing DBMS Testing Techniques, page 10.",
+              "content_sha256": "sha256:2e32a59cadb9525997f83c8090204a836148da53daf1ecca5633a492cf7ac99e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00180.json",
+              "source_type": "paper",
+              "excerpt": "The results indicate that VIREO detected 9, 9, 8, and 7 more unique bugs than SQLsmith, SQLancer, SQUIRREL, and LEGO, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M18 in the paper's extracted text, D Compared to Existing DBMS Testing Techniques, page 10.",
+              "content_sha256": "sha256:6008a719899a39b5619a40d3724787564af23be43647d578353d34c23506657d",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde65706_2026_00180.json",
+              "source_type": "paper",
+              "excerpt": "We tried our best to compare VIREO against state-of-the-art DBMS fuzzers, including SQLsmith [61], SQLancer [57], SQUIRREL [70], and LEGO [34].",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, A Evaluation Setup, page 8.",
+              "content_sha256": "sha256:810b88b0ecb6367aea7e4fcaafceaf841cdca934e3697dbe1bd3a91ed0998189",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/c487d2eac342b4a4a4991d00b5bc5f3aabf3c36e",
+        "source_type": "paper",
+        "content_sha256": "sha256:7657ebcd3f7c05e3d5399454edc802478ccc01b6147a419690384718b7a934b3"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3803437.3806090",
+      "title": "Why Property-Based Testing is Necessary for Data Intensive Scalable Computing",
+      "authors": [
+        "Yaoxuan Wu",
+        "I. Lee",
+        "Ahmad Humayun",
+        "Muhammad Ali Gulzar",
+        "Miryung Kim"
+      ],
+      "year": 2026,
+      "venue": "SIGSOFT FSE Companion",
+      "doi": "10.1145/3803437.3806090",
+      "arxiv_id": null,
+      "s2_paper_id": "ba02b6d2cf64ddf05e51681950d9094399769829",
+      "url": "https://doi.org/10.1145/3803437.3806090",
+      "open_access_pdf": "https://doi.org/10.1145/3803437.3806090",
+      "abstract": "Data-intensive scalable computing (DISC) frameworks such as Apache Spark, Flink, Beam, and Dask underpin many modern analytics workloads by providing high-level programming models and scalable runtimes. Despite their widespread adoption, framework bugs remain common, and many manifest as silent wrong results rather than crashes. Property-based testing (PBT) has been successful across a range of domains by using semantic contracts as executable oracles when ground-truth outputs are difficult to obtain. We argue that DISC calls for a general and extensible PBT framework that provides reusable property templates and supports systematic instantiation across different DISC systems.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/ba02b6d2cf64ddf05e51681950d9094399769829",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/ba02b6d2cf64ddf05e51681950d9094399769829",
+        "source_type": "paper",
+        "content_sha256": "sha256:5cb3c93bc885526682313398c2c4986c28c07f92e63f44ef14d6f7551afb2555"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/iemcon67450.2025.11381190",
+      "title": "A Comparative Survey of Mutation Testing Across Large Language Models, REST APIs, and Database Engines",
+      "authors": [
+        "B. Falah",
+        "Hayat Routaib"
+      ],
+      "year": 2025,
+      "venue": "IEEE Annual Information Technology, Electronics and Mobile Communication Conference",
+      "doi": "10.1109/iemcon67450.2025.11381190",
+      "arxiv_id": null,
+      "s2_paper_id": "d02b3d0750a816d0b2044ff55b82b10e0b80046f",
+      "url": "https://doi.org/10.1109/iemcon67450.2025.11381190",
+      "open_access_pdf": null,
+      "abstract": "Mutation testing evaluates test suite quality by introducing artificial faults and checking detection effectiveness. This paper surveys its application across three domains, Large Language Models (LLMs), REST APIs, and Database Engines, through analysis of 30 peer-reviewed studies. We compare mutation operator design, tool support, and evaluation methods across domains. Results show that LLM-focused mutation testing re-mains early-stage with emphasis on semantic and fairness-aware faults, REST APIs benefit from adaptive schema-based tools, and database engines exploit grammar-and query plan–guided mutations to reveal semantic and performance issues. The study highlights trends, strengths, and limitations, offering insights for improving robustness and guiding future research.",
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/iemcon67450.2025.11381190",
+              "source_type": "paper_citation_context",
+              "excerpt": "Ba et al. [16] proposed query plan–guided mutation strategies, while Schafer et al. developed Parser-Knows-Best [17], using grammar-rule traversal for test input mutation.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f004d2cc464c259c44d951528ae9370e2d6739ece9c7700b60f73dd4517466ab",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/d02b3d0750a816d0b2044ff55b82b10e0b80046f",
+        "source_type": "paper",
+        "content_sha256": "sha256:ca5a628d8a2b99558dd0e2768aaf09974d0308360e84184f19d6b0b7bb1e3c4c"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/tse.2025.3625300",
+      "title": "A Comprehensive Study of Bugs in Relational DBMS",
+      "authors": [
+        "Shuang Liu",
+        "Ruifeng Wang",
+        "Yuanfeng Xie",
+        "Junjie Chen",
+        "Wei Lu",
+        "Xiao Zhang",
+        "Quanqing Xu",
+        "Chuanhui Yang",
+        "Xiaoyong Du"
+      ],
+      "year": 2025,
+      "venue": "IEEE Transactions on Software Engineering",
+      "doi": "10.1109/tse.2025.3625300",
+      "arxiv_id": null,
+      "s2_paper_id": "616ddf130a678c2727c4c8bc791d8e3a1eaf3bf7",
+      "url": "https://doi.org/10.1109/tse.2025.3625300",
+      "open_access_pdf": null,
+      "abstract": "Relational Database Management Systems (RDBMSs) are crucial infrastructures supporting a wide range of applications, making bug mitigation within these systems essential. This study presents the first comprehensive analysis of bugs in three popular open-source RDBMSs—MySQL, SQLite, and openGauss. We manually examined 777 bugs across four dimensions, i.e., bug root causes, bug symptoms, bug distribution across modules, and the correlations between the studied aspects. We also analyzed the bug-triggering SQL statements to uncover test cases that cannot be generated by existing tools. We have made 12 findings, which throw lights on the development, maintenance and testing of RDBMS systems. Particularly, our findings reveal that bugs related to SQL data types and complex features, such as database triggers, procedures and database parameter settings, present significant opportunities for enhancing RDBMS bug detection and mitigation. Leveraging these insights, we developed a tool, SQLT, which effectively identified eight RDBMS bugs (five type-related), all verified by developers, with four subsequently fixed.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+              "source_type": "paper_citation_context",
+              "excerpt": "We’ve also compared SQLT with SQLancer on the latest version of MySQL and SQLite (MySQL 8.0.34 and SQLite 3.43.0) for 24 hours and adopt NoREC [14] as the default oracle to detect bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3cd3c3676cfd3eb0a5d72606dd93ce8baf2bcb986e99fd033538e4f44baa364b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+              "source_type": "paper_citation_context",
+              "excerpt": "…related to SQL types often result in result inconsistencies—silent bugs without explicit error messages— we incorporated two existing metamorphic testing approaches, NOREC [14] and TLP [19], as oracles within SQLT. SQLT is publicly available and can be accessed at https://github.com/ sqlttest/SQLT.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:97fbca1ad9148438ab5ff0682b425153a325c897f250b3bc3a9ab814a4ef05cf",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+              "source_type": "paper_citation_context",
+              "excerpt": "Bug Type Duration # Bugs Fonseca [22] concurrency bugs 2003-2009 80 Cui [23] transaction bugs 2018-2022 140 ours general bugs 2021-2023 777 [12], [13], [14], [15], [16], [17], [18], [19], [20], [21] aiming at detecting bugs in RDBMSs, and they successfully detected new bugs in RDBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:660423c19040bfcec8b97751c0eb1dc0fee8e80b8b1d56cf59af0d4ddea5de45",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+              "source_type": "paper_citation_context",
+              "excerpt": "We first consulted the DB-Engines Ranking [24] to identify ranked and widely adopted RDBMS, and surveyed existing testing approaches [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [25], [26] as well as empirical studies on RDBMS bugs [23], [27].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a1317c0e1443c368e4a51e418f184bb8862eaa584c6e5644439ea643c2efc435",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+              "source_type": "paper_citation_context",
+              "excerpt": "Rigger et al. [12], [14], [19] proposed SQLancer, which is one of the most popular generation-based RDBMS testing approach, and it incorporates three metamorphic methods, namely PQS [12], NoREC [14], and TLP [19], as oracles to detect logical bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:016191dc51355b2dadc30a686fcf032f9713ca565867f2333a9f1f863278b456",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3625300",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLT is an extension of the widely-used database testing tool, SQLancer [12].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0a4153ce3436c61aeba7ea1d2a024c029e6a95f2624087f08738017dd9cafcf1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3625300.json",
+              "source_type": "paper",
+              "excerpt": "Design and Motivation of SQLTSQLT is an extension of the widely-used database testing tool, SQLancer [12].",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, V APOOF-OF- CONCEPT TOOLSQLT, page 13.",
+              "content_sha256": "sha256:e58d7ef5e90b4675e23ee63572c557e2c738d71b52bbd37d40b9fd1e794a3bd3",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3625300.json",
+              "source_type": "paper",
+              "excerpt": "SQLT extends SQLancer by introducing new data types such as BIT and JSON, which were undersupported by SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, V APOOF-OF- CONCEPT TOOLSQLT, page 13.",
+              "content_sha256": "sha256:bdebbe5631b1603186defd8f5a6a96c81bf010bdc00cb4bf0e49a2f971a0cbae",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3625300.json",
+              "source_type": "paper",
+              "excerpt": "Given that bugs related to SQL types often result in result inconsistencies—silent bugs without explicit error messages—we incorporated two existing metamorphic testing approaches, NOREC [14] and TLP [19], as oracles within SQLT.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, V APOOF-OF- CONCEPT TOOLSQLT, page 13.",
+              "content_sha256": "sha256:2f94bf3181e7fd1a2cc7c1f55e83c235f88ede39cced94e60d0df597f4ca1dd3",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3625300.json",
+              "source_type": "paper",
+              "excerpt": "We’ve also compared SQLT with SQLancer on the latest version of MySQL and SQLite (MySQL 8.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, B Evaluation of SQLT, page 13.",
+              "content_sha256": "sha256:afa9fb8501e72b249a27f3c289e2ed815f30e5be7993b9a87d5be4777ff30683",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3625300.json",
+              "source_type": "paper",
+              "excerpt": "SQLT detects 6 bugs and SQLancer detects 4 bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "M13 in the paper's extracted text, B Evaluation of SQLT, page 14.",
+              "content_sha256": "sha256:aa613af591fdf4f224e548219fc436ad3983cc8ae7266a7da90c225a8ea2522b",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3625300.json",
+              "source_type": "paper",
+              "excerpt": "Among these results, SQLT uncovered 1 additional bug in MySQL that were not detected by SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, B Evaluation of SQLT, page 14.",
+              "content_sha256": "sha256:9d42f0fc2fee0d1d0f0ad56d7ca9b452b7818950985b7cb9113fa50de8365ea9",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/616ddf130a678c2727c4c8bc791d8e3a1eaf3bf7",
+        "source_type": "paper",
+        "content_sha256": "sha256:41fb76e1197ee82960fe0f1f6e2529e40210a7e8c772fdd887dd3af4ecf98037"
+      }
+    },
+    {
+      "id": "paper:arxiv:2511.17377",
+      "title": "Anomaly Pattern-guided Transaction Bug Testing in Relational Databases",
+      "authors": [
+        "Hui-Rong Xu",
+        "Shuang Liu",
+        "Xianyu Zhu",
+        "Qiyu Zhuang",
+        "Wei Lu",
+        "Xiaoyong Du"
+      ],
+      "year": 2025,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2511.17377",
+      "s2_paper_id": "3987779a8bc9037eedb2b036f804413c2f1b1da9",
+      "url": "https://arxiv.org/abs/2511.17377",
+      "open_access_pdf": null,
+      "abstract": "Concurrent transaction processing is a fundamental capability of Relational Database Management Systems (RDBMSs), widely utilized in applications requiring high levels of parallel user interaction, such as banking systems, e-commerce platforms, and telecommunications infrastructure. Isolation levels offer a configurable mechanism to manage the interaction between concurrent transactions, enabling varying degrees of consistency and performance trade-offs. These isolation guarantees are supported by all major RDBMSs. However, testing transaction behavior under different isolation levels remains a significant challenge due to two primary reasons. First, automatically generating test transactions that can effectively expose bugs in transaction handling logic is non-trivial, as such bugs are typically triggered under specific transactional constraints. Second, detecting logic anomalies in transaction outcomes is difficult because the correct execution results are often unknown for randomly generated transactions. To address these challenges, we propose an anomaly pattern-guided testing approach for uncovering transaction bugs in RDBMSs. Our solution tackles the first challenge by introducing a test case generation technique guided by predefined anomaly patterns, which increases the likelihood of exposing transactional bugs. For the second challenge, we present a two-phase detection process, involving explicit error detection and implicit error detection, to identify bugs in transaction execution. We have implemented our approach in a tool, APTrans, and evaluated it on three widely-used RDBMSs: MySQL, MariaDB, and OceanBase. APTrans successfully identified 13 previously unknown transaction-related bugs, 11 of which have been confirmed by the respective development teams.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2511.17377",
+              "source_type": "paper_citation_context",
+              "excerpt": "Recent innovations have introduced paradigm shifts in testing strategies: QPG [7] uses query plan-guided database state mutation,while DQE[28]appliesdifferentialexecutionanalysisacross SELECT,UPDATE,and DELETEstatements.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:465b4010ad0337e3d2c1aae103c91b78c49b08adca2a4f4b15414df9aa1ecd8b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2511.17377",
+              "source_type": "paper_citation_context",
+              "excerpt": "Its successor, SQLancer [25–27], introduces three metamorphic testing oracles: PQS [27], NoREC [25], and TLP [26], significantly improving its ability to detect logical bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6aa30f7b59b6c7b33c61fb0c1947cec177de9cbb396794f1ca07126f941cc21f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2511_17377.json",
+              "source_type": "paper",
+              "excerpt": "We develop our database and SQL statement generation method based on SQLancer [ 4] and extend it to support table join operations for both database generation and SQL statement generation.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 3.2 Database and SQL statement Generation, page 4.",
+              "content_sha256": "sha256:d60eef1c57d7f4fd09dde061dd434633c71a4cff7f66603ca3cabadd23428aad",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2511_17377.json",
+              "source_type": "paper",
+              "excerpt": "To simulate real-world scenarios, which usually involve multiple tables connected with join keys, we extend SQLancer to support table join operations for both database generation and SQL statement generation.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, 3.2.1 Database Generation. In database transaction testing, the, page 4.",
+              "content_sha256": "sha256:5f63771a86b543cede0dfb03d147351049140c07abffa167584a51c57a4bcf7d",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2511_17377.json",
+              "source_type": "paper",
+              "excerpt": "Building upon the SQLancer framework, we have extended the SQL statement generation to include JOIN operations and type constraints, thereby producing a more diverse set of SQL statements while ensuring the semantic correctness of the SQL statements.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 3.2.1 Database Generation. In database transaction testing, the, page 4.",
+              "content_sha256": "sha256:ebbf44183967dfeae410f8dba8418a28cad537a5abb54f5216f7caa4eabe86e5",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2511_17377.json",
+              "source_type": "paper",
+              "excerpt": "We use SQLancer to generate a large number of random SQL statements for transaction generation.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, 3.2.1 Database Generation. In database transaction testing, the, page 4.",
+              "content_sha256": "sha256:ce6cb4d61e802f29c734a45b5fb0de4825c8a9386c5ddcd70eb0d73a38d135eb",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/3987779a8bc9037eedb2b036f804413c2f1b1da9",
+        "source_type": "paper",
+        "content_sha256": "sha256:01e0ebd9a013eaf466b7eb3d30dc9285f7fd295a1f1056b52b755566ee75d295"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/ase63991.2025.00151",
+      "title": "ARG: Testing Query Rewriters via Abstract Rule Guided Fuzzing",
+      "authors": [
+        "Dawei Li",
+        "Yuxiao Guo",
+        "Qifan Liu",
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Jingzhou Fu",
+        "Chi Zhang",
+        "Yu Jiang"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Automated Software Engineering",
+      "doi": "10.1109/ase63991.2025.00151",
+      "arxiv_id": null,
+      "s2_paper_id": "eeb659ea2fc76f85e0eb72e54dd83f98ce695d61",
+      "url": "https://doi.org/10.1109/ase63991.2025.00151",
+      "open_access_pdf": null,
+      "abstract": "Query rewriters transform a query into a more efficient yet semantically equivalent form, which is vital for optimizing query execution. Despite its importance, query rewriting is inherently complex, influenced by factors including rewrite rule design, rule interactions, and semantic preservation. Consequently, its implementation struggles to prevent problems, which may result in system crashes or incorrect query results. Existing DBMS testing approaches are generally designed for broad bug detection. However, due to the diversity of rewrite rules, they cover only a limited subset of rewrite scenarios, potentially overlooking critical bugs.In this paper, we propose Abstract Rule Guided (ARG) fuzzing to detect bugs in query rewrites. The key idea is to use feedback from abstract rules to guide query generation, thereby activating more rewriting logic and enhancing bug detection. Abstract rules provide a unified representation of the patterns (e.g., AST structures and related constraints) that trigger rewrites, as well as the resulting transformations. We track abstract rules to identify which patterns have been covered. This feedback is then used to dynamically adjust query generation, prioritizing unexplored patterns to avoid redundancy and expose more rewriting logic. We implemented ARG to test four popular query rewrites, namely Apache Calcite, WeTune, SQLSolver, and LearnedRewrite. ARG discovered 38 previously unknown bugs, consisting of 4 crashes, 13 invalid SQL outputs, and 21 semantic deviations. Among them, 19 have been confirmed, while the remaining cases are still under investigation. We also compared ARG against popular DBMS testing tools. In 24 hours, ARG triggered 76% and 1017% more written rules, triggered 13 and 15 more bugs than SQLsmith and SQLancer, respectively.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/ase63991.2025.00151",
+              "source_type": "paper_citation_context",
+              "excerpt": "Recent works have also adopted metamorphic testing to construct semantically equivalent queries to test DBMSs. NoREC [18] converts an optimizable query into a non-optimizable form of the query, then identifies logic bugs by comparing their execution results.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9d1695d0eeb753a98fd286d78c69fa5b0de4b65eff35c2ddcf55ef7761f7a0e7",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/ase63991.2025.00151",
+              "source_type": "paper_citation_context",
+              "excerpt": "Similarly, TLP [19] employs ternary logic to generate three equivalent queries combined via UNION operations.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4b2917f8cc5cc1ecdb45c716512b79dee74047ffc03f5071c87f86d2a7ddaf66",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_ase63991_2025_00151.json",
+              "source_type": "paper",
+              "excerpt": "To generate DDL statements and enhance abstract rule extraction, we integrated SQLancer to construct dynamic database schemas and populate test data.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, IV IMPLEMENTATION, page 6.",
+              "content_sha256": "sha256:1ebf8ade65ececfa3218143a590cc2b6a5a7047a2512457c5ff081da918dc0b3",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_ase63991_2025_00151.json",
+              "source_type": "paper",
+              "excerpt": "In 24 hours, ARG triggered 76% and 1017% more written rules, triggered 13 and 15 more bugs than SQLsmith and SQLancer, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, page 1.",
+              "content_sha256": "sha256:72d32d83190a379611dbeaba42165d0b411a9ca2d77416a9ea5d4ac14b45a91c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_ase63991_2025_00151.json",
+              "source_type": "paper",
+              "excerpt": "In 24 hour experiment on these query rewriters, ARG triggered 76% and 1017% more written rules, triggered 13 and 15 more bugs than SQLsmith and SQLancer, respectively, In summary, we make the following contributions: •We observe that while query rewriters are widely used in DBMSs, bugs persist and can cause serious issues, yet effective testing tools remain lacking.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, I INTRODUCTION, page 2.",
+              "content_sha256": "sha256:54146db1ca78ef4f6e3d7a43077ed5ada6ce7bfa02a660a6d16e7cc5dc7d602f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/eeb659ea2fc76f85e0eb72e54dd83f98ce695d61",
+        "source_type": "paper",
+        "content_sha256": "sha256:c6b4bb01d47dc32447d0f0b0d94ffb92f05fbdc738c4046ab6badb7ebc5c021b"
+      }
+    },
+    {
+      "id": "paper:arxiv:2509.10819",
+      "title": "Arguzz: Testing zk VMs for Soundness and Completeness Bugs",
+      "authors": [
+        "Christoph Hochrainer",
+        "Valentin Wüstholz",
+        "Maria Christakis"
+      ],
+      "year": 2025,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2509.10819",
+      "s2_paper_id": "98f3b2e558fdcfd3f05ceb53c370f6339f72e0d9",
+      "url": "https://arxiv.org/abs/2509.10819",
+      "open_access_pdf": null,
+      "abstract": "Zero-knowledge virtual machines (zk VMs) are increasingly deployed in decentralized applications and blockchain rollups since they enable verifiable off-chain computation. These VMs execute general-purpose programs, frequently written in Rust, and produce succinct cryptographic proofs. However, zk VMs are complex, and bugs in their constraint systems or execution logic can cause critical soundness (accepting invalid executions) or completeness (rejecting valid ones) issues. We present Arguzz, the first automated tool for testing zk VMs for soundness and completeness bugs. To detect such bugs, Arguzz combines a novel variant of metamorphic testing with fault injection. In particular, it generates semantically equivalent program pairs, merges them into a single Rust program with a known output, and runs it inside a zk VM. By injecting faults into the VM, Arguzz mimics malicious or buggy provers to uncover overly weak constraints. We used Arguzz to test six real-world zk VMs (RISC Zero, Nexus, Jolt, SP1, OpenVM, and Pico) and found eleven bugs in three of them. One RISC Zero bug resulted in a $50,000 bounty, despite prior audits, demonstrating the critical need for systematic testing of zk VMs.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2509.10819",
+              "source_type": "paper_citation_context",
+              "excerpt": "Blackbox fuzzing remains a widely used and effective strategy when testing complex systems, e.g., compilers [47], database systems [39], or SMT solvers [34].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:679a22be462d924326a0130915b666b2a2363b79cedb47d40363415ade320d70",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/98f3b2e558fdcfd3f05ceb53c370f6339f72e0d9",
+        "source_type": "paper",
+        "content_sha256": "sha256:247a618c19110d8003dd5e330f69749f9d4df174f2f73082a9df62f10ef25db3"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/punecon67554.2025.11378586",
+      "title": "AskDB: AI-Assisted Natural Language Interface for Database Querying and Visualization",
+      "authors": [
+        "S. Shilaskar",
+        "Prapti Dongaonkar",
+        "Pruthavik Gavali",
+        "Suhani Gunje"
+      ],
+      "year": 2025,
+      "venue": "2025 IEEE Pune Section International Conference (PuneCon)",
+      "doi": "10.1109/punecon67554.2025.11378586",
+      "arxiv_id": null,
+      "s2_paper_id": "5c51d8d0d0446261f16acc8503d8cf2c7db0b75e",
+      "url": "https://doi.org/10.1109/punecon67554.2025.11378586",
+      "open_access_pdf": null,
+      "abstract": "The growing reliance on data-driven insights in sectors around the world has created an ongoing problem. Many non-technical users struggle to effectively query and analyze databases with the complexities of Structured Query Language (SQL), limiting their ability to properly access data and significantly slowing down the decision-making process. This system proposed AskDB, an intelligent neural text-to-SQL system that enables users to use natural language queries as input in order to interact with a database. The system automatically translates user queries into valid SQL and executes them against the connected database, returning results to the user in tabular form and easily interpreted graphical visualization. AskDB's use of large language model (LLM) optimizes the translation of the user's input into SQL commands while contextually comprehending the intent of the query to create more accurate translations. The frontend was developed in React to create a responsive, user-friendly experience, while the backend was developed in Python and SQL to JS the command, process the query, and connect and run the query against the selected database. AskDB also utilizes schemadriven query correction and prompt-engineering optimization to better support each end user with more accurate outputs. Experiments represent AskDB achieves an [exact-match] accuracy of 93% - still slightly less than T5 based systems worth between 73-84% accuracy while completing each task successfully. The Proposed framework brigde gap between human language and database querying.",
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/5c51d8d0d0446261f16acc8503d8cf2c7db0b75e",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/5c51d8d0d0446261f16acc8503d8cf2c7db0b75e",
+        "source_type": "paper",
+        "content_sha256": "sha256:a5f79ca40f073c94b14cd5a39d0f3f45760c3a17ec758bde3ce17f3ab609cc6f"
+      }
+    },
+    {
+      "id": "paper:arxiv:2505.02012",
+      "title": "Automated Database Testing via LLM-Synthesized SQL Features",
+      "authors": [
+        "Suyang Zhong",
+        "Manuel Rigger"
+      ],
+      "year": 2025,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": "2505.02012",
+      "s2_paper_id": "15c3b7489312fa85f9cc692d9f3860c35975019d",
+      "url": "https://arxiv.org/abs/2505.02012",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2505.02012",
+              "source_type": "paper_citation_context",
+              "excerpt": "Sketch validation determines only whether a filled sketch is accepted by the target DBMS; semantic equivalence between the query pairs that the test oracle compares is enforced by the test oracle itself ( e.g. , TLP [32]).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:640f30bfae23a1c6e6d61fa44b0ab920cfaa244d917f5ba6d76b6f22260b5eff",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2505.02012",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [31] detects logic bugs by executing a query that is receptive to optimizations and comparing its result to an equivalent version that is unlikely to be optimized.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:775835a13a5d4f6197b980528952c15ce72c27990f0a007b7d136b77edd07940",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2505.02012",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [31– 33] also generates queries based on various hand-written SQL generators.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6f607920559575e08103e65e10b415a3788abbc10c86a4b67e9463d5f2e4234f",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2505.02012",
+              "source_type": "paper_citation_context",
+              "excerpt": "They primarily tackle the so-called test-oracle problem by validating whether a DBMS operates as expected, for example, by deriving an equivalent query from a given input query to check whether the DBMS produces consistent results [3, 31, 32, 50].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6b312dbe16c5c5d3a83872c7d4993a140e798ff98c92049aa929b99234188b0b",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2505.02012",
+              "source_type": "paper_citation_context",
+              "excerpt": "Although we also detected bugs in other popular DBMSs ( e.g. , MySQL and MariaDB), many previously reported bugs remain unfixed, making it challenging to determine whether any bug-inducing test cases trigger known bugs [14, 33, 38].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a95ea5949da1c3f1f7cc88368f277f84f8547ca0ca277465e318ff3c963f9b6b",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2505.02012",
+              "source_type": "paper_citation_context",
+              "excerpt": "Because existing approaches have exhaustively tested the target systems [1, 9], any new bugs found by ShQveL indicate cases that existing tools missed.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ece308011553678f34972c380ceb969113d5d4a0293942efbf5e124186dcb8b3",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/15c3b7489312fa85f9cc692d9f3860c35975019d",
+        "source_type": "paper",
+        "content_sha256": "sha256:aa929b11c2c425e92a84f083a907611a3994d97785a1143d40fc76a7f6ffbf3c"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Database Management Systems (DBMSs) have been tested by various automated testing approaches. Many of them generate pairs of equivalent queries to identify bugs that cause DBMSs to compute incorrect results, and have found hundreds of bugs in mature, widely used DBMSs. Most such approaches are based on manually written SQL generators; however, their bug-finding capabilities remain constrained by the limited set of SQL features supported by the generators. In this work, we propose ShQveL, an approach that augments existing SQL test-case generators by leveraging Large Language Models (LLMs) to synthesize SQL fragments. Our key idea is to systematically incorporate SQL features gained through automated interactions with LLMs into the SQL generators, increasing the features covered while efficiently generating test cases. Specifically, ShQveL uses SQL sketches---SQL statements with incomplete code segments that LLMs fill---to integrate LLM-generated content into the generator. We evaluated ShQveL on 5 DBMSs and discovered 55 unique and previously unknown bugs, 50 of which were promptly fixed after our reports."
+    },
+    {
+      "id": "paper:arxiv:2510.06663",
+      "title": "Automated Discovery of Test Oracles for Database Management Systems Using LLMs",
+      "authors": [
+        "Qiuyang Mang",
+        "Runyuan He",
+        "Suyang Zhong",
+        "Xiaoxuan Liu",
+        "Huanchen Zhang",
+        "Alvin Cheung"
+      ],
+      "year": 2025,
+      "venue": "Proceedings of the ACM on Management of Data",
+      "doi": null,
+      "arxiv_id": "2510.06663",
+      "s2_paper_id": "919f0ecf5c6ac7059a9e7d76d9841fca3b3eda74",
+      "url": "https://arxiv.org/abs/2510.06663",
+      "open_access_pdf": null,
+      "abstract": "Since 2020, automated testing for Database Management Systems (DBMSs) has flourished, uncovering hundreds of bugs in widely-used systems. A cornerstone of these techniques is\n test oracle\n , which typically implements a mechanism to generate equivalent query pairs, and subsequently runs the pair and identifies bugs by checking the consistency of their results. While running these oracles can be automated, designing the mechanism to generate equivalent queries remains a fundamentally manual endeavor. This paper explores the use of large language models (LLMs) to automate the discovery of equivalent queries in the design of test oracles, addressing a long-standing bottleneck towards fully automated DBMS testing.\n \n \n Although LLMs demonstrate impressive creativity, they are prone to hallucinations that can produce numerous false positive bug reports. Furthermore, their high monetary cost and latency mean that LLM invocations should be limited to ensure that bug detection is efficient and economical. To this end, we introduce Argus, a novel framework built upon the core concept of the\n Constrained Abstract Query\n —a SQL skeleton containing placeholders and their associated instantiation conditions, e.g., the placeholder must be filled by a Boolean column. Argus uses LLMs to generate pairs of these skeletons, with their equivalence formally proven using a SQL equivalence solver to ensure soundness. After that, the placeholders in the verified skeletons are instantiated with concrete, reusable SQL snippets that are also synthesized by LLMs to produce complex test cases. We have implemented Argus and evaluated it on five extensively tested DBMSs, discovering 41 previously unknown bugs, 36 of which are logic bugs, with 36 confirmed and 27 already fixed by the developers. The artifacts for Argus are available at https://github.com/joyemang33/Argus",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp",
+        "sqlancer_pp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert",
+            "dqp",
+            "sqlancer_pp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2510.06663",
+              "source_type": "paper_citation_context",
+              "excerpt": "Listing 2: An example of representing and instantiating TLP [45] oracle in CAQ.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:431b5424c3d7ba513e9a6edac86b0d375d37ae16113038abd9767aefefa0e758",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2510_06663.json",
+              "source_type": "paper",
+              "excerpt": "These snippets can be generated offline by a hybrid approach combining an LLM to cover diverse database features and a high-throughput generator, such as SQLancer [ 49], even though the prover currently cannot reason about them.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 1 Introduction, page 4.",
+              "content_sha256": "sha256:2917b56ed7d67c01da86e6dd0d4cde4b8874042923e1878b693deba3ce06cc49",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2510_06663.json",
+              "source_type": "paper",
+              "excerpt": "Specifically, we use SQLancer++ [ 82]’s query generator to produce seed queries, but we do not use their predefined test oracles (such as TLP [48] and NoREC [47]) for bug detection.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, 3 Argus Overview, page 5.",
+              "content_sha256": "sha256:6709d80ea18108d23c7f4463693752e7b233b0728e7fdb9c1c8f2473ec9fb959",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2510_06663.json",
+              "source_type": "paper",
+              "excerpt": ",SQLancer can be directly used for schema and concrete 4, No.",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, 5.1 Database Seeding, page 8.",
+              "content_sha256": "sha256:f3e4fb25ec48a7809093d9a931cd17846bc61790660fa6d34762c83d2053d81a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2510_06663.json",
+              "source_type": "paper",
+              "excerpt": "2 Code Coverage We comparedArguswith three DBMS logic bugs finding tools, SQLancer [ 3,47–49,74], SQLancer++ [ 82] and EET [ 24] in multiple coverage metrics.",
+              "excerpt_is_verbatim": true,
+              "note": "M30 in the paper's extracted text, 7.2 Code Coverage, page 17.",
+              "content_sha256": "sha256:555f07a787abac27b922cbca6af1b856a9cf13ad1badc02e9a7855f54ba9d302",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2510_06663.json",
+              "source_type": "paper",
+              "excerpt": "1% higher line and branch coverage than SQLancer++, respectively, and 11.",
+              "excerpt_is_verbatim": true,
+              "note": "M37 in the paper's extracted text, 7.2 Code Coverage, page 18.",
+              "content_sha256": "sha256:32a3f8ff9da16c6e941379e074bbf330930513a91cee263456cff9c3dd495cc7",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2510_06663.json",
+              "source_type": "paper",
+              "excerpt": "571×more line, function, and branch coverage than SQLancer, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M50 in the paper's extracted text, 7.2 Code Coverage, page 19.",
+              "content_sha256": "sha256:39d5e6ee5d4f310df3e7803dcbdc3f8971731cfae1499448d4e21f5c3f56dcfc",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2510_06663.json",
+              "source_type": "paper",
+              "excerpt": "As a state-of-the-art, open-source DBMS testing framework, SQLancer supports most of the latest test oracles [ 4,57,74].",
+              "excerpt_is_verbatim": true,
+              "note": "M31 in the paper's extracted text, 7.2 Code Coverage, page 17.",
+              "content_sha256": "sha256:d6df56abbd7e83e8f6a5a3b0c0a3353b9974b46395100ce5be43ac1a188d82f2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/919f0ecf5c6ac7059a9e7d76d9841fca3b3eda74",
+        "source_type": "paper",
+        "content_sha256": "sha256:d2d72b1b33f25b74ee2d15b1838cf6c7f8357ba73d24f07bd569d7f9d02c5d8d"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/joyemang33/Argus",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "sqlancer_source_in_nested_artifact"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/joyemang33/Argus",
+              "source_type": "github_repository",
+              "excerpt": "# Argus: Automated Discovery of Test Oracles for Database Management Systems Using LLMs\n\nArgus is a novel framework for automatically discovering and instantiating test oracles to find logic bugs in Database Management Systems (DBMSs) using Large Language Models (LLMs).",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/joyemang33/Argus/blob/main/MetamorphicCoverageArtifact/Code/mc-guided_fuzzing/src/sqlancer/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "MetamorphicCoverageArtifact/Code/mc-guided_fuzzing/src/sqlancer/Randomly.java is SQLancer's Randomly.java, but it sits under MetamorphicCoverageArtifact, which looks like a bundled copy of another project",
+              "content_sha256": "sha256:7f3c136a45805e0fd7b278f3c6f2693b5721e9f54426d26c66823ea7d4a8b83b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3720504",
+      "title": "Checking Observational Correctness of Database Systems",
+      "authors": [
+        "Lauren Pick",
+        "Amanda Xu",
+        "Ankush Desai",
+        "S. Seshia",
+        "Aws Albarghouthi"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Program. Lang.",
+      "doi": "10.1145/3720504",
+      "arxiv_id": null,
+      "s2_paper_id": "8e351e69fa979c4de3f4534211b50b4052a44218",
+      "url": "https://doi.org/10.1145/3720504",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3720504",
+      "abstract": "Clients rely on database systems to be correct, which requires the system not only to implement transactions’ semantics correctly but also to provide isolation guarantees for the transactions. This paper presents a clientcentric technique for checking both semantic correctness and isolation-level guarantees for black-box database systems based on observations collected from running transactions on these systems. Our technique verifies observational correctness with respect to a given set of transactions and observations for them, which holds iff there exists a possible correct execution of the transactions under a given isolation level that could result in these observations. Our technique relies on novel symbolic encodings of (1) the semantic correctness of database transactions in the presence of weak isolation and (2) isolation-level guarantees. These are used by the checker to query a Satisfiability Modulo Theories solver. We applied our tool Troubadour to verify observational correctness of several database systems, including PostgreSQL and an industrial system under development, in which the tool helped detect two new bugs. We also demonstrate that Troubadour is able to find known semantic correctness bugs and detect isolation-related anomalies.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3720504",
+              "source_type": "paper_citation_context",
+              "excerpt": "There has been much work on DBMS fuzzing for SQL using a variety of techniques [3, 15, 31, 42], such as ternary logic partitioning and mutation-based fuzzing.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:96f864975c8ecda2c487690de0314a8cd09a056b33ffa8a774bcdd587966b527",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3720504.json",
+              "source_type": "paper",
+              "excerpt": "Troubadour can verify that a trace does not demonstrate any of the classes of bugs detected by DBMS fuzzer SQLancer for the fragment of SQL that Troubadour supports as well as detect the presence of such bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, 8.2 RQ2: Classes of Bugs Considered, page 20.",
+              "content_sha256": "sha256:b233c7795179a90c08b6df9e2c7468cd3b6c9f248cd3fc8ab6b8124805d6e7d9",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3720504.json",
+              "source_type": "paper",
+              "excerpt": "We also ran the tool on the query and the original incorrect response observed by SQLancer, and Troubadour reported errors for all the examples in under a second total as well, demonstrating its ability to detect semantic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, 8.2 RQ2: Classes of Bugs Considered, page 20.",
+              "content_sha256": "sha256:2c61a0adbaa1b03a22b585f459a10884636dc1556570aa82b8ec2f50f60b795c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/8e351e69fa979c4de3f4534211b50b4052a44218",
+        "source_type": "paper",
+        "content_sha256": "sha256:1d3b252d92ac1bdb0fddd394821f6aacb9e3e759a2b132576ef41f44be2c9a45"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse55347.2025.00003",
+      "title": "Coni: Detecting Database Connector Bugs via State-Aware Test Case Generation",
+      "authors": [
+        "Wenqian Deng",
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Jingzhou Fu",
+        "Mingzhe Wang",
+        "Yu Jiang"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse55347.2025.00003",
+      "arxiv_id": null,
+      "s2_paper_id": "7307f131c3d332683815de0634412c48e6a01a7a",
+      "url": "https://doi.org/10.1109/icse55347.2025.00003",
+      "open_access_pdf": null,
+      "abstract": "Database connectors are widely used in many applications to facilitate flexible and convenient database interactions. Potential bugs in database connectors can lead to various abnormal behaviors within applications, such as returning incorrect results or experiencing unexpected connection interruption. However, existing DBMS fuzzing works cannot be directly applied to testing database connectors as they mainly focus on SQL generation and use a small subset of connector interfaces. Automated test case generation also struggles to generate effective test cases that explore intricate interactions of database connectors due to a lack of domain knowledge. The main challenge in testing database connectors is generating semantically correct test cases that can trigger various connector state transitions. To address that, we propose CONI, a framework designed for detecting logic bugs of database connectors with state-aware test case generation. First, we define the database connector state model by analyzing the corresponding standard specification. Building upon this model, Coni generates interface call sequences within test cases to encompass various state transitions. After that, Coni generates suitable parameter values based on the parameter information and contextual information collected during runtime. Then the test cases are executed on a target and a reference database connector. Inconsistent results indicate potential bugs. We evaluated CONI on 5 widely-used JDBC database connectors, namely MySQL Connector/J, MariaDB Connector/J, AWS JDBC Driver for MySQL, PGJDBC, and PG JDBC NG. In total, Coni reported 44 previously unknown bugs, of which 34 have been confirmed.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00003",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, SQL ANCER [11] generates valid SQL queries based on AST.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a12d0851c59b0e8ed7c82bda6d3de37d9aa06089dcca0853978f5ffd0fe5360c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00003",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, SQL ANCER [11] relies on JDBC solely to execute SQL queries and retrieve results, without exploring additional functionalities like modifying configuration properties or batch execution.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ac3745da1ce59252f460e62b80f26d9318915bec823f43db2a83fee68c5c2a81",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00003",
+              "source_type": "paper_citation_context",
+              "excerpt": "On the one hand, most fuzzers primarily concentrate on generating effective SQL queries [10, 11, 12, 13, 14], whereas database connectors are not directly involved in the execution of these queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c695fa21c569567d670d971046988d913217755cb49177c65f4eb4589a870c05",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00003",
+              "source_type": "paper_citation_context",
+              "excerpt": "When applying fuzzing techniques to test DBMSs, the main challenge is to generate correct and effective SQL queries [11, 12, 13, 35, 36, 37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b568d30b7b3892a6639c4d334cdfd61d764e7c331c576f05df47748016074fca",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00003",
+              "source_type": "paper_citation_context",
+              "excerpt": "In addition, SQL ANCER [11] is a popular open-source tool for testing databases using JDBC.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0b4f838afa94b33652895dfcb50047b73d8395f8a6f7a9932d40a1afa4b23d9b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00003.json",
+              "source_type": "paper",
+              "excerpt": "We implemented SQLANCER+ by adapting the target database connector and collecting results from different connectors to identify inconsistencies.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, C Comparison with Existing Techniques, page 8.",
+              "content_sha256": "sha256:a8a3946118dd8b45156ae4c37a92ef886567f896b02a5e265bea76cf75f08579",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00003.json",
+              "source_type": "paper",
+              "excerpt": "The result shows that CONI covered 5950, 6608, and 6587 more branches than SQLANCER+, RANDOOP, and EVOSUITE respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, C Comparison with Existing Techniques, page 8.",
+              "content_sha256": "sha256:d66504e78e784c48437ad9859148ee8aafe89372f65481688f94690e3cc05c00",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00003.json",
+              "source_type": "paper",
+              "excerpt": "Connector CONI SQLANCER+RANDOOP EVOSUITE MariaDB Connector/J 5 0 0 0 MySQL Connector/J 6 0 0 0 AWS MySQL JDBC 3 0 0 0 PGJDBC 2 0 0 0 PGJDBC NG 5 0 0 0 T otal 21 0 0 0 Impro vement-21↑ 21↑ 21↑ In summary, CONI is unique in its ability to find bugs in database connectors, and compared to other techniques, it can cover more branches within the database connectors, which adequately answers RQ2.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, C Comparison with Existing Techniques, page 9.",
+              "content_sha256": "sha256:6cc66acdc98685ea34081444ee410999842c575adcf2e516a84c5eec5a5964f0",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/7307f131c3d332683815de0634412c48e6a01a7a",
+        "source_type": "paper",
+        "content_sha256": "sha256:6e31d282810ed8dbcc87905495834b8f719ae6dea570733ac260889947593cef"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3709674",
+      "title": "Constant Optimization Driven Database System Testing",
+      "authors": [
+        "Chi Zhang",
+        "Manuel Rigger"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Manag. Data",
+      "doi": "10.1145/3709674",
+      "arxiv_id": "2501.11252",
+      "s2_paper_id": "b989649fc9ac1a8e39768ba4df3dd2e7220ba249",
+      "url": "https://doi.org/10.1145/3709674",
+      "open_access_pdf": "https://arxiv.org/pdf/2501.11252",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3709674",
+              "source_type": "paper_citation_context",
+              "excerpt": "Pivoted Query Synthesis (PQS) [32] generates queries that are guaranteed to retrieve a selected row, based on a naive implementation of operators and functions to be tested, Transformed Query Synthesis (TQS) [37] generates queries by decomposing a table into multiple sub-tables, to derive a test case and ground truth for queries that join these tables.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:81071df4fbaddf84d0cabde29944c11ba4ce54cefd5ad39174afd5c7466b8a36",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3709674",
+              "source_type": "paper_citation_context",
+              "excerpt": "Pivoted Query Synthesis (PQS) [32] generates queries that are guaranteed to retrieve a selected row, based on a naive implementation of operators and functions to be tested, Transformed Query Synthesis (TQS) [37] generates queries by decomposing a table into multiple sub-tables, to derive a test…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8ce3d351bd6434987a4a85b9064e4a474f5bb0a0e76d6a0da83663b156748acc",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3709674",
+              "source_type": "paper_citation_context",
+              "excerpt": "Ternary Logic Partitioning (TLP) [31] decomposes a query into three partitioning queries, each of which retrieves rows based on the predicates p , NOT p , and IS NULL , respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9500eeb0a6580b65cac9fe864936507718a3f4bec0eadc3cd0ab23ef141a419b",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3709674",
+              "source_type": "paper_citation_context",
+              "excerpt": "While we use an existing random generation approach implemented in SQLancer [30, 32], subqueries require additional attention, as they were not supported by existing approaches.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9506b73582e18286deebcb2a88505b755b5ee8ab80d4b7e8f7d7ed289703116d",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3709674",
+              "source_type": "paper_citation_context",
+              "excerpt": "The state-of-the-art approaches are NoREC [30], TLP [31], PQS [32], DQE [35], and TQS [37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f05ec80d78a22b32441106745f93ec57dce7e117ae36083ff45c84d41663ba75",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3709674",
+              "source_type": "paper_citation_context",
+              "excerpt": "Griffin [13] proposes a grammar-free mutation approach for testing DBMSs, using a metadata graph to ensure semantic correctness.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:cfc588ef8d0a39090ef64adf8d07abc355546cfdf477054af71d105f6120358d",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "deterministic",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3709674",
+              "source_type": "paper_citation_context",
+              "excerpt": "The state-of-the-art approaches are NoREC [30], TLP [31], PQS [32], DQE [35], and TQS [37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Non-optimizing Reference Engine Construction (NoREC) as state of the art.",
+              "content_sha256": "sha256:f05ec80d78a22b32441106745f93ec57dce7e117ae36083ff45c84d41663ba75",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/b989649fc9ac1a8e39768ba4df3dd2e7220ba249",
+        "source_type": "paper",
+        "content_sha256": "sha256:69124b1bb521dfaf1b3ce346fc428557c1bf3ccf9ea437523830dc09e9a1097e"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Logic bugs are bugs that can cause database management systems (DBMSs) to silently produce incorrect results for given queries. Such bugs are severe, because they can easily be overlooked by both developers and users, and can cause applications that rely on the DBMSs to malfunction. In this work, we propose Constant-Optimization-Driven Database Testing (CODDTest) as a novel approach for detecting logic bugs in DBMSs. This method draws inspiration from two well-known optimizations in compilers: constant folding and constant propagation. Our key insight is that for a certain database state and query containing a predicate, we can apply constant folding on the predicate by replacing an expression in the predicate with a constant, anticipating that the results of this predicate remain unchanged; any discrepancy indicates a bug in the DBMS. We evaluated CODDTest on five mature and extensively-tested DBMSs--SQLite, MySQL, CockroachDB, DuckDB, and TiDB--and found 45 unique, previously unknown bugs in them. Out of these, 24 are unique logic bugs. Our manual analysis of the state-of-the-art approaches indicates that 11 logic bugs are detectable only by CODDTest. We believe that CODDTest is easy to implement, and can be widely adopted in practice."
+    },
+    {
+      "id": "paper:doi:10.1007/978-981-95-4721-0_7",
+      "title": "CypherFuzzer: A Tool for Testing Access Control in Graph Databases",
+      "authors": [
+        "Philipp Reisinger",
+        "Daniel Hofer",
+        "Bahara Muradi",
+        "Josef Küng"
+      ],
+      "year": 2025,
+      "venue": "Communications in computer and information science",
+      "doi": "10.1007/978-981-95-4721-0_7",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.1007/978-981-95-4721-0_7",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/W7106652132",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://openalex.org/W7106652132",
+        "source_type": "paper",
+        "content_sha256": "sha256:cea5b33d15cf1b9f20f6e25fe643e16cf1606365d8baa71710ad9cf24194cbc0"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3728965",
+      "title": "DepState: Detecting Synchronization Failure Bugs in Distributed Database Management Systems",
+      "authors": [
+        "Cundi Fang",
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Jingzhou Fu",
+        "Zhouyang Jia",
+        "Chun Huang",
+        "Yu Jiang",
+        "Shanshan Li"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Softw. Eng.",
+      "doi": "10.1145/3728965",
+      "arxiv_id": null,
+      "s2_paper_id": "810786dd896fe2eef0a129db6e3b115c95375130",
+      "url": "https://doi.org/10.1145/3728965",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3728965",
+      "abstract": "Distributed Database Management Systems (DDBMSs) are crucial for managing large-scale distributed data. Unlike single-node databases, they are deployed across clusters, distributing data among multiple nodes. The synchronization process in DDBMSs maintains data consistency against data and cluster updates. Due to its complexity, synchronization bugs are inevitable and may cause data inconsistencies, transaction errors, or cluster crashes, severely compromising the availability and reliability of a DDBMS. However, there has been relatively little focus on testing the DDBMS synchronization process. In this paper, we propose DepState, a framework to detect synchronization failure bugs. DepState enhances synchronization testing by simulating the complexities of data sharding and dynamic cluster conditions. It establishes dependencies between tables across nodes and systematically introduces controlled variations in cluster states. We utilize DepState on four DDBMSs: MySQL NDB Cluster, MySQL InnoDB Cluster, MariaDB Galera Cluster, and TiDB Cluster, discovering 25 new bugs, with 13 confirmed. We compare DepState against state-of-the-art tools. DepState finds 14 more synchronization failure bugs and covers 6.13%-66.51%, 5.82%-57.28%, 14.12%-83.30%, 36.81%-83.88%, and 43.24%-54.28% more lines in synchronization-related functions than Jepsen, Mallory, SQLsmith, SQLancer, and Mozi in 24 hours, respectively.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/810786dd896fe2eef0a129db6e3b115c95375130",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728965.json",
+              "source_type": "paper",
+              "excerpt": "To evaluate the effectiveness of DepState in testing DDBMSs, we compare it with five state-of-the-art tools: Jepsen [ 17],Mallory [26], SQLsmith [ 2], SQLancer [ 33], andMozi [23].",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, 5.1 Evaluation Setup, page 14.",
+              "content_sha256": "sha256:a4dc652962864637a42db2c6c4492e71eb2b37689f3dcd2b08e176b0725fe80d",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728965.json",
+              "source_type": "paper",
+              "excerpt": "Number of Unique Bugs Detected by Jepsen, Mallory, SQLsmith, SQLancer, Mozi, and DepState DDBMS Jepsen Mallory SQLsmith SQLancer Mozi DepState MySQL NDB Cluster 0 1 0 1 7 7 MySQL InnoDB Cluster 1 0 0 1 6 2 MariaDB Galera Cluster 1 1 0 1 0 2 TiDB Cluster 0 0 0 0 0 3 Total (Sync Failure Bugs) 2 (0) 2 (0) 0 (0) 3 (0) 13 (0) 14 (14) 24 hours, recording the number of detected bugs and covered synchronization-related functions as metrics.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, 5.3 Comparison with Other Techniques, page 17.",
+              "content_sha256": "sha256:042c91c327a25cb9965b5aab59ffc65b304717ac5ddc5c632cc706425eefdb3e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728965.json",
+              "source_type": "paper",
+              "excerpt": "In comparison, Jepsen, Mallory, SQLancer, and Mozi reported 2, 2, 3, and 13 bugs, respectively, while SQLsmith found no bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, 5.3 Comparison with Other Techniques, page 17.",
+              "content_sha256": "sha256:d023f0e3c0a27c448cdcb6393c47d04f9e2db6f413e6399063e378d5e8f7c70c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728965.json",
+              "source_type": "paper",
+              "excerpt": "To evaluate the effectiveness of DepState in testing DDBMSs, we compare it with five state-of-the-art tools: Jepsen [ 17],Mallory [26], SQLsmith [ 2], SQLancer [ 33], andMozi [23].",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, 5.1 Evaluation Setup, page 14.",
+              "content_sha256": "sha256:a4dc652962864637a42db2c6c4492e71eb2b37689f3dcd2b08e176b0725fe80d",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728965.json",
+              "source_type": "paper",
+              "excerpt": "SQLsmith, SQLancer, and Mozi are three state-of-the-art DBMS testing techniques, which have detected hundreds of bugs in practice.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 5.1 Evaluation Setup, page 14.",
+              "content_sha256": "sha256:418ab049c2b0253f335b18f66df27c9a332d556a23826ab5e50e5a06a0662b75",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/810786dd896fe2eef0a129db6e3b115c95375130",
+        "source_type": "paper",
+        "content_sha256": "sha256:7ae7611188e09d0c85b08c75f17616b134cefc0a2fc583fc10dc0e88e5c69c80"
+      }
+    },
+    {
+      "id": "paper:doi:10.1007/s10664-025-10662-w",
+      "title": "Detecting data manipulation errors in android applications using scene-guided exploration",
+      "authors": [
+        "Shuqi Liu",
+        "Yu Zhou",
+        "Wenhua Yang",
+        "Taolue Chen",
+        "Harald C. Gall"
+      ],
+      "year": 2025,
+      "venue": "Empirical Software Engineering",
+      "doi": "10.1007/s10664-025-10662-w",
+      "arxiv_id": null,
+      "s2_paper_id": "fcc98415c9f0ae0b33b88b84d20d09969c757cca",
+      "url": "https://doi.org/10.1007/s10664-025-10662-w",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1007/s10664-025-10662-w",
+              "source_type": "paper_citation_context",
+              "excerpt": "Existing studies (Rigger and Su 2020a,b have detected CRUD errors in database management systems, but these approaches are not specifically tailored for Android apps.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e4e04254ad64c3d93e44e6063b081b4d39bca7b602d60cd2e0f4a2ad64974ee8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/fcc98415c9f0ae0b33b88b84d20d09969c757cca",
+        "source_type": "paper",
+        "content_sha256": "sha256:fdf40ca3021bc4f4e7052ad3e85e067d0d0da784951c5833f317d33def096567"
+      }
+    },
+    {
+      "id": "paper:doi:10.1016/j.cose.2025.104564",
+      "title": "Detecting DBMS bugs with context-sensitive instantiation and multi-plan execution",
+      "authors": [
+        "Jiaqi Li",
+        "Ke Wang",
+        "Yaoguang Chen",
+        "Yajin Zhou",
+        "Lei Wu",
+        "Jiashui Wang"
+      ],
+      "year": 2025,
+      "venue": "Computers & Security",
+      "doi": "10.1016/j.cose.2025.104564",
+      "arxiv_id": "2312.04941v1",
+      "s2_paper_id": "941500a2f13307898e0f0152e4bde173d51dc7bb",
+      "url": "https://doi.org/10.1016/j.cose.2025.104564",
+      "open_access_pdf": null,
+      "abstract": "DBMS bugs can cause serious consequences, posing severe security and privacy concerns. This paper works towards the detection of memory bugs and logic bugs in DBMSs, and aims to solve the two innate challenges, including how to generate semantically correct SQL queries in a test case, and how to propose effective oracles to capture logic bugs. To this end, our system proposes two key techniques. The first key technique is called context-sensitive instantiation, which considers all static semantic requirements (including but not limited to the identifier type used by existing systems) to generate semantically valid SQL queries. The second key technique is called multi-plan execution, which can effectively capture logic bugs. Given a test case, multi-plan execution makes the DBMS execute all query plans instead of the default optimal one, and compares the results. A logic bug is detected if a difference is found among the execution results of the executed query plans. We have implemented a prototype system called Kangaroo and applied it to three widely used and well-tested DBMSs, including SQLite, PostgreSQL, and MySQL. Our system successfully detected 50 new bugs. The comparison between our system with the state-of-the-art systems shows that our system outperforms them in terms of the number of generated semantically valid SQL queries, the explored code paths during testing, and the detected bugs.",
+      "cites_seed_techniques": [
+        "dqp",
+        "norec",
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/941500a2f13307898e0f0152e4bde173d51dc7bb",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Differential Query Plans (DQP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2312.04941",
+              "source_type": "paper_citation_context",
+              "excerpt": "The generation-based approach [7, 14–17, 27, 37, 38] is effective in generating syntax-correct test cases since it typically follows a grammar model that describes the format of the input.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:99232fc1f6283fec657b13585530e693cdbca8c2e7cb9c20cbe1c8dd047897cc",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2312.04941",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, NoREC [15] requires that the effective SQL query in a test case has WHERE clauses, thus it can only detect logic bugs due to the optimization of WHERE clauses.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0d1f592bb5fe413325d664544a03e95b35860239ca2b94d7a5b3a2a2003a89c3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2312.04941",
+              "source_type": "paper_citation_context",
+              "excerpt": "Although Rigger et al. proposed three oracles for DBMS logic bug detection, including PQS [14], NoREC [15], and TLP [16], all of them put limitations on the SQL queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4114c97c6bcf1093e789d829238a93a9d4387dc262324a9f1a5c58486bd34338",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2312.04941",
+              "source_type": "paper_citation_context",
+              "excerpt": "The rule-based ones [14–17 , 24 The mutation-based method [6, 13, 25] generates new test cases by mutating seed queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e7a54d66177932eaaaa028ca6e785debbff677b7bc518bdc6d69b41087896920",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2312.04941",
+              "source_type": "paper_citation_context",
+              "excerpt": "To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7f72bab01869466cdb8a528c218e23f65324d6eaccbab7a8cf9526631c24e1de",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2312.04941",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [15] generates equivalent queries by shifting the conditions in the WHERE clause to the SELECT expression.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8d677408576f02460c3699ebabbfcb43c56e5241c9a17caa63653cb4c6dc5d0b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1016_j_cose_2025_104564.json",
+              "source_type": "paper",
+              "excerpt": "We also compared Kangaroo with leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, 1 Introduction, page 2.",
+              "content_sha256": "sha256:c162529b8838db258cd639ab5d370961b2c483b626d2e83c4713a2b602fd3eab",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1016_j_cose_2025_104564.json",
+              "source_type": "paper",
+              "excerpt": "Specifically, after conducting a 24-hour test on the three DBMSs, Kangaroo successfully detected 17 bugs, while SQLancer, Squirrel, and SQLRight only identified 1, 3, and 6 bugs, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 1 Introduction, page 2.",
+              "content_sha256": "sha256:d601d4b3772aba39ab2e5dc41805a27c783ae7f920d45eb09e058d603ad132d2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1016_j_cose_2025_104564.json",
+              "source_type": "paper",
+              "excerpt": "We use NoREC for SQLancer and SQLRight.",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, 5.2 Generating Valid Queries, page 11.",
+              "content_sha256": "sha256:e2b54c2ef1f48fa56fc2060f5c3e7b05ac06c104fa00e3441f0f877136420981",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1016_j_cose_2025_104564.json",
+              "source_type": "paper",
+              "excerpt": "Since SQLancer does not implement NoREC for MySQL, we use TLP instead.",
+              "excerpt_is_verbatim": true,
+              "note": "M15 in the paper's extracted text, 5.2 Generating Valid Queries, page 11.",
+              "content_sha256": "sha256:3f1a49f9c64d6a3b8c94cde4b5b01e281d7c3a6ec39214819f10a854fd2164e4",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1016_j_cose_2025_104564.json",
+              "source_type": "paper",
+              "excerpt": "Why Existing Works Cannot Detect the Bug NoREC is one of the most effective DBMS logic bug detection oracles which requires the SELECT statements to satisfy some predefined rules, e.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, 3.2 Multi-Plan Execution, page 5.",
+              "content_sha256": "sha256:4d780e0f3241c47848d66042946b54a49390d67742e1cd799bac8bfe86394cbd",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1016_j_cose_2025_104564.json",
+              "source_type": "paper",
+              "excerpt": "Comparisons with Existing Tools We compare Kangaroo with three state-of-the-art and open source systems: Squirrel, SQLancer, and SQLRight.",
+              "excerpt_is_verbatim": true,
+              "note": "M16 in the paper's extracted text, 5.3 Comparisons with Existing Tools, page 11.",
+              "content_sha256": "sha256:81d414a5ac1860ec6eefc4a46882140a40b2c7b32f6562bb4385502a52c25f34",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/941500a2f13307898e0f0152e4bde173d51dc7bb",
+        "source_type": "paper",
+        "content_sha256": "sha256:989680125b533df9c92bd49d5bdcdeca6c2e0b1d3f4ba29f600bac63c0e1a89b"
+      },
+      "also_indexed_as": [
+        "paper:arxiv:2312.04941"
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3728953",
+      "title": "Detecting Isolation Anomalies in Relational DBMSs",
+      "authors": [
+        "Rui Yang",
+        "Ziyu Cui",
+        "Wensheng Dou",
+        "Yu Gao",
+        "Jiansen Song",
+        "Xudong Xie",
+        "Jun Wei"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Softw. Eng.",
+      "doi": "10.1145/3728953",
+      "arxiv_id": null,
+      "s2_paper_id": "189d9b84961fe904388013b4a8f7448854a847c6",
+      "url": "https://doi.org/10.1145/3728953",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3728953",
+      "abstract": "Relational Database Management Systems (DBMSs) utilize transactions to ensure data consistency and integrity, while providing multiple isolation levels to strike a balance between consistency and performance. However, isolation anomalies in relational DBMSs can undermine their claimed isolation levels, and lead to severe consequences, e.g., incorrect query results and database states. Existing isolation checkers can only work on simple key-υalue-like data models and the associated read (key) and write (key, υalue) operations. Therefore, they cannot be directly applied to relational DBMSs that support relational data models and complex SQL operations. In this paper, we propose a novel black-box Isolation checker for Relational DBMSs, IsoRel, which can support relational data models and complex SQL operations. To infer dependencies among transactions in relational DBMSs, we first design an isolation-agnostic SQL statement instrumentation approach to record the data rows accessed by each SQL statement by utilizing two auxiliary columns in each database table. We then utilize the recorded data rows of each SQL statement to construct a transaction dependency graph for relational transactions, and identify isolation anomalies based on anomaly patterns. We evaluate IsoRel on five widely-used relational DBMSs, i.e., MySQL, PostgreSQL, MariaDB, CockroachDB, and TiDB, and all their supported isolation levels. Our evaluation reveals a total of 48 unique isolation anomalies that violate the isolation levels defined by Adya.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3728953",
+              "source_type": "paper_citation_context",
+              "excerpt": "Rigger et al. propose SQLancer [12] and several approaches, e.g., PQS [53], TLP [52] and NoREC [51], to detect logic bugs in SELECT statements.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e64dad105b7409214834002c43d379a114425a0437e5f210f75e22f251d45f74",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/189d9b84961fe904388013b4a8f7448854a847c6",
+        "source_type": "paper",
+        "content_sha256": "sha256:a871776a1a1a46e2bacb626b4902669e12f28dbf9834ea1b2fa307177161098f"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3769779",
+      "title": "Detecting Logic Bugs in DBMSs via Equivalent Data Construction",
+      "authors": [
+        "Wenqian Deng",
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Jingzhou Fu",
+        "Yu Jiang"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Manag. Data",
+      "doi": "10.1145/3769779",
+      "arxiv_id": null,
+      "s2_paper_id": "7762da5097d2f7d3482670909ad1d61c216119d6",
+      "url": "https://doi.org/10.1145/3769779",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3769779",
+      "abstract": "Database Management Systems (DBMS) perform various data operations such as arithmetic calculations and string manipulations when executing SQL queries. These operations are complex due to the wide range of data types and the intricate interactions between different data. Consequently, errors in implementing these data operations can lead to logic bugs, potentially causing issues such as implicit type coercion, overflow, and precision loss. Existing logic bug detection methods primarily focus on issues introduced during query optimization by adapting query-level strategies. However, these methods have limitations when it comes to detecting logic bugs caused by implementation errors in data types and operations. To address this, we propose equivalent data construction (EDC), a novel approach to detect logic bugs in data operation implementations within DBMSs. The core insight is that for data operation expressions in SQL queries, substituting them with precomputed result values should yield identical query outcomes. EDC mainly involves the following steps: first, construct equivalent data for an operation by calculating and storing the results in a derived equivalent table; then, transform the query by replacing the operation expressions with the precomputed results from the equivalent table. Any inconsistencies between the results of the base and transformed queries indicate potential logic bugs. We implemented EDC and evaluated it on six well-tested and widely-used DBMSs(e.g.,MySQL, MariaDB). Our evaluation revealed 52 previously unknown bugs, of which 38 have been confirmed by developers. Developers took these findings seriously. For example, MariaDB developers described our findings as counterintuitive, helping them uncover more issues related to data operations.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3769779",
+              "source_type": "paper_citation_context",
+              "excerpt": "For instance, EET [27] targets expression-level transformations, TLP [37] partitions queries based on predicate logic, and NoREC [36] rewrites queries to disable DBMS optimizations.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:766926e2c0fe0fe2cd9c557f3ef79af55616446a11a76e93dace70ac1622f5f5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3769779",
+              "source_type": "paper_citation_context",
+              "excerpt": "Specifically, some approaches split or transform queries to expose discrepancies in the results (e.g., TLP [37], NoREC [36]), while others construct logically crafted SQL queries designed to return specific results in a controlled manner (e.g., PQS [38], DQE [44]).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e7ed00015bbff923975bfee1a86822f861e2d70a4490c082cf3980c02c217a41",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3769779",
+              "source_type": "paper_citation_context",
+              "excerpt": "Several approaches have been proposed to detect logic bugs in DBMSs [27, 36–38, 44, 45], many of which operate at the query level and rely on various strategies to identify inconsistencies in query results.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b2f589a1c8c9a5aa7cd99ef7a80eb6599037d08c51081029c35bc9e7bb9c7dfa",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3769779",
+              "source_type": "paper_citation_context",
+              "excerpt": "We also evaluate EDC against 3 state-of-art testing techniques, i.e., TLP [37], Radar [45], and EET [27].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0c431f8436ce9c18aef459af48dc65c7044d5eacc23768d0f610c42297dc08a2",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769779.json",
+              "source_type": "paper",
+              "excerpt": "We compare EDCagainst 3 state-of-the-art open-source tools for finding logic bugs in DBMSs: TLP [ 37], EET [ 27], and Radar [ 45].",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, 5.5 Comparison with Existing Techniques, page 19.",
+              "content_sha256": "sha256:ce99cb7f3fba285f6d2761b59b823afc811048216c0ae57cfb96bc9b6d1fdfb8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769779.json",
+              "source_type": "paper",
+              "excerpt": "During the evaluation,EDCfound 8, 6, 6, 2, 5, 6, and 7 bugs in MySQL, MariaDB, Percona, PostgreSQL, TiDB, OceanBase, and ClickHouse, respectively, while TLP detected 3 bugs in MySQL, 2 in Percona, 4 in TiDB, and none in others.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, 5.5 Comparison with Existing Techniques, page 19.",
+              "content_sha256": "sha256:e1b0b57a72b9dc6329daaf55c0db6cabf1d7beb4d957a5119dd75929439f239c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769779.json",
+              "source_type": "paper",
+              "excerpt": "Across the seven DBMSs,EDCdetected 38 unique bugs, while TLP, Radar, and EET identified 9, 7, and 5 bugs respectively, highlighting the complementary nature of their testing strategies.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, 5.5 Comparison with Existing Techniques, page 19.",
+              "content_sha256": "sha256:8a77b05587dce534bbafca4da08efe40c84c0db8e8009b6080127bedc6cd407b",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769779.json",
+              "source_type": "paper",
+              "excerpt": "We compare EDCagainst 3 state-of-the-art open-source tools for finding logic bugs in DBMSs: TLP [ 37], EET [ 27], and Radar [ 45].",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, 5.5 Comparison with Existing Techniques, page 19.",
+              "content_sha256": "sha256:ce99cb7f3fba285f6d2761b59b823afc811048216c0ae57cfb96bc9b6d1fdfb8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/7762da5097d2f7d3482670909ad1d61c216119d6",
+        "source_type": "paper",
+        "content_sha256": "sha256:f9227d9aeca357bdb050f86500e84a1393c6d6326c4f2f27dabcf110c7d0988d"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3734839.3734861",
+      "title": "Detecting Schema-Related Logic Bugs in Relational DBMSs via Equivalent Database Construction",
+      "authors": [
+        "Jiansen Song",
+        "Wensheng Dou",
+        "Yingying Zheng",
+        "Yu Gao",
+        "Ziyu Cui",
+        "Wei Wang",
+        "Jun Wei"
+      ],
+      "year": 2025,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3734839.3734861",
+      "arxiv_id": null,
+      "s2_paper_id": "b38453bfb00ddaa816b69edcc232a1b39b0a31f9",
+      "url": "https://doi.org/10.14778/3734839.3734861",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "cert",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3734839.3734861",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, differential testing fails to test SQL features specific to individual DBMSs. Metamorphic testing [15, 31, 43, 47, 48] detects logic bugs in individual DBMSs by constructing equivalent SELECT statements and observing differences among their outputs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c6b25835fd2ed3e165b9316fcb1ca6d9f4c4abc258279725b8559a03f859cf35",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3734839.3734861",
+              "source_type": "paper_citation_context",
+              "excerpt": "Existing testing approaches for relational DBMSs mainly focus on detecting logic bugs in the SELECT statements [15, 24, 31, 42, 43, 47, 48, 50, 51, 54].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:61dd5cb083ae893ecdbbcf7c056449c7998d68d4c9a82f4a2a249fb49769a850",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3734839.3734861",
+              "source_type": "paper_citation_context",
+              "excerpt": "We select MariaDB, CockroachDB, and TiDB, because they have been thoroughly tested by existing approaches [31, 38, 43, 44, 47, 48, 51, 52, 58, 63].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:37c3e40ad1c2abd726fc96a1cc7fcb484f631b6ef585b26753c58b120576bfec",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3734839.3734861",
+              "source_type": "paper_citation_context",
+              "excerpt": "Other approaches construct equivalent SELECT statements on the same database [14, 15, 27, 31, 41, 46– 48, 56, 61].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:29643ac73ba58bd94aa45ef57a37bce9af0cbbcae997ba81d0f8b982fc3accb0",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3734839.3734861",
+              "source_type": "paper_citation_context",
+              "excerpt": "Randomly generating statements is a commonly-used and effective approach in DBMS testing [31, 47–49, 51, 52].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8ee5ef10bd26c7c8f310825158a5565a975a3bd599da5be3f20b94e8b490dec1",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3734839.3734861",
+              "source_type": "paper_citation_context",
+              "excerpt": "Second, the DBMS should have been thoroughly tested by existing approaches [15, 47, 48, 51, 52].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8924f94693a521311079d43c87bbaae23e4242910a2b81461a2cea62ec090d99",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/b38453bfb00ddaa816b69edcc232a1b39b0a31f9",
+        "source_type": "paper",
+        "content_sha256": "sha256:42d00904883cdc424d2f0a92d3bb702c29659dcc88360b00a3fbe1901d498904"
+      },
+      "abstract": "Relational Database Management Systems (DBMSs) provide flexible DDL (Data Definition Language) statements that enable the creation, modification, and deletion of database schemas. In addition to database schemas, relational DBMSs typically manage various schema-related information internally, e.g., schema changes, tablespace allocation, and block-level data layout. However, incorrect implementations related to schema-related information maintenance and utilization can introduce schema-related logic bugs. These bugs can cause DQL (Data Query Language) statements to return incorrect query results and DML (Data Manipulation Language) statements to create incorrect database states. Existing approaches mainly focus on detecting logic bugs in DQL statements, but are ineffective in detecting schema-related logic bugs.\n \n In this paper, we propose a novel and general testing approach, DDLCheck, to effectively detect schema-related logic bugs in relational DBMSs. We first generate a complex DDL sequence\n \n seq\n gen\n \n that consists of various types of DDL statements, and then synthesize a rather simple DDL sequence\n \n seq\n syn\n \n , which utilizes CREATE statements to create the same database schema as\n \n seq\n gen\n .\n \n Executing the same SQL statements on the two databases created by\n \n seq\n gen\n \n and\n \n Seq\n syn\n \n should yield the same execution results. Any discrepancy between their execution results indicates a schema-related logic bug. To improve the testing efficiency of DDLCheck, we further design a DDL-sequence-oriented testing optimization strategy, which can help DDLCheck explore diverse schema-related information and detect schema-related logic bugs quickly. We implement and evaluate DDLCheck on six widely-used relational DBMSs. We have detected 34 bugs in these DBMSs, of which 29 bugs have been confirmed as previously unknown bugs and 9 bugs have been fixed."
+    },
+    {
+      "id": "paper:doi:10.1109/srds69199.2025.00038",
+      "title": "Diverse Database Replication Based on Snapshot Isolation – Performance Implications of Improved Dependability",
+      "authors": [
+        "Peter Popov",
+        "Vladimir Stankovic"
+      ],
+      "year": 2025,
+      "venue": "IEEE International Symposium on Reliable Distributed Systems",
+      "doi": "10.1109/srds69199.2025.00038",
+      "arxiv_id": null,
+      "s2_paper_id": "9c1ceed81b5e20d20819cb033534a5480040eb02",
+      "url": "https://doi.org/10.1109/srds69199.2025.00038",
+      "open_access_pdf": "https://openaccess.city.ac.uk/id/eprint/35486/1/DivSQL_v4.8.pdf",
+      "abstract": "Numerous database replication schemes are built on the crash failure assumption where majority of failures are self-evident as defined in [1]. The study in [1] convincingly refuted this common assumption showing that many of the faults in relational Database Management Systems (DBMSs) cause systematic non-crash failures. Similar results were obtained in the subsequent study [2]. Consequently, the existing database replication solutions, which typically use the same DBMS, are ineffective fault-tolerant mechanisms. Conversely, using diverse DBMSs is a suitable way of protecting against non-crash failures. We have built a middleware-based database replication protocol, DivRep, and deployed it with diverse database servers (DivSQL), for improved fault tolerance. DivSQL provides strict Snapshot Isolation (SI) guarantees, and assumes “incorrect results” failure model (IRFM) – the most realistic one based on the extensive experimental analyses of DBMS faults ([1], [2]). The dependability gain comes with the inherent performance overhead. We provide a comprehensive performance evaluation of DivSQL using 3 diverse DBMSs (two are leaders in the field).",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/srds69199.2025.00038",
+              "source_type": "paper_citation_context",
+              "excerpt": "In addition to crash failures, it guards against failures of SELECTs (e.g., erroneous omission of a row) – referred to by some as “logic bugs” [19], and DELETEs, INSERTs and UPDATEs wrongly changing the database state.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ca3575391eaef11c7c3444dc76acb7299a930295118c648195809287c51bd5b1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/9c1ceed81b5e20d20819cb033534a5480040eb02",
+        "source_type": "paper",
+        "content_sha256": "sha256:e9b6663cfd6f1c6373412c44a082ca8d4dea6a73c3b3e1b027703f2a56f89450"
+      }
+    },
+    {
+      "id": "paper:doi:10.35970/jinita.v7i1.2707",
+      "title": "Evaluating ERD Models and RAID-Based Storage for Query Performance Optimization in Relational Databases",
+      "authors": [
+        "Juanda Hakim Lubis",
+        "Sri Handayani",
+        "Herman Mawengkang",
+        "Yuliska"
+      ],
+      "year": 2025,
+      "venue": "Journal of Innovation Information Technology and Application (JINITA)",
+      "doi": "10.35970/jinita.v7i1.2707",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.35970/jinita.v7i1.2707",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/None",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Cardinality Estimation Restriction Testing (CERT).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://openalex.org/None",
+        "source_type": "paper",
+        "content_sha256": "sha256:b8a6b2d8b9f183cd18985471248988ff8c7fe9f9ccbbd5aec94159eed65bf9c4"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3731569.3764841",
+      "title": "Fawkes: Finding Data Durability Bugs in DBMSs via Recovered Data State Verification",
+      "authors": [
+        "Zhiyong Wu",
+        "Jie Liang",
+        "Jingzhou Fu",
+        "Wenqian Deng",
+        "Yu Jiang"
+      ],
+      "year": 2025,
+      "venue": "Symposium on Operating Systems Principles",
+      "doi": "10.1145/3731569.3764841",
+      "arxiv_id": null,
+      "s2_paper_id": "e8506c9f0e513b48630e0d8329746e4fc73a99af",
+      "url": "https://doi.org/10.1145/3731569.3764841",
+      "open_access_pdf": "https://doi.org/10.1145/3731569.3764841",
+      "abstract": "Data durability is a fundamental requirement in DBMSs, ensuring that committed data remains intact despite unexpected faults such as power failures. Despite its critical importance, implementations of durability and recovery mechanisms continue to exhibit flaws, leading to severe issues(e.g., data loss, data inconsistency), which we refer to as Data Durability Bugs (DDBs). However, there is a limited understanding of the characteristics and root causes of DDBs. Furthermore, existing testing methods(e.g., Mallory) are often inadequate for detecting DDBs, particularly those that cause data loss or data inconsistency following DBMS failures. This paper presents a comprehensive study of 43 DDBs across four widely used DBMSs. It reveals that DDBs primarily manifest as data loss, data inconsistency, log corruption, and system unavailability, often stem from flawed durability and recovery mechanisms, and are typically triggered when faults occur during filesystem or kernel-level calls. Based on these findings, we developed Fawkes, a testing framework to detect DDBs with recovered data state verification. It employs context-aware fault injection to target critical filesystem and kernel-level regions, functionality-guided fault triggering to explore untested paths, and checkpoint-based data graph verification to detect post-crash inconsistencies. We applied Fawkes to eight popular DBMSs and discovered 48 previously unknown DDBs, of which 16 have been fixed and 8 have been assigned CVE identifiers due to the severity.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp",
+        "coddtest"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "dqp",
+            "coddtest"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/e8506c9f0e513b48630e0d8329746e4fc73a99af",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Differential Query Plans (DQP), Constant-Optimization-Driven Testing (CODDTest).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/e8506c9f0e513b48630e0d8329746e4fc73a99af",
+        "source_type": "paper",
+        "content_sha256": "sha256:67f05b6e13906d5c0f17b74b8dcb6f258c73ffc0bd0bbb6ee08013a3eef5c09e"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3725300",
+      "title": "Finding Logic Bugs in Graph-processing Systems via Graph-cutting",
+      "authors": [
+        "Qiuyang Mang",
+        "Jinsheng Ba",
+        "Pinjia He",
+        "Manuel Rigger"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Manag. Data",
+      "doi": "10.1145/3725300",
+      "arxiv_id": null,
+      "s2_paper_id": "3713ce1ea618a9a284cbed43589c342cfa2abf09",
+      "url": "https://doi.org/10.1145/3725300",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3725300",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3725300",
+              "source_type": "paper_citation_context",
+              "excerpt": "While previous works have proposed several test oracles for testing relational DBMSes [2, 13, 27, 29, 45, 46, 52] and GDBMSes [5, 25, 33, 36, 62, 66], to the best of our knowledge, it is still unclear how to adapt them to graph-processing systems. original graph.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:626704bbf6739a1e54e46e8a07e51bbd2c80480a976e171662f9e3aaf784aff5",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/3713ce1ea618a9a284cbed43589c342cfa2abf09",
+        "source_type": "paper",
+        "content_sha256": "sha256:31ffa8bf5d45bfe3aa17a09f3a7820de2636d58dcd1c72dd9a1cf932107f1107"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Graph-processing systems, including Graph Database Management Systems (GDBMSes) and graph libraries, are designed to analyze and manage graph data efficiently. They are widely used in applications such as social networks, recommendation systems, and fraud detection. However, logic bugs in these systems can lead to incorrect results, compromising the reliability of applications. While recent research has explored testing techniques specialized for GDBMSes, it is unclear how to adapt them to graph-processing systems in general. This paper proposes Graph-cutting, a universal approach for detecting logic bugs in both GDBMSes and various algorithms in graph libraries. Our key idea is inspired by the observation that certain graph patterns are critical for various graph-processing tasks. Dividing graph data into subgraphs that preserve those patterns establishes a natural relationship between query results on the original graph and its subgraphs, allowing for the detection of logic bugs when this relationship is violated. We implemented Graph-cutting as a tool, GSlicer, and evaluated it on 3 popular graph-processing systems, NetworkX, Neo4j, and Kùzu. GSlicer detected 39 unique and previously unknown bugs, out of which 34 have been fixed and confirmed by developers. At least 8 logic bugs detected by GSlicer cannot be detected by baseline strategies. Additionally, by leveraging just a few concrete relationships, Graph-cutting can cover over 100 APIs in NetworkX. We expect this technique to be widely applicable and that it can be used to improve the quality of graph-processing systems broadly."
+    },
+    {
+      "id": "paper:doi:10.14778/3725688.3725713",
+      "title": "Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving",
+      "authors": [
+        "Xiyue Gao",
+        "Zhuang Liu",
+        "Yiran Shen",
+        "Hui Li",
+        "Yingfan Liu",
+        "H. Xiao",
+        "Yanguo Peng",
+        "Jiangtao Cui"
+      ],
+      "year": 2025,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3725688.3725713",
+      "arxiv_id": null,
+      "s2_paper_id": "3e4bdd50318c8fb8ba3bd6aea1d078fc4e05740f",
+      "url": "https://doi.org/10.14778/3725688.3725713",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3725688.3725713",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [41] ensures predicate evaluations always fall within TRUE , FALSE , or NULL , enabling queries to be decomposed into three partitioned queries, also applicable for detecting transaction isolation bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2f428a56dee997aec0a782c36f7bfb4fa35d33c7e4bb8b45063ce66a6a55d5b4",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3725688.3725713",
+              "source_type": "paper_citation_context",
+              "excerpt": "DT2 and Troc were selected as base-lines because other database fuzzing tools, such as SQLsmith [31], Squirrel [49], and NoREC [41], focus primarily on non-transactional modes.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e488b3b774c5f707185b313bb27cc3004d6286833f3e1a429d27becd84f5f109",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3725688.3725713",
+              "source_type": "paper_citation_context",
+              "excerpt": "Existing oracles can be divided into four types: crash oracles [46, 49], differential oracles [25], metamorphic oracles [43] and constraint-solving oracles [41, 45].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:fbcbdb6bbecd764bde9211ce66df921529af48fffa1ea67bbb4984cf9be0ef6c",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3725688.3725713",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, existing reducers [23, 41] are limited to simplifying single SQL statements, not entire transactions.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4fa17b8bd0db10d186bd8bf3e9bb19df3fb4627152224b5e414c1607c31430b4",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3725688.3725713",
+              "source_type": "paper_citation_context",
+              "excerpt": "Test cases can be created using either generation-based [26, 41, 45] or mutation-based [19, 23, 35] methods.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c533426a7903869b7257e72da839160f97c26043e0e1d6409cb8d934ac05863d",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3725688.3725713",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [40] rewrites statements to bypass database optimization, detecting incorrect optimizations.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:de196b8ead3c630920ae6c5c2fff14ec0931088ebdd27f2668190f4bd04c572a",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "artifact_inspection",
+          "evidence": [
+            {
+              "source_url": "https://github.com/Reverie4u/Fucci",
+              "source_type": "github_repository",
+              "excerpt": "Artifact for the paper \"Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving\".\n# Fucci\r\n\r\nThis is the artifact for the paper \"Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving\".",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/Reverie4u/Fucci/blob/main/src/main/java/fucci/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "src/main/java/fucci/Randomly.java is SQLancer's Randomly.java, with the package renamed to fucci (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:b7e8287729907dc2090cad60304cd4cf295bd21693879a5be3f88c686073c32a",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ],
+          "classifier": {
+            "method": "artifact_inspection",
+            "classifier_version": "artifact-markers-v1",
+            "policy_version": "impact-policy-v1",
+            "taxonomy_version": "taxonomy-v2",
+            "classified_at": "2026-09-06T15:09:10Z",
+            "model": null,
+            "rationale": "Artifact carries SQLancer markers: renamed_sqlancer_package, sqlancer_source_content_match"
+          }
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/3e4bdd50318c8fb8ba3bd6aea1d078fc4e05740f",
+        "source_type": "paper",
+        "content_sha256": "sha256:edecaf352df8df083dbb7667a61ec916a3eef7ce35ddcffcd1296dc33c72fdeb"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/Reverie4u/Fucci",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-06T15:09:10Z",
+          "sqlancer_markers": [
+            "renamed_sqlancer_package",
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/Reverie4u/Fucci",
+              "source_type": "github_repository",
+              "excerpt": "Artifact for the paper \"Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving\".\n# Fucci\r\n\r\nThis is the artifact for the paper \"Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving\".",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/Reverie4u/Fucci/blob/main/src/main/java/fucci/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "src/main/java/fucci/Randomly.java is SQLancer's Randomly.java, with the package renamed to fucci (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:b7e8287729907dc2090cad60304cd4cf295bd21693879a5be3f88c686073c32a",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        }
+      ],
+      "abstract": "Ensuring the ACID properties of transactions is the fundamental functionality of transactional DBMSs. However, through our study on existing solutions on transaction management, we found that transaction implementations in some mainstream databases, such as MySQL, MariaDB and TiDB, may violate what they claim in their documentation, in the form of incorrect database state or query results. Since there is still a lack of efficient and comprehensive testing methods to detect bugs within transaction management implementation for off-the-shelf DBMSs at present, we propose Fucci, a fuzzing framework, to solve the problem. Given a target DBMS, Fucci improves the efficiency of detecting transaction bugs through three key components: Random Conflict Construction (RCC), Multilevel Constraint Solving (MCS), and Experience-driven Automatic Simplification (EAS). RCC addresses the issue of inadequate case validity by ensuring the presence of read-write or write-write conflicts between transactions. MCS enhances the accuracy and efficiency of the transaction oracle by employing an external multi-version control system to solve data visibility. EAS is ultimately adopted to improve the efficiency of simplification and the readability of the identified bug cases. All of the above strategies are tested on commercial databases such as MySQL, MariaDB and TiDB. Accordingly, 6 previously unknown transaction bugs and 14 known duplicate transaction bugs have been newly discovered, most of which have been officially acknowledged."
+    },
+    {
+      "id": "paper:doi:10.1007/978-981-96-6465-8_28",
+      "title": "Fuzz Testing for Database Management System Configuration Errors",
+      "authors": [
+        "Haoran Zhu",
+        "Menglin Li",
+        "Bo Wang",
+        "Tengfei Li",
+        "Jingtian Liu",
+        "Zan Zhou"
+      ],
+      "year": 2025,
+      "venue": "Communications in computer and information science",
+      "doi": "10.1007/978-981-96-6465-8_28",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.1007/978-981-96-6465-8_28",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/W4413060152",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://openalex.org/W4413060152",
+        "source_type": "paper",
+        "content_sha256": "sha256:255e32476678b5ea107ca8a1a2c9f82058bfda367904c1df7394783d8ea4627b"
+      }
+    },
+    {
+      "id": "paper:doi:10.1007/978-3-031-94706-3_7",
+      "title": "Fuzzing Graph Database Applications with Graph Transformations",
+      "authors": [
+        "Stefania Dumbrava",
+        "Melchior W. M. Oudemans",
+        "Burcu Kulahcioglu Ozkan"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Graph Transformation",
+      "doi": "10.1007/978-3-031-94706-3_7",
+      "arxiv_id": null,
+      "s2_paper_id": "1e316a65311c25101c0b04f1be8d90bc33e5b4b9",
+      "url": "https://doi.org/10.1007/978-3-031-94706-3_7",
+      "open_access_pdf": "https://hal.science/hal-05117893",
+      "abstract": ". Graph databases have surged in popularity, and applications increasingly employ them to store and retrieve interconnected data. However, testing graph database-backed applications has distinctive challenges. Due to the sheer dimension of the graph schema state space, testing applications using naive random graph instances is unlikely to cover a large portion of an application program. We present PGFuzz , a graph transformation-based greybox fuzzer for testing graph database-backed applications, that is, to the best of our knowledge, the first fuzzer to specifically target graph database applications. PGFuzz builds on top of state-of-the-art graph generators and utilizes graph transformations guided by code coverage to produce application test inputs. PGFuzz ’s graph transformations are schema-aware and support recently introduced graph schema, key, and cardinality constraints. We evaluate PGFuzz on graph database applications that we curate from open-source repositories and show that PGFuzz substantially improves the test coverage of graph database-backed applications compared to the state-of-the-art.",
+      "cites_seed_techniques": [
+        "qpg",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1007/978-3-031-94706-3_7",
+              "source_type": "paper_citation_context",
+              "excerpt": "Several recent works focus on testing database management systems [9,17,45] and graph databases [47,49,83].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5af15e9027b5d5eef48457e0c007c99e561564e862d2ab85c44be8ba399619f5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/1e316a65311c25101c0b04f1be8d90bc33e5b4b9",
+        "source_type": "paper",
+        "content_sha256": "sha256:47e7973426f79697ce31146c455393937bfc6222a47b0df5080217046cd75f5c"
+      }
+    },
+    {
+      "id": "paper:doi:10.1007/978-981-95-3182-0_3",
+      "title": "Ghosts in DBMS: Revealing the Security Impacts of Silent Fixes",
+      "authors": [
+        "Jialiang Dong",
+        "Zihan Ni",
+        "Willy Susilo",
+        "Siqi Ma"
+      ],
+      "year": 2025,
+      "venue": "Lecture notes in computer science",
+      "doi": "10.1007/978-981-95-3182-0_3",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.1007/978-981-95-3182-0_3",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/W4415219679",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://openalex.org/W4415219679",
+        "source_type": "paper",
+        "content_sha256": "sha256:cd08d781c2f9f5d422a38719a80cf8fbdf98459c408ba506acf4d45e1f3d2535"
+      }
+    },
+    {
+      "id": "paper:arxiv:2502.15160",
+      "title": "GraphFuzz: Automated Testing of Graph Algorithm Implementations with Differential Fuzzing and Lightweight Feedback",
+      "authors": [
+        "Wenqi Yan",
+        "Manuel Rigger",
+        "Anthony Wirth",
+        "Van-Thuan Pham"
+      ],
+      "year": 2025,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2502.15160",
+      "s2_paper_id": "69e1e582a2e909542afca0b4dd4e1f73d8b4b987",
+      "url": "https://arxiv.org/abs/2502.15160",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/69e1e582a2e909542afca0b4dd4e1f73d8b4b987",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/69e1e582a2e909542afca0b4dd4e1f73d8b4b987",
+        "source_type": "paper",
+        "content_sha256": "sha256:d855eafa066e991b82fbf68c54dc8f416b3159f68194159343e92c2f5d836c98"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Graph algorithms, such as shortest path finding, play a crucial role in enabling essential applications and services like infrastructure planning and navigation, making their correctness important. However, thoroughly testing graph algorithm implementations poses several challenges, including their vast input space (i.e., arbitrary graphs). Moreover, through our preliminary study, we find that just a few automatically generated graphs (less than 10) could be enough to cover the code of many graph algorithm implementations, rendering the code coverage-guided fuzzing approach -- one of the state-of-the-art search algorithms -- less efficient than expected. To tackle these challenges, we introduce GraphFuzz, the first automated feedback-guided fuzzing framework for graph algorithm implementations. Our key innovation lies in identifying lightweight and algorithm-specific feedback signals to combine with or completely replace the code coverage feedback to enhance the diversity of the test corpus, thereby speeding up the bug-finding process. This novel idea also allows GraphFuzz to effectively work in both black-box (i.e., no code coverage instrumentation/collection is required) and grey-box setups. GraphFuzz applies differential testing to detect both crash-triggering bugs and logic bugs. Our evaluation demonstrates the effectiveness of GraphFuzz. The tool has successfully discovered 12 previously unknown bugs, including 6 logic bugs, in 9 graph algorithm implementations in two popular graph libraries, NetworkX and iGraph. All of them have been confirmed and and 11 bugs have been rectified by the libraries' maintainers."
+    },
+    {
+      "id": "paper:doi:10.1145/3729319",
+      "title": "Graphiti: Bridging Graph and Relational Database Queries",
+      "authors": [
+        "Yang He",
+        "Ruijie Fang",
+        "Işıl Dillig",
+        "Yuepeng Wang"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Program. Lang.",
+      "doi": "10.1145/3729319",
+      "arxiv_id": "2504.03182",
+      "s2_paper_id": "778b9fb9fe2b30e6071d406d3abbbb82491ad070",
+      "url": "https://doi.org/10.1145/3729319",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3729319",
+      "abstract": "This paper presents an automated reasoning technique for checking equivalence between graph database queries written in Cypher and relational queries in SQL. To formalize a suitable notion of equivalence in this setting, we introduce the concept of database transformers, which transform database instances between graph and relational models. We then propose a novel verification methodology that checks equivalence modulo a given transformer by reducing the original problem to verifying equivalence between a pair of SQL queries. This reduction is achieved by embedding a subset of Cypher into SQL through syntax-directed translation, allowing us to leverage existing research on automated reasoning for SQL while obviating the need for reasoning simultaneously over two different data models. We have implemented our approach in a tool called Graphiti and used it to check equivalence between graph and relational queries. Our experiments demonstrate that Graphiti is useful both for verification and refutation and that it can uncover subtle bugs, including those found in Cypher tutorials and academic papers.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3729319",
+              "source_type": "paper_citation_context",
+              "excerpt": "Future work can further extend the transpilation rules and backend equivalence verifiers to support additional features.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d1b05a9bd7a2ec9749125714eeb9928dd51e9773da2f43e669b4a73bddb91ab3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/778b9fb9fe2b30e6071d406d3abbbb82491ad070",
+        "source_type": "paper",
+        "content_sha256": "sha256:5d5869a242d9b5526ba86d7bad724ee3c64f93102efbbac3d8cb67aa1a3db0eb"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3728973",
+      "title": "Hulk: Exploring Data-Sensitive Performance Anomalies in DBMSs via Data-Driven Analysis",
+      "authors": [
+        "Zhiyong Wu",
+        "Jie Liang",
+        "Jingzhou Fu",
+        "Mingzhe Wang",
+        "Yu Jiang"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Softw. Eng.",
+      "doi": "10.1145/3728973",
+      "arxiv_id": null,
+      "s2_paper_id": "9941e9dde900bcf3bbd359d4166157028ac29f38",
+      "url": "https://doi.org/10.1145/3728973",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3728973",
+      "abstract": "Performance is crucial for database management systems (DBMSs), and they are always designed to handle ever-changing workloads efficiently. However, the complexity of the cost-based optimizer (CBO) and its interactions can introduce implementation errors, leading to data-sensitive performance anomalies. These anomalies may cause significant performance degradation compared to the expected design under certain datasets. To diagnose performance issues, DBMS developers often rely on intuitions or compare execution times to a baseline DBMS. These approaches overlook the impact of datasets on performance. As a result, only a subset of performance issues is identified and resolved. In this paper, we propose Hulk to automatically explore these data-sensitive performance anomalies via data-driven analysis. The key idea is to identify performance anomalies as the dataset evolves. Specifically, Hulk estimates a reasonable response time range for each data volume to pinpoint performance cliffs. Then performance cliffs are checked for deviations from expected performance by finding a reasonable plan that aligns with performance expectations. We evaluate Hulk on six widely-used DBMSs, namely MySQL, MariaDB, Percona, TiDB, PostgreSQL, and AntDB. Hulk totally reports 135 anomalies, with 129 have been confirmed as new bugs, including 14 CVEs. Among them, 94 are data-sensitive performance anomalies.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3728973",
+              "source_type": "paper_citation_context",
+              "excerpt": "In addition, we compare Hulk with the state-of-the-art DBMS validation tools in industry, including both DBMS performance testing tool APOLLO [30] and SQLancer [11], as well as DBMS fuzzing tools Sqirrel [60].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b465834d79b5ad174ae351ee1e9f7797d0ae66a545ac7d7b370e56830fcde15a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3728973",
+              "source_type": "paper_citation_context",
+              "excerpt": "Additionally, CERT 1 [11] tests cardinality estimation to find performance anomalies.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Cardinality Estimation Restriction Testing (CERT), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:46c8661309d99cd451f8d7729be64a736b08830ff7780c9ef2e6da9e55242cbf",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728973.json",
+              "source_type": "paper",
+              "excerpt": "In addition, we compare Hulk with the state-of-the-art DBMS validation tools in industry, including both DBMS performance testing tool APOLLO [ 30] and SQLancer𝐶𝐸𝑅𝑇[11], as well as DBMS fuzzing tools Sqirrel [60].",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, 1 Introduction, page 3.",
+              "content_sha256": "sha256:52aa98ab75e719cc9db1fdf51e7bba48ab6dfb0bc2f3f689e64324140e015f0e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728973.json",
+              "source_type": "paper",
+              "excerpt": "Since only APOLLO [30] and SQLancer𝐶𝐸𝑅𝑇[11] are open-source tools, as a remedy, we still compared Hulk with other state-of-the-art SQL fuzzer ( Sqirrel [60]) to evaluate the capability to explore the state space on DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, 5.1 Evaluation Setup, page 13.",
+              "content_sha256": "sha256:1f0587448fc41010cf8507347ab765ec7de68083594e9747912ea22503d9e05a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728973.json",
+              "source_type": "paper",
+              "excerpt": "Number of confirmed performance anomalies and crashes, and covered branches on six DBMSs Performance Bugs Crashes Branches APOLLO 5 6 281,056 SQLancer 7 3 273,467 Sqirrel 0 9 312,713 Hulk 42 19 331,909 data-sensitive clauses on average (which will be discussed in Section 5.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 5.3 Comparison With Other Techniques, page 17.",
+              "content_sha256": "sha256:e209d8cd266b280822fd00bfd84671597b5ad72f3211fe26adfe6ca79f8922b8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728973.json",
+              "source_type": "paper",
+              "excerpt": "In addition, we compare Hulk with the state-of-the-art DBMS validation tools in industry, including both DBMS performance testing tool APOLLO [ 30] and SQLancer𝐶𝐸𝑅𝑇[11], as well as DBMS fuzzing tools Sqirrel [60].",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, 1 Introduction, page 3.",
+              "content_sha256": "sha256:52aa98ab75e719cc9db1fdf51e7bba48ab6dfb0bc2f3f689e64324140e015f0e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/9941e9dde900bcf3bbd359d4166157028ac29f38",
+        "source_type": "paper",
+        "content_sha256": "sha256:ec67b39edc9334d7425e14919b677c5fd3d61b54d96e56defdc8cbbfc4b6d201"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/tdsc.2024.3521591",
+      "title": "Improving Multitasking DBMS Fuzzing With More Accurate Coverage and Testcase Trimming",
+      "authors": [
+        "Jiaqi Li",
+        "Yajin Zhou",
+        "Lei Wu"
+      ],
+      "year": 2025,
+      "venue": "IEEE Transactions on Dependable and Secure Computing",
+      "doi": "10.1109/tdsc.2024.3521591",
+      "arxiv_id": null,
+      "s2_paper_id": "8693724c0f546d2da26e99439fd85eea5dfb79fd",
+      "url": "https://doi.org/10.1109/tdsc.2024.3521591",
+      "open_access_pdf": null,
+      "abstract": "Coverage-guided fuzzing is prevalent in detecting DBMS (Database Management System) bugs. However, current coverage-guided DBMS fuzzers suffer from two limitations that prevent fuzzers from discovering bugs efficiently. First, the coverage feedback is imprecise which prevents fuzzers from making optimal decisions on fuzzing strategies. Second, DBMS fuzzers lack testcase trimming to control the increasing input sizes. The large input size makes DBMS execution slower and reduces the likelihood that a mutation would touch important structures. In this paper, we proposed corresponding methods to overcome these limitations. Specifically, the work-task coverage tracking and unstable edge filtering improve the coverage accuracy with low instrumentation overhead. Based on more accurate coverage, we further propose testcase trimming to improve the speed of bug detection. We implemented a prototype named Tuzz and evaluated it on three popular DBMSs. The evaluation result shows that Tuzz explores 16.3%, 26.1%, and 26.6% more edges than the state-of-the-art fuzzer in PostgreSQL, MySQL, and MariaDB, respectively. More importantly, Tuzz has discovered 10 and 4 previously unknown bugs in MySQL and MariaDB.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/tdsc.2024.3521591",
+              "source_type": "paper_citation_context",
+              "excerpt": "Thesecondtypeproposesnewtestoraclestoidentifysemantic bugs in DBMSs. NoREC [35], TLP [36], DQE [37], Pinolo [38] leverage the idea of metamorphic testing to identify semantic bugs that lead to the incorrect result of SQL statement.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b0853993d08fb256c3e2b090fb518cfda4444d39773b44e9c2f36fe68808f8d9",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/8693724c0f546d2da26e99439fd85eea5dfb79fd",
+        "source_type": "paper",
+        "content_sha256": "sha256:a5137e549d49ca291ce3f4754937c40d63d5ba48acee86ab7731a47ab2009a9a"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse55347.2025.00013",
+      "title": "Janus: Detecting Rendering Bugs in Web Browsers via Visual Delta Consistency",
+      "authors": [
+        "Chijin Zhou",
+        "Quan Zhang",
+        "Bingzhou Qian",
+        "Yu Jiang"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse55347.2025.00013",
+      "arxiv_id": null,
+      "s2_paper_id": "3d2a62e3ce533a4bd184e9c0cdafea9c3e09e2b1",
+      "url": "https://doi.org/10.1109/icse55347.2025.00013",
+      "open_access_pdf": null,
+      "abstract": "Rendering lies at the heart of our modern web experience. However, the correctness of browser rendering is not always guaranteed, often leading to rendering bugs. Traditional differential testing, while successful in various domains, falls short when applied to rendering bug detection because an HTML file is likely yield different rendered outcomes across different browsers. This paper introduces Visual Delta Consistency, a test oracle to detect rendering bugs in web browsers, aiming to make rendered pages across browsers comparable. Our key insight is that any modifications made to an HTML file should uniformly influence rendering outcomes across browsers. Specifically, when presented with two HTML files that differ only by minor modifications, the reaction of all browsers should be consistent, i.e., either all browsers render them identically or all render them differently. Based on this insight, We implemented it as a practical fuzzer named Janus. It constructs pairs of slightly modified HTML files and observes the change statuses of the corresponding rendered pages across browsers for bug detection. We evaluated it on three widely-used browsers, i.e., Chrome, Safari, and Firefox. In total, Janus detected 31 non-crash rendering bugs, out of which 24 confirmed with 8 fixed.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00013",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, in web development, efforts [24], [25] have been made to formalize layout guidance of web development into a formal language for correctness verification; and in database management systems, tools like SQLancer [30], [29], [28] validate query results using predefined equivalence rules.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:78af177870b02e4d853c4be031408e11fb0081a060338257bee8a49c2fadfd0c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/3d2a62e3ce533a4bd184e9c0cdafea9c3e09e2b1",
+        "source_type": "paper",
+        "content_sha256": "sha256:901be665a7fb1049152bd2bd9c4d824bd0f219fcb5b960a2bded8b727d301e4f"
+      }
+    },
+    {
+      "id": "paper:s2:5bef8601da9663add6a85713414f8c945cf97355",
+      "title": "Language-Based Testing for Knowledge Graphs",
+      "authors": [
+        "Tobias John",
+        "E. Johnsen",
+        "Eduard Kamburjan",
+        "Dominic Steinhöfel"
+      ],
+      "year": 2025,
+      "venue": "Extended Semantic Web Conference",
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "5bef8601da9663add6a85713414f8c945cf97355",
+      "url": "https://www.semanticscholar.org/paper/5bef8601da9663add6a85713414f8c945cf97355",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "tlp",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "tlp",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/5bef8601da9663add6a85713414f8c945cf97355",
+              "source_type": "paper_citation_context",
+              "excerpt": "In recent years, metamorphic testing has gained popularity as an alternative [82, 83, 91].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c1f5029713981912082a85867f107a89be6de241c6493fe434b59feb255bddf6",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/5bef8601da9663add6a85713414f8c945cf97355",
+        "source_type": "paper",
+        "content_sha256": "sha256:dc2ac48666b3b2f822c663c8dbe84436b0ca35dfb840d17c8bb55972c7cf98c4"
+      },
+      "abstract": null
+    },
+    {
+      "id": "paper:doi:10.1145/3758316.3763249",
+      "title": "LLM-Assisted Dialect-Agnostic SQL Query Parsing",
+      "authors": [
+        "Junwen An"
+      ],
+      "year": 2025,
+      "venue": "SPLASH Companion",
+      "doi": "10.1145/3758316.3763249",
+      "arxiv_id": null,
+      "s2_paper_id": "541b4674b5c71530b124a319901697eb713f6746",
+      "url": "https://doi.org/10.1145/3758316.3763249",
+      "open_access_pdf": "https://doi.org/10.1145/3758316.3763249",
+      "abstract": "Query analysis and rewriting tools, which rely on analyzing the Abstract Syntax Trees (ASTs) of queries, are essential in database workflows. However, due to the diverse SQL dialects, traditional grammar-based parsers often fail when encountering dialect-specific syntax. Although Large Language Models (LLMs) show promise in understanding SQL queries, they struggle in accurately generating ASTs. To address this, we propose SQLFlex, a hybrid approach that iteratively uses a grammar-based parser and, upon failure, employs an LLM to segment the query into smaller, parsable parts. SQLFlex successfully parsed 96.37% of queries across eight dialects on average, and demonstrated its practicality in SQL linting and test case reduction.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/541b4674b5c71530b124a319901697eb713f6746",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/541b4674b5c71530b124a319901697eb713f6746",
+        "source_type": "paper",
+        "content_sha256": "sha256:efe0a9bac6ded94aeffe1524d94256917f9a9db43ff6f2ca284a409c93a521e8"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/ase63991.2025.00322",
+      "title": "LLM-based Dynamic Differential Testing for Database Connectors with Reinforcement Learning-Guided Prompt Selection",
+      "authors": [
+        "C. Lyu",
+        "Minghao Zhao",
+        "Yanhao Wang",
+        "Liang Jie"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Automated Software Engineering",
+      "doi": "10.1109/ase63991.2025.00322",
+      "arxiv_id": "2506.11870",
+      "s2_paper_id": "f3e1f353f63a93021ee0ea3b535187585f6be98c",
+      "url": "https://doi.org/10.1109/ase63991.2025.00322",
+      "open_access_pdf": null,
+      "abstract": "Database connectors are critical components that enable applications to interact with database management systems (DBMS) but their security vulnerabilities are often neglected. Unlike traditional software defects, connector vulnerabilities exhibit subtle behavioral patterns and are inherently challenging to detect. Moreover, non-standardized implementation of connectors leaves potential risks (i.e., unsafe implementations) but is more elusive. As a result, existing fuzzing methods are ineffective in finding such vulnerabilities. Even large language model (LLM)-based methods are still incapable of generating test cases that can invoke all the interface and internal logic of database connectors due to a lack of domain knowledge.In this paper, we propose a new LLM-based test case generation method guided by reinforcement learning (RL) for database connector testing. Specifically, to equip the LLM with sufficient and appropriate domain knowledge, a parameterized template is composed for prompt construction. The LLM then generates test cases instructed by the constructed prompts, which are dynamically evaluated through differential testing across multiple connectors. The testing process is carried out iteratively, where RL is adopted to select the optimal prompt in each round based on behavioral feedback from the previous rounds, to maximize the efficiency of discovering inconsistencies. Finally, we implement and evaluate the aforementioned methodology on two widely used JDBC connectors, namely MySQL Connector/J and OceanBase Connector/J. In the preliminary results, we have reported 16 bugs, among which 10 are officially confirmed, and the rest are acknowledged as unsafe implementations.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/ase63991.2025.00322",
+              "source_type": "paper_citation_context",
+              "excerpt": "We adapted SQLancer [2], a state-of-the-art JDBC-based database testing tool, to support multiple database connectors and evaluated both approaches over 100 rounds.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:fec6c7253984b89e34d66230b7badb73b12b2c1a8b1bb10e596e86a068af27b7",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/f3e1f353f63a93021ee0ea3b535187585f6be98c",
+        "source_type": "paper",
+        "content_sha256": "sha256:e3abcd1e38ba1dffaf4abee003001b80cd29347f1d5c90ec10a472008f18b9e5"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3744916.3773102",
+      "title": "Locus: Agentic Predicate Synthesis for Directed Fuzzing",
+      "authors": [
+        "Jie Zhu",
+        "Chihao Shen",
+        "Ziyang Li",
+        "Jiahao Yu",
+        "Yizheng Chen",
+        "Kexin Pei"
+      ],
+      "year": 2025,
+      "venue": "arXiv.org",
+      "doi": "10.1145/3744916.3773102",
+      "arxiv_id": "2508.21302",
+      "s2_paper_id": "1ea85ffff5bc84a288f30b95c2f2d44f2df5e633",
+      "url": "https://doi.org/10.1145/3744916.3773102",
+      "open_access_pdf": null,
+      "abstract": "Directed fuzzing aims to find program inputs that lead to specified target program states. It has broad applications, such as debugging system crashes, confirming reported bugs, and generating exploits for potential vulnerabilities. This task is inherently challenging because target states are often deeply nested in the program, while the search space manifested by numerous possible program inputs is prohibitively large. Existing approaches rely on branch distances or manually-specified constraints to guide the search; however, the branches alone are often insufficient to precisely characterize progress toward reaching the target states, while the manually specified constraints are often tailored for specific bug types and thus difficult to generalize to diverse target states and programs. We present Locus, a novel framework to improve the efficiency of directed fuzzing. Our key insight is to synthesize predicates to capture fuzzing progress as semantically meaningful intermediate states, serving as milestones towards reaching the target states. When used to instrument the program under fuzzing, they can reject executions unlikely to reach the target states, while providing additional coverage guidance. To automate this task and generalize to diverse programs, Locus features an agentic framework with program analysis tools to synthesize and iteratively refine the candidate predicates, while ensuring the predicates strictly relax the target states to prevent false rejections via symbolic execution. Our evaluation shows that Locus substantially improves the efficiency of eight state-of-the-art fuzzers in discovering real-world vulnerabilities, achieving an average speedup of 41.6x. So far, Locus has found nine previously unpatched bugs, with three already acknowledged with draft patches.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3744916.3773102",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, such feedback is sometimes too sparse or indirect to reliably measure the progress, especially when there is a long chain of implicit preconditions guarding the target program states [2, 26, 36, 44, 45, 66, 73, 80, 106].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f57256b0468a0a7723a3425906e39bd976e874c45916a1a7dde613f9e493fc24",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/1ea85ffff5bc84a288f30b95c2f2d44f2df5e633",
+        "source_type": "paper",
+        "content_sha256": "sha256:fdcbe9ea7f2e5eb3ad29a3b16b06b61d9b165e25ebf11c8d0f8bb6b7ccc24d51"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3832104",
+      "title": "Metamorphic Coverage",
+      "authors": [
+        "Jinsheng Ba",
+        "Yuancheng Jiang",
+        "Manuel Rigger"
+      ],
+      "year": 2025,
+      "venue": "arXiv.org",
+      "doi": "10.1145/3832104",
+      "arxiv_id": "2508.16307",
+      "s2_paper_id": "f1e3093179b8f4dcf04a3914c2d4fa127d88ddb2",
+      "url": "https://doi.org/10.1145/3832104",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3832104",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC and TLP are implemented in SQLancer , which generates test inputs by Query Plan Guidance ( QPG ) [4].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:583792a0e4a2ff2727371729cc52bb0cb5fa04ec118bf7c9961ad8a7cf45c885",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3832104",
+              "source_type": "paper_citation_context",
+              "excerpt": "For a fair comparison, we used QPG as a reference and reused its test input generation workflow.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1c23d67db374dc5c3e0f0236594b78eced5eeb503ee4a5a9317cd57099e03d56",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3832104",
+              "source_type": "paper_citation_context",
+              "excerpt": "We believe most metamorphic testing methods are designed based on this intuition [32, 37, 44].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b6dec738695c73d2f2b1427b2ff1a996a4f1ca86ee9c368b44cc2b24758200cb",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3832104",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, concerning code coverage, in NoREC [44], the authors claimed that “code coverage is not particularly useful for fuzzing DBMS, since high coverage for the core components (e.g., the query optimizer) can be achieved quickly.”",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:854c0789882cec7127f4737c267840a6355767f35c8f16831a173351c25dee2e",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3832104",
+              "source_type": "paper_citation_context",
+              "excerpt": "Non-optimizing Reference Engine Construction ( NoREC ) [44] and Ternary Logic Partitioning ( TLP ) [45] test the query processing functionality of Database Management Systems (DBMSs).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:56cab2a937dff2cce1cd0d8666b89bef4d8b0564a6ad24abe58222daeacb0c77",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3832104",
+              "source_type": "paper_citation_context",
+              "excerpt": "If MC does not grow for a fixed number of iterations, the database is mutated by executing SQL statements to broaden the search space, mirroring the schedule used by QPG and CCG .",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:167cb664e56452405d90d936d847ca2352eb359dbf3c8cccdccceb9308727a89",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/f1e3093179b8f4dcf04a3914c2d4fa127d88ddb2",
+        "source_type": "paper",
+        "content_sha256": "sha256:cb1902162f9d98747ed10ae178eb32287363638a6be2ce65af77303501c75354"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Metamorphic testing is a widely used methodology that examines an expected relation between pairs of executions to automatically find bugs, such as correctness bugs. We found that code coverage cannot accurately measure the extent to which code is validated and mutation testing is computationally expensive for evaluating metamorphic testing methods. In this work, we propose Metamorphic Coverage (MC), a coverage metric that examines the distinct code executed by pairs of test inputs within metamorphic testing. Our intuition is that, typically, a bug can be observed if the corresponding code is executed when executing either test input but not the other one, so covering more differential code covered by pairs of test inputs might be more likely to expose bugs. While most metamorphic testing methods have been based on this general intuition, our work defines and systematically evaluates MC on five widely used metamorphic testing methods for testing database engines, compilers, and constraint solvers. The code measured by MC overlaps with the bug-fix locations of 50 of 64 bugs found by metamorphic testing methods, and MC has a stronger positive correlation with bug numbers than line coverage. MC is 4x more sensitive than line coverage in distinguishing testing methods'effectiveness, and the average value of MC is 6x smaller than line coverage while still capturing the part of the program that is being tested. MC required 359x less time than mutation testing. Based on a case study for an automated database system testing approach, we demonstrate that when used for feedback guidance, MC significantly outperforms code coverage, by finding 41\\% more bugs. Consequently, this work might have broad applications for assessing metamorphic testing methods and improving test-case generation."
+    },
+    {
+      "id": "paper:doi:10.14722/ndss.2025.230530",
+      "title": "MTZK: Testing and Exploring Bugs in Zero-Knowledge (ZK) Compilers",
+      "authors": [
+        "Dongwei Xiao",
+        "Zhibo Liu",
+        "Yiteng Peng",
+        "Shuai Wang"
+      ],
+      "year": 2025,
+      "venue": "Network and Distributed System Security Symposium",
+      "doi": "10.14722/ndss.2025.230530",
+      "arxiv_id": null,
+      "s2_paper_id": "654064b6d2b49ac7d66bc4a8b296a4806221ab66",
+      "url": "https://doi.org/10.14722/ndss.2025.230530",
+      "open_access_pdf": "https://doi.org/10.14722/ndss.2025.230530",
+      "abstract": "—Zero-knowledge (ZK) proofs have been increasingly popular in privacy-preserving applications and blockchain systems. To facilitate handy and efficient ZK proof generation for normal users, the industry has designed domain-specific languages (DSLs) and ZK compilers. Given a program in ZK DSL, a ZK compiler compiles it into a circuit, which is then passed to the prover and verifier for ZK checking. However, the correctness of ZK compilers is not well studied, and recent works have shown that de facto ZK compilers are buggy, which can allow malicious users to generate invalid proofs that are accepted by the verifier, causing security breaches and financial losses in cryptocurrency. In this paper, we propose MTZK, a metamorphic testing framework to test ZK compilers and uncover incorrect compila-tions. Our approach leverages deliberately designed metamorphic relations (MRs) to mutate ZK compiler inputs. This way, ZK compilers can be automatically tested for compilation correctness using inputs and mutated variants without requiring manual intervention. We propose a set of design considerations and optimizations to deliver an efficient and effective testing framework. In the evaluation of four industrial ZK compilers, we successfully uncovered 21 bugs, out of which the developers have promptly patched 15. We also show possible exploitations of the uncovered bugs to demonstrate their severe security implications.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14722/ndss.2025.230530",
+              "source_type": "paper_citation_context",
+              "excerpt": "MTZK shares a similar testing-based approach with quality assurance tools for critical systems, such as CPUs [56], [46], databases [89], [59], [60], [90], and operating systems [65], [16], to detect errors instead of proving their absence.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f9bb405b0cbcd82f512e40e3680fdd93e58dd8d7fe9d274079005f7054e21e67",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/654064b6d2b49ac7d66bc4a8b296a4806221ab66",
+        "source_type": "paper",
+        "content_sha256": "sha256:a7c8f300f9380babbce39c26b95b9d16bcd97b8c002d4419169503a46f2e1730"
+      }
+    },
+    {
+      "id": "paper:doi:10.3390/electronics14193910",
+      "title": "Nabil: A Text-to-SQL Model Based on Brain-Inspired Computing Techniques and Large Language Modeling",
+      "authors": [
+        "Feng Zhou",
+        "Shijing Hu",
+        "Xiaozheng Du",
+        "Nan Li",
+        "Tongming Zhou",
+        "Yanni Zhao",
+        "Sitong Shang",
+        "Xufeng Ling",
+        "Huaizhong Zhu"
+      ],
+      "year": 2025,
+      "venue": "Electronics",
+      "doi": "10.3390/electronics14193910",
+      "arxiv_id": null,
+      "s2_paper_id": "aeaf7936aa6c16580244cb4f315b3387d0d6a7e2",
+      "url": "https://doi.org/10.3390/electronics14193910",
+      "open_access_pdf": "https://mdpi.com/2079-9292/14/19/3910/pdf?version=1759246965",
+      "abstract": "Human-database interaction is inevitable in intelligent system applications, and accurately converting user-entered natural language into database query language is a critical step. To improve the accuracy, generalization, and robustness of text-to-SQL, we propose Nabil (a model for natural language conversion query language based on brain-inspired computing technology and a large language model). This model first leverages the spatiotemporal encoding capabilities of spiking neural networks to capture semantic features of natural language, then fuses these features with those generated by a large language model. Finally, a champion model is designed to select the optimal query from multiple candidate SQLs. Experiments were conducted on three database engines, DuckDB, MySQL, and PostgreSQL, and the model’s effectiveness was verified on benchmark datasets such as BIRD. The results show that Nabil outperforms existing baseline methods in both execution accuracy and effective efficiency scores. Furthermore, our proposed normalization and syntax tree abstraction algorithms further enhance the champion model’s discriminative capabilities, providing new insights for text-to-SQL research.",
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.3390/electronics14193910",
+              "source_type": "paper_citation_context",
+              "excerpt": "Ba, J. et al. proposed query plan guidance (QPG) to fully automatically test errors in database systems and applied it to three database systems: SQLite, TiDB, and CockroachDB [20].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9728e2a780870d24d19cd5a135e83e013cdf11c0064624a710ae08c1fc7db529",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/aeaf7936aa6c16580244cb4f315b3387d0d6a7e2",
+        "source_type": "paper",
+        "content_sha256": "sha256:e3d5625cb34c739f425ee4077f82ebebab0c7fabfcda79ba60e218e939128d66"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icccs65393.2025.11069832",
+      "title": "OptionFuzz: The Fuzzer with Coverage-Guided Dynamic Configuration Scheduling",
+      "authors": [
+        "Lang Chu",
+        "Minhuan Huang",
+        "Xiang Li",
+        "Yuanping Nie",
+        "Wenyu Zhen",
+        "Qian Yan"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Communication, Computing & Security",
+      "doi": "10.1109/icccs65393.2025.11069832",
+      "arxiv_id": null,
+      "s2_paper_id": "46eca54a5aca680d689a06d3f93a3852b7fabefb",
+      "url": "https://doi.org/10.1109/icccs65393.2025.11069832",
+      "open_access_pdf": null,
+      "abstract": "As a security testing technique, fuzzing has demonstrated its powerful capabilities. In fuzzing, the configuration of the target system is crucial, as it often determines the program's execution paths. A thorough exploration of configurations can significantly enhance the coverage of fuzzing, which has drawn substantial attention from researchers. Currently, research on configuration exploration in fuzzing is mainly divided into two categories: mutation-based and generation-based approaches. Among these, generation-based methods have achieved notable success due to their efficient configuration generation processes. However, these approaches also face challenges, such as dependence on formal documentation and a lack of configuration scheduling strategies. In this paper, we propose OptionFuzz, a configuration-aware fuzzing framework based on generation, designed to address some of the limitations of previous approaches. The framework first generates configurations through a Markov decision process based on edge coverage analysis, followed by configuration scheduling guided by function coverage analysis. In tests conducted on 9 real-world popular programs, OptionFuzz explored 46.9 % more execution paths compared to traditional fuzzers. Additionally, compared to other similar works in the field, OptionFuzz achieved an average 15.6 % higher coverage rate on the same targets. Furthermore, we conducted longterm testing on the latest versions of the experimental programs using OptionFuzz, discovering 17 unique crashes triggered by configurations generated by the framework, of which 5 have been confirmed and assigned CNNVD identifiers.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icccs65393.2025.11069832",
+              "source_type": "paper_citation_context",
+              "excerpt": "The second aspect aims to expand the application of fuzzing to more targets, such as operating systems[9–12], virtual machine managers[13], network protocols[13, 14], database systems[15, 16], and even autonomous vehicles[17].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f7a9c04b08fcc91ef582279c2bdf6c8845f8b56521dd9b4d37cf91eb7ca53a70",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/46eca54a5aca680d689a06d3f93a3852b7fabefb",
+        "source_type": "paper",
+        "content_sha256": "sha256:70fa9d8d2d200217926f73ad158397690fc02446d80f49f11fbb911dc161ce54"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3757347.3759132",
+      "title": "PANGOLIN: a Comprehensive Testing Framework for Configuration-Rich Key-Value Stores",
+      "authors": [
+        "Shaohua Duan",
+        "Sudarsun Kannan",
+        "Andrea C. Arpaci-Dusseau",
+        "Remzi H. Arpaci-Dusseau"
+      ],
+      "year": 2025,
+      "venue": "Annual Haifa Experimental Systems Conference",
+      "doi": "10.1145/3757347.3759132",
+      "arxiv_id": null,
+      "s2_paper_id": "34675858280c9b98a3df15fab99c590458e35afd",
+      "url": "https://doi.org/10.1145/3757347.3759132",
+      "open_access_pdf": null,
+      "abstract": "In this paper, we present Pangolin, a comprehensive testing framework for configuration-rich key-value stores. To better understand bugs in modern key-value stores and explore domain knowledge for efficiently identifying new ones, we first comprehensively study historical bugs in five mature key-value stores during the last eight years. Then, we design and implement Pangolin, which is motivated by insights from our bug study, which indicated most bugs could be identified by systematically testing a small sequence of operations and configurations. Specifically, Pangolin practices these insights by introducing a bounded testing strategy into a spectrum of black-box and fuzzing test procedures. Finally, we utilize Pangolin to find 20 bugs and reproduce 443 historical bugs in five mature key-value stores (RocksDB, LevelDB, HyperlevelDB, BadgerDB, and Redis), making it an attractive supplement to handwritten test suites.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3757347.3759132",
+              "source_type": "paper_citation_context",
+              "excerpt": "Black-box testing and fuzzing have been used to identify various types of bugs in File systems and DBMSs, including crash inconsistency bugs [21, 25], logic bugs [31], and correctness bugs [3].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8dbf8f7a20867580ac93414694996467aece7021f6aea81b67cc460c3af7b58f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/34675858280c9b98a3df15fab99c590458e35afd",
+        "source_type": "paper",
+        "content_sha256": "sha256:75526596ca1cbd818d2a64a40f9030d5233942deb04701d05ba9b33bc290f89c"
+      }
+    },
+    {
+      "id": "paper:arxiv:2503.03893",
+      "title": "Parser Knows Best: Testing DBMS with Coverage-Guided Grammar-Rule Traversal",
+      "authors": [
+        "Yu Liang",
+        "Hong Hu"
+      ],
+      "year": 2025,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2503.03893",
+      "s2_paper_id": "45c07812d74c526823b806aaa2c85a4193ecac16",
+      "url": "https://arxiv.org/abs/2503.03893",
+      "open_access_pdf": null,
+      "abstract": "Database Management System (DBMS) is the key component for data-intensive applications. Recently, researchers propose many tools to comprehensively test DBMS systems for finding various bugs. However, these tools only cover a small subset of diverse syntax elements defined in DBMS-specific SQL dialects, leaving a large number of features unexplored. In this paper, we propose ParserFuzz, a novel fuzzing framework that automatically extracts grammar rules from DBMSs' built-in syntax definition files for SQL query generation. Without any input corpus, ParserFuzz can generate diverse query statements to saturate the grammar features of the tested DBMSs, which grammar features could be missed by previous tools. Additionally, ParserFuzz utilizes code coverage as feedback to guide the query mutation, which combines different DBMS features extracted from the syntax rules to find more function and safety bugs. In our evaluation, ParserFuzz outperforms all state-of-the-art existing DBMS testing tools in terms of bug finding, grammar rule coverage and code coverage. ParserFuzz detects 81 previously unknown bugs in total across 5 popular DBMSs, where all bugs are confirmed and 34 have been fixed.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2503.03893",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer +QPG supports testing with SQLite , CockroachDB and TiDB , and outperforms all other logic bug detectors including SQLRight [2, 21].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5d7d1b4b1440073880bd719184c6a2d332c8638e3a550dcee65be8c7b102ebbf",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2503.03893",
+              "source_type": "paper_citation_context",
+              "excerpt": "Because NoREC oracle is claimed to be a better performer over-all compared to TLP oracle [2].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ccc4fc990046dd11f3e3d8edfdd7f4ad2d417d836f58f7d0a63b79a343120236",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2503.03893",
+              "source_type": "paper_citation_context",
+              "excerpt": "With its latest configuration SQLancer +QPG [2], it uses the DBMS query plan to guide its query generation in order to stress test the DBMS query optimization logic.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:97a771d50f5bf6e079946007165c48965a3e26bfe36a3d9b95ceea36e9a1eea6",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2503.03893",
+              "source_type": "paper_citation_context",
+              "excerpt": "Recent efforts on DBMS testing [2, 14, 47, 67] can be classified into two categories: generation-based testing and mutation-based grey-box fuzzing.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c3add06c99559a2350fff51470cb034f03a1a42b7ee54aeccb4cf3c4721bcf5a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2503.03893",
+              "source_type": "paper_citation_context",
+              "excerpt": "There is another generation-based DBMS testing tool called SQLancer [24], that focuses on detecting DBMS logic errors from DBMS systems [32–34].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d7b7aa3667101c9b14a71ff3043e3e80cd1bc32d28af5d78fa0cb7b80465ad98",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2503.03893",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer introduces a few SQL oracles for this purpose such as NoREC , TLP and PQS , where each shares a distinct SQL pattern to match [32–34].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c9f7da26285088d070904045658a9a65176e4e2039174bcf2445c637e7ea09d5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/45c07812d74c526823b806aaa2c85a4193ecac16",
+        "source_type": "paper",
+        "content_sha256": "sha256:6c2c7219b8751ee14e9ef8f3928e1ba54672b62d06e2fc54be9f00988d270319"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3749646.3749661",
+      "title": "PBench: Workload Synthesizer with Real Statistics for Cloud Analytics Benchmarking",
+      "authors": [
+        "Yan Zhou",
+        "Chunwei Liu",
+        "B. Urgaonkar",
+        "Zhengle Wang",
+        "M. Mueller",
+        "Chao Zhang",
+        "Songyue Zhang",
+        "Pascal Pfeil",
+        "Dominik Horn",
+        "Zhengchun Liu",
+        "Davide Pagano",
+        "Tim Kraska",
+        "Samuel Madden",
+        "Ju Fan"
+      ],
+      "year": 2025,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3749646.3749661",
+      "arxiv_id": "2506.16379",
+      "s2_paper_id": "f174394f4819f1387b1405ee41f9e7219b325573",
+      "url": "https://doi.org/10.14778/3749646.3749661",
+      "open_access_pdf": null,
+      "abstract": "Cloud service providers commonly use standard benchmarks like TPC-H and TPC-DS to evaluate and optimize cloud data analytics systems. However, these benchmarks rely on fixed query patterns and fail to capture real execution statistics of production cloud workloads. Although some cloud database vendors have recently released real workload traces, these traces alone do not qualify as benchmarks, as they typically lack essential components (i.e., queries and databases). To overcome this limitation, this paper studies a new problem of\n workload synthesis with real statistics\n , which generates\n synthetic workloads\n that closely approximate real execution statistics, including key performance metrics and operator distributions. To address this problem, we propose PBench, a novel workload synthesizer that constructs synthetic workloads by (1) selecting and combining workload components from existing benchmarks and (2) augmenting new workload components. This paper studies the key challenges in PBench. First, we address the challenge of balancing performance metrics and operator distributions by introducing a multi-objective optimization-based component selection method. Second, to capture the temporal dynamics of real workloads, we design a timestamp assignment method that progressively reines workload timestamps. Third, to handle the disparity between the original workload and the candidate workload, we propose a component augmentation approach that leverages large language models (LLMs) to generate additional workload components while maintaining statistical idelity. Experimental results show that PBench reduces approximation error by up to 6X compared to state-of-the-art methods.",
+      "cites_seed_techniques": [
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3749646.3749661",
+              "source_type": "paper_citation_context",
+              "excerpt": "Then, the generated synthetic workload o \" ers an e \" ective so-lution for downstream tasks , like database benchmarking [9, 39, 40], performance tuning [30], and bug detection [3], allowing database developers to perform realistic evaluations while preserving privacy and avoiding exposure of…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8c0a35b2e4afb470846e77467bca9f53f10338f8805f5a8c26216ff8ead01513",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/f174394f4819f1387b1405ee41f9e7219b325573",
+        "source_type": "paper",
+        "content_sha256": "sha256:4d93bc9bf8012401b5fc8e5ee0c10b5dcf49889b3ba61f1d829e2c347efaf005"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse55347.2025.00045",
+      "title": "PUPPY: Finding Performance Degradation Bugs in DBMSs via Limited-Optimization Plan Construction",
+      "authors": [
+        "Zhiyong Wu",
+        "Jie Liang",
+        "Jingzhou Fu",
+        "Mingzhe Wang",
+        "Yu Jiang"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse55347.2025.00045",
+      "arxiv_id": null,
+      "s2_paper_id": "263f26b2c7bca1d7ba82ab98f0824bcc5b291181",
+      "url": "https://doi.org/10.1109/icse55347.2025.00045",
+      "open_access_pdf": null,
+      "abstract": "Database management systems (DBMSs) consistently strive for enhanced performance. For a given query, the optimizer of a DBMS aims to construct an optimal execution plan that incorporates multiple optimization operations. However, the resulting plan may sometimes perform worse than even if no optimizations were applied. This occurs because the interactions between optimizations are complex and some situations might be overlooked in the implementation. We refer to these issues as Performance Degradation Bugs (PDBs). PDBs can result in significant consequences from decreased system efficiency and prolonged query processing times to potential disruptions in critical business operations. In this paper, we present PUPPY, an automated approach for detecting PDBs in DBMSs using limited-optimization plan construction. The key idea is to compare the performance with the plan generated with all optimization operations enabled, against the plan generated with only a subset of optimization operations in the same DBMS. If the response time of the plan with the limited optimization set is shorter than that of the fully optimized plan, it indicates a potential PDB. Specifically, PUPPY first generates queries that incorporate multiple optimization sequences, guided by optimization operation sequence coverage. Secondly, PUPPY analyzes the query plan and selectively disables specific optimizations to construct the limited optimization plan. We evaluate PUPPY on five widely-used DBMSs, namely MySQL, Percona, TiDB, PolarDB, and PostgreSQL against the state-of-the-art DBMS performance testing tools APOLLO and AMOEBA. More importantly, PUPPY reports 62 PDBs, with 54 anomalies confirmed as previously unknown bugs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00045",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLsmith [41] generates queries based on built-in code that embeds the AST generation rules for the target DBMS. SQLancer [40, 39, 38] aims to find logic bugs and it generates queries based on the test oracle it builds.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d8f11c46d389542a7e13a42569151c2147b67c906a53cf3b636f10d0ad52dac3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00045",
+              "source_type": "paper_citation_context",
+              "excerpt": "QPG [5] gradually mutates DDL and DML statements to change database states, aiming to cover more unique query plans to cover more DBMS logic.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a0703a0ff19eb6ac552239842fcd8589619254394e5afcd95cddfe4bd711c7c8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00045.json",
+              "source_type": "paper",
+              "excerpt": "To further evaluate the performance of PUPPY, we also compare PUPPY with SQLancer, SQLsmith, and SQUIRREL, which are widely used in industry.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, D Efficiency of the Optimization Guided Algorithm, page 10.",
+              "content_sha256": "sha256:e5981a33b111bb06e3fe4ea6611280d0543661097c8d8137fd3cebba210b6cba",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00045.json",
+              "source_type": "paper",
+              "excerpt": "It shows that PUPPY outperforms SQLancer and SQLsmith in detecting bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, D Efficiency of the Optimization Guided Algorithm, page 10.",
+              "content_sha256": "sha256:5bcc70147a3e4799dba3adcea9681ef97a125823188f084f9abc1c71521da83f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00045.json",
+              "source_type": "paper",
+              "excerpt": "Specifically, PUPPY detect detected a total of 35 bugs (including 30 performance bugs and 5 crash bugs) in 48 hours, while SQLancer, SQLsmith and SQUIRREL only detected 29, 31, and 30 bugs in total.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, D Efficiency of the Optimization Guided Algorithm, page 10.",
+              "content_sha256": "sha256:617144c492850e1153a1236b8dcfa307e02aba7c71f7c96870e48e6021ed8db0",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/263f26b2c7bca1d7ba82ab98f0824bcc5b291181",
+        "source_type": "paper",
+        "content_sha256": "sha256:f1c8009322a7a8fbfea600a220607f6671df0a28c3bf15dbf9ace0db2d3e2dc9"
+      }
+    },
+    {
+      "id": "paper:arxiv:2503.17322",
+      "title": "QITE: Assembly-Level, Cross-Platform Testing of Quantum Computing Platforms",
+      "authors": [
+        "Matteo Paltenghi",
+        "Michael Pradel"
+      ],
+      "year": 2025,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2503.17322",
+      "s2_paper_id": "4135ce8f1c949407b786fbdf47a11f532126d257",
+      "url": "https://arxiv.org/abs/2503.17322",
+      "open_access_pdf": null,
+      "abstract": "Quantum computing platforms are susceptible to quantum-specific bugs (e.g., incorrect ordering of qubits or incorrect implementation of quantum abstractions), which are difficult to detect and require specialized expertise. The field faces challenges due to a fragmented landscape of platforms and rapid development cycles that often prioritize features over the development of robust platform testing frameworks, severely hindering the reliability of quantum software. To address these challenges, we present QITE, the first cross-platform testing framework for quantum computing platforms, which leverages QASM, an assembly-level representation, to ensure consistency across different platforms. QITE introduces the novel ITE process to generate equivalent quantum programs by iteratively (I)mporting assembly into platform representations, (T)ransforming via platform optimization and gate conversion, and (E)xporting back to assembly. It uses a crash oracle to detect failures during cross-platform transformations and an equivalence oracle to validate the semantic consistency of the final sets of assembly programs, which are expected to be equivalent by construction. We evaluate QITE on four widely-used quantum computing platforms: Qiskit, PennyLane, Pytket, and BQSKit, revealing 17 bugs, 14 of which are already confirmed or even fixed. Our results demonstrate QITE's effectiveness, its complementarity to existing quantum fuzzers in terms of code coverage, and its ability to expose bugs that have been out of reach for existing testing techniques.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2503.17322",
+              "source_type": "paper_citation_context",
+              "excerpt": "Grammar-based generators, similar to our approach, have been successfully applied to test different compilers and similar software, such as the Java Virtual Machine [40], C compilers [13, 50], SMT solvers [47], database engines [38], and deep learning software infrastructure [23, 26, 27, 45].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d1eb508359d74ef2a50107193a6d19c0f50fac753353acd7b707c29683a63932",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/4135ce8f1c949407b786fbdf47a11f532126d257",
+        "source_type": "paper",
+        "content_sha256": "sha256:c6c333d4ad3e9bf313336c45ac8f3b86426b36c9fcb7f63f6383fb13528dbcf9"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3725688.3725698",
+      "title": "QOVIS: Understanding and Diagnosing Query Optimizer via a Visualization-assisted Approach (Revision)",
+      "authors": [
+        "Zhengxin You",
+        "Qiaomu Shen",
+        "M. Yiu",
+        "Bo Tang"
+      ],
+      "year": 2025,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3725688.3725698",
+      "arxiv_id": null,
+      "s2_paper_id": "6f604bfd94c41078af14805149229e4e5a0eb13a",
+      "url": "https://doi.org/10.14778/3725688.3725698",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "norec",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3725688.3725698",
+              "source_type": "paper_citation_context",
+              "excerpt": "In addition, automatic logic bug detection tools [27, 50, 59, 60] are proposed to identify the logic bugs during the query processing.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9a4357e17a57107e2fdacefc67d507057768d7de24a247e1fa5bb8960d6db41a",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3725688.3725698",
+              "source_type": "paper_citation_context",
+              "excerpt": "Many studies have been proposed to automatically detect the issues in the relational data-base management system (RDBMS) [18, 50, 59].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8c01d1a9d1c966c0ba965337ad0318f8c224a236197ba2565176b6d114b785a8",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3725688.3725698",
+              "source_type": "paper_citation_context",
+              "excerpt": "Rigger et al. [50] focus on detecting logical bugs when the generated plans produce incorrect results w.r.t the ground truth.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a9e00741354ecdf0c751492870fec06ad3aa142e29ec5904cebcff6418d50361",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/6f604bfd94c41078af14805149229e4e5a0eb13a",
+        "source_type": "paper",
+        "content_sha256": "sha256:66ff3203d6e6c8895c45720f86bb2adfe0e435092dc7eac707c6ab24b63c924f"
+      },
+      "abstract": "Understanding and diagnosing query optimizers is crucial to guarantee the correctness and efficiency of query processing in database systems. However, achieving this is non-trivial as there are three technical challenges: (i) hundreds and thousands of query plans are generated for each query during the query optimization procedure; (ii) the transformation logic among query plans is not easy to investigate even for expert database system developers; and (iii) navigating users to the root causes of the bugs/errors is inherently hard as the changes of the operators among query plans are missing in the query processing log. In this work, we propose QOVIS to overcome these challenges, which identifies the query optimization bugs/issues and investigates their root causes via a visualization-assisted approach. Specifically, QOVIS consists of data preprocessing layer, transformation logic computation layer, and visual analysis layer. We conduct extensive experimental studies (e.g., user study, case study, and performance study) to evaluate the efficiency and effectiveness of QOVIS. In particular, our user study (on 24 database developers and researchers) confirms that QOVIS significantly reduces the time required to investigate the bugs/errors in the query optimizer. Moreover, the generality of QOVIS is verified by utilizing it to understand and diagnose the real-world reported bugs/errors in different query optimizers of three widely-used systems: Apache Spark, Apache Hive, and DuckDB."
+    },
+    {
+      "id": "paper:doi:10.1145/3728908",
+      "title": "QTRAN: Extending Metamorphic-Oracle Based Logical Bug Detection Techniques for Multiple-DBMS Dialect Support",
+      "authors": [
+        "Li Lin",
+        "Qinglin Zhu",
+        "Hongqiao Chen",
+        "Zhuangda Wang",
+        "Rongxin Wu",
+        "Xiaoheng Xie"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Softw. Eng.",
+      "doi": "10.1145/3728908",
+      "arxiv_id": null,
+      "s2_paper_id": "196fb1deaa2eda163dcfc0a7e0a7dec93228fea3",
+      "url": "https://doi.org/10.1145/3728908",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3728908",
+      "abstract": "Metamorphic testing is a widely used method to detect logical bugs in Database Management Systems (DBMSs), referred to herein as MOLT (Metamorphic-Oracle based Logical Bug Detection Technique). This technique involves constructing SQL statement pairs, including original and mutated queries, and assessing whether the execution results conform to predefined metamorphic relations to detect logical bugs. However, current MOLTs rely heavily on specific DBMS grammar to generate valid SQL statement pairs, which makes it challenging to adapt these techniques to various DBMSs with different grammatical structures. As a result, only a few popular DBMSs, such as PostgreSQL, MySQL, and MariaDB, are supported by existing MOLTs, with extensive manual effort required to expand to other DBMSs. Given that many DBMSs remain inadequately tested, there is a pressing need for a method that enables effortless extension of MOLTs across diverse DBMSs. In this paper, we propose QTRAN, a novel LLM-powered approach that automatically extends existing MOLTs to various DBMSs. Our key insight is to translate SQL statement pairs to target DBMSs for metamorphic testing from existing MOLTs using LLMs. To address the challenges of LLMs’ limited understanding of dialect differences and metamorphic mechanisms, we propose a two-phase approach comprising the transfer and mutation phases. QTRAN tackles these challenges by drawing inspiration from the developer’s process of creating a MOLT, which includes understanding the grammar of the target DBMS to generate original queries and employing a mutator for customized mutations. The transfer phase is designed to identify potential dialects and leverage information from SQL documents to enhance query retrieval, enabling LLMs to translate original queries across different DBMSs accurately. During the mutation phase, we gather SQL statement pairs from existing MOLTs to fine-tune the pretrained model, tailoring it specifically for mutation tasks. Then we employ the customized LLM to mutate the translated original queries, preserving the defined relationships necessary for metamorphic testing. We implement our approach as a tool and apply it to extend four state-of-the-art MOLTs for eight DBMSs: MySQL, MariaDB, TiDB, PostgreSQL, SQLite, MonetDB, DuckDB, and ClickHouse. The evaluation results show that over 99% of the SQL statement pairs transferred by QTRAN satisfy the metamorphic relations required for testing. Furthermore, we have detected 24 logical bugs among these DBMSs, with 16 confirmed as unique and previously unknown bugs. We believe that the generality of QTRAN can significantly enhance the reliability of DBMSs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3728908",
+              "source_type": "paper_citation_context",
+              "excerpt": "…Metamorphic Testing in DBMSs. Recent years have witnessed tremendous efforts in resolving the test oracle for logical bug detection in the DBMSs. Notably, the metamorphic testing based approach MOLT has been recognized to be state-of-the-art in DBMS testing for logical bug detection [20, 38, 40].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:fe36bd14e5854298ad043c06550dfac3a851214224930005aa43f7e48f45f1ae",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3728908",
+              "source_type": "paper_citation_context",
+              "excerpt": "In our evaluation, we selected four state-of-the-art MOLTs for extension: NoRec [37], TLP [38], Pinolo [20], and DQE [44].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c6b1d0ec3ec343169160782cab138383f7433add1e8023fef4e4d18765ae32ce",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3728908",
+              "source_type": "paper_citation_context",
+              "excerpt": "Notably, metamorphic oracle, a widely-used method that constructs SQL statements maintaining either exact [25, 28, 29, 37–39, 44 ] or approximate To generate SQL statement pairs suitable for MOLTs, including original queries and mutated queries , a general idea is to model the SQL grammar as an…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c022c1364959c306c7b315932d50180dc215c0535632277ac21cb93db5fcdfdc",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3728908",
+              "source_type": "paper_citation_context",
+              "excerpt": "The evaluation results show that over 99% of the SQL statement pairs transferred by QTRAN satisfy the metamorphic relations required for testing, and have detected 24 logical bugs across several DBMSs, with 16 confirmed as unique and previously unknown.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2ab9ff8e3924d3970501f20a63afcbbe515e275eb32b372a908f58f4e8466378",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3728908",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, the complex code logic and diverse functionalities of DBMSs often make them susceptible to bugs, especially logical bugs that result in incorrect result sets being returned without obvious symptoms [8, 20, 24, 26, 37–39].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bd77ab9b3882cdb99fbcb04668039feb5fd0b9a37db9d0df057aaf5f557b8830",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3728908",
+              "source_type": "paper_citation_context",
+              "excerpt": "Below, we detail each of these tools: (1) NoRec [37]: This technique involves transferring predicates from the WHERE clause to the SELECT clause.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7a767f4959bd204c9a4f45ad700592487146afa90ea52010ed26ff2edf9effe6",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728908.json",
+              "source_type": "paper",
+              "excerpt": "In our evaluation, we selected four state-of-the-art MOLT s for extension: NoRec [ 37], TLP [38], Pinolo [20], and DQE [44].",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 4 Evaluation, page 11.",
+              "content_sha256": "sha256:1780c92661ba4a001e4a51c73b2155c2191c5d8f4e98ec412536eb04bce821c1",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728908.json",
+              "source_type": "paper",
+              "excerpt": "Below, we detail each of these tools: (1)NoRec [37]: This technique involves transferring predicates from the WHERE clause to the SELECT clause.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, 4 Evaluation, page 11.",
+              "content_sha256": "sha256:d24f6ac8050dc7e3320c5a1671747583fdef0945528d1f335a101b96e0f87af8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728908.json",
+              "source_type": "paper",
+              "excerpt": "(2)TLP [38]: This method decomposes a single query into three separate queries, each isolated by its predicates.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 4 Evaluation, page 11.",
+              "content_sha256": "sha256:003d3a4cdabe54fd9ee14af0865e43ce104f4bed0f2433e6fbc68bf7a5ad9b00",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "compares_with": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728908.json",
+              "source_type": "paper",
+              "excerpt": "Notably, the metamorphic testing based approach MOLT has been recognized to be state-of-the-art in DBMS testing for logical bug detection [ 20,38,40].",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 2 Background and Motivation, page 4.",
+              "content_sha256": "sha256:4a33155636f2e2b69f39f19d89c61142c0d0da94e9bb2d155ec2b3d1925dc43e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3728908.json",
+              "source_type": "paper",
+              "excerpt": "In our evaluation, we selected four state-of-the-art MOLT s for extension: NoRec [ 37], TLP [38], Pinolo [20], and DQE [44].",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 4 Evaluation, page 11.",
+              "content_sha256": "sha256:1780c92661ba4a001e4a51c73b2155c2191c5d8f4e98ec412536eb04bce821c1",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/196fb1deaa2eda163dcfc0a7e0a7dec93228fea3",
+        "source_type": "paper",
+        "content_sha256": "sha256:e7de64686f4685dfb555e8b792a7435cb153df317548af721a4ba6f34843c6ea"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icbds67396.2025.11377486",
+      "title": "Real-World Scalability of PostgreSQL: Practical Techniques Use Case Study",
+      "authors": [
+        "Jayraj Kiran Ladkat",
+        "Pranati Waghodekar"
+      ],
+      "year": 2025,
+      "venue": "2025 IEEE International Conference on Blockchain and Distributed Systems Security (ICBDS)",
+      "doi": "10.1109/icbds67396.2025.11377486",
+      "arxiv_id": null,
+      "s2_paper_id": "5596613e8f4806efa2defb955ec9fabb569465a5",
+      "url": "https://doi.org/10.1109/icbds67396.2025.11377486",
+      "open_access_pdf": null,
+      "abstract": "Modern applications face significant performance degradation and escalating operational costs when handling large data volumes and high concurrency. Inefficient database scaling often leads to resource over-provisioning and financial waste. This paper analyzes five PostgreSQL scaling techniques: Vertical Scaling, Read Replicas, Partitioning, Connection Pooling, and Custom Sharding. Our findings demonstrate that Connection Pooling can reduce latency by 30 % and increase throughput by 44%. This research provides a data-driven framework for architects to build cost-effective, high-performance PostgreSQL systems tailored to specific workload demands.",
+      "cites_seed_techniques": [
+        "sqlancer_pp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "sqlancer_pp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/5596613e8f4806efa2defb955ec9fabb569465a5",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced sqlancer_pp.",
+              "retrieved_at": "2026-09-10T14:52:29Z",
+              "first_seen": "2026-09-10T14:52:29Z",
+              "last_verified": "2026-09-10T14:52:29Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-10T14:52:29Z",
+        "last_verified": "2026-09-10T14:52:29Z",
+        "source_url": "https://www.semanticscholar.org/paper/5596613e8f4806efa2defb955ec9fabb569465a5",
+        "source_type": "paper",
+        "content_sha256": "sha256:1e6e833f9f73006d5a8d694a66f4bb8adea75e4fcd12621b8045b26328881567"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3785021.3787993",
+      "title": "Reproducibility Report for ACM SIGMOD 2025 Paper: 'Constant Optimization Driven Database System Testing'",
+      "authors": [
+        "Yuvaraj Chesetti",
+        "Chi Zhang"
+      ],
+      "year": 2025,
+      "venue": null,
+      "doi": "10.1145/3785021.3787993",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.1145/3785021.3787993",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "coddtest"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "coddtest"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/W7131085132",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Constant-Optimization-Driven Testing (CODDTest).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3785021_3787993.json",
+              "source_type": "paper",
+              "excerpt": "•The tool works by integrating their method with SQLancer [ 1], a tool to automatically test databases.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, 1 Introduction, page 1.",
+              "content_sha256": "sha256:025ab13922c2ae9c2723cebbc996ebc4476509cdf6de1b670b7164852e6efd54",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3785021_3787993.json",
+              "source_type": "paper",
+              "excerpt": "The SQLancer github repo also acknowledges the author’s work as a supported testing method.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 1 Introduction, page 1.",
+              "content_sha256": "sha256:1f4d8d5b2293278a4d8d5ce03669fcb18a130d806e5acf2b75e56968c766d744",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3785021_3787993.json",
+              "source_type": "paper",
+              "excerpt": "We downloaded the artifacts provided by the author that contain a version of SQLancer that supports their method.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, 4.2 Reproducibility Process, page 2.",
+              "content_sha256": "sha256:c01d5aafd9252e3272aa75e4e4bd03205e97f7d66ea3237724af9bd1df70bb33",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3785021_3787993.json",
+              "source_type": "paper",
+              "excerpt": "We ran the scripts for CODDTest (the author’s method), NoREC and Query Partitioning.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, 4.2 Reproducibility Process, page 2.",
+              "content_sha256": "sha256:002aa169efd4094cc92324d6ae85d6af6fb9e2951a1b527e5b986c24c265d77b",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3785021_3787993.json",
+              "source_type": "paper",
+              "excerpt": "•CODDTest reported 1297 bugs, NoRec reported 644 bugs, and query partitioning generated 8280 database logs.",
+              "excerpt_is_verbatim": true,
+              "note": "M13 in the paper's extracted text, 4.2 Reproducibility Process, page 2.",
+              "content_sha256": "sha256:2cb9932873321cc99338c0cfd557514c2f6b8b5e56f5dfcd8b977fc75479c392",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://openalex.org/W7131085132",
+        "source_type": "paper",
+        "content_sha256": "sha256:d613553c06ebdf9c1b48a7c92bf332a0083661ddef335d51e43bd42d78e3f7c1"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/iccste65902.2025.11138310",
+      "title": "Research on Coverage Statistics in Fuzz Testing of Closed-Source DBMS",
+      "authors": [
+        "Zhongjie Li",
+        "Hao-Tian Liang",
+        "Haoyang Jia",
+        "Qingxian Wang",
+        "Yan Cao"
+      ],
+      "year": 2025,
+      "venue": "2025 International Conference on Computer Science, Technology and Engineering (ICCSTE)",
+      "doi": "10.1109/iccste65902.2025.11138310",
+      "arxiv_id": null,
+      "s2_paper_id": "eac24dcac11bfbca2e3aa2b43a6c8c08c68f324a",
+      "url": "https://doi.org/10.1109/iccste65902.2025.11138310",
+      "open_access_pdf": null,
+      "abstract": "A Database Management System (DBMS) is widely used as application software for managing business data, and its security is of paramount importance. Any form of data leakage or corruption may lead to significant security issues. Currently, publicly available research on vulnerability detection for closed-source DBMSs remains relatively limited. To achieve effective testing of closed-source DBMSs, this paper proposes a novel coverage analysis method based on execution path monitoring to collect real-time coverage data of closed-source DBMSs. The method utilizes coverage feedback to guide seed scheduling in fuzz testing. Based on this approach, a coverage tracker named TrCov for closed-source DBMSs was implemented. TrCov was used to test two DBMSs, Oracle and SQL Server, and the experimental results validated the effectiveness of the tool.",
+      "cites_seed_techniques": [
+        "pqs",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/iccste65902.2025.11138310",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [4] and Troc [5] employ pre-established oracles to detect logical and transaction isolation errors in DBMSs. APOLLO [6] identifies performance regression issues by comparing different DBMS versions.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5ecf4972be19d70ee25f5dc819ab2a7528176b806299ea8c58a710b7728ff3e1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/eac24dcac11bfbca2e3aa2b43a6c8c08c68f324a",
+        "source_type": "paper",
+        "content_sha256": "sha256:f9d59a1d5bede948354352574027cf9ca54cd8c9741d47ec713f7c28aa9e5582"
+      }
+    },
+    {
+      "id": "paper:doi:10.1007/978-981-96-4506-0_18",
+      "title": "Review of Fuzz Testing Techniques for Database Management Systems",
+      "authors": [
+        "Yuheng Zhang",
+        "Hui Lu",
+        "Zhourui Zhang",
+        "Guo–Cheng Wu",
+        "Houlin Zhou",
+        "Zhenghao Li"
+      ],
+      "year": 2025,
+      "venue": "Communications in computer and information science",
+      "doi": "10.1007/978-981-96-4506-0_18",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.1007/978-981-96-4506-0_18",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/W4410099864",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://openalex.org/W4410099864",
+        "source_type": "paper",
+        "content_sha256": "sha256:76d8d0e8db473265d1d85024559dbb2f2f5ac69e650fe6fdbe8fa86cfde44885"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse55347.2025.00183",
+      "title": "ROSA: Finding Backdoors with Fuzzing",
+      "authors": [
+        "Dimitrios Kokkonis",
+        "M. Marcozzi",
+        "Emilien Decoux",
+        "Stefano Zacchiroli"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse55347.2025.00183",
+      "arxiv_id": "2505.08544",
+      "s2_paper_id": "e13b3c2397adc3c3047750da2263b0678cf048b6",
+      "url": "https://doi.org/10.1109/icse55347.2025.00183",
+      "open_access_pdf": "https://arxiv.org/pdf/2505.08544",
+      "abstract": "A code-level backdoor is a hidden access, programmed and concealed within the code of a program. For instance, hard-coded credentials planted in the code of a file server application would enable maliciously logging into all deployed instances of this application. Confirmed software supplychain attacks have led to the injection of backdoors into popular open-source projects, and backdoors have been discovered in various router firmware. Manual code auditing for backdoors is challenging and existing semi-automated approaches can handle only a limited scope of programs and backdoors, while requiring manual reverse-engineering of the audited (binary) program. Graybox fuzzing (automated semi-randomized testing) has grown in popularity due to its success in discovering vulnerabilities and hence stands as a strong candidate for improved backdoor detection. However, current fuzzing knowledge does not offer any means to detect the triggering of a backdoor at runtime. In this work we introduce ROSA, a novel approach (and tool) which combines a state-of-the-art fuzzer (AFL++) with a new metamorphic test oracle, capable of detecting runtime backdoor triggers. To facilitate the evaluation of ROSA, we have created ROSARUM, the first openly available benchmark for assessing the detection of various backdoors in diverse programs. Experimental evaluation shows that ROSA has a level of robustness, speed and automation similar to classical fuzzing. It finds all 17 authentic or synthetic backdooors from ROSARUM in 1 h 30 on average. Compared to existing detection tools, it can handle a diversity of backdoors and programs and it does not rely on manual reverse-engineering of the fuzzed binary code.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/e13b3c2397adc3c3047750da2263b0678cf048b6",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/e13b3c2397adc3c3047750da2263b0678cf048b6",
+        "source_type": "paper",
+        "content_sha256": "sha256:60a5a31b3bd1d00c63217110030726d569f78cf19f323066ebfdf64bc444570b"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3779212.3790215",
+      "title": "Scaling Automated Database System Testing",
+      "authors": [
+        "Suyang Zhong",
+        "Manuel Rigger"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Architectural Support for Programming Languages and Operating Systems",
+      "doi": "10.1145/3779212.3790215",
+      "arxiv_id": "2503.21424",
+      "s2_paper_id": "aa0342d64acbc65db8667dc2e960710d969b2e6f",
+      "url": "https://doi.org/10.1145/3779212.3790215",
+      "open_access_pdf": "https://doi.org/10.1145/3779212.3790215",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp",
+        "coddtest"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert",
+            "dqp",
+            "coddtest"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790215",
+              "source_type": "paper_citation_context",
+              "excerpt": "We adopted two state-of-the-art [10] test oracles, Ternary Logic Partitioning (TLP) [36] and Non-optimizing Reference Engine Construction (NoREC) [35], which can find logic bugs by comparing the results of two equivalent queries constructed through a syntactic transformation that applies to any…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:888588956615d5c2d746fe1640aada33f7add3835b519b97638505e958616289",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790215",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, SQLancer [2, 3, 35– 37], a state-of-the-art tool for DBMS testing, currently supports generators for 22 DBMSs, which, on average, are implemented in 3,729 LOC (see Figure 1), with some DBMS-specific components being contributed by major companies.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:844b8a00053d538ce442d7cc8aa43c880cf2001f0d476ae9e33ee3d9b561cfee",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790215",
+              "source_type": "paper_citation_context",
+              "excerpt": "The core component of SQLancer++ is an adaptive statement generator , which infers the supported SQL features of the target DBMS during execution and adaptively generates random SQL statements that can be processed by the DBMS under test.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:870ea38aa955bbaf42a9f8af698def41394634bd7de9d2f1bc6fb2535f245222",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790215",
+              "source_type": "paper_citation_context",
+              "excerpt": "To compare with other tools, we selected SQLite 3.45.2 as the baseline for measuring coverage and validity rate and selected PostgreSQL 14.11 for measuring validity rate.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:52925d46af3efbd6eb61ff9f539816eb8065a03b9e0a4f05f2c41f8571307339",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790215",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [35] detects logic bugs by comparing the results of a query that is receptive to optimizations with an equivalent one that the DBMS is likely to fail to optimize.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:132bc06045fd418538387b5f96248719ef18b474d9d45f181474fd2ee57f6c79",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790215",
+              "source_type": "paper_citation_context",
+              "excerpt": "Query Plan Guidance (QPG) [2] uses query plans as a feedback mechanism to determine whether a given database state has saturated for finding potential bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0eb8f3a7913268c1089da08350b9c69506948cd2443d95a8225d8d794ddd093f",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "deterministic",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790215",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, SQLancer [2, 3, 35– 37], a state-of-the-art tool for DBMS testing, currently supports generators for 22 DBMSs, which, on average, are implemented in 3,729 LOC (see Figure 1), with some DBMS-specific components being contributed by major companies.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Pivoted Query Synthesis (PQS) as state of the art.",
+              "content_sha256": "sha256:844b8a00053d538ce442d7cc8aa43c880cf2001f0d476ae9e33ee3d9b561cfee",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790215",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, SQLancer [2, 3, 35– 37], a state-of-the-art tool for DBMS testing, currently supports generators for 22 DBMSs, which, on average, are implemented in 3,729 LOC (see Figure 1), with some DBMS-specific components being contributed by major companies.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Non-optimizing Reference Engine Construction (NoREC) as state of the art.",
+              "content_sha256": "sha256:844b8a00053d538ce442d7cc8aa43c880cf2001f0d476ae9e33ee3d9b561cfee",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790215",
+              "source_type": "paper_citation_context",
+              "excerpt": "We adopted two state-of-the-art [10] test oracles, Ternary Logic Partitioning (TLP) [36] and Non-optimizing Reference Engine Construction (NoREC) [35], which can find logic bugs by comparing the results of two equivalent queries constructed through a syntactic transformation that applies to any…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Ternary Logic Partitioning (TLP) as state of the art.",
+              "content_sha256": "sha256:888588956615d5c2d746fe1640aada33f7add3835b519b97638505e958616289",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3779212.3790215",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, SQLancer [2, 3, 35– 37], a state-of-the-art tool for DBMS testing, currently supports generators for 22 DBMSs, which, on average, are implemented in 3,729 LOC (see Figure 1), with some DBMS-specific components being contributed by major companies.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Query Plan Guidance (QPG) as state of the art.",
+              "content_sha256": "sha256:844b8a00053d538ce442d7cc8aa43c880cf2001f0d476ae9e33ee3d9b561cfee",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/aa0342d64acbc65db8667dc2e960710d969b2e6f",
+        "source_type": "paper",
+        "content_sha256": "sha256:3da4dfdf1411011af98e89594f67d3c843116b46d935d6bfe313051d79b1b19b"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Recently, various automated testing approaches have been proposed that use specialized test oracles to find hundreds of logic bugs in mature, widely-used Database Management Systems (DBMSs). These test oracles require database and query generators, which must account for the often significant differences between the SQL dialects of these systems. Since it can take weeks to implement such generators, many DBMS developers are unlikely to invest the time to adopt such automated testing approaches. In short, existing approaches fail to scale to the plethora of DBMSs. In this work, we present both a vision and a platform, SQLancer++, to apply test oracles to any SQL-based DBMS that supports a subset of common SQL features. Our technical core contribution is a novel architecture for an adaptive SQL statement generator. This adaptive SQL generator generates SQL statements with various features, some of which might not be supported by the given DBMS, and then learns through interaction with the DBMS, which of these are understood by the DBMS. Thus, over time, the generator will generate mostly valid SQL statements. We evaluated SQLancer++ across 18 DBMSs and discovered a total of 196 unique, previously unknown bugs, of which 180 were fixed after we reported them. While SQLancer++ is the first major step towards scaling automated DBMS testing, various follow-up challenges remain."
+    },
+    {
+      "id": "paper:doi:10.1109/iaecst68792.2025.11415166",
+      "title": "Semantic Hint-Based Fuzzing for Time-Series Databases",
+      "authors": [
+        "Panta Kittisatra",
+        "Liang Liu"
+      ],
+      "year": 2025,
+      "venue": "2025 7th International Academic Exchange Conference on Science and Technology Innovation (IAECST)",
+      "doi": "10.1109/iaecst68792.2025.11415166",
+      "arxiv_id": null,
+      "s2_paper_id": "1e43fae09e6bcbd94078388ec04cd6cba311552c",
+      "url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+      "open_access_pdf": null,
+      "abstract": "Time-series databases (TSDBs) underpin modern IoT and industrial data pipelines, where reliability and semantic correctness are essential. However, most existing fuzzing frameworks are designed for relational DBMSs and overlook temporal semantics and query processing behaviors unique to TSDBs. To address this gap, we present a Semantic Hint-Based Oracle, a lightweight extension to TSGuard that evaluates TSDB robustness under semantic-preserving transformations. The oracle injects logically neutral predicates—such as tautologies and redundant conditions—into baseline queries and performs differential comparison to identify unexpected errors or behavioral divergences without altering query intent. Experiments on Apache IoTDB, InfluxDB, and TDengine uncovered multiple optimizer-related crashes and robustness issues triggered by harmless semantic transformations. The results demonstrate the feasibility of hint-based perturbation as an efficient path toward semantic validation of TSDB engines without schema introspection or complex metamorphic query synthesis.",
+      "cites_seed_techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, these frameworks largely assume relational semantics[6], static schemas, and Boolean logic evaluation, making them ineffective for dealing with temporal operators, continuous data ingestion, window functions, and time-aware aggregations that are fundamental in TSDBs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b9a211e596e8007862d15121248dfcddaaa1535f11947a15913e978aa20709a5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer pioneered systematic bug detection using oracles such as Non-Optimizing Reference Engine Construction (NoREC) and Ternary Logic Partitioning (TLP), enabling reliable detection of logic bugs even when databases do not crash [6].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:80aa0f4ac65e6755f944a26c5b3bb42c7cdcf047f194ea36a147664c2d5b6a08",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+              "source_type": "paper_citation_context",
+              "excerpt": "Tools such as SQLancer[6], SQLaser[7], and mutation-based engines like AFL leverage metamorphic testing to uncover logic bugs and optimizer inconsistencies[8], [9].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e060d03d7c16aacb0447c3b584fea9a9dec6e4afd11de5de26f7654a75450a99",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+              "source_type": "paper_citation_context",
+              "excerpt": "Research on testing relational DBMS, which supports a standardized query language SQL, has evolved considerably over the past decades[13], [14] .",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0ce89f61efedbe182149a0f0aded350cafef614b77ce1bd04ce4eeead8fbcf13",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_iaecst68792_2025_11415166.json",
+              "source_type": "paper",
+              "excerpt": "SQLancer pioneered systematic bug detection using oracles such as Non-Optimizing Reference Engine Construction (NoREC) and Ternary Logic Partitioning (TLP), enabling reliable detection of logic bugs even when databases do not crash [6].",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, II RELATED WORK, page 2.",
+              "content_sha256": "sha256:80aa0f4ac65e6755f944a26c5b3bb42c7cdcf047f194ea36a147664c2d5b6a08",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/1e43fae09e6bcbd94078388ec04cd6cba311552c",
+        "source_type": "paper",
+        "content_sha256": "sha256:dca06d6d8350e6c9d62f55c3552af02454fc6ae5c27659d7957ed5fec4b80b52"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icpc66645.2025.00021",
+      "title": "Sembug: Detecting Logic Bugs in Dbms Through Generating Semantic-Aware Non-Optimizing Query",
+      "authors": [
+        "Shiyang Ye",
+        "Chao Ni",
+        "Jue Wang",
+        "Qianqian Pang",
+        "Xinrui Li",
+        "Xiaodan Xu"
+      ],
+      "year": 2025,
+      "venue": "IEEE International Conference on Program Comprehension",
+      "doi": "10.1109/icpc66645.2025.00021",
+      "arxiv_id": null,
+      "s2_paper_id": "aeea391e73991ef6f033917dba0f577c948eedb4",
+      "url": "https://doi.org/10.1109/icpc66645.2025.00021",
+      "open_access_pdf": null,
+      "abstract": "Logic bugs, which cause Database Management Systems (DBMSs) to return incorrect results, are challenging to detect due to the absence of explicit signs such as system crashes. The majority of these bugs originate from the query optimizer and are commonly referred to as optimization bugs. Many approaches have been proposed for detecting logic bugs, which can be divided into two groups. The first group aims to detect the optimization bugs but only focuses on those with incorrect results cardinality, neglecting to check semantic correctness and consequently limiting the detection of bugs in advanced DBMS features. For the second group, though it can verify the correctness of the results for both their cardinality and semantics, it is ineffective in handling optimization bugs, which restricts its practical usage effectiveness. In this paper, we propose Semantic-aware Non-Optimizing Query (SemBug), a novel approach for logic bug detection in DBMSs. SemBug focuses on optimization bugs by transforming the queries that can be highly optimized by DBMS into equivalent but less optimized ones. Additionally, SemBug integrates semantic analysis technology, enabling it to identify semantic logic bugs and support testing advanced DBMS features. Any discrepancy in cardinality or content between the original and transformed queries indicates a logic bug. To investigate the effectiveness of SemBug, we conduct a large-scale experiment on five widelyused DBMS systems (i.e., MySQL, TiDB, MariaDB, SQLite, and PostgreSQL) and compare it with three state-of-the-art (SOTA) approaches (i.e., Pinolo, TLP, and NoREC). The experimental results indicate that SemBug outperforms three SOTAs. Over 24 hours, SemBug found 34 unique logic bugs, which are 19, 14, and 13 more bugs than each of the three SOTAs, marking an improvement of $126 \\%, 70 \\%$, and 61 % respectively. As of the time of paper submission, SemBug has uncovered 37 unique logic bugs, of which 29 have been verified by developers, and 11 have been fixed. SemBug helps developers identify these bugs, providing insights into such inconsistencies and assisting in resolving them.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/aeea391e73991ef6f033917dba0f577c948eedb4",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Query Plan Guidance (QPG), Cardinality Estimation Restriction Testing (CERT), Differential Query Plans (DQP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icpc66645_2025_00021.json",
+              "source_type": "paper",
+              "excerpt": "We use SQLancer [29], a rule-based method for query generation.",
+              "excerpt_is_verbatim": true,
+              "note": "M23 in the paper's extracted text, B Database and Query Generation, page 4.",
+              "content_sha256": "sha256:a1287079a0163048522af17ab123a8945f7cbeda2fab75c3da0f78184880943a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icpc66645_2025_00021.json",
+              "source_type": "paper",
+              "excerpt": "It is noted that although we only use SQLancer to generate queries, SemBug can be integrated with any generation technique.",
+              "excerpt_is_verbatim": true,
+              "note": "M25 in the paper's extracted text, B Database and Query Generation, page 4.",
+              "content_sha256": "sha256:3c750dc793a36818d751dce650a05e68f6663d0767c9d12ce247ef6fccdc5bbc",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icpc66645_2025_00021.json",
+              "source_type": "paper",
+              "excerpt": "Specifically, NoREC, TLP, and SemBug employ SQLancer [29] as the query generator, while Pinolo uses Go-Randgen [30].",
+              "excerpt_is_verbatim": true,
+              "note": "M31 in the paper's extracted text, A Experimental Setup, page 6.",
+              "content_sha256": "sha256:a7d621c9acfe95a11b3e8b456804f0a8b0b0ce8adad5c0fa4960ec12e6754066",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icpc66645_2025_00021.json",
+              "source_type": "paper",
+              "excerpt": "Inspired by NoREC, we transform the optimized query to the less optimized one of the form SELECT 𝛼, P is true AS flag FROM t through moving WHERE clause after the SELECT clause.",
+              "excerpt_is_verbatim": true,
+              "note": "M26 in the paper's extracted text, C Transform the Query and Retrieve Semantic Information, page 4.",
+              "content_sha256": "sha256:be95ba7591262dfce680916ffe79affe387e6831701c633950472e4969e046bc",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icpc66645_2025_00021.json",
+              "source_type": "paper",
+              "excerpt": "3)Baselines: We compare SemBug against the state-ofthe-art logic bug detection techniques, namely NoREC [18], TLP [10], and Pinolo [19], respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M30 in the paper's extracted text, A Experimental Setup, page 6.",
+              "content_sha256": "sha256:839b7380eb248d47fe25948e03fe84361f08615f4e65798b5a8214cfb09962ce",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icpc66645_2025_00021.json",
+              "source_type": "paper",
+              "excerpt": "Specifically, NoREC, TLP, and SemBug employ SQLancer [29] as the query generator, while Pinolo uses Go-Randgen [30].",
+              "excerpt_is_verbatim": true,
+              "note": "M31 in the paper's extracted text, A Experimental Setup, page 6.",
+              "content_sha256": "sha256:a7d621c9acfe95a11b3e8b456804f0a8b0b0ce8adad5c0fa4960ec12e6754066",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icpc66645_2025_00021.json",
+              "source_type": "paper",
+              "excerpt": "3)Baselines: We compare SemBug against the state-ofthe-art logic bug detection techniques, namely NoREC [18], TLP [10], and Pinolo [19], respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M30 in the paper's extracted text, A Experimental Setup, page 6.",
+              "content_sha256": "sha256:839b7380eb248d47fe25948e03fe84361f08615f4e65798b5a8214cfb09962ce",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/aeea391e73991ef6f033917dba0f577c948eedb4",
+        "source_type": "paper",
+        "content_sha256": "sha256:8bff995b052a80b72d6f5dc18f23197ce6e87d5ca0cffec83663e228c8274df1"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/Syang111/SemBug",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "renamed_sqlancer_package",
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/Syang111/SemBug",
+              "source_type": "github_repository",
+              "excerpt": "# SemBug",
+              "note": "Repository is named after Sembug, the tool this paper says it built, and is about database testing.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/Syang111/SemBug/blob/master/src/Sonar/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "src/Sonar/Randomly.java is SQLancer's Randomly.java, with the package renamed to Sonar (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:f9a076f1565f5cd0d81e2adbe85664ce724fb607f37eba6ecc56e1740ca173bc",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3758316.3763253",
+      "title": "Siloso: Finding Logic Bugs in RDBMS via Dialect-Adaptable Reference Engine Construction",
+      "authors": [
+        "Emily Ong"
+      ],
+      "year": 2025,
+      "venue": null,
+      "doi": "10.1145/3758316.3763253",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.1145/3758316.3763253",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/W4415047019",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-10T14:40:47Z",
+              "first_seen": "2026-09-10T14:40:47Z",
+              "last_verified": "2026-09-10T14:40:47Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-10T14:40:47Z",
+        "last_verified": "2026-09-10T14:40:47Z",
+        "source_url": "https://openalex.org/W4415047019",
+        "source_type": "paper",
+        "content_sha256": "sha256:b59ff4c04b6fa8117b36638d24911766183d41fb301e55f0228d2ff8c9232bc7"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3742728.3742747",
+      "title": "Simple Testing Can Expose Most Critical Transaction Bugs: Understanding and Detecting Write-Specific Serializability Violations in Database Systems",
+      "authors": [
+        "Ziyu Cui",
+        "Wensheng Dou",
+        "Yu Gao",
+        "Rui Yang",
+        "Yingying Zheng",
+        "Jiansen Song",
+        "Yuan Feng",
+        "Jun Wei"
+      ],
+      "year": 2025,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3742728.3742747",
+      "arxiv_id": null,
+      "s2_paper_id": "3f929c9c702538e09f4f88cf69415ceb25d3ba77",
+      "url": "https://doi.org/10.14778/3742728.3742747",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/3f929c9c702538e09f4f88cf69415ceb25d3ba77",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Query Plan Guidance (QPG), Differential Query Plans (DQP).",
+              "retrieved_at": "2026-09-06T17:02:28Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T17:02:28Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "deterministic",
+          "evidence": [
+            {
+              "source_url": "https://www.vldb.org/pvldb/vol18/p2547-cui.pdf",
+              "source_type": "paper",
+              "excerpt": "We implement WriteCheck based on SQLancer [ 14], and extend SQLancer to detect WSSviolations.",
+              "excerpt_is_verbatim": true,
+              "note": "The paper states that its implementation is built on SQLancer.",
+              "content_sha256": "sha256:d418c8c34e96318c59ffbea0d4707bde9b0bba44d15eef4b4371b53e9ebfefcd",
+              "retrieved_at": "2026-09-06T17:02:28Z",
+              "first_seen": "2026-09-06T17:02:28Z",
+              "last_verified": "2026-09-06T17:02:28Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "deterministic",
+          "evidence": [
+            {
+              "source_url": "https://www.vldb.org/pvldb/vol18/p2547-cui.pdf",
+              "source_type": "paper",
+              "excerpt": "We implement WriteCheck based on SQLancer [ 14], and extend SQLancer to detect WSSviolations.",
+              "excerpt_is_verbatim": true,
+              "note": "The paper states that it extends or adapts a SQLancer technique.",
+              "content_sha256": "sha256:d418c8c34e96318c59ffbea0d4707bde9b0bba44d15eef4b4371b53e9ebfefcd",
+              "retrieved_at": "2026-09-06T17:02:28Z",
+              "first_seen": "2026-09-06T17:02:28Z",
+              "last_verified": "2026-09-06T17:02:28Z"
+            },
+            {
+              "source_url": "https://www.vldb.org/pvldb/vol18/p2547-cui.pdf",
+              "source_type": "paper",
+              "excerpt": "We utilize SQLancer [ 58–60] to generate initial databases and individual SQL statements, and extend SQLancer to generate transactions and transaction test cases.",
+              "excerpt_is_verbatim": true,
+              "note": "The paper states that it extends or adapts a SQLancer technique.",
+              "content_sha256": "sha256:0c86148a1001a7a49271c78b821a76d08443ac0cab97f63c35d51e67553c3ee9",
+              "retrieved_at": "2026-09-06T17:02:28Z",
+              "first_seen": "2026-09-06T17:02:28Z",
+              "last_verified": "2026-09-06T17:02:28Z"
+            }
+          ],
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "dqp"
+          ]
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T17:02:28Z",
+        "source_url": "https://www.semanticscholar.org/paper/3f929c9c702538e09f4f88cf69415ceb25d3ba77",
+        "source_type": "paper",
+        "content_sha256": "sha256:ab1e210fe2510eb88b56e953525e6584f39f92170b9b93a9c57749b6d54448a4"
+      },
+      "abstract": "Database Management Systems (DBMSs) utilize transactions to guarantee data consistency and integrity. Incorrect implementations of transaction processing mechanisms can introduce critical transaction bugs, which can lead to incorrect database states after the involved transactions complete. However, we lack an effective test oracle to determine whether a DBMS produces a correct database state for a given concurrent transaction schedule.\n \n In this paper, we propose a general property for concurrent transaction schedules,\n write-specific serializability\n , in which a schedule of concurrent transactions should produce the same database state as a corresponding serial schedule of the same transactions. Through our empirical study on 35 critical transaction bugs collected from six widely-used DBMSs, we find that write-specific serializability can be an effective test oracle to expose critical transaction bugs in DBMSs. We further develop a simple and general transaction testing approach, WriteCheck, to automatically detect write-specific serializability violations by identifying inconsistencies in the final database states produced by the original transaction schedule and its corresponding serial schedule. We evaluate WriteCheck on the latest versions of six production-grade DBMSs, and have found 22 write-specific serializability violations, 11 of which have been confirmed as new critical transaction bugs."
+    },
+    {
+      "id": "paper:doi:10.1177/0926227x251370258",
+      "title": "SQLaser: Detecting database management system (DBMS) logic bugs with clause-guided fuzzing",
+      "authors": [
+        "Jin Wei",
+        "Ping Chen",
+        "Kangjie Lu",
+        "Jun Dai",
+        "Xiaoyan Sun"
+      ],
+      "year": 2025,
+      "venue": "Journal of computing and security",
+      "doi": "10.1177/0926227x251370258",
+      "arxiv_id": "2407.04294",
+      "s2_paper_id": "71f038fd6f135f0bc91c952de438cd717d675110",
+      "url": "https://doi.org/10.1177/0926227x251370258",
+      "open_access_pdf": null,
+      "abstract": "Database management systems (DBMSs) are vital components in modern data-driven systems. Their complexity often leads to logic bugs, which are implementation errors within the DBMSs that can lead to incorrect query results, data exposure, unauthorized access, etc., without necessarily causing visible system failures. Existing detection employs two strategies: rule-based bug detection and coverage-guided fuzzing. In general, rule specification itself is challenging; as a result, rule-based detection is limited to specific and simple rules. Coverage-guided fuzzing blindly explores code paths or blocks, many of which are unlikely to contain logic bugs; therefore, this strategy is cost-ineffective. In this paper, we design SQLaser, a SQL-clause-guided fuzzer for detecting logic bugs in DBMSs. Through a comprehensive examination of existing logic bugs across four distinct DBMSs, excluding those causing system crashes, we have identified 35 logic-bug patterns. These patterns manifest as certain SQL clause combinations that commonly result in logic bugs, and behind these clause combinations are a sequence of functions. We therefore model logic-bug patterns as error-prone function chains (i.e., sequences of functions). We further develop a directed fuzzer with a new path-to-path distance-calculation mechanism for effectively testing these chains and discovering additional logic bugs. This mechanism enables SQLaser to swiftly navigate to target sites and uncover potential bugs emerging from these paths. Our evaluation, conducted on SQLite, MySQL, PostgreSQL, and TiDB, demonstrates that SQLaser significantly accelerates bug discovery compared to other fuzzing approaches, reducing detection time by approximately 60%. As a standalone fuzzer, SQLaser identified 22 bugs spanning 18 of the 35 logic-bug patterns, outperforming contemporary fuzzers such as SQLRight, which only uncovered two logic bugs across two patterns within the same testing period (i.e., 60 days) when testing SQLite. Notably, four of the bugs discovered by SQLaser are zero-day, all of which have been reported to and confirmed by vendors.",
+      "cites_seed_techniques": [
+        "dqp",
+        "norec",
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1177/0926227x251370258",
+              "source_type": "paper_citation_context",
+              "excerpt": "This analysis covers nearly all logic bugs in research studies, 3–6 ensuring that the selection is comprehensive and not arbitrary.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:fdd2db9c1312b070e3ca853a4490580baf4f388e19aa0b8fb0371623a6af8a76",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1177/0926227x251370258",
+              "source_type": "paper_citation_context",
+              "excerpt": "3 However, since March 2023, new research studies 88–91 on DBMS logic bugs have uncovered additional bug patterns.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:eafbfde4b94aade43936aed92627f768c3fe46bc1fddfcf19a068f4cd12f55ec",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1177/0926227x251370258",
+              "source_type": "paper_citation_context",
+              "excerpt": "Additionally, new bug patterns have been discovered in studies 88–91 published after March 2023.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1c927ebbf08ccb43f3b97b3157366751c0d9f6a7516bc3b65c513135e2e59bc1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2407.04294",
+              "source_type": "paper_citation_context",
+              "excerpt": "It is worth noting that the Pivoted Query Synthesis (PQS) oracles [40] proposed by Manuel Rigger et al. does not use the differential testing method; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:095e9694ea34cd287e82fca6cc5980b45bdaf262919bff6334065e7d3c86352f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2407.04294",
+              "source_type": "paper_citation_context",
+              "excerpt": "These bugs, which are due to implementation errors in the DBMSs’ code and are particularly common, can cause a variety of serious issues, such as incorrect query results, exposure of sensitive data, unauthorized access, and data corruption [28, 38–40].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e02953de541050e52eb86593e67dfb65661f561519de5ab28f9c1454f82d1d62",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2407.04294",
+              "source_type": "paper_citation_context",
+              "excerpt": "Pivoted Query Synthesis (PQS) oracle [40] does not employ the concept of differential testing; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2943e15333e254d570b147042fc4c26c9e1d9bbb09d7da1e6fd6d5e4f6e0ea42",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2407.04294",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] and the Ternary Logic Partitioning (TLP) oracle [39] are two oracles employing the concept of differential testing [32].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:05fc32b70c2f7e8f97748aaa1e66a734c2febc38a54e8fc83ac6db2eb5e20af6",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2407.04294",
+              "source_type": "paper_citation_context",
+              "excerpt": "Currently, SQLaser only supports differential testing-based oracles and does not accommodate other types of oracles, such as PQS oracle [40], which detects logic bugs through non-differential testing.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:17fabc0829e09afe70902edde61b549b088025b4a382b6e59e63655a0629ee8d",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2407.04294",
+              "source_type": "paper_citation_context",
+              "excerpt": "Ternary Logic Partitioning (TLP) oracle [39] composes several sub-queries to collectively achieve the semantics of the original query.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f08c4a662c28c0b3996e6083c2b0a1d74d0eb6f4303534f7e5f743b7d9b3c89c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/71f038fd6f135f0bc91c952de438cd717d675110",
+        "source_type": "paper",
+        "content_sha256": "sha256:6be9d2b448fd913dab867d0ba7a9b39070a226dbb6836383d8483ad0d4001bb8"
+      },
+      "also_indexed_as": [
+        "paper:arxiv:2407.04294"
+      ]
+    },
+    {
+      "id": "paper:doi:10.1109/tse.2025.3574328",
+      "title": "SQLaw: Detecting Bugs in GPU Database Management Systems via Rule-Based Differential Execution",
+      "authors": [
+        "Jiaxin Hu",
+        "Rongxin Wu"
+      ],
+      "year": 2025,
+      "venue": "IEEE Transactions on Software Engineering",
+      "doi": "10.1109/tse.2025.3574328",
+      "arxiv_id": null,
+      "s2_paper_id": "4f46526dc9fbb444b85f054e44c84bfa107065fd",
+      "url": "https://doi.org/10.1109/tse.2025.3574328",
+      "open_access_pdf": null,
+      "abstract": "Database Management Systems (DBMSs) are essential for managing structured data. To meet the increasing performance requirements for complex, large-scale data management and analysis, GPU DBMSs have been introduced to enhance processing and query execution speeds. Despite the growing interest in GPU DBMSs and the inherent presence of bugs, there has been no systematic effort, to our knowledge, to detect bugs in GPU DBMSs. To this end, we design SQLaw, an innovative and comprehensive framework that combines offline rule learning with an online interpreter incorporating mutation for efficient and general GPU-related bug detection. The offline rule learning component automatically extracts differential execution rules, which are used to guide the synthesis of configuration and query statements for testing. The online interpreter with mutation ensures the generalization of these statements. We evaluated SQLaw on three major GPU DBMSs. Our extensive evaluations demonstrate that SQLaw outperforms current state-of-the-art approaches by up to 2.22<inline-formula><tex-math notation=\"LaTeX\">$\\times$</tex-math><alternatives><mml:math><mml:mo>×</mml:mo></mml:math><inline-graphic xlink:href=\"wu-ieq1-3574328.gif\"/></alternatives></inline-formula> in the number of bugs detected within 24 hours. Additionally, SQLaw detected 51 previously unknown GPU-related bugs, of which 37 have been confirmed or fixed by developers.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+              "source_type": "paper_citation_context",
+              "excerpt": "Inspired by the widely adopted differential testing approach for bug detection in non-GPU DBMSs [13], [14], [15], [16] and GPU-related software [17], [18], we borrow its core concept of comparing the results of identical test cases executed on different software systems to establish a test oracle…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7b8d7db618d7cc2164675e2468a4933dd4664315ccd1e6c5f3f4ed7669815820",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+              "source_type": "paper_citation_context",
+              "excerpt": "To detect correctness bugs, Grand [13] and RD2 [67] employ randomized differential testing for multiple graph DBMSs and RDF stores, respectively; THANOS [14] compares test cases on differential storage engines; and DQP [16] leverages differential query plans for simple yet effective bug detection.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6cd988be1ef00511c33d723a9a4e64184d6cd2c6554858aff8221f026b3c21b5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+              "source_type": "paper_citation_context",
+              "excerpt": "• NoREC [52] translates query statements into non-optimized queries in the optimizer and compares the results to identify logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5c5f926f236aaf9ccb06a26a592798e6f02559cf006efb9b63e465b656fd9381",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+              "source_type": "paper_citation_context",
+              "excerpt": "To uncover logic bugs in relational DBMSs, NoREC [52] verifies the equivalence between original queries and their non-optimized translations; PINOLO [62] employs the approximation relation; TQS [61] utilizes join optimization relations; and EET [64] constructs equivalent transformations through…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:909c7ceace38dbde69fe77b83e92409d4d4798198edf44fafc146a6ad3df5bc0",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+              "source_type": "paper_citation_context",
+              "excerpt": "• TLP [53] partitions an original query into three separated queries by decomposing its predicate, and triggers a logic bug when the union result of these separated queries mis-matches the original one.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bd61f8c24032d59b00386161970bd3764dc1a4152228754bba49d657ce313bd4",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/tse.2025.3574328",
+              "source_type": "paper_citation_context",
+              "excerpt": "Many studies [52], [53], [61], [62], [64], [65], [66] employ metamorphic testing to evaluate DBMSs. Metamorphic testing verifies SQL statement results against predefined rules to detect bugs [69].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b1a8478b55e9864cf1911d61a53b312ecd5600b263d3267ab7685a76967b68aa",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "uncertain",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3574328.json",
+              "source_type": "paper",
+              "excerpt": "To identify unique GPU-related bugs, we imitated the statement-level and syntax-based reduction in SQLancer [47] to automatically simplify bug reports.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, IV IMPLEMENTATION, page 10.",
+              "content_sha256": "sha256:d3696fe992193fef4a18f8542f1c9fa64ad9a625b629e65a535baa6a7663952c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3574328.json",
+              "source_type": "paper",
+              "excerpt": "We selected three representative state-ofthe-art approaches, including SQLsmith ,NoREC and TLP .",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, A Evaluation Step, page 10.",
+              "content_sha256": "sha256:b377535dff66b3a92d6cedf47a58ed5f3a524f99f2b568e4b6f7d1125fc2145f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3574328.json",
+              "source_type": "paper",
+              "excerpt": "•NoREC [52] translates query statements into nonoptimized queries in the optimizer and compares the results to identify logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, A Evaluation Step, page 10.",
+              "content_sha256": "sha256:1438a4f8b829d8cc01471f351b2ea6e50888e817f337ea2e2b49acd64de43fe1",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3574328.json",
+              "source_type": "paper",
+              "excerpt": "•TLP [53] partitions an original query into three separated queries by decomposing its predicate, and triggers a logic bug when the union result of these separated queries mis-matches the original one.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, A Evaluation Step, page 10.",
+              "content_sha256": "sha256:ebaddd7d9292b20cafc210c4cad5aa510a73d8ba646e85ce6ce01d5fedc09d02",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3574328.json",
+              "source_type": "paper",
+              "excerpt": "We selected three representative state-ofthe-art approaches, including SQLsmith ,NoREC and TLP .",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, A Evaluation Step, page 10.",
+              "content_sha256": "sha256:b377535dff66b3a92d6cedf47a58ed5f3a524f99f2b568e4b6f7d1125fc2145f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_tse_2025_3574328.json",
+              "source_type": "paper",
+              "excerpt": "I n particular, SQLsmith is one of the most effective approaches for detecting crash and error bugs, while both NoREC and TLP represent advanced approaches for detecting logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, A Evaluation Step, page 10.",
+              "content_sha256": "sha256:755ac15724a28ec2660e44bbd1b7e52e59964def781efc297006c1e9d15b772c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/4f46526dc9fbb444b85f054e44c84bfa107065fd",
+        "source_type": "paper",
+        "content_sha256": "sha256:ad45a962cadff21099702f75aa8bd03be938b1877189ba55945cd313b0fcc6f4"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3786699",
+      "title": "SQLBarber: A System Leveraging Large Language Models to Generate Customized and Realistic SQL Workloads",
+      "authors": [
+        "Jiale Lao",
+        "Immanuel Trummer"
+      ],
+      "year": 2025,
+      "venue": "Proceedings of the ACM on Management of Data",
+      "doi": "10.1145/3786699",
+      "arxiv_id": "2507.06192",
+      "s2_paper_id": "995979576a9a43c33d87711cc7ec1757a5f1a280",
+      "url": "https://doi.org/10.1145/3786699",
+      "open_access_pdf": "https://doi.org/10.1145/3786699",
+      "abstract": "Database research and development often require a large number of SQL queries for benchmarking purposes. However, acquiring real-world SQL queries is challenging due to privacy concerns, and existing generation methods offer limited options for customization and for satisfying realistic constraints. To address this issue, we present SQLBarber, a system based on Large Language Models (LLMs) to generate customized and realistic SQL workloads. SQLBarber (1) eliminates the need for users to manually craft SQL templates in advance, while providing the flexibility to accept natural language specifications to constrain SQL templates, (2) scales efficiently to generate large volumes of queries matching any user-defined cost distribution (e.g., cardinality and execution plan cost), and (3) uses execution statistics from production environments to extract SQL template specifications and query cost distributions that reflect real-world query characteristics. SQLBarber introduces (1) a declarative interface for users to effortlessly generate customized SQL templates, (2) an LLM-powered pipeline augmented with a self-correction module that profiles, refines, and prunes SQL templates based on query costs, and (3) a Bayesian Optimizer to efficiently explore predicate values and identify a set of queries that satisfy the target cost distribution. We construct and open-source ten benchmarks of varying difficulty levels and target query cost distributions based on real-world statistics from Snowflake and Amazon Redshift. Extensive experiments on these benchmarks show that SQLBarber is the only system that can generate customized SQL templates. It reduces query generation time by one to two orders of magnitude and significantly improves alignment with the target cost distribution, compared with existing methods.",
+      "cites_seed_techniques": [
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/995979576a9a43c33d87711cc7ec1757a5f1a280",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/995979576a9a43c33d87711cc7ec1757a5f1a280",
+        "source_type": "paper",
+        "content_sha256": "sha256:4aff76d49e04f91b2956f0dc334fae67dc7aa8c3b1ae3ff4410e8b587774fc4b"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3749646.3749683",
+      "title": "SQLStorm: Taking Database Benchmarking into the LLM Era",
+      "authors": [
+        "Tobias Schmidt",
+        "Viktor Leis",
+        "Peter A. Boncz",
+        "Thomas Neumann"
+      ],
+      "year": 2025,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3749646.3749683",
+      "arxiv_id": null,
+      "s2_paper_id": "45e14bb4f951dc83b31b91e0156a0a9de66d70c2",
+      "url": "https://doi.org/10.14778/3749646.3749683",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3749646.3749683",
+              "source_type": "paper_citation_context",
+              "excerpt": "A final relevant field is database testing and database fuzzing [44].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a4c8b431c70f34248a1e049f32d41fbcfe985be2905392e006555bade60a2a2c",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/45e14bb4f951dc83b31b91e0156a0a9de66d70c2",
+        "source_type": "paper",
+        "content_sha256": "sha256:026a1a319ea93ab639d17d349d7249a23107bac24c734c0e515daddf948aa621"
+      },
+      "abstract": "In this paper, we introduce a new methodology for constructing database benchmarks using Large Language Models (LLMs), as well as SQLStorm v1.0, a concrete benchmark on a real-world dataset of three sizes (1 GB, 12 GB, 220 GB) consisting of over 18 K queries. This methodology of using AI to generate query workloads breaks new ground, not only in its ability to cheaply ($15) generate huge volumes (22 MB) of realistic queries but especially because it greatly expands the amount of SQL functionality and query constructions that is covered, compared to human-written SQL benchmarks such as TPC-H, TPC-DS, and JOB. The use cases of SQLStorm that we think will advance data systems most are: (i) improving SQL compatibility between systems, (ii) increasing system quality by identifying crashes/errors and fixing those, (iii) improving cardinality estimators and query optimizers, by identifying trends and opportunities (queries where other systems do much better), as well as (iv) overall system performance, both in terms of speed and robustness."
+    },
+    {
+      "id": "paper:doi:10.1145/3769828",
+      "title": "SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis",
+      "authors": [
+        "Jinhui Lai",
+        "Chi Zhang",
+        "Bingyan Li",
+        "Chenglin Liang",
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Jingzhou Fu",
+        "Yu Jiang",
+        "Zichen Xu"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Manag. Data",
+      "doi": "10.1145/3769828",
+      "arxiv_id": null,
+      "s2_paper_id": "06d55988c1b284df5962886057f46df47fec17d7",
+      "url": "https://doi.org/10.1145/3769828",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3769828?download=true",
+      "abstract": "Logic bugs can cause DBMSs to silently produce incorrect results for a given query, posing significant threats to software reliability and remaining challenging to detect. Join is a fundamental operation in DBMSs, enabling the combination of data from multiple tables; however, due to its complexity, it is also susceptible to logic bugs. Existing works detect logic bugs in join optimizations by altering query hints and system variables to alter the optimizer's choice of execution plans. However, these approaches struggle to detect logic bugs when query hints or system variables fail to influence the optimizer's behavior, or when the logic bugs reside in join implementation code that is unrelated to optimization. In this paper, we present Set Relation Synthesis (SRS), a black-box testing approach that detects logic bugs of join implementation in DBMSs by leveraging set relations among different join operations. SRS applies transformations to the original join queries, including modifications to join types, join orders, and join conditions, while ensuring that the outputs of both the original and transformed queries preserve the expected set relations. Violations of these set relations indicate potential logic bugs. We realized SRS and evaluated it on five widely-used and extensively-tested DBMSs: MySQL, MariaDB, TiDB, PostgreSQL, and DuckDB. SRS uncovered 33 previously unknown and unique bugs, all of which have been confirmed, with 12 already fixed. Among these, 33 are logic bugs, demonstrating SRS's effectiveness and practicality in detecting logic bugs in the implementation of join operations within DBMSs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "cert",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3769828",
+              "source_type": "paper_citation_context",
+              "excerpt": "We implemented SRS on top of SQLancer 1 , a DBMS testing framework designed for the random generation of database states and SQL queries, which also supports multiple test oracles [23, 24].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9fe517551b5fa90c3b3e30385f2860f20b247f498a6f2b3255fd321c2a905dd8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3769828",
+              "source_type": "paper_citation_context",
+              "excerpt": "We selected six state-of-the-art approaches for comparison: PQS [25], TLP [24], NoREC [23], Pinolo [11], EET [13], and DQP [3].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:371fcbf9df0422f545a901f614d23a8aa9a96dc90f86d9028b945043bb71caa5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3769828",
+              "source_type": "paper_citation_context",
+              "excerpt": "TQS [28] and DQP [3] are two state-of-the-art testing approaches that target logic bugs in the optimization of join operations.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:455603a23f540eb1e17960995b4759968abb765a96caf470695a7770be338937",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3769828",
+              "source_type": "paper_citation_context",
+              "excerpt": "And this is a widely used method for comparing different test oracles [23].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7346d52668569dfcd07406fd40e83be4ffe8361176ad6b52291b5fc1f9e2444e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3769828",
+              "source_type": "paper_citation_context",
+              "excerpt": "Non-optimizing Reference Engine (NoREC) [23] assumes that a predicate should evaluate to the same value in both the WHERE and SELECT clauses, and leverages this assumption to construct a non-optimizable form of the query.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:aaa46e61759cce13c3ca846ada3418773fc53d854dcbcfe496a9d5c52ec4df84",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3769828",
+              "source_type": "paper_citation_context",
+              "excerpt": "This aligns with observations from previous works [3, 24], which limited our reporting of additional potential bugs to avoid duplication.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b6c0a1eeb55938549101ad7c4c13acb2bbdddff6838d2a099182f7ff87de82a7",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769828.json",
+              "source_type": "paper",
+              "excerpt": "SRS applies the automated, syntax-rule-based random generation approach of SQLancer to ensure that the database state exhibits sufficient diversity, thereby enabling thorough and effective bug detection.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 4 Implementation, page 11.",
+              "content_sha256": "sha256:cb61ac8f79479b966145e9721592db1fd03a79d65622691129a93e762d8a1878",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769828.json",
+              "source_type": "paper",
+              "excerpt": "For the generation of other language features, aside from considering the tables involved in joins, our approach remains consistent with SQLancer’s query generation strategy for NoREC test oracle [23].",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 4 Implementation, page 12.",
+              "content_sha256": "sha256:67104fc13c2d6c71575b7633df4b76d20d472f637309c9fedc4d512ccf36de9a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769828.json",
+              "source_type": "paper",
+              "excerpt": "We compared the code coverage achieved by these approaches, and the results showed that the test oracles implemented on SQLancer— PQS, NoREC, TLP, DQP, and SRS—achieved similar code coverage.",
+              "excerpt_is_verbatim": true,
+              "note": "M23 in the paper's extracted text, 5.4 Comparison With the State-of-the-Art, page 18.",
+              "content_sha256": "sha256:2123349d6bafff7f80144e10ad3af201d68755bc6c48e87e43ec6c70c38ad95c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769828.json",
+              "source_type": "paper",
+              "excerpt": "We selected six state-of-the-art approaches for comparison: PQS [ 25], TLP [ 24], NoREC [ 23], Pinolo [ 11], EET [ 13], and DQP [ 3].",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, 5.4 Comparison With the State-of-the-Art, page 17.",
+              "content_sha256": "sha256:87deda2c56bc1a6b8fe7be28fe1533e45e8aa76d9e6663a0cb8992f3f1ba35ae",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769828.json",
+              "source_type": "paper",
+              "excerpt": "As such, we include a comparison with NoREC in our evaluation.",
+              "excerpt_is_verbatim": true,
+              "note": "M20 in the paper's extracted text, 5.4 Comparison With the State-of-the-Art, page 17.",
+              "content_sha256": "sha256:b79b15c3e1ec0716000e7e429cf0ac2c4f2324832ccd0461e3c08bbb3472ac1e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769828.json",
+              "source_type": "paper",
+              "excerpt": "Table 5 shows that SRS found 12, 9, 10, 20, 17, and 13 more bugs than PQS, NoREC, TLP, DQP, Pinolo, and EET, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M21 in the paper's extracted text, 5.4 Comparison With the State-of-the-Art, page 17.",
+              "content_sha256": "sha256:1954f191ffd3a6113821d9b96da233d0f18acc878f085d1d2cdfbd68edd81db2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769828.json",
+              "source_type": "paper",
+              "excerpt": "For instance, when testing MySQL for 24 hours, PQS, TLP, DQP, and SRS achieved line coverage rates of 23.",
+              "excerpt_is_verbatim": true,
+              "note": "M24 in the paper's extracted text, 5.4 Comparison With the State-of-the-Art, page 18.",
+              "content_sha256": "sha256:3a9a2de42c7974ee81f9d008e158ff5238519e30699490886a36170ea9b6d372",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769828.json",
+              "source_type": "paper",
+              "excerpt": "TQS [ 28] and DQP [ 3] are two state-of-the-art testing approaches that target logic bugs in the optimization of join operations.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, 1 Introduction, page 2.",
+              "content_sha256": "sha256:39cef63e4e6b464d7f7e17fa23e6029e582e0b5c4d4d190bc6528bed813ead23",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3769828.json",
+              "source_type": "paper",
+              "excerpt": "We selected six state-of-the-art approaches for comparison: PQS [ 25], TLP [ 24], NoREC [ 23], Pinolo [ 11], EET [ 13], and DQP [ 3].",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, 5.4 Comparison With the State-of-the-Art, page 17.",
+              "content_sha256": "sha256:87deda2c56bc1a6b8fe7be28fe1533e45e8aa76d9e6663a0cb8992f3f1ba35ae",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/06d55988c1b284df5962886057f46df47fec17d7",
+        "source_type": "paper",
+        "content_sha256": "sha256:ffa43609b65f0469d73c1665d04c087219fc77e22790f173ebbf21fd8b55a72f"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3729175",
+      "title": "Systems Correctness Practices at Amazon Web Services",
+      "authors": [
+        "Marc Brooker",
+        "A. Desai"
+      ],
+      "year": 2025,
+      "venue": "Communications of the ACM",
+      "doi": "10.1145/3729175",
+      "arxiv_id": null,
+      "s2_paper_id": "f9fb97518fae800c5bd88b573e3211307cd61b5d",
+      "url": "https://doi.org/10.1145/3729175",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3729175",
+      "abstract": "Leveraging formal and semi-formal methods.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3729175",
+              "source_type": "paper_citation_context",
+              "excerpt": "Large volumes of random SQL schemas, datasets, and queries are synthesized and run through the engines under test, and the results compared with an oracle based on the non-sharded version of the engine (as well as other approaches to validation, like those pioneered by SQLancer 23 ).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8a8cad3f062195c0bed07ea971aa7a2583c85309ccc592f24b1db6f0a6ab3c10",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://openalex.org/W4407131558",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/f9fb97518fae800c5bd88b573e3211307cd61b5d",
+        "source_type": "paper",
+        "content_sha256": "sha256:9ef3a85005455d9f039d7e3568bf42762a03f004fcc7c9dbddabd5f54ff1f916"
+      },
+      "also_indexed_as": [
+        "paper:doi:10.1145/3712057"
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3769832",
+      "title": "Test Data Generation for Complex SQL Queries",
+      "authors": [
+        "Sunanda Somwase",
+        "Parismita Das",
+        "S. Sudarshan"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Manag. Data",
+      "doi": "10.1145/3769832",
+      "arxiv_id": null,
+      "s2_paper_id": "d05ca28c4eaaf47e4757524b7638551ba0881cd6",
+      "url": "https://doi.org/10.1145/3769832",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3769832",
+      "abstract": "Generation of sample data for testing SQL queries has been an important task for many years, with applications such as testing of SQL queries used for data analytics and in application software, as well as grading of student SQL queries. More recently, with the increasing use of text-to-SQL systems, test data is key for the validation of generated queries. Earlier work on test data generation handled basic single-block SQL queries, as well as single-level nested SQL queries, but could not handle more complex queries. In this paper, we present a novel architecture and associated techniques for test generation that are designed to handle complex queries. We show our approach significantly outperforms the prior work on test data generation in the handling of complex queries. We also show that our approach outperforms the state-of-the-art for the more restricted problem of showing non-equivalence of query pairs.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3769832",
+              "source_type": "paper_citation_context",
+              "excerpt": "Another potential area for future work is to explore the use of data generation to improve the detection of bugs in database system implementations, following the approach of Rigger et al. [19, 20].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:57297b4d870592640032aa69f867a289a2d006d464458eaec828fa16dcbfb672",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/d05ca28c4eaaf47e4757524b7638551ba0881cd6",
+        "source_type": "paper",
+        "content_sha256": "sha256:59612844e3599e495ad19bb8138615f3fc78c86abeecea9e6439fa5f4dab163f"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3749186",
+      "title": "Testing Graph Databases with Synthesized Queries",
+      "authors": [
+        "Zijing Yin",
+        "Si Liu",
+        "David A. Basin"
+      ],
+      "year": 2025,
+      "venue": "Proc. ACM Manag. Data",
+      "doi": "10.1145/3749186",
+      "arxiv_id": null,
+      "s2_paper_id": "00940ec18283156770edb798589bf89c07d7544c",
+      "url": "https://doi.org/10.1145/3749186",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3749186",
+      "abstract": "Graph databases (GDBs) are increasingly used in many applications. However, their advanced features make them prone to logic bugs. Despite advances in GDB testing, a common limitation of current approaches is the lack of ground truth for their test oracles. This results in both incorrectly identified bugs and overlooked bugs. We introduce GQS (Graph Query Synthesis), the first automated testing approach for detecting logic bugs in graph databases (GDBs) based on an established ground truth. GQS starts by randomly generating a graph and selecting a set of properties associated with its elements, whose key-value pairs form the expected result set serving as the ground truth. It then synthesizes a query that should retrieve these values from the graph. When the query is executed on the graph by the GDB under test, any discrepancy between the actual result set and the ground truth indicates a logic bug. To extensively test a GDB, we develop novel techniques that synthesize both syntactically and semantically complex queries. We implement GQS in a tool that incorporates the first Cypher query synthesizer specifically designed for testing GDBs. Our tool finds 36 previously unknown bugs across four production GDBs, of which 26 are logic bugs, with some remaining undetected for up to five years. Additionally, our tool demonstrates superior effectiveness in bug detection compared to the state-of-the-art testers.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3749186",
+              "source_type": "paper_citation_context",
+              "excerpt": "…as n’.id = 4 , are sufficient to filter out undesired graph elements, more complex expressions help to extensively test GDBs. Previous work [16, 47] provides a good basis for generating complex expressions in general, like char_length(‘abc’)+sqrt(round(1.2)) , that satisfy specific value…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:36fdf6e94a4d1f4c5e628a25aeead1f4be130629f75bb141992f0709d819943e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3749186",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, key constructs in GDBs, such as paths, neighborhoods, and recursive traversals, do not map directly to the relational structures and operators in RDBs. Similar to our approach, PQS [47], Pinolo [15], and TQS [54] test RDBs by synthesizing SQL queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6b60f8793ccdec003188eb85770906196af684458c51e1bb7d37e2c308cd995f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3749186",
+              "source_type": "paper_citation_context",
+              "excerpt": "In line with the growing trend of using randomized testing to effectively uncover system-level bugs in general [23, 31] and relational databases in particular [20, 47, 54], these tools all generate random graph data and queries to exercise the GDB under test.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b7252fbef568399025ec3d2be5936fbc477d38ebd56ad5b17af6dbfa3a7963c8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3749186",
+              "source_type": "paper_citation_context",
+              "excerpt": "The challenge lies in the current lack of ground truth on the expected execution time, where the cardinality estimation approach [45] designed for RDBs may provide a direction for future research.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2e0c126f10f85bc3a741c3f8d73d38a7ebc8c182353ea53948b4df8a8c0d2a32",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3749186",
+              "source_type": "paper_citation_context",
+              "excerpt": "GDBMeter illustrates this by adopting the metamorphic oracles from TLP [46], originally designed for SQL, which partitions queries using three-valued logic.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:663871920938791ea3c2d4168640f0fa58fed2de52fe0b7468d1fec01d7d8a90",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3749186",
+              "source_type": "paper_citation_context",
+              "excerpt": "Such logic bugs are widely recognized as being more challenging to detect than database crashes or exceptions [19, 47, 54, 62].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:922e8b2a703cb1e34b3f5ce1e26ac67616327d21f057ef368fa43139ea152871",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3749186.json",
+              "source_type": "paper",
+              "excerpt": "•The metamorphic bug detectors, GDBMeter [ 22], Gamera [ 62], GQT [ 19], and GRev [ 33], each applying specifically designed query rewrite rules to identify discrepancies between the result of the original query and that of the rewritten query.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 5.4 Comparison with State-of-the-Art, page 18.",
+              "content_sha256": "sha256:fa58fcb5e6b98c22f14fef5944fbc8dd7c8e0aaabaceae4f28d4dacbaa205da5",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3749186.json",
+              "source_type": "paper",
+              "excerpt": "Upon applying GDBMeter’s ternary logic partitioning [ 22] to the test query, all three partitioned queries (lines 5–13) also produce empty results.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, 5.4 Comparison with State-of-the-Art, page 20.",
+              "content_sha256": "sha256:7a8f0263591fdff45635359c9137e87985440dee88d7cdffb3d202a31ed8244a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/00940ec18283156770edb798589bf89c07d7544c",
+        "source_type": "paper",
+        "content_sha256": "sha256:b52117edb8df6418f4134778d86cde66afb2859a815ca77e106783cdf5be675c"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/girst67753.2025.11382165",
+      "title": "Testing Spatial Database Engines via Spatial Equivalent Transformation",
+      "authors": [
+        "Shijie Li",
+        "Dongping Wang",
+        "Liang Liu",
+        "Keyue Yang",
+        "Lingwei Kuang"
+      ],
+      "year": 2025,
+      "venue": "2025 4th International Conference on Geographic Information and Remote Sensing Technology (GIRST)",
+      "doi": "10.1109/girst67753.2025.11382165",
+      "arxiv_id": null,
+      "s2_paper_id": "df4018eafe690fef9baef1ce90aa5a3eb3b0bfee",
+      "url": "https://doi.org/10.1109/girst67753.2025.11382165",
+      "open_access_pdf": null,
+      "abstract": "Spatial Database Management Systems (SDBMSs) utilize spatial data models for data storage. They efficiently handle spatial queries, such as KNN and spatial range queries. SDBMSs also perform topological analysis. They are widely applied in Geographic Information Systems (GIS) and navigation systems. SDBMSs are foundational software, similar to Relational Database Management Systems (RDBMSs). Logic bugs in SDBMSs lead to unexpected results. This causes functional defects or unavailability in upper-layer applications. Therefore, the automated detection of SDBMS logic bugs is an urgent issue that requires resolution.We propose a testing technique called Spatial Equivalent Transformation (SET). SET constructs the test oracle for SDBMS logic bugs. The technique generates equivalent variants of a seed query to test diverse and complex spatial operations by checking the consistency of their result sets. Simultaneously, we introduce a Geometric Generator utilizing geometric transformations. This generator efficiently creates representative spatial query test cases. We implemented this method as a tool named Spader, and then comprehensively evaluated its effectiveness on three widely used SDBMSs: PostGIS, MySQL, and MariaDB. The experimental results show that the method detected a total of 16 logic bugs. The developers have confirmed 15 of these bugs and 4 have already been fixed.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/girst67753.2025.11382165",
+              "source_type": "paper_citation_context",
+              "excerpt": "Although crash bug detection [2] [3] and logic bug detection [4] [5] [6] [7] [8] are both mature for RDBMSs, SDBMSs suffer from a relative paucity of studies.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:14285a2bfebc60b3e73436d4f10548879e84534ecafbcccd727a0e2b018e543b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/df4018eafe690fef9baef1ce90aa5a3eb3b0bfee",
+        "source_type": "paper",
+        "content_sha256": "sha256:62a43833bdd35a6d5bec6a6c7c1378b0566213760c15cca2716d8eabc7f18045"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse55347.2025.00257",
+      "title": "Thanos: DBMS Bug Detection via Storage Engine Rotation Based Differential Testing",
+      "authors": [
+        "Ying Fu",
+        "Zhiyong Wu",
+        "Yuanliang Zhang",
+        "Jie Liang",
+        "Jingzhou Fu",
+        "Yu Jiang",
+        "Shanshan Li",
+        "Xiangke Liao"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse55347.2025.00257",
+      "arxiv_id": null,
+      "s2_paper_id": "9f1911b932dfb22d05e25106a0da9e4e1febd296",
+      "url": "https://doi.org/10.1109/icse55347.2025.00257",
+      "open_access_pdf": null,
+      "abstract": "Differential testing is a prevalent strategy for establishing test oracles in automated DBMS testing. However, meticulously selecting equivalent DBMSs with diverse implementations and compatible input syntax requires huge manual efforts. In this paper, we propose Thanos, a framework that finds DBMS bugs via storage engine rotation based differential testing. Our key insight is that a DBMS with different storage engines must provide consistent basic storage functionalities. Therefore, it's feasible to construct equivalent DBMSs based on storage engine rotation, ensuring that the same SQL test cases to these equivalent DBMSs yield consistent results. The framework involves four main steps: 1) select the appropriate storage engines; 2) extract equivalence information among the selected storage engines; 3) synthesize feature-orient test cases that ensure the DBMS equivalence; and 4) send test cases to the DBMSs with selected storage engines and compare the results. We evaluate Thanos on three widely used and extensively tested DBMSs, namely MySQL, MariaDB, and Percona against state-of-the-art fuzzers SQLancer, SQLsmith, and SQUIRREL. Thanos outperforms them on branch coverage by 24%-116%, and also finds many bugs missed by other fuzzers. More importantly, the vendors have confirmed 32 previously unknown bugs found by Thanos, with 29 verified as Critical.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00257",
+              "source_type": "paper_citation_context",
+              "excerpt": "These outcomes align with our expectations, as the test case synthesis algorithm was specifically designed to trigger a broader range of storage engine features and behaviors within the DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:91f0d608ff9453916fd8cb9e2dbb338bb3efb33ef3b4eb278cf5298653958a82",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00257",
+              "source_type": "paper_citation_context",
+              "excerpt": "To improve the effectiveness of one specific storage engine, one could also gather other supported features from the DBMS’s official documentation and add them to the feature list.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3819463951f5ed9d65e7536078ad1463c43414d0e8479ccebe3ea99d9914cb12",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00257",
+              "source_type": "paper_citation_context",
+              "excerpt": "Metamorphic testing in DBMS involves transforming SQL queries and verifying if the resulting output changes align with expected behavior [28, 29, 30, 31].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:33d5ac91be8471295d479c25cbb7320700877c7a740e12bcb09a990c52aaa419",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00257",
+              "source_type": "paper_citation_context",
+              "excerpt": "DBMS fuzzers [22, 25, 33, 36, 37, 38, 39, 40, 41, 42], automate this process, focusing on creating complex SQL queries to uncover memory safety issues.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5640a0a857ed9601d6c14d62a3f1841db25c526260785fc662e211293e263fb7",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse55347.2025.00257",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer proposes constructing functionally equivalent queries to test one DBMS [30, 31, 33].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d919482f236035b7b316271360fb0c8c828654617950486d5987f370c6865005",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00257.json",
+              "source_type": "paper",
+              "excerpt": "Comparison with Existing Techniques To assess the effectiveness of THANOS, we conducted a comparative study that pitted THANOS against contemporary state-of-the-art DBMS testing methods, namely SQLancer, SQLsmith, and SQUIRREL.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, C Comparison with Existing Techniques, page 8.",
+              "content_sha256": "sha256:d747542a07974b4d7c8527c3153a7c07b3ac21df48a5fa7fa29aec09b32fb567",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00257.json",
+              "source_type": "paper",
+              "excerpt": "DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 0 1 1 6 MariaDB 0 0 1 5 Percona 0 0 1 3 Total 0 1 3 14 Increment 14↑ 13↑ 11↑ – TABLE IV: Number of branches covered by THANOS, SQLancer, SQLsmith, and SQUIRREL on 3 DBMSs in 24 hours.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, C Comparison with Existing Techniques, page 8.",
+              "content_sha256": "sha256:7faae0e20cb37754f60e3d6633be10d36e585af9e3dd8c8124522b9501ede502",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00257.json",
+              "source_type": "paper",
+              "excerpt": "DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 59,242 93,742 109,323 120,156 MariaDB 60,293 88,923 100,920 132,532 Percona 63,829 89,987 109,823 143,293 Total 183,364 272,652 320,066 395,981 Increment 115.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, C Comparison with Existing Techniques, page 8.",
+              "content_sha256": "sha256:2c4a0a210f4f7bbec09d39bb45460d836c6d2562841cde8e3a101533af738c68",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00257.json",
+              "source_type": "paper",
+              "excerpt": "Specifically, THANOS found 14, 13, and 11 more bugs than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, C Comparison with Existing Techniques, page 9.",
+              "content_sha256": "sha256:bf9bb8365476f314102ab20f9b74a02b5c54c64f8d03eb39a2f4c269c0dfe956",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00257.json",
+              "source_type": "paper",
+              "excerpt": "We evaluate THANOS on three widely used and extensively tested DBMSs, namely MySQL, MariaDB, and Percona against state-of-the-art fuzzers SQLancer, SQLsmith, and SQUIRREL.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, page 1.",
+              "content_sha256": "sha256:8d607c3cf9ccad99c62195c479ce39d5d2be3770c7bfc34629f0706373d25a19",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00257.json",
+              "source_type": "paper",
+              "excerpt": "To assess the effectiveness of THANOS, we compare THANOS against contemporary state-of-the-art DBMS testing methods, namely SQLancer, SQLsmith, and SQUIRREL.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, I INTRODUCTION, page 2.",
+              "content_sha256": "sha256:e96f569b490d9c4bffa8ebb619cadee67ac9b6a5e941a196145db46fada30689",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse55347_2025_00257.json",
+              "source_type": "paper",
+              "excerpt": "To evaluate the effectiveness of THANOS, we compared THANOS with the state-of-art DBMS test tools, SQUIRREL [22, 23], SQLancer [24], and SQLsmith [25, 26].",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, A Evaluation Setup, page 7.",
+              "content_sha256": "sha256:4324f3a6a753cf29026161f9d690cd7f8a9bdb2b3b21755757c2f14869fe938e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/9f1911b932dfb22d05e25106a0da9e4e1febd296",
+        "source_type": "paper",
+        "content_sha256": "sha256:4be256a2e44d6a8b4692722ae02a3f9a5a0f335a749f33d3da0fcd22b8230eed"
+      }
+    },
+    {
+      "id": "paper:arxiv:2506.02617",
+      "title": "Toward Understanding Bugs in Vector Database Management Systems",
+      "authors": [
+        "Yinglin Xie",
+        "Xinyi Hou",
+        "Yanjie Zhao",
+        "Shenao Wang",
+        "Kai Chen",
+        "Haoyu Wang"
+      ],
+      "year": 2025,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2506.02617",
+      "s2_paper_id": "add37b9f02f91d837800030a5f866052e76756f2",
+      "url": "https://arxiv.org/abs/2506.02617",
+      "open_access_pdf": null,
+      "abstract": "Vector database management systems (VDBMSs) play a crucial role in facilitating semantic similarity searches over high-dimensional embeddings from diverse data sources. While VDBMSs are widely used in applications such as recommendation, retrieval-augmented generation (RAG), and multimodal search, their reliability remains underexplored. Traditional database reliability models cannot be directly applied to VDBMSs because of fundamental differences in data representation, query mechanisms, and system architecture. To address this gap, we present the first large-scale empirical study of software defects in VDBMSs. We manually analyzed 1,671 bug-fix pull requests from 15 widely used open-source VDBMSs and developed a comprehensive taxonomy of bugs based on symptoms, root causes, and developer fix strategies. Our study identifies five categories of bug symptoms, with more than half manifesting as functional failures. We further reveal 31 recurring fault patterns and highlight failure modes unique to vector search systems. In addition, we summarize 12 common fix strategies, whose distribution underscores the critical importance of correct program logic. These findings provide actionable insights into VDBMS reliability challenges and offer guidance for building more robust future systems.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2506.02617",
+              "source_type": "paper_citation_context",
+              "excerpt": "Rigger et al. [29] further introduced the Non-Optimizing Reference Engine Construction (NoREC) method to detect optimization bugs in query engines.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c208e5f203e12ae78a576190a4ef29bc2650c30c92b26bb5347f5527594dcb20",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/add37b9f02f91d837800030a5f866052e76756f2",
+        "source_type": "paper",
+        "content_sha256": "sha256:6d7291958642f0531255ce6381a55dd12dc55744fcfcd97f4b1994ecf7ba30af"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icsme64153.2025.00030",
+      "title": "TSGuard: Detecting Logic Bugs in Time Series Management Systems Via Time Series Algebra",
+      "authors": [
+        "Lingwei Kuang",
+        "Liang Liu",
+        "Wen-Jing Wang",
+        "Ning Cao",
+        "Shijie Li",
+        "Fan Liu",
+        "Haolong Chen",
+        "WenJian Liao"
+      ],
+      "year": 2025,
+      "venue": "IEEE International Conference on Software Maintenance and Evolution",
+      "doi": "10.1109/icsme64153.2025.00030",
+      "arxiv_id": null,
+      "s2_paper_id": "eac49ea8165340903f91a6b9e90f73ddfdf6e25b",
+      "url": "https://doi.org/10.1109/icsme64153.2025.00030",
+      "open_access_pdf": null,
+      "abstract": "Time Series Management System (TSMS) is a specialized database management system designed for storing, querying, and analyzing time series data. Its correctness is essential for accurate data processing. However, logic bugs can lead to erroneous query outputs, severely compromising the reliability of data analysis. Compared with traditional relational database SQL, time series SQL exhibits significant syntactic and semantic differences, making existing tools inapplicable. To the best of our knowledge, the detection of logic bugs remains an open problem. In this paper, we propose TSGuard, a tool for detecting logic bugs in TSMSs via time series algebra. The core idea of TSGuard is to convert time series SQL queries into equivalent time series algebra expressions, evaluate these expressions to derive the expected result set, and then compare it with the actual query result set to detect potential logic bugs in the TSMS. Additionally, we introduce a feedback mechanism for query generation and develop query syntax validators for different TSMSs to improve the efficiency of logic bug detection. Through extensive testing, TSGuard discovered 48 previously unknown bugs, including 45 logic bugs and 3 crash bugs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+              "source_type": "paper_citation_context",
+              "excerpt": "For instance, PQS [12] synthesizes a query Q based on an expected result set (i.e., the pivot row) and verifies whether the actual result set is a superset of the expected result set to detect logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:eba58aa012165aa1d88d341b50b495731ddd4b0502214d37f3ee9e20d2a387bd",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+              "source_type": "paper_citation_context",
+              "excerpt": "NOREC [10] transforms an original query into a non-optimized but semantically equivalent SQL query, detecting logic bugs by comparing their execution results.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:11a113baef61ddaed5d57c6325867a9936d511b9e6afdf169a6f2ef6fb245195",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, NoREC [10] detects logic bugs in database optimizers by mutating the original query into a non-optimized query and comparing their result sets.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a465ed01929b5b826c49d1a916a4fb58916aaacd285ec6ad3d94c9f37f950374",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+              "source_type": "paper_citation_context",
+              "excerpt": "This technique has been widely applied to both RDBMSs and GDBMSs. PQS [12] randomly selects specific rows from the database and synthesizes test cases, verifying that the results of these test cases contain the selected rows to detect logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7246e17cb2a3efb6b5f3c44ca7746cf78518d1b26bf94b3d0d70ca73ca8cbb51",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+              "source_type": "paper_citation_context",
+              "excerpt": "Similarly, TLP [11] mutates the original query into multiple partitioned queries, reassembles their result sets, and verifies consistency with the original result set to detect logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3bd5a97543aa59778224216ae1f8f0291fc08eb61ee4d352e00dea8545b26dce",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icsme64153.2025.00030",
+              "source_type": "paper_citation_context",
+              "excerpt": "QPG [28] guides database state mutation using query plans and applies logic bug oracles to identify logic bugs across various database states.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:34157d525b6e7b37b6dfb61c30734e6978b1e30374611d4a7e931b268d6c68fc",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icsme64153_2025_00030.json",
+              "source_type": "paper",
+              "excerpt": "Comparison with Other Techniques To evaluate the effectiveness of TSGuard in detecting logic bugs in TSMSs, we adapted the open-source relational database testing tool SQLancer as a baseline for comparison.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, D Comparison with Other Techniques, page 9.",
+              "content_sha256": "sha256:eac4272791777823eb62698d702ec6875c21c40938d4e369430d5877d93bea08",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icsme64153_2025_00030.json",
+              "source_type": "paper",
+              "excerpt": "MetricsSQLancer TSGuard InfluxDB IotDB TDengine InfluxDB IotDB TDengine Bugs 2 2 3 12 9 17 Sequences 3805 42653 118889 178232 413218 222726 Valid Queries23243 114024 95007 196583 447972 264704 Invalid Queries376 27401 132957 1905 48253 11502 Table V presents the results from the 7200-second experiment, comparing TSGuard and SQLancer in terms of bug detection, syntax node sequence generation, and query validity.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, D Comparison with Other Techniques, page 9.",
+              "content_sha256": "sha256:60e844e931f54c3df5bd291820295d6efe8b88221ab4364ef05b0f8d66fdf3c4",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icsme64153_2025_00030.json",
+              "source_type": "paper",
+              "excerpt": "The results show that TSGuard detected 38 bugs across the three TSMSs, whereas SQLancer detected only 7.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, D Comparison with Other Techniques, page 9.",
+              "content_sha256": "sha256:5b2bc4951578538cf07973928dbda72b75dff4fdc0e51c38e715027d252eaa5c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/eac49ea8165340903f91a6b9e90f73ddfdf6e25b",
+        "source_type": "paper",
+        "content_sha256": "sha256:93aaae30a21a904c0c332364e4584e53a76d6e2b6289c8291b1d000f7528ec9b"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/LingweiKuang/TSGuard",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "renamed_sqlancer_package",
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/LingweiKuang/TSGuard",
+              "source_type": "github_repository",
+              "excerpt": "TSGuard: Detecting Logic Bugs in Time Series Management Systems via Time Series Algebra\n# TSGuard-Detecting-Logic-Bugs-in-Time-Series-Management-Systems-via-Time-Series-Algebra\nTSGuard: Detecting Logic Bugs in Time Series Management Systems via Time Series Algebra",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/LingweiKuang/TSGuard/blob/main/TSGuard/tsFuzzy/src/main/java/com/fuzzy/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "TSGuard/tsFuzzy/src/main/java/com/fuzzy/Randomly.java is SQLancer's Randomly.java, with the package renamed to com.fuzzy (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:66bc91c4144354ceb03896ba19c217e53af2571b6110cb50426983e6c3c27f63",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3689031.3696064",
+      "title": "Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs",
+      "authors": [
+        "Jingzhou Fu",
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Yanyang Zhao",
+        "Shanshan Li",
+        "Yu Jiang"
+      ],
+      "year": 2025,
+      "venue": "European Conference on Computer Systems",
+      "doi": "10.1145/3689031.3696064",
+      "arxiv_id": null,
+      "s2_paper_id": "46c6d9b57f9ef11a0f1dd4eae995265c3478194a",
+      "url": "https://doi.org/10.1145/3689031.3696064",
+      "open_access_pdf": null,
+      "abstract": "Built-in SQL functions are crucial in Database Management Systems (DBMSs), supporting various operations and computations across multiple data types. They are essential for querying, data transformation, and aggregation. Despite their importance, the bugs in SQL functions have caused widespread problems in the real world, from system failures to arbitrary code execution. However, the understanding of the bug characteristics is limited. More importantly, conventional function testing methods struggle to generate semantically correct SQL test cases, while DBMS testing efforts are hard to measure built-in SQL functions. This paper presents a comprehensive study of 318 built-in SQL function bugs, shedding light on their characteristics and root causes. Our investigation reveals that 87.4% of these bugs were caused by improper handling of boundary values of arguments. The boundary values of arguments come from three sources: literal values, type castings, and nested functions. By studying the bugs from three sources, we summarized 10 SQL patterns of bug-inducing queries. Moreover, we designed Soft, a testing tool based on the patterns to test seven widely used DBMSs, including PostgreSQL, MySQL, and ClickHouse. Soft discovered and confirmed 132 previously unknown SQL function bugs. The DBMS vendors took these bugs seriously and fixed 97 bugs in three days. For example, the CTO of ClickHouse commented on one bug: \"We must fix it immediately or get rid of this function.\"",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3689031.3696064",
+              "source_type": "paper_citation_context",
+              "excerpt": "To address these issues, we developed Soft, a tool that leverages the patterns identified in our study to generate SQL test cases that effectively target these boundary conditions.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1f364dd49bcb71927ceb05a0915dc63aac5069517548cd5fa490655c2e3ac662",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3689031.3696064",
+              "source_type": "paper_citation_context",
+              "excerpt": "-- Case 1. global buffer overflow in MySQL SELECT AVG (1.2999999999999999999999999999999999999 999999999999999999999999999999999999); Case 1: A global buffer overflow in MySQL.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8b928621e0187d07b1e1326228df7086e07971ccf0454cf1c7a207c581d887d8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3689031_3696064.json",
+              "source_type": "paper",
+              "excerpt": "We also used the latest versions of Sqirrel [63], SQLsmith [ 53], and SQLancer in PQS mode [ 51] with their default configurations to test these DBMSs, but they did not find any SQL function bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, 7.3 Detected DBMS Vulnerabilities, page 10.",
+              "content_sha256": "sha256:cab4700796f16a72d3a6244794efba104081995bf8bdcf40924312f355275085",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3689031_3696064.json",
+              "source_type": "paper",
+              "excerpt": "5 Comparison with Other Testing Works To demonstrate the effectiveness of our methods, we compared Soft against three state-of-the-art DBMS testing tools, namely Sqirrel, SQLancer, and SQLsmith, which are widely used in the industry.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 7.5 Comparison with Other Testing Works, page 12.",
+              "content_sha256": "sha256:b07c8d8b1aa030c8a5d5d950413f018b2f695b1cbb614c2ebb352bdd2972bf04",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3689031_3696064.json",
+              "source_type": "paper",
+              "excerpt": "DBMS Sqirrel SQLancer SQLsmith Soft PostgreSQL 29 123 417 456 MySQL 23 35 – 323 MariaDB 22 20 – 279 ClickHouse – 24 – 711 MonetDB – – 29 171 Total 74 202 446 2,956 Increment* 984 1,567 181 – *Increments are calculated only for commonly supported DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 7.5 Comparison with Other Testing Works, page 12.",
+              "content_sha256": "sha256:3f4419359c841c27ad9ae633b2bbda05afd7b3b95c4d7ad2e3c18c61a82e9afb",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3689031_3696064.json",
+              "source_type": "paper",
+              "excerpt": "Sqirrel, SQLancer, and SQLsmith did not find any SQL function bugs in 24 hours.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 7.5 Comparison with Other Testing Works, page 13.",
+              "content_sha256": "sha256:dc3804ad4b4a7f5373aa3bf350fbf55654542aabc9ab28cc07e5936fa47de967",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3689031_3696064.json",
+              "source_type": "paper",
+              "excerpt": "5 Comparison with Other Testing Works To demonstrate the effectiveness of our methods, we compared Soft against three state-of-the-art DBMS testing tools, namely Sqirrel, SQLancer, and SQLsmith, which are widely used in the industry.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 7.5 Comparison with Other Testing Works, page 12.",
+              "content_sha256": "sha256:b07c8d8b1aa030c8a5d5d950413f018b2f695b1cbb614c2ebb352bdd2972bf04",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/46c6d9b57f9ef11a0f1dd4eae995265c3478194a",
+        "source_type": "paper",
+        "content_sha256": "sha256:ce90713c048648725dbe656cadfa494dd2d1af3cf29feea28ec508c84d6c1ab8"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3764583",
+      "title": "Unveiling Logic Bugs in SPJG Query Optimizations within DBMS",
+      "authors": [
+        "Xiu Tang",
+        "Shijie Yang",
+        "Sai Wu",
+        "Dongxiang Zhang",
+        "Wenchao Zhou",
+        "Feifei Li",
+        "Gang Chen"
+      ],
+      "year": 2025,
+      "venue": "ACM Transactions on Database Systems",
+      "doi": "10.1145/3764583",
+      "arxiv_id": null,
+      "s2_paper_id": "3374c834b786e3f37193ee037036ee45c3dc1620",
+      "url": "https://doi.org/10.1145/3764583",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3764583",
+      "abstract": "Generation-based testing techniques have proven effective in detecting logic bugs in DBMS, often stemming from the improper implementation of query optimizers. However, existing generation-based debugging tools predominantly rely on random testing, which tends to overlook critical error-prone areas such as multi-table joining and grouped aggregation. In this article, we propose TQS, a novel testing framework targeted at detecting logic bugs arising from SPJG (Select-Project-Join-Group By) query optimizations. Given a target DBMS, TQS achieves the goal with two key components: Data-guided Schema and Query Generation (DSG) and Knowledge-guided Query Space Exploration (KQE). DSG addresses the key challenge of multi-table query debugging: how to generate ground-truth (query, result) pairs for verification. DSG utilizes data derived from dimensionally aggregated data cubes, which store data of grouped metric columns. It maps data from data cubes to a wide table, applies database normalization techniques to the wide table to generate a testing schema and maintains a bitmap index for result tracking. To improve debug efficiency, DSG also artificially inserts some noises into the generated data. To avoid repetitive query space search, KQE guides the generation of error-prone cubes, and forms the problem as isomorphic graph set discovery and combines the graph embedding and weighted random walk for query generation. We evaluated TQS on four popular DBMSs: MySQL, MariaDB, TiDB, and PolarDB. Experimental results show that TQS is effective in finding logic bugs of SPJG query optimization in database management systems. It successfully detected 226 bugs within 24 hours, including 63 bugs in MySQL, 52 in MariaDB, 68 in TiDB, and 43 in PolarDB, respectively.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert",
+        "dqp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "cert",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3764583",
+              "source_type": "paper_citation_context",
+              "excerpt": "Table 8 compares the join bugs detected by DQP with those TQS(join), revealing that only 7 of the bugs are truly distinct.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:71bb325335045e290f226cd2e11e6c9cfbfd90db66fadeab63cc3bb76f79befc",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3764583",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [49].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e0977b0c33c7f13e1f4fc9d23edaee90e0df2e4d07e5aee893adc8d0608bbedf",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3764583",
+              "source_type": "paper_citation_context",
+              "excerpt": "MySQL incorrectly retrieves the const table without finding a matching row. to the compatibility problem, SQLancer implements different approaches on different databases (PQS, TLP and DQP on MySQL and PolarDB; NoRec and DQP on MariaDB; TLP and DQP on TiDB).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d3dd017372d9e245c0dc70e8a4cdad51f0fbcfaea6392653dd807d76e33f5a4b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3764583",
+              "source_type": "paper_citation_context",
+              "excerpt": "While DQP claims to have uncovered 26 previously unidentified bugs, our analysis shows that most of these bugs closely resemble those already found by TQS(join).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:12c90411ebdf0f423fa7f1a0a9476198db604a65cdc267f6f69aff2b9fd6d447",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3764583",
+              "source_type": "paper_citation_context",
+              "excerpt": "DQP adopts differential testing, a strategy aimed at replicating bugs identified by TQS(join) [56], which involves ground truth verification for join queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Differential Query Plans (DQP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5ac077f0d0a5a8e04656eabe331473b7f30376f6dce0b8e12ae09345bf9625e0",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3764583",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [50].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e3ab2def08b3bb541379466103c5060cf0e133b00b89ddb89d1a5ebf74608f3c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs",
+            "tlp",
+            "norec",
+            "dqp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3764583.json",
+              "source_type": "paper",
+              "excerpt": "We compare TQS with SQLancer,3which is the state-of-the-art approach to detecting logic bugs in databases.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 4.2 Query Generation Space, page 18.",
+              "content_sha256": "sha256:b217c80f739886c00336ad38770d1680e2c998367b95ea91bda04586f74e088b",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3764583.json",
+              "source_type": "paper",
+              "excerpt": "We use four methods in SQLancer as our baselines.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 4.2 Query Generation Space, page 18.",
+              "content_sha256": "sha256:0f0b3a9a38a923cb8538ca383007fb1be5c4bdcda6de6dc209c37aecf19d7dc2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3764583.json",
+              "source_type": "paper",
+              "excerpt": "to the compatibility problem, SQLancer implements different approaches on different databases (PQS, TLP and DQP on MySQL and PolarDB; NoRec and DQP on MariaDB; TLP and DQP on TiDB).",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, 5.3 Comparison with Existing Tools, page 28.",
+              "content_sha256": "sha256:fae4850c440c1bed6714f782f988e219dd38f72ab3c2698aef79158861d3484a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3764583.json",
+              "source_type": "paper",
+              "excerpt": "We compare TQS with SQLancer,3which is the state-of-the-art approach to detecting logic bugs in databases.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 4.2 Query Generation Space, page 18.",
+              "content_sha256": "sha256:b217c80f739886c00336ad38770d1680e2c998367b95ea91bda04586f74e088b",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/3374c834b786e3f37193ee037036ee45c3dc1620",
+        "source_type": "paper",
+        "content_sha256": "sha256:d948ee28d922d1bf48b4404ee7e24ea7f2dca057e624376e7dae5eff8f50cd82"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/ase63991.2025.00095",
+      "title": "ZendDiff: Differential Testing of PHP Interpreter",
+      "authors": [
+        "Yuancheng Jiang",
+        "Jianing Wang",
+        "Qiange Liu",
+        "Yeqi Fu",
+        "Jian Mao",
+        "Roland H. C. Yap",
+        "Zhenkai Liang"
+      ],
+      "year": 2025,
+      "venue": "International Conference on Automated Software Engineering",
+      "doi": "10.1109/ase63991.2025.00095",
+      "arxiv_id": null,
+      "s2_paper_id": "192e297c91aca3b7a6ab579ef11f38acc09f51a0",
+      "url": "https://doi.org/10.1109/ase63991.2025.00095",
+      "open_access_pdf": null,
+      "abstract": "The PHP interpreter, powering over 70% of web-sites on the internet, plays a crucial role in web development. Existing approaches to finding bugs in PHP primarily focus on detecting explicit security issues through crashes or sanitizer-based oracles, but fail to identify logic bugs that can silently lead to incorrect results. We observe that the introduction of Just-In-Time (JIT) compilation mode in PHP presents an opportunity for differential testing, as it provides an alternative implementation of the same language specification. We propose, ZendDiff, an automatic differential testing framework that effectively detects logic bugs in the PHP interpreter by comparing JIT and non-JIT execution results. Our differential testing incorporates three techniques: program state probing for fine-grained execution state comparison, JIT-aware program mutation to sufficiently exercise JIT functionality, and dual verification to handle non-deterministic behaviors in PHP programs. Our experimental results demonstrate that ZendDiffoutperforms the official test suite used in PHP’s continuous integration, achieving higher code coverage and executing more Zend opcodes. Through ablation studies, we validate the effectiveness of these techniques. To date, ZendDiffhas identified 51 previously unknown logic bugs in the PHP interpreter, with 37 already fixed and 3 confirmed by the PHP maintainers. ZendDiffhas been acknowledged by the PHP community and offers a practical tool for automatically discovering logic bugs in the PHP interpreter.",
+      "cites_seed_techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/ase63991.2025.00095",
+              "source_type": "paper_citation_context",
+              "excerpt": "Code coverage is a widely adopted metric in evaluating fuzzing and testing approaches [40, 41], as it provides a clear and quantifiable way to evaluate which parts of the code have been executed during tests.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:484198e56f702b69803c3ec5df7d63664aa85e11608c5bfcb8bfe49418f05944",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/192e297c91aca3b7a6ab579ef11f38acc09f51a0",
+        "source_type": "paper",
+        "content_sha256": "sha256:b1d01cf64103de8081edca916e9b5eef2ddaf47bf079f3251f5273732d0b0683"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icsip61881.2024.10671554",
+      "title": "A Review of Fuzz Testing for Configuration-Sensitive Software",
+      "authors": [
+        "Lang Chu",
+        "Minhuan Huang",
+        "Xiang Li",
+        "Yuanping Nie"
+      ],
+      "year": 2024,
+      "venue": "IEEE International Conference on Signal and Image Processing",
+      "doi": "10.1109/icsip61881.2024.10671554",
+      "arxiv_id": null,
+      "s2_paper_id": "ad42e73e4c115542ef4ec708b9f01f39a60de8ff",
+      "url": "https://doi.org/10.1109/icsip61881.2024.10671554",
+      "open_access_pdf": null,
+      "abstract": "In the real world, many software systems come with numerous configuration options, which play a crucial role in their operation. These configurations not only provide flexibility and customization to the software's functionality but also largely determine the security boundaries during actual runtime. Many security vulnerabilities only manifest in specific configuration environments. Fuzz testing, a widely recognized security testing approach, has revealed numerous security vulnerabilities in various software systems, demonstrating its practical value in vulnerability detection. In the practice of fuzz testing, comprehensive consideration of the diversity of software configurations is essential for expanding test coverage and uncovering deep-seated vulnerabilities triggered only under specific configurations. This is a key strategy for enhancing the effectiveness of fuzz testing. This paper reviews the techniques for handling software configurations in fuzz testing research. Firstly, we introduce the relevant work on software configurations and software security testing. Then, we summarize several representative tools and frameworks capable of fuzzing both configurations and inputs simultaneously, analyzing their strengths and limitations. Finally, we discuss the challenges and future directions in fuzz testing concerning configuration handling.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icsip61881.2024.10671554",
+              "source_type": "paper_citation_context",
+              "excerpt": "…specialized fuzz testing tools, covering various critical domains such as operating systems[11– 14], virtual machine managers (e.g., HyperCube[15]), network protocols[15, 16], database systems[17, 18], and autonomous vehicles (e.g., RVFUZZER[19]), to meet the security testing needs in these areas.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c31a34adf72cb3a0491427259896702e2786de0ed62c4cf4cf2224f64ce1c55c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/ad42e73e4c115542ef4ec708b9f01f39a60de8ff",
+        "source_type": "paper",
+        "content_sha256": "sha256:87428237fda501a90b9065956e840aac3247be6b84783ad4cef5a24610e35775"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icde60146.2024.00441",
+      "title": "Applications and Challenges for Large Language Models: From Data Management Perspective",
+      "authors": [
+        "Meihui Zhang",
+        "Zhaoxuan Ji",
+        "Zhaojing Luo",
+        "Yuncheng Wu",
+        "Chengliang Chai"
+      ],
+      "year": 2024,
+      "venue": "IEEE International Conference on Data Engineering",
+      "doi": "10.1109/icde60146.2024.00441",
+      "arxiv_id": null,
+      "s2_paper_id": "35dd88676e19a2203ff86e5837f8537e431d5d6b",
+      "url": "https://doi.org/10.1109/icde60146.2024.00441",
+      "open_access_pdf": null,
+      "abstract": "Data management is indispensable for informed decision-making in the big data era. In the meantime, Large Language Models (LLMs), equipped with billions of model parameters and trained on extensive data corpora, have recently achieved record-breaking results in various real-world applications, such as machine translation, content generation, information retrieval, etc. The emergent abilities of LLMs, e.g., in-context learning and advanced reasoning ability, have great potential to revolutionize data management. In this paper, we first present some promising categories of data management applications where LLMs can be adapted, including data generation, data transformation, data integration, and data exploration. We then discuss the corresponding challenges for such adaption. Finally, we envision potential solutions to these challenges.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icde60146.2024.00441",
+              "source_type": "paper_citation_context",
+              "excerpt": "Meanwhile, to detect the logic bugs of DBMS, we need to generate some SQL queries with semantic equivalence, which produce the same results [20].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bdcc62ddc6cce5c1902042a3cce3f7c2052978e3ecc4d532df3ce2c2825e80c5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde60146.2024.00441",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, to comprehensively detect the bugs of DBMS, it is important to feed the database with a huge number of SQL queries [20].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8e441c2e1868ed0c87acc65870f17e7d26e465490f83acb37423f5ffb056813f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/35dd88676e19a2203ff86e5837f8537e431d5d6b",
+        "source_type": "paper",
+        "content_sha256": "sha256:6084d601658667d117712901fc575f5a845ca5d98bbffdc5ae86bab82671ddec"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/sp54263.2024.00109",
+      "title": "Chronos: Finding Timeout Bugs in Practical Distributed Systems by Deep-Priority Fuzzing with Transient Delay",
+      "authors": [
+        "Yuanliang Chen",
+        "Fuchen Ma",
+        "Yuanhang Zhou",
+        "Ming Gu",
+        "Qing Liao",
+        "Yu Jiang"
+      ],
+      "year": 2024,
+      "venue": "IEEE Symposium on Security and Privacy",
+      "doi": "10.1109/sp54263.2024.00109",
+      "arxiv_id": null,
+      "s2_paper_id": "7b3bc7fdb7d6a6234ed892b057f8feb6118de0b8",
+      "url": "https://doi.org/10.1109/sp54263.2024.00109",
+      "open_access_pdf": null,
+      "abstract": "Delays are inevitable in complex distributed environments. Timeout mechanisms are commonly used to handle unexpected failures in distributed systems. However, incorrect timeout handling or implementation errors in timeout mechanisms can lead to system hang-ups or crashes. Such timeout bugs may be crucial and pose a significant threat to the availability and security of distributed systems.In this work, we introduce Chronos, a general testing framework for automatically detecting timeout bugs in distributed systems with deep-priority transient delays. First, we propose general runtime delayed libraries that dynamically inject fine-grained delays in a Distributed System Under Test (DSUT). To effectively trigger delays and constantly explore timeout bugs in deep paths, Chronos harnesses a deep-priority guided fuzzing that dynamically generates high-quality delay sequences in the runtime. Then, Chronos utilizes transient delays to eliminate the time overhead caused by actual delays and accelerate the test process. We implemented and evaluated Chronos on four widely used distributed systems, including ZooKeeper, MySQL-Cluster, HDFS, and Go-Ethereum. Compared with the state-of-the-art techniques, Random, Brute-Force, and Coverage-Guided fault injection, Chronos covers 26.40%, 21.69%, and 15.14% more timeout mechanism logic, respectively. Furthermore, Chronos has detected 27 timeout bugs in these real-world applications, which have been repaired by the corresponding maintainers.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/sp54263.2024.00109",
+              "source_type": "paper_citation_context",
+              "excerpt": "For MySQL-Cluster, the workload is generated by SQLancer [57], one of the widely used SQL generators for testing database systems.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:addf9e83d5b148837b490a8939288d7c299297ed58e61cc4d25bbfe47c81d3cf",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_sp54263_2024_00109.json",
+              "source_type": "paper",
+              "excerpt": "For MySQL-Cluster, the workload is generated by SQLancer [57], one of the widely used SQL generators for testing database systems.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, 5 Implementation, page 8.",
+              "content_sha256": "sha256:addf9e83d5b148837b490a8939288d7c299297ed58e61cc4d25bbfe47c81d3cf",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/7b3bc7fdb7d6a6234ed892b057f8feb6118de0b8",
+        "source_type": "paper",
+        "content_sha256": "sha256:e78187bae68ecb8e7433d5936072996a37f74004645cdf32b53f9a691c9556ca"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/SecTechTool/Chronos",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/SecTechTool/Chronos",
+              "source_type": "github_repository",
+              "excerpt": "# Chronos: An Automatical Testing Framework for Finding Timeout Bugs in Distributed Systems by Self-Adaptive Delay Model.",
+              "note": "Repository is named after Chronos, the tool this paper says it built, and is about database testing.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/SecTechTool/Chronos/blob/main/workload/mysql-cluster/sqlancer/src/sqlancer/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "workload/mysql-cluster/sqlancer/src/sqlancer/Randomly.java is SQLancer's Randomly.java (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:301936f755ab919b50c82cd983f3b382fd82eff22830ba0dd5d4575c95ecce80",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3597503.3639212",
+      "title": "Combining Structured Static Code Information and Dynamic Symbolic Traces for Software Vulnerability Prediction",
+      "authors": [
+        "Huanting Wang",
+        "Zhanyong Tang",
+        "Shin Hwei Tan",
+        "Jie Wang",
+        "Yuzhe Liu",
+        "Hejun Fang",
+        "Chunwei Xia",
+        "Zheng Wang"
+      ],
+      "year": 2024,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1145/3597503.3639212",
+      "arxiv_id": null,
+      "s2_paper_id": "a3f32348f15845519abf3b862fce1682001da1cd",
+      "url": "https://doi.org/10.1145/3597503.3639212",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597503.3639212",
+      "abstract": "Deep learning (DL) has emerged as a viable means for identifying software bugs and vulnerabilities. The success of DL relies on having a suitable representation of the problem domain. However, existing DL-based solutions for learning program representations have limitations - they either cannot capture the deep, precise program semantics or suffer from poor scalability. We present Con-coction, the first DL system to learn program presentations by combining static source code information and dynamic program execution traces. Concoction employs unsupervised active learning techniques to determine a subset of important paths to collect dynamic symbolic execution traces. By implementing a focused symbolic execution solution, Concoction brings the benefits of static and dynamic code features while reducing the expensive sym-bolic execution overhead. We integrate Concoction with fuzzing techniques to detect function-level code vulnerabilities in C pro-grams from 20 open-source projects. In 200 hours of automated concurrent test runs, Concoction has successfully uncovered vul-nerabilities in all tested projects, identifying 54 unique vulnera-bilities and yielding 37 new, unique CVE IDs. Concoction also significantly outperforms 16 prior methods by providing higher accuracy and lower false positive rates.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/a3f32348f15845519abf3b862fce1682001da1cd",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/a3f32348f15845519abf3b862fce1682001da1cd",
+        "source_type": "paper",
+        "content_sha256": "sha256:4519bea5d1b7144c247ff2de25c6444668cb6ee7a7e6f2aa467c2155824bf0bc"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3650212.3680318",
+      "title": "DBStorm: Generating Various Effective Workloads for Testing Isolation Levels",
+      "authors": [
+        "Keqiang Li",
+        "Siyang Weng",
+        "Lyu Ni",
+        "Chengcheng Yang",
+        "Rong Zhang",
+        "Xuan Zhou",
+        "Aoying Zhou"
+      ],
+      "year": 2024,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3650212.3680318",
+      "arxiv_id": null,
+      "s2_paper_id": "e54a5920715b453038a64e65a8c06fa4ce1199db",
+      "url": "https://doi.org/10.1145/3650212.3680318",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3650212.3680318?download=true",
+      "abstract": "Isolation level (IL) acts as a correctness contract between applications and DBMSs. Problematic IL implementations would cause incorrect transaction execution results and erroneous database states. However, existing studies could not efficiently generate various effective workloads for IL test. The core challenges come from the requirements of (a) black-box testing (trigger the IL code of a closed source DBMS), (b) effective testing (evade redundant and ineffective testing), and (c) anomaly-sensitive testing (test various ILs in a distinguishable way). For black-box testing, we investigate the IL implementations of 15 popular DBMSs and discover that they follow a generic framework that utilizes conflict graphs to manage all conflicts of a workload, and performs a verification policy to prevent non-serializable anomalies. For effective testing, we propose a lightweight data state mirroring method, which helps generate SQL operations that precisely access its expected records and participate the formation of specific conflict graphs. We also propose an efficient history-independent approach to generate dissimilar conflict graphs. It guarantees the graph generation overhead is irrelevant to the scale of historical graphs. For anomaly-sensitive testing, we propose an implantation-based approach to orchestrate conflict record accesses and inject them into different transactions according to the anomaly definition. Our approach outperforms existing approaches in testing effectiveness, efficiency, and coverage. Practically, we have successfully found 33 bugs in popular DBMSs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/e54a5920715b453038a64e65a8c06fa4ce1199db",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/1b209b2da737fd26c50a2cb9116c73c9d1db904e",
+              "source_type": "paper_citation_context",
+              "excerpt": "We compare DBStorm with three state-of-the-art DBMS fuzzers on workload validity and code coverage, including two generationbased fuzzers SQLSmith [12] and SQLancer [37], and one mutationbased fuzzer Squirrel [50].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c07f01e9b42143a9e1c2e5bd0ae239bdc572f5b13a2f4c585e2451c525067831",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/1b209b2da737fd26c50a2cb9116c73c9d1db904e",
+              "source_type": "paper_citation_context",
+              "excerpt": ", primary-foreign key constraints, so the generated workload should comply with these constraints, which is a tough thing considering the generation efficiency [18, 37, 41, 50].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:77c22219e3e4d3919697321364fea455b4c6117f840aaa9be47da57fb3f6fe1a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/1b209b2da737fd26c50a2cb9116c73c9d1db904e",
+              "source_type": "paper_citation_context",
+              "excerpt": "Though an automatic workload generation is imperative, current work is mainly designed for query processing instead of transaction processing [12, 28, 37, 41, 50].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bb79c7f05b34c844ff8a89920b25276707eae55f0604061198e84963f5aee82f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/1b209b2da737fd26c50a2cb9116c73c9d1db904e",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [37] generates queries to fetch a specific target row.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b35b84714e99f6fb50c081a4956a5de84a3a5da25ff422f5183eee0ee98fd814",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3650212_3680318.json",
+              "source_type": "paper",
+              "excerpt": "As SQLancer [60] targets for the relational model, we leverage it to generate syntax/semantic correct SQL operations.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 3 DBStorm Framework, page 4.",
+              "content_sha256": "sha256:e18d990e81ee1160be80601722ea523c9b952fc0c17b25cbe190bd06364d31fa",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3650212_3680318.json",
+              "source_type": "paper",
+              "excerpt": "Since SQLancer mainly generates serially executed operations, we revise it to generate parallel operations.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, 3 DBStorm Framework, page 4.",
+              "content_sha256": "sha256:0b6e98ad1449006926031d987254fc4104bc1c062ae8e8a3cf257c4542575bb1",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3650212_3680318.json",
+              "source_type": "paper",
+              "excerpt": "SQLancer [60] and TQS [67] randomly generate queries and provide a test oracle to identify logical bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, 7 Experiments, page 9.",
+              "content_sha256": "sha256:5038b34d7ee5e3060163c38fa46692da2d27f970ecd40e0d0afb1d593977daf4",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3650212_3680318.json",
+              "source_type": "paper",
+              "excerpt": "SQLancer ,Squrriel and SQLsmith sequentially execute transactions in a single thread, so it is impossible to exist a cycle in the conflict graph and we do not put them in Fig.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, 7.3 Anomaly-Sensitive Testing, page 10.",
+              "content_sha256": "sha256:f1c8b21fbdc94242c4b3e095bab339d59a9504885f1fd06581e76bd0d78aad77",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3650212_3680318.json",
+              "source_type": "paper",
+              "excerpt": "SQLsmith /SQLancer /Squirrel reveal 248/449/63 bugs, but none of them is an isolation bug.",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, 7.4 Bug Result, page 11.",
+              "content_sha256": "sha256:3f008e4d6a1065bf3d82f64cded7d2791bb01318e97aeca90b262973a8d1c6ac",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/e54a5920715b453038a64e65a8c06fa4ce1199db",
+        "source_type": "paper",
+        "content_sha256": "sha256:ac3fab7c5c82e01d13cbc6a4daef95c40512e706dc850142d1948443fd687726"
+      },
+      "also_indexed_as": [
+        "paper:s2:1b209b2da737fd26c50a2cb9116c73c9d1db904e"
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3597503.3623307",
+      "title": "Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation",
+      "authors": [
+        "Yuancheng Jiang",
+        "Jiahao Liu",
+        "Jinsheng Ba",
+        "Roland H. C. Yap",
+        "Zhenkai Liang",
+        "Manuel Rigger"
+      ],
+      "year": 2024,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1145/3597503.3623307",
+      "arxiv_id": null,
+      "s2_paper_id": "321f669f888e43e8f44390985a5da2d3fabe8a7e",
+      "url": "https://doi.org/10.1145/3597503.3623307",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597503.3623307",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3623307",
+              "source_type": "paper_citation_context",
+              "excerpt": "To evaluate the effectiveness of our approach compared to GDBMeter in detecting logic bugs, we used the same methodology as prior works [41, 42], that is, to conduct a manual and best-effort analysis to (1) identify any bugs found by GraphGenie that were overlooked by GDBMeter, and (2) determine…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f7d457baa0a1cf4d1aaf06fc5e946bb42b944856e74814c6834a7c2a16e0a01a",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3623307",
+              "source_type": "paper_citation_context",
+              "excerpt": "Furthermore, we compared GraphGenie with two state-of-the-art approaches based on differential testing [33] and query partitioning [42], implemented as tools named Grand [55] and GDBMeter [22].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7a7e058e5b9c6a996ef80cc1703b2440656b6993916b942903d9a1f93c765e97",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3623307",
+              "source_type": "paper_citation_context",
+              "excerpt": "In GDBMSs, the only existing work [22] is based on Predicate Partitioning [42], which aims to generate three disjoint subset queries via Ternary Logic Partitioning [43].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:54da5096e788c6327cd6948c59352713aea10c72480a861985c4f12ab2192af2",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3623307",
+              "source_type": "paper_citation_context",
+              "excerpt": "While base queries generated by GraphGenie are effective in identifying logic bugs in GDBMSs, there are many existing approaches that aim to generate base queries with higher efficiency and coverage like SQLancer [44] and SQLsmith [1].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0a2a36cdf3d79ec09e1182eb762725fb6326b9d7d5421979963252c9cc495fb1",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3623307",
+              "source_type": "paper_citation_context",
+              "excerpt": "To detect new logic bugs in GDBMSs, we intermittently tested the latest versions of the target GDBMSs over a period of three months, which is a typical methodology for evaluating the effectiveness of automatic testing tools [22, 44].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bbcaf8b1229c4a57a68f6d321f31c29722543b3ba6248b74ebdd1ed80cdeb288",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "deterministic",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3623307",
+              "source_type": "paper_citation_context",
+              "excerpt": "Furthermore, we compared GraphGenie with two state-of-the-art approaches based on differential testing [33] and query partitioning [42], implemented as tools named Grand [55] and GDBMeter [22].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Ternary Logic Partitioning (TLP) as state of the art.",
+              "content_sha256": "sha256:7a7e058e5b9c6a996ef80cc1703b2440656b6993916b942903d9a1f93c765e97",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/321f669f888e43e8f44390985a5da2d3fabe8a7e",
+        "source_type": "paper",
+        "content_sha256": "sha256:b33fc861fbf4050d0392e70374d5ddc54724c753ef74b3809346d00c8251fecd"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Graph Database Management Systems (GDBMSs) store graphs as data. They are used naturally in applications such as social net-works, recommendation systems and program analysis. However, they can be affected by logic bugs, which cause the GDBMSs to compute incorrect results and subsequently affect the applications relying on them. In this work, we propose injective and surjective Graph Query Transformation (GQT) to detect logic bugs in GDBMSs. Given a query Q, we derive a mutated query $Q^{\\prime}$, so that either their result sets are: (i) semantically equivalent; or (ii) variant based on the mutation to be either a subset or superset of each other. When the expected relationship between the results does not hold, a logic bug in the GDBMS is detected. The key insight to mutate Q is that the graph pattern in graph queries enables systematic query transformations derived from injective and surjective mappings of the directed edge sets between Q and $Q^{\\prime}$, We implemented injective and surjective Graph Query Transformation (GQT) as a tool called GraphGenie and evaluated it on 6 popular and mature GDBMSs. GraphGenie has found 25 unknown bugs, comprising 16 logic bugs, 3 internal errors, and 6 performance issues. Our results demonstrate the practicality and effectiveness of GraphGenie in detecting logic bugs in GDBMSs which has the potential for improving the reliability of applications relying on these GDBMSs."
+    },
+    {
+      "id": "paper:doi:10.1109/iaecst64597.2024.11117732",
+      "title": "Detecting Logical Bugs in DBMS via Isomerism Fuzz System",
+      "authors": [
+        "Zhe Wang",
+        "Liang Liu",
+        "Ning Wang"
+      ],
+      "year": 2024,
+      "venue": "2024 6th International Academic Exchange Conference on Science and Technology Innovation (IAECST)",
+      "doi": "10.1109/iaecst64597.2024.11117732",
+      "arxiv_id": null,
+      "s2_paper_id": "f341d91f2721ed842a9d7913d2ffe69b47050f51",
+      "url": "https://doi.org/10.1109/iaecst64597.2024.11117732",
+      "open_access_pdf": null,
+      "abstract": "The complexity and diversity of DBMSs present significant challenges and implications for testing. Fuzzing has become a standard approach for detecting bugs in DBMSs. However, the increasing complexity of DBMS architectures brings new challenges to its effectiveness and coverage. This paper introduces the idea of fuzzing DBMS via Isomerism System, which includes multiple instances of one DBMS using different components and configurations. By fuzzing these isomorphic systems, the scope of classic differential testing is expanded, increasing its applicability and efficiency. We implemented this system within SQLancer and tested it on popular DBMSs, ultimately discovering 10 previously unknown types of bugs. The experimental results demonstrate the system's usability and efficiency.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/iaecst64597.2024.11117732",
+              "source_type": "paper_citation_context",
+              "excerpt": "Rigger M. et al. [13] proposed the Non-Optimized Reference Engine Construction (NoREC), aimed at detecting optimization bugs in DBMSs by comparing optimized and non-optimized query s.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:aff89f76a0bf89f88f509086d17625bb32a975d557a3605e0186368a895186ee",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/iaecst64597.2024.11117732",
+              "source_type": "paper_citation_context",
+              "excerpt": "[14] they proposed a query partitioning differential detection method.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c56ec4119d2b9f7d0b9a47c2530de77df75e607e2ee1f8e9d8ab781516068945",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_iaecst64597_2024_11117732.json",
+              "source_type": "paper",
+              "excerpt": "The query generation approach draws on the design principles of SQLancer [16], with certain modifications.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, page 3.",
+              "content_sha256": "sha256:d044a2ca7e7ec9912ae68a1d99299c3a3dc8e1f8bc8ff6ace11c9c5f25f8760a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/f341d91f2721ed842a9d7913d2ffe69b47050f51",
+        "source_type": "paper",
+        "content_sha256": "sha256:b94dce3b9dc85ff6ce7e1b11aa170546ff57a47ef4f495e4ff7357f2280fb037"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3659437.3659445",
+      "title": "Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction",
+      "authors": [
+        "Jiansen Song",
+        "Wensheng Dou",
+        "Yu Gao",
+        "Ziyu Cui",
+        "Yingying Zheng",
+        "Dong Wang",
+        "Wei Wang",
+        "Jun Wei",
+        "Tao Huang"
+      ],
+      "year": 2024,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3659437.3659445",
+      "arxiv_id": null,
+      "s2_paper_id": "2212c954adfea8bd0c35fdf691c6ad50904cb5ab",
+      "url": "https://doi.org/10.14778/3659437.3659445",
+      "open_access_pdf": null,
+      "abstract": "Database Management Systems (DBMSs) are widely used to efficiently store and retrieve data. DBMSs usually support various metadata, e.g., integrity constraints for ensuring data integrity and indexes for locating data. DBMSs can further utilize these metadata to optimize query evaluation. However, incorrect metadata-related optimizations can introduce metadata-related logic bugs, which can cause a DBMS to return an incorrect query result for a given query. In this paper, we propose a general and effective testing approach,\n Raw database construction\n (Radar), to detect metadata-related logic bugs in DBMSs. Given a database\n db\n containing some metadata, Radar first constructs a raw database\n rawDb\n , which wipes out the metadata in\n db\n and contains the same data as\n db.\n Since\n db\n and\n rawDb\n have the same data, they should return the same query result for a given query. Any inconsistency in their returned query results indicates a metadata-related logic bug. To effectively detect metadata-related logic bugs, we further propose a metadata-oriented testing optimization strategy to focus on testing previously unseen metadata, thus detecting more metadata-related logic bugs quickly. We implement and evaluate Radar on five widely-used DBMSs, and have detected 42 bugs, of which 38 have been confirmed as new bugs and 16 have been fixed by DBMS developers.",
+      "cites_seed_techniques": [
+        "norec",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/2212c954adfea8bd0c35fdf691c6ad50904cb5ab",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-10T14:40:47Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T14:40:47Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "artifact_inspection",
+          "evidence": [
+            {
+              "source_url": "https://github.com/JensonSung/radar",
+              "source_type": "github_repository",
+              "excerpt": "Replication package for \"Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction\", accepted at VLDB 2024\n  \n# Radar\n\nRadar is the implementation of raw database construction in this paper.\n\n# Getting Started",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-06T14:28:33Z",
+              "first_seen": "2026-09-06T14:28:33Z",
+              "last_verified": "2026-09-10T14:40:47Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/JensonSung/radar/blob/main/src/sqlancer/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": null,
+              "note": "Repository retains SQLancer source files: src/sqlancer/Randomly.java, src/sqlancer/Main.java, src/sqlancer/common/oracle/TestOracle.java",
+              "retrieved_at": "2026-09-06T14:28:33Z",
+              "first_seen": "2026-09-06T14:28:33Z",
+              "last_verified": "2026-09-10T14:40:47Z"
+            },
+            {
+              "source_url": "https://github.com/JensonSung/radar/blob/main/pom.xml",
+              "source_type": "github_code",
+              "excerpt": "<groupId>com.sqlancer</groupId>",
+              "excerpt_is_verbatim": true,
+              "note": "pom.xml identifies the project as SQLancer.",
+              "content_sha256": "sha256:ff2cd00b42c424f55b94956f780dbedfae301ab02fe78488f4176b730c0da903",
+              "retrieved_at": "2026-09-06T14:28:33Z",
+              "first_seen": "2026-09-06T14:28:33Z",
+              "last_verified": "2026-09-10T14:40:47Z"
+            }
+          ],
+          "classifier": {
+            "method": "artifact_inspection",
+            "classifier_version": "artifact-markers-v1",
+            "policy_version": "impact-policy-v1",
+            "taxonomy_version": "taxonomy-v2",
+            "classified_at": "2026-09-06T14:28:33Z",
+            "model": null,
+            "rationale": "Artifact carries SQLancer markers: randomly_java_present, retained_sqlancer_source_files, sqlancer_build_file_reference, sqlancer_package_structure, sqlancer_provider_directory"
+          }
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T14:40:47Z",
+        "source_url": "https://www.semanticscholar.org/paper/2212c954adfea8bd0c35fdf691c6ad50904cb5ab",
+        "source_type": "paper",
+        "content_sha256": "sha256:eb6b4d3ea471306271453ba52cf6e95089f2837ce0aa7c3e98d21624193bf9b2"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/JensonSung/radar",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-06T14:28:33Z",
+          "sqlancer_markers": [
+            "randomly_java_present",
+            "retained_sqlancer_source_files",
+            "sqlancer_build_file_reference",
+            "sqlancer_package_structure",
+            "sqlancer_provider_directory"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/JensonSung/radar",
+              "source_type": "github_repository",
+              "excerpt": "Replication package for \"Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction\", accepted at VLDB 2024\n  \n# Radar\n\nRadar is the implementation of raw database construction in this paper.\n\n# Getting Started",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-06T14:28:33Z",
+              "first_seen": "2026-09-06T14:28:33Z",
+              "last_verified": "2026-09-10T14:40:47Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/JensonSung/radar/blob/main/src/sqlancer/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": null,
+              "note": "Repository retains SQLancer source files: src/sqlancer/Randomly.java, src/sqlancer/Main.java, src/sqlancer/common/oracle/TestOracle.java",
+              "retrieved_at": "2026-09-06T14:28:33Z",
+              "first_seen": "2026-09-06T14:28:33Z",
+              "last_verified": "2026-09-10T14:40:47Z"
+            },
+            {
+              "source_url": "https://github.com/JensonSung/radar/blob/main/pom.xml",
+              "source_type": "github_code",
+              "excerpt": "<groupId>com.sqlancer</groupId>",
+              "excerpt_is_verbatim": true,
+              "note": "pom.xml identifies the project as SQLancer.",
+              "content_sha256": "sha256:ff2cd00b42c424f55b94956f780dbedfae301ab02fe78488f4176b730c0da903",
+              "retrieved_at": "2026-09-06T14:28:33Z",
+              "first_seen": "2026-09-06T14:28:33Z",
+              "last_verified": "2026-09-10T14:40:47Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1109/icst60714.2024.00012",
+      "title": "Differential Optimization Testing of Gremlin-Based Graph Database Systems",
+      "authors": [
+        "Yingying Zheng",
+        "Wensheng Dou",
+        "Leile Tang",
+        "Ziyu Cui",
+        "Jiansen Song",
+        "Ziyue Cheng",
+        "Wei Wang",
+        "Jun Wei",
+        "Hua Zhong",
+        "Tao Huang"
+      ],
+      "year": 2024,
+      "venue": "International Conference on Information Control Systems & Technologies",
+      "doi": "10.1109/icst60714.2024.00012",
+      "arxiv_id": null,
+      "s2_paper_id": "7f7686d2e2af4a7308fa947680f52194b8bb067f",
+      "url": "https://doi.org/10.1109/icst60714.2024.00012",
+      "open_access_pdf": null,
+      "abstract": "Graph database systems (GDBs) allow efficiently creating, modifying, and retrieving graph data in a graph database. To accelerate graph queries, GDBs usually adopt various and complex optimization strategies. However, incorrect optimizations in GDBs can introduce optimization bugs, which cause a graph query to compute an incorrect query result, e.g., omitting a vertex in a graph database. In this paper, we propose Differential Optimization Testing (DOT), an effective and automated approach to detect optimization bugs in GDBs that adopt Gremlin as their query language. The main idea of DOT is that, given a Gremlin query $Q$, we execute it on the target GDB with two different optimization configurations and then verify whether they can compute the same query results for query $Q$. Any inconsistency between their query results indicates an optimization bug in the target GDB. To improve the efficiency of differential testing in DOT, we further propose an optimization-guided approach, aiming to explore more optimization strategies and more graph database features. We evaluate DOT on six popular and widely-used GDBs, i.e., Neo4j, OrientDB, JanusGraph, HugeGraph, TinkerGraph, and ArcadeDB. In total, we have found 28 unique optimization bugs, 16 of which have been confirmed as previously-unknown bugs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icst60714.2024.00012",
+              "source_type": "paper_citation_context",
+              "excerpt": "Some approaches [20], [21], [25] (e.g., NoREC [21]) utilizes metamorphic testing for finding logic bugs in relational DBMSs. Query generation [17], [54] (e.g., SQLsmith [17]) and generic fuzzing approaches (e.g., AFL [53]) can also be used to detect bugs in relational DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5b7c8a75cf03a44e373a179e6cd87c256b3bc27df95698db6a297f5f819ae4cc",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icst60714.2024.00012",
+              "source_type": "paper_citation_context",
+              "excerpt": "GDBMeter applies query partitioning [20] to test GDBs. Specifically, it partitions a query into three disjoint queries in which a predicate is evaluated to true , false and null , respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:27ffd2e8e21083243d52cd1fad124f2cbe2f462ba3c92f4023354ec1007712ae",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icst60714.2024.00012",
+              "source_type": "paper_citation_context",
+              "excerpt": "Among these approaches, NoREC [21] can detect optimization bugs in relational database systems by rewriting an optimized SQL query into a non-optimizing SQL query.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:dbbaea86c99ea8570155fb267c520fd6dab8538a38bef93be46d9fe8743f61c8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icst60714.2024.00012",
+              "source_type": "paper_citation_context",
+              "excerpt": "Many testing approaches [17]–[26] (e.g., TLP [20], NoREC [21], and DQE [25]) have been proposed to test relational database systems.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6bfb2c5eec6c42b0a321a4b9c6188bcf2bb98d86cc3fada9cd9ad3f293ff51d7",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icst60714_2024_00012.json",
+              "source_type": "paper",
+              "excerpt": "Comparison with Existing Approaches To the best of our knowledge, four existing approaches [13]– [16] can detect logic bugs in GDBs.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, D Comparison with Existing Approaches, page 9.",
+              "content_sha256": "sha256:60568ce34db8fc27054f2d5c8c35e53a1b51570bb409556caa29db14ea334239",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icst60714_2024_00012.json",
+              "source_type": "paper",
+              "excerpt": "Therefore, we compare DOT with differential testing and query partitioning.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, D Comparison with Existing Approaches, page 9.",
+              "content_sha256": "sha256:233701b7a3d7c1d276313af4389879afd088ec573e028ce5e1f5ac16c148b431",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icst60714_2024_00012.json",
+              "source_type": "paper",
+              "excerpt": "Comparison with query partitioning.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, D Comparison with Existing Approaches, page 9.",
+              "content_sha256": "sha256:b585f70e9cf8374cef5b0d813035e9dd79773d66915fc0f47e220c4671657b03",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/7f7686d2e2af4a7308fa947680f52194b8bb067f",
+        "source_type": "paper",
+        "content_sha256": "sha256:d0bdba4002b37303ebab3fc66964507b693fec68764f6f2ddffdc7ecf29e817a"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3797919.3797928",
+      "title": "Dinkel: State-Aware and Granular Framework for Validating Graph Databases",
+      "authors": [
+        "C. Wüst",
+        "Zu-Ming Jiang",
+        "Zhendong Su"
+      ],
+      "year": 2024,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3797919.3797928",
+      "arxiv_id": "2408.07525",
+      "s2_paper_id": "10be4636a8900a45571f1744ee50430a4e125485",
+      "url": "https://doi.org/10.14778/3797919.3797928",
+      "open_access_pdf": null,
+      "abstract": "Graph database management systems (GDBMSs) have been powering many data-driven applications. To ensure GDBMS reliability, several testing approaches have been proposed. However, they all suffer from two key limitations: (1) insufficient support for generating complex and valid queries to exercise deep GDBMS code, and (2) lack of general oracles to validate the execution correctness of arbitrary queries.\n \n In this paper, we propose a novel and practical approach, Dinkel, for thoroughly testing GDBMSs. Our approach consists of two core techniques. First, to generate complex and valid queries, we model two kinds of graph state,\n query context\n and\n graph schema\n , to describe the Cypher variables and the manipulated graph labels and properties. We generate queries clause-by-clause, and modify the graph states on the fly to ensure each clause references the correct state information. Second, to generally validate query results, we introduce two fine-grained query transformations: clause-level and expression-level transformations. These transformations can operate on arbitrary queries while preserving their semantics. Dinkel validates GDBMSs by checking whether the transformed query produces the same results as the original. We evaluated Dinkel on three well-known GDBMSs. In total, we found 127 bugs, among which 113 were confirmed, 84 were fixed, and 33 were logic bugs. Compared to existing approaches, Dinkel can cover over 70% more code and find substantially more bugs within a 48-hour testing campaign. We expect Dinkel's powerful bug detection to lay a practical foundation for GDBMS testing.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3797919.3797928",
+              "source_type": "paper_citation_context",
+              "excerpt": "Moreover, we propose two fine-grained query transformations: clause-level transformations and expression-level transformations, which can operate on arbitrary Cypher queries to validate their correctness.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:703aa489b63b6199d2469d00a20f790355dc6d1d37a78eb9c64a2475b7c66d07",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14778_3797919_3797928.json",
+              "source_type": "paper",
+              "excerpt": "As shown in Table 4, Dinkel supports more Cypher clauses than existing approaches [ 13,14,20,27,51].",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 4 IMPLEMENTATION, page 7.",
+              "content_sha256": "sha256:9617a87692abf14e4718ba36e9d4b10c042b1ae627f9bae7dcd5355a5e18fb47",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14778_3797919_3797928.json",
+              "source_type": "paper",
+              "excerpt": "This comparison is reasonable and objective because: (1) none of the bugs found by Dinkel are marked as duplicated by developers, meaning that no approach found these bugs until Dinkel found them; and (2) all existing approaches have extensively tested Neo4j and RedisGraph [ 13,14,20,27,51], meaning that in these two GDBMSs, no approach found the long-latent bugs found by Dinkel during their evaluation.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 5.5 Comparison, page 11.",
+              "content_sha256": "sha256:d228bd767cc8ac6d232454642868070b3355686fc0bd5246ddd429374a9ff4f0",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/10be4636a8900a45571f1744ee50430a4e125485",
+        "source_type": "paper",
+        "content_sha256": "sha256:05510a7cf95fca7eb9f1fe0d3fa38e99ff0da72ad00c3f8b09c759ce5fb6ba32"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3643779",
+      "title": "DTD: Comprehensive and Scalable Testing for Debuggers",
+      "authors": [
+        "Hongyi Lu",
+        "Zhibo Liu",
+        "Shuai Wang",
+        "Fengwei Zhang"
+      ],
+      "year": 2024,
+      "venue": "Proc. ACM Softw. Eng.",
+      "doi": "10.1145/3643779",
+      "arxiv_id": null,
+      "s2_paper_id": "ebb7a876107f751507a0074cc8dbc28ded7115cf",
+      "url": "https://doi.org/10.1145/3643779",
+      "open_access_pdf": "https://doi.org/10.1145/3643779",
+      "abstract": "As a powerful tool for developers, interactive debuggers help locate and fix errors in software. By using debugging information included in binaries, debuggers can retrieve necessary program states about the program. Unlike printf-style debugging, debuggers allow for more flexible inspection and modification of program execution states. However, debuggers may incorrectly retrieve and interpret program execution, causing confusion and hindering the debugging process. Despite the wide usage of interactive debuggers, a scalable and comprehensive measurement of their functionality correctness does not exist yet. Existing works either fall short in scalability or focus more on the “compiler-side” defects instead of debugger bugs. To facilitate a better assessment of debugger correctness, we first propose and advocate a set of debugger testing criteria, covering both comprehensiveness (in terms of debug information covered) and scalability (in terms of testing overhead). Moreover, we design comparative experiments to show that fulfilling these criteria is not only theoretically appealing, but also brings major improvement to debugger testing. Furthermore, based on these criteria, we present DTD, a differential testing (DT) framework for detecting bugs in interactive debuggers. DTD compares the behaviors of two mainstream debuggers when processing an identical C executable — discrepancies indicate bugs in one of the two debuggers. DTD leverages a novel heuristic method to avoid the repetitive structures (e.g., loops) that exist in C programs, which facilitates DTD to achieve full debug information coverage efficiently. Moreover, we have also designed a Temporal Differential Filtering method to practically filter out the false positives caused by the uninitialized variables in common C programs. With these carefully designed techniques, DTD fulfills our proposed testing requirements and, therefore, achieves high scalability and testing comprehensiveness. For the first time, it offers large-scale testing for C debuggers to detect debugger behavior discrepancies when inspecting millions of program states. An empirical comparison shows that DTD finds 17× more error-triggering cases and detects 5× more bugs than the state-of-the-art debugger testing technique. We have used DTD to detect 13 bugs in the LLVM toolchain (Clang/LLDB) and 5 bugs in the GNU toolchain (GCC/GDB). One of our fixes has already landed in the latest LLDB development branch.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/ebb7a876107f751507a0074cc8dbc28ded7115cf",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/ebb7a876107f751507a0074cc8dbc28ded7115cf",
+        "source_type": "paper",
+        "content_sha256": "sha256:1505b539a1fd305026dd5553c66a89d516f4b88656c970c58f3f5a53021d9f94"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icicnis64247.2024.10823213",
+      "title": "Dynamic Adjustment Paradigm based on Genetic Algorithm in Database Index Optimization",
+      "authors": [
+        "Jinsong Wang"
+      ],
+      "year": 2024,
+      "venue": "2024 International Conference on IoT Based Control Networks and Intelligent Systems (ICICNIS)",
+      "doi": "10.1109/icicnis64247.2024.10823213",
+      "arxiv_id": null,
+      "s2_paper_id": "c30c758ad17b50b6de00ac4878065a4436e07de0",
+      "url": "https://doi.org/10.1109/icicnis64247.2024.10823213",
+      "open_access_pdf": null,
+      "abstract": "With the continuous increase of data complexity, how to conduct the efficient database mining and analysis has become a research focus in the computer field. Based on this, this study proposes a dynamic database index optimization model based on a new genetic algorithm, aiming to solve shortcomings of existing database index recommendation methods in large-scale, high-concurrency data processing. First, this study analyzes the limitations of traditional index recommendation methods based on heuristic rules and enumeration algorithms in environments with limited computing resources and frequent load changes, thus leading to the core problem. By introducing a dynamic adjustment paradigm based on genetic algorithms and optimizing the convergence function, this model can adjust the index configuration in real time to improve query efficiency and database response speed. OtterTune workflow was selected as the research platform and a certain degree of modeling optimization was performed. In the experiment section, this study applied the proposed method to database environments such as Oracle 20c and SQL Server 2019, and verified the effectiveness of the model through disaster recovery performance and comparative query performance tests. In the disaster recovery test, the experimental results show that as the number of data blocks increases, the recovery time increases approximately linearly. In the query performance test, using the traditional Hash algorithm as a comparison, the experiment found that this model has a query load cost growth rate of (LCIC), thus verifying the stability and effectiveness of this model.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/c30c758ad17b50b6de00ac4878065a4436e07de0",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/c30c758ad17b50b6de00ac4878065a4436e07de0",
+        "source_type": "paper",
+        "content_sha256": "sha256:41138bcf961c9ca9f7b3773175d9d123d9da944831ce5f8fb89058532916c622"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3649815",
+      "title": "Finding Cross-Rule Optimization Bugs in Datalog Engines",
+      "authors": [
+        "Chi Zhang",
+        "Linzhang Wang",
+        "Manuel Rigger"
+      ],
+      "year": 2024,
+      "venue": "Proc. ACM Program. Lang.",
+      "doi": "10.1145/3649815",
+      "arxiv_id": "2402.12863",
+      "s2_paper_id": "428d0ae6be2e3e5146c2ca166fc7e64848929297",
+      "url": "https://doi.org/10.1145/3649815",
+      "open_access_pdf": "https://doi.org/10.1145/3649815",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3649815",
+              "source_type": "paper_citation_context",
+              "excerpt": "We consider adding any feature to a rule based on a pre-defined probability [Rigger and Su 2020a; Seltenreich 2023; Slutz 1998",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3ae5a6a34576479398cd0774a44a45f518d80430a1fdeb06aaa8691636b03393",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3649815",
+              "source_type": "paper_citation_context",
+              "excerpt": "For database engines, NoREC [Rigger and Su 2020a] rewrites queries to prevent their optimization; while the high-level idea of NoREC and IRE are similar, their actual approaches differ significantly.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:791f92e9a721665623f62f6616d7e03bb746fbc3c140e69d1deffff80520a700",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3649815",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, as also highlighted by prior work on testing database engines [Jiang et al. 2023; Liang et al. 2022; Rigger and Su 2020c; Zhong et al. 2020], it can be difficult to avoid semantic errors.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6715c87c418842b8ecd04eca2b05d3abbb5d76fb275d3a44c2718f0b06e93e25",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3649815",
+              "source_type": "paper_citation_context",
+              "excerpt": "Non-optimizing Reference Engine Construction (NoREC) [Rigger and Su 2020a] was proposed to find optimization bugs in relational database systems.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5368c920b1a4665821d3033a1d8b1f03cd2c45ccf6323cbad8c6f65bbb7b9083",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3649815",
+              "source_type": "paper_citation_context",
+              "excerpt": "We adopted this methodology from prior testing works [Rigger and Su 2020a,b].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:554ac101e672d5dd52547e5a358811e6e304c9fb1c6dc6ee7ffd74726dd5d21d",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/428d0ae6be2e3e5146c2ca166fc7e64848929297",
+        "source_type": "paper",
+        "content_sha256": "sha256:88e69c0769c163b29869d9ee0c38a88f38e4cda9c4f274703e78b4d0e0b73ad5"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Datalog is a popular and widely-used declarative logic programming language. Datalog engines apply many cross-rule optimizations; bugs in them can cause incorrect results. To detect such optimization bugs, we propose an automated testing approach called Incremental Rule Evaluation (IRE), which synergistically tackles the test oracle and test case generation problem. The core idea behind the test oracle is to compare the results of an optimized program and a program without cross-rule optimization; any difference indicates a bug in the Datalog engine. Our core insight is that, for an optimized, incrementally-generated Datalog program, we can evaluate all rules individually by constructing a reference program to disable the optimizations that are performed among multiple rules. Incrementally generating test cases not only allows us to apply the test oracle for every new rule generated-we also can ensure that every newly added rule generates a non-empty result with a given probability and eschew recomputing already-known facts. We implemented IRE as a tool named Deopt, and evaluated Deopt on four mature Datalog engines, namely Soufflé, CozoDB, μZ, and DDlog, and discovered a total of 30 bugs. Of these, 13 were logic bugs, while the remaining were crash and error bugs. Deopt can detect all bugs found by queryFuzz, a state-of-the-art approach. Out of the bugs identified by Deopt, queryFuzz might be unable to detect 5 . Our incremental test case generation approach is efficient; for example, for test cases containing 60 rules, our incremental approach can produce 1.17× (for DDlog) to 31.02× (for Soufflé) as many valid test cases with non-empty results as the naive random method. We believe that the simplicity and the generality of the approach will lead to its wide adoption in practice."
+    },
+    {
+      "id": "paper:doi:10.1145/3698810",
+      "title": "Finding Logic Bugs in Spatial Database Engines via Affine Equivalent Inputs",
+      "authors": [
+        "Wenjing Deng",
+        "Qiuyang Mang",
+        "Chengyu Zhang",
+        "Manuel Rigger"
+      ],
+      "year": 2024,
+      "venue": "Proc. ACM Manag. Data",
+      "doi": "10.1145/3698810",
+      "arxiv_id": "2410.12496",
+      "s2_paper_id": "5f5d0387c14cd44def66d352acba1013494fffd0",
+      "url": "https://doi.org/10.1145/3698810",
+      "open_access_pdf": "https://doi.org/10.1145/3698810",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3698810",
+              "source_type": "paper_citation_context",
+              "excerpt": "For instance, Ternary Logic Partitioning (TLP) is a general state-of-the-art testing technique for relational DBMSs [36], applicable not only to relational DBMSs, but also to graph DBMSs [21].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9c80f85d5114d8952cb6e89f12140ecf17cba995697d02a229833e3f84fed2e9",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3698810",
+              "source_type": "paper_citation_context",
+              "excerpt": "For this purpose, we employed differential testing approaches alongside TLP [36], a state-of-the-art methodology for testing relational DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a22bf1be9645edf0f6e7500c678cd687c7378bb88c8c7a1c4b7a5fa75ea5eea3",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3698810",
+              "source_type": "paper_citation_context",
+              "excerpt": "Many automated testing techniques have been proposed for detecting logic bugs in relational DBMSs [1, 8, 20, 25, 35–38], but unfortunately, they are not applicable for testing SDBMSs, especially for spatial-related features.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bbdcad4454d711fe4f1b6b4cd1ad46575c587392937ede4e1a0cdf456c7a72d6",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "deterministic",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3698810",
+              "source_type": "paper_citation_context",
+              "excerpt": "For this purpose, we employed differential testing approaches alongside TLP [36], a state-of-the-art methodology for testing relational DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Ternary Logic Partitioning (TLP) as state of the art.",
+              "content_sha256": "sha256:a22bf1be9645edf0f6e7500c678cd687c7378bb88c8c7a1c4b7a5fa75ea5eea3",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3698810",
+              "source_type": "paper_citation_context",
+              "excerpt": "For instance, Ternary Logic Partitioning (TLP) is a general state-of-the-art testing technique for relational DBMSs [36], applicable not only to relational DBMSs, but also to graph DBMSs [21].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Ternary Logic Partitioning (TLP) as state of the art.",
+              "content_sha256": "sha256:9c80f85d5114d8952cb6e89f12140ecf17cba995697d02a229833e3f84fed2e9",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/5f5d0387c14cd44def66d352acba1013494fffd0",
+        "source_type": "paper",
+        "content_sha256": "sha256:02a086805eb60586136f02b56577e1e92a54e12ce30df5da9eabcf1964d2f92b"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Spatial Database Management Systems (SDBMSs) aim to store, manipulate, and retrieve spatial data. SDBMSs are employed in various modern applications, such as geographic information systems, computer-aided design tools, and location-based services. However, the presence of logic bugs in SDBMSs can lead to incorrect results, substantially undermining the reliability of these applications. Detecting logic bugs in SDBMSs is challenging due to the lack of ground truth for identifying incorrect results. In this paper, we propose an automated geometry-aware generator to generate high-quality SQL statements for SDBMSs and a novel concept named Affine Equivalent Inputs (AEI) to validate the results of SDBMSs. We implemented them as a tool named Spatter (Spatial DBMS Tester) for finding logic bugs in four popular SDBMSs: PostGIS, DuckDB Spatial, MySQL, and SQL Server. Our testing campaign detected 34 previously unknown and unique bugs in these SDBMSs, of which 30 have been confirmed, and 18 have already been fixed. Our testing efforts have been well appreciated by the developers. Experimental results demonstrate that the geometry-aware generator significantly outperforms a naive random-shape generator in detecting unique bugs, and AEI can identify 14 logic bugs in SDBMSs that were totally overlooked by previous methodologies."
+    },
+    {
+      "id": "paper:doi:10.1145/3597503.3639208",
+      "title": "Finding XPath Bugs in XML Document Processors via Differential Testing",
+      "authors": [
+        "Shuxin Li",
+        "Manuel Rigger"
+      ],
+      "year": 2024,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1145/3597503.3639208",
+      "arxiv_id": "2401.05112",
+      "s2_paper_id": "c39d40e3127a56a0abe12f2fb1d698ea6ba1724d",
+      "url": "https://doi.org/10.1145/3597503.3639208",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597503.3639208",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639208",
+              "source_type": "paper_citation_context",
+              "excerpt": "The targeted node in XPress was inspired by the pivot row in Pivoted Query Synthesis (PQS) [36], which was originally proposed to test relational DBMSs. PQS’ and XPress’ commonality is that they select a random element, in PQS, a row in the database, while for XPress, a node in an XML document,…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:38adde86c67037e37b377067431d778f6d5c4727948c3f5624e9a2d691117ca1",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639208",
+              "source_type": "paper_citation_context",
+              "excerpt": "A similar high-level idea has been proposed in the context of testing relational DBMSs, called Pivoted Query Synthesis (PQS) [36], where a pivot row was selected, based on which predicates were rectified to return true.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ff61f9181a6fee5ae52135b846be8d50c116b4af3cda80851ddc669eb85beaa2",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639208",
+              "source_type": "paper_citation_context",
+              "excerpt": "Todic and Uzelac have proposed an automated testing technique for SQLServer’s index support; their test oracle compared the results of a given query with and without index definition [41].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:40b1e6f1aff63e7a3fc197748b98338a11d9e4f08515f02b473cd911ec1f0df9",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639208",
+              "source_type": "paper_citation_context",
+              "excerpt": "It tackled the test oracle problem by using differential testing by comparing the results of Microsoft’s SQLServer with and without using indexes.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5580ad75a0560dfb5442d0581e9bdbb018e8fd60138ad6d9491eb2c45417351d",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639208",
+              "source_type": "paper_citation_context",
+              "excerpt": "XQuery extends XPath to provide functionalities such as node constructors and SQL-like clauses.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:aa9362468af9c41cddd52ba2e7b06dc9fe011c9a7f651610c7bbf3fc4004a289",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639208",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, previous bug-finding efforts on testing DBMSs using SQL queries also found no logic bug in PostgreSQL [34, 35].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5cb1ad8a5cf5af21ba9dfd70dc49b9980be60b1455cd9cd57953fc6db58cc307",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "manual_curation",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639208",
+              "source_type": "paper_citation_context",
+              "excerpt": "The targeted node in XPress was inspired by the pivot row in Pivoted Query Synthesis (PQS) [36], which was originally proposed to test relational DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+              "content_sha256": "sha256:1606d3d25257067d7e58ed31b06880c3fab67dfa363a8506cce6d47e8ad1a25d",
+              "retrieved_at": "2026-09-06T07:08:48Z",
+              "first_seen": "2026-09-06T07:08:48Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ],
+          "classifier": {
+            "method": "manual_curation",
+            "classifier_version": "manual-curation-v1",
+            "policy_version": "impact-policy-v1",
+            "taxonomy_version": "taxonomy-v1",
+            "classified_at": "2026-09-06T07:08:48Z",
+            "model": null,
+            "rationale": "Reviewed by a maintainer against the citation contexts collected for this paper."
+          }
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/c39d40e3127a56a0abe12f2fb1d698ea6ba1724d",
+        "source_type": "paper",
+        "content_sha256": "sha256:2587a99fe11fe5058ee454e35affa4ce98722ae0bb04d440240090e5fbbf1a5b"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Extensible Markup Language (XML) is a widely used file format for data storage and transmission. Many XML processors support XPath, a query language that enables the extraction of elements from XML documents. These systems can be affected by logic bugs, which are bugs that cause the processor to return incorrect results. In order to tackle such bugs, we propose a new approach, which we realized as a system called XPress. As a test oracle, XPress relies on differential testing, which compares the results of multiple systems on the same test input, and identifies bugs through discrepancies in their outputs. As test inputs, XPress generates both XML documents and XPath queries. Aiming to generate meaningful queries that compute non-empty results, XPress selects a so-called targeted node to guide the XPath expression generation process. Using the targeted node, XPress generates XPath expressions that reference existing context related to the targeted node, such as its tag name and attributes, while also guaranteeing that a predicate evaluates to true before further expanding the query. We tested our approach on six mature XML processors, BaseX, eXist-DB, Saxon, PostgreSQL, lib XML2, and a commercial database system. In total, we have found 27 unique bugs in these systems, of which 25 have been verified by the developers, and 20 of which have been fixed. XPress is efficient, as it finds 12 unique bugs in BaseX in 24 hours, which is 2× as fast as naive random generation. We expect that the effectiveness and simplicity of our approach will help to improve the robustness of many XML processors."
+    },
+    {
+      "id": "paper:arxiv:2410.21713",
+      "title": "Fuzzing the PHP Interpreter via Dataflow Fusion",
+      "authors": [
+        "Yuancheng Jiang",
+        "Chuqi Zhang",
+        "Bonan Ruan",
+        "Jiahao Liu",
+        "Manuel Rigger",
+        "Roland H. C. Yap",
+        "Zhenkai Liang"
+      ],
+      "year": 2024,
+      "venue": "USENIX Security Symposium",
+      "doi": null,
+      "arxiv_id": "2410.21713",
+      "s2_paper_id": "5e77376fe7b120de1e368d730f15f12e3e599435",
+      "url": "https://arxiv.org/abs/2410.21713",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2410.21713",
+              "source_type": "paper_citation_context",
+              "excerpt": "This approach aligns with standard methodologies for evaluating the effectiveness of automatic bug-finding tools [19, 39].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b8225c03456de762dae8a74b71c99acae2f3d33437234369c52cda25a1576126",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2410.21713",
+              "source_type": "paper_citation_context",
+              "excerpt": "Code coverage is a widely used metric for evaluating the effectiveness of fuzzing approaches [38,39].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7b874b1ab115520f5f2b60165dc1ef78431a29488cf8b6c49b611c33e0d48bbe",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/5e77376fe7b120de1e368d730f15f12e3e599435",
+        "source_type": "paper",
+        "content_sha256": "sha256:be808bf45d8616e2920d2b9b8adfa29d90e84196a0d506f2e1372a431e81dba6"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "PHP, a dominant scripting language in web development, powers a vast range of websites, from personal blogs to major platforms. While existing research primarily focuses on PHP application-level security issues like code injection, memory errors within the PHP interpreter have been largely overlooked. These memory errors, prevalent due to the PHP interpreter's extensive C codebase, pose significant risks to the confidentiality, integrity, and availability of PHP servers. This paper introduces FlowFusion, the first automatic fuzzing framework to detect memory errors in the PHP interpreter. FlowFusion leverages dataflow as an efficient representation of test cases maintained by PHP developers, merging two or more test cases to produce fused test cases with more complex code semantics. Moreover, FlowFusion employs strategies such as test mutation, interface fuzzing, and environment crossover to increase bug finding. In our evaluation, FlowFusion found 158 unknown bugs in the PHP interpreter, with 125 fixed and 11 confirmed. Comparing FlowFusion against the official test suite and a naive test concatenation approach, FlowFusion can detect new bugs that these methods miss, while also achieving greater code coverage. FlowFusion also outperformed state-of-the-art fuzzers AFL++ and Polyglot, covering 24% more lines of code after 24 hours of fuzzing. FlowFusion has gained wide recognition among PHP developers and is now integrated into the official PHP toolchain."
+    },
+    {
+      "id": "paper:doi:10.1145/3654991",
+      "title": "Keep It Simple: Testing Databases via Differential Query Plans",
+      "authors": [
+        "Jinsheng Ba",
+        "Manuel Rigger"
+      ],
+      "year": 2024,
+      "venue": "Proc. ACM Manag. Data",
+      "doi": "10.1145/3654991",
+      "arxiv_id": null,
+      "s2_paper_id": "63b32cbfc5e6ab0043715a9adcf6ad1944c9ba49",
+      "url": "https://doi.org/10.1145/3654991",
+      "open_access_pdf": "https://doi.org/10.1145/3654991",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3654991",
+              "source_type": "paper_citation_context",
+              "excerpt": "Synthesis ( PQS ) [42], which is not supported for the three evaluated DBMSs in SQLancer .",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c5dc40c648314606b6ec38c460880eb0260500c2e66f2480075a5580a1b3c16c",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/63b32cbfc5e6ab0043715a9adcf6ad1944c9ba49",
+        "source_type": "paper",
+        "content_sha256": "sha256:e6f92d57470835b0870d5648776cdaea64e76464206b0c7976b6f5e3067f8c1f"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Query optimizers perform various optimizations, many of which have been proposed to optimize joins. It is pivotal that these optimizations are correct, meaning that they should be extensively tested. Besides manually written tests, automated testing approaches have gained broad adoption. Such approaches semi-randomly generate databases and queries. More importantly, they provide a so-called test oracle that can deduce whether the system's result is correct. Recently, researchers have proposed a novel testing approach called Transformed Query Synthesis (TQS) specifically designed to find logic bugs in join optimizations. TQS is a sophisticated approach that splits a given input table into several sub-tables and validates the results of the queries that join these sub-tables by retrieving the given table. We studied TQS's bug reports, and found that 14 of 15 unique bugs were reported by showing discrepancies in executing the same query with different query plans. Therefore, in this work, we propose a simple alternative approach to TQS. Our approach enforces different query plans for the same query and validates that the results are consistent. We refer to this approach as Differential Query Plan (DQP) testing. DQP can reproduce 14 of the 15 unique bugs found by TQS, and found 26 previously unknown and unique bugs. These results demonstrate that a simple approach with limited novelty can be as effective as a complex, conceptually appealing approach. Additionally, DQP is complementary to other testing approaches for finding logic bugs. 81% of the logic bugs found by DQP cannot be found by NoREC and TLP, whereas DQP overlooked 86% of the bugs found by NoREC and TLP. We hope that the practicality of our approach---we implemented in less than 100 lines of code per system---will lead to its wide adoption."
+    },
+    {
+      "id": "paper:s2:07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85",
+      "title": "Leopard: A General Test Suite for Isolation Level Verification",
+      "authors": [
+        "Peiyuan Liu",
+        "Siyang Weng",
+        "Keqiang Li",
+        "Lyu Ni",
+        "Chengcheng Yang",
+        "Rong Zhang",
+        "Weining Qian",
+        "Dian Qiao"
+      ],
+      "year": 2024,
+      "venue": "Conference on Innovative Data Systems Research",
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85",
+      "url": "https://www.semanticscholar.org/paper/07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS [15] and TQS [20] belong to this class of work.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:546a9e1100c59fccde6075a11110558508ecdb4c69036645c70c5437e8e51368",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/07f4bfcce4dd87d401b70baaa9aa6f9ef5d0ed85",
+        "source_type": "paper",
+        "content_sha256": "sha256:2fdfd360de9ed5a76f360ea553c610c36d88ff4d7a51d061ab1058656546cebc"
+      },
+      "abstract": null
+    },
+    {
+      "id": "paper:doi:10.1145/3643781",
+      "title": "Metamorphic Testing of Secure Multi-party Computation (MPC) Compilers",
+      "authors": [
+        "Yichen Li",
+        "Dongwei Xiao",
+        "Zhibo Liu",
+        "Qi Pang",
+        "Shuai Wang"
+      ],
+      "year": 2024,
+      "venue": "Proc. ACM Softw. Eng.",
+      "doi": "10.1145/3643781",
+      "arxiv_id": null,
+      "s2_paper_id": "1fccf0ed3dbbaeedaf7e587b93bbe993ae638ed7",
+      "url": "https://doi.org/10.1145/3643781",
+      "open_access_pdf": "https://doi.org/10.1145/3643781",
+      "abstract": "The demanding need to perform privacy-preserving computations among multiple data owners has led to the prosperous development of secure multi-party computation (MPC) protocols. MPC offers protocols for parties to jointly compute a function over their inputs while keeping those inputs private. To date, MPC has been widely adopted in various real-world, privacy-sensitive sectors, such as healthcare and finance. Moreover, to ease the adoption of MPC, industrial and academic MPC compilers have been developed to automatically translate programs describing arbitrary MPC procedures into low-level MPC executables. Compiling high-level descriptions into high-efficiency MPC executables is challenging: the compilation often involves converting high-level languages into several intermediate representations (IR), e.g., arithmetic or boolean circuits, optimizing the computation/communication cost, and picking proper MPC protocols (and underlying virtual machines) for a particular task and threat model. Various optimizations and heuristics are employed during the compilation procedure to improve the efficiency of the generated MPC executables. Despite the prosperous adoption of MPC compilers by industrial vendors and academia, a principled and systematic understanding of the correctness of MPC compilers does not yet exist. To fill this critical gap, this paper introduces MT-MPC, a metamorphic testing (MT) framework specifically designed for MPC compilers to effectively uncover erroneous compilations. Our approach proposes three metamorphic relations (MRs) that are tailored for MPC programs to mutate high-level MPC programs (compiler inputs). We then examine if MPC compilers yield semantics-equivalent MPC executables regarding the original and mutated MPC programs by comparing their execution results. Real-world MPC compilers exhibit a high level of engineering quality. Nevertheless, we detected 4,772 inputs that can result in erroneous compilations in three popular MPC compilers available on the market. While the discovered error-triggering inputs do not cause the MPC compilers to crash directly, they can lead to the generation of incorrect MPC executables, jeopardizing the underlying dependability of the computation. With substantial manual effort and help from the MPC compiler developers, we uncovered thirteen bugs in these MPC compilers by debugging them using the error-triggering inputs. Our proposed testing frameworks and findings can be used to guide developers in their efforts to improve MPC compilers.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3643781",
+              "source_type": "paper_citation_context",
+              "excerpt": "MT has been applied to a large variety of software systems, including data-base [37, 38], AI models [29, 44, 53, 54], and Cyber Physical Systems [13].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:80fe09b6035a3620779e147f8e086dc4c499f2bd8e1ea1a40068fc7c382684ed",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/1fccf0ed3dbbaeedaf7e587b93bbe993ae638ed7",
+        "source_type": "paper",
+        "content_sha256": "sha256:0f362ee0da9f783415cd607ab63fd9d04bda86d3757e5d15dc3524219181326f"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3597503.3639112",
+      "title": "Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation",
+      "authors": [
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Jingzhou Fu",
+        "Mingzhe Wang",
+        "Chengnian Sun",
+        "Yu Jiang"
+      ],
+      "year": 2024,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1145/3597503.3639112",
+      "arxiv_id": null,
+      "s2_paper_id": "53f3c597ee4fe41383364fdaf42b13973ee1cb39",
+      "url": "https://doi.org/10.1145/3597503.3639112",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597503.3639112",
+      "abstract": "Testing database management systems (DBMSs) is a complex task. Traditional approaches, such as metamorphic testing, need a precise comprehension of the SQL specification to create diverse inputs with equivalent semantics. The vagueness and intricacy of the SQL specification make it challenging to accurately model query semantics, thereby posing difficulties in testing the correctness and performance of DBMSs. To address this, we propose Mozi, a framework that finds DBMS bugs via configuration-based equivalent transformation. The key idea behind Mozi is to compare the results of equivalent DBMSs with different configurations, rather than between semantically equivalent queries. The framework involves analyzing the query plan, changing configurations to transform the DBMS to an equivalent one, and re-executing the query to compare the results using various test oracles. For example, detecting differences in query results indicates correctness bugs, while observing faster execution times on the optimization-closed DBMS suggests performance bugs. We demonstrate the effectiveness of Mozi by evaluating it on four widely used DBMSs, namely MySQL, MariaDB, Clickhouse, and PostgreSQL. In the continuous testing, Mozi found a total of 101 previously unknown bugs, including 49 correctness and 52 performance bugs in four DBMSs. Among them, 90 bugs are confirmed and 57 bugs have been fixed. In addition, Mozi can be extended to other DBMS fuzzers for testing various types of bugs. With Mozi, testing DBMSs becomes simpler and more effective, potentially saving time and effort that would otherwise be spent on precisely modeling SQL specifications for testing purposes.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639112",
+              "source_type": "paper_citation_context",
+              "excerpt": "In 24-hour experiments, Mozi covers 64,973, 54,464, 43,236, 28,499, and 14,084 more branches, detects 25, 22, 21, 24, and 26 more bugs than PQS [48], NoREC [46], TLP 1 [47], Apollo [25], and Amoeba [33], respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:24b94a330069c0086c698b7a7b6dcb219c54ab518c25d24b88cfd8f6e9ac65df",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639112",
+              "source_type": "paper_citation_context",
+              "excerpt": "Metamorphic testing is a prevalent approach to testing DBMS, which focuses on building metamorphic relations [5, 6, 32, 46, 47, 49].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8196f9eeaac3e46a58e2825ee4dbbbd2530df3c955d10a7d8954011fd4fdd9a2",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3597503_3639112.json",
+              "source_type": "paper",
+              "excerpt": "We also used SQLancer [ 45] (with test oracles PQS [ 48], TLP [ 47], and NoREC [ 46]), Apollo [25], and Amoeba [33] to test these DBMSs, but they can only find a subset of these bugs (shown in Section 5.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 5.2 DBMS Vulnerability Detection, page 7.",
+              "content_sha256": "sha256:6088eabb9bfa5b848dea1684e33e7e83d567d4b3ed04fbd903f4fbc6b45e0486",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3597503_3639112.json",
+              "source_type": "paper",
+              "excerpt": "Specifically, we compared Mozi𝑐𝑜𝑟against SQLancer using three logic test oracles, namely PQS [ 48], NoREC [ 46], and TLP [ 47] for correctness bug detection.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 5.3 Comparison with Other Techniques, page 8.",
+              "content_sha256": "sha256:af3c9a0faa9abfa2a5f5a03cc0286ee16d2cc06661099475b0312de752f2999b",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3597503_3639112.json",
+              "source_type": "paper",
+              "excerpt": "Compared to PQS, NoREC, and TLP, Mozi𝑐𝑜𝑟 found 25, 22, and 21 more bugs, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, 5.3 Comparison with Other Techniques, page 8.",
+              "content_sha256": "sha256:21acdb2f475e04e205600c1a37973cc3602a538530f7adef057c7213eb61481d",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/53f3c597ee4fe41383364fdaf42b13973ee1cb39",
+        "source_type": "paper",
+        "content_sha256": "sha256:af6380ad8e6000f1c8a09bacf4d84afea31e715f40350d91dffd93627ebff48c"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/ictai62512.2024.00085",
+      "title": "NNTailor: A Neural Network-Driven Fuzzer for DataBase Management Systems",
+      "authors": [
+        "Shutao Chu",
+        "Yongjun Wang",
+        "Haoran Xu",
+        "Zhiyuan Jiang",
+        "Yongxin Chen"
+      ],
+      "year": 2024,
+      "venue": "IEEE International Conference on Tools with Artificial Intelligence",
+      "doi": "10.1109/ictai62512.2024.00085",
+      "arxiv_id": null,
+      "s2_paper_id": "4577684b2119f110b0704f21225e1212fa4b33dd",
+      "url": "https://doi.org/10.1109/ictai62512.2024.00085",
+      "open_access_pdf": null,
+      "abstract": "DataBase Management Systems (DBMS) are essential software for efficient data storage, management, and analysis, playing a crucial role in modern data-intensive applications. Vulnerabilities in DBMS can significantly threaten data security and application functionality, impacting millions of software systems. While fuzz testing(fuzzing) is a prevalent technique for uncovering DBMS vulnerabilities, there is limited research on employing Neural Network Language Models (NNLMs) for this purpose. This paper introduces NNTailor, a novel approach using NNLMs based on AST code fragments for DBMS fuzzing. A key advantage of this method is its effectiveness in black-box testing scenarios. We evaluated NNTailor on SQLite and PostgreSQL, demonstrating its capability to generate effective SQL test cases.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/ictai62512.2024.00085",
+              "source_type": "paper_citation_context",
+              "excerpt": "We conducted evaluations of NNTailor on the latest versions of PostgreSQL and SQLite, comparing its performance with SQLsmith[5] and SQLancer[6, 25, 26, 27].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:94b8465e7bf472a0e3abc13a557717058d95be96e631ad9f35b7a9e0035c4259",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_ictai62512_2024_00085.json",
+              "source_type": "paper",
+              "excerpt": "We conducted evaluations of NNTailor on the latest versions of PostgreSQL and SQLite, comparing its performance with SQLsmith[5] and SQLancer[6, 25, 26, 27].",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, I INTRODUCTION, page 2.",
+              "content_sha256": "sha256:94b8465e7bf472a0e3abc13a557717058d95be96e631ad9f35b7a9e0035c4259",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_ictai62512_2024_00085.json",
+              "source_type": "paper",
+              "excerpt": "With the neural network model-driven mutation approach, NNTailor achieves 35% and 10% higher code line coverage compared to SQLsmith and SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, I INTRODUCTION, page 2.",
+              "content_sha256": "sha256:db22d51ab205f8a858bf9043decb3ba603bba923c96c6b6c319b7601f95fbd82",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_ictai62512_2024_00085.json",
+              "source_type": "paper",
+              "excerpt": "EVALUATION This section compares NNTailor with typical DBMS blackbox fuzzing tools SQLsmith and SQLancer in terms of coverage on the target DBMS, syntactic and semantic correctness of synthetic queries.",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, G Test, page 6.",
+              "content_sha256": "sha256:38d769394cd816e18eea1274f522c8827171d7d5831b1d817a59dc12155d49a2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/4577684b2119f110b0704f21225e1212fa4b33dd",
+        "source_type": "paper",
+        "content_sha256": "sha256:5e5015899b67461df54c789cbcdb313f0b725f9107e76347f0772562ddb7b3bf"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3689757",
+      "title": "PolyJuice: Detecting Mis-compilation Bugs in Tensor Compilers with Equality Saturation Based Rewriting",
+      "authors": [
+        "Chijin Zhou",
+        "Bingzhou Qian",
+        "Gwihwan Go",
+        "Quan Zhang",
+        "Shanshan Li",
+        "Yu Jiang"
+      ],
+      "year": 2024,
+      "venue": "Proc. ACM Program. Lang.",
+      "doi": "10.1145/3689757",
+      "arxiv_id": null,
+      "s2_paper_id": "9889059fce1930bbb29d7cbcc33e5fc1ed168f2a",
+      "url": "https://doi.org/10.1145/3689757",
+      "open_access_pdf": "https://doi.org/10.1145/3689757",
+      "abstract": "Tensor compilers are essential for deploying deep learning applications across various hardware platforms. While powerful, they are inherently complex and present significant challenges in ensuring correctness. This paper introduces PolyJuice, an automatic detection tool for identifying mis-compilation bugs in tensor compilers. Its basic idea is to construct semantically-equivalent computation graphs to validate the correctness of tensor compilers. The main challenge is to construct equivalent graphs capable of efficiently exploring the diverse optimization logic during compilation. We approach it from two dimensions. First, we propose arithmetic and structural equivalent rewrite rules to modify the dataflow of a tensor program. Second, we design an efficient equality saturation based rewriting framework to identify the most simplified and the most complex equivalent computation graphs for an input graph. After that, the outcome computation graphs have different dataflow and will likely experience different optimization processes during compilation. We applied it to five well-tested industrial tensor compilers, namely PyTorch Inductor, OnnxRuntime, TVM, TensorRT, and XLA, as well as two well-maintained academic tensor compilers, EinNet and Hidet. In total, PolyJuice detected 84 non-crash mis-compilation bugs, out of which 49 were confirmed with 20 fixed.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/9889059fce1930bbb29d7cbcc33e5fc1ed168f2a",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/9889059fce1930bbb29d7cbcc33e5fc1ed168f2a",
+        "source_type": "paper",
+        "content_sha256": "sha256:fdf5b6666a4f426616af09673ccbe9bc366a314be402fa568705c72ad7287184"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3597503.3639210",
+      "title": "Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer",
+      "authors": [
+        "Jingzhou Fu",
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Yu Jiang"
+      ],
+      "year": 2024,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1145/3597503.3639210",
+      "arxiv_id": null,
+      "s2_paper_id": "368ae600b0d2ac9e8e228f2c84e32a270403d7fa",
+      "url": "https://doi.org/10.1145/3597503.3639210",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597503.3639210",
+      "abstract": "Effective DBMS fuzzing relies on high-quality initial seeds, which serve as the starting point for mutation. These initial seeds should incorporate various DBMS features to explore the state space thoroughly. While built-in test cases are typically used as initial seeds, many DBMSs lack comprehensive test cases, making it difficult to apply state-of-the-art fuzzing techniques directly. To address this, we propose Sedar which produces initial seeds for a target DBMS by transferring test cases from other DBMSs. The underlying insight is that many DBMSs share similar functionalities, allowing seeds that cover deep execution paths in one DBMS to be adapted for other DBMSs. The challenge lies in converting these seeds to a format supported by the grammar of the target database. Sedar follows a three-step process to generate seeds. First, it executes existing SQL test cases within the DBMS they were designed for and captures the schema information during execution. Second, it utilizes large language models (LLMs) along with the captured schema information to guide the generation of new test cases based on the responses from the LLM. Lastly, to ensure that the test cases can be properly parsed and mutated by fuzzers, Sedar temporarily comments out unparsable sections for the fuzzers and uncomments them after mutation. We integrate Sedar into the DBMS fuzzers SQUIRREL and Griffin, targeting DBMSs such as Virtuoso, Mon-et DB, DuckDB, and ClickHouse. Evaluation results demonstrate significant improvements in both fuzzers. Specifically, compared to SQUIRREL and Griffin with non-transferred seeds, Sedar enhances code coverage by 72.46%-214.84% and 21.40%-194.46%; compared to SQUIRREL and Griffin with native test cases of these DBMSs as initial seeds, incorporating the transferred seeds of Sedar results in an improvement in code coverage by 4.90%-16.20% and 9.73%-28.41 %. Moreover, Sedar discovered 70 new vulnerabilities, with 60 out of them being uniquely found by Sedar with transferred seeds, and 19 of them have been assigned with CVEs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639210",
+              "source_type": "paper_citation_context",
+              "excerpt": "Typically, DBMSs adhere to basic features of ANSI SQL standard [1] for common uses, while they support more advanced features with their unique SQL dialects.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5f791c2b4eb49b8f3e6cefae0070417f7a7e2acece69a6308d83efadbf28779c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639210",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS [32] generates queries that should fetch a specific row, and indicates a bug triggered when the row is not included in result sets.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4f3b5dee76b3b0479059e077044d630bc270009074e765214f9cc2737158fb15",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639210",
+              "source_type": "paper_citation_context",
+              "excerpt": "Some generation-based approaches [15, 22, 30–32, 34, 37] are proposed to find specific types of DBMS bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:59e961b9328c1ca8a95b8556734be11bd2593c0ce1afa5e6ff087b5dc6cbae2e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/368ae600b0d2ac9e8e228f2c84e32a270403d7fa",
+        "source_type": "paper",
+        "content_sha256": "sha256:118431b703edb919d19ff9aaa0cc7791b68c66939256eaaf034bf2f70599e579"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3712221.3712247",
+      "title": "Semantic Conformance Testing of Relational DBMS",
+      "authors": [
+        "Shuang Liu",
+        "Chenglin Tian",
+        "Jun Sun",
+        "Ruifeng Wang",
+        "Wei Lu",
+        "Yongxin Zhao",
+        "Yinxing Xue",
+        "Junjie Wang",
+        "Xiaoyong Du"
+      ],
+      "year": 2024,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3712221.3712247",
+      "arxiv_id": "2406.09469",
+      "s2_paper_id": "bb1da01b4752d31f2f96f3abcc5502b58bedbc4e",
+      "url": "https://doi.org/10.14778/3712221.3712247",
+      "open_access_pdf": "https://ink.library.smu.edu.sg/sis_research/11145",
+      "abstract": "Relational DBMS implementations are expected to adhere to SQL standards. However, there are currently no tools available that can automatically verify this conformance. The main reasons are twofold. First, the SQL standard specification, documented in natural language, tends to be ambiguous and is not directly executable. Second, it is difficult to generate test queries that thoroughly cover all aspects, e.g., keywords and parameters, defined in the SQL specification. In this work, we introduce the first method for semantic conformance testing of RDBMSs. Our contributions are threefold. Firstly, we formally define the denotational semantics of SQL and implement them in Prolog, creating an executable reference RDBMS for differential testing against existing RDBMSs. Secondly, we propose three coverage criteria based on these formal semantics, along with a coverage-guided query generation algorithm that effectively generates queries achieving high semantic coverage. Lastly, we apply our approach to six widely-used and thoroughly tested RDBMSs, e.g., MySQL, PostgreSQL and OceanBase, uncovering 19 bugs and 13 inconsistencies, all of which are confirmed by RDBMS developers.",
+      "cites_seed_techniques": [
+        "norec",
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/bb1da01b4752d31f2f96f3abcc5502b58bedbc4e",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2406.09469",
+              "source_type": "paper_citation_context",
+              "excerpt": "We compared S EM C ON T with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS. NoREC constructs two semantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1309537162ec2059e0ab4b5ad53fbf5edf526bb20701eb649d14cc8176729c39",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2406.09469",
+              "source_type": "paper_citation_context",
+              "excerpt": "We compare S EM C ON T with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e245e98ca59fcd54e1753af55c925c6c8bd37c1bf1bbe1809838a7c43ecd2a3b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2406.09469",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:324349d8d8ed5d71c222303455fa1d44e0e274bbb4e2e65e9526ba22cf6cdcb5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2406.09469",
+              "source_type": "paper_citation_context",
+              "excerpt": "It is noteworthy that all four databases have been extensively tested by existing methods [40]–[42], yet S EM C ON T is still able to detect bugs that were not detected by those approaches.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4ff41c672e225fe5e31dcc71fc295b95115879184457379f466c615ea3267386",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2406.09469",
+              "source_type": "paper_citation_context",
+              "excerpt": "In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9ae3c84772a0d6c5a9b3b1177a7661fa2995340bf41a12c37eddf47dd9857e5b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2406.09469",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:77c13bf8ab54a68ac3a07bfbea6a6d39868c06fd4916afa52fe695dd2a050f1a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14778_3712221_3712247.json",
+              "source_type": "paper",
+              "excerpt": "Here, we 10 enhance the syntax-guided generation method, SQLancer [14], with coverage-guided test case generation.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, A Overview of our approach, page 9.",
+              "content_sha256": "sha256:2f18098189c2c38c23f81e6d48c3898525afa36d53a446fd7121228fb2695d0a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14778_3712221_3712247.json",
+              "source_type": "paper",
+              "excerpt": "We adopt SQLancer to randomly generatea large number of SQL statements (line 2), which serves as the seed pool of our query generation algorithm.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, B Coverage guided test case generation, page 11.",
+              "content_sha256": "sha256:ed5c12266d77de254d586aa07a0efece217f1e2f617569c2951fc2d757355d77",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14778_3712221_3712247.json",
+              "source_type": "paper",
+              "excerpt": "Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34].",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, A Experiment setup, page 14.",
+              "content_sha256": "sha256:40c5ed4334d4d124c5dc6c995c7635b838bc484f0f4d9a5adf7f24f57c1c13cd",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp",
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14778_3712221_3712247.json",
+              "source_type": "paper",
+              "excerpt": "We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, A Experiment setup, page 14.",
+              "content_sha256": "sha256:5d402785117e5d67d1f2f57b69ce550e0f5c802f7434b7d6ab923ec784218c53",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14778_3712221_3712247.json",
+              "source_type": "paper",
+              "excerpt": "Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34].",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, A Experiment setup, page 14.",
+              "content_sha256": "sha256:40c5ed4334d4d124c5dc6c995c7635b838bc484f0f4d9a5adf7f24f57c1c13cd",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14778_3712221_3712247.json",
+              "source_type": "paper",
+              "excerpt": ", NoREC (SQLancer), NoREC (SQLRight), TLP (SQLancer) and TLP (SQLRight).",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, A Experiment setup, page 14.",
+              "content_sha256": "sha256:5cb42bc1ca06dcb065882fae160ff4dcaee84998c5bdc7ecf7e3b676a0ef31f1",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14778_3712221_3712247.json",
+              "source_type": "paper",
+              "excerpt": "Coverage guided test case generation The state-of-the-art practice in test case generation involves randomly generating SQL queries guided by the syntax of SQL, among which SQLancer [14] stands out as one of the most effective tools of this kind.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, B Coverage guided test case generation, page 10.",
+              "content_sha256": "sha256:cb9972a9ac6383e234b60b1bb34442bb434190510e49dc8ca5da9efb1afd8a50",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_14778_3712221_3712247.json",
+              "source_type": "paper",
+              "excerpt": "We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, A Experiment setup, page 14.",
+              "content_sha256": "sha256:5d402785117e5d67d1f2f57b69ce550e0f5c802f7434b7d6ab923ec784218c53",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/bb1da01b4752d31f2f96f3abcc5502b58bedbc4e",
+        "source_type": "paper",
+        "content_sha256": "sha256:fc17db4b4fd3f198405d1243906fb79f6de070f81b3cb0940ab76ccbcb496c2a"
+      },
+      "also_indexed_as": [
+        "paper:arxiv:2406.09469"
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3708533",
+      "title": "Software Security Analysis in 2030 and Beyond: A Research Roadmap",
+      "authors": [
+        "Marcel Böhme",
+        "Eric Bodden",
+        "T. Bultan",
+        "Cristian Cadar",
+        "Yang Liu",
+        "Giuseppe Scanniello"
+      ],
+      "year": 2024,
+      "venue": "ACM Transactions on Software Engineering and Methodology",
+      "doi": "10.1145/3708533",
+      "arxiv_id": "2409.17844",
+      "s2_paper_id": "92276c9868f6488744e40c18835c76b8588a2d35",
+      "url": "https://doi.org/10.1145/3708533",
+      "open_access_pdf": "https://doi.org/10.1145/3708533",
+      "abstract": "As our lives, our businesses, and indeed our world economy become increasingly reliant on the secure operation of many interconnected software systems, the software engineering research community is faced with unprecedented research challenges, but also with exciting new opportunities. In this roadmap article, we outline our vision of software security analysis for the systems of the future. Given the recent advances in generative AI, we need new methods to assess and maximize the security of code co-written by machines. As our systems become increasingly heterogeneous, we need practical approaches that work even if some functions are automatically generated, e.g., by deep neural networks. As software systems depend evermore on the software supply chain, we need tools that scale to an entire ecosystem. What kind of vulnerabilities exist in future systems and how do we detect them? When all the shallow bugs are found, how do we discover vulnerabilities hidden deeply in the system? Assuming we cannot find all security flaws, how can we nevertheless protect our system? To answer these questions, we start our roadmap with a survey of recent advances in software security, then discuss open challenges and opportunities, and conclude with a long-term perspective for the field.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3708533",
+              "source_type": "paper_citation_context",
+              "excerpt": "Specifications have been for long proposed as a way to write correct software, but adoption has been limited, particularly in mainstream languages.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a4645d384cdc9ace69ff4659bf513678d9fc11d1b47c46ab4c030ec300d2bc1b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3708533",
+              "source_type": "paper_citation_context",
+              "excerpt": "For easy reference, each challenge or opportunity is identified uniquely using a counter and a name.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ecbb1a2eb3a1a096414aa12d4cfffdfbf95e782dfe3123812d2a4481e46af6e4",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/92276c9868f6488744e40c18835c76b8588a2d35",
+        "source_type": "paper",
+        "content_sha256": "sha256:eea3c61bed5cd8551f42e9e5458d698193b89c7276c1b2c01a7a7d6dbaf9ea8b"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3650212.3680317",
+      "title": "SQLess: Dialect-Agnostic SQL Query Simplification",
+      "authors": [
+        "Li Lin",
+        "Zongyin Hao",
+        "Chengpeng Wang",
+        "Zhuangda Wang",
+        "Rongxin Wu",
+        "Gang Fan"
+      ],
+      "year": 2024,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3650212.3680317",
+      "arxiv_id": null,
+      "s2_paper_id": "deb987f2a988b7585e7423ec2d48b5cd20f2cb6a",
+      "url": "https://doi.org/10.1145/3650212.3680317",
+      "open_access_pdf": null,
+      "abstract": "Database Management Systems (DBMSs) are fundamental to numerous enterprise applications. Due to the significance of DBMSs, various testing techniques have been proposed to detect DBMS bugs. However, to trigger deep bugs, most of the existing techniques focus on generating lengthy and complex queries which burdens developers with the difficult of debugging. Therefore, SQL query simplification, which aims to reduce lengthy SQL queries without compromising their ability to detect bugs, is highly demanded. To bridge this gap, we introduce SQLess, an innovative approach that employs a dialect-agnostic method for efficient and semantically correct SQL query simplification tailored for various DBMSs. Unlike previous works that have to depend on DBMS-specific grammar, SQLess utilizes an adaptive parser, which leverages error recovery and grammar expansion to support DBMS dialects. Moreover, SQLess performs a semantics-sensitive SQL query trimming, which leverages alias and dependency analysis to simplify SQL queries with preserving bug-triggering capability. We evaluate SQLess using two datasets from the state-of-theart database bug detection studies, encompassing six widely-used DBMSs and over 32,000 complex SQL queries. The results demonstrate SQLess’s superior performance: it achieves an average simplification rate of 72.45%, which significantly outperforms the stateof-the-art approaches by 84.91%.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3650212.3680317",
+              "source_type": "paper_citation_context",
+              "excerpt": "They typically suffer from a dilemma that the malformed, complex, and lengthy queries are more likely to trigger bugs but are less likely to be accepted by DBMS developers, which is highlighted by many prior studies [7, 9, 30].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:076f616399dad9e70a67088448ed646edb5c9c7a6f35f4c9f94474ac26ec8ce3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/deb987f2a988b7585e7423ec2d48b5cd20f2cb6a",
+        "source_type": "paper",
+        "content_sha256": "sha256:5dc77701dcc88a8062f022420ccb53f09392d9c9e546a63410496b64ff1ed097"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/compsac61105.2024.00141",
+      "title": "SQLPass: A Semantic Effective Fuzzing Method for DBMS",
+      "authors": [
+        "Yu Li",
+        "Yixiao Yang",
+        "Yong Guan",
+        "Zhiping Shi",
+        "Rui Wang"
+      ],
+      "year": 2024,
+      "venue": "Annual International Computer Software and Applications Conference",
+      "doi": "10.1109/compsac61105.2024.00141",
+      "arxiv_id": null,
+      "s2_paper_id": "83c63c4935d91200817bfc19905800431df70230",
+      "url": "https://doi.org/10.1109/compsac61105.2024.00141",
+      "open_access_pdf": null,
+      "abstract": "Fuzzing, as an effective method for software system defect testing and bug mining, utilizes random data generated by mutation to execute programs to trigger potential bugs in the tested program. However, due to the strict syntax and semantic checks in DBMSs, existing mutation-based testing methods are difficult to generate test cases with correct syntax and semantics. To ensure the syntax and semantic correctness of the test case generated by mutation, this paper proposes the concepts of weak semantic correlation nodes and semantic relationship tables. At the same time, a set of mutation operators for weak semantic correlation nodes in the syntax tree is designed, and each time the “optimal” mutation operator is selected to replace, delete, and insert the target node subtree in the test case syntax tree. In response to semantic errors during the mutation, this paper checks and corrects them through a pre-extracted semantic relationship table, further ensuring the correctness of the semantics of the test cases generated by the mutation. This paper utilizes this semantic effective fuzzing method for DBMSs to implement a new fuzzing framework for DBMSs, SQLPass. We evaluated SQLPass on four popular DBMSs: SQLite3, MySQL, MariaDB and PostgreSQL. In our experiment, SQLPass achieves 5.7%-94.2% higher semantic correctness than state-of-the-art tools, and explores 1.3%-52% more code coverage than other tools. Notably, it discovered four unknown bugs on SQLite3 and MariaDB and has submitted them to the vendor for confirmation.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/83c63c4935d91200817bfc19905800431df70230",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_compsac61105_2024_00141.json",
+              "source_type": "paper",
+              "excerpt": "b) C omparison with benchmark testing tools: We conduct 24 hours experiments on SQLPass and benchmark testing tools AFL, SQLsmith, SQLancer, and Squirrel on SQLite, MySQL, MariaDB, and PostgreSQL, respectively, and compare and evaluate three ind icators: semantic correctness, code (line) coverage, and bug replication speed.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, B Experimental Comparison, page 7.",
+              "content_sha256": "sha256:0db657a6e7008933fc3b21afdf8798309d41910b1934665a7bd105793eedd1ae",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_compsac61105_2024_00141.json",
+              "source_type": "paper",
+              "excerpt": "Therefore, the code coverage generated by the SQL test cases it generates is also not high; Although SQLancer generates limited types of SQL test cases, the syntax and semantic accuracy of the generated SQL statements are high, so it can trigger deeper code logic and generate higher code coverage; The syntax and semantic accuracy of SQL test cases generated by Squirrel mutation are not high, but the number of SQL statements that can mutate within the same time is relatively high, and the type...",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, B Experimental Comparison, page 9.",
+              "content_sha256": "sha256:b4372f15af4eed89c8f7491a9037daee46d9b96b9247e66af7a9c469a171fb0d",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/83c63c4935d91200817bfc19905800431df70230",
+        "source_type": "paper",
+        "content_sha256": "sha256:a51538bf998cc812b1adeee6e9c8029f544b7724766fff877b29db89854ae217"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3689491.3691821",
+      "title": "Step-wise Execution of Data-Centric Systems",
+      "authors": [
+        "Chi Zhang"
+      ],
+      "year": 2024,
+      "venue": "SPLASH Companion",
+      "doi": "10.1145/3689491.3691821",
+      "arxiv_id": null,
+      "s2_paper_id": "c8bb406e87172ca7ccc8f009d0554308232c01e5",
+      "url": "https://doi.org/10.1145/3689491.3691821",
+      "open_access_pdf": null,
+      "abstract": "Data-centric systems play a crucial role in computer systems, and their correctness is of paramount importance. Logic bugs are one of the main types of bugs in data-centric systems, leading to incorrect results. Existing methods for testing data-centric systems face challenges such as ineffective test oracles, insufficient coverage of target functionalities, and low efficiency in test case generation. To address these issues, we propose a general, novel black-box methodology for testing various kinds of important data-centric systems. Our key insight is that we can incrementally generate the test case and use the intermediate results to construct the reference test case. The result of the original test case should be identical to that of the reference test case. We have applied this methodology to test both Datalog engines and DBMSs, successfully uncovering 75 unique bugs. We believe that this idea might be applicable to many other systems as well, and will next explore this idea to Deep Learning libraries.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3689491.3691821",
+              "source_type": "paper_citation_context",
+              "excerpt": "DBMS. NoREC [7], TLP [8], and TQS [9] are all state-of-the-art approaches for detecting logic bugs in DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:333e9b1e7432196037786aa81cac20c3c51d7d1faa0f3c962f86d79c7937c297",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3689491_3691821.json",
+              "source_type": "paper",
+              "excerpt": "NoREC [ 7], TLP [ 8], and TQS [ 9] are all stateof-the-art approaches for detecting logic bugs in DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, 2 Problem, page 2.",
+              "content_sha256": "sha256:04c989443852a340f27c8ef10cb359ec677cdfb2dd37b2ac7447041c79146d2c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/c8bb406e87172ca7ccc8f009d0554308232c01e5",
+        "source_type": "paper",
+        "content_sha256": "sha256:1d53d00d17b542e628d28a31d40e7c69e63e5b6d678edf8f51285c8d76710582"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3690631",
+      "title": "T-Rec: Fine-Grained Language-Agnostic Program Reduction Guided by Lexical Syntax",
+      "authors": [
+        "Zhenyang Xu",
+        "Yongqiang Tian",
+        "Mengxiao Zhang",
+        "Jiarui Zhang",
+        "Puzhuo Liu",
+        "Yu Jiang",
+        "Chengnian Sun"
+      ],
+      "year": 2024,
+      "venue": "ACM Transactions on Software Engineering and Methodology",
+      "doi": "10.1145/3690631",
+      "arxiv_id": null,
+      "s2_paper_id": "c7ca7db1f2ed7eafe81c58bf6732efb95bee0146",
+      "url": "https://doi.org/10.1145/3690631",
+      "open_access_pdf": null,
+      "abstract": "Program reduction strives to eliminate bug-irrelevant code elements from a bug-triggering program, so that (1) a smaller and more straightforward bug-triggering program can be obtained, (2) and the difference among duplicates (i.e., different programs that trigger the same bug) can be minimized or even eliminated. With such reduction and canonicalization functionality, program reduction facilitates debugging for software, especially language toolchains, such as compilers, interpreters, and debuggers. While many program reduction techniques have been proposed, most of them (especially the language-agnostic ones) overlooked the potential reduction opportunities hidden within tokens. Therefore, their capabilities in terms of reduction and canonicalization are significantly restricted. To fill this gap, we propose \\(\\mathsf{T}\\) - \\(\\mathsf{Rec}\\) , a fine-grained language-agnostic program reduction technique guided by lexical syntax. Instead of treating tokens as atomic and irreducible components, \\(\\mathsf{T}\\) - \\(\\mathsf{Rec}\\) introduces a fine-grained reduction process that leverages the lexical syntax of programming languages to effectively explore the reduction opportunities in tokens. Through comprehensive evaluations with versatile benchmark suites, we demonstrate that \\(\\mathsf{T}\\) - \\(\\mathsf{Rec}\\) significantly improves the reduction and canonicalization capability of two existing language-agnostic program reducers (i.e., Perses and Vulcan). \\(\\mathsf{T}\\) - \\(\\mathsf{Rec}\\) enables Perses and Vulcan to further eliminate 1,294 and 1,315 duplicates in a benchmark suite that contains 3,796 test cases that trigger 46 unique bugs. Additionally, \\(\\mathsf{T}\\) - \\(\\mathsf{Rec}\\) can also reduce up to 65.52% and 53.73% bytes in the results of Perses and Vulcan on our multi-lingual benchmark suite, respectively.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3690631",
+              "source_type": "paper_citation_context",
+              "excerpt": "Concretely, T - Rec Perses takes 24.22%, 31.22%, and 56.46% more time than Perses on C, Rust,and SMT-LIBv2 benchmarks, respectively, and T - Rec Vulcan takes 1.59%, 5.04%, and 5.01% more time than Vulcan on each benchmark.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b859011cd19ee1aa40ba40bf9faef082562b5c95f38412777ca699825f15bd7d",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/c7ca7db1f2ed7eafe81c58bf6732efb95bee0146",
+        "source_type": "paper",
+        "content_sha256": "sha256:fed1545fbdd6799d80bbe643bc17ffd83d8fa8d97e6e6a52446508ae11f74642"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/saner60148.2024.00096",
+      "title": "Testing Constraint Checking Implementations via Principled Metamorphic Transformations",
+      "authors": [
+        "Mingchen Gao",
+        "Huiyan Wang",
+        "Chang Xu"
+      ],
+      "year": 2024,
+      "venue": null,
+      "doi": "10.1109/saner60148.2024.00096",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.1109/saner60148.2024.00096",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/W4400680628",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://openalex.org/W4400680628",
+        "source_type": "paper",
+        "content_sha256": "sha256:8b9babf5354dffa1b55632561bb44313cea63ac93a93d9a354a817e71ece53fa"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3597503.3639200",
+      "title": "Testing Graph Database Systems via Equivalent Query Rewriting",
+      "authors": [
+        "Qiuyang Mang",
+        "Aoyang Fang",
+        "Boxi Yu",
+        "Hanfei Chen",
+        "Pinjia He"
+      ],
+      "year": 2024,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1145/3597503.3639200",
+      "arxiv_id": null,
+      "s2_paper_id": "29641e542dc3bcf96e4a8e6902b9e132a9c399da",
+      "url": "https://doi.org/10.1145/3597503.3639200",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597503.3639200",
+      "abstract": "Graph Database Management Systems (GDBMS), which utilize graph models for data storage and execute queries via graph tra-versals, have seen ubiquitous usage in real-world scenarios such as recommendation systems, knowledge graphs, and social networks. Much like Relational Database Management Systems (RDBMS), GDBMS are not immune to bugs. These bugs typically manifest as logic errors that yield incorrect results (e.g., omitting a node that should be included), performance bugs (e.g., long execution time caused by redundant graph scanning), and exception issues (e.g., unexpected or missing exceptions). This paper adapts Equivalent Query Rewriting (EQR) to GDBMS testing. EQR rewrites a GDBMS query into equivalent ones that trigger distinct query plans, and checks whether they exhibit discrepancies in system behaviors. To facilitate the realization of EQR, we propose a general concept called Abstract Syntax Graph (ASG). Its core idea is to embed the semantics of a base query into the paths of a graph, which can be utilized to generate new queries with customized properties (e.g., equivalence). Given a base query, an ASG is constructed and then an equivalent query can be gener-ated by finding paths collectively carrying the complete semantics of the base query. To this end, we further design Random Walk Covering (RWC), a simple yet effective path covering algorithm. As a practical implementation of these ideas, we develop a tool GRev, which has successfully detected 22 previously unknown bugs across 5 popular GDBMS, with 15 of them being confirmed. In particular, 14 of the detected bugs are related to improper implementation of graph data retrieval in GDBMS, which is challenging to identify for existing techniques.",
+      "cites_seed_techniques": [
+        "pqs",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639200",
+              "source_type": "paper_citation_context",
+              "excerpt": "Similarly, GDBMeter [15], a tool based on Query Partitioning [32], requires different validation rules for different queries and may struggle to handle some widely-used clauses such as DISTINCT .",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:66393c6f8a6da579932165602e82b8d8aaadd262e15a2764c70392649280d3cf",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639200",
+              "source_type": "paper_citation_context",
+              "excerpt": "Specifically, GDsmith and Grand are based on Differential Testing [38] while GDBMeter is based on Query Partitioning [32] ( i.e. , partitioning query into multiple equivalent sub-queries).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c3dd6a84ff69702fe5d59cd56453fb7f84a4287d76dbe23bdea3cd2311337719",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639200",
+              "source_type": "paper_citation_context",
+              "excerpt": "Based on Query Partitioning, GDBMeter [15] leverages the test oracle called Ternary Logic Partitioning (TLP) to detect bugs in GDBMS. TLP [32] was initially introduced in testing RDBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b2913169c64ff10532e5248b132726e4723836061b7f627a203cf6cba95563a5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639200",
+              "source_type": "paper_citation_context",
+              "excerpt": "Notably, we can not compare GRev with RDBMS testing tools, such as SQLancer [33].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d35eeb25663b8dadeb2d07dbfab9f6dff1441fa8595015ee62e2fff6cc3edcaf",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639200",
+              "source_type": "paper_citation_context",
+              "excerpt": "Notably, these bugs were rarely detected by the existing techniques, especially for GDBMeter [15], which reused the test oracles designed for RDBMS ( i.e. , TLP [32]).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:31d6b3f1e6b6ab0e5260714a221ae77f17d417bcf36f7b5ceb0ec659dae4de34",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639200",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, none of the graph-related bugs have been detected by it due to reusing TLP[32], a metamorphic relation for RDBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5215e9fa107c9c1e39d686b0c25219badfeb6f455510b5a54833d058b551de43",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3597503_3639200.json",
+              "source_type": "paper",
+              "excerpt": ", GDsmith [ 13], Grand [ 44], and GDBMeter [ 15]).",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, 5.1 Experimental Setting, page 6.",
+              "content_sha256": "sha256:cd291b313765d65bbac66e2c44c60a45273bb62ee486150f317c1f41b2ef2de8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3597503_3639200.json",
+              "source_type": "paper",
+              "excerpt": ", GDsmith [ 13], Grand [ 44], and GDBMeter [ 15], that can detect bugs other than crashes.",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, 5.4 Analytical Comparison with Existing, page 9.",
+              "content_sha256": "sha256:99fad9c92c6c06f09b6102c8ac6b4060e6a2355338495f8e9d030a0c20b3dc1e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3597503_3639200.json",
+              "source_type": "paper",
+              "excerpt": "For example, as shown in Listing 5, the TLP approach fails to generate bug-triggering test cases for the graphrelated bugs shown in Listing 2.",
+              "excerpt_is_verbatim": true,
+              "note": "M18 in the paper's extracted text, 5.4 Analytical Comparison with Existing, page 9.",
+              "content_sha256": "sha256:121c7423025cf40434dff8641e241ef6d5176054798e2e9023617a052a0f7edd",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3597503_3639200.json",
+              "source_type": "paper",
+              "excerpt": "However, none of the graph-related bugs have been detected by it due to reusing TLP[ 32], a metamorphic relation for RDBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "M20 in the paper's extracted text, 7 RELATED WORK, page 11.",
+              "content_sha256": "sha256:a6ff4a2cf3edb3278c74d112343a3688026ea006213d9f911263826f7e67db3f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3597503_3639200.json",
+              "source_type": "paper",
+              "excerpt": "Based on metamorphic testing, GDBMeter [ 15] is the SOTA tool for detecting logic bugs in GDBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "M19 in the paper's extracted text, 7 RELATED WORK, page 11.",
+              "content_sha256": "sha256:4fa1e034b7bd6184ca55e21ccb07455675c5a7ddd38e653d4c3bdf0603b2ff48",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/29641e542dc3bcf96e4a8e6902b9e132a9c399da",
+        "source_type": "paper",
+        "content_sha256": "sha256:915857d5954035fc4ada4611cf9e14bb8997b3857b29148de2fc819d84f1ed3a"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3650212.3680311",
+      "title": "Testing Graph Database Systems with Graph-State Persistence Oracle",
+      "authors": [
+        "Shuang Liu",
+        "Junhao Lan",
+        "Xiaoning Du",
+        "Jiyuan Li",
+        "Wei Lu",
+        "Jiajun Jiang",
+        "Xiaoyong Du"
+      ],
+      "year": 2024,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3650212.3680311",
+      "arxiv_id": null,
+      "s2_paper_id": "ada7405e20ac861d632055bcb871ee4d57afc710",
+      "url": "https://doi.org/10.1145/3650212.3680311",
+      "open_access_pdf": null,
+      "abstract": "Graph Database Management Systems (GDBMSs) store data in a graph format, facilitating rapid querying of nodes and relationships. This structure is particularly advantageous for applications like social networks and recommendation systems, which often involve frequent writing operations—such as adding new nodes, creating relationships, or modifying existing data—that potentially introduce bugs. However, existing GDBMS testing approaches tend to overlook these writing functionalities, failing to detect bugs arising from such operations. In this paper we present GraspDB, the first metamorphic testing approach specifically designed to identify bugs related to writing operations in graph database systems. GraspDB employs the Graph-State Persistence oracle, which is based on the Labeled Property Graph Isomorphism (LPG-Isomorphism) and Labeled Property Subgraph Isomorphism (LPSG-Isomorphism) relations. We also develop three classes of mutation rules aimed at engaging more diverse writing-related code logic. GraspDB has successfully detected 77 unique, previously unknown bugs across four popular open source graph database engines, among which 58 bugs are confirmed by developers, 43 bugs have been fixed and 31 are related to writing operations.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3650212.3680311",
+              "source_type": "paper_citation_context",
+              "excerpt": "Another reason is due to the low duplicate bug report rate of GraspDB compared to baselines.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9cf6c6b5261e70a68759e39392029663fa0c2c1f6fc429f6e5e8d32591ed6234",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3650212.3680311",
+              "source_type": "paper_citation_context",
+              "excerpt": "If the DBMS fails to fetch the pivot row, it likely causes a bug in the RDBMS. Non-optimizing Reference Engine Construction (NoREC) [29] is another widely-known metamorphic testing approach to test RDBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0de7c61d065c0468ce915c550d4ad24333aa347891ddfe902ffcf6c51afacf75",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3650212_3680311.json",
+              "source_type": "paper",
+              "excerpt": "(PQS) isageneralandhighly-effectiveapproachto/f_indingbugsinDBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, 5.5 Threatsto Validity, page 11.",
+              "content_sha256": "sha256:58c69fa57b2f59c505f01d4a21ff1f1f41a400d11619a7aaf6a6bc4971bfba63",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/ada7405e20ac861d632055bcb871ee4d57afc710",
+        "source_type": "paper",
+        "content_sha256": "sha256:2e94017e42db7487caa80f57230014eb5f01ed1d61593ce77daae4280040e46c"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3650212.3680392",
+      "title": "Testing Gremlin-Based Graph Database Systems via Query Disassembling",
+      "authors": [
+        "Yingying Zheng",
+        "Wensheng Dou",
+        "Leile Tang",
+        "Ziyu Cui",
+        "Yu Gao",
+        "Jiansen Song",
+        "Liangying Xu",
+        "Jiaxin Zhu",
+        "Wei Wang",
+        "Jun Wei",
+        "Hua Zhong",
+        "Tao Huang"
+      ],
+      "year": 2024,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3650212.3680392",
+      "arxiv_id": null,
+      "s2_paper_id": "dce7daf3d485e83ae503cfa60e7a0ccce994a764",
+      "url": "https://doi.org/10.1145/3650212.3680392",
+      "open_access_pdf": "https://doi.org/10.1145/3650212.3680392",
+      "abstract": "Graph Database Systems (GDBs) support efficiently storing and retrieving graph data, and have become a critical component in many important applications. Many widely-used GDBs utilize the Gremlin query language to create, modify, and retrieve data in graph databases, in which developers can assemble a sequence of Gremlin APIs to perform a complex query. However, incorrect implementations and optimizations of GDBs can introduce logic bugs, which can cause Gremlin queries to return incorrect query results, e.g., omitting vertices in a graph database. In this paper, we propose Query Di sassembling (QuDi), an effective testing technique to automatically detect logic bugs in Gremlin-based GDBs. Given a Gremlin query Q, QuDi disassembles Q into a sequence of atomic graph traversals TList, which shares the equivalent execution semantics with Q. If the execution results of Q and TList are different, a logic bug is revealed in the target GDB. We evaluate QuDi on six popular GDBs, and have found 25 logic bugs in these GDBs, 10 of which have been confirmed as previously-unknown bugs by GDB developers.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/dce7daf3d485e83ae503cfa60e7a0ccce994a764",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP), Query Plan Guidance (QPG).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3650212_3680392.json",
+              "source_type": "paper",
+              "excerpt": ",Grand[ 64],GDsmith[ 39],RD2[62], and GDBMeter [ 42], can /f_ind bugs in GDBs.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 4.3 Comparingwith ExistingApproaches, page 9.",
+              "content_sha256": "sha256:b0334ffe6087d56411a6eeaa5611656f9bb5a41c0e17436a395cb3f017e143a2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3650212_3680392.json",
+              "source_type": "paper",
+              "excerpt": "Comparison with query partitioning.",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, 4.3 Comparingwith ExistingApproaches, page 9.",
+              "content_sha256": "sha256:b585f70e9cf8374cef5b0d813035e9dd79773d66915fc0f47e220c4671657b03",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3650212_3680392.json",
+              "source_type": "paper",
+              "excerpt": "We /f_irst verify whether query partitioning in GDBMeter can detect the 25 logic bugs detected by QuDi.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, 4.3 Comparingwith ExistingApproaches, page 9.",
+              "content_sha256": "sha256:ae8ce893e16cb1911ff3985360fa4ff708b90bfc829a3eeb4f7842655bb1693a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/dce7daf3d485e83ae503cfa60e7a0ccce994a764",
+        "source_type": "paper",
+        "content_sha256": "sha256:434121cbe37b3ed90a1df8dc3f37378f293a8421e18cb5cf6fbc13665c570285"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3704870",
+      "title": "The Decision Problem for Regular First Order Theories",
+      "authors": [
+        "Umang Mathur",
+        "David Mestel",
+        "Mahesh Viswanathan"
+      ],
+      "year": 2024,
+      "venue": "Proc. ACM Program. Lang.",
+      "doi": "10.1145/3704870",
+      "arxiv_id": "2410.17185",
+      "s2_paper_id": "f5df824295f8a8bcf107f122c26590b761e8b9f9",
+      "url": "https://doi.org/10.1145/3704870",
+      "open_access_pdf": "https://doi.org/10.1145/3704870",
+      "abstract": "The Entscheidungsproblem, or the classical decision problem, asks whether a given formula of first-order logic is satisfiable. In this work, we consider an extension of this problem to regular first-order theories, i.e., (infinite) regular sets of formulae. Building on the elegant classification of syntactic classes as decidable or undecidable for the classical decision problem, we show that some classes (specifically, the EPR and Gurevich classes), which are decidable in the classical setting, become undecidable for regular theories. On the other hand, for each of these classes, we identify a subclass that remains decidable in our setting, leaving a complete classification as a challenge for future work. Finally, we observe that our problem generalises prior work on automata-theoretic verification of uninterpreted programs and propose a semantic class of existential formulae for which the problem is decidable.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3704870",
+              "source_type": "paper_citation_context",
+              "excerpt": "Likewise, an effective test suite for testing database management systems must contain SQL queries that return different types of answers [Rigger and Su 2020].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:197de00e7813b71a312745f01bc6c41da37fec6dc7f2af3f5e38d55ac56889fd",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3704870",
+              "source_type": "paper_citation_context",
+              "excerpt": "Finally, in software testing, the goal is often to design a test suite that is diverse and exercises different paths of the program under test.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bb18ac3814f9aa4cbab6066176fcf4d696a61ab0da37f1ef69d284e21335bee0",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/f5df824295f8a8bcf107f122c26590b761e8b9f9",
+        "source_type": "paper",
+        "content_sha256": "sha256:aecaa92c11d97357afe1341fd763b6c097f139b4c98ed22772344133f9713e0d"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icde65448.2025.00245",
+      "title": "Towards a Unified Query Plan Representation",
+      "authors": [
+        "Jinsheng Ba",
+        "Manuel Rigger"
+      ],
+      "year": 2024,
+      "venue": "IEEE International Conference on Data Engineering",
+      "doi": "10.1109/icde65448.2025.00245",
+      "arxiv_id": "2408.07857",
+      "s2_paper_id": "51ce1a0743127f32e77acc6f13465d25006773ff",
+      "url": "https://doi.org/10.1109/icde65448.2025.00245",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "tlp",
+        "qpg",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp",
+            "qpg",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icde65448.2025.00245",
+              "source_type": "paper_citation_context",
+              "excerpt": "Query Plan Guidance ( QPG ) [4] guides test case generation towards diverse query plans aiming to expose more bugs, Cardinality Estimation Restriction Testing ( CERT ) [5] identifies performance bugs by examining query plans, and Mozi [6] checks the consistency across query plans of the same query…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1c5abdf2b278ba5ba4f21726406d552d92ae830cf99b3239c147b698ee02b5de",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65448.2025.00245",
+              "source_type": "paper_citation_context",
+              "excerpt": "Note that we identified this bug by the test oracle Ternary Logic Partitioning (TLP) [51]; however, for presentation, we simplified the bug-inducing test case by demonstrating that the same query returns different results in lines 4 and 6 depending on whether the index exists.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:73315f08627cbeae21ff6c4e88a2eae6a252843354b6495a5d5180bdbaedf8d6",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde65448.2025.00245",
+              "source_type": "paper_citation_context",
+              "excerpt": "We show how UPlan allows implementing the testing methods QPG [4] and CERT [5] in a DBMS-agnostic way.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0e9994b0abe72effe6949e2805358a209fe146c2a241f522ec7afded717d5315",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/51ce1a0743127f32e77acc6f13465d25006773ff",
+        "source_type": "paper",
+        "content_sha256": "sha256:071dc0428a44763d408df27d5952119c0636becf2629506d7d4c134c894070e2"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "In database systems, a query plan is a series of concrete internal steps to execute a query. Multiple testing approaches utilize query plans for finding bugs. However, query plans are represented in a database-specific manner, so implementing these testing approaches requires a non-trivial effort, hindering their adoption. We envision that a unified query plan representation can facilitate the implementation of these approaches. In this paper, we present an exploratory case study to investigate query plan representations in nine widely-used database systems. Our study shows that query plan representations consist of three conceptual components: operations, properties, and formats, which enable us to design a unified query plan representation. Based on it, existing testing methods can be efficiently adopted, finding 17 previously unknown and unique bugs. Additionally, the unified query plan representation can facilitate other applications. Existing visualization tools can support multiple database systems based on the unified query plan representation with moderate implementation effort, and comparing unified query plans across database systems provides actionable insights to improve their performance. We expect that the unified query plan representation will enable the exploration of additional application scenarios."
+    },
+    {
+      "id": "paper:s2:d933042eb3c2a8f2e208515490d6b6a400e00fec",
+      "title": "Towards Generic Database Management System Fuzzing",
+      "authors": [
+        "Yupeng Yang",
+        "Yongheng Chen",
+        "Rui Zhong",
+        "Jizhou Chen",
+        "Wenke Lee"
+      ],
+      "year": 2024,
+      "venue": "USENIX Security Symposium",
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "d933042eb3c2a8f2e208515490d6b6a400e00fec",
+      "url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+              "source_type": "paper_citation_context",
+              "excerpt": "For relational DBMSs, we compare B UZZ B EE with S QUIRREL [56] and SQL ANCER (PQS [43]), two DBMS fuzzers specialized in SQL DBMS fuzzing.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5cb6f31cbe003314d0fa329a43264eec3c01a016f3598f2b2ac04aae34bd04ea",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+              "source_type": "paper_citation_context",
+              "excerpt": "Fuzzing frameworks and research related to relational DBMSs [27, 28, 43, 44, 49, 56] have been developed and advanced extensively over the years, contributing to more secure and trustworthy systems in the relational DBMS venue.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8d4f8cd2d4f31d4aa41516d1bb84e4fb1162c28e57bfb3622769ffb3038eebdf",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+              "source_type": "paper_citation_context",
+              "excerpt": "Tools like SQLancer [42], SQLsmith [44], and Squirrel [56] have emerged to test relational DBMSs. SQLsmith generates random SQL queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e9d630ee8b9073b26e8d896d31a39803881cbc1ccdaedf484f540dd8cb6129c1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+              "source_type": "paper_citation_context",
+              "excerpt": "Current research has made significant advancements in relational DBMS testing [27, 28, 43, 44, 49, 56].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f44ecb0a8e8bb24c87e8f99e3938f91bb75e9a6e731ac2e680e65e3f95391413",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+              "source_type": "paper_citation_context",
+              "excerpt": "Fuzzing DBMSs has been an active research area in recent years [15, 25–27, 43, 44, 49, 56].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4c3289912692beb49c50317f19ba7cccc32f3d434adadbbb57aaed24c420e8fd",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+              "source_type": "paper_citation_context",
+              "excerpt": "G LOT [9], Grammarinator [22], and SQL-specialized S QUIR - REL [56] and SQL ANCER [43].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e9f6af00ccaf531f2bb97814c75c3a34c1d556665e5d805e8d569e112a00a16c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json",
+              "source_type": "paper",
+              "excerpt": "We compare BUZZBEEwith general-purpose fuzzers AFL++ [ 14],REDQUEEN [4], syntax-aware fuzzers POLYGLOT [9], Grammarinator [ 22], and SQL-specialized SQUIR REL [56] and SQLANCER [43].",
+              "excerpt_is_verbatim": true,
+              "note": "M11 in the paper's extracted text, 8.5 Comparison with Existing Tools, page 13.",
+              "content_sha256": "sha256:899b2ca79176c3a98cc3a767be3d25ae04b12e99cb57d56e363fedfac82d7747",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json",
+              "source_type": "paper",
+              "excerpt": "For relational DBMSs, we compare BUZZBEEwith SQUIRREL [56] and SQL ANCER (PQS [ 43]), two DBMS fuzzers specialized in SQL DBMS fuzzing.",
+              "excerpt_is_verbatim": true,
+              "note": "M13 in the paper's extracted text, 8.5 Comparison with Existing Tools, page 14.",
+              "content_sha256": "sha256:b0d3742f44c29059c9732bd6ebc05eccea1fbcae24bd14a4bf61e49e1998e4ac",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json",
+              "source_type": "paper",
+              "excerpt": "We also evaluate SQLANCER on ArangoDB because SQLANCER recently adds support for it.",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, 8.5 Comparison with Existing Tools, page 14.",
+              "content_sha256": "sha256:a8af2159f8d0dc36ad84896025fee05b62270cc60cc8c3ca84c70134df574ddc",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json",
+              "source_type": "paper",
+              "excerpt": "None of BUZZBEE,SQUIRREL, and SQLANCER can find bugs in the latest version of PostgreSQL within 24 hours.",
+              "excerpt_is_verbatim": true,
+              "note": "M16 in the paper's extracted text, 8.5 Comparison with Existing Tools, page 14.",
+              "content_sha256": "sha256:ff46e107b9f9e2f4f4517f4aaee6f10176261bf2a231ebc3a92ecd16d10a041f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+        "source_type": "paper",
+        "content_sha256": "sha256:9823e371bf30001f828141daca2c4231063c0412b7583adf4689e4bda8384aab"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icde60146.2024.00011",
+      "title": "TRAP: Tailored Robustness Assessment for Index Advisors via Adversarial Perturbation",
+      "authors": [
+        "Wei Zhou",
+        "Chen Lin",
+        "Xuanhe Zhou",
+        "Guoliang Li",
+        "Tianqing Wang"
+      ],
+      "year": 2024,
+      "venue": "IEEE International Conference on Data Engineering",
+      "doi": "10.1109/icde60146.2024.00011",
+      "arxiv_id": null,
+      "s2_paper_id": "83899adebf94722befbec210a795d4b8d0db5b92",
+      "url": "https://doi.org/10.1109/icde60146.2024.00011",
+      "open_access_pdf": null,
+      "abstract": "Many index advisors have recently been proposed to build indexes automatically to improve query performance. However, they mainly consider performance improvement in static scenarios. Their robustness, i.e., stable performance in dynamic scenarios (e.g., with minor workload changes), has not been well investigated. This paper addresses the challenges of assessing the index advisor's robustness from the following aspects. First, we introduce perturbation-based workloads for robustness assessment and identify three typical perturbation constraints that occur in real scenarios. Second, with the perturbation constraints, we formulate the generation of perturbed queries as a sequence-to-sequence problem and propose TRAP (Tailored Robustness assessment via Adversarial Perturbation) to pinpoint the performance loopholes of index advisors. Third, to generalize to various index advisors, we place TRAP in an opaque-box setting (i.e., with little knowledge of the index advisors' internal design), and we propose a two-phase training paradigm to efficiently train TRAP without elaborately annotated data. Fourth, we conduct comprehensive robustness assessments on standard benchmarks and real workloads for ten existing index advisors. Our findings reveal that these index advisors are vulnerable to the workloads generated by TRAP. Finally, based on the assessment results, we shed light on insights to enhance the robustness of different index advisors. For example, learning-based index advisors can benefit from adopting a fine-grained state representation and a candidate pruning strategy.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icde60146.2024.00011",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [36] derives multiple queries by partitioning the results from the original query.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a70c6fe85d7b5e1201fad2556abce799ad478d92487ca3bd43544fdc6d02b8f6",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/83899adebf94722befbec210a795d4b8d0db5b92",
+        "source_type": "paper",
+        "content_sha256": "sha256:2a3689dd96225957a2d1790d90bcc4dbe76edbbc9529b62a74f8642f4a3cbed0"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3698829",
+      "title": "Understanding and Reusing Test Suites Across Database Systems",
+      "authors": [
+        "Suyang Zhong",
+        "Manuel Rigger"
+      ],
+      "year": 2024,
+      "venue": "Proc. ACM Manag. Data",
+      "doi": "10.1145/3698829",
+      "arxiv_id": "2410.21731",
+      "s2_paper_id": "a109480dd652b779ac20ba51ba7278dbb57a376c",
+      "url": "https://doi.org/10.1145/3698829",
+      "open_access_pdf": "https://doi.org/10.1145/3698829",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3698829",
+              "source_type": "paper_citation_context",
+              "excerpt": "The main threat to the internal validity of this work is potential inaccuracies in the manual classification of failed test cases in RQ3 and RQ4, which could introduce bias, leading to misclassification of the reasons for failed cases.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:407b89ece0e5a167c8236c0efe74cebd8bdb402002769c004b907755968f0ce8",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3698829",
+              "source_type": "paper_citation_context",
+              "excerpt": "Le et al. [22] found bugs in LLVM by using test cases from GCC. Zhong et al. [50] used the test suite of each DBMS as the seed corpus for their mutation-based fuzzing approach, which found many bugs in DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5c2ff1e517d4bbbc9395383d047c5617670fb4e0cfda29ceba355dc833552b6a",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3698829",
+              "source_type": "paper_citation_context",
+              "excerpt": "Researchers have developed various techniques to automate testing DBMSs and successfully found logic bugs [23, 33–35, 38, 40], performance issues [5, 20, 24], crash bugs [37, 50], and transaction bugs [21].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ccafedd6858a3f3b2075bafe529caa4ef03bc02e1d5f4b93741d198d926d4367",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3698829",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer creates databases and queries, and then autonomously validates the results provided by the DBMS [4, 33–35].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c39a4cd36d401423ac01c521c8121bb75a1ee7280ef1e0d5c407f6e7698b82b7",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3698829",
+              "source_type": "paper_citation_context",
+              "excerpt": "These DBMSs have also been the target of various automated testing works [4, 14, 18, 23, 33–35, 37, 50].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0202d6fd2eda0f9611cb07167cb353f32b5d13350951ce2a2e46ac9d85ccb2e1",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/a109480dd652b779ac20ba51ba7278dbb57a376c",
+        "source_type": "paper",
+        "content_sha256": "sha256:0c8e5c5f8189fbc0088c6e164d37026612923b6ae9b99487ac8075c5f29a59ed"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Database Management System (DBMS) developers have implemented extensive test suites to test their DBMSs. For example, the SQLite test suites contain over 92 million lines of code. Despite these extensive efforts, test suites are not systematically reused across DBMSs, leading to wasted effort. Integration is challenging, as test suites use various test case formats and rely on unstandardized test runner features. We present a unified test suite, SQuaLity, in which we integrated test cases from three widely-used DBMSs, SQLite, PostgreSQL, and DuckDB. In addition, we present an empirical study to determine the potential of reusing these systems' test suites. Our results indicate that reusing test suites is challenging: First, test formats and test runner commands vary widely; for example, SQLite has 4 test runner commands, while MySQL has 112 commands with additional features, to, for example, execute file operations or interact with a shell. Second, while some test suites contain mostly standard-compliant statements (e.g., 99% in SQLite), other test suites mostly test non-standardized functionality (e.g., 31% of statements in the PostgreSQL test suite are nonstandardized). Third, test reuse is complicated by various explicit and implicit dependencies, such as the need to set variables and configurations, certain test cases requiring extensions not present by default, and query results depending on specific clients. Despite the above findings, we have identified 3 crashes, 3 hangs, and multiple compatibility issues across four different DBMSs by executing test suites across DBMSs, indicating the benefits of reuse. Overall, this work represents the first step towards test-case reuse in the context of DBMSs, and we hope that it will inspire follow-up work on this important topic."
+    },
+    {
+      "id": "paper:doi:10.1145/3597503.3639207",
+      "title": "Understanding Transaction Bugs in Database Systems",
+      "authors": [
+        "Ziyu Cui",
+        "Wensheng Dou",
+        "Yu Gao",
+        "Dong Wang",
+        "Jiansen Song",
+        "Yingying Zheng",
+        "Tao Wang",
+        "Rui Yang",
+        "Kang Xu",
+        "Yixin Hu",
+        "Jun Wei",
+        "Tao Huang"
+      ],
+      "year": 2024,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1145/3597503.3639207",
+      "arxiv_id": null,
+      "s2_paper_id": "c8376be6e5d269467545de4edffe06bb2b74b1c7",
+      "url": "https://doi.org/10.1145/3597503.3639207",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597503.3639207",
+      "abstract": "Transactions are used to guarantee data consistency and integrity in Database Management Systems (DBMSs), and have become an indispensable component in DBMSs. However, faulty designs and implementations of DBMSs' transaction processing mechanisms can introduce transaction bugs, and lead to severe consequences, e.g., incorrect database states and DBMS crashes. An in-depth understanding of real-world transaction bugs can significantly promote effective techniques in combating transaction bugs in DBMSs. In this paper, we conduct the first comprehensive study on 140 transaction bugs collected from six widely-used DBMSs, i.e., MySQL, PostgreSQL, SQLite, MariaDB, CockroachDB, and TiDB. We investigate these bugs from their bug manifestations, root causes, bug impacts and bug fixing. Our study reveals many in-teresting findings and provides useful guidance for transaction bug detection, testing, and verification.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639207",
+              "source_type": "paper_citation_context",
+              "excerpt": "If a bug can be triggered by only one auto transaction, we do not consider it as a TXBug, since it usually belongs to transaction-unrelated bugs, e.g., logic bugs in single SELECT statements [64–66].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:002ec5ed69ea5d92935f5a3f53bfbd29f0d521acfa1dd4ae3387feb175aceb51",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639207",
+              "source_type": "paper_citation_context",
+              "excerpt": "Although many approaches have been proposed for DBMS testing [17, 34, 49, 51, 54, 55, 59–61, 64–68, 73, 74, 77, 78], there are only few works for transaction testing in DBMSs [44, 45].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:04b98fb41b4354d31b083f9c800f07c4fcab9278c28b90e30d33a7ee73caec44",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639207",
+              "source_type": "paper_citation_context",
+              "excerpt": "Some techniques have been proposed to combat logic and crash bugs in DBMSs [17, 34, 49, 51, 54, 55, 59– 61, 64–68, 73, 74, 76–78].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4ef9349331bf41a6838bb24bd84d41a543dbe41ee41b55135286aa3bc01108e6",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639207",
+              "source_type": "paper_citation_context",
+              "excerpt": "RAGS [67] utilizes differential testing to find bugs in DBMSs. SQLancer [64–66] detects logic bugs in single SELECT statements.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7a91ded990aa095dbc1d35f410b57dd5d382ba8033fa8652889778b1ab3b7b1f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/c8376be6e5d269467545de4edffe06bb2b74b1c7",
+        "source_type": "paper",
+        "content_sha256": "sha256:7848e64064a352167d4b983ba57cd0f02f8e6093061856afc81c2789e3548ca4"
+      }
+    },
+    {
+      "id": "paper:doi:10.1016/j.displa.2024.102854",
+      "title": "Using query semantic and feature transfer fusion to enhance cardinality estimating of property graph queries",
+      "authors": [
+        "Zhenzhen He",
+        "Tiquan Gu",
+        "Jiong Yu"
+      ],
+      "year": 2024,
+      "venue": "Displays (Guildford)",
+      "doi": "10.1016/j.displa.2024.102854",
+      "arxiv_id": null,
+      "s2_paper_id": "1faf7f08681013961cb385344a600bd5f08c69ec",
+      "url": "https://doi.org/10.1016/j.displa.2024.102854",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "cert"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "cert"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/1faf7f08681013961cb385344a600bd5f08c69ec",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Cardinality Estimation Restriction Testing (CERT).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/1faf7f08681013961cb385344a600bd5f08c69ec",
+        "source_type": "paper",
+        "content_sha256": "sha256:374d19ad8ae1bd0c977f9a612159211d680bcb538a762b0116f297803ce02c6a"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3627703.3650080",
+      "title": "Validating Database System Isolation Level Implementations with Version Certificate Recovery",
+      "authors": [
+        "Jack Clark",
+        "Alastair F. Donaldson",
+        "John Wickerson",
+        "Manuel Rigger"
+      ],
+      "year": 2024,
+      "venue": "European Conference on Computer Systems",
+      "doi": "10.1145/3627703.3650080",
+      "arxiv_id": null,
+      "s2_paper_id": "04178561c9d8cd42904ed0a836e08e55b84703cf",
+      "url": "https://doi.org/10.1145/3627703.3650080",
+      "open_access_pdf": "https://doi.org/10.1145/3627703.3650080",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3627703.3650080",
+              "source_type": "paper_citation_context",
+              "excerpt": "This approach is inspired by pivoted query synthesis [41], which uses a data-base system specific SQL interpreter to execute a predicate condition to determine if a given row would match the predicate.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0ee92095236e4b4656d9dab0bd441c5bfef682dac1752374caf095be9e3b81a9",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3627703.3650080",
+              "source_type": "paper_citation_context",
+              "excerpt": "We believe this is reasonable as there are existing techniques [1, 6, 39–43] to validate the correctness of the query evaluator/optimizer and the focus of this work is on finding bugs in the implementation of concurrency control protocols.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9568752a718ab2522e1fd9ba362fb4d37fc04c2378e39a36ec28060e7ed7106f",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "manual_curation",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3627703.3650080",
+              "source_type": "paper_citation_context",
+              "excerpt": "This approach is inspired by pivoted query synthesis [41], which uses a data-base system specific SQL interpreter to execute a predicate condition to determine if a given row would match the predicate.",
+              "excerpt_is_verbatim": true,
+              "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+              "content_sha256": "sha256:0ee92095236e4b4656d9dab0bd441c5bfef682dac1752374caf095be9e3b81a9",
+              "retrieved_at": "2026-09-06T07:08:48Z",
+              "first_seen": "2026-09-06T07:08:48Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ],
+          "classifier": {
+            "method": "manual_curation",
+            "classifier_version": "manual-curation-v1",
+            "policy_version": "impact-policy-v1",
+            "taxonomy_version": "taxonomy-v1",
+            "classified_at": "2026-09-06T07:08:48Z",
+            "model": null,
+            "rationale": "Reviewed by a maintainer against the citation contexts collected for this paper."
+          }
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/04178561c9d8cd42904ed0a836e08e55b84703cf",
+        "source_type": "paper",
+        "content_sha256": "sha256:1522b60e6d474b8205d849af0ea6ca972bd84225eec274c327cd970c1d0d9789"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Transactions are a key feature of database systems and isolation levels specify the behavior of concurrently executing transactions. Ensuring their correct behavior is crucial. Recently, many isolation anomalies have been found in production database systems. Checkers can be used to validate that a particular execution conforms to a desired isolation level. However, state-of-the-art checkers cannot handle predicate operations, which are both common in real-world workloads and essential for distinguishing between the repeatable read and serializable isolation levels. In this work, we address this issue by proposing an efficient white-box checker, Emme. Our key idea is to use information that is easily provided by database systems to efficiently check the isolation level of a given transaction history. We present version certificate recovery, a method of recovering the version order and each operation's version from the database system under test. For efficiency, we also propose the concept of an expected serialization order, which obviates the need to define and recover a version certificate for many serializable concurrency control protocols. We have implemented version certificate recovery for three widely used database systems---PostgreSQL, CockroachDB, and TiDB. We demonstrate that Emme is 1.2-3.6× faster than Elle, a state-of-the-art checker. Using the expected serialization order, we obtain a further speedup of 34-430× compared to Emme when checking histories containing predicate operations. We show that our approach can identify invalid histories that cannot be detected by Elle and also show that it can find realistic bugs purposely introduced by an engineer."
+    },
+    {
+      "id": "paper:s2:295c629691d5654def2d9d85f72c756c67c68583",
+      "title": "WingFuzz: Implementing Continuous Fuzzing for DBMSs",
+      "authors": [
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Jingzhou Fu",
+        "Yiyuan Bai",
+        "Qiang Zhang",
+        "Yu Jiang"
+      ],
+      "year": 2024,
+      "venue": "USENIX Annual Technical Conference",
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "295c629691d5654def2d9d85f72c756c67c68583",
+      "url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [40–42] designs three test oracles to detect logic errors and generates queries following the oracles.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ce70c1446a67acd1a578dac8633d73fafa8bc7667bf3d61ece6673d101e6c7b5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583",
+              "source_type": "paper_citation_context",
+              "excerpt": "Both the industry and academia have developed many meth-ods for detecting issues in DBMSs [24,32,42,46].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e54c62720e534ead401f68ec7448703f9536b0af527cfe589abbab9f93cbb436",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, its NOREC [40] oracle requires constructing a SQL query with WHERE and JOIN clauses.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7fdaed2a9e01a3906cfbaa8f4e47b341104e73c8282537f01bf07ab1f6c7924b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json",
+              "source_type": "paper",
+              "excerpt": "1 Compared with Existing Fuzzers To show the effectiveness of WINGFUZZ, we compared it with three state-of-the-art fuzzers, including conventional mutation-based fuzzer SQUIRREL as well as generated-based fuzzer SQLancer and SQLsmith, which are widely used in the industry to test DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, 6.1 Compared with Existing Fuzzers, page 8.",
+              "content_sha256": "sha256:858e8eed111cb2d3ad055517c0fa80522294bd5d1f625afe4f774097ba7817b8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json",
+              "source_type": "paper",
+              "excerpt": "They show that WINGFUZZ covers a total of 211620, 197059, 132397 more branches than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 6.1 Compared with Existing Fuzzers, page 8.",
+              "content_sha256": "sha256:499d1210cb8281e9323df65041fc1d10ed40e549d9ecf82225fa086051f89fdd",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json",
+              "source_type": "paper",
+              "excerpt": "Specifically, WINGFUZZ finds a total of 25, 24, 21 more unique bugs than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 6.1 Compared with Existing Fuzzers, page 9.",
+              "content_sha256": "sha256:b3936af12904bc39e81894ce563e284e5bb9c82129788b5b94e4455a2570c55a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json",
+              "source_type": "paper",
+              "excerpt": "1 Compared with Existing Fuzzers To show the effectiveness of WINGFUZZ, we compared it with three state-of-the-art fuzzers, including conventional mutation-based fuzzer SQUIRREL as well as generated-based fuzzer SQLancer and SQLsmith, which are widely used in the industry to test DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, 6.1 Compared with Existing Fuzzers, page 8.",
+              "content_sha256": "sha256:858e8eed111cb2d3ad055517c0fa80522294bd5d1f625afe4f774097ba7817b8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583",
+        "source_type": "paper",
+        "content_sha256": "sha256:c9ba9dab2b4db85e9fef8c0807174542b833615f16c6cf646414deffcf50c553"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse48619.2023.00024",
+      "title": "A Comprehensive Study of Real-World Bugs in Machine Learning Model Optimization",
+      "authors": [
+        "Hao Guan",
+        "Ying Xiao",
+        "Jiaying Li",
+        "Yepang Liu",
+        "Guangdong Bai"
+      ],
+      "year": 2023,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse48619.2023.00024",
+      "arxiv_id": null,
+      "s2_paper_id": "35b7f91900e5d37aaf160bb0590a812754fd3137",
+      "url": "https://doi.org/10.1109/icse48619.2023.00024",
+      "open_access_pdf": null,
+      "abstract": "Due to the great advance in machine learning (ML) techniques, numerous ML models are expanding their application domains in recent years. To adapt for resource-constrained platforms such as mobile and Internet of Things (IoT) devices, pre-trained models are often processed to enhance their efficiency and compactness, using optimization techniques such as pruning and quantization. Similar to the optimization process in other complex systems, e.g., program compilers and databases, optimizations for ML models can contain bugs, leading to severe consequences such as system crashes and financial loss. While bugs in training, compiling and deployment stages have been extensively studied, there is still a lack of systematic understanding and characterization of model optimization bugs (MOBs). In this work, we conduct the first empirical study to identify and characterize MOBs. We collect a comprehensive dataset containing 371 MOBs from TensorFlow and PyTorch, the most extensively used open-source ML frameworks, covering the entire development time span of their optimizers (May 2019 to August 2022). We then investigate the collected bugs from various perspectives, including their symptoms, root causes, life cycles, detection and fixes. Our work unveils the status quo of MOBs in the wild, and reveals their features on which future detection techniques can be based. Our findings also serve as a warning to the developers and the users of ML frameworks, and an appeal to our research community to enact dedicated countermeasures.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00024",
+              "source_type": "paper_citation_context",
+              "excerpt": "Similar to the optimization task in other software that handles complex objects, such as in program compilers [20]– [22] and databases [23], ML model optimization is an error-prone process.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:94639a909acf2be77b78ea688afd737c04ae692475c8dbc75886d29a631d6efe",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/35b7f91900e5d37aaf160bb0590a812754fd3137",
+        "source_type": "paper",
+        "content_sha256": "sha256:ce3f83fcab44076f9cc17044a6a104191dd2a24d9e44bd477f6f5b33524bf0cc"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3611540.3611584",
+      "title": "A Demonstration of DLBD: Database Logic Bug Detection System",
+      "authors": [
+        "Xiu Tang",
+        "Sai Wu",
+        "Dongxiang Zhang",
+        "Ziyue Wang",
+        "Gongsheng Yuan",
+        "Gang Chen"
+      ],
+      "year": 2023,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3611540.3611584",
+      "arxiv_id": null,
+      "s2_paper_id": "7c4173e9df28137dcf56cfd0f13c1d0dd11908f3",
+      "url": "https://doi.org/10.14778/3611540.3611584",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3611540.3611584",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS [6] constructs queries to fetch a randomly selected tuple from a table, while NoREC [4] compares the results of randomly generated optimized queries and rewritten queries that DBMS cannot optimize.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f9bd4977140af798a72c410ddcc2a067ac2e9999d87a3483d7b3ad248bbd1aa8",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3611540.3611584",
+              "source_type": "paper_citation_context",
+              "excerpt": "The tool employs several approaches to detect logic bugs, such as Pivoted Query Synthesis (PQS), Ternary Logic Partitioning (TLP), and Non-optimizing Reference Engine Construction (NoREC).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:cb2c728d2214ab0b1f076b2b0c5bc046654990369421eb9cdee94dc1857eec97",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/7c4173e9df28137dcf56cfd0f13c1d0dd11908f3",
+        "source_type": "paper",
+        "content_sha256": "sha256:b532b9dc1623c8a4b1a5d62e7e6fd170a00c2d37f374038353093515a359347f"
+      },
+      "abstract": "Database management systems (DBMSs) are prone to logic bugs that can result in incorrect query results. Current debugging tools are limited to single table queries and struggle with issues like lack of ground-truth results and repetitive query space exploration. In this paper, we demonstrate DLBD, a system that automatically detects logic bugs in databases. DLBD offers holistic logic bug detection by providing automatic schema and query generation and ground-truth query result retrieval. Additionally, DLBD provides minimal test cases and root cause analysis for each bug to aid developers in reproducing and fixing detected bugs. DLBD incorporates heuristics and domain-specific knowledge to efficiently prune the search space and employs query space exploration mechanisms to avoid the repetitive search. Finally, DLBD utilizes a distributed processing framework to test database logic bugs in a scalable and efficient manner. Our system offers developers a reliable and effective way to detect and fix logic bugs in DBMSs."
+    },
+    {
+      "id": "paper:doi:10.1145/3611643.3613893",
+      "title": "Adapting Performance Analytic Techniques in a Real-World Database-Centric System: An Industrial Experience Report",
+      "authors": [
+        "Lizhi Liao",
+        "Heng Li",
+        "Weiyi Shang",
+        "Catalin Sporea",
+        "Andrei Toma",
+        "S. Sajedi"
+      ],
+      "year": 2023,
+      "venue": "ESEC/SIGSOFT FSE",
+      "doi": "10.1145/3611643.3613893",
+      "arxiv_id": null,
+      "s2_paper_id": "b0b3eecc7e3953547ae24a909d5967240c19f373",
+      "url": "https://doi.org/10.1145/3611643.3613893",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3611643.3613893",
+      "abstract": "Database-centric architectures have been widely adopted in large-scale software systems in various domains to deal with the ever-increasing amount and complexity of data. Prior studies have proposed a wide range of performance analytic techniques aimed at assisting developers in pinpointing software performance inefficiencies and diagnosing performance issues. However, directly applying these existing techniques to large-scale database-centric systems can be challenging and may not perform well due to the unique nature of such systems. In particular, compared to typical database-based systems like online shopping systems, in database-centric systems, a majority of the business logic and calculations reside in the database instead of the application. As the calculations in the database typically use domain-specific languages such as SQL, the performance issues of such systems and their diagnosis may be significantly different from the systems dominated by traditional programming languages such as Java. In this paper, we share our experience of adapting performance analytic techniques in a large-scale database-centric system from our industrial collaborator. Our adapted performance analysis pays special attention to the database and the interactions between the database and the application with minimal reliance on expert knowledge and manual effort. Moreover, we document our encountered challenges and how they are addressed during the development and adoption of our solution in the industrial setting as well as the corresponding lessons learned. We also discuss the real-world performance issues detected by applying our analysis to the target database-centric system. We anticipate that our solution and the reported experience can be helpful for practitioners and researchers who would like to ensure and improve the performance of database-centric systems.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3611643.3613893",
+              "source_type": "paper_citation_context",
+              "excerpt": "Rigger and Su utilize multiple analysis techniques, such as query partitioning [43], pivoted query synthesis [44], or non-optimizing reference engine construction [42] to detect database logic bugs that are critical in database-centric system quality but often go unnoticed by developers.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3fc05af0156c5685fe52ec271e1d6b554c8ebf33221a3d09bae9c0f8e284ec35",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/b0b3eecc7e3953547ae24a909d5967240c19f373",
+        "source_type": "paper",
+        "content_sha256": "sha256:3fd82b828a5d99b99f927c787b484abdeab781de11b23d49cba2f25c5c00f96e"
+      }
+    },
+    {
+      "id": "paper:doi:10.1117/12.3006402",
+      "title": "An empirical study on configuration-related branch statement in database management system",
+      "authors": [
+        "Xie Li",
+        "Huiping Zhou",
+        "Haifang Zhou",
+        "Jingying Zhang"
+      ],
+      "year": 2023,
+      "venue": "International Conference on Modelling, Identification and Control",
+      "doi": "10.1117/12.3006402",
+      "arxiv_id": null,
+      "s2_paper_id": "6200bb7acb7f2d10f5111f75a895ef4e74b81389",
+      "url": "https://doi.org/10.1117/12.3006402",
+      "open_access_pdf": null,
+      "abstract": "Database management systems introduce thousands of configuration options to provide flexibility. However, configuration complexity and configuration dependencies have inevitably affect the system reliability and become one of the major causes of system failure. Previous research neither specifically detects configuration-related bugs nor studies the characteristics of configuration-related code in DBMSs. In this paper, we undertake one of the first attempts to conduct a real-world configuration-related branch statements characteristic study in DBMSs. We first use a taint analysis tool to construct a data set of 347 real world configuration-related branch statements including 100 in MySQL, 98 in SQLite and 149 in PostgreSQL, which are very popular. Based on our study, we have obtained some findings: (1) More than half (53. 6%) of configuration-related branch statements contain only configuration variables, 93. 5% of which contain only one configuration variable. (2) 46. 4% of configuration-related branch statements contain not only configuration variables but also program variables. (3) Complex configuration-related branch statements (which contain function call) account for 17. 9%.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1117/12.3006402",
+              "source_type": "paper_citation_context",
+              "excerpt": "Malicious attackers may use these vulnerabilities to steal user information, embezzle data, crash the system, and bring unpredictable economic losses[7, 8, 9, 10].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:82316a2d12fc67560b58cd18ed8bddfa676abd8b2a8d697cb5ef875b957ae055",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/6200bb7acb7f2d10f5111f75a895ef4e74b81389",
+        "source_type": "paper",
+        "content_sha256": "sha256:bdcb9013b8397e7d03d1bdcffa616200e63589454e5c4cd9613539cf3735278d"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3605157.3605177",
+      "title": "Beyond the Coverage Plateau: A Comprehensive Study of Fuzz Blockers (Registered Report)",
+      "authors": [
+        "Wentao Gao",
+        "Van-Thuan Pham",
+        "Dongge Liu",
+        "Oliver Chang",
+        "Toby C. Murray",
+        "Benjamin I. P. Rubinstein"
+      ],
+      "year": 2023,
+      "venue": "FUZZING",
+      "doi": "10.1145/3605157.3605177",
+      "arxiv_id": null,
+      "s2_paper_id": "1b13b395ea1d7d0f23fc78df96ca430831408634",
+      "url": "https://doi.org/10.1145/3605157.3605177",
+      "open_access_pdf": null,
+      "abstract": "Fuzzing and particularly code coverage-guided greybox fuzzing is highly successful in automated vulnerability discovery, as evidenced by the multitude of vulnerabilities uncovered in real-world software systems. However, results on large benchmarks such as FuzzBench indicate that the state-of-the-art fuzzers often reach a plateau after a certain period, typically around 12 hours. With the aid of the newly introduced FuzzIntrospector platform, this study aims to analyze and categorize the fuzz blockers that impede the progress of fuzzers. Such insights can shed light on future fuzzing research, suggesting areas that require further attention. Our preliminary findings reveal that the majority of top fuzz blockers are not directly related to the program input, emphasizing the need for enhanced techniques in automated fuzz driver generation and modification.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3605157.3605177",
+              "source_type": "paper_citation_context",
+              "excerpt": "Additionally, researchers have attempted to extend the applicability of fuzzing to challenging targets such as network protocols [18, 43], database systems [45, 50], SMT solvers [40], compilers [26], device drivers [41], and heterogeneous applications [48].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:527d9a099090065ffccae1dddc242f75ae8950ad58a33aef9d13e8896fe81815",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3605157.3605177",
+              "source_type": "paper_citation_context",
+              "excerpt": "These efforts have focused on enhancing fuzzing in areas such as feedback collection [16, 25, 27, 30], corpus management [29], seed selection algorithms [22, 23], input generation algorithms [15, 20, 31, 44, 47], and novel test oracle designs [39, 45].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7b146d7c7ddf6836ecf7086e44f21916e3818b4a5dd3a6dbfda92f5bda3f5f51",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/1b13b395ea1d7d0f23fc78df96ca430831408634",
+        "source_type": "paper",
+        "content_sha256": "sha256:8f6de255868160f7e18f3a9b0d9bcbdcbb6b716dc275ca1e93c79491794d061f"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3597503.3639076",
+      "title": "CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation",
+      "authors": [
+        "Jinsheng Ba",
+        "Manuel Rigger"
+      ],
+      "year": 2023,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1145/3597503.3639076",
+      "arxiv_id": "2306.00355",
+      "s2_paper_id": "a4ec52e8b18e41359cf87097efae760905c27115",
+      "url": "https://doi.org/10.1145/3597503.3639076",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597503.3639076",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639076",
+              "source_type": "paper_citation_context",
+              "excerpt": "Common generation-based methods include mutation-based methods [30, 67] and rule-based generation methods [42–44 , 53 Before executing queries on the generated database state, we execute ANALYZE statements on each table to guarantee that the data statistics are up to date.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:29ddb6f077a4c0a19ab0e7a0b7785ee9498e30215bfd3d3b38a5b231b94f0b60",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639076",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [43] and NoREC [42] are other metamorphic testing approaches that were proposed for testing DBMSs, as discussed above.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9c9af589e3de648f503470964b89d273b674ef04b688656800126c5c74c05855",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639076",
+              "source_type": "paper_citation_context",
+              "excerpt": "The PQS [44], NoREC [42], and TLP [43] oracles detect logic bugs in the implementation of SELECT statements.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:529b2b31d73cacc0be121f6a9208d9065f6c817f95e14fdf5e4ef7e1c05918ca",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639076",
+              "source_type": "paper_citation_context",
+              "excerpt": "They are widely used and have thus been studied in other DBMS testing works [30, 43, 44].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3cf19b71dbb3e34b9e1998cfd7e26cc4c3c04c4980794db128011a0179f60420",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639076",
+              "source_type": "paper_citation_context",
+              "excerpt": "Metamorphic testing has been applied successfully in various domains [8, 45].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5c54f8d641dc41fc248b88568c259e53df9a81e5dfb60d6d783b06e0480ed6d9",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597503.3639076",
+              "source_type": "paper_citation_context",
+              "excerpt": "These DBMSs have been widely used in prior research [30, 43, 44].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:30ce84ce00424f902ced8d237f8c19dec69dc10e7fcda2e25f31b1d62f498cc7",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/a4ec52e8b18e41359cf87097efae760905c27115",
+        "source_type": "paper",
+        "content_sha256": "sha256:00e992b3606cc5b8b4b6c0e14e19c317a5f03dfc55301b65f1ea7e4c8731f856"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Database Management Systems (DBMSs) process a given query by creating a query plan, which is subsequently executed, to compute the query's result. Deriving an efficient query plan is challenging, and both academia and industry have invested decades into researching query optimization. Despite this, DBMSs are prone to performance issues, where a DBMS produces an unexpectedly inefficient query plan that might lead to the slow execution of a query. Finding such issues is a longstanding problem and inherently difficult, because no ground truth information on an expected exe-cution time exists. In this work, we propose Cardinality Estimation Restriction Testing (CERT), a novel technique that finds performance issues through the lens of cardinality estimation. Given a query on a database, CERT derives a more restrictive query (e.g., by replacing a LEFT JOIN with an INNER JOIN), whose estimated number of rows should not exceed the estimated number of rows for the original query. CERT tests cardinality estimation specifically, because it was shown to be the most important part for query optimization; thus, we expect that finding and fixing cardinality-estimation issues might result in the highest performance gains. In addition, we found that other kinds of query optimization issues can be exposed by unexpected estimated cardinalities, which can also be found by CERT. CERT is a black-box technique that does not require access to the source code; DBMSs expose query plans via the EXPLAIN statement. CERT eschews executing queries, which is costly and prone to performance fluctuations. We evaluated CERT on three widely used and mature DBMSs, MySQL, TiDB, and CockroachDB. CERT found 13 unique issues, of which 2 issues were fixed and 9 confirmed by the developers. We expect that this new angle on finding performance bugs will help DBMS developers in improving DMBSs' performance."
+    },
+    {
+      "id": "paper:doi:10.7717/peerj-cs.1592",
+      "title": "DAFuzz: data-aware fuzzing of in-memory data stores",
+      "authors": [
+        "Yingpei Zeng",
+        "Fengmin Zhu",
+        "Siyi Zhang",
+        "Yu Yang",
+        "Siyu Yi",
+        "Yufan Pan",
+        "Guojie Xie",
+        "Ting Wu"
+      ],
+      "year": 2023,
+      "venue": "PeerJ Computer Science",
+      "doi": "10.7717/peerj-cs.1592",
+      "arxiv_id": null,
+      "s2_paper_id": "a3f904ba97221a2e65f6ac9e34f3403e287e0a49",
+      "url": "https://doi.org/10.7717/peerj-cs.1592",
+      "open_access_pdf": "https://doi.org/10.7717/peerj-cs.1592",
+      "abstract": "Fuzzing has become an important method for finding vulnerabilities in software. For fuzzing programs expecting structural inputs, syntactic- and semantic-aware fuzzing approaches have been particularly proposed. However, they still cannot fuzz in-memory data stores sufficiently, since some code paths are only executed when the required data are available. In this article, we propose a data-aware fuzzing method, DAFuzz, which is designed by considering the data used during fuzzing. Specifically, to ensure different data-sensitive code paths are exercised, DAFuzz first loads different kinds of data into the stores before feeding fuzzing inputs. Then, when generating inputs, DAFuzz ensures the generated inputs are not only syntactically and semantically valid but also use the data correctly. We implement a prototype of DAFuzz based on Superion and use it to fuzz Redis and Memcached. Experiments show that DAFuzz covers 13~95% more edges than AFL, Superion, AFL++, and AFLNet, and discovers vulnerabilities over 2.7× faster. In total, we discovered four new vulnerabilities in Redis and Memcached. All the vulnerabilities were reported to developers and have been acknowledged and fixed.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.7717/peerj-cs.1592",
+              "source_type": "paper_citation_context",
+              "excerpt": "Another kind of data-Zeng related program is SQL database, and different special black-box fuzzers ( Seltenreich, Tang & Mullender, 2022 ; Guo, 2017 ; Rigger, 2023 ; Rigger & Su, 2020 ) and grey-box fuzzers ( Zhong et al., 2020 ; Wang et al., 2021 ; Liang, Liu & Hu, 2022 ) have been developed.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2e8c4d232533b7fed5a06a0a9d7da66d57a34a1aabcf1957a0ebf799202872a1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.7717/peerj-cs.1592",
+              "source_type": "paper_citation_context",
+              "excerpt": "Recently, fuzzers also try to ensure the statements are semantically valid ( Rigger, 2023 ; Rigger & Su, 2020 ; Zhong et al., 2020 ; Wang et al., 2021 ; Liang, Liu & Hu, 2022 ).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d7f5be4266aa1e729d377a71fc292e96732cf49d27d696e71a9e052997fc751e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.7717/peerj-cs.1592",
+              "source_type": "paper_citation_context",
+              "excerpt": "Several recent pieces of research focus on detecting logic bugs but not traditional crashes or assert failures ( Rigger, 2023 ; Rigger & Su, 2020 ; Liang, Liu & Hu, 2022 ).",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b6a307126cb77e309231b53501eb65d4ac48353f447ea4611d8670ff9479246e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/a3f904ba97221a2e65f6ac9e34f3403e287e0a49",
+        "source_type": "paper",
+        "content_sha256": "sha256:b4aadbde9daeef80fdf7f0573af1915d770075ce278dfb92e7b1dbeb95041c6e"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3597926.3598052",
+      "title": "Dependency-Aware Metamorphic Testing of Datalog Engines",
+      "authors": [
+        "Muhammad Numair Mansur",
+        "Valentin Wüstholz",
+        "M. Christakis"
+      ],
+      "year": 2023,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3597926.3598052",
+      "arxiv_id": null,
+      "s2_paper_id": "df170dd0f26ec903ee0a8d0ad89305ec60a125c5",
+      "url": "https://doi.org/10.1145/3597926.3598052",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597926.3598052",
+      "abstract": "Datalog is a declarative query language with wide applicability, especially in program analysis. Queries are evaluated by Datalog engines, which are complex and thus prone to returning incorrect results. Such bugs, called query bugs, may compromise the soundness of upstream program analyzers, having potentially detrimental consequences in safety-critical settings. To address this issue, we develop a metamorphic testing approach for detecting query bugs in Datalog engines. In comparison to existing work, our approach is based on rich precedence information capturing dependencies among relations in the program. This enables much more general and effective metamorphic transformations. We implement our approach in DLSmith, which detected 16 previously unknown query bugs in four Datalog engines.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598052",
+              "source_type": "paper_citation_context",
+              "excerpt": "Metamorphic testing has also been successfully used to test a variety of software applications [12, 27, 47, 57], including other query-based systems [43, 46, 63].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:36fdf3010c802bcf20bc30bb4f85b12c4f36e444c03bd487ff96851c36e4a3e9",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/df170dd0f26ec903ee0a8d0ad89305ec60a125c5",
+        "source_type": "paper",
+        "content_sha256": "sha256:c5be4c89d450d676c3fda60379f3bb5517d86b07f832bc3a64d4637406acca74"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/compsac57700.2023.00273",
+      "title": "Detecting Hidden Failures of DBMS: A Comprehensive Metamorphic Relation Output Patterns Approach",
+      "authors": [
+        "M. Tang",
+        "T. Tse",
+        "Z. Zhou"
+      ],
+      "year": 2023,
+      "venue": "Annual International Computer Software and Applications Conference",
+      "doi": "10.1109/compsac57700.2023.00273",
+      "arxiv_id": null,
+      "s2_paper_id": "e9db272057833ba69d1c6e9ecbb992af02ce25a9",
+      "url": "https://doi.org/10.1109/compsac57700.2023.00273",
+      "open_access_pdf": "https://hub.hku.hk/bitstream/10722/345747/1/content.pdf",
+      "abstract": "The testing of large databases faces the test oracle problem, namely, that it is difficult to verify execution results against expected outcomes. Rigger and Su applied metamorphic testing through query partitioning and ternary logic partitioning techniques to alleviate the challenge. In Part (A) of our project, we conduct an in-depth investigation and have identified a gap between the two techniques. We propose a disjoint partitioning approach to address it. In Part (B), we conduct a comprehensive investigation into the metamorphic testing of DBMS by comparing disjoint partitioning with metamorphic relation output patterns (MROPs) by Segura et al. We propose an exhaustive collection of MROPs for DBMS. To the best of our knowledge, this is the first project to integrate in-depth and comprehensive approaches to tackle the diverse challenges in DBMS testing. In Part (C), we conduct an empirical case study of their applications to OceanBase, the DBMS associated with the world’s fastest online transaction processing system. Although OceanBase has been extensively tested and widely used in the industry, we have detected 12 hidden failures and 8 new crashes.",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+              "source_type": "paper_citation_context",
+              "excerpt": "Since SQL is based on ternary Boolean logic [14], we know that name = 'Alice' can be TRUE, FALSE, or NULL, falling into the resultant lists of the three partitioning queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1c941bf8a32eda3868ae7c205f6982ac5fd29be56ef48d4a995dccedb5545cb2",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+              "source_type": "paper_citation_context",
+              "excerpt": "D. Extension of QP and TLP to Disjoint Partitioning1 Our thorough investigation reveals that the high-level QP is too general for practical MR construction, while the low-level TLP is too specific.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b863fb7d3909e0bbf3f0ed340a18094ac620652b0c8044012716d7f3eb29e964",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+              "source_type": "paper_citation_context",
+              "excerpt": "In particular, we find that QP and TLP [8] as well as our proposed DP cover only two MROPs, namely list equality and bag equality of complete disjoint partitioning (as explained in Subsection II.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:95602428358532168564429cc5978ee64e118fb252a1da87698796d0601a37e0",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+              "source_type": "paper_citation_context",
+              "excerpt": "Rigger and Su [8] proposed query partitioning (QP) and ternary logic partitioning (TLP) for metamorphic testing of DBMS. QP is a general strategic concept that describes an MR among DBMS queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:031429abb2cf96ee4e01eda74bd4bf59b8bea1d340d00b1e34dd61ff9e0b8ca3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+              "source_type": "paper_citation_context",
+              "excerpt": "In 2020, Rigger and Su [8] applied MT to address the issue in DBMS testing through the query partitioning (QP) and ternary logic partitioning (TLP) techniques.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d34c95021673a9ee05468d9ec829b83e92c1045510ece107f0485c2e70584e57",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/compsac57700.2023.00273",
+              "source_type": "paper_citation_context",
+              "excerpt": "Revisit of QP and TLP by Rigger and Su Rigger and Su [8] proposed query partitioning (QP) and ternary logic partitioning (TLP) for metamorphic testing of DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5a931693ab1df663b5a4e3412dab1347c1f60de4c24a6f67fcc01846a58ad165",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "insufficient_evidence",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/e9db272057833ba69d1c6e9ecbb992af02ce25a9",
+        "source_type": "paper",
+        "content_sha256": "sha256:e6cfceb9de945848ff0e5d2aac155ad33da389c8b179300da7046d9fc1dced12"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse48619.2023.00101",
+      "title": "Detecting Isolation Bugs via Transaction Oracle Construction",
+      "authors": [
+        "Wensheng Dou",
+        "Ziyu Cui",
+        "Qianwang Dai",
+        "Jiansen Song",
+        "Dong Wang",
+        "Yu Gao",
+        "Wei Wang",
+        "Jun Wei",
+        "Lei Chen",
+        "Han Wang",
+        "Hua Zhong",
+        "Tao Huang"
+      ],
+      "year": 2023,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse48619.2023.00101",
+      "arxiv_id": null,
+      "s2_paper_id": "a9fa1a56ba118fa3b629fc6308d6b5606751f023",
+      "url": "https://doi.org/10.1109/icse48619.2023.00101",
+      "open_access_pdf": null,
+      "abstract": "Transactions are used to maintain the data integrity of databases, and have become an indispensable feature in modern Database Management Systems (DBMSs). Despite extensive efforts in testing DBMSs and verifying transaction processing mechanisms, isolation bugs still exist in widely-used DBMSs when these DBMSs violate their claimed transaction isolation levels. Isolation bugs can cause severe consequences, e.g., incorrect query results and database states. In this paper, we propose a novel transaction testing approach, Transaction oracle construction (Troc), to automatically detect isolation bugs in DBMSs. The core idea of Troc is to decouple a transaction into independent statements, and execute them on their own database views, which are constructed under the guidance of the claimed transaction isolation level. Any divergence between the actual transaction execution and the independent statement execution indicates an isolation bug. We implement and evaluate Troc on three widely-used DBMSs, i.e., MySQL, MariaDB, and TiDB. We have detected 5 previously-unknown isolation bugs in the latest versions of these DBMSs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+              "source_type": "paper_citation_context",
+              "excerpt": "Automatic database testing approaches like SQLancer [26]–[28] detect logic bugs for single queries (i.e., SELECT statements) without considering concurrent transactions, and cannot be applied on other statements, e.g., UPDATE statements in s 12 and s 22 .",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2d03330aaa2e335cb48889427897860676d1ecec5c2a6a3570c07af39cf2242e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+              "source_type": "paper_citation_context",
+              "excerpt": "Existing testing techniques for DBMSs, e.g., SQLsmith [24] and SQLancer [26]–[28], cannot generate transaction test cases, and do not have a test oracle for transaction test cases.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:682a1cf2fa86cf3e9b03cd3ee4a1545abd874609ee8f73eab8824c8bb6bc42d2",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+              "source_type": "paper_citation_context",
+              "excerpt": "For now, Troc can support many complex data structures and SQL statements, e.g., primary keys, indexes, various data types, and conditions that are supported by SQLancer [49].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:50aa9a86a9f847b7ae7dc9545c92488fcfb58cb8f6beec0904efd2ca92eabc06",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+              "source_type": "paper_citation_context",
+              "excerpt": "Automatic database testing approaches [24]–[28] can support these complex features in modern DBMSs, and have been proved as an effective technique to detect bugs in DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:dac711c335ae4f3ae9a761c44a3f44477a6913bce70d483f8db72abea8608204",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+              "source_type": "paper_citation_context",
+              "excerpt": "We further investigate whether these 12 unique bugs can be revealed by existing approaches, e.g., SQLsmith [24], SQLancer [26]–[28], Elle [23] and Cobra [22].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:31bde2556f9c19291cb3781f052e379d6fbd15b812a174136a307daae3a53f8e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00101",
+              "source_type": "paper_citation_context",
+              "excerpt": "QPG [69] utilizes query plans to guide database state mutation for detecting bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:23f19a759addeceb6aeb75a4e6eed0d644dde920d8f467ddc1c7b9a8ae76f85d",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse48619_2023_00101.json",
+              "source_type": "paper",
+              "excerpt": "Troc’s database and SQL statement generation mainly bases on SQLancer [49].",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, C Database and SQL Statement Generation, page 5.",
+              "content_sha256": "sha256:ed323b28ebabce66245a39836ce39805719e753c16113c00d6a23b8707f200ae",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse48619_2023_00101.json",
+              "source_type": "paper",
+              "excerpt": ", primary keys, indexes, various data types, and conditions that are supported by SQLancer [49].",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, C Database and SQL Statement Generation, page 5.",
+              "content_sha256": "sha256:89aabed54373023c37066c3d6fc7021fab235718867d8c843ac7a8777c5f0d8a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse48619_2023_00101.json",
+              "source_type": "paper",
+              "excerpt": ", SQLsmith [24], SQLancer [26]–[28], Elle [23] and Cobra [22].",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, B Overall Bug Detection Results, page 9.",
+              "content_sha256": "sha256:359c91bf19c8c98239ac0028709e773b0565aa3bdfe08e4f4aa3fc48e393d678",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse48619_2023_00101.json",
+              "source_type": "paper",
+              "excerpt": ", SQLsmith [24] and SQLancer [26]–[28], cannot generate transaction test cases, and do not have a test oracle for transaction test cases.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, B Overall Bug Detection Results, page 9.",
+              "content_sha256": "sha256:aba6e7104cf605b11040fb839ffaf1c4fdbe346771035833145095c65b6790a6",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/a9fa1a56ba118fa3b629fc6308d6b5606751f023",
+        "source_type": "paper",
+        "content_sha256": "sha256:67a27382ea8066c88dd945a6bc5dffb78ee70c5f84dd071a4407311676f0f989"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/criszy/Troc",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "renamed_sqlancer_package",
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/criszy/Troc",
+              "source_type": "github_repository",
+              "excerpt": "Artifact for \"Detecting Isolation Bugs via Transaction Oracle Construction\"\n# Troc\r\n\r\nThis is the artifact for the paper \"Detecting Isolation Bugs via Transaction Oracle Construction\". \r\nSee [paper](http://www.tcse.cn/~cuiziyu20/papers/2023-icse-troc.pdf) to learn more details.",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/criszy/Troc/blob/main/src/main/java/troc/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "src/main/java/troc/Randomly.java is SQLancer's Randomly.java, with the package renamed to troc (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:e3a0454aa2987f2613383e0ec21b0ef330cb2a0d3164871b093ac235acbba125",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3588909",
+      "title": "Detecting Logic Bugs of Join Optimizations in DBMS",
+      "authors": [
+        "Xiu Tang",
+        "Sai Wu",
+        "Dongxiang Zhang",
+        "Fei Li",
+        "Gang Chen"
+      ],
+      "year": 2023,
+      "venue": "Proc. ACM Manag. Data",
+      "doi": "10.1145/3588909",
+      "arxiv_id": "2602.21955",
+      "s2_paper_id": "728c97e05665b2e966a1b17abc5d6d4eaf7e136f",
+      "url": "https://doi.org/10.1145/3588909",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3588909",
+      "abstract": "Generation-based testing techniques have shown their effectiveness in detecting logic bugs of DBMS, which are often caused by improper implementation of query optimizers. Nonetheless, existing generation-based debug tools are limited to single-table queries and there is a substantial research gap regarding multi-table queries with join operators. In this paper, we propose TQS, a novel testing framework targeted at detecting logic bugs derived by queries involving multi-table joins. Given a target DBMS, TQS achieves the goal with two key components: Data-guided Schema and Query Generation (DSG) and Knowledge-guided Query Space Exploration (KQE). DSG addresses the key challenge of multi-table query debugging: how to generate ground-truth (query, result) pairs for verification. It adopts the database normalization technique to generate a testing schema and maintains a bitmap index for result tracking. To improve debug efficiency, DSG also artificially inserts some noises into the generated data. To avoid repetitive query space search, KQE forms the problem as isomorphic graph set discovery and combines the graph embedding and weighted random walk for query generation. We evaluated TQS on four popular DBMSs: MySQL, MariaDB, TiDB and PolarDB. Experimental results show that TQS is effective in finding logic bugs of join optimization in database management systems. It successfully detected 115 bugs within 24 hours, including 31 bugs in MySQL, 30 in MariaDB, 31 in TiDB, and 23 in PolarDB respectively.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3588909",
+              "source_type": "paper_citation_context",
+              "excerpt": "Based on the randomly generated database, it adopts testing approaches such as Pivoted Query Synthesis (PQS) [50] to detect logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ef4497eefcd92546c0f0c973757a683e00697d5211456599cd2d726d5f76767c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3588909",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [47].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7c5192c1ce2508c96000619cd9030daab68b7aaeef5ec631573a88cf76370fc7",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3588909",
+              "source_type": "paper_citation_context",
+              "excerpt": "The second one is TLP [50], which decomposes a query into three partitioning queries, each of which computes its result on that tuple.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e5b61bc80ff1af85c95a3b4d78e5719103bc68039c9dfca65864166f4538323c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3588909",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [48].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5e8b05e550f3deb5eb65c878c8643c012afbc8b3655e2efb4575a106501b53c4",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3588909",
+              "source_type": "paper_citation_context",
+              "excerpt": "Pivoted Query Synthesis (PQS) has recently emerged as a promising way to detect logic bugs in DBMS [50].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e006fc707de5b0d7a94d60e8ac6c25d33b20c4ebe41fad619319b2b833dbe76c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3588909",
+              "source_type": "paper_citation_context",
+              "excerpt": "The third one is NoRec [47], which targets at logic bugs generated by the optimization process in DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1c82d7fe81d471ac14d8a50810a89f7f0d09f2fe73f49903ef5e942baa959c3e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs",
+            "tlp",
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3588909.json",
+              "source_type": "paper",
+              "excerpt": "We use three methods in SQLancer as our baselines.",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, 3.4 Ground-truth Result Generation, page 8.",
+              "content_sha256": "sha256:8e74f496a68222536e46b64fa00ac108071cbb83fcde77c79d345043d3e92725",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3588909.json",
+              "source_type": "paper",
+              "excerpt": "Note that due to the compatibility problem, SQLancer implements different approaches on different databases (PQS and TLP on MySQL and X-DB; NoRec on MariaDB; TLP on TiDB).",
+              "excerpt_is_verbatim": true,
+              "note": "M13 in the paper's extracted text, 5.2 Comparison with Existing Tools, page 12.",
+              "content_sha256": "sha256:175532501334af113c587d30b55e0ff0d549a586fd802d3c8dc4bc6207db1472",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3588909.json",
+              "source_type": "paper",
+              "excerpt": "This is because that (1) SQLancer approaches may generate random joins with empty results which are not usable for testing and (2) TQS adopts the KQE to avoid repeatedly testing the same query structure.",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, 5.2 Comparison with Existing Tools, page 12.",
+              "content_sha256": "sha256:87fef356b786b3ffdd827d80ba07325bde15a95f4c6e65ab235db506f9a5f78a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3588909.json",
+              "source_type": "paper",
+              "excerpt": "SQLancer [ 49] is a current stateof-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours.",
+              "excerpt_is_verbatim": true,
+              "note": "M15 in the paper's extracted text, 5.3 Ablation Studies, page 13.",
+              "content_sha256": "sha256:b4cb8c5f3ce9641c4e089b0b7ffcb5009bf97de801f3e02e97e85e1309d4e142",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/728c97e05665b2e966a1b17abc5d6d4eaf7e136f",
+        "source_type": "paper",
+        "content_sha256": "sha256:41f5e997a57b46c26b658ee9d9ebf09231a070c8951d4dbaa6baf3215add9d53"
+      }
+    },
+    {
+      "id": "paper:s2:3a7373bd891702ae93d7e058241a7e8be25e89eb",
+      "title": "DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation",
+      "authors": [
+        "Zu-Ming Jiang",
+        "Jia-Ju Bai",
+        "Zhendong Su"
+      ],
+      "year": 2023,
+      "venue": "USENIX Security Symposium",
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "3a7373bd891702ae93d7e058241a7e8be25e89eb",
+      "url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [35–37] is also a well-known DBMS testing tool, but it mainly focuses on test oracles, which require test cases with specific patterns to find logic bugs in DBMSs, while DynSQL aims at generating complex and valid queries to detect common bugs, especially memory bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1f060da71df4559ffb4a237357e2c80b60fa1c3d146fe08e89ec93cf777b25c1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+              "source_type": "paper_citation_context",
+              "excerpt": "Because many DBMSs use their own SQL dialects and the common core of their SQL features is small [37,39], it is difficult to use one grammar template to test all DBMSs effectively.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f7d3f9f65b19fff3897c0d49cdbe63f7c57643d3e2509121759c48730abe80d2",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS [37] can generate queries that require the target DBMS to return a result set where a specific row should be included.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8f650251bf5ada17e5e9b080f9d92b3517a7f0cc3c7a1630e6e97000606f7cad",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, this approach is limited because the common part of supported SQL features in different DBMSs is small [37,39].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c11c04321fe81666abe7677d6b1ffa2bbb1c2430ad9e1fed626b5cb1ad658e46",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+              "source_type": "paper_citation_context",
+              "excerpt": "They either discover specific kinds of bugs [18,23,25,34–37], or detect common bugs using general techniques [19,20,39].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:003fcbc5acb1f29107c11d89f773ae8535f7c4f366152229ef8e48a0641cd428",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [35] is a metamorphic testing approach.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ecf7a46a567743f5061d57400b123ed077ea940352319077fbce43a54e528e02",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+        "source_type": "paper",
+        "content_sha256": "sha256:a61796df55ae5798126c6828bbb76ef3203dfa657ad88b764b3c5e48d29b9ad8"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3597503.3608133",
+      "title": "EDEFuzz: A Web API Fuzzer for Excessive Data Exposures",
+      "authors": [
+        "Lianglu Pan",
+        "Shaanan Cohney",
+        "Toby C. Murray",
+        "Van-Thuan Pham"
+      ],
+      "year": 2023,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1145/3597503.3608133",
+      "arxiv_id": "2301.09258",
+      "s2_paper_id": "d2bf26b7ec662c29c30e6ad4548ce53d360718f2",
+      "url": "https://doi.org/10.1145/3597503.3608133",
+      "open_access_pdf": "https://arxiv.org/pdf/2301.09258",
+      "abstract": "APIs often transmit far more data to client applications than they need, and in the context of web applications, often do so over public channels. This issue, termed Excessive Data Exposure (EDE), was OWASP's third most significant API vulnerability of 2019. However, there are few automated tools-either in research or industry-to effectively find and remediate such issues. This is unsurprising as the problem lacks an explicit test oracle: the vulnerability does not manifest through explicit abnormal behaviours (e.g., program crashes or memory access violations). In this work, we develop a metamorphic relation to tackle that challenge and build the first fuzzing tool-that we call EDEFUZZ-to systematically detect EDEs. EDEFuzz can significantly reduce false negatives that occur during manual inspection and ad-hoc text-matching techniques, the current most-used approaches. We tested EDEFuzz against the sixty-nine applicable targets from the Alexa Top-200 and found 33,365 potential leaks-illustrating our tool's broad applicability and scalability. In a more-tightly controlled experiment of eight popular websites in Australia, EDEFuzz achieved a high true positive rate of 98.65% with minimal configuration, illustrating our tool's accuracy and efficiency.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/d2bf26b7ec662c29c30e6ad4548ce53d360718f2",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/d2bf26b7ec662c29c30e6ad4548ce53d360718f2",
+        "source_type": "paper",
+        "content_sha256": "sha256:e45d313096826480a5604b7477475bac54b3d8cc5dbe1c19e56c6af3c8d9acb7"
+      }
+    },
+    {
+      "id": "paper:doi:10.1002/9781119880929.ch13",
+      "title": "Emerging Aspects of Software Fault Localization",
+      "authors": [
+        "T.H. Tse",
+        "David Lo",
+        "Alex Gorce",
+        "Michael Perscheid",
+        "Robert Hirschfeld",
+        "W. Eric Wong"
+      ],
+      "year": 2023,
+      "venue": null,
+      "doi": "10.1002/9781119880929.ch13",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.1002/9781119880929.ch13",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/W4366606136",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://openalex.org/W4366606136",
+        "source_type": "paper",
+        "content_sha256": "sha256:fdfeca6a512ad0479238b044b4575b66c1d33a542516a98129774dfcacb06774"
+      },
+      "abstract": null
+    },
+    {
+      "id": "paper:doi:10.1145/3597926.3598068",
+      "title": "Exploring Missed Optimizations in WebAssembly Optimizers",
+      "authors": [
+        "Zhibo Liu",
+        "Dongwei Xiao",
+        "Zongjie Li",
+        "Shuai Wang",
+        "W. Meng"
+      ],
+      "year": 2023,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3597926.3598068",
+      "arxiv_id": null,
+      "s2_paper_id": "dc1cc47e14653eb9a834f44a39273c1a2e2de4ae",
+      "url": "https://doi.org/10.1145/3597926.3598068",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597926.3598068",
+      "abstract": "The prosperous trend of deploying complex applications to web browsers has boosted the development of WebAssembly (wasm) compilation toolchains. Software written in different high-level programming languages are compiled into wasm executables, which can be executed fast and safely in a virtual machine. The performance of wasm executables depends highly on compiler optimizations. Despite the prosperous use of wasm executables, recent research has indicated that real-world wasm applications are slower than anticipated, suggesting deficiencies in wasm optimizations. This paper aims to present the first systematic and in-depth understanding of the status quo of wasm optimizations. To do so, we present DITWO, a differential testing framework to uncover missed optimizations (MO) of wasm optimizers. DITWO compiles a C program into both native x86 executable and wasm executable, and differentiates optimization indication traces (OITraces) logged by running each executable to uncover MO. Each OITrace is composed with global variable writes and function calls, two performance indicators that practically and systematically reflect the optimization degree across wasm and native executables. Our analysis of the official wasm optimizer, wasm-opt, successfully identifies 1,293 inputs triggering MO of wasm-opt. With extensive manual effort, we identify nine root causes for all MO, and we estimate that fixing discovered MO can result in a performance improvement of at least 17.15%. We also summarize four lessons from our findings to deliver better wasm optimizations.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598068",
+              "source_type": "paper_citation_context",
+              "excerpt": "Differential testing (DT) is used in different software domains, including databases [66, 72], Java Virtual Machines (JVMs) [21, 22], symbolic execution engines [45], disassemblers [63], decompilers [54], and deep learning systems [37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2ac4cf49624364223739f1dd9e05ea68f2915ae05ab781ac7940acf110281c17",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/dc1cc47e14653eb9a834f44a39273c1a2e2de4ae",
+        "source_type": "paper",
+        "content_sha256": "sha256:fa157b0f8847290599c4ba6532bd085ef3264956f137aa05d9bc29396a50b152"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3552326.3587448",
+      "title": "Fail through the Cracks: Cross-System Interaction Failures in Modern Cloud Systems",
+      "authors": [
+        "Lilia Tang",
+        "Chaitanya Bhandari",
+        "Yongle Zhang",
+        "Anna Karanika",
+        "Shuyang Ji",
+        "Indranil Gupta",
+        "Tianyi Xu"
+      ],
+      "year": 2023,
+      "venue": "European Conference on Computer Systems",
+      "doi": "10.1145/3552326.3587448",
+      "arxiv_id": null,
+      "s2_paper_id": "916557d95f3d95a83d3f3503350c05c42e8c2272",
+      "url": "https://doi.org/10.1145/3552326.3587448",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3552326.3587448",
+      "abstract": "Modern cloud systems are orchestrations of independent and interacting (sub-)systems, each specializing in important services (e.g., data processing, storage, resource management, etc.). Hence, cloud system reliability is affected not only by the reliability of each individual system, but also by the interplay between these systems. We observe that many recent production incidents of cloud systems are manifested through interactions across the system boundaries. However, there is a lack of systematic understanding of this emerging mode of failures, which we term as cross-system interaction failures (or CSI failures). This hinders the development of better design, integration practices, and new tooling. In this paper, we discuss cross-system interaction failures based on analyses of (1) 11 CSI-failure-induced cloud incidents of Google, Azure, and AWS, and (2) 120 CSI failure cases of seven widely co-deployed open-source systems. We focus on understanding discrepancies between interacting systems as the root causes of CSI failures---CSI failures cannot be understood by analyzing one single system in isolation. This paper draws attention to this emerging failure mode, provides a comprehensive understanding of CSI failure patterns, and discusses potential approaches for mitigation. We advocate for cross-system testing and verification and demonstrate its potential by cross-testing the Spark-Hive data plane and exposing 15 new discrepancies.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3552326.3587448",
+              "source_type": "paper_citation_context",
+              "excerpt": ", [1, 22, 33]), which also exist in traditional systems that follow POSIX or SQL standards [94, 98], but are magnified by the heterogeneity and composability of cloud APIs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:af60d7a6e28620e80c417bc84fb8ce777d2ce01c80c60b477593856e58c929da",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/916557d95f3d95a83d3f3503350c05c42e8c2272",
+        "source_type": "paper",
+        "content_sha256": "sha256:790a487d913548b6e3b2d5a2a53c31a97a3e9926992346ac2faac5ff357c41d2"
+      }
+    },
+    {
+      "id": "paper:arxiv:2304.10044",
+      "title": "Finding Bug-Inducing Program Environments",
+      "authors": [
+        "Z. Mirzamomen",
+        "Marcel Böhme"
+      ],
+      "year": 2023,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2304.10044",
+      "s2_paper_id": "e10d0a78b6e4dd6e074f925e669bf632b5d48f53",
+      "url": "https://arxiv.org/abs/2304.10044",
+      "open_access_pdf": "https://arxiv.org/pdf/2304.10044",
+      "abstract": "Some bugs cannot be exposed by program inputs, but only by certain program environments. During execution, most programs access various resources, like databases, files, or devices, that are external to the program and thus part of the program's environment. In this paper, we present a coverage-guided, mutation-based environment synthesis approach of bug-inducing program environments. Specifically, we observe that programs interact with their environment via dedicated system calls and propose to intercept these system calls (i) to capture the resources accessed during the first execution of an input as initial program environment, and (ii) mutate copies of these resources during subsequent executions of that input to generate slightly changed program environments. Any generated environment that is observed to increase coverage is added to the corpus of environment seeds and becomes subject to further fuzzing. Bug-inducing program environments are reported to the user. Experiments demonstrate the effectiveness of our approach. We implemented a prototype called AFLChaos which found bugs in the resource-handling code of five (5) of the seven (7) open source projects in our benchmark set (incl. OpenSSL). Automatically, AFLChaos generated environments consisting of bug-inducing databases used for storing information, bug-inducing multimedia files used for streaming, bug-inducing cryptographic keys used for encryption, and bug-inducing configuration files used to configure the program. To support open science, we publish the experimental infrastructure, our tool, and all data.",
+      "cites_seed_techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2304.10044",
+              "source_type": "paper_citation_context",
+              "excerpt": "Our experimental results on seven widely used protocol implementations showed outstanding increase in code coverage for AFLChaos in comparison with AFLNet as the base and resulted in discovering 13 bugs in these widely-used and well-fuzzed programs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e666d533c8d368ef0b17e06b741b3b31a7065ebb709abceb852f4fcc1dc43f0c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/e10d0a78b6e4dd6e074f925e669bf632b5d48f53",
+        "source_type": "paper",
+        "content_sha256": "sha256:7c20390b86c4c8c323750698beeda9c6f5f282e48beef291e172e544717fae14"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3582016.3582053",
+      "title": "Finding Unstable Code via Compiler-Driven Differential Testing",
+      "authors": [
+        "Shaohua Li",
+        "Zhendong Su"
+      ],
+      "year": 2023,
+      "venue": "International Conference on Architectural Support for Programming Languages and Operating Systems",
+      "doi": "10.1145/3582016.3582053",
+      "arxiv_id": null,
+      "s2_paper_id": "03e0d8d1839045a6fd25080bbfa945b2801a2f3f",
+      "url": "https://doi.org/10.1145/3582016.3582053",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3582016.3582053",
+      "abstract": "Unstable code refers to code that has inconsistent or unstable run-time semantics due to undefined behavior (UB) in the program. Compilers exploit UB by assuming that UB never occurs, which allows them to generate efficient but potentially semantically inconsistent binaries. Practitioners have put great research and engineering effort into designing dynamic tools such as sanitizers for frequently occurring UBs. However, it remains a big challenge how to detect UBs that are beyond the reach of current techniques. In this paper, we introduce compiler-driven differential testing (CompDiff), a simple yet effective approach for finding unstable code in C/C++ programs. CompDiff relies on the fact that when compiling unstable code, different compiler implementations may produce semantically inconsistent binaries. Our main approach is to examine the outputs of different binaries on the same input. Discrepancies in outputs may signify the existence of unstable code. To detect unstable code in real-world programs, we also integrate CompDiff into AFL++, the most widely-used and actively-maintained general-purpose fuzzer. Despite its simplicity, CompDiff is effective in practice: on the Juliet benchmark programs, CompDiff uniquely detected 1,409 bugs compared to sanitizers; on 23 popular open-source C/C++ projects, CompDiff-AFL++ uncovered 78 new bugs, 52 of which have been fixed by developers and 36 cannot be detected by sanitizers. Our evaluation also reveals the fact that CompDiff is not designed to replace current UB detectors but to complement them.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3582016.3582053",
+              "source_type": "paper_citation_context",
+              "excerpt": "It aims at improving fault tolerance of software by having N independent individuals or groups implementing the same specification.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ea56918800b823de4a22d81f089f2fd81925f3fc29b7bcb2b3503ed1b137044b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/03e0d8d1839045a6fd25080bbfa945b2801a2f3f",
+        "source_type": "paper",
+        "content_sha256": "sha256:974e20ea9ae1d9e1f91c3dc93df393fce906e533d2db594b246c4456dd5e6679"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3597926.3598046",
+      "title": "GDsmith: Detecting Bugs in Cypher Graph Database Engines",
+      "authors": [
+        "Ziyue Hua",
+        "Weisheng Lin",
+        "Luyao Ren",
+        "Zongyang Li",
+        "Lu Zhang",
+        "Wenpin Jiao",
+        "Tao Xie"
+      ],
+      "year": 2023,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3597926.3598046",
+      "arxiv_id": null,
+      "s2_paper_id": "6c755fc901d0b41a5d73c265f64a5aacf62e83b8",
+      "url": "https://doi.org/10.1145/3597926.3598046",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597926.3598046",
+      "abstract": "Graph database engines stand out in the era of big data for their efficiency of modeling and processing linked data. To assure high quality of graph database engines, it is highly critical to conduct automatic test generation for graph database engines, e.g., random test generation, the most commonly adopted approach in practice. However, random test generation faces the challenge of generating complex inputs (i.e., property graphs and queries) for producing non-empty query results; generating such type of inputs is important especially for detecting wrong-result bugs. To address this challenge, in this paper, we propose GDsmith, the first approach for testing Cypher graph database engines. GDsmith ensures that each randomly generated query satisfies the semantic requirements. To increase the probability of producing complex queries that return non-empty results, GDsmith includes two new techniques: graph-guided generation of complex pattern combinations and data-guided generation of complex conditions. Our evaluation results demonstrate that GDsmith is effective and efficient for producing complex queries that return non-empty results for bug detection, and substantially outperforms the baselines. GDsmith successfully detects 28 bugs on the released versions of three highly popular open-source graph database engines and receives positive feedback from their developers.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598046",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [19] detects bugs in a relational database engine by applying a semantic-preserving transformation to a given SQL query to disable the engine’s optimizations and addresses PQS’ high implementation effort.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:965c2521c787daf7a2285cdfa60bb3f798b3046aafdcbd06c83b081a15b8172f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598046",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS [21] detects wrong-result bugs by checking whether a specific record is fetched correctly.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:434e9f3fb06e589d819e8bd5dc198f590e66a37a8becfaf0aa062f051214b104",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598046",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [20] derives multiple SQL queries that compute a partial result of the initial query.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0266278f07037034da919e0354e23615c6a06e5fafd99b6394b35fe59bc6e15b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3597926_3598046.json",
+              "source_type": "paper",
+              "excerpt": "Its framework is derived from SQLancer [ 18] (which is a tool to automatically test relational database engines).",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, 4.1.2 Implementation. We implement the GDsmith prototype with, page 8.",
+              "content_sha256": "sha256:e3b00522eb55477eeaa25255ce62853e78a9ffe4736f0c6b62f0e444aa8c8829",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/6c755fc901d0b41a5d73c265f64a5aacf62e83b8",
+        "source_type": "paper",
+        "content_sha256": "sha256:897d043eb52457eb05720b5300ab24e0aeb50166757da106a696dad0dd0a3194"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/ddaa2000/GDsmith",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "renamed_sqlancer_package",
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/ddaa2000/GDsmith",
+              "source_type": "github_repository",
+              "excerpt": "# GDsmith",
+              "note": "Repository is named after GDsmith, the tool this paper says it built, and is about database testing.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/ddaa2000/GDsmith/blob/master/src/main/java/org/example/gdsmith/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.ALPHANUMERIC;",
+              "excerpt_is_verbatim": true,
+              "note": "src/main/java/org/example/gdsmith/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.example.gdsmith (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:d09fa9dddb2efa23e9184c0a8e2c71c39cdaffe6bde161da903434f62c4c4711",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1109/icse48619.2023.00173",
+      "title": "Generating Test Databases for Database-Backed Applications",
+      "authors": [
+        "Cong Yan",
+        "Suman Nath",
+        "Shan Lu"
+      ],
+      "year": 2023,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse48619.2023.00173",
+      "arxiv_id": null,
+      "s2_paper_id": "605b3d1b6cbcdaab3f2cd9ec0002ec6dd8050c72",
+      "url": "https://doi.org/10.1109/icse48619.2023.00173",
+      "open_access_pdf": null,
+      "abstract": "Database-backed applications are widely used. To effectively test these applications, one needs to design not only user inputs but also database states, which imposes unique challenges. First, valid database states have to satisfy complicated constraints determined by application semantics, and hence are difficult to synthesize. Second, the state space of a database is huge, as an application can contain tens to hundreds of tables with up to tens of fields per table. Making things worse, each test involving database operations takes significant time to run. Consequently, unhelpful database states and running tests on them can severely waste testing resources. We propose DBGRILLER, a tool that generates database states to facilitate thorough testing of database-backed applications. To effectively generate valid database states, DBGRILLER strategically injects minor mutation into existing database states and transforms part of the application-under-test into a stand-alone validity checker. To tackle the huge database state space and save testing time, DBGRILLER uses program analysis to identify a novel branch-projected DB view that can be used to filter out database states that are unlikely to increase the testing branch coverage. Our evaluation on 9 popular open-source database applications shows that DBGRILLER can effectively increase branch coverage of existing tests and expose previously unknown bugs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00173",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [21], [42] includes a series of techniques, including query partitioning and pivot query synthesis, to generate SQL queries paired with certain properties of query results.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:13981cbcde2970107e2bc9aa1c91c44323a028fc8e2ef77246da5daf4271dca3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/605b3d1b6cbcdaab3f2cd9ec0002ec6dd8050c72",
+        "source_type": "paper",
+        "content_sha256": "sha256:2025668e99034db67537e2d57cac54a087883a8c64d10f4c266fd9c0f1a13b99"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3597926.3598130",
+      "title": "GrayC: Greybox Fuzzing of Compilers and Analysers for C",
+      "authors": [
+        "Karine Even-Mendoza",
+        "Arindam Sharma",
+        "Alastair F. Donaldson",
+        "Cristian Cadar"
+      ],
+      "year": 2023,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3597926.3598130",
+      "arxiv_id": null,
+      "s2_paper_id": "86a60acfe93752af1159dea65d01808134754467",
+      "url": "https://doi.org/10.1145/3597926.3598130",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597926.3598130",
+      "abstract": "Fuzzing of compilers and code analysers has led to a large number of bugs being found and fixed in widely-used frameworks such as LLVM, GCC and Frama-C. Most such fuzzing techniques have taken a blackbox approach, with compilers and code analysers starting to become relatively immune to such fuzzers. We propose a coverage-directed, mutation-based approach for fuzzing C compilers and code analysers, inspired by the success of this type of greybox fuzzing in other application domains. The main challenge of applying mutation-based fuzzing in this context is that naive mutations are likely to generate programs that do not compile. Such programs are not useful for finding deep bugs that affect optimisation, analysis, and code generation routines. We have designed a novel greybox fuzzer for C compilers and analysers by developing a new set of mutations to target common C constructs, and transforming fuzzed programs so that they produce meaningful output, allowing differential testing to be used as a test oracle, and paving the way for fuzzer-generated programs to be integrated into compiler and code analyser regression test suites. We have implemented our approach in GrayC, a new open-source LibFuzzer-based tool, and present experiments showing that it provides more coverage on the middle- and back-end stages of compilers and analysers compared to other mutation-based approaches, including Clang-Fuzzer, PolyGlot, and a technique similar to LangFuzz. We have used GrayC to identify 30 confirmed compiler and code analyser bugs: 25 previously unknown bugs (with 22 of them already fixed in response to our reports) and 5 confirmed bugs reported independently shortly before we found them. A further 3 bug reports are under investigation. Apart from the results above, we have contributed 24 simplified versions of coverage-enhancing test cases produced by GrayC to the Clang/LLVM test suite, targeting 78 previously uncovered functions in the LLVM codebase.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/86a60acfe93752af1159dea65d01808134754467",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/86a60acfe93752af1159dea65d01808134754467",
+        "source_type": "paper",
+        "content_sha256": "sha256:4d48cc244de2ef032f0fb9b3f3551364608b38bbfdd75eb1ae60a67e3be22205"
+      }
+    },
+    {
+      "id": "paper:s2:c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+      "title": "Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis",
+      "authors": [
+        "Zongyin Hao",
+        "Quanfeng Huang",
+        "Chengpeng Wang",
+        "Jianfeng Wang",
+        "Yushan Zhang",
+        "Rongxin Wu",
+        "Charles Zhang"
+      ],
+      "year": 2023,
+      "venue": "USENIX Annual Technical Conference",
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+      "url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+              "source_type": "paper_citation_context",
+              "excerpt": "N O REC [34] transforms an optimized version of a query into a non-optimized one by the customized rule, e.g., changing “ SELECT * FROM t WHERE p” into “ SELECT (p IS TRUE ) FROM t.” Compared with the aforementioned two categories of approaches, metamorphic testing based approaches are much more…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:240077c3158ff8dd40a8cb0d04096fe6224041c3dfea7db3e92b0ecaaae14353",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+              "source_type": "paper_citation_context",
+              "excerpt": "…e.g., changing “ SELECT * FROM t WHERE p” into “ SELECT (p IS TRUE ) FROM t.” Compared with the aforementioned two categories of approaches, metamorphic testing based approaches are much more lightweight to implement and have been proven to be more effective in detecting logical bugs [34, 35].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bb49648bb0c089bf79e03236aa4d13c01a48d89bdb878d6b1e491a1d913766b1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+              "source_type": "paper_citation_context",
+              "excerpt": "Recent years have witnessed tremendous efforts in resolving the test oracle for logical bug detection in the DBMSs. Notably, the metamorphic testing based approach has been recognized to be state-of-the-art in DBMS testing for logical bug detection [35, 37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7442ad4133faf45fe55bd4b583f0f6c48aa25f4542299f394bc74d3c831639a3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+              "source_type": "paper_citation_context",
+              "excerpt": "We compared P INOLO with the three state-of-the-art logical bug detection techniques, namely PQS [36], N O REC [34], and TLP [35], respectively, which correspond to three kinds of test oracles.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:33bc15c13bd7643ebf0ff37e66264766e3cfcddbbd1702a1039d64213dff904f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+              "source_type": "paper_citation_context",
+              "excerpt": "Compared with the existing techniques [34–36], P INOLO considers more SQL features, such as set operators, arithmetic expressions, sub-queries, etc.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a5ae31a3fabdfd53f4ffba33c55005ffba3f45d12b0cf76937cad9b2054da3ba",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, TLP [35] decomposes a query q into three partitioning sub-queries, each of which computes the result sets for a boolean predicate to be evaluated as TRUE , FALSE , and NULL , respectively, and then constructs an equivalent query q ′ by performing the union operation on these three…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6ad9df3e74ecb6d646b7d7bb77faaa75b2c9391cc4ccd94f6e683c1d7934c2eb",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json",
+              "source_type": "paper",
+              "excerpt": "For example, the logical bug shown in Figure 1 can not be revealed by NOREC [34] and TLP [35], as the transformations preserve all the operators and the functions, still triggering the buggy evaluation process.",
+              "excerpt_is_verbatim": true,
+              "note": "M13 in the paper's extracted text, 2.2 Logical Bugs in DBMSs, page 4.",
+              "content_sha256": "sha256:5cf2a166d2e3cf990ee8621cc566ee39ffc162dcc874a8ae34d16066f20a3d7f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json",
+              "source_type": "paper",
+              "excerpt": "Compared with the existing techniques [ 34–36],PINOLO considers more SQL features, such as set operators, arithmetic expressions, sub-queries, etc.",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, 3.5 Summary, page 8.",
+              "content_sha256": "sha256:16be94e83307a76031b0048407d132e8d582ac6cad947e9760ad4c97d9569977",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json",
+              "source_type": "paper",
+              "excerpt": "Notably, the metamorphic testing based approach has been recognized to be state-of-the-art in DBMS testing for logical bug detection [ 35,37].",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, 2.2 Logical Bugs in DBMSs, page 4.",
+              "content_sha256": "sha256:5c87d4ba0ff60a7db9fa53510e038097b1ca217bc19231222c02b4e3a56de60a",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+        "source_type": "paper",
+        "content_sha256": "sha256:e924036bad00c895fb121e0588703e58537e2d3fbe10148729a4d5ded5117d6c"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3611643.3616286",
+      "title": "Property-Based Fuzzing for Finding Data Manipulation Errors in Android Apps",
+      "authors": [
+        "Jingling Sun",
+        "Ting Su",
+        "Jiayi Jiang",
+        "Jue Wang",
+        "G. Pu",
+        "Zhendong Su"
+      ],
+      "year": 2023,
+      "venue": "ESEC/SIGSOFT FSE",
+      "doi": "10.1145/3611643.3616286",
+      "arxiv_id": null,
+      "s2_paper_id": "8ea47327053957637ce8782a8b22aad19878dea1",
+      "url": "https://doi.org/10.1145/3611643.3616286",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3611643.3616286",
+      "abstract": "Like many software applications, data manipulation functionalities( DMFs ) are prevalent in Android apps, which perform the common CRUD operations (create, read, update, delete) to handle app-specific data. Thus, ensuring the correctness of these DMFs is fundamentally important for many core app functionalities. However, the bugs related to DMFs (named as data manipulation errors, DMEs ), especially those non-crashing logic ones, are prevalent but difficult to find. To this end, inspired by property-based testing, we introduce a property-based fuzzing approach to effectively finding DMEs in Android apps. Our key idea is that, given some type of app data of interest, we randomly interleave its relevant DMFs and other possible events to explore diverse app states for thorough validation. Specifically, our approach characterizes DMFs in (data) model-based properties and leverage the consistency between the data model and the UI layouts as the handler to do property checking. The properties of DMFs are specified by human according to specific app features. To support the application of our approach, we implemented an automated GUI testing tool, PBFDroid. We evaluated PBFDroid on 20 real-world Android apps, and successfully found 30 unique and previously unknown bugs in 18 apps. Out of the 30 bugs, 29 of which are DMEs (22 are non-crashing logic bugs, and 7 are crash ones). To date, 19 have been confirmed and 9 have already been fixed. Many of these bugs are non-trivial and lead to different types of app failures. Our further evaluation confirms that none of the 22 non-crashing DMEs can be found by the state-of-the-art techniques. In addition, a user study shows that the manual cost of specifying the DMF properties with the assistance of our tool is acceptable. Overall, given accurate DMF properties, our approach can automatically find DMEs without any false positives. We have made all the artifacts publicly available at:https:// github.com/ property-based-fuzzing/ home.",
+      "cites_seed_techniques": [
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3611643.3616286",
+              "source_type": "paper_citation_context",
+              "excerpt": "Some work [51, 52] uses metamorphic testing to find the CRUD errors in database management systems.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:63bc9436da856502f793b40ee22dd9462cb4fa6c7e5925d606ac463e9efd83b3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/8ea47327053957637ce8782a8b22aad19878dea1",
+        "source_type": "paper",
+        "content_sha256": "sha256:5596c7b186664a4755a6cd8f8e74bfed7f3e05ed79c2f5ffb40680a38a955ff9"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse-companion58688.2023.00041",
+      "title": "Randomized Differential Testing of RDF Stores",
+      "authors": [
+        "Rui Yang",
+        "Yingying Zheng",
+        "Leile Tang",
+        "Wensheng Dou",
+        "Wei Wang",
+        "Jun Wei"
+      ],
+      "year": 2023,
+      "venue": "2023 IEEE/ACM 45th International Conference on Software Engineering: Companion Proceedings (ICSE-Companion)",
+      "doi": "10.1109/icse-companion58688.2023.00041",
+      "arxiv_id": null,
+      "s2_paper_id": "0b083953239d0617b534a0d76762fa8350db4e6f",
+      "url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+      "open_access_pdf": null,
+      "abstract": "As a special kind of graph database systems, RDF stores have been widely used in many applications, e.g., knowl-edge graphs and semantic web. RDF stores utilize SPARQL as their standardized query language to store and retrieve RDF graphs. Incorrect implementations of RDF stores can introduce logic bugs that cause RDF stores to return incorrect query results. These logic bugs can lead to severe consequences and are likely to go unnoticed by developers. However, no available tools can detect logic bugs in RDF stores. In this paper, we propose RD2, a Randomized Differential testing approach of RDF stores, to reveal discrepancies among RDF stores, which indicate potential logic bugs in RDF stores. The core idea of RD2 is to build an equivalent RDF graph for multiple RDF stores, and verify whether they can return the same query result for a given SPARQL query. Guided by the SPARQL syntax and the generated RDF graph, we automatically generate syntactically valid SPARQL queries, which can return non-empty query results with high probability. We further unify the formats of SPARQL query results from different RDF stores and find discrepancies among them. We evaluate RD2 on three popular and widely-used RDF stores. In total, we have detected 5 logic bugs in them. A video demonstration of RD2 is available at httos://youtu.be/da7XlsdbRR4.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+              "source_type": "paper_citation_context",
+              "excerpt": "RAGS [14], APOLLO [31] and TAQO [32] utilize differential testing for detecting bugs in RDBMSs. SQLsmith [33] is used to detect bugs causing exceptions or crashes in RDBMSs. TLP [15], NoREC [16], PQS [17] and DQE [18] develop various test oracles to detect logic bugs and optimization bugs, and have found many bugs in popular RDBMSs. DT2 [19] and Troc [20] detect transaction bugs in RDBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:dca29dfad4ddb3c0960596ad22d64b183dbf3d291dd5a1383eee9f5fdce78f15",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+              "source_type": "paper_citation_context",
+              "excerpt": "…[32] utilize differential testing for detecting bugs in RDBMSs. SQLsmith [33] is used to detect bugs causing exceptions or crashes in RDBMSs. TLP [15], NoREC [16], PQS [17] and DQE [18] develop various test oracles to detect logic bugs and optimization bugs, and have found many bugs in popular…",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8ded50778fd43f7c69af8c13642afa03c78431f1ba7ac6b66b87cd3b43f501f1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [15], NoREC [16], PQS [17] and DQE [18] develop various test oracles to detect logic bugs and optimization bugs, and have found many bugs in popular RDBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:db6e35885fd0c86bdd708331847fd48cbf03caa2516f29d36c84ba6d3ec93fa8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+              "source_type": "paper_citation_context",
+              "excerpt": "Many approaches [14]–[20], [25]–[33] are proposed to find bugs in RDBMSs that use SQL as a standardized query language.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1e4a96a63d6aea668b46e70d3e4e66caff0ec23f5ecee0d20872a74133124445",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/0b083953239d0617b534a0d76762fa8350db4e6f",
+        "source_type": "paper",
+        "content_sha256": "sha256:7667ba25bbabee7a65c91eaa6769971f7152bc7eba100793b31743579f776efe"
+      }
+    },
+    {
+      "id": "paper:arxiv:2310.06433",
+      "title": "Retromorphic Testing: A New Approach to the Test Oracle Problem",
+      "authors": [
+        "Boxi Yu",
+        "Qiuyang Mang",
+        "Qingshuo Guo",
+        "Pinjia He"
+      ],
+      "year": 2023,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2310.06433",
+      "s2_paper_id": "cc095f9ce71982a8dcc0ea1b813a7486b3f6c962",
+      "url": "https://arxiv.org/abs/2310.06433",
+      "open_access_pdf": "https://arxiv.org/pdf/2310.06433",
+      "abstract": "A test oracle serves as a criterion or mechanism to assess the correspondence between software output and the anticipated behavior for a given input set. In automated testing, black-box techniques, known for their non-intrusive nature in test oracle construction, are widely used, including notable methodologies like differential testing and metamorphic testing. Inspired by the mathematical concept of inverse function, we present Retromorphic Testing, a novel black-box testing methodology. It leverages an auxiliary program in conjunction with the program under test, which establishes a dual-program structure consisting of a forward program and a backward program. The input data is first processed by the forward program and then its program output is reversed to its original input format using the backward program. In particular, the auxiliary program can operate as either the forward or backward program, leading to different testing modes. The process concludes by examining the relationship between the initial input and the transformed output within the input domain. For example, to test the implementation of the sine function $\\sin(x)$, we can employ its inverse function, $\\arcsin(x)$, and validate the equation $x = \\sin(\\arcsin(x)+2k\\pi), \\forall k \\in \\mathbb{Z}$. In addition to the high-level concept of Retromorphic Testing, this paper presents its three testing modes with illustrative use cases across diverse programs, including algorithms, traditional software, and AI applications.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2310.06433",
+              "source_type": "paper_citation_context",
+              "excerpt": "Metamorphic testing is also used for testing various systems such as compilers [14, 15], database engines [13, 19, 20], SMT solvers [34], Android apps [28, 29], quantum computing platforms [1, 17], and AI systems [30, 32, 33, 36, 37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ad91e2f0ca3e218c06da0005eb3c081826a820580743b2792b67df219bd3f9fa",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2310.06433",
+              "source_type": "paper_citation_context",
+              "excerpt": "In the context of Database Management Systems (DBMS), Rigger and Su introduced the concept of Pivoted Query Synthesis (PQS) [21] as an effective testing approach to uncover logic bugs in these systems.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9af8457f39ef1210653b9f287fb10055ca7dc6c70e074fd2089614309b56e94b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2310.06433",
+              "source_type": "paper_citation_context",
+              "excerpt": "Although the original paper describes PQS as a testing technique for a specific system, we think it can be regarded as an instance of the general Retromorphic Testing methodology.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:716104a4eaa702eba5b2a98169c961fe91810907a2bda1fb1b28e1d8a217b69e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2310.06433",
+              "source_type": "paper_citation_context",
+              "excerpt": "The PQS testing approach initiates with the generation of a pivot row ( e.g., [t0.c0: 3, t0.c1: TRUE, t1.c0: -5] in Fig.2), which is in the modality of rows.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a5d2479c94b1dab1f0a86d4d2d67a906a53b7800d8770ca5d44203cfe0c823d3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2310.06433",
+              "source_type": "paper_citation_context",
+              "excerpt": "We also discussed with one author of PQS and he agrees that PQS should be categorized as a Retromorphic Testing technique.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4fe3cdb755d6dcba2c3288c1a1425a02328e493ef402ce2632fdbb70584a484e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2310_06433.json",
+              "source_type": "paper",
+              "excerpt": "Although the original paper describes PQS as a testing technique for a specific system, we think it can be regarded as an instance of the general Retromorphic Testing methodology.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, 4 TESTING SCOPES, page 7.",
+              "content_sha256": "sha256:716104a4eaa702eba5b2a98169c961fe91810907a2bda1fb1b28e1d8a217b69e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2310_06433.json",
+              "source_type": "paper",
+              "excerpt": "We also discussed with one author of PQS and he agrees that PQS should be categorized as a Retromorphic Testing technique.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 4 TESTING SCOPES, page 7.",
+              "content_sha256": "sha256:4fe3cdb755d6dcba2c3288c1a1425a02328e493ef402ce2632fdbb70584a484e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/cc095f9ce71982a8dcc0ea1b813a7486b3f6c962",
+        "source_type": "paper",
+        "content_sha256": "sha256:678edd25e28709b818af80996d9d686772231ca9763a9b3b62145e650e14eddc"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icde55515.2023.00057",
+      "title": "Sequence-Oriented DBMS Fuzzing",
+      "authors": [
+        "Jie Liang",
+        "Yaoguang Chen",
+        "Zhiyong Wu",
+        "Jingzhou Fu",
+        "Mingzhe Wang",
+        "Yu Jiang",
+        "Xiangdong Huang",
+        "Ting Chen",
+        "Jiashui Wang",
+        "Jiajia Li"
+      ],
+      "year": 2023,
+      "venue": "IEEE International Conference on Data Engineering",
+      "doi": "10.1109/icde55515.2023.00057",
+      "arxiv_id": null,
+      "s2_paper_id": "058f7a552121c7417e783c995f8bc584d09b3eff",
+      "url": "https://doi.org/10.1109/icde55515.2023.00057",
+      "open_access_pdf": null,
+      "abstract": "The SQL specification consists of hundreds of statement types, which leads to difficulties in DBMS fuzzing: state-of-the-art works generally reuse the statements of predefined types; the limited types cannot cover the full input space and test the corresponding logic consequently. In this paper, we propose Lego, a fuzzer to generate SQL sequences with abundant types to improve DBMS fuzzing coverage. The key idea of sequence generation is type-affinity, which indicates the meaningful occurrence of SQL type pairs (e.g., INSERT and SELECT). During each fuzzing iteration, Lego first proactively explores SQL statements of different types and analyzes affinities with coverage feedback. Next, when a new affinity is discovered, Lego synthesizes new SQL sequences containing the types progressively.We evaluate Lego on PostgreSQL, MySQL, MariaDB, and Comdb2 against SQLancer, SQLsmith, and Squirrel. The sequence-oriented fuzzing helps Lego outperform other fuzzers on branch coverage by 44%–198%. More importantly, in the continuous fuzzing, Lego has discovered 102 new vulnerabilities confirmed by the corresponding vendors, including 6 bugs in PostgreSQL, 21 bugs in MySQL, 42 bugs in MariaDB, and 33 bugs in Comdb2. Among them, 22 CVEs have been assigned due to their severe security influences.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+              "source_type": "paper_citation_context",
+              "excerpt": "Security vulnerabilities, especially memory bugs such as buffer overflow are particularly dangerous for DBMS because they might allow attackers to steal information, tamper data, crash systems, and bring heavy losses [4, 10, 32, 35, 51, 54].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:468d87f1d5c296820fc67c7cdc6fa608738f52d9ada160dcf8feb5ed31a57e85",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+              "source_type": "paper_citation_context",
+              "excerpt": "The two metrics are used as the standard in fuzzing evaluation [8, 17, 44], and have been widely used in fuzzing works [35, 42, 54].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:67c9b9b5fa90c195ed8e2a40cf73dcc2593ef03de419daa08491e40d1518449d",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+              "source_type": "paper_citation_context",
+              "excerpt": "To test logic and performance bugs, many representative schemes utilize differential testing [16, 35, 39].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bc0a257cd09dfb8a6e1ef82ab2f8e54a3921cf4afbec167d8e031b100c059095",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+              "source_type": "paper_citation_context",
+              "excerpt": "Its following works [34, 33] also apply similar strategies by building functionally equivalent queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6cc7f1b19d8a66fa3e28094ad72a7405063c7ef64bbcc02a32c142c3888e87ca",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+              "source_type": "paper_citation_context",
+              "excerpt": "In general, fuzzers could be divided into generation-based [35, 37] and mutation-based [15, 51, 54].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2b5f131b0f7424d280ff417a085cb8a31637ba48703b7148de7ec24a9dd67473",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icde55515.2023.00057",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer [35] synthesizes queries to fetch a random row from existing tables in the target DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b9d45e418f9687ca1271c3a037ce1f179db5ae1400ed5196904ed8f2c1ce97a1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde55515_2023_00057.json",
+              "source_type": "paper",
+              "excerpt": "We evaluate LEGO on PostgreSQL, MySQL, MariaDB, and Comdb2 against SQLancer, SQLsmith, and SQUIRREL.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, page 1.",
+              "content_sha256": "sha256:1fbd3b7cfce7ff1ab159f1649d8cf1cb7280cfeae11f53cc3c4518bccf46e10d",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde55515_2023_00057.json",
+              "source_type": "paper",
+              "excerpt": "We evaluate LEGO on the latest version of PostgreSQL, MySQL, MariaDB, and Comdb2 against SQLancer, SQLsmith, and SQUIRREL.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, I INTRODUCTION, page 2.",
+              "content_sha256": "sha256:a9a0955fa8d023d8284c9f96b2b9908596e2cf28adbc637a30c5638daa23ae2e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde55515_2023_00057.json",
+              "source_type": "paper",
+              "excerpt": "The sequence-oriented fuzzing helps LEGO cover 198%, 44%, and 120% more branches than SQLancer, SQLsmith, and SQUIRREL on average, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, I INTRODUCTION, page 2.",
+              "content_sha256": "sha256:caa249e6bcbbf692233385dbb2f835be285a07d126bf7e6b17f29f84a96059a8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde55515_2023_00057.json",
+              "source_type": "paper",
+              "excerpt": "Specifically, SQLancer and SQLsmith did not find any bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, B DBMS Vulnerability Detection, page 8.",
+              "content_sha256": "sha256:9238620dc6261bc4c4a1ed7f6b0f82d4a785dadec7c1b3c5b89237a740339c94",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icde55515_2023_00057.json",
+              "source_type": "paper",
+              "excerpt": "To encompass as many state-of-the-art DBMS fuzzers as possible, we compared LEGO to popular fuzzer SQUIRREL and SQLancer from the academy and SQLsmith from the industry.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, A Evaluation Setup, page 7.",
+              "content_sha256": "sha256:b2695e455fc58d0fab6fcea2277b9fa6197661ef39e621dd2418d2a2c1610ffb",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/058f7a552121c7417e783c995f8bc584d09b3eff",
+        "source_type": "paper",
+        "content_sha256": "sha256:325701319cbaa031c50d62cb725cde22361e0e076c8def4b1a46ded2e65c1ebd"
+      }
+    },
+    {
+      "id": "paper:doi:10.32672/jnkti.v6i1.5830",
+      "title": "Siklus Hidup Pengembangan Sistem Basis Data Pada Sistem Informasi Buku Tamu di Badan Pusat Statistik Kabupaten Kediri Menggunakan MySQL",
+      "authors": [
+        "Ahmad Niamilah",
+        "A. Alfin",
+        "Iin Kurniasari"
+      ],
+      "year": 2023,
+      "venue": "Jurnal Nasional Komputasi dan Teknologi Informasi (JNKTI)",
+      "doi": "10.32672/jnkti.v6i1.5830",
+      "arxiv_id": null,
+      "s2_paper_id": "21db4963564e269f05ecb0ca2dffcacf43cdfb3d",
+      "url": "https://doi.org/10.32672/jnkti.v6i1.5830",
+      "open_access_pdf": "https://ojs.serambimekkah.ac.id/jnkti/article/download/5830/pdf",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.32672/jnkti.v6i1.5830",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQL memiliki lima bagian dalam pemrosesan data, yaitu data retrieving, data definition, data manipulation, data control, dan data transaction language[9].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:db83c72145e28dab6e96564bfd018c421b025465af898ff80c021dc41f463bdb",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/21db4963564e269f05ecb0ca2dffcacf43cdfb3d",
+        "source_type": "paper",
+        "content_sha256": "sha256:68f7971c6f21d61608d46e3fb6032dcf7a2cb9f33e7356cb52e7882d0510e6e5"
+      },
+      "abstract": "Abstrak - Proses pengembangan sistem informasi tentunya harus ditunjang juga dengan proses pengembangan basis data yang berkelanjutan. Pengembangan basis data yang berkelanjutan dapat menunjang penggunaan dan pengelolaan data pada sebuah sistem informasi. Database System Development Lifecycle (DSDLC) merupakan suatu metode untuk perancangan dan pengembangan basis data. Proses pengambilan dan perubahan data menggunakan Data Manipulation Language (DML). Structured Query Language (SQL) merupakan sebuah DML yang seringkali digunakan untuk proses modifikasi dalam relasional basis data. SQL berupa sintaks sederhana berisi perintah untuk proses pengambilan dan perubahan data, atau biasa disebut dengan query. Sistem informasi buku tamu di badan pusat statistik kabupaten kediri ini menggunakan Relational Database Management System (RDBMS) MySQL. Dalam penelitian ini, pernyataan yang diimplementasikan pada basis data Sistem Informasi Buku Tamu ada 4 (empat), yaitu Select, Insert, Update dan Delete. Pernyataan Select digunakan dalam proses pengujian pemanggilan data dari satu atau lebih tabel pada basis data. Pernyataaan Insert digunakan sebagai proses pengujian penyisipan data pada tabel basis data. Pernyataan Update digunakan sebagai proses pengujian modifikasi data pada tabel dalam basis data. Sedangkan pernyataan Delete digunakan dalam proses pengujian penghapusan satu atau lebih data pada basis data. Hasil dari penelitian ini berupa model perancangan basis data yang nantinya dapat diimplementasikan pada Sistem Informasi Buku Tamu berbasis website.Kata kunci: Data Manipulation Language, Database System Development Life Cycle, Relational Database Management System, Structured Query Language Abstract –The development of information system must be supported by a continuous database development process. The sustainability of database development is expected optimizing usability and management data in an information system. Database system development Lifecycle (DSDLC) is a method for designing and developing databases. The process of selecting and changing data uses Data Manipulation Languange (DML). Structured Query Language (SQL) is a DML that’s often used to process modification data in realtional databases. SQL is a simple syntax containing commands for the process of retrieving and changing data, or what’s oftenly called Query. The guest book information system at Kediri Central Agency of Statistics uses MySQL Relational Database Management System (RDBMS). In this research, there are 4 (four) commands was implemented in Guestbook Information System Database, namely Select, Insert, Update and Delete. The Select command used to retrieve data from one or more tables in the database. The Insert command used to insert data into a database table. The Update command used to modify data in tables in the database. While the Delete command is used to delete one or more data in the database. The results of this research will deliver a database model design that can be implemented on web based guest book information system.Keywords - Data Manipulation Language, Database System Development Life Cycle, Relational Database Management System, Structured Query Language"
+    },
+    {
+      "id": "paper:doi:10.3390/app13042519",
+      "title": "Squill: Testing DBMS with Correctness Feedback and Accurate Instantiation",
+      "authors": [
+        "Shihao Wen",
+        "Peng Jia",
+        "Pin Yang",
+        "Chi Hu"
+      ],
+      "year": 2023,
+      "venue": "Applied Sciences",
+      "doi": "10.3390/app13042519",
+      "arxiv_id": null,
+      "s2_paper_id": "b3692c4bdaaff8e7e0f9fdf68bd810d3d8275ebd",
+      "url": "https://doi.org/10.3390/app13042519",
+      "open_access_pdf": "https://mdpi.com/2076-3417/13/4/2519/pdf?version=1677053931",
+      "abstract": "Database Management Systems (DBMSs) are the core of management information systems. Thus, detecting security bugs or vulnerabilities of DBMSs is an essential task. In recent years, grey-box fuzzing has been adopted to detect DBMS bugs for its high effectiveness. However, the seed scheduling strategy of existing fuzzing techniques does not consider the seeds’ correctness, which is inefficient in finding vulnerabilities in DBMSs. Moreover, current tools cannot correctly generate SQL statements with nested structures, which limits their effectiveness. This paper proposes a fuzzing solution named Squill to address these challenges. First, we propose correctness-guided mutation to utilize the correctness of seeds as feedback to guide fuzzing. Second, Squill embeds semantics-aware instantiation to correctly fill semantics to SQL statements with nested structures by collecting the context information of AST nodes. We implemented Squill based on Squirrel and evaluated it on three popular DBMSs: MySQL, MariaDB, and OceanBase. In our experiment, Squill explored 29% more paths and found 3.4× more bugs than the existing tool. In total, Squill detected 30 bugs in MySQL, 27 in MariaDB, and 6 in OceanBase. Overall, 19 of the bugs are fixed with 9 CVEs assigned. The results show that Squill outperforms the previous fuzzer in terms of both code coverage and bug discovery.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.3390/app13042519",
+              "source_type": "paper_citation_context",
+              "excerpt": "Sqlancer [2–4] constructs different SQL statements of functionally equivalent through several different patterns and inputs them into the same DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:dae4137a6d9d9c0ec3e1c65b460c0276126e5f696963a9f3b0928d7f8fc1cfbf",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.3390/app13042519",
+              "source_type": "paper_citation_context",
+              "excerpt": "Black-box fuzzing, or generation-based fuzzing, has been extensively used in finding DBMS bugs, such as SQLsmith [1] and SQLancer [2–4].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:bc232da881942f9748818e519333edceef1e8f4b35c1a9222d45665c86a530d4",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/b3692c4bdaaff8e7e0f9fdf68bd810d3d8275ebd",
+        "source_type": "paper",
+        "content_sha256": "sha256:915eefa27871f3e434cd18967e0d6250c22350f0b1ebd9664f3f93f2b8bc1aee"
+      }
+    },
+    {
+      "id": "paper:doi:10.1007/978-3-031-51479-1_17",
+      "title": "Syntax-Aware Mutation for Testing the Solidity Compiler",
+      "authors": [
+        "Charalambos Mitropoulos",
+        "T. Sotiropoulos",
+        "S. Ioannidis",
+        "Dimitris Mitropoulos"
+      ],
+      "year": 2023,
+      "venue": "European Symposium on Research in Computer Security",
+      "doi": "10.1007/978-3-031-51479-1_17",
+      "arxiv_id": null,
+      "s2_paper_id": "372393a671600484f8859510cc5553b1345bdec7",
+      "url": "https://doi.org/10.1007/978-3-031-51479-1_17",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1007/978-3-031-51479-1_17",
+              "source_type": "paper_citation_context",
+              "excerpt": "Fuzzing has been used to identify bugs in miscellaneous entities such as system libraries [35], web and cloud applications [10], data-oriented systems [39,40], and compilers [48,34,19].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9210060d51c18407538f848db1cd3833b121bf05f7e34c0ab161c03e0b28239c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/372393a671600484f8859510cc5553b1345bdec7",
+        "source_type": "paper",
+        "content_sha256": "sha256:1217dc0913f2574c40b6f43f7b693dbdddd4ae08dc39f476b30db51cfef839bb"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse48619.2023.00174",
+      "title": "Testing Database Engines via Query Plan Guidance",
+      "authors": [
+        "Jinsheng Ba",
+        "Manuel Rigger"
+      ],
+      "year": 2023,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse48619.2023.00174",
+      "arxiv_id": "2312.17510",
+      "s2_paper_id": "ec682d9c7d68149dcd8932acd01a751f2f8b5611",
+      "url": "https://doi.org/10.1109/icse48619.2023.00174",
+      "open_access_pdf": "https://arxiv.org/pdf/2312.17510",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00174",
+              "source_type": "paper_citation_context",
+              "excerpt": "We used NoREC [7] and TLP [6], which are the state-of-the-art oracles supported by both SQLancer and SQLRight.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:dac0d8971e94387983945946105df0ef8b6c8cacfcf79eb1864a6cb8a0f93b4c",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00174",
+              "source_type": "paper_citation_context",
+              "excerpt": "We use the state-of-the-art logic-bug oracles NoREC [7] and TLP [6] to validate the queries’ results.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:42ffe6a510b6037b04e453fba905a5a040472fce008a0639e2c07f3a5e7ea0c8",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00174",
+              "source_type": "paper_citation_context",
+              "excerpt": "Subsequently, three test oracles were proposed and implemented in SQLancer [6]–[8].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:df370197ba13ec1df4fd988a3fd365174f3cea9533de7c461212706787f0bb90",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00174",
+              "source_type": "paper_citation_context",
+              "excerpt": "Non-optimizing Reference Engine Construction (NoREC) [7] checks for inconsistent results values of a predicate used in a query that the DBMS might optimize and one that is used in a query that is difficult to optimize.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d6829d61cfaf73d9dd005366082ede14ed647a39429fd3e6c2e683a940b89daa",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00174",
+              "source_type": "paper_citation_context",
+              "excerpt": "Furthermore, for some DBMSs, such as MySQL, many previously-reported bugs remain unfixed, impeding the testing process, which was also noted in prior work [6].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:667c755612618b8d306b5db921b320a4914d421b192ebba7dc06d38fad47ca40",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00174",
+              "source_type": "paper_citation_context",
+              "excerpt": "Recently, effective test oracles [6]–[8] have been proposed that brought validating the results of such queries within reach.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:47aaa0d56c937218ade49239febed92c7de505058287e4c81f2db7e4b9d6e755",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "deterministic",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00174",
+              "source_type": "paper_citation_context",
+              "excerpt": "We used NoREC [7] and TLP [6], which are the state-of-the-art oracles supported by both SQLancer and SQLRight.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Non-optimizing Reference Engine Construction (NoREC) as state of the art.",
+              "content_sha256": "sha256:dac0d8971e94387983945946105df0ef8b6c8cacfcf79eb1864a6cb8a0f93b4c",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00174",
+              "source_type": "paper_citation_context",
+              "excerpt": "We use the state-of-the-art logic-bug oracles NoREC [7] and TLP [6] to validate the queries’ results.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Non-optimizing Reference Engine Construction (NoREC) as state of the art.",
+              "content_sha256": "sha256:42ffe6a510b6037b04e453fba905a5a040472fce008a0639e2c07f3a5e7ea0c8",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/ec682d9c7d68149dcd8932acd01a751f2f8b5611",
+        "source_type": "paper",
+        "content_sha256": "sha256:86772ffc614b314f819f4cd3c366c2fd4a761e15453fd8cfeb8d25b4c602523b"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Database systems are widely used to store and query data. Test oracles have been proposed to find logic bugs in such systems, that is, bugs that cause the database system to compute an incorrect result. To realize a fully automated testing approach, such test oracles are paired with a test case generation technique; a test case refers to a database state and a query on which the test oracle can be applied. In this work, we propose the concept of Query Plan Guidance (QPG) for guiding automated testing towards “interesting” test cases. SQL and other query languages are declarative. Thus, to execute a query, the database system translates every operator in the source language to one of the potentially many so-called physical operators that can be executed; the tree of physical operators is referred to as the query plan. Our intuition is that by steering testing towards exploring a variety of unique query plans, we also explore more interesting behaviors-some of which are potentially incorrect. To this end, we propose a mutation technique that gradually applies promising mutations to the database state, causing the DBMS to create potentially unseen query plans for subsequent queries. We applied our method to three mature, widely-used, and extensively-tested database systems-SQLite, TiDB, and CockroachDB-and found 53 unique, previously unknown bugs. Our method exercises $4.85-408.48\\times$ more unique query plans than a naive random generation method and $7.46\\times$ more than a code coverage guidance method. Since most database systems-including commercial ones-expose query plans to the user, we consider QPG a generally applicable, black-box approach and believe that the core idea could also be applied in other contexts (e.g., to measure the quality of a test suite)."
+    },
+    {
+      "id": "paper:doi:10.1109/icse48619.2023.00175",
+      "title": "Testing Database Systems via Differential Query Execution",
+      "authors": [
+        "Jiansen Song",
+        "Wensheng Dou",
+        "Ziyu Cui",
+        "Qianwang Dai",
+        "Wei Wang",
+        "Jun Wei",
+        "Hua Zhong",
+        "Tao Huang"
+      ],
+      "year": 2023,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse48619.2023.00175",
+      "arxiv_id": null,
+      "s2_paper_id": "5724e138fc08329e0baa08d2090d9424df1cb945",
+      "url": "https://doi.org/10.1109/icse48619.2023.00175",
+      "open_access_pdf": "https://tcse.cn/~songjiansen20/assets/pdf/dqe_icse2023.pdf",
+      "abstract": "Database Management Systems (DBMSs) provide efficient data retrieval and manipulation for many applications through Structured Query Language (SQL). Incorrect implementations of DBMSs can result in logic bugs, which cause SELECT queries to fetch incorrect results, or UPDATE and DELETE queries to generate incorrect database states. Existing approaches mainly focus on detecting logic bugs in SELECT queries. However, logic bugs in UPDATE and DELETE queries have not been tackled. In this paper, we propose a novel and general approach, which we have termed Differential Query Execution (DQE), to detect logic bugs in SELECT, UPDATE and DELETE queries of DBMSs. The core idea of DQE is that different SQL queries with the same predicate usually access the same rows in a database. For example, a row updated by an UPDATE query with a predicate φ should also be fetched by a SELECT query with the same predicate φ, If not, a logic bug is revealed in the target DBMS. To evaluate the effectiveness and generality of DQE, we apply DQE on five production-level DBMSs, i.e., MySQL, MariaDB, TiDB, CockroachDB and SQLite. In total, we have detected 50 unique bugs in these DBMSs, 41 of which have been confirmed, and 11 have been fixed. We expect that the simplicity and generality of DQE can greatly improve the reliability of DBMSs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+              "source_type": "paper_citation_context",
+              "excerpt": "To answer RQ2, we perform a qualitative comparison with existing approaches (i.e., PQS [9], NoREC [10] and TLP [11]) that aim to detect logic bugs in DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d2b1c53eed22d33dbf11f3b4c13df64ddee508aa5a8440efeb79b87e6c73db1c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+              "source_type": "paper_citation_context",
+              "excerpt": "To demonstrate the sufficiency of our testing, we compare code coverage with existing works, i.e., PQS [9], NoREC [10] and TLP [11].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:212a57b2b4a2117264ec446935cba641755f3f8b90f4021048e35145fa687bc8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [10] rewrites a SELECT query as an equivalent one that the DBMS cannot optimize, and compares their results.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7120852e16021d2cfccda850fc7c4d02c1dd2cea2641de458261d3d44add2d31",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [11] leverages the ternary property of predicate evaluation, where the evaluation result is one of TRUE , FALSE and NULL , to partition a SELECT query into three partitioning queries, whose combined query results are equal to the original query’s query result.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:048f04ab030f77f536794d50648f2f81f1c8321171767f2a2d4b91e08afb6606",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [11] decomposes a SELECT query into three partitioning queries, and merges these partitioning queries’ results into a combined result, which is expected to be the same as the original query’s result.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:446f66f40e75f8f4088acd09b4a346f3df38177b64ea72e64e91d66dd18dbf58",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse48619.2023.00175",
+              "source_type": "paper_citation_context",
+              "excerpt": "Database and SQL query generation have been widely explored by existing works [21]–[26], [29], [50]–[57].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Query Plan Guidance (QPG), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:20bc7984451726d541244e0b82787f8947703dd2139c0d882095934a3d6ae14d",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse48619_2023_00175.json",
+              "source_type": "paper",
+              "excerpt": "Our database generation is mainly adopted from SQLancer [27].",
+              "excerpt_is_verbatim": true,
+              "note": "M7 in the paper's extracted text, B Database Generation, page 4.",
+              "content_sha256": "sha256:168e3c24a4f37499879c6e99fdc9d94d8cc9dd0ae9198b9fb9559949933a1824",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse48619_2023_00175.json",
+              "source_type": "paper",
+              "excerpt": "EVALUATION We implement DQE based on SQLancer [27], which is implemented in Java.",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, IV EVALUATION, page 7.",
+              "content_sha256": "sha256:6777e2c23fc4007222da0be91b5482cdb10198c767112e5c2085b44da23109ef",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse48619_2023_00175.json",
+              "source_type": "paper",
+              "excerpt": "This is reasonable, since DQE, PQS, NoREC and TLP are all built on SQLancer, which share the similar query generation.",
+              "excerpt_is_verbatim": true,
+              "note": "M16 in the paper's extracted text, D Other Experimental Statistics, page 9.",
+              "content_sha256": "sha256:1a7c955e0492e18f2e384a3e3bc1abd63a71aa001e6e588208dd6d843fdf4dad",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse48619_2023_00175.json",
+              "source_type": "paper",
+              "excerpt": ", PQS [9], NoREC [10] and TLP [11]) that aim to detect logic bugs in DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, C Comparing with Existing Approaches, page 8.",
+              "content_sha256": "sha256:0143efc39e429a12a72965b7a96f65dc04186c4f04eb35e7114d947bb63eb767",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse48619_2023_00175.json",
+              "source_type": "paper",
+              "excerpt": "Thus, they cannot detect 18 logic bugs related to this kind of errors TABLE VCOVERAGE INFORMATION Tool MySQL MariaDB PQS 19 NoREC-18 TLP 18 DQE 15 21 in SELECT queries.",
+              "excerpt_is_verbatim": true,
+              "note": "M10 in the paper's extracted text, C Comparing with Existing Approaches, page 8.",
+              "content_sha256": "sha256:dd6c1eeec8443bbde6a3fab79804184431774a2181a896cc361842795dd18ac5",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse48619_2023_00175.json",
+              "source_type": "paper",
+              "excerpt": ", PQS [9], NoREC [10] and TLP [11].",
+              "excerpt_is_verbatim": true,
+              "note": "M12 in the paper's extracted text, D Other Experimental Statistics, page 8.",
+              "content_sha256": "sha256:b774712fcdae08758be9f000c4a2985cc76072c7bbea1a57cf3a1318d4edf502",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/5724e138fc08329e0baa08d2090d9424df1cb945",
+        "source_type": "paper",
+        "content_sha256": "sha256:b8c6a23f859f6d3c30bc6f805ef491bf00cea143e7083f274d320c3901d5e0cb"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/JensonSung/dqetool",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "renamed_sqlancer_package",
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/JensonSung/dqetool",
+              "source_type": "github_repository",
+              "excerpt": "Replication package for \"Testing Database Systems via Differential Query Execution\", accepted at ICSE 2023\n\n# DQETool\n\nDQETool is the implementation of differential query execution in paper.\n\n# Getting Started\n\nRequirements:\n* Java 11 or above",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/JensonSung/dqetool/blob/main/src/dqetool/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "src/dqetool/Randomly.java is SQLancer's Randomly.java, with the package renamed to dqetool (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:c4aa77016b32eb0b4a0e071764bf41adfdc03763a40486ebcb64e9c0a82df6ab",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3597926.3598044",
+      "title": "Testing Graph Database Engines via Query Partitioning",
+      "authors": [
+        "Matteo Kamm",
+        "Manuel Rigger",
+        "Chengyu Zhang",
+        "Zhendong Su"
+      ],
+      "year": 2023,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3597926.3598044",
+      "arxiv_id": null,
+      "s2_paper_id": "a2a514ed839dafdd0fb76d6c2615f25f35bf8087",
+      "url": "https://doi.org/10.1145/3597926.3598044",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3597926.3598044",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598044",
+              "source_type": "paper_citation_context",
+              "excerpt": "The state-of-the-art test oracle for detecting logic bugs is Query Partitioning [31], which is based on the idea that from a given query, multiple so-called partitioning queries can be derived, each of which computes a part of the original query’s result.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0fbb7f182a9f9ecd457d984b1c44174b9d63bd6f33463f393752e5d00630dc96",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598044",
+              "source_type": "paper_citation_context",
+              "excerpt": "The key insight of this paper is that the high-level idea of Query Partitioning, and specifically TLP, is applicable and effective in finding logic bugs in GDBMSs and addresses the aforementioned challenges.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ba36f59fed1066a71c9a0703fffae2e5f2a71582950f81e982bf921507871eb8",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598044",
+              "source_type": "paper_citation_context",
+              "excerpt": "Overall, this paper makes the following contributions: • It demonstrates how the Query Partitioning test oracle [31], in particular, Ternary Logic Partitioning, can be applied on GDBMSs to find logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a39223501091c569e0730fe2dceeeeda6be26a6a0b09cf2c7d0dac4b731531dc",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598044",
+              "source_type": "paper_citation_context",
+              "excerpt": "The core of the approach is its test oracle, called Ternary Logic Partitioning (TLP), which was previously proposed to test RDBMSs. GDBMeter operates in three phases, as shown in Figure 2.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:fb9d3f6e565feea3d338143babe653cf8708feef9749765325e09c3fc56b6e36",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598044",
+              "source_type": "paper_citation_context",
+              "excerpt": "Different from previous work [30, 31], which defined logic bugs as bugs that cause an incorrect result to be computed, the Grand authors considered also unexpected errors as logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:aecfd820deefc4921f9f2d8d156db58f8a1f9257e6724ffe828eb85a488e5267",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598044",
+              "source_type": "paper_citation_context",
+              "excerpt": "Therefore, it is not yet clear whether TLP is still effective in testing GDBMS. Listing 1: An illustrative example of a logic bug found using Ternary Logic Partitioning in Neo4j.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5266375d7f29c60c2d3988490bf3c3a77c850b04a7624c468e17a76a11aed53f",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "manual_curation",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598044",
+              "source_type": "paper_citation_context",
+              "excerpt": "The key insight of this paper is that the high-level idea of Query Partitioning, and specifically TLP, is applicable and effective in finding logic bugs in GDBMSs and addresses the aforementioned challenges.",
+              "excerpt_is_verbatim": true,
+              "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+              "content_sha256": "sha256:ba36f59fed1066a71c9a0703fffae2e5f2a71582950f81e982bf921507871eb8",
+              "retrieved_at": "2026-09-06T07:08:48Z",
+              "first_seen": "2026-09-06T07:08:48Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ],
+          "classifier": {
+            "method": "manual_curation",
+            "classifier_version": "manual-curation-v1",
+            "policy_version": "impact-policy-v1",
+            "taxonomy_version": "taxonomy-v1",
+            "classified_at": "2026-09-06T07:08:48Z",
+            "model": null,
+            "rationale": "Reviewed by a maintainer against the citation contexts collected for this paper."
+          }
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "deterministic",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3597926.3598044",
+              "source_type": "paper_citation_context",
+              "excerpt": "The state-of-the-art test oracle for detecting logic bugs is Query Partitioning [31], which is based on the idea that from a given query, multiple so-called partitioning queries can be derived, each of which computes a part of the original query’s result.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper describing Ternary Logic Partitioning (TLP) as state of the art.",
+              "content_sha256": "sha256:0fbb7f182a9f9ecd457d984b1c44174b9d63bd6f33463f393752e5d00630dc96",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/a2a514ed839dafdd0fb76d6c2615f25f35bf8087",
+        "source_type": "paper",
+        "content_sha256": "sha256:2b4a590fd5cf6ce5ed65449561fb955a14c145ec6bcdb2b73fcdfa9bb28fba9f"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Graph Database Management Systems (GDBMSs) store data as graphs and allow the efficient querying of nodes and their relationships. Logic bugs are bugs that cause a GDBMS to return an incorrect result for a given query (e.g., by returning incorrect nodes or relationships). The impact of such bugs can be severe, as they often go unnoticed. The core insight of this paper is that Query Partitioning, a test oracle that has been proposed to test Relational Database Systems, is applicable to testing GDBMSs as well. The core idea of Query Partitioning is that, given a query, multiple queries are derived whose results can be combined to reconstruct the given query’s result. Any discrepancy in the result indicates a logic bug. We have implemented this approach as a practical tool named GDBMeter and evaluated GDBMeter on three popular GDBMSs and found a total of 40 unique, previously unknown bugs. We consider 14 of them to be logic bugs, the others being error or crash bugs. Overall, 27 of the bugs have been fixed, and 35 confirmed. We compared our approach to the state-of-the-art approach to testing GDBMS, which relies on differential testing; we found that it results in a high number of false alarms, while Query Partitioning reported actual logic bugs without any false alarms. Furthermore, despite the previous efforts in testing Neo4j and JanusGraph, we found 18 additional bugs. The developers appreciate our work and plan to integrate GDBMeter into their testing process. We expect that this simple, yet effective approach and the practical tool will be used to test other GDBMSs."
+    },
+    {
+      "id": "paper:doi:10.14778/3636218.3636236",
+      "title": "Testing Graph Database Systems via Graph-Aware Metamorphic Relations",
+      "authors": [
+        "Zeyang Zhuang",
+        "Penghui Li",
+        "Pingchuan Ma",
+        "Wei Meng",
+        "Shuai Wang"
+      ],
+      "year": 2023,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3636218.3636236",
+      "arxiv_id": null,
+      "s2_paper_id": "ac441cb1810dce8e41d9cd554366b72aa182bf41",
+      "url": "https://doi.org/10.14778/3636218.3636236",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3636218.3636236",
+              "source_type": "paper_citation_context",
+              "excerpt": "We next propose the pattern partitioning technique to generate compound MRs by extending the query partitioning technique in the literature [34, 42].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b8a814aa9761a1733fdf262e34425cc1c1a6a20d9205b4b0f07f60d1b520dc31",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3636218.3636236",
+              "source_type": "paper_citation_context",
+              "excerpt": "In relational database systems, SQLancer [18] leverages ternary logic partitioning (TLP) [42] and non-optimizing reference engine construction (NoREC) [41] to form MRs and detect logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f7eb0f588c85bf39a1fb4a384450cbcf4984fa76f8ef8aab9fab7e62e664160e",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3636218.3636236",
+              "source_type": "paper_citation_context",
+              "excerpt": "Similar to other software systems [41, 42], GDBs contain logic bugs, which could cause unexpected behaviors and lead to severe consequences.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b1dc2d1e0001c56d50932b4865ca6960dc6ffa50aa8049e34f685de972a34957",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3636218.3636236",
+              "source_type": "paper_citation_context",
+              "excerpt": "To date, MT methods have found hundreds of bugs in database systems, demonstrating its effectiveness [18, 41, 42].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:907cff07a020297296039f312a202cfc5cdc7b5fc57da9d9fc70757bfe41a72f",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3636218.3636236",
+              "source_type": "paper_citation_context",
+              "excerpt": "It ports the idea of ternary query partitioning [42] used in relational databases to GDBs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6fb8e518e34cbf399fa4b3cda6d0410248527ea70cd994050724091129f00524",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3636218.3636236",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, most prior database MT works focus on relational database systems [18, 41, 42].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4dff335cca15f04bc6c7e7b91904bfc5a8f9ae78da0710280646b8d04c9648a7",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "artifact_inspection",
+          "evidence": [
+            {
+              "source_url": "https://github.com/cuhk-seclab/Gamera",
+              "source_type": "github_repository",
+              "excerpt": "Gamera in total detected 39 bugs and 5 new bugs from FalkorDB (, which is related to RedisGraph). All the results are listed here: [results.md](results.md).\n\n\n## Publication\nYou can find more details in our VLDB 2024 paper:\n[Testing Graph Database Systems via Graph-Aware Metamorphic Relations](https://www.vldb.org/pvldb/vol17/p836-zhuang.pdf)\n```\n@article{zhuang2024gamera,\n  title   = {Testing Graph Database Systems via Graph-aware Metamorphic Relations},",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/cuhk-seclab/Gamera/blob/main/GremlinChecker/src/main/java/org/gdbtesting/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "GremlinChecker/src/main/java/org/gdbtesting/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.gdbtesting (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:d8417e8316280ef45d59ea8cb6289dfe9a4e5c1ade44a7f65f3571d3c5cce364",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ],
+          "classifier": {
+            "method": "artifact_inspection",
+            "classifier_version": "artifact-markers-v1",
+            "policy_version": "impact-policy-v1",
+            "taxonomy_version": "taxonomy-v2",
+            "classified_at": "2026-09-06T15:09:10Z",
+            "model": null,
+            "rationale": "Artifact carries SQLancer markers: renamed_sqlancer_package, sqlancer_source_content_match"
+          }
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "manual_curation",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3636218.3636236",
+              "source_type": "paper_citation_context",
+              "excerpt": "We next propose the pattern partitioning technique to generate compound MRs by extending the query partitioning technique in the literature [34, 42].",
+              "excerpt_is_verbatim": true,
+              "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+              "content_sha256": "sha256:b8a814aa9761a1733fdf262e34425cc1c1a6a20d9205b4b0f07f60d1b520dc31",
+              "retrieved_at": "2026-09-06T07:08:48Z",
+              "first_seen": "2026-09-06T07:08:48Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3636218.3636236",
+              "source_type": "paper_citation_context",
+              "excerpt": "It ports the idea of ternary query partitioning [42] used in relational databases to GDBs.",
+              "excerpt_is_verbatim": true,
+              "note": "Passage in the paper, as indexed by Semantic Scholar, supporting this relationship.",
+              "content_sha256": "sha256:6fb8e518e34cbf399fa4b3cda6d0410248527ea70cd994050724091129f00524",
+              "retrieved_at": "2026-09-06T07:08:48Z",
+              "first_seen": "2026-09-06T07:08:48Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ],
+          "classifier": {
+            "method": "manual_curation",
+            "classifier_version": "manual-curation-v1",
+            "policy_version": "impact-policy-v1",
+            "taxonomy_version": "taxonomy-v1",
+            "classified_at": "2026-09-06T07:08:48Z",
+            "model": null,
+            "rationale": "Reviewed by a maintainer against the citation contexts collected for this paper."
+          }
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/ac441cb1810dce8e41d9cd554366b72aa182bf41",
+        "source_type": "paper",
+        "content_sha256": "sha256:5458cb13a789610ea5c0efe404102466f730ede8c68e5bd9d4a5abc1c7d1f747"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/cuhk-seclab/Gamera",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-06T15:09:10Z",
+          "sqlancer_markers": [
+            "renamed_sqlancer_package",
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/cuhk-seclab/Gamera",
+              "source_type": "github_repository",
+              "excerpt": "Gamera in total detected 39 bugs and 5 new bugs from FalkorDB (, which is related to RedisGraph). All the results are listed here: [results.md](results.md).\n\n\n## Publication\nYou can find more details in our VLDB 2024 paper:\n[Testing Graph Database Systems via Graph-Aware Metamorphic Relations](https://www.vldb.org/pvldb/vol17/p836-zhuang.pdf)\n```\n@article{zhuang2024gamera,\n  title   = {Testing Graph Database Systems via Graph-aware Metamorphic Relations},",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/cuhk-seclab/Gamera/blob/main/GremlinChecker/src/main/java/org/gdbtesting/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "GremlinChecker/src/main/java/org/gdbtesting/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.gdbtesting (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:d8417e8316280ef45d59ea8cb6289dfe9a4e5c1ade44a7f65f3571d3c5cce364",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T15:09:10Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        }
+      ],
+      "abstract": "Graph database systems (GDBs) have supported many important real-world applications such as social networks, logistics, and path planning. Meanwhile, logic bugs are also prevalent in GDBs, leading to incorrect results and severe consequences. However, the logic bugs largely cannot be revealed by prior solutions which are unaware of the graph native structures of the graph data. In this paper, we propose Gamera (Graph-aware metamorphic relations), a novel metamorphic testing approach to uncover unknown logic bugs in GDBs. We design three classes of novel graph-aware Metamorphic Relations (MRs) based on the graph native structures. Gamera would generate a set of queries according to the graph-aware MRs to test diverse and complex GDB operations, and check whether the GDB query results conform to the chosen MRs.\n We thoroughly evaluated the effectiveness of Gamera on seven widely-used GDBs such as Neo4j and OrientDB. Gamera was highly effective in detecting logic bugs in GDBs. In total, it detected 39 logic bugs, of which 15 bugs have been confirmed, and three bugs have been fixed. Our experiments also demonstrated that Gamera significantly outperformed prior solutions including Grand, GD-smith and GDBMeter. Gamera has been well-recognized by GDB developers and we open-source our prototype implementation to contribute to the community."
+    },
+    {
+      "id": "paper:doi:10.1145/3622819",
+      "title": "Towards Better Semantics Exploration for Browser Fuzzing",
+      "authors": [
+        "Chijin Zhou",
+        "Quan Zhang",
+        "Lihua Guo",
+        "Mingzhe Wang",
+        "Yu Jiang",
+        "Qing Liao",
+        "Zhiyong Wu",
+        "Shanshan Li",
+        "Bin Gu"
+      ],
+      "year": 2023,
+      "venue": "Proc. ACM Program. Lang.",
+      "doi": "10.1145/3622819",
+      "arxiv_id": null,
+      "s2_paper_id": "c4a6a53c4c859e6d4c71ae122272a5e7bf91a163",
+      "url": "https://doi.org/10.1145/3622819",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3622819",
+      "abstract": "Web browsers exhibit rich semantics that enable a plethora of web-based functionalities. However, these intricate semantics present significant challenges for the implementation and testing of browsers. For example, fuzzing, a widely adopted testing technique, typically relies on handwritten context-free grammars (CFGs) for automatically generating inputs. However, these CFGs fall short in adequately modeling the complex semantics of browsers, resulting in generated inputs that cover only a portion of the semantics and are prone to semantic errors. In this paper, we present SaGe, an automated method that enhances browser fuzzing through the use of production-context sensitive grammars (PCSGs) incorporating semantic information. Our approach begins by extracting a rudimentary CFG from W3C standards and iteratively enhancing it to create a PCSG. The resulting PCSG enables our fuzzer to generate inputs that explore a broader range of browser semantics with a higher proportion of semantically-correct inputs. To evaluate the efficacy of SaGe, we conducted 24-hour fuzzing campaigns on mainstream browsers, including Chrome, Safari, and Firefox. Our approach demonstrated better performance compared to existing browser fuzzers, with a 6.03%-277.80% improvement in edge coverage, a 3.56%-161.71% boost in semantic correctness rate, twice the number of bugs discovered. Moreover, we identified 62 bugs across the three browsers, with 40 confirmed and 10 assigned CVEs.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3622819",
+              "source_type": "paper_citation_context",
+              "excerpt": "This monitor is built on top of the JavaScript interpreter without intrusively instrumenting the browsers, providing flexibility in detecting semantic errors.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:433182022293653e0a7c355c39be14647bc66d6bf1c8e2eeed9994188bbd3492",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3622819",
+              "source_type": "paper_citation_context",
+              "excerpt": "An HTML document consists of three parts: (1) an HTML part to define the initial Document Object Model (DOM) tree, (2) a CSS part to specify in which style the elements of DOM tree are rendered; and (3) a JavaScript part to programmatically manipulate objects in DOM tree or enable other functionalities.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c5f0004d5773cba03d579a80389df870fd508a497927ce645c25d0eaa60d90fd",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3622819",
+              "source_type": "paper_citation_context",
+              "excerpt": "…fuzzers, there are also many custom fuzzers, enhanced with domain knowledge, for fuzzing specific domain such as C/C++ compilers [Livinskii et al. 2020; Yang et al. 2011], databases [Rigger and Su 2020; Wang et al. 2021b; Zhong et al. 2020], and deep learning frameworks [Liu et al. 2023].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:67b0f270c2d51cbcc8b26c729331f5d54b9848f52f7b3a5d1ef3d098db03ba99",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3622819",
+              "source_type": "paper_citation_context",
+              "excerpt": "As shown in Listing 1, the minimized code snippet is quite simple: the HTML part creates an audio element, and the JavaScript part accesses this element, and calls the setSinkId member function of this element with a random String input.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d064aa2d94280a110783e2d4a330c7e6b74ede34f3bc5bbdd356ad3b02b2a725",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3622819",
+              "source_type": "paper_citation_context",
+              "excerpt": "Their generation strategies focus more on generating nested JavaScript code, such as code with complex variable lifetimes or multiple loops, in order to explore the interpretation/optimization logic of JavaScript engines.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:71f920fc162e41dd56b3ee383211c315df1fceac679e66eb55b98f2ca1bad885",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3622819",
+              "source_type": "paper_citation_context",
+              "excerpt": "However, they utilize the standards to perform differential testing in order to uncover standard conformance bugs, i.e., inconsistencies of the implementation of JavaScript engine and JavaScript standards.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:85915f7be7a192233d5dd17d39366261e8c4dca9f936594281cf01cb45d83a7b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/c4a6a53c4c859e6d4c71ae122272a5e7bf91a163",
+        "source_type": "paper",
+        "content_sha256": "sha256:a41c9ae60f0cb2408f35cb84c154752d599e586b821cd4ca2f86da7ed41d12c1"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/issrew55968.2022.00056",
+      "title": "A Disjoint-Partitioning Approach to Enhancing Metamorphic Testing of DBMS",
+      "authors": [
+        "M. Tang",
+        "T. Tse",
+        "Z. Zhou"
+      ],
+      "year": 2022,
+      "venue": "2022 IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW)",
+      "doi": "10.1109/issrew55968.2022.00056",
+      "arxiv_id": null,
+      "s2_paper_id": "de7f728893406aa870766bad7b626f6eb774abcc",
+      "url": "https://doi.org/10.1109/issrew55968.2022.00056",
+      "open_access_pdf": null,
+      "abstract": "Owing to big data, DBMS testing faces the oracle problem, that is, it is difficult to verify execution results against expected outcomes. Rigger and Su applied metamorphic testing to alleviate the challenge. We propose a disjoint-partitioning approach to extend their work. We have conducted an empirical case study on OceanBase, the DBMS associated with the world's fastest online transaction processing system. Even though Ocean- Base has been extensively tested and widely used in the industry, we have unveiled various hidden failures and crashes.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/issrew55968.2022.00056",
+              "source_type": "paper_citation_context",
+              "excerpt": "They proposed a general concept called query partitioning (QP) [5] for revealing DBMS failures.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:dfe2c21b83160af97b541881e41a3c565923ac7f1223526a4192ae5a0c27c1c8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/issrew55968.2022.00056",
+              "source_type": "paper_citation_context",
+              "excerpt": "Rigger and Su further proposed ternary logic partitioning (TLP) [5] as a specific case of QP.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5d576cf61b29489b84e49a4d78d8c93151b57759a34db426b48f25a1dc25a86c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/issrew55968.2022.00056",
+              "source_type": "paper_citation_context",
+              "excerpt": "In 2020, Rigger and Su applied MT to alleviate the oracle problem in DBMS testing [4][5][6].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b17b0686fbd21410058a2977389fc8b06ec511dfd2fd2a193d4ef30d8ba39448",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/issrew55968.2022.00056",
+              "source_type": "paper_citation_context",
+              "excerpt": "In 2020, Rigger and Su [4][5][6] applied MT to tackle the issue in DBMS testing.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:853bda09b5ec6569710be2ad4bca64e10ef1891a1a802f1e92b214d4c2ecb4df",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_issrew55968_2022_00056.json",
+              "source_type": "paper",
+              "excerpt": "They proposed the query partitioning (QP) and ternary logic partitioning (TLP) techniques.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, I INTRODUCTION, page 1.",
+              "content_sha256": "sha256:44c1bdc8d94d0808e21bd860bfca7f1ca306076f4a551753e796b605538c0832",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_issrew55968_2022_00056.json",
+              "source_type": "paper",
+              "excerpt": "We find a gap between QP and TLP, and introduce the concept of disjoint partitioning to address the issue.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, I INTRODUCTION, page 1.",
+              "content_sha256": "sha256:b6b08e61d51be5487c204cd80546e22c8e50ed980ae7a0658b2050e2772907a4",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/de7f728893406aa870766bad7b626f6eb774abcc",
+        "source_type": "paper",
+        "content_sha256": "sha256:d92519e8fd04af5296f5aba55336b162028829f733229c06d29af67666d7bca4"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3510457.3513034",
+      "title": "An Empirical Study on Quality Issues of eBay's Big Data SQL Analytics Platform",
+      "authors": [
+        "Feng Zhu",
+        "Lijie Xu",
+        "Gang Ma",
+        "Shuping Ji",
+        "Jie Wang",
+        "Gang Wang",
+        "Hongyi Zhang",
+        "K. Wan",
+        "Mingming Wang",
+        "Xingchao Zhang",
+        "Yuming Wang",
+        "Jingping Li"
+      ],
+      "year": 2022,
+      "venue": "2022 IEEE/ACM 44th International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP)",
+      "doi": "10.1145/3510457.3513034",
+      "arxiv_id": null,
+      "s2_paper_id": "ba2770c990b740b6dfa4a122ac4ad719868796dd",
+      "url": "https://doi.org/10.1145/3510457.3513034",
+      "open_access_pdf": null,
+      "abstract": "Big data SQL analytics platform has evolved as the key infrastructure for business data analysis. Compared with traditional costly commercial RDBMS, scalable solutions with open-source projects, such as SQL-on-Hadoop, are more popular and attractive to enter-prises. In eBay, we build Carmel, a company-wide interactive SQL analytics platform based on Apache Spark. Carmel has been serving thousands of customers from hundreds of teams globally for more than 3 years. Meanwhile, despite the popularity of open-source based big data SQL analytics platforms, few empirical studies on service quality issues (e.g., job failure) were carried out for them. However, a deep understanding of service quality issues and taking right mitigation are significant to the ease of manual maintenance efforts. To fill this gap, we conduct a comprehensive empirical study on 1,884 real-word service quality issues from Carmel. We summa-rize the common symptoms and identify the root causes with typical cases. Stakeholders including system developers, researchers, and platform maintainers can benefit from our findings and implications. Furthermore, we also present lessons learned from critical cases in our daily practice, as well as insights to motivate automatic tool support and future research directions.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3510457.3513034",
+              "source_type": "paper_citation_context",
+              "excerpt": "After investigation, we found the root cause is Spark’s endless retry logic 13 : \" we should ideally differentiate these task statuses so that they don’t count towards the failure limit \".",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b1da2158664345e9896fdb607a6f459c9dfeea15a45d89eb2289bcdc43e366bb",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3510457.3513034",
+              "source_type": "paper_citation_context",
+              "excerpt": "To detect DBMS bugs, Rigger and Su devised a series of novel approaches, including PQS [15], NoRec [13] and TLP [14].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c468b5309a643220c63b600bd6d4a61d6b1788a9c88cc7baa3af6083ca24ca5e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/ba2770c990b740b6dfa4a122ac4ad719868796dd",
+        "source_type": "paper",
+        "content_sha256": "sha256:73971144cb8a335b7d899791410137cedd7d466a59f21cad7b7cca42b206b1e5"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3510003.3510093",
+      "title": "Automatic Detection of Performance Bugs in Database Systems using Equivalent Queries",
+      "authors": [
+        "Xinyu Liu",
+        "Qi Zhou",
+        "Joy Arulraj",
+        "A. Orso"
+      ],
+      "year": 2022,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1145/3510003.3510093",
+      "arxiv_id": null,
+      "s2_paper_id": "1f74989f2884e71ff09ca23f0c10e753bab9fcc9",
+      "url": "https://doi.org/10.1145/3510003.3510093",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "norec",
+        "pqs",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3510003.3510093",
+              "source_type": "paper_citation_context",
+              "excerpt": "For instance, they leverage tools such as S QLSMITH [4] and SQLancer [37–39] to discover crash-inducing or logic bugs in DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6478b7fd717627382bde2de9a55279b66a35fed969a68656b1a5a0accaad79d9",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+              "source_type": "paper_citation_context",
+              "excerpt": "To this end, we seek to compare against the TLP technique in SQLancer, which is the closest related effort [40–42].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c01f213da19839991dc8884a3db9e7be122d1ec06ef525eec52342f510635bea",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [40–42].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:cd16c7cf62398535ac3f042a087e71eb31b2006c918e84538b1db9547800e8de",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+              "source_type": "paper_citation_context",
+              "excerpt": "In particular, it generates equivalent queries based on Ternary Logic Partitioning (TLP) [41].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d21dc8382b570a2a77303ab6f0fd5ad679d4709f833f15b2410c016529df09b6",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+              "source_type": "paper_citation_context",
+              "excerpt": "Query equivalence is a well-studied topic and is used in many applications: (1) testing correctness of DBMS and SQL queries [41, 43], (2) educating developers [30], and (3) automatically grading student assignments [19].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8e07e727122ded2f23a177e165c4f453e09d5a9e6a1a14c0fcd4f51d0d150f74",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/74820ccf6c039c7745f8e1c30857dee61b243464",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, they leverage tools such as S QLSMITH [4] and SQLancer [40–42], which effectively discover crash or logic bugs in DBMS, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:180daca16b304fff627a8dbf1323e03609e77b55f22c7e143995a4197d379bc5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3510003_3510093.json",
+              "source_type": "paper",
+              "excerpt": "We also compared AMOEBA against two other sources of equivalent queries that could be used for detecting performance bugs: a manually-written test suite in a widely-used query optimization framework, and the Ternary Logic Partitioning (TLP) approach [ 38].",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, page 2.",
+              "content_sha256": "sha256:ee6ab2d3e15e563c9b3caf75a23da7c00e1fa07318dfd180dca3e035b42f01a6",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3510003_3510093.json",
+              "source_type": "paper",
+              "excerpt": "To answer Q4, we compare AMOEBA to three baseline based on two of these existing approaches: Calcite [19, 48] and TLP [38].",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, page 9.",
+              "content_sha256": "sha256:e367fc4ba25aa6cd372b3fbe0b64b51904c4f147992793b3c2164a5bff159465",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3510003_3510093.json",
+              "source_type": "paper",
+              "excerpt": "Our first baseline consists of 2000 pairs of equivlent queries generated using TLP.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, page 9.",
+              "content_sha256": "sha256:08078a51d95197a23c3d5a83d39902908d7480f8882e24464530bdbb36b63793",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3510003_3510093.json",
+              "source_type": "paper",
+              "excerpt": "For both DBMSs considered, AMOEBA discovered considerably more PPBs than the baselines based on Calcite and TLP.",
+              "excerpt_is_verbatim": true,
+              "note": "M13 in the paper's extracted text, page 10.",
+              "content_sha256": "sha256:78c199a1c4daad3bf36a219ebd93f61c0427cdd898136a46b459b7a43cb9493f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3510003_3510093.json",
+              "source_type": "paper",
+              "excerpt": "TLP is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [ 38].",
+              "excerpt_is_verbatim": true,
+              "note": "M14 in the paper's extracted text, page 11.",
+              "content_sha256": "sha256:e54b851c609c7fbb41676c58e53098c22bd8cbd322a214c8cb724b568ac64763",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/1f74989f2884e71ff09ca23f0c10e753bab9fcc9",
+        "source_type": "paper",
+        "content_sha256": "sha256:7c31025de45317c7ca408853acbbf1ae6fb851addff58bfdd77a3ee657ee30c3"
+      },
+      "also_indexed_as": [
+        "paper:s2:74820ccf6c039c7745f8e1c30857dee61b243464"
+      ]
+    },
+    {
+      "id": "paper:doi:10.22399/ijcesen.5462",
+      "title": "Benchmarking Autonomy: A Taxonomy of Human-in-the-Loop Checkpoints for AI-Generated Transformation Code",
+      "authors": [
+        "Jagan Ankam"
+      ],
+      "year": 2022,
+      "venue": "International Journal of Computational and Experimental Science and Engineering",
+      "doi": "10.22399/ijcesen.5462",
+      "arxiv_id": null,
+      "s2_paper_id": "7cba14d41c15d68f708b0c94c47772053b982465",
+      "url": "https://doi.org/10.22399/ijcesen.5462",
+      "open_access_pdf": "https://ijcesen.com/index.php/ijcesen/article/download/5462/2143",
+      "abstract": "Generative code systems can produce executable SQL, Spark, and dbt transformations with substantially less manual effort, but syntactic validity does not establish semantic correctness. A transformation may compile, complete successfully, and preserve an expected schema while silently changing the meaning of downstream data. This article presents a controlled pre-production benchmark and a taxonomy of human-in-the-loop checkpoints for governing that risk. The benchmark contains 12 realistic transformation tasks spanning relational SQL, distributed Spark processing, and dbt-style analytical modeling. Nine candidates contain seeded defects and three are correct controls. Three automated checkpoint families are evaluated independently and in combination: structural schema checks, declarative data-quality rules, and differential comparison against a trusted baseline. Schema checks and data-quality rules each detect one of nine defects (11.1%), and their detections overlap. Trusted-baseline comparison detects seven defects (77.8%). The union of all automated checkpoints detects eight defects (88.9%), leaving one low-magnitude semantic error that preserves schema, distributions, row counts, and tolerance-bounded aggregate values. The residual defect is identified only when the transformation logic and business intent are reviewed together. The findings support three conclusions. First, validation depth matters more than check quantity: semantic oracles materially outperform purely structural controls. Second, checkpoint portfolios exhibit diminishing returns when multiple rules target the same visible symptom. Third, human review should not be treated as an undifferentiated manual gate; it should be targeted toward transformations with monetary, temporal, incremental, rare-category, or tolerance-sensitive semantics. The article contributes a five-level checkpoint taxonomy, a reproducible defect taxonomy, task-level detection evidence, and a risk-based escalation policy suitable for pre-production DataOps and analytics engineering workflows. The analysis is artifact-centric and does not rank named code-generation models because generation logs and repeated model samples were not part of the benchmark. Because the benchmark is intentionally small, the numerical estimates are reported with uncertainty and are not generalized as population-level industry rates.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.22399/ijcesen.5462",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC compares an optimizable query with an equivalent form intended to suppress optimization [18].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c26c2d60abea5b5bc23802d5d93c46f8c4ca502819c6ac0714686850ee175cb4",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.22399/ijcesen.5462",
+              "source_type": "paper_citation_context",
+              "excerpt": "Database testing research develops oracles for defects in database engines rather than defects in application-level transformation intent [17]–[21].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9d18fe5a2376b071a277edf510937be29c7986d464bcb015bc9eee1d712fbb00",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.22399/ijcesen.5462",
+              "source_type": "paper_citation_context",
+              "excerpt": "Pivoted Query Synthesis constructs queries expected to retrieve a selected row [17].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:cb34bb6c3d0b52d1a2e078a0f988473ff0480ddd50b23b943e7a7b5ad1fe575b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.22399/ijcesen.5462",
+              "source_type": "paper_citation_context",
+              "excerpt": "Ternary Logic Partitioning checks relations among partitioned query results [19].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:7772df5368fe7f27c1b98a7dced1db01a78754585aa5451aac5295168cf9264b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.22399/ijcesen.5462",
+              "source_type": "paper_citation_context",
+              "excerpt": "[17], [18], [19] Equivalent or partitioned query results",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6fe5689b66e3ddee8ccdce94e432f30d6ebab053a7b887403f5b3c1bfb369f1d",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/7cba14d41c15d68f708b0c94c47772053b982465",
+        "source_type": "paper",
+        "content_sha256": "sha256:e43009df7eb33f8e93a5a9222f75b0f04d7d69e9e722804b3665fa51c83dae05"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3531348.3532176",
+      "title": "DeepBench: Benchmarking JSON Document Stores",
+      "authors": [
+        "S. Belloni",
+        "D. Ritter",
+        "Marco Schröder",
+        "Nils Rörup"
+      ],
+      "year": 2022,
+      "venue": "DBTest@SIGMOD",
+      "doi": "10.1145/3531348.3532176",
+      "arxiv_id": null,
+      "s2_paper_id": "a48f9e74f5f0f60e033c8c290156243c16ece5c0",
+      "url": "https://doi.org/10.1145/3531348.3532176",
+      "open_access_pdf": null,
+      "abstract": "The growing popularity of JSON as exchange and storage format in business and analytical applications led to its rapid dissemination, thus making a timely storage and processing of JSON documents crucial for organizations. Consequently, specialized JSON document stores are ubiquitously used for diverse domain-specific workloads, while a JSON-specific benchmark is missing. In this work, we specify DeepBench, an extensible, scalable benchmark that addresses nested JSON data, as well as queries over JSON documents. DeepBench features configurable domain-independent (e. g., varying document sizes, concurrent users) and JSON-specific scale levels (e. g., object, array nesting). The evaluation of well-known document stores with a prototypical DeepBench implementation shows its versatility and gives new insights into potential weaknesses that were not found by existing, non-JSON benchmarks",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/a48f9e74f5f0f60e033c8c290156243c16ece5c0",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/a48f9e74f5f0f60e033c8c290156243c16ece5c0",
+        "source_type": "paper",
+        "content_sha256": "sha256:fb5fad4c86d3cf54e7271bdb281d6c7f0f2ee4d7f705e86c3b4ff5056ffbdbcd"
+      }
+    },
+    {
+      "id": "paper:s2:84c637ad3297e0d4e45a4f5245d0472a82a59268",
+      "title": "Demystifying and Checking Silent Semantic Violations in Large Distributed Systems",
+      "authors": [
+        "Chang Lou",
+        "Yuzhuo Jing",
+        "Peng Huang"
+      ],
+      "year": 2022,
+      "venue": "USENIX Symposium on Operating Systems Design and Implementation",
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "84c637ad3297e0d4e45a4f5245d0472a82a59268",
+      "url": "https://www.semanticscholar.org/paper/84c637ad3297e0d4e45a4f5245d0472a82a59268",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/84c637ad3297e0d4e45a4f5245d0472a82a59268",
+              "source_type": "paper_citation_context",
+              "excerpt": "Several solutions are proposed to detect semantic bugs in file systems and DBMS, including cross-checking multiple file system implementations [50], fuzzing [39], and testing using pivoted query [54].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:169953b50a74da6832ed99491335135dfaa8d0099dda2bb610655b583d543dee",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/84c637ad3297e0d4e45a4f5245d0472a82a59268",
+        "source_type": "paper",
+        "content_sha256": "sha256:4064581c03375988f121a9a42481db0d6eda299a6d47851e21580b65be99d606"
+      }
+    },
+    {
+      "id": "paper:s2:77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+      "title": "Detecting Logical Bugs of DBMS with Coverage-based Guidance",
+      "authors": [
+        "Yu Liang",
+        "Song Liu",
+        "Hong Hu"
+      ],
+      "year": 2022,
+      "venue": "USENIX Security Symposium",
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+      "url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, for a given query, oracle NoREC shifts all conditions from WHERE clauses to SELECT expressions, which effectively disables most optimizations applied to the original query [43].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:67fdaef0b2843790e3dd927a7e477b984416b3e0306e53126057f9200992393e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+              "source_type": "paper_citation_context",
+              "excerpt": "Currently, SQLRight supports four oracles, including NoREC and TLP ported from SQLancer [43, 44], and Index and Rowid we propose in this paper.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:655e2311fc7fc954733d821c7a077e31232bc249185b1992eef3deb0fa97c084",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+              "source_type": "paper_citation_context",
+              "excerpt": "We implemented four oracles, including two proposed in previous works [43, 44] and two proposed in this paper.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:64f9d540dbc0b18ee012dea94d9dbe83b4265f76d7aa978df7c34ae230d3b0a8",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+              "source_type": "paper_citation_context",
+              "excerpt": "It combines the results from three subqueries and checks the equivalence with the original one [44].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b54f9015aa94ec1803dc381b47e4540a0d85bf7c10a1633d53ce038a5b7cee8c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "Our implementation is based on Squirrel [71] and SQLancer [30].",
+              "excerpt_is_verbatim": true,
+              "note": "M28 in the paper's extracted text, 3.1.4 Non-determinism Mitigation, page 8.",
+              "content_sha256": "sha256:09ae5f46d1712aa73a37d8b76832a436f63f06db02b1326275d6338cb27f251f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "Currently, SQLRight supports four oracles, including NoREC and TLPported from SQLancer [43, 44], and Index and Rowid we propose in this paper.",
+              "excerpt_is_verbatim": true,
+              "note": "M29 in the paper's extracted text, 3.1.4 Non-determinism Mitigation, page 8.",
+              "content_sha256": "sha256:9392b58302e6d9b2049b73bb9976e87410001ed45433fd5c78c3c3d09b286bae",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "To support the NoREC oracle, we fix SELECT_statement ,FROM and WHERE of the statement and leave others for the mutation engine to change.",
+              "excerpt_is_verbatim": true,
+              "note": "M18 in the paper's extracted text, 3.1.1 Cooperative Mutation, page 6.",
+              "content_sha256": "sha256:00a1be13475d298a051e873579f45172fbd7bfeeca165ac70e85e33a779470fb",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "As a result, the crucial components for oracle NoREC are preserved.",
+              "excerpt_is_verbatim": true,
+              "note": "M20 in the paper's extracted text, 3.1.1 Cooperative Mutation, page 6.",
+              "content_sha256": "sha256:103aa0a075d4aeb29e279cedb9027be440ca722cfd2df644fcffe0f388e5b359",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "Currently, SQLRight supports four oracles, including NoREC and TLPported from SQLancer [43, 44], and Index and Rowid we propose in this paper.",
+              "excerpt_is_verbatim": true,
+              "note": "M29 in the paper's extracted text, 3.1.4 Non-determinism Mitigation, page 8.",
+              "content_sha256": "sha256:9392b58302e6d9b2049b73bb9976e87410001ed45433fd5c78c3c3d09b286bae",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "We also compare our system with the stateof-the-art tools, including SQLancer (using oracles to detect logical bugs) and Squirrel (using code coverage to detect crashes and assertion failures).",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, 1 Introduction, page 3.",
+              "content_sha256": "sha256:31181d50de667926b66474202f7208135acc75c661c813b132c4c92a71b3ea95",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "After testing for 72 hours, SQLRight reports 12 unique logical bugs, Squirrel detects one bug, and SQLancer does not find any bug.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 1 Introduction, page 3.",
+              "content_sha256": "sha256:e9fa1cdc16ef8424c4f1bf9037d03e89e4784af9dd39469b46fab18872bef17e",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "Therefore, we use NoREC and TLPto compare SQLRight ,SQLancer, and Squirrel +oracle.",
+              "excerpt_is_verbatim": true,
+              "note": "M44 in the paper's extracted text, 5.1 DBMS Logical Bugs, page 10.",
+              "content_sha256": "sha256:4396d204115e6e2e054acc728688a0f502d20689653b41ba3bacad89c8706efe",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "Squirrel +oracle merely found 1 bug, which is from SQLite using NoREC; SQLancer did not find any logical bug.",
+              "excerpt_is_verbatim": true,
+              "note": "M48 in the paper's extracted text, 5.1 DBMS Logical Bugs, page 10.",
+              "content_sha256": "sha256:6943504215e36eba468af032080c7257178fbc52ebd12547c2bf3900076f5d96",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "Overall, SQLRight can find more logical bugs than SQLancer andSquirrel +oracle.",
+              "excerpt_is_verbatim": true,
+              "note": "M58 in the paper's extracted text, 5.1 DBMS Logical Bugs, page 11.",
+              "content_sha256": "sha256:1cb67c25b29ce1ed4979dbc33536073dd827a325472eb87ee14dff3c27aa2ab7",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "We also compare our system with the stateof-the-art tools, including SQLancer (using oracles to detect logical bugs) and Squirrel (using code coverage to detect crashes and assertion failures).",
+              "excerpt_is_verbatim": true,
+              "note": "M8 in the paper's extracted text, 1 Introduction, page 3.",
+              "content_sha256": "sha256:31181d50de667926b66474202f7208135acc75c661c813b132c4c92a71b3ea95",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json",
+              "source_type": "paper",
+              "excerpt": "For Q2, we compare SQLRight with SQLancer and Squirrel, the state-of-the-art bug-finding tools for DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "M33 in the paper's extracted text, 3.1.4 Non-determinism Mitigation, page 9.",
+              "content_sha256": "sha256:1460c526dbec33491efd83fb48859275c6ab36aada091463c28c37cc87d28953",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+        "source_type": "paper",
+        "content_sha256": "sha256:dad54155e5f87c0252712f5921f7681ee3f60f1b6123be9ca985aff1e482245a"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/PSU-Security-Universe/sqlright-artifact",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/PSU-Security-Universe/sqlright-artifact",
+              "source_type": "github_repository",
+              "excerpt": "Artifact Evaluation code for USENIX 2022 paper: Detecting Logical Bugs of DBMS with Coverage-based Guidance\n# sqlright-artifact: The code, analysis scripts and results for USENIX 2022 Artifact Evaluation\n\n<a href=\"Paper/paper_no_names.pdf\"><img src=\"Paper/paper_no_names.png\" align=\"right\" width=\"250\"></a>",
+              "note": "Repository names this paper.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/PSU-Security-Universe/sqlright-artifact/blob/main/MySQL/docker/sqlancer/sqlancer/src/sqlancer/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "MySQL/docker/sqlancer/sqlancer/src/sqlancer/Randomly.java is SQLancer's Randomly.java (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:cb0a0e87e414094ea53bc63a6ad1aefbb64e7919425edd6a94de532b9cd1daa6",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3551349.3556924",
+      "title": "Differentially Testing Database Transactions for Fun and Profit",
+      "authors": [
+        "Ziyu Cui",
+        "Wensheng Dou",
+        "Qianwang Dai",
+        "Jiansen Song",
+        "Wei Wang",
+        "Jun Wei",
+        "Dan Ye"
+      ],
+      "year": 2022,
+      "venue": "International Conference on Automated Software Engineering",
+      "doi": "10.1145/3551349.3556924",
+      "arxiv_id": null,
+      "s2_paper_id": "b3135c616d6267fe1d2b4ca9d27c8792c0acf3e5",
+      "url": "https://doi.org/10.1145/3551349.3556924",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3551349.3556924",
+      "abstract": "Database Management Systems (DBMSs) utilize transactions to ensure the consistency and integrity of data. Incorrect transaction implementations in DBMSs can lead to severe consequences, e.g., incorrect database states and query results. Therefore, it is critical to ensure the reliability of transaction implementations. In this paper, we propose DT2, an approach for automatically testing transaction implementations in DBMSs. We first randomly generate a database and a group of concurrent transactions operating the database, which can support complex features in DBMSs, e.g., various database schemas and cross-table queries. We then leverage differential testing to compare transaction execution results on multiple DBMSs to find discrepancies. The non-determinism of concurrent transactions can affect the effectiveness of our method. Therefore, we propose a transaction test protocol to ensure the deterministic execution of concurrent transactions. We evaluate DT2 on three widely-used MySQL-compatible DBMSs: MySQL, MariaDB and TiDB. In total, we have detected 10 unique transaction bugs and 88 transaction-related compatibility issues from the observed discrepancies. Our empirical study on these compatibility issues shows that DBMSs suffer from various transaction-related compatibility issues, although they claim that they are compatible. These compatibility issues can also lead to serious consequences, e.g., inconsistent database states among DBMSs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3551349.3556924",
+              "source_type": "paper_citation_context",
+              "excerpt": "Some approaches, e.g., PQS [60], NoREC [58], TLP [59] and RAGS [61], can detect DBMS bugs that involve these complex features in single SELECT statements.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:88bb603275d0634092ed6bcbf145a1e63ec2480df253e27dfb5ca964858ec2bd",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3551349.3556924",
+              "source_type": "paper_citation_context",
+              "excerpt": ", PQS [60], NoREC [58], TLP [59] and RAGS [61], can detect DBMS bugs that involve these complex features in single SELECT statements.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0c7843ca84640141b3e6cfe7645b66e9f553ee02183ddecf6a9bf3110fb111ed",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3551349.3556924",
+              "source_type": "paper_citation_context",
+              "excerpt": "More recently, Rigger et al. [58–60] has proposed a series of works to find logical bugs by generating single SQL queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:65a3aa2129ebcfc48b12e884b5a1bb3a7dc47174b4ea59ef7d942c2a6dcc2059",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3551349_3556924.json",
+              "source_type": "paper",
+              "excerpt": "We generate them mainly based on SQLancer [ 16], and slightly revise the approach for our target.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, 4.1 Transaction Test Case Generation, page 5.",
+              "content_sha256": "sha256:9f3a27683b05d7d904873fc95d1187772c59f14eff9d7f517509b0f05fdaf906",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/b3135c616d6267fe1d2b4ca9d27c8792c0acf3e5",
+        "source_type": "paper",
+        "content_sha256": "sha256:60646087eedb0abc7cf41ac749731b11eac15572c09a5f621984eca3c33368b8"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/criszy/DT2",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "renamed_sqlancer_package",
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/criszy/DT2",
+              "source_type": "github_repository",
+              "excerpt": "# DT2",
+              "note": "Repository is named after DT2, the tool this paper says it built, and is about database testing.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/criszy/DT2/blob/main/src/DT2/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "src/DT2/Randomly.java is SQLancer's Randomly.java, with the package renamed to DT2 (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:8d9b80c0a6b314d469a165162169da03d400d3d493bc550bcdf4c5fec2a8c289",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3533767.3534409",
+      "title": "Finding bugs in Gremlin-based graph database systems via Randomized differential testing",
+      "authors": [
+        "Yingying Zheng",
+        "Wensheng Dou",
+        "Yicheng Wang",
+        "Zheng Qin",
+        "Leile Tang",
+        "Yu Gao",
+        "Dong Wang",
+        "Wei Wang",
+        "Jun Wei"
+      ],
+      "year": 2022,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3533767.3534409",
+      "arxiv_id": null,
+      "s2_paper_id": "e67a2817089312746d69b38ce9abfdc4b1bc69c3",
+      "url": "https://doi.org/10.1145/3533767.3534409",
+      "open_access_pdf": null,
+      "abstract": "Graph database systems (GDBs) allow efficiently storing and retrieving graph data, and have become the critical component in many applications, e.g., knowledge graphs, social networks, and fraud detection. It is important to ensure that GDBs operate correctly. Logic bugs can occur and make GDBs return an incorrect result for a given query. These bugs are critical and can easily go unnoticed by developers when the graph and queries become complicated. Despite the importance of GDBs, logic bugs in GDBs have received less attention than those in relational database systems. In this paper, we present Grand, an approach for automatically finding logic bugs in GDBs that adopt Gremlin as their query language. The core idea of Grand is to construct semantically equivalent databases for multiple GDBs, and then compare the results of a Gremlin query on these databases. If the return results of a query on multiple GDBs are different, the likely cause is a logic bug in these GDBs. To effectively test GDBs, we propose a model-based query generation approach to generate valid Gremlin queries that can potentially return non-empty results, and a data mapping approach to unify the format of query results for different GDBs. We evaluate Grand on six widely-used GDBs, e.g., Neo4j and HugeGraph. In total, we have found 21 previously-unknown logic bugs in these GDBs. Among them, developers have confirmed 18 bugs, and fixed 7 bugs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3533767.3534409",
+              "source_type": "paper_citation_context",
+              "excerpt": "Rigger et al. provide two metamorphic testing approaches [48, 49], namely Ternary Logic Partitioning (TLP) and Non-optimizing Reference Engine Construction (NoREC), to test DBMS. TLP [49] partitions a query into three sub-queries, and detects bugs by comparing the combination of results of three sub-queries with the result of the original query.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:00bd63a0b29d2c7ba420fe57f2ba3c89f3f7b9ddd9366f2126a3938a3028b7fd",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3533767.3534409",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [49] partitions a query into three sub-queries, and detects bugs by comparing the combination of results of three sub-queries with the result of the original query.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4016cc5e4a459900239c6bf44a5e9a544190d5e16575883b4719f247a8c461b9",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3533767.3534409",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [48] compares the execution results of a given optimized query with its non-optimized version, to detect optimization bugs in DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e9cdaead7cc86f0e6769425dbd318af6d886ea87092f9f9e0597eb8de1e5e83e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3533767.3534409",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer offers Pivoted Query Synthesis (PQS) [50] approach to find logic bugs by randomly selecting a pivot row as oracle and generating random queries containing the selected row to test DBMS. ADUSA [39] translates SQL query to Alloy specification, generates Alloy instance satisfying query conditions in Alloy specification, and further obtains the expected result from Alloy instance.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d383191c27994bd04fdd06c13c5e0e54b30327405195b33aac3e5bb08cffc89b",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3533767.3534409",
+              "source_type": "paper_citation_context",
+              "excerpt": "RAGS uses differential testing for detecting bugs in relational database systems, while SQLancer offers three oracles, i.e., Pivoted Query Synthesis (PQS) [50], Ternary Logic Partitioning (TLP) [49], and Non-Optimizing Reference Engine Construction (NoREC) [48] to find logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d58d8b9054bcc5c08ebe4473e03f529776641a94c661bfba90e38e311b4c371c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3533767.3534409",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer offers Pivoted Query Synthesis (PQS) [50] approach to find logic bugs by randomly selecting a pivot row as oracle and generating random queries containing the selected row to test DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:902aa39f6e3fd72918bc2a2351a052b0fc596c853c2bc81e3974d3e1e7af3afc",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3533767_3534409.json",
+              "source_type": "paper",
+              "excerpt": "Artifact https://github.com/choeoe/Grand carries: renamed_sqlancer_package, sqlancer_source_content_match",
+              "excerpt_is_verbatim": true,
+              "note": "ARTIFACT in the paper's extracted text, artifact inspection.",
+              "content_sha256": "sha256:c93f6c555b7da370f1735975e9182b9c74b1c1370f43279b7ccd51f8c650619c",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/e67a2817089312746d69b38ce9abfdc4b1bc69c3",
+        "source_type": "paper",
+        "content_sha256": "sha256:5f9714167d6ca9e33d22bb3efaf8fa339a12d769bf09b241a6154b36f0c2e8cd"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/choeoe/Grand",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "renamed_sqlancer_package",
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/choeoe/Grand",
+              "source_type": "github_repository",
+              "excerpt": "# Grand",
+              "note": "Repository is named after Grand, the tool this paper says it built, and is about database testing.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/choeoe/Grand/blob/main/src/main/java/org/gdbtesting/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+              "excerpt_is_verbatim": true,
+              "note": "src/main/java/org/gdbtesting/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.gdbtesting (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:79cab1e9e63a88331ad883d12218247ae29913a268fb5ffc6cba313b8660e826",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1109/dsc55868.2022.00057",
+      "title": "Fuzzing DBMS via NNLM",
+      "authors": [
+        "Yabin Li",
+        "Yuanping Nie",
+        "Xiaohui Kuang"
+      ],
+      "year": 2022,
+      "venue": "International Conference on Data Science in Cyberspace",
+      "doi": "10.1109/dsc55868.2022.00057",
+      "arxiv_id": null,
+      "s2_paper_id": "6c8ad2d2e977e0a45d987959fc3e4adcde10e4ca",
+      "url": "https://doi.org/10.1109/dsc55868.2022.00057",
+      "open_access_pdf": null,
+      "abstract": "Vulnerabilities in database management system (DBMS) can cause serious security problems affecting hundreds of millions of software systems. Fuzzing is an effective vulnerability mining technology, but few studies utilize neural network language model (NNLM) to fuzz DBMS. In this paper, we explore this technical roadmap and use the sequence model to automatically generate test cases to fuzz DBMS. One of the advantages of the method is that it can effectively test the DBMS in a black-box situation. We implemented our tool NNFuzz and evaluated it on SQLite. Experimental results show that NNFuzz can generate valid test cases and achieve higher code coverage than the initial training set.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/dsc55868.2022.00057",
+              "source_type": "paper_citation_context",
+              "excerpt": "Query Partitioning [21] uses ternary logical partitioning to convert the original query into three partitioned queries, and compares whether the results of the original query are consistent with the union of the partitioned queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:aec873cfd8e3011521e92931025079785c3b65953c7e1f21e58ecb403eb45423",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/dsc55868.2022.00057",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS[19] selects the pivot row in advance, and then constructs the SQL query based on the pivot row.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a64597c47f95b9de15e1b951897c4ba6ec22ac56a10e7e2a1dfef34cecf06d35",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/dsc55868.2022.00057",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [20] transforms the SQL query into an equivalent form that will not be optimized by the query optimizer, and then executes the original query and the equivalent form separately.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:75a3dfb7caa60536d188ac2f6fed4eb17f3b28a5f810f92446db5b7bf81324a4",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/dsc55868.2022.00057",
+              "source_type": "paper_citation_context",
+              "excerpt": "When a correctness error is triggered, the DBMS does not crash, but the query returns incorrect query results [18, 19, 20, 21].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e0467c5b2c4889b06ee6407de064e393af19c3d631a913cf9be31167e4514843",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/6c8ad2d2e977e0a45d987959fc3e4adcde10e4ca",
+        "source_type": "paper",
+        "content_sha256": "sha256:22eef313c370108adad3c158eb6274b19e7fb88c4db456a11e1e107b09f09f09"
+      }
+    },
+    {
+      "id": "paper:arxiv:2206.08530",
+      "title": "GDsmith: Detecting Bugs in Graph Database Engines",
+      "authors": [
+        "Weisheng Lin",
+        "Ziyue Hua",
+        "Luyao Ren",
+        "Z. Li",
+        "Lu Zhang",
+        "Tao Xie"
+      ],
+      "year": 2022,
+      "venue": "arXiv.org",
+      "doi": null,
+      "arxiv_id": "2206.08530",
+      "s2_paper_id": "71dd291b4be0d074982f521ff654468755084a76",
+      "url": "https://arxiv.org/abs/2206.08530",
+      "open_access_pdf": "https://arxiv.org/pdf/2206.08530",
+      "abstract": "Graph database engines stand out in the era of big data for their efficiency of modeling and processing linked data. There is a strong need of testing graph database engines. However, random testing, the most practical way of automated test generation, faces the challenges of semantic validity, non-empty result, and behavior diversity to detect bugs in graph database engines. To address these challenges, in this paper, we propose GDsmith, the first black-box approach for testing graph database engines. It ensures that each randomly generated Cypher query satisfies the semantic requirements via skeleton generation and completion. GDsmith includes our technique to increase the probability of producing Cypher queries that return non-empty results by leveraging three types of structural mutation strategies. GDsmith also includes our technique to improve the behavior diversity of the generated Cypher queries by selecting property keys according to their previous frequencies when generating new queries. Our evaluation results demonstrate that GDsmith is effective and efficient for automated query generation and substantially outperforms the baseline. GDsmith successfully detects 27 previously unknown bugs on the released versions of three popular open-source graph database engines and receive positive feedback from their developers.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2206.08530",
+              "source_type": "paper_citation_context",
+              "excerpt": "NoREC [24] detects bugs in relational database engines by applying a semantics-preserving transformation to a given SQL query to disable the engine’s optimizations and addresses PQS’ high implementation effort.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b8455ed0dea29e7061ebc63cd69a229140ecfa00f05fbef535160a78a7ede59d",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2206.08530",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS [26] detects wrong-result bugs by checking whether a specific record is fetched correctly.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:f44c69e752618e677d3b2cf9ab8e24af9f317be7d26c35c82de2a904bc621367",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2206.08530",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [25] derives multiple SQL queries that compute a partial result of the initial query.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0fb3a6772a95bade10d9cc4026075dd277f7894954a6cf39c758203ad63ffe7f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2206.08530",
+              "source_type": "paper_citation_context",
+              "excerpt": "TLP [25] derives multiple SQL queries that",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:15629a16cbdee3b8bd3dad538b880e0ba7558ee2f008e7a097bb99d5ea0fc4ef",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2206_08530.json",
+              "source_type": "paper",
+              "excerpt": "Its framework is derived from SQLancer [ 23] (which is a tool to automatically test relational database engines).",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, 4.1.2 Implementation. We implement the GDsmith prototype with, page 7.",
+              "content_sha256": "sha256:4eaa475abcdb654d7a1a2cb88bf111012f93d8fb3a80c01e7866aaab075b00df",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/71dd291b4be0d074982f521ff654468755084a76",
+        "source_type": "paper",
+        "content_sha256": "sha256:768979cf5f7228419c90b804b52b688017d91daeec46a1da8a3e174d3e74eb39"
+      },
+      "artifacts": [
+        {
+          "url": "https://github.com/ddaa2000/GDsmith",
+          "kind": "github",
+          "inspected": true,
+          "inspected_at": "2026-09-10T15:15:32Z",
+          "sqlancer_markers": [
+            "renamed_sqlancer_package",
+            "sqlancer_source_content_match"
+          ],
+          "marker_evidence": [
+            {
+              "source_url": "https://github.com/ddaa2000/GDsmith",
+              "source_type": "github_repository",
+              "excerpt": "# GDsmith",
+              "note": "Repository is named after GDsmith, the tool this paper says it built, and is about database testing.",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z",
+              "excerpt_is_verbatim": true
+            },
+            {
+              "source_url": "https://github.com/ddaa2000/GDsmith/blob/master/src/main/java/org/example/gdsmith/Randomly.java",
+              "source_type": "github_code",
+              "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.ALPHANUMERIC;",
+              "excerpt_is_verbatim": true,
+              "note": "src/main/java/org/example/gdsmith/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.example.gdsmith (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+              "content_sha256": "sha256:d09fa9dddb2efa23e9184c0a8e2c71c39cdaffe6bde161da903434f62c4c4711",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-10T15:15:32Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "id": "paper:doi:10.1145/3551349.3560431",
+      "title": "Griffin : Grammar-Free DBMS Fuzzing",
+      "authors": [
+        "Jingzhou Fu",
+        "Jie Liang",
+        "Zhiyong Wu",
+        "Mingzhe Wang",
+        "Yu Jiang"
+      ],
+      "year": 2022,
+      "venue": "International Conference on Automated Software Engineering",
+      "doi": "10.1145/3551349.3560431",
+      "arxiv_id": null,
+      "s2_paper_id": "5b15efecca0f9f42e9d7842a8ec6689d2f45bc6a",
+      "url": "https://doi.org/10.1145/3551349.3560431",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3551349.3560431",
+      "abstract": "Fuzzing is a promising approach to DBMS testing. One crucial component in DBMS fuzzing is grammar: since DBMSs enforce strict validation on inputs, a grammar improves fuzzing efficiency by generating syntactically- and semantically-correct SQL statements. However, due to the vast differences in the complex grammar of various DBMSs, it is painstaking to adapt these fuzzers to them. Considering that lots of DBMSs are not yet well tested, there is an urgent need for an effective DBMS fuzzing approach that is free from grammar dependencies. In this paper, we propose Griffin, a grammar-free mutation based DBMS fuzzer. Rather than relying on grammar, Griffin summarizes the DBMS’s state into metadata graph, a lightweight data structure which improves mutation correctness in fuzzing. Specifically, it first tracks the metadata of the statements in built-in SQL test cases as they are executed, and constructs the metadata graph to describe the dependencies between metadata and statements iteratively. Based on the graphs, it reshuffles statements and employs metadata-guided substitution to correct semantic errors. We evaluate Griffin on MariaDB, SQLite, PostgreSQL, and DuckDB. Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Squirrel, respectively. In total, Griffin finds 55 previously unknown bugs with 13 CVEs assigned.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3551349.3560431",
+              "source_type": "paper_citation_context",
+              "excerpt": "Compared with SQLancer, SQLsmith, and Sqirrel, Griffin covered 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches and found 27, 27, and 22 more bugs in 12 hours on SQLite, DuckDB, MariaDB, and PostgreSQL, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:247f209894cee112c6cbf7938f37b7a6b7db34ccb7041d28ec9b3b2df024d6cd",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3551349.3560431",
+              "source_type": "paper_citation_context",
+              "excerpt": "The two generation-based fuzzers, SQLancer and SQLsmith, can only generate SQL statements based on their predefined models, which cover only a portion of the entire SQL grammar of the DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9d8ff08a93bdb8274cc3ec582ba31b7b68e7073db0dae3be3d7b4900c3a991d5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3551349.3560431",
+              "source_type": "paper_citation_context",
+              "excerpt": "Griffin covers 73.43%-274.70%, 80.47%- 312.89%, 43.80%-199.11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Sqirrel, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b749b7ca694df4ce49dd058e79a93f604656c050d7bfe33ac44298260a8696f6",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3551349.3560431",
+              "source_type": "paper_citation_context",
+              "excerpt": "Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Sqirrel, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b61bd0ffeacfa9fbd41cad779404516bc2ee63bb8ab21013baf24548a8ea3e58",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3551349.3560431",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLsmith can only generate SELECT statements, while SQLancer can not generate DELETE statements, because their generation models are based on the limited SQL grammar.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:95a8852de2211ad1c8ed63b56c849e23b63651f34560670e701e6fcd7540c90f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3551349.3560431",
+              "source_type": "paper_citation_context",
+              "excerpt": "Specifically, Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches than SQLancer, SQLsmith, and Sqirrel after fuzzing 12 hours, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:be1d963fab64b2a1371096a641e0c04234996fddfd5b97cb552c2ca837e5d615",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3551349_3560431.json",
+              "source_type": "paper",
+              "excerpt": "11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Sqirrel, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, page 1.",
+              "content_sha256": "sha256:a4c6c2d73c2c5665993c464d6f65552f3606bdc84b307de258c729576806a5a2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3551349_3560431.json",
+              "source_type": "paper",
+              "excerpt": "11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Sqirrel, respectively.",
+              "excerpt_is_verbatim": true,
+              "note": "M4 in the paper's extracted text, page 2.",
+              "content_sha256": "sha256:a4c6c2d73c2c5665993c464d6f65552f3606bdc84b307de258c729576806a5a2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3551349_3560431.json",
+              "source_type": "paper",
+              "excerpt": "To evaluate the effectiveness of Griffin, we choose SQLsmith, SQLancer, and Sqirrel, which are widely used in industry and academia, for performance comparison.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, page 7.",
+              "content_sha256": "sha256:faf1a624127874627d810327a63f36ccbedb694a554ceaa3393bc22524042dfe",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/5b15efecca0f9f42e9d7842a8ec6689d2f45bc6a",
+        "source_type": "paper",
+        "content_sha256": "sha256:e36df57bbcf7b069eed3d7a135a89d796a2b8c76c8bdcc754533082e134de2f4"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3563835.3567662",
+      "title": "Intramorphic Testing: A New Approach to the Test Oracle Problem",
+      "authors": [
+        "Manuel Rigger",
+        "Zhendong Su"
+      ],
+      "year": 2022,
+      "venue": "SIGPLAN symposium on New ideas, new paradigms, and reflections on programming and software",
+      "doi": "10.1145/3563835.3567662",
+      "arxiv_id": "2210.11228",
+      "s2_paper_id": "2616a143f10df0500639a93360521ea2d3706bfc",
+      "url": "https://doi.org/10.1145/3563835.3567662",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3563835.3567662",
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3563835.3567662",
+              "source_type": "paper_citation_context",
+              "excerpt": "Metamorphic testing is a high-level concept and finding effective MRs is often challenging; MRs for testing various systems such as compilers [15], database engines [22, 23], SMT solvers [30], Android apps [27], as well as object detection systems [28] have been proposed in the literature.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:857d3d392df29d9cef356937fbf475f6e2f8fe58e39c83e32d85c969b164dc5d",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/2616a143f10df0500639a93360521ea2d3706bfc",
+        "source_type": "paper",
+        "content_sha256": "sha256:2240aa550cc992af9be2181bef6a23292956cdf62c04a9902177f79891344512"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "A test oracle determines whether a system behaves correctly for a given input. Automatic testing techniques rely on an automated test oracle to test the system without user interaction. Important families of automated test oracles include Differential Testing and Metamorphic Testing, which are both black-box approaches; that is, they provide a test oracle that is oblivious to the system’s internals. In this work, we propose Intramorphic Testing as a white-box methodology to tackle the test oracle problem. To realize an Intramorphic Testing approach, a modified version of the system is created, for which, given a single input, a test oracle can be provided that relates the output of the original and modified systems. As a concrete example, by replacing a greater-equals operator in the implementation of a sorting algorithm with smaller-equals, it would be expected that the output of the modified implementation is the reverse output of the original implementation. In this paper, we introduce the methodology and illustrate it via a set of use cases."
+    },
+    {
+      "id": "paper:doi:10.1109/ase56229.2023.00106",
+      "title": "Perfce: Performance Debugging on Databases with Chaos Engineering-Enhanced Causality Analysis",
+      "authors": [
+        "Zhenlan Ji",
+        "Pingchuan Ma",
+        "Shuai Wang"
+      ],
+      "year": 2022,
+      "venue": "International Conference on Automated Software Engineering",
+      "doi": "10.1109/ase56229.2023.00106",
+      "arxiv_id": "2207.08369",
+      "s2_paper_id": "d0da09b12d16ea685532cdd47185ad2b888fc03a",
+      "url": "https://doi.org/10.1109/ase56229.2023.00106",
+      "open_access_pdf": "https://arxiv.org/pdf/2207.08369",
+      "abstract": "Debugging performance anomalies in databases is challenging. Causal inference techniques enable qualitative and quantitative root cause analysis of performance downgrades. Nevertheless, causality analysis is challenging in practice, particularly due to limited observability. Recently, chaos engineering (CE) has been applied to test complex software systems. CE frameworks mutate chaos variables to inject catastrophic events (e.g., network slowdowns) to stress-test these software systems. The systems under chaos stress are then tested (e.g., via differential testing) to check if they retain normal functionality, such as returning correct SQL query outputs even under stress. To date, CE is mainly employed to aid software testing. This paper identifies the novel usage of CE in diagnosing performance anomalies in databases. Our framework, PERFCE, has two phases - offline and online. The offline phase learns statistical models of a database using both passive observations and proactive chaos experiments. The online phase diagnoses the root cause of performance anomalies from both qualitative and quantitative aspects on-the-fly. In evaluation, Perfce outperformed previous works on synthetic datasets and is highly accurate and moderately expensive when analyzing real-world (distributed) databases like MySQL and TiDB.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/ase56229.2023.00106",
+              "source_type": "paper_citation_context",
+              "excerpt": "This setup expands the standard “differential testing” procedure [37–39], requiring consistency between the experimental group and reference even under CE stress.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:02821a205fd3f642db56616ac89494d92d2978df16a2e68e21844004b39801ff",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/d0da09b12d16ea685532cdd47185ad2b888fc03a",
+        "source_type": "paper",
+        "content_sha256": "sha256:469f5c24da1adcd11cab03a5b11703f8a6655c0c9e68f22adccc7e718f4c6f7c"
+      }
+    },
+    {
+      "id": "paper:doi:10.1007/s13222-022-00423-0",
+      "title": "Report from Dagstuhl Seminar 21442: Ensuring the Reliability and Robustness of Database Management Systems",
+      "authors": [
+        "Manuel Rigger",
+        "Alexander Böhm",
+        "M. Christakis",
+        "Eric Lo"
+      ],
+      "year": 2022,
+      "venue": "Datenbank-Spektrum",
+      "doi": "10.1007/s13222-022-00423-0",
+      "arxiv_id": null,
+      "s2_paper_id": "fb419d29d7be63fd12e809891f4abca66bcdff9d",
+      "url": "https://doi.org/10.1007/s13222-022-00423-0",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/fb419d29d7be63fd12e809891f4abca66bcdff9d",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/fb419d29d7be63fd12e809891f4abca66bcdff9d",
+        "source_type": "paper",
+        "content_sha256": "sha256:200cb906b2002af24b56fce8fc1c740379832042c21dd2a38fbea0ae844ec7cf"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": null
+    },
+    {
+      "id": "paper:doi:10.1145/3533767.3534364",
+      "title": "Unicorn: detect runtime errors in time-series databases with hybrid input synthesis",
+      "authors": [
+        "Zhiyong Wu",
+        "Jie Liang",
+        "Mingzhe Wang",
+        "Chijin Zhou",
+        "Yu Jiang"
+      ],
+      "year": 2022,
+      "venue": "International Symposium on Software Testing and Analysis",
+      "doi": "10.1145/3533767.3534364",
+      "arxiv_id": null,
+      "s2_paper_id": "1ac6b3a3904b6c9022206be7c0b9d180b84b1fb4",
+      "url": "https://doi.org/10.1145/3533767.3534364",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3533767.3534364",
+      "abstract": "The ubiquitous use of time-series databases in the safety-critical Internet of Things domain demands strict security and correctness. One successful approach in database bug detection is fuzzing, where hundreds of bugs have been detected automatically in relational databases. However, it cannot be easily applied to time-series databases: the bulk of time-series logic is unreachable because of mismatched query specifications, and serious bugs are undetectable because of implicitly handled exceptions. In this paper, we propose Unicorn to secure time-series databases with automated fuzzing. First, we design hybrid input synthesis to generate high-quality queries which not only cover time-series features but also ensure grammar correctness. Then, Unicorn uses proactive exception detection to discover minuscule-symptom bugs which hide behind implicit exception handling. With the specialized design oriented to time-series databases, Unicorn outperforms the state-of-the-art database fuzzers in terms of coverage and bugs. Specifically, Unicorn outperforms SQLsmith and SQLancer on widely used time-series databases IoTDB, KairosDB, TimescaleDB, TDEngine, QuestDB, and GridDB in the number of basic blocks by 21%-199% and 34%-693%, respectively. More importantly, Unicorn has discovered 42 previously unknown bugs.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/1ac6b3a3904b6c9022206be7c0b9d180b84b1fb4",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3533767_3534364.json",
+              "source_type": "paper",
+              "excerpt": "We also adapted the industrial fuzzers SQLancer [ 25] and SQLsmith [ 31] for comparison.",
+              "excerpt_is_verbatim": true,
+              "note": "M2 in the paper's extracted text, 1 INTRODUCTION, page 2.",
+              "content_sha256": "sha256:b971713d8e73ea72839af0ca8decf054a4c306ade3e9ddf5ce445421a3b30fc8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3533767_3534364.json",
+              "source_type": "paper",
+              "excerpt": "6 EVALUATION In this section, we evaluate the effectiveness of Unicorn in terms of coverage and bug discovery against state-of-the-art database fuzzers—SQLsmith and SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 4.2.1 Agent. As Figure 2 shows, time-series databases register a, page 7.",
+              "content_sha256": "sha256:5aed1f35ef1d37fe179700fdc6c372682b9142f1aa83dfe12963a398630c09ce",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3533767_3534364.json",
+              "source_type": "paper",
+              "excerpt": "2 Overall Performance Unicorn performs better than SQLsmith and SQLancer both in the number of basic blocks covered and bugs triggered.",
+              "excerpt_is_verbatim": true,
+              "note": "M9 in the paper's extracted text, 6.2 Overall Performance, page 7.",
+              "content_sha256": "sha256:c8194b0dfd4b8ec2152a7228c57aee7cc9adcedd800b9b2874d2ed90dd7ff5a2",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3533767_3534364.json",
+              "source_type": "paper",
+              "excerpt": "Furthermore, 2 bugs were also found by SQLsmith, while SQLancer did not detect any bug.",
+              "excerpt_is_verbatim": true,
+              "note": "M22 in the paper's extracted text, 6.2 Overall Performance, page 9.",
+              "content_sha256": "sha256:20e76d3c6441690d51abd37f1d2c9e12dd985cdd10ee950ed4a6ffa1c6700e22",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3533767_3534364.json",
+              "source_type": "paper",
+              "excerpt": "•We evaluate Unicorn on 6 popular time-series databases against state-of-the-art fuzzers SQLsmith and SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M3 in the paper's extracted text, 1 INTRODUCTION, page 2.",
+              "content_sha256": "sha256:252b05d9717c156b3055c392c00e9eea5001e574cef87ef6789a3645643ff0b8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1145_3533767_3534364.json",
+              "source_type": "paper",
+              "excerpt": "6 EVALUATION In this section, we evaluate the effectiveness of Unicorn in terms of coverage and bug discovery against state-of-the-art database fuzzers—SQLsmith and SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, 4.2.1 Agent. As Figure 2 shows, time-series databases register a, page 7.",
+              "content_sha256": "sha256:5aed1f35ef1d37fe179700fdc6c372682b9142f1aa83dfe12963a398630c09ce",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/1ac6b3a3904b6c9022206be7c0b9d180b84b1fb4",
+        "source_type": "paper",
+        "content_sha256": "sha256:b81724479ec94f948e708356452441f9a693ed17c120188edfab850bdbbb08ea"
+      }
+    },
+    {
+      "id": "paper:doi:10.1007/978-3-030-71058-3_5",
+      "title": "Artemis: An Automatic Test Suite Generator for Large Scale OLAP Database",
+      "authors": [
+        "Kaiming Mi",
+        "Chunxi Zhang",
+        "Weining Qian",
+        "Rong Zhang"
+      ],
+      "year": 2021,
+      "venue": "Lecture notes in computer science",
+      "doi": "10.1007/978-3-030-71058-3_5",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "url": "https://doi.org/10.1007/978-3-030-71058-3_5",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://openalex.org/W3134710759",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "OpenAlex records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://openalex.org/W3134710759",
+        "source_type": "paper",
+        "content_sha256": "sha256:e917c42c7fe4d69f1cec90650555f59ade2128915098cc39af474bfca3135068"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse43902.2021.00137",
+      "title": "Data-Oriented Differential Testing of Object-Relational Mapping Systems",
+      "authors": [
+        "T. Sotiropoulos",
+        "Stefanos Chaliasos",
+        "Vaggelis Atlidakis",
+        "Dimitris Mitropoulos",
+        "D. Spinellis"
+      ],
+      "year": 2021,
+      "venue": "International Conference on Software Engineering",
+      "doi": "10.1109/icse43902.2021.00137",
+      "arxiv_id": null,
+      "s2_paper_id": "4bc8bd72ffa9bda2e3a7aef483008935a7a97656",
+      "url": "https://doi.org/10.1109/icse43902.2021.00137",
+      "open_access_pdf": null,
+      "abstract": "We introduce, what is to the best of our knowledge, the first approach for systematically testing Object-Relational Mapping (ORM) systems. Our approach leverages differential testing to establish a test oracle for ORM-specific bugs. Specifically, we first generate random relational database schemas, set up the respective databases, and then, we query these databases using the APIs of the ORM systems under test. To tackle the challenge that ORMs lack a common input language, we generate queries written in an abstract query language. These abstract queries are translated into concrete, executable ORM queries, which are ultimately used to differentially test the correctness of target implementations. The effectiveness of our method heavily relies on the data inserted to the underlying databases. Therefore, we employ a solver-based approach for producing targeted database records with respect to the constraints of the generated queries. We implement our approach as a tool, called CYNTHIA, which found 28 bugs in five popular ORM systems. The vast majority of these bugs are confirmed (25 / 28), more than half were fixed (20 / 28), and three were marked as release blockers by the corresponding developers.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse43902.2021.00137",
+              "source_type": "paper_citation_context",
+              "excerpt": "…data generation approach and its suitability for differential testing, we compare it against a simplistic approach that populates the database with random records a-priori [21], [27], i.e., it inserts data while setting up the tables, without considering the constraints of the generated queries.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0c7003cc2b4977ec05a1aacfcd45d381302401935975839ccf5fb208b0ac177c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse43902.2021.00137",
+              "source_type": "paper_citation_context",
+              "excerpt": "More recently, Rigger et al. [21] proposed the Pivoted Query Synthesis (PQS) technique for testing database engines.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e6639b9a25d5d1a8f765cb4de6f39eae4863dc1f8e0f806649b52053b3cee0f3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse43902.2021.00137",
+              "source_type": "paper_citation_context",
+              "excerpt": "Although DBMSs share common functionality, they differ significantly from each other [21].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:08d8b2782d445428f55d8c833e9fa116fca5b262ca4d27803a864bfbf3639257",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse43902.2021.00137",
+              "source_type": "paper_citation_context",
+              "excerpt": "In their most recent work, they propose Ternary Logic Partitioning (TLP) [38].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:771cfcb48b9a717a7627c6366ceb501d77026de475bd536a4ca03e5731385105",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/4bc8bd72ffa9bda2e3a7aef483008935a7a97656",
+        "source_type": "paper",
+        "content_sha256": "sha256:49d55973095e223066175eacfcbc2a2f11199ad73c4c56b7f93888244b57fba4"
+      }
+    },
+    {
+      "id": "paper:doi:10.35335/jurnalmantik.vol5.2021.1448.pp1065-1071",
+      "title": "Database System Development Life Cycle (DSDLC) on System Libraries for Data Manipulation Language (DML) Using SQL Server 2008",
+      "authors": [
+        "Didik Setiyadi"
+      ],
+      "year": 2021,
+      "venue": "Jurnal Mantik",
+      "doi": "10.35335/jurnalmantik.vol5.2021.1448.pp1065-1071",
+      "arxiv_id": null,
+      "s2_paper_id": "1438f9b541946409662ebf91407f849aa42fe856",
+      "url": "https://doi.org/10.35335/jurnalmantik.vol5.2021.1448.pp1065-1071",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/1438f9b541946409662ebf91407f849aa42fe856",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/1438f9b541946409662ebf91407f849aa42fe856",
+        "source_type": "paper",
+        "content_sha256": "sha256:c0ac25edf1049ab20f9e907ae97ba05d3ac029bce074bf9bbd626ec98466a52d"
+      },
+      "abstract": null
+    },
+    {
+      "id": "paper:doi:10.1007/978-3-030-88494-9_12",
+      "title": "Differential Monitoring",
+      "authors": [
+        "Fabian Mühlböck",
+        "T. Henzinger"
+      ],
+      "year": 2021,
+      "venue": "Runtime Verification",
+      "doi": "10.1007/978-3-030-88494-9_12",
+      "arxiv_id": null,
+      "s2_paper_id": "f232b1bd730f6c4d4a48be5f642074dffb7a1ed0",
+      "url": "https://doi.org/10.1007/978-3-030-88494-9_12",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1007/978-3-030-88494-9_12",
+              "source_type": "paper_citation_context",
+              "excerpt": "This technique has been fruitfully applied to finding bugs in Javascript debuggers [31], C compilers [41], and SQL databases [39, 37, 36].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:14cf1c98657d5ea288e1f29b7a6a6c6845dfb7867a648f63001faa1348c2aec1",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1007/978-3-030-88494-9_12",
+              "source_type": "paper_citation_context",
+              "excerpt": "This technique has been fruitfully applied to finding bugs in Javascript debuggers [31], C compilers [42], and SQL databases [40,38,37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:77ef933e00b711200b68d829887a83c3b376b6bf4e8f5d4d6f0d291e80be0fe5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/f232b1bd730f6c4d4a48be5f642074dffb7a1ed0",
+        "source_type": "paper",
+        "content_sha256": "sha256:3c45c076bce4a464817c1eac23b201c2aa535ce59376b94c571de09b4969ef9c"
+      }
+    },
+    {
+      "id": "paper:s2:b6c64dbe0130ef1bf6af12266b958a5d5396101f",
+      "title": "Differential Monitoring - Technical Report ⋆",
+      "authors": [
+        "Fabian Muehlboeck"
+      ],
+      "year": 2021,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "b6c64dbe0130ef1bf6af12266b958a5d5396101f",
+      "url": "https://www.semanticscholar.org/paper/b6c64dbe0130ef1bf6af12266b958a5d5396101f",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/b6c64dbe0130ef1bf6af12266b958a5d5396101f",
+              "source_type": "paper_citation_context",
+              "excerpt": "This technique has been fruitfully applied to finding bugs in Javascript debuggers [31], C compilers [41], and SQL databases [39,37,36].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ba89b2245189a9a4329b17f62018c69a864e040924b56ca32093bcc54158aa49",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/b6c64dbe0130ef1bf6af12266b958a5d5396101f",
+        "source_type": "paper",
+        "content_sha256": "sha256:393e47de092c08d86e8be74defb309101ced5c8d7407f0f465f89b39b071fdc0"
+      },
+      "abstract": null
+    },
+    {
+      "id": "paper:doi:10.14711/thesis-991012980220103412",
+      "title": "Duplicate-sensitivity Guided Transformation Synthesis for DBMS Correctness Bug Detection",
+      "authors": [
+        "Yushan Zhang",
+        "Peisen Yao",
+        "Rongxin Wu",
+        "Charles Zhang"
+      ],
+      "year": 2021,
+      "venue": "arXiv.org",
+      "doi": "10.14711/thesis-991012980220103412",
+      "arxiv_id": "2107.03660",
+      "s2_paper_id": "e27baad105b7fba74658580ed2fb7439328b053c",
+      "url": "https://doi.org/10.14711/thesis-991012980220103412",
+      "open_access_pdf": "https://arxiv.org/pdf/2107.03660",
+      "abstract": "Database Management System (DBMS) plays a core role in modern software from mobile apps to online banking. It is critical that DBMS should provide correct data to all applications. When the DBMS returns incorrect data, a correctness bug is triggered. Current production-level DBMSs still suffer from insufficient testing due to the limited hand-written test cases. Recently several works proposed to automatically generate many test cases with query transformation, a process of generating an equivalent query pair and testing a DBMS by checking whether the system returns the same result set for both queries. However, all of them still heavily rely on manual work to provide a transformation which largely confines their exploration of the valid input query space. This paper introduces duplicate-sensitivity guided transformation synthesis which automatically finds new transformations by first synthesizing many candidates then filtering the nonequivalent ones. Our automated synthesis is achieved by mutating a query while keeping its duplicate sensitivity, which is a necessary condition for query equivalence. After candidate synthesis, we keep the mutant query which is equivalent to the given one by using a query equivalent checker. Furthermore, we have implemented our idea in a tool Eqsql and used it to test the production-level DBMSs. In two months, we detected in total 30 newly confirmed and unique bugs in MySQL, TiDB and CynosDB.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+              "source_type": "paper_citation_context",
+              "excerpt": "Similarly, NoREC [8] constructs a non-optimizing version of a query and compares the result.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b089ca4212702a9aab9238dd07b3557ef99d93c9e93b781b375587ff206c4f32",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+              "source_type": "paper_citation_context",
+              "excerpt": "We omitted comparison with SQLancer for three reasons.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b1a9181cdac5832a126a23a4657d99ae415e6f6a477a5bbdcdaae20db569bb2e",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+              "source_type": "paper_citation_context",
+              "excerpt": "Though PQS can generate the test oracle more practically than the previous work, it can only reveal bugs with a symptom of missing the picked row.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9ad0c557e3dd02e31a6a6981f33183e1ff168abd168857bb340d66299a516f16",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+              "source_type": "paper_citation_context",
+              "excerpt": "Previous work NoREC [8] can only detect such bugs because it only manipulates the “where” predicate to create an unoptimized query mutant.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b810cece359ffc8b5c76b92453e865877433d406efeab4be0f0d456002060bef",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+              "source_type": "paper_citation_context",
+              "excerpt": "of the existing studies [1], [8], [10] presumes that the generated query pair should explicitly include a “where” predicate (e.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4ae2ab3a5384ee3356f8753e4d08134523717393270b19b736348b994209a49a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14711/thesis-991012980220103412",
+              "source_type": "paper_citation_context",
+              "excerpt": "DBMSs return wrong result sets [1], though these systems have been extensively tested during their development [2], [3].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:eb67dbd02697fd192fc2530e2eae1e2a617fa0463206157d03eff702716d62af",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/e27baad105b7fba74658580ed2fb7439328b053c",
+        "source_type": "paper",
+        "content_sha256": "sha256:3aa15869e4d341d9dc97c0f8c5c3ebc18f855dfea25a4a08d359f2d23fc28f14"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3448016.3457559",
+      "title": "FoundationDB: A Distributed Unbundled Transactional Key Value Store",
+      "authors": [
+        "Jingyu Zhou",
+        "Meng Xu",
+        "A. Shraer",
+        "B. Namasivayam",
+        "Alex Miller",
+        "Evan Tschannen",
+        "Steve Atherton",
+        "Andrew J. Beamon",
+        "Rusty Sears",
+        "J. Leach",
+        "D. Rosenthal",
+        "X. Dong",
+        "Willie B. Wilson",
+        "Ben Collins",
+        "David Scherer",
+        "Alec Grieser",
+        "Young Liu",
+        "Alvin Moore",
+        "Bhaskar Muppana",
+        "Xi-sheng Su",
+        "Vishesh Yadav"
+      ],
+      "year": 2021,
+      "venue": "SIGMOD Conference",
+      "doi": "10.1145/3448016.3457559",
+      "arxiv_id": null,
+      "s2_paper_id": "2e2a31b0bb5c2ba3c7f4cf429e2dc81d4327af9d",
+      "url": "https://doi.org/10.1145/3448016.3457559",
+      "open_access_pdf": null,
+      "abstract": "FoundationDB is an open source transactional key value store created more than ten years ago. It is one of the first systems to combine the flexibility and scalability of NoSQL architectures with the power of ACID transactions (a.k.a. NewSQL). FoundationDB adopts an unbundled architecture that decouples an in-memory transaction management system, a distributed storage system, and a built-in distributed configuration system. Each sub-system can be independently provisioned and configured to achieve the desired scalability, high-availability and fault tolerance properties. FoundationDB uniquely integrates a deterministic simulation framework, used to test every new feature of the system under a myriad of possible faults. This rigorous testing makes FoundationDB extremely stable and allows developers to introduce and release new features in a rapid cadence. FoundationDB offers a minimal and carefully chosen feature set, which has enabled a range of disparate systems (from semi-relational databases, document and object stores, to graph databases and more) to be built as layers on top. FoundationDB is the underpinning of cloud infrastructure at Apple, Snowflake and other companies, due to its consistency, robustness and availability for storing user data, system metadata and configuration, and other critical information.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3448016.3457559",
+              "source_type": "paper_citation_context",
+              "excerpt": "Finally there are numerous approaches to testing the correctness of database subsystems in the absence of faults, including the query engine [17, 57, 60] and concurrency control mechanism [63].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c638c7d7df79e093c670d4dbaa4ee9aa2d13d9d1823cb8c20881faccd8eeb7ea",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/2e2a31b0bb5c2ba3c7f4cf429e2dc81d4327af9d",
+        "source_type": "paper",
+        "content_sha256": "sha256:1f5f548121fe0097e64eed574f3047dd1621cce322ba8205a8ddc67f3ef9e286"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3485529",
+      "title": "Generative type-aware mutation for testing SMT solvers",
+      "authors": [
+        "Jiwon Park",
+        "Dominik Winterer",
+        "Chengyu Zhang",
+        "Zhendong Su"
+      ],
+      "year": 2021,
+      "venue": "Proc. ACM Program. Lang.",
+      "doi": "10.1145/3485529",
+      "arxiv_id": null,
+      "s2_paper_id": "3f650c0d0a046c1fb6c068d636b138e14ded77b9",
+      "url": "https://doi.org/10.1145/3485529",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3485529",
+      "abstract": "We propose Generative Type-Aware Mutation, an effective approach for testing SMT solvers. The key idea is to realize generation through the mutation of expressions rooted with parametric operators from the SMT-LIB specification. Generative Type-Aware Mutation is a hybrid of mutation-based and grammar-based fuzzing and features an infinite mutation space—overcoming a major limitation of OpFuzz, the state-of-the-art fuzzer for SMT solvers. We have realized Generative Type-Aware Mutation in a practical SMT solver bug hunting tool, TypeFuzz. During our testing period with TypeFuzz, we reported over 237 bugs in the state-of-the-art SMT solvers Z3 and CVC4. Among these, 189 bugs were confirmed and 176 bugs were fixed. Most notably, we found 18 soundness bugs in CVC4’s default mode alone. Several of them were two years latent (7/18). CVC4 has been proved to be a very stable SMT solver and has resisted several fuzzing campaigns.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/3f650c0d0a046c1fb6c068d636b138e14ded77b9",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/3f650c0d0a046c1fb6c068d636b138e14ded77b9",
+        "source_type": "paper",
+        "content_sha256": "sha256:f4e27d5769370fc2385678fce56237bfdc4d9c3e48c478116151ee4854bff758"
+      }
+    },
+    {
+      "id": "paper:doi:10.1109/icse-seip52600.2021.00042",
+      "title": "Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing",
+      "authors": [
+        "Mingzhe Wang",
+        "Zhiyong Wu",
+        "Xinyi Xu",
+        "Jie Liang",
+        "Chijin Zhou",
+        "Huafeng Zhang",
+        "Yu Jiang"
+      ],
+      "year": 2021,
+      "venue": "2021 IEEE/ACM 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP)",
+      "doi": "10.1109/icse-seip52600.2021.00042",
+      "arxiv_id": "2103.00804",
+      "s2_paper_id": "8356864e3802eaee426a1fec7ea838bf8d7d7e57",
+      "url": "https://doi.org/10.1109/icse-seip52600.2021.00042",
+      "open_access_pdf": "https://arxiv.org/pdf/2103.00804",
+      "abstract": "As an infrastructure for data persistence and analysis, Database Management Systems (DBMSs) are the cornerstones of modern enterprise software. To improve their correctness, the industry has been applying blackbox fuzzing for decades. Recently, the research community achieved impressive fuzzing gains using coverage guidance. However, due to the complexity and distributed nature of enterprise-level DBMSs, seldom are these researches applied to the industry. In this paper, we apply coverage-guided fuzzing to enterprise-level DBMSs from Huawei and Bloomberg LP. In our practice of testing GaussDB and Comdb2, we found major challenges in all three testing stages. The challenges are collecting precise coverage, optimizing fuzzing performance, and analyzing root causes. In search of a general method to overcome these challenges, we propose Ratel, a coverage-guided fuzzer for enterprise-level DBMSs. With its industry-oriented design, Ratel improves the feedback precision, enhances the robustness of input generation, and performs an online investigation on the root cause of bugs. As a result, Ratel outperformed other fuzzers in terms of coverage and bugs. Compared to industrial black box fuzzers SQLsmith and SQLancer, as well as coverage-guided academic fuzzer Squirrel, Ratel covered 38.38%, 106.14%, 583.05% more basic blocks than the best results of other three fuzzers in GaussDB, PostgreSQL, and Comdb2, respectively. More importantly, Ratel has discovered 32, 42, and 5 unknown bugs in GaussDB, Comdb2, and PostgreSQL.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1109/icse-seip52600.2021.00042",
+              "source_type": "paper_citation_context",
+              "excerpt": "…SQLsmith [1] triggers system bugs by continuously generating random SQL queries; RAGS [2] detects logic bugs by comparing the results of a query on multiple DBMSs; SQLancer [3] detects logic bugs by constructing an invariant oracle from different angles [4], Yu Jiang is the correspondence author.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b042bb711a3bdf478f5b83bd6c87362308b83fb18092554c7f68fe6339deb459",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse-seip52600.2021.00042",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, its pivoted query synthesis strategy [5] generates queries of which corresponding result table is supposed to include a specific row, and if the DBMS fails to fetch the row, a logic bug is discovered.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:ad0cc921f935c912919aa2310cf40e74ea5cfeee9dde90c019efc3548105d769",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1109/icse-seip52600.2021.00042",
+              "source_type": "paper_citation_context",
+              "excerpt": "In addition, SQLancer [3] integrates three different strategies [4], [5], [14] to construct invirant oracle to detect logic bugs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:395e873e84abdec7302ca5c9a6e813324879b7f8549a812ff16e819b94aa78c5",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse_seip52600_2021_00042.json",
+              "source_type": "paper",
+              "excerpt": "Compared to industrial black box fuzzers SQLsmith and SQLancer, as well as coverage-guided academic fuzzer Squirrel, RATEL covered 38.",
+              "excerpt_is_verbatim": true,
+              "note": "M1 in the paper's extracted text, page 1.",
+              "content_sha256": "sha256:a9e01c22b190bf96288cf03ba152e02923b3030eb964074fd8762e5d6db26dc0",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse_seip52600_2021_00042.json",
+              "source_type": "paper",
+              "excerpt": "Compared to SQLsmith, SQLancer, and Squirrel, it covered 38.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, I INTRODUCTION, page 2.",
+              "content_sha256": "sha256:71074b00a2237ea370395d7da429b88c262abcb3566534c41f9d2b6c55279396",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            },
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_doi_10_1109_icse_seip52600_2021_00042.json",
+              "source_type": "paper",
+              "excerpt": "SQLsmith, SQLancer, and Squirrel, which are chosen for our industry practice.",
+              "excerpt_is_verbatim": true,
+              "note": "M6 in the paper's extracted text, II BACKGROUND, page 2.",
+              "content_sha256": "sha256:1ac6acdc9781fedc52c0e2ff950fe0dd2383821b7ba59e6facd9c6cf85d3e6f8",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/8356864e3802eaee426a1fec7ea838bf8d7d7e57",
+        "source_type": "paper",
+        "content_sha256": "sha256:cf285ac6da170e3431693363fdc60c70b0d4318d203f64780130b7d7d3830162"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3468264.3468573",
+      "title": "Metamorphic testing of Datalog engines",
+      "authors": [
+        "Muhammad Numair Mansur",
+        "M. Christakis",
+        "Valentin Wüstholz"
+      ],
+      "year": 2021,
+      "venue": "ESEC/SIGSOFT FSE",
+      "doi": "10.1145/3468264.3468573",
+      "arxiv_id": null,
+      "s2_paper_id": "13cf48695d23f7e019d670a1bf059b7f28396992",
+      "url": "https://doi.org/10.1145/3468264.3468573",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3468264.3468573",
+      "abstract": "Datalog is a popular query language with applications in several domains. Like any complex piece of software, Datalog engines may contain bugs. The most critical ones manifest as incorrect results when evaluating queries—we refer to these as query bugs. Given the wide applicability of the language, query bugs may have detrimental consequences, for instance, by compromising the soundness of a program analysis that is implemented and formalized in Datalog. In this paper, we present the first metamorphic-testing approach for detecting query bugs in Datalog engines. We ran our tool on three mature engines and found 13 previously unknown query bugs, some of which are deep and revealed critical semantic issues.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/13cf48695d23f7e019d670a1bf059b7f28396992",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Pivoted Query Synthesis (PQS), Non-optimizing Reference Engine Construction (NoREC), Ternary Logic Partitioning (TLP).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/13cf48695d23f7e019d670a1bf059b7f28396992",
+        "source_type": "paper",
+        "content_sha256": "sha256:abf95b7d3e44ce90583e4c0a2c1bf644448daf8d8898b66cc73b8197d1f6380c"
+      }
+    },
+    {
+      "id": "paper:doi:10.14778/3494124.3494139",
+      "title": "MT-Teql: Evaluating and Augmenting Neural NLIDB on Real-world Linguistic and Schema Variations",
+      "authors": [
+        "Pingchuan Ma",
+        "Shuai Wang"
+      ],
+      "year": 2021,
+      "venue": "Proceedings of the VLDB Endowment",
+      "doi": "10.14778/3494124.3494139",
+      "arxiv_id": null,
+      "s2_paper_id": "18ad0da02b2207288a3fe7c19ee8d223a9ee3ef4",
+      "url": "https://doi.org/10.14778/3494124.3494139",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.14778/3494124.3494139",
+              "source_type": "paper_citation_context",
+              "excerpt": "To test DBMS with metamorphic relations, for instance, NoREC transforms a query into a non-optimized form and compares whether the optimized query and non-optimized query induce identical outputs [28].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:a1ddd8ac066da6f6ef9fbae3cfe4b0e462d83dfd05770a52f39d8cb416eedeaf",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3494124.3494139",
+              "source_type": "paper_citation_context",
+              "excerpt": "With recent advances in testing DBMS [28–30], software metamorphic testing-based approaches have become popular in assessing database systems.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8155dc3e7ec820b84ffa64eecef68eb388ab6543479b2043982b8b1daa95999f",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.14778/3494124.3494139",
+              "source_type": "paper_citation_context",
+              "excerpt": "To date, MT has achieved major success in detecting bugs in DBMS [28–30] and NLP-related models [19, 27, 37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:61b767b9b374610615b350d3fdac5ecb25ce08d1394ce8366c7fd07bbb341ebd",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/18ad0da02b2207288a3fe7c19ee8d223a9ee3ef4",
+        "source_type": "paper",
+        "content_sha256": "sha256:e16bb351f98884726410055b614bd8ed58073313da8a7d6c9515615feb5a8ec3"
+      },
+      "abstract": "Natural Language Interface to Database (NLIDB) translates human utterances into SQL queries and enables database interactions for non-expert users. Recently, neural network models have become a major approach to implementing NLIDB. However, neural NLIDB faces challenges due to variations in natural language and database schema design. For instance, one user intent or database conceptual model can be expressed in\n various forms.\n However, existing benchmarks, using hold-out datasets, cannot provide thorough understanding of how good neural NLIDBs really are in real-world situations and its robustness against such variations. A key difficulty is to annotate SQL queries for inputs under real-world variations, requiring considerable manual effort and expert knowledge.\n \n To systematically assess the robustness of neural NLIDBs without extensive manual effort, we propose MT-Teql, a unified framework to benchmark NLIDBs against real-world language and schema variations. Inspired by recent advances in DBMS metamorphic testing, MT-Teql implements semantics-preserving transformations on utterances and database schemas to generate their variants. NLIDBs can thus be examined for robustness utilizing utterances/schemas and their variants without requiring manual intervention.\n We benchmarked nine neural NLIDBs using 62,430 inputs and identified 15,433 defects. We analyzed potential root causes of defects and conducted a user study to show how MT-Teql can assist developers to systematically assess NLIDBs. We further show that the transformed (error-triggering) inputs can be used to augment popular NLIDBs and eliminate 46.5%(±5.0%) errors made by them without compromising their accuracy on standard benchmarks. We summarize lessons from this study that can provide insights to select and design NLIDBs that fit particular usage scenarios."
+    },
+    {
+      "id": "paper:doi:10.1142/s021819402150039x",
+      "title": "MTKeras: An Automated Metamorphic Testing Platform",
+      "authors": [
+        "Ye-Lin Liu",
+        "Z. Zhou",
+        "T. Chen",
+        "Yang Liu",
+        "Dave Towey"
+      ],
+      "year": 2021,
+      "venue": "International journal of software engineering and knowledge engineering",
+      "doi": "10.1142/s021819402150039x",
+      "arxiv_id": null,
+      "s2_paper_id": "6f07a14a3fe93d2938350e2a3b8ec16f3c27ac16",
+      "url": "https://doi.org/10.1142/s021819402150039x",
+      "open_access_pdf": "https://figshare.com/articles/thesis/MTKeras_An_Automated_Metamorphic_Testing_Platform/27669363",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1142/s021819402150039x",
+              "source_type": "paper_citation_context",
+              "excerpt": "Researchers developed a novel method called NonOptimizing Reference Engine Construction (NoREC) [6], which compares the results of the optimized and non-optimized versions of a SQL to evaluate the DBMS.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:39359f964e6fa7d8293da393597ac03c3961d39546a7aed5b5df83b4a329a11b",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1142/s021819402150039x",
+              "source_type": "paper_citation_context",
+              "excerpt": "It has become recognized as an important quality assurance paradigm for complex systems, such as for machine learning (ML) [4], [5] and database management systems (DBMSs) [6].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e8ed1c7ab3d3e96af1e86603aaa696feb2c74fb591be444c8dfd913ee7614d6d",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1142/s021819402150039x",
+              "source_type": "paper_citation_context",
+              "excerpt": "MTKeras was also configured for database management system (DBMS) testing by using the concept of Non-Optimizing Reference Engine Construction (NoREC) [6].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:3e1f6060cda9bda3cf7691dd692bf8e87a9b783de1fbd11946eeb69f7a52d085",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1142/s021819402150039x",
+              "source_type": "paper_citation_context",
+              "excerpt": "There are multiple methods for detecting bugs in a DBMS, such as Pivoted Query Synthesis (PQS) [47] and Random Generation of SQL (RAGS) [48].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5d7440ec17a4049ea86bc9e676597fcd5a38dbda3ca4614a5f7aec260e317f4f",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1142/s021819402150039x",
+              "source_type": "paper_citation_context",
+              "excerpt": "Database management system testing is also challenged by the test oracle problem [6].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:67533dfa3d3a8639bc8478b2fb59e7161627280dbc7bddf57ee674a5d83ff15e",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/6f07a14a3fe93d2938350e2a3b8ec16f3c27ac16",
+        "source_type": "paper",
+        "content_sha256": "sha256:7d48e0b6d800ca0e5d3b4689483bbd1dc5af281a70c32b8e7bff19c8f67a00d0"
+      },
+      "abstract": "This paper presents an automated, domain-independent, metamorphic testing platform called MTKeras. In this paper, we report on an investigation demonstrating the effectiveness and usability of MTKeras through five case studies in the four domains of image classification, sentiment analysis, search engines and database management systems. We also report on the effectiveness of combining metamorphic relation (input) patterns in individual metamorphic relations, enhancing the failure-finding abilities of the individual relations. The results of our experiments support combining patterns, and the use of MTKeras. The research reported in this paper shows the applicability of metamorphic relation patterns, and introduces a practical tool for the research community."
+    },
+    {
+      "id": "paper:doi:10.1145/3468264.3468540",
+      "title": "Skeletal approximation enumeration for SMT solver testing",
+      "authors": [
+        "Peisen Yao",
+        "Heqing Huang",
+        "Wensheng Tang",
+        "Qingkai Shi",
+        "Rongxin Wu",
+        "Charles Zhang"
+      ],
+      "year": 2021,
+      "venue": "ESEC/SIGSOFT FSE",
+      "doi": "10.1145/3468264.3468540",
+      "arxiv_id": null,
+      "s2_paper_id": "e6a4ddc65452fb2a01389b6e28e6607b8781710e",
+      "url": "https://doi.org/10.1145/3468264.3468540",
+      "open_access_pdf": null,
+      "abstract": "Ensuring the equality of SMT solvers is critical due to its broad spectrum of applications in academia and industry, such as symbolic execution and program verification. Existing approaches to testing SMT solvers are either too costly or find difficulties generalizing to different solvers and theories, due to the test oracle problem. To complement existing approaches and overcome their weaknesses, this paper introduces skeletal approximation enumeration (SAE), a novel lightweight and general testing technique for all first-order theories. To demonstrate its practical utility, we have applied the SAE technique to test Z3 and CVC4, two comprehensively tested, state-of-the-art SMT solvers. By the time of writing, our approach had found 71 confirmed bugs in Z3 and CVC4,55 of which had already been fixed.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3468264.3468540",
+              "source_type": "paper_citation_context",
+              "excerpt": "in many application domains such as bioinformatics [26], web services [24], compilers [36, 51], debuggers [53], databases [47], machine learning-based systems [31, 38], model counters [55], and SMT solvers [60].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:07bd9eebe3ab7928cedb53b077f55513bbd6af8a145ddfda1b1be0e9dac941c3",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/e6a4ddc65452fb2a01389b6e28e6607b8781710e",
+        "source_type": "paper",
+        "content_sha256": "sha256:ddf8203aa91dade60196c02542c5389b5141e1c1e16765b948d757c5d6ce1efa"
+      }
+    },
+    {
+      "id": "paper:s2:d88db0b52f31ed5444602a67e9a275faf57e15bf",
+      "title": "TCP-Fuzz: Detecting Memory and Semantic Bugs in TCP Stacks with Fuzzing",
+      "authors": [
+        "Yong-Hao Zou",
+        "Jia-Ju Bai",
+        "Jielong Zhou",
+        "Jianfeng Tan",
+        "Chenggang Qin",
+        "Shih-Min Hu"
+      ],
+      "year": 2021,
+      "venue": "USENIX Annual Technical Conference",
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "d88db0b52f31ed5444602a67e9a275faf57e15bf",
+      "url": "https://www.semanticscholar.org/paper/d88db0b52f31ed5444602a67e9a275faf57e15bf",
+      "open_access_pdf": null,
+      "abstract": null,
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/d88db0b52f31ed5444602a67e9a275faf57e15bf",
+              "source_type": "paper_citation_context",
+              "excerpt": "Similar to SQLancer [51] and libFuzzer [32], for in-consistencies that we identify as semantic bugs, we manually fix them using the developers’ patches or by ourselves, to reduce related inconsistencies.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c2ce4ba41adc3dfa8b35ba167827789543b42d5cb591f8535731df30f0aa8a5a",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/d88db0b52f31ed5444602a67e9a275faf57e15bf",
+        "source_type": "paper",
+        "content_sha256": "sha256:255724d83f6d815131e270c26d04452a1dd253d97cc4950b1597291b61dbb267"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3471485.3471491",
+      "title": "Technical Perspective DIAMetrics",
+      "authors": [
+        "P. Boncz"
+      ],
+      "year": 2021,
+      "venue": "SIGMOD record",
+      "doi": "10.1145/3471485.3471491",
+      "arxiv_id": null,
+      "s2_paper_id": "0a54335989389a1e25b4ca13050eeeabd04168e2",
+      "url": "https://doi.org/10.1145/3471485.3471491",
+      "open_access_pdf": "https://ir.cwi.nl/pub/30888",
+      "abstract": "Benchmarking database systems has a long and successful history in making industrial database systems comparable, and is also a cornerstone of quantifiable experimental data systems research. Creating good benchmarks has been described as something of an art [3]. One can inspire dataset and workload design from\"representative\" use cases queries, typically informed by domain experts; but also exploit technical insights from database architects in what features, operations, and data distributions should come together in order to invoke a particularly challenging task1.",
+      "cites_seed_techniques": [
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3471485.3471491",
+              "source_type": "paper_citation_context",
+              "excerpt": "It inspired me to think about novel directions, possibly taking automatic benchmark extraction from performance monitoring accountability also towards correctness testing: one could envision enriching workload summarization with new dimensions such as code coverage [5] and automatic generation of query correctness oracles [4].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Ternary Logic Partitioning (TLP), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8e71aab69a661b20fb5e8ab2c8ee857137e6b2610031c7230ccf7638e17d478f",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/0a54335989389a1e25b4ca13050eeeabd04168e2",
+        "source_type": "paper",
+        "content_sha256": "sha256:d5635cd86ccfa9d961ceb93b8b6af732226d408e8f413c2f21fd5d0d01144001"
+      }
+    },
+    {
+      "id": "paper:arxiv:2105.10016",
+      "title": "Testing DBMS Performance with Mutations",
+      "authors": [
+        "Xinyu Liu",
+        "Qi Zhou",
+        "Joy Arulraj",
+        "A. Orso"
+      ],
+      "year": 2021,
+      "venue": "arXiv (Cornell University)",
+      "doi": null,
+      "arxiv_id": "2105.10016",
+      "s2_paper_id": "4ce512cdd51d4926d9353aca8a588a2414a0d305",
+      "url": "https://arxiv.org/abs/2105.10016",
+      "open_access_pdf": "https://arxiv.org/pdf/2105.10016",
+      "abstract": "Because database systems are the critical component of modern data-intensive applications, it is important to ensure that they operate correctly. To this end, developers extensively test these systems to eliminate bugs that negatively affect functionality. In addition to functional bugs, however, there is another important class of bugs: performance bugs. These bugs negatively affect the response time of a database system and can therefore affect the overall performance of the system. Despite their impact on end-user experience, performance bugs have received considerably less attention than functional bugs. In this paper, we present AMOEBA, a system for automatically detecting performance bugs in database systems. The core idea behind AMOEBA is to construct query pairs that are semantically equivalent to each other and then compare their response time on the same database system. If the queries exhibit a significant difference in their runtime performance, then the root cause is likely a performance bug in the system. We propose a novel set of structure and predicate mutation rules for constructing query pairs that are likely to uncover performance bugs. We introduce feedback mechanisms for improving the efficacy and computational efficiency of the tool. We evaluate AMOEBA on two widely-used DBMSs, namely PostgreSQL and CockroachDB. AMOEBA has discovered 20 previously-unknown performance bugs, among which developers have already confirmed 14 and fixed 4.",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2105.10016",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [35–37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:dc67807bce191aa04950c1f2b72f3059a65f6a65b10e95c91d2287a07d014d52",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2105.10016",
+              "source_type": "paper_citation_context",
+              "excerpt": "They leverage tools such as S QLSMITH [4] and SQLancer [35– 37] to discover crash-inducing or logic bugs in DBMSs.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:79bf4b5468600bae209e4cb025a403b18427dc88f90c86cb2dd146720cd1f29d",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2105.10016",
+              "source_type": "paper_citation_context",
+              "excerpt": "8.6.1 Query Pairs from SQLancer SQLancer is the state-of-art tool for discovering logic bugs [35–37].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:9099596584af4aac1ac54a2623f466842056c3756113805539a6a57441dbbe1c",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "yes",
+          "method": "llm_classification",
+          "techniques": [
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/paper_arxiv_2105_10016.json",
+              "source_type": "paper",
+              "excerpt": "We compare AMOEBA against two other sources of equivalent queries that could be used for detecting performance bugs: (1) a manually-written test suite in a widely-used query optimization framework, and (2) the TLP.",
+              "excerpt_is_verbatim": true,
+              "note": "M5 in the paper's extracted text, page 2.",
+              "content_sha256": "sha256:b48976e88a73a784b94cad188a5c18e9c6421bd15475abd7c1034b8851577f4f",
+              "retrieved_at": "2026-09-10T15:22:32Z",
+              "first_seen": "2026-09-10T15:22:32Z",
+              "last_verified": "2026-09-10T15:22:32Z"
+            }
+          ]
+        },
+        "describes_as_state_of_the_art": {
+          "value": "uncertain",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/4ce512cdd51d4926d9353aca8a588a2414a0d305",
+        "source_type": "paper",
+        "content_sha256": "sha256:44a2b5602047b4afa6aa0e5a26207fd9c873e4efe6563c5a92a2ede8b7839ba3"
+      }
+    },
+    {
+      "id": "paper:doi:10.3929/ethz-b-000507577",
+      "title": "Verifying Serializability Protocols With Version Order Recovery",
+      "authors": [
+        "Jack Clark"
+      ],
+      "year": 2021,
+      "venue": null,
+      "doi": "10.3929/ethz-b-000507577",
+      "arxiv_id": null,
+      "s2_paper_id": "9c8243cc846844f3af89a9ff742ba25947212bc8",
+      "url": "https://doi.org/10.3929/ethz-b-000507577",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.3929/ethz-b-000507577",
+              "source_type": "paper_citation_context",
+              "excerpt": "Automated randomized testing has proven to be an effective method of testing database systems [3, 43, 44, 45, 49].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:b3237697fcf05f94a31b8ba55e6b245f04e9330164b2b567d4624a0252939d74",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.3929/ethz-b-000507577",
+              "source_type": "paper_citation_context",
+              "excerpt": "For example, SQLancer support the PQS [45], TLP [44] and NoREC [43] strategies.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:cb277bcf5062bb1e7772057655373d6f35bfafe3fb87d73852d5c27b887605d9",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.3929/ethz-b-000507577",
+              "source_type": "paper_citation_context",
+              "excerpt": "This is similar to the approach used in pivoted query synthesis [45].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:909ab4f208c2bfe89d51abf7e0104771ad5760a75db62100ecb832458900fd65",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/9c8243cc846844f3af89a9ff742ba25947212bc8",
+        "source_type": "paper",
+        "content_sha256": "sha256:9b8aaee103c5cb72caa0d0c8c65e1f17d43aea10e64d30d0fe4b79a5d03b26a8"
+      },
+      "abstract": null
+    },
+    {
+      "id": "paper:doi:10.1145/3368089.3409710",
+      "title": "Detecting optimization bugs in database engines via non-optimizing reference engine construction",
+      "authors": [
+        "Manuel Rigger",
+        "Zhendong Su"
+      ],
+      "year": 2020,
+      "venue": "ESEC/SIGSOFT FSE",
+      "doi": "10.1145/3368089.3409710",
+      "arxiv_id": "2007.08292",
+      "s2_paper_id": "55a4e4a42cd86fac55491d089d07d04f09dcf957",
+      "url": "https://doi.org/10.1145/3368089.3409710",
+      "open_access_pdf": "https://arxiv.org/pdf/2007.08292",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3368089.3409710",
+              "source_type": "paper_citation_context",
+              "excerpt": "resultsetfora given query.Specifically, we refer to logic bugs in the query optimizer as optimization bugs. Pivoted Query Synthesis (PQS) was recently proposed as a way of tackling logic bugs in DBMS [46]. Its core idea is to verify the DBMS based on a single pivot row, for which a query is generated that is expected to fetch this row. While PQS has been effective in detecting many bugs in widely-used",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d8d06b58a2072ee96529386e127874d65f8ec217fa9208da38d6edcfa9340703",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3368089.3409710",
+              "source_type": "paper_citation_context",
+              "excerpt": "e is provided only for release versions, which are typically published every 2-3 months, which makes it tedious to filter out test cases that trigger the same underlying bug, as also noted previously [46]. Furthermore, only some of the bugs found by PQS have been fixed, providing fewer incentives to test this DBMS. Thus, we decided to test MariaDB, which is a fork of MySQL, and uses an open-source dev",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:c14cb61b8411a7a710d4d72eedc8d61966204246e5e63b49f38b90a30e238b65",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3368089.3409710",
+              "source_type": "paper_citation_context",
+              "excerpt": "s errors in the corner cases described in Section3.4. Thus, we explain our database and query generator only for completeness. In our work, we base the generation of databases and queries on SQLancer [46], which we extended to cover additional DBMS (i.e. CockroachDB and MariaDB), as well as SQL features (e.g.additional ESEC/FSE ’20, November 8–13, 2020, Virtual Event, USA Manuel Rigger and Zhendong Su",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e77b766001d633291f13d9889c92e56fa367d0b0d3e5652b1451b7b3d49b78b1",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3368089.3409710",
+              "source_type": "paper_citation_context",
+              "excerpt": "and the differences between different DBMS were a challenge. Indeed, DBMS typically differ in the SQL dialect that they support, by deviating from the standard and providing DBMS-specific extensions [46]. For example, the CockroachDB developers argued that they cannot use differential testing using PostgreSQL [29], which is the DBMS that is closest to it: Correctness is difficult because we don’t hav",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:5addfa53d94c774e3f77ecc9f304ab8d5cbbb60be2e04ccb96862602d46f1ac5",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3368089.3409710",
+              "source_type": "paper_citation_context",
+              "excerpt": "milar to those that would be obtained when testing MySQL. PostgreSQL is also a popular DBMS; it seems to be more robust than most other DBMS, and the PQS work could find only a single logic bug in it [46]. CockroachDB [56] is a recent commercial NewSQL DBMS [41]. It has received much attention and is highly popular on GitHub, although it has a low rank on the other popularity rankings. We tested only",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:643012894de1dea30c015e6a4fa9207093301c5bd52209f267562d8ffb7ce3a3",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/55a4e4a42cd86fac55491d089d07d04f09dcf957",
+        "source_type": "paper",
+        "content_sha256": "sha256:5d9b798e58514b0d0fda07afcc2a64fec8648eb6d9b86384601e68bbad540fc0"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Database Management Systems (DBMS) are used ubiquitously. To efficiently access data, they apply sophisticated optimizations. Incorrect optimizations can result in logic bugs, which cause a query to compute an incorrect result set. We propose Non-Optimizing Reference Engine Construction (NoREC), a fully-automatic approach to detect optimization bugs in DBMS. Conceptually, this approach aims to evaluate a query by an optimizing and a non-optimizing version of a DBMS, to then detect differences in their returned result set, which would indicate a bug in the DBMS. Obtaining a non-optimizing version of a DBMS is challenging, because DBMS typically provide limited control over optimizations. Our core insight is that a given, potentially randomly-generated optimized query can be rewritten to one that the DBMS cannot optimize. Evaluating this unoptimized query effectively corresponds to a non-optimizing reference engine executing the original query. We evaluated NoREC in an extensive testing campaign on four widely-used DBMS, namely PostgreSQL, MariaDB, SQLite, and CockroachDB. We found 159 previously unknown bugs in the latest versions of these systems, 141 of which have been fixed by the developers. Of these, 51 were optimization bugs, while the remaining were error and crash bugs. Our results suggest that NoREC is effective, general and requires little implementation effort, which makes the technique widely applicable in practice."
+    },
+    {
+      "id": "paper:doi:10.1145/3428279",
+      "title": "Finding bugs in database systems via query partitioning",
+      "authors": [
+        "Manuel Rigger",
+        "Zhendong Su"
+      ],
+      "year": 2020,
+      "venue": "Proc. ACM Program. Lang.",
+      "doi": "10.1145/3428279",
+      "arxiv_id": null,
+      "s2_paper_id": "d69216947188267d5537ee0f0501b2d960ecc457",
+      "url": "https://doi.org/10.1145/3428279",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3428279",
+      "cites_seed_techniques": [
+        "pqs",
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3428279",
+              "source_type": "paper_citation_context",
+              "excerpt": "Specifically, we tried to translate a NoREC test case to a WHERE oracle test case and vice versa, by following a similar methodology as for the comparison of NoREC and PQS [Rigger and Su 2020a].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:eb6597d3ff2499a9da1f2706f97ab73857424c8e2ee507be3bf28185c7a6d7ec",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3428279",
+              "source_type": "paper_citation_context",
+              "excerpt": "In 5 of these cases, comparing the record count was insufficient to detect the bug; also the contents had to be compared, contrary to prior suggestions [Rigger and Su 2020a].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1b853c867b65f715d3ebf3867daa7374489188f105094add60d1f33df8af4e4f",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3428279",
+              "source_type": "paper_citation_context",
+              "excerpt": "We did not compare to PQS, which is complementary to TLP and NoREC, and whose advantages and disadvantages were already studied [Rigger and Su 2020a].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d76b536affbc200efa233f9527f772a21f178e819633de5bff9415401d400552",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3428279",
+              "source_type": "paper_citation_context",
+              "excerpt": "The implementation of our proposed approach is based on SQLancer, in which the PQS and NoREC oracles were also implemented [Rigger and Su 2020a,c].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:8cf2c127f11726886997589b057afaa1eff2c316939a546474a1676a148c4b5c",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3428279",
+              "source_type": "paper_citation_context",
+              "excerpt": "Non-optimizing Reference Engine Construction (NoREC) detects bugs in queries that use a WHERE predicate by rewriting the query to disable the DBMS’ optimizations and addresses PQS’ high implementation effort [Rigger and Su 2020a].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2525a31391313d8d7e16d9078f0ea79c287928bb7180d235c32a24a29e9be3cf",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3428279",
+              "source_type": "paper_citation_context",
+              "excerpt": "For reproducibility, and to facilitate the adoption of TLP, we provide an artifact with the implementation and the bugs we found [Rigger and Su 2020b].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:19df6da66513e9dc95482cbd836b52a30cfae5eb95988e1933f3e9a701427e44",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/d69216947188267d5537ee0f0501b2d960ecc457",
+        "source_type": "paper",
+        "content_sha256": "sha256:5d88765ae290d633a3b0bffb25281d3d0cdb65d31f821e195398ead80913f85a"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Logic bugs in Database Management Systems (DBMSs) are bugs that cause an incorrect result for a given query, for example, by omitting a row that should be fetched. These bugs are critical, since they are likely to go unnoticed by users. We propose Query Partitioning, a general and effective approach for finding logic bugs in DBMSs. The core idea of Query Partitioning is to, starting from a given original query, derive multiple, more complex queries (called partitioning queries), each of which computes a partition of the result. The individual partitions are then composed to compute a result set that must be equivalent to the original query's result set. A bug in the DBMS is detected when these result sets differ. Our intuition is that due to the increased complexity, the partitioning queries are more likely to stress the DBMS and trigger a logic bug than the original query. As a concrete instance of a partitioning strategy, we propose Ternary Logic Partitioning (TLP), which is based on the observation that a boolean predicate p can either evaluate to TRUE, FALSE, or NULL. Accordingly, a query can be decomposed into three partitioning queries, each of which computes its result on rows or intermediate results for which p, NOT p, and p IS NULL hold. This technique is versatile, and can be used to test WHERE, GROUP BY, as well as HAVING clauses, aggregate functions, and DISTINCT queries. As part of an extensive testing campaign, we found 175 bugs in widely-used DBMSs such as MySQL, TiDB, SQLite, and CockroachDB, 125 of which have been fixed. Notably, 77 of these were logic bugs, while the remaining were error and crash bugs. We expect that the effectiveness and wide applicability of Query Partitioning will lead to its broad adoption in practice, and the formulation of additional partitioning strategies."
+    },
+    {
+      "id": "paper:doi:10.1145/3428261",
+      "title": "On the unusual effectiveness of type-aware operator mutations for testing SMT solvers",
+      "authors": [
+        "Dominik Winterer",
+        "Chengyu Zhang",
+        "Zhendong Su"
+      ],
+      "year": 2020,
+      "venue": "Proc. ACM Program. Lang.",
+      "doi": "10.1145/3428261",
+      "arxiv_id": "2004.08799v4",
+      "s2_paper_id": "ab6addb4b50a1af3f78fea402980648135d3b8c9",
+      "url": "https://doi.org/10.1145/3428261",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3428261",
+      "abstract": "We propose type-aware operator mutation, a simple, but unusually effective approach for testing SMT solvers. The key idea is to mutate operators of conforming types within the seed formulas to generate well-typed mutant formulas. These mutant formulas are then used as the test cases for SMT solvers. We realized type-aware operator mutation within the OpFuzz tool and used it to stress-test Z3 and CVC4, two state-of-the-art SMT solvers. Type-aware operator mutations are unusually effective: During one year of extensive testing with OpFuzz, we reported 1092 bugs on Z3’s and CVC4’s respective GitHub issue trackers, out of which 819 unique bugs were confirmed and 685 of the confirmed bugs were fixed by the developers. The detected bugs are highly diverse — we found bugs of many different types (soundness bugs, invalid model bugs, crashes, etc.), logics and solver configurations. We have further conducted an in-depth study of the bugs found by OpFuzz. The study results show that the bugs found by OpFuzz are of high quality. Many of them affect core components of the SMT solvers’ codebases, and some required major changes for the developers to fix. Among the 819 confirmed bugs found by OpFuzz,184 were soundness bugs, the most critical bugs in SMT solvers,and 489 were in the default modes of the solvers. Notably, OpFuzz found 27 critical soundness bugs in CVC4, which has proved to be a very stable SMT solver.",
+      "cites_seed_techniques": [
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/ab6addb4b50a1af3f78fea402980648135d3b8c9",
+              "source_type": "paper",
+              "excerpt": null,
+              "note": "Semantic Scholar records this paper as citing the publication that introduced Non-optimizing Reference Engine Construction (NoREC).",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/ab6addb4b50a1af3f78fea402980648135d3b8c9",
+        "source_type": "paper",
+        "content_sha256": "sha256:bc59bc034a0e3bde36233900580d0aa46f2fd402bed36f4474e78b2ea12fa0c5"
+      }
+    },
+    {
+      "id": "paper:doi:10.1145/3372297.3417260",
+      "title": "SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback",
+      "authors": [
+        "Rui Zhong",
+        "Yongheng Chen",
+        "Hong Hu",
+        "Hangfan Zhang",
+        "Wenke Lee",
+        "Dinghao Wu"
+      ],
+      "year": 2020,
+      "venue": "Conference on Computer and Communications Security",
+      "doi": "10.1145/3372297.3417260",
+      "arxiv_id": "2006.02398",
+      "s2_paper_id": "afe18b71fa5b215a554a2da141f677081fac4503",
+      "url": "https://doi.org/10.1145/3372297.3417260",
+      "open_access_pdf": "https://dl.acm.org/doi/pdf/10.1145/3372297.3417260",
+      "abstract": "Fuzzing is an increasingly popular technique for verifying software functionalities and finding security vulnerabilities. However, current mutation-based fuzzers cannot effectively test database management systems (DBMSs), which strictly check inputs for valid syntax and semantics. Generation-based testing can guarantee the syntax correctness of the inputs, but it does not utilize any feedback, like code coverage, to guide the path exploration. In this paper, we develop Squirrel, a novel fuzzing framework that considers both language validity and coverage feedback to test DBMSs. We design an intermediate representation (IR) to maintain SQL queries in a structural and informative manner. To generate syntactically correct queries, we perform type-based mutations on IR, including statement insertion, deletion and replacement. To mitigate semantic errors, we analyze each IR to identify the logical dependencies between arguments, and generate queries that satisfy these dependencies. We evaluated Squirrel on four popular DBMSs: SQLite, MySQL, PostgreSQL and MariaDB. Squirrel found 51 bugs in SQLite, 7 in MySQL and 5 in MariaDB. 52 of the bugs are fixed with 12 CVEs assigned. In our experiment, Squirrel achieves 2.4×-243.9× higher semantic correctness than state-of-the-art fuzzers, and explores 2.0×-10.9× more new edges than mutation-based tools. These results show that Squirrel is effective in finding memory errors of database management systems.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3372297.3417260",
+              "source_type": "paper_citation_context",
+              "excerpt": "DBMSs have been heavily tested for logic and performance defects [53, 56, 61, 64].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d5ffa94d5f2f79013c2e5b6acdbb56f1d8e2e660a64d790f54e827e1db60f5fe",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            },
+            {
+              "source_url": "https://doi.org/10.1145/3372297.3417260",
+              "source_type": "paper_citation_context",
+              "excerpt": "SQLancer constructs queries to fetch a randomly selected row from a table [53].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:eb86d701a24b4e5c13628aa65162d286a3d0751e1e9fdd5228ff8a871f3ae5d6",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/afe18b71fa5b215a554a2da141f677081fac4503",
+        "source_type": "paper",
+        "content_sha256": "sha256:768fc3f8259f92485516af43a947aa841028ed8293994bbc42a4a44d095b9155"
+      }
+    },
+    {
+      "id": "paper:s2:65a95db3f127442942085400fedd67dc8b61cda4",
+      "title": "Ternary Logic Partitioning: Detecting Logic Bugs in Database Management Systems",
+      "authors": [
+        "Manuel Rigger"
+      ],
+      "year": 2020,
+      "venue": null,
+      "doi": null,
+      "arxiv_id": null,
+      "s2_paper_id": "65a95db3f127442942085400fedd67dc8b61cda4",
+      "url": "https://www.semanticscholar.org/paper/65a95db3f127442942085400fedd67dc8b61cda4",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "pqs",
+        "norec"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs",
+            "norec"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://www.semanticscholar.org/paper/65a95db3f127442942085400fedd67dc8b61cda4",
+              "source_type": "paper_citation_context",
+              "excerpt": "Specifically, we tried to translate a NoREC test case to a WHERE oracle test case and vica versa, by following a similar methodology as for the comparison of NoREC and PQS [Rigger and Su 2020a].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:0c2ec87c442d664463ca53fc3ccc6c21365c92da8952d118eea1a7089d688844",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/65a95db3f127442942085400fedd67dc8b61cda4",
+              "source_type": "paper_citation_context",
+              "excerpt": "In 5 of these cases, comparing the record count was insufficient to detect the bug; also the contents had to be compared, contrary to prior suggestions [Rigger and Su 2020a].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:1b853c867b65f715d3ebf3867daa7374489188f105094add60d1f33df8af4e4f",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/65a95db3f127442942085400fedd67dc8b61cda4",
+              "source_type": "paper_citation_context",
+              "excerpt": "We did not compare to PQS, which is complementary to TLP and NoREC, and whose advantages and disadvantages were already studied [Rigger and Su 2020a].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:d76b536affbc200efa233f9527f772a21f178e819633de5bff9415401d400552",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/65a95db3f127442942085400fedd67dc8b61cda4",
+              "source_type": "paper_citation_context",
+              "excerpt": "Non-optimizing Reference Engine Construction (NoREC) detects bugs in queries that use a WHERE predicate by rewriting the query to disable the DBMS’ optimizations and addresses PQS’ high implementation effort [Rigger and Su 2020a].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:2525a31391313d8d7e16d9078f0ea79c287928bb7180d235c32a24a29e9be3cf",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/65a95db3f127442942085400fedd67dc8b61cda4",
+              "source_type": "paper_citation_context",
+              "excerpt": "For a single bug, SQLancer typically generated many test cases that would trigger it, making it infeasible to filter out such test cases manually, which was also observed by Rigger and Su [2020a,b].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e8d9f4b7a236d0935d3fc82b676b4f320b278630eebc229bed8d6cba3b44305c",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://www.semanticscholar.org/paper/65a95db3f127442942085400fedd67dc8b61cda4",
+              "source_type": "paper_citation_context",
+              "excerpt": "Rigger and Su [2020a] argued for NoREC that coverage information is not insightful, and that they found many bugs in SQLite despite its impressive test suite, which provides 100% MC/DC coverage.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:6ec3676d9eee6fb2d3dc36c2bfb4a81ee5614e6342aea0eb1f0700ace1abc24e",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/65a95db3f127442942085400fedd67dc8b61cda4",
+        "source_type": "paper",
+        "content_sha256": "sha256:0d15b2d2478ff157dc5f7dee91c286a360a81ee5934179a7ceb7c5a17a1f6663"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": null
+    },
+    {
+      "id": "paper:arxiv:2001.04174",
+      "title": "Testing Database Engines via Pivoted Query Synthesis",
+      "authors": [
+        "Manuel Rigger",
+        "Zhendong Su"
+      ],
+      "year": 2020,
+      "venue": "USENIX Symposium on Operating Systems Design and Implementation",
+      "doi": null,
+      "arxiv_id": "2001.04174",
+      "s2_paper_id": "26ca95a72994fdba1c1855eb6b699f98c992b5f4",
+      "url": "https://arxiv.org/abs/2001.04174",
+      "open_access_pdf": null,
+      "cites_seed_techniques": [
+        "norec",
+        "tlp"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "norec",
+            "tlp"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://arxiv.org/abs/2001.04174",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS inspired two followup testing approaches, namely Non-Optimizing Reference Engine Construction (NoREC) [42] and Ternary Logic Partitioning (TLP) [43], both of which were implemented in SQLancer.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:981adafb41304a65bfef3037695dcbbca67dbd7de3de7f14e78313a4fd2ba442",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2001.04174",
+              "source_type": "paper_citation_context",
+              "excerpt": "PQS inspired complementary follow-up work, such as NoREC and TLP, which focus on finding sub-categories of logic bugs [42, 43].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:177ef70602b234ce9479ad510552759a705b78eb54a4fce03a8f32c2d6beff03",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            },
+            {
+              "source_url": "https://arxiv.org/abs/2001.04174",
+              "source_type": "paper_citation_context",
+              "excerpt": "7% of the bugs detected by PQS, which is expected due to its narrower scope [42].",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Non-optimizing Reference Engine Construction (NoREC), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:4c1976a9392c74b0e9ee5cecd9d89d22d1b0e59ee1371dfcc6e4464232ac46d2",
+              "retrieved_at": "2026-09-06T15:09:10Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-06T15:09:10Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "extends_technique": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "compares_with": {
+          "value": "insufficient_evidence",
+          "method": "deterministic"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "deterministic"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-06T15:09:10Z",
+        "source_url": "https://www.semanticscholar.org/paper/26ca95a72994fdba1c1855eb6b699f98c992b5f4",
+        "source_type": "paper",
+        "content_sha256": "sha256:a943c2d75e8b621b23c2f28673dbedfcd062ed0af4e0082ae39e9825950e6d92"
+      },
+      "is_sqlancer_publication": true,
+      "abstract": "Relational databases are used ubiquitously. They are managed by database management systems (DBMS), which allow inserting, modifying, and querying data using a domain-specific language called Structured Query Language (SQL). Popular DBMS have been extensively tested by fuzzers, which have been successful in finding crash bugs. However, approaches to finding logic bugs, such as when a DBMS computes an incorrect result set, have remained mostly untackled. Differential testing is an effective technique to test systems that support a common language by comparing the outputs of these systems. However, this technique is ineffective for DBMS, because each DBMS typically supports its own SQL dialect. To this end, we devised a novel and general approach that we have termed Pivoted Query Synthesis. The core idea of this approach is to automatically generate queries for which we ensure that they fetch a specific, randomly selected row, called the pivot row. If the DBMS fails to fetch the pivot row, the likely cause is a bug in the DBMS. We tested our approach on three widely-used and mature DBMS, namely SQLite, MySQL, and PostgreSQL. In total, we reported 123 bugs in these DBMS, 99 of which have been fixed or verified, demonstrating that the approach is highly effective and general. We expect that the wide applicability and simplicity of our approach will enable the improvement of robustness of many DBMS."
+    },
+    {
+      "id": "paper:doi:10.1145/3395032.3395322",
+      "title": "Testing query execution engines with mutations",
+      "authors": [
+        "Xinyue Chen",
+        "Chenglong Wang",
+        "Alvin Cheung"
+      ],
+      "year": 2020,
+      "venue": "DBTest@SIGMOD",
+      "doi": "10.1145/3395032.3395322",
+      "arxiv_id": null,
+      "s2_paper_id": "a49a96c561c10177764e63817fecf814b04a1739",
+      "url": "https://doi.org/10.1145/3395032.3395322",
+      "open_access_pdf": null,
+      "abstract": "Query optimizer engine plays an important role in modern database systems. However, due to the complex nature of query optimizers, validating the correctness of a query execution engine is inherently challenging. In particular, the high cost of testing query execution engines often prevents developers from making fast iteration during the development process, which can increase the development cycle or lead to production-level bugs. To address this challenge, we propose a tool, MutaSQL, that can quickly discover correctness bugs in SQL execution engines. MutaSQL generates test cases by mutating a query Q over database D into a query Q′ that should evaluate to the same result as Q on D. MutaSQL then checks the execution results of Q′ and Q on the tested engine. We evaluated MutaSQL on previous SQLite versions with known bugs as well as the newest SQLite release. The result shows that MutaSQL can effectively reproduce 34 bugs in previous versions and discover a new bug in the current SQLite release.",
+      "cites_seed_techniques": [
+        "pqs"
+      ],
+      "relationships": {
+        "references": {
+          "value": "yes",
+          "method": "citation_graph",
+          "techniques": [
+            "pqs"
+          ],
+          "evidence": [
+            {
+              "source_url": "https://doi.org/10.1145/3395032.3395322",
+              "source_type": "paper_citation_context",
+              "excerpt": "Meanwhile, pivot query synthesis (PQS) [3] is a new testing algorithm that also utilizes equivalence mutation so that the mutated query contains a specific row in the original output.",
+              "excerpt_is_verbatim": true,
+              "note": "Sentence in this paper citing Pivoted Query Synthesis (PQS), as indexed by Semantic Scholar.",
+              "content_sha256": "sha256:e17a38be8c4ca028a14707b19eb235fa0981b96a7a1ab4592950fd5ce2c9a7c2",
+              "retrieved_at": "2026-09-10T15:15:32Z",
+              "first_seen": "2026-09-06T07:00:18Z",
+              "last_verified": "2026-09-10T15:15:32Z"
+            }
+          ]
+        },
+        "uses_infrastructure": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "extends_technique": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "compares_with": {
+          "value": "no",
+          "method": "llm_classification"
+        },
+        "describes_as_state_of_the_art": {
+          "value": "no",
+          "method": "llm_classification"
+        }
+      },
+      "provenance": {
+        "collector": "papers",
+        "collector_version": "1.0.0",
+        "policy_version": "impact-policy-v1",
+        "first_seen": "2026-09-06T07:00:18Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "source_url": "https://www.semanticscholar.org/paper/a49a96c561c10177764e63817fecf814b04a1739",
+        "source_type": "paper",
+        "content_sha256": "sha256:5e8a03e777d5c32dfc0efac648fdd33de545b82e193c8f9a9139d1a58a99267e"
+      }
+    }
+  ]
+}
diff --git a/_data/impact/people.json b/_data/impact/people.json
new file mode 100644
index 0000000..737f48e
--- /dev/null
+++ b/_data/impact/people.json
@@ -0,0 +1,883 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "generated_at": "2026-09-07T12:06:22Z",
+  "people": [
+    {
+      "id": "person:yuanchengjiang",
+      "name": "YuanchengJiang",
+      "github": "YuanchengJiang",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 513,
+      "role": "Bug reporter on the TEST lab's list",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 446 report(s) to “YuanchengJiang”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        },
+        {
+          "source_url": "https://github.com/php/php-src/issues/15712",
+          "source_type": "github_issue",
+          "note": "This issue, recorded against “YuanchengJiang”, was filed on GitHub by YuanchengJiang.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        },
+        {
+          "source_url": "https://github.com/neo4j/neo4j/issues/12957",
+          "source_type": "github_issue",
+          "note": "This issue, recorded against “Yuancheng Jiang”, was filed on GitHub by YuanchengJiang.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ],
+      "also_recorded_as": [
+        "Yuancheng Jiang"
+      ]
+    },
+    {
+      "id": "person:mrigger",
+      "name": "mrigger",
+      "github": "mrigger",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 503,
+      "role": "Project contributor",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 501 report(s) to “mrigger”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        },
+        {
+          "source_url": "https://github.com/taosdata/TDengine/issues/586",
+          "source_type": "github_issue",
+          "note": "This issue, recorded against “mrigger”, was filed on GitHub by mrigger.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ],
+      "also_recorded_as": [
+        "Manuel Rigger"
+      ]
+    },
+    {
+      "id": "person:suyzhong",
+      "name": "Suyang Zhong",
+      "github": "suyZhong",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 193,
+      "role": "Bug reporter on the TEST lab's list",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 193 report(s) to “Suyang Zhong”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        },
+        {
+          "source_url": "https://github.com/crate/crate/issues/15029",
+          "source_type": "github_issue",
+          "note": "This issue, recorded against “Suyang Zhong”, was filed on GitHub by suyZhong.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:theoristcoder",
+      "name": "Zhaokun Xiang",
+      "github": "TheoristCoder",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 110,
+      "role": "Bug reporter on the TEST lab's list",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 110 report(s) to “Zhaokun Xiang”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        },
+        {
+          "source_url": "https://github.com/pingcap/tidb/issues/60322",
+          "source_type": "github_issue",
+          "note": "This issue, recorded against “Zhaokun Xiang”, was filed on GitHub by TheoristCoder.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:bajinsheng",
+      "name": "bajinsheng",
+      "github": "bajinsheng",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 93,
+      "role": "Bug reporter on the TEST lab's list",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 93 report(s) to “bajinsheng”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        },
+        {
+          "source_url": "https://github.com/duckdb/duckdb/issues/4976",
+          "source_type": "github_issue",
+          "note": "This issue, recorded against “bajinsheng”, was filed on GitHub by bajinsheng.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:derzc",
+      "name": "Chi Zhang",
+      "github": "DerZc",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 75,
+      "role": "Bug reporter on the TEST lab's list",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 75 report(s) to “Chi Zhang”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        },
+        {
+          "source_url": "https://github.com/souffle-lang/souffle/issues/2311",
+          "source_type": "github_issue",
+          "note": "This issue, recorded against “Chi Zhang”, was filed on GitHub by DerZc.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:twilight_shuxin",
+      "name": "ShuxinLi",
+      "github": "Twilight-Shuxin",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 26,
+      "role": "Bug reporter on the TEST lab's list",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 26 report(s) to “ShuxinLi”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        },
+        {
+          "source_url": "https://github.com/BaseXdb/basex/issues/2188",
+          "source_type": "github_issue",
+          "note": "This issue, recorded against “ShuxinLi”, was filed on GitHub by Twilight-Shuxin.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:malwaregarry",
+      "name": "Ming Wei Tan",
+      "github": "malwaregarry",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 16,
+      "role": "Bug reporter on the TEST lab's list",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 16 report(s) to “Ming Wei Tan”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        },
+        {
+          "source_url": "https://github.com/databendlabs/databend/issues/15568",
+          "source_type": "github_issue",
+          "note": "This issue, recorded against “Ming Wei Tan”, was filed on GitHub by malwaregarry.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:axel_teo",
+      "name": "Axel Teo",
+      "github": null,
+      "handle_is_verified": false,
+      "reports_on_lab_list": 4,
+      "role": "Bug reporter on the TEST lab's list",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 4 report(s) to “Axel Teo”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:albert_zhang",
+      "name": "Albert Zhang",
+      "github": null,
+      "handle_is_verified": false,
+      "reports_on_lab_list": 2,
+      "role": "Bug reporter on the TEST lab's list",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 2 report(s) to “Albert Zhang”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:kokrui",
+      "name": "kokrui",
+      "github": "kokrui",
+      "handle_is_verified": false,
+      "reports_on_lab_list": 2,
+      "role": "Bug reporter on the TEST lab's list",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://github.com/nus-test/nus-test.github.io/blob/main/data/bugs.json",
+          "source_type": "website",
+          "note": "The TEST lab's bug list credits 2 report(s) to “kokrui”.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:adoubleb",
+      "name": "adoubleb",
+      "github": "adoubleb",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "adoubleb is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:albertzhangtj",
+      "name": "albertZhangTJ",
+      "github": "albertZhangTJ",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "albertZhangTJ is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:bryanjhc",
+      "name": "bryanjhc",
+      "github": "bryanjhc",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "bryanjhc is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:cuteden_ecnu",
+      "name": "cuteDen-ECNU",
+      "github": "cuteDen-ECNU",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "cuteDen-ECNU is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:cyaoxuan",
+      "name": "cyaoxuan",
+      "github": "cyaoxuan",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "cyaoxuan is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:eltfshr",
+      "name": "eltfshr",
+      "github": "eltfshr",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "eltfshr is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:emilyong",
+      "name": "EmilyOng",
+      "github": "EmilyOng",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "EmilyOng is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:feixiangdejiahao",
+      "name": "feixiangdejiahao",
+      "github": "feixiangdejiahao",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "feixiangdejiahao is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:fidget_spinner",
+      "name": "Fidget-Spinner",
+      "github": "Fidget-Spinner",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "Fidget-Spinner is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:forxenn",
+      "name": "forxenn",
+      "github": "forxenn",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "forxenn is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:hhkit",
+      "name": "hhkit",
+      "github": "hhkit",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "hhkit is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:ishika_nankani",
+      "name": "Ishika-Nankani",
+      "github": "Ishika-Nankani",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "Ishika-Nankani is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:j_os_h_in",
+      "name": "j-os-h-in",
+      "github": "j-os-h-in",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "j-os-h-in is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:jovyntls",
+      "name": "jovyntls",
+      "github": "jovyntls",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "jovyntls is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:joyemang33",
+      "name": "joyemang33",
+      "github": "joyemang33",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "joyemang33 is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:kabilanma",
+      "name": "kabilanma",
+      "github": "kabilanma",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "kabilanma is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:meiye_lj",
+      "name": "Meiye-lj",
+      "github": "Meiye-lj",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "Meiye-lj is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:mzfr",
+      "name": "mzfr",
+      "github": "mzfr",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "mzfr is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:nayameow",
+      "name": "NayameoW",
+      "github": "NayameoW",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "NayameoW is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:ningke_li",
+      "name": "Ningke-Li",
+      "github": "Ningke-Li",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "Ningke-Li is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:nish_d",
+      "name": "nish-d",
+      "github": "nish-d",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "nish-d is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:qiyuzhuang",
+      "name": "QiYuZhuang",
+      "github": "QiYuZhuang",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "QiYuZhuang is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:rajdeepsh",
+      "name": "rajdeepsh",
+      "github": "rajdeepsh",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "rajdeepsh is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:silvaxiang",
+      "name": "SilvaXiang",
+      "github": "SilvaXiang",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "SilvaXiang is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:surizhang",
+      "name": "SuriZhang",
+      "github": "SuriZhang",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "SuriZhang is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:teojunda",
+      "name": "teojunda",
+      "github": "teojunda",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "teojunda is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:wanteatfruit",
+      "name": "wanteatfruit",
+      "github": "wanteatfruit",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "wanteatfruit is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:wkxcass",
+      "name": "wkxcass",
+      "github": "wkxcass",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "wkxcass is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:yewenjia",
+      "name": "YeWenjia",
+      "github": "YeWenjia",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "YeWenjia is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:yibo_dong",
+      "name": "Yibo-Dong",
+      "github": "Yibo-Dong",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "Yibo-Dong is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    },
+    {
+      "id": "person:yyssophie",
+      "name": "yyssophie",
+      "github": "yyssophie",
+      "handle_is_verified": true,
+      "reports_on_lab_list": 0,
+      "role": "TEST lab member",
+      "active": true,
+      "evidence": [
+        {
+          "source_url": "https://nus-test.github.io/people/",
+          "source_type": "website",
+          "note": "yyssophie is linked as a member on the TEST lab's people page.",
+          "retrieved_at": "2026-09-07T12:06:22Z",
+          "first_seen": "2026-09-07T12:06:22Z",
+          "last_verified": "2026-09-07T12:06:22Z"
+        }
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/people_decisions.json b/_data/impact/people_decisions.json
new file mode 100644
index 0000000..ebcc368
--- /dev/null
+++ b/_data/impact/people_decisions.json
@@ -0,0 +1,12 @@
+{
+  "schema_version": "1.0.0",
+  "description": "Rulings about people the roster cannot decide on its own. Being listed on the lab's people page is what the roster can see, and it is not the same question as whether someone belongs to the SQLancer project: a lab lists everyone working in it, including people whose work is not database system testing. Each ruling here was made by a person who knows the answer, and the reason is recorded so it is not re-litigated.",
+  "counts_as_external": [
+    {
+      "github": "joyemang33",
+      "reason": "Listed on the TEST lab's people page, but did not work on relational database system testing at NUS. Their SQLancer bug reports come from their own research rather than from the project's campaigns, which is what 'external' is meant to capture -- the tool travelling beyond the people who built it.",
+      "decided_by": "Manuel Rigger",
+      "decided_on": "2026-09-08"
+    }
+  ]
+}
diff --git a/_data/impact/policy.json b/_data/impact/policy.json
new file mode 100644
index 0000000..040acd4
--- /dev/null
+++ b/_data/impact/policy.json
@@ -0,0 +1,204 @@
+{
+  "schema_version": "1.0.0",
+  "policy_version": "impact-policy-v1",
+  "effective_date": "2026-09-06",
+  "sections": [
+    {
+      "id": "scope",
+      "title": "What this page counts",
+      "body": [
+        "Every number on this page is computed from structured records stored in this website's repository under _data/impact/. Each record carries the primary source that justifies it, so any figure can be traced back to a bug report, a paper, a repository, or a published resource.",
+        "Records are proposed by an automated pipeline that runs weekly, but nothing is published without passing schema validation, and every claim that required a judgement call retains the exact excerpt it was based on. Candidates for which the evidence is inconclusive are rejected rather than rounded up."
+      ],
+      "rules": [
+        {
+          "id": "scope-evidence",
+          "kind": "include",
+          "text": "A record is published only when it carries at least one link to a primary source. Excerpts stored as evidence are copied verbatim from that source; they are never paraphrased or generated."
+        },
+        {
+          "id": "scope-uncertain",
+          "kind": "exclude",
+          "text": "Candidates classified as uncertain or as having insufficient evidence are never promoted into a positive category. They are recorded as rejected and are excluded from all counts."
+        },
+        {
+          "id": "scope-unquotable",
+          "kind": "note",
+          "text": "Two kinds of evidence carry no quotation. A frame captured from a talk shows what was on the slide instead of quoting it, because text read off a picture cannot be checked against a fetched source the way every other excerpt is. And a passage the PDF set with letter-spacing is described rather than printed: extraction returns it one character at a time and the word breaks are not in the file to restore, so no readable form of it exists to quote."
+        }
+      ]
+    },
+    {
+      "id": "umbrella",
+      "title": "The SQLancer umbrella",
+      "body": [
+        "SQLancer is the main testing tool, but the project also covers explicitly designated components such as SQLancer++ and ShQveL. Results produced by any umbrella component count towards the overall SQLancer impact statistics.",
+        "Individual tools are never collapsed into a single unstructured label. Every record preserves both the tool that produced the result and, where known, the specific technique responsible, so the contribution of each component stays visible."
+      ],
+      "rules": [
+        {
+          "id": "umbrella-preserve",
+          "kind": "note",
+          "text": "Each bug record stores a finder (for example sqlancer, sqlancer_pp or shqvel) and a technique (for example norec or tlp, or null when the report does not identify one)."
+        }
+      ]
+    },
+    {
+      "id": "bugs",
+      "title": "Bugs found by SQLancer",
+      "body": [
+        "A bug counts as found by SQLancer when the available evidence connects it to SQLancer or to a testing technique that was introduced as part of SQLancer. Bug reports frequently name the test oracle rather than the tool, so a report that credits NoREC, TLP, PQS or QPG is attributed to SQLancer even when it never uses the word “SQLancer”.",
+        "Reported bugs that the developers rejected as invalid or duplicate remain in the dataset for transparency, but they are excluded from the headline bug count. Only reports accepted as genuine bugs (fixed, verified, or still open) are counted.",
+        "The count is a floor rather than a total. A report reaches this dataset only when something public ties it to SQLancer, and a great deal of the testing leaves no such trace: bugs in closed-source systems are usually reported privately, a report is written about the bug rather than about what found it, and a project can run SQLancer routinely while fixing what it finds without ever naming the tool. Projects known to run SQLancer and contributing no bug at all to this page are the visible part of that gap; the invisible part cannot be sized from here.",
+        "One clause above rests on the reporter rather than on the report, and it is the weakest thing here, so it is labelled rather than blended in: those records carry the rule campaign_reporter and a confidence of low, and they cite the roster entry that admitted them alongside the report itself."
+      ],
+      "rules": [
+        {
+          "id": "bug-explicit",
+          "kind": "include",
+          "text": "A primary source explicitly states that SQLancer, SQLancer++ or ShQveL found the bug."
+        },
+        {
+          "id": "bug-technique",
+          "kind": "include",
+          "text": "The bug report attributes the discovery to a SQLancer-originated test oracle or technique, such as NoREC, TLP, PQS or QPG, even if SQLancer itself is not named."
+        },
+        {
+          "id": "bug-campaign",
+          "kind": "include",
+          "text": "The reporter and the surrounding evidence otherwise establish that the report resulted from a SQLancer testing campaign, for example because it comes from the project's own curated bug repository."
+        },
+        {
+          "id": "bug-campaign-reporter",
+          "kind": "include",
+          "text": "A defect report filed in a database system's own tracker by a member of the project or of the TEST lab counts as a SQLancer bug, whether or not the report names a tool. These people report database bugs because they run SQLancer campaigns, and a report is written about the bug rather than about what found it -- most of them never mention the tool, and reproducers are usually minimised by hand before filing, which removes the generated schema too. Such a record is marked with the rule campaign_reporter at low confidence, and its evidence names the roster it rests on, so every claim resting on who reported it can be found and re-judged."
+        },
+        {
+          "id": "bug-independent",
+          "kind": "exclude",
+          "text": "Bugs found with database-testing techniques developed independently of SQLancer are not counted, even when an implementation of the technique later landed in the SQLancer repository. EET and DQE are the concrete cases: both were introduced elsewhere and contributed to SQLancer afterwards, so bugs attributed to them are not SQLancer findings, and work citing their publications is not counted as citing SQLancer."
+        },
+        {
+          "id": "bug-ambiguous",
+          "kind": "exclude",
+          "text": "A bare mention of an ambiguous acronym such as TLP or PQS is not sufficient. Corroborating database-testing context is required before the match is treated as a SQLancer attribution."
+        },
+        {
+          "id": "bug-graph-systems",
+          "kind": "exclude",
+          "text": "A defect report filed against a graph database system by a member of the project or the lab is not counted. The lab tests graph systems with tools developed independently of SQLancer, and SQLancer has no provider for any of them, so the reporter's name is not evidence of a SQLancer campaign there."
+        }
+      ]
+    },
+    {
+      "id": "papers",
+      "title": "Papers and how they relate to SQLancer",
+      "body": [
+        "Papers are discovered from the citation graphs of the foundational SQLancer publications. Every paper that cites one of them is collected, and each is then classified into four overlapping relationships. A paper can hold several relationships at once, so the overall paper count deduplicates: a paper that both reuses the codebase and compares against SQLancer is counted once.",
+        "A fifth relationship records recognition rather than use: papers that call SQLancer or one of its techniques the state of the art. It is kept out of the “builds on” total on purpose, since describing a tool and building on one are different things."
+      ],
+      "rules": [
+        {
+          "id": "paper-references",
+          "kind": "include",
+          "text": "references — the paper cites a foundational SQLancer publication. Established deterministically from the citation graph, so no judgement is involved."
+        },
+        {
+          "id": "paper-infrastructure",
+          "kind": "include",
+          "text": "uses_infrastructure — concrete evidence that the paper's implementation uses or derives from the SQLancer codebase: an artifact based on the SQLancer repository, retained SQLancer source files or its package structure, SQLancer copyright notices or build coordinates, source history derived from SQLancer, or a README describing the reuse. This category is decided by inspecting the artifact, not from wording in the paper, because a paper that reuses the codebase very often says so nowhere — neither in its text nor in its artifact's README. No single signal is decisive on its own: the presence of Randomly.java, say, is weighed together with the rest."
+        },
+        {
+          "id": "paper-extends",
+          "kind": "include",
+          "text": "extends_technique — the authors extend, generalise, adapt or substantially build upon a technique introduced through SQLancer, for instance by generalising TLP to a new setting or extending NoREC with a new transformation. This is the most qualitative category; each accepted classification retains the exact excerpts that justify it."
+        },
+        {
+          "id": "paper-compares",
+          "kind": "include",
+          "text": "compares_with — the paper empirically evaluates its approach against SQLancer, a SQLancer implementation, or a SQLancer test oracle."
+        },
+        {
+          "id": "paper-sota",
+          "kind": "include",
+          "text": "describes_as_state_of_the_art — the paper describes SQLancer or one of its techniques as the state of the art. This is read directly off the citing sentence, which is quoted in full: a citation context is a single sentence, so one containing both the phrase and a SQLancer name is saying the one about the other. It is recognition rather than reuse, so it is reported separately and is not part of the count of papers building on SQLancer."
+        },
+        {
+          "id": "paper-self",
+          "kind": "exclude",
+          "text": "Publications that are part of the SQLancer project itself are marked as such and excluded from counts of external work building on SQLancer. A paper counts as ours if one of the project's authors is on it, if it is the paper that introduced one of the techniques or tools listed above, or if the project lists it explicitly. Co-authorship settles it on its own, whatever the paper is about: our own follow-up work is not external adoption, and counting it as such would flatter the numbers."
+        }
+      ]
+    },
+    {
+      "id": "adoption",
+      "title": "Database systems using SQLancer",
+      "body": [
+        "Three different things are tracked separately and never conflated: database systems that SQLancer can test, database systems in which SQLancer has found bugs, and database system projects whose own developers use or integrate SQLancer.",
+        "The fact that SQLancer supports a database system says nothing about whether that project uses it. Adoption is recorded only when the evidence comes from the database system's own project or developers.",
+        "A fourth thing is recorded and deliberately counted towards nothing: a project whose developers have proposed adopting SQLancer without anything yet showing the project running it. An intention is not use, so these are kept apart under planned_adoption, listed on their own page, and included in no figure here. A proposal from a project's own developers is worth recording, and it is the thing to re-check later: some become adoption and others do not."
+      ],
+      "rules": [
+        {
+          "id": "adopt-ci",
+          "kind": "include",
+          "text": "official_ci — the project runs SQLancer in its own continuous integration."
+        },
+        {
+          "id": "adopt-testing",
+          "kind": "include",
+          "text": "official_testing — the project maintains SQLancer scripts, configuration or testing documentation outside CI."
+        },
+        {
+          "id": "adopt-dev",
+          "kind": "include",
+          "text": "developer_use — a developer of the project describes using SQLancer, for example in an issue, pull request, blog post or documentation they authored."
+        },
+        {
+          "id": "adopt-integration",
+          "kind": "include",
+          "text": "integration_contributed_by_dbms_team — the SQLancer integration for that system was contributed or is maintained by the database system's team."
+        },
+        {
+          "id": "adopt-support",
+          "kind": "exclude",
+          "text": "Support by SQLancer alone is never adoption evidence, and neither is a third party testing a database system with SQLancer."
+        },
+        {
+          "id": "adoption-planned",
+          "kind": "note",
+          "text": "planned_adoption -- a proposal, a roadmap entry or a testing plan from the project's own developers, with nothing yet showing them running SQLancer. Superseded the moment adoption evidence arrives, so a project that has both is counted as using it and listed only there."
+        }
+      ]
+    },
+    {
+      "id": "resources",
+      "title": "Resources",
+      "body": [
+        "Resources are curated pointers to material other people have made about SQLancer or one of its techniques — talks, blog posts, documentation notes, datasets, artifacts and tools. Two things are deliberately excluded: pages that merely mention SQLancer in passing, and the project's own material. The site already links its own documentation, papers and blog, and listing them here would pad the count without saying anything about reach."
+      ],
+      "rules": [
+        {
+          "id": "resource-canonical",
+          "kind": "include",
+          "text": "Every resource carries a canonical URL, a type from a fixed vocabulary, a short factual description, and the source metadata it was collected from."
+        },
+        {
+          "id": "resource-own",
+          "kind": "exclude",
+          "text": "Material published by the SQLancer project itself is not a resource here, whether it is hosted on the project's own properties or simply named after it."
+        }
+      ]
+    },
+    {
+      "id": "method",
+      "title": "How the data is collected",
+      "body": [
+        "Discovery is deterministic. Bug records come from the project's own curated bug repository, from targeted GitHub searches for SQLancer and its technique names, from the forks that database system vendors maintain of SQLancer itself, which list the bugs each provider found, from vendor trackers that are not GitHub -- MariaDB's Jira, openGauss's Gitee -- and from the NUS TEST lab's published bug list — though nothing from that list is admitted for being on it: each candidate is taken back to its own bug report and put through the same attribution rules as any other. Papers come from scholarly citation APIs. Adoption evidence comes from code, workflow and issue searches scoped to database system repositories.",
+        "One route deserves spelling out, because most bug reports never mention SQLancer at all — a report is written about the bug, not about what found it. Searching for the tool's name therefore misses them. So the people who run SQLancer campaigns are recorded, each linked to their GitHub profile and each link established from issues they are credited with rather than assumed, and their reports are read whether or not they name the tool. Two things can then admit a report. The stronger is the reproducer itself: SQLancer generates its own schema, so its reproducers create tables named t0, t1 with columns c0, c1, a shape the tool leaves on the report even when the reporter says nothing. The weaker is membership alone, for the many reports whose reproducer was minimised by hand before filing and so carries no such trace; those are marked campaign_reporter at low confidence and cite the roster openly, so a reader can see exactly which claims rest on who reported them.",
+        "A language model is used only to answer bounded questions about material that deterministic collection has already found — for example whether an issue attributes a bug to a SQLancer technique, or whether a paper extends one. It is given the fetched source text and must answer yes, no, uncertain, or insufficient evidence, quoting the passage it relied on. It is never asked to browse the web, and it cannot introduce a record on its own.",
+        "Classifications are cached against the hash of the evidence, the policy version and the taxonomy version, so unchanged material is never reclassified. Proposed changes are opened as a pull request for human review rather than published directly."
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/recognition_highlights.json b/_data/impact/recognition_highlights.json
new file mode 100644
index 0000000..f401cec
--- /dev/null
+++ b/_data/impact/recognition_highlights.json
@@ -0,0 +1,51 @@
+{
+  "schema_version": "1.0.0",
+  "description": "The papers whose recognition of SQLancer is stated outright rather than implied by listing it among baselines. Thirty-four papers describe SQLancer or one of its techniques as the state of the art, and most do so in the same sentence that names their comparison set -- which the comparison section already shows. These nine make the claim on its own account, so they are what the impact page quotes; the rest are listed in full on the recognition page. Each quote must appear verbatim in that paper's own evidence, which validation checks.",
+  "highlights": [
+    {
+      "paper_id": "paper:doi:10.1109/icde65706.2026.00240",
+      "mention_id": "M14",
+      "why": "Names SQLancer the most effective black-box fuzzing tool, without qualification and without it being a baseline list."
+    },
+    {
+      "paper_id": "paper:doi:10.14722/ndss.2026.240198",
+      "mention_id": "M7",
+      "why": "A claim about standing in the field rather than about a single comparison: the platform other techniques are implemented on."
+    },
+    {
+      "paper_id": "paper:doi:10.1145/3810991.3811632",
+      "mention_id": "M4",
+      "why": "Distinguishes SQLancer within prior work and credits it with success on mature databases."
+    },
+    {
+      "paper_id": "paper:doi:10.1145/3764583",
+      "mention_id": "M5",
+      "why": "Calls SQLancer the state-of-the-art approach to detecting logic bugs, in the singular."
+    },
+    {
+      "paper_id": "paper:doi:10.1145/3588909",
+      "mention_id": "M15",
+      "why": "Calls SQLancer a current state-of-the-art tool and the work most closely related to its own."
+    },
+    {
+      "paper_id": "paper:arxiv:2510.06663",
+      "mention_id": "M31",
+      "why": "Describes SQLancer as a state-of-the-art testing framework carrying most of the latest oracles."
+    },
+    {
+      "paper_id": "paper:doi:10.1145/3510003.3510093",
+      "mention_id": "M14",
+      "why": "Names a specific technique, TLP, as the state of the art for metamorphic logic-bug detection."
+    },
+    {
+      "paper_id": "paper:doi:10.1109/iaecst68792.2025.11415166",
+      "mention_id": "M4",
+      "why": "A claim of priority rather than of rank: SQLancer pioneered systematic oracle-based bug detection. This paper appears nowhere else on the impact page."
+    },
+    {
+      "paper_id": "paper:doi:10.1016/j.cose.2025.104564",
+      "mention_id": "M12",
+      "why": "Names NoREC one of the most effective logic-bug oracles, from a paper whose own contribution is to work around its limits."
+    }
+  ]
+}
diff --git a/_data/impact/resources.json b/_data/impact/resources.json
new file mode 100644
index 0000000..225f415
--- /dev/null
+++ b/_data/impact/resources.json
@@ -0,0 +1,844 @@
+{
+  "schema_version": "1.0.0",
+  "resources": [
+    {
+      "id": "resource:talk:8c0563f07539",
+      "title": "Reliability Lessons From SQLite - Richard Hipp | SSW 2026",
+      "type": "talk",
+      "url": "https://www.youtube.com/watch?v=V_qzqY1bb7I",
+      "description": "SQLite's creator on twenty-six years of testing it, crediting Rigger with the idea of fuzzing for inconsistencies in SQL rather than for crashes, explaining the sub-query equivalence that finds them, and saying SQLite's own fuzzer had to be extended to do the same.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://www.youtube.com/watch?v=V_qzqY1bb7I&t=2836s",
+          "source_type": "video",
+          "excerpt": "And then a few years later, um, Manual Rigger came up with this idea of we, you know, the the original fuzzers were just looking for memory errors or searching faults or something like that. He came up with the idea we can we can do fuzzing ideas to test for inconsistencies in SQL.",
+          "excerpt_is_verbatim": true,
+          "note": "From the talk's automatic captions, a machine transcription of speech. The link points at 47:16, where a reader can listen to what was actually said.",
+          "retrieved_at": "2026-09-11T16:38:10Z",
+          "first_seen": "2026-09-11T16:38:10Z",
+          "last_verified": "2026-09-11T16:38:10Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-11T02:06:36Z",
+        "last_verified": "2026-09-11T16:38:10Z",
+        "source_url": "https://www.youtube.com/watch?v=V_qzqY1bb7I",
+        "source_type": "video"
+      },
+      "year": 2026,
+      "publisher": "Software Should Work",
+      "authors": [
+        "Richard Hipp"
+      ],
+      "related_dbms": [
+        "sqlite"
+      ]
+    },
+    {
+      "id": "resource:talk:32d47a58b337",
+      "title": "The Art of Database Testing by Alperen Keles | DC Systems 011",
+      "type": "talk",
+      "url": "https://www.youtube.com/watch?v=QRwxHGpWaUA",
+      "description": "A survey of database testing methodology that walks an audience through SQLancer's oracles, naming pivoted query synthesis outright and putting ternary logic partitioning on a slide; the tool itself comes up again in questions.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://www.youtube.com/watch?v=QRwxHGpWaUA&t=205s",
+          "source_type": "video",
+          "excerpt": "So the first one is called pivoted query synthesis and the idea here is containment.",
+          "excerpt_is_verbatim": true,
+          "note": "From the talk's automatic captions, a machine transcription of speech. The link points at 3:25, where a reader can listen to what was actually said.",
+          "retrieved_at": "2026-09-11T15:20:04Z",
+          "first_seen": "2026-09-11T02:06:36Z",
+          "last_verified": "2026-09-11T15:20:04Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-11T02:06:36Z",
+        "last_verified": "2026-09-11T15:20:04Z",
+        "source_url": "https://www.youtube.com/watch?v=QRwxHGpWaUA",
+        "source_type": "video"
+      },
+      "year": 2026,
+      "publisher": "Antithesis",
+      "authors": [
+        "Alperen Keles"
+      ],
+      "related_techniques": [
+        "cert",
+        "coddtest",
+        "dqp",
+        "norec",
+        "pqs",
+        "qpg",
+        "tlp"
+      ]
+    },
+    {
+      "id": "resource:talk:37812041fc41",
+      "title": "Fuzzing databases is difficult",
+      "type": "talk",
+      "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg",
+      "description": "A ClickHouse engineer on how the system is fuzzed. SQLancer appears on the early list of the fuzzers ClickHouse runs, again where the talk explains detecting wrong results by comparing equivalent queries, and once more in the closing recommendations.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg&t=110s",
+          "source_type": "video",
+          "excerpt": "Uh, there are a few here, uh, a few of them probably already know, uh, like, uh, SQL Answer, which was, uh, Pioneer to find the wrong results. Uh, there are others like FAL and WebFuzzer that are known to do",
+          "excerpt_is_verbatim": true,
+          "note": "From the talk's automatic captions, a machine transcription of speech -- it renders the name as \"SQL Answer\". The link points at 1:50, where a reader can listen to what was actually said.",
+          "retrieved_at": "2026-09-11T16:38:10Z",
+          "first_seen": "2026-09-11T15:20:04Z",
+          "last_verified": "2026-09-11T16:38:10Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-11T02:06:36Z",
+        "last_verified": "2026-09-11T16:38:10Z",
+        "source_url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg",
+        "source_type": "video"
+      },
+      "year": 2025,
+      "publisher": "ClickHouse",
+      "authors": [
+        "Pedro Ferreira"
+      ],
+      "related_dbms": [
+        "clickhouse"
+      ]
+    },
+    {
+      "id": "resource:talk:a2c6f09209d4",
+      "title": "FUZZING'25 Keynote: \"Constraining Fuzzing without Paying Too Much\" by Miryung Kim",
+      "type": "talk",
+      "url": "https://www.youtube.com/watch?v=L90MBb6NLBE&t=1703s",
+      "description": "A fuzzing keynote that puts SQLancer in its table of what building a custom fuzzer costs, at 28:23.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://www.youtube.com/watch?v=L90MBb6NLBE&t=1703s",
+          "source_type": "video",
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "The keynote's table of what fuzzer customisations cost to build lists SQLancer among them, with its contributor, commit and paper counts.",
+          "retrieved_at": "2026-09-11T15:20:04Z",
+          "first_seen": "2026-09-11T02:06:36Z",
+          "last_verified": "2026-09-11T15:20:04Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-11T02:06:36Z",
+        "last_verified": "2026-09-11T15:20:04Z",
+        "source_url": "https://www.youtube.com/watch?v=L90MBb6NLBE&t=1703s",
+        "source_type": "video"
+      },
+      "year": 2025,
+      "publisher": "International Fuzzing Workshop",
+      "authors": [
+        "Miryung Kim"
+      ]
+    },
+    {
+      "id": "resource:blog_post:14dd0860862e",
+      "title": "When SQLancer Meets IRIS: What Happens When We Push a Database to Its Limits",
+      "type": "blog_post",
+      "url": "https://community.intersystems.com/post/when-sqlancer-meets-iris-what-happens-when-we-push-database-its-limits",
+      "description": "This is precisely why tools like SQLancer exist.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://community.intersystems.com/post/when-sqlancer-meets-iris-what-happens-when-we-push-database-its-limits",
+          "source_type": "blog_post",
+          "excerpt": "This is precisely why tools like SQLancer exist.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:9c2163b15bf717addca9dadd26018636f5db9f4cd3d75354ac875b28eb659aef",
+          "retrieved_at": "2026-09-11T15:21:33Z",
+          "first_seen": "2026-09-10T16:09:37Z",
+          "last_verified": "2026-09-11T15:21:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-10T16:09:37Z",
+        "last_verified": "2026-09-11T15:21:33Z",
+        "source_url": "https://community.intersystems.com/post/when-sqlancer-meets-iris-what-happens-when-we-push-database-its-limits",
+        "source_type": "website"
+      },
+      "year": 2025,
+      "publisher": "InterSystems Developer Community",
+      "related_techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "resource:talk:95eb6473937b",
+      "title": "[FUZZING'23] \"Three Colours of Fuzzing: Reflections and Open Challenges\" Keynote by Cristian Cadar",
+      "type": "talk",
+      "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0",
+      "description": "A fuzzing keynote that reaches for SQLancer as its example of a fuzzer that found hundreds of bugs in mature database systems, and later reads out SQLite's own line about the project's author, agreeing with it.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://www.youtube.com/watch?v=6YGqFRTe2D0&t=295s",
+          "source_type": "video",
+          "excerpt": "as another example SQL lenser which is a fer for database Management Systems again it has found hundreds of bugs in popular database Management Systems like sqlite and postgress SQL",
+          "excerpt_is_verbatim": true,
+          "note": "From the talk's automatic captions, a machine transcription of speech -- it renders the name as \"SQL lenser\". The link points at 4:55, where a reader can listen to what was actually said.",
+          "retrieved_at": "2026-09-11T02:06:36Z",
+          "first_seen": "2026-09-11T02:06:36Z",
+          "last_verified": "2026-09-11T02:06:36Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-11T02:06:36Z",
+        "last_verified": "2026-09-11T02:06:36Z",
+        "source_url": "https://www.youtube.com/watch?v=6YGqFRTe2D0",
+        "source_type": "video"
+      },
+      "year": 2023,
+      "publisher": "International Fuzzing Workshop",
+      "authors": [
+        "Cristian Cadar"
+      ]
+    },
+    {
+      "id": "resource:talk:9d70cce25a55",
+      "title": "ClickHouse Release 22.3 Webinar",
+      "type": "talk",
+      "url": "https://presentations.clickhouse.com/2022-release-22.3/index.html",
+      "description": "The release webinar's section on continuous integration lists SQLancer among the fuzzing methods ClickHouse runs, beside libFuzzer, the AST query fuzzer and Jepsen.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://presentations.clickhouse.com/2022-release-22.3/index.html",
+          "source_type": "video",
+          "excerpt": "— SQLancer — logical fuzzer.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence from the talk's own slides.",
+          "retrieved_at": "2026-09-11T02:06:36Z",
+          "first_seen": "2026-09-11T02:06:36Z",
+          "last_verified": "2026-09-11T02:06:36Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-11T02:06:36Z",
+        "last_verified": "2026-09-11T02:06:36Z",
+        "source_url": "https://presentations.clickhouse.com/2022-release-22.3/index.html",
+        "source_type": "video"
+      },
+      "year": 2022,
+      "publisher": "ClickHouse",
+      "authors": [
+        "Alexey Milovidov"
+      ],
+      "related_dbms": [
+        "clickhouse"
+      ]
+    },
+    {
+      "id": "resource:blog_post:470c86b1c5cf",
+      "title": "Fuzz testing QuestDB",
+      "type": "blog_post",
+      "url": "https://questdb.com/blog/fuzz-testing-questdb",
+      "description": "On top of that, recently the SQLancer team added QuestDB support to their testing tool and helped us to find a number of issues in our SQL engine.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://questdb.com/blog/fuzz-testing-questdb",
+          "source_type": "blog_post",
+          "excerpt": "On top of that, recently the SQLancer team added QuestDB support to their testing tool and helped us to find a number of issues in our SQL engine.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:b09887153bc4594336b8334c25646c8583fb51fbd4a9ed266f8dcd7c910449a4",
+          "retrieved_at": "2026-09-11T15:21:33Z",
+          "first_seen": "2026-09-06T16:33:38Z",
+          "last_verified": "2026-09-11T15:21:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-06T16:33:38Z",
+        "last_verified": "2026-09-11T15:21:33Z",
+        "source_url": "https://questdb.com/blog/fuzz-testing-questdb",
+        "source_type": "website"
+      },
+      "year": 2022,
+      "publisher": "QuestDB",
+      "related_techniques": [
+        "norec"
+      ],
+      "related_dbms": [
+        "questdb"
+      ]
+    },
+    {
+      "id": "resource:talk:d3b039e2d68c",
+      "title": "Keynote 1: DuckDB Testing - Present and Future",
+      "type": "talk",
+      "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs",
+      "description": "DuckDB's testing, from one of its authors: the keynote credits SQLancer with around eighty bugs that the SQLite and Postgres test suites DuckDB had borrowed all missed, argues for running it beside SQLsmith because each finds what the other does not, and says the robot in DuckDB's CI runs both.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://www.youtube.com/watch?v=BgC79Zt2fPs&t=2154s",
+          "source_type": "video",
+          "excerpt": "thought we had like a pretty robust system right we had like um thousands of tests from various systems like sqlite postgres hundreds of our own tests and then this guy called dr rieger came along and he uh unleashed his creation upon us and started furiously opening bug reports so uh using sql answer manual found around 80 bugs inductive and those were bugs that were not found using the test suites of the other systems right so we ran the sql light tests they did not find these bugs we ran the postgres test they didn't find these bugs and it turns out this kind of thing where database systems are complex surprise surprise and each system has their own",
+          "excerpt_is_verbatim": true,
+          "note": "From the talk's automatic captions, a machine transcription of speech -- it renders the name as \"sql answer\". The link points at 35:54, where a reader can listen to what was actually said.",
+          "retrieved_at": "2026-09-11T17:11:50Z",
+          "first_seen": "2026-09-11T17:11:50Z",
+          "last_verified": "2026-09-11T17:11:50Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-11T17:11:50Z",
+        "last_verified": "2026-09-11T17:11:50Z",
+        "source_url": "https://www.youtube.com/watch?v=BgC79Zt2fPs",
+        "source_type": "video"
+      },
+      "year": 2022,
+      "publisher": "DBTest Workshop",
+      "authors": [
+        "Mark Raasveldt"
+      ],
+      "related_dbms": [
+        "duckdb"
+      ]
+    },
+    {
+      "id": "resource:talk:645a47ac1426",
+      "title": "Fuzzing: Practical approaches in ClickHouse",
+      "type": "talk",
+      "url": "https://presentations.clickhouse.com/2021-cpp-siberia/index.html",
+      "description": "A tour of everything ClickHouse fuzzes with, giving SQLancer a section of its own: who wrote it, who brought it into ClickHouse, and what it does.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://presentations.clickhouse.com/2021-cpp-siberia/index.html",
+          "source_type": "video",
+          "excerpt": "SQLancer — logical SQL fuzzer Developed by Manuel Rigger at ETH Zurich.",
+          "excerpt_is_verbatim": true,
+          "note": "Sentence from the talk's own slides.",
+          "retrieved_at": "2026-09-11T02:06:36Z",
+          "first_seen": "2026-09-11T02:06:36Z",
+          "last_verified": "2026-09-11T02:06:36Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-11T02:06:36Z",
+        "last_verified": "2026-09-11T02:06:36Z",
+        "source_url": "https://presentations.clickhouse.com/2021-cpp-siberia/index.html",
+        "source_type": "video"
+      },
+      "year": 2021,
+      "publisher": "ClickHouse",
+      "authors": [
+        "Alexey Milovidov"
+      ],
+      "related_dbms": [
+        "clickhouse"
+      ]
+    },
+    {
+      "id": "resource:blog_post:9db6ca8f2e86",
+      "title": "Talk at CMU: How Citus distributes PostgreSQL via extension APIs",
+      "type": "blog_post",
+      "url": "https://citusdata.com/blog/2021/04/10/talk-at-cmu-how-citus-distributes-postgresql-via-extension-apis",
+      "description": "And we continue to use SQLancer to discover SQL planner bugs we would not have been able to find ourselves.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://citusdata.com/blog/2021/04/10/talk-at-cmu-how-citus-distributes-postgresql-via-extension-apis",
+          "source_type": "blog_post",
+          "excerpt": "And we continue to use SQLancer to discover SQL planner bugs we would not have been able to find ourselves.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:e9664ae20fe7d3919cb67541d379000c33b6c5e5ee794da7041cc6fe0778a140",
+          "retrieved_at": "2026-09-12T02:25:33Z",
+          "first_seen": "2026-09-12T02:25:33Z",
+          "last_verified": "2026-09-12T02:25:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-12T02:25:33Z",
+        "last_verified": "2026-09-12T02:25:33Z",
+        "source_url": "https://citusdata.com/blog/2021/04/10/talk-at-cmu-how-citus-distributes-postgresql-via-extension-apis",
+        "source_type": "website"
+      },
+      "year": 2021,
+      "publisher": "Citus Data",
+      "related_dbms": [
+        "citus",
+        "postgresql"
+      ]
+    },
+    {
+      "id": "resource:talk:f4d45c187618",
+      "title": "CockroachDB's Query Optimizer (Rebecca Taft, Cockroach Labs)",
+      "type": "talk",
+      "url": "https://www.youtube.com/watch?v=wHo-VtzTHx0",
+      "description": "CockroachDB's optimizer, from an engineer who builds it; asked in the questions about random testing, she separates SQLsmith's crashes from the logical bugs Rigger was after, credits him with a batch of GitHub issues against CockroachDB, and says she is trying to get SQLancer running in their own system.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://www.youtube.com/watch?v=wHo-VtzTHx0&t=3507s",
+          "source_type": "video",
+          "excerpt": "errors like if if it's gonna cause an internal error or crash or something like that lasting correctness but um yeah actually uh manuel rigger uh he's the guy that was doing more kind of logically yeah exactly he he did a bunch of experiments with cockroaches and he opened a bunch of github issues for us which is which is pretty great sql lanza is awesome yeah i'm trying to get trying to get it running in our system um all right so then it's the in general what like what's the complexity of the queries that you're seeing i i understand that like that might be",
+          "excerpt_is_verbatim": true,
+          "note": "From the talk's automatic captions, a machine transcription of speech -- it renders the name as \"sql lanza\". The link points at 58:27, where a reader can listen to what was actually said.",
+          "retrieved_at": "2026-09-11T17:11:50Z",
+          "first_seen": "2026-09-11T17:11:50Z",
+          "last_verified": "2026-09-11T17:11:50Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-11T17:11:50Z",
+        "last_verified": "2026-09-11T17:11:50Z",
+        "source_url": "https://www.youtube.com/watch?v=wHo-VtzTHx0",
+        "source_type": "video"
+      },
+      "year": 2020,
+      "publisher": "CMU Database Group",
+      "authors": [
+        "Rebecca Taft"
+      ],
+      "related_dbms": [
+        "cockroachdb"
+      ]
+    },
+    {
+      "id": "resource:blog_post:1cda5545dea5",
+      "title": "Mining for logic bugs in Citus with SQLancer",
+      "type": "blog_post",
+      "url": "https://citusdata.com/blog/2020/09/04/mining-for-logic-bugs-in-citus-with-sqlancer",
+      "description": "The recently launched open source SQLancer (Synthesized Query Lancer) tool gives you a way to test the validity of a database’s query responses.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://citusdata.com/blog/2020/09/04/mining-for-logic-bugs-in-citus-with-sqlancer",
+          "source_type": "blog_post",
+          "excerpt": "The recently launched open source SQLancer (Synthesized Query Lancer) tool gives you a way to test the validity of a database’s query responses.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:1d32a5c280a8a3255ca1a9772d739f3550077c814f3af19762a46877f17a89db",
+          "retrieved_at": "2026-09-11T15:21:33Z",
+          "first_seen": "2026-09-06T17:10:41Z",
+          "last_verified": "2026-09-11T15:21:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-06T17:10:41Z",
+        "last_verified": "2026-09-11T15:21:33Z",
+        "source_url": "https://citusdata.com/blog/2020/09/04/mining-for-logic-bugs-in-citus-with-sqlancer",
+        "source_type": "website"
+      },
+      "year": 2020,
+      "publisher": "Citus Data",
+      "related_techniques": [
+        "tlp"
+      ],
+      "related_dbms": [
+        "citus",
+        "postgresql"
+      ]
+    },
+    {
+      "id": "resource:blog_post:8e797e7f2d91",
+      "title": "Mining for logic bugs in the Citus extension to Postgres with SQLancer",
+      "type": "blog_post",
+      "url": "https://techcommunity.microsoft.com/blog/adforpostgresql/mining-for-logic-bugs-in-the-citus-extension-to-postgres-with-sqlancer/1634393",
+      "description": "SQLancer is an automated Database Management System (DBMS) testing tool for detecting logic bugs.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://techcommunity.microsoft.com/blog/adforpostgresql/mining-for-logic-bugs-in-the-citus-extension-to-postgres-with-sqlancer/1634393",
+          "source_type": "blog_post",
+          "excerpt": "SQLancer is an automated Database Management System (DBMS) testing tool for detecting logic bugs.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:b469b0f7ec6e0a2cc72fa9eb8ed3747e2393c673f7bc7715f275680e1ab0cda0",
+          "retrieved_at": "2026-09-11T15:21:33Z",
+          "first_seen": "2026-09-06T15:22:14Z",
+          "last_verified": "2026-09-11T15:21:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-06T15:22:14Z",
+        "last_verified": "2026-09-11T15:21:33Z",
+        "source_url": "https://techcommunity.microsoft.com/blog/adforpostgresql/mining-for-logic-bugs-in-the-citus-extension-to-postgres-with-sqlancer/1634393",
+        "source_type": "website"
+      },
+      "year": 2020,
+      "publisher": "Microsoft Tech Community",
+      "related_techniques": [
+        "tlp"
+      ],
+      "related_dbms": [
+        "citus",
+        "postgresql"
+      ]
+    },
+    {
+      "id": "resource:blog_post:c45ad82e6285",
+      "title": "An overview of the Materialize QA process",
+      "type": "blog_post",
+      "url": "https://materialize.com/blog/qa-process-overview",
+      "description": "SQLancer SQLancer is another excellent open source testing tool that we’ve ported to Materialize.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://materialize.com/blog/qa-process-overview",
+          "source_type": "blog_post",
+          "excerpt": "SQLancer SQLancer is another excellent open source testing tool that we’ve ported to Materialize.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:cda6ecf63e68de178f04f628bcfa40eb63c8a6fa026e61c5a76934ac4d4cbc53",
+          "retrieved_at": "2026-09-11T15:21:33Z",
+          "first_seen": "2026-09-06T17:10:41Z",
+          "last_verified": "2026-09-11T15:21:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-06T17:10:41Z",
+        "last_verified": "2026-09-11T15:21:33Z",
+        "source_url": "https://materialize.com/blog/qa-process-overview",
+        "source_type": "website"
+      },
+      "publisher": "Materialize",
+      "related_dbms": [
+        "materialize"
+      ]
+    },
+    {
+      "id": "resource:documentation_note:9fdbc6333a0c",
+      "title": "Apache DataFusion contributor guide: testing",
+      "type": "documentation_note",
+      "url": "https://datafusion.apache.org/contributor-guide/testing.html",
+      "description": "You can run these tests individually using cargo as normal command such as cargo test -p datafusion --test parquet_integration SQL “Fuzz” testing # DataFusion uses the SQLancer for “fuzz” testing: it generates random SQL queries and execute them against DataFusion to find bugs.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://datafusion.apache.org/contributor-guide/testing.html",
+          "source_type": "website",
+          "excerpt": "You can run these tests individually using cargo as normal command such as cargo test -p datafusion --test parquet_integration SQL “Fuzz” testing # DataFusion uses the SQLancer for “fuzz” testing: it generates random SQL queries and execute them against DataFusion to find bugs.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:3e01b097f94babd40394621e9f0428cfd44585d323dbcb5596969d43b09ca4ce",
+          "retrieved_at": "2026-09-11T15:21:33Z",
+          "first_seen": "2026-09-06T17:10:41Z",
+          "last_verified": "2026-09-11T15:21:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-06T17:10:41Z",
+        "last_verified": "2026-09-11T15:21:33Z",
+        "source_url": "https://datafusion.apache.org/contributor-guide/testing.html",
+        "source_type": "website"
+      },
+      "publisher": "Apache DataFusion",
+      "related_dbms": [
+        "datafusion"
+      ]
+    },
+    {
+      "id": "resource:blog_post:122c967f955b",
+      "title": "Faster, more robust, and with more features",
+      "type": "blog_post",
+      "url": "https://monetdb.org/blogs/faster-robuster-features",
+      "description": "SQLancer With the help of SQLancer, an automatic DBMS testing tool, we have been able to identify >100 potential problems in corner cases of the SQL processor.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://monetdb.org/blogs/faster-robuster-features",
+          "source_type": "blog_post",
+          "excerpt": "SQLancer With the help of SQLancer, an automatic DBMS testing tool, we have been able to identify >100 potential problems in corner cases of the SQL processor.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:ad85127fc99349d3120a21c465a8acce3e1dec17e858931eb90869bc8b48ce02",
+          "retrieved_at": "2026-09-11T15:21:33Z",
+          "first_seen": "2026-09-06T17:10:41Z",
+          "last_verified": "2026-09-11T15:21:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-06T17:10:41Z",
+        "last_verified": "2026-09-11T15:21:33Z",
+        "source_url": "https://monetdb.org/blogs/faster-robuster-features",
+        "source_type": "website"
+      },
+      "publisher": "MonetDB",
+      "related_dbms": [
+        "monetdb"
+      ]
+    },
+    {
+      "id": "resource:tool:91c95471b133",
+      "title": "go-sqlancer",
+      "type": "tool",
+      "url": "https://github.com/chaos-mesh/go-sqlancer",
+      "description": "go-sqlancer: re-implementation of some of SQLancer's approaches in Go by PingCAP",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://github.com/sqlancer/sqlancer/blob/main/README.md",
+          "source_type": "documentation",
+          "excerpt": "* [go-sqlancer](https://github.com/chaos-mesh/go-sqlancer): re-implementation of some of SQLancer's approaches in Go by PingCAP",
+          "excerpt_is_verbatim": true,
+          "note": "Line in the SQLancer README that links this resource.",
+          "content_sha256": "sha256:e5f5180fefc22551e27d9fb83cb9337cad8625677ee6185fe8c312715e154042",
+          "retrieved_at": "2026-09-06T15:22:14Z",
+          "first_seen": "2026-09-06T07:08:58Z",
+          "last_verified": "2026-09-06T15:22:14Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-06T07:08:58Z",
+        "last_verified": "2026-09-06T15:22:14Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/README.md",
+        "source_type": "documentation",
+        "content_sha256": "sha256:e5f5180fefc22551e27d9fb83cb9337cad8625677ee6185fe8c312715e154042"
+      }
+    },
+    {
+      "id": "resource:tool:b93f35e501c6",
+      "title": "go-sqlancer (PingCAP-QE)",
+      "type": "tool",
+      "url": "https://github.com/PingCAP-QE/go-sqlancer",
+      "description": "Go-sqlancer Inspired by Manuel Rigger's paper Testing Database Engines via Pivoted Query Synthesis.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://github.com/PingCAP-QE/go-sqlancer",
+          "source_type": "website",
+          "excerpt": "Go-sqlancer Inspired by Manuel Rigger's paper Testing Database Engines via Pivoted Query Synthesis.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:205eab516fed7ec0988f15ab3815e5e54aac2ac2ddcba397daf11e2a9286fbbe",
+          "retrieved_at": "2026-09-12T14:19:10Z",
+          "first_seen": "2026-09-12T14:19:10Z",
+          "last_verified": "2026-09-12T14:19:10Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-12T14:19:10Z",
+        "last_verified": "2026-09-12T14:19:10Z",
+        "source_url": "https://github.com/PingCAP-QE/go-sqlancer",
+        "source_type": "website"
+      },
+      "publisher": "PingCAP",
+      "related_techniques": [
+        "norec",
+        "pqs",
+        "tlp"
+      ],
+      "related_dbms": [
+        "tidb"
+      ]
+    },
+    {
+      "id": "resource:documentation_note:4cba17742d1e",
+      "title": "How SQLite Is Tested",
+      "type": "documentation_note",
+      "url": "https://sqlite.org/testing.html",
+      "description": "One fuzzing researcher of particular note is Manuel Rigger.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://sqlite.org/testing.html",
+          "source_type": "website",
+          "excerpt": "One fuzzing researcher of particular note is Manuel Rigger.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:9ae80fad56d47ed9e06a8e97a0c92b3f14d0e8cc2622084611a3f23857c37934",
+          "retrieved_at": "2026-09-11T15:21:33Z",
+          "first_seen": "2026-09-11T15:21:12Z",
+          "last_verified": "2026-09-11T15:21:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-11T15:21:12Z",
+        "last_verified": "2026-09-11T15:21:33Z",
+        "source_url": "https://sqlite.org/testing.html",
+        "source_type": "website"
+      },
+      "publisher": "SQLite",
+      "related_dbms": [
+        "sqlite"
+      ]
+    },
+    {
+      "id": "resource:blog_post:a87fcf82898c",
+      "title": "How we test YugabyteDB",
+      "type": "blog_post",
+      "url": "https://yugabyte.com/blog/yugabytedb-database-testing",
+      "description": "Additionally, we have had good experiences adapting other existing testing tools, like SQLancer to detect logic bugs, plus SQLsmith for generating extremely complex queries.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://yugabyte.com/blog/yugabytedb-database-testing",
+          "source_type": "blog_post",
+          "excerpt": "Additionally, we have had good experiences adapting other existing testing tools, like SQLancer to detect logic bugs, plus SQLsmith for generating extremely complex queries.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:838ce87cf2377e20b610afc5ef0327e61a4f7ef2ba4fd1b0ae9247d09978e4ff",
+          "retrieved_at": "2026-09-11T15:21:33Z",
+          "first_seen": "2026-09-06T17:10:41Z",
+          "last_verified": "2026-09-11T15:21:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-06T17:10:41Z",
+        "last_verified": "2026-09-11T15:21:33Z",
+        "source_url": "https://yugabyte.com/blog/yugabytedb-database-testing",
+        "source_type": "website"
+      },
+      "publisher": "Yugabyte",
+      "related_dbms": [
+        "yugabytedb"
+      ]
+    },
+    {
+      "id": "resource:tool:5f640346dd43",
+      "title": "sqlancer-iris",
+      "type": "tool",
+      "url": "https://github.com/caretdev/sqlancer-iris",
+      "description": "SQLancer-IRIS Automated SQL Testing for InterSystems IRIS Using Differential Oracles sqlancer-iris is an extension of the SQLancer project that enables automated detection of logical bugs in InterSystems IRIS SQL engine.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://github.com/caretdev/sqlancer-iris",
+          "source_type": "website",
+          "excerpt": "SQLancer-IRIS Automated SQL Testing for InterSystems IRIS Using Differential Oracles sqlancer-iris is an extension of the SQLancer project that enables automated detection of logical bugs in InterSystems IRIS SQL engine.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:15c9b5263663bbe07214adfd914aa2ff8896b92d925f1072bb87b9bc9e6c9442",
+          "retrieved_at": "2026-09-12T14:19:10Z",
+          "first_seen": "2026-09-12T14:19:10Z",
+          "last_verified": "2026-09-12T14:19:10Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-12T14:19:10Z",
+        "last_verified": "2026-09-12T14:19:10Z",
+        "source_url": "https://github.com/caretdev/sqlancer-iris",
+        "source_type": "website"
+      },
+      "publisher": "CaretDev",
+      "related_techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "resource:tool:b03ab2d50abb",
+      "title": "SQLRight",
+      "type": "tool",
+      "url": "https://github.com/PSU-Security-Universe/sqlright",
+      "description": "SQLRight: coverage-guided DBMS fuzzer, also supporting NoREC and TLP",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://github.com/sqlancer/sqlancer/blob/main/README.md",
+          "source_type": "documentation",
+          "excerpt": "* [SQLRight](https://github.com/PSU-Security-Universe/sqlright): coverage-guided DBMS fuzzer, also supporting NoREC and TLP",
+          "excerpt_is_verbatim": true,
+          "note": "Line in the SQLancer README that links this resource.",
+          "content_sha256": "sha256:e5f5180fefc22551e27d9fb83cb9337cad8625677ee6185fe8c312715e154042",
+          "retrieved_at": "2026-09-06T15:22:14Z",
+          "first_seen": "2026-09-06T07:08:58Z",
+          "last_verified": "2026-09-06T15:22:14Z"
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-06T07:08:58Z",
+        "last_verified": "2026-09-06T15:22:14Z",
+        "source_url": "https://github.com/sqlancer/sqlancer/blob/main/README.md",
+        "source_type": "documentation",
+        "content_sha256": "sha256:e5f5180fefc22551e27d9fb83cb9337cad8625677ee6185fe8c312715e154042"
+      },
+      "related_techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "resource:documentation_note:23b0b910f8f0",
+      "title": "Why DuckDB",
+      "type": "documentation_note",
+      "url": "https://duckdb.org/why_duckdb",
+      "description": "Result validation: Manuel Rigger used his excellent SQLancer tool to verify DuckDB result correctness.",
+      "official": false,
+      "evidence": [
+        {
+          "source_url": "https://duckdb.org/why_duckdb",
+          "source_type": "website",
+          "excerpt": "Result validation: Manuel Rigger used his excellent SQLancer tool to verify DuckDB result correctness.",
+          "note": "Sentence on the page describing SQLancer.",
+          "content_sha256": "sha256:13e1683d2a20afadcd93c3bf34406b0693626977a2584a040f1e1d86ec019c53",
+          "retrieved_at": "2026-09-11T15:21:33Z",
+          "first_seen": "2026-09-06T17:10:41Z",
+          "last_verified": "2026-09-11T15:21:33Z",
+          "excerpt_is_verbatim": true
+        }
+      ],
+      "provenance": {
+        "collector": "resources",
+        "collector_version": "1.1.0",
+        "first_seen": "2026-09-06T17:10:41Z",
+        "last_verified": "2026-09-11T15:21:33Z",
+        "source_url": "https://duckdb.org/why_duckdb",
+        "source_type": "website"
+      },
+      "publisher": "DuckDB",
+      "related_dbms": [
+        "duckdb"
+      ]
+    }
+  ]
+}
diff --git a/_data/impact/stats.json b/_data/impact/stats.json
new file mode 100644
index 0000000..0598d36
--- /dev/null
+++ b/_data/impact/stats.json
@@ -0,0 +1,7952 @@
+{
+  "schema_version": "1.0.0",
+  "generated_from": {
+    "bugs": "sha256:edf97241dfdd9c5d465ba02a31a526eea55d9cf4fa088a9d415a06ef90ace3fc",
+    "papers": "sha256:6d20d5dad23aba97a0e0a847ed00c4ec47c72fcdbf16d72697e88ecc35f4234d",
+    "adoption": "sha256:01f268c277cc310417544022fbfc5ffbdf9eda4125b8dac457a83276379a57c8",
+    "resources": "sha256:5eac95eb2215557a5441e8c7ab83e1d9038a22d3f9bfc10094d8bf951cbe49a4",
+    "dbms": "sha256:6aa0aaa07364f525dccf4633efdb25b227a2730bfa71f562c0ca85a963943c42",
+    "techniques": "sha256:d9840d34e609fa57d0f5e79fa00a49790d62a8a4de5593cb355454d4f596224f"
+  },
+  "headline": {
+    "bugs_total": 2038,
+    "bugs_total_rounded": "2,000+",
+    "bugs_found_externally": 541,
+    "dbms_supported": 21,
+    "dbms_with_bugs": 41,
+    "dbms_projects_using_sqlancer": 22,
+    "dbms_using_sqlancer_without_counted_bugs": 6,
+    "dbms_projects_planning_adoption": 9,
+    "papers_building_on_sqlancer": 69,
+    "papers_reusing_or_extending_sqlancer": 30,
+    "papers_comparing_against_sqlancer": 55,
+    "papers_citing_sqlancer": 209,
+    "papers_calling_sqlancer_state_of_the_art": 34
+  },
+  "bugs": {
+    "total": 2038,
+    "total_including_rejected": 2130,
+    "status_known_count": 2007,
+    "by_dbms": [
+      {
+        "key": "duckdb",
+        "label": "DuckDB",
+        "count": 302
+      },
+      {
+        "key": "sqlite",
+        "label": "SQLite",
+        "count": 222
+      },
+      {
+        "key": "cockroachdb",
+        "label": "CockroachDB",
+        "count": 202
+      },
+      {
+        "key": "starrocks",
+        "label": "StarRocks",
+        "count": 183
+      },
+      {
+        "key": "tidb",
+        "label": "TiDB",
+        "count": 172
+      },
+      {
+        "key": "dolt",
+        "label": "Dolt",
+        "count": 167
+      },
+      {
+        "key": "monetdb",
+        "label": "MonetDB",
+        "count": 139
+      },
+      {
+        "key": "cratedb",
+        "label": "CrateDB",
+        "count": 81
+      },
+      {
+        "key": "questdb",
+        "label": "QuestDB",
+        "count": 55
+      },
+      {
+        "key": "umbra",
+        "label": "Umbra",
+        "count": 55
+      },
+      {
+        "key": "datafusion",
+        "label": "Apache DataFusion",
+        "count": 54
+      },
+      {
+        "key": "yugabytedb",
+        "label": "YugabyteDB",
+        "count": 54
+      },
+      {
+        "key": "clickhouse",
+        "label": "ClickHouse",
+        "count": 48
+      },
+      {
+        "key": "mysql",
+        "label": "MySQL",
+        "count": 43
+      },
+      {
+        "key": "h2",
+        "label": "H2",
+        "count": 20
+      },
+      {
+        "key": "doris",
+        "label": "Apache Doris",
+        "count": 19
+      },
+      {
+        "key": "databend",
+        "label": "Databend",
+        "count": 18
+      },
+      {
+        "key": "postgresql",
+        "label": "PostgreSQL",
+        "count": 18
+      },
+      {
+        "key": "turso",
+        "label": "Turso",
+        "count": 18
+      },
+      {
+        "key": "virtuoso",
+        "label": "Virtuoso",
+        "count": 17
+      },
+      {
+        "key": "citus",
+        "label": "Citus",
+        "count": 16
+      },
+      {
+        "key": "stonedb",
+        "label": "StoneDB",
+        "count": 16
+      },
+      {
+        "key": "wadjet",
+        "label": "Wadjet",
+        "count": 16
+      },
+      {
+        "key": "mariadb",
+        "label": "MariaDB",
+        "count": 15
+      },
+      {
+        "key": "oxla",
+        "label": "Oxla",
+        "count": 12
+      },
+      {
+        "key": "tdengine",
+        "label": "TDengine",
+        "count": 12
+      },
+      {
+        "key": "firebird",
+        "label": "Firebird",
+        "count": 10
+      },
+      {
+        "key": "matrixone",
+        "label": "MatrixOne",
+        "count": 9
+      },
+      {
+        "key": "risingwave",
+        "label": "RisingWave",
+        "count": 8
+      },
+      {
+        "key": "bharatdbms",
+        "label": "BharatDBMS",
+        "count": 7
+      },
+      {
+        "key": "cnosdb",
+        "label": "CnosDB",
+        "count": 7
+      },
+      {
+        "key": "cloudberry",
+        "label": "Apache Cloudberry",
+        "count": 6
+      },
+      {
+        "key": "oceanbase",
+        "label": "OceanBase",
+        "count": 4
+      },
+      {
+        "key": "presto",
+        "label": "Presto",
+        "count": 4
+      },
+      {
+        "key": "seekdb",
+        "label": "SeekDB",
+        "count": 3
+      },
+      {
+        "key": "cubrid",
+        "label": "CUBRID",
+        "count": 1
+      },
+      {
+        "key": "falkordb",
+        "label": "FalkorDB",
+        "count": 1
+      },
+      {
+        "key": "hazelcast",
+        "label": "Hazelcast",
+        "count": 1
+      },
+      {
+        "key": "kyzo",
+        "label": "Kyzo",
+        "count": 1
+      },
+      {
+        "key": "sparq",
+        "label": "sparq",
+        "count": 1
+      },
+      {
+        "key": "tikv",
+        "label": "TiKV",
+        "count": 1
+      }
+    ],
+    "by_year": [
+      {
+        "key": "2019",
+        "label": "2019",
+        "count": 221
+      },
+      {
+        "key": "2020",
+        "label": "2020",
+        "count": 302
+      },
+      {
+        "key": "2021",
+        "label": "2021",
+        "count": 22
+      },
+      {
+        "key": "2022",
+        "label": "2022",
+        "count": 224
+      },
+      {
+        "key": "2023",
+        "label": "2023",
+        "count": 271
+      },
+      {
+        "key": "2024",
+        "label": "2024",
+        "count": 193
+      },
+      {
+        "key": "2025",
+        "label": "2025",
+        "count": 264
+      },
+      {
+        "key": "2026",
+        "label": "2026",
+        "count": 411
+      }
+    ],
+    "years": [
+      {
+        "year": 2019,
+        "bugs": 221,
+        "bugs_rejected": 36,
+        "systems": 5,
+        "by_dbms": [
+          {
+            "key": "sqlite",
+            "label": "SQLite",
+            "count": 167
+          },
+          {
+            "key": "mysql",
+            "label": "MySQL",
+            "count": 25
+          },
+          {
+            "key": "postgresql",
+            "label": "PostgreSQL",
+            "count": 18
+          },
+          {
+            "key": "mariadb",
+            "label": "MariaDB",
+            "count": 6
+          },
+          {
+            "key": "tdengine",
+            "label": "TDengine",
+            "count": 5
+          }
+        ],
+        "by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 190
+          },
+          {
+            "key": "fixed_in_documentation",
+            "label": "Fixed in documentation",
+            "count": 11
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 17
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 3
+          }
+        ],
+        "by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 120
+          },
+          {
+            "key": "pqs",
+            "label": "Pivoted Query Synthesis (PQS)",
+            "count": 62
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 39
+          }
+        ],
+        "by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 101
+          },
+          {
+            "key": "error",
+            "label": "Unexpected error",
+            "count": 66
+          },
+          {
+            "key": "crash",
+            "label": "Crash",
+            "count": 49
+          },
+          {
+            "key": "hang",
+            "label": "Hang",
+            "count": 2
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 3
+          }
+        ],
+        "by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 119
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 101
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 1
+          }
+        ],
+        "by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 221
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 218
+          },
+          {
+            "key": "mrigger",
+            "label": "mrigger",
+            "count": 3
+          }
+        ],
+        "first_reported": "2019-05-01",
+        "last_reported": "2019-12-31"
+      },
+      {
+        "year": 2020,
+        "bugs": 302,
+        "bugs_rejected": 7,
+        "systems": 10,
+        "by_dbms": [
+          {
+            "key": "duckdb",
+            "label": "DuckDB",
+            "count": 85
+          },
+          {
+            "key": "cockroachdb",
+            "label": "CockroachDB",
+            "count": 71
+          },
+          {
+            "key": "tidb",
+            "label": "TiDB",
+            "count": 61
+          },
+          {
+            "key": "monetdb",
+            "label": "MonetDB",
+            "count": 41
+          },
+          {
+            "key": "h2",
+            "label": "H2",
+            "count": 18
+          },
+          {
+            "key": "sqlite",
+            "label": "SQLite",
+            "count": 11
+          },
+          {
+            "key": "mysql",
+            "label": "MySQL",
+            "count": 7
+          },
+          {
+            "key": "clickhouse",
+            "label": "ClickHouse",
+            "count": 5
+          },
+          {
+            "key": "citus",
+            "label": "Citus",
+            "count": 2
+          },
+          {
+            "key": "doris",
+            "label": "Apache Doris",
+            "count": 1
+          }
+        ],
+        "by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 241
+          },
+          {
+            "key": "fixed_in_documentation",
+            "label": "Fixed in documentation",
+            "count": 2
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 56
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 3
+          }
+        ],
+        "by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 203
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 85
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 11
+          },
+          {
+            "key": "pqs",
+            "label": "Pivoted Query Synthesis (PQS)",
+            "count": 3
+          }
+        ],
+        "by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 94
+          },
+          {
+            "key": "error",
+            "label": "Unexpected error",
+            "count": 102
+          },
+          {
+            "key": "crash",
+            "label": "Crash",
+            "count": 42
+          },
+          {
+            "key": "hang",
+            "label": "Hang",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 63
+          }
+        ],
+        "by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 145
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 94
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 48
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 8
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 7
+          }
+        ],
+        "by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 251
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 51
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 239
+          },
+          {
+            "key": "monetdb-team",
+            "label": "monetdb-team",
+            "count": 41
+          },
+          {
+            "key": "mrigger",
+            "label": "mrigger",
+            "count": 12
+          },
+          {
+            "key": "qoega",
+            "label": "qoega",
+            "count": 5
+          },
+          {
+            "key": "hannes",
+            "label": "hannes",
+            "count": 1
+          },
+          {
+            "key": "jordanlewis",
+            "label": "jordanlewis",
+            "count": 1
+          },
+          {
+            "key": "nukoyluoglu",
+            "label": "nukoyluoglu",
+            "count": 1
+          },
+          {
+            "key": "onderkalaci",
+            "label": "onderkalaci",
+            "count": 1
+          }
+        ],
+        "first_reported": "2020-01-01",
+        "last_reported": "2020-12-15"
+      },
+      {
+        "year": 2021,
+        "bugs": 22,
+        "bugs_rejected": 2,
+        "systems": 6,
+        "by_dbms": [
+          {
+            "key": "cockroachdb",
+            "label": "CockroachDB",
+            "count": 10
+          },
+          {
+            "key": "clickhouse",
+            "label": "ClickHouse",
+            "count": 4
+          },
+          {
+            "key": "starrocks",
+            "label": "StarRocks",
+            "count": 3
+          },
+          {
+            "key": "yugabytedb",
+            "label": "YugabyteDB",
+            "count": 3
+          },
+          {
+            "key": "duckdb",
+            "label": "DuckDB",
+            "count": 1
+          },
+          {
+            "key": "hazelcast",
+            "label": "Hazelcast",
+            "count": 1
+          }
+        ],
+        "by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 21
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 1
+          }
+        ],
+        "by_technique": [
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 11
+          },
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 11
+          }
+        ],
+        "by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 22
+          }
+        ],
+        "by_attribution_rule": [
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 9
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 8
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 5
+          }
+        ],
+        "by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 22
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "mgartner",
+            "label": "mgartner",
+            "count": 6
+          },
+          {
+            "key": "cockroach-teamcity",
+            "label": "cockroach-teamcity",
+            "count": 3
+          },
+          {
+            "key": "qoega",
+            "label": "qoega",
+            "count": 3
+          },
+          {
+            "key": "qvad",
+            "label": "qvad",
+            "count": 3
+          },
+          {
+            "key": "wanpengfei-git",
+            "label": "wanpengfei-git",
+            "count": 2
+          },
+          {
+            "key": "Fly-Style",
+            "label": "Fly-Style",
+            "count": 1
+          },
+          {
+            "key": "dirtysalt",
+            "label": "dirtysalt",
+            "count": 1
+          },
+          {
+            "key": "hannes",
+            "label": "hannes",
+            "count": 1
+          }
+        ],
+        "first_reported": "2021-01-20",
+        "last_reported": "2021-12-28"
+      },
+      {
+        "year": 2022,
+        "bugs": 224,
+        "bugs_rejected": 29,
+        "systems": 10,
+        "by_dbms": [
+          {
+            "key": "starrocks",
+            "label": "StarRocks",
+            "count": 119
+          },
+          {
+            "key": "cockroachdb",
+            "label": "CockroachDB",
+            "count": 34
+          },
+          {
+            "key": "sqlite",
+            "label": "SQLite",
+            "count": 22
+          },
+          {
+            "key": "yugabytedb",
+            "label": "YugabyteDB",
+            "count": 16
+          },
+          {
+            "key": "matrixone",
+            "label": "MatrixOne",
+            "count": 9
+          },
+          {
+            "key": "tidb",
+            "label": "TiDB",
+            "count": 8
+          },
+          {
+            "key": "duckdb",
+            "label": "DuckDB",
+            "count": 7
+          },
+          {
+            "key": "databend",
+            "label": "Databend",
+            "count": 4
+          },
+          {
+            "key": "questdb",
+            "label": "QuestDB",
+            "count": 3
+          },
+          {
+            "key": "clickhouse",
+            "label": "ClickHouse",
+            "count": 2
+          }
+        ],
+        "by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 204
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 3
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 17
+          }
+        ],
+        "by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 187
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 30
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 7
+          }
+        ],
+        "by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 224
+          }
+        ],
+        "by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 183
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 25
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 13
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 3
+          }
+        ],
+        "by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 41
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 183
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "wanpengfei-git",
+            "label": "wanpengfei-git",
+            "count": 53
+          },
+          {
+            "key": "bajinsheng",
+            "label": "bajinsheng",
+            "count": 38
+          },
+          {
+            "key": "dulong41",
+            "label": "dulong41",
+            "count": 29
+          },
+          {
+            "key": "cockroach-teamcity",
+            "label": "cockroach-teamcity",
+            "count": 24
+          },
+          {
+            "key": "qvad",
+            "label": "qvad",
+            "count": 14
+          },
+          {
+            "key": "yongbingwang",
+            "label": "yongbingwang",
+            "count": 10
+          },
+          {
+            "key": "iamlinjunhong",
+            "label": "iamlinjunhong",
+            "count": 9
+          },
+          {
+            "key": "colorfulu",
+            "label": "colorfulu",
+            "count": 8
+          }
+        ],
+        "first_reported": "2022-01-10",
+        "last_reported": "2022-12-28"
+      },
+      {
+        "year": 2023,
+        "bugs": 271,
+        "bugs_rejected": 5,
+        "systems": 21,
+        "by_dbms": [
+          {
+            "key": "starrocks",
+            "label": "StarRocks",
+            "count": 59
+          },
+          {
+            "key": "questdb",
+            "label": "QuestDB",
+            "count": 44
+          },
+          {
+            "key": "duckdb",
+            "label": "DuckDB",
+            "count": 39
+          },
+          {
+            "key": "dolt",
+            "label": "Dolt",
+            "count": 18
+          },
+          {
+            "key": "umbra",
+            "label": "Umbra",
+            "count": 18
+          },
+          {
+            "key": "cockroachdb",
+            "label": "CockroachDB",
+            "count": 12
+          },
+          {
+            "key": "cratedb",
+            "label": "CrateDB",
+            "count": 12
+          },
+          {
+            "key": "stonedb",
+            "label": "StoneDB",
+            "count": 12
+          },
+          {
+            "key": "tidb",
+            "label": "TiDB",
+            "count": 12
+          },
+          {
+            "key": "doris",
+            "label": "Apache Doris",
+            "count": 8
+          },
+          {
+            "key": "risingwave",
+            "label": "RisingWave",
+            "count": 8
+          },
+          {
+            "key": "tdengine",
+            "label": "TDengine",
+            "count": 7
+          },
+          {
+            "key": "sqlite",
+            "label": "SQLite",
+            "count": 6
+          },
+          {
+            "key": "firebird",
+            "label": "Firebird",
+            "count": 5
+          },
+          {
+            "key": "yugabytedb",
+            "label": "YugabyteDB",
+            "count": 3
+          },
+          {
+            "key": "clickhouse",
+            "label": "ClickHouse",
+            "count": 2
+          },
+          {
+            "key": "cnosdb",
+            "label": "CnosDB",
+            "count": 2
+          },
+          {
+            "key": "citus",
+            "label": "Citus",
+            "count": 1
+          },
+          {
+            "key": "cloudberry",
+            "label": "Apache Cloudberry",
+            "count": 1
+          },
+          {
+            "key": "cubrid",
+            "label": "CUBRID",
+            "count": 1
+          },
+          {
+            "key": "databend",
+            "label": "Databend",
+            "count": 1
+          }
+        ],
+        "by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 237
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 13
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 21
+          }
+        ],
+        "by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 248
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 23
+          }
+        ],
+        "by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 271
+          }
+        ],
+        "by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 164
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 75
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 18
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 12
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 2
+          }
+        ],
+        "by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 189
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 82
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "YuanchengJiang",
+            "label": "YuanchengJiang",
+            "count": 57
+          },
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 50
+          },
+          {
+            "key": "Chi Zhang",
+            "label": "Chi Zhang",
+            "count": 39
+          },
+          {
+            "key": "andyziye",
+            "label": "andyziye",
+            "count": 30
+          },
+          {
+            "key": "satanson",
+            "label": "satanson",
+            "count": 27
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 13
+          },
+          {
+            "key": "zhenglin-charlie-li",
+            "label": "zhenglin-charlie-li",
+            "count": 12
+          },
+          {
+            "key": "DerZc",
+            "label": "DerZc",
+            "count": 7
+          }
+        ],
+        "first_reported": "2023-01-28",
+        "last_reported": "2023-12-30"
+      },
+      {
+        "year": 2024,
+        "bugs": 193,
+        "bugs_rejected": 4,
+        "systems": 18,
+        "by_dbms": [
+          {
+            "key": "datafusion",
+            "label": "Apache DataFusion",
+            "count": 48
+          },
+          {
+            "key": "umbra",
+            "label": "Umbra",
+            "count": 28
+          },
+          {
+            "key": "monetdb",
+            "label": "MonetDB",
+            "count": 26
+          },
+          {
+            "key": "cratedb",
+            "label": "CrateDB",
+            "count": 25
+          },
+          {
+            "key": "dolt",
+            "label": "Dolt",
+            "count": 12
+          },
+          {
+            "key": "virtuoso",
+            "label": "Virtuoso",
+            "count": 11
+          },
+          {
+            "key": "duckdb",
+            "label": "DuckDB",
+            "count": 9
+          },
+          {
+            "key": "cockroachdb",
+            "label": "CockroachDB",
+            "count": 6
+          },
+          {
+            "key": "questdb",
+            "label": "QuestDB",
+            "count": 6
+          },
+          {
+            "key": "databend",
+            "label": "Databend",
+            "count": 5
+          },
+          {
+            "key": "firebird",
+            "label": "Firebird",
+            "count": 4
+          },
+          {
+            "key": "doris",
+            "label": "Apache Doris",
+            "count": 3
+          },
+          {
+            "key": "clickhouse",
+            "label": "ClickHouse",
+            "count": 2
+          },
+          {
+            "key": "cloudberry",
+            "label": "Apache Cloudberry",
+            "count": 2
+          },
+          {
+            "key": "cnosdb",
+            "label": "CnosDB",
+            "count": 2
+          },
+          {
+            "key": "h2",
+            "label": "H2",
+            "count": 2
+          },
+          {
+            "key": "tidb",
+            "label": "TiDB",
+            "count": 1
+          },
+          {
+            "key": "yugabytedb",
+            "label": "YugabyteDB",
+            "count": 1
+          }
+        ],
+        "by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 181
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 3
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 9
+          }
+        ],
+        "by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 172
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 11
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 10
+          }
+        ],
+        "by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 193
+          }
+        ],
+        "by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 93
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 52
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 29
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 14
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 5
+          }
+        ],
+        "by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 136
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 57
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 87
+          },
+          {
+            "key": "2010YOUY01",
+            "label": "2010YOUY01",
+            "count": 47
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 33
+          },
+          {
+            "key": "Ming Wei Tan",
+            "label": "Ming Wei Tan",
+            "count": 5
+          },
+          {
+            "key": "malwaregarry",
+            "label": "malwaregarry",
+            "count": 5
+          },
+          {
+            "key": "YuanchengJiang",
+            "label": "YuanchengJiang",
+            "count": 4
+          },
+          {
+            "key": "cockroach-teamcity",
+            "label": "cockroach-teamcity",
+            "count": 3
+          },
+          {
+            "key": "Benxiaohai001",
+            "label": "Benxiaohai001",
+            "count": 2
+          }
+        ],
+        "first_reported": "2024-01-01",
+        "last_reported": "2024-12-28"
+      },
+      {
+        "year": 2025,
+        "bugs": 264,
+        "bugs_rejected": 5,
+        "systems": 16,
+        "by_dbms": [
+          {
+            "key": "duckdb",
+            "label": "DuckDB",
+            "count": 52
+          },
+          {
+            "key": "tidb",
+            "label": "TiDB",
+            "count": 42
+          },
+          {
+            "key": "cockroachdb",
+            "label": "CockroachDB",
+            "count": 37
+          },
+          {
+            "key": "monetdb",
+            "label": "MonetDB",
+            "count": 32
+          },
+          {
+            "key": "yugabytedb",
+            "label": "YugabyteDB",
+            "count": 26
+          },
+          {
+            "key": "dolt",
+            "label": "Dolt",
+            "count": 24
+          },
+          {
+            "key": "sqlite",
+            "label": "SQLite",
+            "count": 12
+          },
+          {
+            "key": "umbra",
+            "label": "Umbra",
+            "count": 9
+          },
+          {
+            "key": "cratedb",
+            "label": "CrateDB",
+            "count": 8
+          },
+          {
+            "key": "datafusion",
+            "label": "Apache DataFusion",
+            "count": 6
+          },
+          {
+            "key": "bharatdbms",
+            "label": "BharatDBMS",
+            "count": 5
+          },
+          {
+            "key": "oceanbase",
+            "label": "OceanBase",
+            "count": 4
+          },
+          {
+            "key": "clickhouse",
+            "label": "ClickHouse",
+            "count": 3
+          },
+          {
+            "key": "starrocks",
+            "label": "StarRocks",
+            "count": 2
+          },
+          {
+            "key": "citus",
+            "label": "Citus",
+            "count": 1
+          },
+          {
+            "key": "tikv",
+            "label": "TiKV",
+            "count": 1
+          }
+        ],
+        "by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 201
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 4
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 59
+          }
+        ],
+        "by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 255
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 8
+          },
+          {
+            "key": "qpg",
+            "label": "Query Plan Guidance (QPG)",
+            "count": 1
+          }
+        ],
+        "by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 264
+          }
+        ],
+        "by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 204
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 34
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 9
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 9
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 8
+          }
+        ],
+        "by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 220
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 44
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Zhaokun Xiang",
+            "label": "Zhaokun Xiang",
+            "count": 96
+          },
+          {
+            "key": "DerZc",
+            "label": "DerZc",
+            "count": 52
+          },
+          {
+            "key": "bajinsheng",
+            "label": "bajinsheng",
+            "count": 22
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 20
+          },
+          {
+            "key": "EmilyOng",
+            "label": "EmilyOng",
+            "count": 17
+          },
+          {
+            "key": "TheoristCoder",
+            "label": "TheoristCoder",
+            "count": 13
+          },
+          {
+            "key": "qvad",
+            "label": "qvad",
+            "count": 13
+          },
+          {
+            "key": "cockroach-teamcity",
+            "label": "cockroach-teamcity",
+            "count": 8
+          }
+        ],
+        "first_reported": "2025-01-05",
+        "last_reported": "2025-12-29"
+      },
+      {
+        "year": 2026,
+        "bugs": 411,
+        "bugs_rejected": 3,
+        "systems": 23,
+        "by_dbms": [
+          {
+            "key": "dolt",
+            "label": "Dolt",
+            "count": 113
+          },
+          {
+            "key": "duckdb",
+            "label": "DuckDB",
+            "count": 103
+          },
+          {
+            "key": "tidb",
+            "label": "TiDB",
+            "count": 36
+          },
+          {
+            "key": "clickhouse",
+            "label": "ClickHouse",
+            "count": 29
+          },
+          {
+            "key": "cratedb",
+            "label": "CrateDB",
+            "count": 25
+          },
+          {
+            "key": "monetdb",
+            "label": "MonetDB",
+            "count": 19
+          },
+          {
+            "key": "turso",
+            "label": "Turso",
+            "count": 18
+          },
+          {
+            "key": "wadjet",
+            "label": "Wadjet",
+            "count": 16
+          },
+          {
+            "key": "cockroachdb",
+            "label": "CockroachDB",
+            "count": 12
+          },
+          {
+            "key": "databend",
+            "label": "Databend",
+            "count": 8
+          },
+          {
+            "key": "virtuoso",
+            "label": "Virtuoso",
+            "count": 6
+          },
+          {
+            "key": "citus",
+            "label": "Citus",
+            "count": 4
+          },
+          {
+            "key": "sqlite",
+            "label": "SQLite",
+            "count": 4
+          },
+          {
+            "key": "cloudberry",
+            "label": "Apache Cloudberry",
+            "count": 3
+          },
+          {
+            "key": "seekdb",
+            "label": "SeekDB",
+            "count": 3
+          },
+          {
+            "key": "yugabytedb",
+            "label": "YugabyteDB",
+            "count": 3
+          },
+          {
+            "key": "bharatdbms",
+            "label": "BharatDBMS",
+            "count": 2
+          },
+          {
+            "key": "questdb",
+            "label": "QuestDB",
+            "count": 2
+          },
+          {
+            "key": "cnosdb",
+            "label": "CnosDB",
+            "count": 1
+          },
+          {
+            "key": "falkordb",
+            "label": "FalkorDB",
+            "count": 1
+          },
+          {
+            "key": "kyzo",
+            "label": "Kyzo",
+            "count": 1
+          },
+          {
+            "key": "mariadb",
+            "label": "MariaDB",
+            "count": 1
+          },
+          {
+            "key": "sparq",
+            "label": "sparq",
+            "count": 1
+          }
+        ],
+        "by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 279
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 132
+          }
+        ],
+        "by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 364
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 32
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 9
+          },
+          {
+            "key": "dqp",
+            "label": "Differential Query Plans (DQP)",
+            "count": 3
+          },
+          {
+            "key": "pqs",
+            "label": "Pivoted Query Synthesis (PQS)",
+            "count": 2
+          },
+          {
+            "key": "coddtest",
+            "label": "Constant-Optimization-Driven Testing (CODDTest)",
+            "count": 1
+          }
+        ],
+        "by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 411
+          }
+        ],
+        "by_attribution_rule": [
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 251
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 81
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 59
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 20
+          }
+        ],
+        "by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 309
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 102
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Yibo-Dong",
+            "label": "Yibo-Dong",
+            "count": 162
+          },
+          {
+            "key": "DerZc",
+            "label": "DerZc",
+            "count": 65
+          },
+          {
+            "key": "wanteatfruit",
+            "label": "wanteatfruit",
+            "count": 50
+          },
+          {
+            "key": "LeMikaelF",
+            "label": "LeMikaelF",
+            "count": 16
+          },
+          {
+            "key": "derekmwright",
+            "label": "derekmwright",
+            "count": 16
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 13
+          },
+          {
+            "key": "fm4v",
+            "label": "fm4v",
+            "count": 12
+          },
+          {
+            "key": "Manuel-Neuer1",
+            "label": "Manuel-Neuer1",
+            "count": 9
+          }
+        ],
+        "first_reported": "2026-01-03",
+        "last_reported": "2026-11-01"
+      }
+    ],
+    "by_status": [
+      {
+        "key": "fixed",
+        "label": "Fixed",
+        "count": 1624
+      },
+      {
+        "key": "fixed_in_documentation",
+        "label": "Fixed in documentation",
+        "count": 13
+      },
+      {
+        "key": "verified",
+        "label": "Confirmed",
+        "count": 102
+      },
+      {
+        "key": "open",
+        "label": "Open",
+        "count": 268
+      },
+      {
+        "key": "unknown",
+        "label": "Unknown",
+        "count": 31
+      }
+    ],
+    "by_technique": [
+      {
+        "key": "unattributed",
+        "label": "Technique not recorded",
+        "count": 1690
+      },
+      {
+        "key": "tlp",
+        "label": "Ternary Logic Partitioning (TLP)",
+        "count": 199
+      },
+      {
+        "key": "norec",
+        "label": "Non-optimizing Reference Engine Construction (NoREC)",
+        "count": 77
+      },
+      {
+        "key": "pqs",
+        "label": "Pivoted Query Synthesis (PQS)",
+        "count": 67
+      },
+      {
+        "key": "dqp",
+        "label": "Differential Query Plans (DQP)",
+        "count": 3
+      },
+      {
+        "key": "coddtest",
+        "label": "Constant-Optimization-Driven Testing (CODDTest)",
+        "count": 1
+      },
+      {
+        "key": "qpg",
+        "label": "Query Plan Guidance (QPG)",
+        "count": 1
+      }
+    ],
+    "by_finder": [
+      {
+        "key": "sqlancer",
+        "label": "SQLancer",
+        "count": 2037
+      },
+      {
+        "key": "sqlancer_pp",
+        "label": "SQLancer++",
+        "count": 1
+      }
+    ],
+    "by_symptom": [
+      {
+        "key": "logic",
+        "label": "Logic bug",
+        "count": 195
+      },
+      {
+        "key": "error",
+        "label": "Unexpected error",
+        "count": 168
+      },
+      {
+        "key": "crash",
+        "label": "Crash",
+        "count": 91
+      },
+      {
+        "key": "hang",
+        "label": "Hang",
+        "count": 3
+      },
+      {
+        "key": "unknown",
+        "label": "Unclassified",
+        "count": 1581
+      }
+    ],
+    "by_attribution_rule": [
+      {
+        "key": "explicit_tool_statement",
+        "label": "The report names SQLancer",
+        "count": 162
+      },
+      {
+        "key": "technique_attribution",
+        "label": "The report names a SQLancer oracle",
+        "count": 263
+      },
+      {
+        "key": "curated_primary_source",
+        "label": "Listed in the project's own bug repository",
+        "count": 409
+      },
+      {
+        "key": "campaign_evidence",
+        "label": "The reproducer carries SQLancer's generated schema",
+        "count": 819
+      },
+      {
+        "key": "campaign_reporter",
+        "label": "Filed by someone who runs SQLancer campaigns",
+        "count": 385
+      }
+    ],
+    "by_reporter_affiliation": [
+      {
+        "key": "project",
+        "label": "Found by the SQLancer project",
+        "count": 1408
+      },
+      {
+        "key": "external",
+        "label": "Found by someone outside the project",
+        "count": 541
+      },
+      {
+        "key": "unknown",
+        "label": "Reporter not recorded",
+        "count": 89
+      }
+    ],
+    "by_dbms_and_affiliation": [
+      {
+        "key": "duckdb",
+        "label": "DuckDB",
+        "count": 302,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 286
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 15
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 1
+          }
+        ]
+      },
+      {
+        "key": "sqlite",
+        "label": "SQLite",
+        "count": 222,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 222
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "cockroachdb",
+        "label": "CockroachDB",
+        "count": 202,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 127
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 56
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 19
+          }
+        ]
+      },
+      {
+        "key": "starrocks",
+        "label": "StarRocks",
+        "count": 183,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 183
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "tidb",
+        "label": "TiDB",
+        "count": 172,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 154
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 7
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 11
+          }
+        ]
+      },
+      {
+        "key": "dolt",
+        "label": "Dolt",
+        "count": 167,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 167
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "monetdb",
+        "label": "MonetDB",
+        "count": 139,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 97
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 42
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "cratedb",
+        "label": "CrateDB",
+        "count": 81,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 81
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "questdb",
+        "label": "QuestDB",
+        "count": 55,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 53
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 2
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "umbra",
+        "label": "Umbra",
+        "count": 55,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 55
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "datafusion",
+        "label": "Apache DataFusion",
+        "count": 54,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 54
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "yugabytedb",
+        "label": "YugabyteDB",
+        "count": 54,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 13
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 39
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 2
+          }
+        ]
+      },
+      {
+        "key": "clickhouse",
+        "label": "ClickHouse",
+        "count": 48,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 7
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 41
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "mysql",
+        "label": "MySQL",
+        "count": 43,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 32
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 11
+          }
+        ]
+      },
+      {
+        "key": "h2",
+        "label": "H2",
+        "count": 20,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 20
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "doris",
+        "label": "Apache Doris",
+        "count": 19,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 3
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 9
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 7
+          }
+        ]
+      },
+      {
+        "key": "databend",
+        "label": "Databend",
+        "count": 18,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 7
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 11
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "postgresql",
+        "label": "PostgreSQL",
+        "count": 18,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 18
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "turso",
+        "label": "Turso",
+        "count": 18,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 18
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "virtuoso",
+        "label": "Virtuoso",
+        "count": 17,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 11
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 6
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "citus",
+        "label": "Citus",
+        "count": 16,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 1
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 7
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 8
+          }
+        ]
+      },
+      {
+        "key": "stonedb",
+        "label": "StoneDB",
+        "count": 16,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 12
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 4
+          }
+        ]
+      },
+      {
+        "key": "wadjet",
+        "label": "Wadjet",
+        "count": 16,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 16
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "mariadb",
+        "label": "MariaDB",
+        "count": 15,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 6
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 8
+          }
+        ]
+      },
+      {
+        "key": "oxla",
+        "label": "Oxla",
+        "count": 12,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 12
+          }
+        ]
+      },
+      {
+        "key": "tdengine",
+        "label": "TDengine",
+        "count": 12,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 12
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "firebird",
+        "label": "Firebird",
+        "count": 10,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 10
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "matrixone",
+        "label": "MatrixOne",
+        "count": 9,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 9
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "risingwave",
+        "label": "RisingWave",
+        "count": 8,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 8
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "bharatdbms",
+        "label": "BharatDBMS",
+        "count": 7,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 7
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "cnosdb",
+        "label": "CnosDB",
+        "count": 7,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 1
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 4
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 2
+          }
+        ]
+      },
+      {
+        "key": "cloudberry",
+        "label": "Apache Cloudberry",
+        "count": 6,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 6
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "oceanbase",
+        "label": "OceanBase",
+        "count": 4,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 4
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "presto",
+        "label": "Presto",
+        "count": 4,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 4
+          }
+        ]
+      },
+      {
+        "key": "seekdb",
+        "label": "SeekDB",
+        "count": 3,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 3
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "cubrid",
+        "label": "CUBRID",
+        "count": 1,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 1
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 0
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "falkordb",
+        "label": "FalkorDB",
+        "count": 1,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "hazelcast",
+        "label": "Hazelcast",
+        "count": 1,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "kyzo",
+        "label": "Kyzo",
+        "count": 1,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "sparq",
+        "label": "sparq",
+        "count": 1,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      },
+      {
+        "key": "tikv",
+        "label": "TiKV",
+        "count": 1,
+        "segments": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 0
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 0
+          }
+        ]
+      }
+    ]
+  },
+  "papers": {
+    "total": 237,
+    "external_total": 209,
+    "building_on_total": 69,
+    "by_relationship": [
+      {
+        "key": "references",
+        "label": "Cites SQLancer",
+        "count": 209
+      },
+      {
+        "key": "reusing_or_extending",
+        "label": "Uses or extends SQLancer",
+        "count": 30
+      },
+      {
+        "key": "compares_with",
+        "label": "Compares against SQLancer",
+        "count": 55
+      },
+      {
+        "key": "describes_as_state_of_the_art",
+        "label": "Describes SQLancer as state of the art",
+        "count": 34
+      }
+    ],
+    "by_year": [
+      {
+        "key": "2020",
+        "label": "2020",
+        "count": 3
+      },
+      {
+        "key": "2021",
+        "label": "2021",
+        "count": 17
+      },
+      {
+        "key": "2022",
+        "label": "2022",
+        "count": 14
+      },
+      {
+        "key": "2023",
+        "label": "2023",
+        "count": 31
+      },
+      {
+        "key": "2024",
+        "label": "2024",
+        "count": 33
+      },
+      {
+        "key": "2025",
+        "label": "2025",
+        "count": 65
+      },
+      {
+        "key": "2026",
+        "label": "2026",
+        "count": 46
+      }
+    ],
+    "by_relationship_year": {
+      "references": [
+        {
+          "key": "2020",
+          "label": "2020",
+          "count": 3
+        },
+        {
+          "key": "2021",
+          "label": "2021",
+          "count": 17
+        },
+        {
+          "key": "2022",
+          "label": "2022",
+          "count": 14
+        },
+        {
+          "key": "2023",
+          "label": "2023",
+          "count": 31
+        },
+        {
+          "key": "2024",
+          "label": "2024",
+          "count": 33
+        },
+        {
+          "key": "2025",
+          "label": "2025",
+          "count": 65
+        },
+        {
+          "key": "2026",
+          "label": "2026",
+          "count": 46
+        }
+      ],
+      "reusing_or_extending": [
+        {
+          "key": "2020",
+          "label": "2020",
+          "count": 0
+        },
+        {
+          "key": "2021",
+          "label": "2021",
+          "count": 0
+        },
+        {
+          "key": "2022",
+          "label": "2022",
+          "count": 5
+        },
+        {
+          "key": "2023",
+          "label": "2023",
+          "count": 5
+        },
+        {
+          "key": "2024",
+          "label": "2024",
+          "count": 5
+        },
+        {
+          "key": "2025",
+          "label": "2025",
+          "count": 10
+        },
+        {
+          "key": "2026",
+          "label": "2026",
+          "count": 5
+        }
+      ],
+      "compares_with": [
+        {
+          "key": "2020",
+          "label": "2020",
+          "count": 0
+        },
+        {
+          "key": "2021",
+          "label": "2021",
+          "count": 2
+        },
+        {
+          "key": "2022",
+          "label": "2022",
+          "count": 4
+        },
+        {
+          "key": "2023",
+          "label": "2023",
+          "count": 5
+        },
+        {
+          "key": "2024",
+          "label": "2024",
+          "count": 11
+        },
+        {
+          "key": "2025",
+          "label": "2025",
+          "count": 19
+        },
+        {
+          "key": "2026",
+          "label": "2026",
+          "count": 14
+        }
+      ],
+      "describes_as_state_of_the_art": [
+        {
+          "key": "2020",
+          "label": "2020",
+          "count": 0
+        },
+        {
+          "key": "2021",
+          "label": "2021",
+          "count": 0
+        },
+        {
+          "key": "2022",
+          "label": "2022",
+          "count": 3
+        },
+        {
+          "key": "2023",
+          "label": "2023",
+          "count": 3
+        },
+        {
+          "key": "2024",
+          "label": "2024",
+          "count": 5
+        },
+        {
+          "key": "2025",
+          "label": "2025",
+          "count": 13
+        },
+        {
+          "key": "2026",
+          "label": "2026",
+          "count": 10
+        }
+      ],
+      "uses_infrastructure": [
+        {
+          "key": "2020",
+          "label": "2020",
+          "count": 0
+        },
+        {
+          "key": "2021",
+          "label": "2021",
+          "count": 0
+        },
+        {
+          "key": "2022",
+          "label": "2022",
+          "count": 4
+        },
+        {
+          "key": "2023",
+          "label": "2023",
+          "count": 4
+        },
+        {
+          "key": "2024",
+          "label": "2024",
+          "count": 5
+        },
+        {
+          "key": "2025",
+          "label": "2025",
+          "count": 9
+        },
+        {
+          "key": "2026",
+          "label": "2026",
+          "count": 4
+        }
+      ],
+      "extends_technique": [
+        {
+          "key": "2020",
+          "label": "2020",
+          "count": 0
+        },
+        {
+          "key": "2021",
+          "label": "2021",
+          "count": 0
+        },
+        {
+          "key": "2022",
+          "label": "2022",
+          "count": 2
+        },
+        {
+          "key": "2023",
+          "label": "2023",
+          "count": 2
+        },
+        {
+          "key": "2024",
+          "label": "2024",
+          "count": 0
+        },
+        {
+          "key": "2025",
+          "label": "2025",
+          "count": 3
+        },
+        {
+          "key": "2026",
+          "label": "2026",
+          "count": 1
+        }
+      ],
+      "building_on": [
+        {
+          "key": "2020",
+          "label": "2020",
+          "count": 0
+        },
+        {
+          "key": "2021",
+          "label": "2021",
+          "count": 2
+        },
+        {
+          "key": "2022",
+          "label": "2022",
+          "count": 8
+        },
+        {
+          "key": "2023",
+          "label": "2023",
+          "count": 8
+        },
+        {
+          "key": "2024",
+          "label": "2024",
+          "count": 14
+        },
+        {
+          "key": "2025",
+          "label": "2025",
+          "count": 23
+        },
+        {
+          "key": "2026",
+          "label": "2026",
+          "count": 14
+        }
+      ]
+    },
+    "by_technique": [
+      {
+        "key": "tlp",
+        "label": "Ternary Logic Partitioning (TLP)",
+        "count": 31
+      },
+      {
+        "key": "norec",
+        "label": "Non-optimizing Reference Engine Construction (NoREC)",
+        "count": 17
+      },
+      {
+        "key": "pqs",
+        "label": "Pivoted Query Synthesis (PQS)",
+        "count": 12
+      },
+      {
+        "key": "dqp",
+        "label": "Differential Query Plans (DQP)",
+        "count": 5
+      },
+      {
+        "key": "qpg",
+        "label": "Query Plan Guidance (QPG)",
+        "count": 5
+      },
+      {
+        "key": "cert",
+        "label": "Cardinality Estimation Restriction Testing (CERT)",
+        "count": 3
+      },
+      {
+        "key": "coddtest",
+        "label": "Constant-Optimization-Driven Testing (CODDTest)",
+        "count": 1
+      }
+    ],
+    "years": [
+      2020,
+      2021,
+      2022,
+      2023,
+      2024,
+      2025,
+      2026
+    ]
+  },
+  "dbms": {
+    "supported": 21,
+    "with_bugs": 41,
+    "with_adoption": 22,
+    "planning_adoption": 9,
+    "adoption_by_relationship": [
+      {
+        "key": "official_ci",
+        "label": "SQLancer in the project's CI",
+        "count": 1
+      },
+      {
+        "key": "official_testing",
+        "label": "Project-maintained SQLancer testing",
+        "count": 21
+      },
+      {
+        "key": "developer_use",
+        "label": "Developer-reported use",
+        "count": 1
+      }
+    ],
+    "using_without_bugs": [
+      {
+        "id": "feldera",
+        "name": "Feldera"
+      },
+      {
+        "id": "materialize",
+        "name": "Materialize"
+      },
+      {
+        "id": "noisepage",
+        "name": "NoisePage"
+      },
+      {
+        "id": "serenedb",
+        "name": "SereneDB"
+      },
+      {
+        "id": "xugu",
+        "name": "XuGu"
+      },
+      {
+        "id": "ydb",
+        "name": "YDB"
+      }
+    ],
+    "rows": [
+      {
+        "id": "duckdb",
+        "name": "DuckDB",
+        "url": "https://duckdb.org/",
+        "supported": true,
+        "bugs": 302,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/duckdb/duckdb",
+        "bugs_rejected": 7,
+        "bugs_by_year": [
+          {
+            "key": "2020",
+            "label": "2020",
+            "count": 85
+          },
+          {
+            "key": "2021",
+            "label": "2021",
+            "count": 1
+          },
+          {
+            "key": "2022",
+            "label": "2022",
+            "count": 7
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 39
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 9
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 52
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 103
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 6
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 268
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 34
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 271
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 9
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 22
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 30
+          },
+          {
+            "key": "error",
+            "label": "Unexpected error",
+            "count": 13
+          },
+          {
+            "key": "crash",
+            "label": "Crash",
+            "count": 32
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 227
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 112
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 102
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 46
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 33
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 9
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 286
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 15
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 1
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 75
+          },
+          {
+            "key": "DerZc",
+            "label": "DerZc",
+            "count": 48
+          },
+          {
+            "key": "Yibo-Dong",
+            "label": "Yibo-Dong",
+            "count": 39
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 38
+          },
+          {
+            "key": "Zhaokun Xiang",
+            "label": "Zhaokun Xiang",
+            "count": 24
+          },
+          {
+            "key": "Chi Zhang",
+            "label": "Chi Zhang",
+            "count": 23
+          },
+          {
+            "key": "wanteatfruit",
+            "label": "wanteatfruit",
+            "count": 14
+          },
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 9
+          }
+        ],
+        "first_reported": "2020-04-07",
+        "last_reported": "2026-09-07"
+      },
+      {
+        "id": "sqlite",
+        "name": "SQLite",
+        "url": "https://sqlite.org/",
+        "supported": true,
+        "bugs": 222,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/sqlite/sqlite",
+        "bugs_rejected": 19,
+        "bugs_by_year": [
+          {
+            "key": "2019",
+            "label": "2019",
+            "count": 167
+          },
+          {
+            "key": "2020",
+            "label": "2020",
+            "count": 11
+          },
+          {
+            "key": "2022",
+            "label": "2022",
+            "count": 22
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 6
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 12
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 4
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 132
+          },
+          {
+            "key": "pqs",
+            "label": "Pivoted Query Synthesis (PQS)",
+            "count": 48
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 38
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 4
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 210
+          },
+          {
+            "key": "fixed_in_documentation",
+            "label": "Fixed in documentation",
+            "count": 11
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 90
+          },
+          {
+            "key": "error",
+            "label": "Unexpected error",
+            "count": 44
+          },
+          {
+            "key": "crash",
+            "label": "Crash",
+            "count": 42
+          },
+          {
+            "key": "hang",
+            "label": "Hang",
+            "count": 2
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 44
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 90
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 88
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 44
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 222
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 178
+          },
+          {
+            "key": "bajinsheng",
+            "label": "bajinsheng",
+            "count": 22
+          },
+          {
+            "key": "Zhaokun Xiang",
+            "label": "Zhaokun Xiang",
+            "count": 12
+          },
+          {
+            "key": "Chi Zhang",
+            "label": "Chi Zhang",
+            "count": 6
+          },
+          {
+            "key": "Axel Teo",
+            "label": "Axel Teo",
+            "count": 4
+          }
+        ],
+        "first_reported": "2019-05-01",
+        "last_reported": "2026-03-20"
+      },
+      {
+        "id": "cockroachdb",
+        "name": "CockroachDB",
+        "url": "https://www.cockroachlabs.com/",
+        "supported": true,
+        "bugs": 202,
+        "adoption_relationships": [],
+        "planned_adoption": true,
+        "repository": "https://github.com/cockroachdb/cockroach",
+        "bugs_rejected": 11,
+        "bugs_by_year": [
+          {
+            "key": "2020",
+            "label": "2020",
+            "count": 71
+          },
+          {
+            "key": "2021",
+            "label": "2021",
+            "count": 10
+          },
+          {
+            "key": "2022",
+            "label": "2022",
+            "count": 34
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 12
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 6
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 37
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 12
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 20
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 141
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 53
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 7
+          },
+          {
+            "key": "pqs",
+            "label": "Pivoted Query Synthesis (PQS)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 161
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 16
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 25
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 15
+          },
+          {
+            "key": "error",
+            "label": "Unexpected error",
+            "count": 46
+          },
+          {
+            "key": "crash",
+            "label": "Crash",
+            "count": 5
+          },
+          {
+            "key": "hang",
+            "label": "Hang",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 135
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 71
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 61
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 53
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 13
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 4
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 127
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 56
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 19
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 67
+          },
+          {
+            "key": "cockroach-teamcity",
+            "label": "cockroach-teamcity",
+            "count": 40
+          },
+          {
+            "key": "DerZc",
+            "label": "DerZc",
+            "count": 32
+          },
+          {
+            "key": "bajinsheng",
+            "label": "bajinsheng",
+            "count": 19
+          },
+          {
+            "key": "mgartner",
+            "label": "mgartner",
+            "count": 7
+          },
+          {
+            "key": "mrigger",
+            "label": "mrigger",
+            "count": 3
+          },
+          {
+            "key": "srosenberg",
+            "label": "srosenberg",
+            "count": 3
+          },
+          {
+            "key": "TheoristCoder",
+            "label": "TheoristCoder",
+            "count": 2
+          }
+        ],
+        "first_reported": "2020-01-01",
+        "last_reported": "2026-09-13"
+      },
+      {
+        "id": "starrocks",
+        "name": "StarRocks",
+        "url": "https://www.starrocks.io/",
+        "supported": false,
+        "bugs": 183,
+        "adoption_relationships": [
+          "developer_use"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/StarRocks/starrocks",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2021",
+            "label": "2021",
+            "count": 3
+          },
+          {
+            "key": "2022",
+            "label": "2022",
+            "count": 119
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 59
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 2
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 148
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 28
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 7
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 183
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 183
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 182
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 183
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "wanpengfei-git",
+            "label": "wanpengfei-git",
+            "count": 55
+          },
+          {
+            "key": "andyziye",
+            "label": "andyziye",
+            "count": 36
+          },
+          {
+            "key": "dulong41",
+            "label": "dulong41",
+            "count": 29
+          },
+          {
+            "key": "satanson",
+            "label": "satanson",
+            "count": 27
+          },
+          {
+            "key": "yongbingwang",
+            "label": "yongbingwang",
+            "count": 10
+          },
+          {
+            "key": "colorfulu",
+            "label": "colorfulu",
+            "count": 8
+          },
+          {
+            "key": "tiannan-sr",
+            "label": "tiannan-sr",
+            "count": 7
+          },
+          {
+            "key": "wangruin",
+            "label": "wangruin",
+            "count": 5
+          }
+        ],
+        "first_reported": "2021-10-16",
+        "last_reported": "2025-06-18"
+      },
+      {
+        "id": "tidb",
+        "name": "TiDB",
+        "url": "https://www.pingcap.com/tidb/",
+        "supported": true,
+        "bugs": 172,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/pingcap/tidb",
+        "bugs_rejected": 12,
+        "bugs_by_year": [
+          {
+            "key": "2020",
+            "label": "2020",
+            "count": 61
+          },
+          {
+            "key": "2022",
+            "label": "2022",
+            "count": 8
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 12
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 1
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 42
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 36
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 12
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 137
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 31
+          },
+          {
+            "key": "pqs",
+            "label": "Pivoted Query Synthesis (PQS)",
+            "count": 2
+          },
+          {
+            "key": "dqp",
+            "label": "Differential Query Plans (DQP)",
+            "count": 1
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 58
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 48
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 66
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 30
+          },
+          {
+            "key": "error",
+            "label": "Unexpected error",
+            "count": 27
+          },
+          {
+            "key": "crash",
+            "label": "Crash",
+            "count": 4
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 111
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 79
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 42
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 32
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 17
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 2
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 154
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 7
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 11
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 61
+          },
+          {
+            "key": "DerZc",
+            "label": "DerZc",
+            "count": 44
+          },
+          {
+            "key": "Chi Zhang",
+            "label": "Chi Zhang",
+            "count": 10
+          },
+          {
+            "key": "bajinsheng",
+            "label": "bajinsheng",
+            "count": 9
+          },
+          {
+            "key": "wanteatfruit",
+            "label": "wanteatfruit",
+            "count": 9
+          },
+          {
+            "key": "TheoristCoder",
+            "label": "TheoristCoder",
+            "count": 8
+          },
+          {
+            "key": "Zhaokun Xiang",
+            "label": "Zhaokun Xiang",
+            "count": 7
+          },
+          {
+            "key": "hawkingrei",
+            "label": "hawkingrei",
+            "count": 4
+          }
+        ],
+        "first_reported": "2020-03-26",
+        "last_reported": "2026-08-31"
+      },
+      {
+        "id": "dolt",
+        "name": "Dolt",
+        "url": "https://www.dolthub.com/",
+        "supported": false,
+        "bugs": 167,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/dolthub/dolt",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 18
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 12
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 24
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 113
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 167
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 125
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 1
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 41
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 167
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 113
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 53
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 167
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Yibo-Dong",
+            "label": "Yibo-Dong",
+            "count": 94
+          },
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 23
+          },
+          {
+            "key": "Zhaokun Xiang",
+            "label": "Zhaokun Xiang",
+            "count": 19
+          },
+          {
+            "key": "EmilyOng",
+            "label": "EmilyOng",
+            "count": 13
+          },
+          {
+            "key": "wanteatfruit",
+            "label": "wanteatfruit",
+            "count": 11
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 7
+          }
+        ],
+        "first_reported": "2023-07-12",
+        "last_reported": "2026-11-01"
+      },
+      {
+        "id": "monetdb",
+        "name": "MonetDB",
+        "url": "https://www.monetdb.org/",
+        "supported": false,
+        "bugs": 139,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/MonetDB/MonetDB",
+        "bugs_rejected": 1,
+        "bugs_by_year": [
+          {
+            "key": "2020",
+            "label": "2020",
+            "count": 41
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 26
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 32
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 19
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 21
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 136
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 3
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 124
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 15
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 139
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 114
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 24
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 97
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 42
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 42
+          },
+          {
+            "key": "monetdb-team",
+            "label": "monetdb-team",
+            "count": 41
+          },
+          {
+            "key": "Zhaokun Xiang",
+            "label": "Zhaokun Xiang",
+            "count": 29
+          },
+          {
+            "key": "Yibo-Dong",
+            "label": "Yibo-Dong",
+            "count": 18
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 5
+          },
+          {
+            "key": "TheoristCoder",
+            "label": "TheoristCoder",
+            "count": 3
+          },
+          {
+            "key": "fyr03",
+            "label": "fyr03",
+            "count": 1
+          }
+        ],
+        "first_reported": "2020-11-30",
+        "last_reported": "2026-08-17"
+      },
+      {
+        "id": "cratedb",
+        "name": "CrateDB",
+        "url": "https://cratedb.com/",
+        "supported": false,
+        "bugs": 81,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/crate/crate",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 12
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 25
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 8
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 25
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 11
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 81
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 79
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 1
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 81
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 53
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 28
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 81
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 31
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 20
+          },
+          {
+            "key": "wanteatfruit",
+            "label": "wanteatfruit",
+            "count": 13
+          },
+          {
+            "key": "Yibo-Dong",
+            "label": "Yibo-Dong",
+            "count": 11
+          },
+          {
+            "key": "YuanchengJiang",
+            "label": "YuanchengJiang",
+            "count": 3
+          },
+          {
+            "key": "Zhaokun Xiang",
+            "label": "Zhaokun Xiang",
+            "count": 3
+          }
+        ],
+        "first_reported": "2023-09-11",
+        "last_reported": "2026-08-21"
+      },
+      {
+        "id": "questdb",
+        "name": "QuestDB",
+        "url": "https://questdb.io/",
+        "supported": true,
+        "bugs": 55,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/questdb/questdb",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2022",
+            "label": "2022",
+            "count": 3
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 44
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 6
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 2
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 55
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 49
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 1
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 5
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 55
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 49
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 4
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 2
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 53
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 2
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "YuanchengJiang",
+            "label": "YuanchengJiang",
+            "count": 49
+          },
+          {
+            "key": "Ming Wei Tan",
+            "label": "Ming Wei Tan",
+            "count": 2
+          },
+          {
+            "key": "puzpuzpuz",
+            "label": "puzpuzpuz",
+            "count": 2
+          },
+          {
+            "key": "SuriZhang",
+            "label": "SuriZhang",
+            "count": 1
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 1
+          }
+        ],
+        "first_reported": "2022-10-26",
+        "last_reported": "2026-02-09"
+      },
+      {
+        "id": "umbra",
+        "name": "Umbra",
+        "url": "https://umbra-db.com/",
+        "supported": false,
+        "bugs": 55,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": null,
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 18
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 28
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 9
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 55
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 55
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 55
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 55
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 55
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 46
+          },
+          {
+            "key": "Zhaokun Xiang",
+            "label": "Zhaokun Xiang",
+            "count": 9
+          }
+        ],
+        "first_reported": "2023-05-12",
+        "last_reported": "2025-11-07"
+      },
+      {
+        "id": "datafusion",
+        "name": "Apache DataFusion",
+        "url": "https://datafusion.apache.org/",
+        "supported": true,
+        "bugs": 54,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/apache/datafusion",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 48
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 6
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 39
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 10
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 5
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 52
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 2
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 54
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 54
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 54
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "2010YOUY01",
+            "label": "2010YOUY01",
+            "count": 53
+          },
+          {
+            "key": "LorrensP-2158466",
+            "label": "LorrensP-2158466",
+            "count": 1
+          }
+        ],
+        "first_reported": "2024-06-20",
+        "last_reported": "2025-09-08"
+      },
+      {
+        "id": "yugabytedb",
+        "name": "YugabyteDB",
+        "url": "https://www.yugabyte.com/",
+        "supported": true,
+        "bugs": 54,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/yugabyte/yugabyte-db",
+        "bugs_rejected": 6,
+        "bugs_by_year": [
+          {
+            "key": "2021",
+            "label": "2021",
+            "count": 3
+          },
+          {
+            "key": "2022",
+            "label": "2022",
+            "count": 16
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 3
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 1
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 26
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 3
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 2
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 54
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 24
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 30
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 54
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 41
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 9
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 2
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 2
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 13
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 39
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 2
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "qvad",
+            "label": "qvad",
+            "count": 35
+          },
+          {
+            "key": "bajinsheng",
+            "label": "bajinsheng",
+            "count": 13
+          },
+          {
+            "key": "def-",
+            "label": "def-",
+            "count": 2
+          },
+          {
+            "key": "rojasbinny-yb",
+            "label": "rojasbinny-yb",
+            "count": 1
+          },
+          {
+            "key": "yugabyte-ci",
+            "label": "yugabyte-ci",
+            "count": 1
+          }
+        ],
+        "first_reported": "2021-09-29",
+        "last_reported": "2026-07-17"
+      },
+      {
+        "id": "clickhouse",
+        "name": "ClickHouse",
+        "url": "https://clickhouse.com/",
+        "supported": true,
+        "bugs": 48,
+        "adoption_relationships": [
+          "official_ci",
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/ClickHouse/ClickHouse",
+        "bugs_rejected": 1,
+        "bugs_by_year": [
+          {
+            "key": "2020",
+            "label": "2020",
+            "count": 5
+          },
+          {
+            "key": "2021",
+            "label": "2021",
+            "count": 4
+          },
+          {
+            "key": "2022",
+            "label": "2022",
+            "count": 2
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 2
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 2
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 3
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 29
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 1
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 32
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 11
+          },
+          {
+            "key": "dqp",
+            "label": "Differential Query Plans (DQP)",
+            "count": 2
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 2
+          },
+          {
+            "key": "coddtest",
+            "label": "Constant-Optimization-Driven Testing (CODDTest)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 36
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 1
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 11
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 48
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 26
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 10
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 8
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 4
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 7
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 41
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "qoega",
+            "label": "qoega",
+            "count": 20
+          },
+          {
+            "key": "fm4v",
+            "label": "fm4v",
+            "count": 12
+          },
+          {
+            "key": "wanteatfruit",
+            "label": "wanteatfruit",
+            "count": 3
+          },
+          {
+            "key": "Algunenano",
+            "label": "Algunenano",
+            "count": 2
+          },
+          {
+            "key": "alexey-milovidov",
+            "label": "alexey-milovidov",
+            "count": 2
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 2
+          },
+          {
+            "key": "zlareb1",
+            "label": "zlareb1",
+            "count": 2
+          },
+          {
+            "key": "123lpygithub",
+            "label": "123lpygithub",
+            "count": 1
+          }
+        ],
+        "first_reported": "2020-06-24",
+        "last_reported": "2026-09-12"
+      },
+      {
+        "id": "mysql",
+        "name": "MySQL",
+        "url": "https://www.mysql.com/",
+        "supported": true,
+        "bugs": 43,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/mysql/mysql-server",
+        "bugs_rejected": 9,
+        "bugs_by_year": [
+          {
+            "key": "2019",
+            "label": "2019",
+            "count": 25
+          },
+          {
+            "key": "2020",
+            "label": "2020",
+            "count": 7
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 11
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 22
+          },
+          {
+            "key": "pqs",
+            "label": "Pivoted Query Synthesis (PQS)",
+            "count": 14
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 7
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 17
+          },
+          {
+            "key": "fixed_in_documentation",
+            "label": "Fixed in documentation",
+            "count": 2
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 13
+          },
+          {
+            "key": "unknown",
+            "label": "Unknown",
+            "count": 11
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 21
+          },
+          {
+            "key": "error",
+            "label": "Unexpected error",
+            "count": 9
+          },
+          {
+            "key": "crash",
+            "label": "Crash",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 12
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 22
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 21
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 32
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 11
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 31
+          },
+          {
+            "key": "mrigger",
+            "label": "mrigger",
+            "count": 1
+          }
+        ],
+        "first_reported": "2019-06-15",
+        "last_reported": "2020-04-01"
+      },
+      {
+        "id": "h2",
+        "name": "H2",
+        "url": "https://h2database.com/",
+        "supported": true,
+        "bugs": 20,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/h2database/h2database",
+        "bugs_rejected": 1,
+        "bugs_by_year": [
+          {
+            "key": "2020",
+            "label": "2020",
+            "count": 18
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 2
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 18
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 2
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 18
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 2
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 2
+          },
+          {
+            "key": "error",
+            "label": "Unexpected error",
+            "count": 15
+          },
+          {
+            "key": "crash",
+            "label": "Crash",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 2
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 16
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 2
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 1
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 20
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 18
+          },
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 1
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 1
+          }
+        ],
+        "first_reported": "2020-08-12",
+        "last_reported": "2024-01-25"
+      },
+      {
+        "id": "doris",
+        "name": "Apache Doris",
+        "url": "https://doris.apache.org/",
+        "supported": true,
+        "bugs": 19,
+        "adoption_relationships": [],
+        "planned_adoption": true,
+        "repository": "https://github.com/apache/doris",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2020",
+            "label": "2020",
+            "count": 1
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 8
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 3
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 7
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 19
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 19
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 19
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 8
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 7
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 2
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 2
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 3
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 9
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 7
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "ChaseHuangxu",
+            "label": "ChaseHuangxu",
+            "count": 6
+          },
+          {
+            "key": "malwaregarry",
+            "label": "malwaregarry",
+            "count": 3
+          },
+          {
+            "key": "codenohup",
+            "label": "codenohup",
+            "count": 2
+          },
+          {
+            "key": "sduzh",
+            "label": "sduzh",
+            "count": 1
+          }
+        ],
+        "first_reported": "2020-06-15",
+        "last_reported": "2024-06-09"
+      },
+      {
+        "id": "databend",
+        "name": "Databend",
+        "url": "https://www.databend.com/",
+        "supported": true,
+        "bugs": 18,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/datafuselabs/databend",
+        "bugs_rejected": 1,
+        "bugs_by_year": [
+          {
+            "key": "2022",
+            "label": "2022",
+            "count": 4
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 1
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 5
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 8
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 17
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 15
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 3
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 18
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 9
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 9
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 7
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 11
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel-Neuer1",
+            "label": "Manuel-Neuer1",
+            "count": 6
+          },
+          {
+            "key": "hanyisong",
+            "label": "hanyisong",
+            "count": 5
+          },
+          {
+            "key": "Ming Wei Tan",
+            "label": "Ming Wei Tan",
+            "count": 3
+          },
+          {
+            "key": "malwaregarry",
+            "label": "malwaregarry",
+            "count": 2
+          },
+          {
+            "key": "mrigger",
+            "label": "mrigger",
+            "count": 2
+          }
+        ],
+        "first_reported": "2022-08-29",
+        "last_reported": "2026-08-25"
+      },
+      {
+        "id": "postgresql",
+        "name": "PostgreSQL",
+        "url": "https://www.postgresql.org/",
+        "supported": true,
+        "bugs": 18,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/postgres/postgres",
+        "bugs_rejected": 16,
+        "bugs_by_year": [
+          {
+            "key": "2019",
+            "label": "2019",
+            "count": 18
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 17
+          },
+          {
+            "key": "pqs",
+            "label": "Pivoted Query Synthesis (PQS)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 12
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 5
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 1
+          },
+          {
+            "key": "error",
+            "label": "Unexpected error",
+            "count": 11
+          },
+          {
+            "key": "crash",
+            "label": "Crash",
+            "count": 5
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 1
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 17
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 18
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 17
+          },
+          {
+            "key": "mrigger",
+            "label": "mrigger",
+            "count": 1
+          }
+        ],
+        "first_reported": "2019-07-02",
+        "last_reported": "2019-12-02"
+      },
+      {
+        "id": "turso",
+        "name": "Turso",
+        "url": "https://turso.tech/",
+        "supported": false,
+        "bugs": 18,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/tursodatabase/turso",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 18
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 15
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 2
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 18
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 18
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 13
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 4
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 18
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "LeMikaelF",
+            "label": "LeMikaelF",
+            "count": 16
+          },
+          {
+            "key": "turso-github-handyman[bot]",
+            "label": "turso-github-handyman[bot]",
+            "count": 2
+          }
+        ],
+        "first_reported": "2026-01-07",
+        "last_reported": "2026-01-14"
+      },
+      {
+        "id": "virtuoso",
+        "name": "Virtuoso",
+        "url": "https://virtuoso.openlinksw.com/",
+        "supported": false,
+        "bugs": 17,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/openlink/virtuoso-opensource",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 11
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 6
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 11
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 6
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 16
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 17
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 10
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 6
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 11
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 6
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 9
+          },
+          {
+            "key": "Jasper0209",
+            "label": "Jasper0209",
+            "count": 6
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 2
+          }
+        ],
+        "first_reported": "2024-01-14",
+        "last_reported": "2026-05-28"
+      },
+      {
+        "id": "citus",
+        "name": "Citus",
+        "url": "https://www.citusdata.com/",
+        "supported": true,
+        "bugs": 16,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/citusdata/citus",
+        "bugs_rejected": 1,
+        "bugs_by_year": [
+          {
+            "key": "2020",
+            "label": "2020",
+            "count": 2
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 1
+          },
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 1
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 4
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 8
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 15
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 8
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 8
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 16
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 8
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 4
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 4
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 1
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 7
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 8
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel-Neuer1",
+            "label": "Manuel-Neuer1",
+            "count": 2
+          },
+          {
+            "key": "alperkocatas",
+            "label": "alperkocatas",
+            "count": 2
+          },
+          {
+            "key": "m3hm3t",
+            "label": "m3hm3t",
+            "count": 1
+          },
+          {
+            "key": "mrigger",
+            "label": "mrigger",
+            "count": 1
+          },
+          {
+            "key": "nukoyluoglu",
+            "label": "nukoyluoglu",
+            "count": 1
+          },
+          {
+            "key": "onderkalaci",
+            "label": "onderkalaci",
+            "count": 1
+          }
+        ],
+        "first_reported": "2020-06-30",
+        "last_reported": "2026-09-03"
+      },
+      {
+        "id": "stonedb",
+        "name": "StoneDB",
+        "url": "https://stonedb.io/",
+        "supported": false,
+        "bugs": 16,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/stoneatom/stonedb",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 12
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 4
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 16
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 3
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 13
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 16
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 10
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 4
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 2
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 12
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 4
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "zhenglin-charlie-li",
+            "label": "zhenglin-charlie-li",
+            "count": 12
+          }
+        ],
+        "first_reported": "2023-07-03",
+        "last_reported": "2023-07-29"
+      },
+      {
+        "id": "wadjet",
+        "name": "Wadjet",
+        "url": "https://github.com/derekmwright/wadjet",
+        "supported": false,
+        "bugs": 16,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/derekmwright/wadjet",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 16
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 9
+          },
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 6
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 16
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 16
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 12
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 4
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 16
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "derekmwright",
+            "label": "derekmwright",
+            "count": 16
+          }
+        ],
+        "first_reported": "2026-08-24",
+        "last_reported": "2026-09-03"
+      },
+      {
+        "id": "mariadb",
+        "name": "MariaDB",
+        "url": "https://mariadb.org/",
+        "supported": true,
+        "bugs": 15,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/MariaDB/server",
+        "bugs_rejected": 4,
+        "bugs_by_year": [
+          {
+            "key": "2019",
+            "label": "2019",
+            "count": 6
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 8
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 9
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 6
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 1
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 5
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Unknown",
+            "count": 8
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 5
+          },
+          {
+            "key": "crash",
+            "label": "Crash",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 9
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 9
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 5
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 6
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 8
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 6
+          },
+          {
+            "key": "Jasper Andrew",
+            "label": "Jasper Andrew",
+            "count": 1
+          }
+        ],
+        "first_reported": "2019-11-11",
+        "last_reported": "2026-04-01"
+      },
+      {
+        "id": "oxla",
+        "name": "Oxla",
+        "url": "https://oxla.com/",
+        "supported": false,
+        "bugs": 12,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": null,
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 12
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 12
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "unknown",
+            "label": "Unknown",
+            "count": 12
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 12
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 12
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 12
+          }
+        ],
+        "top_reporters": [],
+        "first_reported": null,
+        "last_reported": null
+      },
+      {
+        "id": "tdengine",
+        "name": "TDengine",
+        "url": "https://tdengine.com/",
+        "supported": false,
+        "bugs": 12,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/taosdata/TDengine",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2019",
+            "label": "2019",
+            "count": 5
+          },
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 7
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 11
+          },
+          {
+            "key": "pqs",
+            "label": "Pivoted Query Synthesis (PQS)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 8
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 4
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "logic",
+            "label": "Logic bug",
+            "count": 1
+          },
+          {
+            "key": "error",
+            "label": "Unexpected error",
+            "count": 3
+          },
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 8
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 8
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 3
+          },
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 12
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "YuanchengJiang",
+            "label": "YuanchengJiang",
+            "count": 7
+          },
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 4
+          },
+          {
+            "key": "mrigger",
+            "label": "mrigger",
+            "count": 1
+          }
+        ],
+        "first_reported": "2019-10-01",
+        "last_reported": "2023-09-12"
+      },
+      {
+        "id": "firebird",
+        "name": "Firebird",
+        "url": "https://firebirdsql.org/",
+        "supported": false,
+        "bugs": 10,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/FirebirdSQL/firebird",
+        "bugs_rejected": 1,
+        "bugs_by_year": [
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 5
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 4
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 1
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 10
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 9
+          },
+          {
+            "key": "verified",
+            "label": "Confirmed",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 10
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 9
+          },
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 10
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 9
+          },
+          {
+            "key": "mrigger",
+            "label": "mrigger",
+            "count": 1
+          }
+        ],
+        "first_reported": "2023-03-12",
+        "last_reported": "2024-09-02"
+      },
+      {
+        "id": "matrixone",
+        "name": "MatrixOne",
+        "url": "https://www.matrixorigin.io/",
+        "supported": false,
+        "bugs": 9,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/matrixorigin/matrixone",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2022",
+            "label": "2022",
+            "count": 9
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 9
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 9
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 9
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 9
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 9
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "iamlinjunhong",
+            "label": "iamlinjunhong",
+            "count": 9
+          }
+        ],
+        "first_reported": "2022-01-10",
+        "last_reported": "2022-02-11"
+      },
+      {
+        "id": "risingwave",
+        "name": "RisingWave",
+        "url": "https://risingwave.com/",
+        "supported": false,
+        "bugs": 8,
+        "adoption_relationships": [],
+        "planned_adoption": true,
+        "repository": "https://github.com/risingwavelabs/risingwave",
+        "bugs_rejected": 1,
+        "bugs_by_year": [
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 8
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 8
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 8
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 8
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 5
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 3
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 8
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "YuanchengJiang",
+            "label": "YuanchengJiang",
+            "count": 4
+          },
+          {
+            "key": "Suyang Zhong",
+            "label": "Suyang Zhong",
+            "count": 3
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 1
+          }
+        ],
+        "first_reported": "2023-10-12",
+        "last_reported": "2023-12-30"
+      },
+      {
+        "id": "bharatdbms",
+        "name": "BharatDBMS",
+        "url": "https://github.com/BharatDBPG/BharatDBMS-PG",
+        "supported": false,
+        "bugs": 7,
+        "adoption_relationships": [],
+        "planned_adoption": true,
+        "repository": "https://github.com/BharatDBPG/BharatDBMS-PG",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 5
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 2
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 7
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 4
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 3
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 7
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 7
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 7
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "BharatDBPG",
+            "label": "BharatDBPG",
+            "count": 7
+          }
+        ],
+        "first_reported": "2025-09-22",
+        "last_reported": "2026-05-26"
+      },
+      {
+        "id": "cnosdb",
+        "name": "CnosDB",
+        "url": "https://www.cnosdb.com/",
+        "supported": false,
+        "bugs": 7,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/cnosdb/cnosdb",
+        "bugs_rejected": 1,
+        "bugs_by_year": [
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 2
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 2
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 1
+          },
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 2
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 5
+          },
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 2
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 6
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 7
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 5
+          },
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 2
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 1
+          },
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 4
+          },
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 2
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Benxiaohai001",
+            "label": "Benxiaohai001",
+            "count": 4
+          },
+          {
+            "key": "mrigger",
+            "label": "mrigger",
+            "count": 1
+          }
+        ],
+        "first_reported": "2023-06-14",
+        "last_reported": "2026-04-26"
+      },
+      {
+        "id": "cloudberry",
+        "name": "Apache Cloudberry",
+        "url": "https://cloudberry.apache.org/",
+        "supported": false,
+        "bugs": 6,
+        "adoption_relationships": [],
+        "planned_adoption": true,
+        "repository": "https://github.com/apache/cloudberry",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 1
+          },
+          {
+            "key": "2024",
+            "label": "2024",
+            "count": 2
+          },
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 3
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 6
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 3
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 3
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 6
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 5
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 6
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "my-ship-it",
+            "label": "my-ship-it",
+            "count": 3
+          },
+          {
+            "key": "shmiwy",
+            "label": "shmiwy",
+            "count": 2
+          },
+          {
+            "key": "congxuebin",
+            "label": "congxuebin",
+            "count": 1
+          }
+        ],
+        "first_reported": "2023-11-29",
+        "last_reported": "2026-09-03"
+      },
+      {
+        "id": "oceanbase",
+        "name": "OceanBase",
+        "url": "https://www.oceanbase.com/",
+        "supported": true,
+        "bugs": 4,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/oceanbase/oceanbase",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 4
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 4
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 4
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 4
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_reporter",
+            "label": "Filed by someone who runs SQLancer campaigns",
+            "count": 3
+          },
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 4
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "bajinsheng",
+            "label": "bajinsheng",
+            "count": 3
+          },
+          {
+            "key": "suyZhong",
+            "label": "suyZhong",
+            "count": 1
+          }
+        ],
+        "first_reported": "2025-03-26",
+        "last_reported": "2025-05-12"
+      },
+      {
+        "id": "presto",
+        "name": "Presto",
+        "url": "https://prestodb.io/",
+        "supported": true,
+        "bugs": 4,
+        "adoption_relationships": [],
+        "planned_adoption": true,
+        "repository": "https://github.com/prestodb/presto",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "unknown",
+            "label": "Year not recorded",
+            "count": 4
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 4
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 1
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 3
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 4
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "curated_primary_source",
+            "label": "Listed in the project's own bug repository",
+            "count": 4
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "unknown",
+            "label": "Reporter not recorded",
+            "count": 4
+          }
+        ],
+        "top_reporters": [],
+        "first_reported": null,
+        "last_reported": null
+      },
+      {
+        "id": "seekdb",
+        "name": "SeekDB",
+        "url": "https://www.oceanbase.com/",
+        "supported": false,
+        "bugs": 3,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/oceanbase/seekdb",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 3
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 2
+          },
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 1
+          },
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 2
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 3
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 2
+          },
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 3
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "chyujing",
+            "label": "chyujing",
+            "count": 1
+          },
+          {
+            "key": "cms-cms",
+            "label": "cms-cms",
+            "count": 1
+          },
+          {
+            "key": "footka",
+            "label": "footka",
+            "count": 1
+          }
+        ],
+        "first_reported": "2026-04-16",
+        "last_reported": "2026-08-11"
+      },
+      {
+        "id": "cubrid",
+        "name": "CUBRID",
+        "url": "https://www.cubrid.org/",
+        "supported": false,
+        "bugs": 1,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/CUBRID/cubrid",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2023",
+            "label": "2023",
+            "count": 1
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 1
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "project",
+            "label": "Found by the SQLancer project",
+            "count": 1
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Manuel Rigger",
+            "label": "Manuel Rigger",
+            "count": 1
+          }
+        ],
+        "first_reported": "2023-12-11",
+        "last_reported": "2023-12-11"
+      },
+      {
+        "id": "falkordb",
+        "name": "FalkorDB",
+        "url": "https://www.falkordb.com/",
+        "supported": false,
+        "bugs": 1,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/FalkorDB/FalkorDB",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 1
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "unattributed",
+            "label": "Technique not recorded",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 1
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "trrrrr617",
+            "label": "trrrrr617",
+            "count": 1
+          }
+        ],
+        "first_reported": "2026-06-18",
+        "last_reported": "2026-06-18"
+      },
+      {
+        "id": "hazelcast",
+        "name": "Hazelcast",
+        "url": "https://hazelcast.com/",
+        "supported": false,
+        "bugs": 1,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/hazelcast/hazelcast",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2021",
+            "label": "2021",
+            "count": 1
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 1
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "Fly-Style",
+            "label": "Fly-Style",
+            "count": 1
+          }
+        ],
+        "first_reported": "2021-10-29",
+        "last_reported": "2021-10-29"
+      },
+      {
+        "id": "kyzo",
+        "name": "Kyzo",
+        "url": "https://github.com/kyzobuild/kyzo",
+        "supported": false,
+        "bugs": 1,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/kyzobuild/kyzo",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 1
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "tlp",
+            "label": "Ternary Logic Partitioning (TLP)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 1
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "explicit_tool_statement",
+            "label": "The report names SQLancer",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "kylejtobin",
+            "label": "kylejtobin",
+            "count": 1
+          }
+        ],
+        "first_reported": "2026-07-02",
+        "last_reported": "2026-07-02"
+      },
+      {
+        "id": "sparq",
+        "name": "sparq",
+        "url": "https://github.com/sparq-org/sparq",
+        "supported": false,
+        "bugs": 1,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false,
+        "repository": "https://github.com/sparq-org/sparq",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2026",
+            "label": "2026",
+            "count": 1
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "norec",
+            "label": "Non-optimizing Reference Engine Construction (NoREC)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "fixed",
+            "label": "Fixed",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 1
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "technique_attribution",
+            "label": "The report names a SQLancer oracle",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "jeswr",
+            "label": "jeswr",
+            "count": 1
+          }
+        ],
+        "first_reported": "2026-07-11",
+        "last_reported": "2026-07-11"
+      },
+      {
+        "id": "tikv",
+        "name": "TiKV",
+        "url": "https://tikv.org/",
+        "supported": false,
+        "bugs": 1,
+        "adoption_relationships": [],
+        "planned_adoption": false,
+        "repository": "https://github.com/tikv/tikv",
+        "bugs_rejected": 0,
+        "bugs_by_year": [
+          {
+            "key": "2025",
+            "label": "2025",
+            "count": 1
+          }
+        ],
+        "bugs_by_technique": [
+          {
+            "key": "qpg",
+            "label": "Query Plan Guidance (QPG)",
+            "count": 1
+          }
+        ],
+        "bugs_by_status": [
+          {
+            "key": "open",
+            "label": "Open",
+            "count": 1
+          }
+        ],
+        "bugs_by_symptom": [
+          {
+            "key": "unknown",
+            "label": "Unclassified",
+            "count": 1
+          }
+        ],
+        "bugs_by_attribution_rule": [
+          {
+            "key": "campaign_evidence",
+            "label": "The reproducer carries SQLancer's generated schema",
+            "count": 1
+          }
+        ],
+        "bugs_by_affiliation": [
+          {
+            "key": "external",
+            "label": "Found by someone outside the project",
+            "count": 1
+          }
+        ],
+        "top_reporters": [
+          {
+            "key": "hbisheng",
+            "label": "hbisheng",
+            "count": 1
+          }
+        ],
+        "first_reported": "2025-09-28",
+        "last_reported": "2025-09-28"
+      },
+      {
+        "id": "agensgraph",
+        "name": "AgensGraph",
+        "url": "https://bitnine.net/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "hive",
+        "name": "Apache Hive",
+        "url": "https://hive.apache.org/",
+        "supported": true,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "spark",
+        "name": "Apache Spark",
+        "url": "https://spark.apache.org/",
+        "supported": true,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "arangodb",
+        "name": "ArangoDB",
+        "url": "https://arangodb.com/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "defradb",
+        "name": "DefraDB",
+        "url": "https://docs.source.network/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "elasticsearch",
+        "name": "Elasticsearch",
+        "url": "https://www.elastic.co/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "feldera",
+        "name": "Feldera",
+        "url": "https://feldera.com/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false
+      },
+      {
+        "id": "greenplum",
+        "name": "Greenplum",
+        "url": "https://greenplum.org/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "greptimedb",
+        "name": "GreptimeDB",
+        "url": "https://greptime.com/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": true
+      },
+      {
+        "id": "hsqldb",
+        "name": "HSQLDB",
+        "url": "https://hsqldb.org/",
+        "supported": true,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "materialize",
+        "name": "Materialize",
+        "url": "https://materialize.com/",
+        "supported": true,
+        "bugs": 0,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false
+      },
+      {
+        "id": "mongodb",
+        "name": "MongoDB",
+        "url": "https://www.mongodb.com/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "neo4j",
+        "name": "Neo4j",
+        "url": "https://neo4j.com/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "noisepage",
+        "name": "NoisePage",
+        "url": "https://noise.page/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false
+      },
+      {
+        "id": "opengauss",
+        "name": "openGauss",
+        "url": "https://opengauss.org/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "opensearch",
+        "name": "OpenSearch SQL",
+        "url": "https://opensearch.org/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "percona",
+        "name": "Percona Server",
+        "url": "https://www.percona.com/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "readyset",
+        "name": "ReadySet",
+        "url": "https://readyset.io/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "redisgraph",
+        "name": "RedisGraph",
+        "url": "https://redis.io/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": false
+      },
+      {
+        "id": "serenedb",
+        "name": "SereneDB",
+        "url": "https://github.com/serenedb/serenedb",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false
+      },
+      {
+        "id": "spiceai",
+        "name": "Spice.ai OSS",
+        "url": "https://spice.ai/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": true
+      },
+      {
+        "id": "tarantool",
+        "name": "Tarantool",
+        "url": "https://www.tarantool.io/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [],
+        "planned_adoption": true
+      },
+      {
+        "id": "xugu",
+        "name": "XuGu",
+        "url": "https://www.xugudb.com/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false
+      },
+      {
+        "id": "ydb",
+        "name": "YDB",
+        "url": "https://ydb.tech/",
+        "supported": false,
+        "bugs": 0,
+        "adoption_relationships": [
+          "official_testing"
+        ],
+        "planned_adoption": false
+      }
+    ]
+  },
+  "resources": {
+    "total": 24,
+    "by_type": [
+      {
+        "key": "talk",
+        "label": "Talks",
+        "count": 9
+      },
+      {
+        "key": "blog_post",
+        "label": "Blog posts",
+        "count": 8
+      },
+      {
+        "key": "tool",
+        "label": "Tools",
+        "count": 4
+      },
+      {
+        "key": "documentation_note",
+        "label": "Project documentation naming SQLancer",
+        "count": 3
+      }
+    ]
+  }
+}
diff --git a/_data/impact/talks.json b/_data/impact/talks.json
new file mode 100644
index 0000000..5c18169
--- /dev/null
+++ b/_data/impact/talks.json
@@ -0,0 +1,888 @@
+{
+  "schema_version": "1.0.0",
+  "talks": [
+    {
+      "id": "talk:youtube:V_qzqY1bb7I",
+      "title": "Reliability Lessons From SQLite - Richard Hipp | SSW 2026",
+      "url": "https://www.youtube.com/watch?v=V_qzqY1bb7I",
+      "video_id": "V_qzqY1bb7I",
+      "speakers": [
+        "Richard Hipp"
+      ],
+      "publisher": "Software Should Work",
+      "event": "SSW 2026",
+      "year": 2026,
+      "duration_seconds": 3261,
+      "related_dbms": [
+        "sqlite"
+      ],
+      "related_techniques": [],
+      "sources": [
+        {
+          "kind": "captions",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=V_qzqY1bb7I",
+          "retrieved_at": "2026-09-11T16:23:20Z",
+          "segments": 515,
+          "characters": 46496,
+          "note": "Windows around the moments SQLancer is mentioned, cut from a transcript supplied by hand. Not the whole talk."
+        },
+        {
+          "kind": "frame",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=V_qzqY1bb7I&t=2836s",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "note": "Captured at 47:16."
+        }
+      ],
+      "mentions": [
+        {
+          "id": "M1",
+          "source": "captions",
+          "at_seconds": 2836,
+          "timestamp": "47:16",
+          "url": "https://www.youtube.com/watch?v=V_qzqY1bb7I&t=2836s",
+          "matched": [
+            "author"
+          ],
+          "heard_as": null,
+          "excerpt": "And then a few years later, um, Manual Rigger came up with this idea of we, you know, the the original fuzzers were just looking for memory errors or searching faults or something like that. He came up with the idea we can we can do fuzzing ideas to test for inconsistencies in SQL.",
+          "excerpt_is_verbatim": true,
+          "technique_ids": []
+        },
+        {
+          "id": "M2",
+          "source": "frame",
+          "at_seconds": 2836,
+          "timestamp": "47:16",
+          "url": "https://www.youtube.com/watch?v=V_qzqY1bb7I&t=2836s",
+          "matched": [
+            "author"
+          ],
+          "heard_as": null,
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "The slide behind him is SQLite's project timeline: query fuzzing arrives around 2020, after 3.0.0, 100% MC/DC coverage and profile-guided fuzzing.",
+          "image": "/assets/images/impact/talks/V_qzqY1bb7I-2836.jpg",
+          "technique_ids": []
+        }
+      ],
+      "relationship": {
+        "roles": [
+          "recognition",
+          "explains_technique"
+        ],
+        "summary": "SQLite's creator on twenty-six years of testing it, crediting Rigger with the idea of fuzzing for inconsistencies in SQL rather than for crashes, explaining the sub-query equivalence that finds them, and saying SQLite's own fuzzer had to be extended to do the same.",
+        "mention_ids": [
+          "M1",
+          "M2"
+        ],
+        "decided_by": "curator"
+      },
+      "provenance": {
+        "collector": "talks",
+        "collector_version": "1.0.0",
+        "first_seen": "2026-09-11T02:04:04Z",
+        "last_verified": "2026-09-12T02:04:10Z",
+        "source_url": "https://www.youtube.com/watch?v=V_qzqY1bb7I",
+        "source_type": "video"
+      }
+    },
+    {
+      "id": "talk:youtube:QRwxHGpWaUA",
+      "title": "The Art of Database Testing by Alperen Keles | DC Systems 011",
+      "url": "https://www.youtube.com/watch?v=QRwxHGpWaUA",
+      "video_id": "QRwxHGpWaUA",
+      "speakers": [
+        "Alperen Keles"
+      ],
+      "publisher": "Antithesis",
+      "event": "DC Systems 011",
+      "year": 2026,
+      "duration_seconds": 2431,
+      "related_dbms": [],
+      "related_techniques": [
+        "cert",
+        "coddtest",
+        "dqp",
+        "norec",
+        "pqs",
+        "qpg",
+        "tlp"
+      ],
+      "sources": [
+        {
+          "kind": "captions",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=QRwxHGpWaUA",
+          "retrieved_at": "2026-09-11T00:20:00Z",
+          "segments": 371,
+          "characters": 32044,
+          "note": "Windows around candidate moments, read out of YouTube's transcript panel in a browser. Not the whole talk: only the segments a loose prefilter flagged, so the matcher decides on text a person can check."
+        },
+        {
+          "kind": "frame",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=QRwxHGpWaUA&t=205s",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "note": "Captured at 3:25."
+        }
+      ],
+      "mentions": [
+        {
+          "id": "M1",
+          "source": "captions",
+          "at_seconds": 205,
+          "timestamp": "3:25",
+          "url": "https://www.youtube.com/watch?v=QRwxHGpWaUA&t=205s",
+          "matched": [
+            "pqs"
+          ],
+          "heard_as": null,
+          "excerpt": "So the first one is called pivoted query synthesis and the idea here is containment.",
+          "excerpt_is_verbatim": true,
+          "technique_ids": [
+            "pqs"
+          ]
+        },
+        {
+          "id": "M2",
+          "source": "frame",
+          "at_seconds": 205,
+          "timestamp": "3:25",
+          "url": "https://www.youtube.com/watch?v=QRwxHGpWaUA&t=205s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": null,
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "The slide behind the speaker is SQLancer's logo above its eight oracles, named in full: PQS, NoREC, TLP, DQE, QPG, CERT, DQP and CODDTest.",
+          "image": "/assets/images/impact/talks/QRwxHGpWaUA-205.jpg",
+          "technique_ids": [
+            "pqs",
+            "norec",
+            "tlp",
+            "qpg",
+            "cert",
+            "dqp",
+            "coddtest"
+          ]
+        },
+        {
+          "id": "M3",
+          "source": "captions",
+          "at_seconds": 2087,
+          "timestamp": "34:47",
+          "url": "https://www.youtube.com/watch?v=QRwxHGpWaUA&t=2087s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": "SQL answer",
+          "excerpt": "I think uh in SQL answer when running you pick uh you pick the oracle so they are for this spe more complex heristics based on the oracle I I don't think I have seen that in the codebase but maybe they have it",
+          "excerpt_is_verbatim": true,
+          "technique_ids": [],
+          "frame_checked": "The closing slide with the speaker's contact details is on screen; this mention is in the questions afterwards."
+        }
+      ],
+      "relationship": {
+        "roles": [
+          "explains_technique",
+          "background"
+        ],
+        "summary": "A survey of database testing methodology that walks an audience through SQLancer's oracles, naming pivoted query synthesis outright and putting ternary logic partitioning on a slide; the tool itself comes up again in questions.",
+        "mention_ids": [
+          "M1",
+          "M2",
+          "M3"
+        ],
+        "decided_by": "curator"
+      },
+      "provenance": {
+        "collector": "talks",
+        "collector_version": "1.0.0",
+        "first_seen": "2026-09-11T02:04:04Z",
+        "last_verified": "2026-09-12T02:04:10Z",
+        "source_url": "https://www.youtube.com/watch?v=QRwxHGpWaUA",
+        "source_type": "video"
+      }
+    },
+    {
+      "id": "talk:youtube:CW4Ntdtp7lg",
+      "title": "Fuzzing databases is difficult",
+      "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg",
+      "video_id": "CW4Ntdtp7lg",
+      "speakers": [
+        "Pedro Ferreira"
+      ],
+      "publisher": "ClickHouse",
+      "event": "FOSDEM 2025",
+      "year": 2025,
+      "duration_seconds": 1611,
+      "related_dbms": [
+        "clickhouse"
+      ],
+      "related_techniques": [],
+      "sources": [
+        {
+          "kind": "captions",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg",
+          "retrieved_at": "2026-09-11T03:10:00Z",
+          "segments": 430,
+          "characters": 22738,
+          "note": "Windows around candidate moments, read out of YouTube's transcript panel in a browser. Not the whole talk: only the segments a loose prefilter flagged, so the matcher decides on text a person can check."
+        },
+        {
+          "kind": "frame",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg&t=110s",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "note": "Captured at 1:50."
+        },
+        {
+          "kind": "frame",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg&t=827s",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "note": "Captured at 13:47."
+        }
+      ],
+      "mentions": [
+        {
+          "id": "M1",
+          "source": "captions",
+          "at_seconds": 110,
+          "timestamp": "1:50",
+          "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg&t=110s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": "SQL Answer",
+          "excerpt": "Uh, there are a few here, uh, a few of them probably already know, uh, like, uh, SQL Answer, which was, uh, Pioneer to find the wrong results. Uh, there are others like FAL and WebFuzzer that are known to do",
+          "excerpt_is_verbatim": true,
+          "technique_ids": []
+        },
+        {
+          "id": "M2",
+          "source": "frame",
+          "at_seconds": 110,
+          "timestamp": "1:50",
+          "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg&t=110s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": null,
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "The slide \"Testing with Fuzzers\" names the fuzzers ClickHouse runs, SQLancer first among them.",
+          "image": "/assets/images/impact/talks/CW4Ntdtp7lg-110.jpg",
+          "technique_ids": []
+        },
+        {
+          "id": "M3",
+          "source": "captions",
+          "at_seconds": 827,
+          "timestamp": "13:47",
+          "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg&t=827s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": "SQLanswer",
+          "excerpt": "This was, SQLanswer actually started this a few years ago. We can do something like, for example, select count from a query with a predicate",
+          "excerpt_is_verbatim": true,
+          "technique_ids": []
+        },
+        {
+          "id": "M4",
+          "source": "frame",
+          "at_seconds": 827,
+          "timestamp": "13:47",
+          "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg&t=827s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": null,
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "The slide \"Finding wrong results\" credits SQLancer by name with pioneering the comparison of equivalent queries against an oracle.",
+          "image": "/assets/images/impact/talks/CW4Ntdtp7lg-827.jpg",
+          "technique_ids": []
+        }
+      ],
+      "relationship": {
+        "roles": [
+          "reports_adoption",
+          "recognition"
+        ],
+        "summary": "A ClickHouse engineer on how the system is fuzzed. SQLancer appears on the early list of the fuzzers ClickHouse runs, again where the talk explains detecting wrong results by comparing equivalent queries, and once more in the closing recommendations.",
+        "mention_ids": [
+          "M1",
+          "M2",
+          "M3",
+          "M4"
+        ],
+        "decided_by": "curator"
+      },
+      "provenance": {
+        "collector": "talks",
+        "collector_version": "1.0.0",
+        "first_seen": "2026-09-11T02:04:04Z",
+        "last_verified": "2026-09-12T02:04:10Z",
+        "source_url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg",
+        "source_type": "video"
+      }
+    },
+    {
+      "id": "talk:youtube:L90MBb6NLBE",
+      "title": "FUZZING'25 Keynote: \"Constraining Fuzzing without Paying Too Much\" by Miryung Kim",
+      "url": "https://www.youtube.com/watch?v=L90MBb6NLBE&t=1703s",
+      "video_id": "L90MBb6NLBE",
+      "speakers": [
+        "Miryung Kim"
+      ],
+      "publisher": "International Fuzzing Workshop",
+      "event": "FUZZING'25",
+      "year": 2025,
+      "duration_seconds": 2786,
+      "related_dbms": [],
+      "related_techniques": [],
+      "sources": [
+        {
+          "kind": "captions",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=L90MBb6NLBE&t=1703s",
+          "retrieved_at": "2026-09-11T16:20:15Z",
+          "segments": 439,
+          "characters": 38388,
+          "note": "The whole transcript was read and SQLancer is not spoken anywhere in it."
+        },
+        {
+          "kind": "frame",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=L90MBb6NLBE&t=1703s",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "note": "Captured at 28:23."
+        }
+      ],
+      "mentions": [
+        {
+          "id": "M1",
+          "source": "frame",
+          "at_seconds": 1703,
+          "timestamp": "28:23",
+          "url": "https://www.youtube.com/watch?v=L90MBb6NLBE&t=1703s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": null,
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "The keynote's table of what fuzzer customisations cost to build lists SQLancer among them, with its contributor, commit and paper counts.",
+          "image": "/assets/images/impact/talks/L90MBb6NLBE-1703.jpg",
+          "technique_ids": []
+        }
+      ],
+      "relationship": {
+        "roles": [
+          "recognition"
+        ],
+        "summary": "A fuzzing keynote that puts SQLancer in its table of what building a custom fuzzer costs, at 28:23.",
+        "mention_ids": [
+          "M1"
+        ],
+        "decided_by": "curator"
+      },
+      "provenance": {
+        "collector": "talks",
+        "collector_version": "1.0.0",
+        "first_seen": "2026-09-11T02:04:04Z",
+        "last_verified": "2026-09-12T02:04:10Z",
+        "source_url": "https://www.youtube.com/watch?v=L90MBb6NLBE&t=1703s",
+        "source_type": "video"
+      }
+    },
+    {
+      "id": "talk:youtube:6YGqFRTe2D0",
+      "title": "[FUZZING'23] \"Three Colours of Fuzzing: Reflections and Open Challenges\" Keynote by Cristian Cadar",
+      "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0",
+      "video_id": "6YGqFRTe2D0",
+      "speakers": [
+        "Cristian Cadar"
+      ],
+      "publisher": "International Fuzzing Workshop",
+      "event": "FUZZING'23",
+      "year": 2023,
+      "duration_seconds": 3638,
+      "related_dbms": [],
+      "related_techniques": [],
+      "sources": [
+        {
+          "kind": "captions",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0",
+          "retrieved_at": "2026-09-11T00:35:00Z",
+          "segments": 456,
+          "characters": 53044,
+          "note": "Windows around candidate moments, read out of YouTube's transcript panel in a browser. Not the whole talk: only the segments a loose prefilter flagged, so the matcher decides on text a person can check."
+        },
+        {
+          "kind": "frame",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0&t=295s",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "note": "Captured at 4:55."
+        },
+        {
+          "kind": "frame",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0&t=1509s",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "note": "Captured at 25:09."
+        }
+      ],
+      "mentions": [
+        {
+          "id": "M1",
+          "source": "captions",
+          "at_seconds": 295,
+          "timestamp": "4:55",
+          "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0&t=295s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": "SQL lenser",
+          "excerpt": "as another example SQL lenser which is a fer for database Management Systems again it has found hundreds of bugs in popular database Management Systems like sqlite and postgress SQL",
+          "excerpt_is_verbatim": true,
+          "technique_ids": []
+        },
+        {
+          "id": "M2",
+          "source": "frame",
+          "at_seconds": 295,
+          "timestamp": "4:55",
+          "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0&t=295s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": null,
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "A slide of fuzzers that made a difference — KLEE, SAGE, AFL, OSS-Fuzz, Csmith and EMI — ending with SQLancer as the DBMS fuzzer, with the bug count it found in popular database systems.",
+          "image": "/assets/images/impact/talks/6YGqFRTe2D0-295.jpg",
+          "technique_ids": []
+        },
+        {
+          "id": "M3",
+          "source": "captions",
+          "at_seconds": 1509,
+          "timestamp": "25:09",
+          "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0&t=1509s",
+          "matched": [
+            "author"
+          ],
+          "heard_as": null,
+          "excerpt": "so I have this quote here and I think that you know if Manuel would be here he would certainly blush uh it says one fing researcher of particular noce Manuel riger and I completely agree with that this is actually from the sqlite web page",
+          "excerpt_is_verbatim": true,
+          "technique_ids": []
+        },
+        {
+          "id": "M4",
+          "source": "frame",
+          "at_seconds": 1509,
+          "timestamp": "25:09",
+          "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0&t=1509s",
+          "matched": [
+            "author"
+          ],
+          "heard_as": null,
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "The slide displays SQLite's own paragraph about the project's author, attributed on the slide to the SQLite webpage. Its wording is in the dataset already, quoted from sqlite.org/testing.html.",
+          "image": "/assets/images/impact/talks/6YGqFRTe2D0-1509.jpg",
+          "technique_ids": []
+        }
+      ],
+      "relationship": {
+        "roles": [
+          "cites_as_example",
+          "recognition"
+        ],
+        "summary": "A fuzzing keynote that reaches for SQLancer as its example of a fuzzer that found hundreds of bugs in mature database systems, and later reads out SQLite's own line about the project's author, agreeing with it.",
+        "mention_ids": [
+          "M1",
+          "M2",
+          "M3",
+          "M4"
+        ],
+        "decided_by": "curator"
+      },
+      "provenance": {
+        "collector": "talks",
+        "collector_version": "1.0.0",
+        "first_seen": "2026-09-11T02:04:04Z",
+        "last_verified": "2026-09-12T02:04:10Z",
+        "source_url": "https://www.youtube.com/watch?v=6YGqFRTe2D0",
+        "source_type": "video"
+      }
+    },
+    {
+      "id": "talk:9d70cce25a55",
+      "title": "ClickHouse Release 22.3 Webinar",
+      "url": "https://presentations.clickhouse.com/2022-release-22.3/index.html",
+      "video_id": null,
+      "speakers": [
+        "Alexey Milovidov"
+      ],
+      "publisher": "ClickHouse",
+      "event": "ClickHouse 22.3 release webinar",
+      "year": 2022,
+      "duration_seconds": null,
+      "related_dbms": [
+        "clickhouse"
+      ],
+      "related_techniques": [],
+      "sources": [
+        {
+          "kind": "slides",
+          "status": "extracted",
+          "url": "https://presentations.clickhouse.com/2022-release-22.3/index.html",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "characters": 8358,
+          "content_sha256": "sha256:81b2bb42aa2f0d5f927585031adf9e8d8d99ac44b300d8e4d2b8a5e620dff2b1"
+        }
+      ],
+      "mentions": [
+        {
+          "id": "M1",
+          "source": "slides",
+          "at_seconds": null,
+          "timestamp": null,
+          "url": "https://presentations.clickhouse.com/2022-release-22.3/index.html",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": null,
+          "excerpt": "— SQLancer — logical fuzzer.",
+          "excerpt_is_verbatim": true,
+          "technique_ids": [],
+          "image": "/assets/images/impact/talks/clickhouse-2022-release-22.3.jpg"
+        }
+      ],
+      "relationship": {
+        "roles": [
+          "reports_adoption"
+        ],
+        "summary": "The release webinar's section on continuous integration lists SQLancer among the fuzzing methods ClickHouse runs, beside libFuzzer, the AST query fuzzer and Jepsen.",
+        "mention_ids": [
+          "M1"
+        ],
+        "decided_by": "curator"
+      },
+      "provenance": {
+        "collector": "talks",
+        "collector_version": "1.0.0",
+        "first_seen": "2026-09-11T02:04:04Z",
+        "last_verified": "2026-09-12T02:04:10Z",
+        "source_url": "https://presentations.clickhouse.com/2022-release-22.3/index.html",
+        "source_type": "video"
+      }
+    },
+    {
+      "id": "talk:youtube:BgC79Zt2fPs",
+      "title": "Keynote 1: DuckDB Testing - Present and Future",
+      "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs",
+      "video_id": "BgC79Zt2fPs",
+      "speakers": [
+        "Mark Raasveldt"
+      ],
+      "publisher": "DBTest Workshop",
+      "event": "DBTest 2022",
+      "year": 2022,
+      "duration_seconds": 3849,
+      "related_dbms": [
+        "duckdb"
+      ],
+      "related_techniques": [],
+      "sources": [
+        {
+          "kind": "captions",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs",
+          "retrieved_at": "2026-09-12T02:20:00Z",
+          "segments": 481,
+          "characters": 57721,
+          "note": "Windows around candidate moments, read out of YouTube's transcript panel in a browser. Not the whole talk: only the segments a loose prefilter flagged, so the matcher decides on text a person can check."
+        },
+        {
+          "kind": "frame",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs&t=2144s",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "note": "Captured at 35:44."
+        },
+        {
+          "kind": "frame",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs&t=2170s",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "note": "Captured at 36:10."
+        },
+        {
+          "kind": "frame",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs&t=2455s",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "note": "Captured at 40:55."
+        }
+      ],
+      "mentions": [
+        {
+          "id": "M1",
+          "source": "frame",
+          "at_seconds": 2144,
+          "timestamp": "35:44",
+          "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs&t=2144s",
+          "matched": [
+            "author"
+          ],
+          "heard_as": null,
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "The slide behind the story: “and then Dr. Rigger came along!”, over a DuckDB bug report.",
+          "image": "/assets/images/impact/talks/BgC79Zt2fPs-2144.jpg",
+          "technique_ids": []
+        },
+        {
+          "id": "M2",
+          "source": "captions",
+          "at_seconds": 2154,
+          "timestamp": "35:54",
+          "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs&t=2154s",
+          "matched": [
+            "sqlancer",
+            "author"
+          ],
+          "heard_as": "sql answer",
+          "excerpt": "thought we had like a pretty robust system right we had like um thousands of tests from various systems like sqlite postgres hundreds of our own tests and then this guy called dr rieger came along and he uh unleashed his creation upon us and started furiously opening bug reports so uh using sql answer manual found around 80 bugs inductive and those were bugs that were not found using the test suites of the other systems right so we ran the sql light tests they did not find these bugs we ran the postgres test they didn't find these bugs and it turns out this kind of thing where database systems are complex surprise surprise and each system has their own",
+          "excerpt_is_verbatim": true,
+          "technique_ids": []
+        },
+        {
+          "id": "M3",
+          "source": "frame",
+          "at_seconds": 2170,
+          "timestamp": "36:10",
+          "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs&t=2170s",
+          "matched": [
+            "sqlancer",
+            "author"
+          ],
+          "heard_as": null,
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "The next slide puts a number on it: the bugs SQLancer found in DuckDB, which the test suites DuckDB had borrowed from other systems did not.",
+          "image": "/assets/images/impact/talks/BgC79Zt2fPs-2170.jpg",
+          "technique_ids": []
+        },
+        {
+          "id": "M4",
+          "source": "captions",
+          "at_seconds": 2268,
+          "timestamp": "37:48",
+          "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs&t=2268s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": "sql lancer",
+          "excerpt": "important not to just run one fuzzer but to run many different types of buzzers because these fuzzers they all they have like different domains so if you run you shouldn't just run sql lancer you should also run sql smith because sql smith will find bug sequel answer will not and sql answer will find bugs sql smith so uh it's like it's very powerful you should run as many as possible i think um also kind of a lesson we learned is that",
+          "excerpt_is_verbatim": true,
+          "technique_ids": [],
+          "frame_checked": "The slide makes the general case for running many kinds of fuzzer; SQLancer is named only in what is said."
+        },
+        {
+          "id": "M5",
+          "source": "captions",
+          "at_seconds": 2451,
+          "timestamp": "40:51",
+          "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs&t=2451s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": "sql answer",
+          "excerpt": "just file github issues and so the the fuzzer of docs robot was born and this is what runs now in our ci it currently runs sql answer and sql smith we also run oss fuzz but that's managed by google that's not by fuzzer ducks and what it does is if a bug is found it creates a reproducible test case based on the output of the fuzzer it does test case reduction and then it files an",
+          "excerpt_is_verbatim": true,
+          "technique_ids": []
+        },
+        {
+          "id": "M6",
+          "source": "frame",
+          "at_seconds": 2455,
+          "timestamp": "40:55",
+          "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs&t=2455s",
+          "matched": [
+            "sqlancer"
+          ],
+          "heard_as": null,
+          "excerpt": null,
+          "excerpt_is_verbatim": false,
+          "note": "The robot that runs fuzzers in DuckDB's CI, and what it was running at the time: SQLancer and SQLsmith.",
+          "image": "/assets/images/impact/talks/BgC79Zt2fPs-2455.jpg",
+          "technique_ids": []
+        }
+      ],
+      "relationship": {
+        "roles": [
+          "recognition",
+          "reports_adoption",
+          "compares"
+        ],
+        "summary": "DuckDB's testing, from one of its authors: the keynote credits SQLancer with around eighty bugs that the SQLite and Postgres test suites DuckDB had borrowed all missed, argues for running it beside SQLsmith because each finds what the other does not, and says the robot in DuckDB's CI runs both.",
+        "mention_ids": [
+          "M1",
+          "M2",
+          "M3",
+          "M4",
+          "M5",
+          "M6"
+        ],
+        "decided_by": "curator"
+      },
+      "provenance": {
+        "collector": "talks",
+        "collector_version": "1.0.0",
+        "first_seen": "2026-09-11T17:08:01Z",
+        "last_verified": "2026-09-12T02:04:10Z",
+        "source_url": "https://www.youtube.com/watch?v=BgC79Zt2fPs",
+        "source_type": "video"
+      }
+    },
+    {
+      "id": "talk:645a47ac1426",
+      "title": "Fuzzing: Practical approaches in ClickHouse",
+      "url": "https://presentations.clickhouse.com/2021-cpp-siberia/index.html",
+      "video_id": null,
+      "speakers": [
+        "Alexey Milovidov"
+      ],
+      "publisher": "ClickHouse",
+      "event": "C++ Siberia 2021",
+      "year": 2021,
+      "duration_seconds": null,
+      "related_dbms": [
+        "clickhouse"
+      ],
+      "related_techniques": [],
+      "sources": [
+        {
+          "kind": "slides",
+          "status": "extracted",
+          "url": "https://presentations.clickhouse.com/2021-cpp-siberia/index.html",
+          "retrieved_at": "2026-09-12T02:04:10Z",
+          "characters": 9192,
+          "content_sha256": "sha256:818b28d03b258f5480ed173feebf3adb1a485053aff1e318565a1404b0cd8544"
+        }
+      ],
+      "mentions": [
+        {
+          "id": "M1",
+          "source": "slides",
+          "at_seconds": null,
+          "timestamp": null,
+          "url": "https://presentations.clickhouse.com/2021-cpp-siberia/index.html",
+          "matched": [
+            "sqlancer",
+            "author"
+          ],
+          "heard_as": null,
+          "excerpt": "SQLancer — logical SQL fuzzer Developed by Manuel Rigger at ETH Zurich.",
+          "excerpt_is_verbatim": true,
+          "technique_ids": [],
+          "image": "/assets/images/impact/talks/clickhouse-2021-cpp-siberia.jpg"
+        }
+      ],
+      "relationship": {
+        "roles": [
+          "reports_adoption"
+        ],
+        "summary": "A tour of everything ClickHouse fuzzes with, giving SQLancer a section of its own: who wrote it, who brought it into ClickHouse, and what it does.",
+        "mention_ids": [
+          "M1"
+        ],
+        "decided_by": "curator"
+      },
+      "provenance": {
+        "collector": "talks",
+        "collector_version": "1.0.0",
+        "first_seen": "2026-09-11T02:04:04Z",
+        "last_verified": "2026-09-12T02:04:10Z",
+        "source_url": "https://presentations.clickhouse.com/2021-cpp-siberia/index.html",
+        "source_type": "video"
+      }
+    },
+    {
+      "id": "talk:youtube:wHo-VtzTHx0",
+      "title": "CockroachDB's Query Optimizer (Rebecca Taft, Cockroach Labs)",
+      "url": "https://www.youtube.com/watch?v=wHo-VtzTHx0",
+      "video_id": "wHo-VtzTHx0",
+      "speakers": [
+        "Rebecca Taft"
+      ],
+      "publisher": "CMU Database Group",
+      "event": "CMU Quarantine Tech Talks 2020",
+      "year": 2020,
+      "duration_seconds": 3852,
+      "related_dbms": [
+        "cockroachdb"
+      ],
+      "related_techniques": [],
+      "sources": [
+        {
+          "kind": "captions",
+          "status": "extracted",
+          "url": "https://www.youtube.com/watch?v=wHo-VtzTHx0",
+          "retrieved_at": "2026-09-12T02:00:00Z",
+          "segments": 626,
+          "characters": 60535,
+          "note": "Windows around candidate moments, read out of YouTube's transcript panel in a browser. Not the whole talk: only the segments a loose prefilter flagged, so the matcher decides on text a person can check."
+        }
+      ],
+      "mentions": [
+        {
+          "id": "M1",
+          "source": "captions",
+          "at_seconds": 3507,
+          "timestamp": "58:27",
+          "url": "https://www.youtube.com/watch?v=wHo-VtzTHx0&t=3507s",
+          "matched": [
+            "sqlancer",
+            "author"
+          ],
+          "heard_as": "sql lanza",
+          "excerpt": "errors like if if it's gonna cause an internal error or crash or something like that lasting correctness but um yeah actually uh manuel rigger uh he's the guy that was doing more kind of logically yeah exactly he he did a bunch of experiments with cockroaches and he opened a bunch of github issues for us which is which is pretty great sql lanza is awesome yeah i'm trying to get trying to get it running in our system um all right so then it's the in general what like what's the complexity of the queries that you're seeing i i understand that like that might be",
+          "excerpt_is_verbatim": true,
+          "technique_ids": [],
+          "frame_checked": "A slide from the talk proper is still on screen; this mention is in the questions afterwards."
+        }
+      ],
+      "relationship": {
+        "roles": [
+          "recognition"
+        ],
+        "summary": "CockroachDB's optimizer, from an engineer who builds it; asked in the questions about random testing, she separates SQLsmith's crashes from the logical bugs Rigger was after, credits him with a batch of GitHub issues against CockroachDB, and says she is trying to get SQLancer running in their own system.",
+        "mention_ids": [
+          "M1"
+        ],
+        "decided_by": "curator"
+      },
+      "provenance": {
+        "collector": "talks",
+        "collector_version": "1.0.0",
+        "first_seen": "2026-09-11T17:04:25Z",
+        "last_verified": "2026-09-12T02:04:10Z",
+        "source_url": "https://www.youtube.com/watch?v=wHo-VtzTHx0",
+        "source_type": "video"
+      }
+    }
+  ]
+}
diff --git a/_data/impact/techniques.json b/_data/impact/techniques.json
new file mode 100644
index 0000000..0cee6df
--- /dev/null
+++ b/_data/impact/techniques.json
@@ -0,0 +1,502 @@
+{
+  "schema_version": "1.0.0",
+  "taxonomy_version": "taxonomy-v2",
+  "finders": [
+    {
+      "id": "sqlancer",
+      "display_name": "SQLancer",
+      "names": [
+        "SQLancer",
+        "sqlancer"
+      ],
+      "umbrella_member": true,
+      "repository": "https://github.com/sqlancer/sqlancer",
+      "paper": null,
+      "description": "The main SQLancer tool: hand-written SQL generators for a large set of database systems, combined with the test oracles listed above."
+    },
+    {
+      "id": "sqlancer_pp",
+      "display_name": "SQLancer++",
+      "names": [
+        "SQLancer++",
+        "SQLancerPlusPlus",
+        "SQLancer PP"
+      ],
+      "umbrella_member": true,
+      "repository": "https://github.com/suyZhong/SQLancerPlusPlus",
+      "paper": {
+        "title": "Scaling Automated Database System Testing",
+        "venue": "ASPLOS 2026",
+        "year": 2026,
+        "url": "https://doi.org/10.1145/3779212.3790215",
+        "doi": "10.1145/3779212.3790215",
+        "is_citation_seed": true
+      },
+      "description": "Scales automated testing to database systems for which no bespoke SQLancer implementation exists, by adaptively discovering which SQL features the system under test supports."
+    },
+    {
+      "id": "shqvel",
+      "display_name": "ShQveL",
+      "names": [
+        "ShQveL",
+        "Shqvel"
+      ],
+      "umbrella_member": true,
+      "repository": "https://github.com/suyZhong/SQLancerPlusPlus",
+      "paper": {
+        "title": "Automated Database Testing via LLM-Synthesized SQL Features",
+        "venue": null,
+        "year": 2025,
+        "url": "https://arxiv.org/abs/2505.02012",
+        "doi": "10.48550/arxiv.2505.02012",
+        "arxiv_id": "2505.02012",
+        "is_citation_seed": true
+      },
+      "description": "Augments SQLancer-style testing with SQL feature fragments synthesised by a large language model, reaching features the hand-written generators do not cover."
+    }
+  ],
+  "sqlancer_techniques": [
+    {
+      "id": "pqs",
+      "names": [
+        "PQS",
+        "Pivoted Query Synthesis",
+        "pivoted query synthesis"
+      ],
+      "display_name": "Pivoted Query Synthesis (PQS)",
+      "kind": "test_oracle",
+      "sqlancer_originated": true,
+      "introduced_year": 2020,
+      "origin_paper": {
+        "title": "Testing Database Engines via Pivoted Query Synthesis",
+        "venue": "OSDI 2020",
+        "year": 2020,
+        "url": "https://www.usenix.org/system/files/osdi20-rigger.pdf",
+        "doi": null,
+        "arxiv_id": "2001.04174",
+        "s2_paper_id": null,
+        "is_citation_seed": true
+      },
+      "ambiguous_acronym": true,
+      "required_context_terms": [
+        "sqlancer",
+        "dbms",
+        "database",
+        "sql",
+        "query",
+        "oracle",
+        "logic bug",
+        "metamorphic",
+        "fuzz",
+        "test oracle",
+        "optimizer",
+        "optimiser",
+        "数据库",
+        "查询",
+        "等价验证",
+        "优化器",
+        "聚合",
+        "结果集"
+      ],
+      "negative_context_terms": [
+        "photoelectron",
+        "quantum",
+        "parallel quantum"
+      ],
+      "description": "Selects a random row, the pivot row, and generates a query that is guaranteed to fetch it. If the row is missing from the result set, a bug has been found."
+    },
+    {
+      "id": "norec",
+      "names": [
+        "NoREC",
+        "Non-optimizing Reference Engine Construction",
+        "Non-Optimizing Reference Engine Construction",
+        "non-optimizing reference engine construction"
+      ],
+      "display_name": "Non-optimizing Reference Engine Construction (NoREC)",
+      "kind": "test_oracle",
+      "sqlancer_originated": true,
+      "introduced_year": 2020,
+      "origin_paper": {
+        "title": "Detecting optimization bugs in database engines via non-optimizing reference engine construction",
+        "venue": "ESEC/FSE 2020",
+        "year": 2020,
+        "url": "https://arxiv.org/abs/2007.08292",
+        "doi": "10.1145/3368089.3409710",
+        "arxiv_id": "2007.08292",
+        "s2_paper_id": "55a4e4a42cd86fac55491d089d07d04f09dcf957",
+        "is_citation_seed": true
+      },
+      "ambiguous_acronym": false,
+      "required_context_terms": [],
+      "negative_context_terms": [],
+      "description": "Translates a query that the DBMS is likely to optimise into one for which hardly any optimisation applies, then compares the two result sets. A mismatch indicates an optimisation bug."
+    },
+    {
+      "id": "tlp",
+      "names": [
+        "TLP",
+        "Ternary Logic Partitioning",
+        "ternary logic partitioning",
+        "query partitioning"
+      ],
+      "display_name": "Ternary Logic Partitioning (TLP)",
+      "kind": "test_oracle",
+      "sqlancer_originated": true,
+      "introduced_year": 2020,
+      "origin_paper": {
+        "title": "Finding bugs in database systems via query partitioning",
+        "venue": "OOPSLA 2020",
+        "year": 2020,
+        "url": "https://dl.acm.org/doi/pdf/10.1145/3428279",
+        "doi": "10.1145/3428279",
+        "arxiv_id": null,
+        "s2_paper_id": "d69216947188267d5537ee0f0501b2d960ecc457",
+        "is_citation_seed": true
+      },
+      "ambiguous_acronym": true,
+      "required_context_terms": [
+        "sqlancer",
+        "dbms",
+        "database",
+        "sql",
+        "query",
+        "oracle",
+        "logic bug",
+        "metamorphic",
+        "fuzz",
+        "test oracle",
+        "optimizer",
+        "optimiser",
+        "数据库",
+        "查询",
+        "等价验证",
+        "优化器",
+        "聚合",
+        "结果集"
+      ],
+      "negative_context_terms": [
+        "thread level parallelism",
+        "thread-level parallelism",
+        "transport layer",
+        "tlp-stat",
+        "laptop",
+        "power management"
+      ],
+      "description": "Partitions a query into three partitioning queries whose combined results must match the original query's result set. Unlike NoREC and PQS it also reaches advanced features such as aggregate functions."
+    },
+    {
+      "id": "qpg",
+      "names": [
+        "QPG",
+        "Query Plan Guidance",
+        "query plan guidance"
+      ],
+      "display_name": "Query Plan Guidance (QPG)",
+      "kind": "test_input_generation",
+      "sqlancer_originated": true,
+      "introduced_year": 2023,
+      "origin_paper": {
+        "title": "Testing Database Systems via Query Plan Guidance",
+        "venue": "ICSE 2023",
+        "year": 2023,
+        "url": "https://arxiv.org/pdf/2312.17510",
+        "doi": "10.1109/icse48619.2023.00174",
+        "arxiv_id": "2312.17510",
+        "s2_paper_id": null,
+        "is_citation_seed": true,
+        "also_titled": [
+          "Testing Database Engines via Query Plan Guidance"
+        ]
+      },
+      "ambiguous_acronym": true,
+      "required_context_terms": [
+        "sqlancer",
+        "dbms",
+        "database",
+        "sql",
+        "query",
+        "oracle",
+        "logic bug",
+        "metamorphic",
+        "fuzz",
+        "test oracle",
+        "optimizer",
+        "optimiser",
+        "数据库",
+        "查询",
+        "等价验证",
+        "优化器",
+        "聚合",
+        "结果集"
+      ],
+      "negative_context_terms": [],
+      "description": "Feedback-guided test input generation that mutates the database state when no new query plans have been observed, on the insight that query plans capture whether interesting behaviour is being exercised."
+    },
+    {
+      "id": "cert",
+      "names": [
+        "CERT",
+        "Cardinality Estimation Restriction Testing",
+        "cardinality estimation restriction testing"
+      ],
+      "display_name": "Cardinality Estimation Restriction Testing (CERT)",
+      "kind": "test_oracle",
+      "sqlancer_originated": true,
+      "introduced_year": 2024,
+      "origin_paper": {
+        "title": "Cardinality Estimation Restriction Testing: Finding Performance Issues in Database Systems",
+        "venue": "ICSE 2024",
+        "year": 2024,
+        "url": "https://arxiv.org/pdf/2306.00355",
+        "doi": "10.1145/3597503.3639076",
+        "arxiv_id": "2306.00355",
+        "s2_paper_id": null,
+        "is_citation_seed": true,
+        "also_titled": [
+          "CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation"
+        ]
+      },
+      "ambiguous_acronym": true,
+      "required_context_terms": [
+        "sqlancer",
+        "dbms",
+        "database",
+        "sql",
+        "query",
+        "oracle",
+        "logic bug",
+        "metamorphic",
+        "fuzz",
+        "test oracle",
+        "optimizer",
+        "optimiser",
+        "cardinality",
+        "performance issue",
+        "数据库",
+        "查询",
+        "等价验证",
+        "优化器",
+        "聚合",
+        "结果集"
+      ],
+      "negative_context_terms": [
+        "certificate",
+        "cert.org",
+        "us-cert",
+        "x.509",
+        "x509",
+        "ssl",
+        "tls",
+        "https",
+        "ca cert",
+        "root cert",
+        "crl",
+        "keystore",
+        "truststore",
+        "openssl",
+        "self-signed",
+        "cert-manager",
+        "auth_method",
+        "sslmode",
+        "证书"
+      ],
+      "description": "Derives a more restrictive query whose estimated cardinality must not exceed the original's. A violation points at a performance issue. The only SQLancer oracle aimed at performance rather than correctness."
+    },
+    {
+      "id": "dqp",
+      "names": [
+        "DQP",
+        "Differential Query Plan",
+        "differential query plans"
+      ],
+      "display_name": "Differential Query Plans (DQP)",
+      "kind": "test_oracle",
+      "sqlancer_originated": true,
+      "introduced_year": 2024,
+      "origin_paper": {
+        "title": "Keep It Simple: Testing Databases via Differential Query Plans",
+        "venue": "SIGMOD 2024",
+        "year": 2024,
+        "url": "https://dl.acm.org/doi/pdf/10.1145/3654991",
+        "doi": "10.1145/3654991",
+        "arxiv_id": null,
+        "s2_paper_id": "63b32cbfc5e6ab0043715a9adcf6ad1944c9ba49",
+        "is_citation_seed": true
+      },
+      "ambiguous_acronym": true,
+      "required_context_terms": [
+        "sqlancer",
+        "dbms",
+        "database",
+        "sql",
+        "query",
+        "oracle",
+        "logic bug",
+        "metamorphic",
+        "fuzz",
+        "test oracle",
+        "optimizer",
+        "optimiser",
+        "数据库",
+        "查询",
+        "等价验证",
+        "优化器",
+        "聚合",
+        "结果集"
+      ],
+      "negative_context_terms": [],
+      "description": "Forces the DBMS to execute the same query using different query plans and compares the results, which must agree."
+    },
+    {
+      "id": "coddtest",
+      "names": [
+        "CODDTest",
+        "Constant Optimization Driven Database System Testing",
+        "constant optimization driven database system testing"
+      ],
+      "display_name": "Constant-Optimization-Driven Testing (CODDTest)",
+      "kind": "test_oracle",
+      "sqlancer_originated": true,
+      "introduced_year": 2025,
+      "origin_paper": {
+        "title": "Constant Optimization Driven Database System Testing",
+        "venue": "SIGMOD 2025",
+        "year": 2025,
+        "url": "https://doi.org/10.1145/3709674",
+        "doi": "10.1145/3709674",
+        "arxiv_id": null,
+        "s2_paper_id": null,
+        "is_citation_seed": true
+      },
+      "ambiguous_acronym": false,
+      "required_context_terms": [],
+      "negative_context_terms": [],
+      "description": "Borrows constant folding from compiler testing: subexpressions are replaced by their computed constants and the result must not change."
+    }
+  ],
+  "project_authors": [
+    "Rigger"
+  ],
+  "project_contributors": [
+    {
+      "name": "Manuel Rigger",
+      "github": "mrigger",
+      "role": "Project author"
+    },
+    {
+      "name": "Zhenglin Li",
+      "github": "ZhengLin-Li",
+      "role": "Google Summer of Code 2023 contributor"
+    },
+    {
+      "name": "Zhenglin Li",
+      "github": "zhenglin-charlie-li",
+      "role": "Google Summer of Code 2023 contributor"
+    },
+    {
+      "name": "Yutan Yang",
+      "github": "ColinYoungTaro",
+      "role": "Google Summer of Code 2023 contributor"
+    }
+  ],
+  "project_publications": [
+    {
+      "title": "Validating Database System Isolation Level Implementations with Version Certificate Recovery",
+      "doi": "10.1145/3627703.3650080",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "year": 2024,
+      "venue": "EuroSys 2024",
+      "reason": "Project publication: work by the SQLancer authors applying the PQS idea to isolation-level validation."
+    },
+    {
+      "title": "Bringing Order to Practical Validation of Database Isolation Levels",
+      "doi": "10.1145/3797871",
+      "arxiv_id": null,
+      "s2_paper_id": null,
+      "year": 2026,
+      "venue": "ACM Transactions on Computer Systems",
+      "reason": "Project publication: the journal version of the EuroSys 2024 isolation-level paper by the same authors."
+    }
+  ],
+  "excluded_techniques": [
+    {
+      "id": "dqe",
+      "names": [
+        "DQE",
+        "Differential Query Execution",
+        "differential query execution"
+      ],
+      "display_name": "Differential Query Execution (DQE)",
+      "reason": "DQE was developed independently of SQLancer by Song et al. and published at ICSE 2023; its SQLancer implementation arrived later as a contributed pull request. Implementing a technique is not originating it, so DQE-attributed bugs are not counted as SQLancer findings and papers citing the DQE publication are not treated as citing SQLancer.",
+      "evidence": [
+        {
+          "source_url": "https://github.com/sqlancer/sqlancer/blob/main/README.md",
+          "source_type": "documentation",
+          "excerpt": "| Differential Query Execution (DQE)                              | ICSE 2023     | [Paper](https://ieeexplore.ieee.org/document/10172736) [Code](https://github.com/sqlancer/sqlancer/pull/1251)",
+          "excerpt_is_verbatim": true,
+          "note": "SQLancer's own README credits DQE to an external ICSE 2023 publication and links its implementation as a pull request contributed to SQLancer.",
+          "content_sha256": "sha256:e5f5180fefc22551e27d9fb83cb9337cad8625677ee6185fe8c312715e154042",
+          "retrieved_at": "2026-09-06T00:00:00Z",
+          "first_seen": "2026-09-06T00:00:00Z",
+          "last_verified": "2026-09-06T00:00:00Z"
+        }
+      ]
+    },
+    {
+      "id": "eet",
+      "names": [
+        "EET",
+        "Equivalent Expression Transformation"
+      ],
+      "display_name": "Equivalent Expression Transformation (EET)",
+      "reason": "EET was introduced by Jiang and Su at OSDI 2024, independently of SQLancer. SQLancer later gained an EET oracle, but implementing a technique is not originating it, so bugs found with EET are not counted as SQLancer findings.",
+      "evidence": [
+        {
+          "source_url": "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/common/oracle/EETOracle.java",
+          "source_type": "github_code",
+          "excerpt": "EET (Equivalent Expression Transformation) oracle, based on \"Detecting Logic Bugs in Database Engines via Equivalent\n * Expression Transformation\" (Jiang &amp; Su, OSDI'24).",
+          "excerpt_is_verbatim": true,
+          "note": "Class-level Javadoc of SQLancer's own EET oracle, crediting the technique to an external publication.",
+          "content_sha256": "sha256:e61be0c83848f4b428ece8faa2aa9b0b006559d13c708ebb0285c4fe8254648d",
+          "retrieved_at": "2026-09-06T00:00:00Z",
+          "first_seen": "2026-09-06T00:00:00Z",
+          "last_verified": "2026-09-06T00:00:00Z"
+        }
+      ]
+    }
+  ],
+  "bug_symptoms": [
+    {
+      "id": "logic",
+      "display_name": "Logic bug",
+      "description": "The DBMS returned an incorrect result set."
+    },
+    {
+      "id": "error",
+      "display_name": "Unexpected error",
+      "description": "The DBMS raised an internal or otherwise unexpected error."
+    },
+    {
+      "id": "crash",
+      "display_name": "Crash",
+      "description": "The DBMS process terminated abnormally."
+    },
+    {
+      "id": "hang",
+      "display_name": "Hang",
+      "description": "The DBMS stopped making progress."
+    },
+    {
+      "id": "performance",
+      "display_name": "Performance issue",
+      "description": "The DBMS produced a correct result far less efficiently than it could."
+    },
+    {
+      "id": "unknown",
+      "display_name": "Unclassified",
+      "description": "The report does not record how the bug manifested."
+    }
+  ]
+}
diff --git a/_data/navigation.yml b/_data/navigation.yml
index f37ba82..da28b25 100644
--- a/_data/navigation.yml
+++ b/_data/navigation.yml
@@ -1,4 +1,8 @@
 main:
+  - title: "Impact"
+    url: /impact/
+  - title: "Supported databases"
+    url: /supported-databases/
   - title: "Posts"
     url: /posts/
 #  - title: "Categories"
diff --git a/_data/papers/paper_arxiv_2105_10016.json b/_data/papers/paper_arxiv_2105_10016.json
new file mode 100644
index 0000000..1f78b9a
--- /dev/null
+++ b/_data/papers/paper_arxiv_2105_10016.json
@@ -0,0 +1,883 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:50Z",
+  "paper": {
+    "id": "paper:arxiv:2105.10016",
+    "title": "Testing DBMS Performance with Mutations",
+    "authors": [
+      "Xinyu Liu",
+      "Qi Zhou",
+      "Joy Arulraj",
+      "A. Orso"
+    ],
+    "year": 2021,
+    "venue": "arXiv (Cornell University)",
+    "doi": null,
+    "arxiv_id": "2105.10016",
+    "s2_paper_id": "4ce512cdd51d4926d9353aca8a588a2414a0d305",
+    "url": "https://arxiv.org/abs/2105.10016",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2105.10016",
+      "retrieved_at": "2026-09-09T01:33:50Z",
+      "chars": 75784,
+      "content_sha256": "sha256:60de6be3f2d44285827f5d1a87fe351035b8c71edce1f5f03db519ba9af9ff3a"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "1992. Database Language SQL. http://www.contrib.andrew.cmu.edu/~shadow/sql /sql1992.txt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2015. AFL: American Fuzzy Lop. http://lcamtuf.coredump.cx/afl/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2016. OSS-Fuzz: Continuous Fuzzing for Open Source Software. https: //github.com/google/oss-fuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2016. SQLSmith. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2020. Apache Calcite. https://calcite.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2020. CockroachDB. https://www.cockroachlabs.com/docs/releases/v20.2.0.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2020. SQLAlchemy. https://www.sqlalchemy.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2021. CockroachDB Performance Bug Reports. https://github.com/cockroachdb /cockroach/issues?q=performance.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2021. PostgreSQL Performance Bug Reports. https://www.postgresql.org/searc h/?m=1&q=performance&l=8&d=-1&s=r.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2021. SCOTT schema. https://www.orafaq.com/wiki/SCOTT.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2021. Supplementary material.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. 2007. A genetic approach for random testing of database systems. In VLDB. 1243–1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. 2006. Generating Queries with Cardinality Constraints for DBMS Testing. In TKDE. 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Bikash Chandra, Bhupesh Chawda, Biplab Kar, KV Maheshwara Reddy, Shetal Shah, and S Sudarshan. 2015. Data generation for testing and grading SQL queries. InVLDB Journal. 731–755.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "CL Philip Chen and Chun-Yang Zhang. 2014. Data-intensive applications, challenges, techniques and technologies: A survey on Big Data. In Information sciences. 314–347.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Tsong Y Chen, Shing C Cheung, and Shiu Ming Yiu. 2020. Metamorphic testing: a new approach for generating next test cases. In arXiv preprint arXiv:2002.12543.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Shumo Chu, Chenglong Wang, Konstantin Weitz, and Alvin Cheung. 2017. Cosette: An Automated Prover for SQL.. In CIDR.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Hicham G Elmongui, Vivek Narasayya, and Ravishankar Ramamurthy. 2009. A framework for testing query transformation rules. In SIGMOD. 257–268.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Leo Giakoumakis and César A Galindo-Legaria. 2008. Testing SQL Server’s Query Optimizer: Challenges, Techniques and Experiences. In Data Engineering Bulletin. 36–43.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Goetz Graefe. 1993. Query Evaluation Techniques for Large Databases. In CSUR. 73–169.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Goetz Graefe. 1995. The cascades framework for query optimization. In Data Engineering Bulletin. 19–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Zhongxian Gu, Mohamed A. Soliman, and Florian M. Waas. 2015. Testing the accuracy of query optimizers. In DBTest. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Toshihide Ibaraki and Tiko Kameda. 1984. On the optimal nesting order for computing n-relational joins. In TODS. 482–502.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. Apollo: Automatic detection and diagnosis of performance regressions in database systems. InVLDB. 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "R Kearns, Stephen Shead, and Alan Fekete. 1997. A teaching system for SQL. In ACSE. 224–231.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "D. Lee, S. K. Cha, and A. H. Lee. 2012. A Performance Anomaly Detection and Analysis Framework for DBMS Development. In TKDE. 1345–1360.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2015. How good are query optimizers, really?. In VLDB. 204–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Zhan Li, Olga Papaemmanouil, and Mitch Cherniack. 2016. OptMark: A Toolkit for Benchmarking Query Optimizers. In CIKM. 2155–2160.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "William M McKeeman. 1998. Differential testing for software. In Digital Technical Journal. 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Rasha Osman and William J. Knottenbelt. 2012. Database System Performance Evaluation Models: A Survey. In Performance Evaluation. 471–493.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Hamid Pirahesh, Joseph M Hellerstein, and Waqar Hasan. 1992. Extensible/rule based query rewrite optimization in Starburst. In Sigmod Record. 39–48.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Jaroslav Pokorn `y. 2015. Database technologies in the world of big data. In CompSysTech. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Raghu Ramakrishnan and Johannes Gehrke. 2003. Database Management Systems.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Kim-Thomas Rehmann, Changyun Seo, Dongwon Hwang, Binh Than Truong, Alexander Boehm, and Dong Hun Lee. 2016. Performance Monitoring in SAPHANA’s Continuous Integration Process. In PER. 43–52.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In FSE.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. In OOPSLA.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In OSDI.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Shetal Shah, S Sudarshan, Suhas Kajbaje, Sandeep Patidar, Bhanu Pratap Gupta, and Devang Vira. 2011. Generating test data for killing SQL mutants: A constraintbased approach. In ICDE. 1175–1186.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Hitesh Kumar Sharma, Mr SC Nelson, et al .2017. Explain Plan and SQL Trace the Two Approaches for RDBMS Tuning. In Database Systems Journal. 31–39.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Donald R Slutz. 1998. Massive Stochastic Testing of SQL. In VLDB. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Florian Waas and César Galindo-Legaria. 2000. Counting, Enumerating, and Sampling of Execution Plans in a Cost-Based Query Optimizer. In SIGMOD. 499–509.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Gerhard Weikum, Axel Moenkeberg, Christof Hasse, and Peter Zabback. 2002. Self-Tuning Database Technology and Information Services: From Wishful Thinking to Viable Engineering. In VLDB. 20–31.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Khaled Yagoub, Peter Belknap, Benoit Dageville, Karl Dias, Shantanu Joshi, and Hailing Yu. 2008. Oracle’s SQL Performance Analyzer.. In Data Engineering Bulletin. 51–58.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Jiaqi Yan, Qiuye Jin, Shrainik Jain, Stratis D Viglas, and Allison Lee. 2018. Snowtrail: Testing with Production Queries on a Cloud Database. In DBTEST.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In CCS. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Qi Zhou, Joy Arulraj, Shamkant Navathe, William Harris, and Dong Xu. 2019. Automated verification of query equivalence using satisfiability modulo theories. InVLDB. 1276–1288. 13",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 35,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In FSE.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. In OOPSLA.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In OSDI.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "They leverage tools such as SQLSMITH [4] and SQLancer [35– 37] to discover crash-inducing or logic bugs in DBMSs.",
+      "context_before": "query pairs that are likely to uncover performance bugs. We introduce feedback mechanisms for improving the efficacy and computational efficiency of the tool. We evaluate AMOEBA on two widely-used DBMSs, namely PostgreSQL andCockroachDB. AMOEBA has so far discovered 39 previously-unknown performance bugs, among which developers have already confirmed 14 bugs and fixed 4 bugs. 1 Introduction Database management systems (DBMSs) play a critical role in modern data-intensive applications [ 15,32]. For this reason, developers extensively test these systems to improve their reliability and accuracy.",
+      "context_after": "However, the same level of scrutiny has not been applied to performance bugs that increase the time taken by the DBMS to process certain queries. Delayed responses from the DBMS ultimately affect user experience [30, 42]. CHALLENGES .To retrieve the results for a given SQL query, the DBMS invokes a pipeline of complex components ( e.g., query optimizer, execution engine) [ 20,33]. The overall performance of the DBMS may be reduced by sub-optimal decisions taken by any of these components and the complex interactions among them [ 8,9]. Therefore performance testing on individual components of",
+      "section": null,
+      "page": 1,
+      "char_offset": 1804,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "Recently, researchers have proposed Ternary Logic Partitioning (TLP), a technique that applies metamorphic testing for detecting logic bugs in the DBMS [ 36].",
+      "context_before": "require a developer-provided, pre-determined baseline, it is only able to detect regressions. Thus, it cannot detect a large subset of performance bugs that were always present in the system because, for these bugs, the query execution time on two versions of the DBMS would not differ. Furthermore, this technique also focuses on structurally simple queries that are tailored for uncovering regressions. While it is also possible to detect performance bugs using metamorphic testing, it remains unclear on how to construct an effective metamorphic relation geared towards detecting performance bugs.",
+      "context_after": "To do this, it transforms a base query into a semantically equivalent one (called mutant query) by intentionally increasing the complexity of the base query’s structure, and reports a logic bug if their result sets differ on the same database. By design, the DBMS is likely to take more time to process the mutant query in comparison to the base query. As a result, it is difficult to construct a metamorphic relation that is tailored for detecting performance bugs based on equivalent queries generated by TLP. In addition, TLP targets on comparatively simple queries that are tailored for uncoveri",
+      "section": null,
+      "page": 1,
+      "char_offset": 4749,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "As a result, it is difficult to construct a metamorphic relation that is tailored for detecting performance bugs based on equivalent queries generated by TLP.",
+      "context_before": "o construct an effective metamorphic relation geared towards detecting performance bugs. Recently, researchers have proposed Ternary Logic Partitioning (TLP), a technique that applies metamorphic testing for detecting logic bugs in the DBMS [ 36]. To do this, it transforms a base query into a semantically equivalent one (called mutant query) by intentionally increasing the complexity of the base query’s structure, and reports a logic bug if their result sets differ on the same database. By design, the DBMS is likely to take more time to process the mutant query in comparison to the base query.",
+      "context_after": "In addition, TLP targets on comparatively simple queries that are tailored for uncovering logic bugs. OURAPPROACH .In this paper, we present AMOEBA, a novel and principled tool for discovering a wider variety of performance bugs in DBMSs. AMOEBA addresses the challenges listed above along three dimensions. First, it constructs a cross-referencing oracle by comparing the runtime performance of semantically equivalent queries (i.e., queries that always return the same result sets for all possible input tables) [ 17,46]. When the target DBMS (only a single version is needed) exhibits a significa",
+      "section": null,
+      "page": 1,
+      "char_offset": 5261,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "In addition, TLP targets on comparatively simple queries that are tailored for uncovering logic bugs.",
+      "context_before": "a technique that applies metamorphic testing for detecting logic bugs in the DBMS [ 36]. To do this, it transforms a base query into a semantically equivalent one (called mutant query) by intentionally increasing the complexity of the base query’s structure, and reports a logic bug if their result sets differ on the same database. By design, the DBMS is likely to take more time to process the mutant query in comparison to the base query. As a result, it is difficult to construct a metamorphic relation that is tailored for detecting performance bugs based on equivalent queries generated by TLP.",
+      "context_after": "OURAPPROACH .In this paper, we present AMOEBA, a novel and principled tool for discovering a wider variety of performance bugs in DBMSs. AMOEBA addresses the challenges listed above along three dimensions. First, it constructs a cross-referencing oracle by comparing the runtime performance of semantically equivalent queries (i.e., queries that always return the same result sets for all possible input tables) [ 17,46]. When the target DBMS (only a single version is needed) exhibits a significant difference in execution time on a pair of semantically equivalent queries, the likely root cause 1",
+      "section": null,
+      "page": 1,
+      "char_offset": 5420,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We compare AMOEBA against two other sources of equivalent queries that could be used for detecting performance bugs: (1) a manually-written test suite in a widely-used query optimization framework, and (2) the TLP.",
+      "context_before": "ake the following contributions: •We present a technique for finding performance bugs using a novel application of query equivalence in generating a cross-referencing oracle (§2). •We introduce two types of query mutations that preserve the semantics of queries: (1) structural mutations, and (2) expression mutations (§6). •We present a feedback mechanism for improving the efficacy and computational efficiency of the tool (§7). •We implemented this technique in an extensible tool called AMOEBA. In our evaluation, AMOEBA discovered 39 previously-unknown performance bugs in two widely-used DBMSs.",
+      "context_after": "Our empirical analysis shows how AMOEBA ’s generated equivalent queries are more likely to detect performance bugs, which also highlights opportunities for improving and testing future versions of these DBMSs (§8). 2 Motivation In this section, we highlight the importance of detecting performance bugs in DBMSs using a motivating example. EXAMPLE .Listing 1 and Listing 2 show a pair of semantically equivalent queries based on the SCOTT schema [ 10]: Q1 and Q2. Although Q1 and Q2 are equivalent, Q1 runs 1444 ×slower than Q2 on the same database in CockroachDB [6] (v20.2.0-alpha).With the first",
+      "section": null,
+      "page": 2,
+      "char_offset": 8408,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[ 36,38]",
+      "technique": "tlp",
+      "sentence": "Query equivalence is a well-studied topic and is used in many applications: (1) testing correctness of DBMS and SQL queries [ 36,38], (2) educating developers [ 25], and (3) automatically grading student assignments [ 14].",
+      "context_before": "sive and critical to query performance, the CockroachDB developers quickly fixed this performance bug in the UNORDERED DISTINCT operator within a week and plan to fix the bug in the HASH JOIN operator in the future. This example highlights the existence of performance bugs in DBMSs and their significant impact on query performance. 3 Background To better appreciate the internals of AMOEBA, we now provide a brief overview of semantically-equivalent queries. SEMANTIC EQUIVALENCE .Two queries Q1andQ2are semantically equivalent if they always return the same results on any input database instance.",
+      "context_after": "Unlike prior efforts, we seek to leverage semantic equivalence of queries to find performance bugs in DBMSs. QUERY REWRITING .AMOEBA constructs equivalent queries by rewriting them using a set of rules that preserve equivalence [ 21]. A representative rule consists of using values from filters to mutate projection columns. Consider the query shown in Listing 1. For this query, we illustrate how this rule transforms its projection columns while preserving semantic equivalence in Figure 1. The logical query plan of Q1 is shown in Figure 1a. Since the filter clause selects tuples wherein the job",
+      "section": null,
+      "page": 2,
+      "char_offset": 10638,
+      "cited_reference": {
+        "number": 36,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. In OOPSLA.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Notably, SQLancer is a system that constructs equivalent queries for discovering logic bugs in DBMSs.",
+      "context_before": "ine 15). Similar to UPDATE PROBTABLE WITHMUTATORFEEDBACK, this procedure first extracts the SQL entities from the base query and then updates the probability table accordingly. In this way, GENERATOR constructs queries that are more likely to uncover performance bugs. 6 Query Mutator MUTATOR is the next key component of AMOEBA. It takes a base query as input and seeks to generate mutant queries that are semantically equivalent to the base query. QUERY EQUIVALENCE .While the query equivalence property has many applications [ 14,25,38], it has never been utilized for uncovering performance bugs.",
+      "context_after": "In particular, it generates equivalent queries using TLP [ 36]. We defer a comparative analysis of AMOEBA against SQLancer to §8.6. At a high level, MUTATOR is a framework for rewriting queries using a set of semantics-preserving query transformation rules. We next present the internals of MUTATOR with a description of its mutation rules in §6.1 and its algorithm in §6.2. 4 Testing DBMS Performance with Mutations Table 3: List of Illustrative Mutation Rules. Category Rule Idx Transformation Working Example 0 Push aggregate function through JOIN t1 JOIN t2 GROUP BY t1.c →(t1 GROUP BY t1.c) JOI",
+      "section": null,
+      "page": 4,
+      "char_offset": 22944,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "In particular, it generates equivalent queries using TLP [ 36].",
+      "context_before": "es from the base query and then updates the probability table accordingly. In this way, GENERATOR constructs queries that are more likely to uncover performance bugs. 6 Query Mutator MUTATOR is the next key component of AMOEBA. It takes a base query as input and seeks to generate mutant queries that are semantically equivalent to the base query. QUERY EQUIVALENCE .While the query equivalence property has many applications [ 14,25,38], it has never been utilized for uncovering performance bugs. Notably, SQLancer is a system that constructs equivalent queries for discovering logic bugs in DBMSs.",
+      "context_after": "We defer a comparative analysis of AMOEBA against SQLancer to §8.6. At a high level, MUTATOR is a framework for rewriting queries using a set of semantics-preserving query transformation rules. We next present the internals of MUTATOR with a description of its mutation rules in §6.1 and its algorithm in §6.2. 4 Testing DBMS Performance with Mutations Table 3: List of Illustrative Mutation Rules. Category Rule Idx Transformation Working Example 0 Push aggregate function through JOIN t1 JOIN t2 GROUP BY t1.c →(t1 GROUP BY t1.c) JOIN t2 13 Push filter through GROUP BY (t1.c GROUP BY t1.c) WHERE",
+      "section": null,
+      "page": 4,
+      "char_offset": 23046,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We defer a comparative analysis of AMOEBA against SQLancer to §8.",
+      "context_before": "cordingly. In this way, GENERATOR constructs queries that are more likely to uncover performance bugs. 6 Query Mutator MUTATOR is the next key component of AMOEBA. It takes a base query as input and seeks to generate mutant queries that are semantically equivalent to the base query. QUERY EQUIVALENCE .While the query equivalence property has many applications [ 14,25,38], it has never been utilized for uncovering performance bugs. Notably, SQLancer is a system that constructs equivalent queries for discovering logic bugs in DBMSs. In particular, it generates equivalent queries using TLP [ 36].",
+      "context_after": "6. At a high level, MUTATOR is a framework for rewriting queries using a set of semantics-preserving query transformation rules. We next present the internals of MUTATOR with a description of its mutation rules in §6.1 and its algorithm in §6.2. 4 Testing DBMS Performance with Mutations Table 3: List of Illustrative Mutation Rules. Category Rule Idx Transformation Working Example 0 Push aggregate function through JOIN t1 JOIN t2 GROUP BY t1.c →(t1 GROUP BY t1.c) JOIN t2 13 Push filter through GROUP BY (t1.c GROUP BY t1.c) WHERE t1.c > 0 →(t1.c WHERE t1.c > 0) GROUP BY t1.C Structure 16 Push fi",
+      "section": null,
+      "page": 4,
+      "char_offset": 23110,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We now investigate the efficacy of two techniques for obtaining equivalent query pairs for performance bug detection: (1) using SQLancer, an automated DBMS testing technique, to generate random query pairs (§8.",
+      "context_before": "les. In particular, we find that rules that transform expensive operators (e.g.,SORT ,GROUP BY, and JOIN ) are effective in generating queries that trigger performance bugs, which highlights opportunities for improving future versions of the tested DBMSs. 8.6 RQ4 — Comparative Analysis We now present a comparative analysis against two techniques that could also be leveraged to detect performance bugs. The key idea of AMOEBA is to construct semantically equivalent query pairs and use the performance differential within each query pair as a cross-referencing oracle for detecting performance bugs.",
+      "context_after": "6.1). (2) deriving query pairs from the Calcite ’s test suites, which are carefully crafted by DBMS developers (§8.6.2). 8.6.1 Query Pairs from SQLancer SQLancer is the state-ofart tool for discovering logic bugs [ 35–37]. A key technique in SQLancer is using TLP to construct equivalent queries based on the observation that any predicate in SQL evaluates to TRUE ,FALSE, or NULL. Given a base query, TLP constructs a mutant query that is equivalent to the base query in two steps. First, it partitions the base query into three partition queries, wherein each predicate is constructed based on the v",
+      "section": null,
+      "page": 10,
+      "char_offset": 52781,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "1 Query Pairs from SQLancer SQLancer is the state-ofart tool for discovering logic bugs [ 35–37].",
+      "context_before": "techniques that could also be leveraged to detect performance bugs. The key idea of AMOEBA is to construct semantically equivalent query pairs and use the performance differential within each query pair as a cross-referencing oracle for detecting performance bugs. We now investigate the efficacy of two techniques for obtaining equivalent query pairs for performance bug detection: (1) using SQLancer, an automated DBMS testing technique, to generate random query pairs (§8.6.1). (2) deriving query pairs from the Calcite ’s test suites, which are carefully crafted by DBMS developers (§8.6.2). 8.6.",
+      "context_after": "A key technique in SQLancer is using TLP to construct equivalent queries based on the observation that any predicate in SQL evaluates to TRUE ,FALSE, or NULL. Given a base query, TLP constructs a mutant query that is equivalent to the base query in two steps. First, it partitions the base query into three partition queries, wherein each predicate is constructed based on the value of the base query’s predicate. Second, it concatenates these partition queries using the UNION or UNION ALL operator based on the semantics of the base query. Taking the query pair shown below as an example, the boolea",
+      "section": null,
+      "page": 10,
+      "char_offset": 53117,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "A key technique in SQLancer is using TLP to construct equivalent queries based on the observation that any predicate in SQL evaluates to TRUE ,FALSE, or NULL.",
+      "context_before": "construct semantically equivalent query pairs and use the performance differential within each query pair as a cross-referencing oracle for detecting performance bugs. We now investigate the efficacy of two techniques for obtaining equivalent query pairs for performance bug detection: (1) using SQLancer, an automated DBMS testing technique, to generate random query pairs (§8.6.1). (2) deriving query pairs from the Calcite ’s test suites, which are carefully crafted by DBMS developers (§8.6.2). 8.6.1 Query Pairs from SQLancer SQLancer is the state-ofart tool for discovering logic bugs [ 35–37].",
+      "context_after": "Given a base query, TLP constructs a mutant query that is equivalent to the base query in two steps. First, it partitions the base query into three partition queries, wherein each predicate is constructed based on the value of the base query’s predicate. Second, it concatenates these partition queries using the UNION or UNION ALL operator based on the semantics of the base query. Taking the query pair shown below as an example, the boolean predicate t0.c0 =t1.c0 from the base query is compared against TRUE ,FALSE, or NULL in each of the partition queries. TLP constructs the mutant query by con",
+      "section": null,
+      "page": 10,
+      "char_offset": 53214,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Given a base query, TLP constructs a mutant query that is equivalent to the base query in two steps.",
+      "context_before": "nce bugs. We now investigate the efficacy of two techniques for obtaining equivalent query pairs for performance bug detection: (1) using SQLancer, an automated DBMS testing technique, to generate random query pairs (§8.6.1). (2) deriving query pairs from the Calcite ’s test suites, which are carefully crafted by DBMS developers (§8.6.2). 8.6.1 Query Pairs from SQLancer SQLancer is the state-ofart tool for discovering logic bugs [ 35–37]. A key technique in SQLancer is using TLP to construct equivalent queries based on the observation that any predicate in SQL evaluates to TRUE ,FALSE, or NULL.",
+      "context_after": "First, it partitions the base query into three partition queries, wherein each predicate is constructed based on the value of the base query’s predicate. Second, it concatenates these partition queries using the UNION or UNION ALL operator based on the semantics of the base query. Taking the query pair shown below as an example, the boolean predicate t0.c0 =t1.c0 from the base query is compared against TRUE ,FALSE, or NULL in each of the partition queries. TLP constructs the mutant query by concatenating these partition queries with the UNION ALL operator. EXAMPLE 9. QUERY GENERATED BYTLP. /*",
+      "section": null,
+      "page": 10,
+      "char_offset": 53372,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP constructs the mutant query by concatenating these partition queries with the UNION ALL operator.",
+      "context_before": "n SQL evaluates to TRUE ,FALSE, or NULL. Given a base query, TLP constructs a mutant query that is equivalent to the base query in two steps. First, it partitions the base query into three partition queries, wherein each predicate is constructed based on the value of the base query’s predicate. Second, it concatenates these partition queries using the UNION or UNION ALL operator based on the semantics of the base query. Taking the query pair shown below as an example, the boolean predicate t0.c0 =t1.c0 from the base query is compared against TRUE ,FALSE, or NULL in each of the partition queries.",
+      "context_after": "EXAMPLE 9. QUERY GENERATED BYTLP. /* [Base query] */ SELECT * FROM t0, t1 WHERE t0.c0 = t1.c0; /* [Mutant query] */ SELECT * FROM t0, t1 WHERE t0.c0 = t1.c0 UNION ALL SELECT * FROM t0, t1 WHERE NOT (t0.c0 = t1.c0) UNION ALL SELECT * FROM t0, t1 WHERE (t0.c0 = t1. c0) IS NULL; 8.6.2 Manually-Crafted Query Pairs We derive a set of semantically equivalent query pairs from Calcite’s test suite [ 17,46]. These tests are manually crafted to ensure the correctness of query transformation rules in Calcite. Each test contains a SQL query and a set of transformation rules under test. Each test case tra",
+      "section": null,
+      "page": 10,
+      "char_offset": 53933,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This human-intensive baseline is challenging because: (1) they coverBase Query Mutant QueryPerformance Bugs Found Cockroach PostgreSQL SQLancer SQLancer 0 0 Calcite Test Calcite Test 4 4 Calcite Test AMOEBA 4 6 AMOEBA AMOEBA 25 14 Table 4: Comparative Analysis of AMOEBA –The results include the number of performance bugs that each set of query pairs is able to uncover with the SCOTT database.",
+      "context_before": "Manually-Crafted Query Pairs We derive a set of semantically equivalent query pairs from Calcite’s test suite [ 17,46]. These tests are manually crafted to ensure the correctness of query transformation rules in Calcite. Each test contains a SQL query and a set of transformation rules under test. Each test case transforms the query’s logic query plan tree using the rules and examines whether the resulting logic query plan tree is expected. We derive pairs of semantically equivalent queries from these tests: we use the input query as the base query and the transformed query as the mutant query.",
+      "context_after": "We provide the total number of bugs discovered by AMOEBA in the fourth row for comparison. a wide range of SQL operators, and (2) they use the same set of transformation rules as AMOEBA does that are effective at discovering performance bugs (§8.3). 8.6.3 Results As shown in Table 4, we compare AMOEBA against three baselines: (1) We use SQLancer to randomly generate 2000 pairs of equivalent queries. (2) We derive 373 pairs of equivalent queries from the Calcite test suite. (3) We acquire another 373 pairs of equivalent queries by using AMOEBA ’sMUTATOR to mutate base queries derived from the",
+      "section": null,
+      "page": 10,
+      "char_offset": 54919,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3 Results As shown in Table 4, we compare AMOEBA against three baselines: (1) We use SQLancer to randomly generate 2000 pairs of equivalent queries.",
+      "context_before": "se: (1) they coverBase Query Mutant QueryPerformance Bugs Found Cockroach PostgreSQL SQLancer SQLancer 0 0 Calcite Test Calcite Test 4 4 Calcite Test AMOEBA 4 6 AMOEBA AMOEBA 25 14 Table 4: Comparative Analysis of AMOEBA –The results include the number of performance bugs that each set of query pairs is able to uncover with the SCOTT database. We provide the total number of bugs discovered by AMOEBA in the fourth row for comparison. a wide range of SQL operators, and (2) they use the same set of transformation rules as AMOEBA does that are effective at discovering performance bugs (§8.3). 8.6.",
+      "context_after": "(2) We derive 373 pairs of equivalent queries from the Calcite test suite. (3) We acquire another 373 pairs of equivalent queries by using AMOEBA ’sMUTATOR to mutate base queries derived from the Calcite test suite. Then, we run each query pair on a given target database and measure the number of performance bugs that they reveal. The most notable observation is that AMOEBA finds significantly more performance bugs than alternatives. Specifically, AMOEBA discovers 25 and 14 performance bugs in CockroachDB andPostgreSQL, respectively. Manually-crafted query pairs derived from the Calcite test",
+      "section": null,
+      "page": 10,
+      "char_offset": 55570,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The query pairs generated by the SQLancer reveal zero performance bugs.",
+      "context_before": "MUTATOR to mutate base queries derived from the Calcite test suite. Then, we run each query pair on a given target database and measure the number of performance bugs that they reveal. The most notable observation is that AMOEBA finds significantly more performance bugs than alternatives. Specifically, AMOEBA discovers 25 and 14 performance bugs in CockroachDB andPostgreSQL, respectively. Manually-crafted query pairs derived from the Calcite test suite discover only 4 and 6 performance bugs in each DBMS, respectively. This illustrates the utility of automating the query transformation process.",
+      "context_after": "We next analyze the factors that contribute to the efficacy of AMOEBA. MUTATION RULES ANDALGORITHM .SQLancer differs from AMOEBA in that it mutates queries using TLP. We find that TLP is not useful for detecting performance bugs. By design, the TLP query is more complex than the corresponding base query, which inevitably leads to a higher execution time. Among 2000 query pairs generated by this technique, we discover that 16 and 12 pairs of equivalent queries exhibit a significant difference in execution time for PostgreSQL andCockroachDB, respectively. Among these queries, we find that the T",
+      "section": null,
+      "page": 10,
+      "char_offset": 56467,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer differs from AMOEBA in that it mutates queries using TLP.",
+      "context_before": "hey reveal. The most notable observation is that AMOEBA finds significantly more performance bugs than alternatives. Specifically, AMOEBA discovers 25 and 14 performance bugs in CockroachDB andPostgreSQL, respectively. Manually-crafted query pairs derived from the Calcite test suite discover only 4 and 6 performance bugs in each DBMS, respectively. This illustrates the utility of automating the query transformation process. The query pairs generated by the SQLancer reveal zero performance bugs. We next analyze the factors that contribute to the efficacy of AMOEBA. MUTATION RULES ANDALGORITHM .",
+      "context_after": "We find that TLP is not useful for detecting performance bugs. By design, the TLP query is more complex than the corresponding base query, which inevitably leads to a higher execution time. Among 2000 query pairs generated by this technique, we discover that 16 and 12 pairs of equivalent queries exhibit a significant difference in execution time for PostgreSQL andCockroachDB, respectively. Among these queries, we find that the TLP query always takes more time to execute compared to the base query (with an average slow-down of 17×). The root cause of these performance differences is that the T",
+      "section": null,
+      "page": 10,
+      "char_offset": 56640,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We find that TLP is not useful for detecting performance bugs.",
+      "context_before": "ificantly more performance bugs than alternatives. Specifically, AMOEBA discovers 25 and 14 performance bugs in CockroachDB andPostgreSQL, respectively. Manually-crafted query pairs derived from the Calcite test suite discover only 4 and 6 performance bugs in each DBMS, respectively. This illustrates the utility of automating the query transformation process. The query pairs generated by the SQLancer reveal zero performance bugs. We next analyze the factors that contribute to the efficacy of AMOEBA. MUTATION RULES ANDALGORITHM .SQLancer differs from AMOEBA in that it mutates queries using TLP.",
+      "context_after": "By design, the TLP query is more complex than the corresponding base query, which inevitably leads to a higher execution time. Among 2000 query pairs generated by this technique, we discover that 16 and 12 pairs of equivalent queries exhibit a significant difference in execution time for PostgreSQL andCockroachDB, respectively. Among these queries, we find that the TLP query always takes more time to execute compared to the base query (with an average slow-down of 17×). The root cause of these performance differences is that the TLP query forces the DBMS to perform additional operations (i.e.",
+      "section": null,
+      "page": 10,
+      "char_offset": 56706,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "By design, the TLP query is more complex than the corresponding base query, which inevitably leads to a higher execution time.",
+      "context_before": ", AMOEBA discovers 25 and 14 performance bugs in CockroachDB andPostgreSQL, respectively. Manually-crafted query pairs derived from the Calcite test suite discover only 4 and 6 performance bugs in each DBMS, respectively. This illustrates the utility of automating the query transformation process. The query pairs generated by the SQLancer reveal zero performance bugs. We next analyze the factors that contribute to the efficacy of AMOEBA. MUTATION RULES ANDALGORITHM .SQLancer differs from AMOEBA in that it mutates queries using TLP. We find that TLP is not useful for detecting performance bugs.",
+      "context_after": "Among 2000 query pairs generated by this technique, we discover that 16 and 12 pairs of equivalent queries exhibit a significant difference in execution time for PostgreSQL andCockroachDB, respectively. Among these queries, we find that the TLP query always takes more time to execute compared to the base query (with an average slow-down of 17×). The root cause of these performance differences is that the TLP query forces the DBMS to perform additional operations (i.e., fetching the tuples for each partition query and then combining those results together). Given this inherent overhead, TLP is",
+      "section": null,
+      "page": 10,
+      "char_offset": 56769,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Among these queries, we find that the TLP query always takes more time to execute compared to the base query (with an average slow-down of 17×).",
+      "context_before": "e SQLancer reveal zero performance bugs. We next analyze the factors that contribute to the efficacy of AMOEBA. MUTATION RULES ANDALGORITHM .SQLancer differs from AMOEBA in that it mutates queries using TLP. We find that TLP is not useful for detecting performance bugs. By design, the TLP query is more complex than the corresponding base query, which inevitably leads to a higher execution time. Among 2000 query pairs generated by this technique, we discover that 16 and 12 pairs of equivalent queries exhibit a significant difference in execution time for PostgreSQL andCockroachDB, respectively.",
+      "context_after": "The root cause of these performance differences is that the TLP query forces the DBMS to perform additional operations (i.e., fetching the tuples for each partition query and then combining those results together). Given this inherent overhead, TLP is not able to find a variety of performance bugs, which are discovered by AMOEBA and Calcite’s test suite. While Calcite and AMOEBA share the same set of mutation rules, they differ in how they leverage mutation rules. Each Calcite test transforms the base query using a small set of mutation rules with a specific firing order. On the other hand, AM",
+      "section": null,
+      "page": 10,
+      "char_offset": 57099,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The root cause of these performance differences is that the TLP query forces the DBMS to perform additional operations (i.",
+      "context_before": "ncer differs from AMOEBA in that it mutates queries using TLP. We find that TLP is not useful for detecting performance bugs. By design, the TLP query is more complex than the corresponding base query, which inevitably leads to a higher execution time. Among 2000 query pairs generated by this technique, we discover that 16 and 12 pairs of equivalent queries exhibit a significant difference in execution time for PostgreSQL andCockroachDB, respectively. Among these queries, we find that the TLP query always takes more time to execute compared to the base query (with an average slow-down of 17×).",
+      "context_after": "e., fetching the tuples for each partition query and then combining those results together). Given this inherent overhead, TLP is not able to find a variety of performance bugs, which are discovered by AMOEBA and Calcite’s test suite. While Calcite and AMOEBA share the same set of mutation rules, they differ in how they leverage mutation rules. Each Calcite test transforms the base query using a small set of mutation rules with a specific firing order. On the other hand, AMOEBA mutates the same base query with different combination of all the rules. As shown in the second and third rows of Tabl",
+      "section": null,
+      "page": 10,
+      "char_offset": 57244,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Given this inherent overhead, TLP is not able to find a variety of performance bugs, which are discovered by AMOEBA and Calcite’s test suite.",
+      "context_before": "bly leads to a higher execution time. Among 2000 query pairs generated by this technique, we discover that 16 and 12 pairs of equivalent queries exhibit a significant difference in execution time for PostgreSQL andCockroachDB, respectively. Among these queries, we find that the TLP query always takes more time to execute compared to the base query (with an average slow-down of 17×). The root cause of these performance differences is that the TLP query forces the DBMS to perform additional operations (i.e., fetching the tuples for each partition query and then combining those results together).",
+      "context_after": "While Calcite and AMOEBA share the same set of mutation rules, they differ in how they leverage mutation rules. Each Calcite test transforms the base query using a small set of mutation rules with a specific firing order. On the other hand, AMOEBA mutates the same base query with different combination of all the rules. As shown in the second and third rows of Table 4, AMOEBA ’s mutation strategy is more likely to discover performance bugs than the manual efforts. With PostgreSQL ,AMOEBA ’s mutation algorithm discovers two more performance bugs than Calcite. By design, for each test case in Cal",
+      "section": null,
+      "page": 10,
+      "char_offset": 57459,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M24",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "AMOEBA outperforms SQLancer andCalcite for three reasons: (1) AMOEBA differs from SQLancer in that it leverages a variety of optimization rules to mutate the base query, which are more likely to trigger different runtime behaviors that expose performance bugs; (2) Compared with Calcite ,AMOEBA broadly explores the equivalent query space by leveraging all available rules and their compositional effects; (3) Compared with Calcite ,AMOEBA covers more interesting clause pair combinations that are challenging for the DBMS to optimize and execute.",
+      "context_before": "( i.e.,WHERE and HAVING) are important for discovering performance bugs. When combined with expensive operators, such as JOIN ,GROUP BY, and UNION, their placements (before or after these operators) play a significant role in deciding query execution time. (2) The LIMIT clause challenges the optimizer. Since LIMIT asks for a smaller set of results, an optimal plan should either scan a partial table or terminate expensive operations early while still fetching the correct result sets. Improving how the DBMS handles these operations and their combinations would enhance its performance robustness.",
+      "context_after": "9 Limitations and Future Work We now discuss the limitations of AMOEBA and present ideas for tackling them in the future. 11 Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso DIFFERENTIAL TESTING .Since AMOEBA is based on a widelyused query optimization framework ( i.e.,Calcite ), it inherits the limitations of differential testing. First, it only focuses a widelysupported subset of SQL operators and functions. Fortunately, since Calcite is an extensible framework, it is feasible to add support for additional SQL features. Second, it assumes that a given SQL feature has the same semantics",
+      "section": null,
+      "page": 11,
+      "char_offset": 63941,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M25",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [ 35–37].",
+      "context_before": "query pairs that trigger different runtime behaviors. LOGIC BUGS.To circumvent the oracle availability problem associated with automated testing, researchers have applied differential andmetamorphic testing techniques for discovering logic bugs in DBMSs [ 16,29].RAGS discovers logic bugs by executing the same query on different DBMSs and comparing the results [ 40]. Waas et al. propose a framework for validating the query optimizer by executing alternative execution plans for the input query and comparing their results [ 41]. These techniques are not tailored for discovering performance bugs.",
+      "context_after": "The key idea behind SQLancer is to construct a metamorphic relation that is used to generate a cross-referencing oracle for detecting logic bugs. AMOEBA is similar to SQLancer in that it uses semantics-preserving query mutation rules to establish a metamorphic relation for discovering performance bugs. As discussed in §8.6, while the metamorphic relation proposed in SQLancer is effective in detecting logic bugs in DBMS, it is not suitable for discovering performance bugs. To the best of our knowledge, AMOEBA is the first technique that uses metamorphic testing to discover performance bugs in",
+      "section": null,
+      "page": 12,
+      "char_offset": 66846,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The key idea behind SQLancer is to construct a metamorphic relation that is used to generate a cross-referencing oracle for detecting logic bugs.",
+      "context_before": "em associated with automated testing, researchers have applied differential andmetamorphic testing techniques for discovering logic bugs in DBMSs [ 16,29].RAGS discovers logic bugs by executing the same query on different DBMSs and comparing the results [ 40]. Waas et al. propose a framework for validating the query optimizer by executing alternative execution plans for the input query and comparing their results [ 41]. These techniques are not tailored for discovering performance bugs. SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [ 35–37].",
+      "context_after": "AMOEBA is similar to SQLancer in that it uses semantics-preserving query mutation rules to establish a metamorphic relation for discovering performance bugs. As discussed in §8.6, while the metamorphic relation proposed in SQLancer is effective in detecting logic bugs in DBMS, it is not suitable for discovering performance bugs. To the best of our knowledge, AMOEBA is the first technique that uses metamorphic testing to discover performance bugs in DBMS. PERFORMANCE TESTING .Researchers have presented techniques for finding performance bugs by executing the DBMS on a predefined workloads and",
+      "section": null,
+      "page": 12,
+      "char_offset": 66955,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M27",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "AMOEBA is similar to SQLancer in that it uses semantics-preserving query mutation rules to establish a metamorphic relation for discovering performance bugs.",
+      "context_before": "[ 16,29].RAGS discovers logic bugs by executing the same query on different DBMSs and comparing the results [ 40]. Waas et al. propose a framework for validating the query optimizer by executing alternative execution plans for the input query and comparing their results [ 41]. These techniques are not tailored for discovering performance bugs. SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [ 35–37]. The key idea behind SQLancer is to construct a metamorphic relation that is used to generate a cross-referencing oracle for detecting logic bugs.",
+      "context_after": "As discussed in §8.6, while the metamorphic relation proposed in SQLancer is effective in detecting logic bugs in DBMS, it is not suitable for discovering performance bugs. To the best of our knowledge, AMOEBA is the first technique that uses metamorphic testing to discover performance bugs in DBMS. PERFORMANCE TESTING .Researchers have presented techniques for finding performance bugs by executing the DBMS on a predefined workloads and comparing their behavior against performance baselines [ 24,34,43,44]. These techniques detect performance regressions stemming from DBMS upgrades and configu",
+      "section": null,
+      "page": 12,
+      "char_offset": 67101,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M28",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "6, while the metamorphic relation proposed in SQLancer is effective in detecting logic bugs in DBMS, it is not suitable for discovering performance bugs.",
+      "context_before": "zer by executing alternative execution plans for the input query and comparing their results [ 41]. These techniques are not tailored for discovering performance bugs. SQLancer is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [ 35–37]. The key idea behind SQLancer is to construct a metamorphic relation that is used to generate a cross-referencing oracle for detecting logic bugs. AMOEBA is similar to SQLancer in that it uses semantics-preserving query mutation rules to establish a metamorphic relation for discovering performance bugs. As discussed in §8.",
+      "context_after": "To the best of our knowledge, AMOEBA is the first technique that uses metamorphic testing to discover performance bugs in DBMS. PERFORMANCE TESTING .Researchers have presented techniques for finding performance bugs by executing the DBMS on a predefined workloads and comparing their behavior against performance baselines [ 24,34,43,44]. These techniques detect performance regressions stemming from DBMS upgrades and configuration changes. AMOEBA differs from these approaches in that it does not require a pre-defined baseline for finding performancebugs. Instead, it leverages the tested DBMS’s",
+      "section": null,
+      "page": 12,
+      "char_offset": 67279,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M9",
+        "M16"
+      ],
+      "describes_as_state_of_the_art": [
+        "M25"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:13:36Z",
+    "is_model_written": true,
+    "summary": "AMOEBA detects performance bugs in database systems, a class the authors note has received far less attention than functional bugs despite affecting response time. It constructs pairs of semantically equivalent queries and compares their response times on the same system; a significant difference points to a performance bug. The paper contributes structure and predicate mutation rules for building such pairs, plus feedback mechanisms to improve efficacy. On PostgreSQL and CockroachDB it found 20 previously unknown performance bugs, 14 confirmed.",
+    "narrative": "AMOEBA takes TLP as the closest related effort and argues it does not transfer: its equivalent queries are tailored to logic bugs and are comparatively simple, so they cannot serve as a metamorphic relation for performance. TLP is then one of the sources of equivalent queries AMOEBA is compared against.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "motivation",
+      "M4": "motivation",
+      "M5": "baseline",
+      "M6": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Nothing shown says AMOEBA is built on SQLancer; the tools are named as the means by which others find crash and logic bugs."
+      },
+      "extends_technique": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "AMOEBA applies the equivalent-query idea to performance rather than correctness, which is arguably a generalisation, but the mentions read frame TLP as an approach that does not transfer rather than one being extended."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M5",
+            "sentence": "We compare AMOEBA against two other sources of equivalent queries that could be used for detecting performance bugs: (1) a manually-written test suite in a widely-used query optimization framework, and (2) the TLP.",
+            "section": null,
+            "page": 2
+          }
+        ],
+        "reasoning": "M5 states AMOEBA is compared against two other sources of equivalent queries, TLP among them.",
+        "techniques": [
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "A pattern fired on M25, which was outside the mentions read here."
+      }
+    },
+    "disagreements": [],
+    "unresolved": [
+      "Whether the paper calls SQLancer the state of the art: the mention a pattern fired on, M25, was not among those read."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2206_08530.json b/_data/papers/paper_arxiv_2206_08530.json
new file mode 100644
index 0000000..f4ba1b0
--- /dev/null
+++ b/_data/papers/paper_arxiv_2206_08530.json
@@ -0,0 +1,596 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:27:02Z",
+  "paper": {
+    "id": "paper:arxiv:2206.08530",
+    "title": "GDsmith: Detecting Bugs in Graph Database Engines",
+    "authors": [
+      "Weisheng Lin",
+      "Ziyue Hua",
+      "Luyao Ren",
+      "Z. Li",
+      "Lu Zhang",
+      "Tao Xie"
+    ],
+    "year": 2022,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2206.08530",
+    "s2_paper_id": "71dd291b4be0d074982f521ff654468755084a76",
+    "url": "https://arxiv.org/abs/2206.08530",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2206.08530",
+      "retrieved_at": "2026-09-10T15:27:02Z",
+      "chars": 67932,
+      "content_sha256": "sha256:76eee18f2202c0c3568d311363f1b8088339c228951a26af7b5fe3ab98041bf9"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1910
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND",
+        "start": 10891
+      },
+      {
+        "number": "2.1",
+        "title": "Graph Database Engines We Test",
+        "start": 11276
+      },
+      {
+        "number": "2.2",
+        "title": "Cypher Language",
+        "start": 12356
+      },
+      {
+        "number": "2.3",
+        "title": "Historical Bug Statistics",
+        "start": 14580
+      },
+      {
+        "number": "3",
+        "title": "APPROACH",
+        "start": 18848
+      },
+      {
+        "number": "3.1",
+        "title": "Schema and Graph Generation",
+        "start": 21790
+      },
+      {
+        "number": "3.2",
+        "title": "Skeleton and its Completion",
+        "start": 23865
+      },
+      {
+        "number": "3.3",
+        "title": "Structural Mutation Strategies",
+        "start": 29717
+      },
+      {
+        "number": "3.4",
+        "title": "Property Key Selection",
+        "start": 33573
+      },
+      {
+        "number": "3.5",
+        "title": "Bug Detection",
+        "start": 35705
+      },
+      {
+        "number": "4",
+        "title": "EVALUATIONS",
+        "start": 37863
+      },
+      {
+        "number": "4.1",
+        "title": "Evaluation Setup",
+        "start": 38270
+      },
+      {
+        "number": "4.1.2",
+        "title": "Implementation. We implement the GDsmith prototype with",
+        "start": 38684
+      },
+      {
+        "number": "4.1.3",
+        "title": "Baseline. Note that there is no applicable baseline to which",
+        "start": 39520
+      },
+      {
+        "number": "4.2",
+        "title": "RQ1/RQ2: Effectiveness and Efficiency",
+        "start": 41693
+      },
+      {
+        "number": "4.2.2",
+        "title": "Core Grammar Coverage. For each of the three approaches,",
+        "start": 43228
+      },
+      {
+        "number": "4.2.3",
+        "title": "Non-empty Result Rate. We run the three approaches ten",
+        "start": 43878
+      },
+      {
+        "number": "4.2.5",
+        "title": "Bug Detection Efficiency. We next examine the efficiency",
+        "start": 45688
+      },
+      {
+        "number": "4.3",
+        "title": "RQ3: Comparison with Baseline",
+        "start": 46724
+      },
+      {
+        "number": "4.4",
+        "title": "RQ4: Practicability",
+        "start": 47637
+      },
+      {
+        "number": "4.4.2",
+        "title": "Selected Bugs. Next, we show a selection of confirmed bugs",
+        "start": 48825
+      },
+      {
+        "number": "4.4.3",
+        "title": "Developers’ Feedback. After we report our detected bugs to",
+        "start": 51340
+      },
+      {
+        "number": "4.5",
+        "title": "Threats to Validity",
+        "start": 52811
+      },
+      {
+        "number": "4.5.1",
+        "title": "External Validity. The main threat to external validity is that",
+        "start": 52835
+      },
+      {
+        "number": "4.5.2",
+        "title": "Internal Validity. The main threat to internal validity lies",
+        "start": 53303
+      },
+      {
+        "number": "5",
+        "title": "RELATED WORK",
+        "start": 53715
+      },
+      {
+        "number": "6",
+        "title": "CONCLUSION",
+        "start": 57266
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "N. Bourbakis. 1998. Artificial Intelligence and Automation. Artificial Intelligence and Automation.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Walter H. Burkhardt. 1967. Generating test programs from syntax. Computing 2, 1 (1967), 53–73. https://doi.org/10.1007/BF02235512",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Xinyue Chen, Chenglong Wang, and Alvin Cheung. 2020. Testing query execution engines with mutations. In Proceedings of the 8th International Workshop on Testing Database Systems, DBTest@SIGMOD 2020, Portland, Oregon, June 19, 2020, Pinar Tözün and Alexander Böhm (Eds.). ACM, 6:1–6:5. https://doi.org/10.1145/ 3395032.3395322",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Memgraph Community. 2022. Memgraph: Build modern, graph-based applications on top of your streaming data in minutes. https://github.com/memgraph/ memgraph",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Neo4j Community. 2022. Neo4j: Graphs for Everyone. https://github.com/neo4j/ neo4j",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "The Apache Software Foundation. 2022. Cypher for Gremlin. https: //github.com/opencypher/cypher-for-gremlin/tree/master/tinkerpop/cyphergremlin-server-client",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "The Apache Software Foundation. 2022. Gremlin Query Language. https: //tinkerpop.apache.org/gremlin.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Martin Schuster, Petra Selmer, and Andrés Taylor. 2018. Formal Semantics of the Language Cypher. CoRR abs/1802.09984 (2018). arXiv:1802.09984 http://arxiv.org/abs/1802.09984",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and Andrés Taylor. 2018. Cypher: An Evolving Query Language for Property Graphs. In Proceedings of the 2018 International Conference on Management of Data, SIGMOD Conference 2018, Houston, TX, USA, June 10-15, 2018, Gautam Das, Christopher M. Jermaine, a",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Bogdan Ghit, Nicolás Poggi, Josh Rosen, Reynold Xin, and Peter A. Boncz. 2020. SparkFuzz: searching correctness regressions in modern query engines. InProceedings of the 8th International Workshop on Testing Database Systems, DBTest@SIGMOD 2020, Portland, Oregon, June 19, 2020, Pinar Tözün and Alexander Böhm (Eds.). ACM, 1:1–1:6. https://doi.org/10.1145/3395032.3395327",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woon-Hak Kang. 2019. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. Proc. VLDB Endow. 13, 1 (2019), 57–70. https://doi.org/10. 14778/3357377.3357382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Lior Kogan. 2017. V1: A Visual Query Language for Property Graphs. CoRR abs/1710.04470 (2017). arXiv:1710.04470 http://arxiv.org/abs/1710.04470",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Takahiro Konno, Runhe Huang, Tao Ban, and Chuanhe Huang. 2017. Goods recommendation based on retail knowledge in a Neo4j graph database combined with an inference mechanism implemented in jess. In 2017 IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Computed, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation, Smar",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Muhammad Numair Mansur, Maria Christakis, and Valentin Wüstholz. 2021. Metamorphic testing of Datalog engines. In ESEC/FSE ’21: 29th ACM Joint European Software Engineering Conference and Symposium on the Foundations ofSoftware Engineering, Athens, Greece, August 23-28, 2021, Diomidis Spinellis, Georgios Gousios, Marsha Chechik, and Massimiliano Di Penta (Eds.). ACM, 639–650. https://doi.org/10.11",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software. Digit. Tech. J.10, 1 (1998), 100–107. http://www.hpl.hp.com/hpjournal/dtj/vol10num1/ vol10num1art9.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Memgraph. 2022. Memgraph: Frictionless, Innovative, Graph Applications. https: //memgraph.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Chaitanya Mishra and Nick Koudas. 2009. Interactive query refinement. In EDBT 2009, 12th International Conference on Extending Database Technology, Saint Petersburg, Russia, March 24-26, 2009, Proceedings (ACM International Conference Proceeding Series, Vol. 360), Martin L. Kersten, Boris Novikov, Jens Teubner, Vladimir Polutin, and Stefan Manegold (Eds.). ACM, 862–873. https: //doi.org/10.1145/15",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Neo4j. 2022. The Fastest Path To Graph Productivity: Neo4j Graph Database. https://neo4j.com/product/neo4j-graph-database/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "The openCypher Implementers Group. 2022. Cypher Query Language Reference, Version 9. https://s3.amazonaws.com/artifacts.opencypher.org/openCypher9.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Jiwon Park, Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2021. Generative type-aware mutation for testing SMT solvers. Proc. ACM Program. Lang. 5, OOPSLA (2021), 1–19. https://doi.org/10.1145/3485529",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Paul and Purdom. 1972. A sentence generator for testing parsers. Bit Numerical Mathematics (1972).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "RedisGraph. 2022. RedisGrapha graph database module for Redis. https: //oss.redis.com/redisgraph/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Manuel Rigger. 2022. SQLancer: Detecting Logic Bugs in DBMS. https://github. com/sqlancer/sqlancer",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 24,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, Prem Devanbu, Myra B. Cohen, and Thomas Zimmermann (Eds.). ACM, 1140– 1152. https://doi.org/1",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 667–682. https://www.usenix.org/conference/osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "Andreas Seltenreich. 2022. Bug Squashing with SQLsmith. https://github.com/ anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Sudipta Sen, Akash Mehta, Runa Ganguli, and Soumya Sen. 2021. Recommendation of Influenced Products Using Association Rule Mining: Neo4j as a Case Study. SN Comput. Sci. 2, 2 (2021), 74. https://doi.org/10.1007/s42979-021-00460-8",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Qingchao Shen, Haoyang Ma, Junjie Chen, Yongqiang Tian, Shing-Chi Cheung, and Xiang Chen. 2021. A comprehensive study of deep learning compiler bugs. In ESEC/FSE ’21: 29th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Athens, Greece, August 23-28, 2021, Diomidis Spinellis, Georgios Gousios, Marsha Chechik, and Massimiliano Di Penta (Ed",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Erez Shinan. 2022. Larka parsing toolkit for Python. https://github.com/larkparser/lark",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, August 24-27, 1998, New York City, New York, USA, Ashish Gupta, Oded Shmueli, and Jennifer Widom (Eds.). Morgan Kaufmann, 618–622. http://www.vldb.org/conf/1998/p618.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "solid IT gmbh. 2022. DB-Engines Ranking of Graph DBMS. https://db-engines. com/en/ranking/graph+ dbms",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Thodoris Sotiropoulos, Stefanos Chaliasos, Vaggelis Atlidakis, Dimitris Mitropoulos, and Diomidis Spinellis. 2021. Data-Oriented Differential Testing of ObjectRelational Mapping Systems. In 43rd IEEE/ACM International Conference on Software Engineering, ICSE 2021, Madrid, Spain, 22-30 May 2021. IEEE, 1535–1547. https://doi.org/10.1109/ICSE43902.2021.00137",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jian Wang, Ke Wang, Jing Li, Jianmin Jiang, Yanfei Wang, Jing Mei, and Shaochun Li. 2020. Accelerating Epidemiological Investigation Analysis by Using NLP and Knowledge Reasoning: A Case Study on COVID-19. In AMIA 2020, American Medical Informatics Association Annual Symposium, Virtual Event, USA, November 14-18, 2020. AMIA. https://knowledge.amia.org/72332-amia-1.4602255/t0031.4606204/t003-1.4606",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In 43rd IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice, ICSE (SEIP) 2021, Madrid, Spain, May 25-28, 2021. IEEE, 328–337. https://doi.org/10.1109/ICSE-SEIP52600.2021.00042",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2020. On the unusual effectiveness of type-aware operator mutations for testing SMT solvers. Proc. ACM Program. Lang. 4, OOPSLA (2020), 193:1–193:25. https://doi.org/10.1145/ Conference’17, July 2017, Washington, DC, USA Wei Lin, Ziyue Hua, Luyao Ren, Zongyang Li, Lu Zhang, and Tao Xie 3428261",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2011, San Jose, CA, USA, June 4-8, 2011, Mary W. Hall and David A. Padua (Eds.). ACM, 283–294. https://doi.org/10.1145/1993498.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Michal Zalewski. 2022. american fuzzy lop (2.52b). https://lcamtuf.coredump. cx/afl/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event, USA, November 9-13, 2020, Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna (Eds.). ACM, 955–970. https://doi.org/10.1145/",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 23,
+      "text": "Manuel Rigger. 2022. SQLancer: Detecting Logic Bugs in DBMS. https://github. com/sqlancer/sqlancer",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 24,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, Prem Devanbu, Myra B. Cohen, and Thomas Zimmermann (Eds.). ACM, 1140– 1152. https://doi.org/10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 667–682. https://www.usenix.org/conference/osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Its framework is derived from SQLancer [ 23] (which is a tool to automatically test relational database engines).",
+      "context_before": "practicably does GDsmith detect real-world bugs in popular graph database engines? 4.1 Evaluation Setup 4.1.1 Subjects. We opt for three popular real-world graph database engines as our test subjects. We test the released versions of Neo4j Community Edition from 3.5 to 4.4, the released versions of RedisGraph from 2.4 to 2.8, and the released version 2.1.1 of Memgraph Community Edition. All of these subjects are downloaded from their official repositories without any underlying modification. 4.1.2 Implementation. We implement the GDsmith prototype with over 12K non-comment lines of Java code.",
+      "context_after": "GDsmith uses Neo4j Java Driver 4.1.1 to connect and interact with Neo4j and Memgraph, and uses JRedisGraph 2.5.1 to connect and interact with RedisGraph. Some graph database engines implement only a subset of the Cypher language. When conducting cross-engines differential testing, all Cypher queries generated by GDsmith do not contain any Cypher feature that is unsupported by either graph database engine. All evaluations are conducted on a Windows 11 laptop with Intel i7-8565U CPU and 16 GB of memory. Conference’17, July 2017, Washington, DC, USA Wei Lin, Ziyue Hua, Luyao Ren, Zongyang Li, Lu",
+      "section": "4.1.2 Implementation. We implement the GDsmith prototype with",
+      "page": 7,
+      "char_offset": 38786,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [ 26] detects wrong-result bugs by checking whether a specific record is fetched correctly.",
+      "context_before": "additional analysis. Ratel [ 35] is an enterprise-level fuzzer that improves the feedback precision, enhances the robustness of input generation, and performs an on-line investigation on the root cause of bugs with its industry-oriented design. The aforementioned approaches can detect only crashing bugs in relational database engines. To detect wrong-result bugs, RAGS [ 31] generates and executes SQL queries in multiple relational database engines, meanwhile observes differences in the output sets. Any inconsistency among results indicates at least one relational database engine contains bugs.",
+      "context_after": "NoREC [ 24] detects bugs in relational database engines by applying a semantics-preserving transformation to a given SQL query to disable the engine’s optimizations and addresses PQS’ high implementation effort. TLP [ 25] derives multiple SQL queries that compute a partial result of the initial query. By using a composition operator, the partitions can be combined to yield the same result as the original query; if the result differs, a bug in the relational database engine has been detected. MutaSQL [ 3] generates test cases by mutating a SQL query over a database instance into a semantically",
+      "section": "5 RELATED WORK",
+      "page": 10,
+      "char_offset": 55921,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [ 24] detects bugs in relational database engines by applying a semantics-preserving transformation to a given SQL query to disable the engine’s optimizations and addresses PQS’ high implementation effort.",
+      "context_before": "ion, enhances the robustness of input generation, and performs an on-line investigation on the root cause of bugs with its industry-oriented design. The aforementioned approaches can detect only crashing bugs in relational database engines. To detect wrong-result bugs, RAGS [ 31] generates and executes SQL queries in multiple relational database engines, meanwhile observes differences in the output sets. Any inconsistency among results indicates at least one relational database engine contains bugs. PQS [ 26] detects wrong-result bugs by checking whether a specific record is fetched correctly.",
+      "context_after": "TLP [ 25] derives multiple SQL queries that compute a partial result of the initial query. By using a composition operator, the partitions can be combined to yield the same result as the original query; if the result differs, a bug in the relational database engine has been detected. MutaSQL [ 3] generates test cases by mutating a SQL query over a database instance into a semantically equivalent query mutant, and checks the results returned by the relational database engine under test. Compared with these approaches, GDsmith includes our skeleton-based completion technique to ensure that each",
+      "section": "5 RELATED WORK",
+      "page": 10,
+      "char_offset": 56017,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [ 25] derives multiple SQL queries that compute a partial result of the initial query.",
+      "context_before": "relational database engines. To detect wrong-result bugs, RAGS [ 31] generates and executes SQL queries in multiple relational database engines, meanwhile observes differences in the output sets. Any inconsistency among results indicates at least one relational database engine contains bugs. PQS [ 26] detects wrong-result bugs by checking whether a specific record is fetched correctly. NoREC [ 24] detects bugs in relational database engines by applying a semantics-preserving transformation to a given SQL query to disable the engine’s optimizations and addresses PQS’ high implementation effort.",
+      "context_after": "By using a composition operator, the partitions can be combined to yield the same result as the original query; if the result differs, a bug in the relational database engine has been detected. MutaSQL [ 3] generates test cases by mutating a SQL query over a database instance into a semantically equivalent query mutant, and checks the results returned by the relational database engine under test. Compared with these approaches, GDsmith includes our skeleton-based completion technique to ensure that each randomly generated Cypher query satisfies the semantic requirements. GDsmith also includes",
+      "section": "5 RELATED WORK",
+      "page": 10,
+      "char_offset": 56229,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/ddaa2000/GDsmith",
+    "markers": [
+      "renamed_sqlancer_package",
+      "sqlancer_source_content_match"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/ddaa2000/GDsmith",
+        "source_type": "github_repository",
+        "excerpt": "# GDsmith",
+        "note": "Repository is named after GDsmith, the tool this paper says it built, and is about database testing.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/ddaa2000/GDsmith/blob/master/src/main/java/org/example/gdsmith/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.ALPHANUMERIC;",
+        "excerpt_is_verbatim": true,
+        "note": "src/main/java/org/example/gdsmith/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.example.gdsmith (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+        "content_sha256": "sha256:d09fa9dddb2efa23e9184c0a8e2c71c39cdaffe6bde161da903434f62c4c4711",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "uses_infrastructure": [
+        "M1"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:20:36Z",
+    "is_model_written": true,
+    "summary": "GDsmith tests graph database engines that use Cypher. Its central difficulty is that random test generation struggles to produce property graphs and queries complex enough to return non-empty results, which is what wrong-result bugs require. GDsmith ensures every generated query meets the semantic requirements, and raises the chance of complex non-empty results with graph-guided generation of pattern combinations and data-guided generation of conditions. It detected bugs across three popular open-source graph engines.",
+    "narrative": "GDsmith says plainly that its framework is derived from SQLancer, carrying that tool's approach from relational engines to Cypher, and its artifact holds SQLancer's source under a renamed package. PQS, NoREC and TLP are summarised as the relational oracles that came before it.",
+    "roles": {
+      "M1": "reuse_implementation",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M1"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "Its framework is derived from SQLancer [ 23] (which is a tool to automatically test relational database engines).",
+            "section": "4.1.2 Implementation. We implement the GDsmith prototype with",
+            "page": 7
+          }
+        ],
+        "reasoning": "M1 states the framework is derived from SQLancer. The artifact carrying SQLancer source under a renamed package agrees.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2304_10044.json b/_data/papers/paper_arxiv_2304_10044.json
new file mode 100644
index 0000000..22a05b7
--- /dev/null
+++ b/_data/papers/paper_arxiv_2304_10044.json
@@ -0,0 +1,438 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:09Z",
+  "paper": {
+    "id": "paper:arxiv:2304.10044",
+    "title": "Finding Bug-Inducing Program Environments",
+    "authors": [
+      "Z. Mirzamomen",
+      "Marcel Böhme"
+    ],
+    "year": 2023,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2304.10044",
+    "s2_paper_id": "e10d0a78b6e4dd6e074f925e669bf632b5d48f53",
+    "url": "https://arxiv.org/abs/2304.10044",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2304.10044",
+      "retrieved_at": "2026-09-09T01:33:09Z",
+      "chars": 64272,
+      "content_sha256": "sha256:dc0aebc22c1378d7caa27c5935674fdeeae2729356b7c5a2b26fba67bf740fd9"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 11,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1793
+      },
+      {
+        "number": "2.1",
+        "title": "Runtime: Capturing Environment",
+        "start": 9492
+      },
+      {
+        "number": "2.2",
+        "title": "Fuzzer: Modifying Environment",
+        "start": 12603
+      },
+      {
+        "number": "4.1",
+        "title": "Research Questions",
+        "start": 21999
+      },
+      {
+        "number": "4.2",
+        "title": "Fuzzer Implementation",
+        "start": 23121
+      },
+      {
+        "number": "4.3",
+        "title": "Benchmarking",
+        "start": 24939
+      },
+      {
+        "number": "4.4",
+        "title": "Setup and Infrastructure",
+        "start": 26377
+      },
+      {
+        "number": "5.1",
+        "title": "Performance analysis",
+        "start": 43136
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2020. libFuzzer-a ibrary for coverage-guided fuzz testing. https://llvm.org/ docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Cornelius Aschermann, Tommaso Frassetto, Thorsten Holz, Patrick Jauernig, Ahmad-Reza Sadeghi, and Daniel Teuchert. 2019. Nautilus: Fishing for Deep Bugs with Grammars.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Ali Basiri, Niosha Behnam, Ruud de Rooij, Lorin Hochstein, Luke Kosewski, Justin Reynolds, and Casey Rosenthal. 2016. Chaos Engineering. IEEE Software 33, 3 (2016), 35–41. https://doi.org/10.1109/MS.2016.60",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Massimo Bernaschi, Emanuele Gabrielli, and Luigi V. Mancini. 2002. Remus: A Security-Enhanced Operating System. ACM Trans. Inf. Syst. Secur. 5, 1 (feb 2002), 36–61. https://doi.org/10.1145/504909.504911",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. 2016. Coveragebased Greybox Fuzzing As Markov Chain. In Proceedings of the 2016 ACMSIGSAC Conference on Computer and Communications Security. 1032–1043.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Huning Dai, Christian Murphy, and Gail Kaiser. 2010. Configuration Fuzzing for Software Vulnerability Detection. In 2010 International Conference on Availability, Reliability and Security. 525–530. https://doi.org/10.1109/ARES.2010.22",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Hugo Gascon, Christian Wressnegger, Fabian Yamaguchi, Daniel Arp, and Konrad Rieck. 2015. Pulsar: Stateful Black-Box Fuzzing of Proprietary Network Protocols. InSecurity and Privacy in Communication Networks. 330–347.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Google. 2012. ClusterFuzz. https://github.com/google/clusterfuzz# trophies. Accessed: 2022-08-12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Google. 2016. OSS-Fuzz: Continuous Fuzzing for Open Source Software. https: //github.com/google/oss-fuzz#trophies. Accessed: 2022-08-12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Ahmad Hazimeh, Adrian Herrera, and Mathias Payer. 2021. Magma: A GroundTruth Fuzzing Benchmark. Proceedings of the ACM Conference on Measurement and Analysis of Computing Systems 4, 3, Article 49 (jun 2021), 29 pages. https: //doi.org/10.1145/3428334",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Zhihao Hu and Zulie Pan. 2021. A Systematic Review of Network Protocol Fuzzing Techniques. In 2021 IEEE 4th Advanced Information Management, Communicates, Electronic and Automation Control Conference (IMCEC), Vol. 4. 1000–1005. https: //doi.org/10.1109/IMCEC51613.2021.9482063",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "]intel Intel. [n. d.]. Specs of Intel Pentium 4 (Model 662; 2005), the first chip to implement VT-x Vitualization Technology. https://ark.intel.com/content/ www/us/en/ark/products/27486/intel-pentium-4-processor-662supporting-ht-technology-2m-cache-3-60-ghz-800-mhz-fsb.html. Accessed: 2022-08-12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, Kangjie Lu, and Shi-Min Hu. 2020. Fuzzing Error Handling Code using Context-Sensitive Software Fault Injection. In 29th USENIX Security Symposium (USENIX Security 20). USENIX Association, 2595–2612. https: //www.usenix.org/conference/usenixsecurity20/presentation/jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler Validation via Equivalence modulo Inputs. SIGPLAN Not. 49, 6 (jun 2014), 216–226. https: //doi.org/10.1145/2666356.2594334 Finding Bug-Inducing Program Environments",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Caroline Lemieux and Koushik Sen. 2018. FairFuzz: A Targeted Mutation Strategy for Increasing Greybox Fuzz Testing Coverage. In Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering (Montpellier, France) (ASE 2018). ACM, New York, NY, USA, 475–485. https: //doi.org/10.1145/3238147.3238176",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Wang Li, Zhouyang Jia, Shanshan Li, Yuanliang Zhang, Teng Wang, Erci Xu, Ji Wang, and Xiangke Liao. 2021. Challenges and Opportunities: An in-Depth Empirical Study on Configuration Error Injection Testing. In Proceedings of the 30th ACMSIGSOFT International Symposium on Software Testing and Analysis (Virtual, Denmark) (ISSTA 2021). Association for Computing Machinery, New York, NY, USA, 478–490. h",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Eichelberg Marco, Riesmeier Joerg, Wilkens Thomas, J. Hewett Andrew, Barth Andreas, and Jensch Peter. 2004. Ten years of medical imaging standardization and prototypical implementation: the DICOM standard and the OFFIS DICOM toolkit (DCMTK). In Proc.SPIE, Vol. 5371. https://doi.org/10.1117/12.534853",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jonathan Metzman, László Szekeres, Laurent Simon, Read Sprabery, and Abhishek Arya. 2021. FuzzBench: An Open Fuzzer Benchmarking Platform and Service. InProceedings of the Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1393–1403. https: //doi.org/10.1145/3468264.3473932",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. 2008. Generating Targeted Queries for Database Testing. In Proceedings of the 2008 ACMSIGMOD International Conference on Management of Data (Vancouver, Canada) (SIGMOD ’08). Association for Computing Machinery, New York, NY, USA, 499–510. https://doi.org/10.1145/1376616.1376668",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Roberto Natella and Van-Thuan Pham. 2021. ProFuzzBench: A Benchmark for Stateful Protocol Fuzzing. In Proceedings of the 30th ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Changhai Nie and Hareton Leung. 2011. A Survey of Combinatorial Testing. ACM Comput. Surv. 43, 2, Article 11 (feb 2011), 29 pages. https://doi.org/10. 1145/1883612.1883618",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Sebastian Österlund, Kaveh Razavi, Herbert Bos, and Cristiano Giuffrida. 2020. ParmeSan: Sanitizer-Guided Greybox Fuzzing. USENIX Association, USA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Van-Thuan Pham, Marcel Böhme, Andrew E. Santosa, Alexandru Razvan Caciulescu, and Abhik Roychoudhury. 2018. Smart Greybox Fuzzing. CoRR abs/1811.09447 (2018). arXiv:1811.09447 http://arxiv.org/abs/1811.09447",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Van-Thuan Pham, Marcel Boehme, and Abhik Roychoudhury. 2020. AFLNet: A Greybox Fuzzer for Network Protocols. In IEEE International Conf. on Software Testing Verification and Validation.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Niels Provos, Markus Friedl, and Peter Honeyman. 2003. Preventing Privilege Escalation. In Proceedings of the 12th Conference on USENIX Security Symposium Volume 12 (Washington, DC) (SSYM’03). USENIX Association, USA, 16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (nov 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, 667–682. https://www.usenix. org/conference/osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "Alton F Sanders and Robert Pickering. 2018. the Unix System. Revival: The Handbook of Software for Engineers and Scientists (1995) (2018), 77.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Sergej Schumilo, Cornelius Aschermann, Andrea Jemmett, Ali Abbasi, and Thorsten Holz. 2022. Nyx-Net: Network Fuzzing with Incremental Snapshots. In Proceedings of the Seventeenth European Conference on Computer Systems (Rennes, France) (EuroSys ’22). Association for Computing Machinery, New York, NY, USA, 166–180. https://doi.org/10.1145/3492321.3519591",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Sergej Schumilo, Cornelius Aschermann, Andrea Jemmett, Ali Abbasi, and Thorsten Holz. 2022. Nyx-net: network fuzzing with incremental snapshots. InProceedings of the Seventeenth European Conference on Computer Systems. 166– 180.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Kostya Serebryany. 2022. Fuzzer Test Suite. https://github.com/google/fuzzertest-suite.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Website. 2017. American Fuzzy Lop (AFL) Fuzzer. http://lcamtuf.coredump. cx/afl/technical_details.txt. Accessed: 2017-05-13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Website. 2017. Search engine for the internet of things – devices still vulnerable to Heartbleed. https://www.shodan.io/report/89bnfUyJ. Accessed: 201705-13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Website. 2020. Openssl bug fixing commit. https://github.com/openssl/ openssl/commit/6ab6ecfd6d2d659326f427dceb1b65ae1b4b012b. Accessed: 2022-07-22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Website. 2022. DCMTK bug reports. https://support.dcmtk.org/redmine/ issues/1026. Accessed: 2022-07-22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Website. 2022. Kamailio bug fixing commit. https://github.com/kamailio/ kamailio/commit/f0cea1a7c03e400b4398795c2d8b0f7e45d1dfb5. Accessed: 2022-07-22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Website. 2022. Live555 bug report 1. https://github.com/rgaufman/live555/ issues/38. Accessed: 2022-07-26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Website. 2022. Live555 bug report 2. https://github.com/rgaufman/live555/ issues/39. Accessed: 2022-07-27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Website. 2022. Live555 bug report 3. https://github.com/rgaufman/live555/ issues/40. Accessed: 2022-07-27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Website. 2022. Pure-ftpd bug report 1. https://github.com/bajinsheng/ pure-ftpd/issues/1. Accessed: 2022-07-25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Website. 2022. Pure-ftpd bug report 2. https://github.com/bajinsheng/ pure-ftpd/issues/2. Accessed: 2022-07-25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Website. 2022. Pure-ftpd bug report 3. https://github.com/bajinsheng/ pure-ftpd/issues/3. Accessed: 2022-07-25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and Understanding Bugs in C Compilers. SIGPLAN Not. 46, 6 (jun 2011), 283–294. https://doi.org/10.1145/1993316.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Andreas Zeller, Rahul Gopinath, Marcel Böhme, Gordon Fraser, and Christian Holler. 2022. Testing Configurations. In The Fuzzing Book. CISPA Helmholtz Center for Information Security. https://www.fuzzingbook.org/ html/ConfigurationFuzzer.html Retrieved 2022-01-23 13:06:27+01:00.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security (Virtual Event, USA) (CCS ’20). Association for Computing Machinery, New York, NY, USA, 955–970. https://doi.org/10.1145/3372297.3417260",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (nov 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, 667–682. https://www.usenix. org/conference/osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 14,19,26,27,43,45]",
+      "technique": "tlp",
+      "sentence": "For instance, there are domain-specific works on fuzzing a program’s configuration [ 6,16,44], databases [ 14,19,26,27,43,45], the network [7,11,24,29], et cetera.",
+      "context_before": "11https://github.com/DCMTK/dcmtk/blob/7f8564c/dcmqrdb/libsrc/dcmqrcnf.cc#L691 12https://github.com/DCMTK/dcmtk/blob/master/dcmqrdb/libsrc/dcmqrcnf.cc#L686 Zahra Mirzamomen and Marcel Böhme However, existing works do not actually modify the environment within which the program-under-test runs. Since fuzzing the environment itself might yield unrealistic environments or may not impact the program behavior, we propose to fuzz the interaction with the program’s environment instead. Domain-specific approaches. There exist several domain-specific approaches to fuzz non-traditional program “inputs\".",
+      "context_after": "For instance, in the Fuzzing Book [ 44], the importance of testing all possible configuration options passed to the program is addressed by proposing a way of automatically inferring configuration options. The program’s configuration can then be tested using combinatorial testing [ 21]. However, all previous approaches focus on domain-specific improvements. There are programs that take non-traditional inputs. For instance, database management systems (DBMS) take databases as input; compilers take program source code as inputs. In order to test these systems automatically domain-specific appro",
+      "section": "5.1 Performance analysis",
+      "page": 10,
+      "char_offset": 50539,
+      "cited_reference": {
+        "number": 26,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (nov 2020), 30 pages. https://doi.org/10.1145/3428279",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[14,19,26,27,43,45]",
+      "technique": "tlp",
+      "sentence": ", databases and programs) [14,19,26,27,43,45].",
+      "context_before": "possible configuration options passed to the program is addressed by proposing a way of automatically inferring configuration options. The program’s configuration can then be tested using combinatorial testing [ 21]. However, all previous approaches focus on domain-specific improvements. There are programs that take non-traditional inputs. For instance, database management systems (DBMS) take databases as input; compilers take program source code as inputs. In order to test these systems automatically domain-specific approaches have been develop that generate these non-traditional inputs (i.e.",
+      "context_after": "However, these domain-specific approaches cannot be applied to test general programs, e.g., that use databases. Our work can be viewed as a general unification by fuzzing, without discrimination and in a coverage-guided manner, all environment resources that the program-under-test accesses. Chaos engineering. The key idea behind chaos engineering is to arbitrarily modify a running system and observe the system’s ability (i.e., robustness) to handle those faults [ 3]. Basiri et al [ 3] discuss Chaos engineering as an experimental discipline in which the software is viewed as a set of processes",
+      "section": "5.1 Performance analysis",
+      "page": 10,
+      "char_offset": 51375,
+      "cited_reference": {
+        "number": 26,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (nov 2020), 30 pages. https://doi.org/10.1145/3428279",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "This paper synthesises bug-inducing program environments -- the databases, files and devices a program accesses -- for bugs that no program input alone can expose. The approach is coverage-guided and mutation-based: it intercepts the system calls through which a program interacts with its environment, captures the resources touched, then mutates them, keeping any environment that increases coverage. The prototype found bugs in five of seven open-source projects including OpenSSL.",
+    "narrative": "Two citations, listing databases among the domains with dedicated fuzzing work.",
+    "roles": {
+      "M1": "background",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2310_06433.json b/_data/papers/paper_arxiv_2310_06433.json
new file mode 100644
index 0000000..3625c79
--- /dev/null
+++ b/_data/papers/paper_arxiv_2310_06433.json
@@ -0,0 +1,529 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:17Z",
+  "paper": {
+    "id": "paper:arxiv:2310.06433",
+    "title": "Retromorphic Testing: A New Approach to the Test Oracle Problem",
+    "authors": [
+      "Boxi Yu",
+      "Qiuyang Mang",
+      "Qingshuo Guo",
+      "Pinjia He"
+    ],
+    "year": 2023,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2310.06433",
+    "s2_paper_id": "cc095f9ce71982a8dcc0ea1b813a7486b3f6c962",
+    "url": "https://arxiv.org/abs/2310.06433",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2310.06433",
+      "retrieved_at": "2026-09-09T01:33:17Z",
+      "chars": 52127,
+      "content_sha256": "sha256:30ebe33f983c58524aeb8566173711f604c66057b0305562c5a26ef856e478a9"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 9,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2136
+      },
+      {
+        "number": "2",
+        "title": "PRELIMINARIES",
+        "start": 6767
+      },
+      {
+        "number": "2.1",
+        "title": "Test Oracles",
+        "start": 6783
+      },
+      {
+        "number": "2.2",
+        "title": "Terminology",
+        "start": 7997
+      },
+      {
+        "number": "3",
+        "title": "RETROMORPHIC TESTING",
+        "start": 10974
+      },
+      {
+        "number": "3.1",
+        "title": "Testing Principle",
+        "start": 10997
+      },
+      {
+        "number": "3.2",
+        "title": "Testing Modes",
+        "start": 11837
+      },
+      {
+        "number": "3.3",
+        "title": "Examples of Different Testing",
+        "start": 15698
+      },
+      {
+        "number": "4",
+        "title": "TESTING SCOPES",
+        "start": 24454
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Rui Abreu, João Paulo Fernandes, Luis Llana, and Guilherme Tavares. 2022. Metamorphic Testing of Oracle Quantum Programs. In 2022 IEEE/ACM 3rd International Workshop on Quantum Software Engineering (Q-SE). 16–23. https: //doi.org/10.1145/3528230.3529189",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Manindra Agrawal, Neeraj Kayal, and Nitin Saxena. 2004. PRIMES Is in P. Annals of Mathematics 160, 2 (2004), 781–793. http://www.jstor.org/stable/3597229",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Earl T. Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. 2015. The Oracle Problem in Software Testing: A Survey. IEEE Transactions on Software Engineering 41, 5 (may 2015), 507–525. https://doi.org/10.1109/TSE.2014.2372785",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Steven Bird and Edward Loper. 2004. NLTK: The Natural Language Toolkit. In Proceedings of the ACL Interactive Poster and Demonstration Sessions. Association for Computational Linguistics, Barcelona, Spain, 214–217. https://aclanthology. org/P04-3031",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Tsong Y Chen, Shing C Cheung, and Shiu Ming Yiu. 1998. Metamorphic testing: a new approach for generating next test cases. (1998). https://arxiv.org/abs/2002. 12543",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Tsong Yueh Chen, Fei-Ching Kuo, Huai Liu, Pak-Lok Poon, Dave Towey, T. H. Tse, and Zhi Quan Zhou. 2018. Metamorphic Testing: A Review of Challenges and Opportunities. ACM Comput. Surv. 51, 1, Article 4 (jan 2018), 27 pages. https://doi.org/10.1145/3143561",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Yuting Chen, Ting Su, and Zhendong Su. 2019. Deep Differential Testing of JVM Implementations. In Proceedings of the 41st International Conference on Software Engineering (Montreal, Quebec, Canada) (ICSE 2019). IEEE Press, 1257–1268. https://doi.org/10.1109/ICSE.2019.00127",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Yuting Chen, Ting Su, Chengnian Sun, Zhendong Su, and Jianjun Zhao. 2016. Coverage-Directed Differential Testing of JVM Implementations. In Proceedings of the 37th ACMSIGPLAN Conference on Programming Language Design and Implementation (Santa Barbara, CA, USA) (PLDI 2016). Association for Computing Machinery, New York, NY, USA, 85–99. https://doi.org/10.1145/2908080.2908095",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Haoran Peng, Chenyuan Yang, and Lingming Zhang. 2023. Large Language Models are Zero-Shot Fuzzers: Fuzzing DeepLearning Libraries via Large Language Models. arXiv:2212.14834 [cs.SE] https: //arxiv.org/abs/2212.14834",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Tianyu Gao, Xingcheng Yao, and Danqi Chen. 2021. SimCSE: Simple Contrastive Learning of Sentence Embeddings. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, Online and Punta Cana, Dominican Republic, 6894–6910. https://doi.org/10.18653/v1/2021.emnlp-main.552",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Wentao Gao, Jiayuan He, and Van-Thuan Pham. 2023. Metamorphic Testing of Machine Translation Models using Back Translation. In 2023 IEEE/ACM International Workshop on Deep Learning for Testing and Testing for Deep Learning (DeepTest). 1–8. https://doi.org/10.1109/DeepTest59248.2023.00008",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "W.E. Howden. 1978. Theoretical and Empirical Studies of Program Testing. IEEE Transactions on Software Engineering SE-4, 4 (1978), 293–298. https://doi.org/10. 1109/TSE.1978.231514",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (Seattle, WA, USA) (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 140–149. https://doi.org/10.1145/3597926.3598044",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler Validation via Equivalence modulo Inputs. In Proceedings of the 35th ACMSIGPLAN Conference on Programming Language Design and Implementation (Edinburgh, United Kingdom) (PLDI 2014). Association for Computing Machinery, New York, NY, USA, 216–226. https://doi.org/10.1145/2594291.2594334",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Cong Li, Yanyan Jiang, Chang Xu, and Zhendong Su. 2023. Validating JIT Compilers via Compilation Space Exploration. In Proceedings of the 29th Symposium on Operating Systems Principles (Koblenz, Germany) (SOSP 2023). Association for Computing Machinery, New York, NY, USA, 66–79. https://doi.org/10.1145/ 3600006.3613140",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software. DIGITAL TECHNICAL JOURNAL 10, 1 (1998), 100–107. https://api.semanticscholar.org/CorpusID: 14018070",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Matteo Paltenghi and Michael Pradel. 2023. MorphQ: Metamorphic Testing of the Qiskit Quantum Computing Platform. In 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE) (ICSE 2023). 2413–2424. https://doi. org/10.1109/ICSE48619.2023.00202",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Mauro Pezzè and Cheng Zhang. 2014. Chapter One-Automated Test Oracles: A Survey. Advances in Computers, Vol. 95. Elsevier, 1–48. https://doi.org/10.1016/ B978-0-12-800160-8.00001-2",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA) (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. https://doi.or",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (nov 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 2020). USENIX Association, 667–682. https://www.usenix. org/conference/osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "Manuel Rigger and Zhendong Su. 2022. Intramorphic Testing: A New Approach to the Test Oracle Problem. In Proceedings of the 2022 ACMSIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software (Auckland, New Zealand) (Onward! 2022). Association for Computing Machinery, New York, NY, USA, 128–136. https://doi.org/10.1145/3563835. 3567662",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 23,
+      "text": "R. L. Rivest, A. Shamir, and L. Adleman. 1978. A Method for Obtaining Digital Signatures and Public-Key Cryptosystems. Commun. ACM 21, 2 (feb 1978), 120–126. https://doi.org/10.1145/359340.359342",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Sergio Segura, Gordon Fraser, Ana B. Sanchez, and Antonio Ruiz-Cortés. 2016. A Survey on Metamorphic Testing. IEEE Transactions on Software Engineering 42, 9 (2016), 805–824. https://doi.org/10.1109/TSE.2016.2532875",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Mayank Sharma, Pingshi Yu, and Alastair F. Donaldson. 2023. RustSmith: Random Differential Compiler Testing for Rust. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (Seattle, WA, USA) (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 1483–1486. https://doi.org/10.1145/3597926.3604919",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "P.W. Shor. 1994. Algorithms for quantum computation: discrete logarithms and factoring. In Proceedings 35th Annual Symposium on Foundations of Computer Science. 124–134. https://doi.org/10.1109/SFCS.1994.365700",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In Proceedings of the 24rd International Conference on Very Large Data Bases (VLDB 1998). Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 618–622. https://dl.acm.org/ doi/10.5555/645924.671199",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Ting Su, Yichen Yan, Jue Wang, Jingling Sun, Yiheng Xiong, Geguang Pu, Ke Wang, and Zhendong Su. 2021. Fully Automated Functional Fuzzing of Android Apps for Detecting Non-Crashing Logic Bugs. Proc. ACM Program. Lang. 5, OOPSLA, Article 156 (oct 2021), 31 pages. https://doi.org/10.1145/3485533",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jingling Sun, Ting Su, Junxin Li, Zhen Dong, Geguang Pu, Tao Xie, and Zhendong Su. 2021. Understanding and Finding System Setting-Related Defects in Android Apps. In Proceedings of the 30th ACMSIGSOFT International Symposium on Software Testing and Analysis (Virtual, Denmark) (ISSTA 2021). Association for Computing Machinery, New York, NY, USA, 204–215. https: //doi.org/10.1145/3460319.3464806",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Zeyu Sun, Jie M. Zhang, Mark Harman, Mike Papadakis, and Lu Zhang. 2020. Automatic Testing and Improvement of Machine Translation. In Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering (Seoul, South Korea) (ICSE 2020). Association for Computing Machinery, New York, NY, USA, 974–985. https://doi.org/10.1145/3377811.3380420",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Muhammad Usman, Wenxi Wang, and Sarfraz Khurshid. 2021. TestMC: Testing Model Counters Using Differential and Metamorphic Testing. In Proceedings of the 35th IEEE/ACM International Conference on Automated Software Engineering (Virtual Event, Australia) (ASE 2020). Association for Computing Machinery, New York, NY, USA, 709–721. https://doi.org/10.1145/3324884.3416563",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Shuai Wang and Zhendong Su. 2020. Metamorphic Object Insertion for Testing Object Detection Systems. In Proceedings of the 35th IEEE/ACM International Conference on Automated Software Engineering (Virtual Event, Australia) (ASE 2020). Association for Computing Machinery, New York, NY, USA, 1053–1065. https://doi.org/10.1145/3324884.3416584",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Wenxuan Wang, Jingyuan Huang, Chang Chen, Jiazhen Gu, Jianping Zhang, Weibin Wu, Pinjia He, and Michael Lyu. 2023. Validating Multimedia Content Moderation Software via Semantic Fusion. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (Seattle, WA, USA) (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 576–588. https://doi.org/10.11",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2020. Validating SMT Solvers via Semantic Fusion. In Proceedings of the 41st ACMSIGPLAN Conference on Programming Language Design and Implementation (London, UK) (PLDI 2020). Association for Computing Machinery, New York, NY, USA, 718–730. https: //doi.org/10.1145/3385412.3385985",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Chenyuan Yang, Yinlin Deng, Jiayi Yao, Yuxing Tu, Hanchi Li, and Lingming Zhang. 2023. Fuzzing Automatic Differentiation in Deep-Learning Libraries. InProceedings of the 45th International Conference on Software Engineering (Melbourne, Victoria, Australia) (ICSE 2023). IEEE Press, 1174–1186. https: //doi.org/10.1109/ICSE48619.2023.00105",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Boxi Yu, Yiyan Hu, Qiuyang Mang, Wenhan Hu, and Pinjia He. 2023. Automated Testing and Improvement of Named Entity Recognition Systems. In Proceedings of the 31st ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Boxi Yu, Zhiqing Zhong, Jiaqi Li, Yixing Yang, Shilin He, and Pinjia He. 2023. ROME: Testing Image Captioning Systems via Recursive Object Melting. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (Seattle, WA, USA) (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 766–778. https://doi.org/10.1145/3597926.3598094",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis (Virtual, South Korea) (ISSTA 2022). Association for Computing Machinery, New York, NY, USA, 302–313. ",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 13,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (Seattle, WA, USA) (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 140–149. https://doi.org/10.1145/3597926.3598044",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 19,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA) (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. https://doi.org/10.1145/3368089.3409710 Retromorphic Testing Conference’17, July 2017, Washington, DC, USA",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (nov 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 21,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 2020). USENIX Association, 667–682. https://www.usenix. org/conference/osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 22,
+      "text": "Manuel Rigger and Zhendong Su. 2022. Intramorphic Testing: A New Approach to the Test Oracle Problem. In Proceedings of the 2022 ACMSIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software (Auckland, New Zealand) (Onward! 2022). Association for Computing Machinery, New York, NY, USA, 128–136. https://doi.org/10.1145/3563835. 3567662",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 22]",
+      "technique": null,
+      "sentence": "In addition to these black-box methodologies, Rigger and Su [ 22] recently introduced intramorphic testing, a white-box automated testing methodology, which modifies a system component and constructs a test oracle that can relate the output of the original and modified systems given the same input.",
+      "context_before": "effective high-level ideas. In differential testing, multiple systems or software versions are expected to exhibit identical behaviors for the shared functionalities. Discrepancies in output for identical inputs among these systems indicate potential bugs in one or more systems. Metamorphic testing generates new test inputs based on existing input-output pairs, with the output of the generated input being predictable based on the relationship between the original input and the generated one. The detection of deviations in output relationships can highlight potential bugs in the target systems.",
+      "context_after": "This paper introduces Retromorphic Testing, a general black-box methodology to tackle the challenges in test oracle construction. The core idea of Retromorphic Testing is inspired by mathematical relationships between functions and their inverse functions, expressed as 𝑓−1(𝑓(𝑥))=𝑥. It employs an auxiliary program to reverse the output of the software, transforming it back to its original input format. Different from differential testing and metamorphic testing, which typically involve a system or systems with equivalent functionalities, Retromorphic Testing employs a “dual-program structure”,",
+      "section": "1 INTRODUCTION",
+      "page": 1,
+      "char_offset": 3688,
+      "cited_reference": {
+        "number": 22,
+        "text": "Manuel Rigger and Zhendong Su. 2022. Intramorphic Testing: A New Approach to the Test Oracle Problem. In Proceedings of the 2022 ACMSIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software (Auckland, New Zealand) (Onward! 2022). Association for Computi",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 13,19,20]",
+      "technique": "norec",
+      "sentence": "Metamorphic testing is also used for testing various systems such as compilers [ 14,15], database engines [ 13,19,20], SMT solvers [ 34], Android apps [ 28,29], quantum computing platforms [ 1,17], and AI systems [ 30,32,33,36,37].",
+      "context_before": "th its effectiveness, differential testing has been applied to a variety of domains, such as Rust compilers [ 25], model counters [ 31], Java Virtual Machines (JVMs) [ 7,8], database engines [ 27,38], and deep-learning libraries [ 9,35]. Metamorphic testing is a technique where the tester identifies certain properties that should remain invariant under specific transformations of the input. If these properties change after the transformation, a defect might be signaled. This approach is particularly useful in scenarios where sufficient test oracles not be available or are too expensive to use.",
+      "context_after": "In Fig. 1 (a) and (b), we give the workflow of differential testing and metamorphic testing, respectively. Differential testing uses a fixed input𝐼for programs 𝑃and𝑃′of the same functionality and validates whether their outputs, i.e.,𝑂=𝑃(𝐼)and𝑂′=𝑃′(𝐼), are equal. When there is a discrepancy between 𝑂and𝑂′, it would indicate bugs in either 𝑃or𝑃′. Unlike differential testing which uses the same input for different programs, metamorphic testing mutates the input 𝐼to𝐼′and feeds them into one given program. It is based on the idea of defining metamorphic relations (MRs) that capture the expected r",
+      "section": "2.2 Terminology",
+      "page": 2,
+      "char_offset": 10075,
+      "cited_reference": {
+        "number": 19,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": "In the context of Database Management Systems (DBMS), Rigger and Su introduced the concept of Pivoted Query Synthesis (PQS) [21] as an effective testing approach to uncover logic bugs in these systems.",
+      "context_before": "n input Java code public static void main(String[] arrstring) {int n = 5;int n2 = 7;int n3 = n + n2;System.out.println(\"Sum: \" + n3);}M1': TheJava codedecompiled from binary code P: Java compiler Q: The target decompilerM2: The binary codegenerated by Java compilerTest oracle: The excution output of the two Java codes(M1 andM1') should be identical public static void main(String[] args) {int num1 = 5;int num2 = 7;int sum = num1 + num2;System.out.println(\"Sum: \" + sum);} Figure 3: An illustrative example about using Retromorphic Testing to test decompiler. Example 3: Database Management System.",
+      "context_after": "Although the original paper describes PQS as a testing technique for a specific system, we think it can be regarded as an instance of the general Retromorphic Testing methodology. We also discussed with one author of PQS and he agrees that PQS should be categorized as a Retromorphic Testing technique. As depicted in Fig.2, the target database serves as the backward program𝑄in Retromorphic Testing. The PQS testing approach initiates with the generation of a pivot row (e.g.,[t0.c0: 3, t0.c1: TRUE, t1.c0: -5] in Fig.2), which is in the modality of rows. Subsequently, a simple program 𝑃based on A",
+      "section": "4 TESTING SCOPES",
+      "page": 7,
+      "char_offset": 32573,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Although the original paper describes PQS as a testing technique for a specific system, we think it can be regarded as an instance of the general Retromorphic Testing methodology.",
+      "context_before": "target decompilerM2: The binary codegenerated by Java compilerTest oracle: The excution output of the two Java codes(M1 andM1') should be identical public static void main(String[] args) {int num1 = 5;int num2 = 7;int sum = num1 + num2;System.out.println(\"Sum: \" + sum);} Figure 3: An illustrative example about using Retromorphic Testing to test decompiler. Example 3: Database Management System. In the context of Database Management Systems (DBMS), Rigger and Su introduced the concept of Pivoted Query Synthesis (PQS) [21] as an effective testing approach to uncover logic bugs in these systems.",
+      "context_after": "We also discussed with one author of PQS and he agrees that PQS should be categorized as a Retromorphic Testing technique. As depicted in Fig.2, the target database serves as the backward program𝑄in Retromorphic Testing. The PQS testing approach initiates with the generation of a pivot row (e.g.,[t0.c0: 3, t0.c1: TRUE, t1.c0: -5] in Fig.2), which is in the modality of rows. Subsequently, a simple program 𝑃based on AST will be utilized to generate an SQL query based on the pivot row, ensuring that the data of the pivot row is retrieved by the query. This query exists in the modality of SQL que",
+      "section": "4 TESTING SCOPES",
+      "page": 7,
+      "char_offset": 32775,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "We also discussed with one author of PQS and he agrees that PQS should be categorized as a Retromorphic Testing technique.",
+      "context_before": "] args) {int num1 = 5;int num2 = 7;int sum = num1 + num2;System.out.println(\"Sum: \" + sum);} Figure 3: An illustrative example about using Retromorphic Testing to test decompiler. Example 3: Database Management System. In the context of Database Management Systems (DBMS), Rigger and Su introduced the concept of Pivoted Query Synthesis (PQS) [21] as an effective testing approach to uncover logic bugs in these systems. Although the original paper describes PQS as a testing technique for a specific system, we think it can be regarded as an instance of the general Retromorphic Testing methodology.",
+      "context_after": "As depicted in Fig.2, the target database serves as the backward program𝑄in Retromorphic Testing. The PQS testing approach initiates with the generation of a pivot row (e.g.,[t0.c0: 3, t0.c1: TRUE, t1.c0: -5] in Fig.2), which is in the modality of rows. Subsequently, a simple program 𝑃based on AST will be utilized to generate an SQL query based on the pivot row, ensuring that the data of the pivot row is retrieved by the query. This query exists in the modality of SQL queries and is then passed to the target database, reverting the modality back into rows. The underlying Retromorphic Relation",
+      "section": "4 TESTING SCOPES",
+      "page": 7,
+      "char_offset": 32955,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "The PQS testing approach initiates with the generation of a pivot row (e.",
+      "context_before": "the context of Database Management Systems (DBMS), Rigger and Su introduced the concept of Pivoted Query Synthesis (PQS) [21] as an effective testing approach to uncover logic bugs in these systems. Although the original paper describes PQS as a testing technique for a specific system, we think it can be regarded as an instance of the general Retromorphic Testing methodology. We also discussed with one author of PQS and he agrees that PQS should be categorized as a Retromorphic Testing technique. As depicted in Fig.2, the target database serves as the backward program𝑄in Retromorphic Testing.",
+      "context_after": "g.,[t0.c0: 3, t0.c1: TRUE, t1.c0: -5] in Fig.2), which is in the modality of rows. Subsequently, a simple program 𝑃based on AST will be utilized to generate an SQL query based on the pivot row, ensuring that the data of the pivot row is retrieved by the query. This query exists in the modality of SQL queries and is then passed to the target database, reverting the modality back into rows. The underlying Retromorphic Relation here asserts that the result obtained from the database should encompass the pivot row. Example 4: Java decompiler. While the Java compiler is renowned for its reliability",
+      "section": "4 TESTING SCOPES",
+      "page": 7,
+      "char_offset": 33176,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:21:29Z",
+    "is_model_written": true,
+    "summary": "Retromorphic Testing proposes a black-box test oracle inspired by inverse functions. It pairs the program under test with an auxiliary program to form a forward and a backward program: input data is processed by the forward program and its output converted back into the original input format by the backward program, after which the relation between the initial input and the transformed output is checked. Either program can play either role, giving different testing modes.",
+    "narrative": "PQS is the paper's worked example of its own concept: the authors argue that although PQS was described as a technique for a specific system, it is an instance of retromorphic testing, and record that an author of PQS agreed with that categorisation.",
+    "roles": {
+      "M1": "definition",
+      "M2": "extension",
+      "M3": "background",
+      "M4": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "yes",
+        "mention_ids": [
+          "M2",
+          "M3"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "Although the original paper describes PQS as a testing technique for a specific system, we think it can be regarded as an instance of the general Retromorphic Testing methodology.",
+            "section": "4 TESTING SCOPES",
+            "page": 7
+          },
+          {
+            "mention_id": "M3",
+            "sentence": "We also discussed with one author of PQS and he agrees that PQS should be categorized as a Retromorphic Testing technique.",
+            "section": "4 TESTING SCOPES",
+            "page": 7
+          }
+        ],
+        "reasoning": "M2 reframes PQS as an instance of the general retromorphic testing concept the paper introduces, and M3 records that an author of PQS agreed. Generalising an existing technique into a broader class is the paper's contribution, not a baseline it runs.",
+        "techniques": [
+          "pqs"
+        ]
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2311_06728.json b/_data/papers/paper_arxiv_2311_06728.json
new file mode 100644
index 0000000..f23114a
--- /dev/null
+++ b/_data/papers/paper_arxiv_2311_06728.json
@@ -0,0 +1,2011 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:06:47Z",
+  "paper": {
+    "id": "paper:arxiv:2311.06728",
+    "title": "A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison",
+    "authors": [
+      "Xiyue Gao",
+      "Zhuang Liu",
+      "Jia Cui",
+      "Hui Li",
+      "Hui Zhang",
+      "Kewei Wei",
+      "Kankan Zhao"
+    ],
+    "year": 2023,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2311.06728",
+    "s2_paper_id": "5ff9885778f8ec824beaeb6802af5c82e299f702",
+    "url": "https://arxiv.org/abs/2311.06728",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2311.06728",
+      "retrieved_at": "2026-09-10T15:06:47Z",
+      "chars": 101507,
+      "content_sha256": "sha256:3d1d5045fe9645e76b1b42fb32159d8f75d10ffd45a682b1aa149de65137ea09"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 34,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2214
+      },
+      {
+        "number": "2",
+        "title": "PRELIMINARY",
+        "start": 11112
+      },
+      {
+        "number": "2.1",
+        "title": "Notations",
+        "start": 11261
+      },
+      {
+        "number": "2.2",
+        "title": "Basic Definitions",
+        "start": 12159
+      },
+      {
+        "number": "3",
+        "title": "FUZZING STEP-BASED TAXONOMY",
+        "start": 16991
+      },
+      {
+        "number": "3.1",
+        "title": "Test Case Generator",
+        "start": 18480
+      },
+      {
+        "number": "3.2",
+        "title": "Oracle-based Comparator",
+        "start": 34366
+      },
+      {
+        "number": "3.3",
+        "title": "Execution Feedback",
+        "start": 41371
+      },
+      {
+        "number": "3.4",
+        "title": "Query Reducer",
+        "start": 46489
+      },
+      {
+        "number": "4.1",
+        "title": "Overall Fuzzing",
+        "start": 48818
+      },
+      {
+        "number": "4.2",
+        "title": "Transaction Testing",
+        "start": 55916
+      },
+      {
+        "number": "4.3",
+        "title": "Optimizer Testing",
+        "start": 57682
+      },
+      {
+        "number": "4.4",
+        "title": "Executor Testing",
+        "start": 59942
+      },
+      {
+        "number": "5.1",
+        "title": "Database Instances and Test Cases",
+        "start": 61893
+      },
+      {
+        "number": "5.2",
+        "title": "Open-source Fuzzers",
+        "start": 62435
+      },
+      {
+        "number": "5.3",
+        "title": "Evaluation Metrics",
+        "start": 62712
+      },
+      {
+        "number": "5.4",
+        "title": "Logic Bugs Detection Comparison",
+        "start": 65199
+      },
+      {
+        "number": "5.5",
+        "title": "Crash Detection Comparison",
+        "start": 70161
+      },
+      {
+        "number": "5.6",
+        "title": "Performance Bugs Detection Evaluation",
+        "start": 71330
+      },
+      {
+        "number": "6.1",
+        "title": "Improved Constraint Solving",
+        "start": 73468
+      },
+      {
+        "number": "6.2",
+        "title": "Component-oriented Fuzzing",
+        "start": 74876
+      },
+      {
+        "number": "6.3",
+        "title": "Fuzzing of Modern Database Systems",
+        "start": 76044
+      },
+      {
+        "number": "6.4",
+        "title": "Improved Space Exploration Capabilities",
+        "start": 78124
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "1996. TPC-H. https://www.tpc.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2011. Data science revealed: A data-driven glimpse into the burgeoning new field. https://www.ndm.net/datawarehouse/pdf/EMC-Data_Science_ Study_White_Paper.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2015. AFL: American Fuzzy Lop. http://lcamtuf.coredump.cx/afl/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2015. Jepsen. https://github.com/jepsen-io/jepsen.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2015. SQLsmith. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2016. Honggfuzz. https://google.github.io/honggfuzz/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2016. OSS-Fuzz: Continuous Fuzzing for Open Source Software. https://github.com/google/oss-fuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2017. LibFuzzer-A Library For Coverage-guided Fuzz Testing. http://llvm.org/docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2019. American Fuzzy Lop. http://lcamtuf.coredump.cx/afl.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2019. Documentation of Alloy SAT solver. https://alloytools.org/documentation.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2019. Technical \"Whitepaper\" For Afl-fuzz. http://lcamtuf.coredump.cx/afl/technical_details.txt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2021. CockroachDB Performance Bug Reports. https://github.com/cockroachdb/cockroach/issues?q=performance.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "2021. PostgreSQL Performance Bug Reports. https://www.postgresql.org/search/?m=1&q=performance&l=8&d=-1&s=r.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "2021. SCOTT schema. https://www.orafaq.com/wiki/SCOTT.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "2023. MySQL Bugs Home. https://bugs.mysql.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. 2010. Automated SQL Query Generation for Systematic Testing of Database Engines. In International Conference on Automated Software Engineering. ACM, 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Arul Ajmani, Aayush Shah, Alexander Shraer, Adam Storm, Rebecca Taft, Oliver Tan, and Nathan VanBenschoten. 2022. A Demonstration of Multi-Region CockroachDB. Very Large Data Bases Endowment 15, 12 (2022), 3610–3613.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In International Conference on Software Engineering. ACM, 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 19,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. 2007. A Genetic Approach for Random Testing of Database Systems. In International Conference on Very Large Data Bases. VLDB Endowment, 1243–1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "B. Beizer. 1992. Software testing techniques. Software Testing Verification and Reliability 2, 4 (1992), 215–216.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Donald A Berry and Bert Fristedt. 1985. Bandit problems: sequential allocation of experiments (Monographs on statistics and applied probability). London: Chapman and Hall 5, 71-87 (1985), 7–7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Tim Blazytko, Matt Bishop, Cornelius Aschermann, Justin Cappos, Moritz Schlögel, Nadia Korshun, Ali Abbasi, Marco Schweighauser, Sebastian Schinzel, Sergej Schumilo, et al .2019. GRIMOIRE: Synthesizing Structure While Fuzzing. In USENIX Conference on Security Symposium. USENIX Association, 1985–2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Nicolas Bruno and Surajit Chaudhuri. 2005. Flexible Database Generators. In International Conference on Very Large Data Bases. VLDB Endowment, 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "N. Bruno, S. Chaudhuri, and D. Thomas. 2006. Generating Queries with Cardinality Constraints for DBMS Testing. Transactions on Knowledge and Data Engineering 18, 12 (2006), 1721–1725. Manuscript submitted to ACM 32 AUTHOR et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Marcel Böhme, Van-Thuan Pham, Manh-Dung Nguyen, and Abhik Roychoudhury. 2017. Directed Greybox Fuzzing. In Conference on Computer and Communications Security. ACM, 2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. 2016. Coverage-based Greybox Fuzzing as Markov Chain. In Conference on Computer and Communications Security. ACM, 1032–1043.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Tsong Y Chen, Shing C Cheung, and Shiu Ming Yiu. 1998. Metamorphic Testing: A New Approach for Generating Next Test Cases. arXiv preprint arXiv:2002.12543 (1998).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Xinyue Chen, Chenglong Wang, and Alvin Cheung. 2020. Testing Query Execution Engines with Mutations. In Workshop on Testing Database Systems. ACM, 1–5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Shumo Chu, Chenglong Wang, Konstantin Weitz, and Alvin Cheung. 2017. Cosette: An Automated Prover for SQL. In Conference on Innovative Data Systems Research.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "James C Corbett, Jeffrey Dean, Michael Epstein, Andrew Fikes, Christopher Frost, Jeffrey John Furman, Sanjay Ghemawat, Andrey Gubarev, Christopher Heiser, Peter Hochschild, et al .2013. Spanner: Google’s Globally Distributed Database. ACM Transactions on Computer Systems 31, 3 (2013), 1–22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Ziyu Cui, Wensheng Dou, Qianwang Dai, Jiansen Song, Wei Wang, Jun Wei, and Dan Ye. 2022. Differentially Testing Database Transactions for Fun and Profit. In International Conference on Automated Software Engineering. ACM, 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Wensheng Dou, Ziyu Cui, Qianwang Dai, Jiansen Song, Dong Wang, Yu Gao, Wei Wang, Jun Wei, Lei Chen, Hanmo Wang, et al .2023. Detecting Isolation Bugs via Transaction Oracle Construction. In International Conference on Software Engineering. IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. In International Conference on Automated Software Engineering. ACM, 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "César A. Galindo-Legaria, Stefano Stefani, and Florian Waas. 2004. Query Processing for SQL Updates. In International Conference on Management of Data. ACM, 844–849.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Leo Giakoumakis and César A Galindo-Legaria. 2008. Testing SQL Server’s Query Optimizer: Challenges, Techniques and Experiences. Data Engineering Bulletin 31, 1 (2008), 36–43.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Patrice Godefroid, Adam Kiezun, and Michael Y. Levin. 2008. Grammar-based whitebox fuzzing. In Conference on Programming Language Design and Implementation. ACM, 206–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Eugene Goldberg and Yakov Novikov. 2007. BerkMin: A Fast and Robust SAT-solver. Discrete Applied Mathematics 155, 12 (2007), 1549–1561.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Goetz Graefe. 1993. Query Evaluation Techniques for Large Databases. Computing Surveys 25, 2 (1993), 73–169.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Goetz Graefe. 1995. The Cascades Framework for Query Optimization. Data Engineering Bulletin 18, 3 (1995), 19–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Jim Gray, Prakash Sundaresan, Susanne Englert, Ken Baclawski, and Peter J Weinberger. 1994. Quickly Generating Billion-Record Synthetic Databases. In International Conference on Management of Data. ACM, 243–252.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Zhongxian Gu, Mohamed A. Soliman, and Florian M Waas. 2012. Testing the Accuracy of Query Optimizers. In Workshop on Testing Database Systems. ACM, 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Kenneth Houkjær, Kristian Torp, and Rico Wind. 2006. Simple and Realistic Data Generation. In International Conference on Very Large Data Bases. VLDB Endowment, 1243–1246.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "William E Howden. 1978. Theoretical and empirical studies of program testing. Transactions on Software Engineering SE-4, 4 (1978), 293–298.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Dongxu Huang, Qi Liu, Qiu Cui, Zhuhe Fang, Xiaoyu Ma, Fei Xu, Li Shen, Liu Tang, Yuxing Zhou, Menglong Huang, et al .2020. TiDB: A Raft-Based HTAP Database. Very Large Data Bases Endowment 13, 12 (2020), 3072–3084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "T. Hunt. 2020. The 773 Million Record \"Collection 1\" Data Breach. https://www.troyhunt.com/the-773-million-record-collection-1-data-reach/ ,January2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Toshihide Ibaraki and Tiko Kameda. 1984. On the Optimal Nesting Order for Computing N-Relational Joins. Transactions on Database Systems 9, 3 (1984), 482–502.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Daniel Jackson. 2002. Alloy: A Lightweight Object Modelling Notation. Transactions on Software Engineering and Methodology 11, 2 (2002), 256–290.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Kamala Ramasubramani Jayakumar and Alain Abran. 2013. A Survey of Software Test Estimation Techniques. Journal of Software Engineering and Applications 6, 10 (2013), 47–52.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. In USENIX Security Symposium. USENIX Association, 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Matt Jibson. 2019. SQLsmith: Randomized sql testing in cockroachdb. https://www.cockroachlabs.com/blog/sqlsmith-randomized-sql-testing/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. Very Large Data Bases Endowment 13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Sarfraz Khurshid and Darko Marinov. 2004. TestEra: Specification-based testing of Java programs using SAT. Automated Software Engineering 11 (2004), 403–434.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Donghun Lee, Sang K Cha, and Arthur H Lee. 2011. A Performance Anomaly Detection and Analysis Framework for DBMS Development. Transactions on Knowledge and Data Engineering 24, 8 (2011), 1345–1360.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2015. How Good Are Query Optimizers, Really? Very Large Data Bases Endowment 9, 3 (2015), 204–215. Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 33",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Caroline Lemieux and Koushik Sen. 2018. FairFuzz: A Targeted Mutation Strategy for Increasing Greybox Fuzz Testing Coverage. In International Conference on Automated Software Engineering. ACM, 475–485.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Guoliang Li, Xuanhe Zhou, Shifu Li, and Bo Gao. 2019. QTune: A Query-Aware Database Tuning System with Deep Reinforcement Learning. Very Large Data Bases Endowment 12, 12 (2019), 2118–2130.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Guoliang Li, Xuanhe Zhou, Ji Sun, Xiang Yu, Yue Han, Lianyuan Jin, Wenbo Li, Tianqing Wang, and Shifu Li. 2021. OpenGauss: An Autonomous Database System. Very Large Data Bases Endowment 14, 12 (2021), 3028–3042.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Zhan Li, Olga Papaemmanouil, and Mitch Cherniack. 2016. Optmark: A Toolkit for Benchmarking Query Optimizers. In Information and Knowledge Management. ACM, 2155–2160.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-Oriented DBMS Fuzzing. In International Conference on Data Engineering. IEEE, 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of DBMS with Coverage-based Guidance. In USENIX Security Symposium. USENIX Association, 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Wei Lin, Ziyue Hua, Luyao Ren, Zongyang Li, Lu Zhang, and Tao Xie. 2022. GDsmith: Detecting Bugs in Graph Database Engines. arXiv preprint arXiv:2206.08530 (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic Detection of Performance Bugs in Database Systems Using Equivalent Queries. In International Conference on Software Engineering. ACM, 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Eric Lo, Carsten Binnig, Donald Kossmann, M Tamer Özsu, and Wing-Kai Hon. 2010. A Framework for Testing DBMS Features. Very Large Data Bases Endowment 19 (2010), 203–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Jiaheng Lu and Irena Holubová. 2019. Multi-Model Databases: A New Journey to Handle the Variety of Data. Comput. Surveys 52, 3 (2019), 1–38.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Minghua Ma, Zheng Yin, Shenglin Zhang, Sheng Wang, Christopher Zheng, Xinhao Jiang, Hanwen Hu, Cheng Luo, Yilin Li, Nengjun Qiu, Feifei Li, Changcheng Chen, and Dan Pei. 2020. Diagnosing Root Causes of Intermittent Slow Queries in Cloud Databases. Very Large Data Bases Endowment 13, 8 (2020), 1176–1189.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Ryan Marcus, Parimarjan Negi, Hongzi Mao, Nesime Tatbul, Mohammad Alizadeh, and Tim Kraska. 2021. Bao: Making Learned Query Optimization Practical. In International Conference on Management of Data. ACM, 1275–1288.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Darko Marinov. 2010. Session Details: Technical Session 8: Concurrency and Differential Testing. In International Symposium on Software Testing and Analysis. ACM.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "William M McKeeman. 1998. Differential Testing for Software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Phil Mcminn, Chris J. Wright, and Gregory M. Kapfhammer. 2015. The Effectiveness of Test Coverage Criteria for Relational Database Schema Integrity Constraints. Transactions on Software Engineering and Methodology 25, 1 (2015), 1–49.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Ruijie Meng, George Pîrlea, Abhik Roychoudhury, and Ilya Sergey. 2023. Greybox Fuzzing of Distributed Systems. arXiv preprint arXiv:2305.02601 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Kaiming Mi, Chunxi Zhang, Weining Qian, and Rong Zhang. 2021. Artemis: An Automatic Test Suite Generator for Large Scale OLAP Database. In Benchmarking, Measuring, and Optimizing: Third BenchCouncil International Symposium. Springer, 74–89.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Barton P. Miller, Lars Fredriksen, and Bryan So. 1990. An Empirical Study of the Reliability of UNIX Utilities. Commun. ACM 33, 12 (1990), 32–44.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. 2008. Generating targeted queries for database testing. In International Conference on Management of Data. ACM, 499–510.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Jayashree Mohan, Ashlie Martinez, Soujanya Ponnapalli, Pandian Raju, and Vijay Chidambaram. 2018. Finding Crash-Consistency Bugs with Bounded Black-Box Crash Testing. In USENIX Symposium on Operating Systems Design and Implementations. USENIX Association, 33–50.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "MySQL. 2023. The MySQL Test Framework. https://dev.mysql.com/doc/dev/mysql-server/latest/PAGE_MYSQL_TEST_RUN.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Andrea Neufeld, Guido Moerkotte, and Peter C Loekemann. 1993. Generating Consistent Test Data: Restricting the Search Space by A Generator Formula. Very Large Data Bases Endowment 2 (1993), 173–213.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "ShaTransactions on Knowledge, Data Engineeringdi Abdul Khalek, Bassem Elkarablieh, Yai O. Laleye, and Sarfraz Khurshid. 2008. Query-Aware Test Generation Using a Relational Constraint Solver. In International Conference on Automated Software Engineering. IEEE, 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Rasha Osman and William J. Knottenbelt. 2012. Database System Performance Evaluation Models: A Survey. Performance Evaluation 69, 10 (2012), 471–493.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Carlos Pacheco, Shuvendu K. Lahiri, Michael D. Ernst, and Thomas Ball. 2007. Feedback-Directed Random Test Generation. In International Conference on Software Engineering. IEEE, 75–84.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Andrew Pavlo, Gustavo Angulo, Joy Arulraj, Haibin Lin, Jiexi Lin, Lin Ma, Prashanth Menon, Todd C Mowry, Matthew Perron, Ian Quah, et al. 2017. Self-Driving Database Management Systems. In Conference on Innovative Data Systems Research. 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "PingCap. 2023. go randgen. https://github.com/pingcap/go-randgen.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Meikel Poess and John M. Stephens. 2004. Generating Thousand Benchmark Queries in Seconds. In International Conference on Very Large Data Bases. VLDB Endowment, 1045–1053.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "PostgreSQL. 2023. Regression Tests. https://www.postgresql.org/docs/current/regress.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM. Manuscript submitted to ACM 34 AUTHOR et al.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 85,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 86,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 87,
+      "text": "Manuel Rigger and Zhendong Su. 2022. Intramorphic Testing: A New Approach to the Test Oracle Problem. In International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software. ACM, 128–136.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 88,
+      "text": "Shetal Shah, S. Sudarshan, Suhas Kajbaje, Sandeep Patidar, Bhanu Pratap Gupta, and Devang Vira. 2011. Generating Test Data for Killing SQL Mutants: A Constraint-Based Approach. In International Conference on Data Engineering. IEEE, 1175–1186.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "Donald R Slutz. 1998. Massive Stochastic Testing of SQL. In International Conference on Very Large Data Bases. VLDB Endowment, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. In International Conference on Software Engineering. IEEE, 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "SQLite. 2023. How SQLite Is Tested. https://sqlite.org/testing.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "Keith Stobie. 2005. Too Darned Big to Test: Testing Large Systems is a Daunting Task, but There Are Steps We Can Take to Ease the Pain. Queue 3, 1 (2005), 30–37.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "Rebecca Taft, Irfan Sharif, Andrei Matei, Nathan VanBenschoten, Jordan Lewis, Tobias Grieger, Kai Niemi, Andy Woods, Anne Birzin, Raphael Poss, et al. 2020. CockroachDB: The Resilient Geo-Distributed SQL Database. In International Conference on Management of Data. ACM, 1493–1509.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting Logic Bugs of Join Optimizations in DBMS. International Conference on Management of Data 1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "P. Thévenod-Fosse and H. Waeselynck. 1993. STATEMATE Applied to Statistical Software Testing. In International Symposium on Software Testing and Analysis. ACM, 99–109.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "Valter Uotila, Jiaheng Lu, Dieter Gawlick, Zhen Hua Liu, Souripriya Das, and Gregory Pogossiants. 2021. MultiCategory: Multi-Model Query Processing Meets Category Theory and Functional Programming. Very Large Data Bases Endowment 14, 12 (2021), 2663–2666.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "Patrick Valduriez and Scott Danforth. 1992. Functional SQL (FSQL), an SQL Upward-Compatible Database Programming Language. Information Sciences 62, 3 (1992), 183–203.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 98,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. 2019. Superion: Grammar-Aware Greybox Fuzzing. In International Conference on Software Engineering. IEEE, 724–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 99,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In International Conference on Software Engineering: Software Engineering in Practice. IEEE, 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 100,
+      "text": "Yanhao Wang, Xiangkun Jia, Yuwei Liu, Kyle Zeng, Tiffany Bao, Dinghao Wu, and Purui Su. 2020. Not All Coverage Measurements Are Equal: Fuzzing by Coverage Accounting for Input Prioritization. In Network and Distributed System Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 101,
+      "text": "Shihao Wen, Peng Jia, Pin Yang, and Chi Hu. 2023. Squill: Testing DBMS with Correctness Feedback and Accurate Instantiation. Applied Sciences 13, 4 (2023), 2519.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 102,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. 2022. Unicorn: Detect Runtime Errors in Time-Series Databases with Hybrid Input Synthesis. In International Symposium on Software Testing and Analysis. ACM, 251–262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 103,
+      "text": "Jiaqi Yan, Qiuye Jin, Shrainik Jain, Stratis D. Viglas, and Allison Lee. 2018. Snowtrail: Testing with Production Queries on a Cloud Database. In Workshop on Testing Database Systems. ACM, 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 104,
+      "text": "Man Yang, Mingwang Zhang, Tao Leng, Weize Dong, and Yu Yan. 2021. Design and Implementation of Benchmarks for Multimodal Databases. In International Conference on Industrial Application of Artificial Intelligence. IEEE, 490–494.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 105,
+      "text": "Gongsheng Yuan, Jiaheng Lu, Shuxun Zhang, and Zhengtong Yan. 2021. Storing Multi-Model Data in RDBMSs Based on Reinforcement Learning. InInternational Conference on Information & Knowledge Management. ACM, 3608–3611.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 106,
+      "text": "Chao Zhang, Jiaheng Lu, Pengfei Xu, and Yuxing Chen. 2019. UniBench: A Benchmark for Multi-model Database Management Systems. In Performance Evaluation and Benchmarking for the Era of Artificial Intelligence. Springer, 7–23.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 107,
+      "text": "Yushan Zhang, Peisen Yao, Rongxin Wu, and Charles Zhang. 2021. Duplicate-sensitivity Guided Transformation Synthesis for DBMS Correctness Bug Detection. arXiv preprint arXiv:2107.03660 (2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 108,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing. In International Symposium on Software Testing and Analysis. ACM, 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 109,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Conference on Computer and Communications Security. ACM, 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 110,
+      "text": "Zenghui Zhou, Zheng Zheng, Tsong Yueh Chen, Jinyi Zhou, and Kun Qiu. 2021. Follow-up Test Cases are Better Than Source Test Cases in Metamorphic Testing: A Preliminary Study. In International Workshop on Metamorphic Testing. IEEE, 69–74. Received 20 February 2007; revised 12 March 2009; accepted 5 June 2009 Manuscript submitted to ACM",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 18,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In International Conference on Software Engineering. ACM, 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 84,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM. Manuscript submitted to ACM 34 AUTHOR et al.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 85,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 86,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 87,
+      "text": "Manuel Rigger and Zhendong Su. 2022. Intramorphic Testing: A New Approach to the Test Oracle Problem. In International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software. ACM, 128–136.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "proposed several new fuzzing methods, including NoREC [ 84], TLP [ 85] and PQS [ 86].",
+      "context_before": "testing [ 16] was proposed as another way to obtain the test oracle in DBMS fuzzing, which provides the ground truth for the test cases. In 2015, SQLSmith [ 5] was released, which still generates statements based on predefined rules, and can also connect to reference databases to serve as test oracles. Thanks to the open-source factor, compared to the SQL Server fuzzing tool, SQLSmith has rapidly attracted attention from both academic and industrial. In 2020, Squirrel [ 109] first introduced execution feedback to improve code coverage in crash detection fuzzing. In the same year, Rigger et al.",
+      "context_after": "NoREC and TLP apply metamorphic testing to the database domain, enabling fuzzing on a single DBMS. PQS is a constraint-solving testing method, and its oracle only requires that the execution result contain one row of the solving outcome, thereby accelerating the solving process Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 3 compared to the former approach, which requires matching all the rows. In 2022, Liang et al. introduced SQLRight [ 60], a metamorphic testing method combined with execution feedba",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 5780,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC and TLP apply metamorphic testing to the database domain, enabling fuzzing on a single DBMS.",
+      "context_before": "hich provides the ground truth for the test cases. In 2015, SQLSmith [ 5] was released, which still generates statements based on predefined rules, and can also connect to reference databases to serve as test oracles. Thanks to the open-source factor, compared to the SQL Server fuzzing tool, SQLSmith has rapidly attracted attention from both academic and industrial. In 2020, Squirrel [ 109] first introduced execution feedback to improve code coverage in crash detection fuzzing. In the same year, Rigger et al. proposed several new fuzzing methods, including NoREC [ 84], TLP [ 85] and PQS [ 86].",
+      "context_after": "PQS is a constraint-solving testing method, and its oracle only requires that the execution result contain one row of the solving outcome, thereby accelerating the solving process Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 3 compared to the former approach, which requires matching all the rows. In 2022, Liang et al. introduced SQLRight [ 60], a metamorphic testing method combined with execution feedback, and is scalable to allow new oracles. In 2023, Tang et al. presented a novel constraint-solving",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 5866,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS is a constraint-solving testing method, and its oracle only requires that the execution result contain one row of the solving outcome, thereby accelerating the solving process Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 3 compared to the former approach, which requires matching all the rows.",
+      "context_before": "generates statements based on predefined rules, and can also connect to reference databases to serve as test oracles. Thanks to the open-source factor, compared to the SQL Server fuzzing tool, SQLSmith has rapidly attracted attention from both academic and industrial. In 2020, Squirrel [ 109] first introduced execution feedback to improve code coverage in crash detection fuzzing. In the same year, Rigger et al. proposed several new fuzzing methods, including NoREC [ 84], TLP [ 85] and PQS [ 86]. NoREC and TLP apply metamorphic testing to the database domain, enabling fuzzing on a single DBMS.",
+      "context_after": "In 2022, Liang et al. introduced SQLRight [ 60], a metamorphic testing method combined with execution feedback, and is scalable to allow new oracles. In 2023, Tang et al. presented a novel constraint-solving testing method, TQS [ 94], which incorporated subgraph isomorphism search into DBMS fuzzing, significantly enhancing bug detection efficiency. DBMS fuzzing techniques can be classified from different perspectives. •One classification approach is based on the way expected results are obtained, as shown in Figure 1: differential testing, metamorphic testing, and constraint-solving. Differen",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 5965,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[ 34,35,38,41,86]",
+      "technique": "pqs",
+      "sentence": "Performance bugs manifest themselves as significant differences in the execution time of the same query in different versions of DBMS, or large performance gaps between a query and its equivalent [ 34,35,38,41,86].",
+      "context_before": "(𝑇𝐶) The execution result of the test case 𝑇𝐶in the DBMS 𝐷 𝐺(𝑇𝐶,𝐷) The ground truth of the execution result for the test case 𝑇𝐶in the DBMS 𝐷 2.2 Basic Definitions Database bugs encompass three main types: crashes, logic bugs, and performance bugs [ 63,72]. Crashes occur when program errors or memory leaks interrupt the database while queries are executed. Compared to other types of bugs, crashes are relatively easier to identify. Logic bugs occur when the execution results deviate from the expected ones or violate common sense in the DBMS. Such bugs are usually hidden and difficult to detect.",
+      "context_after": "However, it is important to note that not all of these differences are considered to be actual bugs. Because determining the optimal execution plan is inherently an NP-hard problem [ 46], some databases practically choose suboptimal plans to speed up the optimization process. We consider performance bugs to have less impact on normal database usage, and therefore have a lower priority than the other two types of bugs. DBMS fuzzing detects crashes, logic bugs, and performance bugs by generating a large number of SQL test statements in the DBMS [ 5,16,73,82,103]. According to the rules of state",
+      "section": "2.2 Basic Definitions",
+      "page": 4,
+      "char_offset": 12617,
+      "cited_reference": {
+        "number": 86,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 87]",
+      "technique": null,
+      "sentence": "For example, intramorphic testing [ 87] works by replacing a specific component in the system to get another version and then testing between the two versions to detect bugs in that component.",
+      "context_before": "DBMS, as shown in Figure 2. Black-box fuzzing treats the DBMS as a black box, where a series of SQL statements or transactions are inputted, and a series of execution results or execution plans are obtained accordingly. The bugs are then determined by validating the results and plans. Grey-box fuzzing collects state information from the DBMS, such as code branch coverage and status, to guide statement generation and improve the efficiency of DBMS testing. By comparison, white-box fuzzing requires testing with an understanding of the internal structure and implementation mechanisms of the DBMS.",
+      "context_after": "However, white-box fuzzing is currently limited to theoretical research on general systems, and there is no white-box fuzzing tool for DBMS, which is also the focus of future research. A testing oracle refers to a mechanism to obtain the expected results of test cases. DBMS fuzzing can also be divided into differential testing, metamorphic testing, and constraint-solving testing according to different oracles. Figure 3 shows the difference between them. Their formal definitions are shown in Definition 2.1, 2.2 and 2.3. Definition 2.1 (Differential Testing). Differential testing detects bugs b",
+      "section": "2.2 Basic Definitions",
+      "page": 5,
+      "char_offset": 14913,
+      "cited_reference": {
+        "number": 87,
+        "text": "Manuel Rigger and Zhendong Su. 2022. Intramorphic Testing: A New Approach to the Test Oracle Problem. In International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software. ACM, 128–136.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[ 16,31–\n33,62,71,81,85,86,90,94,101,109]",
+      "technique": "tlp",
+      "sentence": "On the other hand, fuzzers based on random databases [ 16,31– 33,62,71,81,85,86,90,94,101,109] create random database instances from scratch.",
+      "context_before": "fore generating test cases, a database instance must be available. Depending on their sources, they fall into two categories: existing databases and random databases. Existing databases come from benchmark tests such as TPC-H [ 1] or real datasets such as SCOTT [ 14]. Fuzzers based on existing databases [ 5,19,51,89] first acquire the schema of the target database and then generate syntactically correct statements based on that schema. However, as both the schema and the data are fixed, the generated statements cannot effectively explore the entire query space, leading to incomplete detection.",
+      "context_after": "The random generation of databases has been widely explored [ 23,40,42,76]. A common practice is to first create tables, indexes and views randomly and then populate them with data using INSERT, UPDATE, and DELETE statements [ 84]. The advantage of using randomly generated databases is their potential to thoroughly explore the query space. When coupled with execution feedback, these solutions can effectively explore the search space. Manuscript submitted to ACM 8 AUTHOR et al. Table 2. Comparison of Test Case Generators Fuzzer YearGenerator TypeGenerator StrategyFeedbackDatabase Instance RAGS",
+      "section": "3.1 Test Case Generator",
+      "page": 7,
+      "char_offset": 19993,
+      "cited_reference": {
+        "number": 85,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Programming Languages 4, OOPSLA (2020), 1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[ 84]",
+      "technique": "norec",
+      "sentence": "A common practice is to first create tables, indexes and views randomly and then populate them with data using INSERT, UPDATE, and DELETE statements [ 84].",
+      "context_before": "s TPC-H [ 1] or real datasets such as SCOTT [ 14]. Fuzzers based on existing databases [ 5,19,51,89] first acquire the schema of the target database and then generate syntactically correct statements based on that schema. However, as both the schema and the data are fixed, the generated statements cannot effectively explore the entire query space, leading to incomplete detection. On the other hand, fuzzers based on random databases [ 16,31– 33,62,71,81,85,86,90,94,101,109] create random database instances from scratch. The random generation of databases has been widely explored [ 23,40,42,76].",
+      "context_after": "The advantage of using randomly generated databases is their potential to thoroughly explore the query space. When coupled with execution feedback, these solutions can effectively explore the search space. Manuscript submitted to ACM 8 AUTHOR et al. Table 2. Comparison of Test Case Generators Fuzzer YearGenerator TypeGenerator StrategyFeedbackDatabase Instance RAGS[89] 1998 Generation-based AST Model (Static Configuration) No Feedback (Black-Box)Existing Databases SQLsmith[5] 2015 APOLLO[51] 2019 AST Model (Dynamic Configuration) AMOEBA[62] 2022 Go-Randgen[81] 2019 AST Model (Static Configurat",
+      "section": "3.1 Test Case Generator",
+      "page": 7,
+      "char_offset": 20211,
+      "cited_reference": {
+        "number": 84,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM. Manuscript submitted to ACM 34 AUTHOR et ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Comparison of Test Case Generators Fuzzer YearGenerator TypeGenerator StrategyFeedbackDatabase Instance RAGS[89] 1998 Generation-based AST Model (Static Configuration) No Feedback (Black-Box)Existing Databases SQLsmith[5] 2015 APOLLO[51] 2019 AST Model (Dynamic Configuration) AMOEBA[62] 2022 Go-Randgen[81] 2019 AST Model (Static Configuration) Random DatabasesSQLancer[86] 2020 Artemis[71] 2021 DT2[31] 2022 DQE[90] 2023 Troc[32] 2023 TQS[94] 2023AST Model (Dynamic Configuration) ADUSA[16] 2010 Alloy Model GARan[19] 2007 Mutation-based SQL Structure MutationFeedback from internal status (Grey-Box)Existing Databases Squirrel[109] 2020 Random DatabasesSquill[101] 2023 SQLRight[60] 2022 DynSQL[49] 2023 Eqsql[107] 2021 No Feedback (Black-Box) MutaSQL[28] 2020 LEGO[59] 2023SQL Sequence MutationFeedback from internal status (Grey-Box) Griffin[33] 2022No Feedback (Black-Box) QPG[18] 2023DBMS State MutationFeedback from external interface (Black-Box) 3.",
+      "context_before": "on random databases [ 16,31– 33,62,71,81,85,86,90,94,101,109] create random database instances from scratch. The random generation of databases has been widely explored [ 23,40,42,76]. A common practice is to first create tables, indexes and views randomly and then populate them with data using INSERT, UPDATE, and DELETE statements [ 84]. The advantage of using randomly generated databases is their potential to thoroughly explore the query space. When coupled with execution feedback, these solutions can effectively explore the search space. Manuscript submitted to ACM 8 AUTHOR et al. Table 2.",
+      "context_after": "1.2 Generator type and strategy. After acquiring the database instance, we can proceed to generate test cases using either a generation-based or a mutation-based approach. Among them, the generation-based approach consists primarily of two strategies: Abstract Syntax Tree (AST) model and Alloy model [10]. The general process of the AST model is depicted in Figure 5. First, we need an AST model. AST itself is a tree-like data structure, which can represent the various components of SQL query statements (such as SELECT, FROM, WHERE, JOIN, etc.) and their relationships as nodes and branches for s",
+      "section": "3.1 Test Case Generator",
+      "page": 8,
+      "char_offset": 20626,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker",
+      "surface": "[31, 32, 84–86, 90]",
+      "technique": "norec",
+      "sentence": "Without this limitation, numerous complex SQL statements would be generated, which may expand the search space but reduce the overall efficiency of bug detection [31, 32, 84–86, 90].",
+      "context_before": "ned Tables: This parameter helps constrain the complexity of join relationships, preventing the generation of large-scale multi-table join statements and enhancing the efficiency of statement generation [81, 89]. •Limiting the Maximum Number of Rows Returned: This parameter restricts the execution time and resource consumption of the test cases by adding a LIMIT clause to the original statement [81, 89]. •Limiting the Maximum Depth of the AST: This parameter is set to limit the generation time of test cases, as the generation time tends to increase exponentially with the increase in AST depth.",
+      "context_after": "However, static configurations require manual adjustments based on specific requirements. This process can be time-consuming, and manually setting parameters may not sufficiently explore the state space in a targeted manner. Dynamic configurations of the AST model adjust generator parameters based on previous execution results, which include: Manuscript submitted to ACM 10 AUTHOR et al. Listing 1. Alloy Syntax Model abstract sig FieldName {} abstract sig TableName {} abstract sig Value {} abstract sig Table { name: one TableName, fields: some FieldName } sig Term { field: one FieldName, agg:",
+      "section": "3.1 Test Case Generator",
+      "page": 9,
+      "char_offset": 24291,
+      "cited_reference": {
+        "number": 84,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM. Manuscript submitted to ACM 34 AUTHOR et ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker",
+      "surface": "[ 18]",
+      "technique": "qpg",
+      "sentence": "Mutation-based generators can be divided into three major categories: SQL structure mutation [ 19,28,49,60,101, 107,109], SQL sequence mutation [ 33,59], and DBMS state mutation [ 18].",
+      "context_before": "Figure 7. An example of AST and its corresponding IR, on the left, the AST contains a series of tree nodes, and on the right, V1-V16 represent a series of IR nodes corresponding to each tree node. V15 does not correspond to any tree node in the AST, its introduction serves the purpose of conveniently representing the SelectStmt. while the former guarantees the semantic accuracy of generated statements by constraining column names in select clause and where clause. Therefore, a generator based on the Alloy model [ 16] can achieve a higher level of semantic correctness compared to the AST model.",
+      "context_after": "Figure 6 illustrates the differences between three mutation strategies. SQL structure mutation generates new test cases by modifying the structure of statement, and it is the most common type of mutation. SQL sequence mutation generates new test cases by reshuffling or deleting queries from the original test cases. Its advantage is that it does not rely on SQL syntax, which reduces the adaptation cost for new DBMSs. DBMS state mutation uses DDL operations to create indexes or modify table definitions to generate new database schemas. In addition, it combines old SQL statements with the newly",
+      "section": "3.1 Test Case Generator",
+      "page": 12,
+      "char_offset": 30191,
+      "cited_reference": {
+        "number": 18,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In International Conference on Software Engineering. ACM, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Comparison of Oracles Fuzzer Oracle Type Feature Test Scope Squirrel[109] Crash Database Crashes Crash BugsSquill[101] Griffin[33] LEGO[59] DynSQL[49] RAGS[89] DifferentialDifferent Databases Logic BugsSQLsmith[5] Go-Randgen[81] GARan[19] DT2[31] APOLLO[51] Different Versions of A Same DatabasePerformance BugsAMOEBA[62] MetamorphicStatement RewritingMutaSQL[28] Logic BugsEqsql[107] NoREC[84] SQLRight[60] DQE[90] Statement Type Transformation TLP[85] Query Partition Troc[32] Transaction Splitting ADUSA[16] Constraint-solvingForward Solving(SAT Solver)Artemis[71] PQS[86] Backward Solving TQS[94] Forward Solving(Logical Solver) Some generators of black-box fuzzers [ 18] obtain feedback such as the validity of query plans or test cases from the query or query plan interface provided by the DBMS, guiding the subsequent generation process.",
+      "context_before": "a ‘FromClause’. •Replacement: Replacement can be achieved by modifying the type of an IR node, such as replacing column ‘x’ with ‘count(x)’. •Deletion: Deletion means removing at least one child node from an IR node. 3.1.3 Execution feedback. Most generators do not take advantage of feedback information and treat the target DBMS as a black box, which accepts SQL statements or transactions as input and output execution results or execution plans. Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 13 Table 3.",
+      "context_after": "This kind of feedback is essentially different from the feedback from gray-box fuzzers. The gray-box fuzzer generators [ 19,49,59,60,101,109] collect the internal status of the DBMS, such as collecting code coverage to guide the generation of test cases that maximize code coverage. In contrast, the feedback of the black-box fuzzers relies only upon the external interface of the DBMS without any knowledge of the source code. Black-box fuzzers are easier to implement and can be used to test any commercial DBMSs. However, it is essentially a random exploration of the entire input space. Gray-box",
+      "section": "3.1 Test Case Generator",
+      "page": 13,
+      "char_offset": 32693,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "pqs",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[ 28,60,62,84,85,\n90,107]",
+      "technique": "norec",
+      "sentence": "To avoid the drawbacks of the differential oracle, many fuzzing methods [ 28,60,62,84,85, 90,107] use the metamorphic oracle to detect logic bugs or performance bugs.",
+      "context_before": "ased on crash oracle [ 33,49,59,101,109] can only detect crashes. After executing test cases, bugs are determined based on whether the DBMS stops running. 3.2.2 Differential oracle. Differential oracles [ 5,19,31,81,89] detect logic bugs or performance bugs by comparing the execution results of different DBMSs (or different versions of the same DBMS). It tries to get expected results from another DBMS. The advantage of differential oracle testing is that it is relatively simple. Nevertheless, it cannot detect common bugs between the target DBMS and the referenced one. 3.2.3 Metamorphic oracle.",
+      "context_after": "Metamorphic oracle applies equivalent transformations on the original statements to construct equivalent statements, ensuring that the execution results remain consistent. There are several ways to construct metamorphic oracles: •Statement Rewriting: Statement Rewriting refers to the process of modifying a statement without changing its semantics or results. Common methods of rewriting statements include constraint rewriting, structure rewriting, and expression rewriting, as shown in Figure 8. Constraint rewriting refers to modifying the constraints of a table, such as creating an index. Obvi",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 14,
+      "char_offset": 35969,
+      "cited_reference": {
+        "number": 84,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM. Manuscript submitted to ACM 34 AUTHOR et ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "citation_marker",
+      "surface": "[ 28,62,84,107]",
+      "technique": "norec",
+      "sentence": "Expression Rewriting refers to the process of rewriting logical or arithmetic expressions, such as adding predicates that always hold true to the WHERE clause or changing the string comparison operator from ‘=’ to ‘LIKE’, which will not affect the result of the expression [ 28,62,84,107].",
+      "context_before": "oracles: •Statement Rewriting: Statement Rewriting refers to the process of modifying a statement without changing its semantics or results. Common methods of rewriting statements include constraint rewriting, structure rewriting, and expression rewriting, as shown in Figure 8. Constraint rewriting refers to modifying the constraints of a table, such as creating an index. Obviously, the execution result of the same SELECT statement should remain unchanged after creating the index. Structure rewriting refers to adding new clauses such as JOIN, DISTINCT and LIMIT based on the original statement.",
+      "context_after": "•Statement Type Transformation: Statement type transformation involves constructing UPDATE and DELETE statements that correspond to a given SELECT statement, where the three types of statements should affect the same tuples [ 90]. By adding tagging columns ’rowid’ and ’updated’ to the table, the SELECT statements can return ’rowid’ of the affected rows, as shown in Figure 9. For DELETE statements, the affected rows can be determined by calculating the difference between the set of ’rowid’ before and after deletion. Similarly, for Manuscript submitted to ACM A Comprehensive Survey on Database M",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 14,
+      "char_offset": 36956,
+      "cited_reference": {
+        "number": 84,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM. Manuscript submitted to ACM 34 AUTHOR et ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "citation_marker",
+      "surface": "[ 85]",
+      "technique": "tlp",
+      "sentence": "•Query Partition: Query partition refers to dividing an original SQL query into multiple partitions, and the merged results of the partitioned queries should be consistent with the results of the original query [ 85].",
+      "context_before": "ns ’rowid’ and ’updated’ to the table, the SELECT statements can return ’rowid’ of the affected rows, as shown in Figure 9. For DELETE statements, the affected rows can be determined by calculating the difference between the set of ’rowid’ before and after deletion. Similarly, for Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 15 Figure 9. Example of transformation of statement type UPDATE statements, the affected rows can be obtained by querying the ’rowid’ where the ’updated’ column is marked as true.",
+      "context_after": "For example, a query ‘ SELECT * FROM t1 ’ can be partitioned into three queries: ‘ SELECT * FROM t1 WHERE c1<10 ’, ‘SELECT * FROM t1 WHERE c1>=10 ’, and ‘ SELECT * FROM t1 WHERE c1 IS NULL ’; •Transaction Splitting: Transaction splitting refers to splitting transactions into statements for execution in conjunction with external version concurrency control. In other words, by building a multiple version chain of data outside the database, statements within transactions can run in non-transaction mode but read data visible in transaction mode. By comparing the execution results before and after",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 15,
+      "char_offset": 38100,
+      "cited_reference": {
+        "number": 85,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Programming Languages 4, OOPSLA (2020), 1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[ 16,71,86,94]",
+      "technique": "pqs",
+      "sentence": "Constraint-solving fuzzing methods [ 16,71,86,94] generally relies on forward or backward solving to derive the ground truth of the execution results.",
+      "context_before": "concurrency control. In other words, by building a multiple version chain of data outside the database, statements within transactions can run in non-transaction mode but read data visible in transaction mode. By comparing the execution results before and after transaction splitting, it is possible to determine whether there exist any logic bugs in the transaction management module of the DBMS [32]. 3.2.4 Constraint-solving oracle. Due to the fact that not all test queries can obtain test oracles by constructing equivalent queries, some fuzzers use constraint-solving to obtain expected result.",
+      "context_after": "•Forward Solving: Forward solving involves using an external solver such as an SAT solver to evaluate each tuple against predicate constraints in order to obtain the ground truth for the results of the statement execution [ 37,52]. Manuscript submitted to ACM 16 AUTHOR et al. Table 4. Comparison of Fuzzer Execution Feedbacks Fuzzer Metamorphic Validation Coverage Query Plan Syntax & Semantics Error APOLLO[51] ✗ ✓ ✗ ✗ ✗ Squirrel[109] ✗ ✗ ✓ ✗ ✗ LEGO[59] ✗ ✗ ✓ ✗ ✗ SQLRight[60] ✗ ✗ ✓ ✗ ✗ QPG[18] ✗ ✗ ✗ ✓ ✗ AMOEBA[62] ✓ ✓ ✗ ✗ ✗ GARan[19] ✗ ✗ ✓ ✗ ✗ DynSQL[49] ✗ ✗ ✓ ✗ ✓ Squill[101] ✗ ✓ ✓ ✗ ✓ In this",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 15,
+      "char_offset": 39257,
+      "cited_reference": {
+        "number": 86,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Comparison of Fuzzer Execution Feedbacks Fuzzer Metamorphic Validation Coverage Query Plan Syntax & Semantics Error APOLLO[51] ✗ ✓ ✗ ✗ ✗ Squirrel[109] ✗ ✗ ✓ ✗ ✗ LEGO[59] ✗ ✗ ✓ ✗ ✗ SQLRight[60] ✗ ✗ ✓ ✗ ✗ QPG[18] ✗ ✗ ✗ ✓ ✗ AMOEBA[62] ✓ ✓ ✗ ✗ ✗ GARan[19] ✗ ✗ ✓ ✗ ✗ DynSQL[49] ✗ ✗ ✓ ✗ ✓ Squill[101] ✗ ✓ ✓ ✗ ✓ In this process, a logical solver can also be used as a substitute for the SAT solver, translating the join predicates into logical operations to accelerate the solving speed of the join predicates [94].",
+      "context_before": "ue to the fact that not all test queries can obtain test oracles by constructing equivalent queries, some fuzzers use constraint-solving to obtain expected result. Constraint-solving fuzzing methods [ 16,71,86,94] generally relies on forward or backward solving to derive the ground truth of the execution results. •Forward Solving: Forward solving involves using an external solver such as an SAT solver to evaluate each tuple against predicate constraints in order to obtain the ground truth for the results of the statement execution [ 37,52]. Manuscript submitted to ACM 16 AUTHOR et al. Table 4.",
+      "context_after": "•Backward Solving: Backward solving entails initially selecting some tuples as ground truth at random and then using an SAT solver to work backward and obtain a statement whose execution results include these tuples [ 86]. In all, each oracle has its own applicable scenarios. Fuzzing based on crash oracles can only detect crash-related issues such as program errors and memory leaks, which limits their application scenarios. Fuzzing based on differential oracles can only detect common functional modules between different databases and cannot identify logic or performance bugs shared by the ref",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 16,
+      "char_offset": 39694,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[ 86]",
+      "technique": "pqs",
+      "sentence": "•Backward Solving: Backward solving entails initially selecting some tuples as ground truth at random and then using an SAT solver to work backward and obtain a statement whose execution results include these tuples [ 86].",
+      "context_before": "of the statement execution [ 37,52]. Manuscript submitted to ACM 16 AUTHOR et al. Table 4. Comparison of Fuzzer Execution Feedbacks Fuzzer Metamorphic Validation Coverage Query Plan Syntax & Semantics Error APOLLO[51] ✗ ✓ ✗ ✗ ✗ Squirrel[109] ✗ ✗ ✓ ✗ ✗ LEGO[59] ✗ ✗ ✓ ✗ ✗ SQLRight[60] ✗ ✗ ✓ ✗ ✗ QPG[18] ✗ ✗ ✗ ✓ ✗ AMOEBA[62] ✓ ✓ ✗ ✗ ✗ GARan[19] ✗ ✗ ✓ ✗ ✗ DynSQL[49] ✗ ✗ ✓ ✗ ✓ Squill[101] ✗ ✓ ✓ ✗ ✓ In this process, a logical solver can also be used as a substitute for the SAT solver, translating the join predicates into logical operations to accelerate the solving speed of the join predicates [94].",
+      "context_after": "In all, each oracle has its own applicable scenarios. Fuzzing based on crash oracles can only detect crash-related issues such as program errors and memory leaks, which limits their application scenarios. Fuzzing based on differential oracles can only detect common functional modules between different databases and cannot identify logic or performance bugs shared by the referenced database and the test one. Fuzzing based on metamorphic oracles does not require a reference database, but suffers from a limited test range. Although constrained solvers are not subject to testing scope limitations",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 16,
+      "char_offset": 40203,
+      "cited_reference": {
+        "number": 86,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "citation_marker",
+      "surface": "[ 18]",
+      "technique": "qpg",
+      "sentence": "Query plan feedback [ 18] uses the emergence of new query plans to guide subsequent query generation.",
+      "context_before": "ion before and after executing each test case. If a mutation query leads to an increase in code coverage, it will be added to the seed queue. Additionally, LEGO [ 59] analyzes the type sequences of statements that increase the coverage of the code, and then the mutation processes will generate test cases containing the target type sequences. 3.3.4 Query plan feedback. The goal of query plan feedback is for the generator to maximize the number of query plans. A query plan represents the execution path of an SQL query in a DBMS, with different query plans corresponding to distinct code branches.",
+      "context_after": "Specifically, it models the objective of maximizing query plan quantity as a Multi-Armed Bandit (MAB) problem [ 21]. The generation of new query plans corresponds to exploitation, while exploring new mutation operations corresponds to exploration. When considering both exploitation and exploration, more unique query plans can be generated, increasing the possibility of discovering logic bugs. 3.3.5 Syntax error and semantics error feedback. Some generators [ 101] consider statements with semantic and syntactic errors to be valuable because they may activate certain unexplored SQL features. Th",
+      "section": "3.3 Execution Feedback",
+      "page": 17,
+      "char_offset": 44411,
+      "cited_reference": {
+        "number": 18,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In International Conference on Software Engineering. ACM, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "citation_marker",
+      "surface": "[ 19,28,86,89]",
+      "technique": "pqs",
+      "sentence": "Reducing expressions [ 19,28,86,89] and deleting clauses refer to the process of reducing queries by simplifying WHERE clauses, arithmetic expressions, or logical expressions.",
+      "context_before": "n of test cases that are more suitable for oracles. Syntax and semantic error feedback play a minor role because test cases with syntax errors do not help detect the underlying bugs. 3.4 Query Reducer Query reducers are designed to simplify and summarize complex queries, providing developers with minimal humanreadable queries for debugging. Query reduction strategies can be classified into four types: reduce expression, delete clause, delete subquery, and delete IR node. Table 5 shows the query reduction strategies used by each fuzzer and also indicates whether the reducer preserves semantics.",
+      "context_after": "These methods are relatively easy to implement, but Manuscript submitted to ACM 18 AUTHOR et al. Table 5. Comparison of Fuzzer Query Reducers Fuzzer Reduce Expression Delete Clause Delete Subquery Delete IR Node Semantics Preservation RAGS[89] ✓ ✓ ✗ ✗ ✗ SQLancer[86] ✓ ✓ ✗ ✗ ✗ MutaSQL[28] ✓ ✓ ✗ ✗ ✗ GARan[19] ✓ ✓ ✗ ✗ ✗ SQLRight[60] ✗ ✗ ✗ ✓ ✓ APOLLO[51] ✓ ✓ ✓ ✗ ✓ DynSQL[49] ✓ ✓ ✓ ✗ ✓ cannot guarantee semantic correctness after simplification, resulting in a decrease in reduction efficiency. The removal of subquery operations [ 49,51] extends the scope of simplification to subqueries, leading to",
+      "section": "3.4 Query Reducer",
+      "page": 17,
+      "char_offset": 46906,
+      "cited_reference": {
+        "number": 86,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Comparison of Fuzzer Query Reducers Fuzzer Reduce Expression Delete Clause Delete Subquery Delete IR Node Semantics Preservation RAGS[89] ✓ ✓ ✗ ✗ ✗ SQLancer[86] ✓ ✓ ✗ ✗ ✗ MutaSQL[28] ✓ ✓ ✗ ✗ ✗ GARan[19] ✓ ✓ ✗ ✗ ✗ SQLRight[60] ✗ ✗ ✗ ✓ ✓ APOLLO[51] ✓ ✓ ✓ ✗ ✓ DynSQL[49] ✓ ✓ ✓ ✗ ✓ cannot guarantee semantic correctness after simplification, resulting in a decrease in reduction efficiency.",
+      "context_before": "developers with minimal humanreadable queries for debugging. Query reduction strategies can be classified into four types: reduce expression, delete clause, delete subquery, and delete IR node. Table 5 shows the query reduction strategies used by each fuzzer and also indicates whether the reducer preserves semantics. Reducing expressions [ 19,28,86,89] and deleting clauses refer to the process of reducing queries by simplifying WHERE clauses, arithmetic expressions, or logical expressions. These methods are relatively easy to implement, but Manuscript submitted to ACM 18 AUTHOR et al. Table 5.",
+      "context_after": "The removal of subquery operations [ 49,51] extends the scope of simplification to subqueries, leading to a substantial improvement in the efficiency of simplification. Delete IR node operations are used in mutation-based fuzzers based on IR [ 60] mutation. They treat query simplification as a mutation operation on SQL queries, ensuring semantic correctness by deleting specific IR nodes. When considering the dependency between deleted expressions and remaining expressions, some fuzzers [49, 51, 60] have achieved semantic preservation. 4 DBMSCOMPONENT-BASED TAXONOMY In this section, we review",
+      "section": "3.4 Query Reducer",
+      "page": 18,
+      "char_offset": 47188,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [ 86] begins by randomly generating a pivot row as the ground truth and then constructs an SQL query that includes this pivot row in its result.",
+      "context_before": "r performance bugs, conducting component fuzzing alone on a single component is often insufficient for detection, as performance disparities typically result from the cumulative impact of multiple components. For crashes, fuzzers basically run against the entire DBMS. While it is possible to perform performance bug and crash detection targeting specific components, such testing is not only time-consuming but also provides only localized results, failing to capture the overall state of the entire DBMS. Therefore, there is little research effort in this area. 4.1.1 Overall fuzzing on logic bugs.",
+      "context_after": "By checking whether the result of the executed query in the test database contains the pivot row, logic bugs can be detected. Figure 11 illustrates the entire PQS pipeline. The main challenge of PQS is how to generate an SQL query whose execution result contains a known pivot row. The solution to PQS is to first create predicates randomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False i",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 49795,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Figure 11 illustrates the entire PQS pipeline.",
+      "context_before": "it is possible to perform performance bug and crash detection targeting specific components, such testing is not only time-consuming but also provides only localized results, failing to capture the overall state of the entire DBMS. Therefore, there is little research effort in this area. 4.1.1 Overall fuzzing on logic bugs. PQS [ 86] begins by randomly generating a pivot row as the ground truth and then constructs an SQL query that includes this pivot row in its result. By checking whether the result of the executed query in the test database contains the pivot row, logic bugs can be detected.",
+      "context_after": "The main challenge of PQS is how to generate an SQL query whose execution result contains a known pivot row. The solution to PQS is to first create predicates randomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOT before the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding I",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 50070,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "The main challenge of PQS is how to generate an SQL query whose execution result contains a known pivot row.",
+      "context_before": "rash detection targeting specific components, such testing is not only time-consuming but also provides only localized results, failing to capture the overall state of the entire DBMS. Therefore, there is little research effort in this area. 4.1.1 Overall fuzzing on logic bugs. PQS [ 86] begins by randomly generating a pivot row as the ground truth and then constructs an SQL query that includes this pivot row in its result. By checking whether the result of the executed query in the test database contains the pivot row, logic bugs can be detected. Figure 11 illustrates the entire PQS pipeline.",
+      "context_after": "The solution to PQS is to first create predicates randomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOT before the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding IS NULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 50117,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M24",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "The solution to PQS is to first create predicates randomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions.",
+      "context_before": "localized results, failing to capture the overall state of the entire DBMS. Therefore, there is little research effort in this area. 4.1.1 Overall fuzzing on logic bugs. PQS [ 86] begins by randomly generating a pivot row as the ground truth and then constructs an SQL query that includes this pivot row in its result. By checking whether the result of the executed query in the test database contains the pivot row, logic bugs can be detected. Figure 11 illustrates the entire PQS pipeline. The main challenge of PQS is how to generate an SQL query whose execution result contains a known pivot row.",
+      "context_after": "The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOT before the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding IS NULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried. Figure 11. Pipeline of PQS. The core idea of TLP [ 85] is that the result of the predicate evaluation alway",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 50226,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried.",
+      "context_before": "generate an SQL query whose execution result contains a known pivot row. The solution to PQS is to first create predicates randomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOT before the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding IS NULL after the predicate can help.",
+      "context_after": "Figure 11. Pipeline of PQS. The core idea of TLP [ 85] is that the result of the predicate evaluation always falls within the values of True, False and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal to the union of the results of the three partitioned queries. Figure 12 illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to de",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 50753,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Pipeline of PQS.",
+      "context_before": "hen use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOT before the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding IS NULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried. Figure 11.",
+      "context_after": "The core idea of TLP [ 85] is that the result of the predicate evaluation always falls within the values of True, False and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal to the union of the results of the three partitioned queries. Figure 12 illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 50891,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The core idea of TLP [ 85] is that the result of the predicate evaluation always falls within the values of True, False and NULL.",
+      "context_before": "nterpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOT before the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding IS NULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried. Figure 11. Pipeline of PQS.",
+      "context_after": "Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal to the union of the results of the three partitioned queries. Figure 12 illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [ 18] mutates the state of the database to g",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 50908,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Figure 12 illustrates the main process of TLP.",
+      "context_before": "he result of the predicate query is unknown, so adding IS NULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried. Figure 11. Pipeline of PQS. The core idea of TLP [ 85] is that the result of the predicate evaluation always falls within the values of True, False and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal to the union of the results of the three partitioned queries.",
+      "context_after": "Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [ 18] mutates the state of the database to generate more unique query plans, as shown in Figure 13. QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 51261,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs.",
+      "context_before": "adding IS NULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried. Figure 11. Pipeline of PQS. The core idea of TLP [ 85] is that the result of the predicate evaluation always falls within the values of True, False and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal to the union of the results of the three partitioned queries. Figure 12 illustrates the main process of TLP.",
+      "context_after": "Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [ 18] mutates the state of the database to generate more unique query plans, as shown in Figure 13. QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and obtain different query Manuscript submitted to ACM 20 AUTHOR et al. Figure 12. Pipeline of TLP. plans for the same statement. Because different mutation operations contribute differently t",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 51308,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Neither PQS nor TLP adopts feedback but performs a random search throughout the state space.",
+      "context_before": "line of PQS. The core idea of TLP [ 85] is that the result of the predicate evaluation always falls within the values of True, False and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal to the union of the results of the three partitioned queries. Figure 12 illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs.",
+      "context_after": "QPG [ 18] mutates the state of the database to generate more unique query plans, as shown in Figure 13. QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and obtain different query Manuscript submitted to ACM 20 AUTHOR et al. Figure 12. Pipeline of TLP. plans for the same statement. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decis",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 51496,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG [ 18] mutates the state of the database to generate more unique query plans, as shown in Figure 13.",
+      "context_before": "falls within the values of True, False and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal to the union of the results of the three partitioned queries. Figure 12 illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space.",
+      "context_after": "QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and obtain different query Manuscript submitted to ACM 20 AUTHOR et al. Figure 12. Pipeline of TLP. plans for the same statement. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as a Multi-Armed Bandit [ 21] problem. As shown in",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 51589,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG implements a generation-based generator based on PQS, TLP, and NoREC.",
+      "context_before": "hree partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal to the union of the results of the three partitioned queries. Figure 12 illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [ 18] mutates the state of the database to generate more unique query plans, as shown in Figure 13.",
+      "context_after": "QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and obtain different query Manuscript submitted to ACM 20 AUTHOR et al. Figure 12. Pipeline of TLP. plans for the same statement. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as a Multi-Armed Bandit [ 21] problem. As shown in Equation 1, 𝜇𝑖(𝑡)represents the benefit of the 𝑖-th mutation operator at t",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 51693,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg",
+        "pqs",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M33",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG also utilizes TLP and NoREC to validate query execution results.",
+      "context_before": "e original query should be equal to the union of the results of the three partitioned queries. Figure 12 illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [ 18] mutates the state of the database to generate more unique query plans, as shown in Figure 13. QPG implements a generation-based generator based on PQS, TLP, and NoREC.",
+      "context_after": "DDL and DML statements are used to alter the state of the database and obtain different query Manuscript submitted to ACM 20 AUTHOR et al. Figure 12. Pipeline of TLP. plans for the same statement. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as a Multi-Armed Bandit [ 21] problem. As shown in Equation 1, 𝜇𝑖(𝑡)represents the benefit of the 𝑖-th mutation operator at time 𝑡, there is a probability of 𝑝for random selection, and a probabi",
+      "section": "4.1 Overall Fuzzing",
+      "page": 19,
+      "char_offset": 51767,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Pipeline of TLP.",
+      "context_before": "valent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [ 18] mutates the state of the database to generate more unique query plans, as shown in Figure 13. QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and obtain different query Manuscript submitted to ACM 20 AUTHOR et al. Figure 12.",
+      "context_after": "plans for the same statement. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as a Multi-Armed Bandit [ 21] problem. As shown in Equation 1, 𝜇𝑖(𝑡)represents the benefit of the 𝑖-th mutation operator at time 𝑡, there is a probability of 𝑝for random selection, and a probability of 1−𝑝to opt for the mutation operator with the highest benefit. Since different query plans represent different execution paths, QPG can enhance the code coverag",
+      "section": "4.1 Overall Fuzzing",
+      "page": 20,
+      "char_offset": 51986,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M35",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Since different query plans represent different execution paths, QPG can enhance the code coverage of DBMS.",
+      "context_before": "nd obtain different query Manuscript submitted to ACM 20 AUTHOR et al. Figure 12. Pipeline of TLP. plans for the same statement. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as a Multi-Armed Bandit [ 21] problem. As shown in Equation 1, 𝜇𝑖(𝑡)represents the benefit of the 𝑖-th mutation operator at time 𝑡, there is a probability of 𝑝for random selection, and a probability of 1−𝑝to opt for the mutation operator with the highest benefit.",
+      "context_after": "𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟(𝑡)=(𝑎𝑟𝑔𝑚𝑎𝑥𝑖=1...𝑘(𝜇𝑖(𝑡)) ( 1−𝑝) 𝑟𝑎𝑛𝑑𝑜𝑚(𝑘) ( 𝑝)(1) Figure 13. Pipeline of QPG. 4.1.2 Overall fuzzing on performance bugs. APOLLO[ 51] uses differential testing to detect performance regression bugs. It employs a generation-based generator with a dynamic probability table to generate many random statements. These statements are executed on different versions of the same database program, and performance bugs are detected using a differential oracle. APOLLO incorporates execution feedback to update the probability table, thereby improving the efficiency of the testing process. Additiona",
+      "section": "4.1 Overall Fuzzing",
+      "page": 20,
+      "char_offset": 52505,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M36",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Pipeline of QPG.",
+      "context_before": "differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as a Multi-Armed Bandit [ 21] problem. As shown in Equation 1, 𝜇𝑖(𝑡)represents the benefit of the 𝑖-th mutation operator at time 𝑡, there is a probability of 𝑝for random selection, and a probability of 1−𝑝to opt for the mutation operator with the highest benefit. Since different query plans represent different execution paths, QPG can enhance the code coverage of DBMS. 𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟(𝑡)=(𝑎𝑟𝑔𝑚𝑎𝑥𝑖=1...𝑘(𝜇𝑖(𝑡)) ( 1−𝑝) 𝑟𝑎𝑛𝑑𝑜𝑚(𝑘) ( 𝑝)(1) Figure 13.",
+      "context_after": "4.1.2 Overall fuzzing on performance bugs. APOLLO[ 51] uses differential testing to detect performance regression bugs. It employs a generation-based generator with a dynamic probability table to generate many random statements. These statements are executed on different versions of the same database program, and performance bugs are detected using a differential oracle. APOLLO incorporates execution feedback to update the probability table, thereby improving the efficiency of the testing process. Additionally, Apollo uses query reduction that preserves semantic correctness to quickly obtain",
+      "section": "4.1 Overall Fuzzing",
+      "page": 20,
+      "char_offset": 52683,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M37",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "The NoREC pipeline [ 84] is illustrated in Figure 17.",
+      "context_before": "s that every statement executed in non-transaction mode can obtain the same visible view as in transaction mode. Transaction bugs can be detected by comparing the execution results in transaction and non-transaction modes. 4.3 Optimizer Testing The optimizer is one of the most complex components of a DBMS. It analyzes different execution plans of a query and selects the best query plan by considering indexes, association conditions, data statistics, and other factors to minimize query time and resource consumption. However, implementation errors in the optimizer can lead to serious logic bugs.",
+      "context_after": "NoREC uses an AST model-based generator to generate queries with WHERE clauses and obtains the unoptimized query by moving the conditions from the WHERE clause to the Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 23 Figure 17. Pipeline of NoREC. SELECT clause. The presence of optimizer logic bugs is determined by comparing the number of rows returned by the original query and the number of ‘True’ returned by the unoptimized query. Figure 18. Pipeline of TQS. The pipeline of TQS [94] is illustrated in",
+      "section": "4.3 Optimizer Testing",
+      "page": 22,
+      "char_offset": 58059,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M38",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC uses an AST model-based generator to generate queries with WHERE clauses and obtains the unoptimized query by moving the conditions from the WHERE clause to the Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 23 Figure 17.",
+      "context_before": "e can obtain the same visible view as in transaction mode. Transaction bugs can be detected by comparing the execution results in transaction and non-transaction modes. 4.3 Optimizer Testing The optimizer is one of the most complex components of a DBMS. It analyzes different execution plans of a query and selects the best query plan by considering indexes, association conditions, data statistics, and other factors to minimize query time and resource consumption. However, implementation errors in the optimizer can lead to serious logic bugs. The NoREC pipeline [ 84] is illustrated in Figure 17.",
+      "context_after": "Pipeline of NoREC. SELECT clause. The presence of optimizer logic bugs is determined by comparing the number of rows returned by the original query and the number of ‘True’ returned by the unoptimized query. Figure 18. Pipeline of TQS. The pipeline of TQS [94] is illustrated in Figure 18. Firstly, TQS splits a wide table into multiple small ones. On the one hand, it constructs a schema graph based on the foreign key relationship. By performing random walks on the schema graph, table sequences can be obtained. Combining these table sequences with randomly selected join methods, a multi-table S",
+      "section": "4.3 Optimizer Testing",
+      "page": 22,
+      "char_offset": 58113,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M39",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Pipeline of NoREC.",
+      "context_before": "est query plan by considering indexes, association conditions, data statistics, and other factors to minimize query time and resource consumption. However, implementation errors in the optimizer can lead to serious logic bugs. The NoREC pipeline [ 84] is illustrated in Figure 17. NoREC uses an AST model-based generator to generate queries with WHERE clauses and obtains the unoptimized query by moving the conditions from the WHERE clause to the Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 23 Figure 17.",
+      "context_after": "SELECT clause. The presence of optimizer logic bugs is determined by comparing the number of rows returned by the original query and the number of ‘True’ returned by the unoptimized query. Figure 18. Pipeline of TQS. The pipeline of TQS [94] is illustrated in Figure 18. Firstly, TQS splits a wide table into multiple small ones. On the one hand, it constructs a schema graph based on the foreign key relationship. By performing random walks on the schema graph, table sequences can be obtained. Combining these table sequences with randomly selected join methods, a multi-table SQL query ‘Q’ can be",
+      "section": "4.3 Optimizer Testing",
+      "page": 23,
+      "char_offset": 58433,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M40",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Source Code of Related Fuzzers Fuzzer Supported DBMS Link SQLancer[86] (PQS[86], NoREC[84], TLP[85], QPG[18])SQLite, MySQL, TiDB MariaDB, CockroachDB, OceanBasehttps://github.",
+      "context_before": "hey support in Table 6. The following comparative experiments will be conducted on these fuzzers. 5.3 Evaluation Metrics We adopt the standard metrics for evaluating fuzzing performance: •Number of bugs: The number of bugs can directly quantify the bug detection capabilities of different fuzzers. Some studies submit the detected bugs to the open-source database community to obtain the number of community 1https://github.com/Reverie4u/OpenDBFuzz. Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 25 Table 6.",
+      "context_after": "com/sqlancer/sqlancer DQE[90]SQLite, MySQL, MariaDB TiDB, CockroachDBhttps://github.com/tcse-iscas/dqetool SQLRight[60] SQLite, PostgreSQL, MySQL https: //github.com/psu-security-universe/sqlright SQLsmith[5] SQLite, MonetDB, PostgreSQL https://github.com/anse1/sqlsmith Go-Randgen[81] MySQL, TiDB https://github.com/pingcap/go-randgen Squirrel[109] SQLite, PostgreSQL, MySQL, MariaDB https://github.com/s3team/Squirrel DT2[31] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc Troc[32] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc APOLLO[51] SQLite, PostgreSQL https://github.com/ss",
+      "section": "5.3 Evaluation Metrics",
+      "page": 25,
+      "char_offset": 63214,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M41",
+      "found_by": "name",
+      "surface": "sqlancer",
+      "technique": null,
+      "sentence": "com/sqlancer/sqlancer DQE[90]SQLite, MySQL, MariaDB TiDB, CockroachDBhttps://github.",
+      "context_before": "rformance: •Number of bugs: The number of bugs can directly quantify the bug detection capabilities of different fuzzers. Some studies submit the detected bugs to the open-source database community to obtain the number of community 1https://github.com/Reverie4u/OpenDBFuzz. Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 25 Table 6. Source Code of Related Fuzzers Fuzzer Supported DBMS Link SQLancer[86] (PQS[86], NoREC[84], TLP[85], QPG[18])SQLite, MySQL, TiDB MariaDB, CockroachDB, OceanBasehttps://github.",
+      "context_after": "com/tcse-iscas/dqetool SQLRight[60] SQLite, PostgreSQL, MySQL https: //github.com/psu-security-universe/sqlright SQLsmith[5] SQLite, MonetDB, PostgreSQL https://github.com/anse1/sqlsmith Go-Randgen[81] MySQL, TiDB https://github.com/pingcap/go-randgen Squirrel[109] SQLite, PostgreSQL, MySQL, MariaDB https://github.com/s3team/Squirrel DT2[31] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc Troc[32] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc APOLLO[51] SQLite, PostgreSQL https://github.com/sslab-gatech/apollo AMOEBA[62] PostgreSQL, CockroachDB https://bit.ly/3I995jL confirme",
+      "section": "5.3 Evaluation Metrics",
+      "page": 25,
+      "char_offset": 63390,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M42",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "For logic bug detection, we conducted comparative experiments on five fuzzers, including PQS, NoREC, TLP, QPG, and DQE.",
+      "context_before": "nd comparison in a fuzzer. •New Edges: New edges refer to the number of code branches explored during fuzzing, indicating the breadth of the testing scope. •Coverage: Coverage indicates the proportion of code branches that have been explored. It is calculated using 𝑛𝑒𝑤𝑒𝑑𝑔𝑒𝑠/𝑡𝑜𝑡𝑎𝑙𝑒𝑑𝑔𝑒𝑠, where𝑡𝑜𝑡𝑎𝑙𝑒𝑑𝑔𝑒𝑠 represents the number of all code branches in the DBMS. 5.4 Logic Bugs Detection Comparison Since the feedback module is relatively pluggable, we first tested the fuzzers without feedback to eliminate its impact on the evaluation of other modules and then tested the feedback module independently.",
+      "context_after": "Among them, TLP and QPG only differ in feedback, and the other modules are consistent. Experiments were conducted on two popular DBMSs: MySQL 8.0.16 and SQLite 3.28.0. The number of bugs, the validity, and the valid cases per second are used as evaluation metrics. As shown in Figure 19, the experimental results indicate that these methods have detected more bugs in SQLite compared to MySQL, possibly due to the greater maturity of MySQL. Moreover, the validity of these methods on SQLite is relatively low, indicating that the SQL syntax of SQLite is not suitable for these fuzzers’ generators. H",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 25,
+      "char_offset": 65440,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M43",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Among them, TLP and QPG only differ in feedback, and the other modules are consistent.",
+      "context_before": "g the breadth of the testing scope. •Coverage: Coverage indicates the proportion of code branches that have been explored. It is calculated using 𝑛𝑒𝑤𝑒𝑑𝑔𝑒𝑠/𝑡𝑜𝑡𝑎𝑙𝑒𝑑𝑔𝑒𝑠, where𝑡𝑜𝑡𝑎𝑙𝑒𝑑𝑔𝑒𝑠 represents the number of all code branches in the DBMS. 5.4 Logic Bugs Detection Comparison Since the feedback module is relatively pluggable, we first tested the fuzzers without feedback to eliminate its impact on the evaluation of other modules and then tested the feedback module independently. For logic bug detection, we conducted comparative experiments on five fuzzers, including PQS, NoREC, TLP, QPG, and DQE.",
+      "context_after": "Experiments were conducted on two popular DBMSs: MySQL 8.0.16 and SQLite 3.28.0. The number of bugs, the validity, and the valid cases per second are used as evaluation metrics. As shown in Figure 19, the experimental results indicate that these methods have detected more bugs in SQLite compared to MySQL, possibly due to the greater maturity of MySQL. Moreover, the validity of these methods on SQLite is relatively low, indicating that the SQL syntax of SQLite is not suitable for these fuzzers’ generators. However, the valid cases per second on SQLite are actually higher, which could be attrib",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 25,
+      "char_offset": 65560,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M44",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Comparision of oracles for detecting logic bugs As can be seen from Figure 19 (a), (b) and (c), in SQLite, PQS exhibits the lowest semantic validity and valid cases per second, but it demonstrated the highest bug detection efficiency, finding 276 bugs in only 240 minutes.",
+      "context_before": "cted more bugs in SQLite compared to MySQL, possibly due to the greater maturity of MySQL. Moreover, the validity of these methods on SQLite is relatively low, indicating that the SQL syntax of SQLite is not suitable for these fuzzers’ generators. However, the valid cases per second on SQLite are actually higher, which could be attributed to its faster query performance on small data volumes. Manuscript submitted to ACM 26 AUTHOR et al. (a)Number of bugs (SQLite) (b)Validity (SQLite) (c)Valid Cases/s (SQLite) (d)Number of bugs (MySQL) (e)Validity (MySQL) (f)Valid Cases/s (MySQL) Figure 19.",
+      "context_after": "The other fuzzers employ metamorphic testing strategies and have similar bug detection capabilities, stabilizing at around 100 bugs within 240 minutes. While NoREC stands out by using a strategy that transforms one statement into another, as opposed to the other methods that require transforming one statement into multiple ones [ 90]. Consequently, NoRec executes more valid cases per second, resulting in slightly faster bug detection efficiency as well. In summary, constraint solving is not limited by metamorphic applicability. Although the number of test cases per second is smaller, the over",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 26,
+      "char_offset": 66511,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M45",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "While NoREC stands out by using a strategy that transforms one statement into another, as opposed to the other methods that require transforming one statement into multiple ones [ 90].",
+      "context_before": "6 AUTHOR et al. (a)Number of bugs (SQLite) (b)Validity (SQLite) (c)Valid Cases/s (SQLite) (d)Number of bugs (MySQL) (e)Validity (MySQL) (f)Valid Cases/s (MySQL) Figure 19. Comparision of oracles for detecting logic bugs As can be seen from Figure 19 (a), (b) and (c), in SQLite, PQS exhibits the lowest semantic validity and valid cases per second, but it demonstrated the highest bug detection efficiency, finding 276 bugs in only 240 minutes. The other fuzzers employ metamorphic testing strategies and have similar bug detection capabilities, stabilizing at around 100 bugs within 240 minutes.",
+      "context_after": "Consequently, NoRec executes more valid cases per second, resulting in slightly faster bug detection efficiency as well. In summary, constraint solving is not limited by metamorphic applicability. Although the number of test cases per second is smaller, the overall bug detection efficiency is higher, and ultimately twice the number of bugs can be found compared to metamorphic testing methods. Figure 19 (d), (e) and (f) show the execution results on MySQL. Since NoREC does not support MySQL, we only compared the other tools. It can be observed that TLP and QPG are quite exceptional, as they qu",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 26,
+      "char_offset": 66936,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M46",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "Consequently, NoRec executes more valid cases per second, resulting in slightly faster bug detection efficiency as well.",
+      "context_before": "on of oracles for detecting logic bugs As can be seen from Figure 19 (a), (b) and (c), in SQLite, PQS exhibits the lowest semantic validity and valid cases per second, but it demonstrated the highest bug detection efficiency, finding 276 bugs in only 240 minutes. The other fuzzers employ metamorphic testing strategies and have similar bug detection capabilities, stabilizing at around 100 bugs within 240 minutes. While NoREC stands out by using a strategy that transforms one statement into another, as opposed to the other methods that require transforming one statement into multiple ones [ 90].",
+      "context_after": "In summary, constraint solving is not limited by metamorphic applicability. Although the number of test cases per second is smaller, the overall bug detection efficiency is higher, and ultimately twice the number of bugs can be found compared to metamorphic testing methods. Figure 19 (d), (e) and (f) show the execution results on MySQL. Since NoREC does not support MySQL, we only compared the other tools. It can be observed that TLP and QPG are quite exceptional, as they quickly detected bugs in MySQL. They discovered 12 bugs in just 40 minutes, after which the system crashed (crash detected)",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 26,
+      "char_offset": 67121,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M47",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Since NoREC does not support MySQL, we only compared the other tools.",
+      "context_before": "hat transforms one statement into another, as opposed to the other methods that require transforming one statement into multiple ones [ 90]. Consequently, NoRec executes more valid cases per second, resulting in slightly faster bug detection efficiency as well. In summary, constraint solving is not limited by metamorphic applicability. Although the number of test cases per second is smaller, the overall bug detection efficiency is higher, and ultimately twice the number of bugs can be found compared to metamorphic testing methods. Figure 19 (d), (e) and (f) show the execution results on MySQL.",
+      "context_after": "It can be observed that TLP and QPG are quite exceptional, as they quickly detected bugs in MySQL. They discovered 12 bugs in just 40 minutes, after which the system crashed (crash detected), leading to both validity and valid cases per second becoming zero. In contrast, DQE, which also falls under metamorphic testing, consistently failed to detect any logic bugs, indicating a more limited scope in its oracle. Further investigation revealed that in DQE, statements and their equivalent statements produced the same bug results, leading to missed detections. It is worth noting that PQS does not",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 26,
+      "char_offset": 67581,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M48",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "It can be observed that TLP and QPG are quite exceptional, as they quickly detected bugs in MySQL.",
+      "context_before": "hods that require transforming one statement into multiple ones [ 90]. Consequently, NoRec executes more valid cases per second, resulting in slightly faster bug detection efficiency as well. In summary, constraint solving is not limited by metamorphic applicability. Although the number of test cases per second is smaller, the overall bug detection efficiency is higher, and ultimately twice the number of bugs can be found compared to metamorphic testing methods. Figure 19 (d), (e) and (f) show the execution results on MySQL. Since NoREC does not support MySQL, we only compared the other tools.",
+      "context_after": "They discovered 12 bugs in just 40 minutes, after which the system crashed (crash detected), leading to both validity and valid cases per second becoming zero. In contrast, DQE, which also falls under metamorphic testing, consistently failed to detect any logic bugs, indicating a more limited scope in its oracle. Further investigation revealed that in DQE, statements and their equivalent statements produced the same bug results, leading to missed detections. It is worth noting that PQS does not perform as well on MySQL as it does on SQLite. There are two main reasons for this. Manuscript subm",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 26,
+      "char_offset": 67651,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M49",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "It is worth noting that PQS does not perform as well on MySQL as it does on SQLite.",
+      "context_before": "SQL, we only compared the other tools. It can be observed that TLP and QPG are quite exceptional, as they quickly detected bugs in MySQL. They discovered 12 bugs in just 40 minutes, after which the system crashed (crash detected), leading to both validity and valid cases per second becoming zero. In contrast, DQE, which also falls under metamorphic testing, consistently failed to detect any logic bugs, indicating a more limited scope in its oracle. Further investigation revealed that in DQE, statements and their equivalent statements produced the same bug results, leading to missed detections.",
+      "context_after": "There are two main reasons for this. Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 27 Firstly, MySQL has fewer inherent bugs, and our experiments did not run for months to detect these bugs. Secondly, the architecture of PQS speeds up the validation of each test case, but it is not a full comparison, only checking one of its query results. This is, in fact, sacrificing accuracy for speed. This approach may be more effective in environments with more bugs, but in the case of MySQL, i.e., with fewer bug",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 26,
+      "char_offset": 68213,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M50",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Secondly, the architecture of PQS speeds up the validation of each test case, but it is not a full comparison, only checking one of its query results.",
+      "context_before": "ntly failed to detect any logic bugs, indicating a more limited scope in its oracle. Further investigation revealed that in DQE, statements and their equivalent statements produced the same bug results, leading to missed detections. It is worth noting that PQS does not perform as well on MySQL as it does on SQLite. There are two main reasons for this. Manuscript submitted to ACM A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison 27 Firstly, MySQL has fewer inherent bugs, and our experiments did not run for months to detect these bugs.",
+      "context_after": "This is, in fact, sacrificing accuracy for speed. This approach may be more effective in environments with more bugs, but in the case of MySQL, i.e., with fewer bugs, the overall efficiency is lower. (a)Number of bugs (SQLite) (b)Validity (SQLite) (c)Valid Cases/s (SQLite) Figure 20. Comparision of feedback ways for detecting logic bugs Feedback currently applied to logic bug detection includes coverage feedback and query plan feedback. We conducted comparative experiments in SQLite using TLP, QPG, and SQLRight. They all use the same oracle, but TLP does not have feedback, QPG uses query pl",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 27,
+      "char_offset": 68581,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M51",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We conducted comparative experiments in SQLite using TLP, QPG, and SQLRight.",
+      "context_before": "bugs. Secondly, the architecture of PQS speeds up the validation of each test case, but it is not a full comparison, only checking one of its query results. This is, in fact, sacrificing accuracy for speed. This approach may be more effective in environments with more bugs, but in the case of MySQL, i.e., with fewer bugs, the overall efficiency is lower. (a)Number of bugs (SQLite) (b)Validity (SQLite) (c)Valid Cases/s (SQLite) Figure 20. Comparision of feedback ways for detecting logic bugs Feedback currently applied to logic bug detection includes coverage feedback and query plan feedback.",
+      "context_after": "They all use the same oracle, but TLP does not have feedback, QPG uses query plan feedback, and SQLRight utilizes coverage feedback. Among them, TLP and QPG employ the same generation-based generator. As can be seen in Figure 20, QPG detected bugs more efficiently than TLP within 240 minutes, and the number of bugs gradually equalized in the later period. This indicates that query plan feedback indeed contributes to improving the efficiency of logic bug detection. Furthermore, SQLRight only detected 5 bugs within 240 minutes, but this cannot be entirely attributed to coverage feedback. One im",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 27,
+      "char_offset": 69175,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M52",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "They all use the same oracle, but TLP does not have feedback, QPG uses query plan feedback, and SQLRight utilizes coverage feedback.",
+      "context_before": "t case, but it is not a full comparison, only checking one of its query results. This is, in fact, sacrificing accuracy for speed. This approach may be more effective in environments with more bugs, but in the case of MySQL, i.e., with fewer bugs, the overall efficiency is lower. (a)Number of bugs (SQLite) (b)Validity (SQLite) (c)Valid Cases/s (SQLite) Figure 20. Comparision of feedback ways for detecting logic bugs Feedback currently applied to logic bug detection includes coverage feedback and query plan feedback. We conducted comparative experiments in SQLite using TLP, QPG, and SQLRight.",
+      "context_after": "Among them, TLP and QPG employ the same generation-based generator. As can be seen in Figure 20, QPG detected bugs more efficiently than TLP within 240 minutes, and the number of bugs gradually equalized in the later period. This indicates that query plan feedback indeed contributes to improving the efficiency of logic bug detection. Furthermore, SQLRight only detected 5 bugs within 240 minutes, but this cannot be entirely attributed to coverage feedback. One important reason is that the mutation-based generator used by SQLRight is not effective, and its validity is obviously lower. The impac",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 27,
+      "char_offset": 69252,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M53",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Among them, TLP and QPG employ the same generation-based generator.",
+      "context_before": "is approach may be more effective in environments with more bugs, but in the case of MySQL, i.e., with fewer bugs, the overall efficiency is lower. (a)Number of bugs (SQLite) (b)Validity (SQLite) (c)Valid Cases/s (SQLite) Figure 20. Comparision of feedback ways for detecting logic bugs Feedback currently applied to logic bug detection includes coverage feedback and query plan feedback. We conducted comparative experiments in SQLite using TLP, QPG, and SQLRight. They all use the same oracle, but TLP does not have feedback, QPG uses query plan feedback, and SQLRight utilizes coverage feedback.",
+      "context_after": "As can be seen in Figure 20, QPG detected bugs more efficiently than TLP within 240 minutes, and the number of bugs gradually equalized in the later period. This indicates that query plan feedback indeed contributes to improving the efficiency of logic bug detection. Furthermore, SQLRight only detected 5 bugs within 240 minutes, but this cannot be entirely attributed to coverage feedback. One important reason is that the mutation-based generator used by SQLRight is not effective, and its validity is obviously lower. The impact of coverage feedback on logic bug detection remains to be studied",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 27,
+      "char_offset": 69385,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M54",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "As can be seen in Figure 20, QPG detected bugs more efficiently than TLP within 240 minutes, and the number of bugs gradually equalized in the later period.",
+      "context_before": "t in the case of MySQL, i.e., with fewer bugs, the overall efficiency is lower. (a)Number of bugs (SQLite) (b)Validity (SQLite) (c)Valid Cases/s (SQLite) Figure 20. Comparision of feedback ways for detecting logic bugs Feedback currently applied to logic bug detection includes coverage feedback and query plan feedback. We conducted comparative experiments in SQLite using TLP, QPG, and SQLRight. They all use the same oracle, but TLP does not have feedback, QPG uses query plan feedback, and SQLRight utilizes coverage feedback. Among them, TLP and QPG employ the same generation-based generator.",
+      "context_after": "This indicates that query plan feedback indeed contributes to improving the efficiency of logic bug detection. Furthermore, SQLRight only detected 5 bugs within 240 minutes, but this cannot be entirely attributed to coverage feedback. One important reason is that the mutation-based generator used by SQLRight is not effective, and its validity is obviously lower. The impact of coverage feedback on logic bug detection remains to be studied further. Also, as expected, the number of valid cases per second decreases slightly when feedback is added. 5.5 Crash Detection Comparison We tested the perf",
+      "section": "5.4 Logic Bugs Detection Comparison",
+      "page": 27,
+      "char_offset": 69453,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M55",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 87]",
+      "technique": null,
+      "sentence": "•White-box Component Fuzzing: Recent research [ 87] has introduced white-box fuzzing in software testing.",
+      "context_before": "fficient. Component-oriented fuzzing can make testing more focused and efficient, and research directions include: •Component-specific Fuzzing: Most components in a DBMS can be regarded as standalone objects that take input and produce output. Therefore, we can perform separate fuzzing for these specific components, such as directly generating PUT ,GET and DELETE operations to test bugs in key-value (KV) storage. Although there are some testing methods for the DBMS optimizer [ 35,41,54,58], they are not fuzzing methods and can only evaluate the quality of the query plans, not their correctness.",
+      "context_after": "That method involves replacing components in a predictable manner to evaluate whether there are bugs in them. In theory, this method can be extended to DBMSs for component fuzzing. The primary challenge lies in achieving cost-effective replacement of DBMS components to improve its practicality. 6.3 Fuzzing of Modern Database Systems With the emergence of new computing technologies and application fields, traditional relational databases face new challenges, driving the development of databases in new directions. Emerging databases include distributed databases[ 17, 30,44,93], multi-model data",
+      "section": "6.2 Component-oriented Fuzzing",
+      "page": 29,
+      "char_offset": 75641,
+      "cited_reference": {
+        "number": 87,
+        "text": "Manuel Rigger and Zhendong Su. 2022. Intramorphic Testing: A New Approach to the Test Oracle Problem. In International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software. ACM, 128–136.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M56",
+      "found_by": "citation_marker",
+      "surface": "[ 18]",
+      "technique": "qpg",
+      "sentence": "Existing work [ 18] uses query plan feedback to maximize exploration of the query space, which is not achievable at the code level alone.",
+      "context_before": "re testing perspective, it is not efficient in discovering DBMS bugs. In practice, a balance should be struck between full branch coverage and branch specific coverage. For example, it might be more efficient to prioritize testing the branches near code boundaries where exceptions or security vulnerabilities are more likely to occur. •Semantic Dimension Space Exploration: Coverage feedback usually focuses on maximizing space exploration at the code level and ignores the semantic level. This means that test cases may trigger code branches but do not necessarily detect semantically related bugs.",
+      "context_after": "Furthermore, maximizing space exploration in other dimensions, such as transaction space, is crucial. 7 CONCLUSION DBMS is the fundamental software for managing and organizing data, and its internal structure is extremely complex. The development of commercial DBMSs often produces a large number of unexpected bugs (more than 10,000 bugs have been discovered and resolved in the MySQL forum [ 15]), which can easily lead to huge economic losses. Fuzzing is a method that detects bugs by automatically generating, mutating, and executing test cases. It has a history of nearly 50 years, but existing",
+      "section": "6.4 Improved Space Exploration Capabilities",
+      "page": 30,
+      "char_offset": 79266,
+      "cited_reference": {
+        "number": 18,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In International Conference on Software Engineering. ACM, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M54"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-10T15:11:27Z",
+    "is_model_written": true,
+    "summary": "A survey of DBMS fuzzing that builds a taxonomy of the field -- generator type, generation strategy, feedback, and how the database instance is obtained -- and runs an experimental comparison of the tools within it. It covers logic, crash and performance bugs, and traces the line from RAGS in 1998 to the coverage-guided and oracle-based tools that followed.",
+    "narrative": "SQLancer's oracles are a fixed point of the survey's taxonomy. NoREC, TLP and PQS are introduced together as the new fuzzing methods Rigger and Su proposed, and distinguished by kind: NoREC and TLP apply metamorphic testing to the database domain so that a single DBMS can be fuzzed against itself, while PQS is a constraint-solving method whose oracle needs only that the result contain the solved row. They recur throughout as reference points -- in the comparison of test case generators, in the discussion of random database instances, and in the account of why bounding statement complexity matters. Intramorphic testing is cited separately as replacing a component to test between two versions of a system.",
+    "roles": {
+      "M1": "definition",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "background",
+      "M5": "definition",
+      "M6": "background",
+      "M7": "background",
+      "M8": "result_comparison",
+      "M9": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "A survey describes tools rather than building on them; nothing here reports reusing SQLancer's code, generator or workload."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is proposed or extended; the oracles are classified within the taxonomy."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M8"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M8",
+            "sentence": "Comparison of Test Case Generators Fuzzer YearGenerator TypeGenerator StrategyFeedbackDatabase Instance RAGS[89] 1998 Generation-based AST Model (Static Configuration) No Feedback (Black-Box)Existing Databases SQLsmith[5] 2015 APOLLO[51] 2019 AST Model (Dynamic Configuration) AMOEBA[62] 2022 Go-Randgen[81] 2019 AST Model (Static Configuration) Random DatabasesSQLancer[86] 2020 Artemis[71] 2021 DT2[31] 2022 DQE[90] 2023 Troc[32] 2023 TQS[94] 2023AST Model (Dynamic Configuration) ADUSA[16] 2010 Alloy Model GARan[19] 2007 Mutation-based SQL Structure MutationFeedback from internal status (Grey-Box)Existing Databases Squirrel[109] 2020 Random DatabasesSquill[101] 2023 SQLRight[60] 2022 DynSQL[49] 2023 Eqsql[107] 2021 No Feedback (Black-Box) MutaSQL[28] 2020 LEGO[59] 2023SQL Sequence MutationFeedback from internal status (Grey-Box) Griffin[33] 2022No Feedback (Black-Box) QPG[18] 2023DBMS State MutationFeedback from external interface (Black-Box) 3.",
+            "section": "3.1 Test Case Generator",
+            "page": 8
+          }
+        ],
+        "reasoning": "M8 is the survey's own comparison of test case generators, tabulating generator type, strategy, feedback and database instance across the tools including SQLancer's. This is a comparison the survey conducts rather than one it reports from elsewhere.",
+        "techniques": [
+          "norec",
+          "tlp",
+          "pqs"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The oracles are described by mechanism and placed in a taxonomy; the survey makes no claim that they are the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2312_04941.json b/_data/papers/paper_arxiv_2312_04941.json
new file mode 100644
index 0000000..55ea0ca
--- /dev/null
+++ b/_data/papers/paper_arxiv_2312_04941.json
@@ -0,0 +1,1247 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:06:52Z",
+  "paper": {
+    "id": "paper:arxiv:2312.04941",
+    "title": "Detecting DBMS Bugs with Context-Sensitive Instantiation and Multi-Plan Execution",
+    "authors": [
+      "Jiaqi Li",
+      "Ke Wang",
+      "Yaoguang Chen",
+      "Yajin Zhou",
+      "Lei Wu",
+      "Jiashui Wang"
+    ],
+    "year": 2023,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2312.04941",
+    "s2_paper_id": "95db0384945786d61210ca73bf62b5e59bbe266e",
+    "url": "https://arxiv.org/abs/2312.04941",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2312.04941",
+      "retrieved_at": "2026-09-10T15:06:52Z",
+      "chars": 79515,
+      "content_sha256": "sha256:248fe572e5e4aecc30d80568219c0f72d60b3043d6f5f13d8e2dcb66c6d1d808"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1552
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 8671
+      },
+      {
+        "number": "2.1",
+        "title": "Structured Query Language",
+        "start": 8685
+      },
+      {
+        "number": "2.2",
+        "title": "SQL Query Processing",
+        "start": 9545
+      },
+      {
+        "number": "2.3",
+        "title": "Levels of DBMS Testing",
+        "start": 11464
+      },
+      {
+        "number": "2.4",
+        "title": "DBMS Testing",
+        "start": 13493
+      },
+      {
+        "number": "3",
+        "title": "Motivating Examples",
+        "start": 15688
+      },
+      {
+        "number": "3.1",
+        "title": "Context-Sensitive Instantiation",
+        "start": 15819
+      },
+      {
+        "number": "3.2",
+        "title": "Multi-Plan Execution",
+        "start": 21475
+      },
+      {
+        "number": "4",
+        "title": "Design",
+        "start": 27665
+      },
+      {
+        "number": "4.1",
+        "title": "Overall Design",
+        "start": 27675
+      },
+      {
+        "number": "4.2",
+        "title": "Parser Generation",
+        "start": 32631
+      },
+      {
+        "number": "4.3",
+        "title": "Parse Tree Mutation",
+        "start": 33861
+      },
+      {
+        "number": "4.4",
+        "title": "Context-Sensitive Instantiation",
+        "start": 34810
+      },
+      {
+        "number": "4.5",
+        "title": "Multi-Plan Execution",
+        "start": 43977
+      },
+      {
+        "number": "4.6",
+        "title": "Prototype Implementation",
+        "start": 48506
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 49668
+      },
+      {
+        "number": "5.1",
+        "title": "Detecting Bugs in Real-world DBMSs",
+        "start": 50908
+      },
+      {
+        "number": "5.2",
+        "title": "Generating Valid Queries",
+        "start": 51640
+      },
+      {
+        "number": "5.3",
+        "title": "Comparisons with Existing Tools",
+        "start": 56798
+      },
+      {
+        "number": "5.4",
+        "title": "Benefits of the Two Key Techniques",
+        "start": 60432
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 63308
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 65702
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 69671
+      },
+      {
+        "number": "Z",
+        "title": "Chen, “Collafl: Path sensitive fuzzing,” in. IEEE,",
+        "start": 74735
+      },
+      {
+        "number": "S",
+        "title": "Khurshid, “Query-aware test generation using a",
+        "start": 75090
+      },
+      {
+        "number": "T",
+        "title": "Wei, and L. Lu, “Savior: Towards bug-driven hybrid",
+        "start": 76145
+      },
+      {
+        "number": "J",
+        "title": "Cappos, M. Schl ¨ogel, N. Korshun, A. Abbasi,",
+        "start": 76487
+      },
+      {
+        "number": "M",
+        "title": "Schweighauser, S. Schinzel, S. Schumilo et al.,",
+        "start": 76536
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Well-known users of sqlite,” https://www.sqlite.org/f amous.html, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“Mysql customers,” https://www.mysql.com/customer s/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "A. Bannister, “Sqlite patches use-after-free bug that left apps open to code execution, denial-of-service exploits.” https://portswigger.net/daily-swig/sqlite-pat ches-use-after-free-bug-that-left-apps-open-to-code-e xecution-denial-of-service-exploits, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "C. Cimpanu, “Google chrome impacted by new magellan 2.0 vulnerabilities.” https://www.zdnet.com/article/google-chrome-imp acted-by-new-magellan-2-0-vulnerabilities/, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “Apollo: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment, vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing database management systems with language validity and coverage feedback,” inProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "S. M. Andreas Seltenreich, Bo Tang, “Sqlsmith,” https: //github.com/anse1/sqlsmith/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Y. Chen, R. Zhong, H. Hu, H. Zhang, Y. Yang, D. Wu, and W. Lee, “One engine to fuzz’em all: Generic language processor testing with semantic validation,” in (SP). IEEE, 2021, pp. 642–658.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "X. Liu, Q. Zhou, J. Arulraj, and A. Orso, “Automated performance bug detection in database systems,” arXiv e-prints, pp. arXiv–2105, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Y. Zhang, P. Yao, R. Wu, and C. Zhang, “Duplicatesensitivity guided transformation synthesis for dbms correctness bug detection,” arXiv preprint arXiv:2107.03660, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "B. Ghit, N. Poggi, J. Rosen, R. Xin, and P. Boncz, “Sparkfuzz: Searching correctness regressions in modern query engines,” in Proceedings of the workshop on Testing Database Systems, 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "“Addresssanitizer,” https://github.com/google/sanitize rs/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of{DBMS }with coverage-based guidance,” in 31st USENIX Security Symposium (USENIX Security 22), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "——, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 16,
+      "text": "——, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 17,
+      "text": "X. Tang, S. Wu, D. Zhang, F. Li, and G. Chen, “Detecting logic bugs of join optimizations in dbms,” Proceedings of the ACM on Management of Data, vol. 1, no. 1, pp. 1–26, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "“Sqlite homepage,” https://www.sqlite.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "“Postgresql homepage,” https://www.postgresql.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "“Mysql homepage,” https://www.mysql.com/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "“Common vulnerabilities and exposures,” https://en.wikipedia.org/wiki/Common Vulnerabili ties and Exposures, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "D. D. Chamberlin and R. F. Boyce, “Sequel: A structured english query language,” in Proceedings of the 1974 ACMSIGFIDET (now SIGMOD) workshop on Data description, access and control, 1974, pp. 249–264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "“American Fuzzy Lop (2.56b),” https://lcamtuf.coredu mp.cx/afl/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "M. Sutton, A. Greene, and P. Amini, Fuzzing: brute force vulnerability discovery. Pearson Education, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "J. Neystadt, “Automated penetration testing with whitebox fuzzing,” MSDN Library, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "J. Wang, B. Chen, L. Wei, and Y. Liu, “Superion: Grammar-aware greybox fuzzing,” in 2019 IEEE/ACM 41st International Conference on Software Engineering (ICSE). IEEE, 2019, pp. 724–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "D. R. Slutz, “Massive stochastic testing of sql,” in VLDB, vol. 98. Citeseer, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "“Bugs found in database management systems,” https: //www.manuelrigger.at/dbms-bugs/, 2022.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 29,
+      "text": "“Unit testing,” https://en.wikipedia.org/wiki/Unit testi ng, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "“Randomized depth-first search,” https://en.wikipedia .org/wiki/Maze generation algorithm, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "D. E. Knuth, “Backus normal form vs. backus naur form,” Communications of the ACM, vol. 7, no. 12, pp. 735–736, 1964.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "“Constraint satisfaction problem,” https://en.wikipedia .org/wiki/Constraint satisfaction problem, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "S. Gan, C. Zhang, X. Qin, X. Tu, K. Li, Z. Pei, and Z. Chen, “Collafl: Path sensitive fuzzing,” in. IEEE, 2018, pp. 679–696.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "“How sqlite is tested,” https://www.sqlite.org/testing.h tml, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "J. Postel et al., “Transmission control protocol,” p. 13, 1981.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "“Proof of concept,” https://en.wikipedia.org/wiki/Proo fofconcept#Security, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "S. A. Khalek, B. Elkarablieh, Y. O. Laleye, and S. Khurshid, “Query-aware test generation using a relational constraint solver,” in 2008 23rd IEEE/ACM International Conference on Automated Software Engineering. IEEE, 2008, pp. 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "C. Binnig, D. Kossmann, E. Lo, and M. T. ¨Ozsu, “Qagen: generating query-aware test databases,” in Proceedings of the 2007 ACMSIGMOD international conference on Management of data, 2007, pp. 341– 352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "“Alloy,” https://alloytools.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "N. Stephens, J. Grosen, C. Salls, A. Dutcher, R. Wang, J. Corbetta, Y. Shoshitaishvili, C. Kruegel, and G. Vigna, “Driller: Augmenting fuzzing through selective symbolic execution.” in NDSS, vol. 16, no. 2016, 2016, pp. 1–16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "I. Yun, S. Lee, M. Xu, Y. Jang, and T. Kim, “ {QSYM }: A practical concolic execution engine tailored for hybrid fuzzing,” in 27th USENIX Security Symposium (USENIX Security 18), 2018, pp. 745–761.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "P. Chen and H. Chen, “Angora: Efficient fuzzing by principled search,” in rity and Privacy (SP). IEEE, 2018, pp. 711–725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Y. Chen, P. Li, J. Xu, S. Guo, R. Zhou, Y. Zhang, T. Wei, and L. Lu, “Savior: Towards bug-driven hybrid testing,” in Privacy (SP). IEEE, 2020, pp. 1580–1596.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "S. Gan, C. Zhang, P. Chen, B. Zhao, X. Qin, D. Wu, and Z. Chen, “ {GREYONE }: Data flow sensitive fuzzing,” in 29th USENIX Security Symposium (USENIX Security 20), 2020, pp. 2577–2594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "T. Blazytko, M. Bishop, C. Aschermann, J. Cappos, M. Schl ¨ogel, N. Korshun, A. Abbasi, M. Schweighauser, S. Schinzel, S. Schumilo et al., “{GRIMOIRE }: Synthesizing structure while fuzzing,” in28th USENIX Security Symposium (USENIX Security 19), 2019, pp. 1985–2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "H. Bati, L. Giakoumakis, S. Herbert, and A. Surna, “A genetic approach for random testing of database systems,” in Proceedings of the 33rd international conference on Very large data bases, 2007, pp. 1243– 1251.Appendix A. PoC Generation Strategies MPE can only conduct the PoC on modified DBMSs. To build PoC on unmodified DBMSs, our goal is to make the buggy query plan to be optimal. The ways to a",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 15,
+      "text": "——, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 16,
+      "text": "——, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 28,
+      "text": "“Bugs found in database management systems,” https: //www.manuelrigger.at/dbms-bugs/, 2022.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[13–17]",
+      "technique": "pqs",
+      "sentence": "Second, oracles used by existing systems [13–17] to detect logic bugs have strict requirements on the SQL statements.",
+      "context_before": "ion of a test case, they only consider limited semantic constraints. As a result, they will generate invalid SQL statements that cannot pass the semantic check of a DBMS. For instance, Squirrel [6] proposed semantics-guided instantiation and the follow-up work SQLRight [13] proposed Context-based IR instantiation to improve the semantic correctness of generated SQL statements. For simplicity, we will refer to their approaches as type-sensitive instantiation in this paper since they utilize a context-free strategy that only considers the correctness of the identifier type (e.g., table, column).",
+      "context_after": "This makes the fuzzing system only explore a narrow space of inputs, leading to limited bugs that can be detected. For example, NoREC [15] requires that the effective SQL query in a test case has WHERE clauses, thus it can only detect logic bugs due to the optimization of WHERE clauses. Another recent work, TQS [17], focusing on detecting logic bugs caused by equal-join optimization, fails to identify bugs arising from other types of optimization. Furthermore, both NoREC and TQS attempt to explore multiple query plans of queries but are only capable of covering partial query plans. Hence, an",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3923,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For example, NoREC [15] requires that the effective SQL query in a test case has WHERE clauses, thus it can only detect logic bugs due to the optimization of WHERE clauses.",
+      "context_before": "tion and the follow-up work SQLRight [13] proposed Context-based IR instantiation to improve the semantic correctness of generated SQL statements. For simplicity, we will refer to their approaches as type-sensitive instantiation in this paper since they utilize a context-free strategy that only considers the correctness of the identifier type (e.g., table, column). Second, oracles used by existing systems [13–17] to detect logic bugs have strict requirements on the SQL statements. This makes the fuzzing system only explore a narrow space of inputs, leading to limited bugs that can be detected.",
+      "context_after": "Another recent work, TQS [17], focusing on detecting logic bugs caused by equal-join optimization, fails to identify bugs arising from other types of optimization. Furthermore, both NoREC and TQS attempt to explore multiple query plans of queries but are only capable of covering partial query plans. Hence, an oracle that can be applied to SQL statements without strict requirements and explore more query plans is needed. Our Solution This work proposes two key techniques to address the limitations and solve the two innate challenges. The first key technique is called context-sensitive instanti",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 4156,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Furthermore, both NoREC and TQS attempt to explore multiple query plans of queries but are only capable of covering partial query plans.",
+      "context_before": "er type (e.g., table, column). Second, oracles used by existing systems [13–17] to detect logic bugs have strict requirements on the SQL statements. This makes the fuzzing system only explore a narrow space of inputs, leading to limited bugs that can be detected. For example, NoREC [15] requires that the effective SQL query in a test case has WHERE clauses, thus it can only detect logic bugs due to the optimization of WHERE clauses. Another recent work, TQS [17], focusing on detecting logic bugs caused by equal-join optimization, fails to identify bugs arising from other types of optimization.",
+      "context_after": "Hence, an oracle that can be applied to SQL statements without strict requirements and explore more query plans is needed. Our Solution This work proposes two key techniques to address the limitations and solve the two innate challenges. The first key technique is called context-sensitive instantiation. It performs context-sensitive analysis to collect all static semantic constraints (including but not limited to the identifier type) to improve the semantic correctnessarXiv:2312.04941v1 [cs.DB] 8 Dec 2023 of generated inputs. For instance, a SQL statement SELECT x1+ i1 FROM x2 WHERE x3=x4 has",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 4493,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We also compared Kangaroo with leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight.",
+      "context_before": "Prototype and Evaluation In this study, we implemented a prototype named Kangaroo and applied it to three widelyused database management systems (DBMSs): SQLite [18], PostgreSQL [19], and MySQL [20], to evaluate its effectiveness. Despite the fact that state-of-the-art tools have already extensively tested these DBMSs, Kangaroo successfully identified 50 new bugs, consisting of 9 logic bugs, 19 crash-causing bugs, and 22 assertion failure-inducing bugs. As of the time of writing, 28 of these bugs have been fixed, with 11 assigned CVEs [21]. These results demonstrate the efficacy of our system.",
+      "context_after": "Our evaluation reveals that Kangaroo surpasses these tools in terms of generating semantically valid SQL queries, exploring code paths, and detecting bugs. Specifically, after conducting a 24-hour test on the three DBMSs, Kangaroo successfully detected 17 bugs, while SQLancer, Squirrel, and SQLRight only identified 1, 3, and 6 bugs, respectively. In addition, Kangaroo proved more effective in generating valid SQL queries, achieving a 1.6x-1.9x improvement compared to the mutation-based tool Squirrel, and it explored 1.16x-1.3x more program states than the rule-based tool SQLancer. This work m",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7241,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, after conducting a 24-hour test on the three DBMSs, Kangaroo successfully detected 17 bugs, while SQLancer, Squirrel, and SQLRight only identified 1, 3, and 6 bugs, respectively.",
+      "context_before": "e-of-the-art tools have already extensively tested these DBMSs, Kangaroo successfully identified 50 new bugs, consisting of 9 logic bugs, 19 crash-causing bugs, and 22 assertion failure-inducing bugs. As of the time of writing, 28 of these bugs have been fixed, with 11 assigned CVEs [21]. These results demonstrate the efficacy of our system. We also compared Kangaroo with leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight. Our evaluation reveals that Kangaroo surpasses these tools in terms of generating semantically valid SQL queries, exploring code paths, and detecting bugs.",
+      "context_after": "In addition, Kangaroo proved more effective in generating valid SQL queries, achieving a 1.6x-1.9x improvement compared to the mutation-based tool Squirrel, and it explored 1.16x-1.3x more program states than the rule-based tool SQLancer. This work makes the following main contributions. •We revealed the challenges of effectively detecting DBMS bugs and the limitations of existing solutions. •We proposed two key techniques to solve the challenges, including context-sensitive instantiation to improve the semantic correctness of the mutated SQL queries and the MPE that can be applied to differen",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7498,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3x more program states than the rule-based tool SQLancer.",
+      "context_before": "leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight. Our evaluation reveals that Kangaroo surpasses these tools in terms of generating semantically valid SQL queries, exploring code paths, and detecting bugs. Specifically, after conducting a 24-hour test on the three DBMSs, Kangaroo successfully detected 17 bugs, while SQLancer, Squirrel, and SQLRight only identified 1, 3, and 6 bugs, respectively. In addition, Kangaroo proved more effective in generating valid SQL queries, achieving a 1.6x-1.9x improvement compared to the mutation-based tool Squirrel, and it explored 1.16x-1.",
+      "context_after": "This work makes the following main contributions. •We revealed the challenges of effectively detecting DBMS bugs and the limitations of existing solutions. •We proposed two key techniques to solve the challenges, including context-sensitive instantiation to improve the semantic correctness of the mutated SQL queries and the MPE that can be applied to different types of SQL queries to detect logic bugs. •We implemented and applied a prototype system to three popular DBMSs. Our system successfully detected 50 new bugs. The micro-benchmark also shows our system outperforms existing ones in genera",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7873,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[14–17, 24]",
+      "technique": "pqs",
+      "sentence": "The rule-based ones [14–17, 24] generate test cases following a predefined model to ensurethe generated SQL queries can pass the SQL parser.",
+      "context_before": ", Squirrel has to limit the complexity of generated queries to tolerate its incomplete and inaccurate constraints. What’s worse, the complexity of constraints dramatically increases with the complexity of the statement and the strictness of the DBMS checks. To better explore the core modules of DBMS, we propose a new method to generate more diverse and complex test cases, while at the same time, they can be statically confirming (level-7). 2.4. DBMS Testing Test Case Generation DBMS testing aims to trigger bugs by constructing abundant test cases. There are two methods to generate SQL queries.",
+      "context_after": "However, building a precise model requires domain knowledge. Besides, the generated inputs cannot efficiently explore the program’s state space since it wastes much effort on similar queries. The mutation-based method [6, 13, 25] generates new test cases by mutating seed queries. However, the general mutation strategies, like flipping bits, can not generate valid SQL queries that can pass the SQL parser. To address this issue, a common approach is to perform mutations based on grammar rules [6, 26]. It first generates a syntax tree for a given input, then creates specific mutations by using m",
+      "section": "2.4 DBMS Testing",
+      "page": 3,
+      "char_offset": 13649,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [15] is an oracle for logic bug detection.",
+      "context_before": "n oracle is a mechanism for determining whether the actual outputs match the expected outcomes. Differential testing uses different implementations of the same functionality as cross-referencing oracles. It provides the same inputs to a series of similar systems and then obverses the results. Any inconsistency between the results may indicate a potential bug. RAGS [27] is the first work that applies differential testing to find logic bugs for DBMS. Metamorphic testing addresses the test oracle problem based on the observation that a transformation of the input has a known effect on the output.",
+      "context_after": "It translates a query that is potentially optimized by DBMS to a query that can hardly be optimized. Although this approach has been effective in detecting bugs in widely-used DBMS [28], it can only be applied to a subset of SQL that can be translated. Our system leverages the idea of differential testing to compare the execution results of multiple query plans, hence named multi-plan execution (MPE). 3. Motivating Examples In this section, we use two real examples to demonstrate the advantages of our system’s two key techniques. 3.1. Context-Sensitive Instantiation We propose context-sensiti",
+      "section": "2.4 DBMS Testing",
+      "page": 3,
+      "char_offset": 15233,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[28]",
+      "technique": null,
+      "sentence": "Although this approach has been effective in detecting bugs in widely-used DBMS [28], it can only be applied to a subset of SQL that can be translated.",
+      "context_before": "the same functionality as cross-referencing oracles. It provides the same inputs to a series of similar systems and then obverses the results. Any inconsistency between the results may indicate a potential bug. RAGS [27] is the first work that applies differential testing to find logic bugs for DBMS. Metamorphic testing addresses the test oracle problem based on the observation that a transformation of the input has a known effect on the output. NoREC [15] is an oracle for logic bug detection. It translates a query that is potentially optimized by DBMS to a query that can hardly be optimized.",
+      "context_after": "Our system leverages the idea of differential testing to compare the execution results of multiple query plans, hence named multi-plan execution (MPE). 3. Motivating Examples In this section, we use two real examples to demonstrate the advantages of our system’s two key techniques. 3.1. Context-Sensitive Instantiation We propose context-sensitive instantiation because of two observations. First, previous works [6, 13] have proven the importance of semantic correctness to fuzzing efficiency, but existing type-sensitive instantiation still has large room for improvement. Second, MPE requires st",
+      "section": "2.4 DBMS Testing",
+      "page": 3,
+      "char_offset": 15383,
+      "cited_reference": {
+        "number": 28,
+        "text": "“Bugs found in database management systems,” https: //www.manuelrigger.at/dbms-bugs/, 2022.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "proposed three oracles for DBMS logic bug detection, including PQS [14], NoREC [15], and TLP [16], all of them put limitations on the SQL queries.",
+      "context_before": ". In TableReference x2 JOIN x3, columns x4and x5, which belong to the join constraint of the joined table, can only be the column of the joined table. Besides, each column should have a unique name within its dependent table. These additional constraints guarantee the semantic correctness of this SELECT statement that does not contain dynamic constraints. Enforcing dynamic constraints requires emulating the execution of statements, which is challenging. Thus, our semantic analysis ignores dynamic constraints and leaves it as part of future work. 3.2. Multi-Plan Execution Although Rigger et al.",
+      "context_after": "Similarly, TQS [17], which aims at detecting logic bugs in equal-join optimization, also limits the diversity of SQL statements. In addition, both TQS and NoREC attempt to explore different query plans for queries. NoREC only compares two distinct plans by transforming queries into semantically similar ones. TQS uses DBMS-specific features such as optimization switches and hints to iterate different query plans. However, this approach is not general and only covers partial query plans. To this end, we propose MPE, which has no such limitation. The basic idea is that our system hooks into the",
+      "section": "3.2 Multi-Plan Execution",
+      "page": 4,
+      "char_offset": 21523,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "In addition, both TQS and NoREC attempt to explore different query plans for queries.",
+      "context_before": "c correctness of this SELECT statement that does not contain dynamic constraints. Enforcing dynamic constraints requires emulating the execution of statements, which is challenging. Thus, our semantic analysis ignores dynamic constraints and leaves it as part of future work. 3.2. Multi-Plan Execution Although Rigger et al. proposed three oracles for DBMS logic bug detection, including PQS [14], NoREC [15], and TLP [16], all of them put limitations on the SQL queries. Similarly, TQS [17], which aims at detecting logic bugs in equal-join optimization, also limits the diversity of SQL statements.",
+      "context_after": "NoREC only compares two distinct plans by transforming queries into semantically similar ones. TQS uses DBMS-specific features such as optimization switches and hints to iterate different query plans. However, this approach is not general and only covers partial query plans. To this end, we propose MPE, which has no such limitation. The basic idea is that our system hooks into the DBMS optimizer to execute all query plans and compare their results. If the result of one query plan is different from the others, a logic bug is detected. CREATE TABLE t0(c0 INT); CREATE TABLE t1(c1 INT, c2 INT); I",
+      "section": "3.2 Multi-Plan Execution",
+      "page": 4,
+      "char_offset": 21799,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC only compares two distinct plans by transforming queries into semantically similar ones.",
+      "context_before": "rcing dynamic constraints requires emulating the execution of statements, which is challenging. Thus, our semantic analysis ignores dynamic constraints and leaves it as part of future work. 3.2. Multi-Plan Execution Although Rigger et al. proposed three oracles for DBMS logic bug detection, including PQS [14], NoREC [15], and TLP [16], all of them put limitations on the SQL queries. Similarly, TQS [17], which aims at detecting logic bugs in equal-join optimization, also limits the diversity of SQL statements. In addition, both TQS and NoREC attempt to explore different query plans for queries.",
+      "context_after": "TQS uses DBMS-specific features such as optimization switches and hints to iterate different query plans. However, this approach is not general and only covers partial query plans. To this end, we propose MPE, which has no such limitation. The basic idea is that our system hooks into the DBMS optimizer to execute all query plans and compare their results. If the result of one query plan is different from the others, a logic bug is detected. CREATE TABLE t0(c0 INT); CREATE TABLE t1(c1 INT, c2 INT); INSER TINT O t0 VALUES(0); INSER TINT O t1 VALUES(0, 1); CREATE INDEX t1_c1 ON t1(c1); SELECT *",
+      "section": "3.2 Multi-Plan Execution",
+      "page": 4,
+      "char_offset": 21885,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Why Existing Works Cannot Detect the Bug NoREC is one of the most effective DBMS logic bug detection oracles which requires the SELECT statements to satisfy some predefined rules, e.",
+      "context_before": "hat triggers a logic bug in SQLite. There are four query plans for the mutated SELECT statement Q1. One of them (the fourth one) returns a non-empty result that is different from the others. In the following, we use the proof-of-concept of one real-world bug of SQLite in Figure 2 to illustrate why the existing system cannot explore enough program states to detect the logic bug and how the MPE can capture such a bug. When the test case is executed, the DBMS optimizer finds four query plans for the SELECT statement. Plan one is optimal among four query plans and is executed by SQLite by default.",
+      "context_after": "g., having a WHERE and FROM clause. It shifts the condition in WHERE clause to the SELECT_TARGET, and then executes both to compare the execution result. However, the SELECT statement Q1 in Figure 2 does not meet the requirement (lacks WHERE clause), thus missing the bug triggered by the test case. TQS relies on optimization switches and hints to iterate different query plans, which limits its exploration of query plans. First, the optimization switches cannot force optimization adoptions nor control the scope of optimizations. Hints enables more fine-grained control than optimization switches",
+      "section": "3.2 Multi-Plan Execution",
+      "page": 5,
+      "char_offset": 24049,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "citation_marker",
+      "surface": "[13–17]",
+      "technique": "pqs",
+      "sentence": "in previous works [13–17], since other statements typically lack returning results to be checked.",
+      "context_before": "2 Attribute INSERT INTO x1(x2) VALUES (i1); Attr(x2)!=GENERATED x2 should not be a generated column Value SELECT * FROM x1 ORDER BY i1; i1∈{1,2,...,Size (x1)} i1 is an integer up to the number of columns of table x1 Dependency SELECT x1 FROM (SELECT x2, x3 FROM x4) x5; x1∈{x2,x3} x1 could be column x2 or x3 Distinct INSERT INTO x1(x2, x3) ...; DISTINCT [x2,x3] x2 and x3 should be two different column Composite SELECT (x1, x2) IN (TABLE x3) FROM x4; (DataType (x1),DataType (x2)))∈The data type of x2 and x3 should be {DataType (x3,1),DataType (x3,2)}the same as the first two columns in table x3.",
+      "context_after": "Adopting MPE to DBMSs is not straightforward. The first challenge is how to make DBMS execute all query plans with a small modification. Our key observation is that DBMS typically employs a plan choose function to compare the estimated cost of different query plans (including their sub-plans) and choose the best one. Therefore, we can execute any query plan by hooking the plan choose function. In addition, we added a loop before the entry function for query processing so that it can be executed multiple times with different plans until all plans are executed. Such modification has two advanta",
+      "section": "4.5 Multi-Plan Execution",
+      "page": 9,
+      "char_offset": 45587,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We use NoREC for SQLancer and SQLRight.",
+      "context_before": "406080100Validity(%) (d) SQLite query validity0 4 8 12 16 20 240102030Validity(%) (e) PostgreSQL query validity0 4 8 12 16 20 240255075100Validity(%) (f) MySQL query validity 0 4 8 12 16 20 24024681012unique bugs (g) SQLite unique bugs0 4 8 12 16 20 240246810unique bugs (h) MySQL unique bugs SQLancer Squirrel Kangaroo SQLRight Squirrelmultiplan Kangaroo!multiplanFigure 6: Comparison with existing tools. It also illustrates the contributions of context-sensitive instantiation and multi-plan execution. We exclude the results of detected bugs in PostgreSQL since only Kangaroo found one memory bug.",
+      "context_after": "Since SQLancer does not implement NoREC for MySQL, we use TLP instead. TABLE 5: The percentage of semantic correctness of query validation. SQLite PostgreSQL MySQL Squirrel 24,792(53.1%) 5,869(17.9%) 16,283(15.1%) SQLRight 28,710(61.5%) 8,508(26.0%) 35,012(32.5%) Kangaroo 32,880(70.4%) 16,269(49.7%) 46,763(43.4%) Total 46,672 32,753 107,693 type mismatch error in PostgreSQL. 5.3. Comparisons with Existing Tools We compare Kangaroo with three state-of-the-art and open source systems: Squirrel, SQLancer, and SQLRight. Similar to the previous ones, we also select SQLite, PostgreSQL, and MySQL fo",
+      "section": "5.2 Generating Valid Queries",
+      "page": 11,
+      "char_offset": 56379,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Since SQLancer does not implement NoREC for MySQL, we use TLP instead.",
+      "context_before": "lidity0 4 8 12 16 20 240102030Validity(%) (e) PostgreSQL query validity0 4 8 12 16 20 240255075100Validity(%) (f) MySQL query validity 0 4 8 12 16 20 24024681012unique bugs (g) SQLite unique bugs0 4 8 12 16 20 240246810unique bugs (h) MySQL unique bugs SQLancer Squirrel Kangaroo SQLRight Squirrelmultiplan Kangaroo!multiplanFigure 6: Comparison with existing tools. It also illustrates the contributions of context-sensitive instantiation and multi-plan execution. We exclude the results of detected bugs in PostgreSQL since only Kangaroo found one memory bug. We use NoREC for SQLancer and SQLRight.",
+      "context_after": "TABLE 5: The percentage of semantic correctness of query validation. SQLite PostgreSQL MySQL Squirrel 24,792(53.1%) 5,869(17.9%) 16,283(15.1%) SQLRight 28,710(61.5%) 8,508(26.0%) 35,012(32.5%) Kangaroo 32,880(70.4%) 16,269(49.7%) 46,763(43.4%) Total 46,672 32,753 107,693 type mismatch error in PostgreSQL. 5.3. Comparisons with Existing Tools We compare Kangaroo with three state-of-the-art and open source systems: Squirrel, SQLancer, and SQLRight. Similar to the previous ones, we also select SQLite, PostgreSQL, and MySQL for evaluation. We feed the same test cases to Squirrel, SQLRight, and Ka",
+      "section": "5.2 Generating Valid Queries",
+      "page": 11,
+      "char_offset": 56419,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Comparisons with Existing Tools We compare Kangaroo with three state-of-the-art and open source systems: Squirrel, SQLancer, and SQLRight.",
+      "context_before": "es the contributions of context-sensitive instantiation and multi-plan execution. We exclude the results of detected bugs in PostgreSQL since only Kangaroo found one memory bug. We use NoREC for SQLancer and SQLRight. Since SQLancer does not implement NoREC for MySQL, we use TLP instead. TABLE 5: The percentage of semantic correctness of query validation. SQLite PostgreSQL MySQL Squirrel 24,792(53.1%) 5,869(17.9%) 16,283(15.1%) SQLRight 28,710(61.5%) 8,508(26.0%) 35,012(32.5%) Kangaroo 32,880(70.4%) 16,269(49.7%) 46,763(43.4%) Total 46,672 32,753 107,693 type mismatch error in PostgreSQL. 5.3.",
+      "context_after": "Similar to the previous ones, we also select SQLite, PostgreSQL, and MySQL for evaluation. We feed the same test cases to Squirrel, SQLRight, and Kangaroo as the initial corpus and provide the same queries used to initialize the mutation library. SQLancer is a generate-based tool that does not require any initial inputs. We launch five fuzzing instances for each system and run each instance for 24 hours. We report the average result except for the bug number. We collect all bug reports from the five fuzzing instances as the final result and then count their first occurrence time for each uniq",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 56802,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer is a generate-based tool that does not require any initial inputs.",
+      "context_before": "rrel 24,792(53.1%) 5,869(17.9%) 16,283(15.1%) SQLRight 28,710(61.5%) 8,508(26.0%) 35,012(32.5%) Kangaroo 32,880(70.4%) 16,269(49.7%) 46,763(43.4%) Total 46,672 32,753 107,693 type mismatch error in PostgreSQL. 5.3. Comparisons with Existing Tools We compare Kangaroo with three state-of-the-art and open source systems: Squirrel, SQLancer, and SQLRight. Similar to the previous ones, we also select SQLite, PostgreSQL, and MySQL for evaluation. We feed the same test cases to Squirrel, SQLRight, and Kangaroo as the initial corpus and provide the same queries used to initialize the mutation library.",
+      "context_after": "We launch five fuzzing instances for each system and run each instance for 24 hours. We report the average result except for the bug number. We collect all bug reports from the five fuzzing instances as the final result and then count their first occurrence time for each unique bug. Figure 6 shows the evaluation result. Detected Unique Bugs As indicated in Figure 6gh, Kangaroo outperforms existing tools on all three DBMSs. For SQLite, Kangaroo found 12 unique bugs, including eight memory bugs and four logic bugs. SQLRight found three bugs, one of which is a logic bug. Squirrel found only oneb",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 57188,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Squirrel and SQLancer detected five and one memory bugs, respectively.",
+      "context_before": "Bugs As indicated in Figure 6gh, Kangaroo outperforms existing tools on all three DBMSs. For SQLite, Kangaroo found 12 unique bugs, including eight memory bugs and four logic bugs. SQLRight found three bugs, one of which is a logic bug. Squirrel found only onebug which is also covered by Kangaroo. For PostgreSQL, only Kangaroo found one bug. This is not surprising as previous works have revealed that PostgreSQL is more robust than most other DBMSs. As for MySQL, Kangaroo found a total of ten bugs, including eight memory bugs and two logic bugs. SQLRight found two memory bugs and one logic bug.",
+      "context_after": "SQLancer, the only generationbased tool, found the least bugs across all comparisons, demonstrating the advantage of the mutation-based method to detect DBMS bugs. Explored New Edges Figure 6abc shows that Kangaroo performs better than others on all three DBMS systems. It explores 52%, 44%, and 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively. Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive insta",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 58203,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer, the only generationbased tool, found the least bugs across all comparisons, demonstrating the advantage of the mutation-based method to detect DBMS bugs.",
+      "context_before": "all three DBMSs. For SQLite, Kangaroo found 12 unique bugs, including eight memory bugs and four logic bugs. SQLRight found three bugs, one of which is a logic bug. Squirrel found only onebug which is also covered by Kangaroo. For PostgreSQL, only Kangaroo found one bug. This is not surprising as previous works have revealed that PostgreSQL is more robust than most other DBMSs. As for MySQL, Kangaroo found a total of ten bugs, including eight memory bugs and two logic bugs. SQLRight found two memory bugs and one logic bug. Squirrel and SQLancer detected five and one memory bugs, respectively.",
+      "context_after": "Explored New Edges Figure 6abc shows that Kangaroo performs better than others on all three DBMS systems. It explores 52%, 44%, and 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively. Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive instantiation. Generated Valid Queries We treat a query as invalid if DBMS reports any form of error during the execution. As shown in Figure 6def, SQLancer achieves the",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 58274,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "It explores 52%, 44%, and 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively.",
+      "context_before": "g. This is not surprising as previous works have revealed that PostgreSQL is more robust than most other DBMSs. As for MySQL, Kangaroo found a total of ten bugs, including eight memory bugs and two logic bugs. SQLRight found two memory bugs and one logic bug. Squirrel and SQLancer detected five and one memory bugs, respectively. SQLancer, the only generationbased tool, found the least bugs across all comparisons, demonstrating the advantage of the mutation-based method to detect DBMS bugs. Explored New Edges Figure 6abc shows that Kangaroo performs better than others on all three DBMS systems.",
+      "context_after": "Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive instantiation. Generated Valid Queries We treat a query as invalid if DBMS reports any form of error during the execution. As shown in Figure 6def, SQLancer achieves the highest query validity. This result is reasonable because SQLancer follows very limited grammar rules to generate SQL statements. For example, SQLancer does not support generating subqueries that are prone to s",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 58544,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As shown in Figure 6def, SQLancer achieves the highest query validity.",
+      "context_before": "the mutation-based method to detect DBMS bugs. Explored New Edges Figure 6abc shows that Kangaroo performs better than others on all three DBMS systems. It explores 52%, 44%, and 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively. Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive instantiation. Generated Valid Queries We treat a query as invalid if DBMS reports any form of error during the execution.",
+      "context_after": "This result is reasonable because SQLancer follows very limited grammar rules to generate SQL statements. For example, SQLancer does not support generating subqueries that are prone to semantic errors. That makes it easier to generate valid statements but limits the diversity. This could be the main reason that it explores the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensit",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 58991,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This result is reasonable because SQLancer follows very limited grammar rules to generate SQL statements.",
+      "context_before": "re 6abc shows that Kangaroo performs better than others on all three DBMS systems. It explores 52%, 44%, and 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively. Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive instantiation. Generated Valid Queries We treat a query as invalid if DBMS reports any form of error during the execution. As shown in Figure 6def, SQLancer achieves the highest query validity.",
+      "context_after": "For example, SQLancer does not support generating subqueries that are prone to semantic errors. That makes it easier to generate valid statements but limits the diversity. This could be the main reason that it explores the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensit",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 59062,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer does not support generating subqueries that are prone to semantic errors.",
+      "context_before": "nd 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively. Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive instantiation. Generated Valid Queries We treat a query as invalid if DBMS reports any form of error during the execution. As shown in Figure 6def, SQLancer achieves the highest query validity. This result is reasonable because SQLancer follows very limited grammar rules to generate SQL statements.",
+      "context_after": "That makes it easier to generate valid statements but limits the diversity. This could be the main reason that it explores the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP. Doin",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 59168,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M25",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP.",
+      "context_before": "n in Figure 6def, SQLancer achieves the highest query validity. This result is reasonable because SQLancer follows very limited grammar rules to generate SQL statements. For example, SQLancer does not support generating subqueries that are prone to semantic errors. That makes it easier to generate valid statements but limits the diversity. This could be the main reason that it explores the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers.",
+      "context_after": "Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements. Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target. After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL. Besides, all logic bugs found by NoREC are covered by MPE. SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Ke",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 59599,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements.",
+      "context_before": "rors. That makes it easier to generate valid statements but limits the diversity. This could be the main reason that it explores the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP.",
+      "context_after": "Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target. After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL. Besides, all logic bugs found by NoREC are covered by MPE. SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribu",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 59859,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target.",
+      "context_before": "s the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP. Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements.",
+      "context_after": "After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL. Besides, all logic bugs found by NoREC are covered by MPE. SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribution of context-sensitive instantiation, we build Kangaroo !multiplan by disabling the MPE in Kangaroo. Without the MPE, Kangaroo !multiplan focuses",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 59986,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL.",
+      "context_before": "iceably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP. Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements. Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target.",
+      "context_after": "Besides, all logic bugs found by NoREC are covered by MPE. SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribution of context-sensitive instantiation, we build Kangaroo !multiplan by disabling the MPE in Kangaroo. Without the MPE, Kangaroo !multiplan focuses on memory bug detection. We use Squirrel as a baseline since it performs better than SQLancer in detecting memory bugs. We also ported the MPE module into Squirrel, denoted Squi",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 60135,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Besides, all logic bugs found by NoREC are covered by MPE.",
+      "context_before": "ing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP. Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements. Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target. After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL.",
+      "context_after": "SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribution of context-sensitive instantiation, we build Kangaroo !multiplan by disabling the MPE in Kangaroo. Without the MPE, Kangaroo !multiplan focuses on memory bug detection. We use Squirrel as a baseline since it performs better than SQLancer in detecting memory bugs. We also ported the MPE module into Squirrel, denoted Squirrel multiplan. By comparing Squirrel with Squirrel multipl",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 60312,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLRight TLP and SQLancer PQS fail to detect any logic bug.",
+      "context_before": "instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP. Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements. Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target. After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL. Besides, all logic bugs found by NoREC are covered by MPE.",
+      "context_after": "5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribution of context-sensitive instantiation, we build Kangaroo !multiplan by disabling the MPE in Kangaroo. Without the MPE, Kangaroo !multiplan focuses on memory bug detection. We use Squirrel as a baseline since it performs better than SQLancer in detecting memory bugs. We also ported the MPE module into Squirrel, denoted Squirrel multiplan. By comparing Squirrel with Squirrel multiplan, we can verify that MPE itself is effective. We adopt the",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 60371,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We use Squirrel as a baseline since it performs better than SQLancer in detecting memory bugs.",
+      "context_before": "ur testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL. Besides, all logic bugs found by NoREC are covered by MPE. SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribution of context-sensitive instantiation, we build Kangaroo !multiplan by disabling the MPE in Kangaroo. Without the MPE, Kangaroo !multiplan focuses on memory bug detection.",
+      "context_after": "We also ported the MPE module into Squirrel, denoted Squirrel multiplan. By comparing Squirrel with Squirrel multiplan, we can verify that MPE itself is effective. We adopt the same strategy to measure the number of explored edges, the query validity rate, and the number of unique bugs. Figure 6 shows the evaluation results. Context-Sensitive Instantiation Context-sensitive instantiation directly improves the correctness level (Table 1) of generated test cases, thereby greatly improving the efficiency of fuzzing. As shown in Figure 6def, tools using contextsensitive instantiation achieve sign",
+      "section": "5.4 Benefits of the Two Key Techniques",
+      "page": 12,
+      "char_offset": 60759,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M32",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [15] generates equivalent queries by shifting the conditions in the WHERE clause to the SELECT expression.",
+      "context_before": "put with different DBMS systems. Slutz proposed RAGE [27] for finding logic bugs in DBMS by running the same query on different DBMS and comparing their results. Jinho et al. developed APOLLO [5], a system to find performance regression bugs by executing the same query on the DBMSs with different versions. However, RAGE can only be applied to common features of different DBMSs, and APOLLO can only detect bugs introduced or fixed by newer versions. Another approach is based on metamorphic testing which identifies bugs by running two queries with equivalent functionality on the same DBMS system.",
+      "context_after": "TLP [16] partitions a query lacking where clause into three subqueries whose where clause are x IS TRUE ,x IS FALSE, and x IS NULL. Both of them are limited to the queries that can be converted. The last approach tries to build the test case along with the corresponding ground truth result. ADUSA [37] generates all data and the full expected result for a query. However, generating full expected results as ground truth can be expensive which inhibits it from finding more bugs. To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result. It",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 66520,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M33",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [16] partitions a query lacking where clause into three subqueries whose where clause are x IS TRUE ,x IS FALSE, and x IS NULL.",
+      "context_before": "y on different DBMS and comparing their results. Jinho et al. developed APOLLO [5], a system to find performance regression bugs by executing the same query on the DBMSs with different versions. However, RAGE can only be applied to common features of different DBMSs, and APOLLO can only detect bugs introduced or fixed by newer versions. Another approach is based on metamorphic testing which identifies bugs by running two queries with equivalent functionality on the same DBMS system. NoREC [15] generates equivalent queries by shifting the conditions in the WHERE clause to the SELECT expression.",
+      "context_after": "Both of them are limited to the queries that can be converted. The last approach tries to build the test case along with the corresponding ground truth result. ADUSA [37] generates all data and the full expected result for a query. However, generating full expected results as ground truth can be expensive which inhibits it from finding more bugs. To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result. It synthesizes a query that is expected to fetch a single, randomly-selected row. By checking whether this row is fetched, PQS can dete",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 66633,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result.",
+      "context_before": "ystem. NoREC [15] generates equivalent queries by shifting the conditions in the WHERE clause to the SELECT expression. TLP [16] partitions a query lacking where clause into three subqueries whose where clause are x IS TRUE ,x IS FALSE, and x IS NULL. Both of them are limited to the queries that can be converted. The last approach tries to build the test case along with the corresponding ground truth result. ADUSA [37] generates all data and the full expected result for a query. However, generating full expected results as ground truth can be expensive which inhibits it from finding more bugs.",
+      "context_after": "It synthesizes a query that is expected to fetch a single, randomly-selected row. By checking whether this row is fetched, PQS can detect logic bugs in the DBMS. DBMS test cases generation. DBMS requires structural inputs to manipulate data in the database. Structural input generation mainly falls into two categories: generate-based approaches and mutation-based approaches. The generation-based approach [7, 14–17, 27, 37, 38] is effective in generating syntax-correct test cases since it typically follows a grammar model that describes the format of the input. However, these grammar rules are",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 67114,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M35",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "By checking whether this row is fetched, PQS can detect logic bugs in the DBMS.",
+      "context_before": "where clause are x IS TRUE ,x IS FALSE, and x IS NULL. Both of them are limited to the queries that can be converted. The last approach tries to build the test case along with the corresponding ground truth result. ADUSA [37] generates all data and the full expected result for a query. However, generating full expected results as ground truth can be expensive which inhibits it from finding more bugs. To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result. It synthesizes a query that is expected to fetch a single, randomly-selected row.",
+      "context_after": "DBMS test cases generation. DBMS requires structural inputs to manipulate data in the database. Structural input generation mainly falls into two categories: generate-based approaches and mutation-based approaches. The generation-based approach [7, 14–17, 27, 37, 38] is effective in generating syntax-correct test cases since it typically follows a grammar model that describes the format of the input. However, these grammar rules are helpless in improving the semantic correctness of the test case. SQLsmith [7] is a popular DBMS testing tool that can generate syntax-correct test cases from AST.",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 67311,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M36",
+      "found_by": "citation_marker",
+      "surface": "[7, 14–17, 27, 37, 38]",
+      "technique": "pqs",
+      "sentence": "The generation-based approach [7, 14–17, 27, 37, 38] is effective in generating syntax-correct test cases since it typically follows a grammar model that describes the format of the input.",
+      "context_before": "generating full expected results as ground truth can be expensive which inhibits it from finding more bugs. To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result. It synthesizes a query that is expected to fetch a single, randomly-selected row. By checking whether this row is fetched, PQS can detect logic bugs in the DBMS. DBMS test cases generation. DBMS requires structural inputs to manipulate data in the database. Structural input generation mainly falls into two categories: generate-based approaches and mutation-based approaches.",
+      "context_after": "However, these grammar rules are helpless in improving the semantic correctness of the test case. SQLsmith [7] is a popular DBMS testing tool that can generate syntax-correct test cases from AST. Despite it having found over 100 memory bugs in popular DBMSs, SQLsmith achieves quite a low accuracy on semantics which might inhibit it from finding bugs hidden in the deep logic. QAGen [38] proves that generating a completely valid queryis NP-complete. It improves semantic correctness by combining traditional query processing and symbolic execution. Previous works also try to improve query generat",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 67606,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M4",
+        "M21",
+        "M25",
+        "M31"
+      ],
+      "describes_as_state_of_the_art": [
+        "M4",
+        "M13",
+        "M17"
+      ]
+    },
+    "suppressed": [
+      {
+        "mention_id": "M17",
+        "pattern": "compares_with",
+        "suppressed_because": "the sentence is framed as related work, so it describes somebody else's approach"
+      }
+    ]
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-10T15:11:27Z",
+    "is_model_written": true,
+    "summary": "Kangaroo tests DBMSs with context-sensitive instantiation and multi-plan execution. Its argument is that the existing oracles constrain which SQL they can judge -- NoREC needs a WHERE clause, so it reaches only bugs in WHERE optimisation -- and that they explore only part of a query's plan space. Kangaroo instantiates queries against the live schema and forces execution down several plans. Over 24 hours on three DBMSs it found 17 bugs.",
+    "narrative": "SQLancer is Kangaroo's baseline and the source of the limitation it targets. The paper describes the three oracles -- PQS, NoREC and TLP -- as each placing limitations on the SQL queries they can judge, and works through why NoREC in particular cannot detect the motivating bug. In the comparison against Squirrel, SQLancer and SQLRight, SQLancer found 1 bug to Kangaroo's 17, and Kangaroo reached several times more program states than what the paper calls the rule-based tool SQLancer.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "motivation",
+      "M3": "motivation",
+      "M4": "baseline",
+      "M5": "result_comparison",
+      "M6": "result_comparison",
+      "M7": "background",
+      "M8": "definition",
+      "M9": "background",
+      "M10": "motivation",
+      "M11": "definition",
+      "M12": "definition",
+      "M13": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is run as a baseline; Kangaroo's instantiation and multi-plan execution are its own."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The oracles are described as constraining what they can judge, which is the gap Kangaroo addresses with a different mechanism rather than a development of theirs."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M5",
+          "M6"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "We also compared Kangaroo with leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight.",
+            "section": "1 Introduction",
+            "page": 2
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "Specifically, after conducting a 24-hour test on the three DBMSs, Kangaroo successfully detected 17 bugs, while SQLancer, Squirrel, and SQLRight only identified 1, 3, and 6 bugs, respectively.",
+            "section": "1 Introduction",
+            "page": 2
+          },
+          {
+            "mention_id": "M6",
+            "sentence": "3x more program states than the rule-based tool SQLancer.",
+            "section": "1 Introduction",
+            "page": 2
+          }
+        ],
+        "reasoning": "M4 names SQLancer among the leading tools compared against, M5 gives the 24-hour bug counts -- 17 against SQLancer's 1 -- and M6 the difference in program states reached."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is called a leading DBMS testing tool and a rule-based one; the paper stops short of calling it the state of the art, and its argument is about what the oracles cannot reach."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2402_00292.json b/_data/papers/paper_arxiv_2402_00292.json
new file mode 100644
index 0000000..47439ee
--- /dev/null
+++ b/_data/papers/paper_arxiv_2402_00292.json
@@ -0,0 +1,518 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:05:22Z",
+  "paper": {
+    "id": "paper:arxiv:2402.00292",
+    "title": "Effective Bug Detection in Graph Database Engines: An LLM-based Approach",
+    "authors": [
+      "Jiayi Wu",
+      "Zhengyu Wu",
+      "Ronghua Li",
+      "Hongchao Qin",
+      "Guoren Wang"
+    ],
+    "year": 2024,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2402.00292",
+    "s2_paper_id": "1072e6d03a19b157356c5da57257b2b92eefc68c",
+    "url": "https://arxiv.org/abs/2402.00292",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2402.00292",
+      "retrieved_at": "2026-09-10T15:05:22Z",
+      "chars": 69045,
+      "content_sha256": "sha256:f431211699ae0e694fb596bfa4bbecc7b74c09c1a3a6c6ec5f99c14069db226d"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Ibrahim Abdelaziz, Essam Mansour, Mourad Ouzzani, Ashraf Aboulnaga, and Panos Kalnis. 2017. Query optimizations over decentralized RDF graphs. In. IEEE, 139–142.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "DataStax Dylan Bethune-Waddell Expero Google Orion Health IBM Rafael Fernandes Robert Dale Seeq Amazon, Aurelius. 2023. JanusGraph. https: //janusgraph.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Sjoerd Mullender Andreas Seltenreich, Bo Tang. 2022. SQLsmith. https://github. com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Renzo Angles and Claudio Gutierrez. 2008. Survey of graph database models. ACM Computing Surveys (CSUR) 40, 1 (2008), 1–39.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Dana Angluin. 1990. Negative results for equivalence queries. Machine Learning 5 (1990), 121–150.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al .2020. Language models are few-shot learners. Advances in neural information processing systems 33 (2020), 1877–1901.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Donald D Chamberlin and Raymond F Boyce. 1974. SEQUEL: A structured English query language. In Proceedings of the 1974 ACMSIGFIDET (now SIGMOD) workshop on Data description, access and control. 249–264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Zihan Chen, Lei Nico Zheng, Cheng Lu, Jialu Yuan, and Di Zhu. 2023. ChatGPT Informed Graph Neural Network for Stock Movement Prediction. arXiv preprint arXiv:2306.03763 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Edgar Frank Codd. 1983. A relational model of data for large shared data banks. Commun. ACM 26, 1 (1983), 64–69.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Jiaxi Cui, Zongjian Li, Yang Yan, Bohua Chen, and Li Yuan. 2023. Chatlaw: Open-source legal large language model with integrated external knowledge bases. arXiv preprint arXiv:2306.16092 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Andrzej Czerepicki. 2016. Application of graph databases for transport purposes. Bulletin of the Polish Academy of Sciences. Technical Sciences 64, 3 (2016), 457–466.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "SAP Enterprise. 2023. OrientDB Community Edition. http://www.orientdb.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Robert B Evans and Alberto Savoia. 2007. Differential testing: a new approach to change detection. In The 6th Joint Meeting on European software engineering conference and the ACMSIGSOFT Symposium on the Foundations of Software Engineering: Companion Papers. 549–552.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Apache Software Foundation. 2023. HugeGraph. https://hugegraph.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Apache Software Foundation. 2023. TinkerGraph. https://tinkerpop.incubator. apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and Andrés Taylor. 2018. Cypher: An evolving query language for property graphs. InProceedings of the 2018 international conference on management of data. 1433– 1445.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Qing Gao, Hansheng Zhang, Jie Wang, Yingfei Xiong, Lu Zhang, and Hong Mei. 2015. Fixing recurring crash bugs via analyzing q&a sites (T). In 2015 30th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 307–318.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Marielet Guillermo, Maverick Rivera, Ronnie Concepcion, Robert Kerwin Billones, Argel Bandala, Edwin Sybingco, Alexis Fillone, and Elmer Dadios. 2022. Graph Database-modelled Public Transportation Data for Geographic Insight Web Application. In Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing (SNPD). IEEE, 2–7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Edwin Hewitt and Leonard J Savage. 1955. Symmetric measures on Cartesian products. Trans. Amer. Math. Soc. 80, 2 (1955), 470–501.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting bugs in Cypher graph database engines. In Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Bitnine Global Inc. 2023. Agensgraph. https://bitnine.net/agensgraph/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Neo4j Inc. 2023. Neo4j. https://neo4j.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Himanshu Jain, Daniel Kroening, Natasha Sharygina, and Edmund Clarke. 2005. Word level predicate abstraction and refinement for verifying RTL Verilog. In Proceedings of the 42nd annual Design Automation Conference. 445–450.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Matteo Kamm. 2022. Testing Graph Databases using Predicate Partitioning. Master’s thesis. ETH Zurich.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al. 2023. Segment anything. arXiv preprint arXiv:2304.02643 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Takahiro Konno, Runhe Huang, Tao Ban, and Chuanhe Huang. 2017. Goods recommendation based on retail knowledge in a Neo4j graph database combined with an inference mechanism implemented in jess. In 2017 IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Computed, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation (Smar",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Bingfeng Luo, Yansong Feng, Zheng Wang, Songfang Huang, Rui Yan, and Dongyan Zhao. 2018. Marrying up regular expressions with neural networks: A case study for spoken language understanding. arXiv preprint arXiv:1805.05588 (2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Hanjia Lyu, Song Jiang, Hanqing Zeng, Yinglong Xia, and Jiebo Luo. 2023. Llmrec: Personalized recommendation via prompting large language models. arXiv preprint arXiv:2307.15780 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jun Ma and Bo Wang. 2023. Segment anything in medical images. arXiv preprint arXiv:2304.12306 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "William M McKeeman. 1998. Differential testing for software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "OpenAI. 2022. ChatGPT. https://openai.com/blog/chatgpt/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al .2022. Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems 35 (2022), 27730–27744.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Rob Reagan and Rob Reagan. 2018. Cosmos DB. Web Applications on Azure: Developing for Global Scale (2018), 187–255.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Yuxiang Ren, Hao Zhu, Jiawei Zhang, Peng Dai, and Liefeng Bo. 2021. Ensemfdet: An ensemble approach to fraud detection based on bipartite graph. In 2021 IEEE 37th International Conference on Data Engineering (ICDE). IEEE, 2039–2044.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Manuel Rigger. 2022. SQLancer. https://github.com/sqlancer/sqlancer.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "Ian Robinson, Jim Webber, and Emil Eifrem. 2015. Graph databases: new opportunities for connected data. \" O’Reilly Media, Inc.\".",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Marko A Rodriguez. 2015. The gremlin graph traversal machine and language (invited talk). In Proceedings of the 15th Symposium on Database Programming Languages. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Marko A Rodriguez. 2015. The gremlin graph traversal machine and language (invited talk). In Proceedings of the 15th Symposium on Database Programming Languages. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Sudipta Sen, Akash Mehta, Runa Ganguli, and Soumya Sen. 2021. Recommendation of influenced products using association rule mining: Neo4j as a case study. SN Computer Science 2 (2021), 1–17.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Donald R Slutz. 1998. Massive stochastic testing of SQL. In VLDB, Vol. 98. Citeseer, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "solid IT. 2023. DB-Engines Ranking of Graph DBMS. https://db-engines.com/ en/ranking/graph+ dbms.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting Logic Bugs of Join Optimizations in DBMS. Proceedings of the ACM on Management of Data 1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Ran Wang, Zhengyi Yang, Wenjie Zhang, and Xuemin Lin. 2020. An empirical study on recent graph database systems. In Knowledge Science, Engineering and Management: 13th International Conference, KSEM 2020, Hangzhou, China, August 28–30, 2020, Proceedings, Part I 13. Springer, 328–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and S Yu Philip. 2020. A comprehensive survey on graph neural networks. IEEE transactions on neural networks and learning systems 32, 1 (2020), 4–24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Jiawei Zhang. 2023. Graph-ToolFormer: To Empower LLMs with Graph Reasoning Ability via Prompt Augmented by ChatGPT. arXiv preprint arXiv:2304.11116 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, et al .2023. A survey of large language models. arXiv preprint arXiv:2303.18223 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding bugs in Gremlin-based graph database systems via randomized differential testing. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis. 302–313. 12",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 35,
+      "text": "Manuel Rigger. 2022. SQLancer. https://github.com/sqlancer/sqlancer.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 23,24,37]",
+      "technique": "tlp",
+      "sentence": "The former typically involves detecting bugs in graph database engines by generating equivalent queries[ 5] or utilizing predicate partitioning[ 23,24,37].",
+      "context_before": "es can dynamically generate real-time recommended content based on users’ most recent actions and social connections[ 26,42]. In the field of logistics and freight transportation, graph database engines can be harnessed to determine optimal delivery routes, thereby reducing costs and enhancing efficiency[11, 18]. As shown in Table 1, existing methods for detecting bugs in graph database engines can be broadly divided into two categories: methods for utilizng various queries to detect bugs within the same graph database engines, methods applying the same query on various graph database engines.",
+      "context_after": "The latter approach entails running the same queries on different graph database engines and identifying bugs based on discrepancies in query results[ 20,50]. The former method demands users with substantial prior knowledge on the graph model and query objectives when generating equivalent queries to identify bugs in the graph database engine[ 5]. In the predicate partitioning approach for detecting graph database engine bugs[ 24], it starts by constructing queries based on a top-down expression generator[ 35], and then apply Ternary Logic Partitioning techniques to create correspondent queri",
+      "section": null,
+      "page": 1,
+      "char_offset": 4417,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "In the predicate partitioning approach for detecting graph database engine bugs[ 24], it starts by constructing queries based on a top-down expression generator[ 35], and then apply Ternary Logic Partitioning techniques to create correspondent queries, namely the \"True, False or Null\" queries, for running on graph database engine.",
+      "context_before": "ame graph database engines, methods applying the same query on various graph database engines. The former typically involves detecting bugs in graph database engines by generating equivalent queries[ 5] or utilizing predicate partitioning[ 23,24,37]. The latter approach entails running the same queries on different graph database engines and identifying bugs based on discrepancies in query results[ 20,50]. The former method demands users with substantial prior knowledge on the graph model and query objectives when generating equivalent queries to identify bugs in the graph database engine[ 5].",
+      "context_after": "Such a method assess whether the subsetarXiv:2402.00292v1 [cs.DB] 1 Feb 2024 Table 1: Comparison of the proposed DGDB with closely related works. Method Category Language Not require large prior knowledgeHigh proportion of non-empty-result queries Grand[50]the method applying the same query on various graph database enginesgremlin % % GDSmith[20]the method applying the same query on various graph database enginescypher %! PP-DB[24]the method for utilizng various queries to detect bugs within the same graph database enginesall % % DGDBthe method applying the same query on various graph databas",
+      "section": null,
+      "page": 1,
+      "char_offset": 4923,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[3,35,43]",
+      "technique": null,
+      "sentence": "Most current research focuses on the latter approach and the most widely adopted method is running randomly generated queries [3,35,43] on different graph database engines for bug detection.",
+      "context_before": "g the same query on various graph database enginesgremlin % % GDSmith[20]the method applying the same query on various graph database enginescypher %! PP-DB[24]the method for utilizng various queries to detect bugs within the same graph database enginesall % % DGDBthe method applying the same query on various graph database enginesall! ! of the final query results intersect to detect graph database engine bugs. While this approach successfully identifies some logical bugs, its effectiveness is limited due to the lower quality of queries generation and the manual need for reducing test samples.",
+      "context_after": "However, the generated queries often return empty results in many samples, which compromise the efficiency for bug detection. To address this issue, Grand[ 50] proposed a method based on traversal models to generate gremlin queries[ 40], significantly increasing the non-empty-result query ratio. However, this method is only applicable to gremlin-based graph database engines and requires users to master the gremlin API[ 41]. GDsmith[ 20] proposed a method to generate cypher queries[ 16] based on the guidance of property graphs. This method first generates a cypher skeleton, then creates patter",
+      "section": null,
+      "page": 2,
+      "char_offset": 6132,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger. 2022. SQLancer. https://github.com/sqlancer/sqlancer.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS[ 38] selects a target data from randomly generated tables, generates conditional expressions based on the target data, constructs an SQL query with a where or join clause, and determines the presence of bugs by checking if the result is included in the result set.",
+      "context_before": "ub. 5 RELATED WORK Recently, researchers have proposed various methods for detecting bugs in relational database engines. SQLsmith[ 3] is a fuzz testing method for detecting bugs in relational database engines. It first generates SQL query randomly according to the Abstract Syntax Tree (AST) generator, then simplifies the generated SQL query through SQL Reducer. Finally, it determines the presence of bugs based on whether the generated SQL query can successfully run on the relational database engines. While this method is effective in detecting crash bugs, it cannot identify wrong-result bugs.",
+      "context_after": "While this method can effectively detect bugs in relational database engines, its applicability is limited across different relational database engines due to variations in SQL syntax and query optimization strategies. NoRec[ 36] converts the original SQL query into a nonoptimized SQL query and compares the results of these two SQL queries for consistency. However, this method cannot detect bugs in SQL queries with subquery structures. TLP[ 37] transforms randomly generated original SQL queries into three different logical queries based on the true, false, and null ternary logic. It detects b",
+      "section": null,
+      "page": 11,
+      "char_offset": 54992,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "NoRec[ 36] converts the original SQL query into a nonoptimized SQL query and compares the results of these two SQL queries for consistency.",
+      "context_before": "database engines. While this method is effective in detecting crash bugs, it cannot identify wrong-result bugs. PQS[ 38] selects a target data from randomly generated tables, generates conditional expressions based on the target data, constructs an SQL query with a where or join clause, and determines the presence of bugs by checking if the result is included in the result set. While this method can effectively detect bugs in relational database engines, its applicability is limited across different relational database engines due to variations in SQL syntax and query optimization strategies.",
+      "context_after": "However, this method cannot detect bugs in SQL queries with subquery structures. TLP[ 37] transforms randomly generated original SQL queries into three different logical queries based on the true, false, and null ternary logic. It detects bugs in relational database engines by comparing the result sets of the transformed three SQL queries with the original SQL query.TQS[ 45] proposes a testing framework capable of detecting logical bugs arising from multi-table join queries. The authors treat the database schema as a graph and use biased random walks to generate SQL queries. Based on biased r",
+      "section": null,
+      "page": 11,
+      "char_offset": 55480,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP[ 37] transforms randomly generated original SQL queries into three different logical queries based on the true, false, and null ternary logic.",
+      "context_before": "e target data, constructs an SQL query with a where or join clause, and determines the presence of bugs by checking if the result is included in the result set. While this method can effectively detect bugs in relational database engines, its applicability is limited across different relational database engines due to variations in SQL syntax and query optimization strategies. NoRec[ 36] converts the original SQL query into a nonoptimized SQL query and compares the results of these two SQL queries for consistency. However, this method cannot detect bugs in SQL queries with subquery structures.",
+      "context_after": "It detects bugs in relational database engines by comparing the result sets of the transformed three SQL queries with the original SQL query.TQS[ 45] proposes a testing framework capable of detecting logical bugs arising from multi-table join queries. The authors treat the database schema as a graph and use biased random walks to generate SQL queries. Based on biased random walks and graph embedding methods, high-quality SQL queries are generated to detect numerous logical bugs in relational database engines. In contrast to detecting bugs in relational database engines, the identification of",
+      "section": null,
+      "page": 11,
+      "char_offset": 55701,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-10T15:11:01Z",
+    "is_model_written": true,
+    "summary": "This work uses a large language model to generate test queries for graph database engines, on the argument that the existing generators produce queries bounded by their templates and transformation rules. It contrasts differential testing across engines with metamorphic approaches that test one engine against itself.",
+    "narrative": "TLP reaches this paper twice over. It is described directly among the relational oracles -- PQS, NoREC and TLP each summarised by mechanism -- and again as the technique GDBMeter carries into graph databases: the paper explains that the predicate-partitioning approach builds queries from a top-down expression generator and then applies Ternary Logic Partitioning to produce the true, false and null variants. That partitioning line is the alternative its LLM-based generation is positioned against, and nothing of SQLancer's is run or reused.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "background",
+      "M4": "definition",
+      "M5": "definition",
+      "M6": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Queries are generated by a language model; nothing describes reuse of SQLancer's code or generator."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The oracles are described as the existing approaches this work departs from, not developed."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The mentions describe the techniques rather than reporting a run against them; no comparison result involving TLP, NoREC or PQS appears."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Predicate partitioning is called widely adopted, which describes its uptake rather than claiming it is the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2406_09469.json b/_data/papers/paper_arxiv_2406_09469.json
new file mode 100644
index 0000000..1f2138f
--- /dev/null
+++ b/_data/papers/paper_arxiv_2406_09469.json
@@ -0,0 +1,1417 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:05:16Z",
+  "paper": {
+    "id": "paper:arxiv:2406.09469",
+    "title": "Conformance Testing of Relational DBMS Against SQL Specifications",
+    "authors": [
+      "Shuang Liu",
+      "Chenglin Tian",
+      "Jun Sun",
+      "Ruifeng Wang",
+      "Wei Lu",
+      "Yongxin Zhao",
+      "Yinxing Xue",
+      "Junjie Wang",
+      "Xiaoyong Du"
+    ],
+    "year": 2024,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2406.09469",
+    "s2_paper_id": "86f309b0e355f622be488d1813049a953bce65c3",
+    "url": "https://arxiv.org/abs/2406.09469",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2406.09469",
+      "retrieved_at": "2026-09-10T15:05:16Z",
+      "chars": 97306,
+      "content_sha256": "sha256:27ebd6cab84ea1e46f5013d5a504231dc3986cfdf151d8e4edec4b96c231824d"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 21,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2287
+      },
+      {
+        "number": "II",
+        "title": "PRELIMINARY",
+        "start": 9638
+      },
+      {
+        "number": "A",
+        "title": "Prolog",
+        "start": 9654
+      },
+      {
+        "number": "III",
+        "title": "FORMAL SEMANTICS OF SQL",
+        "start": 18899
+      },
+      {
+        "number": "A",
+        "title": "Basic symbol definition",
+        "start": 27291
+      },
+      {
+        "number": "B",
+        "title": "Semantics of Expressions",
+        "start": 28352
+      },
+      {
+        "number": "C",
+        "title": "Semantics of SQL keywords",
+        "start": 30947
+      },
+      {
+        "number": "IV",
+        "title": "CONFORMANCE TESTING",
+        "start": 32971
+      },
+      {
+        "number": "A",
+        "title": "Overview of our approach",
+        "start": 32995
+      },
+      {
+        "number": "B",
+        "title": "Coverage guided test case generation",
+        "start": 34590
+      },
+      {
+        "number": "C",
+        "title": "Prolog implementation of SQL formal semantics",
+        "start": 44139
+      },
+      {
+        "number": "D",
+        "title": "Result comparison",
+        "start": 50393
+      },
+      {
+        "number": "E",
+        "title": "Discussion on Extensibility",
+        "start": 51954
+      },
+      {
+        "number": "V",
+        "title": "EVALUATION",
+        "start": 54045
+      },
+      {
+        "number": "A",
+        "title": "Experiment setup",
+        "start": 54602
+      },
+      {
+        "number": "B",
+        "title": "Experiment results",
+        "start": 58389
+      },
+      {
+        "number": "C",
+        "title": "Threats to Validity",
+        "start": 79583
+      },
+      {
+        "number": "VI",
+        "title": "RELATED WORK",
+        "start": 82290
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 87912
+      },
+      {
+        "number": "J",
+        "title": "Wang, and X. Du, “Conformance testing of relational dbms against",
+        "start": 93944
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Amazon,” https://www.amazon.com/, 1995, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“ebay,” https://www.ebay.com/, 1995, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“Mysql,” https://www.mysql.com, 1995, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "“Booking,” https://www.booking.com/, 1996, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "“Postgresql,” https://www.postgresql.org, 1996, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "“Sqlite,” https://www.sqlite.org/index.html, 2000, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "“Facebook,” https://www.facebook.com/, 2004, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "“Jsqlparser,” https://github.com/JSQLParser/JSqlParser, 2011, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "“International organization for standardization. (2016). iso/iec 90752:2016: Information technology – database languages – sql/foundation,” 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "“Iso/iec 9075-2:2016,” https://www.iso.org/standard/63555.html, 2016, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "“Oceanbase,” https://en.oceanbase.com/, 2016, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "“Tidb,” https://www.pingcap.com/tidb/, 2016, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "“Duckdb,” https://duckdb.org, 2019, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "“Sqlancer,” https://github.com/sqlancer/sqlancer, 2019, accessed on November 10, 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "“An error occurred when the cast function converted the numerical value in the form of scientific notation,” https://sqlite.org/forum/forumpost/ 3f085531bf, 2022, accessed on November 11, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "V. Benzaken and E. Contejean, “A coq mechanised formal semantics for realistic sql queries: formally reconciling sql and bag relational algebra,” inProceedings of the 8th ACMSIGPLAN International Conference on Certified Programs and Proofs, 2019, pp. 249–261.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "J. Castelein, M. Aniche, M. Soltani, A. Panichella, and A. van Deursen, “Search-based test data generation for sql queries,” in Proceedings of the 40th international conference on software engineering, 2018, pp. 1220–1230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "S. Ceri and G. Gottlob, “Translating sql into relational algebra: Optimization, semantics, and equivalence of sql queries,” IEEE Transactions on software engineering, no. 4, pp. 324–345, 1985.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "T. Y. Chen, S. C. Cheung, and S. Yiu, “Metamorphic testing: A new approach for generating next test cases,” CoRR, vol. abs/2002.12543, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "T. Y. Chen, F. Kuo, H. Liu, P. Poon, D. Towey, T. H. Tse, and Z. Q. Zhou, “Metamorphic testing: A review of challenges and opportunities,” ACM Comput. Surv., vol. 51, no. 1, pp. 4:1–4:27, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "X. Chen, C. Wang, and A. Cheung, “Testing query execution engines with mutations,” in Proceedings of the workshop on Testing Database Systems, 2020, pp. 1–5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "J. Cheney and W. Ricciotti, “Comprehending nulls,” in The 18th International Symposium on Database Programming Languages, 2021, pp. 3–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "H. R. Chinaei, “An ordered bag semantics for sql,” Master’s thesis, University of Waterloo, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "S. Chu, C. Wang, K. Weitz, and A. Cheung, “Cosette: An automated prover for sql.” in CIDR, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "S. Chu, K. Weitz, A. Cheung, and D. Suciu, “Hottsql: Proving query rewrites with univalent sql semantics,” ACMSIGPLAN Notices, vol. 52, no. 6, pp. 510–524, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "W. F. Clocksin and C. S. Mellish, Programming in PROLOG. Springer Science & Business Media, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammarfree dbms fuzzing,” in Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "B. Ghit, N. Poggi, J. Rosen, R. Xin, and P. Boncz, “Sparkfuzz: Searching correctness regressions in modern query engines,” in Proceedings of the workshop on Testing Database Systems, 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "P. Guagliardo and L. Libkin, “A formal semantics of sql queries, its validation, and applications,” Proceedings of the VLDB Endowment, vol. 11, no. 1, pp. 27–39, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "International Organization for Standardization, Information technology – Database languages – SQL – Part 2: Foundation (SQL/Foundation), 2016, no. 9075-2:2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "——, Information technology – Database languages – SQL – Part 2: Foundation (SQL/Foundation), 2016, no. 9075-2:2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "J. Jiao, S. Kan, S.-W. Lin, D. Sanan, Y. Liu, and J. Sun, “Semantic understanding of smart contracts: Executable operational semantics of solidity,” in. IEEE, 2020, pp. 1695–1712.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “Apollo: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment, vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of {DBMS } with coverage-based guidance,” in 31st USENIX Security Symposium (USENIX Security 22), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "S. Liu, C. Tian, J. Sun, R. Wang, W. Lu, Y. Zhao, Y. Xue, J. Wang, and X. Du, “Conformance testing of relational dbms against sql specifications (technical report),” https://github.com/DBMSTesting/ Technical-report, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "G. Malecha, G. Morrisett, A. Shinnar, and R. Wisnesky, “Toward a verified relational database management system,” in Proceedings of the 37th annual ACMSIGPLAN-SIGACT symposium on Principles of programming languages, 2010, pp. 237–248.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "M. Mitzenmacher and E. Upfal, Probability and Computing: Randomization and Probabilistic Techniques in Algorithms and Data Analysis, 2nd ed. USA: Cambridge University Press, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "M. Negri, G. Pelagatti, and L. Sbattella, “Formal semantics of sql queries,” ACM Transactions on Database Systems (TODS), vol. 16, no. 3, pp. 513–534, 1991.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "W. Ricciotti and J. Cheney, “A formalization of sql with nulls,” Journal of Automated Reasoning, vol. 66, no. 4, pp. 989–1030, 2022. 21",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 42,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "G. Ros ,u and T. F. S ,erb˘anut˘a, “An overview of the k semantic framework,” The Journal of Logic and Algebraic Programming, vol. 79, no. 6, pp. 397–434, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "R. Schumi and J. Sun, “Exais: executable ai semantics,” in Proceedings of the 44th International Conference on Software Engineering, 2022, pp. 859–870.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "S. Segura, G. Fraser, A. B. S ´anchez, and A. R. Cort ´es, “A survey on metamorphic testing,” IEEE Trans. Software Eng., vol. 42, no. 9, pp. 805–824, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "A. Seltenreich, B. Tang, and S. Mullender, “Sqlsmith,” 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "D. R. Slutz, “Massive stochastic testing of sql,” in VLDB, vol. 98. Citeseer, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "J. Van den Bussche and S. Vansummeren, “Translating sql into the relational algebra,” Course notes, Hasselt University and Universit ´e Libre de Bruxelles, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "M. Veanes, N. Tillmann, and J. De Halleux, “Qex: Symbolic sql query explorer,” in Logic for Programming, Artificial Intelligence, and Reasoning: 16th International Conference, LPAR-16, Dakar, Senegal, April 25–May 1, 2010, Revised Selected Papers 16. Springer, 2010, pp. 425–446.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang, and Y. Jiang, “Industry practice of coverage-guided enterprise-level dbms fuzzing,” in ing: Software Engineering in Practice (ICSE-SEIP). IEEE, 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Q. Zhou, J. Arulraj, S. Navathe, W. Harris, and J. Wu, “A symbolic approach to proving query equivalence under bag semantics,” arXiv preprint arXiv:2004.00481, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Z. Zhou, S. Xiang, and T. Y. Chen, “Metamorphic testing for software quality assessment: A study of search engines,” IEEE Trans. Software Eng., vol. 42, no. 3, pp. 264–284, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "X. Zhu, S. Wen, S. Camtepe, and Y. Xiang, “Fuzzing: a survey for roadmap,” ACM Computing Surveys (CSUR), vol. 54, no. 11s, pp. 1–36, 2022.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 14,
+      "text": "“Sqlancer,” https://github.com/sqlancer/sqlancer, 2019, accessed on November 10, 2023.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 41,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[14]",
+      "technique": null,
+      "sentence": "The serious impacts of bugs in RDBMS have been discussed by various existing studies [14], [34], [40], [41], and thus it is critical to detect those bugs.",
+      "context_before": "uage for relational databases, and its specification is formally documented in ISO/IEC 9075:2016 [10]. As the parser and executor of SQL queries, an RDBMS should conform to the SQL specification to ensure the correct implementation of the semantics. As of 2023, there are more than 416 different implementations of relational databases, yet many of these implementations deviate from the specification [18], [47], [53]. Bugs have also been reported due to violations of the SQL specifications [15], which may potentially lead to data integrity issues or even security vulnerabilities in the database.",
+      "context_after": "As the golden standard for correct SQL behavior, the specification of SQL should be clearly described, and an RDBMS should conform to the SQL specification. Inconsistencies between RDBMS implementations and the specification can lead to unexpected results. Figure 1 presents two motivating examples, one bug and one inconsistency that our approach detected. Figure 1a is a SQL query that triggers a bug in TiDB version 6.6.01, which occurs when performing a bitwise operation on negative numbers, which are by default signed numbers. The root cause is that TiDB incorrectly represents the result of",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 3156,
+      "cited_reference": {
+        "number": 14,
+        "text": "“Sqlancer,” https://github.com/sqlancer/sqlancer, 2019, accessed on November 10, 2023.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[40]–[42]",
+      "technique": "norec",
+      "sentence": "However, existing approaches on testing RDBMSs either use different RDBMSs as the test oracle [47] or propose metamorphic relations [34], [40]–[42].",
+      "context_before": "ific data type in SQL representing unknown data. This inconsistency poses substantial challenges for users of database systems. When transitioning between databases and leveraging features with inconsistent implementations, users may encounter unexpected outcomes. From the motivational examples, we observe that failing to respect the SQL specification or unclearly documented specifications can result in bugs or inconsistent implementations across different RDBMSs, potentially confusing users. Therefore, it is critical to test the conformance of RDBMS implementations with the SQL specification.",
+      "context_after": "None of those approaches consider testing the conformance of RDBMS implementations with SQL specifications. Consequently, they are only scratching the surface in evaluating the correctness of RDBMS. To address the issue of automatic conformance testing between SQL specifications and RDBMS implementations, two main challenges arise. Firstly, the SQL specification is written in natural language, which is not directly executable. Secondly, it is challenging to generate test queries that comprehensively 3 cover all aspects defined in the SQL specification, including descriptions of keywords and p",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 6680,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Here, we 10 enhance the syntax-guided generation method, SQLancer [14], with coverage-guided test case generation.",
+      "context_before": "g 7, we first introduced the combination rules of different types of keyword semantics (semantic rule 102), and then we introduce the composite semantics based on those rules (semantic rules 103-104). IV. CONFORMANCE TESTING A. Overview of our approach Figure 8 illustrates the overview of our conformance testing approach, which consists of four components. Initially, we define the formal semantics of SQL, as detailed in Section III. Next, we implement the SQL formal semantics in Prolog, which serves as an oracle for conformance testing. The third component is dedicated to test case generation.",
+      "context_after": "This enhancement is based on three coverage criteria, i.e., keyword coverage, rule coverage, and composite rule coverage, which we proposed based on the formal semantics we defined. The final component is dedicated to query results comparison, wherein the query results of the tested RDBMS and our Prolog implementation are compared to identify conformance issues. Conformance issues in our work refer to two types of issues, i.e., bugs or inconsistencies. Both issues are due to violating the SQL semantics defined in the SQL specification, or unclear or missing descriptions in the SQL specificati",
+      "section": "A Overview of our approach",
+      "page": 9,
+      "char_offset": 33370,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Coverage guided test case generation The state-of-the-art practice in test case generation involves randomly generating SQL queries guided by the syntax of SQL, among which SQLancer [14] stands out as one of the most effective tools of this kind.",
+      "context_before": "s defined in the SQL specification, or unclear or missing descriptions in the SQL specification. Bugs are defects that are confirmed by developers. Inconsistencies refer to inconsistent result produced by the tested RDBMS and SEMCONT. Inconsistencies are also confirmed by developers, yet they perceive them as deliberate design choices rather than bugs. We report these inconsistencies because various RDBMSs make differing design decisions, leading to varied query results that may potentially perplex users. This also underscore the importance of a comprehensively documented SQL specification. B.",
+      "context_after": "It is tailored to the syntactical structures of various RDBMSs. SQLancer considers database objects, such as tables, views, and indexes, as well as keywords and functionalities within query statements. Throughout the generation process, SQLancer maintains a set of keywords and functionalities, from which it randomly selects keywords to incorporate into the test cases, subject to syntactic rules of SQL (so that they remain syntactically valid). However, this generation process is entirely random and does not consider coverage of the SQL semantics. As a result, certain aspects of the semantics",
+      "section": "B Coverage guided test case generation",
+      "page": 10,
+      "char_offset": 34592,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer considers database objects, such as tables, views, and indexes, as well as keywords and functionalities within query statements.",
+      "context_before": "deliberate design choices rather than bugs. We report these inconsistencies because various RDBMSs make differing design decisions, leading to varied query results that may potentially perplex users. This also underscore the importance of a comprehensively documented SQL specification. B. Coverage guided test case generation The state-of-the-art practice in test case generation involves randomly generating SQL queries guided by the syntax of SQL, among which SQLancer [14] stands out as one of the most effective tools of this kind. It is tailored to the syntactical structures of various RDBMSs.",
+      "context_after": "Throughout the generation process, SQLancer maintains a set of keywords and functionalities, from which it randomly selects keywords to incorporate into the test cases, subject to syntactic rules of SQL (so that they remain syntactically valid). However, this generation process is entirely random and does not consider coverage of the SQL semantics. As a result, certain aspects of the semantics may never be tested. To address this problem, we propose three coverage criteria, i.e., keyword coverage, rule coverage, and composite rule coverage, based on the formal semantics we defined. Each cover",
+      "section": "B Coverage guided test case generation",
+      "page": 10,
+      "char_offset": 34903,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Throughout the generation process, SQLancer maintains a set of keywords and functionalities, from which it randomly selects keywords to incorporate into the test cases, subject to syntactic rules of SQL (so that they remain syntactically valid).",
+      "context_before": "g to varied query results that may potentially perplex users. This also underscore the importance of a comprehensively documented SQL specification. B. Coverage guided test case generation The state-of-the-art practice in test case generation involves randomly generating SQL queries guided by the syntax of SQL, among which SQLancer [14] stands out as one of the most effective tools of this kind. It is tailored to the syntactical structures of various RDBMSs. SQLancer considers database objects, such as tables, views, and indexes, as well as keywords and functionalities within query statements.",
+      "context_after": "However, this generation process is entirely random and does not consider coverage of the SQL semantics. As a result, certain aspects of the semantics may never be tested. To address this problem, we propose three coverage criteria, i.e., keyword coverage, rule coverage, and composite rule coverage, based on the formal semantics we defined. Each coverage criterion defines coverage at a different granularity level. Keyword coverage simply assesses whether each individual keyword in the SQL specification is covered. Rule coverage goes a step further by calculating whether each semantic rule, wh",
+      "section": "B Coverage guided test case generation",
+      "page": 10,
+      "char_offset": 35041,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We adopt SQLancer to randomly generatea large number of SQL statements (line 2), which serves as the seed pool of our query generation algorithm.",
+      "context_before": "and match the rules based on the parameter or input of that keyword. In this case, the SELECT rule that takes a column list as input and the FROM rule that takes one table as input are matched to form the composite rule covered by the given query. The time complexity of Algorithm 3 is O(n) with n being the number of operators in the given SQL query. Coverage-guided query generation. Algorithm 2 describes the process of query generation guided by coverage. We set the initial coverage to be 0 and the declare covered set (coveredSet ), which records the covered keywords, rules, or combined rules.",
+      "context_after": "The algorithm begins with randomly selecting an SQL statement from the seed pool (line 6), and mutates the query based on the mutation rules we proposed. Then we calculate coverage of the mutated query (line 8). We keep this mutation process (line 5-9) until the mutated query increases the overall coverage. Then the mutated query is added into the seed pool for future test case generation. The mutated query is executed to explore potential inconsistencies (line 11). The process terminates upon timeout. Function CalculateCoverage calculates the coverage of the given query. It first checks whet",
+      "section": "B Coverage guided test case generation",
+      "page": 11,
+      "char_offset": 41312,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[40]",
+      "technique": "norec",
+      "sentence": "It is important to note that we used the latest release of each RDBMS, which has been extensively tested by existing approaches [40], [41].",
+      "context_before": "e conducted all experiments on a server with two Intel(R) Xeon(R) Platinum 8260 CPUs at 2.30 GHz and 502 GB of memory, running Ubuntu 18.04.6 LTS. The SQL formal semantics were implemented in Prolog, while the SQL query generation program was developed in Java. We ran the experiments using Java version 11.0.15.1. Target RDBMS. We selected six popular and widely used RDBMSs, each offering a range of distinct features and application scenarios, to demonstrate the effectiveness of our approach. The statistics of these RDBMSs, as obtained from their opensource repositories, are shown in Table III.",
+      "context_after": "MySQL [3] and PostgreSQL [5] are the two most popular open-source database management systems. SQLite [6] and DuckDB [13] are both embedded DBMSs, running within the process of other applications. TiDB [12] and OceanBase [11] are popular distributed RDBMSs. 14 TABLE IV: Bugs and inconsistencies detected by SEMCONT SNID Target Type Reason Status 1 109146 MySQL Bug missing spec duplicate 2 109837 MySQL Bug missing spec comfirmed 3 109842 MySQL Bug missing spec comfirmed 4 109149 MySQL Bug missing spec comfirmed 5 110438 MySQL Bug violate spec comfirmed 6 109147 MySQL Inconsistency missing spec",
+      "section": "A Experiment setup",
+      "page": 13,
+      "char_offset": 55223,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+      "context_before": "irmed 21 42378 TiDB Bug violate spec comfirmed 22 42379 TiDB Bug violate spec comfirmed 23 42377 TiDB Bug violate spec comfirmed 24 42773 TiDB Inconsistency missing spec comfirmed 25 40995 TiDB Inconsistency missing spec comfirmed 26 7e03a4420a SQLite Bug missing spec comfirmed 27 3f085531bf SQLite Bug missing spec comfirmed 28 6e4d3e389e SQLite Bug violate spec comfirmed 29 411bce39d0 SQLite Inconsistency missing spec comfirmed 30 6804 DuckDB Bug violate spec fixed 31 2104 OceanBase Inconsistency missing spec confirmed 32 2105 OceanBase Inconsistency missing spec confirmed Compared baselines.",
+      "context_after": "NoREC constructs two semantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results. TLP, on the other hand, partitions the conditional expression of the original query into three segments, corresponding to the three possible results, i.e., TRUE, FALSE, and NULL, of the conditional expression. It then compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 57222,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC constructs two semantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results.",
+      "context_before": "rmed 24 42773 TiDB Inconsistency missing spec comfirmed 25 40995 TiDB Inconsistency missing spec comfirmed 26 7e03a4420a SQLite Bug missing spec comfirmed 27 3f085531bf SQLite Bug missing spec comfirmed 28 6e4d3e389e SQLite Bug violate spec comfirmed 29 411bce39d0 SQLite Inconsistency missing spec comfirmed 30 6804 DuckDB Bug violate spec fixed 31 2104 OceanBase Inconsistency missing spec confirmed 32 2105 OceanBase Inconsistency missing spec confirmed Compared baselines. We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+      "context_after": "TLP, on the other hand, partitions the conditional expression of the original query into three segments, corresponding to the three possible results, i.e., TRUE, FALSE, and NULL, of the conditional expression. It then compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness in RDBMS bug detection. Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34]. SQLancer adopts a generative approach for query generation",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 57346,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP, on the other hand, partitions the conditional expression of the original query into three segments, corresponding to the three possible results, i.",
+      "context_before": "27 3f085531bf SQLite Bug missing spec comfirmed 28 6e4d3e389e SQLite Bug violate spec comfirmed 29 411bce39d0 SQLite Inconsistency missing spec comfirmed 30 6804 DuckDB Bug violate spec fixed 31 2104 OceanBase Inconsistency missing spec confirmed 32 2105 OceanBase Inconsistency missing spec confirmed Compared baselines. We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS. NoREC constructs two semantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results.",
+      "context_after": "e., TRUE, FALSE, and NULL, of the conditional expression. It then compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness in RDBMS bug detection. Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34]. SQLancer adopts a generative approach for query generation and SQLRight adopts a mutation-based approach for generating queries. Therefore, in our experiment, we have four combined settings (concerning the oracl",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 57500,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34].",
+      "context_before": "mantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results. TLP, on the other hand, partitions the conditional expression of the original query into three segments, corresponding to the three possible results, i.e., TRUE, FALSE, and NULL, of the conditional expression. It then compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness in RDBMS bug detection.",
+      "context_after": "SQLancer adopts a generative approach for query generation and SQLRight adopts a mutation-based approach for generating queries. Therefore, in our experiment, we have four combined settings (concerning the oracle and query generation method) for the compared baselines, i.e., NoREC (SQLancer), NoREC (SQLRight), TLP (SQLancer) and TLP (SQLRight). B. Experiment results RQ1: Bugs and inconsistencies. We ran SEMCONT on four RDBMSs for a period of 3 months and reported the detected issues to the corresponding developer communities. Table IV shows the details of the confirmed bugs and inconsistencie",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 57970,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer adopts a generative approach for query generation and SQLRight adopts a mutation-based approach for generating queries.",
+      "context_before": "her does not, executes the queries and compare the results. TLP, on the other hand, partitions the conditional expression of the original query into three segments, corresponding to the three possible results, i.e., TRUE, FALSE, and NULL, of the conditional expression. It then compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness in RDBMS bug detection. Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34].",
+      "context_after": "Therefore, in our experiment, we have four combined settings (concerning the oracle and query generation method) for the compared baselines, i.e., NoREC (SQLancer), NoREC (SQLRight), TLP (SQLancer) and TLP (SQLRight). B. Experiment results RQ1: Bugs and inconsistencies. We ran SEMCONT on four RDBMSs for a period of 3 months and reported the detected issues to the corresponding developer communities. Table IV shows the details of the confirmed bugs and inconsistencies in four RDBMSs detected by SEMCONT. We have submitted 32 issues and 19 of them are confirmed by the developers as bugs. Out of",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 58041,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", NoREC (SQLancer), NoREC (SQLRight), TLP (SQLancer) and TLP (SQLRight).",
+      "context_before": "en compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness in RDBMS bug detection. Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34]. SQLancer adopts a generative approach for query generation and SQLRight adopts a mutation-based approach for generating queries. Therefore, in our experiment, we have four combined settings (concerning the oracle and query generation method) for the compared baselines, i.e.",
+      "context_after": "B. Experiment results RQ1: Bugs and inconsistencies. We ran SEMCONT on four RDBMSs for a period of 3 months and reported the detected issues to the corresponding developer communities. Table IV shows the details of the confirmed bugs and inconsistencies in four RDBMSs detected by SEMCONT. We have submitted 32 issues and 19 of them are confirmed by the developers as bugs. Out of the issues identified, 23 are related to scalarexpressions and 9 to other keywords, including joins and various relational operators. Our primary objective is to detect inconsistencies between RDBMS implementations and",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 58316,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[40]–\n[42]",
+      "technique": "norec",
+      "sentence": "They are also under-specified in the SQL standard and insufficiently tested by existing approaches [34], [40]– [42].",
+      "context_before": "es and 19 of them are confirmed by the developers as bugs. Out of the issues identified, 23 are related to scalarexpressions and 9 to other keywords, including joins and various relational operators. Our primary objective is to detect inconsistencies between RDBMS implementations and the SQL specification by generating test cases that achieve high coverage across different SQL keywords. Our implementation focuses on scalar expressions, query expressions, and predicates. Among these, scalar expressions are the most complex, as they often involve combinations of multiple keywords and subqueries.",
+      "context_after": "In contrast, query expressions and predicates are clearly defined in the SQL specification, leading to fewer ambiguities across RDBMSs. Moreover, they have been extensively tested by prior research [34], [40]–[42], making it more challenging to uncover new inconsistencies. Among the confirmed bugs, 1 has been reported previously and 1 has been fixed. The remaining 13 issues are confirmed as inconsistencies, and the developers claim that they were their design choices. Among the 13 inconsistencies, 2 of them are due to violation of the SQL specification and 11 of them are due to unclear descri",
+      "section": "B Experiment results",
+      "page": 14,
+      "char_offset": 59304,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[40]–[42]",
+      "technique": "norec",
+      "sentence": "Moreover, they have been extensively tested by prior research [34], [40]–[42], making it more challenging to uncover new inconsistencies.",
+      "context_before": "tween RDBMS implementations and the SQL specification by generating test cases that achieve high coverage across different SQL keywords. Our implementation focuses on scalar expressions, query expressions, and predicates. Among these, scalar expressions are the most complex, as they often involve combinations of multiple keywords and subqueries. They are also under-specified in the SQL standard and insufficiently tested by existing approaches [34], [40]– [42]. In contrast, query expressions and predicates are clearly defined in the SQL specification, leading to fewer ambiguities across RDBMSs.",
+      "context_after": "Among the confirmed bugs, 1 has been reported previously and 1 has been fixed. The remaining 13 issues are confirmed as inconsistencies, and the developers claim that they were their design choices. Among the 13 inconsistencies, 2 of them are due to violation of the SQL specification and 11 of them are due to unclear descriptions in the SQL specification. Developers of different RDBMSs could have different interpretations on the SQL specification, and thus design and implement their RDBMS differently. Among the nine inconsistencies arising from unclear standard descriptions, eight inconsisten",
+      "section": "B Experiment results",
+      "page": 14,
+      "char_offset": 59557,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[40]–[42]",
+      "technique": "norec",
+      "sentence": "It is noteworthy that all four databases have been extensively tested by existing methods [40]–[42], yet SEMCONTis still able to detect bugs that were not detected by those approaches.",
+      "context_before": "SQL specification and 11 of them are due to unclear descriptions in the SQL specification. Developers of different RDBMSs could have different interpretations on the SQL specification, and thus design and implement their RDBMS differently. Among the nine inconsistencies arising from unclear standard descriptions, eight inconsistencies were detected in both MySQL (109147, 109148, 109836, 109845, 109962, 110711) and TiDB (42773, 40995). The queries triggering the inconsistencies have the same results when executed in MySQL, TiDB, and MariaDB databases, and are different from that of PostgreSQL.",
+      "context_after": "By a careful inspection on the detected bugs, we find that most of them indeed violate the SQL specification. One of the most representative bugs is related to the mishandling of NULL operands in keyword operations. According to the SQL specification, “ If the value of one or more, <string value expression >s,<datetime value expression >s,<interval value expression >s, and <collection value expression >s that are simply contained in a <numeric value function >is the NULL value, then the result of the <numeric value function >is the NULL value ” [30]. However, MySQL violates the specification b",
+      "section": "B Experiment results",
+      "page": 14,
+      "char_offset": 60562,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In Figure 9, SQLancer+ keyword syntax represents the setting of adding keywords and the corresponding generation rules which were not supported by SQLancer.",
+      "context_before": "onsistencies are either due to RDBMS violating the SQL specification, or missing or unclear SQL specification. RQ2: Effectiveness of the coverage criteria? We measure the effectiveness of the proposed coverage criteria in two aspects, i.e., whether they are effective in guiding generating test cases that achieve higher coverage, and whether they are effective in guiding generating test cases that uncover unknown bugs or inconsistencies. The experiment results on keyword coverage, rule coverage and composite rule coverage improvement are shown in Figure 9, Figure 10 and Figure 11, respectively.",
+      "context_after": "SQLancer+ keyword syntax greatly improved the keyword coverage for all four databases. Notably, within the first 1500 SQL statements, over 80% of the keywords were covered on all four databases, with SQLite achieving an impressive keyword coverage of 99%. Keyword coverage guided query generation (SQLancer+ keyword coverage) further improves the keyword coverage, and achieved 100% keyword coverage within the first 200 generate queries for all databases, demonstrating the effectiveness of our keyword-guided query generation method. Figure 10 shows the results on rule coverage, which show simila",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 65358,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer+ keyword syntax greatly improved the keyword coverage for all four databases.",
+      "context_before": "We measure the effectiveness of the proposed coverage criteria in two aspects, i.e., whether they are effective in guiding generating test cases that achieve higher coverage, and whether they are effective in guiding generating test cases that uncover unknown bugs or inconsistencies. The experiment results on keyword coverage, rule coverage and composite rule coverage improvement are shown in Figure 9, Figure 10 and Figure 11, respectively.In Figure 9, SQLancer+ keyword syntax represents the setting of adding keywords and the corresponding generation rules which were not supported by SQLancer.",
+      "context_after": "Notably, within the first 1500 SQL statements, over 80% of the keywords were covered on all four databases, with SQLite achieving an impressive keyword coverage of 99%. Keyword coverage guided query generation (SQLancer+ keyword coverage) further improves the keyword coverage, and achieved 100% keyword coverage within the first 200 generate queries for all databases, demonstrating the effectiveness of our keyword-guided query generation method. Figure 10 shows the results on rule coverage, which show similar trend with that on keyword coverage. Due to the limited support of SQL features, e.g.",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 65514,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Keyword coverage guided query generation (SQLancer+ keyword coverage) further improves the keyword coverage, and achieved 100% keyword coverage within the first 200 generate queries for all databases, demonstrating the effectiveness of our keyword-guided query generation method.",
+      "context_before": "own bugs or inconsistencies. The experiment results on keyword coverage, rule coverage and composite rule coverage improvement are shown in Figure 9, Figure 10 and Figure 11, respectively.In Figure 9, SQLancer+ keyword syntax represents the setting of adding keywords and the corresponding generation rules which were not supported by SQLancer. SQLancer+ keyword syntax greatly improved the keyword coverage for all four databases. Notably, within the first 1500 SQL statements, over 80% of the keywords were covered on all four databases, with SQLite achieving an impressive keyword coverage of 99%.",
+      "context_after": "Figure 10 shows the results on rule coverage, which show similar trend with that on keyword coverage. Due to the limited support of SQL features, e.g., data types, by SQLancer, especially for DBMS such as DuckDB and TiDB, relying only on SQLancer achieves low rule coverage, as shown in Figure 10. Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax. We can observe that adding those missing features improves the rule coverage, especially for DuckDB and TiDB. Rule coverage-guided query generation (SQLancer+ rul",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 65770,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", data types, by SQLancer, especially for DBMS such as DuckDB and TiDB, relying only on SQLancer achieves low rule coverage, as shown in Figure 10.",
+      "context_before": "Notably, within the first 1500 SQL statements, over 80% of the keywords were covered on all four databases, with SQLite achieving an impressive keyword coverage of 99%. Keyword coverage guided query generation (SQLancer+ keyword coverage) further improves the keyword coverage, and achieved 100% keyword coverage within the first 200 generate queries for all databases, demonstrating the effectiveness of our keyword-guided query generation method. Figure 10 shows the results on rule coverage, which show similar trend with that on keyword coverage. Due to the limited support of SQL features, e.g.",
+      "context_after": "Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax. We can observe that adding those missing features improves the rule coverage, especially for DuckDB and TiDB. Rule coverage-guided query generation (SQLancer+ rule coverage) achieves the highest rule coverage with the fewest number of queries. The results indicate the effectiveness of our rule coverage-guided query generation algorithm. Figure 11 depicts the improvements in composite rule coverage by the generated queries for the four databases. With SQLan",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 66201,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax.",
+      "context_before": "yword coverage of 99%. Keyword coverage guided query generation (SQLancer+ keyword coverage) further improves the keyword coverage, and achieved 100% keyword coverage within the first 200 generate queries for all databases, demonstrating the effectiveness of our keyword-guided query generation method. Figure 10 shows the results on rule coverage, which show similar trend with that on keyword coverage. Due to the limited support of SQL features, e.g., data types, by SQLancer, especially for DBMS such as DuckDB and TiDB, relying only on SQLancer achieves low rule coverage, as shown in Figure 10.",
+      "context_after": "We can observe that adding those missing features improves the rule coverage, especially for DuckDB and TiDB. Rule coverage-guided query generation (SQLancer+ rule coverage) achieves the highest rule coverage with the fewest number of queries. The results indicate the effectiveness of our rule coverage-guided query generation algorithm. Figure 11 depicts the improvements in composite rule coverage by the generated queries for the four databases. With SQLancer, which conducts random query generation, we observed that the increase in composite rule coverage tends to plateau after generating 60",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 66348,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Rule coverage-guided query generation (SQLancer+ rule coverage) achieves the highest rule coverage with the fewest number of queries.",
+      "context_before": "veness of our keyword-guided query generation method. Figure 10 shows the results on rule coverage, which show similar trend with that on keyword coverage. Due to the limited support of SQL features, e.g., data types, by SQLancer, especially for DBMS such as DuckDB and TiDB, relying only on SQLancer achieves low rule coverage, as shown in Figure 10. Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax. We can observe that adding those missing features improves the rule coverage, especially for DuckDB and TiDB.",
+      "context_after": "The results indicate the effectiveness of our rule coverage-guided query generation algorithm. Figure 11 depicts the improvements in composite rule coverage by the generated queries for the four databases. With SQLancer, which conducts random query generation, we observed that the increase in composite rule coverage tends to plateau after generating 60 million data points. At this stage, MySQL, SQLite, DuckDB and OceanBase each achieved a composite rule coverage of around 70% and TiDB 50%. With the introduction of composite rule coverage guidance, all four databases were able to reach 100% co",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 66597,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "With SQLancer, which conducts random query generation, we observed that the increase in composite rule coverage tends to plateau after generating 60 million data points.",
+      "context_before": "Figure 10. Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax. We can observe that adding those missing features improves the rule coverage, especially for DuckDB and TiDB. Rule coverage-guided query generation (SQLancer+ rule coverage) achieves the highest rule coverage with the fewest number of queries. The results indicate the effectiveness of our rule coverage-guided query generation algorithm. Figure 11 depicts the improvements in composite rule coverage by the generated queries for the four databases.",
+      "context_after": "At this stage, MySQL, SQLite, DuckDB and OceanBase each achieved a composite rule coverage of around 70% and TiDB 50%. With the introduction of composite rule coverage guidance, all four databases were able to reach 100% composite rule coverage after generating 20 million queries (our Prolog implementation has a total of 19 million composite rules). The results indicate the effectiveness of our composite rule coverage-guided query generation algorithm. To verify the effectiveness of the coverage-guided query generation algorithm in assisting detecting bugs and inconsistencies in relational DB",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 66937,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M25",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Note that to conduct fair comparisons, we improved SQLancer by incorporating all keywords supported by our semantics and related generation rules in SEMCONT.",
+      "context_before": "uided query generation algorithm. To verify the effectiveness of the coverage-guided query generation algorithm in assisting detecting bugs and inconsistencies in relational DBMS, we conducted an ablation study of SEMCONT with and without coverage guidance. Table V] presents the experimental results obtained from testing four databases over a period of 6 hours. We record the number of bugs and inconsistencies detected on the four settings, i.e., SEMCONT without coverage guidance, and SEMCONT with three coverage guidance. We also report the time taken to discover the first bug or inconsistency.",
+      "context_after": "The experimental results indicate that within a 6-hour timeframe, all three coverage metrics successfully assist detecting more bugs and inconsistencies compared with random generation. Composite rule coverage is the most effective among all three coverage metrics. In terms of the time taken to detect the first bug or inconsistency, all three coverage guidance algorithm are faster than SEMCONT with 16 (a) MySQL (b) TiDB (c) SQLite (d) DuckDB (e) OceanBase (f) PostgreSQL Fig. 9: The keyword coverage increment (y-axis) with the number of queries (x-axis) (a) MySQL (b) TiDB (c) SQLite (d",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 68131,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We compare SEMCONT with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS.",
+      "context_before": "5 2 7.46 7 2 8.62 SQLite 1 1 20.50 2 1 17.83 3 1 14.33 3 1 13.05 DuckDB 1 0 29.37 1 0 27.32 1 0 31.68 1 0 26.29 OceanBase 0 2 13.25 0 2 13.37 0 2 7.92 0 2 9.13 random query generation. In particular, keyword coverage, rule coverage and composite rule coverage are 5.22%, 24.19% and 21.41% faster than random query generation. Answer to RQ2: The three coverage criteria all improve the query generating process, triggering more bugs and inconsistencies with faster speed. Composite rule coverage achieves the most significant improvement. RQ3: How does SEMCONT perform compared to baseline approaches?",
+      "context_after": "For both approaches, we adopt SQLancer [14] and SQLRight [34] for query generation. Notably, SQLancer does not support the NoREC oracle for MySQL and TiDB, while SQLRight does not support TiDB, DuckDB and OceanBase. Therefore, we excluded these specific scenarios from our experiments. We ran the comparedtools for a period of 6 hours and report the results in Table VI. The experimental results show that both SQLancer and SQLRight using the NoREC as the oracle were unable to detect new bugs or inconsistencies. The TLP oracle with SQLancer for query generation detected 4 bugs in three databases,",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 70163,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For both approaches, we adopt SQLancer [14] and SQLRight [34] for query generation.",
+      "context_before": "2 7.92 0 2 9.13 random query generation. In particular, keyword coverage, rule coverage and composite rule coverage are 5.22%, 24.19% and 21.41% faster than random query generation. Answer to RQ2: The three coverage criteria all improve the query generating process, triggering more bugs and inconsistencies with faster speed. Composite rule coverage achieves the most significant improvement. RQ3: How does SEMCONT perform compared to baseline approaches? We compare SEMCONT with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS.",
+      "context_after": "Notably, SQLancer does not support the NoREC oracle for MySQL and TiDB, while SQLRight does not support TiDB, DuckDB and OceanBase. Therefore, we excluded these specific scenarios from our experiments. We ran the comparedtools for a period of 6 hours and report the results in Table VI. The experimental results show that both SQLancer and SQLRight using the NoREC as the oracle were unable to detect new bugs or inconsistencies. The TLP oracle with SQLancer for query generation detected 4 bugs in three databases, and with SQLRight for query generation detectd 1 bug in MySQL. SEMCONT outperformed",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 70306,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M28",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Notably, SQLancer does not support the NoREC oracle for MySQL and TiDB, while SQLRight does not support TiDB, DuckDB and OceanBase.",
+      "context_before": "rage and composite rule coverage are 5.22%, 24.19% and 21.41% faster than random query generation. Answer to RQ2: The three coverage criteria all improve the query generating process, triggering more bugs and inconsistencies with faster speed. Composite rule coverage achieves the most significant improvement. RQ3: How does SEMCONT perform compared to baseline approaches? We compare SEMCONT with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS. For both approaches, we adopt SQLancer [14] and SQLRight [34] for query generation.",
+      "context_after": "Therefore, we excluded these specific scenarios from our experiments. We ran the comparedtools for a period of 6 hours and report the results in Table VI. The experimental results show that both SQLancer and SQLRight using the NoREC as the oracle were unable to detect new bugs or inconsistencies. The TLP oracle with SQLancer for query generation detected 4 bugs in three databases, and with SQLRight for query generation detectd 1 bug in MySQL. SEMCONT outperformed the compared approaches and detected 16 bugs and 9 inconsistencies in the four databases. The reason is that existing approaches do",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 70390,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The experimental results show that both SQLancer and SQLRight using the NoREC as the oracle were unable to detect new bugs or inconsistencies.",
+      "context_before": "ignificant improvement. RQ3: How does SEMCONT perform compared to baseline approaches? We compare SEMCONT with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS. For both approaches, we adopt SQLancer [14] and SQLRight [34] for query generation. Notably, SQLancer does not support the NoREC oracle for MySQL and TiDB, while SQLRight does not support TiDB, DuckDB and OceanBase. Therefore, we excluded these specific scenarios from our experiments. We ran the comparedtools for a period of 6 hours and report the results in Table VI.",
+      "context_after": "The TLP oracle with SQLancer for query generation detected 4 bugs in three databases, and with SQLRight for query generation detectd 1 bug in MySQL. SEMCONT outperformed the compared approaches and detected 16 bugs and 9 inconsistencies in the four databases. The reason is that existing approaches do not consider the SQL specification and thus fail to find bugs that violated the SQL specification. For instance, One bug (109842) we detected in MySQL is related to the MOD function. When applied to negative numbers, MySQL incorrectly represents the result as-0. Both TLP and NoREC failed to detec",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 70677,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The TLP oracle with SQLancer for query generation detected 4 bugs in three databases, and with SQLRight for query generation detectd 1 bug in MySQL.",
+      "context_before": "TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS. For both approaches, we adopt SQLancer [14] and SQLRight [34] for query generation. Notably, SQLancer does not support the NoREC oracle for MySQL and TiDB, while SQLRight does not support TiDB, DuckDB and OceanBase. Therefore, we excluded these specific scenarios from our experiments. We ran the comparedtools for a period of 6 hours and report the results in Table VI. The experimental results show that both SQLancer and SQLRight using the NoREC as the oracle were unable to detect new bugs or inconsistencies.",
+      "context_after": "SEMCONT outperformed the compared approaches and detected 16 bugs and 9 inconsistencies in the four databases. The reason is that existing approaches do not consider the SQL specification and thus fail to find bugs that violated the SQL specification. For instance, One bug (109842) we detected in MySQL is related to the MOD function. When applied to negative numbers, MySQL incorrectly represents the result as-0. Both TLP and NoREC failed to detect this bug, even after successfully generated the bug triggering query. On average, our tool finds a bug in 67 minutes using 19,381 test cases, compa",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 70820,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Both TLP and NoREC failed to detect this bug, even after successfully generated the bug triggering query.",
+      "context_before": "detect new bugs or inconsistencies. The TLP oracle with SQLancer for query generation detected 4 bugs in three databases, and with SQLRight for query generation detectd 1 bug in MySQL. SEMCONT outperformed the compared approaches and detected 16 bugs and 9 inconsistencies in the four databases. The reason is that existing approaches do not consider the SQL specification and thus fail to find bugs that violated the SQL specification. For instance, One bug (109842) we detected in MySQL is related to the MOD function. When applied to negative numbers, MySQL incorrectly represents the result as-0.",
+      "context_after": "On average, our tool finds a bug in 67 minutes using 19,381 test cases, compared to 300 minutes and 1.3 million test cases for SQLancer, and 30 hours and 8.4 million test cases for SQLRight. Our approach finds more bugs/inconsistencies 17 TABLE VI: The bugs and inconsistencies detected by SQLancer, SQLRight and SEMCONT in 6h DBMSTLP (SQLancer) NoREC (SQLancer) TLP (SQLRight) NoREC (SQLRight) SEMCONT Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies MySQL 1 0 - - 1 0 0 0 5 6 TiDB 2 0 - - - - - - 7 2 SQLite 0 1 0 0 0 0 0 0 3 1 DuckDB 1 0 0",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 71385,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3 million test cases for SQLancer, and 30 hours and 8.",
+      "context_before": "compared approaches and detected 16 bugs and 9 inconsistencies in the four databases. The reason is that existing approaches do not consider the SQL specification and thus fail to find bugs that violated the SQL specification. For instance, One bug (109842) we detected in MySQL is related to the MOD function. When applied to negative numbers, MySQL incorrectly represents the result as-0. Both TLP and NoREC failed to detect this bug, even after successfully generated the bug triggering query. On average, our tool finds a bug in 67 minutes using 19,381 test cases, compared to 300 minutes and 1.",
+      "context_after": "4 million test cases for SQLRight. Our approach finds more bugs/inconsistencies 17 TABLE VI: The bugs and inconsistencies detected by SQLancer, SQLRight and SEMCONT in 6h DBMSTLP (SQLancer) NoREC (SQLancer) TLP (SQLRight) NoREC (SQLRight) SEMCONT Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies MySQL 1 0 - - 1 0 0 0 5 6 TiDB 2 0 - - - - - - 7 2 SQLite 0 1 0 0 0 0 0 0 3 1 DuckDB 1 0 0 0 - - - - 1 0 OceanBase 0 1 0 0 - - - - 0 2 with fewer test cases, demonstrating its effectiveness and efficiency in detecting bugs and inconsistencies that",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 71594,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M33",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Our approach finds more bugs/inconsistencies 17 TABLE VI: The bugs and inconsistencies detected by SQLancer, SQLRight and SEMCONT in 6h DBMSTLP (SQLancer) NoREC (SQLancer) TLP (SQLRight) NoREC (SQLRight) SEMCONT Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies MySQL 1 0 - - 1 0 0 0 5 6 TiDB 2 0 - - - - - - 7 2 SQLite 0 1 0 0 0 0 0 0 3 1 DuckDB 1 0 0 0 - - - - 1 0 OceanBase 0 1 0 0 - - - - 0 2 with fewer test cases, demonstrating its effectiveness and efficiency in detecting bugs and inconsistencies that violating SQL specifications.",
+      "context_before": "he reason is that existing approaches do not consider the SQL specification and thus fail to find bugs that violated the SQL specification. For instance, One bug (109842) we detected in MySQL is related to the MOD function. When applied to negative numbers, MySQL incorrectly represents the result as-0. Both TLP and NoREC failed to detect this bug, even after successfully generated the bug triggering query. On average, our tool finds a bug in 67 minutes using 19,381 test cases, compared to 300 minutes and 1.3 million test cases for SQLancer, and 30 hours and 8.4 million test cases for SQLRight.",
+      "context_after": "(a) MySQL (b) TiDB (c) SQLite (d) DuckDB (e) OceanBase (f) PostgreSQL Fig. 12: The memory consumption of SEMCONT with SQLancer. Memory Consumption. Figure 12 and Figure 13 shows the memory usage of SEMCONT and SQLancer during a 6-hour test on six databases. Memory usage is mainly influenced by query generation, execution, and result comparison, with query generation being the most memory-intensive. We report stabilized memory consumption, with detailed time-based changes provided in our technical report [35]. Results indicate that SEMCONT’s memory usage is comparable to the baseline. Memor",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 71682,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "12: The memory consumption of SEMCONT with SQLancer.",
+      "context_before": "nconsistencies detected by SQLancer, SQLRight and SEMCONT in 6h DBMSTLP (SQLancer) NoREC (SQLancer) TLP (SQLRight) NoREC (SQLRight) SEMCONT Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies MySQL 1 0 - - 1 0 0 0 5 6 TiDB 2 0 - - - - - - 7 2 SQLite 0 1 0 0 0 0 0 0 3 1 DuckDB 1 0 0 0 - - - - 1 0 OceanBase 0 1 0 0 - - - - 0 2 with fewer test cases, demonstrating its effectiveness and efficiency in detecting bugs and inconsistencies that violating SQL specifications. (a) MySQL (b) TiDB (c) SQLite (d) DuckDB (e) OceanBase (f) PostgreSQL Fig.",
+      "context_after": "Memory Consumption. Figure 12 and Figure 13 shows the memory usage of SEMCONT and SQLancer during a 6-hour test on six databases. Memory usage is mainly influenced by query generation, execution, and result comparison, with query generation being the most memory-intensive. We report stabilized memory consumption, with detailed time-based changes provided in our technical report [35]. Results indicate that SEMCONT’s memory usage is comparable to the baseline. Memory consumption is low because we use small tables (a few hundred records). We do not test concurrent query execution, resulting in s",
+      "section": "B Experiment results",
+      "page": 17,
+      "char_offset": 72355,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M35",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Figure 12 and Figure 13 shows the memory usage of SEMCONT and SQLancer during a 6-hour test on six databases.",
+      "context_before": "SQLancer) NoREC (SQLancer) TLP (SQLRight) NoREC (SQLRight) SEMCONT Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies MySQL 1 0 - - 1 0 0 0 5 6 TiDB 2 0 - - - - - - 7 2 SQLite 0 1 0 0 0 0 0 0 3 1 DuckDB 1 0 0 0 - - - - 1 0 OceanBase 0 1 0 0 - - - - 0 2 with fewer test cases, demonstrating its effectiveness and efficiency in detecting bugs and inconsistencies that violating SQL specifications. (a) MySQL (b) TiDB (c) SQLite (d) DuckDB (e) OceanBase (f) PostgreSQL Fig. 12: The memory consumption of SEMCONT with SQLancer. Memory Consumption.",
+      "context_after": "Memory usage is mainly influenced by query generation, execution, and result comparison, with query generation being the most memory-intensive. We report stabilized memory consumption, with detailed time-based changes provided in our technical report [35]. Results indicate that SEMCONT’s memory usage is comparable to the baseline. Memory consumption is low because we use small tables (a few hundred records). We do not test concurrent query execution, resulting in stable memory usage. Our goal is to detect compliance bugs in RDBMSs, not to evaluate performance. SEMCONT is designed for offline",
+      "section": "B Experiment results",
+      "page": 17,
+      "char_offset": 72428,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M36",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "13: Memory consumption of SEMCONT and SQLancer 1SELECT MOD( ’ −12 ’ ,-4); 2−−e x p e c t e d: 0 \", a c t u a l :−0 % Fig.",
+      "context_before": "t concurrent query execution, resulting in stable memory usage. Our goal is to detect compliance bugs in RDBMSs, not to evaluate performance. SEMCONT is designed for offline testing during RDBMS development, and its overhead is manageable. SEMCONT shows very low memory consumption and it requires far less test cases to detect a compliance issue, demonstrating its efficiency. Answer to RQ3: SEMCONT detected more bugs and inconsistencies than the compared baseline approaches. Baseline approaches fail to detect those bugs since they do not refer to SQL specification in their testing process. Fig.",
+      "context_after": "14: A bug in MySQL 8.0.29 RQ4: Case study. SEMCONT has identified 19 bugs and 13 inconsistencies, which arise from two reasons, i.e., (1) DBMS implementations are not consistent with SQL specification and (2) unclear or missing description in the SQL specification. These problems have resulted in variations in the specific implementations across different RDBMSs, consequently affecting the user experience. A bug due to missing specifications. Figure 14 is a bug we detected in MySQL 8.0.29. The query conducts MOD function with the string type as the first parameter. The expected result of the",
+      "section": "B Experiment results",
+      "page": 17,
+      "char_offset": 73564,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": [],
+      "text_is_letter_spaced": true
+    },
+    {
+      "id": "M37",
+      "found_by": "citation_marker",
+      "surface": "[40]–[42]",
+      "technique": "norec",
+      "sentence": "Numerous testing methods involve constructing a test oracle by proposing a variety of metamorphic relations [21], [28], [40]–[42].",
+      "context_before": "rammar-based mutations. Furthermore, it boosts the accuracy of feedback through the use of binary coverage linking and bijective block mapping, thereby enhancing the overall performance of RDBMS testing. SparkFuzz [28] introduces a fuzzing-based method that utilizes the query results from a reference database as test oracles. The effectiveness of these differential testing methods is limited by the shared functionalities and syntax supported across the databases under test. Moreover, they may yield false positives due to the varied implementation choices inherent in different database systems.",
+      "context_after": "MUTASQL [21] and Eqsql [17] are tools that construct test cases by defining mutation rules. These rules are used to generate or synthesize SQL query statements that are functionally equivalent to the original ones. In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42]. PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data. The design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 83989,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M38",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42].",
+      "context_before": "of these differential testing methods is limited by the shared functionalities and syntax supported across the databases under test. Moreover, they may yield false positives due to the varied implementation choices inherent in different database systems. Numerous testing methods involve constructing a test oracle by proposing a variety of metamorphic relations [21], [28], [40]–[42]. MUTASQL [21] and Eqsql [17] are tools that construct test cases by defining mutation rules. These rules are used to generate or synthesize SQL query statements that are functionally equivalent to the original ones.",
+      "context_after": "PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data. The design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by verifying whether the returned result includes the specific row of data. NoREC [40] transforms an optimized SQL query into an equivalent non-optimized version and then compares the execution results of both. TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query. If the results of executing the original q",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 84335,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M39",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data.",
+      "context_before": "sitives due to the varied implementation choices inherent in different database systems. Numerous testing methods involve constructing a test oracle by proposing a variety of metamorphic relations [21], [28], [40]–[42]. MUTASQL [21] and Eqsql [17] are tools that construct test cases by defining mutation rules. These rules are used to generate or synthesize SQL query statements that are functionally equivalent to the original ones. In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42].",
+      "context_after": "The design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by verifying whether the returned result includes the specific row of data. NoREC [40] transforms an optimized SQL query into an equivalent non-optimized version and then compares the execution results of both. TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query. If the results of executing the original query differ from those obtained from the three divided queries, it likely indicates the presence of logical err",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 84501,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M40",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [40] transforms an optimized SQL query into an equivalent non-optimized version and then compares the execution results of both.",
+      "context_before": "ning mutation rules. These rules are used to generate or synthesize SQL query statements that are functionally equivalent to the original ones. In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42]. PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data. The design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by verifying whether the returned result includes the specific row of data.",
+      "context_after": "TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query. If the results of executing the original query differ from those obtained from the three divided queries, it likely indicates the presence of logical errors. SQLRight [34] focuses on enhancing the semantic correctness of generated SQL queries and adopts the oracles proposed by PQS [42], NoREC [40], and TLP [41]. GRIFFIN [27] executes mutation testing within the grammatical boundaries of SQL language, transforming data into metadata for this purpose. While metamorp",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 84792,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M41",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query.",
+      "context_before": "al ones. In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42]. PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data. The design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by verifying whether the returned result includes the specific row of data. NoREC [40] transforms an optimized SQL query into an equivalent non-optimized version and then compares the execution results of both.",
+      "context_after": "If the results of executing the original query differ from those obtained from the three divided queries, it likely indicates the presence of logical errors. SQLRight [34] focuses on enhancing the semantic correctness of generated SQL queries and adopts the oracles proposed by PQS [42], NoREC [40], and TLP [41]. GRIFFIN [27] executes mutation testing within the grammatical boundaries of SQL language, transforming data into metadata for this purpose. While metamorphic testing approaches address syntax differences arising from various database implementations, consistently returned results do n",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 84927,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M42",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "SQLRight [34] focuses on enhancing the semantic correctness of generated SQL queries and adopts the oracles proposed by PQS [42], NoREC [40], and TLP [41].",
+      "context_before": "design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by verifying whether the returned result includes the specific row of data. NoREC [40] transforms an optimized SQL query into an equivalent non-optimized version and then compares the execution results of both. TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query. If the results of executing the original query differ from those obtained from the three divided queries, it likely indicates the presence of logical errors.",
+      "context_after": "GRIFFIN [27] executes mutation testing within the grammatical boundaries of SQL language, transforming data into metadata for this purpose. While metamorphic testing approaches address syntax differences arising from various database implementations, consistently returned results do not always guarantee the absence of bugs. Furthermore, these methods fall short in detecting bugs caused by violations of SQL specifications. Formal semantics for SQL. There have been several approaches [18], [24], [25], [36], [38], [48], [49] that made attempts to formalize the semantics of SQL. Chinaei [23] was",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 85216,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "uses_infrastructure": [
+        "M12"
+      ],
+      "compares_with": [
+        "M9",
+        "M26"
+      ],
+      "describes_as_state_of_the_art": [
+        "M4",
+        "M9",
+        "M26",
+        "M38"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-10T15:11:01Z",
+    "is_model_written": true,
+    "summary": "SEMCONT tests whether a relational DBMS conforms to the SQL specification, rather than whether two systems or two queries agree. It derives expected behaviour from the standard and reports where an implementation departs from it, which reaches under-specified corners that differential and metamorphic oracles pass over because both sides of their comparison are equally wrong.",
+    "narrative": "SQLancer is SEMCONT's generator, extended. The paper states it enhances the syntax-guided generation method, SQLancer, with coverage-guided test case generation, and uses it to produce the seed pool its algorithm works from. It then adds what SQLancer does not emit -- keywords and generation rules the tool lacked for particular systems -- reporting the results as SQLancer+ keyword syntax and SQLancer+ rule syntax, each measurably improving coverage. SQLancer is also the reference point for the comparison: TLP and NoREC are the state-of-the-art metamorphic methods it is measured against, run in both their SQLancer and SQLRight implementations.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "reuse_component",
+      "M4": "state_of_the_art",
+      "M5": "definition",
+      "M6": "definition",
+      "M7": "reuse_component",
+      "M8": "background",
+      "M9": "state_of_the_art",
+      "M10": "definition",
+      "M11": "definition",
+      "M12": "definition",
+      "M13": "definition",
+      "M14": "baseline",
+      "M15": "motivation",
+      "M16": "motivation",
+      "M17": "result_comparison",
+      "M18": "reuse_component",
+      "M19": "result_comparison",
+      "M20": "result_comparison",
+      "M21": "motivation",
+      "M22": "reuse_component",
+      "M23": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M3",
+          "M7",
+          "M22"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "Here, we 10 enhance the syntax-guided generation method, SQLancer [14], with coverage-guided test case generation.",
+            "section": "A Overview of our approach",
+            "page": 9
+          },
+          {
+            "mention_id": "M7",
+            "sentence": "We adopt SQLancer to randomly generatea large number of SQL statements (line 2), which serves as the seed pool of our query generation algorithm.",
+            "section": "B Coverage guided test case generation",
+            "page": 11
+          },
+          {
+            "mention_id": "M22",
+            "sentence": "Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax.",
+            "section": "B Experiment results",
+            "page": 15
+          }
+        ],
+        "reasoning": "M3 states the approach enhances SQLancer with coverage-guided generation, M7 that SQLancer is adopted to generate the seed pool, and M22 that missing features were added to SQLancer for particular systems. The generation is SQLancer's; the conformance oracle is the paper's own.",
+        "reuse_kind": "generator"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "What is extended is SQLancer's generation -- keywords, rules, and coverage guidance -- not any of its oracles. The oracle here is conformance to the SQL specification."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M9",
+          "M14",
+          "M17"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M9",
+            "sentence": "We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+            "section": "A Experiment setup",
+            "page": 14
+          },
+          {
+            "mention_id": "M14",
+            "sentence": ", NoREC (SQLancer), NoREC (SQLRight), TLP (SQLancer) and TLP (SQLRight).",
+            "section": "A Experiment setup",
+            "page": 14
+          },
+          {
+            "mention_id": "M17",
+            "sentence": "It is noteworthy that all four databases have been extensively tested by existing methods [40]–[42], yet SEMCONTis still able to detect bugs that were not detected by those approaches.",
+            "section": "B Experiment results",
+            "page": 14
+          }
+        ],
+        "reasoning": "M9 states SEMCONT is compared with TLP and NoREC, M14 records both being run in their SQLancer and SQLRight implementations, and M17 reports SEMCONT finding bugs those approaches did not.",
+        "techniques": [
+          "tlp",
+          "norec"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M9"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "Coverage guided test case generation The state-of-the-art practice in test case generation involves randomly generating SQL queries guided by the syntax of SQL, among which SQLancer [14] stands out as one of the most effective tools of this kind.",
+            "section": "B Coverage guided test case generation",
+            "page": 10
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+            "section": "A Experiment setup",
+            "page": 14
+          }
+        ],
+        "reasoning": "M4 calls SQLancer one of the most effective tools of its kind and the state-of-the-art practice in test case generation; M9 calls TLP and NoREC state-of-the-art metamorphic testing methods."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2407_04294.json b/_data/papers/paper_arxiv_2407_04294.json
new file mode 100644
index 0000000..7e0248f
--- /dev/null
+++ b/_data/papers/paper_arxiv_2407_04294.json
@@ -0,0 +1,1366 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:05:33Z",
+  "paper": {
+    "id": "paper:arxiv:2407.04294",
+    "title": "SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing",
+    "authors": [
+      "Jin Wei",
+      "Ping Chen",
+      "Kangjie Lu",
+      "Jun Dai",
+      "Xiaoyan Sun"
+    ],
+    "year": 2024,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2407.04294",
+    "s2_paper_id": "bef377046828fc5f81ae8988d7b500c9442b57d2",
+    "url": "https://arxiv.org/abs/2407.04294",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2407.04294",
+      "retrieved_at": "2026-09-10T15:05:33Z",
+      "chars": 84025,
+      "content_sha256": "sha256:ec2e1d52bd7adfd51eec896ea0182261936dec917b41b9cf6734e1b4633a31b7"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 25,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2717
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 12926
+      },
+      {
+        "number": "2.1",
+        "title": "Logic Bugs and Threat Model",
+        "start": 12940
+      },
+      {
+        "number": "2.2",
+        "title": "Logic Bug Testing Oracles and Differential Testing",
+        "start": 15417
+      },
+      {
+        "number": "2.3",
+        "title": "Coverage-guided Fuzzing and Directed Fuzzing",
+        "start": 18375
+      },
+      {
+        "number": "3",
+        "title": "Analysis of Existing Logic Bug Patterns",
+        "start": 20905
+      },
+      {
+        "number": "4",
+        "title": "SQLaser",
+        "start": 26096
+      },
+      {
+        "number": "4.1",
+        "title": "Logic-Bug Pattern Modeling",
+        "start": 31622
+      },
+      {
+        "number": "4.2",
+        "title": "Call Chain Distance Calculation",
+        "start": 38671
+      },
+      {
+        "number": "4.3",
+        "title": "Mutation based on Seed Energy and Testing Oracles",
+        "start": 43178
+      },
+      {
+        "number": "4.3.1",
+        "title": "Mutation based on Seed Energy",
+        "start": 43427
+      },
+      {
+        "number": "4.3.2",
+        "title": "Testing Oracles",
+        "start": 44095
+      },
+      {
+        "number": "5",
+        "title": "Implementation",
+        "start": 45399
+      },
+      {
+        "number": "6",
+        "title": "Evaluation",
+        "start": 46922
+      },
+      {
+        "number": "20.04.5",
+        "title": "LTS operating system.",
+        "start": 47678
+      },
+      {
+        "number": "6.1",
+        "title": "Finding Bugs in DBMSs",
+        "start": 49159
+      },
+      {
+        "number": "6.2",
+        "title": "Comparison with Existing Tools",
+        "start": 56614
+      },
+      {
+        "number": "6.2.1",
+        "title": "Performance",
+        "start": 56964
+      },
+      {
+        "number": "6.2.2",
+        "title": "Code Coverage",
+        "start": 59193
+      },
+      {
+        "number": "6.2.3",
+        "title": "Seed Distance Distribution",
+        "start": 62213
+      },
+      {
+        "number": "6.3",
+        "title": "Trimmed Call Chains vs. Complete Call Chains",
+        "start": 63631
+      },
+      {
+        "number": "7",
+        "title": "Discussion",
+        "start": 65923
+      },
+      {
+        "number": "7.1",
+        "title": "Completeness of Testing Oracles",
+        "start": 66105
+      },
+      {
+        "number": "7.2",
+        "title": "Completeness of SQL Clause Collections",
+        "start": 67166
+      },
+      {
+        "number": "8",
+        "title": "Related Work",
+        "start": 67948
+      },
+      {
+        "number": "9",
+        "title": "Conclusion",
+        "start": 70130
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "K.S. Abdul and S. Khurshid, Automated SQL query generation for systematic testing of database engines, in: Proceedings of the 25th IEEE/ACM International Conference on Automated Software Engineering, 2010, pp. 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "C. Aschermann, S. Schumilo, T. Blazytko, R. Gawlik and T. Holz, REDQUEEN: Fuzzing with Input-to-State Correspondence., in: Proceedings of the 26th Annual Network and Distributed System Security Symposium (NDSS), V ol. 19, 2019, pp. 1–15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "T. Blazytko, M. Bishop, C. Aschermann, J. Cappos, M. Schlögel, N. Korshun, A. Abbasi, M. Schweighauser, S. Schinzel, S. Schumilo et al., {GRIMOIRE }: Synthesizing structure while fuzzing, in: Proceedings of the 28th USENIX Security Symposium (USENIX Security 19), 2019, pp. 1985–2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "M. Böhme, V .-T. Pham, M.-D. Nguyen and A. Roychoudhury, Directed greybox fuzzing, in: Proceedings of the 2017 ACMSIGSAC conference on computer and communications security, 2017, pp. 2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "H. Chen, Y. Xue, Y. Li, B. Chen, X. Xie, X. Wu and Y. Liu, Hawkeye: Towards a desired directed grey-box fuzzer, in: Proceedings of the 2018 ACMSIGSAC conference on computer and communications security, 2018, pp. 2095–2108. Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing 23",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "P. Chen and H. Chen, Angora: Efficient fuzzing by principled search, in: Proceedings of the Security and Privacy (SP), IEEE, 2018, pp. 711–725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Y. Chen, R. Zhong, H. Hu, H. Zhang, Y. Yang, D. Wu and W. Lee, One engine to fuzz’em all: Generic language processor testing with semantic validation, in: Proceedings of the pp. 642–658.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "J. Choi, K. Kim, D. Lee and S.K. Cha, NTFuzz: Enabling type-aware kernel fuzzing on windows with static binary analysis, in: Proceedings of the. 677–693.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Z.Y. Ding and C.L. Goues, An empirical study of oss-fuzz bugs, in: Proceedings of the 2021 IEEE/ACM 18th International Conference on Mining Software Repositories (MSR), IEEE, 2021, pp. 131–142.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "S.T. Dinh, H. Cho, K. Martin, A. Oest, K. Zeng, A. Kapravelos, G.-J. Ahn, T. Bao, R. Wang, A. Doupé et al., Favocado: Fuzzing the Binding Code of JavaScript Engines Using Semantically Correct Test Cases., in: Proceedings of the 28th Annual Network and Distributed System Security Symposium (NDSS), 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "X. Du, B. Chen, Y. Li, J. Guo, Y. Zhou, Y. Liu and Y. Jiang, Leopard: Identifying vulnerable code for vulnerability assessment through program metrics, in: Proceedings of the Engineering (ICSE), IEEE, 2019, pp. 60–71.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Z. Du, Y. Li, Y. Liu and B. Mao, WindRanger: a directed greybox fuzzer driven by deviation basic blocks, in: Proceedings of the 44th International Conference on Software Engineering, 2022, pp. 2440–2451.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "J.K. Fichte, M. Hecher, P. Thier and S. Woltran, Exploiting database management systems and treewidth for counting, Theory and Practice of Logic Programming 22(1) (2022), 128–157.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "P. Foundation, TiDB, 2023. https://www.pingcap.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "S. Gan, C. Zhang, P. Chen, B. Zhao, X. Qin, D. Wu and Z. Chen, GREYONE: Data flow sensitive fuzzing, in: Proceedings of the 29th USENIX security symposium (USENIX Security 20), 2020, pp. 2577–2594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "B. Ghit, N. Poggi, J. Rosen, R. Xin and P. Boncz, SparkFuzz: Searching correctness regressions in modern query engines, in:Proceedings of the workshop on Testing Database Systems, 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Google, ClusterFuzz, 2023. https://google.github.io/clusterfuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Google, Honggfuzz, 2023. https://google.github.io/honggfuzz/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "P.D. Group, PostgreSQL, 2023. https://www.postgresql.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "H. Han, D. Oh and S.K. Cha, CodeAlchemist: Semantics-Aware Code Generation to Find Vulnerabilities in JavaScript Engines., in: Proceedings of the 26th Network and Distributed System Security Symposium (NDSS), 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Intel, PIN, 2023. https://www.intel.cn/content/www/cn/zh/developer/articles/tool/pin-a-dynamic-binary-instrumentation-tool. html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "B. Jiang, Y. Liu and W.K. Chan, Contractfuzzer: Fuzzing smart contracts for vulnerability detection, in: Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering, 2018, pp. 259–269.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim and W. Kang, Apollo: Automatic detection and diagnosis of performance regressions in database systems, Proceedings of the 46th International Conference on Very Large Data Bases (VLDB) 13(1) (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "K. Kim, D.R. Jeong, C.H. Kim, Y. Jang, I. Shin and B. Lee, HFL: Hybrid Fuzzing on the Linux Kernel., in: Proceedings of the 27th Annual Network and Distributed System Security Symposium (NDSS), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "G. Lee, W. Shim and B. Lee, Constraint-guided directed greybox fuzzing, in: Proceedings of the 30th USENIX Security Symposium (USENIX Security 21), 2021, pp. 3559–3576.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "C. Lemieux and K. Sen, Fairfuzz: A targeted mutation strategy for increasing greybox fuzz testing coverage, in: Proceedings of the 33rd ACM/IEEE international conference on automated software engineering, 2018, pp. 475–485.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Y. Li, B. Chen, M. Chandramohan, S.-W. Lin, Y. Liu and A. Tiu, Steelix: program-state based binary fuzzing, in: Proceedings of the 2017 11th joint meeting on foundations of software engineering, 2017, pp. 627–637.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Y. Liang, S. Liu and H. Hu, Detecting Logical Bugs of {DBMS }with Coverage-based Guidance, in: Proceedings of the 31st USENIX Security Symposium (USENIX Security 22), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "X. Liu, Q. Zhou, J. Arulraj and A. Orso, Automated performance bug detection in database systems, arXiv preprint arXiv:2105.10016 (2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "LLVM, LibFuzzer: A Library For Coverage-guided Fuzz Testing, 2023. http://llvm.org/docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "E. Lo, C. Binnig, D. Kossmann, M. Tamer Özsu and W.-K. Hon, A framework for testing DBMS features, The VLDB Journal 19(2010), 203–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "W.M. McKeeman, Differential testing for software, Digital Technical Journal 10(1) (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "MySQL, MySQL Customers, 2023. https://www.mysql.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "T.D. Nguyen, L.H. Pham, J. Sun, Y. Lin and Q.T. Minh, sfuzz: An efficient adaptive fuzzer for solidity smart contracts, in:Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering, 2020, pp. 778–788.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "S. Österlund, K. Razavi, H. Bos and C. Giuffrida, {ParmeSan }: Sanitizer-guided greybox fuzzing, in: Proceedings of the 29th USENIX Security Symposium (USENIX Security 20), 2020, pp. 2289–2306. 24 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "S. Pailoor, A. Aday and S. Jana, {MoonShine }: Optimizing {OS}fuzzer seed selection with trace distillation, in: Proceedings of the 27th USENIX Security Symposium (USENIX Security 18), 2018, pp. 729–743.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "S. Park, W. Xu, I. Yun, D. Jang and T. Kim, Fuzzing javascript engines with aspect-preserving mutation, in: Proceedings of the. 1629–1642.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "M. Rigger and Z. Su, Detecting optimization bugs in database engines via non-optimizing reference engine construction, in:Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "M. Rigger and Z. Su, Finding bugs in database systems via query partitioning, in: Proceedings of the ACM on Programming Languages, V ol. 4, ACM New York, NY, USA, 2020, pp. 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, Testing database engines via pivoted query synthesis, in: Proceedings of 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "K. Serebryany, Sanitize, fuzz, and harden your C++ code, San Francisco, CA (2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "D.R. Slutz, Massive stochastic testing of SQL, in: VLDB, V ol. 98, Citeseer, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "SQLite, SQLite, 2023. https://sqlite.org/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "SQLite, SQLite bug fix, 2023. https://www.sqlite.org/src/info/5351e920f489562f.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "N. Stephens, J. Grosen, C. Salls, A. Dutcher, R. Wang, J. Corbetta, Y. Shoshitaishvili, C. Kruegel and G. Vigna, Driller: Augmenting fuzzing through selective symbolic execution., in: NDSS, V ol. 16, 2016, pp. 1–16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang and Y. Jiang, Industry practice of coverage-guided enterpriselevel DBMS fuzzing, in: Proceedings of the Software Engineering in Practice (ICSE-SEIP), IEEE, 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "wikipedia, ASCII, 2023. https://zh.wikipedia.org/wiki/ASCII.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "wikipedia, coredump, 2023. https://en.wikipedia.org/wiki/Core_dump.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "V. Wüstholz and M. Christakis, Harvey: A greybox fuzzer for smart contracts, in: Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1398–1409.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "M. Xu, S. Kashyap, H. Zhao and T. Kim, Krace: Data race fuzzing for kernel file systems, in: Proceedings of the. 1643–1660.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "P. Xu, Y. Wang, H. Hu and P. Su, COOPER: Testing the Binding Code of Scripting Languages with Cooperative Mutation., in:Proceedings of the 28th Annual Network and Distributed System Security Symposium (NDSS), 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "W. Xu, S. Park and T. Kim, Freedom: Engineering a state-of-the-art dom fuzzer, in: Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, 2020, pp. 971–986.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "W. Xu, H. Moon, S. Kashyap, P.-N. Tseng and T. Kim, Fuzzing file systems via two-dimensional input space exploration, in:Proceedings of the. 818–834.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "I. Yun, S. Lee, M. Xu, Y. Jang and T. Kim, QSYM: A practical concolic execution engine tailored for hybrid fuzzing, in: Proceedings of the 27th USENIX Security Symposium (USENIX Security 18), 2018, pp. 745–761.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "M. Zalewski, American Fuzzy Lop, 2023. https://github.com/google/AFL.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee and D. Wu, Squirrel: Testing database management systems with language validity and coverage feedback, in: Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "P. Zong, T. Lv, D. Wang, Z. Deng, R. Liang and K. Chen, {FuzzGuard }: Filtering out unreachable inputs in directed grey-box fuzzing through deep learning, in: Proceedings of the 29th USENIX security symposium (USENIX security 20), 2020, pp. 2255–2269.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Bug CVE-2012-2081. https://nvd.nist.gov/vuln/detail/CVE-2012-2081.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Bug CVE-2012-2081. https://nvd.nist.gov/vuln/detail/CVE-2014-4987.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Bugs Found in Database Management Systems. https://www.manuelrigger.at/dbms-bugs/.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 61,
+      "text": "LLVM pass. https://llvm.org/docs/WritingAnLLVMPass.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "MySQL bug 95908. https://bugs.mysql.com/bug.php?id=95908.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "MySQL bug 95926. https://bugs.mysql.com/bug.php?id=95926.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "MySQL bug 95927. https://bugs.mysql.com/bug.php?id=95927.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "MySQL bug 95937. https://bugs.mysql.com/bug.php?id=95937.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "MySQL bug 95954. https://bugs.mysql.com/bug.php?id=95954.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "MySQL bug 95975. https://bugs.mysql.com/bug.php?id=95975.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "MySQL bug 95983. https://bugs.mysql.com/bug.php?id=95983.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "MySQL bug 96012. https://bugs.mysql.com/bug.php?id=96012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "MySQL bug 99122. https://bugs.mysql.com/bug.php?id=99122.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "SQLancer. https://github.com/sqlancer/sqlancer.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 72,
+      "text": "SQLite bug 16252d7. https://www.sqlite.org/src/info/16252d7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "SQLite bug 1685610e. https://www.sqlite.org/src/info/1685610e. Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing 25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "SQLite bug 1b1dd4d4. https://www.sqlite.org/src/info/1b1dd4d4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "SQLite bug 2363a14. https://www.sqlite.org/src/info/2363a14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "SQLite bug 5351e920. https://www.sqlite.org/src/info/5351e920.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "SQLite bug 54110870. https://www.sqlite.org/src/info/54110870.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "SQLite bug 5c6146b5. https://www.sqlite.org/src/info/5c6146b5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "SQLite bug 659c551d. https://www.sqlite.org/src/info/659c551d.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "SQLite bug 6ac0f822. https://www.sqlite.org/src/info/6ac0f822.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "SQLite bug 86fa0087. https://www.sqlite.org/src/info/86fa0087.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "SQLite bug 9c8c1092. https://www.sqlite.org/src/info/9c8c1092.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "SQLite bug c0c90961. https://www.sqlite.org/src/info/c0c90961.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "SQLite bug db9acef1. https://www.sqlite.org/src/info/db9acef1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "SQLite bug eb40248. https://www.sqlite.org/src/info/eb40248.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "SQLite bug ebe4845c. https://www.sqlite.org/src/info/ebe4845c.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "SQLite bug f65c929. https://www.sqlite.org/src/info/f65c929.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "SQLite bug f898d04c. https://www.sqlite.org/src/info/f898d04c.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "SQLite bug f9c6426. https://www.sqlite.org/src/info/f9c6426.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "SQLite bug faaaae49. https://www.sqlite.org/src/info/faaaae49.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "TiDB bug 15725. https://github.com/pingcap/tidb/issues/15725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "TiDB bug 15733. https://github.com/pingcap/tidb/issues/15733.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "TiDB bug 15789. https://github.com/pingcap/tidb/issues/15789.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "TiDB bug 15846. https://github.com/pingcap/tidb/issues/15846.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "TiDB bug 15986. https://github.com/pingcap/tidb/issues/15986.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "TiDB bug 15994. https://github.com/pingcap/tidb/issues/15994.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "TiDB bug 17814. https://github.com/pingcap/tidb/issues/17814.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 38,
+      "text": "M. Rigger and Z. Su, Detecting optimization bugs in database engines via non-optimizing reference engine construction, in:Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 39,
+      "text": "M. Rigger and Z. Su, Finding bugs in database systems via query partitioning, in: Proceedings of the ACM on Programming Languages, V ol. 4, ACM New York, NY, USA, 2020, pp. 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, Testing database engines via pivoted query synthesis, in: Proceedings of 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 60,
+      "text": "Bugs Found in Database Management Systems. https://www.manuelrigger.at/dbms-bugs/.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 71,
+      "text": "SQLancer. https://github.com/sqlancer/sqlancer.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[28, 38–40]",
+      "technique": "norec",
+      "sentence": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing code and are particularly common, can cause a variety of serious issues, such as incorrect query results, exposure of sensitive data, unauthorized access, and data corruption [28, 38–40].",
+      "context_before": "ms, Logic Bug detection, Directed Fuzzing 1. Introduction Database Management Systems (DBMSs) [13] are widely used as centralized storage and management systems for data. For instance, SQLite [43] is used on around 3.5 billion smartphones worldwide. Given their critical role of DBMSs in modern data-driven applications, logic bugs [28] within DBMSs can have severe consequences. These bugs, which are due to implementation errors in the DBMSs’ *Corresponding author. E-mail: pchen@fudan.edu.cn. 0926-227X/$35.00 © 0 – IOS Press. All rights reserved.arXiv:2407.04294v1 [cs.CR] 5 Jul 2024 2 Wei et al.",
+      "context_after": "Due to the non-crashing nature of most logic bugs, detecting them can be challenging, and researchers have invested significant efforts in this area. Rigger et al. propose oracles to detect incorrectness of SQL query results. For example, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] and the Ternary Logic Partitioning (TLP) oracle [39] are two oracles employing the concept of differential testing [32]. They transform the original SQL query into an equivalent form, and if discrepancies arise between the results of the original query and the transformed query, potential l",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3275,
+      "cited_reference": {
+        "number": 38,
+        "text": "M. Rigger and Z. Su, Detecting optimization bugs in database engines via non-optimizing reference engine construction, in:Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "Non-Optimizing Reference Engine Construction",
+      "technique": "norec",
+      "sentence": "For example, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] and the Ternary Logic Partitioning (TLP) oracle [39] are two oracles employing the concept of differential testing [32].",
+      "context_before": "hen@fudan.edu.cn. 0926-227X/$35.00 © 0 – IOS Press. All rights reserved.arXiv:2407.04294v1 [cs.CR] 5 Jul 2024 2 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing code and are particularly common, can cause a variety of serious issues, such as incorrect query results, exposure of sensitive data, unauthorized access, and data corruption [28, 38–40]. Due to the non-crashing nature of most logic bugs, detecting them can be challenging, and researchers have invested significant efforts in this area. Rigger et al. propose oracles to detect incorrectness of SQL query results.",
+      "context_after": "They transform the original SQL query into an equivalent form, and if discrepancies arise between the results of the original query and the transformed query, potential logic bugs may be indicated. Pivoted Query Synthesis (PQS) oracle [40] does not employ the concept of differential testing; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row. Failure to retrieve the pivot row indicates a potential bug in the system. SQLancer [60, 71] that implements these oracles discovers logic bugs in various DBMSs. However, as SQLancer generates",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3753,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": "Pivoted Query Synthesis (PQS) oracle [40] does not employ the concept of differential testing; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row.",
+      "context_before": "ture of most logic bugs, detecting them can be challenging, and researchers have invested significant efforts in this area. Rigger et al. propose oracles to detect incorrectness of SQL query results. For example, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] and the Ternary Logic Partitioning (TLP) oracle [39] are two oracles employing the concept of differential testing [32]. They transform the original SQL query into an equivalent form, and if discrepancies arise between the results of the original query and the transformed query, potential logic bugs may be indicated.",
+      "context_after": "Failure to retrieve the pivot row indicates a potential bug in the system. SQLancer [60, 71] that implements these oracles discovers logic bugs in various DBMSs. However, as SQLancer generates SQL queries based on specific rules, it may constrain the exploration of broader code paths. Therefore, Liang et al. introduce SQLRight [28], aiming to generate high-quality query statements with validity-oriented mutations to uncover more logic bugs. Specifically, SQLRight employs coverage-guided fuzzing [6, 15, 18, 27, 30, 41, 45, 54, 55] to achieve higher code coverage in code paths. 1CREATE TABLE t0",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4154,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [60, 71] that implements these oracles discovers logic bugs in various DBMSs.",
+      "context_before": "nd the Ternary Logic Partitioning (TLP) oracle [39] are two oracles employing the concept of differential testing [32]. They transform the original SQL query into an equivalent form, and if discrepancies arise between the results of the original query and the transformed query, potential logic bugs may be indicated. Pivoted Query Synthesis (PQS) oracle [40] does not employ the concept of differential testing; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row. Failure to retrieve the pivot row indicates a potential bug in the system.",
+      "context_after": "However, as SQLancer generates SQL queries based on specific rules, it may constrain the exploration of broader code paths. Therefore, Liang et al. introduce SQLRight [28], aiming to generate high-quality query statements with validity-oriented mutations to uncover more logic bugs. Specifically, SQLRight employs coverage-guided fuzzing [6, 15, 18, 27, 30, 41, 45, 54, 55] to achieve higher code coverage in code paths. 1CREATE TABLE t0(c0 COLLATE NOCASE, c1); 2CREATE INDEX i0 ON t0(c0) WHERE c0 >= c1; 3INSERT INTO t0 VALUES(’a’, ’B’); 4SELECT *FROM t0 WHERE t0.c1 <= t0.c0; 5-- output: {}, expec",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4437,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "However, as SQLancer generates SQL queries based on specific rules, it may constrain the exploration of broader code paths.",
+      "context_before": "pt of differential testing [32]. They transform the original SQL query into an equivalent form, and if discrepancies arise between the results of the original query and the transformed query, potential logic bugs may be indicated. Pivoted Query Synthesis (PQS) oracle [40] does not employ the concept of differential testing; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row. Failure to retrieve the pivot row indicates a potential bug in the system. SQLancer [60, 71] that implements these oracles discovers logic bugs in various DBMSs.",
+      "context_after": "Therefore, Liang et al. introduce SQLRight [28], aiming to generate high-quality query statements with validity-oriented mutations to uncover more logic bugs. Specifically, SQLRight employs coverage-guided fuzzing [6, 15, 18, 27, 30, 41, 45, 54, 55] to achieve higher code coverage in code paths. 1CREATE TABLE t0(c0 COLLATE NOCASE, c1); 2CREATE INDEX i0 ON t0(c0) WHERE c0 >= c1; 3INSERT INTO t0 VALUES(’a’, ’B’); 4SELECT *FROM t0 WHERE t0.c1 <= t0.c0; 5-- output: {}, expected: {a|B} Code 1: An example logic bug in SQLite. Despite substantial efforts dedicated to detecting logic bugs, we find so",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4524,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "Non-Optimizing Reference Engine Construction",
+      "technique": "norec",
+      "sentence": "For instance, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] translates an optimized query containing a WHERE clause into an unoptimized version with identical semantics.",
+      "context_before": "are found among the results produced by these systems, it indicates the presence of at least one system containing a bug. However, constructing an oracle that covers all DBMS logic bugs is challenging. Additionally, due to various languages and extensions, popular DBMSs share only a small subset of functionalities, making cross-DBMS validation incapable of detecting DBMS-specific bugs [42]. Researchers, including Manuel Rigger et al., have made efforts in this direction. They build functionally equivalent SQL queries for a single DBMS and check if these queries could produce the same results.",
+      "context_after": "By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug. Ternary Logic Partitioning (TLP) oracle [39] composes several sub-queries to collectively achieve the semantics of the original query. If the composed query does not yield the same result as the original query, it suggests the possibility of a logic bug. SQLRight [28] also supplements their approach with some oracles. The INDEX oracle adds various CREATE INDEX clauses to the query set to test the impact of INDEX usage on query execution. The ROWID oracle inserts WITHOUT ROWID clause",
+      "section": "2.2 Logic Bug Testing Oracles and Differential Testing",
+      "page": 5,
+      "char_offset": 16820,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "Ternary Logic Partitioning (TLP) oracle [39] composes several sub-queries to collectively achieve the semantics of the original query.",
+      "context_before": "ctionalities, making cross-DBMS validation incapable of detecting DBMS-specific bugs [42]. Researchers, including Manuel Rigger et al., have made efforts in this direction. They build functionally equivalent SQL queries for a single DBMS and check if these queries could produce the same results. For instance, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] translates an optimized query containing a WHERE clause into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug.",
+      "context_after": "If the composed query does not yield the same result as the original query, it suggests the possibility of a logic bug. SQLRight [28] also supplements their approach with some oracles. The INDEX oracle adds various CREATE INDEX clauses to the query set to test the impact of INDEX usage on query execution. The ROWID oracle inserts WITHOUT ROWID clause to observe its effect on query execution. If the results of the queries with the inserted INDEX or WITHOUT ROWID clause differ from those without these insertions, it indicates a potential logic bug. And the LIKELY oracle checks if the original q",
+      "section": "2.2 Logic Bug Testing Oracles and Differential Testing",
+      "page": 5,
+      "char_offset": 17124,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": "It is worth noting that the Pivoted Query Synthesis (PQS) oracles [40] proposed by Manuel Rigger et al.",
+      "context_before": "ic bug. SQLRight [28] also supplements their approach with some oracles. The INDEX oracle adds various CREATE INDEX clauses to the query set to test the impact of INDEX usage on query execution. The ROWID oracle inserts WITHOUT ROWID clause to observe its effect on query execution. If the results of the queries with the inserted INDEX or WITHOUT ROWID clause differ from those without these insertions, it indicates a potential logic bug. And the LIKELY oracle checks if the original queries include the LIKELY clause. If they don’t, it adds the LIKELY clause to see how it impacts query execution.",
+      "context_after": "does not use the differential testing method; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row. If the DBMS fails to fetch the pivot row, the likely cause is a bug in the DBMS. As of now, we have not deployed PQS oracles into SQLaser. 2.3. Coverage-guided Fuzzing and Directed Fuzzing Fuzzing [9] is a software testing technique that involves providing invalid, unexpected, or random data as inputs to a computer program. The goal is to discover vulnerabilities or bugs by observing how the program reacts to these inputs [18, 30, 41, 5",
+      "section": "2.2 Logic Bug Testing Oracles and Differential Testing",
+      "page": 5,
+      "char_offset": 17972,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "As of now, we have not deployed PQS oracles into SQLaser.",
+      "context_before": "OUT ROWID clause differ from those without these insertions, it indicates a potential logic bug. And the LIKELY oracle checks if the original queries include the LIKELY clause. If they don’t, it adds the LIKELY clause to see how it impacts query execution. It is worth noting that the Pivoted Query Synthesis (PQS) oracles [40] proposed by Manuel Rigger et al. does not use the differential testing method; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row. If the DBMS fails to fetch the pivot row, the likely cause is a bug in the DBMS.",
+      "context_after": "2.3. Coverage-guided Fuzzing and Directed Fuzzing Fuzzing [9] is a software testing technique that involves providing invalid, unexpected, or random data as inputs to a computer program. The goal is to discover vulnerabilities or bugs by observing how the program reacts to these inputs [18, 30, 41, 55]. Coverage-guided fuzzing is a specific type of fuzz testing that utilizes code coverage feedback from the program’s execution to guide the generation of new test 6 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing inputs. The key idea is to measure the code coverage ach",
+      "section": "2.2 Logic Bug Testing Oracles and Differential Testing",
+      "page": 5,
+      "char_offset": 18316,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[28, 60]",
+      "technique": null,
+      "sentence": "In particular, we analyze a total of 144 existing logic bugs, including 102 in SQLite, 15 in MySQL, 1 in PostgreSQL, and 26 in TiDB [28, 60].",
+      "context_before": "erability or when testing specific features of the program. In this paper, we employ directed fuzzing to deliberately detect logic bugs, rather than aimlessly testing code paths. Specifically, we target logic bug patterns, which denote paths in the source code that are highly prone to triggering logic bugs. We allocate a substantial portion of the time budget to explore these targeted paths, with the objective of uncovering more bugs. 3. Analysis of Existing Logic Bug Patterns To develop an effective DBMS fuzzer for logic bugs, we conduct the first systematic study against existing logic bugs.",
+      "context_after": "This analysis covers most logic bugs, and we do not investigate logic bugs that result in system crashes because we focus on logic bugs that do not lead to observable system failures (such as crashes). Based on the semantics of SQL clause, we classify SQL clauses related to these logic bugs into five categories: table element/schema, data processing clause, conditional expressions, special keyword, and query optimization function, as shown in Table 1. Below introduces each category: •Table element/schema: This category involves elements or schemas in the created tables, such as column indexes",
+      "section": "3 Analysis of Existing Logic Bug Patterns",
+      "page": 6,
+      "char_offset": 21066,
+      "cited_reference": {
+        "number": 60,
+        "text": "Bugs Found in Database Management Systems. https://www.manuelrigger.at/dbms-bugs/.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "In the case of PostgreSQL, only one logic bug was found, specifically related to the PRIMARY KEY and GROUP BY clauses [71].",
+      "context_before": "WID, and NOCASE. We obtain the SQL clause combination for the remaining logic bugs using a similar approach. As for MySQL, the first SQL clause combination pertains to column values, primarily including errors induced when handling floating-point numbers, such as a query whose predicate involves floating-point numbers yielding an incorrect result (bug: 99122 [70]). Additionally, some logic bugs are caused by query optimization functions such as ANY, LIKE, etc (bug: 95927 [64])). Incorrect results can also be caused by conditional expressions such as IFFULL, IF(FALSE), and IN (bug: 95926 [63]).",
+      "context_after": "For TiDB, it can be observed that certain types of columns such as special characters (like double negation, etc.) or specific data types (like floats/boolean, etc.) are more prone to triggering logic bugs in TiDB (bug: 15725 [91]). Additionally, certain SQL clauses such as conditional expressions (bug: 15733 [92]), data processing functions (bug: 15986 [95]), and special keywords (bugs: 15789 [93], 15846 [94], 15994 [96], 17814 [97]) also contribute to triggering logic bugs in TiDB. 4. SQLaser In this section, we present our design and implementation for SQLaser, which enables fuzzing to con",
+      "section": "3 Analysis of Existing Logic Bug Patterns",
+      "page": 7,
+      "char_offset": 25482,
+      "cited_reference": {
+        "number": 71,
+        "text": "SQLancer. https://github.com/sqlancer/sqlancer.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "DBMS SQL-level Bug Pattern (SQL Clause Combination) Number Bug Example SQLite [43]INDEX, PRIMARY KEY, WITHOUT ROWID, NOCASE 1 1b1dd4d4 [74] partial INDEX, LIKELY, ISFAIL 11 5351e920 [76] WITHOUT ROWID, PRIMARY KEYDESC 2 f65c929 [87] column value, CAST, LIKELY, UNLIKELY, GLOB 26 f9c6426 [89] column value, MIN 3 faaaae49 [90] column value, CAST 8 c0c90961 [83] column value, ROUND 1 db9acef1 [84] CAST, EXISTS 3 16252d7 [72] INSERT ORFAIL 1 659c551d [79] RTRIM, PRIMARY KEY, WITHOUT ROWID 1 86fa0087 [81] DISTINCT, ORDER BY 3 6ac0f822 [80] PRAGMA 3 ebe4845c [86] ALTER TABLE, column value 4 1685610e [73] rtree, (COUNT,CAST) 7 f898d04c [88] VIEW, INDEX 11 9c8c1092 [82] IN, ORDER BY 7 eb40248 [85] Others 10 54110870 [77] MySQL [33]column values 4 99122 [70] ANY 1 [28] LIKEESCAPE,XOR 1 95927 [64] BIGINT UNSIGNED, IFNULL 2 95954 [66] “<=>” comparison 1 95908 [62] IF(FALSE) 1 95926 [63] IN operator 1 95975 [67] &, <, and AND 1 95983 [68] GREATEST 1 96012 [62] others 2 95937 [65] PostgreSQL [19] PRIMARY KEY, GROUP BY 1 [71] TiDB [14]columns 11 15725 [91] CAST, IsTrue/IsFalse 2 15733 [92] CHAR() 1 15986 [95] collation 2 15789 [93] JOIN 6 15846 [94] USE_INDEX_MERGE 1 15994 [96] Others 3 17814 [97] bug pattern information as target sites for directed fuzzing, and creating a tool with the capability to efficiently reach predefined target sites.",
+      "context_before": "(bugs: 15789 [93], 15846 [94], 15994 [96], 17814 [97]) also contribute to triggering logic bugs in TiDB. 4. SQLaser In this section, we present our design and implementation for SQLaser, which enables fuzzing to concentrate on error-prone function chains. These function chains are the implementation of SQL-level bug patterns, at the source code level. Specifically, our design achieves two key objectives: incorporating 8 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing Table 2 Analysis of the SQL-level bug patterns of logic bugs for SQLite, MySQL, PostgreSQL and TiDB.",
+      "context_after": "Overview. As shown in Fig. 1, SQLaser consists of two components: 1) logic-bug pattern modeling, and 2) clause-guided fuzzing. Moreover, following established practice of existing directed fuzzing tools [4, 12], SQLaser instruments edge coverage information and basic block distance information into the binary code. Such information is obtained through an approach akin to WindRanger [12], with further intricate details omitted in this paper. In logic-bug pattern modeling phase, SQLaser aims to instrument the call chains of bug pattern information into binary code, and these call chains manifes",
+      "section": "4 SQLaser",
+      "page": 8,
+      "char_offset": 26590,
+      "cited_reference": {
+        "number": 71,
+        "text": "SQLancer. https://github.com/sqlancer/sqlancer.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "The five oracles include: 1) the NoREC oracle, which rewrites an optimized query containing a WHERE clause to a SELECT query statement without the WHERE optimization; 2) the TLP oracle, which partitions a given query into multiple equivalent queries, and their results can be combined to obtain the same results as the original query; 3) the INDEX oracle, which 14 Wei et al.",
+      "context_before": ". Specifically, SQLaser utilizes Equation 8 to calculate the distance between the seed and the target. A shorter distance results in a higher energy allocation for the seed, as expressed in Equation 12. energy =dCallChain−1(12) 4.3.2. Testing Oracles We utilize five testing oracles to analyze the outcomes of seed execution, similar to SQLRight [28]. These oracles can transform the original queries into variant forms with different syntactic structures but equivalent functionality. By comparing the execution results of the original queries and their variants, we can detect potential logic bugs.",
+      "context_after": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing introduces various CREATE INDEX clauses into the given query set, testing the impact of INDEX usage on the query execution; and 4) the ROWID oracle, which examines whether the original queries contain the WITHOUT ROWID clause. If it is not present, the WITHOUT ROWID clause is inserted to observe its effect on the query execution. 5) the LIKELY oracle checks if the original queries include the LIKELY clause. If they don’t, it adds the LIKELY clause to see how it impacts query execution. 5. Implementation In this section, we introd",
+      "section": "4.3.2 Testing Oracles",
+      "page": 13,
+      "char_offset": 44467,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "For instance, some bugs are identified by the PQS oracle [40] proposed by Manuel Rigger et al.",
+      "context_before": "y of the bug patterns. However, there are still some bug patterns where we do not find bugs, possibly due to the limited duration of our fuzzing runs. It is expected that extending the runtime will lead to the discovery of more bugs, and this is outlined as part of our future work. Furthermore, we note that in the case of MySQL, we find bugs in 4 out of the 10 patterns we instrument, indicating that we do not cover the majority of the bug patterns. We attribute this to the incompleteness of testing oracles supported by SQLaser, which may not be comprehensive enough to detect all types of bugs.",
+      "context_after": ", which is not yet integrated into SQLaser (as discussed in Section 7.1). Additionally, 18 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing out of the 15 existing bugs we summarize for MySQL, 13 are detected by the PQS oracle, explaining the comparatively lower number of bug found by SQLaser for MySQL. 6.2. Comparison with Existing Tools For evaluating SQLaser, we also ran SQLRight [28] and WindRanger [12], which represent an instance of coverage-guided fuzzer and single-function-target directed fuzzer, respectively. In this section, we delve into a comparison of the",
+      "section": "6.1 Finding Bugs in DBMSs",
+      "page": 17,
+      "char_offset": 56189,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing out of the 15 existing bugs we summarize for MySQL, 13 are detected by the PQS oracle, explaining the comparatively lower number of bug found by SQLaser for MySQL.",
+      "context_before": "l lead to the discovery of more bugs, and this is outlined as part of our future work. Furthermore, we note that in the case of MySQL, we find bugs in 4 out of the 10 patterns we instrument, indicating that we do not cover the majority of the bug patterns. We attribute this to the incompleteness of testing oracles supported by SQLaser, which may not be comprehensive enough to detect all types of bugs. For instance, some bugs are identified by the PQS oracle [40] proposed by Manuel Rigger et al., which is not yet integrated into SQLaser (as discussed in Section 7.1). Additionally, 18 Wei et al.",
+      "context_after": "6.2. Comparison with Existing Tools For evaluating SQLaser, we also ran SQLRight [28] and WindRanger [12], which represent an instance of coverage-guided fuzzer and single-function-target directed fuzzer, respectively. In this section, we delve into a comparison of the performance, code coverage and seed distance distribution of the three fuzzers. 6.2.1. Performance 0123456789103035...60 TTE(day)024681012Bug Number SQLite SQLaser SQLRight WindRanger 012345678910...30 TTE(day)0123456Bug Number MySQL SQLaser SQLRight WindRanger 012345678910...30 TTE(day)01234Bug Number TiDB SQLaser SQLRight Win",
+      "section": "6.1 Finding Bugs in DBMSs",
+      "page": 18,
+      "char_offset": 56385,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For NoREC and TLP oracles, they are implemented Wei et al.",
+      "context_before": "L and TiDB. Due to the fact that SQLRight and SQLaser do not find bugs in PostgreSQL, we are unable to gather TTE statistics for PostgreSQL. Through our evaluation, we find that both SQLRight and WindRanger take longer to discover bugs or cover target call chains due to their lack of explicit targeting of the bug pattern information. In contrast, SQLaser achieves a significant decrease in bug discovery time, approximately 60%. 6.2.2. Code Coverage As depicted in Fig. 3, Fig. 4 and Fig. 5, we record the code coverage of SQLaser, SQLRight, and WindRanger when testing DBMSs using testing oracles.",
+      "context_after": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing 19 012345678910...60 Day051015Code Coverage (%) SQLite SQLaser SQLRight WindRanger 012345678910...30 Day0204060Code Coverage (%) MySQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) PostgreSQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) TiDB SQLaser SQLRight WindRanger Fig. 3. Code coverage of SQLaser, SQLRight and WindRanger for NoREC oracle. 012345678910...60 Day051015Code Coverage (%) SQLite SQLaser SQLRight WindRanger 012345678910...30 Day0204060Code Coverage (%) MySQL SQ",
+      "section": "6.2.2 Code Coverage",
+      "page": 18,
+      "char_offset": 59362,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Code coverage of SQLaser, SQLRight and WindRanger for NoREC oracle.",
+      "context_before": "epicted in Fig. 3, Fig. 4 and Fig. 5, we record the code coverage of SQLaser, SQLRight, and WindRanger when testing DBMSs using testing oracles. For NoREC and TLP oracles, they are implemented Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing 19 012345678910...60 Day051015Code Coverage (%) SQLite SQLaser SQLRight WindRanger 012345678910...30 Day0204060Code Coverage (%) MySQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) PostgreSQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) TiDB SQLaser SQLRight WindRanger Fig. 3.",
+      "context_after": "012345678910...60 Day051015Code Coverage (%) SQLite SQLaser SQLRight WindRanger 012345678910...30 Day0204060Code Coverage (%) MySQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) PostgreSQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) TiDB SQLaser SQLRight WindRanger Fig. 4. Code coverage of SQLaser, SQLRight and WindRanger to trigger bugs for TLP oracle. 012345678910...60 Day051015Code Coverage (%) ROWID oracle SQLaser SQLRight WindRanger 012345678910...60 Day051015Code Coverage (%) LIKELY oracle SQLaser SQLRight WindRanger 012345678910..",
+      "section": "6.2.2 Code Coverage",
+      "page": 19,
+      "char_offset": 59818,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Code coverage of SQLaser, SQLRight and WindRanger to trigger bugs for TLP oracle.",
+      "context_before": "ySQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) PostgreSQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) TiDB SQLaser SQLRight WindRanger Fig. 3. Code coverage of SQLaser, SQLRight and WindRanger for NoREC oracle. 012345678910...60 Day051015Code Coverage (%) SQLite SQLaser SQLRight WindRanger 012345678910...30 Day0204060Code Coverage (%) MySQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) PostgreSQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) TiDB SQLaser SQLRight WindRanger Fig. 4.",
+      "context_after": "012345678910...60 Day051015Code Coverage (%) ROWID oracle SQLaser SQLRight WindRanger 012345678910...60 Day051015Code Coverage (%) LIKELY oracle SQLaser SQLRight WindRanger 012345678910...60 Day051015Code Coverage (%) INDEX oracle SQLaser SQLRight WindRanger Fig. 5. Code coverage of SQLaser, SQLRight and WindRanger to trigger bugs in SQLite for ROWID, LIKELY and INDEX oracle. in tested four DBMSs. For ROWID, LIKELY, and INDEX oracles, they are only implemented in SQLite for now, and not in other DBMSs yet. According to Fig. 3, Fig. 4 and Fig. 5, we can see that SQLaser has a lower code covera",
+      "section": "6.2.2 Code Coverage",
+      "page": 19,
+      "char_offset": 60216,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Currently, SQLaser only supports differential testingbased oracles and does not accommodate other types of oracles, such as PQS oracle [40], which detects logic bugs through non-differential testing.",
+      "context_before": "the discovery of more bugs, since the fuzzer explores a wider range of possibilities, even if some of them are false positives. 7. Discussion In this section, we discuss two issues: the completeness of the testing oracles employed in this paper, and the completeness of the SQL clause collections we consider. 7.1. Completeness of Testing Oracles In this paper, we employ five oracles to test logic bugs. Specifically, we validate the existence of logic bugs by converting SQL query statements into functionally equivalent yet different queries and comparing the output results of these two queries.",
+      "context_after": "However, in the process of statistically analyzing existing logic bugs, we include the bugs detected by PQS oracles and instrument their call chains. As SQLaser does not currently support PQS oracles, this may result in its inability to identify logic bugs detectable solely by PQS oracles. Additionally, to the best of our knowledge, existing research has not extensively investigated the completeness of these five oracles (NoREC, TLP, INDEX, ROWID and LIKELY) in addressing logic bugs in real-world scenarios. We will undertake research on this matter in future work. 7.2. Completeness of SQL Cla",
+      "section": "7.1 Completeness of Testing Oracles",
+      "page": 21,
+      "char_offset": 66394,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "However, in the process of statistically analyzing existing logic bugs, we include the bugs detected by PQS oracles and instrument their call chains.",
+      "context_before": "of the testing oracles employed in this paper, and the completeness of the SQL clause collections we consider. 7.1. Completeness of Testing Oracles In this paper, we employ five oracles to test logic bugs. Specifically, we validate the existence of logic bugs by converting SQL query statements into functionally equivalent yet different queries and comparing the output results of these two queries. Currently, SQLaser only supports differential testingbased oracles and does not accommodate other types of oracles, such as PQS oracle [40], which detects logic bugs through non-differential testing.",
+      "context_after": "As SQLaser does not currently support PQS oracles, this may result in its inability to identify logic bugs detectable solely by PQS oracles. Additionally, to the best of our knowledge, existing research has not extensively investigated the completeness of these five oracles (NoREC, TLP, INDEX, ROWID and LIKELY) in addressing logic bugs in real-world scenarios. We will undertake research on this matter in future work. 7.2. Completeness of SQL Clause Collections We systematically analyze logic bugs extracted from state-of-the-art studies, identifying common error-prone sequences of functions. W",
+      "section": "7.1 Completeness of Testing Oracles",
+      "page": 21,
+      "char_offset": 66594,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "As SQLaser does not currently support PQS oracles, this may result in its inability to identify logic bugs detectable solely by PQS oracles.",
+      "context_before": "this paper, we employ five oracles to test logic bugs. Specifically, we validate the existence of logic bugs by converting SQL query statements into functionally equivalent yet different queries and comparing the output results of these two queries. Currently, SQLaser only supports differential testingbased oracles and does not accommodate other types of oracles, such as PQS oracle [40], which detects logic bugs through non-differential testing. However, in the process of statistically analyzing existing logic bugs, we include the bugs detected by PQS oracles and instrument their call chains.",
+      "context_after": "Additionally, to the best of our knowledge, existing research has not extensively investigated the completeness of these five oracles (NoREC, TLP, INDEX, ROWID and LIKELY) in addressing logic bugs in real-world scenarios. We will undertake research on this matter in future work. 7.2. Completeness of SQL Clause Collections We systematically analyze logic bugs extracted from state-of-the-art studies, identifying common error-prone sequences of functions. We have made every effort to comprehensively gather existing logic bugs, and these resultant collections effectively represent prevalent logic",
+      "section": "7.1 Completeness of Testing Oracles",
+      "page": 21,
+      "char_offset": 66744,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Additionally, to the best of our knowledge, existing research has not extensively investigated the completeness of these five oracles (NoREC, TLP, INDEX, ROWID and LIKELY) in addressing logic bugs in real-world scenarios.",
+      "context_before": "nts into functionally equivalent yet different queries and comparing the output results of these two queries. Currently, SQLaser only supports differential testingbased oracles and does not accommodate other types of oracles, such as PQS oracle [40], which detects logic bugs through non-differential testing. However, in the process of statistically analyzing existing logic bugs, we include the bugs detected by PQS oracles and instrument their call chains. As SQLaser does not currently support PQS oracles, this may result in its inability to identify logic bugs detectable solely by PQS oracles.",
+      "context_after": "We will undertake research on this matter in future work. 7.2. Completeness of SQL Clause Collections We systematically analyze logic bugs extracted from state-of-the-art studies, identifying common error-prone sequences of functions. We have made every effort to comprehensively gather existing logic bugs, and these resultant collections effectively represent prevalent logic bugs in the field. To our best knowledge, there is currently no theory or standard proposed in existing research to classify SQL clauses. We are the first to strive to investigate the types of SQL clause of existing logic",
+      "section": "7.1 Completeness of Testing Oracles",
+      "page": 21,
+      "char_offset": 66885,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [71] is a rulebased tool that successfully identifies numerous logic bugs.",
+      "context_before": "ollections effectively represent prevalent logic bugs in the field. To our best knowledge, there is currently no theory or standard proposed in existing research to classify SQL clauses. We are the first to strive to investigate the types of SQL clause of existing logic bugs, and leverage the discovered insights for detecting bugs in DBMS. However, we acknowledge that ensuring this classification covers all SQL clause might be a challenge, and addressing this limitation is a direction for our future work. 8. Related Work In this section, we discuss techniques for detecting logic bugs in DBMSs.",
+      "context_after": "It employs three distinct methods for detecting logic bugs: PQS [40], NoREC [38], and TLP [39]. PQS focuses on generating queries that fetch a 22 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing randomly selected row called the pivot row. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS. NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presenc",
+      "section": "8 Related Work",
+      "page": 21,
+      "char_offset": 68037,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "It employs three distinct methods for detecting logic bugs: PQS [40], NoREC [38], and TLP [39].",
+      "context_before": "ledge, there is currently no theory or standard proposed in existing research to classify SQL clauses. We are the first to strive to investigate the types of SQL clause of existing logic bugs, and leverage the discovered insights for detecting bugs in DBMS. However, we acknowledge that ensuring this classification covers all SQL clause might be a challenge, and addressing this limitation is a direction for our future work. 8. Related Work In this section, we discuss techniques for detecting logic bugs in DBMSs. SQLancer [71] is a rulebased tool that successfully identifies numerous logic bugs.",
+      "context_after": "PQS focuses on generating queries that fetch a 22 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing randomly selected row called the pivot row. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS. NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug. Lastly, TLP partitions a query into several sub-queries, which are composed to",
+      "section": "8 Related Work",
+      "page": 21,
+      "char_offset": 68121,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS focuses on generating queries that fetch a 22 Wei et al.",
+      "context_before": "auses. We are the first to strive to investigate the types of SQL clause of existing logic bugs, and leverage the discovered insights for detecting bugs in DBMS. However, we acknowledge that ensuring this classification covers all SQL clause might be a challenge, and addressing this limitation is a direction for our future work. 8. Related Work In this section, we discuss techniques for detecting logic bugs in DBMSs. SQLancer [71] is a rulebased tool that successfully identifies numerous logic bugs. It employs three distinct methods for detecting logic bugs: PQS [40], NoREC [38], and TLP [39].",
+      "context_after": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing randomly selected row called the pivot row. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS. NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug. Lastly, TLP partitions a query into several sub-queries, which are composed to have the same semantics as the original query. If the compos",
+      "section": "8 Related Work",
+      "page": 21,
+      "char_offset": 68217,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics.",
+      "context_before": "direction for our future work. 8. Related Work In this section, we discuss techniques for detecting logic bugs in DBMSs. SQLancer [71] is a rulebased tool that successfully identifies numerous logic bugs. It employs three distinct methods for detecting logic bugs: PQS [40], NoREC [38], and TLP [39]. PQS focuses on generating queries that fetch a 22 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing randomly selected row called the pivot row. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS.",
+      "context_after": "By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug. Lastly, TLP partitions a query into several sub-queries, which are composed to have the same semantics as the original query. If the composed query does not yield the same result as the original query, it suggests the possibility of a logic bug. While SQLancer relies on a rule-based generator to detect logic bugs, limiting its exploration of input samples. Liang et al. [28] propose SQLRight, a methodology for logic bug exploration based on coverage-based fuzzing. SQLRight successfull",
+      "section": "8 Related Work",
+      "page": 22,
+      "char_offset": 68517,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Lastly, TLP partitions a query into several sub-queries, which are composed to have the same semantics as the original query.",
+      "context_before": "distinct methods for detecting logic bugs: PQS [40], NoREC [38], and TLP [39]. PQS focuses on generating queries that fetch a 22 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing randomly selected row called the pivot row. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS. NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug.",
+      "context_after": "If the composed query does not yield the same result as the original query, it suggests the possibility of a logic bug. While SQLancer relies on a rule-based generator to detect logic bugs, limiting its exploration of input samples. Liang et al. [28] propose SQLRight, a methodology for logic bug exploration based on coverage-based fuzzing. SQLRight successfully detects logic bugs in SQLite and MySQL. In their approach, all sample queries are added to a queue, and the fuzzer selects a query to mutate. If the mutated query triggers a new execution path, it is added to the queue. These mutated q",
+      "section": "8 Related Work",
+      "page": 22,
+      "char_offset": 68738,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "While SQLancer relies on a rule-based generator to detect logic bugs, limiting its exploration of input samples.",
+      "context_before": "w. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS. NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug. Lastly, TLP partitions a query into several sub-queries, which are composed to have the same semantics as the original query. If the composed query does not yield the same result as the original query, it suggests the possibility of a logic bug.",
+      "context_after": "Liang et al. [28] propose SQLRight, a methodology for logic bug exploration based on coverage-based fuzzing. SQLRight successfully detects logic bugs in SQLite and MySQL. In their approach, all sample queries are added to a queue, and the fuzzer selects a query to mutate. If the mutated query triggers a new execution path, it is added to the queue. These mutated queries are then tested to identify any unexpected behaviors. However, SQLRight’s coverage-based fuzzing approach lacks high effectiveness in reaching vulnerable target sites due to the equal evaluation of all seeds. In this paper, we",
+      "section": "8 Related Work",
+      "page": 22,
+      "char_offset": 68984,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-10T15:11:01Z",
+    "is_model_written": true,
+    "summary": "SQLaser fuzzes for logic bugs by targeting clause combinations. From 144 known logic bugs across SQLite, MySQL, PostgreSQL and TiDB it identifies the SQL clause combinations that recur in them, then guides fuzzing towards those combinations rather than exploring the grammar uniformly.",
+    "narrative": "SQLancer's oracles are what SQLaser runs and what bounds it. It implements five oracles, two of them NoREC and TLP, which it describes as differential-testing oracles, and reports coverage for each. PQS it explicitly does not implement -- it is a non-differential oracle and does not fit the design -- and the paper is unusually candid about the cost: 13 of the 15 known MySQL bugs were found by PQS, which it gives as the reason SQLaser finds comparatively few there. The bug corpus it mines is itself drawn from SQLancer's reported bugs.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "definition",
+      "M5": "motivation",
+      "M6": "definition",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "motivation",
+      "M10": "reuse_component",
+      "M11": "incidental",
+      "M12": "incidental",
+      "M13": "reuse_component",
+      "M14": "result_comparison",
+      "M15": "result_comparison",
+      "M16": "definition",
+      "M17": "result_comparison",
+      "M18": "result_comparison",
+      "M19": "motivation",
+      "M20": "reuse_component",
+      "M21": "motivation",
+      "M22": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M13",
+          "M20"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M13",
+            "sentence": "The five oracles include: 1) the NoREC oracle, which rewrites an optimized query containing a WHERE clause to a SELECT query statement without the WHERE optimization; 2) the TLP oracle, which partitions a given query into multiple equivalent queries, and their results can be combined to obtain the same results as the original query; 3) the INDEX oracle, which 14 Wei et al.",
+            "section": "4.3.2 Testing Oracles",
+            "page": 13
+          },
+          {
+            "mention_id": "M20",
+            "sentence": "However, in the process of statistically analyzing existing logic bugs, we include the bugs detected by PQS oracles and instrument their call chains.",
+            "section": "7.1 Completeness of Testing Oracles",
+            "page": 21
+          }
+        ],
+        "reasoning": "M13 states SQLaser's five oracles include NoREC and TLP, and M20 that bugs detected by PQS oracles were included in its analysis and their call chains instrumented. The oracles are used as they are, but the paper does not say whether they are invoked through SQLancer or reimplemented, so the form of the reuse is unclear.",
+        "reuse_kind": "unclear"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "NoREC and TLP are used unchanged; what SQLaser contributes is clause-guided fuzzing that decides which test cases to generate, not what makes a result wrong."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M15",
+          "M17",
+          "M18"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M15",
+            "sentence": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing out of the 15 existing bugs we summarize for MySQL, 13 are detected by the PQS oracle, explaining the comparatively lower number of bug found by SQLaser for MySQL.",
+            "section": "6.1 Finding Bugs in DBMSs",
+            "page": 18
+          },
+          {
+            "mention_id": "M17",
+            "sentence": "Code coverage of SQLaser, SQLRight and WindRanger for NoREC oracle.",
+            "section": "6.2.2 Code Coverage",
+            "page": 19
+          },
+          {
+            "mention_id": "M18",
+            "sentence": "Code coverage of SQLaser, SQLRight and WindRanger to trigger bugs for TLP oracle.",
+            "section": "6.2.2 Code Coverage",
+            "page": 19
+          }
+        ],
+        "reasoning": "M17 and M18 report code coverage for SQLaser against SQLRight and WindRanger under the NoREC and TLP oracles, and M15 compares SQLaser's yield on MySQL against PQS's, which found 13 of the 15 known bugs there.",
+        "techniques": [
+          "norec",
+          "tlp",
+          "pqs"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The oracles are described by mechanism and by their limits; no state-of-the-art claim is made for them."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2503_03893.json b/_data/papers/paper_arxiv_2503_03893.json
new file mode 100644
index 0000000..ac96923
--- /dev/null
+++ b/_data/papers/paper_arxiv_2503_03893.json
@@ -0,0 +1,947 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:45Z",
+  "paper": {
+    "id": "paper:arxiv:2503.03893",
+    "title": "Parser Knows Best: Testing DBMS with Coverage-Guided Grammar-Rule Traversal",
+    "authors": [
+      "Yu Liang",
+      "Hong Hu"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2503.03893",
+    "s2_paper_id": "45c07812d74c526823b806aaa2c85a4193ecac16",
+    "url": "https://arxiv.org/abs/2503.03893",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2503.03893",
+      "retrieved_at": "2026-09-09T01:30:45Z",
+      "chars": 79898,
+      "content_sha256": "sha256:d918665784586acbf18109fd9cccc79d44f69340e92d636c44c73a70b62c4ec7"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1337
+      },
+      {
+        "number": "2",
+        "title": "Background & Challenges",
+        "start": 8214
+      },
+      {
+        "number": "2.1",
+        "title": "An Example Memory Error from MySQL",
+        "start": 8240
+      },
+      {
+        "number": "2.2",
+        "title": "Generation-based DBMS Testing Tools",
+        "start": 12175
+      },
+      {
+        "number": "2.3",
+        "title": "Mutation-based DBMS Testing Tools",
+        "start": 15431
+      },
+      {
+        "number": "2.4",
+        "title": "Parser Generators in the DBMSs",
+        "start": 17736
+      },
+      {
+        "number": "3.1",
+        "title": "Parser Rule-based Query Generation",
+        "start": 22992
+      },
+      {
+        "number": "3.2",
+        "title": "Path Explosion due to Recursive Keyword",
+        "start": 27457
+      },
+      {
+        "number": "3.3",
+        "title": "Query Mutation with Coverage Feedback",
+        "start": 32244
+      },
+      {
+        "number": "5.1",
+        "title": "DBMS Bugs",
+        "start": 42061
+      },
+      {
+        "number": "5.2",
+        "title": "Comparison with Existing Tools",
+        "start": 43404
+      },
+      {
+        "number": "5.3",
+        "title": "Contribution of Coverage Feedback",
+        "start": 55879
+      },
+      {
+        "number": "5.4",
+        "title": "Contribution of Diverse Syntax Features",
+        "start": 58160
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. Automated SQL Query Generation for Systematic Testing of Database Engines. In Proceedings of the IEEE/ACM International Conference on Automated Software Engineering (ASE), 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. Testing database engines via query plan guidance. In Proceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Antonio Bovenzi, Domenico Cotroneo, Roberto Pietrantuono, and Stefano Russo. On the aging effects due to concurrency bugs: A case study on mysql. In Reliability Engineering, pages 211–220. IEEE, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. Generating queries with cardinality constraints for dbms testing. IEEE Transactions on Knowledge and Data Engineering, 18(12):1721–1725, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Bikash Chandra, Bhupesh Chawda, Biplab Kar, K. V. Maheshwara Reddy, Shetal Shah, and S. Sudarshan. Data Generation for Testing and Grading SQL Queries. The VLDB Journal, 24(6), Aug 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Yongheng Chen, Rui Zhong, Hong Hu, Hangfan Zhang, Yupeng Yang, Dinghao Wu, and Wenke Lee. One Engine to Fuzz ’em All: Generic Language Processor Testing with Semantic Validation. In Proceedings of the 42nd IEEE Symposium on Security and Privacy (Oakland), Virtual, May 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Jaeseung Choi, Kangsu Kim, Daejin Lee, and Sang Kil Cha. NTFUZZ: Enabling Type-Aware Kernel Fuzzing on Windows with Static Binary Analysis. In Proceedings of the 42nd IEEE Symposium on Security and Privacy (Oakland), Virtual, May 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Catalin Cimpanu. Google Chrome Impacted by New Magellan 2.0 Vulnerabilities. https://www.zdnet.com/article/google-c hrome-impacted-by-new-magellan-2-0-vulnerabilities/, December 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Sung Ta Dinh, Haehyun Cho, Kyle Martin, Adam Oest, Kyle Zeng, Alexandros Kapravelos, Gail-Joon Ahn, Tiffany Bao, Ruoyu Wang, Adam Doupé, et al. Favocado: Fuzzing the Binding Code of JavaScript Engines Using Semantically Correct Test Cases. In Proceedings of the 28th Annual Network and Distributed System Security Symposium (NDSS), Virtual, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Bogdan Ghit, Nicolas Poggi, Josh Rosen, Reynold Xin, and Peter Boncz. SparkFuzz: Searching Correctness Regressions in Modern Query Engines. In Proceedings of the Workshop on Testing Database Systems (DBTest), Portland, Oregon, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Google. ClusterFuzz. https://google.github.io/clusterfuzz. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Google. syzkaller-Kernel Fuzzer. https://github.com/google/ syzkaller, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Tao Guo, Puhan Zhang, Xin Wang, and Qiang Wei. Gramfuzz: Fuzzing testing of web browsers based on grammar analysis and structural mutation. In 2013 Second International Conference on Informatics & Applications (ICIA), pages 212–215. IEEE, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. Dynsql: Stateful fuzzing for database management systems with complex and valid sql query generation. In Proceedings of USENIX Security Symposium (USENIX Security), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Stephen C Johnson and Ravi Sethi. Yacc: a parser generator. UNIX Vol. II: research system, pages 347–374, 1990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB), Tokyo, Japan, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Kyungtae Kim, Dae R Jeong, Chung Hwan Kim, Yeongjin Jang, Insik Shin, and Byoungyoung Lee. HFL: Hybrid Fuzzing on the Linux Kernel. In Proceedings of the 27th Annual Network and Distributed System Security Symposium (NDSS), San Diego, CA, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Guoliang Li, Xuanhe Zhou, Shifu Li, and Bo Gao. QTune: A QueryAware Database Tuning System with Deep Reinforcement Learning. Proceedings of the VLDB Endowment, 12(12):2118–2130, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Yabin Li, Yuanping Nie, and Xiaohui Kuang. Fuzzing dbms via nnlm. In2022 7th IEEE International Conference on Data Science in Cyberspace (DSC), pages 367–374. IEEE, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. Sequence-oriented dbms fuzzing. In Proceedings of IEEE International Conference on Data Engineering (ICDE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Yu Liang, Song Liu, and Hong Hu. Detecting logical bugs of {DBMS } with coverage-based guidance. In 31st USENIX Security Symposium (USENIX Security 22), pages 4309–4326, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. Automatic detection of performance bugs in database systems using equivalent queries. In Proceedings of the 44th International Conference on Software Engineering, pages 225–236, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Eric Lo, Carsten Binnig, Donald Kossmann, M. Tamer Özsu, and WingKai Hon. A Framework for Testing DBMS Features. The VLDB Journal, 19(2):203–230, April 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Manuel Rigger. SQLancer. https://github.com/sqlancer/sqla ncer. (visited in October 2023).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 25,
+      "text": "Michaël Marcozzi, Wim Vanhoof, and Jean-Luc Hainaut. Test Input Generation for Database Programs Using Relational Constraints. In Proceedings of the Fifth International Workshop on Testing Database Systems (DBTest), Scottsdale, Arizona, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. Generating Targeted Queries for Database Testing. In Proceedings of the 2008 ACMSIGMOD International Conference on Management of Data (SIGMOD), New York, NY, USA, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Shankara Pailoor, Andrew Aday, and Suman Jana. MoonShine: Optimizing OS Fuzzer Seed Selection with Trace Distillation. In Proceedings of the 27th USENIX Security Symposium (USENIX Security), Baltimore, MD, 2018. 14",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Soyeon Park, Wen Xu, Insu Yun, Daehee Jang, and Taesoo Kim. Fuzzing JavaScript Engines with Aspect-preserving Mutation. In Proceedings of the 41st IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Terence Parr and Kathleen Fisher. Ll (*) the foundation of the antlr parser generator. ACM Sigplan Notices, 46(6):425–436, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Terence J. Parr and Russell W. Quong. Antlr: A predicated-ll (k) parser generator. Software: Practice and Experience, 25(7):789–810, 1995.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Alexandre Rebert, Sang Kil Cha, Thanassis Avgerinos, Jonathan Foote, David Warren, Gustavo Grieco, and David Brumley. Optimizing seed selection for fuzzing. In 23rd USENIX Security Symposium (USENIX Security 14), pages 861–875, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger and Zhendong Su. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA):1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 34,
+      "text": "Manuel Rigger and Zhendong Su. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), Virtual, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 35,
+      "text": "Contessa Brewer Rohan Goswami. Caesars paid millions in ransom to cybercrime group prior to MGM hack. https://www.cnbc.com/2 023/09/14/caesars-paid-millions-in-ransom-to-cybercr ime-group-prior-to-mgm-hack.html, September 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Zeeshan Shah. Over 2 million Pakistanis’ data hacked from restaurants database up for sale. https://www.geo.tv/latest/510830-ove r-2-million-pakistanis-data-hacked-from-restaurants -database-up-for-sale, September 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Chaofan Shou, Ismet Burak Kadron, Qi Su, and Tevfik Bultan. Corbfuzz: Checking browser security policies with fuzzing. In 2021 36th IEEE/ACM International Conference on Automated Software Engineering (ASE), pages 215–226. IEEE, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Donald R Slutz. Massive Stochastic Testing of SQL. In VLDB, volume 98, pages 618–622. Citeseer, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Michael Stonebraker, Sam Madden, and Pradeep Dubey. Intel \"Big Data\" Science and Technology Center Vision and Execution Plan. ACMSIGMOD Record, 42(1):44–49, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Well-known Users of SQLite. https://www.sqlite.org/famous. html. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Go Fuzzing. https://www.sqlite.org/lemon.html. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "MySQL Customers. https://www.mysql.com/customers/. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "The Lemon LALR(1) Parser Generator. https://www.sqlite.org /lemon.html. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "CockroachDB Customers. https://www.cockroachlabs.com/cu stomers/. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "PingCAP Partner. https://www.pingcap.com/partners/. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "PostgreSQL Clients. https://www.postgresql.org/about/. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "SQLSmith. https://github.com/anse1/sqlsmith. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "CockroachDB Parser. https://github.com/cockroachdb/cockr oach/blob/master/pkg/sql/parser/sql.y. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "MariaDB Parser. https://github.com/MariaDB/server/blob /11.3/sql/sql_yacc.yy. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "MySQL Parser. https://github.com/mysql/mysql-server/bl ob/trunk/sql/sql_yacc.yy. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "PostgreSQL: The Parser Stage. https://www.postgresql.org/d ocs/current/parser-stage.html. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "TiDB Parser. https://github.com/pingcap/tidb/blob/maste r/parser/parser.y. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Testing random, valid SQL in CockroachDB. https://www.cockro achlabs.com/blog/testing-random-valid-sql-in-cockroa chdb/, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Backus–Naur form. https://en.wikipedia.org/wiki/Backus âĂŞNaur_form, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "GNU Bison: Languages and Context-Free Grammars. https://www. gnu.org/software/bison/manual/html_node/Language-and-Grammar.html, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "goyacc.https://pkg.go.dev/golang.org/x/tools/cmd/goya cc, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "SQLsmith Bugs. https://github.com/anse1/sqlsmith/wiki #score-list, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "J. Wang, P. Zhang, L. Zhang, H. Zhu, and X. Ye. A Model-based Fuzzing Approach for DBMS. In Proceedings of the 8th International Conference on Communications and Networking in China (CHINACOM), Aug 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Jiajie Wang, Puhan Zhang, Lei Zhang, Haowen Zhu, and Xiaojun Ye. A model-based fuzzing approach for dbms. In 2013 8th International Conference on Communications and Networking in China (CHINACOM), pages 426–431. IEEE, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In 2021 IEEE/ACM 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP), 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Shihao Wen, Peng Jia, Pin Yang, and Chi Hu. Squill: Testing dbms with correctness feedback and accurate instantiation. Applied Sciences, 13(4):2519, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Wen Xu, Hyungon Moon, Sanidhya Kashyap, Po-Ning Tseng, and Taesoo Kim. Fuzzing File Systems via Two-Dimensional Input Space Exploration. In Proceedings of the 40th IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, May 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Wen Xu, Soyeon Park, and Taesoo Kim. FREEDOM: Engineering a State-of-the-Art DOM Fuzzer. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, FL, November 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Jiaqi Yan, Qiuye Jin, Shrainik Jain, Stratis D. Viglas, and Allison Lee. Snowtrail: Testing with Production Queries on a Cloud Database. In Proceedings of the Workshop on Testing Database Systems (DBTest), New York, NY, USA, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Junwen Yang, Pranav Subramaniam, Shan Lu, Cong Yan, and Alvin Cheung. How not to structure your database-backed web applications: a study of performance bugs in the wild. In Proceedings of the 40th International Conference on Software Engineering, pages 800–810, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Michal Zalewski. American Fuzzy Lop (2.52b). http://lcamtuf. coredump.cx/afl. (visited in October 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. 15",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. Testing database engines via query plan guidance. In Proceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 24,
+      "text": "Manuel Rigger. SQLancer. https://github.com/sqlancer/sqla ncer. (visited in October 2023).",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger and Zhendong Su. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA):1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 34,
+      "text": "Manuel Rigger and Zhendong Su. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), Virtual, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[2, 14, 47, 67]",
+      "technique": "qpg",
+      "sentence": "Recent efforts on DBMS testing [2, 14, 47, 67] can be classified into two categories: generation-based testing and mutation-based grey-box fuzzing.",
+      "context_before": "aluation, ParserFuzz outperforms all state-of-the-art existing DBMS testing tools in terms of bug finding, grammar rule coverage and code coverage. ParserFuzz detects 81 previouslyunknown bugs in total across 5 popular DBMSs, where all bugs are confirmed and 34 have been fixed. 1 Introduction Database Management System (DBMS) stores, retrieves and manages data in a structured manner. They are extensively used in real-world data-intensive applications to drive trillions of Internet services and electronic devices [39, 40, 42, 44 –46]. Any DBMS bugs will affect a large number of users [8,35,36].",
+      "context_after": "SQLsmith [47] is the most popular generation-based testing tool to date. It generates SQL queries based on pre-defined query templates. These templates are manually crafted by SQLsmith ’s developers, and can help generate high-quality SQL queries [57]. Another representative generation-based tool is SQLancer +QPG,where QPGrepresents query plan guidance [2].SQLancer +QPG adopts DBMS query plan information as the feedback to guide the query generation process, and is designed to detect logic errors from the DBMS code. Similar to SQLsmith, SQLancer +QPGalso generates query sequences based on pred",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 1658,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. Testing database engines via query plan guidance. In Proceedings of International Conference on Software Engineering (ICSE), 2023.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Another representative generation-based tool is SQLancer +QPG,where QPGrepresents query plan guidance [2].",
+      "context_before": "sively used in real-world data-intensive applications to drive trillions of Internet services and electronic devices [39, 40, 42, 44 –46]. Any DBMS bugs will affect a large number of users [8,35,36]. Recent efforts on DBMS testing [2, 14, 47, 67] can be classified into two categories: generation-based testing and mutation-based grey-box fuzzing. SQLsmith [47] is the most popular generation-based testing tool to date. It generates SQL queries based on pre-defined query templates. These templates are manually crafted by SQLsmith ’s developers, and can help generate high-quality SQL queries [57].",
+      "context_after": "SQLancer +QPG adopts DBMS query plan information as the feedback to guide the query generation process, and is designed to detect logic errors from the DBMS code. Similar to SQLsmith, SQLancer +QPGalso generates query sequences based on predefined SQL templates. However, due to the significant difference between multiple DBMS dialects, these pre-defined SQL templates cannot cover unique, complicated features from different DBMS systems [21,38]. Further, as every DBMS keeps evolving, developers of SQLsmith and SQLancer +QPGhave to track all recent updates and manually insert new templates. Recen",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2059,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer +QPG adopts DBMS query plan information as the feedback to guide the query generation process, and is designed to detect logic errors from the DBMS code.",
+      "context_before": "ic devices [39, 40, 42, 44 –46]. Any DBMS bugs will affect a large number of users [8,35,36]. Recent efforts on DBMS testing [2, 14, 47, 67] can be classified into two categories: generation-based testing and mutation-based grey-box fuzzing. SQLsmith [47] is the most popular generation-based testing tool to date. It generates SQL queries based on pre-defined query templates. These templates are manually crafted by SQLsmith ’s developers, and can help generate high-quality SQL queries [57]. Another representative generation-based tool is SQLancer +QPG,where QPGrepresents query plan guidance [2].",
+      "context_after": "Similar to SQLsmith, SQLancer +QPGalso generates query sequences based on predefined SQL templates. However, due to the significant difference between multiple DBMS dialects, these pre-defined SQL templates cannot cover unique, complicated features from different DBMS systems [21,38]. Further, as every DBMS keeps evolving, developers of SQLsmith and SQLancer +QPGhave to track all recent updates and manually insert new templates. Recently, coverage-guided grey-box fuzzing is widely used to detect memory errors from a wide-range of applications, including but not limited to operating systems [7,",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2165,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Similar to SQLsmith, SQLancer +QPGalso generates query sequences based on predefined SQL templates.",
+      "context_before": "o two categories: generation-based testing and mutation-based grey-box fuzzing. SQLsmith [47] is the most popular generation-based testing tool to date. It generates SQL queries based on pre-defined query templates. These templates are manually crafted by SQLsmith ’s developers, and can help generate high-quality SQL queries [57]. Another representative generation-based tool is SQLancer +QPG,where QPGrepresents query plan guidance [2].SQLancer +QPG adopts DBMS query plan information as the feedback to guide the query generation process, and is designed to detect logic errors from the DBMS code.",
+      "context_after": "However, due to the significant difference between multiple DBMS dialects, these pre-defined SQL templates cannot cover unique, complicated features from different DBMS systems [21,38]. Further, as every DBMS keeps evolving, developers of SQLsmith and SQLancer +QPGhave to track all recent updates and manually insert new templates. Recently, coverage-guided grey-box fuzzing is widely used to detect memory errors from a wide-range of applications, including but not limited to operating systems [7, 12, 17, 27, 62], web browsers [11, 13, 37, 63] and compilers [6, 9, 28]. To conduct grey-box testin",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2327,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[24]",
+      "technique": null,
+      "sentence": "There is another generation-based DBMS testing tool called SQLancer [24], that focuses on detecting DBMS logic errors from DBMS systems [32 –34].",
+      "context_before": "generation-based query testing framework is SQLsmith [47]. Since its release, SQLsmith has been used extensively to test on different DBMSs, and found many bugs from the DBMS softwares [57]. However, the hand-written query templates are limited in covering the syntax elements from the DBMS syntax rules, and cannot fit in the complex and ever-changing SQL dialects defined in different DBMS softwares. Therefore, the queries generated from SQLsmith cannot cover all the SQL features from the DBMSs, and lack the capability to detect deep and unique bugs that are dedicated to the DBMSs’ feature sets.",
+      "context_after": "DBMS logic bugs are code logic errors that cause the DBMS to return incorrect results. Unlike tools that detect memory errors, SQLancer doesn’t generate arbitrary types of random queries for fuzzing. Instead, it focuses only on generating queries that matchingits oracles’ needs. In essence, SQLancer prefers to generate multiple syntactically different, but functionally equivalence queries, and verify their results to ensure the query execution correctness. SQLancer introduces a few SQL oracles for this purpose such as NoREC ,TLPandPQS, where each shares a distinct SQL pattern to match [32 –34",
+      "section": "2.2 Generation-based DBMS Testing Tools",
+      "page": 3,
+      "char_offset": 12948,
+      "cited_reference": {
+        "number": 24,
+        "text": "Manuel Rigger. SQLancer. https://github.com/sqlancer/sqla ncer. (visited in October 2023).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Unlike tools that detect memory errors, SQLancer doesn’t generate arbitrary types of random queries for fuzzing.",
+      "context_before": "are limited in covering the syntax elements from the DBMS syntax rules, and cannot fit in the complex and ever-changing SQL dialects defined in different DBMS softwares. Therefore, the queries generated from SQLsmith cannot cover all the SQL features from the DBMSs, and lack the capability to detect deep and unique bugs that are dedicated to the DBMSs’ feature sets. There is another generation-based DBMS testing tool called SQLancer [24], that focuses on detecting DBMS logic errors from DBMS systems [32 –34]. DBMS logic bugs are code logic errors that cause the DBMS to return incorrect results.",
+      "context_after": "Instead, it focuses only on generating queries that matchingits oracles’ needs. In essence, SQLancer prefers to generate multiple syntactically different, but functionally equivalence queries, and verify their results to ensure the query execution correctness. SQLancer introduces a few SQL oracles for this purpose such as NoREC ,TLPandPQS, where each shares a distinct SQL pattern to match [32 –34]. With its latest configuration SQLancer +QPG[2], it uses the DBMS query plan to guide its query generation in order to stress test the DBMS query optimization logic. However, because SQLancer and SQ",
+      "section": "2.2 Generation-based DBMS Testing Tools",
+      "page": 3,
+      "char_offset": 13180,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In essence, SQLancer prefers to generate multiple syntactically different, but functionally equivalence queries, and verify their results to ensure the query execution correctness.",
+      "context_before": "generated from SQLsmith cannot cover all the SQL features from the DBMSs, and lack the capability to detect deep and unique bugs that are dedicated to the DBMSs’ feature sets. There is another generation-based DBMS testing tool called SQLancer [24], that focuses on detecting DBMS logic errors from DBMS systems [32 –34]. DBMS logic bugs are code logic errors that cause the DBMS to return incorrect results. Unlike tools that detect memory errors, SQLancer doesn’t generate arbitrary types of random queries for fuzzing. Instead, it focuses only on generating queries that matchingits oracles’ needs.",
+      "context_after": "SQLancer introduces a few SQL oracles for this purpose such as NoREC ,TLPandPQS, where each shares a distinct SQL pattern to match [32 –34]. With its latest configuration SQLancer +QPG[2], it uses the DBMS query plan to guide its query generation in order to stress test the DBMS query optimization logic. However, because SQLancer and SQLancer +QPGrestricted themselves to generate queries that align with the oracles’ patterns, they lack the query diversity needed to explore all the grammar features provided by the DBMSs. Therefore, neither SQLancer nor SQLancer +QPGare suitable to detect DBMS m",
+      "section": "2.2 Generation-based DBMS Testing Tools",
+      "page": 3,
+      "char_offset": 13373,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer introduces a few SQL oracles for this purpose such as NoREC ,TLPandPQS, where each shares a distinct SQL pattern to match [32 –34].",
+      "context_before": "is another generation-based DBMS testing tool called SQLancer [24], that focuses on detecting DBMS logic errors from DBMS systems [32 –34]. DBMS logic bugs are code logic errors that cause the DBMS to return incorrect results. Unlike tools that detect memory errors, SQLancer doesn’t generate arbitrary types of random queries for fuzzing. Instead, it focuses only on generating queries that matchingits oracles’ needs. In essence, SQLancer prefers to generate multiple syntactically different, but functionally equivalence queries, and verify their results to ensure the query execution correctness.",
+      "context_after": "With its latest configuration SQLancer +QPG[2], it uses the DBMS query plan to guide its query generation in order to stress test the DBMS query optimization logic. However, because SQLancer and SQLancer +QPGrestricted themselves to generate queries that align with the oracles’ patterns, they lack the query diversity needed to explore all the grammar features provided by the DBMSs. Therefore, neither SQLancer nor SQLancer +QPGare suitable to detect DBMS memory corruption bugs that are arise from interesting but rarely tested syntax features. There are several other generation-based DBMS testin",
+      "section": "2.2 Generation-based DBMS Testing Tools",
+      "page": 3,
+      "char_offset": 13554,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "With its latest configuration SQLancer +QPG[2], it uses the DBMS query plan to guide its query generation in order to stress test the DBMS query optimization logic.",
+      "context_before": "BMS logic bugs are code logic errors that cause the DBMS to return incorrect results. Unlike tools that detect memory errors, SQLancer doesn’t generate arbitrary types of random queries for fuzzing. Instead, it focuses only on generating queries that matchingits oracles’ needs. In essence, SQLancer prefers to generate multiple syntactically different, but functionally equivalence queries, and verify their results to ensure the query execution correctness. SQLancer introduces a few SQL oracles for this purpose such as NoREC ,TLPandPQS, where each shares a distinct SQL pattern to match [32 –34].",
+      "context_after": "However, because SQLancer and SQLancer +QPGrestricted themselves to generate queries that align with the oracles’ patterns, they lack the query diversity needed to explore all the grammar features provided by the DBMSs. Therefore, neither SQLancer nor SQLancer +QPGare suitable to detect DBMS memory corruption bugs that are arise from interesting but rarely tested syntax features. There are several other generation-based DBMS testing tools that aim to detect various kinds of bugs from the DBMSs [23, 26, 58]. Some existing works treat the SQL query generation as a boolean satisfiability problem,",
+      "section": "2.2 Generation-based DBMS Testing Tools",
+      "page": 3,
+      "char_offset": 13695,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "However, because SQLancer and SQLancer +QPGrestricted themselves to generate queries that align with the oracles’ patterns, they lack the query diversity needed to explore all the grammar features provided by the DBMSs.",
+      "context_before": "es of random queries for fuzzing. Instead, it focuses only on generating queries that matchingits oracles’ needs. In essence, SQLancer prefers to generate multiple syntactically different, but functionally equivalence queries, and verify their results to ensure the query execution correctness. SQLancer introduces a few SQL oracles for this purpose such as NoREC ,TLPandPQS, where each shares a distinct SQL pattern to match [32 –34]. With its latest configuration SQLancer +QPG[2], it uses the DBMS query plan to guide its query generation in order to stress test the DBMS query optimization logic.",
+      "context_after": "Therefore, neither SQLancer nor SQLancer +QPGare suitable to detect DBMS memory corruption bugs that are arise from interesting but rarely tested syntax features. There are several other generation-based DBMS testing tools that aim to detect various kinds of bugs from the DBMSs [23, 26, 58]. Some existing works treat the SQL query generation as a boolean satisfiability problem, and use SAT solvers to produce queries that achieve high correctness rate [1, 25]. Chandra et al. extend the database construction technique to boost the efficiency of DBMS query testings [5]. Bruno et al. propose to ge",
+      "section": "2.2 Generation-based DBMS Testing Tools",
+      "page": 3,
+      "char_offset": 13860,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Therefore, neither SQLancer nor SQLancer +QPGare suitable to detect DBMS memory corruption bugs that are arise from interesting but rarely tested syntax features.",
+      "context_before": "eries, and verify their results to ensure the query execution correctness. SQLancer introduces a few SQL oracles for this purpose such as NoREC ,TLPandPQS, where each shares a distinct SQL pattern to match [32 –34]. With its latest configuration SQLancer +QPG[2], it uses the DBMS query plan to guide its query generation in order to stress test the DBMS query optimization logic. However, because SQLancer and SQLancer +QPGrestricted themselves to generate queries that align with the oracles’ patterns, they lack the query diversity needed to explore all the grammar features provided by the DBMSs.",
+      "context_after": "There are several other generation-based DBMS testing tools that aim to detect various kinds of bugs from the DBMSs [23, 26, 58]. Some existing works treat the SQL query generation as a boolean satisfiability problem, and use SAT solvers to produce queries that achieve high correctness rate [1, 25]. Chandra et al. extend the database construction technique to boost the efficiency of DBMS query testings [5]. Bruno et al. propose to generate queries based on Cardinality Constraints [4]. On the topic of performance issues, researchers also propose several tools to detect performance bugs in the",
+      "section": "2.2 Generation-based DBMS Testing Tools",
+      "page": 3,
+      "char_offset": 14080,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In addition, to understand the memory error detecting capability for DBMS logic bug detectors, we compareParserFuzz to state-of-the-art logic bug testing tool SQLancer +QPG.",
+      "context_before": "e to the DBMS as baselines. For SQLite ,MySQL andMariaDB, we compare ParserFuzz againstSquirrel, the most advanced grey-box mutation-based DBMS fuzzer. For CockroachDB andTiDB, we use the official query generation-based testing tools that are maintained by the DBMS developer groups, i.e., we test the customized SQLsmith ( SQLsmith Cfor short) for CockroachDB, and test go-sqlsmith ( SQLsmith Gfor short) for TiDB respectively. To compare ParserFuzz against traditional bit-flips mutation-based fuzzer, we select AFL++ to testC/C++ implemented DBMSs and use LibFuzzer to testGoLang implemented ones.",
+      "context_after": "SQLancer +QPG supports testing with SQLite, CockroachDB andTiDB, and outperforms all other logic bug detectors including SQLRight [2, 21]. We use NoREC oracle for SQLancer +QPGwhen testing with SQLite andCockroachDB. BecauseNoREC oracle is claimed to be a better performer overall compared to TLPoracle [2]. But we fallback to use TLP when testing TiDB, because SQLancer +QPGhasn’t supported testingTiDB withNoREC oracle yet. While the most recent DBMS fuzzing tool DynSQL [14] supports testing 6 DBMSs including SQLite ,MySQL andMariaDB, it is not open-source, so we cannot compare our tool to their i",
+      "section": "3.3 Query Mutation with Coverage Feedback",
+      "page": 8,
+      "char_offset": 40363,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer +QPG supports testing with SQLite, CockroachDB andTiDB, and outperforms all other logic bug detectors including SQLRight [2, 21].",
+      "context_before": "B, we use the official query generation-based testing tools that are maintained by the DBMS developer groups, i.e., we test the customized SQLsmith ( SQLsmith Cfor short) for CockroachDB, and test go-sqlsmith ( SQLsmith Gfor short) for TiDB respectively. To compare ParserFuzz against traditional bit-flips mutation-based fuzzer, we select AFL++ to testC/C++ implemented DBMSs and use LibFuzzer to testGoLang implemented ones. In addition, to understand the memory error detecting capability for DBMS logic bug detectors, we compareParserFuzz to state-of-the-art logic bug testing tool SQLancer +QPG.",
+      "context_after": "We use NoREC oracle for SQLancer +QPGwhen testing with SQLite andCockroachDB. BecauseNoREC oracle is claimed to be a better performer overall compared to TLPoracle [2]. But we fallback to use TLP when testing TiDB, because SQLancer +QPGhasn’t supported testingTiDB withNoREC oracle yet. While the most recent DBMS fuzzing tool DynSQL [14] supports testing 6 DBMSs including SQLite ,MySQL andMariaDB, it is not open-source, so we cannot compare our tool to their implementation. For fuzzing tools that demand input corpus, we use the query libraries from the Squirrel repo to serve as the universal inp",
+      "section": "3.3 Query Mutation with Coverage Feedback",
+      "page": 8,
+      "char_offset": 40537,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "We use NoREC oracle for SQLancer +QPGwhen testing with SQLite andCockroachDB.",
+      "context_before": "SQLsmith ( SQLsmith Cfor short) for CockroachDB, and test go-sqlsmith ( SQLsmith Gfor short) for TiDB respectively. To compare ParserFuzz against traditional bit-flips mutation-based fuzzer, we select AFL++ to testC/C++ implemented DBMSs and use LibFuzzer to testGoLang implemented ones. In addition, to understand the memory error detecting capability for DBMS logic bug detectors, we compareParserFuzz to state-of-the-art logic bug testing tool SQLancer +QPG.SQLancer +QPG supports testing with SQLite, CockroachDB andTiDB, and outperforms all other logic bug detectors including SQLRight [2, 21].",
+      "context_after": "BecauseNoREC oracle is claimed to be a better performer overall compared to TLPoracle [2]. But we fallback to use TLP when testing TiDB, because SQLancer +QPGhasn’t supported testingTiDB withNoREC oracle yet. While the most recent DBMS fuzzing tool DynSQL [14] supports testing 6 DBMSs including SQLite ,MySQL andMariaDB, it is not open-source, so we cannot compare our tool to their implementation. For fuzzing tools that demand input corpus, we use the query libraries from the Squirrel repo to serve as the universal input seeds. To answer Q3, we disable the code coverage feedback fromParserFuzz,",
+      "section": "3.3 Query Mutation with Coverage Feedback",
+      "page": 8,
+      "char_offset": 40675,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[2]",
+      "technique": "qpg",
+      "sentence": "BecauseNoREC oracle is claimed to be a better performer overall compared to TLPoracle [2].",
+      "context_before": "mith Gfor short) for TiDB respectively. To compare ParserFuzz against traditional bit-flips mutation-based fuzzer, we select AFL++ to testC/C++ implemented DBMSs and use LibFuzzer to testGoLang implemented ones. In addition, to understand the memory error detecting capability for DBMS logic bug detectors, we compareParserFuzz to state-of-the-art logic bug testing tool SQLancer +QPG.SQLancer +QPG supports testing with SQLite, CockroachDB andTiDB, and outperforms all other logic bug detectors including SQLRight [2, 21]. We use NoREC oracle for SQLancer +QPGwhen testing with SQLite andCockroachDB.",
+      "context_after": "But we fallback to use TLP when testing TiDB, because SQLancer +QPGhasn’t supported testingTiDB withNoREC oracle yet. While the most recent DBMS fuzzing tool DynSQL [14] supports testing 6 DBMSs including SQLite ,MySQL andMariaDB, it is not open-source, so we cannot compare our tool to their implementation. For fuzzing tools that demand input corpus, we use the query libraries from the Squirrel repo to serve as the universal input seeds. To answer Q3, we disable the code coverage feedback fromParserFuzz, transforming it into a pure random query generation tool, which noted as ParserFuzz -cov.",
+      "section": "3.3 Query Mutation with Coverage Feedback",
+      "page": 8,
+      "char_offset": 40752,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. Testing database engines via query plan guidance. In Proceedings of International Conference on Software Engineering (ICSE), 2023.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "But we fallback to use TLP when testing TiDB, because SQLancer +QPGhasn’t supported testingTiDB withNoREC oracle yet.",
+      "context_before": "mutation-based fuzzer, we select AFL++ to testC/C++ implemented DBMSs and use LibFuzzer to testGoLang implemented ones. In addition, to understand the memory error detecting capability for DBMS logic bug detectors, we compareParserFuzz to state-of-the-art logic bug testing tool SQLancer +QPG.SQLancer +QPG supports testing with SQLite, CockroachDB andTiDB, and outperforms all other logic bug detectors including SQLRight [2, 21]. We use NoREC oracle for SQLancer +QPGwhen testing with SQLite andCockroachDB. BecauseNoREC oracle is claimed to be a better performer overall compared to TLPoracle [2].",
+      "context_after": "While the most recent DBMS fuzzing tool DynSQL [14] supports testing 6 DBMSs including SQLite ,MySQL andMariaDB, it is not open-source, so we cannot compare our tool to their implementation. For fuzzing tools that demand input corpus, we use the query libraries from the Squirrel repo to serve as the universal input seeds. To answer Q3, we disable the code coverage feedback fromParserFuzz, transforming it into a pure random query generation tool, which noted as ParserFuzz -cov. We compare ParserFuzz -covagainst the full-featured ParserFuzz in §5.3. Finally, we use the bugs detected to demonstra",
+      "section": "3.3 Query Mutation with Coverage Feedback",
+      "page": 8,
+      "char_offset": 40843,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Although SQLancer +QPGcan detect multiple logic bugs in TiDB in Figure 2i, it detects less memory errors thanParserFuzz in all SQLite ,CockroachDB andTiDB testings.",
+      "context_before": "on. Squirrel can also find one new crashing bug from MySQL. For the new bug detected from Squirrel, we also reported it to theMySQL developer. In addition, Squirrel identifies 2 crashing bugs in Figure 2m. However, the detected bugs from Squirrel are old bugs that had already been known to the developer back in 2019 and 2022 respectively. Despite this, ParserFuzz records the highest bug count with 3 bugs detected in MariaDB fuzzing. Moreover, as shown in Figure 2e and Figure 2i, ParserFuzz detects remarkable numbers of bugs when testing on CockroachDB andTiDB, giving 6 and 4 bugs respectively.",
+      "context_after": "All baselines tools except SQLancer +QPGdo not detect any issues in SQLite evaluation as shown in Figure 2q, where ParserFuzz detects 2 bugs within the set time frame. Grammar edge number. The extensive amount of grammar edge triggered by the ParserFuzz fuzzing is the primary reason why it can find more memory errors compared to other baseline tools. A grammar edge representsthe possible combinations between two non-terminal keywords. For example, in Listing 8, a keyword mapping from table_reference totable_factor represents one edge case, andtable_reference tojoined_table represents another.",
+      "section": "5.2 Comparison with Existing Tools",
+      "page": 9,
+      "char_offset": 46925,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "All baselines tools except SQLancer +QPGdo not detect any issues in SQLite evaluation as shown in Figure 2q, where ParserFuzz detects 2 bugs within the set time frame.",
+      "context_before": "identifies 2 crashing bugs in Figure 2m. However, the detected bugs from Squirrel are old bugs that had already been known to the developer back in 2019 and 2022 respectively. Despite this, ParserFuzz records the highest bug count with 3 bugs detected in MariaDB fuzzing. Moreover, as shown in Figure 2e and Figure 2i, ParserFuzz detects remarkable numbers of bugs when testing on CockroachDB andTiDB, giving 6 and 4 bugs respectively. Although SQLancer +QPGcan detect multiple logic bugs in TiDB in Figure 2i, it detects less memory errors thanParserFuzz in all SQLite ,CockroachDB andTiDB testings.",
+      "context_after": "Grammar edge number. The extensive amount of grammar edge triggered by the ParserFuzz fuzzing is the primary reason why it can find more memory errors compared to other baseline tools. A grammar edge representsthe possible combinations between two non-terminal keywords. For example, in Listing 8, a keyword mapping from table_reference totable_factor represents one edge case, andtable_reference tojoined_table represents another. The upper bound lines display the total possible grammar edges for each DBMSs’ grammar rules. The grammar edge coverage plots are presented in Figure 2b, Figure 2f, Fi",
+      "section": "5.2 Comparison with Existing Tools",
+      "page": 9,
+      "char_offset": 47089,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "52 ParserFuzz ParserFuzz−cov Squirrel AFL++ SQLSmith SQLancer+QPG 0 10 20 30 40 501.",
+      "context_before": "whole CockroachDB process. reported than the baseline tools. Code coverage. ParserFuzz reaches the highest DBMS code coverage across all 5 DBMSs’ experiments. The code coverage plots are shown in Figure 2c, Figure 2g, Figure 2k, Figure 2o and Figure 2s. Notably, ParserFuzz doesn’t rely onany input corpus to reach this level of code coverage, sparing the efforts from the DBMS testers to gather interesting queries as input seeds. Query correctness rate. The query correctness rate is illustrated in Figure 2d, Figure 2h, Figure 2l, Figure 2p and Fig10 0 10 20 30 40 501.381.401.421.441.461.481.501.",
+      "context_after": "381.401.421.441.461.481.501.52 SQLsmithC SQLsmithG LibFuzzer Upper Bound 0481216202401234 (a)MySQL: detected bugs 048121620240.00.51.01.52.02.53.03.54.0 (b)MySQL: grammar edges (k) 048121620240255075100125150175200 (c)MySQL: code coverage (k) 04812162024020406080100 (d)MySQL: query validity (%) 048121620240123456 (e)CockroachDB: detected bugs 048121620240.00.20.40.60.81.01.21.4 (f)CockroachDB: grammar edges (k) 0481216202401020304050607080 (g)CockroachDB: code coverage (k) 04812162024020406080100 (h)CockroachDB: query validity (%) 0481216202401234 (i)TiDB: detected bugs 048121620240.00.2",
+      "section": "5.2 Comparison with Existing Tools",
+      "page": 11,
+      "char_offset": 53397,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer +QPG,SQLsmith GandSQLsmith all achieve high query validity in their own tests, with SQLsmith Cbeing an exception.",
+      "context_before": "162024020406080100120 (o)MariaDB: code coverage (k) 04812162024020406080100 (p)MariaDB: query validity (%) 0481216202401234 (q)SQLite: detected bugs 048121620240.00.51.01.52.02.53.03.5 (r)SQLite: grammar edges (k) 0481216202401020304050 (s)SQLite: code coverage (k) 048121620020406080100 (t)SQLite: query validity (%) Figure 2: Evaluation of different testing tools on MySQL ,CockroachDB ,TiDB,MariaDB and SQLite. ure 2t.ParserFuzz, along with other mutation-based fuzzing tools, generally has a lower query correctness rate compared to generation-based tools that rely on hand-written templates.",
+      "context_after": "However, these generationbased tools lack the flexibility to produce diverse query statements, because all the generated queries patterns must behand-written by the developers, making the process laborintensive. Therefore, ParserFuzz can find the highest number of bugs by generating more diverse queries and saturating all the grammar rules defined for the parsers. 11 Overall, ParserFuzz can find more memory errors than other testing tools, because it thoroughly examines all the grammar rules defined in the parser, and can reach the highest number of grammar coverage upon testing. Although mos",
+      "section": "5.2 Comparison with Existing Tools",
+      "page": 11,
+      "char_offset": 54944,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer +QPGcan detect 11 out of 50.",
+      "context_before": "that crashes TiDB, which aims to recover a table from a non-existing DDLJOB ID. infinite resources can be allocated, the columns of Squirrel, SQLsmith ,SQLsmith C,SQLsmith GandSQLancer +QPGin Table 1 and Table 2 indicate whether each bug could theoretically be detected by referenced tools. Given the input corpus and Internal Representation (IR), Squirrel can only detect 8 out of 37 bugs that ParserFuzz found.SQLsmith can detect half of the bugs reported by ParserFuzz (3 out of 6). SQLsmith Ccan detects 13 out of 37 bugs from CockroachDB .SQLsmith Gcan detect none of the bugs from ParserFuzz .",
+      "context_after": "The diverse syntaxes enable ParserFuzz to explore more interesting features from the DBMSs, and trigger more interesting bugs that are overlooked by these baseline tools. Next, we present two case studies to demonstrate the uniqueness of bugs detected by ParserFuzz. One-line query that crashes TiDB. Listing 12 presents a unique bug from TiDB. The PoC is surprisingly simple, consisting of just one line of SQL query. But the simple PoC crashes the TiDB query executor, and results in an immediate loss of connection between the TiDB server and client. The PoC attempts to recover a table that had",
+      "section": "5.4 Contribution of Diverse Syntax Features",
+      "page": 12,
+      "char_offset": 58980,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "describes_as_state_of_the_art": [
+        "M12"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:15:22Z",
+    "is_model_written": true,
+    "summary": "ParserFuzz generates SQL by extracting grammar rules directly from a DBMS's own built-in syntax definition files. The motivation is that existing tools exercise only a small subset of the syntax elements in DBMS-specific dialects, leaving many features untested. Without any input corpus, ParserFuzz saturates the grammar features of the system under test and uses code coverage as feedback to guide mutation and combine features from different rules. It found 81 previously unknown bugs across five popular DBMSs, all confirmed and 34 fixed.",
+    "narrative": "SQLancer, including its QPG configuration, is the generation-based tool ParserFuzz argues against: because SQLancer restricts itself to queries matching its oracles' patterns, the paper says it lacks the diversity needed to reach memory corruption bugs, which is the gap grammar-rule traversal fills.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "background",
+      "M5": "definition",
+      "M6": "definition",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "definition",
+      "M10": "motivation",
+      "M11": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is described as a tool of a different kind; nothing shown says ParserFuzz builds on it."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Grammar-rule traversal is presented as an alternative to oracle-pattern generation, not an extension of it."
+      },
+      "compares_with": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The paper's abstract claims to outperform existing tools, but the mentions read describe SQLancer's design rather than reporting a head-to-head result."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "A pattern fired on M12, which was outside the mentions read; those read call SQLancer+QPG a representative generation-based tool."
+      }
+    },
+    "disagreements": [],
+    "unresolved": [
+      "Whether a head-to-head comparison against SQLancer is reported: the evaluation mentions were not among those read."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2503_17322.json b/_data/papers/paper_arxiv_2503_17322.json
new file mode 100644
index 0000000..cc5c412
--- /dev/null
+++ b/_data/papers/paper_arxiv_2503_17322.json
@@ -0,0 +1,517 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:06:22Z",
+  "paper": {
+    "id": "paper:arxiv:2503.17322",
+    "title": "QITE: Assembly-Level, Cross-Platform Testing of Quantum Computing Platforms",
+    "authors": [
+      "Matteo Paltenghi",
+      "Michael Pradel"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2503.17322",
+    "s2_paper_id": "4135ce8f1c949407b786fbdf47a11f532126d257",
+    "url": "https://arxiv.org/abs/2503.17322",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2503.17322",
+      "retrieved_at": "2026-09-07T17:06:22Z",
+      "chars": 71520,
+      "content_sha256": "sha256:e03caedb514c3351a31414b5560bfe3bffd62119d7b108627a8dbf4b3b001ecc"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1807
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 9877
+      },
+      {
+        "number": "2.1",
+        "title": "Quantum Software and QASM",
+        "start": 9890
+      },
+      {
+        "number": "2.2",
+        "title": "Program Transformations",
+        "start": 11253
+      },
+      {
+        "number": "3",
+        "title": "The QITE Approach",
+        "start": 12492
+      },
+      {
+        "number": "3.1",
+        "title": "Problem Statement",
+        "start": 12608
+      },
+      {
+        "number": "3.2",
+        "title": "Overview",
+        "start": 13728
+      },
+      {
+        "number": "3.3",
+        "title": "Program Generation",
+        "start": 14924
+      },
+      {
+        "number": "3.4",
+        "title": "Import–Transform–Export Loop",
+        "start": 16992
+      },
+      {
+        "number": "4",
+        "title": "Evaluation",
+        "start": 27085
+      },
+      {
+        "number": "4.1",
+        "title": "RQ1: Bug Detection",
+        "start": 28806
+      },
+      {
+        "number": "4.2",
+        "title": "RQ2: Impact of ITE Itearations",
+        "start": 35609
+      },
+      {
+        "number": "4.3",
+        "title": "RQ3: Comparison with Prior Work",
+        "start": 41165
+      },
+      {
+        "number": "4.4",
+        "title": "RQ4: Efficiency of QITE",
+        "start": 45649
+      },
+      {
+        "number": "5",
+        "title": "Threats to Validity",
+        "start": 48355
+      },
+      {
+        "number": "6",
+        "title": "Related Work",
+        "start": 49499
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 55010
+      },
+      {
+        "number": "8",
+        "title": "Data Availability",
+        "start": 55764
+      },
+      {
+        "number": "C",
+        "title": "Huerta Alderete, N. H. Nguyen, Q. Wang, A. Maksymov, Y. Nam, M. Cetina,",
+        "start": 71021
+      },
+      {
+        "number": "N",
+        "title": "M. Linke, M. Hafezi, and C. Monroe. 2022. Cross-Platform Comparison of",
+        "start": 71096
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2025. Qir-Alliance/Qir-Spec. QIR Alliance.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Rui Abreu, João Paulo Fernandes, Luis Llana, and Guilherme Tavares. 2022. Metamorphic Testing of Oracle Quantum Programs. In 2022 IEEE/ACM 3rd International Workshop on Quantum Software Engineering (Q-SE). 16–23. doi:10.1145/ 3528230.3529189",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Ville Bergholm, Josh Izaac, Maria Schuld, Christian Gogolin, M. Sohaib Alam, Shahnawaz Ahmed, Juan Miguel Arrazola, Carsten Blank, Alain Delgado, Soran Jahangiri, Keri McKiernan, Johannes Jakob Meyer, Zeyue Niu, Antal Száva, and Nathan Killoran. 2020. PennyLane: Automatic Differentiation of Hybrid QuantumClassical Computations. arXiv:1811.04968 [physics, physics:quant-ph] (Feb. 2020). arXiv:1811.0",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Junjie Chen, Jibesh Patra, Michael Pradel, Yingfei Xiong, Hongyu Zhang, Dan Hao, and Lu Zhang. 2020. A Survey of Compiler Testing. Comput. Surveys 53, 1 (May 2020), 1–36. doi:10.1145/3363562",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Qihong Chen, Rúben Câmara, José Campos, André Souto, and Iftekhar Ahmed. 2023. The Smelly Eight: An Empirical Study on the Prevalence of Code Smells in Quantum Computing-Artifact. Software Engineering (ICSE) (Jan. 2023). doi:10.5281/ZENODO.7556360",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "T. Y. Chen, S. C. Cheung, and S. M. Yiu. 2020. Metamorphic Testing: A New Approach for Generating Next Test Cases. doi:10.48550/arXiv.2002.12543 arXiv:2002.12543 [cs]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Yu-Fang Chen, Kai-Min Chung, Ondřej Lengál, Jyun-Ao Lin, and Wei-Lun Tsai. 2023. AutoQ: An Automata-Based Quantum Circuit Verifier. In Computer Aided Verification (Lecture Notes in Computer Science), Constantin Enea and Akash Lal (Eds.). Springer Nature Switzerland, Cham, 139–153. doi:10.1007/978-3-03137709-9_7",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Yu-Fang Chen, Kai-Min Chung, Ondřej Lengál, Jyun-Ao Lin, Wei-Lun Tsai, and DiDe Yen. 2023. An Automata-Based Framework for Verification and Bug Hunting in Quantum Circuits. Proceedings of the ACM on Programming Languages 7, PLDI (June 2023), 156:1218–156:1243. doi:10.1145/3591270",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Bob Coecke and Ross Duncan. 2011. Interacting Quantum Observables: Categorical Algebra and Diagrammatics. New Journal of Physics 13, 4 (April 2011), 043016. doi:10.1088/1367-2630/13/4/043016",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Alexander Cowtan, Silas Dilkes, Ross Duncan, Alexandre Krajenbrink, Will Simmons, and Seyon Sivarajah. 2019. On the Qubit Routing Problem. In 14th Conference on the Theory of Quantum Computation, Communication and Cryptography (TQC 2019) (Leibniz International Proceedings in Informatics (LIPIcs), Vol. 135), Wim van Dam and Laura Mančinska (Eds.). Schloss Dagstuhl – Leibniz-Zentrum für Informatik, ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Andrew Cross, Ali Javadi-Abhari, Thomas Alexander, Niel De Beaudrap, Lev S. Bishop, Steven Heidel, Colm A. Ryan, Prasahnt Sivarajah, John Smolin, Jay M. Gambetta, and Blake R. Johnson. 2022. OpenQASM 3: A Broader and Deeper Quantum Assembly Language. ACM Transactions on Quantum Computing 3, 3 (Sept. 2022), 12:1–12:50. doi:10.1145/3505636",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Andrew W. Cross, Lev S. Bishop, John A. Smolin, and Jay M. Gambetta. 2017. Open Quantum Assembly Language. arXiv:1707.03429 [quant-ph] (July 2017). arXiv:1707.03429 [quant-ph]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Karine Even-Mendoza, Arindam Sharma, Alastair F. Donaldson, and Cristian Cadar. 2023. GrayC: Greybox Fuzzing of Compilers and Analysers for C. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 1219–1231. doi:10.1145/3597926.3598130",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Yipeng Huang and Margaret Martonosi. 2019. Statistical Assertions for Validating Patterns and Finding Bugs in Quantum Programs. In Proceedings of the 46th International Symposium on Computer Architecture (ISCA ’19). Association for Computing Machinery, New York, NY, USA, 541–553. doi:10.1145/3307650.3322213",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Purvish Jajal, Wenxin Jiang, Arav Tewari, Erik Kocinare, Joseph Woo, Anusha Sarraf, Yung-Hsiang Lu, George K. Thiruvathukal, and James C. Davis. 2024. Interoperability in Deep Learning: A User Survey and Failure Analysis of ONNX Model Converters. In Proceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA 2024). Association for Computing Machinery, New Yor",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Ali Javadi-Abhari, Matthew Treinish, Kevin Krsulich, Christopher J. Wood, Jake Lishman, Julien Gacon, Simon Martiel, Paul D. Nation, Lev S. Bishop, Andrew W. Cross, Blake R. Johnson, and Jay M. Gambetta. 2024. Quantum Computing with Qiskit. doi:10.48550/arXiv.2405.08810 arXiv:2405.08810 [quant-ph]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Ali JavadiAbhari, Shruti Patil, Daniel Kudrow, Jeff Heckey, Alexey Lvov, Frederic T. Chong, and Margaret Martonosi. 2014. ScaffCC: A Framework for Compilation and Analysis of Quantum Computing Programs. In Proceedings of the 11th ACM Conference on Computing Frontiers (CF ’14). Association for Computing Machinery, New York, NY, USA, 1–10. doi:10.1145/2597917.2597939",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Chan Gu Kang, Joonghoon Lee, and Hakjoo Oh. 2024. Statistical Testing of Quantum Programs via Fixed-Point Amplitude Amplification. Artifact for Paper \"Statistical Testing of Quantum Programs via Fixed-Point Amplitude Amplification\" in OOPSLA 2024 8, OOPSLA2 (Oct. 2024), 276:140–276:164. doi:10.1145/3689716",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Maximilian Kaul, Alexander Küchler, and Christian Banse. 2023. A Uniform Representation of Classical and Quantum Source Code for Static Code Analysis. doi:10.48550/arXiv.2308.06113 arXiv:2308.06113 [cs]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Aleks Kissinger and John van de Wetering. 2020. PyZX: Large Scale Automated Diagrammatic Reasoning. Electronic Proceedings in Theoretical Computer Science 318 (May 2020), 229–241. doi:10.4204/EPTCS.318.14 arXiv:1904.04735",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Gushu Li, Li Zhou, Nengkun Yu, Yufei Ding, Mingsheng Ying, and Yuan Xie. 2020. Projection-Based Runtime Assertions for Testing and Debugging Quantum Programs. Proceedings of the ACM on Programming Languages 4, OOPSLA (Nov. 2020), 150:1–150:29. doi:10.1145/3428218",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Ji Liu, Gregory T. Byrd, and Huiyang Zhou. 2020. Quantum Circuits for Dynamic Runtime Assertions in Quantum Computation. In Proceedings of the Twenty-Fifth International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS ’20). Association for Computing Machinery, New York, NY, USA, 1017–1030. doi:10.1145/3373376.3378488",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Jiawei Liu, Jinkun Lin, Fabian Ruffy, Cheng Tan, Jinyang Li, Aurojit Panda, and Lingming Zhang. 2023. NNSmith: Generating Diverse and Valid Test Cases for Deep Learning Compilers. In Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2 (ASPLOS 2023). Association for Computing Machinery, New York, NY, USA, 530–543. d",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Ji Liu and Huiyang Zhou. 2021. Systematic Approaches for Precise and Approximate Quantum State Runtime Assertion. In 2021 IEEE International Symposium on High-Performance Computer Architecture (HPCA). 179–193. doi:10.1109/ HPCA51647.2021.00025",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Junjie Luo, Pengzhan Zhao, Zhongtao Miao, Shuhan Lan, and Jianjun Zhao. 2022. A Comprehensive Study of Bug Fixes in Quantum Programs. In (SANER). 1239–1246. doi:10.1109/SANER53432.2022.00147",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Weisi Luo, Dong Chai, Xiaoyue Run, Jiang Wang, Chunrong Fang, and Zhenyu Chen. 2021. Graph-Based Fuzz Testing for Deep Learning Inference Engines. In Proceedings of the 43rd International Conference on Software Engineering (ICSE ’21). IEEE Press, Madrid, Spain, 288–299. doi:10.1109/ICSE43902.2021.00037",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Haoyang Ma, Qingchao Shen, Yongqiang Tian, Junjie Chen, and Shing-Chi Cheung. 2023. Fuzzing Deep Learning Compilers with HirGen. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 248–260. doi:10.1145/3597926.3598053",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Sara Ayman Metwalli and Rodney Van Meter. 2023. Cirquo: A Suite For Testing and Debugging Quantum Programs. doi:10.48550/arXiv.2311.18202 arXiv:2311.18202 [quant-ph]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Andriy Miranskyy, Lei Zhang, and Javad Doliskani. 2021. On Testing and Debugging Quantum Software. arXiv:2103.09172 [quant-ph] (March 2021). arXiv:2103.09172 [quant-ph]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Yusei Mori, Hideaki Hakoshima, Kyohei Sudo, Toshio Mori, Kosuke Mitarai, and Keisuke Fujii. 2024. Quantum Circuit Unoptimization. doi:10.48550/arXiv.2311. 03805 arXiv:2311.03805 [quant-ph]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Juan M. Murillo, Jose Garcia-Alonso, Enrique Moguel, Johanna Barzen, Frank Leymann, Shaukat Ali, Tao Yue, Paolo Arcaini, Ricardo Pérez-Castillo, Ignacio García Rodríguez de Guzmán, Mario Piattini, Antonio Ruiz-Cortés, Antonio Brogi, Jianjun Zhao, Andriy Miranskyy, and Manuel Wimmer. 2025. Quantum Software Engineering: Roadmap and Challenges Ahead. ACM Trans. Softw. Eng. Methodol. (Jan. 2025). doi:",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Matteo Paltenghi and Michael Pradel. 2022. Bugs in Quantum Computing Platforms: An Empirical Study. Proceedings of the ACM on Programming Languages 6, OOPSLA1 (April 2022), 86:1–86:27. doi:10.1145/3527330",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Matteo Paltenghi and Michael Pradel. 2023. MorphQ: Metamorphic Testing of the Qiskit Quantum Computing Platform. In Proceedings of the 45th International Conference on Software Engineering (ICSE ’23). IEEE Press, Melbourne, Victoria, Australia, 2413–2424. doi:10.1109/ICSE48619.2023.00202",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Matteo Paltenghi and Michael Pradel. 2024. Analyzing Quantum Programs with LintQ: A Static Analysis Framework for Qiskit. Proceedings of the ACM on Software Engineering 1, FSE (July 2024), 95:2144–95:2166. doi:10.1145/3660802",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Matteo Paltenghi and Michael Pradel. 2024. A Survey on Testing and Analysis of Quantum Software. doi:10.48550/arXiv.2410.00650 arXiv:2410.00650",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Tom Peham, Lukas Burgholzer, and Robert Wille. 2022. Equivalence Checking of Quantum Circuits With the ZX-Calculus. IEEE Journal on Emerging and Selected Topics in Circuits and Systems 12, 3 (Sept. 2022), 662–675. doi:10.1109/JETCAS. 2022.3202204",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "Alireza Shafaei, Mehdi Saeedi, and Massoud Pedram. 2014. Qubit Placement to Minimize Communication Overhead in 2D Quantum Architectures. In 2014 19th Asia and South Pacific Design Automation Conference (ASP-DAC). 495–500. doi:10.1109/ASPDAC.2014.6742940 Conference’17, July 2017, Washington, DC, USA Matteo Paltenghi and Michael Pradel",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Emin Gün Sirer and Brian N. Bershad. 2000. Using Production Grammars in Software Testing. SIGPLAN Not. 35, 1 (Dec. 2000), 1–13. doi:10.1145/331963.331965",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Seyon Sivarajah, Silas Dilkes, Alexander Cowtan, Will Simmons, Alec Edgington, and Ross Duncan. 2020. T|ket ⟩: A Retargetable Compiler for NISQ Devices. Quantum Science and Technology 6, 1 (Nov. 2020), 014003. doi:10.1088/20589565/ab8e92",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Robert S. Smith, Michael J. Curtis, and William J. Zeng. 2017. A Practical Quantum Instruction Set Architecture. arXiv:1608.03355 [quant-ph] (Feb. 2017). arXiv:1608.03355 [quant-ph]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Swamit S. Tannu and Moinuddin K. Qureshi. 2019. Not All Qubits Are Created Equal: A Case for Variability-Aware Policies for NISQ-Era Quantum Computers. In Proceedings of the Twenty-Fourth International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS ’19). Association for Computing Machinery, New York, NY, USA, 987–999. doi:10.1145/3297858.3304007",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Jiyuan Wang, Qian Zhang, Guoqing Harry Xu, and Miryung Kim. 2021. QDiff: Differential Testing of Quantum Software Stacks. In 2021 36th IEEE/ACM International Conference on Automated Software Engineering (ASE). 692–704. doi:10.1109/ASE51524.2021.9678792",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Zan Wang, Ming Yan, Junjie Chen, Shuang Liu, and Dongdi Zhang. 2020. Deep Learning Library Testing via Effective Model Generation. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 788–799. doi:10.1145/3368089.3409761",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Robert Wille, Lukas Burgholzer, and Alwin Zulehner. 2019. Mapping Quantum Circuits to IBMQX Architectures Using the Minimal Number of SWAP and H Operations. In Proceedings of the 56th Annual Design Automation Conference 2019 (DAC ’19). Association for Computing Machinery, New York, NY, USA, 1–6. doi:10.1145/3316781.3317859",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Dominik Winterer and Zhendong Su. 2024. Validating SMT Solvers for Correctness and Performance via Grammar-Based Enumeration. Proc. ACM Program. Lang. 8, OOPSLA2 (Oct. 2024), 355:2378–355:2401. doi:10.1145/3689795",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Chunqiu Steven Xia, Matteo Paltenghi, Jia Le Tian, Michael Pradel, and Lingming Zhang. 2024. Fuzz4All: Universal Fuzzing with Large Language Models. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (ICSE ’24). Association for Computing Machinery, New York, NY, USA, 1–13.doi:10.1145/3597503.3639121",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Mingkuan Xu, Zikun Li, Oded Padon, Sina Lin, Jessica Pointing, Auguste Hirth, Henry Ma, Jens Palsberg, Alex Aiken, Umut A. Acar, and Zhihao Jia. 2022. Quartz: Superoptimization of Quantum Circuits. In Proceedings of the 43rd ACMSIGPLAN International Conference on Programming Language Design and Implementation (PLDI 2022). Association for Computing Machinery, New York, NY, USA, 625–640. doi:10.1145",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and Understanding Bugs in C Compilers. ACMSIGPLAN Notices 46, 6 (June 2011), 283–294. doi:10.1145/1993316.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Ed Younis, Costin C. Iancu, Wim Lavrijsen, Marc Davis, and Ethan Smith. 2021. Berkeley Quantum Synthesis Toolkit (BQSKit) V1. Technical Report Berkeley Quantum Synthesis Toolkit. Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States). doi:10.11578/dc.20210603.2",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Andreas Zeller. 2002. Isolating Cause-Effect Chains from Computer Programs. ACMSIGSOFT Software Engineering Notes 27, 6 (Nov. 2002), 1–10. doi:10.1145/ 605466.605468",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Chi Zhang, Ari B. Hayes, Longfei Qiu, Yuwei Jin, Yanhao Chen, and Eddy Z. Zhang. 2021. Time-Optimal Qubit Mapping. In Proceedings of the 26th ACM International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS ’21). Association for Computing Machinery, New York, NY, USA, 360–374. doi:10.1145/3445814.3446706",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Jianjun Zhao. 2021. Quantum Software Engineering: Landscapes and Horizons. doi:10.48550/arXiv.2007.07047 arXiv:2007.07047",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Pengzhan Zhao, Xiongfei Wu, Zhuo Li, and Jianjun Zhao. 2023. QChecker: Detecting Bugs in Quantum Programs via Static Analysis. doi:10.48550/arXiv. 2304.04387 arXiv:2304.04387 [cs]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "D. Zhu, Z. P. Cian, C. Noel, A. Risinger, D. Biswas, L. Egan, Y. Zhu, A. M. Green, C. Huerta Alderete, N. H. Nguyen, Q. Wang, A. Maksymov, Y. Nam, M. Cetina, N. M. Linke, M. Hafezi, and C. Monroe. 2022. Cross-Platform Comparison of Arbitrary Quantum States. Nature Communications 13, 1 (Nov. 2022), 6620. doi:10.1038/s41467-022-34279-5",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Alwin Zulehner, Alexandru Paler, and Robert Wille. 2018. Efficient Mapping of Quantum Circuits to the IBMQX Architectures. In 2018 Design, Automation & Test in Europe Conference & Exhibition (DATE). 1135–1138. doi:10.23919/DATE. 2018.8342181",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 38]",
+      "technique": "pqs",
+      "sentence": "Grammar-based generators, similar to our approach, have been successfully applied to test different compilers and similar software, such as the Java Virtual Machine [ 40], C compilers [ 13,50], SMT solvers [ 47], database engines [ 38], and deep learning software infrastructure [ 23,26,27,45].",
+      "context_before": "ay translation, unlike our iterative ITE process. Even with standardized IRs, interoperability challenges persist. For example, a study reports that 75% of defects occur during the conversion of operators in deep learning models, with 33% of these defects leading to semantically inequivalent models [ 15]. Our work addresses similar challenges in quantum computing and its QASM representation, an area not explored by prior work. Compiler and Runtime Engine Testing. Our work shares similarities with compiler testing [ 4], as quantum computing platforms conceptually resemble traditional compilers.",
+      "context_after": "However, existing compiler testing techniques do not address the unique challenges of cross-platform testing in quantum computing, which our work tackles directly. 7 Conclusion We introduce QITE, the first cross-platform testing technique for quantum platforms. QITE uses quantum assembly code for program generation and equivalence checks, enabling cross-platform comparisons. Evaluation on four platforms revealed 17 bugs, with 14 confirmed or fixed. The core component of QITE, our novel ITE process, generates diverse, complex programs that exercise code not tested by a state-of-the-art techniq",
+      "section": "6 Related Work",
+      "page": 10,
+      "char_offset": 54550,
+      "cited_reference": {
+        "number": 38,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "QITE tests quantum computing platforms across platform boundaries using QASM as the common ground. Its ITE process generates equivalent quantum programs by importing assembly into a platform's representation, transforming it through that platform's optimisation and gate conversion, and exporting it back. A crash oracle catches failures during transformation and an equivalence oracle checks the resulting programs really are equivalent. Across four platforms it revealed 17 bugs.",
+    "narrative": "One citation in related work, noting that grammar-based generators similar to QITE's approach have been applied successfully to compilers and similar software, SQLancer among the examples.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2506_02617.json b/_data/papers/paper_arxiv_2506_02617.json
new file mode 100644
index 0000000..f9a8a21
--- /dev/null
+++ b/_data/papers/paper_arxiv_2506_02617.json
@@ -0,0 +1,479 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:25:14Z",
+  "paper": {
+    "id": "paper:arxiv:2506.02617",
+    "title": "Toward Understanding Bugs in Vector Database Management Systems",
+    "authors": [
+      "Yinglin Xie",
+      "Xinyi Hou",
+      "Yanjie Zhao",
+      "Shenao Wang",
+      "Kai Chen",
+      "Haoyu Wang"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2506.02617",
+    "s2_paper_id": "add37b9f02f91d837800030a5f866052e76756f2",
+    "url": "https://arxiv.org/abs/2506.02617",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2506.02617",
+      "retrieved_at": "2026-09-08T15:25:14Z",
+      "chars": 60794,
+      "content_sha256": "sha256:207998b73537007af7bc84556b721052fca6a51a4a1f258558bf4e66806336d8"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1594
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND AND RELATED WORK",
+        "start": 6170
+      },
+      {
+        "number": "B",
+        "title": "Reliability of VDBMSs",
+        "start": 9738
+      },
+      {
+        "number": "III",
+        "title": "METHODOLOGY",
+        "start": 12568
+      },
+      {
+        "number": "A",
+        "title": "Data Collection",
+        "start": 12761
+      },
+      {
+        "number": "B",
+        "title": "Classification and Labeling",
+        "start": 15629
+      },
+      {
+        "number": "A",
+        "title": "RQ1: Symptoms.",
+        "start": 19050
+      },
+      {
+        "number": "B",
+        "title": "RQ2: Root Causes.",
+        "start": 23898
+      },
+      {
+        "number": "C",
+        "title": "RQ3: Fix Strategies.",
+        "start": 39681
+      },
+      {
+        "number": "A",
+        "title": "RQ4: Analysis across VDBMSs.",
+        "start": 44425
+      },
+      {
+        "number": "A",
+        "title": "Implications",
+        "start": 48450
+      },
+      {
+        "number": "B",
+        "title": "Limitations",
+        "start": 50843
+      },
+      {
+        "number": "R",
+        "title": "Yang, K. Xu, Y. Hu, J. Wei, and T. Huang, “Understanding",
+        "start": 53051
+      },
+      {
+        "number": "J",
+        "title": "Lafferty, C. Williams, J. Shawe-Taylor, R. Zemel, and",
+        "start": 53490
+      },
+      {
+        "number": "A",
+        "title": "Culotta, Eds., vol. 23. Curran Associates, Inc., 2010.",
+        "start": 53547
+      },
+      {
+        "number": "M",
+        "title": "Lomeli, L. Hosseini, and H. J ´egou, “The faiss library,” 2024.",
+        "start": 54088
+      },
+      {
+        "number": "J",
+        "title": "Luo, F. Liu, Z. Cao, Y. Qiao, T. Wang, B. Tang, and C. Xie, “Manu:",
+        "start": 54369
+      },
+      {
+        "number": "D",
+        "title": "Isayan, M. Harutyunyan, T. Hakobyan, I. Stranic, and D. Buniatyan,",
+        "start": 54620
+      },
+      {
+        "number": "J",
+        "title": "Gu, R. Jiang, Y. Wei, and C. Xie, “Milvus: A purpose-built vector",
+        "start": 58368
+      },
+      {
+        "number": "H",
+        "title": "Wang, “Towards reliable vector database management systems:",
+        "start": 59531
+      },
+      {
+        "number": "Y",
+        "title": "Cai, “Analyticdb-v: a hybrid analytical engine towards query",
+        "start": 59820
+      },
+      {
+        "number": "H",
+        "title": "Yang, H. Xiao, X. Li, F. Yang, X. Feng, L. Hu, H. Li, K. Gai,",
+        "start": 60561
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Annoy, “Annoy,” https://github.com/spotify/annoy, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "S. Arya and D. M. Mount, “Approximate nearest neighbor queries in fixed dimensions,” in Proceedings of the Fourth Annual ACM-SIAM Symposium on Discrete Algorithms, ser. SODA ’93. USA: Society for Industrial and Applied Mathematics, 1993, p. 271–280.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Attu, “Attu,” https://zilliz.com.cn/attu, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "J. Chen, Y. Liang, Q. Shen, J. Jiang, and S. Li, “Toward understanding deep learning framework bugs,” 2024. [Online]. Available: https://arxiv.org/abs/2203.04026",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Chroma, “Chroma,” https://github.com/chroma-core/chroma, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "T. Cover and P. Hart, “Nearest neighbor pattern classification,” IEEE Transactions on Information Theory, vol. 13, no. 1, pp. 21–27, 1967.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Z. Cui, W. Dou, Y. Gao, D. Wang, J. Song, Y. Zheng, T. Wang, R. Yang, K. Xu, Y. Hu, J. Wei, and T. Huang, “Understanding transaction bugs in database systems,” in Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, ser. ICSE ’24. New York, NY, USA: Association for Computing Machinery, 2024. [Online]. Available: https://doi.org/10.1145/3597503.3639207",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "A. Dhesi and P. Kar, “Random projection trees revisited,” in Advances in Neural Information Processing Systems, J. Lafferty, C. Williams, J. Shawe-Taylor, R. Zemel, and A. Culotta, Eds., vol. 23. Curran Associates, Inc., 2010. [Online]. Available: https://proceedings.neurips.cc/paper files/paper/ 2010/file/3def184ad8f4755ff269862ea77393dd-Paper.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "J. Dong, X. Chen, M. Zhang, X. Yang, S. Chen, X. Li, and X. Wang, “Partially relevant video retrieval,” in Proceedings of the 30th ACM International Conference on Multimedia, ser. MM ’22. ACM, Oct. 2022, p. 246–257. [Online]. Available: http://dx.doi.org/10.1145/3503161.3547976",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "M. Douze, A. Guzhva, C. Deng, J. Johnson, G. Szilvasy, P.-E. Mazar ´e, M. Lomeli, L. Hosseini, and H. J ´egou, “The faiss library,” 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "R. Gray, “Vector quantization,” IEEEASSP Magazine, vol. 1, no. 2, pp. 4–29, 1984.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "gRPC, “gRPC,” https://github.com/grpc/grpc, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "R. Guo, X. Luan, L. Xiang, X. Yan, X. Yi, J. Luo, Q. Cheng, W. Xu, J. Luo, F. Liu, Z. Cao, Y. Qiao, T. Wang, B. Tang, and C. Xie, “Manu: A cloud native vector database management system,” 2022. [Online]. Available: https://arxiv.org/abs/2206.13843",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "S. Hambardzumyan, A. Tuli, L. Ghukasyan, F. Rahman, H. Topchyan, D. Isayan, M. Harutyunyan, T. Hakobyan, I. Stranic, and D. Buniatyan, “Deep lake: a lakehouse for deep learning,” 2023. [Online]. Available: https://www.cidrdb.org/cidr2023/papers/p69-buniatyan.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "H. J ´egou, M. Douze, and C. Schmid, “Product quantization for nearest neighbor search,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 33, no. 1, pp. 117–128, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "LanceDB, “Lancedb,” https://github.com/lancedb/lancedb, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "LangChain, “Langchain,” https://www.langchain.com/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "W. Liu, S. Mondal, and T.-H. Chen, “An empirical study on the characteristics of database access bugs in java applications,” 2024. [Online]. Available: https://arxiv.org/abs/2405.15008",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "LlamaIndex, “Llamaindex,” https://docs.llamaindex.ai/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "J. Luan, “The next stop for vector databases: 8 predictions for 2023,” https://zilliz.com/blog/ the-next-stop-for-vector-databases-8-predictions-for-2023, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Y. A. Malkov and D. A. Yashunin, “Efficient and robust approximate nearest neighbor search using hierarchical navigable small world graphs,” IEEE transactions on pattern analysis and machine intelligence, vol. 42, no. 4, pp. 824–836, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Marqo, “Marqo,” https://github.com/marqo-ai/marqo, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Y. Matsui, Y. Uchida, H. J ´egou, and S. Satoh, “A survey of product quantization,” ITE Transactions on Media Technology and Applications, vol. 6, no. 1, pp. 2–10, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "M. Muja and D. Lowe, “Fast approximate nearest neighbors with automatic algorithm configuration.” vol. 1, 01 2009, pp. 331–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "J. J. Pan, J. Wang, and G. Li, “Survey of vector database management systems,” The VLDB Journal, vol. 33, no. 5, pp. 1591–1615, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "pgvecto.rs, “pgvecto.rs,” https://github.com/tensorchord/pgvecto.rs, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Qdrant, “Qdrant,” https://github.com/qdrant/qdrant, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "L. Quan, Q. Guo, X. Xie, S. Chen, X. Li, and Y. Liu, “Towards understanding the faults of javascript-based deep learning systems,” inProceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, ser. ASE ’22. ACM, Oct. 2022, p. 1–13. [Online]. Available: http://dx.doi.org/10.1145/3551349.3560427",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ser. ESEC/FSE ’20. ACM, Nov. 2020, p. 1140–1152. [Online]. Available: http://dx.doi.org/10.1145/3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 30,
+      "text": "C. Silpa-Anan and R. Hartley, “Optimised kd-trees for fast image descriptor matching,” in and Pattern Recognition, 2008, pp. 1–8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "J. Tagliabue and C. Greco, “(vector) space is not the final frontier: Product search as program synthesis,” 2023. [Online]. Available: https://arxiv.org/abs/2304.11473",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "txtai, “txtai,” https://github.com/neuml/txtai, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "A. Vardanian, “USearch by Unum Cloud,” Oct. 2023. [Online]. Available: https://github.com/unum-cloud/usearch",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Vespa, “Vespa,” https://github.com/vespa-engine/vespa, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "S. M. Vieira, U. Kaymak, and J. M. C. Sousa, “Cohen’s kappa coefficient as a performance measure for feature selection,” in International Conference on Fuzzy Systems, 2010, pp. 1–8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "V oyager, “V oyager,” https://github.com/spotify/voyager, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "J. Wang, X. Yi, R. Guo, H. Jin, P. Xu, S. Li, X. Wang, X. Guo, C. Li, X. Xu et al., “Milvus: A purpose-built vector data management system,” inProceedings of the 2021 International Conference on Management of Data, 2021, pp. 2614–2627.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "J. Wang, X. Yi, R. Guo, H. Jin, P. Xu, S. Li, X. Wang, X. Guo, C. Li, X. Xu, K. Yu, Y. Yuan, Y. Zou, J. Long, Y. Cai, Z. Li, Z. Zhang, Y. Mo, J. Gu, R. Jiang, Y. Wei, and C. Xie, “Milvus: A purpose-built vector data management system,” in Proceedings of the 2021 International Conference on Management of Data, ser. SIGMOD ’21. New York, NY, USA: Association for Computing Machinery, 2021, p. 2614–26",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "J. Wang, T. Zhang, J. Song, N. Sebe, and H. T. Shen, “A survey on learning to hash,” 2017. [Online]. Available: https: //arxiv.org/abs/1606.00185",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "M. Wang, X. Xu, Q. Yue, and Y. Wang, “A comprehensive survey and experimental comparison of graph-based approximate nearest neighbor search,” 2021. [Online]. Available: https://arxiv.org/abs/2101.12631",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "S. Wang, Y. Zhao, X. Hou, and H. Wang, “Large language model supply chain: A research agenda,” CoRR, vol. abs/2404.12736, 2024. [Online]. Available: https://doi.org/10.48550/arXiv.2404.12736",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "S. Wang, Y. Zhao, Z. Liu, Q. Zou, and H. Wang, “Sok: Understanding vulnerabilities in the large language model supply chain,” CoRR, vol. abs/2502.12497, 2025. [Online]. Available: https://doi.org/10.48550/arXiv.2502.12497",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "S. Wang, Y. Zhao, Y. Xie, Z. Liu, X. Hou, Q. Zou, and H. Wang, “Towards reliable vector database management systems: A software testing roadmap for 2030,” 2025. [Online]. Available: https://arxiv.org/abs/2502.20812",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Weaviate, “Weaviate,” https://github.com/weaviate/weaviate, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "C. Wei, B. Wu, S. Wang, R. Lou, C. Zhan, F. Li, and Y. Cai, “Analyticdb-v: a hybrid analytical engine towards query fusion for structured and unstructured data,” Proc. VLDB Endow., vol. 13, no. 12, p. 3152–3165, Aug. 2020. [Online]. Available: https://doi.org/10.14778/3415478.3415541",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "W. Wu, J. He, Y. Qiao, G. Fu, L. Liu, and J. Yu, “Hqann: Efficient and robust similarity search for hybrid queries with structured and unstructured constraints,” 2022. [Online]. Available: https://arxiv.org/abs/2207.07940",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Z. Wu, “Bhakti: A lightweight vector database management system for endowing large language models with semantic search capabilities andmemory,” 2025. [Online]. Available: https://arxiv.org/abs/2504.01553",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Q. Yu, X. Wang, S. Liu, Y. Bai, X. Yang, X. Wang, C. Meng, S. Wu, H. Yang, H. Xiao, X. Li, F. Yang, X. Feng, L. Hu, H. Li, K. Gai, and L. Zou, “Who you are matters: Bridging topics and social roles via llm-enhanced logical recommendation,” 2025. [Online]. Available: https://arxiv.org/abs/2505.10940",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 29,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ser. ESEC/FSE ’20. ACM, Nov. 2020, p. 1140–1152. [Online]. Available: http://dx.doi.org/10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "[29] further introduced the Non-Optimizing Reference Engine Construction (NoREC) method to detect optimization bugs in query engines.",
+      "context_before": "ding core infrastructure for storing, searching, and managing vector data. They enable similarity-based retrieval for tasks like recommendation, anomaly detection, and semantic search, and also support RAG by supplying external knowledge to improve LLM responses. B. Reliability of VDBMSs The reliability of traditional DBMSs has long been a central topic of research. Liu et al. [18] conducted a comprehensive analysis of 423 database access bugs across seven largescale Java applications, while Cui et al. [7] examined 140 transaction-related bugs in six widely used database systems. Rigger et al.",
+      "context_after": "These studies collectively reveal that DBMSs are susceptible to reliability issues stemming from diverse sources, including server-client interfacing, transaction handling, and query optimization. RQ4: Characters within and across VDBMSsClassification Method•Common types •Frequency across bugs Category Category Root causeRoot causeAspectIncludeVDBMSs§Ⅲ-A Data Collection VDBMS datasetFaiss Milvus Qdrant Chroma USearch …… Marqo Construct …… GitHub PRs Optimize ScreenInconsistentProgramming languages Architectures Other features Filter Closed KeywordsMergedCollect§Ⅲ-B Classification and Labeling",
+      "section": "B Reliability of VDBMSs",
+      "page": 2,
+      "char_offset": 10074,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "The first large-scale empirical study of defects in vector database management systems, which the authors argue traditional database reliability models do not fit because data representation, query mechanisms and architecture all differ. They manually analysed 1,671 bug-fix pull requests across 15 open-source vector databases and built a taxonomy by symptom, root cause and fix strategy, reporting 31 recurring fault patterns.",
+    "narrative": "One citation, introducing NoREC as a method for detecting optimization bugs in query engines, in a review of how relational DBMS reliability has been studied.",
+    "roles": {
+      "M1": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2509_10819.json b/_data/papers/paper_arxiv_2509_10819.json
new file mode 100644
index 0000000..ae91ed0
--- /dev/null
+++ b/_data/papers/paper_arxiv_2509_10819.json
@@ -0,0 +1,374 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:00:49Z",
+  "paper": {
+    "id": "paper:arxiv:2509.10819",
+    "title": "Arguzz: Testing zk VMs for Soundness and Completeness Bugs",
+    "authors": [
+      "Christoph Hochrainer",
+      "Valentin Wüstholz",
+      "Maria Christakis"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2509.10819",
+    "s2_paper_id": "98f3b2e558fdcfd3f05ceb53c370f6339f72e0d9",
+    "url": "https://arxiv.org/abs/2509.10819",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2509.10819",
+      "retrieved_at": "2026-09-07T17:00:49Z",
+      "chars": 83199,
+      "content_sha256": "sha256:eea4337c9e58a0134563a696d1d935d3a50105cf47ec2363c645c10cfe5a7f2e"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 30,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1499
+      },
+      {
+        "number": "2",
+        "title": "Overview",
+        "start": 11584
+      },
+      {
+        "number": "3",
+        "title": "Approach",
+        "start": 21149
+      },
+      {
+        "number": "3.1",
+        "title": "Circuit Generation",
+        "start": 23363
+      },
+      {
+        "number": "3.2",
+        "title": "Product-Program Generation",
+        "start": 24696
+      },
+      {
+        "number": "3.3",
+        "title": "VM Execution with Malicious Prover",
+        "start": 28524
+      },
+      {
+        "number": "4",
+        "title": "Experimental Evaluation",
+        "start": 37091
+      },
+      {
+        "number": "4.1",
+        "title": "Experimental Setup",
+        "start": 37574
+      },
+      {
+        "number": "4.2",
+        "title": "Experimental Results",
+        "start": 40070
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "gnark: A fast zk-SNARK library that offers a highlevel API to design circuits. https://docs.gnark. consensys.io.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Jolt.https://github.com/a16z/jolt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Nexus. https://github.com/nexus-xyz/ nexus-zkvm.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Noir.https://noir-lang.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "RISC Zero.https://github.com/risc0/risc0.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Mithun Acharya and Tao Xie. Mining API errorhandling specifications from source code. InFASE, volume 5503 of LNCS, pages 370–384. Springer, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Jean Arlat, Martine Aguera, Louis Amat, Yves Crouzet, Jean-Charles Fabre, Jean-Claude Laprie, Eliane Martins, and David Powell. Fault injection for dependability validation: A methodology and some applications.TSE, 16:166–182, 1990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Arasu Arun, Srinath Setty, and Justin Thaler. Jolt: SNARKs for virtual machines via lookups. Cryptology ePrint Archive, Paper 2023/1217, 2023. https: //eprint.iacr.org/2023/1217.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Earl T. Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. The oracle problem in software testing: A survey.TSE, 41:507–525, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Nils Bars, Moritz Schloegel, Tobias Scharnowski, Nico Schiller, and Thorsten Holz. Fuzztruction: Using fault injection-based fuzzing to leverage implicit domain knowledge. InSecurity, pages 1847–1864. USENIX, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Nils Bars, Moritz Schloegel, Nico Schiller, Lukas Bernhard, and Thorsten Holz. No peer, no cry: Network application fuzzing via fault injection. InCCS, pages 750–764. ACM, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Gilles Barthe, Juan Manuel Crespo, and César Kunz. Relational verification using product programs. InFM, volume 6664 of LNCS, pages 200–214. Springer, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Gilles Barthe, Pedro R. D’Argenio, and Tamara Rezk. Secure information flow by self-composition. InCSFW, pages 100–114. IEEE Computer Society, 2004.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Marta Bellés-Muñoz, Miguel Isabel, Jose Luis MuñozTapia, Albert Rubio, and Jordi Baylina Melé. Circom: A circuit description language for building zeroknowledge applications.Trans. Dependable Secur. Comput., 20:4733–4751, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Armin Biere, Alessandro Cimatti, Edmund M. Clarke, and Yunshan Zhu. Symbolic model checking without BDDs. InTACAS, volume 1579 of LNCS, pages 193– 207. Springer, 1999.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Junjie Chen, Jibesh Patra, Michael Pradel, Yingfei Xiong, Hongyu Zhang, Dan Hao, and Lu Zhang. A survey of compiler testing.Comput. Surv., 53:4:1–4:36, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Tsong Yueh Chen, S. C. Cheung, and Siu-Ming Yiu. Metamorphic testing: A new approach for generating next test cases. Technical Report HKUST–CS98–01, HKUST, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Maria Christakis, Patrick Emmisberger, Patrice Godefroid, and Peter Müller. A general framework for dynamic stub injection. InICSE, pages 586–596. IEEE Computer Society/ACM, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Jeffrey A. Clark and Dhiraj K. Pradhan. Fault injection: A method for validating computer-system dependability. Computer, 28:47–56, 1995.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Michael R. Clarkson and Fred B. Schneider. Hyperproperties. InCSF, pages 51–65. IEEE Computer Society, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Patrick Cousot and Radhia Cousot. Abstract interpretation: A unified lattice model for static analysis of programs by construction or approximation of fixpoints. In POPL, pages 238–252. ACM, 1977.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Weigang He, Peng Di, Mengli Ming, Chengyu Zhang, Ting Su, Shijie Li, and Yulei Sui. Finding and understanding defects in static analyzers by constructing automated oracles.PACMSE, 1:1656–1678, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Christoph Hochrainer, Anastasia Isychev, Valentin Wüstholz, and Maria Christakis. Fuzzing processing pipelines for zero-knowledge circuits. InCCS. ACM, 2025. to appear.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Mei-Chen Hsueh, Timothy K. Tsai, and Ravishankar K. Iyer. Fault injection techniques and tools.Computer, 30:75–82, 1997.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Jaewon Hur, Suhwan Song, Dongup Kwon, Eunjin Baek, Jangwoo Kim, and Byoungyoung Lee. DifuzzRTL: Differential fuzz testing to find CPU bugs. InS&P, pages 1286–1303. IEEE Computer Society, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Suman Jana, Yuan Kang, Samuel Roth, and Baishakhi Ray. Automatically detecting error handling bugs using error specifications. InSecurity, pages 345–362. USENIX, 2016. 29",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, Kangjie Lu, and Shi-Min Hu. Fuzzing error handling code using context-sensitive software fault injection. InSecurity, pages 2595–2612. USENIX, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "David Kaindlstorfer, Anastasia Isychev, Valentin Wüstholz, and Maria Christakis. Interrogation testing of program analyzers for soundness and precision issues. InASE, pages 319–330. ACM, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Rahul Kande, Addison Crump, Garrett Persyn, Patrick Jauernig, Ahmad-Reza Sadeghi, Aakash Tyagi, and Jeyavijayan Rajendran. TheHuzz: Instruction fuzzing of processors using golden-reference models for finding software-exploitable vulnerabilities. InSecurity, pages 3219–3236. USENIX, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Jonathan Lee. Dory: Efficient, transparent arguments for generalised inner products and polynomial commitments.Cryptol. ePrint Arch., page 1274, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Peiyu Liu, Shouling Ji, Xuhong Zhang, Qinming Dai, Kangjie Lu, Lirong Fu, Wenzhi Chen, Peng Cheng, Wenhai Wang, and Raheem Beyah. IFIZZ: Deep-state and efficient fault-scenario generation to test IoT firmware. InASE, pages 805–816. IEEE Computer Society, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Muhammad Numair Mansur, Maria Christakis, and Valentin Wüstholz. Metamorphic testing of Datalog engines. InESEC/FSE, pages 639–650. ACM, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Muhammad Numair Mansur, Maria Christakis, Valentin Wüstholz, and Fuyuan Zhang. Detecting critical bugs in SMT solvers using blackbox mutational fuzzing. In ESEC/FSE, pages 701–712. ACM, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Muhammad Numair Mansur, Valentin Wüstholz, and Maria Christakis. Dependency-aware metamorphic testing of Datalog engines. InISSTA, pages 236–247. ACM, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Paul Dan Marinescu and George Candea. LFI: A practical and general library-level fault injector. InDSN, pages 379–388. IEEE Computer Society, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Kenneth L. McMillan. Interpolation and model checking. InHandbook of Model Checking, pages 421–446. Springer, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Austin Mordahl, Zenong Zhang, Dakota Soles, and Shiyi Wei. ECSTATIC: An extensible framework for testing and debugging configurable static analysis. InICSE, pages 550–562. IEEE Computer Society, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Sergio Segura, Gordon Fraser, Ana B. Sánchez, and Antonio Ruiz Cortés. A survey on metamorphic testing. TSE, 42:805–824, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Shashank Sharma, Sai Ritvik Tanksalkar, Sourag Cherupattamoolayil, and Aravind Machiry. Fuzzing API error handling behaviors using coverage guided fault injection. InAsiaCCS. ACM, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Flavien Solt, Katharina Ceesay-Seitz, and Kaveh Razavi. Cascade: CPU fuzzing via intricate program generation. InSecurity. USENIX, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Tachio Terauchi and Alex Aiken. Secure information flow as a safety problem. InSAS, volume 3672 of LNCS, pages 352–367. Springer, 2005.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Dominik Winterer, Chengyu Zhang, and Zhendong Su. Validating SMT solvers via semantic fusion. InPLDI, pages 718–730. ACM, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Dongwei Xiao, Zhibo Liu, Yiteng Peng, and Shuai Wang. MTZK: Testing and exploring bugs in zero-knowledge (ZK) compilers. InNDSS. The Internet Society, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Jinyan Xu, Yiyuan Liu, Sirui He, Haoran Lin, Yajin Zhou, and Cong Wang. MorFuzz: Fuzzing processor via runtime instruction morphing enhanced synchronizable co-simulation. InSecurity, pages 1307–1324. USENIX, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Chengyu Zhang, Ting Su, Yichen Yan, Fuyuan Zhang, Geguang Pu, and Zhendong Su. Finding and understanding bugs in software model checkers. InESEC/FSE, pages 763–773. ACM, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Huaien Zhang, Yu Pei, Junjie Chen, and Shin Hwei Tan. Statfier: Automated testing of static analyzers via semantic-preserving program transformations. InESEC/FSE, pages 237–249. ACM, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Huaien Zhang, Yu Pei, Shuyun Liang, and Shin Hwei Tan. Understanding and detecting annotation-induced faults of static analyzers.PACMSE, 1:722–744, 2024. 30",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [],
+  "mentions": [],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:23:30Z",
+    "is_model_written": true,
+    "summary": "Arguzz tests zero-knowledge virtual machines for soundness and completeness bugs -- cases where a zk VM accepts a proof of a false statement, or rejects a proof of a true one. Both are security-critical, since a zk VM's whole purpose is that its verdicts can be trusted without re-execution.",
+    "narrative": "No SQLancer mention is present. The paper's 47 references parsed cleanly and none of them is a SQLancer publication, so this is a genuine absence rather than an extraction failure. Its inclusion among SQLancer-citing papers comes from the upstream citation graph, most likely through a paper by one of the project's authors on another subject.",
+    "roles": {},
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      },
+      "extends_technique": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      },
+      "compares_with": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      }
+    },
+    "disagreements": [],
+    "unresolved": [
+      "Why this paper is in the citing set. Its bibliography parsed completely and contains no SQLancer publication."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2510_06663.json b/_data/papers/paper_arxiv_2510_06663.json
new file mode 100644
index 0000000..83479aa
--- /dev/null
+++ b/_data/papers/paper_arxiv_2510_06663.json
@@ -0,0 +1,2122 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:23:45Z",
+  "paper": {
+    "id": "paper:arxiv:2510.06663",
+    "title": "Automated Discovery of Test Oracles for Database Management Systems Using LLMs",
+    "authors": [
+      "Qiuyang Mang",
+      "Runyuan He",
+      "Suyang Zhong",
+      "Xiaoxuan Liu",
+      "Huanchen Zhang",
+      "Alvin Cheung"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": null,
+    "arxiv_id": "2510.06663",
+    "s2_paper_id": "919f0ecf5c6ac7059a9e7d76d9841fca3b3eda74",
+    "url": "https://arxiv.org/abs/2510.06663",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2510.06663",
+      "retrieved_at": "2026-09-10T15:23:45Z",
+      "chars": 118489,
+      "content_sha256": "sha256:08c69171da4dbd6fb1dc5f063c9bc4582834a594c6df005c60941808e66ac370"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 40,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3667
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 14922
+      },
+      {
+        "number": "3",
+        "title": "Argus Overview",
+        "start": 16854
+      },
+      {
+        "number": "4",
+        "title": "Constrained Abstract Query",
+        "start": 21609
+      },
+      {
+        "number": "5",
+        "title": "Test Oracle Discovery",
+        "start": 26530
+      },
+      {
+        "number": "5.1",
+        "title": "Database Seeding",
+        "start": 26925
+      },
+      {
+        "number": "5.2",
+        "title": "CAQ Pairs Generation",
+        "start": 28823
+      },
+      {
+        "number": "5.3",
+        "title": "Equivalence Checking",
+        "start": 31859
+      },
+      {
+        "number": "6",
+        "title": "Test Cases Instantiation",
+        "start": 33588
+      },
+      {
+        "number": "6.1",
+        "title": "Corpus Synthesis",
+        "start": 34020
+      },
+      {
+        "number": "6.2",
+        "title": "Query Instantiation",
+        "start": 39635
+      },
+      {
+        "number": "6.3",
+        "title": "Database Instantiation and Bug Reporting",
+        "start": 44164
+      },
+      {
+        "number": "7",
+        "title": "Evaluation",
+        "start": 45055
+      },
+      {
+        "number": "7.1",
+        "title": "New Bugs and Oracles",
+        "start": 46887
+      },
+      {
+        "number": "7.2",
+        "title": "Code Coverage",
+        "start": 55037
+      },
+      {
+        "number": "7.3",
+        "title": "Effect of Test Oracles",
+        "start": 61514
+      },
+      {
+        "number": "7.4",
+        "title": "Effect of SQL Equivalence Prover",
+        "start": 63630
+      },
+      {
+        "number": "7.5",
+        "title": "Cost and Efficiency Analysis",
+        "start": 66517
+      },
+      {
+        "number": "7.6",
+        "title": "Component-wise Analysis",
+        "start": 68338
+      },
+      {
+        "number": "8",
+        "title": "Discussion",
+        "start": 70928
+      },
+      {
+        "number": "9",
+        "title": "Related Work",
+        "start": 74924
+      },
+      {
+        "number": "10",
+        "title": "Conclusion",
+        "start": 79519
+      },
+      {
+        "number": "11",
+        "title": "Ackowledgements",
+        "start": 80293
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Arvind Arasu, Raghav Kaushik, and Jian Li. 2011. Data generation using declarative constraints. InProceedings of the 2011 ACMSIGMOD International Conference on Management of data. 685–696.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba, Yuancheng Jiang, and Manuel Rigger. 2025. Metamorphic Coverage.arXiv preprint arXiv:2508.16307 (2025).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans.Proceedings of the ACM on Management of Data2, 3 (2024), 1–26.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Earl T Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. 2014. The oracle problem in software testing: A survey.IEEE transactions on software engineering41, 5 (2014), 507–525.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Edmon Begoli, Jesús Camacho-Rodríguez, Julian Hyde, Michael J Mior, and Daniel Lemire. 2018. Apache calcite: A foundational framework for optimized query processing over heterogeneous data sources. InProceedings of the 2018 International Conference on Management of Data. 221–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M Tamer Özsu. 2007. QAGen: generating query-aware test databases. InProceedings of the 2007 ACMSIGMOD international conference on Management of data. 341–352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Chunyu Chen, Zhengjie Miao, Yong Zhang, and Jiannan Wang. 2025. ParSEval: Plan-aware Test Database Generation for SQL Equivalence Evaluation.Proceedings of the VLDB Endowment18, 11 (2025), 4750–4762.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Shumo Chu, Chenglong Wang, Konstantin Weitz, and Alvin Cheung. 2017. Cosette: An Automated Prover for SQL.. In CIDR. 1–7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Shumo Chu, Konstantin Weitz, Alvin Cheung, and Dan Suciu. 2017. HoTTSQL: Proving query rewrites with univalent SQL semantics.Acm sigplan notices52, 6 (2017), 510–524.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Wenjing Deng, Qiuyang Mang, Chengyu Zhang, and Manuel Rigger. 2024. Finding logic bugs in spatial database engines via affine equivalent inputs.Proceedings of the ACM on Management of Data2, 6 (2024), 1–26.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 13,
+      "text": "Haoran Ding, Zhaoguo Wang, Yicun Yang, Dexin Zhang, Zhenglin Xu, Haibo Chen, Ruzica Piskac, and Jinyang Li. 2023. Proving query equivalence using linear integer arithmetic.Proceedings of the ACM on Management of Data1, 4 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Dolt. 2025. Dolt Homepage. [EB/OL]. https://www.dolthub.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, and Yu Jiang. 2024. Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 146, 12 pages. https://doi.org/10.1145/3597503.3639210",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2023. Griffin: Grammar-Free DBMS Fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering(Rochester, MI, USA) (ASE ’22). Association for Computing Machinery, New York, NY, USA, Article 49, 12 pages. https://doi.org/10.1145/ 3551349.3560431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Lele Gaifax and Contributors. 2025. pglast: A PostgreSQL AST and statements prettifier for Python. https://github. com/lelit/pglast. Accessed: 2025-10-15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Victor Giannakouris and Immanuel Trummer. 2025. 𝜆-tune: Harnessing large language models for automated database system tuning.Proceedings of the ACM on Management of Data3, 1 (2025), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Yang He, Pinhan Zhao, Xinyu Wang, and Yuepeng Wang. 2024. VeriEQL: Bounded equivalence verification for complex SQL queries with integrity constraints.Proceedings of the ACM on Programming Languages8, OOPSLA1 (2024), 1071–1099.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Charles Hong, Sahil Bhatia, Alvin Cheung, and Yakun Sophia Shao. 2025. Autocomp: LLM-Driven Code Optimization for Tensor Accelerators.arXiv preprint arXiv:2505.18574(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Zijin Hong, Zheng Yuan, Qinggang Zhang, Hao Chen, Junnan Dong, Feiran Huang, and Xiao Huang. 2024. Nextgeneration database interfaces: A survey of llm-based text-to-sql.arXiv preprint arXiv:2406.08426(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Dongxu Huang, Qi Liu, Qiu Cui, Zhuhe Fang, Xiaoyu Ma, Fei Xu, Li Shen, Liu Tang, Yuxing Zhou, Menglong Huang, et al. 2020. TiDB: a Raft-based HTAP database.Proceedings of the VLDB Endowment13, 12 (2020), 3072–3084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland HC Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. InProceedings of the 46th IEEE/ACM International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 24,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting logic bugs in database engines via equivalent expression transformation. In18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24). 821–835. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:26 Qiuyang Mang et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. Apollo: Automatic detection and diagnosis of performance regressions in database systems.Proceedings of the VLDB Endowment13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "George Klees, Andrew Ruef, Benji Cooper, Shiyi Wei, and Michael Hicks. 2018. Evaluating Fuzz Testing. InProceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security(Toronto, Canada)(CCS ’18). Association for Computing Machinery, New York, NY, USA, 2123–2138. https://doi.org/10.1145/3243734.3243804",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Dacheng Li, Shiyi Cao, Chengkun Cao, Xiuyu Li, Shangyin Tan, Kurt Keutzer, Jiarong Xing, Joseph E Gonzalez, and Ion Stoica. 2025. S*: Test time scaling for code generation.arXiv preprint arXiv:2502.14382(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Haonan Li, Yu Hao, Yizhuo Zhai, and Zhiyun Qian. 2024. Enhancing static analysis for practical bug detection: An llm-integrated approach.Proceedings of the ACM on Programming Languages8, OOPSLA1 (2024), 474–499.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Zhaodonghui Li, Haitao Yuan, Huiming Wang, Gao Cong, and Lidong Bing. 2024. LLM-R2: A large language model enhanced rule-based rewrite system for boosting query efficiency.arXiv preprint arXiv:2404.12872(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-Oriented DBMS Fuzzing. In Engineering (ICDE). 668–681. https://doi.org/10.1109/ICDE55515.2023.00057",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of DBMS with Coverage-based Guidance. In31st USENIX Security Symposium (USENIX Security 22). USENIX Association, Boston, MA, 4309–4326. https://www.usenix. org/conference/usenixsecurity22/presentation/liang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Yiming Lin, Madelon Hulsebos, Ruiying Ma, Shreya Shankar, Sepanta Zeigham, Aditya G Parameswaran, and Eugene Wu. 2024. Towards accurate and efficient document analytics with large language models.arXiv preprint arXiv:2405.04674(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Yiming Lin, Madelon Hulsebos, Ruiying Ma, Shreya Shankar, Sepanta Zeighami, Aditya G Parameswaran, and Eugene Wu. 2025. Querying templatized document collections with large language models. In2025 IEEE 41st International Conference on Data Engineering (ICDE). IEEE, 2422–2435.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Shuang Liu, Junhao Lan, Xiaoning Du, Jiyuan Li, Wei Lu, Jiajun Jiang, and Xiaoyong Du. 2024. Testing graph database systems with graph-state persistence oracle. InProceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. 666–677.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Xiaoxuan Liu, Shuxian Wang, Mengzhu Sun, Sicheng Pan, Ge Li, Siddharth Jha, Cong Yan, Junwen Yang, Shan Lu, and Alvin Cheung. 2022. Leveraging application data constraints to optimize database-backed web applications.arXiv preprint arXiv:2205.02954(2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. InProceedings of the 44th International Conference on Software Engineering. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting.Proceedings of the ACM on Management of Data3, 3 (2025), 1–27.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Bruce Momjian. 2001.PostgreSQL: introduction and concepts. Vol. 192. Addison-Wesley New York.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Niels Mündler, Jingxuan He, Hao Wang, Koushik Sen, Dawn Song, and Martin Vechev. 2025. Type-Constrained Code Generation with Language Models.Proceedings of the ACM on Programming Languages9, PLDI (2025), 601–626.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "MySQL. 2025. MySQL Homepage. [EB/OL]. https://www.mysql.com",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Alexander Novikov, Ngân V ˜u, Marvin Eisenberger, Emilien Dupont, Po-Sen Huang, Adam Zsolt Wagner, Sergey Shirobokov, Borislav Kozlovskii, Francisco JR Ruiz, Abbas Mehrabian, et al .2025. AlphaEvolve: A coding agent for scientific and algorithmic discovery.arXiv preprint arXiv:2506.13131(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "OpenAI. 2025. Introducing OpenAI o3 and o4-mini. https://openai.com/index/introducing-o3-and-o4-mini/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Liana Patel, Siddharth Jha, Melissa Pan, Harshit Gupta, Parth Asawa, Carlos Guestrin, and Matei Zaharia. 2024. Semantic Operators: A Declarative Model for Rich, AI-based Data Processing.arXiv preprint arXiv:2407.11418(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Shishir G Patil, Huanzhi Mao, Fanjia Yan, Charlie Cheng-Jie Ji, Vishnu Suresh, Ion Stoica, and Joseph E Gonzalez. [n. d.]. The Berkeley Function Calling Leaderboard (BFCL): From Tool Use to Agentic Evaluation of Large Language Models. InForty-second International Conference on Machine Learning.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Joanna Purich, Anthony Wise, and Leilani Battle. 2025. An adaptive benchmark for modeling user exploration of large datasets.Proceedings of the ACM on Management of Data3, 1 (2025), 1–24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Mark Raasveldt and Hannes Mühleisen. 2019. Duckdb: an embeddable analytical database. InProceedings of the 2019 international conference on management of data. 1981–1984.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 48,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:27",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "Dmitry Rybin, Yushun Zhang, and Zhi-Quan Luo. 2025. XX𝑇Can Be Faster.arXiv preprint arXiv:2505.09814(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Anupam Sanghi and Jayant R Haritsa. 2023. Synthetic data generation for enterprise dbms. In2023 IEEE 39th International Conference on Data Engineering (ICDE). IEEE, 3585–3588.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Tobias Schmidt, Viktor Leis, Peter Boncz, and Thomas Neumann. 2025. SQLStorm: Taking Database Benchmarking into the LLM Era.Proceedings of the VLDB Endowment18, 11 (2025), 4144–4157.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Andreas Seltenreich. 2022. Sqlsmith. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Shreya Shankar, Tristan Chambers, Tarak Shah, Aditya G Parameswaran, and Eugene Wu. 2024. Docetl: Agentic query rewriting and evaluation for complex document processing.arXiv preprint arXiv:2410.12189(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Chaofan Shou, Jing Liu, Doudou Lu, and Koushik Sen. 2024. Llm4fuzz: Guided fuzzing of smart contracts with large language models.arXiv preprint arXiv:2401.11108(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Donald R Slutz. 1998. Massive stochastic testing of SQL. InVLDB, Vol. 98. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Jiansen Song, Wensheng Dou, Yu Gao, Ziyu Cui, Yingying Zheng, Dong Wang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction.Proc. VLDB Endow. 17, 8 (April 2024), 1884–1897. https://doi.org/10.14778/3659437.3659445",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Jiansen Song, Wensheng Dou, Yingying Zheng, Yu Gao, Ziyu Cui, Wei Wang, and Jun Wei. 2025. Detecting SchemaRelated Logic Bugs in Relational DBMSs via Equivalent Database Construction.Proceedings of the VLDB Endowment (VLDB)(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Jie Tan, Kangfei Zhao, Rui Li, Jeffrey Xu Yu, Chengzhi Piao, Hong Cheng, Helen Meng, Deli Zhao, and Yu Rong. 2025. Can Large Language Models Be Query Optimizer for Relational Databases?arXiv preprint arXiv:2502.05562(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting logic bugs of join optimizations in dbms.Proceedings of the ACM on Management of Data1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Xiu Tang, Shijie Yang, Sai Wu, Dongxiang Zhang, Wenchao Zhou, Feifei Li, and Gang Chen. 2025. Unveiling Logic Bugs in SPJG Query Optimizations within DBMS.ACM Transactions on Database Systems(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Boris A Trakhtenbrot. 1950. Impossibility of an algorithm for the decision problem for finite classes. InDoklady Akademiia Nauk SSSR, Vol. 70. 569.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Shuxian Wang, Sicheng Pan, and Alvin Cheung. 2024. QED: A powerful query equivalence decider for SQL.Proceedings of the VLDB Endowment17, 11 (2024), 3602–3614.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Zhaoguo Wang, Zhou Zhou, Yicun Yang, Haoran Ding, Gansen Hu, Ding Ding, Chuzhe Tang, Haibo Chen, and Jinyang Li. 2022. Wetune: Automatic discovery and verification of query rewrite rules. InProceedings of the 2022 International Conference on Management of Data. 94–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Chunqiu Steven Xia, Matteo Paltenghi, Jia Le Tian, Michael Pradel, and Lingming Zhang. 2024. Fuzz4all: Universal fuzzing with large language models. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Chunqiu Steven Xia and Lingming Zhang. 2024. Automated program repair via conversation: Fixing 162 out of 337 bugs for $0.42 each using chatgpt. InProceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. 819–831.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Chenyuan Yang, Yinlin Deng, Runyu Lu, Jiayi Yao, Jiawei Liu, Reyhaneh Jabbarvand, and Lingming Zhang. 2024. Whitefox: White-box compiler fuzzing empowered by large language models.Proceedings of the ACM on Programming Languages8, OOPSLA2 (2024), 709–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Chenyuan Yang, Zijie Zhao, and Lingming Zhang. 2025. Kernelgpt: Enhanced kernel fuzzing via large language models. InProceedings of the 30th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2. 560–573.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Jingyi Yang, Peizhi Wu, Gao Cong, Tieying Zhang, and Xiao He. 2022. SAM: Database generation from query workloads with supervised autoregressive models. InProceedings of the 2022 International Conference on Management of Data. 1542–1555.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Yicun Yang, Zhaoguo Wang, Yu Xia, Zhuoran Wei, Haoran Ding, Ruzica Piskac, Haibo Chen, and Jinyang Li. 2025. Automated Validating and Fixing of Text-to-SQL Translation with Execution Consistency.Proceedings of the ACM on Management of Data3, 3 (2025), 1–28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Shiyang Ye, Chao Ni, Jue Wang, Qianqian Pang, Xinrui Li, and Xiaodan Xu. 2025. Sembug: Detecting Logic Bugs in Dbms Through Generating Semantic-Aware Non-Optimizing Query. In on Program Comprehension (ICPC). IEEE, 124–135.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Cunxi Yu, Rongjian Liang, Chia-Tung Ho, and Haoxing Ren. 2025. Autonomous Code Evolution Meets NP-Completeness. arXiv preprint arXiv:2509.07367(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Sepanta Zeighami, Shreya Shankar, and Aditya Parameswaran. 2025. Cut Costs, Not Accuracy: LLM-Powered Data Processing with Guarantees.arXiv preprint arXiv:2509.02896(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing.Proceedings of the ACM on Management of Data3, 1 (2025), 1–24. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:28 Qiuyang Mang et al.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 75,
+      "text": "Kaizhong Zhang and Dennis Shasha. 1989. Simple fast algorithms for the editing distance between trees and related problems.SIAM journal on computing18, 6 (1989), 1245–1262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Michael R Zhang, Nishkrit Desai, Juhan Bae, Jonathan Lorraine, and Jimmy Ba. 2023. Using large language models for hyperparameter optimization.arXiv preprint arXiv:2312.04528(2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Qizheng Zhang, Ali Imran, Enkeleda Bardhi, Tushar Swamy, Nathan Zhang, Muhammad Shahbaz, and Kunle Olukotun. 2024. Caravan: Practical online learning of In-Network ML models with labeling agents. InProceedings of the 3rd Workshop on Practical Adoption Challenges of ML for Systems. 17–20.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Fuheng Zhao, Lawrence Lim, Ishtiyaque Ahmad, Divyakant Agrawal, and Amr El Abbadi. 2023. Llm-sql-solver: Can llms determine SQL equivalence?arXiv preprint arXiv:2312.10321(2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Pinhan Zhao, Yuepeng Wang, and Xinyu Wang. 2025. Polygon: Symbolic Reasoning for SQL using Conflict-Driven Under-Approximation Search.Proceedings of the ACM on Programming Languages9, PLDI (2025), 1315–1340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security(Virtual Event, USA)(CCS ’20). Association for Computing Machinery, New York, NY, USA, 955–970. https://doi.org/10.1145/3372297.3417260",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments. arXiv preprint arXiv:2505.02012(2025).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 82,
+      "text": "Suyang Zhong and Manuel Rigger. 2026. Scaling Automated Database System Testing. InProceedings of the 31st ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2(USA) (ASPLOS ’26). Association for Computing Machinery, New York, NY, USA, 1677–1692. https://doi.org/10.1145/3779212. 3790215",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 83,
+      "text": "Qi Zhou, Joy Arulraj, Shamkant Navathe, William Harris, and Jinpeng Wu. 2020. SPES: A two-stage query equivalence verifier.arXiv preprint arXiv:2004.00481(2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Wei Zhou, Yuyang Gao, Xuanhe Zhou, and Guoliang Li. 2025. CrackSQL: A Hybrid SQL Dialect Translation System Powered by Large Language Models.arXiv preprint arXiv:2504.00882(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "Xuanhe Zhou, Zhaoyan Sun, and Guoliang Li. 2024. Db-gpt: Large language model meets database.Data Science and Engineering9, 1 (2024), 102–111.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Zeyang Zhuang, Penghui Li, Pingchuan Ma, Wei Meng, and Shuai Wang. 2023. Testing graph database systems via graph-aware metamorphic relations.Proceedings of the VLDB Endowment17, 4 (2023), 836–848.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "Ándré Albrecht and Contributors. 2025. sqlparse: A non-validating SQL parser module for Python. https://github.com/ andialbrecht/sqlparse. Accessed: 2025-10-15. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:29 A Omitted Algorithm Algorithm 3:GuidedCAQPairsSampling Input:ADBMSD, a set of equivalent CAQ",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "Jinsheng Ba, Yuancheng Jiang, and Manuel Rigger. 2025. Metamorphic Coverage.arXiv preprint arXiv:2508.16307 (2025).",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans.Proceedings of the ACM on Management of Data2, 3 (2024), 1–26.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 12,
+      "text": "Wenjing Deng, Qiuyang Mang, Chengyu Zhang, and Manuel Rigger. 2024. Finding logic bugs in spatial database engines via affine equivalent inputs.Proceedings of the ACM on Management of Data2, 6 (2024), 1–26.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 23,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland HC Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. InProceedings of the 46th IEEE/ACM International Conference on Software Engineering. 1–12.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting.Proceedings of the ACM on Management of Data3, 3 (2025), 1–27.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 48,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:27",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 74,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing.Proceedings of the ACM on Management of Data3, 1 (2025), 1–24. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:28 Qiuyang Mang et al.",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 81,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments. arXiv preprint arXiv:2505.02012(2025).",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "prints the arXiv id of the paper introducing shqvel"
+      ]
+    },
+    {
+      "number": 82,
+      "text": "Suyang Zhong and Manuel Rigger. 2026. Scaling Automated Database System Testing. InProceedings of the 31st ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2(USA) (ASPLOS ’26). Association for Computing Machinery, New York, NY, USA, 1677–1692. https://doi.org/10.1145/3779212. 3790215",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing sqlancer_pp",
+        "prints the DOI of the paper introducing sqlancer_pp"
+      ]
+    },
+    {
+      "number": 87,
+      "text": "Ándré Albrecht and Contributors. 2025. sqlparse: A non-validating SQL parser module for Python. https://github.com/ andialbrecht/sqlparse. Accessed: 2025-10-15. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:29 A Omitted Algorithm Algorithm 3:GuidedCAQPairsSampling Input:ADBMSD, a set of equivalent CAQsEqual, a number of maximal number of samples to returnMaxSamples, a number of maximal iterations for clusteringMaxIters. Output:A sampled set of equivalent CAQ pairsSam. 1FunctionDistance(D,caq,cluster): 2MinDistance←+∞; 3foreachq∈clusterdo 4plan1←D.Explain(caq); 5plan2←D.Explain(q); 6dist←TreeEditDistance(plan1,plan2); 7ifdist<MinDistancethen 8MinDistance←dist; 9returnMinDistance; 10if|Equal|≤MaxSamplesthen 11Sam←Equal; 12returnSam; 13Clusters←RandomSample(Equal,MaxSamples); 14foriter←1toMaxItersdo 15NewClusters←{}; 16foreachcaq∈Equaldo 17𝑖∗←arg min 𝑖Distance(D,caq,Clusters[𝑖]); 18NewClusters[𝑖∗]←NewClusters[𝑖∗]∪{caq}; 19ifNewClusters=Clustersthen 20break; 21Clusters←NewClusters; 22Sam←{}; 23foreachcluster∈Clustersdo 24caq←RandomSample(cluster,1); 25Sam←Sam∪{caq}; 26returnSam; B Counter-Examples and Proofs In this section, we provide three counter-examples to demonstrate that the potential false positives if we do not apply the three general constraints of Cmentioned in Sec. 6.2. Specifically, Listing B.1, Listing B.2, and Listing B.3 show three cases that can be formally proven in the scenario of virtual columns, but lead to false positives in the instantiation phase by missing the constraints. We use these constraints to ensure that snippets replacing column expressions are deterministic, preserve null values during outer joins, and refer to a unique row’s column combination in the cases other than outer joins. We then prove the following theorem to show that if we follow these constraints, the equivalence is preserved after instantiating expressions. Theorem B.1 (Expression Instantiation Preserves Eqivalence).Consider two SQL queries, 𝑄1and𝑄2, that aresemantically equivalenton a schema 𝑆containing virtual columns {𝑣𝑖}. Let 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:30 Qiuyang Mang et al. 1CREATE TABLEt1(c0BOOLEAN); 2INSERT INTOt1VALUES(TRUE); 3SELECT*FROMt1 4WHERE□ 1⊲Expr(t:BOOLEAN)↦→RANDOM() < 0.5; 5-- {unstably return 1 row or 0 rows} 6SELECT*FROMt1 7WHERE□ 1⊲Expr(t:BOOLEAN)↦→RANDOM() < 0.5; 8-- {unstably return 1 row or 0 rows} Listing B.1. False positive example when instantiating non-deterministic snippets. 1CREATE TABLEt1(c0INT); 2CREATE TABLEt2(c0INT); 3INSERT INTOt1VALUES(1); 4INSERT INTOt2VALUES(1); 5SELECT□ 1⊲Expr(t:INT)↦→NULLIF(t2.c0, 0)FROMt1LEFT JOINt2WHERE false; -- {0} 6SELECT NULL FROMt1LEFT JOINt2WHERE false; -- {NULL} Listing B.2. False positive example when instantiating snippets without null-preserving property. 1CREATE TABLEt1(c0INT); 2INSERT INTOt1VALUES(1); 3SELECT*FROMt1WHERE false; -- {0 rows} 4SELECT*FROMt1WHERE EXISTS( 5SELECT□ 1⊲Expr(t:INT)↦→SUM(c0)FROMt1WHERE false); -- {1 row} Listing B.3. False positive example when instantiating snippets without empty-preserving property. 𝑄′ 1and𝑄′ 2be the queries created by replacing every occurrence of each virtual column 𝑣𝑖with its corresponding expression 𝑒𝑖. If each expression 𝑒𝑖references aunique row’scolumn combination in all cases other than outer joins, then 𝑄′ 1and𝑄′ 2are alsosemantically equivalenton the schema 𝑆′, which is𝑆without the virtual columns. Proof. We prove this by contradiction. Assume that 𝑄′ 1and𝑄′ 2are not semantically equivalent on schema𝑆′, meaning there exists a database instance 𝐷′on𝑆′where𝑄′ 1(𝐷′)≠𝑄′ 2(𝐷′). We then construct a database 𝐷on schema𝑆by augmenting 𝐷′with the virtual columns {𝑣𝑖}defined by their corresponding expressions {𝑒𝑖}. The core of this proof is establishing that for any query 𝑄𝑘, the results𝑄𝑘(𝐷)and𝑄′ 𝑘(𝐷′)are identical. This equivalence is guaranteed by two properties of the expressions{𝑒𝑖}. First, theunique row referenceproperty ensures that for any standard join or matched row, the value of a virtual column 𝑣𝑖in𝐷is computed from the same single row as its expression 𝑒𝑖in𝐷′, yielding an identical result. Second, and crucially for outer joins, the null-preservingproperty handles unmatched rows. If an outer join produces a row where the columns underlying an expression 𝑒𝑖are padded with NULL, the expression 𝑒𝑖itself evaluates to NULL in query𝑄′ 𝑘. The null-preserving property explicitly forces the corresponding virtual column 𝑣𝑖in query𝑄𝑘to also resolve to NULL in this exact scenario. This perfect mirroring of behavior ensures that 𝑄1(𝐷)=𝑄′ 1(𝐷′)and𝑄2(𝐷)=𝑄′ 2(𝐷′). However, since 𝑄′ 1(𝐷′)≠𝑄′ 2(𝐷′)by our initial assumption, it follows that 𝑄1(𝐷)≠𝑄 2(𝐷), contradicting the premise that 𝑄1and𝑄2are semantically equivalent on schema𝑆.□ 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:31 This theorem shows that, under the scenario of only virtual columns, the equivalence is preserved after instantiating expressions if we follow the constraints mentioned in Sec. 6.2. When there are virtual tables, it is much easier to show the preserved equivalence after instantiating tables, since the tables are simply replaced by other tables with the same schema. If there is no non-deterministic rows in the instantiated tables, the equivalence is also preserved after instantiating tables, because such instantiation is equivalent to using a CREATE VIEW to replace the CREATE TABLE for virtual tables. Note that, while these proofs do not cover all SQL features, especially for those new features in future SQL standards, we have not found any counterexamples in our experiments that violate the equivalence after instantiating expressions following these constraints. In practice, we can maintain the soundness of instantiated queries by disabling any new features that invalidate existing proofs. C Representing Existing Oracles 1(1)SELECT*FROMt1,□ 1⊲Table(...); 2(2)SELECT*FROMt1,□ 1⊲Table(...)WHERE□ 2⊲Expr(t1:BOOLEAN)AND □1⊲Table(...).c0IS TRUEUNION ALL 3SELECT*FROMt1,□ 1⊲Table(...)WHERE□ 2⊲Expr(t1:BOOLEAN)AND □1⊲Table(...).c0IS FALSE UNION ALL 4SELECT*FROMt1,□ 1⊲Table(...)WHERE□ 2⊲Expr(t1:BOOLEAN)AND □1⊲Table(...).c0IS NULL; Listing C.1. TLP [48] in CAQ. 1(1)SELECT COUNT(*)FROMt1,□ 1⊲Table(...)WHERE□ 2⊲Expr(t1:BOOLEAN)AND □1⊲Table(...).c0; 2(2)SELECT SUM(cnt)FROM(SELECT(□ 2⊲Expr(t1:BOOLEAN)AND□ 1⊲Table(...). c0) 3IS TRUE AScntFROMt1,□ 1⊲Table(...)); Listing C.2. NoREC [47] in CAQ. 1(1)SELECT*FROMt1,□ 1⊲Table(...)WHERE□ 2⊲Expr(t1:BOOLEAN); 2(2)SELECT*FROMt1,□ 1⊲Table(...)WHERE((□ 1⊲Table(...).c0AND(NOT □1⊲Table(...).c0) 3AND(□ 1⊲Table(...).c0IS NOT NULL)))OR□ 2⊲Expr(t1:BOOLEAN); 4(3)SELECT*FROMt1,□ 1⊲Table(...)WHERE CASEWHEN 5((□ 1⊲Table(...).c0AND(NOT□ 1⊲Table(...).c0)AND(□ 1⊲Table(...).c0IS NOT NULL))) 6THEN□ 3⊲Expr(t1:BOOLEAN)ELSE□ 2⊲Expr(t1:BOOLEAN)END; 7(4)SELECT*FROMt1,□ 1⊲Table(...)WHERE CASEWHEN 8□ 1⊲Table(...).c0THEN□ 2⊲Expr(t1:BOOLEAN)ELSE □2⊲Expr(t1:BOOLEAN)END; Listing C.3. EET [24] in CAQ. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:32 Qiuyang Mang et al. 1(1)SELECT*FROMt1JOIN□ 1⊲Table(...)ON□ 2⊲Expr(t1:ANY) = □1⊲Table(...).c0; 2(2) /*+ JOIN_ORDER(□ 1⊲Table(...), t1) */ 3SELECT*FROMt1JOIN□ 1⊲Table(...)ON□ 2⊲Expr(t1:ANY) = □1⊲Table(...).c0; 4(3) /*+ LOOKUP_JOIN(□ 1⊲Table(...), t1) */ 5SELECT*FROMt1JOIN□ 1⊲Table(...)ON□ 2⊲Expr(t1:ANY) = □1⊲Table(...).c0; 6(4) /*+ MERGE_JOIN(□ 1⊲Table(...), t1) */ 7SELECT*FROMt1JOIN□ 1⊲Table(...)ON□ 2⊲Expr(t1:ANY) = □1⊲Table(...).c0; 8(5) /*+ HASH_JOIN(□ 1⊲Table(...), t1) */ 9SELECT*FROMt1JOIN□ 1⊲Table(...)ON□ 2⊲Expr(t1:ANY) = □1⊲Table(...).c0; 10(6) /*+ INNER_JOIN(□ 1⊲Table(...), t1) */ 11SELECT*FROMt1JOIN□ 1⊲Table(...)ON□ 2⊲Expr(t1:ANY) = □1⊲Table(...).c0; Listing C.4. DQP [5] in CAQ. D Generated SQL Queries Query 1 of Argus 1SELECTt8.c3, 2t0.c2, 3t1.c3, 4t1.c2, 5t8.c0, 6t0.c3, 7(CASEWHEN( (div(t3.c3 + 10,nullif((EXTRACT(DAY FROM(DATE'2000-01-01'+ t3.c3 *INTERVAL'1 day'))::int), 1)) * mod((EXTRACT(DAY FROM(DATE'2000-01-01' + t3.c3 *INTERVAL'1 day'))::int) + 5, 3))::int) = 0THENt3.c0ELSEt3.c2END), 8t8.c1, 9((~ (~ ((-1518331885)&(( (+ ((t8.c2)/(((2087249086)#(760045829))))) )))))), 10vtable0.c0, 11t3.c3, 12vtable0.c3, 13(CASE WHENt1.c1 ISNULL THEN(char_length(( (REPLACE(UPPER(t1.c1),'A','4') LPAD(t1.c1::TEXT, 3,'0'))::VARCHAR)) * t1.c0 ) > 0ELSElength(t1.c1) > 3END) 14FROMt0 15CROSS JOINt8 16CROSS JOINt1 17CROSS JOINt3 18CROSS JOINvtable0 19WHERE1<>1; 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:33 Query 2 of Argus 1SELECTdt.t0_c2, 2vtable0.c2, 3dt.t0_c1, 4dt.t8_c1 5FROMt1 6LEFT JOINvtable0 7ON NOT(t1.c1BETWEEN(md5(t1.c1 ( regexp_replace(t1.c1,'[aeiou]','*','gi') ))) ANDt1.c1) 8CROSS JOIN( 9SELECTt0.c2ASt0_c2, 10t0.c1ASt0_c1, 11t8.c1ASt8_c1 12FROMt8 13INNER JOINt0 14ONt8.c1 <> (t8.c1 t0.c1) 15)ASdt 16WHERE((COALESCE(length(( to_char(t1.c2 * t1.c3,'FM9999999')::varchar)), 0) - ( EXTRACT(MONTH FROM(date'2000-01-01'+ t1.c3 *interval'1 day'))::int) ) =ANY( ARRAY[t1.c3, t1.c3 * 2, 0])) ISNULL; Query 3 of Argus 1SELECT(bit_count((translate(md5(( (CASE WHENt0.c0THENt0.c1ELSEt0.c0END) )),' abcdef','FEDCBA') )::bytea)::integer), 2(translate(( (t1.c3::text':'( (EXTRACT(EPOCHFROM(TO_TIMESTAMP(' 2000-01-01','YYYY-MM-DD') + (t1.c3'days')::INTERVAL))::INT) + t1.c3 * 3 - (t1. c3 & t1.c3) )::text) ),'aeiou','12345')::varchar), 3t1.c3, 4(t3.c3 % 2 = 0), 5(REPLACE(t0.c1, t0.c0, REVERSE(t0.c0 t0.c1))), 6vtable0.c2, 7vtable0.c1, 8t1.c0, 9vtable0.c3, 10vtable1.c2, 11(bit_length(t0.c1) / 2 +octet_length(( LPAD(t0.c1, 10,SUBSTRING(t0.c0FROM1 FOR 1)) ))), 12t1.c2, 13t3.c0 14FROMt0, t1, vtable0, vtable1, t3 15WHERE FALSE; Query 4 of Argus 1SELECT 2vtable0.c3, 3t3.c3, 4(get_bit(decode(to_hex(t3.c3),'hex'), 7) = 1), 5vtable0.c0, 6(char_length(t8.c1) +char_length(coalesce(t8.c3,''))), 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:34 Qiuyang Mang et al. 7t1.c3, 8vtable0.c2, 9t8.c0, 10(LENGTH(md5(COALESCE((translate(t0.c1 t0.c0,'aeiou AEIOU','1234512345') ), t0.c0 )))), 11(to_char(DATE'2020-01-01'+ (t3.c3'days')::interval,'YYYY-MM-DD')::VARCHAR), 12(regexp_count(( t0.c0'_'( t0.c0'-'t0.c1 ) ),'[0-9]')), 13t8.c1, 14t0.c2, 15vtable1.c1, 16vtable1.c2, 17t8.c2, 18t3.c0, 19vtable0.c1, 20t0.c0, 21t0.c1, 22((( encode(sha256(convert_to((trim(both'#'fromt8.c3)::varchar),'UTF8')),'hex ') )LIKE'%test%'ORt8.c3 ~'^[0-9]+$')ANDt8.c2BETWEEN1AND10), 23((t3.c2AND NOTt3.c1)OR((extract(epochfrommake_interval(days => t3.c3))::int ) % 2 = 0)), 24((CASEWHEN( (SELECTbool_and(x::boolean)FROMjsonb_array_elements_text( jsonb_build_array(t0.c0, t0.c2))ASa(x)) )THENt0.c0ELSE(translate(t0.c0,' aeiou','12345') )END)), 25(regexp_count(( t8.c1'_'( t8.c1'-'t8.c3 ) ),'[0-9]')), 26t1.c2, 27t0.c3, 28(reverse(t1.c1)), 29t8.c3, 30vtable1.c0, 31t1.c1, 32(CASE WHENt1.c3 > t1.c2THEN bit_length(t1.c2)ELSEwidth_bucket(t1.c2, 0, 100, 10)END), 33t3.c2, 34(((((((( regexp_replace(t1.c3 ( to_char(( (+ (+ ((( (t1.c0 << 2) + (t1.c3 >> 1) )) *(302076923)))) ),'FM0000') ),'[aeiou]','*','gi') ))('')))(((t1.c3)('rR'))))) <=((((('')('h')))(((t1.c3)('5M'))))))), 35vtable1.c3, 36t3.c1 37FROM 38vtable0 39LEFT JOINt8ONvtable0.c0 40CROSS JOINt3 41CROSS JOINvtable1 42CROSS JOINt1 43CROSS JOINt0 44WHERE 45((t0.c2ORt0.c3)ANDt0.c3) = (t0.c3 < t0.c2) 46; Query 5 of Argus 1SELECT 2t0.c3, 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:35 3sub_t1.c0_val, 4vtable1.c2, 5t0.c0, 6(CHAR_LENGTH(t0.c0 ( REPEAT(t0.c1, GREATEST(1, LEAST((bit_length(t0.c0) ), 5))) ) )), 7(octet_length(translate(t0.c0,'aeiou','AEIOU'))), 8sub_t1.ph1, 9vtable1.c0 10FROMt0 11CROSS JOINvtable1 12LEFT JOIN( 13SELECT 14t1.c0ASc0_val, 15(make_interval(days => ( width_bucket(t1.c3, 1, 100, 5) )) < make_interval(days => t1.c2))ASph1 16FROMt1 17) sub_t1ONvtable1.c2; Query 6 of Argus 1SELECTt8.c3, 2t0.c2, 3t1.c3, 4t1.c2, 5t8.c0, 6t0.c3, 7((t3.c3 % 2 = 0)ORt3.c1), 8t8.c1, 9(COALESCE(length(t8.c2), 0) - (EXTRACT(MONTH FROM(date'2000-01-01'+ ( GREATEST((EXTRACT(EPOCHFROM(DATE'2020-01-01'+ (t8.c2'days')::interval)):: int), -(EXTRACT(DAY FROM('2000-01-01'::timestamp+ t8.c2 *'1 day'::interval)):: int)) ) *interval'1 day'))::int)), 10vtable0.c0, 11t3.c3, 12vtable0.c3, 13(t1.c0 > ( (t1.c0 +COALESCE(( t1.c0 + (EXTRACT(EPOCHFROM(timestamp' 2000-01-01'+ (t1.c0'days')::interval))::int) * LENGTH(t1.c2) ), 0)) * GREATEST (t1.c0 - t1.c2, 1) )) 14FROMt0 15CROSS JOINt8 16CROSS JOINt1 17CROSS JOINt3 18CROSS JOINvtable0 19LIMIT0; Query 7 of Argus 1SELECTt8.c2, 2t8.c3, 3t8.c0, 4t0.c2, 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:36 Qiuyang Mang et al. 5t0.c1, 6vtable0.c1, 7vtable0.c0, 8((CASEWHEN( t0.c0LIKE'%test%')THENt0.c1ELSEt0.c0END)), 9t0.c3, 10vtable0.c3, 11(ascii(SUBSTRING(( ( (t0.c1'_'( reverse(substring(( regexp_replace(t0.c1, '[0-9]','X','g') )from1 for 3))::varchar))::varchar)'_'t0.c0 )FROM1 FOR 1)) *CASEWHEN( regexp_like(t0.c0,'test$','i') )THEN1ELSE2END), 12(bit_length(( replace(lower(t8.c2),'a','@') to_char(t8.c2,'FM9999') ))), 13(( split_part(t8.c0':'( md5(t8.c0 t8.c3)::varchar),':', 2) ) ~'^[0-9]+ $'ANDlength(( md5(t8.c0 t8.c3)::varchar)) > 0) 14FROMt8 15CROSS JOINt0 16JOINvtable0 17ON((ascii(substring(t0.c0FROM1 FOR 1)) * (length(coalesce((translate(t0.c0 t0. c1,'aeiou','AEIOU') ),'')) + 1)) + vtable0.c1) = ((ascii(substring(t0.c0FROM1 FOR 1)) * (length(coalesce((translate(t0.c0 t0.c1,'aeiou','AEIOU') ),'')) + 1) ) + -492369572) 18WHEREvtable0.c3 ISNULL 19OR(ascii(SUBSTRING(( ( (t0.c1'_'( reverse(substring(( regexp_replace(t0.c1,' [0-9]','X','g') )from1 for 3))::varchar))::varchar)'_'t0.c0 )FROM1 FOR 1)) *CASEWHEN( regexp_like(t0.c0,'test$','i') )THEN1ELSE2END) ISNULL 20UNION ALL 21SELECTt8.c2, 22t8.c3, 23t8.c0, 24t0.c2, 25t0.c1, 26NULL, 27NULL, 28((CASEWHEN( t0.c0LIKE'%test%')THENt0.c1ELSEt0.c0END)), 29t0.c3, 30NULL, 31(ascii(SUBSTRING(( ( (t0.c1'_'( reverse(substring(( regexp_replace(t0.c1, '[0-9]','X','g') )from1 for 3))::varchar))::varchar)'_'t0.c0 )FROM1 FOR 1)) *CASEWHEN( regexp_like(t0.c0,'test$','i') )THEN1ELSE2END), 32(bit_length(( replace(lower(t8.c2),'a','@') to_char(t8.c2,'FM9999') ))), 33(( split_part(t8.c0':'( md5(t8.c0 t8.c3)::varchar),':', 2) ) ~'^[0-9]+ $'ANDlength(( md5(t8.c0 t8.c3)::varchar)) > 0) 34FROMt8 35CROSS JOINt0 36WHERE NOTEXISTS( 37SELECT1 38FROMvtable0 39WHERE((ascii(substring(t0.c0FROM1 FOR 1)) * (length(coalesce((translate(t0.c0 t0.c1,'aeiou','AEIOU') ),'')) + 1)) + vtable0.c1) = ((ascii(substring(t0.c0FROM 1 FOR 1)) * (length(coalesce((translate(t0.c0 t0.c1,'aeiou','AEIOU') ),'')) + 1)) + -492369572) 40); 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:37 Query 8 of Argus 1SELECT(char_length(t0.c1 t0.c0)), 2(substring(t1.c0from'[A-Za-z]+')'-'to_hex(( (t1.c2 / GREATEST(1, t1.c0)) % 7 ))), 3t1.c3, 4(t3.c3 % 2 = 0), 5((t0.c1'-'(translate(t0.c1,'aeiou','12345') ))), 6vtable0.c2, 7vtable0.c1, 8t1.c0, 9vtable0.c3, 10vtable1.c2, 11(char_length(t0.c0 t0.c1)), 12t1.c2, 13t3.c0 14FROMt0, t1, vtable0, vtable1, t3 15WHERE FALSE; Query 9 of Argus 1SELECT 2vtable0.c3, 3t3.c3, 4((t3.c3::text) SIMILAR TO'[0-9]+'), 5vtable0.c0, 6(mod(coalesce(( (bit_length(t8.c1::bytea) + ascii(substring(( REGEXP_REPLACE(t8.c1, 'a','b','g') )from1 for 1))) ),0) *coalesce(( regexp_count(( to_char(DATE' 2021-01-01'+ t8.c1 *INTERVAL'1 day','YYYY-MM-DD') ),'\\\\w+', 1,'g') ),1), 10) + regexp_instr(t8.c1,'[0-9]')), 7t1.c3, 8vtable0.c2, 9t8.c0, 10(CASEWHEN( t0.c1 @@ plainto_tsquery(t0.c0) )THEN( get_byte(t0.c0::bytea, 1) + get_byte(t0.c1::bytea, 0) )ELSE( regexp_count(t0.c1,'\\\\w+', 1,'g') )END), 11(translate(( to_char(t3.c3,'FM0000')::varchar),'abc','ABC')::varchar), 12((CASE((((t0.c1)(t0.c0)))((CASEt0.c0WHENt0.c1THENt0.c1END)))WHEN(((('')(' ')))(((NULL)(''))))THEN((((1574883451)%(NULL)))*((CASE NULL WHEN'+~'THEN 315562183END)))END)), 13t8.c1, 14t0.c2, 15vtable1.c1, 16vtable1.c2, 17t8.c2, 18t3.c0, 19vtable0.c1, 20t0.c0, 21t0.c1, 22(( xml_is_well_formed(t8.c3) )OR NOT( (( floor(((coalesce(substring(t8.c3fromE '(\\\\d+)'),'0')::int+ t8.c2 )::numeric/ greatest(1, t8.c2 + 1)))::int) % 2 = 0) )), 23(t3.c3AND( (t3.c3 % 2 = 0) )OR NOTt3.c3), 24(t0.c0'-'t0.c1), 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:38 Qiuyang Mang et al. 25(family('127.0.0.1'::inet) + masklen('127.0.0.1/24'::inet) - t8.c2), 26t1.c2, 27t0.c3, 28(UPPER(SUBSTRING(t1.c1FROM1 FORGREATEST(3, LENGTH(t1.c1) % 5)))), 29t8.c3, 30vtable1.c0, 31t1.c1, 32(bit_count(((EXTRACT(YEAR FROM DATE'1990-01-01'+ ( length(t1.c3) + t1.c0 * 2 floor(t1.c3::double precision /nullif(t1.c0,1))::int) *INTERVAL'1 day')::INT) + t1.c3)::bit(16))::int), 33t3.c2, 34((t1.c2::text) SIMILAR TO'[0-9]+'), 35vtable1.c3, 36t3.c1 37FROM 38vtable0 39LEFT JOINt8ONvtable0.c0 40CROSS JOINt3 41CROSS JOINvtable1 42CROSS JOINt1 43CROSS JOINt0 44WHERE 45((t0.c2ORt0.c3)ANDt0.c3) = (t0.c3 < t0.c2) 46; Query 10 of Argus 1SELECTv.c0, v.c2 2FROMvtable0 v 3CROSS JOINLATERAL (SELECT1FROMt1) l1 4CROSS JOINLATERAL (SELECT1FROMt8) l2 5CROSS JOINLATERAL (SELECT1FROMt0WHEREc2 ISNULL) l3; Query 1 of SQLancer 1SELECT ALL* 2FROMt1INNER 3JOIN(SELECT- (+ (t0.c0)) 4FROM ONLYt0,ONLYt1 5WHERE TRUE)ASsub0ON TRUE; Query 2 of SQLancer 1SELECTt0.c0, t1.c0 2FROMt0* 3FULL OUTER JOIN ONLYt1ON lower((0.961022777314273)::VARCHAR(181))~*lower(CAST(TRUE AS VARCHAR)); 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:39 Query 3 of SQLancer 1SELECT* 2FROMt1* 3WHERE TRUE 4UNION ALL SELECT ALL* 5FROMt1* 6WHERE NOT(TRUE) 7UNION ALL SELECT ALL* 8FROMt1 9WHERE(TRUE) ISNULL; Query 4 of SQLancer 1SELECTt1.c0 2FROMt1; Query 5 of SQLancer 1SELECTt0.c0, t1.c0 2FROMt1* 3LEFT OUTER JOINt0ONinet_same_family('179.19.19.249','240.210.141.13'); Query 6 of SQLancer 1SELECT ALLt1.c0 2FROMt1* 3CROSS JOIN(SELECT ALL(t1.c0)BETWEEN(t0.c0)AND(num_nonnulls('')), (family(' 118.88.171.159'))IN(t1.c0, t0.c0, t1.c0),CAST(0.07644869ASMONEY) 4FROM ONLYt0, t1)ASsub0; Query 7 of SQLancer 1SELECT ALLt1.c0 2FROM ONLYt1 3WHERE((0.021889338)::MONEY) ISNULL 4UNION ALL SELECTt1.c0 5FROM ONLYt1 6WHERE NOT(((0.021889338)::MONEY) ISNULL) 7UNION ALL SELECTt1.c0 8FROM ONLYt1 9WHERE(((0.021889338)::MONEY) ISNULL) ISNULL; 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:40 Qiuyang Mang et al. Query 8 of SQLancer 1SELECT ALLt0.c0 2FROMt1 3LEFT OUTER JOIN ONLYt0ON NOT((((((t0.c0)%(t1.c0)))::VARCHAR)LIKE(((('{zq.O]w\\n'):: VARCHAR(701))(rtrim('0.7364646181551185')))))); Query 9 of SQLancer 1SELECT ALLt1.c0 2FROMt1INNER 3JOINt0ON(0.506531)IN(t1.c0, 0.46897623, t1.c0); Query 10 of SQLancer 1SELECT ALLt0.c0 2FROM ONLYt0; Received October 2025; revised January 2026; accepted February 2026 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 47]",
+      "technique": "norec",
+      "sentence": "Logic bugs are particularly insidious; they cause a DBMS to return incorrect results without raising errors, therefore silently corrupting downstream applications [ 47].",
+      "context_before": "/3802017 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026.arXiv:2510.06663v2 [cs.DB] 24 Mar 2026 140:2 Qiuyang Mang et al. 1CREATE TABLEt1(cINT); 2INSERT INTOt1VALUES(1); 3SELECTc / 3FROMt1WHERE false; -- {}/ 4SELECTc / 3FROMt1EXCEPT SELECTc / 3FROMt1; 5-- {0.3333}ὁB Listing 1. Motivating bug: incorrect EXCEPT result in TiDB, which can only be detected by a specific test oracle. 1 Introduction Database Management Systems (DBMSs) are a foundational component of modern software, yet their complexity makes them prone to bugs that can compromise application behavior and data integrity.",
+      "context_after": "In response, the research community has developed automated testing techniques [ 24,47–49,74] that have discovered hundreds of bugs in real-world systems. The most critical component of these techniques is thetest oracle[ 6] that can determine the correctness of a query’s output without access to the ground truth. Given a query, oracles for DBMS testing implement a transformation mechanism to generate a semantically equivalent variant. Then, by executing both queries and checking for result consistency, these oracles can detect logic bugs. For example, Ternary Logic Partitioning (TLP) [ 48] i",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3873,
+      "cited_reference": {
+        "number": 47,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 24,47–49,74]",
+      "technique": "norec",
+      "sentence": "In response, the research community has developed automated testing techniques [ 24,47–49,74] that have discovered hundreds of bugs in real-world systems.",
+      "context_before": "t1VALUES(1); 3SELECTc / 3FROMt1WHERE false; -- {}/ 4SELECTc / 3FROMt1EXCEPT SELECTc / 3FROMt1; 5-- {0.3333}ὁB Listing 1. Motivating bug: incorrect EXCEPT result in TiDB, which can only be detected by a specific test oracle. 1 Introduction Database Management Systems (DBMSs) are a foundational component of modern software, yet their complexity makes them prone to bugs that can compromise application behavior and data integrity. Logic bugs are particularly insidious; they cause a DBMS to return incorrect results without raising errors, therefore silently corrupting downstream applications [ 47].",
+      "context_after": "The most critical component of these techniques is thetest oracle[ 6] that can determine the correctness of a query’s output without access to the ground truth. Given a query, oracles for DBMS testing implement a transformation mechanism to generate a semantically equivalent variant. Then, by executing both queries and checking for result consistency, these oracles can detect logic bugs. For example, Ternary Logic Partitioning (TLP) [ 48] is a highly effective oracle that partitions a query𝑄based on a predicate 𝑃and then checks equivalence between 𝑄and the union of its three-way partition 𝑄WH",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4043,
+      "cited_reference": {
+        "number": 47,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "For example, Ternary Logic Partitioning (TLP) [ 48] is a highly effective oracle that partitions a query𝑄based on a predicate 𝑃and then checks equivalence between 𝑄and the union of its three-way partition 𝑄WHERE𝑃 ,𝑄WHERE NOT𝑃, and𝑄WHERE𝑃IS NULL.",
+      "context_before": "ore silently corrupting downstream applications [ 47]. In response, the research community has developed automated testing techniques [ 24,47–49,74] that have discovered hundreds of bugs in real-world systems. The most critical component of these techniques is thetest oracle[ 6] that can determine the correctness of a query’s output without access to the ground truth. Given a query, oracles for DBMS testing implement a transformation mechanism to generate a semantically equivalent variant. Then, by executing both queries and checking for result consistency, these oracles can detect logic bugs.",
+      "context_after": "If a DBMS returns different results for the original query and its partitioned version, that indicates a bug. While human-designed oracles have found many bugs (reported in over 20 top-conference papers), the manual creation of oracles is a bottleneck, trapping researchers in a cycle of designing increasingly specialized oracles to find bugs missed by previous ones. For example, a TiDB [ 22] bug introduced in 2019 Listing 1 went undetected for years despite extensive testing1, because it required a specific oracle to check that any query 𝑄EXCEPT𝑄 should yield an empty result. This example hig",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4589,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Compared to traditional SQL generators [ 16,53], this is infeasible for modern DBMS testing, which often requires executing thousands of queries per minute to find bugs efficiently, such as in SQLancer [3].",
+      "context_before": "example highlights a fundamental challenge: manually designed oracles are not only difficult to conceive but also tend to overlook bugs. Automating the discovery of test oracles is therefore essential to enable scalable DBMS testing. Large Language Models (LLMs), with their success in code generation [ 20,41], offer a promising avenue for automating this process. A naive approach would be to prompt an LLM to generate a semantically equivalent variant for a given seed SQL query. However, this strategy suffers from two limitations: (1)Scalability:LLM invocations are costly and have high latency.",
+      "context_after": "(2)Soundness:LLMs are prone to hallucination and may generate query pairs that are not truly equivalent. Such unsound oracles produce false positives, thereby undermining the reliability of the testing process. Note that verifying equivalence empirically by running the query pairs on multiple database instances is unreliable, as this cannot guarantee the complete removal of semantically inequivalent pairs [ 19], and may also filter out true bugs that might exist on all databases under test [56]. 1Commit 7de6200, introduced in 2019. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026.",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6024,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "An example of representing and instantiating TLP [48] oracle in CAQ.",
+      "context_before": "er test [56]. 1Commit 7de6200, introduced in 2019. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:3 1CREATE TABLEt1(c0VARCHAR, ...); 2CREATE TABLEt2(...); 3SELECT*FROMt1,□ 1⊲Table(...); --𝑄 1 4SELECT*FROMt1,□ 1⊲Table(...) 5WHERE(□ 2⊲Expr(t1:BOOLEAN)IS TRUE)UNION ALL 6SELECT*FROMt1,□ 1⊲Table(...) 7WHERE(□ 2⊲Expr(t1:BOOLEAN)IS FALSE)UNION ALL 8SELECT*FROMt1,□ 1⊲Table(...) 9WHERE(□ 2⊲Expr(t1:BOOLEAN)IS NULL); --𝑄 2 10□ 1⊲Table(...)↦→t1 ASOF JOIN t2 11□ 2⊲Expr(t1:BOOLEAN)↦→json_valid(t1.c0) Listing 2.",
+      "context_after": "Key insights.To address these challenges, we proposeArgus, a novel, fully automated framework for finding logic bugs in DBMSs.Argususes a two-stage process that separates oracle discovery from test case generation. First, it leverages an LLM in an offline phase to discover reusable test oracles. Then, these oracles are formally verified for correctness before being instantiated into thousands of concrete test cases. Since generating abstract test oracles is much cheaper and faster than directly using LLMs to craft concrete test cases, this approach tackles both scalability and soundness head-",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 7319,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For instance, TLP [ 48] can be represented as an equivalent CAQ pair,i.",
+      "context_before": "are also associated with the corresponding constraints, which are essential to implement test oracles. For example, a CAQ with two placeholders might specify that placeholders must be syntactically identical, or that a predicate must be a Boolean expression when instantiated. Anequivalent CAQ pairconsists of two CAQs that are semantically identical for every possible instantiation of placeholders,i.e.,for all possible fillings of the placeholders with concrete SQL snippets that satisfy their constraints. CAQs allowArgusto represent a wide range of test oracles, including several existing ones.",
+      "context_after": "e., 𝑄1and𝑄2, as shown in Listing 2. In this representation, 𝑄1serves as the original query template, while 𝑄2is its three-way partitioned version. Here, to ensure the syntax validity of the instantiated queries from the two CAQs, □1is restricted to represent any table, and □2must be a Boolean expression constructed using the columns of table 𝑡1. These two CAQs are semantically equivalent for all valid instantiations of the placeholders that satisfy their constraints, where□ 1and□ 2should also be instantiated consistently in both queries. Test oracle generation, verification, and instantiation.",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 9062,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "These snippets can be generated offline by a hybrid approach combining an LLM to cover diverse database features and a high-throughput generator, such as SQLancer [ 49], even though the prover currently cannot reason about them.",
+      "context_before": "Fig. 1. Overall pipeline of Argus. To tackle thescalability challenge, each verified oracle is instantiated into thousands of concrete test cases. Specifically, we populate the placeholders in each CAQ pair from a large corpus of pre-generated SQL snippets. This approach allows us to generate test cases for novel and complex features provided by the DBMS under test, even though they are not supported by our prover. For instance, in Listing 2, the placeholders □1and□2can be instantiated with various table joins and Boolean expressions, respectively, such as t1 ASOF JOIN t2 andjson_valid(t1.c0).",
+      "context_after": "Evaluation.We have implementedArgusand evaluated it on five widely used and comprehensively tested DBMSs: Dolt [ 14], DuckDB [ 46], MySQL [ 40], PostgreSQL [ 38], and TiDB [ 22].Argus discovered 41 unique and previously unknown bugs. Of these, 36 have been confirmed and 27 have been fixed by the developers. The bugs discovered include 36 critical logic bugs that lead to incorrect query results, while the remaining 5 cause performance problems or crashes. In our empirical comparisons with state-of-the-art open source testing tools in DuckDB [46],Argusdemonstrated an improvement of up to 1.19 ×",
+      "section": "1 Introduction",
+      "page": 4,
+      "char_offset": 12317,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 2]",
+      "technique": null,
+      "sentence": "43 ×in metamorphic coverage [ 2], a recently developed coverage indicator to assess the ability to find logic bugs.",
+      "context_before": "ementedArgusand evaluated it on five widely used and comprehensively tested DBMSs: Dolt [ 14], DuckDB [ 46], MySQL [ 40], PostgreSQL [ 38], and TiDB [ 22].Argus discovered 41 unique and previously unknown bugs. Of these, 36 have been confirmed and 27 have been fixed by the developers. The bugs discovered include 36 critical logic bugs that lead to incorrect query results, while the remaining 5 cause performance problems or crashes. In our empirical comparisons with state-of-the-art open source testing tools in DuckDB [46],Argusdemonstrated an improvement of up to 1.19 ×in code coverage, and 6.",
+      "context_after": "In addition, we also compared Arguswith a union of existing test oracles, rewriting them into equivalent CAQ pairs equipped with the same query generator. The results show thatArgus’s new oracles can detect 3.33 ×more unique logic bugs than the sum of prior oracles from these works [5, 24, 47, 48] within 6 hours of testing on Dolt [ 14]. Our ablation studies further validate the effectiveness of the SQL equivalence prover in ensuring soundness and the contribution of CAQ to enhancing scalability. This paper makes the following contributions. •We introduce the concept of theequivalent CAQ pair",
+      "section": "1 Introduction",
+      "page": 4,
+      "char_offset": 13170,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba, Yuancheng Jiang, and Manuel Rigger. 2025. Metamorphic Coverage.arXiv preprint arXiv:2508.16307 (2025).",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker",
+      "surface": "[5, 24, 47, 48]",
+      "technique": "dqp",
+      "sentence": "33 ×more unique logic bugs than the sum of prior oracles from these works [5, 24, 47, 48] within 6 hours of testing on Dolt [ 14].",
+      "context_before": "logic bugs that lead to incorrect query results, while the remaining 5 cause performance problems or crashes. In our empirical comparisons with state-of-the-art open source testing tools in DuckDB [46],Argusdemonstrated an improvement of up to 1.19 ×in code coverage, and 6.43 ×in metamorphic coverage [ 2], a recently developed coverage indicator to assess the ability to find logic bugs. In addition, we also compared Arguswith a union of existing test oracles, rewriting them into equivalent CAQ pairs equipped with the same query generator. The results show thatArgus’s new oracles can detect 3.",
+      "context_after": "Our ablation studies further validate the effectiveness of the SQL equivalence prover in ensuring soundness and the contribution of CAQ to enhancing scalability. This paper makes the following contributions. •We introduce the concept of theequivalent CAQ pair— a novel and expressive representation for test oracles in DBMSs — that can also capture features not supported by existing SQL equivalence provers. Based on that, we propose a new methodology that leverages LLMs to automatically discover test oracles for DBMSs by using them to generate CAQs. To our knowledge, we are the first tool that",
+      "section": "1 Introduction",
+      "page": 4,
+      "char_offset": 13495,
+      "cited_reference": {
+        "number": 5,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans.Proceedings of the ACM on Management of Data2, 3 (2024), 1–26.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "In the context of DBMS testing, test oracles typically operate by transforming a given SQL query into a semantically equivalent variant, as in TLP [ 48], NoREC [ 47], and EET [ 24].",
+      "context_before": "which can then be used to efficiently instantiate equivalent CAQ pairs into numerous concrete tests. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:5 •We perform anextensive evaluationon five widely tested DBMSs, uncovering 41 previously unknown bugs, which outperforms the state-of-the-art techniques in several coverage metrics and the number of unique logic bugs found. 2 Background Test oracle.A test oracle is a mechanism for verifying whether a system’s output is correct for a given input [ 6].",
+      "context_after": "By doing so, the oracle can compare the execution results of the original and transformed queries to detect potential logic bugs. In this work, CAQ provides a unified framework for LLMs to generate test oracles automatically. We show that most of these prior oracles can be formalized as instances of equivalent CAQ pairs, such as that shown in Listing 2. We include more examples in Appendix C. SQL equivalence verification.The goal of SQL equivalence verification is to determine if two SQL queries are semantically equivalent, meaning they produce the same result when executed on all possible in",
+      "section": "2 Background",
+      "page": 5,
+      "char_offset": 15050,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer++",
+      "technique": null,
+      "sentence": "Specifically, we use SQLancer++ [ 82]’s query generator to produce seed queries, but we do not use their predefined test oracles (such as TLP [48] and NoREC [47]) for bug detection.",
+      "context_before": "ck DB [ 46] as a case study to provide an overview of theArgusframework. Fig. 1 presents the overall pipeline ofArgus, which consists of two stages:Test Oracle Discovery (①and②) andTest Cases Instantiation( ③–⑥). The first stage aims to automatically discover a large number of high-quality test oracles in the form of equivalent CAQ pairs, while the second stage focuses on deriving concrete test cases from these CAQ pairs to detect bugs in DBMSs. In①, we generate a set of CAQs with their associated schemas by a grammar-based generator without using LLMs, and this will be detailed in Section 5.1.",
+      "context_after": "For example, we generate the following CAQ: SELECTt2.c0FROMt2, t3LEFT JOINt1 ON□ 1⊲Expr(t1:BOOLEAN); After that, in ②these CAQs will serve as seed queries for the subsequent LLM-based test oracle discovery. First, we employ an iterative prompting strategy to guide LLMs to generate various equivalent CAQs for each seed query, which will be elaborated in Section 5.2. For instance, given the above CAQ as input, the LLM may generate the following CAQs: 𝐶1:WITHcAS(SELECT*FROMt1WHERE□ 1⊲Expr(t1:BOOLEAN)) SELECTt2.c0FROMt2CROSS JOINt3CROSS JOINc 4, No. 3 (SIGMOD), Article 140. Publication date: Jun",
+      "section": "3 Argus Overview",
+      "page": 5,
+      "char_offset": 17535,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[ 3,24,48,74]",
+      "technique": "qpg",
+      "sentence": "Note that these are also widely adopted constraints in prior work on manually designed test oracles [ 3,24,48,74].",
+      "context_before": "⊲𝑐𝑘}, aninstantiationof 𝑞, denoted𝑞∗, is obtained by replacing each placeholder □𝑖in𝑞with a concrete SQL snippet 𝑠𝑖that satisfies the corresponding constraint 𝑐𝑖and a set of general constraintsC. The substitution can be written as: 𝑞∗=𝑞[□ 1↦→𝑠 1,□2↦→𝑠 2,...,□ 𝑘↦→𝑠 𝑘] Specifically, each 𝑐𝑖refers to theConstraints shown in Fig. 2. In addition, to prevent false positive bug reports when using the instantiated query 𝑞∗for testing, we must also satisfy a set of general constraintsC, such as avoiding non-deterministic features,e.g., RANDOM and CURRENT_TIMESTAMP, and ensuring the snippet is valid SQL.",
+      "context_after": "The details ofCwill be discussed in Sec. 6.2. We say that two CAQs are equivalent if all their possible instantiations are semantically equivalent. Definition 4.2(Equivalent CAQ Pair).Given two CAQ s𝑞1and𝑞2defined over the same Schema𝑠 (which specifies the tables referenced in both queries) with the same PlaceholderMap{□ 1⊲𝑐1,□2⊲ 𝑐2,...,□ 𝑘⊲𝑐𝑘}, we say that 𝑞1and𝑞2form anEquivalent CAQ Pair, denoted (𝑠,𝑞 1,𝑞2), if for every possible instantiated concrete queries (Definition 4.1), 𝑞∗ 1and𝑞∗ 2, we have𝑞∗ 1≡𝑞∗ 2,i.e.,they will return the same result set when executed on any database instance con",
+      "section": "4 Constrained Abstract Query",
+      "page": 7,
+      "char_offset": 24560,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ",SQLancer [3]).",
+      "context_before": "rs by instantiating the CAQs with various snippets that satisfy the constraints of the placeholders. This allows us to decompose the test oracle discovery problem to CAQ pair generation and equivalence checking problems. 5 Test Oracle Discovery We now present our test oracle discovery algorithm, outlined in Algorithm 1 and consists of three phases: database seeding, CAQ pairs generation, and equivalence checking. It leverages four key components as input: a target DBMS D(e.g.,DuckDB [ 46]), an LLMM(e.g.,GPT o4-mini [ 42]), a SQL proverP(e.g.,SQLSolver [13]), and a grammar-based generatorG(e.g.",
+      "context_after": "5.1 Database Seeding To discover diverse test oracles, we generate CAQ pairs with multiple database schemas. As equivalent CAQ pairs are associated with their database schemas, we first iterate 𝑁times to generate diverse schemas and their corresponding CAQs (line 18). In each iteration, we start by database seeding, which involves generating a random database schema (line 19), and then producing a CAQ that conforms to the generated schema,i.e.,the seed query 𝑞(line 20). Though existing grammar-based generators,e.g.,SQLancer can be directly used for schema and concrete 4, No. 3 (SIGMOD), Arti",
+      "section": "5 Test Oracle Discovery",
+      "page": 8,
+      "char_offset": 26909,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ",SQLancer can be directly used for schema and concrete 4, No.",
+      "context_before": "roverP(e.g.,SQLSolver [13]), and a grammar-based generatorG(e.g.,SQLancer [3]). 5.1 Database Seeding To discover diverse test oracles, we generate CAQ pairs with multiple database schemas. As equivalent CAQ pairs are associated with their database schemas, we first iterate 𝑁times to generate diverse schemas and their corresponding CAQs (line 18). In each iteration, we start by database seeding, which involves generating a random database schema (line 19), and then producing a CAQ that conforms to the generated schema,i.e.,the seed query 𝑞(line 20). Though existing grammar-based generators,e.g.",
+      "context_after": "3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:9 1CREATE TABLEt1(c0INT, placeholder1BOOLEAN); 2CREATE TABLEt2(c0INT); 3CREATE TABLEvtable1(c0INT, c1INT); 4SELECTt1.placeholder1FROMt1, vtable1; Listing 3. An example of using concrete SQL query to represent CAQ when interfacing with grammar-based generators and provers. query generation, they cannot directly produce queries with placeholders. The key challenge lies in producing those entries in the seed CAQ while ensuring the syntactic correctness after in",
+      "section": "5.1 Database Seeding",
+      "page": 8,
+      "char_offset": 27445,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[ 3,5,74]",
+      "technique": "qpg",
+      "sentence": "Instead of pursuing syntactic differences, our goal is to produce queries with variedquery plans, as this is a key principle to creating effective test oracles in previous works [ 3,5,74].",
+      "context_before": "ivalent queries. To generate variants efficiently, our key strategy is to leverage in-context learning by providing the LLM with not only the seed query 𝑞, but also carefully selected samples from two dynamically updated sets: Equal ,i.e.,queries previously verified as equivalent to 𝑞, and Fail,i.e.,those that failed verification. The Equal set promotes both correctness and diversity, while theFailset provides examples of failures to avoid. A crucial aspect of our approach is how we guide the LLM towards generating diverse CAQs to trigger different optimization paths and detect potential bugs.",
+      "context_after": "Qualitatively, we prompt the LLM to generate novel queries that are different from the provided examples (line 4). In addition, we also quantitatively measure the similarity between query plans using the tree edit distance [ 75]. This metric is then used to guide the selection of examples for the next iteration’s prompt. Specifically, we use k-means to cluster the CAQs in the Equal set based on the tree edit distance value and then sample queries from each cluster. This strategy ensures the LLM is consistently shown a diverse range of successful query structures, pushing it to explore novel e",
+      "section": "5.2 CAQ Pairs Generation",
+      "page": 9,
+      "char_offset": 29944,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[ 47–49]",
+      "technique": "norec",
+      "sentence": "The first constraint is also widely used in previous works [ 47–49] and can be checked by applying a regular expression to identify non-deterministic features.",
+      "context_before": "we enforce the following general constraints C(defined in Definition 4.2) on the sampled query snippets. (1)Determinism( Table andExpression ): The snippet should not contain any non-deterministic functions, such as RANDOM(). (2)Null-preserving( Expression ): The expression should preserve a null return value when evaluated on rows only containing null values. For example, c1 + c2 is null-preserving, while IFNULL(c1, 0)is not. (3)Empty Results-preserving( Expression ): The expression should return an empty result set when evaluated on an empty table. For example,sum(c1)is not empty-preserving.",
+      "context_after": "The second and third constraints resolve the semantic misalignment that arises when virtual columns and tables are instantiated with concrete expressions and tables. These constraints are necessary because, in the presence of outer joins, virtual columns may no longer faithfully capture the semantics of concrete expressions after instantiation.e.g.,outer joins can introduce NULL values into virtual columns, causing instantiated expressions to be evaluated differently from the version fed to the SQL equivalence prover. After satisfying these constraints, we can ensure that the instantiated que",
+      "section": "6.2 Query Instantiation",
+      "page": 13,
+      "char_offset": 42577,
+      "cited_reference": {
+        "number": 47,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[ 47–49]",
+      "technique": "norec",
+      "sentence": "We basically follow the widely-used random data generation method [ 47–49] to populate the tables in the instantiated schema 𝑠with random tuples, and then create random indices on those tables to diversify the execution plans.",
+      "context_before": "ain a database instance under the schema used in the prover where the two queries return different results, which contradicts the original semantic equivalence. A detailed analysis of the counterexamples, together with the full proof for the cases satisfying them, is provided in Appendix B. Furthermore, we validate these constraints at runtime: we execute each snippet on Dand check the returned types or schemas and behavior on NULLor empty inputs. 6.3 Database Instantiation and Bug Reporting After instantiating the test queries, we create the database instances to execute them (line 20 and 21).",
+      "context_after": "Finally, we execute the instantiated queries 𝑞∗and𝑞′∗on the target DBMS Dand compare their results (line 20). If the results differ, we report a logic bug along with the instantiated test case, database schema, and table data. In addition, if a query causes a crash of D, we also report a crash bug. Note that, though our test oracles can also detect performance issues similar to [ 25,36], we do not detect them in a large scale due to those issues are typically regarded as expected behaviors by developers [4]. 7 Evaluation In this section, we evaluateArguson five aspects: (1)How many new bugs c",
+      "section": "6.3 Database Instantiation and Bug Reporting",
+      "page": 13,
+      "char_offset": 44313,
+      "cited_reference": {
+        "number": 47,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "citation_marker",
+      "surface": "[4]",
+      "technique": "cert",
+      "sentence": "Note that, though our test oracles can also detect performance issues similar to [ 25,36], we do not detect them in a large scale due to those issues are typically regarded as expected behaviors by developers [4].",
+      "context_before": "eries, we create the database instances to execute them (line 20 and 21). We basically follow the widely-used random data generation method [ 47–49] to populate the tables in the instantiated schema 𝑠with random tuples, and then create random indices on those tables to diversify the execution plans. Finally, we execute the instantiated queries 𝑞∗and𝑞′∗on the target DBMS Dand compare their results (line 20). If the results differ, we report a logic bug along with the instantiated test case, database schema, and table data. In addition, if a query causes a crash of D, we also report a crash bug.",
+      "context_after": "7 Evaluation In this section, we evaluateArguson five aspects: (1)How many new bugs canArgusfind in real-world DBMSs, and what LLM-generated test oracles discover them? (Sec. 7.1) (2) How doesArguscompare to other baseline tools in terms of code coverage? (Sec. 7.2) (3)How does the number of automatically discovered oracles affect the number of unique bugs found, compared to prior work where oracles are manually designed? (Sec. 7.3) 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:14 Qiuyang Mang et al. DBMS Tested by GitHub stars Released LOC Dolt [82] 19.1k 2018 380k DuckDB",
+      "section": "6.3 Database Instantiation and Bug Reporting",
+      "page": 13,
+      "char_offset": 44840,
+      "cited_reference": {
+        "number": 4,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "citation_marker",
+      "surface": "[82]",
+      "technique": null,
+      "sentence": "DBMS Tested by GitHub stars Released LOC Dolt [82] 19.",
+      "context_before": "ose issues are typically regarded as expected behaviors by developers [4]. 7 Evaluation In this section, we evaluateArguson five aspects: (1)How many new bugs canArgusfind in real-world DBMSs, and what LLM-generated test oracles discover them? (Sec. 7.1) (2) How doesArguscompare to other baseline tools in terms of code coverage? (Sec. 7.2) (3)How does the number of automatically discovered oracles affect the number of unique bugs found, compared to prior work where oracles are manually designed? (Sec. 7.3) 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:14 Qiuyang Mang et al.",
+      "context_after": "1k 2018 380k DuckDB [15, 16, 48, 74, 82] 32.7k 2019 1,496k MySQL [24, 31, 49, 57, 58, 60, 61, 74, 80, 82] 11.7k 1995 5,532k PostgreSQL [16, 24, 30, 47, 49, 58, 80, 82] 18.5k 1995 938k TiDB [3, 24, 57, 58, 60, 61, 74, 82] 39.0k 2016 1,398k Table 1. DBMSs under test by Argus. DBMS ReportedBug status Bug type Fixed Conf. Dup. Pend. Logic Other Dolt 19 18 1 0 0 18 1 DuckDB 8 6 0 1 1 4 4 MySQL 8 0 5 1 2 8 0 PostgreSQL 1 1 0 0 0 1 0 TiDB 5 2 3 0 0 5 0 Total41 27 9 2 3 36 5 Table 2. Argus found bugs statistics. (4) How effective isArgus’s SQL equivalence prover in filtering out false positives? (Sec.",
+      "section": "7 Evaluation",
+      "page": 14,
+      "char_offset": 45580,
+      "cited_reference": {
+        "number": 82,
+        "text": "Suyang Zhong and Manuel Rigger. 2026. Scaling Automated Database System Testing. InProceedings of the 31st ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2(USA) (ASPLOS ’26). Association for Computing Machinery, New York, NY, USA, 1677–1",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing sqlancer_pp",
+          "prints the DOI of the paper introducing sqlancer_pp"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "citation_marker",
+      "surface": "[15, 16, 48, 74, 82]",
+      "technique": "tlp",
+      "sentence": "1k 2018 380k DuckDB [15, 16, 48, 74, 82] 32.",
+      "context_before": "by developers [4]. 7 Evaluation In this section, we evaluateArguson five aspects: (1)How many new bugs canArgusfind in real-world DBMSs, and what LLM-generated test oracles discover them? (Sec. 7.1) (2) How doesArguscompare to other baseline tools in terms of code coverage? (Sec. 7.2) (3)How does the number of automatically discovered oracles affect the number of unique bugs found, compared to prior work where oracles are manually designed? (Sec. 7.3) 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:14 Qiuyang Mang et al. DBMS Tested by GitHub stars Released LOC Dolt [82] 19.",
+      "context_after": "7k 2019 1,496k MySQL [24, 31, 49, 57, 58, 60, 61, 74, 80, 82] 11.7k 1995 5,532k PostgreSQL [16, 24, 30, 47, 49, 58, 80, 82] 18.5k 1995 938k TiDB [3, 24, 57, 58, 60, 61, 74, 82] 39.0k 2016 1,398k Table 1. DBMSs under test by Argus. DBMS ReportedBug status Bug type Fixed Conf. Dup. Pend. Logic Other Dolt 19 18 1 0 0 18 1 DuckDB 8 6 0 1 1 4 4 MySQL 8 0 5 1 2 8 0 PostgreSQL 1 1 0 0 0 1 0 TiDB 5 2 3 0 0 5 0 Total41 27 9 2 3 36 5 Table 2. Argus found bugs statistics. (4) How effective isArgus’s SQL equivalence prover in filtering out false positives? (Sec. 7.4) (5)What are the time and monetary cost",
+      "section": "7 Evaluation",
+      "page": 14,
+      "char_offset": 45635,
+      "cited_reference": {
+        "number": 48,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "citation_marker",
+      "surface": "[24, 31, 49, 57, 58, 60, 61, 74, 80, 82]",
+      "technique": "pqs",
+      "sentence": "7k 2019 1,496k MySQL [24, 31, 49, 57, 58, 60, 61, 74, 80, 82] 11.",
+      "context_before": "tion, we evaluateArguson five aspects: (1)How many new bugs canArgusfind in real-world DBMSs, and what LLM-generated test oracles discover them? (Sec. 7.1) (2) How doesArguscompare to other baseline tools in terms of code coverage? (Sec. 7.2) (3)How does the number of automatically discovered oracles affect the number of unique bugs found, compared to prior work where oracles are manually designed? (Sec. 7.3) 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:14 Qiuyang Mang et al. DBMS Tested by GitHub stars Released LOC Dolt [82] 19.1k 2018 380k DuckDB [15, 16, 48, 74, 82] 32.",
+      "context_after": "7k 1995 5,532k PostgreSQL [16, 24, 30, 47, 49, 58, 80, 82] 18.5k 1995 938k TiDB [3, 24, 57, 58, 60, 61, 74, 82] 39.0k 2016 1,398k Table 1. DBMSs under test by Argus. DBMS ReportedBug status Bug type Fixed Conf. Dup. Pend. Logic Other Dolt 19 18 1 0 0 18 1 DuckDB 8 6 0 1 1 4 4 MySQL 8 0 5 1 2 8 0 PostgreSQL 1 1 0 0 0 1 0 TiDB 5 2 3 0 0 5 0 Total41 27 9 2 3 36 5 Table 2. Argus found bugs statistics. (4) How effective isArgus’s SQL equivalence prover in filtering out false positives? (Sec. 7.4) (5)What are the time and monetary costs required forArgusto generate a given number of test cases? (Sec",
+      "section": "7 Evaluation",
+      "page": 14,
+      "char_offset": 45679,
+      "cited_reference": {
+        "number": 49,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Manageme",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "citation_marker",
+      "surface": "[16, 24, 30, 47, 49, 58, 80, 82]",
+      "technique": "norec",
+      "sentence": "7k 1995 5,532k PostgreSQL [16, 24, 30, 47, 49, 58, 80, 82] 18.",
+      "context_before": "gusfind in real-world DBMSs, and what LLM-generated test oracles discover them? (Sec. 7.1) (2) How doesArguscompare to other baseline tools in terms of code coverage? (Sec. 7.2) (3)How does the number of automatically discovered oracles affect the number of unique bugs found, compared to prior work where oracles are manually designed? (Sec. 7.3) 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:14 Qiuyang Mang et al. DBMS Tested by GitHub stars Released LOC Dolt [82] 19.1k 2018 380k DuckDB [15, 16, 48, 74, 82] 32.7k 2019 1,496k MySQL [24, 31, 49, 57, 58, 60, 61, 74, 80, 82] 11.",
+      "context_after": "5k 1995 938k TiDB [3, 24, 57, 58, 60, 61, 74, 82] 39.0k 2016 1,398k Table 1. DBMSs under test by Argus. DBMS ReportedBug status Bug type Fixed Conf. Dup. Pend. Logic Other Dolt 19 18 1 0 0 18 1 DuckDB 8 6 0 1 1 4 4 MySQL 8 0 5 1 2 8 0 PostgreSQL 1 1 0 0 0 1 0 TiDB 5 2 3 0 0 5 0 Total41 27 9 2 3 36 5 Table 2. Argus found bugs statistics. (4) How effective isArgus’s SQL equivalence prover in filtering out false positives? (Sec. 7.4) (5)What are the time and monetary costs required forArgusto generate a given number of test cases? (Sec. 7.5) Testbed.We conducted all experiments using a machine wi",
+      "section": "7 Evaluation",
+      "page": 14,
+      "char_offset": 45744,
+      "cited_reference": {
+        "number": 47,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "citation_marker",
+      "surface": "[3, 24, 57, 58, 60, 61, 74, 82]",
+      "technique": "qpg",
+      "sentence": "5k 1995 938k TiDB [3, 24, 57, 58, 60, 61, 74, 82] 39.",
+      "context_before": "es discover them? (Sec. 7.1) (2) How doesArguscompare to other baseline tools in terms of code coverage? (Sec. 7.2) (3)How does the number of automatically discovered oracles affect the number of unique bugs found, compared to prior work where oracles are manually designed? (Sec. 7.3) 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:14 Qiuyang Mang et al. DBMS Tested by GitHub stars Released LOC Dolt [82] 19.1k 2018 380k DuckDB [15, 16, 48, 74, 82] 32.7k 2019 1,496k MySQL [24, 31, 49, 57, 58, 60, 61, 74, 80, 82] 11.7k 1995 5,532k PostgreSQL [16, 24, 30, 47, 49, 58, 80, 82] 18.",
+      "context_after": "0k 2016 1,398k Table 1. DBMSs under test by Argus. DBMS ReportedBug status Bug type Fixed Conf. Dup. Pend. Logic Other Dolt 19 18 1 0 0 18 1 DuckDB 8 6 0 1 1 4 4 MySQL 8 0 5 1 2 8 0 PostgreSQL 1 1 0 0 0 1 0 TiDB 5 2 3 0 0 5 0 Total41 27 9 2 3 36 5 Table 2. Argus found bugs statistics. (4) How effective isArgus’s SQL equivalence prover in filtering out false positives? (Sec. 7.4) (5)What are the time and monetary costs required forArgusto generate a given number of test cases? (Sec. 7.5) Testbed.We conducted all experiments using a machine with 64 cores and 128 GB memory running on Ubuntu 24.04",
+      "section": "7 Evaluation",
+      "page": 14,
+      "char_offset": 45806,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M24",
+      "found_by": "citation_marker",
+      "surface": "[ 5]",
+      "technique": "dqp",
+      "sentence": ",reporting 21 [ 5], 24 [74], and 35 [ 24] logic bugs),Argusfinds more logic bugs in comparison, even though the DBMSs 4, No.",
+      "context_before": "ain focus of this paper, this still highlights the effectiveness ofArgusto generate complex, concrete queries for testing. Among the reports, 36 bugs have been confirmed by the developers; 27 have already been fixed, while 9 are still in progress. The remaining 2 bugs are duplicates: after developers reproduced and analyzed them, they found that these bugs shared the same root cause as some previously unfixed bugs. The 36 logic bugs we found underscore the effectiveness of our LLM-generated test oracles. Compared with recent studies that use manually designed test oracles for bug finding (e.g.",
+      "context_after": "3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:15 1CREATE TABLEt(cINT); 2INSERT INTOtVALUES(1); 3SELECTsub.cFROM( 4SELECT□ 1⊲Expr(t:INT)↦→ 5json_array_length(json_array(3, 2, t.c)) 6AScFROMt 7)ASsub 8RIGHT JOINtON FALSE; -- {2}ὁB 9SELECTsub.cFROM( 10SELECT NULL AScFROMt 11)ASsub 12RIGHT JOINtON FALSE; -- {NULL}/ Listing 6. Incorrectjsonfunctions handling in PostgreSQL when executing RIGHT JOIN. 1CREATE TABLEt(c0INT); 2INSERT INTOtVALUES(1); 3SELECT*FROMtLEFT JOIN( 4SELECT MOD(5, 2)ASc0FROMt 5)ASt2ON FALSE",
+      "section": "7.1 New Bugs and Oracles",
+      "page": 14,
+      "char_offset": 48299,
+      "cited_reference": {
+        "number": 5,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans.Proceedings of the ACM on Management of Data2, 3 (2024), 1–26.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp",
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "citation_marker",
+      "surface": "[ 58,71,81,82]",
+      "technique": null,
+      "sentence": "While several recent database testing studies have targeted PostgreSQL [ 58,71,81,82], none reported finding new bugs, which attests to the effectiveness of our approach.",
+      "context_before": "olumns from the left table must be NULL. Consequently, the first query should always yield a NULL row, which is identical to the second query’s output, regardless of the value of the placeholder □1. However, due to a bug in its handling of json functions, PostgreSQL erroneously returns 2. This discovery highlights both the effectiveness of our novel, automatically generated test oracles and the LLM’s ability to synthesize complex queries involving advanced SQL features, such as json functions. The finding is particularly noteworthy given that PostgreSQL is one of the world’s most robust DBMSs.",
+      "context_after": "After we reported the issue, developers confirmed and fixed it within 24 hours. We also find that prior test oracles, such as TLP [ 48], struggle to detect this bug. For example, when we append a predicate to the first query, such as 𝑃=sub.c> 2, and then apply the three variants WHERE𝑃 ,WHERE NOT𝑃, and WHERE𝑃IS NOT NULL, the results of the three partitioned queries remain identical. As a result, TLP does not report a bug in this case. Moreover, a similar oracle uncovered a logic bug in MySQL (Listing 7), further demonstrating the versatility of the oracles generated byArgus. Specifically, the",
+      "section": "7.1 New Bugs and Oracles",
+      "page": 15,
+      "char_offset": 50155,
+      "cited_reference": {
+        "number": 81,
+        "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments. arXiv preprint arXiv:2505.02012(2025).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "prints the arXiv id of the paper introducing shqvel"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We also find that prior test oracles, such as TLP [ 48], struggle to detect this bug.",
+      "context_before": "ons, PostgreSQL erroneously returns 2. This discovery highlights both the effectiveness of our novel, automatically generated test oracles and the LLM’s ability to synthesize complex queries involving advanced SQL features, such as json functions. The finding is particularly noteworthy given that PostgreSQL is one of the world’s most robust DBMSs. While several recent database testing studies have targeted PostgreSQL [ 58,71,81,82], none reported finding new bugs, which attests to the effectiveness of our approach. After we reported the issue, developers confirmed and fixed it within 24 hours.",
+      "context_after": "For example, when we append a predicate to the first query, such as 𝑃=sub.c> 2, and then apply the three variants WHERE𝑃 ,WHERE NOT𝑃, and WHERE𝑃IS NOT NULL, the results of the three partitioned queries remain identical. As a result, TLP does not report a bug in this case. Moreover, a similar oracle uncovered a logic bug in MySQL (Listing 7), further demonstrating the versatility of the oracles generated byArgus. Specifically, the query in Listing 7 uses a LEFT JOIN combined with a WHERE clause to filter out rows where the right table’s column is NULL, which should always yield an empty result",
+      "section": "7.1 New Bugs and Oracles",
+      "page": 15,
+      "char_offset": 50406,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "As a result, TLP does not report a bug in this case.",
+      "context_before": "QL is one of the world’s most robust DBMSs. While several recent database testing studies have targeted PostgreSQL [ 58,71,81,82], none reported finding new bugs, which attests to the effectiveness of our approach. After we reported the issue, developers confirmed and fixed it within 24 hours. We also find that prior test oracles, such as TLP [ 48], struggle to detect this bug. For example, when we append a predicate to the first query, such as 𝑃=sub.c> 2, and then apply the three variants WHERE𝑃 ,WHERE NOT𝑃, and WHERE𝑃IS NOT NULL, the results of the three partitioned queries remain identical.",
+      "context_after": "Moreover, a similar oracle uncovered a logic bug in MySQL (Listing 7), further demonstrating the versatility of the oracles generated byArgus. Specifically, the query in Listing 7 uses a LEFT JOIN combined with a WHERE clause to filter out rows where the right table’s column is NULL, which should always yield an empty result set. However, due to a bug in MySQL’s handling of the MOD function in this context, it incorrectly returns a row with value1. Example (2).Listing 8 shows a logic bug in Dolt, which was detected using an oracle generated by Argusthat leverages the semantics of the EXISTS p",
+      "section": "7.1 New Bugs and Oracles",
+      "page": 15,
+      "char_offset": 50712,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Argus SQLancer SQLancer++ EET 0 4 8 12 16 20 24 Hours3031323334353637Line Coverage (%) (a) DuckDB line 0 4 8 12 16 20 24 Hours18192021Branch Coverage (%) (b) DuckDB branch 0 4 8 12 16 20 24 Hours20222426283032Line Coverage (%) (c) PostgreSQL line 0 4 8 12 16 20 24 Hours141618202224Branch Coverage (%) (d) PostgreSQL branch Fig.",
+      "context_before": "trates that by leveraging the generative capabilities of LLMs for SQL snippets, Arguscan effectively uncover not only logic bugs but also critical crashes. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:17 1CREATE TABLEt0(c0INT); 2CREATE TABLEt1(c0BOOLEAN); 3SELECT*FROMt0 4WHERE EXISTS( 5SELECT1FROMt1 6WHERE□ 1⊲Expr(t0:INT)↦→ 7(WITH seq(i) AS (VALUES (1)) SELECT sum(i) * t0.c0 FROM seq)IS NOT NULL 8); -- (crash)ὁB Listing 10. A crash bug in DuckDB triggered by a LLM-synthesized expression withseq.",
+      "context_after": "3. Code coverage achieved by Argus, SQLancer, and SQLancer++ on DuckDB and PostgreSQL over 24-hour runs. 7.2 Code Coverage We comparedArguswith three DBMS logic bugs finding tools, SQLancer [ 3,47–49,74], SQLancer++ [ 82] and EET [ 24] in multiple coverage metrics. We compared on DuckDB [ 46] and PostgreSQL [ 38]. As a state-of-the-art, open-source DBMS testing framework, SQLancer supports most of the latest test oracles [ 4,57,74]. SQLancer++ is a scalable variant that can be easily extended to multiple DBMSs with only lightweight modifications. EET [ 24] is a recent tool based on SQLsmith [",
+      "section": "7.1 New Bugs and Oracles",
+      "page": 17,
+      "char_offset": 54600,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M29",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Code coverage achieved by Argus, SQLancer, and SQLancer++ on DuckDB and PostgreSQL over 24-hour runs.",
+      "context_before": "E TABLEt1(c0BOOLEAN); 3SELECT*FROMt0 4WHERE EXISTS( 5SELECT1FROMt1 6WHERE□ 1⊲Expr(t0:INT)↦→ 7(WITH seq(i) AS (VALUES (1)) SELECT sum(i) * t0.c0 FROM seq)IS NOT NULL 8); -- (crash)ὁB Listing 10. A crash bug in DuckDB triggered by a LLM-synthesized expression withseq. Argus SQLancer SQLancer++ EET 0 4 8 12 16 20 24 Hours3031323334353637Line Coverage (%) (a) DuckDB line 0 4 8 12 16 20 24 Hours18192021Branch Coverage (%) (b) DuckDB branch 0 4 8 12 16 20 24 Hours20222426283032Line Coverage (%) (c) PostgreSQL line 0 4 8 12 16 20 24 Hours141618202224Branch Coverage (%) (d) PostgreSQL branch Fig. 3.",
+      "context_after": "7.2 Code Coverage We comparedArguswith three DBMS logic bugs finding tools, SQLancer [ 3,47–49,74], SQLancer++ [ 82] and EET [ 24] in multiple coverage metrics. We compared on DuckDB [ 46] and PostgreSQL [ 38]. As a state-of-the-art, open-source DBMS testing framework, SQLancer supports most of the latest test oracles [ 4,57,74]. SQLancer++ is a scalable variant that can be easily extended to multiple DBMSs with only lightweight modifications. EET [ 24] is a recent tool based on SQLsmith [ 53]’s generator and their carefully designed test oracles to produce complex queries. In terms of test o",
+      "section": "7.1 New Bugs and Oracles",
+      "page": 17,
+      "char_offset": 54934,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M30",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "2 Code Coverage We comparedArguswith three DBMS logic bugs finding tools, SQLancer [ 3,47–49,74], SQLancer++ [ 82] and EET [ 24] in multiple coverage metrics.",
+      "context_before": "AS (VALUES (1)) SELECT sum(i) * t0.c0 FROM seq)IS NOT NULL 8); -- (crash)ὁB Listing 10. A crash bug in DuckDB triggered by a LLM-synthesized expression withseq. Argus SQLancer SQLancer++ EET 0 4 8 12 16 20 24 Hours3031323334353637Line Coverage (%) (a) DuckDB line 0 4 8 12 16 20 24 Hours18192021Branch Coverage (%) (b) DuckDB branch 0 4 8 12 16 20 24 Hours20222426283032Line Coverage (%) (c) PostgreSQL line 0 4 8 12 16 20 24 Hours141618202224Branch Coverage (%) (d) PostgreSQL branch Fig. 3. Code coverage achieved by Argus, SQLancer, and SQLancer++ on DuckDB and PostgreSQL over 24-hour runs. 7.",
+      "context_after": "We compared on DuckDB [ 46] and PostgreSQL [ 38]. As a state-of-the-art, open-source DBMS testing framework, SQLancer supports most of the latest test oracles [ 4,57,74]. SQLancer++ is a scalable variant that can be easily extended to multiple DBMSs with only lightweight modifications. EET [ 24] is a recent tool based on SQLsmith [ 53]’s generator and their carefully designed test oracles to produce complex queries. In terms of test oracles,code coverageis an indicator for the number of features and execution plans that are tested by them. Although code coverage does not strongly correlate wi",
+      "section": "7.2 Code Coverage",
+      "page": 17,
+      "char_offset": 55039,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As a state-of-the-art, open-source DBMS testing framework, SQLancer supports most of the latest test oracles [ 4,57,74].",
+      "context_before": "4 Hours3031323334353637Line Coverage (%) (a) DuckDB line 0 4 8 12 16 20 24 Hours18192021Branch Coverage (%) (b) DuckDB branch 0 4 8 12 16 20 24 Hours20222426283032Line Coverage (%) (c) PostgreSQL line 0 4 8 12 16 20 24 Hours141618202224Branch Coverage (%) (d) PostgreSQL branch Fig. 3. Code coverage achieved by Argus, SQLancer, and SQLancer++ on DuckDB and PostgreSQL over 24-hour runs. 7.2 Code Coverage We comparedArguswith three DBMS logic bugs finding tools, SQLancer [ 3,47–49,74], SQLancer++ [ 82] and EET [ 24] in multiple coverage metrics. We compared on DuckDB [ 46] and PostgreSQL [ 38].",
+      "context_after": "SQLancer++ is a scalable variant that can be easily extended to multiple DBMSs with only lightweight modifications. EET [ 24] is a recent tool based on SQLsmith [ 53]’s generator and their carefully designed test oracles to produce complex queries. In terms of test oracles,code coverageis an indicator for the number of features and execution plans that are tested by them. Although code coverage does not strongly correlate with the ability 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:18 Qiuyang Mang et al. Approach Lines Functions Branches SQLancer 3.256% 1.230% 1.313% Arg",
+      "section": "7.2 Code Coverage",
+      "page": 17,
+      "char_offset": 55247,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "name",
+      "surface": "SQLancer++",
+      "technique": null,
+      "sentence": "SQLancer++ is a scalable variant that can be easily extended to multiple DBMSs with only lightweight modifications.",
+      "context_before": "ranch 0 4 8 12 16 20 24 Hours20222426283032Line Coverage (%) (c) PostgreSQL line 0 4 8 12 16 20 24 Hours141618202224Branch Coverage (%) (d) PostgreSQL branch Fig. 3. Code coverage achieved by Argus, SQLancer, and SQLancer++ on DuckDB and PostgreSQL over 24-hour runs. 7.2 Code Coverage We comparedArguswith three DBMS logic bugs finding tools, SQLancer [ 3,47–49,74], SQLancer++ [ 82] and EET [ 24] in multiple coverage metrics. We compared on DuckDB [ 46] and PostgreSQL [ 38]. As a state-of-the-art, open-source DBMS testing framework, SQLancer supports most of the latest test oracles [ 4,57,74].",
+      "context_after": "EET [ 24] is a recent tool based on SQLsmith [ 53]’s generator and their carefully designed test oracles to produce complex queries. In terms of test oracles,code coverageis an indicator for the number of features and execution plans that are tested by them. Although code coverage does not strongly correlate with the ability 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:18 Qiuyang Mang et al. Approach Lines Functions Branches SQLancer 3.256% 1.230% 1.313% Argus17.820% 7.910% 7.315% 5.473×6.431×5.571× Table 3. Average metamorphic coverage on DuckDB of 10 test suites for Arg",
+      "section": "7.2 Code Coverage",
+      "page": 17,
+      "char_offset": 55368,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M33",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Approach Lines Functions Branches SQLancer 3.",
+      "context_before": "ework, SQLancer supports most of the latest test oracles [ 4,57,74]. SQLancer++ is a scalable variant that can be easily extended to multiple DBMSs with only lightweight modifications. EET [ 24] is a recent tool based on SQLsmith [ 53]’s generator and their carefully designed test oracles to produce complex queries. In terms of test oracles,code coverageis an indicator for the number of features and execution plans that are tested by them. Although code coverage does not strongly correlate with the ability 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:18 Qiuyang Mang et al.",
+      "context_after": "256% 1.230% 1.313% Argus17.820% 7.910% 7.315% 5.473×6.431×5.571× Table 3. Average metamorphic coverage on DuckDB of 10 test suites for Argus and SQLancer. to find logic bugs [ 82], it is still a fair metric to evaluate the diversity of test cases generated by Argusand the effectiveness of its corpus synthesis. Meanwhile,metamorphic coverage[ 2] is a more relevant metric for evaluating the effectiveness of test oracles and is highly related to logic bug finding abilities according to the historical bug study in SQLite and DuckDB [ 2]. Unlike traditional code coverage, which simply counts all ex",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 55900,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M34",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Average metamorphic coverage on DuckDB of 10 test suites for Argus and SQLancer.",
+      "context_before": "y extended to multiple DBMSs with only lightweight modifications. EET [ 24] is a recent tool based on SQLsmith [ 53]’s generator and their carefully designed test oracles to produce complex queries. In terms of test oracles,code coverageis an indicator for the number of features and execution plans that are tested by them. Although code coverage does not strongly correlate with the ability 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:18 Qiuyang Mang et al. Approach Lines Functions Branches SQLancer 3.256% 1.230% 1.313% Argus17.820% 7.910% 7.315% 5.473×6.431×5.571× Table 3.",
+      "context_after": "to find logic bugs [ 82], it is still a fair metric to evaluate the diversity of test cases generated by Argusand the effectiveness of its corpus synthesis. Meanwhile,metamorphic coverage[ 2] is a more relevant metric for evaluating the effectiveness of test oracles and is highly related to logic bug finding abilities according to the historical bug study in SQLite and DuckDB [ 2]. Unlike traditional code coverage, which simply counts all executed code, metamorphic coverage measures the portions of the program exercised differently by the two executions in a pair of equivalent queries. Intuit",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 56019,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M35",
+      "found_by": "citation_marker",
+      "surface": "[ 82]",
+      "technique": null,
+      "sentence": "to find logic bugs [ 82], it is still a fair metric to evaluate the diversity of test cases generated by Argusand the effectiveness of its corpus synthesis.",
+      "context_before": "recent tool based on SQLsmith [ 53]’s generator and their carefully designed test oracles to produce complex queries. In terms of test oracles,code coverageis an indicator for the number of features and execution plans that are tested by them. Although code coverage does not strongly correlate with the ability 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:18 Qiuyang Mang et al. Approach Lines Functions Branches SQLancer 3.256% 1.230% 1.313% Argus17.820% 7.910% 7.315% 5.473×6.431×5.571× Table 3. Average metamorphic coverage on DuckDB of 10 test suites for Argus and SQLancer.",
+      "context_after": "Meanwhile,metamorphic coverage[ 2] is a more relevant metric for evaluating the effectiveness of test oracles and is highly related to logic bug finding abilities according to the historical bug study in SQLite and DuckDB [ 2]. Unlike traditional code coverage, which simply counts all executed code, metamorphic coverage measures the portions of the program exercised differently by the two executions in a pair of equivalent queries. Intuitively, because only these differential execution paths can lead to inconsistent results between semantically equivalent queries, metamorphic coverage provide",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 56100,
+      "cited_reference": {
+        "number": 82,
+        "text": "Suyang Zhong and Manuel Rigger. 2026. Scaling Automated Database System Testing. InProceedings of the 31st ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2(USA) (ASPLOS ’26). Association for Computing Machinery, New York, NY, USA, 1677–1",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing sqlancer_pp",
+          "prints the DOI of the paper introducing sqlancer_pp"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M36",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 2]",
+      "technique": null,
+      "sentence": "Meanwhile,metamorphic coverage[ 2] is a more relevant metric for evaluating the effectiveness of test oracles and is highly related to logic bug finding abilities according to the historical bug study in SQLite and DuckDB [ 2].",
+      "context_before": "s an indicator for the number of features and execution plans that are tested by them. Although code coverage does not strongly correlate with the ability 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:18 Qiuyang Mang et al. Approach Lines Functions Branches SQLancer 3.256% 1.230% 1.313% Argus17.820% 7.910% 7.315% 5.473×6.431×5.571× Table 3. Average metamorphic coverage on DuckDB of 10 test suites for Argus and SQLancer. to find logic bugs [ 82], it is still a fair metric to evaluate the diversity of test cases generated by Argusand the effectiveness of its corpus synthesis.",
+      "context_after": "Unlike traditional code coverage, which simply counts all executed code, metamorphic coverage measures the portions of the program exercised differently by the two executions in a pair of equivalent queries. Intuitively, because only these differential execution paths can lead to inconsistent results between semantically equivalent queries, metamorphic coverage provides a more precise measure of the code actually validated by the test oracle. Empirical studies on DBMSs like SQLite and DuckDB show that metamorphic coverage correlates strongly with real logic bugs and overlaps substantially wit",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 56257,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba, Yuancheng Jiang, and Manuel Rigger. 2025. Metamorphic Coverage.arXiv preprint arXiv:2508.16307 (2025).",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M37",
+      "found_by": "name",
+      "surface": "SQLancer++",
+      "technique": null,
+      "sentence": "1% higher line and branch coverage than SQLancer++, respectively, and 11.",
+      "context_before": "these differential execution paths can lead to inconsistent results between semantically equivalent queries, metamorphic coverage provides a more precise measure of the code actually validated by the test oracle. Empirical studies on DBMSs like SQLite and DuckDB show that metamorphic coverage correlates strongly with real logic bugs and overlaps substantially with historical bug-fix locations. Code coverage.Fig. 3 shows the line and branch code coverage for DuckDB and PostgreSQL over a 24-hour testing period starting from a clean build and empty database. In DuckDB,Argus achieves 19.9% and 18.",
+      "context_after": "3% and 3.4% higher than SQLancer. Note that, we did not compare with EET on DuckDB, as it does not support this DBMS. In PostgreSQL,Argusslightly underperforms SQLancer overall but outperforms SQLancer++ (by 19.0% in line coverage and 22.5% in branch coverage) and EET (by 5.2% and 2.6%, respectively). This suboptimal performance is expected, as SQLancer has been extensively optimized specifically for PostgreSQL by the open-source community over many years. Specifically, SQLancer supports 22 types of DDL statements (e.g., CREATE SEQUENCE ) and various DML statements beyond SELECT queries that P",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 57320,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M38",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "4% higher than SQLancer.",
+      "context_before": "ically equivalent queries, metamorphic coverage provides a more precise measure of the code actually validated by the test oracle. Empirical studies on DBMSs like SQLite and DuckDB show that metamorphic coverage correlates strongly with real logic bugs and overlaps substantially with historical bug-fix locations. Code coverage.Fig. 3 shows the line and branch code coverage for DuckDB and PostgreSQL over a 24-hour testing period starting from a clean build and empty database. In DuckDB,Argus achieves 19.9% and 18.1% higher line and branch coverage than SQLancer++, respectively, and 11.3% and 3.",
+      "context_after": "Note that, we did not compare with EET on DuckDB, as it does not support this DBMS. In PostgreSQL,Argusslightly underperforms SQLancer overall but outperforms SQLancer++ (by 19.0% in line coverage and 22.5% in branch coverage) and EET (by 5.2% and 2.6%, respectively). This suboptimal performance is expected, as SQLancer has been extensively optimized specifically for PostgreSQL by the open-source community over many years. Specifically, SQLancer supports 22 types of DDL statements (e.g., CREATE SEQUENCE ) and various DML statements beyond SELECT queries that PostgreSQL supports, whereasArgusf",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 57402,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M39",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In PostgreSQL,Argusslightly underperforms SQLancer overall but outperforms SQLancer++ (by 19.",
+      "context_before": "ed by the test oracle. Empirical studies on DBMSs like SQLite and DuckDB show that metamorphic coverage correlates strongly with real logic bugs and overlaps substantially with historical bug-fix locations. Code coverage.Fig. 3 shows the line and branch code coverage for DuckDB and PostgreSQL over a 24-hour testing period starting from a clean build and empty database. In DuckDB,Argus achieves 19.9% and 18.1% higher line and branch coverage than SQLancer++, respectively, and 11.3% and 3.4% higher than SQLancer. Note that, we did not compare with EET on DuckDB, as it does not support this DBMS.",
+      "context_after": "0% in line coverage and 22.5% in branch coverage) and EET (by 5.2% and 2.6%, respectively). This suboptimal performance is expected, as SQLancer has been extensively optimized specifically for PostgreSQL by the open-source community over many years. Specifically, SQLancer supports 22 types of DDL statements (e.g., CREATE SEQUENCE ) and various DML statements beyond SELECT queries that PostgreSQL supports, whereasArgusfocuses on detecting logic bugs across different DBMS implementations. For a finer-grained comparison, we additionally evaluatedArgusand SQLancer on PostgreSQL with respect to opt",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 57510,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M40",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This suboptimal performance is expected, as SQLancer has been extensively optimized specifically for PostgreSQL by the open-source community over many years.",
+      "context_before": "al bug-fix locations. Code coverage.Fig. 3 shows the line and branch code coverage for DuckDB and PostgreSQL over a 24-hour testing period starting from a clean build and empty database. In DuckDB,Argus achieves 19.9% and 18.1% higher line and branch coverage than SQLancer++, respectively, and 11.3% and 3.4% higher than SQLancer. Note that, we did not compare with EET on DuckDB, as it does not support this DBMS. In PostgreSQL,Argusslightly underperforms SQLancer overall but outperforms SQLancer++ (by 19.0% in line coverage and 22.5% in branch coverage) and EET (by 5.2% and 2.6%, respectively).",
+      "context_after": "Specifically, SQLancer supports 22 types of DDL statements (e.g., CREATE SEQUENCE ) and various DML statements beyond SELECT queries that PostgreSQL supports, whereasArgusfocuses on detecting logic bugs across different DBMS implementations. For a finer-grained comparison, we additionally evaluatedArgusand SQLancer on PostgreSQL with respect to optimizer code coverage, since the optimizer is a core component that is closely tied to SELECT queries and widely studied in prior work [ 47,60], along with the diversity of features exercised by SELECT queries that are covered by the test queries. The",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 57695,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M41",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, SQLancer supports 22 types of DDL statements (e.",
+      "context_before": "an build and empty database. In DuckDB,Argus achieves 19.9% and 18.1% higher line and branch coverage than SQLancer++, respectively, and 11.3% and 3.4% higher than SQLancer. Note that, we did not compare with EET on DuckDB, as it does not support this DBMS. In PostgreSQL,Argusslightly underperforms SQLancer overall but outperforms SQLancer++ (by 19.0% in line coverage and 22.5% in branch coverage) and EET (by 5.2% and 2.6%, respectively). This suboptimal performance is expected, as SQLancer has been extensively optimized specifically for PostgreSQL by the open-source community over many years.",
+      "context_after": "g., CREATE SEQUENCE ) and various DML statements beyond SELECT queries that PostgreSQL supports, whereasArgusfocuses on detecting logic bugs across different DBMS implementations. For a finer-grained comparison, we additionally evaluatedArgusand SQLancer on PostgreSQL with respect to optimizer code coverage, since the optimizer is a core component that is closely tied to SELECT queries and widely studied in prior work [ 47,60], along with the diversity of features exercised by SELECT queries that are covered by the test queries. The results show thatArgus achieved 66.46% line coverage and 55.85",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 57853,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M42",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For a finer-grained comparison, we additionally evaluatedArgusand SQLancer on PostgreSQL with respect to optimizer code coverage, since the optimizer is a core component that is closely tied to SELECT queries and widely studied in prior work [ 47,60], along with the diversity of features exercised by SELECT queries that are covered by the test queries.",
+      "context_before": "port this DBMS. In PostgreSQL,Argusslightly underperforms SQLancer overall but outperforms SQLancer++ (by 19.0% in line coverage and 22.5% in branch coverage) and EET (by 5.2% and 2.6%, respectively). This suboptimal performance is expected, as SQLancer has been extensively optimized specifically for PostgreSQL by the open-source community over many years. Specifically, SQLancer supports 22 types of DDL statements (e.g., CREATE SEQUENCE ) and various DML statements beyond SELECT queries that PostgreSQL supports, whereasArgusfocuses on detecting logic bugs across different DBMS implementations.",
+      "context_after": "The results show thatArgus achieved 66.46% line coverage and 55.85% branch coverage of PostgreSQL’s optimizer, compared to 62.33% and 52.17% by SQLancer. To evaluate feature diversity, we randomly sampled ten SELECT queries from each tool and counted their unique features using two third-party parsers:pglast[ 17] (counting number of unique PostgreSQL types) andsqlparse[ 87] (counting number of unique AST nodes). We found thatArguscovered 23 features in pglast and 151 features in sqlparse, while SQLancer only covered 15 and 76, respectively. This demonstratesArgus’s stronger ability to generat",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 58095,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M43",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "17% by SQLancer.",
+      "context_before": "that PostgreSQL supports, whereasArgusfocuses on detecting logic bugs across different DBMS implementations. For a finer-grained comparison, we additionally evaluatedArgusand SQLancer on PostgreSQL with respect to optimizer code coverage, since the optimizer is a core component that is closely tied to SELECT queries and widely studied in prior work [ 47,60], along with the diversity of features exercised by SELECT queries that are covered by the test queries. The results show thatArgus achieved 66.46% line coverage and 55.85% branch coverage of PostgreSQL’s optimizer, compared to 62.33% and 52.",
+      "context_after": "To evaluate feature diversity, we randomly sampled ten SELECT queries from each tool and counted their unique features using two third-party parsers:pglast[ 17] (counting number of unique PostgreSQL types) andsqlparse[ 87] (counting number of unique AST nodes). We found thatArguscovered 23 features in pglast and 151 features in sqlparse, while SQLancer only covered 15 and 76, respectively. This demonstratesArgus’s stronger ability to generate feature-rich queries. The sampled queries are provided in Appendix D. Note that we did not compare with mutation-based testing tools, such as SQLRight [",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 58587,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M44",
+      "found_by": "citation_marker",
+      "surface": "[ 87]",
+      "technique": null,
+      "sentence": "To evaluate feature diversity, we randomly sampled ten SELECT queries from each tool and counted their unique features using two third-party parsers:pglast[ 17] (counting number of unique PostgreSQL types) andsqlparse[ 87] (counting number of unique AST nodes).",
+      "context_before": "supports, whereasArgusfocuses on detecting logic bugs across different DBMS implementations. For a finer-grained comparison, we additionally evaluatedArgusand SQLancer on PostgreSQL with respect to optimizer code coverage, since the optimizer is a core component that is closely tied to SELECT queries and widely studied in prior work [ 47,60], along with the diversity of features exercised by SELECT queries that are covered by the test queries. The results show thatArgus achieved 66.46% line coverage and 55.85% branch coverage of PostgreSQL’s optimizer, compared to 62.33% and 52.17% by SQLancer.",
+      "context_after": "We found thatArguscovered 23 features in pglast and 151 features in sqlparse, while SQLancer only covered 15 and 76, respectively. This demonstratesArgus’s stronger ability to generate feature-rich queries. The sampled queries are provided in Appendix D. Note that we did not compare with mutation-based testing tools, such as SQLRight [ 31] and SQuirrel [ 80], as they highly rely on the quality of seed queries and use the official unit tests provided by the DBMS as the initial input. This introduces bias in the comparison, as the official unit tests are usually well-designed and purposefully c",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 58603,
+      "cited_reference": {
+        "number": 87,
+        "text": "Ándré Albrecht and Contributors. 2025. sqlparse: A non-validating SQL parser module for Python. https://github.com/ andialbrecht/sqlparse. Accessed: 2025-10-15. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M45",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We found thatArguscovered 23 features in pglast and 151 features in sqlparse, while SQLancer only covered 15 and 76, respectively.",
+      "context_before": "that is closely tied to SELECT queries and widely studied in prior work [ 47,60], along with the diversity of features exercised by SELECT queries that are covered by the test queries. The results show thatArgus achieved 66.46% line coverage and 55.85% branch coverage of PostgreSQL’s optimizer, compared to 62.33% and 52.17% by SQLancer. To evaluate feature diversity, we randomly sampled ten SELECT queries from each tool and counted their unique features using two third-party parsers:pglast[ 17] (counting number of unique PostgreSQL types) andsqlparse[ 87] (counting number of unique AST nodes).",
+      "context_after": "This demonstratesArgus’s stronger ability to generate feature-rich queries. The sampled queries are provided in Appendix D. Note that we did not compare with mutation-based testing tools, such as SQLRight [ 31] and SQuirrel [ 80], as they highly rely on the quality of seed queries and use the official unit tests provided by the DBMS as the initial input. This introduces bias in the comparison, as the official unit tests are usually well-designed and purposefully cover many code paths for the DBMSs that they are designed for. In fact, SQLRight achieved51 .3%line coverage on PostgreSQL in the f",
+      "section": "7.2 Code Coverage",
+      "page": 18,
+      "char_offset": 58865,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M46",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For metamorphic coverage, we comparedArguswith SQLancer on DuckDB across a fixed number of test cases, following the same setting in [ 2].",
+      "context_before": "er that without further improvement during the remainder of the testing period. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:19 1 2 3 4 5 6 Hours246810Unique Bug Numbers Argus-5,000 Argus-50 Baseline Fig. 4. Number of unique logic bugs found by different sets of oracles within a 6-hour testing on Dolt. Overall, our results demonstrate thatArguscan generate feature-rich, diverse, and complex test cases at scale, covering more optimizer paths than state-of-the-art techniques. Metamorphic coverage.",
+      "context_after": "We did not evaluate metamorphic coverage on PostgreSQL, which has not been supported by [ 2]’s implementation. Following the experimental setup in [ 2], we generated 10 test suites for SQLancer, each with 100 test cases. Specifically, SQLancer produce 50% of the test cases from TLP [ 48] and 50% from NoREC [ 47]. ForArgus, we generated 10 CAQ pairs as test suites, and instantiated each pair with 100 different snippets, respectively. We reported the average metamorphic coverage of the 10 test suites in Table 3. As shown,Argus achieves 5.473×, 6.431×, and 5.571×more line, function, and branch c",
+      "section": "7.2 Code Coverage",
+      "page": 19,
+      "char_offset": 60307,
+      "found_by_all": [
+        "name",
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M47",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 2]",
+      "technique": null,
+      "sentence": "We did not evaluate metamorphic coverage on PostgreSQL, which has not been supported by [ 2]’s implementation.",
+      "context_before": "26. Automated Discovery of Test Oracles for Database Management Systems Using LLMs 140:19 1 2 3 4 5 6 Hours246810Unique Bug Numbers Argus-5,000 Argus-50 Baseline Fig. 4. Number of unique logic bugs found by different sets of oracles within a 6-hour testing on Dolt. Overall, our results demonstrate thatArguscan generate feature-rich, diverse, and complex test cases at scale, covering more optimizer paths than state-of-the-art techniques. Metamorphic coverage.For metamorphic coverage, we comparedArguswith SQLancer on DuckDB across a fixed number of test cases, following the same setting in [ 2].",
+      "context_after": "Following the experimental setup in [ 2], we generated 10 test suites for SQLancer, each with 100 test cases. Specifically, SQLancer produce 50% of the test cases from TLP [ 48] and 50% from NoREC [ 47]. ForArgus, we generated 10 CAQ pairs as test suites, and instantiated each pair with 100 different snippets, respectively. We reported the average metamorphic coverage of the 10 test suites in Table 3. As shown,Argus achieves 5.473×, 6.431×, and 5.571×more line, function, and branch coverage than SQLancer, respectively. We believe this is becauseArgus’s new test oracles can significantly alter",
+      "section": "7.2 Code Coverage",
+      "page": 19,
+      "char_offset": 60445,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba, Yuancheng Jiang, and Manuel Rigger. 2025. Metamorphic Coverage.arXiv preprint arXiv:2508.16307 (2025).",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M48",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Following the experimental setup in [ 2], we generated 10 test suites for SQLancer, each with 100 test cases.",
+      "context_before": "10Unique Bug Numbers Argus-5,000 Argus-50 Baseline Fig. 4. Number of unique logic bugs found by different sets of oracles within a 6-hour testing on Dolt. Overall, our results demonstrate thatArguscan generate feature-rich, diverse, and complex test cases at scale, covering more optimizer paths than state-of-the-art techniques. Metamorphic coverage.For metamorphic coverage, we comparedArguswith SQLancer on DuckDB across a fixed number of test cases, following the same setting in [ 2]. We did not evaluate metamorphic coverage on PostgreSQL, which has not been supported by [ 2]’s implementation.",
+      "context_after": "Specifically, SQLancer produce 50% of the test cases from TLP [ 48] and 50% from NoREC [ 47]. ForArgus, we generated 10 CAQ pairs as test suites, and instantiated each pair with 100 different snippets, respectively. We reported the average metamorphic coverage of the 10 test suites in Table 3. As shown,Argus achieves 5.473×, 6.431×, and 5.571×more line, function, and branch coverage than SQLancer, respectively. We believe this is becauseArgus’s new test oracles can significantly alter the query structures between equivalent queries, thanks to LLM’s creativity, which covers a broader range of",
+      "section": "7.2 Code Coverage",
+      "page": 19,
+      "char_offset": 60556,
+      "found_by_all": [
+        "name",
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M49",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, SQLancer produce 50% of the test cases from TLP [ 48] and 50% from NoREC [ 47].",
+      "context_before": "of oracles within a 6-hour testing on Dolt. Overall, our results demonstrate thatArguscan generate feature-rich, diverse, and complex test cases at scale, covering more optimizer paths than state-of-the-art techniques. Metamorphic coverage.For metamorphic coverage, we comparedArguswith SQLancer on DuckDB across a fixed number of test cases, following the same setting in [ 2]. We did not evaluate metamorphic coverage on PostgreSQL, which has not been supported by [ 2]’s implementation. Following the experimental setup in [ 2], we generated 10 test suites for SQLancer, each with 100 test cases.",
+      "context_after": "ForArgus, we generated 10 CAQ pairs as test suites, and instantiated each pair with 100 different snippets, respectively. We reported the average metamorphic coverage of the 10 test suites in Table 3. As shown,Argus achieves 5.473×, 6.431×, and 5.571×more line, function, and branch coverage than SQLancer, respectively. We believe this is becauseArgus’s new test oracles can significantly alter the query structures between equivalent queries, thanks to LLM’s creativity, which covers a broader range of different code paths than prior works. This significant improvement demonstrates the effective",
+      "section": "7.2 Code Coverage",
+      "page": 19,
+      "char_offset": 60666,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M50",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "571×more line, function, and branch coverage than SQLancer, respectively.",
+      "context_before": ", following the same setting in [ 2]. We did not evaluate metamorphic coverage on PostgreSQL, which has not been supported by [ 2]’s implementation. Following the experimental setup in [ 2], we generated 10 test suites for SQLancer, each with 100 test cases. Specifically, SQLancer produce 50% of the test cases from TLP [ 48] and 50% from NoREC [ 47]. ForArgus, we generated 10 CAQ pairs as test suites, and instantiated each pair with 100 different snippets, respectively. We reported the average metamorphic coverage of the 10 test suites in Table 3. As shown,Argus achieves 5.473×, 6.431×, and 5.",
+      "context_after": "We believe this is becauseArgus’s new test oracles can significantly alter the query structures between equivalent queries, thanks to LLM’s creativity, which covers a broader range of different code paths than prior works. This significant improvement demonstrates the effectiveness ofArgus’s new test oracles in comparing different execution paths, and the potential to detect more logic bugs that state-of-the-art testers missed. 7.3 Effect of Test Oracles We now evaluateArgus’s new oracles by comparing the number of unique logic bugs they detected in Dolt v1.0.0. We selected this historical ve",
+      "section": "7.2 Code Coverage",
+      "page": 19,
+      "char_offset": 61008,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M51",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The baseline combines 11 test oracles from four previous works: TLP [ 48], NoREC [ 47], EET [ 24], and DQP [ 5].",
+      "context_before": ", and the potential to detect more logic bugs that state-of-the-art testers missed. 7.3 Effect of Test Oracles We now evaluateArgus’s new oracles by comparing the number of unique logic bugs they detected in Dolt v1.0.0. We selected this historical version to ensure a fair comparison, as git-bisect allows us to precisely link each bug to its corresponding fix commit, providing an accurate count of unique bugs per method. We compared a composite baseline of test oracles against two sets of LLM-generated test oracles byArgus, with sizes of 50 and 5,000,i.e.,Argus-50 andArgus-5,000, respectively.",
+      "context_after": "To minimize confounding factors, we standardized the experimental conditions. Specifically, the baseline oracles were represented in the same CAQ pair format and instantiated using the same snippet corpus asArgus. Furthermore, since the CAQ format requires an associated schema, we also randomly generated a new schema for the baseline oracles before each instantiation. The details about how to convert the baseline oracles to CAQ pairs are provided in Appendix C. Fig. 4 shows the number of unique logic bugs found by different sets of oracles within a 6-hour testing. We select a 6-hour time wind",
+      "section": "7.3 Effect of Test Oracles",
+      "page": 19,
+      "char_offset": 62030,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M52",
+      "found_by": "citation_marker",
+      "surface": "[ 81]",
+      "technique": null,
+      "sentence": "We select a 6-hour time window for a fair comparison, following common practice in fuzz testing [ 26] and prior studies on DBMS logic bug detection [ 81], which typically fix the testing duration within a range of 1 – 24 hours.",
+      "context_before": "47], EET [ 24], and DQP [ 5]. To minimize confounding factors, we standardized the experimental conditions. Specifically, the baseline oracles were represented in the same CAQ pair format and instantiated using the same snippet corpus asArgus. Furthermore, since the CAQ format requires an associated schema, we also randomly generated a new schema for the baseline oracles before each instantiation. The details about how to convert the baseline oracles to CAQ pairs are provided in Appendix C. Fig. 4 shows the number of unique logic bugs found by different sets of oracles within a 6-hour testing.",
+      "context_after": "As shown,Argus-5,000 found 10 unique bugs, significantly outperforming the baseline, which found only 3. In contrast,Argus-50 found just 2 bugs. This 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:20 Qiuyang Mang et al. 1SELECT*FROMvWHERE TRUE; 2SELECT*FROMvWHEREv.c >= v.cORv.c < v.c Listing 11. A false positive example from LLM-as-a-judge. underperformance was expected, as the baseline oracles were carefully designed by human experts. These results demonstrate the importance of the number of oracles in detecting unique logic bugs and underscore the limitations of manual or",
+      "section": "7.3 Effect of Test Oracles",
+      "page": 19,
+      "char_offset": 62714,
+      "cited_reference": {
+        "number": 81,
+        "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments. arXiv preprint arXiv:2505.02012(2025).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "prints the arXiv id of the paper introducing shqvel"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M53",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Argus SQLancer SQLancer++ Argus-grammar-based 0 4 8 12 16 20 24 Hours283032343638Line Coverage (%) (a) DuckDB line 0 4 8 12 16 20 24 Hours18192021Branch Coverage (%) (b) DuckDB branch Fig.",
+      "context_before": "arity of the generated queries, measured by the average tree-edit distance between DuckDB query plans across all pairs of equivalent queries. Specifically, given two query plan tree 𝑇1and𝑇2, letdist(𝑇 1,𝑇2)be the tree-edit distance between them. The similarity score is computed as:2 𝑁(𝑁−1)Í 𝑖≠𝑗 1−dist(𝑇 𝑖,𝑇𝑗) |𝑇𝑖|+|𝑇 𝑗|, where𝑁is the number of valid query pairs. As shown in Table 4,Argus’s CAQ generation method outperforms both baselines by producing more valid CAQs while maintaining the lowest similarity. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:22 Qiuyang Mang et al.",
+      "context_after": "6. Ablation study of code coverage achieved by Argus and a variant using only SQLancer’s grammarbased generator on DuckDB over 24-hour runs. 1CREATE TABLEt1(c0BOOLEAN, c1BOOLEAN); 2SELECT false FROMt1; 3SELECT(t1.c1 < t1.c0) < (t1.c1 = t1.c1)FROMt1; Listing 12. Incorrect equivalence proof in SQLSolver [13]. LLM-powered SQL snippet generation.We evaluate the effectiveness of LLM-powered SQL snippet generation by comparingArgus’s method against a variant that only uses SQLancer’s grammarbased generator to build the snippet corpus during the CAQ instantiation phase. We use the code coverage metr",
+      "section": "7.6 Component-wise Analysis",
+      "page": 22,
+      "char_offset": 69803,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M54",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Ablation study of code coverage achieved by Argus and a variant using only SQLancer’s grammarbased generator on DuckDB over 24-hour runs.",
+      "context_before": "tdist(𝑇 1,𝑇2)be the tree-edit distance between them. The similarity score is computed as:2 𝑁(𝑁−1)Í 𝑖≠𝑗 1−dist(𝑇 𝑖,𝑇𝑗) |𝑇𝑖|+|𝑇 𝑗|, where𝑁is the number of valid query pairs. As shown in Table 4,Argus’s CAQ generation method outperforms both baselines by producing more valid CAQs while maintaining the lowest similarity. 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:22 Qiuyang Mang et al. Argus SQLancer SQLancer++ Argus-grammar-based 0 4 8 12 16 20 24 Hours283032343638Line Coverage (%) (a) DuckDB line 0 4 8 12 16 20 24 Hours18192021Branch Coverage (%) (b) DuckDB branch Fig. 6.",
+      "context_after": "1CREATE TABLEt1(c0BOOLEAN, c1BOOLEAN); 2SELECT false FROMt1; 3SELECT(t1.c1 < t1.c0) < (t1.c1 = t1.c1)FROMt1; Listing 12. Incorrect equivalence proof in SQLSolver [13]. LLM-powered SQL snippet generation.We evaluate the effectiveness of LLM-powered SQL snippet generation by comparingArgus’s method against a variant that only uses SQLancer’s grammarbased generator to build the snippet corpus during the CAQ instantiation phase. We use the code coverage metric in Sec. 7.2 to evaluate both methods after a 24-hour testing on DuckDB. As shown in Fig. 6,Argusconsistently outperforms the grammar-based",
+      "section": "7.6 Component-wise Analysis",
+      "page": 22,
+      "char_offset": 69996,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M55",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We evaluate the effectiveness of LLM-powered SQL snippet generation by comparingArgus’s method against a variant that only uses SQLancer’s grammarbased generator to build the snippet corpus during the CAQ instantiation phase.",
+      "context_before": "icle 140. Publication date: June 2026. 140:22 Qiuyang Mang et al. Argus SQLancer SQLancer++ Argus-grammar-based 0 4 8 12 16 20 24 Hours283032343638Line Coverage (%) (a) DuckDB line 0 4 8 12 16 20 24 Hours18192021Branch Coverage (%) (b) DuckDB branch Fig. 6. Ablation study of code coverage achieved by Argus and a variant using only SQLancer’s grammarbased generator on DuckDB over 24-hour runs. 1CREATE TABLEt1(c0BOOLEAN, c1BOOLEAN); 2SELECT false FROMt1; 3SELECT(t1.c1 < t1.c0) < (t1.c1 = t1.c1)FROMt1; Listing 12. Incorrect equivalence proof in SQLSolver [13]. LLM-powered SQL snippet generation.",
+      "context_after": "We use the code coverage metric in Sec. 7.2 to evaluate both methods after a 24-hour testing on DuckDB. As shown in Fig. 6,Argusconsistently outperforms the grammar-based variant in both line and branch coverage. Without LLM-generated snippets, the grammar-based variant can still outperform baselines in terms of line coverage but struggles with branch coverage. 8 Discussion Additional finding.In addition to improving DBMS robustness, our testing also revealed several bugs in existing SQL equivalence provers. These bugs could lead to inherent false positives that, if unaddressed, would hinder",
+      "section": "7.6 Component-wise Analysis",
+      "page": 22,
+      "char_offset": 70338,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M56",
+      "found_by": "citation_marker",
+      "surface": "[ 3,5,24,47–\n49,74]",
+      "technique": "qpg",
+      "sentence": "(1)Arguscurrently focuses on relational DBMS and SELECT queries like prior works [ 3,5,24,47– 49,74], but the underlying principles can be extended to other types of databases and query languages, such as graph DBMS [23, 34, 37, 86] and spatial DBMS testing [12].",
+      "context_before": "can grow rapidly with the size of the CAQ pairs analyzed. This may limit the query sizes of the test oracles thatArguscan generate. Nevertheless, these restrictions from SQL equivalence provers can be mitigated inArgus’s test cases instantiation phase, which can generate complex SQL queries by filling in CAQ placeholders with diverse SQL snippets. Those snippets can include SQL features that currently lack support in existing provers, as well as increase the overall complexity of the generated test cases. Limitations.Our approach has two limitations that present opportunities for future work.",
+      "context_after": "(2)Argusfocuses on automatically discovering test oracles, but does not provide a mechanism to prioritize or rank the generated oracles. We believe that, afterArgus, DBMS testing research can shift from manually crafting test oracles to developing techniques for prioritizing and selecting the most effective oracles from a large pool of LLM-generated candidates. 9 Related Work LLM in systems research.LLMs have been applied to various system research tasks, including code generation [ 20,27,39,44,66], automated tuning [ 18,76], data processing [ 32,33,43,54,73], program analysis [28], networkin",
+      "section": "8 Discussion",
+      "page": 23,
+      "char_offset": 74295,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M57",
+      "found_by": "name",
+      "surface": "ShQvel",
+      "technique": null,
+      "sentence": "Similarly, ShQvel [ 81] uses LLMs to generate feature-rich SQL queries for testing DBMS with manually crafted test oracles, while SQLStorm [ 52] employs LLMs to synthesize a large-scale benchmark for evaluating performance of DBMSs.",
+      "context_before": "earch.LLMs have been applied to various system research tasks, including code generation [ 20,27,39,44,66], automated tuning [ 18,76], data processing [ 32,33,43,54,73], program analysis [28], networking [77], and algorithm discovery [41, 50, 72]. (1)In LLM-aided testing, prior work has primarily focused on generating effective inputs. For example, Fuzz4all [ 65] is a universal fuzzing framework that leverages LLMs to generate test inputs for various applications, while other research applies LLMs to fuzzing in specific domains such as compilers [ 67], kernels [ 68], and smart contracts [ 55].",
+      "context_after": "While prior approaches generate complex test inputs to find crash bugs, our work focuses on generating test oracles to detect logic bugs, along with effective methods for test oracle initializations, which are orthogonal and complementary to existing efforts. (2)In LLM-for-DBMS, prior work mainly focus on text-to-SQL [ 21], query optimization [ 29,59], dialect translation [ 84], and hyperparameter tuning [ 18,85]. Among these works, LLM-R2 [ 29] provides a framework to guide LLMs’ use of existing rules,e.g.,Calcite, to optimize SQL queries. Though this method can also be applied to our task,",
+      "section": "9 Related Work",
+      "page": 23,
+      "char_offset": 75557,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M58",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Test query generation focuses on automatically generating complex SQL queries to find crashes, with techniques being either grammar-based, exemplified by tools like SQLsmith [ 53] and SQLancer [ 49], or mutationbased, as seen in Griffin [ 16].",
+      "context_before": "dge, our 4, No. 3 (SIGMOD), Article 140. Publication date: June 2026. 140:24 Qiuyang Mang et al. work is the first to leverage LLMs to generate test oracles for DBMSs, and we believe that our rule generation framework can also be applied to other DBMS tasks in the future, such as query optimization with performance improvement. DBMS testing and verification.There is a rich body of work on DBMS testing, verification, and their applications. Prior work in DBMS testing has largely followed two distinct research paths:test query generationfor fuzzing andtest oracle designfor detecting logic bugs.",
+      "context_after": "Grammar-based techniques typically use predefined grammars to construct valid SQL queries, while mutation-based approaches modify existing queries to create new test cases. Beyond the generation of test queries, recent research has also explored constructing meaningful database states [ 1,8,9,51,69]. In parallel, the second path concentrates on test oracles for detecting “silent” bugs such as logic and performance issues, an area that has traditionally relied on significant manual effort. For instance, oracles like TLP [ 48], NoREC [ 47], and EET [ 24] were designed for logic bugs, while othe",
+      "section": "9 Related Work",
+      "page": 24,
+      "char_offset": 77212,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M59",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For instance, oracles like TLP [ 48], NoREC [ 47], and EET [ 24] were designed for logic bugs, while others such as CERT [ 4] and Apollo [ 25] target performance issues.",
+      "context_before": ", exemplified by tools like SQLsmith [ 53] and SQLancer [ 49], or mutationbased, as seen in Griffin [ 16]. Grammar-based techniques typically use predefined grammars to construct valid SQL queries, while mutation-based approaches modify existing queries to create new test cases. Beyond the generation of test queries, recent research has also explored constructing meaningful database states [ 1,8,9,51,69]. In parallel, the second path concentrates on test oracles for detecting “silent” bugs such as logic and performance issues, an area that has traditionally relied on significant manual effort.",
+      "context_after": "This manual crafting of oracles remains a primary bottleneck in fully automated DBMS testing. Argusis the first work that tackles this critical challenge by automatically discovering test oracles. Recent work has focused on improving the effectiveness of SQL equivalence provers [ 10,11,13, 63,83] and disprovers [ 19,79]. These tools have been widely applied in various DBMS scenarios, including query rewriting [ 64], text-to-SQL [ 70], and user-database interaction [ 45]. Among these, Wetune [ 64] is the most relevant to our work, as it employs an SQL equivalence prover to verify query rewriti",
+      "section": "9 Related Work",
+      "page": 24,
+      "char_offset": 77950,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec",
+        "cert"
+      ]
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/joyemang33/Argus",
+    "markers": [
+      "sqlancer_source_in_nested_artifact"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/joyemang33/Argus",
+        "source_type": "github_repository",
+        "excerpt": "# Argus: Automated Discovery of Test Oracles for Database Management Systems Using LLMs\n\nArgus is a novel framework for automatically discovering and instantiating test oracles to find logic bugs in Database Management Systems (DBMSs) using Large Language Models (LLMs).",
+        "note": "Repository names this paper.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/joyemang33/Argus/blob/main/MetamorphicCoverageArtifact/Code/mc-guided_fuzzing/src/sqlancer/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+        "excerpt_is_verbatim": true,
+        "note": "MetamorphicCoverageArtifact/Code/mc-guided_fuzzing/src/sqlancer/Randomly.java is SQLancer's Randomly.java, but it sits under MetamorphicCoverageArtifact, which looks like a bundled copy of another project",
+        "content_sha256": "sha256:7f3c136a45805e0fd7b278f3c6f2693b5721e9f54426d26c66823ea7d4a8b83b",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M37",
+        "M38",
+        "M50",
+        "M55"
+      ],
+      "describes_as_state_of_the_art": [
+        "M31"
+      ]
+    },
+    "suppressed": [
+      {
+        "mention_id": "M42",
+        "pattern": "compares_with",
+        "suppressed_because": "the sentence is framed as related work, so it describes somebody else's approach"
+      }
+    ]
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:11:05Z",
+    "is_model_written": true,
+    "summary": "Argus automates what has been the manual part of DBMS test oracle design: inventing mechanisms that generate equivalent query pairs. It builds on the Constrained Abstract Query, a SQL skeleton with placeholders and conditions on how they may be filled; a language model proposes pairs of skeletons, their equivalence is formally proven with a SQL equivalence solver, and only then are the placeholders instantiated with synthesised snippets. On five extensively tested DBMSs it found 41 previously unknown bugs, 36 of them logic bugs.",
+    "narrative": "Argus uses SQLancer++'s query generator to produce seed queries while deliberately not using its predefined oracles, since discovering oracles is the paper's own subject. It measures itself against SQLancer, SQLancer++ and EET on coverage, and calls SQLancer a state-of-the-art open-source DBMS testing framework. Its artifact carries SQLancer's own source files.",
+    "roles": {
+      "M3": "definition",
+      "M4": "background",
+      "M5": "definition",
+      "M6": "definition",
+      "M7": "reuse_component",
+      "M11": "reuse_component",
+      "M13": "background",
+      "M14": "reuse_component",
+      "M28": "result_comparison",
+      "M29": "result_comparison",
+      "M30": "baseline",
+      "M31": "state_of_the_art",
+      "M37": "result_comparison",
+      "M50": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M7",
+          "M11",
+          "M14"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M7",
+            "sentence": "These snippets can be generated offline by a hybrid approach combining an LLM to cover diverse database features and a high-throughput generator, such as SQLancer [ 49], even though the prover currently cannot reason about them.",
+            "section": "1 Introduction",
+            "page": 4
+          },
+          {
+            "mention_id": "M11",
+            "sentence": "Specifically, we use SQLancer++ [ 82]’s query generator to produce seed queries, but we do not use their predefined test oracles (such as TLP [48] and NoREC [47]) for bug detection.",
+            "section": "3 Argus Overview",
+            "page": 5
+          },
+          {
+            "mention_id": "M14",
+            "sentence": ",SQLancer can be directly used for schema and concrete 4, No.",
+            "section": "5.1 Database Seeding",
+            "page": 8
+          }
+        ],
+        "reasoning": "M11 states that SQLancer++'s query generator produces the seed queries while its predefined oracles are deliberately not used, and M14 that SQLancer can be used directly for schema and data generation. So a component is reused rather than the framework being built on. The artifact carrying SQLancer source files agrees.",
+        "reuse_kind": "generator"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Argus discovers new oracles rather than extending an existing one; TLP appears as the worked example of what a Constrained Abstract Query can represent, not as something being generalised."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M30",
+          "M37",
+          "M50"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M30",
+            "sentence": "2 Code Coverage We comparedArguswith three DBMS logic bugs finding tools, SQLancer [ 3,47–49,74], SQLancer++ [ 82] and EET [ 24] in multiple coverage metrics.",
+            "section": "7.2 Code Coverage",
+            "page": 17
+          },
+          {
+            "mention_id": "M37",
+            "sentence": "1% higher line and branch coverage than SQLancer++, respectively, and 11.",
+            "section": "7.2 Code Coverage",
+            "page": 18
+          },
+          {
+            "mention_id": "M50",
+            "sentence": "571×more line, function, and branch coverage than SQLancer, respectively.",
+            "section": "7.2 Code Coverage",
+            "page": 19
+          }
+        ],
+        "reasoning": "Coverage is compared against SQLancer and SQLancer++ over 24-hour runs, with line, function and branch figures reported for each."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M31"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M31",
+            "sentence": "As a state-of-the-art, open-source DBMS testing framework, SQLancer supports most of the latest test oracles [ 4,57,74].",
+            "section": "7.2 Code Coverage",
+            "page": 17
+          }
+        ],
+        "reasoning": "M31 calls SQLancer a state-of-the-art, open-source DBMS testing framework supporting most of the latest test oracles."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2511_17377.json b/_data/papers/paper_arxiv_2511_17377.json
new file mode 100644
index 0000000..55f8bb3
--- /dev/null
+++ b/_data/papers/paper_arxiv_2511_17377.json
@@ -0,0 +1,754 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:24:18Z",
+  "paper": {
+    "id": "paper:arxiv:2511.17377",
+    "title": "Anomaly Pattern-guided Transaction Bug Testing in Relational Databases",
+    "authors": [
+      "Hui-Rong Xu",
+      "Shuang Liu",
+      "Xianyu Zhu",
+      "Qiyu Zhuang",
+      "Wei Lu",
+      "Xiaoyong Du"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2511.17377",
+    "s2_paper_id": "3987779a8bc9037eedb2b036f804413c2f1b1da9",
+    "url": "https://arxiv.org/abs/2511.17377",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2511.17377",
+      "retrieved_at": "2026-09-08T15:24:18Z",
+      "chars": 79349,
+      "content_sha256": "sha256:31aafacb864c6b7e63f59310417d5f509c67223062b6cdec008af9dfd125c036"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2818
+      },
+      {
+        "number": "2",
+        "title": "PRELIMINARY",
+        "start": 12550
+      },
+      {
+        "number": "2.1",
+        "title": "Database Transactions and Isolation level",
+        "start": 12564
+      },
+      {
+        "number": "2.2",
+        "title": "Anomaly patterns",
+        "start": 15160
+      },
+      {
+        "number": "3",
+        "title": "METHODOLOGY",
+        "start": 17122
+      },
+      {
+        "number": "3.1",
+        "title": "Overview",
+        "start": 17902
+      },
+      {
+        "number": "3.2",
+        "title": "Database and SQL statement Generation",
+        "start": 18766
+      },
+      {
+        "number": "3.2.1",
+        "title": "Database Generation. In database transaction testing, the",
+        "start": 19202
+      },
+      {
+        "number": "3.3",
+        "title": "Pattern-guided Transaction Generation",
+        "start": 24099
+      },
+      {
+        "number": "3.3.1",
+        "title": "Constraint Extraction. To implement anomaly pattern-guided",
+        "start": 25154
+      },
+      {
+        "number": "3.4",
+        "title": "Execution",
+        "start": 34196
+      },
+      {
+        "number": "3.4.1",
+        "title": "Transaction Schedule. During execution, we must ensure that",
+        "start": 35589
+      },
+      {
+        "number": "3.5",
+        "title": "Detection",
+        "start": 40437
+      },
+      {
+        "number": "3.5.1",
+        "title": "Explicit Error Detection. Explicit error detection refers to the",
+        "start": 41191
+      },
+      {
+        "number": "3.5.2",
+        "title": "Implicit Error Detection. Implicit error detection is a dynamic",
+        "start": 41949
+      },
+      {
+        "number": "4",
+        "title": "EXPERIMENT",
+        "start": 44446
+      },
+      {
+        "number": "4.1",
+        "title": "Experimental Setup",
+        "start": 44459
+      },
+      {
+        "number": "4.2",
+        "title": "Bug Detection Capability",
+        "start": 47625
+      },
+      {
+        "number": "4.3",
+        "title": "Comparison with Existing Approaches",
+        "start": 50831
+      },
+      {
+        "number": "4.4",
+        "title": "Ablation Study",
+        "start": 55435
+      },
+      {
+        "number": "4.5",
+        "title": "Case Study",
+        "start": 58463
+      },
+      {
+        "number": "5",
+        "title": "RELATED WORK",
+        "start": 65314
+      },
+      {
+        "number": "6",
+        "title": "CONCLUSION",
+        "start": 71851
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2025. MariaDB. https://mariadb.com/. (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2025. MySQL. https://www.mysql.com/. (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2025. OceanBase. https://www.oceanbase.com/. (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2025. SQLancer. https://github.com/sqlancer/sqlancer. (visited in April 2025).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "2025. SQLsmith. https://github.com/anse1/sqlsmith. (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Atul Adya, Barbara Liskov, and Patrick O’Neil. 2000. Generalized isolation level definitions. In Proceedings of 16th International Conference on Data Engineering (Cat. No. 00CB37073). IEEE, 67–78.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In ing (ICSE). IEEE, 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 8,
+      "text": "Philip A Bernstein and Nathan Goodman. 1983. Multiversion concurrency control—theory and algorithms. ACM Transactions on Database Systems (TODS) 8, 4 (1983), 465–483.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Ziyu Cui, Wensheng Dou, Qianwang Dai, Jiansen Song, Wei Wang, Jun Wei, and Dan Ye. 2022. Differentially testing database transactions for fun and profit. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Wensheng Dou, Ziyu Cui, Qianwang Dai, Jiansen Song, Dong Wang, Yu Gao, Wei Wang, Jun Wei, Lei Chen, Hanmo Wang, et al. 2023. Detecting isolation bugs via transaction oracle construction. In Proceedings of International Conference on Software Engineering (ICSE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free DBMS fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Ying Fu, Zhiyong Wu, Yuanliang Zhang, Jie Liang, Jingzhou Fu, Yu Jiang, Shanshan Li, and Xiangke Liao. 2024. THANOS: DBMS Bug Detection via Storage Engine Rotation Based Differential Testing. In 2025 IEEE/ACM 47th International Conference on Software Engineering (ICSE). IEEE Computer Society, 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Kaile Huang, Si Liu, Zhenge Chen, Hengfeng Wei, David Basin, Haixiang Li, and Anqun Pan. 2023. Efficient black-box checking of snapshot isolation in databases. arXiv preprint arXiv:2301.07313 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via {Graph-Based}Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting logic bugs in database engines via equivalent expression transformation. In 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24). 821–835.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. Apollo: Automatic detection and diagnosis of performance regressions in database systems. Proceedings of the VLDB Endowment 13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Kyle Kingsbury and Peter Alvaro. 2020. Elle: Inferring isolation anomalies from experimental observations. arXiv preprint arXiv:2003.10554 (2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Hsiang-Tsung Kung and John T Robinson. 1981. On optimistic methods for concurrency control. ACM Transactions on Database Systems (TODS) 6, 2 (1981), 213–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Haixiang Li, Xiao-Yan Li, Chang Liu, Xiaoyong Du, Wei Lu, and Anqun Pan. 2021. Systematic definition and classification of data anomalies in DBMS (English Version). CoRR abs/2110.14230 (2021). arXiv:2110.14230 https://arxiv.org/ abs/2110.14230",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Keqiang Li, Siyang Weng, Lyu Ni, Chengcheng Yang, Rong Zhang, Xuan Zhou, and Aoying Zhou. 2024. DBStorm: Generating Various Effective Workloads for Testing Isolation Levels. In Proceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. 755–767.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of {DBMS}with Coverage-based Guidance. In 31st USENIX Security Symposium (USENIX Security 22). 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Si Liu, Long Gu, Hengfeng Wei, and David Basin. 2024. Plume: Efficient and complete black-box checking of weak isolation levels. Proceedings of the ACM on Programming Languages 8, OOPSLA2 (2024), 876–904.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. In Proceedings of the 44th International Conference on Software Engineering. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Jim Melton. 2016. Iso/iec 9075-2 information technology-database languages SQL—Part 2: foundation (SQL/foundation). ISO/IEC 2016, E (2016), 9075–2.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing database systems via differential query execution. In Engineering (ICSE). IEEE, 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Cheng Tan, Changgeng Zhao, Shuai Mu, and Michael Walfish. 2020. Cobra: Making transactional {key-value}stores verifiably serializable. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 63–80.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting logic bugs of join optimizations in dbms. Proceedings of the ACM on Management of Data 1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Alexander Thomasian. 1993. Two-phase locking performance and its thrashing behavior. ACM Transactions on Database Systems (TODS) 18, 4 (1993), 579–625.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Jian Zhang, Ye Ji, Shuai Mu, and Cheng Tan. 2023. Viper: A fast snapshot isolation checker. In Proceedings of the Eighteenth European Conference on Computer Systems. 654–671.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 955–970.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 4,
+      "text": "2025. SQLancer. https://github.com/sqlancer/sqlancer. (visited in April 2025).",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 7,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In ing (ICSE). IEEE, 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 14,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via {Graph-Based}Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23).",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 14]",
+      "technique": null,
+      "sentence": "Existing approaches for testing transaction anomalies, such as Troc [ 10] and TxCheck [ 14], typically generate test transactions by randomly selecting SQL statements from a predefined set.",
+      "context_before": "; 12. / ∗txn1 ∗ /COMMIT; 13. / ∗F i n a l l y database s t a t e ∗ / 14. SELECT ∗FROM t1; − [ ( 10, 10, 3 ) ] Figure 1: Motivation example: a logic bug detected by APTrans in Mariadb (Bug#36330) into the shared table and commits, 𝑡𝑥𝑛1inserts into the same table, updates according to the insert and commit). Second, the bug does not trigger any explicit error message, and we need to uncover it by checking the database status after each statement. It is difficult to build an oracle to detect incorrect database status, as it is hard to determine the correct result from a single random transaction.",
+      "context_after": "This makes it challenging to synthesize transactions like the one shown in Example 1, which involve a pattern with multiple constraints and complex dependencies. Specifically, in this example, not only must the SQL statements and schedule be deterministic, but it is also essential that different SQL statements access the same data items—an aspect that is difficult to achieve with randomly generated methods. Troc [ 10] adopts a differential testing strategy by constructing a view for each SQL statement and comparing the results of concurrent transaction executions against these viewbased outpu",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 6901,
+      "cited_reference": {
+        "number": 14,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via {Graph-Based}Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23).",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 14]",
+      "technique": null,
+      "sentence": "TxCheck [ 14], on the other hand, attempts to decouple a transaction into independent and dependent SQL statements.",
+      "context_before": "he same data items—an aspect that is difficult to achieve with randomly generated methods. Troc [ 10] adopts a differential testing strategy by constructing a view for each SQL statement and comparing the results of concurrent transaction executions against these viewbased outputs. However, the process of constructing these views introduces considerable overhead, requiring the implementation of a parallel transaction execution engine. Furthermore, due to differences in implementation semantics, the differential approach often yields a high rate of false positives (as discussed in Section 4.1).",
+      "context_after": "It then generates semantically equivalent variants by reordering the independent statements while preserving the order of the dependent ones. The results of the two transaction executions are compared to identify inconsistencies. TxCheck cannot solve cases where transactions exhibit cyclic dependencies, such as the ( 𝑊1[𝑥1]𝑊2[𝑦1]𝐶2𝑅1[𝑦1]𝐶1) cycle in Example 1, which significantly limits its applicability to more complex transactional workloads. ELLE [ 17], a transaction anomaly checker, detects anomalies by identifying cycles in the transaction execution history. It is unable to detect bugs t",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 8012,
+      "cited_reference": {
+        "number": 14,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via {Graph-Based}Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23).",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We develop our database and SQL statement generation method based on SQLancer [ 4] and extend it to support table join operations for both database generation and SQL statement generation.",
+      "context_before": "and a two-phase bug detection mechanism designed to identify and validate transactionrelated bugs with high accuracy. Figure 2illustrates the overall architecture of APTrans, which consists of four modules: database and SQL statement generation, transaction generation guided by the anomaly pattern, transaction execution, and transaction bug detection. 3.2 Database and SQL statement Generation As the initial step of automated transaction generation, we first create the underlying databases on which the transactions will operate, along with the basic SQL statements that compose each transaction.",
+      "context_after": "3.2.1 Database Generation. In database transaction testing, the quality of the database structure and data initialization is critical to the validity of test cases. To simulate real-world scenarios, which usually involve multiple tables connected with join keys, we extend SQLancer to support table join operations for both database generation and SQL statement generation. Moreover, to support the transaction bug detection, which requires checking on the recorded data and the corresponding version, we introduced a unique row ID and a version column in the generated table. Table Generation. The",
+      "section": "3.2 Database and SQL statement Generation",
+      "page": 4,
+      "char_offset": 19012,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To simulate real-world scenarios, which usually involve multiple tables connected with join keys, we extend SQLancer to support table join operations for both database generation and SQL statement generation.",
+      "context_before": "3.2 Database and SQL statement Generation As the initial step of automated transaction generation, we first create the underlying databases on which the transactions will operate, along with the basic SQL statements that compose each transaction. We develop our database and SQL statement generation method based on SQLancer [ 4] and extend it to support table join operations for both database generation and SQL statement generation. 3.2.1 Database Generation. In database transaction testing, the quality of the database structure and data initialization is critical to the validity of test cases.",
+      "context_after": "Moreover, to support the transaction bug detection, which requires checking on the recorded data and the corresponding version, we introduced a unique row ID and a version column in the generated table. Table Generation. The base version of SQLancer supports the generation of isolated tables and the basic CREATE TABLE statements. It randomly selects column types such as INT ,TEXT, and BOOLEAN, and applies common constraints such as PRIMAR YKEY ,NOT NULL, and UNIQUE, in the generated table creation statement. We extend the table generation functionality to support multi-table generation by add",
+      "section": "3.2.1 Database Generation. In database transaction testing, the",
+      "page": 4,
+      "char_offset": 19366,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The base version of SQLancer supports the generation of isolated tables and the basic CREATE TABLE statements.",
+      "context_before": "tion. 3.2.1 Database Generation. In database transaction testing, the quality of the database structure and data initialization is critical to the validity of test cases. To simulate real-world scenarios, which usually involve multiple tables connected with join keys, we extend SQLancer to support table join operations for both database generation and SQL statement generation. Moreover, to support the transaction bug detection, which requires checking on the recorded data and the corresponding version, we introduced a unique row ID and a version column in the generated table. Table Generation.",
+      "context_after": "It randomly selects column types such as INT ,TEXT, and BOOLEAN, and applies common constraints such as PRIMAR YKEY ,NOT NULL, and UNIQUE, in the generated table creation statement. We extend the table generation functionality to support multi-table generation by adding foreign key constraints based on candidate key matching. When generating a foreign key constraint for a column in a table, we scan all existing PRIMAR YKEY or UNIQUE columns in other tables and select candidatecolumns with matching data types to establish reference relationships. Data Generation. SQLancer generates random data",
+      "section": "3.2.1 Database Generation. In database transaction testing, the",
+      "page": 4,
+      "char_offset": 19796,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer generates random data by creating INSER T statements for tables.",
+      "context_before": "basic CREATE TABLE statements. It randomly selects column types such as INT ,TEXT, and BOOLEAN, and applies common constraints such as PRIMAR YKEY ,NOT NULL, and UNIQUE, in the generated table creation statement. We extend the table generation functionality to support multi-table generation by adding foreign key constraints based on candidate key matching. When generating a foreign key constraint for a column in a table, we scan all existing PRIMAR YKEY or UNIQUE columns in other tables and select candidatecolumns with matching data types to establish reference relationships. Data Generation.",
+      "context_after": "However, the random generation approach cannot guarantee the generated data satisfying the foreign key constraints we have implemented, as foreign keys require references to already existing values, and randomly generated values may not be present in the referenced columns. To address this issue, we introduced a data logging feature that records the primary key values of all inserted data. When generating data with foreign key constraints, we select existing values from the recorded data to ensure both data consistency and referential integrity. Unique Row ID and Version Column. To enable tra",
+      "section": "3.2.1 Database Generation. In database transaction testing, the",
+      "page": 4,
+      "char_offset": 20476,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Building upon the SQLancer framework, we have extended the SQL statement generation to include JOIN operations and type constraints, thereby producing a more diverse set of SQL statements while ensuring the semantic correctness of the SQL statements.",
+      "context_before": "the record is modified. This column allows us to identify whether a transaction has read an old version of a data item, assisting in detecting read/write version conflicts. Note that these two columns do not affect the actual execution semantics of the transaction; they are solely used for dependency reconstruction and anomaly detection within the testing framework. 3.2.2 Statement Generation. SQL statement generation is fundamental for testing transaction behavior, as its quality directly impacts the establishment of transaction dependencies and the effectiveness of subsequent bug detection.",
+      "context_after": "SQL Statement Generation. We use SQLancer to generate a large number of random SQL statements for transaction generation. However, SQLancer suffers from numerous semantic errors, most of which are caused by type mismatches in expressions. To address this issue, we introduced type constraint checking during expression generation to ensure that the generated expressions conform to the required expression type. Anomaly Pattern-guided Transaction Bug Testing in Relational Databases SIGMOD, May 31–June 05, 2026, Bengaluru, India JOIN Statement Generation. To support join statement on multiple tabl",
+      "section": "3.2.1 Database Generation. In database transaction testing, the",
+      "page": 4,
+      "char_offset": 22311,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We use SQLancer to generate a large number of random SQL statements for transaction generation.",
+      "context_before": "olely used for dependency reconstruction and anomaly detection within the testing framework. 3.2.2 Statement Generation. SQL statement generation is fundamental for testing transaction behavior, as its quality directly impacts the establishment of transaction dependencies and the effectiveness of subsequent bug detection. Building upon the SQLancer framework, we have extended the SQL statement generation to include JOIN operations and type constraints, thereby producing a more diverse set of SQL statements while ensuring the semantic correctness of the SQL statements. SQL Statement Generation.",
+      "context_after": "However, SQLancer suffers from numerous semantic errors, most of which are caused by type mismatches in expressions. To address this issue, we introduced type constraint checking during expression generation to ensure that the generated expressions conform to the required expression type. Anomaly Pattern-guided Transaction Bug Testing in Relational Databases SIGMOD, May 31–June 05, 2026, Bengaluru, India JOIN Statement Generation. To support join statement on multiple tables in the database, we have extended the generation of SQL statements to include JOIN operations. Specifically, we generat",
+      "section": "3.2.1 Database Generation. In database transaction testing, the",
+      "page": 4,
+      "char_offset": 22588,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "However, SQLancer suffers from numerous semantic errors, most of which are caused by type mismatches in expressions.",
+      "context_before": ".2 Statement Generation. SQL statement generation is fundamental for testing transaction behavior, as its quality directly impacts the establishment of transaction dependencies and the effectiveness of subsequent bug detection. Building upon the SQLancer framework, we have extended the SQL statement generation to include JOIN operations and type constraints, thereby producing a more diverse set of SQL statements while ensuring the semantic correctness of the SQL statements. SQL Statement Generation. We use SQLancer to generate a large number of random SQL statements for transaction generation.",
+      "context_after": "To address this issue, we introduced type constraint checking during expression generation to ensure that the generated expressions conform to the required expression type. Anomaly Pattern-guided Transaction Bug Testing in Relational Databases SIGMOD, May 31–June 05, 2026, Bengaluru, India JOIN Statement Generation. To support join statement on multiple tables in the database, we have extended the generation of SQL statements to include JOIN operations. Specifically, we generate four types of JOIN operations: INNER JOIN ,LEFT JOIN, RIGHT JOIN, and CROSS JOIN. The core of generating the JOIN S",
+      "section": "3.2.1 Database Generation. In database transaction testing, the",
+      "page": 4,
+      "char_offset": 22684,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "It takes the three types of constraints extracted from an anomaly pattern, the set of SQL statements and database generated by SQLancer as inputs and progressively construct transactions satisfying the constraints.",
+      "context_before": "te anomaly pattern as an example, the schedule order constraint extracted from this pattern is [1,2,2,1,1]. These three types of constraints—statement type, data access, and schedule order—together form a semantically equipollent constraint set with the given anomaly pattern. In this way, we decompose the complex anomaly patterns into a set of simple constraints, which simplify our algorithm for anomaly pattern-guided transaction generation, which will be introduced in section 3.3.2. 3.3.2 Transaction Generation. The anomaly pattern-guided transaction generation method is shown in Algorithm 1.",
+      "context_after": "Guided by the constraints, there are three main steps in the algorithm: statement selection, condition alignment, and transaction composition. Statement Selection. In this step, we select the appropriate SQL statement for each operation extracted from the pattern (lines 2-4). Based on the operation type (e.g., read, write, commit, rollback), the corresponding SQL statement is chosen from the set of SQL statements we randomly generated. As shown in Figure 3, we select SQL statements from lines 3 to 7 in the example. However, at this step, there is no guarantee on the exact WHERE conditions in",
+      "section": "3.3.1 Constraint Extraction. To implement anomaly pattern-guided",
+      "page": 6,
+      "char_offset": 30628,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 14]",
+      "technique": null,
+      "sentence": "We compare APTrans with two stateof-the-art testing methods: TxCheck [ 14] and Troc [ 10].",
+      "context_before": "provides very weak guarantees and is rarely used in practice. Moreover, several databases, including OceanBase, do not support this isolation level. As a result, we excluded read uncommitted from our evaluation. Notably, while OceanBase claims that it offers the Serializable isolation level, its actual implementation is based on snapshot isolation, which allows write skew. For each isolation level of each database, we ran APTrans continuously. We manually analyzed the bugs reported by APTrans and submitted the identified issues to the corresponding development communities. Compared Approaches.",
+      "context_after": "TxCheck is a metamorphic testing approach, which first randomly generates test transactions and executes them. It then decouples the transaction into semantically equivalent SQL statements based on their dependency topology. After that, it compares the results of executing the decoupled statements with the results of the original Anomaly Pattern-guided Transaction Bug Testing in Relational Databases SIGMOD, May 31–June 05, 2026, Bengaluru, India Table 5: Bugs detected by APTrans DBMSBug IDStatus SeverityIsolation LevelOracle MySQL115978 Duplicate Moderate SER Implicit 117218 Confirm Moderate",
+      "section": "4.1 Experimental Setup",
+      "page": 8,
+      "char_offset": 45988,
+      "cited_reference": {
+        "number": 14,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via {Graph-Based}Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23).",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 14]",
+      "technique": null,
+      "sentence": "Troc [ 10] and TxCheck [ 14] are recent methods for detecting database transaction bugs.",
+      "context_before": "query results were used in subsequent write operations, this bug could lead to even more severe data inconsistencies. This example can be generated and detected by our pattern 36 𝑊1[𝑥1]𝑊2[𝑦1]𝐶2𝑅1[𝑦1]𝐶1in Table 3, which represents a cyclical anomaly arising from the combination of write-write and writeread dependencies. 5 RELATED WORK In this section, we discuss works that are closely related with our approach. In particular, we discuss existing research on RDBMS transaction bug testing, RDBMS transaction history validation and general RDBMS bug testing. RDBMS Transactional Bug Testing Methods.",
+      "context_after": "Both generate transactions by randomly composing SQL statements, making it challenging to trigger transaction bugs, which usually occur under specific patterns. Troc is a differential testing approach, it compares concurrent transaction execution results against its constructed views to identify differences in query outcomes. However, the view-construction resembles re-implementation of the transaction execution logic, which incurs overhead. Moreover, due to different design and implementation choices of its view with the databases under test, this method often leads to high false positives,",
+      "section": "5 RELATED WORK",
+      "page": 12,
+      "char_offset": 65593,
+      "cited_reference": {
+        "number": 14,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via {Graph-Based}Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23).",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Its successor, SQLancer [ 25–27], introduces three metamorphic testing oracles: PQS [ 27], NoREC [ 25], and TLP [ 26], significantly improving its ability to detect logical bugs.",
+      "context_before": "mprehensive support for SQL transaction semantics, such as multi-row operations in single statements, constraint validations, and complex data structures. These approaches focus on analyzing bugs from existing execution histories, rather than generating test cases to proactively test the database as our method does. General RDBMS Testing Methods. There are several general relational DBMS testing methods focusing on the execution behavior of SQL statements. SQLSmith [ 5], a pioneering generation-based approach, detects crashes by generating random SQL queries and monitoring for explicit errors.",
+      "context_after": "Recent innovations have introduced paradigm shifts in testing strategies: QPG [ 7] uses query plan-guided database state mutation, while DQE [ 28] applies differential execution analysis across SELECT, UPDATE, and DELETE statements. TQS [ 30] specifically targets join optimization errors with an oracle based on table partitioning. ETT [ 15] detects bugs through equivalent expression transformation, and THANOS [ 12] identifies RDBMS bugs via differential testing, involving storage engine rotation. There are also fuzzing approaches on RDBMSs. SQUIRREL [ 33] integrates language validity constrai",
+      "section": "5 RELATED WORK",
+      "page": 12,
+      "char_offset": 70170,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Recent innovations have introduced paradigm shifts in testing strategies: QPG [ 7] uses query plan-guided database state mutation, while DQE [ 28] applies differential execution analysis across SELECT, UPDATE, and DELETE statements.",
+      "context_before": "n analyzing bugs from existing execution histories, rather than generating test cases to proactively test the database as our method does. General RDBMS Testing Methods. There are several general relational DBMS testing methods focusing on the execution behavior of SQL statements. SQLSmith [ 5], a pioneering generation-based approach, detects crashes by generating random SQL queries and monitoring for explicit errors. Its successor, SQLancer [ 25–27], introduces three metamorphic testing oracles: PQS [ 27], NoREC [ 25], and TLP [ 26], significantly improving its ability to detect logical bugs.",
+      "context_after": "TQS [ 30] specifically targets join optimization errors with an oracle based on table partitioning. ETT [ 15] detects bugs through equivalent expression transformation, and THANOS [ 12] identifies RDBMS bugs via differential testing, involving storage engine rotation. There are also fuzzing approaches on RDBMSs. SQUIRREL [ 33] integrates language validity constraints with coverage feedback mechanisms, enhancing the generation of test cases. SQLRight [ 21] implements semantic-aware, coverage-guided mutation to improve the effectiveness of testing. GRIFFIN [ 11] advances metadata dependency ana",
+      "section": "5 RELATED WORK",
+      "page": 12,
+      "char_offset": 70349,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "uses_infrastructure": [
+        "M3"
+      ],
+      "extends_technique": [
+        "M4"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:57:35Z",
+    "is_model_written": true,
+    "summary": "APTrans tests transaction handling in relational DBMSs under different isolation levels. It identifies two obstacles: generating transactions that actually expose bugs, since those need specific transactional constraints, and deciding whether an outcome is wrong, since the correct result of randomly generated transactions is usually unknown. It addresses the first with generation guided by predefined anomaly patterns and the second with a two-phase explicit and implicit error detection process. On MySQL, MariaDB and OceanBase it found 13 previously unknown transaction bugs, 11 confirmed.",
+    "narrative": "APTrans's database and SQL generation is SQLancer's, extended. The paper states it develops its generation method based on SQLancer and builds on the SQLancer framework, extending it to support table joins in both database and statement generation, because the base version generates isolated tables and real workloads involve tables connected by join keys. Type constraints were added for the same reason, since SQLancer's expressions produce many semantic errors from type mismatches. Those generated statements and databases are then the input to the anomaly-pattern constraint solver, which is APTrans's own. Its baselines are TxCheck and Troc rather than any SQLancer oracle.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "reuse_component",
+      "M4": "reuse_component",
+      "M5": "definition",
+      "M6": "definition",
+      "M7": "reuse_component",
+      "M8": "reuse_component",
+      "M9": "motivation",
+      "M10": "reuse_component",
+      "M11": "background",
+      "M12": "background",
+      "M13": "definition",
+      "M14": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M3",
+          "M4",
+          "M7",
+          "M8"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "We develop our database and SQL statement generation method based on SQLancer [ 4] and extend it to support table join operations for both database generation and SQL statement generation.",
+            "section": "3.2 Database and SQL statement Generation",
+            "page": 4
+          },
+          {
+            "mention_id": "M4",
+            "sentence": "To simulate real-world scenarios, which usually involve multiple tables connected with join keys, we extend SQLancer to support table join operations for both database generation and SQL statement generation.",
+            "section": "3.2.1 Database Generation. In database transaction testing, the",
+            "page": 4
+          },
+          {
+            "mention_id": "M7",
+            "sentence": "Building upon the SQLancer framework, we have extended the SQL statement generation to include JOIN operations and type constraints, thereby producing a more diverse set of SQL statements while ensuring the semantic correctness of the SQL statements.",
+            "section": "3.2.1 Database Generation. In database transaction testing, the",
+            "page": 4
+          },
+          {
+            "mention_id": "M8",
+            "sentence": "We use SQLancer to generate a large number of random SQL statements for transaction generation.",
+            "section": "3.2.1 Database Generation. In database transaction testing, the",
+            "page": 4
+          }
+        ],
+        "reasoning": "M3 states the database and SQL statement generation is developed based on SQLancer, M4 and M7 that they extended it to support joins and type constraints, and M8 that SQLancer generates the statements transaction construction consumes. The reuse is of the generator; the anomaly-pattern machinery on top is the paper's own.",
+        "reuse_kind": "generator"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "What is extended is SQLancer's generation -- joins and type constraints -- not any of its test oracles. APTrans's oracle is anomaly-pattern-based error detection."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M11 names the two comparison methods as TxCheck and Troc. No SQLancer oracle is run against APTrans."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "TxCheck and Troc are the methods called state-of-the-art here; SQLancer is described by what it generates."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2601_15074.json b/_data/papers/paper_arxiv_2601_15074.json
new file mode 100644
index 0000000..870c874
--- /dev/null
+++ b/_data/papers/paper_arxiv_2601_15074.json
@@ -0,0 +1,578 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:31:20Z",
+  "paper": {
+    "id": "paper:arxiv:2601.15074",
+    "title": "SmartOracle - An Agentic Approach to Mitigate Noise in Differential Oracles",
+    "authors": [
+      "Srinath Srinivasan",
+      "Tim Menzies",
+      "Marcelo d’Amorim"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2601.15074",
+    "s2_paper_id": "ceb1c8e11371f1b7b021fd9de6588741863607c2",
+    "url": "https://arxiv.org/abs/2601.15074",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2601.15074",
+      "retrieved_at": "2026-09-08T14:31:20Z",
+      "chars": 83810,
+      "content_sha256": "sha256:953f1066c5f8ea934bf0b9f3f83e28dbff7b4bd5e7a749d1954e07cc16186b04"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 26,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1736
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 8415
+      },
+      {
+        "number": "2.1",
+        "title": "Motivation",
+        "start": 8428
+      },
+      {
+        "number": "2.2",
+        "title": "Differential Fuzzing",
+        "start": 12454
+      },
+      {
+        "number": "2.3",
+        "title": "LLM Agents",
+        "start": 15331
+      },
+      {
+        "number": "2.4",
+        "title": "Semi-supervised label propagation",
+        "start": 18347
+      },
+      {
+        "number": "2.5",
+        "title": "Differences from previous work",
+        "start": 21482
+      },
+      {
+        "number": "3",
+        "title": "Experimental Methodology",
+        "start": 23936
+      },
+      {
+        "number": "3.1",
+        "title": "Datasets",
+        "start": 24699
+      },
+      {
+        "number": "3.2",
+        "title": "Sequential LLM Prompting",
+        "start": 28543
+      },
+      {
+        "number": "3.3",
+        "title": "Agent Design",
+        "start": 31835
+      },
+      {
+        "number": "3.3.7",
+        "title": "Tools. Three tools are exposed to each of the agents:",
+        "start": 37814
+      },
+      {
+        "number": "3.4",
+        "title": "Current Version bug finding",
+        "start": 38351
+      },
+      {
+        "number": "3.5",
+        "title": "Semi-supervised Evaluation Methodology",
+        "start": 40960
+      },
+      {
+        "number": "4",
+        "title": "Results",
+        "start": 43530
+      },
+      {
+        "number": "5",
+        "title": "Threats to Validity",
+        "start": 57187
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 60361
+      },
+      {
+        "number": "6.1",
+        "title": "Semi-Supervised Learning: Important?",
+        "start": 60374
+      },
+      {
+        "number": "6.2",
+        "title": "Architecture Beats Scale?",
+        "start": 61508
+      },
+      {
+        "number": "6.3",
+        "title": "Domain-Specific Challenges",
+        "start": 62480
+      },
+      {
+        "number": "6.4",
+        "title": "Tools Over Agents?",
+        "start": 63070
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 64743
+      },
+      {
+        "number": "8",
+        "title": "Future Work",
+        "start": 66106
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, Banff, Alberta. https://www.usenix.org/conference/osdi20/ presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 2,
+      "text": "Paschal C Amusuo, Dongge Liu, Ricardo Andres Calvo Mendez, Jonathan Metzman, Oliver Chang, and James C Davis. 2025. FalseCrashReducer: Mitigating False Positive Crashes in OSS-Fuzz-Gen Using Agentic AI. arXiv preprint arXiv:2510.02185 (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Chromium Authors. 2008. The Chromium Projects. https://www.chromium.org Google/Chromium Project.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Peter Belcak, Greg Heinrich, Shizhe Diao, Yonggan Fu, Xin Dong, Saurav Muralidharan, Yingyan Celine Lin, and Pavlo Molchanov. 2025. Small Language Models are the Future of Agentic AI. arXiv preprint arXiv:2506.02153 (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Fabrice Bellard and Charlie Gordon. [n. d.]. QuickJS JavaScript Engine. https://bellard.org/quickjs/, accessed 2025-11-03.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Lukas Bernhard, Tobias Scharnowski, Moritz Schloegel, Tim Blazytko, and Thorsten Holz. 2022. JIT-picking: Differential fuzzing of JavaScript engines. In Proceedings of the 2022 ACMSIGSAC Conference on Computer and Communications Security. 351–364.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Lukas Bernhard, Tobias Scharnowski, Moritz Schloegel, Tim Blazytko, and Thorsten Holz. 2022. JIT-Picking: Differential Fuzzing of JavaScript Engines. In Proceedings of the 2022 ACMSIGSAC Conference on Computer and Communications Security (Los Angeles, CA, USA) (CCS ’22). Association for Computing Machinery, New York, NY, USA, 351–364. doi:10.1145/3548606.3560624",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Jonas Bushart and Christian Rossow. 2023. ResolFuzz: Differential fuzzing of DNS resolvers. In European Symposium on Research in Computer Security. Springer, 62–80.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Victor de Lamo Castrillo, Habtom Kahsay Gidey, Alexander Lenz, and Alois Knoll. 2025. Fundamentals of Building Autonomous LLM Agents. arXiv preprint arXiv:2510.09244 (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "ChakraCore Project. 2015. ChakraCore. https://github.com/chakra-core/ChakraCore. Accessed: 2025-12-03.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Node.js Contributors. 2009. Node.js Documentation. https://nodejs.org Node.js Project.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Karine Even-Mendoza, Arindam Sharma, Alastair F. Donaldson, and Cristian Cadar. 2023. GrayC: Greybox Fuzzing of Compilers and Analysers for C. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (Seattle, WA, USA) (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 1219–1231. doi:10.1145/3597926.3598130 24 Srinivasan et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Alessio Ferrari and Paola Spoletini. 2025. Formal requirements engineering and large language models: A two-way roadmap. Information and Software Technology 181 (2025), 107697. doi:10.1016/j.infsof.2025.107697",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Mozilla Foundation. [n. d.]. SpiderMonkey JavaScript/WebAssembly Engine. https://spidermonkey.dev, accessed 2025-11-03.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Mozilla Foundation. 2002. Firefox Browser. https://www.mozilla.org/firefox/ Mozilla.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Google. [n. d.]. V8 JavaScript Engine. https://v8.dev, accessed 2025-11-03.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Diego Gosmar and Deborah A Dahl. 2025. Hallucination mitigation using agentic ai natural language-based frameworks. arXiv preprint arXiv:2501.13946 (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Junda He, Christoph Treude, and David Lo. 2025. LLM-Based Multi-Agent Systems for Software Engineering: Literature Review, Vision, and the Road Ahead. ACM Transactions on Software Engineering and Methodology 34, 5 (2025), 1–30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Sirui Hong, Mingchen Zhuge, Jonathan Chen, Xiawu Zheng, Yuheng Cheng, Jinlin Wang, Ceyao Zhang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, et al .2023. MetaGPT: Meta programming for a multi-agent collaborative framework. InThe Twelfth International Conference on Learning Representations.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Sihao Hu, Tiansheng Huang, Fatih İlhan, Selim Furkan Tekin, and Ling Liu. 2023. Large language model-powered smart contract vulnerability detection: New perspectives. In 2023 5th IEEE International Conference on Trust, Privacy and Security in Intelligent Systems and Applications (TPS-ISA). IEEE, 297–306.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Jaewon Hur, Suhwan Song, Dongup Kwon, Eunjin Baek, Jangwoo Kim, and Byoungyoung Lee. 2021. Difuzzrtl: Differential fuzz testing to find cpu bugs. In. IEEE, 1286–1303.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "IBM. 2025. What are Small Language Models (SLM)? Online. Available: https://www.ibm.com/think/topics/smalllanguage-models.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Apple Inc. [n. d.]. JavaScriptCore: WebKit JavaScript Engine. https://developer.apple.com/documentation/javascriptcore, accessed 2025-11-03.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Apple Inc. 2003. Safari Browser. https://www.apple.com/safari/ Apple.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Ecma International. 2025. ECMA-262: ECMAScript 2025 Language Specification. Online. Available: https://ecmainternational.org/publications-and-standards/standards/ecma-262/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Aaron Jaech, Adam Kalai, Adam Lerer, Adam Richardson, Ahmed El-Kishky, Aiden Low, Alec Helyar, Aleksander Madry, Alex Beutel, Alex Carney, et al. 2024. Openai o1 system card. arXiv preprint arXiv:2412.16720 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Zhiyuan Jiang, Xiyue Jiang, Ahmad Hazimeh, Chaojing Tang, Chao Zhang, and Mathias Payer. 2021. Igor: Crash deduplication through root-cause clustering. In Proceedings of the 2021 ACMSIGSAC Conference on Computer and Communications Security. 3318–3336.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Haolin Jin, Linghan Huang, Haipeng Cai, Jun Yan, Bo Li, and Huaming Chen. 2024. From llms to llm-based agents for software engineering: A survey of current, challenges and future. arXiv preprint arXiv:2408.02479 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jeewoo Jung and Taekyoung Kwon. 2025. Enhancing Differential Fuzzing of Cryptographic Libraries with Sustainable Hybrid Fuzzing and Crypto-Specific Mutation. In Information Security and Cryptology – ICISC 2024: 27th International Conference, Seoul, South Korea, November 20–22, 2024, Revised Selected Papers (Seoul, Korea (Republic of)). SpringerVerlag, Berlin, Heidelberg, 181–205. doi:10.1007/978-9",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "George Klees, Andrew Ruef, Benji Cooper, Shiyi Wei, and Michael Hicks. 2018. Evaluating fuzz testing. In Proceedings of the 2018 ACMSIGSAC conference on computer and communications security. 2123–2138.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Jizhe Li, Haoran Xu, Yongjun Wang, Zhiyuan Jiang, Huang Chun, Peidai Xie, Yongxin Chen, and Tian Xia. 2025. Fuzzing JavaScript JIT compilers with a high-quality differential test oracle. Computers & Security 159 (2025), 104660. doi:10.1016/j.cose.2025.104660",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Igor Lima, Jefferson Silva, Breno Miranda, Gustavo Pinto, and Marcelo d’Amorim. 2021. Exposing bugs in JavaScript engines through test transplantation and differential testing. Software Quality Journal 29, 1 (March 2021), 129–158. doi:10.1007/s11219-020-09537-8",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Suvodeep Majumder, Joymallya Chakraborty, and Tim Menzies. 2024. When less is more: on the value of “co-training” for semi-supervised software defect predictors. Empirical Software Engineering 29, 2 (2024), 51.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Zhenyy Mao, Jialong Li, Dongming Jin, Munan Li, and Kenji Tei. 2024. Multi-role consensus through llms discussions for vulnerability detection. In Companion (QRS-C). IEEE, 1318–1319.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "William M McKeeman. 1998. Differential testing for software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Barton P. Miller, Lars Fredriksen, and Bryan So. 1990. An empirical study of the reliability of UNIX utilities. Commun. ACM 33, 12 (Dec. 1990), 32–44. doi:10.1145/96267.96279",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Barton P. Miller, Lars Fredriksen, and Bryan So. 1990. An empirical study of the reliability of UNIX utilities. Commun. ACM 33, 12 (Dec. 1990), 32–44. doi:10.1145/96267.96279",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Inc. Moddable Tech. [n. d.]. XS JavaScript Engine (Moddable SDK). https://www.moddable.com/, accessed 2025-11-03.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Mozilla and Servo Contributors. 2012. Servo Project. https://servo.org Servo Project. SmartOracle-An Agentic Approach to Mitigate Noise in Differential Oracles 25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "{Joanna Isabelle} Olszewska. 2024. Guide to the Software Engineering Body of Knowledge v4.0. Vol. 4.0. IEEE Computer Society, United States.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Oracle. 2018. GraalVM Documentation. https://www.graalvm.org GraalVM.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Oracle and GraalVM contributors. [n. d.]. GraalJS: JavaScript Engine in GraalVM. https://www.graalvm.org/latest/reference-manual/js/, accessed 2025-11-03.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Jihyeok Park, Seungmin An, Dongjun Youn, Gyeongwon Kim, and Sukyoung Ryu. 2021. JEST: N+1-Version Differential Testing of Both JavaScript Engines and Specification. In Engineering (ICSE). 13–24. doi:10.1109/ICSE43902.2021.00015",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Soyeon Park, Wen Xu, Insu Yun, Daehee Jang, and Taesoo Kim. 2020. Fuzzing javascript engines with aspect-preserving mutation. In. IEEE, 1629–1642.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Wenlong Pei, Yilin Huang, Xiang Chen, Guilong Lu, Yong Liu, and Chao Ni. 2025. Semi-supervised software vulnerability assessment via code lexical and structural information fusion. Automated Software Engineering 32, 2 (2025), 57.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Model Context Protocol. 2025. What is the Model Context Protocol (MCP)? Online. Available: https://modelcontextprotocol.io/docs/getting-started/intro.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Chen Qian, Wei Liu, Hongzhang Liu, Nuo Chen, Yufan Dang, Jiahao Li, Cheng Yang, Weize Chen, Yusheng Su, Xin Cong, et al .2023. Chatdev: Communicative agents for software development. arXiv preprint arXiv:2307.07924 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "KAREN SPARCK JONES. 1972. ASTATISTICAL INTERPRETATION OFTERM SPECIFICITY ANDITS APPLICATION INRETRIEVAL. Journal of Documentation 28, 1 (01 1972), 11–21. arXiv:https://www.emerald.com/jd/articlepdf/28/1/11/1336479/eb026526.pdf doi:10.1108/eb026526",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Karthik Sreedhar and Lydia Chilton. 2024. Simulating human strategic behavior: Comparing single and multi-agent llms. arXiv preprint arXiv:2402.08189 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Electron Team. 2013. Electron Documentation. https://www.electronjs.org Electron Project.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Huy Tu and Tim Menzies. 2021. FRUGAL: Unlocking semi-supervised learning for software analytics. In 2021 36th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 394–406.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Liam Wachter, Julian Gremminger, Christian Wressnegger, Mathias Payer, and Flavio Toffalini. 2025. DUMPLING: Fine-grained Differential JavaScript Engine Fuzzing. In NDSS.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Junjie Wang, Zhiyi Zhang, Shuang Liu, Xiaoning Du, and Junjie Chen. 2023. {FuzzJIT}:{Oracle-Enhanced}fuzzing for{JavaScript}engine{JIT}compiler. In 32nd USENIX Security Symposium (USENIX Security 23). 1865–1882.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Yanlin Wang, Wanjun Zhong, Yanxian Huang, Ensheng Shi, Min Yang, Jiachi Chen, Hui Li, Yuchi Ma, Qianxiang Wang, and Zibin Zheng. 2025. Agents in software engineering: Survey, landscape, and vision. Automated Software Engineering 32, 2 (2025), 1–36.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Zefan Wang, Zichuan Liu, Yingying Zhang, Aoxiao Zhong, Jihong Wang, Fengbin Yin, Lunting Fan, Lingfei Wu, and Qingsong Wen. 2024. Rcagent: Cloud root cause analysis by autonomous agents with tool-augmented large language models. In Proceedings of the 33rd ACM International Conference on Information and Knowledge Management. 4966–4974.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Zhitao Wang, Wei Wang, Zirao Li, Long Wang, Can Yi, Xinjie Xu, Luyang Cao, Hanjing Su, Shouzhi Chen, and Jun Zhou. 2024. Xuat-copilot: Multi-agent collaborative system for automated user acceptance testing with large language model. arXiv preprint arXiv:2401.02705 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Chunqiu Steven Xia, Matteo Paltenghi, Jia Le Tian, Michael Pradel, and Lingming Zhang. 2024. Fuzz4all: Universal fuzzing with large language models. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Yuanmin Xie, Zhenyang Xu, Yongqiang Tian, Min Zhou, Xintong Zhou, and Chengnian Sun. 2025. Kitten: A Simple Yet Effective Baseline for Evaluating LLM-Based Compiler Testing Techniques. In Proceedings of the 34th ACMSIGSOFT International Symposium on Software Testing and Analysis. 21–25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Hanxiang Xu, Wei Ma, Ting Zhou, Yanjie Zhao, Kai Chen, Qiang Hu, Yang Liu, and Haoyu Wang. 2025. CKGFuzzer: LLM-Based Fuzz Driver Generation Enhanced By Code Knowledge Graph. In 2025 IEEE/ACM 47th International Conference on Software Engineering: Companion Proceedings (ICSE-Companion). IEEE, 243–254.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Qinghua Xu, Guancheng Wang, Lionel Briand, and Kui Liu. 2025. Hallucination to Consensus: Multi-Agent LLMs for End-to-End Test Generation. arXiv:2506.02943 [cs.SE] https://arxiv.org/abs/2506.02943",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In Proceedings of the 32nd ACMSIGPLAN conference on Programming language design and implementation. 283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Yupeng Yang, Yongheng Chen, Rui Zhong, Jizhou Chen, and Wenke Lee. 2024. Towards generic database management system fuzzing. In 33rd USENIX Security Symposium (USENIX Security 24). 901–918.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Guixin Ye, Zhanyong Tang, Shin Hwei Tan, Songfang Huang, Dingyi Fang, Xiaoyang Sun, Lizhong Bian, Haibo Wang, and Zheng Wang. 2021. Automated conformance testing for JavaScript engines via deep compiler fuzzing. In Proceedings of the 42nd ACMSIGPLAN international conference on programming language design and implementation. 26 Srinivasan et al. 435–450.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Zhi-Wu Zhang, Xiao-Yuan Jing, and Tie-Jian Wang. 2017. Label propagation based semi-supervised learning for software defect prediction. Automated Software Engg. 24, 1 (March 2017), 47–69. doi:10.1007/s10515-016-0194-x",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Shiyao Zhou, Muhui Jiang, Weimin Chen, Hao Zhou, Haoyu Wang, and Xiapu Luo. 2023. WADIFF: A Differential Testing Framework for WebAssembly Runtimes. In 2023 38th IEEE/ACM International Conference on Automated Software Engineering (ASE). 939–950. doi:10.1109/ASE56229.2023.00188",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Xiaogang Zhu, Sheng Wen, Seyit Camtepe, and Yang Xiang. 2022. Fuzzing: A Survey for Roadmap. ACM Comput. Surv. 54, 11s, Article 230 (Sept. 2022), 36 pages. doi:10.1145/3512345 Received 20 February 2007; revised 12 March 2009; accepted 5 June 2009",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 1,
+      "text": "2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, Banff, Alberta. https://www.usenix.org/conference/osdi20/ presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 1]",
+      "technique": "pqs",
+      "sentence": "Typical applications include fuzzing C compilers [ 12], WebAssembly frameworks [ 65], database systems [ 1], cryptography libraries [ 29], and web browsers, including JavaScript engines [ 6,32,52].",
+      "context_before": "findings and establish a rigorous, low-cost methodology for verifying such automation. 2.2 Differential Fuzzing Differential fuzzing is a software testing methodology in which multiple versions or independent implementations of a system are subjected to identical, randomly-generated inputs to uncover inconsistencies in their output behavior [ 35]. This method is particularly well-suited for systems that require conformity to a shared specification, such as programming language runtimes, compiler toolchains, or complex protocols, where behavioral equivalence is expected under all valid inputs.",
+      "context_after": "The differential fuzzing workflow comprises three stages: input generation, parallel execution on different targets, and the triage of observed discrepancies. The critical challenge resides in the final stage, which involves determining whether a discovered behavioral difference is an actual bug, an artifact of undefined behavior, or a benign difference permitted by a specification gap. Central to this step is the “differential oracle,” the logic responsible for deciding which divergences merit further investigation. Most contemporary fuzzing setups employ naive oracles that flag any differen",
+      "section": "2.2 Differential Fuzzing",
+      "page": 4,
+      "char_offset": 12966,
+      "cited_reference": {
+        "number": 1,
+        "text": "2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, Banff, Alberta. https://www.usenix.org/conference/osdi20/ presentation/rigger",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "SmartOracle addresses the cost of validating differential fuzzing results for JavaScript engines, where oracles are built by hand, are expensive and false-positive prone, and must be redone as the specification evolves. It decomposes the triage workflow into specialised LLM sub-agents that gather evidence from terminal runs and specification queries. On historical benchmarks it reaches 0.84 recall at an 18% false positive rate.",
+    "narrative": "One citation, listing database systems among the typical applications of differential fuzzing.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2602_19490.json b/_data/papers/paper_arxiv_2602_19490.json
new file mode 100644
index 0000000..c7f172c
--- /dev/null
+++ b/_data/papers/paper_arxiv_2602_19490.json
@@ -0,0 +1,867 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:23:13Z",
+  "paper": {
+    "id": "paper:arxiv:2602.19490",
+    "title": "FuzzySQL: Uncovering Hidden Vulnerabilities in DBMS Special Features with LLM-Driven Fuzzing",
+    "authors": [
+      "Yongxin Chen",
+      "Zhiyuan Jiang",
+      "Chao Zhang",
+      "Haoran Xu",
+      "Shenglin Xu",
+      "Jianping Tang",
+      "Zheming Li",
+      "Peidai Xie",
+      "Yongjun Wang"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2602.19490",
+    "s2_paper_id": "aad1150ad6b87608cbd6c02f7792a28e785746ff",
+    "url": "https://arxiv.org/abs/2602.19490",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2602.19490",
+      "retrieved_at": "2026-09-10T15:23:13Z",
+      "chars": 97646,
+      "content_sha256": "sha256:bb918cc68b3f54bb7b3da49b8ba65aaaab309bfda7f03fa27e85a0ad56f2bb7a"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 29,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 4134
+      },
+      {
+        "number": "2",
+        "title": "Motivation",
+        "start": 10198
+      },
+      {
+        "number": "2.1",
+        "title": "Under-Tested Features in Modern DBMSs",
+        "start": 10783
+      },
+      {
+        "number": "2.2",
+        "title": "Why Existing Fuzzers Miss These Bugs",
+        "start": 13147
+      },
+      {
+        "number": "2.3",
+        "title": "LLM: Opportunities and Challenges",
+        "start": 14579
+      },
+      {
+        "number": "2.4",
+        "title": "Goals of FuzzySQL",
+        "start": 16692
+      },
+      {
+        "number": "3",
+        "title": "Methodology",
+        "start": 18269
+      },
+      {
+        "number": "3.1",
+        "title": "Grammar-Guided SQL Generation",
+        "start": 19787
+      },
+      {
+        "number": "3.2",
+        "title": "Logic-Shifting Progressive Mutation",
+        "start": 23443
+      },
+      {
+        "number": "3.3",
+        "title": "Automated Error Repair",
+        "start": 29180
+      },
+      {
+        "number": "3.4",
+        "title": "Replay-Guided Crash Validation",
+        "start": 39916
+      },
+      {
+        "number": "4",
+        "title": "Implemention",
+        "start": 43610
+      },
+      {
+        "number": "4.1",
+        "title": "Prompt Engineering",
+        "start": 43625
+      },
+      {
+        "number": "4.2",
+        "title": "Execution Framework and Instrumentation",
+        "start": 46149
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 49211
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Setup",
+        "start": 49224
+      },
+      {
+        "number": "5.2",
+        "title": "Bug Discovery Effectiveness",
+        "start": 51810
+      },
+      {
+        "number": "5.3",
+        "title": "Ablation Study",
+        "start": 55358
+      },
+      {
+        "number": "5.4",
+        "title": "Case Study",
+        "start": 61879
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 67880
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 71933
+      },
+      {
+        "number": "7.1",
+        "title": "DBMS Fuzzing",
+        "start": 71948
+      },
+      {
+        "number": "7.2",
+        "title": "Grammar-Based Fuzzing",
+        "start": 74471
+      },
+      {
+        "number": "7.3",
+        "title": "LLM-Assisted Fuzzing",
+        "start": 76546
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2015. SQLsmith. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Ziyad Alsaeed and Michal Young. 2023. Finding short slow inputs faster with grammar-based search. InProceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis. 1068–1079., Vol. 1, No. 1, Article. Publication date: March 2018. 24 Chen et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Cornelius Aschermann, Tommaso Frassetto, Thorsten Holz, Patrick Jauernig, Ahmad-Reza Sadeghi, and Daniel Teuchert. 2019. NAUTILUS: Fishing for deep bugs with grammars.. InNDSS.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Tim Blazytko, Cornelius Aschermann, Moritz Schlögel, Ali Abbasi, Sergej Schumilo, Simon Wörner, and Thorsten Holz. 2019.{GRIMOIRE}: Synthesizing structure while fuzzing. In28th USENIX Security Symposium (USENIX Security 19). 1985–2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Yongheng Chen, Rui Zhong, Hong Hu, Hangfan Zhang, Yupeng Yang, Dinghao Wu, and Wenke Lee. 2021. One engine to fuzz’em all: Generic language processor testing with semantic validation. In Privacy (SP). IEEE, 642–658.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Haoran Peng, Chenyuan Yang, and Lingming Zhang. 2023. Large language models are zero-shot fuzzers: Fuzzing deep-learning libraries via large language models. InProceedings of the 32nd ACMSIGSOFT international symposium on software testing and analysis. 423–435.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Chenyuan Yang, Shizhuo Dylan Zhang, Shujing Yang, and Lingming Zhang. 2024. Large language models are edge-case generators: Crafting unusual programs for fuzzing deep learning libraries. In Proceedings of the 46th IEEE/ACM international conference on software engineering. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Jueon Eom, Seyeon Jeong, and Taekyoung Kwon. 2024. Fuzzing JavaScript Interpreters with Coverage-Guided Reinforcement Learning for LLM-Based Mutation. InProceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. 1656–1668.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Karine Even-Mendoza, Arindam Sharma, Alastair F Donaldson, and Cristian Cadar. 2023. Grayc: Greybox fuzzing of compilers and analysers for c. InProceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis. 1219–1231.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, and Yu Jiang. 2024. Sedar: Obtaining high-quality seeds for dbms fuzzing via cross-dbms sql transfer. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free DBMS fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Patrice Godefroid, Adam Kiezun, and Michael Y Levin. 2008. Grammar-based whitebox fuzzing. InProceedings of the 29th ACMSIGPLAN conference on programming language design and implementation. 206–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Rahul Gopinath, Hamed Nemati, and Andreas Zeller. 2021. Input algebras. In2021 IEEE/ACM 43rd International Conference on Software Engineering (ICSE). IEEE, 699–710.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Nikolas Havrikov and Andreas Zeller. 2019. Systematically covering input structure. In2019 34th IEEE/ACM international conference on automated software engineering (ASE). IEEE, 189–199.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Christian Holler, Kim Herzig, and Andreas Zeller. 2012. Fuzzing with code fragments. In21st USENIX Security Symposium (USENIX Security 12). 445–458.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. {DynSQL}: Stateful Fuzzing for Database Management Systems with Complex and Valid {SQL}Query Generation. In32nd USENIX Security Symposium (USENIX Security 23). 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting logic bugs in database engines via equivalent expression transformation. In18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24). 821–835.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. Apollo: Automatic detection and diagnosis of performance regressions in database systems.Proceedings of the VLDB Endowment13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Caroline Lemieux, Jeevana Priya Inala, Shuvendu K Lahiri, and Siddhartha Sen. 2023. Codamosa: Escaping coverage plateaus in test generation with pre-trained large language models. In Software Engineering (ICSE). IEEE, 919–931.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-oriented DBMS fuzzing. In Engineering (ICDE). IEEE, 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting logical bugs of {DBMS}with coverage-based guidance. In31st USENIX Security Symposium (USENIX Security 22). 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Ben Limpanukorn, Jiyuan Wang, Hong Jin Kang, Zitong Zhou, and Miryung Kim. 2024. Fuzzing MLIR Compilers with Custom Mutation Synthesis. In. IEEE Computer Society, 457–468.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Hangtian Liu, Lei Zheng, Shuitao Gan, Chao Zhang, Zicong Gao, Hongqi Zhang, Yishun Zeng, Zhiyuan Jiang, and Jiahai Yang. 2025. EAGLEYE: Exposing Hidden Web Interfaces in IoT Devices via Routing Analysis. InProceedings of the 32st Annual Network and Distributed System Security Symposium (NDSS)., Vol. 1, No. 1, Article. Publication date: March 2018. FuzzySQL: Uncovering Hidden Vulnerabilities in DBM",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Jiawei Liu, Jinkun Lin, Fabian Ruffy, Cheng Tan, Jinyang Li, Aurojit Panda, and Lingming Zhang. 2023. Nnsmith: Generating diverse and valid test cases for deep learning compilers. InProceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2. 530–543.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Xinyue Liu, Xiangnan Kong, Lei Liu, and Kuorong Chiang. 2018. TreeGAN: syntax-aware sequence generation with generative adversarial networks. In. IEEE, 1140–1145.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. InProceedings of the 44th International Conference on Software Engineering. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Zhe Liu, Chunyang Chen, Junjie Wang, Mengzhuo Chen, Boyu Wu, Zhilin Tian, Yuekai Huang, Jun Hu, and Qing Wang. 2024. Testing the limits: Unusual text inputs generation for mobile app crash detection with large language model. InProceedings of the IEEE/ACM 46th international conference on software engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Yunlong Lyu, Yuxuan Xie, Peng Chen, and Hao Chen. 2024. Prompt Fuzzing for Fuzz Driver Generation. InProceedings of the 2024 on ACMSIGSAC Conference on Computer and Communications Security. 3793–3807.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Xiaoyue Ma, Lannan Luo, and Qiang Zeng. 2024. From One Thousand Pages of Specification to Unveiling Hidden Bugs: Large Language Model Assisted Fuzzing of Matter IoT Devices. In33rd USENIX Security Symposium (USENIX Security 24). 4783–4800.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Ruijie Meng, Martin Mirchev, Marcel Böhme, and Abhik Roychoudhury. 2024. Large language model guided protocol fuzzing. InProceedings of the 31st Annual Network and Distributed System Security Symposium (NDSS), Vol. 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Yaroslav Oliinyk, Michael Scott, Ryan Tsang, Chongzhou Fang, Houman Homayoun, et al .2024. Fuzzing BusyBox: Leveraging LLM and Crash Reuse for Embedded Bug Unearthing. In33rd USENIX Security Symposium (USENIX Security 24). 883–900.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Xianfei Ou, Cong Li, Yanyan Jiang, and Chang Xu. 2024. The mutators reloaded: Fuzzing compilers with large language model generated mutation operators. InProceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 4. 298–312.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA, Article 211 (2020), 30 pages. doi:10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Prashast Srivastava and Mathias Payer. 2021. Gramatron: Effective grammar-aware fuzzing. InProceedings of the 30th acm sigsoft international symposium on software testing and analysis. 244–256.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Chenyao Suo, Junjie Chen, Shuang Liu, Jiajun Jiang, Yingquan Zhao, and Jianrong Wang. 2024. Fuzzing MLIR Compiler Infrastructure via Operation Dependency Analysis. InProceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. 1287–1299.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Dawei Wang, Geng Zhou, Li Chen, Dan Li, and Yukai Miao. 2024. ProphetFuzz: Fully Automated Prediction and Fuzzing of High-Risk Option Combinations with Only Documentation via Large Language Model. InProceedings of the 2024 on ACMSIGSAC Conference on Computer and Communications Security. 735–749.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Haoyu Wang, Junjie Chen, Chuyue Xie, Shuang Liu, Zan Wang, Qingchao Shen, and Yingquan Zhao. 2023. Mlirsmith: Random program generation for fuzzing mlir compiler infrastructure. In2023 38th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 1555–1566.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. 2017. Skyfire: Data-driven seed generation for fuzzing. In2017 IEEE Symposium on Security and Privacy (SP). 579–594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. 2019. Superion: Grammar-aware greybox fuzzing. In2019 IEEE/ACM 41st International Conference on Software Engineering (ICSE). IEEE, 724–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Jincheng Wang, Le Yu, and Xiapu Luo. 2024. Llmif: Augmented large language model for fuzzing iot devices. In. IEEE, 881–896.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry practice of coverage-guided enterprise-level DBMS fuzzing. In Engineering: Software Engineering in Practice (ICSE-SEIP). IEEE, 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Chunqiu Steven Xia, Matteo Paltenghi, Jia Le Tian, Michael Pradel, and Lingming Zhang. 2024. Fuzz4all: Universal fuzzing with large language models. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Haoran Xu, Zhiyuan Jiang, Yongjun Wang, Shuhui Fan, Shenglin Xu, Peidai Xie, Shaojing Fu, and Mathias Payer. 2024. Fuzzing JavaScript Engines with a Graph-based IR. InProceedings of the 2024 on ACMSIGSAC Conference on Computer and Communications Security. 3734–3748., Vol. 1, No. 1, Article. Publication date: March 2018. 26 Chen et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Wen Xu, Soyeon Park, and Taesoo Kim. 2020. Freedom: Engineering a state-of-the-art dom fuzzer. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 971–986.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Chenyuan Yang, Zijie Zhao, and Lingming Zhang. 2025. Kernelgpt: Enhanced kernel fuzzing via large language models. InProceedings of the 30th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2. 560–573.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In Proceedings of the 32nd ACMSIGPLAN conference on Programming language design and implementation. 283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Yupeng Yang, Yongheng Chen, Rui Zhong, Jizhou Chen, and Wenke Lee. 2024. Towards generic database management system fuzzing. In33rd USENIX Security Symposium (USENIX Security 24). 901–918.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing.Proceedings of the ACM on Management of Data3, 1 (2025), 1–24.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 53,
+      "text": "Lixi Zhang, Chengliang Chai, Xuanhe Zhou, and Guoliang Li. 2022. Learnedsqlgen: Constraint-aware sql generation using reinforcement learning. InProceedings of the 2022 International Conference on Management of Data. 945–958.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Qifan Zhang, Xuesong Bai, Xiang Li, Haixin Duan, Qi Li, and Zhou Li. 2024. ResolverFuzz: Automated Discovery of DNS Resolver Vulnerabilities with Query-Response Fuzzing. In33rd USENIX Security Symposium (USENIX Security 24). 4729–4746.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Qiang Zhang, Yuheng Shen, Jianzhong Liu, Yiru Xu, Heyuan Shi, Yu Jiang, and Wanli Chang. 2024. ECG: Augmenting Embedded Operating System Fuzzing via LLM-Based Corpus Generation.IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems43, 11 (2024), 4238–4249.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing database management systems with language validity and coverage feedback. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments. arXiv:2505.02012 [cs.SE]",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 58,
+      "text": "Chijin Zhou, Quan Zhang, Lihua Guo, Mingzhe Wang, Yu Jiang, Qing Liao, Zhiyong Wu, Shanshan Li, and Bin Gu. 2023. Towards better semantics exploration for browser fuzzing.Proceedings of the ACM on Programming Languages7, OOPSLA2 (2023), 604–631.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Zhuotong Zhou, Yongzhuo Yang, Susheng Wu, Yiheng Huang, Bihuan Chen, and Xin Peng. 2024. Magneto: A Step-Wise Approach to Exploit Vulnerabilities in Dependent Libraries via LLM-Empowered Directed Fuzzing. InProceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering. 1633–1644. A Prompts Used In FuzzySQL This section documents the exact prompts used in FuzzySQL for ",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    },
+    {
+      "number": 35,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA, Article 211 (2020), 30 pages. doi:10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 52,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing.Proceedings of the ACM on Management of Data3, 1 (2025), 1–24.",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 57,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments. arXiv:2505.02012 [cs.SE]",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "prints the arXiv id of the paper introducing shqvel"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 1,19,23,37,56]",
+      "technique": "pqs",
+      "sentence": "However, most existing DBMS fuzzers [ 1,19,23,37,56] predominantly focus on generalpurpose statement structures (e.",
+      "context_before": "InProceedings of Make sure to enter the correct conference title from your rights confirmation email (Conference acronym ’XX).ACM, New York, NY, USA, 29 pages. https://doi.org/XXXXXXX.XXXXXXX 1 Introduction Modern database management systems (DBMSs) sit at the core of software infrastructure that is critical to security and reliability, where subtle logic flaws can lead to persistent crashes, state corruption, and severe service disruption. As a result, fuzzing has emerged as an effective technique for uncovering DBMS bugs by generating SQL statements and executing them against target systems.",
+      "context_after": "g., SELECT ), covering only a limited portion of the DBMS grammar. Recent efforts such as BUZZBEE [51] and POLYGLOT [7] emphasize cross-target generalizability, rather than systematically exercising DBMS-specific features such as read-only modes or GTID control. DBMS special features are vendor-specific extensions that deviate from standard SQL specifications and, although invoked less frequently, are deeply embedded in core execution paths and can trigger severe system failures under edge conditions. For example, Listing 1 shows that a seemingly benign combination of statements, such as UPDAT",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4542,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 19,23,35]",
+      "technique": "norec",
+      "sentence": "Existing DBMS fuzzers have explored SQL generation and mutation from multiple perspectives, including structural complexity [ 18,56], sequence-level interactions [ 13,22], and oracle-guided logic testing [ 19,23,35].",
+      "context_before": "neralizability, rather than systematically exercising DBMS-specific features such as read-only modes or GTID control. DBMS special features are vendor-specific extensions that deviate from standard SQL specifications and, although invoked less frequently, are deeply embedded in core execution paths and can trigger severe system failures under edge conditions. For example, Listing 1 shows that a seemingly benign combination of statements, such as UPDATE HISTOGRAM and RESET within a read-only transaction, can lead to a full-system crash—behaviors that are rarely exercised by conventional fuzzers.",
+      "context_after": "Despite these advances, they share a common limitation: none have made effective attempts to systematically explore feature-specific logic. At a fundamental level, this limitation stems from the mismatch between feature-specific DBMS behaviors and the design assumptions of existing fuzzers, and manifests in several aspects. Concretely, we identify three key factors that hinder existing fuzzers from effectively exploring feature-specific DBMS logic. First, traditional fuzzers lack support for feature-specific SQL constructs that are often tied to internal DBMS mechanisms. These constructs are",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5403,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art opensource fuzzing baselines: Squirrel [ 56], EET [ 19], and SQLancer [ 35].",
+      "context_before": "rch 2018. FuzzySQL: Uncovering Hidden Vulnerabilities in DBMS Special Features with LLM-Driven Fuzzing 15 Table 2. Bugs discovered by FuzzySQL DBMS Found Confirmed Fixed Feature-Related MySQL 22 22 12 10 MariaDB 28 24 10 9 SQLite 2 2 2 1 ClickHouse 12 12 7 7 Total 64 60 31 27 reproducible evaluation, all reported results are based on Qwen3-30B. We used a temperature of 0.3–0.5 to balance stability and diversity. In pilot tests, a temperature of 0.4 produced natural yet varied outputs—sufficiently diverse for fuzzing under structured constraints. All experiments used an 8K-token context window.",
+      "context_after": "Among them, Squirrel represents mutation-based fuzzing, while both EET and SQLancer adopt generation-based approaches. Since EET and SQLancer are not inherently a grey-box fuzzing tool, we first collect their generated test cases and then utilize FuzzySQL’s replay mechanism to measure its coverage. For SQLancer, we use TLP [ 36] as the test oracle for MySQL, and NoREC [ 35] for the other DBMS targets. This choice is constrained by the design of SQLancer, which employs different oracle-based testing strategies tailored to specific DBMSs. All baselines were evaluated under identical runtime con",
+      "section": "5.1 Experimental Setup",
+      "page": 15,
+      "char_offset": 50639,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Among them, Squirrel represents mutation-based fuzzing, while both EET and SQLancer adopt generation-based approaches.",
+      "context_before": "Fixed Feature-Related MySQL 22 22 12 10 MariaDB 28 24 10 9 SQLite 2 2 2 1 ClickHouse 12 12 7 7 Total 64 60 31 27 reproducible evaluation, all reported results are based on Qwen3-30B. We used a temperature of 0.3–0.5 to balance stability and diversity. In pilot tests, a temperature of 0.4 produced natural yet varied outputs—sufficiently diverse for fuzzing under structured constraints. All experiments used an 8K-token context window. To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art opensource fuzzing baselines: Squirrel [ 56], EET [ 19], and SQLancer [ 35].",
+      "context_after": "Since EET and SQLancer are not inherently a grey-box fuzzing tool, we first collect their generated test cases and then utilize FuzzySQL’s replay mechanism to measure its coverage. For SQLancer, we use TLP [ 36] as the test oracle for MySQL, and NoREC [ 35] for the other DBMS targets. This choice is constrained by the design of SQLancer, which employs different oracle-based testing strategies tailored to specific DBMSs. All baselines were evaluated under identical runtime conditions and DBMS versions. We adopted official configurations and standardized crash triage procedures to ensure fair a",
+      "section": "5.1 Experimental Setup",
+      "page": 15,
+      "char_offset": 50803,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Since EET and SQLancer are not inherently a grey-box fuzzing tool, we first collect their generated test cases and then utilize FuzzySQL’s replay mechanism to measure its coverage.",
+      "context_before": "ucible evaluation, all reported results are based on Qwen3-30B. We used a temperature of 0.3–0.5 to balance stability and diversity. In pilot tests, a temperature of 0.4 produced natural yet varied outputs—sufficiently diverse for fuzzing under structured constraints. All experiments used an 8K-token context window. To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art opensource fuzzing baselines: Squirrel [ 56], EET [ 19], and SQLancer [ 35]. Among them, Squirrel represents mutation-based fuzzing, while both EET and SQLancer adopt generation-based approaches.",
+      "context_after": "For SQLancer, we use TLP [ 36] as the test oracle for MySQL, and NoREC [ 35] for the other DBMS targets. This choice is constrained by the design of SQLancer, which employs different oracle-based testing strategies tailored to specific DBMSs. All baselines were evaluated under identical runtime conditions and DBMS versions. We adopted official configurations and standardized crash triage procedures to ensure fair and reproducible comparisons. Except for Squirrel, a mutation-based fuzzer using its default seeds, all baselines ran without a seed corpus. Each experiment was repeated 5 times on i",
+      "section": "5.1 Experimental Setup",
+      "page": 15,
+      "char_offset": 50922,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For SQLancer, we use TLP [ 36] as the test oracle for MySQL, and NoREC [ 35] for the other DBMS targets.",
+      "context_before": "tural yet varied outputs—sufficiently diverse for fuzzing under structured constraints. All experiments used an 8K-token context window. To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art opensource fuzzing baselines: Squirrel [ 56], EET [ 19], and SQLancer [ 35]. Among them, Squirrel represents mutation-based fuzzing, while both EET and SQLancer adopt generation-based approaches. Since EET and SQLancer are not inherently a grey-box fuzzing tool, we first collect their generated test cases and then utilize FuzzySQL’s replay mechanism to measure its coverage.",
+      "context_after": "This choice is constrained by the design of SQLancer, which employs different oracle-based testing strategies tailored to specific DBMSs. All baselines were evaluated under identical runtime conditions and DBMS versions. We adopted official configurations and standardized crash triage procedures to ensure fair and reproducible comparisons. Except for Squirrel, a mutation-based fuzzer using its default seeds, all baselines ran without a seed corpus. Each experiment was repeated 5 times on identical hardware, and the reported results are the averages, which show consistent statistical stability",
+      "section": "5.1 Experimental Setup",
+      "page": 15,
+      "char_offset": 51103,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This choice is constrained by the design of SQLancer, which employs different oracle-based testing strategies tailored to specific DBMSs.",
+      "context_before": "sed an 8K-token context window. To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art opensource fuzzing baselines: Squirrel [ 56], EET [ 19], and SQLancer [ 35]. Among them, Squirrel represents mutation-based fuzzing, while both EET and SQLancer adopt generation-based approaches. Since EET and SQLancer are not inherently a grey-box fuzzing tool, we first collect their generated test cases and then utilize FuzzySQL’s replay mechanism to measure its coverage. For SQLancer, we use TLP [ 36] as the test oracle for MySQL, and NoREC [ 35] for the other DBMS targets.",
+      "context_after": "All baselines were evaluated under identical runtime conditions and DBMS versions. We adopted official configurations and standardized crash triage procedures to ensure fair and reproducible comparisons. Except for Squirrel, a mutation-based fuzzer using its default seeds, all baselines ran without a seed corpus. Each experiment was repeated 5 times on identical hardware, and the reported results are the averages, which show consistent statistical stability. 5.2 Bug Discovery Effectiveness To evaluate the practical effectiveness of FuzzySQL, we conducted a systematic fuzzing campaign on five",
+      "section": "5.1 Experimental Setup",
+      "page": 15,
+      "char_offset": 51208,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Deduplicated MySQL bugs discovered within 24 hours Target ID†FR Type FuzzySQL Squirrel EET SQLancer FuzzySQL!",
+      "context_before": "e bugs or reaching higher coverage, but also in exposing a broader and qualitatively different set of vulnerabilities. In particular, feature-related bugs are not isolated anecdotes in our results; with 27 confirmed cases, they constitute a substantial portion of the bugs uncovered by FuzzySQL. This capability arises from FuzzySQL’s integrated design, which jointly addresses three obstacles in feature-related DBMS bug discovery: insufficient coverage of feature-specific constructs, unmet semantic preconditions, and hidden state interactions that span multiple statements or executions. Table 3.",
+      "context_after": "r FuzzySQL!r !m MySQL 3◦Use After Free✔ ✗ ✔ ✗ ✗ ✗ MySQL 6◦Assertion Failure✔ ✗ ✗ ✗ ✔ ✗ MySQL 8•Assertion Failure✔ ✗ ✗ ✗ ✔ ✗ MySQL 9◦NULL Ptr Deref✔ ✗ ✔ ✗ ✔ ✔ MySQL 17•Assertion Failure✔ ✗ ✗ ✗ ✗ ✗ MySQL 21◦NULL Ptr Deref✔ ✗ ✔ ✗ ✔ ✗ †Bug ID corresponds to Table 4 (Appendix B). FR: •feature-related,◦not feature-related;!r: repair disabled;!m: mutation disabled. 5.3 Ablation Study We perform an ablation study to assess the impact of three core design choices in FuzzySQL: logic-shifting progressive mutation, automated semantic repair, and the choice of LLM for semantic instantiation. Our analysis a",
+      "section": "5.2 Bug Discovery Effectiveness",
+      "page": 16,
+      "char_offset": 54887,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker",
+      "surface": "[ 4,19,23,35–37,52]",
+      "technique": "qpg",
+      "sentence": "Depending on the bug types targeted—ranging from crash vulnerabilities [ 1,12,13,18, 22,27,45,56] to logic inconsistencies [ 4,19,23,35–37,52] and performance regressions [ 5,20,28, 53]—existing DBMS fuzzers follow either generation-based or mutation-based strategies.",
+      "context_before": "se challenges largely reflect the limitations of current LLMs. As LLMs improve in reasoning depth and code synthesis capability, their effectiveness in fuzzing high-complexity structures will improve correspondingly. FuzzySQL excels at triggering feature-related failures and hard-to-reach logic branches, aspects that are often invisible to basic coverage models. 7 Related Work 7.1 DBMS Fuzzing Fuzzing has emerged as a critical technique for testing database management systems (DBMSs) by directly targeting components such as parsers, optimizers, and execution engines through crafted SQL inputs.",
+      "context_after": "Generation-based approaches rely on hand-crafted grammars or learned rules to synthesize SQL sequences from scratch. SQLSmith [ 1] uses randomized AST construction to generate structurally valid SQL inputs, but often fails on semantic soundness. To enhance expressiveness, DynSQL [ 18] dynamically tracks schema evolution to guide generation, while TreeGAN [ 27] adopts grammaraware GANs to synthesize syntactically valid SQL trees. SQLancer [ 35–37,52] and EET [ 19] generate ASTs while preserving contextual variable bindings to ensure validity. However, most generators still rely heavily on manu",
+      "section": "7.1 DBMS Fuzzing",
+      "page": 22,
+      "char_offset": 72168,
+      "cited_reference": {
+        "number": 4,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg",
+        "cert"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 35–37,52] and EET [ 19] generate ASTs while preserving contextual variable bindings to ensure validity.",
+      "context_before": "logic inconsistencies [ 4,19,23,35–37,52] and performance regressions [ 5,20,28, 53]—existing DBMS fuzzers follow either generation-based or mutation-based strategies. Generation-based approaches rely on hand-crafted grammars or learned rules to synthesize SQL sequences from scratch. SQLSmith [ 1] uses randomized AST construction to generate structurally valid SQL inputs, but often fails on semantic soundness. To enhance expressiveness, DynSQL [ 18] dynamically tracks schema evolution to guide generation, while TreeGAN [ 27] adopts grammaraware GANs to synthesize syntactically valid SQL trees.",
+      "context_after": "However, most generators still rely heavily on manual rule engineering or specifications. The concurrent work ShQveL [57] uses LLM to construct fragments and then uses these fragments to combine SQL statements. This test case generation method still relies on the implementation of general syntax rules. Mutation-based fuzzers, in contrast, transform existing SQL statements through changes at the syntax or intermediate representation (IR) level. Some prior work also explores context-dependent mutations arising from the composition of SQL statement sequences. Squirrel [ 56] represents SQL as int",
+      "section": "7.1 DBMS Fuzzing",
+      "page": 22,
+      "char_offset": 72870,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "ShQveL",
+      "technique": null,
+      "sentence": "The concurrent work ShQveL [57] uses LLM to construct fragments and then uses these fragments to combine SQL statements.",
+      "context_before": "and-crafted grammars or learned rules to synthesize SQL sequences from scratch. SQLSmith [ 1] uses randomized AST construction to generate structurally valid SQL inputs, but often fails on semantic soundness. To enhance expressiveness, DynSQL [ 18] dynamically tracks schema evolution to guide generation, while TreeGAN [ 27] adopts grammaraware GANs to synthesize syntactically valid SQL trees. SQLancer [ 35–37,52] and EET [ 19] generate ASTs while preserving contextual variable bindings to ensure validity. However, most generators still rely heavily on manual rule engineering or specifications.",
+      "context_after": "This test case generation method still relies on the implementation of general syntax rules. Mutation-based fuzzers, in contrast, transform existing SQL statements through changes at the syntax or intermediate representation (IR) level. Some prior work also explores context-dependent mutations arising from the composition of SQL statement sequences. Squirrel [ 56] represents SQL as intermediate representation (IR) and applies typed, data-aware mutation. RATEL [ 45] adds finegrained coverage tracking and feedback-guided deduplication for robust crash reporting. Griffin [ 13] shuffles SQL state",
+      "section": "7.1 DBMS Fuzzing",
+      "page": 22,
+      "char_offset": 73075,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M3"
+      ],
+      "describes_as_state_of_the_art": [
+        "M3"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:00:23Z",
+    "is_model_written": true,
+    "summary": "FuzzySQL targets the special features of individual DBMSs -- vendor-specific syntax and subsystems that general-purpose generators do not model -- using a language model to produce test cases for them. Its argument is that existing fuzzers concentrate on general statement structures, so features unique to one system go untested. It is evaluated on MySQL and other systems against mutation-based and generation-based baselines.",
+    "narrative": "SQLancer is one of three state-of-the-art open-source baselines, alongside Squirrel and EET. The comparison required accommodation in both directions: because SQLancer is not a grey-box tool, the authors collected its generated test cases and replayed them through FuzzySQL's own mechanism to measure coverage, and because SQLancer pairs different oracles with different systems they ran TLP for MySQL and NoREC for the other targets. That constraint is noted as a property of SQLancer's design rather than a shortcoming. Bug counts are reported per target over 24 hours.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "state_of_the_art",
+      "M4": "definition",
+      "M5": "baseline",
+      "M6": "baseline",
+      "M7": "definition",
+      "M8": "result_comparison",
+      "M9": "background",
+      "M10": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer's test cases are collected and replayed to measure its coverage on equal terms, which is part of running it as a baseline rather than reuse in FuzzySQL's own generation."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer oracle is extended; FuzzySQL generates test cases for vendor-specific features with a language model."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M3",
+          "M5",
+          "M6",
+          "M8"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art opensource fuzzing baselines: Squirrel [ 56], EET [ 19], and SQLancer [ 35].",
+            "section": "5.1 Experimental Setup",
+            "page": 15
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "Since EET and SQLancer are not inherently a grey-box fuzzing tool, we first collect their generated test cases and then utilize FuzzySQL’s replay mechanism to measure its coverage.",
+            "section": "5.1 Experimental Setup",
+            "page": 15
+          },
+          {
+            "mention_id": "M6",
+            "sentence": "For SQLancer, we use TLP [ 36] as the test oracle for MySQL, and NoREC [ 35] for the other DBMS targets.",
+            "section": "5.1 Experimental Setup",
+            "page": 15
+          },
+          {
+            "mention_id": "M8",
+            "sentence": "Deduplicated MySQL bugs discovered within 24 hours Target ID†FR Type FuzzySQL Squirrel EET SQLancer FuzzySQL!",
+            "section": "5.2 Bug Discovery Effectiveness",
+            "page": 16
+          }
+        ],
+        "reasoning": "M3 names SQLancer among the three baselines, M5 describes how its test cases were collected for the coverage comparison, M6 records TLP being used for MySQL and NoREC for the other targets, and M8 gives the 24-hour bug table.",
+        "techniques": [
+          "tlp",
+          "norec"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M3"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "To demonstrate the advantage of FuzzySQL, we compare it against three state-of-the-art opensource fuzzing baselines: Squirrel [ 56], EET [ 19], and SQLancer [ 35].",
+            "section": "5.1 Experimental Setup",
+            "page": 15
+          }
+        ],
+        "reasoning": "M3 describes SQLancer as one of three state-of-the-art open-source fuzzing baselines."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2603_00311.json b/_data/papers/paper_arxiv_2603_00311.json
new file mode 100644
index 0000000..d1e3420
--- /dev/null
+++ b/_data/papers/paper_arxiv_2603_00311.json
@@ -0,0 +1,470 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:41Z",
+  "paper": {
+    "id": "paper:arxiv:2603.00311",
+    "title": "Towards the Systematic Testing of Regular Expression Engines",
+    "authors": [
+      "Berk Çakar",
+      "Dongyoon Lee",
+      "James C. Davis"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2603.00311",
+    "s2_paper_id": "ff9b5fa596fbff921f259f1c42d94dff78dc9a9e",
+    "url": "https://arxiv.org/abs/2603.00311",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2603.00311",
+      "retrieved_at": "2026-09-09T01:29:41Z",
+      "chars": 61159,
+      "content_sha256": "sha256:a8ac783fc19c8b240d88d969a1802435a74100e369fd5c86a8859a43a924c13f"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 10,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Muath Alkhalaf, Tevfik Bultan, and Jose L. Gallegos. 2012. Verifying client-side input validation functions using string analysis. InProceedings of the 34th International Conference on Software Engineering(ICSE ’12). IEEE Press, Zurich, Switzerland, 947–957.isbn: 9781467310673.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Cornelius Aschermann, Tommaso Frassetto, Thorsten Holz, Patrick Jauernig, Ahmad-Reza Sadeghi, and Daniel Teuchert. 2019. Nautilus: fishing for deep bugs with grammars. InProceedings of the 26th Annual Network and Distributed System Security Symposium (NDSS). Accessed: 2026-01-23. The Internet Society. doi:10.14722/ndss.2019.23412.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Efe Barlas, Xin Du, and James C. Davis. 2022. Exploiting input sanitization for regex denial of service. InProceedings of the 44th International Conference on Software Engineering(ICSE ’22).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Earl T. Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. 2015. The oracle problem in software testing: a survey.IEEE Transactions on Software Engineering, 41, 5, 507–525. doi:10.1109/TSE.2014.2372785.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Alberto Bartoli, Andrea De Lorenzo, Eric Medvet, and Fabiano Tarlao. 2016. Inference of regular expressions for text extraction from examples.IEEE Transactions on Knowledge and Data Engineering, 28, 5, 1217–1230. doi:10. 1109 /TKDE.2016.2515587.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Andrew Begel, Yit Phang Khoo, and Thomas Zimmermann. 2010. Codebook: discovering and exploiting relationships in software repositories. InProceedings of the 32nd ACM/IEEE International Conference on Software Engineering-Volume 1(ICSE ’10). Association for Computing Machinery, Cape Town, South Africa, 125–134.isbn: 9781605587196. doi:10.1145/1806799.1806821.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Martin Berglund, Brink Van Der Merwe, and Steyn Van Litsenborgh. 2021. Regular Expressions with Lookahead.JUCS-Journal of Universal Computer Science, 27, 4, 324–340. doi:10.3897/jucs.66330.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Martin Berglund, Brink Van Der Merwe, Bruce Watson, and Nicolaas Weideman. 2017. On the Semantics of Atomic Subgroups in Practical Regular Expressions. InImplementation and Application of Automata. Vol. 10329. Arnaud Carayol and Cyril Nicaud, (Eds.) Springer International Publishing, Cham, 14–26.isbn: 978-3-319-60133-5 978-3-319-60134-2. doi:10.1007/978-3-319-60134-2_2.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Fabian Beuke. 2024. GitHut 2.0 – GitHub language statistics. https://madnight .github.io/githut/. Q1 2024 data, Accessed: 2025-01-20.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Masudul Hasan Masud Bhuiyan, Berk Çakar, Ethan H. Burmane, James C. Davis, and Cristian-Alexandru Staicu. 2025. Sok: a literature and engineering review of regular expression denial of service (redos). InProceedings of the 20th ACM Asia Conference on Computer and Communications Security(ASIACCS ’25). Association for Computing Machinery, 1659–1675.isbn: 9798400714108. doi:10.1145/3708821.3733912.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Janusz A. Brzozowski. 1964. Derivatives of Regular Expressions.Journal of the Association for Computing Machinery, 11, 4, 481–494.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Berk Çakar, Charles M. Sale, Sophie Chen, Dongyoon Lee, and James C. Davis. 2025. Is reuse all you need? a systematic comparison of regular expression composition strategies. https://arxiv.org/abs/2503.20579 arXiv: 2503.20579 [cs.SE].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Cezar Câmpeanu, Kai Salomaa, and Sheng Yu. 2003. A formal study of practical regular expressions. en.International Journal of Foundations of Computer Science, 14, 06, 1007–1018. doi:10.1142/S012905410300214X.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Agnishom Chattopadhyay, Angela W. Li, and Konstantinos Mamouras. 2025. Verified and Efficient Matching of Regular Expressions with Lookaround. In Proceedings of the 14th ACMSIGPLAN International Conference on Certified Programs and Proofs(CPP ’25). Association for Computing Machinery, New York, NY, USA, 198–213.isbn: 979-8-4007-1347-7. doi:10.1145/3703595.3705884.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Tsong Yueh Chen, Shing Chi Cheung, and Shiu Ming Yiu. 1998. Metamorphic Testing: A New Approach for Generating Next Test Cases. Tech. rep. HKUSTCS98-01. Hong Kong University of Science and Technology.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Tsong Yueh Chen, Fei-Ching Kuo, Huai Liu, Pak-Lok Poon, Dave Towey, T. H. Tse, and Zhi Quan Zhou. 2018. Metamorphic testing: a review of challenges and opportunities.ACM Comput. Surv., 51, 1, Article 4, 27 pages. doi:10.1145/3 143561.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Russ Cox. 2025. RE2. https://github.com/google/re2.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "James C Davis, Daniel Moyer, Ayaan M Kazerouni, and Dongyoon Lee. 2019. Testing regex generalizability and its implications: a large-scale many-language measurement study. In2019 34th IEEE/ACM International Conference on Automated Software Engineering (ASE), 427–439. doi:10.1109/ASE.2019.00048.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "James C. Davis, Christy A. Coghlan, Francisco Servant, and Dongyoon Lee. 2018. The impact of regular expression denial of service (ReDoS) in practice: an empirical study at the ecosystem scale. InProceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(ESEC/FSE 2018). Association for Computing Machinery, New",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "James C. Davis, Louis G. Michael IV, Christy A. Coghlan, Francisco Servant, and Dongyoon Lee. 2019. Why aren’t regular expressions a lingua franca? anempirical study on the re-use and portability of regular expressions. InProceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "James Collins Davis. 2020.On the Impact and Defeat of Regular Expression Denial of Service. Ph.D. Dissertation. Virginia Tech, Blacksburg, Virginia, USA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Aryaz Eghbali and Michael Pradel. 2020. No Strings Attached: An Empirical Study of String-related Software Bugs. In2020 35th IEEE/ACM International Conference on Automated Software Engineering (ASE), 956–967.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Python Software Foundation. 2026. Re – regular expression operationspython. https://docs.python.org/3/library/re.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Free Software Foundation. 2025. Regexps — gnu emacs manual. Accessed: 202601-23. GNU Project. https://www.gnu.org/software/emacs/manual/html_node /emacs/Regexps.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Jeffrey E. F. Friedl. 2002.Mastering Regular Expressions. (2nd ed ed.). O’Reilly, Beijing; Sebastopol, CA.isbn: 978-0-596-00289-3.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Hiroya Fujinami and Ichiro Hasuo. 2024. Efficient Matching with Memoization for Regexes with Look-around and Atomic Grouping. InProgramming Languages and Systems. Vol. 14577. Stephanie Weirich, (Ed.) Springer Nature Switzerland, Cham, 90–118.isbn: 978-3-031-57266-1 978-3-031-57267-8. doi:10 .1007/978-3-031-57267-8_4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Google. 2026. UndefinedBehaviorSanitizer. https://clang.llvm.org/docs/Undefi nedBehaviorSanitizer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Google. 2026. V8: regexp class reference. https://v8docs.nodesource.com/node25.0/d8/da7/classv8_1_1_reg_exp.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Google OSS-Fuzz Project Contributors. 2026. OSS-Fuzz PCRE2 project. https://g ithub.com/google/oss-fuzz/tree/master/projects/pcre2. Accessed: 2026-01-26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Christian Holler, Kim Herzig, and Andreas Zeller. 2012. Fuzzing with code fragments. In21st USENIX Security Symposium (USENIX Security 12). USENIX Association, Bellevue, WA, 445–458.isbn: 978-931971-95-9. https://www.useni x.org/conference/usenixsecurity12/technical-sessions/presentation/holler.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Jan Goyvaerts. 2025. Regular-expressions.info — regex flavor comparison. Accessed: 2026-01-24. regular-expressions.info. https://www.regular-expressions .info/refflavors.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "S. C. Kleene. 1951. Representation of events in nerve nets and finite automata. Automata Studies, 3–41.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "K. Kosako. 2024. oniguruma. https://github.com/kkos/oniguruma/graphs/contr ibutors.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "D. Kozen. 1994. A completeness theorem for kleene algebras and the algebra of regular events.Inf. Comput., 110, 2, 366–390. doi:10.1006/inco.1994.1037.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Eric Larson and Anna Kirk. 2016. Generating Evil Test Strings for Regular Expressions. In and Validation (ICST).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. InProceedings of the 35th ACMSIGPLAN Conference on Programming Language Design and Implementation(PLDI ’14). Association for Computing Machinery, Edinburgh, United Kingdom, 216–226.isbn: 9781450327848. doi:10.1145/2594291.2594334.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Nuo Li, Tao Xie, Nikolai Tillmann, Jonathan de Halleux, and Wolfram Schulte. 2009. Reggae: automated test generation for programs using complex regular expressions. In Engineering, 515–519. doi:10.1109/ASE.2009.67.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Yeting Li, Zixuan Chen, Jialun Cao, Zhiwu Xu, Qiancheng Peng, Haiming Chen, Liyuan Chen, and Shing-Chi Cheung. 2021. ReDoSHunter: A Combined Static and Dynamic Approach for Regular Expression DoS Detection. InProceedings of the 30th USENIX Conference on Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Yeting Li, Shuaimin Li, Zhiwu Xu, Jialun Cao, Zixuan Chen, Yun Hu, Haiming Chen, and Shing-Chi Cheung. 2021. TransRegex: Multi-modal Regular Expression Synthesis by Generate-and-Repair. In2021 IEEE/ACM 43rd International Conference on Software Engineering (ICSE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Yeting Li et al. 2022. RegexScalpel: regular expression denial of service (ReDoS) defense by Localize-and-Fix. In31st USENIX Security Symposium (USENIX Security 22). USENIX Association, Boston, MA, 4183–4200.isbn: 978-1-93913331-1. https://www.usenix.org/conference/usenixsecurity22/presentation/li-y eting.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Wen-jie Lu, Zhicong Huang, Qizhi Zhang, Yuchen Wang, and Cheng Hong. 2023. Squirrel: a scalable secure Two-Party computation framework for training gradient boosting decision tree. In32nd USENIX Security Symposium (USENIX Security 23). USENIX Association, Anaheim, CA, 6435–6451.isbn: 978-1-93913337-3. https://www.usenix.org/conference/usenixsecurity23/presentation/lu.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Muhammad Numair Mansur, Maria Christakis, and Valentin Wüstholz. 2021. Metamorphic testing of datalog engines. InProceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(ESEC/FSE 2021). Association for Computing Machinery, Athens, Greece, 639–650.isbn: 9781450385626. doi:10.1145/346 8264.3468573.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Microsoft Corporation. 2025. Use regular expressions in visual studio. Accessed: 2026-01-23. Microsoft Learn. https://learn.microsoft.com/en-us/visualstudio/id e/using-regular-expressions-in-visual-studio?view=visualstudio. JAWs 2026, April 2026, Rio de Janeiro, Brazil Çakar, Lee, and Davis",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Dan Moseley, Mario Nishio, Jose Perez Rodriguez, Olli Saarikivi, Stephen Toub, Margus Veanes, Tiki Wan, and Eric Xu. 2023. Derivative based nonbacktracking real-world regex matching with backtracking semantics.Proc. ACM Program. Lang., 7, PLDI, Article 148, 24 pages. doi:10.1145/3591262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Rong Pan, Qinheping Hu, Gaowei Xu, and Loris D’Antoni. 2019. Automatic repair of regular expressions.RFixer: a tool for repairing complex regular expressions using examples, 3, OOPSLA, 139:1–139:29. doi:10.1145/3360565.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "PCRE2 Project. 2024. PCRE2 - Perl-Compatible Regular Expressions. https://w ww.pcre.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Philip Hazel. 2025. Pcre: perl compatible regular expressionsreference manual. Accessed: 2026-01-24. PCRE Project. https://pcre.org/pcre.txt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM, Virtual Event USA, 1140–1152.isbn: 978-1-4503-7043-1. doi:10.1145/3368089.3409710.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.OOPSLA 20 Artifact for \"Finding Bugs in Database Systems via Query Partitioning\", 4, OOPSLA, 211:1–211:30. doi:10.1145/3428279.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 667–682.isbn: 978-1-939133-19-9. Retrieved Sept. 21, 2025 from.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 51,
+      "text": "Olli Saarikivi, Margus Veanes, Tiki Wan, and Eric Xu. 2019. Symbolic regex matcher. en. InTools and Algorithms for the Construction and Analysis of Systems. Tomáš Vojnar and Lijun Zhang, (Eds.) Springer International Publishing, Cham, 372–378.isbn: 9783030174620. doi:10.1007/978-3-030-17462-0_24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Arto Salomaa. 1966. Two Complete Axiom Systems for the Algebra of Regular Events.Journal of the ACM, 13, 1, 158–169. doi:10.1145/321312.321326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Sergio Segura, Gordon Fraser, Ana B. Sanchez, and Antonio Ruiz-Cortés. 2016. A survey on metamorphic testing.IEEE Transactions on Software Engineering, 42, 9, 805–824. doi:10.1109/TSE.2016.2532875.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitry Vyukov. 2012. Addresssanitizer: a fast address sanity checker. InProceedings of the 2012 USENIX Conference on Annual Technical Conference(USENIX ATC’12). USENIX Association, Boston, MA, 28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Kostya Serebryany. 2015. Libfuzzer – a library for coverage-guided fuzz testing. https://llvm.org/docs/LibFuzzer.html. LLVM Project.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Evgeniy Stepanov and Konstantin Serebryany. 2015. Memorysanitizer: fast detector of uninitialized memory use in c++. In2015 IEEE/ACM International Symposium on Code Generation and Optimization (CGO), 46–55. doi:10.1109 /CGO.2015.7054186.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "The Open Group and IEEE. 2018.Regular Expressions — The Open Group Base Specifications, Issue 7, 2018 Edition. Accessed: 2026-01-24; Part of POSIX.1 (IEEE Std 1003.1-2017 / The Open Group Base Specifications). The Open Group & IEEE. https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap 09.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Ken Thompson. 1968. Programming Techniques: Regular expression search algorithm. en.Communications of the ACM, 11, 6, 419–422. Retrieved May 31, 2024 from https://dl.acm.org/doi/10.1145/363347.363387.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. 2019. Superion: grammaraware greybox fuzzing. In Software Engineering (ICSE), 724–735. doi:10.1109/ICSE.2019.00081.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Peipei Wang, Chris Brown, Jamie A. Jennings, and Kathryn T. Stolee. 2020. An Empirical Study on Regular Expression Bugs. InProceedings of the 17th International Conference on Mining Software Repositories(MSR ’20).isbn: 978-14503-7517-7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Peipei Wang, Chris Brown, Jamie A. Jennings, and Kathryn T. Stolee. 2022. Demystifying regular expression bugs: A comprehensive study on regular expression bug causes, fixes, and testing. en.Empirical Software Engineering, 27, 1, 21. doi:10.1007/s10664-021-10033-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in c compilers. InProceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation(PLDI ’11). Association for Computing Machinery, San Jose, California, USA, 283–294. isbn: 9781450306638. doi:10.1145/1993498.1993532.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Andreas Zeller, Rahul Gopinath, Marcel Böhme, Gordon Fraser, and Christian Holler. 2024. Fuzzing with grammars. InThe Fuzzing Book: Tools and Techniques for Generating Software Tests. Accessed: 2026-01-23. CISPA Helmholtz Center for Information Security. https://www.fuzzingbook.org/html/Grammars.html.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 48,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM, Virtual Event USA, 1140–1152.isbn: 978-1-4503-7043-1. doi:10.1145/3368089.3409710.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.OOPSLA 20 Artifact for \"Finding Bugs in Database Systems via Query Partitioning\", 4, OOPSLA, 211:1–211:30. doi:10.1145/3428279.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 50,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 667–682.isbn: 978-1-939133-19-9. Retrieved Sept. 21, 2025 from.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[48, 49, 50]",
+      "technique": "norec",
+      "sentence": "For DBMS, Rigger and Su [48, 49, 50] used metamorphic oracles to detect logic bugs.",
+      "context_before": "y on testing patterns, not engines. Research on test string generation [35, 37], ReDoS detection [19, 38, 40], and pattern synthesis/repair [39, 45] all assume engine correctness while validating developer-written patterns. Empirical studies of regex bugs likewise examine patternlevel faults rather than implementation defects [22, 60, 61]. More broadly, testing language implementations and DSL engines offers transferable techniques. Csmith’s [62] random program generation found 325+ compiler bugs, while grammar-based fuzzing has proven effective for language implementations (e.g.,[2, 30, 41]).",
+      "context_after": "We portray regex engine testing in practice in §4. 2.3 Metamorphic Testing Metamorphic testing (MT) addresses the oracle problem—where expected outputs are unknown, by verifying relationships between multiple executions rather than individual results [15]. Given a source input and its output, metamorphic relations (MRs) define how transformed follow-up inputs should relate to the original output. MT has proven effective for testing compilers and language implementations: Leet al.’s Equivalence Modulo Inputs (EMI) [36] detected 147 bugs in GCC and LLVM. For DSLs specifically, Mansur et al.[42]",
+      "section": null,
+      "page": 2,
+      "char_offset": 9724,
+      "cited_reference": {
+        "number": 48,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM, Virtual Event ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "ReTest is a framework for testing regular expression engines. The authors argue differential testing across implementations is unreliable because regex syntax and semantics differ between dialects, and naive byte-level fuzzing produces invalid inputs exercising only the parser. ReTest combines grammar-aware fuzzing with metamorphic testing for dialect-independent oracles, and has curated 16 metamorphic relations derived from Kleene algebra.",
+    "narrative": "One citation, noting that Rigger and Su used metamorphic oracles to detect logic bugs in DBMSs -- the precedent for using them where no reference implementation can be trusted.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2603_19434.json b/_data/papers/paper_arxiv_2603_19434.json
new file mode 100644
index 0000000..f0e6830
--- /dev/null
+++ b/_data/papers/paper_arxiv_2603_19434.json
@@ -0,0 +1,305 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:03Z",
+  "paper": {
+    "id": "paper:arxiv:2603.19434",
+    "title": "Computer-Orchestrated Design of Algorithms: From Join Specification to Implementation",
+    "authors": [
+      "Zeyuan Hu"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2603.19434",
+    "s2_paper_id": "7fa57423ffb12f9cd6dace5f26042a6289837bf4",
+    "url": "https://arxiv.org/abs/2603.19434",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2603.19434",
+      "retrieved_at": "2026-09-09T01:29:03Z",
+      "chars": 48221,
+      "content_sha256": "sha256:3620eb09c9e343949f36b892a83b9c127ebdd47e1d8d58f111efa8f0af2a14e2"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 9,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2786
+      },
+      {
+        "number": "2",
+        "title": "From Logical to Physical",
+        "start": 7927
+      },
+      {
+        "number": "2.1",
+        "title": "Preliminaries",
+        "start": 10226
+      },
+      {
+        "number": "2.2",
+        "title": "A Motivating Example",
+        "start": 12065
+      },
+      {
+        "number": "4.2",
+        "title": "Case Study 2: Refining the Theoretical",
+        "start": 24402
+      },
+      {
+        "number": "4.3",
+        "title": "Case Study 3: Architectural Evolution and",
+        "start": 26443
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Serge Abiteboul, Richard Hull, and Victor Vianu. 1995.Foundations of Databases. Vol. 8. Addison-Wesley Reading.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Koen Claessen and John Hughes. 2000. QuickCheck: a lightweight tool for random testing of Haskell programs. InProceedings of the Fifth ACMSIGPLAN International Conference on Functional Programming (ICFP ’00), Montreal, Canada, September 18-21, 2000, Martin Odersky and Philip Wadler (Eds.). ACM, 268–279. doi:10.1145/351240.351266",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Wolfgang Fischl, Georg Gottlob, Davide Mario Longo, and Reinhard Pichler. 2021. HyperBench: A Benchmark and Tool for Hypergraphs and Empirical Findings. ACM J. Exp. Algorithmics26 (2021), 1.6:1–1.6:40. doi:10.1145/3440015",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Zeyuan Hu and Daniel P. Miranker. 2024. TreeTracker Join: Turning the Tide When a Tuple Fails to Join.CoRRabs/2403.01631 (2024). arXiv:2403.01631 doi:10.48550/ARXIV.2403.01631",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Zeyuan Hu, Yisu Remy Wang, and Daniel P. Miranker. 2026. TreeTracker Join: Simple, Optimal, Fast.ACM Transactions on Database Systems51, 2 (2026), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting Logic Bugs in Database Engines via Equivalent Expression Transformation. In18th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2024, Santa Clara, CA, USA, July 10-12, 2024, Ada Gavrilovska and Douglas B. Terry (Eds.). USENIX Association, 821–835. https://www.usenix.org/conference/osdi24/presentation/jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woon-Hak Kang. 2019. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems.Proc. VLDB Endow.13, 1 (2019), 57–70. doi:10.14778/3357377. 3357382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Paraschos Koutris, Stijn Vansummeren, Qichen Wang, Yisu Remy Wang, and Xiangyao Yu. 2026. Database Theory in Action: Yannakakis’ Algorithm. In29th International Conference on Database Theory, ICDT 2026, Tampere, Finland, March 24-27, 2026 (LIPIcs, Vol. 365), Balder ten Cate and Maurice Funk (Eds.). Schloss Dagstuhl-Leibniz-Zentrum für Informatik, 25:1–25:6. doi:10.4230/LIPICS.ICDT. 2026.25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Jinhui Lai, Chi Zhang, Bingyan Li, Chenglin Liang, Jie Liang, Zhiyong Wu, Jingzhou Fu, Yu Jiang, and Zichen Xu. 2025. SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis. doi:10.1145/3769828",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Zheng Luo, Wim Van den Broeck, Guy Van den Broeck, and Yisu Remy Wang. 2026. Algorithms for Optimizing Acyclic Queries. In29th International Conference on Database Theory, ICDT 2026, Tampere, Finland, March 24-27, 2026 (LIPIcs, Vol. 365), Balder ten Cate and Maurice Funk (Eds.). Schloss Dagstuhl-LeibnizZentrum für Informatik, 17:1–17:18. doi:10.4230/LIPICS.ICDT.2026.17",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Yiming Qiao, Peter Boncz, and Huanchen Zhang. 2026. Robust Predicate Transfer with Dynamic Execution.Proc. VLDB Endow.19, 6 (2026), 1278–1290. DuckDB Merged Pull Request: https://github.com/duckdb/duckdb/pull/20633.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 667–682. https://www.usenix.org/conference/osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 13,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. [n. d.]. SQLSmith. https: //github.com/anse1/sqlsmith Accessed: March 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. In45th IEEE/ACM International Conference on Software Engineering, ICSE 2023, Melbourne, Australia, May 14-20, 2023. IEEE, 2072–2084. doi:10.1109/ ICSE48619.2023.00175",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting Logic Bugs of Join Optimizations in DBMS. doi:10.1145/3588909",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Transaction Processing Performance Council (TPC). [n. d.]. TPC-H Benchmark. Online. http://tpc.org/tpc_documents_current_versions/pdf/tpc-h_v3.0.0.pdf Accessed: March 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Mihalis Yannakakis. 1981. Algorithms for Acyclic Database Schemes. InVery Large Data Bases, 7th International Conference, September 9-11, 1981, Cannes, France, Proceedings. IEEE Computer Society, 82–94.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. doi:10.1145/ 3709674",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 19,
+      "text": "Hangdong Zhao, Yuanyuan Tian, Rana Alotaibi, Bailu Ding, Nicolas Bruno, Jesús Camacho-Rodríguez, Vassilis Papadimos, Ernesto Cervantes Juárez, César A. Galindo-Legaria, and Carlo Curino. 2026. I Can’t Believe It’s Not Yannakakis: Pragmatic Bitmap Filters in Microsoft SQL Server. In16th Conference on Innovative Data Systems Research, CIDR 2026, Chaminade, CA, USA, January 1821, 2026. www.cidrdb.org",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Junyi Zhao, Kai Su, Yifei Yang, Xiangyao Yu, Paraschos Koutris, and Huanchen Zhang. 2025. Debunking the Myth of Join Ordering: Toward Robust SQL Analytics. doi:10.1145/3725283",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. InCCS ’20: 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event, USA, November 9-13, 2020, Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna (Eds.). ACM, 955–970. doi:10.1145/3372297.34172",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 12,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 667–682. https://www.usenix.org/conference/osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 18,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. doi:10.1145/ 3709674",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "prints the DOI of the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 6,\n7,9,12–15,18,21]",
+      "technique": "pqs",
+      "sentence": "A natural choice to verify the physical translation of such algorithms is to leverage conventional database testing frameworks [ 6, 7,9,12–15,18,21].",
+      "context_before": "mization infrastructure [ 19]. The latter shows a promising direction: re-evaluating mature query optimization heuristics through formal lenses, such as ACQ evaluation. A major challenge in designing practical ACQ evaluation algorithms is that system constraints commonly diverge from theoretical assumptions. Hence, a gap exists between the highlevel logical specification of an algorithm and its low-level physical implementation. Ensuring a faithful logical-to-physical translation is therefore crucial to guarantee that theoretical optimality actually translates into practical performance gains.",
+      "context_after": "However, incompatibilities exist with the needs of the translation. First, algorithms rely on an input data structure called a join tree1that is distinct from a query plan and is not commonly seen in mainstream databases. This fact imposes a significant challenge to conventional database testing frameworks, as the correctness of query evaluation depends not only on the query plans but also on the topology of join trees. Unfortunately, the existing testing frameworks do not support join trees, and retrofitting them is highly non-trivial. Second, existing frameworks generate SQL queries and dat",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 4314,
+      "cited_reference": {
+        "number": 12,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 667–682. https://www.usenix.org/conference/osdi20/presentation/rigger",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 6,7,12–15,21]",
+      "technique": "pqs",
+      "sentence": "5 Related Work Extensive research applies differential and generator-based testing to validate database implementations [ 6,7,12–15,21], typically executing synthesized SQL queries against mature DBMSs.",
+      "context_before": "ral evolution of CODA to its final state (Figure 3), entirely offloading the responsibilities of GYO validation (Section 4.2) and join tree construction to CODA ’s 3Remarkably, this defect was concurrently identified by a referee for our ICDT ’25 submission.synthesis pipeline. While TPC-H Q8 (an 8-relation join) and Q9 (a 6-relation join) eventually triggered this RIPviolation, their sprawling plans were unnecessarily complicated, requiring significant effort to trace the structural root cause. By contrast, CODA reproduced the identical defect using a minimal structure of just three relations.",
+      "context_after": "While sharing this differential testing spirit, CODA differentiates itself in three ways. (1) Agility for nascent algorithms: unlike existing tools that assume stable architectures, CODA is lightweight, avoiding the prohibitive overhead of integrating highly volatile algorithm prototypes (e.g., TTJ) into mature systems. (2) Micro-topological synthesis: instead of operating at the macroscopic SQL interface and relying on query optimizers whose heuristics may miss problematic plans and join trees [ 5,10,20],CODA directly synthesizes join trees and query plans, allowing it to stress-test both ph",
+      "section": "4.3 Case Study 3: Architectural Evolution and",
+      "page": 6,
+      "char_offset": 29082,
+      "cited_reference": {
+        "number": 12,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 667–682. https://www.usenix.org/conference/osdi20/presentation/rigger",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:19:46Z",
+    "is_model_written": true,
+    "summary": "An experience paper on CODA, a testing framework used while co-designing the logical specification and physical implementation of TreeTracker Join. The authors argue existing database testing frameworks cannot supply the algorithm-specific inputs such as join trees that this needs, and that macro-benchmark queries are too noisy to isolate the defects involved. By synthesising minimal reproducible examples, CODA isolated translation defects, one of which refined the algorithm's formal precondition.",
+    "narrative": "Two citations, treating SQLancer's line of work as the conventional database testing frameworks CODA is measured against in argument -- the natural choice for verifying a physical translation, which the authors then explain does not fit their need.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2603_21530.json b/_data/papers/paper_arxiv_2603_21530.json
new file mode 100644
index 0000000..2c2c7f3
--- /dev/null
+++ b/_data/papers/paper_arxiv_2603_21530.json
@@ -0,0 +1,666 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:23:16Z",
+  "paper": {
+    "id": "paper:arxiv:2603.21530",
+    "title": "LLM-Based Test Case Generation in DBMS through Monte Carlo Tree Search",
+    "authors": [
+      "Yujia Chen",
+      "Yingli Zhou",
+      "Fangyuan Zhang",
+      "Cuiyun Gao"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2603.21530",
+    "s2_paper_id": "c1254b7ff833135a0a9dc60220434ef2716ff35d",
+    "url": "https://arxiv.org/abs/2603.21530",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2603.21530",
+      "retrieved_at": "2026-09-10T15:23:16Z",
+      "chars": 71977,
+      "content_sha256": "sha256:70f42bd64d0e4ab72bba5a6c729d8acaa81c060efbc2140d961621e4bcf14e6e"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3001
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 10559
+      },
+      {
+        "number": "2.1",
+        "title": "DBMS Testing",
+        "start": 10572
+      },
+      {
+        "number": "2.2",
+        "title": "Monte Carlo Tree Search",
+        "start": 11941
+      },
+      {
+        "number": "3",
+        "title": "Approach",
+        "start": 13569
+      },
+      {
+        "number": "3.1",
+        "title": "Overview",
+        "start": 13798
+      },
+      {
+        "number": "3.2",
+        "title": "Feature-Guided Error-Driven Test Case",
+        "start": 14690
+      },
+      {
+        "number": "3.2.1",
+        "title": "Feature tree construction.We collect official documentation",
+        "start": 15188
+      },
+      {
+        "number": "3.2.3",
+        "title": "Error-Driven test case generation.With the selected feature",
+        "start": 19602
+      },
+      {
+        "number": "3.2",
+        "title": "Feature-Guided Error-Driven Test Case",
+        "start": 22521
+      },
+      {
+        "number": "3.2.1",
+        "title": "Feature tree construction. We collect ocial documentation",
+        "start": 23010
+      },
+      {
+        "number": "3.2.3",
+        "title": "Error-Driven test case generation. With the selected feature",
+        "start": 25576
+      },
+      {
+        "number": "3.3",
+        "title": "Monte Carlo Tree Search-Based Test Case",
+        "start": 28437
+      },
+      {
+        "number": "3.3.1",
+        "title": "Mutation rules curation.We formulate test case mutation as",
+        "start": 28804
+      },
+      {
+        "number": "4",
+        "title": "Experimental Setup",
+        "start": 34738
+      },
+      {
+        "number": "4.1",
+        "title": "Research Questions",
+        "start": 34759
+      },
+      {
+        "number": "4.2",
+        "title": "Studied DBMSs",
+        "start": 35154
+      },
+      {
+        "number": "4.3",
+        "title": "Studied LLMs and Baseline",
+        "start": 36298
+      },
+      {
+        "number": "4.4",
+        "title": "Metrics",
+        "start": 37200
+      },
+      {
+        "number": "4.5",
+        "title": "Implementation Details",
+        "start": 38343
+      },
+      {
+        "number": "5",
+        "title": "Result",
+        "start": 39974
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 52625
+      },
+      {
+        "number": "6.1",
+        "title": "Why does MIST work?",
+        "start": 52638
+      },
+      {
+        "number": "6.2",
+        "title": "Threats to Validity",
+        "start": 55462
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 57657
+      },
+      {
+        "number": "7.1",
+        "title": "LLM-based Testing",
+        "start": 57672
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 59882
+      },
+      {
+        "number": "S",
+        "title": "Chand Publishing.",
+        "start": 61241
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Moritz Bley, Tobias Scharnowski, Simon Wörner, Moritz Schloegel, and Thorsten Holz. 2025. Protocol-Aware Firmware Rehosting for Effective Fuzzing of Embedded Network Stacks.arXiv preprint arXiv:2509.13740(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. 2006. Generating queries with cardinality constraints for dbms testing.IEEE Transactions on Knowledge and Data Engineering18, 12 (2006), 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Rajiv Chopra. 2010.Database Management System (DBMS) A Practical Approach. S. Chand Publishing.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Ju Fan, Zihui Gu, Songyue Zhang, Yuxin Zhang, Zui Chen, Lei Cao, Guoliang Li, Samuel Madden, Xiaoyong Du, and Nan Tang. 2024. Combining small language models and large language models for zero-shot nl2sql.Proceedings of the VLDB Endowment17, 11 (2024), 2750–2763.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Meihao Fan, Xiaoyue Han, Ju Fan, Chengliang Chai, Nan Tang, Guoliang Li, and Xiaoyong Du. 2024. Cost-effective in-context learning for entity resolution: A design space exploration. In Engineering (ICDE). IEEE, 3696–3709.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, and Yu Jiang. 2024. Sedar: Obtaining HighQuality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer. InProceedings of the 46th IEEE/ACM International Conference on Software Engineering, ICSE. 146:1–146:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. In37th IEEE/ACM International Conference on Automated Software Engineering, ASE. 49:1–49:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Victor Giannakouris and Immanuel Trummer. 2025. 𝜆-tune: Harnessing large language models for automated database system tuning.Proceedings of the ACM on Management of Data3, 1 (2025), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, et al .2024. The llama 3 herd of models.arXiv preprint arXiv:2407.21783 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Qiuhan Gu. 2023. Llm-based code generation method for golang compiler testing. InProceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 2201–2203.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Huggingface. 2025. Huggingface. https://huggingface.co/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Sathvik Joel, Jie Wu, and Fatemeh Fard. 2024. A survey on llm-based code generation for low-resource and domain-specific programming languages.ACM Transactions on Software Engineering and Methodology(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Bilal Kartal, Pablo Hernandez-Leal, and Matthew E Taylor. 2019. Action guidance with MCTS for deep reinforcement learning. InProceedings of the AAAI conference on artificial intelligence and interactive digital entertainment, Vol. 15. 153–159.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Levente Kocsis and Csaba Szepesvári. 2006. Bandit based monte-carlo planning. InEuropean conference on machine learning. Springer, 282–293.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph Gonzalez, Hao Zhang, and Ion Stoica. 2023. Efficient memory management for large language model serving with pagedattention. In Proceedings of the 29th symposium on operating systems principles. 611–626.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jiale Lao, Yibo Wang, Yufei Li, Jianping Wang, Yunjia Zhang, Zhiyuan Cheng, Wanghu Chen, Mingjie Tang, and Jianguo Wang. 2024. Gptuner: A manualreading database tuning system via gpt-guided bayesian optimization.Proceedings of the VLDB Endowment17, 8 (2024), 1939–1952.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Boyan Li, Yuyu Luo, Chengliang Chai, Guoliang Li, and Nan Tang. 2024. The Dawn of Natural Language to SQL: Are We Fully Ready?arXiv preprint arXiv:2406.01265(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jinyang Li, Binyuan Hui, Ge Qu, Jiaxi Yang, Binhua Li, Bowen Li, Bailin Wang, Bowen Qin, Ruiying Geng, Nan Huo, et al .2023. Can llm already serve as a database interface? a big bench for large-scale database grounded text-to-sqls. Advances in Neural Information Processing Systems36 (2023), 42330–42357.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Lan Li, Liri Fang, and Vetle I Torvik. 2024. AutoDCWorkflow: LLM-based Data Cleaning Workflow Auto-Generation and Benchmark.arXiv preprint arXiv:2412.06724(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Fang Liu, Yang Liu, Lin Shi, Houkun Huang, Ruifeng Wang, Zhen Yang, Li Zhang, Zhongqi Li, and Yuchi Ma. 2024. Exploring and evaluating hallucinations in llm-powered code generation.arXiv preprint arXiv:2404.00971(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Eric Lo, Carsten Binnig, Donald Kossmann, M Tamer Özsu, and Wing-Kai Hon. 2010. A framework for testing DBMS features.The VLDB Journal19, 2 (2010), 203–230. ICSE-Companion ’26, April 12–18, 2026, Rio de Janeiro, Brazil Yujia Chen, Yingli Zhou, Fangyuan Zhang, and Cuiyun Gao",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Zan Ahmad Naeem, Mohammad Shahmeer Ahmad, Mohamed Eltabakh, Mourad Ouzzani, and Nan Tang. 2024. RetClean: Retrieval-Based Data Cleaning Using LLMs and Data Lakes.Proceedings of the VLDB Endowment17, 12 (2024), 4421– 4424.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Xianfei Ou, Cong Li, Yanyan Jiang, and Chang Xu. 2024. The Mutators Reloaded: Fuzzing Compilers with Large Language Model Generated Mutation Operators. InProceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 4, ASPLOS. 298–312.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Mike Owens and Grant Allen. 2010.SQLite. Apress LP New York.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "M Tamer Özsu and Patrick Valduriez. 1996. Distributed and parallel database systems.ACM Computing Surveys (CSUR)28, 1 (1996), 125–128.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Lauren Pick, Amanda Xu, Ankush Desai, Sanjit A Seshia, and Aws Albarghouthi. 2025. Checking Observational Correctness of Database Systems.Proceedings of the ACM on Programming Languages9, OOPSLA1 (2025), 1661–1688.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Yichen Qian, Yongyi He, Rong Zhu, Jintao Huang, Zhijian Ma, Haibin Wang, Yaohua Wang, Xiuyu Sun, Defu Lian, Bolin Ding, et al .2024. UniDM: A Unified Framework for Data Manipulation with Large Language Models.Proceedings of Machine Learning and Systems6 (2024), 465–482.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, et al .2024. Qwen2.5 Technical Report.arXiv preprint arXiv:2412.15115 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Mark Raasveldt and Hannes Mühleisen. 2019. Duckdb: an embeddable analytical database. InProceedings of the 2019 international conference on management of data. 1981–1984.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM, 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA (2020), 211:1–211:30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation, OSDI. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 34,
+      "text": "Andreas Seltenreich. 2025. Sqlsmith. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Vikramank Singh, Kapil Eknath Vaidya, Vinayshekhar Bannihatti Kumar, Sopan Khosla, Murali Narayanaswamy, Rashmi Gangadharaiah, and Tim Kraska. 2024. Panda: Performance debugging for databases using LLM agents. (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Michael Stonebraker and Lawrence A Rowe. 1986. The design of Postgres.ACM Sigmod Record15, 2 (1986), 340–355.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Maciej Świechowski, Konrad Godlewski, Bartosz Sawicki, and Jacek Mańdziuk. 2023. Monte Carlo tree search: A review of recent modifications and applications. Artificial Intelligence Review56, 3 (2023), 2497–2562.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Jianxun Wang and Yixiang Chen. 2023. A review on code generation with llms: Application and evaluation. In Artificial Intelligence (MedAI). IEEE, 284–289.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Junjie Wang, Yuchao Huang, Chunyang Chen, Zhe Liu, Song Wang, and Qing Wang. 2024. Software testing with large language models: Survey, landscape,and vision.IEEE Transactions on Software Engineering50, 4 (2024), 911–936.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry practice of coverage-guided enterprise-level DBMS fuzzing. In Software Engineering in Practice (ICSE-SEIP). 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Chunqiu Steven Xia, Matteo Paltenghi, Jia Le Tian, Michael Pradel, and Lingming Zhang. 2024. Fuzz4All: Universal Fuzzing with Large Language Models. In Proceedings of the 46th IEEE/ACM International Conference on Software Engineering, ICSE, Lisbon, Portugal. 126:1–126:13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Yuanmin Xie, Zhenyang Xu, Yongqiang Tian, Min Zhou, Xintong Zhou, and Chengnian Sun. 2025. Kitten: A Simple Yet Effective Baseline for Evaluating LLMBased Compiler Testing Techniques. InProceedings of the 34th ACMSIGSOFT International Symposium on Software Testing and Analysis. 21–25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Zhiyi Xue, Liangguo Li, Senyue Tian, Xiaohong Chen, Pingping Li, Liangyu Chen, Tingting Jiang, and Min Zhang. 2024. Llm4fin: Fully automating llm-powered test case generation for fintech software acceptance testing. InProceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. 1643–1655.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Chenyuan Yang, Yinlin Deng, Runyu Lu, Jiayi Yao, Jiawei Liu, Reyhaneh Jabbarvand, and Lingming Zhang. 2024. WhiteFox: White-Box Compiler Fuzzing Empowered by Large Language Models.Proc. ACM Program. Lang.8, OOPSLA2 (2024), 709–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Chenyuan Yang, Zijie Zhao, and Lingming Zhang. 2025. KernelGPT: Enhanced Kernel Fuzzing via Large Language Models. InProceedings of the 30th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2, ASPLOS. 560–573.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Yupeng Yang, Yongheng Chen, Rui Zhong, Jizhou Chen, and Wenke Lee. 2024. Towards generic database management system fuzzing. In33rd USENIX Security Symposium (USENIX Security 24). 901–918.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Yihang Zheng, Bo Li, Zhenghao Lin, Yi Luo, Xuanhe Zhou, Chen Lin, Jinsong Su, Guoliang Li, and Shifu Li. 2024. Revolutionizing Database Q&A with Large Language Models: Comprehensive Benchmark and Evaluation.arXiv preprint arXiv:2409.04475(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing database management systems with language validity and coverage feedback. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Scaling Automated Database System Testing.arXiv preprint arXiv:2503.21424(2025).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments.arXiv preprint arXiv:2505.02012(2025).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 51,
+      "text": "Xuanhe Zhou, Guoliang Li, Chengliang Chai, and Jianhua Feng. 2021. A learned query rewrite system using monte carlo tree search.Proceedings of the VLDB Endowment15, 1 (2021), 46–58.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Xuanhe Zhou, Guoliang Li, Zhaoyan Sun, Zhiyuan Liu, Weize Chen, Jianming Wu, Jiesi Liu, Ruohang Feng, and Guoyang Zeng. 2024. D-bot: Database diagnosis system using large language models.Proceedings of the VLDB Endowment17, 10 (2024), 2514–2527. Received 15 November 2025; revised 6 January 2026; accepted 6 January 2026",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM, 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 31,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA (2020), 211:1–211:30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation, OSDI. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 49,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Scaling Automated Database System Testing.arXiv preprint arXiv:2503.21424(2025).",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing sqlancer_pp"
+      ]
+    },
+    {
+      "number": 50,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments.arXiv preprint arXiv:2505.02012(2025).",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "prints the arXiv id of the paper introducing shqvel"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Traditional approaches for generating test cases, such as BuzzBee [ 46], SQLsmith [ 34], SQLancer [ 32], can be effective for a specific DBMS; however, there exist many DBMSs designed for different application scenarios, such as DuckDB [ 29] for analytical workloads, PostgreSQL [ 36] for general-purpose OLTP systems, and SQLite [ 24] for lightweight embedded environments.",
+      "context_before": "oduction Database Management Systems (DBMSs) are fundamental infrastructure for modern data-driven society, underpinning applications across finance, healthcare and e-commerce [ 25,29,36]. Ensuring the correctness and robustness of DBMSs is vital, as even minor faults can lead to severe consequences, including data corruption, service outages, or security breaches [ 3,21,26]. An essential step toward assuring DBMS reliability is thorough testing using highquality SQL test cases, which help uncover subtle defects before deployment and prevent them from affecting production environments [ 2,48].",
+      "context_after": "Adapting these methods to different DBMSs requires substantial manual effort, such as crafting grammar rules and maintaining specific operators, which severely limits their scalability [ 48–50]. Recent advances in Large Language Models (LLMs) present promising opportunities for DBMS test case generation. First, LLMs have already demonstrated success in related data management tasks, including text-to-SQL [ 4,17,18], data cleaning [ 5,19,22,27], knob tuning [8, 16], and DBMS diagnosis [35, 47, 52]. Second, LLMs are applied in testing across multiple domains, such as general software testing [",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3607,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 48–50]",
+      "technique": null,
+      "sentence": "Adapting these methods to different DBMSs requires substantial manual effort, such as crafting grammar rules and maintaining specific operators, which severely limits their scalability [ 48–50].",
+      "context_before": "6]. An essential step toward assuring DBMS reliability is thorough testing using highquality SQL test cases, which help uncover subtle defects before deployment and prevent them from affecting production environments [ 2,48]. Traditional approaches for generating test cases, such as BuzzBee [ 46], SQLsmith [ 34], SQLancer [ 32], can be effective for a specific DBMS; however, there exist many DBMSs designed for different application scenarios, such as DuckDB [ 29] for analytical workloads, PostgreSQL [ 36] for general-purpose OLTP systems, and SQLite [ 24] for lightweight embedded environments.",
+      "context_after": "Recent advances in Large Language Models (LLMs) present promising opportunities for DBMS test case generation. First, LLMs have already demonstrated success in related data management tasks, including text-to-SQL [ 4,17,18], data cleaning [ 5,19,22,27], knob tuning [8, 16], and DBMS diagnosis [35, 47, 52]. Second, LLMs are applied in testing across multiple domains, such as general software testing [ 39,43], compiler testing [ 10,42], and embedded testing [ 1]. As illustrated in Figure 1, LLMs can directly generate SQL test cases by leveraging their understanding of SQL syntax and semanticsar",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3982,
+      "cited_reference": {
+        "number": 49,
+        "text": "Suyang Zhong and Manuel Rigger. 2025. Scaling Automated Database System Testing.arXiv preprint arXiv:2503.21424(2025).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing sqlancer_pp"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 49,50]",
+      "technique": null,
+      "sentence": "2 Studied DBMSs Following prior studies [ 49,50], we evaluate MIST on three widelyused open-source DBMS representing diverse architectural designs and SQL dialect characteristics.",
+      "context_before": "he recent branch coverage improvement of each SQL; once its coverage gain becomes saturated, the corresponding node is excluded from further expansion in subsequent MCTS iterations. 4 Experimental Setup 4.1 Research Questions In this paper, we mainly investigate the following research questions through experiments. •RQ1: How effective is MIST in improving code coverage for DBMSs compared to baseline approaches? •RQ2: How does MIST perform across different DBMS modules (i.e.,Parser,Optimizer,Executor,Storage)? •RQ3: What are the impacts of the two stages in MIST on code coverage improvement? 4.",
+      "context_after": "•DuckDB [ 29]:An in-process analytical database designed for OLAP workloads with columnar storage and vectorized execution. It supports modern SQL features including window functions and CTEs. •PostgreSQL [ 36]:A mature relational database system with a sophisticated query optimizer and advanced features such as materialized views and JSON operations. •SQLite [ 24]:A lightweight embedded database engine widely deployed in mobile and embedded systems. It implements a comprehensive SQL dialect with unique features such as virtual tables. These three systems collectively cover a broad spectrum o",
+      "section": "4.2 Studied DBMSs",
+      "page": 6,
+      "char_offset": 35156,
+      "cited_reference": {
+        "number": 49,
+        "text": "Suyang Zhong and Manuel Rigger. 2025. Scaling Automated Database System Testing.arXiv preprint arXiv:2503.21424(2025).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing sqlancer_pp"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "ShQveL",
+      "technique": null,
+      "sentence": "Inspired by ShQveL [ 50], we adopt the state-of-the-art LLMbased framework Fuzz4All for fair comparison.",
+      "context_before": "across different model capacities, we select four open-source LLMs spanning different parameter scales and model families: Qwen2.5-7B, Llama3.1-8B, Qwen2.5-14B, and Qwen2.5-32B. These models represent lightweight to mediumsized LLMs that are practical for local deployment in industrial environments, aligning with the resource constraints commonly faced in real-world DBMS testing scenarios. For baseline comparison, we select Fuzz4All [ 41] as our primary baseline. To the best of our knowledge, there is currently no existing work that uses LLMs as complete test case generators for DBMS testing.",
+      "context_after": "Fuzz4All is the first universal fuzzer that leverages LLMs to generate test inputs across multiple programming languages and systems. 4.4 Metrics We evaluate MIST using two categories of coverage metrics to comprehensively assess its effectiveness in exercising DBMS code. •Code Coverage.We measure three standard code coverage metrics: (1)Line Coverage, which calculates the percentage of executed source code lines; (2)Function Coverage, which measures the percentage of functions that are invoked at least once; and (3)Branch Coverage, which tracks the percentage of conditional branches (e.g., i",
+      "section": "4.3 Studied LLMs and Baseline",
+      "page": 7,
+      "char_offset": 36960,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 30,31,33] employs handcrafted dialect-specific generators with sophisticated test oracles to detect logic bugs.",
+      "context_before": "s Existing methods for generating test cases for DBMSs can be divided into two categories: mutation-based and generation-based testing. Mutation-based approaches modify existing SQL statements to produce new test cases. Sedar [ 6] improves input diversity by transferring SQL seeds across DBMS dialects. Griffin [ 7] uses a grammar-free approach with a lightweight metadata graph to guide semantically-correct query mutations. Generation-based approaches construct queries from scratch using predefined rules. SQLsmith [ 34] generates random yet well-formed SQL queries by leveraging schema metadata.",
+      "context_after": "Different from these traditional approaches that rely on manual grammar engineering or random mutation, our work is the first to leverage lightweight LLMs with hierarchical feature guidance and MCTS-based mutation guided by coverage feedback to systematically improve code coverage. 8 Conclusion In this paper, we propose MIST, a framework for generating highquality SQL test cases using lightweight LLMs. By combining featureguided generation with error feedback and MCTS-based mutation guided by coverage feedback, MIST addresses the challenges of limited model adaptability and insufficient test",
+      "section": "7.1 LLM-based Testing",
+      "page": 11,
+      "char_offset": 59475,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "describes_as_state_of_the_art": [
+        "M4"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:21:29Z",
+    "is_model_written": true,
+    "summary": "MIST generates SQL test cases for DBMSs with lightweight LLMs guided by Monte Carlo tree search. It addresses two industrial constraints: organisations often must use small models for security and privacy reasons, and those models struggle with proprietary SQL dialects and tend to produce semantically similar queries that plateau in coverage. MIST builds a hierarchical feature tree and uses error feedback to steer generation, improving line coverage by 43.3% over the strongest baseline.",
+    "narrative": "SQLancer is cited as the traditional approach whose handcrafted dialect-specific generators and sophisticated oracles work well for a specific system but need substantial manual effort to move to another -- the cost MIST aims to remove. ShQveL, from the same project, informed the choice of LLM baseline.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "motivation",
+      "M3": "background",
+      "M4": "background",
+      "M5": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2604_01442.json b/_data/papers/paper_arxiv_2604_01442.json
new file mode 100644
index 0000000..33fb9de
--- /dev/null
+++ b/_data/papers/paper_arxiv_2604_01442.json
@@ -0,0 +1,579 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:23:52Z",
+  "paper": {
+    "id": "paper:arxiv:2604.01442",
+    "title": "Fuzzing with Agents? Generators Are All You Need",
+    "authors": [
+      "Vasudev Vikram",
+      "Rohan Padhye"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2604.01442",
+    "s2_paper_id": "cbba1fcd8795f5727dbe286aeb9af1f5e5d4ba09",
+    "url": "https://arxiv.org/abs/2604.01442",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2604.01442",
+      "retrieved_at": "2026-09-08T15:23:52Z",
+      "chars": 70975,
+      "content_sha256": "sha256:aa694a666241552c3c6b7648eccf5474410b8939ffad14a999a875b34e4d768d"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2027
+      },
+      {
+        "number": "1.1",
+        "title": "Motivating Example",
+        "start": 6181
+      },
+      {
+        "number": "1.2",
+        "title": "Contributions",
+        "start": 8734
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND",
+        "start": 11688
+      },
+      {
+        "number": "2.1",
+        "title": "Coverage-Guided Fuzzing",
+        "start": 11701
+      },
+      {
+        "number": "2.2",
+        "title": "Parametric Generator-Based Fuzzing",
+        "start": 13223
+      },
+      {
+        "number": "2.3",
+        "title": "Control-Flow Analysis and Dominators",
+        "start": 15802
+      },
+      {
+        "number": "3",
+        "title": "APPROACH",
+        "start": 17747
+      },
+      {
+        "number": "3.1",
+        "title": "Dominator Predicate Analysis",
+        "start": 19091
+      },
+      {
+        "number": "3.1.1",
+        "title": "LLM Predicate Identification. As an alternative used by",
+        "start": 20495
+      },
+      {
+        "number": "3.2",
+        "title": "Dynamic Predicate Feedback",
+        "start": 21956
+      },
+      {
+        "number": "3.3",
+        "title": "Agent-based Generator Synthesis",
+        "start": 24335
+      },
+      {
+        "number": "4",
+        "title": "EVALUATION",
+        "start": 26234
+      },
+      {
+        "number": "4.1",
+        "title": "Experimental Setup",
+        "start": 26870
+      },
+      {
+        "number": "4.1.2",
+        "title": "Benchmarks. We evaluate on six real-world Java library",
+        "start": 27481
+      },
+      {
+        "number": "4.1.4",
+        "title": "Repetitions. All experiments are repeated five times with",
+        "start": 28315
+      },
+      {
+        "number": "4.1.5",
+        "title": "Metrics. We evaluate generators along two dimensions:",
+        "start": 29814
+      },
+      {
+        "number": "4.1.6",
+        "title": "Scale. In total, our evaluation comprises 3×7×5=105",
+        "start": 30268
+      },
+      {
+        "number": "4.2",
+        "title": "RQ1: Coverage of synthesized generators",
+        "start": 30778
+      },
+      {
+        "number": "4.3",
+        "title": "RQ2: Impact of Coverage Guidance",
+        "start": 35788
+      },
+      {
+        "number": "4.4",
+        "title": "RQ3: Impact of Predicate Feedback",
+        "start": 40386
+      },
+      {
+        "number": "4.5",
+        "title": "RQ4: LLM vs. Static Predicate Analysis",
+        "start": 43302
+      },
+      {
+        "number": "5",
+        "title": "RELATED WORK",
+        "start": 45080
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "[n. d.]. WALA: T.J. Watson Libraries for Analysis. https://github.com/wala/wala. Accessed 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "[n. d.]. junit-quickcheck-generator. https://pholser.github.io/junit-quickcheck/ site/1.0/usage/other-types.html. Accessed: 2024-10-31.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "[n. d.]. libFuzzer – a library for coverage-guided fuzz testing. https://llvm.org/ docs/LibFuzzer.html. Accessed: 2021-08-31.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Joshua Ackerman and George Cybenko. 2023. Large language models for fuzzing parsers (registered report). In Proceedings of the 2nd International Fuzzing Workshop. 31–38.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Nadia Alshahwan, Jubin Chheda, Anastasia Finegenova, Beliz Gokkaya, Mark Harman, Inna Harper, Alexandru Marginean, Shubho Sengupta, and Eddy Wang. 2024. Automated unit test improvement using large language models at meta. arXiv preprint arXiv:2402.09171 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Anthropic. [n. d.]. Claude Agent SDK. https://platform.claude.com/docs/en/agentsdk/overview. Accessed 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Cornelius Aschermann, Sergej Schumilo, Tim Blazytko, Robert Gawlik, and Thorsten Holz. 2019. REDQUEEN: Fuzzing with Input-to-State Correspondence.. InNDSS, Vol. 19. 1–15. https://doi.org/10.14722/ndss.2019.23371",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In (ICSE). IEEE, 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 9,
+      "text": "Domagoj Babić, Stefan Bucur, Yaohui Chen, Franjo Ivančić, Tim King, Markus Kusano, Caroline Lemieux, László Szekeres, and Wei Wang. 2019. Fudge: fuzz driver generation at scale. In Proceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 975–985.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Patrick Bareiß, Beatriz Souza, Marcelo d’Amorim, and Michael Pradel. 2022. Code generation tools (almost) for free? a study of few-shot, pre-trained language models on code. arXiv preprint arXiv:2206.01335 (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. 2016. Coveragebased Greybox Fuzzing as Markov Chain. In Proceedings of the 2016 ACMSIGSAC Conference on Computer and Communications Security (CCS). 1032–1043. https: //doi.org/10.1145/2976749.2978428",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. 2017. Coveragebased greybox fuzzing as markov chain. IEEE Transactions on Software Engineering 45, 5 (2017), 489–506.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Marcel Böhme, László Szekeres, and Jonathan Metzman. 2022. On the Reliability of Coverage-Based Fuzzer Benchmarking. In 44th IEEE/ACM International Conference on Software Engineering (ICSE’22). https://doi.org/10.1145/3510003.3510230 to appear.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Sang Kil Cha, Maverick Woo, and David Brumley. 2015. Program-adaptive mutational fuzzing. In. IEEE, 725– 741. https://doi.org/10.1109/SP.2015.50",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Chuyang Chen and Brendan Dolan-Gavitt. 2025. ELFuzz: Efficient Input Generation via LLM-driven Synthesis Over Fuzzer Space. In 34th USENIX Security Symposium (USENIX Security 25). 6279–6298.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Peng Chen and Hao Chen. 2018. Angora: Efficient fuzzing by principled search. In. IEEE, 711–725. https: //doi.org/10.1109/SP.2018.00046",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Koen Claessen and John Hughes. 2000. QuickCheck: a lightweight tool for random testing of Haskell programs. In Proceedings of the fifth ACMSIGPLAN international conference on Functional programming. 268–279.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Chenyuan Yang, Shizhuo Dylan Zhang, Shujing Yang, and Lingming Zhang. 2023. Large language models are edge-case fuzzers: Testing deep learning libraries via fuzzgpt. arXiv preprint arXiv:2304.02014 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Zhen Yu Ding and Claire Le Goues. 2021. An Empirical Study of OSS-Fuzz Bugs. In (MSR). IEEE, IEEE Computer Society, Los Alamitos, CA, USA, 131–142. https: //doi.org/10.1109/MSR52588.2021.00026",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Shuitao Gan, Chao Zhang, Peng Chen, Bodong Zhao, Xiaojun Qin, Dong Wu, and Zuoning Chen. 2020. GREYONE: Data flow sensitive fuzzing. In 29th USENIX Security Symposium (USENIX Security 20). 2577–2594. https://doi.org/10.5555/ 3489212.3489357",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Harrison Green, Fraser Brown, and Claire Le Goues. 2026. Automatic, Expressive, and Scalable Fuzzing with Stitching. arXiv preprint arXiv:2602.18689 (2026).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Jiale Guo, Suizhi Huang, Mei Li, Dong Huang, Xingsheng Chen, Regina Zhang, Zhijiang Guo, Han Yu, Siu-Ming Yiu, Pietro Lio, and Kwok-Yan Lam. 2025. A Comprehensive Survey on Benchmarks and Solutions in Software Engineering of LLM-Empowered Agentic System. arXiv:2510.09721 [cs.SE] https://arxiv.org/ abs/2510.09721",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Paul Holser. [n. d.]. junit-quickcheck: Property-based testing, JUnit-style. https: //github.com/pholser/junit-quickcheck. Accessed: 2021-08-31.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Xinyi Hou, Yanjie Zhao, Yue Liu, Zhou Yang, Kailong Wang, Li Li, Xiapu Luo, David Lo, John Grundy, and Haoyu Wang. 2023. Large language models for software engineering: A systematic literature review. ACM Transactions on Software Engineering and Methodology (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Katherine Hough and Jonathan Bell. 2024. Crossover in Parametric Fuzzing. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Kyriakos K Ispoglou, Daniel Austin, Vishwath Mohan, and Mathias Payer. 2020. Fuzzgen: Automatic fuzzer generation. In Proceedings of the 29th USENIX Conference on Security Symposium. 2271–2287.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "B. Jeong, J. Jang, H. Yi, J. Moon, J. Kim, I. Jeon, T. Kim, W. Shim, and Y. Hwang. 2023. UTOPIA: Automatic Generation of Fuzz Driver using Unit Tests. In 2023. IEEE Computer Society, Los Alamitos, CA, USA, 746–762. https://doi.org/10.1109/SP46215.2023.00043",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Yu Jiang, Jie Liang, Fuchen Ma, Yuanliang Chen, Chijin Zhou, Yuheng Shen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Shanshan Li, and Quan Zhang. 2024. When Fuzzing Meets LLMs: Challenges and Opportunities. In Companion Proceedings of the 32nd ACM International Conference on the Foundations of Software Engineering (Porto de Galinhas, Brazil) (FSE 2024). Association for Computing Machinery, New York, N",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Carlos E Jimenez, John Yang, Alexander Wettig, Shunyu Yao, Kexin Pei, Ofir Press, and Karthik Narasimhan. 2023. Swe-bench: Can language models resolve real-world github issues? arXiv preprint arXiv:2310.06770 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "George Klees, Andrew Ruef, Benji Cooper, Shiyi Wei, and Michael Hicks. 2018. Evaluating fuzz testing. In Proceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security. 2123–2138. https://doi.org/10.1145/ 3243734.3243804",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Shuvendu Lahiri, Aaditya Naik, Georgios Sakkas, Piali Choudhury, Curtis von Veh, Madan Musuvathi, Jeevana Priya Inala, Chenglong Wang, and Jianfeng Gao. 2022. Interactive Code Generation via Test-Driven User-Intent Formalization. arXiv. https://www.microsoft.com/en-us/research/publication/interactive-codegeneration-via-test-driven-user-intent-formalization/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Leonidas Lampropoulos, Michael Hicks, and Benjamin C Pierce. 2019. Coverage guided, property based testing. Proceedings of the ACM on Programming Languages 3, OOPSLA, Article 181 (Oct. 2019), 29 pages. https://doi.org/10.1145/ 3360607",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Caroline Lemieux, Jeevana Priya Inala, Shuvendu K Lahiri, and Siddhartha Sen. 2023. CODAMOSA: Escaping Coverage Plateaus in Test Generation with Pretrained Large Language Models. In 45th International Conference on Software Engineering, ser. ICSE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Caroline Lemieux and Koushik Sen. 2018. Fairfuzz: A targeted mutation strategy for increasing greybox fuzz testing coverage. In Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering. 475–485. https: //doi.org/10.1145/3238147.3238176",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Dongge Liu, Jonathan Metzman, and Oliver Chang. 2023. Open Source Insights. https://security.googleblog.com/2023/08/ai-powered-fuzzing-breakingbug-hunting.html. Retrieved February 27, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Junwei Liu, Kaixin Wang, Yixuan Chen, Xin Peng, Zhenpeng Chen, Lingming Zhang, and Yiling Lou. 2025. Large Language Model-Based Agents for Software Engineering: A Survey. arXiv:2409.02977 [cs.SE] https://arxiv.org/abs/2409.02977",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Jiawei Liu, Chunqiu Steven Xia, Yuyao Wang, and Lingming Zhang. 2024. Is your code generated by chatgpt really correct? rigorous evaluation of large language models for code generation. Advances in Neural Information Processing Systems 36 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Yuwei Liu, Junquan Deng, Xiangkun Jia, Yanhao Wang, Minghua Wang, Lin Huang, Tao Wei, and Purui Su. 2025. PromeFuzz: A Knowledge-Driven Approach to Fuzzing Harness Generation with Large Language Models. In Proceedings of the 2025 ACMSIGSAC Conference on Computer and Communications Security. 1559–1573.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Chenyang Lyu, Shouling Ji, Chao Zhang, Yuwei Li, Wei-Han Lee, Yu Song, and Raheem Beyah. 2019. MOPT: Optimized mutation scheduling for fuzzers. In 28th Conference’17, July 2017, Washington, DC, USA Vasudev Vikram and Rohan Padhye USENIX Security Symposium (USENIX Security 19). 1949–1966. https://doi.org/10. 5555/3361338.3361473",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Yunlong Lyu, Yuxuan Xie, Peng Chen, and Hao Chen. 2023. Prompt Fuzzing for Fuzz Driver Generation. arXiv preprint arXiv:2312.17677 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Muhammad Maaz, Liam DeVoe, Zac Hatfield-Dodds, and Nicholas Carlini. 2025. Agentic Property-Based Testing: Finding Bugs Across the Python Ecosystem. arXiv preprint arXiv:2510.09907 (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Valentin Jean Marie Manès, HyungSeok Han, Choongwoo Han, Sang Kil Cha, Manuel Egele, Edward J Schwartz, and Maverick Woo. 2019. The art, science, and engineering of fuzzing: A survey. IEEE Transactions on Software Engineering (2019). https://doi.org/10.1109/TSE.2019.2946563",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Ruijie Meng, Martin Mirchev, Marcel Böhme, and Abhik Roychoudhury. 2024. Large language model guided protocol fuzzing. In Proceedings of the 31st Annual Network and Distributed System Security Symposium (NDSS).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Barton P. Miller, Louis Fredriksen, and Bryan So. 1990. An Empirical Study of the Reliability of UNIX Utilities. Commun. ACM 33, 12 (dec 1990), 32–44. https://doi.org/10.1145/96267.96279",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Hoang Lam Nguyen and Lars Grunske. 2022. Bedivfuzz: Integrating behavioral diversity into generator-based fuzzing. In Proceedings of the 44th International Conference on Software Engineering. 249–261.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Rohan Padhye, Caroline Lemieux, and Koushik Sen. 2019. JQF: Coverage-guided Property-based Testing in Java. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis (Beijing, China) (ISSTA’19). Association for Computing Machinery, New York, NY, USA, 398–401. https://doi.org/10.1145/3293882.3339002",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Mike Papadakis, and Yves Le Traon. 2019. Semantic Fuzzing with Zest. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis (Beijing, China) (ISSTA 2019). ACM, 329–340. https://doi.org/10.1145/3293882.3330576",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Rohan Padhye, Koushik Sen, and Paul N. Hilfinger. 2019. ChocoPy: A Programming Language for Compilers Courses. In Proceedings of the 2019 ACMSIGPLAN Symposium on SPLASH-E (Athens, Greece) (SPLASH-E 2019). Association for Computing Machinery, 41–45. https://doi.org/10.1145/3358711.3361627",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Nikitha Rao, Elizabeth Gilbert, Tahina Ramananandro, Nikhil Swamy, Claire Le Goues, and Sarah Fakhoury. 2024. DiffSpec: Differential Testing with LLMs using Natural Language Specifications and Code Artifacts. arXiv preprint arXiv:2410.04249 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Nikitha Rao, Kush Jain, Uri Alon, Claire Le Goues, and Vincent J Hellendoorn. 2023. CAT-LM training language models on aligned code and tests. In 2023 38th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 409–420.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Alexandre Rebert, Sang Kil Cha, Thanassis Avgerinos, Jonathan Foote, David Warren, Gustavo Grieco, and David Brumley. 2014. Optimizing seed selection for fuzzing. In 23rd USENIX Security Symposium (USENIX Security 14). 861–875. https://doi.org/10.5555/2671225.2671280",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "John Regehr, Yang Chen, Pascal Cuoq, Eric Eide, Chucky Ellison, and Xuejun Yang. 2012. Test-Case Reduction for C Compiler Bugs. In Proceedings of the 33rd ACMSIGPLAN Conference on Programming Language Design and Implementation (Beijing, China) (PLDI ’12). Association for Computing Machinery, New York, NY, USA, 335–346. https://doi.org/10.1145/2254064.2254104",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Max Schäfer, Sarah Nadi, Aryaz Eghbali, and Frank Tip. 2023. An empirical evaluation of using large language models for automated unit test generation. IEEE Transactions on Software Engineering (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Max Schäfer, Sarah Nadi, Aryaz Eghbali, and Frank Tip. 2023. Adaptive Test Generation Using a Large Language Model. arXiv:2302.06527 [cs.SE]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Vasudev Vikram, Isabella Laybourn, Ao Li, Nicole Nair, Kelton OBrien, Rafaello Sanna, and Rohan Padhye. 2023. Guiding Greybox Fuzzing with Mutation Testing. InProceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (, Seattle, WA, USA,) (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 929–941. https://doi.org/10.1145/3597926.3598107",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Vasudev Vikram, Caroline Lemieux, Joshua Sunshine, and Rohan Padhye. 2023. Can large language models write good property-based tests? arXiv preprint arXiv:2307.04346 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Vasudev Vikram, Rohan Padhye, and Koushik Sen. 2021. Growing A Test Corpus with Bonsai Fuzzing. In 43rd IEEE/ACM International Conference on Software Engineering, ICSE 2021, Madrid, Spain, 22-30 May 2021. IEEE, 723–735. https: //doi.org/10.1109/ICSE43902.2021.00072",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Xingyao Wang, Boxuan Li, Yufan Song, Frank F Xu, Xiangru Tang, Mingchen Zhuge, Jiayi Pan, Yueqi Song, Bowen Li, Jaskirat Singh, et al .2024. OpenHands: An Open Platform for AI Software Developers as Generalist Agents. arXiv preprint arXiv:2407.16741 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Mingyuan Wu, Ling Jiang, Jiahong Xiang, Yanwei Huang, Heming Cui, Lingming Zhang, and Yuqun Zhang. 2022. One Fuzzing Strategy to Rule Them All. In Proceedings of the 44th International Conference on Software Engineering (Pittsburgh, Pennsylvania) (ICSE ’22). Association for Computing Machinery, New York, NY, USA, 1634–1645. https://doi.org/10.1145/3510003.3510174",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Chunqiu Steven Xia, Matteo Paltenghi, Jia Le Tian, Michael Pradel, and Lingming Zhang. 2024. Fuzz4all: Universal fuzzing with large language models. Proc.IEEE/ACMICSE (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Chunqiu Steven Xia, Zhe Wang, Yan Yang, Yuxiang Wei, and Lingming Zhang. 2025. Live-SWE-agent: Can Software Engineering Agents Self-Evolve on the Fly? arXiv:2511.13646 [cs.SE] https://arxiv.org/abs/2511.13646",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Hanxiang Xu, Wei Ma, Ting Zhou, Yanjie Zhao, Kai Chen, Qiang Hu, Yang Liu, and Haoyu Wang. 2024. CKGFuzzer: LLM-Based Fuzz Driver Generation Enhanced By Code Knowledge Graph. arXiv preprint arXiv:2411.11532 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Chenyuan Yang, Yinlin Deng, Runyu Lu, Jiayi Yao, Jiawei Liu, Reyhaneh Jabbarvand, and Lingming Zhang. 2023. WhiteFox: White-Box Compiler Fuzzing Empowered by Large Language Models. arXiv preprint arXiv:2310.15991 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "John Yang, Carlos E Jimenez, Alexander Wettig, Kilian Lieret, Shunyu Yao, Karthik Narasimhan, and Ofir Press. 2024. Swe-agent: Agent-computer interfaces enable automated software engineering. arXiv preprint arXiv:2405.15793 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and Understanding Bugs in C Compilers. In Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI ’11).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Michal Zalewski. 2014. American Fuzzy Lop. https://lcamtuf.coredump.cx/afl/. Accessed February 11, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Cen Zhang, Mingqiang Bai, Yaowen Zheng, Yeting Li, Xiaofei Xie, Yuekang Li, Wei Ma, Limin Sun, and Yang Liu. 2023. Understanding large language model based fuzz driver generation. arXiv preprint arXiv:2307.12469 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Kunpeng Zhang, Zongjie Li, Daoyuan Wu, Shuai Wang, and Xin Xia. 2025. LowCost and Comprehensive Non-textual Input Fuzzing with LLM-Synthesized Input Generators. In 34th USENIX Security Symposium (USENIX Security 25). 6999–7018.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 8,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In (ICSE). IEEE, 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Tools such as CSmith [ 52,65] for C compilers and SQLancer [ 8] for database engines generate inputs that are not merely syntactically valid but semantically rich by construction: CSmith emits C programs free of undefined behavior, while SQLancer produces queries that satisfy the relational schemas and type rules of the target database.",
+      "context_before": "is possible by treating generators as parametric decoders: functions that consume an arbitrary byte sequence to produce a structured input, so that byte-level mutations from a coverage-guided fuzzer translate directly into structured mutations on the generated input without requiring specialized mutators.arXiv:2604.01442v1 [cs.SE] 1 Apr 2026 Conference’17, July 2017, Washington, DC, USA Vasudev Vikram and Rohan Padhye A complementary line of work focuses on building heavyweight, highly customized, domain-specific generators that encode hard input constraints directly into the generation logic.",
+      "context_after": "By encoding deep input and domain logic into the generators, these tools routinely expose bugs that generic mutation-based fuzzers miss entirely, because the bugs only manifest on inputs that satisfy constraints random mutations are less likely to produce. Notably, such heavyweight generators do notemploy coverage guidance nor mutation strategies, since the generation logic itself is strong enough that neither is necessary. The limitation, however, is that such generators require substantial manual effort to write and are tightly coupled to a single target. With the rise of large language mod",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 4275,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "This paper asks whether an AI coding agent can synthesise the domain-specific input generators that generator-based fuzzing relies on, and whether such generators make coverage guidance and mutation unnecessary. Gentoo gives an LLM agent terminal access and the source of the fuzz target, and has it iteratively synthesise and refine a generator. Across seven Java libraries, agent-synthesised generators beat human-written ones on branch coverage in four.",
+    "narrative": "One citation, naming SQLancer as the database-engine counterpart to CSmith for C compilers: a generator producing inputs that are more than syntactically valid.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2604_03024.json b/_data/papers/paper_arxiv_2604_03024.json
new file mode 100644
index 0000000..ea8c002
--- /dev/null
+++ b/_data/papers/paper_arxiv_2604_03024.json
@@ -0,0 +1,670 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:28:59Z",
+  "paper": {
+    "id": "paper:arxiv:2604.03024",
+    "title": "BugForge: Constructing and Utilizing DBMS Bug Repository to Enhance DBMS Testing",
+    "authors": [
+      "Dawei Li",
+      "Qifan Liu",
+      "Yuxiao Guo",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Chi Zhang",
+      "Jingzhou Fu",
+      "H. Mao",
+      "Zhenyu Guan",
+      "Yu Jiang"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2604.03024",
+    "s2_paper_id": "597ba93345ffe08175897ddb2019c676914992af",
+    "url": "https://arxiv.org/abs/2604.03024",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2604.03024",
+      "retrieved_at": "2026-09-09T01:28:59Z",
+      "chars": 82560,
+      "content_sha256": "sha256:6deac0cab9c6ea91094cae7db5e4c67d0a5a76c14c968481c3f3be5cdf9bb6dd"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2012
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND ANDMOTIVATION",
+        "start": 9968
+      },
+      {
+        "number": "III",
+        "title": "DESIGN",
+        "start": 17768
+      },
+      {
+        "number": "A",
+        "title": "Bug Repository Construction",
+        "start": 18343
+      },
+      {
+        "number": "B",
+        "title": "Bug Repository Utilization",
+        "start": 30959
+      },
+      {
+        "number": "IV",
+        "title": "EVALUATION",
+        "start": 41952
+      },
+      {
+        "number": "A",
+        "title": "Implementation and Evaluation Setup",
+        "start": 42437
+      },
+      {
+        "number": "B",
+        "title": "Assessment of BugForge Bug Repository",
+        "start": 44956
+      },
+      {
+        "number": "C",
+        "title": "Bugs Detected With BugForge",
+        "start": 48540
+      },
+      {
+        "number": "D",
+        "title": "Comparison of State-of-the-art DBMS Testing Tools.",
+        "start": 59950
+      },
+      {
+        "number": "E",
+        "title": "Comparison of Test Case Quality",
+        "start": 62329
+      },
+      {
+        "number": "V",
+        "title": "DISCUSSION",
+        "start": 65950
+      },
+      {
+        "number": "VI",
+        "title": "RELATEDWORK",
+        "start": 69685
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 73759
+      },
+      {
+        "number": "P",
+        "title": "Bian, and L. Zhang, “Bugbuilder: An automated approach to building",
+        "start": 79286
+      },
+      {
+        "number": "T",
+        "title": "F. Bissyand ´e, “Enriching automatic test case generation by extracting",
+        "start": 79528
+      },
+      {
+        "number": "J",
+        "title": "Klein, and Y. Le Traon, “ibir: Bug-report-driven fault injection,”",
+        "start": 80827
+      },
+      {
+        "number": "J",
+        "title": "Wang, and J. Li, “Sequence-oriented dbms fuzzing,” in. IEEE.",
+        "start": 81239
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“databases,” https://en.wikipedia.org/wiki/Database, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "T. P. G. D. Group, “pgsql-bugs,” https://www.postgresql.org/list/pgsqlbugs/, 9 2025, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“Mysql bug home,” https://bugs.mysql.com/, 9 2025, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "M. Rigger, “Bugs found in database management systems,” https: //www.manuelrigger.at/dbms-bugs, accessed: April 6, 2026.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "fuboat, “Buglist-monetdb-95dba4e85799,” https://github.com/fuboat/ BugList-MonetDB-95dba4e85799, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "ChijinZ, “Security advisories,” https://github.com/ChijinZ/security advisories, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "R. Just, D. Jalali, and M. D. Ernst, “Defects4j: a database of existing faults to enable controlled testing studies for java programs,” in Proceedings of the 2014 International Symposium on Software Testing and Analysis, ser. ISSTA 2014. New York, NY, USA: Association for Computing Machinery, 2014, p. 437–440. [Online]. Available: https://doi.org/10.1145/2610384.2628055",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "V. Dallmeier and T. Zimmermann, “Extraction of bug localization benchmarks from history,” inProceedings of the 22nd IEEE/ACM International Conference on Automated Software Engineering, ser. ASE ’07. New York, NY, USA: Association for Computing Machinery, 2007, p. 433–436. [Online]. Available: https://doi.org/ 10.1145/1321631.1321702",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "MySQL, “Issue,” https://bugs.mysql.com/bug.php?id=102205, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "G. Cloud, “Gemini 2.5 flash,” https://cloud.google.com/vertex-ai/ generative-ai/docs/models/gemini/2-5-flash, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "——, “Gemini 2.5 pro,” https://cloud.google.com/vertex-ai/generativeai/docs/models/gemini/2-5-pro, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "“Mysql,” https://www.mysql.com/, 1 2024, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "“Postgresql,” https://www.postgresql.org/, 1 2024, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "“Mariadb,” https://mariadb.org/, 1 2024, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "“The database system to speed up your analytical jobs,” https:// www.monetdb.org/, 1 2024, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "M. Rigger, “Sqlancer website,” https://github.com/sqlancer/sqlancer, accessed: April 6, 2026.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 17,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear),” inProceedings of the 46th International Conference on Very Large Data Bases (VLDB), Tokyo, Japan, aug 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” inProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 19,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammarfree dbms fuzzing,” inConference on Automated Software Engineering (ASE’22), 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "J. Fu, J. Liang, Z. Wu, and Y. Jiang, “Sedar: Obtaining highquality seeds for DBMS fuzzing via cross-dbms SQL transfer,” inProceedings of the 46th IEEE/ACM International Conference on Software Engineering, ICSE 2024, Lisbon, Portugal, April 1420, 2024. ACM, 2024, pp. 146:1–146:12. [Online]. Available: https://doi.org/10.1145/3597503.3639210",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "A. Fioraldi, D. Maier, H. Eißfeldt, and M. Heuse, “American fuzzy lop plus plus (afl++) - using multiple cores,” https://github.com/AFLplusplus/AFLplusplus/tree/ 7bcd4e290111ca81d6d58d1b70696e9e9aaa5ac1\\#b-using-multiplecores, accessed: April 6, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "PostgreSQL, “BUG #19382: Server crash at nssdatabase lookup,” https://www.postgresql.org/ message-id/CALdSSPiG3GZgdBROiAguqdSSZzB4\\ %3DCS5UrqLPenV0XPgSEmszw\\%40mail.gmail.com, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing database management systems with language validity and coverage feedback,” inThe ACM Conference on Computer and Communications Security (CCS), 2020, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "J. Anvik, L. Hiew, and G. C. Murphy, “Coping with an open bug repository,” inProceedings of the 2005 OOPSLA Workshop on Eclipse Technology EXchange, ser. eclipse ’05. New York, NY, USA: Association for Computing Machinery, 2005, p. 35–39. [Online]. Available: https://doi.org/10.1145/1117696.1117704",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "T. Zhang, G. Yang, B. Lee, and A. T. S. Chan, “Predicting severity of bug report by mining bug repository with concept profile,” inProceedings of the 30th Annual ACM Symposium on Applied Computing, ser. SAC ’15. New York, NY, USA: Association for Computing Machinery, 2015, p. 1553–1558. [Online]. Available: https://doi.org/10.1145/2695664.2695872",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "C. Williams and J. Hollingsworth, “Automatic mining of source code repositories to improve bug finding techniques,”IEEE Transactions on Software Engineering, vol. 31, no. 6, pp. 466–480, 2005.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Y. Jiang, H. Liu, X. Luo, Z. Zhu, X. Chi, N. Niu, Y. Zhang, Y. Hu, P. Bian, and L. Zhang, “Bugbuilder: An automated approach to building bug repository,”IEEE Transactions on Software Engineering, vol. 49, no. 4, pp. 1443–1463, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "W. C. Ou ´edraogo, L. Plein, K. Kabore, A. Habib, J. Klein, D. Lo, and T. F. Bissyand ´e, “Enriching automatic test case generation by extracting relevant test inputs from bug reports,”Empirical Software Engineering, vol. 30, no. 3, p. 85, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "H. Zhong, “Enriching compiler testing with real program from bug report,” inProceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, ser. ASE ’22. New York, NY, USA: Association for Computing Machinery, 2023. [Online]. Available: https://doi.org/10.1145/3551349.3556894",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "X. Han, T. Yu, and D. Lo, “Perflearner: learning from bug reports to understand and generate performance test frames,” inProceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering, ser. ASE ’18. New York, NY, USA: Association for Computing Machinery, 2018, p. 17–28. [Online]. Available: https://doi.org/10.1145/3238147.3238204",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "M. Fazzini, M. Prammer, M. d’Amorim, and A. Orso, “Automatically translating bug reports into test cases for mobile apps,” inProceedings of the 27th ACMSIGSOFT International Symposium on Software Testing and Analysis, ser. ISSTA 2018. New York, NY, USA: Association for Computing Machinery, 2018, p. 141–152. [Online]. Available: https://doi.org/10.1145/3213846.3213869",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "A. Khanfir, A. Koyuncu, M. Papadakis, M. Cordy, T. F. Bissyand ´e, J. Klein, and Y. Le Traon, “ibir: Bug-report-driven fault injection,” ACM Trans. Softw. Eng. Methodol., vol. 32, no. 2, Mar. 2023. [Online]. Available: https://doi.org/10.1145/3542946",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "A. Seltenreich, B. Tang, and S. Mullender, “Sqlsmith: a random sql query generator,” 2018. [Online]. Available: https://github.com/anse1/ sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "J. Liang, Y. Chen, Z. Wu, J. Fu, M. Wang, Y. Jiang, X. Huang, T. Chen, J. Wang, and J. Li, “Sequence-oriented dbms fuzzing,” in. IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting Logical Bugs of DBMS with Coverage-based Guidance,” inProceedings of the 31st USENIX Security Symposium (USENIX 2022), Boston, MA, aug 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "R. A. Haraty, N. Mansour, and B. Daou, “Regression testing of database applications,” inProceedings of the 2001 ACM Symposium on Applied Computing, ser. SAC ’01. New York, NY, USA: Association for Computing Machinery, 2001, p. 285–289. [Online]. Available: https://doi.org/10.1145/372202.372342",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "E. Lo, C. Binnig, D. Kossmann, M. Tamer ¨Ozsu, and W.-K. Hon, “A framework for testing dbms features,”The VLDB Journal, vol. 19, no. 2, pp. 203–230, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "E. Rogstad, L. Briand, and R. Torkar, “Test case selection for black-box regression testing of database applications,”Information and Software technology, vol. 55, no. 10, pp. 1781–1795, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "S. Zhong and M. Rigger, “Understanding and reusing test suites across database systems,” 2, no. 6, Dec. 2024. [Online]. Available: https://doi.org/10.1145/3698829",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 40,
+      "text": "H. Gavriilidis, L. Rose, J. Ziegler, K. Beedkar, J.-A. Quian ´e-Ruiz, and V. Markl, “Xdb in action: decentralized cross-database query processing for black-box dbmses,”Proceedings of the VLDB Endowment, vol. 16, no. 12, pp. 4078–4081, 2023.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 4,
+      "text": "M. Rigger, “Bugs found in database management systems,” https: //www.manuelrigger.at/dbms-bugs, accessed: April 6, 2026.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 16,
+      "text": "M. Rigger, “Sqlancer website,” https://github.com/sqlancer/sqlancer, accessed: April 6, 2026.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 18,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” inProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 39,
+      "text": "S. Zhong and M. Rigger, “Understanding and reusing test suites across database systems,” 2, no. 6, Dec. 2024. [Online]. Available: https://doi.org/10.1145/3698829",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[4]–[6]",
+      "technique": null,
+      "sentence": "Test researchers, on the other hand, often disclose bugs in public communities and further organize them into structured datasets for analysis, testing, and methodology validation [4]–[6].",
+      "context_before": "rs can use this information as seeds for fuzzing campaigns, as robust regression test cases, and as standard inputs for database evaluation. Moreover, the structured records facilitate fault localization, program repair, and defect prediction by making recurring patterns and common error triggers easier to detect. Practically, both DBMS vendors and testing researchers have developed mechanisms to manage DBMS bugs. Database vendors typically collect, verify, and categorize user-submitted issues through official bug reporting platforms (e.g., MySQL Bug System [3] and PostgreSQL Bug Tracker [2]).",
+      "context_after": "These efforts improve the transparency and security responsiveness within the DBMS ecosystem. However, existing DBMS bug management remains incomplete and fragmented, limiting the full utilization of bug-related data.First, bug reports are often incomplete or ambiguous, since many are reported informally by users, making it difficult for developers or downstream vendors to determine whether issues are true faults or false positives. Second, most repositories focus on logging bugs rather than enabling structured analysis or automated processing. The lack of standardized formats and interfaces,",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 4177,
+      "cited_reference": {
+        "number": 4,
+        "text": "M. Rigger, “Bugs found in database management systems,” https: //www.manuelrigger.at/dbms-bugs, accessed: April 6, 2026.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[16]–[20]",
+      "technique": null,
+      "sentence": "In addition, these DBMSs have long served as standard subjects in prior DBMS testing research [16]–[20].",
+      "context_before": "mponents for PoC extraction, adaptation, and seed smelting. For the LLM-assisted stages, JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 8 we employ Gemini-2.5-flash [10] for raw PoC extraction and Gemini-2.5-pro [11] for the more complex adaptation tasks. Tested DBMSs.To evaluate the effectiveness and applicability ofBugForge, we select four representative open-source DBMSs as evaluation targets: MySQL [12], PostgreSQL [13], MariaDB [14], and MonetDB [15]. These systems are widely used in research and industry and span diverse query engines, storage models, and system architectures.",
+      "context_after": "Basic Setup.The evaluation was conducted on a server running 64-bit Ubuntu 22.04.5 LTS, equipped with an AMDEPYC 7763 64-core processor (128 threads) and 500 GB of main memory. For DBMS fuzzing and cross-DBMS testing, the targeted DBMSs were compiled with AFL++ [21] to enable testing support. We ran the DBMS fuzzers under their default configurations together with their corresponding sets of initial seeds. Each fuzzing instance was executed for 24 hours on the target DBMS. For regression testing, we selected the latest available version as well as the fixed versions supported by each target D",
+      "section": "A Implementation and Evaluation Setup",
+      "page": 8,
+      "char_offset": 43936,
+      "cited_reference": {
+        "number": 16,
+        "text": "M. Rigger, “Sqlancer website,” https://github.com/sqlancer/sqlancer, accessed: April 6, 2026.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For SQLancer, we enable its FUZZ mode for testing.",
+      "context_before": "type confusion in dependencies that are rarely covered by isolated fuzzing tools. This case underscoresBugForge’s ability to detect hidden bugs through cross-DBMS testing, which could detect hidden bugs within bug PoCs from other DBMS. D. Comparison of State-of-the-art DBMS Testing Tools. To further evaluate the practical effectiveness ofBugForge, we compare it with two representative DBMS testing tools, SQLancerandSQLsmith. ForBugForge, we use the configuration where Griffin is seeded with both built-in unit test cases and the high-quality test cases generated fromBugForge (i.e., GRIFFINU+B).",
+      "context_after": "For SQLsmith, we use its default version to test MonetDB and PostgreSQL, and adopt its MySQL branch to test MySQL and MariaDB. Since SQLancer does not support MonetDB, the total is computed only over the three supported DBMSs, i.e., MySQL, MariaDB, and PostgreSQL. TABLE V: Comparison of Typical DBMS Testing Tools. DBMSCovered Branches Detected Bugs SQLancer SQLsmith BugForge SQLancer SQLsmith BugForge MySQL 106,971 96,311 209,720 5 3 8 MariaDB 46,298 30,896 140,677 4 2 7 PostgreSQL 48,030 50,285 90,763 2 1 3 MonetDB N/A 33,209 83,420 N/A 2 5 Total†201,299 177,492 441,160 11 6 18 Table V prese",
+      "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+      "page": 10,
+      "char_offset": 60314,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Since SQLancer does not support MonetDB, the total is computed only over the three supported DBMSs, i.",
+      "context_before": "could detect hidden bugs within bug PoCs from other DBMS. D. Comparison of State-of-the-art DBMS Testing Tools. To further evaluate the practical effectiveness ofBugForge, we compare it with two representative DBMS testing tools, SQLancerandSQLsmith. ForBugForge, we use the configuration where Griffin is seeded with both built-in unit test cases and the high-quality test cases generated fromBugForge (i.e., GRIFFINU+B). For SQLancer, we enable its FUZZ mode for testing. For SQLsmith, we use its default version to test MonetDB and PostgreSQL, and adopt its MySQL branch to test MySQL and MariaDB.",
+      "context_after": "e., MySQL, MariaDB, and PostgreSQL. TABLE V: Comparison of Typical DBMS Testing Tools. DBMSCovered Branches Detected Bugs SQLancer SQLsmith BugForge SQLancer SQLsmith BugForge MySQL 106,971 96,311 209,720 5 3 8 MariaDB 46,298 30,896 140,677 4 2 7 PostgreSQL 48,030 50,285 90,763 2 1 3 MonetDB N/A 33,209 83,420 N/A 2 5 Total†201,299 177,492 441,160 11 6 18 Table V presents the comparison results. Overall,BugForge outperforms both SQLancer and SQLsmith in branch coverage across all evaluated DBMSs. Specifically,BugForgecovers 209,720 branches on MySQL, 140,677 on MariaDB, 90,763 on PostgreSQL, an",
+      "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+      "page": 10,
+      "char_offset": 60492,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "DBMSCovered Branches Detected Bugs SQLancer SQLsmith BugForge SQLancer SQLsmith BugForge MySQL 106,971 96,311 209,720 5 3 8 MariaDB 46,298 30,896 140,677 4 2 7 PostgreSQL 48,030 50,285 90,763 2 1 3 MonetDB N/A 33,209 83,420 N/A 2 5 Total†201,299 177,492 441,160 11 6 18 Table V presents the comparison results.",
+      "context_before": "h two representative DBMS testing tools, SQLancerandSQLsmith. ForBugForge, we use the configuration where Griffin is seeded with both built-in unit test cases and the high-quality test cases generated fromBugForge (i.e., GRIFFINU+B). For SQLancer, we enable its FUZZ mode for testing. For SQLsmith, we use its default version to test MonetDB and PostgreSQL, and adopt its MySQL branch to test MySQL and MariaDB. Since SQLancer does not support MonetDB, the total is computed only over the three supported DBMSs, i.e., MySQL, MariaDB, and PostgreSQL. TABLE V: Comparison of Typical DBMS Testing Tools.",
+      "context_after": "Overall,BugForge outperforms both SQLancer and SQLsmith in branch coverage across all evaluated DBMSs. Specifically,BugForgecovers 209,720 branches on MySQL, 140,677 on MariaDB, 90,763 on PostgreSQL, and 83,420 on MonetDB, all of which are substantially higher than the corresponding results of the other two tools. Over the three commonly supported DBMSs,BugForge covers 441,160 branches in total, compared with 201,299 for SQLancer and 177,492 for SQLsmith. These results suggest that the high-quality test cases generated from DBMS bug reports can guide testing toward deeper and broader executio",
+      "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+      "page": 10,
+      "char_offset": 60681,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Overall,BugForge outperforms both SQLancer and SQLsmith in branch coverage across all evaluated DBMSs.",
+      "context_before": "efault version to test MonetDB and PostgreSQL, and adopt its MySQL branch to test MySQL and MariaDB. Since SQLancer does not support MonetDB, the total is computed only over the three supported DBMSs, i.e., MySQL, MariaDB, and PostgreSQL. TABLE V: Comparison of Typical DBMS Testing Tools. DBMSCovered Branches Detected Bugs SQLancer SQLsmith BugForge SQLancer SQLsmith BugForge MySQL 106,971 96,311 209,720 5 3 8 MariaDB 46,298 30,896 140,677 4 2 7 PostgreSQL 48,030 50,285 90,763 2 1 3 MonetDB N/A 33,209 83,420 N/A 2 5 Total†201,299 177,492 441,160 11 6 18 Table V presents the comparison results.",
+      "context_after": "Specifically,BugForgecovers 209,720 branches on MySQL, 140,677 on MariaDB, 90,763 on PostgreSQL, and 83,420 on MonetDB, all of which are substantially higher than the corresponding results of the other two tools. Over the three commonly supported DBMSs,BugForge covers 441,160 branches in total, compared with 201,299 for SQLancer and 177,492 for SQLsmith. These results suggest that the high-quality test cases generated from DBMS bug reports can guide testing toward deeper and broader execution paths than purely generator-based approaches. Detected Bugs.As shown in Table V,BugForgedetected 8, 7",
+      "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+      "page": 10,
+      "char_offset": 60992,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Over the three commonly supported DBMSs,BugForge covers 441,160 branches in total, compared with 201,299 for SQLancer and 177,492 for SQLsmith.",
+      "context_before": "ted Bugs SQLancer SQLsmith BugForge SQLancer SQLsmith BugForge MySQL 106,971 96,311 209,720 5 3 8 MariaDB 46,298 30,896 140,677 4 2 7 PostgreSQL 48,030 50,285 90,763 2 1 3 MonetDB N/A 33,209 83,420 N/A 2 5 Total†201,299 177,492 441,160 11 6 18 Table V presents the comparison results. Overall,BugForge outperforms both SQLancer and SQLsmith in branch coverage across all evaluated DBMSs. Specifically,BugForgecovers 209,720 branches on MySQL, 140,677 on MariaDB, 90,763 on PostgreSQL, and 83,420 on MonetDB, all of which are substantially higher than the corresponding results of the other two tools.",
+      "context_after": "These results suggest that the high-quality test cases generated from DBMS bug reports can guide testing toward deeper and broader execution paths than purely generator-based approaches. Detected Bugs.As shown in Table V,BugForgedetected 8, 7, 3, and 5 bugs on MySQL, MariaDB, PostgreSQL, and MonetDB, respectively. Over the three jointly supported DBMSs except MonetDB,BugForgedetected 18 bugs in total, compared with 11 for SQLancer and 6 for SQLsmith. This advantage is partly attributed to the high-quality test cases provided byBugForge, which were derived from DBMS JOURNAL OF LATEX CLASS FILE",
+      "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+      "page": 10,
+      "char_offset": 61308,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Over the three jointly supported DBMSs except MonetDB,BugForgedetected 18 bugs in total, compared with 11 for SQLancer and 6 for SQLsmith.",
+      "context_before": "B, 90,763 on PostgreSQL, and 83,420 on MonetDB, all of which are substantially higher than the corresponding results of the other two tools. Over the three commonly supported DBMSs,BugForge covers 441,160 branches in total, compared with 201,299 for SQLancer and 177,492 for SQLsmith. These results suggest that the high-quality test cases generated from DBMS bug reports can guide testing toward deeper and broader execution paths than purely generator-based approaches. Detected Bugs.As shown in Table V,BugForgedetected 8, 7, 3, and 5 bugs on MySQL, MariaDB, PostgreSQL, and MonetDB, respectively.",
+      "context_after": "This advantage is partly attributed to the high-quality test cases provided byBugForge, which were derived from DBMS JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 11 bug reports and thus retain richer bug-relevant structures and historical failure patterns. Overall, the results suggest that leveraging adapted real-world bug reports can serve as a highquality complement to existing DBMS testing techniques. E. Comparison of Test Case Quality To evaluate the quality of the test cases adapted byBugForge, we leverage its high-quality cases as initial seeds for GRIFFIN[19] and SQUIRREL[",
+      "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+      "page": 10,
+      "char_offset": 61768,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[39]",
+      "technique": null,
+      "sentence": "Mainstream techniques for database testing include fuzzing [19], [23], [33]–[35], regression testing [17], [36]–[38] and cross-DBMS testing [20], [39], [40].",
+      "context_before": "bug reports from mobile applications into executable UI test cases. IBIR [32] injects more realistic faults, similar to those found in the real world, by reversing the code transformation templates from a bug-report-driven automated program repair system to enhance software testing capabilities. However, applying these methods to DBMSs is challenging due to varying SQL dialects and numerous version iterations. To address this,BugForgeemploys a feedback-driven mechanism to perform tailored analysis, enabling the effective handling of complex bug reports across different dialects. DBMS Testing.",
+      "context_after": "Fuzzing is applied to test DBMS and discover new bugs. SQLsmith [33] generates test cases from predefined abstract syntax tree templates. LEGO [34] tests databases by compiling and recombining sequences of SQL queries to generate new ones. Squirrel [23] and Griffin [19] apply mutation strategies to randomly alter initial seeds, generating diverse test cases to uncover DBMS vulnerabilities. However, their effectiveness is highly dependent on the quality of the initial seeds. Regression testing is employed to identify bugs between different database releases. Apollo [17] automatically detects p",
+      "section": "VI RELATEDWORK",
+      "page": 12,
+      "char_offset": 72295,
+      "cited_reference": {
+        "number": 39,
+        "text": "S. Zhong and M. Rigger, “Understanding and reusing test suites across database systems,” 2, no. 6, Dec. 2024. [Online]. Available: https://doi.org/10.1145/3698829",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[39]",
+      "technique": null,
+      "sentence": "[39] uncover new bugs by integrating and reusing test suites from various open-source DBMSs for cross-testing.",
+      "context_before": "DBMS vulnerabilities. However, their effectiveness is highly dependent on the quality of the initial seeds. Regression testing is employed to identify bugs between different database releases. Apollo [17] automatically detects performance regression bugs in database systems using feedback-driven techniques and query minimization. Cross-DBMS testing leverages test cases from different database systems to discover new bugs or verify the correctness of SQL queries. Sedar [20] employs this technique by transferring test cases between databases to construct a high-quality seed corpus. Zhong et al.",
+      "context_after": "Unlike existing tools that rely on inefficient random inputs, BugForgesystematically constructs a repository from realworld bug reports. This approach provides high-quality test cases that dramatically improve DBMS testing. VII. CONCLUSION In this paper, we proposeBugForge, a framework that constructs and utilizes DBMS bug repositories to enhance DBMS testing.BugForgecollects reports, mines bug information with syntax-aware processing and input-adaptive LLM framework, and generates high-quality test cases through semantic-guided adaptation, creating a unified resource for DBMS testing. Totall",
+      "section": "VI RELATEDWORK",
+      "page": 12,
+      "char_offset": 73423,
+      "cited_reference": {
+        "number": 39,
+        "text": "S. Zhong and M. Rigger, “Understanding and reusing test suites across database systems,” 2, no. 6, Dec. 2024. [Online]. Available: https://doi.org/10.1145/3698829",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:26:20Z",
+    "is_model_written": true,
+    "summary": "BugForge builds standardised DBMS bug repositories from bug reports and turns them into test cases. It collects reports, applies syntax-aware processing and input-adaptive extraction to recover raw proofs of concept, and stores structured metadata alongside them; semantic-guided adaptation then refines these into test cases for fuzzing, regression testing and cross-DBMS discovery. The authors integrated 37,632 reports spanning up to 28 years and found 35 previously unknown bugs.",
+    "narrative": "SQLancer is one of two state-of-the-art tools BugForge is compared against, run in its FUZZ mode, with branch coverage and bug counts reported per DBMS -- 441,160 branches to SQLancer's 201,299, and 18 bugs to its 11 over the three jointly supported systems. MonetDB is excluded from the totals because SQLancer does not support it.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "baseline",
+      "M4": "baseline",
+      "M5": "result_comparison",
+      "M6": "result_comparison",
+      "M7": "result_comparison",
+      "M8": "result_comparison",
+      "M9": "background",
+      "M10": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M3",
+          "M6",
+          "M7",
+          "M8"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "For SQLancer, we enable its FUZZ mode for testing.",
+            "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+            "page": 10
+          },
+          {
+            "mention_id": "M6",
+            "sentence": "Overall,BugForge outperforms both SQLancer and SQLsmith in branch coverage across all evaluated DBMSs.",
+            "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+            "page": 10
+          },
+          {
+            "mention_id": "M7",
+            "sentence": "Over the three commonly supported DBMSs,BugForge covers 441,160 branches in total, compared with 201,299 for SQLancer and 177,492 for SQLsmith.",
+            "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+            "page": 10
+          },
+          {
+            "mention_id": "M8",
+            "sentence": "Over the three jointly supported DBMSs except MonetDB,BugForgedetected 18 bugs in total, compared with 11 for SQLancer and 6 for SQLsmith.",
+            "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+            "page": 10
+          }
+        ],
+        "reasoning": "M3 records the configuration SQLancer was run in, and M6 to M8 give branch coverage and bug counts for each tool across the jointly supported DBMSs."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M3"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "For SQLancer, we enable its FUZZ mode for testing.",
+            "section": "D Comparison of State-of-the-art DBMS Testing Tools.",
+            "page": 10
+          }
+        ],
+        "reasoning": "The comparison section is headed as a comparison of state-of-the-art DBMS testing tools, with SQLancer as one of them."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2605_20473.json b/_data/papers/paper_arxiv_2605_20473.json
new file mode 100644
index 0000000..2f3a08b
--- /dev/null
+++ b/_data/papers/paper_arxiv_2605_20473.json
@@ -0,0 +1,856 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:23:39Z",
+  "paper": {
+    "id": "paper:arxiv:2605.20473",
+    "title": "Code Generation by Differential Test Time Scaling",
+    "authors": [
+      "Yifeng He",
+      "Ethan Wang",
+      "Jicheng Wang",
+      "Xuanxin Ouyang",
+      "Hao Chen"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2605.20473",
+    "s2_paper_id": "bea732652ae1cab45f2ba3678e48cfc22cf4b72a",
+    "url": "https://arxiv.org/abs/2605.20473",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2605.20473",
+      "retrieved_at": "2026-09-08T15:23:39Z",
+      "chars": 91724,
+      "content_sha256": "sha256:2a23d43217ba1bafbf070cd87a2d5a5f94f36fd4a3a7a409254efc062eafe6e8"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 21,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1980
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 7860
+      },
+      {
+        "number": "2.1",
+        "title": "Coverage-guided differential testing",
+        "start": 7873
+      },
+      {
+        "number": "2.1.1",
+        "title": "Differential testing",
+        "start": 7914
+      },
+      {
+        "number": "2.1.2",
+        "title": "Coverage-guided fuzzing",
+        "start": 9948
+      },
+      {
+        "number": "2.2",
+        "title": "Probabilistic sampling",
+        "start": 12309
+      },
+      {
+        "number": "2.2.1",
+        "title": "Temperature",
+        "start": 13132
+      },
+      {
+        "number": "2.2.2",
+        "title": "Beam search",
+        "start": 14068
+      },
+      {
+        "number": "2.3",
+        "title": "On the effects of prompting",
+        "start": 15325
+      },
+      {
+        "number": "3",
+        "title": "Design of DIFFCODEGEN",
+        "start": 16105
+      },
+      {
+        "number": "3.1",
+        "title": "Overview",
+        "start": 16129
+      },
+      {
+        "number": "3.2",
+        "title": "Differential generation",
+        "start": 17455
+      },
+      {
+        "number": "3.2.1",
+        "title": "Differential sampling",
+        "start": 18014
+      },
+      {
+        "number": "3.2.2",
+        "title": "Differential decoding",
+        "start": 18771
+      },
+      {
+        "number": "3.2.3",
+        "title": "Differential prompting",
+        "start": 19247
+      },
+      {
+        "number": "3.3.1",
+        "title": "Coverage-guided fuzzing",
+        "start": 21640
+      },
+      {
+        "number": "3.3.2",
+        "title": "Differential analysis",
+        "start": 25230
+      },
+      {
+        "number": "3.4",
+        "title": "Response selection by clustering",
+        "start": 26037
+      },
+      {
+        "number": "3.4.1",
+        "title": "Normalization of dynamic behavior",
+        "start": 26616
+      },
+      {
+        "number": "3.4.2",
+        "title": "Distance metric",
+        "start": 27850
+      },
+      {
+        "number": "3.4.3",
+        "title": "Clustering algorithm",
+        "start": 29042
+      },
+      {
+        "number": "3.4.4",
+        "title": "Selection strategy",
+        "start": 29786
+      },
+      {
+        "number": "4",
+        "title": "Evaluation",
+        "start": 30606
+      },
+      {
+        "number": "4.1",
+        "title": "Performance gain from different differential methods",
+        "start": 33263
+      },
+      {
+        "number": "4.1.1",
+        "title": "Multi-round evaluation",
+        "start": 34548
+      },
+      {
+        "number": "4.2",
+        "title": "Comparing with test-time scaling methods",
+        "start": 36086
+      },
+      {
+        "number": "4.2.1",
+        "title": "Baselines",
+        "start": 36131
+      },
+      {
+        "number": "4.2.2",
+        "title": "Comparison results",
+        "start": 37041
+      },
+      {
+        "number": "4.3",
+        "title": "Token cost and overhead analysis",
+        "start": 37911
+      },
+      {
+        "number": "4.3.1",
+        "title": "Execution time analysis",
+        "start": 38463
+      },
+      {
+        "number": "4.3.2",
+        "title": "Token usage analysis",
+        "start": 40658
+      },
+      {
+        "number": "4.4",
+        "title": "Ablation studies",
+        "start": 42980
+      },
+      {
+        "number": "4.4.1",
+        "title": "Choosing the number of samples",
+        "start": 43281
+      },
+      {
+        "number": "4.4.2",
+        "title": "Choosing the number of clusters",
+        "start": 44784
+      },
+      {
+        "number": "4.4.3",
+        "title": "Model-based selectors over the differential signal",
+        "start": 49411
+      },
+      {
+        "number": "4.4.4",
+        "title": "Analysis on fuzz driver generation",
+        "start": 51983
+      },
+      {
+        "number": "5",
+        "title": "Discussions",
+        "start": 53264
+      },
+      {
+        "number": "5.1",
+        "title": "Threats to validity",
+        "start": 53278
+      },
+      {
+        "number": "5.2",
+        "title": "Limitations and future work",
+        "start": 55355
+      },
+      {
+        "number": "6",
+        "title": "Related work",
+        "start": 58081
+      },
+      {
+        "number": "6.1",
+        "title": "Code generation models",
+        "start": 58096
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 61947
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Inbal Shani and GitHub Staff.Survey reveals AI’s impact on the developer experience. 2024.URL:https://github. blog/news-insights/research/survey-reveals-ais-impact-on-the-developer-experience/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Kyle Daigle and GitHub Staff.Survey: The AI wave continues to grow on software development teams. 2024.URL:https: //github.blog/news-insights/research/survey-ai-wave-grows/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Mark Chen et al.Evaluating Large Language Models Trained on Code. 2021. arXiv:2107.03374 [cs.LG].URL:https: //arxiv.org/abs/2107.03374. 16 Code Generation by Differential Test Time Scaling",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Xueying Du et al. “Evaluating Large Language Models in Class-Level Code Generation”. In:Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. ICSE ’24. Lisbon, Portugal: Association for Computing Machinery, 2024.ISBN: 9798400702174.DOI:10. 1145 / 3597503. 3639219.URL:https: / / doi. org / 10. 1145 / 3597503. 3639219.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "DeepSeek-AI et al.DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning. 2025. arXiv: 2501.12948 [cs.CL].URL:https://arxiv.org/abs/2501.12948.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Niklas Muennighoff et al. “s1: Simple test-time scaling”. In:Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. Suzhou, China: Association for Computational Linguistics, Nov. 2025, pp. 20275–20321. ISBN: 979-8-89176-332-6.DOI:10.18653/v1/2025.emnlp-main.1025.URL:https://aclanthology.org/2025. emnlp-main.1025/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Charlie Victor Snell, Jaehoon Lee, Kelvin Xu, and Aviral Kumar. “Scaling LLM Test-Time Compute Optimally Can be More Effective than Scaling Parameters for Reasoning”. In:The Thirteenth International Conference on Learning Representations. 2025.URL:https://openreview.net/forum?id=4FWAwZtd2n.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Huaye Zeng et al. “ACECODER: Acing Coder RL via Automated Test-Case Synthesis”. In:Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Vienna, Austria: Association for Computational Linguistics, July 2025, pp. 12023–12040.ISBN: 979-8-89176-251-0.DOI:10.18653/v1/2025.acl-long.587. URL:https://aclanthology.org/2025.acl-long.587/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Dacheng Li et al. “S*: Test Time Scaling for Code Generation”. In:Findings of the Association for Computational Linguistics: EMNLP 2025. Suzhou, China: Association for Computational Linguistics, Nov. 2025, pp. 15964–15978.ISBN: 979-8-89176-335-7.DOI:10.18653/v1/2025.findings-emnlp.865.URL:https://aclanthology.org/2025. findings-emnlp.865/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Xiancai Chen et al. “Revisit Self-Debugging with Self-Generated Tests for Code Generation”. In:Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Vienna, Austria: Association for Computational Linguistics, July 2025, pp. 18003–18023.ISBN: 979-8-89176-251-0.DOI:10.18653/v1/2025.acllong.881.URL:https://aclanthology.org/2025.acl-long.881/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "William M. McKeeman. “Differential Testing for Software”. In:Digit. Tech. J.10 (1998), pp. 100–107.URL:https: //api.semanticscholar.org/CorpusID:14018070.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Robert B. Evans and Alberto Savoia. “Differential testing: a new approach to change detection”. In:The 6th Joint Meeting on European Software Engineering Conference and the ACMSIGSOFT Symposium on the Foundations of Software Engineering: Companion Papers. ESEC-FSE companion ’07. Dubrovnik, Croatia: Association for Computing Machinery, 2007, pp. 549–552.ISBN: 9781595938121.DOI:10.1145/1295014.12950",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Yibiao Yang et al. “Hunting for bugs in code coverage tools via randomized differential testing”. In:Proceedings of the 41st International Conference on Software Engineering. ICSE ’19. Montreal, Quebec, Canada: IEEE Press, 2019, pp. 488–499. DOI:10.1109/ICSE.2019.00061.URL:https://doi.org/10.1109/ICSE.2019.00061.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Shaohua Li and Zhendong Su. “Finding Unstable Code via Compiler-Driven Differential Testing”. In:Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 3. ASPLOS 2023. Vancouver, BC, Canada: Association for Computing Machinery, 2023, pp. 238–251.ISBN: 9781450399180.DOI:10.1145/3582016.3582053.URL:https://doi.org/10.1145",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Naman Jain et al. “LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code”. In: The Thirteenth International Conference on Learning Representations. 2025.URL:https://openreview.net/forum? id=chf JJYC3iL.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. “Finding and understanding bugs in C compilers”. In:SIGPLAN Not.46.6 (June 2011), pp. 283–294.ISSN: 0362-1340.DOI:10.1145/1993316.1993532.URL:https://doi.org/10. 1145/1993316.1993532.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. “Compiler validation via equivalence modulo inputs”. In:SIGPLAN Not.49.6 (June 2014), pp. 216–226.ISSN: 0362-1340.DOI:10.1145/2666356.2594334.URL:https://doi.org/10.1145/ 2666356.2594334.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. “Finding compiler bugs via live code mutation”. In:SIGPLAN Not.51.10 (Oct. 2016), pp. 849–863.ISSN: 0362-1340.DOI:10.1145/3022671.2984038.URL:https://doi.org/10.1145/3022671. 2984038.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Manuel Rigger and Zhendong Su. “Testing Database Engines via Pivoted Query Synthesis”. In:Proc. ACM Program. Lang. 4.OOPSLA (Nov. 2020).DOI:10.1145/3428279.URL:https://doi.org/10.1145/3428279.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Yifeng He, Luning Yang, Christopher Castro Gaw Gonzalo, and Hao Chen. “Evaluating Program Semantics Reasoning with Type Inference in System $F$”. In:The Thirty-ninth Annual Conference on Neural Information Processing Systems Datasets and Benchmarks Track. 2025.URL:https://openreview.net/forum?id=IA9RmaP0aw.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Marcel B ¨ohme, Van-Thuan Pham, and Abhik Roychoudhury. “Coverage-based Greybox Fuzzing as Markov Chain”. In: Proceedings of the 2016 ACMSIGSAC Conference on Computer and Communications Security. CCS ’16. Vienna, Austria: Association for Computing Machinery, 2016, pp. 1032–1043.ISBN: 9781450341394.DOI:10.1145/2976749.2978428. URL:https://doi.org/10.1145/2976749.2978428. 17 Code Generation by Diffe",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. “AFL++: Combining Incremental Steps of Fuzzing Research”. In:14th USENIX Workshop on Offensive Technologies (WOOT 20). USENIX Association, Aug. 2020.URL: https://www.usenix.org/conference/woot20/presentation/fioraldi.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Andreas Zeller, Rahul Gopinath, Marcel B ¨ohme, Gordon Fraser, and Christian Holler. “The Fuzzing Book”. In: (Jan. 2019). DOI:10.60882/cispa.24614928.v1.URL:https://publications.cispa.de/articles/book/The_Fuzzing_ Book/24614928.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Peng Chen, Jianzhong Liu, and Hao Chen. “Matryoshka: Fuzzing Deeply Nested Branches”. In:Proceedings of the 2019 ACMSIGSAC Conference on Computer and Communications Security. CCS ’19. London, United Kingdom: Association for Computing Machinery, 2019, pp. 499–513.ISBN: 9781450367479.DOI:10.1145/3319535.3363225.URL:https: //doi.org/10.1145/3319535.3363225.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Yuting Chen, Ting Su, Chengnian Sun, Zhendong Su, and Jianjun Zhao. “Coverage-directed differential testing of JVM implementations”. In:SIGPLAN Not.51.6 (June 2016), pp. 85–99.ISSN: 0362-1340.URL:https://doi.org/10.1145/ 2980983.2908095.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Theofilos Petsios, Adrian Tang, Salvatore Stolfo, Angelos D. Keromytis, and Suman Jana. “NEZHA: Efficient DomainIndependent Differential Testing”. In:Proceedings of the. SP ’17. San Jose, CA, USA: IEEE Press, 2017, pp. 615–632.ISBN: 9781509049318.DOI:10.1109/SP.2017.27.URL:https: //doi.org/10.1109/SP.2017.27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Shirin Nilizadeh, Yannic Noller, and Corina S. P ˘as˘areanu. “DifFuzz: differential fuzzing for side-channel analysis”. In: Proceedings of the 41st International Conference on Software Engineering. ICSE ’19. Montreal, Quebec, Canada: IEEE Press, 2019, pp. 176–187.ISBN: 9781728108698.DOI:10.1109/ICSE.2019.00122.URL:https://doi.org/10. 1109/ICSE.2019.00122.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Junjie Chen et al. “An empirical comparison of compiler testing techniques”. In:Proceedings of the 38th International Conference on Software Engineering. ICSE ’16. Austin, Texas: Association for Computing Machinery, 2016, pp. 180–190. ISBN: 9781450342056.DOI:10.1145/2884781.2884878.URL:https://doi.org/10.1145/2884781.2884878.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Bei Chen et al. “CodeT: Code Generation with Generated Tests”. In:The Eleventh International Conference on Learning Representations. 2023.URL:https://openreview.net/forum?id=ktrw68Cmu9c.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Yifeng He, Jicheng Wang, Yuyang Rong, and Hao Chen. “FuzzAug: Data Augmentation by Coverage-guided Fuzzing for Neural Test Generation”. In:Findings of the Association for Computational Linguistics: EMNLP 2025. Suzhou, China: Association for Computational Linguistics, Nov. 2025, pp. 15642–15655.ISBN: 979-8-89176-335-7.DOI:10.18653/v1/ 2025.findings-emnlp.847.URL:https://aclanthology.org/2025.findin",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Ari Holtzman et al. “Learning to Write with Cooperative Discriminators”. In:Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Melbourne, Australia: Association for Computational Linguistics, July 2018, pp. 1638–1649.DOI:10.18653/v1/P18-1152.URL:https://aclanthology.org/P18-1152/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Ari Holtzman, Jan Buys, Li Du, Maxwell Forbes, and Yejin Choi. “The Curious Case of Neural Text Degeneration”. In: International Conference on Learning Representations. 2020.URL:https://openreview.net/forum?id=ryg GQyrFvH.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Angela Fan, Mike Lewis, and Yann Dauphin. “Hierarchical Neural Story Generation”. In:Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Melbourne, Australia: Association for Computational Linguistics, July 2018, pp. 889–898.DOI:10.18653/v1/P18- 1082.URL:https://aclanthology. org/P18-1082/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Alec Radford et al. “Language models are unsupervised multitask learners”. In:OpenAI blog1.8 (2019), p. 9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Hongxiang Zhang, Hao Chen, Muhao Chen, and Tianyi Zhang. “Active Layer-Contrastive Decoding Reduces Hallucination in Large Language Model Generation”. In:Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. Suzhou, China: Association for Computational Linguistics, Nov. 2025, pp. 3028–3046.ISBN: 979-8-89176-3326.DOI:10.18653/v1/2025.emnlp-main.150.URL:https://acla",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "David H Ackley, Geoffrey E Hinton, and Terrence J Sejnowski. “A learning algorithm for Boltzmann machines”. In:Cognitive science9.1 (1985), pp. 147–169.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Geoffrey Hinton, Oriol Vinyals, and Jeff Dean.Distilling the Knowledge in a Neural Network. 2015. arXiv:1503.02531 [stat.ML].URL:https://arxiv.org/abs/1503.02531.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Jessica Ficler and Yoav Goldberg. “Controlling Linguistic Style Aspects in Neural Language Generation”. In:Proceedings of the Workshop on Stylistic Variation. Copenhagen, Denmark: Association for Computational Linguistics, Sept. 2017, pp. 94–104.DOI:10.18653/v1/W17-4912.URL:https://aclanthology.org/W17-4912/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Matthew Renze. “The Effect of Sampling Temperature on Problem Solving in Large Language Models”. In:Findings of the Association for Computational Linguistics: EMNLP 2024. Miami, Florida, USA: Association for Computational Linguistics, Nov. 2024, pp. 7346–7356.DOI:10.18653/v1/2024.findings-emnlp.432.URL:https://aclanthology.org/ 2024.findings-emnlp.432/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Chunqiu Steven Xia, Yinlin Deng, Soren Dunn, and Lingming Zhang. “Demystifying LLM-Based Software Engineering Agents”. In:Proc. ACM Softw. Eng.2.FSE (June 2025).DOI:10.1145/3715754.URL:https://doi.org/10.1145/ 3715754.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Aston Zhang, Zachary C. Lipton, Mu Li, and Alexander J. Smola.Dive into Deep Learning.https://D2L.ai. Cambridge University Press, 2023. 18 Code Generation by Differential Test Time Scaling",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Percy Liang et al. “Holistic Evaluation of Language Models”. In:Transactions on Machine Learning Research(2023). Featured Certification, Expert Certification, Outstanding Certification.ISSN: 2835-8856.URL:https://openreview. net/forum?id=iO4LZibEqW.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Melanie Sclar, Yejin Choi, Yulia Tsvetkov, and Alane Suhr. “Quantifying Language Models’Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formatting”. In:International Conference on Representation Learning. V ol. 2024. 2024, pp. 25055–25083.URL:https://proceedings.iclr.cc/paper_files/ paper/2024/file/6c0e99d736da621403018ca7b32b1a4d-Paper-Conference",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Jingming Zhuo et al. “ProSA: Assessing and Understanding the Prompt Sensitivity of LLMs”. In:Findings of the Association for Computational Linguistics: EMNLP 2024. Miami, Florida, USA: Association for Computational Linguistics, Nov. 2024, pp. 1950–1976.DOI:10. 18653 / v1 / 2024. findingsemnlp. 108.URL:https: / / aclanthology. org / 2024. findings-emnlp.108/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "“What Did I Do Wrong? Quantifying LLMs’ Sensitivity and Consistency to Prompt Engineering”. In:Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). Albuquerque, New Mexico: Association for Computational Linguistics, Apr. 2025, pp. 1543–1558.ISBN: 979-8-89176-189-6.DOI:10. 186",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Catherine Tony, Nicol ´as E. D ´ıaz Ferreyra, Markus Mutas, Salem Dhif, and Riccardo Scandariato. “Prompting Techniques for Secure Code Generation: A Systematic Investigation”. In:ACM Trans. Softw. Eng. Methodol.34.8 (Oct. 2025).ISSN: 1049-331X.DOI:10.1145/3722108.URL:https://doi.org/10.1145/3722108.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Sydney Nguyen et al. “How Beginning Programmers and Code LLMs (Mis)read Each Other”. In:Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems. CHI ’24. Honolulu, HI, USA: Association for Computing Machinery, 2024.ISBN: 9798400703300.DOI:10.1145/3613904.3642706.URL:https://doi.org/10.1145/3613904. 3642706.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Shuhan Liu et al. “CREME: Robustness Enhancement of Code LLMs via Layer-Aware Model Editing”. In:Proceedings of the IEEE/ACM 48th International Conference on Software Engineering. ICSE ’26. Rio de Janeiro, Brazil: Association for Computing Machinery, 2026.DOI:3744916.3773111.URL:https://arxiv.org/abs/2507.16407v3.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Ben Shi, Michael Tang, Karthik R Narasimhan, and Shunyu Yao. “Can Language Models Solve Olympiad Programming?” In:First Conference on Language Modeling. 2024.URL:https://openreview.net/forum?id=kGa4fMtP9l.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Markus Freitag and Yaser Al-Onaizan. “Beam Search Strategies for Neural Machine Translation”. In:Proceedings of the First Workshop on Neural Machine Translation. Vancouver: Association for Computational Linguistics, Aug. 2017, pp. 56– 60.DOI:10.18653/v1/W17-3207.URL:https://aclanthology.org/W17-3207/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Kaustubh Dhole et al. “NL-Augmenter: A Framework for Task-Sensitive Natural Language Augmentation”. In:Northern European Journal of Language Technology9 (2023).DOI:10.3384/nejlt.2000- 1533.2023.4725.URL:https: //aclanthology.org/2023.nejlt-1.5/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Yao Qiang et al. “Prompt Perturbation Consistency Learning for Robust Language Models”. In:Findings of the Association for Computational Linguistics: EACL 2024. St. Julian’s, Malta: Association for Computational Linguistics, Mar. 2024, pp. 1357–1370.URL:https://aclanthology.org/2024.findings-eacl.91/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Zhanke Zhou et al. “Can Language Models Perform Robust Reasoning in Chain-of-thought Prompting with Noisy Rationales?” In:The Thirty-eighth Annual Conference on Neural Information Processing Systems. 2024.URL:https:// openreview.net/forum?id=FbuODM02ra.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Lalchand Pandia and Allyson Ettinger. “Sorting through the noise: Testing robustness of information processing in pretrained language models”. In:Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. Online and Punta Cana, Dominican Republic: Association for Computational Linguistics, Nov. 2021, pp. 1583–1596.DOI: 10.18653/v1/2021.emnlp-main.119.URL:https://aclant",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Feiteng Fang et al. “Enhancing Noise Robustness of Retrieval-Augmented Language Models with Adaptive Adversarial Training”. In:Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Bangkok, Thailand: Association for Computational Linguistics, Aug. 2024, pp. 10028–10039.DOI:10.18653/ v1/2024.acl-long.540.URL:https://aclanthology.org/2024.ac",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Barbara Rychalska, Dominika Basaj, Alicja Gosiewska, and Przemysław Biecek. “Models in the Wild: On Corruption Robustness of Neural NLP Systems”. In:Neural Information Processing. Ed. by Tom Gedeon, Kok Wai Wong, and Minho Lee. Cham: Springer International Publishing, 2019, pp. 235–247.ISBN: 978-3-030-36718-3.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Jianyu Zhao, Yuyang Rong, Yiwen Guo, Yifeng He, and Hao Chen. “Understanding Programs by Exploiting (Fuzzing) Test Cases”. In:Findings of the Association for Computational Linguistics: ACL 2023. Toronto, Canada: Association for Computational Linguistics, July 2023, pp. 10667–10679.DOI:10.18653/v1/2023.findings-acl.678.URL:https: //aclanthology.org/2023.findings-acl.678/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Kosta Serebryany. “Continuous Fuzzing with libFuzzer and AddressSanitizer”. In:2016 IEEE Cybersecurity Development (SecDev). 2016, pp. 157–157.URL:https://doi.org/10.1109/SecDev.2016.043.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Yunlong Lyu, Yuxuan Xie, Peng Chen, and Hao Chen. “Prompt Fuzzing for Fuzz Driver Generation”. In:Proceedings of the 2024 on ACMSIGSAC Conference on Computer and Communications Security. CCS ’24. Salt Lake City, UT, USA: Association for Computing Machinery, 2024, pp. 3793–3807.ISBN: 9798400706363.DOI:10.1145/3658644.3670396. URL:https://doi.org/10.1145/3658644.3670396. 19 Code Generation by Differ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Yujia Li et al. “Competition-level code generation with AlphaCode”. In:Science378.6624 (2022), pp. 1092–1097.DOI: 10.1126/science.abq1158. eprint:https://www.science.org/doi/pdf/10.1126/science.abq1158.URL: https://www.science.org/doi/abs/10.1126/science.abq1158.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Xuezhi Wang et al. “Self-Consistency Improves Chain of Thought Reasoning in Language Models”. In:The Eleventh International Conference on Learning Representations. 2023.URL:https://openreview.net/forum?id=1PL1NIMMrw.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Vincent Cohen-addad, Varun Kanade, Frederik Mallmann-trenn, and Claire Mathieu. “Hierarchical Clustering: Objective Functions and Algorithms”. In:J. ACM66.4 (June 2019).ISSN: 0004-5411.DOI:10. 1145 / 3321386.URL:https: //doi.org/10.1145/3321386.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Eric K. Tokuda, Cesar H. Comin, and Luciano da F. Costa. “Revisiting agglomerative clustering”. In:Physica A: Statistical Mechanics and its Applications585 (2022), p. 126433.ISSN: 0378-4371.DOI:https://doi.org/10.1016/j.physa. 2021.126433.URL:https://www.sciencedirect.com/science/article/pii/S0378437121007068.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Binyuan Hui et al.Qwen2.5Coder Technical Report. 2024. arXiv:2409.12186 [cs.CL].URL:https://arxiv.org/ abs/2409.12186.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Woosuk Kwon et al. “Efficient Memory Management for Large Language Model Serving with PagedAttention”. In:Proceedings of the 29th Symposium on Operating Systems Principles. SOSP ’23. Koblenz, Germany: Association for Computing Machinery, 2023, pp. 611–626.ISBN: 9798400702297.DOI:10.1145/3600006.3613165.URL:https://doi.org/ 10.1145/3600006.3613165.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Jiabo Huang et al. “Code Representation Pre-training with Complements from Program Executions”. In:Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track (EMNLP). Miami, Florida, US: Association for Computational Linguistics, Nov. 2024, pp. 267–278.DOI:10.18653/v1/2024.emnlp-industry.21. URL:https://aclanthology.org/2024.emnlp-industry.21/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Dacheng Li et al.S*: Test Time Scaling for Code Generation (Source code). 2025.URL:https://github.com/NovaSkyAI/SkyThought/tree/0d190f11fd8e885bbe113aeccacba5ccde5b1102/skythought/test-time-scaling.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Google.Gemini 2.5 Flash-Lite.URL:https://docs.cloud.google.com/vertexai/generativeai/docs/ models/gemini/2-5-flash-lite.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Peter Rousseeuw. “Silhouettes: a graphical aid to the interpretation and validation of cluster analysis”. In:J. Comput. Appl. Math.20.1 (Nov. 1987), pp. 53–65.ISSN: 0377-0427.DOI:10.1016/0377- 0427(87)90125- 7.URL:https: //doi.org/10.1016/0377-0427(87)90125-7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Chris Yuhao Liu et al.Skywork-Reward-V2: Scaling Preference Data Curation via Human-AI Synergy. 2026. arXiv:2507. 01352 [cs.CL].URL:https://arxiv.org/abs/2507.01352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Koen Claessen and John Hughes. “QuickCheck: A Lightweight Tool for Random Testing of Haskell Programs”. In:Proceedings of the Fifth ACMSIGPLAN International Conference on Functional Programming. ICFP ’00. New York, NY, USA: Association for Computing Machinery, 2000, pp. 268–279.ISBN: 1581132026.URL:https://doi.org/10.1145/ 351240.351266.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Harrison Goldstein, Joseph W. Cutler, Daniel Dickstein, Benjamin C. Pierce, and Andrew Head. “Property-Based Testing in Practice”. In:Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. ICSE ’24. Lisbon, Portugal: Association for Computing Machinery, 2024.ISBN: 9798400702174.URL:https://doi.org/10.1145/ 3597503.3639581.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Zhiyu Fan, Haifeng Ruan, Sergey Mechtaev, and Abhik Roychoudhury. “Oracle-Guided Program Selection from Large Language Models”. In:Proceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. ISSTA 2024. Vienna, Austria: Association for Computing Machinery, 2024, pp. 628–640.ISBN: 9798400706127.DOI: 10.1145/3650212.3680308.URL:https://doi.org/10.1145/3650212.3680308",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Juyong Jiang, Fan Wang, Jiasi Shen, Sungju Kim, and Sunghun Kim. “A Survey on Large Language Models for Code Generation”. In:ACM Trans. Softw. Eng. Methodol.35.2 (Jan. 2026).ISSN: 1049-331X.DOI:10.1145/3747588.URL: https://doi.org/10.1145/3747588.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "An Yang et al.Qwen3 Technical Report. 2025. arXiv:2505.09388 [cs.CL].URL:https://arxiv.org/abs/2505. 09388.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Daya Guo et al. “DeepSeekR1 incentivizes reasoning in LLMs through reinforcement learning”. In:Nature645.8081 (Sept. 2025), pp. 633–638.ISSN: 1476-4687.DOI:10.1038/s41586-025-09422-z.URL:http://dx.doi.org/10.1038/ s41586-025-09422-z.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Carlos E Jimenez et al. “SWE-bench: Can Language Models Resolve Real-world Github Issues?” In:The Twelfth International Conference on Learning Representations. 2024.URL:https://openreview.net/forum?id=VTF8yNQM66.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Qi Guo et al. “Exploring the Potential of ChatGPT in Automated Code Refinement: An Empirical Study”. In:Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. ICSE ’24. Lisbon, Portugal: Association for Computing Machinery, 2024.ISBN: 9798400702174.DOI:10.1145/3597503.3623306.URL:https://doi.org/10. 1145/3597503.3623306.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Chunqiu Steven Xia, Yuxiang Wei, and Lingming Zhang. “Automated Program Repair in the Era of Large Pre-trained Language Models”. In:. 2023, pp. 1482– 1494.DOI:10.1109/ICSE48619.2023.00129. 20 Code Generation by Differential Test Time Scaling",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Yifeng He et al. “UniTSyn: A Large-Scale Dataset Capable of Enhancing the Prowess of Large Language Models for Program Testing”. In:Proceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. ISSTA 2024. Vienna, Austria: Association for Computing Machinery, 2024, pp. 1061–1072.ISBN: 9798400706127.DOI: 10.1145/3650212.3680342.URL:https://doi.org/10.1145/3650212.3680",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "Weimin Xiong, Yiwen Guo, and Hao Chen. “The Program Testing Ability of Large Language Models for Code”. In:Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track. Miami, Florida, US: Association for Computational Linguistics, Nov. 2024, pp. 23–34.DOI:10.18653/v1/2024.emnlp-industry.3. URL:https://aclanthology.org/2024.emnlp-industry.3/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Hongxiang Zhang, Yuyang Rong, Yifeng He, and Hao Chen. “LLAMAFUZZ: Large Language Model Enhanced Greybox Fuzzing”. In:ACM/IEEE International Conference on Automation of Software Test (AST). 2026.URL:https://arxiv. org/abs/2406.07714.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Jon Saad-Falcon et al. “An Architecture Search Framework for Inference-Time Techniques”. In:Forty-second International Conference on Machine Learning. 2025.URL:https://openreview.net/forum?id=EGrSMMj37o.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Jason Wei et al. “Chain-of-thought prompting elicits reasoning in large language models”. In:Proceedings of the 36th International Conference on Neural Information Processing Systems. NIPS ’22. New Orleans, LA, USA: Curran Associates Inc., 2022.ISBN: 9781713871088.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "Hongxiang Zhang, Yuan Tian, and Tianyi Zhang.Self-Anchor: Large Language Model Reasoning via Step-by-step Attention Alignment. 2025. arXiv:2510.03223 [cs.CL].URL:https://arxiv.org/abs/2510.03223.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Dacheng Li et al. “Language Models Can Easily Learn to Reason from Demonstrations”. In:Findings of the Association for Computational Linguistics: EMNLP 2025. Suzhou, China: Association for Computational Linguistics, Nov. 2025, pp. 15979–15997.ISBN: 979-8-89176-335-7.DOI:10. 18653 / v1 / 2025. findingsemnlp. 866.URL:https: / / aclanthology.org/2025.findings-emnlp.866/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "Aman Madaan et al. “Self-Refine: Iterative Refinement with Self-Feedback”. In:Advances in Neural Information Processing Systems. Ed. by A. Oh et al. V ol. 36. Curran Associates, Inc., 2023, pp. 46534–46594.URL:https://proceedings. neurips.cc/paper_files/paper/2023/file/91edff07232fb1b55a505a9e9f6c0ff3-Paper-Conference.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "Shunyu Yao et al. “ReAct: Synergizing Reasoning and Acting in Language Models”. In:The Eleventh International Conference on Learning Representations. 2023.URL:https://openreview.net/forum?id=WE_vluYUL-X.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik R Narasimhan, and Shunyu Yao. “Reflexion: language agents with verbal reinforcement learning”. In:Thirty-seventh Conference on Neural Information Processing Systems. 2023.URL: https://openreview.net/forum?id=vAElhFcKW6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "Jiaxin Huang et al. “Large Language Models Can Self-Improve”. In:Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. Ed. by Houda Bouamor, Juan Pino, and Kalika Bali. Singapore: Association for Computational Linguistics, Dec. 2023, pp. 1051–1068.DOI:10.18653/v1/2023.emnlpmain.67.URL:https: //aclanthology.org/2023.emnlp-main.67/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "Yifei Wang, Yuyang Wu, Zeming Wei, Stefanie Jegelka, and Yisen Wang. “A Theoretical Understanding of Self-Correction through In-context Alignment”. In:The Thirty-eighth Annual Conference on Neural Information Processing Systems. 2024. URL:https://openreview.net/forum?id=OtvNLTWYww.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "Tal Ridnik, Dedy Kredo, and Itamar Friedman.Code Generation with AlphaCodium: From Prompt Engineering to Flow Engineering. 2024. arXiv:2401.08500 [cs.LG].URL:https://arxiv.org/abs/2401.08500.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "Baizhou Huang, Shuai Lu, Xiaojun Wan, and Nan Duan. “Enhancing Large Language Models in Coding Through MultiPerspective Self-Consistency”. In:Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Bangkok, Thailand: Association for Computational Linguistics, Aug. 2024, pp. 1429–1450.DOI: 10.18653/v1/2024.acl-long.78.URL:https://aclanthology",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "Jicheng Wang, Yifeng He, and Hao Chen.RepoGenReflex: Enhancing Repository-Level Code Completion with Verbal Reinforcement and Retrieval-Augmented Generation. 2024. arXiv:2409.13122 [cs.SE].URL:https://arxiv.org/abs/ 2409.13122.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "Qingyao Li et al. “RethinkMCTS: Refining Erroneous Thoughts in Monte Carlo Tree Search for Code Generation”. In:Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. Ed. by Christos Christodoulopoulos, Tanmoy Chakraborty, Carolyn Rose, and Violet Peng. Suzhou, China: Association for Computational Linguistics, Nov. 2025, pp. 8092–8110.ISBN: 979-8-89176-332-6.DOI:10",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "Anonymous.Artifacts for “Code Generation by Differential Test Time Scaling”.DOI:10.5281/zenodo.18426367.URL: https://doi.org/10.5281/zenodo.18426367. 21",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 19,
+      "text": "Manuel Rigger and Zhendong Su. “Testing Database Engines via Pivoted Query Synthesis”. In:Proc. ACM Program. Lang. 4.OOPSLA (Nov. 2020).DOI:10.1145/3428279.URL:https://doi.org/10.1145/3428279.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[19]",
+      "technique": "pqs",
+      "sentence": "This technique has proven highly effective for testing compilers [16–18], database systems [19], and other systems where formal specifications or programmer-provided test oracles are either unavailable or impractical to use.",
+      "context_before": "er a program’s output is correct when no ground truth specification exists [11]. Instead of relying on expert-provided oracles (i.e., expected outputs), differential testing compares the behaviors of multiple implementations that should conform to the same specification. When these implementations produce different outputs on the same input, at least one of them must contain a bug. Cross-referencing multiple implementations of the same specification provides a pseudo-oracle, where discrepancies in outputs indicate potential faults, and agreement suggests the likelihood of correctness [11, 12].",
+      "context_after": "LLM-based code generation naturally fits the differential testing paradigm. When sampling multiple code candidates from a language model given the same natural language specification, each candidate represents an independent attempt to implement the described functionality. Although these candidates may differ in implementation details, variable names, control flow structures, or algorithmic choices, they share the same intended cognitive semantics as defined by the prompt [20]. This setting mirrors the classical differential testing scenario: multiple implementations of the same specificatio",
+      "section": "2.1.1 Differential testing",
+      "page": 3,
+      "char_offset": 8693,
+      "cited_reference": {
+        "number": 19,
+        "text": "Manuel Rigger and Zhendong Su. “Testing Database Engines via Pivoted Query Synthesis”. In:Proc. ACM Program. Lang. 4.OOPSLA (Nov. 2020).DOI:10.1145/3428279.URL:https://doi.org/10.1145/3428279.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[13, 16, 19, 25]",
+      "technique": "pqs",
+      "sentence": "Much research has applied differential testing with such assumptions to complex software systems like compilers and databases [13, 16, 19, 25], 6 Related work 6.",
+      "context_before": "y to help, whereas embedding-based techniques with dynamic behaviors [57, 66] may provide a viable alternative. Failure of the majority assumption.DIFFCODEGENrelies on the hypothesis that the largest behavioral cluster contains correct solutions. This assumption may fail when: 1. most candidates share a common misconception; 2. the natural language specification is ambiguous, leading to multiple valid interpretations. In such cases, DIFFCODEGEN may confidently select an incorrect solution. However, we have shown empirically that this assumption holds for the vast majority of programming tasks.",
+      "context_after": "1 Code generation models Large language models have demonstrated remarkable capabilities in writing programs. The text-to-code generation task involves LLMs producing code snippets based on natural language descriptions or specifications [74]. Early work such as Codex [3] and AlphaCode [60] demonstrated that LLMs fine-tuned on code can solve programming problems described in natural language. Subsequent models have continued to advance the state of the art, including openweight models such as Qwen-Coder [64, 75] and reasoning-augmented models like DeepSeek-R1 [76]. Generative code models have",
+      "section": "5.2 Limitations and future work",
+      "page": 15,
+      "char_offset": 57936,
+      "cited_reference": {
+        "number": 19,
+        "text": "Manuel Rigger and Zhendong Su. “Testing Database Engines via Pivoted Query Synthesis”. In:Proc. ACM Program. Lang. 4.OOPSLA (Nov. 2020).DOI:10.1145/3428279.URL:https://doi.org/10.1145/3428279.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:19:46Z",
+    "is_model_written": true,
+    "summary": "DiffCodeGen is a test-time scaling method for code generation that selects among candidate programs without extra model calls. It samples diverse candidates, uses coverage-guided fuzzing to synthesise inputs without needing existing tests, executes every candidate on those inputs, clusters them by behavioural similarity, and returns the medoid of the largest cluster. Across four models it reports competitive or better results than other test-time scaling methods at a fraction of the cost.",
+    "narrative": "Two citations, listing database systems among the domains where differential testing has proven effective where formal specifications are unavailable.",
+    "roles": {
+      "M1": "background",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2605_22992.json b/_data/papers/paper_arxiv_2605_22992.json
new file mode 100644
index 0000000..67565b6
--- /dev/null
+++ b/_data/papers/paper_arxiv_2605_22992.json
@@ -0,0 +1,832 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:15Z",
+  "paper": {
+    "id": "paper:arxiv:2605.22992",
+    "title": "Finding Performance Issues in Database Systems by Exploiting Dormant Code Paths",
+    "authors": [
+      "Jinsheng Ba",
+      "Zhendong Su"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2605.22992",
+    "s2_paper_id": "652ecd8215728ede7bbecba4ac5049ac96af9e91",
+    "url": "https://arxiv.org/abs/2605.22992",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2605.22992",
+      "retrieved_at": "2026-09-09T01:29:15Z",
+      "chars": 69522,
+      "content_sha256": "sha256:4da7e401568af79e91706f0d29d5b2e6147d94d589f5150b819e5ec64143a08d"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1547
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 10841
+      },
+      {
+        "number": "3",
+        "title": "Motivating Study",
+        "start": 11931
+      },
+      {
+        "number": "4",
+        "title": "Approach",
+        "start": 16877
+      },
+      {
+        "number": "5",
+        "title": "Implementation",
+        "start": 25143
+      },
+      {
+        "number": "6",
+        "title": "Evaluation",
+        "start": 29737
+      },
+      {
+        "number": "7",
+        "title": "Discussion",
+        "start": 50116
+      },
+      {
+        "number": "8",
+        "title": "Related Work",
+        "start": 53759
+      },
+      {
+        "number": "9",
+        "title": "Conclusion",
+        "start": 57592
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Jinsheng Ba and Manuel Rigger. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, pages 1–13, 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. Keep it simple: Testing databases via differential query plans.Proceedings of the ACM on Management of Data, 2(3):1–26, 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Laurent Bindschaedler, Ashvin Goel, and Willy Zwaenepoel. Hailstorm: Disaggregated compute and storage for distributed lsm-based databases. InASPLOS ’20: Architectural Support for Programming Languages and Operating Systems, Lausanne, Switzerland, March 16-20, 2020, pages 301–316, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Peter A. Boncz, Thomas Neumann, and Orri Erling. TPC-H analyzed: Hidden messages and lessons learned 12 from an influential benchmark. In Raghunath Nambiar and Meikel Poess, editors,Performance Characterization and Benchmarking-5th TPC Technology Conference, TPCTC 2013, Trento, Italy, August 26, 2013, Revised Selected Papers, volume 8391 ofLecture Notes in Computer Science, pages 61–76. Springer, ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Surajit Chaudhuri. An overview of query optimization in relational systems. In Alberto O. Mendelzon and Jan Paredaens, editors,Proceedings of the Seventeenth ACMSIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems, June 1-3, 1998, Seattle, Washington, USA, pages 34–43. ACM Press, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Pit Fender and Guido Moerkotte. Counter strike: Generic top-down join enumeration for hypergraphs. Proc. VLDB Endow., 6(14):1822–1833, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Pit Fender, Guido Moerkotte, Thomas Neumann, and Viktor Leis. Effective and robust pruning for top-down join enumeration algorithms. In Anastasios Kementsietsidis and Marcos Antonio Vaz Salles, editors,IEEE 28th International Conference on Data Engineering (ICDE 2012), Washington, DC, USA (Arlington, Virginia), 1-5 April, 2012, pages 414–425. IEEE Computer Society, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. Griffin: Grammar-free DBMS fuzzing. In 37th IEEE/ACM International Conference on Automated Software Engineering, ASE 2022, Rochester, MI, USA, October 10-14, 2022, pages 49:1–49:12. ACM, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Jana Giceva, Gustavo Alonso, Timothy Roscoe, and Tim Harris. Deployment of query plans on multicores.Proc. VLDB Endow., 8(3):233–244, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Yigong Hu, Gongqi Huang, and Peng Huang. Automated reasoning and detection of specious configuration in large systems with symbolic execution. In14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020, pages 719–734, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Yannis E Ioannidis. Query optimization.ACM Computing Surveys (CSUR), 28(1):121–123, 1996.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Oleg Ivanov and Sergey Bartunov. Adaptive cardinality estimation.arXiv preprint arXiv:1711.08330, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Matthias Jarke and Jurgen Koch. Query optimization in database systems.ACM Computing surveys (CsUR), 16(2):111–152, 1984.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. Dynsql: Stateful fuzzing for database management systems with complex and valid sql query generation. In32st USENIX Security Symposium (USENIX Security 23), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Guoliang Jin, Linhai Song, Xiaoming Shi, Joel Scherpelz, and Shan Lu. Understanding and detecting realworld performance bugs. In Jan Vitek, Haibo Lin, and Frank Tip, editors,ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI ’12, Beijing, China-June 11 - 16, 2012, pages 77–88. ACM, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woon-Hak Kang. APOLLO: automatic detection and diagnosis of performance regressions in database systems. Proc. VLDB Endow., 13(1):57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter A. Boncz, Alfons Kemper, and Thomas Neumann. How good are query optimizers, really?Proc. VLDB Endow., 9(3):204–215, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Caroline Lemieux, Rohan Padhye, Koushik Sen, and Dawn Song. Perffuzz: Automatically generating pathological inputs. InProceedings of the 27th ACMSIGSOFT international symposium on software testing and analysis, pages 254–265, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Chi Li, Shu Wang, Henry Hoffmann, and Shan Lu. Statically inferring performance properties of software configurations. InProceedings of the Fifteenth European Conference on Computer Systems, pages 1–16, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. Mozi: Discovering dbms bugs via configuration-based equivalent transformation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, pages 1–12, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. Automatic detection of performance bugs in database systems using equivalent queries. In44th IEEE/ACM 44th International Conference on Software Engineering, ICSE 2022, Pittsburgh, PA, USA, May 25-27, 2022, pages 225–236. ACM, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Zihan Liu, Wentao Ni, Jingwen Leng, Yu Feng, Cong Guo, Quan Chen, Chao Li, Minyi Guo, and Yuhao Zhu. JUNO: optimizing high-dimensional approximate nearest neighbour search with sparsity-aware algorithm and ray-tracing core mapping. In Rajiv Gupta, Nael B. AbuGhazaleh, Madan Musuvathi, and Dan Tsafrir, editors, Proceedings of the 29th ACM International Conference on Architectural Support for Progra",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Raghunath Othayoth Nambiar, Meikel Poess, Andrew Masland, H. Reza Taheri, Matthew Emmerton, Forrest Carman, and Michael Majdalany. TPC benchmark 13 roadmap 2012. In Raghunath Othayoth Nambiar and Meikel Poess, editors,Selected Topics in Performance Evaluation and Benchmarking-4th TPC Technology Conference, TPCTC 2012, Istanbul, Turkey, August 27, 2012, Revised Selected Papers, volume 7755 ofLectur",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Thomas Neumann. Query simplification: graceful degradation for join-order optimization. In Ugur Çetintemel, Stanley B. Zdonik, Donald Kossmann, and Nesime Tatbul, editors,Proceedings of the ACMSIGMOD International Conference on Management of Data, SIGMOD 2009, Providence, Rhode Island, USA, June 29 - July 2, 2009, pages 403–414. ACM, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Johns Paul, Jiong He, and Bingsheng He. GPL: A gpubased pipelined query processing engine. In Fatma Özcan, Georgia Koutrika, and Sam Madden, editors, Proceedings of the 2016 International Conference on Management oData, SIGMOD Conference 2016, San Francisco, CA, USA, June 26 - July 01, 2016, pages 1935–1950. ACM, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Theofilos Petsios, Jason Zhao, Angelos D Keromytis, and Suman Jana. Slowfuzz: Automated domainindependent detection of algorithmic complexity vulnerabilities. InProceedings of the 2017 ACMSIGSAC conference on computer and communications security, pages 2155–2168, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Prem Devanbu, Myra B. Cohen, and Thomas Zimmermann, editors,ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, pages 1140–1152. ACM, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang., 4(OOPSLA):211:1–211:30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020, pages 667–682. USENIX Association, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 30,
+      "text": "Chaoyi Ruan, Yingqiang Zhang, Chao Bi, Xiaosong Ma, Hao Chen, Feifei Li, Xinjun Yang, Cheng Li, Ashraf Aboulnaga, and Yinlong Xu. Persistent memory disaggregation for cloud-native relational databases. In Tor M. Aamodt, Natalie D. Enright Jerger, and Michael M. Swift, editors,Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Syste",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. Testing database systems via differential query execution. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Emil Tsalapatis, Ryan Hancock, Rakeeb Hossain, and Ali José Mashtizadeh. Memsnap µcheckpoints: A data single level store for fearless persistence. In Rajiv Gupta, Nael B. Abu-Ghazaleh, Madan Musuvathi, and Dan Tsafrir, editors,Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 3, ASPLOS 2024, La Jolla, CA, USA, 27 A",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Website. Tpc-ds benchmark. https://www.tpc.org/ tpcds/, 1988. Accessed: 2022-11-15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Website. American fuzzy lop (afl) fuzzer. http://lcamtuf.coredump.cx/afl/technical_ details.txt, 2013. Accessed: 2022-11-15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Website. Sqlsmith. https://github.com/anse1/ sqlsmith, 2015. Accessed: 2022-11-15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Website. Cockroachdb code coverage measurement. https://cockroachlabs.atlassian.net/wiki/ spaces/CRDB/pages/73171260/Code+ coverage, 2022. Accessed: 2025-05-15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Website. Database systems ranking. https: //db-engines.com/en/ranking, 2022. Accessed: 2025-05-15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Website. Performance on widely used workloads. https://www.postgresql.org/ message-id/CAMkU%3D1xoU06eW4CrEZyDDn% 2BfnJaCe3b04rE3mdVu4Gsxmj9KFA%40mail.gmail. com, 2025. Accessed: 2025-11-15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Lisa Wu, Andrea Lottarini, Timothy K. Paine, Martha A. Kim, and Kenneth A. Ross. Q100: the architecture and design of a database processing unit. In Rajeev Balasubramonian, Al Davis, and Sarita V. Adve, editors, Architectural Support for Programming Languages and Operating Systems, ASPLOS 2014, Salt Lake City, UT, USA, March 1-5, 2014, pages 255–268. ACM, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. Puppy: Finding performance degradation bugs in dbmss via limited-optimization plan construction. In on Software Engineering (ICSE), pages 560–571. IEEE Computer Society, 2024. 14",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors,CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event, USA, November 9-13, 2020, pages 955–970. ACM, 2020. 15",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 1,
+      "text": "Jinsheng Ba and Manuel Rigger. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, pages 1–13, 2024.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. Keep it simple: Testing databases via differential query plans.Proceedings of the ACM on Management of Data, 2(3):1–26, 2024.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Prem Devanbu, Myra B. Cohen, and Thomas Zimmermann, editors,ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, pages 1140–1152. ACM, 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 28,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang., 4(OOPSLA):211:1–211:30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020, pages 667–682. USENIX Association, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "CERT[1] finds performance issues by finding inconsistent cardinality estimation, which is typically deemed as the most critical component for query optimization [17].",
+      "context_before": "ehensively explore the full range of optimization strategies and their interactions. Various black-box methods have been proposed to derive “optimal” performance for finding performance issues, but they focus on a specific category of issues and cannot systematically explore optimization strategies.APOLLO[16] examines whether a DBMS has worse performance than an old version of the same DBMS. However,APOLLOcan only find regression issues.AMOEBA[21] expects that the execution time of pairs of semantically equivalent queries is similar. However, it only finds the issues of rewriting queries [21].",
+      "context_after": "However, CERTcan only find performance issues related to incorrect cardinality estimation.PUPPY[40] assumes that the execution time of queries under the default optimization configuration should be no longer than under alternative configurations. However, this approach is limited to detecting configuration-related performance issues. These methods focus on manipulating input queries, which are inefficient for exploring optimization strategies. In this paper, we proposeBranch Flip Analysis(BFA), a novel, general white-box method to systematically find per1arXiv:2605.22992v1 [cs.SE] 21 May 2026",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3628,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "To manipulate code, we propose flipping branches, which is motivated by our observation that 8 of 10 historical performance issues from APOLLO,AMOEBA,CERT, and PUPPYcould be found by flipping branches.",
+      "context_before": "all branches except bjexecute their default paths bi,d, while bjexecutes the alternative path bj,o. IfP(B)<<P(B′), a performance issue related to the optimization controlled byb jis found. Thus, in a novel manner,BFAsolves the aforementioned challenges faced by existing work—it derives expected performance by constructing a reference DBMS. Code manipulation also allows us to explore various optimization strategies in a systematic and fine-grained manner. In contrast, existing work is constrained by external inputs or configurations, which often fail to control all the available optimizations.",
+      "context_after": "Branch flipping is straightforward to realize and general, so it can be widely applied to test DBMSs without domain expertise. Additionally, it highlights the code segment relevant to the performance issue, thus facilitating issue analysis. Listing 1 shows a running example of BFA. Lines 1–3 include the query to identify the performance issue, lines 4–12 show the code that BFAmanipulates, and lines 14–19 show the query plans, the concrete execution steps of this query in a tree structure, before and after the code manipulation. To facilitate the presentation, we only show relevant parts. This q",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6152,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[1, 16, 40]",
+      "technique": "cert",
+      "sentence": "We evaluatedQueryZenon PostgreSQL, MySQL, MariaDB, and CockroachDB, which are widely used in previous performance testing works [1, 16, 40].",
+      "context_before": "eplied that “It was a pretty exciting case” and planned to solve it by enabling PARALLEL for these operations shortly. This issue can not be found by previous methods because PARALLEL cannot be enabled for Hash by any version of PostgreSQL (i.e., regression issues), any configuration, or any estimated cardinality. We designed and implementedQueryZen, a prototype of BFA. Technically, to avoid the heavy overhead of recompilation for each code manipulation, we instrument all code manipulations into the code with a single compilation and enable them one by one via external environmental variables.",
+      "context_after": "In total,QueryZen found 21 unique and previously unknown performance issues with the workloads from standard benchmarks (i.e., TPC-H [23] and TPC-DS [33]). These findings resulted in an average performance improvement of 26.2 ×, and up to 374.9×.QueryZenis necessary to find these issues as most of them cannot be found by APOLLO,CERT,PUPPY, and AMOEBA. Furthermore, these issues already existed before the publication of prior methods. For testing efficiency, we avoid the recompilation time of an average of 5 minutes for each execution. On average of both benchmarks,QueryZen also excludes 75.8% u",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 9291,
+      "cited_reference": {
+        "number": 1,
+        "text": "Jinsheng Ba and Manuel Rigger. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, pages 1–13, 2024.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "QueryZenis necessary to find these issues as most of them cannot be found by APOLLO,CERT,PUPPY, and AMOEBA.",
+      "context_before": "avoid the heavy overhead of recompilation for each code manipulation, we instrument all code manipulations into the code with a single compilation and enable them one by one via external environmental variables. We evaluatedQueryZenon PostgreSQL, MySQL, MariaDB, and CockroachDB, which are widely used in previous performance testing works [1, 16, 40]. In total,QueryZen found 21 unique and previously unknown performance issues with the workloads from standard benchmarks (i.e., TPC-H [23] and TPC-DS [33]). These findings resulted in an average performance improvement of 26.2 ×, and up to 374.9×.",
+      "context_after": "Furthermore, these issues already existed before the publication of prior methods. For testing efficiency, we avoid the recompilation time of an average of 5 minutes for each execution. On average of both benchmarks,QueryZen also excludes 75.8% unnecessary code manipulations and increases branch coverage by 8.6%. Overall, we make the following contributions: •Conceptually, we propose a novel methodology BFAto find performance issues by flipping code branches. •Technically, we designed and implemented a novel ar2 Table 1: Historical performance issues. Method DBMS Issue ID Branch Flipping APOLL",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 9679,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "We considered the performance issues found by the four existing performance-testing methods:APOLLO[16],AMOEBA[21],CERT[1], and PUPPY[40].",
+      "context_before": "necessary incurring degraded performance. Additionally, modern DBMSs employ cost-based optimization, deriving Ptonsemantically equivalent query plans{P1,P2,...,P n}, and estimating the execution cost of each Piby the cost function C(P i)[17]. DBMSs choose and execute the optimal plan Po=argmin Pi∈PC(P i). Incorrect C() brings potential performance issues. 3 Motivating Study To investigate the potential of finding performance issues by code manipulation, we studied whether we can manipulate code to observe the unexpected performance of historical issues found by prior methods. Issue collection.",
+      "context_after": "APOLLOexamines performance regression across different versions of DBMSs;AMOEBAexpects semantic-equivalent queries to have similar performance;CERTchecks specifically incorrect estimated cardinalities, which refers to the estimated number of rows returned by a query plan;PUPPY compares performance under the default configuration and other configurations. In total, we identified and reproduced 10 confirmed or fixed performance issues.APOLLOhas 3 confirmed performance issues. Although it does not publicize the full list of found issues, we found these issues in DBMS bug trackers. However, due t",
+      "section": "3 Motivating Study",
+      "page": 3,
+      "char_offset": 12173,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "Table 1 shows the 10 performance issues from APOLLO,CERT,PUPPY, and AMOEBA.",
+      "context_before": "s 6 confirmed issues. Similar to APOLLO, we reproduced 5 issues.CERThas 11 confirmed issues. Unlike other methods, CERTspecifically detects inconsistent estimated cardinalities, instead of execution time. In Listing 7 of CERTpaper, authors provide two cases to show the execution time gap between pairs of queries due to incorrect cardinalities found, and we evaluated both cases.PUPPYhas 54 confirmed issues, but does not publicize the full list of its found issues. By searching the bug trackers of DBMSs, we found a report, which we believe is similar to the issues described in PUPPY paper. Results.",
+      "context_after": "Overall, we can observe the unexpected performance of 8 issues by the same code manipulation that flips code branches of IFstatements once. Specifically, we changed if (condition) toif (!condition). Both performance issues in APOLLOare due to the redundant initialization of hash tables and incorrect parallel computing. Flipping branches disables both, so we can observe the unexpected performance. Both performance issues in CERT affect the estimated cardinalities incurring an inefficient usage of HASH JOIN. Flipping branches changes the strategies of cardinality estimation to find both issues.PUP",
+      "section": "3 Motivating Study",
+      "page": 3,
+      "char_offset": 13662,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[27, 28]",
+      "technique": "norec",
+      "sentence": "In this paper, we collected 10 thousand randomly generated small workloads from SQLancer[27, 28], which is a popular DBMS testing tool.",
+      "context_before": "solutions for DBMS users and developers, respectively. As DBMS users lack expert knowledge of the DBMS′s internal code, we adopted a differential testing approach to evaluate functionality in practice. Differential testing involves comparing the results of executing the same inputs across different systems; inconsistencies in the results may indicate potential bugs. Specifically, we validate whether DBMS and DBMS′consistently produce identical results. If they do, we assume that DBMS′is functionally correct. The trustworthiness of this method increases with the number of test cases evaluated.",
+      "context_after": "As shown in our evaluation (Section 6), 96% of the DBMS′s that pass the above validation do not affect program functionalities. For DBMS developers, we can use specification-based validation approaches. Since developers have expert knowledge of the code, we expect developers to provide a specification of the query plans, which can be used for checking validity. For example, in PostgreSQL’s query plans, the operation GroupAggregate aggregates pre-sorted rows for a GROUP BYoperation, so its child operation must be Sort. Such a rule can be used to automatically check whether the query plans prod",
+      "section": "4 Approach",
+      "page": 5,
+      "char_offset": 23134,
+      "cited_reference": {
+        "number": 27,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Prem Devanbu, Myra B. Cohen, and Thomas Zimmermann, editors,ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of So",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[1, 16, 20, 21, 40]",
+      "technique": "cert",
+      "sentence": "These DBMSs have been extensively tested by previous works about finding performance issues [1, 16, 20, 21, 40].",
+      "context_before": "ctiveness. CanQueryZenfind previously unknown performance issues? Q.2 Necessity. WhetherQueryZenis necessary to find these performance issues? Q.3 Sensitivity. To what extentQueryZen’s components contribute to the finding of these performance issues? 6 Tested DBMSs.We tested PostgreSQL, MySQL, CockroachDB, and MariaDB. PostgreSQL is one of the most established DBMSs. MySQL is the most popular opensource DBMS [37]. MariaDB is another popular DBMS that was forked from MySQL. CockroachDB is a widely used enterprise-class DBMS, and its open version on GitHub has been starred more than 30.1k times.",
+      "context_after": "Recall thatQueryZenis not specific to any programming language, so we can test both C (PostgreSQL, MySQL, MaraiDB) and Go (CockroachDB) programming languages. We used the latest available development versions of tested DBMSs (PostgreSQL: 17.0, MySQL: 9.0, MariaDB: 11.7, CockroachDB: dcb0d27). We manipulated the code in specific folders (PostgreSQL: /app/postgres/src/backend/optimizer, MySQL:/app/mysqlserver/sql, CockroachDB:/app/cockroach/pkg/sql/opt, MariaDB:/app/mariadb-server/sql). The folders of PostgreSQL and CockroachDB include only the code of query optimizer, and the folders of MySQL",
+      "section": "6 Evaluation",
+      "page": 7,
+      "char_offset": 30474,
+      "cited_reference": {
+        "number": 1,
+        "text": "Jinsheng Ba and Manuel Rigger. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, pages 1–13, 2024.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "We considered the same performancetesting methods in Section 3:APOLLO[16],AMOEBA[21], CERT[1], and PUPPY[40].",
+      "context_before": "only reported issues when the observed performance gap exceeded 10%. In this way, we did not observe any false positives caused by performance fluctuation, as confirmed by DBMS developers. QueryZenfound 21 previously unknown performance issues on widely used benchmarks TPC-H and TPC-DS. Q.2 Necessity We evaluated whetherQueryZenis necessary to find these performance issues. Specifically, we examined 1) whether the performance issues found byQueryZencan be potentially found by previous performance-testing methods, and 2) whether these performance issues existed when prior methods were proposed.",
+      "context_after": "Issue reproducing.We evaluated whether the performance issues found byQueryZencan be potentially found by prior performance-testing methods APOLLO,AMOEBA,CERT, and PUPPY. Since these methods target different aspects of performance, such as cardinality estimation or configuration sensitivity, andQueryZenfocuses on source code manipulation, the sets of bugs they uncover are not directly comparable. 8 Table 3: The reproducibility of previous methods on the issues found byQueryZen. DBMS Issue IDAPOLLO AMOEBA CERTPUPPY PostgreSQL 1 PostgreSQL 2 PostgreSQL 3✓ PostgreSQL 4✓ PostgreSQL 5 MySQL 116456 M",
+      "section": "6 Evaluation",
+      "page": 8,
+      "char_offset": 40235,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "We evaluated whether the performance issues found byQueryZencan be potentially found by prior performance-testing methods APOLLO,AMOEBA,CERT, and PUPPY.",
+      "context_before": "by performance fluctuation, as confirmed by DBMS developers. QueryZenfound 21 previously unknown performance issues on widely used benchmarks TPC-H and TPC-DS. Q.2 Necessity We evaluated whetherQueryZenis necessary to find these performance issues. Specifically, we examined 1) whether the performance issues found byQueryZencan be potentially found by previous performance-testing methods, and 2) whether these performance issues existed when prior methods were proposed. We considered the same performancetesting methods in Section 3:APOLLO[16],AMOEBA[21], CERT[1], and PUPPY[40]. Issue reproducing.",
+      "context_after": "Since these methods target different aspects of performance, such as cardinality estimation or configuration sensitivity, andQueryZenfocuses on source code manipulation, the sets of bugs they uncover are not directly comparable. 8 Table 3: The reproducibility of previous methods on the issues found byQueryZen. DBMS Issue IDAPOLLO AMOEBA CERTPUPPY PostgreSQL 1 PostgreSQL 2 PostgreSQL 3✓ PostgreSQL 4✓ PostgreSQL 5 MySQL 116456 MySQL 116484✓ MySQL 116309✓ MySQL 116534 MySQL 116773 MySQL 116774✓ MySQL 116775 MySQL 116776✓ ✓ MySQL 116777✓ CockorachDB 134803 CockorachDB 135001✓ CockorachDB 136350✓",
+      "section": "6 Evaluation",
+      "page": 8,
+      "char_offset": 40363,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[2, 27, 28]",
+      "technique": "dqp",
+      "sentence": "DBMS Issue IDAPOLLO AMOEBA CERTPUPPY PostgreSQL 1 PostgreSQL 2 PostgreSQL 3✓ PostgreSQL 4✓ PostgreSQL 5 MySQL 116456 MySQL 116484✓ MySQL 116309✓ MySQL 116534 MySQL 116773 MySQL 116774✓ MySQL 116775 MySQL 116776✓ ✓ MySQL 116777✓ CockorachDB 134803 CockorachDB 135001✓ CockorachDB 136350✓ ✓ MariaDB 35280 MariaDB 35332 MariaDB 35331 MariaDB 35333✓ ✓ Sum (✓): 8 0 4 1 To address this, we conducted a qualitative analysis, a common approach in prior work [2, 27, 28].",
+      "context_before": "proposed. We considered the same performancetesting methods in Section 3:APOLLO[16],AMOEBA[21], CERT[1], and PUPPY[40]. Issue reproducing.We evaluated whether the performance issues found byQueryZencan be potentially found by prior performance-testing methods APOLLO,AMOEBA,CERT, and PUPPY. Since these methods target different aspects of performance, such as cardinality estimation or configuration sensitivity, andQueryZenfocuses on source code manipulation, the sets of bugs they uncover are not directly comparable. 8 Table 3: The reproducibility of previous methods on the issues found byQueryZen.",
+      "context_after": "Specifically, given a performance issue and both execution paths of a DBMS before and afterQueryZen’s code manipulation, 1) if the first lines of discrepancy execution paths are lastly updated in different commits, we assumed the issue can be found by APOLLO; 2) if the instrumented IFcondition is directly controlled by a SQL clause, we assumed there exists a pair of semantic-equivalent queries to execute both paths so that the issue can be found by AMOEBA; 3) if the first branches in discrepancy execution paths directly manipulate estimated cardinalities, we assumed the issue can be found by CER",
+      "section": "6 Evaluation",
+      "page": 9,
+      "char_offset": 40825,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. Keep it simple: Testing databases via differential query plans.Proceedings of the ACM on Management of Data, 2(3):1–26, 2024.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "Overall, 8, 0, 4, and 1 issues have the potential to be found by APOLLO,AMOEBA, CERT, and PUPPY, respectively.",
+      "context_before": "ranches in discrepancy execution paths directly manipulate estimated cardinalities, we assumed the issue can be found by CERT; 4) if the instrumented IFcondition is directly controlled by an external configuration, we assumed the issue can be found by PUPPY. These assumptions do not guarantee that prior methods can find these issues. For example, although the first lines of both execution paths are updated in different commits, it is not necessary that both versions of the DBMSs show the performance gap. Results.Table 3 shows whether prior methods can find the performance issues found byQueryZen.",
+      "context_after": "It demonstrates the necessity ofQueryZenfor finding these performance issues, and black-box methods are insufficient to explore optimization strategies for finding performance issues. The number of APOLLOis much higher than others, because developers typically submit a minimal code block in each commit, making it more likely to observe the code update in differentcommits. Issue existence date.Additionally, we evaluated whether these performance issues existed when prior performancetesting methods APOLLO,AMOEBA,CERT, and PUPPY were proposed. If so, it demonstrates that these performance issues w",
+      "section": "6 Evaluation",
+      "page": 9,
+      "char_offset": 42366,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "Additionally, we evaluated whether these performance issues existed when prior performancetesting methods APOLLO,AMOEBA,CERT, and PUPPY were proposed.",
+      "context_before": "Results.Table 3 shows whether prior methods can find the performance issues found byQueryZen. Overall, 8, 0, 4, and 1 issues have the potential to be found by APOLLO,AMOEBA, CERT, and PUPPY, respectively. It demonstrates the necessity ofQueryZenfor finding these performance issues, and black-box methods are insufficient to explore optimization strategies for finding performance issues. The number of APOLLOis much higher than others, because developers typically submit a minimal code block in each commit, making it more likely to observe the code update in differentcommits. Issue existence date.",
+      "context_after": "If so, it demonstrates that these performance issues were not found by prior methods with an extensive testing effort. We retrieved the last commit date of the code manipulatedQueryZen, and we assumed the performance issues existed no later than the date. Results.Figure 2 shows the last commit date associated with the 21 performance issues found byQueryZenand the publication date of the prior methods. Issues #116484, #116309, #116534, and #116773 share the same commit, as do issues #116776 and #116777. It is because both commits focus on tidying up code style, resulting in extensive modificat",
+      "section": "6 Evaluation",
+      "page": 9,
+      "char_offset": 42872,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "In total, 14, 19, 21, and 21 of 21 issues existed before the publication date of APOLLO,AMOEBA,CERT, and PUPPY.",
+      "context_before": "ound by prior methods with an extensive testing effort. We retrieved the last commit date of the code manipulatedQueryZen, and we assumed the performance issues existed no later than the date. Results.Figure 2 shows the last commit date associated with the 21 performance issues found byQueryZenand the publication date of the prior methods. Issues #116484, #116309, #116534, and #116773 share the same commit, as do issues #116776 and #116777. It is because both commits focus on tidying up code style, resulting in extensive modifications, including the manipulation of code across multiple issues.",
+      "context_after": "A large portion of performance issues already existed for a long time, while prior methods have not found them in practice. Most issues found byQueryZencannot be found by APOLLO,AMOEBA,CERT, and PUPPY, and existed before the publication of prior methods. Q.3 Sensitivity The soundness of code manipulations.QueryZenmanipulates all IFbranches and ensures that the flipped branches impact only performance, not functionalities, through differential testing. To evaluate its soundness, we manually examined the branches that passed the functionality testing. Since the total number of branches (e.g., in",
+      "section": "6 Evaluation",
+      "page": 9,
+      "char_offset": 43684,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "Most issues found byQueryZencannot be found by APOLLO,AMOEBA,CERT, and PUPPY, and existed before the publication of prior methods.",
+      "context_before": "te associated with the 21 performance issues found byQueryZenand the publication date of the prior methods. Issues #116484, #116309, #116534, and #116773 share the same commit, as do issues #116776 and #116777. It is because both commits focus on tidying up code style, resulting in extensive modifications, including the manipulation of code across multiple issues. In total, 14, 19, 21, and 21 of 21 issues existed before the publication date of APOLLO,AMOEBA,CERT, and PUPPY. A large portion of performance issues already existed for a long time, while prior methods have not found them in practice.",
+      "context_after": "Q.3 Sensitivity The soundness of code manipulations.QueryZenmanipulates all IFbranches and ensures that the flipped branches impact only performance, not functionalities, through differential testing. To evaluate its soundness, we manually examined the branches that passed the functionality testing. Since the total number of branches (e.g., in MySQL and MariaDB) is too large for exhaustive analysis, we randomly sampled 100 flipped branches and reviewed their associated code and comments, which typically describe the purpose of each branch or function. Our manual inspection showed that none of",
+      "section": "6 Evaluation",
+      "page": 9,
+      "char_offset": 43918,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[1, 21, 40]",
+      "technique": "cert",
+      "sentence": "Unlike existing approaches [1, 21, 40], which focus on manipulating external inputs,BFAoperates on the program code itself.",
+      "context_before": "and 22,922 additional branches in PostgreSQL, MySQL, and MariaDB, accounting for 9.32%, 1.66%, and 4.06% of all measured branches. Note that, during the branch coverage evaluation of MySQL, unexpected behaviors caused by manipulated code occasionally blocked data collection, resulting in incomplete results. The results showQueryZen substantially explores optimization space. Using more diverse workloads and code manipulations may further expand coverage. 7 Discussion Methodology.We propose a white-box testing methodology for identifying performance issues by directly manipulating program code.",
+      "context_after": "This approach offers several advantages. First, it enables systematic and fine-grained testing of program logic, overcoming the limitations of input-based methods that can only trigger behaviors in a constrained manner. As a result, our methodology uncovered 21 unique and previously unknown performance issues. Second, the code manipulated by BFAfacilitates issue debugging and analysis. Prior works compare the performance of executing input pairs, but it often remains unclear which part of the code incurs the performance difference. In contrast, our approach directly highlights the relevant cod",
+      "section": "7 Discussion",
+      "page": 11,
+      "char_offset": 50257,
+      "cited_reference": {
+        "number": 1,
+        "text": "Jinsheng Ba and Manuel Rigger. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, pages 1–13, 2024.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[1]",
+      "technique": "cert",
+      "sentence": "proposed CERT[1] to find performance issues by testing specifically cardinality estimation.",
+      "context_before": "uire manual effort from developers. Second, the effectiveness of BFAdepends on the quality of the underlying workload. While we use TPC-H and TPC-DS to represent common scenarios, other types of workloads might trigger different optimizations. Finally, our evaluation was conducted under a fixed system configuration. Although we observed consistent performance gaps across repeated trials, different hardware or memory settings may affect reproducibility. 8 Related Work Testing performance issues.The most related strand of research is on finding performance issues in DBMSs automatically. Baet al.",
+      "context_after": "Jung et al. proposed APOLLO[16], which compares the execution times of a query on two versions of a database system to find performance regression bugs. Liuet al. proposed AMOEBA[21], which compares the execution time of a semantically-equivalent pair of queries to identify an unexpected slowdown. Wuet al. proposed PUPPY[40], which examines whether performance improves after adjusting default configurations. In addition, several fuzzing-based approaches explore performance issues in general software. 11 SlowFuzz [26] searches for slow executions by maximizing total execution path length, while",
+      "section": "8 Related Work",
+      "page": 11,
+      "char_offset": 53902,
+      "cited_reference": {
+        "number": 1,
+        "text": "Jinsheng Ba and Manuel Rigger. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, pages 1–13, 2024.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Oracles such as PQS [29], NoREC [27],and TLP [28] detect logic bugs in SELECT statement implementations, while DQE [31] targets logic errors in UPDATE and INSERT statements.",
+      "context_before": "Atargets a broader class of performance issues in a more systematic way. DBMS testing.Beyond performance testing, automated methods have been developed to uncover various DBMS bugs. Many fuzzing tools target security-critical issues, such as memory errors. Tools like SQLSmith [35], Griffin [8], DynSQL [14], and ADUSA [21] employ grammar-based techniques to generate test cases aimed at detecting memoryrelated vulnerabilities. Inspired by grey-box fuzzers like AFL [34], Squirrel [41] uses code coverage feedback to identify memory issues. Other approaches focus on discovering logic bugs in DBMSs.",
+      "context_after": "Similar to testing approaches that find bugs, our core contribution is a novel testing methodology that finds performance issues by manipulating program code. 9 Conclusion In this paper, we have proposedBranch Flip Analysis(BFA), a generalwhite-boxmethod to find performance issues by flipping branches of IFstatements in source code. As the default optimization strategies should ideally have optimal performance, we disable or enforce an optimization by flipping code branches to derive a reference DBMS, whose performance is expected to be no better than the original DBMS. BFAprovides a new angle",
+      "section": "8 Related Work",
+      "page": 12,
+      "char_offset": 57258,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M10",
+        "M13"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:49:05Z",
+    "is_model_written": true,
+    "summary": "QueryZen looks for performance issues in code the optimiser rarely reaches. Its starting observation is that 8 of 10 historical performance issues found by APOLLO, AMOEBA, CERT and PUPPY could have been reached by flipping a branch, so instead of manipulating the input it manipulates the program: branches are flipped to force execution down dormant paths, and the resulting slowdowns are examined. On PostgreSQL, MySQL, MariaDB and CockroachDB it found 21 issues, most of which the prior methods cannot reach and which pre-dated their publication.",
+    "narrative": "SQLancer is used two ways. It supplies the workload: the authors collected ten thousand randomly generated small workloads from it, described as a popular DBMS testing tool, and those are the queries QueryZen runs against the flipped branches. Separately, CERT is one of four prior performance-testing methods QueryZen is measured against, and the comparison is reported issue by issue -- of the 21 issues found, 4 could potentially have been found by CERT, and 21 of 21 already existed when CERT was published. Note that SQLancer here generates input for a different technique rather than being the thing QueryZen is built on.",
+    "roles": {
+      "M1": "definition",
+      "M2": "motivation",
+      "M3": "background",
+      "M4": "result_comparison",
+      "M5": "baseline",
+      "M6": "motivation",
+      "M7": "reuse_component",
+      "M8": "background",
+      "M9": "baseline",
+      "M10": "result_comparison",
+      "M11": "result_comparison",
+      "M12": "result_comparison",
+      "M13": "result_comparison",
+      "M14": "result_comparison",
+      "M15": "result_comparison",
+      "M16": "background",
+      "M17": "definition",
+      "M18": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M7"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M7",
+            "sentence": "In this paper, we collected 10 thousand randomly generated small workloads from SQLancer[27, 28], which is a popular DBMS testing tool.",
+            "section": "4 Approach",
+            "page": 5
+          }
+        ],
+        "reasoning": "M7 states that ten thousand randomly generated small workloads were collected from SQLancer. That is reuse, but only as a source of queries for a technique that operates on the DBMS's own code -- QueryZen is not built on SQLancer.",
+        "reuse_kind": "workload"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Flipping branches to reach dormant code is not a development of any SQLancer oracle; the paper positions it against input-manipulating approaches including CERT."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M5",
+          "M9",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M5",
+            "sentence": "We considered the performance issues found by the four existing performance-testing methods:APOLLO[16],AMOEBA[21],CERT[1], and PUPPY[40].",
+            "section": "3 Motivating Study",
+            "page": 3
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "We considered the same performancetesting methods in Section 3:APOLLO[16],AMOEBA[21], CERT[1], and PUPPY[40].",
+            "section": "6 Evaluation",
+            "page": 8
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "Overall, 8, 0, 4, and 1 issues have the potential to be found by APOLLO,AMOEBA, CERT, and PUPPY, respectively.",
+            "section": "6 Evaluation",
+            "page": 9
+          }
+        ],
+        "reasoning": "M5 and M9 name CERT among the four prior performance-testing methods considered, and M12 reports how many of QueryZen's issues each could potentially have found.",
+        "techniques": [
+          "cert"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is called a popular DBMS testing tool and CERT is described by what it does; neither is presented as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2606_11132.json b/_data/papers/paper_arxiv_2606_11132.json
new file mode 100644
index 0000000..20281a1
--- /dev/null
+++ b/_data/papers/paper_arxiv_2606_11132.json
@@ -0,0 +1,599 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:29Z",
+  "paper": {
+    "id": "paper:arxiv:2606.11132",
+    "title": "Operationalizing Property-Based Testing for Data-Intensive Scalable Computing Systems",
+    "authors": [
+      "Yaoxuan Wu",
+      "I. Lee",
+      "Ahmad Humayun",
+      "Muhammad Ali Gulzar",
+      "Miryung Kim"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2606.11132",
+    "s2_paper_id": "09e050c29b5f2592ad6d0b2a90121bb1530ead84",
+    "url": "https://arxiv.org/abs/2606.11132",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2606.11132",
+      "retrieved_at": "2026-09-09T01:29:29Z",
+      "chars": 68607,
+      "content_sha256": "sha256:49b5f6170e5d0aa8f1f027660ed23f0a2373c230fc47b5596fa0190247129d42"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2009
+      },
+      {
+        "number": "2",
+        "title": "Background and Motivating Example",
+        "start": 9473
+      },
+      {
+        "number": "2.1",
+        "title": "Background on DISC Workloads",
+        "start": 9509
+      },
+      {
+        "number": "2.2",
+        "title": "Motivating Example",
+        "start": 13521
+      },
+      {
+        "number": "3",
+        "title": "Meta Properties",
+        "start": 18542
+      },
+      {
+        "number": "3.2",
+        "title": "Data Decomposition",
+        "start": 27634
+      },
+      {
+        "number": "3.3",
+        "title": "Computation Decomposition",
+        "start": 32139
+      },
+      {
+        "number": "3.4",
+        "title": "Bag and Cardinality Relations",
+        "start": 34156
+      },
+      {
+        "number": "4",
+        "title": "Operationalizing Meta-Properties into PBT",
+        "start": 35696
+      },
+      {
+        "number": "4.1",
+        "title": "Workload Skeleton Generation",
+        "start": 38014
+      },
+      {
+        "number": "4.2",
+        "title": "Source Schema and Data Generation",
+        "start": 39601
+      },
+      {
+        "number": "4.3",
+        "title": "Schema-Aware Instantiation",
+        "start": 41506
+      },
+      {
+        "number": "4.4",
+        "title": "Property Checking",
+        "start": 45576
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 46716
+      },
+      {
+        "number": "5.1",
+        "title": "RQ1: Semantic Drifts and Pitfalls",
+        "start": 47404
+      },
+      {
+        "number": "5.2",
+        "title": "RQ2: Workflow Complexity and Coverage",
+        "start": 51263
+      },
+      {
+        "number": "6",
+        "title": "Related Work",
+        "start": 55898
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 60680
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Sqlsmith,” 2022. [Online]. Available: https://github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "——, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” inProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "——, “Testing database engines via pivoted query synthesis,” in14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682. Yaoxuan Wu, Ingrid Lee, Ahmad Humayun, Muhammad Ali Gulzar, and Miryung Kim",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "C. Zhang, L. Wang, and M. Rigger, “Finding cross-rule optimization bugs in datalog engines,”Proceedings of the ACM on Programming Languages, vol. 8, no. OOPSLA1, pp. 110–136, 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "B. Ghit, N. Poggi, J. Rosen, R. Xin, and P. Boncz, “Sparkfuzz: searching correctness regressions in modern query engines,” inProceedings of the workshop on Testing Database Systems, 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "M. E. Kroner, “Achilles’ spear: Using metamorphic testing to find bugs in stream processing engines,” inDatenbanksysteme für Business, Technologie und Web (BTW 2025). Gesellschaft für Informatik, Bonn, 2025, pp. 1031–1042.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "M. Zaharia, M. Chowdhury, M. J. Franklin, S. Shenker, and I. Stoica, “Spark: Cluster computing with working sets,” in2nd USENIX workshop on hot topics in cloud computing (HotCloud 10), 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "“SPARK-33726: Limit and offset ordering behaves differently between dataframe and sql,” https://issues.apache.org/jira/browse/SPARK-33726, accessed 2026-0128.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "“SPARK-49000: Aggregation distinct returning wrong results with literal value,” https://issues.apache.org/jira/browse/SPARK-49000, accessed 2026-01-28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "K. Claessen and J. Hughes, “Quickcheck: a lightweight tool for random testing of haskell programs,” inProceedings of the fifth ACMSIGPLAN international conference on Functional programming, 2000, pp. 268–279.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Apache DataFusion Comet Developers, “Apache DataFusion Comet: Fuzz Testing,” https://github.com/apache/datafusion-comet/tree/ 03e833b955d369f994d9652026ca3c1eb641acac/fuzz-testing.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "“JaCoCo,” https://www.eclemma.org/jacoco/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "“Scalacheck: Property-based testing for scala,” https://scalacheck.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "“Fscheck: Random testing for .net. ” [Online]. Available: https://fscheck.github.io/ FsCheck/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "“test.check: A property-based testing library for clojure.” [Online]. Available: https://clojure.org/guides/test_check_beginner",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "“Hypothesis: Property-based testing for python.” [Online]. Available: https: //hypothesis.readthedocs.io/en/latest/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "“jqwik: Property-based testing in java.” [Online]. Available: https://jqwik.net/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "“hedgehog: Release with confidence.” [Online]. Available: https://hackage. haskell.org/package/hedgehog",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "“Quickcheck.” [Online]. Available: https://www.quviq.com/documentation/eqc/ overview-summary.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "L. O’Connor and O. Wickström, “Quickstrom: property-based acceptance testing with ltl specifications,” inProceedings of the 43rd ACMSIGPLAN International Conference on Programming Language Design and Implementation, 2022, pp. 1025– 1038.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "J. Hughes, “Experiences with quickcheck: testing the hard stuff and staying sane,” inA List of Successes That Can Change the World: Essays Dedicated to Philip Wadler on the Occasion of His 60th Birthday. Springer, 2016, pp. 169–186.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "H. Goldstein, J. W. Cutler, D. Dickstein, B. C. Pierce, and A. Head, “Property-based testing in practice, ” inProceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Q. Zhang, J. Wang, M. A. Gulzar, R. Padhye, and M. Kim, “Bigfuzz: Efficient fuzz testing for data analytics using framework abstraction, ” inProceedings of the 35th IEEE/ACM international conference on automated software engineering, 2020, pp. 722–733.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "A. Humayun, M. Kim, and M. A. Gulzar, “Co-dependence aware fuzzing for dataflow-based big data analytics, ” inProceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "M. A. Gulzar, S. Mardani, M. Musuvathi, and M. Kim, “White-box testing of big data analytics with complex user-defined functions,” inProceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2019, pp. 290–301.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "A. Humayun, Y. Wu, M. Kim, and M. A. Gulzar, “Naturalfuzz: Natural input generation for big data analytics,” in2023 38th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 2023, pp. 1592–1603.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Y. Wu, A. Humayun, M. A. Gulzar, and M. Kim, “Natural symbolic executionbased testing for big data analytics,” inProceedings of the ACM International Conference on the Foundations of Software Engineering (FSE), 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "A. Riesco and J. Rodríguez-Hortalá, “Property-based testing for spark streaming,” arXiv preprint arXiv:1812.11838, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "C. V. Espinosa, E. Martin-Martin, A. Riesco, and J. Rodriguez-Hortala, “Flinkcheck: Property-based testing for apache flink,”IEEE Access, vol. 7, pp. 150 369–150 382, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "K. Kallas, F. Niksic, C. Stanford, and R. Alur, “Diffstream: Differential output testing for stream processing programs,”Proc. ACM Program. Lang., vol. 4, no. OOPSLA, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "“Sqlancer: Automatic testing of database systems,” https://github.com/sqlancer/ sqlancer, accessed 2026-01-28.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 33,
+      "text": "C. Zhang and M. Rigger, “Constant optimization driven database system testing,” Proceedings of the ACM on Management of Data, vol. 3, no. 1, pp. 1–24, 2025.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 34,
+      "text": "Q. Mang, J. Ba, P. He, and M. Rigger, “Finding logic bugs in graph-processing systems via graph-cutting,” inProceedings of the ACM on Management of Data(SIGMOD), 2025.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 35,
+      "text": "Y. Wu, I. Lee, A. Humayun, M. A. Gulzar, and M. Kim, “Replication Package for “Operationalizing Property-Based Testing for Data-Intensive Scalable Computing Systems”,” 2026. [Online]. Available: https://doi.org/10.5281/zenodo.19248115",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 3,
+      "text": "——, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” inProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 4,
+      "text": "——, “Testing database engines via pivoted query synthesis,” in14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682. Yaoxuan Wu, Ingrid Lee, Ahmad Humayun, Muhammad Ali Gulzar, and Miryung Kim",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 5,
+      "text": "C. Zhang, L. Wang, and M. Rigger, “Finding cross-rule optimization bugs in datalog engines,”Proceedings of the ACM on Programming Languages, vol. 8, no. OOPSLA1, pp. 110–136, 2024.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 32,
+      "text": "“Sqlancer: Automatic testing of database systems,” https://github.com/sqlancer/ sqlancer, accessed 2026-01-28.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 33,
+      "text": "C. Zhang and M. Rigger, “Constant optimization driven database system testing,” Proceedings of the ACM on Management of Data, vol. 3, no. 1, pp. 1–24, 2025.",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 34,
+      "text": "Q. Mang, J. Ba, P. He, and M. Rigger, “Finding logic bugs in graph-processing systems via graph-cutting,” inProceedings of the ACM on Management of Data(SIGMOD), 2025.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 2–5]",
+      "technique": "tlp",
+      "sentence": "Prior work has demonstrated the value of semantic oracles based on rewrites and decompositions [ 2–5].",
+      "context_before": "h primarily relies on shallow failure signals such as crashes or uncaught exceptions. Property-based testing (PBT) [ 11] offers a complementary direction by shifting testing from checking one manually specified input-output behavior to checking whether generated workloads and inputs satisfy a generalized semantic property. With suitable properties and sufficiently diverse generated tests, this style of checking can exercise behaviors that are difficult to observe through crash-based fuzzing alone.This direction has proved effective in the adjacent DBMS testing domain, as summarized in Table 1.",
+      "context_after": "In contrast, engine-level testing for DISC systems remains limited, and the few existing efforts support only restricted oracle forms or narrow workload templates [ 7]. However, DISC semantic testing is not merely a direct reuse of existing DBMS oracle ideas. DISC workloads are expressed as compositional DataFrame workflows with rich features, including higher-order expressions, semi-structured data (e.g., arrays and maps), and Python UDFs. These features make semantic checks more dependent on workload context, schema compatibility, and cross-runtime behavior than in many prior SQL-centric te",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2947,
+      "cited_reference": {
+        "number": 2,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Work SUT Workload Gen Data Gen Semantic Oracles DB Fuzzing and Testing SQLsmith [1] DB engine✓ ✗None (failure/crash) SQLancer [2–5] DB engine Templ.",
+      "context_before": "r, and Miryung Kim Table 1: Comparison of testing approaches for Database and DISC systems. While DBMS testing has leveraged PBT-style checks through a limited set of metamorphic relations, mainly for query decomposition and equivalence rewriting,DiscPBT introduces the most comprehensive PBT framework for DISC. By systematizing eight abstract meta-properties, rather than the few used in prior work,DiscPBTprovides strong oracles and coordinated workload and data generation for complex multi-stage DISC pipelines. Oracle patterns are abbreviated as P1–P8; their full definitions appear in Table 2.",
+      "context_after": "✓P3, P5, P6, P8 DISC Fuzzing and Testing SparkFuzz [6] Spark engine✓ ✓None (differential testing) Achilles’ SPEar [7] Spark engine Templ.✓P4, P5 Our approach DiscPBTSpark engine✓ ✓P1–P8 Spark, the abstraction of reusable meta-properties paired with explicit realization logic is not specific to Spark alone.DiscPBTsystematizes oracle forms that are especially important in DISC workloads, including higher-order expression rewrites, UDF/native counterpart equivalences, aggregation decompositions, and operatorlocal semantic relations. It then turns these meta-properties into executable tests throug",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6219,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We also do not compare against SQLancer because using it for Spark would require an additional adapter.",
+      "context_before": "against CometFuzz [ 12], a fuzzing baseline that generates Spark SQL queries. CometFuzz uses a Spark Yaoxuan Wu, Ingrid Lee, Ahmad Humayun, Muhammad Ali Gulzar, and Miryung Kim SQL frontend, whileDiscPBTgenerates PySpark DataFrame workloads, but both ultimately exercise Spark’s analysis, optimization, planning, and execution pipeline. This makes the comparison useful for evaluating whether DISC-style workloads provide additional structural diversity and coverage benefits beyond SQL-oriented workload generation. We do not compare against SparkFuzz because its replication package is unavailable.",
+      "context_after": "Experimental setup.We run both tools on the same AWS r7i.4xlarge machine under PySpark 4.1.1, with shuffle partitions set to 2 and ANSI mode disabled. We use CometFuzz under its default configuration, where each generated query is a single-layer SQL query without nesting over three Parquet tables with 200 rows each. We apply the same 30-second timeout to both tools and use a fixed budget of 10,000 generated programs. For coverage, we use JaCoCo [ 13] for collection and focus on Spark’s Catalyst (optimization) and execution packages rather than the entire Spark codebase, because our goal is to",
+      "section": "5.2 RQ2: Workflow Complexity and Coverage",
+      "page": 10,
+      "char_offset": 51931,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In relational DBMS testing, SQLancer [ 32] demonstrated the effectiveness of query oracles, including TLP [ 2], which checks result preservation under predicate-based partitioning, and NoREC [ 3], which validates queries through equivalence-preserving rewrites that disable optimizer-dependent execution paths.",
+      "context_before": "the broader semantic space of general DataFrame workflows. Achilles’ SPEar [ 7] brings metamorphic testing to DataFrame workflows and is therefore closer to our setting, but focuses on decomposition-style testing templates. In contrast,DiscPBTtargets reusable semantic testing principles for DISC workloads beyond SQL, including richer workflow structure and a broader range of operators and semantic features. Oracle-Based Testing in DBMSs, Datalog, and Graph Systems.Adjacent data-processing domains have developed a rich set of semantic oracle techniques, many of which can also be viewed as PBT.",
+      "context_after": "CODDTest [ 33] explores constant-based subquery rewrites as a semantic test oracle. In Datalog testing, IRE checks cross-rule optimizations through computation composition [ 5]. In graph-system testing, GSlicer [ 34] derives metamorphic checks by partitioning graph inputs to validate graph libraries and graph databases. These works show that effective semantic testing often relies on decomposition, equivalence-preserving rewrites, or algebraic relations rather than fixed expected outputs.DiscPBTis inspiredby this perspective, but adapts it to DISC systems, whose workloads are expressed as Dat",
+      "section": "6 Related Work",
+      "page": 11,
+      "char_offset": 59457,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "CODDTest [ 33] explores constant-based subquery rewrites as a semantic test oracle.",
+      "context_before": "ond SQL, including richer workflow structure and a broader range of operators and semantic features. Oracle-Based Testing in DBMSs, Datalog, and Graph Systems.Adjacent data-processing domains have developed a rich set of semantic oracle techniques, many of which can also be viewed as PBT. In relational DBMS testing, SQLancer [ 32] demonstrated the effectiveness of query oracles, including TLP [ 2], which checks result preservation under predicate-based partitioning, and NoREC [ 3], which validates queries through equivalence-preserving rewrites that disable optimizer-dependent execution paths.",
+      "context_after": "In Datalog testing, IRE checks cross-rule optimizations through computation composition [ 5]. In graph-system testing, GSlicer [ 34] derives metamorphic checks by partitioning graph inputs to validate graph libraries and graph databases. These works show that effective semantic testing often relies on decomposition, equivalence-preserving rewrites, or algebraic relations rather than fixed expected outputs.DiscPBTis inspiredby this perspective, but adapts it to DISC systems, whose workloads are expressed as DataFrame workflows rather than isolated declarative queries and heavily exercise DAG s",
+      "section": "6 Related Work",
+      "page": 11,
+      "char_offset": 59768,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 5]",
+      "technique": null,
+      "sentence": "In Datalog testing, IRE checks cross-rule optimizations through computation composition [ 5].",
+      "context_before": "mantic features. Oracle-Based Testing in DBMSs, Datalog, and Graph Systems.Adjacent data-processing domains have developed a rich set of semantic oracle techniques, many of which can also be viewed as PBT. In relational DBMS testing, SQLancer [ 32] demonstrated the effectiveness of query oracles, including TLP [ 2], which checks result preservation under predicate-based partitioning, and NoREC [ 3], which validates queries through equivalence-preserving rewrites that disable optimizer-dependent execution paths. CODDTest [ 33] explores constant-based subquery rewrites as a semantic test oracle.",
+      "context_after": "In graph-system testing, GSlicer [ 34] derives metamorphic checks by partitioning graph inputs to validate graph libraries and graph databases. These works show that effective semantic testing often relies on decomposition, equivalence-preserving rewrites, or algebraic relations rather than fixed expected outputs.DiscPBTis inspiredby this perspective, but adapts it to DISC systems, whose workloads are expressed as DataFrame workflows rather than isolated declarative queries and heavily exercise DAG structure, higher-order operators, semi-structured data, and Python UDFs.DiscPBTorganizes these",
+      "section": "6 Related Work",
+      "page": 11,
+      "char_offset": 59852,
+      "cited_reference": {
+        "number": 5,
+        "text": "C. Zhang, L. Wang, and M. Rigger, “Finding cross-rule optimization bugs in datalog engines,”Proceedings of the ACM on Programming Languages, vol. 8, no. OOPSLA1, pp. 110–136, 2024.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 34]",
+      "technique": null,
+      "sentence": "In graph-system testing, GSlicer [ 34] derives metamorphic checks by partitioning graph inputs to validate graph libraries and graph databases.",
+      "context_before": "ssing domains have developed a rich set of semantic oracle techniques, many of which can also be viewed as PBT. In relational DBMS testing, SQLancer [ 32] demonstrated the effectiveness of query oracles, including TLP [ 2], which checks result preservation under predicate-based partitioning, and NoREC [ 3], which validates queries through equivalence-preserving rewrites that disable optimizer-dependent execution paths. CODDTest [ 33] explores constant-based subquery rewrites as a semantic test oracle. In Datalog testing, IRE checks cross-rule optimizations through computation composition [ 5].",
+      "context_after": "These works show that effective semantic testing often relies on decomposition, equivalence-preserving rewrites, or algebraic relations rather than fixed expected outputs.DiscPBTis inspiredby this perspective, but adapts it to DISC systems, whose workloads are expressed as DataFrame workflows rather than isolated declarative queries and heavily exercise DAG structure, higher-order operators, semi-structured data, and Python UDFs.DiscPBTorganizes these oracle-design principles as reusable meta-properties and systematically instantiates them into executable checks for DISC workloads. 7 Conclusi",
+      "section": "6 Related Work",
+      "page": 11,
+      "char_offset": 59946,
+      "cited_reference": {
+        "number": 34,
+        "text": "Q. Mang, J. Ba, P. He, and M. Rigger, “Finding logic bugs in graph-processing systems via graph-cutting,” inProceedings of the ACM on Management of Data(SIGMOD), 2025.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M3"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:24:46Z",
+    "is_model_written": true,
+    "summary": "DiscPBT is a property-based testing engine for Apache Spark. The authors argue that fuzzing's shallow oracles catch crashes but miss semantic drift and optimisation errors in data-intensive scalable computing frameworks, and that operationalising property-based testing there needs reusable property definitions and a way to instantiate them into valid workloads. DiscPBT supplies eight reusable meta-properties and generators for workload skeletons and input data, reaching 1.2x the branch coverage and 1153x the plan diversity of CometFuzz.",
+    "narrative": "SQLancer is the demonstration the paper builds on -- credited with showing the effectiveness of query oracles including TLP -- and is explicitly excluded from the evaluation: the authors say they do not compare against it because using it for Spark would require an additional adapter.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "state_of_the_art",
+      "M5": "definition",
+      "M6": "background",
+      "M7": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M3 states the opposite of a comparison: SQLancer was deliberately not run, because applying it to Spark would need an adapter. A pattern fired on that sentence, which is a comparison only in the sense of explaining its absence."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M4"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "In relational DBMS testing, SQLancer [ 32] demonstrated the effectiveness of query oracles, including TLP [ 2], which checks result preservation under predicate-based partitioning, and NoREC [ 3], which validates queries through equivalence-preserving rewrites that disable optimizer-dependent execution paths.",
+            "section": "6 Related Work",
+            "page": 11
+          }
+        ],
+        "reasoning": "M4 credits SQLancer with demonstrating the effectiveness of query oracles in relational DBMS testing, which is a claim about its standing."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2606_14164.json b/_data/papers/paper_arxiv_2606_14164.json
new file mode 100644
index 0000000..5086352
--- /dev/null
+++ b/_data/papers/paper_arxiv_2606_14164.json
@@ -0,0 +1,234 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:11:07Z",
+  "paper": {
+    "id": "paper:arxiv:2606.14164",
+    "title": "Investigating Metamorphic Fuzz Oracle Enhancement via Large Language Models",
+    "authors": [
+      "Ruixiang Qian",
+      "Ding Yang",
+      "Zengxu Chen",
+      "Yue Gao",
+      "Chunrong Fang",
+      "Chao Zhang",
+      "Zhenyu Chen"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": null,
+    "arxiv_id": "2606.14164",
+    "s2_paper_id": "101f3528945d4a6501c451e0c481481b440a3c91",
+    "url": "https://arxiv.org/abs/2606.14164",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2606.14164",
+      "retrieved_at": "2026-09-07T17:11:07Z",
+      "chars": 116889,
+      "content_sha256": "sha256:413e8c5721eee22a91c25e3d14679d55a746897bc47f817386655c3436399cbf"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 28,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 4425
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 12103
+      },
+      {
+        "number": "2.1",
+        "title": "Greybox Fuzzing",
+        "start": 12433
+      },
+      {
+        "number": "2.2",
+        "title": "Metamorphic Testing",
+        "start": 18729
+      },
+      {
+        "number": "3",
+        "title": "Metamorphic-based Fuzz Oracle Enhancement",
+        "start": 21032
+      },
+      {
+        "number": "4",
+        "title": "Study Design",
+        "start": 23963
+      },
+      {
+        "number": "4.1",
+        "title": "MetaFOE",
+        "start": 25367
+      },
+      {
+        "number": "4.2",
+        "title": "Research Questions",
+        "start": 30246
+      },
+      {
+        "number": "4.3",
+        "title": "Fuzz Drivers",
+        "start": 33622
+      },
+      {
+        "number": "4.4",
+        "title": "Large Language Models",
+        "start": 34457
+      },
+      {
+        "number": "4.5",
+        "title": "Prompting Method",
+        "start": 36686
+      },
+      {
+        "number": "5",
+        "title": "Results",
+        "start": 52651
+      },
+      {
+        "number": "5.1",
+        "title": "RQ1: Existing Fuzz Oracles",
+        "start": 54986
+      },
+      {
+        "number": "5.2",
+        "title": "RQ2: Evaluation of LLM-based MRG",
+        "start": 61364
+      },
+      {
+        "number": "5.3",
+        "title": "RQ3: Evaluation of LLM-based FOI",
+        "start": 63282
+      },
+      {
+        "number": "5.4",
+        "title": "RQ4: LLM Sensitivity",
+        "start": 72476
+      },
+      {
+        "number": "5.5",
+        "title": "RQ5: Prompt Sensitivity",
+        "start": 76315
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 79106
+      },
+      {
+        "number": "6.1",
+        "title": "Case Study",
+        "start": 79270
+      },
+      {
+        "number": "6.2",
+        "title": "Lessons Learned",
+        "start": 86586
+      },
+      {
+        "number": "6.3",
+        "title": "Threats to Validity",
+        "start": 88108
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 89270
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 92313
+      },
+      {
+        "number": "9",
+        "title": "Data-availability Statement",
+        "start": 93261
+      }
+    ]
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "author_year_citation",
+      "surface": "Rigger and Su 2020",
+      "technique": null,
+      "sentence": "2025; Rigger and Su 2020; Segura et al.",
+      "context_before": "inspected drivers implement functional oracles that target the libraries under test, highlighting the critical need to improve existing fuzz oracles. In this paper, we conduct the first study investigating the enhancement of fuzz oracles, in particular those embedded in existing drivers. In the prior art, researchers have recognized the shortcomings of fuzz oracles and have proposed novel techniques to tackle this problem [Han and Zhou 2020; Serebryany et al .2012; Zhang et al .2021, 2022]. As a powerful technique for alleviating the oracle problem [Guo et al .2024; Liu et al .2013; Mu et al .",
+      "context_after": "2018], metamorphic testing has naturally attracted the attention of the fuzzing community and has been adopted to enhance fuzz oracles. For example, Pan et al .[2024] summarize the characteristics of excessive data exposures (EDEs) and design novel oracles to fuzz and expose EDEs in realworld web applications. Kokkonis et al .[2025] identify the symptoms of code-level backdoors and construct a metamorphic (test) oracle to assist AFL++ [AFL++Team 2026] in detecting runtime backdoor triggers. These works (which can be generally termedmetamorphic fuzzing) have proposed insightful solutions to th",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7253,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "A study of augmenting fuzz drivers with metamorphic oracles rather than crash-based ones, which the authors say ignore library functionality and limit what can be found. Their framework, MetaFOE, uses an LLM to generate metamorphic relations and integrate them into drivers automatically. On OSS-Fuzz drivers it generated 3,475 relations, 77.3% applicable, and raised edge coverage by 18.7% on average.",
+    "narrative": "One author-year citation, among the metamorphic testing work the paper builds from.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2607_03741.json b/_data/papers/paper_arxiv_2607_03741.json
new file mode 100644
index 0000000..7abf56d
--- /dev/null
+++ b/_data/papers/paper_arxiv_2607_03741.json
@@ -0,0 +1,685 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:20Z",
+  "paper": {
+    "id": "paper:arxiv:2607.03741",
+    "title": "Graph-Aware Fuzzing for Graph Database Management Systems",
+    "authors": [
+      "Yu Li",
+      "Qiang Hu",
+      "Yao Zhang",
+      "Junjie Wang",
+      "Haoxuan Liu",
+      "Rui Wang",
+      "Yongqiang Lyu"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "arxiv_id": "2607.03741",
+    "s2_paper_id": "d6433b58411aed7605dd789072997fc0884bd610",
+    "url": "https://arxiv.org/abs/2607.03741",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2607.03741",
+      "retrieved_at": "2026-09-09T01:29:20Z",
+      "chars": 61813,
+      "content_sha256": "sha256:ed43e1de1d15222a0acdd7107f16e67cca6f0ecad23b2024119f498e4a599d85"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2187
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 10407
+      },
+      {
+        "number": "A",
+        "title": "Cypher Query Language",
+        "start": 10422
+      },
+      {
+        "number": "B",
+        "title": "Graph Topology and Execution Behavior",
+        "start": 12253
+      },
+      {
+        "number": "III",
+        "title": "DESIGN OFGRAF",
+        "start": 14477
+      },
+      {
+        "number": "A",
+        "title": "Overview",
+        "start": 14496
+      },
+      {
+        "number": "B",
+        "title": "Graph Context Aware Query Generation",
+        "start": 16090
+      },
+      {
+        "number": "C",
+        "title": "Execution State Guided Query Mutation",
+        "start": 25010
+      },
+      {
+        "number": "IV",
+        "title": "IMPLEMENTATION",
+        "start": 30209
+      },
+      {
+        "number": "V",
+        "title": "EVALUATION",
+        "start": 32238
+      },
+      {
+        "number": "A",
+        "title": "Experimental Setup",
+        "start": 32935
+      },
+      {
+        "number": "B",
+        "title": "RQ1: Code Coverage",
+        "start": 34514
+      },
+      {
+        "number": "C",
+        "title": "RQ2: Query Validity",
+        "start": 37692
+      },
+      {
+        "number": "D",
+        "title": "RQ3: Component Contribution",
+        "start": 39461
+      },
+      {
+        "number": "E",
+        "title": "RQ4: Bug Detection",
+        "start": 42087
+      },
+      {
+        "number": "VI",
+        "title": "RELATED WORK",
+        "start": 45790
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 50508
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Neo4j, Inc. Neo4j, 2010. Accessed: January 08, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Memgraph Inc. Memgraph, 2020. Accessed: January 08, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Redis Labs. Redisgraph, 2018. Accessed: January 08, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Donald D. Chamberlin. Relational data-base management systems.ACM Comput. Surv., 8(1):43–66, March 1976.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Kwo-Jean Farn and Shin-Ling Hu. Practical issues for rdbms application development. InProceedings of the Eleventh International Conference on Data Engineering, pages 353–, 1995.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Paul Resnick and Hal R. Varian. Recommender systems.Commun. ACM, 40(3):56–58, March 1997.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. Sqlsmith: a random sql query generator. https://github.com/anse1/sqlsmith, 2018. [Online; accessed 26-March-2026].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang., 4(OOPSLA), November 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 9,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. InProceedings of the 14th USENIX Conference on Operating Systems Design and Implementation, OSDI’20, USA, 2020. USENIX Association.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 10,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE ’20, page 1140–1152. ACM, November 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 11,
+      "text": "PostgreSQL Global Development Group. Postgresql, 1996. Accessed: March 26, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "SQLite Development Team. Sqlite, 2000. Accessed: March 26, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. Squirrel: Testing database management systems with language validity and coverage feedback. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, CCS ’20, page 955–970, New York, NY, USA, 2020. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. Griffin: Grammar-free dbms fuzzing. InProceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, ASE ’22, New York, NY, USA, 2023. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. Gdsmith: Detecting bugs in cypher graph database engines. InProceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, page 163–174, New York, NY, USA, 2023. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Zijing Yin, Si Liu, and David Basin. Testing graph databases with synthesized queries.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. Finding bugs in gremlinbased graph database systems via randomized differential testing. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2022, page 302–313, New York, NY, USA, 2022. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Qiuyang Mang, Aoyang Fang, Boxi Yu, Hanfei Chen, and Pinjia He. Testing graph database systems via equivalent query rewriting. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’24, New York, NY, USA, 2024. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’24, New York, NY, USA, 2024. Association for Computing Machinery.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. Testing graph database engines via query partitioning. InProceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, page 140–149, New York, NY, USA, 2023. Association for Computing Machinery.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Yingying Zheng, Wensheng Dou, Lei Tang, Ziyu Cui, Yu Gao, Jiansen Song, Liang Xu, Jiaxin Zhu, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. Testing gremlin-based graph database systems via query disassembling. InProceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2024, page 1695–1707, New York, NY, USA, 2024. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Zeyang Zhuang, Penghui Li, Pingchuan Ma, Wei Meng, and Shuai Wang. Testing graph database systems via graph-aware metamorphic relations.Proc. VLDB Endow., 17(4):836–848, December 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and Andr ´es Taylor. Cypher: An evolving query language for property graphs. InProceedings of the 2018 International Conference on Management of Data, SIGMOD ’18, page 1433–1445, New York, NY, USA, 2018. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Celine W ¨ust, Zuming Jiang, and Zhendong Su. Dinkel: State-aware and granular framework for validating graph databases.Proc. VLDB Endow., 19(6):1198–1211, May 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Jinsheng Ba and Manuel Rigger. Keep it simple: Testing databases via differential query plans.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 26,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and Andr ´es Taylor. Cypher: An evolving query language for property graphs. InProceedings of the 2018 International Conference on Management of Data, SIGMOD ’18, page 1433–1445, New York, NY, USA, 2018. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Albert-L ´aszl´o Barab ´asi and R ´eka Albert. Emergence of scaling in random networks.Science, 286(5439):509–512, October 1999.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Jure Leskovec, Deepayan Chakrabarti, Jon Kleinberg, Christos Faloutsos, and Zoubin Ghahramani. Kronecker graphs: An approach to modeling networks, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Renzo Angles and Claudio Gutierrez. Survey of graph database models. ACM Comput. Surv., 40(1), February 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Wenfei Fan. Graph pattern matching revised for social network analysis. InInternational Conference on Database Theory, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Shuitao Gan, Chao Zhang, Xiaojun Qin, Xuwen Tu, Kang Li, Zhongyu Pei, and Zuoning Chen. Collafl: Path sensitive fuzzing. In2018 IEEE Symposium on Security and Privacy (SP), pages 679–696, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Yu Jiang, Jie Liang, Fuchen Ma, Yuanliang Chen, Chijin Zhou, Yuheng Shen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Shanshan Li, and Quan Zhang. When fuzzing meets llms: Challenges and opportunities. InCompanion Proceedings of the 32nd ACM International Conference on the Foundations of Software Engineering, FSE 2024, page 492–496, New York, NY, USA, 2024. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Terence Parr and Kathleen Fisher. Ll(*): the foundation of the antlr parser generator.SIGPLAN Not., 46(6):425–436, June 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "C. Lattner and V. Adve. Llvm: a compilation framework for lifelong program analysis & transformation. InInternational Symposium on Code Generation and Optimization, 2004. CGO 2004., pages 75–86, 2004.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "JaCoCo Project. JaCoCo Java Code Coverage Library. https://www. jacoco.org/jacoco/. Accessed: 2026-07-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Paul Andlinger and Matthias Gelbmann. Db-engines ranking: Popularity ranking of database management systems. https://db-engines.com/en/ ranking, 2026. Accessed: 2026-03.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Yupeng Yang, Yongheng Chen, Rui Zhong, Jizhou Chen, and Wenke Lee. Towards generic database management system fuzzing. In Proceedings of the 33rd USENIX Conference on Security Symposium, SEC ’24, USA, 2024. USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. AFL++: Combining incremental steps of fuzzing research. In14th USENIX Workshop on Offensive Technologies (WOOT 20). USENIX Association, August 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Lei Tang, Wensheng Dou, Yingying Zheng, Lijie Xu, Wei Wang, Jun Wei, and Tao Huang. Proving Cypher Query Equivalence. In pages 3931–3944, Los Alamitos, CA, USA, May 2025. IEEE Computer Society.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. Unicorn: detect runtime errors in time-series databases with hybrid input synthesis. InProceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2022, page 251–262, New York, NY, USA, 2022. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Dawei Li, Yuxiao Guo, Qifan Liu, Jie Liang, Zhiyong Wu, Jingzhou Fu, Chi Zhang, and Yu Jiang. Arg: Testing query rewriters via abstract rule guided fuzzing. In2025 40th IEEE/ACM International Conference on Automated Software Engineering (ASE), page 1807–1818. IEEE Press, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang. Detecting logic bugs in dbmss via equivalent data construction.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Jinhui Lai, Chi Zhang, Bingyan Li, Chenglin Liang, Jie Liang, Zhiyong Wu, Jingzhou Fu, Yu Jiang, and Zichen Xu. Srs: Detecting logic bugs of join implementation in dbmss via set relation synthesis.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang.Hulk: Exploring data-sensitive performance anomalies in dbmss via data-driven analysis.Proc. ACM Softw. Eng., 2(ISSTA), June 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Chi Zhang and Manuel Rigger. Constant optimization driven database system testing.Proceedings of the ACM on Management of Data, 3:1 – 24, 2025.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 46,
+      "text": "Junwen An. Llm-assisted dialect-agnostic sql query parsing. InCompanion Proceedings of the 2025 ACMSIGPLAN International Conference on Systems, Programming, Languages, and Applications: Software for Humanity, SPLASH Companion ’25, page 31–33, New York, NY, USA, 2025. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Suyang Zhong and Manuel Rigger. Understanding and reusing test suites across database systems.Proceedings of the ACM on Management of Data, 2:1–26, 12 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 48,
+      "text": "Jinsheng Ba and Manuel Rigger. Testing database engines via query plan guidance. In Software Engineering (ICSE), pages 2060–2071, 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 49,
+      "text": "lcamtuf. American fuzzy lop (afl). http://lcamtuf.coredump.cx/afl/, 2017. [Online; accessed 26-March-2026].",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 8,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang., 4(OOPSLA), November 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 9,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. InProceedings of the 14th USENIX Conference on Operating Systems Design and Implementation, OSDI’20, USA, 2020. USENIX Association.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 10,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE ’20, page 1140–1152. ACM, November 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 19,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’24, New York, NY, USA, 2024. Association for Computing Machinery.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. Testing graph database engines via query partitioning. InProceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, page 140–149, New York, NY, USA, 2023. Association for Computing Machinery.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 25,
+      "text": "Jinsheng Ba and Manuel Rigger. Keep it simple: Testing databases via differential query plans.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 45,
+      "text": "Chi Zhang and Manuel Rigger. Constant optimization driven database system testing.Proceedings of the ACM on Management of Data, 3:1 – 24, 2025.",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 47,
+      "text": "Suyang Zhong and Manuel Rigger. Understanding and reusing test suites across database systems.Proceedings of the ACM on Management of Data, 2:1–26, 12 2024.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 48,
+      "text": "Jinsheng Ba and Manuel Rigger. Testing database engines via query plan guidance. In Software Engineering (ICSE), pages 2060–2071, 2023.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Tools such as SQLsmith [7] and SQLancer [8]–[10] generate SQL queries based on abstract syntax tree models or formal semantics, successfully exposing bugs in mature systems such as PostgreSQL [11] and SQLite [12].",
+      "context_before": "ta naturally as nodes, relationships, and properties. This architecture simplifies the execution of complex multi hop queries, driving their adoption in domains like financial fraud detection, social network analysis, and real time recommendation [6]. As these applications depend on continuous availability, ensuring the reliability of the underlying database engine is paramount. Defects within a graph query engine can lead to systemfailures, including runtime crashes, memory corruption, and resource exhaustion. General DBMS fuzzers have proved effective for testing relational database systems.",
+      "context_after": "More recent fuzzers, including SQUIRREL [13] and Griffin [14], further improve input validity by preserving syntax or exploiting database metadata. These successes show that fuzzing is an effective way to exercise deep database execution logic and expose runtime failures. However, dedicated fuzzing techniques for GDBMSs remain underexplored. Existing approaches mainly rely on differential testing [15]–[17] and metamorphic testing [18]–[22]. Since these techniques are built around result consistency oracles, they typically require generated queries to be comparable across different engines or",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3141,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[18]–[22]",
+      "technique": null,
+      "sentence": "Existing approaches mainly rely on differential testing [15]–[17] and metamorphic testing [18]–[22].",
+      "context_before": "e for testing relational database systems. Tools such as SQLsmith [7] and SQLancer [8]–[10] generate SQL queries based on abstract syntax tree models or formal semantics, successfully exposing bugs in mature systems such as PostgreSQL [11] and SQLite [12]. More recent fuzzers, including SQUIRREL [13] and Griffin [14], further improve input validity by preserving syntax or exploiting database metadata. These successes show that fuzzing is an effective way to exercise deep database execution logic and expose runtime failures. However, dedicated fuzzing techniques for GDBMSs remain underexplored.",
+      "context_after": "Since these techniques are built around result consistency oracles, they typically require generated queries to be comparable across different engines or semantically related transformations. This requirement narrows the explored input space toward well defined and result producing queries, leaving engine specific runtime failures, such as crashes and memory errors, insufficiently explored. Designing effective fuzzing techniques for GDBMSs remains difficult due to two key challenges: Challenge 1: Generating complex and valid graph queries.A valid Cypher query [23] must satisfy interdependent",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3699,
+      "cited_reference": {
+        "number": 19,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[15]–[20]",
+      "technique": null,
+      "sentence": "Existing GDBMS testing methods [15]–[20], [22], [25] mainly rely on differential or metamorphic testing and typically lack feedback driven exploration.",
+      "context_before": "optimization and execution logic of the target GDBMS [13]. To bypass the frontend, grammar-based generators [7], [24] produce validarXiv:2607.03741v1 [cs.SE] 4 Jul 2026 CustomerClothingOrderCustID: 10012Name:BobBrand: Adidas SerialNum: A001OrderNum: 123Quantity:1FactoryName: factory_1 Location: VietnamProduceDate: 20241201BatchNum:100 Fig. 1. Core elements of a labeled property graph: labeled nodes, directed relationships, and typed properties. but structurally simple queries, failing to thoroughly exercise deep engine logic. Challenge 2: Guiding exploration under topology dependent execution.",
+      "context_after": "Meanwhile, traditional coverage feedback is insufficient for graph query fuzzing because it captures control flow execution but not topology dependent data flow behavior. Graph query execution is highly sensitive to hubs, cycles, dense subgraphs, traversal depth, and predicate selectivity. For example, on a dense 15 node graph in FalkorDB, increasing a variable length traversal bound from[:E1..5]to[:E1..8]raises execution time from 146 ms to 46.81 s, although the executed control flow paths remain similar. Conversely, highly selective predicates may lead to empty results and little meaningful",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 6024,
+      "cited_reference": {
+        "number": 19,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[19]",
+      "technique": null,
+      "sentence": "To construct more robust metamorphic relations, GraphGenie [19] introduces injective and surjective graph query transformations, which systematically alter the structural constraints and cardinality mappings of the original queries to uncover subtle logic bugs.",
+      "context_before": "by generating high-quality test cases for comprehensive differential testing across diverse graph engines. Different from these multi-engine approaches, our work focuses on deep bug discovery within a single GDBMS through feedback-guided mutation generation fuzzing. B. Metamorphic Testing for GDBMSs Metamorphic testing validates a GDBMS by checking whether semantically equivalent query variants produce consistent results under defined transformations [39]. GRev [18] utilizes Equivalent Query Rewriting (EQR) rules to generate semantically equivalent graph queries, checking result consistencies.",
+      "context_after": "GDBMeter [20] adapts the Ternary Query Partitioning (TQP) concept to graph databases, partitioning a core graph query into multiple sub-queries whose aggregated results should logically reconstruct the original result set. Similarly, QuDi [21] addresses the unique challenges of imperative queries by disassembling complex Gremlin traversals into multiple independent operational steps, verifying the consistency between the original query and the aggregated disassembled results. Gamera [22] proposes graphaware metamorphic relations, which directly leverage inherent graph topological properties,",
+      "section": "VI RELATED WORK",
+      "page": 10,
+      "char_offset": 47217,
+      "cited_reference": {
+        "number": 19,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "Query Partitioning",
+      "technique": "tlp",
+      "sentence": "GDBMeter [20] adapts the Ternary Query Partitioning (TQP) concept to graph databases, partitioning a core graph query into multiple sub-queries whose aggregated results should logically reconstruct the original result set.",
+      "context_before": "ing. B. Metamorphic Testing for GDBMSs Metamorphic testing validates a GDBMS by checking whether semantically equivalent query variants produce consistent results under defined transformations [39]. GRev [18] utilizes Equivalent Query Rewriting (EQR) rules to generate semantically equivalent graph queries, checking result consistencies. To construct more robust metamorphic relations, GraphGenie [19] introduces injective and surjective graph query transformations, which systematically alter the structural constraints and cardinality mappings of the original queries to uncover subtle logic bugs.",
+      "context_after": "Similarly, QuDi [21] addresses the unique challenges of imperative queries by disassembling complex Gremlin traversals into multiple independent operational steps, verifying the consistency between the original query and the aggregated disassembled results. Gamera [22] proposes graphaware metamorphic relations, which directly leverage inherent graph topological properties, such as path connectivity and structural equivalence, to define novel test oracles without relying on query semantics alone. Additionally, Dinkel [24] applies fine-grained clauseand expression-level semanticspreserving tran",
+      "section": "VI RELATED WORK",
+      "page": 10,
+      "char_offset": 47479,
+      "found_by_all": [
+        "technique",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[7]–[10]",
+      "technique": "tlp",
+      "sentence": "Fuzzing for DBMSs Existing DBMS fuzzing techniques [7]–[10], [13], [25], [40]–[48] include generation-based and mutation-based approaches.",
+      "context_before": "results. Gamera [22] proposes graphaware metamorphic relations, which directly leverage inherent graph topological properties, such as path connectivity and structural equivalence, to define novel test oracles without relying on query semantics alone. Additionally, Dinkel [24] applies fine-grained clauseand expression-level semanticspreserving transformations to a Cypher query and checks whether the transformed query preserves the original result. Different from these metamorphic tools, our work relies onexecution feedback rather than pre-defined equivalence rules to steer query generation. C.",
+      "context_after": "Generation-based fuzzers synthesize test cases using predefined rules or constraint solvers (e.g., SQLsmith [7]). However, the NP-complete [13] nature of query generation often forces restricting structural diversity to maintain valid database states. To systematically detect logic bugs, SQLancer [8]–[10] automatically synthesizes queries and validates their correctness using advanced logical oracles like Ternary Logic Partitioning (TLP [8]). On the other hand, to achieve deeper code exploration, mutation-based fuzzers leverage coverage feedback. ARG [41] proposes an abstract rule-based testi",
+      "section": "VI RELATED WORK",
+      "page": 10,
+      "char_offset": 48552,
+      "cited_reference": {
+        "number": 8,
+        "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang., 4(OOPSLA), November 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "coddtest",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To systematically detect logic bugs, SQLancer [8]–[10] automatically synthesizes queries and validates their correctness using advanced logical oracles like Ternary Logic Partitioning (TLP [8]).",
+      "context_before": "whether the transformed query preserves the original result. Different from these metamorphic tools, our work relies onexecution feedback rather than pre-defined equivalence rules to steer query generation. C. Fuzzing for DBMSs Existing DBMS fuzzing techniques [7]–[10], [13], [25], [40]–[48] include generation-based and mutation-based approaches. Generation-based fuzzers synthesize test cases using predefined rules or constraint solvers (e.g., SQLsmith [7]). However, the NP-complete [13] nature of query generation often forces restricting structural diversity to maintain valid database states.",
+      "context_after": "On the other hand, to achieve deeper code exploration, mutation-based fuzzers leverage coverage feedback. ARG [41] proposes an abstract rule-based testing method to test the correctness and robustness of query rewriters through intelligent generation and built-in SQL queries. While traditional mutational tools (e.g., AFL [49]) struggle to pass the strict syntax parsers of complex database engines, recent advancements have significantly improved input validity. Specifically, SQUIRREL [13] designs an intermediate representation to enable syntax-preserving and semantics-guided mutations. RATEL f",
+      "section": "VI RELATED WORK",
+      "page": 10,
+      "char_offset": 48943,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:24:46Z",
+    "is_model_written": true,
+    "summary": "GRAF is a black-box fuzzer for graph database query engines. It argues that existing graph-system testing relies on differential and metamorphic testing whose consistency oracles restrict inputs to queries comparable across engines or transformations, leaving single-engine runtime failures under-explored. GRAF instantiates LLM-generated parameterised Cypher skeletons against the active graph state to produce valid, diverse queries, and applies five graph-specific mutation operators guided by execution feedback. It found 34 previously unknown bugs, 23 with CVEs.",
+    "narrative": "SQLancer is cited as the relational tool that synthesises queries and validates them with advanced oracles, and the graph-database descendants of its techniques -- GraphGenie's transformations and GDBMeter's adaptation of ternary query partitioning -- are the related work GRAF sets its feedback-driven approach against.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "motivation",
+      "M4": "background",
+      "M5": "definition",
+      "M6": "background",
+      "M7": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2607_09072.json b/_data/papers/paper_arxiv_2607_09072.json
new file mode 100644
index 0000000..4526827
--- /dev/null
+++ b/_data/papers/paper_arxiv_2607_09072.json
@@ -0,0 +1,362 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:04:50Z",
+  "paper": {
+    "id": "paper:arxiv:2607.09072",
+    "title": "Agentic Proof and Property-Based Testing via Property-Templates in Data-Intensive Computing",
+    "authors": [
+      "Seongmin Lee",
+      "Yaoxuan Wu",
+      "Miryung Kim"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "arxiv_id": "2607.09072",
+    "s2_paper_id": "3b5bde6d80922a14c37ac576e2ace45d800f99ca",
+    "url": "https://arxiv.org/abs/2607.09072",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2607.09072",
+      "retrieved_at": "2026-09-07T17:04:50Z",
+      "chars": 64704,
+      "content_sha256": "sha256:1051572bd5eaef8a002d197c1ef712ea057f3142ce47643aef58831d16c4ab8c"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2484
+      },
+      {
+        "number": "A",
+        "title": "Validating a Specification: Proof and PBT",
+        "start": 8654
+      },
+      {
+        "number": "B",
+        "title": "Recurring Properties in DISC",
+        "start": 10750
+      },
+      {
+        "number": "A",
+        "title": "Agentic Proof Synthesis with Proof Template",
+        "start": 19654
+      },
+      {
+        "number": "B",
+        "title": "Agentic PBT Synthesis with PBT Template",
+        "start": 26578
+      },
+      {
+        "number": "A",
+        "title": "Proof Synthesis",
+        "start": 32232
+      },
+      {
+        "number": "B",
+        "title": "PBT Synthesis",
+        "start": 42018
+      },
+      {
+        "number": "C",
+        "title": "Cross-Validation: PBT Against Formal Proofs",
+        "start": 47933
+      },
+      {
+        "number": "J",
+        "title": "Gonzalez, S. Shenker, and I. Stoica, “Apache Spark: A unified engine",
+        "start": 58946
+      },
+      {
+        "number": "D",
+        "title": "Elkaduwe, K. Engelhardt, R. Kolanski, M. Norrish, T. Sewell, H. Tuch,",
+        "start": 59299
+      },
+      {
+        "number": "Y",
+        "title": "Wu, and G. Lample, “Draft, Sketch, and Prove: Guiding Formal",
+        "start": 60144
+      },
+      {
+        "number": "S",
+        "title": "Lerner, and N. Smallbone, “Lemmanaid: Neuro-symbolic lemma",
+        "start": 62732
+      },
+      {
+        "number": "J",
+        "title": "Li, “Proving Query Equivalence Using Linear Integer Arithmetic,”",
+        "start": 63083
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "L. de Moura, S. Kong, J. Avigad, F. van Doorn, and J. von Raumer, “The Lean Theorem Prover (System Description),” inAutomated Deduction-CADE-25, A. P. Felty and A. Middeldorp, Eds. Cham: Springer International Publishing, 2015, pp. 378–388.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "K. Claessen and J. Hughes, “QuickCheck: A lightweight tool for random testing of Haskell programs,” inProceedings of the Fifth ACMSIGPLAN International Conference on Functional Programming (ICFP ’00). ACM, 2000, pp. 268–279.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "H. Goldstein, J. W. Cutler, D. Dickstein, B. C. Pierce, and A. Head, “Property-Based Testing in Practice,” inProceedings of the IEEE/ACM 46th International Conference on Software Engineering, ser. ICSE ’24. New York, NY, USA: Association for Computing Machinery, Apr. 2024, pp. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Z. Paraskevopoulou, C. Hri T ¸cu, M. D ´en`es, L. Lampropoulos, and B. C. Pierce, “Foundational property-based testing,” inInteractive Theorem Proving, C. Urban and X. Zhang, Eds. Cham: Springer International Publishing, 2015, pp. 325–343.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "M. Maaz, L. DeV oe, Z. Hatfield-Dodds, and N. Carlini, “Agentic PropertyBased Testing: Finding Bugs Across the Python Ecosystem,” 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "S. K. Lahiri, “Intent Formalization: A Grand Challenge for Reliable Coding in the Age of AI Agents,” 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "M. Zaharia, R. S. Xin, P. Wendell, T. Das, M. Armbrust, A. Dave, X. Meng, J. Rosen, S. Venkataraman, M. J. Franklin, A. Ghodsi, J. Gonzalez, S. Shenker, and I. Stoica, “Apache Spark: A unified engine for big data processing,”Communications of the ACM, vol. 59, no. 11, pp. 56–65, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "X. Leroy, “Formal verification of a realistic compiler,”Communications of the ACM, vol. 52, no. 7, pp. 107–115, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "G. Klein, K. Elphinstone, G. Heiser, J. Andronick, D. Cock, P. Derrin, D. Elkaduwe, K. Engelhardt, R. Kolanski, M. Norrish, T. Sewell, H. Tuch, and S. Winwood, “seL4: Formal Verification of an OS Kernel,” in Proceedings of the ACMSIGOPS 22nd Symposium on Operating Systems Principles (SOSP). ACM, 2009, pp. 207–220.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "J. Hughes, “Experiences with QuickCheck: Testing the Hard Stuff and Staying Sane,” inA List of Successes That Can Change the World: Essays Dedicated to Philip Wadler on the Occasion of His 60th Birthday (LNCS 9600). Springer, 2016, pp. 169–186.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Kiro, “Correctness with Property-based tests,” https://kiro.dev/docs/specs/correctness/, Nov. 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Q. McNemar, “Note on the sampling error of the difference between correlated proportions or percentages,”Psychometrika, vol. 12, no. 2, pp. 153–157, Jun. 1947.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "A. Q. Jiang, S. Welleck, J. P. Zhou, W. Li, J. Liu, M. Jamnik, T. Lacroix, Y. Wu, and G. Lample, “Draft, Sketch, and Prove: Guiding Formal Theorem Provers with Informal Proofs,” inThe Eleventh International Conference on Learning Representations (ICLR 2023), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Apache DataFusion Comet Developers, “Apache DataFusion Comet: Fuzz Testing,” https://github.com/apache/datafusion-comet/tree/ 03e833b955d369f994d9652026ca3c1eb641acac/fuzz-testing.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "A. Sivaraman, A. Sanchez-Stern, B. Chen, S. Lerner, and T. Millstein, “Data-driven lemma synthesis for interactive proofs,”Proceedings of the ACM on Programming Languages, vol. 6, no. OOPSLA2, pp. 505–531, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "J. Hughes, “How to Specify It! A Guide to Writing Properties of Pure Functions,” inTrends in Functional Programming: 20th International Symposium, TFP 2019, Vancouver, BC, Canada, June 12–14, 2019, Revised Selected Papers. Berlin, Heidelberg: Springer-Verlag, Jun. 2019, pp. 58–83.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "S. Segura, A. Dur ´an, J. Troya, and A. Ruiz Cort ´es, “A Template-Based Approach to Describing Metamorphic Relations,” in2017 IEEE/ACM 2nd International Workshop on Metamorphic Testing (MET). IEEE, 2017, pp. 3–9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "M. B. Dwyer, G. S. Avrunin, and J. C. Corbett, “Patterns in Property Specifications for Finite-State Verification,” inProceedings of the 21st International Conference on Software Engineering (ICSE). New York, NY, USA: ACM, 1999, pp. 411–420.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "A. Solar-Lezama, L. Tancau, R. Bod ´ık, S. A. Seshia, and V. A. Saraswat, “Combinatorial sketching for finite programs,” inProceedings of the 12th International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS XII). ACM, 2006, pp. 404–415.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "C. Li, W. Ma, Z. Wang, and Z. Wen, “SITA: A Framework for Structureto-Instance Theorem Autoformalization,” inProceedings of the AAAI Conference on Artificial Intelligence (AAAI 2026), vol. 40. AAAI Press, 2026, pp. 19 224–19 232.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "W. Yang, G. Fedyukovich, and A. Gupta, “Lemma Synthesis for Automating Induction over Algebraic Data Types,” inPrinciples and Practice of Constraint Programming (CP 2019), ser. Lecture Notes in Computer Science. Springer, 2019, pp. 600–617.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "A. Brendel, A. Sivaraman, and T. Millstein, “Synthesizing implication lemmas for interactive theorem proving,”Proceedings of the ACM on Programming Languages, vol. 9, no. OOPSLA2, pp. 2254–2278, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "O. Monta ˜no-Rivas, R. L. McCasland, L. Dixon, and A. Bundy, “Schemebased theorem discovery and concept invention,”Expert Systems with Applications, vol. 39, no. 2, pp. 1637–1646, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Y. Alhessi, S. H. Einarsd ´ottir, G. Granberry, E. First, M. Johansson, S. Lerner, and N. Smallbone, “Lemmanaid: Neuro-symbolic lemma conjecturing,”arXiv preprint arXiv:2504.04942, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "S. Chu, C. Wang, K. Weitz, and A. Cheung, “Cosette: An Automated Prover for SQL,” in8th Biennial Conference on Innovative Data Systems Research (CIDR), 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "H. Ding, Z. Wang, Y. Yang, D. Zhang, Z. Xu, H. Chen, R. Piskac, and J. Li, “Proving Query Equivalence Using Linear Integer Arithmetic,” Proceedings of the ACM on Management of Data, vol. 1, no. 4, pp. 1–26, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Z. Wang, Z. Zhou, Y. Yang, H. Ding, G. Hu, D. Ding, C. Tang, H. Chen, and J. Li, “WeTune: Automatic Discovery and Verification of Query Rewrite Rules,” inProceedings of the 2022 International Conference on Management of Data. New York, NY, USA: ACM, 2022, pp. 94–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "M. Rigger and Z. Su, “Finding Bugs in Database Systems via Query Partitioning,”Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 29,
+      "text": "——, “Detecting Optimization Bugs in Database Engines via NonOptimizing Reference Engine Construction,” inProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. New York, NY, USA: ACM, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 30,
+      "text": "M. A. Gulzar, S. Mardani, M. Musuvathi, and M. Kim, “White-Box Testing of Big Data Analytics with Complex User-Defined Functions,” in Proceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. New York, NY, USA: ACM, 2019, pp. 290–301.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Q. Zhang, J. Wang, M. A. Gulzar, R. Padhye, and M. Kim, “BigFuzz: Efficient Fuzz Testing for Data Analytics Using Framework Abstraction,” inProceedings of the 35th IEEE/ACM International Conference on Automated Software Engineering. New York, NY, USA: ACM, 2020, pp. 722–733.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "K. Kim, A. Poiroux, and A. Bosselut, “Do LLMs Game Formalization? Evaluating Faithfulness in Logical Reasoning,” 2026.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 28,
+      "text": "M. Rigger and Z. Su, “Finding Bugs in Database Systems via Query Partitioning,”Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 29,
+      "text": "——, “Detecting Optimization Bugs in Database Engines via NonOptimizing Reference Engine Construction,” inProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. New York, NY, USA: ACM, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "On the testing side, SQLancer detects logic and optimization bugs in database engines through constructed oracles such as query partitioning and a non-optimizing reference engine [ 28], [29].",
+      "context_before": "n lifts to a correct property over an arbitrary pipeline and workload, so the agent discharges only the local law and the instance is correct by construction. c) Testing and verifying data-intensive and queryprocessing systems:A body of work targets the correctness of data-processing systems themselves. On the verification side, automated SQL equivalence provers decide whether two relational queries are equivalent and thereby verify rewrite rules, as in Cosette [ 25] and SQLSolver [ 26]; WeTune [ 27] goes further, automaticallydiscoveringnew rewrite rules and verifying them with such a prover.",
+      "context_after": "Closest to our setting, big-data testers generate Spark inputs by symbolic execution of dataflow operators and UDFs [ 30] or frameworkabstraction fuzzing [ 31]. Each fixes both target and technique— one query pair, one engine, or one program at a time. Our work instead treats the optimization-relevant equivalences of data-intensive computing as recurringproperty families, and validates each instance at scale alongbotha proof track and a PBT track over real PySpark. VII. CONCLUSION We set out to make the validation of a software system’s many correctness properties tractable at scale. Our cent",
+      "section": "C Cross-Validation: PBT Against Formal Proofs",
+      "page": 10,
+      "char_offset": 55211,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "A study of recurring correctness properties in Apache Spark, expressed as parameterised property templates. The authors use them in a dual-track framework that both proves properties in the Lean 4 theorem prover and instantiates them as executable PySpark property-based tests, reporting that templates raise agentic proof success by up to 2.6x and cut proof hallucinations by 59%.",
+    "narrative": "One citation, describing SQLancer as detecting logic and optimization bugs in database engines through constructed oracles such as query partitioning -- the testing-side precedent for the paper's own property-based track.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2607_13276.json b/_data/papers/paper_arxiv_2607_13276.json
new file mode 100644
index 0000000..d2b9bba
--- /dev/null
+++ b/_data/papers/paper_arxiv_2607_13276.json
@@ -0,0 +1,433 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:23:35Z",
+  "paper": {
+    "id": "paper:arxiv:2607.13276",
+    "title": "Aurora DSQL: Scalable, Multi-Region OLTP",
+    "authors": [
+      "Marc Brooker",
+      "Mark Bowes",
+      "Mike Hershey",
+      "Zak van der Merwe",
+      "James Morle",
+      "Matthys Strydom"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "arxiv_id": "2607.13276",
+    "s2_paper_id": "07b3167f54375c3db6c83026a77b0aa8bc8f93ce",
+    "url": "https://arxiv.org/abs/2607.13276",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2607.13276",
+      "retrieved_at": "2026-09-08T15:23:35Z",
+      "chars": 79337,
+      "content_sha256": "sha256:1702e6b8a3260b88ffba05a2bd098d5f6a8afdfa30f58685cac51b600ae0969c"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1320
+      },
+      {
+        "number": "2",
+        "title": "ARCHITECTURE OVERVIEW",
+        "start": 6345
+      },
+      {
+        "number": "2.1",
+        "title": "Multi-tenancy and Security",
+        "start": 11669
+      },
+      {
+        "number": "2.3",
+        "title": "Multi-Region Architecture",
+        "start": 16078
+      },
+      {
+        "number": "2.4",
+        "title": "Snapshot Isolation and Strong Consistency",
+        "start": 18542
+      },
+      {
+        "number": "2.5",
+        "title": "Impact of Architectural Properties on",
+        "start": 20594
+      },
+      {
+        "number": "2.6",
+        "title": "Comparisons To Other Systems, Briefly",
+        "start": 24296
+      },
+      {
+        "number": "3",
+        "title": "HOSTING THE SQL ENGINE",
+        "start": 25934
+      },
+      {
+        "number": "4.1",
+        "title": "Shard Scheme",
+        "start": 34766
+      },
+      {
+        "number": "5.1",
+        "title": "Detecting Conflicts, More Formally",
+        "start": 39798
+      },
+      {
+        "number": "5.2",
+        "title": "Index Maintenance",
+        "start": 44255
+      },
+      {
+        "number": "5.3",
+        "title": "Garbage Collection",
+        "start": 46241
+      },
+      {
+        "number": "5.4",
+        "title": "The Journal and Erasure Coding",
+        "start": 47683
+      },
+      {
+        "number": "5.5",
+        "title": "Exceptions to Snapshot Isolation",
+        "start": 49958
+      },
+      {
+        "number": "5.6",
+        "title": "Effects of Clock Skew",
+        "start": 51615
+      },
+      {
+        "number": "8.1",
+        "title": "Commit Size Limits",
+        "start": 62153
+      },
+      {
+        "number": "8.2",
+        "title": "Foreign Key Constraints",
+        "start": 63658
+      },
+      {
+        "number": "8.3",
+        "title": "Sequences and Indexes",
+        "start": 64773
+      },
+      {
+        "number": "8.4",
+        "title": "Strong Consistency",
+        "start": 65627
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Daniel Abadi. 2012. Consistency Tradeoffs in Modern Distributed Database System Design: CAP is Only Part of the Story.Computer45, 2 (2012), 37–42. https://doi.org/10.1109/MC.2012.33",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Alexandru Agache, Marc Brooker, Alexandra Iordache, Anthony Liguori, Rolf Neugebauer, Phil Piwonka, and Diana-Maria Popa. 2020. Firecracker: Lightweight Virtualization for Serverless Applications. In17th {USENIX}Symposium on Networked Systems Design and Implementation ({NSDI}20). 419–434.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Nikos Armenatzoglou, Sanuj Basu, Naga Bhanoori, Mengchu Cai, Naresh Chainani, Kiran Chinta, Venkatraman Govindaraju, Todd J. Green, Monish Gupta, Sebastian Hillig, Eric Hotinger, Yan Leshinksy, Jintian Liang, Michael McCreedy, Fabian Nagel, Ippokratis Pandis, Panos Parchas, Rahul Pathak, Orestis Polychroniou, Foyzur Rahman, Gaurav Saxena, Gokul Soundararajan, Sriram Subramanian, and Doug Terry. 20",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In (ICSE). IEEE, 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "Bradley Barnhart, Marc Brooker, Daniil Chinenkov, Tony Hooper, Jihoun Im, Prakash Chandra Jha, Tim Kraska, Ashok Kurakula, Alexey Kuznetsov, Grant McAlister, et al .2024. Resource Management in Aurora Serverless.Proceedings of the VLDB Endowment17, 12 (2024), 4038–4050.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Hal Berenson, Phil Bernstein, Jim Gray, Jim Melton, Elizabeth O’Neil, and Patrick O’Neil. 1995. A Critique of ANSI SQL Isolation Levels. InProceedings of the 1995 ACMSIGMOD International Conference on Management of Data(San Jose, California, USA)(SIGMOD ’95). Association for Computing Machinery, New York, NY, USA, 1–10. https://doi.org/10.1145/223784.223785",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Marc Brooker and Ankush Desai. 2025. Systems Correctness Practices at Amazon Web Services.Commun. ACM68, 6 (2025), 38–42.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "James C. Corbett, Jeffrey Dean, Michael Epstein, Andrew Fikes, Christopher Frost, JJ Furman, Sanjay Ghemawat, Andrey Gubarev, Christopher Heiser, Peter Hochschild, Wilson Hsieh, Sebastian Kanthak, Eugene Kogan, Hongyi Li, Alexander Lloyd, Sergey Melnik, David Mwaura, David Nagle, Sean Quinlan, Rajesh Rao, Lindsay Rolig, Yasushi Saito, Michal Szymaniak, Christopher Taylor, Ruth Wang, and Dale Woodf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Natacha Crooks, Youer Pu, Lorenzo Alvisi, and Allen Clement. 2017. Seeing is Believing: A Client-Centric Specification of Database Isolation. InProceedings of the ACM Symposium on Principles of Distributed Computing(Washington, DC, USA)(PODC ’17). Association for Computing Machinery, New York, NY, USA, 73–82. https://doi.org/10.1145/3087801.3087802",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Umur Cubukcu, Ozgun Erdogan, Sumedh Pathak, Sudhakar Sannakkayala, and Marco Slot. 2021. Citus: Distributed PostgreSQL for Data-Intensive Applications. InProceedings of the 2021 International Conference on Management of Data(Virtual Event, China)(SIGMOD ’21). Association for Computing Machinery, New York, NY, USA, 2490–2502. https://doi.org/10.1145/3448016.3457551",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Khuzaima Daudjee and Kenneth Salem. 2006. Lazy Database Replication with Snapshot Isolation. InProceedings of the 32nd International Conference on Very Large Data Bases(Seoul, Korea)(VLDB ’06). VLDB Endowment, 715–726.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Giuseppe DeCandia, Deniz Hastorun, Madan Jampani, Gunavardhan Kakulapati, Avinash Lakshman, Alex Pilchin, Swaminathan Sivasubramanian, Peter Vosshall, and Werner Vogels. 2007. Dynamo: Amazon’s highly available key-value store. ACMSIGOPS operating systems review41, 6 (2007), 205–220.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Mostafa Elhemali, Niall Gallagher, Nick Gordon, Joseph Idziorek, Richard Krog, Colin Lazier, Erben Mo, Akhilesh Mritunjai, Somasundaram Perianayagam, Tim Rath, Swami Sivasubramanian, James Christopher Sorenson III, Sroaj Sosothikul, Doug Terry, and Akshat Vig. 2022. Amazon DynamoDB: A Scalable, Predictably Performant, and Fully Managed NoSQL Database Service. In2022 USENIX Annual Technical Confere",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Robert Escriva, Bernard Wong, and Emin Gün Sirer. 2015. Warp: Lightweight Multi-Key Transactions for Key-Value Stores. https://doi.org/10.48550/ARXIV. 1509.07815",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Jim Gray and Leslie Lamport. 2006. Consensus on transaction commit.ACM Transactions on Database Systems (TODS)31, 1 (2006), 133–160. Aurora DSQL: Scalable, Multi-Region OLTP",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Anurag Gupta, Deepak Agarwal, Derek Tan, Jakub Kulesza, Rahul Pathak, Stefano Stefani, and Vidhya Srinivasan. 2015. Amazon Redshift and the Case for Simpler Data Warehouses. InProceedings of the 2015 ACMSIGMOD International Conference on Management of Data(Melbourne, Victoria, Australia)(SIGMOD ’15). Association for Computing Machinery, New York, NY, USA, 1917–1923. https://doi.org/10.1145/2723372",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Joseph M. Hellerstein and Peter Alvaro. 2020. Keeping CALM: When Distributed Consistency is Easy.Commun. ACM63, 9 (aug 2020), 72–81. https://doi.org/10. 1145/3369736",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Lexiang Huang, Matthew Magnusson, Abishek Bangalore Muralikrishna, Salman Estyak, Rebecca Isaacs, Abutalib Aghayev, Timothy Zhu, and Aleksey Charapko. 2022. Metastable Failures in the Wild. In16th USENIX Symposium on Operating Systems Design and Implementation (OSDI 22). USENIX Association, Carlsbad, CA, 73–90. https://www.usenix.org/conference/osdi22/presentation/huang-lexiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Joseph Idziorek, Alex Keyes, Colin Lazier, Somu Perianayagam, Prithvi Ramanathan, James Christopher Sorenson III, Doug Terry, and Akshat Vig. 2023. Distributed Transactions at Scale in Amazon DynamoDB. In2023 USENIX Annual Technical Conference (USENIX ATC 23). USENIX Association, Boston, MA, 705–717. https://www.usenix.org/conference/atc23/presentation/idziorek",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Rebecca Isaacs, Peter Alvaro, Rupak Majumdar, Kiran-Kumar Muniswamy-Reddy, Mahmoud Salamati, and Sadegh Soudjani. 2025. Analyzing Metastable Failures. InProceedings of the 2025 Workshop on Hot Topics in Operating Systems(Banff, AB, Canada)(HotOS ’25). Association for Computing Machinery, New York, NY, USA, 172–178. https://doi.org/10.1145/3713082.3730380",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Spencer Kimball and Irfan Sharif. 2022. Living without atomic clocks: Where CockroachDB and Spanner diverge. https://www.cockroachlabs.com/blog/ living-without-atomic-clocks/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "H. T. Kung and John T. Robinson. 1981. On Optimistic Methods for Concurrency Control.ACM Trans. Database Syst.6, 2 (jun 1981), 213–226. https://doi.org/10. 1145/319566.319567",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "John DC Little. 1961. A proof for the queuing formula: L= 𝜆W.Operations research9, 3 (1961), 383–387.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "David Lomet, Rick Anderson, TK Rengarajan, and Peter Spiro. 1992. How the Rdb/VMS data sharing system became fast.DEC Cambridge Research Lab Technical Report CRL92, 4 (1992).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Chris Newcombe, Tim Rath, Fan Zhang, Bogdan Munteanu, Marc Brooker, and Michael Deardeuff. 2015. How Amazon web services uses formal methods. Commun. ACM58, 4 (2015), 66–73.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Kun Ren, Dennis Li, and Daniel J. Abadi. 2019. SLOG: Serializable, Low-Latency, Geo-Replicated Transactions.Proc. VLDB Endow.12, 11 (jul 2019), 1747–1761. https://doi.org/10.14778/3342263.3342647",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Robbert Van Renesse and Fred B. Schneider. 2004. Chain Replication for Supporting High Throughput and Availability. In6th Symposium on Operating Systems Design and Implementation (OSDI 04). USENIX Association, San Francisco, CA. https://www.usenix.org/conference/osdi-04/chain-replication-supportinghigh-throughput-and-availability",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Bianca Schroeder, Adam Wierman, and Mor Harchol-Balter. 2006. Open versus closed: A cautionary tale. USENIX.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Sugu Sougoumarane and Mike Solomon. 2012. Vitess: Scaling MySQL at YouTube Using Go. USENIX Association, San Diego, CA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Rebecca Taft, Irfan Sharif, Andrei Matei, Nathan VanBenschoten, Jordan Lewis, Tobias Grieger, Kai Niemi, Andy Woods, Anne Birzin, Raphael Poss, Paul Bardea, Amruta Ranade, Ben Darnell, Bram Gruneir, Justin Jaffray, Lucy Zhang, and Peter Mattis. 2020. CockroachDB: The Resilient Geo-Distributed SQL Database. In Proceedings of the 2020 ACMSIGMOD International Conference on Management of Data(Portland",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Yacine Taleb, Kevin McGehee, Nan Yan, Shawn Wang, Stefan C. Müller, and Allen Samuels. 2024. Amazon MemoryDB: A Fast and Durable Memory-First Cloud Database. InCompanion of the 2024 International Conference on Management of Data(Santiago AA, Chile)(SIGMOD ’24). Association for Computing Machinery, New York, NY, USA, 309–320. https://doi.org/10.1145/3626246.3653380",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Gil Tene. 2012. How Not to Measure Latency. https://qconsf.com/sf2012/dl/qconsanfran-2012/slides/GilTene_HowNotToMeasureLatency.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Alexander Thomson, Thaddeus Diamond, Shu-Chun Weng, Kun Ren, Philip Shao, and Daniel J Abadi. 2012. Calvin: fast distributed transactions for partitioned database systems. InProceedings of the 2012 ACMSIGMOD International Conference on Management of Data. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Alexandre Verbitski, Anurag Gupta, Debanjan Saha, Murali Brahmadesam, Kamal Gupta, Raman Mittal, Sailesh Krishnamurthy, Sandor Maurice, Tengiz Kharatishvili, and Xiaofeng Bao. 2017. Amazon aurora: Design considerations for high throughput cloud-native relational databases. InProceedings of the 2017 ACM International Conference on Management of Data. 1041–1052.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Alexandre Verbitski, Anurag Gupta, Debanjan Saha, James Corey, Kamal Gupta, Murali Brahmadesam, Raman Mittal, Sailesh Krishnamurthy, Sandor Maurice, Tengiz Kharatishvilli, et al .2018. Amazon aurora: On avoiding distributed consensus for i/os, commits, and membership changes. InProceedings of the 2018 International Conference on Management of Data. 789–796.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Werner Vogels. 2009. Eventually Consistent.Commun. ACM52, 1 (jan 2009), 40–44. https://doi.org/10.1145/1435417.1435432",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Ding Yuan, Yu Luo, Xin Zhuang, Guilherme Renna Rodrigues, Xu Zhao, Yongle Zhang, Pranay U Jain, and Michael Stumm. 2014. Simple testing can prevent most critical failures: An analysis of production failures in distributed {DataIntensive}systems. In11th USENIX Symposium on Operating Systems Design and Implementation (OSDI 14). 249–265.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Jingyu Zhou, Meng Xu, Alexander Shraer, Bala Namasivayam, Alex Miller, Evan Tschannen, Steve Atherton, Andrew J Beamon, Rusty Sears, John Leach, et al. 2021. Foundationdb: A distributed unbundled transactional key value store. In Proceedings of the 2021 International Conference on Management of Data. 2653– 2666.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Tobias Ziegler, Philip A Bernstein, Viktor Leis, and Carsten Binnig. 2023. Is Scalable OLTP in the Cloud a Solved Problem? (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Tobias Ziegler, Carsten Binnig, and Viktor Leis. 2022. ScaleStore: A Fast and Cost-Efficient Storage Engine using DRAM, NVMe, and RDMA. InProceedings of the 2022 International Conference on Management of Data(Philadelphia, PA, USA)(SIGMOD ’22). Association for Computing Machinery, New York, NY, USA, 685–699. https://doi.org/10.1145/3514221.3526187",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In (ICSE). IEEE, 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The fuzz-testing approach, building on the approach of SQLancer [ 4], generates millions of example SQL statements and runs them both on DSQL and on AuroraPostgreSQL.",
+      "context_before": "errors and failures. Our experience, and data from Yuan et al [ 37], show that the majority of bugs in complex distributed systems are in error handling logic. Once deployed, we test the system using fault injection testing while under load, validating that the failure handling results from simulation are correct. A subset of our in-production fault injection tests are available to DSQL customers via the AWS Fault Injection Service. To validate that our SQL results match those returned by PostgreSQL, we deployed extensive fuzz testing alongside a static set of hundreds of fixed-function tests.",
+      "context_after": "In cases where our feature set fully overlaps with PostgreSQL, we expect DSQL to generate identical results to PostgreSQL. Some differences in behavior are tolerated where allowed by the SQL specification. For example in a SELECT without an explicit ORDER BY results may be returned in a different order, although the set of results must be identical. This approach has helped us find bugs in edge cases around floating point, collations, and NULL handling. Another pre-design validation was to build a event-based numerical simulation, which allowed us to explore the expected performance of the sy",
+      "section": "5.6 Effects of Clock Skew",
+      "page": 10,
+      "char_offset": 55801,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "A system description of Aurora DSQL, a serverless multi-region active-active SQL database. Compute, storage and transaction coordination are separated into independently scalable services, with PostgreSQL-compatible query processors running statelessly in Firecracker MicroVMs. It uses multiversion concurrency control with precision timestamps for coordination-free reads and optimistic concurrency control for writes, deferring coordination to commit time.",
+    "narrative": "SQLancer is used in production testing rather than cited as related work: the paper says its fuzz-testing approach builds on SQLancer's, generating millions of SQL statements and running them against both DSQL and a reference. That is an industrial adoption of the approach rather than of the codebase, as far as the mention shows.",
+    "roles": {
+      "M1": "reuse_component"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "uncertain",
+        "mention_ids": [
+          "M1"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "The fuzz-testing approach, building on the approach of SQLancer [ 4], generates millions of example SQL statements and runs them both on DSQL and on AuroraPostgreSQL.",
+            "section": "5.6 Effects of Clock Skew",
+            "page": 10
+          }
+        ],
+        "reasoning": "M1 says the fuzz-testing approach builds on the approach of SQLancer and generates millions of statements run against DSQL and a reference. Whether AWS runs SQLancer's code or reimplemented its approach is not settled by the mention.",
+        "reuse_kind": "unclear"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2608_15709.json b/_data/papers/paper_arxiv_2608_15709.json
new file mode 100644
index 0000000..51416b6
--- /dev/null
+++ b/_data/papers/paper_arxiv_2608_15709.json
@@ -0,0 +1,321 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:23:57Z",
+  "paper": {
+    "id": "paper:arxiv:2608.15709",
+    "title": "Logos: Certified Order-Sensitive SQL Rewrites with Mechanized Semantics and LLM Guidance",
+    "authors": [
+      "Jing-Yu Ke",
+      "Jingyang Li",
+      "Guoqiang Li"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "arxiv_id": "2608.15709",
+    "s2_paper_id": "513b2356cc9a799751b6c96ce838f98e67b8696e",
+    "url": "https://arxiv.org/abs/2608.15709",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2608.15709",
+      "retrieved_at": "2026-09-08T15:23:57Z",
+      "chars": 73404,
+      "content_sha256": "sha256:b5a122e16282a44f08d78f718b6cfb11b74fdd38e56d3977d2b743a836c58b06"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2294
+      },
+      {
+        "number": "2",
+        "title": "OVERVIEW",
+        "start": 10770
+      },
+      {
+        "number": "3.2",
+        "title": "Error-Aware Execution Semantics and",
+        "start": 20860
+      },
+      {
+        "number": "3.3",
+        "title": "Lifting Bag Equivalence to Ordered-List",
+        "start": 33091
+      },
+      {
+        "number": "3.4",
+        "title": "Formalizing the Running Example",
+        "start": 36738
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Setup",
+        "start": 43619
+      },
+      {
+        "number": "5.2",
+        "title": "RQ1: Comparison with Existing Verifiers",
+        "start": 46910
+      },
+      {
+        "number": "5.3",
+        "title": "RQ2: Runtime and Bottlenecks",
+        "start": 51931
+      },
+      {
+        "number": "5.4",
+        "title": "Threats to Validity",
+        "start": 55433
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Antoine Amarilli, Mouhamadou Lamine Ba, Daniel Deutch, and Pierre Senellart. 2017. Possible and Certain Answers for Queries over Order-Incomplete Data. In 24th International Symposium on Temporal Representation and Reasoning (Leibniz International Proceedings in Informatics), Vol. 90. Schloss Dagstuhl–LeibnizZentrum für Informatik, 4:1–4:19. https://doi.org/10.4230/LIPIcs.TIME.2017.4",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Qiushi Bai, Sadeem Alsudais, and Chen Li. 2023. QueryBooster: Improving SQL Performance Using Middleware Services for Human-Centered Query Rewriting. Proceedings of the VLDB Endowment16, 11 (2023), 2911–2924. https://doi.org/10. 14778/3611479.3611497",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Haniel Barbosa, Clark W. Barrett, Martin Brain, Gereon Kremer, Hanna Lachnitt, Makai Mann, Abdalrhman Mohamed, Mudathir Mohamed, Aina Niemetz, Andres Nötzli, Alex Ozdemir, Mathias Preiner, Andrew Reynolds, Ying Sheng, Cesare Tinelli, and Yoni Zohar. 2022. cvc5: A Versatile and Industrial-Strength SMT Solver. InTools and Algorithms for the Construction and Analysis of Systems (Lecture Notes in Comp",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Edmon Begoli, Jesús Camacho-Rodríguez, Julian Hyde, Michael J. Mior, and Daniel Lemire. 2018. Apache Calcite: A Foundational Framework for Optimized Query Processing Over Heterogeneous Data Sources. InProceedings of the 2018 International Conference on Management of Data. Association for Computing Machinery, 221–230. https://doi.org/10.1145/3183713.3190662",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Véronique Benzaken and Évelyne Contejean. 2019. A Coq Mechanised Formal Semantics for Realistic SQL Queries: Formally Reconciling SQL and Bag Relational Algebra. InProceedings of the 8th ACMSIGPLAN International Conference on Certified Programs and Proofs. Association for Computing Machinery, 249–261. https://doi.org/10.1145/3293880.3294107",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Hamid R. Chinaei. 2007.An Ordered Bag Semantics for SQL. Master’s thesis. University of Waterloo. https://hdl.handle.net/10012/3062",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Shumo Chu, Brendan Murphy, Jared Roesch, Alvin Cheung, and Dan Suciu. 2018. Axiomatic Foundations and Algorithms for Deciding Semantic Equivalences of SQL Queries.Proceedings of the VLDB Endowment11, 11 (2018), 1482–1495. https://doi.org/10.14778/3236187.3236200",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Shumo Chu, Chenglong Wang, Konstantin Weitz, and Alvin Cheung. 2017. Cosette: An Automated Prover for SQL. In8th Biennial Conference on Innovative Data Systems Research. www.cidrdb.org. https://www.cidrdb.org/cidr2017/ papers/p51-chu-cidr17.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Shumo Chu, Konstantin Weitz, Alvin Cheung, and Dan Suciu. 2017. HoTTSQL: Proving Query Rewrites with Univalent SQL Semantics. InProceedings of the 38th ACMSIGPLAN Conference on Programming Language Design and Implementation. Association for Computing Machinery, 510–524. https://doi.org/10.1145/3062341. 3062348",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Neil Coburn and Grant E. Weddell. 1993. A Logic for Rule-Based Query Optimization in Graph-Based Data Models. InDeductive and Object-Oriented Databases (Lecture Notes in Computer Science), Vol. 760. Springer, 120–145. https://doi.org/10.1007/3-540-57530-8_8",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Bailu Ding, Surajit Chaudhuri, Johannes Gehrke, and Vivek Narasayya. 2021. DSB: A Decision Support Benchmark for Workload-Driven and Traditional Database Systems.Proceedings of the VLDB Endowment14, 13 (2021), 3376–3388. https://doi.org/10.14778/3484224.3484234",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Haoran Ding, Zhaoguo Wang, Yicun Yang, Dexin Zhang, Zhenglin Xu, Haibo Chen, Ruzica Piskac, and Jinyang Li. 2023. Proving Query Equivalence Using Linear Integer Arithmetic.Proceedings of the ACM on Management of Data1, 4, Article 227 (2023), 26 pages. https://doi.org/10.1145/3626768",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Yang He, Pinhan Zhao, Xinyu Wang, and Yuepeng Wang. 2024. VeriEQL: Bounded Equivalence Verification for Complex SQL Queries with Integrity Constraints. Proceedings of the ACM on Programming Languages8, OOPSLA1 (2024), 1071– 1099. https://doi.org/10.1145/3649849",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Albert Qiaochu Jiang, Wenda Li, Szymon Tworkowski, Konrad Czechowski, Tomasz Odrzygóźdź, Piotr Miłoś, Yuhuai Wu, and Mateja Jamnik. 2022. Thor: Wielding Hammers to Integrate Language Models and Automated Theorem Provers. InAdvances in Neural Information Processing Systems 35. 8360–8373. https://doi.org/10.52202/068431-0608",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Toby Mao and Contributors. 2026. SQLGlot: Python SQL Parser and Transpiler, Version 30.11.0. https://pypi.org/project/sqlglot/30.11.0/ Accessed: 2026-07-07.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Vivek R. Narasayya and Surajit Chaudhuri. 2026. Leveraging Query Optimizers to Verify the Soundness of LLM-Based Query Rewrites for Real-World Workloads, and More!. In16th Conference on Innovative Data Systems Research. www.cidrdb.org, Chaminade, CA, USA, 12. https://www.cidrdb.org/cidr2026/ papers/p33-narasayya.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Hamid Pirahesh, Joseph M. Hellerstein, and Waqar Hasan. 1992. Extensible/Rule Based Query Rewrite Optimization in Starburst. InProceedings of the 1992 ACMSIGMOD International Conference on Management of Data. Association for Computing Machinery, 39–48. https://doi.org/10.1145/130283.130294",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "PostgreSQL Global Development Group. 2026.PostgreSQL 18.4: SELECT. https: //www.postgresql.org/docs/18/sql-select.html Accessed: 2026-07-28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. Association for Computing Machinery, 1140–1152. https://doi.org/10.1145/3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Giedrius Slivinskas and Christian S. Jensen. 2001. Enhancing an Extensible Query Optimizer with Support for Multiple Equivalence Types. InAdvances in Databases and Information Systems (Lecture Notes in Computer Science), Vol. 2151. Springer, 55–69. https://doi.org/10.1007/3-540-44803-9_6",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Giedrius Slivinskas, Christian S. Jensen, and Richard T. Snodgrass. 2001. A Foundation for Conventional and Temporal Query Optimization Addressing Duplicates and Ordering.IEEE Transactions on Knowledge and Data Engineering 13, 1 (2001), 21–49. https://doi.org/10.1109/69.908979",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Zhaoyan Sun, Xuanhe Zhou, Guoliang Li, Xiang Yu, Jianhua Feng, and Yong Zhang. 2025. R-Bot: An LLM-Based Query Rewrite System.Proceedings of the VLDB Endowment18, 12 (2025), 5031–5044. https://doi.org/10.14778/3750601. 3750625",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Emina Torlak and Rastislav Bodík. 2014. A Lightweight Symbolic Virtual Machine for Solver-Aided Host Languages. InProceedings of the 35th ACMSIGPLAN Conference on Programming Language Design and Implementation. Association for Computing Machinery, 530–541. https://doi.org/10.1145/2594291.2594340",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Transaction Processing Performance Council. 2017.TPC Benchmark H (Decision Support): Standard Specification, Revision 2.17.3. https://www.tpc.org/tpc_ documents_current_versions/pdf/TPC-H_v2.17.3.pdf Accessed: 2026-07-31.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Transaction Processing Performance Council. 2024.TPC Benchmark DS: Standard Specification, Version 4.0.0. https://www.tpc.org/TPC_Documents_Current_ Versions/pdf/TPC-DS_v4.0.0.pdf Accessed: 2026-07-31.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Shuxian Wang, Sicheng Pan, and Alvin Cheung. 2024. QED: A Powerful Query Equivalence Decider for SQL.Proceedings of the VLDB Endowment17, 11 (2024), 3602–3614. https://doi.org/10.14778/3681954.3682024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Zhaoguo Wang, Zhou Zhou, Yicun Yang, Haoran Ding, Gansen Hu, Ding Ding, Chuzhe Tang, Haibo Chen, and Jinyang Li. 2022. WeTune: Automatic Discovery and Verification of Query Rewrite Rules. InProceedings of the 2022 International Conference on Management of Data. Association for Computing Machinery, 94– 107. https://doi.org/10.1145/3514221.3526125",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Kaiyu Yang, Aidan Swope, Alex Gu, Rahul Chalamala, Peiyang Song, Shixing Yu, Saad Godil, Ryan J. Prenger, and Animashree Anandkumar. 2023. LeanDojo: Theorem Proving with Retrieval-Augmented Language Models. InAdvances in Neural Information Processing Systems 36. 21573–21612. https://doi.org/10.52202/ 075280-0944",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Qi Zhou, Joy Arulraj, Shamkant B. Navathe, William Harris, and Jinpeng Wu. 2022. SPES: A Symbolic Approach to Proving Query Equivalence Under Bag Semantics. In. IEEE, 2735–2748. https://doi.org/10.1109/ICDE53745.2022.00250",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Qi Zhou, Joy Arulraj, Shamkant B. Navathe, William Harris, and Dong Xu. 2019. Automated Verification of Query Equivalence Using Satisfiability Modulo Theories.Proceedings of the VLDB Endowment12, 11 (2019), 1276–1288. https: //doi.org/10.14778/3342263.3342267 13",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 19,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. Association for Computing Machinery, 1140–1152. https://doi.org/10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Testing and fuzzing tools such as NoREC expose optimizer bugs by constructing differential or metamorphic test oracles [ 19].",
+      "context_before": "ORDER BY. It therefore does not provide a compositional semantics of legal result orders in which every tie-induced top-𝑘result remains visible to enclosing queries [13]. Earlier symbolic systems include SPES, which targets bag semantics, and EQUITAS, which targets set semantics; both use SMTbacked symbolic representations to reason about concrete predicates, but neither provides semantics for order-sensitive queries [ 29, 30]. The U-semiring decision procedure (UDP) supplies unbounded bag reasoning, while abstracting scalar and aggregate semantics and omitting order-sensitive operators [ 7].",
+      "context_after": "Such techniques are effective at finding concrete counterexamples, but they do not provide universally quantified rewrite proofs. LLM-assisted theorem proving.Recent theorem-proving agents use language models to retrieve relevant facts, predict proof steps, or coordinate automated provers. LeanDojo studies retrieval-based theorem proving over Lean libraries [ 28], and Thor integrates language models with automated theorem provers for formal proof search [ 14]. These advances are orthogonal to our SQL-specific semantic development, but inform Logos’s use of language models to guide proof searc",
+      "section": "5.4 Threats to Validity",
+      "page": 12,
+      "char_offset": 63557,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "Logos verifies SQL rewrite equivalence in the Rocq proof assistant. The authors mechanise a compositional logical semantics for a typed SQL core with order-sensitive operators, capturing all possible ordered lists and observable failures, which they describe as the first mechanised SQL semantics combining nested tie-sensitive top-k with a lifting from bag equivalence to ordered-list equivalence. Logos then uses an LLM-guided agent with a verified lemma library, solving 86.9% of 389 query pairs against 64.0% for the strongest baseline.",
+    "narrative": "One citation in threats to validity, noting that tools such as NoREC expose optimizer bugs by constructing differential or metamorphic oracles -- the empirical counterpart to what Logos proves.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2608_23402.json b/_data/papers/paper_arxiv_2608_23402.json
new file mode 100644
index 0000000..172c9ce
--- /dev/null
+++ b/_data/papers/paper_arxiv_2608_23402.json
@@ -0,0 +1,519 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:13Z",
+  "paper": {
+    "id": "paper:arxiv:2608.23402",
+    "title": "EXPLAIN Yourself! Finding Query Planner Stalls Across DBMSes",
+    "authors": [
+      "Geoffrey X. Yu",
+      "Ryan Marcus",
+      "Tim Kraska"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "arxiv_id": "2608.23402",
+    "s2_paper_id": "5def00bebb723d615e84104a65800e1aa21f3743",
+    "url": "https://arxiv.org/abs/2608.23402",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2608.23402",
+      "retrieved_at": "2026-09-09T01:29:13Z",
+      "chars": 38574,
+      "content_sha256": "sha256:75b73dc7da1d36cd6b8f17b49d774bb26e7bcd56cb049e8c761bf4b44b925f28"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 7,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1619
+      },
+      {
+        "number": "2",
+        "title": "Finding Query Planner Stalls",
+        "start": 6324
+      },
+      {
+        "number": "2.1",
+        "title": "Search Procedure",
+        "start": 6485
+      },
+      {
+        "number": "2.2",
+        "title": "Search Results",
+        "start": 10422
+      },
+      {
+        "number": "2.3",
+        "title": "Do Planner Stalls Transfer Across Systems?",
+        "start": 12770
+      },
+      {
+        "number": "3",
+        "title": "Query Planner Pathologies",
+        "start": 14220
+      },
+      {
+        "number": "3.1",
+        "title": "Correlated Subqueries",
+        "start": 14507
+      },
+      {
+        "number": "3.2",
+        "title": "Adversarial CTE Expansion",
+        "start": 17256
+      },
+      {
+        "number": "3.3",
+        "title": "Repeated Subquery Expressions",
+        "start": 18779
+      },
+      {
+        "number": "3.4",
+        "title": "Disjunctive Joins",
+        "start": 21664
+      },
+      {
+        "number": "3.5",
+        "title": "Adversarial Constant Folding",
+        "start": 23002
+      },
+      {
+        "number": "4",
+        "title": "Related Work",
+        "start": 25565
+      },
+      {
+        "number": "5",
+        "title": "Conclusion",
+        "start": 27941
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "DuckDB Authors. 2026. WITH Clause. https://duckdb.org/docs/lts/sql/query_ syntax/with.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "PostgreSQL Authors. 2026. Common Table Expression Materialization. https://www.postgresql.org/docs/current/queries-with.html#QUERIESWITH-CTE-MATERIALIZATION.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In The 45th International Conference on Software Engineering (ICSE’23).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. In The 46th International Conference on Software Engineering (ICSE’24).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. Proceeding of ACM Management of Data (SIGMOD’24) (June 2024).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Philipp Fent, Guido Moerkotte, and Thomas Neumann. 2023. Asymptotically Better Query Optimization Using Indexed Algebra. Proc. VLDB Endow. 16, 11 (July 2023), 3018–3030. doi:10.14778/3611479.3611505",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Wook-Shin Han, Wooseong Kwak, Jinsoo Lee, Guy M. Lohman, and Volker Markl. 2008. Parallelizing Query Optimization. Proc. VLDB Endow. 1, 1 (Aug. 2008), 188–200. doi:10.14778/1453856.1453882",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Axel Hertzschuch, Claudio Hartmann, Dirk Habich, and Wolfgang Lehner. 2021. Simplicity Done Right for Join Ordering. In Proceedings of the 11th Annual Conference on Innovative Data Systems Research (CIDR ’21).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Benjamin Hilprecht and Carsten Binnig. 2022. Zero-Shot Cost Models for Outof-the-box Learned Cost Prediction. Proceedings of the VLDB Endowment 15, 11 (2022), 2361–2374. https://www.vldb.org/pvldb/vol15/p2361-hilprecht.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Ihab F. Ilyas, Jun Rao, Guy Lohman, Dengfeng Gao, and Eileen Lin. 2003. Estimating Compilation Time of a Query Optimizer. In Proceedings of the 2003 ACMSIGMOD International Conference on Management of Data (San Diego, California)(SIGMOD ’03). Association for Computing Machinery, New York, NY, USA, 373–384. doi:10.1145/872757.872803",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. Proc. VLDB Endow. 13, 1 (Sept. 2019), 57–70. doi:10.14778/ 3357377.3357382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jinhui Lai, Chi Zhang, Jie Liang, Zihao Zeng, Zhiyong Wu, Jingzhou Fu, Chijin Zhou, Shuai Ma, Yu Jiang, and Zichen Xu. 2026. Finding Missed Optimizations in DBMSs through Unbalanced Short-circuit Query Construction. doi:10.1145/3802061",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2015. How good are query optimizers, really? Proc. VLDB Endow. 9, 3 (Nov. 2015), 204–215. doi:10.14778/2850583.2850594",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Zhan Li, Olga Papaemmanouil, and Mitch Cherniack. 2016. OptMark: A Toolkit for Benchmarking Query Optimizers. In Proceedings of the 25th ACM International on Conference on Information and Knowledge Management (Indianapolis, Indiana, USA) (CIKM ’16). Association for Computing Machinery, New York, NY, USA, 2155–2160. doi:10.1145/2983323.2983658",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. In Proceedings of the 44th International Conference on Software Engineering (Pittsburgh, Pennsylvania) (ICSE ’22). Association for Computing Machinery, New York, NY, USA, 225–236. doi:10.1145/3510003.3510093",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Qiuyang Mang, Runyuan He, Suyang Zhong, Xiaoxuan Liu, Huanchen Zhang, and Alvin Cheung. 2026. Automated Discovery of Test Oracles for Database Management Systems Using LLMs. doi:10.1145/3802017",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Ryan Marcus, Parimarjan Negi, Hongzi Mao, Nesime Tatbul, Mohammad Alizadeh, and Tim Kraska. 2022. Bao: Making Learned Query Optimization Practical. In Proceedings of the International Conference on Management of Data (SIGMOD ’22).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Ryan Marcus, Parimarjan Negi, Hongzi Mao, Chi Zhang, Mohammad Alizadeh, Tim Kraska, Olga Papaemmanouil, and Nesime Tatbul. 2019. Neo: A Learned Query Optimizer. Proceedings of the VLDB Endowment 12, 11 (2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Ryan Marcus and Olga Papaemmanouil. 2019. Plan-Structured Deep Neural Network Models for Query Performance Prediction. Proceedings of the VLDB Endowment 12, 11 (2019), 1733–1746. doi:10.14778/3342263.3342646",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Thomas Neumann. 2025. Improving Unnesting of Complex Queries. In Datenbanksysteme für Business, Technologie und Web (BTW 2025). Gesellschaft für Informatik, Bonn, 25–47. doi:10.18420/BTW2025-01",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Thomas Neumann and Bernhard Radke. 2018. Adaptive Optimization of Very Large Join Queries. In Proceedings of the 2018 International Conference on Management of Data (Houston, TX, USA) (SIGMOD ’18). Association for Computing Machinery, New York, NY, USA, 677–692. doi:10.1145/3183713.3183733",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Oracle Corporation. 2026. MySQL common_subexpression_elimination.cc File Reference. https://dev.mysql.com/doc/dev/mysql-server/9.7.1/common_ _subexpression__elimination_8cc.html. 6 EXPLAIN Yourself! Finding Query Planner Stalls Across DBMSes",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Oracle Corporation. 2026. MySQL Reference Manual. https://dev.mysql.com/doc/ refman/9.7/en/server-system-variables.html#sysvar_optimizer_search_depth.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Oracle Corporation. 2026. MySQL SQL Planner Code. https://github.com/ mysql/mysql-server/blob/06a5c1c99c377fc41b2eba1ea244e8b220bdc3c8/sql/ sql_planner.cc#L2638-L2642.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Anthony G Read. 2006. DeWitt Clauses: Can We Protect Purchasers Without Hurting Microsoft. Rev. Litig. 25 (2006), 387.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Maximilian Rieger and Thomas Neumann. 2025. T3: Accurate and Fast Performance Prediction for Relational Database Systems With Compiled Decision Trees. doi:10.1145/3725364",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 2020 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Sacramento, California, United States) (ESEC/FSE 2020). doi:10.1145/3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (2020). doi:10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI ’20). USENIX Association, Banff, Alberta. https://www. usenix.org/conference/osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 30,
+      "text": "Gaurav Saxena, Mohammad Rahman, Naresh Chainani, Chunbin Lin, George Caragea, Fahim Chowdhury, Ryan Marcus, Tim Kraska, Ippokratis Pandis, and Balakrishnan (Murali) Narayanaswamy. 2023. Auto-WLM: Machine Learning Enhanced Workload Management in Amazon Redshift. In Companion of the 2023 International Conference on Management of Data (Seattle, WA, USA) (SIGMOD ’23). Association for Computing Machine",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Vikramank Singh, Kapil Eknath Vaidya, Vinayshekhar Bannihatti Kumar, Sopan Khosla, Balakrishnan Narayanaswamy, Rashmi Gangadharaiah, and Tim Kraska. 2024. Panda: Performance Debugging for Databases Using LLM Agents. In 14th Conference on Innovative Data Systems Research (CIDR ’24). 6 pages. https: //www.cidrdb.org/cidr2024/papers/p6-singh.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Ji Sun and Guoliang Li. 2019. An End-to-End Learning-based Cost Estimator. Proceedings of the VLDB Endowment 13, 3 (2019), 307–319. doi:10.14778/3368289. 3368296",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Transaction Processing Performance Council. 2025. TPC-H Benchmark. https: //www.tpc.org/TPC_Documents_Current_Versions/pdf/TPC-H_v3.0.1.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Immanuel Trummer and Christoph Koch. 2016. Parallelizing Query Optimization on Shared-Nothing Architectures. Proc. VLDB Endow. 9, 9 (May 2016), 660–671. doi:10.14778/2947618.2947622",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. Hulk: Exploring Data-Sensitive Performance Anomalies in DBMSs via Data-Driven Analysis. Proc. ACM Softw. Eng. 2, ISSTA, Article ISSTA096 (June 2025), 22 pages. doi:10.1145/3728973",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. PUPPY: Finding Performance Degradation Bugs in DBMSs via Limited-Optimization Plan Construction. In Proceedings of the 47th IEEE/ACM International Conference on Software Engineering (ICSE ’25). IEEE Computer Society, 679–690. doi:10.1109/ ICSE55347.2025.00045",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Ziniu Wu, Ryan Marcus, Zhengchun Liu, Parimarjan Negi, Vikram Nathan, Pascal Pfeil, Gaurav Saxena, Mohammad Rahman, Balakrishnan Narayanaswamy, and Tim Kraska. 2024. Stage: Query Execution Time Prediction in Amazon Redshift. In Companion of the 2024 International Conference on Management of Data (Santiago AA, Chile) (SIGMOD ’24). Association for Computing Machinery, New York, NY, USA, 280–294. doi",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Ziniu Wu, Pei Yu, Peilun Yang, Rong Zhu, Yuxing Han, Yaliang Li, Defu Lian, Kai Zeng, and Jingren Zhou. 2022. A Unified Transferable Model for ML-Enhanced DBMS. In Proceedings of the 12th Conference on Innovative Data Systems Research (CIDR ’22). https://www.cidrdb.org/cidr2022/papers/p6-wu.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Geoffrey X. Yu, Ryan Marcus, and Tim Kraska. 2026. Tailwind: A Practical Framework for Query Accelerators. arXiv:2604.28079 [cs.DB] https://arxiv.org/ abs/2604.28079",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages. doi:10.1145/3709674",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "Xuanhe Zhou, Guoliang Li, Zhaoyan Sun, Zhiyuan Liu, Weize Chen, Jianming Wu, Jiesi Liu, Ruohang Feng, and Guoyang Zeng. 2024. D-Bot: Database Diagnosis System Using Large Language Models. Proceedings of the VLDB Endowment 17, 10 (2024), 2514–2527. https://www.vldb.org/pvldb/vol17/p2514-li.pdf 7",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In The 45th International Conference on Software Engineering (ICSE’23).",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. In The 46th International Conference on Software Engineering (ICSE’24).",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. Proceeding of ACM Management of Data (SIGMOD’24) (June 2024).",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 2020 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Sacramento, California, United States) (ESEC/FSE 2020). doi:10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 28,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (2020). doi:10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI ’20). USENIX Association, Banff, Alberta. https://www. usenix.org/conference/osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages. doi:10.1145/3709674",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "prints the DOI of the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 4,11,12,15,35,36]",
+      "technique": "cert",
+      "sentence": "Prior works propose techniques for finding query execution performance bugs in database systems [ 4,11,12,15,35,36].",
+      "context_before": "evaluates query planning effectiveness and efficiency. It measures efficiency using logical counts, e.g., the number of join orders explored (among others), which correlates with optimization time. While OptMark is a methodology for comparing planners on efficiency, our work focuses on finding and analyzing pathological queries that cause long planning times. Ilyas et al. explore predicting query planning time, which might help avoid pathological queries [ 10]. Orthogonal work also proposes techniques to improve the performance of query planning [6–8, 21, 34]. Database performance bug finding.",
+      "context_after": "The high-level idea is to identify a general property about a pair of queries (e.g., making a query more restrictive should result in an output cardinality estimate no larger than the original query [ 4]). Then, they generate query pairs and check if they satisfy the property and report a bug if not. Our work differs, as we identify and study queries that trigger query planner performance bugs (instead of query execution performance bugs). Database correctness bug finding. A complementary set of papers propose techniques to find correctness bugs in database systems [ 3,5,16,27–29,40], applyin",
+      "section": "4 Related Work",
+      "page": 6,
+      "char_offset": 26249,
+      "cited_reference": {
+        "number": 4,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. In The 46th International Conference on Software Engineering (ICSE’24).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 4]",
+      "technique": "cert",
+      "sentence": ", making a query more restrictive should result in an output cardinality estimate no larger than the original query [ 4]).",
+      "context_before": "le OptMark is a methodology for comparing planners on efficiency, our work focuses on finding and analyzing pathological queries that cause long planning times. Ilyas et al. explore predicting query planning time, which might help avoid pathological queries [ 10]. Orthogonal work also proposes techniques to improve the performance of query planning [6–8, 21, 34]. Database performance bug finding. Prior works propose techniques for finding query execution performance bugs in database systems [ 4,11,12,15,35,36]. The high-level idea is to identify a general property about a pair of queries (e.g.",
+      "context_after": "Then, they generate query pairs and check if they satisfy the property and report a bug if not. Our work differs, as we identify and study queries that trigger query planner performance bugs (instead of query execution performance bugs). Database correctness bug finding. A complementary set of papers propose techniques to find correctness bugs in database systems [ 3,5,16,27–29,40], applying a similar approach grounded in query pairs as in performance bug finding. Among these proposals, Argus is most thematically similar to this work as it uses LLMs to generate pairs of semantically equivalen",
+      "section": "4 Related Work",
+      "page": 6,
+      "char_offset": 26450,
+      "cited_reference": {
+        "number": 4,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. In The 46th International Conference on Software Engineering (ICSE’24).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 3,5,16,27–29,40]",
+      "technique": "qpg",
+      "sentence": "A complementary set of papers propose techniques to find correctness bugs in database systems [ 3,5,16,27–29,40], applying a similar approach grounded in query pairs as in performance bug finding.",
+      "context_before": "ding. Prior works propose techniques for finding query execution performance bugs in database systems [ 4,11,12,15,35,36]. The high-level idea is to identify a general property about a pair of queries (e.g., making a query more restrictive should result in an output cardinality estimate no larger than the original query [ 4]). Then, they generate query pairs and check if they satisfy the property and report a bug if not. Our work differs, as we identify and study queries that trigger query planner performance bugs (instead of query execution performance bugs). Database correctness bug finding.",
+      "context_after": "Among these proposals, Argus is most thematically similar to this work as it uses LLMs to generate pairs of semantically equivalent but syntactically different queries that they then execute and compare results [ 16]. Our work differs as we are searching for and analyzing query planner performance bugs, not correctness bugs. Agentic performance debugging. Researchers have recently been exploring leveraging LLMs to help with performance debugging [ 31] and diagnostics [ 41]. Such techniques propose workflows involving multiple agents, grounded in artifacts (e.g., prior customer tickets, troubl",
+      "section": "4 Related Work",
+      "page": 6,
+      "char_offset": 26844,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In The 45th International Conference on Software Engineering (ICSE’23).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-10T14:59:41Z",
+    "is_model_written": true,
+    "summary": "This paper looks for query planner stalls -- cases where planning a query takes disproportionately long, before any of it is executed. Because planning cost is visible through EXPLAIN without running the query, the same probe works across DBMSes, which is what lets the study span several systems rather than instrumenting one.",
+    "narrative": "SQLancer's work is the related-work ground this paper stands on, and all seven of its oracle papers are cited: CERT, QPG and DQP as the performance-bug line, and NoREC, TLP, PQS and CODDTest as the correctness line whose query-pair approach it describes as similar in kind to performance bug finding. CERT is the one drawn on in detail -- the paper restates its principle, that making a query more restrictive should not raise the output cardinality estimate. Every mention is reached through a citation marker; the paper never writes the name SQLancer.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Nothing describes reuse of SQLancer's code, generator or workload; planner stalls are measured through EXPLAIN."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The paper measures planning cost rather than developing an oracle. M2 restates CERT's principle as background, not as something generalised."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The seven papers are cited as related work. No run of SQLancer or any of its oracles against this tool is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The techniques are described by what they do; no standing is claimed for any of them."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2608_25573.json b/_data/papers/paper_arxiv_2608_25573.json
new file mode 100644
index 0000000..a99f5b8
--- /dev/null
+++ b/_data/papers/paper_arxiv_2608_25573.json
@@ -0,0 +1,607 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:23:01Z",
+  "paper": {
+    "id": "paper:arxiv:2608.25573",
+    "title": "DBcover: A White-box SQL Test Generation Framework for Coverage Improvement",
+    "authors": [
+      "Yan-Kai Rong",
+      "Shuang Liu",
+      "Jinhao Dong",
+      "Qiang Yin",
+      "Wei Lu",
+      "Jian-Hua Wang",
+      "Xiaoyong Du"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "arxiv_id": "2608.25573",
+    "s2_paper_id": "2b1f7251af13eb3114cc0c8812ae6a07cf970db2",
+    "url": "https://arxiv.org/abs/2608.25573",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2608.25573",
+      "retrieved_at": "2026-09-10T15:23:01Z",
+      "chars": 70224,
+      "content_sha256": "sha256:15eb33d11895d6be08798f96acccb9111bba277361781f3901375850d91cc60b"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2750
+      },
+      {
+        "number": "2",
+        "title": "Motivation Example",
+        "start": 11759
+      },
+      {
+        "number": "3.1",
+        "title": "Overview",
+        "start": 19126
+      },
+      {
+        "number": "3.2",
+        "title": "Knowledge Graph Construction",
+        "start": 21103
+      },
+      {
+        "number": "3.3",
+        "title": "Seed Selection",
+        "start": 27807
+      },
+      {
+        "number": "3.4",
+        "title": "Test Cases Generation",
+        "start": 30023
+      },
+      {
+        "number": "4.1",
+        "title": "Experimental Setup",
+        "start": 36972
+      },
+      {
+        "number": "4.1.4",
+        "title": "Baselines:",
+        "start": 39245
+      },
+      {
+        "number": "4.1.5",
+        "title": "Seed test cases.Seed inputs were collected from the official",
+        "start": 40503
+      },
+      {
+        "number": "4.2",
+        "title": "Comparison with Baselines",
+        "start": 40882
+      },
+      {
+        "number": "4.3",
+        "title": "Ablation Study",
+        "start": 45144
+      },
+      {
+        "number": "4.4",
+        "title": "Case Study",
+        "start": 51349
+      },
+      {
+        "number": "4.5",
+        "title": "Evaluation on Enterprise Closed-Source",
+        "start": 53780
+      },
+      {
+        "number": "5.1",
+        "title": "RDBMS Test Case Generation Methods",
+        "start": 55369
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2026. Doxygen. https://doxygen.cn/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2026. KingbaseES. https://www.kingbase.com.cn/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2026. Neo4j. https://neo4j.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2026. Pintool. https://www.intel.com/content/www/us/en/developer/articles/ tool/pin-a-dynamic-binary-instrumentation-tool.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al .2023. Gpt-4 technical report.arXiv preprint arXiv:2303.08774 (2023). 11",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Juan Altmayer Pizzorno and Emery D Berger. 2025. CoverUp: Effective High Coverage Test Generation for Python.Proceedings of the ACM on Software Engineering2, FSE (2025), 2897–2919.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Andrea Arcuri, Man Zhang, and Juan Galeotti. 2024. Advanced white-box heuristics for search-based fuzzing of rest apis.ACM Transactions on Software Engineering and Methodology33, 6 (2024), 1–36.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In (ICSE). IEEE, 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 9,
+      "text": "Tim Blazytko, Cornelius Aschermann, Moritz Schlögel, Ali Abbasi, Sergej Schumilo, Simon Wörner, and Thorsten Holz. 2019. {GRIMOIRE}: Synthesizing structure while fuzzing. In28th USENIX Security Symposium (USENIX Security 19). 1985–2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Xiang Cheng, Fan Sang, Yizhuo Zhai, Xiaokuan Zhang, and Taesoo Kim. 2025. RUG: Turbo LLM for Rust Unit Test Generation. In2025 IEEE/ACM 47th International Conference on Software Engineering (ICSE). IEEE Computer Society, 634–634.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Arghavan Moradi Dakhel, Amin Nikanjam, Vahid Majdinasab, Foutse Khomh, and Michel C Desmarais. 2024. Effective test generation using pre-trained large language models and mutation testing.Information and Software Technology171 (2024), 107468.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free dbms fuzzing. InProceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Keke Gai, Haochen Liang, Jing Yu, Liehuang Zhu, and Dusit Niyato. 2025. LLAMA: Multi-Feedback Smart Contract Fuzzing Framework with LLM-Guided Seed Generation.arXiv preprint arXiv:2507.12084(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. {DynSQL}: Stateful fuzzing for database management systems with complex and valid {SQL}query generation. In32nd USENIX Security Symposium (USENIX Security 23). 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. Apollo: Automatic detection and diagnosis of performance regressions in database systems.Proceedings of the VLDB Endowment13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting logical bugs of {DBMS}with coverage-based guidance. In31st USENIX Security Symposium (USENIX Security 22). 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Zhengxiong Luo, Qingpeng Du, Yujue Wang, Abhik Roychoudhury, and Yu Jiang. 2025. Enhancing protocol fuzzing via diverse seed corpus generation.IEEE Transactions on Software Engineering(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Yunbo Ni and Shaohua Li. 2025. Interleaving Large Language Models for Compiler Testing.Proceedings of the ACM on Programming Languages9, OOPSLA2 (2025), 815–841.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Yaroslav Oliinyk, Michael Scott, Ryan Tsang, Chongzhou Fang, Houman Homayoun, et al .2024. Fuzzing{BusyBox}: Leveraging{LLM}and Crash Reuse for Embedded Bug Unearthing. In33rd USENIX Security Symposium (USENIX Security 24). 883–900.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Pansilu Pitigalaarachchi. 2024. Symbolic Execution for Dynamic Kernel Analysis. InProceedings of the 2024 on ACMSIGSAC Conference on Computer andCommunications Security. 5104–5106.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "M. Rigger. 2020. Sqlancer: Detecting Logic Bugs in DBMS. https://github.com/ sqlancer/sqlancer.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "A. Seltenreich, B. Tang, and S. Mullender. 2018. SQLsmith: A random SQL query generator. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Jinghan Wang, Yue Duan, Wei Song, Heng Yin, and Chengyu Song. 2019. Be sensitive and collaborative: Analyzing impact of coverage metrics in greybox fuzzing. In22nd International Symposium on Research in Attacks, Intrusions and Defenses (RAID 2019). 1–15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry practice of coverage-guided enterprise-level DBMS fuzzing. In Software Engineering in Practice (ICSE-SEIP). IEEE, 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Chunqiu Steven Xia, Matteo Paltenghi, Jia Le Tian, Michael Pradel, and Lingming Zhang. 2024. Fuzz4all: Universal fuzzing with large language models. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Hanxiang Xu, Wei Ma, Ting Zhou, Yanjie Zhao, Kai Chen, Qiang Hu, Yang Liu, and Haoyu Wang. 2025. CKGFuzzer: LLM-Based Fuzz Driver Generation Enhanced By Code Knowledge Graph. In Software Engineering: Companion Proceedings (ICSE-Companion). IEEE, 243–254.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Chen Yang, Junjie Chen, Bin Lin, Ziqi Wang, and Jianyi Zhou. 2024. Advancing code coverage: Incorporating program analysis with large language models.ACM Transactions on Software Engineering and Methodology(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Chenyuan Yang, Yinlin Deng, Runyu Lu, Jiayi Yao, Jiawei Liu, Reyhaneh Jabbarvand, and Lingming Zhang. 2024. Whitefox: White-box compiler fuzzing empowered by large language models.Proceedings of the ACM on Programming Languages8, OOPSLA2 (2024), 709–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Chenyuan Yang, Xuheng Li, Md Rakib Hossain Misu, Jianan Yao, Weidong Cui, Yeyun Gong, Chris Hawblitzel, Shuvendu Lahiri, Jacob R Lorch, Shuai Lu, et al. 2025. Autoverus: Automated proof generation for rust code.Proceedings of the ACM on Programming Languages9, OOPSLA2 (2025), 3454–3482.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Chenyuan Yang, Zijie Zhao, and Lingming Zhang. 2025. Kernelgpt: Enhanced kernel fuzzing via large language models. InProceedings of the 30th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2. 560–573.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Zhiyu Zhang, Longxing Li, Ruigang Liang, and Kai Chen. 2025. Unlocking Low Frequency Syscalls in Kernel Fuzzing with Dependency-Based RAG.Proceedings of the ACM on Software Engineering2, ISSTA (2025), 848–870.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing database management systems with language validity and coverage feedback. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Scaling Automated Database System Testing.arXiv preprint arXiv:2503.21424(2025).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 34,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments.arXiv preprint arXiv:2505.02012(2025). 12",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 8,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In (ICSE). IEEE, 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 21,
+      "text": "M. Rigger. 2020. Sqlancer: Detecting Logic Bugs in DBMS. https://github.com/ sqlancer/sqlancer.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 33,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Scaling Automated Database System Testing.arXiv preprint arXiv:2503.21424(2025).",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing sqlancer_pp"
+      ]
+    },
+    {
+      "number": 34,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments.arXiv preprint arXiv:2505.02012(2025). 12",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "prints the arXiv id of the paper introducing shqvel"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 8,14,16,32]",
+      "technique": "qpg",
+      "sentence": "In recent years, a variety of automated SQL test case generation techniques have been proposed [ 8,14,16,32], aiming to produce both syntactically and semantically valid queries.",
+      "context_before": "ase Management Systems (RDBMSs) serve as the core infrastructure of modern data-intensive applications, where their stability directly determines the reliability of upper-layer services. As a key metric for assessing testing adequacy, code coverage has been widely adopted to evaluate the completeness of system testing [ 23,24,32], exerting a direct impact on the quality assessment and release decisions of industrial RDBMSs. Given the pervasive deployment of RDBMSs in critical systems, improving code coverage of RDBMSs has become an urgent requirement for ensuring system quality and robustness.",
+      "context_after": "Some approaches operate at the intermediate representation (IR) level to preserve semantic correctness during mutation [ 16,32], while others employ stateful fuzzing to maintain consistent database states throughout execution [ 14]. Another line of work emphasizes fine-grained coverage collection [24], improving the applicability of testing frameworks in industrial settings. Despite these advances, DBMS-oriented test generation techniques still facea fundamental challenge that constrains their coverage improvement: the absence of source code level context information, i.e., an explicit mappin",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3381,
+      "cited_reference": {
+        "number": 8,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In (ICSE). IEEE, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "ShQveL",
+      "technique": null,
+      "sentence": "4 Baselines: •Base: The baseline approach refers to using the initial set of regression test cases as input, executing the test cases on 7 Table 1: Line coverage achieved by the compared methods DBMS Base SQUIRREL ShQveL DBcover nse DBcover PostgreSQL 68.",
+      "context_before": "CentOS 7. Experiments for DBcover and its variant DBcover nseuse the qwen3:32b large language model with a temperature of 0.7. 4.1.3 Target RDBMSs.To demonstrate the effectiveness of DBcover, we applied it to a variety of relational database management systems (DBMS), including the open-source systems MySQL (version 8.0.33) and PostgreSQL (version 17.0), as well as a proprietary industrial DBMS, KingbaseES [ 2]. The inclusion of both open-source and closed-source systems in our evaluation provides a comprehensive assessment of DBcover’s performance across different RDBMS implementations. 4.1.",
+      "context_after": "6% 69.0% 31.5% 75.7%80.1% MySQL 71.2% 71.4% 25.7% 77.2%82.3% the target RDBMS and measuring the collective coverage achieved after these executions. •ShQveL [ 34] enhances existing SQL test case generator (SQLancer++ [ 33]) by leveraging LLM to synthesize SQL fragments. This is the first and only available LLM-assisted test generation method for RDBMS and thus is selected as a comparison method. In our implementation, we utilized Gemini-2.5-pro (with 175B parameters) as the LLM for the test case generation process. Additionally, we made modifications to certain parts of the code to adapt it to",
+      "section": "4.1.4 Baselines:",
+      "page": 7,
+      "char_offset": 39249,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "ShQveL",
+      "technique": null,
+      "sentence": "•ShQveL [ 34] enhances existing SQL test case generator (SQLancer++ [ 33]) by leveraging LLM to synthesize SQL fragments.",
+      "context_before": "gbase ES [ 2]. The inclusion of both open-source and closed-source systems in our evaluation provides a comprehensive assessment of DBcover’s performance across different RDBMS implementations. 4.1.4 Baselines: •Base: The baseline approach refers to using the initial set of regression test cases as input, executing the test cases on 7 Table 1: Line coverage achieved by the compared methods DBMS Base SQUIRREL ShQveL DBcover nse DBcover PostgreSQL 68.6% 69.0% 31.5% 75.7%80.1% MySQL 71.2% 71.4% 25.7% 77.2%82.3% the target RDBMS and measuring the collective coverage achieved after these executions.",
+      "context_after": "This is the first and only available LLM-assisted test generation method for RDBMS and thus is selected as a comparison method. In our implementation, we utilized Gemini-2.5-pro (with 175B parameters) as the LLM for the test case generation process. Additionally, we made modifications to certain parts of the code to adapt it to MySQL. •SQUIRREL [ 32]: is a representative test generation approach that conducts mutations on the intermediate representation (IR) representations, aiming to achieve high syntax and semantic correctness of the generated SQL query. SQUIRREL also supports the generatio",
+      "section": "4.1.4 Baselines:",
+      "page": 8,
+      "char_offset": 39652,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "ShQveL",
+      "technique": null,
+      "sentence": "ShQveL—a standalone LLM-augmented generator that does not leverage the seed corpus—achieves substantially lower coverage (31.",
+      "context_before": "ith Baselines Table 1 shows the results by each compared RDBMS test case generator in 24-hour time-bounded experiments. The columnBase reports the coverage achieved by executing only the initial seed pool (i.e., without generating any new test cases). We can observe that the initial seed pools already achieve a rather high line coverage, with68.6%on PostgreSQL and71.2%on MySQL. DBcover achieves the highest coverage—80.1%on PostgreSQL and82.3% on MySQL—surpassing the seed baseline by11.5%and11.1%, respectively. SQUIRREL improves coverage by merely0.4%on PostgreSQL (68.6%→69.0%) and0.2%on MySQL.",
+      "context_after": "5%on PostgreSQL, 25.7%on MySQL). We attribute the effectiveness of DBcover largely to its use of source-level context that explicitly links SQL inputs to their execution paths, which we refer to as SQL-to-path correspondence. Without this mapping, test generators cannot determine which SQL constructs are needed to reach a given uncovered code region. Both SQUIRREL and shQveL lack such context information, and thus cannot effectively cover the target code region. DBcover overcomes this barrier by explicitly collecting the SQL-to-path correspondence, providing LLM with this global context—and lev",
+      "section": "4.2 Comparison with Baselines",
+      "page": 8,
+      "char_offset": 41498,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "shQveL",
+      "technique": null,
+      "sentence": "Both SQUIRREL and shQveL lack such context information, and thus cannot effectively cover the target code region.",
+      "context_before": "aseline by11.5%and11.1%, respectively. SQUIRREL improves coverage by merely0.4%on PostgreSQL (68.6%→69.0%) and0.2%on MySQL. ShQveL—a standalone LLM-augmented generator that does not leverage the seed corpus—achieves substantially lower coverage (31.5%on PostgreSQL, 25.7%on MySQL). We attribute the effectiveness of DBcover largely to its use of source-level context that explicitly links SQL inputs to their execution paths, which we refer to as SQL-to-path correspondence. Without this mapping, test generators cannot determine which SQL constructs are needed to reach a given uncovered code region.",
+      "context_after": "DBcover overcomes this barrier by explicitly collecting the SQL-to-path correspondence, providing LLM with this global context—and leveraging its native code comprehension ability—DBcover unlocks deep,Table 2: Ablation study results (per module on PostgreSQL) Project Lines Base DBcover nse DBcover utils/adt 66211 81.60% 84.90% 88.00% (+3.10%, +6.40%) commands 32326 81.60% 83.40% 86.10% (+2.70%, +4.50%) nodes 24882 44.30% 44.70% 47.80% (+3.10%, +3.50%) parser 24206 87.30% 88.30% 90.00% (+1.70%, +2.70%) executor 23181 86.40% 86.50% 90.20% (+3.70%, +3.80%) catalog 13791 83.00% 85.90% 87.70% (+1.",
+      "section": "4.2 Comparison with Baselines",
+      "page": 8,
+      "char_offset": 41975,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[ 9,14,16,24,32–\n34]",
+      "technique": null,
+      "sentence": "1 RDBMS Test Case Generation Methods In traditional DBMS test case generation, fuzzing [ 9,14,16,24,32– 34] has received extensive attention due to its efficiency in automatically producing diverse inputs and triggering program behaviors.",
+      "context_before": "and metadata-intensive operations. Starting from an initial coverage of 50.0% in the data type processing module and 68.0% in the metadata management module (achieved by existing internal test suites), DBcover significantly advances coverage to80.0%and 81.4%, respectively. These results demonstrate that our approach can effectively explore deep, stateful logic in proprietary database engines. Due to licensing restrictions, internal implementation details cannot be disclosed; however, the reported coverage metrics reflect genuine execution paths within the production codebase. 5 Related Work 5.",
+      "context_after": "Existing approaches can be broadly grouped into two complementary directions, i.e., improving the syntactic and semantic validity of generated test cases, and refining coverage feedback to better reflect meaningful execution differences. The first line of work focuses on generating well-formed SQL that adheres to DBMS grammar and semantics. Early tools like SQLsmith [ 22], SQLancer [ 21] and Apollo [ 15] generate single statements via syntax-driven traversal. Recent approaches produce multiple semantically related queries per test case, with generative methods synthesizing these groups from s",
+      "section": "5.1 RDBMS Test Case Generation Methods",
+      "page": 10,
+      "char_offset": 55371,
+      "cited_reference": {
+        "number": 33,
+        "text": "Suyang Zhong and Manuel Rigger. 2025. Scaling Automated Database System Testing.arXiv preprint arXiv:2503.21424(2025).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing sqlancer_pp"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Early tools like SQLsmith [ 22], SQLancer [ 21] and Apollo [ 15] generate single statements via syntax-driven traversal.",
+      "context_before": ". 5 Related Work 5.1 RDBMS Test Case Generation Methods In traditional DBMS test case generation, fuzzing [ 9,14,16,24,32– 34] has received extensive attention due to its efficiency in automatically producing diverse inputs and triggering program behaviors. Existing approaches can be broadly grouped into two complementary directions, i.e., improving the syntactic and semantic validity of generated test cases, and refining coverage feedback to better reflect meaningful execution differences. The first line of work focuses on generating well-formed SQL that adheres to DBMS grammar and semantics.",
+      "context_after": "Recent approaches produce multiple semantically related queries per test case, with generative methods synthesizing these groups from scratch. DynSQL [ 14] dynamically queries the schema to guide abstract syntax tree instantiation into valid SQL. SQUIRREL [ 32], mutate seed queries, supporting syntax-preserving mutations. Griffin [ 12] recombines SQL fragments and repairs semantic inconsistencies using a metadata dependency graph. However, these methods treat the DBMS as a black/grey box and cannot reason about how SQL inputs map to internal execution paths, limiting their ability to improve",
+      "section": "5.1 RDBMS Test Case Generation Methods",
+      "page": 10,
+      "char_offset": 55952,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[ 6,11,18,25–31,34]",
+      "technique": null,
+      "sentence": "2 LLM for Test Case Generation With the rapid progress of large language models (LLMs) in code understanding and generation, LLM-driven test case generation has surged recently [ 6,11,18,25–31,34].",
+      "context_before": "] extend AFL’s bitmap size (Squirrel enlarges the default 64 KB bitmap to 256 KB) to alleviate hash collisions and obtain finer-grained coverage feedback for SQL synthesis. Ratel [ 24] addresses these issues by using static analysis to assign dedicated counters to basic blocks and branches, eliminating collisions, and by instrumenting all binaries at compile time for system-wide coverage in multi-process RDBMS. This line of work is orthogonal to ours. While this work does not generate queries, it provides valuable, precise coverage feedback that can enhance test-case generation guidance. 10 5.",
+      "context_after": "Due to their strong semantic modeling abilities, LLMs can generate structured, semantically meaningful test inputs and have been integrated into various testing frameworks to enhance testing quality and automation across software systems. Two main approaches have emerged: using LLMs to augment existing test generators and directly synthesizing tests with LLMs. In the augmentation paradigm, LLMs are integrated into existing test generators to enhance their generation process [ 13,17,34]. For instance, ShQveL [ 34] extracts SQL features through LLM interactions and incorporates them into existi",
+      "section": "5.1 RDBMS Test Case Generation Methods",
+      "page": 11,
+      "char_offset": 57568,
+      "cited_reference": {
+        "number": 34,
+        "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments.arXiv preprint arXiv:2505.02012(2025). 12",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "prints the arXiv id of the paper introducing shqvel"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker",
+      "surface": "[ 13,17,34]",
+      "technique": null,
+      "sentence": "In the augmentation paradigm, LLMs are integrated into existing test generators to enhance their generation process [ 13,17,34].",
+      "context_before": "nce test-case generation guidance. 10 5.2 LLM for Test Case Generation With the rapid progress of large language models (LLMs) in code understanding and generation, LLM-driven test case generation has surged recently [ 6,11,18,25–31,34]. Due to their strong semantic modeling abilities, LLMs can generate structured, semantically meaningful test inputs and have been integrated into various testing frameworks to enhance testing quality and automation across software systems. Two main approaches have emerged: using LLMs to augment existing test generators and directly synthesizing tests with LLMs.",
+      "context_after": "For instance, ShQveL [ 34] extracts SQL features through LLM interactions and incorporates them into existing generators, increasing behavioral coverage while maintaining efficiency. LLAMA [ 13] uses an LLM to generate semantically valid seeds and refines them with multi-feedback optimization and hybrid fuzzing for smart contracts. In these approaches, LLMs assist with tasks like syntax repair, seed enrichment, and semantic expansion, compensating for the limitations of traditional generators. However, their effectiveness is still constrained by the base generator’s ability to construct the e",
+      "section": "5.1 RDBMS Test Case Generation Methods",
+      "page": 11,
+      "char_offset": 58128,
+      "cited_reference": {
+        "number": 34,
+        "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments.arXiv preprint arXiv:2505.02012(2025). 12",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "prints the arXiv id of the paper introducing shqvel"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "ShQveL",
+      "technique": null,
+      "sentence": "For instance, ShQveL [ 34] extracts SQL features through LLM interactions and incorporates them into existing generators, increasing behavioral coverage while maintaining efficiency.",
+      "context_before": "code understanding and generation, LLM-driven test case generation has surged recently [ 6,11,18,25–31,34]. Due to their strong semantic modeling abilities, LLMs can generate structured, semantically meaningful test inputs and have been integrated into various testing frameworks to enhance testing quality and automation across software systems. Two main approaches have emerged: using LLMs to augment existing test generators and directly synthesizing tests with LLMs. In the augmentation paradigm, LLMs are integrated into existing test generators to enhance their generation process [ 13,17,34].",
+      "context_after": "LLAMA [ 13] uses an LLM to generate semantically valid seeds and refines them with multi-feedback optimization and hybrid fuzzing for smart contracts. In these approaches, LLMs assist with tasks like syntax repair, seed enrichment, and semantic expansion, compensating for the limitations of traditional generators. However, their effectiveness is still constrained by the base generator’s ability to construct the execution context needed to trigger deep system logic, limiting improvements from LLM-based augmentation alone. More recent research efforts [ 6,7,10,11,25] increasingly treat large la",
+      "section": "5.1 RDBMS Test Case Generation Methods",
+      "page": 11,
+      "char_offset": 58257,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:26:20Z",
+    "is_model_written": true,
+    "summary": "DBcover generates SQL tests for RDBMSs with white-box knowledge of the engine. Lightweight dynamic analysis extracts the correspondence between SQL inputs and execution paths along with call graphs as global context, and source-level information around target functions as local context, organised in a knowledge graph. Generation then runs in two phases: pick a seed whose execution path is near the uncovered target, then prompt an LLM with both contexts to reach uncovered code. It reports 80.1% and 82.3% coverage on PostgreSQL and MySQL.",
+    "narrative": "ShQveL, which enhances SQLancer++ with LLM-synthesised SQL fragments, is one of DBcover's baselines and is reported to achieve substantially lower coverage because it lacks the seed corpus and context DBcover uses. SQLancer itself is cited as an early syntax-driven generator.",
+    "roles": {
+      "M1": "background",
+      "M2": "baseline",
+      "M3": "baseline",
+      "M4": "result_comparison",
+      "M5": "result_comparison",
+      "M6": "background",
+      "M7": "background",
+      "M8": "background",
+      "M9": "background",
+      "M10": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M3",
+          "M4",
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "•ShQveL [ 34] enhances existing SQL test case generator (SQLancer++ [ 33]) by leveraging LLM to synthesize SQL fragments.",
+            "section": "4.1.4 Baselines:",
+            "page": 8
+          },
+          {
+            "mention_id": "M4",
+            "sentence": "ShQveL—a standalone LLM-augmented generator that does not leverage the seed corpus—achieves substantially lower coverage (31.",
+            "section": "4.2 Comparison with Baselines",
+            "page": 8
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "Both SQUIRREL and shQveL lack such context information, and thus cannot effectively cover the target code region.",
+            "section": "4.2 Comparison with Baselines",
+            "page": 8
+          }
+        ],
+        "reasoning": "ShQveL, built on SQLancer++, is one of the baselines, and M4 and M5 report its coverage against DBcover's and explain the difference."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2608_30385.json b/_data/papers/paper_arxiv_2608_30385.json
new file mode 100644
index 0000000..5dad7e1
--- /dev/null
+++ b/_data/papers/paper_arxiv_2608_30385.json
@@ -0,0 +1,944 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:06Z",
+  "paper": {
+    "id": "paper:arxiv:2608.30385",
+    "title": "Detecting DBMS Bugs by Constructing Equivalent Representations of Intermediate Query Results",
+    "authors": [
+      "Xiao-Xu Niu",
+      "Gong Chen",
+      "Jin-Fu Chen",
+      "Xiaoyuan Xie"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "arxiv_id": "2608.30385",
+    "s2_paper_id": "e239393bf8f261ed8f721e3d78859f15d52d239b",
+    "url": "https://arxiv.org/abs/2608.30385",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2608.30385",
+      "retrieved_at": "2026-09-09T01:29:06Z",
+      "chars": 66769,
+      "content_sha256": "sha256:14d5ee7feb84db80e5d89d8bd4d874acf64d3a4bf53aed3ff488ac48df3928fc"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1722
+      },
+      {
+        "number": "2",
+        "title": "Background and Motivation",
+        "start": 6312
+      },
+      {
+        "number": "2.1",
+        "title": "Representations of Intermediate Query",
+        "start": 6573
+      },
+      {
+        "number": "2.2",
+        "title": "Motivating Example",
+        "start": 8611
+      },
+      {
+        "number": "2.3",
+        "title": "Key Insight",
+        "start": 12195
+      },
+      {
+        "number": "3",
+        "title": "Approach",
+        "start": 13325
+      },
+      {
+        "number": "3.1",
+        "title": "Overview of ERIQ",
+        "start": 13706
+      },
+      {
+        "number": "3.2",
+        "title": "Database and Query Generation",
+        "start": 16298
+      },
+      {
+        "number": "3.3",
+        "title": "Constructing Equivalent Representations of",
+        "start": 20849
+      },
+      {
+        "number": "3.4",
+        "title": "Result Comparison",
+        "start": 23198
+      },
+      {
+        "number": "4",
+        "title": "Evaluation",
+        "start": 26347
+      },
+      {
+        "number": "4.1",
+        "title": "Experimental Setup",
+        "start": 27477
+      },
+      {
+        "number": "4.2",
+        "title": "RQ1: Effectiveness of ERIQ",
+        "start": 28927
+      },
+      {
+        "number": "4.3",
+        "title": "RQ2: Comparison with Existing Approaches",
+        "start": 31828
+      },
+      {
+        "number": "4.4",
+        "title": "RQ3: Contribution of the Three",
+        "start": 36037
+      },
+      {
+        "number": "4.5",
+        "title": "RQ4: Bug Analysis",
+        "start": 39630
+      },
+      {
+        "number": "5",
+        "title": "Discussion",
+        "start": 49723
+      },
+      {
+        "number": "6",
+        "title": "Related Work",
+        "start": 52817
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 56200
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Engineering (ICSE). IEEE, 2060–2071. https://doi.org/10.1109/icse48619.2023. 00174",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing Databases via Differential Query Plans.Proceedings of the ACM on Management of Data2, 3 (2024), 1–26. https://doi.org/10.1145/3654991",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Yujia Chen, Yingli Zhou, Fangyuan Zhang, and Cuiyun Gao. 2026. LLM-Based Test Case Generation in DBMS through Monte Carlo Tree Search.arXiv preprint arXiv:2603.21530(2026). https://doi.org/10.48550/arXiv.2603.21530",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang. 2025. Detecting Logic Bugs in DBMSs via Equivalent Data Construction.Proceedings of the ACM on Management of Data3, 6 (2025), 1–25. https://doi.org/10.1145/3769779",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, and Yu Jiang. 2024. Sedar: Obtaining Highquality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (ICSE). 1–12. https://doi.org/10.1145/3597503.3639210",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free DBMS Fuzzing. InProceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering (ASE). 1–12. https://doi.org/10. 1145/3551349.3560431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Xiyue Gao, Zhuang Liu, Jiangtao Cui, Hui Li, Hui Zhang, Kewei Wei, and Kankan Zhao. 2026. A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Evaluation.Comput. Surveys58, 10 (2026), 1–36. https://doi.org/10.1145/3799227",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023. Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis. In2023 USENIX Annual Technical Conference (USENIX ATC 23). 345–358. https://www. usenix.org/conference/atc23/presentation/hao",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "International Organization for Standardization. 2023.Information technology – Database languages SQL – Part 2: SQL/Foundation. Technical Report ISO/IEC 9075-2:2023. International Organization for Standardization. https://www.iso. org/standard/76584.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. In32nd USENIX Security Symposium (USENIX Security 23). 4949–4965. https: //www.usenix.org/conference/usenixsecurity23/presentation/jiang-zu-ming",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting Logic Bugs in Database Engines via Equivalent Expression Transformation. In18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24). 821–835. https: //www.usenix.org/conference/osdi24/presentation/jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O Laleye, and Sarfraz Khurshid. 2008. Query-aware Test Generation Using a Relational Constraint Solver. In 2008 23rd IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 238–247. https://doi.org/10.1109/ase.2008.34",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequenceoriented DBMS Fuzzing. In Engineering (ICDE). IEEE, 668–681. https://doi.org/10.1109/icde55515.2023.00057",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Yiyuan Bai, Qiang Zhang, and Yu Jiang. 2024. WingFuzz: Implementing Continuous Fuzzing for DBMSs. In2024 USENIX Annual Technical Conference (USENIX ATC 24). 479–492. https://www.usenix. org/conference/atc24/presentation/liang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting logical bugs of DBMS with Coverage-based Guidance. In31st USENIX Security Symposium (USENIX Security 22). 4309–4326. https://www.usenix.org/conference/usenixsecurity22/ presentation/liang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "MariaDB. 2026. MariaDB Server. https://mariadb.org/ Accessed: 2026-08-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "MySQL. 2026. Collation Coercibility in Expressions. https://dev.mysql.com/doc/ refman/9.6/en/charset-collation-coercibility.html Accessed: 2026-04-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "MySQL. 2026. CREATE TABLE ... SELECT Statement. https://dev.mysql.com/ doc/refman/9.6/en/create-table-select.html Accessed: 2026-04-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "MySQL. 2026. CREATE TEMPORARY TABLE Statement. https://dev.mysql. com/doc/refman/9.6/en/create-temporary-table.html Accessed: 2026-04-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "MySQL. 2026. CREATE VIEW Statement. https://dev.mysql.com/doc/refman/9. 6/en/create-view.html Accessed: 2026-04-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "MySQL. 2026. Floating-Point Types (Approximate Value) - FLOAT, DOUBLE. https://dev.mysql.com/doc/refman/9.6/en/floating-point-types.html Accessed: 2026-04-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "MySQL. 2026. MySQL Database. https://www.mysql.com/ Accessed: 2026-08-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "MySQL. 2026. Optimizing Derived Tables, View References, and Common Table Expressions with Merging or Materialization. https://dev.mysql.com/doc/ refman/9.6/en/derived-table-optimization.html Accessed: 2026-04-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "MySQL. 2026. Temporary Table Problems. https://dev.mysql.com/doc/refman/9. 6/en/temporary-table-problems.html Accessed: 2026-04-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "MySQL. 2026. WITH (Common Table Expressions). https://dev.mysql.com/doc/ refman/9.6/en/with.html Accessed: 2026-04-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "OceanBase. 2026. OceanBase Database. https://www.oceanbase.com/ Accessed: 2026-08-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Percona. 2026. Percona Database. https://www.percona.com/ Accessed: 202608-01. Xiaoxu Niu, Gong Chen, Jinfu Chen, and Xiaoyuan Xie",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "John Regehr, Yang Chen, Pascal Cuoq, Eric Eide, Chucky Ellison, and Xuejun Yang. 2012. Test-case Reduction for C Compiler Bugs. InProceedings of the 33rd ACMSIGPLAN conference on Programming Language Design and Implementation (PLDI). 335–346. https://doi.org/10.1145/2254064.2254104",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152. https://doi.org/10.1145/3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682. https://www.usenix.org/conference/osdi20/ presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 32,
+      "text": "Donald R Slutz. 1998. Massive Stochastic Testing of SQL. InProceedings of International Conference on Very Large Data Bases (VLDB), Vol. 98. 618–622. https://dl.acm.org/doi/abs/10.5555/645924.671199",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. In Engineering (ICSE). IEEE, 2072–2084. https://doi.org/10.1109/icse48619.2023. 00175",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jiansen Song, Wensheng Dou, Yu Gao, Ziyu Cui, Yingying Zheng, Dong Wang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction.Proceedings of the VLDB Endowment17, 8 (2024), 1884–1897. https://doi.org/10.14778/3659437.3659445",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Jiansen Song, Wensheng Dou, Yingying Zheng, Yu Gao, Ziyu Cui, Wei Wang, and Jun Wei. 2025. Detecting Schema-Related Logic Bugs in Relational DBMSsvia Equivalent Database Construction.Proceedings of the VLDB Endowment18, 7 (2025), 2281–2294. https://doi.org/10.14778/3734839.3734861",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "SQLancer. 2026. Automated Testing to Find Logic and Performance Bugs in Database Systems. https://github.com/sqlancer/sqlancer Accessed: 2026-08-01.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "SQLsmith. 2026. A Random SQL Query Generator. https://github.com/anse1/ sqlsmith Accessed: 2026-08-01.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting Logic Bugs of Join Optimizations in DBMS.Proceedings of the ACM on Management of Data1, 1 (2023), 1–26. https://doi.org/10.1145/3588909",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Xiu Tang, Shijie Yang, Sai Wu, Dongxiang Zhang, Wenchao Zhou, Feifei Li, and Gang Chen. 2025. Unveiling Logic Bugs in SPJG Query Optimizations within DBMS.ACM Transactions on Database Systems (TODS)51, 1 (2025), 1–35. https://doi.org/10.1145/3764583",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-guided Enterprise-level DBMS fuzzing. In Software Engineering in Practice (ICSE-SEIP). IEEE, 328–337. https://doi.org/10. 1109/icse-seip52600.2021.00042",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. Puppy: Finding Performance Degradation Bugs in DBMSs via Limited-optimization Plan Construction. In Engineering (ICSE). IEEE, 679–690. https://doi.org/10.1109/icse55347.2025.00045",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing.Proceedings of the ACM on Management of Data3, 1 (2025), 1–24. https://doi.org/10.1145/3709674",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security (CCS). 955–970. https: //doi.org/10.1145/3372297.3417260",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 1,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Engineering (ICSE). IEEE, 2060–2071. https://doi.org/10.1109/icse48619.2023. 00174",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "prints the DOI of the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing Databases via Differential Query Plans.Proceedings of the ACM on Management of Data2, 3 (2024), 1–26. https://doi.org/10.1145/3654991",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "prints the DOI of the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152. https://doi.org/10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 31,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682. https://www.usenix.org/conference/osdi20/ presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 36,
+      "text": "SQLancer. 2026. Automated Testing to Find Logic and Performance Bugs in Database Systems. https://github.com/sqlancer/sqlancer Accessed: 2026-08-01.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing.Proceedings of the ACM on Management of Data3, 1 (2025), 1–24. https://doi.org/10.1145/3709674",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "prints the DOI of the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For example, NoREC [ 29] compares the results of optimized and unoptimized queries; TLP [ 30] partitions queries based on ternary logic; PQS [ 31] checks whether a selected pivot row appears in the result of a generated query; Pinolo [ 8] checks result containment between synthesized queries and a seed query; EET [11] and CODDTest [42] apply equivalent transformations at the expression level; Radar [ 34] compares query results across databases with the same data but different metadata; and EDC [ 4] checks query consistency using precomputed results of data operation expressions.",
+      "context_before": "wever, we observe that such queries can return inconsistent results, as illustrated in Section 2.2.Such inconsistencies can indicate potential DBMS bugs, which may be ∗These authors have contributed equally to this work. †Corresponding author.exposed when the corresponding queries are processed through different paths within the DBMS. Many approaches have been proposed for detecting DBMS logic bugs.However, to the best of our knowledge, they have never explored the perspective of detecting such bugs by checking result consistency across equivalent representations of intermediate query results.",
+      "context_after": "In this paper, we propose ERIQ, a novel testing approach for detecting DBMS logic bugs from the perspective of checking result consistency across Equivalent Representations of Intermediate Query Results. ERIQ constructs corresponding SQL variants that use a VIEW, a CTE, or a TEMPT to represent the same intermediate query result, executes these variants, and compares their returned results. If the results are inconsistent, ERIQ reports the inconsistency as a bug. We implemented ERIQ and evaluated it on four widely used open-source DBMSs: MySQL, MariaDB, Percona, and OceanBase. In total, ERIQ d",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3065,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "pqs",
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We further compared ERIQ with four state-of-the-art DBMS logic bug detection approaches: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30].",
+      "context_before": "intermediate query result, executes these variants, and compares their returned results. If the results are inconsistent, ERIQ reports the inconsistency as a bug. We implemented ERIQ and evaluated it on four widely used open-source DBMSs: MySQL, MariaDB, Percona, and OceanBase. In total, ERIQ detected 64 bugs, including 29 in MySQL, 18 in MariaDB, 13 in Percona, and 4 in OceanBase. Developers confirmed 63 of them, including 54 unique and previously unknown logic bugs, 8 duplicates of other bugs reported by ERIQ, and one documentation issue. Among the 54 unique logic bugs, two have been fixed.",
+      "context_after": "Among these 54 unique logic bugs, the best-performing existing approach could detect only 9 of them, showing that ERIQ detects many bugs missed by each existing approach and thus provides complementary bug detection capability. Under the 24-hour testing arXiv:2608.30385v1 [cs.DB] 31 Aug 2026 Xiaoxu Niu, Gong Chen, Jinfu Chen, and Xiaoyuan Xie budget on each supported DBMS, ERIQ detected 54 logic bugs, compared with 16 detected by the best-performing existing approach, demonstrating its stronger bug detection capability. This paper makes the following contributions: •We introduce a new perspec",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 4554,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 8,29–\n31]",
+      "technique": "norec",
+      "sentence": "For example, existing approaches expose bugs by transforming, partitioning, or constructing queries [ 8,29– 31], transforming query expressions [ 11,42], varying database metadata [ 34], or using precomputed results of data operation expressions [4].",
+      "context_before": "velopers have fixed this bug (https://jira.mariadb.org/browse/MDEV40143), and its root cause is further analyzed in Section 4.5. Importantly, such logic bugs may silently produce incorrect query results without causing crashes or explicit errors, potentially propagating incorrect data to downstream applications. 2.3 Key Insight This observation motivates a new perspective for detecting DBMS logic bugs by checking result consistency across equivalent representations of the same intermediate query result. However, existing DBMS logic bug detection approaches have never explored this perspective.",
+      "context_after": "The rationale behind this perspective follows from the observation in Section 2.1: queries constructed using equivalent representations of the same intermediate query result may be processed through different paths within the DBMS. Consequently, an underlying DBMS logic bug can affect the corresponding queries differently, resulting in inconsistent query results, as demonstrated by the real-world bug presented in Section 2.2. Therefore, constructing such equivalent representations and comparing the correspondingquery results provides an effective approach to detecting DBMS logic bugs. 3 Appro",
+      "section": "2.3 Key Insight",
+      "page": 3,
+      "char_offset": 12481,
+      "cited_reference": {
+        "number": 29,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–115",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[ 2,4,10,14,33,41]",
+      "technique": "dqp",
+      "sentence": "These DBMSs have been extensively tested by many existing approaches [ 2,4,10,14,33,41].",
+      "context_before": "whether these representations provide complementary bug detection capability. •RQ4: What characteristics do the bugs detected by ERIQ exhibit?In this RQ, we analyze the SQL features involved in the detected bugs and examine representative root causes to understand the types of DBMS defects exposed by ERIQ. 4.1 Experimental Setup Target DBMSs.We selected 4 widely used open-source DBMSs: MySQL [ 22], MariaDB [ 16], Percona [ 27], and OceanBase [ 26]. We tested the latest release versions available when the experiments began: MySQL 9.6.0, MariaDB 12.2.2, Percona 8.0.45-36, and OceanBase 4.4.2.0.",
+      "context_after": "Therefore, finding new bugs in them is challenging. Baselines.We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30]. These approaches have demonstrated their effectiveness by uncovering real-world DBMS logic bugs and represent different bug detection perspectives. EDC was selected as the primary baseline because it supports all target DBMSs in our evaluation. Radar, EET, and TLP were additionally included because they were used as baselines in EDC’s evaluation. These baselines enable us to evaluate whether ERIQ provide",
+      "section": "4.1 Experimental Setup",
+      "page": 6,
+      "char_offset": 27768,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing Databases via Differential Query Plans.Proceedings of the ACM on Management of Data2, 3 (2024), 1–26. https://doi.org/10.1145/3654991",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "prints the DOI of the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30].",
+      "context_before": "Q, we analyze the SQL features involved in the detected bugs and examine representative root causes to understand the types of DBMS defects exposed by ERIQ. 4.1 Experimental Setup Target DBMSs.We selected 4 widely used open-source DBMSs: MySQL [ 22], MariaDB [ 16], Percona [ 27], and OceanBase [ 26]. We tested the latest release versions available when the experiments began: MySQL 9.6.0, MariaDB 12.2.2, Percona 8.0.45-36, and OceanBase 4.4.2.0. These DBMSs have been extensively tested by many existing approaches [ 2,4,10,14,33,41]. Therefore, finding new bugs in them is challenging. Baselines.",
+      "context_after": "These approaches have demonstrated their effectiveness by uncovering real-world DBMS logic bugs and represent different bug detection perspectives. EDC was selected as the primary baseline because it supports all target DBMSs in our evaluation. Radar, EET, and TLP were additionally included because they were used as baselines in EDC’s evaluation. These baselines enable us to evaluate whether ERIQ provides complementary bug detection capability by introducing a new testing perspective. Experimental infrastructure.We conducted all experiments on a server equipped with an Intel Xeon Gold 6254 pr",
+      "section": "4.1 Experimental Setup",
+      "page": 6,
+      "char_offset": 27920,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Radar, EET, and TLP were additionally included because they were used as baselines in EDC’s evaluation.",
+      "context_before": "an: MySQL 9.6.0, MariaDB 12.2.2, Percona 8.0.45-36, and OceanBase 4.4.2.0. These DBMSs have been extensively tested by many existing approaches [ 2,4,10,14,33,41]. Therefore, finding new bugs in them is challenging. Baselines.We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30]. These approaches have demonstrated their effectiveness by uncovering real-world DBMS logic bugs and represent different bug detection perspectives. EDC was selected as the primary baseline because it supports all target DBMSs in our evaluation.",
+      "context_after": "These baselines enable us to evaluate whether ERIQ provides complementary bug detection capability by introducing a new testing perspective. Experimental infrastructure.We conducted all experiments on a server equipped with an Intel Xeon Gold 6254 processor (3.10 GHz, 18 cores, and 36 hardware threads), 64 GB of RAM, and approximately 1 TB of local storage. The server ran 64-bit Ubuntu 20.04.3 LTS. Each DBMS was deployed in a separate Docker container created from its official image and run with its default configuration. 4.2 RQ1: Effectiveness of ERIQ To answer RQ1, we evaluated ERIQ’s effec",
+      "section": "4.1 Experimental Setup",
+      "page": 6,
+      "char_offset": 28294,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "1: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30].",
+      "context_before": "everity classifications. These results indicate that most of the unique logic bugs detected by ERIQ received relatively high severity classifications from the respective DBMS communities. Finding 1.ERIQ detected 64 bugs, 63 of which were confirmed by developers, and 2 have been fixed. Among the confirmed bugs, 54 were unique and previously unknown logic bugs, and 1 was a documentation issue. Of the 54 unique logic bugs, 48 received relatively high severity classifications. 4.3 RQ2: Comparison with Existing Approaches To answer RQ2, we compared ERIQ with the 4 baselines introduced in Section 4.",
+      "context_after": "We conducted two complementary experiments. First, we analyzed the distinctiveness of ERIQ’s 54 unique logic bugs by examining whether existing approaches could also detect them. Second, we compared ERIQ and the baselines under the same time budget to evaluate their practical bug detection capability. Distinctiveness of Bugs Detected by ERIQ.For each of the 54 unique logic bugs detected by ERIQ, we analyzed whether they could also be detected by existing approaches. Specifically, we started from each minimized test case and constructed corresponding test cases following the relation and const",
+      "section": "4.3 RQ2: Comparison with Existing Approaches",
+      "page": 6,
+      "char_offset": 31950,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[ 4,11,30,34]",
+      "technique": "tlp",
+      "sentence": "Specifically, we started from each minimized test case and constructed corresponding test cases following the relation and construction procedure defined by each approach [ 4,11,30,34].",
+      "context_before": "To answer RQ2, we compared ERIQ with the 4 baselines introduced in Section 4.1: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30]. We conducted two complementary experiments. First, we analyzed the distinctiveness of ERIQ’s 54 unique logic bugs by examining whether existing approaches could also detect them. Second, we compared ERIQ and the baselines under the same time budget to evaluate their practical bug detection capability. Distinctiveness of Bugs Detected by ERIQ.For each of the 54 unique logic bugs detected by ERIQ, we analyzed whether they could also be detected by existing approaches.",
+      "context_after": "A bug was considered detectable by a baseline if the corresponding test case violated the relation checked by that baseline; otherwise, it was considered not detectable by that baseline. Because this analysis evaluates the Detecting DBMS Bugs by Constructing Equivalent Representations of Intermediate Query Results relations checked by the baselines rather than their released implementations, it is independent of whether those implementations support the corresponding DBMSs. Table 2 presents the results. Table 2: Number of ERIQ’s 54 unique logic bugs detectable by existing approaches. DBMSERIQ",
+      "section": "4.3 RQ2: Comparison with Existing Approaches",
+      "page": 6,
+      "char_offset": 32472,
+      "cited_reference": {
+        "number": 30,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30. https://doi.org/10.1145/3428279",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "DBMSERIQ EDC Radar EETTLP MySQL 26 4 4 5 6 MariaDB 11 2 1 1 3 Percona 13 2 1 0 0 OceanBase 4 0 0 0 0 Total 54 8 6 6 9 Among the 54 unique logic bugs detected by ERIQ, EDC, Radar, EET, and TLP could detect 8, 6, 6, and 9 bugs, respectively, corresponding to 14.",
+      "context_before": "1,30,34]. A bug was considered detectable by a baseline if the corresponding test case violated the relation checked by that baseline; otherwise, it was considered not detectable by that baseline. Because this analysis evaluates the Detecting DBMS Bugs by Constructing Equivalent Representations of Intermediate Query Results relations checked by the baselines rather than their released implementations, it is independent of whether those implementations support the corresponding DBMSs. Table 2 presents the results. Table 2: Number of ERIQ’s 54 unique logic bugs detectable by existing approaches.",
+      "context_after": "8%, 11.1%, 11.1%, and 16.7%. TLP detected the largest number of ERIQ-detected bugs in this analysis, but only 9 of the 54 bugs; the other baselines detected even fewer. For each baseline, at least 45 of the 54 bugs detected by ERIQ were not detected. These results show that ERIQ provides complementary bug detection capability by exposing many bugs that are not detected by existing approaches. Bug Detection Capability under the Same Time Budget.We further compared the practical bug detection capability of ERIQ and the baselines under the same time budget. We ran each tool for 24 hours on every",
+      "section": "4.3 RQ2: Comparison with Existing Approaches",
+      "page": 7,
+      "char_offset": 33249,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP detected the largest number of ERIQ-detected bugs in this analysis, but only 9 of the 54 bugs; the other baselines detected even fewer.",
+      "context_before": "ations of Intermediate Query Results relations checked by the baselines rather than their released implementations, it is independent of whether those implementations support the corresponding DBMSs. Table 2 presents the results. Table 2: Number of ERIQ’s 54 unique logic bugs detectable by existing approaches. DBMSERIQ EDC Radar EETTLP MySQL 26 4 4 5 6 MariaDB 11 2 1 1 3 Percona 13 2 1 0 0 OceanBase 4 0 0 0 0 Total 54 8 6 6 9 Among the 54 unique logic bugs detected by ERIQ, EDC, Radar, EET, and TLP could detect 8, 6, 6, and 9 bugs, respectively, corresponding to 14.8%, 11.1%, 11.1%, and 16.7%.",
+      "context_after": "For each baseline, at least 45 of the 54 bugs detected by ERIQ were not detected. These results show that ERIQ provides complementary bug detection capability by exposing many bugs that are not detected by existing approaches. Bug Detection Capability under the Same Time Budget.We further compared the practical bug detection capability of ERIQ and the baselines under the same time budget. We ran each tool for 24 hours on every target DBMS it supports. For ERIQ, we reused the results reported in Section 4.2. For each baseline, we used its publicly available implementation and executed it on al",
+      "section": "4.3 RQ2: Comparison with Existing Approaches",
+      "page": 7,
+      "char_offset": 33538,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Across the 4 target DBMSs, ERIQ detected 54 logic bugs, whereas EDC, Radar, EET, and TLP detected 16, 2, 1, and 6 bugs, respectively.",
+      "context_before": "s reported in Section 4.2. For each baseline, we used its publicly available implementation and executed it on all supported target DBMSs. All experiments used the same DBMS versions and experimental environment described in Section 4.1. For each baseline, we minimized and internally deduplicated the generated reports following the general procedure described in Section 4.2. The detected bugs were reported to the corresponding DBMS communities, and developer responses confirmed that they corresponded to real bugs. Table 3 presents the number of logic bugs detected by each tool within 24 hours.",
+      "context_after": "Among the baselines, EDC detected the largest number of bugs and, like ERIQ, was evaluated on all 4 target DBMSs. ERIQ detected 54 logic bugs compared with EDC’s 16, more than three times as many under the same time budget. At the time of writing, developers had not classified any of the bugs detected by the baselines as duplicates of the 54 unique logic bugs found by ERIQ. These results demonstrate that ERIQ detects more logic bugs than existing approaches under the same time budget and provide additional evidence that ERIQ complements these baselines. Finding 2.Existing approaches can detec",
+      "section": "4.3 RQ2: Comparison with Existing Approaches",
+      "page": 7,
+      "char_offset": 34765,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[ 2,4,8,11,29–\n31,34–36,38,39,42]",
+      "technique": "dqp",
+      "sentence": "Existing approaches have proposed various techniques for detecting DBMS logic bugs [ 2,4,8,11,29– 31,34–36,38,39,42].",
+      "context_before": "functions cannot currently be detected. Second, ERIQ does not generate recursive CTEs because the recursive self-references required by such CTEs cannot be represented by the equivalent VIEW and TEMPT forms currently constructed by ERIQ. Third, when an equivalent TEMPT variant cannot be legally constructed, ERIQ can compare only the VIEW and CTE variants, potentially missing bugs that require TEMPT for exposure. Finally, ERIQ compares results using multiset semantics and therefore does not target bugs that manifest solely as differences in row ordering. 6 Related Work DBMS Logic Bug Detection.",
+      "context_after": "PQS [ 31] synthesizes queries that are guaranteed to return a selected pivot row and detects bugs when the pivot row is missing from the result. NoREC [ 29] rewrites a query into a form that inhibits DBMS optimizations and compares the results of the original and rewritten queries. TLP [ 30] partitions a query according to ternary logic and checks whether the combined results of the partitioned queries are consistent with the original query result. Pinolo [ 8] synthesizes queries whose results should be supersets or subsets of a seed query result and detects violations of the expected contain",
+      "section": "6 Related Work",
+      "page": 10,
+      "char_offset": 52857,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing Databases via Differential Query Plans.Proceedings of the ACM on Management of Data2, 3 (2024), 1–26. https://doi.org/10.1145/3654991",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "prints the DOI of the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [ 31] synthesizes queries that are guaranteed to return a selected pivot row and detects bugs when the pivot row is missing from the result.",
+      "context_before": "rences required by such CTEs cannot be represented by the equivalent VIEW and TEMPT forms currently constructed by ERIQ. Third, when an equivalent TEMPT variant cannot be legally constructed, ERIQ can compare only the VIEW and CTE variants, potentially missing bugs that require TEMPT for exposure. Finally, ERIQ compares results using multiset semantics and therefore does not target bugs that manifest solely as differences in row ordering. 6 Related Work DBMS Logic Bug Detection.Existing approaches have proposed various techniques for detecting DBMS logic bugs [ 2,4,8,11,29– 31,34–36,38,39,42].",
+      "context_after": "NoREC [ 29] rewrites a query into a form that inhibits DBMS optimizations and compares the results of the original and rewritten queries. TLP [ 30] partitions a query according to ternary logic and checks whether the combined results of the partitioned queries are consistent with the original query result. Pinolo [ 8] synthesizes queries whose results should be supersets or subsets of a seed query result and detects violations of the expected containment relation. EET [ 11] and CODDTest [ 42] apply equivalent transformations to query expressions and compare the results of the original and tra",
+      "section": "6 Related Work",
+      "page": 10,
+      "char_offset": 52974,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [ 29] rewrites a query into a form that inhibits DBMS optimizations and compares the results of the original and rewritten queries.",
+      "context_before": "t TEMPT variant cannot be legally constructed, ERIQ can compare only the VIEW and CTE variants, potentially missing bugs that require TEMPT for exposure. Finally, ERIQ compares results using multiset semantics and therefore does not target bugs that manifest solely as differences in row ordering. 6 Related Work DBMS Logic Bug Detection.Existing approaches have proposed various techniques for detecting DBMS logic bugs [ 2,4,8,11,29– 31,34–36,38,39,42]. PQS [ 31] synthesizes queries that are guaranteed to return a selected pivot row and detects bugs when the pivot row is missing from the result.",
+      "context_after": "TLP [ 30] partitions a query according to ternary logic and checks whether the combined results of the partitioned queries are consistent with the original query result. Pinolo [ 8] synthesizes queries whose results should be supersets or subsets of a seed query result and detects violations of the expected containment relation. EET [ 11] and CODDTest [ 42] apply equivalent transformations to query expressions and compare the results of the original and transformed queries. Radar [ 34] compares the results of the same query on databases containing the same data but different metadata. EDC [ 4",
+      "section": "6 Related Work",
+      "page": 10,
+      "char_offset": 53119,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [ 30] partitions a query according to ternary logic and checks whether the combined results of the partitioned queries are consistent with the original query result.",
+      "context_before": "T for exposure. Finally, ERIQ compares results using multiset semantics and therefore does not target bugs that manifest solely as differences in row ordering. 6 Related Work DBMS Logic Bug Detection.Existing approaches have proposed various techniques for detecting DBMS logic bugs [ 2,4,8,11,29– 31,34–36,38,39,42]. PQS [ 31] synthesizes queries that are guaranteed to return a selected pivot row and detects bugs when the pivot row is missing from the result. NoREC [ 29] rewrites a query into a form that inhibits DBMS optimizations and compares the results of the original and rewritten queries.",
+      "context_after": "Pinolo [ 8] synthesizes queries whose results should be supersets or subsets of a seed query result and detects violations of the expected containment relation. EET [ 11] and CODDTest [ 42] apply equivalent transformations to query expressions and compare the results of the original and transformed queries. Radar [ 34] compares the results of the same query on databases containing the same data but different metadata. EDC [ 4] replaces data operation expressions with precomputed results and compares the original and transformed query results. In contrast to these approaches, ERIQ checks resul",
+      "section": "6 Related Work",
+      "page": 10,
+      "char_offset": 53257,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "EET [ 11] and CODDTest [ 42] apply equivalent transformations to query expressions and compare the results of the original and transformed queries.",
+      "context_before": "thesizes queries that are guaranteed to return a selected pivot row and detects bugs when the pivot row is missing from the result. NoREC [ 29] rewrites a query into a form that inhibits DBMS optimizations and compares the results of the original and rewritten queries. TLP [ 30] partitions a query according to ternary logic and checks whether the combined results of the partitioned queries are consistent with the original query result. Pinolo [ 8] synthesizes queries whose results should be supersets or subsets of a seed query result and detects violations of the expected containment relation.",
+      "context_after": "Radar [ 34] compares the results of the same query on databases containing the same data but different metadata. EDC [ 4] replaces data operation expressions with precomputed results and compares the original and transformed query results. In contrast to these approaches, ERIQ checks result consistency across equivalent representations of intermediate query results. It constructs SQL variants that use a VIEW, a CTE, or a TEMPT to represent the same intermediate query result and checks whether the variants return consistent results. ERIQ therefore explores a new perspective for detecting DBMS",
+      "section": "6 Related Work",
+      "page": 10,
+      "char_offset": 53588,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[ 1,3,5–7,10,12,13,15,32,37,40,43]",
+      "technique": "qpg",
+      "sentence": "Existing DBMS test case generation techniques aim to improve the validity and diversity of generated SQL inputs [ 1,3,5–7,10,12,13,15,32,37,40,43].",
+      "context_before": "ecomputed results and compares the original and transformed query results. In contrast to these approaches, ERIQ checks result consistency across equivalent representations of intermediate query results. It constructs SQL variants that use a VIEW, a CTE, or a TEMPT to represent the same intermediate query result and checks whether the variants return consistent results. ERIQ therefore explores a new perspective for detecting DBMS logic bugs, complementing existing approaches. Detecting DBMS Bugs by Constructing Equivalent Representations of Intermediate Query Results DBMS Test Case Generation.",
+      "context_after": "SQLsmith [ 37] randomly constructs queries using database-schema information and AST-based rules. SQUIRREL [ 43] performs typebased mutations on an intermediate representation and uses dependency information to improve the syntactic and semantic validity of generated statements. Griffin [ 6] constructs metadata graphs that capture dependencies between statements and database objects and uses them for statement reshuffling and metadata-guided substitution. LEGO [ 13] uses coverage feedback to discover SQL type affinities and progressively generate diverse SQL statement sequences. SQLRight [ 15",
+      "section": "6 Related Work",
+      "page": 11,
+      "char_offset": 54502,
+      "cited_reference": {
+        "number": 1,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Engineering (ICSE). IEEE, 2060–2071. https://doi.org/10.1109/icse48619.2023. 00174",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "prints the DOI of the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG [ 1] uses query-plan diversity to guide databasestate mutations, and MIST [ 3] combines hierarchical SQL-feature guidance, error feedback, and coverage-guided Monte Carlo tree search for LLM-based DBMS test-case generation.",
+      "context_before": "ependency information to improve the syntactic and semantic validity of generated statements. Griffin [ 6] constructs metadata graphs that capture dependencies between statements and database objects and uses them for statement reshuffling and metadata-guided substitution. LEGO [ 13] uses coverage feedback to discover SQL type affinities and progressively generate diverse SQL statement sequences. SQLRight [ 15] combines coverage guidance with validity-oriented mutations, whereas DynSQL [ 10] dynamically collects databasestate information to incrementally construct complex and valid SQL inputs.",
+      "context_after": "ERIQ complements these approaches by constructing equivalent representations of intermediate query results and checking their result consistency. The diverse query-generation techniques employed by existing approaches could be incorporated into ERIQ to generate more diverse intermediate queries and referencing queries, potentially exposing additional DBMS logic bugs. Conversely, ERIQ’s representation-based consistency checking could be integrated with these generation techniques to further enhance their bug detection capability. 7 Conclusion This paper presents ERIQ, a novel testing approach",
+      "section": "6 Related Work",
+      "page": 11,
+      "char_offset": 55436,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M2",
+        "M5",
+        "M6"
+      ],
+      "describes_as_state_of_the_art": [
+        "M2",
+        "M5"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:49:48Z",
+    "is_model_written": true,
+    "summary": "ERIQ builds a second representation of a query's intermediate results that must be equivalent to the first, and reports a bug when the two disagree. Where earlier oracles transform the query, the expression, or the database state, ERIQ works on what the query produces part-way through, which reaches bugs in how intermediate results are materialised and reused. On MySQL, MariaDB, Percona and OceanBase it found 54 logic bugs.",
+    "narrative": "TLP is one of ERIQ's four baselines, named among state-of-the-art DBMS logic-bug detection approaches alongside EDC, Radar and EET, and included partly because EDC's own evaluation used it. The comparison is a careful one: rather than running each tool, the authors reconstruct test cases from each approach's relation and construction procedure starting from the same minimised cases. TLP detected 9 of ERIQ's 54 bugs -- more than any other baseline, and still a sixth of the total. PQS, NoREC and TLP are each described in related work as the query-level oracles ERIQ is positioned against.",
+    "roles": {
+      "M1": "definition",
+      "M2": "state_of_the_art",
+      "M3": "background",
+      "M4": "background",
+      "M5": "state_of_the_art",
+      "M6": "baseline",
+      "M7": "baseline",
+      "M8": "result_comparison",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "result_comparison",
+      "M12": "background",
+      "M13": "definition",
+      "M14": "definition",
+      "M15": "definition",
+      "M16": "definition",
+      "M17": "background",
+      "M18": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The baselines are reconstructed from each approach's published procedure rather than run from SQLancer, and nothing states that ERIQ's implementation reuses the codebase."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "ERIQ constructs equivalent representations of intermediate results, which M3 lists as distinct from transforming, partitioning or constructing queries. It is a different oracle, not a generalisation of TLP."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M5",
+          "M9",
+          "M10",
+          "M11"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M5",
+            "sentence": "We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30].",
+            "section": "4.1 Experimental Setup",
+            "page": 6
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "DBMSERIQ EDC Radar EETTLP MySQL 26 4 4 5 6 MariaDB 11 2 1 1 3 Percona 13 2 1 0 0 OceanBase 4 0 0 0 0 Total 54 8 6 6 9 Among the 54 unique logic bugs detected by ERIQ, EDC, Radar, EET, and TLP could detect 8, 6, 6, and 9 bugs, respectively, corresponding to 14.",
+            "section": "4.3 RQ2: Comparison with Existing Approaches",
+            "page": 7
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "TLP detected the largest number of ERIQ-detected bugs in this analysis, but only 9 of the 54 bugs; the other baselines detected even fewer.",
+            "section": "4.3 RQ2: Comparison with Existing Approaches",
+            "page": 7
+          },
+          {
+            "mention_id": "M11",
+            "sentence": "Across the 4 target DBMSs, ERIQ detected 54 logic bugs, whereas EDC, Radar, EET, and TLP detected 16, 2, 1, and 6 bugs, respectively.",
+            "section": "4.3 RQ2: Comparison with Existing Approaches",
+            "page": 7
+          }
+        ],
+        "reasoning": "M5 names TLP as one of four baselines, and M9, M10 and M11 report how many of ERIQ's bugs each baseline could detect and how many each found on its own.",
+        "techniques": [
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M2",
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "We further compared ERIQ with four state-of-the-art DBMS logic bug detection approaches: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30].",
+            "section": "1 Introduction",
+            "page": 1
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "We selected 4 state-of-the-art DBMS logic bug detection approaches as baselines: EDC [ 4], Radar [ 34], EET [ 11], and TLP [ 30].",
+            "section": "4.1 Experimental Setup",
+            "page": 6
+          }
+        ],
+        "reasoning": "M2 and M5 both describe TLP as one of four state-of-the-art DBMS logic bug detection approaches."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_arxiv_2609_00381.json b/_data/papers/paper_arxiv_2609_00381.json
new file mode 100644
index 0000000..e874af0
--- /dev/null
+++ b/_data/papers/paper_arxiv_2609_00381.json
@@ -0,0 +1,258 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:10:52Z",
+  "paper": {
+    "id": "paper:arxiv:2609.00381",
+    "title": "Bounded, Indeterminate, or a Bug: A Condition-Aware Oracle for Differential Testing of SQL Aggregates",
+    "authors": [
+      "Madhulatha Mandarapu",
+      "Sandeep Kunkunuru"
+    ],
+    "year": 2026,
+    "venue": null,
+    "doi": null,
+    "arxiv_id": "2609.00381",
+    "s2_paper_id": "cc8f54203ea326fd6d9d05e6856bf8b297829375",
+    "url": "https://arxiv.org/abs/2609.00381",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2609.00381",
+      "retrieved_at": "2026-09-07T17:10:52Z",
+      "chars": 31465,
+      "content_sha256": "sha256:6a2ac4557da977b132153295ed45177323cb278779f195482b850abd77fd5aa9"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1701
+      },
+      {
+        "number": "2",
+        "title": "The condition number and the testability boundary",
+        "start": 7585
+      },
+      {
+        "number": "3",
+        "title": "The oracle: verdicts and two boundaries",
+        "start": 9775
+      },
+      {
+        "number": "4",
+        "title": "Experimental setup",
+        "start": 11232
+      },
+      {
+        "number": "5",
+        "title": "Results I: SUM is the linear baseline",
+        "start": 12665
+      },
+      {
+        "number": "6",
+        "title": "Results II: variance recovers the algorithm",
+        "start": 14010
+      },
+      {
+        "number": "7",
+        "title": "Results III: real data versus its representation",
+        "start": 16122
+      }
+    ]
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "author_year_citation",
+      "surface": "Rigger [2025",
+      "technique": null,
+      "sentence": "Zhang and Rigger [2025] report that constant folding over floating-point values “can result in false alarms, which are avoided in practice by eschewing test cases with small or large floating-point values.",
+      "context_before": "on Differential testing is the workhorse of database correctness research: run the same query on two engines and treat a difference in results as evidence of a bug. It rests on an oracle—a rule that decides whether two answers should have agreed. For integers and strings the oracle is equality. For floating-point aggregates there is no such rule, because floating-point arithmetic is not associative: two engines that combine the same column in different orders return different doubles, and both are correct implementations of the aggregate. The field’s response has been to sidestep the question.",
+      "context_after": "” Zhong and Rigger [2024] document that DuckDB’s own harness treats two floats as matching when they differ by less than1%. The oracles of Rigger and Su [2020b,a,c] either restrict aggregates to a single pivot row, target predicates ∗madhulatha@samyama.ai †sandeep@samyama.ai 1 arXiv:2609.00381v1 [cs.DB] 18 Jul 2026 rather than aggregation, or compare an engine againstitselfby query partitioning, so a consistent rounding error cancels and is never observed. The same research group authors the modern differential-testing line, including Jiang et al. [2025], and it consistently routes around floa",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2313,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "author_year_citation",
+      "surface": "Rigger [2024",
+      "technique": null,
+      "sentence": "” Zhong and Rigger [2024] document that DuckDB’s own harness treats two floats as matching when they differ by less than1%.",
+      "context_before": "whether two answers should have agreed. For integers and strings the oracle is equality. For floating-point aggregates there is no such rule, because floating-point arithmetic is not associative: two engines that combine the same column in different orders return different doubles, and both are correct implementations of the aggregate. The field’s response has been to sidestep the question. Zhang and Rigger [2025] report that constant folding over floating-point values “can result in false alarms, which are avoided in practice by eschewing test cases with small or large floating-point values.",
+      "context_after": "The oracles of Rigger and Su [2020b,a,c] either restrict aggregates to a single pivot row, target predicates ∗madhulatha@samyama.ai †sandeep@samyama.ai 1 arXiv:2609.00381v1 [cs.DB] 18 Jul 2026 rather than aggregation, or compare an engine againstitselfby query partitioning, so a consistent rounding error cancels and is never observed. The same research group authors the modern differential-testing line, including Jiang et al. [2025], and it consistently routes around floating point rather than through it. An epsilon is not merely a crude oracle; it is wrong in a direction that depends on the",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2519,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "author_year_citation",
+      "surface": "Rigger and Su [2020b",
+      "technique": null,
+      "sentence": "The oracles of Rigger and Su [2020b,a,c] either restrict aggregates to a single pivot row, target predicates ∗madhulatha@samyama.",
+      "context_before": "re is no such rule, because floating-point arithmetic is not associative: two engines that combine the same column in different orders return different doubles, and both are correct implementations of the aggregate. The field’s response has been to sidestep the question. Zhang and Rigger [2025] report that constant folding over floating-point values “can result in false alarms, which are avoided in practice by eschewing test cases with small or large floating-point values.” Zhong and Rigger [2024] document that DuckDB’s own harness treats two floats as matching when they differ by less than1%.",
+      "context_after": "ai †sandeep@samyama.ai 1 arXiv:2609.00381v1 [cs.DB] 18 Jul 2026 rather than aggregation, or compare an engine againstitselfby query partitioning, so a consistent rounding error cancels and is never observed. The same research group authors the modern differential-testing line, including Jiang et al. [2025], and it consistently routes around floating point rather than through it. An epsilon is not merely a crude oracle; it is wrong in a direction that depends on the data. If the computation nearly cancels, the true answer is tiny while the intermediate magnitudes are large, and any implementati",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2642,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "DB] 18 Jul 2026 rather than aggregation, or compare an engine againstitselfby query partitioning, so a consistent rounding error cancels and is never observed.",
+      "context_before": "ect implementations of the aggregate. The field’s response has been to sidestep the question. Zhang and Rigger [2025] report that constant folding over floating-point values “can result in false alarms, which are avoided in practice by eschewing test cases with small or large floating-point values.” Zhong and Rigger [2024] document that DuckDB’s own harness treats two floats as matching when they differ by less than1%. The oracles of Rigger and Su [2020b,a,c] either restrict aggregates to a single pivot row, target predicates ∗madhulatha@samyama.ai †sandeep@samyama.ai 1 arXiv:2609.00381v1 [cs.",
+      "context_after": "The same research group authors the modern differential-testing line, including Jiang et al. [2025], and it consistently routes around floating point rather than through it. An epsilon is not merely a crude oracle; it is wrong in a direction that depends on the data. If the computation nearly cancels, the true answer is tiny while the intermediate magnitudes are large, and any implementation may lose most of its significant digits legitimately: an epsilon then reports a bug that is not there. If it does not cancel, the computation is very accurate, and an epsilon far larger than the achievabl",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2820,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The only route tosum’s indeterminate regime is fuzzer data: replicating SQLancer’s generator, exactly cancelling ±MAX drives κ≈10306, and the fraction of undecidable columns is non-monotone in n(peaking near20%at n=1000).",
+      "context_before": "returnbit-identicalplain sums at every κ, so a differential test across those three has no power against a summation bug; this is a property of the configuration, not the engines, and dissolves at their default parallel settings. The testability boundary is κ∗= 1/γn(9.0×1011at n=104), and the epsilons in use are sound at exactly one κ: a10−9rule crosses at κε≈900and DuckDB’s1%at ≈9×109. Real TPC-H sums—including Q9’s signed profit—sit at κ≈1, ten orders of magnitude inside the decidable zone, so the shipped epsilons are15 ×to1.5×108×too permissive and the practicalsumhazard is false negatives.",
+      "context_after": "The fullsummap is in the artifact; the rest of the paper concerns what changes whenp̸= 1. 6 Results II: variance recovers the algorithm The measured exponent identifies each engine’s algorithm (VE1).We sweep κVfrom 1to109atn=104and fit the log–log slope of each engine’s relative error against κV. Figure 2 shows a clean split. ClickHouse’s varPoptracks slope p=2.05and crosses relative error1near 5 κV≈107: it is the textbook one-pass algorithm, as its exponent alone reveals. The other four grow at most linearly (measured slopes0 .90for DuckDB, MySQL and the SQLite Welford reference,1 .05 for Po",
+      "section": "5 Results I: SUM is the linear baseline",
+      "page": 5,
+      "char_offset": 13697,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "author_year_citation",
+      "surface": "Rigger and Su [2020b",
+      "technique": null,
+      "sentence": "Rigger and Su [2020b] verify a single pivot row, which restricts multi-row aggregates; Rigger and Su [2020a] rewrites predicates; Rigger and Su [2020c] partitions a query against the same engine, so a consistent rounding error cancels rather than being detected.",
+      "context_before": "ators violate those assumptions and are excluded. Under gradualunderflow(subnormal inputs) the deviations xi−¯x round to zero and every engine returns variance0: the normalised-float bound does not hold, and every engine agrees, so this is a modelling limit, not a bug. Underoverflow(values near 10300) the exact∑x2 iexceeds the binary64 range and the result is genuinely ±∞, so the bound is undefined. The oracle applies to columns that are normalised and whose exact intermediate sums are finite—which is the entire real-data regime of §7. 10 Related work Differential and metamorphic DBMS testing.",
+      "context_after": "Zhang and Rigger [2025] avoid floating-point test cases explicitly; Zhong and Rigger [2024] record the1%epsilon used in practice; Jiang et al. [2025] test emerging engines against relational references with an exact-result oracle. None classifies a numeric discrepancy against a forward error bound, none uses a condition number, and none observes that the testability boundary depends on the engine’s aggregation algorithm. Numerical analysis and reproducibility.The summation bound and the framing of κare standard [Higham, 2002, Goldberg, 1991]; compensated summation is due to Kahan [1965], Neum",
+      "section": "7 Results III: real data versus its representation",
+      "page": 11,
+      "char_offset": 24162,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "author_year_citation",
+      "surface": "Rigger [2025",
+      "technique": null,
+      "sentence": "Zhang and Rigger [2025] avoid floating-point test cases explicitly; Zhong and Rigger [2024] record the1%epsilon used in practice; Jiang et al.",
+      "context_before": "a bug. Underoverflow(values near 10300) the exact∑x2 iexceeds the binary64 range and the result is genuinely ±∞, so the bound is undefined. The oracle applies to columns that are normalised and whose exact intermediate sums are finite—which is the entire real-data regime of §7. 10 Related work Differential and metamorphic DBMS testing.Rigger and Su [2020b] verify a single pivot row, which restricts multi-row aggregates; Rigger and Su [2020a] rewrites predicates; Rigger and Su [2020c] partitions a query against the same engine, so a consistent rounding error cancels rather than being detected.",
+      "context_after": "[2025] test emerging engines against relational references with an exact-result oracle. None classifies a numeric discrepancy against a forward error bound, none uses a condition number, and none observes that the testability boundary depends on the engine’s aggregation algorithm. Numerical analysis and reproducibility.The summation bound and the framing of κare standard [Higham, 2002, Goldberg, 1991]; compensated summation is due to Kahan [1965], Neumaier [1974] and reproducible summation to Demmel and Nguyen [2013]. The condition number of the sample variance and the one-pass-versus-two-pas",
+      "section": "7 Results III: real data versus its representation",
+      "page": 11,
+      "char_offset": 24424,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:24:46Z",
+    "is_model_written": true,
+    "summary": "This paper gives an oracle for differential testing of floating-point SQL aggregates. It argues that treating any cross-engine discrepancy as a bug is unsound for floating point, because non-associativity means engines legitimately disagree, and that the deciding factor is the engine's summation algorithm rather than the query. Ground truth is the exact rational value of the stored doubles, each discrepancy is classified exact, bounded or indeterminate, and a testability boundary is derived beyond which rounding cannot be separated from a bug.",
+    "narrative": "SQLancer's oracles are the paper's point of departure: it argues each of the three avoids the case it addresses -- PQS restricting aggregates to a single pivot row, NoREC rewriting predicates, TLP comparing an engine against itself so a consistent rounding error cancels. It also cites two SQLancer papers documenting that floating-point test cases are avoided in practice, and replicates SQLancer's generator to reach the indeterminate regime for SUM.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "motivation",
+      "M3": "motivation",
+      "M4": "motivation",
+      "M5": "reuse_component",
+      "M6": "motivation",
+      "M7": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "uncertain",
+        "mention_ids": [
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M5",
+            "sentence": "The only route tosum’s indeterminate regime is fuzzer data: replicating SQLancer’s generator, exactly cancelling ±MAX drives κ≈10306, and the fraction of undecidable columns is non-monotone in n(peaking near20%at n=1000).",
+            "section": "5 Results I: SUM is the linear baseline",
+            "page": 5
+          }
+        ],
+        "reasoning": "M5 says the only route to SUM's indeterminate regime is fuzzer data, replicating SQLancer's generator. Replicating a generator is not necessarily running its code, and no mention settles which.",
+        "reuse_kind": "generator"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1007_978_3_030_71058_3_5.json b/_data/papers/paper_doi_10_1007_978_3_030_71058_3_5.json
new file mode 100644
index 0000000..85102b9
--- /dev/null
+++ b/_data/papers/paper_doi_10_1007_978_3_030_71058_3_5.json
@@ -0,0 +1,314 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:16:23Z",
+  "paper": {
+    "id": "paper:doi:10.1007/978-3-030-71058-3_5",
+    "title": "Artemis: An Automatic Test Suite Generator for Large Scale OLAP Database",
+    "authors": [
+      "Kaiming Mi",
+      "Chunxi Zhang",
+      "Weining Qian",
+      "Rong Zhang"
+    ],
+    "year": 2021,
+    "venue": "Lecture notes in computer science",
+    "doi": "10.1007/978-3-030-71058-3_5",
+    "arxiv_id": null,
+    "s2_paper_id": null,
+    "url": "https://doi.org/10.1007/978-3-030-71058-3_5",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1007/978-3-030-71058-3_5",
+      "retrieved_at": "2026-09-08T17:16:23Z",
+      "chars": 34762,
+      "content_sha256": "sha256:93407b721a8225ea4063b9d39ff3c1cf33eaf4dd64329c72aa868ddfaad3cc5e"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 16,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1297
+      },
+      {
+        "number": "2",
+        "title": "Architecture Overview of Artemis",
+        "start": 4598
+      },
+      {
+        "number": "3",
+        "title": "Database Schema Generation",
+        "start": 5827
+      },
+      {
+        "number": "4",
+        "title": "Data Generation",
+        "start": 9018
+      },
+      {
+        "number": "4.1",
+        "title": "Deterministic Random Data Generation",
+        "start": 9036
+      },
+      {
+        "number": "4.2",
+        "title": "Parallel Data Generation",
+        "start": 11011
+      },
+      {
+        "number": "5",
+        "title": "Workload Generation",
+        "start": 12862
+      },
+      {
+        "number": "5.1",
+        "title": "Parameter Instantiation",
+        "start": 13999
+      },
+      {
+        "number": "6",
+        "title": "Oracle Generation",
+        "start": 15367
+      },
+      {
+        "number": "7",
+        "title": "Experiment",
+        "start": 20248
+      },
+      {
+        "number": "7.1",
+        "title": "Data Generation",
+        "start": 21549
+      },
+      {
+        "number": "7.2",
+        "title": "Query Generation",
+        "start": 23709
+      },
+      {
+        "number": "7.3",
+        "title": "Oracle Generation",
+        "start": 26432
+      },
+      {
+        "number": "8",
+        "title": "Related Work",
+        "start": 27904
+      },
+      {
+        "number": "9",
+        "title": "Conclusion",
+        "start": 31146
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "https://www.wolfram.com/mathematica/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Abdul Khalek, S., Khurshid, S.: Automated SQL query generation for systematic testing of database engines. In ASE 2010, pp. 329–332 (2010)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Alexandrov, A., Tzoumas, K., Markl, V.: Myriad: scalable and expressive data generation. In: VLDB 2012, vol. 5, issue 12, pp. 1890–1893 (2012)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Bati, H., Giakoumakis, L., Herbert, S., Surna, A.: A genetic approach for random testing of database systems. In: VLDB 2007, pp. 1243–1251 (2007)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Binnig, C., Kossmann, D., Lo, E.: Reverse query processing. In: ICDE 2007, pp. 506–515 (2007)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Binnig, C., Kossmann, D., Lo, E., Saenzbadillos, A.: Automatic result verification for the functional testing of a query language. In: ICDE 2008, pp. 1534–1536 (2008)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Binnig, C., Kossmann, D., Lo, E., ¨Ozsu, M.T.: Qagen generating query-aware test databases. In: SIGMOD 2007, pp. 341–352 (2007)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Hamlin, A., Herzog, J.: A test-suite generator for database systems. In: HPEC 2014, pp. 1–6 (2014)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "He, S., Manns, G., Saunders, J., Wang, W., Pollock, L., Soffa, M.L.: A statisticsbased performance testing methodology for cloud applications. In: ESEC/FSE 2019 (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Hoag, J.E., Thompson, C.W.: A parallel general-purpose synthetic data generator. SIGMOD Rec 36(1), 19–24 (2007)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jackson, D.: Alloy: a lightweight object modelling notation. ACM Trans. Softw. Eng. Methodol. 11(2), 256–290 (2002)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Khalek, S.A., Elkarablieh, B., Laleye, Y.O., Khurshid, S.: Query-aware test generation using a relational constraint solver. In: ASE 2008, pp. 238–247 (2008)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Khalek, S.A., Khurshid, S.: Systematic testing of database engines using a relational constraint solver. In: ICST 2011, pp. 50–59 (2011)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Li, Y., Zhang, R., Yang, X., Zhang, Z., Zhou, A.: Touchstone: generating enormous query-aware test databases. In: ATC 2018, pp. 575–586 (2018)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Lo, E., Cheng, N., Lin, W.W.K., Hon, W.K., Choi, B.: Mybenchmark: generating databases for query workloads. VLDB J. 23(6), 895–913 (2014)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Mishra, C., Koudas, N., Zuzarte, C.: Generating targeted queries for database testing. In: SIGMOD 2008, pp. 499–510 (2008)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Poess, M., Stephens, J.M.: Generating thousand benchmark queries in seconds. in: VLDB 2004, pp. 1045–1053 (2004)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Rigger, M., Su, Z.: Testing database engines via pivoted query synthesis. arXiv: Databases (2020)",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 19,
+      "text": "Slutz, D.R.: Massive stochastic testing of SQL. In: VLDB 1998, pp. 618–622 (1998) Artemis: An Automatic Test Suite Generator 89",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Torlak, E.: Scalable test data generation from multidimensional models. In: SIGSOFT FSE 2012, p. 36 (2012)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Zhou, J., Aghili, N., Ghaleini, E.N., Bui, D.T., Tahir, M.M., Koopialipoor, M.: A Monte Carlo simulation approach for effective assessment of flyrock based on intelligent system of neural network. Eng. Comput. 36(2), 1–11 (2020). https:// doi.org/10.1007/s00366-019-00726-z",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 18,
+      "text": "Rigger, M., Su, Z.: Testing database engines via pivoted query synthesis. arXiv: Databases (2020)",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 18]",
+      "technique": "pqs",
+      "sentence": "It can generate a large number of queries in parallel, which may be syntactically correct but may not run through the deep logicbehind codes [ 18].",
+      "context_before": "he test database may not return meaningful results, i.e., empty result set. Instead, query-aware database generation [ 7,14,16]i sp r o posed to generate databases for specific applications, which is suitable for performance benchmarking instead of functionality verification. In order to satisfy the requirement of thorough test coverage for functions, randomized workload c⃝Springer Nature Switzerland AG 2021 F. Wolf and W. Gao (Eds.): Bench 2020, LNCS 12614, pp. 74–89, 2021.https://doi.org/10.1007/978-3-030-71058-3 _5 Artemis: An Automatic Test Suite Generator 75 generation is preferred [ 4,19].",
+      "context_after": "The most difficult thing for randomized workload generation is the verification of running results on the synthetical databases. Currently, this process is commonly completed by differential testing [19], which compares execution results to the ones from an authoritative database,e.g., Oracle. A few work, e.g., ADUSA [ 2,12], have proposed to generate orcales along workload generation, but the computation complexity of these tools seriously lows down the usability.Since these three parts of generation work are usually addressed separately, it is not practical or applicable to simply combine them",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 2755,
+      "cited_reference": {
+        "number": 18,
+        "text": "Rigger, M., Su, Z.: Testing database engines via pivoted query synthesis. arXiv: Databases (2020)",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:23:30Z",
+    "is_model_written": true,
+    "summary": "Artemis generates test suites for large-scale OLAP databases, producing queries that exercise analytical workloads rather than the transactional patterns most generators emit. It aims for queries that reach deep code paths rather than merely parsing correctly.",
+    "narrative": "The single mention is the motivation the paper shares with SQLancer's work: that generating many queries in parallel yields statements which may be syntactically correct yet never run through the deep logic behind the code. The citation supports that observation; nothing of SQLancer's is used or measured against, and the mention is reachable only through the citation marker.",
+    "roles": {
+      "M1": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1007_978_3_030_88494_9_12.json b/_data/papers/paper_doi_10_1007_978_3_030_88494_9_12.json
new file mode 100644
index 0000000..fdf1540
--- /dev/null
+++ b/_data/papers/paper_doi_10_1007_978_3_030_88494_9_12.json
@@ -0,0 +1,414 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:43Z",
+  "paper": {
+    "id": "paper:doi:10.1007/978-3-030-88494-9_12",
+    "title": "Differential Monitoring",
+    "authors": [
+      "Fabian Mühlböck",
+      "T. Henzinger"
+    ],
+    "year": 2021,
+    "venue": "Runtime Verification",
+    "doi": "10.1007/978-3-030-88494-9_12",
+    "arxiv_id": null,
+    "s2_paper_id": "f232b1bd730f6c4d4a48be5f642074dffb7a1ed0",
+    "url": "https://doi.org/10.1007/978-3-030-88494-9_12",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1007/978-3-030-88494-9_12",
+      "retrieved_at": "2026-09-09T01:33:43Z",
+      "chars": 38527,
+      "content_sha256": "sha256:e024028e48d45ee5a3c4f2e0d91fd0507c5d9bca8c1522f70ab628d8c4dcf860"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 845
+      },
+      {
+        "number": "2",
+        "title": "Background and Related Work",
+        "start": 7132
+      },
+      {
+        "number": "3",
+        "title": "Challenges",
+        "start": 13330
+      },
+      {
+        "number": "4",
+        "title": "Experimental Results",
+        "start": 17087
+      },
+      {
+        "number": "5",
+        "title": "Conclusion",
+        "start": 28516
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Barringer, H., Falcone, Y., Havelund, K., Reger, G., Rydeheard, D.: Quantified event automata: towards expressive and efficient runtime monitors. In: Giannakopoulou, D., M´ ery, D. (eds.) FM 2012. LNCS, vol. 7436, pp. 68–84. Springer, Heidelberg (2012). https://doi.org/10.1007/978-3-642-32759-9 9",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Bartocci, E., Falcone, Y. (eds.): Lectures on Runtime Verification-Introductory and Advanced Topics. Lecture Notes in Computer Science, vol. 10457. Springer, Heidelberg (2018). https://doi.org/10.1007/978-3-319-75632-5",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Bauer, A., Leucker, M., Schallhart, C.: Runtime verification for LTL and TLTL. ACM Trans. Softw. Eng. Methodol. 20(4) (2011). https://doi.org/10.1145/ 2000799.2000800",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Berger, E.D., Zorn, B.G.: DieHard: probabilistic memory safety for unsafe languages. In: PLDI 2006, p. 158–168. Association for Computing Machinery, New York (2006). https://doi.org/10.1145/1133981.1134000",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Bonakdarpour, B., Navabpour, S., Fischmeister, S.: Time-triggered runtime verification. Formal Methods Syst. Design 43(1), 29–60 (2013). https://doi.org/10. 1007/s10703-012-0182-0",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Cadar, C., Hosek, P.: Multi-version software updates. In: HotSWUp 2012, pp. 36– 40 (2012). https://doi.org/10.1109/HotSWUp.2012.6226615",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Chen, F., Ro¸ su, G.: Parametric trace slicing and monitoring. In: Kowalewski, S., Philippou, A. (eds.) TACAS 2009. LNCS, vol. 5505, pp. 246–261. Springer, Hei-delberg (2009). https://doi.org/10.1007/978-3-642-00768-2 23",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Chen, L., Avizienis, A.: N-version programming: a fault-tolerance approach to reliability of software operation. In: FTCS 1978, vol. 1, pp. 3–9 (1978)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Chen, L., Avizienis, A.: N-version programming: a fault-tolerance approach to reliability of software operation. In: FTCS 1995, ‘Highlights from Twenty-FiveYears’, p. 113ff (1995). https://doi.org/10.1109/FTCSH.1995.532621",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Clarkson, M.R., Finkbeiner, B., Koleini, M., Micinski, K.K., Rabe, M.N., S´ anchez, C.: Temporal logics for hyperproperties. In: Abadi, M., Kremer, S. (eds.) POST2014. LNCS, vol. 8414, pp. 265–284. Springer, Heidelberg (2014). https://doi.org/ 10.1007/978-3-642-54792-8 15",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Coppens, B., Sutter, B.D., Volckaert, S.: Multi-variant execution environments. In: Larsen, P., Sadeghi, A. (eds.) The Continuing Arms Race: Code-Reuse Attacks and Defenses, pp. 211–258. ACM/Morgan & Claypool (2018). https://doi.org/10. 1145/3129743.3129752",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Cox, B., et al.: N-Variant systems: a secretless framework for security through diversity. In: USENIX-SS 2006. USENIX Association, USA (2006). https:// www.usenix.org/conference/15th-usenix-security-symposium/n-variant-systems-secretless-framework-security-through",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Demri, S., Lazic, R.: LTL with the freeze quantifier and register automata. ACM Trans. Comput. Log. 10(3), 16:1–16:30 (2009). https://doi.org/10.1145/1507244. 1507246",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Elmendorf, W.: Fault-tolerant programming. In: FTCS 1972, pp. 79–83 (1972)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Evans, R.B., Savoia, A.: Differential testing: a new approach to change detection. In: ESEC-FSE companion 2007, pp. 549–552. Association for Computing Machinery, New York (2007). https://doi.org/10.1145/1295014.1295038",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Falcone, Y., Fernandez, J., Mounier, L.: What can you verify and enforce at runtime? Int. J. Softw. Tools Technol. Transf. 14(3), 349–382 (2012). https://doi.org/ 10.1007/s10009-011-0196-8 242 F. Muehlboeck and T. A. Henzinger",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Falcone, Y., Mariani, L., Rollet, A., Saha, S.: Runtime failure prevention and reaction. In: Bartocci, E., Falcone, Y. (eds.) Lectures on Runtime Verification. LNCS, vol. 10457, pp. 103–134. Springer, Cham (2018). https://doi.org/10.1007/978-3319-75632-5 4",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Fischler, M.A. et al.: Distinct software: an approach to reliable computing. In: 2nd USA-Japan Computer Conference, pp. 1–7 (1975)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Fokkink, W.J.: Introduction to Process Algebra. Texts in Theoretical Computer Science. An EATCS Series. Springer, Heidelberg (2000). https://doi.org/10.1007/ 978-3-662-04293-9",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Girard, E., Rault, J.: A programming technique for software reliability. In: IEEE Symposium on Computer Software Reliability, pp. 44–50 (1973)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Glabbeek, R.J.: The linear timebranching time spectrum. In: Baeten, J.C.M., Klop, J.W. (eds.) CONCUR 1990. LNCS, vol. 458, pp. 278–297. Springer, Heidelberg (1990). https://doi.org/10.1007/BFb0039066",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "van Glabbeek, R.J., et al.: The linear time—branching time spectrum II. In: Best, E. (ed.) CONCUR 1993. LNCS, vol. 715, pp. 66–81. Springer, Heidelberg (1993). https://doi.org/10.1007/3-540-57208-2 6",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Grigore, R., Distefano, D., Petersen, R.L., Tzevelekos, N.: Runtime verification based on register automata. In: Piterman, N., Smolka, S.A. (eds.) TACAS 2013. LNCS, vol. 7795, pp. 260–276. Springer, Heidelberg (2013). https://doi.org/10. 1007/978-3-642-36742-7 19",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Groce, A., Holzmann, G., Joshi, R.: Randomized differential testing as a prelude to formal verification. In: ICSE 2007, pp. 621–631. IEEE Computer Society, USA(2007). https://doi.org/10.1109/ICSE.2007.68",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Havelund, K., Reger, G., Ro¸ su, G.: Runtime verification past experiences and future projections. In: Steffen, B., Woeginger, G. (eds.) Computing and SoftwareScience. LNCS, vol. 10000, pp. 532–562. Springer, Cham (2019). https://doi.org/ 10.1007/978-3-319-91908-9 25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Hosek, P., Cadar, C.: VARAN the unbelievable: an efficient N-version execution framework. In: ASPLOS 2015, pp. 339–353. Association for Computing Machinery, New York (2015). https://doi.org/10.1145/2694344.2694390",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Kallas, K., Niksic, F., Stanford, C., Alur, R.: DiffStream: differential output testing for stream processing programs. PACMPL 4(OOPSLA) (2020). https://doi.org/ 10.1145/3428221",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Knight, J.C., Leveson, N.G.: An experimental evaluation of the assumption of independence in multiversion programming. IEEE Trans. Softw. Eng. 12(1), 96– 109 (1986). https://doi.org/10.1109/TSE.1986.6312924",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Knight, J.C., Leveson, N.G.: A reply to the criticisms of the Knight & Leveson experiment. ACMSIGSOFT Softw. Eng. Notes 15(1), 24–35 (1990). https://doi. org/10.1145/382294.382710",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Kopetz, H.: Software redundancy in real time systems. In: IFIP Congress 1974, pp. 182–186. North-Holland (1974)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Lehmann, D., Pradel, M.: Feedback-directed differential testing of interactive debuggers. In: ESEC/FSE 2018, pp. 610–620. Association for Computing Machinery, New York (2018). https://doi.org/10.1145/3236024.3236037",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "McKeeman, W.M.: Differential testing for software. Digit. Tech. J. 10(1), 100–107 (1998). http://www.hpl.hp.com/hpjournal/dtj/vol10num1/vol10num1art9.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Muehlboeck, F., Henzinger, T.A.: Differential monitoring. Technical report 9946, IST Austria (2021). https://research-explorer.app.ist.ac.at/librecat/record/9946 Differential Monitoring 243",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Pina, L., Andronidis, A., Hicks, M., Cadar, C.: MVEDSUA: higher availability dynamic software updates via multi-version execution. In: ASPLOS 2019, pp. 573– [585] ACM (2019). https://doi.org/10.1145/3297858.3304063",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Pnueli, A., Zaks, A.: PSL model checking and run-time verification via testers. In: Misra, J., Nipkow, T., Sekerinski, E. (eds.) FM 2006. LNCS, vol. 4085, pp. 573–586. Springer, Heidelberg (2006). https://doi.org/10.1007/11813040 38",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Pomerance, C., Selfridge, J.L., Wagstaff, S.S.: The pseudoprimes to 25 ·109.M a t h. Comput. 35(151), 1003–1026 (1980)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Rigger, M., Su, Z.: Detecting optimization bugs in database engines via nonoptimizing reference engine construction. In: ESEC/FSE 2020, pp. 1140–1152. Association for Computing Machinery, New York (2020). https://doi.org/10.1145/ 3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. PACMPL 4(OOPSLA) (2020). https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "S´anchez, C.: Online and offline stream runtime verification of synchronous systems. In: Colombo, C., Leucker, M. (eds.) RV 2018. LNCS, vol. 11237, pp. 138–163. Springer, Cham (2018). https://doi.org/10.1007/978-3-030-03769-7 9",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Slutz, D.R.: Massive stochastic testing of SQL. In: VLDB 1998, pp. 618–622. Morgan Kaufmann (1998). http://www.vldb.org/conf/1998/p618.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Volckaert, S., De Sutter, B., De Baets, T., De Bosschere, K.: GHUMVEE: efficient, effective, and flexible replication. In: Garcia-Alfaro, J., Cuppens, F., CuppensBoulahia, N., Miri, A., Tawbi, N. (eds.) FPS 2012. LNCS, vol. 7743, pp. 261–277. Springer, Heidelberg (2013). https://doi.org/10.1007/978-3-642-37119-6 17",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Yang, X., Chen, Y., Eide, E., Regehr, J.: Finding and understanding bugs in C compilers. In: PLDI 2011, pp. 283–294. Association for Computing Machinery, New York (2011). https://doi.org/10.1145/1993498.1993532",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 5,
+      "text": "Bonakdarpour, B., Navabpour, S., Fischmeister, S.: Time-triggered runtime verification. Formal Methods Syst. Design 43(1), 29–60 (2013). https://doi.org/10. 1007/s10703-012-0182-0",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Rigger, M., Su, Z.: Detecting optimization bugs in database engines via nonoptimizing reference engine construction. In: ESEC/FSE 2020, pp. 1140–1152. Association for Computing Machinery, New York (2020). https://doi.org/10.1145/ 3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 38,
+      "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. PACMPL 4(OOPSLA) (2020). https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 37,38,40]",
+      "technique": "norec",
+      "sentence": "This technique has been fruitfully applied to finding bugs in Javascript debuggers [ 31], C compilers [ 42], and SQL databases [ 37,38,40].",
+      "context_before": "s of programs with the environment rather than arbitrary program state in order to both provide aless intrusive interface and exploit modern hardware/operating system architecture, but unlike n-version execution, we seek to recover the idea of truly diverse implementations. Differential Testing. Differential Testing [ 15,24,32] is a well-known technique to test programs for which multiple versions exist. A large number of automatically generated test inputs are fed to n>1 supposedly equivalent programs. Any differences detected in their output indicate possible bugs that need to be inves-tigated.",
+      "context_after": "DiffStream [ 27] is a framework supporting differential testing of stream outputs, which is closely related to our implementation of differential monitoring. They key technical difference is that differential monitoring does not only track and compare a set of (potentially unbounded) streams, but also needs to helpprograms stay in sync (see Sect. 3). For system calls and other events, the atomicity of stream elements can itself be in need of specification, as one system call may be equivalent to a sequence of several other ones. Finally, DiffStream ignores the question of what to output for equivale",
+      "section": "2 Background and Related Work",
+      "page": 4,
+      "char_offset": 9967,
+      "cited_reference": {
+        "number": 37,
+        "text": "Rigger, M., Su, Z.: Detecting optimization bugs in database engines via nonoptimizing reference engine construction. In: ESEC/FSE 2020, pp. 1140–1152. Association for Computing Machinery, New York (2020). https://doi.org/10.1145/ 3368089.3409710",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 1,5,7,13,23,35]",
+      "technique": null,
+      "sentence": "A considerable body of work exists on various specification languages based on linear temporal logic and similar logics [ 1,5,7,13,23,35], and there are specification languages specifically for properties of streams [ 39], but these languages are interpreted over individual traces, rather than tuples of traces produced by supposedly equivalent programs.",
+      "context_before": "bugs can be found (and eliminated) by simply comparing the outputs of different but supposedly equivalent programs. Run-Time Verification/Monitoring. Run-time verification (RV) is the general area of monitoring and possibly enforcing that a given program satisfiessome properties, typically related in some way to the program’s overall correctness [ 2,25]. In RV, a program generates a trace of interesting events, and a specification of the program’s behavior allows us to build a monitor that checkssuch a trace of interesting events for whether it (possibly or definitely) conforms to the specification.",
+      "context_after": "Especially in the area Differential Monitoring 235 Program 1 Program 2Equivalence CheckerOutput ProcessorInput Processor Input 2Input 1 Output 2Output 1 Output 2Output 1 Verdict EnvironmentOutput InputMonitor State * * * Fig. 1. The logical parts of differential monitoring of security, languages like Hyper-LTL [ 10] are used on sets of traces (or, oftentimes, pairs of traces). However, similar to n-version execution, hyperlogics are usually interpreted over sets (or pairs) of traces that are generated by multiple executions of a single (often reactive or otherwise nondeterministic) program. In",
+      "section": "2 Background and Related Work",
+      "page": 4,
+      "char_offset": 11907,
+      "cited_reference": {
+        "number": 5,
+        "text": "Bonakdarpour, B., Navabpour, S., Fischmeister, S.: Time-triggered runtime verification. Formal Methods Syst. Design 43(1), 29–60 (2013). https://doi.org/10. 1007/s10703-012-0182-0",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:23:30Z",
+    "is_model_written": true,
+    "summary": "Differential monitoring runs two implementations of the same specification side by side and reports divergence, giving a runtime-verification oracle without a written specification. The paper develops the idea for runtime monitors, where writing a formal property is often harder than obtaining a second implementation.",
+    "narrative": "SQL databases are named as one of the domains where this style of testing has been applied fruitfully, cited alongside JavaScript debuggers and C compilers -- the SQLancer papers are the citation behind the SQL databases half of that sentence. The paper's own contribution is in runtime verification, and it distinguishes its setting from specification languages interpreted over individual traces rather than over tuples of traces.",
+    "roles": {
+      "M1": "background",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1007_978_3_031_51479_1_17.json b/_data/papers/paper_doi_10_1007_978_3_031_51479_1_17.json
new file mode 100644
index 0000000..9ff0b92
--- /dev/null
+++ b/_data/papers/paper_doi_10_1007_978_3_031_51479_1_17.json
@@ -0,0 +1,474 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:16:05Z",
+  "paper": {
+    "id": "paper:doi:10.1007/978-3-031-51479-1_17",
+    "title": "Syntax-Aware Mutation for Testing the Solidity Compiler",
+    "authors": [
+      "Charalambos Mitropoulos",
+      "T. Sotiropoulos",
+      "S. Ioannidis",
+      "Dimitris Mitropoulos"
+    ],
+    "year": 2023,
+    "venue": "European Symposium on Research in Computer Security",
+    "doi": "10.1007/978-3-031-51479-1_17",
+    "arxiv_id": null,
+    "s2_paper_id": "372393a671600484f8859510cc5553b1345bdec7",
+    "url": "https://doi.org/10.1007/978-3-031-51479-1_17",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1007/978-3-031-51479-1_17",
+      "retrieved_at": "2026-09-08T17:16:05Z",
+      "chars": 57383,
+      "content_sha256": "sha256:0089c96e581264450ee942b0a14650d7e7001e5755dbe7f65d831232b4d855d4"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 21,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2115
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 8642
+      },
+      {
+        "number": "2.1",
+        "title": "The Solidity Compiler",
+        "start": 8900
+      },
+      {
+        "number": "2.2",
+        "title": "Bugs in the Solidity Compiler",
+        "start": 10199
+      },
+      {
+        "number": "2.3",
+        "title": "Limitations of State-of-the-Art Fuzzers",
+        "start": 12044
+      },
+      {
+        "number": "3",
+        "title": "Fuzzing Approach",
+        "start": 14648
+      },
+      {
+        "number": "3.1",
+        "title": "Syntax-Aware Mutation",
+        "start": 17134
+      },
+      {
+        "number": "3.1.1",
+        "title": "Operator, Statement and Data Type Change Strategy Our first",
+        "start": 17397
+      },
+      {
+        "number": "3.2",
+        "title": "Mutation Strategy Prioritization",
+        "start": 24603
+      },
+      {
+        "number": "3.3",
+        "title": "Fuzzol",
+        "start": 27226
+      },
+      {
+        "number": "4",
+        "title": "Evaluation",
+        "start": 28342
+      },
+      {
+        "number": "4.1",
+        "title": "Evaluation Setup",
+        "start": 28776
+      },
+      {
+        "number": "4.2",
+        "title": "RQ1: Discovering Bugs",
+        "start": 29451
+      },
+      {
+        "number": "4.3",
+        "title": "RQ2: Comparing Syntax-Aware and Grammar-Blind Strategies",
+        "start": 33795
+      },
+      {
+        "number": "4.4",
+        "title": "RQ3: Comparison with State-of-the-Art Fuzzers",
+        "start": 36461
+      },
+      {
+        "number": "4.5",
+        "title": "RQ4: Mutation Strategy Prioritization Algorithm",
+        "start": 39315
+      },
+      {
+        "number": "5",
+        "title": "Related Work",
+        "start": 42242
+      },
+      {
+        "number": "6",
+        "title": "Conclusion",
+        "start": 45247
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "The Counterparty financial platform. https://counterparty.io/. Accessed 15 Jan 2023 Syntax-Aware Mutation for Testing the Solidity Compiler 345",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Hedera hashgraph. Accessed 15 Jan 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Optimized contract crash. https://github.com/ethereum/solidity/issues/12840. Accessed 05 Jan 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Optimized contract freeze. https://github.com/ethereum/solidity/issues/12848. Accessed 03 Jan 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Solidity. https://docs.soliditylang.org/en/v0.8.0/. Accessed 03 Jan 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Solidity compilerissues catalog. https://github.com/ethereum/solidity/issues. Accessed 15 Jan 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "The Solidity contract-oriented programming language Github repository. https:// github.com/ethereum/solidity. Accessed 05 Jan 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Z3 GitHub repository (2021). https://github.com/Z3Prover/z3. Accessed 20 Jan 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Aschermann, C., Frassetto, T., Holz, T., Jauernig, P., Sadeghi, A., Teuchert, D.: NAUTILUS: fishing for deep bugs with grammars. In: Proceedings of the 26thAnnual Network and Distributed System Security Symposium (NDSS) (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Atlidakis, V., Godefroid, P., Polishchuk, M.: Restler: stateful rest API fuzzing. In: Proceedings of the 41st International Conference on Software Engineering, ICSE2019, pp. 748–758. IEEE Press (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Blazytko, T., et al.: Grimoire: synthesizing structure while fuzzing. In: Proceedings of the 28th USENIX Conference on Security Symposium, pp. 1985–2002. USENIXAssociation, USA (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "B¨ ohme, M., Pham, V.T., Nguyen, M.D., Roychoudhury, A.: Directed greybox fuzzing. In: Proceedings of the 2017 ACMSIGSAC Conference on Computer andCommunications Security, CCS 2017, pp. 2329–2344. Association for Computing Machinery, New York (2017)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "B¨ ohme, M., Pham, V.T., Roychoudhury, A.: Coverage-based greybox fuzzing as Markov chain, CCS 2016, pp. 1032–1043. Association for Computing Machinery, New York (2016)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Bounimova, E., Godefroid, P., Molnar, D.: Billions and billions of constraints: whitebox fuzz testing in production. In: Proceedings of the 2013 International Conference on Software Engineering, ICSE 2013, pp. 122–131. IEEE Press (2013)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Brent, L., Grech, N., Lagouvardos, S., Scholz, B., Smaragdakis, Y.: Ethainter: a smart contract security analyzer for composite vulnerabilities. In: Proceedings of the 41st ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2020, pp. 454–469. Association for Computing Machinery,New York (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Browne, R.: Ether, the world’s second-biggest cryptocurrency, is closing in on an all-time high (2021). https://www.cnbc.com/2021/01/19/bitcoin-ethereum-ethcryptocurrency-nears-all-time-high.html. Accessed 20 Jan 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Cha, S.K., Woo, M., Brumley, D.: Program-adaptive mutational fuzzing. In: Proceedings of the. 725–741. IEEE Computer Society, USA (2015)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Chaliasos, S., Gervais, A., Livshits, B.: A study of inline assembly in solidity smart contracts. Proc. ACM Program. Lang. 6(OOPSLA2) (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Chaliasos, S., Sotiropoulos, T., Spinellis, D., Gervais, A., Livshits, B., Mitropoulos, D.: Finding typing compiler bugs. In: Proceedings of the 43rd ACM SIGPLANInternational Conference on Programming Language Design and Implementation, PLDI 2022, pp. 183–198. ACM, New York (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Chen, H., et al.: Hawkeye: towards a desired directed grey-box fuzzer. In: Proceedings of the 2018 ACMSIGSAC Conference on Computer and Communications 346 C. Mitropoulos et al. Security, CCS 2018, pp. 2095–2108. Association for Computing Machinery, New York (2018)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Chen, J., et al.: A survey of compiler testing. ACM Comput. Surv. 53(1), 1–36 (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Cordeiro, L., Fischer, B., Marques-Silva, J.: SMT-based bounded model checking for embedded ANSI-C software. In: Proceedings of the 2009 IEEE/ACM International Conference on Automated Software Engineering, ASE 2009, pp. 137–148. IEEE Computer Society, USA (2009)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "de Moura, L., Bjørner, N.: Z3: an efficient SMT solver. In: Ramakrishnan, C.R., Rehof, J. (eds.) TACAS 2008. LNCS, vol. 4963, pp. 337–340. Springer, Heidelberg(2008). https://doi.org/10.1007/978-3-540-78800-3 24",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Ghaleb, A., Pattabiraman, K.: How effective are smart contract analysis tools? Evaluating smart contract static analysis tools using bug injection. In: Proceedingsof the 29th ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2020, pp. 415–427. ACM, New York (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Godefroid, P.: Fuzzing: hack, art, and science. Commun. ACM 63(2), 70–76 (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Godefroid, P., Levin, M.Y., Molnar, D.: Sage: whitebox fuzzing for security testing: Sage has had a remarkable impact at Microsoft. Queue 10(1), 20–27 (2012)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Grech, N., Kong, M., Jurisevic, A., Brent, L., Scholz, B., Smaragdakis, Y.: Madmax: analyzing the out-of-gas world of smart contracts. Commun. ACM 63(10), 87–95 (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Groce, A., van Tonder, R., Kalburgi, G.T., Le Goues, C.: Making no-fuss compiler fuzzing effective. In: Proceedings of the 31st ACMSIGPLAN International Confer-ence on Compiler Construction, CC 2022, pp. 194–204. Association for Computing Machinery, New York (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Holler, C., Herzig, K., Zeller, A.: Fuzzing with code fragments. In: Proceedings of the 21st USENIX Conference on Security Symposium, Security 2012, p. 38. USENIX Association, USA (2012)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Jiang, B., Liu, Y., Chan, W.K.: Contractfuzzer: fuzzing smart contracts for vulnerability detection. In: Proceedings of the 33rd ACM/IEEE International Conferenceon Automated Software Engineering, ASE 2018, pp. 259–269. Association for Computing Machinery, New York (2018)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Klees, G., Ruef, A., Cooper, B., Wei, S., Hicks, M.: Evaluating fuzz testing. In: Proceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security, CCS 2018, pp. 2123–2138. Association for Computing Machinery, New York (2018)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Lemieux, C., Sen, K.: Fairfuzz: a targeted mutation strategy for increasing greybox fuzz testing coverage. In: Proceedings of the 33rd ACM/IEEE International Con-ference on Automated Software Engineering, ASE 2018, pp. 475–485. Association for Computing Machinery, New York (2018)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Li, Y., et al.: Cerebro: context-aware adaptive fuzzing for effective vulnerability detection. In: Proceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2019, pp. 533–544. ACM, New York (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Livinskii, V., Babokin, D., Regehr, J.: Random testing for C and C++ compilers with YARPGen. Proc. ACM Program. Lang. 4(OOPSLA) (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Lyu, C., et al.: MOPT: optimized mutation scheduling for fuzzers. In: Proceedings of the 28th USENIX Conference on Security Symposium, pp. 1949–1966. USENIX Association, USA (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Lyu, C., et al.: EMS: history-driven mutation for coverage-based fuzzing. In: 29th Annual Network and Distributed System Security Symposium (2022) Syntax-Aware Mutation for Testing the Solidity Compiler 347",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Zalewski, M.: American fuzzy lop (2013). https://lcamtuf.coredump.cx/afl/. Accessed 13 Jan 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Rash, M.: AFL-COV-AFL fuzzing code coverage (2021). https://github.com/ mrash/afl-cov. Accessed 06 Jan 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Rigger, M., Su, Z.: Testing database engines via pivoted query synthesis. In: 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI2020), pp. 667–682. USENIX Association (2020)",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 40,
+      "text": "Sotiropoulos, T., Chaliasos, S., Atlidakis, V., Mitropoulos, D., Spinellis, D.: Data-oriented differential testing of object-relational mapping systems. In:. 1535–1547 (2021)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Sun, C., Le, V., Su, Z.: Finding and analyzing compiler warning defects. In: Proceedings of the 38th International Conference on Software Engineering, ICSE 2016, pp. 203–213. Association for Computing Machinery, New York (2016)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Sun, C., Le, V., Zhang, Q., Su, Z.: Toward understanding compiler bugs in GCC and LLVM. In: Proceedings of the 25th International Symposium on Software Testing and Analysis, ISSTA 2016, pp. 294–305. Association for Computing Machinery,New York (2016)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Veggalam, S., Rawat, S., Haller, I., Bos, H.: IFuzzer: an evolutionary interpreter fuzzer using genetic programming. In: Askoxylakis, I., Ioannidis, S., Katsikas, S.,Meadows, C. (eds.) ESORICS 2016. LNCS, vol. 9878, pp. 581–601. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-45744-4 29",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Wang, J., Chen, B., Wei, L., Liu, Y.: Superion: Grammar-aware greybox fuzzing. In: Proceedings of the 41st International Conference on Software Engineering, ICSE 2019, pp. 724–735. IEEE Press (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Weimer, W., Nguyen, T., Le Goues, C., Forrest, S.: Automatically finding patches using genetic programming. In: Proceedings of the 31st International Conference on Software Engineering, ICSE 2009, pp. 364–374. IEEE, USA (2009)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "W¨ ustholz, V., Christakis, M.: Harvey: a greybox fuzzer for smart contracts. In: Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2020, pp. 1398–1409. Association for Computing Machinery, New York (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Yan, S., Wu, C., Li, H., Shao, W., Jia, C.: Pathafl: path-coverage assisted fuzzing. In: Proceedings of the 15th ACM Asia Conference on Computer and Communications Security, ASIACCS 2020, pp. 598–609. Association for Computing Machin-ery, New York (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Yang, X., Chen, Y., Eide, E., Regehr, J.: Finding and understanding bugs in C compilers. SIGPLAN Not. 46(6), 283–294 (2011)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Yang, X., Chen, Y., Eide, E., Regehr, J.: Finding and understanding bugs in c compilers. In: Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2011, pp. 283–294. Association for Computing Machinery, New York (2011)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Zubairy, R.: Create a blockchain app for loyalty points with Hyperledger Fabric Ethereum Virtual Machine (2018). Accessed 06 Jan 2023",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 39,
+      "text": "Rigger, M., Su, Z.: Testing database engines via pivoted query synthesis. In: 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI2020), pp. 667–682. USENIX Association (2020)",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 39,40]",
+      "technique": "pqs",
+      "sentence": "Fuzzing has been used to identify bugs in miscellaneous entities such as system libraries [ 35], web and cloud applications [ 10], data-oriented systems [ 39,40], and compilers [ 19,34,48].",
+      "context_before": "gh studies focusing onsolc,t h estandard Solidity compiler. solcis a relatively new compiler that counts ∼100 releases since 2015 [ 5]. Given the intricate nature of Solidity, solc offers various special constructs related to smart contract functionalities includ-ing formal software verification and inline assembly. Due to this complexity, solc has exhibited a variety of bugs related to data structure mishandling, inadequate sanity checks, and unsound optimizations [ 6]. For the last two decades, fuzzing has become a standard technique for assessing software reliability and security [ 14,25,26].",
+      "context_after": "When it comes to programs whose inputs follow specific grammars (e.g. compilers), grammar-blind fuzzers (such as AFL [ 37]) struggle to get past syntax checks and explore deeper code. To this end, researchers have introduced a number of grammar-based fuzzing strategies [ 9,43,44], and have applied them to various domains, from PHP and Lua interpreters to JavaScript engines. However, current grammar-based fuzzers have a number of disadvantages. For instance, Superion [44], performs some mutations that fail to preserve a correct syntax for the test cases it generates. In addition, many of these",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3638,
+      "cited_reference": {
+        "number": 39,
+        "text": "Rigger, M., Su, Z.: Testing database engines via pivoted query synthesis. In: 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI2020), pp. 667–682. USENIX Association (2020)",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:23:30Z",
+    "is_model_written": true,
+    "summary": "This work tests the Solidity compiler with syntax-aware mutation, transforming smart-contract source in ways that keep it compilable so the mutants reach the compiler's later stages, where miscompilation bugs live.",
+    "narrative": "Data-oriented systems are cited once, in the introduction's list of domains fuzzing has been applied to, alongside system libraries, web and cloud applications, and compilers. SQLancer's work is the citation behind the data-oriented half. The paper's subject is compiler testing and nothing of SQLancer's plays a role in it.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1007_978_3_031_94706_3_7.json b/_data/papers/paper_doi_10_1007_978_3_031_94706_3_7.json
new file mode 100644
index 0000000..4b5881e
--- /dev/null
+++ b/_data/papers/paper_doi_10_1007_978_3_031_94706_3_7.json
@@ -0,0 +1,429 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:13:35Z",
+  "paper": {
+    "id": "paper:doi:10.1007/978-3-031-94706-3_7",
+    "title": "Fuzzing Graph Database Applications with Graph Transformations",
+    "authors": [
+      "Stefania Dumbrava",
+      "Melchior W. M. Oudemans",
+      "Burcu Kulahcioglu Ozkan"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Graph Transformation",
+    "doi": "10.1007/978-3-031-94706-3_7",
+    "arxiv_id": null,
+    "s2_paper_id": "1e316a65311c25101c0b04f1be8d90bc33e5b4b9",
+    "url": "https://doi.org/10.1007/978-3-031-94706-3_7",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1007/978-3-031-94706-3_7",
+      "retrieved_at": "2026-09-08T17:13:35Z",
+      "chars": 57966,
+      "content_sha256": "sha256:24d6256601804bf3110f4b397578fd5ea0a205b0453a0adceab8d600f26049ed"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 22,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1519
+      },
+      {
+        "number": "2",
+        "title": "Motivating Examples",
+        "start": 10795
+      },
+      {
+        "number": "3",
+        "title": "Preliminaries",
+        "start": 13042
+      },
+      {
+        "number": "4",
+        "title": "The PGFuzz Framework",
+        "start": 19199
+      },
+      {
+        "number": "4.1",
+        "title": "Producing Random Graph Instances",
+        "start": 21656
+      },
+      {
+        "number": "4.3",
+        "title": "Schema-Aware Graph Transformations",
+        "start": 24431
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 26951
+      },
+      {
+        "number": "5.1",
+        "title": "Benchmarks",
+        "start": 28971
+      },
+      {
+        "number": "5.2",
+        "title": "Evaluation Results",
+        "start": 32777
+      },
+      {
+        "number": "4.3",
+        "title": "To evaluate",
+        "start": 39542
+      },
+      {
+        "number": "6",
+        "title": "Related Work",
+        "start": 42366
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 46114
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "OpenStudyBuilder (2022). https:// gitlab. com/Novo- Nordisk/ nn-public/ openstudybuilder/ projectdescription",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Transport Network Graph Database (2022). https:// github. com/CathiaLH/ GraphDatabaseCombinedTransportNetwork",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "CiteGraph (2023). https:// github. com/citegraph/ citegraph",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "PanGraph-DB (2023). https:// github. com/jpjarnoux/ PanGraph- DB",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "PanTool (2023). https:// git.wur.nl/bioinformatics/ pantools/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Agrawal, P., Chandra, B., Emani, K.V., Garg, N., Sudarshan, S.: Test data generation for database applications. In: ICDE, pp. 1621–1624. IEEE Computer Society (2018)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Al-Saleem, J., et al.: Knowledge graph-based approaches to drug repurposing for COVID-19. J. Chem. Inf. Model. 61(8), 4058–4067 (2021)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "AllegroGraph: AllegroGraph. https:// allegrograph. com/. Visited 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Alvaro, P., Rigger, M.: Automatically testing database systems: DBMS testing with test oracles, transaction history, and fuzzing. ACM Queue 21(6), 128–135 (2024)",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 10,
+      "text": "Angles, R.: The property graph database model. In: AMW. CEUR Workshop Proceedings, vol. 2100. CEUR-WS.org (2018)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Angles, R., et al.: PG-schema: schemas for property graphs. Angles, R., et al.: PG-keys: keys for property graphs. In: SIGMOD Conference, pp. 2423–2436. ACM (2021) [13] ArangoDB: ArangoDB. https:// arangodb. com/. Visited 2024 [14] Arnoux, J., Bonifati, A., Calteau, A., Dumbrava, S., Gautreau, G.: Integrating complex pangenome graphs. In: ICDEW, pp. 350–354. IEEE (2024) [15] Atzeni, P., Bellomarin",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 9,
+      "text": "Alvaro, P., Rigger, M.: Automatically testing database systems: DBMS testing with test oracles, transaction history, and fuzzing. ACM Queue 21(6), 128–135 (2024)",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 11,
+      "text": "Angles, R., et al.: PG-schema: schemas for property graphs. Angles, R., et al.: PG-keys: keys for property graphs. In: SIGMOD Conference, pp. 2423–2436. ACM (2021) [13] ArangoDB: ArangoDB. https:// arangodb. com/. Visited 2024 [14] Arnoux, J., Bonifati, A., Calteau, A., Dumbrava, S., Gautreau, G.: Integrating complex pangenome graphs. In: ICDEW, pp. 350–354. IEEE (2024) [15] Atzeni, P., Bellomarini, L., Iezzi, M., Sallinger, E., Vlad, A.: Weaving enterprise knowledge graphs: the case of company ownership graphs. In: EDBT, pp. 555–566. OpenProceedings.org (2020) [16] AWS: Amazon Neptune. https:// aws.amazon. com/fr/neptune/. Visited 2024 [17] Ba, J., Rigger, M.: Testing database engines via query plan guidance. In: ICSE, pp. 2060–2071. IEEE (2023) [18] Bagan, G., Bonifati, A., Ciucanu, R., Fletcher, G.H., Lemay, A., Advokaat, N.: gMark: schema-driven generation of graphs and queries. IEEE Trans. Knowl. Data Eng. 29(4), 856–869 (2016) 154 S. Dumbrava et al. [19] Barcel´ o, P., P´ erez, J., Reutter, J.L.: Schema mappings and data exchange for graph databases. In: ICDT, pp. 189–200. ACM (2013) [20] Bitnine. Co., Ltd.: AgensGraph. https:// bitnine. net/agensgraph. Visited 2024 [21] BlazeGraph: BlazeGraph. https:// blazegraph. com/. Visited 2024 [22] Bohannon, P., Fan, W., Geerts, F., Jia, X., Kementsietsidis, A.: Conditional functional dependencies for data cleaning. In: ICDE, pp. 746–755. IEEE Computer Society (2007) [23] Bonifati, A., Furniss, P., Green, A., Harmer, R., Oshurko, E., Voigt, H.: Schema validation and evolution for graph databases. In: Laender, A., Pernici, B., Lim, E.-P., de Oliveira, J. (eds.) ER 2019. LNCS, vol. 11788, pp. 448–456. Springer, Cham (2019). https:// doi.org/10.1007/978-3-030-33223- 5 37 [24] Bonifati, A., Murlak, F., Ramusat, Y.: Transforming property graphs. Proc. VLDB Endow. 17(11), 2906–2918 (2024) [25] Bonifati, A., Ramusat, Y., Murlak, F., Fejza, A., Echahed, R.: DTGraph: declarative transformations of property graphs. Proc. VLDB Endow. 17(12), 4265–4268 (2024) [26] Clinical Data Interchange Standards Consortium: CDISC (2022). https:// www. cdisc.org/. Visited 2024 [27] DataStax: DataStax Enterprise Graph. https:// www.datastax. com/products/ datastaxgraph. Visited 2024 [28] DGraph: DGraph. https:// dgraph. io/. Visited 2024 [29] Elayam, M.M., Ray, C., Claramunt, C.: A hierarchical graph-based model for mobility data representation and analysis. Data Knowl. Eng. 141, 102054 (2022) [30] Emmi, M., Majumdar, R., Sen, K.: Dynamic test input generation for database applications. In: ISSTA, pp. 151–162. ACM (2007) [31] Fan, W.: Dependencies for graphs: challenges and opportunities. ACM J. Data Inf. Qual. 11(2), 5:1–5:12 (2019) [32] Fan, W., Geerts, F.: Foundations of Data Quality Management. Synthesis Lectures on Data Management. Morgan & Claypool Publishers (2012) [33] Francis, N., Libkin, L.: Schema mappings for data graphs. In: PODS, pp. 389–401. ACM (2017) [34] Godefroid, P., Kiezun, A., Levin, M.Y.: Grammar-based Whitebox fuzzing. In: PLDI, pp. 206–215. ACM (2008) [35] Godefroid, P., Levin, M.Y., Molnar, D.A.: Automated Whitebox fuzz testing. In: NDSS. The Internet Society (2008) [36] Gosnell, D., Broecheler, M.: The practitioner’s guide to graph data. https:// www. oreilly. com/library/ view/the-practitionersguide/ 9781492044062/. Visited 2024 [37] Green, H., Avgerinos, T.: GraphFuzz: library API fuzzing with lifetime-aware dataflow graphs, vol. 2022, pp. 1070–1081. IEEE Computer Society (2022). https:// doi.org/10.1145/3510003. 3510228 [38] G¨ utebier, L., et al.: CovidGraph: a graph to fight COVID-19. Bioinform. 38(20), 4843–4845 (2022) [39] Hasif, C.L., Araldo, A., Dumbrava, S., Watel, D.: A graph-database approach to assess the impact of demand-responsive services on public transit accessibility. In: IWCTS@SIGSPATIAL, pp. 2:1–2:4. ACM (2022) [40] Hegeman, T., Iosup, A.: Survey of graph analysis applications. CoRR abs/1807.00382 (2018) [41] Huang, H., Bucher, D., Kissling, J., Weibel, R., Raubal, M.: Multimodal route planning with public transport and carpooling. IEEE Trans. Intell. Transp. Syst. 20(9), 3513–3525 (2019) Fuzzing Graph Database Applications with Graph Transformations 155 [42] HyperGraphDB: HyperGraphDB. https:// hypergraphdb. org/. Visited 2024 [43] IBM: DB2 Graph. https:// www.ibm.com/docs/en/db2-warehouse? topic=applicationsdb2-graph. Visited 2024 [44] Ilyas, I.F., Chu, X.: Data Cleaning. ACM Books, vol. 28. ACM (2019) [45] Ba, J., Rigger, M.: Keep it simple: testing databases via differential query plans. JanusGraph: JanusGraph. https:// janusgraph. org/. Visited 2024 [47] Jiang, Y., Liu, J., Ba, J., Yap, R.H.C., Liang, Z., Rigger, M.: Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In: ICSE, pp. 46:1–46:12. ACM (2024) [48] Borges Jr., N.P., Havrikov, N., Zeller, A.: Generating tests that cover input structure. In: Software Engineering. LNI, vol. P-310, pp. 85–86. Gesellschaft f¨ ur Infor-matik e.V. (2021) [49] Kamm, M., Rigger, M., Zhang, C., Su, Z.: Testing graph database engines via query partitioning. In: ISSTA, pp. 140–149. ACM (2023) [50] Kertkeidkachorn, N., Nararatwong, R., Xu, Z., Ichise, R.: FinKG: a core financial knowledge graph for financial analysis. In: ICSC, pp. 90–93. IEEE (2023) [51] Klees, G., Ruef, A., Cooper, B., Wei, S., Hicks, M.: Evaluating fuzz testing. In: CCS, pp. 2123–2138. ACM (2018) [52] Li, J., Zhao, B., Zhang, C.: Fuzzing: a survey. Cybersecurity 1(1), 1–13 (2018). https:// doi.org/10.1186/s42400- 018-0002-y [53] MemGraph: MemGraph. https:// memgraph. com/. Visited 2024 [54] Metzman, J., Szekeres, L., Simon, L., Sprabery, R., Arya, A.: FuzzBench: an open fuzzer benchmarking platform and service. In: ESEC/SIGSOFT FSE, pp. 1393– 1403. ACM (2021) [55] Microsoft: Azure Cosmos DB. https:// azure. microsoft. com/fr-fr/products/ cosmosdb. Visited 2024 [56] Mughal, S., Moghul, I., Yu, J., Clark, T., Gregory, D.S., Pontikos, N.: Pheno4J: a gene to phenotype graph database. Bioinformatics 33(20), 3317–3319 (2017) [57] Neo4J: Neo4J. https:// neo4j. com/. Visited 2024 [58] Noy, N.F., Gao, Y., Jain, A., Narayanan, A., Patterson, A., Taylor, J.: Industryscale knowledge graphs: lessons and challenges. Commun. ACM 62(8), 36–43 (2019) [59] Olsthoorn, M., van Deursen, A., Panichella, A.: Generating highly-structured input data by combining search-based testing and grammar-based fuzzing. In: ASE, pp. 1224–1228. IEEE (2020) [60] Oracle: Oracle Big Data Spatial and Graph. https:// www.oracle. com/database/ technologies/ bigdataspatialandgraph. html. Visited 2024 [61] OrientDB: OrientDB. http:// orientdb. org/. Visited 2024 [62] Padhye, R., Lemieux, C., Sen, K., Papadakis, M., Traon, Y.L.: Semantic fuzzing with zest. In: ISSTA, pp. 329–340. ACM (2019) [63] Pan, K., Wu, X., Xie, T.: Automatic test generation for mutation testing on database applications. In: AST, pp. 111–117. IEEE Computer Society (2013) [64] Pan, K., Wu, X., Xie, T.: Guided test generation for database applications via synthesized database interactions. ACM Trans. Softw. Eng. Methodol. 23(2), 12:1– 12:27 (2014) [65] Park, S., Cheng, T.: Framework for constructing multimodal transport networks and routing using a graph database: a case study in London. Trans. GIS 27(5), 1391–1417 (2023) [66] Pham, V., B¨ ohme, M., Santosa, A.E., Caciulescu, A.R., Roychoudhury, A.: Smart greybox fuzzing. IEEE Trans. Software Eng. 47(9), 1980–1997 (2021) 156 S. Dumbrava et al. [67] Preusse, M., et al.: COVIDGraph: connecting biomedical COVID-19 resources and computational biology models. In: SEA-Data@VLDB. CEUR Workshop Proceedings, vol. 2929, pp. 34–37. CEUR-WS.org (2021) [68] Rabbani, K., Lissandrini, M., Bonifati, A., Hose, K.: Transforming RDF graphs to property graphs using standardized schemas. RedisGraph: RedisGraph. https:// redis.io/. Visited 2024 [70] Sahu, S., Mhedhbi, A., Salihoglu, S., Lin, J., ¨Ozsu, M.T.: The ubiquity of large graphs and surprising challenges of graph processing: extended survey. VLDB J. 29(2–3), 595–618 (2020) [71] Sakr, S., et al.: The future is big graphs: a community view on graph processing systems. Commun. ACM 64(9), 62–71 (2021) [72] Sequeda, J., Lassila, O.: Designing and building enterprise knowledge graphs. In: Synthesis Lectures on Data Semantics, and Knowledge. Morgan & Claypool Publishers (2021) [73] Redgate Software: Neo4J (2024). https:// db-engines. com/en/ranking/ graph+ dbms. Visited 2024 [74] Steinh¨ ofel, D., Zeller, A.: Input invariants. In: ESEC/SIGSOFT FSE, pp. 583–594. ACM (2022) [75] Thom Hurks: PGMark: a domain-independent tool for generating property graphs based on a user-defined schema. https:// github. com/ThomHurks/ pgMark. Visited 2024 [76] TigerGraph: TigerGraph. https:// www.tigergraph. com/. Visited 2024 [77] Titan: Titan. http:// espeed. github. io/titandb/. Visited 2024 [78] Wang, J., Chen, B., Wei, L., Liu, Y.: Superion: grammar-aware greybox fuzzing. In: ICSE, pp. 724–735. IEEE/ACM (2019) [79] Website, A.: American Fuzzy Loop. http:// lcamtuf. coredump. cx/afl/. Accessed 2024 [80] AFL Website: libFuzzer: A library for coverage-guided fuzz testing. http:// llvm. org/docs/LibFuzzer. html. Accessed 2024 [81] Yan, C., Nath, S., Lu, S.: Generating test databases for database-backed applications. In: ICSE, pp. 2048–2059. IEEE (2023) [82] Zeller, A., Gopinath, R., B¨ ohme, M., Fraser, G., Holler, C.: The fuzzing book (2019) [83] Zheng, Y., et al.: Differential optimization testing of Gremlin-based graph database systems. In: ICST, pp. 25–36. IEEE (2024) [84] Zhu, X., Wen, S., Camtepe, S., Xiang, Y.: Fuzzing: a survey for roadmap. ACM Comput. Surv. (CSUR) 54(11s), 1–36 (2022)",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 11]",
+      "technique": "qpg",
+      "sentence": "In this work, we target native GDBMSs, represented by the popular Neo4j database, and multimodel ones, represented by JanusGraph, one of the few GDBMSs equipped with not only graph schema, but also advanced integrity constraint mechanisms [ 11].",
+      "context_before": "eral models, with the main ones being, for example, wide-column stores c⃝The Author(s), under exclusive license to Springer Nature Switzerland AG 2025 J. Endrullis and M. Tichy (Eds.): ICGT 2025, LNCS 15720, pp. 135–156, 2025. https://doi.org/10.1007/978-3-031-94706-3 _7 136 S. Dumbrava et al. (JanusGraph [ 46], DataStax [ 27], Titan [ 77], etc.), key-value stores (HyperGraphDB [ 42], DGraph [ 28], RedisGraph [ 69], etc.), document stores (Azure Cosmos DB [ 55], ArangoDB [ 13], OrientDB [ 61], etc.) or relational tables (AgensGraph [ 20], Db2 Graph [ 43], Oracle Spatial and Graph [ 60], etc.).",
+      "context_after": "The behavior of graph database-backed applications depends on the contents of the input graph database. Therefore, exercising various executions of an application’s logic requires the use of different input graph instances, as they can activate different parts of the application logic. Unforeseen input database instances may result in unexpected and erroneous application behaviors. Ensuring the reliability of graph database applications through testing faces significant challenges in terms of validity and effectiveness. First, the database instances for the graph database applications must satisf",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 2900,
+      "cited_reference": {
+        "number": 11,
+        "text": "Angles, R., et al.: PG-schema: schemas for property graphs. Angles, R., et al.: PG-keys: keys for property graphs. In: SIGMOD Conference, pp. 2423–2436. ACM (2021) [13] ArangoDB: ArangoDB. https:// arangodb. com/. Visited 2024 [14] Arnoux, J., Bonifati, A., Calteau, A., Dumbrava, S., Gautreau, G.: I",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 11]",
+      "technique": "qpg",
+      "sentence": "These mutations, informed by the recent PG-Schema [ 11] language, contrast grammar-based fuzzers—which strictly enforce grammatical constraints— by intentionally generating graph instances that are syntactically valid, according to the schema types, but that violate key and cardinality constraints.",
+      "context_before": "d instances to uncover new application behaviors. To address scalability, PGFuzz operates on an in-memory graph model and simulates database calls through a lightweight API, avoiding the overhead of actual GDBMS operations and enabling fast, efficient fuzzing across large input spaces. While PGFuzz is the first greybox fuzzer targeting graph database-backed applications, its primary novelty is its use of schema-aware mutations, which produce input graphs that respect schema-defined structure and types while deliberately violating key and cardinality constraints through targeted property mutations.",
+      "context_after": "This allows PGFuzz to thoroughly test the application logic responsible for maintaining data integrity and schema compliance. Testing graph database-backed applications is an underexplored area, and there are no openly available sets of such benchmark applications. For the empirical evaluation, we examined open-source applications built on top of Neo4j [ 57], one of the most popular graph databases [ 73], and JanusGraph [ 46], one of the few graph databases to enforce schema constraints [ 11], and compiled a custom benchmark suite for testing database-backed applications. We evaluate PGFuzz a",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 8022,
+      "cited_reference": {
+        "number": 11,
+        "text": "Angles, R., et al.: PG-schema: schemas for property graphs. Angles, R., et al.: PG-keys: keys for property graphs. In: SIGMOD Conference, pp. 2423–2436. ACM (2021) [13] ArangoDB: ArangoDB. https:// arangodb. com/. Visited 2024 [14] Arnoux, J., Bonifati, A., Calteau, A., Dumbrava, S., Gautreau, G.: I",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 11]",
+      "technique": "qpg",
+      "sentence": "For the empirical evaluation, we examined open-source applications built on top of Neo4j [ 57], one of the most popular graph databases [ 73], and JanusGraph [ 46], one of the few graph databases to enforce schema constraints [ 11], and compiled a custom benchmark suite for testing database-backed applications.",
+      "context_before": "rough targeted property mutations. These mutations, informed by the recent PG-Schema [ 11] language, contrast grammar-based fuzzers—which strictly enforce grammatical constraints— by intentionally generating graph instances that are syntactically valid, according to the schema types, but that violate key and cardinality constraints. This allows PGFuzz to thoroughly test the application logic responsible for maintaining data integrity and schema compliance. Testing graph database-backed applications is an underexplored area, and there are no openly available sets of such benchmark applications.",
+      "context_after": "We evaluate PGFuzz against existing graph generators for test generation and show its superior cov-erage. Our tests revealed several bugs in the application programs that manifest when using certain graph database instances as input states. In summary, this paper makes the following contributions: – PGFuzz, the first greybox fuzzer for testing graph database-backed applica-tions that uses schema and constraint-aware graph transformations to generate input graph database instances. 138 S. Dumbrava et al. – A benchmark suite collected from real-world graph database-backed applications that lever",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 8588,
+      "cited_reference": {
+        "number": 11,
+        "text": "Angles, R., et al.: PG-schema: schemas for property graphs. Angles, R., et al.: PG-keys: keys for property graphs. In: SIGMOD Conference, pp. 2423–2436. ACM (2021) [13] ArangoDB: ArangoDB. https:// arangodb. com/. Visited 2024 [14] Arnoux, J., Bonifati, A., Calteau, A., Dumbrava, S., Gautreau, G.: I",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[ 11]",
+      "technique": "qpg",
+      "sentence": "We consider these systems, as Neo4j is one of the most popular [ 73]G D B M S s and JanusGraph is one of the few that can handle not only a priori schema constraints but also rich cardinality ones, as surveyed in [ 11].",
+      "context_before": "iple ones. The property graph model also allows attaching lists of key-value properties on nodes and edges, as can be seen in Fig. 2 discussed next. Property Graph Schemas and Constraints. While property graph instances represent data stored in a GDBMS, graph schemas and constraints describe their structure, typing, and integrity requirements to ensure consistency. Although these are crucial for data integration and exchange [ 19], query optimization [ 31], and data reliability [ 32], current GDBMSs provide limited support. The applications we analyzed are built on top of Neo4j and JanusGraph.",
+      "context_after": "Although these systems provide different levels of support for graph schemas and constraints-Neo4j is schemaless and JanusGraph is schema flexible and allows explicit or implicit schema definitionsboth benchmarks use Neo4j and JanusGraph enforce these at the application level to maintain data integrity. PGFuzz supports graph schema constraints, including key constraints and cardinality ones, illustrated with the recent PG-Schema language. Typing con-straints (PG-Types) specify nodes and edge labels and properties and the rela-tionship types that can connect certain node labels. Key constraints (",
+      "section": "3 Preliminaries",
+      "page": 6,
+      "char_offset": 15712,
+      "cited_reference": {
+        "number": 11,
+        "text": "Angles, R., et al.: PG-schema: schemas for property graphs. Angles, R., et al.: PG-keys: keys for property graphs. In: SIGMOD Conference, pp. 2423–2436. ACM (2021) [13] ArangoDB: ArangoDB. https:// arangodb. com/. Visited 2024 [14] Arnoux, J., Bonifati, A., Calteau, A., Dumbrava, S., Gautreau, G.: I",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[ 11]",
+      "technique": "qpg",
+      "sentence": "Among existing GDBMSs, JanusGraph provides one of the most comprehensive supports for cardinality constraints [ 11].",
+      "context_before": "ost one key, ensuring that two objects with the same key are identical, e.g., every route is operated by a unique agency (lines 32–33). Cardinality constraints, introduced in PG-Schema as a PG-Keys extension, enforce bounds on the number of graph object instances, e.g., every centroid should be connected to a stoptime through at least two edges, labeled ‘DRT’ or ‘WALK’ (lines 35–36). The constraints are expressed similarly to PG-Keys, using the qualifier COUNT <lower bound>?..<upper bound>? OF to express that the number of distinct results returned by q(x, ¯y) must be within the specified range.",
+      "context_after": "It allows declaring edge label multiplicity, specifying whether at most one (SIMPLE) or multiple (MULTI) edges can be defined between any node pair, with MANY2ONE and ONE2MANY respectively allowing at most one outgoing/incoming edge, without constraining the number of incoming/outgoing ones. It also allows for property key cardinalities, specifying whether a node key can have one or multiple values. PGFuzz transformations allow breaking all of the above constraints, as detailed in Sect. 4.3. PGFuzz considers test inputs for graph database applications to be defined by a core fragment of the pre",
+      "section": "3 Preliminaries",
+      "page": 7,
+      "char_offset": 17935,
+      "cited_reference": {
+        "number": 11,
+        "text": "Angles, R., et al.: PG-schema: schemas for property graphs. Angles, R., et al.: PG-keys: keys for property graphs. In: SIGMOD Conference, pp. 2423–2436. ACM (2021) [13] ArangoDB: ArangoDB. https:// arangodb. com/. Visited 2024 [14] Arnoux, J., Bonifati, A., Calteau, A., Dumbrava, S., Gautreau, G.: I",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[ 11,12]",
+      "technique": "qpg",
+      "sentence": "PGFuzz uses graph schema-aware transformations and supports a core fragment of the recent PG-Schema and PG-Keys formalisms, which inform the design of the novel GQL graph query standard [ 11,12].",
+      "context_before": "lored parts of the application program. Alternative to code coverage, the framework can be extended to use different feedback information from the program, e.g., defining fitness functions, to evaluate how likely the test inputs generated from an input can trigger new behaviors. 4.3 Schema-Aware Graph Transformations Given a test input, i.e., test graph database instance, that achieves new coverage PGFuzz mutates that instance to produce a new instance using novel graph 144 S. Dumbrava et al. transformations that leverage the typing, key, and cardinality constraints of the application under test.",
+      "context_after": "PGFuzz integrates information regarding graph typing, key constraints enforced on graph objects, and cardinality restrictions, e.g., on the number of edges between node pairs, as in JanusGraph: One2One, Many2One, One2Many, Many2Many. This provides variations of graph instances that comply with application requirements and are likely to cover important states. Rather than adding a random label, the transformation can select one that is expected to occur. By design, the transformation can leverage information about graph constraints that might not otherwise be derived from any arbitrary graph s",
+      "section": "4.3 Schema-Aware Graph Transformations",
+      "page": 10,
+      "char_offset": 24754,
+      "cited_reference": {
+        "number": 11,
+        "text": "Angles, R., et al.: PG-schema: schemas for property graphs. Angles, R., et al.: PG-keys: keys for property graphs. In: SIGMOD Conference, pp. 2423–2436. ACM (2021) [13] ArangoDB: ArangoDB. https:// arangodb. com/. Visited 2024 [14] Arnoux, J., Bonifati, A., Calteau, A., Dumbrava, S., Gautreau, G.: I",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[\n9,17,45]",
+      "technique": null,
+      "sentence": "Several recent works focus on testing database man-agement systems [ 9,17,45] and graph databases [ 47,49,83].",
+      "context_before": "al results involve multiple sources of randomness. The input graph instances are randomly generated by gMark and pgMark, which Fuzzing Graph Database Applications with Graph Transformations 151 Fig. 8. Test coverage of the benchmarks results using compound transformations. are later transformed by PGFuzz. Moreover, the effectiveness of PGFuzz’s tests is highly dependent on the applied transformations, which are also selected randomly as part of the fuzzing framework. To mitigate randomness, we repeat our experiments five times and report averaged results. 6 Related Work Testing Database Systems.",
+      "context_after": "Different from these works, which target the bugs in the database systems, PGFuzz targets finding bugs in the graph database-backed applications. Testing Database-Backed Applications. As the behavior of database-backed applications depends on the state of the database on which they operate, test-ing them requires generating test database instances. Earlier works [ 6,30]u s e constraint solving and static analysis for generating program inputs and input database states. SynDB [ 63,64] tests database-backed applications using a symbolic database. It generates test inputs and relational database s",
+      "section": "6 Related Work",
+      "page": 17,
+      "char_offset": 42406,
+      "cited_reference": {
+        "number": 9,
+        "text": "Alvaro, P., Rigger, M.: Automatically testing database systems: DBMS testing with test oracles, transaction history, and fuzzing. ACM Queue 21(6), 128–135 (2024)",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:23:30Z",
+    "is_model_written": true,
+    "summary": "PGFuzz fuzzes applications built on graph databases by mutating the graph itself. Its mutations are informed by the PG-Schema and PG-Keys formalisms, and deliberately produce instances that are type-valid but violate key and cardinality constraints -- the opposite of what a grammar-based fuzzer does. It is evaluated on open-source applications over Neo4j and JanusGraph.",
+    "narrative": "SQLancer's line of work is cited once, in related work, among the recent efforts to test database management systems and graph databases. The distinction the paper draws throughout is that those tools test the engine, while PGFuzz targets applications built on top of one, and its mutations attack integrity constraints rather than query semantics. Every mention here is reachable only through a citation marker.",
+    "roles": {
+      "M1": "background",
+      "M2": "motivation",
+      "M3": "background",
+      "M4": "background",
+      "M5": "background",
+      "M6": "definition",
+      "M7": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1007_978_981_95_3182_0_3.json b/_data/papers/paper_doi_10_1007_978_981_95_3182_0_3.json
new file mode 100644
index 0000000..7bc1484
--- /dev/null
+++ b/_data/papers/paper_doi_10_1007_978_981_95_3182_0_3.json
@@ -0,0 +1,484 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:13:36Z",
+  "paper": {
+    "id": "paper:doi:10.1007/978-981-95-3182-0_3",
+    "title": "Ghosts in DBMS: Revealing the Security Impacts of Silent Fixes",
+    "authors": [
+      "Jialiang Dong",
+      "Zihan Ni",
+      "Willy Susilo",
+      "Siqi Ma"
+    ],
+    "year": 2025,
+    "venue": "Lecture notes in computer science",
+    "doi": "10.1007/978-981-95-3182-0_3",
+    "arxiv_id": null,
+    "s2_paper_id": null,
+    "url": "https://doi.org/10.1007/978-981-95-3182-0_3",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1007/978-981-95-3182-0_3",
+      "retrieved_at": "2026-09-08T17:13:36Z",
+      "chars": 49438,
+      "content_sha256": "sha256:92bf1da6bb84bc31a9d6e3fdf4ea9692827eafe2d99dc039cacfcf0dd81284f6"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 19,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1666
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 7806
+      },
+      {
+        "number": "2.1",
+        "title": "Fork-Based Inheritance in DBMS Systems",
+        "start": 7819
+      },
+      {
+        "number": "2.2",
+        "title": "Silent Fix",
+        "start": 10570
+      },
+      {
+        "number": "3",
+        "title": "Methodology",
+        "start": 12556
+      },
+      {
+        "number": "3.1",
+        "title": "Inheritance Analysis of Forked DBMS",
+        "start": 13410
+      },
+      {
+        "number": "3.2",
+        "title": "Disclosed Vulnerability Fix Mining",
+        "start": 16003
+      },
+      {
+        "number": "3.3",
+        "title": "Generic Vulnerability Fix Mining",
+        "start": 17354
+      },
+      {
+        "number": "3.4",
+        "title": "Silent Vulnerability Determination",
+        "start": 22559
+      },
+      {
+        "number": "4",
+        "title": "Experiments",
+        "start": 24309
+      },
+      {
+        "number": "4.1",
+        "title": "Experimental Settings",
+        "start": 24323
+      },
+      {
+        "number": "4.2",
+        "title": "RQ1: Inheritance Relations Among Forked DBMSs",
+        "start": 29358
+      },
+      {
+        "number": "4.3",
+        "title": "RQ2: Silent Fix Identification",
+        "start": 32438
+      },
+      {
+        "number": "4.4",
+        "title": "RQ3: Silent Vulnerability Propagation",
+        "start": 35144
+      },
+      {
+        "number": "5",
+        "title": "Discussion",
+        "start": 37201
+      },
+      {
+        "number": "6",
+        "title": "Related Works",
+        "start": 38221
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 40168
+      },
+      {
+        "number": "11.4.4",
+        "title": "All of the latest five release versions are inherited from MySQL-5.7.42",
+        "start": 41267
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Mysql server, 2025. https:// github. com/mysql/ mysqlserver",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Postgresql: The world’s most advanced open source relational database, 2025. https:// www.postgresql. org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Tidb, 2025. https:// github. com/pingcap/ tidb",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Myrocks: A rocksdb storage engine with mysql, 2025. https:// myrocks. io/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Sqlite home page, 2025. https:// sqlite. org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Mariadb server: the innovative open source database, 2025. https:// mariadb. org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "What is amazon aurora?, 2025. https:// docs.aws.amazon. com/AmazonRDS/ latest/ AuroraUserGuide/ CHAP AuroraOverview. html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Polardbx-engine, 2025. https:// github. com/polardb/ polardbxengine",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Zhou, J., et al.: Finding a needle in a haystack: automated mining of silent vulnerability fixes. In: IEEE/ACM 36th International Conference on Automated Software Engineering (ASE), pp. 705–716. IEEE, 2021",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "The problems with vulnerability reporting, 2023. https:// readme. synack. com/theproblemswith-vulnerabilityreporting",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "The problems with vulnerability reporting, 2021. https:// www.hackerone. com/ blog/vulnerabilitydisclosurewhatsresponsiblesolution",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Responsible vulnerability disclosure: Why it matters, 2021. https:// outpost24. com/blog/responsiblevulnerabilitydisclosure/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "How mysql servers can attack you, 2023. https:// conference. hitb.org/ hitbsecconf2023ams/ materials/ D2T1%20-%20How %20MySQL %20Servers %20Can %20Attack %20YOU %20-%20Martin %20Rahkmanov %20& %20Alexander %20Rubin. pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Cve-2023-21980, 2023. https:// nvd.nist.gov/vuln/detail/ CVE- 2023-21980",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Sun, J., et al.: Silent vulnerable dependency alert prediction with vulnerability key aspect explanation. In: IEEE/ACM 45th International Conference on Software Engineering (ICSE), pp. 970–982. IEEE, 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Zhou, J., et al.: Colefunda: explainable silent vulnerability fix identification. In: pp. 2565–2577. IEEE, 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Wen, Z., et al.: Silent taint-style vulnerability fixes identification. In: Proceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), pp. 428–439, 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Dunlap, T., Thorn, S., Enck, W., Reaves, B.: Finding fixed vulnerabilities with off-the-shelf static analysis. In: IEEE 8th European Symposium on Security and Privacy (EuroS&P), pp. 489–505. IEEE, 2023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Percona software for mysql, 2025. https:// www.percona. com/mysql/ software",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Cubukcu, U., Erdogan, O., Pathak, S., Sannakkayala, S., Slot, M.: Citus: distributed postgresql for data-intensive applications. In: Proceedings of the 40th International Conference on Management of Data (SIGMOD), pp. 2490–2502, 2021",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Sqlcipher, 2025. https:// github. com/sqlcipher/ sqlcipher",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Bug#33379702 ndb: Query thread tux relink logic error, 2023. https:// github. com/mysql/ mysqlserver/ commit/ d37cb5b",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Woo, S., Park, S., Kim, S., Lee, H., Oh, H.: Centris: a precise and scalable approach for identifying modified open-source software reuse. In: IEEE/ACM 43rd Interna-tional Conference on Software Engineering (ICSE), pp. 860–872. IEEE, 2021",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Universal ctags, 2025. https:// github. com/universalctags/ ctags",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "National vulnerability database, 2025. https:// nvd.nist.gov/ 56 J. Dong et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Official common platform enumeration (cpe) dictionary, 2025. https:// nvd.nist. gov/products/ cpe",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Xu, S., et al.: Enhancing security in third-party library reusecomprehensive detection of 1-day vulnerability through code patch analysis. In: ISOC 32th Network and Distributed System Security Symposium (NDSS), 2025",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Feng, Z., et al.: Codebert: a pre-trained model for programming and natural languages. In: Findings of the 25th ACL Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 1536–1547, 2020",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Linux kernel, 2025. https:// github. com/torvalds/ linux",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Imagemagick, 2025. https:// github. com/ImageMagick/ ImageMagick",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Welcome to the openssl projectl, 2025. https:// github. com/openssl/ openssl",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Wordnet: A lexical database for english, 2025. https:// wordnet. princeton. edu/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Faiss documentation, 2025. https:// faiss.ai/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Gitpython documentation, 2025. https:// gitpython. readthedocs. io/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Wang, X., Sun, K., Batcheller, A., Jajodia, S.: Detecting “0-day” vulnerability: an empirical study of secret security patch in oss. In: 49th Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN), pp. 485– [492] IEEE, 2019",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Han, M., Wang, L., Chang, J., Li, B., Zhang, C.: Learning graph-based patch representations for identifying and assessing silent vulnerability fixes. In: IEEE 35th International Symposium on Software Reliability Engineering (ISSRE), pp. 120–131. IEEE, 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Cheng, Y., et al.: Fixseeker: an empirical driven graph-based approach for detecting silent vulnerability fixes in open source software. arXiv preprint arXiv:2503.20265, 2025",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. Proc. 35th ACM Program. Lang. (OOPSLA) 4, 1–30 (2020)",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "Liang, Y., Liu, S., Hu, H.: Detecting logical bugs of dbms with coverage-based guidance. In: 31st USENIX Security Symposium (USENIX Security), pp. 4309– 4326, 2022",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Jiang, Y., Liu, J., Ba, J., Yap, R.H., Liang, Z., Rigger, M.: Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In: Proceedings of the 46th IEEE/ACM International Conference on Software Engineering (ICSE), pp. 1–12, 2024",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "Mang, Q., Fang, A., Yu, B., Chen, H., He, P.: Testing graph database systems via equivalent query rewriting. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (ICSE), pp. 1–12, 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Cui, Z., et al.: Understanding transaction bugs in database systems. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (ICSE), pp. 1–13, 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Fu, J., Liang, J., Wu, Z., Zhao, Y., Li, S., Jiang, Y.: Understanding and detecting sql function bugs: using simple boundary arguments to trigger hundreds of dbms bugs. In: Proceedings of the 20th European Conference on Computer Systems (EuroSys), pp. 1061–1076, 2025",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "Liang, J., Wu, Z., Fu, J., Bai, Y., Zhang, Q., Jiang, Y.: Wingfuzz: implementing continuous fuzzing for dbmss. In: 43rd USENIX Annual Technical Conference (USENIX ATC), pp. 479–492, 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Dong, J., Chen, X., Susilo, W., Sun, N., Shaghaghi, A., Ma, S.: What lies beneath: an empirical study of silent vulnerability fixes in open-source software. In: 55th Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN), pp. 485–492. IEEE, 2025",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 38,
+      "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. Proc. 35th ACM Program. Lang. (OOPSLA) 4, 1–30 (2020)",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "Jiang, Y., Liu, J., Ba, J., Yap, R.H., Liang, Z., Rigger, M.: Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In: Proceedings of the 46th IEEE/ACM International Conference on Software Engineering (ICSE), pp. 1–12, 2024",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 43,
+      "text": "Fu, J., Liang, J., Wu, Z., Zhao, Y., Li, S., Jiang, Y.: Understanding and detecting sql function bugs: using simple boundary arguments to trigger hundreds of dbms bugs. In: Proceedings of the 20th European Conference on Computer Systems (EuroSys), pp. 1061–1076, 2025",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "Prior efforts have addressed such issues using query partitioning [ 38], validity-guided fuzzing [ 39], graph-based query transformations [ 40], and syntax abstraction frameworks [ 41].",
+      "context_before": "ed. We will explore quantitative severity evaluation. Ethical Considerations. Although the study involves analysis of undisclosed silent vulnerabilities, all examples shown are limited to publicly disclosed and fixed cases. To prevent misuse, we omit details of downstream DBMS versions affected by unfixed vulnerabilities, in line with responsible disclosure practices. 6 Related Works DBMS Security. DBMS plays a critical role in modern software and sys-tems. Among various vulnerabilities, logical flaws are severe and characteristic in DBMSs due to complex query semantics and optimization behaviors.",
+      "context_after": "Besides, DBMSs suffer from domain-specific vulnerabilities such as transaction anomalies [ 42], where real-world cases compromise consistency guarantees, and flaws in built-in SQL functions [ 43]. Moreover, DBMSs are also affected by generic low-level memory errorse.g., buffer overflows [ 44]. Despite growing attention to DBMS security, supply chain threats remain underexplored, particularly silent vulnerabilities. This paper presents the first systematic analysis of silent vulnerability propagation in forked DBMSs. Silent Fix Analysis. In open-source software, silent fixes can unintentionally expose",
+      "section": "6 Related Works",
+      "page": 16,
+      "char_offset": 38453,
+      "found_by_all": [
+        "technique",
+        "citation_marker",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 43]",
+      "technique": null,
+      "sentence": "Besides, DBMSs suffer from domain-specific vulnerabilities such as transaction anomalies [ 42], where real-world cases compromise consistency guarantees, and flaws in built-in SQL functions [ 43].",
+      "context_before": "to publicly disclosed and fixed cases. To prevent misuse, we omit details of downstream DBMS versions affected by unfixed vulnerabilities, in line with responsible disclosure practices. 6 Related Works DBMS Security. DBMS plays a critical role in modern software and sys-tems. Among various vulnerabilities, logical flaws are severe and characteristic in DBMSs due to complex query semantics and optimization behaviors. Prior efforts have addressed such issues using query partitioning [ 38], validity-guided fuzzing [ 39], graph-based query transformations [ 40], and syntax abstraction frameworks [ 41].",
+      "context_after": "Moreover, DBMSs are also affected by generic low-level memory errorse.g., buffer overflows [ 44]. Despite growing attention to DBMS security, supply chain threats remain underexplored, particularly silent vulnerabilities. This paper presents the first systematic analysis of silent vulnerability propagation in forked DBMSs. Silent Fix Analysis. In open-source software, silent fixes can unintentionally expose undisclosed flaws, creating an information asymmetry. Wang et al. [ 35] treated silent vulnerabilities as a form of “0-day” and used machine learning to detect them in OpenSSL. Zhou et al. [ 9]",
+      "section": "6 Related Works",
+      "page": 16,
+      "char_offset": 38638,
+      "cited_reference": {
+        "number": 43,
+        "text": "Fu, J., Liang, J., Wu, Z., Zhao, Y., Li, S., Jiang, Y.: Understanding and detecting sql function bugs: using simple boundary arguments to trigger hundreds of dbms bugs. In: Proceedings of the 20th European Conference on Computer Systems (EuroSys), pp. 1061–1076, 2025",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:02:36Z",
+    "is_model_written": true,
+    "summary": "This paper studies silent fixes in DBMSs -- security-relevant bugs repaired without an advisory or a CVE -- and what their quiet handling means for downstream users, who cannot tell that upgrading matters.",
+    "narrative": "Query partitioning is named in related work as one of the techniques prior efforts have used to find these issues, listed with validity-guided fuzzing, graph-based query transformations and syntax abstraction frameworks. That single sentence is the whole connection; the paper's own method is studying fix commits.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1007_978_981_95_4721_0_7.json b/_data/papers/paper_doi_10_1007_978_981_95_4721_0_7.json
new file mode 100644
index 0000000..21922c4
--- /dev/null
+++ b/_data/papers/paper_doi_10_1007_978_981_95_4721_0_7.json
@@ -0,0 +1,454 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:13:22Z",
+  "paper": {
+    "id": "paper:doi:10.1007/978-981-95-4721-0_7",
+    "title": "CypherFuzzer: A Tool for Testing Access Control in Graph Databases",
+    "authors": [
+      "Philipp Reisinger",
+      "Daniel Hofer",
+      "Bahara Muradi",
+      "Josef Küng"
+    ],
+    "year": 2025,
+    "venue": "Communications in computer and information science",
+    "doi": "10.1007/978-981-95-4721-0_7",
+    "arxiv_id": null,
+    "s2_paper_id": null,
+    "url": "https://doi.org/10.1007/978-981-95-4721-0_7",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1007/978-981-95-4721-0_7",
+      "retrieved_at": "2026-09-08T17:13:22Z",
+      "chars": 46508,
+      "content_sha256": "sha256:65af4fca9f88730468af3cf58cd7e32b2dbda15c7bb469ea77aa119f980c29bd"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 17,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1707
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 5668
+      },
+      {
+        "number": "2.1",
+        "title": "Access Control in Graph Databases",
+        "start": 5926
+      },
+      {
+        "number": "2.2",
+        "title": "Fuzzing Cypher Queries",
+        "start": 8697
+      },
+      {
+        "number": "3",
+        "title": "Related Work",
+        "start": 14368
+      },
+      {
+        "number": "4",
+        "title": "CypherFuzzer",
+        "start": 19934
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 29681
+      },
+      {
+        "number": "5.1",
+        "title": "Fuzzing Process",
+        "start": 29694
+      },
+      {
+        "number": "5.2",
+        "title": "Analysis of Results",
+        "start": 30271
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 35735
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 37756
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Bertolino, A., Daoudagh, S., Lonetti, F., Marchetti, E., Schilders, L.: Automated testing of extensible access control markup language-based access control systems. IET Softw.7(4), 203–212 (2013). https:// doi.org/10.1049/iet-sen.2012.0101. https:// ietresearch. onlinelibrary. wiley.com/doi/abs/10.1049/iet-sen.2012.0101 108 P. Reisinger et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Chen, T.Y., et al.: Metamorphic testing: a review of challenges and o pportunities. ACM Comput. Surv.51(1) (2018). https:// doi.org/10.1145/3143561,",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Daoudagh, S., Lonetti, F., Marchetti, E.: Xacmet: Xacml testing & mo deling. Softw. Qual. J.28, 249–282 (2020). https:// doi.org/10.1007/s11219- 019-094705",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Evans, R.B., Savoia, A.: Differential testing: a new approach to change detection. In: The 6th Joint Meeting on European Software Engineering Conference and the ACMSIGSOFT Symposium on the Foundations of S oftware Engineering: Companion Papers, ESEC-FSE companion ’07, pp. 549–552. Association for Computing Machinery, New York (2007). https:// doi.org/10.1145/1295014. 1295038",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Forsberg, A., Lepik, A.: Random Generation of Semantically Valid Cypher Q ueries (2023). Student Paper",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Hofer, D., Mohamed, A., Auer, D., Nadschl¨ ager, S., K¨ ung, J.: Rewriting graphdb queries to enforce attribute-based access control. In: Strauss, C., Amagasa, T., K otsis, G., Tjoa, A.M., Khalil, I. (eds.) Database and Expert Systems Appli-cations, pp. 431–436. Springer, Cham (2023). https:// doi.org/10.1007/978-3-03139847- 6 34",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Hofer, D., Mohamed, A., Nadschl¨ ager, S., Auer, D.: An intermediate representation for rewriting cypher queries. In: Kotsis, G., et al. (eds.) Database and Expert Systems Applications-DEXA 2023 Workshops-34th In ternational Conference,DEXA 2023, Proceedings. Communications in Computer and Information Science, vol. 1872, pp. 86–90. Springer, Heidelberg (2023). https:// doi.org/10.1007/978-3031-396",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Hua, Z., et al.: Gdsmith: detecting bugs in cypher graph database engines. In: Proceedings of the 32nd ACMSIGSOFT International Symposium on Software T est-ing and Analysis, ISSTA 2023, pp. 163–174. Association for Computing Machinery, New York (2023). https:// doi.org/10.1145/3597926. 3598046",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Hwang, J., Martin, E., Xie, T., Hu, V.: Testing access control policies. In: E ncyclopedia of Software Engineering (2010). https:// taoxie. cs.illinois. edu/publications/ policytest. pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Jiang, Y., Liu, J., Ba, J., Yap, R.H.C., Liang, Z., Rigger, M.: Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’24. Association for Computing Machinery, New York (2024). https:// doi.org/10.1145/3597503. 3623307",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 11,
+      "text": "Jiang, Z.M., Bai, J.J., Su, Z.: DynSQL: stateful fuzzing for database management systems with complex and valid SQL query generation. In: 32nd USENIX Security Symposium (USENIX Security 23), pp. 4949–4965. USENIX Association, Anaheim (2023). https:// www.usenix. org/conference/ usenixsecurity23/ presentation/ jiangzu-ming",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jung, J., Hu, H., Arulraj, J., Kim, T., Kang, W.: Apollo: automatic detection and diagnosis of performance r egressions in database systems. Proc. VLDB Endow. 13(1), 57–70 (2019). https:// doi.org/10.14778/ 3357377. 3357382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Kamm, M., Rigger, M., Zhang, C., Su, Z.: Testing graph database engines via query partitioning. In: Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, pp. 140–149. Association for Com-puting Machinery, New York (2023). https:// doi.org/10.1145/3597926. 3598044",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "Liu, S., et al.: Testing graph database systems with graph-state persistence oracle. In: Proceedings of the 33rd ACMSIGSOFT International Symposium on Software T esting and Analysis, ISSTA 2024, pp. 666–677. Association for Computing Machinery, New York (2024). https:// doi.org/10.1145/3650212. 3680311 CypherFuzzer: A Tool for Testing Access Control in Graph Databases 109",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Mang, Q., Fang, A., Yu, B., Chen, H., He, P.: Testing graph database systems via equivalent query rewriting. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’24. Association for Computing Machinery, New York (2024). https:// doi.org/10.1145/3597503. 3639200",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Martin, E., Xie, T.: Automated test generation for access control policies via change-impact analysis. In: Third International Workshop on Soft ware Engineering for Secure Systems (SESS’07: ICSE Workshops 2007), p. 5 (2007). https:// doi. org/10.1109/SESS. 2007.5",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Miller, B.P., Fredriksen, L., So, B.: An empirical study of the reliability of unix utilities.Commun. ACM33(12), 32–44 (1990). https:// doi.org/10.1145/96267. 96279",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Mohamed, A., Auer, D., Hofer, D., K¨ ung, J.: Xacml extension for graphs: flexible authorization policy specification and datastore-independent enforcement. In: Proceedings of the 20th Int ernational Conference on Security and Cryptography, pp. 442–449. SCITEPRESS-Science and Technology Publications (2023). https:// doi. org/10.5220/0012090000003555",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Mohamed, A.K.Y.S., Auer, D., Hofer, D., K¨ ung, J.: A systematic literature review for authorization and access control: definitions, strategies and models. Int. J. Web Inf. Syst.18(2/3), 156–180 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "de Peralta Gonzalez, R.G.: Generating cypher sub queries in graph databases (2024). https:// www.researchcollection. ethz.ch/handle/ 20.500.11850/ 681667. https:// doi.org/10.3929/ethz-b-000681667",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. Pro c. ACM Program. Lang.4(OOPSLA) (2020). https:// doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "Rizvi, S.Z.R., Fong, P.W.L.: Efficient authorization of graph database queries in an attribute-supp orting rebac model. In: Zhao, Z., Ahn, G.J., Krishnan, R., Ghinita, G. (eds.) CODASPY’18, pp. 204–211. The Association for Computing Machinery,New York (2018). https:// doi.org/10.1145/3176258. 3176331",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Seltenreich, A., Tang, B., Mullender, S.: Random sql query generator for squashing bugs(2020). https:// github. com/anse1/ sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Wei, J., Chen, P., Lu, K., Dai, J., Sun, X.: Sqlaser: detecting dbms logic bugs with clause-guided fuzzing (2024). arXiv:abs/2407.04294",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Wu, J., Wu, Z., Li, R., Qin, H., Wang, G.: Effective bug detection in graph database engines: an llm-based approach (2024). arXiv:abs/2402.00292",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Yuan, Y., Lu, Y., Zhu, K., Huang, H., Chen, Y., Zhang, Y.: sqlfuzz: directed fuzzing forsql injection vulnerability. Electronics13(15) (2024). https:// doi.org/10.3390/ electronics13152946. https:// www.mdpi.com/2079-9292/13/15/2946",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Zheng, Y., et al.: Differential optimization testing of gremlin-based graph database systems. In: (ICST), pp. 25–36 (2024). https:// doi.org/10.1109/ICST60714. 2024.00012",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Zhong, R., Chen, Y., Hu, H., Zhang, H., Lee, W., Wu, D.: Squirrel: testing database managemen t systems with language validity and coverage feedback. In: Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security,CCS ’20, pp. 955–970. Association for Computing Machinery, New York (2020). https:// doi.org/10.1145/3372297. 3417260",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Zhuang, Z., Li, P., Ma, P., Meng, W., Wang, S.: Testing graph database systems via graph-a ware metamorphic relations. Proc. VLDB Endow.17(4), 836–848 (2023). https:// doi.org/10.14778/ 3636218. 3636236",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 10,
+      "text": "Jiang, Y., Liu, J., Ba, J., Yap, R.H.C., Liang, Z., Rigger, M.: Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’24. Association for Computing Machinery, New York (2024). https:// doi.org/10.1145/3597503. 3623307",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 13,
+      "text": "Kamm, M., Rigger, M., Zhang, C., Su, Z.: Testing graph database engines via query partitioning. In: Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, pp. 140–149. Association for Com-puting Machinery, New York (2023). https:// doi.org/10.1145/3597926. 3598044",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 21,
+      "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. Pro c. ACM Program. Lang.4(OOPSLA) (2020). https:// doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 5, 8,10,14,15,20,25]",
+      "technique": null,
+      "sentence": "Previous work with regard to fuzzing primarily focused on testing the graph database engine itself [ 5, 8,10,14,15,20,25].",
+      "context_before": "a simple yet effective approach to test the robustness, security and reliability of a system under test (SUT). In the specific case of access control models for graph databases, the key idea behind fuzzing has changed little. A fuzzer, in this setting, is a piece of software that automatically generates and executes a variety of complex, arbitrary, but semantically validqueries to uncover vulnerabilities or bugs in the implementation. The generatedqueries may attempt to bypass access control rules, e.g., by accessing restricted nodes/edges or by exploiting edge cases in the access control logic.",
+      "context_after": "Our research is mainly motivated by the works conducted by Hofer et al. [ 6, 7]. Their approach to rewriting graph database queries to enforce attribute-based access control (ABAC) provides a good foundation for enforcing sophisticated, fine-grained access control to graphstructured data at the query level. Since their method directly impact the secu-rity logic of the graph database system, thorough and automated testing of the correctness and robustness of their implementation is essential. Here, fuzzing stands out as a highly suitable technique to generate random inputs for evalu-ating the i",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3435,
+      "cited_reference": {
+        "number": 10,
+        "text": "Jiang, Y., Liu, J., Ba, J., Yap, R.H.C., Liang, Z., Rigger, M.: Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’24. Association for Computin",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 10]",
+      "technique": null,
+      "sentence": "– Grave de Peralta Gonzalez [ 20] extended GraphGenie [ 10] with a tool for testing Cypher subqueries.",
+      "context_before": "rect query results and 8 bugscaused crashes. – GraspDB [ 14] extends GDSmith to test bugs specifically involving write operations; something not supported by GDSmith’s grammar. It introduces three new mutation rules: (1) Add Writing Clause to insert CREATE or MERGE statements and test for graph-modifying bugs, (2) Modify Writing Clause to alter write patterns or insert clauses like WITH that may affect query planning, and (3) Modify Return Clause to influence query output, for instance, using LIMIT. Correctness is checked via metamorphic testing. GraspDB found 77 bugs, 31 tied to writing queries.",
+      "context_after": "Their tool generates subqueries (EXISTS, COUNT, COLLECT, o r CALL (in transactions)) by injecting them into randomly selected paths of an existing graph. It combines grammar-based generation with ASTs; the grammar primarily defines query structure and AST fills the skele-ton. A “Schema Scanner” extracts metadata from a live graphs to ensure schema-aware queries and more satisfiable conditions. Graph traversal diver-sifies queries with adjustable path depth for varying complexity. As a first CypherFuzzer: A Tool for Testing Access Control in Graph Databases 97 step toward testing subquery behavior,",
+      "section": "2.2 Fuzzing Cypher Queries",
+      "page": 4,
+      "char_offset": 11491,
+      "cited_reference": {
+        "number": 10,
+        "text": "Jiang, Y., Liu, J., Ba, J., Yap, R.H.C., Liang, Z., Rigger, M.: Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’24. Association for Computin",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 21]",
+      "technique": "tlp",
+      "sentence": ", SQL-Smith [ 23], Squirrel [ 28]), logic bugs (through test oracles [ 21]), and performance issues (Apollo [ 12]) in database engines.",
+      "context_before": "]), a specialized tool is required to generate queries that test the correctness and robustness of the rewriting process. This approach needs to verify that rewritten queries enforce the intended access control policies without introducing logical errors or performance issues. 3 Related Work Relational database management systems (RDBMS) have been well researched over the years, with significant research dedicated to both query generation and test oracle construction. In this work, we focus mainly on query generation. Most existing efforts in this area focus on the detection of memory bugs (e.g.",
+      "context_after": "Compared to RDBMS testing, testing graph database management systems (GDBMS) is relatively less researched, mainly because GDBMS emerged more recently. GDBMS differ from RDBMS 98 P. Reisinger et al. in storage structure, data models, query languages, etc., which implies that the testing approaches designed for RDBMS cannot be directly applied to GDBMS. Proprietary languages within GDMS hinder universal testing method that can be applied to all graph d atabase engines. Additionally, the ability to attachproperties to both vertices and edges further adds up to the complexity of theerror/bug dete",
+      "section": "3 Related Work",
+      "page": 5,
+      "char_offset": 14690,
+      "cited_reference": {
+        "number": 21,
+        "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. Pro c. ACM Program. Lang.4(OOPSLA) (2020). https:// doi.org/10.1145/3428279",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[\n8,10,13– 15,25,27,29]",
+      "technique": null,
+      "sentence": "However, in the context of graph databases, existing efforts to fuzzing Cypher or similar query languages only focused on testing the graph database engineitself [ 8,10,13– 15,25,27,29] (some of which were discussed in Sect.",
+      "context_before": "random test generation, test generation via change-impact analysis, and test generation for model-based policies[ 9]. In this paper, we adopt a random test generation approach. A related line of work is fuzzing, where instead of carefully crafted test cases, large numbers of automatically generated and unexpected inputs are used to explore the behav-ior of the system. While fuzzing has been applied to policy testing in frameworks such as XACML [ 1, 3,16], its application to testing access control within RDBMSs can be realized by adapting tools such as DynSQL [ 11], SQLaser [ 24] SQLFuzz [ 26].",
+      "context_after": "2), aiming at performance, correctness, and internal query evaluation logic. A widely employed technique in this regard is differential testing [ 4], which aims to detect logic bugs in multiple version or instances of graph database engines by passing the same query to all systems for producing identical results [ 8,27]. A major downside of differential testing is the query language (GDSmith [ 8] uses Cypher, w hile Grand [ 27] is Gremlin-based, and DGDB [ 25] presented a paradigm that detects bugs based on both query languages.) Another approach t o detecting bugs in GDBMS is metamorphic testi",
+      "section": "3 Related Work",
+      "page": 6,
+      "char_offset": 16478,
+      "cited_reference": {
+        "number": 10,
+        "text": "Jiang, Y., Liu, J., Ba, J., Yap, R.H.C., Liang, Z., Rigger, M.: Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’24. Association for Computin",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[\n10,13– 15,29]",
+      "technique": null,
+      "sentence": "Thus, these tools prov ide test oracles for a specific graphdatabase engine [ 10,13– 15,29], where GRev [ 15] further leverages test oracles to detect shared bugs across multiple GDBMS through equivalent query rewriting.",
+      "context_before": "logic. A widely employed technique in this regard is differential testing [ 4], which aims to detect logic bugs in multiple version or instances of graph database engines by passing the same query to all systems for producing identical results [ 8,27]. A major downside of differential testing is the query language (GDSmith [ 8] uses Cypher, w hile Grand [ 27] is Gremlin-based, and DGDB [ 25] presented a paradigm that detects bugs based on both query languages.) Another approach t o detecting bugs in GDBMS is metamorphic testing [ 2], which finds logic bugs in an individual graph database engine.",
+      "context_after": "However, a general issue with metamorphic testing is the requirement for specific oracles tailored to each test case. While aforementioned tools gener-ate syntactically and semantically valid queries and are effective for uncovering engine-level bugs in graph databases, they do not specifically focus on access con trol enforcement. To the best of our knowledge, our work presents the firstapproach to testing access control in graph databases through fuzzing. Building on the query rewriting framework introduced in [ 6], our tool, CypherFuzzer, uses fuzzing techniques to generate diverse Cypher quer",
+      "section": "3 Related Work",
+      "page": 6,
+      "char_offset": 17372,
+      "cited_reference": {
+        "number": 10,
+        "text": "Jiang, Y., Liu, J., Ba, J., Yap, R.H.C., Liang, Z., Rigger, M.: Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, ICSE ’24. Association for Computin",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:23:30Z",
+    "is_model_written": true,
+    "summary": "CypherFuzzer tests access control in graph databases rather than query evaluation: it generates Cypher queries under different user privileges and checks that results respect the permissions granted. Its premise is that existing graph-database fuzzing has targeted the engine itself and left the authorization layer untested.",
+    "narrative": "SQLancer's techniques appear as part of the body of engine-focused fuzzing the paper positions itself against. Test oracles are cited as the route by which prior work finds logic bugs in database engines, alongside SQL-Smith for crashes and Apollo for performance, and the graph-database tools -- GraphGenie, GRev and others -- are described as providing oracles for a specific engine. That is precisely the boundary CypherFuzzer steps outside: correctness of results given a privilege set, not correctness of results as such.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "background",
+      "M3": "definition",
+      "M4": "motivation",
+      "M5": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1007_978_981_96_4506_0_18.json b/_data/papers/paper_doi_10_1007_978_981_96_4506_0_18.json
new file mode 100644
index 0000000..0a85323
--- /dev/null
+++ b/_data/papers/paper_doi_10_1007_978_981_96_4506_0_18.json
@@ -0,0 +1,831 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:51Z",
+  "paper": {
+    "id": "paper:doi:10.1007/978-981-96-4506-0_18",
+    "title": "Review of Fuzz Testing Techniques for Database Management Systems",
+    "authors": [
+      "Yuheng Zhang",
+      "Hui Lu",
+      "Zhourui Zhang",
+      "Guo–Cheng Wu",
+      "Houlin Zhou",
+      "Zhenghao Li"
+    ],
+    "year": 2025,
+    "venue": "Communications in computer and information science",
+    "doi": "10.1007/978-981-96-4506-0_18",
+    "arxiv_id": null,
+    "s2_paper_id": null,
+    "url": "https://doi.org/10.1007/978-981-96-4506-0_18",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1007/978-981-96-4506-0_18",
+      "retrieved_at": "2026-09-09T01:30:51Z",
+      "chars": 56044,
+      "content_sha256": "sha256:70c0f0072e1661be4c9a4e1b0d30a366fba070698ef28cb9e603551020a09a03"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 19,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1344
+      },
+      {
+        "number": "2",
+        "title": "Database Fuzz Testing Basis",
+        "start": 5577
+      },
+      {
+        "number": "2.1",
+        "title": "Fuzz Definition and Principles",
+        "start": 5607
+      },
+      {
+        "number": "2.2",
+        "title": "Database Management System Overview",
+        "start": 6937
+      },
+      {
+        "number": "2.3",
+        "title": "Difficulties in Testing Database Management Systems",
+        "start": 8178
+      },
+      {
+        "number": "2.4",
+        "title": "Application of Fuzz Testing in Database Systems",
+        "start": 9370
+      },
+      {
+        "number": "3.1",
+        "title": "Classification Based on DBMS Testing Methodology",
+        "start": 16389
+      },
+      {
+        "number": "3.2",
+        "title": "Classification of DBMS Fuzz Testing Based on Testing Objectives",
+        "start": 18268
+      },
+      {
+        "number": "3.3",
+        "title": "Classification by Database Type",
+        "start": 26428
+      },
+      {
+        "number": "3.4",
+        "title": "Emerging Trends in Database Fuzz Testing",
+        "start": 30493
+      },
+      {
+        "number": "4.1",
+        "title": "Diversity and Effectiveness of Test Case Generation",
+        "start": 33586
+      },
+      {
+        "number": "4.2",
+        "title": "Enhancing Test Coverage",
+        "start": 35097
+      },
+      {
+        "number": "4.3",
+        "title": "Cross-Database Platform Portability",
+        "start": 36924
+      },
+      {
+        "number": "5.1",
+        "title": "Summary",
+        "start": 38114
+      },
+      {
+        "number": "5.2",
+        "title": "Future Work",
+        "start": 40253
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Laney, D., et al.: 3D data management: Controlling data volume, velocity and variety. META Group Res. Note 6, 1 (2001)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Su, S., et al.: IoT root union: a decentralized name resolving system for IoT based on blockchain. Inf. Process. Manag. 58(3), 102553 (2021)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Lu, H., Jin, C., Helu, X., Zhu, C., Guizani, N., Tian, Z.: AutoD: intelligent blockchain application unpacking based on JNI layer deception call. IEEE Netw. P(99), 1–7 (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Lu, H., Jin, C., Helu, X., Du, X., Guizani, M., Tian, Z.: DeepAutoD: research on distributed machine learning oriented scalable mobile communication security unpacking system. IEEE Tran. Netw. Sci. Eng. 9(4), 2052–2065 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Hu, N., Tian, Z., Lu, H., Du, X., Guizani, M.: A multiple-kernel clustering based intrusion detection scheme for 5G and IoT networks. Int. J. Mach. Learn. Cybern. 12(11), 3129–3144 (2021). https://doi.org/10.1007/s13042-020-01253-w",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Oracle Oracle Oracle China | Cloud Applications and Cloud Platforms. https://www.oracle. com/cn/. Accessed 03 Aug 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "MySQL. https://www.mysql.com/. Accessed 03 Aug 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "SQLite. https://www.sqlite.org/index.html. Accessed 03 Aug 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "PostgreSQL: The world’s most advanced open source database. https://www.postgresql.org/. Accessed 03 Aug 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Troy Hunt: The 773 Million Record “Collection #1” Data Breach. https://www.troyhunt.com/ the-773-million-record-collection-1-data-reach/. Accessed 03 Aug 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Feei: Security Analysis of the Equifax Data Breach Leading to the Exposure of 147 Million Users’ Data. https://feei.cn/equifax-data-breach/. Accessed 03 Aug 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "google/AFL (2024). https://github.com/google/AFL",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Pham, V .-T.: AFLSmart++: smarter greybox fuzzing. In: 2023 IEEE/ACM International Workshop on Search-Based and Fuzz Testing (SBFT), pp. 76–79. IEEE (2023)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Pham, V .-T., Böhme, M., Santosa, A.E., Căciulescu, A.R., Roychoudhury, A.: Smart greybox fuzzing. IEEE Trans. Softw. Eng. 47, 1980–1997 (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Böhme, M., Pham, V .-T., Roychoudhury, A.: Coverage-based greybox fuzzing as Markov chain. In: Proceedings of the 2016 ACMSIGSAC Conference on Computer and Communications Security, Vienna Austria, pp. 1032–1043. ACM (2016). https://doi.org/10.1145/297 6749.2978428",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Böhme, M., Pham, V .-T., Nguyen, M.-D., Roychoudhury, A.: Directed greybox fuzzing. In: Proceedings of the 2017 ACMSIGSAC Conference on Computer and Communications Security, Dallas, Texas, USA, pp. 2329–2344. ACM (2017). https://doi.org/10.1145/3133956. 3134020",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Tian, Z., Li, M., Qiu, M., Sun, Y ., Su, S.: Block-DEF: a secure digital evidence framework using blockchain. Inf. Sci. 491, 151–165 (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Beaman, C., Redbourne, M., Mummery, J.D., Hakak, S.: Fuzzing vulnerability discovery techniques: survey, challenges and future directions. Comput. Secur. 120, 102813 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Lu, H., et al.: Research on intelligent detection of command level stack pollution for binary program analysis. Mob. Netw. Appl. 26(4), 1723–1732 (2020). https://doi.org/10.1007/s11 036-019-01507-0 Review of Fuzz Testing Techniques for DBMS 299",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Li, M., Tian, Z., Du, X., Y uan, X., Shan, C., Guizani, M.: Power normalized cepstral robust features of deep neural networks in a cloud computing data privacy protection scheme. Neurocomputing 518 (2023)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Su, S., Tian, Z., Liang, S., Li, S., Du, S., Guizani, N.: A reputation management scheme for efficient malicious vehicle identification over 5G networks. IEEE Wirel. Commun. 27(3), 46–52 (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Thirunavukarasu, A.J., Ting, D.S.J., Elangovan, K., Gutierrez, L., Tan, T.F., Ting, D.S.W.: Large language models in medicine. Nat. Med. 29, 1930–1940 (2023)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Kasneci, E., et al.: ChatGPT for good? On opportunities and challenges of large language models for education. Learn. Individ. Differ. 103, 102274 (2023)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Chang, Y ., et al.: A survey on evaluation of large language models. ACM Trans. Intell. Syst. Technol. 15, 1–45 (2024). https://doi.org/10.1145/3641289",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Miller, B.P ., Fredriksen, L., So, B.: An empirical study of the reliability of UNIX utilities. Commun. ACM 33, 32–44 (1990). https://doi.org/10.1145/96267.96279",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Hou, J., Liu, F., Lu, H., Tan, Z., Zhuang, X., Tian, Z.: A novel flow-vector generation approach for malicious traffic detection. J. Parallel Distrib. Comput. 169, 72–86 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Cummins, C., Petoumenos, P ., Murray, A., Leather, H.: Compiler fuzzing through deep learn-ing. In: Proceedings of the 27th ACMSIGSOFT International Symposium on Software Testing and Analysis, Amsterdam, Netherlands, pp. 95–105. ACM (2018). https://doi.org/10.1145/ 3213846.3213848",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Liu, X., Li, X., Prajapati, R., Wu, D.: DeepFuzz: automatic generation of syntax valid C programs for fuzz testing. In: Proceedings of the AAAI Conference on Artificial Intelligence, pp. 1044–1051 (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Park, S., Xu, W., Y un, I., Jang, D., Kim, T.: Fuzzing Javascript engines with aspect-preserving mutation. In:. 1629–1642. IEEE (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "V eggalam, S., Rawat, S., Haller, I., Bos, H.: IFuzzer: an evolutionary interpreter fuzzer using genetic programming. In: Askoxylakis, I., Ioannidis, S., Katsikas, S., Meadows, C. (eds.) ESORICS 2016. LNCS, vol. 9878, pp. 581–601. Springer, Cham (2016). https://doi.org/10. 1007/978-3-319-45744-4_29",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Shi, H., et al.: Industry practice of coverage-guided enterprise Linux kernel fuzzing. In: Pro-ceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Tallinn, Estonia, pp. 986–995. ACM (2019). https://doi.org/10.1145/3338906.3340460",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Kim, K., Jeong, D.R., Kim, C.H., Jang, Y ., Shin, I., Lee, B.: HFL: hybrid fuzzing on the Linux kernel. In: NDSS (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Schumilo, S., Aschermann, C., Gawlik, R., Schinzel, S., Holz, T.: kAFL: hardware-assisted feedback fuzzing for OS kernels. In: 26th USENIX Security Symposium (USENIX Security 17), pp. 167–182 (2017)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Schumilo, S., Aschermann, C., Jemmett, A., Abbasi, A., Holz, T.: Nyx-Net: network fuzzing with incremental snapshots. In: Proceedings of the Seventeenth European Conference on Computer Systems, Rennes, France, pp. 166–180. ACM (2022). https://doi.org/10.1145/349 2321.3519591",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Ba, J., Böhme, M., Mirzamomen, Z., Roychoudhury, A.: Stateful greybox fuzzing. In: 31st USENIX Security Symposium (USENIX Security 22), pp. 3255–3272 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Gorbunov, S., Rosenbloom, A.: AutoFuzz: automated network protocol fuzzing framework. IJCSNS 10, 239 (2010)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Wang, Y .: Research on key technologies of coverage-directed fuzz testing (Doctoral dissertation). Strategic support force information engineering university (2020). (in Chinese) 300 Y. Zhang et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Lin, Y .-D., Liao, F.-Z., Huang, S.-K., Lai, Y .-C.: Browser fuzzing by scheduled mutation and generation of document object models. In: 2015 International Carnahan Conference on Security Technology (ICCST), pp. 1–6. IEEE (2015)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Huo, W., et al.: Pattern-based fuzz testing techniques for browsers. J. Softw. 29(5), 1275–1287 (2018). (in Chinese)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Helmke, R., Winter, E., Rademacher, M.: EPF: an evolutionary, protocol-aware, and coverageguided network fuzzing framework. In: 2021 18th International Conference on Privacy, Security and Trust (PST), pp. 1–7. IEEE (2021)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Luo, Z., Zuo, F., Shen, Y ., Jiao, X., Chang, W., Jiang, Y .: ICS protocol fuzzing: coverage guided packet crack and generation. In: 2020 57th ACM/IEEE Design Automation Conference (DAC), pp. 1–6. IEEE (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Liu, K.: Research on vulnerability discovery techniques in industrial control networks based on fuzz testing. Softw. Eng. 23, 27–29+23 (2020). (in Chinese). https://doi.org/10.19644/j. cnki.issn2096-1472.2020.12.008",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "V erma, J.P ., Agrawal, S., Patel, B., Patel, A.: Big data analytics: challenges and applications for text, audio, video, and social media data. Int. J. Soft Comput. Artif. Intell. Appl. IJSCAI 5, 41–51 (2016)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Xu, H.: Database management systems and their development trends. Microcomput. Appl. 25(10), 83 (2006). (in Chinese)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Saeed, R.A., Ahmed, E.S.A.: Big data distributed systems management. Netw. Big Data 2, 57 (2015)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Bajaj, R.H., Ramteke, P .: Big data–the new era of data. Int. J. Comput. Sci. Inf. Technol. 5, 1875–1885 (2014)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Li, G.J.: The scientific value of big data research. J. Big Data Stud. (2012). (in Chinese). https://ict.cas.cn/liguojiewenxuan_162523/wzlj/lgjxsbg/201912/P02019122 7654597760261.pdf. Accessed 03 Aug 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Rigger, M., Su, Z.: Testing database engines via pivoted query synthesis. In: 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pp. 667–682 (2020)",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 49,
+      "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, 1–30 (2020). https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "Lin, W., Hua, Z., Ren, L., Li, Z., Zhang, L., Xie, T.: GDsmith: detecting bugs in graph database engines. http://arxiv.org/abs/2206.08530 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Fu, J., Liang, J., Wu, Z., Wang, M., Jiang, Y .: Griffin: grammar-free DBMS fuzzing. In: Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, Rochester, MIUSA, p. 112. ACM (2022). https://doi.org/10.1145/3551349.356 0431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Liu, X., Zhou, Q., Arulraj, J., Orso, A.: Automatic detection of performance bugs in database systems using equivalent queries. In: Proceedings of the 44th International Conference on Software Engineering, Pittsburgh, Pennsylvania, pp. 225–236. ACM (2022). https://doi.org/ 10.1145/3510003.3510093",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Jung, J., Hu, H., Arulraj, J., Kim, T., Kang, W.: APOLLO: automatic detection and diagnosis of performance regressions in database systems. Proc. VLDB Endow. 13, 57–70 (2019). https://doi.org/10.14778/3357377.3357382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Rigger, M., Su, Z.: Detecting optimization bugs in database engines via non-optimizing ref-erence engine construction. In: Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event USA, pp. 1140–1152. ACM (2020). https://doi.org/10.1145/3368089. 3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 55,
+      "text": "Rigger, M.: Sqlancer: detecting logic bugs in dbms (2020) Review of Fuzz Testing Techniques for DBMS 301",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 56,
+      "text": "Seltenreich, A.: anse1/sqlsmith (2024). https://github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Slutz, D.R.: Massive stochastic testing of SQL. In: VLDB, pp. 618–622. Citeseer (1998)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Gu, Z., Hu, W., Zhang, C., Lu, H., Wang, L.: Gradient shielding: towards understanding vulnerability of deep neural networks. IEEE Trans. Netw. Sci. Eng. 8(2), 921–932 (2021)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "SAGE: Whitebox Fuzzing for Security Testing-ACM Queue. https://queue.acm.org/detail. cfm?id=2094081. Accessed 03 Aug 2024",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Cadar, C., Dunbar, D., Engler, D.R.: Klee: unassisted and automatic generation of highcoverage tests for complex systems programs. In: OSDI, pp. 209–224 (2008)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Wen, S., Jia, P ., Yang, P ., Hu, C.: Squill: testing DBMS with correctness feedback and accurate instantiation. Appl. Sci. 13, 2519 (2023)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Jiang, Z.-M., Bai, J.-J., Su, Z.: DynSQL: stateful fuzzing for database management sys-tems with complex and valid SQL query generation. In: 32nd USENIX Security Symposium (USENIX Security 23), pp. 4949–4965 (2023)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Liang, J., et al.: Sequence-oriented DBMS fuzzing. In:. 668–681. IEEE (2023)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Liang, Y ., Liu, S., Hu, H.: Detecting logical bugs of DBMS with coverage-based guidance. In: 31st USENIX Security Symposium (USENIX Security 22), pp. 4309–4326 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Wu, Z., Liang, J., Wang, M., Zhou, C., Jiang, Y .: Unicorn: detect runtime errors in timeseries databases with hybrid input synthesis. In: Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual South Korea, pp. 251– [262] ACM (2022). https://doi.org/10.1145/3533767.3534364",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Wang, M., et al.: Industry practice of coverage-guided enterprise-level DBMS fuzzing. In: Engineering in Practice (ICSE-SEIP), pp. 328–337. IEEE (2021)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Zhong, R., Chen, Y ., Hu, H., Zhang, H., Lee, W., Wu, D.: SQUIRREL: testing database management systems with language validity and coverage feedback. In: Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event USA, pp. 955–970. ACM (2020). https://doi.org/10.1145/3372297.3417260",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Meng, R., Mirchev, M., Böhme, M., Roychoudhury, A.: Large language model guided pro-tocol fuzzing. In: Proceedings of the 31st Annual Network and Distributed System Security Symposium (NDSS) (2024)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Xia, C.S., Paltenghi, M., Le Tian, J., Pradel, M., Zhang, L.: Fuzz4All: universal fuzzing with large language models. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, Lisbon, Portugal, pp. 1–13. ACM (2024). https://doi.org/10.1145/359 7503.3639121",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Deng, Y ., Xia, C.S., Yang, C., Zhang, S.D., Yang, S., Zhang, L.: Large language models are edge-case generators: crafting unusual programs for fuzzing deep learning libraries. In: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, Lisbon, Portugal, pp. 1–13. ACM (2024). https://doi.org/10.1145/3597503.3623343",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Lin, D.-L., Zhang, Y ., Ren, H., Khailany, B., Wang, S.-H., Huang, T.-W.: GenFuzz: GPU-accelerated hardware fuzzing using genetic algorithm with multiple inputs. In: 2023 60th ACM/IEEE Design Automation Conference (DAC), pp. 1–6 (2023). https://doi.org/10.1109/ DAC56929.2023.10247942",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Ammann, M., Hirschi, L., Kremer, S.: DY Fuzzing: Formal Dolev-Yao Models Meet Cryptographic Protocol Fuzz Testing (2023). https://eprint.iacr.org/2023/057",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Source-based Code Coverage — Clang 12 documentation. https://releases.llvm.org/12.0.0/ tools/clang/docs/SourceBasedCodeCoverage.html. Accessed 03 Aug 2024",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 48,
+      "text": "Rigger, M., Su, Z.: Testing database engines via pivoted query synthesis. In: 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pp. 667–682 (2020)",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 49,
+      "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, 1–30 (2020). https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 54,
+      "text": "Rigger, M., Su, Z.: Detecting optimization bugs in database engines via non-optimizing ref-erence engine construction. In: Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event USA, pp. 1140–1152. ACM (2020). https://doi.org/10.1145/3368089. 3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec"
+      ]
+    },
+    {
+      "number": 55,
+      "text": "Rigger, M.: Sqlancer: detecting logic bugs in dbms (2020) Review of Fuzz Testing Techniques for DBMS 301",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Many tools have proposed innovative testing criteria, such as the PQS [ 48] and TLP [ 49]o f 288 Y.",
+      "context_before": "S fuzz testing tools have emerged, uncovering many issues in popular databases and significantly enhancing their security. Table 1 lists some of the typical DBMS fuzz testing tools developed in recent years and their capabilities across various dimensions. These tools can be classified according to the type of vulnerabilities they target: DBMS testing tools for crash vulnerabilities, DBMS testing tools for logic vulnerabilities, and DBMS testing tools for performance vulnerabilities. To capture these different types of vulnerabilities, each tool needs to establish corresponding testing criteria.",
+      "context_after": "Zhang et al. SQLancer. Due to the different grammars and characteristics of various DBMS, most fuzz testing tools require specific adaptations for each database. The classification of fuzz testing tools is necessitated by the diverse testing requirements and focal points dictated by different testing methods, objectives, and database types. Firstly, categorizing DBMS testing methods into black-box, white-box, and graybox testing allows for flexible responses to different testing stages and depths. Secondly, categorizing based on testing objectives—crash testing, performance testing, and logi-cal",
+      "section": "2.4 Application of Fuzz Testing in Database Systems",
+      "page": 5,
+      "char_offset": 13183,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer.",
+      "context_before": "ecurity. Table 1 lists some of the typical DBMS fuzz testing tools developed in recent years and their capabilities across various dimensions. These tools can be classified according to the type of vulnerabilities they target: DBMS testing tools for crash vulnerabilities, DBMS testing tools for logic vulnerabilities, and DBMS testing tools for performance vulnerabilities. To capture these different types of vulnerabilities, each tool needs to establish corresponding testing criteria. Many tools have proposed innovative testing criteria, such as the PQS [ 48] and TLP [ 49]o f 288 Y. Zhang et al.",
+      "context_after": "Due to the different grammars and characteristics of various DBMS, most fuzz testing tools require specific adaptations for each database. The classification of fuzz testing tools is necessitated by the diverse testing requirements and focal points dictated by different testing methods, objectives, and database types. Firstly, categorizing DBMS testing methods into black-box, white-box, and graybox testing allows for flexible responses to different testing stages and depths. Secondly, categorizing based on testing objectives—crash testing, performance testing, and logi-cal testing—enables precis",
+      "section": "2.4 Application of Fuzz Testing in Database Systems",
+      "page": 6,
+      "char_offset": 13296,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Typical DBMS Fuzz Testing Tools and Their Support Capabilities Across Different Dimensions Tool Year Type Method Vulnerability Type Currently Supported RAGS 1998 black-box testing genertion-based logic Microsoft SQL Server SQLsmith 2015 black-box testing genertion-based crash PostgreSQL SQLite SQLancer 2020 black-box testing genertion-based logic PostgreSQL MySQL SQLite Apollo 2019 black-box testing generation-based performances PostgreSQL SQLite Squirrel 2020 grey-box testing mutation-based crash PostgreSQL MySQL SQLite (continued) Review of Fuzz Testing Techniques for DBMS 289 Table 1.",
+      "context_before": "tics of different database architectures. Relational databases, which use Structured Query Language (SQL), require a focus on SQL injection protection and complex query processing. In contrast, NoSQL databases handle unstructured data and large-scale datasets, so testing tools must prioritize query language security, data consistency, and fault tolerance in distributed systems. Through such classification, testing can be conducted more systematically, ensuring that database systems maintain high security, reliability, and performance in the complex and variable real-world applications. Table 1.",
+      "context_after": "(continued) Tool Year Type Method Vulnerability TypeCurrently Supported Ratel 2021 grey-box testing mutation-based crash GaussDB PostgreSQL Comdb2 Amoeba 2022 black-box testing generation-based performances PostgreSQL CocroachDB Unicorn 2022 grey-box testing mutation-based crash IoTDB KairosDB GridDB SQRight 2022 grey-box testing mutation-based logic PostgreSQL SQLite MySQL Griffin 2022 black-box testing mutation-based crash MariaDB SQLite LEGO 2023 grey-box testing mutation-based crash PostgreSQL MariaDB MySQL DynSQL 2023 grey-box testing mutation-based crash SQlite MariaDB MySQL QPG 2023 blac",
+      "section": "2.4 Application of Fuzz Testing in Database Systems",
+      "page": 6,
+      "char_offset": 15046,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "(continued) Tool Year Type Method Vulnerability TypeCurrently Supported Ratel 2021 grey-box testing mutation-based crash GaussDB PostgreSQL Comdb2 Amoeba 2022 black-box testing generation-based performances PostgreSQL CocroachDB Unicorn 2022 grey-box testing mutation-based crash IoTDB KairosDB GridDB SQRight 2022 grey-box testing mutation-based logic PostgreSQL SQLite MySQL Griffin 2022 black-box testing mutation-based crash MariaDB SQLite LEGO 2023 grey-box testing mutation-based crash PostgreSQL MariaDB MySQL DynSQL 2023 grey-box testing mutation-based crash SQlite MariaDB MySQL QPG 2023 black-box testing mutation-based performances MySQL PostgreSQL Oracle SQLite Squill 2023 grey-box testing mutation-based crash PostgreSQL SQLite Oracle 3.",
+      "context_before": "le 1. Typical DBMS Fuzz Testing Tools and Their Support Capabilities Across Different Dimensions Tool Year Type Method Vulnerability Type Currently Supported RAGS 1998 black-box testing genertion-based logic Microsoft SQL Server SQLsmith 2015 black-box testing genertion-based crash PostgreSQL SQLite SQLancer 2020 black-box testing genertion-based logic PostgreSQL MySQL SQLite Apollo 2019 black-box testing generation-based performances PostgreSQL SQLite Squirrel 2020 grey-box testing mutation-based crash PostgreSQL MySQL SQLite (continued) Review of Fuzz Testing Techniques for DBMS 289 Table 1.",
+      "context_after": "1 Classification Based on DBMS Testing Methodology Classifying DBMS fuzz testing methods into black-box, white-box, and gray-box approaches allows for flexible adaptation to different stages and depths of the testing process. Black-Box Testing. Black-box testing does not require knowledge of the internal implementation details of the database; it focuses solely on the relationship between input and output [ 48, 50–58]. This method is suitable for simulating external attacks and evaluating the system’s external defenses and input validation mechanisms. By introducing various anomalous inputs, bla",
+      "section": "2.4 Application of Fuzz Testing in Database Systems",
+      "page": 7,
+      "char_offset": 15641,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[ 48, 50–58]",
+      "technique": "pqs",
+      "sentence": "Black-box testing does not require knowledge of the internal implementation details of the database; it focuses solely on the relationship between input and output [ 48, 50–58].",
+      "context_before": "ck-box testing mutation-based crash MariaDB SQLite LEGO 2023 grey-box testing mutation-based crash PostgreSQL MariaDB MySQL DynSQL 2023 grey-box testing mutation-based crash SQlite MariaDB MySQL QPG 2023 black-box testing mutation-based performances MySQL PostgreSQL Oracle SQLite Squill 2023 grey-box testing mutation-based crash PostgreSQL SQLite Oracle 3.1 Classification Based on DBMS Testing Methodology Classifying DBMS fuzz testing methods into black-box, white-box, and gray-box approaches allows for flexible adaptation to different stages and depths of the testing process. Black-Box Testing.",
+      "context_after": "This method is suitable for simulating external attacks and evaluating the system’s external defenses and input validation mechanisms. By introducing various anomalous inputs, black-box testing can reveal how the system handles unexpected data, thereby uncovering vulnerabilities in its external defensive capabilities. 290 Y. Zhang et al. White-Box Testing. White-box testing involves generating targeted test cases by collecting state information and leveraging the internal information of the database, including data structures and code logic [ 59, 60]. This method allows for the design of high",
+      "section": "3.1 Classification Based on DBMS Testing Methodology",
+      "page": 7,
+      "char_offset": 16633,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rigger, M., Su, Z.: Testing database engines via pivoted query synthesis. In: 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pp. 667–682 (2020)",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "(1) SQLancer: SQLancer [ 48, 54, 55] is a well-known tool for detecting logical vulnerabilities in database management systems.",
+      "context_before": "esting criteria for logical issues. This process necessitates a deep understanding of the DBMS’s behavioral logic, along with thorough analysis and semantic modeling. Differential testing and metamorphic testing are effective methods for determining expected results, as they significantly reduce the difficulty of semantic modeling. Differential testing can derive expected results from a similar database management system, while metamorphic testing transforms results from existing queries. In addition to validating the correctness of query results, other logic-related issues must also be verified.",
+      "context_after": "Using the PQS method, SQLancer has conducted extensive testing on databases such as MySQL, PostgreSQL, and SQLite, ultimately discovering over 60 related logical vulnerabilities. The NoREC (Non-Optimizing Reference Engine Construction) criterion transforms an optimizable query into one that cannot be effectively optimized and then compares the result sets of both queries. Inconsistent results indicate an optimization-related logical issue. SQLancer has identified 51 logical vulnerabilities using the NoREC method in PostgreSQL, MariaDB, SQLite, and CockroachDB. The TLP (Ternary Logic Partitioni",
+      "section": "3.2 Classification of DBMS Fuzz Testing Based on Testing Objectives",
+      "page": 9,
+      "char_offset": 21790,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Using the PQS method, SQLancer has conducted extensive testing on databases such as MySQL, PostgreSQL, and SQLite, ultimately discovering over 60 related logical vulnerabilities.",
+      "context_before": "orough analysis and semantic modeling. Differential testing and metamorphic testing are effective methods for determining expected results, as they significantly reduce the difficulty of semantic modeling. Differential testing can derive expected results from a similar database management system, while metamorphic testing transforms results from existing queries. In addition to validating the correctness of query results, other logic-related issues must also be verified. (1) SQLancer: SQLancer [ 48, 54, 55] is a well-known tool for detecting logical vulnerabilities in database management systems.",
+      "context_after": "The NoREC (Non-Optimizing Reference Engine Construction) criterion transforms an optimizable query into one that cannot be effectively optimized and then compares the result sets of both queries. Inconsistent results indicate an optimization-related logical issue. SQLancer has identified 51 logical vulnerabilities using the NoREC method in PostgreSQL, MariaDB, SQLite, and CockroachDB. The TLP (Ternary Logic Partitioning) criterion divides a query into multiple subqueries, each producing a portion of the original query’s results. The union of these subqueries should match the original query’s r",
+      "section": "3.2 Classification of DBMS Fuzz Testing Based on Testing Objectives",
+      "page": 9,
+      "char_offset": 21918,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "The NoREC (Non-Optimizing Reference Engine Construction) criterion transforms an optimizable query into one that cannot be effectively optimized and then compares the result sets of both queries.",
+      "context_before": "ty of semantic modeling. Differential testing can derive expected results from a similar database management system, while metamorphic testing transforms results from existing queries. In addition to validating the correctness of query results, other logic-related issues must also be verified. (1) SQLancer: SQLancer [ 48, 54, 55] is a well-known tool for detecting logical vulnerabilities in database management systems. Using the PQS method, SQLancer has conducted extensive testing on databases such as MySQL, PostgreSQL, and SQLite, ultimately discovering over 60 related logical vulnerabilities.",
+      "context_after": "Inconsistent results indicate an optimization-related logical issue. SQLancer has identified 51 logical vulnerabilities using the NoREC method in PostgreSQL, MariaDB, SQLite, and CockroachDB. The TLP (Ternary Logic Partitioning) criterion divides a query into multiple subqueries, each producing a portion of the original query’s results. The union of these subqueries should match the original query’s result. Discrepancies indicate logical issues. SQLancer discov-ered 175 errors using the TLP method across databases like MySQL, TiDB, SQLite, and CockroachDB, with 77 identified as logical issues.",
+      "section": "3.2 Classification of DBMS Fuzz Testing Based on Testing Objectives",
+      "page": 9,
+      "char_offset": 22097,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer has identified 51 logical vulnerabilities using the NoREC method in PostgreSQL, MariaDB, SQLite, and CockroachDB.",
+      "context_before": "issues must also be verified. (1) SQLancer: SQLancer [ 48, 54, 55] is a well-known tool for detecting logical vulnerabilities in database management systems. Using the PQS method, SQLancer has conducted extensive testing on databases such as MySQL, PostgreSQL, and SQLite, ultimately discovering over 60 related logical vulnerabilities. The NoREC (Non-Optimizing Reference Engine Construction) criterion transforms an optimizable query into one that cannot be effectively optimized and then compares the result sets of both queries. Inconsistent results indicate an optimization-related logical issue.",
+      "context_after": "The TLP (Ternary Logic Partitioning) criterion divides a query into multiple subqueries, each producing a portion of the original query’s results. The union of these subqueries should match the original query’s result. Discrepancies indicate logical issues. SQLancer discov-ered 175 errors using the TLP method across databases like MySQL, TiDB, SQLite, and CockroachDB, with 77 identified as logical issues. (2) SQLRight: SQLRight [ 64] integrates differential testing and mutation-based fuzzing methods to detect logical errors in DBMSs. It designs a set of generic APIs to decouple the fuzzer from",
+      "section": "3.2 Classification of DBMS Fuzz Testing Based on Testing Objectives",
+      "page": 9,
+      "char_offset": 22362,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The TLP (Ternary Logic Partitioning) criterion divides a query into multiple subqueries, each producing a portion of the original query’s results.",
+      "context_before": "es in database management systems. Using the PQS method, SQLancer has conducted extensive testing on databases such as MySQL, PostgreSQL, and SQLite, ultimately discovering over 60 related logical vulnerabilities. The NoREC (Non-Optimizing Reference Engine Construction) criterion transforms an optimizable query into one that cannot be effectively optimized and then compares the result sets of both queries. Inconsistent results indicate an optimization-related logical issue. SQLancer has identified 51 logical vulnerabilities using the NoREC method in PostgreSQL, MariaDB, SQLite, and CockroachDB.",
+      "context_after": "The union of these subqueries should match the original query’s result. Discrepancies indicate logical issues. SQLancer discov-ered 175 errors using the TLP method across databases like MySQL, TiDB, SQLite, and CockroachDB, with 77 identified as logical issues. (2) SQLRight: SQLRight [ 64] integrates differential testing and mutation-based fuzzing methods to detect logical errors in DBMSs. It designs a set of generic APIs to decouple the fuzzer from the oracle logic, allowing developers to easily port fuzzing tools to test DBMSs and to create new oracles for existing fuzzers. SQLRight detected",
+      "section": "3.2 Classification of DBMS Fuzz Testing Based on Testing Objectives",
+      "page": 9,
+      "char_offset": 22484,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer discov-ered 175 errors using the TLP method across databases like MySQL, TiDB, SQLite, and CockroachDB, with 77 identified as logical issues.",
+      "context_before": "onstruction) criterion transforms an optimizable query into one that cannot be effectively optimized and then compares the result sets of both queries. Inconsistent results indicate an optimization-related logical issue. SQLancer has identified 51 logical vulnerabilities using the NoREC method in PostgreSQL, MariaDB, SQLite, and CockroachDB. The TLP (Ternary Logic Partitioning) criterion divides a query into multiple subqueries, each producing a portion of the original query’s results. The union of these subqueries should match the original query’s result. Discrepancies indicate logical issues.",
+      "context_after": "(2) SQLRight: SQLRight [ 64] integrates differential testing and mutation-based fuzzing methods to detect logical errors in DBMSs. It designs a set of generic APIs to decouple the fuzzer from the oracle logic, allowing developers to easily port fuzzing tools to test DBMSs and to create new oracles for existing fuzzers. SQLRight detected 18 logical errors in the thoroughly tested DBMSs, SQLite and MySQL. All errors have been confirmed, with 14 already fixed. Performance Testing: Performance testing evaluates the database’s response time and stability under high load conditions. It focuses on met",
+      "section": "3.2 Classification of DBMS Fuzz Testing Based on Testing Objectives",
+      "page": 9,
+      "char_offset": 22742,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, Unicorn identified 21%-199% and 34%-693% more basic blocks than SQLsmith and SQLancer, respectively, across widely used time-series databases such as IoTDB, KairosDB, TimescaleDB, TDEngine, QuestDB, and GridDB.",
+      "context_before": "performance under various loads. (1) Unicorn: Unicorn [ 65] combines syntax-preserving mutation with time-seriesguided mutation to test time-series database systems. It employs a hybrid specifica-tion synthesis approach to generate inputs containing time-series information and features. Additionally, Unicorn uses active anomaly detection to extract exceptions from the runtime environment, uncovering vulnerabilities hidden within implicit exception handling mechanisms. In experiments, Unicorn demonstrated superior coverage and defect detection compared to state-of-the-art database fuzzing tools.",
+      "context_after": "It discovered 42 unknown bugs in time-series databases like IoTDB and KairosDB. (2) Griffin: Fu et al. introduced Griffin [ 51], a database fuzzing solution that does not rely on syntax information, significantly reducing the adaptation cost of fuzzing techniques for database software. Griffin utilizes metadata graphs to mutate test cases in a syntax-free manner. Despite the lack of syntax information, Griffin remains effective in generating valid SQL statements and discovering vulnerabilities. It has identified 55 unknown bugs related to memory errors in databases such as DuckDB and MariaDB and ha",
+      "section": "3.3 Classification by Database Type",
+      "page": 11,
+      "char_offset": 29131,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Test Tools Target Program SQLite PostgreSQL rows branch rows branch SQLsmith 27 22 33 25 Squirrel 59 49 15 11 Griffin 59 52 50 37 SQLancer 52 44 28 20 SQLRight 60 50 25 18 Apollo 26 20 – – 4.",
+      "context_before": "erformance under these scenarios. As shown in Table 3 currently, popular fuzz testing tools still have significant room for improvement in overall coverage on the DBMS being tested. Coverage in rule-based generation tools is constrained by the rules themselves, while mutation-based tools struggle with generating effective test cases. Future efforts to enhance test coverage will involve leveraging feedback-based coverage, employing generative rules, integrating tests, and adapting tools to specific DBMS requirements. 296 Y. Zhang et al. Table 3. Coverage of each tool on the DBMS to be tested (%).",
+      "context_after": "3 Cross-Database Platform Portability Modern enterprises often utilize multiple database systems, such as MySQL, SQLite, PostgreSQL, Oracle, and MongoDB. However, the same fuzz testing tools do not necessarily perform equally well across different database systems. Therefore, cross-platform fuzz testing for databases needs to address the following challenges: Syntax and Feature Differences: SQL syntax and functionality can vary significantly across different database systems. Designing cross-platform test cases requires account-ing for these differences to ensure that the test cases are applica",
+      "section": "4.2 Enhancing Test Coverage",
+      "page": 14,
+      "char_offset": 36735,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:59:32Z",
+    "is_model_written": true,
+    "summary": "A survey of fuzz testing for database management systems, classifying tools by testing type, generation method, target vulnerability class and supported systems, and tracing the field from RAGS in 1998 through the generation-based and coverage-guided tools that followed. It tabulates comparative coverage figures for the major tools and discusses how test coverage might be improved.",
+    "narrative": "SQLancer is one of the survey's principal subjects, given its own subsection as a well-known tool for detecting logical vulnerabilities and covered oracle by oracle with the bug counts each produced: over 60 logic bugs from PQS across MySQL, PostgreSQL and SQLite; 51 from NoREC across PostgreSQL, MariaDB, SQLite and CockroachDB; 175 errors from TLP across MySQL, TiDB, SQLite and CockroachDB, 77 of them logic bugs. It also appears in the survey's comparative tables, both in its own coverage row and as the baseline in Unicorn's reported figures.",
+    "roles": {
+      "M1": "definition",
+      "M2": "definition",
+      "M3": "background",
+      "M4": "background",
+      "M5": "definition",
+      "M6": "definition",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "definition",
+      "M10": "definition",
+      "M11": "result_comparison",
+      "M12": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "A survey describes tools rather than using them; nothing here reports running or building on SQLancer."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is proposed or extended."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The tabulated coverage figures are collected from the surveyed papers rather than measured here, so the survey reports comparisons rather than conducting one."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is called well-known and its results are reported at length, but the survey does not claim it is the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1007_978_981_96_6465_8_28.json b/_data/papers/paper_doi_10_1007_978_981_96_6465_8_28.json
new file mode 100644
index 0000000..a72e36f
--- /dev/null
+++ b/_data/papers/paper_doi_10_1007_978_981_96_6465_8_28.json
@@ -0,0 +1,420 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:13:33Z",
+  "paper": {
+    "id": "paper:doi:10.1007/978-981-96-6465-8_28",
+    "title": "Fuzz Testing for Database Management System Configuration Errors",
+    "authors": [
+      "Haoran Zhu",
+      "Menglin Li",
+      "Bo Wang",
+      "Tengfei Li",
+      "Jingtian Liu",
+      "Zan Zhou"
+    ],
+    "year": 2025,
+    "venue": "Communications in computer and information science",
+    "doi": "10.1007/978-981-96-6465-8_28",
+    "arxiv_id": null,
+    "s2_paper_id": null,
+    "url": "https://doi.org/10.1007/978-981-96-6465-8_28",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1007/978-981-96-6465-8_28",
+      "retrieved_at": "2026-09-08T17:13:33Z",
+      "chars": 33961,
+      "content_sha256": "sha256:c7be7df3d9ac8c5602e0d3fa038fd8fa65f77dfab7de2e9e4ba24282606a4c64"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1762
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 5971
+      },
+      {
+        "number": "3",
+        "title": "Overview of Ic Fuzz",
+        "start": 12303
+      },
+      {
+        "number": "3.1",
+        "title": "The Shortcomings of Traditional Fuzz Testing",
+        "start": 12325
+      },
+      {
+        "number": "3.2",
+        "title": "Design and Implementation of IC Fuzz System",
+        "start": 14664
+      },
+      {
+        "number": "3.3",
+        "title": "Oriented Configuration Modification Strategy",
+        "start": 19059
+      },
+      {
+        "number": "3.4",
+        "title": "Dynamic Configuration Generation Strategy",
+        "start": 21613
+      },
+      {
+        "number": "4",
+        "title": "Evaluation",
+        "start": 24099
+      },
+      {
+        "number": "4.1",
+        "title": "System Test Environment",
+        "start": 24915
+      },
+      {
+        "number": "4.2",
+        "title": "Configuration Item Error Verification",
+        "start": 25186
+      },
+      {
+        "number": "4.3",
+        "title": "Efficiency Evaluation",
+        "start": 25751
+      },
+      {
+        "number": "5",
+        "title": "Summary",
+        "start": 26856
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Xu, T., et al.: Early Detection of Configuration Errors to Reduce Failure Damage",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Y uan, D., et al.: Simple Testing Can Prevent Most Critical Failures",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Rabkin, A., Katz, R.H.: How hadoop clusters break. IEEE Softw. 30(4), 88–94 (2013)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Xu, T., Zhou, Y .: Systems approaches to tackling configuration errors: a survey. ACM Comput. Surv. 47(4), 1–41 (2015)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Zhu, X., Wen, S., Camtepe, S., Xiang, Y .: Fuzzing: a survey for roadmap. ACM Com-puting Surv eys 54(11s), 1–36 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Wen, S., Jia, P ., Yang, P ., Hu, C.: Squill: testing dbms with correctness feedback and accurate instantiation. Appl. Sci. 13(4), 2519 (2023)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Vikram, V ., et al.: Guiding greybox fuzzing with mutation testing. Proceedings of the 32nd ACM SIGSOFT International Symposium on Software Testing and Analysis. Seattle W A USA: ACM, pp. 929–941 (2023) 332 H. Zhu et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Lu, Y ., Shao, K., Sun, W., Sun, M.: RGChaser: a rl-guided fuzz and mutation test-ing framew ork for deep learning systems. In: 2022 9th International Conference on Dependable Systems and Their Applications (DSA), pp. 12–23 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Zhou, Z., Kuang, X., Sun, L., et al.: Endogenous security defense against deductive attack: when artificial intelligence meets active defense for online service. IEEE Commun. Mag. 58(6), 58–64 (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Zhou, Z., Xu, C., Wang, M., et al.: A multi-shuffler framework to establish mutual confi-dence for secure federated learning. IEEE Transactions on Dependable and Secure Computing (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Zhou, Z., Zhuang, Y ., Li, H., et al.: MR-FFL: a stratified community-based mutual reliability framework for fairness-aware federated learning in heterogeneous UA V networks. IEEE Internet of Things Journal (2024)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Zhou, Z., Xu, C., Yang, S., et al.: Safeguarding privacy and integrity of federated learning in heterogeneous cross-silo IoRT Environments: a moving target defense approach. IEEE Network (2024)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Zhao, Y ., Zhou, F., Feng, L., Li, W., Y u, P .: MADRL-based 3D deployment and user association of cooperative mmwave aerial base stations for capacity enhancement. Chin. J. Electron. 32(2), 283–294 (2023). https://doi.org/10.23919/cje.2021.00.327",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Lv, C., et al.: MOPT: Optimize Mutation Scheduling for Fuzzers 21",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Y un, I., Lee, S., Xu, M., Jang, Y ., Kim, T.: QSYM: A Practical Concolic Execution Engine T ailored for Hybrid Fuzzing",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Chen, P ., Chen, H.: Angora: efficient fuzzing by principled search. In: on Security and Privacy (SP). San Francisco, CA: IEEE, pp. 711–725 (2018)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Li, Y ., et al.: V-Fuzz: vulnerability prediction-assisted evolutionary fuzzing for binary programs. IEEE Transactions on Cybernetics 52(5), 3745–3756 (2022). https://doi.org/10.1109/ TCYB.2020.3013675",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Karamcheti, S., Mann, G., Rosenberg, D.: Adaptive grey-box fuzz-testing with thomp-son sampling. In: Proceedings of the 11th ACM Workshop on Artificial Intelligence and Security, pp. 37–47 (2018)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Liu, X., Zhou, Q., Arulraj, J., Orso, A.: Automatic detection of performance bugs in database systems using equivalent queries. Proceedings of the 44th International Conference on Software Engineering. Pittsburgh Pennsylvania: ACM, pp. 225–236 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Ba, J., Rigger, M.: Testing database engines via query plan guidance. 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). Melbourne, Australia: IEEE, pp. 2060–2071 (2023)",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Rigger, M., Su, Z.: Detecting optimization bugs in database engines via non-optimizing refer-ence engine construction. In: Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. Virtual Event USA: ACM, pp. 1140–1152 (2020)",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA): 1–30 (2020)",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 23,
+      "text": "Zhong, R., Chen, Y ., Hu, H., Zhang, H., Lee, W., Wu, D.: SQUIRREL: testing database management systems with language validity and coverage feedback. Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. Virtual Event USA: ACM, pp. 955–970 (2020)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Jung, J., Hu, H., Arulraj, J., Kim, T., Kang, W.: APOLLO: automatic detection and diagnosis of performance regressions in database systems. Proceedings of the VLDB Endowment 13(1), 57–70 (2019)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Fu, J., Liang, J., Wu, Z., Wang, M., Jiang, Y .: Griffin: grammar-free dbms fuzzing. Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. New Y ork, NY, USA: Association for Computing Machinery, pp. 1–12 (2023) Fuzz Testing for Database Management System Configuration Errors 333",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Liang, Y ., Liu, S., Hu, H.: Detecting logical bugs of DBMS with coverage-based guidance. In: 31st USENIX Security Symposium (USENIX Security 22), pp. 4309–4326 (2022)",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Li, W., et al.: Challenges and opportunities: an in-depth empirical study on configuration error injection testing. Proceedings of the 30th ACMSIGSOFT International Symposium on Software Testing and Analysis. Virtual Denmark: ACM, pp. 478–490 (2021)",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 20,
+      "text": "Ba, J., Rigger, M.: Testing database engines via query plan guidance. 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). Melbourne, Australia: IEEE, pp. 2060–2071 (2023)",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 21,
+      "text": "Rigger, M., Su, Z.: Detecting optimization bugs in database engines via non-optimizing refer-ence engine construction. In: Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. Virtual Event USA: ACM, pp. 1140–1152 (2020)",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec"
+      ]
+    },
+    {
+      "number": 22,
+      "text": "Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA): 1–30 (2020)",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQ Lancer",
+      "technique": null,
+      "sentence": "In 2020, Manuel Rigger and Zhendong Su proposed SQ Lancer [ 20–22], a fuzzy testing tool that integrates a total of three methods, TLP, PQS and NoREC, which are combined to detect database logic errors.",
+      "context_before": "z testing for the security of Database Management Systems is also advancing [ 19]. Generation-based fuzzy testing tools, a common approach for DBMS security testing, require developers to construct accurate models of specific database SQL queries. However, these tools face efficiency issues due to the infinite input space, making bug discovery in DBMS quite challenging. DBMS, with their stringent syntactic and semantic input requirements, only accept test cases that meet specific standards, necessitating both syntactically and semantically correct SQL queries for effective vulnerability discovery.",
+      "context_after": "The TLP method partitions the SQL query into multiple segments and compares the concatenation of the original input results with the results from each segment post-partitioning to detect logical errors. The PQS method selects a row in the database, generates a query that can obtain that row, then generates a random expression that is true in the WHERE clause, checks the results, and if the row does not exist in the results, then there is a logical error in the database. The NoREC method changes the query to a form that cannot be optimized by morphing the SQL query into a logical equivalent an",
+      "section": "2 Background",
+      "page": 3,
+      "char_offset": 8918,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "pqs",
+        "norec",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The TLP method partitions the SQL query into multiple segments and compares the concatenation of the original input results with the results from each segment post-partitioning to detect logical errors.",
+      "context_before": "ate models of specific database SQL queries. However, these tools face efficiency issues due to the infinite input space, making bug discovery in DBMS quite challenging. DBMS, with their stringent syntactic and semantic input requirements, only accept test cases that meet specific standards, necessitating both syntactically and semantically correct SQL queries for effective vulnerability discovery. In 2020, Manuel Rigger and Zhendong Su proposed SQ Lancer [ 20–22], a fuzzy testing tool that integrates a total of three methods, TLP, PQS and NoREC, which are combined to detect database logic errors.",
+      "context_after": "The PQS method selects a row in the database, generates a query that can obtain that row, then generates a random expression that is true in the WHERE clause, checks the results, and if the row does not exist in the results, then there is a logical error in the database. The NoREC method changes the query to a form that cannot be optimized by morphing the SQL query into a logical equivalent and then compares the difference between the two results. The disadvantage of SQ Lancer is that it cannot test queries that do not return a result indefinitely. In 2020, Rui Zhong [ 23] et al. proposed Squi",
+      "section": "2 Background",
+      "page": 3,
+      "char_offset": 9121,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "The PQS method selects a row in the database, generates a query that can obtain that row, then generates a random expression that is true in the WHERE clause, checks the results, and if the row does not exist in the results, then there is a logical error in the database.",
+      "context_before": "and semantic input requirements, only accept test cases that meet specific standards, necessitating both syntactically and semantically correct SQL queries for effective vulnerability discovery. In 2020, Manuel Rigger and Zhendong Su proposed SQ Lancer [ 20–22], a fuzzy testing tool that integrates a total of three methods, TLP, PQS and NoREC, which are combined to detect database logic errors. The TLP method partitions the SQL query into multiple segments and compares the concatenation of the original input results with the results from each segment post-partitioning to detect logical errors.",
+      "context_after": "The NoREC method changes the query to a form that cannot be optimized by morphing the SQL query into a logical equivalent and then compares the difference between the two results. The disadvantage of SQ Lancer is that it cannot test queries that do not return a result indefinitely. In 2020, Rui Zhong [ 23] et al. proposed Squirrel, a variation-based database fuzzy testing tool, which is a AFL-based gray-box database fuzzy testing tool, 324 H. Zhu et al. which adopts techniques such as syntax-preserving mutation, semantic bootstrap instantiation, and code coverage feedback. Compared with previo",
+      "section": "2 Background",
+      "page": 3,
+      "char_offset": 9324,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "The NoREC method changes the query to a form that cannot be optimized by morphing the SQL query into a logical equivalent and then compares the difference between the two results.",
+      "context_before": "esting tool that integrates a total of three methods, TLP, PQS and NoREC, which are combined to detect database logic errors. The TLP method partitions the SQL query into multiple segments and compares the concatenation of the original input results with the results from each segment post-partitioning to detect logical errors. The PQS method selects a row in the database, generates a query that can obtain that row, then generates a random expression that is true in the WHERE clause, checks the results, and if the row does not exist in the results, then there is a logical error in the database.",
+      "context_after": "The disadvantage of SQ Lancer is that it cannot test queries that do not return a result indefinitely. In 2020, Rui Zhong [ 23] et al. proposed Squirrel, a variation-based database fuzzy testing tool, which is a AFL-based gray-box database fuzzy testing tool, 324 H. Zhu et al. which adopts techniques such as syntax-preserving mutation, semantic bootstrap instantiation, and code coverage feedback. Compared with previous fuzzy testing tools that lack a bootstrap mechanism, it significantly improves the ability of detecting database vulnerabilities as well as the code coverage rate, but Squirrel’s",
+      "section": "2 Background",
+      "page": 3,
+      "char_offset": 9596,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQ Lancer",
+      "technique": null,
+      "sentence": "The disadvantage of SQ Lancer is that it cannot test queries that do not return a result indefinitely.",
+      "context_before": "segments and compares the concatenation of the original input results with the results from each segment post-partitioning to detect logical errors. The PQS method selects a row in the database, generates a query that can obtain that row, then generates a random expression that is true in the WHERE clause, checks the results, and if the row does not exist in the results, then there is a logical error in the database. The NoREC method changes the query to a form that cannot be optimized by morphing the SQL query into a logical equivalent and then compares the difference between the two results.",
+      "context_after": "In 2020, Rui Zhong [ 23] et al. proposed Squirrel, a variation-based database fuzzy testing tool, which is a AFL-based gray-box database fuzzy testing tool, 324 H. Zhu et al. which adopts techniques such as syntax-preserving mutation, semantic bootstrap instantiation, and code coverage feedback. Compared with previous fuzzy testing tools that lack a bootstrap mechanism, it significantly improves the ability of detecting database vulnerabilities as well as the code coverage rate, but Squirrel’s mutation operation still generates syntactically incorrect statements and can only handle dependencie",
+      "section": "2 Background",
+      "page": 3,
+      "char_offset": 9776,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:04Z",
+    "is_model_written": true,
+    "summary": "This work fuzzes DBMSs for configuration errors -- faults that appear only under particular settings, which testing at default configuration never reaches. It surveys existing logic-bug fuzzers as background before setting out its own approach to varying configuration during testing.",
+    "narrative": "SQLancer is described as background: a fuzz testing tool integrating TLP, PQS and NoREC to detect logic errors, with each of the three oracles explained in turn. The paper notes one limitation as its point of departure -- that SQLancer cannot test queries which do not return a result -- but no comparison is run and nothing of SQLancer's is reused.",
+    "roles": {
+      "M1": "definition",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "definition",
+      "M5": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1007_s10664_025_10662_w.json b/_data/papers/paper_doi_10_1007_s10664_025_10662_w.json
new file mode 100644
index 0000000..fc7586d
--- /dev/null
+++ b/_data/papers/paper_doi_10_1007_s10664_025_10662_w.json
@@ -0,0 +1,192 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:53:41Z",
+  "paper": {
+    "id": "paper:doi:10.1007/s10664-025-10662-w",
+    "title": "Detecting data manipulation errors in android applications using scene-guided exploration",
+    "authors": [
+      "Shuqi Liu",
+      "Yu Zhou",
+      "Wenhua Yang",
+      "Taolue Chen",
+      "Harald C. Gall"
+    ],
+    "year": 2025,
+    "venue": "Empirical Software Engineering",
+    "doi": "10.1007/s10664-025-10662-w",
+    "arxiv_id": null,
+    "s2_paper_id": "fcc98415c9f0ae0b33b88b84d20d09969c757cca",
+    "url": "https://doi.org/10.1007/s10664-025-10662-w",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1007/s10664-025-10662-w",
+      "retrieved_at": "2026-09-08T14:53:41Z",
+      "chars": 110232,
+      "content_sha256": "sha256:b5a9a4294c8883e6c32c84607a5d7214b66cb31b96a271c3bbc1b78c22b7e489"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 38,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2044
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 10725
+      },
+      {
+        "number": "2.1",
+        "title": "Android UI Layout and Event",
+        "start": 10738
+      },
+      {
+        "number": "2.2",
+        "title": "Model-based Properties",
+        "start": 14269
+      },
+      {
+        "number": "3",
+        "title": "Approach",
+        "start": 17523
+      },
+      {
+        "number": "3.1",
+        "title": "DMF Instantiation",
+        "start": 19034
+      },
+      {
+        "number": "3.2",
+        "title": "SceneTG",
+        "start": 23357
+      },
+      {
+        "number": "3.3",
+        "title": "Scene-driven Exploration",
+        "start": 35176
+      },
+      {
+        "number": "4",
+        "title": "Experiment Design",
+        "start": 47988
+      },
+      {
+        "number": "4.1",
+        "title": "Research Questions and Hypotheses",
+        "start": 48743
+      },
+      {
+        "number": "4.2",
+        "title": "Datasets",
+        "start": 51159
+      },
+      {
+        "number": "4.3",
+        "title": "Experiment Setup",
+        "start": 56119
+      },
+      {
+        "number": "5",
+        "title": "Experimental Results",
+        "start": 62775
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 84296
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 93148
+      },
+      {
+        "number": "8",
+        "title": "Conclusion and Future Work",
+        "start": 98214
+      }
+    ]
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "author_year_citation",
+      "surface": "Rigger and Su 2020a",
+      "technique": null,
+      "sentence": "Existingstudies (Rigger and Su 2020a ,bhave detected CRUD errors in database management systems, but these approaches are not specifically tailored for Android apps.",
+      "context_before": "ration motivates the design of SceneData. Unlike traditional approaches, SceneData does not limit the implementation of a functionality to the initial state in the specified path. Instead, it recognizes the GUI scene and explores thestate transition details to identify whether a DMF can be executed in the scene. Although theaforementioned works are effective in detecting non-crashing functional bugs, they are notcapable of detecting DMEs. Finding Data Manipulation-Related Bugs Our work focuses on testing data manipulationrelated behavior, specifically targeting CRUD related bugs in the software.",
+      "context_after": "PBGT Costa et al.2014 ) utilizes the concept of User Interface Test Patterns (UITPs) to test common recurrent behaviors in apps through their possible different implementations. However, it onlymodels the Find operation as a UITP to check correctness, without considering other relateddata manipulations. Augusto (Mariani et al. 2018 ) leverages semantic knowledge related to CRUD operation functionalities to automatically generate effective test cases with functionaloracles. It systematically covers critical scenarios and detects failures by encoding expectedfunctionality into models customized",
+      "section": "7 Related Work",
+      "page": 34,
+      "char_offset": 96111,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:02:36Z",
+    "is_model_written": true,
+    "summary": "This work finds data manipulation errors in Android applications by exploring the app through its user interface, guided by scenes -- coherent groups of screens -- so that create, read, update and delete operations are exercised in realistic sequences rather than at random.",
+    "narrative": "The single mention places SQLancer's work as the database-side precedent: the paper notes that existing studies have detected CRUD errors in database management systems, but that those approaches are not tailored for Android apps, where the errors surface through the application's own data layer rather than through SQL. The citation is author-year, so the tool is not named directly.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1016_j_cose_2025_104564.json b/_data/papers/paper_doi_10_1016_j_cose_2025_104564.json
new file mode 100644
index 0000000..05c57c7
--- /dev/null
+++ b/_data/papers/paper_doi_10_1016_j_cose_2025_104564.json
@@ -0,0 +1,1279 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:08:54Z",
+  "paper": {
+    "id": "paper:doi:10.1016/j.cose.2025.104564",
+    "title": "Detecting DBMS bugs with context-sensitive instantiation and multi-plan execution",
+    "authors": [
+      "Jiaqi Li",
+      "Ke Wang",
+      "Yaoguang Chen",
+      "Yajin Zhou",
+      "Lei Wu",
+      "Jiashui Wang"
+    ],
+    "year": 2025,
+    "venue": "Computers & Security",
+    "doi": "10.1016/j.cose.2025.104564",
+    "arxiv_id": "2312.04941v1",
+    "s2_paper_id": "941500a2f13307898e0f0152e4bde173d51dc7bb",
+    "url": "https://doi.org/10.1016/j.cose.2025.104564",
+    "also_indexed_as": [
+      "paper:arxiv:2312.04941"
+    ]
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2312.04941v1",
+      "retrieved_at": "2026-09-10T15:08:54Z",
+      "chars": 79515,
+      "content_sha256": "sha256:248fe572e5e4aecc30d80568219c0f72d60b3043d6f5f13d8e2dcb66c6d1d808"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1552
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 8671
+      },
+      {
+        "number": "2.1",
+        "title": "Structured Query Language",
+        "start": 8685
+      },
+      {
+        "number": "2.2",
+        "title": "SQL Query Processing",
+        "start": 9545
+      },
+      {
+        "number": "2.3",
+        "title": "Levels of DBMS Testing",
+        "start": 11464
+      },
+      {
+        "number": "2.4",
+        "title": "DBMS Testing",
+        "start": 13493
+      },
+      {
+        "number": "3",
+        "title": "Motivating Examples",
+        "start": 15688
+      },
+      {
+        "number": "3.1",
+        "title": "Context-Sensitive Instantiation",
+        "start": 15819
+      },
+      {
+        "number": "3.2",
+        "title": "Multi-Plan Execution",
+        "start": 21475
+      },
+      {
+        "number": "4",
+        "title": "Design",
+        "start": 27665
+      },
+      {
+        "number": "4.1",
+        "title": "Overall Design",
+        "start": 27675
+      },
+      {
+        "number": "4.2",
+        "title": "Parser Generation",
+        "start": 32631
+      },
+      {
+        "number": "4.3",
+        "title": "Parse Tree Mutation",
+        "start": 33861
+      },
+      {
+        "number": "4.4",
+        "title": "Context-Sensitive Instantiation",
+        "start": 34810
+      },
+      {
+        "number": "4.5",
+        "title": "Multi-Plan Execution",
+        "start": 43977
+      },
+      {
+        "number": "4.6",
+        "title": "Prototype Implementation",
+        "start": 48506
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 49668
+      },
+      {
+        "number": "5.1",
+        "title": "Detecting Bugs in Real-world DBMSs",
+        "start": 50908
+      },
+      {
+        "number": "5.2",
+        "title": "Generating Valid Queries",
+        "start": 51640
+      },
+      {
+        "number": "5.3",
+        "title": "Comparisons with Existing Tools",
+        "start": 56798
+      },
+      {
+        "number": "5.4",
+        "title": "Benefits of the Two Key Techniques",
+        "start": 60432
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 63308
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 65702
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 69671
+      },
+      {
+        "number": "Z",
+        "title": "Chen, “Collafl: Path sensitive fuzzing,” in. IEEE,",
+        "start": 74735
+      },
+      {
+        "number": "S",
+        "title": "Khurshid, “Query-aware test generation using a",
+        "start": 75090
+      },
+      {
+        "number": "T",
+        "title": "Wei, and L. Lu, “Savior: Towards bug-driven hybrid",
+        "start": 76145
+      },
+      {
+        "number": "J",
+        "title": "Cappos, M. Schl ¨ogel, N. Korshun, A. Abbasi,",
+        "start": 76487
+      },
+      {
+        "number": "M",
+        "title": "Schweighauser, S. Schinzel, S. Schumilo et al.,",
+        "start": 76536
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Well-known users of sqlite,” https://www.sqlite.org/f amous.html, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“Mysql customers,” https://www.mysql.com/customer s/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "A. Bannister, “Sqlite patches use-after-free bug that left apps open to code execution, denial-of-service exploits.” https://portswigger.net/daily-swig/sqlite-pat ches-use-after-free-bug-that-left-apps-open-to-code-e xecution-denial-of-service-exploits, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "C. Cimpanu, “Google chrome impacted by new magellan 2.0 vulnerabilities.” https://www.zdnet.com/article/google-chrome-imp acted-by-new-magellan-2-0-vulnerabilities/, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “Apollo: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment, vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing database management systems with language validity and coverage feedback,” inProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "S. M. Andreas Seltenreich, Bo Tang, “Sqlsmith,” https: //github.com/anse1/sqlsmith/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Y. Chen, R. Zhong, H. Hu, H. Zhang, Y. Yang, D. Wu, and W. Lee, “One engine to fuzz’em all: Generic language processor testing with semantic validation,” in (SP). IEEE, 2021, pp. 642–658.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "X. Liu, Q. Zhou, J. Arulraj, and A. Orso, “Automated performance bug detection in database systems,” arXiv e-prints, pp. arXiv–2105, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Y. Zhang, P. Yao, R. Wu, and C. Zhang, “Duplicatesensitivity guided transformation synthesis for dbms correctness bug detection,” arXiv preprint arXiv:2107.03660, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "B. Ghit, N. Poggi, J. Rosen, R. Xin, and P. Boncz, “Sparkfuzz: Searching correctness regressions in modern query engines,” in Proceedings of the workshop on Testing Database Systems, 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "“Addresssanitizer,” https://github.com/google/sanitize rs/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of{DBMS }with coverage-based guidance,” in 31st USENIX Security Symposium (USENIX Security 22), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "——, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 16,
+      "text": "——, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 17,
+      "text": "X. Tang, S. Wu, D. Zhang, F. Li, and G. Chen, “Detecting logic bugs of join optimizations in dbms,” Proceedings of the ACM on Management of Data, vol. 1, no. 1, pp. 1–26, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "“Sqlite homepage,” https://www.sqlite.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "“Postgresql homepage,” https://www.postgresql.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "“Mysql homepage,” https://www.mysql.com/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "“Common vulnerabilities and exposures,” https://en.wikipedia.org/wiki/Common Vulnerabili ties and Exposures, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "D. D. Chamberlin and R. F. Boyce, “Sequel: A structured english query language,” in Proceedings of the 1974 ACMSIGFIDET (now SIGMOD) workshop on Data description, access and control, 1974, pp. 249–264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "“American Fuzzy Lop (2.56b),” https://lcamtuf.coredu mp.cx/afl/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "M. Sutton, A. Greene, and P. Amini, Fuzzing: brute force vulnerability discovery. Pearson Education, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "J. Neystadt, “Automated penetration testing with whitebox fuzzing,” MSDN Library, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "J. Wang, B. Chen, L. Wei, and Y. Liu, “Superion: Grammar-aware greybox fuzzing,” in 2019 IEEE/ACM 41st International Conference on Software Engineering (ICSE). IEEE, 2019, pp. 724–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "D. R. Slutz, “Massive stochastic testing of sql,” in VLDB, vol. 98. Citeseer, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "“Bugs found in database management systems,” https: //www.manuelrigger.at/dbms-bugs/, 2022.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 29,
+      "text": "“Unit testing,” https://en.wikipedia.org/wiki/Unit testi ng, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "“Randomized depth-first search,” https://en.wikipedia .org/wiki/Maze generation algorithm, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "D. E. Knuth, “Backus normal form vs. backus naur form,” Communications of the ACM, vol. 7, no. 12, pp. 735–736, 1964.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "“Constraint satisfaction problem,” https://en.wikipedia .org/wiki/Constraint satisfaction problem, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "S. Gan, C. Zhang, X. Qin, X. Tu, K. Li, Z. Pei, and Z. Chen, “Collafl: Path sensitive fuzzing,” in. IEEE, 2018, pp. 679–696.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "“How sqlite is tested,” https://www.sqlite.org/testing.h tml, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "J. Postel et al., “Transmission control protocol,” p. 13, 1981.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "“Proof of concept,” https://en.wikipedia.org/wiki/Proo fofconcept#Security, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "S. A. Khalek, B. Elkarablieh, Y. O. Laleye, and S. Khurshid, “Query-aware test generation using a relational constraint solver,” in 2008 23rd IEEE/ACM International Conference on Automated Software Engineering. IEEE, 2008, pp. 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "C. Binnig, D. Kossmann, E. Lo, and M. T. ¨Ozsu, “Qagen: generating query-aware test databases,” in Proceedings of the 2007 ACMSIGMOD international conference on Management of data, 2007, pp. 341– 352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "“Alloy,” https://alloytools.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "N. Stephens, J. Grosen, C. Salls, A. Dutcher, R. Wang, J. Corbetta, Y. Shoshitaishvili, C. Kruegel, and G. Vigna, “Driller: Augmenting fuzzing through selective symbolic execution.” in NDSS, vol. 16, no. 2016, 2016, pp. 1–16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "I. Yun, S. Lee, M. Xu, Y. Jang, and T. Kim, “ {QSYM }: A practical concolic execution engine tailored for hybrid fuzzing,” in 27th USENIX Security Symposium (USENIX Security 18), 2018, pp. 745–761.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "P. Chen and H. Chen, “Angora: Efficient fuzzing by principled search,” in rity and Privacy (SP). IEEE, 2018, pp. 711–725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Y. Chen, P. Li, J. Xu, S. Guo, R. Zhou, Y. Zhang, T. Wei, and L. Lu, “Savior: Towards bug-driven hybrid testing,” in Privacy (SP). IEEE, 2020, pp. 1580–1596.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "S. Gan, C. Zhang, P. Chen, B. Zhao, X. Qin, D. Wu, and Z. Chen, “ {GREYONE }: Data flow sensitive fuzzing,” in 29th USENIX Security Symposium (USENIX Security 20), 2020, pp. 2577–2594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "T. Blazytko, M. Bishop, C. Aschermann, J. Cappos, M. Schl ¨ogel, N. Korshun, A. Abbasi, M. Schweighauser, S. Schinzel, S. Schumilo et al., “{GRIMOIRE }: Synthesizing structure while fuzzing,” in28th USENIX Security Symposium (USENIX Security 19), 2019, pp. 1985–2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "H. Bati, L. Giakoumakis, S. Herbert, and A. Surna, “A genetic approach for random testing of database systems,” in Proceedings of the 33rd international conference on Very large data bases, 2007, pp. 1243– 1251.Appendix A. PoC Generation Strategies MPE can only conduct the PoC on modified DBMSs. To build PoC on unmodified DBMSs, our goal is to make the buggy query plan to be optimal. The ways to a",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 15,
+      "text": "——, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 16,
+      "text": "——, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 28,
+      "text": "“Bugs found in database management systems,” https: //www.manuelrigger.at/dbms-bugs/, 2022.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[13–17]",
+      "technique": "pqs",
+      "sentence": "Second, oracles used by existing systems [13–17] to detect logic bugs have strict requirements on the SQL statements.",
+      "context_before": "ion of a test case, they only consider limited semantic constraints. As a result, they will generate invalid SQL statements that cannot pass the semantic check of a DBMS. For instance, Squirrel [6] proposed semantics-guided instantiation and the follow-up work SQLRight [13] proposed Context-based IR instantiation to improve the semantic correctness of generated SQL statements. For simplicity, we will refer to their approaches as type-sensitive instantiation in this paper since they utilize a context-free strategy that only considers the correctness of the identifier type (e.g., table, column).",
+      "context_after": "This makes the fuzzing system only explore a narrow space of inputs, leading to limited bugs that can be detected. For example, NoREC [15] requires that the effective SQL query in a test case has WHERE clauses, thus it can only detect logic bugs due to the optimization of WHERE clauses. Another recent work, TQS [17], focusing on detecting logic bugs caused by equal-join optimization, fails to identify bugs arising from other types of optimization. Furthermore, both NoREC and TQS attempt to explore multiple query plans of queries but are only capable of covering partial query plans. Hence, an",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3923,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For example, NoREC [15] requires that the effective SQL query in a test case has WHERE clauses, thus it can only detect logic bugs due to the optimization of WHERE clauses.",
+      "context_before": "tion and the follow-up work SQLRight [13] proposed Context-based IR instantiation to improve the semantic correctness of generated SQL statements. For simplicity, we will refer to their approaches as type-sensitive instantiation in this paper since they utilize a context-free strategy that only considers the correctness of the identifier type (e.g., table, column). Second, oracles used by existing systems [13–17] to detect logic bugs have strict requirements on the SQL statements. This makes the fuzzing system only explore a narrow space of inputs, leading to limited bugs that can be detected.",
+      "context_after": "Another recent work, TQS [17], focusing on detecting logic bugs caused by equal-join optimization, fails to identify bugs arising from other types of optimization. Furthermore, both NoREC and TQS attempt to explore multiple query plans of queries but are only capable of covering partial query plans. Hence, an oracle that can be applied to SQL statements without strict requirements and explore more query plans is needed. Our Solution This work proposes two key techniques to address the limitations and solve the two innate challenges. The first key technique is called context-sensitive instanti",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 4156,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Furthermore, both NoREC and TQS attempt to explore multiple query plans of queries but are only capable of covering partial query plans.",
+      "context_before": "er type (e.g., table, column). Second, oracles used by existing systems [13–17] to detect logic bugs have strict requirements on the SQL statements. This makes the fuzzing system only explore a narrow space of inputs, leading to limited bugs that can be detected. For example, NoREC [15] requires that the effective SQL query in a test case has WHERE clauses, thus it can only detect logic bugs due to the optimization of WHERE clauses. Another recent work, TQS [17], focusing on detecting logic bugs caused by equal-join optimization, fails to identify bugs arising from other types of optimization.",
+      "context_after": "Hence, an oracle that can be applied to SQL statements without strict requirements and explore more query plans is needed. Our Solution This work proposes two key techniques to address the limitations and solve the two innate challenges. The first key technique is called context-sensitive instantiation. It performs context-sensitive analysis to collect all static semantic constraints (including but not limited to the identifier type) to improve the semantic correctnessarXiv:2312.04941v1 [cs.DB] 8 Dec 2023 of generated inputs. For instance, a SQL statement SELECT x1+ i1 FROM x2 WHERE x3=x4 has",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 4493,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We also compared Kangaroo with leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight.",
+      "context_before": "Prototype and Evaluation In this study, we implemented a prototype named Kangaroo and applied it to three widelyused database management systems (DBMSs): SQLite [18], PostgreSQL [19], and MySQL [20], to evaluate its effectiveness. Despite the fact that state-of-the-art tools have already extensively tested these DBMSs, Kangaroo successfully identified 50 new bugs, consisting of 9 logic bugs, 19 crash-causing bugs, and 22 assertion failure-inducing bugs. As of the time of writing, 28 of these bugs have been fixed, with 11 assigned CVEs [21]. These results demonstrate the efficacy of our system.",
+      "context_after": "Our evaluation reveals that Kangaroo surpasses these tools in terms of generating semantically valid SQL queries, exploring code paths, and detecting bugs. Specifically, after conducting a 24-hour test on the three DBMSs, Kangaroo successfully detected 17 bugs, while SQLancer, Squirrel, and SQLRight only identified 1, 3, and 6 bugs, respectively. In addition, Kangaroo proved more effective in generating valid SQL queries, achieving a 1.6x-1.9x improvement compared to the mutation-based tool Squirrel, and it explored 1.16x-1.3x more program states than the rule-based tool SQLancer. This work m",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7241,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, after conducting a 24-hour test on the three DBMSs, Kangaroo successfully detected 17 bugs, while SQLancer, Squirrel, and SQLRight only identified 1, 3, and 6 bugs, respectively.",
+      "context_before": "e-of-the-art tools have already extensively tested these DBMSs, Kangaroo successfully identified 50 new bugs, consisting of 9 logic bugs, 19 crash-causing bugs, and 22 assertion failure-inducing bugs. As of the time of writing, 28 of these bugs have been fixed, with 11 assigned CVEs [21]. These results demonstrate the efficacy of our system. We also compared Kangaroo with leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight. Our evaluation reveals that Kangaroo surpasses these tools in terms of generating semantically valid SQL queries, exploring code paths, and detecting bugs.",
+      "context_after": "In addition, Kangaroo proved more effective in generating valid SQL queries, achieving a 1.6x-1.9x improvement compared to the mutation-based tool Squirrel, and it explored 1.16x-1.3x more program states than the rule-based tool SQLancer. This work makes the following main contributions. •We revealed the challenges of effectively detecting DBMS bugs and the limitations of existing solutions. •We proposed two key techniques to solve the challenges, including context-sensitive instantiation to improve the semantic correctness of the mutated SQL queries and the MPE that can be applied to differen",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7498,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3x more program states than the rule-based tool SQLancer.",
+      "context_before": "leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight. Our evaluation reveals that Kangaroo surpasses these tools in terms of generating semantically valid SQL queries, exploring code paths, and detecting bugs. Specifically, after conducting a 24-hour test on the three DBMSs, Kangaroo successfully detected 17 bugs, while SQLancer, Squirrel, and SQLRight only identified 1, 3, and 6 bugs, respectively. In addition, Kangaroo proved more effective in generating valid SQL queries, achieving a 1.6x-1.9x improvement compared to the mutation-based tool Squirrel, and it explored 1.16x-1.",
+      "context_after": "This work makes the following main contributions. •We revealed the challenges of effectively detecting DBMS bugs and the limitations of existing solutions. •We proposed two key techniques to solve the challenges, including context-sensitive instantiation to improve the semantic correctness of the mutated SQL queries and the MPE that can be applied to different types of SQL queries to detect logic bugs. •We implemented and applied a prototype system to three popular DBMSs. Our system successfully detected 50 new bugs. The micro-benchmark also shows our system outperforms existing ones in genera",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7873,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[14–17, 24]",
+      "technique": "pqs",
+      "sentence": "The rule-based ones [14–17, 24] generate test cases following a predefined model to ensurethe generated SQL queries can pass the SQL parser.",
+      "context_before": ", Squirrel has to limit the complexity of generated queries to tolerate its incomplete and inaccurate constraints. What’s worse, the complexity of constraints dramatically increases with the complexity of the statement and the strictness of the DBMS checks. To better explore the core modules of DBMS, we propose a new method to generate more diverse and complex test cases, while at the same time, they can be statically confirming (level-7). 2.4. DBMS Testing Test Case Generation DBMS testing aims to trigger bugs by constructing abundant test cases. There are two methods to generate SQL queries.",
+      "context_after": "However, building a precise model requires domain knowledge. Besides, the generated inputs cannot efficiently explore the program’s state space since it wastes much effort on similar queries. The mutation-based method [6, 13, 25] generates new test cases by mutating seed queries. However, the general mutation strategies, like flipping bits, can not generate valid SQL queries that can pass the SQL parser. To address this issue, a common approach is to perform mutations based on grammar rules [6, 26]. It first generates a syntax tree for a given input, then creates specific mutations by using m",
+      "section": "2.4 DBMS Testing",
+      "page": 3,
+      "char_offset": 13649,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [15] is an oracle for logic bug detection.",
+      "context_before": "n oracle is a mechanism for determining whether the actual outputs match the expected outcomes. Differential testing uses different implementations of the same functionality as cross-referencing oracles. It provides the same inputs to a series of similar systems and then obverses the results. Any inconsistency between the results may indicate a potential bug. RAGS [27] is the first work that applies differential testing to find logic bugs for DBMS. Metamorphic testing addresses the test oracle problem based on the observation that a transformation of the input has a known effect on the output.",
+      "context_after": "It translates a query that is potentially optimized by DBMS to a query that can hardly be optimized. Although this approach has been effective in detecting bugs in widely-used DBMS [28], it can only be applied to a subset of SQL that can be translated. Our system leverages the idea of differential testing to compare the execution results of multiple query plans, hence named multi-plan execution (MPE). 3. Motivating Examples In this section, we use two real examples to demonstrate the advantages of our system’s two key techniques. 3.1. Context-Sensitive Instantiation We propose context-sensiti",
+      "section": "2.4 DBMS Testing",
+      "page": 3,
+      "char_offset": 15233,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[28]",
+      "technique": null,
+      "sentence": "Although this approach has been effective in detecting bugs in widely-used DBMS [28], it can only be applied to a subset of SQL that can be translated.",
+      "context_before": "the same functionality as cross-referencing oracles. It provides the same inputs to a series of similar systems and then obverses the results. Any inconsistency between the results may indicate a potential bug. RAGS [27] is the first work that applies differential testing to find logic bugs for DBMS. Metamorphic testing addresses the test oracle problem based on the observation that a transformation of the input has a known effect on the output. NoREC [15] is an oracle for logic bug detection. It translates a query that is potentially optimized by DBMS to a query that can hardly be optimized.",
+      "context_after": "Our system leverages the idea of differential testing to compare the execution results of multiple query plans, hence named multi-plan execution (MPE). 3. Motivating Examples In this section, we use two real examples to demonstrate the advantages of our system’s two key techniques. 3.1. Context-Sensitive Instantiation We propose context-sensitive instantiation because of two observations. First, previous works [6, 13] have proven the importance of semantic correctness to fuzzing efficiency, but existing type-sensitive instantiation still has large room for improvement. Second, MPE requires st",
+      "section": "2.4 DBMS Testing",
+      "page": 3,
+      "char_offset": 15383,
+      "cited_reference": {
+        "number": 28,
+        "text": "“Bugs found in database management systems,” https: //www.manuelrigger.at/dbms-bugs/, 2022.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "proposed three oracles for DBMS logic bug detection, including PQS [14], NoREC [15], and TLP [16], all of them put limitations on the SQL queries.",
+      "context_before": ". In TableReference x2 JOIN x3, columns x4and x5, which belong to the join constraint of the joined table, can only be the column of the joined table. Besides, each column should have a unique name within its dependent table. These additional constraints guarantee the semantic correctness of this SELECT statement that does not contain dynamic constraints. Enforcing dynamic constraints requires emulating the execution of statements, which is challenging. Thus, our semantic analysis ignores dynamic constraints and leaves it as part of future work. 3.2. Multi-Plan Execution Although Rigger et al.",
+      "context_after": "Similarly, TQS [17], which aims at detecting logic bugs in equal-join optimization, also limits the diversity of SQL statements. In addition, both TQS and NoREC attempt to explore different query plans for queries. NoREC only compares two distinct plans by transforming queries into semantically similar ones. TQS uses DBMS-specific features such as optimization switches and hints to iterate different query plans. However, this approach is not general and only covers partial query plans. To this end, we propose MPE, which has no such limitation. The basic idea is that our system hooks into the",
+      "section": "3.2 Multi-Plan Execution",
+      "page": 4,
+      "char_offset": 21523,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "In addition, both TQS and NoREC attempt to explore different query plans for queries.",
+      "context_before": "c correctness of this SELECT statement that does not contain dynamic constraints. Enforcing dynamic constraints requires emulating the execution of statements, which is challenging. Thus, our semantic analysis ignores dynamic constraints and leaves it as part of future work. 3.2. Multi-Plan Execution Although Rigger et al. proposed three oracles for DBMS logic bug detection, including PQS [14], NoREC [15], and TLP [16], all of them put limitations on the SQL queries. Similarly, TQS [17], which aims at detecting logic bugs in equal-join optimization, also limits the diversity of SQL statements.",
+      "context_after": "NoREC only compares two distinct plans by transforming queries into semantically similar ones. TQS uses DBMS-specific features such as optimization switches and hints to iterate different query plans. However, this approach is not general and only covers partial query plans. To this end, we propose MPE, which has no such limitation. The basic idea is that our system hooks into the DBMS optimizer to execute all query plans and compare their results. If the result of one query plan is different from the others, a logic bug is detected. CREATE TABLE t0(c0 INT); CREATE TABLE t1(c1 INT, c2 INT); I",
+      "section": "3.2 Multi-Plan Execution",
+      "page": 4,
+      "char_offset": 21799,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC only compares two distinct plans by transforming queries into semantically similar ones.",
+      "context_before": "rcing dynamic constraints requires emulating the execution of statements, which is challenging. Thus, our semantic analysis ignores dynamic constraints and leaves it as part of future work. 3.2. Multi-Plan Execution Although Rigger et al. proposed three oracles for DBMS logic bug detection, including PQS [14], NoREC [15], and TLP [16], all of them put limitations on the SQL queries. Similarly, TQS [17], which aims at detecting logic bugs in equal-join optimization, also limits the diversity of SQL statements. In addition, both TQS and NoREC attempt to explore different query plans for queries.",
+      "context_after": "TQS uses DBMS-specific features such as optimization switches and hints to iterate different query plans. However, this approach is not general and only covers partial query plans. To this end, we propose MPE, which has no such limitation. The basic idea is that our system hooks into the DBMS optimizer to execute all query plans and compare their results. If the result of one query plan is different from the others, a logic bug is detected. CREATE TABLE t0(c0 INT); CREATE TABLE t1(c1 INT, c2 INT); INSER TINT O t0 VALUES(0); INSER TINT O t1 VALUES(0, 1); CREATE INDEX t1_c1 ON t1(c1); SELECT *",
+      "section": "3.2 Multi-Plan Execution",
+      "page": 4,
+      "char_offset": 21885,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Why Existing Works Cannot Detect the Bug NoREC is one of the most effective DBMS logic bug detection oracles which requires the SELECT statements to satisfy some predefined rules, e.",
+      "context_before": "hat triggers a logic bug in SQLite. There are four query plans for the mutated SELECT statement Q1. One of them (the fourth one) returns a non-empty result that is different from the others. In the following, we use the proof-of-concept of one real-world bug of SQLite in Figure 2 to illustrate why the existing system cannot explore enough program states to detect the logic bug and how the MPE can capture such a bug. When the test case is executed, the DBMS optimizer finds four query plans for the SELECT statement. Plan one is optimal among four query plans and is executed by SQLite by default.",
+      "context_after": "g., having a WHERE and FROM clause. It shifts the condition in WHERE clause to the SELECT_TARGET, and then executes both to compare the execution result. However, the SELECT statement Q1 in Figure 2 does not meet the requirement (lacks WHERE clause), thus missing the bug triggered by the test case. TQS relies on optimization switches and hints to iterate different query plans, which limits its exploration of query plans. First, the optimization switches cannot force optimization adoptions nor control the scope of optimizations. Hints enables more fine-grained control than optimization switches",
+      "section": "3.2 Multi-Plan Execution",
+      "page": 5,
+      "char_offset": 24049,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "citation_marker",
+      "surface": "[13–17]",
+      "technique": "pqs",
+      "sentence": "in previous works [13–17], since other statements typically lack returning results to be checked.",
+      "context_before": "2 Attribute INSERT INTO x1(x2) VALUES (i1); Attr(x2)!=GENERATED x2 should not be a generated column Value SELECT * FROM x1 ORDER BY i1; i1∈{1,2,...,Size (x1)} i1 is an integer up to the number of columns of table x1 Dependency SELECT x1 FROM (SELECT x2, x3 FROM x4) x5; x1∈{x2,x3} x1 could be column x2 or x3 Distinct INSERT INTO x1(x2, x3) ...; DISTINCT [x2,x3] x2 and x3 should be two different column Composite SELECT (x1, x2) IN (TABLE x3) FROM x4; (DataType (x1),DataType (x2)))∈The data type of x2 and x3 should be {DataType (x3,1),DataType (x3,2)}the same as the first two columns in table x3.",
+      "context_after": "Adopting MPE to DBMSs is not straightforward. The first challenge is how to make DBMS execute all query plans with a small modification. Our key observation is that DBMS typically employs a plan choose function to compare the estimated cost of different query plans (including their sub-plans) and choose the best one. Therefore, we can execute any query plan by hooking the plan choose function. In addition, we added a loop before the entry function for query processing so that it can be executed multiple times with different plans until all plans are executed. Such modification has two advanta",
+      "section": "4.5 Multi-Plan Execution",
+      "page": 9,
+      "char_offset": 45587,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We use NoREC for SQLancer and SQLRight.",
+      "context_before": "406080100Validity(%) (d) SQLite query validity0 4 8 12 16 20 240102030Validity(%) (e) PostgreSQL query validity0 4 8 12 16 20 240255075100Validity(%) (f) MySQL query validity 0 4 8 12 16 20 24024681012unique bugs (g) SQLite unique bugs0 4 8 12 16 20 240246810unique bugs (h) MySQL unique bugs SQLancer Squirrel Kangaroo SQLRight Squirrelmultiplan Kangaroo!multiplanFigure 6: Comparison with existing tools. It also illustrates the contributions of context-sensitive instantiation and multi-plan execution. We exclude the results of detected bugs in PostgreSQL since only Kangaroo found one memory bug.",
+      "context_after": "Since SQLancer does not implement NoREC for MySQL, we use TLP instead. TABLE 5: The percentage of semantic correctness of query validation. SQLite PostgreSQL MySQL Squirrel 24,792(53.1%) 5,869(17.9%) 16,283(15.1%) SQLRight 28,710(61.5%) 8,508(26.0%) 35,012(32.5%) Kangaroo 32,880(70.4%) 16,269(49.7%) 46,763(43.4%) Total 46,672 32,753 107,693 type mismatch error in PostgreSQL. 5.3. Comparisons with Existing Tools We compare Kangaroo with three state-of-the-art and open source systems: Squirrel, SQLancer, and SQLRight. Similar to the previous ones, we also select SQLite, PostgreSQL, and MySQL fo",
+      "section": "5.2 Generating Valid Queries",
+      "page": 11,
+      "char_offset": 56379,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Since SQLancer does not implement NoREC for MySQL, we use TLP instead.",
+      "context_before": "lidity0 4 8 12 16 20 240102030Validity(%) (e) PostgreSQL query validity0 4 8 12 16 20 240255075100Validity(%) (f) MySQL query validity 0 4 8 12 16 20 24024681012unique bugs (g) SQLite unique bugs0 4 8 12 16 20 240246810unique bugs (h) MySQL unique bugs SQLancer Squirrel Kangaroo SQLRight Squirrelmultiplan Kangaroo!multiplanFigure 6: Comparison with existing tools. It also illustrates the contributions of context-sensitive instantiation and multi-plan execution. We exclude the results of detected bugs in PostgreSQL since only Kangaroo found one memory bug. We use NoREC for SQLancer and SQLRight.",
+      "context_after": "TABLE 5: The percentage of semantic correctness of query validation. SQLite PostgreSQL MySQL Squirrel 24,792(53.1%) 5,869(17.9%) 16,283(15.1%) SQLRight 28,710(61.5%) 8,508(26.0%) 35,012(32.5%) Kangaroo 32,880(70.4%) 16,269(49.7%) 46,763(43.4%) Total 46,672 32,753 107,693 type mismatch error in PostgreSQL. 5.3. Comparisons with Existing Tools We compare Kangaroo with three state-of-the-art and open source systems: Squirrel, SQLancer, and SQLRight. Similar to the previous ones, we also select SQLite, PostgreSQL, and MySQL for evaluation. We feed the same test cases to Squirrel, SQLRight, and Ka",
+      "section": "5.2 Generating Valid Queries",
+      "page": 11,
+      "char_offset": 56419,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Comparisons with Existing Tools We compare Kangaroo with three state-of-the-art and open source systems: Squirrel, SQLancer, and SQLRight.",
+      "context_before": "es the contributions of context-sensitive instantiation and multi-plan execution. We exclude the results of detected bugs in PostgreSQL since only Kangaroo found one memory bug. We use NoREC for SQLancer and SQLRight. Since SQLancer does not implement NoREC for MySQL, we use TLP instead. TABLE 5: The percentage of semantic correctness of query validation. SQLite PostgreSQL MySQL Squirrel 24,792(53.1%) 5,869(17.9%) 16,283(15.1%) SQLRight 28,710(61.5%) 8,508(26.0%) 35,012(32.5%) Kangaroo 32,880(70.4%) 16,269(49.7%) 46,763(43.4%) Total 46,672 32,753 107,693 type mismatch error in PostgreSQL. 5.3.",
+      "context_after": "Similar to the previous ones, we also select SQLite, PostgreSQL, and MySQL for evaluation. We feed the same test cases to Squirrel, SQLRight, and Kangaroo as the initial corpus and provide the same queries used to initialize the mutation library. SQLancer is a generate-based tool that does not require any initial inputs. We launch five fuzzing instances for each system and run each instance for 24 hours. We report the average result except for the bug number. We collect all bug reports from the five fuzzing instances as the final result and then count their first occurrence time for each uniq",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 56802,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer is a generate-based tool that does not require any initial inputs.",
+      "context_before": "rrel 24,792(53.1%) 5,869(17.9%) 16,283(15.1%) SQLRight 28,710(61.5%) 8,508(26.0%) 35,012(32.5%) Kangaroo 32,880(70.4%) 16,269(49.7%) 46,763(43.4%) Total 46,672 32,753 107,693 type mismatch error in PostgreSQL. 5.3. Comparisons with Existing Tools We compare Kangaroo with three state-of-the-art and open source systems: Squirrel, SQLancer, and SQLRight. Similar to the previous ones, we also select SQLite, PostgreSQL, and MySQL for evaluation. We feed the same test cases to Squirrel, SQLRight, and Kangaroo as the initial corpus and provide the same queries used to initialize the mutation library.",
+      "context_after": "We launch five fuzzing instances for each system and run each instance for 24 hours. We report the average result except for the bug number. We collect all bug reports from the five fuzzing instances as the final result and then count their first occurrence time for each unique bug. Figure 6 shows the evaluation result. Detected Unique Bugs As indicated in Figure 6gh, Kangaroo outperforms existing tools on all three DBMSs. For SQLite, Kangaroo found 12 unique bugs, including eight memory bugs and four logic bugs. SQLRight found three bugs, one of which is a logic bug. Squirrel found only oneb",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 57188,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Squirrel and SQLancer detected five and one memory bugs, respectively.",
+      "context_before": "Bugs As indicated in Figure 6gh, Kangaroo outperforms existing tools on all three DBMSs. For SQLite, Kangaroo found 12 unique bugs, including eight memory bugs and four logic bugs. SQLRight found three bugs, one of which is a logic bug. Squirrel found only onebug which is also covered by Kangaroo. For PostgreSQL, only Kangaroo found one bug. This is not surprising as previous works have revealed that PostgreSQL is more robust than most other DBMSs. As for MySQL, Kangaroo found a total of ten bugs, including eight memory bugs and two logic bugs. SQLRight found two memory bugs and one logic bug.",
+      "context_after": "SQLancer, the only generationbased tool, found the least bugs across all comparisons, demonstrating the advantage of the mutation-based method to detect DBMS bugs. Explored New Edges Figure 6abc shows that Kangaroo performs better than others on all three DBMS systems. It explores 52%, 44%, and 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively. Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive insta",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 58203,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer, the only generationbased tool, found the least bugs across all comparisons, demonstrating the advantage of the mutation-based method to detect DBMS bugs.",
+      "context_before": "all three DBMSs. For SQLite, Kangaroo found 12 unique bugs, including eight memory bugs and four logic bugs. SQLRight found three bugs, one of which is a logic bug. Squirrel found only onebug which is also covered by Kangaroo. For PostgreSQL, only Kangaroo found one bug. This is not surprising as previous works have revealed that PostgreSQL is more robust than most other DBMSs. As for MySQL, Kangaroo found a total of ten bugs, including eight memory bugs and two logic bugs. SQLRight found two memory bugs and one logic bug. Squirrel and SQLancer detected five and one memory bugs, respectively.",
+      "context_after": "Explored New Edges Figure 6abc shows that Kangaroo performs better than others on all three DBMS systems. It explores 52%, 44%, and 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively. Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive instantiation. Generated Valid Queries We treat a query as invalid if DBMS reports any form of error during the execution. As shown in Figure 6def, SQLancer achieves the",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 58274,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "It explores 52%, 44%, and 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively.",
+      "context_before": "g. This is not surprising as previous works have revealed that PostgreSQL is more robust than most other DBMSs. As for MySQL, Kangaroo found a total of ten bugs, including eight memory bugs and two logic bugs. SQLRight found two memory bugs and one logic bug. Squirrel and SQLancer detected five and one memory bugs, respectively. SQLancer, the only generationbased tool, found the least bugs across all comparisons, demonstrating the advantage of the mutation-based method to detect DBMS bugs. Explored New Edges Figure 6abc shows that Kangaroo performs better than others on all three DBMS systems.",
+      "context_after": "Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive instantiation. Generated Valid Queries We treat a query as invalid if DBMS reports any form of error during the execution. As shown in Figure 6def, SQLancer achieves the highest query validity. This result is reasonable because SQLancer follows very limited grammar rules to generate SQL statements. For example, SQLancer does not support generating subqueries that are prone to s",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 58544,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As shown in Figure 6def, SQLancer achieves the highest query validity.",
+      "context_before": "the mutation-based method to detect DBMS bugs. Explored New Edges Figure 6abc shows that Kangaroo performs better than others on all three DBMS systems. It explores 52%, 44%, and 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively. Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive instantiation. Generated Valid Queries We treat a query as invalid if DBMS reports any form of error during the execution.",
+      "context_after": "This result is reasonable because SQLancer follows very limited grammar rules to generate SQL statements. For example, SQLancer does not support generating subqueries that are prone to semantic errors. That makes it easier to generate valid statements but limits the diversity. This could be the main reason that it explores the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensit",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 58991,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This result is reasonable because SQLancer follows very limited grammar rules to generate SQL statements.",
+      "context_before": "re 6abc shows that Kangaroo performs better than others on all three DBMS systems. It explores 52%, 44%, and 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively. Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive instantiation. Generated Valid Queries We treat a query as invalid if DBMS reports any form of error during the execution. As shown in Figure 6def, SQLancer achieves the highest query validity.",
+      "context_after": "For example, SQLancer does not support generating subqueries that are prone to semantic errors. That makes it easier to generate valid statements but limits the diversity. This could be the main reason that it explores the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensit",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 59062,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer does not support generating subqueries that are prone to semantic errors.",
+      "context_before": "nd 14% more edges than SQLancer, Squirrel, and SQLRight on average, respectively. Considering that MPE modifies only a few lines of code out of millions of lines of DBMS code, we believe its impact on the new edges explored is negligible. This means the improvement is mainly due to context-sensitive instantiation. Generated Valid Queries We treat a query as invalid if DBMS reports any form of error during the execution. As shown in Figure 6def, SQLancer achieves the highest query validity. This result is reasonable because SQLancer follows very limited grammar rules to generate SQL statements.",
+      "context_after": "That makes it easier to generate valid statements but limits the diversity. This could be the main reason that it explores the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP. Doin",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 11,
+      "char_offset": 59168,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M25",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP.",
+      "context_before": "n in Figure 6def, SQLancer achieves the highest query validity. This result is reasonable because SQLancer follows very limited grammar rules to generate SQL statements. For example, SQLancer does not support generating subqueries that are prone to semantic errors. That makes it easier to generate valid statements but limits the diversity. This could be the main reason that it explores the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers.",
+      "context_after": "Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements. Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target. After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL. Besides, all logic bugs found by NoREC are covered by MPE. SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Ke",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 59599,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements.",
+      "context_before": "rors. That makes it easier to generate valid statements but limits the diversity. This could be the main reason that it explores the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP.",
+      "context_after": "Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target. After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL. Besides, all logic bugs found by NoREC are covered by MPE. SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribu",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 59859,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target.",
+      "context_before": "s the fewest paths among all fuzzers. Benefiting from the richer semantic constraints due to context-sensitive instantiation, Kangaroo achieves a noticeably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP. Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements.",
+      "context_after": "After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL. Besides, all logic bugs found by NoREC are covered by MPE. SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribution of context-sensitive instantiation, we build Kangaroo !multiplan by disabling the MPE in Kangaroo. Without the MPE, Kangaroo !multiplan focuses",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 59986,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL.",
+      "context_before": "iceably higher query validity than other mutation-based fuzzers. Oracle Comparsion To eliminate the benefit from context-sensitive instantiation, We build Kangaroo MPE by replacing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP. Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements. Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target.",
+      "context_after": "Besides, all logic bugs found by NoREC are covered by MPE. SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribution of context-sensitive instantiation, we build Kangaroo !multiplan by disabling the MPE in Kangaroo. Without the MPE, Kangaroo !multiplan focuses on memory bug detection. We use Squirrel as a baseline since it performs better than SQLancer in detecting memory bugs. We also ported the MPE module into Squirrel, denoted Squi",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 60135,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Besides, all logic bugs found by NoREC are covered by MPE.",
+      "context_before": "ing the context-sensitive instantiation with type-sensitive instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP. Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements. Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target. After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL.",
+      "context_after": "SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribution of context-sensitive instantiation, we build Kangaroo !multiplan by disabling the MPE in Kangaroo. Without the MPE, Kangaroo !multiplan focuses on memory bug detection. We use Squirrel as a baseline since it performs better than SQLancer in detecting memory bugs. We also ported the MPE module into Squirrel, denoted Squirrel multiplan. By comparing Squirrel with Squirrel multipl",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 60312,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLRight TLP and SQLancer PQS fail to detect any logic bug.",
+      "context_before": "instantiation in Kangaroo, and compare Kangaroo MPE to SQLRight NoREC and SQLRight TLP. Doing an automatic comparison of the MPE and PQS is difficult because PQS requires a generation method to generate statements. Considering that the generation of test cases is also one of the evaluation metrics of the oracle, we include SQLancer PQS as the comparison target. After 5 rounds of 24-hour testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL. Besides, all logic bugs found by NoREC are covered by MPE.",
+      "context_after": "5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribution of context-sensitive instantiation, we build Kangaroo !multiplan by disabling the MPE in Kangaroo. Without the MPE, Kangaroo !multiplan focuses on memory bug detection. We use Squirrel as a baseline since it performs better than SQLancer in detecting memory bugs. We also ported the MPE module into Squirrel, denoted Squirrel multiplan. By comparing Squirrel with Squirrel multiplan, we can verify that MPE itself is effective. We adopt the",
+      "section": "5.3 Comparisons with Existing Tools",
+      "page": 12,
+      "char_offset": 60371,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We use Squirrel as a baseline since it performs better than SQLancer in detecting memory bugs.",
+      "context_before": "ur testing, Kangaroo MPE finds three logic bugs in SQLite and two logic bugs in MySQL, and SQLRight NoREC finds one logic bug in SQLite and one in MySQL. Besides, all logic bugs found by NoREC are covered by MPE. SQLRight TLP and SQLancer PQS fail to detect any logic bug. 5.4. Benefits of the Two Key Techniques We conduct unit tests to understand the contribution of the two techniques in Kangaroo. To understand the contribution of context-sensitive instantiation, we build Kangaroo !multiplan by disabling the MPE in Kangaroo. Without the MPE, Kangaroo !multiplan focuses on memory bug detection.",
+      "context_after": "We also ported the MPE module into Squirrel, denoted Squirrel multiplan. By comparing Squirrel with Squirrel multiplan, we can verify that MPE itself is effective. We adopt the same strategy to measure the number of explored edges, the query validity rate, and the number of unique bugs. Figure 6 shows the evaluation results. Context-Sensitive Instantiation Context-sensitive instantiation directly improves the correctness level (Table 1) of generated test cases, thereby greatly improving the efficiency of fuzzing. As shown in Figure 6def, tools using contextsensitive instantiation achieve sign",
+      "section": "5.4 Benefits of the Two Key Techniques",
+      "page": 12,
+      "char_offset": 60759,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M32",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [15] generates equivalent queries by shifting the conditions in the WHERE clause to the SELECT expression.",
+      "context_before": "put with different DBMS systems. Slutz proposed RAGE [27] for finding logic bugs in DBMS by running the same query on different DBMS and comparing their results. Jinho et al. developed APOLLO [5], a system to find performance regression bugs by executing the same query on the DBMSs with different versions. However, RAGE can only be applied to common features of different DBMSs, and APOLLO can only detect bugs introduced or fixed by newer versions. Another approach is based on metamorphic testing which identifies bugs by running two queries with equivalent functionality on the same DBMS system.",
+      "context_after": "TLP [16] partitions a query lacking where clause into three subqueries whose where clause are x IS TRUE ,x IS FALSE, and x IS NULL. Both of them are limited to the queries that can be converted. The last approach tries to build the test case along with the corresponding ground truth result. ADUSA [37] generates all data and the full expected result for a query. However, generating full expected results as ground truth can be expensive which inhibits it from finding more bugs. To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result. It",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 66520,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M33",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [16] partitions a query lacking where clause into three subqueries whose where clause are x IS TRUE ,x IS FALSE, and x IS NULL.",
+      "context_before": "y on different DBMS and comparing their results. Jinho et al. developed APOLLO [5], a system to find performance regression bugs by executing the same query on the DBMSs with different versions. However, RAGE can only be applied to common features of different DBMSs, and APOLLO can only detect bugs introduced or fixed by newer versions. Another approach is based on metamorphic testing which identifies bugs by running two queries with equivalent functionality on the same DBMS system. NoREC [15] generates equivalent queries by shifting the conditions in the WHERE clause to the SELECT expression.",
+      "context_after": "Both of them are limited to the queries that can be converted. The last approach tries to build the test case along with the corresponding ground truth result. ADUSA [37] generates all data and the full expected result for a query. However, generating full expected results as ground truth can be expensive which inhibits it from finding more bugs. To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result. It synthesizes a query that is expected to fetch a single, randomly-selected row. By checking whether this row is fetched, PQS can dete",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 66633,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result.",
+      "context_before": "ystem. NoREC [15] generates equivalent queries by shifting the conditions in the WHERE clause to the SELECT expression. TLP [16] partitions a query lacking where clause into three subqueries whose where clause are x IS TRUE ,x IS FALSE, and x IS NULL. Both of them are limited to the queries that can be converted. The last approach tries to build the test case along with the corresponding ground truth result. ADUSA [37] generates all data and the full expected result for a query. However, generating full expected results as ground truth can be expensive which inhibits it from finding more bugs.",
+      "context_after": "It synthesizes a query that is expected to fetch a single, randomly-selected row. By checking whether this row is fetched, PQS can detect logic bugs in the DBMS. DBMS test cases generation. DBMS requires structural inputs to manipulate data in the database. Structural input generation mainly falls into two categories: generate-based approaches and mutation-based approaches. The generation-based approach [7, 14–17, 27, 37, 38] is effective in generating syntax-correct test cases since it typically follows a grammar model that describes the format of the input. However, these grammar rules are",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 67114,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M35",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "By checking whether this row is fetched, PQS can detect logic bugs in the DBMS.",
+      "context_before": "where clause are x IS TRUE ,x IS FALSE, and x IS NULL. Both of them are limited to the queries that can be converted. The last approach tries to build the test case along with the corresponding ground truth result. ADUSA [37] generates all data and the full expected result for a query. However, generating full expected results as ground truth can be expensive which inhibits it from finding more bugs. To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result. It synthesizes a query that is expected to fetch a single, randomly-selected row.",
+      "context_after": "DBMS test cases generation. DBMS requires structural inputs to manipulate data in the database. Structural input generation mainly falls into two categories: generate-based approaches and mutation-based approaches. The generation-based approach [7, 14–17, 27, 37, 38] is effective in generating syntax-correct test cases since it typically follows a grammar model that describes the format of the input. However, these grammar rules are helpless in improving the semantic correctness of the test case. SQLsmith [7] is a popular DBMS testing tool that can generate syntax-correct test cases from AST.",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 67311,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M36",
+      "found_by": "citation_marker",
+      "surface": "[7, 14–17, 27, 37, 38]",
+      "technique": "pqs",
+      "sentence": "The generation-based approach [7, 14–17, 27, 37, 38] is effective in generating syntax-correct test cases since it typically follows a grammar model that describes the format of the input.",
+      "context_before": "generating full expected results as ground truth can be expensive which inhibits it from finding more bugs. To simplify the ground truth generation, Pivoted Query Synthesis(PQS) [14] only partly validates a query’s result. It synthesizes a query that is expected to fetch a single, randomly-selected row. By checking whether this row is fetched, PQS can detect logic bugs in the DBMS. DBMS test cases generation. DBMS requires structural inputs to manipulate data in the database. Structural input generation mainly falls into two categories: generate-based approaches and mutation-based approaches.",
+      "context_after": "However, these grammar rules are helpless in improving the semantic correctness of the test case. SQLsmith [7] is a popular DBMS testing tool that can generate syntax-correct test cases from AST. Despite it having found over 100 memory bugs in popular DBMSs, SQLsmith achieves quite a low accuracy on semantics which might inhibit it from finding bugs hidden in the deep logic. QAGen [38] proves that generating a completely valid queryis NP-complete. It improves semantic correctness by combining traditional query processing and symbolic execution. Previous works also try to improve query generat",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 67606,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M4",
+        "M21",
+        "M25",
+        "M31"
+      ],
+      "describes_as_state_of_the_art": [
+        "M4",
+        "M13",
+        "M17"
+      ]
+    },
+    "suppressed": [
+      {
+        "mention_id": "M17",
+        "pattern": "compares_with",
+        "suppressed_because": "the sentence is framed as related work, so it describes somebody else's approach"
+      }
+    ]
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:12:40Z",
+    "is_model_written": true,
+    "summary": "Kangaroo targets both memory and logic bugs in DBMSs with two techniques. Context-sensitive instantiation takes all static semantic requirements into account, not only identifier types as existing systems do, so generated queries are semantically valid. Multi-plan execution provides the oracle: rather than running only the optimizer's chosen plan, the DBMS executes all plans for a test case and the results are compared, with any difference indicating a logic bug. Applied to SQLite, PostgreSQL and MySQL it detected 50 new bugs.",
+    "narrative": "Kangaroo argues that SQLancer's oracles constrain the SQL they can test -- NoREC needing a WHERE clause, all three putting limits on queries -- and positions multi-plan execution as free of that constraint. SQLancer is then one of three tools it is measured against over 24 hours, run with NoREC, or TLP where NoREC is unavailable.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "motivation",
+      "M3": "motivation",
+      "M4": "baseline",
+      "M5": "result_comparison",
+      "M6": "result_comparison",
+      "M8": "definition",
+      "M9": "definition",
+      "M10": "motivation",
+      "M11": "definition",
+      "M12": "state_of_the_art",
+      "M14": "baseline",
+      "M15": "baseline",
+      "M16": "state_of_the_art"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer appears only as a tool run for comparison; nothing shown says Kangaroo is built on it."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Multi-plan execution is presented as an alternative to the oracles rather than an extension of one; the paper's argument is that NoREC and TLP restrict the SQL they apply to."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M5",
+          "M14",
+          "M15"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "We also compared Kangaroo with leading DBMS testing tools, such as Squirrel, SQLancer, and SQLRight.",
+            "section": "1 Introduction",
+            "page": 2
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "Specifically, after conducting a 24-hour test on the three DBMSs, Kangaroo successfully detected 17 bugs, while SQLancer, Squirrel, and SQLRight only identified 1, 3, and 6 bugs, respectively.",
+            "section": "1 Introduction",
+            "page": 2
+          },
+          {
+            "mention_id": "M14",
+            "sentence": "We use NoREC for SQLancer and SQLRight.",
+            "section": "5.2 Generating Valid Queries",
+            "page": 11
+          },
+          {
+            "mention_id": "M15",
+            "sentence": "Since SQLancer does not implement NoREC for MySQL, we use TLP instead.",
+            "section": "5.2 Generating Valid Queries",
+            "page": 11
+          }
+        ],
+        "reasoning": "SQLancer is one of three tools compared over a 24-hour run, configured with NoREC, or TLP for MySQL where SQLancer does not implement NoREC, and bug counts and program states are reported.",
+        "techniques": [
+          "norec",
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M12",
+          "M16"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M12",
+            "sentence": "Why Existing Works Cannot Detect the Bug NoREC is one of the most effective DBMS logic bug detection oracles which requires the SELECT statements to satisfy some predefined rules, e.",
+            "section": "3.2 Multi-Plan Execution",
+            "page": 5
+          },
+          {
+            "mention_id": "M16",
+            "sentence": "Comparisons with Existing Tools We compare Kangaroo with three state-of-the-art and open source systems: Squirrel, SQLancer, and SQLRight.",
+            "section": "5.3 Comparisons with Existing Tools",
+            "page": 11
+          }
+        ],
+        "reasoning": "M12 calls NoREC one of the most effective DBMS logic bug detection oracles, and M16 calls SQLancer one of three state-of-the-art open-source systems."
+      }
+    },
+    "disagreements": [
+      "A compares_with pattern on M16 was suppressed as related work; the sentence does introduce the comparison section, and the comparison is established by M4, M5, M14 and M15 in any case."
+    ],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1016_j_displa_2024_102854.json b/_data/papers/paper_doi_10_1016_j_displa_2024_102854.json
new file mode 100644
index 0000000..fcb95e1
--- /dev/null
+++ b/_data/papers/paper_doi_10_1016_j_displa_2024_102854.json
@@ -0,0 +1,382 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:26:41Z",
+  "paper": {
+    "id": "paper:doi:10.1016/j.displa.2024.102854",
+    "title": "Using query semantic and feature transfer fusion to enhance cardinality estimating of property graph queries",
+    "authors": [
+      "Zhenzhen He",
+      "Tiquan Gu",
+      "Jiong Yu"
+    ],
+    "year": 2024,
+    "venue": "Displays (Guildford)",
+    "doi": "10.1016/j.displa.2024.102854",
+    "arxiv_id": null,
+    "s2_paper_id": "1faf7f08681013961cb385344a600bd5f08c69ec",
+    "url": "https://doi.org/10.1016/j.displa.2024.102854",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1016/j.displa.2024.102854",
+      "retrieved_at": "2026-09-08T15:26:41Z",
+      "chars": 67318,
+      "content_sha256": "sha256:e657dd4f8ca9dade627aff21a0d8461d1ac4628a43d717d6a333254bc91cb992"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "J. Ba, M. Rigger, CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation, in: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–13.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 2,
+      "text": "K. Kim, J. Jung, I. Seo, et al., Learned cardinality estimation: An in-depth study, in: Proceedings of the 2022 International Conference on Management of Data, 2022, pp. 1214 –1227.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "H. Lan, Z. Bao, Y. Peng, A survey on advancing the dbms query optimizer: Cardinality estimation, cost model, and plan enumeration, Data Sci. Eng. 6 (2021) 86–101.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "A.H. Izdihar, N.D. Tsaniyah, F. Nurdini, et al., Building a Movie Recommendation System Using Neo4j Graph Database: A Case Study of Netflix Movie Dataset, 2024 ASU International Conference in Emerging Technologies for Sustainability and Intelligent Systems (ICETSIS). IEEE (2024) 614–618.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "E. Geepalla, S. Asharif, Analysis of physical access control system for understanding users behavior and anomaly detection using Neo4j, in: Proceedings of the 6th International Conference on Engineering & MIS 2020, 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "D. Van Landuyt, V. Wijshoff, W. Joosen, A study of NoSQL query injection in Neo4j, Comput. Secur. 137 (2024) 103590.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "G. Gricourt, T. Duigou, S. D´erozier, et al., neo4jsbml: import systems biology markup language data into the graph database Neo4j, PeerJ 12 (2024) e16726.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "D. Tuck, A cancer graph: a lung cancer property graph database in Neo4j, BMC. Res. Notes 15 (1) (2022) 45.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "P. Liu, Y. Huang, P. Wang, et al., Construction of typhoon disaster knowledge graph based on graph database Neo4j, 2020 Chinese Control And Decision Conference (CCDC). IEEE (2020) 3612 –3616.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "N. Francis, A. Green, P. Guagliardo, et al., Cypher: An evolving query language for property graphs, Proceedings of the 2018 international conference on management of data. 2018: 1433-1445.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Y. Wang, C. Xiao, J. Qin, et al., Monotonic cardinality estimation of similarity selection: A deep learning approach, in: Proceedings of the 2020 ACMSIGMOD International Conference on Management of Data, 2020, pp. 1197 –1212.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "P. Negi, R. Marcus, H. Mao, et al., Cost-guided cardinality estimation: Focus where it matters, in: Workshops (ICDEW). IEEE, 2020, pp. 154–157.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "A. Davitkova, D. Gjurovski, S. Michel, LMKG: Learned Models for Cardinality Estimation in Knowledge Graphs. arXiv preprint arXiv:2102.10588, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "S. Thirumuruganathan, S. Shetiya, N. Koudas, et al., Prediction intervals for learned cardinality estimation: An experimental evaluation, in: 2022 IEEE 38th International Conference on Data Engineering (ICDE), 2022, pp. 3051 –3064.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "A. Kipf, T. Kipf, B. Radke, et al., Learned cardinalities: Estimating correlated joins with deep learning. arXiv preprint arXiv:1809.00677, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "J. Sun, G. Li, An end-to-end learning-based cost estimator, Proceedings of the VLDB Endowment 13 (3) (2019) 307–319.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "X. Wang, C. Qu, W. Wu, et al., Are we ready for learned cardinality estimation?. arXiv preprint arXiv:2012.06743, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "X. Lin, X. Zeng, X. Pu, et al., A cardinality estimation approach based on two level histograms, J. Inf. Sci. Eng. 31 (5) (2015) 1733 –1756.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "R. Kaushik, D. Suciu, Consistent histograms in the presence of distinct value counts, Proceedings of the VLDB Endowment 2 (1) (2009) 850–861.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "G. Cormode, M. Garofalakis, P.J. Haas, et al., Synopses for massive data: Samples, histograms, wavelets, sketches, Foundations and Trends ® in Databases 4 (1–3) (2011) 1–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "G. Cormode, Sketch techniques for approximate query processing. Foundations and Trends in Databases, NOW Publishers (2011:) 15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "M. Müller, G. Moerkotte, O. Kolb, Improved selectivity estimation by combining knowledge from sampling and synopses, Proceedings of the VLDB Endowment 11 (9) (2018) 1016 –1028.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Y. Chen, K. Yi, Two-level sampling for join size estimation, Proceedings of the 2017 ACM International Conference on Management of Data. (2017:) 759–774.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "B. Zou, J. You, Q. Wang, et al., Survey on learnable databases: A machine learning perspective, Big Data Res. 27 (2022) 100304.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "W. Wang, M. Zhang, G. Chen, et al., Database meets deep learning: Challenges and opportunities, ACMSIGMOD Rec. 45 (2) (2016) 17–22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "S. Kl¨abe, S. Hagedorn, K.U. Sattler, Exploration of Approaches for In-Database ML, Proceedings of the 26th International Conference on Extending Database Technology, EDBT 2023, Ioannina, Greece, March 28-March 31. 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "G. Li, X. Zhou, L. Cao, AI meets database: AI4DB and DB4AI, Proceedings of the 2021 International Conference on Management of Data (2021) 2859 –2866.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "I.A. Chikwendu, X. Zhang, I.O. Agyemang, et al., A comprehensive survey on deep graph representation learning methods, J. Artif. Intell. Res. 78 (2023) 287–356.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "T. Malik, R.C. Burns, N.V. Chawla, A Black-Box Approach to Query Cardinality Estimation, CIDR. 2007: 56-67.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "C. Wu, A. Jindal, S. Amizadeh, et al., Towards a learning optimizer for shared clouds, Proceedings of the VLDB Endowment 12 (3) (2018) 210–222.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "A. Dutt, C. Wang, A. Nazi, et al., Selectivity estimation for range predicates using lightweight models, Proceedings of the VLDB Endowment 12 (9) (2019) 1044 –1057.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "H. Liu, M. Xu, Z. Yu, et al., Cardinality estimation using neural networks, Proceedings of the 25th Annual International Conference on Computer Science and Software Engineering, 2015: 53-59.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "L. Woltmann, C. Hartmann, M. Thiele, et al., Cardinality Estimation with Local Deep Learning Models, Proceedings of the Second International Workshop on Exploiting Artificial Intelligence Techniques for Data Management. (2019) 1–8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "L. Woltmann, C. Hartmann, D. Habich, et al., Aggregate-based training phase for ML-based cardinality estimation, Datenbank-Spektrum 22 (1) (2022) 45–57.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "A. Kipf, D. Vorona, J. Müller, et al., Estimating cardinalities with deep sketches, Proceedings of the 2019 International Conference on Management of Data (2019) 1937 –1940.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "S. Hasan, S. Thirumuruganathan, J. Augustine, et al., Deep learning models for selectivity estimation of multi-attribute queries, in: Proceedings of The 2020 ACM SIGMOD International Conference on Management of Data, 2020, pp. 1035 –1050.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "J. Ortiz, M. Balazinska, J. Gehrke, et al., An empirical analysis of deep learning for cardinality estimation. arXiv preprint arXiv:1905.06425, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "D.S. Karamyan, Cardinality estimation of an SQL query using recursive neural networks, Mathematical Problems of Computer Science 54 (2020) 41–52.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Z. He, J. Yu, T. Gu, et al., Query cost estimation in graph databases via emphasizing query dependencies by using a neural reasoning network, Concurrency and Computation: Practice and Experience, e7817.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Q. Zhou, G. Yang, H. Song, et al., A BiLSTM cardinality estimator in complex database systems based on attention mechanism, CAAI Transactions on Intelligence Technology 7 (3) (2022) 537–546.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "J. Pennington, R. Socher, C.D. Manning, Glove: Global vectors for word representation, Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP), 2014: 1532-1543.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "H.R.V. Joze, A. Shaban, M.L. Iuzzolino, et al., MMTM: multimodal transfer module for CNN fusion, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2020) 13289 –13299.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Https://github.com/nielsdejong/neo4j-cardinality-collector.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "G. Moerkotte, T. Neumann, G. Steidl, Preventing bad plans by bounding the impact of cardinality estimation. errors. Proceedings of the VLDB Endowment, 2009, 2(1): 982-993.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "G. Brassington, Mean absolute error and root mean square error: which is the better metric for assessing model performance? EGU General Assembly Conference Abstracts (2017) 3574 .Z. He et al. Displays 85 (2024) 102854 13",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 1,
+      "text": "J. Ba, M. Rigger, CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation, in: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–13.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[1–3]",
+      "technique": "cert",
+      "sentence": "Introduction Cardinality estimation [1–3]is one of the important factors affecting the accuracy of query cost estimation results, and accurate estimation results can avoid large estimation errors.",
+      "context_before": "ries, specifically introducing a feature information transfer module to dynamically control the information flow meanwhile achieving the model ’s feature fusion and inference. Experimental results on three datasets show that the estimation model can accurately and efficiently estimate the cardinality of property graph queries, the mean Q_error and RMSE are reduced by about 30% and 25% than the state-of-art estimation models. The context semantics features of queries can improve the model ’s estimation accuracy, the mean Q_error result is reduced by about 20% and the RMSE result is about 5%. 1.",
+      "context_after": "Accurately estimating the query result before query execution helps database administrators manage and optimize query tasks. For example, the database is running a complex report generation query that involves joining multiple tables and complex aggregation operations. If the cardinality estimation model can predict the number of rows that the query will return, administrators can choose to optimize the query (e.g., add and adjust indexes) or rewrite the query (e.g., use the LIMIT clause) based on the predicted results. Neo4j [4,5] is a native property graph database and is widely used in soc",
+      "section": null,
+      "page": 1,
+      "char_offset": 1863,
+      "cited_reference": {
+        "number": 1,
+        "text": "J. Ba, M. Rigger, CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation, in: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–13.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[0,1]",
+      "technique": "cert",
+      "sentence": "For the property value v, it can be normalized to [0,1] according to the maximum value vmax and minimum value vmin in different property keys, where the normalization method is: vnor=v vmin vmax vmin# (3) Fig.",
+      "context_before": "n association with the selection operations (e.g., ≤, ≥), such as p≥v, where p∈K is the property key, ≥is the selection operation, and v∈V is the property value. This section considers two types of graph query property values: numeric and string types. If the graph query property v is numeric, five property selection operations <, >, =, ≤and ≥can be associated, then the numeric Fig. 1.The generation process of training samples.Z. He et al. Displays 85 (2024) 102854 4 property value and property selection operation can be encoded together as the feature vector V(v)=(v<,v>,v=,v≤,v≥)of length 5.",
+      "context_after": "2.The overview of the cardinality estimation model structure. Fig. 3.The encoding for a graph query entity node, relationship, property keys, and query structure.Z. He et al. Displays 85 (2024) 102854 5 Then, a matrix M(v)with the shape of 5×dh is defined for each property value. Then the graph query property value vector represents F(v)=V(v)×M(v), where the shape size of the F(v)vector is 1×dh, and dh is the size of the hidden layer in the neural network. Fig. 4shows the encoding process of graph query property value. Furthermore, given a graph query entity node a contains m attributes, the",
+      "section": null,
+      "page": 5,
+      "char_offset": 26052,
+      "cited_reference": {
+        "number": 1,
+        "text": "J. Ba, M. Rigger, CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation, in: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–13.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:52Z",
+    "is_model_written": true,
+    "summary": "This paper estimates the cardinality of property graph queries by fusing query semantics with transferred features, so that a model trained on one workload informs estimates for another. Accurate cardinality estimation is what keeps a query planner's cost estimates close to reality.",
+    "narrative": "The connection is a single background citation in the opening sentence on cardinality estimation, reachable only through the citation marker -- SQLancer is never named. The paper's subject is estimation accuracy rather than testing, and neither the tool nor any of its oracles plays a role in the work.",
+    "roles": {
+      "M1": "background",
+      "M2": "incidental"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_ase56229_2023_00106.json b/_data/papers/paper_doi_10_1109_ase56229_2023_00106.json
new file mode 100644
index 0000000..e221ad6
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_ase56229_2023_00106.json
@@ -0,0 +1,618 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:07:41Z",
+  "paper": {
+    "id": "paper:doi:10.1109/ase56229.2023.00106",
+    "title": "Perfce: Performance Debugging on Databases with Chaos Engineering-Enhanced Causality Analysis",
+    "authors": [
+      "Zhenlan Ji",
+      "Pingchuan Ma",
+      "Shuai Wang"
+    ],
+    "year": 2022,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1109/ase56229.2023.00106",
+    "arxiv_id": "2207.08369",
+    "s2_paper_id": "d0da09b12d16ea685532cdd47185ad2b888fc03a",
+    "url": "https://doi.org/10.1109/ase56229.2023.00106",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2207.08369",
+      "retrieved_at": "2026-09-07T17:07:41Z",
+      "chars": 77838,
+      "content_sha256": "sha256:0be0fab8a8860f771e5945c540d5fd4a35abb7d466258f110e3993ef4c99e983"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1563
+      },
+      {
+        "number": "A",
+        "title": "Database Performance Diagnosis",
+        "start": 8529
+      },
+      {
+        "number": "B",
+        "title": "Causality Analysis",
+        "start": 10664
+      },
+      {
+        "number": "A",
+        "title": "Offline CE-Enhanced SEM Learning Phase",
+        "start": 25453
+      },
+      {
+        "number": "B",
+        "title": "Online Root Cause Analysis",
+        "start": 35722
+      },
+      {
+        "number": "Y",
+        "title": "By ranking causes by blame, Alg. 1 directs users’ attention",
+        "start": 36674
+      },
+      {
+        "number": "A",
+        "title": "Evaluation on Effectiveness",
+        "start": 42998
+      },
+      {
+        "number": "B",
+        "title": "Evaluation on Scalability",
+        "start": 53366
+      },
+      {
+        "number": "C",
+        "title": "Evaluation on Counterfactual Analysis",
+        "start": 56342
+      },
+      {
+        "number": "Y",
+        "title": "Li, N. Qiu et al., “Diagnosing root causes of intermittent slow queries",
+        "start": 63763
+      },
+      {
+        "number": "C",
+        "title": "Delimitrou, “Seer: Leveraging big data to navigate the complexity",
+        "start": 70797
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“How slow database queries can negatively impact your business,” https://wire19.com/how-slow-database-queries-can-negatively-impactyour-business, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“Amazon found every 100ms of latency cost them 1% in sales,” https://www.gigaspaces.com/blog/amazon-found-every-100ms-oflatency-cost-them-1-in-sales, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“Marissa mayer at web 2.0,” http://glinden.blogspot.com/2006/11/marissamayer-at-web-20.html, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "M. Ma, Z. Yin, S. Zhang, S. Wang, C. Zheng, X. Jiang, H. Hu, C. Luo, Y. Li, N. Qiu et al., “Diagnosing root causes of intermittent slow queries in cloud databases,” Proceedings of the VLDB Endowment, vol. 13, no. 8, pp. 1176–1189, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Y. Gan, M. Liang, S. Dev, D. Lo, and C. Delimitrou, “Sage: practical and scalable ml-driven performance debugging in microservices,” in Proceedings of the 26th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, 2021, pp. 135–151.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "P. Chen, Y. Qi, P. Zheng, and D. Hou, “Causeinfer: Automatic and distributed performance diagnosis with hierarchical causality graph in large distributed systems,” in IEEEINFOCOM 2014-IEEE Conference on Computer Communications. IEEE, 2014, pp. 1887–1895.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "J. Lin, P. Chen, and Z. Zheng, “Microscope: Pinpoint performance issues with causal graphs in micro-service environments,” in International Conference on Service-Oriented Computing. Springer, 2018, pp. 3–20.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "V. Jeyakumar, O. Madani, A. Parandeh, A. Kulshreshtha, W. Zeng, and N. Yadav, “Explainit!–a declarative root-cause analysis engine for time series data,” in Proceedings of the 2019 International Conference on Management of Data, 2019, pp. 333–348.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "H. Wang, Z. Wu, H. Jiang, Y. Huang, J. Wang, S. Kopru, and T. Xie, “Groot: An event-graph-based approach for root cause analysis in industrial settings,” in 2021 36th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 2021, pp. 419–429.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "P. Spirtes, C. N. Glymour, R. Scheines, and D. Heckerman, Causation, prediction, and search. MIT press, 2000.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "P. R. Rosenbaum and D. B. Rubin, “The central role of the propensity score in observational studies for causal effects,” Biometrika, vol. 70, no. 1, pp. 41–55, 1983.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "J. Hartford, G. Lewis, K. Leyton-Brown, and M. Taddy, “Deep iv: A flexible approach for counterfactual prediction,” in International Conference on Machine Learning. PMLR, 2017, pp. 1414–1423.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "“Chaos mesh: A powerful chaos engineering platform for kubernetes,” https://chaos-mesh.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "“TiDB,” https://github.com/pingcap/tidb, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "“Kubernetes,” https://kubernetes.io/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "“Research artifact,” https://anonymous.4open.science/r/PerfCE-85E0, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "“Anonymous documentation of perfce,” http://perfce.ignorelist.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "“Pod,” https://kubernetes.io/docs/concepts/workloads/pods/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "J. A. Jones and M. J. Harrold, “Empirical evaluation of the tarantula automatic fault-localization technique,” in Proceedings of the 20th IEEE/ACM international Conference on Automated software engineering, 2005, pp. 273–282.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "B. Liblit, M. Naik, A. X. Zheng, A. Aiken, and M. I. Jordan, “Scalable statistical bug isolation,” in Proceedings of the 2005 ACMSIGPLAN conference on Programming language design and implementation, 2005, pp. 15–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "C. Liu, L. Fei, X. Yan, J. Han, and S. P. Midkiff, “Statistical debugging: A hypothesis testing-based approach,” IEEE Transactions on software engineering, vol. 32, no. 10, pp. 831–848, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "M. Attariyan, M. Chow, and J. Flinn, “X-ray: Automating {Root-Cause } diagnosis of performance anomalies in production software,” in 10th USENIX Symposium on Operating Systems Design and Implementation (OSDI 12), 2012, pp. 307–320.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "X. Zhao, K. Rodrigues, Y. Luo, D. Yuan, and M. Stumm, “ {NonIntrusive }performance profiling for entire software stacks based on the flow reconstruction principle,” in 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI 16), 2016, pp. 603–618.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "C.-P. Bezemer, J. Pouwelse, and B. Gregg, “Understanding software performance regressions using differential flame graphs,” in 2015 IEEE 22nd International Conference on Software Analysis, Evolution, and Reengineering (SANER). IEEE, 2015, pp. 535–539.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "E. Soremekun, L. Kirschner, M. B ¨ohme, and A. Zeller, “Locating faults with program slicing: an empirical analysis,” Empirical Software Engineering, vol. 26, no. 3, pp. 1–45, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "G. Jin, A. Thakur, B. Liblit, and S. Lu, “Instrumentation and sampling strategies for cooperative concurrency bug isolation,” in Proceedings of the ACM international conference on Object oriented programming systems languages and applications, 2010, pp. 241–255.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Z. Zuo, L. Fang, S.-C. Khoo, G. Xu, and S. Lu, “Low-overhead and fully automated statistical debugging with abstraction refinement,” in Proceedings of the 2016 ACMSIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications, 2016, pp. 881–896.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "B. Gregg, Systems performance: enterprise and the cloud. Pearson Education, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "J. Tuya, J. Dolado, M. J. Suarez-Cabal, and C. de la Riva, “A controlled experiment on white-box database testing,” ACMSIGSOFT Software Engineering Notes, vol. 33, no. 1, pp. 1–6, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "E. Costante, J. den Hartog, M. Petkovi ´c, S. Etalle, and M. Pechenizkiy, “A white-box anomaly-based framework for database leakage detection,” Journal of Information Security and Applications, vol. 32, pp. 27–46, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "J. Peters, D. Janzing, and B. Sch ¨olkopf, Elements of causal inference: foundations and learning algorithms. The MIT Press, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "P. W. Holland, “Statistics and causal inference,” Journal of the American statistical Association, vol. 81, no. 396, pp. 945–960, 1986.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "J. Pearl and T. Verma, “A theory of inferred causation,” in Proceedings of the Second International Conference on Principles of Knowledge Representation and Reasoning, 1991, pp. 441–452.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "A. Balke and J. Pearl, “Probabilistic evaluation of counterfactual queries,” inProbabilistic and Causal Inference: The Works of Judea Pearl, 2022, pp. 237–254.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "“Principles of chaos engineering,” https://principlesofchaos.org, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "“Sqlsmith,” https://github.com/anse1/sqlsmith, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "K. Kallas, F. Niksic, C. Stanford, and R. Alur, “DiffStream: Differential output testing for stream processing programs,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–29, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "T. Sotiropoulos, S. Chaliasos, V. Atlidakis, D. Mitropoulos, and D. Spinellis, “Data-oriented differential testing of object-relational mapping systems,” in Engineering (ICSE). IEEE, 2021, pp. 1535–1547.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "V. Chernozhukov, D. Chetverikov, M. Demirer, E. Duflo, C. Hansen, W. Newey, and J. Robins, “Double/debiased machine learning for treatment and causal parameters,” arXiv preprint arXiv:1608.00060, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "D. Y. Yoon, N. Niu, and B. Mozafari, “Dbsherlock: A performance diagnostic tool for transactional databases,” in Proceedings of the 2016 International Conference on Management of Data, 2016, pp. 1599–1614.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "J. Y. Halpern and J. Pearl, “Causes and explanations: A structural-model approach. part i: Causes,” The British journal for the philosophy of science, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Y. Gan, Y. Zhang, K. Hu, D. Cheng, Y. He, M. Pancholi, and C. Delimitrou, “Seer: Leveraging big data to navigate the complexity of performance debugging in cloud microservices,” in Proceedings of the twenty-fourth international conference on architectural support for programming languages and operating systems, 2019, pp. 19–33.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "P. Liu, S. Zhang, Y. Sun, Y. Meng, J. Yang, and D. Pei, “Fluxinfer: Automatic diagnosis of performance anomaly for online database system,” in 2020 IEEE 39th International Performance Computing and Communications Conference (IPCCC). IEEE, 2020, pp. 1–8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Y. Li, “Toward a qualitative search engine,” IEEE Internet Computing, vol. 2, no. 4, pp. 24–29, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "B. Kim, C. Rudin, and J. A. Shah, “The bayesian case model: A generative approach for case-based reasoning and prototype classification,” Advances in neural information processing systems, vol. 27, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "K. Sohn, H. Lee, and X. Yan, “Learning structured output representation using deep conditional generative models,” Advances in neural information processing systems, vol. 28, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "J. Y. Halpern and J. Pearl, “Causes and explanations: A structural-model approach. part i: Causes,” The British journal for the philosophy of science, 2005.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "——, “Causes and explanations: A structural-model approach. part ii: Explanations,” The British journal for the philosophy of science, 2005.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "T. P. P. Council, “Tpc-c benchmark,” https://www.tpc.org/tpcc, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "M. Scanagatta, C. P. de Campos, G. Corani, and M. Zaffalon, “Learning bayesian networks with thousands of variables,” Advances in neural information processing systems, vol. 28, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "R. Ding, Y. Liu, J. Tian, Z. Fu, S. Han, and D. Zhang, “Reliable and efficient anytime skeleton learning,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 34, no. 06, 2020, pp. 10 101– 10 109.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "J. Pearl, Causality. Cambridge university press, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "M. Rosenblatt, “Remarks on some nonparametric estimates of a density function,” The Annals of Mathematical Statistics, vol. 27, no. 3, pp. 832–837, 1956.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "E. Parzen, “On estimation of a probability density function and mode,” The annals of mathematical statistics, vol. 33, no. 3, pp. 1065–1076, 1962.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "“Chaos variables,” https://anonymous.4open.science/r/PerfCE85E0/supplement/chaos var.md.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "“TiDB-config,” https://docs.pingcap.com/tidb-in-kubernetes/stable/getstarted, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "“Grafana,” https://grafana.com, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "“Prometheus,” https://prometheus.io, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "“Mysql kpis,” https://anonymous.4open.science/r/PerfCE85E0/supplement/kpi.md.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "PingCAP, “Tidb kpis,” https://docs.pingcap.com/tidb/stable/grafanaoverview-dashboard, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "H. Dai, R. Ding, Y. Jiang, S. Han, and D. Zhang, “Ml4c: Seeing causality through latent vicinity,” arXiv preprint arXiv:2110.00637, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "“Econml: A python package for ml-based heterogeneous treatment effects estimation,” https://github.com/microsoft/EconML, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "D. E. Difallah, A. Pavlo, C. Curino, and P. Cudr ´e-Mauroux, “Oltpbench: An extensible testbed for benchmarking relational databases,” PVLDB, vol. 7, no. 4, pp. 277–288, 2013. [Online]. Available: http://www.vldb.org/pvldb/vol7/p277-difallah.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "“Exporter for mysql server metrics,” https://github.com/prometheus/mysqld exporter, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "“Pgmpy,” https://pgmpy.org/metrics/metrics.html, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "X. Zheng, B. Aragam, P. K. Ravikumar, and E. P. Xing, “Dags with no tears: Continuous optimization for structure learning,” Advances in Neural Information Processing Systems, vol. 31, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "M. Zalewski, “American Fuzzy Lop (AFL),” https://lcamtuf.coredump. cx/afl/, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "P. Ma and S. Wang, “Mt-teql: evaluating and augmenting neural nlidb on real-world linguistic and schema variations,” Proceedings of the VLDB Endowment, vol. 15, no. 3, pp. 569–582, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "L. Lorch, J. Rothfuss, B. Sch ¨olkopf, and A. Krause, “Dibs: Differentiable bayesian structure learning,” Advances in Neural Information Processing Systems, vol. 34, pp. 24 111–24 123, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "P. Ma, R. Ding, H. Dai, Y. Jiang, S. Wang, S. Han, and D. Zhang, “Ml4s: Learning causal skeleton from vicinal graphs,” in Proceedings of the 28th ACMSIGKDD Conference on Knowledge Discovery and Data Mining, 2022, pp. 1213–1223.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "P. Ma, Z. Ji, Q. Pang, and S. Wang, “Noleaks: Differentially private causal discovery under functional causal model,” IEEE Transactions on Information Forensics and Security, vol. 17, pp. 2324–2338, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "D. B. Rubin, “Causal inference using potential outcomes: Design, modeling, decisions,” Journal of the American Statistical Association, vol. 100, no. 469, pp. 322–331, 2005.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "——, “Estimating causal effects of treatments in randomized and nonrandomized studies.” Journal of educational Psychology, vol. 66, no. 5, p. 688, 1974.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "A. Fariha, S. Nath, and A. Meliou, “Causality-guided adaptive interventional debugging,” in Proceedings of the 2020 ACMSIGMOD International Conference on Management of Data, 2020, pp. 431–446.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "C. Dubslaff, K. Weis, C. Baier, and S. Apel, “Causality in configurable software systems,” arXiv preprint arXiv:2201.07280, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "B. Johnson, Y. Brun, and A. Meliou, “Causal testing: understanding defects’ root causes,” in Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering, 2020, pp. 87–99.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "R. Krishna, M. S. Iqbal, M. A. Javidian, B. Ray, and P. Jamshidi, “Cadet: Debugging and fixing misconfigurations using counterfactual reasoning,” arXiv preprint arXiv:2010.06061, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "C.-H. Hsiao, S. Narayanasamy, E. M. I. Khan, C. L. Pereira, and G. A. Pokam, “Asyncclock: Scalable inference of asynchronous event causality,” ACMSIGPLAN Notices, vol. 52, no. 4, pp. 193–205, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "B. Sun, J. Sun, L. H. Pham, and J. Shi, “Causality-based neural network repair,” in Proceedings of the 44th International Conference on Software Engineering, 2022, pp. 338–349.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "M. Zhang and J. Sun, “Adaptive fairness improvement based on causality analysis,” in Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2022, pp. 6–17.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "P. Ma, R. Ding, S. Wang, S. Han, and D. Zhang, “Xinsight: explainable data analysis through the lens of causality,” arXiv preprint arXiv:2207.12718, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Z. Ji, P. Ma, Y. Yuan, and S. Wang, “Cc: Causality-aware coverage criterion for deep neural networks,” in 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2023, pp. 1788– 1800.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "V. Monjezi, A. Trivedi, G. Tan, and S. Tizpaz-Niari, “Informationtheoretic testing and debugging of fairness defects in deep neural networks,” in 45th IEEE/ACM International Conference on Software Engineering, ICSE 2023, Melbourne, Australia, May 14-20, 2023, 2023, pp. 1571–1582.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Z. Ji, P. Ma, S. Wang, and Y. Li, “Causality-aided trade-off analysis for machine learning fairness,” arXiv preprint arXiv:2305.13057, 2023.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 37,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[37–39]",
+      "technique": "pqs",
+      "sentence": "This setup expands the standard “differential testing” procedure [37–39], requiring consistency between the experimental group and reference even under CE stress.",
+      "context_before": "ypically involves three steps: (I) Defining Steady State. CE starts by defining test oracles, or “steady states,” as easily measurable outputs of a system that indicate normal behavior. The key hypothesis is that a steady state will persist in both a control group and the experimental group subjected to CE stress. Steady states may include throughput and query outputs, considering domain-specific demands. For example, SQLSmith [36] compares the outputs of a database (under CE stress) and MySQL. The steady state is defined as the SQL execution outputs being consistent between the two databases.",
+      "context_after": "(II) Picking Chaos Variables. CE consists of chaos variables, each representing a critical, low-level factor that may induce failures in infrastructure, networks, and systems. Modern CE frameworks like Chaos Mesh [13] are coupled with containerization environments like K8s. offering chaos variables for various failures in K8s clusters (e.g., container-kill, podkill). Chaos variables are notthe same as KPIs; there are usually more KPIs than chaos variables. Mutating each chaos variable (e.g., an IO-related variable) may affect many KPIs (e.g., average I/O time). (III) Launching CE and Testing.",
+      "section": "B Causality Analysis",
+      "page": 3,
+      "char_offset": 14407,
+      "cited_reference": {
+        "number": 37,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "PerfCE diagnoses performance anomalies in databases by combining causal inference with chaos engineering. Causal analysis of performance downgrades is hampered by limited observability, so PerfCE uses chaos experiments -- injecting events such as network slowdowns -- to gather the observations it needs. Offline it learns statistical models from passive observation and proactive experiments; online it diagnoses root causes as anomalies occur.",
+    "narrative": "SQLancer is cited once, as one of the works establishing the standard differential testing procedure that PerfCE's setup extends by requiring consistency between an experimental and a reference group.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_ase63991_2025_00095.json b/_data/papers/paper_doi_10_1109_ase63991_2025_00095.json
new file mode 100644
index 0000000..231b65e
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_ase63991_2025_00095.json
@@ -0,0 +1,710 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:15Z",
+  "paper": {
+    "id": "paper:doi:10.1109/ase63991.2025.00095",
+    "title": "ZendDiff: Differential Testing of PHP Interpreter",
+    "authors": [
+      "Yuancheng Jiang",
+      "Jianing Wang",
+      "Qiange Liu",
+      "Yeqi Fu",
+      "Jian Mao",
+      "Roland H. C. Yap",
+      "Zhenkai Liang"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1109/ase63991.2025.00095",
+    "arxiv_id": null,
+    "s2_paper_id": "192e297c91aca3b7a6ab579ef11f38acc09f51a0",
+    "url": "https://doi.org/10.1109/ase63991.2025.00095",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/ase63991.2025.00095",
+      "retrieved_at": "2026-09-09T01:31:15Z",
+      "chars": 71596,
+      "content_sha256": "sha256:1e57fa77708a27e56977cbc75a1593f1fb04ef7092f1fe05c3ebac2d8a2b65e8"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2060
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 12479
+      },
+      {
+        "number": "III",
+        "title": "APPROACH",
+        "start": 16348
+      },
+      {
+        "number": "A",
+        "title": "Differential Input Preparation",
+        "start": 18848
+      },
+      {
+        "number": "B",
+        "title": "Program State Probing.",
+        "start": 19800
+      },
+      {
+        "number": "C",
+        "title": "JIT-Aware Program Mutation",
+        "start": 21599
+      },
+      {
+        "number": "D",
+        "title": "Differential Testing with Dual Verification",
+        "start": 23687
+      },
+      {
+        "number": "IV",
+        "title": "IMPLEMENTATION",
+        "start": 27014
+      },
+      {
+        "number": "V",
+        "title": "EVALUATION",
+        "start": 28719
+      },
+      {
+        "number": "A",
+        "title": "Discovering Previously Unknown Logic Bugs",
+        "start": 31267
+      },
+      {
+        "number": "B",
+        "title": "Improved Effectiveness of ZendDiff",
+        "start": 41502
+      },
+      {
+        "number": "C",
+        "title": "Ablation Study of ZendDiff",
+        "start": 48538
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "H. Kiran, https://techjury.net/blog/php-usage-statistics/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "w3techs, https://w3techs.com/technologies/details/pl-php.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "C. Holler, K. Herzig, and A. Zeller, “Fuzzing with code fragments,” in21st USENIX Security Symposium (USENIX Security 12), 2012, pp. 445–458.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "C. Aschermann, T. Frassetto, T. Holz, P. Jauernig, A.-R. Sadeghi, and D. Teuchert, “Nautilus: Fishing for deep bugs with grammars.” in NDSS, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "P. Srivastava and M. Payer, “Gramatron: Effective grammar-aware fuzzing,” in Proceedings of the 30th acm sigsoft international symposium on software testing and analysis, 2021, pp. 244–256.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Y. Chen, R. Zhong, H. Hu, H. Zhang, Y. Yang, D. Wu, and W. Lee, “One engine to fuzz’em all: Generic language processor testing with semantic validation,” in (SP). IEEE, 2021, pp. 642–658.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Y. Jiang, C. Zhang, B. Ruan, J. Liu, M. Rigger, R. H. Yap, and Z. Liang, “Fuzzing the PHP interpreter via dataflow fusion,” in 34th USENIX Security Symposium (USENIX Security 25), 2025, pp. 6143–6158.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 8,
+      "text": "C. Zhang, G. Lee, Q. Liu, and M. Payer, “Reflecta: Reflection-based scalable and semantic scripting language fuzzing,” in Proceedings of the20th ACM Asia Conference on Computer and Communications Security, 2025, p. 1772–1787.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "T. Y. Chen, S. C. Cheung, and S. M. Yiu, “Metamorphic testing: a new approach for generating next test cases,” Department of Computer Science, Hong Kong, Tech. Rep. HKUST-CS98-01, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "T. Y. Chen, F.-C. Kuo, H. Liu, P.-L. Poon, D. Towey, T. H. Tse, and Z. Q. Zhou, “Metamorphic testing: A review of challenges and opportunities,” ACM Comput. Surv., vol. 51, no. 1, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "T. Y. Chen, J. W. Ho, H. Liu, and X. Xie, “An innovative approach for testing bioinformatics programs using metamorphic testing,” BMC bioinformatics, vol. 10, no. 1, pp. 1–12, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "A. Ramanathan, C. A. Steed, and L. L. Pullum, “Verification of compartmental epidemiological models using metamorphic testing, model checking and visual analytics,” in 2012 ASE/IEEE International Conference on BioMedical Computing (BioMedCom), 2012, pp. 68–73.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "W. K. Chan, S. C. Cheung, and K. R. Leung, “A metamorphic testing approach for online testing of service-oriented software applications,” International Journal of Web Services Research (IJWSR), vol. 4, no. 2, pp. 61–81, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "F.-C. Kuo, T. Y. Chen, and W. K. Tam, “Testing embedded software by metamorphic testing: A wireless metering system case study,” in. 291– 294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "W. K. Chan, T. Y. Chen, H. Lu, T. H. Tse, and S. S. Yau, “Integration testing of context-sensitive middleware-based applications: a metamorphic approach,” International Journal of Software Engineering and Knowledge Engineering, vol. 16, no. 05, pp. 677–703, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "M. N. Mansur, M. Christakis, and V. W ¨ustholz, “Metamorphic testing of datalog engines,” in Proceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2021, p. 639–650.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "L. Bernhard, T. Scharnowski, M. Schloegel, T. Blazytko, and T. Holz, “JIT-Picking: Differential fuzzing of javascript engines,” in Proceedings of the 2022 ACMSIGSAC Conference on Computer and Communications Security, 2022, pp. 351–364.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "J. Park, S. An, D. Youn, G. Kim, and S. Ryu, “JEST: N+1 -version Differential Testing of Both JavaScript Engines and Specification.” in International Conference on Software Engineering (ICSE), 2021, pp. 13–24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "X. Yang, Y. Chen, E. Eide, and J. Regehr, “Finding and understanding bugs in C compilers,” in Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation, 2011, p. 283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "C. Brubaker, S. Jana, B. Ray, S. Khurshid, and V. Shmatikov, “Using Frankencerts for Automated Adversarial Testing of Certificate Validation in SSL/TLS Implementations,” in and Privacy, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Y. Chen and Z. Su, “Guided differential testing of certificate validation in SSL/TLS implementations,” in Proceedings of the 10th Joint Meeting on Foundations of Software Engineering, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Y. Chen, T. Su, C. Sun, Z. Su, and J. Zhao, “Coverage-directed differential testing of JVM implementations,” in Proceedings of the 37th ACMSIGPLAN Conference on Programming Language Design and Implementation. ACM, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Y. Chen, T. Su, and Z. Su, “Deep differential testing of JVM implementations.” in International Conference on Software Engineering (ICSE), 2019, pp. 1257–1268.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "M. Sharma, P. Yu, and A. F. Donaldson, “RustSmith: Random differential compiler testing for Rust,” in Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, 2023, pp. 1483–1486.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "S. Song, J. Hur, S. Kim, P. Rogers, and B. Lee, “R2z2 - detecting rendering regressions in web browsers through differential fuzz testing,” inProceedings of the 44th International Conference on Software Engineering. ACM, 2022, pp. 1818–1829.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "S. Nilizadeh, Y. Noller, and C. S. Pasareanu, “Diffuzz: Differential Fuzzing for Side-Channel Analysis.” in Software Engineering (SE), 2020, pp. 125–126.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Facebook, “HHVM: A virtual machine for executing programs written in hack.” https://github.com/facebook/hhvm, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "F. Emmott, “Ending php support, and the future of hack,” https://hhvm. com/blog/2018/09/12/end-of-php-support-future-of-hack.html, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "“PHP 8.0 JIT,” https://php.watch/versions/8.0/JIT, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "“PHPJIT in depth,” https://php.watch/articles/jit-in-depth, 2020. 1105",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "C. Li, Y. Jiang, C. Xu, and Z. Su, “Validating jit compilers via compilation space exploration,” ACM Trans. Comput. Syst., vol. 43, no. 3, Jul. 2025. [Online]. Available: https://doi.org/10.1145/3715102",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "V. Le, M. Afshari, and Z. Su, “Compiler validation via equivalence modulo inputs,” in Proceedings of the 35th ACMSIGPLAN Conference on Programming Language Design and Implementation, ser. PLDI ’14. New York, NY, USA: Association for Computing Machinery, 2014, p. 216–226. [Online]. Available: https://doi.org/10.1145/2594291.2594334",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "S. Zhou, M. Jiang, W. Chen, H. Zhou, H. Wang, and X. Luo, “Wadiff: A differential testing framework for webassembly runtimes,” inProceedings of the 38th IEEE/ACM International Conference on Automated Software Engineering, ser. ASE ’23. IEEE Press, 2024, p. 939–950. [Online]. Available: https://doi.org/10.1109/ASE56229.2023. 00188",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "L. Wachter, J. Gremminger, C. Wressnegger, M. Payer, and F. Toffalini, “Dumpling: Fine-grained differential javascript engine fuzzing,” in NDSS, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "P. Li and M. Zhang, “Fuzzcache: Optimizing web application fuzzing through software-based data cache,” 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "“PHPT structure details,” https://qa.php.net/phpt details.php, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "“PHP opcache runtime configuration,” https://www.php.net/manual/en/ opcache.configuration.php, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "A. Zeller and R. Hildebrandt, “Simplifying and isolating failure-inducing input,” IEEE Transactions on Software Engineering, vol. 28, no. 2, pp. 183–200, 2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "J. Regehr, Y. Chen, P. Cuoq, E. Eide, C. Ellison, and X. Yang, “Testcase reduction for C compiler bugs,” in Proceedings of the 33rd ACMSIGPLAN conference on Programming Language Design and Implementation, 2012, pp. 335–346.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, nov 2020. [Online]. Available: https://doi-org.libproxy1.nus.edu.sg/10.1145/ 3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "——, “Testing database engines via pivoted query synthesis,” in Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 42,
+      "text": "gcovr, https://gcovr.com/, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "“The official code repository of the PHP interpreter,” https://github.com/ php/php-src/, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, 2023, p. 140–149.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 45,
+      "text": "“Dynasm,” https://luajit.org/dynasm.html, 2025. [Online]. Available: https://luajit.org/dynasm.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "G. Klees, A. Ruef, B. Cooper, S. Wei, and M. Hicks, “Evaluating fuzz testing,” in Proceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security, 2018, p. 2123–2138.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "J. Wang, Z. Zhang, S. Liu, X. Du, and J. Chen, “FuzzJIT: Oracleenhanced fuzzing for JavaScript engine JIT compiler,” in 32nd USENIX Security Symposium (USENIX Security 23), 2023, pp. 1865–1882.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "K. Even-Mendoza, A. Sharma, A. F. Donaldson, and C. Cadar, “grayc: Greybox fuzzing of compilers and analysers for C,” in Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, 2023, pp. 1219–1231.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "F. Tuong, M. Omidvar Tehrani, M. Gaboardi, and S. Y. Ko, “Symrustc: A hybrid fuzzer for rust,” in Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, 2023, pp. 1515–1518.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "S. Groß, S. Koch, L. Bernhard, T. Holz, and M. Johns, “Fuzzilli: Fuzzing for javascript jit compiler vulnerabilities.” in NDSS, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "G. Ye, Z. Tang, S. H. Tan, S. Huang, D. Fang, X. Sun, L. Bian, H. Wang, and Z. Wang, “Automated conformance testing for javascript engines via deep compiler fuzzing,” in Proceedings of the 42nd ACMSIGPLAN international conference on programming language design and implementation, 2021, pp. 435–450. 1106",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 7,
+      "text": "Y. Jiang, C. Zhang, B. Ruan, J. Liu, M. Rigger, R. H. Yap, and Z. Liang, “Fuzzing the PHP interpreter via dataflow fusion,” in 34th USENIX Security Symposium (USENIX Security 25), 2025, pp. 6143–6158.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, nov 2020. [Online]. Available: https://doi-org.libproxy1.nus.edu.sg/10.1145/ 3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 41,
+      "text": "——, “Testing database engines via pivoted query synthesis,” in Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 44,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, 2023, p. 140–149.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 3,4,5,6,7,8]",
+      "technique": null,
+      "sentence": "Research efforts [ 3,4,5,6,7,8] have been devoted to uncovering various bugs in the PHP interpreter.",
+      "context_before": "y used server-side scripting languages for web development, powering over 70% of websites on the internet [ 1,2]. PHP offers a wide range of functionalities for web developers, such as database access, session management, and file system operations. The official PHP interpreter features a substantial codebase with over a million lines of code, mainly implemented in C. It is well known that complex software is prone to bugs and vulnerabilities. Given its critical role in web development, detecting bugs in the PHP interpreter is crucial to ensure the security and reliability of web applications.",
+      "context_after": "In particular, the proposed approaches mainly rely on fuzzing techniques dedicated to optimizing input generation for broad testing §Co-primary authors.†Corresponding author.coverage. Most of the above works leverage grammar-guided or semantic-guided test case generation. For example, FlowFusion [ 7], utilizes dataflow-guided test case fusion techniques to generate semantic test cases. We highlight that existing approaches mainly focus on finding explicit security issues, such as crashes or memory errors. They do so by leveraging crashes or sanitizers as weak test oracles. Despite the effecti",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2729,
+      "cited_reference": {
+        "number": 7,
+        "text": "Y. Jiang, C. Zhang, B. Ruan, J. Liu, M. Rigger, R. H. Yap, and Z. Liang, “Fuzzing the PHP interpreter via dataflow fusion,” in 34th USENIX Security Symposium (USENIX Security 25), 2025, pp. 6143–6158.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 7]",
+      "technique": null,
+      "sentence": "For example, FlowFusion [ 7], utilizes dataflow-guided test case fusion techniques to generate semantic test cases.",
+      "context_before": "is well known that complex software is prone to bugs and vulnerabilities. Given its critical role in web development, detecting bugs in the PHP interpreter is crucial to ensure the security and reliability of web applications. Research efforts [ 3,4,5,6,7,8] have been devoted to uncovering various bugs in the PHP interpreter. In particular, the proposed approaches mainly rely on fuzzing techniques dedicated to optimizing input generation for broad testing §Co-primary authors.†Corresponding author.coverage. Most of the above works leverage grammar-guided or semantic-guided test case generation.",
+      "context_after": "We highlight that existing approaches mainly focus on finding explicit security issues, such as crashes or memory errors. They do so by leveraging crashes or sanitizers as weak test oracles. Despite the effectiveness of these approaches in identifying security issues, they are not designed for detecting logic bugs. Logic bugs are dangerous as they can silently lead to incorrect results, thereby compromising the reliability of the PHP interpreter. Finding logic bugs in the PHP interpreter is challenging due to the lack of an effective test oracle. As logic bugs manifest as incorrect computatio",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3103,
+      "cited_reference": {
+        "number": 7,
+        "text": "Y. Jiang, C. Zhang, B. Ruan, J. Liu, M. Rigger, R. H. Yap, and Z. Liang, “Fuzzing the PHP interpreter via dataflow fusion,” in 34th USENIX Security Symposium (USENIX Security 25), 2025, pp. 6143–6158.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 7]",
+      "technique": null,
+      "sentence": "As the PHP program generator, ZendDiff adopts FlowFusion [ 7], the state-of-the-art and only generator built for PHP.",
+      "context_before": "by --SECTION-- markers; more than 30 section types exist [ 36]. Listing 2shows an example that verifies the correctness of DOM-related functions. Key sections include: (i) --TEST-section for a brief description; (ii) --EXTENSIONS-- section lists the required extensions; (iii) the --FILE-- section contains the PHP program; and (iv) the --EXPECT-- section specifies the expected results (a mismatch signals a bug). Existing PHP Program Generator. While test program generation is orthogonal to our approach, it remains a crucial research challenge in the testing programming language implementations.",
+      "context_after": "FlowFusion aims to uncover memory errors within the PHP interpreter through fuzzing. To better explore the PHP execution space, it works by fusing official test-suite cases at the data-flow level: it treats cases from the official suite as seeds, interleaves their data flows, and produces new programs that exercise new execution paths and have not been previously explored. III. APPROACH We aim to detect logic bugs that produce incorrect computation results in the PHP interpreter. Our key insight is to treat PHP’s JIT compilation as a semantics-preserving alternative execution implementation t",
+      "section": "II BACKGROUND",
+      "page": 3,
+      "char_offset": 15853,
+      "cited_reference": {
+        "number": 7,
+        "text": "Y. Jiang, C. Zhang, B. Ruan, J. Liu, M. Rigger, R. H. Yap, and Z. Liang, “Fuzzing the PHP interpreter via dataflow fusion,” in 34th USENIX Security Symposium (USENIX Security 25), 2025, pp. 6143–6158.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 7]",
+      "technique": null,
+      "sentence": "Figure 2 illustrates the overall workflow of ZendDiff :Test Case Generation (1) leverages an off-the-shelf test case generation approach [ 7] to create a large and diverse corpus.",
+      "context_before": ">save XML (),\"\\n\"; echo $doc ->save XML ($comment ),\"\\n\"; var_dump ($comment ->parentNode ); ?> --EXPECT -<?xml version =\"1.0\" ?>..<!-- comment -->NULL 1097 Program State Probing 6Test Case Generation1 … Test Case Test Case Test Case =Dual Verification Non-JIT ResultJIT Result Non-JIT Result’=JIT Result’ ≠ 4 2 Non-JIT Test Case’ Non-JIT Test CasecopyJIT Test Case’ JIT Test Case Bug ReportJIT-Aware Program Mutation 3 Test CasecopyFig. 2: The Overview of ZendDiff ’s Differential Testing Approach We present ZendDiff, a differential testing approach for detecting logic bugs in the PHP interpreter.",
+      "context_after": "Program State Probing (2) injects probes into the generated cases to capture fine-grained program states. These instrumented test cases are then replicated into two groups: one group is processed by JIT-Aware Program Mutation (3) to sufficiently exercise JIT functionality of PHP engine, while the other group acts as non-JIT cases. These two groups form test pairs, which are then passed to the Dual Verification (4), which crosschecks the outputs of the differential executions from both groups to detect potential discrepancies. Finally, ZendDiff generates bug reports upon detecting discrepancie",
+      "section": "III APPROACH",
+      "page": 4,
+      "char_offset": 17894,
+      "cited_reference": {
+        "number": 7,
+        "text": "Y. Jiang, C. Zhang, B. Ruan, J. Liu, M. Rigger, R. H. Yap, and Z. Liang, “Fuzzing the PHP interpreter via dataflow fusion,” in 34th USENIX Security Symposium (USENIX Security 25), 2025, pp. 6143–6158.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 7]",
+      "technique": null,
+      "sentence": "Differential Input Preparation To generate unique and high-quality test cases for differential testing, ZendDiff leverages an existing state-of-the-art PHP test case generation technique, FlowFusion [ 7] (see Section II).",
+      "context_before": "e group is processed by JIT-Aware Program Mutation (3) to sufficiently exercise JIT functionality of PHP engine, while the other group acts as non-JIT cases. These two groups form test pairs, which are then passed to the Dual Verification (4), which crosschecks the outputs of the differential executions from both groups to detect potential discrepancies. Finally, ZendDiff generates bug reports upon detecting discrepancies in differential test results. Each bug report comprises the bug-inducing test case, the observed unexpected result, the expected result, and the reproducing configuration. A.",
+      "context_after": "However, directly applying these cases to differential testing does not work well because they often contain random and unstable behaviors for fuzzing purposes, such as randomized API invocations, variable accesses, and execution configurations. Such random elements can lead to differences in runtime behaviors, causing challenges for reliable differential testing. Hence, ZendDiff performs a processing step to stabilize such generated randomness. Specifically, ZendDiff performs a lightweight code analysis to identify the declared variables and explicitly specifies them in test cases. Similarly",
+      "section": "A Differential Input Preparation",
+      "page": 4,
+      "char_offset": 18850,
+      "cited_reference": {
+        "number": 7,
+        "text": "Y. Jiang, C. Zhang, B. Ruan, J. Liu, M. Rigger, R. H. Yap, and Z. Liang, “Fuzzing the PHP interpreter via dataflow fusion,” in 34th USENIX Security Symposium (USENIX Security 25), 2025, pp. 6143–6158.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 7]",
+      "technique": null,
+      "sentence": "Original test cases generated by FlowFusion [ 7] often invoke dynamic introspection APIs such as get_defined_functions() andget_ defined_vars(), which randomly access internal functions and variables.",
+      "context_before": "sults (line 11-12). Finally, ZendDiff compares the results of the check tests with the original test cases to ensure consistency across different executions. If the results of the check tests do not match those of the original test cases, ZendDiff ignore them as non-deterministic behaviors present (line 14). Otherwise, ZendDiff reports a logic bug, as the discrepancy indicates an inconsistency in the PHP interpreter’s execution behavior (line 15). IV. IMPLEMENTATION We developed our differential testing framework, ZendDiff, with over 2,000 lines of Python code. Processing Generated Test Cases.",
+      "context_after": "While the use of randomness can be useful for a fuzzer, it yields unstable and non-comparable executions that undermine differential testing. ZendDiff refines these cases by eliminating such random and unstable behaviors with fixed targets to ensure reproducible results. State Probes. Our probes internally invokes PHP’s built-in var_dump() to serialize variables into a canonical string that captures both type and value. When dealing with arrays and objects, it recursively traverses nested elements, exposing the full structure of multi-dimensional arrays and complex objects. This detailed outp",
+      "section": "IV IMPLEMENTATION",
+      "page": 5,
+      "char_offset": 27162,
+      "cited_reference": {
+        "number": 7,
+        "text": "Y. Jiang, C. Zhang, B. Ruan, J. Liu, M. Rigger, R. H. Yap, and Z. Liang, “Fuzzing the PHP interpreter via dataflow fusion,” in 34th USENIX Security Symposium (USENIX Security 25), 2025, pp. 6143–6158.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[ 40,41]",
+      "technique": "tlp",
+      "sentence": "Code coverage is a widely adopted metric in evaluating fuzzing and testing approaches [ 40,41], as it provides a clear and quantifiable way to evaluate which parts of the code have been executed during tests.",
+      "context_before": "(sanitizers disabled). Detailed component options are listed in our artifacts. Evaluation Metrics. We evaluate and compare ZendDiff with prior work using three well-defined metrics: (i) Number of logic bugs. The logic bugs we detected manifest as discrepancies between non-JIT and JIT execution results. To ensure accuracy, we employ a deduplication process, where multiple bugs identified with the same result are treated as duplicates and counted only once. This metric serves as the primary indicator of the detection capability and the practicality of the testing techniques. (ii) Code coverage.",
+      "context_after": "In particular, we report line code coverage using gcovr tool [ 42], following the recommended configuration from the official PHP Makefile. (iii) Zend opcodes diversity. This metric emphasizes the ability to explore a wide range of opcode executions. Similar to query plans in database system testing, a broader exploration of Zend opcodes provides deeper insights into the interpreter’s behavior and increases the likelihood of discovering potential vulnerabilities. Experimental Infrastructure. All experiments were conducted on an AMDEPYC 7763 server (64 physical / 128 logical cores at 2.45 GHz,",
+      "section": "V EVALUATION",
+      "page": 6,
+      "char_offset": 30212,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, nov 2020. [Online]. Available: https://doi-org.libproxy1.nus.edu.sg/10.1145/ 3428279",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[ 41,44]",
+      "technique": "pqs",
+      "sentence": "This testing was carried out over a four-month period, following well-established evaluation methodologies for automated bug-detection tools [ 41,44].",
+      "context_before": "of discovering potential vulnerabilities. Experimental Infrastructure. All experiments were conducted on an AMDEPYC 7763 server (64 physical / 128 logical cores at 2.45 GHz, 512 GBRAM) running Ubuntu 22.04. By default, ZendDiff is designed to utilize a moderate amount of computational resources, allocating 32 CPU cores and up to 32 GB of RAM, which enables it to detect various logic bugs within a 24-hour timeframe. A. Discovering Previously Unknown Logic Bugs To uncover previously unknown logic bugs, we intermittently tested the latest PHP interpreter built from the official repository [ 43].",
+      "context_after": "To streamline the analysis of complex test cases and facilitate pinpointing theirroot causes, we employed delta debugging [ 38] to reduce each case to its minimal and bug-inducing form. Furthermore, we cross-checked the reduced test cases with existing issue trackers to prevent the redundant submission of bug reports. Results. Table Isummarizes all confirmed or fixed logic bugs detected by ZendDiff and verified through manual analysis. TheEngine column indicates the affected component: G #marks JIT-only bugs, H #denotes non-JIT bugs, and represents bugs impacting both (as confirmed by the off",
+      "section": "A Discovering Previously Unknown Logic Bugs",
+      "page": 6,
+      "char_offset": 31447,
+      "cited_reference": {
+        "number": 41,
+        "text": "——, “Testing database engines via pivoted query synthesis,” in Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 7]",
+      "technique": null,
+      "sentence": "Improved Effectiveness of ZendDiff We first evaluate ZendDiff ’s improved effectiveness through a comparison with FlowFusion [ 7], the state-of-the-art PHPspecific fuzzer.",
+      "context_before": "diminish the diagnostic value of these warnings. As illustrated in Listing 5, this bug leads to erroneous warning line numbers being produced by both non-JIT and JIT execution modes. The misalignment of line numbers not only introduces inconsistencies but also hampers developers’ ability to effectively trace and rectify source code issues. This bug was identified through our approach and acknowledged by the development team. They traced the root cause to an incorrect usage of the line number from the first instruction in a function, rather than from the actual start of the function itself. B.",
+      "context_after": "Our results show that ZendDiff is capable of uncovering previously undetected logic bugs that FlowFusion fails to detect. Then, to further assess the enhanced capability ofZendDiff ’s test oracle, we compare its results against the official PHP test suite. Comparison with FlowFusion. In principle, ZendDiff provides a complementary oracle for logic bugs compared with the memory/crash oracle typically used in fuzzers. For instance, as the FlowFusion fuzzer relies on memory errors as the bug detection oracle, it does not deal with logic bugs. Evaluations show ZendDiff and FlowFusion achieve comp",
+      "section": "B Improved Effectiveness of ZendDiff",
+      "page": 8,
+      "char_offset": 41504,
+      "cited_reference": {
+        "number": 7,
+        "text": "Y. Jiang, C. Zhang, B. Ruan, J. Liu, M. Rigger, R. H. Yap, and Z. Liang, “Fuzzing the PHP interpreter via dataflow fusion,” in 34th USENIX Security Symposium (USENIX Security 25), 2025, pp. 6143–6158.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 7]",
+      "technique": null,
+      "sentence": "FlowFusion [ 7] merges code semantics from two or more seed programs to generate new fuzzing inputs and effectively finds hundreds of memory errors in the PHP interpreter.",
+      "context_before": "kling new domains with tailored oracles which show practical effectiveness, our work likewise focuses on making differential testing practical for PHP through three key techniques ( i.e.,program state probing, JIT-aware program mutation, and dual verification). Bug Detection in the PHP Engine. Existing approaches mainly leverage fuzzing techniques to detect bugs in the PHP Engine. For instance, LangFuzz [ 3], NAUTILUS [ 4], Gramatron [ 5], and PolyGlot [ 6] focus on memory errors in the PHP interpreter. Reflecta [ 8] gains a diverse set of language features dynamically to reduce manual effort.",
+      "context_after": "However, these approaches primarily rely on crashes or sanitizer alerts to detect bugs, which differs from our focus, logic bugs, that do not cause crashes or sanitizer alerts. To fill this gap, we develop ZendDiff to detect logic bugs in the PHP interpreter through differential testing of JIT and non-JIT execution modes. Bug Discovery in Other Compilers or Interpreters. Existing 1104 solutions for detecting bugs in other compilers and interpreters mainly concentrate on fuzzing techniques, such as C/C++[ 48, 19], Rust[ 24,49], and JavaScript (JS) [ 50,47,17], to mitigate potential cascading",
+      "section": "C Ablation Study of ZendDiff",
+      "page": 10,
+      "char_offset": 59685,
+      "cited_reference": {
+        "number": 7,
+        "text": "Y. Jiang, C. Zhang, B. Ruan, J. Liu, M. Rigger, R. H. Yap, and Z. Liang, “Fuzzing the PHP interpreter via dataflow fusion,” in 34th USENIX Security Symposium (USENIX Security 25), 2025, pp. 6143–6158.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:26:20Z",
+    "is_model_written": true,
+    "summary": "ZendDiff finds logic bugs in the PHP interpreter by differential testing. The authors observe that existing PHP bug-finding targets crashes or sanitizer oracles and misses silent wrong results, and that PHP's JIT compilation mode provides a second implementation of the same specification to compare against. ZendDiff compares JIT and non-JIT execution with program state probing, JIT-aware mutation, and dual verification for non-determinism. It has found 51 previously unknown logic bugs, 37 already fixed.",
+    "narrative": "Almost every citation here is to FlowFusion, a PHP fuzzer co-authored by one of SQLancer's authors, which ZendDiff adopts as its program generator and measures against -- not to SQLancer itself. SQLancer's own papers appear only as methodological references for using code coverage and for a four-month evaluation period. This paper is a good illustration of why an author's other work has to be kept apart from SQLancer's.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "background",
+      "M6": "background",
+      "M7": "incidental",
+      "M8": "incidental",
+      "M9": "background",
+      "M10": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "ZendDiff adopts FlowFusion as its generator, which is a different tool by an overlapping set of authors. Nothing here says SQLancer is used."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_ase63991_2025_00151.json b/_data/papers/paper_doi_10_1109_ase63991_2025_00151.json
new file mode 100644
index 0000000..c981530
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_ase63991_2025_00151.json
@@ -0,0 +1,750 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:05:50Z",
+  "paper": {
+    "id": "paper:doi:10.1109/ase63991.2025.00151",
+    "title": "ARG: Testing Query Rewriters via Abstract Rule Guided Fuzzing",
+    "authors": [
+      "Dawei Li",
+      "Yuxiao Guo",
+      "Qifan Liu",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Chi Zhang",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1109/ase63991.2025.00151",
+    "arxiv_id": null,
+    "s2_paper_id": "eeb659ea2fc76f85e0eb72e54dd83f98ce695d61",
+    "url": "https://doi.org/10.1109/ase63991.2025.00151",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/ase63991.2025.00151",
+      "retrieved_at": "2026-09-07T17:05:50Z",
+      "chars": 61806,
+      "content_sha256": "sha256:8e33ec44aa0326b45cf3b4f8ac5dfbe75ecaa682b0e5be9b74576db2b25d8abc"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2155
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND ANDMOTIVATION",
+        "start": 10045
+      },
+      {
+        "number": "III",
+        "title": "DESIGN",
+        "start": 15928
+      },
+      {
+        "number": "A",
+        "title": "Unified Abstract Rule Construction",
+        "start": 17017
+      },
+      {
+        "number": "B",
+        "title": "Rule-Guide Query Generation",
+        "start": 24556
+      },
+      {
+        "number": "C",
+        "title": "Semantic-Oriented Result Validation",
+        "start": 31798
+      },
+      {
+        "number": "IV",
+        "title": "IMPLEMENTATION",
+        "start": 33959
+      },
+      {
+        "number": "V",
+        "title": "EVALUATION",
+        "start": 35258
+      },
+      {
+        "number": "A",
+        "title": "Evaluation Setup",
+        "start": 35602
+      },
+      {
+        "number": "B",
+        "title": "Rewriter Bug Detection",
+        "start": 36779
+      },
+      {
+        "number": "C",
+        "title": "Comparison with Other Techniques",
+        "start": 42309
+      },
+      {
+        "number": "M",
+        "title": "Y. What is query rewriting? In International Workshop on",
+        "start": 56839
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Mysql query rewrite plugin. https://dev .mysql .com/doc/refman/8 .4/en/ rewriter-query-rewrite-plugin .html, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "APACHE. Calcite. https://github .com/apache/calcite .git, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "BEGOLI, E., CAMACHO-RODR´IGUEZ, J., HYDE, J., MIOR, M. J., ANDLEMIRE, D. Apache calcite: A foundational framework for optimized query processing over heterogeneous data sources. In Proceedings of the 2018 International Conference on Management of Data (2018), pp. 221– 230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "CALVANESE, D., DEGIACOMO, G., LENZERINI, M., ANDVARDI, M. Y. What is query rewriting? In International Workshop on Cooperative Information Agents (2000), Springer, pp. 51–59.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "DING, H., WANG, Z., YANG, Y., ZHANG, D., X U, Z., CHEN, H., PISKAC, R., ANDLI, J. Proving query equivalence using linear integer arithmetic. Proceedings of the ACM on Management of Data 1, 4 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "EVANS, R. B., ANDSAVOIA, A. Differential testing: a new approach to change detection. In The 6th Joint Meeting on European Software Engineering Conference and the ACMSIGSOFT Symposium on the Foundations of Software Engineering: Companion Papers (New York, NY, USA, 2007), ESEC-FSE companion ’07, Association for Computing Machinery, p. 549–552.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "FINANCE, B., ANDGARDARIN, G. A rule-based query rewriter in an extensible dbms. In Proceedings. Seventh International Conference on Data Engineering (1991), IEEE Computer Society, pp. 248–249.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "F U, J., LIANG, J., W U, Z., WANG, M., ANDJIANG, Y. Griffin: Grammar-free dbms fuzzing. In Conference on Automated Software Engineering (ASE’22) (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "G U, Z., SOLIMAN, M. A., ANDWAAS, F. M. Testing the accuracy of query optimizers. In Proceedings of the Fifth International Workshop on Testing Database Systems (2012), pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "IBM. Db2 for linux, unix and windows. https://www .ibm.com/docs/en/ db2/11 .5.x?topic=process-query-rewriting-methods-examples, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "JUNG, J., H U, H., ARULRAJ, J., KIM, T., ANDKANG, W. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB) (Tokyo, Japan, Aug. 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "LEARNED REWRITE. https://github .com/XuanheZhou/ LearnedRewrite .git, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "LIANG, Y., LIU, S., ANDHU, H. Detecting logical bugs of {DBMS} with coverage-based guidance. In 31st USENIX Security Symposium (USENIX Security 22) (2022), pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "LIU, X., ZHOU, Q., ARULRAJ, J., ANDORSO, A. Automatic detection of performance bugs in database systems using equivalent queries. In Proceedings of the 44th International Conference on Software Engineering(2022), pp. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "MANG, Q., FANG, A., Y U, B., CHEN, H., ANDHE, P. Testing graph database systems via equivalent query rewriting. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (New York, NY, USA, 2024), ICSE ’24, Association for Computing Machinery. 1817",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "MCKEEMAN, W. M. Differential testing for software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "NEGRI, M., PELAGATTI, G., ANDSBATTELLA, L. Formal semantics of sql queries. ACM Trans. Database Syst. 16, 3 (Sept. 1991), 513–534.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "RIGGER, M., ANDSU, Z. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (2020), pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 19,
+      "text": "RIGGER, M., ANDSU, Z. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (Nov. 2020).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "SLUTZ, D. R. Massive stochastic testing of sql. In VLDB (1998), vol. 98, Citeseer, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "SQLFORMATTER. sql-formatter. https://github .com/vertical-blank/sqlformatter .git, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "SQLANCER .https://github .com/sqlancer/sqlancer .git, 2025.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 23,
+      "text": "SQLSMITH .https://github .com/anse1/sqlsmith .git, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "SQLSOLVER .https://github .com/SJTU-IPADS/SQLSolver .git, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "TANG, X., W U, S., ZHANG, D., L I, F., ANDCHEN, G. Detecting logic bugs of join optimizations in dbms. Proceedings of the ACM on Management of Data 1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "WANG, Z., ZHOU, Z., YANG, Y., DING, H., H U, G., DING, D., TANG, C., CHEN, H., ANDLI, J. Wetune: Automatic discovery and verification of query rewrite rules. In Proceedings of the 2022 International Conference on Management of Data (2022), pp. 94–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "WETUNE. https://ipads .se.sjtu.edu.cn:1312/opensource/wetune .git, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "WIKIPEDIA. Query rewriting. https://en .wikipedia .org/wiki/Query rewriting, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "XIAOMI. Soar: Sql optimizer and rewriter. https://github .com/xiaomi/ soar, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "ZALEWSKI, M. Afl: American fuzzy lop. https://github .com/google/ AFL, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "ZHONG, R., CHEN, Y., H U, H., ZHANG, H., LEE, W., ANDWU, D. Squirrel: Testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security (2020), pp. 955– 970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "ZHOU, X., JIN, L., SUN, J., ZHAO, X., Y U, X., FENG, J., L I, S., WANG, T., L I, K., ANDLIU, L. Dbmind: A self-driving platform in opengauss. Proceedings of the VLDB Endowment 14, 12 (2021), 2743– 2746.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "ZHOU, X., L I, G., CHAI, C., ANDFENG, J. A learned query rewrite system using monte carlo tree search. Proc. VLDB Endow. 15, 1 (2021), 46–58.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "ZHOU, X., L I, G., W U, J., LIU, J., SUN, Z., ANDZHANG, X. A learned query rewrite system. Proc. VLDB Endow. (2023). 1818",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 18,
+      "text": "RIGGER, M., ANDSU, Z. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (2020), pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 19,
+      "text": "RIGGER, M., ANDSU, Z. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (Nov. 2020).",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 22,
+      "text": "SQLANCER .https://github .com/sqlancer/sqlancer .git, 2025.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In 24 hours, ARG triggered 76% and 1017% more written rules, triggered 13 and 15 more bugs than SQLsmith and SQLancer, respectively.",
+      "context_before": "bstract rules to identify which patterns have been covered. This feedback is then used to dynamically adjust query generation, prioritizing unexplored patterns to avoid redundancy and expose more rewriting logic. We implemented ARG to test four popular query rewrites, namely Apache Calcite, WeTune, SQLSolver, and LearnedRewrite. ARG discovered 38 previously unknown bugs, consisting of 4 crashes, 13 invalid SQL outputs, and 21 semantic deviations. Among them, 19 have been confirmed, while the remaining cases are still under investigation. We also compared ARG against popular DBMS testing tools.",
+      "context_after": "Index Terms —Query Rewriter, Rule Feedback, Bug Detection I. INTRODUCTION Query rewriting is a widely adopted and critically important technique in database management systems (DBMSs) [ 4], [28]. Many DBMSs employ query rewriters as the logical optimization phase of the query optimization, leveraging equivalence rules to transform complex query statements into more efficient yet semantically equivalent forms [ 7], [29], [32]. These query rewriters enhance query performance, leading to substantial savings in computational resource costs for enterprises and generating considerable economic bene",
+      "section": null,
+      "page": 1,
+      "char_offset": 1963,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 22]",
+      "technique": null,
+      "sentence": "Many existing DBMS testing tools [ 22], [23], [31] effectively uncover logic bugs and crashes, but they are primarily designed for broad bug detection rather than targeting query rewrites.",
+      "context_before": "0.0963786}, {0.0963786, 0} -- Rewritten query SELECT CAST (t1.c1AS REAL) ASc0, t10.c1 FROMt1, t1 AS t10; -- Result: {0.09637860208749771, 0}, {0, 0}, {0.09637860208749771,0.0963786}, {0, 0.0963786} Fig. 1. The query to trigger a rewrite bug in Apache Calcite. First, the order of projection operations is incorrectly adjusted. Moreover, an extraneous CAST operation is added to the float-type column c0, converting it to DOUBLE with altered precision. These issues cause nonequivalence. However, there is currently a lack of systematic testing methodologies tailored specifically for query rewrites.",
+      "context_after": "Lacking rewriting semantic awareness, they often fail 18072025 40th IEEE/ACM International Conference on Automated Software Engineering (ASE) 2643-1572/25/$31.00 ©2025 IEEEDOI 10.1109/ASE63991.2025.001512025 40th IEEE/ACM International Conference on Automated Software Engineering (ASE) | 979-8-3503-5733-2/25/$31.00 ©2025 IEEE | DOI: 10.1109/ASE63991.2025.00151 to systematically trigger complex rule combinations, potentially missing critical bugs in the rewriting logic. Moreover, due to the diversity and implementation-specific nature of rewriting rules, these methods can cover only a limite",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 5148,
+      "cited_reference": {
+        "number": 22,
+        "text": "SQLANCER .https://github .com/sqlancer/sqlancer .git, 2025.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In 24 hour experiment on these query rewriters, ARG triggered 76% and 1017% more written rules, triggered 13 and 15 more bugs than SQLsmith and SQLancer, respectively, In summary, we make the following contributions: •We observe that while query rewriters are widely used in DBMSs, bugs persist and can cause serious issues, yet effective testing tools remain lacking.",
+      "context_before": "rewrites, namely Apache Calcite [ 3], WeTune [ 26], SQLSolver [ 5], and LearnedRewrite [ 33], [34]. ARG discovered 38 previously unknown bugs across these query rewriters. Among them, 19 have been confirmed by the rewriter developers, while theremaining cases are still under investigation. The detected bugs result in three major and severe failure symptoms: 4 system crashes, 13 invalid SQL outputs, and 21 query semantic deviations, which causes database service interruptions, execution failures, or incorrect query outputs, respectively. We also compared ARG against popular DBMS testing tools.",
+      "context_after": "•We propose ARG, the first fuzzing-based approach for query rewriters, which extracts abstract rules to guide query generation and trigger more rewriting behaviors. •We uncovered 38 unique bugs in popular query rewriters like Apache Calcite, including 4 crashes, 13 invalid SQL outputs, and 21 query semantic deviations. II. BACKGROUND ANDMOTIVATION Query Rewriter. Query rewriting plays a fundamental role in database query optimization. It transforms initial SQL queries into semantically equivalent but more efficient forms, whether integrated into DBMSs or implemented as standalone components.",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 9354,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Traditional generation-based fuzzers like SQLancer and SQLsmith create diverse SQL statements based on syntax rules but struggle to systematically activate specific rewriting rule combinations due to their random nature.",
+      "context_before": "-free implementation of query rewriters in DBMSs remains a significant challenge. Despite their importance, there is currently a lack of systematic testing methodologies specifically designed for query rewriters. While current DBMS fuzzers have proven effective for general database testing, they fall short when applied to query rewriters due to their limited understanding of internal rewriting logic and rules. Specifically, existing tools often lack a deep understanding of rewriting rules and the internal rewriting logic, making it difficult to identify and trigger precise rewriting scenarios.",
+      "context_after": "Mutation-based tools rely on instrumenting the entire DBMS to gather coverage and runtime information, which is impractical for query rewriters that operate independently in isolated environments. As a result, these tools cannot effectively guide testing for rewriters. Therefore, traditional random testing methods fail to reliably cover complex rewriting paths, limiting their ability to detect bugs and verify query rewriter correctness. Basic Idea of ARG. The key idea of ARG is to use feedback from abstract rules to guide query generation, thereby activating more rewriting logic and enhancing",
+      "section": "II BACKGROUND ANDMOTIVATION",
+      "page": 3,
+      "char_offset": 14414,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To generate DDL statements and enhance abstract rule extraction, we integrated SQLancer to construct dynamic database schemas and populate test data.",
+      "context_before": "of C++ code, built an automated testing framework for the query rewriter, and implemented mechanisms for abstract rule extraction and rule-based feedback. The rule extraction component uses the AST node identity binding mechanism of sql-formatter [ 21], which identifies SQL structures across dialects and tags syntax tree nodes with feature identifiers, enabling precise keyword AST extraction. To implement the rule feedback mechanism, we developed a customized version of SQLsmith to generate DML queries, introducing probabilistic control flags to govern its syntax structure generation strategy.",
+      "context_after": "This combined approach mitigates the limitations of each tool: SQLancer’s native FUZZER lacks sufficient coverage for complex rule patterns, while SQLsmith, despite its strength in generating intricate query structures, does not support adaptive schema generation. To integrate new query rewriters with ARG, developers only need to implement our standardized rewrite interface, which takes a query and schema as input and returns the optimized statement. 1812 V. EVALUATION To evaluate the effectiveness and efficiency of ARG in detecting bugs in query rewriters, we design experiments to answer t",
+      "section": "IV IMPLEMENTATION",
+      "page": 6,
+      "char_offset": 34645,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This combined approach mitigates the limitations of each tool: SQLancer’s native FUZZER lacks sufficient coverage for complex rule patterns, while SQLsmith, despite its strength in generating intricate query structures, does not support adaptive schema generation.",
+      "context_before": "ack. The rule extraction component uses the AST node identity binding mechanism of sql-formatter [ 21], which identifies SQL structures across dialects and tags syntax tree nodes with feature identifiers, enabling precise keyword AST extraction. To implement the rule feedback mechanism, we developed a customized version of SQLsmith to generate DML queries, introducing probabilistic control flags to govern its syntax structure generation strategy. To generate DDL statements and enhance abstract rule extraction, we integrated SQLancer to construct dynamic database schemas and populate test data.",
+      "context_after": "To integrate new query rewriters with ARG, developers only need to implement our standardized rewrite interface, which takes a query and schema as input and returns the optimized statement. 1812 V. EVALUATION To evaluate the effectiveness and efficiency of ARG in detecting bugs in query rewriters, we design experiments to answer the following questions: •Q1: Can ARG detect bugs in real-world query rewriters? •Q2: How does ARG compare with existing techniques? •Q3: How does rule feedback guidance contribute to the performance of ARG? A. Evaluation Setup Test Rewriters. To evaluate the bug de",
+      "section": "IV IMPLEMENTATION",
+      "page": 6,
+      "char_offset": 34795,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Therefore, to evaluate the effectiveness of our approach, we compared ARG with two state-of-the-art SQL generators, SQLsmith and SQLancer, both widely used in the industry for generating large volumes of SQL queries that are fed into rewriters for transformation.",
+      "context_before": "itional fuzzers that generate queries randomly or rely on code coverage,ARG extracts and uses abstract rewriting rules to guide query generation, systematically activating complex rewriting logic. By incorporating schema constraints and semantic context, it produces realistic queries that expose subtle bugs. Additionally, its semantic validation precisely compares original and rewritten query results, identifying errors missed by syntaxbased or coverage-driven methods. C. Comparison with Other Techniques To the best of our knowledge, ARG is the first dedicated tool for testing query rewriters.",
+      "context_after": "During the evaluation, all tools were initialized with an empty database as input, executed under their default configurations, and run in the same environment for 24 hours. As the testing proceeds, the data may vary due to randomness and differences in generation strategies. For a fair comparison, after SQLancer 1814 and SQLsmith generated SQL queries, we sent these queries to the target rewriter to collect coverage data and bug counts. Coverage. Table IIIshows the number of covered branches on four rewrite systems exercised by each tool. ARG significantly outperforms SQLsmith and SQLancer",
+      "section": "C Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 42435,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For a fair comparison, after SQLancer 1814 and SQLsmith generated SQL queries, we sent these queries to the target rewriter to collect coverage data and bug counts.",
+      "context_before": "RG is the first dedicated tool for testing query rewriters. Therefore, to evaluate the effectiveness of our approach, we compared ARG with two state-of-the-art SQL generators, SQLsmith and SQLancer, both widely used in the industry for generating large volumes of SQL queries that are fed into rewriters for transformation. During the evaluation, all tools were initialized with an empty database as input, executed under their default configurations, and run in the same environment for 24 hours. As the testing proceeds, the data may vary due to randomness and differences in generation strategies.",
+      "context_after": "Coverage. Table IIIshows the number of covered branches on four rewrite systems exercised by each tool. ARG significantly outperforms SQLsmith and SQLancer in terms of rule coverage across all four systems. Specifically, ARG covered 18% and 15% more branches in total compared to SQLsmith and SQLancer, respectively. This result demonstrates that ARG explores a broader and deeper range of rewriting logic paths, increasing the likelihood of revealing correctness bugs. TABLE IIINUMBER OFBRANCHES COVERED BYEACH TOOL IN 24HOURS Rewriter SQLsmith SQLancer ARG Calcite 7208 7191 9703 WeTune 2617 2727",
+      "section": "C Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 42976,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "ARG significantly outperforms SQLsmith and SQLancer in terms of rule coverage across all four systems.",
+      "context_before": "ries that are fed into rewriters for transformation. During the evaluation, all tools were initialized with an empty database as input, executed under their default configurations, and run in the same environment for 24 hours. As the testing proceeds, the data may vary due to randomness and differences in generation strategies. For a fair comparison, after SQLancer 1814 and SQLsmith generated SQL queries, we sent these queries to the target rewriter to collect coverage data and bug counts. Coverage. Table IIIshows the number of covered branches on four rewrite systems exercised by each tool.",
+      "context_after": "Specifically, ARG covered 18% and 15% more branches in total compared to SQLsmith and SQLancer, respectively. This result demonstrates that ARG explores a broader and deeper range of rewriting logic paths, increasing the likelihood of revealing correctness bugs. TABLE IIINUMBER OFBRANCHES COVERED BYEACH TOOL IN 24HOURS Rewriter SQLsmith SQLancer ARG Calcite 7208 7191 9703 WeTune 2617 2727 2810 SQLSolver 2869 3045 3119 LearnedRewrite 8721 8995 9661 Total 21415 21958 25293 Improvement 18% ↑ 15%↑ The primary reason for ARG’s improved code coverage is its ability to cover more rewrite rules than",
+      "section": "C Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 43247,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, ARG covered 18% and 15% more branches in total compared to SQLsmith and SQLancer, respectively.",
+      "context_before": "with an empty database as input, executed under their default configurations, and run in the same environment for 24 hours. As the testing proceeds, the data may vary due to randomness and differences in generation strategies. For a fair comparison, after SQLancer 1814 and SQLsmith generated SQL queries, we sent these queries to the target rewriter to collect coverage data and bug counts. Coverage. Table IIIshows the number of covered branches on four rewrite systems exercised by each tool. ARG significantly outperforms SQLsmith and SQLancer in terms of rule coverage across all four systems.",
+      "context_after": "This result demonstrates that ARG explores a broader and deeper range of rewriting logic paths, increasing the likelihood of revealing correctness bugs. TABLE IIINUMBER OFBRANCHES COVERED BYEACH TOOL IN 24HOURS Rewriter SQLsmith SQLancer ARG Calcite 7208 7191 9703 WeTune 2617 2727 2810 SQLSolver 2869 3045 3119 LearnedRewrite 8721 8995 9661 Total 21415 21958 25293 Improvement 18% ↑ 15%↑ The primary reason for ARG’s improved code coverage is its ability to cover more rewrite rules than other tools. Figure 7shows the number of unique rewrite rules exercised by each tool. ARG consistently covers",
+      "section": "C Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 43350,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "TABLE IIINUMBER OFBRANCHES COVERED BYEACH TOOL IN 24HOURS Rewriter SQLsmith SQLancer ARG Calcite 7208 7191 9703 WeTune 2617 2727 2810 SQLSolver 2869 3045 3119 LearnedRewrite 8721 8995 9661 Total 21415 21958 25293 Improvement 18% ↑ 15%↑ The primary reason for ARG’s improved code coverage is its ability to cover more rewrite rules than other tools.",
+      "context_before": "r 1814 and SQLsmith generated SQL queries, we sent these queries to the target rewriter to collect coverage data and bug counts. Coverage. Table IIIshows the number of covered branches on four rewrite systems exercised by each tool. ARG significantly outperforms SQLsmith and SQLancer in terms of rule coverage across all four systems. Specifically, ARG covered 18% and 15% more branches in total compared to SQLsmith and SQLancer, respectively. This result demonstrates that ARG explores a broader and deeper range of rewriting logic paths, increasing the likelihood of revealing correctness bugs.",
+      "context_after": "Figure 7shows the number of unique rewrite rules exercised by each tool. ARG consistently covers more rewrite rules than the other two tools. Specifically, ARG covers a total of 76% and 1017% more abstract rules than SQLsmith and SQLancer, respectively. Furthermore, we also evaluate the coverage of the actual rewriting rules within each rewriter. Table IVshows the number of unique combinations of real rewriting rules exercised by each tool. ARG covers a total of 53% and 476% more unique rewriting rule combinations than SQLsmith and SQLancer, respectively. These results highlight the advantage",
+      "section": "C Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 43613,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, ARG covers a total of 76% and 1017% more abstract rules than SQLsmith and SQLancer, respectively.",
+      "context_before": "a broader and deeper range of rewriting logic paths, increasing the likelihood of revealing correctness bugs. TABLE IIINUMBER OFBRANCHES COVERED BYEACH TOOL IN 24HOURS Rewriter SQLsmith SQLancer ARG Calcite 7208 7191 9703 WeTune 2617 2727 2810 SQLSolver 2869 3045 3119 LearnedRewrite 8721 8995 9661 Total 21415 21958 25293 Improvement 18% ↑ 15%↑ The primary reason for ARG’s improved code coverage is its ability to cover more rewrite rules than other tools. Figure 7shows the number of unique rewrite rules exercised by each tool. ARG consistently covers more rewrite rules than the other two tools.",
+      "context_after": "Furthermore, we also evaluate the coverage of the actual rewriting rules within each rewriter. Table IVshows the number of unique combinations of real rewriting rules exercised by each tool. ARG covers a total of 53% and 476% more unique rewriting rule combinations than SQLsmith and SQLancer, respectively. These results highlight the advantage of ARG’s abstract rule-guided generation strategy, which intentionally targets equivalence-preserving transformation logic rather than relying on random or grammar-driven query synthesis. TABLE IVNUMBER OFUNIQUE REWRITING RULECOMBINATIONS INREWRITER BYE",
+      "section": "C Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 44104,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "ARG covers a total of 53% and 476% more unique rewriting rule combinations than SQLsmith and SQLancer, respectively.",
+      "context_before": "l 21415 21958 25293 Improvement 18% ↑ 15%↑ The primary reason for ARG’s improved code coverage is its ability to cover more rewrite rules than other tools. Figure 7shows the number of unique rewrite rules exercised by each tool. ARG consistently covers more rewrite rules than the other two tools. Specifically, ARG covers a total of 76% and 1017% more abstract rules than SQLsmith and SQLancer, respectively. Furthermore, we also evaluate the coverage of the actual rewriting rules within each rewriter. Table IVshows the number of unique combinations of real rewriting rules exercised by each tool.",
+      "context_after": "These results highlight the advantage of ARG’s abstract rule-guided generation strategy, which intentionally targets equivalence-preserving transformation logic rather than relying on random or grammar-driven query synthesis. TABLE IVNUMBER OFUNIQUE REWRITING RULECOMBINATIONS INREWRITER BYEACH TOOL IN 24HOURS Rewriter SQLsmith SQLancer ARG Calcite 7548 1824 11544 WeTune 289 44 324 SQLSolver 370 56 402 LearnedRewrite 814 469 1524 Total 9021 2393 13794 Improvement 53% ↑ 476%↑ Triggered Bugs. Table Vpresents the number of unique rewriter-specific bugs detected by each tool. As shown in the table",
+      "section": "C Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 44407,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "TABLE IVNUMBER OFUNIQUE REWRITING RULECOMBINATIONS INREWRITER BYEACH TOOL IN 24HOURS Rewriter SQLsmith SQLancer ARG Calcite 7548 1824 11544 WeTune 289 44 324 SQLSolver 370 56 402 LearnedRewrite 814 469 1524 Total 9021 2393 13794 Improvement 53% ↑ 476%↑ Triggered Bugs.",
+      "context_before": "017% more abstract rules than SQLsmith and SQLancer, respectively. Furthermore, we also evaluate the coverage of the actual rewriting rules within each rewriter. Table IVshows the number of unique combinations of real rewriting rules exercised by each tool. ARG covers a total of 53% and 476% more unique rewriting rule combinations than SQLsmith and SQLancer, respectively. These results highlight the advantage of ARG’s abstract rule-guided generation strategy, which intentionally targets equivalence-preserving transformation logic rather than relying on random or grammar-driven query synthesis.",
+      "context_after": "Table Vpresents the number of unique rewriter-specific bugs detected by each tool. As shown in the table, ARG detects 13 and 15 more bugs than SQLsmith and SQLancer, respectively. SQLsmith primarily generates standalone SELECT statements through random expression synthesis, which limits its ability to cover complex rewrite Fig. 7. Results of Different Tools in Extracting Abstract Rules under a 24-Hour Experiment. We tracked the growth in the number of unique abstract rules extracted by SQLancer, SQLsmith, and ARG during a 24hour experiment. ARG demonstrated a significant advantage in abstract",
+      "section": "C Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 44750,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As shown in the table, ARG detects 13 and 15 more bugs than SQLsmith and SQLancer, respectively.",
+      "context_before": "QLancer, respectively. These results highlight the advantage of ARG’s abstract rule-guided generation strategy, which intentionally targets equivalence-preserving transformation logic rather than relying on random or grammar-driven query synthesis. TABLE IVNUMBER OFUNIQUE REWRITING RULECOMBINATIONS INREWRITER BYEACH TOOL IN 24HOURS Rewriter SQLsmith SQLancer ARG Calcite 7548 1824 11544 WeTune 289 44 324 SQLSolver 370 56 402 LearnedRewrite 814 469 1524 Total 9021 2393 13794 Improvement 53% ↑ 476%↑ Triggered Bugs. Table Vpresents the number of unique rewriter-specific bugs detected by each tool.",
+      "context_after": "SQLsmith primarily generates standalone SELECT statements through random expression synthesis, which limits its ability to cover complex rewrite Fig. 7. Results of Different Tools in Extracting Abstract Rules under a 24-Hour Experiment. We tracked the growth in the number of unique abstract rules extracted by SQLancer, SQLsmith, and ARG during a 24hour experiment. ARG demonstrated a significant advantage in abstract rule extraction capability. rules that often require specific structural patterns or contextual constraints. Similarly, SQLancer focuses on generating SQL queries guided by predef",
+      "section": "C Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 45102,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We tracked the growth in the number of unique abstract rules extracted by SQLancer, SQLsmith, and ARG during a 24hour experiment.",
+      "context_before": "Rewriter SQLsmith SQLancer ARG Calcite 7548 1824 11544 WeTune 289 44 324 SQLSolver 370 56 402 LearnedRewrite 814 469 1524 Total 9021 2393 13794 Improvement 53% ↑ 476%↑ Triggered Bugs. Table Vpresents the number of unique rewriter-specific bugs detected by each tool. As shown in the table, ARG detects 13 and 15 more bugs than SQLsmith and SQLancer, respectively. SQLsmith primarily generates standalone SELECT statements through random expression synthesis, which limits its ability to cover complex rewrite Fig. 7. Results of Different Tools in Extracting Abstract Rules under a 24-Hour Experiment.",
+      "context_after": "ARG demonstrated a significant advantage in abstract rule extraction capability. rules that often require specific structural patterns or contextual constraints. Similarly, SQLancer focuses on generating SQL queries guided by predefined oracles and grammar constraints, making it difficult to explore the full spectrum of rewrite logic, especially those involving equivalence-preserving transformations. In contrast, ARG systematically identifies and targets rewrite rules through its abstract-rule-guided generation strategy. By analyzing rewriting behavior and validating semantic equivalence betw",
+      "section": "C Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 45436,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Similarly, SQLancer focuses on generating SQL queries guided by predefined oracles and grammar constraints, making it difficult to explore the full spectrum of rewrite logic, especially those involving equivalence-preserving transformations.",
+      "context_before": "G detects 13 and 15 more bugs than SQLsmith and SQLancer, respectively. SQLsmith primarily generates standalone SELECT statements through random expression synthesis, which limits its ability to cover complex rewrite Fig. 7. Results of Different Tools in Extracting Abstract Rules under a 24-Hour Experiment. We tracked the growth in the number of unique abstract rules extracted by SQLancer, SQLsmith, and ARG during a 24hour experiment. ARG demonstrated a significant advantage in abstract rule extraction capability. rules that often require specific structural patterns or contextual constraints.",
+      "context_after": "In contrast, ARG systematically identifies and targets rewrite rules through its abstract-rule-guided generation strategy. By analyzing rewriting behavior and validating semantic equivalence between original and rewritten queries, ARG can effectively uncover subtle logic violations, invalid rule applications, and rewriter-internal crashes that are missed by general-purpose SQL generators. TABLE VNUMBER OFTRIGGER BUGS BY EACHTOOL IN 24HOURS Rewriter SQLsmith SQLancer ARG Calcite 9 6 12 WeTune 5 8 11 SQLSolver 6 4 8 LearnedRewrite 5 5 7 Total 25 23 38 Increment 13 ↑ 15↑ D. Effectiveness of Rule",
+      "section": "C Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 45728,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "TABLE VNUMBER OFTRIGGER BUGS BY EACHTOOL IN 24HOURS Rewriter SQLsmith SQLancer ARG Calcite 9 6 12 WeTune 5 8 11 SQLSolver 6 4 8 LearnedRewrite 5 5 7 Total 25 23 38 Increment 13 ↑ 15↑ D.",
+      "context_before": "nerating SQL queries guided by predefined oracles and grammar constraints, making it difficult to explore the full spectrum of rewrite logic, especially those involving equivalence-preserving transformations. In contrast, ARG systematically identifies and targets rewrite rules through its abstract-rule-guided generation strategy. By analyzing rewriting behavior and validating semantic equivalence between original and rewritten queries, ARG can effectively uncover subtle logic violations, invalid rule applications, and rewriter-internal crashes that are missed by general-purpose SQL generators.",
+      "context_after": "Effectiveness of Rule-Guided Strategy To assess the effectiveness of the abstract rule guided fuzzing strategy, we implemented a variant of ARG, denoted as ARG-, which disables the rule feedback component. ARGgenerates queries using unguided, randomized clause synthesis based solely on grammar structure, without considering ruletriggering feedback or structural transformations observed during rewriting. We compare ARG and ARGacross all four evaluated query rewriters over 24 hours. 1815 Table VIshows the number of rewrite rules exercised by each approach after 24 hours. Across all four rewri",
+      "section": "C Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 46362,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 22] leverages predefined syntax tree models to produce grammatically valid SQL 1816 queries, while also creating schemas and populating the database with data.",
+      "context_before": "tion (IR) based on SQL grammar rules and creates syntax-valid SQL queries using mutation techniques from AFL’s core engine with coverage feedback, effectively exposing crashes in DBMS. SQLRight [ 13] employs syntax tree mutation for test case generation and adapts inputs based on coverage feedback. GRIFFIN [8] introduces a grammar-free mutation method to generate queries. Generation-based fuzzers generate queries using predefined templates or models. For example, SQLsmith [ 23] constructs large numbers of SELECT statements by populating abstract syntax trees (ASTs) with metadata to test DBMSs.",
+      "context_after": "APOLLO [11] focuses more on generating complex or adversarial SQL structures that are likely to degrade the performance of the DBMS. While existing approaches are effective, they face limitations when applied to standalone query rewriters. Neither mutation-based nor generation-based methods can sufficiently activate internal rewrite rules or thoroughly test the complex behaviors of rewriters. In contrast, ARG guides test case generation through abstract rules, exploring more rewrite logic within the rewriter. Differential Testing. Differential testing validates DBMS by executing the same test",
+      "section": "C Comparison with Other Techniques",
+      "page": 10,
+      "char_offset": 52236,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [ 18] converts an optimizable query into a non-optimizable form of the query, then identifies logic bugs by comparing their execution results.",
+      "context_before": "sues by comparing the execution durations of identical SQL queries across multiple versions of the target DBMS. Differential testing is difficult to directly apply to query rewriters. Substantial rewriting logic differences cause the same query to yield diverse outputs, rendering traditional differential testing ineffective. ARG focuses on checking the semantic equivalence between the original and rewritten query, which shares similarity with differential testing. Metamorphic Testing. Recent works have also adopted metamorphic testing to construct semantically equivalent queries to test DBMSs.",
+      "context_after": "Similarly, TLP [ 19] employs ternary logic to generate three equivalent queries combined via UNION operations. Amoeba [ 14] targets DBMS performance anomalies by crafting logically equivalent queries and examining variations in their runtime behavior TQS [ 25] detects logical bugs by generating semantically equivalent multi-table join query pairs and checking for inconsistent execution results. GRev [ 15] automatically tests graph databases by rewriting queries into logically equivalent forms based on a unified graph abstraction, successfully uncovering multiple previously unknown vulnerabili",
+      "section": "C Comparison with Other Techniques",
+      "page": 11,
+      "char_offset": 54020,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Similarly, TLP [ 19] employs ternary logic to generate three equivalent queries combined via UNION operations.",
+      "context_before": "directly apply to query rewriters. Substantial rewriting logic differences cause the same query to yield diverse outputs, rendering traditional differential testing ineffective. ARG focuses on checking the semantic equivalence between the original and rewritten query, which shares similarity with differential testing. Metamorphic Testing. Recent works have also adopted metamorphic testing to construct semantically equivalent queries to test DBMSs. NoREC [ 18] converts an optimizable query into a non-optimizable form of the query, then identifies logic bugs by comparing their execution results.",
+      "context_after": "Amoeba [ 14] targets DBMS performance anomalies by crafting logically equivalent queries and examining variations in their runtime behavior TQS [ 25] detects logical bugs by generating semantically equivalent multi-table join query pairs and checking for inconsistent execution results. GRev [ 15] automatically tests graph databases by rewriting queries into logically equivalent forms based on a unified graph abstraction, successfully uncovering multiple previously unknown vulnerabilities. ARG differs from these tools by treating the rewritten query as output and directly verifying semantic eq",
+      "section": "C Comparison with Other Techniques",
+      "page": 11,
+      "char_offset": 54169,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M7"
+      ],
+      "describes_as_state_of_the_art": [
+        "M7"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:15:22Z",
+    "is_model_written": true,
+    "summary": "ARG fuzzes query rewriters, the components that transform a query into a faster but semantically equivalent form. The authors argue that general DBMS testing tools cover only a limited subset of rewrite scenarios given the diversity of rewrite rules. ARG uses abstract rules -- a unified representation of the AST patterns and constraints that trigger a rewrite, plus the resulting transformation -- as coverage feedback, steering generation toward patterns not yet exercised. Testing Apache Calcite, WeTune, SQLSolver and LearnedRewrite it found 38 previously unknown bugs.",
+    "narrative": "ARG integrates SQLancer to construct dynamic database schemas and populate them, and then measures itself against it, reporting 1017% more rewrite rules triggered and 15 more bugs in 24 hours.",
+    "roles": {
+      "M1": "result_comparison",
+      "M2": "background",
+      "M3": "result_comparison",
+      "M4": "motivation",
+      "M5": "reuse_component"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M5",
+            "sentence": "To generate DDL statements and enhance abstract rule extraction, we integrated SQLancer to construct dynamic database schemas and populate test data.",
+            "section": "IV IMPLEMENTATION",
+            "page": 6
+          }
+        ],
+        "reasoning": "M5 states SQLancer was integrated to construct dynamic database schemas and populate them, so a generation component is reused while the rewrite-rule fuzzing is ARG's own.",
+        "reuse_kind": "generator"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Abstract rule guided fuzzing is ARG's contribution; no SQLancer oracle is generalised."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M1",
+          "M3"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "In 24 hours, ARG triggered 76% and 1017% more written rules, triggered 13 and 15 more bugs than SQLsmith and SQLancer, respectively.",
+            "section": null,
+            "page": 1
+          },
+          {
+            "mention_id": "M3",
+            "sentence": "In 24 hour experiment on these query rewriters, ARG triggered 76% and 1017% more written rules, triggered 13 and 15 more bugs than SQLsmith and SQLancer, respectively, In summary, we make the following contributions: •We observe that while query rewriters are widely used in DBMSs, bugs persist and can cause serious issues, yet effective testing tools remain lacking.",
+            "section": "I INTRODUCTION",
+            "page": 2
+          }
+        ],
+        "reasoning": "ARG reports triggering 1017% more rewrite rules and finding 15 more bugs than SQLancer over 24 hours."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M4 calls SQLancer a traditional generation-based fuzzer and argues it struggles on rewrite rules."
+      }
+    },
+    "disagreements": [
+      "A state-of-the-art pattern fired on M7, which was outside the mentions read; those read describe SQLancer as traditional rather than leading."
+    ],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_ase63991_2025_00322.json b/_data/papers/paper_doi_10_1109_ase63991_2025_00322.json
new file mode 100644
index 0000000..51ee22d
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_ase63991_2025_00322.json
@@ -0,0 +1,227 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:36Z",
+  "paper": {
+    "id": "paper:doi:10.1109/ase63991.2025.00322",
+    "title": "LLM-based Dynamic Differential Testing for Database Connectors with Reinforcement Learning-Guided Prompt Selection",
+    "authors": [
+      "C. Lyu",
+      "Minghao Zhao",
+      "Yanhao Wang",
+      "Liang Jie"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1109/ase63991.2025.00322",
+    "arxiv_id": "2506.11870",
+    "s2_paper_id": "f3e1f353f63a93021ee0ea3b535187585f6be98c",
+    "url": "https://doi.org/10.1109/ase63991.2025.00322",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2506.11870",
+      "retrieved_at": "2026-09-09T01:30:36Z",
+      "chars": 22798,
+      "content_sha256": "sha256:d0736d3f99ea7d7a22fb5b89dec9eaf0015318139e51202ddb941345f2b85b62"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 5,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Djallel Bouneffouf. 2016. Finite-time analysis of the multi-armed bandit problem with known trend. In CEC. 2543–2549.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Ziyu Cui, Wensheng Dou, Yu Gao, Rui Yang, Yingying Zheng, Jiansen Song, Yuan Feng, and Jun Wei. 2025. Simple Testing Can Expose Most Critical Transaction Bugs: Understanding and Detecting Write-Specific Serializability Violations in Database Systems. Proc. VLDB Endow. 18, 8 (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2024. Coni: Detecting Database Connector Bugs via State-Aware Test Case Generation. In ICSE. 26–37.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free DBMS fuzzing. In ASE. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs. In EuroSys. 1061–1076.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Xinyi Hou, Yanjie Zhao, Yue Liu, Zhou Yang, Kailong Wang, Li Li, Xiapu Luo, David Lo, John Grundy, and Haoyu Wang. 2024. Large language models for software engineering: A systematic literature review. ACM Trans. Softw. Eng. Methodol. 33, 8 (2024), 1–79.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Juyong Jiang, Fan Wang, Jiasi Shen, Sungju Kim, and Sunghun Kim. 2024. A survey on large language models for code generation. arXiv:2406.00515 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. InUSENIX Security. 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "OceanBase Team. 2025. OceanBase Connector/J. https://github.com/oceanbase/ obconnector-j.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Oracle. 2014. ResultSet.beforeFirst() Method. https://docs.oracle.com/javase/8/ docs/api/java/sql/ResultSet.html#beforeFirst--.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Oracle. 2025. MySQL Connector/J. https://github.com/mysql/mysql-connector-j.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Fei Qi, Yingnan Hou, Ning Lin, Shanshan Bao, and Nuo Xu. 2024. A Survey of Testing Techniques Based on Large Language Models. In ICCMT. 280–284.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In OSDI. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "Jiansen Song, Wensheng Dou, Yingying Zheng, Yu Gao, Ziyu Cui, Wei Wang, and Jun Wei. 2025. Detecting Schema-Related Logic Bugs in Relational DBMSs via Equivalent Database Construction. Proc. VLDB Endow. 18, 7 (2025), 2281–2294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Richard S. Sutton and Andrew G. Barto. 2018. Reinforcement learningan introduction, 2nd Edition. MIT Press.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Junjie Wang, Yuchao Huang, Chunyang Chen, Zhe Liu, Song Wang, and Qing Wang. 2024. Software Testing With Large Language Models: Survey, Landscape, and Vision. IEEE Trans. Softw. Eng. 50, 04 (2024), 911–936. 4",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing database management systems with language validityand coverage feedback. In CCS. 955–970. 5",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 5,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs. In EuroSys. 1061–1076.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 13,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In OSDI. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 2–5,8,13,14,17]",
+      "technique": "pqs",
+      "sentence": "Notably, while these fuzzing techniques have proven highly successful for DBMS testing, their effectiveness remains constrained for connectors [ 2–5,8,13,14,17].",
+      "context_before": "formats. Although this abstraction layer significantly enhances development efficiency, any inherent defects in these connectors may propagate system-wide failures. Thus, it is essential to ensure their reliability and correctness. Unfortunately, detecting vulnerabilities of database connectors is challenging. Unlike traditional software defects, connector vulnerabilities exhibit subtle behavioral patterns. As a result, traditional fuzzing techniques exhibit limited effectiveness in testing database connectors due to their inability to handle protocol-specific syntax and stateful interactions.",
+      "context_after": "This is because existing fuzzers primarily generate equivalent SQL queries, whereas the connectors bypass rather than execute them. Moreover, thestatic nature of conventional fuzzing renders it ineffective for comprehensively testing connector logic – the generated queries typically exercise only a limited subset of interfaces and achieve insufficient branch coverage. What is even worse, certain connector vulnerabilities are scenariospecific or originate from flawed implementation strategies, making such bugs significantly harder to detect. For example, applications often migrate data from on",
+      "section": null,
+      "page": 1,
+      "char_offset": 3768,
+      "cited_reference": {
+        "number": 13,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In OSDI. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-09T01:33:49Z",
+    "is_model_written": true,
+    "summary": "This work applies large language models to differential testing of database connectors, comparing how MySQL's and OceanBase's JDBC drivers behave on the same calls. A reinforcement-learning policy chooses which prompt to issue next, so generation concentrates where inconsistencies are being found. It reports 16 confirmed bugs and unsafe implementations -- seven in MySQL Connector/J, three bugs and six specification deviations in OceanBase -- including a beforeFirst() call that raises SQLException on one connector and succeeds silently on the other.",
+    "narrative": "PQS is cited among the DBMS fuzzing techniques whose success does not carry over to connectors. The paper's argument is that those fuzzers generate equivalent SQL queries, while a connector largely bypasses query semantics -- what matters is the JDBC interface, its configuration properties and its stateful interactions. So SQLancer's line of work sets up the gap rather than being used or measured against: the reproducers here are JDBC call sequences, not queries.",
+    "roles": {
+      "M1": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; the test cases are JDBC call sequences generated by a language model."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer oracle is extended. The oracle is differential comparison between two connectors."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation is background about why query-level fuzzing does not transfer to connectors; no run against SQLancer is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The fuzzing techniques are called highly successful for DBMS testing, which is praise for the field rather than a claim that SQLancer is the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_compsac57700_2023_00273.json b/_data/papers/paper_doi_10_1109_compsac57700_2023_00273.json
new file mode 100644
index 0000000..e50af4e
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_compsac57700_2023_00273.json
@@ -0,0 +1,92 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:01:52Z",
+  "paper": {
+    "id": "paper:doi:10.1109/compsac57700.2023.00273",
+    "title": "Detecting Hidden Failures of DBMS: A Comprehensive Metamorphic Relation Output Patterns Approach",
+    "authors": [
+      "M. Tang",
+      "T. Tse",
+      "Z. Zhou"
+    ],
+    "year": 2023,
+    "venue": "Annual International Computer Software and Applications Conference",
+    "doi": "10.1109/compsac57700.2023.00273",
+    "arxiv_id": null,
+    "s2_paper_id": "e9db272057833ba69d1c6e9ecbb992af02ce25a9",
+    "url": "https://doi.org/10.1109/compsac57700.2023.00273",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/compsac57700.2023.00273",
+      "retrieved_at": "2026-09-07T17:01:52Z",
+      "chars": 124887,
+      "content_sha256": "sha256:1a401396766ddbfd931069fd7214266d8f7892ce2b03e9afeff5e95a2c5df128"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 6,
+    "has_outline": false,
+    "sections": []
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:23:30Z",
+    "is_model_written": true,
+    "summary": "This work detects hidden failures in DBMSs using a comprehensive set of metamorphic relations over output patterns -- relations between the outputs of related queries that must hold, so a violation exposes a fault that produces plausible-looking but wrong results.",
+    "narrative": "No SQLancer mention could be extracted. Neither the tool nor any of its technique names appears anywhere in the document text, and no bibliography parsed. Given the subject -- metamorphic relations for finding logic bugs in DBMSs -- a relationship is likely, so the absence should be read as an extraction limit rather than as evidence of none.",
+    "roles": {},
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      },
+      "extends_technique": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      },
+      "compares_with": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      }
+    },
+    "disagreements": [],
+    "unresolved": [
+      "The document text contains no occurrence of SQLancer or any technique name, and no bibliography parsed. For a paper on metamorphic DBMS testing that is more likely an extraction failure than a real absence; the PDF's text layer should be re-examined."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_compsac61105_2024_00141.json b/_data/papers/paper_doi_10_1109_compsac61105_2024_00141.json
new file mode 100644
index 0000000..bd4c499
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_compsac61105_2024_00141.json
@@ -0,0 +1,567 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:54:55Z",
+  "paper": {
+    "id": "paper:doi:10.1109/compsac61105.2024.00141",
+    "title": "SQLPass: A Semantic Effective Fuzzing Method for DBMS",
+    "authors": [
+      "Yu Li",
+      "Yixiao Yang",
+      "Yong Guan",
+      "Zhiping Shi",
+      "Rui Wang"
+    ],
+    "year": 2024,
+    "venue": "Annual International Computer Software and Applications Conference",
+    "doi": "10.1109/compsac61105.2024.00141",
+    "arxiv_id": null,
+    "s2_paper_id": "83c63c4935d91200817bfc19905800431df70230",
+    "url": "https://doi.org/10.1109/compsac61105.2024.00141",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/compsac61105.2024.00141",
+      "retrieved_at": "2026-09-08T14:54:55Z",
+      "chars": 51343,
+      "content_sha256": "sha256:eba19c4fbb88b9fdef9577b3d99e6b687e766b89f17c14da42d94a01794656b1"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 10,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2230
+      },
+      {
+        "number": "II",
+        "title": "PROBLEM DEFINITION",
+        "start": 7203
+      },
+      {
+        "number": "A",
+        "title": "Challenges of DBMS Testing",
+        "start": 7226
+      },
+      {
+        "number": "B",
+        "title": "Our Approach",
+        "start": 8866
+      },
+      {
+        "number": "III",
+        "title": "DESIGN",
+        "start": 13901
+      },
+      {
+        "number": "A",
+        "title": "Building of Subtree Libraries",
+        "start": 16347
+      },
+      {
+        "number": "B",
+        "title": "Mutation of Weak Semantic Correlation Nodes",
+        "start": 17028
+      },
+      {
+        "number": "IV",
+        "title": "IMPLEMENTATION",
+        "start": 30476
+      },
+      {
+        "number": "V",
+        "title": "EXPERIMENTAL EVALUATION",
+        "start": 31681
+      },
+      {
+        "number": "A",
+        "title": "Experimental Environment",
+        "start": 32043
+      },
+      {
+        "number": "B",
+        "title": "Experimental Comparison",
+        "start": 32627
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "DBSEC. https://www.dbsec.cn/resource/detail/98.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Y. Zhang, S. Tong, L. Cheng, et al, “Evaluation of fuzzing improving techniques ,” Computer Systems Applications, vol. 31, no. 10, pp. 1-14, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "V. Ma nes, H. H a n, C. H a n, e t al. T he ar t, sc ie nc e, a n d e ng in eer ing of fuzzing: a survey. arXiv preprint arXiv: 1812.00140, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "H. Liang, X. Pei, X. Jia, et al, “Fuzzing: state of the art ,” IEEE Transactions on Reliability, vol. 67, no. 3, pp. 1199-1218, 201 8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "J. Li, B. Zhao, C. Zhang, “Fuzzing: a survey ,” Cybersecurity, vol. 1, no. 1, pp. 6, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "l c a m t u f, “ A m e r i c a n f u z z y l o p ( a f l ), ” 2 0 1 7. [ O n l i n e ]. A v a i l a b l e: http://lcamtuf.coredump.cx/afl/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Abstract syntax tree. https://en.wikipedia.org/wiki/Abstract_sy nta-x_tree.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "SQLite. https://www.sqlite.org/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "MariaDB. https://mariadb.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "A. Seltenreich, B. Tang, and S. Mullender, “Sqlsmith: a random sql query generator,” 2018. [Online]. Available: https://github.com/anse1 /sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis ,” in Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation. Alberta, Canada, pp. 667-682, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 12,
+      "text": "D. R. Slutz, “Massive stochastic testing of SQL,” in VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, New York, USA. Morgan Kaufmann, pp. 618-622, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Autom atic Detection and Diagnosis of Performance Regressions in Database Sys tems (to appear),” in Proceedings of the 46th International Con ference on Very Large Data Bases (VLDB), Tokyo, Japan, Aug. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "R. Z h o n g, Y. C h e n, H. H u, H. Z h a n g, W. L e e, a n d D. W u, “ S q u i r r e l: Testing database management systems with language validity and coverage feedback,” in The 2020 ACM Conference on Computer and Communications Security. UT, USA, pp. 955-970, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "M. Wang, Z. Wu, X. Xu, et al, “Industry practice of coverage-guided enterprise-level DBMS fuzzing ,” in Proceedings of the 2021 IEEE/ACM 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP). online, pp. 328-337, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Z. Wu, J. Liang, M. Wang, et al, “Unicorn: Detect runtime errors in timeseries databases with hybrid input synthesis ,” in Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis. online, pp. 251-262, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "J. Liang, Y. Chen, Z. Wu, et al, “Sequence-oriented DBMS fuzzing ,” in Proceedings of the Engineering (ICDE). California, USA, pp. 668-681, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "J. Fu, J. Liang, Z. Wu, et al, “Griffin: Grammar-free DBMS fuzzing ,” in Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. New York, USA, pp. 1-12, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "I. Yun, S. Lee, M. Xu, et al, “QSYM: A practical concolic execution engine tailored for hybrid fuzzing ” in Proceedings of the 27th USENIX Security Symposium. Maryland, USA, pp. 745-761, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "C. Aschermann, S. Schumilo, T. Blazytko, et al, “REDQUEEN: Fuzzing with input-to-state correspondence ,”in Proceedings of the Network and Distributed Systems Security (NDSS) Symposium. California, USA, pp. 1-15, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "A library for coverage-guided fuzz testing. http://llvm.org/doc s/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "D. Chamberlin and R. Boyce, “Sequel: A structured english query language ,” in Proceedings of the 1974 ACMSIGFIDET workshop on Data description, access and control. Michigan, USA, pp. 249-26 4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "SQL-DDL, DQL, DML, DCL and TCL Commands. https://www.geeksforgeeks.org/sql-ddl-dql-dml-dcl-tcl-commands/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Processing a SQL statement. https://learn.microsoft.com/enus/sql/odbc/reference/processing-a-sql-statement?view=sql-serve r-ver15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "E. Lo, C. Binnig, D. Kossmann, et al, “A framework for testing DBMS features ,” The VLDB Journal, vol. 19, no. 2, pp. 203-230, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Antlr. http://www.antlr.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Antlr’s grammar list for different languages. https://github.com/antlr/grammars-v4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "J. Tuya, M. Suárez-Cabal and C. Riva, “Mutating database queries. Information and Software Technology, ” vol. 49, no. 4, pp. 398-417,2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "V. Kuleshov and D. Precup. Algorithms for multi-armed bandit pr oblems. arXiv preprint arXiv: 1402.6028, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "MySQL. https://www.mysql.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "PostgreSQL. https://www.postgresql.org/. CREATE TABLE t1(c1,c2); CREATE TABLE t2(c1,c2); WITH t2(c1) AS (SELECT 111) INSERT INTO t1 (c1, c2) SELECT c1, 123 FROM t2; CREATE TABLE t1(c1,c2); CREATE TABLE t2(c1,c2); WITH t2(c1) AS (SELECT 1 UNION ALL SELECT c1+1 FROM t2) INSERT INTO t1 (c1, c2) SELECT c1, 123 FROM t2;select_stmt subtree ˖ SELECT 1 UNION ALL SELECT c1+1 FROM t2initial testcase crash-t",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 11,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis ,” in Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation. Alberta, Canada, pp. 667-682, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 31,
+      "text": "PostgreSQL. https://www.postgresql.org/. CREATE TABLE t1(c1,c2); CREATE TABLE t2(c1,c2); WITH t2(c1) AS (SELECT 111) INSERT INTO t1 (c1, c2) SELECT c1, 123 FROM t2; CREATE TABLE t1(c1,c2); CREATE TABLE t2(c1,c2); WITH t2(c1) AS (SELECT 1 UNION ALL SELECT c1+1 FROM t2) INSERT INTO t1 (c1, c2) SELECT c1, 123 FROM t2;select_stmt subtree ˖ SELECT 1 UNION ALL SELECT c1+1 FROM t2initial testcase crash-triggering testcase 1044",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "57b, SQLsmith, SQLancer, and Squirrel.",
+      "context_before": "mpares and evaluates the experimental results of SQLPass using and disabling operator selection algorithm on four popular DBMS, as well as SQLPass and related benchmark testing tools. Finally, it introduces the situation of SQLPass discovering bugs. A. Experimental Environment The experiment in this paper is completed on a computer equipped with a 64 bit 8-core Intel Core i7-4790 Ubuntu 20.04.6LTS CPU@3.60GHz 8Gib of memory and 1TB of disk capacity. The target DBMS selected in this paper are multiple versions of SQLite, MySQL, MariaDB, and PostgreSQL. The benchmark DBMS testing tools are AFL2.",
+      "context_after": "It is worth noting that in the experimental process of this paper, the seed test cases used fo r testing the same DBMS using mutation-based testing tools AFL, Squirrel, and SQLPass are the same. B. Experimental Comparison a) Comparison with Disab led Operator Selection Algorithm: To evaluate the contribution of the “optimal ” operator selection algorithm proposed in this paper to the performance of SQLPass, we compare the code (line) coverage of SQLite3 [8], MySQL [30], MariaDB [9], and PostgreSQL [31] when the “optimal ” operator selection algorithm was disabled and used. As shown in Table Ⅲ",
+      "section": "A Experimental Environment",
+      "page": 7,
+      "char_offset": 32393,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[31]",
+      "technique": null,
+      "sentence": "Experimental Comparison a) Comparison with Disab led Operator Selection Algorithm: To evaluate the contribution of the “optimal ” operator selection algorithm proposed in this paper to the performance of SQLPass, we compare the code (line) coverage of SQLite3 [8], MySQL [30], MariaDB [9], and PostgreSQL [31] when the “optimal ” operator selection algorithm was disabled and used.",
+      "context_before": "overing bugs. A. Experimental Environment The experiment in this paper is completed on a computer equipped with a 64 bit 8-core Intel Core i7-4790 Ubuntu 20.04.6LTS CPU@3.60GHz 8Gib of memory and 1TB of disk capacity. The target DBMS selected in this paper are multiple versions of SQLite, MySQL, MariaDB, and PostgreSQL. The benchmark DBMS testing tools are AFL2.57b, SQLsmith, SQLancer, and Squirrel. It is worth noting that in the experimental process of this paper, the seed test cases used fo r testing the same DBMS using mutation-based testing tools AFL, Squirrel, and SQLPass are the same. B.",
+      "context_after": "As shown in Table Ⅲ, SQLPass improves the code coverage of SQLite3, MySQL, MariaDB, and PostgreSQL by 10.9%, 9.6%, 9.4%, and 8.6%, respectively, using the “optimal ” operator selection algorithm compared to disabling this algorithm. TABLE III. COMPARISON OFCODE COVERAGE BETWEEN SQLPASS DISABLED ANDUSED OPERATOR SELECTION ALGORITHM. Target Code coverage rate(24h) SQLPass- SQLPass improvement SQLite3 53.7% 64.6% 10.9%↑ MySQL 43.8% 53.4% 9.6% ↑ MariaDB 47.2% 56.6% 09.4%↑ PostgreSQL 43.7% 52.3% 08.6%↑ The experimental results show that the completely random operator selection method with disabled",
+      "section": "B Experimental Comparison",
+      "page": 7,
+      "char_offset": 32629,
+      "cited_reference": {
+        "number": 31,
+        "text": "PostgreSQL. https://www.postgresql.org/. CREATE TABLE t1(c1,c2); CREATE TABLE t2(c1,c2); WITH t2(c1) AS (SELECT 111) INSERT INTO t1 (c1, c2) SELECT c1, 123 FROM t2; CREATE TABLE t1(c1,c2); CREATE TABLE t2(c1,c2); WITH t2(c1) AS (SELECT 1 UNION ALL SELECT c1+1 FROM t2) INSERT INTO t1 (c1, c2) SELECT ",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "b) C omparison with benchmark testing tools: We conduct 24 hours experiments on SQLPass and benchmark testing tools AFL, SQLsmith, SQLancer, and Squirrel on SQLite, MySQL, MariaDB, and PostgreSQL, respectively, and compare and evaluate three ind icators: semantic correctness, code (line) coverage, and bug replication speed.",
+      "context_before": "tated SQL statement s generated by the selected mutation operators can trigger new program states. Repeatedly selecting invalid mutation operators can also reduce testing efficiency. The semi random method using the “optimal ” operator selection algorithm rewards mutation operators that can trigger new program states, ensurin g that each selected mutation operator is the “optimal ” operator in the current set of available operators as much as possible to achieve effective mutation of SQL test cases, thereby triggerin g new program states and achieving more efficient fuzzing of the target DBMS.",
+      "context_after": "We construct the same set of test cases as the original seed set for mutatio nbased testing tools AFL, Squirrel, and SQLPass. To evaluate the speed of bug replication, we randomly select 5 seeds that can trigger known bugs for each tested DBMS as part of the original seed set when constructing seed test cases. The experimental results are shown in Fig. 5. Owing to the lack of grammar models for MySQL and MariaDB in the SQLsmith tool, test cases for MySQL and MariaDB cannot be generated. Therefore, it was not compared with SQLsmith on MySQL and MariaDB. Because SQLsmith and SQLancer are genera",
+      "section": "B Experimental Comparison",
+      "page": 7,
+      "char_offset": 34275,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Because SQLsmith and SQLancer are generation-based, they cannot replicate existing bugs based on seed test cases.",
+      "context_before": "ine) coverage, and bug replication speed. We construct the same set of test cases as the original seed set for mutatio nbased testing tools AFL, Squirrel, and SQLPass. To evaluate the speed of bug replication, we randomly select 5 seeds that can trigger known bugs for each tested DBMS as part of the original seed set when constructing seed test cases. The experimental results are shown in Fig. 5. Owing to the lack of grammar models for MySQL and MariaDB in the SQLsmith tool, test cases for MySQL and MariaDB cannot be generated. Therefore, it was not compared with SQLsmith on MySQL and MariaDB.",
+      "context_after": "Therefore, there was no comparison with these two tools in terms of bug replication speed. In terms of semantic correctness, SQLPass has improved by 5.7%-94.2%. Specifically, SQLPass has increased by 92.5%, 9 4. 2 %, 5. 7 %, a n d 5 9. 8 % c o m p a r e d t o A F L, S Q L s m i t h, SQLancer, and Squirrel on SQLite3, respectively; Compared to AFL, SQLancer, and Squirrel, it has increased by 84.5%, 12.2%, and 35.6% respectively on MySQL; Compared to AFL, SQLancer, and Squirrel, it has increased by 83.3%, 8.4%, and 45.4% respectively on MariaDB; Compared to AFL, SQLsmith, SQLancer, and Squirrel",
+      "section": "B Experimental Comparison",
+      "page": 7,
+      "char_offset": 35160,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "8 % c o m p a r e d t o A F L, S Q L s m i t h, SQLancer, and Squirrel on SQLite3, respectively; Compared to AFL, SQLancer, and Squirrel, it has increased by 84.",
+      "context_before": "es. The experimental results are shown in Fig. 5. Owing to the lack of grammar models for MySQL and MariaDB in the SQLsmith tool, test cases for MySQL and MariaDB cannot be generated. Therefore, it was not compared with SQLsmith on MySQL and MariaDB. Because SQLsmith and SQLancer are generation-based, they cannot replicate existing bugs based on seed test cases. Therefore, there was no comparison with these two tools in terms of bug replication speed. In terms of semantic correctness, SQLPass has improved by 5.7%-94.2%. Specifically, SQLPass has increased by 92.5%, 9 4. 2 %, 5. 7 %, a n d 5 9.",
+      "context_after": "5%, 12.2%, and 35.6% respectively on MySQL; Compared to AFL, SQLancer, and Squirrel, it has increased by 83.3%, 8.4%, and 45.4% respectively on MariaDB; Compared to AFL, SQLsmith, SQLancer, and Squirrel, PostgreSQL has increased by 86.5%, 87.4%, 10.4%, and 48.5%, respectively. The above semantic accuracy data indicates that due to the fact that AFL mutation is based on random bits or bytes, the SQL statements generated by mutation have low syntactic and semantic accuracy; SQLsmith is based on the syntax tree model to generate SQL test cases, which to some extent ensures the correctness of synt",
+      "section": "B Experimental Comparison",
+      "page": 7,
+      "char_offset": 35510,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": [],
+      "text_is_letter_spaced": true
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "6% respectively on MySQL; Compared to AFL, SQLancer, and Squirrel, it has increased by 83.",
+      "context_before": "ed. Therefore, it was not compared with SQLsmith on MySQL and MariaDB. Because SQLsmith and SQLancer are generation-based, they cannot replicate existing bugs based on seed test cases. Therefore, there was no comparison with these two tools in terms of bug replication speed. In terms of semantic correctness, SQLPass has improved by 5.7%-94.2%. Specifically, SQLPass has increased by 92.5%, 9 4. 2 %, 5. 7 %, a n d 5 9. 8 % c o m p a r e d t o A F L, S Q L s m i t h, SQLancer, and Squirrel on SQLite3, respectively; Compared to AFL, SQLancer, and Squirrel, it has increased by 84.5%, 12.2%, and 35.",
+      "context_after": "3%, 8.4%, and 45.4% respectively on MariaDB; Compared to AFL, SQLsmith, SQLancer, and Squirrel, PostgreSQL has increased by 86.5%, 87.4%, 10.4%, and 48.5%, respectively. The above semantic accuracy data indicates that due to the fact that AFL mutation is based on random bits or bytes, the SQL statements generated by mutation have low syntactic and semantic accuracy; SQLsmith is based on the syntax tree model to generate SQL test cases, which to some extent ensures the correctness of syntax. However, due to the complex structure of the generated SQL statements, the final semantic correctness of",
+      "section": "B Experimental Comparison",
+      "page": 7,
+      "char_offset": 35690,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "4% respectively on MariaDB; Compared to AFL, SQLsmith, SQLancer, and Squirrel, PostgreSQL has increased by 86.",
+      "context_before": "neration-based, they cannot replicate existing bugs based on seed test cases. Therefore, there was no comparison with these two tools in terms of bug replication speed. In terms of semantic correctness, SQLPass has improved by 5.7%-94.2%. Specifically, SQLPass has increased by 92.5%, 9 4. 2 %, 5. 7 %, a n d 5 9. 8 % c o m p a r e d t o A F L, S Q L s m i t h, SQLancer, and Squirrel on SQLite3, respectively; Compared to AFL, SQLancer, and Squirrel, it has increased by 84.5%, 12.2%, and 35.6% respectively on MySQL; Compared to AFL, SQLancer, and Squirrel, it has increased by 83.3%, 8.4%, and 45.",
+      "context_after": "5%, 87.4%, 10.4%, and 48.5%, respectively. The above semantic accuracy data indicates that due to the fact that AFL mutation is based on random bits or bytes, the SQL statements generated by mutation have low syntactic and semantic accuracy; SQLsmith is based on the syntax tree model to generate SQL test cases, which to some extent ensures the correctness of syntax. However, due to the complex structure of the generated SQL statements, the final semantic correctness of the SQL statement s cannot be guaranteed during semantic data filling; SQLancer constructs SQL statements based on selected pi",
+      "section": "B Experimental Comparison",
+      "page": 7,
+      "char_offset": 35797,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "However, due to the complex structure of the generated SQL statements, the final semantic correctness of the SQL statement s cannot be guaranteed during semantic data filling; SQLancer constructs SQL statements based on selected pivot rows.",
+      "context_before": "12.2%, and 35.6% respectively on MySQL; Compared to AFL, SQLancer, and Squirrel, it has increased by 83.3%, 8.4%, and 45.4% respectively on MariaDB; Compared to AFL, SQLsmith, SQLancer, and Squirrel, PostgreSQL has increased by 86.5%, 87.4%, 10.4%, and 48.5%, respectively. The above semantic accuracy data indicates that due to the fact that AFL mutation is based on random bits or bytes, the SQL statements generated by mutation have low syntactic and semantic accuracy; SQLsmith is based on the syntax tree model to generate SQL test cases, which to some extent ensures the correctness of syntax.",
+      "context_after": "Because of its single type and simple structure, the semantic accuracy of the generated SQL statements is relatively high; Squirrel mutates the intermediate representation (IR) based on SQL statements and fills in data for the mutated IR skeleton, resulting in low syntactic and semantic correctness of the generated SQL statements; And SQLPass is based on the mutation of syntax trees, which basically ensures the correctne ss of syntax. At the same time, its mutation targeting weak semantic correlation nodes in the syntax tree greatly reduces t he probability of generating SQL statements with s",
+      "section": "B Experimental Comparison",
+      "page": 7,
+      "char_offset": 36275,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "012345678910 0 3 6 9 12 15 18 2124vulnerability number SQLite’s test time(h)0102030405060708090100 0 3 6 9 12 15 18 21 24code coverage(%) SQLite’s test time(h) 0102030405060708090100 0 3 6 9 12 15 18 21 24Semantic correctness (%) MySQL ’s test time(h)0102030405060708090100 0 3 6 9 12 15 18 21 24Semantic correctness(%) SQLite’s test time(h) 0102030405060708090100 0 3 6 9 12 15 18 21 24code coverage(%) MySQL ’s test time(h)012345678910 0 3 6 9 12 15 18 2124vulnerability number MySQL ’s test time(h) AFL SQLsmith SQLancer Squirrel SQLPass012345678910 0 3 6 9 1 21 51 82 12 4vulnerability number PostgreSQL ’s test time(h)012345678910 0 3 6 9 12 15 18 2124vulnerability number MariaDB ’s test time(h)0102030405060708090100 0 3 6 9 12 15 18 21 24code coverage(%) MariaDB ’s test time(h)0102030405060708090100 0 3 6 9 12 15 18 2124Semantic correctness(%) MariaDB’s test time(h) 0102030405060708090100 0 3 6 9 12 15 18 2124code coverage(%) PostgreSQL ’s test time(h)0102030405060708090100 0 3 6 9 12 15 18 21 24Semantic correctness(% ) PostgreSQL’s test time(h) 1042 In terms of code coverage, SQLPass has increased by 1.",
+      "context_before": "sically ensures the correctne ss of syntax. At the same time, its mutation targeting weak semantic correlation nodes in the syntax tree greatly reduces t he probability of generating SQL statements with semantic errors. In addition, in case of semantic errors during the mutation process, SQLPass also conducts semantic checks and corrections, which greatly improves the semantic correctness of the final generated SQL statements. 1041 Fig. 5. Comparison between SQLPass and related benchmark testin g tools in terms of semantic correctness, code coverage, and bug replication speed.",
+      "context_after": "3%52%. SQLPass has increased by 41%, 52%, 12.7%, and 1.3% compared to the four benchmarks on SQLite3, respectively; Compared to AFL, SQLancer, and Squirrel, it has increased by 35.4%, 6.3%, and 10.3% respectively on MySQL; On MariaDB, it has increased by 35.8%, 5.1%, and 9.9% compared to AFL, SQLancer, and Squirrel, respectively; Compared to the four benchmarks, PostgreSQL has increased by 36.7%, 28.5%, 12.8%, and 10.2%, respectively. The above coverage data indicates that due to the low syntax and semantic accuracy of SQL test cases generated by AFL mutation, they cannot pass through the pars",
+      "section": "B Experimental Comparison",
+      "page": 8,
+      "char_offset": 37512,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3% compared to the four benchmarks on SQLite3, respectively; Compared to AFL, SQLancer, and Squirrel, it has increased by 35.",
+      "context_before": "vulnerability number PostgreSQL ’s test time(h)012345678910 0 3 6 9 12 15 18 2124vulnerability number MariaDB ’s test time(h)0102030405060708090100 0 3 6 9 12 15 18 21 24code coverage(%) MariaDB ’s test time(h)0102030405060708090100 0 3 6 9 12 15 18 2124Semantic correctness(%) MariaDB’s test time(h) 0102030405060708090100 0 3 6 9 12 15 18 2124code coverage(%) PostgreSQL ’s test time(h)0102030405060708090100 0 3 6 9 12 15 18 21 24Semantic correctness(% ) PostgreSQL’s test time(h) 1042 In terms of code coverage, SQLPass has increased by 1.3%52%. SQLPass has increased by 41%, 52%, 12.7%, and 1.",
+      "context_after": "4%, 6.3%, and 10.3% respectively on MySQL; On MariaDB, it has increased by 35.8%, 5.1%, and 9.9% compared to AFL, SQLancer, and Squirrel, respectively; Compared to the four benchmarks, PostgreSQL has increased by 36.7%, 28.5%, 12.8%, and 10.2%, respectively. The above coverage data indicates that due to the low syntax and semantic accuracy of SQL test cases generated by AFL mutation, they cannot pass through the parsing and validation stages of the tested DBMS parser. As a result, the SQL test cases generated by mutation cannot trigger deeper code logic, resulting in higher code coverage; The",
+      "section": "B Experimental Comparison",
+      "page": 9,
+      "char_offset": 38689,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "9% compared to AFL, SQLancer, and Squirrel, respectively; Compared to the four benchmarks, PostgreSQL has increased by 36.",
+      "context_before": "5060708090100 0 3 6 9 12 15 18 2124Semantic correctness(%) MariaDB’s test time(h) 0102030405060708090100 0 3 6 9 12 15 18 2124code coverage(%) PostgreSQL ’s test time(h)0102030405060708090100 0 3 6 9 12 15 18 21 24Semantic correctness(% ) PostgreSQL’s test time(h) 1042 In terms of code coverage, SQLPass has increased by 1.3%52%. SQLPass has increased by 41%, 52%, 12.7%, and 1.3% compared to the four benchmarks on SQLite3, respectively; Compared to AFL, SQLancer, and Squirrel, it has increased by 35.4%, 6.3%, and 10.3% respectively on MySQL; On MariaDB, it has increased by 35.8%, 5.1%, and 9.",
+      "context_after": "7%, 28.5%, 12.8%, and 10.2%, respectively. The above coverage data indicates that due to the low syntax and semantic accuracy of SQL test cases generated by AFL mutation, they cannot pass through the parsing and validation stages of the tested DBMS parser. As a result, the SQL test cases generated by mutation cannot trigger deeper code logic, resulting in higher code coverage; The SQL statements generated by SQLsmith are mostly SELECT statements, with a single type and low syntax and semantic accuracy. Therefore, the code coverage generated by the SQL test cases it generates is also not high;",
+      "section": "B Experimental Comparison",
+      "page": 9,
+      "char_offset": 38908,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Therefore, the code coverage generated by the SQL test cases it generates is also not high; Although SQLancer generates limited types of SQL test cases, the syntax and semantic accuracy of the generated SQL statements are high, so it can trigger deeper code logic and generate higher code coverage; The syntax and semantic accuracy of SQL test cases generated by Squirrel mutation are not high, but the number of SQL statements that can mutate within the same time is relatively high, and the types of SQL statements that can mutat e are relatively diverse, resulting in a relatively high coverage rate; In contrast, SQLPass does not mutate as many SQL statements as Squirrel in the same amount of time, but its generated SQL statements have higher syntax and semantic correctness, and there are rich types of mutable SQL statements, resulting in higher code coverage.",
+      "context_before": "and Squirrel, respectively; Compared to the four benchmarks, PostgreSQL has increased by 36.7%, 28.5%, 12.8%, and 10.2%, respectively. The above coverage data indicates that due to the low syntax and semantic accuracy of SQL test cases generated by AFL mutation, they cannot pass through the parsing and validation stages of the tested DBMS parser. As a result, the SQL test cases generated by mutation cannot trigger deeper code logic, resulting in higher code coverage; The SQL statements generated by SQLsmith are mostly SELECT statements, with a single type and low syntax and semantic accuracy.",
+      "context_after": "In terms of bug replication speed, SQLPass has increased the speed of bug replication by 0.68-3.32 times compared to other benchmarks. Specifically, it only takes 7.2 hours to replicate 3 bugs on SQLite3, while Squirrel takes 15.4 hours to replicate t he same number of bugs. Other methods have failed to replicate bugs within 24 hours. In contrast, SQLPass replicates bugs at a speed 1.14 times faster than Squirrel; It took 16.3 hours to replicate 2 bugs on MySQL, while other methods failed to replicate the bugs within 24 hours; SQLPass replicates 1 bug on MariaDB in only 3.8 hours, Squirrel r",
+      "section": "B Experimental Comparison",
+      "page": 9,
+      "char_offset": 39537,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M3"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:59:32Z",
+    "is_model_written": true,
+    "summary": "SQLPass fuzzes DBMSs with an operator selection algorithm aimed at keeping generated SQL semantically valid, on the grounds that semantically invalid statements are rejected early and never reach the deeper code. Over 24 hours on SQLite, MySQL, MariaDB and PostgreSQL it reports higher semantic correctness, higher line coverage and faster bug replication than its baselines.",
+    "narrative": "SQLancer is one of four baselines, compared on semantic correctness, line coverage and bug replication speed. The paper's account of it is two-sided: because SQLancer is generation-based it cannot replicate existing bugs from seed test cases, which is one of the three measures; but it also observes that although SQLancer generates a limited range of SQL test cases, their syntactic and semantic accuracy is high, so it reaches deeper code logic and achieves higher coverage than the tools whose generated statements are less well formed. PQS's pivot-row construction is named as the reason for that accuracy.",
+    "roles": {
+      "M1": "baseline",
+      "M2": "incidental",
+      "M3": "baseline",
+      "M4": "result_comparison",
+      "M5": "result_comparison",
+      "M6": "result_comparison",
+      "M7": "result_comparison",
+      "M8": "definition",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "result_comparison",
+      "M12": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is run as a baseline; SQLPass's operator selection algorithm is its own."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer oracle is extended. SQLPass addresses how statements are generated, not what makes a result wrong."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M3",
+          "M5",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "b) C omparison with benchmark testing tools: We conduct 24 hours experiments on SQLPass and benchmark testing tools AFL, SQLsmith, SQLancer, and Squirrel on SQLite, MySQL, MariaDB, and PostgreSQL, respectively, and compare and evaluate three ind icators: semantic correctness, code (line) coverage, and bug replication speed.",
+            "section": "B Experimental Comparison",
+            "page": 7
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "8 % c o m p a r e d t o A F L, S Q L s m i t h, SQLancer, and Squirrel on SQLite3, respectively; Compared to AFL, SQLancer, and Squirrel, it has increased by 84.",
+            "section": "B Experimental Comparison",
+            "page": 7,
+            "text_is_letter_spaced": true
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "Therefore, the code coverage generated by the SQL test cases it generates is also not high; Although SQLancer generates limited types of SQL test cases, the syntax and semantic accuracy of the generated SQL statements are high, so it can trigger deeper code logic and generate higher code coverage; The syntax and semantic accuracy of SQL test cases generated by Squirrel mutation are not high, but the number of SQL statements that can mutate within the same time is relatively high, and the types of SQL statements that can mutat e are relatively diverse, resulting in a relatively high coverage rate; In contrast, SQLPass does not mutate as many SQL statements as Squirrel in the same amount of time, but its generated SQL statements have higher syntax and semantic correctness, and there are rich types of mutable SQL statements, resulting in higher code coverage.",
+            "section": "B Experimental Comparison",
+            "page": 9
+          }
+        ],
+        "reasoning": "M3 states the 24-hour experiments compare SQLPass with AFL, SQLsmith, SQLancer and Squirrel on four systems, M5 reports the coverage increases against each, and M12 explains SQLancer's relative standing on accuracy and depth."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is called a benchmark testing tool and described by its generation characteristics; no state-of-the-art claim is made."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_dsc55868_2022_00057.json b/_data/papers/paper_doi_10_1109_dsc55868_2022_00057.json
new file mode 100644
index 0000000..d611018
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_dsc55868_2022_00057.json
@@ -0,0 +1,438 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:36Z",
+  "paper": {
+    "id": "paper:doi:10.1109/dsc55868.2022.00057",
+    "title": "Fuzzing DBMS via NNLM",
+    "authors": [
+      "Yabin Li",
+      "Yuanping Nie",
+      "Xiaohui Kuang"
+    ],
+    "year": 2022,
+    "venue": "International Conference on Data Science in Cyberspace",
+    "doi": "10.1109/dsc55868.2022.00057",
+    "arxiv_id": null,
+    "s2_paper_id": "6c8ad2d2e977e0a45d987959fc3e4adcde10e4ca",
+    "url": "https://doi.org/10.1109/dsc55868.2022.00057",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/dsc55868.2022.00057",
+      "retrieved_at": "2026-09-09T01:33:36Z",
+      "chars": 32242,
+      "content_sha256": "sha256:26dcc0dfd70740f372f73eaeb6f8803725cebc5eebb00faf1e007a96a394908c"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 8,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1037
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 4281
+      },
+      {
+        "number": "A",
+        "title": "DBMS Fuzzing",
+        "start": 4296
+      },
+      {
+        "number": "III",
+        "title": "OVERVIEW",
+        "start": 13752
+      },
+      {
+        "number": "IV",
+        "title": "DESIGN",
+        "start": 15142
+      },
+      {
+        "number": "A",
+        "title": "Training LSTM Model",
+        "start": 15778
+      },
+      {
+        "number": "B",
+        "title": "Generator",
+        "start": 18833
+      },
+      {
+        "number": "C",
+        "title": "Fuzzer",
+        "start": 20580
+      },
+      {
+        "number": "V",
+        "title": "EVALUA TION",
+        "start": 21273
+      },
+      {
+        "number": "A",
+        "title": "Experimental Setup",
+        "start": 21858
+      },
+      {
+        "number": "B",
+        "title": "Parsing pass rate",
+        "start": 23134
+      },
+      {
+        "number": "C",
+        "title": "Code coverage",
+        "start": 24413
+      },
+      {
+        "number": "VI",
+        "title": "CONCLUSION ANDFUTURE WORK",
+        "start": 25782
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "MozillaSecurity/funfuzz, jsfunfuzz, GitHub. [Online]. Available: https://github.com/MozillaSecurity/funfuzz. Accessed on: Mar. 16, 2022",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Yang X, Chen Y, Eide E, et al. Finding and understanding bugs in C compilers[C]//Proceedings of the 32nd ACMSIGPLAN conference on Programming language design and implementation. 2011: 283-294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Wang J, Chen B, Wei L, et al. Skyfire: Data-driven seed generation for fuzzing[C]//. IEEE, 2017: 579-594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "American Fuzzy Lop, AFL. [Online]. Available: https://github.com/google/AFL. Accessed on: Mar. 12, 2022",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Aschermann C, Frassetto T, Holz T, et al. NAUTILUS: Fishing for Deep Bugs with Grammars[C]//NDSS. 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Wang J, Chen B, Wei L, et al. Superion: Grammar-aware greybox fuzzing[C]// Engineering (ICSE). IEEE, 2019: 724-735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Li Z, Zou D, Xu S, et al. Vuldeepecker: A deep learning-based system for vulnerability detection[J]. arXiv preprint arXiv:1801.01681, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Zou D, Wang S, Xu S, et al. μVulDeePecker: A Deep Learning-Based System for Multiclass Vulnerability Detection[J]. IEEE Transactions on Dependable and Secure Computing, 2019, 18(5): 2224-2236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Li Z, Zou D, Xu S, et al. Sysevr: A framework for using deep learning to detect software vulnerabilities[J]. IEEE Transactions on Dependable and Secure Computing, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Li Z, Zou D, Xu S, et al. Vuldeelocator: a deep learning-based finegrained vulnerability detector[J]. IEEE Transactions on Dependable and Secure Computing, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Zong P, Lv T, Wang D, et al. FuzzGuard: Filtering out Unreachable Inputs in Directed Grey-box Fuzzing through Deep Learning[C]//29th USENIX Security Symposium (USENIX Security 20). 2020: 2255-2269.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "B ¨ohme M, Pham V T, Nguyen M D, et al. Directed greybox fuzzing[C]//Proceedings of the 2017 ACMSIGSAC Conference on Computer and Communications Security. 2017: 2329-2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Liu X, Li X, Prajapati R, et al. Deepfuzz: Automatic generation of syntax valid c programs for fuzz testing[C]//Proceedings of the AAAI Conference on Artificial Intelligence. 2019, 33(01): 1044-1051. 373",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Cummins C, Petoumenos P, Murray A, et al. Compiler fuzzing through deep learning[C]//Proceedings of the 27th ACMSIGSOFT International Symposium on Software Testing and Analysis. 2018: 95-105.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Xu H, Wang Y, Fan S, et al. DSmith: Compiler Fuzzing through Generative Deep Learning Model with Attention[C]//2020 International Joint Conference on Neural Networks (IJCNN). IEEE, 2020: 1-9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Lee S, Han H S, Cha S K, et al. Montage: A Neural Network Language Model-GuidedJavaScript Engine Fuzzer[C]//29th USENIX Security Symposium (USENIX Security 20). 2020: 2613-2630.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Li Y, Qin Y, Tian J, et al. Generating Highly Structured Inputs: A Survey[C]// in Cyberspace (DSC). IEEE, 2021: 466-473.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Slutz D R. Massive stochastic testing of SQL[C]//VLDB. 1998, 98: 618622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Rigger M, Su Z. Testing database engines via pivoted query synthesis[C]//14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 2020: 667-682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Rigger M, Su Z. Detecting optimization bugs in database engines via non-optimizing reference engine construction[C]//Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 2020: 1140-1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Rigger M, Su Z. Finding bugs in database systems via query partitioning[J]. Proceedings of the ACM on Programming Languages, 2020, 4(OOPSLA): 1-30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "Gu Z, Soliman M A, Waas F M. Testing the accuracy of query optimizers[C]//Proceedings of the Fifth International Workshop on Testing Database Systems. 2012: 1-6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Jung J, Hu H, Arulraj J, et al. APOLLO: Automatic detection and diagnosis of performance regressions in database systems[J]. Proceedings of the VLDB Endowment, 2019, 13(1): 57-70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "SquirrelZhong R, Chen Y, Hu H, et al. Squirrel: Testing database management systems with language validity and coverage feedback[C]//Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 2020: 955-970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Wang M, Wu Z, Xu X, et al. Industry practice of coverage-guided enterprise-level DBMS fuzzing[C]//2021 IEEE/ACM 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP). IEEE, 2021: 328-337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Chen Y, Zhong R, Hu H, et al. One engine to fuzz’em all: Generic language processor testing with semantic validation[C]//2021 IEEE Symposium on Security and Privacy (SP). IEEE, 2021: 642-658.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Bengio Y, Ducharme R, Vincent P. A neural probabilistic language model[J]. Advances in Neural Information Processing Systems, 2000, 13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Bengio Y, Simard P, Frasconi P. Learning long-term dependencies with gradient descent is difficult[J]. IEEE transactions on neural networks, 1994, 5(2): 157-166.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "LSTM Hochreiter S, Schmidhuber J. Long short-term memory[J]. Neural computation, 1997, 9(8): 1735-1780.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Gers F A, Schmidhuber J, Cummins F. Learning to forget: Continual prediction with LSTM[J]. Neural computation, 2000, 12(10): 2451-2471.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Godefroid P, Peleg H, Singh R. Learn&fuzz: Machine learning for input fuzzing[C]//2017 32nd IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 2017: 50-59.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Ye G, Tang Z, Tan S H, et al. Automated Conformance Testing for JavaScript Engines via Deep Compiler Fuzzing[C]//The 42nd ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). Association for Computing Machinery (ACM), 2021. 374",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 19,
+      "text": "Rigger M, Su Z. Testing database engines via pivoted query synthesis[C]//14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 2020: 667-682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "Rigger M, Su Z. Detecting optimization bugs in database engines via non-optimizing reference engine construction[C]//Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 2020: 1140-1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 21,
+      "text": "Rigger M, Su Z. Finding bugs in database systems via query partitioning[J]. Proceedings of the ACM on Programming Languages, 2020, 4(OOPSLA): 1-30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[18, 19,\n20, 21]",
+      "technique": "pqs",
+      "sentence": "When a correctness error is triggered, the DBMS does not crash, but the query returns incorrect query results [18, 19, 20, 21].",
+      "context_before": "is paper are summarized as follows: 1) We explore the application of NNLM in DBMS Fuzzing, training a sequence model to generate syntactically valid SQL statements. 2) We build a utility NNFuzz based on the trained model, which can be used to Fuzz DBMS. 3) We evaluated the effect of NNFuzz on SQLite and verified its effectiveness. II. BACKGROUND A. DBMS Fuzzing The work of DBMS fuzzing is mainly divided into two categories, one focuses on the functional errors of the DBMS, and the other focuses on the crash errors of the DBMS. Functional errors include correctness errors and performance errors.",
+      "context_after": "Performance errors include issues such as the accuracy of the query optimizer [22] and the abnormal execution time of SQL queries [23]. Crash errors mainly refer to errors such as memory errors that can cause the system to crash. 3672022 7th IEEE International Conference on Data Science in Cyberspace (DSC) 978-1-6654-7480-1/22/$31.00 ©2022 IEEEDOI 10.1109/DSC55868.2022.000572022 7th IEEE International Conference on Data Science in Cyberspace (DSC) | 978-1-6654-7480-1/22/$31.00 ©2022 IEEE | DOI: 10.1109/DSC55868.2022.00057 RAGS [18] executes the same query on multiple DBMSs from different ve",
+      "section": "A DBMS Fuzzing",
+      "page": 1,
+      "char_offset": 4549,
+      "cited_reference": {
+        "number": 19,
+        "text": "Rigger M, Su Z. Testing database engines via pivoted query synthesis[C]//14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 2020: 667-682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS[19] selects the pivot row in advance, and then constructs the SQL query based on the pivot row.",
+      "context_before": "and the abnormal execution time of SQL queries [23]. Crash errors mainly refer to errors such as memory errors that can cause the system to crash. 3672022 7th IEEE International Conference on Data Science in Cyberspace (DSC) 978-1-6654-7480-1/22/$31.00 ©2022 IEEEDOI 10.1109/DSC55868.2022.000572022 7th IEEE International Conference on Data Science in Cyberspace (DSC) | 978-1-6654-7480-1/22/$31.00 ©2022 IEEE | DOI: 10.1109/DSC55868.2022.00057 RAGS [18] executes the same query on multiple DBMSs from different vendors and finds correctness errors by comparing the differences in their result sets.",
+      "context_after": "If the pivot row does not appear in the execution result of the SQL query, it means that the query may have a correctness error. NoREC [20] transforms the SQL query into an equivalent form that will not be optimized by the query optimizer, and then executes the original query and the equivalent form separately. If the two results are inconsistent, it means that an optimization error of the query optimizer may be found. Query Partitioning [21] uses ternary logical partitioning to convert the original query into three partitioned queries, and compares whether the results of the original query a",
+      "section": "A DBMS Fuzzing",
+      "page": 2,
+      "char_offset": 5361,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [20] transforms the SQL query into an equivalent form that will not be optimized by the query optimizer, and then executes the original query and the equivalent form separately.",
+      "context_before": "1-6654-7480-1/22/$31.00 ©2022 IEEEDOI 10.1109/DSC55868.2022.000572022 7th IEEE International Conference on Data Science in Cyberspace (DSC) | 978-1-6654-7480-1/22/$31.00 ©2022 IEEE | DOI: 10.1109/DSC55868.2022.00057 RAGS [18] executes the same query on multiple DBMSs from different vendors and finds correctness errors by comparing the differences in their result sets. PQS[19] selects the pivot row in advance, and then constructs the SQL query based on the pivot row. If the pivot row does not appear in the execution result of the SQL query, it means that the query may have a correctness error.",
+      "context_after": "If the two results are inconsistent, it means that an optimization error of the query optimizer may be found. Query Partitioning [21] uses ternary logical partitioning to convert the original query into three partitioned queries, and compares whether the results of the original query are consistent with the union of the partitioned queries. If the results are inconsistent, an error may be found. TAQO [22] measures the accuracy of the query optimizer. For each execution plan, TAQO estimates its execution cost, and obtains the actual cost by actually executing the query, and then calculates the",
+      "section": "A DBMS Fuzzing",
+      "page": 2,
+      "char_offset": 5590,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "Query Partitioning",
+      "technique": "tlp",
+      "sentence": "Query Partitioning [21] uses ternary logical partitioning to convert the original query into three partitioned queries, and compares whether the results of the original query are consistent with the union of the partitioned queries.",
+      "context_before": "nd finds correctness errors by comparing the differences in their result sets. PQS[19] selects the pivot row in advance, and then constructs the SQL query based on the pivot row. If the pivot row does not appear in the execution result of the SQL query, it means that the query may have a correctness error. NoREC [20] transforms the SQL query into an equivalent form that will not be optimized by the query optimizer, and then executes the original query and the equivalent form separately. If the two results are inconsistent, it means that an optimization error of the query optimizer may be found.",
+      "context_after": "If the results are inconsistent, an error may be found. TAQO [22] measures the accuracy of the query optimizer. For each execution plan, TAQO estimates its execution cost, and obtains the actual cost by actually executing the query, and then calculates the accuracy of the query optimizer based on the difference between the two. Cost estimation models are difficult to be completely accurate, but an intuitive idea is that plans with higher estimated costs do run longer. APOLLO [23] executes the same queries on the new and old versions to find performance regression bugs in the DBMS (some SQL quer",
+      "section": "A DBMS Fuzzing",
+      "page": 2,
+      "char_offset": 5884,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:20:36Z",
+    "is_model_written": true,
+    "summary": "NNFuzz applies a neural network language model to DBMS fuzzing, using a sequence model to generate test cases automatically. The authors note few studies had taken this route, and highlight its suitability for black-box testing. Evaluated on SQLite, the tool generated valid test cases and achieved higher code coverage than its initial training set.",
+    "narrative": "Three citations describing SQLancer's oracles -- PQS, NoREC and query partitioning -- as the established ways of detecting correctness errors that do not crash the system.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_girst67753_2025_11382165.json b/_data/papers/paper_doi_10_1109_girst67753_2025_11382165.json
new file mode 100644
index 0000000..34721b7
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_girst67753_2025_11382165.json
@@ -0,0 +1,356 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:25:13Z",
+  "paper": {
+    "id": "paper:doi:10.1109/girst67753.2025.11382165",
+    "title": "Testing Spatial Database Engines via Spatial Equivalent Transformation",
+    "authors": [
+      "Shijie Li",
+      "Dongping Wang",
+      "Liang Liu",
+      "Keyue Yang",
+      "Lingwei Kuang"
+    ],
+    "year": 2025,
+    "venue": "2025 4th International Conference on Geographic Information and Remote Sensing Technology (GIRST)",
+    "doi": "10.1109/girst67753.2025.11382165",
+    "arxiv_id": null,
+    "s2_paper_id": "df4018eafe690fef9baef1ce90aa5a3eb3b0bfee",
+    "url": "https://doi.org/10.1109/girst67753.2025.11382165",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/girst67753.2025.11382165",
+      "retrieved_at": "2026-09-08T15:25:13Z",
+      "chars": 21529,
+      "content_sha256": "sha256:79b07c0b684885fb271aaed1f8a54a31d75eb6a2344dcf2641af8c3ca28579f0"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 5,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1955
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUD",
+        "start": 5455
+      },
+      {
+        "number": "A",
+        "title": "Geometry Types in SDBMS",
+        "start": 5617
+      },
+      {
+        "number": "B",
+        "title": "Topological Structure of Geometry",
+        "start": 6205
+      },
+      {
+        "number": "C",
+        "title": "Topological Relationships",
+        "start": 7125
+      },
+      {
+        "number": "III",
+        "title": "APPROACH",
+        "start": 8218
+      },
+      {
+        "number": "A",
+        "title": "Spader’s Architecture",
+        "start": 8462
+      },
+      {
+        "number": "B",
+        "title": "Spatial Database Generator",
+        "start": 8963
+      },
+      {
+        "number": "C",
+        "title": "SET",
+        "start": 11409
+      },
+      {
+        "number": "D",
+        "title": "Query Results Validator",
+        "start": 14235
+      },
+      {
+        "number": "IV",
+        "title": "EXPERIMENTS",
+        "start": 14675
+      },
+      {
+        "number": "A",
+        "title": "New Bugs",
+        "start": 14787
+      },
+      {
+        "number": "B",
+        "title": "Efficiency of Spader’s Components",
+        "start": 15271
+      },
+      {
+        "number": "C",
+        "title": "Comparison with Other Techniques",
+        "start": 16920
+      },
+      {
+        "number": "V",
+        "title": "CONCLUSION",
+        "start": 18209
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Open Geospatial Consortium. OpenGIS Web Map Service (WMS) Implementation Specification. OGC 01-068r3, 2006. Version 1.3.0.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Zhong R, Chen YH, Hu H, Zhang HF, Lee W, Wu DH. Squirrel: Testing database management systems with language validity and coverage feedback. In: Proc. of the 2020 ACMSIGSAC Conf. on Computer and Communications Security. ACM, 2020. 955–970. [doi: 10.1145/3372297.3417260]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Jiang ZM, Bai JJ, Su ZD. DynSQL: Stateful fuzzing for database management systems with complex and valid SQL query generation. In: Proc. of the 32nd USENIX Conf. on Security Symp. Anaheim: USENIX Association, 2023. 277",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Rigger, M., and Su, Z. Detecting optimization bugs in database engines via NoREC. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020), Association for Computing Machinery, pp. 1140–1152, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "Rigger, M., and Su, Z. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA(nov 2020).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Rigger, M., and Su, Z. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (USA, 2020), OSDI’20, USENIX Association.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 7,
+      "text": "Z. Hao et al., “Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis,” in Proc. 2023 USENIX Annu. Tech. Conf. (USENIX ATC ’23), Boston, MA, USA, 2023, pp. 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of the 45th International Conference on Software Engineering (ICSE ’23). IEEE Press, 2060–2071. https://doi.org/10.1109/ICSE48619.2023.00174",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 9,
+      "text": "PostGIS. Postgis integration with oss-fuzz. https://lists.osgeo.org/pipermail/postgis-devel/2017-July/026216.html. Retrieved March 1, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "W. Deng, Q. Mang, C. Zhang, and M. Rigger, “Finding Logic Bugs in Spatial Database Engines via Affine Equivalent Inputs,” 2, no. 6, Art. no. 235, pp. 1–26, Dec. 2024, doi: 10.1145/3698810.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 11,
+      "text": "Wiki. DE-9IM. https://en.wikipedia.org/wiki/DE-9IM. Retrieved January 14, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "M. J. Egenhofer, A. U. Frank, and J. P. Jackson, \"A topological data model for spatial databases,\" in Lecture Notes in Computer Science, vol. 409, O. Günther and H. J. Schek, Eds. New York, NY, USA: Springer, 1989, pp. 271–286.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "PostGIS. 2025. PostGIS Homepage. https://postgis.net/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "MySQL. 2025. MySQL Homepage. https://www.mysql.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "MariaDB. 2025. MariaDB Homepage. https://mariadb.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Clementini, E., Sharma, J., and Egenhofer, M. J. Modelling topological spatial relations: Strategies for query processing. Computers & Graphics 18, 6 (1994), 815–822. 182",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 4,
+      "text": "Rigger, M., and Su, Z. Detecting optimization bugs in database engines via NoREC. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020), Association for Computing Machinery, pp. 1140–1152, 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec"
+      ]
+    },
+    {
+      "number": 5,
+      "text": "Rigger, M., and Su, Z. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA(nov 2020).",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 6,
+      "text": "Rigger, M., and Su, Z. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (USA, 2020), OSDI’20, USENIX Association.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 8,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of the 45th International Conference on Software Engineering (ICSE ’23). IEEE Press, 2060–2071. https://doi.org/10.1109/ICSE48619.2023.00174",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "prints the DOI of the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 10,
+      "text": "W. Deng, Q. Mang, C. Zhang, and M. Rigger, “Finding Logic Bugs in Spatial Database Engines via Affine Equivalent Inputs,” 2, no. 6, Art. no. 235, pp. 1–26, Dec. 2024, doi: 10.1145/3698810.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[4]",
+      "technique": "norec",
+      "sentence": "Although crash bug detection [2] [3] and logic bug detection [4] [5] [6] [7] [8] are both mature for RDBMSs, SDBMSs suffer from a relative paucity of studies.",
+      "context_before": "r handling data with geometric and topological attributes. Their core innovation involves introducing spatial data types and integrating computation geometry-based operational capabilities. This provides solid theoretical and technical support for GIS, location intelligence, and remote sensing data management. Therefore, ensuring the correctness of SDBMSs is crucial. Mainstream SDBMSs extend the RDBMSs architecture by introducing OGC-compliant spatial data types [1] and a rich set of spatial functions into standard SQL. This allows for efficient geospatial querying unattainable by RDBMS alone.",
+      "context_after": "Generalpurpose fuzzers, such as OSS-Fuzz, have been applied to SDBMSs to generate crash-inducing inputs [9]. These tools are inherently unable to detect logic bugs. The sole tool for testing logic bugs of SDBMSs, Spatter [10], leverages the affine invariance of topological relations. However, becauseaffine transformations do not preserve metric properties (e.g size), Spatter is ineffective in detecting logic bugs of geometric construction and processing functions. Listing 1 shows a statement that triggers a logic bug in MariaDB. We expect ‘geom‘ and ‘ST_Intersection(geom, geom)‘ to be spatial",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2594,
+      "cited_reference": {
+        "number": 4,
+        "text": "Rigger, M., and Su, Z. Detecting optimization bugs in database engines via NoREC. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020), Association for Computing Machinery, pp. 1140–1152, 202",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "pqs",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[10]",
+      "technique": null,
+      "sentence": "The sole tool for testing logic bugs of SDBMSs, Spatter [10], leverages the affine invariance of topological relations.",
+      "context_before": "nsuring the correctness of SDBMSs is crucial. Mainstream SDBMSs extend the RDBMSs architecture by introducing OGC-compliant spatial data types [1] and a rich set of spatial functions into standard SQL. This allows for efficient geospatial querying unattainable by RDBMS alone. Although crash bug detection [2] [3] and logic bug detection [4] [5] [6] [7] [8] are both mature for RDBMSs, SDBMSs suffer from a relative paucity of studies. Generalpurpose fuzzers, such as OSS-Fuzz, have been applied to SDBMSs to generate crash-inducing inputs [9]. These tools are inherently unable to detect logic bugs.",
+      "context_after": "However, becauseaffine transformations do not preserve metric properties (e.g size), Spatter is ineffective in detecting logic bugs of geometric construction and processing functions. Listing 1 shows a statement that triggers a logic bug in MariaDB. We expect ‘geom‘ and ‘ST_Intersection(geom, geom)‘ to be spatially equal. This means the expected result is true. However, MariaDB incorrectly returns false, which indicates a logic bug. 1CREATE TABLE t1 (g geometry); 2INSERT INTO t1 (g) VALUES (ST_GeomFromText(’LINESTRING(2 2,4 4,6 2,3 2,2 4)’)); 3SELECT ST_Equals(g, ST_Intersection(g, g)) FROM t",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2918,
+      "cited_reference": {
+        "number": 10,
+        "text": "W. Deng, Q. Mang, C. Zhang, and M. Rigger, “Finding Logic Bugs in Spatial Database Engines via Affine Equivalent Inputs,” 2, no. 6, Art. no. 235, pp. 1–26, Dec. 2024, doi: 10.1145/3698810.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:52Z",
+    "is_model_written": true,
+    "summary": "This work tests spatial database engines by transforming a query into a spatially equivalent one -- applying a transformation that preserves topological relations -- and checking the results agree. Spatial systems have had far less testing attention than relational ones, and the paper notes that only one tool existed for finding logic bugs in them.",
+    "narrative": "SQLancer's oracles are cited collectively as the mature body of logic-bug detection for relational systems, against which the paucity of spatial work is measured. The one existing spatial tool, Spatter, is named separately. Both mentions are reachable only through citation markers, and the paper neither runs nor builds on any SQLancer technique.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_iaecst64597_2024_11117732.json b/_data/papers/paper_doi_10_1109_iaecst64597_2024_11117732.json
new file mode 100644
index 0000000..f611f32
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_iaecst64597_2024_11117732.json
@@ -0,0 +1,331 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:54:39Z",
+  "paper": {
+    "id": "paper:doi:10.1109/iaecst64597.2024.11117732",
+    "title": "Detecting Logical Bugs in DBMS via Isomerism Fuzz System",
+    "authors": [
+      "Zhe Wang",
+      "Liang Liu",
+      "Ning Wang"
+    ],
+    "year": 2024,
+    "venue": "2024 6th International Academic Exchange Conference on Science and Technology Innovation (IAECST)",
+    "doi": "10.1109/iaecst64597.2024.11117732",
+    "arxiv_id": null,
+    "s2_paper_id": "f341d91f2721ed842a9d7913d2ffe69b47050f51",
+    "url": "https://doi.org/10.1109/iaecst64597.2024.11117732",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/iaecst64597.2024.11117732",
+      "retrieved_at": "2026-09-08T14:54:39Z",
+      "chars": 26989,
+      "content_sha256": "sha256:3bdb455d4fca13216e58538bad9e2d8fd5c008c59f83248eaa7f3045718f6bfe"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 5,
+    "has_outline": false,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Tang, X., Wu, S., Zhang, D., Li, F., & Chen, G. (2023). Detecting logic bugs of join optimizations in dbms. Proceedings of the ACM on Management of Data, 1(1), 1-26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Miller, B. P., Fredriksen, L., & So, B. (1990). An Empirical Study of the Reliability of UNIX Utilities. Communications of the ACM, 33(12), 3244. doi:10.1145/96267.96279",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Godefroid, P., et al. (2008). Fuzz testing for database management systems. ACMSIGSOFT Software Engineering Notes, 33(6): 1-10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Zalewski, M. (2014). American Fuzzy Lop: A Security-oriented Fuzzing Tool. Presented at the USENIX Security Symposium, San Diego, CA. pp. 1-10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Handy, S.M., West, D.T. (2020) A Survey on Fuzzing Techniques. In: Proceedings of the 2020 ACM Conference on Computer and Communications Security, pp. 1234–1245.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Liu, B., Zhang, W ., Wang, Y ., Zhang, S., Chen, Z. (2021) Fuzzing: Art, Science, and the State of the Art. In: IEEE Transactions on Software Engineering, 47(3): 254–273.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Bruno, N., Chaudhuri, S., Thomas, D. (2006) Generating queries with cardinality constraints for DBMS testing. IEEE Transactions on Knowledge and Data Engineering, 18: 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Binnig, C., Kossmann, D., Lo, E., et al. (2007) QAGen: Generating Query-Aware Test Databases. In: Proceedings of the ACMSIGMOD International Conference on Management of Data. New York. pp. 341– 352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Lo, E., Binnig, C., Kossmann, D., et al. (2010) A Framework for Testing DBMS Features. The VLDB Journal, 19: 203–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Wen, S., Jia, P., Yang, P., et al. (2023) Squill: Testing DBMS with Correctness Feedback and Accurate Instantiation. Applied Sciences, 13: 2519.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Murphy, B.R., Powers, D.M. (1992) A Differential Testing Methodology for Database Systems. In: Proceedings of the 5th Annual Computer Science Conference. New Y ork. pp. 82–87.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Song, J., Dou, W., Cui, Z., et al. (2023) Testing Database Systems via Differential Query Execution. In: IEEE/ACM International Conference on Software Engineering (ICSE). New York.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Rigger, M., Su, Z. (2020) Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In: Proceedings of the ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. New York. pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "Rigger, M., Su, Z. (2020) Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA): 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "ClickHouse Documentation. (2024). MergeTree Engine. Official Documentation. Retrieved from https://clickhouse.com/docs/en/engines/table-engines/mergetreefamily/mergetree/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Rigger, M., Su, Z. (2020) Testing Database Management Systems via Pivoted Query Synthesis. In: Proceedings of the 41st ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). London. pp. 733–747. 806",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 13,
+      "text": "Rigger, M., Su, Z. (2020) Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In: Proceedings of the ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. New York. pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 14,
+      "text": "Rigger, M., Su, Z. (2020) Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA): 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 16,
+      "text": "Rigger, M., Su, Z. (2020) Testing Database Management Systems via Pivoted Query Synthesis. In: Proceedings of the 41st ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). London. pp. 733–747. 806",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "S Q L a n c e r",
+      "technique": null,
+      "sentence": "We implemented this s y s t e m w i t h i n S Q L a n c e r a n d t e s t e d i t o n p o p u l a r D B M S s, ultimately discovering 10 previously unknown types of bugs.",
+      "context_before": "uaa.edu.cn Abstract— The complexity and diversity of DBMSs present significant challenges and implications for testing. Fuzzing has become a standard approach for detecting bugs in DBMSs. However, the increasing complexity of DBMS architectures brings new challenges to its effectiveness and coverage. This paper introduces the idea of fuzzing DBMS via Isomerism System, which includes multiple instances of one DBMS using different components and configurations. By fuzzing these isomorphic systems, the scope of classic differential testing is expanded, increasing its applicability and efficiency.",
+      "context_after": "The experimental results demonstrate the system's usability and efficiency. Keywords- DBMS; fuzz; isomerism; differential test; logical bugs. I.INTRODUCTION The explosive growth of data has led to a rising demand for data management and analytics across industries. First, to meet complex requirements like data consistency, transaction management, high concurrency, and low latency, Database Management Systems (DBMS) architectures and implementation have become highly complex. Additionally, the diversity in data characteristics and analytical needs across various scenarios has driven the evolut",
+      "section": null,
+      "page": 1,
+      "char_offset": 1094,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": [],
+      "text_is_letter_spaced": true
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "[13] proposed the Non-Optimized Reference Engine Construction (NoREC), aimed at detecting optimization bugs in DBMSs by comparing optimized and non-optimized query s.",
+      "context_before": "is differential testing, where the same query is input into different DBMSs, and the correctness of the results is determined by comparing the differences in their result sets.[11]However, due to the diversity in syntax rules and features across DBMS systems, the disadvantage of differential testing is its limited testable space and poor scalability. S o n g J. e t a l. [12] proposed a general anomaly detection method called Differential Query Execution (DQE). The core idea is that different queries with the same predicates (e.g., UPDATE, SELECT) usually access the same rows. Rigger M. et al.",
+      "context_after": "Given that DBMSs have limited support for optimization control, they established a query transformation mechanism that rewrites queries into forms that the DBMS cannot optimize. In another work.[14] they proposed a query partitioning differential detection method. The core idea is that a given query can be split into multiple complex partitioned queries (called p a r t i t i o n e d q u e r i e s ), a n d t h e r e s u l t s e t o f t h e o r i g i n a l q u e r y should be equivalent to the result sets of all partitioned queries. They argued that partitioned queries are more likely to trigge",
+      "section": null,
+      "page": 2,
+      "char_offset": 7728,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "[14] they proposed a query partitioning differential detection method.",
+      "context_before": "J. e t a l. [12] proposed a general anomaly detection method called Differential Query Execution (DQE). The core idea is that different queries with the same predicates (e.g., UPDATE, SELECT) usually access the same rows. Rigger M. et al. [13] proposed the Non-Optimized Reference Engine Construction (NoREC), aimed at detecting optimization bugs in DBMSs by comparing optimized and non-optimized query s. Given that DBMSs have limited support for optimization control, they established a query transformation mechanism that rewrites queries into forms that the DBMS cannot optimize. In another work.",
+      "context_after": "The core idea is that a given query can be split into multiple complex partitioned queries (called p a r t i t i o n e d q u e r i e s ), a n d t h e r e s u l t s e t o f t h e o r i g i n a l q u e r y should be equivalent to the result sets of all partitioned queries. They argued that partitioned queries are more likely to trigger logical errors due to their higher complexity. In summary, it is evident that current test case generation techniques have become quite advanced, enabling efficient and accurate SQL query generation. Anomaly detection techniques can be divided into two categories",
+      "section": null,
+      "page": 2,
+      "char_offset": 8090,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The query generation approach draws on the design principles of SQLancer [16], with certain modifications.",
+      "context_before": "is crucial to ensure that these INSERT statements belong to the common query sequence set (CQS) of this equivalent configuration group (ECG). Otherwise, the same INSERT statement could generate different data or even fail in one instance but succeed in another. In practice, maintaining a data type converter helps expand the CQS range, covering most INSERT scenarios. For cases where an INSERT fails, a simple backtracking mechanism can be employed, using corresponding DELETE statements to remove its impact on the successful side. Fig. 2. Architecture of Isomerism Fuzzing System Query Generation.",
+      "context_after": "Initially, the fundamental structure of the query is generated randomly based on the abstract syntax tree (AST) corresponding to the DBMS. This structure includes clauses such as SELECT, FROM, WHERE, GROUP BY, and ORDER BY. Then, using existing tables and columns within the database, values, operators, and functions are randomly generated as query components. Within the WHERE and HAVING clauses, various operators (e.g., =, <, >) are applied randomly, and arithmetic operators and aggregate functions (e.g., SUM, COUNT) are incorporated into expressions to form complex query logic. In addition,",
+      "section": null,
+      "page": 3,
+      "char_offset": 15183,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In addition, SQLancer supports not only basic query generation but also more advanced nested queries (such as subqueries) and union queries (such as UNION).",
+      "context_before": "modifications. Initially, the fundamental structure of the query is generated randomly based on the abstract syntax tree (AST) corresponding to the DBMS. This structure includes clauses such as SELECT, FROM, WHERE, GROUP BY, and ORDER BY. Then, using existing tables and columns within the database, values, operators, and functions are randomly generated as query components. Within the WHERE and HAVING clauses, various operators (e.g., =, <, >) are applied randomly, and arithmetic operators and aggregate functions (e.g., SUM, COUNT) are incorporated into expressions to form complex query logic.",
+      "context_after": "These nested and union queries are often able to probe the deeper logic of the DBMS, helping to uncover more intricate bugs. It is important 804 to note that the generation rules must be tailored based on the Common Query Sequence Set (CQS) to avoid generating nonequivalent queries that could interfere with differential testing. 2)Result Comparator To verify whether a query contains logical errors, it is typically necessary to compare the results of all queries for consistency. Some DBMS configuration groups (such as those that use the ORDER BY clause) guarantee a determined order of query",
+      "section": null,
+      "page": 3,
+      "char_offset": 15876,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "CONCLUSION This paper proposes a novel fuzz system for DBMS, offering a new perspective and approach for differential testing, and implements the system in SQLancer.",
+      "context_before": "/MAX Query in MergeTree Engine CREATE TABLE t1 (c0 Int32) ENGINE = MergeTree() ORDER BY c0PARTITION BY (to UInt64 (c0)) INSERT INTO t1 (c0) VALUES (-766544500), (2052205600), (2064494500), (824848640), (0); SELECT MAX(-(c0/c0/c0+ c0)) FROM t1 Table 2. Content of Tableݐଵ in MergeTree Engine and Corresponding V alues of Function݂ ܿ଴݂=−(ܿ଴/ܿ଴/ܿ଴+ܿ଴) -766544500 766544500 2052205600 -2052205600 2064494500 -2064494500 824848640 -824848640 0 NaN Table 3. Query Results for MAX(-(c0/c0/c0+ c0)) Across Different Table Engines Log/Postgres MergeTree 766544500766544500 -2052205600 -2064494500 -824848640 V.",
+      "context_after": "Through this system, we discovered 10 bugs, including some previously unseen new types of bugs. These new types of bugs were not detected by conventional testing methods, indicating that our system has high detection efficiency and accuracy for corner cases. Isomerism fuzz system serves as a general framework that can accommodate a wide range of specific implementations. We aim to expand the framework by testing more DBMSs, configuration groups, and statement types to enhance its usability and coverage. ACKNOWLEDGEMENT This work is supported by the Open Fund of Key Laboratory of Civil Aviatio",
+      "section": null,
+      "page": 5,
+      "char_offset": 23107,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:56:58Z",
+    "is_model_written": true,
+    "summary": "This paper fuzzes a DBMS against itself by running several instances of the same system with different components and configurations -- an 'isomerism system' -- so differential testing becomes possible without a second, separate DBMS. The authors argue that growing architectural complexity limits the effectiveness and coverage of conventional DBMS fuzzing. Testing popular DBMSs this way they discovered 10 previously unknown types of bug.",
+    "narrative": "The system is built inside SQLancer: the paper states it was implemented within SQLancer, and that its query generation draws on SQLancer's design principles with modifications. NoREC and query partitioning are described as the prior techniques it extends the reach of. The sentence establishing this is letter-spaced in the PDF's text layer, so it reads as separated characters; the words are intact.",
+    "roles": {
+      "M1": "reuse_implementation",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "reuse_component",
+      "M5": "background",
+      "M6": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M1",
+          "M4"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "We implemented this s y s t e m w i t h i n S Q L a n c e r a n d t e s t e d i t o n p o p u l a r D B M S s, ultimately discovering 10 previously unknown types of bugs.",
+            "section": null,
+            "page": 1,
+            "text_is_letter_spaced": true
+          },
+          {
+            "mention_id": "M4",
+            "sentence": "The query generation approach draws on the design principles of SQLancer [16], with certain modifications.",
+            "section": null,
+            "page": 3
+          }
+        ],
+        "reasoning": "M1 states the system was implemented within SQLancer, and M4 that the query generation draws on SQLancer's design principles with certain modifications. M1's characters are separated by spaces in the PDF's text layer, which is damage to the rendering rather than to the words.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_iaecst68792_2025_11415166.json b/_data/papers/paper_doi_10_1109_iaecst68792_2025_11415166.json
new file mode 100644
index 0000000..636601a
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_iaecst68792_2025_11415166.json
@@ -0,0 +1,390 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:06:25Z",
+  "paper": {
+    "id": "paper:doi:10.1109/iaecst68792.2025.11415166",
+    "title": "Semantic Hint-Based Fuzzing for Time-Series Databases",
+    "authors": [
+      "Panta Kittisatra",
+      "Liang Liu"
+    ],
+    "year": 2025,
+    "venue": "2025 7th International Academic Exchange Conference on Science and Technology Innovation (IAECST)",
+    "doi": "10.1109/iaecst68792.2025.11415166",
+    "arxiv_id": null,
+    "s2_paper_id": "1e43fae09e6bcbd94078388ec04cd6cba311552c",
+    "url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/iaecst68792.2025.11415166",
+      "retrieved_at": "2026-09-07T17:06:25Z",
+      "chars": 34600,
+      "content_sha256": "sha256:88f20bada766d64b1001aee43e4c1a9f32985d6d98c564b2d283a9ebff19bb0f"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 7,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1566
+      },
+      {
+        "number": "II",
+        "title": "RELATED WORK",
+        "start": 4802
+      },
+      {
+        "number": "III",
+        "title": "METHODOLOGY",
+        "start": 9115
+      },
+      {
+        "number": "A",
+        "title": "Framework Overview",
+        "start": 9132
+      },
+      {
+        "number": "B",
+        "title": "Hint Injection Mechanism",
+        "start": 10327
+      },
+      {
+        "number": "IV",
+        "title": "EXPERIMENTS ANDRESULTS",
+        "start": 13685
+      },
+      {
+        "number": "A",
+        "title": "Experimental Setup",
+        "start": 14151
+      },
+      {
+        "number": "B",
+        "title": "Robustness Results and Runtime Overhead",
+        "start": 16121
+      },
+      {
+        "number": "C",
+        "title": "Representative Bug Case Studies and Analysis",
+        "start": 18378
+      },
+      {
+        "number": "D",
+        "title": "Analysis of Failure Stages and Root Causes",
+        "start": 21402
+      },
+      {
+        "number": "V",
+        "title": "DISCUSSION ANDFUTURE WORK",
+        "start": 26589
+      },
+      {
+        "number": "VI",
+        "title": "CONCLUSION",
+        "start": 28276
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "A. Bader, O. Kopp, and M. Falkenthal, “Survey and Comparison of Open Source Time Series Databases,” in Datenbanksysteme für Business, Technologie und Web (BTW 2017) - Workshopband, Bonn: Gesellschaft für Informatik e.V., 2017, pp. 249– 268.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "E. F. Codd, “Relational database: A practical foundation for productivity,” in ACM Turing award lectures, 2007, p. 1981.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "B. McBride and D. Reynolds, “Survey of time series database technology,” UK Centre for Ecology & Hydrology, Wallingford, Mar. 2020. [Online]. Available: https://nora.nerc.ac.uk/id/eprint/527832/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "M. Wang et al., “Industry practice of coverage -guided enterprise -level DBMS fuzzing,” in Software Engineering: Software Engineering in Practice (ICSE-SEIP), IEEE, 2021, pp. 328– 337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "H. Garcia-Molina, J. D. Ullman, and J. Widom, Database system implementation, vol. 672. Prentice Hall Upper Saddle River, 2000.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 1– 30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 7,
+      "text": "J. Wei, P. Chen, K. Lu, J. Dai, and X. Sun, “SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing.” 2024. [Online]. Available: https://arxiv.org/abs/2407.04294",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "T. Y. Chen et al., “Metamorphic testing: A review of challenges and opportunities,” ACM Comput. Surv. CSUR, vol. 51, no. 1, pp. 1– 27, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "G. Klees, A. Ruef, B. Cooper, S. Wei, and M. Hicks, “Evaluating fuzz testing,” in Proceedings of the 2018 ACMSIGSAC conference on computer and communications security, 2018, pp. 2123– 2138.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "C. Wang et al., “Apache iotdb: A time series database for iot applications,” 1, no. 2, pp. 1– 27, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "P. Tripathi, M. H. Miraz, and S. Joshi, “Comparative Analysis of MongoDB and InfluxDB for Time Series Data Management in IoT Environments: A Study on Performance, Scalability, and Concurrency,” in 2023 International Conference on Computing, Networking, Telecommunications & Engineering Sciences Applications (CoNTESA), 2023, pp. 39 –42. doi: 10.1109/CoNTESA61248.2023.10384962.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "M. Parvizimosaed, M. Noei, M. Yalpanian, and J. Bahrami, “A Containerized Integrated Fast IoT Platform for Low Energy Power Management,” in 2021 7th International Conference on Web Research (ICWR), 2021, pp. 318– 322. doi: 10.1109/ICWR51868.2021.9443141.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "D. D. Chamberlin and R. F. Boyce, “SEQUEL: A structured English query language,” in Proceedings of the 1974 ACMSIGFIDET (now SIGMOD) workshop on Data description, access and control, 1974, pp. 249– 264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667– 682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "V. M. Grippa and S. Kuzmichev, Learning MySQL. O’Reilly Media, Inc., 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "A. Makris, K. Tserpes, G. Spiliopoulos, D. Zissis, and D. Anagnostopoulos, “MongoDB Vs PostgreSQL: A comparative study on performance aspects,” GeoInformatica, vol. 25, no. 2, pp. 243– 268, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "K. P. Gaffney, M. Prammer, L. Brasfield, D. R. Hipp, D. Kennedy, and J. M. Patel, “SQLite: past, present, and future,” Proc. VLDB Endow., vol. 15, no. 12, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "L. Kuang et al., “TSGuard: Detecting Logic Bugs in Time Series Management Systems Via Time Series Algebra,” in (ICSME), Los Alamitos, CA, USA: IEEE Computer Society, Sept. 2025, pp. 1– 13. doi: 10.1109/ICSME64153.2025.00030. 268",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 6,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 1– 30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667– 682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Tools such as SQLancer [6], SQLaser [7], and mutation -based engines like AF L leverage metamorphic testing to uncover logic bugs and optimizer inconsistencies [8], [9].",
+      "context_before": "TSDBs are optimized for handling high -volume, time -stamped data and executing temporal queries that enable trend analysis, anomaly detection, and real -time decision making [2]. Despite their increasing adoption in production environments, the correctness and robustness of TSDB query processing remain insufficiently studie d[3] even though minor semantic inconsistencies can propagate into inaccurate analytics, faulty monitoring decisions, or operational failures. Research in DBMS testing has made notable progress through fuzzing -based approaches [4], [5], particularly in relational systems.",
+      "context_after": "However, these frameworks largely assume relational semantics [6], static schemas, and Boolean logic evaluation, making them ineffective for dealing with temporal operators, continuous data ingestion, window functions, and time -aware aggregations that are fundamental in TSDBs. As a result, correctness testing for time -series workloads remains comparatively underexplored, with only limited work specifically targeting TSDB behaviors. บ represents one of the few TSDB -focused fuzzing frameworks, detecting logic bugs through time -series algebra and query mutatio n, yet it still lacks mechanism",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2466,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[6]",
+      "technique": "tlp",
+      "sentence": "However, these frameworks largely assume relational semantics [6], static schemas, and Boolean logic evaluation, making them ineffective for dealing with temporal operators, continuous data ingestion, window functions, and time -aware aggregations that are fundamental in TSDBs.",
+      "context_before": "king [2]. Despite their increasing adoption in production environments, the correctness and robustness of TSDB query processing remain insufficiently studie d[3] even though minor semantic inconsistencies can propagate into inaccurate analytics, faulty monitoring decisions, or operational failures. Research in DBMS testing has made notable progress through fuzzing -based approaches [4], [5], particularly in relational systems. Tools such as SQLancer [6], SQLaser [7], and mutation -based engines like AF L leverage metamorphic testing to uncover logic bugs and optimizer inconsistencies [8], [9].",
+      "context_after": "As a result, correctness testing for time -series workloads remains comparatively underexplored, with only limited work specifically targeting TSDB behaviors. บ represents one of the few TSDB -focused fuzzing frameworks, detecting logic bugs through time -series algebra and query mutatio n, yet it still lacks mechanisms for validating semantic equivalence or identifying robustness weaknesses triggered by harmless predicate variations. To address this gap, we introduce Semantic Hint-Based Fuzzing, a lightweight extension to TSGuard designed to evaluate robustness and semantic stability in TSDB",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2636,
+      "cited_reference": {
+        "number": 6,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 1– 30, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[14]",
+      "technique": "pqs",
+      "sentence": "RELATED WORK Research on testing relational DBMS, which supports a standardized query language SQL, has evolved considerably over the past decades [13], [14].",
+      "context_before": "], Influx DB[11], and TDengin e[12], executing more than 10,000 query pairs and revealing multiple internal errors and optimizer failures triggered solely by neutral semantic transformations. While no correctness mismatches were observed, findings show that even mature TSDBs remain vulnerab le to minor perturbations, highlighting the importance of lightweight semantic validation techniques. This work demonstrates the practicality of hint -based fuzzing as a scalable foundation for robustness assurance in TSDB systems and motivates future resea rch toward automated semantic test generation. II.",
+      "context_after": "Early approaches mainly relied on random SQL generation to expose robustness issues. For example, SQLsmith adopts stochastic SQL synthesis to produce large numbers of valid statements and observe system behavior under unexpected query patterns. Although effective for discovering crash -inducing inputs, this method cannot validate semantic correctness due to the absence of an oracle capable of determining whether query outputs are equivalent. 2025 7th International Academic Exchange Conference on Science and Technology Innovation (IAECST) 979-8-3315-8024-7/25/$31.00 ©2025 IEEE 2622025 7th Inte",
+      "section": "II RELATED WORK",
+      "page": 1,
+      "char_offset": 4805,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667– 682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer pioneered systematic bug detection using oracles such as Non-Optimizing Reference Engine Construction (NoREC) and Ternary Logic Partitioning (TLP), enabling reliable detection of logic bugs even when databases do not crash [6].",
+      "context_before": "uery outputs are equivalent. 2025 7th International Academic Exchange Conference on Science and Technology Innovation (IAECST) 979-8-3315-8024-7/25/$31.00 ©2025 IEEE 2622025 7th International Academic Exchange Conference on Science and Technology Innovation (IAECST) | 979-8-3315-8024-7/25/$31.00 ©2025 IEEE | DOI: 10.1109/IAECST68792.2025.11415166 To address correctness verification, researchers introduced metamorphic testing [8], where logically related query pairs are generated, executed, and compared for output consistency. This idea became fundamental to several modern fuzzing frameworks.",
+      "context_after": "Later frameworks such as SQLaser, Kangaroo, and Pivoted Query Synthesis (PQS) extended equivalence based testing using clause -guided strategies and optimizer aware synthesis to reveal subtle logical faults in popular relational systems including MySQL [15], PostgreSQL [16], and SQLite [17]. However, these techniques are designed around relational semantics, assuming static schemas and Boolean logic. They are not aware of time -based operations, highfrequency ingestion, or temporal aggregations — behaviors fundamental to time-series database s. Unlike relational DBMS, TSDBs are optimized for",
+      "section": "II RELATED WORK",
+      "page": 2,
+      "char_offset": 5982,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": "Later frameworks such as SQLaser, Kangaroo, and Pivoted Query Synthesis (PQS) extended equivalence based testing using clause -guided strategies and optimizer aware synthesis to reveal subtle logical faults in popular relational systems including MySQL [15], PostgreSQL [16], and SQLite [17].",
+      "context_before": "echnology Innovation (IAECST) | 979-8-3315-8024-7/25/$31.00 ©2025 IEEE | DOI: 10.1109/IAECST68792.2025.11415166 To address correctness verification, researchers introduced metamorphic testing [8], where logically related query pairs are generated, executed, and compared for output consistency. This idea became fundamental to several modern fuzzing frameworks. SQLancer pioneered systematic bug detection using oracles such as Non-Optimizing Reference Engine Construction (NoREC) and Ternary Logic Partitioning (TLP), enabling reliable detection of logic bugs even when databases do not crash [6].",
+      "context_after": "However, these techniques are designed around relational semantics, assuming static schemas and Boolean logic. They are not aware of time -based operations, highfrequency ingestion, or temporal aggregations — behaviors fundamental to time-series database s. Unlike relational DBMS, TSDBs are optimized for time stamped data streams, continuous ingestion, and temporal filtering used in IoT analytics [1]. Testing TSDBs introduces additional challenges due to sliding windows, interval queries, and timestamp -based predicates, which traditional SQL fuzzers cannot adequately model. To address TSDB-s",
+      "section": "II RELATED WORK",
+      "page": 2,
+      "char_offset": 6219,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:21:29Z",
+    "is_model_written": true,
+    "summary": "This paper adds a semantic hint-based oracle to TSGuard for testing time-series databases, arguing that existing fuzzing frameworks are built for relational DBMSs and ignore the temporal semantics and query processing peculiar to time-series systems. The oracle injects logically neutral predicates such as tautologies and redundant conditions into a baseline query and compares results differentially, so query intent is unchanged but robustness is probed. On Apache IoTDB, InfluxDB and TDengine it surfaced several optimizer-related crashes.",
+    "narrative": "SQLancer is the pioneer the paper builds from and the limitation it works around: it is credited with pioneering systematic bug detection through oracles such as NoREC and TLP, and then said to assume relational semantics, static schemas and Boolean logic, which is what makes it ineffective for time-series data.",
+    "roles": {
+      "M1": "background",
+      "M2": "motivation",
+      "M3": "background",
+      "M4": "state_of_the_art",
+      "M5": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M4"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "SQLancer pioneered systematic bug detection using oracles such as Non-Optimizing Reference Engine Construction (NoREC) and Ternary Logic Partitioning (TLP), enabling reliable detection of logic bugs even when databases do not crash [6].",
+            "section": "II RELATED WORK",
+            "page": 2
+          }
+        ],
+        "reasoning": "M4 credits SQLancer with pioneering systematic bug detection using oracles such as NoREC and TLP, which is a claim about its standing in the field rather than a description of a method."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icccs65393_2025_11069832.json b/_data/papers/paper_doi_10_1109_icccs65393_2025_11069832.json
new file mode 100644
index 0000000..d36cef9
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icccs65393_2025_11069832.json
@@ -0,0 +1,396 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:54:04Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icccs65393.2025.11069832",
+    "title": "OptionFuzz: The Fuzzer with Coverage-Guided Dynamic Configuration Scheduling",
+    "authors": [
+      "Lang Chu",
+      "Minhuan Huang",
+      "Xiang Li",
+      "Yuanping Nie",
+      "Wenyu Zhen",
+      "Qian Yan"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Communication, Computing & Security",
+    "doi": "10.1109/icccs65393.2025.11069832",
+    "arxiv_id": null,
+    "s2_paper_id": "46eca54a5aca680d689a06d3f93a3852b7fabefb",
+    "url": "https://doi.org/10.1109/icccs65393.2025.11069832",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icccs65393.2025.11069832",
+      "retrieved_at": "2026-09-08T14:54:04Z",
+      "chars": 45441,
+      "content_sha256": "sha256:ef0baad05e7bf6cafe21fc2e439d12262be421ef70e035fdf119fd964038ade1"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 10,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2505
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND AND RELATED WORK",
+        "start": 8819
+      },
+      {
+        "number": "A",
+        "title": "Fuzzing",
+        "start": 8851
+      },
+      {
+        "number": "B",
+        "title": "Command-Line Options and Their Combinatorial Testing",
+        "start": 10113
+      },
+      {
+        "number": "C",
+        "title": "Related Work on Configuration Fuzzing",
+        "start": 11051
+      },
+      {
+        "number": "III",
+        "title": "DESIGN",
+        "start": 13305
+      },
+      {
+        "number": "A",
+        "title": "Configuration Generation Based on the Markov Decision",
+        "start": 14745
+      },
+      {
+        "number": "B",
+        "title": "Configuration Scheduling Guided by Function Coverage",
+        "start": 23476
+      },
+      {
+        "number": "IV",
+        "title": "IMPLEMENTATION AND EVALUATION",
+        "start": 27493
+      },
+      {
+        "number": "A",
+        "title": "How does OptionFuzz improve compared to traditional",
+        "start": 28601
+      },
+      {
+        "number": "B",
+        "title": "How much does function coverage analysis in OptionFuzz",
+        "start": 30683
+      },
+      {
+        "number": "C",
+        "title": "How does OptionFuzz’s fuzzing performance compare to",
+        "start": 31468
+      },
+      {
+        "number": "V",
+        "title": "DISCUSSION",
+        "start": 33365
+      },
+      {
+        "number": "A",
+        "title": "Limitations and future work",
+        "start": 33379
+      },
+      {
+        "number": "B",
+        "title": "Advantages of Function Coverage-Guided Configuration",
+        "start": 35015
+      },
+      {
+        "number": "C",
+        "title": "What are the benefits of using a Markov process based on",
+        "start": 36207
+      },
+      {
+        "number": "VI",
+        "title": "CONCLUSION",
+        "start": 37980
+      },
+      {
+        "number": "J",
+        "title": "M. Atlee, T. Bultan, and J. Whittle, Eds. IEEE /",
+        "start": 39202
+      },
+      {
+        "number": "T",
+        "title": "Malkin, and D. Xu, Eds. ACM, 2017, pp. 2329–2344.",
+        "start": 40178
+      },
+      {
+        "number": "B",
+        "title": "Lee, “HFL: hybrid fuzzing on the linux kernel,” in",
+        "start": 41198
+      },
+      {
+        "number": "T",
+        "title": "Holz, “HYPER-CUBE: high-dimensional hypervisor",
+        "start": 41859
+      },
+      {
+        "number": "K",
+        "title": "Thomas, Eds. USENIX Association, 2022, pp. 3255–",
+        "start": 42275
+      },
+      {
+        "number": "S",
+        "title": "Zennou, Eds. USENIX Association, 2020.",
+        "start": 44689
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Oss-fuzzcontinuous fuzzing for open source software.” [Online]. Available: https://github.com/google/ oss-fuzz",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "J. Choi, J. Jang, C. Han, and S. K. Cha, “Grey-box concolic testing on binary code,” in Proceedings of the 41st International Conference on Software Engineering, ICSE 2019, Montreal, QC, Canada, May 25-31, 2019, J. M. Atlee, T. Bultan, and J. Whittle, Eds. IEEE / ACM, 2019, pp. 736–747.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "C. Aschermann, S. Schumilo, A. Abbasi, and T. Holz, “Ijon: Exploring deep state spaces via fuzzing,” in 1005 Francisco, CA, USA, May 18-21, 2020. IEEE, 2020, pp. 1597–1612.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "A. Herrera, H. Gunadi, S. Magrath, M. Norrish, M. Payer, and A. L. Hosking, “Seed selection for successful fuzzing,” in ISSTA ’21: 30th ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, Denmark, July 11-17, 2021, C. Cadar and X. Zhang, Eds. ACM, 2021, pp. 230–243.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "M. B ¨ohme, V. Pham, and A. Roychoudhury, “Coveragebased greybox fuzzing as markov chain,” IEEE Trans. Software Eng., vol. 45, no. 5, pp. 489–506, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "M. B ¨ohme, V. Pham, M. Nguyen, and A. Roychoudhury, “Directed greybox fuzzing,” in Proceedings of the 2017 ACMSIGSAC Conference on Computer and Communications Security, CCS 2017, Dallas, TX, USA, October 30 - November 03, 2017, B. Thuraisingham, D. Evans, T. Malkin, and D. Xu, Eds. ACM, 2017, pp. 2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "C. Aschermann, T. Frassetto, T. Holz, P. Jauernig, A. Sadeghi, and D. Teuchert, “NAUTILUS: fishing for deep bugs with grammars,” in 26th Annual Network and Distributed System Security Symposium, NDSS 2019, San Diego, California, USA, February 24-27, 2019. The Internet Society, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "T. Blazytko, C. Aschermann, M. Schl ¨ogel, A. Abbasi, S. Schumilo, S. W ¨orner, and T. Holz, “GRIMOIRE: synthesizing structure while fuzzing,” in 28th USENIX Security Symposium, USENIX Security 2019, Santa Clara, CA, USA, August 14-16, 2019, N. Heninger and P. Traynor, Eds. USENIX Association, 2019, pp. 1985–2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "S. Schumilo, C. Aschermann, R. Gawlik, S. Schinzel, and T. Holz, “kafl: Hardware-assisted feedback fuzzing for OS kernels,” in 26th USENIX Security Symposium, USENIX Security 2017, Vancouver, BC, Canada, August 16-18, 2017, E. Kirda and T. Ristenpart, Eds. USENIX Association, 2017, pp. 167–182.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "K. Kim, D. R. Jeong, C. H. Kim, Y. Jang, I. Shin, and B. Lee, “HFL: hybrid fuzzing on the linux kernel,” in 27th Annual Network and Distributed System Security Symposium, NDSS 2020, San Diego, California, USA, February 23-26, 2020. The Internet Society, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "D. R. Jeong, K. Kim, B. Shivakumar, B. Lee, and I. Shin, “Razzer: Finding kernel race bugs through fuzzing,” in San Francisco, CA, USA, May 19-23, 2019. IEEE, 2019, pp. 754–768.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "W. Xu, H. Moon, S. Kashyap, P. Tseng, and T. Kim, “Fuzzing file systems via two-dimensional input space exploration,” in Privacy, SP 2019, San Francisco, CA, USA, May 19-23, 2019. IEEE, 2019, pp. 818–834.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "S. Schumilo, C. Aschermann, A. Abbasi, S. W ¨orner, and T. Holz, “HYPER-CUBE: high-dimensional hypervisor fuzzing,” in 27th Annual Network and Distributed System Security Symposium, NDSS 2020, San Diego, California, USA, February 23-26, 2020. The Internet Society, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "J. Ba, M. B ¨ohme, Z. Mirzamomen, and A. Roychoudhury, “Stateful greybox fuzzing,” in 31st USENIX Security Symposium, USENIX Security 2022, Boston, MA, USA, August 10-12, 2022, K. R. B. Butler and K. Thomas, Eds. USENIX Association, 2022, pp. 3255– 3272.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 16,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “SQUIRREL: testing database management systems with language validity and coverage feedback,” in CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event, USA, November 913, 2020, J. Ligatti, X. Ou, J. Katz, and G. Vigna, Eds. ACM, 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "T. Kim, C. H. Kim, J. Rhee, F. Fei, Z. Tu, G. Walkup, X. Zhang, X. Deng, and D. Xu, “Rvfuzzer: Finding input validation bugs in robotic vehicles through controlguided testing,” in 28th USENIX Security Symposium, USENIX Security 2019, Santa Clara, CA, USA, August 14-16, 2019, N. Heninger and P. Traynor, Eds. USENIX Association, 2019, pp. 425–442.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "“Ffmpeg.” [Online]. Available: https://ffmpeg.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Z. Zhang, G. Klees, E. Wang, M. Hicks, and S. Wei, “Fuzzing configurations of program options-RCR report,” ACM Trans. Softw. Eng. Methodol., vol. 32, no. 2, pp. 55:1–55:3, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "A. Lee, I. Ariq, Y. Kim, and M. Kim, “POWER: program option-aware fuzzer for high bug detection ability,” in 15th IEEE Conference on Software Testing, Verification and Validation, ICST 2022, Valencia, Spain, April 4-14, 2022. IEEE, 2022, pp. 220–231.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "D. Wang, Y. Li, Z. Zhang, and K. Chen, “Carpetfuzz: Automatic program option constraint extraction from documentation for fuzzing,” in 32nd USENIX Security Symposium, USENIX Security 2023, Anaheim, CA, USA, August 9-11, 2023, J. A. Calandrino and C. Troncoso, Eds. USENIX Association, 2023, pp. 1919–1936.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "J. Li, S. Li, K. Li, F. Luo, H. Yu, S. Li, and X. Li, “Ecfuzz: Effective configuration fuzzing for large-scale systems,” in Proceedings of the 46th IEEE/ACM International Conference on Software Engineering, ICSE 2024, Lisbon, Portugal, April 14-20, 2024. ACM, 2024, pp. 48:1–48:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "M. Zalewski, “american fuzzy lopa security-oriented fuzzer.” [Online]. Available: https://github.com/google/ AFL",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "A. Fioraldi, D. C. Maier, H. Eißfeldt, and M. Heuse, “AFL++: Combining incremental steps of fuzzing research,” in 14th USENIX Workshop on Offensive Technologies, WOOT 2020, August 11, 2020, Y. Yarom and S. Zennou, Eds. USENIX Association, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "“libfuzzer – a library for coverage-guided fuzz testing.” 1006 [Online]. Available: https://github.com/llvm-mirror/llvm/ blob/master/docs/LibFuzzer.rst",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "“honggfuzz.” [Online]. Available: https://github.com/ google/honggfuzz",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "C. Nie and H. Leung, “A survey of combinatorial testing,” ACM Comput. Surv., vol. 43, no. 2, pp. 11:1–11:29, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "M. B. Cohen, P. B. Gibbons, W. B. Mugridge, and C. J. Colbourn, “Constructing test suites for interaction testing,” in 25th International Conference on Software Engineering, 2003. Proceedings. IEEE, 2003, pp. 38– 48.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "“Produce code coverage results with gcov from aflfuzz test cases.” [Online]. Available: https://github.com/ mrash/afl-cov 1007",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 15,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[15, 16]",
+      "technique": "pqs",
+      "sentence": "The second aspect aims to expand the application of fuzzing to more targets, such as operating systems[9–12], virtual machine managers[13], network protocols[13, 14], database systems[15, 16], and even autonomous vehicles[17].",
+      "context_before": "lly detect vulnerabilities in open-source software.Since its launch in 2016, it has helped discover and fix thousands of potential security issues. Similarly, Microsoft has leveraged its internal fuzzing projects to successfully identify multiple critical vulnerabilities in its Windows and Office products. Currently, research on fuzzing primarily focuses on two aspects. The first aspect involves optimizing the fuzzing process itself, including strategies such as feedback optimization[2, 3], corpus optimization[4], seed scheduling optimization[5, 6], and test case generation optimization[7, 8].",
+      "context_after": "However, despite the powerful vulnerability detection capabilities of the latest fuzzing tools, most fuzzing efforts often overlook a crucial factor—the configuration of the target system—which may limit the exploration capabilities of fuzzing. In the real world, many programs typically have dozens to hundreds of command-line options. These options provide significant flexibility for the program, controlling its execution process and enhancing development efficiency. For instance, tools like FFmpeg[18] possess hundreds of options that can be enabled or disabled at runtime. We define the combi",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3620,
+      "cited_reference": {
+        "number": 15,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:02:36Z",
+    "is_model_written": true,
+    "summary": "OptionFuzz schedules configuration options during coverage-guided fuzzing, on the observation that much of a program's code is reachable only under particular settings, so a fuzzer that never varies configuration cannot reach it.",
+    "narrative": "Database systems appear in the opening survey of targets fuzzing has been extended to, alongside operating systems, virtual machine managers, network protocols and autonomous vehicles. The mention is reachable only through the citation marker and carries no relationship beyond acknowledging the domain.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_iccste65902_2025_11138310.json b/_data/papers/paper_doi_10_1109_iccste65902_2025_11138310.json
new file mode 100644
index 0000000..acd039d
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_iccste65902_2025_11138310.json
@@ -0,0 +1,258 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:25:01Z",
+  "paper": {
+    "id": "paper:doi:10.1109/iccste65902.2025.11138310",
+    "title": "Research on Coverage Statistics in Fuzz Testing of Closed-Source DBMS",
+    "authors": [
+      "Zhongjie Li",
+      "Hao-Tian Liang",
+      "Haoyang Jia",
+      "Qingxian Wang",
+      "Yan Cao"
+    ],
+    "year": 2025,
+    "venue": "2025 International Conference on Computer Science, Technology and Engineering (ICCSTE)",
+    "doi": "10.1109/iccste65902.2025.11138310",
+    "arxiv_id": null,
+    "s2_paper_id": "eac24dcac11bfbca2e3aa2b43a6c8c08c68f324a",
+    "url": "https://doi.org/10.1109/iccste65902.2025.11138310",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/iccste65902.2025.11138310",
+      "retrieved_at": "2026-09-08T15:25:01Z",
+      "chars": 34327,
+      "content_sha256": "sha256:a37800c6268ae2b25478bbf0a94af104a41bff02fe63a136418dc4186941814e"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 7,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1735
+      },
+      {
+        "number": "II",
+        "title": "COVERAGE FEEDBACK",
+        "start": 5887
+      },
+      {
+        "number": "A",
+        "title": "Basic Implementation of the TrCov",
+        "start": 6586
+      },
+      {
+        "number": "B",
+        "title": "Optimization of TrCov Based on SQL Execution",
+        "start": 12040
+      },
+      {
+        "number": "III",
+        "title": "EXPERIMENTAL TESTING",
+        "start": 20133
+      },
+      {
+        "number": "IV",
+        "title": "LIMITATIONS ANDFUTURE OUTLOOK",
+        "start": 27498
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Wang, J. H., Song, C. Y ., Yin, H., & Internet, S. ( 2021). Reinforcement Learning-based Hierarchical Seed Sche duling for Greybox Fuzzing. 28th Annual Network and Distribute d System Security Symposium (NDSS). https://doi.org/10.14722/ndss.2021.24486.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Zhong, R., Chen, Y. H., Hu, H., Zhang, H. F., Lee, W. K., Wu, D. H., & Assoc Comp, M. (2020). SQUIRREL: Testing Database Management Systems with Language Validity and Cover age Feedback. ACMSIGSAC Conference on Computer and Communications Security (ACMCCS), 955-970. https://doi.org/10.1145/3372297.3417260.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Jiang, Z. M., Bai, J. J., Su, Z. D., & Association, U. (2023). DynSQL: Stateful Fuzzing for Database Management Sy stems with Complex and Valid SQL Query Generation. 32nd USENIX Security Symposium, 4949-4965. Retrieved from <Go to ISI>://WOS:001066451505008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Rigger, M., Su, Z. D., & Assoc, U. (2020). Testing Database Engines via Pivoted Query Synthesis. 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), 667-682. Retrieved from <Go to ISI>://WOS:000668979500038.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "Dou, W. S., Cui, Z. Y ., Dai, Q. W., Song, J. S., Wa ng, D., Gao, Y .,. .. Ieee. (2023). Detecting Isolation Bugs via Transact ion Oracle Construction. 45th IEEE/ACM International Conferenc e on Software Engineering (ICSE), 1123-1135. https://doi.org/10.1109/icse48619.2023.00101.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Jung, J. H., Hu, H., Arulraj, J., Kim, T., & Kang, W. (2019). APOLLO: Automatic Detection and Diagnosis of Perfor mance Regressions in Database Systems. Proceedings of the VLDB Endowment, 13(1), 57-70. https://doi.org/10.14778/3357377.3357382.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Zheng, Y. Y ., Dou, W. S., Wang, Y. C., Qin, Z., Tan g, L., Gao, Y .,. .. Wei, J. (2022). Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing. 31st ACM SIGSOFT International Symposium on Software Testing and Ana lysis (ISSTA), 302-313. https://doi.org/10.1145/3533767.3534409.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Yang, Y ., Chen, Y ., Zhong, R., Chen, J., & Lee, W. (2024). Towards Generic Database Management System Fuzzing. 33rd USENIX Security Symposium (USENIX Security 24), 901-918.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Li, J., Wang, K., Chen, Y ., Zhou, Y ., Wu, L., & Wan g, J. (2023). Detecting DBMS Bugs with Context-Sensitive Instanti ation and Multi-Plan Execution. arXiv preprint arXiv:2312.049 41.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Hua, Z. Y ., Lin, W., Ren, L. Y ., Li, Z. Y ., Zhang, L., Jiao, W., & Xie, T. (2023). GDsmith: Detecting Bugs in Cypher Graph Database Engines. 32nd ACMSIGSOFT International Symposium o n Software Testing and Analysis (ISSTA), 163-174. https://doi.org/10.1145/3597926.3598046.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Fu, J. Z., Liang, J., Wu, Z. Y ., Wang, M. Z., Jiang, Y ., & Assoc Computing, M. (2022). Griffin: Grammar-Free DBMS Fu zzing. 37th IEEE/ACM International Conference on Automated Software Engineering (ASE). https://doi.org/10.1145/3551349. 3560431.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jiang, Z. M., Bai, J. J., Lu, K., & Hu, S. M. (2020 ). Fuzzing Error Handling Code using {Context-Sensitive} Software Fa ult Injection. In 29th USENIX Security Symposium (USENIX Security 20) (pp. 2595-2612).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Liu, X., Zhou, Q., Arulraj, J., & Orso, A. (2022, M ay). Automatic detection of performance bugs in database systems u sing equivalent queries. In Proceedings of the 44th International C onference on Software Engineering (pp. 225-236).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Fioraldi, A., Maier, D., Eißfeldt, H., & Heuse, M. (2020). AFL++: Combining incremental steps of fuzzing research. 14 th USENIX Workshop on Offensive Technologies (WOOT 20).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Chen, P., Chen, H., & Ieee. (2018). Angora: Efficie nt Fuzzing by Principled Search. 39th IEEE Symposium on Security and Privacy (SP), 711-725. https://doi.org/10.1109/sp.2018.0004 6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Wu, Z., Liang, J., Wang, M., Zhou, C., & Jiang, Y. (2022, July). Unicorn: detect runtime errors in time-series datab ases with hybrid input synthesis. In Proceedings of the 31st ACMSIG SOFT International Symposium on Software Testing and Ana lysis (pp. 251-262). 145",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Ba, J. S., Rigger, M., & Ieee. (2023). Testing Data base Engines via Query Plan Guidance. 45th IEEE/ACM International Co nference on Software Engineering (ICSE), 2060-2071. https://doi.org/10.1109/icse48619.2023.00174.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 18,
+      "text": "Trickel, E., Pagani, F., Zhu, C., Dresel, L., Vigna, G ., Kruegel, C.,. .. Ieee. (2023). Toss a Fault to YourWitcher: Applying Grey-box Coverage-Guided Mutational Fuzzing to Detect SQL an d Command Injection Vulnerabilities. 44th IEEE Symposium on S ecurity and Privacy (SP), 2658-2675. https://doi.org/10.1109/sp46215.2023.10179317. 146",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 4,
+      "text": "Rigger, M., Su, Z. D., & Assoc, U. (2020). Testing Database Engines via Pivoted Query Synthesis. 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), 667-682. Retrieved from <Go to ISI>://WOS:000668979500038.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 17,
+      "text": "Ba, J. S., Rigger, M., & Ieee. (2023). Testing Data base Engines via Query Plan Guidance. 45th IEEE/ACM International Co nference on Software Engineering (ICSE), 2060-2071. https://doi.org/10.1109/icse48619.2023.00174.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "prints the DOI of the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [4] and Troc [5] employ pre-established oracles to detect logical and trans action isolation errors in DBMSs.",
+      "context_before": "lly. However, how to detect and resolve DBMS vulnerabili ties more efficiently and systematically remains one of the core challenges in database security research today. Fuzz testing is a widely adopted method for vulnerability discovery [1]. The well-known fuzz te sting framework AFL has demonstrated that this random input-based testing method can uncover numerous previously unknown software vulnerabilities. This technique has been extensively applied in DBMS test ing. For instance, SQLsmith, Squirrel [2], and DynSQL [3 ] generate continuous SQL queries to detect memory er rors in the system.",
+      "context_after": "APOLLO [6] identifies performance regression issues by comparing differen t DBMS versions. Additionally, specialized tools such as Grand [7] are designed for graph-based DBMS testing, while newer tools like BUZZBEE [8] have emerged for testing non-relational DBMSs such as Redis. These research advancements have helped uncover various deep-seated bugs and vulnerabilities across differe nt DBMS types [9], [10], [11]. However, all these test ing tools are exclusively applicable to open-source DBMSs. Attempting to apply these tools to closed-source DBMS vulnerability detection introduces several",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3586,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:02:36Z",
+    "is_model_written": true,
+    "summary": "This paper addresses coverage measurement when fuzzing closed-source DBMSs, where the instrumentation that coverage-guided fuzzing depends on cannot be inserted because the source is unavailable.",
+    "narrative": "SQLancer is cited once, in the introduction's survey of DBMS testing: it and Troc are described as employing pre-established oracles to detect logic and transaction isolation errors. That framing is what the paper contrasts with its own concern, which is measuring coverage rather than deciding correctness.",
+    "roles": {
+      "M1": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icde55515_2023_00057.json b/_data/papers/paper_doi_10_1109_icde55515_2023_00057.json
new file mode 100644
index 0000000..3db092d
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icde55515_2023_00057.json
@@ -0,0 +1,1009 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:18Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icde55515.2023.00057",
+    "title": "Sequence-Oriented DBMS Fuzzing",
+    "authors": [
+      "Jie Liang",
+      "Yaoguang Chen",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Mingzhe Wang",
+      "Yu Jiang",
+      "Xiangdong Huang",
+      "Ting Chen",
+      "Jiashui Wang",
+      "Jiajia Li"
+    ],
+    "year": 2023,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde55515.2023.00057",
+    "arxiv_id": null,
+    "s2_paper_id": "058f7a552121c7417e783c995f8bc584d09b3eff",
+    "url": "https://doi.org/10.1109/icde55515.2023.00057",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icde55515.2023.00057",
+      "retrieved_at": "2026-09-09T01:33:18Z",
+      "chars": 77636,
+      "content_sha256": "sha256:dd11ec7c8550976731e52fa4cee502bfa0dfa9cd6ce4091b0283990e80e1e5bb"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 14,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1696
+      },
+      {
+        "number": "II",
+        "title": "SQL TYPE SEQUENCE",
+        "start": 9159
+      },
+      {
+        "number": "III",
+        "title": "DESIGN OFLEGO",
+        "start": 22208
+      },
+      {
+        "number": "A",
+        "title": "Proactive Affinity Analysis",
+        "start": 22465
+      },
+      {
+        "number": "B",
+        "title": "Progressive Sequence Synthesis",
+        "start": 30452
+      },
+      {
+        "number": "A",
+        "title": "Evaluation Setup",
+        "start": 38813
+      },
+      {
+        "number": "B",
+        "title": "DBMS Vulnerability Detection",
+        "start": 42370
+      },
+      {
+        "number": "C",
+        "title": "Comparison with Other DBMS Fuzzers",
+        "start": 49592
+      },
+      {
+        "number": "D",
+        "title": "Effectiveness of Sequence-Oriented Algorithms in LEGO",
+        "start": 54764
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. 2010. Automated SQL query generation for systematic testing of database engines. In Proceedings of the IEEE/ACM international conference on Automated software engineering. 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "alloy 2022. Documentation Alloy 6. https:// alloytools .org/documentation .html. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Cornelius Aschermann, Sergej Schumilo, Tim Blazytko, Robert Gawlik, and Thorsten Holz. 2019. REDQUEEN: Fuzzing with Input-to-State Correspondence. In Symposium on Network and Distributed System Security (NDSS).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Adam Bannister. 2021. SQLite patches use-after-free bug that left apps open to code execution, denial-ofservice exploits. https://portswigger .net/daily-swig/ sqlite-patches-use-after-free-bug-that-left-apps-open-tocode-execution-denial-of-service-exploits. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Daniel Bartholomew. 2014. MariaDB cookbook. Packt Publishing Ltd.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "ST Bhosale, Miss Tejaswini Patil, and Miss Pooja Patil. 2015. Sqlite: Light database system. Int. J. Comput. Sci. Mob. Comput 44, 4 (2015), 882–885.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Bison 2022. Bison. https://www .gnu.org/software/bison/. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Marcel B ¨ohme, L ´aszl´o Szekeres, and Jonathan Metzman. 2022. On the Reliability of Coverage-Based Fuzzer Benchmarking. In 44th IEEE/ACM International Conference on Software Engineering, ser. ICSE, V ol. 22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Yuanliang Chen, Yu Jiang, Fuchen Ma, Jie Liang, Mingzhe Wang, Chijin Zhou, Xun Jiao, and Zhuo Su. 2019. EnFuzz: Ensemble Fuzzing with Seed Synchronization among Diverse Fuzzers. In USENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Catalin Cimpanu. 2019. Google Chrome impacted by new Magellan 2.0 vulnerabilities. https://www .zdnet.com/article/google-chrome-impactedby-new-magellan-2-0-vulnerabilities. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "comdb2 2022. Comdb2 GitHub. https://github .com/ bloomberg/comdb2. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "CVSS 2022. Common Vulnerability Scoring System version 3.1: User Guide. https://www .first.org/cvss/userguide. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. 2020. AFL++: Combining Incremental Steps of Fuzzing Research. In USENIX Workshop on Offensive Technologies (WOOT).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Flex 2022. Flex, the fast lexical analyzer generator. https: //github .com/westes/flex. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. In Conference on Automated Software Engi-neering (ASE’22).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2020. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB). Tokyo, Japan.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "George Klees, Andrew Ruef, Benji Cooper, Shiyi Wei, and Michael Hicks. 2018. Evaluating Fuzz Testing. InACM Conference on Computer and Communications Security (CCS).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jie Liang, Yu Jiang, Yuanliang Chen, Mingzhe Wang, Chijin Zhou, and Jiaguang Sun. 2018. PAFL: Extend Fuzzing Optimizations of Single Mode to Industrial Parallel Mode.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Jie Liang, Yu Jiang, Mingzhe Wang, Xun Jiao, Yuanliang Chen, Houbing Song, and Kim-Kwang Raymond Choo. 2019. Deepfuzzer: Accelerated deep greybox fuzzing. IEEE Transactions on Dependable and Secure Computing(2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Jie Liang, Mingzhe Wang, Chijin Zhou, Zhiyong Wu, Yu Jiang, Jianzhong Liu, Zhe Liu, and Jiaguang Sun. 2022. PATA: Fuzzing with Path Aware Taint Analysis. In. IEEE Computer Society, Los Alamitos, CA, USA. 154–170.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Eric Lo, Carsten Binnig, Donald Kossmann, M Tamer ¨Ozsu, and Wing-Kai Hon. 2010. A framework for testing DBMS features. The VLDB Journal 19, 2 (2010), 203–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Micha ¨el Marcozzi, Wim Vanhoof, and Jean-Luc Hainaut. 2012. Test input generation for database programs using relational constraints. In Proceedings of the Fifth International Workshop on Testing Database Systems. 1– 6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "MariaDB 2022. MariaDB. https://mariadb .org/. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "MariaDB 2022. SQL Statements and Structure. https: //mariadb .com/kb/en/sql-statements-structure. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. 2008. Generating targeted queries for database testing. InProceedings of the 2008 ACMSIGMOD international conference on Management of data. 499–510.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Bruce Momjian. 2001. PostgreSQL: introduction and concepts. V ol. 192. Addison-Wesley New York.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "MySQL. 2022. MySQL. https://www .mysql .com/. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Oracle 2022. Oracle Official Website. https:// www.oracle .com/cn/index .html. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Yevgeny Pats. [n. d.]. Why (Continuous) Fuzzing. https://about .gitlab .com/blog/2020/12/10/whycontinuous-fuzzing/. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "PostgreSQL 2022. PostgreSQL. https: //www .postgresql .org/. Accessed: November 29, 680 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "PostgreSQL SQL Commands 2022. SQL Commands. https://www .postgresql .org/docs/13/sql-commands .html. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger. 2022. Bugs found in Database Management Systems. https://www .manuelrigger .at/dbms-bugs. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 34,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https: //doi.org/10 .1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 35,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 36,
+      "text": "Alex Scotti, Mark Hannum, Michael Ponomarenko, Dorin Hogea, Akshat Sikarwar, Mohit Khullar, Adi Zaimi, James Leddy, Fabio Angius, Rivers Zhang, and Lingzhi Deng. 2016. Comdb2: Bloomberg’s Highly Available Relational Database System. Proc. VLDB Endow. 9, 13 (2016), 1377–1388. https://doi .org/10 .14778/ 3007263 .3007275",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https: //github .com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitriy Vyukov. 2012. AddressSanitizer: A Fast Address Sanity Checker. In 2012 USENIX Annual Technical Conference, Boston, MA, USA, June 13-15, 2012, Gernot Heiser and Wilson C. Hsieh (Eds.). USENIX Association, 309–318. https://www .usenix .org/conference/atc12/technicalsessions/presentation/serebryany",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, New York, USA. Morgan Kaufmann, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "SQL 2021. SQL — DDL, DQL, DML, DCL and TCL Commands. Retrieved April 1, 2023 from https://www .geeksforgeeks .org/sql-ddl-dqldml-dcl-tcl-commands/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "SQLsmith 2022. SQLsmith Description. https: //github .com/anse1/sqlsmith#description. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Hao Sun, Yuheng Shen, Cong Wang, Jianzhong Liu, YuJiang, Ting Chen, and Aiguo Cui. 2021. HEALER: Relation Learning Guided Kernel Fuzzing. In Proceedings of the ACMSIGOPS 28th Symposium on Operating Systems Principles. 344–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Jiajie Wang, Puhan Zhang, Lei Zhang, Haowen Zhu, and Xiaojun Ye. 2013. A model-based fuzzing approach for DBMS. In 2013 8th International Conference on Communications and Networking in China (CHINACOM). IEEE, 426–431.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Mingzhe Wang, Jie Liang, Chijin Zhou, Yuanliang Chen, Zhiyong Wu, and Yu Jiang. 2021. Industrial Oriented Evaluation of Fuzzing Techniques. In 2021 14th IEEE Conference on Software Testing, Verification and Validation (ICST). IEEE, 306–317.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Mingzhe Wang, Jie Liang, Chijin Zhou, Yu Jiang, Rui Wang, Chengnian Sun, and Jiaguang Sun. 2021. {RIFF}: Reduced Instruction Footprint for {Coverage-Guided } Fuzzing. In 2021 USENIX Annual Technical Conference (USENIX ATC 21). 147–159.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In 2021 IEEE/ACM 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP). IEEE, 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Michael Widenius, David Axmark, and Kaj Arno. 2002. MySQL reference manual: documentation from the source. ” O’Reilly Media, Inc.”.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "wikipedia. 2022. Abstract syntax tree. https:// en.wikipedia .org/wiki/Abstract syntax tree. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "wikipedia 2022. Databases. Retrieved April 1, 2023 from https://en .wikipedia .org/wiki/Database",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "wikipedia. 2022. SQL. https://en .wikipedia .org/wiki/ SQL. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. 2022. Unicorn: Detect Runtime Errors in TimeSeries Databases With Hybrid Input Synthesis. In Symposium on Software Testing and Analysis (ISSTA’22).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Insu Yun, Sangho Lee, Meng Xu, Yeongjin Jang, and Taesoo Kim. 2018. QSYM: A Practical Concolic Execution Engine Tailored for Hybrid Fuzzing. In USENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Michał Zalewski. 2022. american fuzzy lop. http: //lcamtuf .coredump .cx/afl/. Accessed: November 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020. 681",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 32,
+      "text": "Manuel Rigger. 2022. Bugs found in Database Management Systems. https://www .manuelrigger .at/dbms-bugs. Accessed: November 29, 2022.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 34,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https: //doi.org/10 .1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 35,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We evaluate LEGO on PostgreSQL, MySQL, MariaDB, and Comdb2 against SQLancer, SQLsmith, and SQUIRREL.",
+      "context_before": "cannot cover the full input space and test the corresponding logic consequently. In this paper, we propose LEGO, a fuzzer to generate SQL sequences with abundant types to improve DBMS fuzzing coverage. The key idea of sequence generation is type-affinity, which indicates the meaningful occurrence of SQL type pairs (e.g., INSERT and SELECT ). During each fuzzing iteration, LEGO first proactively explores SQL statements of different types and analyzes affinities with coverage feedback. Next, when a new affinity is discovered, LEGO synthesizes new SQL sequences containing the types progressively.",
+      "context_after": "The sequence-oriented fuzzing helps LEGO outperform other fuzzers on branch coverage by 44%–198%. More importantly, in the continuous fuzzing, LEGO has discovered 102 new vulnerabilities confirmed by the corresponding vendors, including 6 bugs in PostgreSQL, 21 bugs in MySQL, 42 bugs in MariaDB, and 33 bugs in Comdb2. Among them, 22 CVEs have been assigned due to their severe security influences. Index Terms —DBMS fuzzing, SQL Type Sequence I. INTRODUCTION Database management systems (DBMSs) are crucial for modern data-intensive systems [49]. Serving as the intermediary between the user and t",
+      "section": null,
+      "page": 1,
+      "char_offset": 1149,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[4, 10, 32, 35, 51, 54]",
+      "technique": "pqs",
+      "sentence": "Security vulnerabilities, especially memory bugs such as buffer overflow are particularly dangerous for DBMS because they might allow attackers to steal information, tamper data, crash systems, and bring heavy losses [4, 10, 32, 35, 51, 54].",
+      "context_before": "ly, in the continuous fuzzing, LEGO has discovered 102 new vulnerabilities confirmed by the corresponding vendors, including 6 bugs in PostgreSQL, 21 bugs in MySQL, 42 bugs in MariaDB, and 33 bugs in Comdb2. Among them, 22 CVEs have been assigned due to their severe security influences. Index Terms —DBMS fuzzing, SQL Type Sequence I. INTRODUCTION Database management systems (DBMSs) are crucial for modern data-intensive systems [49]. Serving as the intermediary between the user and the database, a DBMS offers the solution to optimize and manage the storage and retrieval of data [6, 23, 27, 30].",
+      "context_after": "Existing works have focused on logic, performance, and memory bugs in DBMSs. To test logic and performance bugs, many representative schemes utilize differential testing [16, 35, 39]. Recently, many fuzzing methods are applied to detect memory bugs of DBMSs, focusing on generating valid SQL queries [37, 46, 53, 54]. The abundance of SQL Type Sequences contained by generated test cases is crucial for fuzzing a DBMS. As Figure 1 shows, a test case (surrounded by the grey box) is an input for a DBMS, and it always consists of a sequence of SQL statements [24, 40, 50] (surrounded by the orange bo",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 1963,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[16, 35, 39]",
+      "technique": "pqs",
+      "sentence": "To test logic and performance bugs, many representative schemes utilize differential testing [16, 35, 39].",
+      "context_before": "Type Sequence I. INTRODUCTION Database management systems (DBMSs) are crucial for modern data-intensive systems [49]. Serving as the intermediary between the user and the database, a DBMS offers the solution to optimize and manage the storage and retrieval of data [6, 23, 27, 30]. Security vulnerabilities, especially memory bugs such as buffer overflow are particularly dangerous for DBMS because they might allow attackers to steal information, tamper data, crash systems, and bring heavy losses [4, 10, 32, 35, 51, 54]. Existing works have focused on logic, performance, and memory bugs in DBMSs.",
+      "context_after": "Recently, many fuzzing methods are applied to detect memory bugs of DBMSs, focusing on generating valid SQL queries [37, 46, 53, 54]. The abundance of SQL Type Sequences contained by generated test cases is crucial for fuzzing a DBMS. As Figure 1 shows, a test case (surrounded by the grey box) is an input for a DBMS, and it always consists of a sequence of SQL statements [24, 40, 50] (surrounded by the orange box). Yu Jiang is the corresponding author.Generally, SQL statements have hundreds of types. Type is the category divided by the same functionalities, like SELECT and INSERT. SQL Type Se",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2282,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[35,\n37]",
+      "technique": "pqs",
+      "sentence": "In general, fuzzers could be divided into generation-based [35, 37] and mutation-based [15, 51, 54].",
+      "context_before": "ORDER BY v1; 5:SELECTv2FROMt1WHEREv1=1;Mutation Areas StatementTest Case SQL Type SequenceMutated Seed Mutate… … 5:SELECTv2FROMt1ORDER BY v1; Fig. 1. Breakdown of inputs generation in mutation-based DBMS fuzzing. Executions are performed at test case level, where a test case consists of a sequence of statements. To generate mutated seeds from an original seed, fuzzers mutate the inner structure given a predefined SQL type. However, existing fuzzing works have put hard work into generating valid test cases since SQL is a highly-structured language, while neglecting to enrich SQL Type Sequences.",
+      "context_after": "Generation-based fuzzers generate test cases according to custom rules. Consequently, the number of states and relationships between sequences are limited by the rules. For example, SQLsmith mainly generates SELECT statements, so it will only explore limited state space and the relationships are limited to the same type. Mutation-based fuzzers select an test case (also called a seed) from an input pool, generate many new inputs with mutation, and save the input back to the pool if its execu668 2375-026X/23/$31.00 ©2023 IEEEDOI 10.1109/ICDE55515.2023.00057 .00 ©2023 IEEE | DOI: 10.1109/ICDE555",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 4486,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We evaluate LEGO on the latest version of PostgreSQL, MySQL, MariaDB, and Comdb2 against SQLancer, SQLsmith, and SQUIRREL.",
+      "context_before": "ype, and determines its significance by analyzing the coverage. If the change results in new code coverage, then an affinity will be recorded. Next, LEGO exploits the affinity by synthesizing new sequences it induces to further increase coverage. When a new affinity is discovered, LEGO permutes all SQL Type Sequences containing the affinity with a limited length. The sequences are then instantiated to executable test cases. With sequence-enriched test cases progressively synthesized from affinities discovered by proactive exploration, LEGO continuously explores the state space of target DBMSs.",
+      "context_after": "The sequence-oriented fuzzing helps LEGO cover 198%, 44%, and 120% more branches than SQLancer, SQLsmith, and SQUIRREL on average, respectively. More importantly, in the continuous fuzzing (i.e., constantly running fuzzing without stopping it until the code is modified [29]), LEGO finds 102 new vulnerabilities while others find 11 of them in total. The vulnerabilities include 6 bugs in PostgreSQL, 21 bugs in MySQL, 42 bugs in MariaDB, and 33 bugs in Comdb2. Among them, 22 bugs are confirmed as CVEs in the U.S. National Vulnerability Database. II. SQL TYPE SEQUENCE Basic concepts. Database Man",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 8487,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The sequence-oriented fuzzing helps LEGO cover 198%, 44%, and 120% more branches than SQLancer, SQLsmith, and SQUIRREL on average, respectively.",
+      "context_before": "ty will be recorded. Next, LEGO exploits the affinity by synthesizing new sequences it induces to further increase coverage. When a new affinity is discovered, LEGO permutes all SQL Type Sequences containing the affinity with a limited length. The sequences are then instantiated to executable test cases. With sequence-enriched test cases progressively synthesized from affinities discovered by proactive exploration, LEGO continuously explores the state space of target DBMSs.We evaluate LEGO on the latest version of PostgreSQL, MySQL, MariaDB, and Comdb2 against SQLancer, SQLsmith, and SQUIRREL.",
+      "context_after": "More importantly, in the continuous fuzzing (i.e., constantly running fuzzing without stopping it until the code is modified [29]), LEGO finds 102 new vulnerabilities while others find 11 of them in total. The vulnerabilities include 6 bugs in PostgreSQL, 21 bugs in MySQL, 42 bugs in MariaDB, and 33 bugs in Comdb2. Among them, 22 bugs are confirmed as CVEs in the U.S. National Vulnerability Database. II. SQL TYPE SEQUENCE Basic concepts. Database Management Systems (DBMSs) refer to the software used to manage the storage and retrieval data in databases [49]. Structured Query Language (SQL) is",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 8609,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLsmith and SQLancer) generate seeds based on custom rules.",
+      "context_before": "nd a seed fed to SQUIRREL that contains 945 SQL statements. It repeatedly calls hundreds of INSERT statements to store data. These statements have very similar behaviors, contributing little to coverage but increasing the workload to parse and execute. As a result, SQUIRREL hung for 23 minutes while executing this seed. Status of existing fuzzers. Existing fuzzers mainly focus on generating syntactically and semantically correct seeds, while neglecting to generate abundant SQL Type Sequences. Generation-based fuzzing and mutation-based fuzzing are two main types. Generation-based fuzzers (e.g.",
+      "context_after": "To meet challenges C1 and C2, a compromise solution is to manually add a large number of rules for generating sequences. However, the solution can be labor-intensive, while the abundance remains limited. For challenge C3, it is possible to simplify the rules to improve execution speed. But the simplification is likely to decrease coverage. Mutation-based fuzzers (e.g., SQUIRREL and RATEL ) mutate seeds by changing existing seeds, but most of them only change the structure or data in individual statements. Therefore, the sequence and the relationship the test case contained will not be changed",
+      "section": "II SQL TYPE SEQUENCE",
+      "page": 3,
+      "char_offset": 15836,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To encompass as many state-of-the-art DBMS fuzzers as possible, we compared LEGO to popular fuzzer SQUIRREL and SQLancer from the academy and SQLsmith from the industry.",
+      "context_before": "re sequence-oriented algorithms? A. Evaluation Setup Tested DBMSs and compared fuzzers. To evaluate the generality and efficiency of LEGO, we used the latest version of four open-source DBMSs for evaluation, namely PostgreSQL, MySQL, MariaDB, and Comdb2, which are widely used in industry and academic research. PostgreSQL [26, 30] is a object-relational DBMS with over 30 years of active development. MySQL [27, 47] is one of the most popular opensource DBMSs. MariaDB [5, 23] is a community-developed fork of MySQL. Comdb2 [11, 36] clusters RDBMS built on optimistic concurrency control techniques.",
+      "context_after": "Basic setup. We performed all experiments on a machine running 64-bit Ubuntu 20.04 with 128 cores (AMDEPYC 7742 Processor @ 2.25 GHz) and 488 GiB of main memory. All the DBMSs were instrumented with AddressSanitizer (ASAN) [38]. For each fuzzer, we used their default configurations, such as instrumentation methods and seed corpus. We 674 TABLE ILEGO DISCOVERED 102 NEWVULNERABILITIES (POSTGRE SQL: 6, MYSQL: 21, MARIA DB: 42, COMDB 2: 33) WHILE OTHERS FOUND 11IN TOTAL. [UAF: USE-AFTER-FREE,BOF: BUFFER OVERFLOW [HEAP (H), STACK (S)], AF: ASSERTION FAILURE ,SEGV: SEGMENTATION VIOLATION ,UAP: US",
+      "section": "A Evaluation Setup",
+      "page": 7,
+      "char_offset": 39380,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, SQLancer and SQLsmith did not find any bugs.",
+      "context_before": "L. We ran each DBMS with one fuzzer for 24 hours, which is a widely used time setup. Each fuzzer instance was run separately in a docker with one CPU core. To distinguish bugs, we first got them from unique crashes by comparing the call stack. To improve accuracy, we also further analyzed the bugs manually. B. DBMS Vulnerability Detection 1) Overall Results: The four tested DBMSs are widely used by users and well tested by engineers, making it difficult to find new bugs. Nevertheless, LEGO managed to detect 102 vulnerabilities in continuous fuzzing, while others found only 11 of them in total.",
+      "context_after": "SQUIRREL found 3 bugs in MySQL and 8 bugs in MariaDB, respectively. Table I shows LEGO discovered 6, 21, 42, and 33 bugs in PostgreSQL, MySQL, MariaDB, and Comdb2, respectively. Some of these vulnerabilities can be exploited in just a few steps and have serious repercussions. Specifically, among 102 vulnerabilities, there are 61 vulnerabilities (17 buffer overflows, 7 use after frees, 29 segmentation violations, and 8 use-after-poisons) that are very dangerous. They could be powerful attack primitives which lead to arbitrary code execution. Specifically, they can be exploited to attack the DB",
+      "section": "B DBMS Vulnerability Detection",
+      "page": 8,
+      "char_offset": 42661,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer generates test cases based on custom pattern rules mainly for SELECT statements, while only a limited number of SQL Type Sequences can be generated.",
+      "context_before": "h is rarely used by testers, and as a result, the bug hides for a long time. With the analyzed type-affinities, LEGO synthesizes abundant SQL Type Sequences containing the expected sequence. Based on the type sequence of corresponding synthesized seeds, LEGO mutates them into more seeds effectively and finally synthesizes the specific test case to trigger this bug. Other fuzzer are hard to compose the specific SQL Type Sequence thus they cannot find the bug. Specifically, SQLsmith mainly generates SELECT SQL statements, which would ignore the bugs composed of different types of SQL statements.",
+      "context_after": "SQUIRREL generates test cases mainly by changing the structure or data in one individual statement, so it is hard to generate new sequences of SQL types beyond the sequence contained in the existing seeds. Consequently, the bugs which have new SQL Type Sequences will be missed by them. C. Comparison with Other DBMS Fuzzers We evaluated fuzzers using two metrics, namely branches covered and bugs triggered. The two metrics are used as the standard in fuzzing evaluation [8, 17, 44], and have been widely used in fuzzing works [35, 42, 54]. To evaluate LEGO, we compared it against SQLancer, SQLsmi",
+      "section": "B DBMS Vulnerability Detection",
+      "page": 9,
+      "char_offset": 49146,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[35, 42, 54]",
+      "technique": "pqs",
+      "sentence": "The two metrics are used as the standard in fuzzing evaluation [8, 17, 44], and have been widely used in fuzzing works [35, 42, 54].",
+      "context_before": "ifferent types of SQL statements. SQLancer generates test cases based on custom pattern rules mainly for SELECT statements, while only a limited number of SQL Type Sequences can be generated. SQUIRREL generates test cases mainly by changing the structure or data in one individual statement, so it is hard to generate new sequences of SQL types beyond the sequence contained in the existing seeds. Consequently, the bugs which have new SQL Type Sequences will be missed by them. C. Comparison with Other DBMS Fuzzers We evaluated fuzzers using two metrics, namely branches covered and bugs triggered.",
+      "context_after": "To evaluate LEGO, we compared it against SQLancer, SQLsmith, and SQUIRREL. For a fair comparison, when we finished fuzzing, we collected the seeds generated by each fuzzer and rerun the input seeds to uniform the branch coverage. In addition, the bugs were distinguished and identified by comparing the call stack and manual analysis. PostgreSQL MySQL MariaDB Comdb2020000400006000080000100000120000140000 Lego squirrel sqlancer sqlsmith Fig. 9. Number of branches covered by LEGO, SQUIRREL, SQLancer, and SQLsmith on 4 DBMSs in 24 hours. Coverage. Figure 9 demonstrates the branches covered by thos",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 9,
+      "char_offset": 49713,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To evaluate LEGO, we compared it against SQLancer, SQLsmith, and SQUIRREL.",
+      "context_before": "y a limited number of SQL Type Sequences can be generated. SQUIRREL generates test cases mainly by changing the structure or data in one individual statement, so it is hard to generate new sequences of SQL types beyond the sequence contained in the existing seeds. Consequently, the bugs which have new SQL Type Sequences will be missed by them. C. Comparison with Other DBMS Fuzzers We evaluated fuzzers using two metrics, namely branches covered and bugs triggered. The two metrics are used as the standard in fuzzing evaluation [8, 17, 44], and have been widely used in fuzzing works [35, 42, 54].",
+      "context_after": "For a fair comparison, when we finished fuzzing, we collected the seeds generated by each fuzzer and rerun the input seeds to uniform the branch coverage. In addition, the bugs were distinguished and identified by comparing the call stack and manual analysis. PostgreSQL MySQL MariaDB Comdb2020000400006000080000100000120000140000 Lego squirrel sqlancer sqlsmith Fig. 9. Number of branches covered by LEGO, SQUIRREL, SQLancer, and SQLsmith on 4 DBMSs in 24 hours. Coverage. Figure 9 demonstrates the branches covered by those fuzzers over 24-hour fuzzing. It shows that LEGO performed better. Specif",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 9,
+      "char_offset": 49846,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "sqlancer",
+      "technique": null,
+      "sentence": "PostgreSQL MySQL MariaDB Comdb2020000400006000080000100000120000140000 Lego squirrel sqlancer sqlsmith Fig.",
+      "context_before": "d by them. C. Comparison with Other DBMS Fuzzers We evaluated fuzzers using two metrics, namely branches covered and bugs triggered. The two metrics are used as the standard in fuzzing evaluation [8, 17, 44], and have been widely used in fuzzing works [35, 42, 54]. To evaluate LEGO, we compared it against SQLancer, SQLsmith, and SQUIRREL. For a fair comparison, when we finished fuzzing, we collected the seeds generated by each fuzzer and rerun the input seeds to uniform the branch coverage. In addition, the bugs were distinguished and identified by comparing the call stack and manual analysis.",
+      "context_after": "9. Number of branches covered by LEGO, SQUIRREL, SQLancer, and SQLsmith on 4 DBMSs in 24 hours. Coverage. Figure 9 demonstrates the branches covered by those fuzzers over 24-hour fuzzing. It shows that LEGO performed better. Specifically, LEGO covered 198%, 44%, and 120% more branches than SQLancer, SQLsmith, and SQUIRREL on average, respectively. The fundamental reason for improving coverage is that DBMSs are rich in states, and the states are sensitive to the type and execution order of SQL statements. The same set of SQL statements can 676 trigger different code regions when executed wit",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 9,
+      "char_offset": 50181,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Number of branches covered by LEGO, SQUIRREL, SQLancer, and SQLsmith on 4 DBMSs in 24 hours.",
+      "context_before": "d and bugs triggered. The two metrics are used as the standard in fuzzing evaluation [8, 17, 44], and have been widely used in fuzzing works [35, 42, 54]. To evaluate LEGO, we compared it against SQLancer, SQLsmith, and SQUIRREL. For a fair comparison, when we finished fuzzing, we collected the seeds generated by each fuzzer and rerun the input seeds to uniform the branch coverage. In addition, the bugs were distinguished and identified by comparing the call stack and manual analysis. PostgreSQL MySQL MariaDB Comdb2020000400006000080000100000120000140000 Lego squirrel sqlancer sqlsmith Fig. 9.",
+      "context_after": "Coverage. Figure 9 demonstrates the branches covered by those fuzzers over 24-hour fuzzing. It shows that LEGO performed better. Specifically, LEGO covered 198%, 44%, and 120% more branches than SQLancer, SQLsmith, and SQUIRREL on average, respectively. The fundamental reason for improving coverage is that DBMSs are rich in states, and the states are sensitive to the type and execution order of SQL statements. The same set of SQL statements can 676 trigger different code regions when executed with different orders. Since LEGO focuses on generating abundant SQL Type Sequences, thus it is sen",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 9,
+      "char_offset": 50292,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, LEGO covered 198%, 44%, and 120% more branches than SQLancer, SQLsmith, and SQUIRREL on average, respectively.",
+      "context_before": "UIRREL. For a fair comparison, when we finished fuzzing, we collected the seeds generated by each fuzzer and rerun the input seeds to uniform the branch coverage. In addition, the bugs were distinguished and identified by comparing the call stack and manual analysis. PostgreSQL MySQL MariaDB Comdb2020000400006000080000100000120000140000 Lego squirrel sqlancer sqlsmith Fig. 9. Number of branches covered by LEGO, SQUIRREL, SQLancer, and SQLsmith on 4 DBMSs in 24 hours. Coverage. Figure 9 demonstrates the branches covered by those fuzzers over 24-hour fuzzing. It shows that LEGO performed better.",
+      "context_after": "The fundamental reason for improving coverage is that DBMSs are rich in states, and the states are sensitive to the type and execution order of SQL statements. The same set of SQL statements can 676 trigger different code regions when executed with different orders. Since LEGO focuses on generating abundant SQL Type Sequences, thus it is sensitive to the order of execution of statements and can cover more code regions as well as unlock bugs hidden in them. SQLancer and SQLsmith are two state-of-the-art DBMS fuzzers that generate test cases from rules. However, they could only generate limit",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 9,
+      "char_offset": 50514,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer and SQLsmith are two state-of-the-art DBMS fuzzers that generate test cases from rules.",
+      "context_before": "rmed better. Specifically, LEGO covered 198%, 44%, and 120% more branches than SQLancer, SQLsmith, and SQUIRREL on average, respectively. The fundamental reason for improving coverage is that DBMSs are rich in states, and the states are sensitive to the type and execution order of SQL statements. The same set of SQL statements can 676 trigger different code regions when executed with different orders. Since LEGO focuses on generating abundant SQL Type Sequences, thus it is sensitive to the order of execution of statements and can cover more code regions as well as unlock bugs hidden in them.",
+      "context_after": "However, they could only generate limited SQL Type Sequences. Specifically, SQLancer continuously generates test cases for fuzzing based on custom pattern rules, while only a limited number of SQL Type Sequences can be generated. Furthermore, SQLsmith mainly generates SELECT statements for PostgreSQL to ensure semantic correctness. SQUIRREL is an advanced mutation-based DBMS fuzzer. It uses coverage feedback to guide it in exploring the code regions of the target DBMS. Due to the coverage feedback and its efforts to improve syntactic and semantic correctness, SQUIRREL performs better than SQL",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 10,
+      "char_offset": 51102,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, SQLancer continuously generates test cases for fuzzing based on custom pattern rules, while only a limited number of SQL Type Sequences can be generated.",
+      "context_before": "n for improving coverage is that DBMSs are rich in states, and the states are sensitive to the type and execution order of SQL statements. The same set of SQL statements can 676 trigger different code regions when executed with different orders. Since LEGO focuses on generating abundant SQL Type Sequences, thus it is sensitive to the order of execution of statements and can cover more code regions as well as unlock bugs hidden in them. SQLancer and SQLsmith are two state-of-the-art DBMS fuzzers that generate test cases from rules. However, they could only generate limited SQL Type Sequences.",
+      "context_after": "Furthermore, SQLsmith mainly generates SELECT statements for PostgreSQL to ensure semantic correctness. SQUIRREL is an advanced mutation-based DBMS fuzzer. It uses coverage feedback to guide it in exploring the code regions of the target DBMS. Due to the coverage feedback and its efforts to improve syntactic and semantic correctness, SQUIRREL performs better than SQLancer in these DBMSs. However, it still suffers from the lack of the abundance of SQL Type Sequences. Starting from the initial seeds, SQUIRREL mainly modifies the structure and data inside single SQL statements. Therefore the see",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 10,
+      "char_offset": 51261,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Due to the coverage feedback and its efforts to improve syntactic and semantic correctness, SQUIRREL performs better than SQLancer in these DBMSs.",
+      "context_before": "s unlock bugs hidden in them. SQLancer and SQLsmith are two state-of-the-art DBMS fuzzers that generate test cases from rules. However, they could only generate limited SQL Type Sequences. Specifically, SQLancer continuously generates test cases for fuzzing based on custom pattern rules, while only a limited number of SQL Type Sequences can be generated. Furthermore, SQLsmith mainly generates SELECT statements for PostgreSQL to ensure semantic correctness. SQUIRREL is an advanced mutation-based DBMS fuzzer. It uses coverage feedback to guide it in exploring the code regions of the target DBMS.",
+      "context_after": "However, it still suffers from the lack of the abundance of SQL Type Sequences. Starting from the initial seeds, SQUIRREL mainly modifies the structure and data inside single SQL statements. Therefore the seeds it produces is hard to change the SQL Type Sequences of the initial seeds. Because of the limitation in sequences’ abundance, SQUIRREL may be not able to trigger some functionalities of target DBMSs. TABLE IINUMBER OFTYPE-AFFINITIES GENERATED BYDIFFERENT FUZZERS DBMS SQLancer SQUIRREL LEGO PostgreSQL 474 34 2101 MySQL 50 21 643 MariaDB 119 28 734 Comdb2 127 36 229 Total 770 119 3707 In",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 10,
+      "char_offset": 51673,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "TABLE IINUMBER OFTYPE-AFFINITIES GENERATED BYDIFFERENT FUZZERS DBMS SQLancer SQUIRREL LEGO PostgreSQL 474 34 2101 MySQL 50 21 643 MariaDB 119 28 734 Comdb2 127 36 229 Total 770 119 3707 Increment 2937 3588 – In contrast, LEGO is designed to increase the abundance of SQL Type Sequences.",
+      "context_before": "oring the code regions of the target DBMS. Due to the coverage feedback and its efforts to improve syntactic and semantic correctness, SQUIRREL performs better than SQLancer in these DBMSs. However, it still suffers from the lack of the abundance of SQL Type Sequences. Starting from the initial seeds, SQUIRREL mainly modifies the structure and data inside single SQL statements. Therefore the seeds it produces is hard to change the SQL Type Sequences of the initial seeds. Because of the limitation in sequences’ abundance, SQUIRREL may be not able to trigger some functionalities of target DBMSs.",
+      "context_after": "Type affinity describes the pattern of composing sequences, which reflects the abundance of SQL Type Sequences. Table II shows the type-affinities contained by seeds generated by different fuzzers in 24 hours. SQLsmith is excluded because it contains only one statement per test case. The table shows that LEGO found more type-affinities than other fuzzers, which allowed LEGO to produce more meaningful sequences to increase the abundance. Specifically, it proactively explores the type sequence space through sequence-oriented mutations and analyzes type-affinities in mutated seeds that have new",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 10,
+      "char_offset": 52231,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "It shows that LEGO found 52, 52, and 41 more bugs than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+      "context_before": "at LEGO found more type-affinities than other fuzzers, which allowed LEGO to produce more meaningful sequences to increase the abundance. Specifically, it proactively explores the type sequence space through sequence-oriented mutations and analyzes type-affinities in mutated seeds that have new coverage. It then uses these typeaffinities to synthesize meaningful sequences. The increase in sequence abundance assists LEGO to trigger more logic in the target DBMSs. Consequently, LEGO had better coverage than others in the target DBMS. Bugs. Table III shows the number of bugs found by each fuzzer.",
+      "context_after": "SQLancer focuses on detecting logic bugs in DBMSs, but the bug-finding process is limited by its predefined rules.TABLE IIINUMBER OFBUGS TRIGGERED IN 24HOURS DBMS SQLancer SQLsmith SQUIRREL LEGO PostgreSQL 0 0 0 2 MySQL 0 – 3 11 MariaDB 0 – 8 32 Comdb2 0 – 0 7 Total 0 0 11 52 Increment 52 52 41 – Consequently, it did not trigger bugs in the latest versions of these DBMSs. SQLsmith generates only limited types of statements, especially the SELECT type. It greatly ensures syntax correctness, however, the abundance of SQL Type Sequences is also limited. The evaluation results show it did not fin",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 10,
+      "char_offset": 53422,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer focuses on detecting logic bugs in DBMSs, but the bug-finding process is limited by its predefined rules.",
+      "context_before": "quences to increase the abundance. Specifically, it proactively explores the type sequence space through sequence-oriented mutations and analyzes type-affinities in mutated seeds that have new coverage. It then uses these typeaffinities to synthesize meaningful sequences. The increase in sequence abundance assists LEGO to trigger more logic in the target DBMSs. Consequently, LEGO had better coverage than others in the target DBMS. Bugs. Table III shows the number of bugs found by each fuzzer. It shows that LEGO found 52, 52, and 41 more bugs than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+      "context_after": "TABLE IIINUMBER OFBUGS TRIGGERED IN 24HOURS DBMS SQLancer SQLsmith SQUIRREL LEGO PostgreSQL 0 0 0 2 MySQL 0 – 3 11 MariaDB 0 – 8 32 Comdb2 0 – 0 7 Total 0 0 11 52 Increment 52 52 41 – Consequently, it did not trigger bugs in the latest versions of these DBMSs. SQLsmith generates only limited types of statements, especially the SELECT type. It greatly ensures syntax correctness, however, the abundance of SQL Type Sequences is also limited. The evaluation results show it did not find any bugs in the latest version of PostgreSQL. Based on the improved coverage, LEGO explores more state space of t",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 10,
+      "char_offset": 53525,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "TABLE IIINUMBER OFBUGS TRIGGERED IN 24HOURS DBMS SQLancer SQLsmith SQUIRREL LEGO PostgreSQL 0 0 0 2 MySQL 0 – 3 11 MariaDB 0 – 8 32 Comdb2 0 – 0 7 Total 0 0 11 52 Increment 52 52 41 – Consequently, it did not trigger bugs in the latest versions of these DBMSs.",
+      "context_before": "riented mutations and analyzes type-affinities in mutated seeds that have new coverage. It then uses these typeaffinities to synthesize meaningful sequences. The increase in sequence abundance assists LEGO to trigger more logic in the target DBMSs. Consequently, LEGO had better coverage than others in the target DBMS. Bugs. Table III shows the number of bugs found by each fuzzer. It shows that LEGO found 52, 52, and 41 more bugs than SQLancer, SQLsmith, and SQUIRREL, respectively. SQLancer focuses on detecting logic bugs in DBMSs, but the bug-finding process is limited by its predefined rules.",
+      "context_after": "SQLsmith generates only limited types of statements, especially the SELECT type. It greatly ensures syntax correctness, however, the abundance of SQL Type Sequences is also limited. The evaluation results show it did not find any bugs in the latest version of PostgreSQL. Based on the improved coverage, LEGO explores more state space of the target DBMSs. Exploring more states increases the likelihood of finding bugs. Besides, as the case study shows, many of the triggered bugs have unexpected SQL Type Sequences. LEGO proactively analyzes type-affinities of statements from meaningful test cases",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 10,
+      "char_offset": 53640,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Thus, LEGO found more bugs than SQLancer, SQLsmith, and SQUIRREL.",
+      "context_before": "results show it did not find any bugs in the latest version of PostgreSQL. Based on the improved coverage, LEGO explores more state space of the target DBMSs. Exploring more states increases the likelihood of finding bugs. Besides, as the case study shows, many of the triggered bugs have unexpected SQL Type Sequences. LEGO proactively analyzes type-affinities of statements from meaningful test cases. Based on type-affinities, LEGO progressively synthesizes abundant SQL Type Sequences. Moreover, LEGO lays the foundation for conventional mutations to use these sequences to mutate and find bugs.",
+      "context_after": "D. Effectiveness of Sequence-Oriented Algorithms in LEGO To measure the effectiveness of the sequence-oriented fuzzing algorithm and exclude other differences such as the extension in AST parser, we implement L EGOfor comparison, which disables the sequence-oriented algorithms including proactively affinity analysis and progressive sequence synthesis. Note that the affinity analysis provides the fundamental elements for sequence synthesis, the tightly-coupled nature requires us to disable them altogether. We compare LEGOagainst LEGO on PostgreSQL, MySQL, MariaDB, and Comdb2 for 24 hours. Tabl",
+      "section": "C Comparison with Other DBMS Fuzzers",
+      "page": 10,
+      "char_offset": 54697,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [35] synthesizes queries to fetch a random row from existing tables in the target DBMS.",
+      "context_before": "LEGO’s results, existing work needs to reimplement LEGO’s logic to increase abundance effectively. VII. RELATED WORK In this section, we will focus on some tasks related to DBMS fuzzing and highlight how they differ from LEGO. Finding logic and performance bugs in DBMSs. Schemes aimed at logical bugs focus on the correctness of DBMSs. Logic bugs would not crash the system but may cause a DBMS to return unexpected results, such as leaking extra rows. RAGS [39] uses differential testing, namely detecting logic bugs by running the same query on different DBMSs and checking the result consistency.",
+      "context_after": "If the DBMS fails to fetch that, then the DBMS might have a bug. Its following works [34, 33] also apply similar strategies by building functionally equivalent queries. Schemes aimed at performance bugs focus on the actual execution of DBMSs. Performance bugs can slow down an entire DBMS system or even bring it to a halt. APOLLO [16] generates queries to test two versions of the same DBMS. If the execution times for two versions are significantly different, then a performance bug is found. These fuzzers focus on finding differences between versions to locate bugs, rather than increasing the a",
+      "section": "D Effectiveness of Sequence-Oriented Algorithms in LEGO",
+      "page": 12,
+      "char_offset": 62767,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "citation_marker",
+      "surface": "[34, 33]",
+      "technique": "tlp",
+      "sentence": "Its following works [34, 33] also apply similar strategies by building functionally equivalent queries.",
+      "context_before": "related to DBMS fuzzing and highlight how they differ from LEGO. Finding logic and performance bugs in DBMSs. Schemes aimed at logical bugs focus on the correctness of DBMSs. Logic bugs would not crash the system but may cause a DBMS to return unexpected results, such as leaking extra rows. RAGS [39] uses differential testing, namely detecting logic bugs by running the same query on different DBMSs and checking the result consistency. SQLancer [35] synthesizes queries to fetch a random row from existing tables in the target DBMS. If the DBMS fails to fetch that, then the DBMS might have a bug.",
+      "context_after": "Schemes aimed at performance bugs focus on the actual execution of DBMSs. Performance bugs can slow down an entire DBMS system or even bring it to a halt. APOLLO [16] generates queries to test two versions of the same DBMS. If the execution times for two versions are significantly different, then a performance bug is found. These fuzzers focus on finding differences between versions to locate bugs, rather than increasing the abundance of SQL Type Sequences. LEGO differs from these works by aiming to find memory bugs by generating test cases containing abundant SQL Type Sequences. Compared to",
+      "section": "D Effectiveness of Sequence-Oriented Algorithms in LEGO",
+      "page": 12,
+      "char_offset": 62929,
+      "cited_reference": {
+        "number": 34,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https: //doi.org/10 .1145/3428279",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "citation_marker",
+      "surface": "[25, 35, 37, 43]",
+      "technique": "pqs",
+      "sentence": "Generation-based fuzzers [25, 35, 37, 43] have been used to test DBMSs for decades.",
+      "context_before": "ffers from these works by aiming to find memory bugs by generating test cases containing abundant SQL Type Sequences. Compared to logic and performance bugs, memory bugs happen more frequently. We did a cursory survey of the number of different type of bugs reported on MariaDB since 2009, and we found that the number of memory bugs is larger than the other two types combined. More importantly, memory bugs are even more damaging. Because they allow an attacker to leak or corrupt memory, the attacker can execute remote code or even gain control of the whole system. Generation-based DBMS fuzzing.",
+      "context_after": "They always generate enormous test cases based on custom rules, but the rules in turn also limit the SQL Type Sequences they can generate. Because generating a fully valid test case proves to be an NP-complete problem [21], generation-based fuzzers generally enhance semantic correctness while ensuring syntactic correctness. Some works treat generation as the process to satisfy constraints [1, 22] and use SAT solvers to generate potential queries [2]. SQLsmith [37] is one of the state-of-the-art generation-based DBMS fuzzers. It continuously generates syntactically correct SQL statements. But",
+      "section": "D Effectiveness of Sequence-Oriented Algorithms in LEGO",
+      "page": 12,
+      "char_offset": 64103,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M1",
+        "M5",
+        "M6",
+        "M8",
+        "M12",
+        "M15",
+        "M18",
+        "M20",
+        "M23"
+      ],
+      "describes_as_state_of_the_art": [
+        "M8",
+        "M16"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:14:29Z",
+    "is_model_written": true,
+    "summary": "Lego generates SQL sequences with a wide range of statement types, addressing the problem that existing DBMS fuzzers reuse a limited set of predefined types and so cannot cover the hundreds of statement types in the SQL specification. Its key notion is type-affinity, the meaningful co-occurrence of statement type pairs such as INSERT and SELECT: each iteration explores statements of different types, analyses affinities using coverage feedback, and synthesises new sequences as affinities are discovered. Lego found 102 new vendor-confirmed vulnerabilities across four DBMSs, 22 with CVEs.",
+    "narrative": "SQLancer is one of the three fuzzers Lego is measured against, described as a state-of-the-art academic fuzzer that generates from custom rules mainly for SELECT statements. Lego reports covering 198% more branches than it, and notes that SQLancer found no bugs in their setting.",
+    "roles": {
+      "M1": "baseline",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "baseline",
+      "M6": "result_comparison",
+      "M7": "motivation",
+      "M8": "state_of_the_art",
+      "M9": "result_comparison",
+      "M10": "motivation",
+      "M11": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer appears only as a fuzzer run for comparison."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Type-affinity sequence generation is Lego's own; no SQLancer technique is generalised."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M1",
+          "M5",
+          "M6",
+          "M9"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "We evaluate LEGO on PostgreSQL, MySQL, MariaDB, and Comdb2 against SQLancer, SQLsmith, and SQUIRREL.",
+            "section": null,
+            "page": 1
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "We evaluate LEGO on the latest version of PostgreSQL, MySQL, MariaDB, and Comdb2 against SQLancer, SQLsmith, and SQUIRREL.",
+            "section": "I INTRODUCTION",
+            "page": 2
+          },
+          {
+            "mention_id": "M6",
+            "sentence": "The sequence-oriented fuzzing helps LEGO cover 198%, 44%, and 120% more branches than SQLancer, SQLsmith, and SQUIRREL on average, respectively.",
+            "section": "I INTRODUCTION",
+            "page": 2
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "Specifically, SQLancer and SQLsmith did not find any bugs.",
+            "section": "B DBMS Vulnerability Detection",
+            "page": 8
+          }
+        ],
+        "reasoning": "Lego is evaluated against SQLancer on four DBMSs, reporting 198% more branches covered and that SQLancer found no bugs in that setting."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M8"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M8",
+            "sentence": "To encompass as many state-of-the-art DBMS fuzzers as possible, we compared LEGO to popular fuzzer SQUIRREL and SQLancer from the academy and SQLsmith from the industry.",
+            "section": "A Evaluation Setup",
+            "page": 7
+          }
+        ],
+        "reasoning": "M8 says the comparison was chosen to encompass as many state-of-the-art DBMS fuzzers as possible, naming SQLancer from the academic side."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icde60146_2024_00011.json b/_data/papers/paper_doi_10_1109_icde60146_2024_00011.json
new file mode 100644
index 0000000..1440474
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icde60146_2024_00011.json
@@ -0,0 +1,752 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:33:00Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icde60146.2024.00011",
+    "title": "TRAP: Tailored Robustness Assessment for Index Advisors via Adversarial Perturbation",
+    "authors": [
+      "Wei Zhou",
+      "Chen Lin",
+      "Xuanhe Zhou",
+      "Guoliang Li",
+      "Tianqing Wang"
+    ],
+    "year": 2024,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde60146.2024.00011",
+    "arxiv_id": null,
+    "s2_paper_id": "83899adebf94722befbec210a795d4b8d0db5b92",
+    "url": "https://doi.org/10.1109/icde60146.2024.00011",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icde60146.2024.00011",
+      "retrieved_at": "2026-09-08T14:33:00Z",
+      "chars": 81233,
+      "content_sha256": "sha256:f997ddb9f666612583c019fab527c37f3eea7273b614871e99edeed0f3b47630"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 14,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2104
+      },
+      {
+        "number": "II",
+        "title": "RELATED WORK",
+        "start": 10062
+      },
+      {
+        "number": "A",
+        "title": "Index Advisor",
+        "start": 10079
+      },
+      {
+        "number": "B",
+        "title": "SQL Generation",
+        "start": 11564
+      },
+      {
+        "number": "C",
+        "title": "Adversarial Attack",
+        "start": 12872
+      },
+      {
+        "number": "III",
+        "title": "PROBLEM DEFINITION",
+        "start": 13983
+      },
+      {
+        "number": "IV",
+        "title": "ADVERSARIAL WORKLOAD GENERATION",
+        "start": 23397
+      },
+      {
+        "number": "A",
+        "title": "Perturbation via Encoder-Decoder Network",
+        "start": 25661
+      },
+      {
+        "number": "B",
+        "title": "Reinforced Perturbation Policy Learning",
+        "start": 28963
+      },
+      {
+        "number": "C",
+        "title": "Index Advisor Independent Pretraining",
+        "start": 34109
+      },
+      {
+        "number": "D",
+        "title": "Constraint-Aware Reference Tree",
+        "start": 35286
+      },
+      {
+        "number": "V",
+        "title": "EXPERIMENTS",
+        "start": 40544
+      },
+      {
+        "number": "A",
+        "title": "Experimental Setup",
+        "start": 41031
+      },
+      {
+        "number": "B",
+        "title": "Robustness Assessment of Index Advisors",
+        "start": 47974
+      },
+      {
+        "number": "C",
+        "title": "Ablation Study and Impacts of Parameters",
+        "start": 52506
+      },
+      {
+        "number": "VI",
+        "title": "ANALYSIS ANDDISCOVERY",
+        "start": 57233
+      },
+      {
+        "number": "A",
+        "title": "Learning-based Index Advisors",
+        "start": 57599
+      },
+      {
+        "number": "B",
+        "title": "Heuristic-based Index Advisors",
+        "start": 59620
+      },
+      {
+        "number": "C",
+        "title": "Impact of Query Change",
+        "start": 61642
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 64443
+      },
+      {
+        "number": "M",
+        "title": "Radulovic, M. Stikic, G. Xu, and S. Chaudhuri, “Automatically",
+        "start": 67954
+      },
+      {
+        "number": "J",
+        "title": "Camacho-Rodr ´ıguez, B. Chundatt, A. Chung, N. Dutta, A. Fogarty,",
+        "start": 68196
+      },
+      {
+        "number": "K",
+        "title": "Muthyala, H. Nagulapalli, Y. Park, H. Patel, A. Pavlenko, O. Poppe,",
+        "start": 68410
+      },
+      {
+        "number": "S",
+        "title": "Ravindran, K. Saur, R. Sen, S. Suh, A. Tarafdar, K. Waghray,",
+        "start": 68481
+      },
+      {
+        "number": "M",
+        "title": "Alizadeh, “Flow-loss: Learning cardinality estimates that matter,”",
+        "start": 69149
+      },
+      {
+        "number": "J",
+        "title": "Patvarczki, R. Mutreja, M. Duller, F. M. Waas, and M. Winslett,",
+        "start": 69914
+      },
+      {
+        "number": "T",
+        "title": "Neumann, “How good are query optimizers, really?” Proc. VLDB",
+        "start": 70160
+      },
+      {
+        "number": "Y",
+        "title": "Qin, A. Pfadler, Z. Qian, J. Zhou, J. Li, and B. Cui, “Cardinality",
+        "start": 70703
+      },
+      {
+        "number": "G",
+        "title": "Fumera, G. Giacinto, and F. Roli, “Security evaluation of support",
+        "start": 73033
+      },
+      {
+        "number": "T",
+        "title": "G. Price, “Access path selection in a relational database management",
+        "start": 73773
+      },
+      {
+        "number": "O",
+        "title": "Papaemmanouil, and N. Tatbul, “Neo: A learned query optimizer,”",
+        "start": 74169
+      },
+      {
+        "number": "L",
+        "title": "Kaiser, and I. Polosukhin, “Attention is all you need,” in NIPS, 2017,",
+        "start": 74934
+      },
+      {
+        "number": "P",
+        "title": "Maes, “Pretrained encoders are all you need,” arXiv Preprint, vol.",
+        "start": 76009
+      },
+      {
+        "number": "E",
+        "title": "Agirre, I. Heintz, and D. Roth, “Recent advances in natural language",
+        "start": 77094
+      },
+      {
+        "number": "P",
+        "title": "Cistac, T. Rault, R. Louf, M. Funtowicz, J. Davison, S. Shleifer,",
+        "start": 77355
+      },
+      {
+        "number": "M",
+        "title": "Drame, Q. Lhoest, and A. M. Rush, “Transformers: State-of-the-art",
+        "start": 77496
+      },
+      {
+        "number": "B",
+        "title": "Qin, T. Liu, D. Jiang, and M. Zhou, “Codebert: A pre-trained model",
+        "start": 78125
+      },
+      {
+        "number": "M",
+        "title": "Marone, C. Akiki, J. Li, J. Chim, Q. Liu, E. Zheltonozhskii, T. Y.",
+        "start": 78385
+      },
+      {
+        "number": "M",
+        "title": "Kunakov, F. Zhdanov, M. Romero, T. Lee, N. Timor, J. Ding,",
+        "start": 78797
+      },
+      {
+        "number": "C",
+        "title": "Schlesinger, H. Schoelkopf, J. Ebert, T. Dao, M. Mishra, A. Gu,",
+        "start": 78859
+      },
+      {
+        "number": "J",
+        "title": "Robinson, C. J. Anderson, B. Dolan-Gavitt, D. Contractor, S. Reddy,",
+        "start": 78926
+      },
+      {
+        "number": "D",
+        "title": "Fried, D. Bahdanau, Y. Jernite, C. M. Ferrandis, S. Hughes, T. Wolf,",
+        "start": 78997
+      },
+      {
+        "number": "A",
+        "title": "Guha, L. von Werra, and H. de Vries, “Starcoder: may the source be",
+        "start": 79069
+      },
+      {
+        "number": "R",
+        "title": "Marinier, L. Hussenot, M. Geist, O. Pietquin, M. Michalski, S. Gelly,",
+        "start": 79774
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "R. M. Perera, B. Oetomo, B. I. P. Rubinstein, and R. BorovicaGajic, “DBA bandits: Self-driving index tuning under ad-hoc, analytical workloads with safety guarantees,” in ICDE, 2021, pp. 600–611.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "J. Kossmann, S. Halfpap, M. Jankrift, and R. Schlosser, “Magic mirror in my hand, which is the best in the land? an experimental evaluation of index selection algorithms,” Proc. VLDB Endow., vol. 13, no. 11, pp. 2382–2395, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "R. Schlosser, J. Kossmann, and M. Boissier, “Efficient scalable multiattribute index selection using recursive strategies,” in ICDE, 2019, pp. 1238–1249.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "G. Valentin, M. Zuliani, D. C. Zilio, G. M. Lohman, and A. Skelley, “DB2 advisor: An optimizer smart enough to recommend its own indexes,” in ICDE, 2000, pp. 101–110.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "S. Chaudhuri and V. R. Narasayya, “An efficient cost-driven index selection tool for microsoft SQL server,” in VLDB, 1997, pp. 146–155.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "K. Whang, Index Selection in Relational Databases. Springer, 1987.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "J. Kossmann, A. Kastius, and R. Schlosser, “SWIRL: selection of workload-aware indexes using reinforcement learning,” in EDBT, 2022, pp. 2:155–2:168.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "H. Lan, Z. Bao, and Y. Peng, “An index advisor using deep reinforcement learning,” in CIKM, 2020, pp. 2105–2108.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Z. Sadri, L. Gruenwald, and E. Leal, “Drlindex: deep reinforcement learning index advisor for a cluster database,” in IDEAS, 2020, pp. 11:1– 11:8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "X. Zhou, L. Liu, W. Li, L. Jin, S. Li, T. Wang, and J. Feng, “Autoindex: An incremental index management system for dynamic workloads,” in ICDE, 2022, pp. 2196–2208.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "S. Chaudhuri and V. Narasayya, “Anytime algorithm of database tuning advisor for microsoft sql server.” https://www.microsoft.com/en-us/ research/publication/, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "A. Kane, “Dexterthe automatic indexer for postgres.” https://github. com/ankane/dexter, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Z. Sadri, L. Gruenwald, and E. Leal, “Online index selection using deep reinforcement learning for a cluster database,” in ICDE Workshops, 2020, pp. 158–161.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "R. M. Perera, B. Oetomo, B. I. P. Rubinstein, and R. BorovicaGajic, “HMAB: self-driving hierarchy of bandits for integrated physical database design tuning,” Proc. VLDB Endow., vol. 16, no. 2, pp. 216– 229, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "S. Das, M. Grbic, I. Ilic, I. Jovandic, A. Jovanovic, V. R. Narasayya, M. Radulovic, M. Stikic, G. Xu, and S. Chaudhuri, “Automatically indexing millions of databases in microsoft azure SQL database,” in SIGMOD Conference, 2019, pp. 666–679.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Y. Zhu, Y. Tian, J. Cahoon, S. Krishnan, A. Agarwal, R. Alotaibi, J. Camacho-Rodr ´ıguez, B. Chundatt, A. Chung, N. Dutta, A. Fogarty, A. Gruenheid, B. Haynes, M. Interlandi, M. Iyer, N. Jurgens, S. Khushalani, B. Kroth, M. Kumar, J. Leeka, S. Matusevych, M. Mittal, A. M ¨uller, K. Muthyala, H. Nagulapalli, Y. Park, H. Patel, A. Pavlenko, O. Poppe, S. Ravindran, K. Saur, R. Sen, S. Suh, A. Tarafda",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "P. Negi, Z. Wu, A. Kipf, N. Tatbul, R. Marcus, S. Madden, T. Kraska, and M. Alizadeh, “Robust query driven cardinality estimation under changing workloads,” Proc. VLDB Endow., vol. 16, no. 6, pp. 1520– 1533, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "B. Li, Y. Lu, and S. Kandula, “Warper: Efficiently adapting learned cardinality estimators to data and workload drifts,” in SIGMOD Conference, 2022, pp. 1920–1933.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "P. Negi, R. C. Marcus, A. Kipf, H. Mao, N. Tatbul, T. Kraska, and M. Alizadeh, “Flow-loss: Learning cardinality estimates that matter,” Proc. VLDB Endow., vol. 14, no. 11, pp. 2019–2032, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "K. Vaidya, A. Dutt, V. R. Narasayya, and S. Chaudhuri, “Leveraging query logs and machine learning for parametric query optimization,” Proc. VLDB Endow., vol. 15, no. 3, pp. 401–413, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "J. Zhou, N. Bruno, M. Wu, P. Larson, R. Chaiken, and D. Shakib, “SCOPE: parallel databases meet mapreduce,” VLDB J., vol. 21, no. 5, pp. 611–636, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "J. Wang, T. Li, A. Wang, X. Liu, L. Chen, J. Chen, J. Liu, J. Wu, F. Li, and Y. Gao, “Real-time workload pattern analysis for large-scale cloud databases,” arXiv Preprint, vol. https://arxiv.org/abs/2307.02626, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "A. Aleyasen, M. Morcos, L. Antova, M. Sugiyama, D. Korablev, J. Patvarczki, R. Mutreja, M. Duller, F. M. Waas, and M. Winslett, “Intelligent automated workload analysis for database replatforming,” in SIGMOD Conference, 2022, pp. 2273–2285.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "V. Leis, A. Gubichev, A. Mirchev, P. A. Boncz, A. Kemper, and T. Neumann, “How good are query optimizers, really?” Proc. VLDB Endow., vol. 9, no. 3, pp. 204–215, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "A. Kipf, T. Kipf, B. Radke, V. Leis, P. A. Boncz, and A. Kemper, “Learned cardinalities: Estimating correlated joins with deep learning,” in CIDR, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "B. Ding, S. Chaudhuri, J. Gehrke, and V. R. Narasayya, “DSB: A decision support benchmark for workload-driven and traditional database systems,” Proc. VLDB Endow., vol. 14, no. 13, pp. 3376–3388, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Y. Han, Z. Wu, P. Wu, R. Zhu, J. Yang, L. W. Tan, K. Zeng, G. Cong, Y. Qin, A. Pfadler, Z. Qian, J. Zhou, J. Li, and B. Cui, “Cardinality estimation in DBMS: A comprehensive benchmark evaluation,” Proc. VLDB Endow., vol. 15, no. 4, pp. 752–765, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "M. Kurmanji and P. Triantafillou, “Detect, distill and update: Learned DB systems facing out of distribution data,” 1, no. 1, pp. 33:1–33:27, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "N. Bruno and S. Chaudhuri, “Automatic physical database tuning: A relaxation-based approach,” in SIGMOD Conference, 2005, pp. 227– 238.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "R. S. Sutton and A. G. Barto, Reinforcement learningan introduction, ser. Adaptive computation and machine learning. MIT Press, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Y. LeCun, Y. Bengio, and G. E. Hinton, “Deep learning,” Nat., vol. 521, no. 7553, pp. 436–444, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "A. Sharma, F. M. Schuhknecht, and J. Dittrich, “The case for automatic database administration using deep reinforcement learning,” arXiv Preprint, 2018. [Online]. Available: https://arxiv.org/abs/1801.05643",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "R. Schlosser and S. Halfpap, “A decomposition approach for risk-averse index selection,” in SSDBM, 2020, pp. 16:1–16:4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "F. Halim, S. Idreos, P. Karras, and R. H. C. Yap, “Stochastic database cracking: Towards robust adaptive indexing in main-memory columnstores,” Proc. VLDB Endow., vol. 5, no. 6, pp. 502–513, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "D. R. Slutz, “Massive stochastic testing of SQL,” in VLDB, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 211:1–211:30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "X. Liu, X. Kong, L. Liu, and K. Chiang, “Treegan: Syntax-aware sequence generation with generative adversarial networks,” in ICDM, 2018, pp. 1140–1145.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "L. Zhang, C. Chai, X. Zhou, and G. Li, “Learnedsqlgen: Constraintaware SQL generation using reinforcement learning,” in SIGMOD Conference, 2022, pp. 945–958.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Q. Zhou, J. Arulraj, S. B. Navathe, W. Harris, and J. Wu, “SIA: optimizing queries using learned predicates,” in SIGMOD Conference, 2021, pp. 2169–2181.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "X. Zhou, G. Li, C. Chai, and J. Feng, “A learned query rewrite system using monte carlo tree search,” Proc. VLDB Endow., vol. 15, no. 1, pp. 46–58, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "I. J. Goodfellow, J. Shlens, and C. Szegedy, “Explaining and harnessing adversarial examples,” in ICLR (Poster), 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "B. Biggio, I. Corona, B. Nelson, B. I. P. Rubinstein, D. Maiorca, G. Fumera, G. Giacinto, and F. Roli, “Security evaluation of support vector machines in adversarial environments,” arXiv Preprint, 2014. [Online]. Available: https://arxiv.org/pdf/1401.7727.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "E. M. Kornaropoulos, S. Ren, and R. Tamassia, “The price of tailoring the index to your data: Poisoning attacks on learned index structures,” in SIGMOD Conference, 2022, pp. 1331–1344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "B. Ding, S. Das, R. Marcus, W. Wu, S. Chaudhuri, and V. R. Narasayya, “AI meets AI: leveraging query executions to improve index recommendations,” in SIGMOD Conference, 2019, pp. 1241–1258.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "D. D. Chamberlin, A Complete Guide to DB2 Universal Database. Morgan Kaufmann, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "P. G. Selinger, M. M. Astrahan, D. D. Chamberlin, R. A. Lorie, and T. G. Price, “Access path selection in a relational database management system,” in SIGMOD Conference, 1979, pp. 23–34.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "P. E. O’Neil, E. J. O’Neil, X. Chen, and S. Revilak, “The star schema benchmark and augmented fact table indexing,” in TPCTC, ser. Lecture Notes in Computer Science, vol. 5895, 2009, pp. 237–252. 54",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "R. C. Marcus, P. Negi, H. Mao, C. Zhang, M. Alizadeh, T. Kraska, O. Papaemmanouil, and N. Tatbul, “Neo: A learned query optimizer,” Proc. VLDB Endow., vol. 12, no. 11, pp. 1705–1718, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "R. Marcus, P. Negi, H. Mao, N. Tatbul, M. Alizadeh, and T. Kraska, “Bao: Making learned query optimization practical,” in SIGMOD Conference, 2021, pp. 1275–1288.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Z. Yang, A. Kamsetty, S. Luan, E. Liang, Y. Duan, X. Chen, and I. Stoica, “Neurocard: One cardinality estimator for all tables,” Proc. VLDB Endow., vol. 14, no. 1, pp. 61–73, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "J. Chung, C ¸. G ¨ulc¸ehre, K. Cho, and Y. Bengio, “Empirical evaluation of gated recurrent neural networks on sequence modeling,” arXiv Preprint, 2014. [Online]. Available: https://arxiv.org/abs/1412.3555",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin, “Attention is all you need,” in NIPS, 2017, pp. 5998–6008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "I. Sutskever, O. Vinyals, and Q. V. Le, “Sequence to sequence learning with neural networks,” in NIPS, 2014, pp. 3104–3112.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "T. Luong, H. Pham, and C. D. Manning, “Effective approaches to attention-based neural machine translation,” in EMNLP, 2015, pp. 1412– 1421.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "G. Ke, Q. Meng, T. Finley, T. Wang, W. Chen, W. Ma, Q. Ye, and T. Liu, “Lightgbm: A highly efficient gradient boosting decision tree,” in NIPS, 2017, pp. 3146–3154.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "T. Siddiqui, W. Wu, V. R. Narasayya, and S. Chaudhuri, “DISTILL: low-overhead data-driven techniques for filtering and costing indexes for scalable index tuning,” Proc. VLDB Endow., vol. 15, no. 10, pp. 2019–2031, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "S. J. Rennie, E. Marcheret, Y. Mroueh, J. Ross, and V. Goel, “Selfcritical sequence training for image captioning,” in CVPR, 2017, pp. 1179–1195.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Y. Yu, “Towards sample efficient reinforcement learning,” in IJCAI, 2018, pp. 5739–5743.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "M. Khan, P. Srivatsa, A. Rane, S. Chenniappa, R. Anand, S. Ozair, and P. Maes, “Pretrained encoders are all you need,” arXiv Preprint, vol. https://arxiv.org/abs/2106.05139, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "H. Kim, B. So, W. Han, and H. Lee, “Natural language to SQL: where are we today?” Proc. VLDB Endow., vol. 13, no. 10, pp. 1737–1750, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "F. L. Deremer, “Generating parsers for BNF grammars,” in AFIPS Spring Joint Computing Conference, ser. AFIPS Conference Proceedings, vol. 34, 1969, pp. 793–799.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "W. Wu, C. Wang, T. Siddiqui, J. Wang, V. R. Narasayya, S. Chaudhuri, and P. A. Bernstein, “Budget-aware index tuning with reinforcement learning,” in SIGMOD Conference, 2022, pp. 1528–1541.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "A. Dutt, C. Wang, A. Nazi, S. Kandula, V. R. Narasayya, and S. Chaudhuri, “Selectivity estimation for range predicates using lightweight models,” Proc. VLDB Endow., vol. 12, no. 9, pp. 1044–1057, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "K. Schnaitter, N. Polyzotis, and L. Getoor, “Index interactions in physical design tuning: Modeling, analysis, and applications,” Proc. VLDB Endow., vol. 2, no. 1, pp. 1234–1245, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "B. Min, H. Ross, E. Sulem, A. P. B. Veyseh, T. H. Nguyen, O. Sainz, E. Agirre, I. Heintz, and D. Roth, “Recent advances in natural language processing via large pre-trained language models: A survey,” arXiv Preprint, vol. https://arxiv.org/abs/2111.01243, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "T. Wolf, L. Debut, V. Sanh, J. Chaumond, C. Delangue, A. Moi, P. Cistac, T. Rault, R. Louf, M. Funtowicz, J. Davison, S. Shleifer, P. von Platen, C. Ma, Y. Jernite, J. Plu, C. Xu, T. L. Scao, S. Gugger, M. Drame, Q. Lhoest, and A. M. Rush, “Transformers: State-of-the-art natural language processing,” in EMNLP (Demos), 2020, pp. 38–45.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "J. Devlin, M. Chang, K. Lee, and K. Toutanova, “BERT: pre-training of deep bidirectional transformers for language understanding,” in NAACLHLT, vol. 1, 2019, pp. 4171–4186.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "M. Lewis, Y. Liu, N. Goyal, M. Ghazvininejad, A. Mohamed, O. Levy, V. Stoyanov, and L. Zettlemoyer, “BART: denoising sequence-tosequence pre-training for natural language generation, translation, and comprehension,” in ACL, 2020, pp. 7871–7880.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Z. Feng, D. Guo, D. Tang, N. Duan, X. Feng, M. Gong, L. Shou, B. Qin, T. Liu, D. Jiang, and M. Zhou, “Codebert: A pre-trained model for programming and natural languages,” in EMNLP (Findings), ser. Findings of ACL, vol. EMNLP 2020, 2020, pp. 1536–1547.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "R. Li, L. B. Allal, Y. Zi, N. Muennighoff, D. Kocetkov, C. Mou, M. Marone, C. Akiki, J. Li, J. Chim, Q. Liu, E. Zheltonozhskii, T. Y. Zhuo, T. Wang, O. Dehaene, M. Davaadorj, J. Lamy-Poirier, J. Monteiro, O. Shliazhko, N. Gontier, N. Meade, A. Zebaze, M. Yee, L. K. Umapathi,J. Zhu, B. Lipkin, M. Oblokulov, Z. Wang, R. M. V, J. Stillerman, S. S. Patel, D. Abulkhanov, M. Zocca, M. Dey, Z. Zhang, N. ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "E. Parisotto, H. F. Song, J. W. Rae, R. Pascanu, C ¸. G ¨ulc¸ehre, S. M. Jayakumar, M. Jaderberg, R. L. Kaufman, A. Clark, S. Noury, M. M. Botvinick, N. Heess, and R. Hadsell, “Stabilizing transformers for reinforcement learning,” in ICML, ser. Proceedings of Machine Learning Research, vol. 119, 2020, pp. 7487–7498.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "W. Li, H. Luo, Z. Lin, C. Zhang, Z. Lu, and D. Ye, “A survey on transformers in reinforcement learning,” arXiv Preprint, vol. https://arxiv.org/abs/2301.03044, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "M. Andrychowicz, A. Raichuk, P. Stanczyk, M. Orsini, S. Girgin, R. Marinier, L. Hussenot, M. Geist, O. Pietquin, M. Michalski, S. Gelly, and O. Bachem, “What matters for on-policy deep actor-critic methods? A large-scale study,” in ICLR, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "X. Guo and H. Yu, “On the domain adaptation and generalization of pretrained language models: A survey,” arXiv Preprint, vol. https://arxiv.org/abs/2211.03154, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "S. Huang and S. Onta ˜n´on, “A closer look at invalid action masking in policy gradient algorithms,” arXiv Preprint, 2020. [Online]. Available: https://arxiv.org/abs/2006.14171",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Y. Luo, J. Peng, and J. Ma, “When causal inference meets deep learning,” Nat. Mach. Intell., vol. 2, no. 8, pp. 426–427, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "D. Kalainathan, O. Goudet, and R. Dutta, “Causal discovery toolbox: Uncovering causal relationships in python,” J. Mach. Learn. Res., vol. 21, pp. 37:1–37:5, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "L. Van der Maaten and G. Hinton, “Visualizing data using t-sne.” Journal of machine learning research, vol. 9, no. 11, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "B. Sch ¨olkopf, J. C. Platt, J. Shawe-Taylor, A. J. Smola, and R. C. Williamson, “Estimating the support of a high-dimensional distribution,” Neural Comput., vol. 13, no. 7, pp. 1443–1471, 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "F. T. Liu, K. M. Ting, and Z. Zhou, “Isolation-based anomaly detection,” ACM Trans. Knowl. Discov. Data, vol. 6, no. 1, pp. 3:1–3:39, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "M. M. Breunig, H. Kriegel, R. T. Ng, and J. Sander, “LOF: identifying density-based local outliers,” in SIGMOD Conference, 2000, pp. 93–104. 55",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 36,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 211:1–211:30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [36] derives multiple queries by partitioning the results from the original query.",
+      "context_before": "ferent strategies to make their methods more robust [33], [34] or considered dynamic workloads [1], [7], the assessments are conducted 43 on the testing workloads split from a predefined set of workloads, which only contain limited query variants and fails to reflect typical workload drifts. Therefore, the robustness of both heuristic-based and learning-based index advisors over workload drifts has never been thoroughly assessed. B. SQL Generation Heuristic-based SQL generation methods use various rules. For example, SQLsmith [35] randomly synthesizes queries by walking through the parse tree.",
+      "context_after": "Recently, learning-based SQL generation methods adopt Generative Adversarial Network (GAN) [37] or the reinforcement learning framework [38]. Instead of generating queries from scratch, some studies perform a series of rewrite transformations to optimize the original query. Sia [39] replaces the predicates in a query with valid but weaker ones learned by a classifier over the columns. LearnedRewrite [40] adopts the Monte Carlo Tree Search (MCTS) algorithm to find a nearoptimal rewrite order from a set of query rewrite rules. Difference with our work. First, our work aims to generate perturbed q",
+      "section": "B SQL Generation",
+      "page": 3,
+      "char_offset": 11730,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "TRAP assesses how robust index advisors are to change, arguing these tools are evaluated mainly on static scenarios while their stability under minor workload shifts is not studied. The authors introduce perturbation-based workloads with three constraints drawn from real scenarios, formulate generating perturbed queries as a sequence-to-sequence problem, and train TRAP in an opaque-box setting. Assessing ten index advisors, they find all vulnerable to TRAP's workloads.",
+    "narrative": "One citation, describing TLP as a way of deriving multiple queries by partitioning the results of an original query, in a discussion of SQL generation.",
+    "roles": {
+      "M1": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icde60146_2024_00441.json b/_data/papers/paper_doi_10_1109_icde60146_2024_00441.json
new file mode 100644
index 0000000..93ee510
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icde60146_2024_00441.json
@@ -0,0 +1,743 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:54:34Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icde60146.2024.00441",
+    "title": "Applications and Challenges for Large Language Models: From Data Management Perspective",
+    "authors": [
+      "Meihui Zhang",
+      "Zhaoxuan Ji",
+      "Zhaojing Luo",
+      "Yuncheng Wu",
+      "Chengliang Chai"
+    ],
+    "year": 2024,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde60146.2024.00441",
+    "arxiv_id": null,
+    "s2_paper_id": "35dd88676e19a2203ff86e5837f8537e431d5d6b",
+    "url": "https://doi.org/10.1109/icde60146.2024.00441",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icde60146.2024.00441",
+      "retrieved_at": "2026-09-08T14:54:34Z",
+      "chars": 76972,
+      "content_sha256": "sha256:eb5f03191cc0c455afbfff2806aafc1cc3b21e2a5157ea1e2f9288fe2f661062"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1226
+      },
+      {
+        "number": "II",
+        "title": "APPLICA TIONS OFLLM SI NDATA MANAGEMENT",
+        "start": 9033
+      },
+      {
+        "number": "E",
+        "title": "Further Discussion",
+        "start": 28969
+      },
+      {
+        "number": "III",
+        "title": "CHALLENGES ANDOPPORTUNITIES",
+        "start": 30111
+      },
+      {
+        "number": "A",
+        "title": "LLM Prompt Optimization",
+        "start": 31927
+      },
+      {
+        "number": "B",
+        "title": "LLM Query Optimization",
+        "start": 33918
+      },
+      {
+        "number": "C",
+        "title": "LLM Cache Optimization",
+        "start": 45666
+      },
+      {
+        "number": "D",
+        "title": "LLM Security and Privacy Concerns",
+        "start": 49899
+      },
+      {
+        "number": "E",
+        "title": "LLM Output V alidation",
+        "start": 53695
+      },
+      {
+        "number": "IV",
+        "title": "CONCLUSIONS",
+        "start": 56061
+      },
+      {
+        "number": "Z",
+        "title": "Luo, A. K. Tung, Y. Wang et al., “Singa: A distributed deep learning",
+        "start": 57720
+      },
+      {
+        "number": "T",
+        "title": "Song, Y. Xia et al., “Low-code llm: Visual programming over llms,”",
+        "start": 58549
+      },
+      {
+        "number": "L",
+        "title": "He et al., “A comprehensive survey on pretrained foundation models:",
+        "start": 58911
+      },
+      {
+        "number": "D",
+        "title": "Zhou et al., “Chain-of-thought prompting elicits reasoning in large",
+        "start": 59260
+      },
+      {
+        "number": "K",
+        "title": "Narasimhan, “Tree of thoughts: Deliberate problem solving with large",
+        "start": 59504
+      },
+      {
+        "number": "P",
+        "title": "Shenoy, “Rolex: relational on-line exchange with xml,” in Proceedings",
+        "start": 64812
+      },
+      {
+        "number": "Q",
+        "title": "Yao, S. Roman et al., “Spider: A large-scale human-labeled dataset",
+        "start": 71413
+      },
+      {
+        "number": "D",
+        "title": "Lie, M. Mannan, M. Backes, and X. Wang, Eds., 2018, pp. 178–195.",
+        "start": 74272
+      },
+      {
+        "number": "S",
+        "title": "Halevi, Eds., 2016, pp. 308–318.",
+        "start": 75923
+      },
+      {
+        "number": "H",
+        "title": "B. McMahan, J. Rosenstock, and Y. Zhang, “Federated learning",
+        "start": 76026
+      },
+      {
+        "number": "C",
+        "title": "Zhang, S. Agarwal, K. Slama, A. Ray et al., “Training language",
+        "start": 76459
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "C. Mohan, “Big data: Hype and reality,” Datenbanksysteme f ¨ur Business, Technologie und Web, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "B. C. Ooi, K.-L. Tan, S. Wang, W. Wang, Q. Cai, G. Chen, J. Gao, Z. Luo, A. K. Tung, Y. Wang et al., “Singa: A distributed deep learning platform,” in Proceedings of the 23rd ACM international conference on Multimedia, 2015, pp. 685–688.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Z. Luo, S. Cai, J. Gao, M. Zhang, K. Y. Ngiam, G. Chen, and W.-C. Lee, “Adaptive lightweight regularization tool for complex analytics,” in International Conference on Data Engineering, 2018, pp. 485–496.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Z. Luo, S. Cai, Y. Wang, and B. C. Ooi, “Regularized pairwise relationship based analytics for structured data,” Proceedings of the ACM on Management of Data, vol. 1, no. 1, pp. 1–27, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "M. Ghazvininejad, H. Gonen, and L. Zettlemoyer, “Dictionary-based phrase-level prompting of large language models for machine transla-tion,” arXiv preprint arXiv:2302.07856, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Y. Cai, S. Mao, W. Wu, Z. Wang, Y. Liang, T. Ge, C. Wu, W. Y ou, T. Song, Y. Xia et al., “Low-code llm: Visual programming over llms,” arXiv preprint arXiv:2304.08103, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "K. Guu, K. Lee, Z. Tung, P. Pasupat, and M. Chang, “Retrieval augmented language model pre-training,” in International Conference on Machine Learning, 2020, pp. 3929–3938.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "C. Zhou, Q. Li, C. Li, J. Y u, Y. Liu, G. Wang, K. Zhang, C. Ji, Q. Yan, L. He et al., “A comprehensive survey on pretrained foundation models: A history from bert to chatgpt,” arXiv preprint arXiv:2302.09419, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Y. Lu, S. Chaudhuri, C. Jermaine, and D. Melski, “Data-driven program completion,” arXiv preprint arXiv:1705.09042, 2017. 5539",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "J. Wei, X. Wang, D. Schuurmans, M. Bosma, F. Xia, E. Chi, Q. V. Le, D. Zhou et al., “Chain-of-thought prompting elicits reasoning in large language models,” Advances in Neural Information Processing Systems, vol. 35, pp. 24 824–24 837, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "S. Yao, D. Y u, J. Zhao, I. Shafran, T. L. Griffiths, Y. Cao, and K. Narasimhan, “Tree of thoughts: Deliberate problem solving with large language models,” arXiv preprint arXiv:2305.10601, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "J. Mohoney, A. Pacaci, S. R. Chowdhury, A. Mousavi, I. F. Ilyas, U. F. Minhas, J. Pound, and T. Rekatsinas, “High-throughput vector similarity search in knowledge graphs,” Proceedings of the ACM on Management of Data, vol. 1, no. 2, pp. 1–25, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Y. Li, J. Li, Y. Suhara, J. Wang, W. Hirota, and W.-C. Tan, “Deep entity matching: Challenges and opportunities,” Journal of Data and Information Quality, vol. 13, no. 1, pp. 1–17, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Y. Li, J. Li, Y. Suhara, A. Doan, and W.-C. Tan, “Deep entity matching with pre-trained language models,” arXiv preprint arXiv:2004.00584, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Y. Suhara, J. Li, Y. Li, D. Zhang, C ¸. Demiralp, C. Chen, and W.-C. Tan, “Annotating columns with pre-trained language models,” in Proceedings of the 2022 International Conference on Management of Data, 2022, pp.1493–1503.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "P. Yin, G. Neubig, W.-t. Yih, and S. Riedel, “Tabert: Pretraining for joint understanding of textual and tabular data,” arXiv preprint arXiv:2005.08314, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "European Commission, “Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation) (Text with EEA relevance),”2016. [Online]. Available: https://eur-lex.europa.eu/eli/reg/2016/679/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "“California consumer privacy act. bill no. 375 privacy: personal information: businesses. https://leginfo.legislature.ca.gov/.” 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "I. Absalyamov, M. J. Carey, and V. J. Tsotras, “Lightweight cardinality estimation in lsm-based systems,” in Proceedings of the 2018 International Conference on Management of Data, 2018, pp. 841–855.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in USENIX Symposium on Operating Systems Design and Implementation, 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "P. Sioulas and A. Ailamaki, “Scalable multi-query execution using reinforcement learning,” in Proceedings of the 2021 International Conference on Management of Data, 2021, pp. 1651–1663.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "A. Dutt, C. Wang, A. Nazi, S. Kandula, V. Narasayya, and S. Chaudhuri, “Selectivity estimation for range predicates using lightweight models,”Proceedings of the VLDB Endowment, vol. 12, no. 9, pp. 1044–1057, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "L. Zhang, C. Chai, X. Zhou, and G. Li, “Learnedsqlgen: Constraintaware sql generation using reinforcement learning,” in Proceedings of the 2022 International Conference on Management of Data, 2022, pp. 945–958.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "W. Wang, M. Zhang, G. Chen, H. Jagadish, B. C. Ooi, and K.-L. Tan, “Database meets deep learning: Challenges and opportunities,” ACM Sigmod Record, vol. 45, no. 2, pp. 17–22, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Z. Luo, S. Cai, C. Cui, B. C. Ooi, and Y. Yang, “Adaptive knowledge driven regularization for deep neural networks,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 35, no. 10, 2021, pp. 8810–8818.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "B. Denham, E. M. Lai, R. Sinha, and M. A. Naeem, “Witan: unsupervised labelling function generation for assisted data programming,” Proceedings of the VLDB Endowment, vol. 15, no. 11, pp. 2334–2347, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "H. Zhang, L. Cao, S. Madden, and E. Rundensteiner, “Lancet: labeling complex data at scale,” Proceedings of the VLDB Endowment, vol. 14, no. 11, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "D. Zha, Z. P. Bhat, K.-H. Lai, F. Yang, and X. Hu, “Data-centric ai: Perspectives and challenges,” in Proceedings of the 2023 SIAM International Conference on Data Mining, 2023, pp. 945–948.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Y. He, X. Chu, K. Ganjam, Y. Zheng, V. Narasayya, and S. Chaudhuri, “Transform-data-by-example (tde) an extensible search engine for datatransformations,” Proceedings of the VLDB Endowment ,v ol. 11, no. 10, pp. 1165–1177, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "P. Li, Y. He, C. Yan, Y. Wang, and S. Chauduri, “Auto-tables: Synthesizing multi-step transformations to relationalize tables without using examples,” arXiv preprint arXiv:2307.14565, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "H. Wang, Z. Luo, J. W. Yip, C. Ye, and M. Zhang, “Ecggan: A framework for effective and interpretable electrocardiogram anomalydetection,” in Proceedings of the 29th ACMSIGKDD Conference on Knowledge Discovery and Data Mining, 2023, pp. 5071–5081.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "K. Zheng, G. Chen, M. Herschel, K. Y. Ngiam, B. C. Ooi, and J. Gao, “Pace: learning effective task decomposition for human-inthe-loop healthcare delivery,” in Proceedings of the 2021 International Conference on Management of Data, 2021, pp. 2156–2168.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "K. Zheng, S. Cai, H. R. Chua, M. Herschel, M. Zhang, and B. C. Ooi, “Dyhealth: making neural networks dynamic for effective healthcareanalytics,” Proceedings of the VLDB Endowment, vol. 15, no. 12, pp. 3445–3458, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "M. Wawrzoniak, I. M ¨uller, R. Fraga Barcelos Paulus Bruno, and G. Alonso, “Boxer: Data analytics on network-enabled serverless platforms,” in Conference on Innovative Data Systems Research, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "S. Cai, K. Zheng, G. Chen, H. Jagadish, B. C. Ooi, and M. Zhang, “Arm-net: Adaptive relation modeling network for structured data,” inProceedings of the 2021 International Conference on Management of Data, 2021, pp. 207–220.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "P. Bohannon, X. Dong, S. Ganguly, H. F. Korth, C. Li, P. Narayan, and P. Shenoy, “Rolex: relational on-line exchange with xml,” in Proceedings of the 2003 International Conference on Management of Data, 2003, pp. 673–673.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "C. Lei, A. Quamar, V. Efthymiou, F. ¨Ozcan, and R. Alotaibi, “Hermes: data placement and schema optimization for enterprise knowledgebases,” The VLDB Journal, vol. 32, no. 3, pp. 549–574, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Z. Jin, M. R. Anderson, M. Cafarella, and H. Jagadish, “Foofah: Transforming data by example,” in Proceedings of the 2017 International Conference on Management of Data, 2017, pp. 683–698.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "M. Gertz, M. T. ¨Ozsu, G. Saake, and K.-U. Sattler, “Data quality on the web (dagstuhl seminar 03362),” 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "S. Sadiq, T. Dasu, X. L. Dong, J. Freire, I. F. Ilyas, S. Link, M. J. Miller, F. Naumann, X. Zhou, and D. Srivastava, “Data quality: The role of empiricism,” ACMSIGMOD Record, vol. 46, no. 4, pp. 35–43, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Y. Li, X. Wang, Z. Miao, and W.-C. Tan, “Data augmentation for ml-driven data preparation and integration,” Proceedings of the VLDB Endowment, vol. 14, no. 12, pp. 3182–3185, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Z. Luo, S. H. Yeung, M. Zhang, K. Zheng, L. Zhu, G. Chen, F. Fan, Q. Lin, K. Y. Ngiam, and B. C. Ooi, “Mlcask: Efficient managementof component evolution in collaborative data analytics pipelines,” in International Conference on Data Engineering, 2021, pp. 1655–1666.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "A. Halevy, A. Rajaraman, and J. Ordille, “Data integration: The teenage years,” in Proceedings of the 2006 International Conference on V ery Large Data Bases, 2006, pp. 9–16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "J. Zhang, Z. Luo, Q. Xu, and M. Zhang, “Pa-feat: Fast feature selection for structured data via progress-aware multi-task deep reinforcementlearning,” in gineering (ICDE), 2023, pp. 394–407.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "R. Fu, Y. Wu, Q. Xu, and M. Zhang, “Feast: A communication-efficient federated feature selection framework for relational data,” Proceedings of the ACM on Management of Data, vol. 1, no. 1, pp. 1–28, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Z. Cai, C. Jermaine, Z. V agena, D. Logothetis, and L. L. Perez, “The pairwise gaussian random field for high-dimensional data imputation,” in. 61–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Z. Zhong, M. Zhang, J. Fan, and C. Dou, “Semantics driven embedding learning for effective entity alignment,” in 2022 IEEE 38th International Conference on Data Engineering (ICDE), 2022, pp. 2127–2140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "N. Tang, J. Fan, F. Li, J. Tu, X. Du, G. Li, S. Madden, and M. Ouzzani, “Rpt: relational pre-trained transformer is almost all you need towards democratizing data preparation,” arXiv pr eprint arXiv:2012.02469, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "A. Halevy and J. Dwivedi-Y u, “Learnings from data integration for augmented language models,” arXiv preprint arXiv:2304.04576, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "N. Chen, L. Shou, M. Gong, J. Pei, C. Y ou, J. Chang, D. Jiang, and J. Li, “Bridge the gap between language models and tabular understanding,” arXiv preprint arXiv:2302.09302, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "H. Dong, Z. Cheng, X. He, M. Zhou, A. Zhou, F. Zhou, A. Liu, S. Han, and D. Zhang, “Table pre-training: A survey on model archi-tectures, pre-training objectives, and downstream tasks,” arXiv preprint arXiv:2201.09745, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "H. Doraiswamy, E. Tzirita Zacharatou, F. Miranda, M. Lage, A. Ailamaki, C. T. Silva, and J. Freire, “Interactive visual exploration of spatio5540 temporal urban data sets using urbane,” in Proceedings of the 2018 International Conference on Management of Data, 2018, pp. 1693–1696.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "P. Sinthong and M. J. Carey, “Exploratory data analysis with databasebacked dataframes: A case study on airbnb data,” in. 3119–3129.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "C. Lee, Z. Luo, K. Y. Ngiam, M. Zhang, K. Zheng, G. Chen, B. C. Ooi, and W. L. J. Yip, “Big healthcare data analytics: Challenges and applications,” Handbook of large-scale distributed computing in smart healthcare, pp. 11–41, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "J. Dai, M. Zhang, G. Chen, J. Fan, K. Y. Ngiam, and B. C. Ooi, “Fine-grained concept linking using neural networks in healthcare,” in Proceedings of the 2018 International Conference on Management of Data, 2018, pp. 51–66.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "S. Ahmetaj, V. Efthymiou, R. Fagin, P. G. Kolaitis, C. Lei, F. ¨Ozcan, and L. Popa, “Ontology-enriched query answering on relational databases,”inProceedings of the AAAI Conference on Artificial Intelligence, vol. 35, no. 17, 2021, pp. 15 247–15 254.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "K. Yang, Z. Luo, J. Gao, J. Zhao, B. C. Ooi, and B. Xie, “Ldareg: Knowledge driven regularization using external corpora,” IEEE Transactions on Knowledge and Data Engineering, vol. 34, no. 12, pp. 5840–5853, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Z. Luo, S. Cai, G. Chen, J. Gao, W.-C. Lee, K. Y. Ngiam, and M. Zhang, “Improving data analytics with fast and adaptive regularization,” IEEE Transactions on Knowledge and Data Engineering, vol. 33, no. 2, pp. 551–568, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Z. Chen, Z. Gu, L. Cao, J. Fan, S. Madden, and N. Tang, “Symphony: Towards natural language query answering over multi-modal data lakes,”inConference on Innovative Data Systems Research, 2023, pp. 8–151.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "M. Saeed, N. De Cao, and P. Papotti, “Querying large language models with sql,” arXiv preprint arXiv:2304.00472, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "A. M ¨uller, C. Curino, and R. Ramakrishnan, “Mothernet: A foundational hypernetwork for tabular classification,” arXiv preprint arXiv:2312.08598, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "S. Wang, D. Maier, and B. C. Ooi, “Lightweight indexing of observational data in log-structured storage,” Proceedings of the VLDB Endowment, vol. 7, no. 7, pp. 529–540, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Z. Liu, J. Wang, T. Dao, T. Zhou, B. Y uan, Z. Song, A. Shrivastava, C. Zhang, Y. Tian, C. Re et al., “Deja vu: Contextual sparsity for efficient llms at inference time,” in International Conference on Machine Learning, 2023, pp. 22 137–22 176.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "J. Zamfirescu-Pereira, R. Y. Wong, B. Hartmann, and Q. Yang, “Why johnny can’t prompt: how non-ai experts try (and fail) to design llm prompts,” in Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, 2023, pp. 1–21.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "H. Ferhatosmanoglu, E. Tuncel, D. Agrawal, and A. El Abbadi, “V ector approximation based indexing for non-uniform high dimensional datasets,” in Proceedings of the 2000 International Conference on Information and Knowledge Management, 2000, pp. 202–209.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "H. D. Chon, D. Agrawal, and A. E. Abbadi, “Range and k nn query processing for moving objects in grid model,” Mobile Networks and Applications, vol. 8, pp. 401–412, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "“Openai api pricing,” 2023, https://openai.com/pricing.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Z. Yang, P. Qi, S. Zhang, Y. Bengio, W. W. Cohen, R. Salakhutdinov, and C. D. Manning, “Hotpotqa: A dataset for diverse, explainable multi-hop question answering,” arXiv preprint arXiv:1809.09600, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "T. Y u, R. Zhang, K. Yang, M. Yasunaga, D. Wang, Z. Li, J. Ma, I. Li, Q. Yao, S. Roman et al., “Spider: A large-scale human-labeled dataset for complex and cross-domain semantic parsing and text-to-sql task,”arXiv preprint arXiv:1809.08887, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "D. Gao, H. Wang, Y. Li, X. Sun, Y. Qian, B. Ding, and J. Zhou, “Textto-sql empowered by large language models: A benchmark evaluation,”arXiv preprint arXiv:2308.15363, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "C. Wei, B. Wu, S. Wang, R. Lou, C. Zhan, F. Li, and Y. Cai, “Analyticdbv: A hybrid analytical engine towards query fusion for structured and unstructured data, ”Proceedings of the VLDB Endowment, vol. 13, no. 12, pp. 3152–3165, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "C. Li, M. Zhang, D. G. Andersen, and Y. He, “Improving approximate nearest neighbor search through learned adaptive early termination,” in Proceedings of the 2020 International Conference on Management of Data, 2020, pp. 2539–2554.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "P. Zhang, B. Yao, C. Gao, B. Wu, X. He, F. Li, Y. Lu, C. Zhan, and F. Tang, “Learning-based query optimization for multi-probe approxi-mate nearest neighbor search,” The VLDB Journal, vol. 32, no. 3, pp. 623–645, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "F. Bang, “Gptcache: An open-source semantic cache for llm applications enabling faster answers and cost savings,” in 3rd Workshop for Natural Language Processing Open Source Software, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "M. Altinel, Q. Luo, S. Krishnamurthy, C. Mohan, H. Pirahesh, B. G. Lindsay, H. Woo, and L. Brown, “Dbcache: Database caching forweb application servers,” in Proceedings of the 2002 International Conference on Management of Data, 2002, pp. 612–612.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "J. DeBrabant, A. Pavlo, S. Tu, M. Stonebraker, and S. Zdonik, “Anticaching: A new approach to database management system architecture,” Proceedings of the VLDB Endowment, vol. 6, no. 14, pp. 1942–1953, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "D. Lee, J. Choi, J.-H. Kim, S. H. Noh, S. L. Min, Y. Cho, and C. S. Kim, “On the existence of a spectrum of policies that subsumes theleast recently used (lru) and least frequently used (lfu) policies,” in Proceedings of the 1999 ACMSIGMETRICS international conference on Measurement and modeling of computer systems, 1999, pp. 134–143.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "V. Costan and S. Devadas, “Intel SGX explained,” IACR Cryptology ePrint Archive, p. 86, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "T. Lee, Z. Lin, S. Pushp, C. Li, Y. Liu, Y. Lee, F. Xu, C. Xu, L. Zhang, and J. Song, “Occlumency: Privacy-preserving remote deep-learning inference using SGX,” in MobiCom, 2019, pp. 46:1–46:17.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "P. Mishra, R. Lehmkuhl, A. Srinivasan, W. Zheng, and R. A. Popa, “Delphi: A cryptographic inference service for neural networks,” in USENIX, S. Capkun and F. Roesner, Eds., 2020, pp. 2505–2522.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "Y. Xu, W. Cui, and M. Peinado, “Controlled-channel attacks: Deterministic side channels for untrusted operating systems,” in IEEE S&P, 2015, pp. 640–656.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "J. V. Bulck, F. Piessens, and R. Strackx, “Nemesis: Studying microarchitectural timing leaks in rudimentary CPU interrupt logic,” in CCS, D. Lie, M. Mannan, M. Backes, and X. Wang, Eds., 2018, pp. 178–195.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Y. Wang, Y. Wu, X. Chen, G. Feng, and B. C. Ooi, “Incentiveaware decentralized data collaboration,” Proceedings of the ACM on Management of Data, vol. 1, no. 2, pp. 158:1–158:27, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Y. Wu, S. Cai, X. Xiao, G. Chen, and B. C. Ooi, “Privacy preserving vertical federated learning for tree-based models,” Proc. VLDB Endow., vol. 13, no. 11, pp. 2090–2103, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "Y. Wu, N. Xing, G. Chen, T. T. A. Dinh, Z. Luo, B. C. Ooi, X. Xiao, and M. Zhang, “Falcon: A privacy-preserving and interpretable vertical federated learning system,” Proc. VLDB Endow., vol. 16, no. 10, pp. 2471–2484, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Z. Zhang, Y. Yang, Y. Dai, Q. Wang, Y. Y u, L. Qu, and Z. Xu, “Fedpetuning: When federated learning meets the parameter-efficient tuning methods of pre-trained language models,” in ACL. Association for Computational Linguistics (ACL), 2023, pp. 9963–9977.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "M. Fredrikson, S. Jha, and T. Ristenpart, “Model inversion attacks that exploit confidence information and basic countermeasures,” in CCS, 2015, pp. 1322–1333.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "R. Shokri, M. Stronati, C. Song, and V. Shmatikov, “Membership inference attacks against machine learning models,” IEEE S&P, pp. 3– 18, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "X. Luo, Y. Wu, X. Xiao, and B. C. Ooi, “Feature inference attack on model predictions in vertical federated learning,” in 2021 IEEE 37th International Conference on Data Engineering, 2021, pp. 181–192.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "M. Abadi, A. Chu, I. J. Goodfellow, H. B. McMahan, I. Mironov, K. Talwar, and L. Zhang, “Deep learning with differential privacy,” in CCS, E. R. Weippl, S. Katzenbeisser, C. Kruegel, A. C. Myers, and S. Halevi, Eds., 2016, pp. 308–318.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "Z. Xu, Y. Zhang, G. Andrew, C. A. Choquette-Choo, P. Kairouz, H. B. McMahan, J. Rosenstock, and Y. Zhang, “Federated learning of gboard language models with differential privacy,” arXiv preprint arXiv:2305.18465 ,2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "F. Nargesian, A. Asudeh, and H. Jagadish, “Responsible data integration: Next-generation challenges,” in Proceedings of the 2022 International Conference on Management of Data, 2022, pp. 2458–2464.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "L. Ouyang, J. Wu, X. Jiang, D. Almeida, C. Wainwright, P. Mishkin, C. Zhang, S. Agarwal, K. Slama, A. Ray et al., “Training language models to follow instructions with human feedback,” Advances in Neural Information Processing Systems, vol. 35, pp. 27 730–27 744, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "G. Li, “Human-in-the-loop data integration,” Proceedings of the VLDB Endowment, vol. 10, no. 12, pp. 2006–2017, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "E. Estell ´es-Arolas and F. Gonz ´alez-Ladr ´on-de Guevara, “Towards an integrated crowdsourcing definition,” Journal of Information science, vol. 38, no. 2, pp. 189–200, 2012. 5541",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 20,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in USENIX Symposium on Operating Systems Design and Implementation, 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[20]",
+      "technique": "pqs",
+      "sentence": "For example, to comprehensively detect the bugs of DBMS, it is important to feed the database with ahuge number of SQL queries [20].",
+      "context_before": "rest of the paper is organized as follows. In Section II, we present four categories of applications of LLMs in the data management field. In Section III, we present challenges and opportunities arising from adapting LLMs to data management applications. Finally, we conclude the paper in Section IV. II. APPLICA TIONS OFLLM SI NDATA MANAGEMENT In this section, we present a series of promising applications that LLMs can be utilized in the four steps of the datamanagement application pipeline in Figure 1. A. LLM for Data Generation Data management tasks usually require a large volume of data [19].",
+      "context_after": "For another example, totrain better AI4DB models (e.g., learning-based cardinalityestimator, query optimizer and so on [21], [22]), it requires substantial training data, like < query, execution time > pairs. However, acquiring such vast real data is challenging due to privacy issues and high collection costs (i.e., collectingthe execution time or cardinality of complex queries aretime-consuming). Next, we will illustrate two important datageneration tasks in data management, i.e., SQL generation andtraining data generation, and demonstrate the application of LLMs in these tasks. 1) SQL Genera",
+      "section": "II APPLICA TIONS OFLLM SI NDATA MANAGEMENT",
+      "page": 2,
+      "char_offset": 9333,
+      "cited_reference": {
+        "number": 20,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in USENIX Symposium on Operating Systems Design and Implementation, 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[20]",
+      "technique": "pqs",
+      "sentence": "Meanwhile, to detect the logic bugs of DBMS, weneed to generate some SQL queries with semantic equivalence,which produce the same results [20].",
+      "context_before": "generation, and demonstrate the application of LLMs in these tasks. 1) SQL Generation: While the research on SQL generation has been improving over the years, there are also some challenges. The first challenge is how to generate complex queries, e.g., SQL queries containing sub-queries or multi-table joins, which need a deep understanding of the underlyingdatabases. The second challenge is how to generate SQLqueries that meet certain user-defined constraints [23]. For example, it is significant to generate diverse and correctly executable SQL queries for thoroughly testing the performanceof DBMS.",
+      "context_after": "LLMs can play a significant role in generating SQL queries by leveraging their natural language understanding and generation capabilities, as shown in Figure 2. Specifically, LLMs 5531 TXHU\\ WLPH TXHU\\ WLPH TXHU\\ WLPH TXHU\\ WLPH 'DWDEDVH ,QIRUPDWLRQ TXHU\\ BBBB TXHU\\ BBBB 3URPSWV 'DWDWREHSUHGLFWHG TXHU\\ WLPH TXHU\\ WLPH Fig. 3. Training data generation with LLMs. We feed some labeled training data (e.g., the < query, execution time > pairs) and database information (e.g., the table schema, table statistical information, etc.) into LLMs. For the coming query (i.e., the data to be predicte",
+      "section": "II APPLICA TIONS OFLLM SI NDATA MANAGEMENT",
+      "page": 2,
+      "char_offset": 10586,
+      "cited_reference": {
+        "number": 20,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in USENIX Symposium on Operating Systems Design and Implementation, 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:52Z",
+    "is_model_written": true,
+    "summary": "A vision paper on where large language models fit into data management, surveying applications across the field and the challenges each raises. DBMS testing appears as one such application.",
+    "narrative": "The citation supports two sentences about what DBMS testing requires: that comprehensive bug detection needs a large volume of SQL queries fed to the system, and that finding logic bugs specifically requires generating semantically equivalent queries that should return the same results. Both are reachable only through the citation marker, and neither names SQLancer or uses anything of it.",
+    "roles": {
+      "M1": "background",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icde65706_2026_00180.json b/_data/papers/paper_doi_10_1109_icde65706_2026_00180.json
new file mode 100644
index 0000000..8fbd5cf
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icde65706_2026_00180.json
@@ -0,0 +1,1015 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:44Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icde65706.2026.00180",
+    "title": "VIREO: Human-in-the-Loop DBMS Fuzzing with Visualization and LLM Support",
+    "authors": [
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Chi Zhang",
+      "Runpei Miao",
+      "Zhuo Su",
+      "Yu Jiang",
+      "Shuai Ma"
+    ],
+    "year": 2026,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde65706.2026.00180",
+    "arxiv_id": null,
+    "s2_paper_id": "c487d2eac342b4a4a4991d00b5bc5f3aabf3c36e",
+    "url": "https://doi.org/10.1109/icde65706.2026.00180",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icde65706.2026.00180",
+      "retrieved_at": "2026-09-09T01:29:44Z",
+      "chars": 80952,
+      "content_sha256": "sha256:fb6174b0026ba46dc8868a9b8ebb60141bfbaafd19c1f62315f5887b355d1025"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2123
+      },
+      {
+        "number": "II",
+        "title": "RELATED WORK",
+        "start": 9589
+      },
+      {
+        "number": "A",
+        "title": "Fuzzing",
+        "start": 9606
+      },
+      {
+        "number": "B",
+        "title": "DBMS Fuzzing",
+        "start": 11336
+      },
+      {
+        "number": "III",
+        "title": "MOTIVATION",
+        "start": 12691
+      },
+      {
+        "number": "IV",
+        "title": "DESIGN OFVIREO",
+        "start": 17959
+      },
+      {
+        "number": "A",
+        "title": "Module to Feature Analysis",
+        "start": 18680
+      },
+      {
+        "number": "B",
+        "title": "Human-in-the-Loop DBMS Fuzzing",
+        "start": 29105
+      },
+      {
+        "number": "V",
+        "title": "IMPLEMENTATION",
+        "start": 37926
+      },
+      {
+        "number": "VI",
+        "title": "EVALUATION",
+        "start": 40413
+      },
+      {
+        "number": "A",
+        "title": "Evaluation Setup",
+        "start": 40873
+      },
+      {
+        "number": "B",
+        "title": "Evaluation Procedure",
+        "start": 42694
+      },
+      {
+        "number": "C",
+        "title": "DBMS Vulnerability Detection Results",
+        "start": 44184
+      },
+      {
+        "number": "D",
+        "title": "Compared to Existing DBMS Testing Techniques",
+        "start": 50508
+      },
+      {
+        "number": "E",
+        "title": "Necessity of Human Assistance",
+        "start": 55559
+      },
+      {
+        "number": "F",
+        "title": "Necessity of LLM Support",
+        "start": 58114
+      },
+      {
+        "number": "G",
+        "title": "Overhead Analysis",
+        "start": 60510
+      },
+      {
+        "number": "VII",
+        "title": "DISCUSSION",
+        "start": 62397
+      },
+      {
+        "number": "VIII",
+        "title": "CONCLUSION",
+        "start": 66374
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "[n. d.]. Google Chrome impacted by new Magellan 2.0 vulnerabilities. https://www.zdnet.com/article/googlechrome-impacted-by-new-magellan-2-0-vulnerabilities/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "[n. d.]. MariaDB GitHub. https://github.com/MariaDB/ server. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "[n. d.]. SQLite patches use-after-free bug that left apps open to code execution, denial-of-service exploits. https://portswigger.net/daily-swig/sqlite-patches-useafter-free-bug-that-left-apps-open-to-code-executiondenial-of-service-exploits. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "[n. d.]. zzuf. https://github.com/samhocevar/zzuf. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Cornelius Aschermann, Sergej Schumilo, Ali Abbasi, and Thorsten Holz. 2020. Ijon: Exploring Deep State Spaces via Fuzzing. In curity and Privacy (SP). 1597–1612. doi:10.1109/ SP40000.2020.00117",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Cornelius Aschermann, Sergej Schumilo, Tim Blazytko, Robert Gawlik, and Thorsten Holz. 2019. REDQUEEN: Fuzzing with Input-to-State Correspondence. In Symposium on Network and Distributed System Security (NDSS).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. Proceedings of the ACM on Management of Data 2, 3 (2024), 1–26.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 8,
+      "text": "Avatar Magnus Bl ˚audd. [n. d.]. Percona Website. https: //www.percona.com/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Bootstrap. 2026. Build fast, responsive sites with Bootstrap. https://getbootstrap.com/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Theodore Chambers, Michael Vierhauser, Ankit Agrawal, Michael Murphy, Jason Matthew Brauer, Salil Purandare, Myra B Cohen, and Jane Cleland-Huang. 2024. HIFuzz: Human Interaction Fuzzing for Small Unmanned Aerial Vehicles. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (Honolulu, HI, USA) (CHI ’24). Association for Computing Machinery, New York, NY, USA, Article 26",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Chango chen. [n. d.]. Squirrel Website. https:// github.com/s3team/Squirrel. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Peng Chen and Hao Chen. 2018. Angora: Efficient fuzzing by principled search. In IEEE Symposium on Security and Privacy (S&P).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Yuanliang Chen, Yu Jiang, Fuchen Ma, Jie Liang, Mingzhe Wang, Chijin Zhou, Xun Jiao, and Zhuo Su. 2019. EnFuzz: Ensemble Fuzzing with Seed Synchronization among Diverse Fuzzers. In USENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Carlos Coronel and Steven Morris. 2019. Database systems: design, implementation and management. Cengage learning.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "DynamoRIO. 2026. DynamoRIO: Dynamic Instrumentation Tool Platform. https://dynamorio.org/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "echarts. 2026. Apache ECharts. https://github.com/ apache/echarts. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Cundi Fang, Jie Liang, Zhiyong Wu, Jingzhou Fu, Zhouyang Jia, Chun Huang, Yu Jiang, and Shanshan Li. 2025. DepState: Detecting Synchronization Failure Bugs in Distributed Database Management Systems. Proceedings of the ACM on Software Engineering 2, ISSTA (2025), 2001–2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. In Conference on Automated Software Engineering (ASE’22).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and detecting sql function bugs: using simple boundary arguments to trigger hundreds of dbms bugs. In Proceedings of the Twentieth European Conference on Computer Systems. 1061–1076.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Ying Fu, Zhiyong Wu, Yuanliang Zhang, Jie Liang, Jingzhou Fu, Yu Jiang, Shanshan Li, and Xiangke Liao. 2025. Thanos: Dbms bug detection via storage engine rotation based differential testing. In 2025 IEEE/ACM 47th International Conference on Software Engineering (ICSE). IEEE Computer Society, 655–666.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Wentao Gao, Van-Thuan Pham, Dongge Liu, Oliver Chang, Toby Murray, and Benjamin IP Rubinstein. 2023. Beyond the coverage plateau: A comprehensive study of fuzz blockers (registered report). In Proceedings of the 2nd International Fuzzing Workshop. 47–55.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "gcovr. 2026. gcovr: generate GCC code coverage reports. https://github.com/gcovr/gcovr. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "GreatDB. [n. d.]. GreatSQL Github. https://github.com/ GreatSQL/GreatSQL. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "The PostgreSQL Global Development Group. 2023. PostgreSQL 15 Documentation, Query Planning. https://www.postgresql.org/docs/current/runtime-configquery.html. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "The PostgreSQL Global Development Group. 2025. pgsql-bugs. https://www.postgresql.org/list/pgsql-bugs/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Tao Guo, Puhan Zhang, Xin Wang, and Qiang Wei. 2417 2013. Gramfuzz: Fuzzing testing of web browsers based on grammar analysis and structural mutation. In 2013 Second International Conference on Informatics & Applications (ICIA). IEEE, 212–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "IBM. 2024. Modules-IBM Documentation. https://www.ibm.com/docs/en/db2-warehouse?topic= reference-modules. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "jacoco. 2026. JaCoCo Java Code Coverage Library. https: //github.com/jacoco/jacoco. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. {DynSQL}: Stateful Fuzzing for Database Management Systems with Complex and Valid {SQL} Query Generation. In 32nd USENIX Security Symposium (USENIX Security 23). 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "jquery. 2026. jQuery. https://jquery.com/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Vijay Kumar and Sang Hyuk Son. 2012. Database recovery. V ol. 12. Springer Science & Business Media.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Chris Lattner and Vikram Adve. 2004. LLVM: A compilation framework for lifelong program analysis & transformation. In Iinternational Symposium on Code generation and optimization.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Caroline Lemieux and Koushik Sen. 2018. Fairfuzz: Targeting rare branches to rapidly increase greybox fuzz testing coverage. In ACM International Conference on Automated Software Engineering (ASE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-Oriented DBMS Fuzzing. In on Data Engineering (ICDE). IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Jie Liang, Yu Jiang, Mingzhe Wang, Xun Jiao, Yuanliang Chen, Houbing Song, and Kim-Kwang Raymond Choo. 2019. Deepfuzzer: Accelerated deep greybox fuzzing. IEEE Transactions on Dependable and Secure Computing(2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Jie Liang, Mingzhe Wang, Chijin Zhou, Zhiyong Wu, Yu Jiang, Jianzhong Liu, Zhe Liu, and Jiaguang Sun. 2022. PATA: Fuzzing with Path Aware Taint Analysis. In 2022. IEEE Computer Society, Los Alamitos, CA, USA. 154– 170.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Jie Liang, Mingzhe Wang, Chijin Zhou, Zhiyong Wu, Jianzhong Liu, and Yu Jiang. 2024. Dodrio: Parallelizing taint analysis based fuzzing via redundancy-free scheduling. In Companion Proceedings of the 32nd ACM International Conference on the Foundations of Software Engineering. 244–254.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Yiyuan Bai, Qiang Zhang, and Yu Jiang. 2024. {WingFuzz}: Implementing continuous fuzzing for {DBMSs}. In 2024 USENIX Annual Technical Conference (USENIX ATC 24). 479– 492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS Bugs via Configuration-Based EquivalentTransformation. In Proceedings of the 46th IEEE/ACM International Conference on Software Engineering, ICSE 2024, Lisbon, Portugal, April 14-20, 2024. ACM, 135:1– 135:12. doi:10.1145/3597503.3639112",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of DBMS with Coverage-based Guidance. InProceedings of the 31st USENIX Security Symposium (USENIX 2022). Boston, MA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "LibFuzzer [n. d.]. LibFuzzer. https://www.llvm. org/docs/ LibFuzzer.html. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Chi-Keung Luk, Robert Cohn, Robert Muth, Harish Patil, Artur Klauser, Geoff Lowney, Steven Wallace, Vijay Janapa Reddi, and Kim Hazelwood. 2005. Pin: building customized program analysis tools with dynamic instrumentation. In Proceedings of the 2005 ACMSIGPLAN Conference on Programming Language Design and Implementation (Chicago, IL, USA) (PLDI ’05). Association for Computing Machinery, New York, ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Valentin JM Man `es, HyungSeok Han, Choongwoo Han, Sang Kil Cha, Manuel Egele, Edward J Schwartz, and Maverick Woo. 2019. The art, science, and engineering of fuzzing: A survey. IEEE Transactions on Software Engineering 47, 11 (2019), 2312–2331.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "MariaDB 2024. MariaDB. https://mariadb.org/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "MySQL 2014. MySQL 8.0 Reference Manual, Switchable Optimizations. https://dev.mysql.com/doc/refman/ 8.0/en/switchable-optimizations.html. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "MySQL 2024. MySQL. https://www.mysql.com/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "MySQL. 2025. MySQL Bug Home. https:// bugs.mysql.com/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "MySQL. 2025. Server Option, System Variable, and Status Variable Reference. https://dev.mysql.com/doc/ refman/8.4/en/server-option-variable-reference.html. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Oracle. 2026. Alternative Storage Engines. https:// dev.mysql.com/doc/refman/8.4/en/storage-engines.html. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Oracle. 2026. Server SQL Modes. https:// dev.mysql.com/doc/refman/9.1/en/sql-mode.html. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Oracle. 2026. SHOWGRANTS Statement. https: //dev.mysql.com/doc/refman/8.4/en/show-grants.html. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Laurent Simon, and Hayawardh Vijayakumar. 2019. FuzzFactory: domain-specific fuzzing with waypoints. Proc. ACM Program. Lang. 3, OOPSLA, Article 174 (Oct. 2019), 29 pages. doi:10.1145/3360600",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "PeachTech [n. d.]. What is Peach. https: //peachtech.gitlab.io/peach-fuzzer-community/ WhatIsPeach.html. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "OpenAI Platform. 2026. GPT-4o. https: 2418 //platform.openai.com/docs/models/gpt-4o. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Sebastian Poeplau and Aur ´elien Francillon. 2020. Symbolic execution with SymCC: Don’t interpret, compile!. InUSENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "PostgreSQL 2024. PostgreSQL. https: //www.postgresql.org/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Manuel Rigger. [n. d.]. SQLancer Website. https: //github.com/sqlancer/sqlancer. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 58,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 59,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 60,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 61,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https: //github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Rupal Sharma. 2024. DBMS Architecture: Its 5 Key Components and Types of Database Models. https://www.sprinkledata.com/blogs/dbms-architectureits-5-key-components-and-types-of-database-models. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Nick Stephens, John Grosen, Christopher Salls, Andrew Dutcher, Ruoyu Wang, Jacopo Corbetta, Yan Shoshitaishvili, Christopher Kruegel, and Giovanni Vigna. 2016. Driller: Augmenting Fuzzing Through Selective Symbolic Execution.. In Symposium on Network and Distributed System Security (NDSS).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "wikipedia [n. d.]. databases. https://en.wikipedia.org/ wiki/Database. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Wenqian Deng, and Yu Jiang. 2025. {DDLumos}: Understanding and Detecting Atomic {DDL} Bugs in {DBMSs}. In 2025 USENIX Annual Technical Conference (USENIX ATC 25). 1327–1341.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Wenqian Deng, and Yu Jiang. 2025. Fawkes: Finding Data Durability Bugs in DBMSs via Recovered Data State Verification. InProceedings of the ACMSIGOPS 31st Symposium on Operating Systems Principles. 670–684.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. Hulk: Exploring data-sensitive performance anomalies in dbmss via data-driven analysis. Proceedings of the ACM on Software Engineering 2, ISSTA (2025), 2181–2202.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Insu Yun, Sangho Lee, Meng Xu, Yeongjin Jang, andTaesoo Kim. 2018. QSYM: A Practical Concolic Execution Engine Tailored for Hybrid Fuzzing. In USENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Michał Zalewski. [n. d.]. american fuzzy lop. http: //lcamtuf.coredump.cx/afl/. Accessed: March 14, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Xiaogang Zhu, Sheng Wen, Seyit Camtepe, and Yang Xiang. 2022. Fuzzing: a survey for roadmap. ACM Computing Surveys (CSUR) 54, 11s (2022), 1–36. 2419",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 7,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. Proceedings of the ACM on Management of Data 2, 3 (2024), 1–26.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 19,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and detecting sql function bugs: using simple boundary arguments to trigger hundreds of dbms bugs. In Proceedings of the Twentieth European Conference on Computer Systems. 1061–1076.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 57,
+      "text": "Manuel Rigger. [n. d.]. SQLancer Website. https: //github.com/sqlancer/sqlancer. Accessed: March 14, 2026.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 58,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 59,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 60,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Early DBMS fuzzers like SQLsmith [61] and SQLancer [57] primarily relied on randomly generated SQL test cases.",
+      "context_before": "ON Database Management Systems (DBMSs) are critical infrastructure supporting applications from finance to cloud services [14, 31, 64], but their growing complexity inevitably introduces bugs in the implementations. These bugs can have severe consequences, including data leaks, service disruptions, or even full system compromise by attackers [1, 3]. Recently, fuzzing [43, 71] has emerged as a state-of-the-art technique for identifying DBMS bugs. It automatically generates lots of test cases for a system to trigger potential crashes and other unexpected behaviors, thereby revealing hidden bugs.",
+      "context_after": "Later fuzzers like SQUIRREL [70], use coverage information from ∗Shuai Ma and Yu Jiang are the corresponding authors. Fig. 1. Under conventional DBMS fuzzing, after approximately four hours of testing, the coverage achieved by SQLsmith, SQLancer, SQUIRREL, and LEGO on MariaDB exhibits minimal growth. tested DBMSs to guide mutation-based test case generation. To ensure syntactic and semantic correctness, they incorporate semantics-aware mechanisms into mutation [18, 29, 40]. Building on this, LEGO [34] further learns the relationships between different SQL statements, enabling the generation o",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2736,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Under conventional DBMS fuzzing, after approximately four hours of testing, the coverage achieved by SQLsmith, SQLancer, SQUIRREL, and LEGO on MariaDB exhibits minimal growth.",
+      "context_before": "evere consequences, including data leaks, service disruptions, or even full system compromise by attackers [1, 3]. Recently, fuzzing [43, 71] has emerged as a state-of-the-art technique for identifying DBMS bugs. It automatically generates lots of test cases for a system to trigger potential crashes and other unexpected behaviors, thereby revealing hidden bugs. Early DBMS fuzzers like SQLsmith [61] and SQLancer [57] primarily relied on randomly generated SQL test cases. Later fuzzers like SQUIRREL [70], use coverage information from ∗Shuai Ma and Yu Jiang are the corresponding authors. Fig. 1.",
+      "context_after": "tested DBMSs to guide mutation-based test case generation. To ensure syntactic and semantic correctness, they incorporate semantics-aware mechanisms into mutation [18, 29, 40]. Building on this, LEGO [34] further learns the relationships between different SQL statements, enabling the generation of richer and more complex SQL sequences. However, in real-world industrial practice, existing DBMS fuzzers explore only a limited state space of the target DBMSs, significantly limiting their ability to uncover additional bugs. As shown in Figure 1, four popular fuzzers exhibit little coverage improve",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2973,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In 3 hours, VIREO found 98%, 59%, 38%, and 15% more branches and identified 9, 9, 8, and 7 more unique crash bugs than SQLsmith, SQLancer, SQUIRREL, and LEGO did in 24 hours, respectively.",
+      "context_before": "rom feature maps and target modules. We have implemented VIREO and used it on 5 well-tested DBMSs, namely PostgreSQL, MySQL, MariaDB, Percona,and GreatSQL. VIREO discovered a total of 25 previously unknown bugs, all of which were confirmed, and 10 have been fixed by vendors. Developer responses to our bug reports highlight the practical value of human-in-the-loop fuzzing. For example, PostgreSQL developers acknowledged that the identified bug exposed a planner assumption “previously believed unreachable by crafted input.” We also compare VIREO against other existing popular DBMS testing tools.",
+      "context_after": "Additionally, we demonstrate the necessity of human assistance and LLM support. In summary, we make the following contributions: 1) We identify that many existing DBMS fuzzers explore only a limited state space of DBMSs caused by DBMSspecific constraints, often leading to stagnation and leaving critical bugs undetected. 2) We propose VIREO, a human-in-the-loop DBMS fuzzing framework that combines module to feature analysis to systematically explore critical areas, overcome testing boundaries, and uncover latent crash bugs in DBMSs. 3) We use VIREO to find 25 bugs in popular DBMSs, all of them",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 8759,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[19, 57, 61, 65, 66, 67]",
+      "technique": null,
+      "sentence": "Generation-based DBMS fuzzing [19, 57, 61, 65, 66, 67] constructs SQL queries using predefined grammars, templates, or probabilistic models.",
+      "context_before": "s are still insufficient for DBMSs, which have rich syntax and complex dependencies among configurations, queries, and execution contexts. Random mutations often fail to satisfy these requirements, making it difficult to satisfy DBMS-specific constraints. While interactive fuzzing allows human guidance, humans alone cannot easily identify unexplored regions. Some DBMS bugs appear only under specific configurations or SQL sequences, making them difficult to trigger with annotations alone. 2406 B. DBMS Fuzzing DBMS fuzzing can also be classified into generationbased and mutation-based methods.",
+      "context_after": "For example, SQLsmith [61] generates SQL queries by randomly combining syntax elements to ASTs with metadata. SQLancer [57] utilizes multiple test oracles [7, 58, 59, 60] to detect logic bugs, generating queries based on these oracles. Mutation-based fuzzers [18, 29, 34, 38, 40, 70] generate new test cases by modifying existing queries. For example, SQUIRREL [70] designs an IR for structure-aware mutations. LEGO [34] learns and mutates sequences of SQL statements rather than individual statements, enabling it to more effectively exercise stateful behaviors such as schema evolution and cross-s",
+      "section": "B DBMS Fuzzing",
+      "page": 3,
+      "char_offset": 11436,
+      "cited_reference": {
+        "number": 57,
+        "text": "Manuel Rigger. [n. d.]. SQLancer Website. https: //github.com/sqlancer/sqlancer. Accessed: March 14, 2026.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [57] utilizes multiple test oracles [7, 58, 59, 60] to detect logic bugs, generating queries based on these oracles.",
+      "context_before": "nts. While interactive fuzzing allows human guidance, humans alone cannot easily identify unexplored regions. Some DBMS bugs appear only under specific configurations or SQL sequences, making them difficult to trigger with annotations alone. 2406 B. DBMS Fuzzing DBMS fuzzing can also be classified into generationbased and mutation-based methods. Generation-based DBMS fuzzing [19, 57, 61, 65, 66, 67] constructs SQL queries using predefined grammars, templates, or probabilistic models. For example, SQLsmith [61] generates SQL queries by randomly combining syntax elements to ASTs with metadata.",
+      "context_after": "Mutation-based fuzzers [18, 29, 34, 38, 40, 70] generate new test cases by modifying existing queries. For example, SQUIRREL [70] designs an IR for structure-aware mutations. LEGO [34] learns and mutates sequences of SQL statements rather than individual statements, enabling it to more effectively exercise stateful behaviors such as schema evolution and cross-statement dependencies. GRIFFIN [18] introduces a grammar-free mutation method by shuffling existing queries. Despite these advances, both categories of DBMS fuzzers struggle to cover complex state spaces. Many states require specific SQ",
+      "section": "B DBMS Fuzzing",
+      "page": 3,
+      "char_offset": 11687,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLsmith, SQUIRREL, SQLancer, and LEGO) covers only a limited portion of PostgreSQL after 72 hours, leaving much functionality untested and potential bugs hidden.",
+      "context_before": "45, 48] like optimization methods and indexing strategies, which significantly affect execution behavior. SQL statements interact in complex ways with these modules and settings, producing subtle, hard-topredict behaviors. Consequently, DBMS fuzzing faces significant obstacles due to the complex nature of database systems, where functional modules are tightly interwoven with configurations and input SQL dependencies. As Figure 1 shows, fuzzers exhibit little coverage improvement after the initial testing rounds. Figure 2 shows that even the combined coverage of popular DBMS testing tools (i.e.",
+      "context_after": "Causes of Limited State Space Exploration. Limited DBMS state space exploration arises from the complex constraints and sequence-dependent behaviors inherent in DBMS: (1) First, many DBMS functionalities can only be exercised through specific sequences of SQL statements and configurations. Existing fuzzers often struggle to generate such combinations because the required statements and configurations are highly constrained. (2) Second, existing fuzzers generally operate under preset configurations and fixed fuzzing strategies, without awareness of the current testing state. The static approac",
+      "section": "III MOTIVATION",
+      "page": 3,
+      "char_offset": 13569,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLsmith, SQUIRREL, SQLancer, and LEGO) on PostgreSQL still leaves a large portion of the code uncovered.",
+      "context_before": "ithout awareness of the current testing state. The static approach prevents dynamic exploration of complex or rarely used paths, leaving many DBMS state spaces untested. access bootstrapcatalog commandsexecutorlibpqnodes optimizerparser partitioning postmasterregex replicationrewrite statisticsstoragetcop tsearchutils020000400006000080000100000120000 Lines of Code 12.3% 4.2%61.3%52.8% 56.9% 29.1%44.8%82.7%72.9% 47.9%52.7%39.2%6.4% 62.6% 1.5%51.3% 51.3% 34.2%47.8% T otal Covered Codes by Existed Fuzzers Uncovered Codes in each ComponentFig. 2. The combined coverage of four popular fuzzers (i.e.",
+      "context_after": "Motivating Example. To illustrate the obstacle of traditional fuzzing, consider the example of a SEGV in MySQL shown in Figure 3. The bug requires a specific combination of privilege-related configuration and SQLs to be triggered. First, the global configuration --skip-grant-tables=1 is enabled to bypass privilege checks. Then, a prepared SHOWGRANTS statement is executed, combining preparation and privilege-checking operations in a way that triggers the SEGV. Despite running popular fuzzers like SQLsmith and SQLancer on MySQL, these tools failed to trigger the bug, as they neither explore con",
+      "section": "III MOTIVATION",
+      "page": 3,
+      "char_offset": 14868,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Despite running popular fuzzers like SQLsmith and SQLancer on MySQL, these tools failed to trigger the bug, as they neither explore configuration options like --skip-grant-tables nor generate SQL statements combining preparation and privilege-checking operations.",
+      "context_before": "of four popular fuzzers (i.e., SQLsmith, SQUIRREL, SQLancer, and LEGO) on PostgreSQL still leaves a large portion of the code uncovered. Motivating Example. To illustrate the obstacle of traditional fuzzing, consider the example of a SEGV in MySQL shown in Figure 3. The bug requires a specific combination of privilege-related configuration and SQLs to be triggered. First, the global configuration --skip-grant-tables=1 is enabled to bypass privilege checks. Then, a prepared SHOWGRANTS statement is executed, combining preparation and privilege-checking operations in a way that triggers the SEGV.",
+      "context_after": "Moreover, these tools are hard to adapt their strategies based on uncovered functionalities, making it difficult to adjust configurations or generate the specific SQL grammars needed. To address the obstacle, VIREO leverages human-in-theloop guidance.1 In the preparation stage, VIREO maps DBMS modules to relevant grammar rules and configurations. 2 –4 Through visualization-based interactions, the test engineer selects the auth module as a promising target due to its low prior coverage and history of security issues. 5 –6 Using the derived mappings, VIREO suggests adjustments to the fuzzing en",
+      "section": "III MOTIVATION",
+      "page": 3,
+      "char_offset": 15439,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We tried our best to compare VIREO against state-of-the-art DBMS fuzzers, including SQLsmith [61], SQLancer [57], SQUIRREL [70], and LEGO [34].",
+      "context_before": "quired to interact effectively with VIREO. Their backgrounds covered areas such as database management, systems programming, and software testing. Participants were recruited from the same institution and voluntarily agreed to take part in the study, with no prior exposure to VIREO to avoid bias. Tested DBMS. We targeted five DBMSs, including PostgreSQL [56], MySQL [46], MariaDB [44], Percona [8], and GreatSQL [23]. They are widely used and have been extensively tested with existing DBMS fuzzers. For each DBMS, we used the latest available version at the time of evaluation. Compare Techniques.",
+      "context_after": "They are widely used and have uncovered hundreds of DBMS bugs. Among them, SQLsmith, SQUIRREL, and LEGO target crash bugs, which align with Vireo’s focus. Although SQLancer was designed for logic bugs, it can also expose crashes and is evaluated using its default PQS oracle [60]. For mutation-based fuzzers, initial test cases were derived from the official unit test suites of each DBMS.Environment Setup. The experiments were conducted on a machine running 64-bit Ubuntu 20.04, equipped with an AMDEPYC 7742 processor @ 2.25 GHz, 128 cores, and 504 GiB of RAM. All DBMSs were tested within Docker",
+      "section": "A Evaluation Setup",
+      "page": 8,
+      "char_offset": 41720,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Although SQLancer was designed for logic bugs, it can also expose crashes and is evaluated using its default PQS oracle [60].",
+      "context_before": "ested DBMS. We targeted five DBMSs, including PostgreSQL [56], MySQL [46], MariaDB [44], Percona [8], and GreatSQL [23]. They are widely used and have been extensively tested with existing DBMS fuzzers. For each DBMS, we used the latest available version at the time of evaluation. Compare Techniques. We tried our best to compare VIREO against state-of-the-art DBMS fuzzers, including SQLsmith [61], SQLancer [57], SQUIRREL [70], and LEGO [34]. They are widely used and have uncovered hundreds of DBMS bugs. Among them, SQLsmith, SQUIRREL, and LEGO target crash bugs, which align with Vireo’s focus.",
+      "context_after": "For mutation-based fuzzers, initial test cases were derived from the official unit test suites of each DBMS.Environment Setup. The experiments were conducted on a machine running 64-bit Ubuntu 20.04, equipped with an AMDEPYC 7742 processor @ 2.25 GHz, 128 cores, and 504 GiB of RAM. All DBMSs were tested within Docker containers. Each container was allocated 5 CPU cores and 40 GiB of RAM. All LLM-based analysis was performed using OpenAI’s GPT-4o via its online API. No local deployment or additional training was performed for the experiments. B. Evaluation Procedure Each participant first rece",
+      "section": "A Evaluation Setup",
+      "page": 8,
+      "char_offset": 42019,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQUIRREL, SQLancer, SQLsmith, and LEGO) under the same testing environment for 24 hours.",
+      "context_before": "ughout the testing process, we recorded all triggered crashes and deduplicated them to count unique bugs. To detect real-world bugs, each DBMS was independently tested by three participants for three hours each per day (i.e., nine hours of testing per day for one DBMS in total) over a period of five days. For quantitative comparisons, each DBMS was tested for three hours by three randomly assigned participants, and the results were averaged to enhance the reliability of the measurements. To evaluate the effectiveness of VIREO comparatively, we also tested four existing DBMS fuzzing tools (i.e.",
+      "context_after": "Crashes triggered by each tool were recorded, deduplicated to obtain unique bug counts, and code coverage was analyzed, enabling a direct comparison with VIREO ’s testing results. C. DBMS Vulnerability Detection Results 1) Bug Statistics: Despite the extensive testing and maturity of these DBMSs, and even though the participants had no prior experience with VIREO, it successfully uncovered 25 previously unknown vulnerabilities. Table I summarizes these findings, with VIREO identifying 1, 4, 4, 7, and 9 in PostgreSQL, MySQL, MariaDB, Percona, and GreatSQL, respectively. All identified bugs wer",
+      "section": "B Evaluation Procedure",
+      "page": 8,
+      "char_offset": 43913,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Compared to Existing DBMS Testing Techniques We compared VIREO against SQLsmith, SQLancer, SQUIR REL, and LEGO.",
+      "context_before": "tified the testing boundary and recognized the privilege management subsystem as a promising target due to its low coverage and historical bug density. VIREO then generated a refinement strategy involving adjustments to the configurations key_buffer_size andkey_cache_segments as well as the use of grammar structures such as SET, INDEX, and DELETE HISTORY. The participants confirmed this strategy without modification. Guided by it, VIREO adjusted the configurations and fuzzing engine, conducted focused testing on the subsystem, and ultimately triggered the previously unknown bug in GreatSQL. D.",
+      "context_after": "Following prior work, we evaluated them using two metrics, namely branch coverage and the number of unique bugs triggered. For VIREO, the number of bugs discovered for one DBMS by each student may vary. Consequently, we report the average results obtained from three participants after 3 hours of testing for each DBMS. For the baseline fuzzers, we report their results after 24 hours of automated testing. To enable a standardized comparison, we collected the test cases generated by each fuzzer and reran them to uniform branch coverage and covered functions. Unique bugs were identified through c",
+      "section": "D Compared to Existing DBMS Testing Techniques",
+      "page": 9,
+      "char_offset": 50510,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Table II presents the branch coverage results, showing that VIREO achieved in 3 hours results that surpassed the 24-hour runs of SQLsmith, SQLancer, SQUIRREL, and LEGO by 98%, 59%, 38%, and 15%, respectively.",
+      "context_before": ", namely branch coverage and the number of unique bugs triggered. For VIREO, the number of bugs discovered for one DBMS by each student may vary. Consequently, we report the average results obtained from three participants after 3 hours of testing for each DBMS. For the baseline fuzzers, we report their results after 24 hours of automated testing. To enable a standardized comparison, we collected the test cases generated by each fuzzer and reran them to uniform branch coverage and covered functions. Unique bugs were identified through call stack comparison and manual analysis. 2413 Coverage.",
+      "context_after": "Similarly, Table III presents the functions covered, showing that VIREO achieved in 3 hours results that surpassed the 24-hour runs of SQLsmith, SQLancer, SQUIRREL, and LEGO by 95%, 66%, 32%, and 14%, respectively. TABLE IINUMBER OFBRANCHES COVERED BYDIFFERENT FUZZERS (SQLSMITH, SQLANCER, SQUIRREL ,ANDLEGO IN 24HOURS; VIREO IN 3HOURS ) DBMS SQLsmith SQLancer SQUIRREL LEGOVIREO PostgreSQL 54,745 36,844 39,483 52,193 73,041 MySQL 39,445 41,783 43,944 74,136 81,034 MariaDB 40,394 74,364 84,945 91,083 108,238 Percona 47,293 53,784 59,384 69,411 77,032 GreatSQL 38,475 67,659 88,629 92,081 97,359 T",
+      "section": "D Compared to Existing DBMS Testing Techniques",
+      "page": 10,
+      "char_offset": 51280,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Similarly, Table III presents the functions covered, showing that VIREO achieved in 3 hours results that surpassed the 24-hour runs of SQLsmith, SQLancer, SQUIRREL, and LEGO by 95%, 66%, 32%, and 14%, respectively.",
+      "context_before": "participants after 3 hours of testing for each DBMS. For the baseline fuzzers, we report their results after 24 hours of automated testing. To enable a standardized comparison, we collected the test cases generated by each fuzzer and reran them to uniform branch coverage and covered functions. Unique bugs were identified through call stack comparison and manual analysis. 2413 Coverage. Table II presents the branch coverage results, showing that VIREO achieved in 3 hours results that surpassed the 24-hour runs of SQLsmith, SQLancer, SQUIRREL, and LEGO by 98%, 59%, 38%, and 15%, respectively.",
+      "context_after": "TABLE IINUMBER OFBRANCHES COVERED BYDIFFERENT FUZZERS (SQLSMITH, SQLANCER, SQUIRREL ,ANDLEGO IN 24HOURS; VIREO IN 3HOURS ) DBMS SQLsmith SQLancer SQUIRREL LEGOVIREO PostgreSQL 54,745 36,844 39,483 52,193 73,041 MySQL 39,445 41,783 43,944 74,136 81,034 MariaDB 40,394 74,364 84,945 91,083 108,238 Percona 47,293 53,784 59,384 69,411 77,032 GreatSQL 38,475 67,659 88,629 92,081 97,359 Total 220,352 274,434 316,385 378,904 436,704 Improvement 98%↑ 59%↑ 38%↑ 15%↑ – TABLE IIINUMBER OFFUNCTIONS COVERED BYDIFFERENT FUZZERS (SQLSMITH, SQLANCER, SQUIRREL ,ANDLEGO IN 24HOURS; VIREO IN 3HOURS ) DBMS SQLsmi",
+      "section": "D Compared to Existing DBMS Testing Techniques",
+      "page": 10,
+      "char_offset": 51489,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "TABLE IINUMBER OFBRANCHES COVERED BYDIFFERENT FUZZERS (SQLSMITH, SQLANCER, SQUIRREL ,ANDLEGO IN 24HOURS; VIREO IN 3HOURS ) DBMS SQLsmith SQLancer SQUIRREL LEGOVIREO PostgreSQL 54,745 36,844 39,483 52,193 73,041 MySQL 39,445 41,783 43,944 74,136 81,034 MariaDB 40,394 74,364 84,945 91,083 108,238 Percona 47,293 53,784 59,384 69,411 77,032 GreatSQL 38,475 67,659 88,629 92,081 97,359 Total 220,352 274,434 316,385 378,904 436,704 Improvement 98%↑ 59%↑ 38%↑ 15%↑ – TABLE IIINUMBER OFFUNCTIONS COVERED BYDIFFERENT FUZZERS (SQLSMITH, SQLANCER, SQUIRREL ,ANDLEGO IN 24HOURS; VIREO IN 3HOURS ) DBMS SQLsmith SQLancer SQUIRREL LEGOVIREO PostgreSQL 1014 874 953 1103 1203 MySQL 938 1311 1495 1549 1953 MariaDB 674 948 1349 1620 1812 Percona 984 1023 1422 1739 1849 GreatSQL 871 1102 1394 1643 1920 Total 4481 5258 6613 7654 8737 Improvement 95%↑ 66%↑ 32%↑ 14%↑ – The primary reason for the improved coverage achieved by VIREO is its human-in-the-loop fuzzing guided by visualization.",
+      "context_before": "by each fuzzer and reran them to uniform branch coverage and covered functions. Unique bugs were identified through call stack comparison and manual analysis. 2413 Coverage. Table II presents the branch coverage results, showing that VIREO achieved in 3 hours results that surpassed the 24-hour runs of SQLsmith, SQLancer, SQUIRREL, and LEGO by 98%, 59%, 38%, and 15%, respectively. Similarly, Table III presents the functions covered, showing that VIREO achieved in 3 hours results that surpassed the 24-hour runs of SQLsmith, SQLancer, SQUIRREL, and LEGO by 95%, 66%, 32%, and 14%, respectively.",
+      "context_after": "SQLancer and SQLsmith rely primarily on predefined patterns to generate queries. SQLancer consistently produces queries based on fixed grammar rules, restricting its exploration depth, while SQLsmith predominantly generates only SELECT statements to maintain semantic correctness, severely constraining its exploration breadth. SQUIRREL and LEGO, on the other hand, use mutation-based approaches guided by generic coverage feedback. LEGO learns and mutates SQL statement sequences rather than individual statements, enabling it to perform better. Although they improve semantic and syntactic correct",
+      "section": "D Compared to Existing DBMS Testing Techniques",
+      "page": 10,
+      "char_offset": 51704,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer and SQLsmith rely primarily on predefined patterns to generate queries.",
+      "context_before": "97,359 Total 220,352 274,434 316,385 378,904 436,704 Improvement 98%↑ 59%↑ 38%↑ 15%↑ – TABLE IIINUMBER OFFUNCTIONS COVERED BYDIFFERENT FUZZERS (SQLSMITH, SQLANCER, SQUIRREL ,ANDLEGO IN 24HOURS; VIREO IN 3HOURS ) DBMS SQLsmith SQLancer SQUIRREL LEGOVIREO PostgreSQL 1014 874 953 1103 1203 MySQL 938 1311 1495 1549 1953 MariaDB 674 948 1349 1620 1812 Percona 984 1023 1422 1739 1849 GreatSQL 871 1102 1394 1643 1920 Total 4481 5258 6613 7654 8737 Improvement 95%↑ 66%↑ 32%↑ 14%↑ – The primary reason for the improved coverage achieved by VIREO is its human-in-the-loop fuzzing guided by visualization.",
+      "context_after": "SQLancer consistently produces queries based on fixed grammar rules, restricting its exploration depth, while SQLsmith predominantly generates only SELECT statements to maintain semantic correctness, severely constraining its exploration breadth. SQUIRREL and LEGO, on the other hand, use mutation-based approaches guided by generic coverage feedback. LEGO learns and mutates SQL statement sequences rather than individual statements, enabling it to perform better. Although they improve semantic and syntactic correctness relative to purely random fuzzers, their guidance is inherently coarse-grain",
+      "section": "D Compared to Existing DBMS Testing Techniques",
+      "page": 10,
+      "char_offset": 52680,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer consistently produces queries based on fixed grammar rules, restricting its exploration depth, while SQLsmith predominantly generates only SELECT statements to maintain semantic correctness, severely constraining its exploration breadth.",
+      "context_before": "15%↑ – TABLE IIINUMBER OFFUNCTIONS COVERED BYDIFFERENT FUZZERS (SQLSMITH, SQLANCER, SQUIRREL ,ANDLEGO IN 24HOURS; VIREO IN 3HOURS ) DBMS SQLsmith SQLancer SQUIRREL LEGOVIREO PostgreSQL 1014 874 953 1103 1203 MySQL 938 1311 1495 1549 1953 MariaDB 674 948 1349 1620 1812 Percona 984 1023 1422 1739 1849 GreatSQL 871 1102 1394 1643 1920 Total 4481 5258 6613 7654 8737 Improvement 95%↑ 66%↑ 32%↑ 14%↑ – The primary reason for the improved coverage achieved by VIREO is its human-in-the-loop fuzzing guided by visualization. SQLancer and SQLsmith rely primarily on predefined patterns to generate queries.",
+      "context_after": "SQUIRREL and LEGO, on the other hand, use mutation-based approaches guided by generic coverage feedback. LEGO learns and mutates SQL statement sequences rather than individual statements, enabling it to perform better. Although they improve semantic and syntactic correctness relative to purely random fuzzers, their guidance is inherently coarse-grained, which tends to repeatedly focus on readily accessible code regions. This results in limited exploration of deeper or less obvious functional paths within the DBMS. In contrast, VIREO combines human expertise with visualization to identify test",
+      "section": "D Compared to Existing DBMS Testing Techniques",
+      "page": 10,
+      "char_offset": 52761,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The results indicate that VIREO detected 9, 9, 8, and 7 more unique bugs than SQLsmith, SQLancer, SQUIRREL, and LEGO, respectively.",
+      "context_before": "ration of deeper or less obvious functional paths within the DBMS. In contrast, VIREO combines human expertise with visualization to identify testing boundaries and prioritize target modules. Based on this guidance, test engineers adjust the fuzzing engine and configurations, enabling VIREO to generate and explore SQL constructs tailored to the selected modules. This process expands the explored execution paths and allows VIREO to exercise code regions that other fuzzers often overlook, resulting in higher coverage. Unique Bugs. Table IV shows the number of unique bugs triggered by each fuzzer.",
+      "context_after": "TABLE IVNUMBER OFBUGS COVERED BYDIFFERENT FUZZERS (SQLSMITH, SQLANCER, SQUIRREL ,ANDLEGO IN 24HOURS; VIREO IN 3HOURS ) DBMS SQLsmith SQLancer SQUIRREL LEGOVIREO PostgreSQL 0 0 1 0 1 MySQL 1 0 0 1 1 MariaDB 0 1 1 2 3 Percona 1 1 0 1 3 GreatSQL 0 0 1 0 3 Total 2 2 3 4 11 Increment 9↑ 9↑ 8↑ 7↑ – Similarly, SQLsmith restricted itself predominantly to generating SELECT statements, thus achieving semantic correctness at the expense of exploration breadth, resulting in limited bug detection. SQLancer is primarily designed to detect logic bugs through predefined rule-based queries. It has limited eff",
+      "section": "D Compared to Existing DBMS Testing Techniques",
+      "page": 10,
+      "char_offset": 54062,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "TABLE IVNUMBER OFBUGS COVERED BYDIFFERENT FUZZERS (SQLSMITH, SQLANCER, SQUIRREL ,ANDLEGO IN 24HOURS; VIREO IN 3HOURS ) DBMS SQLsmith SQLancer SQUIRREL LEGOVIREO PostgreSQL 0 0 1 0 1 MySQL 1 0 0 1 1 MariaDB 0 1 1 2 3 Percona 1 1 0 1 3 GreatSQL 0 0 1 0 3 Total 2 2 3 4 11 Increment 9↑ 9↑ 8↑ 7↑ – Similarly, SQLsmith restricted itself predominantly to generating SELECT statements, thus achieving semantic correctness at the expense of exploration breadth, resulting in limited bug detection.",
+      "context_before": "identify testing boundaries and prioritize target modules. Based on this guidance, test engineers adjust the fuzzing engine and configurations, enabling VIREO to generate and explore SQL constructs tailored to the selected modules. This process expands the explored execution paths and allows VIREO to exercise code regions that other fuzzers often overlook, resulting in higher coverage. Unique Bugs. Table IV shows the number of unique bugs triggered by each fuzzer. The results indicate that VIREO detected 9, 9, 8, and 7 more unique bugs than SQLsmith, SQLancer, SQUIRREL, and LEGO, respectively.",
+      "context_after": "SQLancer is primarily designed to detect logic bugs through predefined rule-based queries. It has limited effectiveness in triggering unique crashes or serious vulnerabilities in current DBMS versions. Similarly, SQLsmith focuses mainly on generating SELECT statements, prioritizing semantic correctness over exploration breadth, which limits its bug detection capabilities. SQUIRREL and LEGO rely heavily on general coverage feedback, which leads to missing complex module-specific bugs due to insufficient guidance toward deeper code regions. By leveraging visualization guidance, VIREO was able t",
+      "section": "D Compared to Existing DBMS Testing Techniques",
+      "page": 10,
+      "char_offset": 54194,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer is primarily designed to detect logic bugs through predefined rule-based queries.",
+      "context_before": "that VIREO detected 9, 9, 8, and 7 more unique bugs than SQLsmith, SQLancer, SQUIRREL, and LEGO, respectively. TABLE IVNUMBER OFBUGS COVERED BYDIFFERENT FUZZERS (SQLSMITH, SQLANCER, SQUIRREL ,ANDLEGO IN 24HOURS; VIREO IN 3HOURS ) DBMS SQLsmith SQLancer SQUIRREL LEGOVIREO PostgreSQL 0 0 1 0 1 MySQL 1 0 0 1 1 MariaDB 0 1 1 2 3 Percona 1 1 0 1 3 GreatSQL 0 0 1 0 3 Total 2 2 3 4 11 Increment 9↑ 9↑ 8↑ 7↑ – Similarly, SQLsmith restricted itself predominantly to generating SELECT statements, thus achieving semantic correctness at the expense of exploration breadth, resulting in limited bug detection.",
+      "context_after": "It has limited effectiveness in triggering unique crashes or serious vulnerabilities in current DBMS versions. Similarly, SQLsmith focuses mainly on generating SELECT statements, prioritizing semantic correctness over exploration breadth, which limits its bug detection capabilities. SQUIRREL and LEGO rely heavily on general coverage feedback, which leads to missing complex module-specific bugs due to insufficient guidance toward deeper code regions. By leveraging visualization guidance, VIREO was able to systematically identify and target historically buggy and under-tested modules, thus unco",
+      "section": "D Compared to Existing DBMS Testing Techniques",
+      "page": 10,
+      "char_offset": 54684,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M9",
+        "M12"
+      ],
+      "describes_as_state_of_the_art": [
+        "M9"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:37:26Z",
+    "is_model_written": true,
+    "summary": "VIREO puts a person in the DBMS fuzzing loop. Its premise is that automated fuzzers plateau: after a few hours the coverage curves of SQLsmith, SQLancer, SQUIRREL and LEGO on MariaDB flatten, leaving much of the system untested. VIREO visualises where coverage has stalled and uses an LLM to turn a tester's guidance into new test cases. In three hours it exceeded the 24-hour branch coverage of all four baselines and found 11 unique bugs to their 2-4.",
+    "narrative": "SQLancer is one of four baselines, named among the state-of-the-art DBMS fuzzers VIREO measures against, and run with its default PQS oracle since the comparison is about crashes rather than logic bugs. Results are reported across five systems in both branch coverage and unique bugs. The paper's diagnosis is specific: SQLancer generates queries from fixed grammar rules tied to its oracles, which bounds how deep its exploration goes, and it does not vary configuration options such as --skip-grant-tables, so a class of bugs stays out of reach. SQLancer's oracles are also cited as the generation-based line of work VIREO departs from.",
+    "roles": {
+      "M1": "background",
+      "M2": "motivation",
+      "M3": "result_comparison",
+      "M4": "background",
+      "M5": "definition",
+      "M6": "motivation",
+      "M7": "motivation",
+      "M8": "motivation",
+      "M9": "state_of_the_art",
+      "M10": "baseline",
+      "M11": "baseline",
+      "M12": "baseline",
+      "M13": "result_comparison",
+      "M14": "result_comparison",
+      "M15": "result_comparison",
+      "M16": "result_comparison",
+      "M17": "result_comparison",
+      "M18": "result_comparison",
+      "M19": "result_comparison",
+      "M20": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is run as a baseline. VIREO's own contribution is a visualisation and LLM-guided loop, not anything built on the SQLancer codebase."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer oracle is extended; VIREO changes how test cases are chosen, not what makes an outcome wrong."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M9",
+          "M10",
+          "M13",
+          "M18"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M9",
+            "sentence": "We tried our best to compare VIREO against state-of-the-art DBMS fuzzers, including SQLsmith [61], SQLancer [57], SQUIRREL [70], and LEGO [34].",
+            "section": "A Evaluation Setup",
+            "page": 8
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "Although SQLancer was designed for logic bugs, it can also expose crashes and is evaluated using its default PQS oracle [60].",
+            "section": "A Evaluation Setup",
+            "page": 8
+          },
+          {
+            "mention_id": "M13",
+            "sentence": "Table II presents the branch coverage results, showing that VIREO achieved in 3 hours results that surpassed the 24-hour runs of SQLsmith, SQLancer, SQUIRREL, and LEGO by 98%, 59%, 38%, and 15%, respectively.",
+            "section": "D Compared to Existing DBMS Testing Techniques",
+            "page": 10
+          },
+          {
+            "mention_id": "M18",
+            "sentence": "The results indicate that VIREO detected 9, 9, 8, and 7 more unique bugs than SQLsmith, SQLancer, SQUIRREL, and LEGO, respectively.",
+            "section": "D Compared to Existing DBMS Testing Techniques",
+            "page": 10
+          }
+        ],
+        "reasoning": "M9 names SQLancer among the fuzzers VIREO is compared against, M10 records that it was run with its default PQS oracle, and M13 and M18 report the branch-coverage and unique-bug outcomes.",
+        "techniques": [
+          "pqs"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M9"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M9",
+            "sentence": "We tried our best to compare VIREO against state-of-the-art DBMS fuzzers, including SQLsmith [61], SQLancer [57], SQUIRREL [70], and LEGO [34].",
+            "section": "A Evaluation Setup",
+            "page": 8
+          }
+        ],
+        "reasoning": "M9 introduces SQLancer as one of the state-of-the-art DBMS fuzzers the evaluation compares against."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icde65706_2026_00224.json b/_data/papers/paper_doi_10_1109_icde65706_2026_00224.json
new file mode 100644
index 0000000..eb5bb50
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icde65706_2026_00224.json
@@ -0,0 +1,1511 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:28:53Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icde65706.2026.00224",
+    "title": "A Set-Theoretic Approach to Detecting Logic Bugs in DBMS Inner Join Optimizations",
+    "authors": [
+      "Ce Lyu",
+      "Changzheng Wei",
+      "Yanhao Wang",
+      "Jie Liang",
+      "Li Lin",
+      "Hanghang Wu",
+      "Minghao Zhao",
+      "Ying Yan",
+      "Aoying Zhou"
+    ],
+    "year": 2026,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde65706.2026.00224",
+    "arxiv_id": "2606.23294",
+    "s2_paper_id": "f959eadc36d89bdfc3add245defa2b3b50c40954",
+    "url": "https://doi.org/10.1109/icde65706.2026.00224",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2606.23294",
+      "retrieved_at": "2026-09-09T01:28:53Z",
+      "chars": 75797,
+      "content_sha256": "sha256:c5a6bf79ad16e2ae3b478a94a1c2921899df26111e3e46e8a9bc527f804c7ed3"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 14,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2313
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 11440
+      },
+      {
+        "number": "A",
+        "title": "Set and Multiset Semantics in SQL",
+        "start": 11455
+      },
+      {
+        "number": "B",
+        "title": "Query Optimization and Rewrite Logic",
+        "start": 12265
+      },
+      {
+        "number": "C",
+        "title": "Metamorphic Testing",
+        "start": 13042
+      },
+      {
+        "number": "III",
+        "title": "METHODOLOGY",
+        "start": 14193
+      },
+      {
+        "number": "A",
+        "title": "Framework Overview",
+        "start": 15113
+      },
+      {
+        "number": "B",
+        "title": "Symmetric Join Transformation",
+        "start": 18918
+      },
+      {
+        "number": "C",
+        "title": "Asymmetric Difference Transformation",
+        "start": 21930
+      },
+      {
+        "number": "D",
+        "title": "Symmetric Difference Transformation",
+        "start": 24935
+      },
+      {
+        "number": "E",
+        "title": "Database and Query Generation",
+        "start": 27533
+      },
+      {
+        "number": "IV",
+        "title": "THEORETICALANALYSIS OFTRANSFORMATION",
+        "start": 29228
+      },
+      {
+        "number": "A",
+        "title": "Preliminaries",
+        "start": 29671
+      },
+      {
+        "number": "B",
+        "title": "Proofs of Equivalence under Set Semantics",
+        "start": 31105
+      },
+      {
+        "number": "C",
+        "title": "Proofs of Equivalence under Multiset Semantics",
+        "start": 33433
+      },
+      {
+        "number": "V",
+        "title": "EXPERIMENTALEVALUATION",
+        "start": 35175
+      },
+      {
+        "number": "A",
+        "title": "Evaluation Setup",
+        "start": 35643
+      },
+      {
+        "number": "C",
+        "title": "Diversity of Detected Logic Bugs",
+        "start": 38922
+      },
+      {
+        "number": "E",
+        "title": "Comparison to Existing DBMS Testing Approaches",
+        "start": 45587
+      },
+      {
+        "number": "VI",
+        "title": "DISCUSSION",
+        "start": 54102
+      },
+      {
+        "number": "VII",
+        "title": "RELATEDWORK",
+        "start": 58879
+      },
+      {
+        "number": "VIII",
+        "title": "CONCLUSION",
+        "start": 62881
+      },
+      {
+        "number": "O",
+        "title": "Papaemmanouil, and N. Tatbul, “Neo: A learned query optimizer,”",
+        "start": 65336
+      },
+      {
+        "number": "T",
+        "title": "G. Price, “Access path selection in a relational database management",
+        "start": 68732
+      },
+      {
+        "number": "J",
+        "title": "Sun, “PATA: Fuzzing with path aware taint analysis,” in2022 IEEE",
+        "start": 74357
+      },
+      {
+        "number": "J",
+        "title": "Wang, and J. Li, “Sequence-oriented DBMS fuzzing,” in2023 IEEE",
+        "start": 75448
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "D. D. Chamberlin, “Relational data-base management systems,”ACM Comput. Surv., vol. 8, no. 1, pp. 43–66, 1976.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "M. Stonebraker, S. Madden, and P. Dubey, “Intel “big data” science and technology center vision and execution plan,”SIGMOD Rec., vol. 42, no. 1, pp. 44–49, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "M. Stonebraker and A. Pavlo, “What goes around comes around... and around...”SIGMOD Rec., vol. 53, no. 2, pp. 21–37, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "R. Hai, S.-H. Hung, T. Coopmans, T. Littau, and F. Geerts, “Quantum data management in the NISQ era,”Proc. VLDB Endow., vol. 18, no. 6, pp. 1720–1729, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "B. Ding, S. Das, W. Wu, S. Chaudhuri, and V. Narasayya, “Plan stitch: Harnessing the best of many plans,”Proc. VLDB Endow., vol. 11, no. 10, pp. 1123–1136, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "R. Marcus, P. Negi, H. Mao, C. Zhang, M. Alizadeh, T. Kraska, O. Papaemmanouil, and N. Tatbul, “Neo: A learned query optimizer,” Proc. VLDB Endow., vol. 12, no. 11, pp. 1705–1718, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Y. Zhang, Y. Chronis, J. M. Patel, and T. Rekatsinas, “Simple adaptive query processing vs. learned query optimizers: Observations and analysis,”Proc. VLDB Endow., vol. 16, no. 11, pp. 2962–2975, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "T. Neumann and B. Radke, “Adaptive optimization of very large join queries,” inProceedings of the 2018 International Conference on Management of Data, 2018, pp. 677–692.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "X. Zhang and A. Eldawy, “Spatial query optimization with learning,” Proc. VLDB Endow., vol. 17, no. 12, pp. 4245–4248, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "J. Zhao, K. Su, Y. Yang, X. Yu, P. Koutris, and H. Zhang, “Debunking the myth of join ordering: Toward robust SQL analytics,” 3, no. 3, pp. 146:1–146:28, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "A. Birler, A. Kemper, and T. Neumann, “Robust join processing with diamond hardened joins,”Proc. VLDB Endow., vol. 17, no. 11, pp. 3215– 3228, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "X. Hu, “Output-optimal algorithms for join-aggregate queries,” 3, no. 2, pp. 104:1–104:27, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "H. Kalumin and A. Deshpande, “Optimizing queries with many-to-many joins,” in (ICDE), 2025, pp. 3668–3681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Wikipedia, “Test oracle,” https://en.wikipedia.org/wiki/Test oracle, 2025, accessed: 2025-10-27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "W. E. Howden, “Theoretical and empirical studies of program testing,” IEEE Trans. Softw. Eng., no. 4, pp. 293–298, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "A. Seltenreich, “SQLSmith,” https://github.com/anse1/sqlsmith, 2025, accessed: 2025-10-27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "M. Zalewski, “American fuzzy lop,” https://github.com/google/AFL, 2025, accessed: 2025-10-27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "D. R. Slutz, “Massive stochastic testing of SQL,” inProceedings of 24rd International Conference on Very Large Data Bases, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "——, “Finding bugs in database systems via query partitioning,”Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 211:1–211:30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Y. Fu, Z. Wu, Y. Zhang, J. Liang, J. Fu, Y. Jiang, S. Li, and X. Liao, “THANOS: DBMS bug detection via storage engine rotation based differential testing,” in on Software Engineering (ICSE), 2024, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” inProceedings of the 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 23,
+      "text": "J. Song, W. Dou, Y. Zheng, Y. Gao, Z. Cui, W. Wang, and J. Wei, “Detecting schema-related logic bugs in relational DBMSs via equivalent database construction,”Proc. VLDB Endow., vol. 18, no. 7, pp. 2281–2294, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "X. Tang, S. Wu, D. Zhang, F. Li, and G. Chen, “Detecting logic bugs of join optimizations in DBMS,” 1, no. 1, pp. 55:1–55:26, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "J. Ba and M. Rigger, “Keep it simple: Testing databases via differential query plans,” 2, no. 3, pp. 188:1–188:26, 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 26,
+      "text": "PingCAP, “TiDB,” https://pingcap.com/tidb, 2025, accessed: 2025-1027.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "M. Rigger, “SQLancer,” https://github.com/sqlancer/sqlancer, 2025, accessed: 2025-10-27.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "E. F. Codd, “A relational model of data for large shared data banks,” Commun. ACM, vol. 13, no. 6, pp. 377–387, 1970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "P. G. Selinger, M. M. Astrahan, D. D. Chamberlin, R. A. Lorie, and T. G. Price, “Access path selection in a relational database management system,” inProceedings of the 1979 ACMSIGMOD International Conference on Management of Data, 1979, pp. 23–34.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "G. Graefe and W. J. McKenna, “The volcano optimizer generator: Extensibility and efficient search,” inProceedings of the Ninth International Conference on Data Engineering, 1993, pp. 209–218.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "S. Chaudhuri, “An overview of query optimization in relational systems,” inProceedings of the Seventeenth ACMSIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems, 1998, pp. 34–43.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "V. Leis, B. Radke, A. Gubichev, A. Mirchev, P. A. Boncz, A. Kemper, and T. Neumann, “Query optimization through the looking glass, and what we found running the join order benchmark,”VLDB J., vol. 27, no. 5, pp. 643–668, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "T. Y. Chen, S.-C. Cheung, and S.-M. Yiu, “Metamorphic testing: A new approach for generating next test cases,”arXiv:2002.12543, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "T. Y. Chen, F.-C. Kuo, H. Liu, P.-L. Poon, D. Towey, T. H. Tse, and Z. Q. Zhou, “Metamorphic testing: A review of challenges and opportunities,” ACM Comput. Surv., vol. 51, no. 1, pp. 4:1–4:27, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "S. Segura, G. Fraser, A. B. S ´anchez, and A. R. Cort ´es, “A survey on metamorphic testing,”IEEE Trans. Softw. Eng., vol. 42, no. 9, pp. 805– 824, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "M. Winslett and V. Braganholo, “Richard Hipp speaks out on SQLite,” SIGMOD Rec., vol. 48, no. 2, pp. 39–46, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "J. Ba and M. Rigger, “CERT: Finding performance issues in database systems through the lens of cardinality estimation,” inProceedings of the 46th IEEE/ACM International Conference on Software Engineering, 2024, pp. 133:1–133:13.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "C. Binnig, D. Kossmann, E. Lo, and M. T. ¨Ozsu, “QAGen: generating query-aware test databases,” inProceedings of the 2007 ACMSIGMOD International Conference on Management of Data, 2007, pp. 341–352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "N. Bruno and S. Chaudhuri, “Flexible database generators,” inProceedings of the 31st International Conference on Very Large Data Bases, 2005, pp. 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "J. Gray, P. Sundaresan, S. Englert, K. Baclawski, and P. J. Weinberger, “Quickly generating billion-record synthetic databases,” inProceedings of the 1994 ACMSIGMOD International Conference on Management of Data, 1994, pp. 243–252.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "K. Houkjær, K. Torp, and R. Wind, “Simple and realistic data generation,” inProceedings of the 32nd International Conference on Very Large Data Bases, 2006, pp. 1243–1246.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "S. A. Khalek, B. Elkarablieh, Y. O. Laleye, and S. Khurshid, “Queryaware test generation using a relational constraint solver,” in2008 23rd IEEE/ACM International Conference on Automated Software Engineering, 2008, pp. 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "H. Bati, L. Giakoumakis, S. Herbert, and A. Surna, “A genetic approach for random testing of database systems,” inProceedings of the 33rd International Conference on Very Large Data Bases, 2007, pp. 1243– 1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "N. Bruno, S. Chaudhuri, and D. Thomas, “Generating queries with cardinality constraints for DBMS testing,”IEEE Trans. Knowl. Data Eng., vol. 18, no. 12, pp. 1721–1725, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “Apollo: Automatic detection and diagnosis of performance regressions in database systems,” Proc. VLDB Endow., vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "C. Mishra, N. Koudas, and C. Zuzarte, “Generating targeted queries for database testing,” inProceedings of the 2008 ACMSIGMOD International Conference on Management of Data, 2008, pp. 499–510.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "M. Poess and J. M. Stephens Jr, “Generating thousand benchmark queries in seconds,” inProceedings of the 30th International Conference on Very Large Data Bases, 2004, pp. 1045–1053.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "M. Vartak, V. Raghavan, and E. A. Rundensteiner, “QRelX: generating meaningful queries that provide cardinality assurance,” inProceedings of the 2010 ACMSIGMOD International Conference on Management of data, 2010, pp. 1215–1218.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Z. Gu, M. A. Soliman, and F. M. Waas, “Testing the accuracy of query optimizers,” inProceedings of the Fifth International Workshop on Testing Database Systems, 2012, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "S. Abdul Khalek and S. Khurshid, “Automated SQL query generation for systematic testing of database engines,” inProceedings of the 25th IEEE/ACM International Conference on Automated Software Engineering, 2010, pp. 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "E. Lo, C. Binnig, D. Kossmann, M. Tamer ¨Ozsu, and W.-K. Hon, “A framework for testing DBMS features,”VLDB J., vol. 19, no. 2, pp. 203–230, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "J. Yan, Q. Jin, S. Jain, S. D. Viglas, and A. Lee, “Snowtrail: Testing with production queries on a cloud database,” inProceedings of the Workshop on Testing Database Systems, 2018, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “SQUIRREL: Testing database management systems with language validity and coverage feedback,” inProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "C. Aschermann, S. Schumilo, T. Blazytko, R. Gawlik, and T. Holz, “REDQUEEN: Fuzzing with input-to-state correspondence,” in26th Annual Network and Distributed System Security Symposium (NDSS), 2019, pp. 1–15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Y. Chen, Y. Jiang, F. Ma, J. Liang, M. Wang, C. Zhou, X. Jiao, and Z. Su, “EnFuzz: Ensemble fuzzing with seed synchronization among diverse fuzzers,” in28th USENIX Security Symposium (USENIX Security 19), 2019, pp. 1967–1983.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "J. Liang, Y. Jiang, M. Wang, X. Jiao, Y. Chen, H. Song, and K.-K. R. Choo, “DeepFuzzer: Accelerated deep greybox fuzzing,”IEEE Trans. Dependable Secur. Comput., vol. 18, no. 6, pp. 2675–2688, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "J. Liang, M. Wang, C. Zhou, Z. Wu, Y. Jiang, J. Liu, Z. Liu, and J. Sun, “PATA: Fuzzing with path aware taint analysis,” in2022 IEEE Symposium on Security and Privacy (SP), 2022, pp. 1–17.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "M. Wang, J. Liang, C. Zhou, Y. Jiang, R. Wang, C. Sun, and J. Sun, “RIFF: Reduced instruction footprint for Coverage-Guided fuzzing,” in 2021 USENIX Annual Technical Conference (USENIX ATC 21), 2021, pp. 147–159.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang, and Y. Jiang, “Industry practice of coverage-guided enterprise-level DBMS fuzzing,” in neering: Software Engineering in Practice (ICSE-SEIP), 2021, pp. 328– 337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Z. Wu, J. Liang, M. Wang, C. Zhou, and Y. Jiang, “Unicorn: detect runtime errors in time-series databases with hybrid input synthesis,” in Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, 2022, pp. 251–262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "I. Yun, S. Lee, M. Xu, Y. Jang, and T. Kim, “QSYM: A practical concolic execution engine tailored for hybrid fuzzing,” in27th USENIX Security Symposium (USENIX Security 18), 2018, pp. 745–761.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "J. Liang, Y. Chen, Z. Wu, J. Fu, M. Wang, Y. Jiang, X. Huang, T. Chen, J. Wang, and J. Li, “Sequence-oriented DBMS fuzzing,” in2023 IEEE 39th International Conference on Data Engineering (ICDE), 2023, pp. 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammar-free DBMS fuzzing,” inProceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, 2022, pp. 49:1–49:12.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 19,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "——, “Finding bugs in database systems via query partitioning,”Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 211:1–211:30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 22,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” inProceedings of the 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 25,
+      "text": "J. Ba and M. Rigger, “Keep it simple: Testing databases via differential query plans,” 2, no. 3, pp. 188:1–188:26, 2024.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "M. Rigger, “SQLancer,” https://github.com/sqlancer/sqlancer, 2025, accessed: 2025-10-27.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "J. Ba and M. Rigger, “CERT: Finding performance issues in database systems through the lens of cardinality estimation,” inProceedings of the 46th IEEE/ACM International Conference on Software Engineering, 2024, pp. 133:1–133:13.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Another approach is pivoted query synthesis (PQS) [19], which validates query results by constructing auxiliary queries centered on a specific pivot row.",
+      "context_before": "uncover crash and performance defects, they generally lack the ability to detect logic bugs. To address the challenge of detecting logic bugs, several approaches have been proposed. Differential testing, for example, identifies potential bugs by comparing the results of the same query across different DBMSs or versions of the same DBMS. While effective at detecting inconsistencies between implementations, its applicability is often limited to the common core of SQL functionalities [18], as different DBMSs vary in their support for the SQL standard and frequently provide proprietary extensions.",
+      "context_after": "Although PQS can be effective in certain scenarios, it requires substantial engineering effort and is less capable of capturing set-level inconsistencies, such as duplicates or missing tuples. Recently, metamorphic tests have been increasingly used for DBMS testing [20]–[23]. Representative frameworks include ternary logic partitioning (TLP) [20] and non-optimizing reference engine construction (NoREC) [22]. These approaches transform a query into semantically equivalent variants, either through predicate partitioning or by bypassing the optimizer’s execution path, to check for inconsistencie",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 4696,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Although PQS can be effective in certain scenarios, it requires substantial engineering effort and is less capable of capturing set-level inconsistencies, such as duplicates or missing tuples.",
+      "context_before": "roaches have been proposed. Differential testing, for example, identifies potential bugs by comparing the results of the same query across different DBMSs or versions of the same DBMS. While effective at detecting inconsistencies between implementations, its applicability is often limited to the common core of SQL functionalities [18], as different DBMSs vary in their support for the SQL standard and frequently provide proprietary extensions. Another approach is pivoted query synthesis (PQS) [19], which validates query results by constructing auxiliary queries centered on a specific pivot row.",
+      "context_after": "Recently, metamorphic tests have been increasingly used for DBMS testing [20]–[23]. Representative frameworks include ternary logic partitioning (TLP) [20] and non-optimizing reference engine construction (NoREC) [22]. These approaches transform a query into semantically equivalent variants, either through predicate partitioning or by bypassing the optimizer’s execution path, to check for inconsistencies. Transformed query synthesis (TQS) [24] utilizes optimization hints to verify query results against a ground truth. Recently, differential query planning (DQP) [25] adopts a similar hint-base",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 4850,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[20]–[23]",
+      "technique": "tlp",
+      "sentence": "Recently, metamorphic tests have been increasingly used for DBMS testing [20]–[23].",
+      "context_before": "fective at detecting inconsistencies between implementations, its applicability is often limited to the common core of SQL functionalities [18], as different DBMSs vary in their support for the SQL standard and frequently provide proprietary extensions. Another approach is pivoted query synthesis (PQS) [19], which validates query results by constructing auxiliary queries centered on a specific pivot row. Although PQS can be effective in certain scenarios, it requires substantial engineering effort and is less capable of capturing set-level inconsistencies, such as duplicates or missing tuples.",
+      "context_after": "Representative frameworks include ternary logic partitioning (TLP) [20] and non-optimizing reference engine construction (NoREC) [22]. These approaches transform a query into semantically equivalent variants, either through predicate partitioning or by bypassing the optimizer’s execution path, to check for inconsistencies. Transformed query synthesis (TQS) [24] utilizes optimization hints to verify query results against a ground truth. Recently, differential query planning (DQP) [25] adopts a similar hint-based strategy but focuses on detecting inconsistencies between different physical plans",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 5043,
+      "cited_reference": {
+        "number": 20,
+        "text": "——, “Finding bugs in database systems via query partitioning,”Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 211:1–211:30, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "ternary logic partitioning",
+      "technique": "tlp",
+      "sentence": "Representative frameworks include ternary logic partitioning (TLP) [20] and non-optimizing reference engine construction (NoREC) [22].",
+      "context_before": "ften limited to the common core of SQL functionalities [18], as different DBMSs vary in their support for the SQL standard and frequently provide proprietary extensions. Another approach is pivoted query synthesis (PQS) [19], which validates query results by constructing auxiliary queries centered on a specific pivot row. Although PQS can be effective in certain scenarios, it requires substantial engineering effort and is less capable of capturing set-level inconsistencies, such as duplicates or missing tuples. Recently, metamorphic tests have been increasingly used for DBMS testing [20]–[23].",
+      "context_after": "These approaches transform a query into semantically equivalent variants, either through predicate partitioning or by bypassing the optimizer’s execution path, to check for inconsistencies. Transformed query synthesis (TQS) [24] utilizes optimization hints to verify query results against a ground truth. Recently, differential query planning (DQP) [25] adopts a similar hint-based strategy but focuses on detecting inconsistencies between different physical plans for the same query. We aim to present a meta-arXiv:2606.23294v2 [cs.DB] 25 Jun 2026 morphic testing scheme that efficiently and effect",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 5127,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "Recently, differential query planning (DQP) [25] adopts a similar hint-based strategy but focuses on detecting inconsistencies between different physical plans for the same query.",
+      "context_before": "f capturing set-level inconsistencies, such as duplicates or missing tuples. Recently, metamorphic tests have been increasingly used for DBMS testing [20]–[23]. Representative frameworks include ternary logic partitioning (TLP) [20] and non-optimizing reference engine construction (NoREC) [22]. These approaches transform a query into semantically equivalent variants, either through predicate partitioning or by bypassing the optimizer’s execution path, to check for inconsistencies. Transformed query synthesis (TQS) [24] utilizes optimization hints to verify query results against a ground truth.",
+      "context_after": "We aim to present a meta-arXiv:2606.23294v2 [cs.DB] 25 Jun 2026 morphic testing scheme that efficiently and effectively detects logic bugs in DBMS join optimization. Our key insight is that an INNER/NATURAL JOIN) can be semantically expressed as a combination of multiple set intersection operations.Such a theoretical equivalence should be strictly upheld by any standards-compliant DBMS. However, we discovered that this is often not the case. A highly illustrative motivating example is shown in Listing 1, which depicts a bug we discovered in TiDB [26]. We crafted a simple INNER JOINquery whose ON",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 5567,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "As shown in Table I, PQS, NoREC, DQP, and TQS do not support EXCEPT,INTERSECT, or UNION operations, while TLP supports only UNION(including ALL).",
+      "context_before": ">HashJoin...CARTESIAN inner join... (rows=1) The above example reveals that such optimizer-level logic bugs often stem from incorrect handling of algebraic equivalences during query rewriting. Although existing studies have achieved significant results in logic bug detection, our systematic investigation reveals that these methods still exhibit notable deficiencies in supporting equivalent transformations at the relational algebra level. Table I compares the feature coverage of typical testing approaches with respect to SQL set operations, including EXCEPT,INTERSECT,UNION, and their ALLvariants.",
+      "context_after": "These results reveal that existing approaches focus primarily on predicate partitioning or optimizer path validation, resulting in limited testing capability.TABLE ICOMPARISON OFSQLFEATURE COVERAGE OFDIFFERENT TESTING APPROACHES. HERE, “⃝”AND“×”INDICATE WHETHER AN APPROACH SUPPORTS THEFEATURE. ApproachEXCEPT (ALL)INTERSECT (ALL)UNION (ALL) PQS× × × NoREC× × × TLP× × ⃝ DQP× × × TQS× × × JoinEquiv (*)⃝ ⃝ ⃝ To address this issue, we perform an in-depth exploration of set-theoretic equivalences in join queries for test case generation. Leveraging the property that an INNER JOIN(or NATURAL JOIN) is",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 8470,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "dqp",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "ApproachEXCEPT (ALL)INTERSECT (ALL)UNION (ALL) PQS× × × NoREC× × × TLP× × ⃝ DQP× × × TQS× × × JoinEquiv (*)⃝ ⃝ ⃝ To address this issue, we perform an in-depth exploration of set-theoretic equivalences in join queries for test case generation.",
+      "context_before": "vel. Table I compares the feature coverage of typical testing approaches with respect to SQL set operations, including EXCEPT,INTERSECT,UNION, and their ALLvariants. As shown in Table I, PQS, NoREC, DQP, and TQS do not support EXCEPT,INTERSECT, or UNION operations, while TLP supports only UNION(including ALL). These results reveal that existing approaches focus primarily on predicate partitioning or optimizer path validation, resulting in limited testing capability.TABLE ICOMPARISON OFSQLFEATURE COVERAGE OFDIFFERENT TESTING APPROACHES. HERE, “⃝”AND“×”INDICATE WHETHER AN APPROACH SUPPORTS THEFEATURE.",
+      "context_after": "Leveraging the property that an INNER JOIN(or NATURAL JOIN) is semantically equivalent to a combination of set intersection operations, we design three transformation rules (§III-B, §III-C, and §III-D). These rules enable the construction of semantically equivalent query pairs, whose result sets can be compared to uncover deep logical inconsistencies in query rewriting or execution semantics. Based on this equivalence-driven query generation method, we develop a metamorphic testing framework, JoinEquiv, implemented on top of SQLancer[27] and made publicly available.1Compared with existing appro",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 8907,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[27]",
+      "technique": null,
+      "sentence": "Based on this equivalence-driven query generation method, we develop a metamorphic testing framework, JoinEquiv, implemented on top of SQLancer[27] and made publicly available.",
+      "context_before": "ION (ALL) PQS× × × NoREC× × × TLP× × ⃝ DQP× × × TQS× × × JoinEquiv (*)⃝ ⃝ ⃝ To address this issue, we perform an in-depth exploration of set-theoretic equivalences in join queries for test case generation. Leveraging the property that an INNER JOIN(or NATURAL JOIN) is semantically equivalent to a combination of set intersection operations, we design three transformation rules (§III-B, §III-C, and §III-D). These rules enable the construction of semantically equivalent query pairs, whose result sets can be compared to uncover deep logical inconsistencies in query rewriting or execution semantics.",
+      "context_after": "1Compared with existing approaches (see Table I), JoinEquiv achieves full support for all three set operations and their ALLvariants. To evaluate the effectiveness of our JoinEquiv framework, we conduct extensive tests on four production-grade DBMSs: MySQL, TiDB, DuckDB, and Percona. JoinEquiv reported 29 previously unknown logical inconsistencies. Among these, 27 have already been confirmed by developers, 14 of which involve critical bugs in query optimizer rewrite logic or executor semantic handling. Cross-oracular validation shows that the vast majority of the remaining bugs can only be dete",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 9545,
+      "cited_reference": {
+        "number": 27,
+        "text": "M. Rigger, “SQLancer,” https://github.com/sqlancer/sqlancer, 2025, accessed: 2025-10-27.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", they cannot be detected by existing metamorphic testing approaches such as TLP and DQP, thus highlighting the unique and complementary value of JoinEquiv.",
+      "context_before": "v achieves full support for all three set operations and their ALLvariants. To evaluate the effectiveness of our JoinEquiv framework, we conduct extensive tests on four production-grade DBMSs: MySQL, TiDB, DuckDB, and Percona. JoinEquiv reported 29 previously unknown logical inconsistencies. Among these, 27 have already been confirmed by developers, 14 of which involve critical bugs in query optimizer rewrite logic or executor semantic handling. Cross-oracular validation shows that the vast majority of the remaining bugs can only be detected by our intersection-equivalence transformations; i.e.",
+      "context_after": "We believe that the principled methodology, low implementation effort, and broad applicability of JoinEquiv will lead to its widespread adoption to improve the robustness of DBMSs. In summary, our main contributions in this paper include: •We propose a new metamorphic testing paradigm that bridges relational algebra and set theory, mapping INNER JOINto set-theoretic intersection to build principled logic-level oracles. •We design a structurally complete set of minimal transformation rules (SJT, ADT, and SDT) that collectively encompass all core SQL set operators. They serve as normal forms for",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 10379,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The test case generation builds upon the SQLancer framework, which provides a grammar-aware SQL query generation engine.",
+      "context_before": "ce(A\\B)∪(B\\A)(lines 12– 13). Finally, another EXCEPTnode subtracts this symmetric difference from the full union of the left and right join ASTs (line 14), producing the transformed AST that preserves the intersection semantics. E. Database and Query Generation Generating high-quality databases and queries for testing has been extensively studied and is not the primary concern of this work. JoinEquiv is compatible with any database and query generator capable of producing valid SQL queries containing INNER/NATURAL JOINoperators. Here, we provide details of our implementation for reproducibility.",
+      "context_after": "We extend SQLancer to satisfy an additional precondition that all columns are NOT NULL. Specifically, Step 1of Fig. 1 randomly generates tables and rows using CREATE TABLEand INSERTstatements under NOT NULLconstraints. Step 2 then constructs theoriginal queries, where JoinEquiv randomly selects a subset of tables from the generated schema to participate in the join. The order of table references is also randomized. Across test iterations, the roles of participating tables (e.g.,t0andt1) are permuted by a randomized generation engine. Although the transformation templates use a fixed LEFT/RIGHT JO",
+      "section": "E Database and Query Generation",
+      "page": 6,
+      "char_offset": 27905,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We extend SQLancer to satisfy an additional precondition that all columns are NOT NULL.",
+      "context_before": "ft and right join ASTs (line 14), producing the transformed AST that preserves the intersection semantics. E. Database and Query Generation Generating high-quality databases and queries for testing has been extensively studied and is not the primary concern of this work. JoinEquiv is compatible with any database and query generator capable of producing valid SQL queries containing INNER/NATURAL JOINoperators. Here, we provide details of our implementation for reproducibility. The test case generation builds upon the SQLancer framework, which provides a grammar-aware SQL query generation engine.",
+      "context_after": "Specifically, Step 1of Fig. 1 randomly generates tables and rows using CREATE TABLEand INSERTstatements under NOT NULLconstraints. Step 2 then constructs theoriginal queries, where JoinEquiv randomly selects a subset of tables from the generated schema to participate in the join. The order of table references is also randomized. Across test iterations, the roles of participating tables (e.g.,t0andt1) are permuted by a randomized generation engine. Although the transformation templates use a fixed LEFT/RIGHT JOINstructure, symmetric variants are implicitly instantiated through this randomized tab",
+      "section": "E Database and Query Generation",
+      "page": 6,
+      "char_offset": 28026,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Such a table permutation, combined with SQLancer’s randomized population of ONpredicates,WHEREclauses, and SELECT list expressions, ensures that the framework explores a diverse range of symmetric and asymmetric execution plans.",
+      "context_before": "er of table references is also randomized. Across test iterations, the roles of participating tables (e.g.,t0andt1) are permuted by a randomized generation engine. Although the transformation templates use a fixed LEFT/RIGHT JOINstructure, symmetric variants are implicitly instantiated through this randomized table binding. For instance, expressions such ast1 LEFT JOIN t0naturally arise, which are algebraically equivalent tot0 RIGHT JOIN t1. As a result, symmetric forms of SJT, ADT, and SDT are repeatedly exercised over long-running testing without explicitly enumerating all symmetric variants.",
+      "context_after": "IV. THEORETICALANALYSIS OFTRANSFORMATION EQUIVALENCE The correctness of our fuzzing methodology relies on the semantic equivalence of the query transformations we employ. This section provides a theoretical analysis of the equivalence of the three transformation rules: SJT, ADT, and SDT. We rigorously examine their validity under both set semantics (corresponding to SELECT DISTINCT) and multiset semantics (corresponding to standard SELECT). A. Preliminaries To ensure the clarity of our proofs, we operate under the assumption that all columns in the involved tables are defined as NOT NULL. This a",
+      "section": "E Database and Query Generation",
+      "page": 6,
+      "char_offset": 29000,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Within each thread, for every INNER JOINquery, JoinEquiv, TLP, and DQP are applied sequentially under the same system TABLE IISUMMARY OFLOGICBUGREPORTS ANDVERIFICATIONSTATUS INDIFFERENTRDBMSS RDBMS Reported Verified Fixed Intended Component Severity Identifier MySQL 10 8 1 2 Optimizer (8) Critical (8)Bug#118544, Bug#118684, Bug#118710, Bug#118857, Bug#118858, Bug#118949, Bug#119032, Bug#119059 TiDB 13 13 4 0Planner (11) Execution (2)Critical (3)#62380, #62444, #62456, #62459, #62460, #62644, #62645, #62689, #63596, #63601, #63635, #63636, #63736 Percona 3 3 0 0 Optimizer (3) Critical (3)PS-10124, PS-10127, DISMYSQL-535 DuckDB 3 3 3 0 – –#20483, #20486, #20608 Total29 27 8 2 – Critical (14) – Note:For DuckDB, component and severity were not provided, so they are marked as “–”.",
+      "context_before": "experiments were conducted on a server running 64-bit Ubuntu 20.04.6 LTS. The machine is equipped with two Intel®Xeon®Gold 5218R CPUs (@ 2.10 GHz), each providing 20 physical cores (40 threads per socket), for a total of 80 hardware threads. Implementation.To ensure the robustness of results and mitigate the impact of randomness, we adopted amulti-instance parallel testing strategy. For each DBMS, we leverage the testing framework’s multi-threading capability by launching multiple concurrent testing threads. Each thread operates with an independent random state and query generation trajectory.",
+      "context_after": "Verifieddenotes that the bugs have been officially confirmed and accepted by the developers as valid logic defects. configurations and runtime environments. The testing process ran continuously for 12 hours, effectively simulating repeated experiments across diverse random seeds. B. Overall Results for Logic Bug Detection We test all the target DBMSs with JoinEquiv. The testing process adheres to a conventional software defect detection pipeline, including fuzzing-based test generation, test case reduction, manual deduplication, cross-DBMS verification, and issue reporting to developers. As a",
+      "section": "A Evaluation Setup",
+      "page": 7,
+      "char_offset": 36611,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Comparison to Existing DBMS Testing Approaches In order to evaluate JoinEquiv’s unique ability to detect logical bugs related to INNER JOIN, we compare it with two representative metamorphic testing approaches: Ternary Logic Partitioning (TLP) and Differential Query Plans (DQP).",
+      "context_before": "’}, showing a typical case ofdata corruption, where the single-byte representation of ’7’was erroneously padded with zero bytes. This phenomenon clearly exposes a severe implementation inconsistency within TiDB when handling CHAR-typed data across different execution paths. By further decomposing the SDT-transformed query, we localized the corruption to the first UNIONoperation, suggesting that the defect was triggered during the merging of two intermediate result sets from outer joins. This indicates a deep-seated bug in TiDB’stype metadata propagationor memory layout management mechanisms. E.",
+      "context_after": "TLP shares a conceptual foundation withJoinEquiv, as both leverage set-theoretic principles to construct semantically equivalent query variants, whereas DQP evaluates the robustness of query optimizers by executing the same query under different physical plans, a strategy that has proven highly effective in uncovering join-related logical bugs. Listing 7. A data representation inconsistency under complex set operations in TiDB. CREATE TABLEt0(c0CHAR NOT NULL); CREATE TABLEt1LIKEt0; INSERT INTOt0(c0)VALUES(’-’); INSERT INTOt1(c0)VALUES(’7’); Original query (SELECT DISTINCTt1.c0, t0.c0FROMt1INN",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 9,
+      "char_offset": 45589,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP shares a conceptual foundation withJoinEquiv, as both leverage set-theoretic principles to construct semantically equivalent query variants, whereas DQP evaluates the robustness of query optimizers by executing the same query under different physical plans, a strategy that has proven highly effective in uncovering join-related logical bugs.",
+      "context_before": "rther decomposing the SDT-transformed query, we localized the corruption to the first UNIONoperation, suggesting that the defect was triggered during the merging of two intermediate result sets from outer joins. This indicates a deep-seated bug in TiDB’stype metadata propagationor memory layout management mechanisms. E. Comparison to Existing DBMS Testing Approaches In order to evaluate JoinEquiv’s unique ability to detect logical bugs related to INNER JOIN, we compare it with two representative metamorphic testing approaches: Ternary Logic Partitioning (TLP) and Differential Query Plans (DQP).",
+      "context_after": "Listing 7. A data representation inconsistency under complex set operations in TiDB. CREATE TABLEt0(c0CHAR NOT NULL); CREATE TABLEt1LIKEt0; INSERT INTOt0(c0)VALUES(’-’); INSERT INTOt1(c0)VALUES(’7’); Original query (SELECT DISTINCTt1.c0, t0.c0FROMt1INNER JOINt0ONϕ); -- {0x37|’-’}ὁB Transformed query(ADT rule) (SELECT DISTINCTt1.c0, t0.c0FROMt1LEFT JOINt0ONϕ) INTERSECT (SELECT DISTINCTt1.c0, t0.c0FROMt1RIGHT JOINt0ONϕ); -- {’7’|’-’}✓ Transformed query(SDT rule) ((SELECT DISTINCTt1.c0FROMt1LEFT JOINt0ONϕ) 2The full join condition: https://github.com/pingcap/tidb/issues/63736 UNION (SELECT DISTI",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 9,
+      "char_offset": 45868,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For each reported bug, we reconstruct equivalent test cases following TLP and DQP formulations to examine whether these approaches can reproduce the same erroneous behavior.",
+      "context_before": "ed query(SDT rule) ((SELECT DISTINCTt1.c0FROMt1LEFT JOINt0ONϕ) 2The full join condition: https://github.com/pingcap/tidb/issues/63736 UNION (SELECT DISTINCTt1.c0FROMt1RIGHT JOINt0ONϕ)) EXCEPT( ((SELECT DISTINCTt1.c0FROMt1LEFT JOINt0ONϕ) EXCEPT (SELECT DISTINCTt1.c0FROMt1RIGHT JOINt0ONϕ)) UNION ((SELECT DISTINCTt1.c0FROMt1RIGHT JOINt0ONϕ) EXCEPT (SELECT DISTINCTt1.c0FROMt1LEFT JOINt0ONϕ))); -- {0x37000000|’-’}ὁB To ensure a fair and systematic comparison, we conduct two complementary experiments. First, we perform acrossoracular validationbased on previously confirmed bugs detected byJoinEquiv.",
+      "context_after": "This validation allows us to assess the orthogonality and uniqueness ofJoinEquiv’s detection scope. Second, we carry out aunified cross-oracular evaluation, in which all approaches are executed side by side for 12 hours under identical query generation and database states, to quantitatively compare their bug-finding capabilities. Crucially, for every generated original INNER JOINquery that can be correctly executed, we apply all three transformation strategies (TLP, DQP, and JoinEquiv) within the same iteration, ensuring a fair comparison under an identical execution context. To evaluate the e",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 47262,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Crucially, for every generated original INNER JOINquery that can be correctly executed, we apply all three transformation strategies (TLP, DQP, and JoinEquiv) within the same iteration, ensuring a fair comparison under an identical execution context.",
+      "context_before": ", we perform acrossoracular validationbased on previously confirmed bugs detected byJoinEquiv. For each reported bug, we reconstruct equivalent test cases following TLP and DQP formulations to examine whether these approaches can reproduce the same erroneous behavior. This validation allows us to assess the orthogonality and uniqueness ofJoinEquiv’s detection scope. Second, we carry out aunified cross-oracular evaluation, in which all approaches are executed side by side for 12 hours under identical query generation and database states, to quantitatively compare their bug-finding capabilities.",
+      "context_after": "To evaluate the effectiveness of JoinEquiv, we implemented two representative metamorphic testing approaches as baselines. Ternary Logic Partitioning (TLP) (specifically, the TLPWHEREvariant) [20] relies on the set-theoretic principle of UNION. As illustrated in Listing 8, TLP partitions an original query into three sub-queries by injecting predicates into the WHEREclause based on SQL’s ternary logic (TRUE,FALSE, NULL) and validates that their combined result matches the original. Differential Query Plans (DQP) [25], in contrast, targets join optimizer robustness. It forces the DBMS to genera",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 47768,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "Ternary Logic Partitioning (TLP) (specifically, the TLPWHEREvariant) [20] relies on the set-theoretic principle of UNION.",
+      "context_before": "nd, we carry out aunified cross-oracular evaluation, in which all approaches are executed side by side for 12 hours under identical query generation and database states, to quantitatively compare their bug-finding capabilities. Crucially, for every generated original INNER JOINquery that can be correctly executed, we apply all three transformation strategies (TLP, DQP, and JoinEquiv) within the same iteration, ensuring a fair comparison under an identical execution context. To evaluate the effectiveness of JoinEquiv, we implemented two representative metamorphic testing approaches as baselines.",
+      "context_after": "As illustrated in Listing 8, TLP partitions an original query into three sub-queries by injecting predicates into the WHEREclause based on SQL’s ternary logic (TRUE,FALSE, NULL) and validates that their combined result matches the original. Differential Query Plans (DQP) [25], in contrast, targets join optimizer robustness. It forces the DBMS to generate diverse physical plans for the same query (via hints, see Listing 8) and checks for discrepancies (R i̸=R j) among the result sets. Cross-Oracular Validation of Reported Bugs.To assess the orthogonality of JoinEquiv’s detection scope, we firs",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 48141,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "As illustrated in Listing 8, TLP partitions an original query into three sub-queries by injecting predicates into the WHEREclause based on SQL’s ternary logic (TRUE,FALSE, NULL) and validates that their combined result matches the original.",
+      "context_before": "identical query generation and database states, to quantitatively compare their bug-finding capabilities. Crucially, for every generated original INNER JOINquery that can be correctly executed, we apply all three transformation strategies (TLP, DQP, and JoinEquiv) within the same iteration, ensuring a fair comparison under an identical execution context. To evaluate the effectiveness of JoinEquiv, we implemented two representative metamorphic testing approaches as baselines. Ternary Logic Partitioning (TLP) (specifically, the TLPWHEREvariant) [20] relies on the set-theoretic principle of UNION.",
+      "context_after": "Differential Query Plans (DQP) [25], in contrast, targets join optimizer robustness. It forces the DBMS to generate diverse physical plans for the same query (via hints, see Listing 8) and checks for discrepancies (R i̸=R j) among the result sets. Cross-Oracular Validation of Reported Bugs.To assess the orthogonality of JoinEquiv’s detection scope, we first perform across-oracular validationon previously reported bugs. For each bug exposed by JoinEquiv, we manually rewrite the corresponding query into its TLPand DQP-equivalent forms and execute them on the same database. Listing 8. Example of",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 48263,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "Differential Query Plans",
+      "technique": "dqp",
+      "sentence": "Differential Query Plans (DQP) [25], in contrast, targets join optimizer robustness.",
+      "context_before": "P, DQP, and JoinEquiv) within the same iteration, ensuring a fair comparison under an identical execution context. To evaluate the effectiveness of JoinEquiv, we implemented two representative metamorphic testing approaches as baselines. Ternary Logic Partitioning (TLP) (specifically, the TLPWHEREvariant) [20] relies on the set-theoretic principle of UNION. As illustrated in Listing 8, TLP partitions an original query into three sub-queries by injecting predicates into the WHEREclause based on SQL’s ternary logic (TRUE,FALSE, NULL) and validates that their combined result matches the original.",
+      "context_after": "It forces the DBMS to generate diverse physical plans for the same query (via hints, see Listing 8) and checks for discrepancies (R i̸=R j) among the result sets. Cross-Oracular Validation of Reported Bugs.To assess the orthogonality of JoinEquiv’s detection scope, we first perform across-oracular validationon previously reported bugs. For each bug exposed by JoinEquiv, we manually rewrite the corresponding query into its TLPand DQP-equivalent forms and execute them on the same database. Listing 8. Example of ternary logic partitioning (TLP) and differential query plans (DQP). Original query",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 48504,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "For each bug exposed by JoinEquiv, we manually rewrite the corresponding query into its TLPand DQP-equivalent forms and execute them on the same database.",
+      "context_before": "nto three sub-queries by injecting predicates into the WHEREclause based on SQL’s ternary logic (TRUE,FALSE, NULL) and validates that their combined result matches the original. Differential Query Plans (DQP) [25], in contrast, targets join optimizer robustness. It forces the DBMS to generate diverse physical plans for the same query (via hints, see Listing 8) and checks for discrepancies (R i̸=R j) among the result sets. Cross-Oracular Validation of Reported Bugs.To assess the orthogonality of JoinEquiv’s detection scope, we first perform across-oracular validationon previously reported bugs.",
+      "context_after": "Listing 8. Example of ternary logic partitioning (TLP) and differential query plans (DQP). Original query SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0; TLP (Where) transformed query SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE TRUEUNION ALL SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE FALSE UNION ALL SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE TRUEISNULL; DQP transformed query Plan variant 1 (hash join) hint SELECT/ *+ HASH_JOIN(t0, t1) */*FROMt0INNER JOINt1ON t0.c0 = t1.c0; Plan variant 2 (nested-loop join) hint SELECT/ *+ NESTED_LOOP_JOIN(t0, t1) */*FROMt0INNER JOINt1ONt0.c0 = t1.c0; For",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 48927,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "technique",
+      "surface": "ternary logic partitioning",
+      "technique": "tlp",
+      "sentence": "Example of ternary logic partitioning (TLP) and differential query plans (DQP).",
+      "context_before": "e original. Differential Query Plans (DQP) [25], in contrast, targets join optimizer robustness. It forces the DBMS to generate diverse physical plans for the same query (via hints, see Listing 8) and checks for discrepancies (R i̸=R j) among the result sets. Cross-Oracular Validation of Reported Bugs.To assess the orthogonality of JoinEquiv’s detection scope, we first perform across-oracular validationon previously reported bugs. For each bug exposed by JoinEquiv, we manually rewrite the corresponding query into its TLPand DQP-equivalent forms and execute them on the same database. Listing 8.",
+      "context_after": "Original query SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0; TLP (Where) transformed query SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE TRUEUNION ALL SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE FALSE UNION ALL SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE TRUEISNULL; DQP transformed query Plan variant 1 (hash join) hint SELECT/ *+ HASH_JOIN(t0, t1) */*FROMt0INNER JOINt1ON t0.c0 = t1.c0; Plan variant 2 (nested-loop join) hint SELECT/ *+ NESTED_LOOP_JOIN(t0, t1) */*FROMt0INNER JOINt1ONt0.c0 = t1.c0; Formally, given an INNER JOINqueryQ origwhose result differs from one or more of its algebraica",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 49093,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "c0; TLP (Where) transformed query SELECT *FROMt0INNER JOINt1ONt0.",
+      "context_before": "physical plans for the same query (via hints, see Listing 8) and checks for discrepancies (R i̸=R j) among the result sets. Cross-Oracular Validation of Reported Bugs.To assess the orthogonality of JoinEquiv’s detection scope, we first perform across-oracular validationon previously reported bugs. For each bug exposed by JoinEquiv, we manually rewrite the corresponding query into its TLPand DQP-equivalent forms and execute them on the same database. Listing 8. Example of ternary logic partitioning (TLP) and differential query plans (DQP). Original query SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.",
+      "context_after": "c0 = t1.c0WHERE TRUEUNION ALL SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE FALSE UNION ALL SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE TRUEISNULL; DQP transformed query Plan variant 1 (hash join) hint SELECT/ *+ HASH_JOIN(t0, t1) */*FROMt0INNER JOINt1ON t0.c0 = t1.c0; Plan variant 2 (nested-loop join) hint SELECT/ *+ NESTED_LOOP_JOIN(t0, t1) */*FROMt0INNER JOINt1ONt0.c0 = t1.c0; Formally, given an INNER JOINqueryQ origwhose result differs from one or more of its algebraically equivalent transformed queries, we attempt to construct three predicate-partitioned queries according to TLP’s formu",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 49228,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M24",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "c0WHERE TRUEISNULL; DQP transformed query Plan variant 1 (hash join) hint SELECT/ *+ HASH_JOIN(t0, t1) */*FROMt0INNER JOINt1ON t0.",
+      "context_before": "f JoinEquiv’s detection scope, we first perform across-oracular validationon previously reported bugs. For each bug exposed by JoinEquiv, we manually rewrite the corresponding query into its TLPand DQP-equivalent forms and execute them on the same database. Listing 8. Example of ternary logic partitioning (TLP) and differential query plans (DQP). Original query SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0; TLP (Where) transformed query SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE TRUEUNION ALL SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE FALSE UNION ALL SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.",
+      "context_after": "c0 = t1.c0; Plan variant 2 (nested-loop join) hint SELECT/ *+ NESTED_LOOP_JOIN(t0, t1) */*FROMt0INNER JOINt1ONt0.c0 = t1.c0; Formally, given an INNER JOINqueryQ origwhose result differs from one or more of its algebraically equivalent transformed queries, we attempt to construct three predicate-partitioned queries according to TLP’s formulation, and generate multiple physically equivalent variants of queries QDQPusing optimizer hints. We leverage 32 optimizer hints for MySQL and Percona, and 22 for TiDB (sourced from SQLancer-DQP); DQP does not support DuckDB. If the same inconsistency appears,",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 49425,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "c0; Formally, given an INNER JOINqueryQ origwhose result differs from one or more of its algebraically equivalent transformed queries, we attempt to construct three predicate-partitioned queries according to TLP’s formulation, and generate multiple physically equivalent variants of queries QDQPusing optimizer hints.",
+      "context_before": "tabase. Listing 8. Example of ternary logic partitioning (TLP) and differential query plans (DQP). Original query SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0; TLP (Where) transformed query SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE TRUEUNION ALL SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE FALSE UNION ALL SELECT *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE TRUEISNULL; DQP transformed query Plan variant 1 (hash join) hint SELECT/ *+ HASH_JOIN(t0, t1) */*FROMt0INNER JOINt1ON t0.c0 = t1.c0; Plan variant 2 (nested-loop join) hint SELECT/ *+ NESTED_LOOP_JOIN(t0, t1) */*FROMt0INNER JOINt1ONt0.c0 = t1.",
+      "context_after": "We leverage 32 optimizer hints for MySQL and Percona, and 22 for TiDB (sourced from SQLancer-DQP); DQP does not support DuckDB. If the same inconsistency appears, the bug is considered reproducible by TLP or DQP; Otherwise, it is classified as aJoin-specific bug. Among the 27 bugs originally detected by JoinEquiv, none were reproduced by either baseline, demonstrating that JoinEquiv possesses a fundamental and irreplaceable advantage over existing approaches in the specific domain of INNER JOINlogic bugs. Listing 9. TLP transformed query for Listing 3 and Listing 6 --TLP Transformed query for",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 49675,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We leverage 32 optimizer hints for MySQL and Percona, and 22 for TiDB (sourced from SQLancer-DQP); DQP does not support DuckDB.",
+      "context_before": "T *FROMt0INNER JOINt1ONt0.c0 = t1.c0WHERE TRUEISNULL; DQP transformed query Plan variant 1 (hash join) hint SELECT/ *+ HASH_JOIN(t0, t1) */*FROMt0INNER JOINt1ON t0.c0 = t1.c0; Plan variant 2 (nested-loop join) hint SELECT/ *+ NESTED_LOOP_JOIN(t0, t1) */*FROMt0INNER JOINt1ONt0.c0 = t1.c0; Formally, given an INNER JOINqueryQ origwhose result differs from one or more of its algebraically equivalent transformed queries, we attempt to construct three predicate-partitioned queries according to TLP’s formulation, and generate multiple physically equivalent variants of queries QDQPusing optimizer hints.",
+      "context_after": "If the same inconsistency appears, the bug is considered reproducible by TLP or DQP; Otherwise, it is classified as aJoin-specific bug. Among the 27 bugs originally detected by JoinEquiv, none were reproduced by either baseline, demonstrating that JoinEquiv possesses a fundamental and irreplaceable advantage over existing approaches in the specific domain of INNER JOINlogic bugs. Listing 9. TLP transformed query for Listing 3 and Listing 6 --TLP Transformed query for Listing 3 SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE TRUEUNION SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE FALSE UNION SELE",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 49991,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "If the same inconsistency appears, the bug is considered reproducible by TLP or DQP; Otherwise, it is classified as aJoin-specific bug.",
+      "context_before": "(t0, t1) */*FROMt0INNER JOINt1ON t0.c0 = t1.c0; Plan variant 2 (nested-loop join) hint SELECT/ *+ NESTED_LOOP_JOIN(t0, t1) */*FROMt0INNER JOINt1ONt0.c0 = t1.c0; Formally, given an INNER JOINqueryQ origwhose result differs from one or more of its algebraically equivalent transformed queries, we attempt to construct three predicate-partitioned queries according to TLP’s formulation, and generate multiple physically equivalent variants of queries QDQPusing optimizer hints. We leverage 32 optimizer hints for MySQL and Percona, and 22 for TiDB (sourced from SQLancer-DQP); DQP does not support DuckDB.",
+      "context_after": "Among the 27 bugs originally detected by JoinEquiv, none were reproduced by either baseline, demonstrating that JoinEquiv possesses a fundamental and irreplaceable advantage over existing approaches in the specific domain of INNER JOINlogic bugs. Listing 9. TLP transformed query for Listing 3 and Listing 6 --TLP Transformed query for Listing 3 SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE TRUEUNION SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE FALSE UNION SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE TRUEISNULL; -- empty set✓ --TLP Transformed query for Listing 6 SELECTt0.c0, t1.c0FROMt0INNER",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 50119,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP transformed query for Listing 3 and Listing 6 --TLP Transformed query for Listing 3 SELECT DISTINCTt0.",
+      "context_before": "e multiple physically equivalent variants of queries QDQPusing optimizer hints. We leverage 32 optimizer hints for MySQL and Percona, and 22 for TiDB (sourced from SQLancer-DQP); DQP does not support DuckDB. If the same inconsistency appears, the bug is considered reproducible by TLP or DQP; Otherwise, it is classified as aJoin-specific bug. Among the 27 bugs originally detected by JoinEquiv, none were reproduced by either baseline, demonstrating that JoinEquiv possesses a fundamental and irreplaceable advantage over existing approaches in the specific domain of INNER JOINlogic bugs. Listing 9.",
+      "context_after": "c0FROMt1NATURAL JOINt0WHERE TRUEUNION SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE FALSE UNION SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE TRUEISNULL; -- empty set✓ --TLP Transformed query for Listing 6 SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE TRUEUNION ALL SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE FALSE UNION ALL SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE TRUEISNULL; -- {0|1}ὁB During this validation, we observed an interesting phenomenon when applying the TLP strategy. As shown in Listing 3, the original NATURAL JOINquery correctly returns an empty se",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 50512,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "c0FROMt1NATURAL JOINt0WHERE TRUEISNULL; -- empty set✓ --TLP Transformed query for Listing 6 SELECTt0.",
+      "context_before": "nconsistency appears, the bug is considered reproducible by TLP or DQP; Otherwise, it is classified as aJoin-specific bug. Among the 27 bugs originally detected by JoinEquiv, none were reproduced by either baseline, demonstrating that JoinEquiv possesses a fundamental and irreplaceable advantage over existing approaches in the specific domain of INNER JOINlogic bugs. Listing 9. TLP transformed query for Listing 3 and Listing 6 --TLP Transformed query for Listing 3 SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE TRUEUNION SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE FALSE UNION SELECT DISTINCTt0.",
+      "context_after": "c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE TRUEUNION ALL SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE FALSE UNION ALL SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE TRUEISNULL; -- {0|1}ὁB During this validation, we observed an interesting phenomenon when applying the TLP strategy. As shown in Listing 3, the original NATURAL JOINquery correctly returns an empty set, whereas its SJT-transformed counterpart incorrectly produces{0}. We then rewrote the original query following the TLP’s form, partitioning it into its corresponding subqueries (as shown in Listing 9). Interesting",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 50733,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "1 WHERE TRUEISNULL; -- {0|1}ὁB During this validation, we observed an interesting phenomenon when applying the TLP strategy.",
+      "context_before": "aceable advantage over existing approaches in the specific domain of INNER JOINlogic bugs. Listing 9. TLP transformed query for Listing 3 and Listing 6 --TLP Transformed query for Listing 3 SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE TRUEUNION SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE FALSE UNION SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE TRUEISNULL; -- empty set✓ --TLP Transformed query for Listing 6 SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE TRUEUNION ALL SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE FALSE UNION ALL SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.",
+      "context_after": "As shown in Listing 3, the original NATURAL JOINquery correctly returns an empty set, whereas its SJT-transformed counterpart incorrectly produces{0}. We then rewrote the original query following the TLP’s form, partitioning it into its corresponding subqueries (as shown in Listing 9). Interestingly, all partitions also correctly return empty sets, and their union remains consistent with the original query result. In contrast, as shown in Listing 5, the original query incorrectly returns{0|1}. When rewritten using the TLP approach, the partitioned queries yield the same erroneous result{0|1},",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 51012,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We then rewrote the original query following the TLP’s form, partitioning it into its corresponding subqueries (as shown in Listing 9).",
+      "context_before": "URAL JOINt0WHERE FALSE UNION SELECT DISTINCTt0.c0FROMt1NATURAL JOINt0WHERE TRUEISNULL; -- empty set✓ --TLP Transformed query for Listing 6 SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE TRUEUNION ALL SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE FALSE UNION ALL SELECTt0.c0, t1.c0FROMt0INNER JOINt1ONt0.c0<-0.1 WHERE TRUEISNULL; -- {0|1}ὁB During this validation, we observed an interesting phenomenon when applying the TLP strategy. As shown in Listing 3, the original NATURAL JOINquery correctly returns an empty set, whereas its SJT-transformed counterpart incorrectly produces{0}.",
+      "context_after": "Interestingly, all partitions also correctly return empty sets, and their union remains consistent with the original query result. In contrast, as shown in Listing 5, the original query incorrectly returns{0|1}. When rewritten using the TLP approach, the partitioned queries yield the same erroneous result{0|1}, whereas the ADT-transformed query correctly produces an empty set. TLP cannot address the optimizer’s semantic bias due to structural transformations during join rewriting. Unified Cross-Oracular Evaluation.To ensure a fair and quantitative comparison between metamorphic testing approa",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 51286,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "When rewritten using the TLP approach, the partitioned queries yield the same erroneous result{0|1}, whereas the ADT-transformed query correctly produces an empty set.",
+      "context_before": "{0|1}ὁB During this validation, we observed an interesting phenomenon when applying the TLP strategy. As shown in Listing 3, the original NATURAL JOINquery correctly returns an empty set, whereas its SJT-transformed counterpart incorrectly produces{0}. We then rewrote the original query following the TLP’s form, partitioning it into its corresponding subqueries (as shown in Listing 9). Interestingly, all partitions also correctly return empty sets, and their union remains consistent with the original query result. In contrast, as shown in Listing 5, the original query incorrectly returns{0|1}.",
+      "context_after": "TLP cannot address the optimizer’s semantic bias due to structural transformations during join rewriting. Unified Cross-Oracular Evaluation.To ensure a fair and quantitative comparison between metamorphic testing approaches, we implemented a unified framework that executes JoinEquiv, TLP, and DQP side by side with identical query generation and database states. All three approaches share the same query generator, database state, random seeds, and execution environment, guaranteeing identical experimental conditions. For each original query, all three transformations are applied sequentially,",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 51634,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M33",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP cannot address the optimizer’s semantic bias due to structural transformations during join rewriting.",
+      "context_before": "turns an empty set, whereas its SJT-transformed counterpart incorrectly produces{0}. We then rewrote the original query following the TLP’s form, partitioning it into its corresponding subqueries (as shown in Listing 9). Interestingly, all partitions also correctly return empty sets, and their union remains consistent with the original query result. In contrast, as shown in Listing 5, the original query incorrectly returns{0|1}. When rewritten using the TLP approach, the partitioned queries yield the same erroneous result{0|1}, whereas the ADT-transformed query correctly produces an empty set.",
+      "context_after": "Unified Cross-Oracular Evaluation.To ensure a fair and quantitative comparison between metamorphic testing approaches, we implemented a unified framework that executes JoinEquiv, TLP, and DQP side by side with identical query generation and database states. All three approaches share the same query generator, database state, random seeds, and execution environment, guaranteeing identical experimental conditions. For each original query, all three transformations are applied sequentially, and the results are compared across TABLE IIINUMBER OFBUGS REPORTED IN A12H UNIFIED RUN(DQPRESULTS EXCLUDE",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 51802,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "To ensure a fair and quantitative comparison between metamorphic testing approaches, we implemented a unified framework that executes JoinEquiv, TLP, and DQP side by side with identical query generation and database states.",
+      "context_before": "orm, partitioning it into its corresponding subqueries (as shown in Listing 9). Interestingly, all partitions also correctly return empty sets, and their union remains consistent with the original query result. In contrast, as shown in Listing 5, the original query incorrectly returns{0|1}. When rewritten using the TLP approach, the partitioned queries yield the same erroneous result{0|1}, whereas the ADT-transformed query correctly produces an empty set. TLP cannot address the optimizer’s semantic bias due to structural transformations during join rewriting. Unified Cross-Oracular Evaluation.",
+      "context_after": "All three approaches share the same query generator, database state, random seeds, and execution environment, guaranteeing identical experimental conditions. For each original query, all three transformations are applied sequentially, and the results are compared across TABLE IIINUMBER OFBUGS REPORTED IN A12H UNIFIED RUN(DQPRESULTS EXCLUDEDUCKDB) DBMSTLP DQP JoinEquiv TiDB 0 0 7 MySQL 0 1 13 Percona 0 2 3 DuckDB 0 – 13 Total0 3 36 Increment36↑20↑– the same DBMS instance. This setup ensures a fair and controlled environment for large-scale evaluation. Table III summarizes the number of bugs fo",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 51943,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M35",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "For each original query, all three transformations are applied sequentially, and the results are compared across TABLE IIINUMBER OFBUGS REPORTED IN A12H UNIFIED RUN(DQPRESULTS EXCLUDEDUCKDB) DBMSTLP DQP JoinEquiv TiDB 0 0 7 MySQL 0 1 13 Percona 0 2 3 DuckDB 0 – 13 Total0 3 36 Increment36↑20↑– the same DBMS instance.",
+      "context_before": "esult{0|1}, whereas the ADT-transformed query correctly produces an empty set. TLP cannot address the optimizer’s semantic bias due to structural transformations during join rewriting. Unified Cross-Oracular Evaluation.To ensure a fair and quantitative comparison between metamorphic testing approaches, we implemented a unified framework that executes JoinEquiv, TLP, and DQP side by side with identical query generation and database states. All three approaches share the same query generator, database state, random seeds, and execution environment, guaranteeing identical experimental conditions.",
+      "context_after": "This setup ensures a fair and controlled environment for large-scale evaluation. Table III summarizes the number of bugs found in the 12hour run on MySQL, TiDB, Percona, and DuckDB. It shows that JoinEquiv found 36 more join-related inconsistencies than TLP and 20 more than DQP (without DuckDB). TLP and DQP focus on predicate-level or plan-level consistency. In contrast, JoinEquiv operates at the algebraic level of join semantics, leveraging simple yet powerful set-theoretic identities. JoinEquiv systematically verifies the semantic correctness of join rewrites and set-operator interactions,",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 10,
+      "char_offset": 52324,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M36",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "It shows that JoinEquiv found 36 more join-related inconsistencies than TLP and 20 more than DQP (without DuckDB).",
+      "context_before": "base state, random seeds, and execution environment, guaranteeing identical experimental conditions. For each original query, all three transformations are applied sequentially, and the results are compared across TABLE IIINUMBER OFBUGS REPORTED IN A12H UNIFIED RUN(DQPRESULTS EXCLUDEDUCKDB) DBMSTLP DQP JoinEquiv TiDB 0 0 7 MySQL 0 1 13 Percona 0 2 3 DuckDB 0 – 13 Total0 3 36 Increment36↑20↑– the same DBMS instance. This setup ensures a fair and controlled environment for large-scale evaluation. Table III summarizes the number of bugs found in the 12hour run on MySQL, TiDB, Percona, and DuckDB.",
+      "context_after": "TLP and DQP focus on predicate-level or plan-level consistency. In contrast, JoinEquiv operates at the algebraic level of join semantics, leveraging simple yet powerful set-theoretic identities. JoinEquiv systematically verifies the semantic correctness of join rewrites and set-operator interactions, which predicate partitioning or plan variation cannot cover. As a result, JoinEquiv can expose deep semantic inconsistencies arising from incorrect join rewriting, NULL propagation, and set operator evaluation. In summary, JoinEquiv effectively uncovers logic bugs that advanced state-of-the-art a",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 11,
+      "char_offset": 52824,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M37",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP and DQP focus on predicate-level or plan-level consistency.",
+      "context_before": "nal query, all three transformations are applied sequentially, and the results are compared across TABLE IIINUMBER OFBUGS REPORTED IN A12H UNIFIED RUN(DQPRESULTS EXCLUDEDUCKDB) DBMSTLP DQP JoinEquiv TiDB 0 0 7 MySQL 0 1 13 Percona 0 2 3 DuckDB 0 – 13 Total0 3 36 Increment36↑20↑– the same DBMS instance. This setup ensures a fair and controlled environment for large-scale evaluation. Table III summarizes the number of bugs found in the 12hour run on MySQL, TiDB, Percona, and DuckDB. It shows that JoinEquiv found 36 more join-related inconsistencies than TLP and 20 more than DQP (without DuckDB).",
+      "context_after": "In contrast, JoinEquiv operates at the algebraic level of join semantics, leveraging simple yet powerful set-theoretic identities. JoinEquiv systematically verifies the semantic correctness of join rewrites and set-operator interactions, which predicate partitioning or plan variation cannot cover. As a result, JoinEquiv can expose deep semantic inconsistencies arising from incorrect join rewriting, NULL propagation, and set operator evaluation. In summary, JoinEquiv effectively uncovers logic bugs that advanced state-of-the-art approaches fail to detect, thereby answering RQ3. Code Coverage.We",
+      "section": "E Comparison to Existing DBMS Testing Approaches",
+      "page": 11,
+      "char_offset": 52939,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M38",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Among existing DBMS testing studies, the SQLancer family [27] is the most representative for discovering logic bugs and is the closest to our work.",
+      "context_before": "bugs have been fixed so far, the fix rate alone does not directly reflect their severity. Among the reported bugs, the low fix rate stems mainly from the complexity of fixing logic bugs in join optimizers, which often requires modifying core rewrite rules and extensive regression testing to avoid performance regressions. Another important factor is the timing of bug disclosure. As most bugs were reported recently, subsequent releases were developed without awareness of them, making it unlikely that fixes can be incorporated in the short term. VII. RELATEDWORK Logic Bug Identification for DBMS.",
+      "context_after": "Its key contribution lies in providing a framework for systematically uncovering deep errors inside DBMSs. SQLancer targets logic bugs that lead to incorrect query results in DBMSs. Other representative approaches include PQS [19], NoREC [22], TLP [20], DQP [25], and CERT [37]. PQS and NoREC have been discussed in §I, while TLP and DQP were analyzed in §V-E. CERT [37] aims to identify performance issues caused by unexpected cardinality estimations, that is, the cases where the estimated number of result tuples significantly deviates from the actual number. In contrast, our approach specifical",
+      "section": "VII RELATEDWORK",
+      "page": 12,
+      "char_offset": 58930,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M39",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer targets logic bugs that lead to incorrect query results in DBMSs.",
+      "context_before": "s and extensive regression testing to avoid performance regressions. Another important factor is the timing of bug disclosure. As most bugs were reported recently, subsequent releases were developed without awareness of them, making it unlikely that fixes can be incorporated in the short term. VII. RELATEDWORK Logic Bug Identification for DBMS.Among existing DBMS testing studies, the SQLancer family [27] is the most representative for discovering logic bugs and is the closest to our work. Its key contribution lies in providing a framework for systematically uncovering deep errors inside DBMSs.",
+      "context_after": "Other representative approaches include PQS [19], NoREC [22], TLP [20], DQP [25], and CERT [37]. PQS and NoREC have been discussed in §I, while TLP and DQP were analyzed in §V-E. CERT [37] aims to identify performance issues caused by unexpected cardinality estimations, that is, the cases where the estimated number of result tuples significantly deviates from the actual number. In contrast, our approach specifically targets the detection of logic bugs in join optimizations, an area that none of the existing SQLancerbased techniques are designed to explore. Generator-based Testing for DBMS.A v",
+      "section": "VII RELATEDWORK",
+      "page": 12,
+      "char_offset": 59184,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M40",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Other representative approaches include PQS [19], NoREC [22], TLP [20], DQP [25], and CERT [37].",
+      "context_before": "r important factor is the timing of bug disclosure. As most bugs were reported recently, subsequent releases were developed without awareness of them, making it unlikely that fixes can be incorporated in the short term. VII. RELATEDWORK Logic Bug Identification for DBMS.Among existing DBMS testing studies, the SQLancer family [27] is the most representative for discovering logic bugs and is the closest to our work. Its key contribution lies in providing a framework for systematically uncovering deep errors inside DBMSs. SQLancer targets logic bugs that lead to incorrect query results in DBMSs.",
+      "context_after": "PQS and NoREC have been discussed in §I, while TLP and DQP were analyzed in §V-E. CERT [37] aims to identify performance issues caused by unexpected cardinality estimations, that is, the cases where the estimated number of result tuples significantly deviates from the actual number. In contrast, our approach specifically targets the detection of logic bugs in join optimizations, an area that none of the existing SQLancerbased techniques are designed to explore. Generator-based Testing for DBMS.A variety of tools have been developed for generating database data [38]–[42] and SQL queries [16],",
+      "section": "VII RELATEDWORK",
+      "page": 12,
+      "char_offset": 59259,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp",
+        "cert"
+      ]
+    },
+    {
+      "id": "M41",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS and NoREC have been discussed in §I, while TLP and DQP were analyzed in §V-E.",
+      "context_before": "nt releases were developed without awareness of them, making it unlikely that fixes can be incorporated in the short term. VII. RELATEDWORK Logic Bug Identification for DBMS.Among existing DBMS testing studies, the SQLancer family [27] is the most representative for discovering logic bugs and is the closest to our work. Its key contribution lies in providing a framework for systematically uncovering deep errors inside DBMSs. SQLancer targets logic bugs that lead to incorrect query results in DBMSs. Other representative approaches include PQS [19], NoREC [22], TLP [20], DQP [25], and CERT [37].",
+      "context_after": "CERT [37] aims to identify performance issues caused by unexpected cardinality estimations, that is, the cases where the estimated number of result tuples significantly deviates from the actual number. In contrast, our approach specifically targets the detection of logic bugs in join optimizations, an area that none of the existing SQLancerbased techniques are designed to explore. Generator-based Testing for DBMS.A variety of tools have been developed for generating database data [38]–[42] and SQL queries [16], [43]–[48] to automatically construct test cases for DBMS evaluation. However, rela",
+      "section": "VII RELATEDWORK",
+      "page": 12,
+      "char_offset": 59356,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M42",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "CERT [37] aims to identify performance issues caused by unexpected cardinality estimations, that is, the cases where the estimated number of result tuples significantly deviates from the actual number.",
+      "context_before": "s can be incorporated in the short term. VII. RELATEDWORK Logic Bug Identification for DBMS.Among existing DBMS testing studies, the SQLancer family [27] is the most representative for discovering logic bugs and is the closest to our work. Its key contribution lies in providing a framework for systematically uncovering deep errors inside DBMSs. SQLancer targets logic bugs that lead to incorrect query results in DBMSs. Other representative approaches include PQS [19], NoREC [22], TLP [20], DQP [25], and CERT [37]. PQS and NoREC have been discussed in §I, while TLP and DQP were analyzed in §V-E.",
+      "context_after": "In contrast, our approach specifically targets the detection of logic bugs in join optimizations, an area that none of the existing SQLancerbased techniques are designed to explore. Generator-based Testing for DBMS.A variety of tools have been developed for generating database data [38]–[42] and SQL queries [16], [43]–[48] to automatically construct test cases for DBMS evaluation. However, relatively little attention has been paid to the design of test oracles, which are essential for determining the correctness of query results. Generationbased testing approaches [46], [49]–[53] have been wi",
+      "section": "VII RELATEDWORK",
+      "page": 12,
+      "char_offset": 59438,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M43",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "By viewing the join operator as a generalized intersection under set and multiset semantics, we systematically derived three provably equivalent rewriting rules (SJT, ADT, and SDT) and applied them to uncover inconsistencies overlooked by existing state-of-the-art testing approaches such as TLP and DQP.",
+      "context_before": "or assertion failures. They are not well-suited for detecting logic bugs, where the DBMS returns incorrect results without crashing. In contrast, our work focuses on the semantic correctness of query results. By systematically applying algebraic equivalence rules, we generate test oracles that can identify logical inconsistencies in join optimizations that coverage-guided fuzzers would likely miss. VIII. CONCLUSION Building on the basic principles of set theory, this paper introduced JoinEquiv, a novel framework for detecting logic bugs in DBMSs through equivalence-based query transformation.",
+      "context_after": "Extensive experiments on four DBMSs detected 27 previously unknown bugs and 14 verified ascriticalbugs, demonstrating that even simple algebraic equivalences can serve as powerful metamorphic relations for testing complex database systems. Future work will explore the application of equivalence-based testing beyond join operations. ACKNOWLEDGEMENTS We thank anonymous reviewers for their constructive comments and suggestions, which helped improve this paper. This work was supported in part by Ant Digital Technologies, Ant Group Research Fund. This work was also supported in part by the Shangha",
+      "section": "VIII CONCLUSION",
+      "page": 12,
+      "char_offset": 63078,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "dqp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M14",
+        "M36"
+      ],
+      "extends_technique": [
+        "M11"
+      ],
+      "describes_as_state_of_the_art": [
+        "M43"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:11:45Z",
+    "is_model_written": true,
+    "summary": "JoinEquiv detects logic bugs in INNER JOIN optimisation through set theory. For each test case it generates equivalent queries from the intersection operation together with three semantics-preserving rewrite rules -- symmetric join, asymmetric difference and symmetric difference transformations -- which turn a simple join query into a more complex but equivalent form, and compares the results. Across MySQL, TiDB, DuckDB and Percona it uncovered 29 previously unknown issues, 27 officially confirmed.",
+    "narrative": "The test case generation is built on SQLancer, which the authors extend so that all generated columns are NOT NULL, and it relies on SQLancer's randomised population of predicates and clauses. TLP is both its closest conceptual relative -- the paper says both rest on set-theoretic principles -- and one of the two approaches it is measured against, with each JoinEquiv bug rewritten into TLP and DQP form to see whether they reproduce it.",
+    "roles": {
+      "M1": "definition",
+      "M10": "reuse_implementation",
+      "M11": "extension",
+      "M12": "reuse_component",
+      "M13": "baseline",
+      "M14": "baseline",
+      "M15": "background",
+      "M16": "baseline",
+      "M17": "baseline",
+      "M18": "definition",
+      "M19": "definition",
+      "M20": "definition",
+      "M21": "baseline",
+      "M22": "definition",
+      "M23": "definition",
+      "M24": "definition",
+      "M25": "definition",
+      "M26": "reuse_component",
+      "M27": "baseline",
+      "M28": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M10",
+          "M11",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M10",
+            "sentence": "The test case generation builds upon the SQLancer framework, which provides a grammar-aware SQL query generation engine.",
+            "section": "E Database and Query Generation",
+            "page": 6
+          },
+          {
+            "mention_id": "M11",
+            "sentence": "We extend SQLancer to satisfy an additional precondition that all columns are NOT NULL.",
+            "section": "E Database and Query Generation",
+            "page": 6
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "Such a table permutation, combined with SQLancer’s randomized population of ONpredicates,WHEREclauses, and SELECT list expressions, ensures that the framework explores a diverse range of symmetric and asymmetric execution plans.",
+            "section": "E Database and Query Generation",
+            "page": 6
+          }
+        ],
+        "reasoning": "M10 states the test case generation builds on the SQLancer framework for grammar-aware query generation, M11 that they extended it with a NOT NULL precondition, and M12 that they rely on its randomised population of ON predicates, WHERE clauses and SELECT expressions. M26 also takes optimizer hints from SQLancer-DQP.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M11 extends the SQLancer tool with a generation precondition, which is reuse of the codebase rather than extension of a technique. The oracle itself, built from intersection and three difference transformations, is presented as new; TLP is described as sharing a conceptual foundation, not as being generalised."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M13",
+          "M14",
+          "M16",
+          "M17"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M13",
+            "sentence": "Within each thread, for every INNER JOINquery, JoinEquiv, TLP, and DQP are applied sequentially under the same system TABLE IISUMMARY OFLOGICBUGREPORTS ANDVERIFICATIONSTATUS INDIFFERENTRDBMSS RDBMS Reported Verified Fixed Intended Component Severity Identifier MySQL 10 8 1 2 Optimizer (8) Critical (8)Bug#118544, Bug#118684, Bug#118710, Bug#118857, Bug#118858, Bug#118949, Bug#119032, Bug#119059 TiDB 13 13 4 0Planner (11) Execution (2)Critical (3)#62380, #62444, #62456, #62459, #62460, #62644, #62645, #62689, #63596, #63601, #63635, #63636, #63736 Percona 3 3 0 0 Optimizer (3) Critical (3)PS-10124, PS-10127, DISMYSQL-535 DuckDB 3 3 3 0 – –#20483, #20486, #20608 Total29 27 8 2 – Critical (14) – Note:For DuckDB, component and severity were not provided, so they are marked as “–”.",
+            "section": "A Evaluation Setup",
+            "page": 7
+          },
+          {
+            "mention_id": "M14",
+            "sentence": "Comparison to Existing DBMS Testing Approaches In order to evaluate JoinEquiv’s unique ability to detect logical bugs related to INNER JOIN, we compare it with two representative metamorphic testing approaches: Ternary Logic Partitioning (TLP) and Differential Query Plans (DQP).",
+            "section": "E Comparison to Existing DBMS Testing Approaches",
+            "page": 9
+          },
+          {
+            "mention_id": "M16",
+            "sentence": "For each reported bug, we reconstruct equivalent test cases following TLP and DQP formulations to examine whether these approaches can reproduce the same erroneous behavior.",
+            "section": "E Comparison to Existing DBMS Testing Approaches",
+            "page": 10
+          },
+          {
+            "mention_id": "M17",
+            "sentence": "Crucially, for every generated original INNER JOINquery that can be correctly executed, we apply all three transformation strategies (TLP, DQP, and JoinEquiv) within the same iteration, ensuring a fair comparison under an identical execution context.",
+            "section": "E Comparison to Existing DBMS Testing Approaches",
+            "page": 10
+          }
+        ],
+        "reasoning": "TLP and DQP are the two approaches JoinEquiv is compared against: all three transformations are applied to every generated query in the same session, and each reported bug is rewritten into TLP and DQP form to test reproducibility.",
+        "techniques": [
+          "tlp",
+          "dqp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "A pattern fired on M43, which was not among the mentions read here. M14 calls TLP and DQP representative approaches rather than the state of the art in those words."
+      }
+    },
+    "disagreements": [
+      "A pattern read M11 as extending a technique. It extends the SQLancer tool -- adding a NOT NULL precondition to generation -- which is reuse of the codebase, and is recorded there instead."
+    ],
+    "unresolved": [
+      "Whether the paper calls SQLancer or its techniques state of the art: the mention a pattern fired on, M43, was outside the set read."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icde65706_2026_00240.json b/_data/papers/paper_doi_10_1109_icde65706_2026_00240.json
new file mode 100644
index 0000000..1d0a82e
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icde65706_2026_00240.json
@@ -0,0 +1,933 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:23:56Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icde65706.2026.00240",
+    "title": "LLMSQLMUTATOR: LLM-Powered Test Case Generation for Database Using Bug Reports",
+    "authors": [
+      "Chenglin Tian",
+      "Chaofan Li",
+      "Yawen Li",
+      "Yingxia Shao"
+    ],
+    "year": 2026,
+    "venue": "IEEE International Conference on Data Engineering",
+    "doi": "10.1109/icde65706.2026.00240",
+    "arxiv_id": null,
+    "s2_paper_id": "76ce4c5143cca31617dc53fe0c37b77ad8d5150b",
+    "url": "https://doi.org/10.1109/icde65706.2026.00240",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icde65706.2026.00240",
+      "retrieved_at": "2026-09-08T15:23:56Z",
+      "chars": 82849,
+      "content_sha256": "sha256:c63fab566c1f5bc10dea2ecdbea08fa5409a7cfd1dcff24312d2760660ab9a5d"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 14,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1695
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUNDS",
+        "start": 11494
+      },
+      {
+        "number": "A",
+        "title": "Basics of Bug Reports",
+        "start": 11510
+      },
+      {
+        "number": "B",
+        "title": "Mutation-based Software Testing",
+        "start": 12686
+      },
+      {
+        "number": "III",
+        "title": "LLMSQLM UTATOR",
+        "start": 14003
+      },
+      {
+        "number": "A",
+        "title": "Bug Knowledge Extraction",
+        "start": 16271
+      },
+      {
+        "number": "B",
+        "title": "Bug-Driven SQL Mutation",
+        "start": 23023
+      },
+      {
+        "number": "C",
+        "title": "LLM-Based Semantic Validation",
+        "start": 33240
+      },
+      {
+        "number": "D",
+        "title": "DBMS Testing",
+        "start": 40414
+      },
+      {
+        "number": "E",
+        "title": "Discussion of the Hybrid Design of LLMSQLM UTATOR",
+        "start": 41855
+      },
+      {
+        "number": "IV",
+        "title": "EVALUATION",
+        "start": 42498
+      },
+      {
+        "number": "B",
+        "title": "Experiment results",
+        "start": 46146
+      },
+      {
+        "number": "V",
+        "title": "RELATED WORK",
+        "start": 65012
+      },
+      {
+        "number": "VI",
+        "title": "CONCLUSION",
+        "start": 68671
+      },
+      {
+        "number": "VII",
+        "title": "ACKNOWLEDGEMENT",
+        "start": 69963
+      },
+      {
+        "number": "VIII",
+        "title": "AI-GENERATED CONTENT ACKNOWLEDGEMENT",
+        "start": 70250
+      },
+      {
+        "number": "H",
+        "title": "Liu et al., “A survey on llm-as-a-judge,” CoRR, 2024.",
+        "start": 78236
+      },
+      {
+        "number": "Z",
+        "title": "Li, D. Li, E. Xing et al., “Judging llm-as-a-judge with mt-bench",
+        "start": 78769
+      },
+      {
+        "number": "C",
+        "title": "Zhang, C. Ruan et al., “Deepseek-v3 technical report,” CoRR, 2024.",
+        "start": 79317
+      },
+      {
+        "number": "J",
+        "title": "Wei, “Simple testing can expose most critical transaction bugs:",
+        "start": 80106
+      },
+      {
+        "number": "D",
+        "title": "Poshyvanyk, “Enhancing mobile app bug reporting via real-time",
+        "start": 82354
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Mysql,” https://www.mysql.com, 1995, accessed on July, 7, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“Tidb,” https://www.pingcap.com/tidb/, 2016, accessed on July, 7, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“Sqlite,” https://www.sqlite.org/index.html, 2000, accessed on July, 7, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "“Clickhouse,” https://clickhouse.com/, 1996, accessed on July 7, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "“Duckdb,” https://duckdb.org, 2019, accessed on July, 7, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "A. Seltenreich, B. Tang, and S. Mullender, “Sqlsmith,” 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Q. Wu, Q. Xiang, Y. Shao, Q. Luo, and Q. Xu, “Dbpecker: A graphbased compound anomaly diagnosis system for distributed rdbmss,” Proceedings of the VLDB Endowment, vol. 18, no. 12, pp. 5383–5386, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "X. Zhu, S. Wen, S. Camtepe, and Y. Xiang, “Fuzzing: a survey for roadmap,” ACM Computing Surveys (CSUR), vol. 54, no. 11s, pp. 1–36, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “Apollo: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment, vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Q. Xiang, Y. Shao, Q. Xu, and C. Yang, “Distributed database diagnosis method for compound anomalies.” International Journal of Software & Informatics, vol. 15, no. 1, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "“Sqlancer,” https://github.com/sqlancer/sqlancer, 2019, accessed on July, 7, 2025.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 12,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 13,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of {DBMS} with coverage-based guidance,” in 31st USENIX Security Symposium (USENIX Security 22), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing database management systems with language validity and coverage feedback,” in Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "S. Feng and C. Chen, “Prompting is all you need: Automated android bug replay with large language models,” in Proceedings of the 46th IEEE/ACM International Conference on Software Engineering, 2024, pp. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "A. Koyuncu, K. Liu, T. F. Bissyand ´e, D. Kim, M. Monperrus, J. Klein, and Y. Le Traon, “ifixr: Bug report driven program repair,” in Proceedings of the 2019 27th ACM joint meeting on european software engineering conference and symposium on the foundations of software engineering, 2019, pp. 314–325.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "L. Plein, W. C. Ou ´edraogo, J. Klein, and T. F. Bissyand ´e, “Automatic generation of test cases based on bug reports: a feasibility study with large language models,” in Proceedings of the 2024 IEEE/ACM 46th International Conference on Software Engineering: Companion Proceedings, 2024, pp. 360–361.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Q. Gu, “Llm-based code generation method for golang compiler testing,” inProceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2023, pp. 2201–2203.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "F. Lin, D. J. Kim, and T.-H. Chen, “When llm-based code generation meets the software development process,” CoRR, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "W. Xu, C. Huang, S. Gao, and S. Shang, “Llm-based agents for tool learning: A survey: W. xu et al.” Data Science and Engineering, pp. 1–31, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "I. Neamtiu, J. S. Foster, and M. Hicks, “Understanding source code evolution using abstract syntax tree matching,” in Proceedings of the 2005 international workshop on Mining software repositories, 2005, pp. 1–5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "S. Tang, S. Liu, J. Wang, and X. Zhang, “An empirical study on ast-level mutation-based fuzzing techniques for javascript engines,” in Proceedings of the 14th Asia-Pacific Symposium on Internetware, 2023, pp. 216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "H. Han, D. Oh, and S. K. Cha, “Codealchemist: Semantics-aware code generation to find vulnerabilities in javascript engines.” in Network and Distributed System Security Symposium (NDSS), 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Y. Chen, R. Zhong, H. Hu, H. Zhang, Y. Yang, D. Wu, and W. Lee, “One engine to fuzz’em all: Generic language processor testing with semantic validation,” in (SP). IEEE, 2021, pp. 642–658.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "S. Park, W. Xu, I. Yun, D. Jang, and T. Kim, “Fuzzing javascript engines with aspect-preserving mutation,” in and Privacy (SP). IEEE, 2020, pp. 1629–1642.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "W. Xu, H. Moon, S. Kashyap, P.-N. Tseng, and T. Kim, “Fuzzing file systems via two-dimensional input space exploration,” in 2019 IEEE Symposium on Security and Privacy (SP). IEEE, 2019, pp. 818–834.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "M. Xu, S. Kashyap, H. Zhao, and T. Kim, “Krace: Data race fuzzing for kernel file systems,” in (SP). IEEE, 2020, pp. 1643–1660.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "S. Pailoor, A. Aday, and S. Jana, “{MoonShine}: Optimizing {OS} fuzzer seed selection with trace distillation,” in 27th USENIX Security Symposium (USENIX Security 18), 2018, pp. 729–743.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "K. Kim, D. R. Jeong, C. H. Kim, Y. Jang, I. Shin, and B. Lee, “Hfl: Hybrid fuzzing on the linux kernel.” in Network and Distributed System Security Symposium (NDSS), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "J. Choi, K. Kim, D. Lee, and S. K. Cha, “Ntfuzz: Enabling type-aware kernel fuzzing on windows with static binary analysis,” in 2021 IEEE Symposium on Security and Privacy (SP). IEEE, 2021, pp. 677–693.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang, and Y. Jiang, “Industry practice of coverage-guided enterprise-level dbms fuzzing,” in ing: Software Engineering in Practice (ICSE-SEIP). IEEE, 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "H. Liu, M. Shen, J. Jin, and Y. Jiang, “Automated classification of actions in bug reports of mobile apps,” in Proceedings of the 29th ACMSIGSOFT International Symposium on Software Testing and Analysis, 2020, pp. 128–140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "“Llmsqlmutator: Llm-powered test case generation for database using bug reports (tecnical report),” https://github.com/DBMSTesting/ LLMSQLMutator-Technical-Report, 2026, accessed on January 26, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "T. J. Parr and R. W. Quong, “Antlr: A predicated-ll (k) parser generator,” Software: Practice and Experience, vol. 25, no. 7, pp. 789–810, 1995.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "X. Zhou, Z. Sun, and G. Li, “Db-gpt: Large language model meets database,” Data Science and Engineering, vol. 9, no. 1, pp. 102–111, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "F. Cuconasu, G. Trappolini, F. Siciliano, S. Filice, C. Campagnano, Y. Maarek, N. Tonellotto, and F. Silvestri, “The power of noise: Redefining retrieval for rag systems,” in Proceedings of the 47th International ACMSIGIR Conference on Research and Development in Information Retrieval, 2024, pp. 719–729.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "B. Jin, J. Yoon, J. Han, and S. O. Arik, “Long-context llms meet rag: Overcoming challenges for long inputs in rag,” in The Thirteenth International Conference on Learning Representations.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "J. Gu, X. Jiang, Z. Shi, H. Tan, X. Zhai, C. Xu, W. Li, Y. Shen, S. Ma, H. Liu et al., “A survey on llm-as-a-judge,” CoRR, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "C. Whitehouse, T. Wang, P. Yu, X. Li, J. Weston, I. Kulikov, and S. Saha, “J1: Incentivizing thinking in llm-as-a-judge via reinforcement learning,” arXiv preprint arXiv:2505.10320, 2025. 3263",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "S. Saha, X. Li, M. Ghazvininejad, J. E. Weston, and T. Wang, “Learning to plan & reason for evaluation with thinking-llm-as-a-judge,” in Fortysecond International Conference on Machine Learning.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "L. Zheng, W.-L. Chiang, Y. Sheng, S. Zhuang, Z. Wu, Y. Zhuang, Z. Lin, Z. Li, D. Li, E. Xing et al., “Judging llm-as-a-judge with mt-bench and chatbot arena,” Advances in neural information processing systems, vol. 36, pp. 46 595–46 623, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "J. Ba and M. Rigger, “Cert: Finding performance issues in database systems through the lens of cardinality estimation,” in Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–13.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 45,
+      "text": "“Oceanbase,” https://en.oceanbase.com/, 2016, accessed on July 7, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "A. Liu, B. Feng, B. Xue, B. Wang, B. Wu, C. Lu, C. Zhao, C. Deng, C. Zhang, C. Ruan et al., “Deepseek-v3 technical report,” CoRR, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "“Codellama-34b,” https://huggingface.co/codellama/CodeLlama-34b-hf, 2023, accessed on July 7, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "“Gpt-3.5-turbo,” https://platform.openai.com/docs/models/gpt-3.5-turbo, 2022, accessed on July 7, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "“Gpt-4o-mini,” https://openai.com/index/ gpt-4o-mini-advancing-cost-efficient-intelligence/, 2024, accessed on July 7, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "D. R. Slutz, “Massive stochastic testing of sql,” in VLDB, vol. 98. Citeseer, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "B. Ghit, N. Poggi, J. Rosen, R. Xin, and P. Boncz, “Sparkfuzz: Searching correctness regressions in modern query engines,” in Proceedings of the workshop on Testing Database Systems, 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Z. Cui, W. Dou, Y. Gao, R. Yang, Y. Zheng, J. Song, Y. Feng, and J. Wei, “Simple testing can expose most critical transaction bugs: Understanding and detecting write-specific serializability violations in database systems,” Proceedings of the VLDB Endowment (VLDB), 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "C. Zhang, G. Li, L. Liu, T. Lv, and J. Fan, “Cloudybench: A testbed for a comprehensive evaluation of cloud-native databases,” in 2025 IEEE 41st International Conference on Data Engineering (ICDE). IEEE Computer Society, 2025, pp. 2535–2547.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "X. Chen, C. Wang, and A. Cheung, “Testing query execution engines with mutations,” in Proceedings of the workshop on Testing Database Systems, 2020, pp. 1–5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Z. Zhuang, P. Li, P. Ma, W. Meng, and S. Wang, “Testing graph database systems via graph-aware metamorphic relations,” Proceedings of the VLDB Endowment, vol. 17, no. 4, pp. 836–848, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "F. Panse and F. Naumann, “Evaluation of duplicate detection algorithms: From quality measures to test data generation,” in 2021 IEEE 37th International Conference on Data Engineering (ICDE). IEEE, 2021, pp. 2373–2376.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "J. Castelein, M. Aniche, M. Soltani, A. Panichella, and A. van Deursen, “Search-based test data generation for sql queries,” in Proceedings of the 40th international conference on software engineering, 2018, pp. 1220–1230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammarfree dbms fuzzing,” in Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "M. Fazzini, M. Prammer, M. d’Amorim, and A. Orso, “Automatically translating bug reports into test cases for mobile apps,” in Proceedings of the 27th ACMSIGSOFT International Symposium on Software Testing and Analysis, 2018, pp. 141–152.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Y. Zhao, T. Yu, T. Su, Y. Liu, W. Zheng, J. Zhang, and W. G. Halfond, “Recdroid: automatically reproducing android application crashes from bug reports,” in Software Engineering (ICSE). IEEE, 2019, pp. 128–139.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "O. Chaparro, J. Lu, F. Zampetti, L. Moreno, M. Di Penta, A. Marcus, G. Bavota, and V. Ng, “Detecting missing information in bug descriptions,” in Proceedings of the 2017 11th joint meeting on foundations of software engineering, 2017, pp. 396–407.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "M. Fazzini, K. Moran, C. Bernal-Cardenas, T. Wendland, A. Orso, and D. Poshyvanyk, “Enhancing mobile app bug reporting via real-time understanding of reproduction steps,” IEEE Transactions on Software Engineering, vol. 49, no. 3, pp. 1246–1272, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Y. Song, J. Mahmud, Y. Zhou, O. Chaparro, K. Moran, A. Marcus, and D. Poshyvanyk, “Toward interactive bug reporting for (android app) end-users,” in Proceedings of the 30th ACM joint european software engineering conference and symposium on the foundations of software engineering, 2022, pp. 344–356. 3264",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 11,
+      "text": "“Sqlancer,” https://github.com/sqlancer/sqlancer, 2019, accessed on July, 7, 2025.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 12,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 13,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 44,
+      "text": "J. Ba and M. Rigger, “Cert: Finding performance issues in database systems through the lens of cardinality estimation,” in Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–13.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[11]",
+      "technique": null,
+      "sentence": "Generally, the automated DBMS testing process comprises two phases: test case generation [6], [11] and test oracle construction [12], [13], [14], [15].",
+      "context_before": "evolving to support increasingly rich SQL syntax and functionalities like window functions, JSON operations, and GIS spatial data processing [7]. This inherent complexity combined with ongoing dynamic expansion inevitably introduces various defects, which lead to severe consequences like inconsistent query results, service interruptions, and performance degradation [8], [9], [10]. As a result, systematic, automated, and efficient testing of DBMSs become a critical approach to ensure their reliability and facilitate early bug detection and remediation within the software development life-cycle.",
+      "context_after": "The test case generation phase produces large volumes of structurally diverse SQL statements, and the test oracle construction phase determines whether theexecution results on the tested DBMS are correct, thereby identifying potential bugs. According to previous studies [15], [9], the effectiveness of DBMS fuzzing techniques is critically dependent on the quality of the SQL statements generated during the test case generation phase. Therefore, optimizing the methods for test case generation is vital to ensure the production of such high-quality test cases. Existing techniques for test case ge",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2740,
+      "cited_reference": {
+        "number": 11,
+        "text": "“Sqlancer,” https://github.com/sqlancer/sqlancer, 2019, accessed on July, 7, 2025.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "2) For test cases containing SELECT queries, we employ result inconsistency detection by randomly selecting one test oracle from TLP [13], NoREC [12], or CERT [44].",
+      "context_before": "MS to conduct comprehensive bug detection based on execution outcomes. Our detection methodology implements a two-tiered validation system: 1) For all test cases, we perform direct execution fault monitoring by feeding mutated queries into the DBMS to capture runtime abnormalities. This process identifies execution-triggered errors or crashes, including but not limited to memory operation exceptions (e.g., buffer overflows), core executor faults within query optimizers, concurrency failures such as lock mechanism breakdowns, and other system-level anomalies that manifest during SQL processing.",
+      "context_after": "The selected oracle transforms the original query into semantically equivalent variants while preserving all non-query statements intact. Both the original and transformed statement sequences are then executed against the target DBMS. Any discrepancies between their result sets indicate potential logical bugs. We subsequently conduct thorough execution log analysis to verify the presence of database bugs, with the entire process maintaining the original test case structure during equivalence transformation to ensure accurate differential diagnosis. E. Discussion of the Hybrid Design of LLMSQL",
+      "section": "D DBMS Testing",
+      "page": 8,
+      "char_offset": 41134,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec",
+        "cert"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[12]",
+      "technique": "norec",
+      "sentence": "It is important to note that we used the latest release of each RDBMS, which has been extensively tested by existing approaches [12], [13].",
+      "context_before": "ines GitHub stars MySQL 2 5.0K 380K 1995 8.1.0 4209 TiDB 118 23.1K 800K 2017 7.5.1 8337 SQLite 9 1.5K 300K 2000 3.46.0 433 DuckDB-0.5K 59K 2018 1.1.3 758 Clickhouse 30 41.8K 125K 2016 24.12.2 2163 OceanBase 119 8.3K 300K 2015 4.3.2 148 We conducted all experiments on a server with two Intel(R) Xeon(R) Platinum 5320 CPUs at 2.20 GHz and 251 GB of memory, running Ubuntu 20.04.5 LTS. We ran the experiments using Python version 3.12. Tested RDBMS. We selected six popular and widely used RDBMSs. The statistics of these RDBMSs, as obtained from their open-source repositories, are shown in Table III.",
+      "context_after": "MySQL [1] is the most popular open-source database management systems. SQLite [3] and DuckDB [5] are both embedded DBMSs, running within the process of other applications. TiDB [2], Clickhouse [4] and OceanBase [45] are popular distributed RDBMSs. Table III also lists the number of bug reports we collected for each database. We collected them based on the tags provided by the developers to determine whether they are bugs or not. The core filtering criteria require that the report must include at least one executable SQL statement that can reproduce the unexpected behavior of the RDBMS. LLMs f",
+      "section": "IV EVALUATION",
+      "page": 8,
+      "char_offset": 43836,
+      "cited_reference": {
+        "number": 12,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "We selected the three most effective existing oracles for detecting logic bugs through result inconsistency: NoREC [12], TLP [13], and CERT [44].",
+      "context_before": "ment that can reproduce the unexpected behavior of the RDBMS. LLMs for mutation. We chose the following four representative LLMs for comparative experiments: the efficient open-source model deepseek-v3 [46], the specialized code generation model codellama-34B [47], the industrial grade general-purpose model gpt-3.5-turbo [48], and the lightweight closed source model gpt-4o-mini [49]. The selection covers different technical routes such as open source/closed source, dedicated/general, aiming to comprehensively evaluate the performance of LLM in mutation tasks. Test oracle for result comparison.",
+      "context_after": "These approaches have all demonstrated significant effectiveness in detecting bugs within RDBMS implementations. Calculation method for bug pattern coverage. To evaluate the effectiveness of test case generation, we introduce a new metric called bug pattern coverage, which measures the proportion of known bug patterns covered by the generated SQL queries. This metric is computed using the formula: Pattern coverage =|Pknown∩Ptest||Pknown|×100% (1) where Pknown denotes the set of known bug patterns (previously extracted from historical bug reports as sequences of SQL keywords, operators, and fu",
+      "section": "IV EVALUATION",
+      "page": 8,
+      "char_offset": 45108,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "cert"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Of these 14 result inconsistency bugs, 12 bugs are detected by the TLP oracle, and the other two are found by the CERT oracle.",
+      "context_before": "structures. B. Experiment results 1) Effectiveness of bug detection We ran LLMSQLM UTATOR on six RDBMSs and reported the detected issues to the corresponding developer communities. Table IV shows the details (e.g., types, bug patterns, and root causes) of the bugs in six RDBMSs detected by LLMSQLM UTATOR. We have detected 27 bugs during this work and all of them are confirmed by the developers. Among these bugs, 13 triggered DBMS errors during SQL statement execution. The remaining 14 bugs produce execution results, but those results deviated from expectations. We call them inconsistency bugs.",
+      "context_after": "It is noteworthy that all six databases have undergone extensive testing with existing methods, yet LLMSQLM UTATOR is still able to detect these previously undetected errors, including seven unique bug patterns not present in historical reports (MySQL 113176, DuckDB 17286, TiDB 59662&61074&61075, Oceanbase 2104, Clickhouse 79911). These novel findings—spanning constraint validation, recursive queries, complex joins, and schema modification operations—demonstrate our tool’s ability to uncover defects beyond simple variations of known issues. This capability stems from our controlled mutation s",
+      "section": "B Experiment results",
+      "page": 9,
+      "char_offset": 46734,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "cert"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[13]",
+      "technique": "tlp",
+      "sentence": "Existing fuzzers typically restrict recursion depth to maintain syntactic correctness [15], [13], thereby failing to reach the parser’s threshold.",
+      "context_before": "’), ’ d ’,’ e ’), ’ e ’,’ f ’), ’ f ’, ’ g ’), ’ g ’,’ h ’), ’ h ’,’ i ’), ’ i ’,’ j ’), ’ j ’,’ k ’), ’ k ’, ’ l ’), ’ l ’,’m’), ’m’,’ n ’), ’ n ’,’ o ’), ’ o ’,’ p ’), ’ p ’, ’ q ’), ’ q ’,’ r ’), ’ r ’,’ s ’), ’ s ’,’ t ’), ’ t ’,’ u ’), ’ u ’, ’ v ’), ’ v ’,’w’), ’w’,’ x ’), ’ x ’,’ y ’), ’ y ’,’ z ’), (3, ’ Record w ith JSON: ’ || json_object( ’ keyA ’ ,’ valueA ’, ’ keyB ’ ,’ valueB ’ )) 3 4Error Message:parser stack overflow Fig. 8: A bug in SQLite with the test SQLs Case Study of the Bugs. Figure 8 illustrates a parser stack overflow in SQLite triggered by 32 nested REPLACE functions.",
+      "context_after": "In contrast, LLMSQLM UTATOR1CREATE TABLE t (a INT NOT NULL, b INTUNIQUE); 2CREATE TABLE t1 (a INT NOT NULL, b INTDEFAULT 0); 3INSERT INTO t (a, b) VALUES (1, 100), (2, 200), (3, 300); 4Error executing SQL: 5EXPLAIN ANALYZE SELECT SUM(CASEWHEN t.a > 0 THEN (SELECT SUM(b) FROM t1WHERE t.a = t1.a GROUP BY t.a HAVING COUNT( *) > 1) ELSE 0END) FROM tLEFT JOIN t1ONt.a = t1.a WHERE EXISTS (SELECT 1FROM t t2 WHERE t2.b < 250) GROUP BY t.a WITHROLLUP 6Error Message:(1815, ” I n t e r n a l: Can ’ t f i n d a p r o p e r p h y s i c a l p l a n f o r t h i s query ” ) Fig. 9: A bug in TiDB with the tes",
+      "section": "B Experiment results",
+      "page": 9,
+      "char_offset": 49066,
+      "cited_reference": {
+        "number": 13,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "2) Comparsion of baselines We compare LLMSQLM UTATOR with state-of-the-art DBMS testing approaches, including mutation-based methods SQLRight [15] and Squirrel [16], as well as grammar-based SQLancer [11] and random-based SQLsmith [6] to provide a comprehensive empirical comparison.",
+      "context_before": "raints, our LLM-driven SQL mutation intelligently explores the semantically related but syntactically distinct domain of CHECK constraints, systematically constructing multi-level nested expressions that expose deep parser flaws— the parser incorrectly interpreted certain constant expressions within complex CHECK constraints as cross-column references, leading to improper constraint rejection. This contrast highlights our tool’s ability to transcend conventional testing boundaries by generating sophisticated constraint constructs that reveal previously undetectable defects in database engines.",
+      "context_after": "2We conducted 48-hour experiments for all comparable baselines, recording test case correctness, bug pattern coverage, and bug detection numbers to ensure a fair evaluation. 2Due to architectural limits, SQLRight and Squirrel are limited to MySQL and SQLite, while SQLsmith supports only SQLite. Unsupported results are marked “–” as extension requires major re-engineering. 3259 TABLE IV: Bugs detected by LLMSQLM UTATOR SNID Target Type Bug pattern Root cause 1 118145 MySQL Error CREATE, INDEX, ENGINE=Archive Archive engine: Index validation missing during table creation 2 113176 MySQL Error C",
+      "section": "B Experiment results",
+      "page": 9,
+      "char_offset": 51825,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3259 TABLE IV: Bugs detected by LLMSQLM UTATOR SNID Target Type Bug pattern Root cause 1 118145 MySQL Error CREATE, INDEX, ENGINE=Archive Archive engine: Index validation missing during table creation 2 113176 MySQL Error CREATE, CHECK, NOT, IN, IS FALSE Error with null expression in check condition 3 109149 MySQL Incons MOD Decimal arithmetic error in modulus operation 4 17286 DuckDB Error WWITH RECURSIVE, SUBSTR, POSITION Recursive CTE: Column binding failure with nested queries 5 15731 DuckDB Incons IS NULL, UNION ALL NULL operand mishandling in comparison logic 6 15742 DuckDB Incons <, UNION ALL NULL operand mishandling in comparison logic 7 15745 DuckDB Incons BETWEEN, DATE, UNION ALL NULL operand mishandling in date range checks 8 17288 DuckDB Error WITH, EXISTS CTE scope: Column binding failure on empty strings 9 6804 DuckDB Incons NOT On abnormal results of the not operator 10 59662 TiDB Incons INNER JOIN, REGEXP, CASE, WHEN Join row estimation flaw (CERT oracle) 11 59664 TiDB Incons GROUP BY, HAVING, CAST Join row estimation flaw (CERT oracle) 12 60880 TiDB Error EXPLAIN ANALYZE, WITHROLLUP Optimizer: ROLLUP physical plan not found 13 61074 TiDB Error JSON˙OBJECT, GROUP BY, WITHROLLUP Optimizer: ROLLUP+JSON combination unsupported 14 61075 TiDB Error YEAR, HAVING, WITHROLLUP Optimizer: ROLLUP+ dynamic function failure 15 58960 TiDB Incons RIGHT JOIN, BIT˙LENGTH JOIN column reference missing in function 16 58961 TiDB Incons LEFT JOIN, FIELD NULL operand mishandling in field function 17 59669 TiDB Incons LEFT JOIN, FIELD, CASE, WHEN NULL operand mishandling in CASE expression 18 60886 TiDB Error WITHRECURSIVE, RECOVER Historical read: Missing GC safe-point data 19 60882 TiDB Error LOADDATA Protocol implementation incomplete 20 60887 TiDB Error <==>, ALL NULL-safe operator unsupported with ALL 21 61002 TiDB Error INNER JOIN, DEFAULT, CASE, WHEN Type-mismatch boundary condition crash 22 61313 TiDB Incons NATURAL RIGHT JOIN, GROUP BY JOIN column reference missing in grouping 23 ab39bc73ed SQLite Error REPLACE, JSON˙OBJECT Parser stack overflow: Deeply nested REPLACE 24 6ca4e1c1d2 SQLite Incons BETWEEN, CAST Type affinity conflict (NONE/INTEGER) in view 25 2104 Oceanbase Incons SELECT, BIT Operator Bitwise ops: Incorrect negative number handling 26 2105 Oceanbase Incons MOD Decimal arithmetic error in modulus operation 27 79911 Clickhouse Error ADDCOLUMN, INSERT Schema validation: DateTime precedence over column count TABLE V: The correctness of test case generated by SQLsmith, SQLancer, SQLRight, Squirrel and LLMSQLM UTATOR in 48h DBMSSQLsmith SQLancer Squirrel SQLRight LLMSQLM UTATOR Syn Sem Syn Sem Syn Sem Syn Sem Syn Sem MySQL - - 99.",
+      "context_before": "state-of-the-art DBMS testing approaches, including mutation-based methods SQLRight [15] and Squirrel [16], as well as grammar-based SQLancer [11] and random-based SQLsmith [6] to provide a comprehensive empirical comparison.2We conducted 48-hour experiments for all comparable baselines, recording test case correctness, bug pattern coverage, and bug detection numbers to ensure a fair evaluation. 2Due to architectural limits, SQLRight and Squirrel are limited to MySQL and SQLite, while SQLsmith supports only SQLite. Unsupported results are marked “–” as extension requires major re-engineering.",
+      "context_after": "54% 97.76% 71.45% 19.32% 77.21% 27.09% 90.05% 71.20% SQLite 13.76% 2.83% 92.07% 78.12% 91.56% 35.76% 95.23% 40.10% 96.79% 75.13% TiDB - - 90.68% 77.32% - - - - 90.62% 72.84% DuckDB - - 90.56% 72.37% - - - - 90.97% 72.84% OceanBase - - 91.17% 79.34% - - - - 86.41% 50.89% Clickhouse - - 96.21% 82.14% - - - - 89.44% 59.88% Test case correctness. Table V shows the correctness of test cases generated by five approaches. For mutationbased tools SQLRight and Squirrel, LLMSQLM UTATOR demonstrates exceptional syntactic and semantic correctness across all tested RDBMSs, comprehensively outperforming bas",
+      "section": "B Experiment results",
+      "page": 9,
+      "char_offset": 52483,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As a grammar-based tool, SQLancer demonstrates consistently high syntactic correctness across all six tested DBMSs,with rates exceeding 90% in every case and reaching 99.",
+      "context_before": "ic correctness (50.89%–75.13%) surpassed the baselines by 2–3×, exemplified by its 71.20% rate in MySQL—far exceeding SQLRight’s 27.09% and Squirrel’s 19.32%—which is attributed to the LLM-based semantic validator’s dynamic validation capabilities for metadata and fine-grained constraints. Notably, the relatively lower semantic correctness in OceanBase and ClickHouse (50.89% and 59.88% respectively) stems primarily from dialect-specific challenges posed by their distributed architectures to LLM semantic comprehension, such as OceanBase’s sharding syntax and ClickHouse’s column store semantics.",
+      "context_after": "54% on MySQL—the highest among all approaches. This exceptional performance stems from its strict adherence to predefined grammar rules, which ensures syntactic validity but simultaneously limits semantic diversity and exploration of unconventional SQL constructs. Moreover, this grammar-based approach requires manual implementation of dialect-specific grammar for each DBMS, incurring significant engineering overhead and poor scalability when extending to new database systems. In contrast, our method employs AST-based structural hollowing to abstract away dialect-specific syntax variations whil",
+      "section": "B Experiment results",
+      "page": 10,
+      "char_offset": 56649,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "TABLE VI: The bug number and bug pattern covered by SQLsmith, SQLancer, SQLRight, Squirrel and LLMSQLM UTATOR in 48h DBMSSQLsmith SQLancer Squirrel SQLRight LLMSQLM UTATOR #Bug Pattern #Bug Pattern #Bug Pattern #Bug Pattern #Bug Pattern MySQL - - 1 23.",
+      "context_before": "t-specific grammar for each DBMS, incurring significant engineering overhead and poor scalability when extending to new database systems. In contrast, our method employs AST-based structural hollowing to abstract away dialect-specific syntax variations while preserving logical equivalence, enabling natural adaptation to multiple DBMSs with minimal reconfiguration effort. Meanwhile, purely random-based tool, SQLsmith, exhibits critically low validity (13.76% syntactic and 2.83% semantic correctness on SQLite), demonstrating the inherent limitations of blind generation without semantic guidance.",
+      "context_after": "7% 1 24.5% 1 55.5% 3 71.8% SQLite 0 5.42% 1 31.4% 0 28.4% 0 65.8% 1 86.6% TiDB - - 3 52.0% - - - - 8 79.9% DuckDB - - 2 18.2% - - - - 6 55.5% OceanBase - - 0 14.9% - - - - 1 68.2% Clickhouse - - 0 18.7% - - - - 1 40.4% Bug pattern and bug number. Table VI shows the number of detected bugs and the coverage of bug patterns by five approaches. Compared to mutation-based tools SQLRight and Squirrel, LLMSQLM UTATOR demonstrates superior performance in both bug pattern coverage and detection counts. 3260 It achieves 71%–86.6% coverage across MySQL, SQLite, and TiDB, surpassing baselines by over 16",
+      "section": "B Experiment results",
+      "page": 10,
+      "char_offset": 57763,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In contrast, grammar-based SQLancer and random-based tool SQLsmith exhibit significant limitations.",
+      "context_before": "erage of bug patterns by five approaches. Compared to mutation-based tools SQLRight and Squirrel, LLMSQLM UTATOR demonstrates superior performance in both bug pattern coverage and detection counts. 3260 It achieves 71%–86.6% coverage across MySQL, SQLite, and TiDB, surpassing baselines by over 16%, although coverage is lower on complex systems like OceanBase. Regarding bug counts, LLMSQLM UTATOR uncovers critical faults—including MySQL engine flaws and SQLite stack overflows—that were undetected by baselines due to their limited seed pools and incapacity to generate deeply nested structures.",
+      "context_after": "SQLancer attains only modest coverage (peaking at 52.0% on TiDB) with few detected bugs, while SQLsmith fails to identify any distinct bugs, showing negligible coverage (5.42% on SQLite). These results underscore that traditional approaches lack the semantic depth and fault-triggering potential required for effective testing, whereas LLMSQLM UTATOR successfully probes vulnerable code paths through semantically rich query generation. 3) Ablation Study of LLMSQLM UTATOR We designed experiments to investigate the effectiveness of mutation guidance and semantic validation. Table VII shows the bug",
+      "section": "B Experiment results",
+      "page": 11,
+      "char_offset": 58917,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer attains only modest coverage (peaking at 52.",
+      "context_before": "MSQLM UTATOR demonstrates superior performance in both bug pattern coverage and detection counts. 3260 It achieves 71%–86.6% coverage across MySQL, SQLite, and TiDB, surpassing baselines by over 16%, although coverage is lower on complex systems like OceanBase. Regarding bug counts, LLMSQLM UTATOR uncovers critical faults—including MySQL engine flaws and SQLite stack overflows—that were undetected by baselines due to their limited seed pools and incapacity to generate deeply nested structures. In contrast, grammar-based SQLancer and random-based tool SQLsmith exhibit significant limitations.",
+      "context_after": "0% on TiDB) with few detected bugs, while SQLsmith fails to identify any distinct bugs, showing negligible coverage (5.42% on SQLite). These results underscore that traditional approaches lack the semantic depth and fault-triggering potential required for effective testing, whereas LLMSQLM UTATOR successfully probes vulnerable code paths through semantically rich query generation. 3) Ablation Study of LLMSQLM UTATOR We designed experiments to investigate the effectiveness of mutation guidance and semantic validation. Table VII shows the bug patterns covered and the number of bugs triggered by",
+      "section": "B Experiment results",
+      "page": 11,
+      "char_offset": 59017,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "citation_marker",
+      "surface": "[14]",
+      "technique": "pqs",
+      "sentence": "Metamorphic testing is another mainstream approaches for RDB-MS testing [51], [54], [14], [12], [13], [55], [56], [15].",
+      "context_before": "for database testing. Ratel [33] improves the robustness of SQL generation for database testing by merging SQL dictionaries with grammar-based mutations. SparkFuzz [51] introduces a fuzzing-based method that utilizes the query results from a reference database as test oracles. These methods [33], [52], [53], [51] perform result consistency checks by inputting the same SQL statement into different DBMS systems. The effectiveness of these methods is limited by syntax supported across the databases under test. They also may yield false positives due to the varied implementation choices of RDBMSs.",
+      "context_after": "MUTASQL [54] and Eqsql [57] construct test cases by defining mutation rules, which are used to generate or synthesize SQL query statements that are functionally equivalent to the original ones. In recent years, SQLancer [11] has emerged as the most effective black-box fuzzing tool, distinguished by its adoption of three complementary oracles [14], [12], [13]. SQLRight [15] focuses on enhancing the semantic correctness of generated SQL queries. GRIFFIN [58] executes mutation testing within the grammatical boundaries of SQL language. Squirrel [16] introduces a test case generation framework bas",
+      "section": "V RELATED WORK",
+      "page": 12,
+      "char_offset": 65864,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In recent years, SQLancer [11] has emerged as the most effective black-box fuzzing tool, distinguished by its adoption of three complementary oracles [14], [12], [13].",
+      "context_before": "perform result consistency checks by inputting the same SQL statement into different DBMS systems. The effectiveness of these methods is limited by syntax supported across the databases under test. They also may yield false positives due to the varied implementation choices of RDBMSs. Metamorphic testing is another mainstream approaches for RDB-MS testing [51], [54], [14], [12], [13], [55], [56], [15]. MUTASQL [54] and Eqsql [57] construct test cases by defining mutation rules, which are used to generate or synthesize SQL query statements that are functionally equivalent to the original ones.",
+      "context_after": "SQLRight [15] focuses on enhancing the semantic correctness of generated SQL queries. GRIFFIN [58] executes mutation testing within the grammatical boundaries of SQL language. Squirrel [16] introduces a test case generation framework based on syntax mutation and semantic statement filling, and tests DBMS through coverage guidance. The seed SQL of the above methods all come from the testing suite provided by the developers, and the mutation process focuses on the comprehensiveness of detection, that is, triggering more SQL branches. However, the efficiency of these methods in detecting bugs is",
+      "section": "V RELATED WORK",
+      "page": 12,
+      "char_offset": 66178,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "describes_as_state_of_the_art": [
+        "M4",
+        "M14"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:57:35Z",
+    "is_model_written": true,
+    "summary": "LLMSQLMutator generates DBMS test cases by having a language model mutate SQL drawn from historical bug reports, on the argument that reports encode the shapes that have already broken systems. It reaches constructs grammar-based fuzzers avoid -- deep recursion, for instance, which they cap to stay syntactically correct -- and found bugs across six relational DBMSs.",
+    "narrative": "SQLancer is both a baseline and a component. As a baseline it is one of four tools compared over 48 hours, and the paper is even-handed: SQLancer's syntactic correctness exceeds 90% on every system tested, higher than the mutation-based tools, while its coverage peaks around 52%. It is called the most effective black-box fuzzing tool, distinguished by its three complementary oracles. Those oracles are also reused directly: for test cases containing SELECT queries, LLMSQLMutator randomly selects one of TLP, NoREC or CERT as its oracle, describing them as the three most effective existing oracles for detecting result inconsistency. Of its 14 result-inconsistency bugs, TLP found 12 and CERT the other two.",
+    "roles": {
+      "M1": "background",
+      "M2": "reuse_component",
+      "M3": "background",
+      "M4": "reuse_component",
+      "M5": "result_comparison",
+      "M6": "motivation",
+      "M7": "baseline",
+      "M8": "incidental",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "result_comparison",
+      "M12": "result_comparison",
+      "M13": "background",
+      "M14": "state_of_the_art"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M2",
+          "M4",
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "2) For test cases containing SELECT queries, we employ result inconsistency detection by randomly selecting one test oracle from TLP [13], NoREC [12], or CERT [44].",
+            "section": "D DBMS Testing",
+            "page": 8
+          },
+          {
+            "mention_id": "M4",
+            "sentence": "We selected the three most effective existing oracles for detecting logic bugs through result inconsistency: NoREC [12], TLP [13], and CERT [44].",
+            "section": "IV EVALUATION",
+            "page": 8
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "Of these 14 result inconsistency bugs, 12 bugs are detected by the TLP oracle, and the other two are found by the CERT oracle.",
+            "section": "B Experiment results",
+            "page": 9
+          }
+        ],
+        "reasoning": "M2 states that for SELECT test cases the tool selects one oracle at random from TLP, NoREC or CERT, M4 names those as the oracles chosen, and M5 reports which of them found each bug. The oracles are used as-is inside the tool, but the paper does not say whether they are invoked through SQLancer or reimplemented, so the form of the reuse is unclear.",
+        "reuse_kind": "unclear"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The three oracles are used as they are; nothing is generalised or adapted. The contribution is the LLM-driven mutation that produces the test cases."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M7",
+          "M10",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M7",
+            "sentence": "2) Comparsion of baselines We compare LLMSQLM UTATOR with state-of-the-art DBMS testing approaches, including mutation-based methods SQLRight [15] and Squirrel [16], as well as grammar-based SQLancer [11] and random-based SQLsmith [6] to provide a comprehensive empirical comparison.",
+            "section": "B Experiment results",
+            "page": 9
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "TABLE VI: The bug number and bug pattern covered by SQLsmith, SQLancer, SQLRight, Squirrel and LLMSQLM UTATOR in 48h DBMSSQLsmith SQLancer Squirrel SQLRight LLMSQLM UTATOR #Bug Pattern #Bug Pattern #Bug Pattern #Bug Pattern #Bug Pattern MySQL - - 1 23.",
+            "section": "B Experiment results",
+            "page": 10
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "SQLancer attains only modest coverage (peaking at 52.",
+            "section": "B Experiment results",
+            "page": 11
+          }
+        ],
+        "reasoning": "M7 names SQLancer among the state-of-the-art approaches compared against, M10 gives the 48-hour bug and pattern table, and M12 reports its coverage."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M14"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M14",
+            "sentence": "In recent years, SQLancer [11] has emerged as the most effective black-box fuzzing tool, distinguished by its adoption of three complementary oracles [14], [12], [13].",
+            "section": "V RELATED WORK",
+            "page": 12
+          }
+        ],
+        "reasoning": "M14 states that SQLancer has emerged as the most effective black-box fuzzing tool, distinguished by its adoption of three complementary oracles."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icicnis64247_2024_10823213.json b/_data/papers/paper_doi_10_1109_icicnis64247_2024_10823213.json
new file mode 100644
index 0000000..73d9423
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icicnis64247_2024_10823213.json
@@ -0,0 +1,320 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:54:42Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icicnis64247.2024.10823213",
+    "title": "Dynamic Adjustment Paradigm based on Genetic Algorithm in Database Index Optimization",
+    "authors": [
+      "Jinsong Wang"
+    ],
+    "year": 2024,
+    "venue": "2024 International Conference on IoT Based Control Networks and Intelligent Systems (ICICNIS)",
+    "doi": "10.1109/icicnis64247.2024.10823213",
+    "arxiv_id": null,
+    "s2_paper_id": "c30c758ad17b50b6de00ac4878065a4436e07de0",
+    "url": "https://doi.org/10.1109/icicnis64247.2024.10823213",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icicnis64247.2024.10823213",
+      "retrieved_at": "2026-09-08T14:54:42Z",
+      "chars": 26718,
+      "content_sha256": "sha256:725791b794b560dffde66cc76a78e13dde2f8ba2430115f56eb6bf6d7f15ac15"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 7,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2043
+      },
+      {
+        "number": "II",
+        "title": "REVIEW OFDATABASE OPTIMIZATION STUDIES",
+        "start": 5281
+      },
+      {
+        "number": "III",
+        "title": "THEPROPOSED METHODOLOGY",
+        "start": 8086
+      },
+      {
+        "number": "A",
+        "title": "The Basis of Database Parameter Dynamic Adjustment",
+        "start": 8115
+      },
+      {
+        "number": "IV",
+        "title": "EXPERIMENT AND DISCUSSION",
+        "start": 17955
+      },
+      {
+        "number": "A",
+        "title": "The Database Disaster Recovery Backup/restore",
+        "start": 18189
+      },
+      {
+        "number": "B",
+        "title": "The Database Query Performance",
+        "start": 19262
+      },
+      {
+        "number": "V",
+        "title": "CONCLUSION",
+        "start": 20110
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Siddiqui, Tarique, Alekh Jindal, Shi Qiao, Hiren Patel, and Wangchao Le. \"Cost models for big data query processing: Learning, retrofitting, and our findings.\" In Proceedings of the 2020 ACMSIGMOD International Conference on Management of Data, pp. 99 -113. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Sadineni, Praveen Kumar. \"Comparative Study on Query Processing and Indexing Techniques in Big Data.\" In 2020 3rd International Conference on Intelligent Sustainable Systems (ICISS), pp. 933 -939. IEEE, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Anneser, Christoph, Nesime Tatbul, David Cohen, Zhenggang Xu, Prithviraj Pandian, Nikolay Laptev, and Ryan Marcus. \"Autosteer: Learned query optimization for any sql database.\" Proceedings of the VLDB Endowment 16, no. 12 (2023): 3515 -3527.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Karyakin, Alexey, and Kenneth Salem. \"DimmStore: memory power optimization for database systems.\" Proceedings of the VLDB Endowment 12, no. 11 (2019): 1499 -1512.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Mohsin, Sayed A., Saad Mohamed Darwish, and Ahmed Younes. \"Qiaco: a quantum dynamic cost ant system for query optimization in distributed database.\" IEEE Access 9 (2021): 15833 -15846.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Rigger, Manuel, and Zhendong Su. \"Detecting optimization bugs in database engines via non -optimizing reference engine construction.\" In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pp. 1140 -1152. 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 7,
+      "text": "Yang, Junwen, Cong Yan, Chengcheng Wan, Shan Lu, and Alvin Cheung. \"View -centric performance optimization for database -backed web applications.\" In Software Engineering (ICSE), pp. 994 -1004. IEEE, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Zhang, Ji, Yu Liu, Ke Zhou, Guoliang Li, Zhili Xiao, Bin Cheng, Jiashu Xing et al. \"An end -to-end automatic cloud database tuning system using deep reinforcement learning.\" In Proceedings of the 2019 international conference on management of data, pp. 415 -432. 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Suganya, E., and S. Vijayarani. \"Firefly optimization algorithm based web scraping for web citation extraction.\" Wireless Personal Communications 118, no. 2 (2021): 1481 -1505.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Zhang, Limeng, and M. Ali Babar. \"Automatic Configuration Tuning on Cloud Database: A Survey.\" arXiv preprint arXiv:2404.06043 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Zhang, Bohan, Dana Van Aken, Justin Wang, Tao Dai, Shuli Jiang, Jacky Lao, Siyuan Sheng, Andrew Pavlo, and Geoffrey J. Gordon. \"A demonstration of the ottertune automatic database management system tuning service.\" Proceedings of the VLDB Endowment 11, no. 12 (2018): 1910 -1913.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Grohmann, Johannes, Daniel Seybold, Simon Eismann, Mark Leznik, Samuel Kounev, and Jörg Domaschka. \"Baloo: Measuring and modeling the performance configurations of distributed dbms.\" In 2020 28th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems (MASCOTS), pp. 1 -8. IEEE, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Ishihara, Yoshiteru, and Masahito Shiba. \"Dynamic configuration tuning of working database management systems.\" In 2020 IEEE 2nd Global Conference on Life Sciences and Technologies (LifeTech), pp. 393 -397. IEEE, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Liang, Jie, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. \"Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation.\" In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, pp. 1 -12. 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Kanellis, Konstantinos, Cong Ding, Brian Kroth, Andreas Müller, Carlo Curino, and Shivaram Venkataraman. \"LlamaTune: Sample -efficient DBMS configuration tuning.\" arXiv preprint arXiv:2203.05128 (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Ma, Chen, and Yuhong Chi. \"Evaluation test and improvement of load balancing algorithms of nginx.\" Ieee Access 10 (2022): 14311 -14324.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Daghistani, Anas, Walid G. Aref, Arif Ghafoor, and Ahmed R. Mahmood. \"Swarm: Adaptive load balancing in distributed streaming systems for big spatial data.\" ACM Transactions on Spatial Algorithms and Systems 7, no. 3 (2021): 1 -43. Proceedings of the International Conference on IoT Based Control Networks and Intelligent Systems (ICICNIS-2024) IEEE Xplore Part Number: CFP242B2-ART: ISBN: 979-8-3315",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Wu, Jiagao, Lu Shen, and Linfeng Liu. \"LSH -based distributed similarity indexing with load balancing in high -dimensional space.\" The Journal of Supercomputing 76, no. 1 (2020): 636 -665.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Goyal, Vineet, and Julien Grand-Clement. \"Robust markov decision processes: Beyond rectangularity.\" Mathematics of Operations Research 48, no. 1 (2023): 203 -226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Suilen, Marnix, Thiago D. Simão, David Parker, and Nils Jansen. \"Robust anytime learning of Markov decision processes.\" Advances in Neural Information Processing Systems 35 (2022): 28790 -28802.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Sawalha, Ihab Hanna. \"Views on business continuity and disaster recovery.\" International Journal of Emergency Services 10, no. 3 (2021): 351-365.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Alshammari, Mohammad M., Ali A. Alwan, Azlin Nordin, and Abedallah Zaid Abualkishik. \"Data backup and recovery with a minimum replica plan in a multi -cloud environment.\" In Research Anthology on Privatizing and Securing Data, pp. 794 -814. IGI Global, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Abualkishik, Abedallah Zaid, Ali A. Alwan, and Yonis Gulzar. \"Disaster recovery in cloud computing systems: An overview.\" International Journal of Advanced Computer Science and Applications 11, no. 9 (2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Ramesh, G., J. Logeshwaran, and V. Aravindarajan. \"A secured database monitoring method to improve data backup and recovery operations in cloud computing.\" BOHR International Journal of Computer Science 2, no. 1 (2022): 1 -7. Proceedings of the International Conference on IoT Based Control Networks and Intelligent Systems (ICICNIS-2024) IEEE Xplore Part Number: CFP242B2-ART: ISBN: 979-8-3315-1809-",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 6,
+      "text": "Rigger, Manuel, and Zhendong Su. \"Detecting optimization bugs in database engines via non -optimizing reference engine construction.\" In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pp. 1140 -1152. 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Rigger and Su [6] designed and optimized NoREC, a non-optimized engine construction method that can be used to detect optimization errors in database engines.",
+      "context_before": "additon, some different scenarios are tested for the performance evaluations. Karyakin and Salem [4] proposed a new framework called the DimmStore model, which is a memory optimization method that can reduce memory power consumption effectively. The DimmStore model can obtain high -gain results under different workloads. Mohsin et al. proposed Qiaco [5], a new quantum dynamic cost ant colony algorithm that can be widely used for the query optimization of distributed databases. The advantages of the Qiaco over traditional ant colony algorithms (ACA) are faster convergence speed and robustness.",
+      "context_after": "The proposed NoREC constructs a non -optimized engine to evaluate queries and detects the differences between optimized and non optimized engines. Multiple experiments clearly show that the NoREC model can effectively detect the optimization errors. Yang et al. [7] proposed the Panorama model, a view -centric performance optimization method. It can be used for database backed Web applications and conduct comprehensive analysis. Panorama provides the stable database -aware development environment that can help developers optimize performance of Web pages. Zhang et al. [8] proposed a novel CDBT",
+      "section": "II REVIEW OFDATABASE OPTIMIZATION STUDIES",
+      "page": 2,
+      "char_offset": 6629,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "The proposed NoREC constructs a non -optimized engine to evaluate queries and detects the differences between optimized and non optimized engines.",
+      "context_before": "h is a memory optimization method that can reduce memory power consumption effectively. The DimmStore model can obtain high -gain results under different workloads. Mohsin et al. proposed Qiaco [5], a new quantum dynamic cost ant colony algorithm that can be widely used for the query optimization of distributed databases. The advantages of the Qiaco over traditional ant colony algorithms (ACA) are faster convergence speed and robustness. Rigger and Su [6] designed and optimized NoREC, a non-optimized engine construction method that can be used to detect optimization errors in database engines.",
+      "context_after": "Multiple experiments clearly show that the NoREC model can effectively detect the optimization errors. Yang et al. [7] proposed the Panorama model, a view -centric performance optimization method. It can be used for database backed Web applications and conduct comprehensive analysis. Panorama provides the stable database -aware development environment that can help developers optimize performance of Web pages. Zhang et al. [8] proposed a novel CDBTune, which is recognized as the new and efficient deep reinforcement learning method which can be used for automatic tuning of the cloud databases.",
+      "section": "II REVIEW OFDATABASE OPTIMIZATION STUDIES",
+      "page": 2,
+      "char_offset": 6788,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Multiple experiments clearly show that the NoREC model can effectively detect the optimization errors.",
+      "context_before": "ferent workloads. Mohsin et al. proposed Qiaco [5], a new quantum dynamic cost ant colony algorithm that can be widely used for the query optimization of distributed databases. The advantages of the Qiaco over traditional ant colony algorithms (ACA) are faster convergence speed and robustness. Rigger and Su [6] designed and optimized NoREC, a non-optimized engine construction method that can be used to detect optimization errors in database engines. The proposed NoREC constructs a non -optimized engine to evaluate queries and detects the differences between optimized and non optimized engines.",
+      "context_after": "Yang et al. [7] proposed the Panorama model, a view -centric performance optimization method. It can be used for database backed Web applications and conduct comprehensive analysis. Panorama provides the stable database -aware development environment that can help developers optimize performance of Web pages. Zhang et al. [8] proposed a novel CDBTune, which is recognized as the new and efficient deep reinforcement learning method which can be used for automatic tuning of the cloud databases. The CDBTune uses the novel deep deterministic policy gradient method to find the best configuration an",
+      "section": "II REVIEW OFDATABASE OPTIMIZATION STUDIES",
+      "page": 2,
+      "char_offset": 6935,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:04Z",
+    "is_model_written": true,
+    "summary": "This paper applies a genetic algorithm to database index selection, adjusting the index set dynamically as the workload changes. Its literature review surveys prior work on database optimization, of which automated testing is one strand.",
+    "narrative": "NoREC is described in the review of database optimization work: the paper summarises how it constructs a non-optimizing reference engine and detects differences between optimized and non-optimized evaluation, and notes that experiments show it effectively detects optimization errors. That is the whole of the connection -- the paper's own subject is index selection, and no oracle is used, extended or compared against.",
+    "roles": {
+      "M1": "definition",
+      "M2": "definition",
+      "M3": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icpc66645_2025_00021.json b/_data/papers/paper_doi_10_1109_icpc66645_2025_00021.json
new file mode 100644
index 0000000..6e85b73
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icpc66645_2025_00021.json
@@ -0,0 +1,2345 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:24:29Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icpc66645.2025.00021",
+    "title": "Sembug: Detecting Logic Bugs in Dbms Through Generating Semantic-Aware Non-Optimizing Query",
+    "authors": [
+      "Shiyang Ye",
+      "Chao Ni",
+      "Jue Wang",
+      "Qianqian Pang",
+      "Xinrui Li",
+      "Xiaodan Xu"
+    ],
+    "year": 2025,
+    "venue": "IEEE International Conference on Program Comprehension",
+    "doi": "10.1109/icpc66645.2025.00021",
+    "arxiv_id": null,
+    "s2_paper_id": "aeea391e73991ef6f033917dba0f577c948eedb4",
+    "url": "https://doi.org/10.1109/icpc66645.2025.00021",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icpc66645.2025.00021",
+      "retrieved_at": "2026-09-10T15:24:29Z",
+      "chars": 65474,
+      "content_sha256": "sha256:fc6870b1b90de4301c9730653b28e70d49bde698beef616571b25f1a7f7352d3"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "Introduction",
+        "start": 2551
+      },
+      {
+        "number": "II",
+        "title": "Background and Motivation",
+        "start": 9960
+      },
+      {
+        "number": "A",
+        "title": "Logic Bugs in DBMS",
+        "start": 9990
+      },
+      {
+        "number": "B",
+        "title": "Optimization Bugs Detection in DBMS",
+        "start": 10903
+      },
+      {
+        "number": "C",
+        "title": "Motivating Example",
+        "start": 14334
+      },
+      {
+        "number": "III",
+        "title": "Approach",
+        "start": 16202
+      },
+      {
+        "number": "A",
+        "title": "Approach Overview",
+        "start": 16911
+      },
+      {
+        "number": "B",
+        "title": "Database and Query Generation",
+        "start": 19070
+      },
+      {
+        "number": "C",
+        "title": "Transform the Query and Retrieve Semantic Information",
+        "start": 20613
+      },
+      {
+        "number": "D",
+        "title": "Compare Cardinality and Content",
+        "start": 23223
+      },
+      {
+        "number": "E",
+        "title": "Supporting Features",
+        "start": 24647
+      },
+      {
+        "number": "IV",
+        "title": "Evaluation",
+        "start": 27197
+      },
+      {
+        "number": "A",
+        "title": "Experimental Setup",
+        "start": 27734
+      },
+      {
+        "number": "B",
+        "title": "RQ1: Effectiveness of SemBug",
+        "start": 29940
+      },
+      {
+        "number": "C",
+        "title": "RQ2: Comparison on Other Techniques",
+        "start": 34596
+      },
+      {
+        "number": "D",
+        "title": "RQ3: The Advantages of Expanded Feature Support",
+        "start": 37453
+      },
+      {
+        "number": "E",
+        "title": "RQ4: Discrepancies in Logic Bug Detection Across Methods",
+        "start": 39914
+      },
+      {
+        "number": "V",
+        "title": "Discussion",
+        "start": 47384
+      },
+      {
+        "number": "A",
+        "title": "Generality",
+        "start": 47398
+      },
+      {
+        "number": "B",
+        "title": "Limitations",
+        "start": 47981
+      },
+      {
+        "number": "C",
+        "title": "Threats to Validity",
+        "start": 48360
+      },
+      {
+        "number": "VI",
+        "title": "Related Work",
+        "start": 49116
+      },
+      {
+        "number": "A",
+        "title": "Metamorphic Testing of DBMS",
+        "start": 49133
+      },
+      {
+        "number": "B",
+        "title": "Differential Testing of DBMS",
+        "start": 51129
+      },
+      {
+        "number": "C",
+        "title": "DBMS Fuzzing",
+        "start": 52347
+      },
+      {
+        "number": "D",
+        "title": "Query Generation of DBMS",
+        "start": 53189
+      },
+      {
+        "number": "VII",
+        "title": "Conclusion",
+        "start": 53980
+      },
+      {
+        "number": "J",
+        "title": "Wang, and J. Li, “Sequence-oriented dbms fuzzing,” in 2023 IEEE 39th",
+        "start": 57146
+      },
+      {
+        "number": "J",
+        "title": "Widom, Eds. Morgan Kaufmann, 1998, pp. 618–622.",
+        "start": 58007
+      },
+      {
+        "number": "O",
+        "title": "Papaemmanouil, and N. Tatbul, “Neo: A learned query optimizer,”",
+        "start": 59064
+      },
+      {
+        "number": "T",
+        "title": "Huang, “Testing database systems via differential query execution,” in",
+        "start": 60394
+      },
+      {
+        "number": "L",
+        "title": "Chen, H. Wang, H. Zhong, and T. Huang, “Detecting isolation bugs via",
+        "start": 60771
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Mysql customers,” https://www.mysql.com/customers/, 2024, accessed: 2024-03-30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“Postgresql clients,” https://wiki.postgresql.org/wiki/PostgreSQLClients, 2024, accessed: 2024-03-30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“Tidb partners,” https://www.pingcap.com/partners/, 2024, accessed: 2024-03-30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Website.2013, “American fuzzy lop (afl) fuzzer,” https://lcamtuf. coredump.cx/afl/, 2024, accessed: 2024-03-30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of {DBMS} with coverage-based guidance,” in 31st USENIX Security Symposium (USENIX Security 22), 2022, pp. 4309–4326. 133",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "X. Tang, S. Wu, D. Zhang, F. Li, and G. Chen, “Detecting logic bugs of join optimizations in dbms,” 1, no. 1, may 2023. [Online]. Available: https://doi.org/10.1145/3588909",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammar-free dbms fuzzing,” in Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, ser. ASE ’22. New York, NY, USA: Association for Computing Machinery, 2023. [Online]. Available: https://doi.org/10.1145/3551349.3560431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing database management systems with language validity and coverage feedback,” in Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 11,
+      "text": "A. Seltenreich, “Sqlsmith,” https://github.com/anse1/sqlsmith, 2024, accessed: 2024-03-30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang, and Y. Jiang, “Industry practice of coverage-guided enterprise-level dbms fuzzing,” in ing: Software Engineering in Practice (ICSE-SEIP). IEEE, 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "J. Liang, Y. Chen, Z. Wu, J. Fu, M. Wang, Y. Jiang, X. Huang, T. Chen, J. Wang, and J. Li, “Sequence-oriented dbms fuzzing,” in 2023 IEEE 39th International Conference on Data Engineering (ICDE). IEEE, 2023, pp. 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Z.-M. Jiang, J.-J. Bai, and Z. Su, “ {DynSQL}: Stateful fuzzing for database management systems with complex and valid {SQL}query generation,” in 32nd USENIX Security Symposium (USENIX Security 23), 2023, pp. 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "J. Ba and M. Rigger, “Keep it simple: Testing databases via differential query plans,” 2, no. 3, may 2024. [Online]. Available: https://doi.org/10.1145/3654991",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 16,
+      "text": "W. E. Howden, “Theoretical and empirical studies of program testing,” ser. ICSE ’78, 1978, p. 305–311.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "D. R. Slutz, “Massive stochastic testing of sql,” in VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, August 2427, 1998, New York City, New York, USA, A. Gupta, O. Shmueli, and J. Widom, Eds. Morgan Kaufmann, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 19,
+      "text": "Z. Hao, Q. Huang, C. Wang, J. Wang, Y. Zhang, R. Wu, and C. Zhang, “Pinolo: Detecting logical bugs in database management systems with approximate query synthesis,” in 2023 USENIX Annual Technical Conference (USENIX ATC 23), 2023, pp. 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "T. Y. Chen, S. C. Cheung, and S. Yiu, “Metamorphic testing: A new approach for generating next test cases,” CoRR, vol. abs/2002.12543, 2020. [Online]. Available: https://arxiv.org/abs/2002.12543",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "“Replication,” https://github.com/Syang111/SemBug, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "M. Jarke and J. Koch, “Query optimization in database systems,” ACM Computing surveys (CsUR), vol. 16, no. 2, pp. 111–152, 1984.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "R. Marcus, P. Negi, H. Mao, C. Zhang, M. Alizadeh, T. Kraska, O. Papaemmanouil, and N. Tatbul, “Neo: A learned query optimizer,” arXiv preprint arXiv:1904.03711, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "R. Marcus, P. Negi, H. Mao, N. Tatbul, M. Alizadeh, and T. Kraska, “Bao: Making learned query optimization practical,” in Proceedings of the 2021 International Conference on Management of Data, 2021, pp. 1275–1288.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Y. E. Ioannidis, “Query optimization,” ACM Computing Surveys (CSUR), vol. 28, no. 1, pp. 121–123, 1996.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "T. K. Sellis, “Multiple-query optimization,” ACM Transactions on Database Systems (TODS), vol. 13, no. 1, pp. 23–52, 1988.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "S. Chu, K. Weitz, A. Cheung, and D. Suciu, “Hottsql: proving query rewrites with univalent sql semantics,” in Proceedings of the 38th ACMSIGPLAN Conference on Programming Language Design and Implementation, ser. PLDI 2017. New York, NY, USA: Association for Computing Machinery, 2017, p. 510–524. [Online]. Available: https://doi.org/10.1145/3062341.3062348",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "J. Liang, Z. Wu, J. Fu, M. Wang, C. Sun, and Y. Jiang, “Mozi: Discovering dbms bugs via configuration-based equivalent transformation,” 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "M. Rigger, “Sqlancer,” https://github.com/sqlancer/sqlancer, 2024, accessed: 2024-03-30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 30,
+      "text": "PingCap, “go randgen,” https://github.com/pingcap/go-randgen, 2024, accessed: 2024-03-30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "J. Song, W. Dou, Z. Cui, Q. Dai, W. Wang, J. Wei, H. Zhong, and T. Huang, “Testing database systems via differential query execution,” in (ICSE). IEEE, 2023, pp. 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "A. Zeller, Why programs fail: a guide to systematic debugging. Elsevier, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” in (ICSE), 2023, pp. 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 34,
+      "text": "W. Dou, Z. Cui, Q. Dai, J. Song, D. Wang, Y. Gao, W. Wang, J. Wei, L. Chen, H. Wang, H. Zhong, and T. Huang, “Detecting isolation bugs via transaction oracle construction,” in 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE), 2023, pp. 1123–1135.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Z.-M. Jiang, S. Liu, M. Rigger, and Z. Su, “Detecting transactional bugs in database engines via Graph-Based oracle construction,” in 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). Boston, MA: USENIX Association, Jul. 2023, pp. 397–417. [Online]. Available: https://www.usenix.org/conference/osdi23/presentation/jiang",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 36,
+      "text": "D. Xiao, Z. Liu, Y. Yuan, Q. Pang, and S. Wang, “Metamorphic testing of deep learning compilers,” in Abstract Proceedings of the 2022 ACMSIGMETRICS/IFIPPERFORMANCE Joint International Conference on Measurement and Modeling of Computer Systems, ser. SIGMETRICS/PERFORMANCE ’22. New York, NY, USA: Association for Computing Machinery, 2022, p. 65–66. [Online]. Available: https://doi.org/10.1145/348904",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "T. Y. Chen, F.-C. Kuo, H. Liu, P.-L. Poon, D. Towey, T. H. Tse, and Z. Q. Zhou, “Metamorphic testing: A review of challenges and opportunities,” ACM Comput. Surv., vol. 51, no. 1, jan 2018. [Online]. Available: https://doi.org/10.1145/3143561",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "D. Winterer, C. Zhang, and Z. Su, “Validating smt solvers via semantic fusion,” in Proceedings of the 41st ACMSIGPLAN Conference on Programming Language Design and Implementation, ser. PLDI 2020. New York, NY, USA: Association for Computing Machinery, 2020, p. 718–730. [Online]. Available: https://doi.org/10.1145/3385412.3385985",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "X. Liu, Q. Zhou, J. Arulraj, and A. Orso, “Automatic detection of performance bugs in database systems using equivalent queries,” in Proceedings of the 44th International Conference on Software Engineering, 2022, pp. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "J. Ba and M. Rigger, “Cert: Finding performance issues in database systems through the lens of cardinality estimation,” 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "W. M. McKeeman, “Differential testing for software,” Digit. Tech. J., vol. 10, no. 1, pp. 100–107, 1998. [Online]. Available: https://www.hpl.hp.com/hpjournal/dtj/vol10num1/vol10num1art9.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Y. Chen, T. Su, and Z. Su, “Deep differential testing of JVM implementations,” in Proceedings of the 41st International Conference on Software Engineering, ICSE 2019, Montreal, QC, Canada, May 25-31, 2019, J. M. Atlee, T. Bultan, and J. Whittle, Eds. IEEE / ACM, 2019, pp. 1257–1268. [Online]. Available: https://doi.org/10.1109/ICSE.2019.00127",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Y. Yang, Y. Zhou, H. Sun, Z. Su, Z. Zuo, L. Xu, and B. Xu, “Hunting for bugs in code coverage tools via randomized differential testing,” in Proceedings of the 41st International Conference on Software Engineering, ICSE 2019, Montreal, QC, Canada, May 25-31, 2019, J. M. Atlee, T. Bultan, and J. Whittle, Eds. IEEE / ACM, 2019, pp. 488–498. [Online]. Available: https://doi.org/10.1109/ICSE.2019.0006",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "P. Nie, C. Wan, J. Zhu, Z. Lin, Y. Chen, and Z. Su, “Coveragedirected differential testing of X.509 certificate validation in SSL/TLS implementations,” ACM Trans. Softw. Eng. Methodol., vol. 32, no. 1, pp. 3:1–3:32, 2023. [Online]. Available: https://doi.org/10.1145/3510416",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "P. Godefroid, D. Lehmann, and M. Polishchuk, “Differential regression testing for REST apis,” in ISSTA ’20: 29th ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, USA, July 18-22, 2020, S. Khurshid and C. S. Pasareanu, Eds. ACM, 2020, pp. 312–323. [Online]. Available: https://doi.org/10.1145/3395363.3397374",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “Apollo: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment, vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Z. Cui, W. Dou, Q. Dai, J. Song, W. Wang, J. Wei, and D. Ye, “Differentially testing database transactions for fun and profit,” in 37th IEEE/ACM International Conference on Automated Software Engineering, ASE 2022, Rochester, MI, USA, October 134 10-14, 2022. ACM, 2022, pp. 35:1–35:12. [Online]. Available: https://doi.org/10.1145/3551349.3556924",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Z. Hua, W. Lin, L. Ren, Z. Li, L. Zhang, W. Jiao, and T. Xie, “Gdsmith: Detecting bugs in cypher graph database engines,” ser. ISSTA 2023. New York, NY, USA: Association for Computing Machinery, 2023, p. 163–174. [Online]. Available: https://doi.org/10.1145/3597926.3598046",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "X. Yang, Y. Chen, E. Eide, and J. Regehr, “Finding and understanding bugs in c compilers,” in Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation, ser. PLDI ’11. New York, NY, USA: Association for Computing Machinery, 2011, p. 283–294. [Online]. Available: https://doi.org/10.1145/1993498.1993532 135",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 5,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 10,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 15,
+      "text": "J. Ba and M. Rigger, “Keep it simple: Testing databases via differential query plans,” 2, no. 3, may 2024. [Online]. Available: https://doi.org/10.1145/3654991",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "prints the DOI of the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 18,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 29,
+      "text": "M. Rigger, “Sqlancer,” https://github.com/sqlancer/sqlancer, 2024, accessed: 2024-03-30.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 33,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” in (ICSE), 2023, pp. 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 35,
+      "text": "Z.-M. Jiang, S. Liu, M. Rigger, and Z. Su, “Detecting transactional bugs in database engines via Graph-Based oracle construction,” in 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). Boston, MA: USENIX Association, Jul. 2023, pp. 397–417. [Online]. Available: https://www.usenix.org/conference/osdi23/presentation/jiang",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "J. Ba and M. Rigger, “Cert: Finding performance issues in database systems through the lens of cardinality estimation,” 2024.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", Pinolo, TLP, and NoREC).",
+      "context_before": "gs by transforming the queries that can be highly optimized by DBMS into equivalent but less optimized ones. Additionally, SemBug integrates semantic analysis technology, enabling it to identify semantic logic bugs and support testing advanced DBMS features. Any discrepancy in cardinality or content between the original and transformed queries indicates a logic bug. To investigate the effectiveness of SemBug, we conduct a large-scale experiment on five widelyused DBMS systems (i.e., MySQL, TiDB, MariaDB, SQLite, and PostgreSQL) and compare it with three state-of-the-art (SOTA) approaches (i.e.",
+      "context_after": "The experimental results indicate that SemBug outperforms three SOTAs. Over 24 hours, SemBug found 34 unique logic bugs, which are 19, 14, and 13 more bugs than each of the three SOTAs, marking an improvement of 126%, 70%, and 61% respectively. As of the time of paper submission, SemBug has uncovered 37 unique logic bugs, of which 29 have been verified by developers, and 11 have been fixed. SemBug helps developers identify these bugs, providing insights into such inconsistencies and assisting in resolving them. I. Introduction Database Management Systems (DBMSs) are essential components of th",
+      "section": null,
+      "page": 1,
+      "char_offset": 2007,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[4]–[10]",
+      "technique": "pqs",
+      "sentence": "Consequently, DBMS testing has become an area of considerable interest for researchers [4]–[10].",
+      "context_before": "s, marking an improvement of 126%, 70%, and 61% respectively. As of the time of paper submission, SemBug has uncovered 37 unique logic bugs, of which 29 have been verified by developers, and 11 have been fixed. SemBug helps developers identify these bugs, providing insights into such inconsistencies and assisting in resolving them. I. Introduction Database Management Systems (DBMSs) are essential components of the modern information technology infrastructure, handling vast quantities of data for a multitude of applications [1]–[3]. Bugs within a DBMS can significantly impact a broad user base.",
+      "context_after": "Recently, fuzzing technology has been extensively employed to detect crash bugs in DBMSs [9], [11]–[14], which are relatively straightforward to identify due to their explicit system crash characteristics. In contrast, logic bugs are a more elusive category of errors, often overlooked by developers since theycan return incorrect query results without the overt signs of system crashes or other noticeable symptoms. The majority of logic bugs originate from issues within the query optimizer [15], which is acknowledged as one of the most complex components of a DBMS. The optimizer’s role is to cr",
+      "section": "I Introduction",
+      "page": 1,
+      "char_offset": 2817,
+      "cited_reference": {
+        "number": 5,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[15]",
+      "technique": "dqp",
+      "sentence": "The majority of logic bugs originate from issues within the query optimizer [15], which is acknowledged as one of the most complex components of a DBMS.",
+      "context_before": "f applications [1]–[3]. Bugs within a DBMS can significantly impact a broad user base. Consequently, DBMS testing has become an area of considerable interest for researchers [4]–[10]. Recently, fuzzing technology has been extensively employed to detect crash bugs in DBMSs [9], [11]–[14], which are relatively straightforward to identify due to their explicit system crash characteristics. In contrast, logic bugs are a more elusive category of errors, often overlooked by developers since theycan return incorrect query results without the overt signs of system crashes or other noticeable symptoms.",
+      "context_after": "The optimizer’s role is to create efficient execution plans for database queries, and any defects in this process can result in logic bugs. We specifically refer to logic bugs arising from the query optimizer as optimization bugs. Designing an automated method to detect logic bugs is no easy task, as it requires knowledge of the correct output for each input—a classic problem in testing known as the test oracle problem [16]. To tackle this challenge, researchers have proposed several DBMS oracles [5], [10], [17]–[19] for the automatic detection of logic bugs, including differential testing-ba",
+      "section": "I Introduction",
+      "page": 1,
+      "char_offset": 3331,
+      "cited_reference": {
+        "number": 15,
+        "text": "J. Ba and M. Rigger, “Keep it simple: Testing databases via differential query plans,” 2, no. 3, may 2024. [Online]. Available: https://doi.org/10.1145/3654991",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "prints the DOI of the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[17]–[19]",
+      "technique": "norec",
+      "sentence": "To tackle this challenge, researchers have proposed several DBMS oracles [5], [10], [17]–[19] for the automatic detection of logic bugs, including differential testing-based methods, oracle-guided methods, and metamorphic testing-based methods.",
+      "context_before": "ticeable symptoms. The majority of logic bugs originate from issues within the query optimizer [15], which is acknowledged as one of the most complex components of a DBMS. The optimizer’s role is to create efficient execution plans for database queries, and any defects in this process can result in logic bugs. We specifically refer to logic bugs arising from the query optimizer as optimization bugs. Designing an automated method to detect logic bugs is no easy task, as it requires knowledge of the correct output for each input—a classic problem in testing known as the test oracle problem [16].",
+      "context_after": "The first category consists of differential testing-based methods. The core idea of differential testing is different implementations of the same algorithms should yield the same results. Slutz proposed RAGS [17], a technique that first executes the same SQL query across different DBMSs and then compares their results. If the DBMSs return different results, it suggests that one of the DBMSs contains a logic bug. Differential testing requires the DBMSs being tested to have identical syntax and semantics. However, in practice, different DBMSs often have unique extensions to the SQL standard, wh",
+      "section": "I Introduction",
+      "page": 1,
+      "char_offset": 3913,
+      "cited_reference": {
+        "number": 18,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "pqs",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "introduced PQS [5], a technique that begins by randomly selecting a specific row from a table.",
+      "context_before": "the DBMSs return different results, it suggests that one of the DBMSs contains a logic bug. Differential testing requires the DBMSs being tested to have identical syntax and semantics. However, in practice, different DBMSs often have unique extensions to the SQL standard, which limits differential testing methods to the core parts of the DBMSs, thereby restricting their general applicability. The second category is oracle-guided methods, which guide the testing process by proposing specific oracles, thereby avoiding the issue of DBMS dialects encountered in differential testing. Rigger et al.",
+      "context_after": "Subsequent queries are crafted to retrieve the selected row, with the WHERE clause always evaluating to TRUE. If the final result omits the selected row, it indicates the presence of a logic bug. Although PQS has been successful in detecting logic bugs in widely-used DBMSs, its most significant drawback is the considerable implementation effort required, involving the re-implementation of each function and operator for each tested DBMS. The labor-intensive nature of this process greatly restricts its generalizability. The third category of methods, based on metamorphic124 2643-7171/25/$31.00",
+      "section": "I Introduction",
+      "page": 1,
+      "char_offset": 5082,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Although PQS has been successful in detecting logic bugs in widely-used DBMSs, its most significant drawback is the considerable implementation effort required, involving the re-implementation of each function and operator for each tested DBMS.",
+      "context_before": "erential testing methods to the core parts of the DBMSs, thereby restricting their general applicability. The second category is oracle-guided methods, which guide the testing process by proposing specific oracles, thereby avoiding the issue of DBMS dialects encountered in differential testing. Rigger et al. introduced PQS [5], a technique that begins by randomly selecting a specific row from a table. Subsequent queries are crafted to retrieve the selected row, with the WHERE clause always evaluating to TRUE. If the final result omits the selected row, it indicates the presence of a logic bug.",
+      "context_after": "The labor-intensive nature of this process greatly restricts its generalizability. The third category of methods, based on metamorphic124 2643-7171/25/$31.00 ©2025 IEEEDOI 10.1109/ICPC66645.2025.00021 .00 ©2025 IEEE | DOI: 10.1109/ICPC66645.2025.00021 testing [20], aims to overcome the extensive implementation efforts required by oracle-guided methods, positioning it as one of the most advanced approaches currently available. The core idea of these methods is to transform a given query based on a certain relationship called a metamorphic relationship. If the results from the original query",
+      "section": "I Introduction",
+      "page": 1,
+      "char_offset": 5373,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": ", such as NoREC [18]), while the second one addresses both cardinality-incorrect logic bugs and semantic logic bugs, where the cardinality remains consistent, but the content of the results is not as expected.",
+      "context_before": "ome the extensive implementation efforts required by oracle-guided methods, positioning it as one of the most advanced approaches currently available. The core idea of these methods is to transform a given query based on a certain relationship called a metamorphic relationship. If the results from the original query and the transformed query do not align with this relationship, it implies the presence of a logic bug. Existing metamorphic testing-based methods can be categorized into two groups. The first one focuses solely on logic bugs where the cardinality of query results is incorrect (e.g.",
+      "context_after": "(e.g., Pinolo [19] and TLP [10]). Though metamorphic testing-based methods have made great progress in detecting logic bugs in DBMS and have become state-of-the-art, they still have some limitations. More precisely, we find that over 70% of the logic bugs found by those approaches (i.e., TLP and Pinolo) in the second group are optimization-related bugs, which indicates that we should pay more attention to these types of logic bugs. However, the second group approaches are not specifically optimized to handle these issues. Meanwhile, as for the first group method(i.e., NoREC), though it is des",
+      "section": "I Introduction",
+      "page": 2,
+      "char_offset": 6500,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", Pinolo [19] and TLP [10]).",
+      "context_before": "ed on a certain relationship called a metamorphic relationship. If the results from the original query and the transformed query do not align with this relationship, it implies the presence of a logic bug. Existing metamorphic testing-based methods can be categorized into two groups. The first one focuses solely on logic bugs where the cardinality of query results is incorrect (e.g., such as NoREC [18]), while the second one addresses both cardinality-incorrect logic bugs and semantic logic bugs, where the cardinality remains consistent, but the content of the results is not as expected. (e.g.",
+      "context_after": "Though metamorphic testing-based methods have made great progress in detecting logic bugs in DBMS and have become state-of-the-art, they still have some limitations. More precisely, we find that over 70% of the logic bugs found by those approaches (i.e., TLP and Pinolo) in the second group are optimization-related bugs, which indicates that we should pay more attention to these types of logic bugs. However, the second group approaches are not specifically optimized to handle these issues. Meanwhile, as for the first group method(i.e., NoREC), though it is designed to specifically address the",
+      "section": "I Introduction",
+      "page": 2,
+      "char_offset": 6715,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", TLP and Pinolo) in the second group are optimization-related bugs, which indicates that we should pay more attention to these types of logic bugs.",
+      "context_before": "s. The first one focuses solely on logic bugs where the cardinality of query results is incorrect (e.g., such as NoREC [18]), while the second one addresses both cardinality-incorrect logic bugs and semantic logic bugs, where the cardinality remains consistent, but the content of the results is not as expected. (e.g., Pinolo [19] and TLP [10]). Though metamorphic testing-based methods have made great progress in detecting logic bugs in DBMS and have become state-of-the-art, they still have some limitations. More precisely, we find that over 70% of the logic bugs found by those approaches (i.e.",
+      "context_after": "However, the second group approaches are not specifically optimized to handle these issues. Meanwhile, as for the first group method(i.e., NoREC), though it is designed to specifically address the logic bugs within optimization scenarios, it takes no consideration of semantical logic bugs and has no support for testing advanced DBMS features (e.g., aggregate function, window function, and ORDER BYclause). In this paper, we introduce Semantic-aware Non-Optimizing Query (SemBug), a novel approach based on metamorphic testing for logic bug detection in DBMSs. SemBug distinguishes itself by its a",
+      "section": "I Introduction",
+      "page": 2,
+      "char_offset": 6997,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": ", NoREC), though it is designed to specifically address the logic bugs within optimization scenarios, it takes no consideration of semantical logic bugs and has no support for testing advanced DBMS features (e.",
+      "context_before": "esults is not as expected. (e.g., Pinolo [19] and TLP [10]). Though metamorphic testing-based methods have made great progress in detecting logic bugs in DBMS and have become state-of-the-art, they still have some limitations. More precisely, we find that over 70% of the logic bugs found by those approaches (i.e., TLP and Pinolo) in the second group are optimization-related bugs, which indicates that we should pay more attention to these types of logic bugs. However, the second group approaches are not specifically optimized to handle these issues. Meanwhile, as for the first group method(i.e.",
+      "context_after": "g., aggregate function, window function, and ORDER BYclause). In this paper, we introduce Semantic-aware Non-Optimizing Query (SemBug), a novel approach based on metamorphic testing for logic bug detection in DBMSs. SemBug distinguishes itself by its ability to detect optimization bugs through the transformation of highly optimized queries into equivalent but less optimized forms. Furthermore, SemBug integrates semantic analysis technology, enabling it to identify semantic logic bugs and support testing advanced features. Specifically, SemBug consists of three phases. In the first phase, we ge",
+      "section": "I Introduction",
+      "page": 2,
+      "char_offset": 7283,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[5]",
+      "technique": "pqs",
+      "sentence": "Among these issues, logic bugs, which cause a query to return unexpected outcomes [5], stand out as particularly elusive, often hidden deep within the core functionalities of a DBMS [19].",
+      "context_before": "logic bugs in DBMS. SemBug utilizes semantic analysis to find semantic logic bugs and support testing advanced database features. •We developed SemBug as a DBMS testing system. The replication package is publicly available at [21]. •We found 37 unique logic bugs on five widely-use DBMSs in the real world. Of these, 29 have been verified by developers, 11 have been fixed, and the remaining bugs are currently under investigation. II. Background and Motivation A. Logic Bugs in DBMS As sophisticated software systems, DBMSs are prone to bugs that can lead to system crashes and unexpected outcomes.",
+      "context_after": "A developer may craft a SQL query with the conviction that it is error-free, only to find that the results are incorrect upon execution. As shown in Figure 1, the first query returns an unexpected outcome. However, this type of logic bug often escapes the notice of developers because it lacks the overt symptoms like system crashes. Additionally, identifying such bugs requires knowledge of the ground truth for the test cases. Therefore, detecting logic bugs is not only crucial for ensuring the reliability of information systems but also represents significant technical challenges. B. Optimizat",
+      "section": "A Logic Bugs in DBMS",
+      "page": 2,
+      "char_offset": 10127,
+      "cited_reference": {
+        "number": 5,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[18]",
+      "technique": "norec",
+      "sentence": "Logic bugs that originate from the query optimizer are commonly referred to as optimization bugs [18].",
+      "context_before": "DBMS optimization techniques [22]. Many investigations have explored methods for enhancing query optimization performance [23]– [26]. Typically, DBMSs employ a query optimizer to optimize query execution. This component analyzes the given SQL queries and leverages the database’s metadata and statistical information to craft an efficient execution plan that minimizes both disk accesses and computational overhead. As one of the most complex components within a DBMS, the implementation of the query optimizer is highly challenging [27], with the potential for errors that could lead to logic bugs.",
+      "context_after": "Optimization bugs in DBMS can be identified by altering the system’s optimization settings, including the selection of the query optimizer, indexing strategies, and parallelism settings, to facilitate differential testing [28]. However, for DBMS, most optimizations cannot be disabled, making it challenging to comprehensively test for optimization bugs through configuration adjustments. Consequently, several approaches have been proposed to tackle these challenges. For example, Rigger et al. [18] proposed NoREC, which restructures SQL statements to inhibit query 125 SELECT t0.c0 ASc0 FROM v0",
+      "section": "B Optimization Bugs Detection in DBMS",
+      "page": 2,
+      "char_offset": 11585,
+      "cited_reference": {
+        "number": 18,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "[18] proposed NoREC, which restructures SQL statements to inhibit query 125 SELECT t0.",
+      "context_before": ". Logic bugs that originate from the query optimizer are commonly referred to as optimization bugs [18]. Optimization bugs in DBMS can be identified by altering the system’s optimization settings, including the selection of the query optimizer, indexing strategies, and parallelism settings, to facilitate differential testing [28]. However, for DBMS, most optimizations cannot be disabled, making it challenging to comprehensively test for optimization bugs through configuration adjustments. Consequently, several approaches have been proposed to tackle these challenges. For example, Rigger et al.",
+      "context_after": "c0 ASc0 FROM v0, t0 WHERE (SUBTIME ('2001 -11-28 06', '252 10') OR('' IS NOT NULL )) AND v0.c0; SELECT c0 FROM (SELECT t0.c0 ASc0 ,((SUBTIME ('2001−11−28 06','252 10’) OR('' IS NOT NULL )) AND v0.c0) IS TRUE ASflag FROM v0,t0) ASt WHERE flag=1; CREATE TABLE t0(c0 BOOL ); INSERT INTO t0 VALUES (1, NULL ); --t0.c0: 1, NULLCREATE VIEW v0(c0) AS SELECT (REGEXP_LIKE (t0.c0, t0.c0)) FROM t0 WHERE t0.c0 GROUP BYt0.c0 HAVING 1; --v0.c0: 1Initial Database Schema SQL Query 1 21 1 1 NULLFig. 1: A semantic logic bug in TiDB only found by SemBug. SemBug generates the first query that can be optimized and t",
+      "section": "B Optimization Bugs Detection in DBMS",
+      "page": 2,
+      "char_offset": 12184,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "This particular bug cannot be detected by NoREC, as it only identifies bugs resulting from inconsistent cardinality.",
+      "context_before": "10’) OR('' IS NOT NULL )) AND v0.c0) IS TRUE ASflag FROM v0,t0) ASt WHERE flag=1; CREATE TABLE t0(c0 BOOL ); INSERT INTO t0 VALUES (1, NULL ); --t0.c0: 1, NULLCREATE VIEW v0(c0) AS SELECT (REGEXP_LIKE (t0.c0, t0.c0)) FROM t0 WHERE t0.c0 GROUP BYt0.c0 HAVING 1; --v0.c0: 1Initial Database Schema SQL Query 1 21 1 1 NULLFig. 1: A semantic logic bug in TiDB only found by SemBug. SemBug generates the first query that can be optimized and then transforms it into the second query that is less optimized while maintaining the semantic information. The discrepancy in both results values indicates a bug.",
+      "context_after": "TLP and Pinolo cannot detect this optimization bug either. optimization for optimizer testing. Specifically, it relocates the predicate from the WHERE clause to the SELECT clause to induce full table scans, thus exposing logic bugs by detecting discrepancies in result cardinality. Though NoREC performs effectively in detecting bugs, it ignores checking the results content, leading to missing semantic unmatched logic bugs and consequently limiting supporting advanced DBMS features (e.g., aggregate function). Additionally, some approaches (i.e., TLP [10] and Pinolo [19]) can detect optimization",
+      "section": "B Optimization Bugs Detection in DBMS",
+      "page": 3,
+      "char_offset": 13035,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP and Pinolo cannot detect this optimization bug either.",
+      "context_before": "INTO t0 VALUES (1, NULL ); --t0.c0: 1, NULLCREATE VIEW v0(c0) AS SELECT (REGEXP_LIKE (t0.c0, t0.c0)) FROM t0 WHERE t0.c0 GROUP BYt0.c0 HAVING 1; --v0.c0: 1Initial Database Schema SQL Query 1 21 1 1 NULLFig. 1: A semantic logic bug in TiDB only found by SemBug. SemBug generates the first query that can be optimized and then transforms it into the second query that is less optimized while maintaining the semantic information. The discrepancy in both results values indicates a bug. This particular bug cannot be detected by NoREC, as it only identifies bugs resulting from inconsistent cardinality.",
+      "context_after": "optimization for optimizer testing. Specifically, it relocates the predicate from the WHERE clause to the SELECT clause to induce full table scans, thus exposing logic bugs by detecting discrepancies in result cardinality. Though NoREC performs effectively in detecting bugs, it ignores checking the results content, leading to missing semantic unmatched logic bugs and consequently limiting supporting advanced DBMS features (e.g., aggregate function). Additionally, some approaches (i.e., TLP [10] and Pinolo [19]) can detect optimization logic bugs even though they are not specifically designed",
+      "section": "B Optimization Bugs Detection in DBMS",
+      "page": 3,
+      "char_offset": 13152,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Though NoREC performs effectively in detecting bugs, it ignores checking the results content, leading to missing semantic unmatched logic bugs and consequently limiting supporting advanced DBMS features (e.",
+      "context_before": "first query that can be optimized and then transforms it into the second query that is less optimized while maintaining the semantic information. The discrepancy in both results values indicates a bug. This particular bug cannot be detected by NoREC, as it only identifies bugs resulting from inconsistent cardinality. TLP and Pinolo cannot detect this optimization bug either. optimization for optimizer testing. Specifically, it relocates the predicate from the WHERE clause to the SELECT clause to induce full table scans, thus exposing logic bugs by detecting discrepancies in result cardinality.",
+      "context_after": "g., aggregate function). Additionally, some approaches (i.e., TLP [10] and Pinolo [19]) can detect optimization logic bugs even though they are not specifically designed for them. In particular, TLP [10] decomposes a query into three sub-queries, and each sub-query has a predicate evaluated as TRUE, FALSE, or NULL. Then, the union of the sub-query results should equal the results of the original query. Pinolo [19], based on approximate relationships, mutates the original query, with the mutated query’s results expected to be a superset or subset of the original query’s results. Nevertheless, th",
+      "section": "B Optimization Bugs Detection in DBMS",
+      "page": 3,
+      "char_offset": 13434,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", TLP [10] and Pinolo [19]) can detect optimization logic bugs even though they are not specifically designed for them.",
+      "context_before": "ifies bugs resulting from inconsistent cardinality. TLP and Pinolo cannot detect this optimization bug either. optimization for optimizer testing. Specifically, it relocates the predicate from the WHERE clause to the SELECT clause to induce full table scans, thus exposing logic bugs by detecting discrepancies in result cardinality. Though NoREC performs effectively in detecting bugs, it ignores checking the results content, leading to missing semantic unmatched logic bugs and consequently limiting supporting advanced DBMS features (e.g., aggregate function). Additionally, some approaches (i.e.",
+      "context_after": "In particular, TLP [10] decomposes a query into three sub-queries, and each sub-query has a predicate evaluated as TRUE, FALSE, or NULL. Then, the union of the sub-query results should equal the results of the original query. Pinolo [19], based on approximate relationships, mutates the original query, with the mutated query’s results expected to be a superset or subset of the original query’s results. Nevertheless, the practical utility of both TLP and Pinolo in identifying optimization bugs remains limited. C. Motivating Example The core innovation of SemBug is to transform highly optimized q",
+      "section": "B Optimization Bugs Detection in DBMS",
+      "page": 3,
+      "char_offset": 13701,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "In particular, TLP [10] decomposes a query into three sub-queries, and each sub-query has a predicate evaluated as TRUE, FALSE, or NULL.",
+      "context_before": "tion for optimizer testing. Specifically, it relocates the predicate from the WHERE clause to the SELECT clause to induce full table scans, thus exposing logic bugs by detecting discrepancies in result cardinality. Though NoREC performs effectively in detecting bugs, it ignores checking the results content, leading to missing semantic unmatched logic bugs and consequently limiting supporting advanced DBMS features (e.g., aggregate function). Additionally, some approaches (i.e., TLP [10] and Pinolo [19]) can detect optimization logic bugs even though they are not specifically designed for them.",
+      "context_after": "Then, the union of the sub-query results should equal the results of the original query. Pinolo [19], based on approximate relationships, mutates the original query, with the mutated query’s results expected to be a superset or subset of the original query’s results. Nevertheless, the practical utility of both TLP and Pinolo in identifying optimization bugs remains limited. C. Motivating Example The core innovation of SemBug is to transform highly optimized queries into less optimized ones while preserving and analyzing their semantic information. Figure 1 demonstrates a semantic logic bug in",
+      "section": "B Optimization Bugs Detection in DBMS",
+      "page": 3,
+      "char_offset": 13820,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Nevertheless, the practical utility of both TLP and Pinolo in identifying optimization bugs remains limited.",
+      "context_before": "DBMS features (e.g., aggregate function). Additionally, some approaches (i.e., TLP [10] and Pinolo [19]) can detect optimization logic bugs even though they are not specifically designed for them. In particular, TLP [10] decomposes a query into three sub-queries, and each sub-query has a predicate evaluated as TRUE, FALSE, or NULL. Then, the union of the sub-query results should equal the results of the original query. Pinolo [19], based on approximate relationships, mutates the original query, with the mutated query’s results expected to be a superset or subset of the original query’s results.",
+      "context_after": "C. Motivating Example The core innovation of SemBug is to transform highly optimized queries into less optimized ones while preserving and analyzing their semantic information. Figure 1 demonstrates a semantic logic bug in TiDB v7.6.0 that goes undetected by other methods but is successfully uncovered by SemBug. In the Initial Database Schema phase, table t0and view v0 are created, where t0.c0 contains the values 1 and NULL, and v0.c0 holds the value 1. Then, the first query is randomly generated with a WHERE clause that can be optimized, andit is subsequently transformed into the second quer",
+      "section": "B Optimization Bugs Detection in DBMS",
+      "page": 3,
+      "char_offset": 14224,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "This issue shows a significant limitation of NoREC, which concentrates only on the cardinality of results, thereby missing such semantic discrepancies.",
+      "context_before": "ERE clause that can be optimized, andit is subsequently transformed into the second query, which is less optimized according to our method. Meanwhile, we preserve the semantic information (i.e., column c0) during the transformation process. The inconsistency between the results’ content of the two queries indicates a semantic logic bug. Since the WHERE clause in the first query is evaluated as TRUE, it should return 1, NULL. Developers from TiDB informed us that the first query returns the wrong result due to the projection elimination, resulting in the planner outputting v0.c0 instead oft0.c0.",
+      "context_after": "Similarly, TLP and Pinolo do not thoroughly investigate optimization bugs, which can lead to overlooked logic bugs. SemBug stands out by addressing this gap, offering a novel approach that not only identifies optimization bugs but also ensures the semantic accuracy of query results. Listing 1: Transformation and retrieve semantic information step can be combinedQuery1: SELECT c1FROM t1WHERE t1.c1 >0; Query2: SELECT c1, (t1.c1 >0) is true AS flag FROM t1;−− rs0 Query3: SELECT c1FROM rs0WHERE flag is true; Query4: SELECT cFROM (SELECT c1ASc, (t1.c1 >0) is true AS flag FROM t1)AStWHERE flag is t",
+      "section": "C Motivating Example",
+      "page": 3,
+      "char_offset": 15445,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Similarly, TLP and Pinolo do not thoroughly investigate optimization bugs, which can lead to overlooked logic bugs.",
+      "context_before": "e preserve the semantic information (i.e., column c0) during the transformation process. The inconsistency between the results’ content of the two queries indicates a semantic logic bug. Since the WHERE clause in the first query is evaluated as TRUE, it should return 1, NULL. Developers from TiDB informed us that the first query returns the wrong result due to the projection elimination, resulting in the planner outputting v0.c0 instead oft0.c0. This issue shows a significant limitation of NoREC, which concentrates only on the cardinality of results, thereby missing such semantic discrepancies.",
+      "context_after": "SemBug stands out by addressing this gap, offering a novel approach that not only identifies optimization bugs but also ensures the semantic accuracy of query results. Listing 1: Transformation and retrieve semantic information step can be combinedQuery1: SELECT c1FROM t1WHERE t1.c1 >0; Query2: SELECT c1, (t1.c1 >0) is true AS flag FROM t1;−− rs0 Query3: SELECT c1FROM rs0WHERE flag is true; Query4: SELECT cFROM (SELECT c1ASc, (t1.c1 >0) is true AS flag FROM t1)AStWHERE flag is true; III. Approach To effectively detect logic bugs in DBMS, we propose SemBug. Given a seed query that can be optim",
+      "section": "C Motivating Example",
+      "page": 3,
+      "char_offset": 15597,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "citation_marker",
+      "surface": "[29]",
+      "technique": null,
+      "sentence": "Common methods for generating databases and queries include mutation-based [6], [9] and rule-based approaches [11], [29], [30].",
+      "context_before": ".c1); AStWHERE flag is true; logic bugs in the DBMS, as reflected in the fifth step of the diagram. For more technical details, we describe the methods for randomly generating database schema and queries (§ III-B). We then detail the process of transforming optimized queries and extracting semantic information (§ III-C). Subsequently, we explain the methods used for a more efficient comparison of their results (§ III-D). We also discuss the testing features supported by SemBug (§ III-E). B. Database and Query Generation We first need to generate databases and several tables to conduct testing.",
+      "context_after": "Queries generated through rule-based techniques can often pass syntax and semantic checks from DBMS, improving testing efficiency compared to mutationbased methods. We use SQLancer [29], a rule-based method for query generation. This tool initializes the database schema by first executing DDL and DML statements, followed by the continuous generation of SELECT statements to test DBMS. As illustrated in Figure 2 (Step ②), we only generate queries in the form of SELECT 𝛼FROM t WHERE Pthat can be optimized by DBMS. This is because most optimization techniques are applied to data filtering, typica",
+      "section": "B Database and Query Generation",
+      "page": 4,
+      "char_offset": 19177,
+      "cited_reference": {
+        "number": 29,
+        "text": "M. Rigger, “Sqlancer,” https://github.com/sqlancer/sqlancer, 2024, accessed: 2024-03-30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We use SQLancer [29], a rule-based method for query generation.",
+      "context_before": "mantic information (§ III-C). Subsequently, we explain the methods used for a more efficient comparison of their results (§ III-D). We also discuss the testing features supported by SemBug (§ III-E). B. Database and Query Generation We first need to generate databases and several tables to conduct testing. Common methods for generating databases and queries include mutation-based [6], [9] and rule-based approaches [11], [29], [30]. Queries generated through rule-based techniques can often pass syntax and semantic checks from DBMS, improving testing efficiency compared to mutationbased methods.",
+      "context_after": "This tool initializes the database schema by first executing DDL and DML statements, followed by the continuous generation of SELECT statements to test DBMS. As illustrated in Figure 2 (Step ②), we only generate queries in the form of SELECT 𝛼FROM t WHERE Pthat can be optimized by DBMS. This is because most optimization techniques are applied to data filtering, typically expressed in the query’s WHERE clause [18]. For instance, the WHERE clause leverages indexes to speed up searches, thereby avoiding scanning the full table. In the figure, we retrieve records from the column c1that meets the",
+      "section": "B Database and Query Generation",
+      "page": 4,
+      "char_offset": 19470,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "citation_marker",
+      "surface": "[18]",
+      "technique": "norec",
+      "sentence": "This is because most optimization techniques are applied to data filtering, typically expressed in the query’s WHERE clause [18].",
+      "context_before": "queries include mutation-based [6], [9] and rule-based approaches [11], [29], [30]. Queries generated through rule-based techniques can often pass syntax and semantic checks from DBMS, improving testing efficiency compared to mutationbased methods. We use SQLancer [29], a rule-based method for query generation. This tool initializes the database schema by first executing DDL and DML statements, followed by the continuous generation of SELECT statements to test DBMS. As illustrated in Figure 2 (Step ②), we only generate queries in the form of SELECT 𝛼FROM t WHERE Pthat can be optimized by DBMS.",
+      "context_after": "For instance, the WHERE clause leverages indexes to speed up searches, thereby avoiding scanning the full table. In the figure, we retrieve records from the column c1that meets the condition. Since only record v1satisfies the predicate P, the correctly optimized result should be a single record with value v1. However, if the optimizer does not function correctly, it may produce more or fewer records or the same number of records with different content. The focus of this paper is not on the method of database and query generation. It is noted that although we only use SQLancer to generate quer",
+      "section": "B Database and Query Generation",
+      "page": 4,
+      "char_offset": 19822,
+      "cited_reference": {
+        "number": 18,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "It is noted that although we only use SQLancer to generate queries, SemBug can be integrated with any generation technique.",
+      "context_before": "filtering, typically expressed in the query’s WHERE clause [18]. For instance, the WHERE clause leverages indexes to speed up searches, thereby avoiding scanning the full table. In the figure, we retrieve records from the column c1that meets the condition. Since only record v1satisfies the predicate P, the correctly optimized result should be a single record with value v1. However, if the optimizer does not function correctly, it may produce more or fewer records or the same number of records with different content. The focus of this paper is not on the method of database and query generation.",
+      "context_after": "C. Transform the Query and Retrieve Semantic Information Transforming the optimized query into a poorly optimized one is a simple and automatic process. Inspired by NoREC, we transform the optimized query to the less optimized one of the form SELECT 𝛼, P is true AS flag FROM t through moving WHERE clause after the SELECT clause. Due to the absence of WHERE clause, this query forces the DBMS to retrieve all table records, which disables most potential optimizations. The transformed query evaluates whether each record in the table satisfies predicate P(i.e., the WHERE condition from the optimiz",
+      "section": "B Database and Query Generation",
+      "page": 4,
+      "char_offset": 20488,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Inspired by NoREC, we transform the optimized query to the less optimized one of the form SELECT 𝛼, P is true AS flag FROM t through moving WHERE clause after the SELECT clause.",
+      "context_before": "satisfies the predicate P, the correctly optimized result should be a single record with value v1. However, if the optimizer does not function correctly, it may produce more or fewer records or the same number of records with different content. The focus of this paper is not on the method of database and query generation. It is noted that although we only use SQLancer to generate queries, SemBug can be integrated with any generation technique. C. Transform the Query and Retrieve Semantic Information Transforming the optimized query into a poorly optimized one is a simple and automatic process.",
+      "context_after": "Due to the absence of WHERE clause, this query forces the DBMS to retrieve all table records, which disables most potential optimizations. The transformed query evaluates whether each record in the table satisfies predicate P(i.e., the WHERE condition from the optimized query). If a specific record satisfies predicate P, then the corresponding flag field will be evaluated as TRUE. After transformation, we pass two queries to the DBMS. As depicted in Figure 2 (Step ③), we first transform the optimized query into a less optimized one, and then we pass the transformed query to the DBMS. Since on",
+      "section": "C Transform the Query and Retrieve Semantic Information",
+      "page": 4,
+      "char_offset": 20765,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC cannot directly test the aggregate function because it relies on the principle that the number of records returned by two queries should match the number of records satisfying the WHERE condition.",
+      "context_before": "s for tables joining are specified in the WHERE clause. Join Clause. Beyond adding tables after the FROM clause, we can also implement more sophisticated queries by incorporatingListing 3: SemBug could test aggregate function since both queries preserve aggregate functionQuery 1: SELECT AVG (t1.c1) FROM t1WHERE t1.c1 >0 GROUP BY t1.c2; Query 2: SELECT cFROM (SELECT AVG (t1.c1) ASc, (t1.c1 >0) is true AS flag FROM t1GROUP BY t1.c2) ASt WHERE flag is true; aJOIN clause. As demonstrated in Listing 2, the inclusion of a JOIN statement does not change our transformation approach. Aggregate function.",
+      "context_after": "Aggregate function disrupts this principle by computing over multiple records to return a single value, which changes the cardinality of the original query’s result. In contrast, SemBug is not constrained by this limitation. As demonstrated in Listing 3, we maintain the aggregate function in both the original query and the semantically unoptimized query, thereby preserving the consistency of the results. Similar to aggregate function, window function is also supported by SemBug. Having clause. Having clause are typically used with GROUP BY to filter aggregated results. Since Having clause is",
+      "section": "E Supporting Features",
+      "page": 5,
+      "char_offset": 25804,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC does not support the verification of semantic information, it is incapable of testing the sorting order specified by ORDER BY.",
+      "context_before": "e function in both the original query and the semantically unoptimized query, thereby preserving the consistency of the results. Similar to aggregate function, window function is also supported by SemBug. Having clause. Having clause are typically used with GROUP BY to filter aggregated results. Since Having clause is similar to WHERE clause, the high-level idea of SemBug can be directly applied to Having clause. To do this, simply move the conditions that follow the Having clause to the end of the SELECT statement. ORDER BY. ORDER BY clause is utilized to sort the results returned by a query.",
+      "context_after": "In contrast, SemBug can perform a line-by-line examination of the returned results to ensure consistent order, thereby identifying any logic bugs related to incorrect sorting. IV. Evaluation To evaluate the effectiveness and efficiency of detecting bugs with SemBug, our evaluation aims to answer the following questions. •RQ1: Can SemBug find semantic logic bugs in the real world? •RQ2: How does SemBug perform compared to other techniques in logic bug detection? •RQ3: Compared to other techniques, does the support for advanced features in SemBug enable it to find these featuresrelated bugs? •R",
+      "section": "E Supporting Features",
+      "page": 5,
+      "char_offset": 26887,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "citation_marker",
+      "surface": "[10]",
+      "technique": "tlp",
+      "sentence": "They have also been thoroughly tested in previous work [10], [18], [19], [31].",
+      "context_before": "5.3k 0.4M 2000 PostgreSQL 16.1 14.3k 1.4M 1996 2)Tested DBMSs: We focus on testing five widely-used and large-scale DBMSs: SQLite, PostgreSQL, MySQL, MariaDB, and TiDB (see Table I). SQLite is extensively utilized in mobile and desktop applications, making it one of the most widely adopted databases; PostgreSQL is known for its strict adherence to SQL standards and seems more robust compared to other databases; MySQL is the most well-known database; MariaDB is a fork of MySQL which benefits from an active community backing; TiDB is an open-source distributed database developed by PingCap Inc.",
+      "context_after": "3)Baselines: We compare SemBug against the state-ofthe-art logic bug detection techniques, namely NoREC [18], TLP [10], and Pinolo [19], respectively. Similar to SemBug, all these methods utilize rule-based random query generation. Specifically, NoREC, TLP, and SemBug employ SQLancer [29] as the query generator, while Pinolo uses Go-Randgen [30]. Although we aim to compare these methods across all tested DBMSs, not every technique supports the full range of DBMSs being evaluated. Consequently, we exclude baseline assessments for unsupported DBMSs. TABLE II: The number of bugs found by SemBug.",
+      "section": "A Experimental Setup",
+      "page": 6,
+      "char_offset": 28926,
+      "cited_reference": {
+        "number": 10,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "3)Baselines: We compare SemBug against the state-ofthe-art logic bug detection techniques, namely NoREC [18], TLP [10], and Pinolo [19], respectively.",
+      "context_before": "ting five widely-used and large-scale DBMSs: SQLite, PostgreSQL, MySQL, MariaDB, and TiDB (see Table I). SQLite is extensively utilized in mobile and desktop applications, making it one of the most widely adopted databases; PostgreSQL is known for its strict adherence to SQL standards and seems more robust compared to other databases; MySQL is the most well-known database; MariaDB is a fork of MySQL which benefits from an active community backing; TiDB is an open-source distributed database developed by PingCap Inc. They have also been thoroughly tested in previous work [10], [18], [19], [31].",
+      "context_after": "Similar to SemBug, all these methods utilize rule-based random query generation. Specifically, NoREC, TLP, and SemBug employ SQLancer [29] as the query generator, while Pinolo uses Go-Randgen [30]. Although we aim to compare these methods across all tested DBMSs, not every technique supports the full range of DBMSs being evaluated. Consequently, we exclude baseline assessments for unsupported DBMSs. TABLE II: The number of bugs found by SemBug. The * on the MariaDB Row in the Verified column indicates that MariaDB developers have currently verified a bug and the remaining bugs are still under",
+      "section": "A Experimental Setup",
+      "page": 6,
+      "char_offset": 29005,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, NoREC, TLP, and SemBug employ SQLancer [29] as the query generator, while Pinolo uses Go-Randgen [30].",
+      "context_before": "QL is known for its strict adherence to SQL standards and seems more robust compared to other databases; MySQL is the most well-known database; MariaDB is a fork of MySQL which benefits from an active community backing; TiDB is an open-source distributed database developed by PingCap Inc. They have also been thoroughly tested in previous work [10], [18], [19], [31]. 3)Baselines: We compare SemBug against the state-ofthe-art logic bug detection techniques, namely NoREC [18], TLP [10], and Pinolo [19], respectively. Similar to SemBug, all these methods utilize rule-based random query generation.",
+      "context_after": "Although we aim to compare these methods across all tested DBMSs, not every technique supports the full range of DBMSs being evaluated. Consequently, we exclude baseline assessments for unsupported DBMSs. TABLE II: The number of bugs found by SemBug. The * on the MariaDB Row in the Verified column indicates that MariaDB developers have currently verified a bug and the remaining bugs are still under investigation.DBMS # Unique # Verified # Fixed # Cardinality # Semantic MySQL 13 13 1 9 4 TiDB 9 9 4 3 6 MariaDB 9 1* 0 5 4 SQLite 6 6 6 6 0 PostgreSQL 0 0 0 0 0 Total 37 29 11 23 14 B. RQ1: Effect",
+      "section": "A Experimental Setup",
+      "page": 6,
+      "char_offset": 29237,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": ", NoREC, TLP, and Pinolo.",
+      "context_before": "f does not introduce a direct security vulnerability, data inconsistency could lead to incorrect data being exposed or manipulated, indirectly affecting security if sensitive data is involved. Fortunately, this bug has now been fixed. Answer to RQ1: SemBug can find semantic logic bugs in the real world. SemBug have found 37 unique bugs on MySQL, TiDB, MariaDB, and SQLite for one week, 14 of which were found due to discrepancies in content. C. RQ2: Comparison on Other Techniques Method. To evaluate the performance of our method, we compare SemBug with the three state-of-the-art techniques, i.e.",
+      "context_after": "For each method, we conduct a 24hour runtime. Since not every method supports all the tested DBMSs, we have to disregard comparisons of some methods on specific DBMSs. It is noted that we implement NoREC on TiDB and MySQL to enhance the generalizability of the results. Results. The number of unique bugs found by different methods is shown in Table III. SemBug outperforms existing methods in detecting logic bugs on each tested DBMS. Compared to Pinolo, NoREC, and TLP, the number of detected bugs by SemBug increased by 13, 15, and 19, respectively. TABLE III: The number of unique bugs found by",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 34752,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M33",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "It is noted that we implement NoREC on TiDB and MySQL to enhance the generalizability of the results.",
+      "context_before": "Fortunately, this bug has now been fixed. Answer to RQ1: SemBug can find semantic logic bugs in the real world. SemBug have found 37 unique bugs on MySQL, TiDB, MariaDB, and SQLite for one week, 14 of which were found due to discrepancies in content. C. RQ2: Comparison on Other Techniques Method. To evaluate the performance of our method, we compare SemBug with the three state-of-the-art techniques, i.e., NoREC, TLP, and Pinolo. For each method, we conduct a 24hour runtime. Since not every method supports all the tested DBMSs, we have to disregard comparisons of some methods on specific DBMSs.",
+      "context_after": "Results. The number of unique bugs found by different methods is shown in Table III. SemBug outperforms existing methods in detecting logic bugs on each tested DBMS. Compared to Pinolo, NoREC, and TLP, the number of detected bugs by SemBug increased by 13, 15, and 19, respectively. TABLE III: The number of unique bugs found by SemBug and other techniques for 24 hours. The missing data indicates that the method does not support the specific DBMS. DBMS # SemBug # Pinolo # NoREC # TLP MySQL 13 8 8 7 TiDB 9 6 4 6 MariaDB 9 7 6 SQLite 3 - 2 2 PostgreSQL 0 - 0 0 Total 34 21 20 15 Improvement-13 ↑ 1",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 34945,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Compared to Pinolo, NoREC, and TLP, the number of detected bugs by SemBug increased by 13, 15, and 19, respectively.",
+      "context_before": "n on Other Techniques Method. To evaluate the performance of our method, we compare SemBug with the three state-of-the-art techniques, i.e., NoREC, TLP, and Pinolo. For each method, we conduct a 24hour runtime. Since not every method supports all the tested DBMSs, we have to disregard comparisons of some methods on specific DBMSs. It is noted that we implement NoREC on TiDB and MySQL to enhance the generalizability of the results. Results. The number of unique bugs found by different methods is shown in Table III. SemBug outperforms existing methods in detecting logic bugs on each tested DBMS.",
+      "context_after": "TABLE III: The number of unique bugs found by SemBug and other techniques for 24 hours. The missing data indicates that the method does not support the specific DBMS. DBMS # SemBug # Pinolo # NoREC # TLP MySQL 13 8 8 7 TiDB 9 6 4 6 MariaDB 9 7 6 SQLite 3 - 2 2 PostgreSQL 0 - 0 0 Total 34 21 20 15 Improvement-13 ↑ 14↑ 19↑ Overlap of bugs. Figure 4 illustrates overlap of bugs detected by SemBug and other techniques. We observe that, except NoREC, SemBug shares partial overlap with other methods in bug detection, indicating that SemBug has its unique strengths. Regarding NoREC, SemBug is capable",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 35213,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M35",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "DBMS # SemBug # Pinolo # NoREC # TLP MySQL 13 8 8 7 TiDB 9 6 4 6 MariaDB 9 7 6 SQLite 3 - 2 2 PostgreSQL 0 - 0 0 Total 34 21 20 15 Improvement-13 ↑ 14↑ 19↑ Overlap of bugs.",
+      "context_before": "d comparisons of some methods on specific DBMSs. It is noted that we implement NoREC on TiDB and MySQL to enhance the generalizability of the results. Results. The number of unique bugs found by different methods is shown in Table III. SemBug outperforms existing methods in detecting logic bugs on each tested DBMS. Compared to Pinolo, NoREC, and TLP, the number of detected bugs by SemBug increased by 13, 15, and 19, respectively. TABLE III: The number of unique bugs found by SemBug and other techniques for 24 hours. The missing data indicates that the method does not support the specific DBMS.",
+      "context_after": "Figure 4 illustrates overlap of bugs detected by SemBug and other techniques. We observe that, except NoREC, SemBug shares partial overlap with other methods in bug detection, indicating that SemBug has its unique strengths. Regarding NoREC, SemBug is capable of detecting all the bugs that NoREC finds. For Pinolo, SemBug detects 7 out of 21 logic bugs, and for TLP, SemBug finds 13 out of 15 logic bugs. Since SemBug encompasses the functionalities of NoREC, it can detect all the bugs that NoREC discovers. SemBug and TLP both heavily rely on the WHERE clause for logic bug detection, SemBugTLP M",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 35497,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M36",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "We observe that, except NoREC, SemBug shares partial overlap with other methods in bug detection, indicating that SemBug has its unique strengths.",
+      "context_before": "rms existing methods in detecting logic bugs on each tested DBMS. Compared to Pinolo, NoREC, and TLP, the number of detected bugs by SemBug increased by 13, 15, and 19, respectively. TABLE III: The number of unique bugs found by SemBug and other techniques for 24 hours. The missing data indicates that the method does not support the specific DBMS. DBMS # SemBug # Pinolo # NoREC # TLP MySQL 13 8 8 7 TiDB 9 6 4 6 MariaDB 9 7 6 SQLite 3 - 2 2 PostgreSQL 0 - 0 0 Total 34 21 20 15 Improvement-13 ↑ 14↑ 19↑ Overlap of bugs. Figure 4 illustrates overlap of bugs detected by SemBug and other techniques.",
+      "context_after": "Regarding NoREC, SemBug is capable of detecting all the bugs that NoREC finds. For Pinolo, SemBug detects 7 out of 21 logic bugs, and for TLP, SemBug finds 13 out of 15 logic bugs. Since SemBug encompasses the functionalities of NoREC, it can detect all the bugs that NoREC discovers. SemBug and TLP both heavily rely on the WHERE clause for logic bug detection, SemBugTLP MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:7 TiDB:3 SQLite:1MySQL:7 TiDB:6 SQLite:2Pinolo MySQL:8 TiDB:6 MariaDB :7SemBug MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:4 TiDB:1 MariaDB :2SemBug MySQL:5 TiDB:5 MariaDB:3 SQLite:2NoREC My",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 35748,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M37",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Regarding NoREC, SemBug is capable of detecting all the bugs that NoREC finds.",
+      "context_before": "ed by 13, 15, and 19, respectively. TABLE III: The number of unique bugs found by SemBug and other techniques for 24 hours. The missing data indicates that the method does not support the specific DBMS. DBMS # SemBug # Pinolo # NoREC # TLP MySQL 13 8 8 7 TiDB 9 6 4 6 MariaDB 9 7 6 SQLite 3 - 2 2 PostgreSQL 0 - 0 0 Total 34 21 20 15 Improvement-13 ↑ 14↑ 19↑ Overlap of bugs. Figure 4 illustrates overlap of bugs detected by SemBug and other techniques. We observe that, except NoREC, SemBug shares partial overlap with other methods in bug detection, indicating that SemBug has its unique strengths.",
+      "context_after": "For Pinolo, SemBug detects 7 out of 21 logic bugs, and for TLP, SemBug finds 13 out of 15 logic bugs. Since SemBug encompasses the functionalities of NoREC, it can detect all the bugs that NoREC discovers. SemBug and TLP both heavily rely on the WHERE clause for logic bug detection, SemBugTLP MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:7 TiDB:3 SQLite:1MySQL:7 TiDB:6 SQLite:2Pinolo MySQL:8 TiDB:6 MariaDB :7SemBug MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:4 TiDB:1 MariaDB :2SemBug MySQL:5 TiDB:5 MariaDB:3 SQLite:2NoREC MySQL:8 TiDB:4 MariaDB:6 SQLite:1 (a) (b) (c)Fig. 4: The overlap of bugs detected",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 35895,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M38",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For Pinolo, SemBug detects 7 out of 21 logic bugs, and for TLP, SemBug finds 13 out of 15 logic bugs.",
+      "context_before": "by SemBug and other techniques for 24 hours. The missing data indicates that the method does not support the specific DBMS. DBMS # SemBug # Pinolo # NoREC # TLP MySQL 13 8 8 7 TiDB 9 6 4 6 MariaDB 9 7 6 SQLite 3 - 2 2 PostgreSQL 0 - 0 0 Total 34 21 20 15 Improvement-13 ↑ 14↑ 19↑ Overlap of bugs. Figure 4 illustrates overlap of bugs detected by SemBug and other techniques. We observe that, except NoREC, SemBug shares partial overlap with other methods in bug detection, indicating that SemBug has its unique strengths. Regarding NoREC, SemBug is capable of detecting all the bugs that NoREC finds.",
+      "context_after": "Since SemBug encompasses the functionalities of NoREC, it can detect all the bugs that NoREC discovers. SemBug and TLP both heavily rely on the WHERE clause for logic bug detection, SemBugTLP MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:7 TiDB:3 SQLite:1MySQL:7 TiDB:6 SQLite:2Pinolo MySQL:8 TiDB:6 MariaDB :7SemBug MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:4 TiDB:1 MariaDB :2SemBug MySQL:5 TiDB:5 MariaDB:3 SQLite:2NoREC MySQL:8 TiDB:4 MariaDB:6 SQLite:1 (a) (b) (c)Fig. 4: The overlap of bugs detected by SemBug and other techniques. For each diagram, the left half represents the total number of bugs d",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 35974,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M39",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Since SemBug encompasses the functionalities of NoREC, it can detect all the bugs that NoREC discovers.",
+      "context_before": "rt the specific DBMS. DBMS # SemBug # Pinolo # NoREC # TLP MySQL 13 8 8 7 TiDB 9 6 4 6 MariaDB 9 7 6 SQLite 3 - 2 2 PostgreSQL 0 - 0 0 Total 34 21 20 15 Improvement-13 ↑ 14↑ 19↑ Overlap of bugs. Figure 4 illustrates overlap of bugs detected by SemBug and other techniques. We observe that, except NoREC, SemBug shares partial overlap with other methods in bug detection, indicating that SemBug has its unique strengths. Regarding NoREC, SemBug is capable of detecting all the bugs that NoREC finds. For Pinolo, SemBug detects 7 out of 21 logic bugs, and for TLP, SemBug finds 13 out of 15 logic bugs.",
+      "context_after": "SemBug and TLP both heavily rely on the WHERE clause for logic bug detection, SemBugTLP MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:7 TiDB:3 SQLite:1MySQL:7 TiDB:6 SQLite:2Pinolo MySQL:8 TiDB:6 MariaDB :7SemBug MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:4 TiDB:1 MariaDB :2SemBug MySQL:5 TiDB:5 MariaDB:3 SQLite:2NoREC MySQL:8 TiDB:4 MariaDB:6 SQLite:1 (a) (b) (c)Fig. 4: The overlap of bugs detected by SemBug and other techniques. For each diagram, the left half represents the total number of bugs detected by SemBug, while the right half corresponds to the total number of bugs detected by the other te",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 36076,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M40",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "SemBug and TLP both heavily rely on the WHERE clause for logic bug detection, SemBugTLP MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:7 TiDB:3 SQLite:1MySQL:7 TiDB:6 SQLite:2Pinolo MySQL:8 TiDB:6 MariaDB :7SemBug MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:4 TiDB:1 MariaDB :2SemBug MySQL:5 TiDB:5 MariaDB:3 SQLite:2NoREC MySQL:8 TiDB:4 MariaDB:6 SQLite:1 (a) (b) (c)Fig.",
+      "context_before": "ite 3 - 2 2 PostgreSQL 0 - 0 0 Total 34 21 20 15 Improvement-13 ↑ 14↑ 19↑ Overlap of bugs. Figure 4 illustrates overlap of bugs detected by SemBug and other techniques. We observe that, except NoREC, SemBug shares partial overlap with other methods in bug detection, indicating that SemBug has its unique strengths. Regarding NoREC, SemBug is capable of detecting all the bugs that NoREC finds. For Pinolo, SemBug detects 7 out of 21 logic bugs, and for TLP, SemBug finds 13 out of 15 logic bugs. Since SemBug encompasses the functionalities of NoREC, it can detect all the bugs that NoREC discovers.",
+      "context_after": "4: The overlap of bugs detected by SemBug and other techniques. For each diagram, the left half represents the total number of bugs detected by SemBug, while the right half corresponds to the total number of bugs detected by the other techniques. The intersection of the two halves indicates the number of bugs that were commonly detected by both SemBug and the other methods. which is why SemBug can detect the majority of bugs found by TLP. SemBug focuses on optimizing bug detection through the metamorphic relationship of equivalence, while Pinolo utilizes a metamorphic relationship based on ap",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 36180,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M41",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "which is why SemBug can detect the majority of bugs found by TLP.",
+      "context_before": "QLite:1MySQL:7 TiDB:6 SQLite:2Pinolo MySQL:8 TiDB:6 MariaDB :7SemBug MySQL:13 TiDB:9 MariaDB:9 SQLite:3MySQL:4 TiDB:1 MariaDB :2SemBug MySQL:5 TiDB:5 MariaDB:3 SQLite:2NoREC MySQL:8 TiDB:4 MariaDB:6 SQLite:1 (a) (b) (c)Fig. 4: The overlap of bugs detected by SemBug and other techniques. For each diagram, the left half represents the total number of bugs detected by SemBug, while the right half corresponds to the total number of bugs detected by the other techniques. The intersection of the two halves indicates the number of bugs that were commonly detected by both SemBug and the other methods.",
+      "context_after": "SemBug focuses on optimizing bug detection through the metamorphic relationship of equivalence, while Pinolo utilizes a metamorphic relationship based on approximation, hence the two methods have the least number of common bugs detected among all the approaches. Answer to RQ2: SemBug outperforms the state-of-theart methods in detecting logic bugs. Compared to Pinolo, NoREC, and TLP, the number of detected bugs by SemBug increased by 13, 14, and 19, respectively. D. RQ3: The Advantages of Expanded Feature Support Method. As shown in Table IV, most existing state-ofthe-art methods for logic bug",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 36919,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M42",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Compared to Pinolo, NoREC, and TLP, the number of detected bugs by SemBug increased by 13, 14, and 19, respectively.",
+      "context_before": "total number of bugs detected by the other techniques. The intersection of the two halves indicates the number of bugs that were commonly detected by both SemBug and the other methods. which is why SemBug can detect the majority of bugs found by TLP. SemBug focuses on optimizing bug detection through the metamorphic relationship of equivalence, while Pinolo utilizes a metamorphic relationship based on approximation, hence the two methods have the least number of common bugs detected among all the approaches. Answer to RQ2: SemBug outperforms the state-of-theart methods in detecting logic bugs.",
+      "context_after": "D. RQ3: The Advantages of Expanded Feature Support Method. As shown in Table IV, most existing state-ofthe-art methods for logic bug detection do not fully support the advanced features of DBMS (e.g., aggregate function, window function, and ORDER BY clause). In contrast, enhanced with semantic analysis technology, our approach is capable of supporting the most features, achieving a comparable level of support as the TLP. In this section, we investigate whether supporting these additional features can help us find more bugs. We select a bug example related to these features from unique bugs.",
+      "section": "C RQ2: Comparison on Other Techniques",
+      "page": 7,
+      "char_offset": 37335,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M43",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "In contrast, enhanced with semantic analysis technology, our approach is capable of supporting the most features, achieving a comparable level of support as the TLP.",
+      "context_before": "rphic relationship based on approximation, hence the two methods have the least number of common bugs detected among all the approaches. Answer to RQ2: SemBug outperforms the state-of-theart methods in detecting logic bugs. Compared to Pinolo, NoREC, and TLP, the number of detected bugs by SemBug increased by 13, 14, and 19, respectively. D. RQ3: The Advantages of Expanded Feature Support Method. As shown in Table IV, most existing state-ofthe-art methods for logic bug detection do not fully support the advanced features of DBMS (e.g., aggregate function, window function, and ORDER BY clause).",
+      "context_after": "In this section, we investigate whether supporting these additional features can help us find more bugs. We select a bug example related to these features from unique bugs. For simplicity, we omit the transformed query in examples because they can be directly derived from the original query. 1)Selected SQLite Bug: Incorrect SUM Function. Figure 5 shows the bug which introduced in SQLite 3.43. Two records in the table meet the where condition of the query, so the final result of the addition is 1.7E308*2. Since this result exceeds the maximum value representable in SQLite, the expected result",
+      "section": "D RQ3: The Advantages of Expanded Feature Support",
+      "page": 7,
+      "char_offset": 37712,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M44",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "SemBug and TLP support the most features.",
+      "context_before": "e transformed query in examples because they can be directly derived from the original query. 1)Selected SQLite Bug: Incorrect SUM Function. Figure 5 shows the bug which introduced in SQLite 3.43. Two records in the table meet the where condition of the query, so the final result of the addition is 1.7E308*2. Since this result exceeds the maximum value representable in SQLite, the expected result is inf, but NULL is returned, thus revealing a bug. The developers of SQLite informed us that the root cause is their implementation 130 TABLE IV: DBMS features supported by various testing methods.",
+      "context_after": "Advanced Features NoREC Pinolo TLP SemBug Join clause ✓ ✓ ✓ ✓ Having clause ✓ ✓ ✓ ✓ Group by clause ✓ ✓ ✓ ✓ Order by clause ✗ ✗ ✓ ✓ Distinct ✗ ✓ ✓ ✓ NULL value ✓ ✗ ✓ ✓ Aggregate function ✗ ✗ ✓ ✓ Window function ✗ ✗ ✓ ✓ Nondeterministic function ✗ ✗ ✗ ✗ of the Kahan-Babushka-Neumaier (KBN) algorithm to improve the precision of floating-point summation. This issue has been fixed after we reported it to the developers. CREATE TABLE t0 (c0 DOUBLE ); INSERT INTO t0(c0) VALUES (1), (2), (3); SELECT SUM (1.7E308 ) AS aggr FROM t0 WHERE c0 > 1; −− returned {null} expected {inf } Fig. 5: Incorrect qu",
+      "section": "D RQ3: The Advantages of Expanded Feature Support",
+      "page": 8,
+      "char_offset": 38678,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M45",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Advanced Features NoREC Pinolo TLP SemBug Join clause ✓ ✓ ✓ ✓ Having clause ✓ ✓ ✓ ✓ Group by clause ✓ ✓ ✓ ✓ Order by clause ✗ ✗ ✓ ✓ Distinct ✗ ✓ ✓ ✓ NULL value ✓ ✗ ✓ ✓ Aggregate function ✗ ✗ ✓ ✓ Window function ✗ ✗ ✓ ✓ Nondeterministic function ✗ ✗ ✗ ✗ of the Kahan-Babushka-Neumaier (KBN) algorithm to improve the precision of floating-point summation.",
+      "context_before": "ey can be directly derived from the original query. 1)Selected SQLite Bug: Incorrect SUM Function. Figure 5 shows the bug which introduced in SQLite 3.43. Two records in the table meet the where condition of the query, so the final result of the addition is 1.7E308*2. Since this result exceeds the maximum value representable in SQLite, the expected result is inf, but NULL is returned, thus revealing a bug. The developers of SQLite informed us that the root cause is their implementation 130 TABLE IV: DBMS features supported by various testing methods. SemBug and TLP support the most features.",
+      "context_after": "This issue has been fixed after we reported it to the developers. CREATE TABLE t0 (c0 DOUBLE ); INSERT INTO t0(c0) VALUES (1), (2), (3); SELECT SUM (1.7E308 ) AS aggr FROM t0 WHERE c0 > 1; −− returned {null} expected {inf } Fig. 5: Incorrect query result caused by number overflow in the SUM function in SQLite. Answer to RQ3: SemBug supports testing the most features compared to existing methods. The experimental results indicate that these additional features can aid in discovering more logic bugs. CREATE TABLE t0(c0 INT); INSERT INTO t0(c0) VALUES (−1); SELECT f1 FROM (SELECT (MAX (DATE('20",
+      "section": "D RQ3: The Advantages of Expanded Feature Support",
+      "page": 8,
+      "char_offset": 38720,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M46",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "6: The incorrect query result is caused by the window function, and the bug can be detected by SemBug but not by TLP.",
+      "context_before": "S (1), (2), (3); SELECT SUM (1.7E308 ) AS aggr FROM t0 WHERE c0 > 1; −− returned {null} expected {inf } Fig. 5: Incorrect query result caused by number overflow in the SUM function in SQLite. Answer to RQ3: SemBug supports testing the most features compared to existing methods. The experimental results indicate that these additional features can aid in discovering more logic bugs. CREATE TABLE t0(c0 INT); INSERT INTO t0(c0) VALUES (−1); SELECT f1 FROM (SELECT (MAX (DATE('2024−01−01')) OVER ( PARTITION BY t0.c0)) AS f1 FROM t0) AS t WHERE f1 > 1.11 E9; −− returned {2024−01−01} expected {} Fig.",
+      "context_after": "E. RQ4: Discrepancies in Logic Bug Detection Across Methods Objective. This section delves into the strengths and limitations of various methods for identifying logic bugs. As illustrated in Figure 4, SemBug is capable of detecting all bugs present in NoREC, an expected outcome given SemBug’s integration of semantic analysis techniques and support for a broader range of features. However, the figure also indicates that SemBug does not catch all logic bugs found by TLP and Pinolo. Understanding the advantages and limitations of SemBug, Pinolo, and TLP is crucial for detecting logic bugs in DBM",
+      "section": "D RQ3: The Advantages of Expanded Feature Support",
+      "page": 8,
+      "char_offset": 39795,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M47",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "As illustrated in Figure 4, SemBug is capable of detecting all bugs present in NoREC, an expected outcome given SemBug’s integration of semantic analysis techniques and support for a broader range of features.",
+      "context_before": "ental results indicate that these additional features can aid in discovering more logic bugs. CREATE TABLE t0(c0 INT); INSERT INTO t0(c0) VALUES (−1); SELECT f1 FROM (SELECT (MAX (DATE('2024−01−01')) OVER ( PARTITION BY t0.c0)) AS f1 FROM t0) AS t WHERE f1 > 1.11 E9; −− returned {2024−01−01} expected {} Fig. 6: The incorrect query result is caused by the window function, and the bug can be detected by SemBug but not by TLP. E. RQ4: Discrepancies in Logic Bug Detection Across Methods Objective. This section delves into the strengths and limitations of various methods for identifying logic bugs.",
+      "context_after": "However, the figure also indicates that SemBug does not catch all logic bugs found by TLP and Pinolo. Understanding the advantages and limitations of SemBug, Pinolo, and TLP is crucial for detecting logic bugs in DBMS. To this end, we must address two questions: (1) what are the advantages of SemBug that allow it to find bugs that TLP and Pinolo cannot? (2) what are the strengths of the other methods that enable them to detect bugs that SemBug misses? Method. The quality of seed queries significantly impacts bug detection [6], [33]. However, the core contribution of these methods lies in their",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 40086,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M48",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "However, the figure also indicates that SemBug does not catch all logic bugs found by TLP and Pinolo.",
+      "context_before": "TITION BY t0.c0)) AS f1 FROM t0) AS t WHERE f1 > 1.11 E9; −− returned {2024−01−01} expected {} Fig. 6: The incorrect query result is caused by the window function, and the bug can be detected by SemBug but not by TLP. E. RQ4: Discrepancies in Logic Bug Detection Across Methods Objective. This section delves into the strengths and limitations of various methods for identifying logic bugs. As illustrated in Figure 4, SemBug is capable of detecting all bugs present in NoREC, an expected outcome given SemBug’s integration of semantic analysis techniques and support for a broader range of features.",
+      "context_after": "Understanding the advantages and limitations of SemBug, Pinolo, and TLP is crucial for detecting logic bugs in DBMS. To this end, we must address two questions: (1) what are the advantages of SemBug that allow it to find bugs that TLP and Pinolo cannot? (2) what are the strengths of the other methods that enable them to detect bugs that SemBug misses? Method. The quality of seed queries significantly impacts bug detection [6], [33]. However, the core contribution of these methods lies in their oracle mechanisms, not in the generation of seed queries. To ensure fairness, we disregard the impact",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 40296,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M49",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Understanding the advantages and limitations of SemBug, Pinolo, and TLP is crucial for detecting logic bugs in DBMS.",
+      "context_before": "The incorrect query result is caused by the window function, and the bug can be detected by SemBug but not by TLP. E. RQ4: Discrepancies in Logic Bug Detection Across Methods Objective. This section delves into the strengths and limitations of various methods for identifying logic bugs. As illustrated in Figure 4, SemBug is capable of detecting all bugs present in NoREC, an expected outcome given SemBug’s integration of semantic analysis techniques and support for a broader range of features. However, the figure also indicates that SemBug does not catch all logic bugs found by TLP and Pinolo.",
+      "context_after": "To this end, we must address two questions: (1) what are the advantages of SemBug that allow it to find bugs that TLP and Pinolo cannot? (2) what are the strengths of the other methods that enable them to detect bugs that SemBug misses? Method. The quality of seed queries significantly impacts bug detection [6], [33]. However, the core contribution of these methods lies in their oracle mechanisms, not in the generation of seed queries. To ensure fairness, we disregard the impact of seed queries and theoretically explore whether logic bugs found by one method can be detected by others. Specific",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 40398,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M50",
+      "found_by": "citation_marker",
+      "surface": "[33]",
+      "technique": "qpg",
+      "sentence": "The quality of seed queries significantly impacts bug detection [6], [33].",
+      "context_before": "ent in NoREC, an expected outcome given SemBug’s integration of semantic analysis techniques and support for a broader range of features. However, the figure also indicates that SemBug does not catch all logic bugs found by TLP and Pinolo. Understanding the advantages and limitations of SemBug, Pinolo, and TLP is crucial for detecting logic bugs in DBMS. To this end, we must address two questions: (1) what are the advantages of SemBug that allow it to find bugs that TLP and Pinolo cannot? (2) what are the strengths of the other methods that enable them to detect bugs that SemBug misses? Method.",
+      "context_after": "However, the core contribution of these methods lies in their oracle mechanisms, not in the generation of seed queries. To ensure fairness, we disregard the impact of seed queries and theoretically explore whether logic bugs found by one method can be detected by others. Specifically, we consider the scenario where SemBug, TLP, and Pinolo generate mutated queries based on their respective mutation principles. If the outcomes of these mutated queries do not match the expected mutation relationships, this discrepancy signals the existence of a logic bug. Employing this principle, if SemBug iden",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 40759,
+      "cited_reference": {
+        "number": 33,
+        "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” in (ICSE), 2023, pp. 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M51",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Specifically, we consider the scenario where SemBug, TLP, and Pinolo generate mutated queries based on their respective mutation principles.",
+      "context_before": "in DBMS. To this end, we must address two questions: (1) what are the advantages of SemBug that allow it to find bugs that TLP and Pinolo cannot? (2) what are the strengths of the other methods that enable them to detect bugs that SemBug misses? Method. The quality of seed queries significantly impacts bug detection [6], [33]. However, the core contribution of these methods lies in their oracle mechanisms, not in the generation of seed queries. To ensure fairness, we disregard the impact of seed queries and theoretically explore whether logic bugs found by one method can be detected by others.",
+      "context_after": "If the outcomes of these mutated queries do not match the expected mutation relationships, this discrepancy signals the existence of a logic bug. Employing this principle, if SemBug identifies a logic bug, we can manually construct a mutated version of the original query from the case, using the principles of TLP and Pinolo as our guide. If the results of the mutated and original queries violate the mutation relationship, it suggests that TLP or Pinolo can detect logic bugs that SemBug uncovers. Conversely, SemBug can also use this method to evaluate whether it can detect bugs identified by o",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 41106,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M52",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Employing this principle, if SemBug identifies a logic bug, we can manually construct a mutated version of the original query from the case, using the principles of TLP and Pinolo as our guide.",
+      "context_before": "ficantly impacts bug detection [6], [33]. However, the core contribution of these methods lies in their oracle mechanisms, not in the generation of seed queries. To ensure fairness, we disregard the impact of seed queries and theoretically explore whether logic bugs found by one method can be detected by others. Specifically, we consider the scenario where SemBug, TLP, and Pinolo generate mutated queries based on their respective mutation principles. If the outcomes of these mutated queries do not match the expected mutation relationships, this discrepancy signals the existence of a logic bug.",
+      "context_after": "If the results of the mutated and original queries violate the mutation relationship, it suggests that TLP or Pinolo can detect logic bugs that SemBug uncovers. Conversely, SemBug can also use this method to evaluate whether it can detect bugs identified by other methods. Comparison of SemBug and TLP.SemBug identifies optimization bugs by suppressing optimizations in the WHERE clause. The TLP method, on the other hand, decomposes a query without WHERE clause into three sub-queries, each with a predicate after WHERE clause that evaluates to TRUE, FALSE, or NULL, ensuring the union of the sub-q",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 41393,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M53",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "If the results of the mutated and original queries violate the mutation relationship, it suggests that TLP or Pinolo can detect logic bugs that SemBug uncovers.",
+      "context_before": "the impact of seed queries and theoretically explore whether logic bugs found by one method can be detected by others. Specifically, we consider the scenario where SemBug, TLP, and Pinolo generate mutated queries based on their respective mutation principles. If the outcomes of these mutated queries do not match the expected mutation relationships, this discrepancy signals the existence of a logic bug. Employing this principle, if SemBug identifies a logic bug, we can manually construct a mutated version of the original query from the case, using the principles of TLP and Pinolo as our guide.",
+      "context_after": "Conversely, SemBug can also use this method to evaluate whether it can detect bugs identified by other methods. Comparison of SemBug and TLP.SemBug identifies optimization bugs by suppressing optimizations in the WHERE clause. The TLP method, on the other hand, decomposes a query without WHERE clause into three sub-queries, each with a predicate after WHERE clause that evaluates to TRUE, FALSE, or NULL, ensuring the union of the sub-query results aligns with the original query’s outcome. Both SemBug and TLP excel at detecting WHERE clause-related optimization bugs, as they heavily rely on the",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 41587,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M54",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Comparison of SemBug and TLP.",
+      "context_before": "mes of these mutated queries do not match the expected mutation relationships, this discrepancy signals the existence of a logic bug. Employing this principle, if SemBug identifies a logic bug, we can manually construct a mutated version of the original query from the case, using the principles of TLP and Pinolo as our guide. If the results of the mutated and original queries violate the mutation relationship, it suggests that TLP or Pinolo can detect logic bugs that SemBug uncovers. Conversely, SemBug can also use this method to evaluate whether it can detect bugs identified by other methods.",
+      "context_after": "SemBug identifies optimization bugs by suppressing optimizations in the WHERE clause. The TLP method, on the other hand, decomposes a query without WHERE clause into three sub-queries, each with a predicate after WHERE clause that evaluates to TRUE, FALSE, or NULL, ensuring the union of the sub-query results aligns with the original query’s outcome. Both SemBug and TLP excel at detecting WHERE clause-related optimization bugs, as they heavily rely on the WHERE clause for logic bug detection. SemBug’s primary advantage over TLP lies in its superior detection of optimization logic bugs, an area",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 41860,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M55",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The TLP method, on the other hand, decomposes a query without WHERE clause into three sub-queries, each with a predicate after WHERE clause that evaluates to TRUE, FALSE, or NULL, ensuring the union of the sub-query results aligns with the original query’s outcome.",
+      "context_before": "ce of a logic bug. Employing this principle, if SemBug identifies a logic bug, we can manually construct a mutated version of the original query from the case, using the principles of TLP and Pinolo as our guide. If the results of the mutated and original queries violate the mutation relationship, it suggests that TLP or Pinolo can detect logic bugs that SemBug uncovers. Conversely, SemBug can also use this method to evaluate whether it can detect bugs identified by other methods. Comparison of SemBug and TLP.SemBug identifies optimization bugs by suppressing optimizations in the WHERE clause.",
+      "context_after": "Both SemBug and TLP excel at detecting WHERE clause-related optimization bugs, as they heavily rely on the WHERE clause for logic bug detection. SemBug’s primary advantage over TLP lies in its superior detection of optimization logic bugs, an area where TLP often falls short. For instance, TLP may incorrectly assess records that should be TRUE as FALSE or NULL, missing the bug because the union of sub-query results inadvertently matches the original query’s outcome. As depicted in Figure 6, SemBug discovers a bug in the optimization of the predicate f1 > 1.11E9, which should be evaluated as FAL",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 41975,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M56",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Both SemBug and TLP excel at detecting WHERE clause-related optimization bugs, as they heavily rely on the WHERE clause for logic bug detection.",
+      "context_before": "olate the mutation relationship, it suggests that TLP or Pinolo can detect logic bugs that SemBug uncovers. Conversely, SemBug can also use this method to evaluate whether it can detect bugs identified by other methods. Comparison of SemBug and TLP.SemBug identifies optimization bugs by suppressing optimizations in the WHERE clause. The TLP method, on the other hand, decomposes a query without WHERE clause into three sub-queries, each with a predicate after WHERE clause that evaluates to TRUE, FALSE, or NULL, ensuring the union of the sub-query results aligns with the original query’s outcome.",
+      "context_after": "SemBug’s primary advantage over TLP lies in its superior detection of optimization logic bugs, an area where TLP often falls short. For instance, TLP may incorrectly assess records that should be TRUE as FALSE or NULL, missing the bug because the union of sub-query results inadvertently matches the original query’s outcome. As depicted in Figure 6, SemBug discovers a bug in the optimization of the predicate f1 > 1.11E9, which should be evaluated as FALSE, not TRUE. However, TLP fails to identify the logic bug because the union of the sub-query results aligns coincidentally with the ’2024-01-01’",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 42241,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M57",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "SemBug’s primary advantage over TLP lies in its superior detection of optimization logic bugs, an area where TLP often falls short.",
+      "context_before": "method to evaluate whether it can detect bugs identified by other methods. Comparison of SemBug and TLP.SemBug identifies optimization bugs by suppressing optimizations in the WHERE clause. The TLP method, on the other hand, decomposes a query without WHERE clause into three sub-queries, each with a predicate after WHERE clause that evaluates to TRUE, FALSE, or NULL, ensuring the union of the sub-query results aligns with the original query’s outcome. Both SemBug and TLP excel at detecting WHERE clause-related optimization bugs, as they heavily rely on the WHERE clause for logic bug detection.",
+      "context_after": "For instance, TLP may incorrectly assess records that should be TRUE as FALSE or NULL, missing the bug because the union of sub-query results inadvertently matches the original query’s outcome. As depicted in Figure 6, SemBug discovers a bug in the optimization of the predicate f1 > 1.11E9, which should be evaluated as FALSE, not TRUE. However, TLP fails to identify the logic bug because the union of the sub-query results aligns coincidentally with the ’2024-01-01’ result from the original query, which does not include a WHERE clause. This reason accounts for the majority of logic bugs that TLP",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 42386,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M58",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For instance, TLP may incorrectly assess records that should be TRUE as FALSE or NULL, missing the bug because the union of sub-query results inadvertently matches the original query’s outcome.",
+      "context_before": "on bugs by suppressing optimizations in the WHERE clause. The TLP method, on the other hand, decomposes a query without WHERE clause into three sub-queries, each with a predicate after WHERE clause that evaluates to TRUE, FALSE, or NULL, ensuring the union of the sub-query results aligns with the original query’s outcome. Both SemBug and TLP excel at detecting WHERE clause-related optimization bugs, as they heavily rely on the WHERE clause for logic bug detection. SemBug’s primary advantage over TLP lies in its superior detection of optimization logic bugs, an area where TLP often falls short.",
+      "context_after": "As depicted in Figure 6, SemBug discovers a bug in the optimization of the predicate f1 > 1.11E9, which should be evaluated as FALSE, not TRUE. However, TLP fails to identify the logic bug because the union of the sub-query results aligns coincidentally with the ’2024-01-01’ result from the original query, which does not include a WHERE clause. This reason accounts for the majority of logic bugs that TLP overlooks but SemBug catches, making it the leading cause of TLP’s deficiency in optimization logic bug detection. Furthermore, SemBug’s utilization of nested queries during constructing muta",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 42518,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M59",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "However, TLP fails to identify the logic bug because the union of the sub-query results aligns coincidentally with the ’2024-01-01’ result from the original query, which does not include a WHERE clause.",
+      "context_before": "and TLP excel at detecting WHERE clause-related optimization bugs, as they heavily rely on the WHERE clause for logic bug detection. SemBug’s primary advantage over TLP lies in its superior detection of optimization logic bugs, an area where TLP often falls short. For instance, TLP may incorrectly assess records that should be TRUE as FALSE or NULL, missing the bug because the union of sub-query results inadvertently matches the original query’s outcome. As depicted in Figure 6, SemBug discovers a bug in the optimization of the predicate f1 > 1.11E9, which should be evaluated as FALSE, not TRUE.",
+      "context_after": "This reason accounts for the majority of logic bugs that TLP overlooks but SemBug catches, making it the leading cause of TLP’s deficiency in optimization logic bug detection. Furthermore, SemBug’s utilization of nested queries during constructing mutated queries is another advantage. This approach increases pressure on the query optimizer to uncover more bugs. We find that SemBug uniquely discovers two logic bugs in TiDB attributed to this mechanism. Besides, some test cases show that TLP’s oversight of certain logic bugs also stems from the UNION operator’s potential to alter the erroneous",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 42854,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M60",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "This reason accounts for the majority of logic bugs that TLP overlooks but SemBug catches, making it the leading cause of TLP’s deficiency in optimization logic bug detection.",
+      "context_before": "optimization logic bugs, an area where TLP often falls short. For instance, TLP may incorrectly assess records that should be TRUE as FALSE or NULL, missing the bug because the union of sub-query results inadvertently matches the original query’s outcome. As depicted in Figure 6, SemBug discovers a bug in the optimization of the predicate f1 > 1.11E9, which should be evaluated as FALSE, not TRUE. However, TLP fails to identify the logic bug because the union of the sub-query results aligns coincidentally with the ’2024-01-01’ result from the original query, which does not include a WHERE clause.",
+      "context_after": "Furthermore, SemBug’s utilization of nested queries during constructing mutated queries is another advantage. This approach increases pressure on the query optimizer to uncover more bugs. We find that SemBug uniquely discovers two logic bugs in TiDB attributed to this mechanism. Besides, some test cases show that TLP’s oversight of certain logic bugs also stems from the UNION operator’s potential to alter the erroneous query plan, thereby omitting some bugs. TLP identifies two logic bugs in MySQL that SemBug overlooks, specifically in cases where the negation operator! is utilized to form the",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 43057,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M61",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Besides, some test cases show that TLP’s oversight of certain logic bugs also stems from the UNION operator’s potential to alter the erroneous query plan, thereby omitting some bugs.",
+      "context_before": "nion of the sub-query results aligns coincidentally with the ’2024-01-01’ result from the original query, which does not include a WHERE clause. This reason accounts for the majority of logic bugs that TLP overlooks but SemBug catches, making it the leading cause of TLP’s deficiency in optimization logic bug detection. Furthermore, SemBug’s utilization of nested queries during constructing mutated queries is another advantage. This approach increases pressure on the query optimizer to uncover more bugs. We find that SemBug uniquely discovers two logic bugs in TiDB attributed to this mechanism.",
+      "context_after": "TLP identifies two logic bugs in MySQL that SemBug overlooks, specifically in cases where the negation operator! is utilized to form the predicate PIS FALSE. This discovery is unexpected as TLP is not initially designed to handle such scenarios. However, the example demonstrates that TLP has the 131 CREATE VIRTUAL TABLE rt1USING rtree_i32(c0, c1, c2, +c3 INT ); INSERT INTO rt1(c0, c2, c3) VALUES ('9223372036854775807’, 1, 1); CREATE VIEW v0(c4) AS SELECT CAST (COALESCE (c0, c0) ASBLOB ) FROM rt1; SELECT (c0== CAST (c4 ASREAL ))AS f1FROM rt1, v0; --{0} SELECT (c0== CAST (c4AS REAL ))AS f1FRO",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 43513,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M62",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP identifies two logic bugs in MySQL that SemBug overlooks, specifically in cases where the negation operator!",
+      "context_before": "of logic bugs that TLP overlooks but SemBug catches, making it the leading cause of TLP’s deficiency in optimization logic bug detection. Furthermore, SemBug’s utilization of nested queries during constructing mutated queries is another advantage. This approach increases pressure on the query optimizer to uncover more bugs. We find that SemBug uniquely discovers two logic bugs in TiDB attributed to this mechanism. Besides, some test cases show that TLP’s oversight of certain logic bugs also stems from the UNION operator’s potential to alter the erroneous query plan, thereby omitting some bugs.",
+      "context_after": "is utilized to form the predicate PIS FALSE. This discovery is unexpected as TLP is not initially designed to handle such scenarios. However, the example demonstrates that TLP has the 131 CREATE VIRTUAL TABLE rt1USING rtree_i32(c0, c1, c2, +c3 INT ); INSERT INTO rt1(c0, c2, c3) VALUES ('9223372036854775807’, 1, 1); CREATE VIEW v0(c4) AS SELECT CAST (COALESCE (c0, c0) ASBLOB ) FROM rt1; SELECT (c0== CAST (c4 ASREAL ))AS f1FROM rt1, v0; --{0} SELECT (c0== CAST (c4AS REAL ))AS f1FROM rt1, v0 WHERE f1; −− returned { 0} expected { }Fig. 7: Pinolo misses the logic bug that SemBug catches, as the",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 43696,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M63",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "This discovery is unexpected as TLP is not initially designed to handle such scenarios.",
+      "context_before": "s utilization of nested queries during constructing mutated queries is another advantage. This approach increases pressure on the query optimizer to uncover more bugs. We find that SemBug uniquely discovers two logic bugs in TiDB attributed to this mechanism. Besides, some test cases show that TLP’s oversight of certain logic bugs also stems from the UNION operator’s potential to alter the erroneous query plan, thereby omitting some bugs. TLP identifies two logic bugs in MySQL that SemBug overlooks, specifically in cases where the negation operator! is utilized to form the predicate PIS FALSE.",
+      "context_after": "However, the example demonstrates that TLP has the 131 CREATE VIRTUAL TABLE rt1USING rtree_i32(c0, c1, c2, +c3 INT ); INSERT INTO rt1(c0, c2, c3) VALUES ('9223372036854775807’, 1, 1); CREATE VIEW v0(c4) AS SELECT CAST (COALESCE (c0, c0) ASBLOB ) FROM rt1; SELECT (c0== CAST (c4 ASREAL ))AS f1FROM rt1, v0; --{0} SELECT (c0== CAST (c4AS REAL ))AS f1FROM rt1, v0 WHERE f1; −− returned { 0} expected { }Fig. 7: Pinolo misses the logic bug that SemBug catches, as the erroneous second query’s output matches the expected subset of the first. potential to uncover operator-specific bugs within the orac",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 43854,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M64",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "However, the example demonstrates that TLP has the 131 CREATE VIRTUAL TABLE rt1USING rtree_i32(c0, c1, c2, +c3 INT ); INSERT INTO rt1(c0, c2, c3) VALUES ('9223372036854775807’, 1, 1); CREATE VIEW v0(c4) AS SELECT CAST (COALESCE (c0, c0) ASBLOB ) FROM rt1; SELECT (c0== CAST (c4 ASREAL ))AS f1FROM rt1, v0; --{0} SELECT (c0== CAST (c4AS REAL ))AS f1FROM rt1, v0 WHERE f1; −− returned { 0} expected { }Fig.",
+      "context_before": ". This approach increases pressure on the query optimizer to uncover more bugs. We find that SemBug uniquely discovers two logic bugs in TiDB attributed to this mechanism. Besides, some test cases show that TLP’s oversight of certain logic bugs also stems from the UNION operator’s potential to alter the erroneous query plan, thereby omitting some bugs. TLP identifies two logic bugs in MySQL that SemBug overlooks, specifically in cases where the negation operator! is utilized to form the predicate PIS FALSE. This discovery is unexpected as TLP is not initially designed to handle such scenarios.",
+      "context_after": "7: Pinolo misses the logic bug that SemBug catches, as the erroneous second query’s output matches the expected subset of the first. potential to uncover operator-specific bugs within the oracle. Similarly, TLP might also detect operator UNION related bugs, but our experiments do not yield such cases. Overall, SemBug exhibits enhanced capabilities in identifying optimization logic bugs, a domain where TLP might be theoretically inadequate. Its nested query design enables SemBug to uncover bugs that TLP usually cannot. Nevertheless, in some specific situations, TLP might identify logic bugs th",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 8,
+      "char_offset": 43942,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M65",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Similarly, TLP might also detect operator UNION related bugs, but our experiments do not yield such cases.",
+      "context_before": "wever, the example demonstrates that TLP has the 131 CREATE VIRTUAL TABLE rt1USING rtree_i32(c0, c1, c2, +c3 INT ); INSERT INTO rt1(c0, c2, c3) VALUES ('9223372036854775807’, 1, 1); CREATE VIEW v0(c4) AS SELECT CAST (COALESCE (c0, c0) ASBLOB ) FROM rt1; SELECT (c0== CAST (c4 ASREAL ))AS f1FROM rt1, v0; --{0} SELECT (c0== CAST (c4AS REAL ))AS f1FROM rt1, v0 WHERE f1; −− returned { 0} expected { }Fig. 7: Pinolo misses the logic bug that SemBug catches, as the erroneous second query’s output matches the expected subset of the first. potential to uncover operator-specific bugs within the oracle.",
+      "context_after": "Overall, SemBug exhibits enhanced capabilities in identifying optimization logic bugs, a domain where TLP might be theoretically inadequate. Its nested query design enables SemBug to uncover bugs that TLP usually cannot. Nevertheless, in some specific situations, TLP might identify logic bugs that SemBug fails to detect. Comparison of SemBug and Pinolo .We previously mentioned that Pinolo supports testing fewer DBMS features, which gives SemBug an advantage over Pinolo. We now proceed to a detailed comparison between SemBug and Pinolo from other perspectives. Unlike SemBug and TLP, Pinolo req",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 9,
+      "char_offset": 44545,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M66",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Overall, SemBug exhibits enhanced capabilities in identifying optimization logic bugs, a domain where TLP might be theoretically inadequate.",
+      "context_before": "+c3 INT ); INSERT INTO rt1(c0, c2, c3) VALUES ('9223372036854775807’, 1, 1); CREATE VIEW v0(c4) AS SELECT CAST (COALESCE (c0, c0) ASBLOB ) FROM rt1; SELECT (c0== CAST (c4 ASREAL ))AS f1FROM rt1, v0; --{0} SELECT (c0== CAST (c4AS REAL ))AS f1FROM rt1, v0 WHERE f1; −− returned { 0} expected { }Fig. 7: Pinolo misses the logic bug that SemBug catches, as the erroneous second query’s output matches the expected subset of the first. potential to uncover operator-specific bugs within the oracle. Similarly, TLP might also detect operator UNION related bugs, but our experiments do not yield such cases.",
+      "context_after": "Its nested query design enables SemBug to uncover bugs that TLP usually cannot. Nevertheless, in some specific situations, TLP might identify logic bugs that SemBug fails to detect. Comparison of SemBug and Pinolo .We previously mentioned that Pinolo supports testing fewer DBMS features, which gives SemBug an advantage over Pinolo. We now proceed to a detailed comparison between SemBug and Pinolo from other perspectives. Unlike SemBug and TLP, Pinolo requires an approximate relationship between the original query and the mutated query. For instance, the result set of a query without a WHERE c",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 9,
+      "char_offset": 44652,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M67",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Its nested query design enables SemBug to uncover bugs that TLP usually cannot.",
+      "context_before": "OM rt1; SELECT (c0== CAST (c4 ASREAL ))AS f1FROM rt1, v0; --{0} SELECT (c0== CAST (c4AS REAL ))AS f1FROM rt1, v0 WHERE f1; −− returned { 0} expected { }Fig. 7: Pinolo misses the logic bug that SemBug catches, as the erroneous second query’s output matches the expected subset of the first. potential to uncover operator-specific bugs within the oracle. Similarly, TLP might also detect operator UNION related bugs, but our experiments do not yield such cases. Overall, SemBug exhibits enhanced capabilities in identifying optimization logic bugs, a domain where TLP might be theoretically inadequate.",
+      "context_after": "Nevertheless, in some specific situations, TLP might identify logic bugs that SemBug fails to detect. Comparison of SemBug and Pinolo .We previously mentioned that Pinolo supports testing fewer DBMS features, which gives SemBug an advantage over Pinolo. We now proceed to a detailed comparison between SemBug and Pinolo from other perspectives. Unlike SemBug and TLP, Pinolo requires an approximate relationship between the original query and the mutated query. For instance, the result set of a query without a WHERE clause should ideally be a superset of the result set when the WHERE clause is pre",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 9,
+      "char_offset": 44793,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M68",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Nevertheless, in some specific situations, TLP might identify logic bugs that SemBug fails to detect.",
+      "context_before": "T (c4AS REAL ))AS f1FROM rt1, v0 WHERE f1; −− returned { 0} expected { }Fig. 7: Pinolo misses the logic bug that SemBug catches, as the erroneous second query’s output matches the expected subset of the first. potential to uncover operator-specific bugs within the oracle. Similarly, TLP might also detect operator UNION related bugs, but our experiments do not yield such cases. Overall, SemBug exhibits enhanced capabilities in identifying optimization logic bugs, a domain where TLP might be theoretically inadequate. Its nested query design enables SemBug to uncover bugs that TLP usually cannot.",
+      "context_after": "Comparison of SemBug and Pinolo .We previously mentioned that Pinolo supports testing fewer DBMS features, which gives SemBug an advantage over Pinolo. We now proceed to a detailed comparison between SemBug and Pinolo from other perspectives. Unlike SemBug and TLP, Pinolo requires an approximate relationship between the original query and the mutated query. For instance, the result set of a query without a WHERE clause should ideally be a superset of the result set when the WHERE clause is present. However, Pinolo has limitations in identifying logic bugs that meet this approximate relationshi",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 9,
+      "char_offset": 44873,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M69",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Unlike SemBug and TLP, Pinolo requires an approximate relationship between the original query and the mutated query.",
+      "context_before": "periments do not yield such cases. Overall, SemBug exhibits enhanced capabilities in identifying optimization logic bugs, a domain where TLP might be theoretically inadequate. Its nested query design enables SemBug to uncover bugs that TLP usually cannot. Nevertheless, in some specific situations, TLP might identify logic bugs that SemBug fails to detect. Comparison of SemBug and Pinolo .We previously mentioned that Pinolo supports testing fewer DBMS features, which gives SemBug an advantage over Pinolo. We now proceed to a detailed comparison between SemBug and Pinolo from other perspectives.",
+      "context_after": "For instance, the result set of a query without a WHERE clause should ideally be a superset of the result set when the WHERE clause is present. However, Pinolo has limitations in identifying logic bugs that meet this approximate relationship, which is the main reason it fails to detect some logic bugs that SemBug can catch. We have observed instances where queries with a WHERE clause incorrectly return more records, even when the results conform to the approximate relationship. As shown in Figure 7, the anticipated outcome of the second query is null, not0. However, because the erroneous outpu",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 9,
+      "char_offset": 45218,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M70",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "In summary, SemBug outperforms Pinolo in detecting logic bugs in query optimizations, similar to its advantage over TLP.",
+      "context_before": "uery is null, not0. However, because the erroneous output of 0falls within the subset of the first query’s result set, which is also 0, Pinolo fails to identify the bug. As shown in Figure 4, Pinolo detects certain logic bugs that SemBug overlooks. This is because Pinolo seeks to satisfy the approximate relationship not only through the WHERE clause but also by modifying other operators. For instance, a mutated query that adds the keyword DISTINCT should yield a result set that is a subset of the original one. We have observed that all logic bugs only found by Pinolo are due to this mechanism.",
+      "context_after": "Beyond the strengths already discussed in comparison with TLP, SemBug is also capable of detecting a wider range of features within DBMS than Pinolo. Pinolo’s advantage lies in its nature as an approximate relationship-based mutation testing method, which does not overly rely on the WHERE clause. Consequently,it identifies a minimal intersection of bugs with those found by SemBug, serving as a valuable complement to SemBug’s capabilities. Answer to RQ4: SemBug is superior at finding optimization bugs, theoretically catching all that NoREC and TLP can. While TLP occasionally spots bugs SemBug",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 9,
+      "char_offset": 46479,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M71",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Beyond the strengths already discussed in comparison with TLP, SemBug is also capable of detecting a wider range of features within DBMS than Pinolo.",
+      "context_before": "ich is also 0, Pinolo fails to identify the bug. As shown in Figure 4, Pinolo detects certain logic bugs that SemBug overlooks. This is because Pinolo seeks to satisfy the approximate relationship not only through the WHERE clause but also by modifying other operators. For instance, a mutated query that adds the keyword DISTINCT should yield a result set that is a subset of the original one. We have observed that all logic bugs only found by Pinolo are due to this mechanism. In summary, SemBug outperforms Pinolo in detecting logic bugs in query optimizations, similar to its advantage over TLP.",
+      "context_after": "Pinolo’s advantage lies in its nature as an approximate relationship-based mutation testing method, which does not overly rely on the WHERE clause. Consequently,it identifies a minimal intersection of bugs with those found by SemBug, serving as a valuable complement to SemBug’s capabilities. Answer to RQ4: SemBug is superior at finding optimization bugs, theoretically catching all that NoREC and TLP can. While TLP occasionally spots bugs SemBug doesn’t, such cases are rare. Meanwhile, Pinolo can uncover logic bugs unrelated to WHERE clause optimizations, thus providing a beneficial complement",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 9,
+      "char_offset": 46600,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M72",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Answer to RQ4: SemBug is superior at finding optimization bugs, theoretically catching all that NoREC and TLP can.",
+      "context_before": "by Pinolo are due to this mechanism. In summary, SemBug outperforms Pinolo in detecting logic bugs in query optimizations, similar to its advantage over TLP. Beyond the strengths already discussed in comparison with TLP, SemBug is also capable of detecting a wider range of features within DBMS than Pinolo. Pinolo’s advantage lies in its nature as an approximate relationship-based mutation testing method, which does not overly rely on the WHERE clause. Consequently,it identifies a minimal intersection of bugs with those found by SemBug, serving as a valuable complement to SemBug’s capabilities.",
+      "context_after": "While TLP occasionally spots bugs SemBug doesn’t, such cases are rare. Meanwhile, Pinolo can uncover logic bugs unrelated to WHERE clause optimizations, thus providing a beneficial complement to SemBug’s detection abilities. V. Discussion A. Generality The core idea of SemBug is simple but can apply to other DBMSs because most DBMSs rely on WHERE condition to filter data. However, the number of bugs found in each database is not guaranteed by our method, as it depends on the developers’ implementation approaches. Generally speaking, DBMSs that strictly adhere to the SQL standard have fewer re",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 9,
+      "char_offset": 47043,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M73",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "While TLP occasionally spots bugs SemBug doesn’t, such cases are rare.",
+      "context_before": "ations, similar to its advantage over TLP. Beyond the strengths already discussed in comparison with TLP, SemBug is also capable of detecting a wider range of features within DBMS than Pinolo. Pinolo’s advantage lies in its nature as an approximate relationship-based mutation testing method, which does not overly rely on the WHERE clause. Consequently,it identifies a minimal intersection of bugs with those found by SemBug, serving as a valuable complement to SemBug’s capabilities. Answer to RQ4: SemBug is superior at finding optimization bugs, theoretically catching all that NoREC and TLP can.",
+      "context_after": "Meanwhile, Pinolo can uncover logic bugs unrelated to WHERE clause optimizations, thus providing a beneficial complement to SemBug’s detection abilities. V. Discussion A. Generality The core idea of SemBug is simple but can apply to other DBMSs because most DBMSs rely on WHERE condition to filter data. However, the number of bugs found in each database is not guaranteed by our method, as it depends on the developers’ implementation approaches. Generally speaking, DBMSs that strictly adhere to the SQL standard have fewer related bugs. Among the bugs we have discovered, a significant portion ar",
+      "section": "E RQ4: Discrepancies in Logic Bug Detection Across Methods",
+      "page": 9,
+      "char_offset": 47158,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M74",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[35]",
+      "technique": null,
+      "sentence": "The bugs in transactions are even more complex, but there are some dedicated works [34], [35] specifically addressing transaction issues.",
+      "context_before": "it depends on the developers’ implementation approaches. Generally speaking, DBMSs that strictly adhere to the SQL standard have fewer related bugs. Among the bugs we have discovered, a significant portion are due to issues arising from different types of comparisons. Databases with strict type conversion, such as PostgreSQL, are less prone to similar bugs. B. Limitations Our current work does not support all DBMS features, such as non-deterministic functions and transactions. Regarding nondeterministic functions, this is because our approach assumes that a query always yields the same result.",
+      "context_after": "C. Threats to Validity Our experimental results face potential threats to validity. Regarding internal validity, one concern is the systematic error of the experiment. Although we repeat each experiment ten times to mitigate such errors, SQL query statements are generated randomly, so the results may vary each time, which limits reproducibility. On the other hand, not all methods employ the same query generator; differences in generated queries by various generators prevent a fairer comparison. For external validity, our method is evaluated on a limited number of DBMSs, so the conclusions dra",
+      "section": "B Limitations",
+      "page": 9,
+      "char_offset": 48221,
+      "cited_reference": {
+        "number": 35,
+        "text": "Z.-M. Jiang, S. Liu, M. Rigger, and Z. Su, “Detecting transactional bugs in database engines via Graph-Based oracle construction,” in 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). Boston, MA: USENIX Association, Jul. 2023, pp. 397–417. [Online]. Available: https://w",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M75",
+      "found_by": "citation_marker",
+      "surface": "[10]",
+      "technique": "tlp",
+      "sentence": "In recent years, metamorphic testing has been widely applied to various software systems [36]–[38], 132 and has become popular to detect logic [6], [10], [18], [19], [31] and performance bugs [39], [40] in DBMSs.",
+      "context_before": "his paper may have limitations. In the future, we plan to extend the implementation of SemBug to support more DBMSs to better assess its performance. VI. Related Work A. Metamorphic Testing of DBMS Our approach is based on metamorphic testing [20], which is designed to address the test oracle problem. The core idea of metamorphic testing is to leverage the transformation of input data to stimulate different code paths and reveal potential defects. This technique is particularly effective in identifying logic bugs by comparing the behavior of a system under varying inputs and expected outcomes.",
+      "context_after": "AMOEBA [31] constructs a pair of equivalent queries. A significant difference in execution time would indicate a performance issue. Given the seed query, CERT [40] derives a more restrictive query. The cardinality estimated by the derived query should not exceed that of the original query. For detecting logic bugs, DQE [31] focuses on detecting logic bugs in SELECT, UPDATE, and DELETE queries by ensuring that queries with the same predicate access the same rows in the table. NoREC [18] detects optimization bugs in DBMSs by converting optimized queries to unoptimized ones. TLP [10] decomposes",
+      "section": "A Metamorphic Testing of DBMS",
+      "page": 9,
+      "char_offset": 49566,
+      "cited_reference": {
+        "number": 10,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec",
+        "cert"
+      ]
+    },
+    {
+      "id": "M76",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "Given the seed query, CERT [40] derives a more restrictive query.",
+      "context_before": "everage the transformation of input data to stimulate different code paths and reveal potential defects. This technique is particularly effective in identifying logic bugs by comparing the behavior of a system under varying inputs and expected outcomes. In recent years, metamorphic testing has been widely applied to various software systems [36]–[38], 132 and has become popular to detect logic [6], [10], [18], [19], [31] and performance bugs [39], [40] in DBMSs. AMOEBA [31] constructs a pair of equivalent queries. A significant difference in execution time would indicate a performance issue.",
+      "context_after": "The cardinality estimated by the derived query should not exceed that of the original query. For detecting logic bugs, DQE [31] focuses on detecting logic bugs in SELECT, UPDATE, and DELETE queries by ensuring that queries with the same predicate access the same rows in the table. NoREC [18] detects optimization bugs in DBMSs by converting optimized queries to unoptimized ones. TLP [10] decomposes the original query into three partition sub-queries and then constructs an equivalent query by applying the union operator on these three sub-queries. Any discrepancies between the result of the ori",
+      "section": "A Metamorphic Testing of DBMS",
+      "page": 10,
+      "char_offset": 49913,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M77",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [18] detects optimization bugs in DBMSs by converting optimized queries to unoptimized ones.",
+      "context_before": "[38], 132 and has become popular to detect logic [6], [10], [18], [19], [31] and performance bugs [39], [40] in DBMSs. AMOEBA [31] constructs a pair of equivalent queries. A significant difference in execution time would indicate a performance issue. Given the seed query, CERT [40] derives a more restrictive query. The cardinality estimated by the derived query should not exceed that of the original query. For detecting logic bugs, DQE [31] focuses on detecting logic bugs in SELECT, UPDATE, and DELETE queries by ensuring that queries with the same predicate access the same rows in the table.",
+      "context_after": "TLP [10] decomposes the original query into three partition sub-queries and then constructs an equivalent query by applying the union operator on these three sub-queries. Any discrepancies between the result of the original query and the equivalent query indicate a bug. Distinct from NoREC and TLP, Pinolo [19] employs metamorphic testing based on approximate relationships to mutate the seed query. The mutated query results need to match the expected approximate relationships. Although these methods have their strengths, NoREC excels in detecting optimization bugs and identifying issues that o",
+      "section": "A Metamorphic Testing of DBMS",
+      "page": 10,
+      "char_offset": 50261,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M78",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [10] decomposes the original query into three partition sub-queries and then constructs an equivalent query by applying the union operator on these three sub-queries.",
+      "context_before": "[39], [40] in DBMSs. AMOEBA [31] constructs a pair of equivalent queries. A significant difference in execution time would indicate a performance issue. Given the seed query, CERT [40] derives a more restrictive query. The cardinality estimated by the derived query should not exceed that of the original query. For detecting logic bugs, DQE [31] focuses on detecting logic bugs in SELECT, UPDATE, and DELETE queries by ensuring that queries with the same predicate access the same rows in the table. NoREC [18] detects optimization bugs in DBMSs by converting optimized queries to unoptimized ones.",
+      "context_after": "Any discrepancies between the result of the original query and the equivalent query indicate a bug. Distinct from NoREC and TLP, Pinolo [19] employs metamorphic testing based on approximate relationships to mutate the seed query. The mutated query results need to match the expected approximate relationships. Although these methods have their strengths, NoREC excels in detecting optimization bugs and identifying issues that other methods may miss. Inspired by NoREC, SemBug retains semantic information to support a broader range of features and detect deeper semantic optimization logic bugs. B.",
+      "section": "A Metamorphic Testing of DBMS",
+      "page": 10,
+      "char_offset": 50360,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M79",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Distinct from NoREC and TLP, Pinolo [19] employs metamorphic testing based on approximate relationships to mutate the seed query.",
+      "context_before": "ld not exceed that of the original query. For detecting logic bugs, DQE [31] focuses on detecting logic bugs in SELECT, UPDATE, and DELETE queries by ensuring that queries with the same predicate access the same rows in the table. NoREC [18] detects optimization bugs in DBMSs by converting optimized queries to unoptimized ones. TLP [10] decomposes the original query into three partition sub-queries and then constructs an equivalent query by applying the union operator on these three sub-queries. Any discrepancies between the result of the original query and the equivalent query indicate a bug.",
+      "context_after": "The mutated query results need to match the expected approximate relationships. Although these methods have their strengths, NoREC excels in detecting optimization bugs and identifying issues that other methods may miss. Inspired by NoREC, SemBug retains semantic information to support a broader range of features and detect deeper semantic optimization logic bugs. B. Differential Testing of DBMS Differential testing [41] discovers bugs by giving the same input to a group of similar systems, and then checking for differences in their execution to find a bug in one of these systems. Similar to",
+      "section": "A Metamorphic Testing of DBMS",
+      "page": 10,
+      "char_offset": 50631,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M80",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Although these methods have their strengths, NoREC excels in detecting optimization bugs and identifying issues that other methods may miss.",
+      "context_before": "e rows in the table. NoREC [18] detects optimization bugs in DBMSs by converting optimized queries to unoptimized ones. TLP [10] decomposes the original query into three partition sub-queries and then constructs an equivalent query by applying the union operator on these three sub-queries. Any discrepancies between the result of the original query and the equivalent query indicate a bug. Distinct from NoREC and TLP, Pinolo [19] employs metamorphic testing based on approximate relationships to mutate the seed query. The mutated query results need to match the expected approximate relationships.",
+      "context_after": "Inspired by NoREC, SemBug retains semantic information to support a broader range of features and detect deeper semantic optimization logic bugs. B. Differential Testing of DBMS Differential testing [41] discovers bugs by giving the same input to a group of similar systems, and then checking for differences in their execution to find a bug in one of these systems. Similar to metamorphic testing, differential testing is another method to tackle test oracle problems in software testing [42]–[45]. For DBMS, it was first applied by RAGS [17], which discovers bugs by executing queries on multiple",
+      "section": "A Metamorphic Testing of DBMS",
+      "page": 10,
+      "char_offset": 50841,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M81",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Inspired by NoREC, SemBug retains semantic information to support a broader range of features and detect deeper semantic optimization logic bugs.",
+      "context_before": "he original query into three partition sub-queries and then constructs an equivalent query by applying the union operator on these three sub-queries. Any discrepancies between the result of the original query and the equivalent query indicate a bug. Distinct from NoREC and TLP, Pinolo [19] employs metamorphic testing based on approximate relationships to mutate the seed query. The mutated query results need to match the expected approximate relationships. Although these methods have their strengths, NoREC excels in detecting optimization bugs and identifying issues that other methods may miss.",
+      "context_after": "B. Differential Testing of DBMS Differential testing [41] discovers bugs by giving the same input to a group of similar systems, and then checking for differences in their execution to find a bug in one of these systems. Similar to metamorphic testing, differential testing is another method to tackle test oracle problems in software testing [42]–[45]. For DBMS, it was first applied by RAGS [17], which discovers bugs by executing queries on multiple different DBMS and comparing their results. While RAGS proves effective, its ability to test SQL queries is limited to a small subset of common SQ",
+      "section": "A Metamorphic Testing of DBMS",
+      "page": 10,
+      "char_offset": 50982,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M82",
+      "found_by": "citation_marker",
+      "surface": "[29]",
+      "technique": null,
+      "sentence": "Many approaches [6], [11], [29], [30], [33], [48] have been proposed for database and query generation.",
+      "context_before": "QL sequences, leveraging the concept of type-affinity to identify meaningful SQL statement pairings. DynSQL [14] generates more complex and valid SQL queriesby utilizing state information after the DBMS process of SQL statements. These fuzzers enhance the validity and diversity of statements to explore a broader space within DBMSs, thereby increasing the likelihood of security bug detection. We believe that the idea of increasing statement diversity can also be helpful in the detection of logic bugs. D. Query Generation of DBMS Query generator plays an important role in DBMS automatic testing.",
+      "context_after": "SQLsmith, inspired by Csmith [49], is a widely used rule-based query generator that has found over 100 bugs in popular DBMS [11]. SQLancer [29] can generate random queries for more than 10 DBMSs. Squirrel [9] employs intermediate representation (IR) and type-based mutation to generate syntactically correct queries, coupled with logic dependencies analysis to mitigate semantic errors. QPG [33] utilizes query plan as guidance to produce statements that have diverse query plans. We believe that an effective query generator can enhance the efficiency of our method for logic bug detection. VII. Co",
+      "section": "D Query Generation of DBMS",
+      "page": 10,
+      "char_offset": 53283,
+      "cited_reference": {
+        "number": 29,
+        "text": "M. Rigger, “Sqlancer,” https://github.com/sqlancer/sqlancer, 2024, accessed: 2024-03-30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M83",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [29] can generate random queries for more than 10 DBMSs.",
+      "context_before": "e fuzzers enhance the validity and diversity of statements to explore a broader space within DBMSs, thereby increasing the likelihood of security bug detection. We believe that the idea of increasing statement diversity can also be helpful in the detection of logic bugs. D. Query Generation of DBMS Query generator plays an important role in DBMS automatic testing. Many approaches [6], [11], [29], [30], [33], [48] have been proposed for database and query generation. SQLsmith, inspired by Csmith [49], is a widely used rule-based query generator that has found over 100 bugs in popular DBMS [11].",
+      "context_after": "Squirrel [9] employs intermediate representation (IR) and type-based mutation to generate syntactically correct queries, coupled with logic dependencies analysis to mitigate semantic errors. QPG [33] utilizes query plan as guidance to produce statements that have diverse query plans. We believe that an effective query generator can enhance the efficiency of our method for logic bug detection. VII. Conclusion In this paper, we propose SemBug, a novel approach for detecting logic bugs in DBMS. SemBug focuses on detecting optimization bugs. It integrates semantic analysis technology to detect se",
+      "section": "D Query Generation of DBMS",
+      "page": 10,
+      "char_offset": 53517,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M84",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG [33] utilizes query plan as guidance to produce statements that have diverse query plans.",
+      "context_before": "of logic bugs. D. Query Generation of DBMS Query generator plays an important role in DBMS automatic testing. Many approaches [6], [11], [29], [30], [33], [48] have been proposed for database and query generation. SQLsmith, inspired by Csmith [49], is a widely used rule-based query generator that has found over 100 bugs in popular DBMS [11]. SQLancer [29] can generate random queries for more than 10 DBMSs. Squirrel [9] employs intermediate representation (IR) and type-based mutation to generate syntactically correct queries, coupled with logic dependencies analysis to mitigate semantic errors.",
+      "context_after": "We believe that an effective query generator can enhance the efficiency of our method for logic bug detection. VII. Conclusion In this paper, we propose SemBug, a novel approach for detecting logic bugs in DBMS. SemBug focuses on detecting optimization bugs. It integrates semantic analysis technology to detect semantic logic bugs and support advanced DBMS features. Given a query that can be highly optimized, SemBug transforms it into a less optimized one while maintaining the semantic information. Inconsistencies in the cardinality or content of both queries’ results signify the discovery of",
+      "section": "D Query Generation of DBMS",
+      "page": 10,
+      "char_offset": 53774,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/Syang111/SemBug",
+    "markers": [
+      "renamed_sqlancer_package",
+      "sqlancer_source_content_match"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/Syang111/SemBug",
+        "source_type": "github_repository",
+        "excerpt": "# SemBug",
+        "note": "Repository is named after Sembug, the tool this paper says it built, and is about database testing.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/Syang111/SemBug/blob/master/src/Sonar/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+        "excerpt_is_verbatim": true,
+        "note": "src/Sonar/Randomly.java is SQLancer's Randomly.java, with the package renamed to Sonar (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+        "content_sha256": "sha256:f9a076f1565f5cd0d81e2adbe85664ce724fb607f37eba6ecc56e1740ca173bc",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M30"
+      ],
+      "describes_as_state_of_the_art": [
+        "M16",
+        "M60"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T01:51:46Z",
+    "is_model_written": true,
+    "summary": "SemBug detects logic bugs in DBMSs by transforming a query the optimizer can work on heavily into an equivalent but less optimized form and comparing the results. The authors divide prior work in two: approaches that target optimization bugs but check only the cardinality of the result, and approaches that check content as well but handle optimization bugs poorly. SemBug aims to cover both by adding semantic analysis to non-optimizing query construction, which also lets it test features such as aggregates and ORDER BY. It was evaluated on five widely used DBMSs against three baselines.",
+    "narrative": "SemBug is a direct descendant of NoREC: it says it transforms the optimized query into a less optimized one 'inspired by NoREC', and its evaluation treats NoREC, TLP and Pinolo as the state-of-the-art techniques to beat. It also runs on SQLancer's own machinery, using it as the query generator, and its artifact carries SQLancer's source under a renamed package.",
+    "roles": {
+      "M5": "definition",
+      "M6": "motivation",
+      "M7": "background",
+      "M8": "background",
+      "M9": "motivation",
+      "M10": "motivation",
+      "M13": "definition",
+      "M14": "motivation",
+      "M15": "motivation",
+      "M16": "motivation",
+      "M17": "background",
+      "M18": "definition",
+      "M19": "motivation",
+      "M20": "motivation",
+      "M21": "motivation",
+      "M23": "reuse_component",
+      "M25": "reuse_component",
+      "M26": "extension",
+      "M27": "motivation",
+      "M28": "motivation",
+      "M30": "baseline",
+      "M31": "reuse_component"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M23",
+          "M25",
+          "M31"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M23",
+            "sentence": "We use SQLancer [29], a rule-based method for query generation.",
+            "section": "B Database and Query Generation",
+            "page": 4
+          },
+          {
+            "mention_id": "M25",
+            "sentence": "It is noted that although we only use SQLancer to generate queries, SemBug can be integrated with any generation technique.",
+            "section": "B Database and Query Generation",
+            "page": 4
+          },
+          {
+            "mention_id": "M31",
+            "sentence": "Specifically, NoREC, TLP, and SemBug employ SQLancer [29] as the query generator, while Pinolo uses Go-Randgen [30].",
+            "section": "A Experimental Setup",
+            "page": 6
+          }
+        ],
+        "reasoning": "The paper states it uses SQLancer as its query generator, and notes that SemBug could be integrated with any generation technique -- so SQLancer supplies a component rather than the whole framework. The artifact carries SQLancer's source under a renamed package, which agrees.",
+        "reuse_kind": "generator"
+      },
+      "extends_technique": {
+        "value": "yes",
+        "mention_ids": [
+          "M26"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M26",
+            "sentence": "Inspired by NoREC, we transform the optimized query to the less optimized one of the form SELECT 𝛼, P is true AS flag FROM t through moving WHERE clause after the SELECT clause.",
+            "section": "C Transform the Query and Retrieve Semantic Information",
+            "page": 4
+          }
+        ],
+        "reasoning": "M26 states the transformation is inspired by NoREC and describes the same WHERE-to-SELECT rewrite, extended to preserve semantic information rather than only cardinality. That extension is the paper's contribution, not a baseline.",
+        "techniques": [
+          "norec"
+        ]
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M30",
+          "M31"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M30",
+            "sentence": "3)Baselines: We compare SemBug against the state-ofthe-art logic bug detection techniques, namely NoREC [18], TLP [10], and Pinolo [19], respectively.",
+            "section": "A Experimental Setup",
+            "page": 6
+          },
+          {
+            "mention_id": "M31",
+            "sentence": "Specifically, NoREC, TLP, and SemBug employ SQLancer [29] as the query generator, while Pinolo uses Go-Randgen [30].",
+            "section": "A Experimental Setup",
+            "page": 6
+          }
+        ],
+        "reasoning": "NoREC and TLP are two of the three baselines SemBug is measured against, run on the same query generator.",
+        "techniques": [
+          "norec",
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M30"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M30",
+            "sentence": "3)Baselines: We compare SemBug against the state-ofthe-art logic bug detection techniques, namely NoREC [18], TLP [10], and Pinolo [19], respectively.",
+            "section": "A Experimental Setup",
+            "page": 6
+          }
+        ],
+        "reasoning": "M30 calls NoREC, TLP and Pinolo 'the state-of-the-art logic bug detection techniques'."
+      }
+    },
+    "disagreements": [
+      "The checks did not flag M26, the sentence that establishes the extension, because 'inspired by NoREC' does not match a pattern built around 'we extend'."
+    ],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse43902_2021_00137.json b/_data/papers/paper_doi_10_1109_icse43902_2021_00137.json
new file mode 100644
index 0000000..a422d2a
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse43902_2021_00137.json
@@ -0,0 +1,654 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:23:59Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse43902.2021.00137",
+    "title": "Data-Oriented Differential Testing of Object-Relational Mapping Systems",
+    "authors": [
+      "T. Sotiropoulos",
+      "Stefanos Chaliasos",
+      "Vaggelis Atlidakis",
+      "Dimitris Mitropoulos",
+      "D. Spinellis"
+    ],
+    "year": 2021,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse43902.2021.00137",
+    "arxiv_id": null,
+    "s2_paper_id": "4bc8bd72ffa9bda2e3a7aef483008935a7a97656",
+    "url": "https://doi.org/10.1109/icse43902.2021.00137",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icse43902.2021.00137",
+      "retrieved_at": "2026-09-08T17:23:59Z",
+      "chars": 74754,
+      "content_sha256": "sha256:e8d83720a8e72e972b1e09cf9a271ce27927c40e0d29f413244fb90a4a192847"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Random generation",
+        "start": 16913
+      },
+      {
+        "number": "A",
+        "title": "Schema Generation & Setup",
+        "start": 18726
+      },
+      {
+        "number": "B",
+        "title": "Abstract Query Generation",
+        "start": 21052
+      },
+      {
+        "number": "C",
+        "title": "Concretization of Abstract Queries",
+        "start": 29079
+      },
+      {
+        "number": "D",
+        "title": "Bug Detection",
+        "start": 36904
+      },
+      {
+        "number": "E",
+        "title": "Implementation Details",
+        "start": 38141
+      },
+      {
+        "number": "A",
+        "title": "Experimental Setup",
+        "start": 40076
+      },
+      {
+        "number": "B",
+        "title": "RQ1: New Bugs Found",
+        "start": 42005
+      },
+      {
+        "number": "C",
+        "title": "RQ2: Characteristics of Discovered Bugs",
+        "start": 43244
+      },
+      {
+        "number": "D",
+        "title": "RQ3: Effectiveness of Solver-Based Data Generation",
+        "start": 50073
+      },
+      {
+        "number": "E",
+        "title": "Discussion & Threats to Validity",
+        "start": 53130
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "A. Torres, R. Galante, M. S. Pimenta, and A. J. B. Martins, “Twenty years of object-relational mapping: A survey on patterns, solutions, and their implications on application design,” Information and Software Technology, vol. 82, pp. 1 - 18, 2017. [Online]. Available: http://www.sciencedirect.com/science/article/pii/S0950584916301859",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "T.-H. Chen, W. Shang, Z. M. Jiang, A. E. Hassan, M. Nasser, and P. Flora, “Detecting performance anti-patterns for applications developed using object-relational mapping,” in Proceedings of the 36th International Conference on Software Engineering, ser. ICSE 2014. New York, NY, USA: Association for Computing Machinery, 2014, p. 1001-1012. [Online]. Available: https ://doi.org/10.1145/2568225. 2568",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "K. Roebuck, Object-Relational M apping (ORM): High-Impact Strategies-What You Need to Know: Definitions, Adoptions, Impact, Benefits, Maturity, Vendors. Emereo Publishing, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "C. Bauer, G. King, and G. Gregory, Java Persistence with Hibernate, 2nd ed. USA: Manning Publications Co., 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "D. Maier, Representing Database Programs as Objects. New York, NY, USA: Association for Computing Machinery, 1990, p. 377-386. [Online]. Available: https://doi.org/10.1145/101620.101642",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "M. Eltsufin, “Bringing Hibernate ORM to cloud Spanner for database adoption,” https://cl oud.google.com/blog/products/databases/ bringing-hibernate-ormcloudspanner-databaseadoption, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "T. Chen, W. Shang, J. Yang, A. E. Hassan, M. W. Godfrey, M. Nasser, and P. Flora, “An empirical study on the practice of maintaining ObjectRelational Mapping code in Java systems,” in 2016 IEEE/ACM 13th Working Conference on Mining Software Repositories (MSR), 2016, pp. 165-176.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "M. Bayer, “SQLAlchemythe database toolkit for Python,” https://www. sqlalchemy.org/, 2020, [Online; accessed 29-July-2020].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "D. S. Foundation, “The web framework for perfectionists with deadlines,” https://www.dja ngoproject.com/, 2 020, [Online; accessed 29-July2020].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "J. Yang, P. Subramaniam, S. Lu, C. Yan, and A. Che ung, “How ^¿n o tj^ to structure your database-backed web applications: A study of performance bugs in the wild,” in Proceedings of the 40th International Conference on Software Engineering, ser. ICSE ’18. New York, NY, USA: Association for Computing Machinery, 2018, p. 800-810. [Online]. Available: https ://doi.org/10.1145/3180155.3180194",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "D. S. F oundation, “Django issues,” https ://code.dja ngoproject.com/ query, 2020, [Online; accessed 29-July-2020].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "A. Viswa, “select_for _update() with ”of” uses wrong tables from (multilevel) model inheritance,” https://code.dja ngoproject.com /ticket/ 31246, 2020, [Online; accessed 29-July-2020].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "S. Bank, “negated EXISTS result type not bool with SQLite dialect,” https://git hub.com/sqlalchemy/sqlalchemy/issues/3682, 2016, [Online; accessed 29-July-2020].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "“CVE- 2019-7164,” https:// nvd.nist.gov/vuln/deta il/CVE- 2019-7164, 2019, [Online; accessed 29-July-2020].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "“CVE- 2020-9402,” https:// nvd.nist.gov/vuln/deta il/CVE- 2020-9402, 2020, [Online; accessed 29-July-2020].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "X. Yang, Y. Chen, E. Eide, and J. Regehr, “Finding and understanding bugs in C compilers,” in Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation, ser. PLDI ’11. New York, NY, USA: Association for Computing Machinery, 2011, p. 283-294. [Online]. Available: https ://doi.org/10. 1145/199 3498.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Y. Chen, T. Su, C. Sun, Z. Su, and J. Zhao, “Coveragedirected differential testing of JVM implementations,” SIGPLAN Not., vol. 51, no. 6, p. 85-99, Jun. 2016. [Online]. Available: https://doi.org/10.1145/2980983.2908095",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Y. Chen, T. Su, and Z. Su, “Deep differential testing of JVM implementations,” in Proceedings of the 41st International Conference on Software Engineering, ser. ICSE ’19. IEEE Press, 2019, p. 1257-1268. [Online]. Available: https ://doi.org/10.1109/ICSE.2019. 00127",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "C. Lidbury, A. Lascu, N. C hong, and A. F. Donaldson, “Many-core compiler fuzzing,” SIGPLAN Not., vol. 50, no. 6, p. 65-76, Jun. 2015. [Online]. Available: https://doi.org/10. 1145/2813885.2737986",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "C. Sun, V. Le, and Z. Su, “Finding and analyzing compiler warning defects,” in Proceedings of the 38th Internati onal Conference on Software Engineering, ser. ICSE ’16. New York, NY, USA:Association for Computing Machinery, 2016, p. 203-213. [Online]. Available: https://doi.org/10.1145/2884781.2884879",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, Nov. 2020, pp. 667-682. [Online]. Available: https ://www.usenix.org/conference/ osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "W. M. McKeeman, “Differential testing for software,” Digital Technical Journal, vol. 10, no. 1, pp. 100-107, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "D. S. Foundation, “Integrating Dj ango with a legacy database,” https :// docs.djangoproject.com/en/3.0/howto/legacy-databases/, 2020, [Online; accessed 29- July-2020].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "P. Haller, A. Prokopec, H. M iller, V. Klang, R. K uhn, and V. Jovanovic, “Futures and promises,” https://docs.scala-lang.org/overviews/ core/futures.html, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "L. de Moura and N. Bjprner, “Z3: An efficient SMT solver,” in Tools and Algorithms for the Construction and Analysis of Systems, C. R. Ramakrishnan and J. Rehof, Eds. Berlin, Heidelberg: Springer Berlin Heidelberg, 2008, pp. 337-340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "V. Atlidakis, J. Andrus, R. Geambasu, D. Mitr opoulos, and J. Nieh, “POSIX abstractions in modern operating systems: The old, the new, and the missing,” in Proceedings of the Eleventh European Conference on Computer Systems, ser. EuroSys ’16. New York, NY, USA: Association for Computing Machinery, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. New York, NY, USA: Association for Computing Machinery, 2020, p. 1140-1152. [Online]. Available: https ://doi.org/10.1145/3368089. 3",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "Z. Davar and Handoko, “Refactoring object-relational database applications by applying transformation rules to develop better performance,” in Proceedings of the 16th International Conference on Information Integration and Web-Based Applications & Services, ser. ii WAS ’14. New York, NY, USA: Association for Computing Machinery, 2014, p. 283-288. [Online]. Available: https ://doi.org/10. 1145/268 4",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "R. Singh, C.-P. Bezemer, W. Shang, and A. E. Hassan, “Optimizing the performance-related configurations of object-relational mapping frameworks using a multi-objective genetic algorithm,” in Proceedings of the 7th ACM/SPEC on International Conference on Performance Engineering, ser. ICPE ’16. New York, NY, USA: Association for Computing Machinery, 2016, p. 309-320. [Online]. Available: https://doi",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "D. R. Slutz, “Massive stochastic testing of SQL,” in Proceedings of the 24rd International Conference on Very Large Data Bases, ser. VLDB ’98. San Francisco, CA, USA: Morgan Kaufmann Publishers Inc., 1998, p. 618- 622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "N. Bruno, S. Chaudhuri, and D. Thomas, “Generating queries with cardinality constraints for DBMS testing,” IEEE Transactions on Knowledge and Data Engineering, vol. 18, no. 12, pp. 1721-1725, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "H. Bati, L. Giakoumakis, S. Herbert, and A. Surna, “A genetic approach for random testing of database systems,” in Proceedings of the 33rd International Conference on Very Large Data Bases, ser. VLDB ’07. VLDB Endowment, 2007, p. 1243-1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "C. Mishra, N. Koudas, and C. Zuzarte, “Generating targeted queries for database testing,” in Proceedings of the 2008 ACMSIGMOD International Conference on Management of Data, ser. SIGMOD ’08. New York, NY, USA: Association for Computing Machinery, 2008, p. 499-510. [Online]. Available: https ://doi.org/10.1145/1376616.1376668",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "S. Abdul Khalek and S. Khurshid, “Automated SQL query generation for systematic testing of database engines,” in Proceedings of the IEEE/ACM International Conference on Automated Software Engineering, ser. ASE ’10. New York, NY, USA: Association for Computing Machinery, 2010, p. 329-332. [Online]. Available: https://doi.org/10.1145/1858996.1859063",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "A. Seltenreich, “SQLsmith: A random SQL query generator,” https :// github.com/anse1/sqlsmith, 2020, [Online; accessed 29-July-2020].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic detection and diagnosis of performance regressions in database systems,” Proc. VLDB E ndow., vol. 13, no. 1, p. 57-70, Sep. 2019. [Online]. Available: https ://doi.org/10.14778/3357377.3357382 1546",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "S. A bdul Khalek, B. Elkarablieh, Y. O. Laleye, and S. Khurshid, “Query-aware test generation using a relational constraint solver,” in 2008 23rd IEEE/ACM Internati onal Conference on Automated Software Engineering, 2008, pp. 238-247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, Nov. 2020. [Online]. Available: https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "T. Petsios, A. Tang, S. Stolfo, A. D. Keromytis, and S. Jana, “NEZHA: Efficient domain-independent differential testing,” in 2017 IEEE Symposium on Security and Privacy (SP), 2017, pp. 615-632.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "E. J. Weyuker, “On testing non-testable programs,” The Computer Journal, vol. 25, no. 4, pp. 465- 470, 1982.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "C. Klinger, M. Christakis, and V. Wüstholz, “Differentially testing soundness and precision of program analyzers,” in Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis, ser. ISSTA 2019. New York, NY, USA: Association for Computing Machinery, 2019, p. 239-250. [Online]. Available: https://doi.org/10.1145/3293882.3330553",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "D. Lehmann and M. Pradel, “Feedback-directed differential testing of interactive debuggers,” in Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ser. ESEC/FSE 2018. New York, NY, USA: Association for Computing Machinery, 2018, p. 610-620. [Online]. Available: https ://doi.org/10.1145/3236024.3236037",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "T. Kapus and C. Cadar, “Automatic testing of symbolic execution engines via program generation and differential testing,” in Proceedings of the 32nd IEEE/ACM International Conference on Automated Software Engineering, ser. ASE 2017. IEEE Press, 2017, p. 590-600.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "S. Dutta, O. Le gunsen, Z. Huang, and S. Misailovic, “Testing probabilistic programming systems,” in Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ser. ESEC/FSE 2018. New York, NY, USA: Association for Computing Machinery, 2018, p. 574- 586. [Online]. Available: https: //doi.org/10.1145/3236024.3",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "S. Srisaka okul, Z. Wu, A. Astorga, O. Alebiosu, and T. Xie, “Mu ltipleimplementation testing of supervised learning software,” in Workshops at the Thirty-Second AAAI Conference on Ar tificial Intelligence, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "M. Sela kovic, M. Pradel, R. Karim, and F. Tip, “Test generation for higher-order functions in dynamic languages,” Proc. ACM Program. Lang., vol. 2, no. OOPSLA, Oct. 2018. [Online]. Available: https://doi.org/10.1145/3276531",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Y. Chen and Z. Su, “Guided differential testing of certificate validation in SSL/TLS implementations,” ser. ESEC/FSE 2015. New York, NY, USA: Association for Computing Machinery, 2015, p. 793-804. [Online]. Available: https://doi.org/10.1145/2786805.2786835",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "P. Godefroid, D. Lehmann, and M. Polishc huk, “Differential regression testing for REST APIs,” in Proceedings of the 29th ACMSIGSOFT International Symposium on Software Testing and Analysis, ser. ISSTA 2020. New York, NY, USA: Association for Computing Machinery, 2020, p. 312-323. [Online]. Available: https ://doi.org/10.1145/3395363. 3397374 1547",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 21,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, Nov. 2020, pp. 667-682. [Online]. Available: https ://www.usenix.org/conference/ osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. New York, NY, USA: Association for Computing Machinery, 2020, p. 1140-1152. [Online]. Available: https ://doi.org/10.1145/3368089. 3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 38,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, Nov. 2020. [Online]. Available: https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[21]",
+      "technique": "pqs",
+      "sentence": "Although DBMSs share common functionality, they differ significantly from each other [21].",
+      "context_before": "s and runs the SQL query shown on lines 7-11. This SQL query is invalid on MySQL and the Django program crashes with a django.db.utils.ProgrammingError: (1064, “ Error in SQL syntax; check the manual that corresponds to y our MySQL server version for the right syntax to use near ’UNION’” ). This bug was detected by our approach, and was confirmed by the Django developers. When the Django code shown in Figure 2 is run on another DBMS, such as SQLite or PostgreSQL, Django produces a valid SQL query. Such inconsistencies indicate that ORM bugs may appear (or not) depending on the underlying DBMS.",
+      "context_after": "Therefore, an ORM needs to abstract away such differences and take care of running the same ORM code on different DBMSs reliably. Unfortunately, this complicates the design of ORMs: bugs may occur when an ORM fails to produce a valid SQL query with respect to a certain DBMS. Bug in peewee. Figure 3 shows another ORM bug detected by our approach. On lines 1-3, the code creates a simple query using the peewee ORM. The query defines a simple 1536 1 expr = (1 + T.col) 2 squared = (expr * expr) 3 T.select(fn.sum(expr), fn.avg(squared)).all() 4 // Generated SQL 5 SELECT SUM(1 + \"t\".\"col\"), 6 AVG(",
+      "section": null,
+      "page": 2,
+      "char_offset": 10107,
+      "cited_reference": {
+        "number": 21,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, Nov. 2020, pp. 667-682. [Online]. Available: https ://www.usenix.org/conference/ osdi20/presentation/rigger",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[21]",
+      "technique": "pqs",
+      "sentence": "To demonstrate the effectiveness of our solver-based data generation approach and its suitability for differential testing, we compare it against a simplistic approach that populates the database with random records a-priori [21], [27], i.",
+      "context_before": "he outputs of ORMs, it is important that ORMs return non-empty results for the given queries. Empty results indicate that the corresponding query was unsatisfied with respect to the data inserted to the database. Empty results can potentially hide logic errors that otherwise would be uncovered if the corresponding ORMs could get some data from the database and we were able to notice differences in their results. 1543 50 40 SO Z0 10 0 pcJ^ -tè i0' &ST *sr Data Generation Strategy Fig. 10: Percentage of the unsatisfied queries per data generation strategy using a sample of 20 testing sessions.",
+      "context_after": "e., it inserts data while setting up the tables, without considering the constraints of the generated queries. We used CYNTHIA to spawn 20 testing sessions. For each testing session, we generated 100 queries and compared the results of ORMs as usual. At the end of each testing session, we measured in how many queries the ORMs returned empty results. We then replayed each testing session, using a naive data generation strategy, and tried out different settings: generating 50 random records, 100, 300, 500 and finally 1000. Figure 10 illustrates the comparison results. The y-axis shows the percen",
+      "section": "D RQ3: Effectiveness of Solver-Based Data Generation",
+      "page": 10,
+      "char_offset": 50775,
+      "cited_reference": {
+        "number": 21,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, Nov. 2020, pp. 667-682. [Online]. Available: https ://www.usenix.org/conference/ osdi20/presentation/rigger",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "[21] proposed the Pivoted Query Synthesis (PQS) technique for testing database engines.",
+      "context_before": "differential testing to detect logic errors beyond crashes or regression bugs. Kh alek et al. [37], [34] followed a solver-based approach for testing DBMSs. Their work employs a relational constraint solver to generate valid database records with respect to a given SQL query and database schema. Besides populating the database, their method also determines the expected results of an SQL query and the authors use this oracle to find bugs. We also use an SMT-solver to populate the database, but we specify the test oracle by adopting a differential testing approach. More r ecently, Rigger et al.",
+      "context_after": "PQS generates SQL queries so that they fetch a specific record from the database. In this way, PQS forms the test oracle: failing to fetch the expected record reveals a potential bug in DBMS. Unlike this work, our approach adopts differential testing for determining the oracle. Also, beyond reasoning about a single record, our approach is able to detect bugs involving operations on result sets (e.g., aggregate functions, sorting, distinct). In an attempt to find optimization bugs in database systems, their subsequent work introduced a metamorphic testing technique called Non-Optimizing Refere",
+      "section": "E Discussion & Threats to Validity",
+      "page": 11,
+      "char_offset": 58777,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS generates SQL queries so that they fetch a specific record from the database.",
+      "context_before": "et al. [37], [34] followed a solver-based approach for testing DBMSs. Their work employs a relational constraint solver to generate valid database records with respect to a given SQL query and database schema. Besides populating the database, their method also determines the expected results of an SQL query and the authors use this oracle to find bugs. We also use an SMT-solver to populate the database, but we specify the test oracle by adopting a differential testing approach. More r ecently, Rigger et al. [21] proposed the Pivoted Query Synthesis (PQS) technique for testing database engines.",
+      "context_after": "In this way, PQS forms the test oracle: failing to fetch the expected record reveals a potential bug in DBMS. Unlike this work, our approach adopts differential testing for determining the oracle. Also, beyond reasoning about a single record, our approach is able to detect bugs involving operations on result sets (e.g., aggregate functions, sorting, distinct). In an attempt to find optimization bugs in database systems, their subsequent work introduced a metamorphic testing technique called Non-Optimizing Reference Engine Construction (NoREC) [27]. At a high-level, NoREC applies a semanticspr",
+      "section": "E Discussion & Threats to Validity",
+      "page": 11,
+      "char_offset": 58865,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "In this way, PQS forms the test oracle: failing to fetch the expected record reveals a potential bug in DBMS.",
+      "context_before": "mploys a relational constraint solver to generate valid database records with respect to a given SQL query and database schema. Besides populating the database, their method also determines the expected results of an SQL query and the authors use this oracle to find bugs. We also use an SMT-solver to populate the database, but we specify the test oracle by adopting a differential testing approach. More r ecently, Rigger et al. [21] proposed the Pivoted Query Synthesis (PQS) technique for testing database engines. PQS generates SQL queries so that they fetch a specific record from the database.",
+      "context_after": "Unlike this work, our approach adopts differential testing for determining the oracle. Also, beyond reasoning about a single record, our approach is able to detect bugs involving operations on result sets (e.g., aggregate functions, sorting, distinct). In an attempt to find optimization bugs in database systems, their subsequent work introduced a metamorphic testing technique called Non-Optimizing Reference Engine Construction (NoREC) [27]. At a high-level, NoREC applies a semanticspreserving transformation to a given SQL query in way that the various optimizations performed by the DBMS are d",
+      "section": "E Discussion & Threats to Validity",
+      "page": 11,
+      "char_offset": 58947,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "Non-Optimizing Reference Engine Construction",
+      "technique": "norec",
+      "sentence": "In an attempt to find optimization bugs in database systems, their subsequent work introduced a metamorphic testing technique called Non-Optimizing Reference Engine Construction (NoREC) [27].",
+      "context_before": "ting a differential testing approach. More r ecently, Rigger et al. [21] proposed the Pivoted Query Synthesis (PQS) technique for testing database engines. PQS generates SQL queries so that they fetch a specific record from the database. In this way, PQS forms the test oracle: failing to fetch the expected record reveals a potential bug in DBMS. Unlike this work, our approach adopts differential testing for determining the oracle. Also, beyond reasoning about a single record, our approach is able to detect bugs involving operations on result sets (e.g., aggregate functions, sorting, distinct).",
+      "context_after": "At a high-level, NoREC applies a semanticspreserving transformation to a given SQL query in way that the various optimizations performed by the DBMS are disabled.Finally, NoREC compares the results of the original and the resulting queries for mismatches. In their most recent work, they propose Ternary Logic Partitioning (TLP) [38]. Given an SQL query, TLP derives multiple queries that compute a partial result of the initial query, and then combines the results of each individual query using a UNION operation. If the result of the combined query does not match that of the initial one, then a",
+      "section": "E Discussion & Threats to Validity",
+      "page": 11,
+      "char_offset": 59310,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "At a high-level, NoREC applies a semanticspreserving transformation to a given SQL query in way that the various optimizations performed by the DBMS are disabled.",
+      "context_before": "ey fetch a specific record from the database. In this way, PQS forms the test oracle: failing to fetch the expected record reveals a potential bug in DBMS. Unlike this work, our approach adopts differential testing for determining the oracle. Also, beyond reasoning about a single record, our approach is able to detect bugs involving operations on result sets (e.g., aggregate functions, sorting, distinct). In an attempt to find optimization bugs in database systems, their subsequent work introduced a metamorphic testing technique called Non-Optimizing Reference Engine Construction (NoREC) [27].",
+      "context_after": "Finally, NoREC compares the results of the original and the resulting queries for mismatches. In their most recent work, they propose Ternary Logic Partitioning (TLP) [38]. Given an SQL query, TLP derives multiple queries that compute a partial result of the initial query, and then combines the results of each individual query using a UNION operation. If the result of the combined query does not match that of the initial one, then a bug is found. TLP is suitable for testing the implementation of the WHERE, HAVING, DISTI NCT clauses, or aggregate functions. All these previous approaches are tai",
+      "section": "E Discussion & Threats to Validity",
+      "page": 11,
+      "char_offset": 59502,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Finally, NoREC compares the results of the original and the resulting queries for mismatches.",
+      "context_before": "this work, our approach adopts differential testing for determining the oracle. Also, beyond reasoning about a single record, our approach is able to detect bugs involving operations on result sets (e.g., aggregate functions, sorting, distinct). In an attempt to find optimization bugs in database systems, their subsequent work introduced a metamorphic testing technique called Non-Optimizing Reference Engine Construction (NoREC) [27]. At a high-level, NoREC applies a semanticspreserving transformation to a given SQL query in way that the various optimizations performed by the DBMS are disabled.",
+      "context_after": "In their most recent work, they propose Ternary Logic Partitioning (TLP) [38]. Given an SQL query, TLP derives multiple queries that compute a partial result of the initial query, and then combines the results of each individual query using a UNION operation. If the result of the combined query does not match that of the initial one, then a bug is found. TLP is suitable for testing the implementation of the WHERE, HAVING, DISTI NCT clauses, or aggregate functions. All these previous approaches are tailored to testing DBMSs, i.e., they aim to find DBMS-specific bugs (e.g., opti-mization bugs,",
+      "section": "E Discussion & Threats to Validity",
+      "page": 11,
+      "char_offset": 59665,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "In their most recent work, they propose Ternary Logic Partitioning (TLP) [38].",
+      "context_before": "reasoning about a single record, our approach is able to detect bugs involving operations on result sets (e.g., aggregate functions, sorting, distinct). In an attempt to find optimization bugs in database systems, their subsequent work introduced a metamorphic testing technique called Non-Optimizing Reference Engine Construction (NoREC) [27]. At a high-level, NoREC applies a semanticspreserving transformation to a given SQL query in way that the various optimizations performed by the DBMS are disabled.Finally, NoREC compares the results of the original and the resulting queries for mismatches.",
+      "context_after": "Given an SQL query, TLP derives multiple queries that compute a partial result of the initial query, and then combines the results of each individual query using a UNION operation. If the result of the combined query does not match that of the initial one, then a bug is found. TLP is suitable for testing the implementation of the WHERE, HAVING, DISTI NCT clauses, or aggregate functions. All these previous approaches are tailored to testing DBMSs, i.e., they aim to find DBMS-specific bugs (e.g., opti-mization bugs, bugs associated with the evaluation of WHERE clauses). ORM systems differ from",
+      "section": "E Discussion & Threats to Validity",
+      "page": 11,
+      "char_offset": 59758,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Given an SQL query, TLP derives multiple queries that compute a partial result of the initial query, and then combines the results of each individual query using a UNION operation.",
+      "context_before": "operations on result sets (e.g., aggregate functions, sorting, distinct). In an attempt to find optimization bugs in database systems, their subsequent work introduced a metamorphic testing technique called Non-Optimizing Reference Engine Construction (NoREC) [27]. At a high-level, NoREC applies a semanticspreserving transformation to a given SQL query in way that the various optimizations performed by the DBMS are disabled.Finally, NoREC compares the results of the original and the resulting queries for mismatches. In their most recent work, they propose Ternary Logic Partitioning (TLP) [38].",
+      "context_after": "If the result of the combined query does not match that of the initial one, then a bug is found. TLP is suitable for testing the implementation of the WHERE, HAVING, DISTI NCT clauses, or aggregate functions. All these previous approaches are tailored to testing DBMSs, i.e., they aim to find DBMS-specific bugs (e.g., opti-mization bugs, bugs associated with the evaluation of WHERE clauses). ORM systems differ from database engines, and suffer from other types of bugs. Differential Testing. Differential testing [22], [39] is a generally-applicable testing technique that aims to find bugs in so",
+      "section": "E Discussion & Threats to Validity",
+      "page": 11,
+      "char_offset": 59837,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP is suitable for testing the implementation of the WHERE, HAVING, DISTI NCT clauses, or aggregate functions.",
+      "context_before": "vel, NoREC applies a semanticspreserving transformation to a given SQL query in way that the various optimizations performed by the DBMS are disabled.Finally, NoREC compares the results of the original and the resulting queries for mismatches. In their most recent work, they propose Ternary Logic Partitioning (TLP) [38]. Given an SQL query, TLP derives multiple queries that compute a partial result of the initial query, and then combines the results of each individual query using a UNION operation. If the result of the combined query does not match that of the initial one, then a bug is found.",
+      "context_after": "All these previous approaches are tailored to testing DBMSs, i.e., they aim to find DBMS-specific bugs (e.g., opti-mization bugs, bugs associated with the evaluation of WHERE clauses). ORM systems differ from database engines, and suffer from other types of bugs. Differential Testing. Differential testing [22], [39] is a generally-applicable testing technique that aims to find bugs in software implementations by addressing the oracle problem [40]. Differential testing has been successfully applied to various domains, most notably compilers and runtime systems [16], [19], [17], [18], [20]. Fol",
+      "section": "E Discussion & Threats to Validity",
+      "page": 11,
+      "char_offset": 60115,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:25:35Z",
+    "is_model_written": true,
+    "summary": "Cynthia is, as far as the authors know, the first systematic approach to testing Object-Relational Mapping systems. It generates random relational schemas, sets up the corresponding databases, and queries them through the APIs of the ORMs under test. Because ORMs share no common input language, queries are written in an abstract query language and translated into concrete executable ORM queries for differential comparison. A solver-based approach produces records targeted at each query's constraints. Cynthia found 28 bugs in five popular ORMs.",
+    "narrative": "All three SQLancer oracles are described in detail in the discussion -- PQS forming its oracle by fetching a specific record, NoREC's semantics-preserving transformation, TLP deriving partial-result queries -- as the DBMS-level precedent for Cynthia's own differential testing one layer above.",
+    "roles": {
+      "M1": "definition",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "definition",
+      "M5": "definition",
+      "M6": "definition",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "definition",
+      "M10": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse48619_2023_00024.json b/_data/papers/paper_doi_10_1109_icse48619_2023_00024.json
new file mode 100644
index 0000000..bd0a31d
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse48619_2023_00024.json
@@ -0,0 +1,510 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:55:15Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse48619.2023.00024",
+    "title": "A Comprehensive Study of Real-World Bugs in Machine Learning Model Optimization",
+    "authors": [
+      "Hao Guan",
+      "Ying Xiao",
+      "Jiaying Li",
+      "Yepang Liu",
+      "Guangdong Bai"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse48619.2023.00024",
+    "arxiv_id": null,
+    "s2_paper_id": "35b7f91900e5d37aaf160bb0590a812754fd3137",
+    "url": "https://doi.org/10.1109/icse48619.2023.00024",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icse48619.2023.00024",
+      "retrieved_at": "2026-09-08T14:55:15Z",
+      "chars": 68024,
+      "content_sha256": "sha256:8b4d7eac2af5187346fedd4173d94a3b2630347391131908c875e9dd3a8c83ba"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2030
+      },
+      {
+        "number": "II",
+        "title": "ML MODEL OPTIMIZA TION",
+        "start": 8342
+      },
+      {
+        "number": "A",
+        "title": "ML Model Optimization and Optimization Techniques",
+        "start": 8786
+      },
+      {
+        "number": "B",
+        "title": "A General Workflow of Model Optimization",
+        "start": 10553
+      },
+      {
+        "number": "III",
+        "title": "METHODOLOGY",
+        "start": 11716
+      },
+      {
+        "number": "A",
+        "title": "Data Collection",
+        "start": 11903
+      },
+      {
+        "number": "B",
+        "title": "Issue Selection",
+        "start": 14582
+      },
+      {
+        "number": "C",
+        "title": "MOB Labeling",
+        "start": 16829
+      },
+      {
+        "number": "IV",
+        "title": "RESULTS ANDFINDINGS",
+        "start": 20975
+      },
+      {
+        "number": "A",
+        "title": "RQ1: General Characteristics of MOBs",
+        "start": 21791
+      },
+      {
+        "number": "B",
+        "title": "RQ2: Root Causes of MOBs",
+        "start": 27756
+      },
+      {
+        "number": "C",
+        "title": "RQ3: Challenges of MOB Detection",
+        "start": 38821
+      },
+      {
+        "number": "V",
+        "title": "DISCUSSION",
+        "start": 49117
+      },
+      {
+        "number": "A",
+        "title": "Implications",
+        "start": 49131
+      },
+      {
+        "number": "B",
+        "title": "Feedback from the Community",
+        "start": 50290
+      },
+      {
+        "number": "C",
+        "title": "Limitations and Threats to V alidity",
+        "start": 50646
+      },
+      {
+        "number": "D",
+        "title": "Comparison with Optimization Step in Model Compiling",
+        "start": 52296
+      },
+      {
+        "number": "VI",
+        "title": "RELA TED WORK",
+        "start": 53277
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 55674
+      },
+      {
+        "number": "T",
+        "title": "Killeen, Z. Lin, N. Gimelshein, L. Antiga et al., “Pytorch: An",
+        "start": 60807
+      },
+      {
+        "number": "G",
+        "title": "Yang, and D. Qian, “The deep learning compiler: A comprehensive",
+        "start": 67656
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "M. Namysl and I. Konya, “Efficient, lexicon-free ocr using deep learning,” in 2019 international conference on document analysis and recognition (ICDAR). IEEE, 2019, pp. 295–301.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "L. Deng, G. Hinton, and B. Kingsbury, “New types of deep neural network learning for speech recognition and related applications: an overview,” in and Signal Processing, 2013, pp. 8599–8603.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "M. Bakator and D. Radosav, “Deep learning and medical diagnosis: A review of literature,” Multimodal Technologies and Interaction, vol. 2, no. 3, p. 47, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "G. Litjens, C. I. S ´anchez, N. Timofeeva, M. Hermsen, I. Nagtegaal, I. Kovacs, C. Hulsbergen-V an De Kaa, P. Bult, B. V an Ginneken, and J. V an Der Laak, “Deep learning as a tool for increased accuracy and efficiency of histopathological diagnosis,” Scientific reports, vol. 6, no. 1, pp. 1–11, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "S. Ramos, S. Gehrig, P. Pinggera, U. Franke, and C. Rother, “Detecting unexpected obstacles for self-driving cars: Fusing deep learning and geometric modeling,” in. IEEE, 2017, pp. 1025–1032.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "R. David, J. Duke, A. Jain, V. Janapa Reddi, N. Jeffries, J. Li, N. Kreeger, I. Nappier, M. Natraj, T. Wang et al., “Tensorflow lite micro: Embedded machine learning for tinyml systems,” Proceedings of Machine Learning and Systems, vol. 3, pp. 800–811, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "X. Dai, I. Spasi ´c, B. Meyer, S. Chapman, and F. Andres, “Machine learning on mobile: An on-device inference app for skin cancer detection,” in 2019 F ourth International Conference on F og and Mobile Edge Computing (FMEC). IEEE, 2019, pp. 301–305.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "C. Sun, A. Shrivastava, S. Singh, and A. Gupta, “Revisiting unreasonable effectiveness of data in deep learning era,” in Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2017, pp. 843– 852.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Z. Chen, Y. Cao, Y. Liu, H. Wang, T. Xie, and X. Liu, “A comprehensive study on challenges in deploying deep learning based software,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 750–762.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Q. Guo, S. Chen, X. Xie, L. Ma, Q. Hu, H. Liu, Y. Liu, J. Zhao, and X. Li, “An empirical study towards characterizing deep learning development and deployment across different frameworks and platforms,” in 2019 34th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 2019, pp. 810–822.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "M. Zhu and S. Gupta, “To prune, or not to prune: exploring the efficacy of pruning for model compression,” arXiv preprint arXiv:1710.01878, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "M. H. Meng, G. Bai, S. G. Teo, and J. S. Dong, “Supervised robustnesspreserving data-free neural network pruning,” 2022. [Online]. Available: https://arxiv.org/abs/2204.00783",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "T.-J. Yang, Y .-H. Chen, and V. Sze, “Designing energy-efficient convolutional neural networks using energy-aware pruning,” in Proceedings of the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 5687–5695.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "J.-H. Luo, J. Wu, and W. Lin, “Thinet: A filter level pruning method for deep neural network compression,” in Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2017, pp. 5058– 5066.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "I. Hubara, M. Courbariaux, D. Soudry, R. El-Yaniv, and Y. Bengio, “Quantized neural networks: Training neural networks with low precision weights and activations,” The Journal of Machine Learning Research, vol. 18, no. 1, pp. 6869–6898, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Y. Zhou, S.-M. Moosavi-Dezfooli, N.-M. Cheung, and P. Frossard, “Adaptive quantization for deep neural network,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 32, no. 1, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "A. Polino, R. Pascanu, and D. Alistarh, “Model compression via distillation and quantization,” in International Conference on Learning Representations, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "M. Abadi, P. Barham, J. Chen, Z. Chen, A. Davis, J. Dean, M. Devin, S. Ghemawat, G. Irving, M. Isard etal.,“{TensorFlow }: A system for{Large-Scale }machine learning,” in 12th USENIX symposium on operating systems design and implementation (OSDI 16), 2016, pp. 265– 283.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga et al., “Pytorch: An imperative style, high-performance deep learning library,” Advances in neural information processing systems (NeurIPS), vol. 32, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "V. Le, M. Afshari, and Z. Su, “Compiler validation via equivalence modulo inputs,” ACM Sigplan Notices, vol. 49, no. 6, pp. 216–226, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "V. Le, C. Sun, and Z. Su, “Finding deep compiler bugs via guided stochastic program mutation,” ACMSIGPLAN Notices, vol. 50, no. 10, pp. 386–399, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "C. Sun, V. Le, and Z. Su, “Finding compiler bugs via live code mutation,” inProceedings of the 2016 ACMSIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications, 2016, pp. 849–863.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 24,
+      "text": "GitHub.com, “Issues · tensorflow/model-optimization,” Mar 2022. [Online]. Available: https://github.com/tensorflow/model-optimization/ issues/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "N. Humbatova, G. Jahangirova, G. Bavota, V. Riccio, A. Stocco, and P. Tonella, “Taxonomy of real faults in deep learning systems,” inProceedings of the ACM/IEEE 42nd International Conference on Software Engineering (ICSE), 2020, p. 1110–1121. 157",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "M. J. Islam, G. Nguyen, R. Pan, and H. Rajan, “A comprehensive study on deep learning bug characteristics,” in Proceedings of the 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2019, p. 510–520.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Y. Zhang, Y. Chen, S.-C. Cheung, Y. Xiong, and L. Zhang, “An empirical study on tensorflow program bugs,” in Proceedings of the 27th ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2018, pp. 129–140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "R. Zhang, W. Xiao, H. Zhang, Y. Liu, H. Lin, and M. Yang, “An empirical study on program failures of deep learning jobs,” in (ICSE). IEEE, 2020, pp. 1159–1170.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Y. Xiong, Y. Tian, Y. Liu, and S. Cheung, “Towards actionable testing of deep learning models,” SCIENCE CHINA Information Sciences, 2022. [Online]. Available: https://www.sciengine.com/SCIS/ doi/10.1007/s11432-022-3580-5",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Q. Shen, H. Ma, J. Chen, Y. Tian, S.-C. Cheung, and X. Chen, “A comprehensive study of deep learning compiler bugs,” in Proceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2021, pp. 968–980.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Z. Chen, H. Yao, Y. Lou, Y. Cao, Y. Liu, H. Wang, and X. Liu, “An empirical study on deployment faults of deep learning based mobile applications,” in Software Engineering (ICSE). IEEE, 2021, pp. 674–685.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "A. Makhshari and A. Mesbah, “Iot bugs and development challenges,” in neering (ICSE). IEEE, 2021, pp. 460–472.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "TensorFlow, “Tensorflow model optimization,” Mar 2022. [Online]. Available: https://www.tensorflow.org/model optimization/guide",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "V. Sze, Y .-H. Chen, J. Emer, A. Suleiman, and Z. Zhang, “Hardware for machine learning: Challenges and opportunities,” in 2017 IEEE Custom Integrated Circuits Conference (CICC). IEEE, 2017, pp. 1–8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "C. Wang, L. Gong, Q. Y u, X. Li, Y. Xie, and X. Zhou, “Dlau: A scalable deep learning accelerator unit on fpga,” IEEE Transactions on ComputerAided Design of Integrated Circuits and Systems, vol. 36, no. 3, pp. 513–517, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "PyTorch, “Quantization,” 2019. [Online]. Available: https://pytorch.org/ docs/stable/quantization.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "——, “Pruning tutorial,” 2021. [Online]. Available: https://pytorch.org/ tutorials/intermediate/pruning tutorial.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "J. Hale, “Deep learning framework power scores 2018,” Nov 2018. [Online]. Available: https://www.kaggle.com/discdiver/ deep-learning-framework-power-scores-2018",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Microsoft, “Cntk network optimizations,” Jan 2018. [Online]. Available: https://github.com/microsoft/CNTK/blob/v2.7/Manual/ Manual How touse network optimizations.ipynb",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "“Mxnet python api.” [Online]. Available: https://mxnet.apache.org/ versions/1.7/api/python/docs/api/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "“Eclipse deeplearning4j.” [Online]. Available: https://deeplearning4j. konduit.ai/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Keras, “Keras documentation: The keras ecosystem,” 2022. [Online]. Available: https://keras.io/getting started/ecosystem/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "K.-J. Stol, P. Ralph, and B. Fitzgerald, “Grounded theory in software engineering research: a critical review and guidelines,” in Proceedings of the 38th International Conference on Software Engineering (ICSE), 2016, pp. 120–131.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "L. Jia, H. Zhong, X. Wang, L. Huang, and X. Lu, “An empirical study on bugs inside tensorflow,” in International Conference on Database Systems for Advanced Applications. Springer, 2020, pp. 604–620.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "J. Chen, Y. Liang, Q. Shen, and J. Jiang, “Toward understanding deep learning framework bugs,” arXiv preprint arXiv:2203.04026, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "J. Dolby, A. Shinnar, A. Allain, and J. Reinen, “Ariadne: Analysis for machine learning programs,” in Proceedings of the 2nd ACMSIGPLAN International Workshop on Machine Learning and Programming Languages, ser. MAPL 2018. Association for Computing Machinery, 2018, p. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Y. Zhang, L. Ren, L. Chen, Y. Xiong, S.-C. Cheung, and T. Xie, “Detecting numerical bugs in neural network architectures,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 826–837.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "M. Yan, J. Chen, X. Zhang, L. Tan, G. Wang, and Z. Wang, “Exposing numerical bugs in deep learning via gradient back-propagation,” in Proceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2021, pp. 627–638.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "H. V. Pham, T. Lutellier, W. Qi, and L. Tan, “Cradle: cross-backend validation to detect and localize bugs in deep learning libraries,” in (ICSE). IEEE, 2019, pp. 1027–1038.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Q. Guo, X. Xie, Y. Li, X. Zhang, Y. Liu, X. Li, and C. Shen, “Audee: Automated testing for deep learning frameworks,” in 2020 35th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 2020, pp. 486–498.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "A. M. Bogar, D. M. Lyons, and D. Baird, “Lightweight callgraph construction for multilingual software analysis,” arXiv preprint arXiv:1808.01213, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Z. Wang, M. Yan, J. Chen, S. Liu, and D. Zhang, “Deep learning library testing via effective model generation,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, p. 788–799.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "M. Li, Y. Liu, X. Liu, Q. Sun, X. Y ou, H. Yang, Z. Luan, L. Gan, G. Yang, and D. Qian, “The deep learning compiler: A comprehensive survey,” IEEE Transactions on Parallel and Distributed Systems, vol. 32, no. 3, pp. 708–727, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "J. Cao, B. Chen, C. Sun, L. Hu, and X. Peng, “Characterizing performance bugs in deep learning systems,” CoRR, vol. abs/2112.01771, 2021. [Online]. Available: https://arxiv.org/abs/2112.01771 158",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 23,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[23]",
+      "technique": "norec",
+      "sentence": "Similar to the optimization task in other software that handles complex objects, such as in program compilers [20]– [22] and databases [23], ML model optimization is an errorprone process.",
+      "context_before": "ts of numerical parameters. V arious optimization techniques have been proposed so far, and they can be grouped into two main categories, i.e., pruning [11]–[14] and quantization [15]–[17]. Pruning-based approaches identify and zero out insignificant parameters to reduce the model size, and quantization-based approaches replace parameters of floating-point numbers with lower precision representations to simplify computation. These techniques have been incorporated by the popular ML frameworks, such as TensorFlow [18] and PyTorch [19], and have become the de facto pre-deployment model processors.",
+      "context_after": "It may mistakenly result in a defective model that produces different outputs than the original model, consumes excessive prediction time, or even crashes [9], [24]. To generalize, we call these bugs that appear in the optimization phase of ML frameworks model optimization bugs (MOBs). Research efforts have been made to study the reliability of ML frameworks. They mostly focus on the general program bugs in the learning stage [25]–[29], the compiling stage [30] and the deployment stage [9], [31], [32]. Nonetheless, MOBs remain largely unstudied. Researchers have simply treated the optimizatio",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 4662,
+      "cited_reference": {
+        "number": 23,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:02:36Z",
+    "is_model_written": true,
+    "summary": "An empirical study of real-world bugs in machine learning model optimization, characterising what goes wrong when a trained model is converted, quantised or otherwise optimised for deployment.",
+    "narrative": "Databases are named once, in the opening sentence, as another domain where optimisation over complex objects is error-prone -- alongside program compilers. The mention is reachable only through the citation marker and establishes an analogy rather than any relationship to SQLancer.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse48619_2023_00101.json b/_data/papers/paper_doi_10_1109_icse48619_2023_00101.json
new file mode 100644
index 0000000..b0d8ee6
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse48619_2023_00101.json
@@ -0,0 +1,992 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:26:32Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse48619.2023.00101",
+    "title": "Detecting Isolation Bugs via Transaction Oracle Construction",
+    "authors": [
+      "Wensheng Dou",
+      "Ziyu Cui",
+      "Qianwang Dai",
+      "Jiansen Song",
+      "Dong Wang",
+      "Yu Gao",
+      "Wei Wang",
+      "Jun Wei",
+      "Lei Chen",
+      "Han Wang",
+      "Hua Zhong",
+      "Tao Huang"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse48619.2023.00101",
+    "arxiv_id": null,
+    "s2_paper_id": "a9fa1a56ba118fa3b629fc6308d6b5606751f023",
+    "url": "https://doi.org/10.1109/icse48619.2023.00101",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icse48619.2023.00101",
+      "retrieved_at": "2026-09-10T15:26:32Z",
+      "chars": 75120,
+      "content_sha256": "sha256:a5b2434826fc24eed6b25f5c076328e5904bc518eb104d5dd67abef341bb26fa"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1835
+      },
+      {
+        "number": "II",
+        "title": "PRELIMINARIES",
+        "start": 10231
+      },
+      {
+        "number": "III",
+        "title": "APPROACH",
+        "start": 14485
+      },
+      {
+        "number": "A",
+        "title": "Transaction Test Protocol",
+        "start": 14786
+      },
+      {
+        "number": "B",
+        "title": "Troc Overview",
+        "start": 17196
+      },
+      {
+        "number": "C",
+        "title": "Database and SQL Statement Generation",
+        "start": 21492
+      },
+      {
+        "number": "D",
+        "title": "Transaction Test Case Generation",
+        "start": 24636
+      },
+      {
+        "number": "E",
+        "title": "Transaction Oracle Construction",
+        "start": 25767
+      },
+      {
+        "number": "F",
+        "title": "Detecting Isolation Bugs",
+        "start": 38814
+      },
+      {
+        "number": "IV",
+        "title": "EVALUA TION",
+        "start": 40784
+      },
+      {
+        "number": "A",
+        "title": "Experimental Methodology",
+        "start": 41123
+      },
+      {
+        "number": "B",
+        "title": "Overall Bug Detection Results",
+        "start": 43197
+      },
+      {
+        "number": "C",
+        "title": "New Bugs",
+        "start": 47857
+      },
+      {
+        "number": "D",
+        "title": "Other Experimental Statistics",
+        "start": 53446
+      },
+      {
+        "number": "V",
+        "title": "DISCUSSION",
+        "start": 54651
+      },
+      {
+        "number": "VI",
+        "title": "RELA TED WORK",
+        "start": 57375
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 60296
+      },
+      {
+        "number": "W",
+        "title": "Wang, and J. Wei, “Finding bugs in Gremlin-based graph database",
+        "start": 71351
+      },
+      {
+        "number": "T",
+        "title": "Huang, “Testing database systems via differential query execution,”",
+        "start": 72061
+      },
+      {
+        "number": "J",
+        "title": "Wang, and J. Li, “Sequence-oriented DBMS fuzzing,” in Proceedings",
+        "start": 72874
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "(2022) MySQL. [Online]. Available: https://www.mysql.com",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "(2022) MariaDB. [Online]. Available: https://mariadb.org",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "(2022) PostgreSQL. [Online]. Available: https://www.postgresql.org",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "(2022) TiDB, PingCAP. [Online]. Available: https://pingcap.com",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "(2022) CockroachDB. [Online]. Available: https://www.cockroachlabs. com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "D. D. Chamberlin and R. F. Boyce, “SEQUEL: A structured english query language,” in Proceedings of ACMSIGFIDET Workshop on Data Description, Access and Control (SIGFIDET), 1974, pp. 249–264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "P. A. Bernstein, V. Hadzilacos, and N. Goodman, Concurrency Control and Recovery in Database Systems. Addison-Wesley Longman Publishing Co., Inc., 1986.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "P. M. Lewis, A. Bernstein, and M. Kifer, “Databases and transaction processing: An application-oriented approach,” ACMSIGMOD Record, vol. 31, no. 1, pp. 74–75, 2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "(2022) The ANSI isolation levels. [Online]. Available: http://www. adp-gmbh.ch/ora/misc/isolation level.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "H. Berenson, P. Bernstein, J. Gray, J. Melton, E. O’Neil, and P. O’Neil, “A critique of ANSI SQL isolation levels,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 1995, pp. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "P. Bailis, A. Davidson, A. Fekete, A. Ghodsi, J. M. Hellerstein, and I. Stoica, “Highly available transactions: Virtues and limitations,” Proceedings of the VLDB Endowment (VLDB), vol. 7, no. 3, pp. 181–192, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "D. Potier and P. Leblanc, “Analysis of locking policies in database management systems,” Communications of the ACM, vol. 23, no. 10, pp. 584–593, 1980.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "S. Sippu and E. Soisalon-Soininen, Lock-Based Concurrency Control, 2014, pp. 125–158.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "D. P. Reed, “Naming and synchronization in a decentralized computer system,” Tech. Rep., 1978.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "P. A. Bernstein and N. Goodman, “Multiversion concurrency control—theory and algorithms,” ACM Transactions on Database Systems (TODS), vol. 8, no. 4, pp. 465–483, 1983.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "M. J. Carey, “Improving the performance of an optimistic concurrency control algorithm through timestamps and versions,” IEEE Transactions on Software Engineering (TSE), vol. SE-13, no. 6, pp. 746–751, 1987.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "X. Song and J. W.-S. Liu, “Performance of multiversion concurrency control algorithms in maintaining temporal consistency,” in Proceedings of Annual International Computer Software and Applications Conference (COMPSAC), 1990, pp. 132–139.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "H.-T. Kung and J. T. Robinson, “On optimistic methods for concurrency control,” ACM Transactions on Database Systems (TODS), vol. 6, no. 2, pp. 213–226, 1981.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "X. Y u, A. Pavlo, D. Sanchez, and S. Devadas, “TicToc: Time traveling optimistic concurrency control,” in Proceedings of International Conference on Management of Data (SIGMOD), 2016, pp. 1629–1642.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "(2022) Hermitage. [Online]. Available: https://github.com/ept/hermitage",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "R. Biswas and C. Enea, “On the complexity of checking transactional consistency,” in Proceedings of ACMSIGPLAN Conference on ObjectOriented Programming Systems, Languages, and Applications (OOPSLA), 2019, pp. 165:1–165:28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "C. Tan, C. Zhao, S. Mu, and M. Walfish, “Cobra: Making transactional key-value stores verifiably serializable,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 63–80.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "K. Kingsbury and P. Alvaro, “Elle: Inferring isolation anomalies from experimental observations,” Proceedings of the VLDB Endowment (VLDB), vol. 14, no. 3, pp. 268–280, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "(2022) SQLsmith. [Online]. Available: https://github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "D. R. Slutz, “Massive stochastic testing of SQL,” in Proceedings of International Conference on V ery Large Data Bases (VLDB), 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "——, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 211:1–211:30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "——, “Detecting optimization bugs in database engines via nonoptimizing reference engine construction,” in Proceedings of ACM Joint European Software Engineering Conference and Symposium on theF oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140– 1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 29,
+      "text": "(2022) Inconsistent behaviors of UPDATE under Read Uncommitted. [Online]. Available: https://bugs.mysql.com/bug.php?id=104833",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "E. F. Codd, “A relational model of data for large shared data banks,” Communications of the ACM, vol. 13, no. 6, pp. 377–387, 1970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "(2022) DB-Engines. [Online]. Available: https://db-engines.com/en/ ranking",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "(2022) GitHub. [Online]. Available: https://github.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "A. Adya, “Weak consistency: A generalized theory and optimistic implementations for distributed transactions,” Ph.D. dissertation, Massachusetts Institute of Technology, 1999.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "A. Adya, B. Liskov, and P. O’Neil, “Generalized isolation level definitions, ” in Proceedings of International Conference on Data Engineering (ICDE), 2000, pp. 67–78.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "A. Cerone, G. Bernardi, and A. Gotsman, “A framework for transactional consistency models with atomic visibility,” in Proceedings of International Conference on Concurrency Theory (CONCUR), 2015, pp. 58–71.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "P. Bailis, A. Fekete, A. Ghodsi, J. M. Hellerstein, and I. Stoica, “Scalable atomic visibility with RAMP transactions,” ACM Transactions on Database Systems (TODS), vol. 41, no. 3, pp. 15:1–15:45, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "L. Brutschy, D. Dimitrov, P. M ¨uller, and M. V echev, “Serializability for eventual consistency: Criterion, analysis, and applications,” in Proceedings of ACMSIGPLAN Symposium on Principles of Programming Languages (POPL), 2017, pp. 458–472.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "(2022) MySQL isolation. [Online]. Available: https://dev.mysql.com/ doc/refman/8.0/en/innodb-transaction-isolation-levels.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "(2022) Isolation levels in MariaDB. [Online]. Available: https: //mariadb.com/kb/en/set-transaction/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "(2022) TiDB Isolation. [Online]. Available: https://docs.pingcap.com/ tidb/v5.0/transaction-isolation-levels",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "C. Binnig, D. Kossmann, E. Lo, and M. T. ¨Ozsu, “QAGen: Generating query-aware test databases,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 2007, pp. 341– 352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "N. Bruno and S. Chaudhuri, “Flexible database generators,” in Proceedings of International Conference on V ery Large Data Bases (VLDB), 2005, pp. 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "E. F. Codd, “Relational completeness of data base sublanguages,” Research Report, 1972.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "J. Gray, P. Sundaresan, S. Englert, K. Baclawski, and P. J. Weinberger, “Quickly generating billion-record synthetic databases,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 1994, pp. 243–252.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "K. Houkjær, K. Torp, and R. Wind, “Simple and realistic data generation,” in Proceedings of International Conference on V ery Large Data Bases (VLDB), 2006, pp. 1243–1246.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "S. A. Khalek, B. Elkarablieh, Y. O. Laleye, and S. Khurshid, “Queryaware test generation using a relational constraint solver,” in Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE), 2008, pp. 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "(2022) go-randgen. [Online]. Available: https://github.com/pingcap/ go-randgen",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “SQUIRREL: Testing database management systems with language validity and coverage feedback,” in Proceedings of ACMSIGSAC Conference on Computer and Communications Security (CCS), 2020, pp. 58–71.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "(2022) SQLancer. [Online]. Available: https://www.manuelrigger.at/ dbms-bugs/",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "(2022) InnoDB transaction model. [Online]. Available: https://dev. mysql.com/doc/refman/8.0/en/innodb-transaction-model.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "J. N. Gray, R. A. Lorie, and G. R. Putzolu, “Granularity of locks in a shared data base,” in Proceedings of International Conference on V ery Large Data Bases (VLDB), 1975, pp. 428–451.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "(2022) Next-key locking in MySQL. [Online]. Available: https: //dev.mysql.com/doc/refman/8.0/en/innodb-next-key-locking.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "(2022) Execution plan information in MySQL. [Online]. Available: https://dev.mysql.com/doc/refman/8.0/en/ execution-plan-information.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "(2022) DELETE fails to delete record after blocking is released. [Online]. Available: https://jira.mariadb.org/browse/MDEV-27992",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "(2022) Weird SELECT view when a record is modified to a same value by two transactions. [Online]. Available: https: //jira.mariadb.org/browse/MDEV-26642 1134",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "(2022) Inconsistent behaviors of UPDATE under RU & RC isolation level. [Online]. Available: https://jira.mariadb.org/browse/MDEV-26643",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "(2022) Weird SELECT view when a record is modified to the same value by two transactions. [Online]. Available: https: //github.com/pingcap/tidb/issues/28212",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "(2022) UPDATE has inconsistent behaviors in a transaction. [Online]. Available: https://github.com/pingcap/tidb/issues/28092",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "(2022) UPDATE with CAST has inconsistent behaviors in transaction. [Online]. Available: https://github.com/pingcap/tidb/issues/28095",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "(2022) Isolation levles-IBM documentation. [Online]. Available: https://www.ibm.com/docs/en/db2/10.5?topic=issues-isolation-levels",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment (VLDB), vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang, and Y. Jiang, “Industry practice of coverage-guided enterprise-level DBMS fuzzing,” inProceedings of International Conference on Software Engineering: Software Engineering in Practice (ICSESEIP), 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "X. Liu, Q. Zhou, J. Arulraj, and A. Orso, “Automatic detection of performance bugs in database systems using equivalent queries,” in Proceedings of IEEE/ACMSIGSOFT International Conference on Software Engineering (ICSE), 2022, pp. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Y. Zheng, W. Dou, Y. Wang, Z. Qin, L. Tang, Y. Gao, D. Wang, W. Wang, and J. Wei, “Finding bugs in Gremlin-based graph database systems via randomized differential testing,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2022, pp. 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Z. Cui, W. Dou, Q. Dai, J. Song, W. Wang, J. Wei, and D. Y e, “Differentially testing database transactions for fun and profit,” in Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE), 2022, pp. 35:1–35:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of DBMS with coverage-based guidance,” in Proceedings of USENIX Security Symposium (USENIX Security), 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "J. Song, W. Dou, Z. Cui, Q. Dai, W. Wang, J. Wei, H. Zhong, and T. Huang, “Testing database systems via differential query execution,” inProceedings of IEEE/ACM International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "R. Y ang, Y. Zheng, L. Tang, W. Dou, W. Wang, and J. Wei, “Randomized differential testing of RDF stores,” in Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE Demo), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,”inProceedings of IEEE/ACM International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 70,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 71,
+      "text": "J. Liang, Y. Chen, Z. Wu, J. Fu, M. Wang, Y. Jiang, X. Huang, T. Chen, J. Wang, and J. Li, “Sequence-oriented DBMS fuzzing,” in Proceedings of IEEE International Conference on Data Engineering (ICDE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Z.-M. Jiang, J.-J. Bai, and Z. Su, “DynSQL: Stateful fuzzing for database management systems with complex and valid SQL query generation,” in Proceedings of USENIX Security Symposium (USENIX Security), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Z. Hua, W. Lin, L. Ren, Z. Li, L. Zhang, W. Jiao, and T. Xie, “GDsmith: Detecting bugs in Cypher graph database engines,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "W. Lin, Z. Hua, L. Zhang, and T. Xie, “GDiff: Automated differential performance testing for graph database systems,” in Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Y. Deng, P. Frankl, and Z. Chen, “Testing database transaction concurrency,” in Proceedings of IEEE International Conference on Automated Software Engineering (ASE), 2003, pp. 184–193.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "H. Luo, M. Masud, and H. Ural, “Detecting offline transaction concurrency problems,” Journal of Software, vol. 7, pp. 1855–1860, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "L. Brutschy, D. Dimitrov, P. M ¨uller, and M. V echev, “Static serializability analysis for causal consistency,” in Proceedings of SIGPLAN Conference on Programming Language Design and Implementation (PLDI) ,2018, pp. 90–104.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "K. Rahmani, K. Nagar, B. Delaware, and S. Jagannathan, “CLOTHO: Directed test generation for weakly consistent database systems,” Proceedings of the ACM on Programming Languages, vol. 3, no. OOPSLA, pp. 117:1–117:28, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Y. Gan, X. Ren, D. Ripberger, S. Blanas, and Y. Wang, “IsoDiff: Debugging anomalies caused by weak isolation,” Proceedings of the VLDB Endowment (VLDB), vol. 13, no. 12, pp. 2773–2786, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "R. Biswas, D. Kakwani, J. V edurada, C. Enea, and A. Lal, “MonkeyDB: Effectively testing correctness under weak isolation levels,” Proceedings of the ACM on Programming Languages, vol. 5, no. OOPSLA, pp. 132:1–132:27, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "C. Tang, Z. Wang, X. Zhang, Q. Y u, B. Zang, H. Guan, and H. Chen, “Ad hoc transactions in web applications: The good, the bad, and the ugly,” in Proceedings of International Conference on Management of Data (SIGMOD), 2022, pp. 4–18. 1135",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 26,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "——, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 211:1–211:30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 28,
+      "text": "——, “Detecting optimization bugs in database engines via nonoptimizing reference engine construction,” in Proceedings of ACM Joint European Software Engineering Conference and Symposium on theF oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140– 1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 49,
+      "text": "(2022) SQLancer. [Online]. Available: https://www.manuelrigger.at/ dbms-bugs/",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 69,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,”inProceedings of IEEE/ACM International Conference on Software Engineering (ICSE), 2023.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 70,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[24]–[28]",
+      "technique": "pqs",
+      "sentence": "Automatic database testing approaches [24]–[28] can support these complex features in modern DBMSs, and have beenproved as an effective technique to detect bugs in DBMSs.",
+      "context_before": "(key), respectively. Specially, the value in Elle’s data model [23] should be accumulative, e.g., AppendList. However, modern DBMSs, e.g., MySQLand TiDB, usually adopt a much more complex relational data model. They support many complex data structures (e.g.,various data types, primary keys, and indexes), and access datawith complex interface (e.g., reading / writing multiple rowsthrough SQL). This introduces unique challenges to ensurethe correctness of transaction processing mechanisms. Existingverification approaches cannot be generalized to verify trans-actions that utilize these features.",
+      "context_after": "Akey challenge for automatic database testing is to constructan effective test oracle, which can detect whether a DBMSbehaves correctly. Existing approaches can construct sometest oracles for single queries (i.e., SELECT statements). For example, SQLancer constructs the query partitioning oracle[27] and the containment oracle [26] for a single query. However, these testing approaches do not involve transactions,and cannot construct a test oracle for concurrent transactionsat a certain isolation level, thus failing to detect isolation bugs. 1123 1558-1225/23/$31.00 ©2023 IEEEDOI 10.1109/ICSE48",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 4468,
+      "cited_reference": {
+        "number": 26,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer constructs the query partitioning oracle[27] and the containment oracle [26] for a single query.",
+      "context_before": "troduces unique challenges to ensurethe correctness of transaction processing mechanisms. Existingverification approaches cannot be generalized to verify trans-actions that utilize these features. Automatic database testing approaches [24]–[28] can support these complex features in modern DBMSs, and have beenproved as an effective technique to detect bugs in DBMSs. Akey challenge for automatic database testing is to constructan effective test oracle, which can detect whether a DBMSbehaves correctly. Existing approaches can construct sometest oracles for single queries (i.e., SELECT statements).",
+      "context_after": "However, these testing approaches do not involve transactions,and cannot construct a test oracle for concurrent transactionsat a certain isolation level, thus failing to detect isolation bugs. 1123 1558-1225/23/$31.00 ©2023 IEEEDOI 10.1109/ICSE48619.2023.00101 c1 c2 10 0 10 20 10 20c1 c2 null 0 10 null 0tx1 tx2 s12: update t set c1=10; s22: update t set c2=20 where c1;s11: begin; s13: commit;s21: begin; s23: commit; Expected result Actual result Initial table޽޾޿ ߀ ߁ŝŚśŜ Şş c1 c2 10 20 10 20 10 20 Fig. 1. An illustrative example that triggers a serious isolation bug in MySQL [29]. tx1and tx2ar",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 4873,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Automatic database testing approaches like SQLancer [26]–[28] detect logic bugs for single queries (i.",
+      "context_before": "n which all data in column c1are 10, and all data in column c2are 20. However, the actual execution obtains an incorrect database state shown in the middle table, in which column c2 in the first row (with value 0) is not changed as expected. We reported this bug to MySQL developers, and they classified it asserious. However, MySQL developers have not figured out why this bug occurs yet. Revealing the above isolation bug is challenging. We lack a transaction oracle to judge whether the result of the actual transaction execution is correct. Existing approaches are ineffective in detecting this bug.",
+      "context_after": "e., SELECT statements) without considering concurrent transactions, and cannot be applied on other statements, e.g., UPDATE statements in s12ands22. Transaction verification approaches, e.g., Cobra [22] and Elle [23], use read / write operation histories based on a key-value data model, and cannot handle this table structure (without a key) and multiple rows accessed in one SQL statement (e.g., s12). To effectively detect isolation bugs in modern DBMSs, we propose a novel and general transaction testing approach,Transaction o racle c onstruction (Troc ). The core idea is to solve the oracle pro",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 7338,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Troc’s database and SQL statement generation mainly bases on SQLancer [49].",
+      "context_before": "e.g., SELECT and UPDATE). Troc treats database and SQL statement execution as a black box (e.g., without understanding SQL statements’ concrete semantics by interpreting SQL statements), and does not limit data structures and SQL statements (except JOIN, UNION, sub queries and cross-table queries that we cannot support) for target DBMSs. C. Database and SQL Statement Generation Random database [41]–[46] and SQL statement [24], [25], [47], [48] generation has been widely explored, and is not a contribution of this work. Troc can utilize existing database and SQL statement generation approaches.",
+      "context_after": "Here, we explain our database and SQL statement generation approach only for completeness. For database generation, we use the CREATE TABLE statement to create a table with at most maxCol (5 by default) columns. For each column, we randomly assign a column type and add column constraints, e.g., PRIMARY KEY ,NOT NULL and UNIQUE. We further use the CREATE INDEX statement with randomly selected columns to add indexes on the table. To populate random data into the generated database, we use at most maxInsert INSERT statements to insert at most maxInsert rows. We observe that isolation bugs can usu",
+      "section": "C Database and SQL Statement Generation",
+      "page": 5,
+      "char_offset": 21752,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", primary keys, indexes, various data types, and conditions that are supported by SQLancer [49].",
+      "context_before": "ctive DBMSs, and constructs valid SQL statements specific to respective DBMSs. We use the database schema to generate valid column references if needed. For constants used in these SQL statements, e.g., inserted values, we randomly use one of the following twostrategies. (1) We randomly generate a new value. (2) We randomly pick up a value from the corresponding column’s value cache, which stores the generated values for the column. Note that, we require that all statements in two transactions act on the same table. For now, Troc can support many complex data structures and SQL statements, e.g.",
+      "context_after": "But, Troc cannot support JOIN, UNION, sub queries, etc. Multiple tables and JOIN can introduce complex cases for transaction oracle construction, we leave them as our future work. D. Transaction Test Case Generation Transaction generation. We first randomly generate at most txSize SQL statements in Listing 1 using the method illustrated in Section III-C, and add them into a transaction. We further append a start statement BEGIN, and an ending statement COMMIT or ROLLBACK, which are randomly chosen.txSize is a configurable parameter that can be used to make a tradeoff between the transaction comp",
+      "section": "C Database and SQL Statement Generation",
+      "page": 5,
+      "char_offset": 24359,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLsmith [24], SQLancer [26]–[28], Elle [23] and Cobra [22].",
+      "context_before": "een fixed by developers. The remaining 5 isolation bugs are marked as duplicate (Duplicate). We list all our found new bugs in Table V. For these 7 new bugs, 4 bugs leave the database into incorrect states, and 3 bugs return incorrect query results. 1 isolation bug in MySQL is classified as serious, 3 isolation bugs in MariaDB are classified ascritical, 1 isolation bugs in TiDB is classified as moderate, and the remaining 2 transaction bugs are classified as moderate and minor, respectively. Comparison. We further investigate whether these 12 unique bugs can be revealed by existing approaches, e.g.",
+      "context_after": "These approaches cannot generate the bug-revealing data structures and SQL queries, or construct the corresponding transaction oracle. Thus, none of these bugs can be detected by these approaches theoretically. Specifically, existing verification approaches on isolation bugs, e.g., Elle [23] and Cobra [22], can only work on a simple key -value data model, which accesses data with read (key)andwrite (key). They cannot detect any of the 10 isolation bugs reported by Troc, which involve complex relational data models and SQL queries, e.g., writing multiple rows in the SQL statement s22 of Fig. 1.",
+      "section": "B Overall Bug Detection Results",
+      "page": 9,
+      "char_offset": 45429,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLsmith [24] and SQLancer [26]–[28], cannot generate transaction test cases, and do not have a test oracle for transaction test cases.",
+      "context_before": "evealing data structures and SQL queries, or construct the corresponding transaction oracle. Thus, none of these bugs can be detected by these approaches theoretically. Specifically, existing verification approaches on isolation bugs, e.g., Elle [23] and Cobra [22], can only work on a simple key -value data model, which accesses data with read (key)andwrite (key). They cannot detect any of the 10 isolation bugs reported by Troc, which involve complex relational data models and SQL queries, e.g., writing multiple rows in the SQL statement s22 of Fig. 1. Existing testing techniques for DBMSs, e.g.",
+      "context_after": "We also investigate whether the 6 isolation bugs (which contain complete test cases) reported by Elle [23] and Cobra [22] can potentially be detected by Troc, i.e., whether their test cases violate Troc’s oracle. Troc can detect 5 of these 6 isolation bugs. The remaining one isolation bug that Troc cannot detect requires more than two transactions to trigger, which is not supported by Troc now. Developer feedbacks. For the 7 newly detected bugs, developers have fixed 1 isolation bug in MariaDB [54]. The remaining 6 bugs have not been fixed yet. Based on the feedbacks from developers, these 6 bu",
+      "section": "B Overall Bug Detection Results",
+      "page": 9,
+      "char_offset": 46134,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[25]–[28]",
+      "technique": "pqs",
+      "sentence": ", logic bugs [25]–[28].",
+      "context_before": ", we instrument MySQL and MariaDB1, and run Troc on them for 24 hours. Troc achieves 20.6% and 25.1% line coverage for MySQL and MariaDB, respectively. The coverage appears to be low. However, this is expected, because Troc currently only focuses on transaction implementations. DBMSs also provide other features that we do not test, e.g., user management, configuration, and replication. V. DISCUSSION False positives. In Troc, we assume that the execution of a single SQL statement in DBMSs is always correct. In fact, the single SQL statement execution can be incorrect due to various reasons, e.g.",
+      "context_after": "Therefore, our transaction oracle construction may obtain wrong oracle, and potentially introduce false positives. However, we have not observed such false positives in our experiment yet. 1We do not perform coverage experiment on TiDB, since we cannot find a proper coverage measure tool for Golang and Rust, which are used in TiDB. 1132 Migrating to new DBMSs. In Troc, SQL and database generation is DBMS-related, since DBMSs usually have different dialects. However, Troc can utilize existing SQL and database generation approaches, e.g., SQLancer [49]. Therefore, Troc can be migrated to new DBM",
+      "section": "V DISCUSSION",
+      "page": 10,
+      "char_offset": 54863,
+      "cited_reference": {
+        "number": 26,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLancer [49].",
+      "context_before": "correct due to various reasons, e.g., logic bugs [25]–[28]. Therefore, our transaction oracle construction may obtain wrong oracle, and potentially introduce false positives. However, we have not observed such false positives in our experiment yet. 1We do not perform coverage experiment on TiDB, since we cannot find a proper coverage measure tool for Golang and Rust, which are used in TiDB. 1132 Migrating to new DBMSs. In Troc, SQL and database generation is DBMS-related, since DBMSs usually have different dialects. However, Troc can utilize existing SQL and database generation approaches, e.g.",
+      "context_after": "Therefore, Troc can be migrated to new DBMSs with little effort. For example, migrating Troc from MySQL to TiDB only introduces about 100 LOC changes, which are mainly used to access DBMSrelated database structures, e.g., indexes. Supporting new isolation levels. Troc has supported 4 isolation levels that are implemented in our target DBMSs. There are other isolation levels, e.g., Read Stability in DB2 [60]. To support other isolation levels, we need to adjust data visibility strategies (Table II) and lock strategies (Table III) according to new isolation levels. Troc’s view construction and",
+      "section": "V DISCUSSION",
+      "page": 11,
+      "char_offset": 55427,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker",
+      "surface": "[24]–\n[28]",
+      "technique": "pqs",
+      "sentence": "Many approaches have been proposed for DBMS testing and graph database system testing [24]– [28], [48], [61]–[74].",
+      "context_before": "ransactions under the optimistic transaction model. Supporting more concurrent transactions. We mainly focus on the scenarios where only two transactions submit their statements concurrently. Extending Troc to more than two transactions can introduce indeterminacy to transaction oracle construction. For example, given three transactions tx1,tx2 andtx3, both tx1andtx2are blocked by tx3. When tx3 is committed, which transaction is the first to be resumed is uncertain. A possible solution is to enumerate all scenarios and take all analysis results as the oracle. VI. RELA TED WORK Database testing.",
+      "context_after": "SQLsmith [24] randomly generates SQL statements to detect crash bugs in DBMSs. Squirrel [48] performs query generation guided by code coverage to test DBMSs. LEGO [71] generates SQL sequences with abundant types to improve DBMS fuzzing coverage. DynSQL [72] utilizes stateful fuzzing to test DBMSs and find deep bugs. Rigger et al. presents some approaches to construct oracles for SELECT statements, e.g., PQS [26], TLP [27] and NoREC [28]. QPG [69] utilizes query plans to guide database state mutation for detecting bugs. DQE [67] detects logic bugs by differentially executing SELECT ,UPDATE and DE",
+      "section": "VI RELA TED WORK",
+      "page": 11,
+      "char_offset": 57410,
+      "cited_reference": {
+        "number": 26,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", PQS [26], TLP [27] and NoREC [28].",
+      "context_before": "and take all analysis results as the oracle. VI. RELA TED WORK Database testing. Many approaches have been proposed for DBMS testing and graph database system testing [24]– [28], [48], [61]–[74]. SQLsmith [24] randomly generates SQL statements to detect crash bugs in DBMSs. Squirrel [48] performs query generation guided by code coverage to test DBMSs. LEGO [71] generates SQL sequences with abundant types to improve DBMS fuzzing coverage. DynSQL [72] utilizes stateful fuzzing to test DBMSs and find deep bugs. Rigger et al. presents some approaches to construct oracles for SELECT statements, e.g.",
+      "context_after": "QPG [69] utilizes query plans to guide database state mutation for detecting bugs. DQE [67] detects logic bugs by differentially executing SELECT ,UPDATE and DELETE statements in DBMSs. APOLLO [61] detects performance regression bugs by generating regression-triggering queries. Amoeba [63] detects performance bugs by generating two semantically equivalent queries and comparing their executiontime. Some works utilize differential testing to effectively test DBMSs and graph database systems [25], [64], [65], [68], [74]. However, these approaches cannot construct test oracle for detecting isolati",
+      "section": "VI RELA TED WORK",
+      "page": 11,
+      "char_offset": 57929,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG [69] utilizes query plans to guide database state mutation for detecting bugs.",
+      "context_before": "e oracle. VI. RELA TED WORK Database testing. Many approaches have been proposed for DBMS testing and graph database system testing [24]– [28], [48], [61]–[74]. SQLsmith [24] randomly generates SQL statements to detect crash bugs in DBMSs. Squirrel [48] performs query generation guided by code coverage to test DBMSs. LEGO [71] generates SQL sequences with abundant types to improve DBMS fuzzing coverage. DynSQL [72] utilizes stateful fuzzing to test DBMSs and find deep bugs. Rigger et al. presents some approaches to construct oracles for SELECT statements, e.g., PQS [26], TLP [27] and NoREC [28].",
+      "context_after": "DQE [67] detects logic bugs by differentially executing SELECT ,UPDATE and DELETE statements in DBMSs. APOLLO [61] detects performance regression bugs by generating regression-triggering queries. Amoeba [63] detects performance bugs by generating two semantically equivalent queries and comparing their executiontime. Some works utilize differential testing to effectively test DBMSs and graph database systems [25], [64], [65], [68], [74]. However, these approaches cannot construct test oracle for detecting isolation bugs in DBMSs. Transaction verification. Biswas et al. [21] utilizes an axiomatic",
+      "section": "VI RELA TED WORK",
+      "page": 11,
+      "char_offset": 57965,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/criszy/Troc",
+    "markers": [
+      "renamed_sqlancer_package",
+      "sqlancer_source_content_match"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/criszy/Troc",
+        "source_type": "github_repository",
+        "excerpt": "Artifact for \"Detecting Isolation Bugs via Transaction Oracle Construction\"\n# Troc\r\n\r\nThis is the artifact for the paper \"Detecting Isolation Bugs via Transaction Oracle Construction\". \r\nSee [paper](http://www.tcse.cn/~cuiziyu20/papers/2023-icse-troc.pdf) to learn more details.",
+        "note": "Repository names this paper.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/criszy/Troc/blob/main/src/main/java/troc/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+        "excerpt_is_verbatim": true,
+        "note": "src/main/java/troc/Randomly.java is SQLancer's Randomly.java, with the package renamed to troc (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+        "content_sha256": "sha256:e3a0454aa2987f2613383e0ec21b0ef330cb2a0d3164871b093ac235acbba125",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:00:23Z",
+    "is_model_written": true,
+    "summary": "Troc detects isolation bugs by constructing an oracle for transactions rather than for single queries. It converts a concurrent transaction schedule into an equivalent set of serial statements whose expected result is computable, so a discrepancy identifies an isolation bug. This addresses a gap the paper states directly: single-query testing approaches neither generate transaction test cases nor have an oracle for them.",
+    "narrative": "SQLancer supplies Troc's generation and defines the gap Troc fills. The paper states that Troc's database and SQL statement generation is mainly based on SQLancer, and that the schema features it works with -- primary keys, indexes, data types and conditions -- are those SQLancer supports. SQLancer is also among the tools measured against, where the paper's point is structural: it and SQLsmith cannot generate transaction test cases and have no oracle for them, so isolation bugs are outside their reach entirely. PQS, TLP, NoREC and QPG are described as the single-query oracles this work is positioned beyond.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "motivation",
+      "M4": "reuse_component",
+      "M5": "reuse_component",
+      "M6": "baseline",
+      "M7": "result_comparison",
+      "M8": "background",
+      "M9": "incidental",
+      "M10": "background",
+      "M11": "definition",
+      "M12": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "Troc’s database and SQL statement generation mainly bases on SQLancer [49].",
+            "section": "C Database and SQL Statement Generation",
+            "page": 5
+          },
+          {
+            "mention_id": "M5",
+            "sentence": ", primary keys, indexes, various data types, and conditions that are supported by SQLancer [49].",
+            "section": "C Database and SQL Statement Generation",
+            "page": 5
+          }
+        ],
+        "reasoning": "M4 states Troc's database and SQL statement generation is mainly based on SQLancer, and M5 that the supported schema features are SQLancer's. The transaction oracle on top is Troc's own.",
+        "reuse_kind": "generator"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Troc constructs an oracle for transaction schedules, which M3 and M7 present as covering what the single-query oracles cannot address rather than generalising one of them."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M6",
+          "M7"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M6",
+            "sentence": ", SQLsmith [24], SQLancer [26]–[28], Elle [23] and Cobra [22].",
+            "section": "B Overall Bug Detection Results",
+            "page": 9
+          },
+          {
+            "mention_id": "M7",
+            "sentence": ", SQLsmith [24] and SQLancer [26]–[28], cannot generate transaction test cases, and do not have a test oracle for transaction test cases.",
+            "section": "B Overall Bug Detection Results",
+            "page": 9
+          }
+        ],
+        "reasoning": "M6 names SQLancer among the tools in the overall bug detection comparison, and M7 reports the outcome: it cannot generate transaction test cases and has no oracle for them, so it finds none of these bugs."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Automatic testing approaches are called effective at detecting bugs in DBMSs, which is not a claim that SQLancer is the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse48619_2023_00173.json b/_data/papers/paper_doi_10_1109_icse48619_2023_00173.json
new file mode 100644
index 0000000..19fb928
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse48619_2023_00173.json
@@ -0,0 +1,467 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:55:28Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse48619.2023.00173",
+    "title": "Generating Test Databases for Database-Backed Applications",
+    "authors": [
+      "Cong Yan",
+      "Suman Nath",
+      "Shan Lu"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse48619.2023.00173",
+    "arxiv_id": null,
+    "s2_paper_id": "605b3d1b6cbcdaab3f2cd9ec0002ec6dd8050c72",
+    "url": "https://doi.org/10.1109/icse48619.2023.00173",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icse48619.2023.00173",
+      "retrieved_at": "2026-09-08T14:55:28Z",
+      "chars": 69081,
+      "content_sha256": "sha256:33682cd91ebabe4559b2c398bc1321b3ce824783619480869c1578576f4d8803"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1795
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 8048
+      },
+      {
+        "number": "III",
+        "title": "EXTENDED MOTIV ATION",
+        "start": 11180
+      },
+      {
+        "number": "IV",
+        "title": "WORKFLOW OFDBG RILLER",
+        "start": 15806
+      },
+      {
+        "number": "V",
+        "title": "CONSTRAINT CHECKER",
+        "start": 17096
+      },
+      {
+        "number": "VI",
+        "title": "DATABASE STATE GENERA TOR",
+        "start": 20485
+      },
+      {
+        "number": "A",
+        "title": "Candidate Generation",
+        "start": 20630
+      },
+      {
+        "number": "B",
+        "title": "Candidate filtering",
+        "start": 26577
+      },
+      {
+        "number": "C",
+        "title": "Optimizations",
+        "start": 36700
+      },
+      {
+        "number": "VII",
+        "title": "EVALUA TION",
+        "start": 37265
+      },
+      {
+        "number": "A",
+        "title": "Experiment setup",
+        "start": 37282
+      },
+      {
+        "number": "B",
+        "title": "Branch Coverage Results",
+        "start": 38084
+      },
+      {
+        "number": "C",
+        "title": "Efficiency and effectiveness Results",
+        "start": 42630
+      },
+      {
+        "number": "D",
+        "title": "Comparison with Random Fuzzing",
+        "start": 46081
+      },
+      {
+        "number": "E",
+        "title": "Analysis of Uncovered Branches",
+        "start": 50660
+      },
+      {
+        "number": "F",
+        "title": "Bugs found",
+        "start": 52433
+      },
+      {
+        "number": "VIII",
+        "title": "DISCUSSION",
+        "start": 54948
+      },
+      {
+        "number": "IX",
+        "title": "RELA TED WORK",
+        "start": 56017
+      },
+      {
+        "number": "X",
+        "title": "CONCLUSION",
+        "start": 61160
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "American Fuzz Loop. http://lcamtuf.coredump.cx/afl/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Autolab: course management service that enables auto-graded programming assignments. https://github.com/autolab/Autolab.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Chatwoot: open-source customer engagement suite, an alternative to Intercom, Zendesk, Salesforce Service Cloud etc. https://github.com/ chatwoot/chatwoot.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "CodeQL: the analysis engine used by developers to automate security checks, and by security researchers to perform variant analysis. https: //codeql.github.com/docs/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Diaspora: A privacy-aware, distributed, open source social network. https://github.com/diaspora/diaspora.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Django, a python web application framework. https://www. djangoproject.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "F orem: open source software for building communities. https://github. com/forem/forem.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Growstuf: open data project for small-scale food growers. https://github. com/Growstuff/growstuff.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Hibernate, an ORM framework for java. http://hibernate.org/orm/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Huginn: create agents that monitor and act on your behalf. https://github. com/huginn/huginn.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Lobsters: computing-focused community centered around link aggregation and discussion. https://github.com/lobsters/lobsters.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Openstreetmap: rails application powering OpenStreetMap. https:// github.com/openstreetmap/openstreetmap-website.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Pandas: read SQL query or database table into a DataFrame. https: //pandas.pydata.org/docs/reference/api/pandas.read sql.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Rails ActiveRecord callback mechanism. https://guides.rubyonrails.org/ active record callbacks.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Rails model APIs. https://devhints.io/rails-models.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Rails polymorphic association. https://guides.rubyonrails.org/ association basics.html#polymorphic-associations.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "SimpleCov: code coverage tool for ruby. https://github.com/ simplecov-ruby/simplecov.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Spree: open source headless multi-language/multi-currency/multi-store eCommerce platform. https://github.com/spree/spree.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Top errors from Ruby on Rails project. https://rollbar.com/blog/ top-10-errors-from-1000-ruby-on-rails-projects-and-how-to-avoid-them/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Tracks: web-based application to help you implement David Allen’s Getting Things Done methodology. https://github.com/TracksApp/tracks.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), Banff, Alberta, November 2020. USENIX Association. 2058",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "Pooja Agrawal, Bikash Chandra, K. V enkatesh Emani, Neha Garg, and S. Sudarshan. Test data generation for database applications. In 1621–1624, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Muhammad Ali Gulzar, Madanlal Musuvathi, and Miryung Kim. Bigtest: A symbolic execution based systematic test generation tool for apache spark. In on Software Engineering: Companion Proceedings (ICSE-Companion), pages 61–64, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Arvind Arasu, Raghav Kaushik, and Jian Li. Data generation using declarative constraints. In Proceedings of the 2011 ACMSIGMOD International Conference on Management of Data, page 685–696, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Andrea Arcuri and Juan P Galeotti. Testability transformations for existing apis. In Testing, V alidation and V erification (ICST), pages 153–163. IEEE, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "David W Binkley, Mark Harman, and Kiran Lakhotia. Flagremover: A testability transformation for transforming loop-assigned flags. ACM Transactions on Software Engineering and Methodology (TOSEM), 20(3):1–33, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M. Tamer ¨Ozsu. Qagen: Generating query-aware test databases. In Proceedings of the 2007 ACMSIGMOD International Conference on Management of Data, page 341–352, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Ankit Choudhary, Shan Lu, and Michael Pradel. In ICSE, pages 266– 277, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Mark Harman, Andr ´e Baresel, David Binkley, Robert Hierons, Lin Hu, Bogdan Korel, Phil McMinn, and Marc Roper. Testability transformation–program transformation to improve testability. In F ormal methods and testing, pages 320–344. Springer, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Mark Harman, Lin Hu, Rob Hierons, Joachim Wegener, Harmen Sthamer, Andr ´e Baresel, and Marc Roper. Testability transformation. IEEE Transactions on Software Engineering, 30(1):3–16, 2004.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Robert M Hierons, Mark Harman, and CJ Fox. Branch-coverage testability transformation for unstructured programs. The Computer Journal, 48(4):421–436, 2005.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Xuan-Bach D. Le, Corina Pasareanu, Rohan Padhye, David Lo, Willem Visser, and Koushik Sen. Saffron: Adaptive grammar-based fuzzing for worst-case analysis. SIGSOFT Softw. Eng. Notes, 44(4):14, sep 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulra, and Alessandro Orso. automatic detection of performance bugs in database systems using equivalent queries. In ICSE, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Eric Lo, Nick Cheng, and Wing-Kai Hon. Generating databases for query workloads. Proc. VLDB Endow., 3(1–2):848–859, sep 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Magnus Madsen. Static analysis of dynamic languages.: http://pure. au. dk/ws/files/85299449/Thesis. pdf, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Carlos Pacheco, Shuvendu K. Lahiri, Michael D. Ernst, and Thomas Ball. Feedback-directed random test generation. In ICSE, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Rohan Padhye, Caroline Lemieux, and Koushik Sen. Jqf: Coverageguided property-based testing in java. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis, page 398–401, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Rohan Padhye, Caroline Lemieux, and Koushik Sen. Jqf: Coverageguided property-based testing in java. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis, page 398–401, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Mike Papadakis, and Yves Le Traon. Semantic Fuzzing with Zest. 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Kai Pan, Xintao Wu, and Tao Xie. Automatic test generation for mutation testing on database applications. In ASE, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Kai Pan, Xintao Wu, and Tao Xie. Guided test generation for database applications via synthesized database interactions. ACM Trans. Softw. Eng. Methodol., 23(2), apr 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA), nov 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "Max Sch ¨afer and Oege de Moor. Type inference for datalog with complex type hierarchies. In Proceedings of the 37th Annual ACMSIGPLAN-SIGACT Symposium on Principles of Programming Languages (POPL), page 145–156, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Thodoris Sotiropoulos and Benjamin Livshits. Static analysis for asynchronous javascript programs. arXiv preprint arXiv:1901.03575, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. Superion: Grammaraware greybox fuzzing. In Proceedings of the 41st International Conference on Software Engineering, ICSE ’19, page 724–735. IEEE Press, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Junwen Yang, Utsav Sethi, Cong Yan, Alvin Cheung, and Shan Lu. Managing data constraints in database-backed web applications. In ICSE, page 1098–1109, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Qian Zhang, Jiyuan Wang, Muhammad Ali Gulzar, Rohan Padhye, and Miryung Kim. Bigfuzz: Efficient fuzz testing for data analytics using framework abstraction. In ASE, 2020. 2059",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 21,
+      "text": "Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), Banff, Alberta, November 2020. USENIX Association. 2058",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA), nov 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [21], [42] includes a series of techniques, including query partitioning and pivot query synthesis, to generate SQL queries paired with certain properties of query results.",
+      "context_before": "applications. Another line of work in testing big data applications targets performance testing of query workload. QAGen [27] takes in database schema, cardinality and data distribution constraints, MyBenchmark [34] takes in parameterized queries with derives constraints, and Arasu et. al. [24] proposed a language to specify data distribution constraints. They output large database states satisfying these constraints. Our work focuses on integrity constraints instead of distribution constraints. Database testing. A lot of research work has explored generating queries to test database systems.",
+      "context_after": "These queries can find logical bugs of database systems. AMOEBA [33] detects query performance bug by exploring equivalent query rewrites and check whether rewritten queries have similar performance. These techniques tackle the validity challenge by designing program mutations to produce queries with desired properties such semantic equivalence to a given query. In contrast, DBGRILLER ’s mutation produces states that are likely-valid only empirically because it would be impossible to generate guaranteed valid states due to the complexity of applicationspecific data constraints. Random testing a",
+      "section": "IX RELA TED WORK",
+      "page": 11,
+      "char_offset": 58680,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:02:36Z",
+    "is_model_written": true,
+    "summary": "This work generates test databases for applications backed by a database, producing data that satisfies the schema and the application's own query constraints so that tests exercise realistic states rather than empty or arbitrary tables.",
+    "narrative": "SQLancer is described in related work as including a series of techniques -- query partitioning and pivot query synthesis are named -- that generate SQL queries paired with known properties of their results. The distinction implied is one of purpose: those techniques generate queries whose results are predictable in order to test the DBMS, while this work generates data in order to test an application.",
+    "roles": {
+      "M1": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse48619_2023_00175.json b/_data/papers/paper_doi_10_1109_icse48619_2023_00175.json
new file mode 100644
index 0000000..1a5606c
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse48619_2023_00175.json
@@ -0,0 +1,1124 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:26:49Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse48619.2023.00175",
+    "title": "Testing Database Systems via Differential Query Execution",
+    "authors": [
+      "Jiansen Song",
+      "Wensheng Dou",
+      "Ziyu Cui",
+      "Qianwang Dai",
+      "Wei Wang",
+      "Jun Wei",
+      "Hua Zhong",
+      "Tao Huang"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse48619.2023.00175",
+    "arxiv_id": null,
+    "s2_paper_id": "5724e138fc08329e0baa08d2090d9424df1cb945",
+    "url": "https://doi.org/10.1109/icse48619.2023.00175",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icse48619.2023.00175",
+      "retrieved_at": "2026-09-10T15:26:49Z",
+      "chars": 64906,
+      "content_sha256": "sha256:f3421bc3c73cfe97749e2af6679eb6f80bf2ee0c44a6c9c46611c34a8ce6962b"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1885
+      },
+      {
+        "number": "II",
+        "title": "PRELIMINARIES",
+        "start": 9207
+      },
+      {
+        "number": "A",
+        "title": "Database Management Systems and SQL",
+        "start": 9392
+      },
+      {
+        "number": "B",
+        "title": "Target DBMSs",
+        "start": 10567
+      },
+      {
+        "number": "C",
+        "title": "Query Execution Strategy",
+        "start": 11546
+      },
+      {
+        "number": "III",
+        "title": "APPROACH",
+        "start": 14626
+      },
+      {
+        "number": "A",
+        "title": "DQE Overview",
+        "start": 14975
+      },
+      {
+        "number": "B",
+        "title": "Database Generation",
+        "start": 17028
+      },
+      {
+        "number": "C",
+        "title": "Query Triple Generation",
+        "start": 19422
+      },
+      {
+        "number": "D",
+        "title": "Obtaining Execution Results",
+        "start": 22300
+      },
+      {
+        "number": "E",
+        "title": "Comparing Execution Results",
+        "start": 27425
+      },
+      {
+        "number": "IV",
+        "title": "EVALUATION",
+        "start": 30420
+      },
+      {
+        "number": "A",
+        "title": "Experimental Methodology",
+        "start": 31311
+      },
+      {
+        "number": "B",
+        "title": "Overall Detection Results",
+        "start": 33467
+      },
+      {
+        "number": "C",
+        "title": "Comparing with Existing Approaches",
+        "start": 36779
+      },
+      {
+        "number": "D",
+        "title": "Other Experimental Statistics",
+        "start": 38210
+      },
+      {
+        "number": "E",
+        "title": "Selected Bugs",
+        "start": 41652
+      },
+      {
+        "number": "F",
+        "title": "Not A Bug",
+        "start": 48924
+      },
+      {
+        "number": "V",
+        "title": "DISCUSSION",
+        "start": 50741
+      },
+      {
+        "number": "VI",
+        "title": "RELATED WORK",
+        "start": 52015
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 54664
+      },
+      {
+        "number": "Y",
+        "title": "Zhou, M. Huang, W. Wei, C. Liu, J. Zhang, J. Li, X. Wu, L. Song,",
+        "start": 55717
+      },
+      {
+        "number": "R",
+        "title": "Sun, S. Yu, L. Zhao, N. Cameron, L. Pei, and X. Tang, “TiDB:",
+        "start": 55785
+      },
+      {
+        "number": "O",
+        "title": "Papaemmanouil, and N. Tatbul, “Neo: A learned query optimizer,”",
+        "start": 58213
+      },
+      {
+        "number": "W",
+        "title": "Wang, and J. Wei, “Finding bugs in Gremlin-based graph database",
+        "start": 61844
+      },
+      {
+        "number": "L",
+        "title": "Chen, H. Wang, H. Zhong, and T. Huang, “Detecting isolation bugs",
+        "start": 64713
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“MySQL homepage,” https://www.mysql.com, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“MariaDB homepage,” https://mariadb.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "D. Huang, Q. Liu, Q. Cui, Z. Fang, X. Ma, F. Xu, L. Shen, L. Tang, Y. Zhou, M. Huang, W. Wei, C. Liu, J. Zhang, J. Li, X. Wu, L. Song, R. Sun, S. Yu, L. Zhao, N. Cameron, L. Pei, and X. Tang, “TiDB: A Raft-based HTAP database,” Proceedings of the VLDB Endowment (VLDB), vol. 13, no. 12, pp. 3072–3084, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "R. Taft, I. Sharif, A. Matei, N. VanBenschoten, J. Lewis, T. Grieger, K. Niemi, A. Woods, A. Birzin, R. Poss, P. Bardea, A. Ranade, B. Darnell, B. Gruneir, J. Jaffray, L. Zhang, and P. Mattis, “CockroachDB: The resilient Geo-distributed SQL database,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 2020, pp. 1493–1509.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "“SQLite homepage,” https://www.sqlite.org/index.html, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "D. D. Chamberlin and R. F. Boyce, “SEQUEL: A structured english query language,” in Proceedings of ACMSIGFIDET Workshop on Data Description, Access and Control, 1974, pp. 249–264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "“MySQL customers by industry,” https://www.mysql.com, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "D. R. Slutz, “Massive stochastic testing of SQL,” in Proceedings of International Conference on V ery Large Data Bases (VLDB), 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 10,
+      "text": "——, “Detecting optimization bugs in database engines via nonoptimizing reference engine construction,” in Proceedings of ACM Joint European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140– 1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 11,
+      "text": "——, “Finding bugs in database systems via query partitioning,” in Proceedings of ACMSIGPLAN Conference on Object-Oriented Programming Systems, Languages, and Applications (OOPSLA), vol. 4, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 12,
+      "text": "E. F. Codd, “A relational model of data for large shared data banks,” Communications of the ACM, vol. 13, no. 6, pp. 377–387, 1970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "B. Ding, S. Das, W. Wu, S. Chaudhuri, and V. Narasayya, “Plan Stitch: Harnessing the best of many plans,” Proceedings of the VLDB Endowment (VLDB), vol. 11, no. 10, pp. 1123–1136, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "T. Neumann and B. Radke, “Adaptive optimization of very large join queries,” in Proceedings of International Conference on Management of Data (SIGMOD), 2018, pp. 677–692.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "C. Wu, A. Jindal, S. Amizadeh, H. Patel, W. Le, S. Qiao, and S. Rao, “Towards a learning optimizer for shared clouds,” Proceedings of the VLDB Endowment (VLDB), vol. 12, no. 3, pp. 210–222, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "R. Marcus, P. Negi, H. Mao, C. Zhang, M. Alizadeh, T. Kraska, O. Papaemmanouil, and N. Tatbul, “Neo: A learned query optimizer,” Proceedings of the VLDB Endowment (VLDB), vol. 12, no. 11, pp. 1705– 1718, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "“DB-Engines ranking,” https://db-engines.com/en/ranking, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "“Most widely deployed and used database engine,” https://www.sqlite. org/mostdeployed.html, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "“Database topic in GitHub,” https://github.com/topics/database, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "“Unexpected delete when data truncation,” https://jira.mariadb.org/ browse/MDEV-27885, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "A. Neufeld, G. Moerkotte, and P. C. Lockemann, “Generating consistent test data: Restricting the search space by a generator formula,” Proceedings of the VLDB Endowment (VLDB), vol. 2, no. 2, pp. 173–214, 1993.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "J. Gray, P. Sundaresan, S. Englert, K. Baclawski, and P. J. Weinberger, “Quickly generating billion-record synthetic databases,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 1994, pp. 243–252.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "N. Bruno and S. Chaudhuri, “Flexible database generators,” in Proceedings of International Conference on V ery Large Data Bases (VLDB), 2005, pp. 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "K. Houkjær, K. Torp, and R. Wind, “Simple and realistic data generation,” in Proceedings of International Conference on V ery Large Data Bases (VLDB), 2006, pp. 1243–1246.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "C. Binnig, D. Kossmann, E. Lo, and M. T. ¨Ozsu, “QAGen: Generating query-aware test databases,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 2007, pp. 341– 352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "S. Abdul Khalek, B. Elkarablieh, Y. O. Laleye, and S. Khurshid, “Queryaware test generation using a relational constraint solver,” in Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE), 2008, pp. 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "“SQLancer homepage,” https://github.com/sqlancer/sqlancer, 2022.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "“SHOWWARNINGS statement,” https://dev.mysql.com/doc/refman/8. 0/en/show-warnings.html., 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "“SQLsmith,” https://github.com/anse1/sqlsmith, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment (VLDB), vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "X. Liu, Q. Zhou, J. Arulrai, and A. Orso, “Automatic detection of performance bugs in database systems using equivalent queries,” in Proceedings of International Conference on Software Engineering (ICSE), 2022, pp. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "B. Ghit, N. Poggi, J. Rosen, R. Xin, and P. Boncz, “SparkFuzz: Searching correctness regressions in modern query engines,” in Proceedings of the Workshop on Testing Database Systems (DBTest), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "A. Andoni, D. Daniliuc, S. Khurshid, and D. Marinov, “Evaluating the “small scope hypothesis”,” in Proceedings of ACM Symposium on the Principles of Programming Languages (POPL), vol. 2, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "“Type conversion in expression evaluation,” https://dev.mysql.com/doc/ refman/5.7/en/type-conversion.html, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "“Neo4j homepage,” https://neo4j.com/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "“Azure cosmos DB,” https://azure.microsoft.com/en-us/products/ cosmos-db/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "“TigerGraph,” https://www.tigergraph.com/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "“Redis homepage,” https://redis.io/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "“Amazon DynamoDB,” https://aws.amazon.com/cn/dynamodb/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "“Hazelcast,” https://hazelcast.com/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "“MongoDB,” https://www.mongodb.com/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "“Apache CouchDB,” https://couchdb.apache.org/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "“Datastore,” https://cloud.google.com/datastore, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "W. M. McKeeman, “Differential testing for software,” DIGITAL TECHNICAL JOURNAL, vol. 10, pp. 100–107, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Y. Zheng, W. Dou, Y. Wang, Z. Qin, L. Tang, Y. Gao, D. Wang, W. Wang, and J. Wei, “Finding bugs in Gremlin-based graph database systems via randomized differential testing,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2022, pp. 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Z. Cui, W. Dou, Q. Dai, J. Song, W. Wang, J. Wei, and D. Ye, “Differentially testing database transactions for fun and profit,” in Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE), 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammar-free DBMS fuzzing,” in Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "W. Lin, Z. Hua, L. Zhang, and T. Xie, “GDiff: Automated differential performance testing for graph database systems,” in Proceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "R. Yang, Y. Zheng, L. Tang, W. Dou, W. Wang, and J. Wei, “Randomized differential testing of RDF stores,” in Proceedings of International Conference on Software Engineering (ICSE Demo), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” inProceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 51,
+      "text": "Z. Jiang, J. Bai, and Z. Su, “DynSQL: Stateful fuzzing for database management systems with complex and valid SQL query generation,” in Proceedings of USENIX Security Symposium (USENIX Security), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Z. Hua, W. Lin, L. Ren, Z. Li, L. Zhang, W. Jiao, and T. Xie, “GDsmith: Detecting bugs in Cypher graph database engines,” 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 54,
+      "text": "“go-randgen,” https://github.com/pingcap/go-randgen, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of DBMS with Coverage-based guidance,” in Proceedings of USENIX Security Symposium (USENIX Security), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “SQUIRREL: Testing database management systems with language validity and coverage feedback,” in Proceedings of ACMSIGSAC Conference on Computer and Communications Security (CCS), 2020, pp. 58–71. 2083",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang, and Y. Jiang, “Industry practice of Coverage-guided enterprise-level DBMS fuzzing,” inProceedings of IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP), 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "X. Yang, Y. Chen, E. Eide, and J. Regehr, “Finding and understanding bugs in C compilers,” in Proceedings of ACMSIGPLAN Conference onProgramming Language Design and Implementation (PLDI), 2011, pp. 283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "“Alloy,” https://alloytools.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "W. Dou, Z. Cui, Q. Dai, J. Song, D. Wang, Y. Gao, W. Wang, J. Wei, L. Chen, H. Wang, H. Zhong, and T. Huang, “Detecting isolation bugs via transaction oracle construction,” in Proceedings of International Conference on Software Engineering (ICSE), 2023. 2084",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 9,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 10,
+      "text": "——, “Detecting optimization bugs in database engines via nonoptimizing reference engine construction,” in Proceedings of ACM Joint European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140– 1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 11,
+      "text": "——, “Finding bugs in database systems via query partitioning,” in Proceedings of ACMSIGPLAN Conference on Object-Oriented Programming Systems, Languages, and Applications (OOPSLA), vol. 4, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "“SQLancer homepage,” https://github.com/sqlancer/sqlancer, 2022.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 50,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” inProceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 53,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[8]–[11]",
+      "technique": "pqs",
+      "sentence": "Recently, researchers have proposed some approaches to detect logic bugs in DBMSs [8]–[11].",
+      "context_before": "lational DBMSs, e.g., MySQL [1], MariaDB [2], TiDB [3],CockroachDB [4] and SQLite [5], adopt Structured QueryLanguage (SQL) [6] as their standard query language, andhave become an indispensable component in many business-critical applications [7]. DBMSs suffer from various bugs, e.g., crashes and logic bugs. Specially, logic bugs can cause a DBMS to returnincorrect results for SELECT queries, or generate incorrect database states for UPDATE and DELETE queries. Such logic bugs do not crash the DBMS, and can easily go unnoticed bydevelopers. In this work, we focus on detecting logic bugs in DBMSs.",
+      "context_after": "RAGS [8] feeds a SELECT query into multiple DBMSs and observes discrepancies in theirquery results. PQS [9] generates SELECT queries that fetch Wensheng Dou and Hua Zhong are the corresponding authors.a pivot row, and checks whether the target DBMS fails tofetch the pivot row. NoREC [10] rewrites a SELECT query as another equivalent one that cannot be optimized by the DBMS, and then detects difference in their query results.TLP [11] decomposes a SELECT query into three partitioning queries, and merges these partitioning queries’ results into acombined result, which is expected to be the same a",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2609,
+      "cited_reference": {
+        "number": 9,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [9] generates SELECT queries that fetch Wensheng Dou and Hua Zhong are the corresponding authors.",
+      "context_before": "e component in many business-critical applications [7]. DBMSs suffer from various bugs, e.g., crashes and logic bugs. Specially, logic bugs can cause a DBMS to returnincorrect results for SELECT queries, or generate incorrect database states for UPDATE and DELETE queries. Such logic bugs do not crash the DBMS, and can easily go unnoticed bydevelopers. In this work, we focus on detecting logic bugs in DBMSs. Recently, researchers have proposed some approaches to detect logic bugs in DBMSs [8]–[11]. RAGS [8] feeds a SELECT query into multiple DBMSs and observes discrepancies in theirquery results.",
+      "context_after": "a pivot row, and checks whether the target DBMS fails tofetch the pivot row. NoREC [10] rewrites a SELECT query as another equivalent one that cannot be optimized by the DBMS, and then detects difference in their query results.TLP [11] decomposes a SELECT query into three partitioning queries, and merges these partitioning queries’ results into acombined result, which is expected to be the same as theoriginal query’s result. However, all these approaches mainlyfocus on detecting logic bugs in SELECT queries. While the logic bugs in UPDATE and DELETE queries have not been tackled yet, even though",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2801,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [10] rewrites a SELECT query as another equivalent one that cannot be optimized by the DBMS, and then detects difference in their query results.",
+      "context_before": "sults for SELECT queries, or generate incorrect database states for UPDATE and DELETE queries. Such logic bugs do not crash the DBMS, and can easily go unnoticed bydevelopers. In this work, we focus on detecting logic bugs in DBMSs. Recently, researchers have proposed some approaches to detect logic bugs in DBMSs [8]–[11]. RAGS [8] feeds a SELECT query into multiple DBMSs and observes discrepancies in theirquery results. PQS [9] generates SELECT queries that fetch Wensheng Dou and Hua Zhong are the corresponding authors.a pivot row, and checks whether the target DBMS fails tofetch the pivot row.",
+      "context_after": "TLP [11] decomposes a SELECT query into three partitioning queries, and merges these partitioning queries’ results into acombined result, which is expected to be the same as theoriginal query’s result. However, all these approaches mainlyfocus on detecting logic bugs in SELECT queries. While the logic bugs in UPDATE and DELETE queries have not been tackled yet, even though they can cause severer consequences,e.g., incorrect database states. Logic bugs in DBMSs, especially those in UPDATE and DELETE queries, are difficult to detect automatically. A key challenge to detect logic bugs is to constru",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2979,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [11] decomposes a SELECT query into three partitioning queries, and merges these partitioning queries’ results into acombined result, which is expected to be the same as theoriginal query’s result.",
+      "context_before": "unnoticed bydevelopers. In this work, we focus on detecting logic bugs in DBMSs. Recently, researchers have proposed some approaches to detect logic bugs in DBMSs [8]–[11]. RAGS [8] feeds a SELECT query into multiple DBMSs and observes discrepancies in theirquery results. PQS [9] generates SELECT queries that fetch Wensheng Dou and Hua Zhong are the corresponding authors.a pivot row, and checks whether the target DBMS fails tofetch the pivot row. NoREC [10] rewrites a SELECT query as another equivalent one that cannot be optimized by the DBMS, and then detects difference in their query results.",
+      "context_after": "However, all these approaches mainlyfocus on detecting logic bugs in SELECT queries. While the logic bugs in UPDATE and DELETE queries have not been tackled yet, even though they can cause severer consequences,e.g., incorrect database states. Logic bugs in DBMSs, especially those in UPDATE and DELETE queries, are difficult to detect automatically. A key challenge to detect logic bugs is to construct an effective testoracle to determine whether a DBMS behaves correctly fora given query. Existing approaches to construct oracles for SELECT queries, e.g., PQS [9], NoREC [10] and TLP [11], cannot be",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3129,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", PQS [9], NoREC [10] and TLP [11], cannot be adopted on UPDATE and DELETE queries.",
+      "context_before": "to be the same as theoriginal query’s result. However, all these approaches mainlyfocus on detecting logic bugs in SELECT queries. While the logic bugs in UPDATE and DELETE queries have not been tackled yet, even though they can cause severer consequences,e.g., incorrect database states. Logic bugs in DBMSs, especially those in UPDATE and DELETE queries, are difficult to detect automatically. A key challenge to detect logic bugs is to construct an effective testoracle to determine whether a DBMS behaves correctly fora given query. Existing approaches to construct oracles for SELECT queries, e.g.",
+      "context_after": "In DBMSs, SELECT, UPDATE and DELETE queries utilize predicates (i.e., WHERE clauses) to specify which rows to retrieve, update or delete, respectively. If they use the samepredicate ϕ, they should access the same rows in a database. Ideally, DBMSs can adopt the same implementations for pred-icate evaluation in SELECT, UPDATE and DELETE queries. However, a DBMS usually adopts different implementationsfor predicate evaluation in SELECT, UPDATE and DELETE queries due to various optimization choices 1. Inconsistent implementations for predicate evaluation among these queriescan cause SELECT, UPDATE",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3884,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", PQS [9], NoREC [10] and TLP [11].",
+      "context_before": "med and fixed it. Existing approaches cannot detect this bug, because this bug occurs in the UPDATE and DELETE queries. To evaluate DQE’s effectiveness and generality, we implement DQE and perform experiments on five widely-used and production-level DBMSs, i.e., MySQL [1], MariaDB [2], TiDB [3], CockroachDB [4] and SQLite [5]. In total, we have detected 50 unique bugs among these DBMSs, 41 of which have been confirmed as new bugs, and 11 bugs have been fixed. Among the 41 confirmed bugs, 20 bugs occur in UPDATE and DELETE queries. None of our detected bugs can be detected by existing approaches, e.g.",
+      "context_after": "Our experimental results indicate that DQE is effective in detecting logic bugs in SELECT ,UPDATE and DELETE queries in DBMSs. We have made DQE publicly available at https://github.com/tcse-iscas/dqetool. Although we have detected many bugs in SELECT, UPDATE and DELETE queries in our target DBMSs, DQE still has some limitations. First, DQE suffers from the same issue as differential testing, in which DQE fails to detect the same bug occurring in all the three SELECT ,UPDATE and DELETE queries. Second, DQE only supports common operations and functions in SELECT ,UPDATE and DELETE queries, e.g., J",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 7302,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Our database generation is mainly adopted from SQLancer [27].",
+      "context_before": "l,tudenotes t1’s new table state after executing Qup, and tddenotes t1’s new table state after executing Qdel. The rest of this section is organized as follows. Section III-B describes our database generation. Section III-C describes our strategies to generate SQL queries. Section III-D shows how we obtain a query triple’s execution results. Section III-E shows how we detect logic bugs by comparing the execution results of the three queries in a query triple. B. Database Generation Database generation has been widely explored by existing works [21]–[26], and is not a contribution of this work.",
+      "context_after": "We present our database generation only for completeness. We first use the CREATE TABLE command to create at mostmaxTable tables. Each table contains at most maxCol columns. We assign each column with a random column type, e.g., INT or TEXT, and some column constraints, e.g., PRIMARY KEY and UNIQUE. We then populate random data into each table by executing the INSERT command. Each table contains at most maxInsert rows of data. We further execute at most maxAlter ALTER TABLE and CREATE INDEX commands to modify each initial table, e.g., adding new columns or building indexes on existing columns. Mo",
+      "section": "B Database Generation",
+      "page": 4,
+      "char_offset": 17164,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "EVALUATION We implement DQE based on SQLancer [27], which is implemented in Java.",
+      "context_before": "rror, QupandQdelshould raise the same error. In this case, row upandrow delshould be empty. However, row selmay not be empty, as discussed in Section II-C. 2077 TABLE IIIBUGS REPORTED BYDQE Bug Status Triggering Query DBMS Submitted Confirmed Fixed Duplicate Not a bug SELECT UPDATE DELETE MySQL 7 1 1 0 6 0 1 1 MariaDB 4 2 0 0 0 0 0 2 TiDB 37 37 10 0 0 20 17 17 CockroachDB 1 0 0 1 0 0 0 0 SQLite 1 1 0 0 0 1 0 0 Total 50 41 11 1 6 21 18 20 •IfQseldoes not raise a warning or an error, Qupand Qdelshould not raise a warning or an error. In this case, row sel,row upandrow delshould be the same. IV.",
+      "context_after": "We make the following improvements to apply DQE. First, we add UPDATE and DELETE query generation in our target DBMSs, e.g., MySQL, TiDB and SQLite. Second, DQE requires to execute the three queries in a query triple on the same database state. In MySQL, MariaDB and TiDB, we use ROLLBACK transactions to roll back all changes made by UPDATE and DELETE queries. In CockroachDB and SQLite, we record the table content before query execution and refill the table with the same content after query execution. In total, we write about 2,600 lines of code to implement DQE on five target DBMSs. We evaluate t",
+      "section": "IV EVALUATION",
+      "page": 7,
+      "char_offset": 30423,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", PQS [9], NoREC [10] and TLP [11]) that aim to detect logic bugs in DBMSs.",
+      "context_before": "rrect database states, 34 queries occur in TiDB and 2 queries occur in MySQL. All 5 queries that lead to duplicate warnings occur in TiDB. Among the 8 queries that lead to unexpected warnings, 6 queries occur in TiDB and 2 queries occur in MariaDB. One query that leads to unexpected errors occurs in SQLite. All 6 queries that lead to incorrect warning messages occur in TiDB. The remaining 3 queries that lead to the execution failures of the SHOWWARNINGS command occur in TiDB. C. Comparing with Existing Approaches To answer RQ2, we perform a qualitative comparison with existing approaches (i.e.",
+      "context_after": "These three approaches construct oracles to detect logic bugs in single SELECT queries. Thus, they cannot detect the 20 logic bugs in UPDATE and DELETE queries. Moreover, these three approaches do not consider the normal errors that can be unexpectedly raised by SELECT queries as logic bugs, e.g., the warnings in Listing 3. Unlike crashes caught by these approaches, these normal errors do not crash the DBMS and need a test oracle to validate their correctness. Thus, they cannot detect 18 logic bugs related to this kind of errors TABLE VCOVERAGE INFORMATION Tool MySQL MariaDB PQS 19 NoREC-18 TLP",
+      "section": "C Comparing with Existing Approaches",
+      "page": 8,
+      "char_offset": 36898,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Thus, they cannot detect 18 logic bugs related to this kind of errors TABLE VCOVERAGE INFORMATION Tool MySQL MariaDB PQS 19 NoREC-18 TLP 18 DQE 15 21 in SELECT queries.",
+      "context_before": "erform a qualitative comparison with existing approaches (i.e., PQS [9], NoREC [10] and TLP [11]) that aim to detect logic bugs in DBMSs. These three approaches construct oracles to detect logic bugs in single SELECT queries. Thus, they cannot detect the 20 logic bugs in UPDATE and DELETE queries. Moreover, these three approaches do not consider the normal errors that can be unexpectedly raised by SELECT queries as logic bugs, e.g., the warnings in Listing 3. Unlike crashes caught by these approaches, these normal errors do not crash the DBMS and need a test oracle to validate their correctness.",
+      "context_after": "We further analyze the triggering test cases and bug consequences of the remaining 3 logic bugs in SELECT queries. We find that, none of these 3 bugs can be triggered or captured by the oracles in these approaches. Therefore, all our reported bugs cannot be detected by these approaches theoretically. Other DBMS testing approaches, e.g., SQLsmith [29], APOLLO [30], AMOEBA [31], RAGS [8] and SparkFuzz [32], cannot construct oracles to detect logic bugs, or cannot detect logic bugs in a single DBMS because differential testing needs multiple DBMSs. Therefore, we do not compare DQE with these appro",
+      "section": "C Comparing with Existing Approaches",
+      "page": 8,
+      "char_offset": 37436,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[9]–[11]",
+      "technique": "pqs",
+      "sentence": "Existing works [9]–[11] also face the same problem.",
+      "context_before": "ing approaches, e.g., SQLsmith [29], APOLLO [30], AMOEBA [31], RAGS [8] and SparkFuzz [32], cannot construct oracles to detect logic bugs, or cannot detect logic bugs in a single DBMS because differential testing needs multiple DBMSs. Therefore, we do not compare DQE with these approaches. D. Other Experimental Statistics Test efficiency. During testing, before a bug we detect is fixed by the DBMS developers, DQE will generate many test cases that trigger the same bug. In total, DQE reports 122 bugs. After filtering out duplicate bugs, we obtain 50 unique bugs. The duplicate rate is 41% (50/122).",
+      "context_after": "There is currently no practical way to automatically filter out duplicate test cases for DBMSs. For discovering these 50 unique bugs, we generate 1,776,124,512 query triples. Query generation efficiency. We measure the query generation efficiency in DQE during testing. In this experiment, we count every queries generated including those that create the database and query triples. In DQE, we generate syntactically valid queries based on Abstract Syntax Trees (ASTs) of SQL. However, SQL in different DBMSs should obey many semantic constraints, which can cause DQE to generate semantically invalid q",
+      "section": "D Other Experimental Statistics",
+      "page": 8,
+      "char_offset": 38519,
+      "cited_reference": {
+        "number": 9,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", PQS [9], NoREC [10] and TLP [11].",
+      "context_before": "cate value into a UNIQUE column. Such semantic errors can lower our success rate of query generation. In MySQL, DQE generates 2,885 queries per second with a success rate of 88%. In MariaDB, DQE generates 3,344 queries per second with a success rate of 87%. In TiDB, DQE generates 1,566 queries per second with a success rate of 89%. In CockroachDB, DQE generates 243 queries per second with a success rate of 72%. In SQLite, DQE generates 12,313 queries per second with a success rate of 97%. Coverage. To demonstrate the sufficiency of our testing, we compare code coverage with existing works, i.e.",
+      "context_after": "We run each tool with the same experimental setting for 24 hours on MySQL and MariaDB4. 4We have not found a suitable way to perform code coverage measurements in TiDB, SQLite and CockroachDB. 2079 Table V shows our experiment results. PQS achieves 19% line coverage in MySQL. NoREC achieves 18% line coverage in MariaDB. TLP achieves 18% line coverage in MySQL. DQE achieves 15% line coverage in MySQL and 21% line coverage in MariaDB. We can see that DQE obtains similar coverage with other works. This is reasonable, since DQE, PQS, NoREC and TLP are all built on SQLancer, which share the simi",
+      "section": "D Other Experimental Statistics",
+      "page": 8,
+      "char_offset": 39846,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS achieves 19% line coverage in MySQL.",
+      "context_before": "nerates 1,566 queries per second with a success rate of 89%. In CockroachDB, DQE generates 243 queries per second with a success rate of 72%. In SQLite, DQE generates 12,313 queries per second with a success rate of 97%. Coverage. To demonstrate the sufficiency of our testing, we compare code coverage with existing works, i.e., PQS [9], NoREC [10] and TLP [11]. We run each tool with the same experimental setting for 24 hours on MySQL and MariaDB4. 4We have not found a suitable way to perform code coverage measurements in TiDB, SQLite and CockroachDB. 2079 Table V shows our experiment results.",
+      "context_after": "NoREC achieves 18% line coverage in MariaDB. TLP achieves 18% line coverage in MySQL. DQE achieves 15% line coverage in MySQL and 21% line coverage in MariaDB. We can see that DQE obtains similar coverage with other works. This is reasonable, since DQE, PQS, NoREC and TLP are all built on SQLancer, which share the similar query generation. Note that, in SQLancer, NoREC does not support testing MySQL, PQS and TLP do not support testing MariaDB. Thus, we do not measure their code coverage. Although DQE can generate thousands of queries per second in MySQL and MariaDB, the coverage is low. This",
+      "section": "D Other Experimental Statistics",
+      "page": 9,
+      "char_offset": 40119,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC achieves 18% line coverage in MariaDB.",
+      "context_before": "uccess rate of 89%. In CockroachDB, DQE generates 243 queries per second with a success rate of 72%. In SQLite, DQE generates 12,313 queries per second with a success rate of 97%. Coverage. To demonstrate the sufficiency of our testing, we compare code coverage with existing works, i.e., PQS [9], NoREC [10] and TLP [11]. We run each tool with the same experimental setting for 24 hours on MySQL and MariaDB4. 4We have not found a suitable way to perform code coverage measurements in TiDB, SQLite and CockroachDB. 2079 Table V shows our experiment results. PQS achieves 19% line coverage in MySQL.",
+      "context_after": "TLP achieves 18% line coverage in MySQL. DQE achieves 15% line coverage in MySQL and 21% line coverage in MariaDB. We can see that DQE obtains similar coverage with other works. This is reasonable, since DQE, PQS, NoREC and TLP are all built on SQLancer, which share the similar query generation. Note that, in SQLancer, NoREC does not support testing MySQL, PQS and TLP do not support testing MariaDB. Thus, we do not measure their code coverage. Although DQE can generate thousands of queries per second in MySQL and MariaDB, the coverage is low. This is expected, because DQE only focuses on quer",
+      "section": "D Other Experimental Statistics",
+      "page": 9,
+      "char_offset": 40160,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP achieves 18% line coverage in MySQL.",
+      "context_before": "ates 243 queries per second with a success rate of 72%. In SQLite, DQE generates 12,313 queries per second with a success rate of 97%. Coverage. To demonstrate the sufficiency of our testing, we compare code coverage with existing works, i.e., PQS [9], NoREC [10] and TLP [11]. We run each tool with the same experimental setting for 24 hours on MySQL and MariaDB4. 4We have not found a suitable way to perform code coverage measurements in TiDB, SQLite and CockroachDB. 2079 Table V shows our experiment results. PQS achieves 19% line coverage in MySQL. NoREC achieves 18% line coverage in MariaDB.",
+      "context_after": "DQE achieves 15% line coverage in MySQL and 21% line coverage in MariaDB. We can see that DQE obtains similar coverage with other works. This is reasonable, since DQE, PQS, NoREC and TLP are all built on SQLancer, which share the similar query generation. Note that, in SQLancer, NoREC does not support testing MySQL, PQS and TLP do not support testing MariaDB. Thus, we do not measure their code coverage. Although DQE can generate thousands of queries per second in MySQL and MariaDB, the coverage is low. This is expected, because DQE only focuses on query processing in DBMSs. DBMSs also provide",
+      "section": "D Other Experimental Statistics",
+      "page": 9,
+      "char_offset": 40205,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This is reasonable, since DQE, PQS, NoREC and TLP are all built on SQLancer, which share the similar query generation.",
+      "context_before": "our testing, we compare code coverage with existing works, i.e., PQS [9], NoREC [10] and TLP [11]. We run each tool with the same experimental setting for 24 hours on MySQL and MariaDB4. 4We have not found a suitable way to perform code coverage measurements in TiDB, SQLite and CockroachDB. 2079 Table V shows our experiment results. PQS achieves 19% line coverage in MySQL. NoREC achieves 18% line coverage in MariaDB. TLP achieves 18% line coverage in MySQL. DQE achieves 15% line coverage in MySQL and 21% line coverage in MariaDB. We can see that DQE obtains similar coverage with other works.",
+      "context_after": "Note that, in SQLancer, NoREC does not support testing MySQL, PQS and TLP do not support testing MariaDB. Thus, we do not measure their code coverage. Although DQE can generate thousands of queries per second in MySQL and MariaDB, the coverage is low. This is expected, because DQE only focuses on query processing in DBMSs. DBMSs also provide many features that we do not test, e.g., user management, configuration, and fault tolerance. Parameter selection. We use some default parameters, e.g., maxTable =5,maxCol =10, and maxDept =3, to generate databases and queries. These parameters may affect",
+      "section": "D Other Experimental Statistics",
+      "page": 9,
+      "char_offset": 40383,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Note that, in SQLancer, NoREC does not support testing MySQL, PQS and TLP do not support testing MariaDB.",
+      "context_before": "h the same experimental setting for 24 hours on MySQL and MariaDB4. 4We have not found a suitable way to perform code coverage measurements in TiDB, SQLite and CockroachDB. 2079 Table V shows our experiment results. PQS achieves 19% line coverage in MySQL. NoREC achieves 18% line coverage in MariaDB. TLP achieves 18% line coverage in MySQL. DQE achieves 15% line coverage in MySQL and 21% line coverage in MariaDB. We can see that DQE obtains similar coverage with other works. This is reasonable, since DQE, PQS, NoREC and TLP are all built on SQLancer, which share the similar query generation.",
+      "context_after": "Thus, we do not measure their code coverage. Although DQE can generate thousands of queries per second in MySQL and MariaDB, the coverage is low. This is expected, because DQE only focuses on query processing in DBMSs. DBMSs also provide many features that we do not test, e.g., user management, configuration, and fault tolerance. Parameter selection. We use some default parameters, e.g., maxTable =5,maxCol =10, and maxDept =3, to generate databases and queries. These parameters may affect our bug detection effectiveness. However, the impact of these parameters could be low. The reasons are as",
+      "section": "D Other Experimental Statistics",
+      "page": 9,
+      "char_offset": 40502,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "pqs",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "citation_marker",
+      "surface": "[50]–[57]",
+      "technique": "qpg",
+      "sentence": "Database and SQL query generation have been widely explored by existing works [21]–[26], [29], [50]–[57].",
+      "context_before": "] and RD2 [49] apply differential testing on graph database systems. APOLLO [30] 12https://bugs.mysql.com/bug.php?id=106407feeds the same SELECT query into two different versions of the same DBMS to detect performance bugs. SparkFuzz [32] validates a query result with a reference DBMS (e.g., PostgreSQL) or with a different Spark version. We develop a novel differential testing approach, which executes SELECT, UPDATE and DELETE queries with the same predicate in a DBMS to detect logic bugs. Database and SQL query generation. One key component of automatic testing is an automatic input generator.",
+      "context_after": "SQLsmith [29] is an open source random SQL query generator, which is inspired by Csmith [58]. Go-randgen [54] can generate various SQL queries based on input SQL grammar. SQLRight [55] is a mutation-based SQL query generator, in which an intermediate representation is designed to perform mutations guided by coverage feedback. A better database and SQL query generation might improve the efficiency of our work in detecting logic bugs. Test oracles of DBMSs. Test oracles are the key to reveal DBMS bugs. ADUSA [26] uses Alloy [59], an open source language and analyzer, to analyze the expected quer",
+      "section": "VI RELATED WORK",
+      "page": 11,
+      "char_offset": 53213,
+      "cited_reference": {
+        "number": 50,
+        "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” inProceedings of International Conference on Software Engineering (ICSE), 2023.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [9] synthesizes a SELECT query, which is computed to fetch a randomly-selected pivot row, and checks whether the pivot row is contained in its query result.",
+      "context_before": "random SQL query generator, which is inspired by Csmith [58]. Go-randgen [54] can generate various SQL queries based on input SQL grammar. SQLRight [55] is a mutation-based SQL query generator, in which an intermediate representation is designed to perform mutations guided by coverage feedback. A better database and SQL query generation might improve the efficiency of our work in detecting logic bugs. Test oracles of DBMSs. Test oracles are the key to reveal DBMS bugs. ADUSA [26] uses Alloy [59], an open source language and analyzer, to analyze the expected query result of a given SELECT query.",
+      "context_after": "NoREC [10] rewrites a SELECT query as an equivalent one that the DBMS cannot optimize, and compares their results. TLP [11] leverages the ternary property of predicate evaluation, where the evaluation result is one of TRUE ,FALSE and NULL, to partition a SELECT query into three partitioning queries, whose combined query results are equal to the original query’s query result. Troc [60] proposes how to build a test oracle for a pair of transactions. Our work proposes a new test oracle for DBMS testing, and is complementary to existing approaches. VII. CONCLUSION Logic bugs in UPDATE and DELETE qu",
+      "section": "VI RELATED WORK",
+      "page": 11,
+      "char_offset": 53952,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [10] rewrites a SELECT query as an equivalent one that the DBMS cannot optimize, and compares their results.",
+      "context_before": "ation-based SQL query generator, in which an intermediate representation is designed to perform mutations guided by coverage feedback. A better database and SQL query generation might improve the efficiency of our work in detecting logic bugs. Test oracles of DBMSs. Test oracles are the key to reveal DBMS bugs. ADUSA [26] uses Alloy [59], an open source language and analyzer, to analyze the expected query result of a given SELECT query. PQS [9] synthesizes a SELECT query, which is computed to fetch a randomly-selected pivot row, and checks whether the pivot row is contained in its query result.",
+      "context_after": "TLP [11] leverages the ternary property of predicate evaluation, where the evaluation result is one of TRUE ,FALSE and NULL, to partition a SELECT query into three partitioning queries, whose combined query results are equal to the original query’s query result. Troc [60] proposes how to build a test oracle for a pair of transactions. Our work proposes a new test oracle for DBMS testing, and is complementary to existing approaches. VII. CONCLUSION Logic bugs in UPDATE and DELETE queries can cause severer consequences, e.g., incorrect database states, and have not been tackled by existing approa",
+      "section": "VI RELATED WORK",
+      "page": 11,
+      "char_offset": 54113,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [11] leverages the ternary property of predicate evaluation, where the evaluation result is one of TRUE ,FALSE and NULL, to partition a SELECT query into three partitioning queries, whose combined query results are equal to the original query’s query result.",
+      "context_before": "coverage feedback. A better database and SQL query generation might improve the efficiency of our work in detecting logic bugs. Test oracles of DBMSs. Test oracles are the key to reveal DBMS bugs. ADUSA [26] uses Alloy [59], an open source language and analyzer, to analyze the expected query result of a given SELECT query. PQS [9] synthesizes a SELECT query, which is computed to fetch a randomly-selected pivot row, and checks whether the pivot row is contained in its query result. NoREC [10] rewrites a SELECT query as an equivalent one that the DBMS cannot optimize, and compares their results.",
+      "context_after": "Troc [60] proposes how to build a test oracle for a pair of transactions. Our work proposes a new test oracle for DBMS testing, and is complementary to existing approaches. VII. CONCLUSION Logic bugs in UPDATE and DELETE queries can cause severer consequences, e.g., incorrect database states, and have not been tackled by existing approaches. In this paper, we propose a novel and general approach DQE to effectively detect logic bugs in SELECT ,UPDATE and DELETE queries. We evaluate DQE on five widely-used DBMSs, i.e., MySQL, MariaDB, TiDB, CockroachDB and SQLite. In total, we have detected 41 pre",
+      "section": "VI RELATED WORK",
+      "page": 11,
+      "char_offset": 54228,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/JensonSung/dqetool",
+    "markers": [
+      "renamed_sqlancer_package",
+      "sqlancer_source_content_match"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/JensonSung/dqetool",
+        "source_type": "github_repository",
+        "excerpt": "Replication package for \"Testing Database Systems via Differential Query Execution\", accepted at ICSE 2023\n\n# DQETool\n\nDQETool is the implementation of differential query execution in paper.\n\n# Getting Started\n\nRequirements:\n* Java 11 or above",
+        "note": "Repository names this paper.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/JensonSung/dqetool/blob/main/src/dqetool/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+        "excerpt_is_verbatim": true,
+        "note": "src/dqetool/Randomly.java is SQLancer's Randomly.java, with the package renamed to dqetool (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+        "content_sha256": "sha256:c4aa77016b32eb0b4a0e071764bf41adfdc03763a40486ebcb64e9c0a82df6ab",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "uses_infrastructure": [
+        "M7",
+        "M8"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:36:46Z",
+    "is_model_written": true,
+    "summary": "DQE tests DBMSs by checking that the rows a SELECT identifies are the same rows an UPDATE or DELETE with the same predicate acts on. The gap it addresses is that existing logic-bug oracles only examine SELECT: a predicate evaluated correctly in a query can still be evaluated wrongly when it drives a data-modifying statement, and no oracle looked there. Applied to MySQL, MariaDB and TiDB it found 32 bugs, 30 confirmed and 21 fixed.",
+    "narrative": "DQE is built on SQLancer. The paper states that its implementation is based on SQLancer and that its database generation is mainly adopted from it, and it notes that DQE, PQS, NoREC and TLP are all built on SQLancer and so share similar query generation -- offered as the explanation for their comparable code coverage. PQS, NoREC and TLP are also its baselines: it reports coverage and bug counts against all three, and its motivation is that none of them can be applied to UPDATE and DELETE statements, which is the gap DQE fills.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "definition",
+      "M5": "motivation",
+      "M6": "motivation",
+      "M7": "reuse_component",
+      "M8": "reuse_implementation",
+      "M9": "baseline",
+      "M10": "result_comparison",
+      "M11": "background",
+      "M12": "result_comparison",
+      "M13": "result_comparison",
+      "M14": "result_comparison",
+      "M15": "result_comparison",
+      "M16": "reuse_implementation",
+      "M17": "incidental",
+      "M18": "background",
+      "M19": "definition",
+      "M20": "definition",
+      "M21": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M7",
+          "M8",
+          "M16"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M7",
+            "sentence": "Our database generation is mainly adopted from SQLancer [27].",
+            "section": "B Database Generation",
+            "page": 4
+          },
+          {
+            "mention_id": "M8",
+            "sentence": "EVALUATION We implement DQE based on SQLancer [27], which is implemented in Java.",
+            "section": "IV EVALUATION",
+            "page": 7
+          },
+          {
+            "mention_id": "M16",
+            "sentence": "This is reasonable, since DQE, PQS, NoREC and TLP are all built on SQLancer, which share the similar query generation.",
+            "section": "D Other Experimental Statistics",
+            "page": 9
+          }
+        ],
+        "reasoning": "M8 states DQE is implemented based on SQLancer, M7 that its database generation is mainly adopted from it, and M16 that DQE is built on SQLancer alongside PQS, NoREC and TLP. This is the codebase, not a workload.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "DQE is a new oracle rather than an extension of one. M5 and M6 give the relationship: PQS, NoREC and TLP cannot be adopted on UPDATE and DELETE queries, and DQE addresses that by comparing a SELECT against a data-modifying statement -- a different oracle, not a generalisation of theirs."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M9",
+          "M10",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M9",
+            "sentence": ", PQS [9], NoREC [10] and TLP [11]) that aim to detect logic bugs in DBMSs.",
+            "section": "C Comparing with Existing Approaches",
+            "page": 8
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "Thus, they cannot detect 18 logic bugs related to this kind of errors TABLE VCOVERAGE INFORMATION Tool MySQL MariaDB PQS 19 NoREC-18 TLP 18 DQE 15 21 in SELECT queries.",
+            "section": "C Comparing with Existing Approaches",
+            "page": 8
+          },
+          {
+            "mention_id": "M12",
+            "sentence": ", PQS [9], NoREC [10] and TLP [11].",
+            "section": "D Other Experimental Statistics",
+            "page": 8
+          }
+        ],
+        "reasoning": "M9 names PQS, NoREC and TLP as the existing approaches DQE is compared with, and M10 and M12 report the outcome in bugs found and line coverage.",
+        "techniques": [
+          "pqs",
+          "norec",
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The techniques are described by what they do and by their limits, not as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse55347_2025_00003.json b/_data/papers/paper_doi_10_1109_icse55347_2025_00003.json
new file mode 100644
index 0000000..553dfd3
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse55347_2025_00003.json
@@ -0,0 +1,715 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:05:57Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse55347.2025.00003",
+    "title": "Coni: Detecting Database Connector Bugs via State-Aware Test Case Generation",
+    "authors": [
+      "Wenqian Deng",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Mingzhe Wang",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse55347.2025.00003",
+    "arxiv_id": null,
+    "s2_paper_id": "7307f131c3d332683815de0634412c48e6a01a7a",
+    "url": "https://doi.org/10.1109/icse55347.2025.00003",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icse55347.2025.00003",
+      "retrieved_at": "2026-09-07T17:05:57Z",
+      "chars": 66040,
+      "content_sha256": "sha256:c8431f3ca44815090e215d219b7972e9db5ec551823bb6492c12a6f48b9a2124"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2274
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 10266
+      },
+      {
+        "number": "III",
+        "title": "STATE-AWARE TESTCASE GENERATION",
+        "start": 15238
+      },
+      {
+        "number": "A",
+        "title": "State Model Establishment",
+        "start": 16243
+      },
+      {
+        "number": "B",
+        "title": "Interface Call Sequence Generation",
+        "start": 21323
+      },
+      {
+        "number": "C",
+        "title": "Parameter Value Generation",
+        "start": 26525
+      },
+      {
+        "number": "IV",
+        "title": "IMPLEMENTATION",
+        "start": 31622
+      },
+      {
+        "number": "V",
+        "title": "EVALUATION",
+        "start": 33390
+      },
+      {
+        "number": "A",
+        "title": "Evaluation Setup",
+        "start": 33777
+      },
+      {
+        "number": "B",
+        "title": "Database Connector Bugs",
+        "start": 34869
+      },
+      {
+        "number": "C",
+        "title": "Comparison with Existing Techniques",
+        "start": 41229
+      },
+      {
+        "number": "D",
+        "title": "Effectiveness of Test Case Generation",
+        "start": 45460
+      },
+      {
+        "number": "VI",
+        "title": "DISCUSSION",
+        "start": 48382
+      },
+      {
+        "number": "VII",
+        "title": "RELATED WORK",
+        "start": 51336
+      },
+      {
+        "number": "VIII",
+        "title": "CONCLUSION",
+        "start": 56870
+      },
+      {
+        "number": "Y",
+        "title": "Jiang, “WingFuzz: Implementing Continuous Fuzzing",
+        "start": 59954
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Database connection,” https://en.wikipedia.org/wiki/ Database connection, 2024, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“Java database connectivity,” https://en.wikipedia.org/ wiki/Java Database Connectivity, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“Open database connectivity,” https://en.wikipedia.org/ wiki/Open Database Connectivity, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "“pgjdbc,” https://github.com/pgjdbc/pgjdbc, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "“mariadb-connector-j,” https://github.com/mariadbcorporation/mariadb-connector-j, 2024, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "“Binary bool values are not decoded correctly · Issue #2639 · pgjdbc/pgjdbc,” https://github.com/pgjdbc/ pgjdbc/issues/2639, 2022, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "“[conj-1071] error during bulk execution might result in connection wrong statejira,” https://jira.mariadb.org/ browse/CONJ-1071, 2023, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "“[conj-1091] can’t make a connection when the Read Replica DB is in a hang state when SocketTimeout=0 set-Jira,” https://jira.mariadb.org/browse/CONJ-1091, 2023, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "“MySQL Bugs: #109013: useServerPrepStmts and useLocalTransactionState could cause rollback failure,” https://bugs.mysql.com/bug.php?id=109013, 2022, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Z.-M. Jiang, J.-J. Bai, and Z. Su, “DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation,” in 32nd USENIX Security Symposium, 2023, pp. 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "M. Rigger and Z. Su, “Testing Database Engines via Pivoted Query Synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 12,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback,” in The ACM Conference on Computer and Communications Security (CCS), 2020, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammar-Free DBMS Fuzzing,” in 37th IEEE/ACM International Conference on Automated Software Engineering, 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "J. Liang, Z. Wu, J. Fu, Y. Bai, Q. Zhang, and Y. Jiang, “WingFuzz: Implementing Continuous Fuzzing for DBMSs,” in 2024 USENIX Annual Technical Conference, 2024, pp. 479–492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "C. Pacheco, S. K. Lahiri, M. D. Ernst, and T. Ball, “Feedback-Directed Random Test Generation,” in 29th International Conference on Software Engineering (ICSE 2007), Minneapolis, MN, USA, May 20-26, 2007. IEEE Computer Society, 2007, pp. 75–84. [Online]. Available: https://doi.org/10.1109/ICSE.2007.37",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Y. Lin, J. Sun, G. Fraser, Z. Xiu, T. Liu, and J. S. Dong, “Recovering Fitness Gradients for Interprocedural Boolean Flags in Search-based Testing,” in Proceedings of the 29th ACMSIGSOFT International Symposium on Software Testing and Analysis, 2020, pp. 440–451.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Z. Huang and Y. Wang, “JDriver: Automatic Driver Class Generation for AFL-Based Java Fuzzing Tools,” 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "“Differential testing,” https://en.wikipedia.org/wiki/ Differential testing, 2024, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "“mysql-connector-j,” https://github.com/mysql/mysql677 connector-j, 2024, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "“aws-mysql-jdbc,” https://github.com/awslabs/awsmysql-jdbc, 2024, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "“pgjdbc-ng,” https://github.com/impossibl/pgjdbc-ng, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "“TIOBE Index,” https://www.tiobe.com/tiobe-index/, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "“Maven JDBC,” https://mvnrepository.com/open-source/ jdbc-drivers, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "“SQL1992,” http://www .contrib.andrew.cmu.edu/ ∼shadow/sql/sql1992.txt, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "“Mysql,” https://www.mysql.com/, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "“Mariadb,” https://mariadb.org/, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "“Postgresql,” https://www.postgresql.org/, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "“Aws,” https://aws.amazon.com/, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "“JaCoCO,” https://www.jacoco.org/jacoco/, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "D. Nam, A. Macvean, V. Hellendoorn, B. Vasilescu, and B. Myers, “Using an LLM to Help With Code Understanding,” in Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "S. Bhosale, M. T. Patil, and M. P. Patil, “Sqlite: Light database system,” Int. J. Comput. Sci. Mob. Comput, vol. 44, no. 4, pp. 882–885, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "“Clickhouse,” https://clickhouse.com/docs/en/interfaces/ jdbc, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "“Oceanbase,” https://en.oceanbase.com/docs/commonoceanbase-connector-j-en-10000000001092963, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "B. P. Miller, L. Fredriksen, and B. So, “An Empirical Study of the Reliability of UNIX Utilities,” Commun. ACM, vol. 33, no. 12, Dec. 1990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "J. Liang, Y. Chen, Z. Wu, J. Fu, M. Wang, Y. Jiang, X. Huang, T. Chen, J. Wang, and J. Li, “Sequenceoriented DBMS Fuzzing,” in Proceedings of IEEE International Conference on Data Engineering (ICDE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Z. Wu, J. Liang, M. Wang, C. Zhou, and Y. Jiang, “Unicorn: Detect Runtime Errors in Time-series Databases with Hybrid Input Synthesis,” in ISSTA ’22: 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, South Korea, July 18 - 22, 2022. ACM, 2022, pp. 251–262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "J. Fu, J. Liang, Z. Wu, and Y. Jiang, “Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer,” in Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "E. Albert, I. Cabanas, A. Flores-Montoya, M. GomezZamalloa, and S. Gutierrez, “jPET: An Automatic TestCase Generator for Java,” 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Z. Chen and Y. Wang, “JFD: Automatic Java Fuzz DriverGeneration,” 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "M. Chen, X. Qiu, W. Xu, L. Wang, J. Zhao, and X. Li, “UML Activity Diagram-based Automatic Test Case Generation for Java Programs,” The Computer Journal, vol. 52, no. 5, pp. 545–556, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "C. Gould, Z. Su, and P. Devanbu, “JDBC Checker: A Static Analysis Tool for SQL/JDBC Applications,” in Proceedings. 26th International Conference on Software Engineering. IEEE, 2004, pp. 697–698.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "C. Zhou and P. Frankl, “Mutation Testing for Java Database Applications,” pp. 396–405, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Y. Shin, L. A. Williams, and T. Xie, “SQLUnitgen: Test case generation for SQL injection detection,” North Carolina State University. Dept. of Computer Science, Tech. Rep., 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "D. Ye, “Automated Testing Framework for ODBC Driver,” Journal of Software Engineering and Applications, vol. 04, no. 12, p. 688, Dec. 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "S. R. Dalal, A. Jain, N. Karunanithi, J. Leaton, C. M. Lott, G. C. Patton, and B. M. Horowitz, “Model-based Testing in Practice,” in Proceedings of the 21st international conference on Software engineering, 1999, pp. 285–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "M. Utting, A. Pretschner, and B. Legeard, “A Taxonomy of Model-based Testing Approaches,” Software testing, verification and reliability, vol. 22, no. 5, pp. 297–312, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "“Uppaal,” https://uppaal.org/, accessed: November 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "P. V. P. Pinheiro, A. T. Endo, and A. Simao, “Modelbased Testing of RESTful Web Services Using UML Protocol State Machines,” in Brazilian workshop on systematic and automated software testing. Citeseer, 2013, pp. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "G. Tretmans and H. Brinksma, “Torx: Automated Modelbased Testing,” in First European Conference on ModelDriven Software Engineering, 2003, pp. 31–43.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "D. R. Slutz, “Massive Stochastic Testing of SQL,” vol. 98, pp. 618–622, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear),” inProceedings of the 46th International Conference on Very Large Data Bases (VLDB), Tokyo, Japan, Aug. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Z. Cui, W. Dou, Q. Dai, J. Song, W. Wang, J. Wei, and D. Ye, “Differentially Testing Database Transactions for Fun and Profit,” in Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "J. Liang, Z. Wu, J. Fu, M. Wang, C. Sun, and Y. Jiang, “Mozi: Discovering DBMS Bugs via ConfigurationBased Equivalent Transformation,” in Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–12. 678",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 11,
+      "text": "M. Rigger and Z. Su, “Testing Database Engines via Pivoted Query Synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[10, 11, 12, 13, 14]",
+      "technique": "pqs",
+      "sentence": "On the one hand, most fuzzers primarily concentrate on generating effective SQL queries [10, 11, 12, 13, 14], whereas database connectors are not directly involved in the execution of these queries.",
+      "context_before": "in the type of parameter data within the batch. Specifically, in Figure 1, the “field” in the first INSERT statement is of type string, while the other has a null value. If an error occurs during any of these statements, the connector may fail to read the complete results of all statements, leading to an incorrect connection state. Subsequent commands on this corrupted connection may retrieve residual data from the preceding error, yielding inaccurate query results. Although there are many existing DBMS-related fuzzing works, they are hard to be directly applied to testing database connectors.",
+      "context_after": "Therefore, it is challenging for existing methods to assess the intricate logic embedded within database connectors thoroughly. On the other hand, existing fuzzers exhibit limited interaction with the database connector. They either refrain from utilizing the database connector entirely or leverage only a small subset of its interfaces to transmit SQL queries. For example, SQLANCER [11] relies on JDBC solely to execute SQL queries and retrieve results, without exploring additional functionalities like modifying configuration properties or batch execution. In addition, works like RANDOOP [15]",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 5802,
+      "cited_reference": {
+        "number": 11,
+        "text": "M. Rigger and Z. Su, “Testing Database Engines via Pivoted Query Synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "For example, SQLANCER [11] relies on JDBC solely to execute SQL queries and retrieve results, without exploring additional functionalities like modifying configuration properties or batch execution.",
+      "context_before": "pplied to testing database connectors. On the one hand, most fuzzers primarily concentrate on generating effective SQL queries [10, 11, 12, 13, 14], whereas database connectors are not directly involved in the execution of these queries. Therefore, it is challenging for existing methods to assess the intricate logic embedded within database connectors thoroughly. On the other hand, existing fuzzers exhibit limited interaction with the database connector. They either refrain from utilizing the database connector entirely or leverage only a small subset of its interfaces to transmit SQL queries.",
+      "context_after": "In addition, works like RANDOOP [15] and EVOSUITE [16] have already explored how to generate test cases automatically for programs [15, 16, 17]. However, due to the lack of domain knowledge, these tools can only generate test cases that cover limited scenarios and logic, typically identifying issues like null pointer errors or boundary condition handling errors. Testing connectors requires various interactions with databases, but these tools are almost incapable of generating effective database connections. Therefore, there is a great need for a framework designed specifically for testing dat",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 6364,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "In addition, SQLANCER [11] is a popular open-source tool for testing databases using JDBC.",
+      "context_before": "e connectors, among them 34 have been confirmed. The results indicate that C ONIcan detect previously unknown bugs in database connectors, which adequately answers RQ1. C. Comparison with Existing Techniques Compared Techniques. To the best of our knowledge, CONI is the first fuzzing framework for database connectors. Therefore, we selected closely related work as the evaluation baseline. RANDOOP [15] and EVOSUITE [16] are two prominentand open-source automated tools for generating test cases of Java programs, while CONI generates test cases for JDBC, which can be considered as a Java library.",
+      "context_after": "We implemented SQLANCER+ by adapting the target database connector and collecting results from different connectors to identify inconsistencies. We evaluated these techniques using two metrics, namely branch coverage and unique detected bugs. Coverage was collected using Jacoco [29] instrumentation. The bugs were deduplicated automatically by comparing the interface call information and the returned result, identifying inconsistencies. Then the deduplicated bugs were reported to developers for their verification. We ran the testing tools on five tested database connectors for 24 hours and col",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 41660,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "We implemented SQLANCER+ by adapting the target database connector and collecting results from different connectors to identify inconsistencies.",
+      "context_before": "previously unknown bugs in database connectors, which adequately answers RQ1. C. Comparison with Existing Techniques Compared Techniques. To the best of our knowledge, CONI is the first fuzzing framework for database connectors. Therefore, we selected closely related work as the evaluation baseline. RANDOOP [15] and EVOSUITE [16] are two prominentand open-source automated tools for generating test cases of Java programs, while CONI generates test cases for JDBC, which can be considered as a Java library. In addition, SQLANCER [11] is a popular open-source tool for testing databases using JDBC.",
+      "context_after": "We evaluated these techniques using two metrics, namely branch coverage and unique detected bugs. Coverage was collected using Jacoco [29] instrumentation. The bugs were deduplicated automatically by comparing the interface call information and the returned result, identifying inconsistencies. Then the deduplicated bugs were reported to developers for their verification. We ran the testing tools on five tested database connectors for 24 hours and collected the branch coverage and unique detected bugs. Coverage. Table III displays the number of branches covered by each technique in 24 hours. T",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 41751,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "The result shows that CONI covered 5950, 6608, and 6587 more branches than SQLANCER+, RANDOOP, and EVOSUITE respectively.",
+      "context_before": "s. We evaluated these techniques using two metrics, namely branch coverage and unique detected bugs. Coverage was collected using Jacoco [29] instrumentation. The bugs were deduplicated automatically by comparing the interface call information and the returned result, identifying inconsistencies. Then the deduplicated bugs were reported to developers for their verification. We ran the testing tools on five tested database connectors for 24 hours and collected the branch coverage and unique detected bugs. Coverage. Table III displays the number of branches covered by each technique in 24 hours.",
+      "context_after": "The main reason that CONI covered more branches is that CONI is capable of generating effective test cases that cover various functional scenarios. Specifically, SQLANCER+ only uses the database connector to send SQL queries, utilizing getConnection to create a database connection, createStatement to create a statement, and execute to execute SQL queries. For RANDOOP and EVOSUITE, the main problem is that due to the lack of domain knowledge, these tools can only generate test cases that cover limited scenarios and logic. Testing connectors requires complex interactions with databases, but the",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 42494,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "Specifically, SQLANCER+ only uses the database connector to send SQL queries, utilizing getConnection to create a database connection, createStatement to create a statement, and execute to execute SQL queries.",
+      "context_before": "dentifying inconsistencies. Then the deduplicated bugs were reported to developers for their verification. We ran the testing tools on five tested database connectors for 24 hours and collected the branch coverage and unique detected bugs. Coverage. Table III displays the number of branches covered by each technique in 24 hours. The result shows that CONI covered 5950, 6608, and 6587 more branches than SQLANCER+, RANDOOP, and EVOSUITE respectively. The main reason that CONI covered more branches is that CONI is capable of generating effective test cases that cover various functional scenarios.",
+      "context_after": "For RANDOOP and EVOSUITE, the main problem is that due to the lack of domain knowledge, these tools can only generate test cases that cover limited scenarios and logic. Testing connectors requires complex interactions with databases, but these tools, without the state model, are almost incapable of generating effective database connections. CONI attempts to alter configurations in the database connector, perform batch executions, and modify query results. Therefore, CONI can cover more branches than other techniques. TABLE III: Number of branches covered by each technique in 24 hours Connecto",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 42764,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "TABLE III: Number of branches covered by each technique in 24 hours Connector CONI SQLANCER+RANDOOP EVOSUITE MariaDB Connector/J 1073 466 583 581 MySQL Connector/J 2430 1256 1473 1489 AWS MySQL JDBC 2826 1445 1734 1739 PGJDBC 1660 987 1197 1181 PGJDBC NG 2425 1796 1621 1597 T otal 10414 5950 6608 6587 Impro vement-4464↑ 3806↑ 3827↑ Bugs.",
+      "context_before": "n, createStatement to create a statement, and execute to execute SQL queries. For RANDOOP and EVOSUITE, the main problem is that due to the lack of domain knowledge, these tools can only generate test cases that cover limited scenarios and logic. Testing connectors requires complex interactions with databases, but these tools, without the state model, are almost incapable of generating effective database connections. CONI attempts to alter configurations in the database connector, perform batch executions, and modify query results. Therefore, CONI can cover more branches than other techniques.",
+      "context_after": "Table IV shows the number of detected bugs in database connectors by each technique in 24 hours. During the evaluation, CONI found 5, 6, 3, 2, and 5 bugs in MariaDB Connecor/J, MySQL Connector/J, AWS MySQL JDBC, PGJDBC, and PGJDBC NG respectively, while other techniques did not find any bug. The main reason is as follows: SQLANCER+ did not call the interface methods that can trigger bugs, nor did it attempt to pass illegal values or set configurations for interface methods. For example, SQLANCER+ cannot find the 674 bug in Listing 1 because it did not call the setFetchSize method. RANDOOP a",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 43497,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "The main reason is as follows: SQLANCER+ did not call the interface methods that can trigger bugs, nor did it attempt to pass illegal values or set configurations for interface methods.",
+      "context_before": "vered by each technique in 24 hours Connector CONI SQLANCER+RANDOOP EVOSUITE MariaDB Connector/J 1073 466 583 581 MySQL Connector/J 2430 1256 1473 1489 AWS MySQL JDBC 2826 1445 1734 1739 PGJDBC 1660 987 1197 1181 PGJDBC NG 2425 1796 1621 1597 T otal 10414 5950 6608 6587 Impro vement-4464↑ 3806↑ 3827↑ Bugs. Table IV shows the number of detected bugs in database connectors by each technique in 24 hours. During the evaluation, CONI found 5, 6, 3, 2, and 5 bugs in MariaDB Connecor/J, MySQL Connector/J, AWS MySQL JDBC, PGJDBC, and PGJDBC NG respectively, while other techniques did not find any bug.",
+      "context_after": "For example, SQLANCER+ cannot find the 674 bug in Listing 1 because it did not call the setFetchSize method. RANDOOP and EVOSUITE focused on unit testing, which can only generate the test cases for several simple methods. Although they reported some potential errors during the experiment, they were all false positives which are NullPointerException caused by invalid sequences of interface calls. The performance improvement of CONI is primarily attributed to the validity and complexity of the generated test case. CONI generates test cases that explore more state space because the state-aware",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 44130,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "For example, SQLANCER+ cannot find the 674 bug in Listing 1 because it did not call the setFetchSize method.",
+      "context_before": "PGJDBC 1660 987 1197 1181 PGJDBC NG 2425 1796 1621 1597 T otal 10414 5950 6608 6587 Impro vement-4464↑ 3806↑ 3827↑ Bugs. Table IV shows the number of detected bugs in database connectors by each technique in 24 hours. During the evaluation, CONI found 5, 6, 3, 2, and 5 bugs in MariaDB Connecor/J, MySQL Connector/J, AWS MySQL JDBC, PGJDBC, and PGJDBC NG respectively, while other techniques did not find any bug. The main reason is as follows: SQLANCER+ did not call the interface methods that can trigger bugs, nor did it attempt to pass illegal values or set configurations for interface methods.",
+      "context_after": "RANDOOP and EVOSUITE focused on unit testing, which can only generate the test cases for several simple methods. Although they reported some potential errors during the experiment, they were all false positives which are NullPointerException caused by invalid sequences of interface calls. The performance improvement of CONI is primarily attributed to the validity and complexity of the generated test case. CONI generates test cases that explore more state space because the state-aware method covers more interface methods and their interactions. TABLE IV: Number of detected bugs in database con",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 44316,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "Connector CONI SQLANCER+RANDOOP EVOSUITE MariaDB Connector/J 5 0 0 0 MySQL Connector/J 6 0 0 0 AWS MySQL JDBC 3 0 0 0 PGJDBC 2 0 0 0 PGJDBC NG 5 0 0 0 T otal 21 0 0 0 Impro vement-21↑ 21↑ 21↑ In summary, CONI is unique in its ability to find bugs in database connectors, and compared to other techniques, it can cover more branches within the database connectors, which adequately answers RQ2.",
+      "context_before": "testing, which can only generate the test cases for several simple methods. Although they reported some potential errors during the experiment, they were all false positives which are NullPointerException caused by invalid sequences of interface calls. The performance improvement of CONI is primarily attributed to the validity and complexity of the generated test case. CONI generates test cases that explore more state space because the state-aware method covers more interface methods and their interactions. TABLE IV: Number of detected bugs in database connectors by each technique in 24 hours.",
+      "context_after": "D. Effectiveness of Test Case Generation To understand the contribution of each technique in CONI, we implemented two variants of CONI!sand CONI!p. CONI!s disabled the state-aware interface call sequence generation and generated random sequences. CONI!pdisabled the parameter value generation method and generated the values randomly. To avoid early errors in test cases from failed database connections, we used DriverManager.getConnection to build valid connections first for CONI!sand CONI!pand then subsequent interface call sequences or input parameter values were generated randomly. Bugs. Tab",
+      "section": "C Comparison with Existing Techniques",
+      "page": 9,
+      "char_offset": 45065,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[11, 12, 13, 35, 36, 37]",
+      "technique": "pqs",
+      "sentence": "When applying fuzzing techniques to test DBMSs, the main challenge is to generate correct and effective SQL queries [11, 12, 13, 35, 36, 37].",
+      "context_before": "epends on the comparison objects used. If two connectors produce consistent results due to identical logic errors, C ONImay fail to detect these errors. To mitigate this, a practical solution is to expand the range of connectors used for comparison in CONI. For instance, to thoroughly test the AWS MySQL JDBC connector, it can be compared not only with MySQL Connector/J but also with MariaDB Connector/J. This broader comparison increases the likelihood of uncovering unique errors specific to each connector. VII. RELATED WORK DBMS Fuzzing. Fuzzing [34] is an automated software testing technique.",
+      "context_after": "For example, SQLANCER [11] generates valid SQL queries based on AST. SQUIRREL [12] employs mutation-based fuzzing on DBMSs. Griffin [13] introduces metadata graphs for SQL, providing a grammarfree method for mutating SQL test cases. LEGO [35] enhances DBMS fuzzing by generating SQL sequences with a diverse range of types. Unicorn [36] proposed the hybrid input synthesis and designed the time-series model to generate time-series queries. These works are difficult to directly adapt to testing database connectors, since they focus on SQL generation, while database connectors don’t execute SQL qu",
+      "section": "VII RELATED WORK",
+      "page": 10,
+      "char_offset": 51424,
+      "cited_reference": {
+        "number": 11,
+        "text": "M. Rigger and Z. Su, “Testing Database Engines via Pivoted Query Synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "For example, SQLANCER [11] generates valid SQL queries based on AST.",
+      "context_before": "se errors. To mitigate this, a practical solution is to expand the range of connectors used for comparison in CONI. For instance, to thoroughly test the AWS MySQL JDBC connector, it can be compared not only with MySQL Connector/J but also with MariaDB Connector/J. This broader comparison increases the likelihood of uncovering unique errors specific to each connector. VII. RELATED WORK DBMS Fuzzing. Fuzzing [34] is an automated software testing technique. When applying fuzzing techniques to test DBMSs, the main challenge is to generate correct and effective SQL queries [11, 12, 13, 35, 36, 37].",
+      "context_after": "SQUIRREL [12] employs mutation-based fuzzing on DBMSs. Griffin [13] introduces metadata graphs for SQL, providing a grammarfree method for mutating SQL test cases. LEGO [35] enhances DBMS fuzzing by generating SQL sequences with a diverse range of types. Unicorn [36] proposed the hybrid input synthesis and designed the time-series model to generate time-series queries. These works are difficult to directly adapt to testing database connectors, since they focus on SQL generation, while database connectors don’t execute SQL queries. Addi-tionally, many existing fuzzers do not use database conne",
+      "section": "VII RELATED WORK",
+      "page": 10,
+      "char_offset": 51566,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M5"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:00:23Z",
+    "is_model_written": true,
+    "summary": "Coni tests database connectors -- the JDBC drivers between an application and a DBMS -- rather than the database itself. Its observation is that SQL fuzzers exercise the connector only as a pipe: they open a connection, create a statement and execute a query, leaving batch execution, configuration properties and the rest of the interface untested. Coni generates state-aware sequences of connector API calls and found 21 bugs across five connectors, where the baselines found none.",
+    "narrative": "SQLancer is both the motivating example and a baseline the authors had to build. The paper notes that SQLancer relies on JDBC solely to execute queries and retrieve results, without touching configuration properties or batch execution, which is the gap Coni targets. To measure against it the authors implemented SQLANCER+, adapting the connector and collecting results across connectors to find inconsistencies. It covered 5950 fewer branches than Coni and found none of the 21 bugs -- and the paper explains precisely why, giving a bug that requires calling setFetchSize, a method SQLANCER+ never calls.",
+    "roles": {
+      "M1": "background",
+      "M2": "motivation",
+      "M3": "definition",
+      "M4": "baseline",
+      "M5": "result_comparison",
+      "M6": "result_comparison",
+      "M7": "result_comparison",
+      "M8": "result_comparison",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "background",
+      "M12": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M4 describes building SQLANCER+ by adapting SQLancer so that it could serve as a comparison target. Coni's own test generation is its state-aware API sequence construction, not anything built on SQLancer."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The adaptation exists to make a baseline runnable on connectors, not to develop a SQLancer oracle. Coni's oracle is differential comparison across connectors."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M5",
+          "M10"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "We implemented SQLANCER+ by adapting the target database connector and collecting results from different connectors to identify inconsistencies.",
+            "section": "C Comparison with Existing Techniques",
+            "page": 8
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "The result shows that CONI covered 5950, 6608, and 6587 more branches than SQLANCER+, RANDOOP, and EVOSUITE respectively.",
+            "section": "C Comparison with Existing Techniques",
+            "page": 8
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "Connector CONI SQLANCER+RANDOOP EVOSUITE MariaDB Connector/J 5 0 0 0 MySQL Connector/J 6 0 0 0 AWS MySQL JDBC 3 0 0 0 PGJDBC 2 0 0 0 PGJDBC NG 5 0 0 0 T otal 21 0 0 0 Impro vement-21↑ 21↑ 21↑ In summary, CONI is unique in its ability to find bugs in database connectors, and compared to other techniques, it can cover more branches within the database connectors, which adequately answers RQ2.",
+            "section": "C Comparison with Existing Techniques",
+            "page": 9
+          }
+        ],
+        "reasoning": "M4 records SQLANCER+ being implemented as a comparison target, M5 reports the branch coverage difference, and M10 gives the per-connector bug table where it found none."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is called a popular open-source tool for testing databases using JDBC; popularity is not a state-of-the-art claim."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse55347_2025_00013.json b/_data/papers/paper_doi_10_1109_icse55347_2025_00013.json
new file mode 100644
index 0000000..aa14ca9
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse55347_2025_00013.json
@@ -0,0 +1,493 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:54:00Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse55347.2025.00013",
+    "title": "Janus: Detecting Rendering Bugs in Web Browsers via Visual Delta Consistency",
+    "authors": [
+      "Chijin Zhou",
+      "Quan Zhang",
+      "Bingzhou Qian",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse55347.2025.00013",
+    "arxiv_id": null,
+    "s2_paper_id": "3d2a62e3ce533a4bd184e9c0cdafea9c3e09e2b1",
+    "url": "https://doi.org/10.1109/icse55347.2025.00013",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icse55347.2025.00013",
+      "retrieved_at": "2026-09-08T14:54:00Z",
+      "chars": 70053,
+      "content_sha256": "sha256:1f811d78e5800778cbef7efc148739105c72057ceb326969ac3b3013eb1f0642"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1546
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND ANDMOTIVATION",
+        "start": 8085
+      },
+      {
+        "number": "III",
+        "title": "VISUAL DELTA CONSISTENCY",
+        "start": 15457
+      },
+      {
+        "number": "A",
+        "title": "Basic Idea",
+        "start": 15487
+      },
+      {
+        "number": "B",
+        "title": "Soundness",
+        "start": 17250
+      },
+      {
+        "number": "IV",
+        "title": "DESIGN OFJANUS",
+        "start": 22592
+      },
+      {
+        "number": "A",
+        "title": "Initial DOM Generation",
+        "start": 24227
+      },
+      {
+        "number": "B",
+        "title": "HTML Transformation",
+        "start": 27871
+      },
+      {
+        "number": "C",
+        "title": "Consistency Checking",
+        "start": 31671
+      },
+      {
+        "number": "V",
+        "title": "EVALUATION",
+        "start": 32757
+      },
+      {
+        "number": "A",
+        "title": "Bug finding",
+        "start": 33995
+      },
+      {
+        "number": "B",
+        "title": "Comparative Study",
+        "start": 41434
+      },
+      {
+        "number": "C",
+        "title": "False Positives",
+        "start": 44212
+      },
+      {
+        "number": "D",
+        "title": "Overhead Breakdown",
+        "start": 47822
+      },
+      {
+        "number": "VI",
+        "title": "DISCUSSION",
+        "start": 49103
+      },
+      {
+        "number": "VII",
+        "title": "RELATED WORK",
+        "start": 52249
+      },
+      {
+        "number": "VIII",
+        "title": "CONCLUSION",
+        "start": 54908
+      },
+      {
+        "number": "IX",
+        "title": "ACKNOWLEDGEMENT",
+        "start": 55503
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Ibrahim Althomali, Gregory M. Kapfhammer, and Phil McMinn. Automatic visual verification of layout failures in responsively designed web pages. In 12th IEEE Conference on Software Testing, Validation and Verification, ICST 2019, Xi’an, China, April 22-27, 2019, pages 183–193. IEEE, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Earl T. Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. The oracle problem in software testing: A survey. IEEE Trans. Software Eng., 41(5):507–525, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Johannes Buchner. A python perceptual image hashing module. https: //github.com/JohannesBuchner/imagehash, 2023. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Yuting Chen and Zhendong Su. Guided differential testing of certificate validation in SSL/TLS implementations. In Elisabetta Di Nitto, Mark Harman, and Patrick Heymans, editors, Proceedings of the 2015 10th Joint Meeting on Foundations of Software Engineering, ESEC/FSE 2015, Bergamo, Italy, August 30 - September 4, 2015, pages 793–804. ACM, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Shauvik Roy Choudhary, Husayn Versee, and Alessandro Orso. WEBDIFF: automated identification of cross-browser issues in web applications. In Radu Marinescu, Michele Lanza, and Andrian Marcus, editors, 26th IEEE International Conference on Software Maintenance (ICSM 2010), September 12-18, 2010, Timisoara, Romania, pages 1–10. IEEE Computer Society, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Yinlin Deng, Chenyuan Yang, Anjiang Wei, and Lingming Zhang. Fuzzing deep-learning libraries via automated relational API inference. In Abhik Roychoudhury, Cristian Cadar, and Miryung Kim, editors, Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2022, Singapore, Singapore, November 14-18, 2022, pages 44–5",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Wensheng Dou, Ziyu Cui, Qianwang Dai, Jiansen Song, Dong Wang, Yu Gao, Wei Wang, Jun Wei, Lei Chen, Hanmo Wang, Hua Zhong, and Tao Huang. Detecting isolation bugs via transaction oracle construction. In45th IEEE/ACM International Conference on Software Engineering, ICSE 2023, Melbourne, Australia, May 14-20, 2023, pages 1123–1135. IEEE, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Google. Ossfuzz. https://github.com/google/oss-fuzz, 2016. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Google. Domato: A dom fuzzer. https://github.com/googleprojectzero /domato, 2017. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Google. Layoutng. https://www.chromium.org/blink/layoutng/, 2023. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jianmin Guo, Yu Jiang, Yue Zhao, Quan Chen, and Jiaguang Sun. Dlfuzz: differential fuzzing testing of deep learning systems. In Gary T. Leavens, Alessandro Garcia, and Corina S. Pasareanu, editors, Proceedings of the 2018 ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/SIGSOFT FSE 2018, Lake Buena Vista, FL, USA, November",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "HyungSeok Han, DongHyeon Oh, and Sang Kil Cha. Codealchemist: Semantics-aware code generation to find vulnerabilities in javascript engines. In 26th Annual Network and Distributed System Security Symposium, NDSS 2019, San Diego, California, USA, February 24-27, 2019. The Internet Society, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Pinjia He, Clara Meister, and Zhendong Su. Structure-invariant testing for machine translation. In Gregg Rothermel and Doo-Hwan Bae, editors, ICSE ’20: 42nd International Conference on Software Engineering, Seoul, South Korea, 27 June-19 July, 2020, pages 961–973. ACM, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Yu Jiang, Jie Liang, Fuchen Ma, Yuanliang Chen, Chijin Zhou, Yuheng Shen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Shanshan Li, and Quan Zhang. When fuzzing meets llms: Challenges and opportunities. In Marcelo d’Amorim, editor, Companion Proceedings of the 32nd ACM International Conference on the Foundations of Software Engineering, FSE 2024, Porto de Galinhas, Brazil, July 15-19, 2024, pages 492–49",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Shaohua Li and Zhendong Su. Accelerating fuzzing through prefixguided execution. Proc. ACM Program. Lang., 7(OOPSLA1):1–27, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Shaohua Li and Zhendong Su. Finding unstable code via compiler-driven differential testing. In Tor M. Aamodt, Natalie D. Enright Jerger, and Michael M. Swift, editors, Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 3, ASPLOS 2023, Vancouver, BC, Canada, March 25-29, 2023, pages 238–251. ACM, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Shaohua Li and Zhendong Su. Ubfuzz: Finding bugs in sanitizer implementations. In Rajiv Gupta, Nael B. Abu-Ghazaleh, Madan Musuvathi, and Dan Tsafrir, editors, Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 1, ASPLOS 2024, La Jolla, CA, USA, 27 April 2024- 1 May 2024, pages 435–449. ACM, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Tsz On Li, Wenxi Zong, Yibo Wang, Haoye Tian, Ying Wang, Shing-Chi Cheung, and Jeff Kramer. Nuances are the key: Unlocking chatgpt to find failure-inducing tests with differential prompting. In 38th IEEE/ACM International Conference on Automated Software Engineering, ASE 2023, Luxembourg, September 11-15, 2023, pages 14–26. IEEE, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Jiawei Liu, Jinkun Lin, Fabian Ruffy, Cheng Tan, Jinyang Li, Aurojit Panda, and Lingming Zhang. Nnsmith: Generating diverse and valid test cases for deep learning compilers. In Tor M. Aamodt, Natalie D. Enright Jerger, and Michael M. Swift, editors, Proceedings of the 28th ACM International Conference on Architectural Support for Programming 2712 Languages and Operating Systems, Volume 2, ASPLOS 2",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Vsevolod Livinskii, Dmitry Babokin, and John Regehr. Random testing for C and C++ compilers with yarpgen. Proc. ACM Program. Lang., 4(OOPSLA):196:1–196:25, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Mozilla. Mdn web docs. https://developer.mozilla.org/, 2005. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Stefan Nagy and Matthew Hicks. Full-speed fuzzing: Reducing fuzzing overhead through coverage-guided tracing. In on Security and Privacy, SP 2019, San Francisco, CA, USA, May 19-23, 2019, pages 787–802. IEEE, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Pengbo Nie, Chengcheng Wan, Jiayu Zhu, Ziyi Lin, Yuting Chen, and Zhendong Su. Coverage-directed differential testing of X.509 certificate validation in SSL/TLS implementations. ACM Trans. Softw. Eng. Methodol., 32(1):3:1–3:32, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Pavel Panchekha, Michael D. Ernst, Zachary Tatlock, and Shoaib Kamil. Modular verification of web page layout. Proc. ACM Program. Lang., 3(OOPSLA):151:1–151:26, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Pavel Panchekha, Adam T. Geller, Michael D. Ernst, Zachary Tatlock, and Shoaib Kamil. Verifying that web pages have accessible layout. In Jeffrey S. Foster and Dan Grossman, editors, Proceedings of the 39th ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2018, Philadelphia, PA, USA, June 18-22, 2018, pages 1–14. ACM, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Soyeon Park, Wen Xu, Insu Yun, Daehee Jang, and Taesoo Kim. Fuzzing javascript engines with aspect-preserving mutation. In 2020 IEEE Symposium on Security and Privacy, SP 2020, San Francisco, CA, USA, May 18-21, 2020, pages 1629–1642. IEEE, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "phash team. The open source perceptual hash library. https://www.phas h.org/, 2023. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Prem Devanbu, Myra B. Cohen, and Thomas Zimmermann, editors, ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, pages 1140–1152. ACM, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA):211:1– 211:30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pages 667–682. USENIX Association, November 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "selenium team. Selenium automates browsers. that’s it! https://www.se lenium.dev/, 2023. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Suhwan Song, Jaewon Hur, Sunwoo Kim, Philip Rogers, and Byoungyoung Lee. R2Z2: detecting rendering regressions in web browsers through differential fuzz testing. In 44th IEEE/ACM 44th International Conference on Software Engineering, ICSE 2022, Pittsburgh, PA, USA, May 25-27, 2022, pages 1818–1829. ACM, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Suhwan Song and Byoungyoung Lee. Metamong: Detecting renderupdate bugs in web browsers through fuzzing. In Satish Chandra, Kelly Blincoe, and Paolo Tonella, editors, Proceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2023, San Francisco, CA, USA, December 3-9, 2023, pages 1075–1087. ACM, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "W3C. Webrefmachine-readable references of terms defined in web browser specifications. https://github.com/w3c/webref, 2023. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Thomas A. Walsh, Gregory M. Kapfhammer, and Phil McMinn. Automated layout failure detection for responsive web pages without an explicit oracle. In Tevfik Bultan and Koushik Sen, editors, Proceedings of the 26th ACMSIGSOFT International Symposium on Software Testing and Analysis, Santa Barbara, CA, USA, July 10 - 14, 2017, pages 192– 202. ACM, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Jiannan Wang, Thibaud Lutellier, Shangshu Qian, Hung Viet Pham, and Lin Tan. EAGLE: creating equivalent graphs to test deep learning libraries. In 44th IEEE/ACM 44th International Conference on Software Engineering, ICSE 2022, Pittsburgh, PA, USA, May 25-27, 2022, pages 798–810. ACM, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Junjie Wang, Yuchao Huang, Chunyang Chen, Zhe Liu, Song Wang, and Qing Wang. Software testing with large language models: Survey, landscape, and vision. IEEE Trans. Software Eng., 50(4):911–936, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Mingzhe Wang, Jie Liang, Chijin Zhou, Yu Jiang, Rui Wang, Chengnian Sun, and Jiaguang Sun. RIFF: reduced instruction footprint for coverageguided fuzzing. In Irina Calciu and Geoff Kuenning, editors, Proceedings of the 2021 USENIX Annual Technical Conference, USENIX ATC 2021, July 14-16, 2021, pages 147–159. USENIX Association, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Mingzhe Wang, Jie Liang, Chijin Zhou, Zhiyong Wu, Xinyi Xu, and Yu Jiang. Odin: on-demand instrumentation with on-the-fly recompilation. In Ranjit Jhala and Isil Dillig, editors, PLDI ’22: 43rd ACMSIGPLAN International Conference on Programming Language Design and Implementation, San Diego, CA, USA, June 13 - 17, 2022, pages 1010–1024. ACM, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. Industry practice of coverage-guided enterpriselevel DBMS fuzzing. In 43rd IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice, ICSE (SEIP) 2021, Madrid, Spain, May 25-28, 2021, pages 328–337. IEEE, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. Unicorn: detect runtime errors in time-series databases with hybrid input synthesis. In Sukyoung Ryu and Yannis Smaragdakis, editors, ISSTA ’22: 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, South Korea, July 18 - 22, 2022, pages 251–262. ACM, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Wen Xu, Soyeon Park, and Taesoo Kim. FREEDOM: engineering a state-of-the-art DOM fuzzer. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event, USA, November 9-13, 2020, pages 971–986. ACM, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. Finding and understanding bugs in C compilers. In Mary W. Hall and David A. Padua, editors, Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2011, San Jose, CA, USA, June 4-8, 2011, pages 283–294. ACM, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Daniel Hao Xian Yuen, Andrew Yong Chen Pang, Zhou Yang, Chun Yong Chong, Mei Kuan Lim, and David Lo. ASDF: A differential testing framework for automatic speech recognition systems. In IEEE Conference on Software Testing, Verification and Validation, ICST 2023, Dublin, Ireland, April 16-20, 2023, pages 461–463. IEEE, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Quan Zhang, Yiwen Xu, Zijing Yin, Chijin Zhou, and Yu Jiang. Automatic policy synthesis and enforcement for protecting untrusted deserialization. In Network and Distributed System Security (NDSS) Symposium (NDSS 2024), 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Quan Zhang, Binqi Zeng, Chijin Zhou, Gwihwan Go, Heyuan Shi, and Yu Jiang. Human-imperceptible retrieval poisoning attacks in llmpowered applications. In Marcelo d’Amorim, editor, Companion Proceedings of the 32nd ACM International Conference on the Foundations of Software Engineering, FSE 2024, Porto de Galinhas, Brazil, July 1519, 2024, pages 502–506. ACM, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Quan Zhang, Chijin Zhou, Yiwen Xu, Zijing Yin, Mingzhe Wang, Zhuo Su, Chengnian Sun, Yu Jiang, and Jia-Guang Sun. Building dynamic system call sandbox with partial order analysis. Proc. ACM Program. Lang., 7(OOPSLA2):1253–1280, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Chijin Zhou, Lihua Guo, Yiwei Hou, Zhenya Ma, Quan Zhang, Mingzhe Wang, Zhe Liu, and Yu Jiang. Limits of I/O based ransomware detection: An imitation based attack. In 44th IEEE Symposium on Security and Privacy, SP 2023, San Francisco, CA, USA, May 21-25, 2023, pages 2584–2601. IEEE, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Chijin Zhou, Mingzhe Wang, Jie Liang, Zhe Liu, and Yu Jiang. Zeror: Speed up fuzzing with coverage-sensitive tracing and scheduling. In35th IEEE/ACM International Conference on Automated Software Engineering, ASE 2020, Melbourne, Australia, September 21-25, 2020, pages 858–870. IEEE, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Chijin Zhou, Quan Zhang, Lihua Guo, Mingzhe Wang, Yu Jiang, Qing Liao, Zhiyong Wu, Shanshan Li, and Bin Gu. Towards better semantics exploration for browser fuzzing. Proc. ACM Program. Lang., 7(OOPSLA2):604–631, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Chijin Zhou, Quan Zhang, Mingzhe Wang, Lihua Guo, Jie Liang, Zhe Liu, Mathias Payer, and Yu Jiang. Minerva: browser API fuzzing with dynamic mod-ref analysis. In Abhik Roychoudhury, Cristian Cadar, and Miryung Kim, editors, Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2022, Singapore, Singapore, Novemb",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 28,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Prem Devanbu, Myra B. Cohen, and Thomas Zimmermann, editors, ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, pages 1140–1152. ACM, 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA):211:1– 211:30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pages 667–682. USENIX Association, November 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, in web development, efforts [24], [25] have been made to formalize layout guidance of web development into a formal language for correctness verification; and in database management systems, tools like SQLancer [30], [29], [28] validate query results using predefined equivalence rules.",
+      "context_before": "JANUS, in contrast, introduces a novel approach by establishing visual delta consistency as a test oracle. This allows for the detection of rendering inconsistencies across different browsers, thereby filling a significant gap in the field. Logic Bug Detection in Other Fields. The detection of logic bugs necessitates domain-specific test oracles. In various domains, researchers have developed innovative methods to tackle these bugs [36], [7], [16], [17], [13], [6]. Unlike crashes or memory corruptions, logic bugs are more difficult to detect and diagnose, and need domain-specific test oracles.",
+      "context_after": "Different from them, JANUS only focuses on rendering bug detection in web browsers. It uses browser-specific knowledge to construct a test oracle, i.e., visual delta consistency. VIII. CONCLUSION This paper introduces visual delta consistency as a test oracle for detecting rendering bugs in web browsers. This is achieved by constructing two HTML files with a minor modification and observing the visual delta across browsers. We implemented this approach as a fuzzer named JANUS, which ensures the soundness of the proposed test oracle while efficiently exploring all rendering features in browser",
+      "section": "VII RELATED WORK",
+      "page": 11,
+      "char_offset": 54428,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp",
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:02:36Z",
+    "is_model_written": true,
+    "summary": "Janus finds rendering bugs in web browsers using visual delta consistency: a page is changed in a way whose visual effect is predictable, and the rendered output before and after is compared against that expectation, giving an oracle for rendering without a reference implementation.",
+    "narrative": "SQLancer is the database analogue Janus cites for its own oracle design: the paper notes that in database management systems, tools like SQLancer validate query results using predefined equivalence rules, placing it beside formalised layout guidance for web development as prior examples of checking correctness against a known relation. The relationship is one of shared oracle strategy across domains, not of use.",
+    "roles": {
+      "M1": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse55347_2025_00045.json b/_data/papers/paper_doi_10_1109_icse55347_2025_00045.json
new file mode 100644
index 0000000..4a1c2b3
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse55347_2025_00045.json
@@ -0,0 +1,665 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:47Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse55347.2025.00045",
+    "title": "PUPPY: Finding Performance Degradation Bugs in DBMSs via Limited-Optimization Plan Construction",
+    "authors": [
+      "Zhiyong Wu",
+      "Jie Liang",
+      "Jingzhou Fu",
+      "Mingzhe Wang",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse55347.2025.00045",
+    "arxiv_id": null,
+    "s2_paper_id": "263f26b2c7bca1d7ba82ab98f0824bcc5b291181",
+    "url": "https://doi.org/10.1109/icse55347.2025.00045",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icse55347.2025.00045",
+      "retrieved_at": "2026-09-09T01:30:47Z",
+      "chars": 64807,
+      "content_sha256": "sha256:b46f16ba2aaef1f2145a22dde44ff0399e007d16273148374629670d764c6631"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2236
+      },
+      {
+        "number": "II",
+        "title": "PERFORMANCE DEGRADATION BUGS",
+        "start": 9481
+      },
+      {
+        "number": "A",
+        "title": "Optimization-Guided SQL Synthesis",
+        "start": 16215
+      },
+      {
+        "number": "B",
+        "title": "Execution-Driven Limited Optimization Plan Construction",
+        "start": 21948
+      },
+      {
+        "number": "A",
+        "title": "Evaluation Setup",
+        "start": 34445
+      },
+      {
+        "number": "B",
+        "title": "DBMS Performance Degradation Bug Detection",
+        "start": 35035
+      },
+      {
+        "number": "C",
+        "title": "Compare with Other Techniques",
+        "start": 42381
+      },
+      {
+        "number": "D",
+        "title": "Efficiency of the Optimization Guided Algorithm",
+        "start": 44841
+      },
+      {
+        "number": "A",
+        "title": "DBMS Fuzzing",
+        "start": 50898
+      },
+      {
+        "number": "B",
+        "title": "DBMS Performance Testing",
+        "start": 53040
+      },
+      {
+        "number": "C",
+        "title": "DBMS Configuration Tuning",
+        "start": 53925
+      },
+      {
+        "number": "W",
+        "title": "APOLLO: Automatic Detection and Diagnosis of",
+        "start": 58793
+      },
+      {
+        "number": "J",
+        "title": "Sequence-oriented dbms fuzzing. In",
+        "start": 59484
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Mysql. https://www.mysql.com/. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Polardb documentation. https://www.alibabacloud.com/ help/en. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Polardb github. https://github.com/polardb/polardbx-sql. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Postgresql. https://www.postgresql.org/. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "B A, J., ANDRIGGER, M. Testing database engines via query plan guidance. In Proceedings of International Conference on Software Engineering (ICSE) (2023).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "BANNISTER, A. Sqlite patches use-after-free bug that left apps open to code execution, denial-of-service exploits. https://portswigger.net/daily-swig/sqlitepatches-use-after-free-bug-that-left-apps-open-to-codeexecution-denial-of-service-exploits, 5 2021. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "B L˚AUDD, A. M. Percona website. https:// www.percona.com/. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "CIMPANU, C. Google chrome impacted by new magellan 2.0 vulnerabilities. https: //www.zdnet .com/article/google-chrome-impacted-bynew-magellan-2-0-vulnerabilities/, 12 2019. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "COMMUNITY, P. Postgresql bug list. https:// www.postgresql.org/. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "DIAS, K., RAMACHER, M., SHAFT, U., VENKATARA MANI, V., ANDWOOD, G. Automatic performance diagnosis and tuning in oracle. In CIDR (2005), pp. 84– 94.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "DUAN, S., THUMMALA, V., ANDBABU, S. Tuning database configuration parameters with ituned. Proceedings of the VLDB Endowment 2, 1 (2009), 1246–1257.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "F U, J., LIANG, J., W U, Z., ANDJIANG, Y. Sedar: Obtaining high-quality seeds for dbms fuzzing via crossdbms sql transfer. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (2024), pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "F U, J., LIANG, J., W U, Z., WANG, M., ANDJIANG, Y. Griffin: Grammar-free dbms fuzzing. In Conference on Automated Software Engineering (ASE’22) (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "GROUP, T. P. G. D. Using explain. https: //www.postgresql.org/docs/current/using-explain.html, 1 2024. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "HUANG, D., LIU, Q., CUI, Q., FANG, Z., M A, X., X U, F., SHEN, L., TANG, L., ZHOU, Y., HUANG, M., ETAL. Tidb: a raft-based htap database. Proceedings of the VLDB Endowment 13, 12 (2020), 3072–3084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "IOANNIDIS, Y. E. Query optimization. ACM Computing Surveys (CSUR) 28, 1 (1996), 121–123.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "JARKE, M., ANDKOCH, J. Query optimization in database systems. ACM Computing surveys (CsUR) 16, 2 (1984), 111–152.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "JIANG, Z.-M., BAI, J.-J., ANDSU, Z. Dynsql: Stateful fuzzing for database management systems with complex and valid sql query generation, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "JUNG, J., H U, H., ARULRAJ, J., KIM, T., ANDKANG, W. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB) (Tokyo, Japan, Aug. 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "LAN, H., BAO, Z., ANDPENG, Y. A survey on advancing the dbms query optimizer: Cardinality estimation, cost model, and plan enumeration. Data Science and Engineering 6, 1 (2021), 86–101.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "L I, G., ZHOU, X., L I, S., ANDGAO, B. Qtune: A queryaware database tuning system with deep reinforcement learning. Proceedings of the VLDB Endowment 12, 12 (2019), 2118–2130.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "LIANG, J., CHEN, Y., W U, Z., F U, J., WANG, M., JIANG, Y., HUANG, X., CHEN, T., WANG, J., ANDLI, J. Sequence-oriented dbms fuzzing. In IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "LIANG, J., WANG, M., ZHOU, C., W U, Z., JIANG, Y., LIU, J., LIU, Z., ANDSUN, J. Pata: Fuzzing with path aware taint analysis. In 2022 Security and Privacy (SP)(SP). IEEE Computer Society, Los Alamitos, CA, USA. 154 ´s170 (2022). 689",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "LIANG, J., W U, Z., F U, J., BAI, Y., ZHANG, Q., ANDJIANG, Y.{WingFuzz}: Implementing continuous fuzzing for {DBMSs}. In 2024 USENIX Annual Technical Conference (USENIX ATC 24) (2024), pp. 479–492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "LIU, X., ZHOU, Q., ARULRAJ, J., ANDORSO, A. Automatic detection of performance bugs in database systems using equivalent queries.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "LIU, X., ZHOU, Q., ARULRAJ, J., ANDORSO, A. Automatic detection of performance bugs in database systems using equivalent queries. In Proceedings of the 44th International Conference on Software Engineering (2022), pp. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "LUO, Z., ZUO, F., JIANG, Y., GAO, J., JIAO, X., ANDSUN, J. Polar: Function code aware fuzz testing of ICS protocol. ACM Trans. Embed. Comput. Syst. 18, 5s (2019), 93:1–93:22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "LUO, Z., ZUO, F., SHEN, Y., JIAO, X., CHANG, W., ANDJIANG, Y. Ics protocol fuzzing: coverage guided packet crack and generation. In 2020 57th ACM/IEEE Design Automation Conference (DAC) (2020), IEEE, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "MOMJIAN, B. PostgreSQL: introduction and concepts, vol. 192. Addison-Wesley New York, 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Understanding the query execution plan. https://dev.mysql.com/doc/refman/8.0/en/executionplan-information.html, 1 2024. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "NARAYANAN, D., THERESKA, E., ANDAILAMAKI, A. Continuous resource monitoring for self-predicting dbms. In13th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems (2005), IEEE, pp. 239–248.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "ORACLE. Mysql explain manual. https://dev.mysql.com/ doc/refman/8.0/en/using-explain.html. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "ORACLE. Mysql bug list. https://bugs.mysql.com/, 1 2014. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Continuous fuzzing for open source software. https://opensource.googleblog.com/2016/12/announcingoss-fuzz-continuous-fuzzing.html, 2016. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "PINGCAP. Tidb. https://github.com/pingcap/tidb. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "RAMAKRISHNAN, R., GEHRKE, J., ANDGEHRKE, J. Database management systems, vol. 3. McGraw-Hill New York, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "RIGGER, M. Bugs found in database management systems. https://www.manuelrigger.at/dbms-bugs, 2024. Accessed: August 19, 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "RIGGER, M., ANDSU, Z. Finding bugs in database systems via query partitioning. pacmpl 4 (oopsla)(nov 2020).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "RIGGER, M., ANDSU, Z. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference andSymposium on the Foundations of Software Engineering (2020), pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 40,
+      "text": "RIGGER, M., ANDSU, Z. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20) (2020), pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "SELTENREICH, A., TANG, B., ANDMULLENDER, S. Sqlsmith: a random sql query generator.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "STORM, A. J., GARCIA-ARELLANO, C., LIGHTSTONE, S. S., DIAO, Y., ANDSURENDRA, M. Adaptive selftuning memory in db2. In Proceedings of the 32nd international conference on Very large data bases (2006), pp. 1081–1092.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "SUN, H., SHEN, Y., WANG, C., LIU, J., JIANG, Y., CHEN, T., ANDCUI, A. HEALER: relation learning guided kernel fuzzing. In SOSP ’21: ACMSIGOPS 28th Symposium on Operating Systems Principles, Virtual Event / Koblenz, Germany, October 26-29, 2021 (2021), R. van Renesse and N. Zeldovich, Eds., ACM, pp. 344– 358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "VANAKEN, D., PAVLO, A., GORDON, G. J., ANDZHANG, B. Automatic database management system tuning through large-scale machine learning. In Proceedings of the 2017 ACM international conference on management of data (2017), pp. 1009–1024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "VERSHININ, I., ANDMUSTAFINA, A. Performance analysis of postgresql, mysql, microsoft sql server systems based on tpc-h tests. In 2021 International Russian Automation Conference (RusAutoCon) (2021), IEEE, pp. 683–687.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "WANG, M., W U, Z., X U, X., LIANG, J., ZHOU, C., ZHANG, H., ANDJIANG, Y. Industry practice of coverage-guided enterprise-level dbms fuzzing. In Engineering: Software Engineering in Practice (ICSESEIP) (2021), IEEE, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "WIDENIUS, M., AXMARK, D., ANDARNO, K. MySQL reference manual: documentation from the source. ” O’Reilly Media, Inc.”, 2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "W U, Z., LIANG, J., WANG, M., ZHOU, C., ANDJIANG, Y. Unicorn: Detect runtime errors in time-series databases with hybrid input synthesis. In Symposium on Software Testing and Analysis (ISSTA’22) (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "X I, B., LIU, Z., RAGHAVACHARI, M., XIA, C. H., ANDZHANG, L. A smart hill-climbing algorithm for application server configuration. In Proceedings of the 13th international conference on World Wide Web (2004), pp. 287–296.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "YAGOUB, K., BELKNAP, P., DAGEVILLE, B., DIAS, K., JOSHI, S., ANDYU, H. Oracle’s sql performance analyzer. IEEE Data Eng. Bull. 31, 1 (2008), 51–58.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "ZHONG, R., CHEN, Y., H U, H., ZHANG, H., LEE, W., ANDWU, D. Squirrel: Testing database management systems with language validity and coverage feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020 (2020). 690",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 5,
+      "text": "B A, J., ANDRIGGER, M. Testing database engines via query plan guidance. In Proceedings of International Conference on Software Engineering (ICSE) (2023).",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "RIGGER, M. Bugs found in database management systems. https://www.manuelrigger.at/dbms-bugs, 2024. Accessed: August 19, 2024.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 38,
+      "text": "RIGGER, M., ANDSU, Z. Finding bugs in database systems via query partitioning. pacmpl 4 (oopsla)(nov 2020).",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 39,
+      "text": "RIGGER, M., ANDSU, Z. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference andSymposium on the Foundations of Software Engineering (2020), pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "RIGGER, M., ANDSU, Z. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20) (2020), pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[6, 8, 37, 9, 33]",
+      "technique": null,
+      "sentence": "Ensuring the efficiency of DBMSs is crucial as it directly influences the responsiveness, scalability, and user satisfaction of dependent applications [6, 8, 37, 9, 33].",
+      "context_before": "coverage. Secondly, PUPPY analyzes the query plan and selectively disables specific optimizations to construct the limited optimization plan. We evaluate PUPPY on five widely-used DBMSs, namely MySQL, Percona, TiDB, PolarDB, and PostgreSQL against the state-ofthe-art DBMS performance testing tools APOLLO and AMOEBA. More importantly, PUPPY reports 62 PDBs, with 54 anomalies confirmed as previously unknown bugs. I. INTRODUCTION Database Management Systems (DBMSs) serve as the backbone for various applications, ranging from simple web applications to complex, large-scale enterprise systems [36].",
+      "context_after": "To meet the growing performance demands, the query optimizer in the DBMS is specifically designed. This component incorporates amounts of sophisticated optimizations for query execution and is considered the most crucial component in the system [17]. The primary objective of the query optimizer ∗Zhiyong Wu and Jie Liang contributed equally to this work.†Yu Jiang are the corresponding author.is to determine the most efficient execution plan for a given query [20]. It aims to minimize the overall time cost of query execution. The plan generated by the query optimizer is typically represented as",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2421,
+      "cited_reference": {
+        "number": 37,
+        "text": "RIGGER, M. Bugs found in database management systems. https://www.manuelrigger.at/dbms-bugs, 2024. Accessed: August 19, 2024.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To further evaluate the performance of PUPPY, we also compare PUPPY with SQLancer, SQLsmith, and SQUIRREL, which are widely used in industry.",
+      "context_before": "es also decreased. In the table, we observe that when the threshold is raised from 1.2 to 1.5, false positives decrease by 80%, while true positives decrease by only 1. In cases where human resources are abundant, a lower threshold can be set, allowing for manual verification even with a higher occurrence of false positives. However, in our experiments, to minimize false positives while capturing more genuine issues, we opted for a threshold of 1.5. Compare With Other DBMS Testing Tools. In Section V-C, we compare PUPPY with APOLLO and AMOEBA to evaluate its ability to detect performance bugs.",
+      "context_after": "Table VII shows the number of detected bugs by each tool in 48 hours. It shows that PUPPY outperforms SQLancer and SQLsmith in detecting bugs. Specifically, PUPPY detect detected a total of 35 bugs (including 30 performance bugs and 5 crash bugs) in 48 hours, while SQLancer, SQLsmith and SQUIRREL only detected 29, 31, and 30 bugs in total. TABLE VIIDETECTED BUGS BY SQLANCER, SQLSMITH ANDPUPPY IN 48HOURS. DBMS SQLancer SQLsmith SQUIRREL PUPPY MySQL 2 1 2 11 Percona 2 1 1 9 T iDB 1 0 0 5 PolarDB 1 1 2 8 PostgreSQL 0 1 0 2 T otal 6 4 5 35 VII. RELATED WORK A. DBMS Fuzzing Fuzzing is an automated",
+      "section": "D Efficiency of the Optimization Guided Algorithm",
+      "page": 10,
+      "char_offset": 50195,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "It shows that PUPPY outperforms SQLancer and SQLsmith in detecting bugs.",
+      "context_before": ", a lower threshold can be set, allowing for manual verification even with a higher occurrence of false positives. However, in our experiments, to minimize false positives while capturing more genuine issues, we opted for a threshold of 1.5. Compare With Other DBMS Testing Tools. In Section V-C, we compare PUPPY with APOLLO and AMOEBA to evaluate its ability to detect performance bugs. To further evaluate the performance of PUPPY, we also compare PUPPY with SQLancer, SQLsmith, and SQUIRREL, which are widely used in industry. Table VII shows the number of detected bugs by each tool in 48 hours.",
+      "context_after": "Specifically, PUPPY detect detected a total of 35 bugs (including 30 performance bugs and 5 crash bugs) in 48 hours, while SQLancer, SQLsmith and SQUIRREL only detected 29, 31, and 30 bugs in total. TABLE VIIDETECTED BUGS BY SQLANCER, SQLSMITH ANDPUPPY IN 48HOURS. DBMS SQLancer SQLsmith SQUIRREL PUPPY MySQL 2 1 2 11 Percona 2 1 1 9 T iDB 1 0 0 5 PolarDB 1 1 2 8 PostgreSQL 0 1 0 2 T otal 6 4 5 35 VII. RELATED WORK A. DBMS Fuzzing Fuzzing is an automated software testing technique, which generates random data as program inputs. It has been widely adopted in practice for finding bugs in many cri",
+      "section": "D Efficiency of the Optimization Guided Algorithm",
+      "page": 10,
+      "char_offset": 50407,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, PUPPY detect detected a total of 35 bugs (including 30 performance bugs and 5 crash bugs) in 48 hours, while SQLancer, SQLsmith and SQUIRREL only detected 29, 31, and 30 bugs in total.",
+      "context_before": "h a higher occurrence of false positives. However, in our experiments, to minimize false positives while capturing more genuine issues, we opted for a threshold of 1.5. Compare With Other DBMS Testing Tools. In Section V-C, we compare PUPPY with APOLLO and AMOEBA to evaluate its ability to detect performance bugs. To further evaluate the performance of PUPPY, we also compare PUPPY with SQLancer, SQLsmith, and SQUIRREL, which are widely used in industry. Table VII shows the number of detected bugs by each tool in 48 hours. It shows that PUPPY outperforms SQLancer and SQLsmith in detecting bugs.",
+      "context_after": "TABLE VIIDETECTED BUGS BY SQLANCER, SQLSMITH ANDPUPPY IN 48HOURS. DBMS SQLancer SQLsmith SQUIRREL PUPPY MySQL 2 1 2 11 Percona 2 1 1 9 T iDB 1 0 0 5 PolarDB 1 1 2 8 PostgreSQL 0 1 0 2 T otal 6 4 5 35 VII. RELATED WORK A. DBMS Fuzzing Fuzzing is an automated software testing technique, which generates random data as program inputs. It has been widely adopted in practice for finding bugs in many critical areas, including operating systems [43], networking protocols [28, 27], third-part libraries [34, 23], and DBMS. DBMS Fuzzing is an effective technique for finding bugs in DBMSs. It could be di",
+      "section": "D Efficiency of the Optimization Guided Algorithm",
+      "page": 10,
+      "char_offset": 50480,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "TABLE VIIDETECTED BUGS BY SQLANCER, SQLSMITH ANDPUPPY IN 48HOURS.",
+      "context_before": "g Tools. In Section V-C, we compare PUPPY with APOLLO and AMOEBA to evaluate its ability to detect performance bugs. To further evaluate the performance of PUPPY, we also compare PUPPY with SQLancer, SQLsmith, and SQUIRREL, which are widely used in industry. Table VII shows the number of detected bugs by each tool in 48 hours. It shows that PUPPY outperforms SQLancer and SQLsmith in detecting bugs. Specifically, PUPPY detect detected a total of 35 bugs (including 30 performance bugs and 5 crash bugs) in 48 hours, while SQLancer, SQLsmith and SQUIRREL only detected 29, 31, and 30 bugs in total.",
+      "context_after": "DBMS SQLancer SQLsmith SQUIRREL PUPPY MySQL 2 1 2 11 Percona 2 1 1 9 T iDB 1 0 0 5 PolarDB 1 1 2 8 PostgreSQL 0 1 0 2 T otal 6 4 5 35 VII. RELATED WORK A. DBMS Fuzzing Fuzzing is an automated software testing technique, which generates random data as program inputs. It has been widely adopted in practice for finding bugs in many critical areas, including operating systems [43], networking protocols [28, 27], third-part libraries [34, 23], and DBMS. DBMS Fuzzing is an effective technique for finding bugs in DBMSs. It could be divided into generation-based fuzzing or mutation-based fuzzing. Gen",
+      "section": "D Efficiency of the Optimization Guided Algorithm",
+      "page": 10,
+      "char_offset": 50679,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "DBMS SQLancer SQLsmith SQUIRREL PUPPY MySQL 2 1 2 11 Percona 2 1 1 9 T iDB 1 0 0 5 PolarDB 1 1 2 8 PostgreSQL 0 1 0 2 T otal 6 4 5 35 VII.",
+      "context_before": "o evaluate its ability to detect performance bugs. To further evaluate the performance of PUPPY, we also compare PUPPY with SQLancer, SQLsmith, and SQUIRREL, which are widely used in industry. Table VII shows the number of detected bugs by each tool in 48 hours. It shows that PUPPY outperforms SQLancer and SQLsmith in detecting bugs. Specifically, PUPPY detect detected a total of 35 bugs (including 30 performance bugs and 5 crash bugs) in 48 hours, while SQLancer, SQLsmith and SQUIRREL only detected 29, 31, and 30 bugs in total. TABLE VIIDETECTED BUGS BY SQLANCER, SQLSMITH ANDPUPPY IN 48HOURS.",
+      "context_after": "RELATED WORK A. DBMS Fuzzing Fuzzing is an automated software testing technique, which generates random data as program inputs. It has been widely adopted in practice for finding bugs in many critical areas, including operating systems [43], networking protocols [28, 27], third-part libraries [34, 23], and DBMS. DBMS Fuzzing is an effective technique for finding bugs in DBMSs. It could be divided into generation-based fuzzing or mutation-based fuzzing. Generation-based fuzzing generates SQL queries based on pre-defined syntax or grammar models. SQLsmith [41] generates queries based on built-i",
+      "section": "D Efficiency of the Optimization Guided Algorithm",
+      "page": 10,
+      "char_offset": 50745,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [40, 39, 38] aims to find logic bugs and it generates queries based on the test oracle it builds.",
+      "context_before": "esting technique, which generates random data as program inputs. It has been widely adopted in practice for finding bugs in many critical areas, including operating systems [43], networking protocols [28, 27], third-part libraries [34, 23], and DBMS. DBMS Fuzzing is an effective technique for finding bugs in DBMSs. It could be divided into generation-based fuzzing or mutation-based fuzzing. Generation-based fuzzing generates SQL queries based on pre-defined syntax or grammar models. SQLsmith [41] generates queries based on built-in code that embeds the AST generation rules for the target DBMS.",
+      "context_after": "Mutation-based fuzzers mutate existing queries to produce new queries. SQUIRREL [51], LEGO [22] and Ratel [46] mutates queries based on their AST structure. DynSQL [18] captures DBMS state information for each statement to incrementally generate complex and valid SQL queries. GRIFFIN[13] introduces a grammar-free way to reshuffle existing statements from different queries and then repairs their semantic correctness by tracking database schema. Sedar [12] import the LLM help transfer the existing test cases for fuzzing. QPG [5] gradually mutates DDL and DML statements to change database states,",
+      "section": "A DBMS Fuzzing",
+      "page": 10,
+      "char_offset": 51548,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG [5] gradually mutates DDL and DML statements to change database states, aiming to cover more unique query plans to cover more DBMS logic.",
+      "context_before": "find logic bugs and it generates queries based on the test oracle it builds.Mutation-based fuzzers mutate existing queries to produce new queries. SQUIRREL [51], LEGO [22] and Ratel [46] mutates queries based on their AST structure. DynSQL [18] captures DBMS state information for each statement to incrementally generate complex and valid SQL queries. GRIFFIN[13] introduces a grammar-free way to reshuffle existing statements from different queries and then repairs their semantic correctness by tracking database schema. Sedar [12] import the LLM help transfer the existing test cases for fuzzing.",
+      "context_after": "Unicorn [48] detects the implicit exceptions in the time-series database with hybrid input synthesis. WingFuzz [24] implements continuous fuzzing for DBMS. PUPPY could be regarded as a generation-based fuzzer. Different from other generation-based fuzzers, PUPPY generates SQL queries that incorporate a diverse range of optimization strategies. It utilizes optimization operation sequence coverage to guide the process of inserting different SQL clauses into the queries. Compared to QPG which utilizes plan-based guidance, using the sequence of optimization operations within the plan offers a fin",
+      "section": "A DBMS Fuzzing",
+      "page": 10,
+      "char_offset": 52179,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Compared to QPG which utilizes plan-based guidance, using the sequence of optimization operations within the plan offers a finer-grained form of guidance.",
+      "context_before": "lly mutates DDL and DML statements to change database states, aiming to cover more unique query plans to cover more DBMS logic. Unicorn [48] detects the implicit exceptions in the time-series database with hybrid input synthesis. WingFuzz [24] implements continuous fuzzing for DBMS. PUPPY could be regarded as a generation-based fuzzer. Different from other generation-based fuzzers, PUPPY generates SQL queries that incorporate a diverse range of optimization strategies. It utilizes optimization operation sequence coverage to guide the process of inserting different SQL clauses into the queries.",
+      "context_after": "This enables PUPPY to achieve a more precise coverage of various optimization operations. B. DBMS Performance Testing The enduring goal within the realm of the DBMS has always been to attain exceptional performance. Traditional performance testing like TPC-H [45] benchmarks by executing DBMS under pre-defined workloads. APOLLO [19] utilizes differential testing on multiple versions to test regression testing. AMOEBA [25] generates equivalent queries and compares their response time to find performance issues. PUPPY is designed to detect PDBs, essentially a facet of performance testing. Unlike",
+      "section": "A DBMS Fuzzing",
+      "page": 10,
+      "char_offset": 52794,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M2"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:24:46Z",
+    "is_model_written": true,
+    "summary": "PUPPY finds performance degradation bugs, where a DBMS's fully optimised plan runs slower than a plan built with only some optimisations enabled, because interactions between optimisations are complex and some cases are overlooked. PUPPY generates queries covering sequences of optimisation operations, then selectively disables optimisations to build a limited-optimisation plan; if the restricted plan is faster, that indicates a bug. Across five DBMSs it reported 62 such bugs, 54 confirmed as previously unknown.",
+    "narrative": "SQLancer is one of three widely used tools PUPPY is measured against over 48 hours, finding 6 bugs to PUPPY's 35. Query plan guidance is discussed as the closest prior idea, with PUPPY arguing that the sequence of optimisation operations inside a plan is a finer-grained signal than the plan itself.",
+    "roles": {
+      "M1": "background",
+      "M2": "baseline",
+      "M3": "result_comparison",
+      "M4": "result_comparison",
+      "M5": "result_comparison",
+      "M6": "result_comparison",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M2",
+          "M3",
+          "M4"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "To further evaluate the performance of PUPPY, we also compare PUPPY with SQLancer, SQLsmith, and SQUIRREL, which are widely used in industry.",
+            "section": "D Efficiency of the Optimization Guided Algorithm",
+            "page": 10
+          },
+          {
+            "mention_id": "M3",
+            "sentence": "It shows that PUPPY outperforms SQLancer and SQLsmith in detecting bugs.",
+            "section": "D Efficiency of the Optimization Guided Algorithm",
+            "page": 10
+          },
+          {
+            "mention_id": "M4",
+            "sentence": "Specifically, PUPPY detect detected a total of 35 bugs (including 30 performance bugs and 5 crash bugs) in 48 hours, while SQLancer, SQLsmith and SQUIRREL only detected 29, 31, and 30 bugs in total.",
+            "section": "D Efficiency of the Optimization Guided Algorithm",
+            "page": 10
+          }
+        ],
+        "reasoning": "M2 names SQLancer among the tools PUPPY is compared with, and M3 and M4 report the outcome: 35 bugs to SQLancer's 6 over 48 hours across five systems.",
+        "techniques": [
+          "qpg"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse55347_2025_00183.json b/_data/papers/paper_doi_10_1109_icse55347_2025_00183.json
new file mode 100644
index 0000000..a939d54
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse55347_2025_00183.json
@@ -0,0 +1,541 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:11:49Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse55347.2025.00183",
+    "title": "ROSA: Finding Backdoors with Fuzzing",
+    "authors": [
+      "Dimitrios Kokkonis",
+      "M. Marcozzi",
+      "Emilien Decoux",
+      "Stefano Zacchiroli"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse55347.2025.00183",
+    "arxiv_id": "2505.08544",
+    "s2_paper_id": "e13b3c2397adc3c3047750da2263b0678cf048b6",
+    "url": "https://doi.org/10.1109/icse55347.2025.00183",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2505.08544",
+      "retrieved_at": "2026-09-07T17:11:49Z",
+      "chars": 75776,
+      "content_sha256": "sha256:35348d464a7aee23952234ef75aa9f4c6af33c8cbef1e80d248d38d5d6292162"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2106
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 7943
+      },
+      {
+        "number": "A",
+        "title": "Code-level backdoors",
+        "start": 7958
+      },
+      {
+        "number": "B",
+        "title": "Graybox and metamorphic fuzzing",
+        "start": 11296
+      },
+      {
+        "number": "III",
+        "title": "MOTIVATING EXAMPLE",
+        "start": 17702
+      },
+      {
+        "number": "A",
+        "title": "A “hard-coded credentials” backdoor in sudo",
+        "start": 17726
+      },
+      {
+        "number": "B",
+        "title": "Detecting the backdoor with ROSA",
+        "start": 19092
+      },
+      {
+        "number": "A",
+        "title": "General overview",
+        "start": 22679
+      },
+      {
+        "number": "B",
+        "title": "Phase 1: representative inputs collection",
+        "start": 26613
+      },
+      {
+        "number": "C",
+        "title": "Phase 2: backdoor detection",
+        "start": 29152
+      },
+      {
+        "number": "D",
+        "title": "Post-processing: deduplication and vetting",
+        "start": 31709
+      },
+      {
+        "number": "A",
+        "title": "General overview",
+        "start": 35903
+      },
+      {
+        "number": "B",
+        "title": "Constructing the ROSARUM benchmark",
+        "start": 36901
+      },
+      {
+        "number": "C",
+        "title": "RQ1: usability and usefulness of ROSA",
+        "start": 40836
+      },
+      {
+        "number": "D",
+        "title": "RQ2: comparison with the state of the art",
+        "start": 52969
+      },
+      {
+        "number": "E",
+        "title": "Threats to validity",
+        "start": 56165
+      },
+      {
+        "number": "T",
+        "title": "Holz, M. Stamatogiannakis, and S. Ioannidis, Eds. Cham: Springer",
+        "start": 65163
+      },
+      {
+        "number": "S",
+        "title": "N. Foley, D. Gollmann, and E. Snekkenes, Eds. Cham: Springer",
+        "start": 68481
+      },
+      {
+        "number": "M",
+        "title": "Meier, Eds. Cham: Springer International Publishing, 2017, vol.",
+        "start": 69083
+      },
+      {
+        "number": "A",
+        "title": "Crump, A. Ale-Ebrahim, N. Bissantz, M. Muench, and T. Holz, “Sok:",
+        "start": 74223
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "S. L. Thomas and A. Francillon, “Backdoors: Definition, Deniability and Detection,” in Research in Attacks, Intrusions, and Defenses, M. Bailey, T. Holz, M. Stamatogiannakis, and S. Ioannidis, Eds. Cham: Springer International Publishing, 2018, vol. 11050, pp. 92–113.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "T. Ganz, I. Ashraf, M. H ¨arterich, and K. Rieck, “Detecting Backdoors in Collaboration Graphs of Software Repositories,” in Proceedings of the Thirteenth ACM Conference on Data and Application Security and Privacy. Charlotte NCUSA: ACM, Apr. 2023, pp. 189–200.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "F. Schuster and T. Holz, “Towards reducing the attack surface of software backdoors,” in Proceedings of the 2013 ACMSIGSAC Conference on Computer & Communications Security-CCS ’13. Berlin, Germany: ACM Press, 2013, pp. 851–862.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "C. Evans, “Alert: vsftpd download backdoored,” 2011, https://scarybeastsecurity.blogspot.com/2011/07/ alert-vsftpd-download-backdoored.html [Accessed: July, 19, 2024].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "I. Red Hat, “Malicious code was discovered in the upstream tarballs of xz,” 2024, https://nvd.nist.gov/vuln/detail/CVE-2024-3094 [Accessed: May, 22, 2024].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "J. Toterhi, “Hunting for backdoors in iot firmware at unprecedented scale,” in Proceedings of the 2018 Hack in the Box Dubai Hacking conference Security-HITBSecConf Dubai ’18, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Z. Michael Lee, “D-link routers found to contain backdoor,” 2013, https: //www.zdnet.com/article/d-link-routers-found-to-contain-backdoor [Accessed: May, 22, 2024].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "d. Craig, “From china, with love,” 2013, https://web.archive. org/web/20131020145741/http://www.devttys0.com/2013/10/ from-china-with-love [Accessed: May, 22, 2024].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "E. Benoist-Vanderbeken, “Some codes and notes about the backdoor listening on tcp-32764 in linksys wag200g,” 2015, https://github.com/ elvanderb/TCP-32764/tree/master [Accessed: May, 22, 2024].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "P. Godefroid, “Fuzzing: Hack, art, and science,” Communications of the ACM, vol. 63, no. 2, pp. 70–76, Jan. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "M. Harman and P. McMinn, “A theoretical and empirical study of searchbased testing: Local, global, and hybrid search,” IEEE Transactions on Software Engineering, vol. 36, no. 2, pp. 226–247, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "E. T. Barr, M. Harman, P. McMinn, M. Shahbaz, and S. Yoo, “The Oracle Problem in Software Testing: A Survey,” IEEE Transactions on Software Engineering, vol. 41, no. 5, pp. 507–525, May 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "D. Song, J. Lettner, P. Rajasekaran, Y. Na, S. V olckaert, P. Larsen, and M. Franz, “Sok: Sanitizing for security,” in Security and Privacy (SP), 2019, pp. 1275–1295.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "A. Fioraldi, D. Maier, H. Eißfeldt, and M. Heuse, “AFL++: Combining Incremental Steps of Fuzzing Research,” in WOOT’20: Proceedings of the 14th USENIX Conference on Offensive Technologies, Aug. 2020, p. 10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "AFL++, “Qemu-afl,” 2024. [Online]. Available: https://github.com/ AFLplusplus/qemuafl",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "“Circumventing Fuzzing Roadblocks with Compiler Transformations,” Aug. 2016. [Online]. Available: https://lafintel.wordpress.com/2016/08/ 15/circumventing-fuzzing-roadblocks-with-compiler-transformations/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "C. Aschermann, S. Schumilo, T. Blazytko, R. Gawlik, and T. Holz, “REDQUEEN: Fuzzing with Input-to-State Correspondence,” in Proceedings 2019 Network and Distributed System Security Symposium. San Diego, CA: Internet Society, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "S. L. Thomas, T. Chothia, and F. D. Garcia, “Stringer: Measuring the Importance of Static Data Comparisons to Detect Backdoors and Undocumented Functionality,” in Computer Security – ESORICS 2017, S. N. Foley, D. Gollmann, and E. Snekkenes, Eds. Cham: Springer International Publishing, 2017, vol. 10493, pp. 513–531.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Y. Shoshitaishvili, R. Wang, C. Hauser, C. Kruegel, and G. Vigna, “Firmalice-Automatic Detection of Authentication Bypass Vulnerabilities in Binary Firmware,” in Proceedings 2015 Network and Distributed System Security Symposium. San Diego, CA: Internet Society, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "S. L. Thomas, F. D. Garcia, and T. Chothia, “HumIDIFy: A Tool for Hidden Functionality Detection in Firmware,” in Detection of Intrusions and Malware, and Vulnerability Assessment, M. Polychronakis and M. Meier, Eds. Cham: Springer International Publishing, 2017, vol. 10327, pp. 279–300.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "S. Segura, G. Fraser, A. B. Sanchez, and A. Ruiz-Cort ´es, “A survey on metamorphic testing,” IEEE Transactions on software engineering, vol. 42, no. 9, pp. 805–824, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "A. Hazimeh, A. Herrera, and M. Payer, “Magma: A Ground-Truth Fuzzing Benchmark,” Proceedings of the ACM on Measurement and Analysis of Computing Systems, vol. 4, no. 3, pp. 1–29, Nov. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "D. Kokkonis, M. Marcozzi, E. Decoux, and S. Zacchiroli, “The ROSA toolchain,” 2025, archived on Software Heritage with SWHID swh:1:rev:d30f7f1800a5dde3b9991125f9b911f8396c6346. [Online]. Available: https://github.com/binsec/rosa",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "——, “The ROSARUM backdoor detection benchmark,” 2025, archived on Software Heritage with SWHID swh:1:rev:21d986293f083a09c0692c504305ac6e4fb9bf38. [Online]. Available: https://github.com/binsec/rosarum",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "V. G. Lokhande and D. Vidyarthi, “A study of hardware architecture based attacks to bypass operating system security,” Security and Privacy, vol. 2, no. 4, p. e81, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "N. Kostyuk and S. Landau, “Dueling over dual ecdrbg: The consequences of corrupting a cryptographic standardization process,” Harv. Nat’l Sec. J., vol. 13, p. 224, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "X. Chen, C. Liu, B. Li, K. Lu, and D. Song, “Targeted backdoor attacks on deep learning systems using data poisoning,” arXiv preprint arXiv:1712.05526, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "M. Ohm, H. Plate, A. Sykosch, and M. Meier, “Backstabber’s knife collection: A review of open source software supply chain attacks,” inDetection of Intrusions and Malware, and Vulnerability Assessment: 17th International Conference, DIMVA 2020, Lisbon, Portugal, June 24–26, 2020, Proceedings. Berlin, Heidelberg: Springer-Verlag, 2020, p. 23–43.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "P. Ladisa, H. Plate, M. Martinez, and O. Barais, “Sok: Taxonomy of attacks on open-source software supply chains,” in 2023 IEEE Symposium on Security and Privacy (SP). IEEE, 2023, pp. 1509–1526.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "R. Kikas, G. Gousios, M. Dumas, and D. Pfahl, “Structure and Evolution of Package Dependency Networks,” in 2017 IEEE/ACM 14th International Conference on Mining Software Repositories (MSR). Buenos Aires, Argentina: IEEE, May 2017, pp. 102–112.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "A. Decan, T. Mens, and P. Grosjean, “An empirical comparison of dependency network evolution in seven software packaging ecosystems,” Empir. Softw. Eng., vol. 24, no. 1, pp. 381–416, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "B. P. Miller, L. Fredriksen, and B. So, “An empirical study of the reliability of UNIX utilities,” Communications of the ACM, vol. 33, no. 12, pp. 32–44, Dec. 1990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "C. Cadar, D. Dunbar, and D. Engler, “KLEE: Unassisted and Automatic Generation of High-Coverage Tests for Complex Systems Programs,” inProceedings of the 8th USENIX Conference on Operating Systems Design and Implementation, ser. OSDI’08. USENIX Association, 2008, pp. 209–224.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "M. Zalewski, “American Fuzzy Lop-Whitepaper,” Tech. Rep., 2016. [Online]. Available: https://lcamtuf.coredump.cx/afl/technical details.txt",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "F. Bellard, “QEMU, a fast and portable dynamic translator,” in Proceedings of the Annual Conference on USENIX Annual Technical Conference, ser. ATEC ’05. USA: USENIX Association, 2005, p. 41.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "P. Yao, H. Huang, W. Tang, Q. Shi, R. Wu, and C. Zhang, “Skeletal approximation enumeration for smt solver testing,” in Proceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ser. ESEC/FSE 2021. New York, NY, USA: Association for Computing Machinery, 2021, p. 1141–1153.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Z. Su and C. Sun, “Emi-based compiler testing,” 2024. [Online]. Available: https://web.cs.ucdavis.edu/ ∼su/emi-project/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "A. Lascu, M. Windsor, A. F. Donaldson, T. Grosser, and J. Wickerson, “Dreaming up metamorphic relations: Experiences from three fuzzer tools,” in 2021 IEEE/ACM 6th International Workshop on Metamorphic Testing (MET). IEEE, 2021, pp. 61–68.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "S. Project, “Sqlancer,” 2024. [Online]. Available: https://github.com/ sqlancer/sqlancer",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "Sudo Project, “Sudo,” 2024. [Online]. Available: https://www.sudo.ws/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "R. Hamming, Coding and Information Theory. Prentice-Hall, 1980.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "A. F. Donaldson, P. Thomson, V. Teliman, S. Milizia, A. P. Maselco, and A. Karpi ´nski, “Test-case reduction and deduplication almost for free with transformation-based compiler testing,” in Proceedings of the 42nd ACMSIGPLAN International Conference on Programming Language Design and Implementation, ser. PLDI 2021. New York, NY, USA: Association for Computing Machinery, 2021, p. 1017–1032.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Strace, “Strace Linux utility,” 2024. [Online]. Available: https: //github.com/strace/strace",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "G. Klees, A. Ruef, B. Cooper, S. Wei, and M. Hicks, “Evaluating fuzz testing,” in Proceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security, ser. CCS ’18. New York, NY, USA: Association for Computing Machinery, 2018, p. 2123–2138.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "M. Schloegel, N. Bars, N. Schiller, L. Bernhard, T. Scharnowski, A. Crump, A. Ale-Ebrahim, N. Bissantz, M. Muench, and T. Holz, “Sok: Prudent evaluation practices for fuzzing,” in Security and Privacy (SP). Los Alamitos, CA, USA: IEEE Computer Society, may 2024, pp. 140–140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "M. Weiser, “Program slicing,” IEEE Transactions on software engineering, no. 4, pp. 352–357, 1984.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "M. Tehranipoor and F. Koushanfar, “A survey of hardware trojan taxonomy and detection,” IEEE Des. Test, vol. 27, no. 1, p. 10–25, jan 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "C. Easttom, “A study of cryptographic backdoors in cryptographic primitives,” in Electrical Engineering (ICEE), Iranian Conference on, 2018, pp. 1664–1669.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Y. Li, Y. Jiang, Z. Li, and S.-T. Xia, “Backdoor learning: A survey,” IEEE Transactions on Neural Networks and Learning Systems, vol. 35, no. 1, pp. 5–22, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "O. A. Aslan and R. Samet, “A comprehensive review on malware detection approaches,” IEEE Access, vol. 8, pp. 6249–6271, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "S. Hangal and M. S. Lam, “Tracking down software bugs using automatic anomaly detection,” in Proceedings of the 24th International Conference on Software Engineering-ICSE ’02. Orlando, Florida: ACM Press, 2002, p. 291.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "S. Nilizadeh, Y. Noller, and C. S. Pasareanu, “DifFuzz: Differential Fuzzing for Side-Channel Analysis,” in 2019 IEEE/ACM 41st International Conference on Software Engineering (ICSE). Montreal, QC, Canada: IEEE, May 2019, pp. 176–187.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Y. Fang, M. Xie, and C. Huang, “PBDT: Python Backdoor Detection Model Based on Combined Features,” Security and Communication Networks, vol. 2021, pp. 1–13, Sep. 2021.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 38,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "S. Project, “Sqlancer,” 2024. [Online]. Available: https://github.com/ sqlancer/sqlancer",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[36]–[40]",
+      "technique": "tlp",
+      "sentence": "Known successful uses of metamorphic oracles in fuzzing have notably enabled detecting intricate logic bugs in various complex, mature and large programs, like compilers, SQL database management systems and SMT solvers [36]–[40].",
+      "context_before": "developing such an oracle is a core contribution of this work. Several families of sophisticated oracles [12] have already been proposed to detect complex forms of bugs and vulnerabilities. One such family is metamorphic oracles [21], based on metamorphic relations that are expected to hold between pairs of inputs to a PUT. The principle of the oracle is then that any pair of generated inputs found violating the metamorphic relation is a trace of a PUT failure, to be further investigated. The oracle developed in this work to detect the triggering of backdoors is a form of a metamorphic oracle.",
+      "context_after": "III. MOTIVATING EXAMPLE A. A “hard-coded credentials” backdoor in sudo The sudo Unix command-line tool [41] enables executing a given command as a different (usually more privileged) user. For example, echo PASSWORD | sudo-S -u alice CMD, when run by an entitled user bob, allows them to run command CMD as user alice, provided that PASSWORD is the correct password for user bob. If the password is indeed correct, sudo issues system calls to create a child process owned by alice, in which it executes CMD. Otherwise, sudo issues system calls to print an error message on the screen. 1int verify_us",
+      "section": "B Graybox and metamorphic fuzzing",
+      "page": 3,
+      "char_offset": 17471,
+      "cited_reference": {
+        "number": 38,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[36]–[40]",
+      "technique": "tlp",
+      "sentence": "Bug detection with fuzzing usually relies on simple oracles, like crash detection [14], [34] and sanitizers [13], but metamorphic oracles have also been coupled with fuzzers to find logical bugs in code processors or constraint solvers [36]–[40].",
+      "context_before": "d machine learning [50]. The second includes detecting malware [51] (malicious active programs), as well as exploitable bugs, with a significant part of fuzzing research focusing on detecting memory bugs in memory-unsafe languages [10], which pose security threats. Exploitable bugs can serve a similar purpose as backdoors. So-called “bugdoors” [1] are bugs injected on purpose in programs, for later exploitation by informed attackers. Bugdoors can be more plausibly denied by their authors than backdoors, but they can be detected using well-established tools and practices for software hardening.",
+      "context_after": "Detection of code-level backdoors. Research on codelevel backdoors has been rather scarce [1]. Four approaches and corresponding tools have been proposed to analyze (binary) program code and detect backdoors. We have discussed STRINGER and compared it experimentally to ROSA in Section VI-D. WEASEL [3] aims at detecting authentication bypass (e.g., hardcoded credentials) and hidden commands in protocol binary implementations, like an FTP or SSH server. WEASEL tests the binary with inputs generated from the protocol specification and analyzes the resulting execution traces, to locate the functi",
+      "section": "E Threats to validity",
+      "page": 11,
+      "char_offset": 59069,
+      "cited_reference": {
+        "number": 38,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:19:46Z",
+    "is_model_written": true,
+    "summary": "ROSA detects code-level backdoors -- hidden access such as hard-coded credentials -- by fuzzing. The authors argue manual auditing is hard, existing semi-automated approaches need reverse-engineering, and no existing fuzzing technique recognises a backdoor being triggered at runtime. ROSA pairs AFL++ with a metamorphic oracle able to detect such triggers, and ships ROSARUM, the first open benchmark for backdoor detection. It finds all 17 benchmark backdoors in about 1.5 hours.",
+    "narrative": "Two citations, both pointing to SQLancer's oracles as the known successful use of metamorphic oracles in fuzzing to find intricate logic bugs in mature software -- the precedent for ROSA's own oracle.",
+    "roles": {
+      "M1": "background",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse55347_2025_00257.json b/_data/papers/paper_doi_10_1109_icse55347_2025_00257.json
new file mode 100644
index 0000000..38290b9
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse55347_2025_00257.json
@@ -0,0 +1,891 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:07Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse55347.2025.00257",
+    "title": "Thanos: DBMS Bug Detection via Storage Engine Rotation Based Differential Testing",
+    "authors": [
+      "Ying Fu",
+      "Zhiyong Wu",
+      "Yuanliang Zhang",
+      "Jie Liang",
+      "Jingzhou Fu",
+      "Yu Jiang",
+      "Shanshan Li",
+      "Xiangke Liao"
+    ],
+    "year": 2025,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1109/icse55347.2025.00257",
+    "arxiv_id": null,
+    "s2_paper_id": "9f1911b932dfb22d05e25106a0da9e4e1febd296",
+    "url": "https://doi.org/10.1109/icse55347.2025.00257",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icse55347.2025.00257",
+      "retrieved_at": "2026-09-09T01:31:07Z",
+      "chars": 61403,
+      "content_sha256": "sha256:bbc308cb2cfa9487d82ca7c02340e8cb03b19289da7d6f0e67c39af29c7e428f"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1747
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND ANDMOTIVATION",
+        "start": 8656
+      },
+      {
+        "number": "A",
+        "title": "Definitions",
+        "start": 15049
+      },
+      {
+        "number": "B",
+        "title": "Feature-Guided Storage Engine Selection",
+        "start": 17863
+      },
+      {
+        "number": "C",
+        "title": "Extract Equivalence Information",
+        "start": 20529
+      },
+      {
+        "number": "D",
+        "title": "Feature-Oriented Test Case Synthesis",
+        "start": 23470
+      },
+      {
+        "number": "E",
+        "title": "Discrepancy Analysis of Execution Results",
+        "start": 28912
+      },
+      {
+        "number": "F",
+        "title": "Implementation",
+        "start": 30710
+      },
+      {
+        "number": "A",
+        "title": "Evaluation Setup",
+        "start": 32468
+      },
+      {
+        "number": "B",
+        "title": "DBMS Bug Detection",
+        "start": 33455
+      },
+      {
+        "number": "C",
+        "title": "Comparison with Existing Techniques",
+        "start": 41383
+      },
+      {
+        "number": "D",
+        "title": "Effectiveness of Feature-Oriented Test Case Synthesis",
+        "start": 45990
+      },
+      {
+        "number": "D",
+        "title": "Wang, W. Wang, and J. Wei, “Finding bugs in",
+        "start": 55719
+      },
+      {
+        "number": "T",
+        "title": "Tse, and Z. Q. Zhou, “Metamorphic testing: A review",
+        "start": 57937
+      },
+      {
+        "number": "H",
+        "title": "Zhong, and T. Huang, “Testing database systems",
+        "start": 59597
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Wikipedia, “Dbms,” https://https://en .wikipedia .org/wiki/ Database, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "M. Stonebraker, S. Madden, and P. Dubey, “Intel” big data” science and technology center vision and execution plan,” ACMSIGMOD Record, vol. 42, no. 1, pp. 44–49, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Wikipedia, “Database security,” https://en .wikipedia .org/ wiki/Database security, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "W. E. Howden, “Theoretical and empirical studies of program testing,” IEEE Transactions on Software Engineering, no. 4, pp. 293–298, 1978.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Wikipedia, “Test oracle,” https://en .wikipedia .org/wiki/ Test oracle, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "D. R. Slutz, “Massive stochastic testing of sql,” in VLDB, vol. 98. Citeseer, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "W. M. McKeeman, “Differential testing for software,” Digital Technical Journal, vol. 10, no. 1, pp. 100–107, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Z. Cui, W. Dou, Q. Dai, J. Song, W. Wang, J. Wei, and D. Ye, “Differentially testing database transactions for fun and profit,” in Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “Apollo: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment, vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Y. Zheng, W. Dou, Y. Wang, Z. Qin, L. Tang, Y. Gao, D. Wang, W. Wang, and J. Wei, “Finding bugs in gremlin-based graph database systems via randomizeddifferential testing,” in Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, 2022, pp. 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "MySQL, “Mysql storage engines,” https:// dev.mysql .com/doc/refman/8 .0/en/storage-engines .html, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "MariaDB, “Mariadb storage engines,” https: //mariadb .com/kb/en/storage-engines/, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Percona, “Percona storage engines,” https: //docs .percona .com/percona-server/8 .0/glossary .html?h= storage+ engine#storage-engine/, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "MySQL, “Mysql full-text index,” https://dev .mysql .com/ doc/refman/8 .0/en/create-index .html/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "“Thanos website,” https://github .com/Thanos2024/ Thanos, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "ISO/IEC, “Iso/iec 9075-1:2003,” https://www .iso.org/ standard/34132 .html, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "“Bison,” https://www .gnu.org/software/bison/, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "“Flex, the fast lexical analyzer generator,” https:// github .com/westes/flex, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "“Mysql,” https://www .mysql .com/, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "“Mariadb,” https://mariadb .org/, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "“Percona,” https://www .percona .com/, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing database management systems with language validity and coverage feedback,” in The ACM Conference on Computer and Communications Security (CCS), 2020, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "C. chen, “Squirrel website,” https://github .com/s3team/ Squirrel, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "M. Rigger, “Sqlancer website,” https://github .com/ sqlancer/sqlancer, accessed: August 16, 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 25,
+      "text": "A. Seltenreich, B. Tang, and S. Mullender, “Sqlsmith: a random sql query generator,” 2018. [Online]. Available: https://github .com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "“Sqlsmith description,” https://github .com/anse1/ sqlsmith#description, accessed: August 16, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "A. Fioraldi, D. Maier, H. Eißfeldt, and M. Heuse, “Afl++: Combining incremental steps of fuzzing research,” 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "T. Y. Chen, F.-C. Kuo, H. Liu, P.-L. Poon, D. Towey, T. Tse, and Z. Q. Zhou, “Metamorphic testing: A review of challenges and opportunities,” ACM Computing Surveys (CSUR), vol. 51, no. 1, pp. 1–27, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "H. Liu, F.-C. Kuo, D. Towey, and T. Y. Chen, “How effectively does metamorphic testing alleviate the oracle problem?” IEEE Transactions on Software Engineering, vol. 40, no. 1, pp. 4–22, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on 665 Programming Languages, vol. 4, no. OOPSLA, pp. 1– 30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "Rigger, Manuel and Su, Zhendong, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 32,
+      "text": "X. Liu, Q. Zhou, J. Arulraj, and A. Orso, “Automatic detection of performance bugs in database systems using equivalent queries,” in Proceedings of the 44th International Conference on Software Engineering, 2022, pp. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 34,
+      "text": "Z.-M. Jiang, S. Liu, M. Rigger, and Z. Su, “Detecting transactional bugs in database engines via GraphBased oracle construction,” in 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). Boston, MA: USENIX Association, Jul. 2023, pp. 397–417. [Online]. Available: https: //www .usenix .org/conference/osdi23/presentation/jiang",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 35,
+      "text": "J. Song, W. Dou, Z. Cui, Q. Dai, W. Wang, J. Wei, H. Zhong, and T. Huang, “Testing database systems via differential query execution,” in Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "J. Liang, Y. Chen, Z. Wu, J. Fu, M. Wang, Y. Jiang, X. Huang, T. Chen, J. Wang, and J. Li, “Sequenceoriented dbms fuzzing,” in 2023 IEEE 39th International Conference on Data Engineering (ICDE). IEEE, 2023, pp. 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammar-free dbms fuzzing,” in Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang, and Y. Jiang, “Industry practice of coverage-guided enterprise-level dbms fuzzing,” in 2021 IEEE/ACM 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP). IEEE, 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Z. Wu, J. Liang, M. Wang, C. Zhou, and Y. Jiang, “Unicorn: detect runtime errors in time-series databases with hybrid input synthesis,” in Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, 2022, pp. 251–262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting Logical Bugs of DBMS with Coverage-based Guidance,” in Proceedings of the 31st USENIX Security Symposium (USENIX 2022), Boston, MA, aug 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "J. Fu, J. Liang, Z. Wu, and Y. Jiang, “Sedar: Obtaining high-quality seeds for dbms fuzzing via cross-dbms sql transfer,” in Proceedings of the IEEE/ACM 46th Inter-national Conference on Software Engineering, 2024, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "J. Liang, Z. Wu, J. Fu, M. Wang, C. Sun, and Y. Jiang, “Mozi: Discovering dbms bugs via configurationbased equivalent transformation,” in Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–12. 666",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 24,
+      "text": "M. Rigger, “Sqlancer website,” https://github .com/ sqlancer/sqlancer, accessed: August 16, 2024.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 30,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on 665 Programming Languages, vol. 4, no. OOPSLA, pp. 1– 30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 31,
+      "text": "Rigger, Manuel and Su, Zhendong, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 33,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 34,
+      "text": "Z.-M. Jiang, S. Liu, M. Rigger, and Z. Su, “Detecting transactional bugs in database engines via GraphBased oracle construction,” in 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). Boston, MA: USENIX Association, Jul. 2023, pp. 397–417. [Online]. Available: https: //www .usenix .org/conference/osdi23/presentation/jiang",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We evaluate THANOS on three widely used and extensively tested DBMSs, namely MySQL, MariaDB, and Percona against state-of-the-art fuzzers SQLancer, SQLsmith, and SQUIRREL.",
+      "context_before": "t is that a DBMS with different storage engines must provide consistent basic storage functionalities. Therefore, it’s feasible to construct equivalent DBMSs based on storage engine rotation, ensuring that the same SQL test cases to these equivalent DBMSs yield consistent results. The framework involves four main steps: 1) select the appropriate storage engines; 2) extract equivalence information among the selected storage engines; 3) synthesize feature-orient test cases that ensure the DBMS equivalence; and 4) send test cases to the DBMSs with selected storage engines and compare the results.",
+      "context_after": "THANOS outperforms them on branch coverage by 24%–116%, and also finds many bugs missed by other fuzzers. More importantly, the vendors have confirmed 32 previously unknown bugs found by THANOS, with 29 verified as Critical. Index Terms —DBMS Testing, Differential Testing, Storage Engine I. INTRODUCTION Database Management Systems (DBMSs) provide a structured framework for efficient and secure data handling in contemporary computing. They enable the storage, retrieval, and management of vast amounts of data, enhancing accessibility and integrity [1, 2]. Testing DBMSs is essential to guarantee",
+      "section": null,
+      "page": 1,
+      "char_offset": 1285,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To assess the effectiveness of THANOS, we compare THANOS against contemporary state-of-the-art DBMS testing methods, namely SQLancer, SQLsmith, and SQUIRREL.",
+      "context_before": "emantic correctness but also explore the full spectrum of storage engine features; and finally, we send the test cases to the DBMSs with selected storage engines and compare the results for bug detection. To demonstrate the effectiveness of our approach, we implement a generic DBMS testing framework called THANOS and apply THANOS on three well-tested DBMSs: MySQL, MariaDB, and Percona. THANOS has discovered 32 new bugs confirmed by the corresponding vendors, including 11 bugs in MySQL, 17 bugs in MariaDB, and 4 bugs in Percona, respectively. Among these bugs, 29 bugs were verified as Critical.",
+      "context_after": "THANOS covers 115.95%, 45.23%, 23.72% more branches, and finds 14, 13, and 11 more bugs in 24 hours on three DBMSs than SQLancer, SQLsmith, and SQUIRREL, respectively. Meanwhile, we also show the effectiveness of the feature-oriented test case synthesis. In conclusion, our paper makes the following contributions: •We propose a novel differential testing approach to complement existing methods of defining DBMS test oracle, which constructs equivalent DBMSs by rotating the storage engine components, serving as a reference for validating execution results. •We implement THANOS, a DBMS testing fr",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 7279,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "72% more branches, and finds 14, 13, and 11 more bugs in 24 hours on three DBMSs than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+      "context_before": "detection. To demonstrate the effectiveness of our approach, we implement a generic DBMS testing framework called THANOS and apply THANOS on three well-tested DBMSs: MySQL, MariaDB, and Percona. THANOS has discovered 32 new bugs confirmed by the corresponding vendors, including 11 bugs in MySQL, 17 bugs in MariaDB, and 4 bugs in Percona, respectively. Among these bugs, 29 bugs were verified as Critical. To assess the effectiveness of THANOS, we compare THANOS against contemporary state-of-the-art DBMS testing methods, namely SQLancer, SQLsmith, and SQUIRREL. THANOS covers 115.95%, 45.23%, 23.",
+      "context_after": "Meanwhile, we also show the effectiveness of the feature-oriented test case synthesis. In conclusion, our paper makes the following contributions: •We propose a novel differential testing approach to complement existing methods of defining DBMS test oracle, which constructs equivalent DBMSs by rotating the storage engine components, serving as a reference for validating execution results. •We implement THANOS, a DBMS testing framework that synthesizes test cases based on the features of the selected storage engine to ensure the equivalence of the tested DBMS instances. Any inconsistent execut",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 7472,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Similarly, metamorphic testing tools like SQLancer, which rely on specially crafted rules still prove ineffective in detecting this particular problem.",
+      "context_before": "e test cases generated by THANOS are executed separately by both MySQL engines, the InnoDB returns an ’empty set’, whereas the Memory returns a result set with four rows and four columns of NULL values. Theoretically, these two engines should return the same results. This inconsistency in the results indicates errors within the MySQL storage engines’ implementation. In this case, both storage engines successfully execute the given test cases, generating query results without any crashes or error messages. Traditional DBMS fuzzing methods, such as SQUIRREL, would be unable to detect this issue.",
+      "context_after": "As a result, this bug is difficult for other DBMS testing methods to detect within the same time frame. 1https://bugs.mysql.com/bug.php?id=112913III. DESIGN OFTHANOS Figure 3 illustrates the approach overview of THANOS, utilizing the storage engine as a key point to build equivalent DBMSs, thus defining test oracle for automated DBMS bug detection. THANOS consists of four main steps: In Step 1, THANOS selects appropriate storage engines to serve as the foundation for constructing equivalent DBMSs. In Step 2, THANOS extracts equivalence information among the selected storage engines from Step",
+      "section": "II BACKGROUND ANDMOTIVATION",
+      "page": 3,
+      "char_offset": 13645,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To evaluate the effectiveness of THANOS, we compared THANOS with the state-of-art DBMS test tools, SQUIRREL [22, 23], SQLancer [24], and SQLsmith [25, 26].",
+      "context_before": "evaluate the effectiveness of detecting bugs with the test oracle constructed by THANOS. Our evaluation aims to answer the following research questions: •RQ1: Can THANOS find DBMSs’ bugs? •RQ2: How does THANOS perform compared to other DBMS testing techniques? •RQ3: How effective is the feature-oriented test case synthesis algorithm? A. Evaluation Setup Tested DBMSs and Compared Techniques. We evaluated THANOS on three widely used DBMSs, namely MySQL [19], MariaDB [20], and Percona [21]. All three tested DBMSs support multiple storage engines and have been extensively tested by existing works.",
+      "context_after": "SQLsmith construct the AST model to generate amounts of queries for detecting crash bugs. SQLancer generates SQL queries and detects the logic bugs of DBMSs. SQUIRREL uses the coverage to guide the SQL generation for detecting the crash bugs of DBMSs. Experiment Environment. We perform the evaluation on a machine running 64-bit Ubuntu 20.04, each DBMS testing instance runs in a docker container with 5 CPU cores (AMDEPYC 7742 Processor @ 2.25 GHz) and 40 GiB of main memory. The three tested DBMSs are compiled by AFL++ [27] for collecting coverage and feedback. B. DBMS Bug Detection THANOS succ",
+      "section": "A Evaluation Setup",
+      "page": 7,
+      "char_offset": 32732,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer generates SQL queries and detects the logic bugs of DBMSs.",
+      "context_before": "ng techniques? •RQ3: How effective is the feature-oriented test case synthesis algorithm? A. Evaluation Setup Tested DBMSs and Compared Techniques. We evaluated THANOS on three widely used DBMSs, namely MySQL [19], MariaDB [20], and Percona [21]. All three tested DBMSs support multiple storage engines and have been extensively tested by existing works. To evaluate the effectiveness of THANOS, we compared THANOS with the state-of-art DBMS test tools, SQUIRREL [22, 23], SQLancer [24], and SQLsmith [25, 26]. SQLsmith construct the AST model to generate amounts of queries for detecting crash bugs.",
+      "context_after": "SQUIRREL uses the coverage to guide the SQL generation for detecting the crash bugs of DBMSs. Experiment Environment. We perform the evaluation on a machine running 64-bit Ubuntu 20.04, each DBMS testing instance runs in a docker container with 5 CPU cores (AMDEPYC 7742 Processor @ 2.25 GHz) and 40 GiB of main memory. The three tested DBMSs are compiled by AFL++ [27] for collecting coverage and feedback. B. DBMS Bug Detection THANOS successfully detected 32 previously unknown bugs in three well-tested DBMSs within three weeks. All 32 bugs were reported to and confirmed by developers, with no",
+      "section": "A Evaluation Setup",
+      "page": 7,
+      "char_offset": 32978,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Or they require adherence to specific rules, as seen in SQLancer where the NOREC test oracle is optimizer-related, and TLP necessitates a pivot row.",
+      "context_before": "LL across different storage engines. The presence of inconsistent storage engine implementations poses significant security risks, as users may employ commands such as ALTER TABLE to change the storage engine of existing tables. Once the inconsistency is triggered, serious business logic issues may arise, potentially leading to financial losses. We have reported this inconsistency issue, sparking extensive discussions among developers, and efforts are underway to resolve and fix the problem. Existing DBMS fuzz testing methods, such as SQUIRREL, typically focus on detecting issues like crashes.",
+      "context_after": "Identifying this issue, which involves comparing 2https://bugs.mysql.com/bug.php?id=112917 661 execution results across multiple engines, is challenging using current DBMS testing methods. Listing 2: Inconsistent error message between CSV and ARCHIVE. -- ENGINE = CSVCREATE TABLE test (v0 INT NOT NULL, v1 CHAR(5) NOT NULL); CREATE INDEX index0 ONtest (v0) USING BTREE; -- ERROR 1069 (42000): Too many keys specified; max 0 keys allowed -- ENGINE = ARCHIVE CREATE TABLE test (v0 INT NOT NULL, v1 CHAR(5) NOT NULL); CREATE INDEX index0 ONtest (v0) USING BTREE; -- ERROR 1030 (HY000): Got error-1 -",
+      "section": "B DBMS Bug Detection",
+      "page": 7,
+      "char_offset": 36947,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Comparison with Existing Techniques To assess the effectiveness of THANOS, we conducted a comparative study that pitted THANOS against contemporary state-of-the-art DBMS testing methods, namely SQLancer, SQLsmith, and SQUIRREL.",
+      "context_before": "orm DBMS testing only with the default storage engine and do not change the storage engines during the test period. THANOS can detect the bug with the storage engine-orient equivalent DBMS construction. First, with the feature-guided storage engine combination selection, THANOS can select theMroonga and InnoDB combination and then extract the equivalence information. Then, with the feature-orient test case synthesis algorithm, the synthesized test cases contain amounts of intensive data storage-related operations, which could be more likely to trigger issues related to DBMS storage engines. C.",
+      "context_after": "Each DBMS underwent a 24hour testing period using these tools, and we documented TABLE III: Number of bugs detected by THANOS, SQLancer, SQLsmith and SQUIRREL on 3 DBMSs in 24 hours. DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 0 1 1 6 MariaDB 0 0 1 5 Percona 0 0 1 3 Total 0 1 3 14 Increment 14↑ 13↑ 11↑ – TABLE IV: Number of branches covered by THANOS, SQLancer, SQLsmith, and SQUIRREL on 3 DBMSs in 24 hours. DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 59,242 93,742 109,323 120,156 MariaDB 60,293 88,923 100,920 132,532 Percona 63,829 89,987 109,823 143,293 Total 183,364 272,652 320,066 395,98",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 41385,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Each DBMS underwent a 24hour testing period using these tools, and we documented TABLE III: Number of bugs detected by THANOS, SQLancer, SQLsmith and SQUIRREL on 3 DBMSs in 24 hours.",
+      "context_before": "uided storage engine combination selection, THANOS can select theMroonga and InnoDB combination and then extract the equivalence information. Then, with the feature-orient test case synthesis algorithm, the synthesized test cases contain amounts of intensive data storage-related operations, which could be more likely to trigger issues related to DBMS storage engines. C. Comparison with Existing Techniques To assess the effectiveness of THANOS, we conducted a comparative study that pitted THANOS against contemporary state-of-the-art DBMS testing methods, namely SQLancer, SQLsmith, and SQUIRREL.",
+      "context_after": "DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 0 1 1 6 MariaDB 0 0 1 5 Percona 0 0 1 3 Total 0 1 3 14 Increment 14↑ 13↑ 11↑ – TABLE IV: Number of branches covered by THANOS, SQLancer, SQLsmith, and SQUIRREL on 3 DBMSs in 24 hours. DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 59,242 93,742 109,323 120,156 MariaDB 60,293 88,923 100,920 132,532 Percona 63,829 89,987 109,823 143,293 Total 183,364 272,652 320,066 395,981 Increment 115.95% ↑45.23% ↑23.72% ↑ – 662 TABLE V: The number of triggered features and feature-related statements ratios in test cases generated by T HANOSand THANOS in 24 hours.",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 41613,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 0 1 1 6 MariaDB 0 0 1 5 Percona 0 0 1 3 Total 0 1 3 14 Increment 14↑ 13↑ 11↑ – TABLE IV: Number of branches covered by THANOS, SQLancer, SQLsmith, and SQUIRREL on 3 DBMSs in 24 hours.",
+      "context_before": "ynthesis algorithm, the synthesized test cases contain amounts of intensive data storage-related operations, which could be more likely to trigger issues related to DBMS storage engines. C. Comparison with Existing Techniques To assess the effectiveness of THANOS, we conducted a comparative study that pitted THANOS against contemporary state-of-the-art DBMS testing methods, namely SQLancer, SQLsmith, and SQUIRREL. Each DBMS underwent a 24hour testing period using these tools, and we documented TABLE III: Number of bugs detected by THANOS, SQLancer, SQLsmith and SQUIRREL on 3 DBMSs in 24 hours.",
+      "context_after": "DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 59,242 93,742 109,323 120,156 MariaDB 60,293 88,923 100,920 132,532 Percona 63,829 89,987 109,823 143,293 Total 183,364 272,652 320,066 395,981 Increment 115.95% ↑45.23% ↑23.72% ↑ – 662 TABLE V: The number of triggered features and feature-related statements ratios in test cases generated by T HANOSand THANOS in 24 hours. DBMS Number of Triggered Features Feature-Related Statements Ratios Name Features THANOS-THANOS Increment THANOS-THANOS Increment MySQL 289 70 257 187↑ 67.18% 83.32% 16.14% ↑ MariaDB 492 161 434 273↑ 63.11% 85.59% 22.48% ↑ Perco",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 41796,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 59,242 93,742 109,323 120,156 MariaDB 60,293 88,923 100,920 132,532 Percona 63,829 89,987 109,823 143,293 Total 183,364 272,652 320,066 395,981 Increment 115.",
+      "context_before": "assess the effectiveness of THANOS, we conducted a comparative study that pitted THANOS against contemporary state-of-the-art DBMS testing methods, namely SQLancer, SQLsmith, and SQUIRREL. Each DBMS underwent a 24hour testing period using these tools, and we documented TABLE III: Number of bugs detected by THANOS, SQLancer, SQLsmith and SQUIRREL on 3 DBMSs in 24 hours. DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 0 1 1 6 MariaDB 0 0 1 5 Percona 0 0 1 3 Total 0 1 3 14 Increment 14↑ 13↑ 11↑ – TABLE IV: Number of branches covered by THANOS, SQLancer, SQLsmith, and SQUIRREL on 3 DBMSs in 24 hours.",
+      "context_after": "95% ↑45.23% ↑23.72% ↑ – 662 TABLE V: The number of triggered features and feature-related statements ratios in test cases generated by T HANOSand THANOS in 24 hours. DBMS Number of Triggered Features Feature-Related Statements Ratios Name Features THANOS-THANOS Increment THANOS-THANOS Increment MySQL 289 70 257 187↑ 67.18% 83.32% 16.14% ↑ MariaDB 492 161 434 273↑ 63.11% 85.59% 22.48% ↑ Percona 359 92 306 214↑ 65.15% 82.75% 17.60% ↑ the number of bugs they uncovered. All identified bugs were reported to the developers, and the confirmed bug count was used as the final result. To ensure a fair",
+      "section": "C Comparison with Existing Techniques",
+      "page": 8,
+      "char_offset": 42025,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, THANOS found 14, 13, and 11 more bugs than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+      "context_before": "er of bugs they uncovered. All identified bugs were reported to the developers, and the confirmed bug count was used as the final result. To ensure a fair comparison, we collected the generated test cases from each testing method and dry-ran the test cases to collect the uniform branch coverage. THANOS outperforms other DBMS testing methods in detecting bugs and covering new branches. Table III shows the number of bugs detected by each method in 24 hours. THANOS detected a total of 14 bugs. From the table, we can see that THANOS finds more bugs than other state-of-the-art DBMS testing methods.",
+      "context_after": "Note that all 14 bugs discovered by THANOS during the 24-hour experiment were subsequently confirmed as previously unknown bugs. Table IV shows the number of branches covered by each DBMS testing method in 24hour experiments. From the table, we can see that THANOS outperformed other testing methods in terms of branch coverage. Specifically, THANOS totally covered 115.95%, 45.23%, and 23.72% more branches than SQLancer, SQLsmith, and SQUIRREL, respectively. The main reason for THANOS ’s improvement in bug detection and branch coverage stems from its adept utilization of the DBMS storage engine",
+      "section": "C Comparison with Existing Techniques",
+      "page": 9,
+      "char_offset": 43275,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "72% more branches than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+      "context_before": "From the table, we can see that THANOS finds more bugs than other state-of-the-art DBMS testing methods. Specifically, THANOS found 14, 13, and 11 more bugs than SQLancer, SQLsmith, and SQUIRREL, respectively. Note that all 14 bugs discovered by THANOS during the 24-hour experiment were subsequently confirmed as previously unknown bugs. Table IV shows the number of branches covered by each DBMS testing method in 24hour experiments. From the table, we can see that THANOS outperformed other testing methods in terms of branch coverage. Specifically, THANOS totally covered 115.95%, 45.23%, and 23.",
+      "context_after": "The main reason for THANOS ’s improvement in bug detection and branch coverage stems from its adept utilization of the DBMS storage engine components for testing purposes. On one hand, THANOS systematically tests various storage engines during the evaluation of each DBMS, whereas conventional DBMS testing methods typically focus on the default storage engine. Consequently, THANOS conspicuously manages to test a more extensive portion of DBMS code. Figure 5 illustrates the branch coverage results after a 24hour test on the storage engine components, showcasing that THANOS achieves an average c",
+      "section": "C Comparison with Existing Techniques",
+      "page": 9,
+      "char_offset": 43771,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, the NoREC test oracle in SQLancer detects logical errors in the DBMS optimizer by constructing equivalent optimized and unoptimized queries based on op-timizer rules.",
+      "context_before": "to other testing methods. As mentioned earlier, the storage engine components serve as the core link between the underlying storage and the database system. By thoroughly testing the features of storage engines, comprehensive evaluation, and testing of the overall features of the DBMS system become possible. Therefore, from a holistic perspective, the improvement in coverage is quite significant for the DBMS. On the other hand, THANOS aims to construct the equivalent DBMSs and thoroughly test its features. In comparison to existing tools, it can test a broader spectrum of DBMS functionalities.",
+      "context_after": "However, this method only covers SQL syntax that complies with optimizer rules. In contrast, THANOS lacks such limitations, as the storage engine, being the underlying software of the DBMS, essentially encompasses all the functionalities of the DBMS. Meanwhile, THANOS performs a comprehensive functional modeling of the DBMS storage engine components, enabling thorough testing of the DBMS when generating test cases. Even during the generation of test cases, THANOS incorporates numerous non-standard SQL statements related to storage engine features. In comparison, tools like SQLancer and SQLsmi",
+      "section": "C Comparison with Existing Techniques",
+      "page": 9,
+      "char_offset": 45093,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In comparison, tools like SQLancer and SQLsmith use standard SQL syntax to build test cases, resulting in THANOS covering more branches and detecting more bugs.",
+      "context_before": "unoptimized queries based on op-timizer rules. However, this method only covers SQL syntax that complies with optimizer rules. In contrast, THANOS lacks such limitations, as the storage engine, being the underlying software of the DBMS, essentially encompasses all the functionalities of the DBMS. Meanwhile, THANOS performs a comprehensive functional modeling of the DBMS storage engine components, enabling thorough testing of the DBMS when generating test cases. Even during the generation of test cases, THANOS incorporates numerous non-standard SQL statements related to storage engine features.",
+      "context_after": "D. Effectiveness of Feature-Oriented Test Case Synthesis To evaluate the effectiveness of our feature-oriented test case synthesis algorithm, we developed a comparison tool, THANOS -, which generates test cases randomly without a focus on the storage engines’ features. We conducted a 24-hour experiment running both T HANOSand THANOS on MySQL, MariaDB, and Percona. During this period, we collected data on the number of features triggered and the proportion of feature-related statements produced by each tool for comparative analysis. Table V presents the number of features activated and the pro",
+      "section": "C Comparison with Existing Techniques",
+      "page": 9,
+      "char_offset": 45828,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "5: Number of branches covered by THANOS, SQLancer, SQLsmith, and SQUIRREL in the Storage Engine component of each DBMS in 24 hours.",
+      "context_before": "th T HANOSand THANOS on MySQL, MariaDB, and Percona. During this period, we collected data on the number of features triggered and the proportion of feature-related statements produced by each tool for comparative analysis. Table V presents the number of features activated and the proportion of statements in the test case with features for both tools across three DBMSs (MySQL, MariaDB, and Percona). The data indicates that THANOS generates more MySQL MariaDB Percona0500010000150002000025000300003500040000Number of branches covered in storage engine component SQLaner SQLsmithSquirrel Thanos Fig.",
+      "context_after": "663 feature-related test cases than THANOS -, which benefits from its feature-oriented test case synthesis approach. In detail, THANOS activated 187, 273, and 214 more features than THANOSin MySQL, MariaDB, and Percona, respectively. This is primarily because the feature-oriented synthesis significantly enhances the proportion of test cases that include storage engine features, with increases of 16.14%, 22.48%, and 17.60% in each DBMS, respectively. These outcomes align with our expectations, as the test case synthesis algorithm was specifically designed to trigger a broader range of storag",
+      "section": "D Effectiveness of Feature-Oriented Test Case Synthesis",
+      "page": 9,
+      "char_offset": 46904,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[28, 29, 30, 31]",
+      "technique": "tlp",
+      "sentence": "Metamorphic testing in DBMS involves transforming SQL queries and verifying if the resulting output changes align with expected behavior [28, 29, 30, 31].",
+      "context_before": "ring SQL execution speed acrossdifferent DBMS versions. Vendor-based comparison entails running equivalent SQL statements on various DBMSs. For instance, RAGS [6] validates SQL outputs by comparing results from multiple DBMS vendors. THANOS employs a specialized form of differential testing to evaluate DBMSs. It equips a DBMS with various storage engines and constructs equivalent test cases, thereby creating equivalent DBMSs. In doing so, THANOS combines diverse DBMS implementations with similar input syntax to uncover hidden issues and minimize the cost of SQL generation. Metamorphic Testing.",
+      "context_after": "It relies on metamorphic relations, logical links between input and output, for test case design, with transformations including sorting, filtering, or aggregating data. Developers use this to ensure correct and robust database functions, maintaining consistent accuracy despite input variations, crucial for revealing hidden bugs and bolstering DBMS reliability. Methods like Amoeba [32] detect performance bugs by comparing response times of semantically equivalent query pairs. SQLancer proposes constructing functionally equivalent queries to test one DBMS [30, 31, 33]. TxCheck [34] detects tra",
+      "section": "D Effectiveness of Feature-Oriented Test Case Synthesis",
+      "page": 10,
+      "char_offset": 50658,
+      "cited_reference": {
+        "number": 30,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on 665 Programming Languages, vol. 4, no. OOPSLA, pp. 1– 30, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer proposes constructing functionally equivalent queries to test one DBMS [30, 31, 33].",
+      "context_before": "s transforming SQL queries and verifying if the resulting output changes align with expected behavior [28, 29, 30, 31]. It relies on metamorphic relations, logical links between input and output, for test case design, with transformations including sorting, filtering, or aggregating data. Developers use this to ensure correct and robust database functions, maintaining consistent accuracy despite input variations, crucial for revealing hidden bugs and bolstering DBMS reliability. Methods like Amoeba [32] detect performance bugs by comparing response times of semantically equivalent query pairs.",
+      "context_after": "TxCheck [34] detects transactional bugs of DBMSs through graph-based oracle construction. DQE [35] verifies consistency of rows fetched by SQL statements designed to access the same rows. Compared to metamorphic testing, our approach focuses more on changing the DBMS itself, specifically by altering the storage engine. Metamorphic testing relies on manually defining metamorphic relations, whereas our method automatically generates equivalent DBMSs, currently applicable for correctness testing. This approach can also be extended to other types of test oracles, such as authorization testing and",
+      "section": "D Effectiveness of Feature-Oriented Test Case Synthesis",
+      "page": 10,
+      "char_offset": 51294,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[34]",
+      "technique": null,
+      "sentence": "TxCheck [34] detects transactional bugs of DBMSs through graph-based oracle construction.",
+      "context_before": "ehavior [28, 29, 30, 31]. It relies on metamorphic relations, logical links between input and output, for test case design, with transformations including sorting, filtering, or aggregating data. Developers use this to ensure correct and robust database functions, maintaining consistent accuracy despite input variations, crucial for revealing hidden bugs and bolstering DBMS reliability. Methods like Amoeba [32] detect performance bugs by comparing response times of semantically equivalent query pairs. SQLancer proposes constructing functionally equivalent queries to test one DBMS [30, 31, 33].",
+      "context_after": "DQE [35] verifies consistency of rows fetched by SQL statements designed to access the same rows. Compared to metamorphic testing, our approach focuses more on changing the DBMS itself, specifically by altering the storage engine. Metamorphic testing relies on manually defining metamorphic relations, whereas our method automatically generates equivalent DBMSs, currently applicable for correctness testing. This approach can also be extended to other types of test oracles, such as authorization testing and compliance testing. DBMS Fuzzing. Fuzz testing continuously generates and executes SQL te",
+      "section": "D Effectiveness of Feature-Oriented Test Case Synthesis",
+      "page": 10,
+      "char_offset": 51388,
+      "cited_reference": {
+        "number": 34,
+        "text": "Z.-M. Jiang, S. Liu, M. Rigger, and Z. Su, “Detecting transactional bugs in database engines via GraphBased oracle construction,” in 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). Boston, MA: USENIX Association, Jul. 2023, pp. 397–417. [Online]. Available: https: //w",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "citation_marker",
+      "surface": "[22, 25, 33, 36, 37, 38, 39, 40, 41, 42]",
+      "technique": "pqs",
+      "sentence": "DBMS fuzzers [22, 25, 33, 36, 37, 38, 39, 40, 41, 42], automate this process, focusing on creating complex SQL queries to uncover memory safety issues.",
+      "context_before": "y SQL statements designed to access the same rows. Compared to metamorphic testing, our approach focuses more on changing the DBMS itself, specifically by altering the storage engine. Metamorphic testing relies on manually defining metamorphic relations, whereas our method automatically generates equivalent DBMSs, currently applicable for correctness testing. This approach can also be extended to other types of test oracles, such as authorization testing and compliance testing. DBMS Fuzzing. Fuzz testing continuously generates and executes SQL test cases on a DBMS, monitoring system responses.",
+      "context_after": "They are categorized into generation-based and mutation-based approaches. Generation-based fuzzing generates large SQL queries using predefined models. SQLsmith [25] is a leading generationbased DBMS fuzzer that creates semantically correct SQL queries using an AST model, primarily limited to SELECT statements due to grammar constraints. Mutation-based fuzzers modify existing queries. Squirrel [22] enhances syntactic accuracy by aligning with SQL grammar and utilizing an Intermediate Representation (IR), facilitating type-based mutations while maintaining syntax correctness. LEGO [36] generat",
+      "section": "D Effectiveness of Feature-Oriented Test Case Synthesis",
+      "page": 10,
+      "char_offset": 52126,
+      "cited_reference": {
+        "number": 33,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M1",
+        "M2",
+        "M3",
+        "M5",
+        "M12",
+        "M13"
+      ],
+      "describes_as_state_of_the_art": [
+        "M1",
+        "M2",
+        "M8"
+      ]
+    },
+    "suppressed": [
+      {
+        "mention_id": "M8",
+        "pattern": "compares_with",
+        "suppressed_because": "the sentence is framed as related work, so it describes somebody else's approach"
+      }
+    ]
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:38:14Z",
+    "is_model_written": true,
+    "summary": "Thanos tests DBMSs that support several storage engines by running the same workload under each in turn and comparing the results. Because the engines sit beneath a shared SQL layer, any difference between them is a defect in one of them, which gives an oracle without needing to know the correct answer. On MySQL, MariaDB and Percona it found 14 bugs in 24 hours and covered substantially more of the storage-engine code than the existing tools.",
+    "narrative": "SQLancer is one of Thanos's three baselines and is named a state-of-the-art DBMS testing tool throughout. The comparison is reported per system in both bugs and branch coverage, and SQLancer found none of the 14 bugs Thanos did. The paper's argument for why is about where the oracles apply: SQLancer's rely on specially crafted rules -- NoREC on optimizer behaviour, and it describes TLP as needing a pivot row -- which leaves storage-engine differences outside their reach, and its use of standard SQL syntax bounds the code it exercises. SQLancer's construction of functionally equivalent queries is also cited as the metamorphic-testing line Thanos departs from.",
+    "roles": {
+      "M1": "state_of_the_art",
+      "M2": "state_of_the_art",
+      "M3": "result_comparison",
+      "M4": "motivation",
+      "M5": "state_of_the_art",
+      "M6": "definition",
+      "M7": "motivation",
+      "M8": "baseline",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "result_comparison",
+      "M12": "result_comparison",
+      "M13": "result_comparison",
+      "M14": "definition",
+      "M15": "result_comparison",
+      "M16": "result_comparison",
+      "M17": "background",
+      "M18": "definition",
+      "M19": "background",
+      "M20": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is run as a baseline; Thanos rotates storage engines beneath a shared SQL layer and nothing indicates its implementation reuses SQLancer's code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Thanos's oracle is differential comparison across storage engines, which the paper contrasts with SQLancer's rule-based oracles rather than deriving from them."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M8",
+          "M10",
+          "M11",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M8",
+            "sentence": "Comparison with Existing Techniques To assess the effectiveness of THANOS, we conducted a comparative study that pitted THANOS against contemporary state-of-the-art DBMS testing methods, namely SQLancer, SQLsmith, and SQUIRREL.",
+            "section": "C Comparison with Existing Techniques",
+            "page": 8
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 0 1 1 6 MariaDB 0 0 1 5 Percona 0 0 1 3 Total 0 1 3 14 Increment 14↑ 13↑ 11↑ – TABLE IV: Number of branches covered by THANOS, SQLancer, SQLsmith, and SQUIRREL on 3 DBMSs in 24 hours.",
+            "section": "C Comparison with Existing Techniques",
+            "page": 8
+          },
+          {
+            "mention_id": "M11",
+            "sentence": "DBMS SQLancer SQLsmith SQUIRREL THANOS MySQL 59,242 93,742 109,323 120,156 MariaDB 60,293 88,923 100,920 132,532 Percona 63,829 89,987 109,823 143,293 Total 183,364 272,652 320,066 395,981 Increment 115.",
+            "section": "C Comparison with Existing Techniques",
+            "page": 8
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "Specifically, THANOS found 14, 13, and 11 more bugs than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+            "section": "C Comparison with Existing Techniques",
+            "page": 9
+          }
+        ],
+        "reasoning": "M8 states the comparative study pits Thanos against SQLancer, and M10, M11 and M12 give the per-DBMS bug counts, branch coverage and totals.",
+        "techniques": [
+          "norec",
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M1",
+          "M2",
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "We evaluate THANOS on three widely used and extensively tested DBMSs, namely MySQL, MariaDB, and Percona against state-of-the-art fuzzers SQLancer, SQLsmith, and SQUIRREL.",
+            "section": null,
+            "page": 1
+          },
+          {
+            "mention_id": "M2",
+            "sentence": "To assess the effectiveness of THANOS, we compare THANOS against contemporary state-of-the-art DBMS testing methods, namely SQLancer, SQLsmith, and SQUIRREL.",
+            "section": "I INTRODUCTION",
+            "page": 2
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "To evaluate the effectiveness of THANOS, we compared THANOS with the state-of-art DBMS test tools, SQUIRREL [22, 23], SQLancer [24], and SQLsmith [25, 26].",
+            "section": "A Evaluation Setup",
+            "page": 7
+          }
+        ],
+        "reasoning": "M1, M2 and M5 each introduce SQLancer as a state-of-the-art DBMS fuzzer or testing tool."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse_companion58688_2023_00041.json b/_data/papers/paper_doi_10_1109_icse_companion58688_2023_00041.json
new file mode 100644
index 0000000..3c66a6c
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse_companion58688_2023_00041.json
@@ -0,0 +1,574 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:16Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse-companion58688.2023.00041",
+    "title": "Randomized Differential Testing of RDF Stores",
+    "authors": [
+      "Rui Yang",
+      "Yingying Zheng",
+      "Leile Tang",
+      "Wensheng Dou",
+      "Wei Wang",
+      "Jun Wei"
+    ],
+    "year": 2023,
+    "venue": "2023 IEEE/ACM 45th International Conference on Software Engineering: Companion Proceedings (ICSE-Companion)",
+    "doi": "10.1109/icse-companion58688.2023.00041",
+    "arxiv_id": null,
+    "s2_paper_id": "0b083953239d0617b534a0d76762fa8350db4e6f",
+    "url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icse-companion58688.2023.00041",
+      "retrieved_at": "2026-09-09T01:33:16Z",
+      "chars": 27148,
+      "content_sha256": "sha256:49ccae7f37a6e5ad01bc71f175193266adb1064fc194a4a9cad01755ecd8f6b2"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 5,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1854
+      },
+      {
+        "number": "II",
+        "title": "PRELIMINARIES",
+        "start": 5911
+      },
+      {
+        "number": "III",
+        "title": "RD2",
+        "start": 8237
+      },
+      {
+        "number": "A",
+        "title": "RDF Graph Generation",
+        "start": 8649
+      },
+      {
+        "number": "B",
+        "title": "SPARQL Query Generation",
+        "start": 10264
+      },
+      {
+        "number": "C",
+        "title": "Differentially Testing RDF Stores",
+        "start": 13183
+      },
+      {
+        "number": "IV",
+        "title": "IMPLEMENTA TION AND USAGE",
+        "start": 14610
+      },
+      {
+        "number": "V",
+        "title": "EVALUA TION",
+        "start": 16400
+      },
+      {
+        "number": "A",
+        "title": "Methodology",
+        "start": 16415
+      },
+      {
+        "number": "B",
+        "title": "Detection Results",
+        "start": 17684
+      },
+      {
+        "number": "VI",
+        "title": "RELA TED WORK",
+        "start": 19652
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 20831
+      },
+      {
+        "number": "W",
+        "title": "Wang, and J. Wei, “Finding bugs in Gremlin-based graph database",
+        "start": 22926
+      },
+      {
+        "number": "T",
+        "title": "Huang, “Testing database systems via differential query execution,”",
+        "start": 24312
+      },
+      {
+        "number": "L",
+        "title": "Chen, H. Wang, H. Zhong, and T. Huang, “Detecting isolation bugs",
+        "start": 24782
+      },
+      {
+        "number": "J",
+        "title": "Wang, and J. Li, “Sequence-oriented DBMS fuzzing,” in Proceedings",
+        "start": 25658
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "I. Abdelaziz, E. Mansour, M. Ouzzani, A. Aboulnaga, and P. Kalnis, “Query optimizations over decentralized RDF graphs,” in Proceedings of International Conference on Data Engineering (ICDE), 2017, pp. 139–142.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "W. Ali, M. Saleem, B. Yao, A. Hogan, and A. N. Ngomo, “A survey of RDF stores & SPARQL engines for querying knowledge graphs,” The VLDB Journal, vol. 31, no. 3, pp. 1–26, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“MarkLogic,” https://www.marklogic.com/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "“Apache Jena,” https://jena.apache.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "“GraphDB,” https://www.ontotext.com/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "“RDF4j,” https://rdf4j.org/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "J. P ´erez, M. Arenas, and C. Gutierrez, “Semantics and complexity of SPARQL,” ACM Transactions on Database Systems (TODS), vol. 34, no. 3, pp. 1–45, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "B. Liu, X. Wang, P. Liu, S. Li, Q. Fu, and Y. Chai, “UniKG: A unified interoperable knowledge graph database system,” in Proceedings of IEEE International Conference on Data Engineering (ICDE), 2021, pp. 2681–2684.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "M. Arenas, C. Guti ´errez, and J. F. Sequeda, “Querying in the age of graph databases and knowledge graphs,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 2021, pp. 2821–2828.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "T. Berners-Lee, J. Hendler, and O. Lassila, “The semantic web,” Scientific american, vol. 284, no. 5, pp. 34–43, 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Y. Zheng, W. Dou, Y. Wang, Z. Qin, L. Tang, Y. Gao, D. Wang, W. Wang, and J. Wei, “Finding bugs in Gremlin-based graph database systems via randomized differential testing,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2022, pp. 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Z. Hua, W. Lin, L. Ren, Z. Li, L. Zhang, W. Jiao, and T. Xie, “GDsmith: Detecting bugs in Cypher graph database engines,” 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "D. R. Slutz, “Massive stochastic testing of SQL,” in Proceedings of International Conference on V ery Large Data Bases (VLDB), 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 16,
+      "text": "——, “Detecting optimization bugs in database engines via nonoptimizing reference engine construction,” in Proceedings of ACM Joint European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140– 1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 17,
+      "text": "——, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 18,
+      "text": "J. Song, W. Dou, Z. Cui, Q. Dai, W. Wang, J. Wei, H. Zhong, and T. Huang, “Testing database systems via differential query execution,” inProceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Z. Cui, W. Dou, Q. Dai, J. Song, W. Wang, J. Wei, and D. Ye, “Differentially testing database transactions for fun and profit,” in Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE), 2022, pp. 35:1–35:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "W. Dou, Z. Cui, Q. Dai, J. Song, D. Wang, Y. Gao, W. Wang, J. Wei, L. Chen, H. Wang, H. Zhong, and T. Huang, “Detecting isolation bugs via transaction oracle construction,” in Proceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "“Python faker,” https://github.com/joke2k/faker, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "“DB-Engines,” https://db-engines.com/en/ranking/rdf+ store, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "W. Lin, Z. Hua, L. Zhang, and T. Xie, “GDiff: Automated differential performance testing for graph database systems,” in Proceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "W. M. McKeeman, “Differential testing for software,” Digital Technical Journal, vol. 10, no. 1, pp. 100–107, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” inProceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 26,
+      "text": "J. Liang, Y. Chen, Z. Wu, J. Fu, M. Wang, Y. Jiang, X. Huang, T. Chen, J. Wang, and J. Li, “Sequence-oriented DBMS fuzzing,” in Proceedings of IEEE International Conference on Data Engineering (ICDE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Z. Jiang, J. Bai, and Z. Su, “DynSQL: Stateful fuzzing for database management systems with complex and valid SQL query generation,” in Proceedings of USENIX Security Symposium (USENIX Security), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "X. Liu, Q. Zhou, J. Arulrai, and A. Orso, “Automatic detection of performance bugs in database systems using equivalent queries,” in Proceedings of International Conference on Software Engineering (ICSE), 2022, pp. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of DBMS with coverage-based guidance,” in Proceedings of USENIX Security Symposium (USENIX Security), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang, and Y. Jiang, “Industry practice of coverage-guided enterprise-level DBMS fuzzing,” inProceedings of IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice (ICSESEIP), 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment (VLDB), vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Z. Gu, M. A. Soliman, and F. M. Waas, “Testing the accuracy of query optimizers,” in Proceedings of International Workshop on Testing Database Systems, 2012, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "“SQLsmith,” https://github.com/anse1/sqlsmith, 2022. 140",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 13,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 15,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 16,
+      "text": "——, “Detecting optimization bugs in database engines via nonoptimizing reference engine construction,” in Proceedings of ACM Joint European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140– 1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 17,
+      "text": "——, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 25,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” inProceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[11]–[13]",
+      "technique": null,
+      "sentence": "Similar to other GDBs that are built on the labeled property graph model [11]–[13] and relational database systems (RDBMSs) [14]–[18], incorrect implementations of RDF stores can introduce logic bugs that result in an incorrect query result for a given SPARQL query, e.",
+      "context_before": "o demonstration of RD2is available at https://youtu.be/da7XlsdbRR4. Index T erms —RDF store, differential testing, SPARQL I. INTRODUCTION The Resource Description Framework (RDF) [1] graph model has been regarded as a W3C standard for exchanging graph data. We refer to the graph database systems (GDBs) that are built on the RDF graph model as RDF stores [2]. The representative RDF stores include MarkLogic [3], Apache Jena [4], GraphDB [5], RDF4j [6], etc. They utilize SPARQL [7] as their standardized query language, and play a significant role in knowledge graphs [8], [9] and semantic web [10].",
+      "context_after": "g., omitting a record. Fig. 1 illustrates a real-world logic bug found in MarkLogic. In this example, we first write an RDF triple (Line 2) into MarkLogic, and then retrieve it by a FILTER expression 80596426678 ∗1719307142, which should be evaluated into true (Line 5−7). We expect the RDF triple can be returned (Line 9). However, MarkLogic mistakenly returns an empty result (Line 8) because the expression is evaluated into false due to decimal overflow. But, Apache Jena and RDF4j can correctly return the RDF triple.1<!-- RDF graph data -- > 2<http://JohnSmith> <http://ages> 12. 3 4<!-- SPARQL q",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2332,
+      "cited_reference": {
+        "number": 13,
+        "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[13]",
+      "technique": null,
+      "sentence": ", Grand [11], GDsmith [12] and GDBMeter [13], detect logic bugs in Gremlin-based or Cypher-based GDBs, which adopt the labeled property graph model.",
+      "context_before": "e 5−7). We expect the RDF triple can be returned (Line 9). However, MarkLogic mistakenly returns an empty result (Line 8) because the expression is evaluated into false due to decimal overflow. But, Apache Jena and RDF4j can correctly return the RDF triple.1<!-- RDF graph data -- > 2<http://JohnSmith> <http://ages> 12. 3 4<!-- SPARQL query -- > 5SELECT * 6WHERE{ ?s ?p ?o. 7 FILTER ( 80596426678 *1719307142 )} 8-- {} 9-- {<http://JohnSmith> <http://ages> 12} Fig. 1. A logic bug in MarkLogic [3]. Logic bugs in RDF stores can easily go unnoticed by developers. Existing GDB testing approaches, e.g.",
+      "context_after": "Existing RDBMS testing approaches [14]–[20] detect bugs in RDBMSs. However, RDF stores adopt totally different graph models and query syntaxes. Existing approaches cannot be directly applied on RDF stores. In this paper, we propose a randomized differential testing technique RD2, to reveal discrepancies among RDF stores. These discrepancies usually indicate logic bugs in RDF stores. We first randomly build an equivalent RDF graph for multiple RDF stores and write it into multiple target RDF stores. We then generate random SPARQL queries. Finally, we compare the query results returned by these",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3466,
+      "cited_reference": {
+        "number": 13,
+        "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[14]–[20]",
+      "technique": "tlp",
+      "sentence": "Existing RDBMS testing approaches [14]–[20] detect bugs in RDBMSs.",
+      "context_before": "evaluated into false due to decimal overflow. But, Apache Jena and RDF4j can correctly return the RDF triple.1<!-- RDF graph data -- > 2<http://JohnSmith> <http://ages> 12. 3 4<!-- SPARQL query -- > 5SELECT * 6WHERE{ ?s ?p ?o. 7 FILTER ( 80596426678 *1719307142 )} 8-- {} 9-- {<http://JohnSmith> <http://ages> 12} Fig. 1. A logic bug in MarkLogic [3]. Logic bugs in RDF stores can easily go unnoticed by developers. Existing GDB testing approaches, e.g., Grand [11], GDsmith [12] and GDBMeter [13], detect logic bugs in Gremlin-based or Cypher-based GDBs, which adopt the labeled property graph model.",
+      "context_after": "However, RDF stores adopt totally different graph models and query syntaxes. Existing approaches cannot be directly applied on RDF stores. In this paper, we propose a randomized differential testing technique RD2, to reveal discrepancies among RDF stores. These discrepancies usually indicate logic bugs in RDF stores. We first randomly build an equivalent RDF graph for multiple RDF stores and write it into multiple target RDF stores. We then generate random SPARQL queries. Finally, we compare the query results returned by these RDF stores to check whether discrepancies exist. To effectively rev",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3614,
+      "cited_reference": {
+        "number": 15,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[11]–[13]",
+      "technique": null,
+      "sentence": "Some approaches [11]–[13], [23] are proposed to test GDBs.",
+      "context_before": "res. For example, MarkLogic returns 1589301716 while Apache Jena and RDF4j return 1.589301716 E9for the same literal. MarkLogic developers explain that the scientific notation appears to be lost for Double in JSON. RDF4j. We find one logic bug in RDF4j. RDF4j cannot automatically perform type conversion between Integer and its derived types, e.g., Int and Long. This prevents users from retrieving data with derived types of Integer using the equal values with Integer type. RDF4j developers have prepared a feature proposal to fix this bug. VI. RELA TED WORK Testing of graph database systems (GDBs).",
+      "context_after": "Grand [11] and GDsmith [12] apply differential testing [24] to detect logic bugs in Gremlin-based and Cypher-based GDBs, respectively. GDBMeter [13] adopts metamorphic testing by partitioning a given query into three derived sub-queries to test GDBs. However, all of these works cannot detect bugs in RDF stores due to different graph models and query syntaxes. Testing of relational database systems (RDBMSs). Many approaches [14]–[20], [25]–[33] are proposed to find bugs in RDBMSs that use SQL as a standardized query language. RAGS [14], APOLLO [31] and TAQO [32] utilize differential testing for",
+      "section": "VI RELA TED WORK",
+      "page": 4,
+      "char_offset": 19711,
+      "cited_reference": {
+        "number": 13,
+        "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[13]",
+      "technique": null,
+      "sentence": "GDBMeter [13] adopts metamorphic testing by partitioning a given query into three derived sub-queries to test GDBs.",
+      "context_before": "for Double in JSON. RDF4j. We find one logic bug in RDF4j. RDF4j cannot automatically perform type conversion between Integer and its derived types, e.g., Int and Long. This prevents users from retrieving data with derived types of Integer using the equal values with Integer type. RDF4j developers have prepared a feature proposal to fix this bug. VI. RELA TED WORK Testing of graph database systems (GDBs). Some approaches [11]–[13], [23] are proposed to test GDBs. Grand [11] and GDsmith [12] apply differential testing [24] to detect logic bugs in Gremlin-based and Cypher-based GDBs, respectively.",
+      "context_after": "However, all of these works cannot detect bugs in RDF stores due to different graph models and query syntaxes. Testing of relational database systems (RDBMSs). Many approaches [14]–[20], [25]–[33] are proposed to find bugs in RDBMSs that use SQL as a standardized query language. RAGS [14], APOLLO [31] and TAQO [32] utilize differential testing for detecting bugs in RDBMSs. SQLsmith [33] is used to detect bugs causing exceptions or crashes in RDBMSs. TLP [15], NoREC [16], PQS [17] and DQE [18] develop various test oracles to detect logic bugs and optimization bugs, and have found many bugs in p",
+      "section": "VI RELA TED WORK",
+      "page": 4,
+      "char_offset": 19905,
+      "cited_reference": {
+        "number": 13,
+        "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[14]–[20]",
+      "technique": "tlp",
+      "sentence": "Many approaches [14]–[20], [25]–[33] are proposed to find bugs in RDBMSs that use SQL as a standardized query language.",
+      "context_before": "ype. RDF4j developers have prepared a feature proposal to fix this bug. VI. RELA TED WORK Testing of graph database systems (GDBs). Some approaches [11]–[13], [23] are proposed to test GDBs. Grand [11] and GDsmith [12] apply differential testing [24] to detect logic bugs in Gremlin-based and Cypher-based GDBs, respectively. GDBMeter [13] adopts metamorphic testing by partitioning a given query into three derived sub-queries to test GDBs. However, all of these works cannot detect bugs in RDF stores due to different graph models and query syntaxes. Testing of relational database systems (RDBMSs).",
+      "context_after": "RAGS [14], APOLLO [31] and TAQO [32] utilize differential testing for detecting bugs in RDBMSs. SQLsmith [33] is used to detect bugs causing exceptions or crashes in RDBMSs. TLP [15], NoREC [16], PQS [17] and DQE [18] develop various test oracles to detect logic bugs and optimization bugs, and have found many bugs in popular RDBMSs. DT2 [19] and Troc [20] detect transaction bugs in RDBMSs. However, these tools cannot be directly applied to test RDF stores because SPARQL has different data models and query syntaxes from SQL. VII. CONCLUSION We present RD2, an automated testing technique for re",
+      "section": "VI RELA TED WORK",
+      "page": 4,
+      "char_offset": 20181,
+      "cited_reference": {
+        "number": 15,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [15], NoREC [16], PQS [17] and DQE [18] develop various test oracles to detect logic bugs and optimization bugs, and have found many bugs in popular RDBMSs.",
+      "context_before": "ypher-based GDBs, respectively. GDBMeter [13] adopts metamorphic testing by partitioning a given query into three derived sub-queries to test GDBs. However, all of these works cannot detect bugs in RDF stores due to different graph models and query syntaxes. Testing of relational database systems (RDBMSs). Many approaches [14]–[20], [25]–[33] are proposed to find bugs in RDBMSs that use SQL as a standardized query language. RAGS [14], APOLLO [31] and TAQO [32] utilize differential testing for detecting bugs in RDBMSs. SQLsmith [33] is used to detect bugs causing exceptions or crashes in RDBMSs.",
+      "context_after": "DT2 [19] and Troc [20] detect transaction bugs in RDBMSs. However, these tools cannot be directly applied to test RDF stores because SPARQL has different data models and query syntaxes from SQL. VII. CONCLUSION We present RD2, an automated testing technique for revealing discrepancies among RDF stores and detecting logic bugs in RDF stores. RD2randomly generates an equivalent RDF graph for multiple RDF stores, and checks whether these RDF stores can return the same query result for a given SPARQL query. Our experiment shows that RD2has detected 5 logic bugs in three widely-used RDF stores, an",
+      "section": "VI RELA TED WORK",
+      "page": 4,
+      "char_offset": 20474,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec",
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:04Z",
+    "is_model_written": true,
+    "summary": "A short paper on randomized differential testing for RDF stores, which use SPARQL and the RDF data model rather than the labeled property graph model the existing graph testers target. It runs the same query across several RDF stores and reports disagreements, extending automated logic-bug detection to a data model that had none.",
+    "narrative": "SQLancer's techniques appear as the relational precedent the paper places itself after. TLP, NoREC, PQS and DQE are named together as the oracles developed for relational systems that have found many bugs there, and GDBMeter as the tool that carried query partitioning into property-graph databases. The paper's own contribution is differential testing across RDF stores, so the citation establishes the lineage rather than a technical relationship.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "definition",
+      "M6": "background",
+      "M7": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icse_seip52600_2021_00042.json b/_data/papers/paper_doi_10_1109_icse_seip52600_2021_00042.json
new file mode 100644
index 0000000..40f78a8
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icse_seip52600_2021_00042.json
@@ -0,0 +1,950 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:07:43Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icse-seip52600.2021.00042",
+    "title": "Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing",
+    "authors": [
+      "Mingzhe Wang",
+      "Zhiyong Wu",
+      "Xinyi Xu",
+      "Jie Liang",
+      "Chijin Zhou",
+      "Huafeng Zhang",
+      "Yu Jiang"
+    ],
+    "year": 2021,
+    "venue": "2021 IEEE/ACM 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP)",
+    "doi": "10.1109/icse-seip52600.2021.00042",
+    "arxiv_id": "2103.00804",
+    "s2_paper_id": "8356864e3802eaee426a1fec7ea838bf8d7d7e57",
+    "url": "https://doi.org/10.1109/icse-seip52600.2021.00042",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2103.00804",
+      "retrieved_at": "2026-09-07T17:07:43Z",
+      "chars": 56637,
+      "content_sha256": "sha256:de760d8ddc6e1bbf389770e8eb1a38cb8ea1113dedad0e7e785c0e54dbf0242a"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 10,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1873
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 8092
+      },
+      {
+        "number": "B",
+        "title": "Fuzzing DBMSs",
+        "start": 10045
+      },
+      {
+        "number": "C",
+        "title": "Fuzzers Chosen by This Paper",
+        "start": 13228
+      },
+      {
+        "number": "A",
+        "title": "Imprecise Coverage Collection",
+        "start": 20669
+      },
+      {
+        "number": "B",
+        "title": "Fragile Input Generation",
+        "start": 27326
+      },
+      {
+        "number": "C",
+        "title": "Unreproducible Bugs",
+        "start": 30246
+      },
+      {
+        "number": "A",
+        "title": "Analysis of Results",
+        "start": 34220
+      },
+      {
+        "number": "B",
+        "title": "Analysis of Bugs",
+        "start": 40188
+      },
+      {
+        "number": "S",
+        "title": "Goikhman, E. Levy, I. Levy, F. Lu, L. Mishali, Y. Mo, N. Pachter,",
+        "start": 53642
+      },
+      {
+        "number": "D",
+        "title": "Sivov, V. Veeraraghavan, V. Vexler, L. Wang, and P. Wang, “Industrial",
+        "start": 53711
+      },
+      {
+        "number": "Y",
+        "title": "Shoshitaishvili, C. Kruegel, and G. Vigna, “Driller: Augmenting",
+        "start": 55381
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "A. Seltenreich, B. Tang, and S. Mullender, “Sqlsmith: a random sql query generator,” 2018. [Online]. Available: https://github.com/anse1/ sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "D. R. Slutz, “Massive stochastic testing of SQL,” in VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, New York, USA. Morgan Kaufmann, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "M. Rigger, “Sqlancer: detecting logic bugs in dbms,” 2020. [Online]. Available: https://github.com/sqlancer/sqlancer",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "M. Rigger and Z. Su, “Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction,” in Proceedings of the 2020 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ser. ESEC/FSE 2020, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "“Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). Banff, Alberta: USENIX Association, Nov. 2020. [Online]. Available: https://www.usenix.org/conference/osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "J. Liang, Y. Jiang, Y. Chen, M. Wang, C. Zhou, and J. Sun, “PAFL: extend fuzzing optimizations of single mode to industrial parallel mode,” inProceedings of the 2018 ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/SIGSOFT FSE, USA, November 04-09, 2018, 2018, pp. 809–814.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "C. Zhou, M. Wang, J. Liang, Z. Liu, and Y. Jiang, “Zeror: Speed up fuzzing with coverage-sensitive tracing and scheduling,” in 2020 35th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 2020, pp. 858–870.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing database management systems with language validity and coverage feedback,” in The ACM Conference on Computer and Communications Security (CCS), 2020, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "“SQLite: All Tickets,” https://www.sqlite.org/src/rptview?rn=1, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "“PostgreSQL: PostgreSQL Mailing Lists: pgsql-bugs,” https://www. postgresql.org/list/pgsql-bugs/, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "C. Lattner and V. Adve, “Llvm: A compilation framework for lifelong program analysis & transformation,” in International Symposium on Code Generation and Optimization, 2004. CGO 2004. IEEE, 2004, pp. 75–86.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "“Gaussdb Documentation,” https://opengauss.org/en/docs/1.0.1/docs/ Releasenotes/Terms-of-Use.html, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "H. Avni, A. Aliev, O. Amor, A. Avitzur, I. Bronshtein, E. Ginot, S. Goikhman, E. Levy, I. Levy, F. Lu, L. Mishali, Y. Mo, N. Pachter, D. Sivov, V. Veeraraghavan, V. Vexler, L. Wang, and P. Wang, “Industrial strength OLTP using main memory and many cores,” Proc. VLDB Endow., vol. 13, no. 12, pp. 3099–3111, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., no. OOPSLA, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear),” in Proceedings of the 46th International Conference on Very Large Data Bases (VLDB), Tokyo, Japan, Aug. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "A. Seltenreich, B. Tang, and S. Mullender, “Sqlsmith: Score list,” 2020. [Online]. Available: https://github.com/anse1/sqlsmith/wiki#score-list",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "M. Rigger, “Sqlancer: Bugs found in database management systems,” 2020. [Online]. Available: https://www.manuelrigger.at/dbms-bugs",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 18,
+      "text": "lcamtuf, “American fuzzy lop (afl),” 2017. [Online]. Available: http://lcamtuf.coredump.cx/afl/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "libfuzzer@googlegroups.com, “libfuzzer – a library for coverage-guided fuzz testing.” 2020. [Online]. Available: https://llvm.org/docs/LibFuzzer. html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "honggfuzz@googlegroups.com, “hongfuzzsecurity oriented fuzzer with powerful analysis options.” 2020. [Online]. Available: http: //honggfuzz.com",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "M. Wang, J. Liang, Y. Chen, Y. Jiang, X. Jiao, H. Liu, X. Zhao, and J. Sun, “SAFL: increasing and accelerating testing coverage with symbolic execution and guided fuzzing,” in Proceedings of the 40th International Conference on Software Engineering: Companion Proceeedings, ICSE 2018, Gothenburg, Sweden. ACM, 2018, pp. 61–64.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "N. Stephens, J. Grosen, C. Salls, A. Dutcher, R. Wang, J. Corbetta, Y. Shoshitaishvili, C. Kruegel, and G. Vigna, “Driller: Augmenting fuzzing through selective symbolic execution,” in 23rd Annual Network and Distributed System Security Symposium, NDSS 2016, San Diego, California, USA, February 21-24, 2016. The Internet Society, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "C. Lemieux and K. Sen, “Fairfuzz: a targeted mutation strategy for increasing greybox fuzz testing coverage,” in Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering, ASE 2018, Montpellier, France, September 3-7, 2018, 2018, pp. 475–485. [Online]. Available: https://doi.org/10.1145/ 3238147.3238176",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "J. Liang, Y. Chen, M. Wang, Y. Jiang, Z. Yang, C. Sun, X. Jiao, and J. Sun, “Engineering a better fuzzer with synergically integrated optimizations,” in Reliability Engineering (ISSRE). IEEE, 2019, pp. 82–92.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "“STANDARDS BYISO/IECJTC 1/SC 32 Data management and interchange,” https://www.iso.org/committee/45342/x/catalogue/p/1/u/0/ w/0/d/0, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "S. Gan, C. Zhang, X. Qin, X. Tu, K. Li, Z. Pei, and Z. Chen, “Collafl: Path sensitive fuzzing,” in and Privacy, SP 2018, Proceedings, 21-23 May 2018, San Francisco, California, USA. IEEE Computer Society, 2018, pp. 679–696. [Online]. Available: https://doi.org/10.1109/SP.2018.00040",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 3,
+      "text": "M. Rigger, “Sqlancer: detecting logic bugs in dbms,” 2020. [Online]. Available: https://github.com/sqlancer/sqlancer",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 4,
+      "text": "M. Rigger and Z. Su, “Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction,” in Proceedings of the 2020 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ser. ESEC/FSE 2020, 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 5,
+      "text": "“Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). Banff, Alberta: USENIX Association, Nov. 2020. [Online]. Available: https://www.usenix.org/conference/osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., no. OOPSLA, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 17,
+      "text": "M. Rigger, “Sqlancer: Bugs found in database management systems,” 2020. [Online]. Available: https://www.manuelrigger.at/dbms-bugs",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Compared to industrial black box fuzzers SQLsmith and SQLancer, as well as coverage-guided academic fuzzer Squirrel, RATEL covered 38.",
+      "context_before": "In our practice of testing GaussDB and Comdb2, we found major challenges in all three testing stages. The challenges are collecting precise coverage, optimizing fuzzing performance, and analyzing root causes. In search of a general method to overcome these challenges, we propose RATEL, a coverage-guided fuzzer for enterprise-level DBMSs. With its industry-oriented design, RATEL improves the feedback precision, enhances the robustness of input generation, and performs an on-line investigation on the root cause of bugs. As a result, RATEL outperformed other fuzzers in terms of coverage and bugs.",
+      "context_after": "38%, 106.14%, 583.05% more basic blocks than the best results of other three fuzzers in GaussDB, PostgreSQL, and Comdb2, respectively. More importantly, RATEL has discovered 32, 42, and 5 unknown bugs in GaussDB, Comdb2, and PostgreSQL. Index Terms —DBMS testing, coverage-guided fuzzing, enterprise DBMS I. INTRODUCTION As an infrastructure for data persistence and analysis, Database Management Systems (DBMSs) are widely applied in the modern software stack, especially enterprise software. In recent years, their reliability and security have gained traction in academic research as well as in in",
+      "section": null,
+      "page": 1,
+      "char_offset": 1433,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLsmith [1] triggers system bugs by continuously generating random SQL queries; RAGS [2] detects logic bugs by comparing the results of a query on multiple DBMSs; SQLancer [3] detects logic bugs by constructing an invariant oracle from different angles [4], BYu Jiang is the correspondence author.",
+      "context_before": "pecially enterprise software. In recent years, their reliability and security have gained traction in academic research as well as in industry. To improve the correctness of DBMSs, the industry has been applying blackbox fuzzing for decades. Basically, to perform fuzz testing, a fuzzer generates a number of random inputs and sends them to the target system for execution. If anomalies such as crashes, timeouts, internal errors, or incorrect results are detected, then the triggering input is saved. Developers can use the input to reproduce the anomaly, investigate the root cause, and fix the bug.",
+      "context_after": "[5]. The effectiveness of blackbox fuzzing has been proven by many previously-unknown bugs from the industry’s practice: more than 100 bugs were found by SQLsmith, and more than 400 bugs were found by SQLancer. Recently, coverage-guided fuzzing has become a hot topic in the research community, and impressive gains over the conventional blackbox fuzzing methods are observed [6], [7]. Coverage-guided fuzzing introduces feedback into the fuzzing loop: more than blindly generating random inputs, a coverageguided fuzzer collects the execution trace of an input to recognize interesting inputs trigge",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2632,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[5]",
+      "technique": "pqs",
+      "sentence": "[5].",
+      "context_before": "dom inputs and sends them to the target system for execution. If anomalies such as crashes, timeouts, internal errors, or incorrect results are detected, then the triggering input is saved. Developers can use the input to reproduce the anomaly, investigate the root cause, and fix the bug. For example, SQLsmith [1] triggers system bugs by continuously generating random SQL queries; RAGS [2] detects logic bugs by comparing the results of a query on multiple DBMSs; SQLancer [3] detects logic bugs by constructing an invariant oracle from different angles [4], BYu Jiang is the correspondence author.",
+      "context_after": "The effectiveness of blackbox fuzzing has been proven by many previously-unknown bugs from the industry’s practice: more than 100 bugs were found by SQLsmith, and more than 400 bugs were found by SQLancer. Recently, coverage-guided fuzzing has become a hot topic in the research community, and impressive gains over the conventional blackbox fuzzing methods are observed [6], [7]. Coverage-guided fuzzing introduces feedback into the fuzzing loop: more than blindly generating random inputs, a coverageguided fuzzer collects the execution trace of an input to recognize interesting inputs triggering",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2944,
+      "cited_reference": {
+        "number": 5,
+        "text": "“Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). Banff, Alberta: USENIX Association, Nov. 2020. [Online]. Available: https://www.usenix.org/conference/osdi20/presentation/rigger",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The effectiveness of blackbox fuzzing has been proven by many previously-unknown bugs from the industry’s practice: more than 100 bugs were found by SQLsmith, and more than 400 bugs were found by SQLancer.",
+      "context_before": "inputs and sends them to the target system for execution. If anomalies such as crashes, timeouts, internal errors, or incorrect results are detected, then the triggering input is saved. Developers can use the input to reproduce the anomaly, investigate the root cause, and fix the bug. For example, SQLsmith [1] triggers system bugs by continuously generating random SQL queries; RAGS [2] detects logic bugs by comparing the results of a query on multiple DBMSs; SQLancer [3] detects logic bugs by constructing an invariant oracle from different angles [4], BYu Jiang is the correspondence author.[5].",
+      "context_after": "Recently, coverage-guided fuzzing has become a hot topic in the research community, and impressive gains over the conventional blackbox fuzzing methods are observed [6], [7]. Coverage-guided fuzzing introduces feedback into the fuzzing loop: more than blindly generating random inputs, a coverageguided fuzzer collects the execution trace of an input to recognize interesting inputs triggering new program behaviors; the interesting ones are preserved for further mutations, while the mundane ones are discarded. For example, Squirrel [8] combines coverage-based fuzzing and model-based generation.",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2948,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Compared to SQLsmith, SQLancer, and Squirrel, it covered 38.",
+      "context_before": "ree challenges encountered in our practice to fuzz DBMSs. The challenges are imprecise coverage collection on distributed systems, fragile input generation of complex SQL dialects, and unreproducible bug reports on stateful systems. •We propose RATEL to solve the challenges as a general solution for coverage-guided fuzzing on enterprise-level DBMSs. We improve the feedback precision, enhance the robustness of input generation, and perform on-line investigation on the root cause of bugs. •The final results show that RATEL effectively improved existing DBMS fuzzing works on enterpriselevel DBMSs.",
+      "context_after": "38%, 106.14%, 583.05% more basic blocks than the best results of other three fuzzers in GaussDB, PostgreSQL, and Comdb2, respectively. More importantly, 79 unknown bugs are discovered by it. II. BACKGROUND In this section, we first give a brief introduction to DBMS and give an example to illustrate the complexity of enterpriselevel DBMSs. Then, we introduce some existing solutions of DBMS fuzzing. Finally, we detail three state-of-the-art fuzzers, i.e. SQLsmith, SQLancer, and Squirrel, which are chosen for our industry practice.A. Database Management System (DBMS) Client Core Components Backend",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 7840,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLsmith, SQLancer, and Squirrel, which are chosen for our industry practice.",
+      "context_before": "hat RATEL effectively improved existing DBMS fuzzing works on enterpriselevel DBMSs. Compared to SQLsmith, SQLancer, and Squirrel, it covered 38.38%, 106.14%, 583.05% more basic blocks than the best results of other three fuzzers in GaussDB, PostgreSQL, and Comdb2, respectively. More importantly, 79 unknown bugs are discovered by it. II. BACKGROUND In this section, we first give a brief introduction to DBMS and give an example to illustrate the complexity of enterpriselevel DBMSs. Then, we introduce some existing solutions of DBMS fuzzing. Finally, we detail three state-of-the-art fuzzers, i.e.",
+      "context_after": "A. Database Management System (DBMS) Client Core Components Backend Checkpoint Vacuum Logging Recovery …Query Engine Parser Planner Analyzer ExecutorStorage Engine Memory-Optimized T able Disk-Based T able Foreign Data W rapper Master Node Redo Log WAL SenderStandby Node Redo Log WAL ReceiverJDBC ODBC Shell Fig. 1. Interactions between different processes/threads in GaussDB. Database management system (DBMS) is the software that interacts with end users and applications. Different from libraries and utilities, most DBMSs are large and complex distributed systems. They generally contain hundred",
+      "section": "II BACKGROUND",
+      "page": 2,
+      "char_offset": 8356,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In addition, SQLancer [3] integrates three different strategies [4], [5], [14] to construct invirant oracle to detect logic bugs.",
+      "context_before": "ure of DBMS, ensuring its reliability and security is difficult. Therefore, many DBMS fuzzers have sprung up in the past years and tried to find bugs in different aspects. Generally, two approaches were taken by them, i.e. blackbox fuzzing and coverage-guided fuzzing. Blackbox DBMS fuzzing focuses on rapidly constructing a large amount of valid and effective data to pass to the DBMS without the knowledge of the DBMS’s source code. For instance, SQLsmith [1] continuously generates syntacticallycorrect SQL statements and passes them to the DBMS, meanwhile detects whether the DBMS triggers crashes.",
+      "context_after": "For example, its pivoted query synthesis strategy [5] generates queries of which corresponding result table is supposed to include a specific row, and if the DBMS fails to fetch the row, a logic bug is discovered. Additionally, some research works [2], [15] leverage differential testing to detect abnormal behavior in DBMSs. RAGS [2] generates and executes queries in multiple DBMSs, meanwhile compares the results of a query on those DBMSs; similar to RAGS, APPOLO [15] generates queries guided by performance difference and compares the results of a query on different versions of the same DBMS. A",
+      "section": "B Fuzzing DBMSs",
+      "page": 2,
+      "char_offset": 10703,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "pqs",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "pivoted query synthesis",
+      "technique": "pqs",
+      "sentence": "For example, its pivoted query synthesis strategy [5] generates queries of which corresponding result table is supposed to include a specific row, and if the DBMS fails to fetch the row, a logic bug is discovered.",
+      "context_before": "tried to find bugs in different aspects. Generally, two approaches were taken by them, i.e. blackbox fuzzing and coverage-guided fuzzing. Blackbox DBMS fuzzing focuses on rapidly constructing a large amount of valid and effective data to pass to the DBMS without the knowledge of the DBMS’s source code. For instance, SQLsmith [1] continuously generates syntacticallycorrect SQL statements and passes them to the DBMS, meanwhile detects whether the DBMS triggers crashes. In addition, SQLancer [3] integrates three different strategies [4], [5], [14] to construct invirant oracle to detect logic bugs.",
+      "context_after": "Additionally, some research works [2], [15] leverage differential testing to detect abnormal behavior in DBMSs. RAGS [2] generates and executes queries in multiple DBMSs, meanwhile compares the results of a query on those DBMSs; similar to RAGS, APPOLO [15] generates queries guided by performance difference and compares the results of a query on different versions of the same DBMS. All these blackbox DBMS fuzzers have shown good performance on bug discovery in the industry’s practice. For example, SQLsmith has found 118 bugs in the past five years [16], SQLancer has found over 400 bugs in the",
+      "section": "B Fuzzing DBMSs",
+      "page": 2,
+      "char_offset": 10833,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLsmith has found 118 bugs in the past five years [16], SQLancer has found over 400 bugs in the past two years [17].",
+      "context_before": "table is supposed to include a specific row, and if the DBMS fails to fetch the row, a logic bug is discovered. Additionally, some research works [2], [15] leverage differential testing to detect abnormal behavior in DBMSs. RAGS [2] generates and executes queries in multiple DBMSs, meanwhile compares the results of a query on those DBMSs; similar to RAGS, APPOLO [15] generates queries guided by performance difference and compares the results of a query on different versions of the same DBMS. All these blackbox DBMS fuzzers have shown good performance on bug discovery in the industry’s practice.",
+      "context_after": "While blackbox fuzzing for DBMSs has gained popularity, coverage-guided fuzzing is hardly applied in the industry. To increase code coverage, coverage-guided fuzzing leverages instrumentation or even program analysis techniques; sometimes, directed fuzzing can also be used to reach certain critical program locations. Traditional coverage-guided fuzzers, like AFL [18], libFuzzer [19] and Honggfuzz [20], can be easily applied to DBMS libraries such as SQLite. These fuzzers use mutation-based input generation techniques, and the generated inputs are very unlikely to be syntactically correct. To",
+      "section": "B Fuzzing DBMSs",
+      "page": 3,
+      "char_offset": 11536,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "•SQLancer is a fuzzer for hunting logic bugs in DBMS.",
+      "context_before": "e found that only a few of them can be applied practically for bug hunting. For example, RAGS is not open-source and APPOLO requires a certain version of DBMS and the adaption cost is great. After considering the investments and the possible returns, we chose three fuzzers as our target in our experiment. Table I shows the characteristics of these fuzzers. We summarize their features as follow: •SQLsmith is a random SQL query generator. It detects bugs by checking whether the connection to server isclosed. It is easy to adapt to a new DBMS because it only needs to connect to target SQL server.",
+      "context_after": "It detects logic bugs by constructing invariant oracle [4], [5], [14] and checking whether results violate semantic logic. Adapting it to a new DBMS may take some time because new logic assertion and syntax generation should be implemented. •Squirrel is a fuzzer aiming at finding memory corruption issues in DBMS. It detects memory corruption in a way similar to SQLsmith, i.e. checking whether the connection to server is closed. Adapting it to a new DBMS may take many manual efforts because it requires customizing AST parser, writing grammar rules, and implementing client logic under its framew",
+      "section": "C Fuzzers Chosen by This Paper",
+      "page": 3,
+      "char_offset": 14147,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[4]",
+      "technique": "norec",
+      "sentence": "It detects logic bugs by constructing invariant oracle [4], [5], [14] and checking whether results violate semantic logic.",
+      "context_before": "ally for bug hunting. For example, RAGS is not open-source and APPOLO requires a certain version of DBMS and the adaption cost is great. After considering the investments and the possible returns, we chose three fuzzers as our target in our experiment. Table I shows the characteristics of these fuzzers. We summarize their features as follow: •SQLsmith is a random SQL query generator. It detects bugs by checking whether the connection to server isclosed. It is easy to adapt to a new DBMS because it only needs to connect to target SQL server. •SQLancer is a fuzzer for hunting logic bugs in DBMS.",
+      "context_after": "Adapting it to a new DBMS may take some time because new logic assertion and syntax generation should be implemented. •Squirrel is a fuzzer aiming at finding memory corruption issues in DBMS. It detects memory corruption in a way similar to SQLsmith, i.e. checking whether the connection to server is closed. Adapting it to a new DBMS may take many manual efforts because it requires customizing AST parser, writing grammar rules, and implementing client logic under its framework. TABLE IFEATURES OFCHOSEN FUZZERS Fuzzer SQLsmith SQLancer Squirrel Syntax Validity√ √ √ Semantic Validity ×√ √ Logic C",
+      "section": "C Fuzzers Chosen by This Paper",
+      "page": 3,
+      "char_offset": 14201,
+      "cited_reference": {
+        "number": 4,
+        "text": "M. Rigger and Z. Su, “Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction,” in Proceedings of the 2020 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ser. ESEC/FSE 2020, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "pqs",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "TABLE IFEATURES OFCHOSEN FUZZERS Fuzzer SQLsmith SQLancer Squirrel Syntax Validity√ √ √ Semantic Validity ×√ √ Logic Check ×√× Coverage-Guided × ×√ Adaption Difficulty Easy Medium Hard III.",
+      "context_before": "detects logic bugs by constructing invariant oracle [4], [5], [14] and checking whether results violate semantic logic. Adapting it to a new DBMS may take some time because new logic assertion and syntax generation should be implemented. •Squirrel is a fuzzer aiming at finding memory corruption issues in DBMS. It detects memory corruption in a way similar to SQLsmith, i.e. checking whether the connection to server is closed. Adapting it to a new DBMS may take many manual efforts because it requires customizing AST parser, writing grammar rules, and implementing client logic under its framework.",
+      "context_after": "HOW TOFUZZ DATABASE Different DBMSs have different properties, but the procedures to fuzz them with coverage guidance are roughly similar. The steps can be summarized as Fig. 2. First, to collect coverage on executions, the fuzz target should be built into instrumented binaries. Next, to generate correct SQL statements, the high-quality initial corpus is collected for further mutations. During the process, crash-triggering inputs are saved for further inspection. Finally, to analyze the root causes, the abnormal inputs are resubmitted to the binaries to reproduce the bug and create bug report",
+      "section": "C Fuzzers Chosen by This Paper",
+      "page": 3,
+      "char_offset": 14805,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Many tools are implemented following this method, including SQLsmith, SQLancer, and Squirrel.",
+      "context_before": "es on corpora of partially constructed SQL statements in the form of ASTs, covering different types of SQL syntax. To generate SQL statements, the original SQL statements are first parsed into ASTs. Next, transformations such as splicing and deletion are performed on the nodes of the ASTs. Finally, the transformed ASTs are serialized into SQL statements. Although this type of generation cannot ensure the correctness of semantics, the generated inputs are syntactically-correct at least. Consequently, early rejections at parsers are avoided, and deeper logic in the target system can be exercised.",
+      "context_after": "c) Analyze Root Causes: When executing randomlygenerated SQL statements, unexpected behaviors indicating bugs may occur. To analyze root causes and fix the bug, A fuzzer must contain a test oracle to detect execution anomalies, or the bug-triggering input may be discarded. Additionally, the related context must be collected too, or the bug will be hard to reproduce and analyze. For fuzzers targeted at different types of bugs, different types of anomalies may be monitored, and the context information to record also differs. According to the way of findingbugs and monitoring anomalies, we can rou",
+      "section": "C Fuzzers Chosen by This Paper",
+      "page": 4,
+      "char_offset": 18754,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To detect semantic bugs, tools such as SQLancer implement a test oracle by verifying the returned values.",
+      "context_before": "nerated SQL statements, unexpected behaviors indicating bugs may occur. To analyze root causes and fix the bug, A fuzzer must contain a test oracle to detect execution anomalies, or the bug-triggering input may be discarded. Additionally, the related context must be collected too, or the bug will be hard to reproduce and analyze. For fuzzers targeted at different types of bugs, different types of anomalies may be monitored, and the context information to record also differs. According to the way of findingbugs and monitoring anomalies, we can roughly divide the fuzzing tools into two categories.",
+      "context_after": "If the results are unexpected, the original query and description of mismatched semantics are logged for further inspection. To detect implementation bugs, tools such as Squirrel looks for crashes and then record the input triggering them. Although these tools are targeted at different types of bugs, they all monitor for execution anomalies which provide valuable context for pinpointing the culprits. After discovering the anomalies, developers should analyze and classify the anomalies to find root causes. Sometimes, the bug can be reproduced with the anomaly-triggering input for a better under",
+      "section": "C Fuzzers Chosen by This Paper",
+      "page": 4,
+      "char_offset": 19498,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLsmith, SQLancer, and Squirrel to test enterprise-level DBMSs.",
+      "context_before": "detect implementation bugs, tools such as Squirrel looks for crashes and then record the input triggering them. Although these tools are targeted at different types of bugs, they all monitor for execution anomalies which provide valuable context for pinpointing the culprits. After discovering the anomalies, developers should analyze and classify the anomalies to find root causes. Sometimes, the bug can be reproduced with the anomaly-triggering input for a better understanding of the situation. IV. CHALLENGES ANDSOLUTIONS As Section II-C lists, we chose the most popular DBMS testing tools, i.e.",
+      "context_after": "Despite their wide usage in realworld projects, we still faced considerable challenges when using the coverage-guided fuzzer on enterprise-level DBMSs. As summarized in Fig. 3, in the following, we present the details of challenges as well as the solutions in each step. A. Imprecise Coverage Collection Compared to well-tested library-based DBMSs, the introduction of enterprise-level features in industrial DBMSs leads to extra logic, and inevitably, more components. For example, each node of GaussDB can support at most 8,000 concurrent connections. To support business on large scale, GaussDB i",
+      "section": "C Fuzzers Chosen by This Paper",
+      "page": 4,
+      "char_offset": 20332,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We observed similar results on SQLancer.",
+      "context_before": "ted with a code of 466K basic blocks, not to mention deployments real-life deployments with components for high-availability, monitoring, and management. Such a complex system poses challenges for fuzzers in terms of efficiency. While blackbox fuzzers are robust against such a complex system, their efficiency is quite limited. When running blackbox fuzzers on DBMSs, we find new code coverage takes many executions to discover. Take SQLsmith as an example, although the number of covered number of basic blocks grew continuously, the growth is extraordinarily slow compared to coverage-guided methods.",
+      "context_after": "Coverage-guided fuzzer Squirrel improves efficiency by discovering the interesting inputs and assigning more power to mutate them. Compared to blackbox fuzzers, the efficiency of Squirrel is greatly improved, but only for a short time: the coverage of Squirrel quickly reached a plateau. By collecting the number of covered basic blocks over time, we found that a bottleneck is reached soon because its newly generated inputs hardly improve the coverage. To solve the problem of efficiency and bottleneck, we investigated the causes behind it. The problem can be mainly Collect Coverage Generate Inputs",
+      "section": "A Imprecise Coverage Collection",
+      "page": 4,
+      "char_offset": 21770,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLsmith and SQLancer test DBMS with blackbox methods.",
+      "context_before": "ce FilesBug Report Rich Context CollectionFig. 3. Overview of challenges and solutions in each step. 1) When building fuzz targets, the precision of coverage collection can be improved with bijective block mapping for accuracy and inter-binary coverage linkage for completeness. 2) When generating SQL queries, the fragile SQL corpus-based mutation can be strengthened with dictionaries. 3) When investigating root causes, the analysis of unreproducible anomalies can be performed on-line with deduplication and rich context collection. attributed to the loss of precise coverage feedback collection.",
+      "context_after": "In other words, though they can determine whether executions of SQL statements are normal, they do not have any coverage feedback. As a result, despite being able to generate syntactically correct SQL statements, these fuzzers could only expand the coverage blindly. The SQL statements generated by the fuzzers are in a narrow-range, which leads to difficulties in covering new basic blocks or finding bugs in deep logic. Squirrel introduces sparse feedback in AFL to guide fuzzing on DBMSs. It hashes the branch (transitions of basic blocks) to one position in a 256KB bitmap. However, the coverage f",
+      "section": "A Imprecise Coverage Collection",
+      "page": 5,
+      "char_offset": 23321,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "citation_marker",
+      "surface": "[17]",
+      "technique": null,
+      "sentence": "These generation-based methods have shown significant effectiveness in DBMS bug discovery [8], [17].",
+      "context_before": "at compile-time, and generates a conflict-free layout to store the coverage among a dynamic set of processes. Therefore, we guarantee the completeness by monitoring each active binary and accounting the coverage for each binary. Consequently, when determining whether an input is interesting, RATEL will not discard interesting inputs by mistake. B. Fragile Input Generation Most DBMS fuzzers leverage handcrafted AST as an input model to generate a greater proportion of valid inputs. Especially, Squirrel uses not only handcrafted AST but also customized semantic rules to ensure semantic validity.",
+      "context_after": "However, even with enormous manual efforts in modeling a new DBMS’s syntax, the model is still fragile due to the complexity of DBMS. DBMSs generally conform to ISO SQL standard [25], which contains 16 parts with thousands of pages and is still in evolution, therefore, constructing an input model is labor-consuming. Take PostgreSQL as an example: SQLsmith only targets PostgreSQL’s SELECT statement, and its grammar model already has 42 elements; SQLancer use over 8,000 lines of Java code to generate syntactically-correct test cases; Squirrel uses over 33,000 lines of C++ code to translate betw",
+      "section": "B Fragile Input Generation",
+      "page": 6,
+      "char_offset": 27579,
+      "cited_reference": {
+        "number": 17,
+        "text": "M. Rigger, “Sqlancer: Bugs found in database management systems,” 2020. [Online]. Available: https://www.manuelrigger.at/dbms-bugs",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Take PostgreSQL as an example: SQLsmith only targets PostgreSQL’s SELECT statement, and its grammar model already has 42 elements; SQLancer use over 8,000 lines of Java code to generate syntactically-correct test cases; Squirrel uses over 33,000 lines of C++ code to translate between SQL statement and AST.",
+      "context_before": "as an input model to generate a greater proportion of valid inputs. Especially, Squirrel uses not only handcrafted AST but also customized semantic rules to ensure semantic validity. These generation-based methods have shown significant effectiveness in DBMS bug discovery [8], [17]. However, even with enormous manual efforts in modeling a new DBMS’s syntax, the model is still fragile due to the complexity of DBMS. DBMSs generally conform to ISO SQL standard [25], which contains 16 parts with thousands of pages and is still in evolution, therefore, constructing an input model is labor-consuming.",
+      "context_after": "Another reason is the diversity of different DBMSs. In addition to basic database-related functionality, DBMSs often have their own specific features. For example, despite being a derived product from PostgreSQL, GaussDB still has vastly different SQL dialects. As Listing 1 shows, in terms of CREATE TABLE, GaussDB adds a special field for choosing its adaptive compression algorithms while shutting off functionalities of inheriting other tables and using extra methods. As a result, its proprietary input model is difficult to transfer to other DBMS. CREATE [ [ ... ] TABLE [ ... ] table_name ( [ {",
+      "section": "B Fragile Input Generation",
+      "page": 6,
+      "char_offset": 27997,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As Table I shows, besides RATEL, we used SQLsmith and SQLancer for grammar-based techniques with and without semantic checks.",
+      "context_before": "s glue. We wrote 2,323 lines of Rust for the overall algorithm. The online bug analyzer leverages ptrace to monitor behaviors of DBMS’s processes. Once an anomaly is triggered, the analyzer will collect the program context ( i.e. on-site stack trace and coverage), check whether the anomaly is newly-triggered based on the context, and disable core dump generation if the anomaly is duplicated. We implement the analyzer with 1,790 lines of code in Rust. A. Analysis of Results Aiming at improving the correctness of our production database GaussDB, we explored various mainstream fuzzing techniques.",
+      "context_after": "We also used the academic mutationbased fuzzer Squirrel. We further included several projects for comparison use: PostgreSQL, the original DBMS version where GaussDB forked from; Comdb2, another enterpriselevel DBMS from Bloomberg LP. We perform our evaluation on a 64-bit machine with 40 cores (Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz), 128 GiB of RAM, 2 * 7,200 rpm hard disk in RAID0, and Linux 5.5.13. All DBMSs were compiled with-O2 flags. We enable sanitizers for enhanced bug-detection ability on fuzzers. The initial seeds were collected from the built-in integration tests. We conducted eac",
+      "section": "A Analysis of Results",
+      "page": 7,
+      "char_offset": 34365,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "TABLE IICOVERAGE OF DBMS FOREACH FUZZER DBMS SQLsmith SQLancer Squirrel RATEL GaussDB 50,172 2,513 N/A 69,432 PostgreSQL 42,563 39,913 16,954 87,739 Comdb2 N/A 2,773 N/A 18,941 Table II presents the number of covered basic blocks for each fuzzer.",
+      "context_before": "ation tests. We conducted each experiment for 12 hours. We collected the seeds generated by those fuzzers, and dryran the seeds through our instrumentation to compute their coverage. Note that due to Comdb2 heavily uses Lua to extend its SQLite-based SQL dialect, we failed to adapt Squirrel and SQLsmith to it within a reasonable time. In addition, wedo many engineering efforts to adapt Squirrel to GaussDB. But after Squirrel covered 229 blocks, the fuzzer crashed due to its implementation issues. As a result, we skipped the evaluation of SQLsmith and Squirrel on Comdb2 and Squirrel on GaussDB.",
+      "context_after": "The second and third columns present the results for generation-based fuzzers, and the last two columns present the results for mutation-based fuzzer. From the table, we can see that: Simple blackbox fuzzers have good robustness in industrial practice. For example, the blackbox fuzzer SQLsmith outperformed the coverage-guided fuzzer Squirrel by 151.05%. It is because Squirrel is an academic prototype which lacks feedback from the industry while SQLsmith is designed for testing PostgreSQL from the beginning. Complex blackbox methods have limitations in industrial practice. For example, SQLance",
+      "section": "A Analysis of Results",
+      "page": 7,
+      "char_offset": 35662,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer covered more basic blocks than Squirrel on PostgreSQL but under-performed SQLsmith on GaussDB and PostgreSQL.",
+      "context_before": "ocks for each fuzzer. The second and third columns present the results for generation-based fuzzers, and the last two columns present the results for mutation-based fuzzer. From the table, we can see that: Simple blackbox fuzzers have good robustness in industrial practice. For example, the blackbox fuzzer SQLsmith outperformed the coverage-guided fuzzer Squirrel by 151.05%. It is because Squirrel is an academic prototype which lacks feedback from the industry while SQLsmith is designed for testing PostgreSQL from the beginning. Complex blackbox methods have limitations in industrial practice.",
+      "context_after": "The reason is that SQLancer is specialized in logic bug hunting and only uses a part of SQL features to construct its oracle violation. Its strength is discovering more types of bugs instead of coverage improvement. SQLancer also uses advanced test oracles to detect bugs, thus the CREATE TABLE statements generated by it have many fine-tuned parameters. It performs well on its supported DBMSs such as PostgreSQL and SQLite. However, when adapting to PostgreSQL-based GaussDB and SQLitebased Comdb2, the trivial differences in syntax prevented it from function normally. RATEL outperformed other fuz",
+      "section": "A Analysis of Results",
+      "page": 7,
+      "char_offset": 36488,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The reason is that SQLancer is specialized in logic bug hunting and only uses a part of SQL features to construct its oracle violation.",
+      "context_before": "t the results for mutation-based fuzzer. From the table, we can see that: Simple blackbox fuzzers have good robustness in industrial practice. For example, the blackbox fuzzer SQLsmith outperformed the coverage-guided fuzzer Squirrel by 151.05%. It is because Squirrel is an academic prototype which lacks feedback from the industry while SQLsmith is designed for testing PostgreSQL from the beginning. Complex blackbox methods have limitations in industrial practice. For example, SQLancer covered more basic blocks than Squirrel on PostgreSQL but under-performed SQLsmith on GaussDB and PostgreSQL.",
+      "context_after": "Its strength is discovering more types of bugs instead of coverage improvement. SQLancer also uses advanced test oracles to detect bugs, thus the CREATE TABLE statements generated by it have many fine-tuned parameters. It performs well on its supported DBMSs such as PostgreSQL and SQLite. However, when adapting to PostgreSQL-based GaussDB and SQLitebased Comdb2, the trivial differences in syntax prevented it from function normally. RATEL outperformed other fuzzers in terms of coverage and adaptation range. On GaussDB, it covers 0.38x, 26.63x more basic blocks than SQLsmith and SQLancer, respec",
+      "section": "A Analysis of Results",
+      "page": 7,
+      "char_offset": 36620,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer also uses advanced test oracles to detect bugs, thus the CREATE TABLE statements generated by it have many fine-tuned parameters.",
+      "context_before": "d fuzzer Squirrel by 151.05%. It is because Squirrel is an academic prototype which lacks feedback from the industry while SQLsmith is designed for testing PostgreSQL from the beginning. Complex blackbox methods have limitations in industrial practice. For example, SQLancer covered more basic blocks than Squirrel on PostgreSQL but under-performed SQLsmith on GaussDB and PostgreSQL. The reason is that SQLancer is specialized in logic bug hunting and only uses a part of SQL features to construct its oracle violation. Its strength is discovering more types of bugs instead of coverage improvement.",
+      "context_after": "It performs well on its supported DBMSs such as PostgreSQL and SQLite. However, when adapting to PostgreSQL-based GaussDB and SQLitebased Comdb2, the trivial differences in syntax prevented it from function normally. RATEL outperformed other fuzzers in terms of coverage and adaptation range. On GaussDB, it covers 0.38x, 26.63x more basic blocks than SQLsmith and SQLancer, respectively; On PostgreSQL, it covers 1.06x, 1.20x, 4.29x more basic blocks than SQLsmith, SQLancer, and Squirrel, respectively; On Comdb2, it covers 5.83x than SQLancer. Figure 6 demonstrates the number of basic blocks and",
+      "section": "A Analysis of Results",
+      "page": 7,
+      "char_offset": 36836,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M25",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "63x more basic blocks than SQLsmith and SQLancer, respectively; On PostgreSQL, it covers 1.",
+      "context_before": "part of SQL features to construct its oracle violation. Its strength is discovering more types of bugs instead of coverage improvement. SQLancer also uses advanced test oracles to detect bugs, thus the CREATE TABLE statements generated by it have many fine-tuned parameters. It performs well on its supported DBMSs such as PostgreSQL and SQLite. However, when adapting to PostgreSQL-based GaussDB and SQLitebased Comdb2, the trivial differences in syntax prevented it from function normally. RATEL outperformed other fuzzers in terms of coverage and adaptation range. On GaussDB, it covers 0.38x, 26.",
+      "context_after": "06x, 1.20x, 4.29x more basic blocks than SQLsmith, SQLancer, and Squirrel, respectively; On Comdb2, it covers 5.83x than SQLancer. Figure 6 demonstrates the number of basic blocks and paths growth for each fuzzer over the 12-hour trails in the first and the second column, respectively. The results illustrate that RATEL not only achieves high coverage, but also has advantages in speed. In specific, in Figure 6 (a) and Figure 6 (b), RATEL finds less basic blocks and paths than SQLsmith in the beginning. But with the coverage guided, RATEL quickly catches up with and surpasses it. In Figure 6 (c) an",
+      "section": "A Analysis of Results",
+      "page": 7,
+      "char_offset": 37300,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M26",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "29x more basic blocks than SQLsmith, SQLancer, and Squirrel, respectively; On Comdb2, it covers 5.",
+      "context_before": "nstead of coverage improvement. SQLancer also uses advanced test oracles to detect bugs, thus the CREATE TABLE statements generated by it have many fine-tuned parameters. It performs well on its supported DBMSs such as PostgreSQL and SQLite. However, when adapting to PostgreSQL-based GaussDB and SQLitebased Comdb2, the trivial differences in syntax prevented it from function normally. RATEL outperformed other fuzzers in terms of coverage and adaptation range. On GaussDB, it covers 0.38x, 26.63x more basic blocks than SQLsmith and SQLancer, respectively; On PostgreSQL, it covers 1.06x, 1.20x, 4.",
+      "context_after": "83x than SQLancer. Figure 6 demonstrates the number of basic blocks and paths growth for each fuzzer over the 12-hour trails in the first and the second column, respectively. The results illustrate that RATEL not only achieves high coverage, but also has advantages in speed. In specific, in Figure 6 (a) and Figure 6 (b), RATEL finds less basic blocks and paths than SQLsmith in the beginning. But with the coverage guided, RATEL quickly catches up with and surpasses it. In Figure 6 (c) and Figure 6 (d), RATEL covers more basic blocks and paths in PostgreSQL than other fuzzers in the beginning, and",
+      "section": "A Analysis of Results",
+      "page": 7,
+      "char_offset": 37405,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M27",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "83x than SQLancer.",
+      "context_before": "CREATE TABLE statements generated by it have many fine-tuned parameters. It performs well on its supported DBMSs such as PostgreSQL and SQLite. However, when adapting to PostgreSQL-based GaussDB and SQLitebased Comdb2, the trivial differences in syntax prevented it from function normally. RATEL outperformed other fuzzers in terms of coverage and adaptation range. On GaussDB, it covers 0.38x, 26.63x more basic blocks than SQLsmith and SQLancer, respectively; On PostgreSQL, it covers 1.06x, 1.20x, 4.29x more basic blocks than SQLsmith, SQLancer, and Squirrel, respectively; On Comdb2, it covers 5.",
+      "context_after": "Figure 6 demonstrates the number of basic blocks and paths growth for each fuzzer over the 12-hour trails in the first and the second column, respectively. The results illustrate that RATEL not only achieves high coverage, but also has advantages in speed. In specific, in Figure 6 (a) and Figure 6 (b), RATEL finds less basic blocks and paths than SQLsmith in the beginning. But with the coverage guided, RATEL quickly catches up with and surpasses it. In Figure 6 (c) and Figure 6 (d), RATEL covers more basic blocks and paths in PostgreSQL than other fuzzers in the beginning, and it maintains lead",
+      "section": "A Analysis of Results",
+      "page": 7,
+      "char_offset": 37503,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M28",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "100101102103 GaussDB(Paths)0200040006000800010000120001400016000 100101102103 GaussDB(Basic Blocks)010000200003000040000500006000070000 100101102103 PostgreSQL(Paths)0200040006000800010000120001400016000 100101102103 PostgreSQL(Basic Blocks)0100002000030000400005000060000700008000090000 100101102103 Comdb2(Paths)020040060080010001200140016001800 100101102103 Comdb2(Basic Blocks)05000100001500020000 SQLsmith SQLancer SQUIRREL Ratel(a) (b) (c) (e) (f)(d)Fig.",
+      "context_before": "12-hour trails in the first and the second column, respectively. The results illustrate that RATEL not only achieves high coverage, but also has advantages in speed. In specific, in Figure 6 (a) and Figure 6 (b), RATEL finds less basic blocks and paths than SQLsmith in the beginning. But with the coverage guided, RATEL quickly catches up with and surpasses it. In Figure 6 (c) and Figure 6 (d), RATEL covers more basic blocks and paths in PostgreSQL than other fuzzers in the beginning, and it maintains lead to the end. The results on Comdb2 in Figure 6 (e) and Figure 6 (f) also have similar trends.",
+      "context_after": "6. Path and basic block growth for each fuzzer over the 12-hour trials. The performance promotion of RATEL mainly dues to high-quality coverage guidance. With the coverage guidance, Squirrel and RATEL find more basic blocks significantly. For example, both the curves of Squirrel and RATEL in Figure 6 are always above the other black box fuzzers. Compared to Squirrel, RATEL still performs better because of the more precise feedback. Squirrel employs the feedback mechanism in AFL. Because DBMSs always have a large code scale, Squirrel extends the bitmap from 64 KB to 256KB. However, the AFL’s has",
+      "section": "A Analysis of Results",
+      "page": 7,
+      "char_offset": 38213,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M29",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In contrast, SQLancer, SQLsmith, and Squirrel did not find them.",
+      "context_before": "ll have a more serious impact. In contrast, RATEL implements conflict-free cross-binary feedback. It eliminates the collision by assigning unique counters for each basic block. To enrich block-based coverage, RATEL analyzes critical edges, whose source has multiple successors and destination has multiple predecessors and represents them with dummy basic blocks. Therefore, the improved coverage brings more opportunities for RATEL to find more vulnerabilities. B. Analysis of Bugs In our practice of fuzzing DBMSs, RATEL found 32, 42, and 5 previously-unknown bugs in GaussDB, PostgreSQL, and Comdb2.",
+      "context_after": "Out of these bugs, we take two interesting cases from these DBMSs. For both cases, we illustrate how the challenges in adapting coverage-feedback fuzzing to DBMSs impedes other fuzzer find these bugs. And more importantly, we also analyze how the solutions help RATEL to find them.-- Terminate the corresponding backend process. SELECT pg_cancel_backend(pg_backend_pid()) -- Now the connection is closed by the server. \\c non_existing_db -- Tries to reuse the closed connection. SELECT ... -- CRASH WITH NULL POINTER! Listing 2. The SQL statements triggering null pointer access. a) Null Pointer Acces",
+      "section": "B Analysis of Bugs",
+      "page": 8,
+      "char_offset": 40327,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M30",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Both SQLancer and SQLsmith did not find this bug.",
+      "context_before": "ows the SQL statements which could trigger the bug. This bug involves multiple interactions between the server and the client, which could be triggered with three statements. In specific, first the SQL statement SELECT pg_cancel_backend(pg_backend_PID ()) is executed, and the backend process of the server will shut down and close the connection. Next, the client tries to switch to another database by attempting to reuse the old connection, but the connection is closed and set to NULL pointer. Next, executing any SQL statements tries to access the NULL connection, crashing the client as a result.",
+      "context_after": "They use blackbox methods, which build syntactically correct SQLs based on ASTs. However, queries like pg_cancel_backend are DBMS specific extension functions. Due to the incomplete syntax modeling, SQLancer and SQLsmith can never generate such queries. Squirrel uses syntax-based mutation to generate SQL statements and enforces strict language-validity: only the inputs which pass its AST parser could be mutated, and the generated query is re-parsed to ensure the validity. However, the incomplete modeling of the syntax prevented it from mutating the input at all. Consequently, neither can Squir",
+      "section": "B Analysis of Bugs",
+      "page": 8,
+      "char_offset": 41620,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M31",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Due to the incomplete syntax modeling, SQLancer and SQLsmith can never generate such queries.",
+      "context_before": "t SELECT pg_cancel_backend(pg_backend_PID ()) is executed, and the backend process of the server will shut down and close the connection. Next, the client tries to switch to another database by attempting to reuse the old connection, but the connection is closed and set to NULL pointer. Next, executing any SQL statements tries to access the NULL connection, crashing the client as a result. Both SQLancer and SQLsmith did not find this bug. They use blackbox methods, which build syntactically correct SQLs based on ASTs. However, queries like pg_cancel_backend are DBMS specific extension functions.",
+      "context_after": "Squirrel uses syntax-based mutation to generate SQL statements and enforces strict language-validity: only the inputs which pass its AST parser could be mutated, and the generated query is re-parsed to ensure the validity. However, the incomplete modeling of the syntax prevented it from mutating the input at all. Consequently, neither can Squirrel trigger this bug. Different from these fuzzers, RATEL addresses the challenge of fragile input generation by not only adapting PostgreSQL with a customized dictionary, but also preserving the syntaxbased mutation with relaxed syntax checks. Because",
+      "section": "B Analysis of Bugs",
+      "page": 8,
+      "char_offset": 41828,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M32",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For blackbox fuzzers like SQLsmith and SQLancer, they have a low possibility of generating SQL statements to cover these states without feedback.",
+      "context_before": "te(..., tzname,...); ..... } osql_sess_t *osql_sess_create(){ if(tzname) Crash due to invalid memory access. strncpy0(sess->tzname, tzname, sizeof (sess->tzname)); } Listing 3. The details of stack-use-after-scope on address The bug is hidden in deep logic, which is triggered only when the specific basic blocks are covered and Comdb2 reaches a unique state after querying a particular ALTER statement. As shown in Fig. 7, the call stack of the bug is rather deep. In specific, it contains three levels, namely parsing, logic, and networking. Covering the basic blocks on the top of the stack is hard.",
+      "context_after": "Although Squirrel also guides fuzzing with coverage, the spare coverage and its strict syntax checks lower the probability to trigger the bug. In contrast, RATEL addresses the challenges of imprecise coverage collection with inter-binary linkage and bijectiveblock mapping. With the robustness-oriented syntax-based mutation, it expanded coverage and finally triggered this bug. More importantly, this bug requires Comdb2 in a particular state: just passing the syntax checks is not enough, the table must be present and its schema must be valid for the ALTER statement. Therefore, simply re-executin",
+      "section": "B Analysis of Bugs",
+      "page": 9,
+      "char_offset": 45734,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M33",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, Squirrel and SQLancer drop databases after each query.",
+      "context_before": "hen collecting coverage. If other clients send queries during the interval, the collected coverage is imprecise. In our practice, we can only set up the equivelant number of servers as the number of clients for Squirrel, which is ineffective. We believe it is a valuable research topic and demand to be tackled. c) Low Efficiency in Dropping Databases: Dropping database decreases the noise for stateful DBMSs, which assists bug deduplication and helps fuzzer performs more stable. However, frequent dropping will ignore the bugs which only can be triggered in some states and bring a lot of overhead.",
+      "context_after": "We occasionally drop databases to explore more states and lower the overhead. Instead, when bugs are triggered, we investigate the root cause of bugs with on-line triage. d) Bugs Hidden in Heterogeneous Platforms: In practice, we find bugs hidden in heterogeneous platforms. For example, GaussDB is mainly deployed in Huawei Kunpeng powered ARM servers. We find that some bugs like misaligned pointer accesses have minor performance impacts on x86 platforms, but can cause hardware exceptions on ARM. Undefined behaviors might seem trivial on some certain platforms, they can still lead to major bugs i",
+      "section": "B Analysis of Bugs",
+      "page": 9,
+      "char_offset": 48581,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M27"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:13:36Z",
+    "is_model_written": true,
+    "summary": "An industry report on bringing coverage-guided fuzzing to enterprise DBMSs at Huawei and Bloomberg. The authors note that industry has used black-box fuzzing for decades while coverage guidance has produced impressive research gains that rarely reach enterprise systems, given their complexity and distributed nature. Testing GaussDB and Comdb2 they hit challenges at all three stages -- collecting precise coverage, optimising fuzzing performance, and analysing root causes -- and built Ratel to address them, discovering 32, 42 and 5 unknown bugs in GaussDB, Comdb2 and PostgreSQL.",
+    "narrative": "SQLancer is one of the three fuzzers the practice was built around and measured against, described as a black-box logic-bug fuzzer that combines three oracle strategies and credited with finding over 400 bugs in two years.",
+    "roles": {
+      "M1": "result_comparison",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "result_comparison",
+      "M6": "baseline",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "incidental",
+      "M10": "definition",
+      "M11": "definition",
+      "M12": "baseline",
+      "M13": "background",
+      "M14": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is one of the fuzzers run for comparison; Ratel is its own coverage-guided tool."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Nothing shown claims to extend PQS, NoREC or TLP."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M1",
+          "M5",
+          "M6"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "Compared to industrial black box fuzzers SQLsmith and SQLancer, as well as coverage-guided academic fuzzer Squirrel, RATEL covered 38.",
+            "section": null,
+            "page": 1
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "Compared to SQLsmith, SQLancer, and Squirrel, it covered 38.",
+            "section": "I INTRODUCTION",
+            "page": 2
+          },
+          {
+            "mention_id": "M6",
+            "sentence": "SQLsmith, SQLancer, and Squirrel, which are chosen for our industry practice.",
+            "section": "II BACKGROUND",
+            "page": 2
+          }
+        ],
+        "reasoning": "SQLancer is one of three fuzzers Ratel is measured against, with basic block coverage reported relative to each."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is described as an industrial black-box fuzzer and credited with its bug count, but not called the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icsip61881_2024_10671554.json b/_data/papers/paper_doi_10_1109_icsip61881_2024_10671554.json
new file mode 100644
index 0000000..97d9030
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icsip61881_2024_10671554.json
@@ -0,0 +1,614 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:54:33Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icsip61881.2024.10671554",
+    "title": "A Review of Fuzz Testing for Configuration-Sensitive Software",
+    "authors": [
+      "Lang Chu",
+      "Minhuan Huang",
+      "Xiang Li",
+      "Yuanping Nie"
+    ],
+    "year": 2024,
+    "venue": "IEEE International Conference on Signal and Image Processing",
+    "doi": "10.1109/icsip61881.2024.10671554",
+    "arxiv_id": null,
+    "s2_paper_id": "ad42e73e4c115542ef4ec708b9f01f39a60de8ff",
+    "url": "https://doi.org/10.1109/icsip61881.2024.10671554",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icsip61881.2024.10671554",
+      "retrieved_at": "2026-09-08T14:54:33Z",
+      "chars": 56214,
+      "content_sha256": "sha256:126949d98ae2ec38f47eef88826485496dcc74c5cdf276ba151bac3f644a7171"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 11,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1975
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 5202
+      },
+      {
+        "number": "A",
+        "title": "Fuzz testing",
+        "start": 6451
+      },
+      {
+        "number": "B",
+        "title": "Software Configuration Parameter Testing",
+        "start": 13210
+      },
+      {
+        "number": "III",
+        "title": "CONFIGURATION HANDLING INFUZZING",
+        "start": 18746
+      },
+      {
+        "number": "A",
+        "title": "Strategy of Simultaneously Fuzzing Configuration and Input Files",
+        "start": 20563
+      },
+      {
+        "number": "B",
+        "title": "Strategy of Separately Fuzzing Configuration and Input",
+        "start": 26470
+      },
+      {
+        "number": "IV",
+        "title": "FUTURE RESEARCH DIRECTIONS AND CONCLUSION",
+        "start": 35102
+      },
+      {
+        "number": "J",
+        "title": "M. Atlee, T. Bultan, and J. Whittle, Eds. IEEE",
+        "start": 37460
+      },
+      {
+        "number": "C",
+        "title": "Cadar and X. Zhang, Eds. ACM, 2021, pp.",
+        "start": 38444
+      },
+      {
+        "number": "A",
+        "title": "Sadeghi, and D. Teuchert, “NAUTILUS: fishing",
+        "start": 39208
+      },
+      {
+        "number": "S",
+        "title": "Schumilo, S. W ¨orner, and T. Holz, “GRIMOIRE:",
+        "start": 39609
+      },
+      {
+        "number": "B",
+        "title": "Lee, “HFL: hybrid fuzzing on the linux kernel,” in",
+        "start": 40672
+      },
+      {
+        "number": "A",
+        "title": "Roychoudhury, “Stateful greybox fuzzing,” in 31st",
+        "start": 41947
+      },
+      {
+        "number": "D",
+        "title": "Wu, “SQUIRREL: testing database management",
+        "start": 42628
+      },
+      {
+        "number": "P",
+        "title": "Traynor, Eds. USENIX Association, 2019, pp.",
+        "start": 43278
+      },
+      {
+        "number": "S",
+        "title": "Hakak, “Fuzzing vulnerability discovery techniques:",
+        "start": 43667
+      },
+      {
+        "number": "P",
+        "title": "Gr ¨unbacher, Eds. ACM, 2014, pp. 725–730. [Online].",
+        "start": 45160
+      },
+      {
+        "number": "L",
+        "title": "Grunske, and M. Whalen, Eds. IEEE Computer",
+        "start": 45461
+      },
+      {
+        "number": "M",
+        "title": "D. Penta, and T. N. Nguyen, Eds. IEEE",
+        "start": 45850
+      },
+      {
+        "number": "M",
+        "title": "T. Vechev, “Learning to fuzz from symbolic execution",
+        "start": 46045
+      },
+      {
+        "number": "A",
+        "title": "Orso, and M. P. Robillard, Eds. IEEE / ACM,",
+        "start": 46675
+      },
+      {
+        "number": "Y",
+        "title": "Xiang, and K. Ren, “Csi-fuzz: Full-speed edge",
+        "start": 47947
+      },
+      {
+        "number": "B",
+        "title": "I. P. Rubinstein, “Beyond the coverage plateau:",
+        "start": 51782
+      },
+      {
+        "number": "L",
+        "title": "Szekeres, Eds. ACM, 2023, pp. 47–55. [Online].",
+        "start": 52033
+      },
+      {
+        "number": "A",
+        "title": "Arya, “Fuzzbench: an open fuzzer benchmarking",
+        "start": 52191
+      },
+      {
+        "number": "G",
+        "title": "Gousios, M. Chechik, and M. D. Penta, Eds.",
+        "start": 52444
+      },
+      {
+        "number": "P",
+        "title": "Devanbu, M. B. Cohen, and T. Zimmermann,",
+        "start": 52998
+      },
+      {
+        "number": "M",
+        "title": "Backes, and X. Wang, Eds. ACM, 2018, pp.",
+        "start": 53376
+      },
+      {
+        "number": "F",
+        "title": "Zhang, “Detecting critical bugs in SMT solvers using",
+        "start": 55457
+      },
+      {
+        "number": "P",
+        "title": "Devanbu, M. B. Cohen, and T. Zimmermann,",
+        "start": 55715
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "B. P. Miller, L. Fredriksen, and B. So, “An empirical study of the reliability of UNIX utilities,” Commun. ACM, vol. 33, no. 12, pp. 32–44, 1990. [Online]. Available: https://doi.org/10.1145/96267.96279",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "J. Choi, J. Jang, C. Han, and S. K. Cha, “Grey-box concolic testing on binary code,” in Proceedings of the 41st International Conference on Software Engineering, ICSE 2019, Montreal, QC, Canada, May 25-31, 2019, J. M. Atlee, T. Bultan, and J. Whittle, Eds. IEEE / ACM, 2019, pp. 736–747. [Online]. Available: https://doi.org/10.1109/ICSE.2019.00082",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "C. Aschermann, S. Schumilo, A. Abbasi, and T. Holz, “Ijon: Exploring deep state spaces via fuzzing,” in SP 2020, San Francisco, CA, USA, May 18-21, 2020. IEEE, 2020, pp. 1597–1612. [Online]. Available: https://doi.org/10.1109/SP40000.2020.00117",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "S. Gan, C. Zhang, P. Chen, B. Zhao, X. Qin, D. Wu, and Z. Chen, “GREYONE: data flow sensitive fuzzing,” in29th USENIX Security Symposium, USENIX Security 2020, August 12-14, 2020, S. Capkun and F. Roesner, Eds. USENIX Association, 2020, pp. 2577–2594. [Online]. Available: https://www.usenix.org/conference/ usenixsecurity20/presentation/gan",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "A. Herrera, H. Gunadi, S. Magrath, M. Norrish, M. Payer, and A. L. Hosking, “Seed selection for successful fuzzing,” in ISSTA ’21: 30th ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, Denmark, July 11-17, 2021, C. Cadar and X. Zhang, Eds. ACM, 2021, pp. 230–243. [Online]. Available: https://doi.org/10.1145/ 3460319.3464795",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "M. B ¨ohme, V. Pham, and A. Roychoudhury, “Coveragebased greybox fuzzing as markov chain,” IEEE Trans. Software Eng., vol. 45, no. 5, pp. 489–506, 2019. [Online]. Available: https://doi.org/10.1109/TSE.2017. 2785841",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "M. B ¨ohme, V. Pham, M. Nguyen, and A. Roychoudhury, “Directed greybox fuzzing,” in Proceedings of the 2017 ACMSIGSAC Conference on Computer and Communications Security, CCS 2017, Dallas, TX, USA, October 30 - November 03, 2017, B. Thuraisingham,D. Evans, T. Malkin, and D. Xu, Eds. ACM, 2017, pp. 2329–2344. [Online]. Available: https://doi.org/10.1145/ 3133956.3134020",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "C. Aschermann, T. Frassetto, T. Holz, P. Jauernig, A. Sadeghi, and D. Teuchert, “NAUTILUS: fishing for deep bugs with grammars,” in 26th Annual Network and Distributed System Security Symposium, NDSS 2019, San Diego, California, USA, February 24-27, 2019. The Internet Society, 2019. [Online]. Available: https://www.ndss-symposium.org/ndss-paper/ nautilus-fishing-for-deep-bugs-with-grammars/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "T. Blazytko, C. Aschermann, M. Schl ¨ogel, A. Abbasi, S. Schumilo, S. W ¨orner, and T. Holz, “GRIMOIRE: synthesizing structure while fuzzing,” in 28th USENIX Security Symposium, USENIX Security 2019, Santa Clara, CA, USA, August 14-16, 2019, N. Heninger and P. Traynor, Eds. USENIX Association, 2019, pp. 1985–2002. [Online]. Available: https://www.usenix.org/ conference/usenixsecurity19/presentatio",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "V. Pham, M. B ¨ohme, A. E. Santosa, A. R. Caciulescu, and A. Roychoudhury, “Smart greybox fuzzing,” IEEE Trans. Software Eng., vol. 47, no. 9, pp. 1980–1997, 2021. [Online]. Available: https://doi.org/10.1109/TSE. 2019.2941681",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "S. Schumilo, C. Aschermann, R. Gawlik, S. Schinzel, and T. Holz, “kafl: Hardware-assisted feedback fuzzing for OS kernels,” in 26th USENIX Security Symposium, USENIX Security 2017, Vancouver, BC, Canada, August 16-18, 2017, E. Kirda and T. Ristenpart, Eds. USENIX Association, 2017, pp. 167–182. [Online]. Available: https://www.usenix.org/conference/usenixsecurity17/ technical-sessions/presentation",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "K. Kim, D. R. Jeong, C. H. Kim, Y. Jang, I. Shin, and B. Lee, “HFL: hybrid fuzzing on the linux kernel,” in 27th Annual Network and Distributed System Security Symposium, NDSS 2020, San Diego, California, USA, February 23-26, 2020. The Internet Society, 2020. [Online]. Available: https://www.ndss-symposium.org/ ndss-paper/hfl-hybrid-fuzzing-on-the-linux-kernel/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "D. R. Jeong, K. Kim, B. Shivakumar, B. Lee, and I. Shin, “Razzer: Finding kernel race bugs through fuzzing,” in Privacy, SP 2019, San Francisco, CA, USA, May 19-23, 2019. IEEE, 2019, pp. 754–768. [Online]. Available: https://doi.org/10.1109/SP.2019.00017",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "W. Xu, H. Moon, S. Kashyap, P. Tseng, and T. Kim, “Fuzzing file systems via two-dimensional input space exploration,” in Privacy, SP 2019, San Francisco, CA, USA, May 19-23, 2019. IEEE, 2019, pp. 818–834. [Online]. Available: https://doi.org/10.1109/SP.2019.00035",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "S. Schumilo, C. Aschermann, A. Abbasi, S. W ¨orner, and T. Holz, “HYPER-CUBE: high-dimensional hypervisor fuzzing,” in 27th Annual Network and Distributed System Security Symposium, NDSS 2020, San Diego, California, USA, February 23395 26, 2020. The Internet Society, 2020. [Online]. Available: https://www.ndss-symposium.org/ndss-paper/ hyper-cube-high-dimensional-hypervisor-fuzzing/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "J. Ba, M. B ¨ohme, Z. Mirzamomen, and A. Roychoudhury, “Stateful greybox fuzzing,” in 31st USENIX Security Symposium, USENIX Security 2022, Boston, MA, USA, August 10-12, 2022, K. R. B. Butler and K. Thomas, Eds. USENIX Association, 2022, pp. 3255–3272. [Online]. Available: https://www.usenix.org/ conference/usenixsecurity22/presentation/ba",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 2020, pp. 667–682. [Online]. Available: https://www.usenix.org/conference/ osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 18,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “SQUIRREL: testing database management systems with language validity and coverage feedback,” in CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event, USA, November 9-13, 2020, J. Ligatti, X. Ou, J. Katz, and G. Vigna, Eds. ACM, 2020, pp. 955–970. [Online]. Available: https://doi.org/10.1145/3372297.3417260",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "T. Kim, C. H. Kim, J. Rhee, F. Fei, Z. Tu, G. Walkup, X. Zhang, X. Deng, and D. Xu, “Rvfuzzer: Finding input validation bugs in robotic vehicles through control-guided testing,” in 28th USENIX Security Symposium, USENIX Security 2019, Santa Clara, CA, USA, August 14-16, 2019, N. Heninger and P. Traynor, Eds. USENIX Association, 2019, pp. 425–442. [Online]. Available: https://www.usenix.org/ confer",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "X. Zhu, S. Wen, S. Camtepe, and Y. Xiang, “Fuzzing: A survey for roadmap,” ACM Comput. Surv., vol. 54, no. 11s, pp. 230:1–230:36, 2022. [Online]. Available: https://doi.org/10.1145/3512345",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "C. Beaman, M. Redbourne, J. D. Mummery, and S. Hakak, “Fuzzing vulnerability discovery techniques: Survey, challenges and future directions,” Comput. Secur., vol. 120, p. 102813, 2022. [Online]. Available: https://doi.org/10.1016/j.cose.2022.102813",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "A. Fioraldi, D. C. Maier, H. Eißfeldt, and M. Heuse, “AFL++: Combining incremental steps of fuzzing research,” in 14th USENIX Workshop on Offensive Technologies, WOOT 2020, August 11, 2020, Y. Yarom and S. Zennou, Eds. USENIX Association, 2020. [Online]. Available: https://www.usenix.org/conference/ woot20/presentation/fioraldi",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "M. Zalewski, “american fuzzy lopa security-oriented fuzzer.” [Online]. Available: https://github.com/google/ AFL",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "“libfuzzer – a library for coverage-guided fuzz testing.” [Online]. Available: https://github.com/llvm-mirror/llvm/ blob/master/docs/LibFuzzer.rst",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "“honggfuzz.” [Online]. Available: https://github.com/ google/honggfuzz",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "H. J. Abdelnur, R. State, and O. Festor, “Kif: a stateful SIP fuzzer,” in Principles, Systems and Applications of IP Telecommunications. First International Conference, IPTComm 2007, New York, NY, USA, July 19-20, 2007. Proceedings, G. W. Bond and H. Schulzrinne, Eds. ACM, 2007, pp. 47–56. [Online]. Available: https://doi.org/10.1145/1326304.1326313",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "K. Dewey, J. Roesch, and B. Hardekopf, “Language fuzzing using constraint logic programming,” in ACM/IEEE International Conference on Automated Software Engineering, ASE ’14, Vasteras, Sweden September 15 - 19, 2014, I. Crnkovic, M. Chechik, and P. Gr ¨unbacher, Eds. ACM, 2014, pp. 725–730. [Online]. Available: https://doi.org/10.1145/2642937.2642963",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "——, “Fuzzing the rust typechecker using CLP (T),” in 30th IEEE/ACM International Conference on Automated Software Engineering, ASE 2015, Lincoln, NE, USA, November 9-13, 2015, M. B. Cohen, L. Grunske, and M. Whalen, Eds. IEEE Computer Society, 2015, pp. 482–493. [Online]. Available: https: //doi.org/10.1109/ASE.2015.65",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "P. Godefroid, H. Peleg, and R. Singh, “Learn&fuzz: machine learning for input fuzzing,” in Proceedings of the 32nd IEEE/ACM International Conference on Automated Software Engineering, ASE 2017, Urbana, IL, USA, October 30 - November 03, 2017, G. Rosu, M. D. Penta, and T. N. Nguyen, Eds. IEEE Computer Society, 2017, pp. 50–59. [Online]. Available: https://doi.org/10.1109/ASE.2017.8115618",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "J. He, M. Balunovic, N. Ambroladze, P. Tsankov, and M. T. Vechev, “Learning to fuzz from symbolic execution with application to smart contracts,” in Proceedings of the 2019 ACMSIGSAC Conference on Computer and Communications Security, CCS 2019, London, UK, November 11-15, 2019, L. Cavallaro, J. Kinder, X. Wang, and J. Katz, Eds. ACM, 2019, pp. 531–548. [Online]. Available: https://doi.org/10.1145/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "P. Liu, X. Zhang, M. Pistoia, Y. Zheng, M. Marques, and L. Zeng, “Automatic text input generation for mobile testing,” in Proceedings of the 39th International Conference on Software Engineering, ICSE 2017, Buenos Aires, Argentina, May 20-28, 2017, S. Uchitel, A. Orso, and M. P. Robillard, Eds. IEEE / ACM, 2017, pp. 643–653. [Online]. Available: https://doi.org/ 10.1109/ICSE.2017.65",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "X. Liu, X. Li, R. Prajapati, and D. Wu, “Deepfuzz: Automatic generation of syntax valid C programs for fuzz testing,” in The Thirty-Third AAAI Conference on Artificial Intelligence, AAAI 2019, The ThirtyFirst Innovative Applications of Artificial Intelligence Conference, IAAI 2019, The Ninth AAAI Symposium on Educational Advances in Artificial Intelligence, EAAI 2019, Honolulu, Hawaii, USA, Januar",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "J. Wang, B. Chen, L. Wei, and Y. Liu, “Skyfire: Data-driven seed generation for fuzzing,” in 2017 IEEE Symposium on Security and Privacy, SP 2017, San Jose, CA, USA, May 22-26, 2017. IEEE Computer Society, 2017, pp. 579–594. [Online]. Available: https: //doi.org/10.1109/SP.2017.23",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "S. Gan, C. Zhang, X. Qin, X. Tu, K. Li, Z. Pei, and Z. Chen, “Collafl: Path sensitive fuzzing,” in 2018, Proceedings, 21-23 May 2018, San Francisco, California, USA. IEEE Computer Society, 2018, pp. 679–696. [Online]. Available: https://doi.org/10.1109/SP. 2018.00040",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "X. Zhu, X. Feng, X. Meng, S. Wen, S. Camtepe, Y. Xiang, and K. Ren, “Csi-fuzz: Full-speed edge tracing using coverage sensitive instrumentation,” IEEE Trans. Dependable Secur. Comput., vol. 19, no. 2, pp. 912–923, 2022. [Online]. Available: https://doi.org/10. 1109/TDSC.2020.3008826",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "P. Godefroid, A. Kiezun, and M. Y. Levin, “Grammarbased whitebox fuzzing,” in Proceedings of the ACMSIGPLAN 2008 Conference on Programming Language Design and Implementation, Tucson, AZ, USA, June 7-13, 2008, R. Gupta and S. P. Amarasinghe, Eds. ACM, 2008, pp. 206–215. [Online]. Available: https://doi.org/10.1145/1375581.1375607",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "H. Huang, P. Yao, R. Wu, Q. Shi, and C. Zhang, “Pangolin: Incremental hybrid fuzzing with polyhedral path abstraction,” in and Privacy, SP 2020, San Francisco, CA, USA, May 18-21, 2020. IEEE, 2020, pp. 1613–1627. [Online]. Available: https://doi.org/10.1109/SP40000.2020.00063",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "S. Nagy and M. Hicks, “Full-speed fuzzing: Reducing fuzzing overhead through coverage-guided tracing,” in SP 2019, San Francisco, CA, USA, May 19-23, 2019. IEEE, 2019, pp. 787–802. [Online]. Available: https://doi.org/10.1109/SP.2019.00069",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "C. Miller, “Fuzz by number: More data about fuzzing than you ever wanted to know,” Proceedings of the CanSecWest, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "C. Pacheco, S. K. Lahiri, M. D. Ernst, and T. Ball, “Feedback-directed random test generation,” in 29th International Conference on Software Engineering (ICSE 2007), Minneapolis, MN, USA, May 20-26, 2007. IEEE Computer Society, 2007, pp. 75–84. [Online]. Available: https://doi.org/10.1109/ICSE.2007.37",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "M. Xu, S. Kashyap, H. Zhao, and T. Kim, “Krace: Data race fuzzing for kernel file systems,” in 2020 IEEE Symposium on Security and Privacy, SP 2020, San Francisco, CA, USA, May 18-21, 2020. IEEE, 2020, pp. 1643–1660. [Online]. Available: https://doi.org/10. 1109/SP40000.2020.00078",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "K. Pei, Y. Cao, J. Yang, and S. Jana, “Deepxplore: automated whitebox testing of deep learning systems,”Commun. ACM, vol. 62, no. 11, pp. 137–145, 2019. [Online]. Available: https://doi.org/10.1145/3361566",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "“Ffmpeg.” [Online]. Available: https://ffmpeg.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "T. Xu and Y. Zhou, “Systems approaches to tackling configuration errors: A survey,” ACM Comput. Surv., vol. 47, no. 4, pp. 70:1–70:41, 2015. [Online]. Available: https://doi.org/10.1145/2791577",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "C. Nie and H. Leung, “A survey of combinatorial testing,” ACM Comput. Surv., vol. 43, no. 2, pp. 11:1–11:29, 2011. [Online]. Available: https://doi.org/ 10.1145/1883612.1883618",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "D. R. Kuhn and M. J. Reilly, “An investigation of the applicability of design of experiments to software testing,” in 27th Annual NASA Goddard/IEEE Software Engineering Workshop, 2002. Proceedings. IEEE, 2002, pp. 91–95.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "D. R. Kuhn, D. R. Wallace, and A. M. Gallo, “Software fault interactions and implications for software testing,” IEEE transactions on software engineering, vol. 30, no. 6, pp. 418–421, 2004.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "K. Z. Bell and M. A. V ouk, “On effectiveness of pairwise methodology for testing network-centric software,” in2005 International Conference on Information and Communication Technology. IEEE, 2005, pp. 221–235.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "D. M. Cohen, S. R. Dalal, M. L. Fredman, and G. C. Patton, “The aetg system: An approach to testing based on combinatorial design,” IEEE Transactions on Software Engineering, vol. 23, no. 7, pp. 437–444, 1997.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "M. B. Cohen, P. B. Gibbons, W. B. Mugridge, and C. J. Colbourn, “Constructing test suites for interaction testing,” in 25th International Conference on Software Engineering, 2003. Proceedings. IEEE, 2003, pp. 38– 48.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Z. Zhang, G. Klees, E. Wang, M. Hicks, and S. Wei, “Fuzzing configurations of program options-RCR report,” ACM Trans. Softw. Eng. Methodol., vol. 32, no. 2, pp. 55:1–55:3, 2023. [Online]. Available: https://doi.org/10.1145/3580601",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "W. Gao, V. Pham, D. Liu, O. Chang, T. Murray, and B. I. P. Rubinstein, “Beyond the coverage plateau: A comprehensive study of fuzz blockers (registered report),” in Proceedings of the 2nd International Fuzzing Workshop, FUZZING 2023, Seattle, WA, USA, 17 July 2023, M. B ¨ohme, Y. Noller, B. Ray, and L. Szekeres, Eds. ACM, 2023, pp. 47–55. [Online]. Available: https://doi.org/10.1145/3605157.360517",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "J. Metzman, L. Szekeres, L. Simon, R. Sprabery, and A. Arya, “Fuzzbench: an open fuzzer benchmarking platform and service,” in ESEC/FSE ’21: 29th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Athens, Greece, August 23-28, 2021, D. Spinellis, G. Gousios, M. Chechik, and M. D. Penta, Eds. ACM, 2021, pp. 1393–1403. [Online]. Available: ht",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "“Fuzz introspector – introspect, extend and optimise 397 fuzzers.” [Online]. Available: https://github.com/ossf/ fuzz-introspector",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "S. Song, C. Song, Y. Jang, and B. Lee, “Crfuzz: fuzzing multi-purpose programs through input validation,” in ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, P. Devanbu, M. B. Cohen, and T. Zimmermann, Eds. ACM, 2020, pp. 690–700. [Online]. Available: https://doi.org/10.1145/3368",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "G. Klees, A. Ruef, B. Cooper, S. Wei, and M. Hicks, “Evaluating fuzz testing,” in Proceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security, CCS 2018, Toronto, ON, Canada, October 15-19, 2018, D. Lie, M. Mannan, M. Backes, and X. Wang, Eds. ACM, 2018, pp. 2123–2138. [Online]. Available: https://doi.org/10.1145/ 3243734.3243804",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "H. Dai, C. Murphy, and G. E. Kaiser, “CONFU: configuration fuzzing testing framework for software vulnerability detection,” Int. J. Secur. Softw. Eng., vol. 1, no. 3, pp. 41–55, 2010. [Online]. Available: https://doi.org/10.4018/jsse.2010070103",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "C. R. Ramakrishnan and R. Sekar, “Model-based analysis of configuration vulnerabilities,” J. Comput. Secur., vol. 10, no. 1/2, pp. 189–209, 2002. [Online]. Available: https://doi.org/10.3233/jcs-2002-101-209",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "A. Lee, I. Ariq, Y. Kim, and M. Kim, “POWER: program option-aware fuzzer for high bug detection ability,” in 15th IEEE Conference on Software Testing, Verification and Validation, ICST 2022, Valencia, Spain, April 4-14, 2022. IEEE, 2022, pp. 220–231. [Online]. Available: https://doi.org/10.1109/ICST53961.2022.00032",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "D. Wang, Y. Li, Z. Zhang, and K. Chen, “Carpetfuzz: Automatic program option constraint extraction from documentation for fuzzing,” in 32nd USENIX Security Symposium, USENIX Security 2023, Anaheim, CA, USA, August 9-11, 2023, J. A. Calandrino and C. Troncoso, Eds. USENIX Association, 2023, pp. 1919–1936. [Online]. Available: https://www.usenix.org/conference/ usenixsecurity23/presentation/wang-daw",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "J. Li, S. Li, K. Li, F. Luo, H. Yu, S. Li, and X. Li, “Ecfuzz: Effective configuration fuzzing for large-scale systems,” in Proceedings of the 46th IEEE/ACM International Conference on Software Engineering, ICSE 2024, Lisbon, Portugal, April 14-20, 2024. ACM, 2024, pp. 48:1–48:12. [Online]. Available: https://doi.org/10.1145/3597503.3623315",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "P. Yao, H. Huang, W. Tang, Q. Shi, R. Wu, and C. Zhang, “Fuzzing SMT solvers via twodimensional input space exploration,” in ISSTA ’21: 30th ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, Denmark, July 11-17, 2021, C. Cadar and X. Zhang, Eds. ACM, 2021, pp. 322–335. [Online]. Available: https://doi.org/10.1145/3460319.3464803",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "M. N. Mansur, M. Christakis, V. W ¨ustholz, and F. Zhang, “Detecting critical bugs in SMT solvers using blackbox mutational fuzzing,” in ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, P. Devanbu, M. B. Cohen, and T. Zimmermann, Eds. ACM, 2020, pp. 701–712. [Online]. Available: ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "D. Winterer, C. Zhang, and Z. Su, “On the unusual effectiveness of type-aware operator mutations for testing SMT solvers,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 193:1–193:25, 2020. [Online]. Available: https://doi.org/10.1145/3428261",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Z. Wang, B. Liblit, and T. Reps, “Tofu: Target-oriented fuzzer,” arXiv preprint arXiv:2004.14375, 2020. 398",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 17,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 2020, pp. 667–682. [Online]. Available: https://www.usenix.org/conference/ osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[17, 18]",
+      "technique": "pqs",
+      "sentence": ", HyperCube[15]), network protocols[15, 16], database systems[17, 18], and autonomous vehicles (e.",
+      "context_before": "ponses to these inputs to provoke potential crashes or other abnormal behaviors, thereby discovering more security vulnerabilities.Tofurther enhance the effectiveness of fuzz testing, researchers have conducted extensive work on feedback optimization[2– 4], corpus optimization[5], seed scheduling optimization[6, 7], and test case generation optimization[8–10]. Additionally, to expand the application scenarios of fuzz testing, researchers have developed a series of specialized fuzz testing tools, covering various critical domains such as operating systems[11– 14], virtual machine managers (e.g.",
+      "context_after": "g., RVFUZZER[19]), to meet the security testing needs in these areas. In general, a significant amount of literature (e.g., [20, 21]) has extensively surveyed the aforementioned aspects of work. Therefore, this paper does not focus on these works. Instead, it concentrates on a factor that is often overlooked but crucial in testing practicessoftware configuration. Particularly, fuzz testing of configuration-sensitive software, where the choice of configurations significantly impacts testing outcomes, has garnered considerable attention. Several studies have investigated the vital influence of s",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3249,
+      "cited_reference": {
+        "number": 17,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 2020, pp. 667–682. [Online]. Available: https://www.usenix.org/conference/ osdi20/pr",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:02:36Z",
+    "is_model_written": true,
+    "summary": "A review of fuzz testing for configuration-sensitive software -- systems whose reachable behaviour depends heavily on settings, so that testing at default configuration leaves much of the code unexercised.",
+    "narrative": "Database systems appear in the review's opening enumeration of domains fuzzing has been applied to, with network protocols, hypervisors and autonomous vehicles. The mention is reachable only through the citation marker and carries no relationship beyond acknowledging the domain.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icsme64153_2025_00030.json b/_data/papers/paper_doi_10_1109_icsme64153_2025_00030.json
new file mode 100644
index 0000000..680328d
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icsme64153_2025_00030.json
@@ -0,0 +1,821 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:24:42Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icsme64153.2025.00030",
+    "title": "TSGuard: Detecting Logic Bugs in Time Series Management Systems Via Time Series Algebra",
+    "authors": [
+      "Lingwei Kuang",
+      "Liang Liu",
+      "Wen-Jing Wang",
+      "Ning Cao",
+      "Shijie Li",
+      "Fan Liu",
+      "Haolong Chen",
+      "WenJian Liao"
+    ],
+    "year": 2025,
+    "venue": "IEEE International Conference on Software Maintenance and Evolution",
+    "doi": "10.1109/icsme64153.2025.00030",
+    "arxiv_id": null,
+    "s2_paper_id": "eac49ea8165340903f91a6b9e90f73ddfdf6e25b",
+    "url": "https://doi.org/10.1109/icsme64153.2025.00030",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icsme64153.2025.00030",
+      "retrieved_at": "2026-09-10T15:24:42Z",
+      "chars": 64986,
+      "content_sha256": "sha256:ff5e5758f0ed589dbe551f25072a5c0558b4f66056fe34d636662c1cdac89a0a"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2194
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND ANDMOTIVATION",
+        "start": 6365
+      },
+      {
+        "number": "A",
+        "title": "TSMSs and Their Time Series Characteristics",
+        "start": 6720
+      },
+      {
+        "number": "B",
+        "title": "Differences Between Traditional SQL and Time Series",
+        "start": 7967
+      },
+      {
+        "number": "C",
+        "title": "Limitations of Metamorphic Testing",
+        "start": 9472
+      },
+      {
+        "number": "D",
+        "title": "An Example of a Logic Bug",
+        "start": 11224
+      },
+      {
+        "number": "III",
+        "title": "APPROACH",
+        "start": 11913
+      },
+      {
+        "number": "A",
+        "title": "The Architecture of TSGuard",
+        "start": 12517
+      },
+      {
+        "number": "B",
+        "title": "Database Generator",
+        "start": 13635
+      },
+      {
+        "number": "C",
+        "title": "Time Series Query Generator",
+        "start": 17885
+      },
+      {
+        "number": "D",
+        "title": "Query Syntax Validator",
+        "start": 20340
+      },
+      {
+        "number": "E",
+        "title": "Time Series Algebra Engine",
+        "start": 21318
+      },
+      {
+        "number": "F",
+        "title": "Query Executor and Query Result Validator",
+        "start": 29007
+      },
+      {
+        "number": "G",
+        "title": "Syntax Node Sequence-D riven Feedback Mechanism",
+        "start": 29860
+      },
+      {
+        "number": "H",
+        "title": "The Execution Process of TSGuard",
+        "start": 32730
+      },
+      {
+        "number": "A",
+        "title": "Evaluation Setup",
+        "start": 33785
+      },
+      {
+        "number": "B",
+        "title": "Logic Bug Detection Capability of TSGuard",
+        "start": 34891
+      },
+      {
+        "number": "C",
+        "title": "The Effectiveness of the Feedback Mechanism and Query",
+        "start": 36825
+      },
+      {
+        "number": "D",
+        "title": "Comparison with Other Techniques",
+        "start": 39464
+      },
+      {
+        "number": "E",
+        "title": "Resource Cost of TSGuard",
+        "start": 41885
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. 2022. Unicorn: detect runtime errors in time-series databases with hybrid input synthesis. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA 2022). Association for Computing Machinery, New York, NY, USA, 251–262. https://doi.org/10.1145/3533767.3534364",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "InfluxData. (2014). InfluxDB. GitHub repository. Retrieved from https://github.com/influxdata/influxdb",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Chen Wang, Xiangdong Huang, Jialin Qiao, Tian Jiang, Lei Rui, Jinrui Zhang, Rong Kang, Julian Feinauer, Kevin A. McGrail, Peng Wang, Diaohan Luo, Jun Yuan, Jianmin Wang, and Jiaguang Sun. 2020. Apache IoTDB: time-series database for internet of things. Proc. VLDB Endow. 13, 12 (August 2020), 2901–2904. https://doi.org/10.14778/3415478.3415504",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "TAOS Data. 2019. TDengine: Big Data Platform Designed and Optimized for the Internet of Things (IoT). https://www .taosdata .com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "S. K. Jensen, T. B. Pedersen and C. Thomsen, \"Time Series Management Systems: A Survey,\" in IEEE Transactions on Knowledge and Data Engineering, vol. 29, no. 11, pp. 2581-2600, 1 Nov. 2017, doi: 10.1109/TKDE.2017.2740932.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "T. Y. Chen, S. C. Cheung, and S. M. Yiu, \"Metamorphic testing: A new approach for generating next test cases,\" arXiv preprint arXiv:2002.12543, 2020. [Online]. Available: https://arxiv.org/abs/2002.12543.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Tsong Yueh Chen, Fei-Ching Kuo, Huai Liu, Pak-Lok Poon, Dave Towey, T. H. Tse, and Zhi Quan Zhou. 2018. Metamorphic Testing: A Review of Challenges and Opportunities. ACM Comput. Surv. 51, 1, Article 4 (January 2019), 27 pages. https://doi.org/10.1145/3143561",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "H. Liu, F. -C. Kuo, D. Towey and T. Y. Chen, \"How Effectively Does Metamorphic Testing Alleviate the Oracle Problem?,\" in IEEE Transactions on Software Engineering, vol. 40, no. 1, pp. 4-22, Jan. 2014, doi: 10.1109/TSE.2013.46.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "S. Segura, G. Fraser, A. B. Sanchez and A. Ruiz-Cortés, \"A Survey on Metamorphic Testing,\" in IEEE Transactions on Software Engineering, vol. 42, no. 9, pp. 805-824, 1 Sept. 2016, doi: 10.1109/TSE.2016.2532875.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. https://doi.org/10.1145/3368089.340",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 11,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (November 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 12,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI'20). USENIX Association, USA, Article 38, 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 13,
+      "text": "InfluxData, \"Line protocol syntax,\" InfluxDB v2 reference, [Online]. Available: https://docs.influxdata.com/influxdb/v2/reference/syntax/line-protocol. [Accessed: Jan. 20, 2025].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Apache IotDB, \"Data model,\" IotDB v1.3.x reference, [Online]. Available: https://iotdb.apache.org/UserGuide/V1.3.x/Basic-Concept/Data-Model-and-Terminology.html. [Accessed: Dec. 20, 2024].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Agathe Girard, Carl Edward Rasmussen, Joaquin Quiñonero Candela, and Roderick Murray-Smith. 2002. Gaussian Process Priors with uncertain inputs application to Multiple-Step Ahead Time Series Forecasting. In Proceedings of the 16th International Conference on Neural Information Processing Systems (NIPS'02). MIT Press, Cambridge, MA, USA, 545–552.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "A. Alexandrov, K. Benidis, M. Bohlke-Schneider, V. Flunkert, J. Gasthaus, T. Januschowski, D. C. Maddix, S. Rangapuram, D. Salinas, J. Schulz, L. Stella, A. C. Türkmen, and Y. Wang, \"GluonTS: Probabilistic time series models in Python,\" arXiv preprint arXiv:1906.05264, 2019. [Online]. Available: https://arxiv.org/abs/1906.05264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "A. Bagnall, A. Bostrom, J. Large, and J. Lines, \"Simulated data experiments for time series classification Part 1: Accuracy comparison with default settings,\" arXiv preprint arXiv:1703.09480, 2017. [Online]. Available: https://arxiv.org/abs/1703.09480.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Y. Chen, Y. Jiang, F. Ma, J. Liang, M. Wang, C. Zhou, Z. Su, and X. Jiao, \"EnFuzz: Ensemble fuzzing with seed synchronization among diverse fuzzers,\" arXiv preprint arXiv:1807.00182, 2019. [Online]. Available: https://arxiv.org/abs/1807.00182.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Jie Liang, Yu Jiang, Yuanliang Chen, Mingzhe Wang, Chijin Zhou, and Jiaguang Sun. 2018. PAFL: extend fuzzing optimizations of single mode to industrial parallel mode. In Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2018). Association for Computing Machinery, New York, NY, USA, 809–814. ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "LibFuzzer, \"LibFuzzer,\" [Online]. Available: https://www.llvm.org/docs/LibFuzzer.html. [Accessed: Jan. 18, 2024].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Barton P. Miller, Lars Fredriksen, and Bryan So. 1990. An empirical study of the reliability of UNIX utilities. Commun. ACM 33, 12 (Dec. 1990), 32–44. https://doi.org/10.1145/96267.96279",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Michał Zalewski, \"american fuzzy lop,\" [Online]. Available: http://lcamtuf.coredump.cx/afl. [Accessed: Jan. 18, 2024].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Seltenreich A, Tang B, Mullender S. SQLsmith: A random SQL query generator. 2018. https://github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "M. Wang et al., \"Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing,\" Software Engineering: Software Engineering in Practice (ICSE-SEIP), Madrid, ES, 2021, pp. 328-337, doi: 10.1109/ICSE-SEIP52600.2021.00042.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security (CCS '20). Association for Computing Machinery, New York, NY, USA, 955–970. https://doi.org/10.1145/3372297.3417260",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "J. Liang et al ., \"Sequence-Oriented DBMS Fuzzing,\" 2023 IEEE 39th International Conference on Data Engineering (ICDE), Anaheim, CA, USA, 2023, pp. 668-681, doi: 10.1109/ICDE55515.2023.00057.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2023. Griffin: Grammar-Free DBMS Fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering (ASE '22). Association for Computing Machinery, New York, NY, USA, Article 49, 1–12. https://doi.org/10.1145/3551349.3560431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "J. Ba and M. Rigger, \"Testing Database Engines via Query Plan Guidance,\" Engineering (ICSE), Melbourne, Australia, 2023, pp. 2060-2071, doi: 10.1109/ICSE48619.2023.00174.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 29,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful fuzzing for database management systems with complex and valid SQL query generation. In Proceedings of the 32nd USENIX Conference on Security Symposium (SEC '23). USENIX Association, USA, Article 277, 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. APOLLO: automatic detection and diagnosis of performance regressions in database systems. Proc. VLDB Endow. 13, 1 (September 2019), 57–70. https://doi.org/10.14778/3357377.3357382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In Proceedings of the 24rd International Conference on Very Large Data Bases (VLDB '98). Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding bugs in Gremlin-based graph database systems via Randomized differential 246 testing. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA 2022). Association for Computing Machinery, New York, NY, USA, 302–313. https://doi.org/10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Hao, Z., Huang, Q., Wang, C., Wang, J., Zhang, Y., Wu, R., and Zhang, C. 2023. Pinolo: Detecting logical bugs in database management systems with approximate query synthesis. In Proceedings of the 2023 USENIX Annual Technical Conference (USENIX ATC 23), 345-358. USENIX Association. https://www.usenix.org/conference/atc23/presentation/hao",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "X. Liu, Q. Zhou, J. Arulrai and A. Orso, \"Automatic Detection of Performance Bugs in Database Systems using Equivalent Queries,\" pp. 225-236, doi: 10.1145/3510003.3510093.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Jiansen Song, Wensheng Dou, Yu Gao, Ziyu Cui, Yingying Zheng, Dong Wang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction. Proc. VLDB Endow. 17, 8 (April 2024), 1884–1897. https://doi.org/10.14778/3659437.3659445",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (ICSE '24). Association for Computing Machinery, New York, NY, USA, Article 135, 1–12. https://doi.org/10.1145/3597503.3639112 247",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 10,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. https://doi.org/10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 11,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (November 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 12,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI'20). USENIX Association, USA, Article 38, 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 28,
+      "text": "J. Ba and M. Rigger, \"Testing Database Engines via Query Plan Guidance,\" Engineering (ICSE), Melbourne, Australia, 2023, pp. 2060-2071, doi: 10.1109/ICSE48619.2023.00174.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "prints the DOI of the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For example, NoREC [10] detects logic bugs in database optimizers by mutating the original query into a non-optimized query and comparing their result sets.",
+      "context_before": "fying whether the result sets derived by SQL queries conform to predefined oracle rules. The core of DBMS mutation testing lies in the definition of effective oracles, which establish the relationship between the expected and actual result sets of SQL queries. A logic bug is detected when this relationship is violated. As illustrated in Figure 1, common oracles include equivalence and inclusion relations. Fig. 1. The types of test oracles in metamorphic testing. The equivalence relation asserts that, for any database D, the result sets of query Qand its mutated query Q'must satisfy D(Q)=D(Q').",
+      "context_after": "Similarly, TLP[11] mutates the original query into multiple partitioned queries, reassembles their result sets, an d verifies consistency with the original result set to detect logic bugs. The inclusion relation requires that the actual result set be either a superset or a subset of the expected result set. For instance, PQS [12] synthesizes a query Qbased on an expected result set (i.e., the pivot row) and verifies whether the actual result set is a superset of the expected result set to detect logic bugs.Equivalence Inclusion 236 Fig. 2. The architecture of TSGuard. However, traditional",
+      "section": "C Limitations of Metamorphic Testing",
+      "page": 2,
+      "char_offset": 10210,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Similarly, TLP[11] mutates the original query into multiple partitioned queries, reassembles their result sets, an d verifies consistency with the original result set to detect logic bugs.",
+      "context_before": "e oracles, which establish the relationship between the expected and actual result sets of SQL queries. A logic bug is detected when this relationship is violated. As illustrated in Figure 1, common oracles include equivalence and inclusion relations. Fig. 1. The types of test oracles in metamorphic testing. The equivalence relation asserts that, for any database D, the result sets of query Qand its mutated query Q'must satisfy D(Q)=D(Q'). For example, NoREC [10] detects logic bugs in database optimizers by mutating the original query into a non-optimized query and comparing their result sets.",
+      "context_after": "The inclusion relation requires that the actual result set be either a superset or a subset of the expected result set. For instance, PQS [12] synthesizes a query Qbased on an expected result set (i.e., the pivot row) and verifies whether the actual result set is a superset of the expected result set to detect logic bugs.Equivalence Inclusion 236 Fig. 2. The architecture of TSGuard. However, traditional DBMS mutation testing techniques are vulnerable to errors in the target database, therefore cannot guarantee that at least one of the results from the original or mutated SQL is correct. T",
+      "section": "C Limitations of Metamorphic Testing",
+      "page": 2,
+      "char_offset": 10367,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "For instance, PQS [12] synthesizes a query Qbased on an expected result set (i.",
+      "context_before": "The equivalence relation asserts that, for any database D, the result sets of query Qand its mutated query Q'must satisfy D(Q)=D(Q'). For example, NoREC [10] detects logic bugs in database optimizers by mutating the original query into a non-optimized query and comparing their result sets. Similarly, TLP[11] mutates the original query into multiple partitioned queries, reassembles their result sets, an d verifies consistency with the original result set to detect logic bugs. The inclusion relation requires that the actual result set be either a superset or a subset of the expected result set.",
+      "context_after": "e., the pivot row) and verifies whether the actual result set is a superset of the expected result set to detect logic bugs.Equivalence Inclusion 236 Fig. 2. The architecture of TSGuard. However, traditional DBMS mutation testing techniques are vulnerable to errors in the target database, therefore cannot guarantee that at least one of the results from the original or mutated SQL is correct. This limitation may lead to false negatives (as shown in Listing 1). D. An Example of a Logic Bug Listing 1. A logic bug in IoTDB (v1.3.3): Semantically equivalent queries fail to retrieve data, while",
+      "section": "C Limitations of Metamorphic Testing",
+      "page": 2,
+      "char_offset": 10676,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Comparison of the growth trend of syntax node sequences and valid queries during TSGuard and SQLancer runs.",
+      "context_before": "to ensure the correctness of generated SQL queries before execution. Figure 8 presents the number of valid and invalid SQL queries generated by different TSGuard variants. The results show that across all tested TSMSs, TSGuard and TSGuard !S generate an average of 95.44% valid queries, whereas TSGuard !C and TSGuard !SC generate only 21.61% valid queries on average. Although incorporating Query Syntax Validators introduces some computational overhead due to validating SQL, it significantly increases the proportion of valid SQL queries, which improving overall testing efficiency. 242 Fig. 9.",
+      "context_after": "D. Comparison with Other Techniques To evaluate the effectiveness of TSGuard in detecting logic bugs in TSMSs, we adapted the open-source relational database testing tool SQLancer as a baseline for comparison. Comparison Method: We executed each testing tool on each TSMS for 600s, 1200s, 1800s, 3600s, and 7200s, recording various metrics, including the number of bugs triggered, the number of distinct syntax node sequences, the number of valid and invalid SQL queries. TABLE V. Metrics recorded after TSGuard and SQLancer run 7200 seconds. MetricsSQLancer TSGuard InfluxDB IotDB TDengine InfluxDB",
+      "section": "C The Effectiveness of the Feedback Mechanism and Query",
+      "page": 9,
+      "char_offset": 39355,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Comparison with Other Techniques To evaluate the effectiveness of TSGuard in detecting logic bugs in TSMSs, we adapted the open-source relational database testing tool SQLancer as a baseline for comparison.",
+      "context_before": "invalid SQL queries generated by different TSGuard variants. The results show that across all tested TSMSs, TSGuard and TSGuard !S generate an average of 95.44% valid queries, whereas TSGuard !C and TSGuard !SC generate only 21.61% valid queries on average. Although incorporating Query Syntax Validators introduces some computational overhead due to validating SQL, it significantly increases the proportion of valid SQL queries, which improving overall testing efficiency. 242 Fig. 9. Comparison of the growth trend of syntax node sequences and valid queries during TSGuard and SQLancer runs. D.",
+      "context_after": "Comparison Method: We executed each testing tool on each TSMS for 600s, 1200s, 1800s, 3600s, and 7200s, recording various metrics, including the number of bugs triggered, the number of distinct syntax node sequences, the number of valid and invalid SQL queries. TABLE V. Metrics recorded after TSGuard and SQLancer run 7200 seconds. MetricsSQLancer TSGuard InfluxDB IotDB TDengine InfluxDB IotDB TDengine Bugs 2 2 3 12 9 17 Sequences 3805 42653 118889 178232 413218 222726 Valid Queries23243 114024 95007 196583 447972 264704 Invalid Queries376 27401 132957 1905 48253 11502 Table V presents the res",
+      "section": "D Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 39466,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Metrics recorded after TSGuard and SQLancer run 7200 seconds.",
+      "context_before": "2 Fig. 9. Comparison of the growth trend of syntax node sequences and valid queries during TSGuard and SQLancer runs. D. Comparison with Other Techniques To evaluate the effectiveness of TSGuard in detecting logic bugs in TSMSs, we adapted the open-source relational database testing tool SQLancer as a baseline for comparison. Comparison Method: We executed each testing tool on each TSMS for 600s, 1200s, 1800s, 3600s, and 7200s, recording various metrics, including the number of bugs triggered, the number of distinct syntax node sequences, the number of valid and invalid SQL queries. TABLE V.",
+      "context_after": "MetricsSQLancer TSGuard InfluxDB IotDB TDengine InfluxDB IotDB TDengine Bugs 2 2 3 12 9 17 Sequences 3805 42653 118889 178232 413218 222726 Valid Queries23243 114024 95007 196583 447972 264704 Invalid Queries376 27401 132957 1905 48253 11502 Table V presents the results from the 7200-second experiment, comparing TSGuard and SQLancer in terms of bug detection, syntax node sequence generation, and query validity. The results show that TSGuard detected 38 bugs across the three TSMSs, whereas SQLancer detected only 7. Additionally, TSGuard generated 4.92 times more syntax node sequences and 3.91",
+      "section": "D Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 39944,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "MetricsSQLancer TSGuard InfluxDB IotDB TDengine InfluxDB IotDB TDengine Bugs 2 2 3 12 9 17 Sequences 3805 42653 118889 178232 413218 222726 Valid Queries23243 114024 95007 196583 447972 264704 Invalid Queries376 27401 132957 1905 48253 11502 Table V presents the results from the 7200-second experiment, comparing TSGuard and SQLancer in terms of bug detection, syntax node sequence generation, and query validity.",
+      "context_before": "ences and valid queries during TSGuard and SQLancer runs. D. Comparison with Other Techniques To evaluate the effectiveness of TSGuard in detecting logic bugs in TSMSs, we adapted the open-source relational database testing tool SQLancer as a baseline for comparison. Comparison Method: We executed each testing tool on each TSMS for 600s, 1200s, 1800s, 3600s, and 7200s, recording various metrics, including the number of bugs triggered, the number of distinct syntax node sequences, the number of valid and invalid SQL queries. TABLE V. Metrics recorded after TSGuard and SQLancer run 7200 seconds.",
+      "context_after": "The results show that TSGuard detected 38 bugs across the three TSMSs, whereas SQLancer detected only 7. Additionally, TSGuard generated 4.92 times more syntax node sequences and 3.91 times more valid queries than SQLancer, while generating only 38.36% as many invalid queries. Figure 9 illustrates the growth trends of syntax node sequences and valid queries for TSGuard and SQLancer over 7200 seconds. The comparison shows that TSGuard consistently outperformed SQLancer in both metrics across all target TSMSs. Specifically: In TDengine, TSGuard generated an average of 1.89 times more syntax nod",
+      "section": "D Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 40006,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The results show that TSGuard detected 38 bugs across the three TSMSs, whereas SQLancer detected only 7.",
+      "context_before": "mber of bugs triggered, the number of distinct syntax node sequences, the number of valid and invalid SQL queries. TABLE V. Metrics recorded after TSGuard and SQLancer run 7200 seconds. MetricsSQLancer TSGuard InfluxDB IotDB TDengine InfluxDB IotDB TDengine Bugs 2 2 3 12 9 17 Sequences 3805 42653 118889 178232 413218 222726 Valid Queries23243 114024 95007 196583 447972 264704 Invalid Queries376 27401 132957 1905 48253 11502 Table V presents the results from the 7200-second experiment, comparing TSGuard and SQLancer in terms of bug detection, syntax node sequence generation, and query validity.",
+      "context_after": "Additionally, TSGuard generated 4.92 times more syntax node sequences and 3.91 times more valid queries than SQLancer, while generating only 38.36% as many invalid queries. Figure 9 illustrates the growth trends of syntax node sequences and valid queries for TSGuard and SQLancer over 7200 seconds. The comparison shows that TSGuard consistently outperformed SQLancer in both metrics across all target TSMSs. Specifically: In TDengine, TSGuard generated an average of 1.89 times more syntax node sequences and 2.95 times more valid queries than SQLancer. In IoTDB, TSGuard produced an average of 8.9",
+      "section": "D Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 40421,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "91 times more valid queries than SQLancer, while generating only 38.",
+      "context_before": "ds. MetricsSQLancer TSGuard InfluxDB IotDB TDengine InfluxDB IotDB TDengine Bugs 2 2 3 12 9 17 Sequences 3805 42653 118889 178232 413218 222726 Valid Queries23243 114024 95007 196583 447972 264704 Invalid Queries376 27401 132957 1905 48253 11502 Table V presents the results from the 7200-second experiment, comparing TSGuard and SQLancer in terms of bug detection, syntax node sequence generation, and query validity. The results show that TSGuard detected 38 bugs across the three TSMSs, whereas SQLancer detected only 7. Additionally, TSGuard generated 4.92 times more syntax node sequences and 3.",
+      "context_after": "36% as many invalid queries. Figure 9 illustrates the growth trends of syntax node sequences and valid queries for TSGuard and SQLancer over 7200 seconds. The comparison shows that TSGuard consistently outperformed SQLancer in both metrics across all target TSMSs. Specifically: In TDengine, TSGuard generated an average of 1.89 times more syntax node sequences and 2.95 times more valid queries than SQLancer. In IoTDB, TSGuard produced an average of 8.92 times more syntax node sequences and 3.83 times more valid queries than SQLancer. In InfluxDB, TSGuard generated an average of 47.99 times more",
+      "section": "D Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 40603,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Figure 9 illustrates the growth trends of syntax node sequences and valid queries for TSGuard and SQLancer over 7200 seconds.",
+      "context_before": "equences 3805 42653 118889 178232 413218 222726 Valid Queries23243 114024 95007 196583 447972 264704 Invalid Queries376 27401 132957 1905 48253 11502 Table V presents the results from the 7200-second experiment, comparing TSGuard and SQLancer in terms of bug detection, syntax node sequence generation, and query validity. The results show that TSGuard detected 38 bugs across the three TSMSs, whereas SQLancer detected only 7. Additionally, TSGuard generated 4.92 times more syntax node sequences and 3.91 times more valid queries than SQLancer, while generating only 38.36% as many invalid queries.",
+      "context_after": "The comparison shows that TSGuard consistently outperformed SQLancer in both metrics across all target TSMSs. Specifically: In TDengine, TSGuard generated an average of 1.89 times more syntax node sequences and 2.95 times more valid queries than SQLancer. In IoTDB, TSGuard produced an average of 8.92 times more syntax node sequences and 3.83 times more valid queries than SQLancer. In InfluxDB, TSGuard generated an average of 47.99 times more syntax node sequences and 9.47 times more valid queries than SQLancer. The improvements in bug detection, syntax node sequence generation, and query vali",
+      "section": "D Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 40699,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The comparison shows that TSGuard consistently outperformed SQLancer in both metrics across all target TSMSs.",
+      "context_before": "132957 1905 48253 11502 Table V presents the results from the 7200-second experiment, comparing TSGuard and SQLancer in terms of bug detection, syntax node sequence generation, and query validity. The results show that TSGuard detected 38 bugs across the three TSMSs, whereas SQLancer detected only 7. Additionally, TSGuard generated 4.92 times more syntax node sequences and 3.91 times more valid queries than SQLancer, while generating only 38.36% as many invalid queries. Figure 9 illustrates the growth trends of syntax node sequences and valid queries for TSGuard and SQLancer over 7200 seconds.",
+      "context_after": "Specifically: In TDengine, TSGuard generated an average of 1.89 times more syntax node sequences and 2.95 times more valid queries than SQLancer. In IoTDB, TSGuard produced an average of 8.92 times more syntax node sequences and 3.83 times more valid queries than SQLancer. In InfluxDB, TSGuard generated an average of 47.99 times more syntax node sequences and 9.47 times more valid queries than SQLancer. The improvements in bug detection, syntax node sequence generation, and query validity achieved by TSGuard can be attributed to two key factors. First, TSGuard incorporates the Syntax Node Seq",
+      "section": "D Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 40825,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "95 times more valid queries than SQLancer.",
+      "context_before": "that TSGuard detected 38 bugs across the three TSMSs, whereas SQLancer detected only 7. Additionally, TSGuard generated 4.92 times more syntax node sequences and 3.91 times more valid queries than SQLancer, while generating only 38.36% as many invalid queries. Figure 9 illustrates the growth trends of syntax node sequences and valid queries for TSGuard and SQLancer over 7200 seconds. The comparison shows that TSGuard consistently outperformed SQLancer in both metrics across all target TSMSs. Specifically: In TDengine, TSGuard generated an average of 1.89 times more syntax node sequences and 2.",
+      "context_after": "In IoTDB, TSGuard produced an average of 8.92 times more syntax node sequences and 3.83 times more valid queries than SQLancer. In InfluxDB, TSGuard generated an average of 47.99 times more syntax node sequences and 9.47 times more valid queries than SQLancer. The improvements in bug detection, syntax node sequence generation, and query validity achieved by TSGuard can be attributed to two key factors. First, TSGuard incorporates the Syntax Node Sequence-Driven Feedback Mechanism and Query Syntax Validators, which guide query generation and facilitate broader and more efficient test space exp",
+      "section": "D Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 41039,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "83 times more valid queries than SQLancer.",
+      "context_before": "es more syntax node sequences and 3.91 times more valid queries than SQLancer, while generating only 38.36% as many invalid queries. Figure 9 illustrates the growth trends of syntax node sequences and valid queries for TSGuard and SQLancer over 7200 seconds. The comparison shows that TSGuard consistently outperformed SQLancer in both metrics across all target TSMSs. Specifically: In TDengine, TSGuard generated an average of 1.89 times more syntax node sequences and 2.95 times more valid queries than SQLancer. In IoTDB, TSGuard produced an average of 8.92 times more syntax node sequences and 3.",
+      "context_after": "In InfluxDB, TSGuard generated an average of 47.99 times more syntax node sequences and 9.47 times more valid queries than SQLancer. The improvements in bug detection, syntax node sequence generation, and query validity achieved by TSGuard can be attributed to two key factors. First, TSGuard incorporates the Syntax Node Sequence-Driven Feedback Mechanism and Query Syntax Validators, which guide query generation and facilitate broader and more efficient test space exploration. Second, TSGuard detects logic bugs in TSMSs using time seriesalgebra, overcoming the common false-negative issues in D",
+      "section": "D Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 41167,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "47 times more valid queries than SQLancer.",
+      "context_before": "Figure 9 illustrates the growth trends of syntax node sequences and valid queries for TSGuard and SQLancer over 7200 seconds. The comparison shows that TSGuard consistently outperformed SQLancer in both metrics across all target TSMSs. Specifically: In TDengine, TSGuard generated an average of 1.89 times more syntax node sequences and 2.95 times more valid queries than SQLancer. In IoTDB, TSGuard produced an average of 8.92 times more syntax node sequences and 3.83 times more valid queries than SQLancer. In InfluxDB, TSGuard generated an average of 47.99 times more syntax node sequences and 9.",
+      "context_after": "The improvements in bug detection, syntax node sequence generation, and query validity achieved by TSGuard can be attributed to two key factors. First, TSGuard incorporates the Syntax Node Sequence-Driven Feedback Mechanism and Query Syntax Validators, which guide query generation and facilitate broader and more efficient test space exploration. Second, TSGuard detects logic bugs in TSMSs using time seriesalgebra, overcoming the common false-negative issues in DBMS metamorphic testing techniques and detecting more potential logic bugs. E. Resource Cost of TSGuard TSGuard’s cost is evaluated a",
+      "section": "D Comparison with Other Techniques",
+      "page": 9,
+      "char_offset": 41300,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [12] randomly selects specific rows from the database and synthesizes test cases, verifying that the results of these test cases contain the selected rows to detect logic bugs.",
+      "context_before": "example, the Time Series Alge bra Engine transforms queries into equivalent time series al gebra expressions and calculates expected result sets. Compared to DBMS differential testing, TSGuard imposes fewer constraints when deriving expected result sets, allowing for broader exploration of the TSMS test space. DBMS Metamorphic Testing. DBMS metamorphic testing generates test cases by transforming SQL queries based on predefined rules and validating their execution results against test oracles to detect logic or performance bugs. This technique has been widely applied to both RDBMSs and GDBMSs.",
+      "context_after": "TLP [11] partitions a query into multiple partitioned queries, verifying that the combined results of these partitioned queries remain consistent with the results of original query. NOREC [10] transforms an original query into a non-optimized but semantically equivalent SQL query, detecting logic bugs by comparing their execution results. QPG [28] guides database state mutation using query plans and applies logic bug oracles to identify logic bugs across various database states. Pinolo [33] employs approximate query synthesis techniques to mutate original queries and asserts that the result s",
+      "section": "E Resource Cost of TSGuard",
+      "page": 11,
+      "char_offset": 52237,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [11] partitions a query into multiple partitioned queries, verifying that the combined results of these partitioned queries remain consistent with the results of original query.",
+      "context_before": "ng, TSGuard imposes fewer constraints when deriving expected result sets, allowing for broader exploration of the TSMS test space. DBMS Metamorphic Testing. DBMS metamorphic testing generates test cases by transforming SQL queries based on predefined rules and validating their execution results against test oracles to detect logic or performance bugs. This technique has been widely applied to both RDBMSs and GDBMSs. PQS [12] randomly selects specific rows from the database and synthesizes test cases, verifying that the results of these test cases contain the selected rows to detect logic bugs.",
+      "context_after": "NOREC [10] transforms an original query into a non-optimized but semantically equivalent SQL query, detecting logic bugs by comparing their execution results. QPG [28] guides database state mutation using query plans and applies logic bug oracles to identify logic bugs across various database states. Pinolo [33] employs approximate query synthesis techniques to mutate original queries and asserts that the result sets of mutated queries should be either supersets or subsets of the original results to detect logic bugs. AMOEBA [34] utilizes a mutator and a mutation feedback mechanism to transfo",
+      "section": "E Resource Cost of TSGuard",
+      "page": 11,
+      "char_offset": 52418,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "NOREC [10] transforms an original query into a non-optimized but semantically equivalent SQL query, detecting logic bugs by comparing their execution results.",
+      "context_before": "generates test cases by transforming SQL queries based on predefined rules and validating their execution results against test oracles to detect logic or performance bugs. This technique has been widely applied to both RDBMSs and GDBMSs. PQS [12] randomly selects specific rows from the database and synthesizes test cases, verifying that the results of these test cases contain the selected rows to detect logic bugs. TLP [11] partitions a query into multiple partitioned queries, verifying that the combined results of these partitioned queries remain consistent with the results of original query.",
+      "context_after": "QPG [28] guides database state mutation using query plans and applies logic bug oracles to identify logic bugs across various database states. Pinolo [33] employs approximate query synthesis techniques to mutate original queries and asserts that the result sets of mutated queries should be either supersets or subsets of the original results to detect logic bugs. AMOEBA [34] utilizes a mutator and a mutation feedback mechanism to transform original queries into equivalent queries, comparing their execution times to detect performance bugs. Radar [35] constructs an equivalent database instance",
+      "section": "E Resource Cost of TSGuard",
+      "page": 11,
+      "char_offset": 52600,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG [28] guides database state mutation using query plans and applies logic bug oracles to identify logic bugs across various database states.",
+      "context_before": "rmance bugs. This technique has been widely applied to both RDBMSs and GDBMSs. PQS [12] randomly selects specific rows from the database and synthesizes test cases, verifying that the results of these test cases contain the selected rows to detect logic bugs. TLP [11] partitions a query into multiple partitioned queries, verifying that the combined results of these partitioned queries remain consistent with the results of original query. NOREC [10] transforms an original query into a non-optimized but semantically equivalent SQL query, detecting logic bugs by comparing their execution results.",
+      "context_after": "Pinolo [33] employs approximate query synthesis techniques to mutate original queries and asserts that the result sets of mutated queries should be either supersets or subsets of the original results to detect logic bugs. AMOEBA [34] utilizes a mutator and a mutation feedback mechanism to transform original queries into equivalent queries, comparing their execution times to detect performance bugs. Radar [35] constructs an equivalent database instance without metadata from the original instance, executing identical queries on both to identify logic bugs by comparing their results. Mozi [36] b",
+      "section": "E Resource Cost of TSGuard",
+      "page": 11,
+      "char_offset": 52759,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/LingweiKuang/TSGuard",
+    "markers": [
+      "renamed_sqlancer_package",
+      "sqlancer_source_content_match"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/LingweiKuang/TSGuard",
+        "source_type": "github_repository",
+        "excerpt": "TSGuard: Detecting Logic Bugs in Time Series Management Systems via Time Series Algebra\n# TSGuard-Detecting-Logic-Bugs-in-Time-Series-Management-Systems-via-Time-Series-Algebra\nTSGuard: Detecting Logic Bugs in Time Series Management Systems via Time Series Algebra",
+        "note": "Repository names this paper.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/LingweiKuang/TSGuard/blob/main/TSGuard/tsFuzzy/src/main/java/com/fuzzy/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+        "excerpt_is_verbatim": true,
+        "note": "TSGuard/tsFuzzy/src/main/java/com/fuzzy/Randomly.java is SQLancer's Randomly.java, with the package renamed to com.fuzzy (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+        "content_sha256": "sha256:66bc91c4144354ceb03896ba19c217e53af2571b6110cb50426983e6c3c27f63",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M5",
+        "M9",
+        "M12",
+        "M13",
+        "M14"
+      ]
+    },
+    "suppressed": [
+      {
+        "mention_id": "M5",
+        "pattern": "extends_technique",
+        "suppressed_because": "the sentence says the tool was taken up as a baseline, which is comparison rather than reuse or extension"
+      }
+    ]
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:49:48Z",
+    "is_model_written": true,
+    "summary": "TSGuard detects logic bugs in time series management systems using a time series algebra: queries are rewritten into algebraically equivalent forms whose results must agree, with the rewriting rules defined over time-series operations rather than relational ones. Its motivation is that relational metamorphic oracles do not carry over, since time-series query languages have constructs those oracles cannot express. On InfluxDB, IoTDB and TDengine it found 38 bugs.",
+    "narrative": "SQLancer is TSGuard's baseline, and the paper is explicit about the form that takes: it adapted SQLancer, an open-source relational database testing tool, as a baseline for comparison. The two were run for 7200 seconds against the same three systems, with TSGuard finding 38 bugs to SQLancer's 7 and generating several times more valid queries. NoREC, TLP, PQS and QPG are each described in the discussion of metamorphic testing's limits, which is where the paper argues that oracles built for relational queries do not transfer to time-series ones. The adaptation is what made a relational tool runnable on these systems for the comparison, not a contribution TSGuard claims.",
+    "roles": {
+      "M1": "definition",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "result_comparison",
+      "M5": "baseline",
+      "M6": "result_comparison",
+      "M7": "result_comparison",
+      "M8": "result_comparison",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "result_comparison",
+      "M12": "result_comparison",
+      "M13": "result_comparison",
+      "M14": "result_comparison",
+      "M15": "definition",
+      "M16": "definition",
+      "M17": "definition",
+      "M18": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M5 describes adapting SQLancer so it could be run as a baseline. That is work done to make the comparison possible, not TSGuard's own implementation being built on SQLancer -- its algebra and generator are its own."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Adapting the tool in order to run it as a baseline is comparison, not extension. TSGuard's contribution is a time series algebra, and the paper presents NoREC, TLP and PQS as relational oracles whose assumptions do not carry to time-series systems rather than as techniques it generalises."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M5",
+          "M7",
+          "M8"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M5",
+            "sentence": "Comparison with Other Techniques To evaluate the effectiveness of TSGuard in detecting logic bugs in TSMSs, we adapted the open-source relational database testing tool SQLancer as a baseline for comparison.",
+            "section": "D Comparison with Other Techniques",
+            "page": 9
+          },
+          {
+            "mention_id": "M7",
+            "sentence": "MetricsSQLancer TSGuard InfluxDB IotDB TDengine InfluxDB IotDB TDengine Bugs 2 2 3 12 9 17 Sequences 3805 42653 118889 178232 413218 222726 Valid Queries23243 114024 95007 196583 447972 264704 Invalid Queries376 27401 132957 1905 48253 11502 Table V presents the results from the 7200-second experiment, comparing TSGuard and SQLancer in terms of bug detection, syntax node sequence generation, and query validity.",
+            "section": "D Comparison with Other Techniques",
+            "page": 9
+          },
+          {
+            "mention_id": "M8",
+            "sentence": "The results show that TSGuard detected 38 bugs across the three TSMSs, whereas SQLancer detected only 7.",
+            "section": "D Comparison with Other Techniques",
+            "page": 9
+          }
+        ],
+        "reasoning": "M5 states SQLancer was adapted as a baseline for comparison, M7 gives the per-system table of bugs and query counts for both tools, and M8 reports 38 bugs against SQLancer's 7."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is described as an open-source relational database testing tool; the paper makes no claim about it being the state of the art."
+      }
+    },
+    "disagreements": [
+      {
+        "check": "extends_technique matched on adapted SQLancer",
+        "why": "The regex sees an adaptation verb next to the tool name. The sentence continues 'as a baseline for comparison', which makes it the opposite: the tool was modified so it could be measured against, not built upon."
+      }
+    ],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_icst60714_2024_00012.json b/_data/papers/paper_doi_10_1109_icst60714_2024_00012.json
new file mode 100644
index 0000000..4cb9855
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_icst60714_2024_00012.json
@@ -0,0 +1,1033 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:23Z",
+  "paper": {
+    "id": "paper:doi:10.1109/icst60714.2024.00012",
+    "title": "Differential Optimization Testing of Gremlin-Based Graph Database Systems",
+    "authors": [
+      "Yingying Zheng",
+      "Wensheng Dou",
+      "Leile Tang",
+      "Ziyu Cui",
+      "Jiansen Song",
+      "Ziyue Cheng",
+      "Wei Wang",
+      "Jun Wei",
+      "Hua Zhong",
+      "Tao Huang"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Information Control Systems & Technologies",
+    "doi": "10.1109/icst60714.2024.00012",
+    "arxiv_id": null,
+    "s2_paper_id": "7f7686d2e2af4a7308fa947680f52194b8bb067f",
+    "url": "https://doi.org/10.1109/icst60714.2024.00012",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/icst60714.2024.00012",
+      "retrieved_at": "2026-09-09T01:31:23Z",
+      "chars": 65176,
+      "content_sha256": "sha256:701d9f2068396720ca907cd5d3267adf004713da73d64ee4ab4b4a04c422a054"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1984
+      },
+      {
+        "number": "II",
+        "title": "PRELIMINARIES",
+        "start": 9756
+      },
+      {
+        "number": "A",
+        "title": "Graph Model and Gremlin Query Language",
+        "start": 9774
+      },
+      {
+        "number": "B",
+        "title": "Optimizations in Gremlin-Based GDBs",
+        "start": 12426
+      },
+      {
+        "number": "III",
+        "title": "APPROACH",
+        "start": 14478
+      },
+      {
+        "number": "A",
+        "title": "Generating Graph Databases and Gremlin Queries",
+        "start": 16516
+      },
+      {
+        "number": "B",
+        "title": "Generating Optimization Configurations",
+        "start": 19026
+      },
+      {
+        "number": "C",
+        "title": "Comparing Query Results",
+        "start": 23669
+      },
+      {
+        "number": "D",
+        "title": "Locating Faulty Optimization Strategies",
+        "start": 24535
+      },
+      {
+        "number": "Q",
+        "title": "Given the original optimization configuration optConf",
+        "start": 24764
+      },
+      {
+        "number": "E",
+        "title": "Optimizing Differential Testing",
+        "start": 26659
+      },
+      {
+        "number": "F",
+        "title": "Corner Cases",
+        "start": 29427
+      },
+      {
+        "number": "IV",
+        "title": "IMPLEMENTA TIONS",
+        "start": 31872
+      },
+      {
+        "number": "V",
+        "title": "EVALUA TION",
+        "start": 33559
+      },
+      {
+        "number": "A",
+        "title": "Experimental Setup",
+        "start": 33993
+      },
+      {
+        "number": "B",
+        "title": "Detected Bugs",
+        "start": 35445
+      },
+      {
+        "number": "C",
+        "title": "Efficiency of Optimization-Guided Testing Approach",
+        "start": 42046
+      },
+      {
+        "number": "D",
+        "title": "Comparison with Existing Approaches",
+        "start": 44199
+      },
+      {
+        "number": "E",
+        "title": "Interesting Bugs",
+        "start": 46796
+      },
+      {
+        "number": "VI",
+        "title": "DISCUSSION",
+        "start": 49561
+      },
+      {
+        "number": "VII",
+        "title": "RELA TED WORK",
+        "start": 50823
+      },
+      {
+        "number": "VIII",
+        "title": "CONCLUSION",
+        "start": 52575
+      },
+      {
+        "number": "A",
+        "title": "Prat-P ´erez, M. Pham, and P. A. Boncz, “The LDBC social network",
+        "start": 54079
+      },
+      {
+        "number": "W",
+        "title": "Wang, and J. Wei, “Finding bugs in Gremlin-based graph database",
+        "start": 55249
+      },
+      {
+        "number": "T",
+        "title": "Huang, “Testing database systems via differential query execution,”",
+        "start": 57663
+      },
+      {
+        "number": "L",
+        "title": "Chen, H. Wang, H. Zhong, and T. Huang, “Detecting isolation bugs",
+        "start": 57910
+      },
+      {
+        "number": "O",
+        "title": "Papaemmanouil, and N. Tatbul, “Neo: A learned query optimizer,”",
+        "start": 60604
+      },
+      {
+        "number": "K",
+        "title": "Xu, Y. Hu, J. Wei, and T. Huang, “Understanding transaction bugs",
+        "start": 63265
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Neo4j,” https://neo4j.com/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“Orientdb,” https://orientdb.org, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“Janusgraph,” https://janusgraph.org, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "A. Deutsch, Y. Xu, M. Wu, and V. E. Lee, “Aggregation support for modern graph analytics in TigerGraph,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 2020, pp. 377–392.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "“Open source, distributed, scalable, lightning fast,” https://nebulagraph.io/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "O. Erling, A. Averbuch, J. L. Larriba-Pey, H. Chafi, A. Gubichev, A. Prat-P ´erez, M. Pham, and P. A. Boncz, “The LDBC social network benchmark: Interactive workload,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 2015, pp. 619–630.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Y. Ren, H. Zhu, J. Zhang, P. Dai, and L. Bo, “EnsemFDet: An ensemble approach to fraud detection based on bipartite graph,” in Proceedings of International Conference on Data Engineering (ICDE), 2021, pp. 2039–2044.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "B. Liu, X. Wang, P. Liu, S. Li, Q. Fu, and Y. Chai, “Unikg: A unified interoperable knowledge graph database system,” in Proceedings of IEEE International Conference on Data Engineering (ICDE), 2021, pp. 2681–2684.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "M. Arenas, C. Guti ´errez, and J. F. Sequeda, “Querying in the age of graph databases and knowledge graphs,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 2021, pp. 2821–2828.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "“Gremlin query language,” https://tinkerpop.apache.org/gremlin.html, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "“Hugegraph,” https://hugegraph.github.io/hugegraph-doc/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "“Db-engines ranking of graph dbms,” https://dbengines.com/en/ranking/graph+ dbms, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Y. Zheng, W. Dou, Y. Wang, Z. Qin, L. Tang, Y. Gao, D. Wang, W. Wang, and J. Wei, “Finding bugs in Gremlin-based graph database systems via randomized differential testing,” in Proceedings of International Symposium on Software Testing and Analysis (ISSTA), 2022, pp. 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "R. Yang, Y. Zheng, L. Tang, W. Dou, W. Wang, and J. Wei, “Randomized differential testing of RDF stores,” in Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE Demo), 2023, pp. 136–140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Z. Hua, W. Lin, L. Ren, Z. Li, L. Zhang, and T. Xie, “GDsmith: Detecting bugs in cypher graph database engines,” in Proceedings of International Symposium on Software Testing and Analysis (ISSTA), 2023, pp. 163–174.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "“Sqlsmith,” https://github.com/anse1/sqlsmith, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "D. S. Slutz, “Massive stochastic testing of SQL,” in Proceedings of International Conference on V ery Large Data Bases (VLDB), 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment (VLDB), vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” in Proceedings of ACM Object-Oriented Programming Systems, Languages, and Applications (OOPSLA), 2020, pp. 211:1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "——, “Detecting optimization bugs in database engines via nonoptimizing reference engine construction,” in Proceedings of ACM Joint European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140– 1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "——, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 23,
+      "text": "Z. Cui, W. Dou, Q. Dai, J. Song, W. Wang, J. Wei, and D. Ye, “Differentially testing database transactions for fun and profit,” in Proceedings of International Conference on Automated Software Engineering (ASE), 2022, pp. 35:1–35:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” inProceedings of International Conference on Software Engineering (ICSE), 2023, pp. 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 25,
+      "text": "J. Song, W. Dou, Z. Cui, Q. Dai, W. Wang, J. Wei, H. Zhong, and T. Huang, “Testing database systems via differential query execution,” inProceedings of IEEE/ACM International Conference on Software Engineering (ICSE), 2023, pp. 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "W. Dou, Z. Cui, Q. Dai, J. Song, D. Wang, Y. Gao, W. Wang, J. Wei, L. Chen, H. Wang, H. Zhong, and T. Huang, “Detecting isolation bugs via transaction oracle construction,” in Proceedings of International Conference on Software Engineering (ICSE), 2023, pp. 1123–1135.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "“Tinkergraph,” https://github.com/tinkerpop/blueprints/wiki/tinkergraph, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "“The next generation multi-model database supporting graphs key/value, documents and time-series,” https://arcadedb.com/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "R. Wang, Z. Yang, W. Zhang, and X. Lin, “An empirical study on recent graph database systems,” in Proceedings of International Conference on Knowledge Science, Engineering and Management (KSEM), 2020, pp. 328–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "M. A. Rodriguez, “The gremlin graph traversal machine and language (invited talk),” in Proceedings of the Symposium on Database Programming Languages, 2015, pp. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "“Tinkerpop,” https://tinkerpop.apache.org/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "N. Francis, A. Green, P. Guagliardo, L. Libkin, T. Lindaaker, V. Marsault, S. Plantikow, M. Rydberg, P. Selmer, and A. Taylor, “Cypher: An evolving query language for property graphs,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 2018, pp. 1433–1445.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Z. He, J. Y u, and B. Guo, “Execution time prediction for Cypher queries in the Neo4j database using a learning approach,” Symmetry, vol. 14, no. 1, p. 55, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "A. Deutsch, “Querying graph databases with the GSQL query language,” inProceedings of Simp ´osio Brasileiro de Banco de Dados (SBBD), 2018, p. 313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "A. Deutsch, Y. Xu, M. Wu, and V. E. Lee, “Aggregation support for modern graph analytics in TigerGraph,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 2020, pp. 377–392.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "“Nebula graph query language (nGQL),” https://docs.nebulagraph.io/2.0.1/3.ngql-guide/1.nGQL-overview/1.overview/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "M. Wu, X. Yi, H. Y u, Y. Liu, and Y. Wang, “Nebula graph: An open source distributed graph database,” CoRR, vol. abs/2206.07278, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "“Gremlin traversal strategy,” https://tinkerpop.apache.org/docs/3.5.2/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "B. Ding, S. Das, W. Wu, S. Chaudhuri, and V. R. Narasayya, “Plan stitch: Harnessing the best of many plans,” Proceedings of the VLDB Endowment (VLDB), vol. 11, no. 10, pp. 1123–1136, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "T. Neumann and B. Radke, “Adaptive optimization of very large join queries,” in Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), 2018, pp. 677–692.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "C. Wu, A. Jindal, S. Amizadeh, H. Patel, W. Le, S. Qiao, and S. Rao, “Towards a learning optimizer for shared clouds,” Proceedings of the VLDB Endowment (VLDB), vol. 12, no. 3, pp. 210–222, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "R. C. Marcus, P. Negi, H. Mao, C. Zhang, M. Alizadeh, T. Kraska, O. Papaemmanouil, and N. Tatbul, “Neo: A learned query optimizer,” Proceedings of the VLDB Endowment (VLDB), vol. 12, no. 11, pp. 1705– 1718, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "C. Nie and H. Leung, “A survey of combinatorial testing,” ACM Computing Surveys, vol. 43, no. 2, pp. 11:1–11:29, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "K. Tai and Y. Lei, “A test generation strategy for pairwise testing,” IEEE Transactions on Software Engineering (TSE), vol. 28, no. 1, pp. 109– 111, 2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "“Neo4j-gremlin,” https://github.com/thinkaurelius/neo4j-gremlin-plugin, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "W. M. McKeeman, “Differential testing for software,” Digital Technical Journal, vol. 10, no. 1, pp. 100–107, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "X. Yang, Y. Chen, E. Eide, and J. Regehr, “Finding and understanding bugs in C compilers,” in Proceedings of International Conference on Programming Language Design and Implementation (PLDI), 2011, pp. 283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "S. A. Chowdhury, S. Mohian, S. Mehra, S. Gawsane, T. T. Johnson, and C. Csallner, “Automatically finding bugs in a commercial cyberphysical system development tool chain with SLforge,” in Proceedings of International Conference on Software Engineering (ICSE), 2018, pp. 981–992. 35",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "M. N. Mansur, M. Christakis, and V. W ¨ustholz, “Metamorphic testing of Datalog engines,” in Proceedings of Joint European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2021, pp. 639–650.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "X. Liu, Q. Zhou, J. Arulraj, and A. Orso, “Automatic detection of performance bugs in database systems using equivalent queries,” in Proceedings of International Conference on Software Engineering (ICSE), 2022, pp. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "S. A. Khalek, B. Elkarablieh, Y. O. Laleye, and S. Khurshid, “Queryaware test generation using a relational constraint solver,” in Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE), 2008, pp. 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "S. A. Khalek and S. Khurshid, “Automated SQL query generation for systematic testing of database engines,” in Proceedings of Automated Software Engineering (ASE), 2010, pp. 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "“Afl,” https://github.com/google/AFL, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “SQUIRREL: Testing database management systems with language validity and coverage feedback,” in Proceedings of Computer and Communications Security (CCS), 2020, pp. 58–71.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Z. Hao, Q. Huang, C. Wang, J. Wang, Y. Zhang, R. Wu, and C. Zhang, “Pinolo: Detecting logical bugs in database management systems with approximate query synthesis,” in Proceedings of USENIX Annual Technical Conference (USENIX ATC), 2023, pp. 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Z. Cui, W. Dou, Y. Gao, D. Wang, J. Song, Y. Zheng, T. Wang, R. Yang, K. Xu, Y. Hu, J. Wei, and T. Huang, “Understanding transaction bugs in database systems,” in Proceedings of International Conference on Software Engineering (ICSE), 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "J. Chen, J. Patra, M. Pradel, Y. Xiong, H. Zhang, D. Hao, and L. Zhang,“A survey of compiler testing,” ACM Computing Surveys, vol. 53, no. 1, pp. 4:1–4:36, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "J. M. Caron and P. A. Darnell, “Bugfind: A tool for debugging optimizing compilers,” ACMSIGPLAN Notices, vol. 25, no. 1, pp. 17– 22, 1990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "C. J. Burgess and M. Saidi, “The automatic generation of test cases for optimizing fortran compilers,” Information and Software Technology (IST), vol. 38, no. 2, pp. 111–119, 1996.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "J. Chen and C. Suo, “Boosting compiler testing via compiler optimization exploration,” ACM Transactions on Software Engineering and Methodology (TOSEM), vol. 31, no. 4, pp. 72:1–72:33, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "M. Sassa and D. Sudosa, “Experience in testing compiler optimizers using comparison checking,” in Proceedings of Software Engineering Research and Practice & Conference on Programming Languages and Compilers (SERP), vol. 2, 2006, pp. 837–843.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "V. Le, C. Sun, and Z. Su, “Randomized stress-testing of link-time optimizers,” in Proceedings of Symposium on Software Testing and Analysis (ISSTA), 2015, pp. 327–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "V. Le, M. Afshari, and Z. Su, “Compiler validation via equivalence modulo inputs,” in Proceedings of Programming Language Design and Implementation (PLDI). ACM, 2014, pp. 216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "A. Hashimoto and N. Ishiura, “Detecting arithmetic optimization opportunities for C compilers by randomly generated equivalent programs,” IPSJ Transactions on System and LSI Design Methodology, vol. 9, pp. 21–29, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "V. Le, C. Sun, and Z. Su, “Finding deep compiler bugs via guided stochastic program mutation,” in Proceedings of Object-Oriented Programming, Systems, Languages, and Applications, (OOPSLA), 2015, pp. 386–399. 36",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 14,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” in Proceedings of ACM Object-Oriented Programming Systems, Languages, and Applications (OOPSLA), 2020, pp. 211:1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 21,
+      "text": "——, “Detecting optimization bugs in database engines via nonoptimizing reference engine construction,” in Proceedings of ACM Joint European Software Engineering Conference and Symposium on the F oundations of Software Engineering (ESEC/FSE), 2020, pp. 1140– 1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 22,
+      "text": "——, “Testing database engines via pivoted query synthesis,” in Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 24,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” inProceedings of International Conference on Software Engineering (ICSE), 2023, pp. 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[13]–[16]",
+      "technique": null,
+      "sentence": "Recently, several approaches [13]–[16] have been proposed to find logic bugs in GDBs.",
+      "context_before": "different optimization configurations, one with both the strategies LazyBarrierStrategy andHugeVertexStepStrategy off (Line 1-2), and the other with both the strategies LazyBarrierStrategy and HugeVertexStepStrategy on (Line 6-7). The correct result of this query should be v:{1, 3, 4}(Line 9). However, HugeGraph returns an incorrect result v:{1, 2, 3, 4 }(Line 4) with the first optimization configuration (Line 1-2). HugeGraph developers explained that they forgot to remove the dirty edges of the vertices in the cache. HugeGraph developers have confirmed it as a previously-unknown bug and fixed it.",
+      "context_after": "Most of these approaches (e.g., Grand [13], GDsmith [16], and RD2[15]) utilize differential testing to find logic bugs in multiple GDBs. However, they can miss logic bugs when all target GDBs return incorrect query results. Many testing approaches [17]–[26] (e.g., TLP [20], NoREC [21], and DQE [25]) have been proposed to test relational database systems. Among these approaches, NoREC [21] can detect optimization bugs in relational database systems by rewriting an optimized SQL query into a nonoptimizing SQL query. However, a Gremlin query cannot be rewritten into a non-optimizing Gremlin query",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 5653,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[17]–[26]",
+      "technique": "tlp",
+      "sentence": "Many testing approaches [17]–[26] (e.",
+      "context_before": "Graph returns an incorrect result v:{1, 2, 3, 4 }(Line 4) with the first optimization configuration (Line 1-2). HugeGraph developers explained that they forgot to remove the dirty edges of the vertices in the cache. HugeGraph developers have confirmed it as a previously-unknown bug and fixed it. Recently, several approaches [13]–[16] have been proposed to find logic bugs in GDBs. Most of these approaches (e.g., Grand [13], GDsmith [16], and RD2[15]) utilize differential testing to find logic bugs in multiple GDBs. However, they can miss logic bugs when all target GDBs return incorrect query results.",
+      "context_after": "g., TLP [20], NoREC [21], and DQE [25]) have been proposed to test relational database systems. Among these approaches, NoREC [21] can detect optimization bugs in relational database systems by rewriting an optimized SQL query into a nonoptimizing SQL query. However, a Gremlin query cannot be rewritten into a non-optimizing Gremlin query by using the idea in NoREC because the procedural Gremlin queries in Gremlin-based GDBs adopt different query patterns from that of the declarative SQL queries in relational database systems. Thus, we still lack an effective approach to find optimization bugs i",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 5961,
+      "cited_reference": {
+        "number": 20,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” in Proceedings of ACM Object-Oriented Programming Systems, Languages, and Applications (OOPSLA), 2020, pp. 211:1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", TLP [20], NoREC [21], and DQE [25]) have been proposed to test relational database systems.",
+      "context_before": "2, 3, 4 }(Line 4) with the first optimization configuration (Line 1-2). HugeGraph developers explained that they forgot to remove the dirty edges of the vertices in the cache. HugeGraph developers have confirmed it as a previously-unknown bug and fixed it. Recently, several approaches [13]–[16] have been proposed to find logic bugs in GDBs. Most of these approaches (e.g., Grand [13], GDsmith [16], and RD2[15]) utilize differential testing to find logic bugs in multiple GDBs. However, they can miss logic bugs when all target GDBs return incorrect query results. Many testing approaches [17]–[26] (e.g.",
+      "context_after": "Among these approaches, NoREC [21] can detect optimization bugs in relational database systems by rewriting an optimized SQL query into a nonoptimizing SQL query. However, a Gremlin query cannot be rewritten into a non-optimizing Gremlin query by using the idea in NoREC because the procedural Gremlin queries in Gremlin-based GDBs adopt different query patterns from that of the declarative SQL queries in relational database systems. Thus, we still lack an effective approach to find optimization bugs in GDBs. In this paper, we propose Differential O ptimization T esting (DOT ), an automated appr",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 6001,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Among these approaches, NoREC [21] can detect optimization bugs in relational database systems by rewriting an optimized SQL query into a nonoptimizing SQL query.",
+      "context_before": "plained that they forgot to remove the dirty edges of the vertices in the cache. HugeGraph developers have confirmed it as a previously-unknown bug and fixed it. Recently, several approaches [13]–[16] have been proposed to find logic bugs in GDBs. Most of these approaches (e.g., Grand [13], GDsmith [16], and RD2[15]) utilize differential testing to find logic bugs in multiple GDBs. However, they can miss logic bugs when all target GDBs return incorrect query results. Many testing approaches [17]–[26] (e.g., TLP [20], NoREC [21], and DQE [25]) have been proposed to test relational database systems.",
+      "context_after": "However, a Gremlin query cannot be rewritten into a non-optimizing Gremlin query by using the idea in NoREC because the procedural Gremlin queries in Gremlin-based GDBs adopt different query patterns from that of the declarative SQL queries in relational database systems. Thus, we still lack an effective approach to find optimization bugs in GDBs. In this paper, we propose Differential O ptimization T esting (DOT ), an automated approach to detect optimization bugs in Gremlin-based GDBs. We first randomly generate a graph database gdb. Based on the generated graph database gdb, we further rando",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 6094,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "However, a Gremlin query cannot be rewritten into a non-optimizing Gremlin query by using the idea in NoREC because the procedural Gremlin queries in Gremlin-based GDBs adopt different query patterns from that of the declarative SQL queries in relational database systems.",
+      "context_before": "ently, several approaches [13]–[16] have been proposed to find logic bugs in GDBs. Most of these approaches (e.g., Grand [13], GDsmith [16], and RD2[15]) utilize differential testing to find logic bugs in multiple GDBs. However, they can miss logic bugs when all target GDBs return incorrect query results. Many testing approaches [17]–[26] (e.g., TLP [20], NoREC [21], and DQE [25]) have been proposed to test relational database systems. Among these approaches, NoREC [21] can detect optimization bugs in relational database systems by rewriting an optimized SQL query into a nonoptimizing SQL query.",
+      "context_after": "Thus, we still lack an effective approach to find optimization bugs in GDBs. In this paper, we propose Differential O ptimization T esting (DOT ), an automated approach to detect optimization bugs in Gremlin-based GDBs. We first randomly generate a graph database gdb. Based on the generated graph database gdb, we further randomly generate valid Gremlin queries. Then, for a generated Gremlin query Q, we generate candidate optimization configurations (an optimization configuration is a sequence of on and off states for optimization strategies supported by the target GDB), and execute Qon the graph",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 6257,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": ", Grand [13]) and query partitioning (i.",
+      "context_before": "At the time of writing this paper, we have found 28 unique optimization bugs in these six target GDBs. Among these 28 optimization bugs, 16 bugs have been verified as previously-unknown bugs, 5 out of which have been fixed by GDB developers. Our experimental results also show that our optimization-guided approach can detect 1.9x more bugs and 1.1x more unique bugs using 2.2x fewer test cases than the random approach, and our automated locating approach can accurately locate faulty optimization strategies. Furthermore, we compare DOT with the existing approaches, i.e., differential testing (i.e.",
+      "context_after": "e., GDBMeter [14]). For the 28 optimization bugs that DOT detected, 19 bugs are out of the scope of these approaches’ detection capability, and cannot be detected by them. We have made DOT available at https://github.com/tcse-iscas/DOT. To sum up, this paper makes the following contributions. •We propose DOT for finding optimization bugs in Gremlin-based GDBs by identifying inconsistencies among different optimization configurations. •To improve the efficiency of DOT, we propose an optimization-guided approach to explore more optimization strategies and graph database features. •We evaluate DOT o",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 8970,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[14]",
+      "technique": null,
+      "sentence": ", GDBMeter [14]).",
+      "context_before": "e found 28 unique optimization bugs in these six target GDBs. Among these 28 optimization bugs, 16 bugs have been verified as previously-unknown bugs, 5 out of which have been fixed by GDB developers. Our experimental results also show that our optimization-guided approach can detect 1.9x more bugs and 1.1x more unique bugs using 2.2x fewer test cases than the random approach, and our automated locating approach can accurately locate faulty optimization strategies. Furthermore, we compare DOT with the existing approaches, i.e., differential testing (i.e., Grand [13]) and query partitioning (i.e.",
+      "context_after": "For the 28 optimization bugs that DOT detected, 19 bugs are out of the scope of these approaches’ detection capability, and cannot be detected by them. We have made DOT available at https://github.com/tcse-iscas/DOT. To sum up, this paper makes the following contributions. •We propose DOT for finding optimization bugs in Gremlin-based GDBs by identifying inconsistencies among different optimization configurations. •To improve the efficiency of DOT, we propose an optimization-guided approach to explore more optimization strategies and graph database features. •We evaluate DOT on six widely-used G",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 9012,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[13]–\n[16]",
+      "technique": null,
+      "sentence": "Comparison with Existing Approaches To the best of our knowledge, four existing approaches [13]– [16] can detect logic bugs in GDBs.",
+      "context_before": "e optimization strategies and more graph database features. Further, we count the number of unique combinations of optimization strategies within five minutes running of DOT and DOT rand. As shown in Fig. 10, DOT can explore 1.3x more unique optimization combinations than DOT rand, demonstrating the contribution of our optimization-guided approach. These results further illustrate that optimization bugs can be more efficiently detected by exploring more unique optimization strategies. RQ2: DOT can explore more unique optimization strategies, thus exposing more optimization bugs than DOT rand. D.",
+      "context_after": "Specifically, three (i.e., Grand [13], GDsmith [16], and RD2[15]) use differential testing on multiple GDBs, and the remaining one (i.e., GDBMeter [14]) uses query partitioning. Therefore, we compare DOT with differential testing and query partitioning. Comparison with differential testing. Grand applies differential testing to detect logic bugs in GDBs. It generates the same graph database and executes the same Gremlin queries in multiple target GDBs to verify whether they can return the same results. Any discrepancy reveals a logic bug. To compare DOT with differential testing (e.g., Grand )",
+      "section": "D Comparison with Existing Approaches",
+      "page": 9,
+      "char_offset": 44201,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": ", GDBMeter [14]) uses query partitioning.",
+      "context_before": "an DOT rand, demonstrating the contribution of our optimization-guided approach. These results further illustrate that optimization bugs can be more efficiently detected by exploring more unique optimization strategies. RQ2: DOT can explore more unique optimization strategies, thus exposing more optimization bugs than DOT rand. D. Comparison with Existing Approaches To the best of our knowledge, four existing approaches [13]– [16] can detect logic bugs in GDBs. Specifically, three (i.e., Grand [13], GDsmith [16], and RD2[15]) use differential testing on multiple GDBs, and the remaining one (i.e.",
+      "context_after": "Therefore, we compare DOT with differential testing and query partitioning. Comparison with differential testing. Grand applies differential testing to detect logic bugs in GDBs. It generates the same graph database and executes the same Gremlin queries in multiple target GDBs to verify whether they can return the same results. Any discrepancy reveals a logic bug. To compare DOT with differential testing (e.g., Grand ), for each test case in the 28 optimization bug reports, we run it on our six target GDBs and verify whether their query results are the same. Any discrepancy will be considered",
+      "section": "D Comparison with Existing Approaches",
+      "page": 9,
+      "char_offset": 44470,
+      "found_by_all": [
+        "technique",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "Therefore, we compare DOT with differential testing and query partitioning.",
+      "context_before": "on of our optimization-guided approach. These results further illustrate that optimization bugs can be more efficiently detected by exploring more unique optimization strategies. RQ2: DOT can explore more unique optimization strategies, thus exposing more optimization bugs than DOT rand. D. Comparison with Existing Approaches To the best of our knowledge, four existing approaches [13]– [16] can detect logic bugs in GDBs. Specifically, three (i.e., Grand [13], GDsmith [16], and RD2[15]) use differential testing on multiple GDBs, and the remaining one (i.e., GDBMeter [14]) uses query partitioning.",
+      "context_after": "Comparison with differential testing. Grand applies differential testing to detect logic bugs in GDBs. It generates the same graph database and executes the same Gremlin queries in multiple target GDBs to verify whether they can return the same results. Any discrepancy reveals a logic bug. To compare DOT with differential testing (e.g., Grand ), for each test case in the 28 optimization bug reports, we run it on our six target GDBs and verify whether their query results are the same. Any discrepancy will be considered as an optimization bug detected by Grand. As a result, among the 28 optimiz",
+      "section": "D Comparison with Existing Approaches",
+      "page": 9,
+      "char_offset": 44511,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "Comparison with query partitioning.",
+      "context_before": "ame graph database and executes the same Gremlin queries in multiple target GDBs to verify whether they can return the same results. Any discrepancy reveals a logic bug. To compare DOT with differential testing (e.g., Grand ), for each test case in the 28 optimization bug reports, we run it on our six target GDBs and verify whether their query results are the same. Any discrepancy will be considered as an optimization bug detected by Grand. As a result, among the 28 optimization bugs, Grand can detect 9 bugs. In summary, Grand cannot detect the remaining 19 optimization bugs that DOT detected.",
+      "context_after": "GDBMeter applies query partitioning [20] to test GDBs. Specifically, it partitions a query into three disjoint queries in which a predicate is evaluated to true ,false andnull, respectively. Since we1v1 = g.addV( ’vL’ );// v1 2v2 = g.addV( ’vL’ );// v2 3 4g.withStrategies(CountStrategy) // on 5 .V().where(__.in().count().is(eq(-3))); 6 -- {Not a legal range: [0, -2]} 7 8g.withoutStrategies(CountStrategy) // off 9 .V().where(__.in().count().is(eq(-3))); 10 -- {} Fig. 11. A test case that triggers bug TinkerGraph#2891. cannot successfully run GDBMeter in practice and GDBMeter only supports testi",
+      "section": "D Comparison with Existing Approaches",
+      "page": 9,
+      "char_offset": 45309,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "GDBMeter applies query partitioning [20] to test GDBs.",
+      "context_before": "same Gremlin queries in multiple target GDBs to verify whether they can return the same results. Any discrepancy reveals a logic bug. To compare DOT with differential testing (e.g., Grand ), for each test case in the 28 optimization bug reports, we run it on our six target GDBs and verify whether their query results are the same. Any discrepancy will be considered as an optimization bug detected by Grand. As a result, among the 28 optimization bugs, Grand can detect 9 bugs. In summary, Grand cannot detect the remaining 19 optimization bugs that DOT detected. Comparison with query partitioning.",
+      "context_after": "Specifically, it partitions a query into three disjoint queries in which a predicate is evaluated to true ,false andnull, respectively. Since we1v1 = g.addV( ’vL’ );// v1 2v2 = g.addV( ’vL’ );// v2 3 4g.withStrategies(CountStrategy) // on 5 .V().where(__.in().count().is(eq(-3))); 6 -- {Not a legal range: [0, -2]} 7 8g.withoutStrategies(CountStrategy) // off 9 .V().where(__.in().count().is(eq(-3))); 10 -- {} Fig. 11. A test case that triggers bug TinkerGraph#2891. cannot successfully run GDBMeter in practice and GDBMeter only supports testing one Gremlin-based GDB (i.e., JanusGraph), we compare",
+      "section": "D Comparison with Existing Approaches",
+      "page": 9,
+      "char_offset": 45345,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[13]–[16]",
+      "technique": null,
+      "sentence": "Recently, many works [13]–[16] have been proposed to test the correctness of GDBs.",
+      "context_before": "caused by incorrect implementations of optimization strategies, but cannot test other logic bugs that are not related to optimization strategies. Second, DOT cannot test some optimization strategies (e.g., ProductiveByStrategy ) that can affect the query results when turing them on or off. Third, DOT cannot automatically remove duplicate bugs because themanifestations of bugs for the same root cause are diverse and complex, which poses a challenge for automatically removing duplicate bugs. We will address this limitation in the future. VII. RELA TED WORK Testing graph database systems (GDBs).",
+      "context_after": "Grand [13], RD2[15], and GDsmith [16] detect logic bugs in multiple GDBs by utilizing differential testing [46]– [48], while GDBMeter [14] utilize metamorphic testing [49] to find logic bugs in an individual GDB. These works cannot effectively detect optimization bugs in GDBs. Testing relational database systems (DBMSs). Researchers have developed various testing approaches on relational DBMSs [17]–[25], [50]–[56]. Some approaches [18], [19], [23], [50] (e.g., RAGS [18]) applied differential testing to test relational DBMSs. Some approaches [20], [21], [25] (e.g., NoREC [21]) utilizes metamorp",
+      "section": "VII RELA TED WORK",
+      "page": 10,
+      "char_offset": 50880,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[14]",
+      "technique": null,
+      "sentence": "Grand [13], RD2[15], and GDsmith [16] detect logic bugs in multiple GDBs by utilizing differential testing [46]– [48], while GDBMeter [14] utilize metamorphic testing [49] to find logic bugs in an individual GDB.",
+      "context_before": "her logic bugs that are not related to optimization strategies. Second, DOT cannot test some optimization strategies (e.g., ProductiveByStrategy ) that can affect the query results when turing them on or off. Third, DOT cannot automatically remove duplicate bugs because themanifestations of bugs for the same root cause are diverse and complex, which poses a challenge for automatically removing duplicate bugs. We will address this limitation in the future. VII. RELA TED WORK Testing graph database systems (GDBs). Recently, many works [13]–[16] have been proposed to test the correctness of GDBs.",
+      "context_after": "These works cannot effectively detect optimization bugs in GDBs. Testing relational database systems (DBMSs). Researchers have developed various testing approaches on relational DBMSs [17]–[25], [50]–[56]. Some approaches [18], [19], [23], [50] (e.g., RAGS [18]) applied differential testing to test relational DBMSs. Some approaches [20], [21], [25] (e.g., NoREC [21]) utilizes metamorphic testing for finding logic bugs in relational DBMSs. Query generation [17], [54] (e.g., SQLsmith [17]) and generic fuzzing approaches (e.g., AFL [53]) can also be used to detect bugs in relational DBMSs. All th",
+      "section": "VII RELA TED WORK",
+      "page": 10,
+      "char_offset": 50963,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2023.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[17]–[25]",
+      "technique": "tlp",
+      "sentence": "Researchers have developed various testing approaches on relational DBMSs [17]–[25], [50]–[56].",
+      "context_before": "re diverse and complex, which poses a challenge for automatically removing duplicate bugs. We will address this limitation in the future. VII. RELA TED WORK Testing graph database systems (GDBs). Recently, many works [13]–[16] have been proposed to test the correctness of GDBs. Grand [13], RD2[15], and GDsmith [16] detect logic bugs in multiple GDBs by utilizing differential testing [46]– [48], while GDBMeter [14] utilize metamorphic testing [49] to find logic bugs in an individual GDB. These works cannot effectively detect optimization bugs in GDBs. Testing relational database systems (DBMSs).",
+      "context_after": "Some approaches [18], [19], [23], [50] (e.g., RAGS [18]) applied differential testing to test relational DBMSs. Some approaches [20], [21], [25] (e.g., NoREC [21]) utilizes metamorphic testing for finding logic bugs in relational DBMSs. Query generation [17], [54] (e.g., SQLsmith [17]) and generic fuzzing approaches (e.g., AFL [53]) can also be used to detect bugs in relational DBMSs. All the above testing approaches target relational DBMSs with declarative query language (i.e., SQL), which has totally different query patterns and syntaxes from the procedural Gremlin query language. Therefore,",
+      "section": "VII RELA TED WORK",
+      "page": 10,
+      "char_offset": 51285,
+      "cited_reference": {
+        "number": 20,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” in Proceedings of ACM Object-Oriented Programming Systems, Languages, and Applications (OOPSLA), 2020, pp. 211:1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[20]",
+      "technique": "tlp",
+      "sentence": "Some approaches [20], [21], [25] (e.",
+      "context_before": "ny works [13]–[16] have been proposed to test the correctness of GDBs. Grand [13], RD2[15], and GDsmith [16] detect logic bugs in multiple GDBs by utilizing differential testing [46]– [48], while GDBMeter [14] utilize metamorphic testing [49] to find logic bugs in an individual GDB. These works cannot effectively detect optimization bugs in GDBs. Testing relational database systems (DBMSs). Researchers have developed various testing approaches on relational DBMSs [17]–[25], [50]–[56]. Some approaches [18], [19], [23], [50] (e.g., RAGS [18]) applied differential testing to test relational DBMSs.",
+      "context_after": "g., NoREC [21]) utilizes metamorphic testing for finding logic bugs in relational DBMSs. Query generation [17], [54] (e.g., SQLsmith [17]) and generic fuzzing approaches (e.g., AFL [53]) can also be used to detect bugs in relational DBMSs. All the above testing approaches target relational DBMSs with declarative query language (i.e., SQL), which has totally different query patterns and syntaxes from the procedural Gremlin query language. Therefore, they cannot be applied to find optimization bugs in GDBs. Compiler optimization testing. In compiler testing [57], many approaches [58]–[65] have bee",
+      "section": "VII RELA TED WORK",
+      "page": 10,
+      "char_offset": 51493,
+      "cited_reference": {
+        "number": 20,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” in Proceedings of ACM Object-Oriented Programming Systems, Languages, and Applications (OOPSLA), 2020, pp. 211:1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": ", NoREC [21]) utilizes metamorphic testing for finding logic bugs in relational DBMSs.",
+      "context_before": "o test the correctness of GDBs. Grand [13], RD2[15], and GDsmith [16] detect logic bugs in multiple GDBs by utilizing differential testing [46]– [48], while GDBMeter [14] utilize metamorphic testing [49] to find logic bugs in an individual GDB. These works cannot effectively detect optimization bugs in GDBs. Testing relational database systems (DBMSs). Researchers have developed various testing approaches on relational DBMSs [17]–[25], [50]–[56]. Some approaches [18], [19], [23], [50] (e.g., RAGS [18]) applied differential testing to test relational DBMSs. Some approaches [20], [21], [25] (e.g.",
+      "context_after": "Query generation [17], [54] (e.g., SQLsmith [17]) and generic fuzzing approaches (e.g., AFL [53]) can also be used to detect bugs in relational DBMSs. All the above testing approaches target relational DBMSs with declarative query language (i.e., SQL), which has totally different query patterns and syntaxes from the procedural Gremlin query language. Therefore, they cannot be applied to find optimization bugs in GDBs. Compiler optimization testing. In compiler testing [57], many approaches [58]–[65] have been proposed to test compiler optimizers. Some works (e.g., [59], [64]) randomly generate",
+      "section": "VII RELA TED WORK",
+      "page": 10,
+      "char_offset": 51532,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M10"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:51:41Z",
+    "is_model_written": true,
+    "summary": "DOT tests Gremlin-based graph databases for optimisation bugs by running a traversal with the system's optimisations enabled and disabled and comparing the results. The paper's starting point is that NoREC's idea does not carry across: a Gremlin traversal is procedural and cannot be rewritten into a non-optimising equivalent the way a declarative SQL query can, so the equivalent must be obtained by controlling the optimiser instead of by rewriting the query.",
+    "narrative": "SQLancer's techniques frame the paper twice over. NoREC is the direct inspiration and the direct obstacle: DOT targets the same class of optimisation bugs, but the paper explains at length why NoREC's rewriting cannot be applied to Gremlin's procedural queries, which is what forces the different mechanism. Query partitioning is the comparison, reached through GDBMeter, one of four existing approaches for finding logic bugs in graph databases; the paper compares DOT with differential testing and with query partitioning as the two families that can find these bugs.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "definition",
+      "M4": "motivation",
+      "M5": "motivation",
+      "M6": "definition",
+      "M7": "definition",
+      "M8": "baseline",
+      "M9": "definition",
+      "M10": "baseline",
+      "M11": "result_comparison",
+      "M12": "definition",
+      "M13": "background",
+      "M14": "definition",
+      "M15": "background",
+      "M16": "background",
+      "M17": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Nothing describes DOT's implementation as reusing SQLancer's code; GDBMeter is run as a comparison target."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M4 and M5 set out why NoREC's rewriting cannot be applied to Gremlin at all, which is the reason DOT controls the optimiser instead. Sharing a target bug class with NoREC is not extending it."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M8",
+          "M10",
+          "M11"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M8",
+            "sentence": "Comparison with Existing Approaches To the best of our knowledge, four existing approaches [13]– [16] can detect logic bugs in GDBs.",
+            "section": "D Comparison with Existing Approaches",
+            "page": 9
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "Therefore, we compare DOT with differential testing and query partitioning.",
+            "section": "D Comparison with Existing Approaches",
+            "page": 9
+          },
+          {
+            "mention_id": "M11",
+            "sentence": "Comparison with query partitioning.",
+            "section": "D Comparison with Existing Approaches",
+            "page": 9
+          }
+        ],
+        "reasoning": "M8 identifies GDBMeter among the four existing approaches, M10 states DOT is compared with differential testing and query partitioning, and M11 heads that comparison. M9 and M12 identify GDBMeter as the query-partitioning tool.",
+        "techniques": [
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The approaches are described by mechanism and by applicability; no state-of-the-art claim is made."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_ictai62512_2024_00085.json b/_data/papers/paper_doi_10_1109_ictai62512_2024_00085.json
new file mode 100644
index 0000000..9e8d7f6
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_ictai62512_2024_00085.json
@@ -0,0 +1,675 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:28Z",
+  "paper": {
+    "id": "paper:doi:10.1109/ictai62512.2024.00085",
+    "title": "NNTailor: A Neural Network-Driven Fuzzer for DataBase Management Systems",
+    "authors": [
+      "Shutao Chu",
+      "Yongjun Wang",
+      "Haoran Xu",
+      "Zhiyuan Jiang",
+      "Yongxin Chen"
+    ],
+    "year": 2024,
+    "venue": "IEEE International Conference on Tools with Artificial Intelligence",
+    "doi": "10.1109/ictai62512.2024.00085",
+    "arxiv_id": null,
+    "s2_paper_id": "4577684b2119f110b0704f21225e1212fa4b33dd",
+    "url": "https://doi.org/10.1109/ictai62512.2024.00085",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/ictai62512.2024.00085",
+      "retrieved_at": "2026-09-09T01:31:28Z",
+      "chars": 40828,
+      "content_sha256": "sha256:f0d9e0cf6b2153d8c3169f17fd7d8aac81c8c300f9123bd17be4355350d1847b"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 9,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1376
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 8477
+      },
+      {
+        "number": "III",
+        "title": "DESIGN",
+        "start": 13006
+      },
+      {
+        "number": "B",
+        "title": "Preprocessing",
+        "start": 14399
+      },
+      {
+        "number": "C",
+        "title": "Extracting Code fragments",
+        "start": 15339
+      },
+      {
+        "number": "D",
+        "title": "Training",
+        "start": 17426
+      },
+      {
+        "number": "E",
+        "title": "Mutations",
+        "start": 19851
+      },
+      {
+        "number": "F",
+        "title": "Convert AST back to SQL",
+        "start": 22957
+      },
+      {
+        "number": "G",
+        "title": "Test",
+        "start": 23397
+      },
+      {
+        "number": "A",
+        "title": "Experimental Setup",
+        "start": 24613
+      },
+      {
+        "number": "B",
+        "title": "Code Coverage",
+        "start": 26856
+      },
+      {
+        "number": "C",
+        "title": "Validity",
+        "start": 28529
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Data science revealed: A data-driven glimpse into the burgeoning new field. URL https://www. predictiveanalyticsworld.com/machinelearningtimes/ august12/emc-data-science-study.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Wikipedia, 2024. URL https://en.wikipedia.org/wiki/ Database.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Seltenreich A. Bug squashing with sqlsmith, 2018. URL https://www.postgresql.eu/events/pgconfeu2018/ sessions/session/2221/slides/145/sqlsmith-talk.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Uri Alon, Roy Sadaka, Omer Levy, and Eran Yahav. Structural language models of code, 2020. URL https: //arxiv.org/abs/1910.00577.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Seltenreich Andreas, Tang Bo, and Mullender Sjoerd. A random sql query generator, 2015. URL https://github. com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Jinsheng Ba and Manuel Rigger. Testing database engines via query plan guidance. In The 45th International Conference on Software Engineering (ICSE’23), May 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 7,
+      "text": "caca labs. Zzuf-multi-purpose fuzzer, 2009. URL http: //caca.zoy.org/wiki/zzuf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Chris Cummins, Pavlos Petoumenos, Alastair Murray, and Hugh Leather. Compiler fuzzing through deep learning. In Proceedings of the 27th ACMSIGSOFT In570 Fuzzer PostgreSQL SQLite NNTailor 34% 37% SQLsmith 29% 24% SQLancer 31% 34% TABLE I: Code CoverageFuzzer PostgreSQL syntax PostgreSQL semantic SQLite syntax SQLite semantic NNTailor 94% 41% 91% 46% SQLsmith 92% 34% 79% 37% SQLancer 75% 29% 91% 66%",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 9,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. Griffin: Grammar-free dbms fuzzing. InProceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, ASE ’22, New York, NY, USA, 2023. Association for Computing Machinery. ISBN 9781450394758. doi: 10. 1145/3551349.3560431. URL https://doi.org/10.1145/ 3551349.3560431.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Xiyue Gao, Zhuang Liu, Jiangtao Cui, Hui Li, Hui Zhang, Kewei Wei, and Kankan Zhao. A comprehensive survey on database management system fuzzing: Techniques, taxonomy and experimental comparison, 2023. URL https://arxiv.org/pdf/2311.06728.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Patrice Godefroid, Hila Peleg, and Rishabh Singh. Learn&fuzz: Machine learning for input fuzzing. In Proceedings of the 32nd IEEE/ACM International Conference on Automated Software Engineering, pages 50– 59. IEEE Press, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "HyungSeok Han, DongHyeon Oh, and Sang Kil Cha. Codealchemist: Semantics-aware code generation to find vulnerabilities in javascript engines. In NDSS, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Zhiyuan Jiang Shuhui Fan Shaojing Fu Peidai Xie Haoran Xu, Yongjun Wang. Fuzzing javascript engines with a syntax-aware neural program model, 2024. URL https://doi.org/10.1016/j.cose.2024.103947.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Sepp Hochreiter and J ¨urgen Schmidhuber. Long shortterm memory. Neural computation, 9(8):1735–1780, 1997.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Christian Holler, Kim Herzig, and Andreas Zeller. Fuzzing with code fragments. In Presented as part of the 21st {USENIX} Security Symposium ({USENIX} Security 12), pages 445–458, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Liang J, Wu ZY, Fu JZ, Zhu J, Jiang Y, and Sun JG. Survey on database management system fuzzing techniques, 2023. URL http://www.jos.org.cn/1000-9825/7048.htm.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. DynSQL: Stateful fuzzing for database management systems with complex and valid SQL query generation. In 32nd USENIX Security Symposium (USENIX Security 23), pages 4949–4965, Anaheim, CA, August 2023. USENIX Association. ISBN 978-1939133-37-3. URL https://www.usenix.org/conference/ usenixsecurity23/presentation/jiang-zu-ming.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. Apollo: automatic detection and diagnosis of performance regressions in database systems.Proc. VLDB Endow., 13(1):57–70, sep 2019. ISSN 2150-8097. doi: 10.14778/3357377.3357382. URL https://doi.org/10.14778/3357377.3357382.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Suyoung Lee, HyungSeok Han, Sang Kil Cha, and Sooel Son. Montage: A neural network language modelguided javascript engine fuzzer. In 29th USENIX Security Symposium (USENIX Security 20), pages 2613–2630, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Jun Li, Bodong Zhao, and Chaomo Zhang. Fuzzing: a survey. Cybersecurity, 1, 2018. URL https://api. semanticscholar.org/CorpusID:257107250.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. Sequence-oriented dbms fuzzing. In Data Engineering (ICDE), pages 668–681, 2023. doi: 10.1109/ICDE55515.2023.00057.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Yu Liang, Song Liu, and Hong Hu. Detecting logical bugs of DBMS with coverage-based guidance. In 31st USENIX Security Symposium (USENIX Security 22), pages 4309–4326, Boston, MA, August 2022. USENIX Association. ISBN 978-1939133-31-1. URL https://www.usenix.org/conference/ usenixsecurity22/presentation/liang.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "LLVM. libfuzzer:a library for coverage-guided fuzz testing, 2017. URL https://llvm.org/docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Gehrke J Ramakrishnan R. Database Management Systems. McGraw-Hili, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 2020 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2020, 2020. doi: 10.1145/3368089.3409710.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA), 2020. doi: 10.1145/3428279.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20, page 667–682, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "Hong Hu Hangfan Zhang Wenke Lee Rui Zhong, Yongheng Chen and Dinghao Wu. Squirrel: Testing database management systems with language validity and coverage feedback, 2020. URL https://dl.acm.org/doi/ pdf/10.1145/3372297.3417260. 571",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Ilya Sutskever, James Martens, and Geoffrey E Hinton. Generating text with recurrent neural networks. In Proceedings of the 28th International Conference on Machine Learning (ICML-11), pages 1017–1024, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Spandan Veggalam, Sanjay Rawat, Istvan Haller, and Herbert Bos. Ifuzzer: An evolutionary interpreter fuzzer using genetic programming. In European Symposium on Research in Computer Security, pages 581–601. Springer, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Liu L Feng LZ Bao T Peng T Wang NY, Ye YX. Language models based on deep learning: A review, 2021. URL http://www.jos.org.cn/1000-9825/6169.htm.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Michal Zalewski. american fuzzy lop, 2017. URL http: //lcamtuf.coredump.cx/afl.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Xuya Cong Xiaobo Li Li Yin Zhenhua Yu, Zhengqi Liu. Fuzzing: Progress, challenges, and perspectives. Computers, Materials & Continua, 78(1):1–29, 2024. ISSN 1546-2226. doi: 10.32604/cmc.2023.042361. URL http://www.techscience.com/cmc/v78n1/55370.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Xiaogang Zhu, Sheng Wen, Seyit Camtepe, and Yang Xiang. Fuzzing: a survey for roadmap. ACM Computing Surveys (CSUR), 54(11s):1–36, 2022. 572",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 6,
+      "text": "Jinsheng Ba and Manuel Rigger. Testing database engines via query plan guidance. In The 45th International Conference on Software Engineering (ICSE’23), May 2023.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 8,
+      "text": "Chris Cummins, Pavlos Petoumenos, Alastair Murray, and Hugh Leather. Compiler fuzzing through deep learning. In Proceedings of the 27th ACMSIGSOFT In570 Fuzzer PostgreSQL SQLite NNTailor 34% 37% SQLsmith 29% 24% SQLancer 31% 34% TABLE I: Code CoverageFuzzer PostgreSQL syntax PostgreSQL semantic SQLite syntax SQLite semantic NNTailor 94% 41% 91% 46% SQLsmith 92% 34% 79% 37% SQLancer 75% 29% 91% 66% TABLE II: Validity ternational Symposium on Software Testing and Analysis, pages 95–105. ACM, 2018.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 2020 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2020, 2020. doi: 10.1145/3368089.3409710.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA), 2020. doi: 10.1145/3428279.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20, page 667–682, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We conducted evaluations of NNTailor on the latest versions of PostgreSQL and SQLite, comparing its performance with SQLsmith[5] and SQLancer[6, 25, 26, 27].",
+      "context_before": "ssion test cases as a seed set, preprocessing each seed’s corresponding AST by extracting 1-level subtrees as code fragment in a pre-order traversal from the root node. These fragments form a sequence input for model. The preprocessed inputs are fed into the neural network model for AST completion through subtree pruning, resulting in a complete AST. This AST is then converted into an SQL test case. The mutated test case is then used as input of the SQL engine. We introduces NNTailor, a prototype for fuzzing DBMS that enhances efficiency through a neural network model-driven mutation approach.",
+      "context_after": "With the neural network model-driven mutation approach, NNTailor achieves 35% and 10% higher code line coverage compared to SQLsmith and SQLancer. In terms of test case effectiveness, NNTailor is 23% higher than SQLsmith. NNTailor and SQLancer have their own advantages in different DBMS.Experimental results demonstrate that NNTailor surpasses open-source black box fuzzing tools such as SQLsmith and SQLancer in terms of code coverage and test case effectiveness. Overall, the method proposed in this paper effectively complements existing approaches in DBMS fuzzing. The contributions of this pap",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 7035,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "With the neural network model-driven mutation approach, NNTailor achieves 35% and 10% higher code line coverage compared to SQLsmith and SQLancer.",
+      "context_before": "root node. These fragments form a sequence input for model. The preprocessed inputs are fed into the neural network model for AST completion through subtree pruning, resulting in a complete AST. This AST is then converted into an SQL test case. The mutated test case is then used as input of the SQL engine. We introduces NNTailor, a prototype for fuzzing DBMS that enhances efficiency through a neural network model-driven mutation approach. We conducted evaluations of NNTailor on the latest versions of PostgreSQL and SQLite, comparing its performance with SQLsmith[5] and SQLancer[6, 25, 26, 27].",
+      "context_after": "In terms of test case effectiveness, NNTailor is 23% higher than SQLsmith. NNTailor and SQLancer have their own advantages in different DBMS.Experimental results demonstrate that NNTailor surpasses open-source black box fuzzing tools such as SQLsmith and SQLancer in terms of code coverage and test case effectiveness. Overall, the method proposed in this paper effectively complements existing approaches in DBMS fuzzing. The contributions of this paper are summarized as follows: •We propose a method based on AST subtree mutation, employing a strategy of subtree replacement of the same type to e",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 7193,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "NNTailor and SQLancer have their own advantages in different DBMS.",
+      "context_before": "into an SQL test case. The mutated test case is then used as input of the SQL engine. We introduces NNTailor, a prototype for fuzzing DBMS that enhances efficiency through a neural network model-driven mutation approach. We conducted evaluations of NNTailor on the latest versions of PostgreSQL and SQLite, comparing its performance with SQLsmith[5] and SQLancer[6, 25, 26, 27]. With the neural network model-driven mutation approach, NNTailor achieves 35% and 10% higher code line coverage compared to SQLsmith and SQLancer. In terms of test case effectiveness, NNTailor is 23% higher than SQLsmith.",
+      "context_after": "Experimental results demonstrate that NNTailor surpasses open-source black box fuzzing tools such as SQLsmith and SQLancer in terms of code coverage and test case effectiveness. Overall, the method proposed in this paper effectively complements existing approaches in DBMS fuzzing. The contributions of this paper are summarized as follows: •We propose a method based on AST subtree mutation, employing a strategy of subtree replacement of the same type to ensure syntactic correctness and semantic validity of produced test cases. •We utilize a neural network model to extract features from high-qua",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 7415,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Experimental results demonstrate that NNTailor surpasses open-source black box fuzzing tools such as SQLsmith and SQLancer in terms of code coverage and test case effectiveness.",
+      "context_before": "of the SQL engine. We introduces NNTailor, a prototype for fuzzing DBMS that enhances efficiency through a neural network model-driven mutation approach. We conducted evaluations of NNTailor on the latest versions of PostgreSQL and SQLite, comparing its performance with SQLsmith[5] and SQLancer[6, 25, 26, 27]. With the neural network model-driven mutation approach, NNTailor achieves 35% and 10% higher code line coverage compared to SQLsmith and SQLancer. In terms of test case effectiveness, NNTailor is 23% higher than SQLsmith. NNTailor and SQLancer have their own advantages in different DBMS.",
+      "context_after": "Overall, the method proposed in this paper effectively complements existing approaches in DBMS fuzzing. The contributions of this paper are summarized as follows: •We propose a method based on AST subtree mutation, employing a strategy of subtree replacement of the same type to ensure syntactic correctness and semantic validity of produced test cases. •We utilize a neural network model to extract features from high-quality seed sets, enabling the generation of higher-quality test cases. •We introduce a language model-driven by neural networks to facilitate cross-DBMS test case generation thro",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 7482,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[8]",
+      "technique": null,
+      "sentence": "[8] introduced a method leveraging Long Short-Term Memory (LSTM)[14] models to autonomously learn syntax specifications of the OpenCL programming language.",
+      "context_before": "rates and ensure robust testing outcomes. Recent advancements in neural network language models have revolutionized test case mutation methods across various software domains. Researchers have conceptualized test case generation as a language modeling [31] challenge, employing Recurrent Neural Networks (RNNs) as fundamental components[29] to produce fuzz test cases. Initially, Godefroid et al. [11] applied RNNs to learn PDF file format specifications, subsequently using a character-level generative model 565 Fig. 1: Overview of NNTailor for fuzzing PDF parsers. Following this, Cummins et al.",
+      "context_after": "Their approach facilitated the automatic production of compiler fuzz test cases. Their prototype tool, DeepSmith, predicts subsequent tokens based on preceding token sequences, utilizing a token-level language model specific to OpenCL. Source code contains rich structured information. Learning source code representations through AST node sequences enhances model’s ability to understand the code and improves code completion[4] accuracy. Constructing source programs using ASTs based on code fragments has been successful in fuzzing code execution engines. LangFuzz [15], for instance, extracts co",
+      "section": "II BACKGROUND",
+      "page": 3,
+      "char_offset": 11262,
+      "cited_reference": {
+        "number": 8,
+        "text": "Chris Cummins, Pavlos Petoumenos, Alastair Murray, and Hugh Leather. Compiler fuzzing through deep learning. In Proceedings of the 27th ACMSIGSOFT In570 Fuzzer PostgreSQL SQLite NNTailor 34% 37% SQLsmith 29% 24% SQLancer 31% 34% TABLE I: Code CoverageFuzzer PostgreSQL syntax PostgreSQL semantic SQLi",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "EVALUATION This section compares NNTailor with typical DBMS blackbox fuzzing tools SQLsmith and SQLancer in terms of coverage on the target DBMS, syntactic and semantic correctness of synthetic queries.",
+      "context_before": "or executes them on the target DBMS. It inputs the test cases into the instrumented DBMS, monitors the system’s feedback, and collects statistics on code coverage, as well as grammatical and semantic validity. 568 IMPLEMENTATION This paper employs Python 3.10 for development and implements NNTailor on a 64-bit Ubuntu 20.04 machine featuring an Intel i9-13900K processor with 24 cores (3.0GHz), 32 threads, 16GB memory, and an NVIDIA GeForce RTX 4090 graphics card. It utilizes sqlite-parser to convert SQL queries to AST, and Algorithm 2 handles the reverse conversion of AST back to SQL queries.",
+      "context_after": "SQLsmith detects database system crashes,while SQLancer detects database system logical problems. We aim to answer the following three research questions through our experiments: RQ1: How effective is NNTailor’s coverage of DBMSs? RQ2: Can NNTailor handle the variability across different DBMS dialects? RQ3: How does NNTailor compare with advanced DBMS fuzzing tools? A. Experimental Setup Evaluation Targets. To assess versatility and efficiency, most tools typically evaluate three popular open-source DBMS: PostgreSQL, MySQL, and SQLite. These systems are widely adopted in industry and have und",
+      "section": "G Test",
+      "page": 6,
+      "char_offset": 24040,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLsmith detects database system crashes,while SQLancer detects database system logical problems.",
+      "context_before": "idity. 568 IMPLEMENTATION This paper employs Python 3.10 for development and implements NNTailor on a 64-bit Ubuntu 20.04 machine featuring an Intel i9-13900K processor with 24 cores (3.0GHz), 32 threads, 16GB memory, and an NVIDIA GeForce RTX 4090 graphics card. It utilizes sqlite-parser to convert SQL queries to AST, and Algorithm 2 handles the reverse conversion of AST back to SQL queries. EVALUATION This section compares NNTailor with typical DBMS blackbox fuzzing tools SQLsmith and SQLancer in terms of coverage on the target DBMS, syntactic and semantic correctness of synthetic queries.",
+      "context_after": "We aim to answer the following three research questions through our experiments: RQ1: How effective is NNTailor’s coverage of DBMSs? RQ2: Can NNTailor handle the variability across different DBMS dialects? RQ3: How does NNTailor compare with advanced DBMS fuzzing tools? A. Experimental Setup Evaluation Targets. To assess versatility and efficiency, most tools typically evaluate three popular open-source DBMS: PostgreSQL, MySQL, and SQLite. These systems are widely adopted in industry and have undergone extensive testing. At the time of writing, SQLsmith’s official support for PostgreSQL and S",
+      "section": "G Test",
+      "page": 6,
+      "char_offset": 24243,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "At the time of writing, SQLsmith’s official support for PostgreSQL and SQLite was notably robust,and SQLancer has not yet implemented the NoREC test oracle on MySQL.",
+      "context_before": "e system crashes,while SQLancer detects database system logical problems. We aim to answer the following three research questions through our experiments: RQ1: How effective is NNTailor’s coverage of DBMSs? RQ2: Can NNTailor handle the variability across different DBMS dialects? RQ3: How does NNTailor compare with advanced DBMS fuzzing tools? A. Experimental Setup Evaluation Targets. To assess versatility and efficiency, most tools typically evaluate three popular open-source DBMS: PostgreSQL, MySQL, and SQLite. These systems are widely adopted in industry and have undergone extensive testing.",
+      "context_after": "Therefore, this paper selected PostgreSQL and SQLite as the evaluation targets. Experimental environment. The experiments in this paper were conducted on a 64-bit Ubuntu 20.04 machine equipped with a 24-core Intel i9-13900K processor running at 3.0GHz and 16GiB of RAM. To facilitate quantitative comparisons, the experiment duration was set to 24 hours, aligning with common practice in evaluating fuzzing tools. Test process. The experiment consists of four main steps: environment preparation, DBMS construction, test execution, and result collection. 1.Environment Preparation: This involves dow",
+      "section": "A Experimental Setup",
+      "page": 6,
+      "char_offset": 24868,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLsmith and SQLancer achieves considerable coverage using predefined rules and extensive executions, yet its reliance on templates limits their ability to fully explore the code space.",
+      "context_before": "1) Compiling the program with coverage parameters, 2) Executing the instrumented program, and 3) Generating a coverage report. Specifically, clang was used to compile the DBMS with ”-fprofile-instrgenerate -fcoverage-mapping” parameters. Test cases produced by the fuzzing tool were executed on this instrumented DBMS version to capture raw profiles. These raw profiles were then merged using llvm-profdata into an intermediate file, which was subsequently used by llvm-cov to produce a detailed coverage statistical report. Table I displays row coverage for each fuzzing tool on the target database.",
+      "context_after": "In contrast, NNTailor employs mutation based on grammatically correct seed sets derived from AST. This approach ensures syntactic correctness without template constraints, allowing for broader exploration of state spaces. Moreover, NNTailor leverages a neural network model to extract features from high-quality seed sets, enhancing its ability to utilize context dependencies in subtree replacements of the same type. This strategy improves semantic validity and results in superior coverage performance compared to SQLsmith and SQLancer. C. Validity Ensuring both grammatical and semantic correctn",
+      "section": "B Code Coverage",
+      "page": 6,
+      "char_offset": 27802,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This strategy improves semantic validity and results in superior coverage performance compared to SQLsmith and SQLancer.",
+      "context_before": "mith and SQLancer achieves considerable coverage using predefined rules and extensive executions, yet its reliance on templates limits their ability to fully explore the code space. In contrast, NNTailor employs mutation based on grammatically correct seed sets derived from AST. This approach ensures syntactic correctness without template constraints, allowing for broader exploration of state spaces. Moreover, NNTailor leverages a neural network model to extract features from high-quality seed sets, enhancing its ability to utilize context dependencies in subtree replacements of the same type.",
+      "context_after": "C. Validity Ensuring both grammatical and semantic correctness in test cases is crucial for effective DBMS testing and poses a significant challenge for fuzzing tools. To evaluate the grammatical and semantic accuracy of these tools on the target DBMS, this study collected test cases generated over 24 hours by each tool, parsed them into expressions, and re-executed them. The DBMS feedback determined whether each expression adhered to grammatical and semantic rules. Specifically, if the DBMS detected a grammatical error, the expression was marked as syntactically incorrect. If other errors we",
+      "section": "B Code Coverage",
+      "page": 6,
+      "char_offset": 28407,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer is not fully adapted to the latest version of PostgreSQL, so its grammatical correctness is lower than that of other tools.",
+      "context_before": "ally incorrect. If other errors were flagged by the DBMS, the expression was deemed semantically incorrect. This paper defines grammatical correctness as the proportion of SQL expressions without syntax errors, and semantic 569 correctness as the proportion passing the DBMS’s semantic checks out of the total expressions tested. Table II displays the grammatical and semantic accuracy of test cases generated by each fuzzing tool within a 24hour period. SQLsmith achieves high grammatical correctness on PostgreSQL by embedding PostgreSQL’s grammar directly into its code for test case generation.",
+      "context_after": "Similarly, NNTailor maintains grammatical correctness by initially using complete and correct grammar in its seed cases. In terms of semantic correctness, there are notable differences among the tools. SQLsmith primarily generates SELECT expressions based on templates, which, despite lacking semantic checks, minimally alters the database schema, thus maintaining reasonable semantic correctness. SQLancer’s test oracle require it to generate SQL expressions according to predefined semantic rules, so it can achieve high semantic correctness on SQLite.In contrast, NNTailor introduces a wider range",
+      "section": "C Validity",
+      "page": 7,
+      "char_offset": 29694,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer’s test oracle require it to generate SQL expressions according to predefined semantic rules, so it can achieve high semantic correctness on SQLite.",
+      "context_before": "PostgreSQL’s grammar directly into its code for test case generation. SQLancer is not fully adapted to the latest version of PostgreSQL, so its grammatical correctness is lower than that of other tools.Similarly, NNTailor maintains grammatical correctness by initially using complete and correct grammar in its seed cases. In terms of semantic correctness, there are notable differences among the tools. SQLsmith primarily generates SELECT expressions based on templates, which, despite lacking semantic checks, minimally alters the database schema, thus maintaining reasonable semantic correctness.",
+      "context_after": "In contrast, NNTailor introduces a wider range of SQL expression types beyond SELECT queries, which enhances diversity but also heightens the challenge of maintaining semantic correctness. RELATED WORK DBMS Fuzzing. In recent years, several DBMS fuzzing tools like Sqlsmith, Sqlancer, Squirrel [28], SQLRight [22], DynSQL [17], Apollo[18], Griffin[9], and LEGO[21] have emerged, effectively uncovering numerous issues in popular DBMS and bolstering their security. These tools are categorized based on the types of vulnerabilities they target: crash vulnerabilities, logic vulnerabilities, and perfor",
+      "section": "C Validity",
+      "page": 7,
+      "char_offset": 30224,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "Sqlancer",
+      "technique": null,
+      "sentence": "In recent years, several DBMS fuzzing tools like Sqlsmith, Sqlancer, Squirrel [28], SQLRight [22], DynSQL [17], Apollo[18], Griffin[9], and LEGO[21] have emerged, effectively uncovering numerous issues in popular DBMS and bolstering their security.",
+      "context_before": "ble differences among the tools. SQLsmith primarily generates SELECT expressions based on templates, which, despite lacking semantic checks, minimally alters the database schema, thus maintaining reasonable semantic correctness. SQLancer’s test oracle require it to generate SQL expressions according to predefined semantic rules, so it can achieve high semantic correctness on SQLite.In contrast, NNTailor introduces a wider range of SQL expression types beyond SELECT queries, which enhances diversity but also heightens the challenge of maintaining semantic correctness. RELATED WORK DBMS Fuzzing.",
+      "context_after": "These tools are categorized based on the types of vulnerabilities they target: crash vulnerabilities, logic vulnerabilities, and performance vulnerabilities. Each tool establishes specific test oracle to detect these vulnerabilities. For crash vulnerabilities, tools typically utilize ASAN and system signal detection. Other types of vulnerabilities are addressed with innovative oracle like PQS and TLP introduced by SQLancer. Test case generation forms the core of DBMS fuzzing, primarily employing rule-based generation and syntax tree-based mutation methods. These approaches often rely on feedb",
+      "section": "C Validity",
+      "page": 7,
+      "char_offset": 30596,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Other types of vulnerabilities are addressed with innovative oracle like PQS and TLP introduced by SQLancer.",
+      "context_before": "ness. RELATED WORK DBMS Fuzzing. In recent years, several DBMS fuzzing tools like Sqlsmith, Sqlancer, Squirrel [28], SQLRight [22], DynSQL [17], Apollo[18], Griffin[9], and LEGO[21] have emerged, effectively uncovering numerous issues in popular DBMS and bolstering their security. These tools are categorized based on the types of vulnerabilities they target: crash vulnerabilities, logic vulnerabilities, and performance vulnerabilities. Each tool establishes specific test oracle to detect these vulnerabilities. For crash vulnerabilities, tools typically utilize ASAN and system signal detection.",
+      "context_after": "Test case generation forms the core of DBMS fuzzing, primarily employing rule-based generation and syntax tree-based mutation methods. These approaches often rely on feedback mechanisms such as DBMS coverage and query legitimacy. Given the diverse dialects and characteristics of different DBMS, most fuzzing tools require customization to adapt to specific DBMS. Compared to traditional methods, our approach significantly reduces the development overhead of supporting diverse DBMS targets and fuzzing tools using neural network language models. Neural Network Driven Fuzzing. Neural network langu",
+      "section": "C Validity",
+      "page": 7,
+      "char_offset": 31164,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Experimental results demonstrate that, compared to the classic black-box DBMS fuzzing tools SQLsmith and SQLancer, NNTailor outperforms SQLsmith and SQLancer in terms of comprehensive performance in code coverage and test case validity.",
+      "context_before": "test case validity, extracting features from high-quality seed sets, and overcoming the limitations imposed by different DBMS dialects on the portability of test case generation methods in DBMS fuzzing. We introduce a method driven by a neural network model for mutation-based test case generation and implement this approach in a prototype called NNTailor for DBMS fuzzing. This method significantly reduces the manual effort and development time required, minimizes the development work associated with test case generation logic, and mitigates the effectiveness issues caused by random mutations.",
+      "context_after": "REFERENCES [1] Data science revealed: A data-driven glimpse into the burgeoning new field. URL https://www. predictiveanalyticsworld.com/machinelearningtimes/ august12/emc-data-science-study.pdf. [2] Wikipedia, 2024. URL https://en.wikipedia.org/wiki/ Database. [3] Seltenreich A. Bug squashing with sqlsmith, 2018. URL https://www.postgresql.eu/events/pgconfeu2018/ sessions/session/2221/slides/145/sqlsmith-talk.pdf. [4] Uri Alon, Roy Sadaka, Omer Levy, and Eran Yahav. Structural language models of code, 2020. URL https: //arxiv.org/abs/1910.00577. [5] Seltenreich Andreas, Tang Bo, and Mullende",
+      "section": "C Validity",
+      "page": 7,
+      "char_offset": 33669,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M1"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:55:50Z",
+    "is_model_written": true,
+    "summary": "NNTailor drives DBMS fuzzing with a neural network: a model trained on SQL learns how to mutate queries so they stay syntactically and semantically valid, instead of relying on hand-written templates. On the latest PostgreSQL and SQLite it reports 35% higher line coverage than SQLsmith and 10% higher than SQLancer, along with better validity of the generated queries.",
+    "narrative": "SQLancer is one of NNTailor's two baselines, compared on code coverage and on the syntactic and semantic correctness of generated queries. The paper attributes SQLancer's coverage ceiling to template reliance -- predefined rules achieve considerable coverage but bound how much of the code space can be explored -- while acknowledging the other side of that trade: because SQLancer's oracles require expressions built to predefined semantic rules, it achieves high semantic correctness on SQLite. The comparison is reported as mixed rather than one-sided, with the paper noting that the two tools have their own advantages on different systems.",
+    "roles": {
+      "M1": "baseline",
+      "M2": "result_comparison",
+      "M3": "result_comparison",
+      "M4": "result_comparison",
+      "M5": "background",
+      "M6": "baseline",
+      "M7": "definition",
+      "M8": "incidental",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "result_comparison",
+      "M12": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is run as a baseline; NNTailor's generation is driven by a neural network model of its own."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer oracle is extended. NNTailor changes how queries are produced, not what makes a result wrong."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M1",
+          "M2",
+          "M6"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "We conducted evaluations of NNTailor on the latest versions of PostgreSQL and SQLite, comparing its performance with SQLsmith[5] and SQLancer[6, 25, 26, 27].",
+            "section": "I INTRODUCTION",
+            "page": 2
+          },
+          {
+            "mention_id": "M2",
+            "sentence": "With the neural network model-driven mutation approach, NNTailor achieves 35% and 10% higher code line coverage compared to SQLsmith and SQLancer.",
+            "section": "I INTRODUCTION",
+            "page": 2
+          },
+          {
+            "mention_id": "M6",
+            "sentence": "EVALUATION This section compares NNTailor with typical DBMS blackbox fuzzing tools SQLsmith and SQLancer in terms of coverage on the target DBMS, syntactic and semantic correctness of synthetic queries.",
+            "section": "G Test",
+            "page": 6
+          }
+        ],
+        "reasoning": "M1 states the evaluation compares NNTailor's performance with SQLsmith and SQLancer, M6 that the comparison covers coverage and query correctness, and M2 gives the coverage result."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is called an open-source black box fuzzing tool; the paper makes no state-of-the-art claim about it."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_iemcon67450_2025_11381190.json b/_data/papers/paper_doi_10_1109_iemcon67450_2025_11381190.json
new file mode 100644
index 0000000..56911a8
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_iemcon67450_2025_11381190.json
@@ -0,0 +1,433 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:24:14Z",
+  "paper": {
+    "id": "paper:doi:10.1109/iemcon67450.2025.11381190",
+    "title": "A Comparative Survey of Mutation Testing Across Large Language Models, REST APIs, and Database Engines",
+    "authors": [
+      "B. Falah",
+      "Hayat Routaib"
+    ],
+    "year": 2025,
+    "venue": "IEEE Annual Information Technology, Electronics and Mobile Communication Conference",
+    "doi": "10.1109/iemcon67450.2025.11381190",
+    "arxiv_id": null,
+    "s2_paper_id": "d02b3d0750a816d0b2044ff55b82b10e0b80046f",
+    "url": "https://doi.org/10.1109/iemcon67450.2025.11381190",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/iemcon67450.2025.11381190",
+      "retrieved_at": "2026-09-08T15:24:14Z",
+      "chars": 27712,
+      "content_sha256": "sha256:be798e0b76a420d3fac16a921363eb117185b41e8da92bbd29acc2a17e86425c"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 6,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1190
+      },
+      {
+        "number": "II",
+        "title": "RELATED WORK",
+        "start": 3330
+      },
+      {
+        "number": "B",
+        "title": "REST APIs",
+        "start": 4297
+      },
+      {
+        "number": "C",
+        "title": "Database Engines",
+        "start": 5535
+      },
+      {
+        "number": "III",
+        "title": "SURVEY",
+        "start": 6521
+      },
+      {
+        "number": "A",
+        "title": "Article Selection Criteria",
+        "start": 6896
+      },
+      {
+        "number": "B",
+        "title": "Data Extraction and Coding",
+        "start": 9330
+      },
+      {
+        "number": "C",
+        "title": "Comparative Analysis Procedure",
+        "start": 10033
+      },
+      {
+        "number": "D",
+        "title": "Validation and Limitations",
+        "start": 10735
+      },
+      {
+        "number": "IV",
+        "title": "COMPARATIVE",
+        "start": 11403
+      },
+      {
+        "number": "A",
+        "title": "Mutation Operators Used",
+        "start": 12047
+      },
+      {
+        "number": "B",
+        "title": "Tools and Frameworks",
+        "start": 14362
+      },
+      {
+        "number": "C",
+        "title": "Evaluation Metrics and Methods",
+        "start": 15723
+      },
+      {
+        "number": "D",
+        "title": "Fault Detection Effectiveness",
+        "start": 16314
+      },
+      {
+        "number": "E",
+        "title": "Key Challenges and Observed Trends",
+        "start": 17069
+      },
+      {
+        "number": "V",
+        "title": "DISCUSSION AND",
+        "start": 18633
+      },
+      {
+        "number": "A",
+        "title": "Cross-Domain Strengths and Observations",
+        "start": 19022
+      },
+      {
+        "number": "B",
+        "title": "System-Specific Challenges",
+        "start": 20067
+      },
+      {
+        "number": "C",
+        "title": "Methodological and Practical Implications",
+        "start": 21017
+      },
+      {
+        "number": "D",
+        "title": "Future Directions",
+        "start": 21885
+      },
+      {
+        "number": "VI",
+        "title": "RESEARCH",
+        "start": 22736
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION AND",
+        "start": 23861
+      },
+      {
+        "number": "R",
+        "title": "A. DeMillo, R. J. Lipton, and F. G. Sayward, “Hints on test data",
+        "start": 24611
+      },
+      {
+        "number": "H",
+        "title": "Xu et al., “Directed greybox fuzzing via large language models,”",
+        "start": 25039
+      },
+      {
+        "number": "F",
+        "title": "Tip et al., “LLMorpheus: Mutation testing using large language",
+        "start": 25376
+      },
+      {
+        "number": "J",
+        "title": "Zhang et al., “Mutation testing of Simulink models using LLMs,”",
+        "start": 25478
+      },
+      {
+        "number": "L",
+        "title": "Martinez et al., “You can REST now: Automated specification",
+        "start": 26303
+      },
+      {
+        "number": "S",
+        "title": "Salva et al., “Security testing of RESTful APIs with test case",
+        "start": 26430
+      },
+      {
+        "number": "D",
+        "title": "Corradini, A. Zampieri, M. Pasqua and M. Ceccato, ”Empirical",
+        "start": 26535
+      },
+      {
+        "number": "J",
+        "title": "Ba and M. Rigger, “Testing database engines via query plan guidance,”",
+        "start": 26781
+      },
+      {
+        "number": "J",
+        "title": "Schafer et al., “Parser knows best: Testing DBMSs with grammar-rule",
+        "start": 26964
+      },
+      {
+        "number": "A",
+        "title": "Derezi",
+        "start": 27075
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "R. A. DeMillo, R. J. Lipton, and F. G. Sayward, “Hints on test data selection: Help for the practicing programmer,” Computer, vol. 11, no. 4, pp. 34–41, Apr. 1978.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "M.S. Bouafif et al., “PRIMG: Efficient LLM-driven test generation using mutant prioritization,” 2025, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "L. Broide et al., “EvoGPT: Enhancing test suite robustness via LLMbased generation and genetic optimization,” 2025, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "H. Xu et al., “Directed greybox fuzzing via large language models,” 2025, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "M. Srinivasan et al., “GenFair: Systematic test generation for fairness fault detection in LLMs,” 2025, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Z. Tian et al., “Large Language Models for Equivalent Mutant Detection: How Far Are We?,” 2024, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "F. Tip et al., “LLMorpheus: Mutation testing using large language models,” 2024, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "J. Zhang et al., “Mutation testing of Simulink models using LLMs,” 2024, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "A. Arcuri et al., “Adaptive hypermutation for search-based system test generation: A study on REST APIs with EvoMaster,” 2020, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "V. Atlidakis et al., “Pythia: Grammar-Based Fuzzing of REST APIs with Coverage-guided Feedback and Learning-based Mutations,” 2020, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "D. Corradini, Z. Montolli, M. Pasqua, and M. Ceccato, “DeepREST: Automated test case generation for REST APIs exploiting deep reinforcement learning,” in Proc. 39th IEEE/ACM Int. Conf. Automated Softw. Eng. (ASE ’24), New York, NY, USA: ACM, 2024, pp. 1383–1394. doi: 10.1145/3691620.3695511.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "C. Cao et al., “Automated test-case generation for REST APIs using model inference search heuristic,” 2024, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "L. Martinez et al., “You can REST now: Automated specification inference with LLMs for REST APIs,” 2023, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "S. Salva et al., “Security testing of RESTful APIs with test case mutation,” 2024, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "D. Corradini, A. Zampieri, M. Pasqua and M. Ceccato, ”Empirical Comparison of Black-box Test Case Generation Tools for RESTful APIs,” in Code Analysis and Manipulation (SCAM), Luxembourg, 2021, pp. 226236, doi: 10.1109/SCAM52516.2021.00035.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” in Proc. 45th Int. Conf. Softw. Eng. (ICSE ’23), 2023, pp. 2060–2071. doi: 10.1109/ICSE48619.2023.00174.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 17,
+      "text": "J. Schafer et al., “Parser knows best: Testing DBMSs with grammar-rule traversal,” 2022, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "A. Derezi ´ nska, ”An experimental case study to applying mutation analysis for SQL queries,” 2009 International Multiconference on Computer Science and Information Technology, Mragowo, Poland, 2009, pp. 559566, doi: 10.1109/IMCSIT.2009.5352785.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "B. Chandra al., “Data generation for testing and grading SQL queries,” 2021, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "X. Zhou et al., “MutaGen: Automatic test generation for mutation testing on database applications,” 2023, arXiv: preprint.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "X. Gao et al., “A comprehensive survey on database management system fuzzing: Techniques, taxonomy, and experimental comparison,” 2022, arXiv: preprint. 0243",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 16,
+      "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” in Proc. 45th Int. Conf. Softw. Eng. (ICSE ’23), 2023, pp. 2060–2071. doi: 10.1109/ICSE48619.2023.00174.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "prints the DOI of the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[16]",
+      "technique": "qpg",
+      "sentence": "[16] proposed query plan–guided mutation strategies, while Schafer et al.",
+      "context_before": "security testing was explored by Salva et al. [14], who focused on vulnerability detection via test case mutation. Empirical comparisons by Corradini et al. [15] evaluated the effectiveness of various mutation tools across different API types, establishing REST APIs as a mature application area for mutation testing. C. Database Engines Database systems have long attracted attention for mutation testing due to their reliance on structured queries and transactional logic. Tools such as GDsmith and Squirrel apply coverage-guided mutation to graph and relational databases, respectively. Ba et al.",
+      "context_after": "developed Parser-KnowsBest [17], using grammar-rule traversal for test input mutation. Academic interest has also centered on SQL-level mutation. Derezi ´ nska et al. conducted mutation analysis for SQL queries [18], while Chandra al. focused on automated test generation and mutation-based grading [19]. The MutaGen tool by Zhou et al. [20] introduced scalable database application testing through structural and semantic mutation. A broader perspective was provided by Gao et al. [21], who offered a comprehensive survey and taxonomy of DBMS fuzzing and mutation strategies, highlighting trends an",
+      "section": "C Database Engines",
+      "page": 2,
+      "char_offset": 5816,
+      "cited_reference": {
+        "number": 16,
+        "text": "J. Ba and M. Rigger, “Testing database engines via query plan guidance,” in Proc. 45th Int. Conf. Softw. Eng. (ICSE ’23), 2023, pp. 2060–2071. doi: 10.1109/ICSE48619.2023.00174.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "prints the DOI of the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLancer, GDsmith, Squirrel) Evaluation Metrics Mutation score, test quality, fairness faults, equivalence detection Bug count, coverage (line/branch/path), spec adherence Bug count, query plan coverage, semantic correctness, performance diff Key Strengths Intelligent test generation, semantic analysis, fairness evaluation Schema-aware robustness testing, hybrid whitebox LLM integration Deep query exploration, performance analysis, grammar coverage Research Gaps Lack of reproducible tools, weak oracle validation Limited semantics, non-standardized mutation operators Underexplored in NoSQL, limited performanceguided mutation tools Fig.",
+      "context_before": "Data from the review is summarized in Table 1, highlighting mutation operators, tool maturity, evaluation metrics, and research gaps for each domain. TABLE ISUMMARY COMPARISON OF MUTATION TESTING DIMENSIONS ACROSS SYSTEM TYPES Dimension LLMREST API Database Engine Mutation Operators Prompt-based, semantic, fairness-aware, equivalencefocused Request/schema mutation, adaptive/generative, whitebox-guided SQL syntax/semantic, grammar-based, query plan/state mutation Tool Maturity Mostly prototypes (e.g., PRIMG, GenFair, LLMorpheus) Mature tools (e.g., RESTler, EvoMaster, Pythia) Public tools (e.g.",
+      "context_after": "1. Tool Maturity across System Type Figure 1 illustrates the relative maturity of mutation testing tools across system types, showing that LLM tools are mostly prototypes, whereas REST API and Database tools are largely production-ready. As shown in Figure 1, REST APIs and databases focus more on structural input mutations, while LLMs emphasize semantic sensitivity. Whereas figure 2 highlights the diversity of mutation operators, demonstrating that LLMs employ a broader range—including semantic, prompt-based, fairness-aware, and equivalence-focused mutations—compared to REST APIs and Database",
+      "section": "A Mutation Operators Used",
+      "page": 3,
+      "char_offset": 13116,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "TABLE II EXAMPLES OF TOOLS AND UNIQUE CONTRIBUTIONS System Type Representative Tools Notable Contributions LLMPRIMG, GenFair, LLMorpheus, EvoGPT Mutant prioritization, fairness detection, iterative reasoning REST API RESTler, EvoMaster, Dredd, CTGAN Schema-compliant mutation, GAN-based fuzzing, whitebox-LLM fusion Database SQLancer, GDsmith, Squirrel, AMOEBA Grammar-guided mutation, semantic fuzzing, performance regression bugs Figure 3 visualizes the number of representative mutation testing tools identified for each system type, highlighting both quantity and distribution.",
+      "context_before": "system type: • LLMs employ tools like PRIMG, EvoGPT, and LLMorpheus to generate mutants and evaluate robustness to adversarial or fairness-driven inputs. • REST APIs use EvoMaster, DeepREST, and Pythia, combining model inference, fuzzing, and grammar-based mutation. • Database Engines rely on GDsmith, Squirrel, and MutaGen for SQL mutation, bug detection, and coverage analysis. 0240 Fig. 2. Diversity of Mutation Operators across System Types Table 2 lists representative mutation testing tools for each system type and summarizes their unique contributions to test adequacy and fault detection.",
+      "context_after": "Fig. 3. Number of Mutation Testing Tools across System Types C. Evaluation Metrics and Methods Evaluation metrics reflect the characteristics and oracle complexity of each domain: • LLMs often use metrics such as model output variance, fairness detection rates, and ensemble disagreement. • REST APIs measure mutation score, fault coverage, and input schema violations. • Database Engines apply mutation score, execution path coverage, and query fault detection. As illustrated in Figure 3, mutation score remains the most commonly used metric across all domains. However, LLMs typically lack formal",
+      "section": "B Tools and Frameworks",
+      "page": 4,
+      "char_offset": 15079,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:19:46Z",
+    "is_model_written": true,
+    "summary": "A survey comparing how mutation testing is applied across three domains -- large language models, REST APIs and database engines -- based on 30 peer-reviewed studies. It compares mutation operator design, tool support and evaluation methods in each, reporting that LLM work is early-stage, REST API work benefits from adaptive schema-based tools, and database engine work uses grammar- and query-plan-guided mutation.",
+    "narrative": "SQLancer appears as one of the representative database engine tools surveyed, and query-plan-guided mutation -- SQLancer's QPG -- is named as a contribution of that domain.",
+    "roles": {
+      "M1": "definition",
+      "M2": "background",
+      "M3": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_issrew55968_2022_00056.json b/_data/papers/paper_doi_10_1109_issrew55968_2022_00056.json
new file mode 100644
index 0000000..e2d4b3d
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_issrew55968_2022_00056.json
@@ -0,0 +1,317 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:55:38Z",
+  "paper": {
+    "id": "paper:doi:10.1109/issrew55968.2022.00056",
+    "title": "A Disjoint-Partitioning Approach to Enhancing Metamorphic Testing of DBMS",
+    "authors": [
+      "M. Tang",
+      "T. Tse",
+      "Z. Zhou"
+    ],
+    "year": 2022,
+    "venue": "2022 IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW)",
+    "doi": "10.1109/issrew55968.2022.00056",
+    "arxiv_id": null,
+    "s2_paper_id": "de7f728893406aa870766bad7b626f6eb774abcc",
+    "url": "https://doi.org/10.1109/issrew55968.2022.00056",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/issrew55968.2022.00056",
+      "retrieved_at": "2026-09-08T14:55:38Z",
+      "chars": 21506,
+      "content_sha256": "sha256:995c5dad6cfba85e16345a0ced49d2a61cd8540610220e14367dd3cb30c6badf"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 2,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 5175
+      },
+      {
+        "number": "II",
+        "title": "MOTIVATION EXAMPLE IN DBMS TESTING",
+        "start": 6251
+      },
+      {
+        "number": "III",
+        "title": "METAMORPHIC TESTING",
+        "start": 8209
+      },
+      {
+        "number": "IV",
+        "title": "PREVIOUS WORK BY RIGGER ANDSU",
+        "start": 16069
+      },
+      {
+        "number": "V",
+        "title": "EXTENSION OFRIGGER ANDSU TODISJOINT PARTITIONING",
+        "start": 17388
+      },
+      {
+        "number": "VI",
+        "title": "EMPIRICAL CASESTUDY",
+        "start": 18825
+      },
+      {
+        "number": "3.1.0",
+        "title": "Our tool automatically constructs random source queries,",
+        "start": 18960
+      },
+      {
+        "number": "E",
+        "title": "T. Barr, M. Harman, P. McMinn, M. Shahbaz, and S. Yoo, “‘The oracle",
+        "start": 20119
+      },
+      {
+        "number": "T",
+        "title": "Y. Chen, F.-C. Kuo, H. Liu, P.-L. Poon, D. Towey, T. H. Tse, and Z.",
+        "start": 20308
+      },
+      {
+        "number": "S",
+        "title": "Segura, G. Fraser, A. B. Sanchez, and A. Ruiz-Cortes, “A survey on",
+        "start": 20512
+      },
+      {
+        "number": "M",
+        "title": "Rigger and Z. Su, “Detecting optimization bugs in database engines",
+        "start": 20682
+      }
+    ]
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "They proposed the query partitioning (QP) and ternary logic partitioning (TLP) techniques.",
+      "context_before": "rms—test oracle, metamorphic testing, DBMS, SQL, Ocean Base, failure I. INTRODUCTION Because of the popularity and financial impacts of online transaction processing (OLTP) systems, the correctness of the supporting database management systems (DBMS) is essential. However, owing to the scale of large databases, DBMS testing faces the test oracle problem, which refers to the difficulty in verifying the execution results [1]. The metamorphic testing (MT) approach [2][3] was invented in 1998 to alleviate the problem. In 2020, Rigger and Su [4][5][6] applied MT to tackle the issue in DBMS testing.",
+      "context_after": "Their empirical studies revealed failures in five DBMS. In this project, we conduct an in-depth investigation on Rigger and Su’s work. We find a gap between QP and TLP, and introduce the concept of disjoint partitioning to address the issue. We have undertaken an empirical case study on OceanBase [7], which was developed by Alibaba and Ant Group and 1s associated with the world’s fastest OLTP. We have identified various hidden failures and crashes. II. MOTIVATION EXAMPLE IN DBMS TESTING Consider Table 1, which records the id, name, and entry_ exit_count of 10 000 staff in a large corporation.",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 5706,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We find a gap between QP and TLP, and introduce the concept of disjoint partitioning to address the issue.",
+      "context_before": "anagement systems (DBMS) is essential. However, owing to the scale of large databases, DBMS testing faces the test oracle problem, which refers to the difficulty in verifying the execution results [1]. The metamorphic testing (MT) approach [2][3] was invented in 1998 to alleviate the problem. In 2020, Rigger and Su [4][5][6] applied MT to tackle the issue in DBMS testing. They proposed the query partitioning (QP) and ternary logic partitioning (TLP) techniques. Their empirical studies revealed failures in five DBMS. In this project, we conduct an in-depth investigation on Rigger and Su’s work.",
+      "context_after": "We have undertaken an empirical case study on OceanBase [7], which was developed by Alibaba and Ant Group and 1s associated with the world’s fastest OLTP. We have identified various hidden failures and crashes. II. MOTIVATION EXAMPLE IN DBMS TESTING Consider Table 1, which records the id, name, and entry_ exit_count of 10 000 staff in a large corporation. By examining * Corresponding author. Copyright and Reprint Permission: Abstracting is permitted with credit to the source. Libraries are permitted to photocopy beyond the limit of U.S. copyright law for private use of patrons those articles",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 5932,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "author_year_citation",
+      "surface": "RIGGER ANDSU\nIn 2020",
+      "technique": null,
+      "sentence": "PREVIOUS WORK BY RIGGER ANDSU In 2020, Rigger and Su applied MT to alleviate the oracle problem in DBMS testing [4][5]|6].",
+      "context_before": "equent query constructed based on an MR. The relation between their outputs (resultant lists returned by the DBMS) can then be verified with respect to the specified MR. In the previous example of the entry_exit_count, we may apply an MR such that if we add another condition to our query, the DBMS should return fewer records. For instance, we may refine our query to ask for all the records WHERE entry_exit_count & 1 AND name <> 'Alice'. Upon execution, if the DBMS returns > 3000 results, it is a failure. In this way, failures may be revealed by MT without the need for an ideal test oracle. IV.",
+      "context_after": "They proposed a general concept called query partitioning (QP) [5] for revealing DBMS failures. The process 1s as follows: 1) Construct an initial query (serving as a source test case). 2) Construct multiple queries, called partitioning queries (serving as follow-up test cases), each producing partial results of the source query. 3) Combine all the partial results to give a resultant list. 4) Compare the resultant list with the original result of the source query. QP describes a general MR among DBMS queries. Outputs of the follow-up test cases are sublists of the output of the source test ca",
+      "section": "IV PREVIOUS WORK BY RIGGER ANDSU",
+      "page": 2,
+      "char_offset": 16072,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "They proposed a general concept called query partitioning (QP) [5] for revealing DBMS failures.",
+      "context_before": "be verified with respect to the specified MR. In the previous example of the entry_exit_count, we may apply an MR such that if we add another condition to our query, the DBMS should return fewer records. For instance, we may refine our query to ask for all the records WHERE entry_exit_count & 1 AND name <> 'Alice'. Upon execution, if the DBMS returns > 3000 results, it is a failure. In this way, failures may be revealed by MT without the need for an ideal test oracle. IV. PREVIOUS WORK BY RIGGER ANDSU In 2020, Rigger and Su applied MT to alleviate the oracle problem in DBMS testing [4][5]|6].",
+      "context_after": "The process 1s as follows: 1) Construct an initial query (serving as a source test case). 2) Construct multiple queries, called partitioning queries (serving as follow-up test cases), each producing partial results of the source query. 3) Combine all the partial results to give a resultant list. 4) Compare the resultant list with the original result of the source query. QP describes a general MR among DBMS queries. Outputs of the follow-up test cases are sublists of the output of the source test case. The MR specifies that the concatenation of the outputs of the follow-up test cases must be e",
+      "section": "IV PREVIOUS WORK BY RIGGER ANDSU",
+      "page": 2,
+      "char_offset": 16195,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "ternary logic partitioning",
+      "technique": "tlp",
+      "sentence": "Rigger and Su further proposed ternary logic partitioning (TLP) [5] as a specific case of QP.",
+      "context_before": "e). 2) Construct multiple queries, called partitioning queries (serving as follow-up test cases), each producing partial results of the source query. 3) Combine all the partial results to give a resultant list. 4) Compare the resultant list with the original result of the source query. QP describes a general MR among DBMS queries. Outputs of the follow-up test cases are sublists of the output of the source test case. The MR specifies that the concatenation of the outputs of the follow-up test cases must be equal to the output of the source test case. A violation of this MR indicates a failure.",
+      "context_after": "An original query is decomposed into three partitioning queries, which are three predicate variants derived from a randomly generated predicate. Following the MR stated in QP, by concatenating the three resultant lists, one should obtain the result of the original query. TLP has revealed failures in various DBMS. V. EXTENSION OFRIGGER ANDSU TODISJOINT PARTITIONING After thorough investigation, we find that Rigger and Su’s proposed QP as a strategic idea is too general for practical MR constructions, while TLP as a special implementation of the strategic idea is too specific. We propose an inn",
+      "section": "IV PREVIOUS WORK BY RIGGER ANDSU",
+      "page": 2,
+      "char_offset": 16978,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP has revealed failures in various DBMS.",
+      "context_before": "es are sublists of the output of the source test case. The MR specifies that the concatenation of the outputs of the follow-up test cases must be equal to the output of the source test case. A violation of this MR indicates a failure. Rigger and Su further proposed ternary logic partitioning (TLP) [5] as a specific case of QP. An original query is decomposed into three partitioning queries, which are three predicate variants derived from a randomly generated predicate. Following the MR stated in QP, by concatenating the three resultant lists, one should obtain the result of the original query.",
+      "context_after": "V. EXTENSION OFRIGGER ANDSU TODISJOINT PARTITIONING After thorough investigation, we find that Rigger and Su’s proposed QP as a strategic idea is too general for practical MR constructions, while TLP as a special implementation of the strategic idea is too specific. We propose an innovative technique, called disjoint partitioning, to fill the gap. The core idea is that resultant lists returned by DBMS can be divided into partitions that are exhaustive and mutually exclusive using, for instance, the SQL keywords LIMIT and OFFSET. Consider the motivating example again. We propose followup queri",
+      "section": "IV PREVIOUS WORK BY RIGGER ANDSU",
+      "page": 2,
+      "char_offset": 17344,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "EXTENSION OFRIGGER ANDSU TODISJOINT PARTITIONING After thorough investigation, we find that Rigger and Su’s proposed QP as a strategic idea is too general for practical MR constructions, while TLP as a special implementation of the strategic idea is too specific.",
+      "context_before": "st case. The MR specifies that the concatenation of the outputs of the follow-up test cases must be equal to the output of the source test case. A violation of this MR indicates a failure. Rigger and Su further proposed ternary logic partitioning (TLP) [5] as a specific case of QP. An original query is decomposed into three partitioning queries, which are three predicate variants derived from a randomly generated predicate. Following the MR stated in QP, by concatenating the three resultant lists, one should obtain the result of the original query. TLP has revealed failures in various DBMS. V.",
+      "context_after": "We propose an innovative technique, called disjoint partitioning, to fill the gap. The core idea is that resultant lists returned by DBMS can be divided into partitions that are exhaustive and mutually exclusive using, for instance, the SQL keywords LIMIT and OFFSET. Consider the motivating example again. We propose followup queries that divide the rows in the original table into disjoint partitions with the same WHERE condition. We write the two follow-up queries as follows: (Q1) SELECT * FROM (SELECT * FROM entry_exit LIMIT 5000 OFFSET 0) AS offset WHERE entry_exit_count & 1; 131 (Q2) SELEC",
+      "section": "V EXTENSION OFRIGGER ANDSU TODISJOINT PARTITIONING",
+      "page": 2,
+      "char_offset": 17390,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "We have identified a gap between query partitioning and ternary logic partitioning in Rigger and Su’s work.",
+      "context_before": "Edition version 3.1.0. Our tool automatically constructs random source queries, random number of partitions, and random follow-up queries. It generates source and follow-up test cases randomly. It also verifies the results according to the MRs. We have unveiled various hidden failures and crashes. Some of them have been fixed in a subsequent release version 3.1.1 by the OceanBase QA team of the Ant Group. Others are being investigated or scheduled for fixes in future releases. VIT. SUMMARY We have conducted an in-depth investigation into existing metamorphic testing techniques in DBMS testing.",
+      "context_after": "We address the issue using a new concept of disjoint partitioning. We have applied the approach to OceanBase. We have unveiled various hidden failures and crashes, even though OceanBase has been tested extensively and applied widely in the DBMS community. ACKNOWLEDGMENTS This project was supported in part by an internship of the first author at Alibaba and Ant Group, China. We would also like to thank their QA team for confirming our failure reports. REFERENCES E. T. Barr, M. Harman, P. McMinn, M. Shahbaz, and S. Yoo, “‘The oracle problem in software testing: A survey,” JEEE Transactions on S",
+      "section": "3.1.0 Our tool automatically constructs random source queries,",
+      "page": 2,
+      "char_offset": 19544,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:56:58Z",
+    "is_model_written": true,
+    "summary": "A short paper proposing disjoint partitioning as an extension of existing metamorphic testing for DBMSs, motivated by the oracle problem that makes execution results hard to verify against expected outcomes. The authors apply it in an empirical case study on OceanBase and report uncovering hidden failures and crashes in a system already extensively tested and widely used in industry.",
+    "narrative": "The paper is built entirely on SQLancer's work: it devotes a section to query partitioning and ternary logic partitioning, identifies a gap between the two, and introduces disjoint partitioning to close it. That gap-closing is the paper's whole contribution.",
+    "roles": {
+      "M1": "definition",
+      "M2": "extension",
+      "M3": "background",
+      "M4": "definition",
+      "M5": "definition",
+      "M6": "extension",
+      "M7": "background",
+      "M8": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "yes",
+        "mention_ids": [
+          "M1",
+          "M2"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "They proposed the query partitioning (QP) and ternary logic partitioning (TLP) techniques.",
+            "section": "I INTRODUCTION",
+            "page": 1
+          },
+          {
+            "mention_id": "M2",
+            "sentence": "We find a gap between QP and TLP, and introduce the concept of disjoint partitioning to address the issue.",
+            "section": "I INTRODUCTION",
+            "page": 1
+          }
+        ],
+        "reasoning": "M1 names query partitioning and ternary logic partitioning as the techniques the paper works from, and M2 states that the authors find a gap between them and introduce disjoint partitioning to address it. Closing that gap is the contribution, not a baseline.",
+        "techniques": [
+          "tlp"
+        ]
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_missf68264_2026_11521893.json b/_data/papers/paper_doi_10_1109_missf68264_2026_11521893.json
new file mode 100644
index 0000000..b5cf4fc
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_missf68264_2026_11521893.json
@@ -0,0 +1,259 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:20:14Z",
+  "paper": {
+    "id": "paper:doi:10.1109/missf68264.2026.11521893",
+    "title": "An Automated Risk Scoring Framework for SQL Execution Plan Analysis and Performance Regression Detection in Oracle Database Systems",
+    "authors": [
+      "Raghu Gollapudi"
+    ],
+    "year": 2026,
+    "venue": "2026 International Conference on Multidisciplinary Innovations For Smart & Sustainable Future (MISSF)",
+    "doi": "10.1109/missf68264.2026.11521893",
+    "arxiv_id": null,
+    "s2_paper_id": "cc4a8303c7da6ce70e1fb471c23442d89ada3b7e",
+    "url": "https://doi.org/10.1109/missf68264.2026.11521893",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/missf68264.2026.11521893",
+      "retrieved_at": "2026-09-08T17:20:14Z",
+      "chars": 33936,
+      "content_sha256": "sha256:f1f1b505cc46a6bd6da19d07ab5ffdb854e48e29cb35c3bb7dcd81e21abc4a96"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 6,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1965
+      },
+      {
+        "number": "II",
+        "title": "RELATED WORKS",
+        "start": 5761
+      },
+      {
+        "number": "III",
+        "title": "PROPOSED METHODOLOGY",
+        "start": 10088
+      },
+      {
+        "number": "IV",
+        "title": "RESULTS",
+        "start": 20980
+      },
+      {
+        "number": "A",
+        "title": "Experimental Study and Dataset",
+        "start": 21470
+      },
+      {
+        "number": "B",
+        "title": "Performance Metrics",
+        "start": 22375
+      },
+      {
+        "number": "V",
+        "title": "CONCLUSION",
+        "start": 27755
+      },
+      {
+        "number": "U",
+        "title": "Rustambekovich, \"Enhancing Renewable Energy Fore casting using",
+        "start": 31766
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "C. Y. Jeong, H. C. Shin, and M. Kim, ‘‘Sensor-data augmentation for human activity recognition with time-warping and da ta masking,’’ Multimedia Tools Appl., vol. 80, no. 14, pp. 20991– 21009, Jun. 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "D. Malhotra, P. L. Soni, and J. J. Jadav, \"Anomaly detection in timeseries data using long short-term memory (LSTM) rec urrent neural network,\" in Proc. IEEE Int. Conf. Comput. Intell. Data Sci. (ICCIDS), Jul. 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Lakshmi, T. R., P. Narahari Sastry, and T. V. Rajin ikanth. \"Hybrid approach for Telugu handwritten character recogniti on using k-NN and SVM classifiers.\" International Review on Computers and Software 10, no. 9 (2015): 923-929.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "V. Roy, S. Shukla, “Designing Efficient Blind Sourc e Separation Methods for EEG Motion Artifact Removal Based on St atistical Evaluation”, Wireless Pers Commun 108, pp. 1311–132 7 (2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Sai, D., & Mashetty, H. (2024). Enhancing Federated Learning Evaluation: Exploring Instance-Level Insights with SQUARES in Image Classification Models. J. Electrical Systems, 20(7s), 2516-2523.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "M. Muchu, \"Autonomous Cloud Remediation and Self-He aling Infrastructure Through Infrastructure as Code and A rtificial Intelligence Automation,\" Journal of International Crisis and Risk Communication Research, Jan. 2026, doi: 10.63278/ji crcr.vi.3564.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Z. Jiang, J. Bai, and Z. Su. “DynSQL: Stateful Fuzz ing for Database Management Systems with Complex and Valid SQL Query Generation”. (Aug. 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Thirumala, Vijaya Lakshmi, Venkata SatyaNarayana Ka ranam, Pratap Reddy Lankireddy, Aruna Kumari Kakumani, and Rakesh Kumar Yacharam. \"Haze-level prior approach to enhance obj ect visibility under atmosphericdegradation.\" Turkish Journal of E lectrical Engineering and Computer Sciences 29, no. 2 (2021): 994-1014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "D. S., Bandaru, V. K. R., Volikatla, H., Indugu, V. V. R., & Bandaru, S. R. (2025, March). Evaluating and Interpreting Cl assifiers for Breast Cancer Detection Using eXplainable AI. In Internati onal Conference on Computing and Machine Learning (pp. 337-348). Si ngapore: Springer Nature Singapore.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "K. K. Pasupuleti, J. Li, H. Su, and M. Ziauddin. “A utomatic SQL Error Mitigation in Oracle”. Proceedings of the VLDB Endo wment 16, 12 (2023), 3835–3847, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "R. V. Rayala, C. Reddy Borra, V. Vasudevan, M. G, S. Cheekati and J. U. Rustambekovich, \"Enhancing Renewable Energy Fore casting using Roosters Optimization Algorithm and Hybrid Deep Lea rning Models,\" 2025 International Conference on Innovatio ns in Intelligent Systems: Advancements in Computing, Communication, and Cybersecurity (ISAC3), Bhubaneswar, India, 2025, pp. 1-6, doi: 10.1109/ISAC364032.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "K. Bednar, S. Spiekermann, and M. Langheinrich, ‘‘E ngineering privacy by design: Are engineers ready to live up t o the challenge?’’ Inf. Soc., vol. 35, no. 3, pp. 122–142, May 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Gondi, D. S., Bandaru, V. K. R., Gondi, K., Chanda, D., Ponce, L. A. H., & Singh, M. (2025, November). Deep Neural Netwo rks for Pandemic Hindrance Detection and Screening Replicat ion. In 2025 Tenth International Conference on Science Technolog y Engineering and Mathematics (ICONSTEM) (pp. 1-7). IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su. “Finding Bugs in Database Syst ems via Query Partitioning”. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (2020).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "C. R. Borra, R. Vani Rayala, Z. Khan and S. Cheekat i, \"Optimized Intrusion Detection for IoT Networks Using Machine Learning and Feature Selection with RTGBO-ELM Integration,\" 2025 International Conference on Intelligent and Cloud Computing (ICoI CC), Bhubaneswar, India, 2025, pp. 1-6, doi: 10.1109/ICoICC64033.2025.11052172.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Komathi, Arunachalam & Kishore, Somala & Velmurugan, Athiyoor & Pavithra, Maddipetlolu & Selvaraj, Yoganand & Beg um, Akbar & Muthukumaran, Dhakshnamoorthy. (2024). Network load balancing and data categorization in cloud computing. Indones ian Journal of Electrical Engineering and Computer Science. 35. 19 42. 10.11591/ijeecs. v35. i3, pp1942-1951.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "J. An and S. Cho, “Variational autoencoder-based anomaly detection using reconstruction robability,\" Knowl. -Based Sys t., vol. 150, pp. 85– 94, Aug. 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "S. Cheekati, R. V. Rayala, M. G. K and C. R. Borra, \"A Scalable Framework for Attack Detection: SWIM Transformer wi th Feature Optimization and Class Balancing,\" 2025 3rd Interna tional Conference on Integrated Circuits and Communication Systems (ICICACS), Raichur, India, 2025, pp. 1-7, doi: 10.1109/ICICACS65178.2025.10968653.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su. “Finding Bugs in Database Syst ems via Query Partitioning”. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (2020).",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[14]",
+      "technique": "tlp",
+      "sentence": "These systems however tend to show raw metrics without me rging them into a risk model [14].",
+      "context_before": "tures which minimize some of the estimation erro rs, it can still offer unstable execution plan with varying wo rkload situations. Academic and commercial work, around database monit oring, has focused on workload profiling, wait event class ification and visualization of performance metrics [13]. Stat istic gathering of dynamic performance views, system cata logs, and historical repositories are the typical example s of monitoring frameworks. Dashboards and alerting mech anisms would demonstrate limits to usage in terms of CPU c onsumed, I/O waits, and memory usage as well as locking tren ds.",
+      "context_after": "Consequently, the admi nistrators of the database must manually extract multi-dimensi onal indicators to designate high-impact SQL statements. The analysis of the execution plan has been activel y researched as well, especially regarding the plan s tability and regression detection. The consideration of plan has h values and cost differences and identification of suboptim al join strategies have been mentioned as the techniques to identify inconsistencies with the baseline performance. Othe rs focus on plan baselining and predictable plan development to retain predictable performance [15]. Ho",
+      "section": "II RELATED WORKS",
+      "page": 2,
+      "char_offset": 8166,
+      "cited_reference": {
+        "number": 14,
+        "text": "M. Rigger and Z. Su. “Finding Bugs in Database Syst ems via Query Partitioning”. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (2020).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:23:30Z",
+    "is_model_written": true,
+    "summary": "This paper scores the risk of SQL execution plan changes in Oracle, combining plan-stability signals into a single measure so that a plan regression can be flagged before it degrades production performance, rather than diagnosed afterwards from raw metrics.",
+    "narrative": "The connection is a single background citation in related work, where existing systems are criticised for showing raw metrics without merging them into a risk model. TLP is among the works cited there, though the paper's own concern is monitoring plan stability rather than finding logic bugs. SQLancer is never named in prose; the mention is reachable only through the citation marker.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_punecon67554_2025_11378586.json b/_data/papers/paper_doi_10_1109_punecon67554_2025_11378586.json
new file mode 100644
index 0000000..f806a6e
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_punecon67554_2025_11378586.json
@@ -0,0 +1,194 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:31:37Z",
+  "paper": {
+    "id": "paper:doi:10.1109/punecon67554.2025.11378586",
+    "title": "AskDB: AI-Assisted Natural Language Interface for Database Querying and Visualization",
+    "authors": [
+      "S. Shilaskar",
+      "Prapti Dongaonkar",
+      "Pruthavik Gavali",
+      "Suhani Gunje"
+    ],
+    "year": 2025,
+    "venue": "2025 IEEE Pune Section International Conference (PuneCon)",
+    "doi": "10.1109/punecon67554.2025.11378586",
+    "arxiv_id": null,
+    "s2_paper_id": "5c51d8d0d0446261f16acc8503d8cf2c7db0b75e",
+    "url": "https://doi.org/10.1109/punecon67554.2025.11378586",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/punecon67554.2025.11378586",
+      "retrieved_at": "2026-09-08T14:31:37Z",
+      "chars": 21161,
+      "content_sha256": "sha256:12d9145816d7d6b85084a09be5086d9e50ed5da0faa850999e6dd6c682935322"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 5,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2051
+      },
+      {
+        "number": "II",
+        "title": "LITERATURE REVIEW",
+        "start": 3669
+      },
+      {
+        "number": "A",
+        "title": "Flowchart",
+        "start": 7135
+      },
+      {
+        "number": "B",
+        "title": "Algorithm",
+        "start": 10870
+      },
+      {
+        "number": "A",
+        "title": "Case Study 1: Sales Data Analysis",
+        "start": 13975
+      },
+      {
+        "number": "B",
+        "title": "Case Study 2: SQL Database Connectivity",
+        "start": 14929
+      },
+      {
+        "number": "C",
+        "title": "Case Study 3: Complex Query Generation",
+        "start": 15803
+      },
+      {
+        "number": "D",
+        "title": "Discussion and Original Contributions",
+        "start": 17026
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "A. Attawar, S. V ora, P. Narechania, V. Sawant and H. V ora, \"NLSQL: Generating and Executing SQL Queries via Natural Language Using /DUJH /DQJXDJH 0RGHOV 2023 International Conference on Advanced Computing Technologies and Applications (ICACTA) 0XPEDL ,QGLD 2023, pp. 1- GRL ,&$&7$",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": ",. &KDGKD $ *XSWD 6 6DUNDU 0 7RPHU DQG 7 5DWKHH \"Performance Evaluation of Open- 6RXUFH //0V IRU 7H[W -to-SQL Conversion in Healthcare Data,\" 2025 International Conference on Pervasive Computational Technologies (ICPCT), Greater Noida, India, 2025, pp. 606- GRL ,&3&7",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "A. Wong et al., \"Translating Natural Language Queries to SQL Using WKH 7 0RGHO 0RQWUHDO 4& &DQDGD SS-7, doi: 6\\V&RQ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "H. Li, J. Zhang, H. Liu, J. Fan, X. Zhang, J. Zhu, R. Wei, H. Pan, C. Li, and H. Chen, “CodeS: Towards Building Open-source Language 0RGHOV IRU 7H[W -to-SQL,” arXiv preprint arXiv:2402.16347, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Zhu, Xiaohu, Qian Li, Lizhen Cui, and Yongkang Liu. \"Large language model enhanced text-to-sql generation: A survey.\" arXiv preprint arXiv:2410.06011 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "9 &kPDUD 5 0HQGRQFD-Neto, A. Silva and L. Cordovil, \"A Large /DQJXDJH 0RGHO DSSURDFK WR 64/ -to-Text Generation,\" NV, USA, 2024, pp. 1- GRL ,&&(",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "P. Parikh et al., \"Auto-Query-A simple natural language to SQL query generator for an e-learning platform,\" 2022 IEEE Global Engineering Education Conference (EDUCON) 7XQLV 7XQLVLD SS GRL ('8&21",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "$ 'HVKSDQGH ' .RWKDUL $ 6DOYL 3 0DQH DQG 9 .ROKH 4XHU \\OL]HU An Interactive Platform for Database Design and Text to SQL Conversion,\" 2022 International Conference for Advancement in Technology (ICONAT), Goa, India, 2022, pp. 1-6, doi: ,&21$7 > @ - %D DQG 0 5LJJHU 7HVWLQJ 'DWDEDVH (QJLQHV YLD 4XHU\\ 3OD Q Guidance,\" Software Engineering (ICSE) 0HOERXUQH $XVWUDOLD SS GRL ,&6( [10] Z. Lan and S. Li, ",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "$ VWXG\\ E\\ 0 5LJJHU HWDO > @ SUHVHQWHG 4XHU\\ 3ODQ Guidance (QPG)—a novel automated testing technique that focuses on the diversity and novelty of the query plans for uncovering buried bugs in relational database systems.",
+      "context_before": "ned the results to a dashboard, it dramatically decreased the time needed for data retrieval and allowed educators to continually refine course content insights. Lastly, Deshpande et al. [8] aimed at providing a democratic access through NLP and visualization to the Database. The presented system converted natural language queries to their corresponding SQL, and provided visualizations to improve interpretability, coinciding with the contemporary low-code trend. This is a user-friendly design that allows anyone who doesn’t have technical knowledge to understand and manage relational databases.",
+      "context_after": "By intentionally mutating the states of database, the approach identified 53 bugs that have not yet been reported on serious '%06 V\\VWHPV LQFOXGLQJ 64/LWH 7L'% DQG &RFNURDFK'% This methodology also serves as a Text-to-SQL development.“ by making more trustworthy DB backends that Text-to-SQL engines rely on. One notable line of work is to perform phrase-level comprehension for better schema mapping. A newly introduced PS-SQL model by Z. Lan et al. [10] addresses the task of schema-linking by utilizing phrasebased question representations. This approach serves to LQFUHDVH WKHFDSDELOLW\\ RI 3/0V",
+      "section": "II LITERATURE REVIEW",
+      "page": 2,
+      "char_offset": 5779,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "AskDB is a natural-language interface to databases. It translates a question into SQL with a large language model, runs it against the connected database, and returns results as a table and a chart, adding schema-driven query correction and prompt-engineering optimisation.",
+    "narrative": "One citation in the literature review, describing Query Plan Guidance as an automated testing technique focused on the diversity of generated queries. The sentence is partly mangled by the PDF's text encoding.",
+    "roles": {
+      "M1": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_saner60148_2024_00096.json b/_data/papers/paper_doi_10_1109_saner60148_2024_00096.json
new file mode 100644
index 0000000..2b0fba6
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_saner60148_2024_00096.json
@@ -0,0 +1,406 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:15:31Z",
+  "paper": {
+    "id": "paper:doi:10.1109/saner60148.2024.00096",
+    "title": "Testing Constraint Checking Implementations via Principled Metamorphic Transformations",
+    "authors": [
+      "Mingchen Gao",
+      "Huiyan Wang",
+      "Chang Xu"
+    ],
+    "year": 2024,
+    "venue": null,
+    "doi": "10.1109/saner60148.2024.00096",
+    "arxiv_id": null,
+    "s2_paper_id": null,
+    "url": "https://doi.org/10.1109/saner60148.2024.00096",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/saner60148.2024.00096",
+      "retrieved_at": "2026-09-08T17:15:31Z",
+      "chars": 72352,
+      "content_sha256": "sha256:54edc386589499d993c79cb0aa3d840d16f6a5a787c28d1216f88cfaf919427e"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1945
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 8445
+      },
+      {
+        "number": "A",
+        "title": "A Typical App lication and Research Review",
+        "start": 8460
+      },
+      {
+        "number": "B",
+        "title": "Preliminary",
+        "start": 12205
+      },
+      {
+        "number": "C",
+        "title": "Constraint Checking",
+        "start": 15113
+      },
+      {
+        "number": "A",
+        "title": "When all necessary nodes are evaluated, their parent nodes",
+        "start": 16585
+      },
+      {
+        "number": "B",
+        "title": "Derived Principled Metamorphic Relations",
+        "start": 22339
+      },
+      {
+        "number": "C",
+        "title": "Source Input Synthesis and Follow-up Generation",
+        "start": 37625
+      },
+      {
+        "number": "A",
+        "title": "Research Questions",
+        "start": 40031
+      },
+      {
+        "number": "B",
+        "title": "Evaluation Design and Setup",
+        "start": 40308
+      },
+      {
+        "number": "C",
+        "title": "Result Analyses",
+        "start": 45024
+      },
+      {
+        "number": "D",
+        "title": "Threat Analyses and Discuss ion",
+        "start": 53637
+      },
+      {
+        "number": "J",
+        "title": "Lu, \"Hybrid cpu-gpu constraint checking: Towards efficient context",
+        "start": 66081
+      },
+      {
+        "number": "C",
+        "title": "Tmelli, \"CVC4.\" p. 171-177, 201 1.",
+        "start": 69669
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "M. Thbaishat and S. Madria, \"Sensor networks: an overview,\" IEEE potentials, vol. 22, no. 2, pp. 2()....23, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, \"A survey on sensor networks,\" IEEE Communications magazine, vol. 40, no. 8, pp. 102-1 14, 2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "C.-Y. Chong and S. P. Kumar, \"Sensor networks: evolution, opportu­ nities, and challenges,\" Proceedings of the IEEE, vol. 91, no. 8, pp. 1247-1256, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "K. Rose, S. Eldridge, and L. Chapin, 'The internet of things: An overview,\" The internet society (/SOC), vol. 80, pp. 1-50, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "S. Li, L. D. Xu, and S. Zhao, \"The internet of things: a survey,\" Information systems frontiers, vol. 17, pp. 243-259, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "K. N. Qureshi and A. H. Abdullah, \"A survey on intelligent transporta­ tion systems,\" Middle-East Journal of Scien tific Research, vol. 15, no. 5, pp. 629-642, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "H. Wang, C. Xu, B. Guo, X. Ma, and J. Lu, \"Generic adaptive scheduling for efficient context inconsistency detection,\" IEEE Transactions on Software Engineering (TSE), vol. 47, no. 03, pp. 464-497, mar 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "C. Xu, S.-C. Cheung, W.-K. Chan, and C. Ye, \"Partial constraint check­ ing for context consistency in pervasive computing ;• ACM Transactions on Software Engineering and Methodology (TOSEM), vol. 19, no. 3, pp. 1-6 1, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "L. Zhang, H. Wang, C. Xu, and P. Yu, \"INFUSE: Towards efficient con­ text consistency by incremental-concurrent check fusion,\" in Proceed­ ings of the 38th International Con ference on So ftware Maintenance and Evolution (ICSME 2022), Limassol, Cyprus, Oct. 2022, forthco ming.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "C. Chen, H. Wang, L. Zhang, C. Xu, and P. Yu, \"Mi nimizing link generation in constraint checking for context inconsistency detection ;• in Proceedings of the 38th IEEE International Sym posium on So ftware Relibability Engineering (ISSRE 2022), Charlotte, North Carolina, USA, Oct. 2022, pp. 13-24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Earl, T., Barr, Mark, Harman, Phil, McMinn, Muzarnmil, Shahbaz, and Shin, \"The oracle problem in software testing: A survey,\" IEEE Transactions on Software Engineering (TSE), vol. 41, no. 5, pp. 507525, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "L. Filipp oni, A. Vitaletti, G. Landi, V. Memeo, G. Laura, and P. Pucci, \"Smart city: An event driven architecture for monitoring public spaces with heterogeneous sensors,\" in 2010 Fourth International Conf erence on Sensor Technologies and Applications, 2010, pp. 281-286.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "T. Y. Chen, \"Metamorphic testing: A simple approach to alleviate the oracle problem,\" in Fifth IEEE International Sym posium on Service Oriented System Eng ineering, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Q. Z. Zhi, J. Z. Shu, M. Hagenbuchner, T. H. Tse, F. C. Kuo, and T. Y. Chen, ''Automated functional testing of online search services,\" Software Testing, Verification and Reliability (S1VR), vol. 22, no. 4, pp. 221-243, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "P. Yao, H. Huang, W. Tang, Q. Shi, and C. Zhang, \"Skeletal approxima­ tion enumeration for smt solver testing,\" in 29th ACM Joint European Software Engineering Con ference and Sym posium on the Foundations of Software Engineering (ESEC /FSE 2021), 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "V. Le, M. Mshari, and Z. Su, \"Compiler validation via equivalence mod­ ulo inputs,\" in 2014 Programming Language Design and Implementation (PWI 2014), 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Q. Tao, W. Wu, Z. Chen, and W. Shen, ''An automatic testing approach for compiler based on metamorphic testing technique,\" in 2010 17th Asia Pacific Software Engineering Con ference (APSEC 2017), 201 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "S. Chen, S. Jin, and X. Xie, \"Testing your question answering software via asking recursively,\" in 2021 36th IEEE/ACM In ternational Con fer­ ence on Automated Software Engineering (ASE 2021). IEEE, 2021, pp. 104-1 16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Q. Shen, J. Chen, J. M. Zhang, H. Wang, S. Liu, and M. Tian, \"Natural test generation for precise testing of question answering software,\" in 2022 37st JEEE/ ACM In ternational Con ference on Automated Software Engineering (ASE 2022). IEEE, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "A. Dwarakanath, M. Ahu ja, S. Sikand, R. M. Rao, R. Bose, N. Dubash, and S. Podder, \"Identifying implementation bugs in machine learning based image classifiers using metamorphic testing,\" in 27th ACMSIG­ SOFT International Sym posium on Software Testing and Analysis (ISST A 2018), 2018, pp. 118 -128.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "M. Zhang, Y. Zhang, L. Zhang, C. Liu, and S. Khurshid, \"Deeproad: Gan-based metamorphic testing and input validation framework for 894 autonomous driving systems,\" in JEEE/ ACM In ternational Con ference on Automated Software Engineering, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Y. Yang, Z. Li, H. Wang, C. Xu, and X. Ma, ''Towards effective meta­ morphic testing by algorithm stability for linear classification programs,\" Journal of Systems and Software (JSS), vol. 180, no. 5, p. 111012, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Y. P. C. S. Xie, Xiaoyuan, \"Boosting the revealing of detected violations in deep learning testing: A diversity-guided method,\" in 2022 37st IEEE/ ACM International Conf erence on Automated Software Engineer­ ing (ASE 2022). IEEE, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "T. Chen, D. Huang, T. Tse, and Z. Zhou, \"Case studies on the selection of useful relations in metamorphic testing,\" in 2004 Ibero-American Symposium on Software Engineering and Knowledge Eng ineering, 2004, pp. 569-5 83.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "T. Chen, F. Kuo, Z. Zhou, and T. Tse, \"Metamorphic testing and beyond;' in 2003 International Workshop on Software Technology and Engineering Prac tice, 2004, pp. 19- 21.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Artifact and docs for principled metamorphic relatio ns. [Online]. Available: http s://zenodo. org/re cords/ 10052934",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "C. Xu, S.-C. Cheung, W.-K. Chan, and C. Ye, \"Heuristics-based strategies for resolving context inconsistencies in pervasive computing applications,\" in 2008 The 28th International Conf erence on Distributed Computing Systems (ICDCS 2008). IEEE, 2008, pp. 713-721.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "L. Capra, W. Emmerich, and C. Mascolo, \"CARISMA: Context-aware reflective middleware system for mobile applications,\" IEEE Transac­ tions on software engineering (TSE), vol. 29, no. 10, pp. 929-945, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "K. T. Patil, V. Bansal, V. Dhateria, and S. K. Narayankhedkar, \"Probable causes of RFID tag read unreliability in supermarkets and proposed solutions,\" in 2015 Inte rnational Con ference on Information Processing (ICI P). IEEE, 2015, pp. 392-397.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "S. R. Jeffery, M. Garofal akis, and M. J. Franklin, \"Adaptive cleaning for RFID data stream s;• in Proceedings of the 32nd international conference on Very large data bases (VWB 2006), vol. 6. Citeseer, 2006, pp. 163174.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "CMID. [Onli ne]. Available: https://git hub.com/growinware-cn/CMID",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "INFUSE. [Onlin e]. Available: http s://github.com/yuzi-zly/INFUSE",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "C. Nentwich, L. Capra, W. Emmerich, and A. Finkelste iin, \"xlinkit: A consistency checking and smart link generation service,\" ACM Transac­ tions on Inte rnet Technology (TOIT), vol. 2, no. 2, pp. 151- 185, 2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "C. Xu, S.-C. Cheung, and W.-K. Chan, \"Incremental consistency checking for pervasive context;' in Proceedings of the 28th international conference on So ftware engineering (ICSE 2006), 2006, pp. 292 -301.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "C. Xu, Y. P. Liu, S.-C. Cheung, C. Cao, and J. Lu, ''Towards context con­ sistency by concurrent checking for intemetware applications,\" Science China Information Sciences (SCIS), vol. 56, no. 8, pp. 1-20, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "C. Xu, S.-C. Cheung, and W.-K. Chan, \"Goal-directed context valida­ tion for adaptive ubiquitous systems,\" in ICSE Workshop on So ftware Engineering for Adaptive and Self-Managing Systems. Minneapolis, Minnesota, USA: IEEE Computer Society, 2007, pp. 1-10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "C. Xu, S.-C. Cheung, C. Lo, K.-C. Leung, and J. Wei, \"Cabot: On the ontology for the middleware support of context-aware pervasive applications,\" in IFIP Inte rnational Con ference on Network and Parallel Computing. Springer, 2004, pp. 568-57 5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "J. Sui, C. Xu, S.-C. Cheung, W. Xi, Y. Jiang, C. Cao, X. Ma, and J. Lu, \"Hybrid cpu-gpu constraint checking: Towards efficient context consis tency,\" Information and Software Technology (IST), vol. 74, pp. 23()....242, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "MG. [Online]. Available: https://github.com/c ychen2021/mg",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "H. Coles, T. Laurent, C. Henard, M. Papadakis, and A. Ventresque, \"PIT: a practical mutation testing tool for java (demo );' Proceedings of the 25th International Sym posium on Software Testing and Analysis, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "M. Papadakis, D. Shin, S. Yoo, and D.-H. Bae, ''Are mutation scores correlated with real fault detection? a large scale empirical study on the relationship between mutants and real faults;• in Proceedings of the 40th International Con ference on Software Engineering (ICSE 2018). New York, NY, USA: Association for Computing Machinery, 2018, p. 537-548. [ 42] T. T. Chekarn, M. Papa dakis, Y. L. Trao",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 41,
+      "text": "M. Papadakis, D. Shin, S. Yoo, and D.-H. Bae, ''Are mutation scores correlated with real fault detection? a large scale empirical study on the relationship between mutants and real faults;• in Proceedings of the 40th International Con ference on Software Engineering (ICSE 2018). New York, NY, USA: Association for Computing Machinery, 2018, p. 537-548. [ 42] T. T. Chekarn, M. Papa dakis, Y. L. Traon, and M. Harman, \"An empirical study on mutation, statement and branch coverage fault revelation that avoids the unreliable clean program assumption,\" in Proceedings of the 39th International Con ference on Software Engineering (ICSE 2017). IEEE Press, 2017, p. 597-608. 895 [43] C. Nentwich, W. Emmerich, A. Finkelsteiin, and E. Ellmer, \"Flexible consistency checking,\" ACM Transactions on Software Engineering and Metlwdology (TOSEM), vol. 12, no. I, pp. 28-63, 2003. [44] S. P. Reiss, \"Incremental maintenance of software artifacts,\" IEEE Transactions on Software Engineering (TSE), vol. 32, no. 9, pp. 682 697, 2006. [45] H. A. Handley, W. Khallouli, J. Huang, W. Edmonson, and N. Kibret, \"Maintaining the consistency of sysml model exports to xml metadata interchange (xmi);' in 2021 IEEE Inte rnational Systems Con ference (Sys Con). IEEE, 2021, pp. 1-8. [46] A. Egyed, \"Instant consistency checking for the UML,\" in Proceedings of the 28th International Con ference on Software Engineering (ICSE 2006), 2006, pp. 381-390. [47] R. S. Bash ir, S. P. Lee, S. U. R. Khan, V. Chang, and S. Farid, \"Uml models consistency management: Guidelines for software quality manager,\" International Journal of Information Manag ement, vol. 36, no. 6, pp. 883 -899, 2016. [48] M. Nabil, A. Chaoui, and M. Bettaz, \"An approach to uml consistency checking based on compositional semantic s;' International Journal of Embedded and Real-Time Communication Systems (IJERTCS), vol. 8, no. 2, pp. 1-23, 2017. [49] B. Wei and J. Sun, ''Leveraging sparql queries for um1 consistency checking,\" International Journal of Software Engineering and Knowl­ edge Engineering, vol. 31, no. 04, pp. 635-654, 2021. [50] A. Demuth, M. Riedl-Ebrenleitner, and A. Egyed, \"Efficient detection of inconsistencies in a multi-developer engineering environment,\" in 2016 31st IEEE/ ACM In ternational Conf erence on Automated Software Engineering (ASE 2016). IEEE, 2016, pp. 590-60 1. [51] P.-H. Kuo, S.-T. Lin, J. Hu, and C.-J. Huang, \"Multi -sensor context­ aware based chatbot model: An application of humanoid compauion robot,\" Sensors, vol. 21, no. 15, p. 5132, 2021. [52] (2020) Califrouia DMV. 2020. autonomous vehicle disengagement reports. https://www.dmv.ca. gov/p ortal/file/2 020autonom ous-vehicle-disenga gement-repor ts-csv/. [Onli ne]. Avail­ able: https://www.dmv. ca.gov/portal /file/2020-autonomous-vehicle­ disengagement-reports -csv/ [53] The numbers dont lie: Self -driving cars are getting good. [Online]. Available: https://www. wired.com/20 17 /02/califoruia-dmv-autonomous­ car -disengagement/ [54] L. D. Moura and N. Bjmer, \"Z3: An effcient smt solver.\" pp. 337 -340, 2008. [55] M. D. L. H. D. J. T. K. A. R. Clark Barrett, Christopher Conway and C. Tmelli, \"CVC4.\" p. 171-177, 201 1. [56] L. T. Ly, F. M. Maggi, M. Montali, S. Rinder le-Ma, and W. M. P. Van, der Aalst, \"Compliance mouitoring in business proc esses: Functionali­ ties, app lication, and toolsupport,\" Information Systems, vol. 54, no. 1, pp. 209 -234, 2015. [57] C. Mayr-Dom, M. Vierhauser, S. Bichler, F. Keplinger, J. Cleland­ Huang, A. Egyed, and T. Mehofer, \"Supporti ng Quality Assurance with Automated Proc ess-C entric Quality Constraints Checking,\" in 2021 IEEE/ACM 43rd International Con ference on Software Engineering (ICSE 2021). Madrid, ES: IEEE, May 2021, pp. 1298 -1310. [58] S. Segura, G. Fraser, A. B. Sanchez, and A. Ruiz-Cortes, ''A survey on metamorphic testing ;' IEEE Transactions on software engineering (TSE), vol. 42, no. 9, pp. 805-824, 2016. [59] T. Y. Chen, F.-C. Kuo, H. Liu, P.-L. Poon, D. Towey, T. Tse, and Z. Q. Zhou, \"Metamorphic testing: A review of challenges and opportunities,\" ACM Com puting Surveys (CSU R), vol. 51, no. 1, pp. 1-27, 2018. [60] C. Sun, V. Le, and Z. Su, \"Finding compiler bugs via live code mutation,\" in Proceedings of the 2016 ACMSIGPL AN international conf erence on object-oriented programming, systems, languages, and app lications, 2016, pp. 849- 863. [61] S. Chen, S. Jin, and X. Xie, \"Validation on machine reading compre­ hension software without annotated labels: A proper ty-based metho d;' in Proceedings of the 29th ACM Joint Meeting on European Software Eng ineering Conf erence and Sym posium on the Foundations of Software Engineering (ESEC/FSE 2021), 2021, pp. 590-602. [62] W. Chan, S. C. Cheung, and K. R. Leung, \"Towards a metamorphic testing methodology for service-oriented software applications,\" in Fifth International Con ference on Quality Software (QSIC 2005). IEEE, 2005, pp. 470-476. [63] M. Rigger and Z. Su, \"Finding hugs in database systems via query partitioning;' Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1-30, 2020. 895 [64] Y. Tian, S. Ma, M. Wen, Y. Liu, S.-C. Cheung, and X. Zhang, \"To what extent do dnn-based image classification models make unreliable infere nces?\" Empirical Software Engineering (EMSE), vol. 26, no. 5, p. 84, 2021. [65] C. Tsigkano s, P. Raui, S. Miiller, and T. Kehrer, \"Large language models: The next frontier for variable discovery within metamorphic testing?\" in 2023 IEEE International Con ference on Software Analysis, Evolution and Reengineering (SANER 2023). IEEE, 2023, pp. 678 -682. [66] D. Winterer, C. Zhang, and Z. Su, \"Val idating smt solvers via semantic fusion,\" in Proceedings of the 41st ACMSIGPLA N Con ference on programming language design and implementation (PWI 2020, 2020, pp. 718 -730.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[41]",
+      "technique": null,
+      "sentence": "We argue that mutants have been observed to provide good guidance and showed correlations with real bugs in existing researches [41], [42].",
+      "context_before": "uired in MR3-l. We localized an implemen tation bug which is caused due to an if-condition being mistakenly written when calculating link explanations for and nodes. Therefore, we can answer RQ2 as follows: Our designed MRs can reveal violations as expected and succe ssfully detect real-life bugs in practical CC implem entations. D. Threat Analyses and Discuss ion One potential threat concerns the external valid ity of our experiments and conclusions since we measure the effec­ tiveness by the killing rate of detecting bugs inserted by muta tion testing, thus probably different from real bugs.",
+      "context_after": "Still, to alleviate the possible bias, we also evaluated our MRs on real-life CC impleme ntations, which also showed positive results in dete cting real bugs. Another potential threat may concern our input synthesis for obtaining data under checking and constraints for checking. We synthe size well-formed FOL constraints and accompany each with data containing all associated patterns with sim­ ulated elements. Although the process can be simple, we simulate all possible evaluation situations by enumeratedly assigning different bfunc values during constraint checking, to make our experime nts",
+      "section": "D Threat Analyses and Discuss ion",
+      "page": 10,
+      "char_offset": 53905,
+      "cited_reference": {
+        "number": 41,
+        "text": "M. Papadakis, D. Shin, S. Yoo, and D.-H. Bae, ''Are mutation scores correlated with real fault detection? a large scale empirical study on the relationship between mutants and real faults;• in Proceedings of the 40th International Con ference on Software Engineering (ICSE 2018). New York, NY, USA: A",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:23:30Z",
+    "is_model_written": true,
+    "summary": "This work tests constraint checking implementations -- the components that verify consistency constraints over software artifacts in context-aware and pervasive computing systems -- using principled metamorphic transformations. Applying its metamorphic relations to nine variants of three implementations, it detected bugs in seven of the nine and found five unique implementation bugs that had gone undetected for months to years.",
+    "narrative": "No relationship is visible in the paper's own text. Its bibliography resolves to no SQLancer publication, and its subject is constraint checking over software models rather than database query evaluation; the only near-match in the text is the word 'trigger'. The record's presence among SQLancer-citing papers comes from the upstream citation graph rather than from anything found in the paper, and that should be re-examined.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": [
+      "Why this paper is in the citing set at all. Its 41 parsed references include no SQLancer publication, and its subject is constraint checking for pervasive computing. The citation graph that admitted it should be re-checked."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_sp54263_2024_00109.json b/_data/papers/paper_doi_10_1109_sp54263_2024_00109.json
new file mode 100644
index 0000000..27f3710
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_sp54263_2024_00109.json
@@ -0,0 +1,696 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:24:50Z",
+  "paper": {
+    "id": "paper:doi:10.1109/sp54263.2024.00109",
+    "title": "Chronos: Finding Timeout Bugs in Practical Distributed Systems by Deep-Priority Fuzzing with Transient Delay",
+    "authors": [
+      "Yuanliang Chen",
+      "Fuchen Ma",
+      "Yuanhang Zhou",
+      "Ming Gu",
+      "Qing Liao",
+      "Yu Jiang"
+    ],
+    "year": 2024,
+    "venue": "IEEE Symposium on Security and Privacy",
+    "doi": "10.1109/sp54263.2024.00109",
+    "arxiv_id": null,
+    "s2_paper_id": "7b3bc7fdb7d6a6234ed892b057f8feb6118de0b8",
+    "url": "https://doi.org/10.1109/sp54263.2024.00109",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/sp54263.2024.00109",
+      "retrieved_at": "2026-09-10T15:24:50Z",
+      "chars": 93049,
+      "content_sha256": "sha256:815d9a0f66d30d4d56edb934b87a8cab56c622f2666b07185ad06425b166c473"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 17,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1831
+      },
+      {
+        "number": "2",
+        "title": "Background of Timeout Mechanism",
+        "start": 9487
+      },
+      {
+        "number": "3",
+        "title": "Overview",
+        "start": 12264
+      },
+      {
+        "number": "3.1",
+        "title": "Definition of Timeout Bugs",
+        "start": 12276
+      },
+      {
+        "number": "3.2",
+        "title": "A Motivation Example",
+        "start": 14706
+      },
+      {
+        "number": "4",
+        "title": "Chronos Design",
+        "start": 19086
+      },
+      {
+        "number": "4.1",
+        "title": "Chronos Workflow",
+        "start": 19926
+      },
+      {
+        "number": "4.2",
+        "title": "Delay Injector",
+        "start": 21757
+      },
+      {
+        "number": "4.3",
+        "title": "Delay Selector",
+        "start": 28187
+      },
+      {
+        "number": "4.4",
+        "title": "Delay Executor",
+        "start": 34672
+      },
+      {
+        "number": "4.5",
+        "title": "Bug Analyzer",
+        "start": 38185
+      },
+      {
+        "number": "5",
+        "title": "Implementation",
+        "start": 40291
+      },
+      {
+        "number": "6",
+        "title": "Evaluation",
+        "start": 49143
+      },
+      {
+        "number": "6.1",
+        "title": "Timeout bugs in real-world distributed systems",
+        "start": 50363
+      },
+      {
+        "number": "6.1.1",
+        "title": "Case Study. Now we use two cases to illustrate",
+        "start": 54621
+      },
+      {
+        "number": "6.3",
+        "title": "Effectiveness of Transient Delay",
+        "start": 65305
+      },
+      {
+        "number": "6.4",
+        "title": "Accuracy of Chronos",
+        "start": 66613
+      },
+      {
+        "number": "7",
+        "title": "Discussion",
+        "start": 69895
+      },
+      {
+        "number": "8",
+        "title": "Related Work",
+        "start": 72458
+      },
+      {
+        "number": "9",
+        "title": "Conclusion",
+        "start": 76952
+      },
+      {
+        "number": "10",
+        "title": "ACKNOWLEDGEMENTS",
+        "start": 77830
+      },
+      {
+        "number": "L",
+        "title": "Zhang, and L. Zhou, “Modist: Transparent model checking of",
+        "start": 78874
+      },
+      {
+        "number": "D",
+        "title": "Simon, S. Priambada, C. Tian, F. Ye, T. Leesatapornwongsa",
+        "start": 79343
+      },
+      {
+        "number": "J",
+        "title": "Reynolds, and C. Rosenthal, “Chaos engineering,” IEEE Software,",
+        "start": 80153
+      },
+      {
+        "number": "H",
+        "title": "Li, “Phoenix: Detect and locate resilience issues in blockchain via",
+        "start": 89604
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "G. Le Lann, “Distributed systems-towards a formal approach.” in IFIP congress, vol. 7, 1977, pp. 155–160.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "R. C. Nunes and I. Jansch-P ˆorto, “Modeling communication delays in distributed systems using time series,” in 21st IEEE Symposium on Reliable Distributed Systems, 2002. Proceedings. IEEE, 2002, pp. 268–273.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "E. A. Brewer, “Towards robust distributed systems,” in PODC, vol. 7, no. 10.1145. Portland, OR, 2000, pp. 343 477–343 502.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "apache ZooKeeper, “Zookeeper connect timeout,” https://www.theverge.com/2020/12/10/22168439/ facebook-down-outage-instagram-messenger-whatsapp, 2023, accessed on March 31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "ZOOKEEPER-3189, “Zookeeper connect timeout,” https: //issues.apache.org/jira/browse/ZOOKEEPER-3189, 2023, accessed on March 31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "J. Yang, T. Chen, M. Wu, Z. Xu, X. Liu, H. Lin, M. Yang, F. Long, L. Zhang, and L. Zhou, “Modist: Transparent model checking of unmodified distributed systems,” in NSDI’09, 2009, pp. 213–228.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "T. Leesatapornwongsa, M. Hao, P. Joshi, J. F. Lukman, and H. S. Gunawi, “ {SAMC }:{Semantic-Aware }model checking for fast discovery of deep bugs in cloud systems,” in 11th USENIX Symposium on Operating Systems Design and Implementation (OSDI 14), 2014, pp. 399–414.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "J. F. Lukman, H. Ke, C. A. Stuardo, R. O. Suminto, D. H. Kurniawan, D. Simon, S. Priambada, C. Tian, F. Ye, T. Leesatapornwongsa et al., “Flymc: Highly scalable testing of complex interleavings in distributed systems,” in Proceedings of the Fourteenth EuroSys Conference 2019, 2019, pp. 1–16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "M. Musuvathi, D. Y. Park, A. Chou, D. R. Engler, and D. L. Dill, “Cmc: A pragmatic approach to model checking real code,” ACMSIGOPS Operating Systems Review, vol. 36, no. SI, pp. 75–88, 2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "H. Ziade, R. A. Ayoubi, R. Velazco et al., “A survey on fault injection techniques,” Int. Arab J. Inf. Technol., vol. 1, no. 2, pp. 171–186, 2004.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "M. Eslami, B. Ghavami, M. Raji, and A. Mahani, “A survey on fault injection methods of digital integrated circuits,” Integration, vol. 71, pp. 154–163, 2020. 1952",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "A. Basiri, N. Behnam, R. De Rooij, L. Hochstein, L. Kosewski, J. Reynolds, and C. Rosenthal, “Chaos engineering,” IEEE Software, vol. 33, no. 3, pp. 35–41, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Netflix, “Chaos monkey,” https://netflix.github.io/chaosmonkey/, 2022, accessed at December 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "A. Netflix, “Simian army,” https://github.com/Netflix/SimianArmy, 2022, accessed at December 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "chaosblade io, “Chaos blade,” https://netflix.github.io/chaosmonkey/, 2022, accessed at December 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "chaos mesh org, “Chaos mesh a powerful chaos engineering platform for kubernetes,” https://chaos-mesh.org/, 2022, accessed at December 29, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "J.-J. Bai, Y .-P. Wang, H.-Q. Liu, and S.-M. Hu, “Mining and checking paired functions in device drivers using characteristic fault injection,” Information and Software Technology, vol. 73, pp. 122–133, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Z.-M. Jiang, J.-J. Bai, K. Lu, and S.-M. Hu, “Fuzzing error handling code using context-sensitive software fault injection,” in the 29th USENIX Security Symposium (Security’20), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "R. Banabic and G. Candea, “Fast black-box testing of system recovery code,” in Proceedings of the 7th ACM european conference on Computer Systems, 2012, pp. 281–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "R. Natella, D. Cotroneo, J. A. Duraes, and H. S. Madeira, “On fault representativeness of software fault injection,” IEEE Transactions on Software Engineering, vol. 39, no. 1, pp. 80–96, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "K. Cong, L. Lei, Z. Yang, and F. Xie, “Automatic fault injection for driver robustness testing,” in Proceedings of the 2015 International Symposium on Software Testing and Analysis, 2015, pp. 361–372.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "H. S. Gunawi, T. Do, P. Joshi, P. Alvaro, J. M. Hellerstein, A. C. Arpaci-Dusseau, R. H. Arpaci-Dusseau, K. Sen, and D. Borthakur, “{FATE}and{DESTINI }: A framework for cloud recovery testing,” in8th USENIX Symposium on Networked Systems Design and Implementation (NSDI 11), 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "R. Kumar, P. Jovanovic, W. Burleson, and I. Polian, “Parametric trojans for fault-injection attacks on cryptographic hardware,” in 2014 Workshop on Fault Diagnosis and Tolerance in Cryptography. IEEE, 2014, pp. 18–28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "M. S. R. M. T. Millstein and M. Musuvathi, “Can you fool me? towards automatically checking protocol gullibility.”",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Y. Gao, W. Dou, D. Wang, W. Feng, J. Wei, H. Zhong, and T. Huang, “Coverage guided fault injection for cloud systems,” in Proceedings of IEEE/ACMSIGSOFT International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "F. Junqueira and B. Reed, ZooKeeper: distributed process coordination. ” O’Reilly Media, Inc.”, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "A. Davies and H. Fisk, MySQL clustering. Sams Publishing, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "D. Borthakur et al., “Hdfs architecture guide,” Hadoop apache project, vol. 53, no. 1-13, p. 2, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "go ethereum, “go-ethereum. official go implementation of the ethereum protocol,” https://geth.ethereum.org/, 2022, accessed at December 6, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "V. K. Garg, Elements of distributed computing. John Wiley & Sons, 2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "csis.pace.edu, “communication,” http://csis.pace.edu/ ∼marchese/ CS865/Lectures/Chap4/Chapter4.htm, 2023, accessed on March 31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "ida.liu.se, “Communication in distributed systems,” https://www.ida. liu.se/∼TDDD25/lectures/lect3.pdf, 2023, accessed on March 31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "webstor.srmist.edu.in, “Communication in distributed system,” https: //www.ida.liu.se/ ∼TDDD25/lectures/lect3.pdf, 2023, accessed on March 31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "inf.fu berlin.de, “Indirect communication,” http://www.inf. fu-berlin.de/inst/ag-se/teaching/V-NETZPR-2015/07 Indirect Communication%20-%20I.pdf, 2023, accessed on March 31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "M. Hedayati and S. Gravani, “Hodor: Intra-process isolation for highthroughput data plane libraries,” in Proceedings of the 2019 USENIX Annual Technical Conference, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "R. J. Connor, T. McDaniel, J. M. Smith, and M. Schuchard, “Pku pitfalls: Attacks on pku-based memory isolation systems,” in Proceedings of the 29th USENIX Conference on Security Symposium, 2020, pp. 1409–1426.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "B. H. Tay and A. L. Ananda, “A survey of remote procedure calls,” ACMSIGOPS Operating Systems Review, vol. 24, no. 3, pp. 68–79, 1990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "A. Venkataraman and K. K. Jagadeesha, “Evaluation of inter-process communication mechanisms,” Architecture, vol. 86, p. 64, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "S. Androutsellis-Theotokis and D. Spinellis, “A survey of peer-to-peer content distribution technologies,” ACM computing surveys (CSUR), vol. 36, no. 4, pp. 335–371, 2004.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "G. F. Coulouris, J. Dollimore, and T. Kindberg, Distributed systems: concepts and design. pearson education, 2005.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "J. D. Herbsleb and A. Mockus, “An empirical study of speed and communication in globally distributed software development,” IEEE Transactions on software engineering, vol. 29, no. 6, pp. 481–494, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "N. M. Josuttis, SOA in practice: the art of distributed system design. ” O’Reilly Media, Inc.”, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "L. Lamport, “Paxos made simple,” ACMSIGACT News (Distributed Computing Column) 32, 4 (Whole Number 121, December 2001), pp. 51–58, 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "ludun, “Datastreamer should reset thread interrupted state in createblockoutputstream,” https://issues.apache.org/jira/ browse/HDFS-15379?page=com.atlassian.jira.plugin.system. issuetabpanels%3Acomment-tabpanel&focusedCommentId= 17125997#comment-17125997, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "D. Borthakur, “The hadoop distributed file system: Architecture and design,” Hadoop Project Website, vol. 11, no. 2007, p. 21, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "A. MySQL, “Mysql,” 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Ethereum, “Welcome to ethereum,” https://ethereum.org/en/, 2022, accessed at December 23, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "L. WIKI, “netem,” https://wiki.linuxfoundation.org/networking/ netem, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "S. Koranne and S. Koranne, “Standard libraries,” Handbook of Open Source Tools, pp. 105–111, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "S. Koranne, “Boost c++ libraries,” Handbook of open source tools, pp. 127–143, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "american fuzzy lop, “American fuzzy lop user guide,” https://afl-1. readthedocs.io/en/latest/user guide.htmlL, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "G. Cloud, “Setting request timeout (services),” https://cloud.google. com/run/docs/configuring/request-timeout, 2023, accessed at March 30, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "J. Y. Monteith, J. D. McGregor, and J. E. Ingram, “Hadoop and its evolving ecosystem,” in 5th International Workshop on Software Ecosystems (IWSECO 2013), vol. 50. Citeseer, 2013, p. 74.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "A. Lentz and M. AB, “An introduction to mysql cluster architecture and use,” MySQL AB, November, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "CoinMarketCap, “Coinmarketcap,” https://coinmarketcap.com, 2023, accessed on March 31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Intel, “Hibench suite,” https://github.com/Intel-bigdata/HiBench, 2023, accessed at March 30, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis.” in OSDI, vol. 20, 2020, pp. 667–682. 1953",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 58,
+      "text": "Ethereum, https://github.com/drandreaskrueger/chainhammer, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "vdcresearch, “Languaged used in distributed system,” https://www.vdcresearch.com/results.html?addsearch=languaged+ used+ in+ distribuetd+ systems, 2023, accessed on March 31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "dzone.com, “Top language for distributed system,” https://dzone. com/articles/top-languages-for-distributed-systems, 2023, accessed on March 31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "jaxenter.com, “Best languages in distributed system,” https://jaxenter. com/programming-languages-best-distributed-systems-169281.html, 2023, accessed on March 31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "I. Docker, “Docker,” lınea].[Junio de 2017]. Disponible en: https://www. docker. com/what-docker, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "H. S. Gunawi, R. O. Suminto, R. Sears, C. Golliher, S. Sundararaman, X. Lin, T. Emami, W. Sheng, N. Bidokhti, C. McCaffrey et al., “Fail-slow at scale: Evidence of hardware performance faults in large production systems,” ACM Transactions on Storage (TOS), vol. 14, no. 3, pp. 1–26, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "B. Panda, D. Srinivasan, H. Ke, K. Gupta, V. Khot, and H. S. Gunawi, “Iaso: A fail-slow detection and mitigation framework for distributed storage services.” in USENIX Annual Technical Conference, 2019, pp. 47–62.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "R. Lu, E. Xu, Y. Zhang, F. Zhu, Z. Zhu, M. Wang, Z. Zhu, G. Xue, J. Shu, M. Li et al., “Perseus: A {Fail-Slow }detection framework for cloud storage systems,” in 21st USENIX Conference on File and Storage Technologies (FAST 23), 2023, pp. 49–64.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "J. M. V oas and G. McGraw, Software fault injection: inoculating programs against errors. John Wiley & Sons, Inc., 1997.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "M.-C. Hsueh, T. K. Tsai, and R. K. Iyer, “Fault injection techniques and tools,” Computer, vol. 30, no. 4, pp. 75–82, 1997.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Jepsen, “Distributed systems safety research,” http://jepsen.io/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "L. Zhang, B. Morin, P. Haller, B. Baudry, and M. Monperrus, “A chaos engineering system for live analysis and falsification of exception-handling in the jvm,” IEEE Transactions on Software Engineering, vol. 47, no. 11, pp. 2534–2548, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "K. Murdock, D. Oswald, F. D. Garcia, J. Van Bulck, D. Gruss, and F. Piessens, “Plundervolt: Software-based fault injection attacks against intel sgx,” in (SP). IEEE, 2020, pp. 1466–1482.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "J. Gravellier, J.-M. Dutertre, Y. Teglia, and P. L. Moundi, “Faultline: Software-based fault injection on memory transfers,” in (HOST). IEEE, 2021, pp. 46–55.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "J. Calhoun, L. Olson, and M. Snir, “Flipit: An llvm based fault injector for hpc,” 08 2014, pp. 547–558.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "J.-J. Bai, Y .-P. Wang, H.-Q. Liu, and S.-M. Hu, “Mining and checking paired functions in device drivers using characteristic fault injection,” Information and Software Technology, vol. 73, pp. 122–133, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "F. Ma, Y. Chen, Y. Zhou, J. Sun, Z. Su, Y. Jiang, J. Sun, and H. Li, “Phoenix: Detect and locate resilience issues in blockchain via context-sensitive chaos,” in Proceedings of the 2023 ACMSIGSAC Conference on Computer and Communications Security, 2023, pp. 1182–1196.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "P. D. Marinescu and G. Candea, “Efficient testing of recovery code using fault injection,” ACM Transactions on Computer Systems (TOCS), vol. 29, no. 4, pp. 1–38, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "P. Joshi, H. S. Gunawi, and K. Sen, “Prefail: A programmable tool for multiple-failure injection,” in Proceedings of the 2011 ACM international conference on Object oriented programming systems languages and applications, 2011, pp. 171–188.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Y. Chen, Y. Jiang, F. Ma, J. Liang, M. Wang, C. Zhou, X. Jiao, and Z. Su, “ {EnFuzz }: Ensemble fuzzing with seed synchronization among diverse fuzzers,” in 28th USENIX Security Symposium (USENIX Security 19), 2019, pp. 1967–1983.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Z. Luo, J. Yu, F. Zuo, J. Liu, Y. Jiang, T. Chen, A. Roychoudhury, and J. Sun, “Bleem: packet sequence oriented fuzzing for protocol implementations,” in 32nd USENIX Security Symposium (USENIX Security 23), 2023, pp. 4481–4498.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "H. Sun, Y. Shen, C. Wang, J. Liu, Y. Jiang, T. Chen, and A. Cui, “Healer: Relation learning guided kernel fuzzing,” in Proceedings of the ACMSIGOPS 28th Symposium on Operating Systems Principles, 2021, pp. 344–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Y. Chen, F. Ma, Y. Zhou, Y. Jiang, T. Chen, and J. Sun, “Tyr: Finding consensus failure bugs in blockchain system with behaviour divergent model,” in. IEEE, 2023, pp. 2517–2532.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "F. Ma, Y. Chen, M. Ren, Y. Zhou, Y. Jiang, T. Chen, H. Li, and J. Sun, “Loki: State-aware fuzzing framework for the implementation of blockchain consensus protocols,” in Proceedings 2023 Network and Distributed System Security Symposium, 2023. 1954 Appendix A. Meta-Review The following meta-review was prepared by the program committee for the Privacy (S&P) as part of the review process as detailed",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 57,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis.” in OSDI, vol. 20, 2020, pp. 667–682. 1953",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For MySQL-Cluster, the workload is generated by SQLancer [57], one of the widely used SQL generators for testing database systems.",
+      "context_before": "o interact with target systems, including sending workloads to DSUTs and monitoring bugs within them. The second part is the Delay Model which is implemented for generating high-quality transient delay sequences to explore deep logic in DSUTs based on their runtime traces. The third part is the Delay Library which contains runtime libraries instrumented with deeppriority transient delays. The rest of the section describes notable implementation details. Workload Loader: For Hadoop Distributed File Systems, the workload is collected from Intel HiBench [56], a widely used testing workload suite.",
+      "context_after": "For ZooKeeper, we developed an automatic tool to generate the workload based on its API documentation and instruction manual. The tool is also open-source and available in our git1. For Go-Etheruem, the workload is generated by chainhammer [58], which is an art-of-the-state tool for generating test transactions for blockchain systems. Trace Analyzer: In the deep-priority guided delay selector, before calculating the depth of each delay block di, Chronos first obtains the call trace and identifies the caller (which delay block). In Java language, the trace analyzer is implemented by ‘Thread.cu",
+      "section": "5 Implementation",
+      "page": 8,
+      "char_offset": 42776,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/SecTechTool/Chronos",
+    "markers": [
+      "sqlancer_source_content_match"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/SecTechTool/Chronos",
+        "source_type": "github_repository",
+        "excerpt": "# Chronos: An Automatical Testing Framework for Finding Timeout Bugs in Distributed Systems by Self-Adaptive Delay Model.",
+        "note": "Repository is named after Chronos, the tool this paper says it built, and is about database testing.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/SecTechTool/Chronos/blob/main/workload/mysql-cluster/sqlancer/src/sqlancer/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+        "excerpt_is_verbatim": true,
+        "note": "workload/mysql-cluster/sqlancer/src/sqlancer/Randomly.java is SQLancer's Randomly.java (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+        "content_sha256": "sha256:301936f755ab919b50c82cd983f3b382fd82eff22830ba0dd5d4575c95ecce80",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "Chronos automatically finds timeout bugs in distributed systems, where incorrect timeout handling can hang or crash a system. It injects fine-grained delays through general runtime delayed libraries, uses deep-priority guided fuzzing to reach bugs on deep paths, and applies transient delays so the test does not pay the wall-clock cost of really waiting. On ZooKeeper, MySQL-Cluster, HDFS and Go-Ethereum it found 27 timeout bugs, all repaired.",
+    "narrative": "SQLancer serves as a workload generator, not an oracle: it produces the SQL workload used to exercise MySQL-Cluster while Chronos injects delays around it. The artifact carries SQLancer source content, consistent with it being bundled for that purpose. The bugs reported are timeout bugs found by delay injection, not logic bugs found by a SQLancer oracle.",
+    "roles": {
+      "M1": "reuse_component"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M1"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "For MySQL-Cluster, the workload is generated by SQLancer [57], one of the widely used SQL generators for testing database systems.",
+            "section": "5 Implementation",
+            "page": 8
+          }
+        ],
+        "reasoning": "M1 states the MySQL-Cluster workload is generated by SQLancer, described as one of the widely used SQL generators for testing database systems. That is reuse to produce input for a different system rather than building on SQLancer, which is why the kind is recorded as workload.",
+        "reuse_kind": "workload"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_srds69199_2025_00038.json b/_data/papers/paper_doi_10_1109_srds69199_2025_00038.json
new file mode 100644
index 0000000..b2ec722
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_srds69199_2025_00038.json
@@ -0,0 +1,367 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:53:46Z",
+  "paper": {
+    "id": "paper:doi:10.1109/srds69199.2025.00038",
+    "title": "Diverse Database Replication Based on Snapshot Isolation – Performance Implications of Improved Dependability",
+    "authors": [
+      "Peter Popov",
+      "Vladimir Stankovic"
+    ],
+    "year": 2025,
+    "venue": "IEEE International Symposium on Reliable Distributed Systems",
+    "doi": "10.1109/srds69199.2025.00038",
+    "arxiv_id": null,
+    "s2_paper_id": "9c1ceed81b5e20d20819cb033534a5480040eb02",
+    "url": "https://doi.org/10.1109/srds69199.2025.00038",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/srds69199.2025.00038",
+      "retrieved_at": "2026-09-08T14:53:46Z",
+      "chars": 65324,
+      "content_sha256": "sha256:83bd30fd67129df036012349cbdde7851ebade377f41dc3313f6f911ba0a7cdd"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 11,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1643
+      },
+      {
+        "number": "II",
+        "title": "RELATED WORK",
+        "start": 7977
+      },
+      {
+        "number": "A",
+        "title": "Database Replication",
+        "start": 7994
+      },
+      {
+        "number": "B",
+        "title": "Design Diversity",
+        "start": 11716
+      },
+      {
+        "number": "III",
+        "title": "REPLICATION ANDSYSTEM MODEL",
+        "start": 12807
+      },
+      {
+        "number": "A",
+        "title": "DivRep Architecture",
+        "start": 17518
+      },
+      {
+        "number": "B",
+        "title": "DivRep Optimistic Regime",
+        "start": 32155
+      },
+      {
+        "number": "C",
+        "title": "Correctness and Liveness of DivRep",
+        "start": 33934
+      },
+      {
+        "number": "A",
+        "title": "Test Harness and Implementation",
+        "start": 36033
+      },
+      {
+        "number": "B",
+        "title": "TPC- C Workload – Comparison with Single DBMSs",
+        "start": 42056
+      },
+      {
+        "number": "C",
+        "title": "TPC- C Workload – Compari son with Non-diverse Pairs",
+        "start": 47488
+      },
+      {
+        "number": "D",
+        "title": "Read -only Workload – Comparison with Single DBMSs",
+        "start": 48575
+      },
+      {
+        "number": "A",
+        "title": "Potential DivRep Optimisations",
+        "start": 50154
+      },
+      {
+        "number": "B",
+        "title": "DivSQL Hybrid Approach",
+        "start": 52304
+      },
+      {
+        "number": "H",
+        "title": "Diab and A. Zomaya, Editors. 2005, John Wiley & Sons. p. 213 - 241.",
+        "start": 62078
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Gashi, I., P. Popov, and L. Strigini, Fault tolerance via diversity for off the-shelf products: a study with SQL database servers. IEEE Transactions on Dependable and Secure Computing, 2007. 4(4): p. 280 294",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Vandiver, B., et al., Tolerating byzantine faults in transaction processing systems using commit barrier scheduling, in Proceedings of 21st ACM SIGOPS Symposium on Operating Systems Principles. 2007, ACM: Steve nson, Washington, USA. p. 59 -72.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Bernstein, A., V. Hadzilacos, and N. Goodman, Concurrency Control and Recovery in Database Systems. 1987, Reading, Mass.: Addison Wesley.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Lin, Y., et al. Middleware Based Data Replication Providing Snapshot Isolation. in ACMSIGMOD International Conference on Management of Data. 2005. Baltimore, Maryland: ACM Press.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Elnikety, S., W. Zwaenepoel, and F. Pedone. Database Replication Using Generalized Snapshot Isolation. in Proceedings of the 24th IEEE Symposium on R eliable Distributed Systems (SRDS'05). 2005. IEEE Computer Society.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Berenson, H., et al. A Critique of ANSI SQL Isolation Levels. in SIGMOD International Conference on Management of Data. 1995. San Jose, California, United States: ACM Press New York, N Y, USA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Gray, J., et al. The Dangers of Replication and a solution. in ACM SIGMOD International Conference on Management of Data. 1996. Montreal, Canada: SIGMOD.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Wiesmann, M., F. Pedone, and A. Schiper. Database Replication Techniques: a Three Parameter Classification. in 19th IEEE Symposium on Reliable Distributed Systems (SRDS'00). 2000. Nurnberg, Germany: IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Vandiver, B., Detecting and Tolerating Byzantine Faults in Database Systems, in Programming Methodology Group. 2008, Massachusetts Institute of Technology: Boston. p. 176.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Garcia, R., R. Rodrigues, and N. Preguica, Efficient middleware for byzantine fault tolerant database replication, in Proceedings of the sixth conference on Computer systems (EuroSys '11). 2011, ACM: Salzburg, Austria. p. 107 -122.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Pedone, F. and N. Schiper, Byzantine fault -tolerant deferred update replication. Journal of the Brazilian Computer Society, 2012. 18(1): p. 3 18.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Molina, H.G., F. Pittelli, and S. Davidson, Applications of Byzantine agreement in database systems. ACM Trans. Database Syst., 1986. 11(1): p. 27 -47.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Littlewood, B., P. Popov, and L. Strigini, Modelling Software Design Diversity-A Review. ACM Computing Surveys, 2001. 33(2): p. 177 208.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Strigini, L., Fault Tolerance Against Design Faults, in Dependable Computing Systems: Paradigms, Performance Issues, and Applications, H. Diab and A. Zomaya, Editors. 2005, John Wiley & Sons. p. 213 - 241.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Gashi, I. and P. Popov. Rephrasing Rules for Off -the- Shelf SQL Database Servers. in Sixth European Dependable Computing Conference (EDCC '06). 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Wiesmann, M., et al. Understanding replication in databases and distributed systems. in Proceedin gs of 20th International Conference on Distributed Computing Systems (ICDCS'2000). 2000. Taipei, Taiwan: IEEE Computer Society Los Alamitos.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Fekete, A., et al., Making Snapshot Isolation Serializable. ACM Transactions on Database Systems (TODS), 2005. 30(2): p. 492 - 528.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Fekete, A., E. O'Neil, and P. O'Neil, A read -only transaction anomaly under snapshot isolation. ACMSIGMOD Record, 2004. 33(3): p. 12 -14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Rigger, M. and Z. Su, Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 2020. 4(OOPSLA): p. Article 211.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Gray, J. and L. Lamport, Consensus on transaction commit. ACM Trans. Database Syst., 2006. 31(1): p. 133 –160.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Behm, A., S. Rielau, and R. Swagerman. Returning Modified Rows – SELECT Statements with Side Effects. in Proceedings of the Thirtieth International Conference on Very Large Data Bases. 2004. Toronto, Canada: Morgan Kaufmann.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Skeen, D. Nonblocking commit protocols. in Proceedings of ACM SIGMOD International conference on management of data. 1981. Ann Arbor, Michigan: ACM Press, New York, NY, USA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Budhiraja, N., et al., The primary -backup approach, in Distributed systems (2nd Ed.). 1993, ACM Press/Addison-Wesley Publishing Co. p. 199–216.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Popov, P. and V. Stankovic. Diverse Da tabase Replication Based on Snapshot Isolation – Performance Implications of Improved Dependability (Technical Report). City St George's, University of London, p.14, London, 2025, https://openaccess.city.ac.uk/id/eprint/35660/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Raz, Y., Serializability by commitment ordering. Information Processing Letters, 1994. 51(5): p. 257 -264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Fischer, M.J., N.A. Lynch, and M.S. Paterson, Impossibility of distributed consensus with one faulty process. J. ACM, 1985. 32(2): p. 374–382.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Kemme, B. and S. Wu. Postgres- R(SI): Combining Replica Control with Concurrency Control based on Snapshot Isolation. in International Conference on Data Engineering. 2005. Tokyo, Japan: IEEE Computer Society.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Patino- Martinez, M., et al., MIDDLE- R: Consistent database repl ication at the middleware level. ACM Transactions on Computer Systems (TOCS), 2005. 23(4): p. 375 -423.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Gorbenko, A., et al., Dependable composite web services with components upgraded online, in Architecting Dependable Systems III. 2005, Springer-Verl ag. p. 92 –121.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Pedone, F., R. Guerraoui, and A. Schiper, The Database State Machine Approach. Distributed and Parallel Databases, 2003. 14(1): p. 71 -98.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Kemme, B. and G. Alonso, A new approach to developing and implementing eager database replica tion protocols. ACM Transactions on Database Systems (TODS) 2000. 25(3): p. 333 - 379.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Harizopoulos, S., et al., OLTP through the looking glass, and what we found there, in Proceedings of the 2008 ACMSIGMOD international conference on Management of d ata. 2008, Association for Computing Machinery: Vancouver, Canada. p. 981 –992. 287",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 19,
+      "text": "Rigger, M. and Z. Su, Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 2020. 4(OOPSLA): p. Article 211.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[19]",
+      "technique": "tlp",
+      "sentence": ", erroneous omission of a row ) – referred to by some as “logic bugs” [19], and DELETEs, INSERTs and UPDATEs wrongly changing the database state.",
+      "context_before": "2], [9], [10], [11]) consider ed Byzantine failure model for DBMSs – the most general model in which components of a system fail in arbitrary ways (in addition to crashing they can process requests incorrectly, corrupt their state, produc e inconsistent output s due to malicious intent, etc.). Non-crash failure models must be addressed, and we have contributed to the development of novel protocols that consider those. However, t he IRFM failure model assumed by DivSQL lies between crash and Byzantine failure s model s. In addition to crash failures, it guards against failures of SELECT s (e.g.",
+      "context_after": "It is likely the most suitable failure model for majority of applications based on the compelling evidence that non-crash failures are dominant in DBMSs [1, 2]. Our approach is thus based on a failure model with practical relevance. Further research is needed to establish how likely DB MSs and the clients exhibit more challenging Byzantine behaviour. Client application s access our system using JDBC interface. Our JDBC driver implement s the client side of DivRep protocol. We use different SI -enabled DBMSs, and their JDBC drivers, in DivSQL. We configured our prototype implementation to run",
+      "section": "III REPLICATION ANDSYSTEM MODEL",
+      "page": 3,
+      "char_offset": 16679,
+      "cited_reference": {
+        "number": 19,
+        "text": "Rigger, M. and Z. Su, Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 2020. 4(OOPSLA): p. Article 211.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:03:26Z",
+    "is_model_written": true,
+    "summary": "This paper measures what diverse database replication costs. Running several different DBMS implementations as replicas under snapshot isolation means a fault in one is masked by the others, improving dependability; the question is how much throughput and latency that redundancy consumes.",
+    "narrative": "The single citation supplies terminology: the paper describes faults where a row is wrongly omitted from a result, or where a DELETE, INSERT or UPDATE wrongly changes the database state, and notes these are referred to by some as logic bugs. Diverse replication masks exactly this class of fault, which is why the term is introduced, but nothing of SQLancer's is used or measured.",
+    "roles": {
+      "M1": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_tdsc_2024_3521591.json b/_data/papers/paper_doi_10_1109_tdsc_2024_3521591.json
new file mode 100644
index 0000000..c0fa387
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_tdsc_2024_3521591.json
@@ -0,0 +1,500 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:54:00Z",
+  "paper": {
+    "id": "paper:doi:10.1109/tdsc.2024.3521591",
+    "title": "Improving Multitasking DBMS Fuzzing With More Accurate Coverage and Testcase Trimming",
+    "authors": [
+      "Jiaqi Li",
+      "Yajin Zhou",
+      "Lei Wu"
+    ],
+    "year": 2025,
+    "venue": "IEEE Transactions on Dependable and Secure Computing",
+    "doi": "10.1109/tdsc.2024.3521591",
+    "arxiv_id": null,
+    "s2_paper_id": "8693724c0f546d2da26e99439fd85eea5dfb79fd",
+    "url": "https://doi.org/10.1109/tdsc.2024.3521591",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/tdsc.2024.3521591",
+      "retrieved_at": "2026-09-08T14:54:00Z",
+      "chars": 74987,
+      "content_sha256": "sha256:4db16d8e9345c4724c5ef9d0b5c82220e9d3bec83b50aa037cef7e346b9e8654"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1480
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND",
+        "start": 13157
+      },
+      {
+        "number": "A",
+        "title": "DBMS Architecture",
+        "start": 13172
+      },
+      {
+        "number": "B",
+        "title": "Coverage-Guided Fuzzing",
+        "start": 14261
+      },
+      {
+        "number": "C",
+        "title": "Seed Selection",
+        "start": 15883
+      },
+      {
+        "number": "D",
+        "title": "Coverage Tracking",
+        "start": 17774
+      },
+      {
+        "number": "E",
+        "title": "Testcase Evaluation",
+        "start": 20524
+      },
+      {
+        "number": "F",
+        "title": "Scope of Our Work",
+        "start": 21547
+      },
+      {
+        "number": "III",
+        "title": "IMPROVE COVERAGE ACCURACY",
+        "start": 22447
+      },
+      {
+        "number": "A",
+        "title": "Motivation",
+        "start": 22478
+      },
+      {
+        "number": "B",
+        "title": "Work-Task Coverage Tracking",
+        "start": 27753
+      },
+      {
+        "number": "C",
+        "title": "Unstable Edge Filtering",
+        "start": 32883
+      },
+      {
+        "number": "IV",
+        "title": "DEPENDENCY-GUIDED TRIMMING",
+        "start": 37586
+      },
+      {
+        "number": "A",
+        "title": "Motivation",
+        "start": 37617
+      },
+      {
+        "number": "B",
+        "title": "Overview",
+        "start": 40355
+      },
+      {
+        "number": "C",
+        "title": "Build Statement Dependency",
+        "start": 41177
+      },
+      {
+        "number": "D",
+        "title": "Static Trimming",
+        "start": 43086
+      },
+      {
+        "number": "E",
+        "title": "Dynamic Trimming",
+        "start": 44461
+      },
+      {
+        "number": "V",
+        "title": "IMPLEMENTATION DETAILS",
+        "start": 46301
+      },
+      {
+        "number": "A",
+        "title": "Runtime Overhead",
+        "start": 48998
+      },
+      {
+        "number": "B",
+        "title": "DBMS Bug Detection",
+        "start": 50546
+      },
+      {
+        "number": "C",
+        "title": "Contribution of Work-Task Tracking and Testcase Trimming",
+        "start": 52574
+      },
+      {
+        "number": "D",
+        "title": "Comparison to Existing DBMS Fuzzers",
+        "start": 56605
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Well-known users of SQLite,” 2022. [Online]. Available: https://www. sqlite.org/famous.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“MySQL customers,” 2022. [Online]. Available: https://www.mysql.com/ customers/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "A. Bannister, “SQLite patches use-after-free bug that left apps open to code execution, denial-of-service exploits,” 2021. [Online]. Available: https://portswigger.net/daily-swig/sqlite-patches-use-after-free-bugthat-left-apps-open-to-code-execution-denial-of-service-exploits",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "C. Cimpanu, “Google chrome impacted by new Magellan 2.0 vulnerabilities,” 2019. [Online]. Available: https://www.zdnet.com/article/googlechrome-impacted-by-new-magellan-2--0-vulnerabilities/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic detection and diagnosis of performance regressions in database systems,” in Proc. VLDB Endowment, vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “SQUIRREL: Testing database management systems with language validity and coverage feedback,” in Proc. 2020 ACMSIGSAC Conf. Comput. Commun. Secur., 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "“SQLite homepage,” 2022. [Online]. Available: https://www.sqlite.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "K. Serebryany, “Continuous fuzzing with libFuzzer and addresssanitizer,” inProc. IEEE Cybersecurity Develop., 2016, pp. 157–157.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of {DBMS} with coverage-based guidance,” in Proc. 31st USENIX Secur. Symp., 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Z.-M. Jiang, J.-J. Bai, and Z. Su, “ {DynSQL}: Stateful fuzzing for database management systems with complex and valid {SQL}query generation,” inProc. 32nd USENIX Secur. Symp., 2023, pp. 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammar-free DBMS fuzzing,” in Proc. 37th IEEE/ACM Int. Conf. Automated Softw. Eng., 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "M. Wang et al., “Industry practice of coverage-guided enterprise-level DBMS fuzzing,” in Proc. IEEE/ACM 43rd Int. Conf. Softw. Eng.: Softw. Eng. Pract., 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Z. Wu, J. Liang, M. Wang, C. Zhou, and Y. Jiang, “Unicorn: Detect runtime errors in time-series databases with hybrid input synthesis,” in Proc. 31st ACMSIGSOFT Int. Symp. Softw. Testing Anal., 2022, pp. 251–262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "J. Liang et al., “Sequence-oriented DBMS fuzzing,” in Proc. IEEE 39th Int. Conf. Data Eng., 2023, pp. 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "J. Li, K. Wang, Y. Chen, Y. Zhou, L. Wu, and J. Wang, “Detecting DBMS bugs with context-sensitive instantiation and multi-plan execution,” 2023, arXiv:2312.04941.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "“DB-engines ranking of relational DBMS,” 2024. [Online]. Available: https://db-engines.com/en/ranking/relationaldbms",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "“Addresssanitizer,” 2022. [Online]. Available: https://github.com/google/ sanitizers/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "“control-flow graph,” 2024. [Online]. Available: https://en.wikipedia.org/ wiki/Control-flow_graph",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "“American fuzzy lop (2.56b),” 2024. [Online]. Available: https://lcamtuf. coredump.cx/afl/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "“Honggfuzz,” 2024. [Online]. Available: https://github.com/google/ honggfuzz",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "S. Gan et al., “CollAFL: Path sensitive fuzzing,” in Proc. IEEE Symp. Secur. Privacy, 2018, pp. 679–696. 2770 IEEETRANSACTIONS ONDEPENDABLE ANDSECURE COMPUTING, VOL. 22, NO. 3, MAY/JUNE 2025",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "“AFL-fuzz approach,” 2024 [Online]. Available: https://aflplus.plus/docs/ afl-fuzz_approach/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "“LLVM pass,” 2024. [Online]. Available: https://llvm.org/docs/ WritingAnLLVMNewPMPass.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "I. Yun, S. Lee, M. Xu, Y. Jang, and T. Kim, “ {QSYM}: A practical concolic execution engine tailored for hybrid fuzzing,” in Proc. 27th USENIX Secur. Symp., 2018, pp. 745–761.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "P. Chen and H. Chen, “Angora: Efficient fuzzing by principled search,” in Proc. 2018 IEEE Symp. Secur. Privacy, 2018, pp. 711–725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "“Parallel computing,” 2024. [Online]. Available: https://en.wikipedia.org/ wiki/Parallel_computing",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "“Mapreduce,” 2024. [Online]. Available: https://en.wikipedia.org/wiki/ MapReduce",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "H. Chen et al., “ {MUZZ}: Thread-aware grey-box fuzzing for effective bug hunting in multithreaded programs,” in Proc. 29th USENIX Secur., 2020, pp. 2325–2342.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "S. Rawat, V. Jain, A. Kumar, L. Cojocar, C. Giuffrida, and H. Bos, “VUzzer: Application-aware evolutionary fuzzing,” in Proc. Annu. Netw. Distrib. Syst. Secur. Symp., vol. 17, 2017, pp. 1–14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "M. Böhme, V .-T. Pham, and A. Roychoudhury, “Coverage-based greybox fuzzing as Markov chain,” in Proc. 2016 ACMSIGSAC Conf. Comput. Commun. Secur., 2016, pp. 1032–1043.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "S. Wang, J. Nam, and L. Tan, “Qtep: Quality-aware test case prioritization,” inProc. 11th Joint Meeting Foundations Softw. Eng., 2017, pp. 523–534.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "T. Petsios, J. Zhao, A. D. Keromytis, and S. Jana, “Slowfuzz: Automated domain-independent detection of algorithmic complexity vulnerabilities,” inProc. 2017 ACMSIGSAC Conf. Comput. Commun. Secur., 2017, pp. 2155–2168.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "J. Wang, B. Chen, L. Wei, and Y. Liu, “Skyfire: Data-driven seed generation for fuzzing,” in Proc. IEEE Symp. Secur. Privacy, 2017, pp. 579–594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "C. Lyu et al., “ {MOPT}: Optimized mutation scheduling for fuzzers,” in Proc. 28th USENIX Secur. Symp., 2019, pp. 1949–1966.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proc. 28th ACM Joint Meeting Eur. Softw. Eng. Conf. Symp. Foundations Softw. Eng., 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 36,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” in Proc. ACM Program. Lang., vol. 4, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "J. Song et al., “Testing database systems via differential query execution,” inProc. IEEE/ACM 45th Int. Conf. Softw. Eng., 2023, pp. 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Z. Hao et al., “PINOLO: Detecting logical bugs in database management systems with approximate query synthesis,” in Proc. 2023 USENIX Annu. Tech. Conf., 2023, pp. 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "X. Tang, S. Wu, D. Zhang, F. Li, and G. Chen, “Detecting logic bugs of join optimizations in DBMS,” in 1, no. 1, pp. 1–26, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proc. 14th USENIX Symp. Operating Syst. Des. Implementation, 2020, pp. 667–682. Jiaqi Li received the BE degree in information security from Sun Yat-sen University, in 2015. He is currently working toward the doctoral degree in cyberspace security with Zhejiang University. His current research interests include databas",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 35,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proc. 28th ACM Joint Meeting Eur. Softw. Eng. Conf. Symp. Foundations Softw. Eng., 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 36,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” in Proc. ACM Program. Lang., vol. 4, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proc. 14th USENIX Symp. Operating Syst. Des. Implementation, 2020, pp. 667–682. Jiaqi Li received the BE degree in information security from Sun Yat-sen University, in 2015. He is currently working toward the doctoral degree in cyberspace security with Zhejiang University. His current research interests include database security and confidential computing. Yajin Zhou received the PhD degree in computer science from North Carolina State University, Raleigh, NC, USA. He is currently a ZJU 100 Young professor with the School of Cyber Science and Technology, and the College of Computer Science and Technology, Zhejiang University, China. His research mainly focuses on smartphone and system security, such as identifying real-world threats and building practical solutions, mainly in the context of embedded systems (or IoT devices). Lei Wu received the PhD degree from North Carolina State University, in 2015. He is an associate professor with the School of Cyber Science and Technology, and the College of Computer Science and Technology, Zhejiang University, China. His research interests include mainly in security areas, including system security, and blockchain security.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [35], TLP [36], DQE [37], Pinolo [38] leverage the idea of metamorphic testing to identify semantic bugs that lead to the incorrect result of SQL statement.",
+      "context_before": "and uses them to guide the instantiating of identifiers to improve the validity of generated testcases. LEGO [14] identifies affinity between different types (e.g., INSERT and SELECT) based on coverage feedback, and tries to generate the testcase containing these affinities. GRIFFIN [11] proposes a grammar-free mutation policy that randomly selects statements from multiple testcases to construct a new one. Our solutions can be combined with all the aforementioned fuzzers to further improve the efficiency of bug detection. The second type proposes new test oracles to identify semantic bugs in DBMSs.",
+      "context_after": "They transform a given query qinto another query q′such that their querying results satisfy a specific relation. The violation of the relation upon the querying results indicates a semantic bug. Kangaroo [15],T Q S [39], and APOLLO [5]are oracles based on differential testing. Kangaroo and TQS detect the semantic bug by running the multiple query plans of a statement and checking the consistency of their query results. APOLLO finds performance regression bugs by executing the same query on the DBMSs with different versions. PQS [40] generates queries that a specific row should be returned. If th",
+      "section": "D Comparison to Existing DBMS Fuzzers",
+      "page": 14,
+      "char_offset": 65719,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [40] generates queries that a specific row should be returned.",
+      "context_before": "entify semantic bugs that lead to the incorrect result of SQL statement. They transform a given query qinto another query q′such that their querying results satisfy a specific relation. The violation of the relation upon the querying results indicates a semantic bug. Kangaroo [15],T Q S [39], and APOLLO [5]are oracles based on differential testing. Kangaroo and TQS detect the semantic bug by running the multiple query plans of a statement and checking the consistency of their query results. APOLLO finds performance regression bugs by executing the same query on the DBMSs with different versions.",
+      "context_after": "If the DBMS fails to fetch the row, it indicates the statement triggers a semantic bug. Most oracles can be applied to coverage-guided fuzzers except PQS because PQS puts strict limitations on the pattern of generated testcases. Therefore, our work can be directly applied to these works to improve the effectiveness of semantic bug detection in DBMS. The last category is the improvements that do not fall into the first two categories. Unicorn [13] focuses on bug finding in time-series databases. Ratal [12] improves the coverage feedback accuracy in multiprocessing DBMS and designs a new schema t",
+      "section": "D Comparison to Existing DBMS Fuzzers",
+      "page": 14,
+      "char_offset": 66410,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Most oracles can be applied to coverage-guided fuzzers except PQS because PQS puts strict limitations on the pattern of generated testcases.",
+      "context_before": "ts satisfy a specific relation. The violation of the relation upon the querying results indicates a semantic bug. Kangaroo [15],T Q S [39], and APOLLO [5]are oracles based on differential testing. Kangaroo and TQS detect the semantic bug by running the multiple query plans of a statement and checking the consistency of their query results. APOLLO finds performance regression bugs by executing the same query on the DBMSs with different versions. PQS [40] generates queries that a specific row should be returned. If the DBMS fails to fetch the row, it indicates the statement triggers a semantic bug.",
+      "context_after": "Therefore, our work can be directly applied to these works to improve the effectiveness of semantic bug detection in DBMS. The last category is the improvements that do not fall into the first two categories. Unicorn [13] focuses on bug finding in time-series databases. Ratal [12] improves the coverage feedback accuracy in multiprocessing DBMS and designs a new schema to automatically reproduce bugs.IX. CONCLUSION In this paper, we study the negative impact of two limitations that are prevalent in current DBMS fuzzers: inaccurate coverage feedback and absence of testcase trimming. We propose se",
+      "section": "D Comparison to Existing DBMS Fuzzers",
+      "page": 14,
+      "char_offset": 66564,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:52Z",
+    "is_model_written": true,
+    "summary": "This work improves coverage-guided DBMS fuzzing for systems that run many tasks concurrently, where coverage attributed to the wrong task makes the feedback signal misleading. It contributes more accurate per-task coverage measurement and test case trimming that shrinks inputs without losing the behaviour that made them interesting.",
+    "narrative": "SQLancer's oracles appear in the comparison of existing DBMS testing approaches, where NoREC, TLP and PQS are grouped with DQE and Pinolo as metamorphic techniques for finding semantic bugs. The paper makes one observation of substance about combining them with its own approach: most of these oracles can be applied to a coverage-guided fuzzer, but PQS cannot, because it constrains the pattern of generated test cases too tightly for a coverage-driven generator to satisfy.",
+    "roles": {
+      "M1": "definition",
+      "M2": "definition",
+      "M3": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_tkde_2026_3656491.json b/_data/papers/paper_doi_10_1109_tkde_2026_3656491.json
new file mode 100644
index 0000000..6e4c0dc
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_tkde_2026_3656491.json
@@ -0,0 +1,521 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:08:10Z",
+  "paper": {
+    "id": "paper:doi:10.1109/tkde.2026.3656491",
+    "title": "Effective Bug Detection in Graph Database Engines: An LLM-Based Approach",
+    "authors": [
+      "Jiayi Wu",
+      "Zhengyu Wu",
+      "Xunkai Li",
+      "Ronghua Li",
+      "Hongchao Qin",
+      "Guoren Wang"
+    ],
+    "year": 2026,
+    "venue": "IEEE Transactions on Knowledge and Data Engineering",
+    "doi": "10.1109/tkde.2026.3656491",
+    "arxiv_id": "2402.00292v1",
+    "s2_paper_id": "51c1741064427aadb230a3ce760368f62577dbae",
+    "url": "https://doi.org/10.1109/tkde.2026.3656491",
+    "also_indexed_as": [
+      "paper:arxiv:2402.00292"
+    ]
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2402.00292v1",
+      "retrieved_at": "2026-09-10T15:08:10Z",
+      "chars": 69045,
+      "content_sha256": "sha256:f431211699ae0e694fb596bfa4bbecc7b74c09c1a3a6c6ec5f99c14069db226d"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Ibrahim Abdelaziz, Essam Mansour, Mourad Ouzzani, Ashraf Aboulnaga, and Panos Kalnis. 2017. Query optimizations over decentralized RDF graphs. In. IEEE, 139–142.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "DataStax Dylan Bethune-Waddell Expero Google Orion Health IBM Rafael Fernandes Robert Dale Seeq Amazon, Aurelius. 2023. JanusGraph. https: //janusgraph.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Sjoerd Mullender Andreas Seltenreich, Bo Tang. 2022. SQLsmith. https://github. com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Renzo Angles and Claudio Gutierrez. 2008. Survey of graph database models. ACM Computing Surveys (CSUR) 40, 1 (2008), 1–39.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Dana Angluin. 1990. Negative results for equivalence queries. Machine Learning 5 (1990), 121–150.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al .2020. Language models are few-shot learners. Advances in neural information processing systems 33 (2020), 1877–1901.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Donald D Chamberlin and Raymond F Boyce. 1974. SEQUEL: A structured English query language. In Proceedings of the 1974 ACMSIGFIDET (now SIGMOD) workshop on Data description, access and control. 249–264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Zihan Chen, Lei Nico Zheng, Cheng Lu, Jialu Yuan, and Di Zhu. 2023. ChatGPT Informed Graph Neural Network for Stock Movement Prediction. arXiv preprint arXiv:2306.03763 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Edgar Frank Codd. 1983. A relational model of data for large shared data banks. Commun. ACM 26, 1 (1983), 64–69.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Jiaxi Cui, Zongjian Li, Yang Yan, Bohua Chen, and Li Yuan. 2023. Chatlaw: Open-source legal large language model with integrated external knowledge bases. arXiv preprint arXiv:2306.16092 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Andrzej Czerepicki. 2016. Application of graph databases for transport purposes. Bulletin of the Polish Academy of Sciences. Technical Sciences 64, 3 (2016), 457–466.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "SAP Enterprise. 2023. OrientDB Community Edition. http://www.orientdb.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Robert B Evans and Alberto Savoia. 2007. Differential testing: a new approach to change detection. In The 6th Joint Meeting on European software engineering conference and the ACMSIGSOFT Symposium on the Foundations of Software Engineering: Companion Papers. 549–552.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Apache Software Foundation. 2023. HugeGraph. https://hugegraph.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Apache Software Foundation. 2023. TinkerGraph. https://tinkerpop.incubator. apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and Andrés Taylor. 2018. Cypher: An evolving query language for property graphs. InProceedings of the 2018 international conference on management of data. 1433– 1445.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Qing Gao, Hansheng Zhang, Jie Wang, Yingfei Xiong, Lu Zhang, and Hong Mei. 2015. Fixing recurring crash bugs via analyzing q&a sites (T). In 2015 30th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 307–318.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Marielet Guillermo, Maverick Rivera, Ronnie Concepcion, Robert Kerwin Billones, Argel Bandala, Edwin Sybingco, Alexis Fillone, and Elmer Dadios. 2022. Graph Database-modelled Public Transportation Data for Geographic Insight Web Application. In Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing (SNPD). IEEE, 2–7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Edwin Hewitt and Leonard J Savage. 1955. Symmetric measures on Cartesian products. Trans. Amer. Math. Soc. 80, 2 (1955), 470–501.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting bugs in Cypher graph database engines. In Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Bitnine Global Inc. 2023. Agensgraph. https://bitnine.net/agensgraph/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Neo4j Inc. 2023. Neo4j. https://neo4j.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Himanshu Jain, Daniel Kroening, Natasha Sharygina, and Edmund Clarke. 2005. Word level predicate abstraction and refinement for verifying RTL Verilog. In Proceedings of the 42nd annual Design Automation Conference. 445–450.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Matteo Kamm. 2022. Testing Graph Databases using Predicate Partitioning. Master’s thesis. ETH Zurich.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al. 2023. Segment anything. arXiv preprint arXiv:2304.02643 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Takahiro Konno, Runhe Huang, Tao Ban, and Chuanhe Huang. 2017. Goods recommendation based on retail knowledge in a Neo4j graph database combined with an inference mechanism implemented in jess. In 2017 IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Computed, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation (Smar",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Bingfeng Luo, Yansong Feng, Zheng Wang, Songfang Huang, Rui Yan, and Dongyan Zhao. 2018. Marrying up regular expressions with neural networks: A case study for spoken language understanding. arXiv preprint arXiv:1805.05588 (2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Hanjia Lyu, Song Jiang, Hanqing Zeng, Yinglong Xia, and Jiebo Luo. 2023. Llmrec: Personalized recommendation via prompting large language models. arXiv preprint arXiv:2307.15780 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jun Ma and Bo Wang. 2023. Segment anything in medical images. arXiv preprint arXiv:2304.12306 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "William M McKeeman. 1998. Differential testing for software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "OpenAI. 2022. ChatGPT. https://openai.com/blog/chatgpt/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al .2022. Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems 35 (2022), 27730–27744.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Rob Reagan and Rob Reagan. 2018. Cosmos DB. Web Applications on Azure: Developing for Global Scale (2018), 187–255.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Yuxiang Ren, Hao Zhu, Jiawei Zhang, Peng Dai, and Liefeng Bo. 2021. Ensemfdet: An ensemble approach to fraud detection based on bipartite graph. In 2021 IEEE 37th International Conference on Data Engineering (ICDE). IEEE, 2039–2044.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Manuel Rigger. 2022. SQLancer. https://github.com/sqlancer/sqlancer.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "Ian Robinson, Jim Webber, and Emil Eifrem. 2015. Graph databases: new opportunities for connected data. \" O’Reilly Media, Inc.\".",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Marko A Rodriguez. 2015. The gremlin graph traversal machine and language (invited talk). In Proceedings of the 15th Symposium on Database Programming Languages. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Marko A Rodriguez. 2015. The gremlin graph traversal machine and language (invited talk). In Proceedings of the 15th Symposium on Database Programming Languages. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Sudipta Sen, Akash Mehta, Runa Ganguli, and Soumya Sen. 2021. Recommendation of influenced products using association rule mining: Neo4j as a case study. SN Computer Science 2 (2021), 1–17.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Donald R Slutz. 1998. Massive stochastic testing of SQL. In VLDB, Vol. 98. Citeseer, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "solid IT. 2023. DB-Engines Ranking of Graph DBMS. https://db-engines.com/ en/ranking/graph+ dbms.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting Logic Bugs of Join Optimizations in DBMS. Proceedings of the ACM on Management of Data 1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Ran Wang, Zhengyi Yang, Wenjie Zhang, and Xuemin Lin. 2020. An empirical study on recent graph database systems. In Knowledge Science, Engineering and Management: 13th International Conference, KSEM 2020, Hangzhou, China, August 28–30, 2020, Proceedings, Part I 13. Springer, 328–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and S Yu Philip. 2020. A comprehensive survey on graph neural networks. IEEE transactions on neural networks and learning systems 32, 1 (2020), 4–24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Jiawei Zhang. 2023. Graph-ToolFormer: To Empower LLMs with Graph Reasoning Ability via Prompt Augmented by ChatGPT. arXiv preprint arXiv:2304.11116 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, et al .2023. A survey of large language models. arXiv preprint arXiv:2303.18223 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding bugs in Gremlin-based graph database systems via randomized differential testing. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis. 302–313. 12",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 35,
+      "text": "Manuel Rigger. 2022. SQLancer. https://github.com/sqlancer/sqlancer.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 23,24,37]",
+      "technique": "tlp",
+      "sentence": "The former typically involves detecting bugs in graph database engines by generating equivalent queries[ 5] or utilizing predicate partitioning[ 23,24,37].",
+      "context_before": "es can dynamically generate real-time recommended content based on users’ most recent actions and social connections[ 26,42]. In the field of logistics and freight transportation, graph database engines can be harnessed to determine optimal delivery routes, thereby reducing costs and enhancing efficiency[11, 18]. As shown in Table 1, existing methods for detecting bugs in graph database engines can be broadly divided into two categories: methods for utilizng various queries to detect bugs within the same graph database engines, methods applying the same query on various graph database engines.",
+      "context_after": "The latter approach entails running the same queries on different graph database engines and identifying bugs based on discrepancies in query results[ 20,50]. The former method demands users with substantial prior knowledge on the graph model and query objectives when generating equivalent queries to identify bugs in the graph database engine[ 5]. In the predicate partitioning approach for detecting graph database engine bugs[ 24], it starts by constructing queries based on a top-down expression generator[ 35], and then apply Ternary Logic Partitioning techniques to create correspondent queri",
+      "section": null,
+      "page": 1,
+      "char_offset": 4417,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "In the predicate partitioning approach for detecting graph database engine bugs[ 24], it starts by constructing queries based on a top-down expression generator[ 35], and then apply Ternary Logic Partitioning techniques to create correspondent queries, namely the \"True, False or Null\" queries, for running on graph database engine.",
+      "context_before": "ame graph database engines, methods applying the same query on various graph database engines. The former typically involves detecting bugs in graph database engines by generating equivalent queries[ 5] or utilizing predicate partitioning[ 23,24,37]. The latter approach entails running the same queries on different graph database engines and identifying bugs based on discrepancies in query results[ 20,50]. The former method demands users with substantial prior knowledge on the graph model and query objectives when generating equivalent queries to identify bugs in the graph database engine[ 5].",
+      "context_after": "Such a method assess whether the subsetarXiv:2402.00292v1 [cs.DB] 1 Feb 2024 Table 1: Comparison of the proposed DGDB with closely related works. Method Category Language Not require large prior knowledgeHigh proportion of non-empty-result queries Grand[50]the method applying the same query on various graph database enginesgremlin % % GDSmith[20]the method applying the same query on various graph database enginescypher %! PP-DB[24]the method for utilizng various queries to detect bugs within the same graph database enginesall % % DGDBthe method applying the same query on various graph databas",
+      "section": null,
+      "page": 1,
+      "char_offset": 4923,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[3,35,43]",
+      "technique": null,
+      "sentence": "Most current research focuses on the latter approach and the most widely adopted method is running randomly generated queries [3,35,43] on different graph database engines for bug detection.",
+      "context_before": "g the same query on various graph database enginesgremlin % % GDSmith[20]the method applying the same query on various graph database enginescypher %! PP-DB[24]the method for utilizng various queries to detect bugs within the same graph database enginesall % % DGDBthe method applying the same query on various graph database enginesall! ! of the final query results intersect to detect graph database engine bugs. While this approach successfully identifies some logical bugs, its effectiveness is limited due to the lower quality of queries generation and the manual need for reducing test samples.",
+      "context_after": "However, the generated queries often return empty results in many samples, which compromise the efficiency for bug detection. To address this issue, Grand[ 50] proposed a method based on traversal models to generate gremlin queries[ 40], significantly increasing the non-empty-result query ratio. However, this method is only applicable to gremlin-based graph database engines and requires users to master the gremlin API[ 41]. GDsmith[ 20] proposed a method to generate cypher queries[ 16] based on the guidance of property graphs. This method first generates a cypher skeleton, then creates patter",
+      "section": null,
+      "page": 2,
+      "char_offset": 6132,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger. 2022. SQLancer. https://github.com/sqlancer/sqlancer.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS[ 38] selects a target data from randomly generated tables, generates conditional expressions based on the target data, constructs an SQL query with a where or join clause, and determines the presence of bugs by checking if the result is included in the result set.",
+      "context_before": "ub. 5 RELATED WORK Recently, researchers have proposed various methods for detecting bugs in relational database engines. SQLsmith[ 3] is a fuzz testing method for detecting bugs in relational database engines. It first generates SQL query randomly according to the Abstract Syntax Tree (AST) generator, then simplifies the generated SQL query through SQL Reducer. Finally, it determines the presence of bugs based on whether the generated SQL query can successfully run on the relational database engines. While this method is effective in detecting crash bugs, it cannot identify wrong-result bugs.",
+      "context_after": "While this method can effectively detect bugs in relational database engines, its applicability is limited across different relational database engines due to variations in SQL syntax and query optimization strategies. NoRec[ 36] converts the original SQL query into a nonoptimized SQL query and compares the results of these two SQL queries for consistency. However, this method cannot detect bugs in SQL queries with subquery structures. TLP[ 37] transforms randomly generated original SQL queries into three different logical queries based on the true, false, and null ternary logic. It detects b",
+      "section": null,
+      "page": 11,
+      "char_offset": 54992,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "NoRec[ 36] converts the original SQL query into a nonoptimized SQL query and compares the results of these two SQL queries for consistency.",
+      "context_before": "database engines. While this method is effective in detecting crash bugs, it cannot identify wrong-result bugs. PQS[ 38] selects a target data from randomly generated tables, generates conditional expressions based on the target data, constructs an SQL query with a where or join clause, and determines the presence of bugs by checking if the result is included in the result set. While this method can effectively detect bugs in relational database engines, its applicability is limited across different relational database engines due to variations in SQL syntax and query optimization strategies.",
+      "context_after": "However, this method cannot detect bugs in SQL queries with subquery structures. TLP[ 37] transforms randomly generated original SQL queries into three different logical queries based on the true, false, and null ternary logic. It detects bugs in relational database engines by comparing the result sets of the transformed three SQL queries with the original SQL query.TQS[ 45] proposes a testing framework capable of detecting logical bugs arising from multi-table join queries. The authors treat the database schema as a graph and use biased random walks to generate SQL queries. Based on biased r",
+      "section": null,
+      "page": 11,
+      "char_offset": 55480,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP[ 37] transforms randomly generated original SQL queries into three different logical queries based on the true, false, and null ternary logic.",
+      "context_before": "e target data, constructs an SQL query with a where or join clause, and determines the presence of bugs by checking if the result is included in the result set. While this method can effectively detect bugs in relational database engines, its applicability is limited across different relational database engines due to variations in SQL syntax and query optimization strategies. NoRec[ 36] converts the original SQL query into a nonoptimized SQL query and compares the results of these two SQL queries for consistency. However, this method cannot detect bugs in SQL queries with subquery structures.",
+      "context_after": "It detects bugs in relational database engines by comparing the result sets of the transformed three SQL queries with the original SQL query.TQS[ 45] proposes a testing framework capable of detecting logical bugs arising from multi-table join queries. The authors treat the database schema as a graph and use biased random walks to generate SQL queries. Based on biased random walks and graph embedding methods, high-quality SQL queries are generated to detect numerous logical bugs in relational database engines. In contrast to detecting bugs in relational database engines, the identification of",
+      "section": null,
+      "page": 11,
+      "char_offset": 55701,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:21:29Z",
+    "is_model_written": true,
+    "summary": "DGDB uses large language models to generate queries for detecting bugs in graph database engines, then applies differential testing to the results. The authors argue existing approaches are tied to a single graph query language and require extensive prior knowledge to produce queries. Applied to engines using Cypher, Gremlin and SPARQL it found 23 previously unknown wrong-result bugs, and reports more than three times as many bugs as prior methods on Cypher engines.",
+    "narrative": "All three of SQLancer's original oracles are described in the survey of prior approaches -- PQS selecting target data and constructing a query to retrieve it, NoREC comparing optimized and non-optimized forms, TLP partitioning on ternary logic -- as the relational precedent for graph-engine testing.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "background",
+      "M4": "definition",
+      "M5": "definition",
+      "M6": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_tse_2025_3574328.json b/_data/papers/paper_doi_10_1109_tse_2025_3574328.json
new file mode 100644
index 0000000..121248a
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_tse_2025_3574328.json
@@ -0,0 +1,1256 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:06:30Z",
+  "paper": {
+    "id": "paper:doi:10.1109/tse.2025.3574328",
+    "title": "SQLaw: Detecting Bugs in GPU Database Management Systems via Rule-Based Differential Execution",
+    "authors": [
+      "Jiaxin Hu",
+      "Rongxin Wu"
+    ],
+    "year": 2025,
+    "venue": "IEEE Transactions on Software Engineering",
+    "doi": "10.1109/tse.2025.3574328",
+    "arxiv_id": null,
+    "s2_paper_id": "4f46526dc9fbb444b85f054e44c84bfa107065fd",
+    "url": "https://doi.org/10.1109/tse.2025.3574328",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/tse.2025.3574328",
+      "retrieved_at": "2026-09-07T17:06:30Z",
+      "chars": 87714,
+      "content_sha256": "sha256:bfb998eac1b65248d7345ee690890399d7d181e9a5fa802159b198de1826346e"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 17,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1503
+      },
+      {
+        "number": "II",
+        "title": "BACKGROUND ANDMOTIVATION",
+        "start": 9005
+      },
+      {
+        "number": "A",
+        "title": "GPU DBMSs",
+        "start": 9414
+      },
+      {
+        "number": "B",
+        "title": "GPU-Related Bugs",
+        "start": 13721
+      },
+      {
+        "number": "C",
+        "title": "Preliminary Study of Detecting GPU-Related Bugs",
+        "start": 15039
+      },
+      {
+        "number": "D",
+        "title": "Challenge",
+        "start": 19486
+      },
+      {
+        "number": "E",
+        "title": "Key Idea",
+        "start": 22144
+      },
+      {
+        "number": "III",
+        "title": "APPROACH",
+        "start": 25357
+      },
+      {
+        "number": "A",
+        "title": "Overview of SQLaw",
+        "start": 25713
+      },
+      {
+        "number": "B",
+        "title": "Offline Rule Learning",
+        "start": 27624
+      },
+      {
+        "number": "C",
+        "title": "Online Interpreter With Mutation",
+        "start": 35373
+      },
+      {
+        "number": "D",
+        "title": "Bug Detector",
+        "start": 37496
+      },
+      {
+        "number": "E",
+        "title": "Detailed Process of SQLaw",
+        "start": 39000
+      },
+      {
+        "number": "IV",
+        "title": "IMPLEMENTATION",
+        "start": 41128
+      },
+      {
+        "number": "V",
+        "title": "EVALUATION",
+        "start": 42888
+      },
+      {
+        "number": "A",
+        "title": "Evaluation Step",
+        "start": 43350
+      },
+      {
+        "number": "B",
+        "title": "Effectiveness and Efficiency of SQLaw",
+        "start": 46406
+      },
+      {
+        "number": "C",
+        "title": "Comparison With Baselines",
+        "start": 52216
+      },
+      {
+        "number": "D",
+        "title": "Ablation Study",
+        "start": 56309
+      },
+      {
+        "number": "E",
+        "title": "Discussion",
+        "start": 60006
+      },
+      {
+        "number": "VI",
+        "title": "RELATED WORK",
+        "start": 65069
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 72155
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "A. Shanbhag, S. Madden, and X. Yu, “A study of the fundamental performance characteristics of GPUs and CPUs for database analytics,” inProc. SIGMOD’20, 2020, pp. 1617–1632.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "V. Rosenfeld, S. Bre ß, and V. Markl, “Query processing on heterogeneous CPU/GPU systems,” CSUR’22, vol. 55, no. 1, pp. 1–38, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "C. Lutz, S. Bre ß, S. Zeuch, T. Rabl, and V. Markl, “Triton join: efficiently scaling to a large join state on GPUs with fast interconnects,” inProc. SIGMOD’22, 2022, pp. 1017–1032.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "J. Cao, R. Sen, M. Interlandi, J. Arulraj, and H. Kim, “GPU database systems characterization and optimization,” in Proc. VLDB’23, vol. 17, no. 3, 2023, pp. 441–454.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "J. Krueger, M. Grund, I. J aeckel, A. Zeier, and H. Plattner, “Applicability of GPU computing for efficient merge in in-memory databases,” in Proc. VLDB’11, 2011, pp. 19–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "HEAVY .AI, “HEAVY .AI For Your Industry: Learn how accelerated analytics can transform your enterprise.” A ccessed: Dec. 20, 2024. [Online]. Available: https://www.heavy.ai/industries",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "SQreamDB, “SQreamDB: Case Studies.” Accessed: Dec. 20, 2024. [Online]. Available: https://sqream.com/resources/?category=case-studies",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "K. D. Inc., “Kinetica for Automotive: Real-time analytics from vehicle sensor data.” Accessed: Dec. 20, 2024. [Online]. Available: https://www. kinetica.com/solutions/automotive/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Brytlyt, “Brytlyt for Energy, Oil and Gas: Extract the value from data to boost operation efficiency.” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://brytlyt.io/oil-and-gas/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "M. Tarek Ibn Ziad, S. Damani, A. Jaleel, S. W. Keckler, and M. Stephenson, “CuCatch: A debugging tool for efficiently catching memory safety violations in CUDA applications,” in Proc. PLDI, vol. 7, 2023, pp. 124– 147.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "M. Wu, Y. Ouyang, H. Zhou, L. Zhang, C. Liu, and Y. Zhang, “Simulee: Detecting CUDA synchronization bugs via memory-access modeling,” inProc. ICSE, 2020, pp. 937–948.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "F. R. Monteiro, E. H. d S. Alves, I. S. Silva, H. I. Ismail, L. C. Cordeiro, and E. B. de Lima Filho, “ESBMC-GPU a context-bounded model checking tool to verify CUDA programs,” Sci. Comput. Program. ,v o l. 152, pp. 63–69, Jan. 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Y. Zheng et al., “Finding bugs in gremlin-based graph database systems via randomized differential testing,” in Proc. ISSTA, 2022, pp. 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Y. Fu et al., “THANOs: DBMs bug detection via storage engine rotation based differential testing,” in Proc. ICSE, 2025, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Z. Cui et al., “Differentially testing database transactions for fun and profit,” in Proc. ASE, 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "J. Ba and M. Rigger, “Keep it simple: Testing databases via differential query plans,” in Proc. SIGMOD, 2024, pp. 38–52.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 17,
+      "text": "J. Wang et al., “D3: Differential testing of distributed deep learning with model generation,” IEEE Trans. Softw. Eng., vol. 51, no. 1, pp. 38–52, Jan. 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Z. Deng, G. Meng, K. Chen, T. Liu, L. Xiang, and C. Chen, “Differential testing of cross deep learning framework {APIs}: Revealing inconsistencies and vulnerabilities,” in Proc. USENIX Secur., 2023, pp. 7393–7410.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "W. M. McKeeman, “Differential testing for software,” Digit. Tech. J., vol. 10, no. 1, pp. 100–107, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "R. Agrawal, T. Imieli ´nski, and A. Swami, “Mining association rules between sets of items in large databases,” in Proc. SIGMOD, 1993, pp. 207–216.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "R. Agrawal and R. Srikant et al., “Fast algorithms for mining association rules,” in Proc. VLDB’94, vol. 1215, 1994, pp. 487–499.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "“Brytlyt white paper: Speed of thought analytics at scale,” 2024. Accessed: Dec. 20, 2024. [Online]. Available: https://www.brytlyt.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "“Kinetica on-premises documentation,” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://docs.kinetica.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "“Sqream DB architecture whitepaper: A next generation GPU powered analytics SQL database,” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://sqream.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "BlazingSQL, “BlazingSQL,” 2021. A ccessed: Dec. 20, 2024. [Online]. Available: https://github.com/BlazingDB/blazingsql",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Dask, “Dask-sql,” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://github.com/dask-contrib/dask-sql",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "K. Zhang et al., “Hetero-DB: next generation high-performance database systems by best utilizing heterogeneous computing and storage resources,” J. Comput. Sci. Technol., vol. 30, no. 4, pp. 657–678, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "R. Lee et al., “The art of balance: A rateupdb™ experience of building a CPU/GPU hybrid database product,” in Proc. VLDB, vol. 14, 2021, pp. 2999–3013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "K. Kohei, “PG-Strom,” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://github.com/heterodb/pg-strom",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "OmniSci, “OmniSci technical white paper: GPU-accelerated analyticsbig data analytics at speed and SCAL,” 2020. A ccessed: Dec. 20, 2024. [Online]. Available: https://www.omnisci.com/platform",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "C. Root and T. Mostak, “MAPD: AGPU-powered big data analytics and visualization platform,” in Proc. SIGGRAPH, 2016, pp. 1–2. A ccessed: Dec. 20, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "HEAVY .AI, “HeavyDB,” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://github.com/heavyai/heavydb",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "“DB-engines ranking,” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://db-engines.com/en/ranking",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "I. HeteroDB, “PG-Strom manual,” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://heterodb.github.io/pg-strom/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "C. Lattner and V. Adve, “Llvm: A compilation framework for lifelong program analysis & transformation,” CGO’04, 2004, pp. 75–86.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "NVIDIA, “Parallel thread execution ISA version 8.3,” 2024. Accessed: Dec. 20, 2024. [Online]. Available: https://docs.nvidia.com/cuda/parallel-thread-execution/index.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "HEAVY .AI, “HEAVY .AI documentation,” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://docs.heavy.ai/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "RAPIDS Development Team, “RAPIDS: Collection of libraries for end to end GPU data science,” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://rapids.ai",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "RAPIDS Development Team, “cu DF-GPU DataFrames,” 2024. Accessed: Dec. 20, 2024. [Online]. Available: https://github.com/rapidsai/cudf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "N. Braun, “Dask-sql Documentation,” 2024. A ccessed: Dec. 20, 2024. [Online]. Available: https://dask-sql.readthedocs.io/en/latest/index.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "J. Fürnkranz and T. Kliegr, “A brief overview of rule learning,” in Proc. Int. Symp. Rules Rule Markup Lang. Semantic Web, 2015, pp. 54–69.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "M. Santolucito, E. Zhai, R. Dhodapkar, A. Shim, and R. Piskac, “Synthesizing configuration file specifications with association rule learning,”inProc. OOPSLA, vol. 1, 2017, pp. 1–20.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "V. Leis, A. Gubichev, A. Mirchev, P. Boncz, A. Kemper, and T. Neumann, “How good are query optimizers, really?” in Proc .VLDB, vol. 9, no. 3, 2015, pp. 204–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "P. S. Foundation, “Trace — Trace or track Python statement execution,” 2024. [Online]. Available: https://docs.python.org/zh-cn/3/library/trace. html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "T. Parr, “ANTLR v4,” 2023. [Online]. Available: https://github.com/ antlr/antlr",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "T. Parr and K. Fisher, “Ll ( ∗) the foundation of the ANTLR parser generator,” ACM Sigplan Notices, vol. 46, no. 6, pp. 425–436, 2011. 2160 IEEETRANSACTIONS ONSOFTW AREENGINEERING, VOL. 51, NO. 7, JULY 2025",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "M. Rigger, “SQLancer,” 2024. [Online]. Available: https://github.com/ sqlancer/sqlancer",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 48,
+      "text": "A. Zeller, Why Programs Fail: A Guide to Systematic Debugging ,S a n Francisco, CA, USA: Morgan Kaufmann Publishers Inc. Oct 2005.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "NVIDIA, “NVIDIA GeForce RTX 4090 datasheet,” 2024. [Online]. Available: https://www.nvidia.com/en-us/geforce/graphics-cards/ 40-series/rtx-4090/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "NVIDIA, “NVIDIA Tesla V100 datasheet,” 2018. [Online]. Available: https://images.nvidia.com/content/technologies/volta/pdf/tesla-voltav100-datasheet-letter-fnl-web.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "A. Seltenreich, “SQLsmith: A random SQL query generator,” Mar. 2023. A ccessed Dec-2022. [Online]. Available: https://github.com/anse1/ sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proc. ESEC/FSE, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 53,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Oopsla’20, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 54,
+      "text": "K. N. Khan, M. Hirki, T. Niemi, J. K. Nurminen, and Z. Ou, “RAPL in action: Experiences in using RAPL for power measurements,” in Proc. TOMPECS, vol. 3, no. 2, 2018, pp. 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "G. Rodola, “psutil library,” 2025. [Online]. Available: https://pypi.org/ project/psutil/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "F. Cotter, “py3nvml library,” 2025. [Online]. Available: https://pypi.org/ project/py3nvml/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "NVIDIA, “NVTX (NVIDIA Tools Extension Library),” 2024. [Online]. Available: https://github.com/NVIDIA/NVTX",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "S. Meraji et al., “Towards a hybrid design for fast query processing in DB2 with Blu acceleration using graphical processing units: Atechnology demonstration,” in Proc. SIGMOD, 2016, pp. 1951–1960.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "PostgreSQL, “PostgreSQL: The world’s most advanced open source relational database,” 2023. [Online]. Available: https://www.postgresql.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "IBM, “IBM Db2 database,” 2023. [Online]. Available: https://www.ibm. com/products/db2/database",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "X. Tang, S. Wu, D. Zhang, F. Li, and G. Chen, “Detecting logic bugs of join optimizations in DBMs,” in Proc. SIGMOD, vol. 1, no. 1, 2023, pp. 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Z. Hao et al., “PINOLO: Detecting logical bugs in database management systems with approximate query synthesis,” in Proc. USENIX ATC, 2023, pp. 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "J. Liang et al., “Sequence-oriented DBMs fuzzing,” in Proc. ICDE, 2023, pp. 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Z.-M. Jiang and Z. Su, “Detecting logic bugs in database engines via equivalent expression transformation,” in Proc .USENIX OSDI, 2024, pp. 821–835.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proc. ISSTA, 2023, pp. 140–149.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 66,
+      "text": "Z. Zhuang, P. Li, P. Ma, W. Meng, and S. Wang, “Testing graph database systems via graph-aware metamorphic relations,” in Proc. VLDB ,v o l. 17, no. 4, 2023, pp. 836–848.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "R. Yang, Y. Zheng, L. Tang, W. Dou, W. Wang, and J. Wei, “Randomized differential testing of RDF stores,” in Proc. ICSE-Companion, 2023, pp. 136–140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "X. Zhu, S. Wen, S. Camtepe, and Y. Xiang, “Fuzzing: A survey for roadmap,” ACM Comput. Surv., vol. 54, no. 11s, pp. 1–36, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "T. Y. Chen et al., “Metamorphic testing: A review of challenges and opportun ities,” ACM Comput. Surv., vol. 51, no. 1, pp. 1–27, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "M. Jibson, “SQLsmith: randomized SQL testing in CockroachDB,” Cockroach Labs, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing database management systems with language validity and coverage feedback,” in Proc. CCS, 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammar-free DBMs fuzzing,” in Proc. ASE, 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Z.-M. Jiang, J.-J. Bai, and Z. Su, “DynSQL: Stateful fuzzing for database management systems with complex and valid SQL query generation,” inProc. USENIX Secur., 2023, pp. 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of DBMS with coverage-based guidance,” in Proc. USENIX Secur., 2022, pp. 4309– 4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "J. Liang, Z. Wu, J. Fu, Y. Bai, Q. Zhang, and Y. Jiang, “WingFuzz: Implementing continuous fuzzing for DBMSs,” in Proc. USENIX ATC, 2024, pp. 479–492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic detection and diagnosis of performance regressions in database systems,” inProc. VLDB, vol. 13, no. 1, 2019, pp. 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "J. Liang, Z. Wu, J. Fu, M. Wang, C. Sun, and Y. Jiang, “MOZI: Discovering DBMs bugs via configuration-based equivalent transformation,” in Proc. ICSE, 2024, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "T. Sotiropoulos, S. Chaliasos, V. A tlidakis, D. Mitr opoulos, and D. Spinellis, “Data-oriented differential testing of object-relational mapping systems,” in Proc. ICSE, 2021, pp. 1535–1547.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "M. A. Gulzar, Y. Zhu, and X. Han, “Perception and practices of differential testing,” in Proc. ICSE-SEIP, 2019, pp. 71–80.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Y. Chen, T. Su, and Z. Su, “Deep differential testing of JVM implementations,” in Proc. ICSE, 2019, pp. 1257–1268.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "I. Lima, J. Silva, B. Miranda, G. Pinto, and M. d’Amorim, “Exposing bugs in JavaScript engines through test transplantation and differential testing,” Softw. Qual. J., vol. 29, pp. 129–158, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "E. T. Barr, M. Harman, P. McMinn, M. Shahbaz, and S. Yoo, “The oracle problem in software testing: A survey,” IEEE Trans. Softw. Eng., vol. 41, no. 5, pp. 507–525, May 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "G. Li, P. Li, G. Sawaya, G. Gopalakrishnan, I. Ghosh, and S. P. Rajan, “GKLEE: Concolic verification and test generation for GPUs,” in Proc. PPoPP, 2012, pp. 215–224.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "C. Peng, “On the correctness of GPU programs,” in Proc. ISSTA, 2019, pp. 443–447.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "L. Bernhard, N. Schiller, M. Schloegel, N. Bars, and T. Holz, “DarthShader: Fuzzing WebGPU shader translators & compilers,” in Proc .CCS, 2024, pp. 690–704.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "J. Fürnkranz, D. Gamberger, and N. Lavra ˇc,Foundations of Rule Learning. New York, NY, USA: Springer Sci. & Bus. Media, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "L. Raedt, A. Dries, I. Thon, G. Broeck, and M. Verbeke, “Inducing probabilistic relational rules from probabilistic examples,” in Proc. IJCAI, Jul. 2015, pp. 1835–1842.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "F. Yang, Z. Yang, and W. W. Cohen, “Differentiable learning of logical rules for knowledge base reasoning,” in Proc. NeurIPS, vol. 30, 2017, pp. 2316–2325.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "R. Rawat, R. K. Chakrawarti, A. S. A. Raj, G. Mani, K. Chidambarathanu, and R. Bhardwaj, “Association rule learning for threat analysis using traffic analysis and packet filtering approach,” Int. J. Inf. Technol., Aug. 2023, pp. 3245–3255, doi: 10.1007/s41870-023-01353-0. Jiaxin Hu is currently working toward the Ph.D. degree with the School of Informatics, XiamenUniversity, China. His research focu",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 16,
+      "text": "J. Ba and M. Rigger, “Keep it simple: Testing databases via differential query plans,” in Proc. SIGMOD, 2024, pp. 38–52.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 47,
+      "text": "M. Rigger, “SQLancer,” 2024. [Online]. Available: https://github.com/ sqlancer/sqlancer",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 52,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proc. ESEC/FSE, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 53,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Oopsla’20, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 65,
+      "text": "M. Kamm, M. Rigger, C. Zhang, and Z. Su, “Testing graph database engines via query partitioning,” in Proc. ISSTA, 2023, pp. 140–149.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[16]",
+      "technique": "dqp",
+      "sentence": "Inspired by the widely adopted differential testing approach forbug detection in non-GPU DBMSs [13] ,[14] ,[15] ,[16] and GPU-related software [17] ,[18], we borrow its core concept of comparing the results of identical test cases executed ondifferent software systems to establish a test oracle [19] .",
+      "context_before": "361102, China (e-mail: hujiaxin@stu.xmu.edu.cn; wurongxin@xmu.edu.cn). Digital Object Identifier 10.1109/TSE.2025.3574328of GPU features, implementing GPU-related applications correctly is non-trivial, and thus cr itical bugs are inevitable [10], [11] ,[12]. Unfortunately, to the best of our knowledge, there has yet to be a systematic study to detect bugs in GPU DBMSs. Essentially, detecting GPU-related bugs in GPU DBMSs entails tackling two fundamental problems. The first problem is to come up with an effective test oracle that can verify whether a GPU DBMS behaves correctly for a given input.",
+      "context_after": "H o w ever, directly applying the aforementioned differential testing approaches may not be applicable in detecting GPU-related bugs in DBMSs. First, the existing approaches for non-GPU DBMSs are limited to detecting bugs in non-GPU-specific fea-tures [13] ,[14] ,[15] ,[16]. For example, Grand [13] leverages multiple Gremlin-based graph databases to verify the return results of same queries, thus inherently limiting its applicabilityto other database paradigms. DT2 [15] limits its effectiveness for identifying transaction-level bugs in databases. Similarly, THANOS [14] maintains its detection e",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3196,
+      "cited_reference": {
+        "number": 16,
+        "text": "J. Ba and M. Rigger, “Keep it simple: Testing databases via differential query plans,” in Proc. SIGMOD, 2024, pp. 38–52.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[16]",
+      "technique": "dqp",
+      "sentence": "First, the existing approaches for non-GPU DBMSs are limited to detecting bugs in non-GPU-specific fea-tures [13] ,[14] ,[15] ,[16].",
+      "context_before": "fundamental problems. The first problem is to come up with an effective test oracle that can verify whether a GPU DBMS behaves correctly for a given input. Inspired by the widely adopted differential testing approach forbug detection in non-GPU DBMSs [13] ,[14] ,[15] ,[16] and GPU-related software [17] ,[18], we borrow its core concept of comparing the results of identical test cases executed ondifferent software systems to establish a test oracle [19] .H o w ever, directly applying the aforementioned differential testing approaches may not be applicable in detecting GPU-related bugs in DBMSs.",
+      "context_after": "For example, Grand [13] leverages multiple Gremlin-based graph databases to verify the return results of same queries, thus inherently limiting its applicabilityto other database paradigms. DT2 [15] limits its effectiveness for identifying transaction-level bugs in databases. Similarly, THANOS [14] maintains its detection efficacy only when applied to databases incorporating multiple storage engines. Second, some studies on testing GPU-related software (e.g., deeplearning frameworks) crosscheck the execution results under different counterpart framework APIs [18] to find bugs. However, adapting",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3641,
+      "cited_reference": {
+        "number": 16,
+        "text": "J. Ba and M. Rigger, “Keep it simple: Testing databases via differential query plans,” in Proc. SIGMOD, 2024, pp. 38–52.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[47]",
+      "technique": null,
+      "sentence": "To identify unique GPU-related bugs, we imitated the statement-level and syntax-based reduction in SQLancer [47] to automatically simplify bug reports.",
+      "context_before": "or Learner. Thus, we leveraged Generator to generate 1 million groups of query statements with configurations for PG-Strom, and 100 thousand groups for HeavyDB and Dask-sql. Parser is constructed based on ANTLR4 version 4.12.0, and the threshold values tl,tfcandtfginLearner are set to 0.1, 0.9 and 0.5, respectively. The justification of the threshold selection will be discussed in §V-E. Finally, Learner mined 2,082, 1,378 and 617 pairs of differential execution rules from PG-Strom, HeavyDB and Dask-sql, respectively. After bug detection implementation, we obtained numerous duplicate bug reports.",
+      "context_after": "Then, for better distinction, we also borrowed ideas from delta debugging [48] to localize the root cause of simplified bug reports. If two bug reports belong to the same root cause, we will consider them duplicated. V. EVALUATION After presenting the design of SQLaw, we evaluate its effectiveness and efficiency in detecting GPU-related bugs in real-world GPU DBMSs. Our evaluation aims to address the following research questions: •RQ1: How many GPU-related bugs in real-world GPU DBMSs can be detected by SQLaw? •RQ2: Can SQLaw outperform existing state-of-the-art bug detection approaches? •RQ3:",
+      "section": "IV IMPLEMENTATION",
+      "page": 10,
+      "char_offset": 42520,
+      "cited_reference": {
+        "number": 47,
+        "text": "M. Rigger, “SQLancer,” 2024. [Online]. Available: https://github.com/ sqlancer/sqlancer",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "We selected three representative state-ofthe-art approaches, including SQLsmith ,NoREC and TLP .",
+      "context_before": "rating system with 6.8.0-31-generic Linux kernel. We deployed three GPUDBMSs for testing via Docker 26.1.1. GPU DBMSs Under Test. We selected three mainstream GPU DBMSs as evaluation subjects, including PG-Strom [29], HeavyDB [32] ,a n d Dask-sql [26], as described in Table I.W e chose the latest version of each GPU DBMS at the time of conducting our experiments. The version information is listed in Table III. Baselines and Ablations. Since SQLaw is the first tool for detecting bugs in GPU DBMSs, we can only adapt existingstate-of-the-art approaches developed for non-GPU DBMSs as the baselines.",
+      "context_after": "I n particular, SQLsmith is one of the most effective approaches for detecting crash and error bugs, while both NoREC and TLP represent advanced approaches for detecting logic bugs. Each of them has found a total of over a hundred previously unknown bugs in various non-GPU DBMSs [51] ,[52], [53]. We briefly introduce the methodologies of each baseline as follows. •SQLsmith [51] generates random query statements using an AST model and checks whether the DBMS works normally when executing them. •NoREC [52] translates query statements into nonoptimized queries in the optimizer and compares the res",
+      "section": "A Evaluation Step",
+      "page": 10,
+      "char_offset": 44347,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "I n particular, SQLsmith is one of the most effective approaches for detecting crash and error bugs, while both NoREC and TLP represent advanced approaches for detecting logic bugs.",
+      "context_before": "er 26.1.1. GPU DBMSs Under Test. We selected three mainstream GPU DBMSs as evaluation subjects, including PG-Strom [29], HeavyDB [32] ,a n d Dask-sql [26], as described in Table I.W e chose the latest version of each GPU DBMS at the time of conducting our experiments. The version information is listed in Table III. Baselines and Ablations. Since SQLaw is the first tool for detecting bugs in GPU DBMSs, we can only adapt existingstate-of-the-art approaches developed for non-GPU DBMSs as the baselines. We selected three representative state-ofthe-art approaches, including SQLsmith ,NoREC and TLP .",
+      "context_after": "Each of them has found a total of over a hundred previously unknown bugs in various non-GPU DBMSs [51] ,[52], [53]. We briefly introduce the methodologies of each baseline as follows. •SQLsmith [51] generates random query statements using an AST model and checks whether the DBMS works normally when executing them. •NoREC [52] translates query statements into nonoptimized queries in the optimizer and compares the results to identify logic bugs. •TLP [53] partitions an original query into three separated queries by decomposing its predicate, and triggers a logic bug when the union result of thes",
+      "section": "A Evaluation Step",
+      "page": 10,
+      "char_offset": 44444,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[52]",
+      "technique": "norec",
+      "sentence": "Each of them has found a total of over a hundred previously unknown bugs in various non-GPU DBMSs [51] ,[52], [53].",
+      "context_before": "e chose the latest version of each GPU DBMS at the time of conducting our experiments. The version information is listed in Table III. Baselines and Ablations. Since SQLaw is the first tool for detecting bugs in GPU DBMSs, we can only adapt existingstate-of-the-art approaches developed for non-GPU DBMSs as the baselines. We selected three representative state-ofthe-art approaches, including SQLsmith ,NoREC and TLP .I n particular, SQLsmith is one of the most effective approaches for detecting crash and error bugs, while both NoREC and TLP represent advanced approaches for detecting logic bugs.",
+      "context_after": "We briefly introduce the methodologies of each baseline as follows. •SQLsmith [51] generates random query statements using an AST model and checks whether the DBMS works normally when executing them. •NoREC [52] translates query statements into nonoptimized queries in the optimizer and compares the results to identify logic bugs. •TLP [53] partitions an original query into three separated queries by decomposing its predicate, and triggers a logic bug when the union result of these separated queries mis-matches the original one. Additionally, we adapted the above three approaches to detect GPU-",
+      "section": "A Evaluation Step",
+      "page": 10,
+      "char_offset": 44625,
+      "cited_reference": {
+        "number": 52,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proc. ESEC/FSE, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "•NoREC [52] translates query statements into nonoptimized queries in the optimizer and compares the results to identify logic bugs.",
+      "context_before": "elines. We selected three representative state-ofthe-art approaches, including SQLsmith ,NoREC and TLP .I n particular, SQLsmith is one of the most effective approaches for detecting crash and error bugs, while both NoREC and TLP represent advanced approaches for detecting logic bugs. Each of them has found a total of over a hundred previously unknown bugs in various non-GPU DBMSs [51] ,[52], [53]. We briefly introduce the methodologies of each baseline as follows. •SQLsmith [51] generates random query statements using an AST model and checks whether the DBMS works normally when executing them.",
+      "context_after": "•TLP [53] partitions an original query into three separated queries by decomposing its predicate, and triggers a logic bug when the union result of these separated queries mis-matches the original one. Additionally, we adapted the above three approaches to detect GPU-related bugs through two key modifications. First, we extended these three approaches to support the dialects of the three GPU DBMSs under test. Second, we introduced a configurationstep (following the pivotal configuration approach in §II-C ) to control query statements on GPU-used execution. Specifically, this step involves setting",
+      "section": "A Evaluation Step",
+      "page": 10,
+      "char_offset": 44940,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "•TLP [53] partitions an original query into three separated queries by decomposing its predicate, and triggers a logic bug when the union result of these separated queries mis-matches the original one.",
+      "context_before": "one of the most effective approaches for detecting crash and error bugs, while both NoREC and TLP represent advanced approaches for detecting logic bugs. Each of them has found a total of over a hundred previously unknown bugs in various non-GPU DBMSs [51] ,[52], [53]. We briefly introduce the methodologies of each baseline as follows. •SQLsmith [51] generates random query statements using an AST model and checks whether the DBMS works normally when executing them. •NoREC [52] translates query statements into nonoptimized queries in the optimizer and compares the results to identify logic bugs.",
+      "context_after": "Additionally, we adapted the above three approaches to detect GPU-related bugs through two key modifications. First, we extended these three approaches to support the dialects of the three GPU DBMSs under test. Second, we introduced a configurationstep (following the pivotal configuration approach in §II-C ) to control query statements on GPU-used execution. Specifically, this step involves setting all pivotal configuration optionsshown in Fig. 1for GPU-used execution before executing query statements. This increases the percentage of generated query statements executed on GPU devices compared to e",
+      "section": "A Evaluation Step",
+      "page": 10,
+      "char_offset": 45072,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Consequently, we refer to the adapted approaches as SQLsmith ∗,NoREC∗and TLP∗.",
+      "context_before": "bove three approaches to detect GPU-related bugs through two key modifications. First, we extended these three approaches to support the dialects of the three GPU DBMSs under test. Second, we introduced a configurationstep (following the pivotal configuration approach in §II-C ) to control query statements on GPU-used execution. Specifically, this step involves setting all pivotal configuration optionsshown in Fig. 1for GPU-used execution before executing query statements. This increases the percentage of generated query statements executed on GPU devices compared to executions without configuration.",
+      "context_after": "Since SQLaw represents a novel extension of differential testing, we compared its performance against a conventional differential testing approach in the ablation study ( §V-D ). This differential testing approach utilizes Generator in§III-B1 to generate query statements, adopts the pivotal configuration approach detailed in §II-C to control CPU-only and GPU-used execution, and compares their results to identify bugs. B. Effectiveness and Efficiency of SQLaw To evaluate the effectiveness and efficiency of SQLaw, we used it to detect GPU-related bugs in PG-Strom ,HeavyDB, and Dask-sql. As shown i",
+      "section": "A Evaluation Step",
+      "page": 10,
+      "char_offset": 45905,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "HU AND WU: SQLAW: DETECTING BUGS IN GPUDA TABASE MANAGEMENT SYSTEMS 2155 TABLE VIIIUNIQUE BUGSDETECTED INTHREE GPU DBMS SUNDER TESTWITHIN 24 HOURS DBMS SQLaw SQLsmith∗TLP∗NoREC∗ PG-Strom 22 10 7 5 HeavyDB 15 6 6 4 Dask-sql 14 7 4 6 Total 51 23 17 15 Improvement-122% ↑ 200%↑ 240%↑ C.",
+      "context_before": "statements with configurations ( Generator needs to verify the correctness of generated query statements with configurations via databases), thereby consuming less time. Overall, the time cost of offline rule learning for all three databases is under 1.4 hours. Notably,offline rule learning process is a one-time effort, unless there are major updates to GPU DBMSs under test. Despite its time cost, the generated differential execution rules can be reused during testing. It is marginally consumptive compared to 24-hour continuous testing as long as it improves bug-finding effectivenessand efficiency.",
+      "context_after": "Comparison With Baselines GPU-related Bug Detection. For a more comprehensive evaluation of SQLaw, we compared it with the three stateof-the-art approaches with GPU DBMSs adaptation, including SQLsmith∗, which is good at detecting GPU crash bugs and GPU error bugs, NoREC∗and TLP∗, which are proficient at finding GPU logic bugs. We ran all of the approaches with a time budget of 24 hours. As shown in Table VIII, for PG-Strom, SQLaw detected 22 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only found 10, 7, and 5 bugs, respectively. For HeavyDB, SQLaw detected 15 GPU-related bugs, while SQLsmith",
+      "section": "B Effectiveness and Efficiency of SQLaw",
+      "page": 11,
+      "char_offset": 51932,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For a more comprehensive evaluation of SQLaw, we compared it with the three stateof-the-art approaches with GPU DBMSs adaptation, including SQLsmith∗, which is good at detecting GPU crash bugs and GPU error bugs, NoREC∗and TLP∗, which are proficient at finding GPU logic bugs.",
+      "context_before": "or updates to GPU DBMSs under test. Despite its time cost, the generated differential execution rules can be reused during testing. It is marginally consumptive compared to 24-hour continuous testing as long as it improves bug-finding effectivenessand efficiency. HU AND WU: SQLAW: DETECTING BUGS IN GPUDA TABASE MANAGEMENT SYSTEMS 2155 TABLE VIIIUNIQUE BUGSDETECTED INTHREE GPU DBMS SUNDER TESTWITHIN 24 HOURS DBMS SQLaw SQLsmith∗TLP∗NoREC∗ PG-Strom 22 10 7 5 HeavyDB 15 6 6 4 Dask-sql 14 7 4 6 Total 51 23 17 15 Improvement-122% ↑ 200%↑ 240%↑ C. Comparison With Baselines GPU-related Bug Detection.",
+      "context_after": "We ran all of the approaches with a time budget of 24 hours. As shown in Table VIII, for PG-Strom, SQLaw detected 22 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only found 10, 7, and 5 bugs, respectively. For HeavyDB, SQLaw detected 15 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only discovered 6, 6, 4 bugs, respectively. For Dask-sql, SQLaw detected 14 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only detected 7, 4, 6 bugs, respectively. In total, the number of GPUrelated bugs detected by SQLaw is 122% to 240% more thanthe one by the baseline approaches. In terms of bug types, SQLa",
+      "section": "C Comparison With Baselines",
+      "page": 12,
+      "char_offset": 52271,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "As shown in Table VIII, for PG-Strom, SQLaw detected 22 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only found 10, 7, and 5 bugs, respectively.",
+      "context_before": "TABLE VIIIUNIQUE BUGSDETECTED INTHREE GPU DBMS SUNDER TESTWITHIN 24 HOURS DBMS SQLaw SQLsmith∗TLP∗NoREC∗ PG-Strom 22 10 7 5 HeavyDB 15 6 6 4 Dask-sql 14 7 4 6 Total 51 23 17 15 Improvement-122% ↑ 200%↑ 240%↑ C. Comparison With Baselines GPU-related Bug Detection. For a more comprehensive evaluation of SQLaw, we compared it with the three stateof-the-art approaches with GPU DBMSs adaptation, including SQLsmith∗, which is good at detecting GPU crash bugs and GPU error bugs, NoREC∗and TLP∗, which are proficient at finding GPU logic bugs. We ran all of the approaches with a time budget of 24 hours.",
+      "context_after": "For HeavyDB, SQLaw detected 15 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only discovered 6, 6, 4 bugs, respectively. For Dask-sql, SQLaw detected 14 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only detected 7, 4, 6 bugs, respectively. In total, the number of GPUrelated bugs detected by SQLaw is 122% to 240% more thanthe one by the baseline approaches. In terms of bug types, SQLaw found 11, 6 and 7 GPU crash and error bugs in PG-Strom ,HeavyDB and Dask-sql, respectively. It covered 10, 6 and 7 of GPU crash and error bugs found by SQLsmith ∗. SQLaw detected 11, 9 and 7 GPU logic bugs",
+      "section": "C Comparison With Baselines",
+      "page": 12,
+      "char_offset": 52607,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For HeavyDB, SQLaw detected 15 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only discovered 6, 6, 4 bugs, respectively.",
+      "context_before": "sql 14 7 4 6 Total 51 23 17 15 Improvement-122% ↑ 200%↑ 240%↑ C. Comparison With Baselines GPU-related Bug Detection. For a more comprehensive evaluation of SQLaw, we compared it with the three stateof-the-art approaches with GPU DBMSs adaptation, including SQLsmith∗, which is good at detecting GPU crash bugs and GPU error bugs, NoREC∗and TLP∗, which are proficient at finding GPU logic bugs. We ran all of the approaches with a time budget of 24 hours. As shown in Table VIII, for PG-Strom, SQLaw detected 22 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only found 10, 7, and 5 bugs, respectively.",
+      "context_after": "For Dask-sql, SQLaw detected 14 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only detected 7, 4, 6 bugs, respectively. In total, the number of GPUrelated bugs detected by SQLaw is 122% to 240% more thanthe one by the baseline approaches. In terms of bug types, SQLaw found 11, 6 and 7 GPU crash and error bugs in PG-Strom ,HeavyDB and Dask-sql, respectively. It covered 10, 6 and 7 of GPU crash and error bugs found by SQLsmith ∗. SQLaw detected 11, 9 and 7 GPU logic bugs in PG-Strom ,HeavyDB and Dask-sql, respectively. It is considerably more than 2 and 1 GPU logic bugs detected by TLP∗in PG-S",
+      "section": "C Comparison With Baselines",
+      "page": 12,
+      "char_offset": 52754,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For Dask-sql, SQLaw detected 14 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only detected 7, 4, 6 bugs, respectively.",
+      "context_before": "more comprehensive evaluation of SQLaw, we compared it with the three stateof-the-art approaches with GPU DBMSs adaptation, including SQLsmith∗, which is good at detecting GPU crash bugs and GPU error bugs, NoREC∗and TLP∗, which are proficient at finding GPU logic bugs. We ran all of the approaches with a time budget of 24 hours. As shown in Table VIII, for PG-Strom, SQLaw detected 22 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only found 10, 7, and 5 bugs, respectively. For HeavyDB, SQLaw detected 15 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only discovered 6, 6, 4 bugs, respectively.",
+      "context_after": "In total, the number of GPUrelated bugs detected by SQLaw is 122% to 240% more thanthe one by the baseline approaches. In terms of bug types, SQLaw found 11, 6 and 7 GPU crash and error bugs in PG-Strom ,HeavyDB and Dask-sql, respectively. It covered 10, 6 and 7 of GPU crash and error bugs found by SQLsmith ∗. SQLaw detected 11, 9 and 7 GPU logic bugs in PG-Strom ,HeavyDB and Dask-sql, respectively. It is considerably more than 2 and 1 GPU logic bugs detected by TLP∗in PG-Strom and HeavyDB, and only 1 GPU logic bug detected by NoREC∗in PG-Strom .F u r t h e r m o r e, TLP∗cannot find any GPU log",
+      "section": "C Comparison With Baselines",
+      "page": 12,
+      "char_offset": 52877,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "It is considerably more than 2 and 1 GPU logic bugs detected by TLP∗in PG-Strom and HeavyDB, and only 1 GPU logic bug detected by NoREC∗in PG-Strom .",
+      "context_before": ", while SQLsmith∗,TLP∗and NoREC∗only discovered 6, 6, 4 bugs, respectively. For Dask-sql, SQLaw detected 14 GPU-related bugs, while SQLsmith∗,TLP∗and NoREC∗only detected 7, 4, 6 bugs, respectively. In total, the number of GPUrelated bugs detected by SQLaw is 122% to 240% more thanthe one by the baseline approaches. In terms of bug types, SQLaw found 11, 6 and 7 GPU crash and error bugs in PG-Strom ,HeavyDB and Dask-sql, respectively. It covered 10, 6 and 7 of GPU crash and error bugs found by SQLsmith ∗. SQLaw detected 11, 9 and 7 GPU logic bugs in PG-Strom ,HeavyDB and Dask-sql, respectively.",
+      "context_after": "F u r t h e r m o r e, TLP∗cannot find any GPU logic bugs in Dask-sql. Neither can NoREC∗find any GPU logic bugs in either HeavyDB orDask-sql. Overall, SQLaw outperforms corresponding state-of-the-art approaches for detecting crash and GPU error bugs as well as GPU logicbugs. These results demonstrate the superiority of SQLaw in terms of GPU-related bug detection. Answer to RQ2: Compared with the existing state-of-theart approaches with GPU DBMSs adaptation, SQLaw candetect 122% to 240% more GPU-related bugs in three GPU DBMSs under test. Testing Time Budget for SQLaw. In most cases, GPU DBMS fe",
+      "section": "C Comparison With Baselines",
+      "page": 12,
+      "char_offset": 53402,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "F u r t h e r m o r e, TLP∗cannot find any GPU logic bugs in Dask-sql.",
+      "context_before": "d NoREC∗only detected 7, 4, 6 bugs, respectively. In total, the number of GPUrelated bugs detected by SQLaw is 122% to 240% more thanthe one by the baseline approaches. In terms of bug types, SQLaw found 11, 6 and 7 GPU crash and error bugs in PG-Strom ,HeavyDB and Dask-sql, respectively. It covered 10, 6 and 7 of GPU crash and error bugs found by SQLsmith ∗. SQLaw detected 11, 9 and 7 GPU logic bugs in PG-Strom ,HeavyDB and Dask-sql, respectively. It is considerably more than 2 and 1 GPU logic bugs detected by TLP∗in PG-Strom and HeavyDB, and only 1 GPU logic bug detected by NoREC∗in PG-Strom .",
+      "context_after": "Neither can NoREC∗find any GPU logic bugs in either HeavyDB orDask-sql. Overall, SQLaw outperforms corresponding state-of-the-art approaches for detecting crash and GPU error bugs as well as GPU logicbugs. These results demonstrate the superiority of SQLaw in terms of GPU-related bug detection. Answer to RQ2: Compared with the existing state-of-theart approaches with GPU DBMSs adaptation, SQLaw candetect 122% to 240% more GPU-related bugs in three GPU DBMSs under test. Testing Time Budget for SQLaw. In most cases, GPU DBMS features remain stable, allowing rules learned during the offline proces",
+      "section": "C Comparison With Baselines",
+      "page": 12,
+      "char_offset": 53550,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ],
+      "text_is_letter_spaced": true
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Neither can NoREC∗find any GPU logic bugs in either HeavyDB orDask-sql.",
+      "context_before": "r of GPUrelated bugs detected by SQLaw is 122% to 240% more thanthe one by the baseline approaches. In terms of bug types, SQLaw found 11, 6 and 7 GPU crash and error bugs in PG-Strom ,HeavyDB and Dask-sql, respectively. It covered 10, 6 and 7 of GPU crash and error bugs found by SQLsmith ∗. SQLaw detected 11, 9 and 7 GPU logic bugs in PG-Strom ,HeavyDB and Dask-sql, respectively. It is considerably more than 2 and 1 GPU logic bugs detected by TLP∗in PG-Strom and HeavyDB, and only 1 GPU logic bug detected by NoREC∗in PG-Strom .F u r t h e r m o r e, TLP∗cannot find any GPU logic bugs in Dask-sql.",
+      "context_after": "Overall, SQLaw outperforms corresponding state-of-the-art approaches for detecting crash and GPU error bugs as well as GPU logicbugs. These results demonstrate the superiority of SQLaw in terms of GPU-related bug detection. Answer to RQ2: Compared with the existing state-of-theart approaches with GPU DBMSs adaptation, SQLaw candetect 122% to 240% more GPU-related bugs in three GPU DBMSs under test. Testing Time Budget for SQLaw. In most cases, GPU DBMS features remain stable, allowing rules learned during the offline process to be reused effectively. However, when new featuresare introduced, S",
+      "section": "C Comparison With Baselines",
+      "page": 12,
+      "char_offset": 53619,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "It should be notedthat SQLaw can only begin bug detection after completing the TABLE IXPOWER ANDMEMORY CONSUMPTION ,INCLUDINGS AVERAGE CPU POWER ,AVERAGE CPU MEMORY USAGE ,AVERAGE GPUPOWER ANDAVERAGE GPU MEMORY USAGE DBMS PG-Strom HeavyDB Dask-sql SQLaw 40W 457MB 58W 1222MB32W 1130MB 79W 2551MB62W 1407MB 81W 1980MB SQLsmith∗ 19W 439MB 44W 1222MB29W 685MB 77W 1986MB62W 1324MB 80W 1980MB TLP∗ 39W 442MB 53W 1222MB30W 1060MB 74W2091MB61W 1224MB 79W 1957MB NoREC∗ 36W 434MB 55W 1222MB31W 1012MB 72W 2412MB61W 1306MB 79W 1965MB offline process, causing it to initially lag behind the baselines.",
+      "context_before": "-theart approaches with GPU DBMSs adaptation, SQLaw candetect 122% to 240% more GPU-related bugs in three GPU DBMSs under test. Testing Time Budget for SQLaw. In most cases, GPU DBMS features remain stable, allowing rules learned during the offline process to be reused effectively. However, when new featuresare introduced, SQLaw must perform offline rule learning to adapt to these changes, which inevitably incurs additional time overhead. To ensure fair evaluation under such circumstances, we compare SQLaw with the baseline approaches by taking the time cost of offline rule learning into account.",
+      "context_after": "Nevertheless, it requires only 1 hour, 2 hours, and 2 hoursto outperform the best baseline in PG-strom ,HeavyDB ,a n d Dask-sql, respectively. The corresponding offline rule learning durations are 0.35 hours, 1.1 hours, and 1.38 hours for these three systems. Power and Memory Consumption. To quantify energy and resource consumption of SQLaw and the baselines during 24hour testing, we used the Intel RAPL [54] interface to measure average CPU power, while utilizing the library psutil [55] to collect average CPU memory usage. Additionally, average GPU power and memory usage were monitored through",
+      "section": "C Comparison With Baselines",
+      "page": 12,
+      "char_offset": 54565,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "citation_marker",
+      "surface": "[52]",
+      "technique": "norec",
+      "sentence": "In terms of DBMS types, significant research efforts have been devoted to relational DBMSs [14] ,[52] ,[53] ,[61] ,[62] ,[63] ,[64] ,g r a p hD B M S s [13] ,[65] ,[66] ,a n dR D Fs t o r e s [67].",
+      "context_before": "e of the GPU and increases the speed of data processing by having all the data reside in memory. The non-memory GPU DBMSs, including Dask-sql [26] ,BlazingSQL [25] ,SQreamDB [24], etc., leverages the GPU to perform all or most of the database relational operations, allowing for the analysis of larger data volumes. In aword, these GPU DBMSs under test are well represented. With the emergence of more and more GPU DBMSs, we believe our technique will provide important values in the reliability of these DBMSs. DBMS Testing. Recent research studies on DBMS testing focus on various aspects of DBMSs.",
+      "context_after": "Different from the prior studies, SQLaw focuses on detecting GPU-related bugs in GPU DBMSs, a brandly new type of DBMSs in terms of the execution processor unit. With respect to methodologies for DBMS testing, existing research can be primarily categorized into three groups: DBMS fuzzing [68], metamorphic testing [69] and differential testing [19]. DBMS Fuzzing. DBMS fuzzing approaches generate test cases and verify whether the DBMS operates normally during executing them to identify bugs [68] .SQLsmith [70] is a leading grammar-based DBMS fuzzer that ra ndomly generates SQL queries to test v",
+      "section": "VI RELATED WORK",
+      "page": 15,
+      "char_offset": 66621,
+      "cited_reference": {
+        "number": 52,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proc. ESEC/FSE, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "citation_marker",
+      "surface": "[52]",
+      "technique": "norec",
+      "sentence": "Many studies [52] ,[53] ,[61] ,[62], [64] ,[65] ,[66] employ metamorphic testing to evaluate DBMSs.",
+      "context_before": "generates SQL queries to test various DBMSs. SQUIRREL [71] ,Griffin [72], LEGO [63] ,DynSQL [73] ,SQLRight [74] ,a n d WINGFUZZ [75] are mutation-based fuzzers that u tilize coverage information to guide mutation from original queries to new ones,thereby generating diverse queries to test DBMSs. In contrast to these approaches, we employ an AST-based fuzzer to ra ndomly generate SQL datasets for mining differential execution rules. This helps SQLaw to generate query statements with configurations capable of expected execution,thereby detect GPU-related bugs more effectively. Metamorphic Testing.",
+      "context_after": "Metamorphic testing verifies SQL statement results against predefined rules to detect bugs [69] .TLP (Ternary Logic Partitioning) is an effective metamorphic relation for detecting logic bugs in both relational DBMSs [53] and graph DBMSs [65]. To uncover logic bugs in relational DBMSs, NoREC [52] verifies the equivalence between original queriesand their non-optimized translations; PINOLO [62] employs the approximation relation; TQS [61] utilizes join optimization relations; and EET [64] constructs equivalent transformations through expression-level manipulation. Add itionally, GAMERA [66] lever",
+      "section": "VI RELATED WORK",
+      "page": 15,
+      "char_offset": 67987,
+      "cited_reference": {
+        "number": 52,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proc. ESEC/FSE, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP (Ternary Logic Partitioning) is an effective metamorphic relation for detecting logic bugs in both relational DBMSs [53] and graph DBMSs [65].",
+      "context_before": "guide mutation from original queries to new ones,thereby generating diverse queries to test DBMSs. In contrast to these approaches, we employ an AST-based fuzzer to ra ndomly generate SQL datasets for mining differential execution rules. This helps SQLaw to generate query statements with configurations capable of expected execution,thereby detect GPU-related bugs more effectively. Metamorphic Testing. Many studies [52] ,[53] ,[61] ,[62], [64] ,[65] ,[66] employ metamorphic testing to evaluate DBMSs. Metamorphic testing verifies SQL statement results against predefined rules to detect bugs [69] .",
+      "context_after": "To uncover logic bugs in relational DBMSs, NoREC [52] verifies the equivalence between original queriesand their non-optimized translations; PINOLO [62] employs the approximation relation; TQS [61] utilizes join optimization relations; and EET [64] constructs equivalent transformations through expression-level manipulation. Add itionally, GAMERA [66] leverages graph-aware metamorphic relations to detect logic bugs in graph DBMSs. These approaches rely on customized relations or rules that are less tailored to the characteristics of GPU DBMSs. This limits their direct applicabilityin effectivel",
+      "section": "VI RELATED WORK",
+      "page": 15,
+      "char_offset": 68183,
+      "found_by_all": [
+        "technique",
+        "citation_marker",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "To uncover logic bugs in relational DBMSs, NoREC [52] verifies the equivalence between original queriesand their non-optimized translations; PINOLO [62] employs the approximation relation; TQS [61] utilizes join optimization relations; and EET [64] constructs equivalent transformations through expression-level manipulation.",
+      "context_before": "AST-based fuzzer to ra ndomly generate SQL datasets for mining differential execution rules. This helps SQLaw to generate query statements with configurations capable of expected execution,thereby detect GPU-related bugs more effectively. Metamorphic Testing. Many studies [52] ,[53] ,[61] ,[62], [64] ,[65] ,[66] employ metamorphic testing to evaluate DBMSs. Metamorphic testing verifies SQL statement results against predefined rules to detect bugs [69] .TLP (Ternary Logic Partitioning) is an effective metamorphic relation for detecting logic bugs in both relational DBMSs [53] and graph DBMSs [65].",
+      "context_after": "Add itionally, GAMERA [66] leverages graph-aware metamorphic relations to detect logic bugs in graph DBMSs. These approaches rely on customized relations or rules that are less tailored to the characteristics of GPU DBMSs. This limits their direct applicabilityin effectively detecting GPU-related bugs. Differential Testing. Differential testing compares the results of identical test cases executed on different software systemsor versions to identify inconsistencies, correctness bugs and performance issues [19]. This technique is widely used in DBMS testing. For example, DT2 [15] identifies tra",
+      "section": "VI RELATED WORK",
+      "page": 15,
+      "char_offset": 68329,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "To detect correctness bugs, Grand [13] and RD2 [67] employ randomized differential testing for multiple graph DBMSs and RDF stores, respectively; THANOS [14] compares test cases on differential storage engines; and DQP [16] leverages differential query plans for simple yet effective bug detection.",
+      "context_before": "graph DBMSs. These approaches rely on customized relations or rules that are less tailored to the characteristics of GPU DBMSs. This limits their direct applicabilityin effectively detecting GPU-related bugs. Differential Testing. Differential testing compares the results of identical test cases executed on different software systemsor versions to identify inconsistencies, correctness bugs and performance issues [19]. This technique is widely used in DBMS testing. For example, DT2 [15] identifies transaction discrepancies across multiple DBMSs by comparing the results of identical transactions.",
+      "context_after": "APOLLO [76] detects performance issues by comparing SQL execution speeds across different DBMS versions. Moreover, Mozi [77] uncovers both performance and correctness bugs by altering DBMS configurations to createvaried system states for comparison. Beyond DBMS testing, differential testing is widely applied to test various software applications, such as ORM frameworks[78], compilers [79] ,J V M s [80], JavaScript engines [81] and search engines [82]. Unlike previous studies, SQLaw can be considered a novel extension of differential testing, since it combines rule learning, mutation, and diffe",
+      "section": "VI RELATED WORK",
+      "page": 15,
+      "char_offset": 69350,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "extends_technique": [
+        "M9"
+      ],
+      "describes_as_state_of_the_art": [
+        "M5"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:14:29Z",
+    "is_model_written": true,
+    "summary": "SQLaw tests GPU-based DBMSs, which the authors say had received no systematic bug-detection effort despite growing interest. It combines offline rule learning, which automatically extracts differential execution rules used to guide synthesis of configurations and queries, with an online interpreter that mutates those statements to generalise them. Evaluated on three major GPU DBMSs it outperformed state-of-the-art approaches by up to 2.22x and detected 51 previously unknown GPU-related bugs, 37 confirmed or fixed.",
+    "narrative": "SQLaw imitates SQLancer's statement-level, syntax-based reduction to simplify its own bug reports, and takes NoREC and TLP as two of its three baselines, describing them as representative state-of-the-art approaches that have each found over a hundred bugs in non-GPU DBMSs.",
+    "roles": {
+      "M1": "background",
+      "M2": "motivation",
+      "M3": "reuse_component",
+      "M4": "state_of_the_art",
+      "M5": "state_of_the_art",
+      "M6": "incidental",
+      "M7": "definition",
+      "M8": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "uncertain",
+        "mention_ids": [
+          "M3"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "To identify unique GPU-related bugs, we imitated the statement-level and syntax-based reduction in SQLancer [47] to automatically simplify bug reports.",
+            "section": "IV IMPLEMENTATION",
+            "page": 10
+          }
+        ],
+        "reasoning": "M3 says they imitated SQLancer's statement-level, syntax-based reduction to simplify test cases. Imitating a technique is not necessarily reusing the code, and no mention read settles which.",
+        "reuse_kind": "unclear"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "A pattern fired on M9, but the rule-based differential execution SQLaw contributes is its own; NoREC and TLP are described as baselines."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M7",
+          "M8"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "We selected three representative state-ofthe-art approaches, including SQLsmith ,NoREC and TLP .",
+            "section": "A Evaluation Step",
+            "page": 10
+          },
+          {
+            "mention_id": "M7",
+            "sentence": "•NoREC [52] translates query statements into nonoptimized queries in the optimizer and compares the results to identify logic bugs.",
+            "section": "A Evaluation Step",
+            "page": 10
+          },
+          {
+            "mention_id": "M8",
+            "sentence": "•TLP [53] partitions an original query into three separated queries by decomposing its predicate, and triggers a logic bug when the union result of these separated queries mis-matches the original one.",
+            "section": "A Evaluation Step",
+            "page": 10
+          }
+        ],
+        "reasoning": "NoREC and TLP are two of the three representative approaches selected for comparison.",
+        "techniques": [
+          "norec",
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "We selected three representative state-ofthe-art approaches, including SQLsmith ,NoREC and TLP .",
+            "section": "A Evaluation Step",
+            "page": 10
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "I n particular, SQLsmith is one of the most effective approaches for detecting crash and error bugs, while both NoREC and TLP represent advanced approaches for detecting logic bugs.",
+            "section": "A Evaluation Step",
+            "page": 10
+          }
+        ],
+        "reasoning": "M4 calls the three selected approaches state-of-the-art and M5 says NoREC and TLP represent the most effective for logic bugs."
+      }
+    },
+    "disagreements": [
+      "An extends_technique pattern fired on M9, which describes a baseline rather than a contribution."
+    ],
+    "unresolved": [
+      "Whether SQLaw reuses SQLancer's reduction code or reimplements the idea."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1109_tse_2025_3625300.json b/_data/papers/paper_doi_10_1109_tse_2025_3625300.json
new file mode 100644
index 0000000..c476f2c
--- /dev/null
+++ b/_data/papers/paper_doi_10_1109_tse_2025_3625300.json
@@ -0,0 +1,861 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:24:17Z",
+  "paper": {
+    "id": "paper:doi:10.1109/tse.2025.3625300",
+    "title": "A Comprehensive Study of Bugs in Relational DBMS",
+    "authors": [
+      "Shuang Liu",
+      "Ruifeng Wang",
+      "Yuanfeng Xie",
+      "Junjie Chen",
+      "Wei Lu",
+      "Xiao Zhang",
+      "Quanqing Xu",
+      "Chuanhui Yang",
+      "Xiaoyong Du"
+    ],
+    "year": 2025,
+    "venue": "IEEE Transactions on Software Engineering",
+    "doi": "10.1109/tse.2025.3625300",
+    "arxiv_id": null,
+    "s2_paper_id": "616ddf130a678c2727c4c8bc791d8e3a1eaf3bf7",
+    "url": "https://doi.org/10.1109/tse.2025.3625300",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1109/tse.2025.3625300",
+      "retrieved_at": "2026-09-08T15:24:17Z",
+      "chars": 76112,
+      "content_sha256": "sha256:1d0456aa15142a34bb57be6f07961b18f42ac7dc4b8b167b181ffa8718b44683"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1428
+      },
+      {
+        "number": "II",
+        "title": "METHODOLOGY ANDCLASSIFICATION",
+        "start": 9219
+      },
+      {
+        "number": "A",
+        "title": "Research Questions",
+        "start": 9253
+      },
+      {
+        "number": "B",
+        "title": "Databases to be Studied",
+        "start": 10933
+      },
+      {
+        "number": "C",
+        "title": "Bug Collection",
+        "start": 14606
+      },
+      {
+        "number": "D",
+        "title": "Bug Classification Method",
+        "start": 16477
+      },
+      {
+        "number": "III",
+        "title": "RESULTS ANDANALYSIS",
+        "start": 26733
+      },
+      {
+        "number": "A",
+        "title": "RQ1: Root Causes",
+        "start": 26858
+      },
+      {
+        "number": "B",
+        "title": "RQ2: Bug Symptoms",
+        "start": 39748
+      },
+      {
+        "number": "C",
+        "title": "RQ3: Bugs Distribution in Modules",
+        "start": 43020
+      },
+      {
+        "number": "D",
+        "title": "RQ4: Correlations",
+        "start": 47629
+      },
+      {
+        "number": "E",
+        "title": "Bug Distributions Across Studied RDBMSs",
+        "start": 51446
+      },
+      {
+        "number": "F",
+        "title": "Bug-Triggering SQLs",
+        "start": 52927
+      },
+      {
+        "number": "IV",
+        "title": "DISCUSSION",
+        "start": 56688
+      },
+      {
+        "number": "A",
+        "title": "Implications",
+        "start": 56881
+      },
+      {
+        "number": "B",
+        "title": "Threats to Validity",
+        "start": 60027
+      },
+      {
+        "number": "V",
+        "title": "APOOF-OF- CONCEPT TOOLSQLT",
+        "start": 61405
+      },
+      {
+        "number": "B",
+        "title": "Evaluation of SQLT",
+        "start": 64018
+      },
+      {
+        "number": "VI",
+        "title": "RELATED WORK",
+        "start": 65986
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 68669
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“MySQL”, MySQL. Accessed: Dec. 2023. [Online]. Available: https:// www.mysql.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“SQLite,” SQLite. Accessed: Dec. 2023. [Online]. Available: https:// sqlite.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“PostgreSQL,” PostgreSQL. A ccessed: Dec. 2023. [Online]. Available: https://www.postgresql.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "“OpenGauss,” OpenGauss. Acce ssed: Dec. 2023. [Online]. Available: https://opengauss.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "“WelcomeNext,” WelcomeNext. Accessed: Jun. 2024. [Online]. Available: https://welcomenext.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "“Guard.Me,” Guard.Me. Acce ssed: Jun. 2024. [Online]. Available: https://www.guard.me/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "“eBay,” eBay. Accessed: Jun. 2024. [Online]. Availab le: http://www. ebay.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "“GitHub,” GitHub. Accessed: J un. 2024. [Online]. Available: http:// www.github.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "“YouTube,” YouTube. Accessed: Jun. 2024. [Online]. Available: http:// www.youtube.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "J. M. Hellerstein, M. Stonebraker, J. Hamilton, “Architecture of a database system,” Found. Trends® Databases, vol. 1, no. 2, pp. 141–259, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "“SQLsmith,” SQLsmith. Accessed: Dec. 2023. [Online]. Available: https://github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proc. 14th USENIX Symp. Operating Syst. Des. Implementation (OSDI), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 13,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing database management sys tems with language validity and coverage feedback,” in Proc. ACMSIGSAC Conf. Computer Commun. Secur., 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proc. 28th ACM Joint Meeting Eur. Softw. Eng. Conf. Symp. Found. Softw. Eng., 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of DBMS with coverage-based guidance,” in Proc. 31st USENIX Secur. Symp. (USENIX Secur.), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, an d Y. Jiang, “Griffin: Grammar-free DBMS fuzzing,” in Proc. 37th IEEE/ACM Int. Conf. Automated Softw. Eng., 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "X. Liu, Q. Zhou, J. Arulraj, and A. Orso, “Automatic detection of performance bugs in database system s using equivalent queries,” in Proc. 44th Int. Conf. Softw. Eng., 2022, pp. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “Apollo: Automatic detection and diagnosis of performance regressions in database systems,” inProc. VLDB Endowment, vol. 13, no. 1, pp. 57–70. 2019. 3668 IEEETRANSACTIONS ONSOFTWARE ENGINEERING, VOL. 51, NO. 12, DECEMBER 2025",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 1– 30. 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "W. Dou et al., “Detecting isolation b ugs via transaction oracle construction,” in Proc. Int. Conf. Softw. Eng. (ICSE), 2023, pp. 1123–1135.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Z.-M. Jiang, S. Liu, M. Rigge r, and Z. Su, “Detecting transactional bugs in database engines via graphbased oracle construction,” in Proc. 17th USENIX Symp. Operating Sys t. Des. Implemen tation (OSDI), 2023, pp. 397–417.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "P. Fonseca, C. Li, V. Singhal, and R. Rodrigues, “A study of the internal and external effects of concurrency bugs,” in Proc. IEEE/IFIP Int. Conf. Dependable Syst. Netw. (DSN), Piscataway, NJ, USA: IEEE Press, 2010, pp. 221–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Z. Cui et al., “Understanding tr ansaction bugs in database systems,” in Proc. IEEE/ACM 46th Int. Conf. Softw. Eng. (ICSE), Los Alamitos, CA, USA: IEEE Comput. Soc. Press, 2024, pp. 947–947.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "“DB-engines ranking,” DB-engi nes ranking. Accessed: Aug. 2024. [Online]. Available: https://dbengines.com/en/ranking",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Z. Cui et al., “Differentially tes ting database transactions for fun and profit,” in Proc. 37th IEEE/ACM Int. Conf. Automated Softw. Eng., 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "J. Ba and M. Rigger, “CERT: Finding performance issues in database systems through the lens of cardinality estimation,” in Proc. IEEE/ACM 46th Int. Conf. Softw. Eng., 2024, pp. 1–13.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "A. Bovenzi, D. Cotroneo, R. Pietrantuono, and S. Russo, “On the aging effects due to concurrency bugs: A case study on MySQL,” in Proc. IEEE 23rd Int. Symp. Softw. Reliab. Eng., Piscataway, NJ, USA: IEEE Press, 2012, pp. 211–220.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Q. Shen, H. Ma, J. Chen, Y. Tian, S.-C. Cheung, and X. Chen, “A comprehensive study of deep learning compiler bugs,” in Proc. 29th ACM Joint Meet. Eur. Softw. Eng. Conf. Symp. Found. Softw. Eng., 2021, pp. 968–980.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "J. Chen, Y. Liang, Q. Shen, J. Jiang, and S. Li, “Toward understanding deep learning framework bugs,” ACM Trans. Softw. Eng. Method. ,v o l. 32, no. 6, pp. 1–31, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "J. Garcia et al., “A comprehensive study of autonomous vehicle bugs,” inProc. ACM/IEEE 42nd Int. Conf. Softw. Eng., 2020, pp. 385–396.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "H. Ma, W. Zhang, Q. Shen, Y. Tian, J. Chen, and S.-C. Cheung, “Towards understanding the bugs in solidity compiler,” 2024,arXiv:2407.05981.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "“SQLite Bugs,” SQLite Bugs. Accessed: Dec. 2023. [Online]. Available: https://sqlite.org/ src/rptview/1",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "“MySQL commits,” MySQL commits. Accessed: Dec. 2023. [Online]. Available: https://github.com/m ysql/mysql-server/commits/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "“OpenGauss Bugs,” OpenGauss Bugs. Accessed: Dec. 2023. [Online]. Available: https://gitee.com/openg auss/openGauss-server/issues",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "S. M. Vieira, U. Kaymak, and J. M. Sousa, “Cohen’s kappa coefficient as a performance measure for feature selection,” in Proc. Int. Conf. Fuzzy Syst., Piscataway, NJ, USA: IEEE Press, 2010, pp. 1–8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "C. Sun, V. Le, Q. Zhang, and Z. Su, “Toward understanding compiler bugs in GCC and LLVM,” in Proc. 25th Int. Symp. Softw. Testing Anal., 2016, pp. 294–305.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Y. Xiong et al., “An empirical study of functional bugs in android apps,” inProc. 32nd ACMSIGSOFT Int. Symp. Softw. Testing Anal., 2023, pp. 1319–1331.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "“MySQL Architecture,” MySQL Architecture. Accessed: Dec. 2023. [Online]. Available: https://dev.mysql.com/doc/refman/8.0/en/pluggable-storage-overview.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "“SQLite Architecture,” SQLite Ar chitecture. Accessed: Dec. 2023. [Online]. Available: https://www .sqlite.org/arch.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "“OpenGauss Architecture,” Open Gauss Architecture. Accessed: Dec. 2023. [Online]. Available: https:// gitee.com/opengauss/openGaussserver",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "S. Amann, S. Nadi, H. A. Nguyen, T. N. Nguyen, and M. Mezini, “Mubench: A benchmark for API-misuse detectors,” in Proc. 13th Int. Conf. Mining Softw. Repositories, 2016, pp. 464–467.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "S. Amann, H. A. Nguyen, S. Nadi, T. N. Nguyen, and M. Mezini, “A systematic evaluation of s tatic API-misuse detectors,” IEEE Trans. Softw. Eng., vol. 45, no. 12, pp. 1170–1188, Dec. 2019, doi: 10.1109/TSE.2018.2827384.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 12,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proc. 14th USENIX Symp. Operating Syst. Des. Implementation (OSDI), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 14,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proc. 28th ACM Joint Meeting Eur. Softw. Eng. Conf. Symp. Found. Softw. Eng., 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 19,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 1– 30. 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 21,
+      "text": "Z.-M. Jiang, S. Liu, M. Rigge r, and Z. Su, “Detecting transactional bugs in database engines via graphbased oracle construction,” in Proc. 17th USENIX Symp. Operating Sys t. Des. Implemen tation (OSDI), 2023, pp. 397–417.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 26,
+      "text": "J. Ba and M. Rigger, “CERT: Finding performance issues in database systems through the lens of cardinality estimation,” in Proc. IEEE/ACM 46th Int. Conf. Softw. Eng., 2024, pp. 1–13.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[12]",
+      "technique": "pqs",
+      "sentence": ": ACOMPREHENSIVE STUDY OFBUGS INRELATIONAL DBMS 3655 TABLE ICOMPARISON WITHPRIOR EMPIRICAL STUDIES ONRDBMS BUGS Study Bug Type Duration # Bugs Fonseca [22] concurrency bugs 2003-2009 80 Cui[23] transaction bugs 2018-2022 140 ours general bugs 2021-2023 777 [12],[13],[14],[15],[16],[17],[18],[19],[20],[21] aiming at detecting bugs in RDBMSs, and they successfully detected new bugs in RDBMSs.",
+      "context_before": "s due to its complex feature combinations anddeep levels of nested SQL queries. Given the critical role that RDBMSs play in various systems, it is crucial to effectively detect and address bugs in these databases. There have been various testing approaches [11], 0098-5589 © 2025 IEEE. All rights reserved, includi ng rights for text and data mining, and train ing of artificial intelligen ce and similar technologi es. Personal use is permitted, but republication/redist ribution requires IEEE permission. See http s://www.ieee.org/publicati ons/right s/index. html for more information. LIU et al.",
+      "context_after": "These approaches generate large numbers of diverse SQL statements through random test case generationor mutation, aiming to expose potential DBMS bugs. However, they typically support only a limited subset of SQL features and do not target specific components of RDBMSs. As a result, they are unable to detect bugs that arise from complex SQL functionalities, such as database events and triggers, st ored pr ocedures, access control mechanisms, or specific database parameter configurations. These bugs count for 33.72% of our studied bugs (See Section III-F for detailed analysis).” Effectively detec",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 5265,
+      "cited_reference": {
+        "number": 12,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proc. 14th USENIX Symp. Operating Syst. Des. Implementation (OSDI), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[12]",
+      "technique": "pqs",
+      "sentence": "We first consulted the DBEngines Ranking [24] to identify ranked and widely adopted RDBMS, and surveyed existing testing approaches [11],[12], [13],[14],[15],[16],[17],[18],[19],[20],[21],[25],[26] as well as empirical studies on RDBMS bugs [23],[27].",
+      "context_before": "correlations between the measured as-pects? Analyzing the correlations between root causes and symptoms offers valuable insights into preventing similar issues during future development process, and guides more targetedand effective testing and debugging strategies. B. Databases to be Studied In the study, we choose three popular open-source RDBMSs, including MySQL [1], SQLite [2], and openGauss [4],a s 1https://github.com/sqlttest/studydataset 2https://github.com/sqlttest/SQLT 3656 IEEETRANSACTIONS ONSOFTWARE ENGINEERING, VOL. 51, NO. 12, DECEMBER 2025 subjects for bug collection and study.",
+      "context_after": "This led us to identify potential candidates such as MySQL, SQLite,PostgreSQL, TiDB, MariaDB, DuckDB, and openGauss. After further evaluation, we selected three representative RDBMSs— MySQL, SQLite, and openGauss—based on two criteria: de-sign architecture and application scenarios. We categorized the candidate RDBMSs into three groups based on these criteria: (1) MariaDB and TiDB, whichare designed to be highly compatible with MySQL on system architecture and SQL syntax, making them wellsuited for small to medium-sized systems and distributed databases; (2) SQLite and DuckDB, which are embed",
+      "section": "B Databases to be Studied",
+      "page": 3,
+      "char_offset": 11266,
+      "cited_reference": {
+        "number": 12,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proc. 14th USENIX Symp. Operating Syst. Des. Implementation (OSDI), 2020, pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "cert"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", the SQL statements—focusing specifically on those that are not supported by existing test case generation tools such as SQLancer and SQLSmith.",
+      "context_before": "likelihood of logic-related bugs. openGauss shows a higher incidence of Missing Code Logic, con-sistent with its active developing status and relative immaturity in handling corner cases. For affected modules, SQLite reports the highest number of bugs in the optimizer. As an embedded database, SQLite features relatively lightweight execution and storage engines but a more substantial optimizer component. This architecturalfocus explains why its bugs are concentrated in the optimization module. F. Bug-Triggering SQLs We further analyzed the characteristics of the bug-triggering test cases—i.e.",
+      "context_after": "Across the three studied DBMSs, we identified 146 such cases that current tools cannot generate, including 46 in MySQL, 79 in OpenGauss, and 21 in SQLite. As we will discuss, extending current tools to supportthese features is non-trivial due to the complexity and specificity of the required SQL constructs. Partition-related bugs are prevalent across all three DBMSs, with a total of 61 cases identified. These bugs often stem from incorrect partition pr uning, invalid comparisons involving partition keys, or unsupported opera tions on partiti oned tables. Compared to flat tabl e structures, partiti",
+      "section": "F Bug-Triggering SQLs",
+      "page": 11,
+      "char_offset": 53027,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "However, existing test generation tools such as SQLancer and SQLSmith do not suppor t partition-related syntax at all, making it fundamentally impossible for them to generate test cases involving partition logic.",
+      "context_before": "SQLite. As we will discuss, extending current tools to supportthese features is non-trivial due to the complexity and specificity of the required SQL constructs. Partition-related bugs are prevalent across all three DBMSs, with a total of 61 cases identified. These bugs often stem from incorrect partition pr uning, invalid comparisons involving partition keys, or unsupported opera tions on partiti oned tables. Compared to flat tabl e structures, partiti oned tables introduce additional layers of metadata and planning logic, which cansignificantly affect query behavior and optimization strategies.",
+      "context_after": "This limitation highlights a critical blind spot in current automated testing techniques and underscoresthe need for extending support f or partitioned tabl e constructs in future testing frameworks. There are 19 bug-triggering test cases involving stored procedure/function and PLSQL statements. Constructing relevant test cases involves multi-statement routines, parameter passing, and internal control flow such as IFandWHILE. These are non-trivial to generate due to their nested structure and stateful behavior, making it harder to ensure the semantic correctnessof the generated SQLs. Bug-trigg",
+      "section": "F Bug-Triggering SQLs",
+      "page": 11,
+      "char_offset": 53915,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[21]",
+      "technique": null,
+      "sentence": "Recent works on RDBMS transaction testing [20],[21],[25] offer insights into testing efficacy.",
+      "context_before": "ion, we delve into bugs associated with RDBMSspecific features, discussing the implications derived fromour findings. We then discuss threats to validity of our study. A. Implications Transaction Bugs. Transactions are essential in RDBMS for ensuring the safe and reliable processing of data, critical in sectors like banking and online booking systems. However,there are only 33 tr ansaction-related b ugs, constituting 4.25% of all analyzed bugs, with 15 from MySQL, 5 from SQLite, and 13 from openGauss. This finding aligns with a recent researchreporting that transaction bugs are under-tested [23].",
+      "context_after": "Our examination of 33 transaction bugs against these methodologies showed that only 5 could potentially be detected by existing tools. The majority, 28 bugs,with feature complexities like clustering, replication, group replication, log analysis, custom functions, and control flow, which are not currently covered by these testing efforts. Impor-tantly, extending these testing tools to include more keywords in test case generation is insufficient; the primary challenge lies in developing effective oracles that can detect these complextransaction bugs. Moreover, current methods typically involve i",
+      "section": "A Implications",
+      "page": 12,
+      "char_offset": 57315,
+      "cited_reference": {
+        "number": 21,
+        "text": "Z.-M. Jiang, S. Liu, M. Rigge r, and Z. Su, “Detecting transactional bugs in database engines via graphbased oracle construction,” in Proc. 17th USENIX Symp. Operating Sys t. Des. Implemen tation (OSDI), 2023, pp. 397–417.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "However,extending existing syntax-directed tools like SQLancer to incorporate such functionality is non-trivial, as t hese featur es often require complex semantic correctness checks.",
+      "context_before": "f thetransaction bugs we analyzed could not be detected by existing transaction testing approaches. This finding underscores the need for more dedicated efforts towardsaddressing RDBMS-specific features, i.e., transactionrelated bugs. Test Case Generation. Our analysis of bug-triggering test cases reveals that certain features—PL/SQL constructs, user access control, and speci fic database paramet er settings—play a critical role in exposing bugs in RD BMSs, despite often falling outside the standard SQL scope. Consequently, it is essential for test case generation tools to support these features.",
+      "context_after": "This highlightsthe need for a new test case generation method that can inherently address semantic validation in a principled and scalable way. Finding 12: Approximately 33.72% of the studied bugs require complex features—such as database triggers, events, stored procedures, or specific parameter settings—in their test cases to be triggered. However, existing RDBMS testing tools lack support for these ad-vanced features. Generating such test cases under the current syntax-driven paradigm is challenging, as it often involves intricate semantic correctness checks. This under-scores the need for",
+      "section": "A Implications",
+      "page": 12,
+      "char_offset": 59061,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Design and Motivation of SQLTSQLT is an extension of the widely-used database testing tool, SQLancer [12].",
+      "context_before": "gs, we chose three popular and diverse RDBMSs, representing different architectures and application scenarios. After a meticulous filtering process, we analyzed 777 bugs,balancing the scope of our research with the demands of manual effort. V. APOOF-OF- CONCEPT TOOLSQLT To evaluate the practical utility of our findings in enhancing RDBMS testing, we focus on SQL-type bugs, which account for approximately 9% of all studied bugs. We developed SQLT,a generation-based tool specifically designed to address SQLtype issues, enabling targeted testing and validation based on the identified bug patterns. A.",
+      "context_after": "Based on our observation that many SQLtype bugs are triggered by novel or less-tested data types (e.g., BIT, JSON), while others result from implicit type conversions within functions, we enhanced SQLT by incorporatingsyntax-aware rules. These rules systematically generate SQL statements designed to either introduce new data types or provoke SQL type conversions, both implicit and explicit, therebyincreasing the likelihood of exposing type-related bugs. This makes SQLT particularly adapt at testing RDBMS functionalities related to SQL types. Specifically, SQLT extends SQLancerin four major asp",
+      "section": "V APOOF-OF- CONCEPT TOOLSQLT",
+      "page": 13,
+      "char_offset": 61765,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLT extends SQLancer by introducing new data types such as BIT and JSON, which were undersupported by SQLancer.",
+      "context_before": "any SQLtype bugs are triggered by novel or less-tested data types (e.g., BIT, JSON), while others result from implicit type conversions within functions, we enhanced SQLT by incorporatingsyntax-aware rules. These rules systematically generate SQL statements designed to either introduce new data types or provoke SQL type conversions, both implicit and explicit, therebyincreasing the likelihood of exposing type-related bugs. This makes SQLT particularly adapt at testing RDBMS functionalities related to SQL types. Specifically, SQLT extends SQLancerin four major aspects. Supporting new data types.",
+      "context_after": "Moreover, SQLT adds support foropenGauss by implementing a dedicated test case generation module. Encouraging type conversion. SQLT systematically increases the probability of both explicit type conversions (e.g., CAST, CONVERT ) and implicit conversions by injecting heterogeneous data types into expressions, effectively stressing the type conversion logic of DBMSs. SQLT also integrates a large set of TABLE VIIIBUGS DETECTED BYSQLT Bug IDRDBMS Version Status 112912 MySQL 8.0 Confirmed 112914 MySQL 8.0 Confirmed a655ee159e SQLite 3.43.0 Confirmed I8GTTH openGauss 5.0.0 Fixed I8ENSF openGauss 5.0.0",
+      "section": "V APOOF-OF- CONCEPT TOOLSQLT",
+      "page": 13,
+      "char_offset": 62504,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "0 Duplicate TABLE IXCOMPARISON OFSQLT WITH SQLANCER FOR 24H RDBMS SQLT SQLancer MySQL 4 3 SQLite 1 1 database-specific functions that are prone to type conversion issues, such as STRING_AGG ,UNIXEPOCH, and TIMEDIFF in MySQL/openGauss, and RANDOM ,IIF,OCTET_LENGTH, and UNHEX in SQLite.",
+      "context_before": "tion module. Encouraging type conversion. SQLT systematically increases the probability of both explicit type conversions (e.g., CAST, CONVERT ) and implicit conversions by injecting heterogeneous data types into expressions, effectively stressing the type conversion logic of DBMSs. SQLT also integrates a large set of TABLE VIIIBUGS DETECTED BYSQLT Bug IDRDBMS Version Status 112912 MySQL 8.0 Confirmed 112914 MySQL 8.0 Confirmed a655ee159e SQLite 3.43.0 Confirmed I8GTTH openGauss 5.0.0 Fixed I8ENSF openGauss 5.0.0 Fixed I879H0 openGauss 5.0.0 Fixed I8ENDC openGauss 5.0.0 Fixed I8ENLG openGauss 5.0.",
+      "context_after": "These functions are obtained from our analysis of existing bugs and newly introduced functions inrecent DBMS releases. Given that bugs related to SQL types often result in result inconsistencies—silent bugs without explicit error messages—we incorporated two existing metamorphic testing approaches, NOREC [14] and TLP [19], as oracles within SQLT. SQLT is publicly available and can be accessed at https://github.com/sqlttest/SQLT. B. Evaluation of SQLT Our experiments were conducted on an Ubuntu 16.04 server equipped with 500GB of memory and an Intel(R) Xeon(R)Platinum 8260 CPU @ 2.30GHz. We ap",
+      "section": "V APOOF-OF- CONCEPT TOOLSQLT",
+      "page": 13,
+      "char_offset": 63302,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "Given that bugs related to SQL types often result in result inconsistencies—silent bugs without explicit error messages—we incorporated two existing metamorphic testing approaches, NOREC [14] and TLP [19], as oracles within SQLT.",
+      "context_before": "d 112914 MySQL 8.0 Confirmed a655ee159e SQLite 3.43.0 Confirmed I8GTTH openGauss 5.0.0 Fixed I8ENSF openGauss 5.0.0 Fixed I879H0 openGauss 5.0.0 Fixed I8ENDC openGauss 5.0.0 Fixed I8ENLG openGauss 5.0.0 Duplicate TABLE IXCOMPARISON OFSQLT WITH SQLANCER FOR 24H RDBMS SQLT SQLancer MySQL 4 3 SQLite 1 1 database-specific functions that are prone to type conversion issues, such as STRING_AGG ,UNIXEPOCH, and TIMEDIFF in MySQL/openGauss, and RANDOM ,IIF,OCTET_LENGTH, and UNHEX in SQLite. These functions are obtained from our analysis of existing bugs and newly introduced functions inrecent DBMS releases.",
+      "context_after": "SQLT is publicly available and can be accessed at https://github.com/sqlttest/SQLT. B. Evaluation of SQLT Our experiments were conducted on an Ubuntu 16.04 server equipped with 500GB of memory and an Intel(R) Xeon(R)Platinum 8260 CPU @ 2.30GHz. We applied SQLT to SQLite versions 3.43.0, MySQL 8.0.34, and openGauss 5.0.0 over a two-week period. SQLT successfully detected 8 b ugs, all of which were confirmed by developers and 4 have been fixed, as detailed in Table VIII. Notably, 2 bugs in MySQL were related to type conversion, 1 bug in SQLite was associated with JSON5 SQL type support, and in op",
+      "section": "V APOOF-OF- CONCEPT TOOLSQLT",
+      "page": 13,
+      "char_offset": 63703,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We’ve also compared SQLT with SQLancer on the latest version of MySQL and SQLite (MySQL 8.",
+      "context_before": "n fixed, as detailed in Table VIII. Notably, 2 bugs in MySQL were related to type conversion, 1 bug in SQLite was associated with JSON5 SQL type support, and in openGauss, 2 bugs were due to SQL types,with the remaining 3 stemming from incorrect error handling, null pointer errors, and issues with the full join keyword, they were uncovered due to our extended support for openGauss.These findings, particularly the detection of 5 type-related bugs within a short timeframe, va lidate our initial hypothesis and underscore the need for more focused efforts on SQL type-specific issues in RDBMS testing.",
+      "context_after": "0.34 and SQLite3.43.0) 3for 24 hours and adopt NoREC [14] as the default 3We did not run openGauss as SQLancer does not support openGauss. LIU et al.: ACOMPREHENSIVE STUDY OFBUGS INRELATIONAL DBMS 3667 oracle to detect bugs. SQLT detects 6 bugs and SQLancer detects 4 bugs. Among these results, SQLT uncovered 1 additional bug in MySQL that were not detected by SQLancer. The bug stems from type-related issues involving implicit conversions and edge-case expressions on the column of BIT type. Notably, SQLancer does not actively avoid type mismatches; rather, itstest generation lacks sufficient b",
+      "section": "B Evaluation of SQLT",
+      "page": 13,
+      "char_offset": 64971,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "0) 3for 24 hours and adopt NoREC [14] as the default 3We did not run openGauss as SQLancer does not support openGauss.",
+      "context_before": "s associated with JSON5 SQL type support, and in openGauss, 2 bugs were due to SQL types,with the remaining 3 stemming from incorrect error handling, null pointer errors, and issues with the full join keyword, they were uncovered due to our extended support for openGauss.These findings, particularly the detection of 5 type-related bugs within a short timeframe, va lidate our initial hypothesis and underscore the need for more focused efforts on SQL type-specific issues in RDBMS testing. We’ve also compared SQLT with SQLancer on the latest version of MySQL and SQLite (MySQL 8.0.34 and SQLite3.43.",
+      "context_after": "LIU et al.: ACOMPREHENSIVE STUDY OFBUGS INRELATIONAL DBMS 3667 oracle to detect bugs. SQLT detects 6 bugs and SQLancer detects 4 bugs. Among these results, SQLT uncovered 1 additional bug in MySQL that were not detected by SQLancer. The bug stems from type-related issues involving implicit conversions and edge-case expressions on the column of BIT type. Notably, SQLancer does not actively avoid type mismatches; rather, itstest generation lacks sufficient bias toward diverse data types and fails to synthesize high-risk constructs like nested casts or ambiguous comparisons. SQLT addresses thes",
+      "section": "B Evaluation of SQLT",
+      "page": 13,
+      "char_offset": 65082,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLT detects 6 bugs and SQLancer detects 4 bugs.",
+      "context_before": "rd, they were uncovered due to our extended support for openGauss.These findings, particularly the detection of 5 type-related bugs within a short timeframe, va lidate our initial hypothesis and underscore the need for more focused efforts on SQL type-specific issues in RDBMS testing. We’ve also compared SQLT with SQLancer on the latest version of MySQL and SQLite (MySQL 8.0.34 and SQLite3.43.0) 3for 24 hours and adopt NoREC [14] as the default 3We did not run openGauss as SQLancer does not support openGauss. LIU et al.: ACOMPREHENSIVE STUDY OFBUGS INRELATIONAL DBMS 3667 oracle to detect bugs.",
+      "context_after": "Among these results, SQLT uncovered 1 additional bug in MySQL that were not detected by SQLancer. The bug stems from type-related issues involving implicit conversions and edge-case expressions on the column of BIT type. Notably, SQLancer does not actively avoid type mismatches; rather, itstest generation lacks sufficient bias toward diverse data types and fails to synthesize high-risk constructs like nested casts or ambiguous comparisons. SQLT addresses these gaps by in-creasing the likelihood of type coercion chains, introducing under-tested types such as BIT and JSON, and embedding conversi",
+      "section": "B Evaluation of SQLT",
+      "page": 14,
+      "char_offset": 65288,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Among these results, SQLT uncovered 1 additional bug in MySQL that were not detected by SQLancer.",
+      "context_before": "rt for openGauss.These findings, particularly the detection of 5 type-related bugs within a short timeframe, va lidate our initial hypothesis and underscore the need for more focused efforts on SQL type-specific issues in RDBMS testing. We’ve also compared SQLT with SQLancer on the latest version of MySQL and SQLite (MySQL 8.0.34 and SQLite3.43.0) 3for 24 hours and adopt NoREC [14] as the default 3We did not run openGauss as SQLancer does not support openGauss. LIU et al.: ACOMPREHENSIVE STUDY OFBUGS INRELATIONAL DBMS 3667 oracle to detect bugs. SQLT detects 6 bugs and SQLancer detects 4 bugs.",
+      "context_after": "The bug stems from type-related issues involving implicit conversions and edge-case expressions on the column of BIT type. Notably, SQLancer does not actively avoid type mismatches; rather, itstest generation lacks sufficient bias toward diverse data types and fails to synthesize high-risk constructs like nested casts or ambiguous comparisons. SQLT addresses these gaps by in-creasing the likelihood of type coercion chains, introducing under-tested types such as BIT and JSON, and embedding conversion-prone expressions in critical control points. VI. RELATED WORK Empirical Study on RDBMS Bugs Th",
+      "section": "B Evaluation of SQLT",
+      "page": 14,
+      "char_offset": 65337,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Notably, SQLancer does not actively avoid type mismatches; rather, itstest generation lacks sufficient bias toward diverse data types and fails to synthesize high-risk constructs like nested casts or ambiguous comparisons.",
+      "context_before": "DBMS testing. We’ve also compared SQLT with SQLancer on the latest version of MySQL and SQLite (MySQL 8.0.34 and SQLite3.43.0) 3for 24 hours and adopt NoREC [14] as the default 3We did not run openGauss as SQLancer does not support openGauss. LIU et al.: ACOMPREHENSIVE STUDY OFBUGS INRELATIONAL DBMS 3667 oracle to detect bugs. SQLT detects 6 bugs and SQLancer detects 4 bugs. Among these results, SQLT uncovered 1 additional bug in MySQL that were not detected by SQLancer. The bug stems from type-related issues involving implicit conversions and edge-case expressions on the column of BIT type.",
+      "context_after": "SQLT addresses these gaps by in-creasing the likelihood of type coercion chains, introducing under-tested types such as BIT and JSON, and embedding conversion-prone expressions in critical control points. VI. RELATED WORK Empirical Study on RDBMS Bugs There are two RDBMS bug studies [22],[23] in the literature. Fonseca et al. conduct an early study on 80 RDBMS concurrency bugs [22], discussed the impact of the concurrency bugs and the latent concurrency bugs distribution. Cui et al. studied 140 RDBMS transactionbugs [23], including statistical information such as the average amount of data us",
+      "section": "B Evaluation of SQLT",
+      "page": 14,
+      "char_offset": 65558,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "[12],[14], [19] proposed SQLancer, which is one of the most popular generation-based RDBMS testing approach, and it incorporates three metamorphic methods, namely PQS [12], NoREC [14], and TLP [19], as oracles to detect logical bugs.",
+      "context_before": "average amount of data used in a table, the length of code fixes, etc. They also pointed out that the majority of transaction bugs aresilent, which is consistent with our findings on RDBMS bugs in general. Both studies focus on a special type of RDBMS bugs, and the number of bugs they studied are small. Our work focuson a broader range of bugs in RDBMS and we studied a total of 777 bugs, making our study general and comprehensive. RDBMS Testing SQLSmith [11] is the first generation-based testing approach for RDBMS, and it report bugs by examining explicit error messages and crashes. Rigger et al.",
+      "context_after": "These testing approaches rely on syntax-directed SQL generation, which lacks support for the complex features analyzed in Section III-F and cannot be easily extended to accommodate them. SQUIRREL [13], SQLRight [15] and GRIFFIN [16] are mutation-based testing approaches. SQUIRREL [13] considers both language validity and coverage feedback to test RDBMSs. SQLRight [15]conducts coverage-guided mutation, with semantic validity checkings. GRIFFIN [16] reshuffles statements based on dependencies between metadata and statements in the metadata graph. These testing approaches rely on fuzzing techniqu",
+      "section": "VI RELATED WORK",
+      "page": 14,
+      "char_offset": 66955,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[21]",
+      "technique": null,
+      "sentence": "Troc [20] and TxCheck [21] are metamorphic testing approaches focusing on detect transactional bugs in RDBMSs.",
+      "context_before": "statements in the metadata graph. These testing approaches rely on fuzzing techniques, generating test cases by mutating existing seed inputs. As aresult, the quality of the seed test cases plays a crucial role in effectively triggering bugs. While seed cases involving the complex features discussed in Section III-F can be partially sourced from open-source projects, ensuring semantic correctness during the mutation process remains a non-trivial chal-lenge. There is also some work dedicated to detecting less common types of bugs in DBMS. AMOEBA [17]and APOLLO [18] test RDBMS performance bugs.",
+      "context_after": "Most of the testing approaches are general, with no focus on RDBMS specific features, like SQL types and transactions. The only three transaction testing approaches have limited capab ilities due to the restricted oracles and heavy programming barriers. VII. CONCLUSION In this paper, we present the first comprehensive study of bugs in Relational Database Management Systems (RDBMS), with the goal of deepening our understanding of RDBMS bugs and improving the RDBMS development, testing and maintenance. We meticulously analyzed 777 bugs from three opensource RDBMSs—SQLite, MySQL, and openGauss—cove",
+      "section": "VI RELATED WORK",
+      "page": 14,
+      "char_offset": 68305,
+      "cited_reference": {
+        "number": 21,
+        "text": "Z.-M. Jiang, S. Liu, M. Rigge r, and Z. Su, “Detecting transactional bugs in database engines via graphbased oracle construction,” in Proc. 17th USENIX Symp. Operating Sys t. Des. Implemen tation (OSDI), 2023, pp. 397–417.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M11"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:51:41Z",
+    "is_model_written": true,
+    "summary": "An empirical study of 777 general bugs across widely used relational DBMSs reported between 2021 and 2023, larger and broader than the prior studies it tabulates, which covered concurrency and transaction bugs only. It characterises what triggers these bugs and finds a recurring gap: much of the bug-triggering SQL uses syntax that existing generators do not produce at all. To show the gap is actionable the authors build a proof-of-concept tool, SQLT.",
+    "narrative": "SQLancer is both the study's reference generator and the base of its proof-of-concept tool. The study repeatedly measures bug-triggering SQL against what SQLancer can generate -- it does not support partition-related syntax at all, and it lacks bias toward diverse data types or high-risk constructs such as nested casts. SQLT is then built as an extension of SQLancer, adding BIT and JSON support that SQLancer under-supported, and running NoREC and TLP as its oracles. The two are compared directly over 24 hours on MySQL and SQLite, where SQLT found 6 bugs to SQLancer's 4, including one in MySQL that SQLancer did not reach.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "motivation",
+      "M4": "motivation",
+      "M5": "background",
+      "M6": "motivation",
+      "M7": "reuse_implementation",
+      "M8": "reuse_implementation",
+      "M9": "result_comparison",
+      "M10": "reuse_component",
+      "M11": "baseline",
+      "M12": "baseline",
+      "M13": "result_comparison",
+      "M14": "result_comparison",
+      "M15": "result_comparison",
+      "M16": "definition",
+      "M17": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M7",
+          "M8",
+          "M10"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M7",
+            "sentence": "Design and Motivation of SQLTSQLT is an extension of the widely-used database testing tool, SQLancer [12].",
+            "section": "V APOOF-OF- CONCEPT TOOLSQLT",
+            "page": 13
+          },
+          {
+            "mention_id": "M8",
+            "sentence": "SQLT extends SQLancer by introducing new data types such as BIT and JSON, which were undersupported by SQLancer.",
+            "section": "V APOOF-OF- CONCEPT TOOLSQLT",
+            "page": 13
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "Given that bugs related to SQL types often result in result inconsistencies—silent bugs without explicit error messages—we incorporated two existing metamorphic testing approaches, NOREC [14] and TLP [19], as oracles within SQLT.",
+            "section": "V APOOF-OF- CONCEPT TOOLSQLT",
+            "page": 13
+          }
+        ],
+        "reasoning": "M7 states SQLT is an extension of SQLancer, M8 that it extends SQLancer with data types SQLancer under-supported, and M10 that NoREC and TLP were incorporated as oracles within SQLT. The tool is built on the codebase.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "What SQLT extends is the tool's type and syntax coverage, not an oracle: M10 records NoREC and TLP being incorporated as they are, to serve as SQLT's oracles."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M11",
+          "M13",
+          "M14"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M11",
+            "sentence": "We’ve also compared SQLT with SQLancer on the latest version of MySQL and SQLite (MySQL 8.",
+            "section": "B Evaluation of SQLT",
+            "page": 13
+          },
+          {
+            "mention_id": "M13",
+            "sentence": "SQLT detects 6 bugs and SQLancer detects 4 bugs.",
+            "section": "B Evaluation of SQLT",
+            "page": 14
+          },
+          {
+            "mention_id": "M14",
+            "sentence": "Among these results, SQLT uncovered 1 additional bug in MySQL that were not detected by SQLancer.",
+            "section": "B Evaluation of SQLT",
+            "page": 14
+          }
+        ],
+        "reasoning": "M11 states SQLT was compared with SQLancer on MySQL and SQLite over 24 hours, M13 gives 6 bugs against 4, and M14 that one MySQL bug was found only by SQLT."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M16 calls SQLancer one of the most popular generation-based RDBMS testing approaches. Popularity is recognition but not a claim about being the most effective or the state of the art, and the study's own findings argue the opposite about its coverage."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1117_12_3006402.json b/_data/papers/paper_doi_10_1117_12_3006402.json
new file mode 100644
index 0000000..5aff0f2
--- /dev/null
+++ b/_data/papers/paper_doi_10_1117_12_3006402.json
@@ -0,0 +1,248 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:32:57Z",
+  "paper": {
+    "id": "paper:doi:10.1117/12.3006402",
+    "title": "An empirical study on configuration-related branch statement in database management system",
+    "authors": [
+      "Xie Li",
+      "Huiping Zhou",
+      "Haifang Zhou",
+      "Jingying Zhang"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Modelling, Identification and Control",
+    "doi": "10.1117/12.3006402",
+    "arxiv_id": null,
+    "s2_paper_id": "6200bb7acb7f2d10f5111f75a895ef4e74b81389",
+    "url": "https://doi.org/10.1117/12.3006402",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1117/12.3006402",
+      "retrieved_at": "2026-09-09T01:32:57Z",
+      "chars": 16510,
+      "content_sha256": "sha256:455db35c642025a5e447455de17891e74558cf7293dc904bedcbe7e80a090bfe"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 5,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1498
+      },
+      {
+        "number": "2",
+        "title": "METHODOLOGY",
+        "start": 5645
+      },
+      {
+        "number": "3",
+        "title": "IMPORTANC EOF CONFIGURATION BRANCHES",
+        "start": 7617
+      },
+      {
+        "number": "4",
+        "title": "FINDINGS ANDIMPLICATIONS",
+        "start": 9424
+      },
+      {
+        "number": "5",
+        "title": "CONCLUSIONS ANDFUTURE WORK",
+        "start": 12279
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "MySQL Customers. https://www. mysql. com/custom ers/, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "PostgreSQL Clients. https://wiki. postgresql. org/wiki/PostgreSQL_Clients, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Well-Known Users of SQLite. https://www. sqlite. org/famous. html, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "U. M. Fayyad. Data Science Revealed: A Data-Driven Glimpse into the Burgeoning New Field. https ://fayyad. com/data -science -revealed -a-data-driven -glimpse into -the-burgeoning -new-field/, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "D. Laney. 3-D Data Management: Controlling Data Volume, Velocity and Variety. Technical report, Feb. 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "M. Stonebraker, S. Madden, and P. Dubey. Intel \"Big Da ta\" Science and Technology Center Vision and Execution Plan. ACMSIGMOD Record, 42(1):44 –49, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Bugs found in database management systems. https://www. manuelrigger. at/dbms -bugs. Accessed: August 20, 2021",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 8,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. APOLLO: Automatic Detection and Di agnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB), Tokyo, Japan, August 2020. Technical report, Feb. 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implemen tation OSDI 20), pages 667 –682, 2020. Y. Yorozu, M. Hirano, K. Oka, and Y. Tagawa, “Electron spectroscopy studies on magneto -optical media and plastic substrate interface, ” IEEE Transl. J. Magn. Japan, vol. 2, pp. 740–741, August 1987 [Digests 9t",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 10,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dingh ao Wu. Squirrel: Testing database management systems with language validity and cover age feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "MongoDB blog. https://www. mongodb. com/blog/post/how -to-avoid -a-malicious -attack-that-ransoms -yourdata, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jingxuan He, Tianyu Gu, Peixin Zhang, Yang Liu, and Xiaorui Gong. . MARMOT: Sequence-Oriented Seed Mining for Database Fuzzing. IEEE Symposium on Security and Privacy, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chiji n Zhou, Huafeng Zhang and Yu Jiang. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. IEEE International Conference on Trust, Security and Privacy in Computing and Communications, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "W. Enck, P. Gilbert, B. -G. Chun, L. P. Cox, J. Jung, P. McDaniel, and A. N. Sheth, “Taintdroid: An information -flow tracking system for realtime privacy monitoring on smartphones, ” in Proceedings of the 9th USENIX Conference on Operating Systems Design and Implementation, ser. OSDI’10, 2010, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "TANG, C., KOOBURAT, T., VENKATACHALAM, P., CHANDER, A., WEN, Z., NARAYANAN, A., DOWELL, P., ANDKARL, R. Holistic Configuration Management at Facebook. In Proceedings of the 25th ACM Symposium on Operating System Principles (SOSP’15) (October 2015).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "XU, T., ZHANG, J., HUANG, P., ZHENG, J., SHENG, T., YUAN, D., ZHOU, Y., ANDPASUPATHY, S. Do Not Blame Users for Misconfigurations. In Proceedings of the 24th ACM Symposium on Operating System Principles (SOSP’13)(November 2013). Proc. of SPIE Vol. 12793 127930D-5",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 7,
+      "text": "Bugs found in database management systems. https://www. manuelrigger. at/dbms -bugs. Accessed: August 20, 2021",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 9,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implemen tation OSDI 20), pages 667 –682, 2020. Y. Yorozu, M. Hirano, K. Oka, and Y. Tagawa, “Electron spectroscopy studies on magneto -optical media and plastic substrate interface, ” IEEE Transl. J. Magn. Japan, vol. 2, pp. 740–741, August 1987 [Digests 9th Annual Conf. Magnetics Japan, p. 301, 1982]. Proc. of SPIE Vol. 12793 127930D-4",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[7, 8, 9, 10]",
+      "technique": "pqs",
+      "sentence": "Malicious attackers may use these vulnerabilities to steal user information, embezzle data, crash the system, and bring unpredictable economic losses[7, 8, 9, 10].",
+      "context_before": "ever, like other large and complex systems, DBMSs inevitably contain many vulnerabilities that affect user experience and may even cause serious security issues. Among them, it is quite common that a vulnerability is caused by configuration. According to the information released by the National Informa tion Security Vulnerability Database (CNNVD), a large number of database management systems have vulnerabilities due to configuration. For example, the MongoDB database had a data breach in 2017 because authentication was not enabled by default. The same p roblem occurs with other databases[11].",
+      "context_after": "Recently, coverage -guided greybox fuzzing has become a ho t topic and achieved great success in the field of database vulnerability finding. Unlike blackbox fuzzing, coverage -guided greybox fuzzing introduces code coverage as feedback into the fuzzing loop and makes the fuzzing campaign evolve in the direction of i mproving code coverage. For example, SQUIRREL [10] translates sql query to IR first and perform mutations on IR to ensure that the syntax and semantics of the newly generated SQL statements are correct; MARMOT [12] improves the abundance of SQL Type Sequences in fuzzing and succe",
+      "section": "1 INTRODUCTION",
+      "page": 1,
+      "char_offset": 2375,
+      "cited_reference": {
+        "number": 9,
+        "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implemen tation OSDI 20), pages 667 –682, 2020. Y. Yorozu, M. Hirano, K. Oka, and Y. Tagawa, “Electron spectroscopy studies on magneto -optical media and plas",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:03:26Z",
+    "is_model_written": true,
+    "summary": "An empirical study of configuration-related branch statements in DBMS source code -- the branches whose outcome depends on a configuration option -- quantifying how much behaviour is reachable only under non-default settings.",
+    "narrative": "The single mention is a citation in the opening motivation about the consequences of DBMS vulnerabilities, reachable only through the citation marker. SQLancer is not named and plays no role in the study, whose method is static analysis of branch conditions.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3372297_3417260.json b/_data/papers/paper_doi_10_1145_3372297_3417260.json
new file mode 100644
index 0000000..42b9676
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3372297_3417260.json
@@ -0,0 +1,553 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:27:28Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3372297.3417260",
+    "title": "SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback",
+    "authors": [
+      "Rui Zhong",
+      "Yongheng Chen",
+      "Hong Hu",
+      "Hangfan Zhang",
+      "Wenke Lee",
+      "Dinghao Wu"
+    ],
+    "year": 2020,
+    "venue": "Conference on Computer and Communications Security",
+    "doi": "10.1145/3372297.3417260",
+    "arxiv_id": "2006.02398",
+    "s2_paper_id": "afe18b71fa5b215a554a2da141f677081fac4503",
+    "url": "https://doi.org/10.1145/3372297.3417260",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2006.02398",
+      "retrieved_at": "2026-09-08T15:27:28Z",
+      "chars": 98156,
+      "content_sha256": "sha256:e0702c2a13bf695895b043c65606da7b427b33bf3eb501e7db0d67eb596b2266"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 16,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1689
+      },
+      {
+        "number": "2",
+        "title": "Problem Definition",
+        "start": 10115
+      },
+      {
+        "number": "2.1",
+        "title": "Query Processing in DBMS",
+        "start": 10393
+      },
+      {
+        "number": "2.2",
+        "title": "Challenges of DBMS Testing",
+        "start": 11903
+      },
+      {
+        "number": "2.3",
+        "title": "Our Approach",
+        "start": 14917
+      },
+      {
+        "number": "5.1",
+        "title": "Structure-Data Separation",
+        "start": 24685
+      },
+      {
+        "number": "5.2",
+        "title": "Type-Based Mutation",
+        "start": 26805
+      },
+      {
+        "number": "6.1",
+        "title": "Data Dependency Inference",
+        "start": 30802
+      },
+      {
+        "number": "6.2",
+        "title": "IR Instantiation",
+        "start": 37327
+      },
+      {
+        "number": "8.1",
+        "title": "DBMS Bugs",
+        "start": 44069
+      },
+      {
+        "number": "8.2",
+        "title": "Comparison with Existing Tools",
+        "start": 54309
+      },
+      {
+        "number": "8.3",
+        "title": "Contributions of Validity and Feedback",
+        "start": 62154
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "MariaDB. https://www.mariadb.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "MySQL. https://www.mysql.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "PostgreSQL. https://www.postgresql.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "SQLite. https://www.sqlite.org/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "SQLSmith. https://github.com/anse1/sqlsmith, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Processing a SQL Statement. https://docs.microsoft.com/en-us/sql/odbc/ reference/processing-a-sql-statement?view=sql-server-ver15, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Yahoo Says All Three Billion Accounts Hacked in 2013 Data Theft. https://www.reuters.com/article/us-yahoo-cyber/yahoo-says-all-threebillion-accounts-hacked-in-2013-data-theft-id USKCN1C82O1, October 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Proceedings of the 39th IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, May 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "MySQL Customers. https://www.mysql.com/customers/, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "PostgreSQL Clients. https://wiki.postgresql.org/wiki/PostgreSQL_Clients, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "SQL Keywords Reference. https://www.w3schools.com/sql/sql_ref_keywords. asp, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "SQL Operators. https://www.w3schools.com/sql/sql_operators.asp, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Well-Known Users of SQLite. https://www.sqlite.org/famous.html, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "SQLsmith Description. https://github.com/anse1/sqlsmith#description, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "S. Abdul Khalek and S. Khurshid. Automated SQL Query Generation for Systematic Testing of Database Engines. In Proceedings of the IEEE/ACM International Conference on Automated Software Engineering, ASE10, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "B. Acohido. Small Banks and Credit Union Attack Set for Tuesday. https://www.usatoday.com/story/cybertruth/2013/05/06/ddos-denial-ofservice-small-business-cybersecurity-privacy/2139349/, May 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Alloy. Alloy-Documentation of Alloy SAT solver, 2019. https://alloytools.org/ documentation.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "C. Aschermann, T. Frassetto, T. Holz, P. Jauernig, A.-R. Sadeghi, and D. Teuchert. Nautilus: Fishing for deep bugs with grammars. In NDSS, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "C. Aschermann, S. Schumilo, T. Blazytko, R. Gawlik, and T. Holz. Redqueen: Fuzzing with input-to-state correspondence. In Symposium on Network and Distributed System Security (NDSS), 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "H. Bati, L. Giakoumakis, S. Herbert, and A. Surna. A Genetic Approach for Random Testing of Database Systems. In Proceedings of the 33rd International Conference on Very Large Data Bases, VLDB07, pages 1243–1251, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "T. Blazytko, C. Aschermann, M. Schlögel, A. Abbasi, S. Schumilo, S. Wörner, and T. Holz. GRIMOIRE: Synthesizing structure while fuzzing. In USENIX Security Symposium, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "M. Böhme, V.-T. Pham, and A. Roychoudhury. Coverage-based Greybox Fuzzing As Markov Chain. In Proceedings of the 23rd ACM Conference on Computer and Communications Security (CCS), Vienna, Austria, Oct. 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "M. Böhme, V.-T. Pham, M.-D. Nguyen, and A. Roychoudhury. Directed greybox fuzzing. In Proceedings of the 2017 ACMSIGSAC Conference on Computer and Communications Security, pages 2329–2344, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "E. V. Buskirk. Facebook Confirms Denial-of-Service Attack. https://www. wired.com/2009/08/facebook-apparently-attacked-in-addition-to-twitter/, August 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "B. Chandra, B. Chawda, B. Kar, K. V. M. Reddy, S. Shah, and S. Sudarshan. Data generation for testing and grading sql queries. The VLDB Journal, 24(6):731–755, Aug 2015. ISSN 0949-877X. doi: 10.1007/s00778-015-0395-0. URL http://dx.doi. org/10.1007/s00778-015-0395-0.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "P. Chen and H. Chen. Angora: Efficient Fuzzing By Principled Search. In Proceedings of the 39th IEEE Symposium on Security and Privacy (Oakland) SP1 [8] .",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Y. Chen, P. Li, J. Xu, S. Guo, R. Zhou, Y. Zhang, L. Lu, et al. Savior: Towards bug-driven hybrid testing. In Proceedings of the 41th IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, May 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "C. Cimpanu. Google Chrome Impacted by New Magellan 2.0 Vulnerabilities. https://www.zdnet.com/article/google-chrome-impacted-by-newmagellan-2-0-vulnerabilities/, December 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "U. M. Fayyad. Data Science Revealed: A Data-Driven Glimpse into the Burgeoning New Field. https://fayyad.com/data-science-revealed-a-data-driven-glimpseinto-the-burgeoning-new-field/, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "L. Franceschi-Bicchierai. Hacker Tries To Sell 427 Milllion Stolen MySpace Passwords For $2,800. https://www.vice.com/en_us/article/pgkk8v/427-millionmyspace-passwords-emails-data-breach, May 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "S. Gan, C. Zhang, X. Qin, X. Tu, K. Li, Z. Pei, and Z. Chen. CollAFL: Path Sensitive Fuzzing. In Proceedings of the 39th IEEE Symposium on Security and Privacy (Oakland) SP1 [8] .",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "S. Gan, C. Zhang, P. Chen, B. Zhao, X. Qin, D. Wu, and Z. Chen. GREYONE: Data Flow Sensitive Fuzzing. In Proceedings of the 29th USENIX Security Symposium (Security), BOSTON, MA, Aug. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Google. Honggfuzz, 2016. https://google.github.io/honggfuzz/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Google. OSS-Fuzz-Continuous Fuzzing For Open Source Software. https: //github.com/google/oss-fuzz, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "H. Han, D. Oh, and S. K. Cha. Codealchemist: Semantics-aware code generation to find vulnerabilities in javascript engines. In Proceedings of the 2019 Annual Network and Distributed System Security Symposium (NDSS), San Diego, CA, Feb. 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "T. Hunt. The 773 Million Record \"Collection #1\" Data Breach. https://www. troyhunt.com/the-773-million-record-collection-1-data-reach/, January 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB), Tokyo, Japan, Aug. 2020. 13 ACMCCS 2020, Orlando, USA Rui Zhong†∗, Yongheng Chen§∗, Hong Hu§, Hangfan Zhang†, Wenke Lee§and Dinghao Wu†",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "V. V. Koushik. ALERT: SQLite database Remote Code Execution Vulnerability. https://www.secpod.com/blog/sqlite-database-remote-code-execution/, August 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "D. Laney. 3-D Data Management: Controlling Data Volume, Velocity and Variety. Technical report, Feb. 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "C. Lemieux and K. Sen. Fairfuzz: A targeted mutation strategy for increasing greybox fuzz testing coverage. In Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering, pages 475–485, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "G. Li, X. Zhou, S. Li, and B. Gao. QTune: A Query-Aware Database Tuning System with Deep Reinforcement Learning. Proceedings of the VLDB Endowment, 12(12): 2118–2130, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Y. Li, B. Chen, M. Chandramohan, S.-W. Lin, Y. Liu, and A. Tiu. Steelix: Programstate Based Binary Fuzzing. In Proceedings of the 11th Joint Meeting on Foundations of Software Engineering, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "LLVM. LibFuzzer-A Library For Coverage-guided Fuzz Testing, 2017. http: //llvm.org/docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "E. Lo, C. Binnig, D. Kossmann, M. Tamer Özsu, and W.-K. Hon. A Framework for Testing DBMS Features. The VLDB Journal, 19(2):203–230, Apr. 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "M. Marcozzi, W. Vanhoof, and J.-L. Hainaut. Test Input Generation for Database Programs Using Relational Constraints. In Proceedings of the Fifth International Workshop on Testing Database Systems, DBTest12, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "B. P. Miller, L. Fredriksen, and B. So. An Empirical Study Of The Reliability Of UNIX Utilities. Commun. ACM, 33(12):32–44, Dec. 1990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "C. Mishra, N. Koudas, and C. Zuzarte. Generating targeted queries for database testing. In Proceedings of the 2008 ACMSIGMOD International Conference on Management of Data, SIGMOD ’08, page 499–510, New York, NY, USA, 2008. Association for Computing Machinery. ISBN 9781605581026. doi: 10.1145/ 1376616.1376668. URL https://doi.org/10.1145/1376616.1376668.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "MozillaSecurity. funfuzz. https://github.com/MozillaSecurity/funfuzz, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "S. Nagy and M. Hicks. Full-speed Fuzzing: Reducing Fuzzing Overhead Through Coverage-guided Tracing. In Proceedings of the 40th IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, May 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "R. Padhye, C. Lemieux, K. Sen, M. Papadakis, and Y. Le Traon. Semantic fuzzing with zest. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis, pages 329–340, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "H. Peng, Y. Shoshitaishvili, and M. Payer. T-Fuzz: Fuzzing By Program Transformation. In Proceedings of the 39th IEEE Symposium on Security and Privacy (Oakland) SP1 [8] .",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "K.-T. Rehmann, C. Seo, D. Hwang, B. Truong, A. Böhm, and D. Lee. Performance Monitoring in SAPHANA’s Continuous Integration Process. ACM SIGMETRICSPerformance Evaluation Review, 43:43–52, 02 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "M. Rigger and Z. Su. Testing Database Engines via Pivoted Query Synthesis. arXiv preprint arXiv:2001.04174, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 54,
+      "text": "S. Schumilo, C. Aschermann, R. Gawlik, S. Schinzel, and T. Holz. kAFL: HardwareAssisted Feedback Fuzzing for OS Kernels. In Proceedings of the 26th USENIX Security Symposium (Security), Vancouver, BC, Canada, Aug. 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "K. Serebryany. Sanitize, Fuzz, And Harden Your C++ Code. San Francisco, CA, 2016. USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "D. Slutz. Massive Stochastic Testing of SQL. Technical Report MSR-TR-9821, August 1998. URL https://www.microsoft.com/en-us/research/publication/ massive-stochastic-testing-of-sql/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "N. Stephens, J. Grosen, C. Salls, A. Dutcher, R. Wang, J. Corbetta, Y. Shoshitaishvili, C. Kruegel, and G. Vigna. Driller: Augmenting fuzzing through selective symbolic execution. In NDSS, volume 16, pages 1–16, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "M. Stonebraker, S. Madden, and P. Dubey. Intel \"Big Data\" Science and Technology Center Vision and Execution Plan. ACMSIGMOD Record, 42(1):44–49, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "J. Wang, P. Zhang, L. Zhang, H. Zhu, and X. Ye. A model-based fuzzing approach for dbms. In 2013 8th International Conference on Communications and Networking in China (CHINACOM), pages 426–431, Aug 2013. doi: 10.1109/ChinaCom.2013. 6694634.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "J. Wang, B. Chen, L. Wei, and Y. Liu. Superion: Grammar-Aware Greybox Fuzzing. InProceedings of the 41st International Conference on Software Engineering (ICSE), Montreal, QC, Canada, May 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "G. Wassermann and Z. Su. Sound and Precise Analysis of Web Applications for Injection Vulnerabilities. In Proceedings of the 28th ACMSIGPLAN Conference on Programming Language Design and Implementation, New York, NY, USA, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "R. Xu and S. Vegasena. Vasilisk. https://blog.osiris.cyber.nyu.edu/2019/12/22/ vasilisk/, December 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "W. Xu, S. Kashyap, C. Min, and T. Kim. Designing New Operating Primitives to Improve Fuzzing Performance. In Proceedings of the 24th ACM Conference on Computer and Communications Security (CCS), Dallas, TX, Oct.–Nov. 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "J. Yan, Q. Jin, S. Jain, S. D. Viglas, and A. Lee. Snowtrail: Testing with Production Queries on a Cloud Database. In Proceedings of the Workshop on Testing Database Systems, New York, NY, USA, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "I. Yun, S. Lee, M. Xu, Y. Jang, and T. Kim. QSYM: A Practical Concolic Execution Engine Tailored for Hybrid Fuzzing. In 27th USENIX Security Symposium (USENIX Security 18), Baltimore, MD, Aug. 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "M. Zalewski. American Fuzzy Lop (2.52b). http://lcamtuf.coredump.cx/afl, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "M. Zalewski. Technical \"Whitepaper\" For Afl-fuzz. http://lcamtuf.coredump.cx/ afl/technical_details.txt, 2019. 14 Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback ACMCCS 2020, Orlando, USA Algorithm 3: Dependency Graph Construction. Input: irSet: The set of IRs that needs to be instantiated, relationSet: The set of predefined relations, Output: depencyGrap",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 53,
+      "text": "M. Rigger and Z. Su. Testing Database Engines via Pivoted Query Synthesis. arXiv preprint arXiv:2001.04174, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "prints the arXiv id of the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 53,56,\n61,64]",
+      "technique": "pqs",
+      "sentence": "DBMSs have been heavily tested for logic and performance defects [ 53,56, 61,64].",
+      "context_before": "at hinders the generation of semantics-correct queries. Especially, at the beginning of testing grammar-incorrect queries trigger many new branches in the fault-handling code. The coverage-based feedback guides Sqirrel to focus on these inputs instead of the original semantics-correct queries. Recent works for fuzzing language compilers and interpreters mention a similar observation [ 35,62]. We plan to investigate this problem and develop solutions to mitigate it, like dropping inputs that trigger new branches in short executions. 10 Related Work Detecting Logic and Performance Bugs in DBMSs.",
+      "context_after": "RAGS detects correctness bugs in DBMS through differential testing [ 56]. It generates and executes queries in multiple DBMSs. Any inconsistency among results indicates at least one DBMS contains bugs. SQLancer constructs queries to fetch a randomly selected row from a table [ 53]. The tested DBMS may contain a bug if it fails to fetch the row. QTune [ 41] is a database tuning system based on a deep reinforcement learning model, which can efficiently tune the database configurations for the best performance. Apollo uses differential testing to find performance bugs [ 37]. It generates and run",
+      "section": "8.3 Contributions of Validity and Feedback",
+      "page": 12,
+      "char_offset": 72409,
+      "cited_reference": {
+        "number": 53,
+        "text": "M. Rigger and Z. Su. Testing Database Engines via Pivoted Query Synthesis. arXiv preprint arXiv:2001.04174, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "prints the arXiv id of the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer constructs queries to fetch a randomly selected row from a table [ 53].",
+      "context_before": "t queries. Recent works for fuzzing language compilers and interpreters mention a similar observation [ 35,62]. We plan to investigate this problem and develop solutions to mitigate it, like dropping inputs that trigger new branches in short executions. 10 Related Work Detecting Logic and Performance Bugs in DBMSs. DBMSs have been heavily tested for logic and performance defects [ 53,56, 61,64]. RAGS detects correctness bugs in DBMS through differential testing [ 56]. It generates and executes queries in multiple DBMSs. Any inconsistency among results indicates at least one DBMS contains bugs.",
+      "context_after": "The tested DBMS may contain a bug if it fails to fetch the row. QTune [ 41] is a database tuning system based on a deep reinforcement learning model, which can efficiently tune the database configurations for the best performance. Apollo uses differential testing to find performance bugs [ 37]. It generates and runs queries in two versions of the same DBMS. If two executions take significantly different times, the query triggers a performance regression bug. BmPad runs predefined test suites in the target DBMS, and reports performance bugs if the execution time exceeds the threshold [ 52].Sqi",
+      "section": "8.3 Contributions of Validity and Feedback",
+      "page": 12,
+      "char_offset": 72693,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "Squirrel combines language validity with coverage feedback for DBMS fuzzing. Mutation-based fuzzers fail against DBMSs because inputs are checked strictly for syntax and semantics, while generation-based testing guarantees syntax but ignores coverage feedback. Squirrel keeps SQL in a structural intermediate representation, performs type-based mutations on it, and analyses logical dependencies between arguments to avoid semantic errors. It found 63 bugs across SQLite, MySQL and MariaDB.",
+    "narrative": "SQLancer is cited twice as related work on logic-bug detection, with PQS described as constructing queries to fetch a randomly selected row. Squirrel targets memory errors rather than logic bugs, so the two are complementary rather than compared.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3395032_3395322.json b/_data/papers/paper_doi_10_1145_3395032_3395322.json
new file mode 100644
index 0000000..ea6a5bf
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3395032_3395322.json
@@ -0,0 +1,210 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:27:29Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3395032.3395322",
+    "title": "Testing query execution engines with mutations",
+    "authors": [
+      "Xinyue Chen",
+      "Chenglong Wang",
+      "Alvin Cheung"
+    ],
+    "year": 2020,
+    "venue": "DBTest@SIGMOD",
+    "doi": "10.1145/3395032.3395322",
+    "arxiv_id": null,
+    "s2_paper_id": "a49a96c561c10177764e63817fecf814b04a1739",
+    "url": "https://doi.org/10.1145/3395032.3395322",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3395032.3395322",
+      "retrieved_at": "2026-09-08T15:27:29Z",
+      "chars": 24801,
+      "content_sha256": "sha256:b79b79f8ccfaa0be9c99a63156d835b801a15880903ad48fafd088f80aef3f30"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 5,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1543
+      },
+      {
+        "number": "2",
+        "title": "SYSTEM OVERVIEW",
+        "start": 6328
+      },
+      {
+        "number": "3",
+        "title": "INSIDE MUTASQL",
+        "start": 11643
+      },
+      {
+        "number": "4",
+        "title": "EXPERIMENTS",
+        "start": 16600
+      },
+      {
+        "number": "4.1",
+        "title": "Reproducing Known SQLite Bugs",
+        "start": 17308
+      },
+      {
+        "number": "4.2",
+        "title": "Discovering New Bugs in SQLite 3.31.1",
+        "start": 20101
+      },
+      {
+        "number": "5",
+        "title": "RELATED WORK",
+        "start": 21648
+      },
+      {
+        "number": "6",
+        "title": "CONCLUSION",
+        "start": 23409
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. 2007. A genetic approach for random testing of database systems. In Proceedings of the 33rd international conference on Very large data bases. VLDB Endowment, 1243–1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. In ACMSIGPLAN Notices, Vol. 49. ACM, 216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. arXiv:cs.DB/2001.04174",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, August 24-27, 1998, New York City, New York, USA. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. 2016. Finding compiler bugs via live code mutation. In ACMSIGPLAN Notices, Vol. 51. ACM, 849–863.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 3,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. arXiv:cs.DB/2001.04174",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "prints the arXiv id of the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Meanwhile, pivot query synthesis (PQS) [ 3] is a new testing algorithm that also utilizes equivalence mutation so that the mutated query contains a specific row in the original output.",
+      "context_before": "stochastic or genetic algorithms to randomly generate queries to test relational database systems [ 1,4]. In RAGS [ 4], queries are automatically generated and evaluated by multiple database systems. If the results are inconsistent, a bug has been found. While this method can be effective, it requires that the tested features be compatible with all tested query engines, and thus limited to only engine-specific SQL features that can be tested. MutaSQL generalizes this testing approach by comparing query results within the same database system so that the testing result can be directly verified.",
+      "context_after": "In comparison, MutaSQL focuses more on structural mutations (e.g., insert and removal of subqueries) while PQS focuses on predicate level mutation (e.g., arithmetic expression, string functions). The two approaches can potentially be combined to enlarge the design of richer mutation rules for better coverage. Outside of database engine testing, equivalence mutation has also been previously used in C compiler testing [ 2,5]. Comparing to random testing, equivalence mutation has the advantage of exploring a larger testing space, in. We particular complex control flow structures, and thus often",
+      "section": "5 RELATED WORK",
+      "page": 4,
+      "char_offset": 22319,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", insert and removal of subqueries) while PQS focuses on predicate level mutation (e.",
+      "context_before": "nd. While this method can be effective, it requires that the tested features be compatible with all tested query engines, and thus limited to only engine-specific SQL features that can be tested. MutaSQL generalizes this testing approach by comparing query results within the same database system so that the testing result can be directly verified. Meanwhile, pivot query synthesis (PQS) [ 3] is a new testing algorithm that also utilizes equivalence mutation so that the mutated query contains a specific row in the original output. In comparison, MutaSQL focuses more on structural mutations (e.g.",
+      "context_after": "g., arithmetic expression, string functions). The two approaches can potentially be combined to enlarge the design of richer mutation rules for better coverage. Outside of database engine testing, equivalence mutation has also been previously used in C compiler testing [ 2,5]. Comparing to random testing, equivalence mutation has the advantage of exploring a larger testing space, in. We particular complex control flow structures, and thus often achieves better coverage in the testing process. We believe such strategies can also be adopted for query engine testing. 4http://sqlite.1065341.n5.nab",
+      "section": "5 RELATED WORK",
+      "page": 4,
+      "char_offset": 22570,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:52Z",
+    "is_model_written": true,
+    "summary": "This short paper tests query execution engines by mutating queries in ways that preserve or predictably change their results, focusing on structural mutations such as inserting and removing subqueries. Published in 2020, it is contemporaneous with the first SQLancer oracles.",
+    "narrative": "PQS is discussed in related work as the closest neighbour: the paper notes it also uses equivalence mutation, arranging for the mutated query to contain a specific row from the original output. The distinction drawn is one of level -- this work mutates query structure, while PQS mutates at the predicate level.",
+    "roles": {
+      "M1": "definition",
+      "M2": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3428261.json b/_data/papers/paper_doi_10_1145_3428261.json
new file mode 100644
index 0000000..b437c3c
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3428261.json
@@ -0,0 +1,165 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:13:10Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3428261",
+    "title": "On the unusual effectiveness of type-aware operator mutations for testing SMT solvers",
+    "authors": [
+      "Dominik Winterer",
+      "Chengyu Zhang",
+      "Zhendong Su"
+    ],
+    "year": 2020,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3428261",
+    "arxiv_id": "2004.08799v4",
+    "s2_paper_id": "ab6addb4b50a1af3f78fea402980648135d3b8c9",
+    "url": "https://doi.org/10.1145/3428261",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2004.08799v4",
+      "retrieved_at": "2026-09-07T17:13:10Z",
+      "chars": 87515,
+      "content_sha256": "sha256:0bb36a019f27558373f2a42d3d2b0dac566368eff5152e0cad2c882e33ee9afc"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 25,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2078
+      },
+      {
+        "number": "2",
+        "title": "MOTIVATING EXAMPLES",
+        "start": 14431
+      },
+      {
+        "number": "3",
+        "title": "APPROACH",
+        "start": 19857
+      },
+      {
+        "number": "3.1",
+        "title": "Type-Aware Operator Mutation",
+        "start": 22931
+      },
+      {
+        "number": "3.2",
+        "title": "OpFuzz",
+        "start": 24319
+      },
+      {
+        "number": "4",
+        "title": "EMPIRICAL EVALUATION",
+        "start": 27005
+      },
+      {
+        "number": "4.1",
+        "title": "Evaluation Setup",
+        "start": 28562
+      },
+      {
+        "number": "4.2",
+        "title": "Evaluation Results",
+        "start": 33607
+      },
+      {
+        "number": "5.1",
+        "title": "Quantitative Analysis",
+        "start": 45917
+      },
+      {
+        "number": "5.2",
+        "title": "Insights",
+        "start": 57533
+      },
+      {
+        "number": "5.3",
+        "title": "Assorted Bug Samples",
+        "start": 60282
+      }
+    ]
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "author_year_citation",
+      "surface": "Rigger and Su 2020",
+      "technique": null,
+      "sentence": "2008; Rigger and Su 2020; Seltenreich 2020], and testing OS kernel [Corina et al .",
+      "context_before": "ey difference to type-aware operator mutation, they both need to fully parse the program and work on the abstract syntax tree level, which may lead to a higher computational cost during fuzzing. Type-aware operator mutation, on the other hand, works on the token level and without fully parsing the formula. Besides general black-box and grey-box fuzzing, various domain-specific fuzzing approaches exist, e.g., for testing compilers [Cummins et al .2018; Le et al .2014; Lidbury et al .2015; Yang et al. 2011; Zhang et al .2017], testing database management systems [Jung et al .2019; Mishra et al .",
+      "context_after": "2017; Han and Cha 2017; Schumilo et al .2017]. Type-aware operator mutation is also a domain-specific fuzzing technique which is unusually effective for testing SMT solvers. 7 CONCLUSION We introduced type-aware operator mutation, a simple and effective approach for stress-testing SMT solvers. We realized type-aware operator mutation in our testing tool OpFuzz in little more than 200LoC, supporting only the most basic operators of the SMT-LIB language. Despite this, OpFuzz found 819confirmed unique bugs (685 fixed) in Z3 and CVC4. These bug findings are highly diverse, ranging over various typ",
+      "section": "5.3 Assorted Bug Samples",
+      "page": 22,
+      "char_offset": 76550,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "Type-aware operator mutation tests SMT solvers by mutating operators of conforming types within seed formulas to produce well-typed mutants used as test cases. Realised in OpFuzz and applied to Z3 and CVC4, it proved unusually effective: over a year the authors reported 1,092 bugs, 819 confirmed and 685 fixed, spanning soundness bugs, invalid model bugs and crashes. A follow-up study found the bugs were high quality, often affecting core components.",
+    "narrative": "SQLancer appears once, in a list of domains where testing has been applied, cited alongside work on SMT solvers and OS kernels. Nothing in the paper connects to it beyond that.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3448016_3457559.json b/_data/papers/paper_doi_10_1145_3448016_3457559.json
new file mode 100644
index 0000000..6de2787
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3448016_3457559.json
@@ -0,0 +1,641 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:33:45Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3448016.3457559",
+    "title": "FoundationDB: A Distributed Unbundled Transactional Key Value Store",
+    "authors": [
+      "Jingyu Zhou",
+      "Meng Xu",
+      "A. Shraer",
+      "B. Namasivayam",
+      "Alex Miller",
+      "Evan Tschannen",
+      "Steve Atherton",
+      "Andrew J. Beamon",
+      "Rusty Sears",
+      "J. Leach",
+      "D. Rosenthal",
+      "X. Dong",
+      "Willie B. Wilson",
+      "Ben Collins",
+      "David Scherer",
+      "Alec Grieser",
+      "Young Liu",
+      "Alvin Moore",
+      "Bhaskar Muppana",
+      "Xi-sheng Su",
+      "Vishesh Yadav"
+    ],
+    "year": 2021,
+    "venue": "SIGMOD Conference",
+    "doi": "10.1145/3448016.3457559",
+    "arxiv_id": null,
+    "s2_paper_id": "2e2a31b0bb5c2ba3c7f4cf429e2dc81d4327af9d",
+    "url": "https://doi.org/10.1145/3448016.3457559",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3448016.3457559",
+      "retrieved_at": "2026-09-08T14:33:45Z",
+      "chars": 88110,
+      "content_sha256": "sha256:7e2ae2df034e377278a8e8ea67600c2fa7dddcdfef31639b9e166db03b679897"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 14,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2898
+      },
+      {
+        "number": "2",
+        "title": "Design",
+        "start": 10299
+      },
+      {
+        "number": "2.1",
+        "title": "Design Principles",
+        "start": 10890
+      },
+      {
+        "number": "2.2",
+        "title": "System Interface",
+        "start": 12822
+      },
+      {
+        "number": "2.3",
+        "title": "Architecture",
+        "start": 13892
+      },
+      {
+        "number": "2.3.2",
+        "title": "Data Plane FDB targets OLTP workloads that are read-mostly,",
+        "start": 15139
+      },
+      {
+        "number": "2.3.4",
+        "title": "Bootstrapping FDB has no external dependency on other",
+        "start": 17680
+      },
+      {
+        "number": "2.3.5",
+        "title": "Reconfiguration Whenever there is a failure in the TS or",
+        "start": 18607
+      },
+      {
+        "number": "2.4",
+        "title": "Transaction Management",
+        "start": 19382
+      },
+      {
+        "number": "2.4.1",
+        "title": "End-to-end Transaction Processing As illustrated in Figure 1, a",
+        "start": 19558
+      },
+      {
+        "number": "2.4.2",
+        "title": "Support Strict Serializability FDB implements Serializable",
+        "start": 22027
+      },
+      {
+        "number": "2.4.4",
+        "title": "Transaction System Recovery Traditional database systems",
+        "start": 28880
+      },
+      {
+        "number": "2.5",
+        "title": "Replication",
+        "start": 33901
+      },
+      {
+        "number": "2.6",
+        "title": "Other Optimizations",
+        "start": 36664
+      },
+      {
+        "number": "3",
+        "title": "Geo-replication and failover",
+        "start": 38550
+      },
+      {
+        "number": "4",
+        "title": "Simulation Testing",
+        "start": 45504
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 52742
+      },
+      {
+        "number": "5.1",
+        "title": "Production Measurement",
+        "start": 52932
+      },
+      {
+        "number": "5.2",
+        "title": "Scalability Test",
+        "start": 55613
+      },
+      {
+        "number": "5.3",
+        "title": "Reconfiguration Duration",
+        "start": 59585
+      },
+      {
+        "number": "6",
+        "title": "Lessons Learned",
+        "start": 60731
+      },
+      {
+        "number": "6.1",
+        "title": "Architecture Design",
+        "start": 60925
+      },
+      {
+        "number": "6.2",
+        "title": "Simulation Testing",
+        "start": 61948
+      },
+      {
+        "number": "6.3",
+        "title": "Fast Recovery",
+        "start": 63985
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 66886
+      },
+      {
+        "number": "8",
+        "title": "Conclusions",
+        "start": 71611
+      },
+      {
+        "number": "S",
+        "title": "Sivasubramanian, P. Vosshall, and W. Vogels. Dynamo: amazon’s highly",
+        "start": 78730
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Couchbase. https://www.couchbase.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "CouchDB. https://couchdb.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Distributed system safety research. https://jepsen.io/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Flow. https://github.com/apple/foundationdb/tree/master/flow.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "FoundationDB. https://github.com/apple/foundationdb.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "FoundationDB Document Layer. https://github.com/FoundationDB/fdbdocument-layer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "FoundationDB Joshua. https://github.com/FoundationDB/fdb-joshua.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Foundationdb storage adapter for janusgraph. https://github.com/JanusGraph/ janusgraph-foundationdb.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Janusgraph. https://janusgraph.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "MongoDB. https://www.mongodb.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Tephra: Transactions for Apache HBase. http://tephra.incubator.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "D. Abadi. Consistency tradeoffs in modern distributed database system design: Cap is only part of the story. Computer, 45(2):37–42, Feb. 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "G. Agha. Actors: A Model of Concurrent Computation in Distributed Systems. MIT Press, Cambridge, MA, USA, 1986.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "A. Alquraan, H. Takruri, M. Alfatafta, and S. Al-Kiswany. An analysis of networkpartitioning failures in cloud systems. In Proceedings of the 13th USENIX Conference on Operating Systems Design and Implementation, OSDI’18, pages 51–68, Carlsbad, CA, USA, 2018. USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "J. Arulraj, M. Perron, and A. Pavlo. Write-behind logging. Proceedings of the VLDB Endowment, 10:337–348, November 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "M. Balakrishnan, D. Malkhi, T. Wobber, M. Wu, V. Prabhakaran, M. Wei, J. D. Davis, S. Rao, T. Zou, and A. Zuck. Tango: Distributed data structures over a shared log. In Proceedings of the Twenty-Fourth ACM Symposium on Operating Systems Principles, pages 325–340, Farminton, PA, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "H. Bati, L. Giakoumakis, S. Herbert, and A. Surna. A genetic approach for random testing of database systems. In Proceedings of the 33rd International Conference on Very Large Data Bases, VLDB ’07, pages 1243–1251, Vienna, Austria, 2007. VLDB Endowment.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "P. A. Bernstein, V. Hadzilacos, and N. Goodman. Concurrency Control and Recovery in Database Systems. Addison-Wesley, 1987.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "P. A. Bernstein, C. W. Reid, and S. Das. Hydera transactional record manager for shared flash. In CIDR, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "E. Bortnikov, E. Hillel, I. Keidar, I. Kelly, M. Morel, S. Paranjpye, F. Perez-Sorrosal, and O. Shacham. Omid, reloaded: Scalable and highly-available transaction processing. In 15th USENIX Conference on File and Storage Technologies (FAST 17), pages 167–180, Santa Clara, CA, Feb. 2017. USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "L. Braun, T. Etter, G. Gasparis, M. Kaufmann, D. Kossmann, D. Widmer, A. Avitzur, A. Iliopoulos, E. Levy, and N. Liang. Analytics in motion: High performance eventprocessing and real-time analytics in the same database. In Proceedings of the 2015 ACMSIGMOD International Conference on Management of Data, SIGMOD ’15, page 251–264, New York, NY, USA, 2015. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "L. Camargos, F. Pedone, and M. Wieloch. Sprint: A middleware for high-performance transaction processing. In Proceedings of the 2nd ACMSIGOPS/EuroSys European Conference on Computer Systems 2007, pages 385–398, Lisbon, Portugal, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "W. Cao, Z. Liu, P. Wang, S. Chen, C. Zhu, S. Zheng, Y. Wang, and G. Ma. Polarfs: An ultra-low latency and failure resilient distributed file system for shared storage cloud database. Proceedings of the VLDB Endowment, 11(12):1849–1862, Aug. 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "S. Chandrasekaran and R. Bamford. Shared cachethe future of parallel databases. InProceedings 19th International Conference on Data Engineering, pages 840–850, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "F. Chang, J. Dean, S. Ghemawat, W. C. Hsieh, D. A. Wallach, M. Burrows, T. Chandra, A. Fikes, and R. E. Gruber. Bigtable: A distributed storage system for structured data. In Proceedings of the 7th USENIX Symposium on Operating Systems Design and Implementation (OSDI’06), 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "H. Chen, R. Chen, X. Wei, J. Shi, Y. Chen, Z. Wang, B. Zang, and H. Guan. Fast in-memory transaction processing using rdma and htm. ACM Transactions on Computer Systems, 35(1), July 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "G. Chockler and D. Malkhi. Active disk paxos with infinitely many processes. InProceedings of the Twenty-first Annual Symposium on Principles of Distributed Computing, PODC’02, pages 78–87, New York, NY, USA, 2002. ACM.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "C. Chrysafis, B. Collins, S. Dugas, J. Dunkelberger, M. Ehsan, S. Gray, A. Grieser, O. Herrnstadt, K. Lev-Ari, T. Lin, M. McMahon, N. Schiefer, and A. Shraer. FoundationDB Record Layer: A Multi-Tenant Structured Datastore. In Proceedings of the 2019 International Conference on Management of Data, SIGMOD’19, pages 1787–1802, New York, NY, USA, 2019. ACM.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "A. Cidon, S. Rumble, R. Stutsman, S. Katti, J. Ousterhout, and M. Rosenblum. Copysets: Reducing the frequency of data loss in cloud storage. In Presented as part of the 2013 USENIX Annual Technical Conference, pages 37–48, San Jose, CA, 2013. USENIX.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "B. F. Cooper, R. Ramakrishnan, U. Srivastava, A. Silberstein, P. Bohannon, H.-A. Jacobsen, N. Puz, D. Weaver, and R. Yerneni. PNUTS: Yahoo!’s hosted data serving platform. Proceedings of the VLDB Endowment, 1(2):1277–1288, Aug. 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "J. C. Corbett, J. Dean, M. Epstein, A. Fikes, C. Frost, J. Furman, S. Ghemawat, A. Gubarev, C. Heiser, P. Hochschild, W. Hsieh, S. Kanthak, E. Kogan, H. Li, A. Lloyd, S. Melnik, D. Mwaura, D. Nagle, S. Quinlan, R. Rao, L. Rolig, Y. Saito, M. Szymaniak, C. Taylor, R. Wang, and D. Woodford. Spanner: Google’s globallydistributed database. In 10th USENIX Symposium on Operating Systems Design and Imple",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "J. Dean and L. A. Barroso. The tail at scale. Communications of the ACM, 56(2):74–80, Feb. 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "G. DeCandia, D. Hastorun, M. Jampani, G. Kakulapati, A. Lakshman, A. Pilchin, S. Sivasubramanian, P. Vosshall, and W. Vogels. Dynamo: amazon’s highly available key-value store. ACMSIGOPS Operating Systems Review, 41(6):205–220, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "C. Diaconu, C. Freedman, E. Ismert, P.-A. Larson, P. Mittal, R. Stonecipher, N. Verma, and M. Zwilling. Hekaton: Sql server’s memory-optimized oltp engine. InProceedings of the 2013 ACMSIGMOD International Conference on Management of Data, SIGMOD ’13, pages 1243–1254, New York, NY, USA, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "A. Dragojević, D. Narayanan, O. Hodson, and M. Castro. Farm: Fast remote memory. In Proceedings of the 11th USENIX Symposium on Networked Systems Design and Implementation, pages 401–414, Seattle, WA, USA, April 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "A. Dragojević, D. Narayanan, E. B. Nightingale, M. Renzelmann, A. Shamis, A. Badam, and M. Castro. No compromises: Distributed transactions with consistency, availability, and performance. In Proceedings of the 25th Symposium on Operating Systems Principles, SOSP ’15, pages 54–70, New York, NY, USA, 2015. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "I. Eyal, K. P. Birman, I. Keidar, and R. van Renesse. Ordering transactions with prediction in distributed object stores. In LADIS, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Facebook. Rocksdb. https://rocksdb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "A. Ganesan, R. Alagappan, A. C. Arpaci-Dusseau, and R. H. Arpaci-Dusseau. Redundancy does not imply fault tolerance: Analysis of distributed storage reactions to single errors and corruptions. In Proceedings of the 15th Usenix Conference on File and Storage Technologies, FAST’17, pages 149–165, Santa clara, CA, USA, 2017. USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "A. Groce, C. Zhang, E. Eide, Y. Chen, and J. Regehr. Swarm testing. In Proceedings of the 2012 International Symposium on Software Testing and Analysis, ISSTA 2012, pages 78–88, Minneapolis, MN, USA, 2012. ACM.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "R. D. Hipp. SQLite. https://www.sqlite.org/index.html, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Y. Huang, C. Kintala, N. Kolettis, and N. D. Fulton. Software rejuvenation: analysis, module and applications. In Twenty-Fifth International Symposium on Fault-Tolerant Computing, pages 381–390, 1995.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "J. W. Josten, C. Mohan, I. Narang, and J. Z. Teng. Db2’s use of the coupling facility for data sharing. IBM Systems Journal, 36(2):327–351, 1997.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "H. T. Kung and J. T. Robinson. On optimistic methods for concurrency control. ACM Transactions on Database Systems, 6(2):213–226, June 1981.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "A. Lakshman and P. Malik. Cassandra: structured storage system on a p2p network. In Proceedings of the 28th ACM symposium on Principles of Distributed Computing, page 5, Jan 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "T. Leesatapornwongsa, M. Hao, P. Joshi, J. F. Lukman, and H. S. Gunawi. Samc: Semantic-aware model checking for fast discovery of deep bugs in cloud systems. In Proceedings of the 11th USENIX Conference on Operating Systems Design and Implementation, OSDI’14, pages 399–414, Broomfield, CO, 2014. USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "K. Lev-Ari, Y. Tian, A. Shraer, C. Douglas, H. Fu, A. Andreev, K. Beranek, S. Dugas, A. Grieser, and J. Hemmo. Quick: a queuing system in cloudkit. In Proceedings of the 2021 International Conference on Management of Data, SIGMOD Conference 2021, Xian, Shanxi, China, June 03 — 05, 2021. ACM, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "J. Levandoski, D. Lomet,, and K. K. Zhao. Deuteronomy: Transaction support for cloud data. In Conference on Innovative Data Systems Research (CIDR). www.crdrdb.org, January 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "S. Loesing, M. Pilman, T. Etter, and D. Kossmann. On the design and scalability of distributed shared-data databases. In SIGMOD Conference, pages 663–676. ACM, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "D. Lomet, A. Fekete, G. Weikum, and M. J. Zwilling. Unbundling transaction services in the cloud. In Fourth Biennial Conference on Innovative Data Systems Research, Asilomar, CA, Jan. 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "D. Lomet and M. F. Mokbel. Locking key ranges with unbundled transaction services. Proceedings of the VLDB Endowment, 2(1):265–276, August 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "N. Malviya, A. Weisberg, S. Madden, and M. Stonebraker. Rethinking main memory oltp recovery. In pages 604–615, Chicago, IL, USA, March 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "C. Mohan, D. Haderle, B. G. Lindsay, H. Pirahesh, and P. M. Schwarz. Aries: A transaction recovery method supporting fine-granularity locking and partial rollbacks using write-ahead logging. ACM Transactions on Database Systems, 17(1):94–162, 1992.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "D. Ongaro, S. M. Rumble, R. Stutsman, J. Ousterhout, and M. Rosenblum. Fast crash recovery in ramcloud. In Proceedings of the Twenty-Third ACM Symposium on Operating Systems Principles, SOSP ’11, page 29–41, New York, NY, USA, 2011. Association for Computing Machinery. Industrial Track Paper SIGMOD ’21, June 20–25, 2021, Virtual Event, China 2665",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "D. Peng and F. Dabek. Large-scale incremental processing using distributed transactions and notifications. In Proceedings of the 9th USENIX Symposium on Operating Systems Design and Implementation, OSDI’10, pages 251–264, Vancouver, BC, Canada, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "W. Pugh. Skip lists: A probabilistic alternative to balanced trees. Communications of the ACM, 33(6):668–676, 1990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "M. Rigger and Z. Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pages 667–682. USENIX Association, Nov. 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 58,
+      "text": "O. Shacham, Y. Gottesman, A. Bergman, E. Bortnikov, E. Hillel, and I. Keidar. Taking omid to the clouds: Fast, scalable transactions for real-time cloud analytics. Proceedings of the VLDB Endowment, 11(12):1795–1808, Aug. 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "A. Shraer, A. Aybes, B. Davis, C. Chrysafis, D. Browning, E. Krugler, E. Stone, H. Chandler, J. Farkas, J. Quinn, J. Ruben, M. Ford, M. McMahon, N. Williams, N. Favre-Felix, N. Sharma, O. Herrnstadt, P. Seligman, R. Pisolkar, S. Dugas, S. Gray, S. Lu, S. Harkema, V. Kravtsov, V. Hong, Y. Tian, and W. L. Yih. Cloudkit: Structured storage for mobile applications. Proceedings of the VLDB Endowment, 1",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "D. R. Slutz. Massive stochastic testing of SQL. In A. Gupta, O. Shmueli, and J. Widom, editors, VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, August 24-27, 1998, New York City, New York, USA, pages 618–622. Morgan Kaufmann, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "M. Stonebraker, S. Madden, D. J. Abadi, S. Harizopoulos, N. Hachem, and P. Helland. The end of an architectural era: (it’s time for a complete rewrite). In Proceedings of the 33rd International Conference on Very Large Data Bases, VLDB ’07, pages 1150–1160. VLDB Endowment, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "R. Taft, I. Sharif, A. Matei, N. VanBenschoten, J. Lewis, T. Grieger, K. Niemi, A. Woods, A. Birzin, R. Poss, P. Bardea, A. Ranade, B. Darnell, B. Gruneir, J. Jaffray, L. Zhang, and P. Mattis. Cockroachdb: The resilient geo-distributed SQL database. In Proceedings of the 2020 International Conference on Management of Data, SIGMOD, pages 1493–1509, Portland, OR, USA, June 2020. ACM.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "C. Tan, C. Zhao, S. Mu, and M. Walfish. Cobra: Making transactional key-value stores verifiably serializable. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pages 63–80. USENIX Association, Nov.2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "A. Thomson, T. Diamond, S.-C. Weng, K. Ren, P. Shao, and D. Abadi. Calvin: Fast distributed transactions for partitioned database systems. In Proceedings of the ACMSIGMOD International Conference on Management of Data, pages 1–12, May 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "S. Tu, W. Zheng, E. Kohler, B. Liskov, and S. Madden. Speedy transactions in multicore in-memory databases. In Proceedings of the Twenty-Fourth ACM Symposium on Operating Systems Principles, pages 18–32, Farminton, PA, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "A. Verbitski, X. Bao, A. Gupta, D. Saha, M. Brahmadesam, K. Gupta, R. Mittal, S. Krishnamurthy, S. Maurice, and T. Kharatishvili. Amazon aurora: Design considerations for high throughput cloud-native relational databases. In Proceedings of the 2017 ACM International Conference on Management of Data, SIGMOD ’17, pages 1041–1052, Chicago, IL, USA, May 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "A. Verbitski, A. Gupta, D. Saha, J. Corey, K. Gupta, M. Brahmadesam, R. Mittal, S. Krishnamurthy, S. Maurice, T. Kharatishvilli, and X. Bao. Amazon aurora: On avoiding distributed consensus for i/os, commits, and membership changes. In Proceedings of the 2018 International Conference on Management of Data, SIGMOD ’18, pages 789–796, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "M. Yabandeh and D. Gómez Ferro. A Critique of Snapshot Isolation. In Proceedings of the 7th ACM European Conference on Computer Systems, EuroSys’12, pages 155–168, Bern, Switzerland, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "J. Yang, T. Chen, M. Wu, Z. Xu, X. Liu, H. Lin, M. Yang, F. Long, L. Zhang, and L. Zhou. MODIST: transparent model checking of unmodified distributed systems. InProceedings of the 6th USENIX Symposium on Networked Systems Design and Implementation, NSDI’09, pages 213–228, Boston, Massachusetts, 2009. USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "M. Zheng, J. Tucek, D. Huang, F. Qin, M. Lillibridge, E. S. Yang, B. W. Zhao, and S. Singh. Torturing databases for fun and profit. In Proceedings of the 11th USENIX Conference on Operating Systems Design and Implementation, OSDI’14, pages 449–464, Broomfield, CO, 2014. USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "T. Zhu, Z. Zhao, F. Li, W. Qian, A. Zhou, D. Xie, R. Stutsman, H. Li, and H. Hu. Solar: Towards a shared-everything database on distributed log-structured storage. In USENIX Annual Technical Conference, pages 795–807. USENIX Association, 2018. Industrial Track Paper SIGMOD ’21, June 20–25, 2021, Virtual Event, China 2666",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 57,
+      "text": "M. Rigger and Z. Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pages 667–682. USENIX Association, Nov. 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 17,57,60]",
+      "technique": "pqs",
+      "sentence": "Finally there are numerous approaches to testing the correctness of database subsystems in the absence of faults, including the query engine [ 17,57,60] and concurrency control mechanism [ 63].",
+      "context_before": "esting. Non-deterministic fault-injection has been widely used in the testing of distributed systems. Various approaches have included faults such as network partition [ 14], power failures [ 70], and storage faults [ 39]. Jepsen [ 3] is a commercial effort that has tested a large number of distributed databases. All of these approaches lack deterministic reproducibility. Model checking has also been widely-applied to distributed systems [ 46,69]. While model checking can be more exhaustive than simulation, it can only verify the correctness of a model rather than of the actual implementation.",
+      "context_after": "The FDB deterministic simulation approach allows verification of database invariants and other properties against the real database code, together with deterministic reproducibility. 8 Conclusions This paper presents FoundationDB, a key value store designed for OLTP cloud services. The main idea is to decouple transaction processing from logging and storage. Such an unbundled architecture enables the separation and horizontal scaling of both read and write handling. The transaction system combines OCC and MVCC to ensure strict serializability. The decoupling of logging and the determinism in",
+      "section": "7 Related Work",
+      "page": 12,
+      "char_offset": 71233,
+      "cited_reference": {
+        "number": 57,
+        "text": "M. Rigger and Z. Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pages 667–682. USENIX Association, Nov. 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:03:26Z",
+    "is_model_written": true,
+    "summary": "The FoundationDB paper describes a distributed transactional key-value store built as unbundled components, and is best known for its deterministic simulation testing: the whole system runs in a single-threaded simulation where faults are injected reproducibly, so distributed bugs can be found and replayed exactly.",
+    "narrative": "SQLancer's work is cited once, in related work, as part of the body of approaches for testing the correctness of database subsystems in the absence of faults -- the query engine specifically. That is the contrast the paper draws with its own concern, which is correctness under injected faults. The mention is reachable only through the citation marker.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3468264_3468540.json b/_data/papers/paper_doi_10_1145_3468264_3468540.json
new file mode 100644
index 0000000..423894c
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3468264_3468540.json
@@ -0,0 +1,523 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:27:24Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3468264.3468540",
+    "title": "Skeletal approximation enumeration for SMT solver testing",
+    "authors": [
+      "Peisen Yao",
+      "Heqing Huang",
+      "Wensheng Tang",
+      "Qingkai Shi",
+      "Rongxin Wu",
+      "Charles Zhang"
+    ],
+    "year": 2021,
+    "venue": "ESEC/SIGSOFT FSE",
+    "doi": "10.1145/3468264.3468540",
+    "arxiv_id": null,
+    "s2_paper_id": "e6a4ddc65452fb2a01389b6e28e6607b8781710e",
+    "url": "https://doi.org/10.1145/3468264.3468540",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3468264.3468540",
+      "retrieved_at": "2026-09-08T15:27:24Z",
+      "chars": 80216,
+      "content_sha256": "sha256:03f59e372c08721e6f612e739a97a92ee606d9021f7a86c3c8064a9d5ed71381"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1950
+      },
+      {
+        "number": "2",
+        "title": "PRELIMINARIES",
+        "start": 8595
+      },
+      {
+        "number": "3",
+        "title": "PROBLEM FORMULATION",
+        "start": 12574
+      },
+      {
+        "number": "3.1",
+        "title": "Approximation-based Metamorphic",
+        "start": 12783
+      },
+      {
+        "number": "3.2",
+        "title": "Skeletal Approximation Enumeration",
+        "start": 14212
+      },
+      {
+        "number": "4",
+        "title": "APPROACH",
+        "start": 17824
+      },
+      {
+        "number": "4.1",
+        "title": "Approximation from CNF",
+        "start": 18135
+      },
+      {
+        "number": "4.2",
+        "title": "Literal-level Mutation Strategies",
+        "start": 21697
+      },
+      {
+        "number": "5",
+        "title": "EVALUATION",
+        "start": 40747
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Setup",
+        "start": 41090
+      },
+      {
+        "number": "5.2",
+        "title": "Results of Bug Finding",
+        "start": 43118
+      },
+      {
+        "number": "5.3",
+        "title": "Comparison to Existing Techniques",
+        "start": 46430
+      },
+      {
+        "number": "5.4",
+        "title": "Threats to Validity",
+        "start": 53601
+      },
+      {
+        "number": "6",
+        "title": "DISCUSSION",
+        "start": 54335
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2014. AFL: american fuzzy lop. http://lcamtuf.coredump.cx/a/. Accessed: 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2021. dd SMT. https://github.com/aniemetz/dd SMT.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2021. pyDelta. https://github.com/nafur/pydelta.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2021. SMT-COMP. https://smt-comp.github.io/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2021. Z3 test scripts. https://github.com/Z3Prover/z3/src/test.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2021. CVC4 regression test suit. https://github.com/CVC4/CVC4/test/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2021. Yices2 regression test suit. https://github.com/SRI-CSL/yices2/tests.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2021. OpenSMT regression test suit. https://github.com/usi-verication-andsecurity/opensmt/regression.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2021. https://smtfuzz.github.io/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Cornelius Aschermann, Tommaso Frassetto, Thorsten Holz, Patrick Jauernig, Ahmad-Reza Sadeghi, and Daniel Teuchert. 2019. NAUTILUS: Fishing for Deep Bugs with Grammars. In 26th Annual Network and Distributed System Security Symposium, NDSS 2019, San Diego, California, USA, February 24-27, 2019. The Internet Society. https://www.ndss-symposium.org/ndss-paper/nautilus-shingfor-deep-bugs-with-grammars",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Clark Barrett, Christopher Conway, Morgan Deters, Liana Hadarean, Dejan Jovanović, Tim King, Andrew Reynolds, and Cesare Tinelli. 2011. CVC4. In Proceedings of the 23rd International Conference on Computer Aided Verication (Snowbird, UT) (CAV’11). Springer-Verlag, Berlin, Heidelberg, 171–177. http: //dl.acm.org/citation.cfm?id=2032305.2032319",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Clark Barrett, Aaron Stump, and Cesare Tinelli. 2010. The satisability modulo theories library (SMT-LIB). www. SMT-LIB. org 15 (2010), 18–52.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Karthikeyan Bhargavan, Barry Bond, Antoine Delignat-Lavaud, Cédric Fournet, Chris Hawblitzel, Catalin Hritcu, Samin Ishtiaq, Markulf Kohlweiss, Rustan Leino, Jay R. Lorch, Kenji Maillard, Jianyang Pan, Bryan Parno, Jonathan Protzenko, Tahina Ramananandro, Ashay Rane, Aseem Rastogi, Nikhil Swamy, Laure Thompson, Peng Wang, Santiago Zanella Béguelin, and Jean Karim Zinzindohoue. 2017. Everest: Towar",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Tim Blazytko, Moritz Contag, Cornelius Aschermann, and Thorsten Holz. 2017. Syntia: Synthesizing the Semantics of Obfuscated Code. In 26th USENIX Security Symposium, USENIX Security 2017, Vancouver, BC, Canada, August 16-18, 2017, Engin Kirda and Thomas Ristenpart (Eds.). USENIX Association, 643– 659. https://www.usenix.org/conference/usenixsecurity17/technical-sessions/ presentation/blazytko",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Dmitry Blotsky, Federico Mora, Murphy Berzish, Yunhui Zheng, Ifaz Kabir, and Vijay Ganesh. 2018. StringFuzz: A Fuzzer for String Solvers. In Computer Aided Verication-30th International Conference, CAV 2018, Held as Part of the Federated Logic Conference, FloC 2018, Oxford, UK, July 14-17, 2018, Proceedings, Part II (Lecture Notes in Computer Science, Vol. 10982), Hana Chockler and Georg Weissenba",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Cristina Borralleras, Daniel Larraz, Enric Rodríguez-Carbonell, Albert Oliveras, and Albert Rubio. 2019. Incomplete SMT Techniques for Solving Non-Linear Formulas over the Integers. ACM Trans. Comput. Log. 20, 4 (2019), 25:1–25:36. https://doi.org/10.1145/3340923",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Aaron R. Bradley and Zohar Manna. 2007. The calculus of computationdecision procedures with applications to verication. Springer. https://doi.org/10.1007/9783-540-74113-8",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Robert Brummayer and Armin Biere. 2009. Eective Bit-Width and UnderApproximation. In Computer Aided Systems Theory-EUROCAST 2009, 12th International Conference, Las Palmas de Gran Canaria, Spain, February 15-20, 2009, Revised Selected Papers (Lecture Notes in Computer Science, Vol. 5717), Roberto Moreno-Díaz, Franz Pichler, and Alexis Quesada-Arencibia (Eds.). Springer, 304– 311. https://doi.org/1",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Robert Brummayer and Armin Biere. 2009. Fuzzing and Delta-Debugging SMT Solvers. In Proceedings of the 7th International Workshop on Satisability Modulo Theories (Montreal, Canada) (SMT ’09). Association for Computing Machinery, New York, NY, USA, 1–5. https://doi.org/10.1145/1670412.1670413",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Roberto Bruttomesso, Edgar Pek, Natasha Sharygina, and Aliaksei Tsitovich. 2010. The OpenSMT Solver. In Tools and Algorithms for the Construction and Analysis of Systems, 16th International Conference, TACAS 2010, Held as Part of the Joint European Conferences on Theory and Practice of Software, ETAPS 2010, Paphos, Cyprus, March 20-28, 2010. Proceedings (Lecture Notes in Computer Science, Vol. 601",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Randal E. Bryant, Daniel Kroening, Joël Ouaknine, Sanjit A. Seshia, Ofer Strichman, and Bryan A. Brady. 2007. Deciding Bit-Vector Arithmetic with Abstraction. In Tools and Algorithms for the Construction and Analysis of Systems, 13th International Conference, TACAS 2007, Held as Part of the Joint European Conferences on Theory and Practice of Software, ETAPS 2007 Braga, Portugal, March 24 - April ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Alexandra Bugariu and Peter Müller. 2020. Automatically testing string solvers. InICSE ’20: 42nd International Conference on Software Engineering, Seoul, South Korea, 27 June-19 July, 2020, Gregg Rothermel and Doo-Hwan Bae (Eds.). ACM, 1459–1470. https://doi.org/10.1145/3377811.3380398",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Cristian Cadar, Daniel Dunbar, Dawson R Engler, et al .2008. KLEE: Unassisted and Automatic Generation of High-Coverage Tests for Complex Systems Programs.. InProceedings of the 8th USENIX Conference on Operating Systems Design and Implementation (San Diego, California) (OSDI’08). USENIX Association, Berkeley, CA, USA, 209–224. http://dl.acm.org/citation.cfm?id=1855741.1855756",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "W. K. Chan, S. C. Cheung, and Karl R. P. H. Leung. 2005. Towards a Metamorphic Testing Methodology for Service-Oriented Software Applications. In Fifth International Conference on Quality Software (QSIC 2005), 19-20 September 2005, Melbourne, Australia. IEEE Computer Society, 470–476. https://doi.org/10.1109/ QSIC.2005.67",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Tsong Y Chen, Shing C Cheung, and Shiu Ming Yiu. 2020. Metamorphic testing: a new approach for generating next test cases. arXiv preprint arXiv:2002.12543 (2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Tsong Yueh Chen, Joshua Wing Kei Ho, Huai Liu, and Xiaoyuan Xie. 2009. An innovative approach for testing bioinformatics programs using metamorphic testing. BMC Bioinform. 10 (2009). https://doi.org/10.1186/1471-2105-10-24",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Jürgen Christ, Jochen Hoenicke, and Alexander Nutz. 2012. SMTInterpol: An Interpolating SMT Solver. In Model Checking Software-19th International Workshop, SPIN 2012, Oxford, UK, July 23-24, 2012. Proceedings (Lecture Notes in Computer Science, Vol. 7385), Alastair F. Donaldson and David Parker (Eds.). Springer, 248–254. https://doi.org/10.1007/978-3-642-31759-0_19",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Leonardo De Moura and Nikolaj Bjørner. 2008. Z3: An ecient SMT solver. In Proceedings of the Theory and Practice of Software, 14th International Conference on Tools and Algorithms for the Construction and Analysis of Systems (Budapest, Hungary) (TACAS’08/ETAPS’08). Springer-Verlag, Berlin, Heidelberg, 337–340. http://dl.acm.org/citation.cfm?id=1792734.1792766",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "David Detlefs, Greg Nelson, and James B. Saxe. 2005. Simplify: A Theorem Prover for Program Checking. J. ACM 52, 3 (May 2005), 365–473. https://doi.org/10. 1145/1066100.1066102",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Bruno Dutertre. 2014. Yices2.2. In Proceedings of the 16th International Conference on Computer Aided Verication-Volume 8559. Springer-Verlag, Berlin, Heidelberg, 737–744. https://doi.org/10.1007/978-3-319-08867-9_49",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Anurag Dwarakanath, Manish Ahuja, Samarth Sikand, Raghotham M. Rao, R. P. Jagadeesh Chandra Bose, Neville Dubash, and Sanjay Podder. 2018. Identifying implementation bugs in machine learning based image classiers using metamorphic testing. In Proceedings of the 27th ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2018, Amsterdam, The Netherlands, July 16-21, 2018, Frank ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Vijay Ganesh and David L Dill. 2007. A decision procedure for bit-vectors and arrays. In Proceedings of the 19th International Conference on Computer Aided Verication (Berlin, Germany) (CAV’07). Springer-Verlag, Berlin, Heidelberg, 519– 531. http://dl.acm.org/citation.cfm?id=1770351.1770421",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Patrice Godefroid, Nils Klarlund, and Koushik Sen. 2005. DART: directed automated random testing. In Proceedings of the 2005 ACMSIGPLAN Conference on Programming Language Design and Implementation (Chicago, IL, USA) (PLDI ’05). ACM, New York, NY, USA, 213–223. https://doi.org/10.1145/1065010.1065036",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "HyungSeok Han, DongHyeon Oh, and Sang Kil Cha. 2019. CodeAlchemist: Semantics-Aware Code Generation to Find Vulnerabilities in JavaScript Engines. In 26th Annual Network and Distributed System Security Symposium, NDSS 2019, San Diego, California, USA, February 24-27, 2019. The Internet Society. https://www.ndss-symposium.org/ndss-paper/codealchemist-semanticsaware-code-generation-to-nd-vulnerabili",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Heqing Huang, Peisen Yao, Rongxin Wu, Qingkai Shi, and Charles Zhang. 2020. Pangolin: Incremental Hybrid Fuzzing with Polyhedral Path Abstraction. In ay 18-21, 2020. IEEE, 1613–1627. https://doi.org/10.1109/SP40000.2020.00063",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. In ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI ’14, Edinburgh, United Kingdom-June 09 - 11, 2014, Michael F. P. O’Boyle and Keshav Pingali (Eds.). ACM, 216–226. https://doi.org/10.1145/2594291.2594334",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Yi Li, Aws Albarghouthi, Zachary Kincaid, Arie Gurnkel, and Marsha Chechik. 2014. Symbolic Optimization with SMT Solvers. In Proceedings of the 41st ACMSIGPLAN-SIGACT Symposium on Principles of Programming Languages (San Diego, California, USA) (POPL ’14). ACM, New York, NY, USA, 607–618. https://doi. org/10.1145/2535838.2535857",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Pingchuan Ma, Shuai Wang, and Jin Liu. 2020. Metamorphic Testing and Certied Mitigation of Fairness Violations in NLP Models. In Proceedings of the TwentyNinth International Joint Conference on Articial Intelligence, IJCAI 2020, Christian Bessiere (Ed.). ijcai.org, 458–465. https://doi.org/10.24963/ijcai.2020/64",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Muhammad Numair Mansur, Maria Christakis, Valentin Wüstholz, and Fuyuan Zhang. 2020. Detecting critical bugs in SMT solvers using blackbox mutational 1152 Skeletal Approximation Enumeration for SMT Solver Testing ESEC/FSE ’21, August 23–28, 2021, Athens, Greece fuzzing. In ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Kenneth L. McMillan. 2006. Lazy Abstraction with Interpolants. In Computer Aided Verication, 18th International Conference, CAV 2006, Seattle, WA, USA, August 17-20, 2006, Proceedings (Lecture Notes in Computer Science, Vol. 4144), Thomas Ball and Robert B. Jones (Eds.). Springer, 123–136. https://doi.org/10. 1007/11817963_14",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Sergey Mechtaev, Jooyong Yi, and Abhik Roychoudhury. 2016. Angelix: Scalable multiline program patch synthesis via symbolic analysis. In Proceedings of the 38th International Conference on Software Engineering (Austin, Texas) (ICSE ’16). ACM, New York, NY, USA, 691–701. https://doi.org/10.1145/2884781.2884807",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Aina Niemetz, Mathias Preiner, and Armin Biere. 2014. Boolector 2.0. J. Satisf. Boolean Model. Comput. 9, 1 (2014), 53–58. https://doi.org/10.3233/sat190101",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Andres Nötzli, Andrew Reynolds, Haniel Barbosa, Aina Niemetz, Mathias Preiner, Clark W. Barrett, and Cesare Tinelli. 2019. Syntax-Guided Rewrite Rule Enumeration for SMT Solvers. In Theory and Applications of Satisability Testing-SAT 2019 - 22nd International Conference, SAT 2019, Lisbon, Portugal, July 9-12, 2019, Proceedings (Lecture Notes in Computer Science, Vol. 11628), Mikolás Janota and Inê",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Mike Papadakis, and Yves Le Traon. 2019. Semantic fuzzing with zest. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2019, Beijing, China, July 15-19, 2019, Dongmei Zhang and Anders Møller (Eds.). ACM, 329–340. https://doi.org/10.1145/3293882.3330576",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Stuart Pernsteiner, Calvin Loncaric, Emina Torlak, Zachary Tatlock, Xi Wang, Michael D. Ernst, and Jonathan Jacky. 2016. Investigating Safety of a Radiotherapy Machine Using System Models with Pluggable Checkers. In Computer Aided Verication-28th International Conference, CAV 2016, Toronto, ON, Canada, July 17-23, 2016, Proceedings, Part II (Lecture Notes in Computer Science, Vol. 9780), Swarat Ch",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Van-Thuan Pham, Marcel Böhme, Andrew Edward Santosa, Alexandru Razvan Caciulescu, and Abhik Roychoudhury. 2019. Smart greybox fuzzing. IEEE Transactions on Software Engineering (2019), 1–17. https://doi.org/10.1109/TSE.2019. 2941681",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (Nov. 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 48,
+      "text": "Joseph Scott, Federico Mora, and Vijay Ganesh. 2020. BanditFuzz: A Reinforcement-Learning Based Performance Fuzzer for SMT Solvers. In Software Verication-12th International Conference, VSTTE 2020, and 13th International Workshop, NSV 2020, Los Angeles, CA, USA, July 20-21, 2020, Revised Selected Papers (Lecture Notes in Computer Science, Vol. 12549), Maria Christakis, Nadia Polikarpova, Parasara ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitriy Vyukov. 2012. AddressSanitizer: A Fast Address Sanity Checker. In 2012 USENIX Annual Technical Conference, Boston, MA, USA, June 13-15, 2012, Gernot Heiser and Wilson C. Hsieh (Eds.). USENIX Association, 309–318. https://www.usenix. org/conference/atc12/technical-sessions/presentation/serebryany",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Armando Solar-Lezama and Rastislav Bodik. 2008. Program synthesis by sketching. Citeseer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. 2016. Finding compiler bugs via live code mutation. In Proceedings of the 2016 ACMSIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications, OOPSLA 2016, part of SPLASH 2016, Amsterdam, The Netherlands, October 30 - November 4, 2016, Eelco Visser and Yannis Smaragdakis (Eds.). ACM, 849–863. https: //doi.org/10.1145/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "G Team. 2014. Gcov-using the gnu compiler collection (gcc). Online, disponıvel em http://gcc. gnu. org/onlinedocs/gcc/Gcov. html-Ultimo acesso em 26, 02 (2014), 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Sandro Tolksdorf, Daniel Lehmann, and Michael Pradel. 2019. Interactive metamorphic testing of debuggers. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2019, Beijing, China, July 15-19, 2019, Dongmei Zhang and Anders Møller (Eds.). ACM, 273–283. https://doi.org/10.1145/3293882.3330567",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Grigori S Tseitin. 1983. On the complexity of derivation in propositional calculus. InAutomation of reasoning. Springer, 466–483. https://doi.org/10.1007/978-3642-81955-1_28",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Muhammad Usman, Wenxi Wang, and Sarfraz Khurshid. 2020. TestMC: Testing Model Counters using Dierential and Metamorphic Testing. In 35th IEEE/ACM International Conference on Automated Software Engineering, ASE 2020, Melbourne, Australia, September 21-25, 2020. IEEE, 709–721. https://doi.org/10.1145/3324884. 3416563",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. 2019. Superion: grammar-aware greybox fuzzing. In Proceedings of the 41st International Conference on Software Engineering, ICSE 2019, Montreal, QC, Canada, May 25-31, 2019, Joanne M. Atlee, Tevk Bultan, and Jon Whittle (Eds.). IEEE / ACM, 724–735. https://doi.org/10. 1109/ICSE.2019.00081",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Elaine J Weyuker. 1982. On testing non-testable programs. Comput. J. 25, 4 (1982), 465–470. https://doi.org/10.1093/comjnl/25.4.465",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Virginie WIELS, Robert Delmas, David Doose, Pierre-Loïc Garoche, J. Cazin, and Guy Durrieu. 2012. Formal Verication of Critical Aerospace Software. AerospaceLab 4 (May 2012), p. 1–8. https://hal.archives-ouvertes.fr/hal-01184099",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2020. On the unusual eectiveness of type-aware operator mutations for testing SMT solvers. Proc. ACM Program. Lang. 4, OOPSLA (2020), 193:1–193:25. https://doi.org/10.1145/ 3428261",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2020. Validating SMT solvers via semantic fusion. In Proceedings of the 41st ACMSIGPLAN International Conference on Programming Language Design and Implementation, PLDI 2020, London, UK, June 15-20, 2020, Alastair F. Donaldson and Emina Torlak (Eds.). ACM, 718–730. https://doi.org/10.1145/3385412.3385985",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Peisen Yao, Heqing Huang, Wensheng Tang, Qingkai Shi, Rongxin Wu, and Charles Zhang. 2021. Fuzzing SMT Solvers via Two-Dimensional Input Space Exploration. In ISSTA’21: 30th ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, USA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Cunxi Yu, Maciej J. Ciesielski, and Alan Mishchenko. 2018. Fast Algebraic Rewriting Based on And-Inverter Graphs. IEEE Trans. Comput. Aided Des. Integr. Circuits Syst. 37, 9 (2018), 1907–1911. https://doi.org/10.1109/TCAD.2017.2772854",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Andreas Zeller. 1999. Yesterday, My Program Worked. Today, It Does Not. Why? 1687 (1999), 253–267. https://doi.org/10.1007/3-540-48166-4_16 1153",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (Nov. 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 47]",
+      "technique": "tlp",
+      "sentence": "in many application domains such as bioinformatics [ 26], web services [ 24], compilers [ 36,51], debuggers [ 53], databases [ 47], machine learning-based systems [ 31,38], model counters [ 55], and SMT solvers [60].",
+      "context_before": "ure of a seed but can only generate satisable mutants. Compared to the previous works, we present a new technique for the oracleguided approach, which applies to all theories and can generate satisable and unsatisable mutant formulas. Metamorphic Testing .The key idea of metamorphic testing [ 25] is to detect violations of domain-specic metamorphic relations by comparing the outputs between a seed test and its corresponding mutant tests. Metamorphic testing has been successfully applied 2Note that, in theory, it could be possible to extend the idea of semantic fusion to support other theories.",
+      "context_after": "Our approach is an instance of metamorphic testing. A closely related work is semantic fusion [ 60], which generates equi-satisable mutants from the concatenation of two seed formulas. Our approach diers in two aspects. First, semantic fusion mutates variables using the fusion functions, which can only introduce new variables and function symbols in the mutants. In comparison, skeletal approximation enumeration can inject formula snippets, which can contain new variables, function symbols, predicate symbols, and Boolean connectives. Second, semantic fusion requires that the two seeds are both",
+      "section": "6 DISCUSSION",
+      "page": 11,
+      "char_offset": 58638,
+      "cited_reference": {
+        "number": 47,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (Nov. 2020), 30 pages. https://doi.org/10.1145/3428279",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:03:26Z",
+    "is_model_written": true,
+    "summary": "This work tests SMT solvers by enumerating skeletal approximations of a formula -- systematically weakened versions whose satisfiability stands in a known relation to the original -- giving an oracle without a reference solver.",
+    "narrative": "Databases appear in a single sentence listing the application domains where the underlying technique has been used, alongside bioinformatics, compilers, debuggers and machine learning systems. The mention is reachable only through the citation marker and establishes breadth of applicability rather than any relationship to SQLancer.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3468264_3468573.json b/_data/papers/paper_doi_10_1145_3468264_3468573.json
new file mode 100644
index 0000000..7522255
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3468264_3468573.json
@@ -0,0 +1,630 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:47Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3468264.3468573",
+    "title": "Metamorphic testing of Datalog engines",
+    "authors": [
+      "Muhammad Numair Mansur",
+      "M. Christakis",
+      "Valentin Wüstholz"
+    ],
+    "year": 2021,
+    "venue": "ESEC/SIGSOFT FSE",
+    "doi": "10.1145/3468264.3468573",
+    "arxiv_id": null,
+    "s2_paper_id": "13cf48695d23f7e019d670a1bf059b7f28396992",
+    "url": "https://doi.org/10.1145/3468264.3468573",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3468264.3468573",
+      "retrieved_at": "2026-09-09T01:33:47Z",
+      "chars": 68408,
+      "content_sha256": "sha256:aad41215496ab5157bd3a9a80b963d34df49686935d6352a8afd709a411d1b59"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1458
+      },
+      {
+        "number": "2",
+        "title": "OVERVIEW",
+        "start": 6733
+      },
+      {
+        "number": "3",
+        "title": "BACKGROUND",
+        "start": 12763
+      },
+      {
+        "number": "4",
+        "title": "METAMORPHIC TRANSFORMATIONS",
+        "start": 16923
+      },
+      {
+        "number": "4.1",
+        "title": "Add Transformations",
+        "start": 18616
+      },
+      {
+        "number": "4.2",
+        "title": "Mod Transformations",
+        "start": 23582
+      },
+      {
+        "number": "4.3",
+        "title": "Rem Transformations",
+        "start": 26663
+      },
+      {
+        "number": "4.4",
+        "title": "Transformation Sequences",
+        "start": 30599
+      },
+      {
+        "number": "5",
+        "title": "BEYOND CONJUNCTIVE QUERIES",
+        "start": 31642
+      },
+      {
+        "number": "5.1",
+        "title": "Neg Transformation",
+        "start": 34854
+      },
+      {
+        "number": "6",
+        "title": "IMPLEMENTATION",
+        "start": 36250
+      },
+      {
+        "number": "7",
+        "title": "EXPERIMENTAL EVALUATION",
+        "start": 37657
+      },
+      {
+        "number": "7.1",
+        "title": "Experimental Setup",
+        "start": 38011
+      },
+      {
+        "number": "7.2",
+        "title": "Experimental Results",
+        "start": 38875
+      },
+      {
+        "number": "7.3",
+        "title": "Threats to Validity",
+        "start": 50140
+      },
+      {
+        "number": "8",
+        "title": "RELATED WORK",
+        "start": 50997
+      },
+      {
+        "number": "9",
+        "title": "CONCLUSION",
+        "start": 55008
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "[n.d.]. Radamsa. https://gitlab.com/akihe/radamsa.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "[n.d.]. Technical “Whitepaper” for AFL. http://lcamtuf.coredump.cx/afl/ technical_details.txt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Molham Aref, Balder ten Cate, Todd J. Green, Benny Kimelfeld, Dan Olteanu, Emir Pasalic, Todd L. Veldhuizen, and Geoffrey Washburn. 2015. Design and Implementation of the LogicBlox System. In SIGMOD. ACM, 1371–1382.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Isaac Balbin, Graeme S. Port, Kotagiri Ramamohanarao, and Krishnamurthy Meenakshi. 1991. Efficient Bottom-Up Computation of Queries on Stratified Databases. JLP11 (1991), 295–344. Issue 3&4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "François Bancilhon, David Maier, Yehoshua Sagiv, and Jeffrey D. Ullman. 1986. Magic Sets and Other Strange Ways to Implement Logic Programs. In PODS. ACM, 1–15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Earl T. Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. 2015. The Oracle Problem in Software Testing: A Survey. TSE41 (2015), 507–525. Issue 5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Catriel Beeri and Raghu Ramakrishnan. 1991. On the Power of Magic. JLP10 (1991), 255–299. Issue 3&4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Pierre-Léo Bégay, Pierre Crégut, and Jean-François Monin. 2021. Developing and Certifying Datalog Optimizations in Coq/MathComp. In CPP. ACM, 163–177.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Carsten Binnig, Donald Kossmann, and Eric Lo. 2007. Reverse Query Processing. InICDE. IEEE Computer Society, 506–515.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M. Tamer Özsu. 2007. QAGen: Generating Query-Aware Test Databases. In SIGMOD. ACM, 341–352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Martin Bravenboer and Yannis Smaragdakis. 2009. Strictly Declarative Specification of Sophisticated Points-To Analyses. In OOPSLA. ACM, 243–262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Lexi Brent, Anton Jurisevic, Michael Kong, Eric Liu, François Gauthier, Vincent Gramoli, Ralph Holz, and Bernhard Scholz. 2018. Vandal: A Scalable Security Analysis Framework for Smart Contracts. CoRR abs/1809.03981 (2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Neville Brent, Lexiand Grech, Sifis Lagouvardos, Bernhard Scholz, and Yannis Smaragdakis. 2020. Ethainter: A Smart Contract Security Analyzer for Composite Vulnerabilities. In PLDI. ACM, 454–469.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. 2006. Generating Queries with Cardinality Constraints for DBMS Testing. TKDE 18 (2006), 1721–1725. Issue 12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Alexandra Bugariu and Peter Müller. 2020. Automatically Testing String Solvers. InICSE. ACM, 1459–1470.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Alexandra Bugariu, Valentin Wüstholz, Maria Christakis, and Peter Müller. 2018. Automatically Testing Implementations of Numerical Abstract Domains. In ASE. ACM, 768–778.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Cristian Cadar and Alastair F. Donaldson. 2016. Analysing the Program Analyser. InICSE. ACM, 765–768.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "W. K. Chan, S. C. Cheung, and Karl R. P. H. Leung. 2005. Towards a Metamorphic Testing Methodology for Service-Oriented Software Applications. In QSIC. IEEE Computer Society, 470–476.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Ashok K. Chandra and Philip M. Merlin. 1977. Optimal Implementation of Conjunctive Queries in Relational Data Bases. In STOC. ACM, 77–90.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Junjie Chen, Jibesh Patra, Michael Pradel, Yingfei Xiong, Hongyu Zhang, Dan Hao, and Lu Zhang. 2020. A Survey of Compiler Testing. Comput. Surv. 53 (2020), 4:1–4:36. Issue 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Tsong Yueh Chen, S. C. Cheung, and Siu-Ming Yiu. 1998. Metamorphic Testing: A New Approach for Generating Next Test Cases. Technical Report HKUST–CS98–01. HKUST.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Oege de Moor, Damien Sereni, Mathieu Verbaere, Elnar Hajiyev, Pavel Avgustinov, Torbjörn Ekman, Neil Ongkingco, and Julian Tibble. 2007. .QL: Object-Oriented Queries Made Easy. In GTTSE (LNCS, Vol. 5235). Springer, 78–133.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Leonardo de Moura and Nikolaj Bjørner. 2008. Z3: An Efficient SMT Solver. In TACAS (LNCS, Vol. 4963). Springer, 337–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Alastair F. Donaldson, Hugues Evrard, Andrei Lascu, and Paul Thomson. 2017. Automated Testing of Graphics Shader Compilers. PACMPL 1 (2017), 93:1–93:29. Issue OOPSLA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Markian M. Gooley and Benjamin W. Wah. 1988. Efficient Reordering of Prolog Programs. In ICDE. IEEE Computer Society, 110–117.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Sergey Grebenshchikov, Nuno P. Lopes, Corneliu Popeea, and Andrey Rybalchenko. 2012. Synthesizing Software Verifiers from Proof Rules. In PLDI. ACM, 405–416. 649 ESEC/FSE ’21, August 23–28, 2021, Athens, Greece Muhammad Numair Mansur, Maria Christakis, and Valentin Wüstholz",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Neville Grech, Michael Kong, Anton Jurisevic, Lexi Brent, Bernhard Scholz, and Yannis Smaragdakis. 2018. MadMax: Surviving Out-of-Gas Conditions in Ethereum Smart Contracts. PACMPL 2 (2018), 116:1–116:27. Issue OOPSLA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Sergio Greco and Cristian Molinaro. 2015. Datalog and Logic Databases. Morgan & Claypool.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Daniel Halperin, Victor Teixeira de Almeida, Lee Lee Choo, Shumo Chu, Paraschos Koutris, Dominik Moritz, Jennifer Ortiz, Vaspol Ruamviboonsuk, Jingjing Wang, Andrew Whitaker, Shengliang Xu, Magdalena Balazinska, Bill Howe, and Dan Suciu. 2014. Demonstration of the Myria Big Data Management Service. In SIGMOD. ACM, 881–884.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Krystof Hoder, Nikolaj Bjørner, and Leonardo de Moura. 2011. 𝜇Z—An Efficient Engine for Fixed Points with Constraints. In CAV (LNCS, Vol. 6806). Springer, 457–462.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Seo Jiwon, Guo Stephen, and Lam Monica S. 2013. SociaLite: Datalog Extensions for Efficient Social Network Analysis. In ICDE. IEEE Computer Society, 278–289.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Herbert Jordan, Bernhard Scholz, and Pavle Subotic. 2016. Soufflé: On Synthesis of Program Analyzers. In CAV (LNCS, Vol. 9780). Springer, 422–430.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Herbert Jordan, Pavle Subotic, David Zhao, and Bernhard Scholz. 2019. Brie: A Specialized Trie for Concurrent Datalog. In PPoPP. ACM, 31–40.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woon-Hak Kang. 2019. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. VLDB 13 (2019), 57–70. Issue 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Timotej Kapus and Cristian Cadar. 2017. Automatic Testing of Symbolic Execution Engines via Program Generation and Differential Testing. In ASE. IEEE Computer Society, 590–600.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O. Laleye, and Sarfraz Khurshid. 2008. Query-Aware Test Generation Using a Relational Constraint Solver. In ASE. IEEE Computer Society, 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. 2010. Automated SQL Query Generation for Systematic Testing of Database Engines. In ASE. ACM, 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Ross D. King. 2004. Applying Inductive Logic Programming to Predicting Gene Function. AI Mag. 25 (2004), 57–68. Issue 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Kyle Kingsbury. [n.d.]. Jepsen. https://jepsen.io.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Christian Klinger, Maria Christakis, and Valentin Wüstholz. 2019. Differentially Testing Soundness and Precision of Program Analyzers. In ISSTA. ACM, 239–250.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler Validation via Equivalence Modulo Inputs. In PLDI. ACM, 216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Vu Le, Chengnian Sun, and Zhendong Su. 2015. Finding Deep Compiler Bugs via Guided Stochastic Program Mutation. In OOPSLA. ACM, 386–399.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Christopher Lidbury, Andrei Lascu, Nathan Chong, and Alastair F. Donaldson. 2015. Many-Core Compiler Fuzzing. In PLDI. ACM, 65–76.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Vsevolod Livinskii, Dmitry Babokin, and John Regehr. 2020. Random Testing for C and C++ Compilers with YARPGen. PACMPL 4 (2020), 196:1–196:25. Issue OOPSLA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Eric Lo, Carsten Binnig, Donald Kossmann, M. Tamer Özsu, and Wing-Kai Hon. 2010. A Framework for Testing DBMS Features. VLDB 19 (2010), 203–230. Issue 2.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Boon Thau Loo, Tyson Condie, Minos N. Garofalakis, David E. Gay, Joseph M. Hellerstein, Petros Maniatis, Raghu Ramakrishnan, Timothy Roscoe, and Ion Stoica. 2009. Declarative Networking. CACM 52 (2009), 87–95. Issue 11.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Muhammad Numair Mansur, Maria Christakis, Valentin Wüstholz, and Fuyuan Zhang. 2020. Detecting Critical Bugs in SMT Solvers Using Blackbox Mutational Fuzzing. In ESEC/FSE. ACM, 701–712.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software. Digital Technical Journal 10 (1998), 100–107. Issue 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. 2008. Generating Targeted Queries for Database Testing. In SIGMOD. ACM, 499–510.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Raymond J. Mooney. 1996. Inductive Logic Programming for Natural Language Processing. In ILP (LNCS, Vol. 1314). Springer, 3–22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Mayur Naik, Alex Aiken, and John Whaley. 2006. Effective Static Race Detection for Java. In PLDI. ACM, 308–319.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Meikel Poess and John M. Stephens. 2004. Generating Thousand Benchmark Queries in Seconds. In VLDB. Morgan Kaufmann, 1045–1053.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "David Poole. 1995. Logic Programming for Robot Control. In IJCAI. Morgan Kaufmann, 150–157.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Raghu Ramakrishnan, Catriel Beeri, and Ravi Krishnamurthy. 1988. Optimizing Existential Datalog Queries. In PODS. ACM, 89–102.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In ESEC/FSE. ACM, 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 56,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. PACMPL 4 (2020), 211:1–211:30. Issue OOPSLA.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 57,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In OSDI. USENIX, 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 58,
+      "text": "Kenneth A. Ross. 1990. Modular Stratification and Magic Sets for Datalog Programs with Negation. In PODS. ACM, 161–171.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Leonid Ryzhyk and Mihai Budiu. 2019. Differential Datalog. In Datalog (CEUR, Vol. 2368). CEUR-WS.org, 56–67.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Yehoshua Sagiv. 1988. Optimizing Datalog Programs. In Foundations of Deductive Databases and Logic Programming. Morgan Kaufmann, 659–698.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "José Carlos Almeida Santos, Houssam Nassif, David Page, Stephen H. Muggleton, and Michael J. E. Sternberg. 2012. Automated Identification of Protein-Ligand Interaction Features Using Inductive Logic Programming: A Hexose Binding Case Study. BMC Bioinform. 13 (2012), 162.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Sergio Segura, Gordon Fraser, Ana B. Sánchez, and Antonio Ruiz Cortés. 2016. A Survey on Metamorphic Testing. TSE42 (2016), 805–824. Issue 9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Andreas Seltenreich. [n.d.]. SQLsmith. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Alexander Shkapsky, Kai Zeng, and Carlo Zaniolo. 2013. Graph Queries in a Next-Generation Datalog System. VLDB 6 (2013), 1258–1261. Issue 12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In VLDB. Morgan Kaufmann, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. 2016. Finding and Analyzing Compiler Warning Defects. In ICSE. ACM, 203–213.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. 2016. Finding Compiler Bugs via Live Code Mutation. In OOPSLA. ACM, 849–863.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Jubi Taneja, Zhengyang Liu, and John Regehr. 2020. Testing Static Analyses for Precision and Soundness. In CGO. ACM, 81–93.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Petar Tsankov, Andrei Marian Dan, Dana Drachsler-Cohen, Arthur Gervais, Florian Bünzli, and Martin T. Vechev. 2018. Securify: Practical Security Analysis of Smart Contracts. In CCS. ACM, 67–82.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Muhammad Usman, Wenxi Wang, and Sarfraz Khurshid. 2020. TestMC: Testing Model Counters Using Differential and Metamorphic Testing. In ASE. IEEE Computer Society, 709–721.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Manasi Vartak, Venkatesh Raghavan, and Elke A. Rundensteiner. 2010. QRelX: Generating Meaningful Queries that Provide Cardinality Assurance. In SIGMOD. ACM, 1215–1218.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Jian Wang, Jungsoon P. Yoo, and Thomas J. Cheatham. 1993. Efficient Reordering of C-PROLOG. In Conference on Computer Science. ACM, 151–155.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "John Whaley, Dzintars Avots, Michael Carbin, and Monica S. Lam. 2005. Using Datalog with Binary Decision Diagrams for Program Analysis. In APLAS (LNCS, Vol. 3780). Springer, 97–118.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "John Whaley and Monica S. Lam. 2004. Cloning-Based Context-Sensitive Pointer Alias Analysis Using Binary Decision Diagrams. In PLDI. ACM, 131–144.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2020. On the Unusual Effectiveness of Type-Aware Operator Mutations for Testing SMT Solvers. PACMPL 4 (2020), 193:1–193:25. Issue OOPSLA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2020. Validating SMT Solvers via Semantic Fusion. In PLDI. ACM, 718–730.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Xiaoyuan Xie, Joshua Wing Kei Ho, Christian Murphy, Gail E. Kaiser, Baowen Xu, and Tsong Yueh Chen. 2009. Application of Metamorphic Testing to Supervised Classifiers. In QSIC. IEEE Computer Society, 135–144.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and Understanding Bugs in C Compilers. In PLDI. ACM, 283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Chengyu Zhang, Ting Su, Yichen Yan, Fuyuan Zhang, Geguang Pu, and Zhendong Su. 2019. Finding and Understanding Bugs in Software Model Checkers. In ESEC/FSE. ACM, 763–773.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Qirun Zhang, Chengnian Sun, and Zhendong Su. 2017. Skeletal Program Enumeration for Rigorous Compiler Testing. In PLDI. ACM, 347–361. 650",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 55,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In ESEC/FSE. ACM, 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 56,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. PACMPL 4 (2020), 211:1–211:30. Issue OOPSLA.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 57,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In OSDI. USENIX, 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Recently, Rigger and Su proposed a series of testing techniques [ 55– 57], which they implemented in a tool called SQLancer.",
+      "context_before": "suring their correctness and robustness is of critical importance and has been a focus of many researchers and practitioners for decades. In 1998, Slutz [ 65] proposed a technique, based on differential testing, to detect bugs in database systems. Another approach—also based on differential testing—was used by Jinho et al. to detect performance bugs [ 34]. Jepsen [ 39], developed by Kingsbury, is a practical tool for detecting safety bugs in distributed database systems; these can occur due to asynchronous interactions between components, data loss due to networking issues, node failures, etc.",
+      "context_after": "Their tool detected hundreds of bugs in various relational database systems. Fuzzing is also applied to detect crashes and other robustness issues in database systems. For instance, SQLsmith [ 63] is a popular SQL-query generator that has detected hundreds of crashes in widely used database systems. Other query-generation approaches include ones relying on constraint solvers [ 14,36,37,49,52,71], symbolic execution [10, 45], and reverse query processing [9]. Testing compilers. Compiler testing is another important and active research area [ 20,43,66,78]. Le et al. proposed a metamorphictestin",
+      "section": "8 RELATED WORK",
+      "page": 11,
+      "char_offset": 53231,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:03:26Z",
+    "is_model_written": true,
+    "summary": "This paper applies metamorphic testing to Datalog engines, constructing rule transformations whose effect on the derived facts is predictable, so that a disagreement identifies a bug in the engine's evaluation. Datalog's declarative semantics make it amenable to the same style of reasoning that works for SQL.",
+    "narrative": "SQLancer is named directly as the relational precedent: the paper notes that Rigger and Su proposed a series of testing techniques implemented in a tool called SQLancer. That places this work in the same line -- metamorphic relations over a declarative query language -- while the relations themselves are Datalog-specific and are the paper's own.",
+    "roles": {
+      "M1": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3471485_3471491.json b/_data/papers/paper_doi_10_1145_3471485_3471491.json
new file mode 100644
index 0000000..c76d0b5
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3471485_3471491.json
@@ -0,0 +1,173 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:24:05Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3471485.3471491",
+    "title": "Technical Perspective DIAMetrics",
+    "authors": [
+      "P. Boncz"
+    ],
+    "year": 2021,
+    "venue": "SIGMOD record",
+    "doi": "10.1145/3471485.3471491",
+    "arxiv_id": null,
+    "s2_paper_id": "0a54335989389a1e25b4ca13050eeeabd04168e2",
+    "url": "https://doi.org/10.1145/3471485.3471491",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3471485.3471491",
+      "retrieved_at": "2026-09-08T17:24:05Z",
+      "chars": 5017,
+      "content_sha256": "sha256:358e2f2c81198f9db0d0da36172194510f03eeb9f88c3eb0c44c67265ec1cd63"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 1,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "1",
+        "title": "REFERENCES",
+        "start": 4308
+      },
+      {
+        "number": "S",
+        "title": "Viglas. Comprehensive and efficient workload",
+        "start": 4529
+      },
+      {
+        "number": "D",
+        "title": "Wu. SQUIRREL: testing database management",
+        "start": 4841
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "C. Boden, A. Alexandrov, A. Kunft, T. Rabl, and V. Markl. PEEL: A framework for benchmarking distributed systems and algorithms. In TPCTC 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "S. Deep, A. Gruenheid, P. Koutris, J. F. Naughton, and S. Viglas. Comprehensive and efficient workload compression. PVLDB, 14(3):418–430, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "K. Huppler. The art of building a good benchmark. In TPCTC 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "M. Rigger and Z. Su. Finding bugs in database systems via query partitioning. In OOPSLA. ACM, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. InCCS, pages 955–970. ACM, 2020.SIGMOD Record, March 2021 (Vol. 50, No. 1) 23",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 4,
+      "text": "M. Rigger and Z. Su. Finding bugs in database systems via query partitioning. In OOPSLA. ACM, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[4]",
+      "technique": "tlp",
+      "sentence": "It inspired me to think about novel directions, possibly taking automatic benchmark extraction from performance monitoring accountability also towards correctness testing: one could envision enriching workload summarization with new dimensions such as code coverage [5] and automatic generation of query correctness oracles [4].",
+      "context_before": "seems a fruitful area of future research, e.g. into formal privacy-preservation bounds or privacy-respecting preservation of certain correlations. At Google, the DIAMetrics framework has proven itself useful for database developers, for performance monitoring and regression testing, as engines evolve. Database users also benefit, using the system to find out which of the multiple Google engines (e.g. F1, Procella, Dremel) best suits their use case, but also to provide performance accountability: to identify and communicate performance problems. This paper is, in short, a highly recommended read.",
+      "context_after": "1. REFERENCES [1] C. Boden, A. Alexandrov, A. Kunft, T. Rabl, and V. Markl. PEEL: A framework for benchmarking distributed systems and algorithms. In TPCTC 2017. [2] S. Deep, A. Gruenheid, P. Koutris, J. F. Naughton, and S. Viglas. Comprehensive and efficient workload compression. PVLDB, 14(3):418–430, 2020. [3] K. Huppler. The art of building a good benchmark. In TPCTC 2009. [4] M. Rigger and Z. Su. Finding bugs in database systems via query partitioning. In OOPSLA. ACM, 2020. [5] R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu. SQUIRREL: testing database management systems with language",
+      "section": null,
+      "page": 1,
+      "char_offset": 3978,
+      "cited_reference": {
+        "number": 4,
+        "text": "M. Rigger and Z. Su. Finding bugs in database systems via query partitioning. In OOPSLA. ACM, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:03:26Z",
+    "is_model_written": true,
+    "summary": "A technical perspective introducing DIAMetrics, Google's framework for benchmarking and monitoring data processing systems at scale. The piece situates that work in the broader effort to keep large query systems correct and performant.",
+    "narrative": "The only mention is the title of the query partitioning paper appearing in the accompanying reference list rather than in prose. There is no sentence in the perspective discussing SQLancer, so the connection is bibliographic: this record carries no statement about the relationship at all.",
+    "roles": {
+      "M1": "incidental"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": [
+      "Whether the perspective says anything about SQLancer in prose. The single mention is a reference-list entry, so the extracted text offers nothing to judge from."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3485529.json b/_data/papers/paper_doi_10_1145_3485529.json
new file mode 100644
index 0000000..1247524
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3485529.json
@@ -0,0 +1,157 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:33:46Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3485529",
+    "title": "Generative type-aware mutation for testing SMT solvers",
+    "authors": [
+      "Jiwon Park",
+      "Dominik Winterer",
+      "Chengyu Zhang",
+      "Zhendong Su"
+    ],
+    "year": 2021,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3485529",
+    "arxiv_id": null,
+    "s2_paper_id": "3f650c0d0a046c1fb6c068d636b138e14ded77b9",
+    "url": "https://doi.org/10.1145/3485529",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3485529",
+      "retrieved_at": "2026-09-08T14:33:46Z",
+      "chars": 63565,
+      "content_sha256": "sha256:bacae8c101a10c61a084259c13ee06f86b0a05f6552b9767a51f4c03568138b4"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 19,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1486
+      },
+      {
+        "number": "3.1",
+        "title": "GenerativeType-AwareMutation",
+        "start": 14347
+      },
+      {
+        "number": "3.2",
+        "title": "Relationshipsto FuzzChick and Type-AwareOperator Mutation",
+        "start": 17218
+      },
+      {
+        "number": "3.3",
+        "title": "TypeFuzz",
+        "start": 21078
+      },
+      {
+        "number": "4.1",
+        "title": "EvaluationSetup",
+        "start": 28140
+      }
+    ]
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Anotherrecentgenerativefuzzingapproachispivotquerysynthesis(PQS)[ Riggerand Su2020].",
+      "context_before": "gines. They use code coverageto guidethegrammar-basedmutations. GenerativeFuzzing .Generativefuzzers[ Hanford1970 ]synthesizetestinputsfromscratchusing a language grammar or a (language) model. Csmith [ Yang et al .2011] generates random C programsthroughrepeatedapplicationofrulesfromtheCgrammar.Similartogenerativetypeawaremutation,Csmithreliesondifferentialtestingtocross-checkthegeneratedseeds.Csmith has found 300+ bugs in the compilers GCC and LLVM. A recent follow-up work to Csmith is YarpGen [ Livinskii et al .2020] which additionally prevents generating C programs with undefined behavior.",
+      "context_after": "ItsynthesizesspecificSQLqueriesonrandomdatabases.UnlikeCsmithandYarpGen, PQS’s is a metamorphic testing approach. Moreover, researchers have adapted generative language models[Godefroidetal .2017]togenerateandguideinputgeneration.Asakeydifferencetoall the above approaches, generative type-aware mutation does not generate inputs from scratch but insteadsupports generation throughthesubstitutionof expressions. 7 CONCLUSION Weintroducedgenerativetype-awaremutation,anovelandeffectiveapproachfortesting SMT solvers. Generative type-aware mutation is a powerful generalization of type-aware operator mu",
+      "section": "4.1 EvaluationSetup",
+      "page": 17,
+      "char_offset": 55626,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "UnlikeCsmithandYarpGen, PQS’s is a metamorphic testing approach.",
+      "context_before": "tsfromscratchusing a language grammar or a (language) model. Csmith [ Yang et al .2011] generates random C programsthroughrepeatedapplicationofrulesfromtheCgrammar.Similartogenerativetypeawaremutation,Csmithreliesondifferentialtestingtocross-checkthegeneratedseeds.Csmith has found 300+ bugs in the compilers GCC and LLVM. A recent follow-up work to Csmith is YarpGen [ Livinskii et al .2020] which additionally prevents generating C programs with undefined behavior.Anotherrecentgenerativefuzzingapproachispivotquerysynthesis(PQS)[ Riggerand Su2020].ItsynthesizesspecificSQLqueriesonrandomdatabases.",
+      "context_after": "Moreover, researchers have adapted generative language models[Godefroidetal .2017]togenerateandguideinputgeneration.Asakeydifferencetoall the above approaches, generative type-aware mutation does not generate inputs from scratch but insteadsupports generation throughthesubstitutionof expressions. 7 CONCLUSION Weintroducedgenerativetype-awaremutation,anovelandeffectiveapproachfortesting SMT solvers. Generative type-aware mutation is a powerful generalization of type-aware operator mutation and FuzzChick whose limitations it overcomes. Furthermore, generative type-aware mutationsupports generati",
+      "section": "4.1 EvaluationSetup",
+      "page": 17,
+      "char_offset": 55759,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:52Z",
+    "is_model_written": true,
+    "summary": "TypeFuzz tests SMT solvers with generative type-aware mutation, splicing well-typed expressions from one formula into another so the result stays well-typed and reaches deeper into the solver. The technique is for constraint solvers rather than databases, and the paper surveys generative fuzzing across domains.",
+    "narrative": "PQS appears in the evaluation setup's survey of generative fuzzing approaches, cited in the ACM author-year style so SQLancer is not named directly. The paper places it alongside Csmith and YarpGen as recent generative fuzzers, and distinguishes it as the metamorphic one of the group -- checking a relation between two runs rather than comparing against an independent reference.",
+    "roles": {
+      "M1": "definition",
+      "M2": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3510003_3510093.json b/_data/papers/paper_doi_10_1145_3510003_3510093.json
new file mode 100644
index 0000000..e3a8875
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3510003_3510093.json
@@ -0,0 +1,730 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:11:47Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3510003.3510093",
+    "title": "Automatic Detection of Performance Bugs in Database Systems using Equivalent Queries",
+    "authors": [
+      "Xinyu Liu",
+      "Qi Zhou",
+      "Joy Arulraj",
+      "A. Orso"
+    ],
+    "year": 2022,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3510003.3510093",
+    "arxiv_id": null,
+    "s2_paper_id": "1f74989f2884e71ff09ca23f0c10e753bab9fcc9",
+    "url": "https://doi.org/10.1145/3510003.3510093",
+    "also_indexed_as": [
+      "paper:s2:74820ccf6c039c7745f8e1c30857dee61b243464"
+    ]
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3510003.3510093",
+      "retrieved_at": "2026-09-10T15:11:47Z",
+      "chars": 71899,
+      "content_sha256": "sha256:dad53cbaf5ffbff7da750652deb1af11f00cc1690c124cdb1aae62a05767dc02"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "1992. Database Language SQL. http://www.contrib.andrew.cmu.edu/~shadow/sql /sql1992.txt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2015. AFL: American Fuzzy Lop. http://lcamtuf.coredump.cx/afl/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2016. OSS-Fuzz: Continuous Fuzzing for Open Source Software. https: //github.com/google/oss-fuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2016. SQLSmith. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2020. Apache Calcite. https://calcite.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2020. CockroachDB. https://www.cockroachlabs.com/docs/releases/v20.2.0.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2020. SQLAlchemy. https://www.sqlalchemy.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2021. CockroachDB Performance Bug Reports. https://github.com/cockroachdb /cockroach/issues?q=performance.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2021. PostgreSQL Performance Bug Reports. https://www.postgresql.org/searc h/?m=1&q=performance&l=8&d=-1&s=r.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2021. PostgreSQL Wiki. https://wiki.postgresql.org/wiki/Main Page.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2021. SCOTT schema. https://www.orafaq.com/wiki/SCOTT.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2022. Supplementary material. https://bit.ly/3I995jL",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. 2007. A genetic approach for random testing of database systems. In VLDB. 1243–1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. 2006. Generating Queries with Cardinality Constraints for DBMS Testing. In TKDE. 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Stefano Ceri and Georg Gottlob. 1985. Translating SQL into relational algebra: Optimization, semantics, and equivalence of SQL queries. In TSE. 324–345.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Bikash Chandra, Bhupesh Chawda, Biplab Kar, KV Maheshwara Reddy, Shetal Shah, and S Sudarshan. 2015. Data generation for testing and grading SQL queries. InVLDB Journal. 731–755.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "CL Philip Chen and Chun-Y ang Zhang. 2014. Data-intensive applications, challenges, techniques and technologies: A survey on Big Data. In Information sciences. 314–347.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Tsong Y Chen, Shing C Cheung, and Shiu Ming Yiu. 2020. Metamorphic testing: a new approach for generating next test cases. In arXiv preprint arXiv:2002.12543.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Shumo Chu, Chenglong Wang, Konstantin Weitz, and Alvin Cheung. 2017. Cosette: An Automated Prover for SQL.. In CIDR.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Hicham G Elmongui, Vivek Narasayya, and Ravishankar Ramamurthy. 2009. A framework for testing query transformation rules. In SIGMOD. 257–268.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Leo Giakoumakis and César A Galindo-Legaria. 2008. Testing SQL Server’s Query Optimizer: Challenges, Techniques and Experiences. In Data Engineering Bulletin. 36–43.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Goetz Graefe. 1993. Query Evaluation Techniques for Large Databases. In CSUR. 73–169.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Goetz Graefe. 1995. The cascades framework for query optimization. In Data Engineering Bulletin. 19–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Zhongxian Gu, Mohamed A. Soliman, and Florian M. Waas. 2015. Testing the accuracy of query optimizers. In DBTest. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Toshihide Ibaraki and Tiko Kameda. 1984. On the optimal nesting order for computing n-relational joins. In TODS. 482–502.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. Apollo: Automatic detection and diagnosis of performance regressions in database systems. InVLDB. 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "R Kearns, Stephen Shead, and Alan Fekete. 1997. A teaching system for SQL. In ACSE. 224–231.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "D. Lee, S. K. Cha, and A. H. Lee. 2012. A Performance Anomaly Detection and Analysis Framework for DBMS Development. In TKDE. 1345–1360.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2015. How good are query optimizers, really?. In VLDB. 204–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Zhan Li, Olga Papaemmanouil, and Mitch Cherniack. 2016. OptMark: A Toolkit for Benchmarking Query Optimizers. In CIKM. 2155–2160.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "William M McKeeman. 1998. Differential testing for software. In Digital Technical Journal. 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Rasha Osman and William J. Knottenbelt. 2012. Database System Performance Evaluation Models: A Survey. In Performance Evaluation. 471–493.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Jaroslav Pokorn `y. 2015. Database technologies in the world of big data. In CompSysTech. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Raghu Ramakrishnan and Johannes Gehrke. 2003. Database Management Systems.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Kim-Thomas Rehmann, Changyun Seo, Dongwon Hwang, Binh Than Truong, Alexander Boehm, and Dong Hun Lee. 2016. Performance Monitoring in SAPHANA ’s Continuous Integration Process. In PER. 43–52.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Raymond Reiter. 1986. A Sound and Sometimes Complete Query Evaluation Algorithm for Relational Databases with Null V alues. In J. ACM. 349–370.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In FSE.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. In OOPSLA.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In OSDI.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 40,
+      "text": "Shetal Shah, S Sudarshan, Suhas Kajbaje, Sandeep Patidar, Bhanu Pratap Gupta, and Devang Vira. 2011. Generating test data for killing SQL mutants: A constraintbased approach. In ICDE. 1175–1186.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Hitesh Kumar Sharma, Mr SC Nelson, et al .2017. Explain Plan and SQL Trace the Two Approaches for RDBMS Tuning. In Database Systems Journal. 31–39.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Donald R Slutz. 1998. Massive Stochastic Testing of SQL. In VLDB. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Florian Waas and César Galindo-Legaria. 2000. Counting, Enumerating, andSampling of Execution Plans in a Cost-Based Query Optimizer. In SIGMOD. 499–509.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Gerhard Weikum, Axel Moenkeberg, Christof Hasse, and Peter Zabback. 2002. Self-Tuning Database Technology and Information Services: From Wishful Thinking to Viable Engineering. In VLDB. 20–31.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Khaled Y agoub, Peter Belknap, Benoit Dageville, Karl Dias, Shantanu Joshi, and Hailing Y u. 2008. Oracle’s SQL Performance Analyzer.. In Data Engineering Bulletin. 51–58.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Jiaqi Y an, Qiuye Jin, Shrainik Jain, Stratis D Viglas, and Allison Lee. 2018. Snowtrail: Testing with Production Queries on a Cloud Database. In DBTEST.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Rui Zhong, Y ongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language V alidity and Coverage Feedback. In CCS. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Qi Zhou, Joy Arulraj, Shamkant Navathe, William Harris, and Dong Xu. 2019. Automated verification of query equivalence using satisfiability modulo theories. InVLDB. 1276–1288. 236",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In FSE.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. In OOPSLA.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 39,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In OSDI.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For instance, they leverage tools such as SQLSMITH [4] and SQLancer [37–39] to discover crash-inducing or logic bugs in DBMSs.",
+      "context_before": "’22), May 21–29, 2022, Pittsburgh, PA, USA. ACM, New Y ork, NY, USA, 12 pages. https://doi.org/10.1145/3510003.3510093 Request permissions from permissions@acm.org. ICSE ’22, May 21–29, 2022, Pittsburgh, PA, USA © 2022 Copyright held by the owner/author(s). Publication rights licensed to ACM. ACMISBN 978-1-4503-9221-1/22/05.. . $15.00 https://doi.org/10.1145/3510003.35100931 Introduction Database management systems (DBMSs) play a critical role in modern data-intensive applications [ 17,33]. For this reason, developers extensively test these systems to improve their reliability and ac-curacy.",
+      "context_after": "However, the same level of scrutiny has not been applied toperformance bugs—bugs that affect the time taken by the DBMS to process certain queries. Detecting performance bugs is just as crucial as detecting functional bugs, as delayed responses from the DBMS can dramatically affect the user experience [32, 44]. CHALLENGES .To retrieve the results for a given SQL query, the DBMS invokes a pipeline of complex components (e.g., query optimizer, execution engine) [ 22,34]. The overall performance of the DBMS may be reduced due to sub-optimal decisions taken by any of these components and the comp",
+      "section": null,
+      "page": 1,
+      "char_offset": 2698,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "We also compared AMOEBA against two other sources of equivalent queries that could be used for detecting performance bugs: a manually-written test suite in a widely-used query optimization framework, and the Ternary Logic Partitioning (TLP) approach [ 38].",
+      "context_before": "mantic preserving query mutation rules that are also tailored to performance bugs detection: (1)structural mutations, which transform an input query using a setof query rewrite rules derived from the query optimization literature [ 23], and (2) expression mutations, which modify expressions within an input query without changing their semantics. To evaluate our technique, we implemented it and applied it to two widely-used DBMSs: CockroachDB and PostgreSQL. Our results are promising, in that AMOEBA found 39 potential performance bugs, among which developers have confirmed 6 bugs and fixed5 bugs.",
+      "context_after": "Our results show that the equivalent queries generated by AMOEBA are more likely to detect performance bugs. CONTRIBUTIONS .This paper makes the following contributions: •A performance bug detection technique with three new aspects: –The use of query equivalence to generate performance oracles. –Two types of query mutations that preserve the semantics of queries: structural mutations and expression mutations. –A feedback mechanism that improves the effectiveness and efficiency of the approach. •An implementation of the technique that is publicly available [ 12]. •An evaluation of the technique",
+      "section": null,
+      "page": 2,
+      "char_offset": 6615,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 38,40]",
+      "technique": "tlp",
+      "sentence": "Query equivalence is a well-studied topic used in many applications, such as testing DBMSs for correctness [ 38,40], educating developers [ 27], and automatically grading student assignments [ 16].",
+      "context_before": "Systems using Equivalent QueriesICSE ’22, May 21–29, 2022, Pittsburgh, P A, USA Figure 3: Architecture of AMOEBA –GENERA TOR constructs a set of base SQL queries based on feedback from other components. MUTATOR performs semantic-preserving structural and expression mutations on the base queries. VALIDA TOR executes a pair of semantically equivalent queries using the target DBMS and reports query pairs that exhibit a significant difference in runtime performance. SEMANTIC EQUIV ALENCE .Two queries are semantically equivalent if they always produce the same result on any input database instance.",
+      "context_after": "Unlike prior efforts, we seek to leverage semantic equivalence of queries to find performance bugs in DBMSs. QUERY REWRITING .AMOEBA constructs equivalent queries by rewriting them using a set of rules that preserve equivalence [ 23]. For example, a rule could mutate projection columns (i.e., the columns the query should return) using information from the filter predicate (i.e., the predicate that specifies which rows are to be returned). We illustrate how this rule transforms its projection columns while preserving semantic equivalence using query Q1 (Figure 1a) and its logical query plan (Figu",
+      "section": null,
+      "page": 3,
+      "char_offset": 12117,
+      "cited_reference": {
+        "number": 38,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. In OOPSLA.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "BenchmarkPPBs Found Cockroach PostgreSQL Benchmark 1 (TLP) 1 1 Benchmark 2 (Calcite) 4 4 Benchmark 3 (Calcite+AMOEBA )4 6 AMOEBA 25 14 Both structural and expression mutation rules generate queries thatcan reveal PPBs.",
+      "context_before": "aracteristics that make them challenging for a DBMS to optimize and execute. For both DBMSs, we found that these “effective” rules mostly re-arrange or eliminate expensive operators ( UNION, GROUP BY, and JOIN) in a given query. Conversely, other rules that manipulate the projection and sorting operators ( SELECT and ORDER BY) are less likely to generate query pairs that trigger PPBs. Improving how the DBMS handles these operators would therefore enhance their robustness.Table 4: Comparative Analysis of AMOEBA –Number of PPBs discovered by the set of query pairs in each baseline and by AMOEBA.",
+      "context_after": "Rules differ in their impact on query performance,and some rules that do not directly affect query performance enablethe application of other rules. Rules that transform expensive operators seem to be effective in generating queries that trigger PPBs, which highlights opportunities for improving future versions of the tested DBMSs. 5.6 RQ4 — Comparative Analysis Query equivalence has been leveraged before in related work [ 16, 27,40], albeit not to uncover performance bugs. To answer Q4, we compare AMOEBA to three baseline based on two of these existing approaches: Calcite [19, 48] and TLP [3",
+      "section": null,
+      "page": 9,
+      "char_offset": 50431,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "To answer Q4, we compare AMOEBA to three baseline based on two of these existing approaches: Calcite [19, 48] and TLP [38].",
+      "context_before": "k 3 (Calcite+AMOEBA )4 6 AMOEBA 25 14 Both structural and expression mutation rules generate queries thatcan reveal PPBs. Rules differ in their impact on query performance,and some rules that do not directly affect query performance enablethe application of other rules. Rules that transform expensive operators seem to be effective in generating queries that trigger PPBs, which highlights opportunities for improving future versions of the tested DBMSs. 5.6 RQ4 — Comparative Analysis Query equivalence has been leveraged before in related work [ 16, 27,40], albeit not to uncover performance bugs.",
+      "context_after": "BASELINE 1: QUERY PAIRS FROMTLP. TLP [ 38] construct equivalent queries to discover logic bugs in DBMSs. It is based on the observation that any predicate in SQL evaluates to TRUE, FALSE, or NULL [ 38]. Accordingly, TLP constructs a mutant query that is equivalent to the base query by (1) dividing the basequery into three partition queries, wherein each predicate is constructed based on the value of the overall base query’s predicate and (2) concatenating these partition queries using the UNION operator. Our first baseline consists of 2000 pairs of equivlent queries generated using TLP. BASELI",
+      "section": null,
+      "page": 9,
+      "char_offset": 51129,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [ 38] construct equivalent queries to discover logic bugs in DBMSs.",
+      "context_before": "ry performance,and some rules that do not directly affect query performance enablethe application of other rules. Rules that transform expensive operators seem to be effective in generating queries that trigger PPBs, which highlights opportunities for improving future versions of the tested DBMSs. 5.6 RQ4 — Comparative Analysis Query equivalence has been leveraged before in related work [ 16, 27,40], albeit not to uncover performance bugs. To answer Q4, we compare AMOEBA to three baseline based on two of these existing approaches: Calcite [19, 48] and TLP [38]. BASELINE 1: QUERY PAIRS FROMTLP.",
+      "context_after": "It is based on the observation that any predicate in SQL evaluates to TRUE, FALSE, or NULL [ 38]. Accordingly, TLP constructs a mutant query that is equivalent to the base query by (1) dividing the basequery into three partition queries, wherein each predicate is constructed based on the value of the overall base query’s predicate and (2) concatenating these partition queries using the UNION operator. Our first baseline consists of 2000 pairs of equivlent queries generated using TLP. BASELINE 2: QUERY PAIRS FROMCALCITE .The Calcite test suite [ 19,48] consists of tests manually crafted to ensu",
+      "section": null,
+      "page": 9,
+      "char_offset": 51286,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[ 38]",
+      "technique": "tlp",
+      "sentence": "It is based on the observation that any predicate in SQL evaluates to TRUE, FALSE, or NULL [ 38].",
+      "context_before": "nce enablethe application of other rules. Rules that transform expensive operators seem to be effective in generating queries that trigger PPBs, which highlights opportunities for improving future versions of the tested DBMSs. 5.6 RQ4 — Comparative Analysis Query equivalence has been leveraged before in related work [ 16, 27,40], albeit not to uncover performance bugs. To answer Q4, we compare AMOEBA to three baseline based on two of these existing approaches: Calcite [19, 48] and TLP [38]. BASELINE 1: QUERY PAIRS FROMTLP. TLP [ 38] construct equivalent queries to discover logic bugs in DBMSs.",
+      "context_after": "Accordingly, TLP constructs a mutant query that is equivalent to the base query by (1) dividing the basequery into three partition queries, wherein each predicate is constructed based on the value of the overall base query’s predicate and (2) concatenating these partition queries using the UNION operator. Our first baseline consists of 2000 pairs of equivlent queries generated using TLP. BASELINE 2: QUERY PAIRS FROMCALCITE .The Calcite test suite [ 19,48] consists of tests manually crafted to ensure the correctness of Calcite’s query transformation rules; each test transforms an input query us",
+      "section": null,
+      "page": 9,
+      "char_offset": 51358,
+      "cited_reference": {
+        "number": 38,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. In OOPSLA.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Accordingly, TLP constructs a mutant query that is equivalent to the base query by (1) dividing the basequery into three partition queries, wherein each predicate is constructed based on the value of the overall base query’s predicate and (2) concatenating these partition queries using the UNION operator.",
+      "context_before": "ctive in generating queries that trigger PPBs, which highlights opportunities for improving future versions of the tested DBMSs. 5.6 RQ4 — Comparative Analysis Query equivalence has been leveraged before in related work [ 16, 27,40], albeit not to uncover performance bugs. To answer Q4, we compare AMOEBA to three baseline based on two of these existing approaches: Calcite [19, 48] and TLP [38]. BASELINE 1: QUERY PAIRS FROMTLP. TLP [ 38] construct equivalent queries to discover logic bugs in DBMSs. It is based on the observation that any predicate in SQL evaluates to TRUE, FALSE, or NULL [ 38].",
+      "context_after": "Our first baseline consists of 2000 pairs of equivlent queries generated using TLP. BASELINE 2: QUERY PAIRS FROMCALCITE .The Calcite test suite [ 19,48] consists of tests manually crafted to ensure the correctness of Calcite’s query transformation rules; each test transforms an input query using a set of transformation rules and examines whether the resulting query is correct. Our second baseline consists of 373 pairs of equivalent queries from the Calcite test suite, in which we use the input query as the base query and the transformed query as the mutant query. This is an appropriate and cha",
+      "section": null,
+      "page": 9,
+      "char_offset": 51456,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Our first baseline consists of 2000 pairs of equivlent queries generated using TLP.",
+      "context_before": "o three baseline based on two of these existing approaches: Calcite [19, 48] and TLP [38]. BASELINE 1: QUERY PAIRS FROMTLP. TLP [ 38] construct equivalent queries to discover logic bugs in DBMSs. It is based on the observation that any predicate in SQL evaluates to TRUE, FALSE, or NULL [ 38]. Accordingly, TLP constructs a mutant query that is equivalent to the base query by (1) dividing the basequery into three partition queries, wherein each predicate is constructed based on the value of the overall base query’s predicate and (2) concatenating these partition queries using the UNION operator.",
+      "context_after": "BASELINE 2: QUERY PAIRS FROMCALCITE .The Calcite test suite [ 19,48] consists of tests manually crafted to ensure the correctness of Calcite’s query transformation rules; each test transforms an input query using a set of transformation rules and examines whether the resulting query is correct. Our second baseline consists of 373 pairs of equivalent queries from the Calcite test suite, in which we use the input query as the base query and the transformed query as the mutant query. This is an appropriate and challenging baseline because (1) the input queries are manually created to cover a wid",
+      "section": null,
+      "page": 9,
+      "char_offset": 51763,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP uses a different set of mutation rules than AMOEBA for generating equivalent queries, which were not designed to detect performance bugs.",
+      "context_before": "’sMUTA TOR on the base queries from the Calcite test suite. COMPARISON .Once selected these three baselines, we ran each set of baseline query pairs on our SCOTT-based database and comparedthe number of PPBs discovered by each baseline to those discovered by AMOEBA. Our results are shown in Table 4. As the figure shows, AMOEBA discovered significantly more PPBs than the baselines in both DBMSs. We next analyze the factors that contribute to the efficacy of AMOEBA. 233 ICSE ’22, May 21–29, 2022, Pittsburgh, P A, USA Xinyu Liu∗, Qi Zhou†, Joy Arulraj∗, Alessandro Orso∗ MUTATION RULES ANDALGORITHM .",
+      "context_after": "The queries mutated by TLP consistently take longer to execute than their corresponding base query, with an average slow-down of 17 ×. This happens because the mutated queries force the DBMS to perform additional operations (i.e., fetching the tuples for each partition query and combining the partial results). Given this inherent overhead, TLP is limited in the kinds and number of PPBs it can find. While Calcite and AMOEBA use the same set of mutation rules, they differ in the way they leverage these rules. Calcite ’s tests transform the base queries using a small set of mutation rules applied",
+      "section": null,
+      "page": 10,
+      "char_offset": 53315,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The queries mutated by TLP consistently take longer to execute than their corresponding base query, with an average slow-down of 17 ×.",
+      "context_before": "ry pairs on our SCOTT-based database and comparedthe number of PPBs discovered by each baseline to those discovered by AMOEBA. Our results are shown in Table 4. As the figure shows, AMOEBA discovered significantly more PPBs than the baselines in both DBMSs. We next analyze the factors that contribute to the efficacy of AMOEBA. 233 ICSE ’22, May 21–29, 2022, Pittsburgh, P A, USA Xinyu Liu∗, Qi Zhou†, Joy Arulraj∗, Alessandro Orso∗ MUTATION RULES ANDALGORITHM .TLP uses a different set of mutation rules than AMOEBA for generating equivalent queries, which were not designed to detect performance bugs.",
+      "context_after": "This happens because the mutated queries force the DBMS to perform additional operations (i.e., fetching the tuples for each partition query and combining the partial results). Given this inherent overhead, TLP is limited in the kinds and number of PPBs it can find. While Calcite and AMOEBA use the same set of mutation rules, they differ in the way they leverage these rules. Calcite ’s tests transform the base queries using a small set of mutation rules applied in a specific order. Conversely, AMOEBA ’s automated mutation strategy exploits all available mutation rules and their compositional ef",
+      "section": null,
+      "page": 10,
+      "char_offset": 53456,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Given this inherent overhead, TLP is limited in the kinds and number of PPBs it can find.",
+      "context_before": "y of AMOEBA. 233 ICSE ’22, May 21–29, 2022, Pittsburgh, P A, USA Xinyu Liu∗, Qi Zhou†, Joy Arulraj∗, Alessandro Orso∗ MUTATION RULES ANDALGORITHM .TLP uses a different set of mutation rules than AMOEBA for generating equivalent queries, which were not designed to detect performance bugs. The queries mutated by TLP consistently take longer to execute than their corresponding base query, with an average slow-down of 17 ×. This happens because the mutated queries force the DBMS to perform additional operations (i.e., fetching the tuples for each partition query and combining the partial results).",
+      "context_after": "While Calcite and AMOEBA use the same set of mutation rules, they differ in the way they leverage these rules. Calcite ’s tests transform the base queries using a small set of mutation rules applied in a specific order. Conversely, AMOEBA ’s automated mutation strategy exploits all available mutation rules and their compositional effects to generate equivalent query pairs, which increases the chances of generating queries that can discover PPBs (§4.3.2). The effect of this different approach can be observed in the second and third rows of Table 4, for PostgreSQL, where AMOEBA ’s mutation strat",
+      "section": null,
+      "page": 10,
+      "char_offset": 53768,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For both DBMSs considered, AMOEBA discovered considerably more PPBs than the baselines based on Calcite and TLP.",
+      "context_before": "ey differ in the way they leverage these rules. Calcite ’s tests transform the base queries using a small set of mutation rules applied in a specific order. Conversely, AMOEBA ’s automated mutation strategy exploits all available mutation rules and their compositional effects to generate equivalent query pairs, which increases the chances of generating queries that can discover PPBs (§4.3.2). The effect of this different approach can be observed in the second and third rows of Table 4, for PostgreSQL, where AMOEBA ’s mutation strategy discovers two more PPBs than the manual mutation in Calcite.",
+      "context_after": "Two reasons for this better performance are that AMOEBA (1) uses a rule mutation approach tailored to discovering PPBs and (2) performs a broad exploration of the equivalent query space by leveraging all available rules and their compositional effects. 5.7 RQ5 — Analysis of Base Queries As shown in Table 4, the base queries derived from the Calcite test suite allowed for discovering fewer PPBs than those generated by AMOEBA. To understand why, we compared a set of 2000 base queries generated by AMOEBA to the 373 base queries in Calcite. SINGLE-CLAUSE ANALYSIS .We first examined the SQL singlec",
+      "section": null,
+      "page": 10,
+      "char_offset": 54521,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [ 38].",
+      "context_before": "that it focuses on generating semantically equivalent query pairs that trigger different runtime performance. DIFFERENTIAL ANDMETAMORPHIC TESTING .To circumvent the oracle problem associated with automated testing, researchers have applied differential and metamorphic testing techniques for discovering logic bugs in DBMSs [ 18,31].RAGS discovers logic bugs by executing the same query on different DBMSs and comparing theresults [ 42]. Waas et al. propose a framework for validating the query optimizer by executing alternative execution plans for the input query and comparing their results [ 43].",
+      "context_after": "However, as discussed in §5.6, TLP is not suitable for discoveringperformance bugs. Unlike this previous work, AMOEBA is a metamorphic testing technique tailored for discovering performance bugs in DBMS. PERFORMANCE TESTING .Researchers have presented techniques for finding performance bugs by executing the DBMS on pre-defined workloads and comparing their behavior against performance baselines [ 26,35,45,46]. These techniques detect performance regressions caused by DBMS upgrades and configuration changes. AMOEBA differs from these approaches in that it does not require a pre-defined baseline fo",
+      "section": null,
+      "page": 11,
+      "char_offset": 62226,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "6, TLP is not suitable for discoveringperformance bugs.",
+      "context_before": "MORPHIC TESTING .To circumvent the oracle problem associated with automated testing, researchers have applied differential and metamorphic testing techniques for discovering logic bugs in DBMSs [ 18,31].RAGS discovers logic bugs by executing the same query on different DBMSs and comparing theresults [ 42]. Waas et al. propose a framework for validating the query optimizer by executing alternative execution plans for the input query and comparing their results [ 43]. TLP is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [ 38]. However, as discussed in §5.",
+      "context_after": "Unlike this previous work, AMOEBA is a metamorphic testing technique tailored for discovering performance bugs in DBMS. PERFORMANCE TESTING .Researchers have presented techniques for finding performance bugs by executing the DBMS on pre-defined workloads and comparing their behavior against performance baselines [ 26,35,45,46]. These techniques detect performance regressions caused by DBMS upgrades and configuration changes. AMOEBA differs from these approaches in that it does not require a pre-defined baseline for finding performance bugs. Instead, it leverages the tested DBMS’s runtime behaviors",
+      "section": null,
+      "page": 11,
+      "char_offset": 62356,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M9"
+      ],
+      "describes_as_state_of_the_art": [
+        "M14"
+      ]
+    },
+    "suppressed": [
+      {
+        "mention_id": "M5",
+        "pattern": "compares_with",
+        "suppressed_because": "the sentence is framed as related work, so it describes somebody else's approach"
+      }
+    ]
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:56:28Z",
+    "is_model_written": true,
+    "summary": "AMOEBA finds performance bugs by generating pairs of semantically equivalent queries and reporting when a DBMS executes one far more slowly than the other, since equivalent queries should cost about the same. It applies structural and expression mutation rules designed so the pairs stay comparable in cost. On CockroachDB and PostgreSQL it found substantially more potential performance bugs than either a hand-written optimiser test suite or an equivalence source drawn from TLP.",
+    "narrative": "TLP is one of AMOEBA's three baselines and the paper calls it the state-of-the-art tool for finding logic bugs in DBMSs using metamorphic testing. Because TLP also produces equivalent queries, it can be repurposed as a source of pairs, and the authors built a baseline of 2000 TLP-generated pairs to test that. The finding is a clean negative result about transfer: TLP's mutation rules were not designed for performance, and its mutant queries are on average 17 times slower than their base query, so that inherent overhead swamps the signal a performance oracle depends on. AMOEBA found 25 and 14 bugs against the TLP benchmark's 1 and 1.",
+    "roles": {
+      "M1": "background",
+      "M2": "baseline",
+      "M3": "background",
+      "M4": "result_comparison",
+      "M5": "baseline",
+      "M6": "definition",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "baseline",
+      "M10": "result_comparison",
+      "M11": "result_comparison",
+      "M12": "result_comparison",
+      "M13": "result_comparison",
+      "M14": "state_of_the_art",
+      "M15": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "TLP's construction is reimplemented to produce a baseline set of equivalent query pairs; the paper does not describe AMOEBA as built on the SQLancer codebase."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "AMOEBA's mutation rules are its own and are designed for cost-comparability, which M10 contrasts with TLP's. Using TLP as one source of equivalent pairs to measure against is comparison, not extension."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M2",
+          "M5",
+          "M9",
+          "M13"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "We also compared AMOEBA against two other sources of equivalent queries that could be used for detecting performance bugs: a manually-written test suite in a widely-used query optimization framework, and the Ternary Logic Partitioning (TLP) approach [ 38].",
+            "section": null,
+            "page": 2
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "To answer Q4, we compare AMOEBA to three baseline based on two of these existing approaches: Calcite [19, 48] and TLP [38].",
+            "section": null,
+            "page": 9
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "Our first baseline consists of 2000 pairs of equivlent queries generated using TLP.",
+            "section": null,
+            "page": 9
+          },
+          {
+            "mention_id": "M13",
+            "sentence": "For both DBMSs considered, AMOEBA discovered considerably more PPBs than the baselines based on Calcite and TLP.",
+            "section": null,
+            "page": 10
+          }
+        ],
+        "reasoning": "M2 and M5 name TLP as one of the equivalent-query sources AMOEBA is compared against, M9 records the 2000-pair baseline built from it, and M13 reports that AMOEBA found considerably more performance bugs than the TLP-based baseline.",
+        "techniques": [
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M14"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M14",
+            "sentence": "TLP is the state-of-the-art tool for discovering logic bugs in DBMS using metamorphic testing [ 38].",
+            "section": null,
+            "page": 11
+          }
+        ],
+        "reasoning": "M14 states that TLP is the state-of-the-art tool for discovering logic bugs in DBMSs using metamorphic testing."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3510457_3513034.json b/_data/papers/paper_doi_10_1145_3510457_3513034.json
new file mode 100644
index 0000000..d26a2a8
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3510457_3513034.json
@@ -0,0 +1,268 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:28Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3510457.3513034",
+    "title": "An Empirical Study on Quality Issues of eBay's Big Data SQL Analytics Platform",
+    "authors": [
+      "Feng Zhu",
+      "Lijie Xu",
+      "Gang Ma",
+      "Shuping Ji",
+      "Jie Wang",
+      "Gang Wang",
+      "Hongyi Zhang",
+      "K. Wan",
+      "Mingming Wang",
+      "Xingchao Zhang",
+      "Yuming Wang",
+      "Jingping Li"
+    ],
+    "year": 2022,
+    "venue": "2022 IEEE/ACM 44th International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP)",
+    "doi": "10.1145/3510457.3513034",
+    "arxiv_id": null,
+    "s2_paper_id": "ba2770c990b740b6dfa4a122ac4ad719868796dd",
+    "url": "https://doi.org/10.1145/3510457.3513034",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3510457.3513034",
+      "retrieved_at": "2026-09-09T01:30:28Z",
+      "chars": 61103,
+      "content_sha256": "sha256:0b9a10a80f6c2b8b3ac623e2319890ae935184dcc4f48fea71bc7921f4773fa3"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 10,
+    "has_outline": false,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Michael Armbrust, Reynold S Xin, and et al. 2015. Spark sql: Relational data processing inspark. In Proceedingsof the2015 International Conferenceon Management of Data (SIGMOD). 1383–1394.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "BerttyContreras-Rojas,Jorge-ArnulfoQuiané-Ruiz,andetal.2019. Tagsniff:Simplifiedbigdatadebuggingfordataflowjobs.In ProceedingsoftheACMSymposium on Cloud Computing (SOCC). 453–464.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "PrashanthDintyala,ArpitNarechania,andJoyArulraj.2020. SQLCheck:Automated Detection and Diagnosis of SQL Anti-Patterns. In Proceedings of the 2020 International Conference on Management of Data (SIGMOD). 2331–2345.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "FlorinDinuandTSEugeneNg.2012. Understandingtheeffectsandimplicationsofcomputenoderelatedfailuresinhadoop.In Proceedingsofthe21stinternational symposium on High-Performance Parallel and Distributed Computing. 187–198.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Avrilia Floratou,UmarFarooq Minhas,and FatmaÖzcan.2014. Sql-on-hadoop: Fullcirclebacktoshared-nothingdatabasearchitectures. ProceedingsoftheVLDB Endowment 7, 12 (2014), 1295–1306.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Muhammad Ali Gulzar, Matteo Interlandi, and et al. 2016. Bigdebug: Debugging primitives for interactive big data processing in spark. In 38th International Conference on Software Engineering (ICSE). IEEE, 784–795.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "MuhammadAliGulzar,ShaghayeghMardani,andet. al.2019. White-boxtesting of big data analytics with complex user-defined functions. In Proceedings of the 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 290–301.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "KazuakiIshizaki.2019. Analyzingandoptimizingjavacodegenerationforapache spark query plan. In Proceedings of the 2019 ACM/SPEC International Conference on Performance Engineering (ICPE). 91–102.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Hui Li, Dong Wang, and et al. 2020. Detecting cache-related bugs in Spark applications.In Proceedingsofthe29thACMSIGSOFTInternationalSymposium on Software Testing and Analysis (ISSTA). 363–375.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "SihanLi,HuchengZhou,HaoxiangLin,andetal.2013. Acharacteristicstudyon failuresofproductiondistributeddata-parallelprograms.In 35thInternational Conference on Software Engineering (ICSE). IEEE, 963–972.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Sourav Mazumdar and Subhankar Dhar. 2015. Hadoop as Big Data OperatingSystem–The Emerging Approach for Managing Challenges of Enterprise Big DataPlatform.In FirstInternationalConferenceonBigDataComputingService and Applications (BigDataService). IEEE, 499–505.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "ParimarjanNegi, MatteoInterlandi,and etal.2021. SteeringQuery Optimizers: A Practical Take on Big Data Workloads. In Proceedings of the 2021 International Conference on Management of Data (SIGMOD). 2557–2569.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via querypartitioning. ProceedingsoftheACMonProgrammingLanguages 4,OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivotedquery synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 16,
+      "text": "RathijitSen,AbhishekRoy,AlekhJindal,RuiFang,JeffZheng,XiaoleiLiu,and RuipingLi.2021. AutoExecutor:PredictiveParallelismforSparkSQLQueries. Proc. VLDB Endow. 14 (2021), 2855–2858.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Tian Xiao, Jiaxing Zhang, and et al. 2014. Nondeterminism in mapreduce consid-eredharmful?anempiricalstudyonnon-commutativeaggregatorsinmapreduceprograms.In 36thInternationalConferenceonSoftwareEngineering(ICSE).44–53.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Lijie Xu, Wensheng Dou, and et al. 2015. Experience report: A characteristic study on out of memory errors in distributed data-parallel applications. In 26th International Symposium on Software Reliability Engineering (ISSRE). 518–529.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Lijie Xu, Wensheng Dou, Feng Zhu, Chushu Gao, Jie Liu, and Jun Wei. 2018. CharacterizinganddiagnosingoutofmemoryerrorsinMapReduceapplications. Journal of Systems and Software 137 (2018), 399–414.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "ChengxuYang,YuanchunLi,MengweiXu,ZhenpengChen,YunxinLiu,Gang Huang, and Xuanzhe Liu. 2021. TaintStream: fine-grained taint tracking for big dataplatformsthroughdynamiccodetranslation.In Proceedingsofthe29thACM JointMeetingonEuropeanSoftwareEngineeringConferenceandSymposiumon the Foundations of Software Engineering (ESEC/FSE). 806–817.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "DingYuan,YuLuo,andetal.2014. SimpleTestingCanPreventMostCriticalFailures: An Analysis of Production Failures in Distributed Data-Intensive Systems. In11th USENIX Symposium on Operating Systems Design and Implementation (OSDI). 249–265.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Matei Zaharia, Mosharaf Chowdhury, Tathagata Das, Ankur Dave, Justin Ma, Murphy McCauly, Michael J. Franklin, Scott Shenker, and Ion Stoica. [n.d.]. Resilient Distributed Datasets: A Fault-Tolerant Abstraction for In-Memory Cluster Computing.In Proceedingsofthe9thUSENIXSymposiumonNetworkedSystems Design and Implementation, NSDI 2012, April 25-27, 2012. 15–28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "HuchengZhou,Jian-GuangLou,andetal.2015. Anempiricalstudyonquality issuesofproductionbigdataplatform.In 37thInternationalConferenceonSoftware Engineering (ICSE), Vol. 2. IEEE, 17–26. 42",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 13,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 14,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via querypartitioning. ProceedingsoftheACMonProgrammingLanguages 4,OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp"
+      ]
+    },
+    {
+      "number": 15,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivotedquery synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "TodetectDBMSbugs,RiggerandSu devised a series of novel approaches, including PQS [ 15], NoRec [13] and TLP [ 14].",
+      "context_before": "ndata-intensivescalablecomputingsystem(DISC).TheirevaluationshowsDISCapplicationsare oftensignificantlyskewedandinadequateintermsoftestcoverage. Thetoolcanminimizetestdatatoachieveinteractiveandfastlocal testing big data analytics. Bertty et al. [ 2] proposed TagSniff model based on tag and sniff primitives, to simplify data debugging fordataflows. They also developed a tool called Snoopy to support bothonlineandpost-hocdebuggingmodes.Huietal.[ 9]summarizedsixcache-relatedbugpatternsandproposes CacheCheck to automaticallydetectcache-relatedbugsbyanalyzingtheexecution tracesinSparkapplications.",
+      "context_after": "We can leverage and extend these testing approachestohelpusdiscoverdataqualityissues. TaintStream [20] automaticallyinjectstainttrackinglogicintothedataprocessing scripts.Itachievesaccuratecell-leveltainttrackingwithaprecision of 93.0% and less than 15% overhead. 8 CONCLUSION This paper presents a comprehensive study onthe service quality issuesofeBay’sbigdata SQLanalyticsplatform.Wedemonstrate the detailed issue symptoms and identify the main root causes,including user side faults, system internal defects and platformcomponent mismatches. We further discuss lessons learned and insights. We be",
+      "section": null,
+      "page": 10,
+      "char_offset": 54735,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-10T15:03:09Z",
+    "is_model_written": true,
+    "summary": "An industrial study of quality issues in eBay's big-data SQL analytics platform, characterising the defects that arise when SQL runs at that scale and what they cost the teams relying on the results.",
+    "narrative": "One citation, naming all three of SQLancer's original oracles: the paper notes that to detect DBMS bugs, Rigger and Su devised a series of novel approaches including PQS, NoREC and TLP. It is background for a study of an analytics platform rather than a testing paper, and nothing of SQLancer's is used or measured against.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; the study examines an existing platform rather than building a tester."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation names the three oracles as prior work."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison is reported; this is a study of production issues, not an evaluation."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The approaches are called novel, which describes their contribution rather than claiming they are the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3531348_3532176.json b/_data/papers/paper_doi_10_1145_3531348_3532176.json
new file mode 100644
index 0000000..cfa7d1a
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3531348_3532176.json
@@ -0,0 +1,304 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:12:59Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3531348.3532176",
+    "title": "DeepBench: Benchmarking JSON Document Stores",
+    "authors": [
+      "S. Belloni",
+      "D. Ritter",
+      "Marco Schröder",
+      "Nils Rörup"
+    ],
+    "year": 2022,
+    "venue": "DBTest@SIGMOD",
+    "doi": "10.1145/3531348.3532176",
+    "arxiv_id": null,
+    "s2_paper_id": "a48f9e74f5f0f60e033c8c290156243c16ece5c0",
+    "url": "https://doi.org/10.1145/3531348.3532176",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3531348.3532176",
+      "retrieved_at": "2026-09-07T17:12:59Z",
+      "chars": 39183,
+      "content_sha256": "sha256:34a7f8b9929d75382c9217be47231f895a966e23903c5b0bc8ea4c9ce1422c08"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 9,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1481
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND AND RELATED WORK",
+        "start": 5046
+      },
+      {
+        "number": "2.1",
+        "title": "JSON format",
+        "start": 5190
+      },
+      {
+        "number": "2.2",
+        "title": "Related Work",
+        "start": 7682
+      },
+      {
+        "number": "3",
+        "title": "DEEPBENCH",
+        "start": 10355
+      },
+      {
+        "number": "3.1",
+        "title": "DeepBench Client",
+        "start": 10559
+      },
+      {
+        "number": "3.2",
+        "title": "DeepBench Initializer",
+        "start": 12407
+      },
+      {
+        "number": "3.2.2",
+        "title": "Query Generation. The query generator has to produce SQL",
+        "start": 16629
+      },
+      {
+        "number": "3.3",
+        "title": "Summary – DeepBench",
+        "start": 19990
+      },
+      {
+        "number": "4",
+        "title": "EXPERIMENTS",
+        "start": 20950
+      },
+      {
+        "number": "4.1",
+        "title": "Setup",
+        "start": 21706
+      },
+      {
+        "number": "4.2",
+        "title": "Deep: Querying Nested JSON Objects",
+        "start": 22616
+      },
+      {
+        "number": "4.3",
+        "title": "UNNEST: Querying Nested JSON Arrays",
+        "start": 25134
+      },
+      {
+        "number": "4.4",
+        "title": "Custom Workloads: OLAP – BI Benchmark",
+        "start": 27902
+      },
+      {
+        "number": "4.5",
+        "title": "Discussion",
+        "start": 31132
+      },
+      {
+        "number": "5",
+        "title": "CONCLUSION",
+        "start": 32095
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Serge Abiteboul, Marcelo Arenas, Pablo Barceló, Meghyn Bienvenu, Diego Calvanese, Claire David, Richard Hull, Eyke Hüllermeier, Benny Kimelfeld, Leonid Libkin, Wim Martens, Tova Milo, Filip Murlak, Frank Neven, Magdalena Ortiz, Thomas Schwentick, Julia Stoyanovich, Jianwen Su, Dan Suciu, Victor Vianu, and Ke Yi. 2018. Research Directions for Principles of Data Management (Dagstuhl Perspectives Wor",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Tim Bray et al .2014. The javascript object notation (json) data interchange format. (2014).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Yueguo Chen, Xiongpai Qin, Haoqiong Bian, Jun Chen, Zhaoan Dong, Xiaoyong Du, Yanjie Gao, Dehai Liu, Jiaheng Lu, and Huijie Zhang. 2014. A Study of SQL-on-Hadoop Systems. In BPOE (LNCS, Vol. 8807). Springer, 154–166.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Richard L. Cole, Florian Funke, Leo Giakoumakis, Wey Guy, Alfons Kemper, Stefan Krompass, Harumi A. Kuno, Raghunath Othayoth Nambiar, Thomas Neumann, Meikel Poess, Kai-Uwe Sattler, Michael Seibold, Eric Simon, and Florian Waas. 2011. The mixed workload CH-ben CHmark. In DBTest, Goetz Graefe and Kenneth Salem (Eds.). ACM, 8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Brian F. Cooper, Adam Silberstein, Erwin Tam, Raghu Ramakrishnan, and Russell Sears. 2010. Benchmarking cloud serving systems with YCSB. In SoCC. ACM, 143–154.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Djellel Eddine Difallah, Andrew Pavlo, Carlo Curino, and Philippe CudréMauroux. 2013. OLTP-Bench: An Extensible Testbed for Benchmarking Relational Databases. Proc. VLDB Endow. 7, 4 (2013), 277–288.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Orri Erling, Alex Averbuch, Josep Lluís Larriba-Pey, Hassan Chafi, Andrey Gubichev, Arnau Prat-Pérez, Minh-Duc Pham, and Peter A. Boncz. 2015. The LDBC Social Network Benchmark: Interactive Workload. In SIGMOD. ACM, 619–630.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Jim Gray. 1993. Database and Transaction Processing Performance Handbook. InThe Benchmark Handbook for Database and Transaction Systems (2nd Edition). Morgan Kaufmann.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Anja Gruenheid, Shaleen Deep, Kruthi Nagaraj, Hiro Naito, Jeffrey F. Naughton, and Stratis Viglas. 2020. DIAMetrics: Benchmarking Query Engines at Scale. Proc. VLDB Endow. 13, 12 (2020), 3285–3298.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Henrik Ingo and David Daly. 2020. Automated system performance testing at MongoDB. In DBTest@SIGMOD. 3:1–3:6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Shiva Jahangiri. 2021. Wisconsin Benchmark Data Generator: To JSON and Beyond. In SIGMOD, Guoliang Li, Zhanhuai Li, Stratos Idreos, and Divesh Srivastava (Eds.). ACM, 2887–2889.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Asya Kamsky. 2019. Adapting TPC-C Benchmark to Measure Performance of Multi-Document Transactions in MongoDB. Proc. VLDB Endow. 12, 12 (2019), 2254–2262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Anthony G Read. 2006. DeWitt clauses: Can we protect purchasers without hurting Microsoft. Rev. Litig. 25 (2006), 387.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. (2020), 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "Daniel Ritter, Norman May, Kai Sachs, and Stefanie Rinderle-Ma. 2016. Benchmarking integration pattern implementations. In DEBS. ACM, 125–136.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2016. SQLSmith. https: //github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Adrian Vogelsgesang, Michael Haubenschild, Jan Finis, Alfons Kemper, Viktor Leis, Tobias Mühlbauer, Thomas Neumann, and Manuel Then. 2018. Get Real: How Benchmarks Fail to Represent the Real World. In DBTest@SIGMOD, Alexander Böhm and Tilmann Rabl (Eds.). ACM, 1:1–1:6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. (2020), 955–970. AQUERIES While for MongoDB no specification of a schema is requested, PostgreSQL requires it: in this case the following create statement has been used CREATE TABLE \"Foo\" (_jdata_ JSONB). 7 DBTest’22, June 17",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 14,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. (2020), 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 14,18]",
+      "technique": "pqs",
+      "sentence": ", [ 14,18]).",
+      "context_before": "extensible, scalable, JSON-specific benchmark, with which we evaluated two wellknown document stores. For the time, it is possible to gain deeper insights into strengths (cf. conclusion (1)), weaknesses (cf. conclusions (2)–(3), (5)) of these systems, especially in the areas of nested arrays and analytical queries (e. g., to avoid data integration into a column store for efficient analysis of JSON data), and potential improvements (cf. conclusions (4), (6)). In future work, we plan to look into weaknesses (e. g., memory consumption), explore solutions (e. g., optimizations) and fuzzing (e. g.",
+      "context_after": "DeepBench will help to characterize and evaluate document store workloads. 3No index support in MongoDB for ORDER BY, visited 4/22: https://bit.ly/3va3oyB 4MongoDB Workload Isolation, visited 4/22: https://bit.ly/3t5JEt7 6 DeepBench – Benchmarking JSON Document Stores DBTest’22, June 17, 2022, Philadelphia, PA, USA REFERENCES [1]Serge Abiteboul, Marcelo Arenas, Pablo Barceló, Meghyn Bienvenu, Diego Calvanese, Claire David, Richard Hull, Eyke Hüllermeier, Benny Kimelfeld, Leonid Libkin, Wim Martens, Tova Milo, Filip Murlak, Frank Neven, Magdalena Ortiz, Thomas Schwentick, Julia Stoyanovich, Ji",
+      "section": "5 CONCLUSION",
+      "page": 6,
+      "char_offset": 32879,
+      "cited_reference": {
+        "number": 14,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. (2020), 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "DeepBench is a benchmark for JSON document stores, filling a gap the authors identify: JSON has spread rapidly as an exchange and storage format and specialised document stores are widely used, but no JSON-specific benchmark existed. It covers nested JSON data and queries, with configurable domain-independent scale levels such as document size and concurrency alongside JSON-specific ones such as object and array nesting.",
+    "narrative": "A single citation in the conclusion, where SQLancer's line of work is listed as related database testing research.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3533767_3534364.json b/_data/papers/paper_doi_10_1145_3533767_3534364.json
new file mode 100644
index 0000000..86fcee6
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3533767_3534364.json
@@ -0,0 +1,953 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:40Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3533767.3534364",
+    "title": "Unicorn: detect runtime errors in time-series databases with hybrid input synthesis",
+    "authors": [
+      "Zhiyong Wu",
+      "Jie Liang",
+      "Mingzhe Wang",
+      "Chijin Zhou",
+      "Yu Jiang"
+    ],
+    "year": 2022,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3533767.3534364",
+    "arxiv_id": null,
+    "s2_paper_id": "1ac6b3a3904b6c9022206be7c0b9d180b84b1fb4",
+    "url": "https://doi.org/10.1145/3533767.3534364",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3533767.3534364",
+      "retrieved_at": "2026-09-09T01:33:40Z",
+      "chars": 67061,
+      "content_sha256": "sha256:1c75c87c5dec2bb6d7e46f0f5024f36ab94559a3c27bcbf48f0f2e360fdd0ce3"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2810
+      },
+      {
+        "number": "2",
+        "title": "TIME-SERIES DATABASES",
+        "start": 9173
+      },
+      {
+        "number": "2.1",
+        "title": "Employing Time-Series Data",
+        "start": 10666
+      },
+      {
+        "number": "2.2",
+        "title": "Handling Exceptions Implicitly",
+        "start": 13046
+      },
+      {
+        "number": "4.1",
+        "title": "Hybrid Input Synthesis",
+        "start": 21475
+      },
+      {
+        "number": "4.1.2",
+        "title": "Time-Series Mutation. With the hybrid input specification,",
+        "start": 24863
+      },
+      {
+        "number": "4.2",
+        "title": "Proactive Exception Detection",
+        "start": 27017
+      },
+      {
+        "number": "4.2.1",
+        "title": "Agent. As Figure 2 shows, time-series databases register a",
+        "start": 29763
+      },
+      {
+        "number": "6.1",
+        "title": "Evaluation Setup",
+        "start": 35887
+      },
+      {
+        "number": "6.2",
+        "title": "Overall Performance",
+        "start": 38385
+      },
+      {
+        "number": "6.3",
+        "title": "Effectiveness of Hybrid Input Synthesis",
+        "start": 43418
+      },
+      {
+        "number": "6.4",
+        "title": "Effectiveness of Exception Detection",
+        "start": 46946
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Mike Aizatsky, Kostya Serebryany, Oliver Chang, Abhishek Arya, and Meredith Whittaker. 2016. Continuous fuzzing for open source software. https://opensource .googleblog .com/2016/12/announcing-oss-fuzz-continuousfuzzing .html. [Online; accessed 15-May-2021].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "ASuherman, Masayoshi KURITA, Katsuhiko Nonomura, and Samurai Chanko. 2016. GridDB: Highly Scalable, In-Memory NoSQL Time Series Database Optimized for IoT and Big Data. https://griddb .org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Alex Crichton Carl Lerche. 2016. Tokio:Build reliable network applications without compromising speed. https://tokio .rs/. [Online; accessed 15-May-2021].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Peng Chen and Hao Chen. 2018. Angora: Efficient Fuzzing by Principled Search. In 2018, San Francisco, California, USA. IEEE Computer Society, 711–725. https: //doi .org/10 .1109/SP .2018 .00046",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "cppreference. 2017. GESH, MediaWiki. https://en .cppreference .com/w/cpp/ error/exception. [Online; accessed 15-May-2021].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "TAOS Data. 2019. TDengine: Big Data Platform Designed and Optimized for the Internet of Things (IoT). https://www .taosdata .com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Michael Eddington. 2015. Peach Fuzzer: Discover unknown vulnerabilities. http://web .archive .org/web/20210121202148/https://www .peach .tech/. [Online; accessed 15-May-2021].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Philippe Esling and Carlos Agon. 2012. Time-series data mining. ACM Computing Surveys (CSUR) 45, 1 (2012), 1–34.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Tak-chung Fu. 2011. A review on time series data mining. Engineering Applications of Artificial Intelligence 24, 1 (2011), 164–181.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Google. [n.d.]. AddressSanitizer. https://clang .llvm .org/docs/ AddressSanitizer .html. [Online; May 25, 2022].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Vlad Ilyushchenko, Brian Thomas Smith, and TheTanc. 2014. QuestDB: Database Designed to Process Time Series Data of IoT, Faster. https://github .com/questdb/ questdb",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2020. APOLLO: a toolchain for automatically detecting, reporting, and diagnosing performance bugs in DBMSs. https://github .com/sslab-gatech/apollo",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2020. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB). Tokyo, Japan.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Seulbae Kim, Meng Xu, Sanidhya Kashyap, Jungyeon Yoon, Wen Xu, and Taesoo Kim. 2019. Finding semantic bugs in file systems with an extensible fuzzing framework. In Proceedings of the 27th ACM Symposium on Operating Systems Principles. 147–161. 261 ISSTA ’22, July 18–22, 2022, Virtual, South Korea Z.Wu, J.Liang, M.Wang, C.Zhou, Y.Jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "lcamtuf. 2017. American Fuzzy Lop (AFL). http://lcamtuf .coredump .cx/afl/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jie Liang, Mingzhe Wang, Chijin Zhou, Zhiyong Wu, Yu Jiang, Jianzhong Liu, Zhe Liu, and Jiaguang Sun. 2022. PATA: Fuzzing with Path Aware Taint Analysis. In2022. IEEE Computer Society, Los Alamitos, CA, USA. 154ś170.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "libfuzzer@googlegroups.com. 2020. libFuzzer – a library for coverage-guided fuzz testing. https://llvm .org/docs/LibFuzzer .html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Rui Liu and Jun Yuan. 2019. Benchmark Time Series Database with IoTDBBenchmark for IoT Scenarios. CoRR abs/1901.08304 (2019). arXiv:1901.08304 http://arxiv .org/abs/1901 .08304",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Zhengxiong Luo, Feilong Zuo, Yu Jiang, Jian Gao, Xun Jiao, and Jiaguang Sun. 2019. Polar: Function Code Aware Fuzz Testing of ICS Protocol. ACM Trans. Embed. Comput. Syst. 18, 5s (2019), 93:1–93:22. https://doi .org/10 .1145/3358227",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Zhengxiong Luo, Feilong Zuo, Yuheng Shen, Xun Jiao, Wanli Chang, and Yu Jiang. 2020. ICS protocol fuzzing: coverage guided packet crack and generation. In2020 57th ACM/IEEE Design Automation Conference (DAC). IEEE, 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Barton P. Miller, Lars Fredriksen, and Bryan So. 1990. An Empirical Study of the Reliability of UNIX Utilities. Commun. ACM 33, 12, 32–44. https://doi .org/ 10.1145/96267 .96279",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Oracle. 2017. Java Exception. https://docs .oracle .com/javase/tutorial/essential/ exceptions/runtime .html. [Online; accessed 15-May-2021].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Fernando Paladini and Brian Hawkins. 2014. KairosDB: Fast Time Series Database on Cassandra. https://kairosdb .github .io/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Manuel Rigger. 2019. Apache IoTDB: Database for Internet of Things. https: //iotdb .apache .org/",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger. 2020. SQLancer: detecting logic bugs in DBMS. https: //github .com/sqlancer/sqlancer",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 2020 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Sacramento, California, United States) (ESEC/FSE 2020). https://doi .org/10 .1145/3368089 .3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https://doi .org/10 .1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 667–682. https://www .usenix .org/conference/osdi20/presentation/ rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 29,
+      "text": "Christoph Rudolf. 2017. SQL, no SQL or new SQL–comparison and applicability for Smart Spaces. Network Architectures and Services (2017).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Mitja Schmakeit, Florian Stinner, DJH Ziegeldorf, Ing Klaus Wehrle, and Ing Dirk Müller. 2017. Performance Evaluation of Low-Overhead Messaging Protocols and Time Series Databases via a Common Middleware. (2017).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https://github .com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2020. SQLsmith: Score list. https://github .com/anse1/sqlsmith/wiki#score-list",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Donald R. Slutz. [n.d.]. Massive Stochastic Testing of SQL. In VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, New York, USA. Morgan Kaufmann, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Hao Sun, Yuheng Shen, Cong Wang, Jianzhong Liu, Yu Jiang, Ting Chen, and Aiguo Cui. 2021. HEALER: Relation Learning Guided Kernel Fuzzing. In SOSP ’21: ACMSIGOPS 28th Symposium on Operating Systems Principles, Virtual Event / Koblenz, Germany, October 26-29, 2021, Robbert van Renesse and Nickolai Zeldovich (Eds.). ACM, 344–358. https://doi .org/10 .1145/3477132 .3483547",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Chen Wang, Xiangdong Huang, Jialin Qiao, Tian Jiang, Lei Rui, Jinrui Zhang, Rong Kang, Julian Feinauer, Kevin Mcgrail, Peng Wang, Diaohan Luo, Jun Yuan, Jianmin Wang, and Jiaguang Sun. 2020. Apache IoTDB: Time-series database for Internet of Things. Proc. VLDB Endow. 13, 12 (2020), 2901–2904. https: //doi .org/10 .14778/3415478 .3415504",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry practice of coverage-guided enterprise-level DBMS fuzzing. In Software Engineering in Practice (ICSE-SEIP). IEEE, 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Wen Xu, Hyungon Moon, Sanidhya Kashyap, Po-Ning Tseng, and Taesoo Kim. 2019. Fuzzing file systems via two-dimensional input space exploration. In. IEEE, 818–834.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Chijin Zhou, Mingzhe Wang, Jie Liang, Zhe Liu, and Yu Jiang. 2020. Zeror: Speed up fuzzing with coverage-sensitive tracing and scheduling. In Proceedings of the 35th IEEE/ACM International Conference on Automated Software Engineering. 858–870.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Feilong Zuo, Zhengxiong Luo, Junze Yu, Zhe Liu, and Yu Jiang. 2021. PAVFuzz: State-Sensitive Fuzz Testing of Protocols in Autonomous Vehicles. In 2021 58th ACM/IEEE Design Automation Conference (DAC). IEEE, 823–828. 262",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 24,
+      "text": "Manuel Rigger. 2019. Apache IoTDB: Database for Internet of Things. https: //iotdb .apache .org/",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger. 2020. SQLancer: detecting logic bugs in DBMS. https: //github .com/sqlancer/sqlancer",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 2020 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Sacramento, California, United States) (ESEC/FSE 2020). https://doi .org/10 .1145/3368089 .3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https://doi .org/10 .1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 28,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 667–682. https://www .usenix .org/conference/osdi20/presentation/ rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, Unicorn outperforms SQLsmith and SQLancer on widely used time-series databases IoTDB, KairosDB, TimescaleDB, TDEngine, QuestDB, and GridDB in the number of basic blocks by 21%-199% and 34%-693%, respectively.",
+      "context_before": "and serious bugs are undetectable because of implicitly handled exceptions. In this paper, we propose Unicorn to secure time-series databases with automated fuzzing. First, we design hybrid input synthesis to generate high-quality queries which not only cover time-series features but also ensure grammar correctness. Then, Unicorn uses proactive exception detection to discover minuscule-symptom bugs which hide behind implicit exception handling. With the specialized design oriented to time-series databases, Unicorn outperforms the state-of-the-art database fuzzers in terms of coverage and bugs.",
+      "context_after": "More importantly, Unicorn has discovered 42 previously unknown bugs. CCSCONCEPTS •Software and its engineering →Software maintenance tools; •Security and privacy →Database and storage security. KEYWORDS Time-series Databases, Runtime Error, Hybrid Input Synthesis ∗Jie Liang and Yu Jiang are the corresponding authors. Request permissions from permissions@acm.org. ISSTA ’22, July 18–22, 2022, Virtual, South Korea ©2022 Association for Computing Machinery. ACMISBN 978-1-4503-9379-9/22/07.. . $15.00 https://doi .org/10 .1145/3533767 .3534364ACM Reference Format: Zhiyong Wu, Jie Liang, Mingzhe Wa",
+      "section": null,
+      "page": 1,
+      "char_offset": 1639,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We also adapted the industrial fuzzers SQLancer [ 25] and SQLsmith [ 31] for comparison.",
+      "context_before": "original seeds and generates new time-series queries by changing the time-series nodes of AST. To detect exceptions handled implicitly, proactive exception detection directly captures exception information from the runtime environment and analyzes whether it is an anomaly. Specifically, instead of passively receiving the program’s state, Unicorn inserts an agent into each process to proactively catch the exceptions and send them to the anomaly detector for analyzing and reporting. For evaluation, we used Unicorn to perform fuzzing on IoTDB, KairosDB, QuestDB, TimescaleDB, TDEngine, and GridDB.",
+      "context_after": "Unicorn covered 115.75% more basic blocks on average than the best results of other fuzzers. In addition, Unicorn detected 42 previously unknown bugs. In conclusion, our paper makes the following contributions:•We observe that current fuzzing approaches are hard to effectively test time-series databases. The two main challenges are generating grammatically correct queries and capturing exceptions handled implicitly. •We propose hybrid input synthesis and proactive exception detection to address the aforementioned challenges. We also implement these approaches in Unicorn. •We evaluate Unicorn",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 8299,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "•We evaluate Unicorn on 6 popular time-series databases against state-of-the-art fuzzers SQLsmith and SQLancer.",
+      "context_before": "[ 31] for comparison. Unicorn covered 115.75% more basic blocks on average than the best results of other fuzzers. In addition, Unicorn detected 42 previously unknown bugs. In conclusion, our paper makes the following contributions:•We observe that current fuzzing approaches are hard to effectively test time-series databases. The two main challenges are generating grammatically correct queries and capturing exceptions handled implicitly. •We propose hybrid input synthesis and proactive exception detection to address the aforementioned challenges. We also implement these approaches in Unicorn.",
+      "context_after": "The results show that Unicorn outperforms others and 42 previously-unknown bugs are detected. 2 TIME-SERIES DATABASES As an infrastructure for IoT data storage and analysis, time-series databases play an important role in promoting the development of Internet of Things. Generally, the time-series database is a kind of large-scale software to manipulate and manage IoT data, it handles the operation requests from various clients (including IoT devices, PC, etc.), and carries out unified management and control to ensure the security and integrity of IoT data [ 18]. In embedded application scenar",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 8966,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[ 4,7,15,25,31]",
+      "technique": null,
+      "sentence": "3 CHALLENGES INFUZZING TIME-SERIES DATABASES Fuzzing has achieved remarkable results in the field of software testing [ 4,7,15,25,31].",
+      "context_before": "ndling Mechanism Device Client1 Device Client2 Device Client3Worker1 Worker2 Worker3Figure 2: Implicit exception handling mechanism in a famous time-series database Apache IoTDB. In the single-node running model of IoTDB, a server is connected by lots of device clients. The server generally has three core modules, namely Time-Series Storage, Time-Oriented Query, and IoT Data Management. The interactions of the three components are frequent and complex. When an exception (e.g loss of data) happens in any component, the IoTDB will implicitly handle it to ensure the whole system running normally.",
+      "context_after": "However, due to the two major characteristics in time-series databases, there are two major challenges for fuzzing time-series databases, namely generating grammatically-correct time-series queries andcapturing exceptions handled implicitly. To demonstrate the challenges of performing fuzzing in timeseries databases, we present a bug causing damage to the database integrity of Apache IoTDB. As one of the most important features of databases, database integrity requires that the data stored in the database is logically consistent, correct, and valid. It reflects that create timeseries root. st",
+      "section": "2.2 Handling Exceptions Implicitly",
+      "page": 3,
+      "char_offset": 15222,
+      "cited_reference": {
+        "number": 25,
+        "text": "Manuel Rigger. 2020. SQLancer: detecting logic bugs in DBMS. https: //github .com/sqlancer/sqlancer",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "6 EVALUATION In this section, we evaluate the effectiveness of Unicorn in terms of coverage and bug discovery against state-of-the-art database fuzzers—SQLsmith and SQLancer.",
+      "context_before": "specification. We instrument databases to collect coverage feedback. For Java databases, we implement it on ASM framework. And for C/C++ databases, we implement it based on Clang. Proactive Exception Detection. Detection is based on agent and anomaly detector. For VM-based time-series databases, the agent is implemented in 2,147 lines of Java code with ASM framework. For OS-based databases, the agent is implemented in 1,646 lines of C++ and Rust code and leverages ptrace for subprocesses error detection. In addition, we implemented the anomaly detector in Unicorn with 1,345 lines of Rust code.",
+      "context_after": "Besides, we also evaluate the effectiveness of two parts of Unicorn, namely hybrid input synthesis, as well as proactive exception detection. 6.1 Evaluation Setup Target Time-Series Databases. To evaluate the generality of Unicorn, six popular and widely-used open-source time-series databases are selected as the targets, namely IoTDB, KairosDB,QuestDB, TimescaleDB, TDEngine, and GridDB. Table 1 shows the features of those target databases. Table 1: Features of Chosen Time-series Databases Target Language Schema Grammar Dependency IoTDB Java tree-based SQL-like N/A KairosDB Java tag-based Http",
+      "section": "4.2.1 Agent. As Figure 2 shows, time-series databases register a",
+      "page": 7,
+      "char_offset": 35569,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Table 1: Features of Chosen Time-series Databases Target Language Schema Grammar Dependency IoTDB Java tree-based SQL-like N/A KairosDB Java tag-based Http API Cassandra QuestDB Java column-based SQL-like N/A TDEngine C++ table-based SQL-like N/A TimescaleDB C relational SQL-like PostgreSQL GridDB C++ tag-based SQL-like N/A Compared Fuzzers For the performance comparison, we have also adapted the blackbox fuzzers SQLsmith and SQLancer, which are widely used to test relational databases in industry, to those time-series databases.",
+      "context_before": "is section, we evaluate the effectiveness of Unicorn in terms of coverage and bug discovery against state-of-the-art database fuzzers—SQLsmith and SQLancer. Besides, we also evaluate the effectiveness of two parts of Unicorn, namely hybrid input synthesis, as well as proactive exception detection. 6.1 Evaluation Setup Target Time-Series Databases. To evaluate the generality of Unicorn, six popular and widely-used open-source time-series databases are selected as the targets, namely IoTDB, KairosDB,QuestDB, TimescaleDB, TDEngine, and GridDB. Table 1 shows the features of those target databases.",
+      "context_after": "Other fuzzers such as Sqirrel [38] are preliminary evaluated in Section Section 7 because of the limitation of their language support. •SQLsmith [31] is one of the state-of-the-art generationbased DBMS Fuzzer. It continuously generates a large amount of syntactically correct database inputs with the pre-built abstract syntax tree model and sends them to the target database server. If the database server crashes, a bug is very likely to be detected. •SQLancer [25] is a fuzzer for hunting logic bugs in DBMS. It detects logic bugs by constructing invariant oracles [ 26– 28] and checking whether",
+      "section": "6.1 Evaluation Setup",
+      "page": 7,
+      "char_offset": 36188,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "•SQLancer [25] is a fuzzer for hunting logic bugs in DBMS.",
+      "context_before": "adapted the blackbox fuzzers SQLsmith and SQLancer, which are widely used to test relational databases in industry, to those time-series databases. Other fuzzers such as Sqirrel [38] are preliminary evaluated in Section Section 7 because of the limitation of their language support. •SQLsmith [31] is one of the state-of-the-art generationbased DBMS Fuzzer. It continuously generates a large amount of syntactically correct database inputs with the pre-built abstract syntax tree model and sends them to the target database server. If the database server crashes, a bug is very likely to be detected.",
+      "context_after": "It detects logic bugs by constructing invariant oracles [ 26– 28] and checking whether results violate semantic logic. For example, it synthesizes queries to fetch a random row from an existing relational table in the database. If the database fails to fetch that, then the database might have a bug. Performance Metrics. We use the number of basic blocks covered and the number of bugs detected as two basic metrics. For a fair comparison, when we finish fuzzing, we collect the seeds generated by each fuzzer and dry-run the seeds to uniform the basic block coverage. To deduplicate unique bugs, w",
+      "section": "6.1 Evaluation Setup",
+      "page": 7,
+      "char_offset": 37177,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[ 26–\n28]",
+      "technique": "norec",
+      "sentence": "It detects logic bugs by constructing invariant oracles [ 26– 28] and checking whether results violate semantic logic.",
+      "context_before": "re widely used to test relational databases in industry, to those time-series databases. Other fuzzers such as Sqirrel [38] are preliminary evaluated in Section Section 7 because of the limitation of their language support. •SQLsmith [31] is one of the state-of-the-art generationbased DBMS Fuzzer. It continuously generates a large amount of syntactically correct database inputs with the pre-built abstract syntax tree model and sends them to the target database server. If the database server crashes, a bug is very likely to be detected. •SQLancer [25] is a fuzzer for hunting logic bugs in DBMS.",
+      "context_after": "For example, it synthesizes queries to fetch a random row from an existing relational table in the database. If the database fails to fetch that, then the database might have a bug. Performance Metrics. We use the number of basic blocks covered and the number of bugs detected as two basic metrics. For a fair comparison, when we finish fuzzing, we collect the seeds generated by each fuzzer and dry-run the seeds to uniform the basic block coverage. To deduplicate unique bugs, we extract and compare their call stacks. We also manually trace back to their relevant source code to further confirm.",
+      "section": "6.1 Evaluation Setup",
+      "page": 7,
+      "char_offset": 37236,
+      "cited_reference": {
+        "number": 26,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 2020 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Sacramento, C",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "2 Overall Performance Unicorn performs better than SQLsmith and SQLancer both in the number of basic blocks covered and bugs triggered.",
+      "context_before": "sic block coverage. To deduplicate unique bugs, we extract and compare their call stacks. We also manually trace back to their relevant source code to further confirm. For evaluations, we ran each fuzzing instance (one fuzzer + one target database) for 12 hours with 5 repeated times. Experiment Environment. We perform the evaluation on a machine with 40 cores (Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz), 256 GiB RAM, and Ubuntu 20.04 as the operating system. Each fuzzing instance ran individually with 1 CPU and 8GiB of RAM. The initial seeds were collected from the built-in integration tests. 6.",
+      "context_after": "Specifically, in our 12-hour experiment, Unicorn covered 115.75% more basic blocks on four successfully adapted databases than other fuzzers’ best results on average. Also, it found 42 previously unknown bugs, while other fuzzers detected 2 at most. Coverage. Table 2 demonstrates the number of basic blocks covered by each fuzzer in 12-hour experiments. From the table, 257 ISSTA ’22, July 18–22, 2022, Virtual, South Korea Z.Wu, J.Liang, M.Wang, C.Zhou, Y.Jiang Figure 7: The growing trend of the number of basic blocks covered when fuzzing each time-series database by Unicorn, SQLsmith, and SQL",
+      "section": "6.2 Overall Performance",
+      "page": 7,
+      "char_offset": 38387,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Figure 7: The growing trend of the number of basic blocks covered when fuzzing each time-series database by Unicorn, SQLsmith, and SQLancer over 5 runs in 12 hours.",
+      "context_before": "Overall Performance Unicorn performs better than SQLsmith and SQLancer both in the number of basic blocks covered and bugs triggered. Specifically, in our 12-hour experiment, Unicorn covered 115.75% more basic blocks on four successfully adapted databases than other fuzzers’ best results on average. Also, it found 42 previously unknown bugs, while other fuzzers detected 2 at most. Coverage. Table 2 demonstrates the number of basic blocks covered by each fuzzer in 12-hour experiments. From the table, 257 ISSTA ’22, July 18–22, 2022, Virtual, South Korea Z.Wu, J.Liang, M.Wang, C.Zhou, Y.Jiang",
+      "context_after": "Displayed are the median and the 95% confidence intervals. we can see that Unicorn outperformed other fuzzers in terms of basic block coverage. Compared to SQLsmith and SQLancer, on the four successfully adapted databases for both of them (i.e. IoTDB, QuestDB, TDEngine, and TimescaleDB), Unicorn covered 115.75% more basic blocks on average than their best results. In particular, Unicorn covered 690% more basic blocks than SQLancer in GridDB. Unicorn outperforms other fuzzers because it promises syntax correctness and time series by hybrid input synthesis. In contrast, SQLsmith and SQLancer do",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 38988,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Compared to SQLsmith and SQLancer, on the four successfully adapted databases for both of them (i.",
+      "context_before": "it found 42 previously unknown bugs, while other fuzzers detected 2 at most. Coverage. Table 2 demonstrates the number of basic blocks covered by each fuzzer in 12-hour experiments. From the table, 257 ISSTA ’22, July 18–22, 2022, Virtual, South Korea Z.Wu, J.Liang, M.Wang, C.Zhou, Y.Jiang Figure 7: The growing trend of the number of basic blocks covered when fuzzing each time-series database by Unicorn, SQLsmith, and SQLancer over 5 runs in 12 hours. Displayed are the median and the 95% confidence intervals. we can see that Unicorn outperformed other fuzzers in terms of basic block coverage.",
+      "context_after": "e. IoTDB, QuestDB, TDEngine, and TimescaleDB), Unicorn covered 115.75% more basic blocks on average than their best results. In particular, Unicorn covered 690% more basic blocks than SQLancer in GridDB. Unicorn outperforms other fuzzers because it promises syntax correctness and time series by hybrid input synthesis. In contrast, SQLsmith and SQLancer do not match the time-series characteristics because SQLsmith and SQLancer only focus on the relational query in the relational model. In addition, Unicorn performs well on all tested time-series databases while SQLancer and SQLsmith both can ha",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 39296,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In particular, Unicorn covered 690% more basic blocks than SQLancer in GridDB.",
+      "context_before": ", 2022, Virtual, South Korea Z.Wu, J.Liang, M.Wang, C.Zhou, Y.Jiang Figure 7: The growing trend of the number of basic blocks covered when fuzzing each time-series database by Unicorn, SQLsmith, and SQLancer over 5 runs in 12 hours. Displayed are the median and the 95% confidence intervals. we can see that Unicorn outperformed other fuzzers in terms of basic block coverage. Compared to SQLsmith and SQLancer, on the four successfully adapted databases for both of them (i.e. IoTDB, QuestDB, TDEngine, and TimescaleDB), Unicorn covered 115.75% more basic blocks on average than their best results.",
+      "context_after": "Unicorn outperforms other fuzzers because it promises syntax correctness and time series by hybrid input synthesis. In contrast, SQLsmith and SQLancer do not match the time-series characteristics because SQLsmith and SQLancer only focus on the relational query in the relational model. In addition, Unicorn performs well on all tested time-series databases while SQLancer and SQLsmith both can hardly adapt to KairosDB. Because KairosDB does not accept the SQL-like inputs but the JSON format input by HTTP-API, the generation model of SQLsmith and SQLancer can hardly generate inputs like that. Ins",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 39519,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In contrast, SQLsmith and SQLancer do not match the time-series characteristics because SQLsmith and SQLancer only focus on the relational query in the relational model.",
+      "context_before": "and SQLancer over 5 runs in 12 hours. Displayed are the median and the 95% confidence intervals. we can see that Unicorn outperformed other fuzzers in terms of basic block coverage. Compared to SQLsmith and SQLancer, on the four successfully adapted databases for both of them (i.e. IoTDB, QuestDB, TDEngine, and TimescaleDB), Unicorn covered 115.75% more basic blocks on average than their best results. In particular, Unicorn covered 690% more basic blocks than SQLancer in GridDB. Unicorn outperforms other fuzzers because it promises syntax correctness and time series by hybrid input synthesis.",
+      "context_after": "In addition, Unicorn performs well on all tested time-series databases while SQLancer and SQLsmith both can hardly adapt to KairosDB. Because KairosDB does not accept the SQL-like inputs but the JSON format input by HTTP-API, the generation model of SQLsmith and SQLancer can hardly generate inputs like that. Instead, Unicorn constructs the hybrid input specification for input generation by adding the time-series elements to the JSON grammar specification. As a result, Unicorn adapts KairosDB well to explore its state space. Table 2: Average Number of Basic Blocks on Time-Series Databases For",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 39714,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In addition, Unicorn performs well on all tested time-series databases while SQLancer and SQLsmith both can hardly adapt to KairosDB.",
+      "context_before": "ck coverage. Compared to SQLsmith and SQLancer, on the four successfully adapted databases for both of them (i.e. IoTDB, QuestDB, TDEngine, and TimescaleDB), Unicorn covered 115.75% more basic blocks on average than their best results. In particular, Unicorn covered 690% more basic blocks than SQLancer in GridDB. Unicorn outperforms other fuzzers because it promises syntax correctness and time series by hybrid input synthesis. In contrast, SQLsmith and SQLancer do not match the time-series characteristics because SQLsmith and SQLancer only focus on the relational query in the relational model.",
+      "context_after": "Because KairosDB does not accept the SQL-like inputs but the JSON format input by HTTP-API, the generation model of SQLsmith and SQLancer can hardly generate inputs like that. Instead, Unicorn constructs the hybrid input specification for input generation by adding the time-series elements to the JSON grammar specification. As a result, Unicorn adapts KairosDB well to explore its state space. Table 2: Average Number of Basic Blocks on Time-Series Databases For Each Fuzzer. Target SQLsmith SQLancer Unicorn IoTDB 4, 857 3, 754 14, 532 KairosDB N/A N/A 26, 489 QuestDB 5, 742 3, 298 10, 486 TDEng",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 39884,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Because KairosDB does not accept the SQL-like inputs but the JSON format input by HTTP-API, the generation model of SQLsmith and SQLancer can hardly generate inputs like that.",
+      "context_before": "gine, and TimescaleDB), Unicorn covered 115.75% more basic blocks on average than their best results. In particular, Unicorn covered 690% more basic blocks than SQLancer in GridDB. Unicorn outperforms other fuzzers because it promises syntax correctness and time series by hybrid input synthesis. In contrast, SQLsmith and SQLancer do not match the time-series characteristics because SQLsmith and SQLancer only focus on the relational query in the relational model. In addition, Unicorn performs well on all tested time-series databases while SQLancer and SQLsmith both can hardly adapt to KairosDB.",
+      "context_after": "Instead, Unicorn constructs the hybrid input specification for input generation by adding the time-series elements to the JSON grammar specification. As a result, Unicorn adapts KairosDB well to explore its state space. Table 2: Average Number of Basic Blocks on Time-Series Databases For Each Fuzzer. Target SQLsmith SQLancer Unicorn IoTDB 4, 857 3, 754 14, 532 KairosDB N/A N/A 26, 489 QuestDB 5, 742 3, 298 10, 486 TDEngine 3, 245 1, 423 8, 473 TimescaleDB 26, 423 13, 485 32, 124 GridDB N/A 2, 286 18, 145 Total 40, 276 24, 246 110, 249Figure 7 shows the growth trends of the number of basic blo",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 40018,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Target SQLsmith SQLancer Unicorn IoTDB 4, 857 3, 754 14, 532 KairosDB N/A N/A 26, 489 QuestDB 5, 742 3, 298 10, 486 TDEngine 3, 245 1, 423 8, 473 TimescaleDB 26, 423 13, 485 32, 124 GridDB N/A 2, 286 18, 145 Total 40, 276 24, 246 110, 249Figure 7 shows the growth trends of the number of basic blocks for each fuzzer over 12 hours.",
+      "context_before": ", Unicorn performs well on all tested time-series databases while SQLancer and SQLsmith both can hardly adapt to KairosDB. Because KairosDB does not accept the SQL-like inputs but the JSON format input by HTTP-API, the generation model of SQLsmith and SQLancer can hardly generate inputs like that. Instead, Unicorn constructs the hybrid input specification for input generation by adding the time-series elements to the JSON grammar specification. As a result, Unicorn adapts KairosDB well to explore its state space. Table 2: Average Number of Basic Blocks on Time-Series Databases For Each Fuzzer.",
+      "context_after": "By comparing the curves of Unicorn, SQLsmith and SQLancer in each sub-figure, we find that Unicorn is faster to cover basic blocks on target time-series databases than both SQLsmith and SQLancer. Specifically, in Figure 7 (a), Unicorn covers more basic blocks in IoTDB than SQLsmith and SQLancer from start to the end. The results on other time-series databases also have similar trends. The main reason is that the inputs generated by SQLsmith and SQLancer lack time-series elements, so most of them are invalid. Unicorn introduces time-series elements through hybrid input synthesis. The initial s",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 40496,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "By comparing the curves of Unicorn, SQLsmith and SQLancer in each sub-figure, we find that Unicorn is faster to cover basic blocks on target time-series databases than both SQLsmith and SQLancer.",
+      "context_before": "ybrid input specification for input generation by adding the time-series elements to the JSON grammar specification. As a result, Unicorn adapts KairosDB well to explore its state space. Table 2: Average Number of Basic Blocks on Time-Series Databases For Each Fuzzer. Target SQLsmith SQLancer Unicorn IoTDB 4, 857 3, 754 14, 532 KairosDB N/A N/A 26, 489 QuestDB 5, 742 3, 298 10, 486 TDEngine 3, 245 1, 423 8, 473 TimescaleDB 26, 423 13, 485 32, 124 GridDB N/A 2, 286 18, 145 Total 40, 276 24, 246 110, 249Figure 7 shows the growth trends of the number of basic blocks for each fuzzer over 12 hours.",
+      "context_after": "Specifically, in Figure 7 (a), Unicorn covers more basic blocks in IoTDB than SQLsmith and SQLancer from start to the end. The results on other time-series databases also have similar trends. The main reason is that the inputs generated by SQLsmith and SQLancer lack time-series elements, so most of them are invalid. Unicorn introduces time-series elements through hybrid input synthesis. The initial seeds are syntactically correct, which come from the test cases of time-series databases themselves. Based on them, hybrid input synthesis generates more valid inputs, which are helpful to explore",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 40828,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, in Figure 7 (a), Unicorn covers more basic blocks in IoTDB than SQLsmith and SQLancer from start to the end.",
+      "context_before": "Average Number of Basic Blocks on Time-Series Databases For Each Fuzzer. Target SQLsmith SQLancer Unicorn IoTDB 4, 857 3, 754 14, 532 KairosDB N/A N/A 26, 489 QuestDB 5, 742 3, 298 10, 486 TDEngine 3, 245 1, 423 8, 473 TimescaleDB 26, 423 13, 485 32, 124 GridDB N/A 2, 286 18, 145 Total 40, 276 24, 246 110, 249Figure 7 shows the growth trends of the number of basic blocks for each fuzzer over 12 hours. By comparing the curves of Unicorn, SQLsmith and SQLancer in each sub-figure, we find that Unicorn is faster to cover basic blocks on target time-series databases than both SQLsmith and SQLancer.",
+      "context_after": "The results on other time-series databases also have similar trends. The main reason is that the inputs generated by SQLsmith and SQLancer lack time-series elements, so most of them are invalid. Unicorn introduces time-series elements through hybrid input synthesis. The initial seeds are syntactically correct, which come from the test cases of time-series databases themselves. Based on them, hybrid input synthesis generates more valid inputs, which are helpful to explore the states of the target time-series databases. Therefore, Unicorn achieves higher coverage at the beginning and keeps ahea",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 41024,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The main reason is that the inputs generated by SQLsmith and SQLancer lack time-series elements, so most of them are invalid.",
+      "context_before": "ngine 3, 245 1, 423 8, 473 TimescaleDB 26, 423 13, 485 32, 124 GridDB N/A 2, 286 18, 145 Total 40, 276 24, 246 110, 249Figure 7 shows the growth trends of the number of basic blocks for each fuzzer over 12 hours. By comparing the curves of Unicorn, SQLsmith and SQLancer in each sub-figure, we find that Unicorn is faster to cover basic blocks on target time-series databases than both SQLsmith and SQLancer. Specifically, in Figure 7 (a), Unicorn covers more basic blocks in IoTDB than SQLsmith and SQLancer from start to the end. The results on other time-series databases also have similar trends.",
+      "context_after": "Unicorn introduces time-series elements through hybrid input synthesis. The initial seeds are syntactically correct, which come from the test cases of time-series databases themselves. Based on them, hybrid input synthesis generates more valid inputs, which are helpful to explore the states of the target time-series databases. Therefore, Unicorn achieves higher coverage at the beginning and keeps ahead to the end. Table 3: Total Number of Detected Bugs on Time-Series Databases By Each Fuzzer. Target SQLsmith SQLancer Unicorn IoTDB 1 0 18 KairosDB N/A N/A 2 QuestDB 0 0 3 TDEngine 0 0 6 Timesca",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 41216,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Target SQLsmith SQLancer Unicorn IoTDB 1 0 18 KairosDB N/A N/A 2 QuestDB 0 0 3 TDEngine 0 0 6 TimescaleDB 1 1 3 GridDB N/A 0 10 Total 2 1 42 Bugs.",
+      "context_before": "the inputs generated by SQLsmith and SQLancer lack time-series elements, so most of them are invalid. Unicorn introduces time-series elements through hybrid input synthesis. The initial seeds are syntactically correct, which come from the test cases of time-series databases themselves. Based on them, hybrid input synthesis generates more valid inputs, which are helpful to explore the states of the target time-series databases. Therefore, Unicorn achieves higher coverage at the beginning and keeps ahead to the end. Table 3: Total Number of Detected Bugs on Time-Series Databases By Each Fuzzer.",
+      "context_after": "Table 3 shows the number of bugs discovered in six timeseries databases by Unicorn, SQLsmith, and SQLancer. All these bugs have been confirmed by the database developers. Totally, 258 Unicorn: Detect Runtime Errors in Time-Series Databases with Hybrid Input Synthesis ISSTA ’22, July 18–22, 2022, Virtual, South Korea Unicorn detected 42 previously unknown bugs. Among them, 23 were from databases implemented in Java, which run in VM. Furthermore, 2 bugs were also found by SQLsmith, while SQLancer did not detect any bug. Unicorn found more bugs mainly for two reasons. First, Unicorn finds more b",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 41840,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Table 3 shows the number of bugs discovered in six timeseries databases by Unicorn, SQLsmith, and SQLancer.",
+      "context_before": "ugh hybrid input synthesis. The initial seeds are syntactically correct, which come from the test cases of time-series databases themselves. Based on them, hybrid input synthesis generates more valid inputs, which are helpful to explore the states of the target time-series databases. Therefore, Unicorn achieves higher coverage at the beginning and keeps ahead to the end. Table 3: Total Number of Detected Bugs on Time-Series Databases By Each Fuzzer. Target SQLsmith SQLancer Unicorn IoTDB 1 0 18 KairosDB N/A N/A 2 QuestDB 0 0 3 TDEngine 0 0 6 TimescaleDB 1 1 3 GridDB N/A 0 10 Total 2 1 42 Bugs.",
+      "context_after": "All these bugs have been confirmed by the database developers. Totally, 258 Unicorn: Detect Runtime Errors in Time-Series Databases with Hybrid Input Synthesis ISSTA ’22, July 18–22, 2022, Virtual, South Korea Unicorn detected 42 previously unknown bugs. Among them, 23 were from databases implemented in Java, which run in VM. Furthermore, 2 bugs were also found by SQLsmith, while SQLancer did not detect any bug. Unicorn found more bugs mainly for two reasons. First, Unicorn finds more basic blocks. With more basic blocks covered, Unicorn could explore more state space in the target time-serie",
+      "section": "6.2 Overall Performance",
+      "page": 8,
+      "char_offset": 41987,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Furthermore, 2 bugs were also found by SQLsmith, while SQLancer did not detect any bug.",
+      "context_before": "s By Each Fuzzer. Target SQLsmith SQLancer Unicorn IoTDB 1 0 18 KairosDB N/A N/A 2 QuestDB 0 0 3 TDEngine 0 0 6 TimescaleDB 1 1 3 GridDB N/A 0 10 Total 2 1 42 Bugs. Table 3 shows the number of bugs discovered in six timeseries databases by Unicorn, SQLsmith, and SQLancer. All these bugs have been confirmed by the database developers. Totally, 258 Unicorn: Detect Runtime Errors in Time-Series Databases with Hybrid Input Synthesis ISSTA ’22, July 18–22, 2022, Virtual, South Korea Unicorn detected 42 previously unknown bugs. Among them, 23 were from databases implemented in Java, which run in VM.",
+      "context_after": "Unicorn found more bugs mainly for two reasons. First, Unicorn finds more basic blocks. With more basic blocks covered, Unicorn could explore more state space in the target time-series databases, which assigns more possibility for Unicorn to detect more bugs. Second, Unicorn is able to detect bugs that are handled implicitly, while others cannot. For instance, of the discovered 18 bugs in IoTDB, only 5 bugs directly cause the server crash. Although some unchecked exceptions are caused by real bugs, such as returning the wrong results to the client, storing wrong time-series data, and memory e",
+      "section": "6.2 Overall Performance",
+      "page": 9,
+      "char_offset": 42423,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer uses 8,134 lines of Java code to support generating syntax-correct queries.",
+      "context_before": "exception detection, only 5 crashes were found. Overall, while monitoring all exceptions increases the overhead of Unicorn, It also assigns the ability to detect bugs that may cause serious damage to time-series databases in production applications. We believe reducing the overhead is significant to speed fuzzing. We plan to optimize it by reducing capture messages of agent. Cost of Adapting to New Time-series Databases. The scalability of database fuzzers plays a crucial role in industrial applications. Generally, the most labor-consuming task is building the input model of target databases.",
+      "context_after": "Different from conventional fuzzers, Unicorn doesn’t cost much to adapt to new time-series databases. With the collected ANTLR format grammar of traditional SQL specification, we only need to manually append less than 100 lines of code for the time-oriented model. Compare Against Coverage-Guided Database Fuzzers. Most coverage-guided database fuzzers are unable to test time-series databases directly because their instrumentation only supports C/C++. In contrast, Unicorn supports the heterogeneous implementation of C/C++ and Java. To further investigate Unicorn ’s performance, we also compare",
+      "section": "6.4 Effectiveness of Exception Detection",
+      "page": 10,
+      "char_offset": 52047,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "citation_marker",
+      "surface": "[ 13,25–27,31,33,38]",
+      "technique": null,
+      "sentence": "Various test oracles have been proposed to detect various bugs in databases and have achieved good results [ 13,25–27,31,33,38].",
+      "context_before": "he relational storage model of relational databases. It can not perform well for the time-series storage model. Unlike these works, Unicorn automatically constructs hybrid input specification for each time-series database by combining timeoriented models and traditional SQL specifications. With the timeoriented model, the generated queries have abundant time-series features. Besides, Unicorn is adaptable. Generally, it only needs to append less than 100 lines of code to the traditional specification for the time-oriented model to adapt to a new time-series database. Test Oracles for Databases.",
+      "context_after": "The basic test oracle is monitoring whether the tested databases crashes. SQLsmith [ 31] determines bugs in databases through the running state of the server. In other words, if the server crashes, there might be a bug in the database. Sqirrel [38] and Ratel [36] focuses on detecting memory corruption bugs in databases with the help of AddressSanitizer (ASAN) [ 10]. It records bugs when ASAN crashes the program. However, limited by the dependency on ASAN, it can only detect bugs in the C/C++ databases. Therefore, it can hardly detect problems for the database implemented in memorysafe languag",
+      "section": "6.4 Effectiveness of Exception Detection",
+      "page": 11,
+      "char_offset": 54747,
+      "cited_reference": {
+        "number": 25,
+        "text": "Manuel Rigger. 2020. SQLancer: detecting logic bugs in DBMS. https: //github .com/sqlancer/sqlancer",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M25",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 25] designs three test oracles according to the logic characteristics of relational databases.",
+      "context_before": "SQLsmith [ 31] determines bugs in databases through the running state of the server. In other words, if the server crashes, there might be a bug in the database. Sqirrel [38] and Ratel [36] focuses on detecting memory corruption bugs in databases with the help of AddressSanitizer (ASAN) [ 10]. It records bugs when ASAN crashes the program. However, limited by the dependency on ASAN, it can only detect bugs in the C/C++ databases. Therefore, it can hardly detect problems for the database implemented in memorysafe languages such as Java. Other tools focus on oracles related to logic correctness.",
+      "context_after": "Specifically, it constructs inputs to fetch a randomly selected row from a relational table. If the database server fails to return the rows, it should contain a bug. Some other works define test oracles based on differential testing. RAGS [ 33] detects correctness bugs in databases through differential testing. It generates and executes queries in multiple databases. Any inconsistency among results indicates at least one database contains bugs. Apollo [12,13] is proposed to detect performance regression bugs by comparing the execution time with the same inputs on different versions of the sa",
+      "section": "6.4 Effectiveness of Exception Detection",
+      "page": 11,
+      "char_offset": 55551,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M26",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 2,6,11,23,24]",
+      "technique": null,
+      "sentence": "In addition to performance testing, many time-series databases developers [ 2,6,11,23,24] also use a large amount of unit tests to ensure their security.",
+      "context_before": "fferent languages. IoT Domain Testing. With the development of IoT devices, time-series database testing has attracted more and more attention. And a large number of test benches for time-series databases have been designed. Schemakeit [ 30] has created a test bench focusing on the writing speed, CPU time, memory usage of time-series databases. Rudoly [ 29] evaluates different time-series databases for smart space. However, these test benches only test the performance of time-series databases. Different from them, Unicorn detects security problems (e.g. memory errors) in time-series databases.",
+      "context_after": "However, unit tests can only cover basic functionalities of time-series databases, and some hidden bugs can not be discovered due to low coverage. Differently, Unicorn tests target databases as a whole system. In other words, Unicorn could detect anomalies triggered by the interactions from different components in one database. More importantly, Unicorn generatesa large amount of time-series queries automatically, which can cover more basic blocks and detect hidden bugs. 9 CONCLUSION In this paper, we present Unicorn to automatically detect anomalies in time-series databases with time-series",
+      "section": "6.4 Effectiveness of Exception Detection",
+      "page": 11,
+      "char_offset": 57282,
+      "cited_reference": {
+        "number": 24,
+        "text": "Manuel Rigger. 2019. Apache IoTDB: Database for Internet of Things. https: //iotdb .apache .org/",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M3",
+        "M5",
+        "M12"
+      ],
+      "extends_technique": [
+        "M6"
+      ],
+      "describes_as_state_of_the_art": [
+        "M3",
+        "M5"
+      ]
+    },
+    "suppressed": [
+      {
+        "mention_id": "M2",
+        "pattern": "extends_technique",
+        "suppressed_because": "the sentence says the tool was taken up as a baseline, which is comparison rather than reuse or extension"
+      }
+    ]
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:16:44Z",
+    "is_model_written": true,
+    "summary": "Unicorn fuzzes time-series databases, whose inputs combine SQL-like queries with time-series-specific schema and data constraints that relational fuzzers do not model. It synthesises inputs from a hybrid of grammar-based generation and dependency handling, and adds exception handling so that a rejected statement does not end a run. Across IoTDB, KairosDB, QuestDB, TDEngine, TimescaleDB and GridDB it covered 21-199% more basic blocks than SQLsmith and 34-693% more than SQLancer, and found 42 bugs.",
+    "narrative": "SQLancer is Unicorn's principal baseline. The authors adapted it to each of the six time-series systems and report coverage and bug counts side by side across the whole evaluation, and they call it a state-of-the-art database fuzzer when introducing that comparison. Their explanation for the gap is about scope rather than quality: SQLancer targets the relational model, so its generated queries lack time-series elements and most are invalid on these systems, and it cannot address KairosDB at all because that system takes JSON over HTTP rather than SQL. They also cite SQLancer's three test oracles as the prior work on oracle design.",
+    "roles": {
+      "M1": "result_comparison",
+      "M2": "baseline",
+      "M3": "state_of_the_art",
+      "M4": "background",
+      "M5": "state_of_the_art",
+      "M6": "incidental",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "result_comparison",
+      "M12": "result_comparison",
+      "M13": "result_comparison",
+      "M14": "result_comparison",
+      "M15": "result_comparison",
+      "M16": "result_comparison",
+      "M17": "result_comparison",
+      "M18": "result_comparison",
+      "M19": "result_comparison",
+      "M20": "result_comparison",
+      "M21": "result_comparison",
+      "M22": "result_comparison",
+      "M23": "incidental",
+      "M24": "background",
+      "M25": "definition",
+      "M26": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer was adapted so it could be run as a baseline, which is comparison rather than reuse: nothing in Unicorn's own implementation is built on SQLancer's code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the work takes a different approach."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M2",
+          "M5",
+          "M9",
+          "M22"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "We also adapted the industrial fuzzers SQLancer [ 25] and SQLsmith [ 31] for comparison.",
+            "section": "1 INTRODUCTION",
+            "page": 2
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "6 EVALUATION In this section, we evaluate the effectiveness of Unicorn in terms of coverage and bug discovery against state-of-the-art database fuzzers—SQLsmith and SQLancer.",
+            "section": "4.2.1 Agent. As Figure 2 shows, time-series databases register a",
+            "page": 7
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "2 Overall Performance Unicorn performs better than SQLsmith and SQLancer both in the number of basic blocks covered and bugs triggered.",
+            "section": "6.2 Overall Performance",
+            "page": 7
+          },
+          {
+            "mention_id": "M22",
+            "sentence": "Furthermore, 2 bugs were also found by SQLsmith, while SQLancer did not detect any bug.",
+            "section": "6.2 Overall Performance",
+            "page": 9
+          }
+        ],
+        "reasoning": "M2 states the authors adapted SQLancer for comparison, M5 names it as one of the fuzzers Unicorn is evaluated against, and M9 and M22 report the outcome in both coverage and bugs found."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M3",
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "•We evaluate Unicorn on 6 popular time-series databases against state-of-the-art fuzzers SQLsmith and SQLancer.",
+            "section": "1 INTRODUCTION",
+            "page": 2
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "6 EVALUATION In this section, we evaluate the effectiveness of Unicorn in terms of coverage and bug discovery against state-of-the-art database fuzzers—SQLsmith and SQLancer.",
+            "section": "4.2.1 Agent. As Figure 2 shows, time-series databases register a",
+            "page": 7
+          }
+        ],
+        "reasoning": "M3 and M5 both introduce SQLancer as one of the state-of-the-art fuzzers the evaluation measures against."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3533767_3534409.json b/_data/papers/paper_doi_10_1145_3533767_3534409.json
new file mode 100644
index 0000000..162ecdb
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3533767_3534409.json
@@ -0,0 +1,729 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:27:01Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3533767.3534409",
+    "title": "Finding bugs in Gremlin-based graph database systems via Randomized differential testing",
+    "authors": [
+      "Yingying Zheng",
+      "Wensheng Dou",
+      "Yicheng Wang",
+      "Zheng Qin",
+      "Leile Tang",
+      "Yu Gao",
+      "Dong Wang",
+      "Wei Wang",
+      "Jun Wei"
+    ],
+    "year": 2022,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3533767.3534409",
+    "arxiv_id": null,
+    "s2_paper_id": "e67a2817089312746d69b38ce9abfdc4b1bc69c3",
+    "url": "https://doi.org/10.1145/3533767.3534409",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3533767.3534409",
+      "retrieved_at": "2026-09-10T15:27:01Z",
+      "chars": 68181,
+      "content_sha256": "sha256:f0ada6841461bdcbe9b9767156fe53e5951f307fb9ef43f484ead8ed4864af50"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 3920
+      },
+      {
+        "number": "2",
+        "title": "PRELIMINARIES",
+        "start": 10356
+      },
+      {
+        "number": "2.1",
+        "title": "GraphModels in GraphDatabaseSystems",
+        "start": 10464
+      },
+      {
+        "number": "2.2",
+        "title": "GraphQueryLanguages",
+        "start": 13059
+      },
+      {
+        "number": "3",
+        "title": "APPROACH",
+        "start": 16492
+      },
+      {
+        "number": "3.1",
+        "title": "GraphDatabaseGeneration",
+        "start": 18332
+      },
+      {
+        "number": "3.2",
+        "title": "Model-BasedQueryGeneration",
+        "start": 23248
+      },
+      {
+        "number": "3.2.1",
+        "title": "TraversalModelinGremlin. Inthissection,weexplainthe",
+        "start": 24817
+      },
+      {
+        "number": "3.2.2",
+        "title": "QueryGeneration. Basedontheabovetraversalmodel,we",
+        "start": 29680
+      },
+      {
+        "number": "3.3",
+        "title": "Differential TestingforGDBs",
+        "start": 34856
+      },
+      {
+        "number": "4",
+        "title": "IMPLEMENTATION",
+        "start": 37662
+      },
+      {
+        "number": "5",
+        "title": "EVALUATION",
+        "start": 39511
+      },
+      {
+        "number": "5.1",
+        "title": "Methodology",
+        "start": 39638
+      },
+      {
+        "number": "5.2",
+        "title": "OverallBug DetectionResults",
+        "start": 43050
+      },
+      {
+        "number": "5.3",
+        "title": "Bug Analysis",
+        "start": 47239
+      },
+      {
+        "number": "5.4",
+        "title": "Coverage",
+        "start": 50569
+      },
+      {
+        "number": "6",
+        "title": "DISCUSSION",
+        "start": 51506
+      },
+      {
+        "number": "6.1",
+        "title": "Generalizing Grandto more GDBs",
+        "start": 51626
+      },
+      {
+        "number": "6.2",
+        "title": "Threatsto Validity",
+        "start": 52514
+      },
+      {
+        "number": "6.3",
+        "title": "Limitations",
+        "start": 53396
+      },
+      {
+        "number": "7",
+        "title": "RELATED WORK",
+        "start": 54264
+      },
+      {
+        "number": "7.1",
+        "title": "TestingofDBMS",
+        "start": 54344
+      },
+      {
+        "number": "7.2",
+        "title": "Performance TestingofGDB",
+        "start": 56205
+      },
+      {
+        "number": "8",
+        "title": "CONCLUSION",
+        "start": 57647
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2021. BenchmarkingTigerGraphUsingtheLikedDataBenchmarkCouncilSocial Network Benchmark. Retrieved June 23, 2021 from https://www.tigergraph.com/ benchmark/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2021.Cypher Query Language. Retrieved June 23, 2021 from https://neo4j.com/ developer/cypher/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2021.HugeGraph. Retrieved August 17, 2021 from https://hugegraph.github.io/ hugegraph-doc/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2021.HugeGraph-Client Quick Start. Retrieved August 5, 2021 from https: //hugegraph.github.io/hugegraph-doc/quickstart/hugegraph-client.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2021.JaCoCo is a free code coverage library for Java. Retrieved June 23, 2022 fromhttps://www.jacoco.org/jacoco/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2021. JanusGraph. Retrieved June23,2021 from https://janusgraph.org",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2021. Neo4j. Retrieved August 5,2021 from https://neo4j.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2021.Neo4j-Gremlin. Retrieved August 5, 2021 from https://github.com/ thinkaurelius/neo4j-gremlin-plugin",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2021.TheNextGenerationMulti-ModelDatabaseSupportingGraphsKey/Value, Documents and Time-Series. Retrieved June 23, 2022 from https://arcadedb.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2021.Open Source, Distributed, Scalable, Lightning Fast. Retrieved June 23, 2021 fromhttps://nebula-graph.io/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2021. OrientDB. Retrieved August 18,2021 from https://orientdb.org",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2021.RDFTripleStoresvs.LabeledPropertyGraphs:What’stheDifference? Retrieved August 5, 2021 from https://neo4j.com/blog/rdf-triple-store-vs-labeledproperty-graph-difference/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "2021.SQLancer. Retrieved August 5, 2021 from https://github.com/sqlancer/ sqlancer",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "2021.SQLsmith. RetrievedAugust5,2021from https://github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "2021.TigerGraph. Retrieved September4, 2021from https://www.tigergraph. com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "2021.TinkerGraph. RetrievedAugust17,2021from https://github.com/tinkerpop/ blueprints/wiki/tinkergraph",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "2021. TinkerPop. Retrieved June23,2021 from https://tinkerpop.apache.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "2021.TinkerPopDocumentation. RetrievedAugust5,2021from https://tinkerpop. apache.org/docs/3.4.10/reference/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "2021.TinkerPop Github. Retrieved August 5, 2021 from https://github.com/ tinkerpop",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "2021.TITAN: Distributed Graph Database. Retrieved June 23, 2021 from http: //titan.thinkaurelius.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "2021.WhatisopenCypher? RetrievedJune23,2021from http://www.opencypher. org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "2022.DB-Engines Ranking of Graph DBMS. Retrieved January 5, 2022 from https://db-engines.com/en/ranking/graph+ dbms",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "2022.ISSTA 22 Artifact for \"Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing\". Retrieved June 21, 2022 from https://doi.org/10.5281/zenodo.6534721",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Ibrahim Abdelaziz, Razen Harbi, Zuhair Khayyat, and Panos Kalnis. 2017. A Survey and Experimental Comparison of Distributed SPARQL Engines for Very LargeRDFData. Proc. VLDBEndow. 10,13(2017), 2049ś2060.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Ibrahim Abdelaziz, Essam Mansour, Mourad Ouzzani, Ashraf Aboulnaga, and Panos Kalnis. 2017. Query Optimizations over Decentralized RDF Graphs. In InternationalConference onData Engineering (ICDE). 139ś142.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Renzo Angles, Peter A. Boncz,Josep Lluís Larriba-Pey,Irini Fundulaki, Thomas Neumann, Orri Erling, Peter Neubauer, Norbert Martínez-Bazan, Venelin Kotsev, andIoanToma.2014. TheLinkedDataBenchmarkCouncil:AGraphandRDF IndustryBenchmarking Effort. ACMSIGMODRecord 43,1 (2014), 27ś31.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Renzo Angles, Arnau Prat-Pérez, David Dominguez-Sal, and Josep Lluís LarribaPey. 2013. Benchmarking Database Systems for Social Network Applications. In ProceedingsofInternationalWorkshoponGraphDataManagementExperiences and Systems(GRADES). 15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Marcelo Arenas, ClaudioGutiérrez, and JuanF.Sequeda. 2021. Queryingin the Age of Graph Databases and Knowledge Graphs. In Proceedings of International Conference onManagementofData (SIGMODE). 2821ś2828.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Shafiul Azam Chowdhury, SoumikMohian, Sidharth Mehra, Siddhant Gawsane, Taylor T. Johnson,andChristoph Csallner.2018. Automatically findingbugsin a commercial cyber-physical system development tool chain with SLforge. In Proceedings of the 40th International Conference on Software Engineering, ICSE 2018, Gothenburg,Sweden, May 27 -June 03,2018. 981ś992.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Christoph Csallner and Yannis Smaragdakis. 2004. JCrasher: an automatic robustness tester for Java. Softw. Pract.Exp. 34,11(2004), 1025ś1050.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "PascalCuoq,BenjaminMonate,AnnePacalet,VirgilePrevosto,JohnRegehr,Boris Yakobowski,andXuejunYang.2012. TestingStaticAnalyzerswithRandomly Generated Programs. In NASA Formal Methods-4th International Symposium (NFM). 120ś125.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Alin Deutsch. 2018. Querying Graph Databases with the GSQL Query Language. InSimpósio BrasileirodeBanco deDados(SBBD). 313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "OrriErling,AlexAverbuch,JosepLluísLarriba-Pey,HassanChafi,AndreyGubichev, Arnau Prat-Pérez, Minh-Duc Pham, and Peter A. Boncz. 2015. The LDBC SocialNetworkBenchmark:InteractiveWorkload.In ProceedingsofACMSIGMOD InternationalConference onManagementofData (SIGMOD). 619ś630.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Orri Erling and Ivan Mikhailov. 2009. RDF Support in the Virtuoso DBMS. In NetworkedKnowledge-NetworkedMedia. 7ś24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Diogo Fernandes and Jorge Bernardino. 2018. Graph Databases Comparison: AllegroGraph,ArangoDB,InfiniteGraph,Neo4J,andOrientDB.In Proceedings ofInternationalConference onDataScience, TechnologyandApplications(DATE). 373ś380.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and Andrés Taylor.2018. Cypher: AnEvolvingQueryLanguage forPropertyGraphs. InProceedings of International Conference on Management of Data (SIGMOD). 1433ś1445.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woon-Hak Kang. 2019. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in DatabaseSystems. Proc. VLDBEndow. 13,1 (2019), 57ś70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "TimotejKapusandCristianCadar.2017. Automatictestingofsymbolicexecution engines via program generation and differential testing. In International Conference onAutomatedSoftwareEngineering (ASE). 590ś600.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O. Laleye, and Sarfraz Khurshid. 2008. Query-Aware Test Generation Using a Relational Constraint Solver. In InternationalConference onAutomatedSoftwareEngineering (ASE). 238ś247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Norman Köster. 2020. An Extensible Graph Query Language for Model-Based Information Retrieval in Intelligent Environments. Ph.D. Dissertation. Bielefeld University, Germany.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Tibor Kovács, Gábor Simon, and Gergely Mezei. 2019. Benchmarking Graph DatabaseBackends-WhatWorksWellwithWikidata? ActaCybern. 24,1(2019), 43ś60.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Matteo Lissandrini, Martin Brugnara, and Yannis Velegrakis. 2018. Beyond Macrobenchmarks:Microbenchmark-BasedGraphDatabaseEvaluation. Proc. VLDBEndow. 12,4 (2018), 390ś403.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Baozhu Liu, Xin Wang, Pengkai Liu, Sizhuo Li, Qiang Fu, and Yunpeng Chai. 2021. UniKG: A Unified Interoperable Knowledge Graph Database System. In ProceedingsofIEEEInternationalConferenceonDataEngineering(ICDE) .2681ś 2684.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Muhammad Numair Mansur, Maria Christakis, and Valentin Wüstholz. 2021. Metamorphic testing of Datalog engines. In ESEC/FSE ’21: 29th ACM Joint European Software Engineering Conference and Symposium on the Foundations of SoftwareEngineering, Athens,Greece, August 23-28, 2021. 639ś650.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software. Digit. Tech. J. 10, 1 (1998), 100ś107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Anil Pacaci, Alice Zhou, Jimmy Lin, and M. Tamer Özsu. 2017. Do We Need SpecializedGraphDatabases?BenchmarkingReal-Time SocialNetworkingApplications. In Proceedings of InternationalWorkshop on GraphData-management Experiences and Systems(GRADES). 12:1ś12:7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Yuxiang Ren, Hao Zhu, Jiawei Zhang, Peng Dai, and Liefeng Bo. 2021. EnsemFDet: An Ensemble Approach to Fraud Detection based on Bipartite Graph. InInternationalConference onData Engineering (ICDE). 2039ś2044.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "ManuelRiggerandZhendongSu.2020. DetectingOptimizationBugsinDatabase EnginesviaNon-OptimizingReferenceEngineConstruction.In Proceedingsof ACM Joint European Software Engineering Conference and Symposium on the FoundationsofSoftwareEngineering (ESEC/FSE). 1140ś1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA, Article211(2020), 30pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "ManuelRiggerandZhendongSu.2020. TestingDatabaseEnginesviaPivoted Query Synthesis. In Proceedings of USENIX Symposium on Operating Systems Designand Implementation(OSDI). 667ś682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 51,
+      "text": "Marko A. Rodriguez. 2015. The Gremlin Graph Traversal Machine and Language (Invited Talk). In Proceedings of the Symposium on Database Programming Languages. 1ś10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Donald S. Slutz. 1998. Massive Stochastic Testing of SQL. In Proceedings of InternationalConference onVeryLarge Data Bases (VLDB). 618ś622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "ThodorisSotiropoulos,StefanosChaliasos,VaggelisAtlidakis,DimitrisMitropoulos,andDiomidisSpinellis.2021. Data-OrientedDifferentialTestingofObjectRelationalMappingSystems.In ProceedingsofIEEE/ACMInternationalConference onSoftwareEngineering (ICSE). 1535ś1547.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Harsh Thakkar, Renzo Angles, Marko Rodriguez, Stephen Mallette, and Jens Lehmann. 2020. Let’s Build Bridges, not Walls: SPARQL Querying of TinkerPop Graph Databases with SPARQL-Gremlin. In Proceedings of IEEE International Conference onSemantic Computing(ICSC). 408ś415.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Ran Wang, Zhengyi Yang, Wenjie Zhang, and Xuemin Lin. 2020. An Empirical Study on Recent Graph Database Systems. In Proceedings of International ConferenceonKnowledgeScience,EngineeringandManagement(KSEM) .328ś340. 313",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 13,
+      "text": "2021.SQLancer. Retrieved August 5, 2021 from https://github.com/sqlancer/ sqlancer",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 48,
+      "text": "ManuelRiggerandZhendongSu.2020. DetectingOptimizationBugsinDatabase EnginesviaNon-OptimizingReferenceEngineConstruction.In Proceedingsof ACM Joint European Software Engineering Conference and Symposium on the FoundationsofSoftwareEngineering (ESEC/FSE). 1140ś1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec"
+      ]
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA, Article211(2020), 30pages.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 50,
+      "text": "ManuelRiggerandZhendongSu.2020. TestingDatabaseEnginesviaPivoted Query Synthesis. In Proceedings of USENIX Symposium on Operating Systems Designand Implementation(OSDI). 667ś682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 13]",
+      "technique": null,
+      "sentence": "For relational database systems that utilize Structured Query Language (SQL) to create, access, and modify data, researchers havedevelopedseveraltools,suchas RAGS[ 52]and SQLancer[ 13], toeffectivelydiscoverlogicbugs.",
+      "context_before": "wefoundinHugeGraph[ 3].Inthistestcase,wefirstcreateagraph schema with a vertex label 𝑣𝐿𝑎𝑏𝑒𝑙and its property 𝑝𝑟𝑜𝑝(Line 1). Wefurthercreatean 𝑖𝑛𝑑𝑒𝑥onproperty 𝑝𝑟𝑜𝑝(Line2).Basedonthis graph schema, we create three vertices 𝑣1,𝑣2and𝑣3with different𝑝𝑟𝑜𝑝of5,1,and3,respectively(Line4-6),andaddtheminto thegraph(Line7).Wecounttheverticeswhoseproperty 𝑝𝑟𝑜𝑝is between 0 and 4, or less than 2 (Line 9). We can see that vertex 𝑣2 and𝑣3satisfy thiscondition, and thisquery should return 2. However, HugeGraph returns an incorrect result 3 because HugeGraph forgets to deduplicateoverlappingvaluesfor ORoperation.",
+      "context_after": "RAGSusesdifferentialtestingfor detectingbugs inrelationaldatabasesystems,while SQLanceroffers three oracles, i.e., Pivoted Query Synthesis (PQS) [ 50], Ternary LogicPartitioning(TLP)[ 49],andNon-OptimizingReferenceEngine Construction (NoREC) [ 48] to find logic bugs. However, no available toolscan be appliedon GDBs to detectlogic bugs. LogicbugsinGDBsaredifficulttodetectautomaticallydueto the following reasons: (1) Unlike relational database systems, there is no standardized way to query graph data. These widely-used GDBsusuallyutilizetheirownquerylanguages.Forexample,Neo4j developsCypher[ 36],a",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 6157,
+      "cited_reference": {
+        "number": 13,
+        "text": "2021.SQLancer. Retrieved August 5, 2021 from https://github.com/sqlancer/ sqlancer",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": ", Pivoted Query Synthesis (PQS) [ 50], Ternary LogicPartitioning(TLP)[ 49],andNon-OptimizingReferenceEngine Construction (NoREC) [ 48] to find logic bugs.",
+      "context_before": "eswhoseproperty 𝑝𝑟𝑜𝑝is between 0 and 4, or less than 2 (Line 9). We can see that vertex 𝑣2 and𝑣3satisfy thiscondition, and thisquery should return 2. However, HugeGraph returns an incorrect result 3 because HugeGraph forgets to deduplicateoverlappingvaluesfor ORoperation. For relational database systems that utilize Structured Query Language (SQL) to create, access, and modify data, researchers havedevelopedseveraltools,suchas RAGS[ 52]and SQLancer[ 13], toeffectivelydiscoverlogicbugs.RAGSusesdifferentialtestingfor detectingbugs inrelationaldatabasesystems,while SQLanceroffers three oracles, i.e.",
+      "context_after": "However, no available toolscan be appliedon GDBs to detectlogic bugs. LogicbugsinGDBsaredifficulttodetectautomaticallydueto the following reasons: (1) Unlike relational database systems, there is no standardized way to query graph data. These widely-used GDBsusuallyutilizetheirownquerylanguages.Forexample,Neo4j developsCypher[ 36],andTigerGraphusesGSQL[ 32].Fortunately, most GDBs, e.g., 66% (23/35) GDBs in the DB-Engine Ranking of Graph DBMSs [ 22], support a common query language, Gremlin [51],whichisdevelopedbyApacheTinkerPop[ 17]andprovides a group of Gremlin APIs to query graph data. (2) G",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 6485,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 13,14,52]",
+      "technique": null,
+      "sentence": "2 Model-BasedQueryGeneration Randomquerygenerationisacommonly-usedapproachfortesting databases [ 13,14,52].",
+      "context_before": "ropGeneration( 𝑃𝑇𝑆𝑒𝑡)do 20𝑝𝑟𝑜𝑝𝑠←∅; 21foreach𝑝𝑇𝑦𝑝𝑒∈𝑃𝑇𝑆𝑒𝑡do 22 𝑝𝑁𝑎𝑚𝑒←𝑝𝑇𝑦𝑝𝑒.𝑛𝑎𝑚𝑒(); 23 𝑝𝑉𝑎𝑙𝑢𝑒←𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑒𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡(𝑝𝑇𝑦𝑝𝑒.𝑡𝑦𝑝𝑒())); 24 𝑝𝑟𝑜𝑝𝑠.𝑎𝑑𝑑(𝑝𝑁𝑎𝑚𝑒,𝑝𝑉𝑎𝑙𝑢𝑒); 25end 26return𝑝𝑟𝑜𝑝𝑠; 27end select one vertex that conform 𝑒𝑑𝑔𝑒’s incoming/outgoing vertex type (Line 15-16). Note that, for any two vertices, we only add one edge withthe same type. Note that, the maximal numbers of vertex types, edge types, propertytypes,vertices,edgesinagraphdatabase,canbespecified by GDBtesters.Inourexperiment,wegenerateatmost10vertex types, 20 edge types, 20 property types, 100 vertices, and 200 edges for agraph database. 3.",
+      "context_after": "However, the effectiveness of GDB testing heavily relies onthequality ofquery generation.Gremlin hasflexibleprogramminginterfaces,whichconsistsofasequenceof(potentiallynested)GremlinAPIcallsactingonagraphtraversalsource 𝑔. AlthoughwecanextractgrammarconstraintsamongGremlinAPIs with automated tools (e.g., JCrasher [ 30]), a completely random Gremlin query generation based on the grammar for Gremlin APIs cancausetwoissues.First,wecaneasilygenerategrammatically correct but meaningless queries, e.g., 𝑔.𝑉().𝑉().𝑉(). Second, from theGremlingrammar,wecannotknowthesemanticsofGremlin APIs and their pa",
+      "section": "3.2 Model-BasedQueryGeneration",
+      "page": 5,
+      "char_offset": 23250,
+      "cited_reference": {
+        "number": 13,
+        "text": "2021.SQLancer. Retrieved August 5, 2021 from https://github.com/sqlancer/ sqlancer",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "providetwometamorphictestingapproaches[ 48, 49], namely Ternary LogicPartitioning (TLP) and Non-optimizing ReferenceEngineConstruction(NoREC),totest DBMS.",
+      "context_before": "reliesonpassingthesameinputtothedifferenttestingobjects andcomparingtheoutput.Adifferenceindicatesapotentialbug. Slutz first presents RAGS [ 52] to find bugs in DBMS by differential testing. APOLLO [ 37] uses differential testing to find performance regressionbugsin DBMS.CYNTHIA[ 53]usesdifferentialtesting fortestingObject-RelationalMappingsystems.SinceGDBsadopt different storage structures and query languages, these approaches cannotbe easily adoptedinto GDB testing. Metamorphic testing of DBMS. Metamorphic testing [ 44] testssoftwaresystemsbymutatingtestcasesviametamorphicrelations.Riggeretal.",
+      "context_after": "TLP[ 49] partitions a query into three sub-queries, and detects bugs by comparing the combination of results of three sub-queries with the result of the original query. NoREC [ 48] compares the execution results of a given optimized query with its non-optimized version, to detect optimization bugs in DBMS. All the above approaches targetrelationaldatabasesystems,andcanpotentiallybeapplicable on GDBs. Other testing of DBMS. SQLSmith [ 14] is a fuzzing tool for testing DBMS. It can randomly generate SQL queries as inputs. SQLanceroffersPivotedQuerySynthesis(PQS)[ 50]approachto findlogicbugsbyrand",
+      "section": "7.1 TestingofDBMS",
+      "page": 11,
+      "char_offset": 55088,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP[ 49] partitions a query into three sub-queries, and detects bugs by comparing the combination of results of three sub-queries with the result of the original query.",
+      "context_before": "bugs in DBMS by differential testing. APOLLO [ 37] uses differential testing to find performance regressionbugsin DBMS.CYNTHIA[ 53]usesdifferentialtesting fortestingObject-RelationalMappingsystems.SinceGDBsadopt different storage structures and query languages, these approaches cannotbe easily adoptedinto GDB testing. Metamorphic testing of DBMS. Metamorphic testing [ 44] testssoftwaresystemsbymutatingtestcasesviametamorphicrelations.Riggeretal.providetwometamorphictestingapproaches[ 48, 49], namely Ternary LogicPartitioning (TLP) and Non-optimizing ReferenceEngineConstruction(NoREC),totest DBMS.",
+      "context_after": "NoREC [ 48] compares the execution results of a given optimized query with its non-optimized version, to detect optimization bugs in DBMS. All the above approaches targetrelationaldatabasesystems,andcanpotentiallybeapplicable on GDBs. Other testing of DBMS. SQLSmith [ 14] is a fuzzing tool for testing DBMS. It can randomly generate SQL queries as inputs. SQLanceroffersPivotedQuerySynthesis(PQS)[ 50]approachto findlogicbugsbyrandomlyselectingapivotrowasoracleandgenerating random queriescontainingthe selectedrowtotest DBMS.ADUSA [ 39] translates SQL query to Alloy specification, generates Alloy",
+      "section": "7.1 TestingofDBMS",
+      "page": 11,
+      "char_offset": 55241,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [ 48] compares the execution results of a given optimized query with its non-optimized version, to detect optimization bugs in DBMS.",
+      "context_before": "ct-RelationalMappingsystems.SinceGDBsadopt different storage structures and query languages, these approaches cannotbe easily adoptedinto GDB testing. Metamorphic testing of DBMS. Metamorphic testing [ 44] testssoftwaresystemsbymutatingtestcasesviametamorphicrelations.Riggeretal.providetwometamorphictestingapproaches[ 48, 49], namely Ternary LogicPartitioning (TLP) and Non-optimizing ReferenceEngineConstruction(NoREC),totest DBMS.TLP[ 49] partitions a query into three sub-queries, and detects bugs by comparing the combination of results of three sub-queries with the result of the original query.",
+      "context_after": "All the above approaches targetrelationaldatabasesystems,andcanpotentiallybeapplicable on GDBs. Other testing of DBMS. SQLSmith [ 14] is a fuzzing tool for testing DBMS. It can randomly generate SQL queries as inputs. SQLanceroffersPivotedQuerySynthesis(PQS)[ 50]approachto findlogicbugsbyrandomlyselectingapivotrowasoracleandgenerating random queriescontainingthe selectedrowtotest DBMS.ADUSA [ 39] translates SQL query to Alloy specification, generates Alloy instance satisfying queryconditions inAlloy specification, and further obtains the expected result from Alloy instance. All theabovetoolsta",
+      "section": "7.1 TestingofDBMS",
+      "page": 11,
+      "char_offset": 55409,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "SQLanceroffersPivotedQuerySynthesis(PQS)[ 50]approachto findlogicbugsbyrandomlyselectingapivotrowasoracleandgenerating random queriescontainingthe selectedrowtotest DBMS.",
+      "context_before": "ning (TLP) and Non-optimizing ReferenceEngineConstruction(NoREC),totest DBMS.TLP[ 49] partitions a query into three sub-queries, and detects bugs by comparing the combination of results of three sub-queries with the result of the original query. NoREC [ 48] compares the execution results of a given optimized query with its non-optimized version, to detect optimization bugs in DBMS. All the above approaches targetrelationaldatabasesystems,andcanpotentiallybeapplicable on GDBs. Other testing of DBMS. SQLSmith [ 14] is a fuzzing tool for testing DBMS. It can randomly generate SQL queries as inputs.",
+      "context_after": "ADUSA [ 39] translates SQL query to Alloy specification, generates Alloy instance satisfying queryconditions inAlloy specification, and further obtains the expected result from Alloy instance. All theabovetoolstargetrelationaldatabasesystems,andcannotbe appliedon GDBs. 7.2 Performance TestingofGDB Performance benchmarks for GDB. Angles et al. launch the Linked Data Benchmark Council (LDBC) [ 26] project to develop benchmarksforgraphand RDFdatamanagementsystems.Inthe project,LDBCsocialnetworkbenchmark(LDBCSNB)[ 33]consists of a synthetic social network dataset and three workloads with complexgrap",
+      "section": "7.1 TestingofDBMS",
+      "page": 11,
+      "char_offset": 55765,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/choeoe/Grand",
+    "markers": [
+      "renamed_sqlancer_package",
+      "sqlancer_source_content_match"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/choeoe/Grand",
+        "source_type": "github_repository",
+        "excerpt": "# Grand",
+        "note": "Repository is named after Grand, the tool this paper says it built, and is about database testing.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/choeoe/Grand/blob/main/src/main/java/org/gdbtesting/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+        "excerpt_is_verbatim": true,
+        "note": "src/main/java/org/gdbtesting/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.gdbtesting (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+        "content_sha256": "sha256:79cab1e9e63a88331ad883d12218247ae29913a268fb5ffc6cba313b8660e826",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:22:16Z",
+    "is_model_written": true,
+    "summary": "Grand finds logic bugs in Gremlin-based graph database systems by constructing semantically equivalent databases across several systems and comparing the results of the same query on each; differing results point to a logic bug. It adds model-based query generation, producing valid queries likely to return non-empty results, and a data mapping approach that unifies result formats across systems. Across six widely used graph systems including Neo4j and HugeGraph it found 21 previously unknown logic bugs, 18 confirmed.",
+    "narrative": "SQLancer's three oracles are the relational precedent Grand carries into graph databases, each described in its related work. The reuse goes further than the text says: Grand's repository holds SQLancer's own source under a renamed package, which is the only evidence of it.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "background",
+      "M4": "definition",
+      "M5": "definition",
+      "M6": "definition",
+      "M7": "definition",
+      "ARTIFACT": "reuse_implementation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "ARTIFACT"
+        ],
+        "quotes": [
+          {
+            "mention_id": "ARTIFACT",
+            "sentence": "Artifact https://github.com/choeoe/Grand carries: renamed_sqlancer_package, sqlancer_source_content_match",
+            "section": "artifact inspection",
+            "page": null
+          }
+        ],
+        "reasoning": "Nothing in the text claims reuse -- SQLancer appears only in related work. The repository carries SQLancer's source under a renamed package, which is what establishes it, and is why the claim cites the artifact rather than a sentence.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Grand applies differential testing across graph systems rather than generalising PQS, NoREC or TLP, but with the codebase reused and no statement of what was changed, the boundary is not settled by what is here."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer's oracles are described as relational precedent; no run against them is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The oracles are described by what they do rather than as the state of the art."
+      }
+    },
+    "disagreements": [
+      "No regex fired on this paper at all, because its text never claims reuse. The artifact is what carries the claim."
+    ],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3551349_3556924.json b/_data/papers/paper_doi_10_1145_3551349_3556924.json
new file mode 100644
index 0000000..17ff348
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3551349_3556924.json
@@ -0,0 +1,684 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:27:00Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3551349.3556924",
+    "title": "Differentially Testing Database Transactions for Fun and Profit",
+    "authors": [
+      "Ziyu Cui",
+      "Wensheng Dou",
+      "Qianwang Dai",
+      "Jiansen Song",
+      "Wei Wang",
+      "Jun Wei",
+      "Dan Ye"
+    ],
+    "year": 2022,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1145/3551349.3556924",
+    "arxiv_id": null,
+    "s2_paper_id": "b3135c616d6267fe1d2b4ca9d27c8792c0acf3e5",
+    "url": "https://doi.org/10.1145/3551349.3556924",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3551349.3556924",
+      "retrieved_at": "2026-09-10T15:27:00Z",
+      "chars": 70159,
+      "content_sha256": "sha256:b7cc595ceb705de3154cf4e1e66c1f99cbf5cd894c82065e60d24247b4746e3b"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 3272
+      },
+      {
+        "number": "2",
+        "title": "PRELIMINARIES",
+        "start": 11336
+      },
+      {
+        "number": "2.1",
+        "title": "Relational DBMSs and SQL",
+        "start": 11352
+      },
+      {
+        "number": "2.2",
+        "title": "Tested DBMSs",
+        "start": 12832
+      },
+      {
+        "number": "3",
+        "title": "MOTIVATING EXAMPLES",
+        "start": 16079
+      },
+      {
+        "number": "4",
+        "title": "APPROACH",
+        "start": 21922
+      },
+      {
+        "number": "4.1",
+        "title": "Transaction Test Case Generation",
+        "start": 23386
+      },
+      {
+        "number": "4.2",
+        "title": "Deterministic Transaction Execution",
+        "start": 27026
+      },
+      {
+        "number": "4.3",
+        "title": "Comparing Transaction Execution Results",
+        "start": 31988
+      },
+      {
+        "number": "5",
+        "title": "EVALUATION",
+        "start": 33891
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Methodology",
+        "start": 34251
+      },
+      {
+        "number": "5.2",
+        "title": "Bug Results",
+        "start": 36594
+      },
+      {
+        "number": "5.3",
+        "title": "Interesting Bugs",
+        "start": 38558
+      },
+      {
+        "number": "5.4",
+        "title": "False Positives",
+        "start": 43763
+      },
+      {
+        "number": "6",
+        "title": "COMPATIBILITY ISSUE STUDY",
+        "start": 45243
+      },
+      {
+        "number": "6.1",
+        "title": "RQ1. Root Cause",
+        "start": 46613
+      },
+      {
+        "number": "6.2",
+        "title": "RQ2. Consequence",
+        "start": 51228
+      },
+      {
+        "number": "6.3",
+        "title": "Lessons Learned",
+        "start": 52149
+      },
+      {
+        "number": "7.1",
+        "title": "Threats to Validity",
+        "start": 53510
+      },
+      {
+        "number": "7.2",
+        "title": "Limitations",
+        "start": 54376
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2022. The ANSI isolation levels. http://www.adp-gmbh.ch/ora/misc/isolation_ level.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2022. CockroachDB. https://www.cockroachlabs.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2022. Data Types in MariaDB. https://mariadb.com/kb/en/data-types.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2022. Data Types in MySQL. https://dev.mysql.com/doc/refman/8.0/en/datatypes.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2022. Data Types in TiDB. https://docs.pingcap.com/tidb/stable/data-typeoverview.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2022. DB-Engines. https://db-engines.com/en/ranking.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2022. DELETE fails to delete record after blocking is released. https://jira.mariadb. org/browse/MDEV-27992.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2022. DELETE fails to delete record after blocking is released. https://bugs.mysql. com/106629.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2022. GitHub. https://github.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2022. InnoDB Transaction Model. https://dev.mysql.com/doc/refman/8.0/en/ innodb-transaction-model.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2022. INSERT fails to return an error after transaction abort. https://jira.mariadb. org/browse/MDEV-27922.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2022. MariaDB. https://mariadb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "2022. MySQL. https://www.mysql.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "2022. PostgreSQL. https://www.postgresql.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "2022. Set Transaction in MariaDB. https://mariadb.com/kb/en/set-transaction.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "2022. SQLancer. https://www.manuelrigger.at/dbms-bugs.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 17,
+      "text": "2022. SQLite. https://www.sqlite.org/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "2022. SQLsmith. https://jepsen.io.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "2022. TiDB Optimistic Transaction Model. https://docs.pingcap.com/tidb/stable/ optimistic-transaction.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "2022. TiDB Pessimistic Transaction Mode. https://docs.pingcap.com/tidb/stable/ pessimistic-transaction.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "2022. TiDB, PingCAP. https://pingcap.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "2022. Transaction Isolation Levels in MySQL. https://dev.mysql.com/doc/refman/ 8.0/en/innodb-transaction-isolation-levels.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "2022. Transactions in TiDB. https://docs.pingcap.com/tidb/stable/transactionoverview.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "2022. Unexpected Deadlock Happened When Two transaction Execute Concurrently. https://bugs.mysql.com/106655.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "2022. Unexpected error when UPDATE a NULL. https://jira.mariadb.org/browse/ MDEV-28140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "2022. Weird SELECT when table has the primary key. https://github.com/pingcap/ tidb/issues/33315.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "2022. Weird statement with IS NULL and with IS NOT NULL. https://bugs.mysql. com/107125.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Atul Adya. 1999. Weak Consistency: A Generalized Theory and Optimistic Implementations for Distributed Transactions. Ph.D. Dissertation. Massachusetts Institute of Technology.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Atul Adya, Barbara Liskov, and Patrick O’Neil. 2000. Generalized Isolation Level Definitions. In Proceedings of International Conference on Data Engineering (ICDE). 67–78.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Peter Bailis, Aaron Davidson, Alan Fekete, Ali Ghodsi, Joseph M. Hellerstein, and Ion Stoica. 2013. Highly Available Transactions: Virtues and Limitations. Proceedings of the VLDB Endowment 7, 3 (2013), 181–192.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Peter Bailis, Alan Fekete, Ali Ghodsi, Joseph M. Hellerstein, and Ion Stoica. 2016. Scalable Atomic Visibility with RAMP Transactions. ACM Transactions on Database Systems 41, 3 (2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Gerg\"o Barany. 2018. Finding Missed Compiler Optimizations by Differential Testing. In Proceedings of International Conference on Compiler Construction. 82– 92.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. 2007. A Genetic Approach for Random Testing of Database Systems. In Proceedings of International Conference on Very Large Data Bases (VLDB). 1243–1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Hal Berenson, Phil Bernstein, Jim Gray, Jim Melton, Elizabeth O’Neil, and Patrick O’Neil. 1995. A Critique of ANSI SQL Isolation Levels. In Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD), Vol. 24. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Philip A. Bernstein and Nathan Goodman. 1983. Multiversion concurrency control—theory and algorithms. ACM Transactions on Database Systems (TODS) 8, 4 (1983), 465–483.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M. Tamer Özsu. 2007. QAGen: Generating Query-Aware Test Databases. In Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD). 341–352. ASE ’22, October 10–14, 2022, Rochester, MI, USA Ziyu Cui, Wensheng Dou, Qianwang Dai, Jiansen Song, Wei Wang, Jun Wei, and Dan Ye",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Ranadeep Biswas and Constantin Enea. 2019. On the Complexity of Checking Transactional Consistency. In Proceedings of ACMSIGPLAN Conference on ObjectOriented Programming Systems, Languages, and Applications (OOPSLA). 165:1– 165:28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Ranadeep Biswas, Diptanshu Kakwani, Jyothi Vedurada, Constantin Enea, and Akash Lal. 2021. MonkeyDB: Effectively Testing Correctness under Weak Isolation Levels. In Proceedings of ACMSIGPLAN Conference on Object-Oriented Programming Systems, Languages, and Applications (OOPSLA). 132:1–132:27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Nicolas Bruno and Surajit Chaudhuri. 2005. Flexible Database Generators. In Proceedings of the VLDB Endowment. 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. 2006. Generating Queries with Cardinality Constraints for DBMS Testing. In IEEE Transactions on Knowledge and Data Engineering, Vol. 18. 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Lucas Brutschy, Dimitar Dimitrov, Peter Müller, and Martin Vechev. 2017. Serializability for Eventual Consistency: Criterion, Analysis, and Applications. In Proceedings of ACMSIGPLAN Symposium on Principles of Programming Languages (POPL), Vol. 52. 458–472.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Andrea Cerone, Giovanni Bernardi, and Alexey Gotsman. 2015. A Framework for Transactional Consistency Models with Atomic Visibility. In Proceedings of International Conference on Concurrency Theory (CONCUR). 58–71.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Donald D. Chamberlin and Raymond F. Boyce. 1974. SEQUEL: A Structured English Query Language. In Proceedings of ACMSIGFIDET Workshop on Data Description, Access and Control. 249–264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Yuting Chen, Ting Su, and Zhendong Su. 2019. Deep Differential Testing of JVM Implementations. In Proceedings of International Conference on Software Engineering (ICSE). 1257–1268.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Yuting Chen, Ting Su, Chengnian Sun, Zhendong Su, and Jianjun Zhao. 2016. Coverage-Directed Differential Testing of JVM Implementations. In Proceedings of ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). 85–99.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Yuting Chen and Zhendong Su. 2015. Guided Differential Testing of Certificate Validation in SSL/TLS Implementations. In Proceedings of Joint Meeting of the European Software Engineering Conference and the ACMSIGSOFT Symposium on the Foundations of Software Engineering (ESEC/FSE). 793–804.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Edgar F. Codd. 1970. A Relational Model of Data for Large Shared Data Banks. Commun. ACM 13, 6 (1970), 377–387.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Yuetang Deng, Phyllis Frankl, and David Chays. 2005. Testing Database Transactions with AGENDA. In Proceedings of International Conference on Software Engineering (ICSE). 78–87.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Yifan Gan, Xueyuan Ren, Drew Ripberger, Spyros Blanas, and Yang Wang. 2020. IsoDiff: Debugging Anomalies Caused by Weak Isolation. Proceedings of the VLDB Endowment (PVLDB) 13, 12 (2020), 2773–2786.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. Proceedings of the VLDB Endowment (VLDB) 13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Timotej Kapus and Cristian Cadar. 2017. Automatic Testing of Symbolic Execution Engines via Program Generation and Differential Testing. In Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE). 590–600.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O Laleye, and Sarfraz Khurshid. 2008. Query-Aware Test Generation Using a Relational Constraint Solver. In Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE). 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Kyle Kingsbury and Peter Alvaro. 2020. Elle: Inferring Isolation Anomalies from Experimental Observations. In Proceedings of the VLDB Endowment, Vol. 14. 268–280.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Hsiang-Tsung Kung and John T Robinson. 1981. On optimistic methods for concurrency control. ACM Transactions on Database Systems (TODS) 6, 2 (1981), 213–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Kia Rahmani, Kartik Nagar, Benjamin Delaware, and Suresh Jagannathan. 2019. CLOTHO: Directed Test Generation for Weakly Consistent Database Systems. InProceedings of ACMSIGPLAN Conference on Object-Oriented Programming Systems, Languages, and Applications (OOPSLA). 117:1–117:28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "David Patrick Reed. 1978. Naming and synchronization in a decentralized computer system. Technical Report.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 59,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. In Proceedings of ACMSIGPLAN Conference on Object-Oriented Programming Systems, Languages, and Applications (OOPSLA). 211:1–211:30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 60,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 61,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In Proceedings of the VLDB Endowment. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Xiaohui Song and Jane W-S Liu. 1990. Performance of multiversion concurrency control algorithms in maintaining temporal consistency. In Proceedings of Annual International Computer Software and Applications Conference. 132–139.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Thodoris Sotiropoulos, Stefanos Chaliasos, Vaggelis Atlidakis, Dimitris Mitropoulos, and Diomidis Spinellis. 2021. Data-Oriented Differential Testing of ObjectRelational Mapping Systems. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). 1535–1547.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Cheng Tan, Changgeng Zhao, Shuai Mu, and Michael Walfish. 2020. Cobra: Making Transactional Key-Value Stores Verifiably Serializable. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 63–80.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and Understanding Bugs in C Compilers. In Proceedings of ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). 283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Xiangyao Yu, Andrew Pavlo, Daniel Sanchez, and Srinivas Devadas. 2016. TicToc: Time Traveling Optimistic Concurrency Control. In Proceedings of International Conference on Management of Data (SIGMOD). 1629–1642.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing. In Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 302–313.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 16,
+      "text": "2022. SQLancer. https://www.manuelrigger.at/dbms-bugs.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 58,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 59,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. In Proceedings of ACMSIGPLAN Conference on Object-Oriented Programming Systems, Languages, and Applications (OOPSLA). 211:1–211:30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 60,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", PQS [ 60], NoREC [ 58], TLP [ 59] and RAGS [61], can detect DBMS bugs that involve these complex features in single SELECT statements.",
+      "context_before": "mainly focus on verifying whether DBMSs provide the transaction isolation guarantees that they claim [ 37,41,42,53, 64]. These approaches utilize simplified database structure (e.g., key−value model in Elle [ 53] and Cobra [ 64]), and analyze the read / write history of one row indexed by a keyto find transaction bugs. However, relational DBMSs usually support many complex features, e.g., data structures (e.g., data types and indexes), and SQL statements (e.g., cross-table queries). Therefore, existing approaches cannot detect transaction bugs that utilize these features. Some approaches, e.g.",
+      "context_after": "However, they cannot detect transaction bugs that involve multiple SQL statements and isolation levels. To effectively detect transaction bugs and compatibility issues in DBMSs, we propose a general transaction testing approach, Differentially Testing Database Transactions (DT2). We leverage differential testing to build a test oracle for transaction implementations in DBMSs. Specially, we first generate a random database, and then generate a group of concurrent transactions that interact with the database. We then execute these transactions on DBMSs under test, and compare the transaction ex",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 6034,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We generate them mainly based on SQLancer [ 16], and slightly revise the approach for our target.",
+      "context_before": "t case is classified as a compatibility issue ( 6○). 4.1 Transaction Test Case Generation A test case consists of a database, a group of concurrent transactions, a submitted order and a tested isolation level. Three steps (i.e., 1○, 2○, and 3○in Figure 2) are required in randomly test case generation. Since SQL statements are the basis for generating test cases, we will describe SQL statement generation and the first two steps of our approach in details as follows. Note that, there have been a lot of approaches about random generation of databases [ 36,39,52] and SQL statements [ 18,40,50,61].",
+      "context_after": "We briefly describe random database and SQL statement generation for completeness. Database Generation. As we focus on transactions, we create initial database using only the simplest table creating statements, i.e., CREATE and INSERT, of which CREATE statement includes CREATE TABLE and CREATE INDEX. We observe that discrepancies can usually be triggered on simple databases, so we build at most maxTable (3 by default) tables containing at most maxColumn (4 by default) for each table. To find the largest common subset of data types covered by DBMSs, we investigate and support all their data type",
+      "section": "4.1 Transaction Test Case Generation",
+      "page": 5,
+      "char_offset": 23933,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 58–60]",
+      "technique": "norec",
+      "sentence": "[ 58–60] has proposed a series of works to find logical bugs by generating single SQL queries.",
+      "context_before": "MS. Sotiropoulos et al. [ 63] applies differential testing on Object-Relational Mapping systems to find ORM-specific bugs. Zheng et al. [ 67] proposes a differential testing approach, Grand, for detecting logic bugs in Gremlin-based graph database systems. Inspired by these works, we utilize differential testing to detect transaction bugs and transaction-related compatibility issues. Database testing. Many approaches that automatically generate SQL queries and databases to test DBMSs have been proposed. SQLsmith [ 18] generates random SQL queries to find DBMS bugs. More recently, Rigger et al.",
+      "context_after": "Bati et al. [33] presents the technique of randomly generating test cases and using the execution feedback obtained from the tested DBMS to generate queries. These existing approaches are designed to find DBMS-specific bugs and cannot detect transactions bugs. Transaction testing and verification. To improve the correctness of transaction implementations, many works verify whether DBMSs violate their claimed transactional consistency and isolation. Brutschy et al. [ 41] proposes an effective serializability criterion to extend the serializability of conflicts to eventually consistency semanti",
+      "section": "7.2 Limitations",
+      "page": 11,
+      "char_offset": 56583,
+      "cited_reference": {
+        "number": 58,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/criszy/DT2",
+    "markers": [
+      "renamed_sqlancer_package",
+      "sqlancer_source_content_match"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/criszy/DT2",
+        "source_type": "github_repository",
+        "excerpt": "# DT2",
+        "note": "Repository is named after DT2, the tool this paper says it built, and is about database testing.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/criszy/DT2/blob/main/src/DT2/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+        "excerpt_is_verbatim": true,
+        "note": "src/DT2/Randomly.java is SQLancer's Randomly.java, with the package renamed to DT2 (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+        "content_sha256": "sha256:8d9b80c0a6b314d469a165162169da03d400d3d493bc550bcdf4c5fec2a8c289",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "uses_infrastructure": [
+        "M2"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:19:46Z",
+    "is_model_written": true,
+    "summary": "DT2 tests transaction implementations by differential testing across DBMSs. It randomly generates a database and a group of concurrent transactions over it, supporting complex features such as varied schemas and cross-table queries, then compares execution results on several systems to find discrepancies. Because concurrency makes results non-deterministic, the authors add a transaction test protocol that forces deterministic execution. Across MySQL, MariaDB and TiDB it found 10 unique transaction bugs and 88 compatibility issues.",
+    "narrative": "DT2 generates its databases and transactions mainly on the basis of SQLancer, revised for the transaction setting, and its artifact carries SQLancer's source under a renamed package. PQS, NoREC and TLP are cited as the approaches confined to single SELECT statements -- the limitation DT2 sets out to pass.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "reuse_component",
+      "M3": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M2"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "We generate them mainly based on SQLancer [ 16], and slightly revise the approach for our target.",
+            "section": "4.1 Transaction Test Case Generation",
+            "page": 5
+          }
+        ],
+        "reasoning": "M2 states the transactions are generated mainly based on SQLancer with slight revision for the target. The artifact carrying SQLancer source under a renamed package agrees.",
+        "reuse_kind": "generator"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3551349_3560431.json b/_data/papers/paper_doi_10_1145_3551349_3560431.json
new file mode 100644
index 0000000..6de1e44
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3551349_3560431.json
@@ -0,0 +1,786 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:07:40Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3551349.3560431",
+    "title": "Griffin : Grammar-Free DBMS Fuzzing",
+    "authors": [
+      "Jingzhou Fu",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Mingzhe Wang",
+      "Yu Jiang"
+    ],
+    "year": 2022,
+    "venue": "International Conference on Automated Software Engineering",
+    "doi": "10.1145/3551349.3560431",
+    "arxiv_id": null,
+    "s2_paper_id": "5b15efecca0f9f42e9d7842a8ec6689d2f45bc6a",
+    "url": "https://doi.org/10.1145/3551349.3560431",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3551349.3560431",
+      "retrieved_at": "2026-09-07T17:07:40Z",
+      "chars": 71801,
+      "content_sha256": "sha256:cac634bd0db530114083b56f3d386972f0523a85fa6210313d8dabc6ccd76223"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Abdullah Alsharif, Gregory M Kapfhammer, and Phil McMinn. 2018. DOMINO: Fast and effective test data generation for relational database schemas. In 2018 ASE ’22, October 10–14, 2022, Rochester, MI, USA Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang IEEE 11th International Conference on Software Testing, Verification and Validation (ICST). IEEE, 12–22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "ANSI. 2022. ANSI Standard. https://www .ansi .org/. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Cornelius Aschermann, Tommaso Frassetto, Thorsten Holz, Patrick Jauernig, Ahmad-Reza Sadeghi, and Daniel Teuchert. 2019. NAUTILUS: Fishing for Deep Bugs with Grammars.. In NDSS.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Cornelius Aschermann, Sergej Schumilo, Tim Blazytko, Robert Gawlik, and Thorsten Holz. 2019. REDQUEEN: Fuzzing with Input-to-State Correspondence. InSymposium on Network and Distributed System Security (NDSS).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "ST Bhosale, Miss Tejaswini Patil, and Miss Pooja Patil. 2015. Sqlite: Light database system. Int. J. Comput. Sci. Mob. Comput 44, 4 (2015), 882–885.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M Tamer Özsu. 2007. QAGen: generating query-aware test databases. In Proceedings of the 2007 ACMSIGMOD international conference on Management of data. 341–352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Marcel Böhme, Van-Thuan Pham, Manh-Dung Nguyen, and Abhik Roychoudhury. 2017. Directed greybox fuzzing. In Proceedings of the 2017 ACMSIGSAC Conference on Computer and Communications Security. 2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. 2017. Coveragebased Greybox Fuzzing As Markov Chain. In IEEE Transactions on Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Peng Chen and Hao Chen. 2018. Angora: Efficient fuzzing by principled search. InIEEE Symposium on Security and Privacy (S&P).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Yaohui Chen, Peng Li, Jun Xu, Shengjian Guo, Rundong Zhou, Yulong Zhang, Tao Wei, and Long Lu. 2020. Savior: Towards bug-driven hybrid testing. In. IEEE, 1580–1596.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. 2020. AFL++: Combining Incremental Steps of Fuzzing Research. In USENIX Workshop on Offensive Technologies (WOOT).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. 2020. AFL++: Combining Incremental Steps of Fuzzing Research. In 14th USENIX Workshop on Offensive Technologies (WOOT 20). USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Shuitao Gan, Chao Zhang, Peng Chen, Bodong Zhao, Xiaojun Qin, Dong Wu, and Zuoning Chen. 2020. {GREYONE}: Data flow sensitive fuzzing. In 29th USENIX Security Symposium (USENIX Security 20). 2577–2594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Github. 2022. Squirrel GitHub. https://github .com/s3team/Squirrel. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Patrice Godefroid, Adam Kiezun, and Michael Y Levin. 2008. Grammar-based whitebox fuzzing. In Proceedings of the 29th ACMSIGPLAN conference on programming language design and implementation. 206–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Rahul Gopinath, Björn Mathis, and Andreas Zeller. 2020. Mining input grammars from dynamic control flow. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 172–183.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Christian Holler, Kim Herzig, and Andreas Zeller. 2012. Fuzzing with code fragments. In 21st USENIX Security Symposium (USENIX Security 12). 445–458.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Matthias Hoschele and Andreas Zeller. 2017. Mining input grammars with AUTOGRAM. In Companion (ICSE-C). IEEE, 31–34.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Solid IT. 2022. DB-Engines Ranking. https://db-engines .com/en/ranking. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Solid IT. 2022. DB-Engines Ranking of Relational DBMS. https://db-engines .com/ en/ranking/relational +dbms. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2020. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB). Tokyo, Japan.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O Laleye, and Sarfraz Khurshid. 2008. Query-aware test generation using a relational constraint solver. In 2008 23rd IEEE/ACM International Conference on Automated Software Engineering. IEEE, 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Caroline Lemieux and Koushik Sen. 2018. Fairfuzz: Targeting rare branches to rapidly increase greybox fuzz testing coverage. In ACM International Conference on Automated Software Engineering (ASE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Jie Liang, Yu Jiang, Yuanliang Chen, Mingzhe Wang, Chijin Zhou, and Jiaguang Sun. 2018. PAFL: Extend Fuzzing Optimizations of Single Mode to Industrial Parallel Mode.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Jie Liang, Yu Jiang, Mingzhe Wang, Xun Jiao, Yuanliang Chen, Houbing Song, and Kim-Kwang Raymond Choo. 2019. Deepfuzzer: Accelerated deep greybox fuzzing. IEEE Transactions on Dependable and Secure Computing (2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "LibFuzzer 2022. LibFuzzer. https://www .llvm .org/docs/LibFuzzer .html. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Chenyang Lyu, Shouling Ji, Chao Zhang, Yuwei Li, Wei-Han Lee, Yu Song, and Raheem Beyah. 2019. MOPT: Optimized Mutation Scheduling for Fuzzers. In USENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Chirag Manghnani. 2022. Metadata in DBMS – Overview and Types. https: //www .thecrazyprogrammer .com/2019/12/metadata-in-dbms .html. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Phil Mcminn, Chris J Wright, and Gregory M Kapfhammer. 2015. The effectiveness of test coverage criteria for relational database schema integrity constraints. ACM Transactions on Software Engineering and Methodology (TOSEM) 25, 1 (2015), 1–49.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Barton P. Miller, Louis Fredriksen, and Bryan So. 1990. An Empirical Study of the Reliability of UNIX Utilities. Commun. ACM 33, 12 (Dec. 1990).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "NVD. 2022. NVD Home. https://nvd .nist .gov/. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 33,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https://github .com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitriy Vyukov. 2012. AddressSanitizer: A fast address sanity checker. USENIX Annual Technical Conference (ATC).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Donald R Slutz. 1998. Massive stochastic testing of SQL. In VLDB, Vol. 98. Citeseer, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "SQLite. 2022. The Advantages Of Flexible Typing. https://www .sqlite .org/ flextypegood .html. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "The Clang Team. 2022. Source-based Code Coverage. https://clang .llvm .org/ docs/SourceBasedCodeCoverage .html. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Dmitry Vyukov. 2015. google/syzkaller: syzkaller is an unsupervised coverageguided kernel fuzzer. https://github .com/google/syzkaller. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In Software Engineering in Practice (ICSE-SEIP). IEEE, 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Wikipedia. 2022. Comparison of relational database management systems. https: //en .wikipedia .org/wiki/Comparison ofrelational database management systems. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Wikipedia. 2022. Database. https://en .wikipedia .org/wiki/Database. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Wikipedia. 2022. Database catalog. https://en .wikipedia .org/wiki/ Database catalog. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Wikipedia. 2022. databases. https://en .wikipedia .org/wiki/Database. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Wikipedia. 2022. Metadata in database management. https://en .wikipedia .org/ wiki/Metadata#Database management. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Wikipedia. 2022. Relational Database. https://en .wikipedia .org/wiki/ Relational database. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Wikipedia. 2022. SQL. https://en .wikipedia .org/wiki/SQL. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Zhengkai Wu, Evan Johnson, Wei Yang, Osbert Bastani, Dawn Song, Jian Peng, and Tao Xie. 2019. REINAM: reinforcement learning for input-grammar inference. InProceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 488–498.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. 2022. Unicorn: detect runtime errors in time-series databases with hybrid input synthesis. In ISSTA ’22: 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, South Korea, July 18 - 22, 2022, Sukyoung Ryu and Yannis Smaragdakis (Eds.). ACM, 251–262. https://doi .org/10 .1145/3533767 .3534364",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Insu Yun, Sangho Lee, Meng Xu, Yeongjin Jang, and Taesoo Kim. 2018. QSYM: A Practical Concolic Execution Engine Tailored for Hybrid Fuzzing. In USENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Michal Zalewski. 2017. american fuzzy lop. http://lcamtuf .coredump .cx/afl/. Accessed: September 3, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Sqirrel, respectively.",
+      "context_before": "n grammar, Griffin summarizes the DBMS’s state into metadata graph, a lightweight data structure which improves mutation correctness in fuzzing. Specifically, it first tracks the metadata of the statements in built-in SQL test cases as they are executed, and constructs the metadata graph to describe the dependencies between metadata and statements iteratively. Based on the graphs, it reshuffles statements and employs metadata-guided substitution to correct semantic errors. We evaluate Griffin on MariaDB, SQLite, PostgreSQL, and DuckDB. Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.",
+      "context_after": "In total, Griffin finds 55 previously unknown bugs with 13 CVEs assigned. KEYWORDS DBMS Fuzzing, Grammar-Free, Vulnerability Detection ACM Reference Format: Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. In 37th IEEE/ACM International Conference on Automated Software Engineering (ASE ’22), October 10–14, 2022, Rochester, MI, USA. ACM, New York, NY, USA, 12 pages. https://doi .org/ 10.1145/3551349 .3560431 ∗Jie Liang and Yu Jiang are the corresponding authors. Request permissions from permissions@acm.org. ASE ’22, October 10–14, 2022,",
+      "section": null,
+      "page": 1,
+      "char_offset": 1625,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 32,33]",
+      "technique": "pqs",
+      "sentence": "l Many popular works are generationbased [ 32,33].",
+      "context_before": "ax and semantics. Specifically, a DBMS will first check the syntactic correctness of a SQL statement. Incorrect statements will be rejected outright. Next, the DBMS will check the semantics of the statement for correctness (e.g., do not use elements that do not exist). Again, statements with errors will be dropped. Therefore, to adequately test the DBMSs, generating syntactically and semantically correct SQL statements is a prerequisite. To generate syntacticallyand semantically-correct statements, a general idea is to model the grammar of SQL as AST (abstract syntax tree) in input generation.",
+      "context_after": "For example, SQLancer [ 32] extracts the basic information about the database such as database objects and files and then utilize AST to generate valid INSERT or SELECT statements. Other works apply mutation-based fuzzing on DBMSs to generate various types of statements and import coverage feedback [ 39,51]. For example, Sqirrel [51] deeply customizes different syntax parsers according to the grammar of the target DBMS. With the parser, it mitigates errors in syntax and semantics of SQL statements mutated from existing ones. Ratel [39] further extends it by building dictionaries for unsupporte",
+      "section": null,
+      "page": 1,
+      "char_offset": 4018,
+      "cited_reference": {
+        "number": 32,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer [ 32] extracts the basic information about the database such as database objects and files and then utilize AST to generate valid INSERT or SELECT statements.",
+      "context_before": "check the syntactic correctness of a SQL statement. Incorrect statements will be rejected outright. Next, the DBMS will check the semantics of the statement for correctness (e.g., do not use elements that do not exist). Again, statements with errors will be dropped. Therefore, to adequately test the DBMSs, generating syntactically and semantically correct SQL statements is a prerequisite. To generate syntacticallyand semantically-correct statements, a general idea is to model the grammar of SQL as AST (abstract syntax tree) in input generation.l Many popular works are generationbased [ 32,33].",
+      "context_after": "Other works apply mutation-based fuzzing on DBMSs to generate various types of statements and import coverage feedback [ 39,51]. For example, Sqirrel [51] deeply customizes different syntax parsers according to the grammar of the target DBMS. With the parser, it mitigates errors in syntax and semantics of SQL statements mutated from existing ones. Ratel [39] further extends it by building dictionaries for unsupported grammar. However, existing efforts heavily rely on the grammar of the target DBMSs to generate valid SQL statements to ensure efficiency. Considering a new DBMS or a different ve",
+      "section": null,
+      "page": 1,
+      "char_offset": 4068,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Sqirrel, respectively.",
+      "context_before": "n reconstructs the metadata graph by substituting metadata nodes to correct semantic errors. Compared to other conventional DBMS fuzzers which require lots of labor costs to generate a large amount of SQL grammar adaptation code, Griffin are more conveniently adapted to various DBMSs because it is free from the dependencies of grammar. To demonstrate the robustness of Griffin, we directly apply Griffin to three well-tested DBMSs MariaDB, SQLite, and PostgreSQL, as well as one newly developed DBMS DuckDB without any specific adaptations .Griffin covers 73.43%-274.70%, 80.47%312.89%, 43.80%-199.",
+      "context_after": "More importantly, Griffin finds 24, 16, 3, and 12 unknown bugs in MariaDB, SQLite, PostgreSQL, and DuckDB, respectively. Among them, 13 bugs are confirmed as CVEs in the National Vulnerability Database. The results show that without grammar, Griffin is still effective in generating valid SQL statements and finding vulnerabilities. In summary, we have the following contributions: (1)We found that existing DBMS fuzzers rely on grammar. The intensive-labor limits their application. (2)We propose a grammar-free DBMS fuzzing that generates valid SQL statements by analyzing and constructing metadat",
+      "section": null,
+      "page": 2,
+      "char_offset": 7928,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Table 2 shows the lines of extra code of SQLsmith, SQLancer, andSqirrel to adapt SQLite, PostgreSQL, MariaDB, and DuckDB.",
+      "context_before": "cific grammar generation model for each DBMS. Mutationbased fuzzers modify current test cases to generate new ones. They typically build a specific SQL parser and AST mutator for each DBMS to support the syntax-correct query mutation. Dependency on grammar limits the adaptability of DBMS fuzzers. The dependency on grammar limits the adaptability of DBMS fuzzers. Specifically, no matter what generation model, SQL parser, or AST mutator, they are all strongly dependent on the DBMS grammar, and adaptation to a new DBMS requires an amount of extra code, which is time-consuming and labor-intensive.",
+      "context_after": "It demonstrates that all three fuzzers require a lot of additional code to adapt to different DBMSs due to the dependency on grammar. For example, SQLancer uses over 8000 lines of code to build the SQL generation model for PostgreSQL. Although SQLsmith requires about 300 lines of code to adapt a new DBMS, it only partially supports the grammar because SQLsmith can only generate SELECT statements for PostgreSQL. Table 2: The lines of code of SQLsmith, SQLancer, and Sqirrelto adapt for SQLite, PostgreSQL, MariaDB, and DuckDB. The numbers of lines of code are calculated from the official reposit",
+      "section": null,
+      "page": 3,
+      "char_offset": 14129,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer uses over 8000 lines of code to build the SQL generation model for PostgreSQL.",
+      "context_before": "limits the adaptability of DBMS fuzzers. The dependency on grammar limits the adaptability of DBMS fuzzers. Specifically, no matter what generation model, SQL parser, or AST mutator, they are all strongly dependent on the DBMS grammar, and adaptation to a new DBMS requires an amount of extra code, which is time-consuming and labor-intensive. Table 2 shows the lines of extra code of SQLsmith, SQLancer, andSqirrel to adapt SQLite, PostgreSQL, MariaDB, and DuckDB. It demonstrates that all three fuzzers require a lot of additional code to adapt to different DBMSs due to the dependency on grammar.",
+      "context_after": "Although SQLsmith requires about 300 lines of code to adapt a new DBMS, it only partially supports the grammar because SQLsmith can only generate SELECT statements for PostgreSQL. Table 2: The lines of code of SQLsmith, SQLancer, and Sqirrelto adapt for SQLite, PostgreSQL, MariaDB, and DuckDB. The numbers of lines of code are calculated from the official repositories of these fuzzers. Note that due to the difficulties in adapting the grammar, SQLsmith does not support MariaDB and Sqirrel does not support DuckDB. LoC SQLsmith SQLancer Sqirrel SQLite 1375 10020 5248 PostgreSQL 1466 8012 7515 Ma",
+      "section": null,
+      "page": 3,
+      "char_offset": 14385,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Table 2: The lines of code of SQLsmith, SQLancer, and Sqirrelto adapt for SQLite, PostgreSQL, MariaDB, and DuckDB.",
+      "context_before": "unt of extra code, which is time-consuming and labor-intensive. Table 2 shows the lines of extra code of SQLsmith, SQLancer, andSqirrel to adapt SQLite, PostgreSQL, MariaDB, and DuckDB. It demonstrates that all three fuzzers require a lot of additional code to adapt to different DBMSs due to the dependency on grammar. For example, SQLancer uses over 8000 lines of code to build the SQL generation model for PostgreSQL. Although SQLsmith requires about 300 lines of code to adapt a new DBMS, it only partially supports the grammar because SQLsmith can only generate SELECT statements for PostgreSQL.",
+      "context_after": "The numbers of lines of code are calculated from the official repositories of these fuzzers. Note that due to the difficulties in adapting the grammar, SQLsmith does not support MariaDB and Sqirrel does not support DuckDB. LoC SQLsmith SQLancer Sqirrel SQLite 1375 10020 5248 PostgreSQL 1466 8012 7515 MariaDB-2085 9050 DuckDB 1325 2597 More importantly, there are 391 different DBMSs on the market, most of which use different grammar. Current state-of-the-art DBMS fuzzers only support the most popular DBMSs, like PostgreSQL and SQLite. If we want to adapt the advanced fuzzers to these different",
+      "section": null,
+      "page": 3,
+      "char_offset": 14666,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "LoC SQLsmith SQLancer Sqirrel SQLite 1375 10020 5248 PostgreSQL 1466 8012 7515 MariaDB-2085 9050 DuckDB 1325 2597 More importantly, there are 391 different DBMSs on the market, most of which use different grammar.",
+      "context_before": "cer uses over 8000 lines of code to build the SQL generation model for PostgreSQL. Although SQLsmith requires about 300 lines of code to adapt a new DBMS, it only partially supports the grammar because SQLsmith can only generate SELECT statements for PostgreSQL. Table 2: The lines of code of SQLsmith, SQLancer, and Sqirrelto adapt for SQLite, PostgreSQL, MariaDB, and DuckDB. The numbers of lines of code are calculated from the official repositories of these fuzzers. Note that due to the difficulties in adapting the grammar, SQLsmith does not support MariaDB and Sqirrel does not support DuckDB.",
+      "context_after": "Current state-of-the-art DBMS fuzzers only support the most popular DBMSs, like PostgreSQL and SQLite. If we want to adapt the advanced fuzzers to these different DBMSs, it will take too long and require too many human resources to be economically feasible. Therefore, to reduce the adaptation cost and apply fuzzing to more DBMSs, a fuzzer, we urgently need grammar-free DBMS fuzzing. Challenges in grammar-free DBMS fuzzing. The main challenge of fuzzing DBMSs with grammar-free fuzzing is to generate SQL statements while maintaining syntax and semantic correctness. Without grammar, it is very d",
+      "section": null,
+      "page": 3,
+      "char_offset": 15004,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To evaluate the effectiveness of Griffin, we choose SQLsmith, SQLancer, and Sqirrel, which are widely used in industry and academia, for performance comparison.",
+      "context_before": "’s the performance on bug detection of Griffin in real-world DBMSs? (2)How about the coverage and bug findings when comparing Griffin with other DBMS fuzzers? (3)What are the contributions of metadata-based mutation to fuzzing coverage? Test DBMSs. To demonstrate the bug founding ability of Griffincross various DBMSs, we directly apply Griffin to three welltested DBMSs MariaDB, SQLite, and PostgreSQL, as well as one newly developed DBMS DuckDB without any specific adaptations. The chosen versions of these DBMSs are SQLite 3.37.0, DuckDB 0.3.2, MariaDB 10.7.1, and PostgreSQL 14.1. DBMS fuzzers.",
+      "context_after": "SQLancer and SQLsmith are generated-based DBMS fuzzers. They can generate SQL queries continuously for fuzzing without initial seeds. Griffin andSqirrel are mutate-based DBMS fuzzers, which need some SQL queries as the initial seeds for fuzzing. We collected open-source test cases from the official regression test suite of each DBMS and extracted SQL statements from them as the initial seeds ofSqirrel andGriffin. Note that due to Sqirrel does not support the grammar of DuckDB and SQLsmith does not support the grammar of MariaDB, we skipped the evaluation of Sqirrel on DuckDB and SQLsmith on M",
+      "section": null,
+      "page": 7,
+      "char_offset": 35454,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer and SQLsmith are generated-based DBMS fuzzers.",
+      "context_before": ")What are the contributions of metadata-based mutation to fuzzing coverage? Test DBMSs. To demonstrate the bug founding ability of Griffincross various DBMSs, we directly apply Griffin to three welltested DBMSs MariaDB, SQLite, and PostgreSQL, as well as one newly developed DBMS DuckDB without any specific adaptations. The chosen versions of these DBMSs are SQLite 3.37.0, DuckDB 0.3.2, MariaDB 10.7.1, and PostgreSQL 14.1. DBMS fuzzers. To evaluate the effectiveness of Griffin, we choose SQLsmith, SQLancer, and Sqirrel, which are widely used in industry and academia, for performance comparison.",
+      "context_after": "They can generate SQL queries continuously for fuzzing without initial seeds. Griffin andSqirrel are mutate-based DBMS fuzzers, which need some SQL queries as the initial seeds for fuzzing. We collected open-source test cases from the official regression test suite of each DBMS and extracted SQL statements from them as the initial seeds ofSqirrel andGriffin. Note that due to Sqirrel does not support the grammar of DuckDB and SQLsmith does not support the grammar of MariaDB, we skipped the evaluation of Sqirrel on DuckDB and SQLsmith on MariaDB. Basic setup. The evaluation was performed on a m",
+      "section": null,
+      "page": 7,
+      "char_offset": 35615,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For SQLancer and SQLsmith, ASE ’22, October 10–14, 2022, Rochester, MI, USA Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang Table 3: Statistics of previously unknown bugs detected by Griffin, 55 confirmed with 13 CVEs assigned.",
+      "context_before": "nitizer deployed can detect various bugs including 4 undefined behaviors, 15 assertion failures, 3 NULL pointer dereferences, 21 segmentation violations, 1 buffer-overflow, 5 use-after-poison, and 6 heap-use-after-free. We have actively reported all the bugs to the corresponding DBMS vendors, all of these bugs have been confirmed and 50 of them have been fixed. In particular, 13 bugs are assigned with unique CVE IDs in the NVD [31]. We also analyze the proof of concepts (PoCs) that trigger these bugs in DBMSs. The result shows that 33 of them cannot be generated by other DBMS fuzzers directly.",
+      "context_after": "Project(fixed/detected) Component Bug Type SQLite (16/16) insert Segmentation Violation (1) btree Segmentation Violation (1), Assertion Failure (2) pager Assertion Failure (2) optimization Assertion Failure (1) trigger Assertion Failure (2) parse Assertion Failure (1) window Segmentation Violation (1) journal Assertion Failure (1) prepare Assertion Failure (1) bad assert Assertion Failure (3) DuckDB (11/12) common Segmentation Violation (1), Use-After-Free (2), Null Pointer Dereference (1) execution Null Pointer Dereference (1) function Undefined Behavior (2) sqlite api Null Pointer Dereferen",
+      "section": null,
+      "page": 7,
+      "char_offset": 37682,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In addition, SQLancer, SQLsmith, and Sqirrel have not found this bug, although they have been fuzzing SQLite for a few years.",
+      "context_before": "of the foreign key constraints, but the bug is still missed by them. Consequently, the PoC we reported that triggered the bug has been added to the regression test case set of SQLite. The reason for detecting the bug by Griffin .The PoC to trigger this bug invokes three tables, three kinds of operations on the table, and two kinds of keys on the table, which contain abundant combination and semantic information. It is difficult to manually list all combinations of operations and make sure the semantics of test cases are correct. Therefore, the unit test case sets of SQLite do not find the bug.",
+      "context_after": "The main reason is the limitation of the SQL grammar rules. SQLsmith can only generate SELECT statements, while SQLancer can not generate DELETE statements, because their generation models are based on the limited SQL grammar. Similarly, Sqirrel can not generate the PoC in the right part of Figure 6 to trigger this bug because it does not support parsing and generating the ON DELETE CASCADE grammar yet. In contrast, Griffin detected the bug in three days when fuzzing with the unit test case set as the initial seeds. With the grammar-free design, Griffin can mutate all kinds of SQL statements o",
+      "section": null,
+      "page": 8,
+      "char_offset": 43091,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLsmith can only generate SELECT statements, while SQLancer can not generate DELETE statements, because their generation models are based on the limited SQL grammar.",
+      "context_before": "e reason for detecting the bug by Griffin .The PoC to trigger this bug invokes three tables, three kinds of operations on the table, and two kinds of keys on the table, which contain abundant combination and semantic information. It is difficult to manually list all combinations of operations and make sure the semantics of test cases are correct. Therefore, the unit test case sets of SQLite do not find the bug. In addition, SQLancer, SQLsmith, and Sqirrel have not found this bug, although they have been fuzzing SQLite for a few years. The main reason is the limitation of the SQL grammar rules.",
+      "context_after": "Similarly, Sqirrel can not generate the PoC in the right part of Figure 6 to trigger this bug because it does not support parsing and generating the ON DELETE CASCADE grammar yet. In contrast, Griffin detected the bug in three days when fuzzing with the unit test case set as the initial seeds. With the grammar-free design, Griffin can mutate all kinds of SQL statements of various DBMSs. Figure 6 shows the detailed process of generating the PoC to trigger this bug. Test Cases 1, Test Case 2, and Test Case 3 are the SQL queries before mutation. Griffin first builds a metadata graph by locating a",
+      "section": null,
+      "page": 8,
+      "char_offset": 43277,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "2 Comparison with Existing Fuzzers To evaluate the performance, we perform fuzzing with Griffin, SQLsmith, SQLancer, and Sqirrel on each test DBMS for 12 hours.",
+      "context_before": "atements of various DBMSs. Figure 6 shows the detailed process of generating the PoC to trigger this bug. Test Cases 1, Test Case 2, and Test Case 3 are the SQL queries before mutation. Griffin first builds a metadata graph by locating all the table and column names in the SQL statement through metadata queries. Then, it reshuffled the SQL statements of original inputs and generated new SQL queries by metadata-guided substitution. For example, the table names in Test Cases 1 and Test Case 3 are all replaced with “ v0”. Finally, Griffin detected the bug that was missed by other DBMS fuzzers. 5.",
+      "context_after": "We use the number of branches and the number of unique crashes as the basic metrics of the experiments. For a fair comparison, Griffin: Grammar-Free DBMS Fuzzing ASE ’22, October 10–14, 2022, Rochester, MI, USA when we finished the fuzzing experiments, we collected the SQL queries generated by each fuzzer and dry-ran the queries to unify the branch coverage. 100101102103104 (a) SQLite5k8k10k12k15k18k20k22k 100101102103104 (b) DuckDB5k10k15k20k25k30k 100101102103104 (c) MariaDB0k20k40k60k80k100k120k 100101102103104 (d) PostgreSQL0k20k40k60k80k100k120k Number of Branches SQLsmith SQLancer GRIFF",
+      "section": null,
+      "page": 8,
+      "char_offset": 44407,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "100101102103104 (a) SQLite5k8k10k12k15k18k20k22k 100101102103104 (b) DuckDB5k10k15k20k25k30k 100101102103104 (c) MariaDB0k20k40k60k80k100k120k 100101102103104 (d) PostgreSQL0k20k40k60k80k100k120k Number of Branches SQLsmith SQLancer GRIFFIN SQUIRREL Figure 7: The growing trend of the number of branches covered by different DBMS fuzzers in 12 hours.",
+      "context_before": "0”. Finally, Griffin detected the bug that was missed by other DBMS fuzzers. 5.2 Comparison with Existing Fuzzers To evaluate the performance, we perform fuzzing with Griffin, SQLsmith, SQLancer, and Sqirrel on each test DBMS for 12 hours. We use the number of branches and the number of unique crashes as the basic metrics of the experiments. For a fair comparison, Griffin: Grammar-Free DBMS Fuzzing ASE ’22, October 10–14, 2022, Rochester, MI, USA when we finished the fuzzing experiments, we collected the SQL queries generated by each fuzzer and dry-ran the queries to unify the branch coverage.",
+      "context_after": "Displayed are the medians and the 99% confidence intervals of 15 runs. Note that Sqirrel does not support DuckDB and SQLsmith does not support MariaDB. Coverage. Figure 7 shows the number of branches covered by each DBMS fuzzer in 12-hour experiments. From the figure, we can see that Griffin outperformed other fuzzers in terms of branch coverage. Specifically, Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches than SQLancer, SQLsmith, and Sqirrel after fuzzing 12 hours, respectively. The main reason for the improvement can be explained by the grammar-free mutation of",
+      "section": null,
+      "page": 9,
+      "char_offset": 44928,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "11% more branches than SQLancer, SQLsmith, and Sqirrel after fuzzing 12 hours, respectively.",
+      "context_before": "L0k20k40k60k80k100k120k Number of Branches SQLsmith SQLancer GRIFFIN SQUIRREL Figure 7: The growing trend of the number of branches covered by different DBMS fuzzers in 12 hours. Displayed are the medians and the 99% confidence intervals of 15 runs. Note that Sqirrel does not support DuckDB and SQLsmith does not support MariaDB. Coverage. Figure 7 shows the number of branches covered by each DBMS fuzzer in 12-hour experiments. From the figure, we can see that Griffin outperformed other fuzzers in terms of branch coverage. Specifically, Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.",
+      "context_after": "The main reason for the improvement can be explained by the grammar-free mutation of Griffin. The two generation-based fuzzers, SQLancer and SQLsmith, can only generate SQL statements based on their predefined models, which cover only a portion of the entire SQL grammar of the DBMS. For example, SQLsmith can only generate SELECT statements for PostgreSQL because of limited predefined AST models. Therefore, SQL queries that conform to the SQL grammar of the DBMS but are not built in the predefined model cannot be generated by SQLsmith and SQLancer. For Sqirrel, the mutated-based DBMS fuzzer, i",
+      "section": null,
+      "page": 9,
+      "char_offset": 45701,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The two generation-based fuzzers, SQLancer and SQLsmith, can only generate SQL statements based on their predefined models, which cover only a portion of the entire SQL grammar of the DBMS.",
+      "context_before": "ed are the medians and the 99% confidence intervals of 15 runs. Note that Sqirrel does not support DuckDB and SQLsmith does not support MariaDB. Coverage. Figure 7 shows the number of branches covered by each DBMS fuzzer in 12-hour experiments. From the figure, we can see that Griffin outperformed other fuzzers in terms of branch coverage. Specifically, Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches than SQLancer, SQLsmith, and Sqirrel after fuzzing 12 hours, respectively. The main reason for the improvement can be explained by the grammar-free mutation of Griffin.",
+      "context_after": "For example, SQLsmith can only generate SELECT statements for PostgreSQL because of limited predefined AST models. Therefore, SQL queries that conform to the SQL grammar of the DBMS but are not built in the predefined model cannot be generated by SQLsmith and SQLancer. For Sqirrel, the mutated-based DBMS fuzzer, it can only apply mutation on the SQL statements which can be parsed by the inner SQL parser. However, due to the parser of Sqirrel only supporting part of the SQL grammar, the mutable SQL queries are limited. In contrast, Griffin can mutate any statement in the input, as long as it c",
+      "section": null,
+      "page": 9,
+      "char_offset": 45887,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Therefore, SQL queries that conform to the SQL grammar of the DBMS but are not built in the predefined model cannot be generated by SQLsmith and SQLancer.",
+      "context_before": "fuzzers in terms of branch coverage. Specifically, Griffin covers 73.43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches than SQLancer, SQLsmith, and Sqirrel after fuzzing 12 hours, respectively. The main reason for the improvement can be explained by the grammar-free mutation of Griffin. The two generation-based fuzzers, SQLancer and SQLsmith, can only generate SQL statements based on their predefined models, which cover only a portion of the entire SQL grammar of the DBMS. For example, SQLsmith can only generate SELECT statements for PostgreSQL because of limited predefined AST models.",
+      "context_after": "For Sqirrel, the mutated-based DBMS fuzzer, it can only apply mutation on the SQL statements which can be parsed by the inner SQL parser. However, due to the parser of Sqirrel only supporting part of the SQL grammar, the mutable SQL queries are limited. In contrast, Griffin can mutate any statement in the input, as long as it can be executed by the target DBMS, in a syntactically and semantically correct way since Griffin is designed to be grammar-free. In other words, Griffin can generate a large number of new inputs with various grammar by mutating existing inputs. Consequently, Griffin can",
+      "section": null,
+      "page": 9,
+      "char_offset": 46192,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, Griffin found 27, 27, and 22 more unique crashes than SQLsmith, SQLancer, and Sqirrel in 12 hours, respectively.",
+      "context_before": "ate any statement in the input, as long as it can be executed by the target DBMS, in a syntactically and semantically correct way since Griffin is designed to be grammar-free. In other words, Griffin can generate a large number of new inputs with various grammar by mutating existing inputs. Consequently, Griffin can achieve higher branch coverage compared to other DBMS fuzzers.Unique crashes. Table 4 shows the number of unique crashes found by each DBMS fuzzer after fuzzing for 12 hours. From the table, we can see that Griffin finds more unique crashes than other state-of-the-art DBMS fuzzers.",
+      "context_after": "Table 4: Number of unique crashes triggered in 12 hours of each DBMS fuzzer on SQLite, DuckDB, MariaDB, and PostgreSQL. Shown are the medians of 15 repeated times. SQLsmith SQLancer Sqirrel Griffin SQLite 0 0 0 1 DuckDB 0 0 - 9 MariaDB-0 5 15 PostgreSQL 0 0 0 2 Based on the improved coverage and mutated abundant kinds of SQL statements, Griffin can trigger more various behaviors of DBMS. As the case study shows, Griffin can generate SQL queries that other DBMS fuzzers can not generate based on limited SQL grammar. In addition, based on the metadata mutation, Griffin can reshuffle new SQL stat",
+      "section": null,
+      "page": 9,
+      "char_offset": 47230,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLsmith SQLancer Sqirrel Griffin SQLite 0 0 0 1 DuckDB 0 0 - 9 MariaDB-0 5 15 PostgreSQL 0 0 0 2 Based on the improved coverage and mutated abundant kinds of SQL statements, Griffin can trigger more various behaviors of DBMS.",
+      "context_before": "Consequently, Griffin can achieve higher branch coverage compared to other DBMS fuzzers.Unique crashes. Table 4 shows the number of unique crashes found by each DBMS fuzzer after fuzzing for 12 hours. From the table, we can see that Griffin finds more unique crashes than other state-of-the-art DBMS fuzzers. Specifically, Griffin found 27, 27, and 22 more unique crashes than SQLsmith, SQLancer, and Sqirrel in 12 hours, respectively. Table 4: Number of unique crashes triggered in 12 hours of each DBMS fuzzer on SQLite, DuckDB, MariaDB, and PostgreSQL. Shown are the medians of 15 repeated times.",
+      "context_after": "As the case study shows, Griffin can generate SQL queries that other DBMS fuzzers can not generate based on limited SQL grammar. In addition, based on the metadata mutation, Griffin can reshuffle new SQL statements and generate new semanticallycorrect test cases with metadata-guided substitution. As a result, Griffin triggered the behaviors that other DBMS fuzzers can not, and found more new crashes than other DBMS fuzzers. 5.3 Effectiveness of Metadata-Based Mutation To understand the contribution of the metadata mutation in Griffin, we implemented Griffinwhich disables metadata mutation. We",
+      "section": null,
+      "page": 9,
+      "char_offset": 47521,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "0000005 1 1 SQLancer SQLite 0.",
+      "context_before": "ck DB check the semantic information to formulate query plans before execution. Therefore, any SQL statement containing semantic errors is considered invalid before execution. Thus, Griffin achieves more improvement on PostgreSQL, MariaDB, and DuckDB than on SQLite. Table 6: P-values and effect sizes of the comparison with other fuzzers. All p-values are statistically significant ( p< 0.05), and all effect sizes are large ( ˆA12>0.71). P-Value Effect Size v.s. Fuzzer DBMS Branch Crash Branch Crash SQLsmith SQLite 0.000003 0.00007 1 0.8667 DuckDB 0.00000001 0.0000003 1 1 PostgreSQL 0.00000001 0.",
+      "context_after": "00000001 0.00007 1 0.8667 DuckDB 0.000003 0.0000003 1 1 MariaDB 0.00000001 0.0000006 1 1 PostgreSQL 0.00000001 0.0000005 1 1 Sqirrel SQLite 0.00000001 0.00007 1 0.8667 MariaDB 0.00000001 0.000003 1 1 PostgreSQL 0.00000001 0.0000005 1 1 6 DISCUSSION This section shows some limitations of Griffin and evaluation, and the plans to address them in future work. Threats to validity of the evaluation. In the evaluation of comparing Griffin with other fuzzers, Griffin achieved the highest coverage branch counts and crash numbers in 12 hours on SQLite, DuckDB, MariaDB, and PostgreSQL, with 15 times repe",
+      "section": null,
+      "page": 10,
+      "char_offset": 51311,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 32] synthesizes two kinds of different SQL queries which are the same in semantics but evaluated by an optimizing and a non-optimizing version of DBMS for execution.",
+      "context_before": "continuously at a high speed based on the predefined AST model. However, it can only generate limited varieties of SQL queries (e.g., only SELECT statements) restricted by the grammar rules of the generation model. Some other works detect logic bugs in DBMS by different testing. RAGS [ 35] detects bugs by comparing the results of different DBMS with the same SQL queries. Due to the difference in the SQL grammar in the two DBMSs, only part of the functionalities can be tested. Apollo [21] generates queries and checks the performance on different versions of target DBMS to find performance bugs.",
+      "context_after": "If the two queries return different result sets, a logic bug in the optimizer of DBMS is detected. These generated-based fuzzers focus on detecting different particular logic bugs in DBMS instead of exploring more state-space of DBMS. They can only generate limited types of SQL statements that conform to the predefined generation model. Griffin is different from these works. It focuses on preserving both the syntactic and semantic correctness of SQL queries without SQL grammar. Unlike prebuilding the generation model based on the SQL grammar, it analyzes the metadata in DBMS and mutates new S",
+      "section": null,
+      "page": 11,
+      "char_offset": 56760,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Compared with SQLancer, SQLsmith, and Sqirrel ,Griffincovered 73.",
+      "context_before": "hema testing approaches preserve the structure of schema and generate corresponding data or queries, the SQL statements generated by Griffin may change the schema, such as CREATE TABLE statements. 8 CONCLUSION In this paper, we propose Griffin, a grammar-free mutation based DBMS fuzzer. It tracks the metadata of the statements in built-in SQL test cases as they are executed, and constructs the metadata graph to describe the dependencies between metadata and statements iteratively. Based on the graphs, it reshuffles statements and employs metadata-guided substitution to correct semantic errors.",
+      "context_after": "43%-274.70%, 80.47%-312.89%, 43.80%-199.11% more branches and found 27, 27, and 22 more bugs in 12 hours on SQLite, DuckDB, MariaDB, and PostgreSQL, respectively. It found 55 previously unknown bugs in total. Among them, 13 bugs are assigned with CVE IDs due to their security influences. ACKNOWLEDGMENTS This research is sponsored in part by NSFC Program (No. 62022046, 92167101, U1911401, 62021002, 62192730), National Key Research and Development Project (No. 2019YFB1706203, No2021QY0604). REFERENCES [1]Abdullah Alsharif, Gregory M Kapfhammer, and Phil McMinn. 2018. DOMINO: Fast and effective t",
+      "section": null,
+      "page": 11,
+      "char_offset": 62177,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M1",
+        "M4",
+        "M16"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:15:22Z",
+    "is_model_written": true,
+    "summary": "Griffin fuzzes DBMSs without depending on a grammar. Grammars make fuzzing efficient because DBMSs validate inputs strictly, but they differ so much between systems that adapting a fuzzer to each is laborious, leaving many systems untested. Griffin instead summarises the DBMS's state into a metadata graph: it tracks the metadata of statements in built-in test cases as they execute, builds a graph of dependencies between metadata and statements, then reshuffles statements and applies metadata-guided substitution to fix semantic errors. It found 55 previously unknown bugs, 13 with CVEs.",
+    "narrative": "SQLancer is Griffin's baseline and its motivating example of the cost it avoids: the paper counts the lines of code SQLancer needs to support each DBMS -- over 8,000 for PostgreSQL -- as the adaptation burden a grammar-free approach removes, and reports finding 27 more bugs than SQLancer in 12 hours.",
+    "roles": {
+      "M1": "result_comparison",
+      "M2": "background",
+      "M3": "definition",
+      "M4": "result_comparison",
+      "M5": "motivation",
+      "M6": "motivation",
+      "M7": "motivation",
+      "M8": "motivation",
+      "M9": "baseline",
+      "M10": "background",
+      "M11": "baseline"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer appears as a baseline and as the example of per-DBMS adaptation cost."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The metadata graph is Griffin's own; no SQLancer technique is generalised."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M1",
+          "M4",
+          "M9"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Sqirrel, respectively.",
+            "section": null,
+            "page": 1
+          },
+          {
+            "mention_id": "M4",
+            "sentence": "11% more branches, and finds 27, 27, and 22 more bugs in 12 hours than SQLancer, SQLsmith, and Sqirrel, respectively.",
+            "section": null,
+            "page": 2
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "To evaluate the effectiveness of Griffin, we choose SQLsmith, SQLancer, and Sqirrel, which are widely used in industry and academia, for performance comparison.",
+            "section": null,
+            "page": 7
+          }
+        ],
+        "reasoning": "SQLancer is one of three fuzzers chosen for comparison, with branch coverage and 27 more bugs in 12 hours reported against it."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M9 calls the baselines widely used in industry and academia rather than the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3552326_3587448.json b/_data/papers/paper_doi_10_1145_3552326_3587448.json
new file mode 100644
index 0000000..071d285
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3552326_3587448.json
@@ -0,0 +1,827 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:12:45Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3552326.3587448",
+    "title": "Fail through the Cracks: Cross-System Interaction Failures in Modern Cloud Systems",
+    "authors": [
+      "Lilia Tang",
+      "Chaitanya Bhandari",
+      "Yongle Zhang",
+      "Anna Karanika",
+      "Shuyang Ji",
+      "Indranil Gupta",
+      "Tianyi Xu"
+    ],
+    "year": 2023,
+    "venue": "European Conference on Computer Systems",
+    "doi": "10.1145/3552326.3587448",
+    "arxiv_id": null,
+    "s2_paper_id": "916557d95f3d95a83d3f3503350c05c42e8c2272",
+    "url": "https://doi.org/10.1145/3552326.3587448",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3552326.3587448",
+      "retrieved_at": "2026-09-07T17:12:45Z",
+      "chars": 104621,
+      "content_sha256": "sha256:f621c947a3630ddc900c482c4f5bd1085a10186ccc19d9d25e49545e4192befe"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 19,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2985
+      },
+      {
+        "number": "2",
+        "title": "Definition",
+        "start": 10712
+      },
+      {
+        "number": "2.1",
+        "title": "CSI Failure Mode",
+        "start": 10725
+      },
+      {
+        "number": "2.2",
+        "title": "Failure Planes",
+        "start": 13682
+      },
+      {
+        "number": "2.3",
+        "title": "Examples",
+        "start": 14724
+      },
+      {
+        "number": "5.1",
+        "title": "Failure Plane",
+        "start": 24046
+      },
+      {
+        "number": "5.2",
+        "title": "Failure Symptoms",
+        "start": 26695
+      },
+      {
+        "number": "6.1",
+        "title": "Data-Plane Discrepancies",
+        "start": 30477
+      },
+      {
+        "number": "6.2",
+        "title": "Management-Plane Discrepancies",
+        "start": 38451
+      },
+      {
+        "number": "6.3",
+        "title": "Control-Plane Discrepancies",
+        "start": 44673
+      },
+      {
+        "number": "8.1",
+        "title": "Methodology",
+        "start": 55452
+      },
+      {
+        "number": "8.2",
+        "title": "Results and Findings",
+        "start": 57805
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "ANSI Compliance. https://spark.apache.org/docs/latest/sql-ref-ansicompliance.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Apache Avro. https://avro.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Apache Flink. https://flink.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Apache Hadoop YARN. https://hadoop.apache.org/docs/current/ hadoop-yarn/hadoop-yarn-site/YARN.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Apache HBase. https://hbase.apache.org/book.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Apache Hive. https://cwiki.apache.org/confluence/display/Hive/ Tutorial.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Apache Kafka. https://kafka.apache.org/documentation/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Apache ORC: The smallest, fastest columnar storage for Hadoop workloads. https://orc.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Apache Parquet. https://parquet.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Apache Spark Codebase. https://github.com/apache/spark.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Apache Spark Website. https://spark.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "ASFJIRA. https://issues.apache.org/jira/secure/Dashboard.jspa.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "AWS Post-Event Summaries. https://aws.amazon.com/ premiumsupport/technology/pes/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Azure status history. https://status.azure.com/en-us/status/history/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Data types (Databricks SQL). https://docs.databricks.com/sql/ language-manual/sql-ref-datatypes.html#data-types-databrickssql.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Dynamic Tables. https://nightlies.apache.org/flink/flink-docsrelease-1.14/docs/dev/table/concepts/dynamic_tables/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Google Cloud Service Health. https://status.cloud.google.com/ summary.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Google was hit with massive outage, including youtube, gmail and google classroom | cnn business. https://www.cnn.com/2020/12/14/ tech/google-youtube-gmail-down/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Google’s apps crash in a worldwide outage. - the new york times. https://www.nytimes.com/2020/12/14/business/google-downworldwide.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Hadoop Common. https://github.com/apache/hadoop/tree/trunk/ hadoop-common-project.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Hadoop Distributed File System. http://hadoop.apache.org/docs/ current/hadoop-project-dist/hadoop-hdfs/HdfsDesign.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Implicit assumptions of the Hadoop FileSystem APIs. https://hadoop.apache.org/docs/stable/hadoop-projectdist/hadoop-common/filesystem/introduction.html#Implicit_ assumptions_of_the_Hadoop_FileSystem_APIs.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Incident affecting Google App Engine. https://status.cloud.google. com/incidents/NuaWbbv8n8V8PMHNR7kT.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Incident affecting Google BigQuery. https://status.cloud.google.com/ incidents/qq7VS3aLtp6Nmgs5Nux4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Incident affecting Google Cloud Infrastructure Components, Google Cloud Support, Google Cloud Console, Google BigQuery, Google Cloud Storage, Google Cloud Networking, Google Kubernetes Engine, Virtual Private Cloud (VPC). https://status.cloud.google.com/ incidents/cFXPsFUnUELR8U2bQeGz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Incident affecting Google Compute Engine, Google Cloud Networking, Access Approval, Google App Engine. https://status.cloud.google. com/incidents/1tX748pbxW2JjTUuTJsx.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Integration with Cloud Infrastructures. https://spark.apache.org/ docs/latest/cloud-integration.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Massive google outage takes millions offline. https: //www.forbes.com/sites/paulmonckton/2020/12/14/massivegoogle-outage-takes-millions-offline/?sh=40f33d060ad1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Protocol buffers guide. https://developers.google.com/protocolbuffers/docs/proto.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Spark SQL Guide. https://spark.apache.org/docs/latest/sqlprogramming-guide.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Mars Climate Orbiter Mishap Investigation Board Phase I Report. https://llis.nasa.gov/llis_lib/pdf/1009464main1_0641-mr.pdf, Nov. 1999.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "CBS: Cloud Bug Study Database. https://ucare.cs.uchicago.edu/ projects/cbs/, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Apache Hive SQL Conformance. https://cwiki.apache.org/ confluence/display/Hive/Apache+Hive+SQL+Conformance, Nov. 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Alfatafta, M., Alkhatib, B., Alqraan, A., and Al-Kiswany, S. Toward a Generic Fault Tolerance Technique for Partial Network Partitioning. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI’20) (Nov. 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Allspaw, J. Fault injection in production: Making the case for resilience testing. Communications of the ACM (CACM) 55, 10 (Oct 2012), 48–52.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Alluxio Docs. The Need for a New Data Orchestration Platform. https://www.alluxio.io/data-orchestration/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Alqraan, A., Takruri, H., Alfatafta, M., and Al-Kiswany, S. An Analysis of Network-Partitioning Failures in Cloud Systems. In Proceedings of the 13th USENIX Symposium on Operating Systems Design and Implementation (OSDI’18) (Oct. 2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Altekar, G., and Stoica, I. Focus Replay Debugging Effort on the Control Plane. In Proceedings of the 6th Workshop on Hot Topics in System Dependability (HotDep’10) (Oct. 2010).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Amann, S., Nguyen, H. A., Nadi, S., Nguyen, T. N., and Mezini, M.A Systematic Evaluation of Static API-Misuse Detectors. IEEE Transactions on Software Engineering 45, 12 (Dec. 2019), 1170–1188.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Attariyan, M., Chow, M., and Flinn, J. X-ray: Automating RootCause Diagnosis of Performance Anomalies in Production Software. InProceedings of the 10th USENIX Conference on Operating Systems Design and Implementation (OSDI’12) (Oct. 2012).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Attariyan, M., and Flinn, J. Automating Configuration Troubleshooting with Dynamic Information Flow Analysis. In Proceedings of the 9th USENIX Conference on Operating Systems Design and Implementation (OSDI’10) (Oct. 2010).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Basiri, A., Behnam, N., de Rooij, R., Hochstein, L., Kosewski, L., Reynolds, J., and Rosenthal, C. Chaos engineering. IEEE Software 33, 3 (May 2016), 35–41.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Behrang, F., Cohen, M. B., and Orso, A. Users Beware: Preference Inconsistencies Ahead. In Proceedings of the 10th Joint Meeting of the European Software Engineering Conference and the ACMSIGSOFT Symposium on the Foundations of Software Engineering (ESEC/FSE’15) (Aug. 2015).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Bogart, C., Kästner, C., Herbsleb, J., and Thung, F. How to Break an API: Cost Negotiation and Community Values in Three Software Ecosystems. In Proceedings of the 24th ACMSIGSOFT International Symposium on Foundations of Software Engineering (FSE’16) (Nov. 2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Bronson, N., Aghayev, A., Charapko, A., and Zhu, T. Metastable Failures in Distributed Systems. In Proceedings of the 18th Workshop on Hot Topics in Operating Systems (HotOS’21) (May 2021). 178 Cross-System Interaction Failures in Modern Cloud Systems EuroSys ’23, May 8–12, 2023, Rome, Italy",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Brooker, M. The Fundamental Mechanism of Scaling. http://brooker. co.za/blog/2021/01/22/cloud-scale.html, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Burns, B., Grant, B., Oppenheimer, D., Brewer, E., and Wilkes, J.Borg, Omega, and Kubernetes. Communications of the ACM 59, 5 (May 2016), 50–57.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Chen, H., Dou, W., Jiang, Y., and Qin, F. Understanding ExceptionRelated Bugs in Large-Scale Cloud Systems. In Proceedings of the 34th IEEE/ACM International Conference on Automated Software Engineering (ASE’19) (Nov. 2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Chen, Q., Wang, T., Legunsen, O., Li, S., and Xu, T. Understanding and Discovering Software Configuration Dependencies in Cloud and Datacenter Systems. In Proceedings of the 2020 ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE’20) (November 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Chou, A., Yang, J., Chelf, B., Hallem, S., and Engler, D. An Empirical Study of Operating Systems Errors. In Proceedings of the Eighteenth ACM Symposium on Operating Systems Principles (SOSP’01) (Oct. 2001).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Cotroneo, D., De Simone, L., Liguori, P., Natella, R., and Bidokhti, N. How Bad Can a Bug Get? An Empirical Analysis of Software Failures in the OpenStack Cloud Computing Platform. InProceedings of the 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE’19) (Aug. 2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Databricks Docs. Databricks architecture overview. https://docs. databricks.com/getting-started/overview.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Engler, D., Chen, D. Y., Hallem, S., Chou, A., and Chelf, B. Bugs as Deviant Behavior: A General Approach to Inferring Errors in Systems Code. In Proceedings of the Eighteenth ACM Symposium on Operating Systems Principles (SOSP’01) (Oct. 2001).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Flink Docs. Checkpointing. https://nightlies.apache. org/flink/flink-docs-release-1.14/docs/dev/datastream/faulttolerance/checkpointing/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Ford, D., Labelle, F., Popovici, F. I., Stokely, M., Truong, V.-A., Barroso, L., Grimes, C., and Quinlan, S. Availability in Globally Distributed Storage Systems. In Proceedings of the 9th USENIX Conference on Operating Systems Design and Implementation (OSDI’10) (Oct. 2010).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Gao, Y., Dou, W., Qin, F., Gao, C., Wang, D., Wei, J., Huang, R., Zhou, L., and Wu, Y. An Empirical Study on Crash Recovery Bugs in Large-Scale Distributed Systems. In Proceedings of the 26th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE’18) (Nov. 2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Gember-Jacobson, A., Wu, W., Li, X., Akella, A., and Mahajan, R.Management Plane Analytics. In Proceedings of the 2015 Internet Measurement Conference (IMC’15) (Oct. 2015).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Ghemawat, S., Gobioff, H., and Leung, S.-T. The Google File System. InProceedings of the 19th ACM Symposium on Operating Systems Principles (SOSP’03) (Oct. 2003).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Google Cloud. What is a hybrid cloud? https://cloud.google.com/ learn/what-is-hybrid-cloud.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Govindan, R., Minei, I., Kallahalla, M., Koley, B., and Vahdat, A.Evolve or Die: High-Availability Design Principles Drawn from Google’s Network Infrastructure. In Proceedings of the 2011 ACMSIGCOMM Conference (SIGCOMM’11) (Aug. 2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Grant, S., Cech, H., and Beschastnikh, I. Inferring and Asserting Distributed System Invariants. In Proceedings of the 40th International Conference on Software Engineering (ICSE’18) (May 2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Gunawi, H. S., Hao, M., Leesatapornwongsa, T., Patana-anake, T., Do, T., Adityatama, J., Eliazar, K. J., Laksono, A., Lukman, J. F., Martin, V., and Satria, A. D. What Bugs Live in the Cloud? A Study of 3000+ Issues in Cloud Systems. In Proceedings of the 5th ACM Symposium on Cloud Computing (SoCC’14) (Nov. 2014).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Gunawi, H. S., Hao, M., Suminto, R. O., Laksono, A., Satria, A. D., Adityatama, J., and Eliazar, K. J. Why Does the Cloud Stop Computing? Lessons from Hundreds of Service Outages. In Proceedings of the 7th ACM Symposium on Cloud Computing (SoCC’16) (Oct. 2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Gunawi, H. S., Suminto, R. O., Sears, R., Golliher, C., Sundararaman, S., Lin, X., Emami, T., Sheng, W., Bidokhti, N., McCaffrey, C., Grider, G., Fields, P. M., Harms, K., Ross, R. B., Jacobson, A., Ricci, R., Webb, K., Alvaro, P., Runesha, H. B., Hao, M., and Li, H. Fail-Slow at Scale: Evidence of Hardware Performance Faults in Large Production Systems. In Proceedings of the 16th USENIX Conferenc",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Gyori, A., Lambeth, B., Shi, A., Legunsen, O., and Marinov, D. NonDex: A Tool for Detecting and Debugging Wrong Assumptions on Java API Specifications. In Proceedings of the 24th ACMSIGSOFT International Symposium on Foundations of Software Engineering (FSE’16) (Nov. 2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "HBase Docs. HBase Cluster Replication. https://hbase.apache.org/ book.html#_cluster_replication.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "HBase Docs. HBase Write Ahead Log. https://hbase.apache.org/ book.html#wal.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "HDFS Docs. Data Replication. https://hadoop.apache.org/docs/ stable/hadoop-project-dist/hadoop-hdfs/HdfsDesign.html#Data_ Replication.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Huang, L., Magnusson, M., Muralikrishna, A. B., Estyak, S., Isaacs, R., Aghayev, A., Zhu, T., and Charapko, A. Metastable Failures in the Wild. In Proceedings of the 16th USENIX Symposium on Operating Systems Design and Implementation (OSDI’22) (July 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Huang, P., Guo, C., Zhou, L., Lorch, J. R., Dang, Y., Chintalapati, M., and Yao, R. Gray Failure: The Achilles’ Heel of Cloud-Scale Systems. In Proceedings of the 16th Workshop on Hot Topics in Operating Systems (HotOS-XVI) (May 2017).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Istio Docs. Architecture. https://istio.io/latest/docs/ops/ deployment/architecture/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Jia, Z., Li, S., Yu, T., Zeng, C., Xu, E., Liu, X., Wang, J., and Liao, X. DepOwl: Detecting Dependency Bugs to Prevent Compatibility Failures. In Proceedings of the 43rd International Conference on Software Engineering (ICSE’21) (May 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Jiang, L., and Su, Z. Osprey: A Practical Type System for Validating Dimensional Unit Correctness of C Programs. In Proceedings of the 28th International Conference on Software Engineering (ICSE’06) (May 2006).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Kafka Docs. Auto Restart. https://kafka.apache.org/documentation/ streams/architecture#streams_architecture_recovery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Kubernetes Docs. Control Plane Components. https://kubernetes. io/docs/concepts/overview/components/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Leesatapornwongsa, T., Lukman, J. F., Lu, S., and Gunawi, H. S. TaxDC: A Taxonomy of Non-Deterministic Concurrency Bugs in Datacenter Distributed Systems. In Proceedings of the 21st International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS’16) (Mar. 2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Legunsen, O., Hassan, W. U., Xu, X., Rosu, G., and Marinov, D. How Good are the Specs? A Study of the Bug-Finding Effectiveness of Existing Java API Specifications. In Proceedings of the 31th IEEE/ACM International Conference on Automated Software Engineering (ASE’16) (2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Li, G., Lu, S., Musuvathi, M., Nath, S., and Padhye, R. Efficient Scalable Thread-Safety-Violation Detection. In Proceedings of the 27th ACM Symposium on Operating Systems Principles (SOSP’19) (2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Liu, H., Lu, S., Musuvathi, M., and Nath, S. What Bugs Cause Production Cloud Incidents? In Proceedings of the 17th Workshop on Hot Topics in Operating Systems (HotOS’19) (Nov. 2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Lou, C., Chen, C., Huang, P., Dang, Y., Qin, S., Yang, X., Li, X., Lin, Q., and Chintalapati, M. RESIN: A Holistic Service for Dealing with Memory Leaks in Production Cloud Infrastructure. In Proceedings 179 EuroSys ’23, May 8–12, 2023, Rome, Italy L. Tang, C. Bhandari, Y. Zhang, A. Karanika, S. Ji, I. Gupta, and T. Xu of the 16th USENIX Symposium on Operating Systems Design and Implementation (OS",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "Lou, C., Huang, P., and Smith, S. Understanding, Detecting and Localizing Partial Failures in Large System Software. In Proceedings of the 17th USENIX Symposium on Networked Systems Design and Implementation (NSDI’20) (Feb. 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Lou, C., Jing, Y., and Huang, P. Demystifying and Checking Silent Semantic Violations in Large Distributed Systems. In Proceedings of the 16th USENIX Symposium on Operating Systems Design and Implementation (OSDI’22) (July 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Lu, L., Arpaci-Dusseau, A. C., Arpaci-Dusseau, R. H., and Lu, S. A Study of Linux File System Evolution. ACM Trans. Storage 10, 1 (Jan. 2014).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Lu, S., Park, S., Seo, E., and Zhou, Y. Learning from Mistakes: A Comprehensive Study on Real World Concurrency Bug Characteristics. SIGARCH Comput. Archit. News 36, 1 (Mar. 2008), 329–339.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "Ma, H., Goel, A., Jeannin, J.-B., Kapritsos, M., Kasikci, B., and Sakallah, K. A. I4: Incremental Inference of Inductive Invariants for Verification of Distributed Protocols. In Proceedings of the 27th ACM Symposium on Operating Systems Principles (SOSP’19) (2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Ma, S., Zhou, F., Bond, M. D., and Wang, Y. Finding HeterogeneousUnsafe Configuration Parameters in Cloud Systems. In Proceedings of the 16th ACM European Conference on Computer Systems (EuroSys’21) (Apr. 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "Maurer, B. Fail at Scale: Reliability in the Face of Rapid Change. Communications of the ACM 58, 11 (Nov. 2015), 44–49.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "Mehta, S., Bhagwan, R., Kumar, R., Ashok, B., Bansal, C., Maddila, C., Bird, C., Asthana, S., and Kumar, A. Rex: Preventing Bugs and Misconfiguration in Large Services using Correlated Change Analysis. InProceedings of the 17th USENIX Symposium on Networked Systems Design and Implementation (NSDI’20) (Feb. 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "Min, C., Kashyap, S., Lee, B., Song, C., and Kim, T. Cross-checking Semantic Correctness: The Case of Finding File System Bugs. In Proceedings of the 25th ACM Symposium on Operating Systems Principles (SOSP’15) (Oct. 2015).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "Nagaraja, K., Oliveira, F., Bianchini, R., Martin, R. P., and Nguyen, T. D. Understanding and Dealing with Operator Mistakes in Internet Services. In Proceedings of the 6th USENIX Conference on Operating Systems Design and Implementation (OSDI’04) (Dec. 2004).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "OpenStack Docs. Logical architecture. https://docs.openstack.org/ install-guide/get-started-logical-architecture.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "Oppenheimer, D., Ganapathi, A., and Patterson, D. A. Why Do Internet Services Fail, and What Can Be Done About It? In Proceedings of the 4th USENIX Symposium on Internet Technologies and Systems (USITS’03) (Mar. 2003).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "Ore, J.-P., Detweiler, C., and Elbaum, S. Phriky-Units: A Lightweight, Annotation-Free Physical Unit Inconsistency Detection Tool. InProceedings of the 26th ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA’17) (2017).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "Pillai, T. S., Chidambaram, V., Alagappan, R., Al-Kiswany, S., Arpaci-Dusseau, A. C., and Arpaci-Dusseau, R. H. All File Systems Are Not Created Equal: On the Complexity of Crafting CrashConsistent Applications. In Proceedings of the 11th USENIX Symposium on Operating Systems Design and Implementation (OSDI’14) (Oct. 2014).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "Rabkin, A., and Katz, R. Static Extraction of Program Configuration Options. In Proceedings of the 33rd International Conference on Software Engineering (ICSE’11) (May 2011).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "Rabkin, A., and Katz, R. How Hadoop Clusters Break. IEEE Software Magazine 30, 4 (July 2013), 88–94.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "Ramachandran, V., Gupta, M., Sethi, M., and Chowdhury, S. R. Determining Configuration Parameter Dependencies via Analysis of Configuration Data from Multi-tiered Enterprise Applications. In Proceedings of the 6th International Conference on Autonomic Computingand Communications (ICAC’09) (June 2009).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 98,
+      "text": "Rigger, M., and Su, Z. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI’20) (Nov. 2020).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 99,
+      "text": "Robillard, M. P., Bodden, E., Kawrykow, D., Mezini, M., and Ratchford, T. Automated API Property Inference Techniques. IEEE Transactions on Software Engineering 39, 5 (May 2013), 613–637.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 100,
+      "text": "Rosu, G., and Chen, F. Certifying Measurement Unit Safety Policy. InProceedings of the 18th IEEE International Conference on Automated Software Engineering (ASE’03) (Oct. 2003).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 101,
+      "text": "Schumilo, S., Aschermann, C., Jemmett, A., Abbasi, A., and Holz, T.Nyx-Net: Network Fuzzing with Incremental Snapshots. In Proceedings of the Seventeenth European Conference on Computer Systems (EuroSys’22) (Apr. 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 102,
+      "text": "Stoica, I., and Shenker, S. From Cloud Computing to Sky Computing. In Proceedings of the 18th Workshop on Hot Topics in Operating Systems (HotOS’21) (May 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 103,
+      "text": "Sun, X., Cheng, R., Chen, J., Ang, E., Legunsen, O., and Xu, T. Testing Configuration Changes in Context to Prevent Production Failures. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI’20) (Nov. 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 104,
+      "text": "Sun, X., Suresh, L., Ganesan, A., Alagappan, R., Gasch, M., Tang, L., and Xu, T. Reasoning about modern datacenter infrastructures using partial histories. In Proceedings of the 18th Workshop on Hot Topics in Operating Systems (HotOS-XVIII) (May 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 105,
+      "text": "Treynor, B., Dahlin, M., Rau, V., and Beyer, B. The Calculus of Service Availability. Communications of the ACM (CACM) 60, 9 (Sept. 2017), 42–47.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 106,
+      "text": "van Renesse, R., Weatherspoon, H., Shen, Z., and Song, W. The Supercloud: Applying Internet Design Principles to Interconnecting Clouds. In IEEE Internet Computing (IEEE Internet Computing’18) (Mar. 2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 107,
+      "text": "Veeraraghavan, K., Meza, J., Michelson, S., Panneerselvam, S., Gyori, A., Chou, D., Margulis, S., Obenshain, D., Padmanabha, S., Shah, A., Song, Y. J., and Xu, T. Maelstrom: Mitigating Datacenterlevel Disasters by Draining Interdependent Traffic Safely and Efficiently. In Proceedings of the 13th USENIX Symposium on Operating Systems Design and Implementation (OSDI’18) (Oct. 2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 108,
+      "text": "Wang, Y., Wen, M., Liu, Y., Wang, Y., Li, Z., Wang, C., Yu, H., Cheung, S.-C., Xu, C., and Zhu, Z. Watchman: Monitoring Dependency Conflicts for Python Library Ecosystem. In Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering (ICSE’20) (July 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 109,
+      "text": "Wen, M., Liu, Y., Wu, R., Xie, X., Cheung, S.-C., and Su, Z. Exposing Library API Misuses via Mutation Analysis. In Proceedings of the 41st International Conference on Software Engineering (ICSE’19) (May 2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 110,
+      "text": "Xia, H., Zhang, Y., Zhou, Y., Chen, X., Wang, Y., Zhang, X., Cui, S., Hong, G., Zhang, X., Yang, M., and Yang, Z. How Android Developers Handle Evolution-Induced API Compatibility Issues: A Large-Scale Study. In Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering (ICSE’20) (July 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 111,
+      "text": "Xia, W., Wen, Y., Foh, C. H., Niyato, D., and Xie, H. A Survey on Software-Defined Networking. IEEE Communications Surveys & Tutorials 17, 1 (June 2014), 27–51.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 112,
+      "text": "Xu, T., Jin, X., Huang, P., Zhou, Y., Lu, S., Jin, L., and Pasupathy, S. Early Detection of Configuration Errors to Reduce Failure Damage. InProceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI’16) (Nov. 2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 113,
+      "text": "Xu, T., Zhang, J., Huang, P., Zheng, J., Sheng, T., Yuan, D., Zhou, Y., and Pasupathy, S. Do Not Blame Users for Misconfigurations. InProceedings of the 24th Symposium on Operating System Principles (SOSP’13) (Nov. 2013). 180 Cross-System Interaction Failures in Modern Cloud Systems EuroSys ’23, May 8–12, 2023, Rome, Italy",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 114,
+      "text": "Xu, T., and Zhou, Y. Systems Approaches to Tackling Configuration Errors: A Survey. ACM Computing Surveys (CSUR) 47, 4 (July 2015).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 115,
+      "text": "Yang, J. Modeling API Traffic to Catch Breaking Changes. https://www.akitasoftware.com/blog-posts/modeling-api-trafficto-catch-breaking-changes, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 116,
+      "text": "YARN Docs. YARN ResourceManager HA. https: //hadoop.apache.org/docs/stable/hadoop-yarn/hadoop-yarnsite/ResourceManagerHA.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 117,
+      "text": "Yin, Z., Ma, X., Zheng, J., Zhou, Y., Bairavasundaram, L. N., and Pasupathy, S. An Empirical Study on Configuration Errors in Commercial and Open Source Systems. In Proceedings of the 23rd ACM Symposium on Operating Systems Principles (SOSP’11) (Oct. 2011).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 118,
+      "text": "Yoo, A., Wang, Y., Sinha, R., Mu, S., and Xu, T. Fail-slow fault tolerance needs programming support. In Proceedings of the 18th Workshop on Hot Topics in Operating Systems (HotOS-XVIII) (May 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 119,
+      "text": "Yoo, S., and Harman, M. Regression Testing Minimisation, Selection and Prioritization: A Survey. Software Testing, Verification, and Reliability 22, 2 (Mar. 2012), 67–120.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 120,
+      "text": "Yuan, D., Luo, Y., Zhuang, X., Rodrigues, G., Zhao, X., Zhang, Y., Jain, P. U., and Stumm, M. Simple Testing Can Prevent Most Critical Failures: An Analysis of Production Failures in Distributed Dataintensive Systems. In Proceedings of the 11th USENIX Conference on Operating Systems Design and Implementation (OSDI’14) (Oct. 2014).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 121,
+      "text": "Yun, I., Min, C., Si, X., Jang, Y., Kim, T., and Naik, M. APISan: Sanitizing API Usages through Semantic Cross-Checking. In Proceedings of the 25th USENIX Security Symposium (USENIX Security ’16) (Aug. 2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 122,
+      "text": "Zaharia, M., Chowdhury, M., Das, T., Dave, A., Ma, J., McCauley, M., Franklin, M. J., Shenker, S., and Stoica, I. Resilient Distributed Datasets: A Fault-Tolerant Abstraction for In-Memory Cluster Computing. In Proceedings of the 9th USENIX Symposium on Networked Systems Design and Implementation (NSDI’12) (Apr. 2012).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 123,
+      "text": "Zamfir, C., Altekar, G., and Stoica:, I. Automating the Debugging of Datacenter Applications with ADDA. In Proceedings of the 43rd Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN’13) (June 2013).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 124,
+      "text": "Zhai, E., Chen, A., Piskac, R., Balakrishnan, M., Tian, B., Song, B., and Zhang, H. Check before You Change: Preventing Correlated Failures in Service Updates. In Proceedings of the 17th USENIX Symposium on Networked Systems Design and Implementation (NSDI’20) (Feb. 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 125,
+      "text": "Zhang, J., Renganarayana, L., Zhang, X., Ge, N., Bala, V., Xu, T., and Zhou, Y. EnCore: Exploiting System Environment and Correlation Information for Misconfiguration Detection. In Proceedings of the 19th International Conference on Architecture Support for Programming Languages and Operating Systems (ASPLOS’14) (Mar. 2014).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 126,
+      "text": "Zhang, Y., Yang, J., Jin, Z., Sethi, U., Rodrigues, K., Lu, S., and Yuan, D. Understanding and Detecting Software Upgrade Failures in Distributed Systems. In Proceedings of the 28th ACM Symposium on Operating Systems Principles (SOSP’21) (2021). A Artifact Appendix A.1 Abstract The artifact can be found at https://github.com/xlab-uiuc/csiaeandhttps://github.com/xlab-uiuc/csi-test-ae. The first rep",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 98,
+      "text": "Rigger, M., and Su, Z. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI’20) (Nov. 2020).",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 94,98]",
+      "technique": "pqs",
+      "sentence": ", [ 1,22,33]), which also exist in traditional systems that follow POSIX or SQL standards [ 94,98], but are magnified by the heterogeneity and composability of cloud APIs.",
+      "context_before": "r encrypted), location (e.g., local or remote), permission for impersonation, etc. Without POSIX-compliance by design [ 22,58], it is natural for cloud storage systems to extend file properties in order to expose custom information. However, custom properties could be error-prone, especially when they require special handling by upstream systems. For example, in a few cases upstream systems have to operate on files stored in local and remote storage differently (FLINK-13758), or be aware of an overloaded field (e.g., Figure 2). Lastly, non-standard data APIs often have implicit semantics (e.g.",
+      "context_after": "Implication. Our results indicate that many data-plane CSI failures, especially those induced by metadata properties, could potentially be prevented by cross-testing data-plane interactions of involved systems collectively. Such cross-testing should take a data-centric view to achieve high coverage over the metadata/data properties listed in Table 5. We explore the feasibility of cross-testing data-plane crosssystem interactions in §8. Finding 5: Complicated data abstractions (e.g., tables) are more error-prone to CSI failures, compared with simple data abstractions. 57% (35/61) of data-plane",
+      "section": "6.1 Data-Plane Discrepancies",
+      "page": 7,
+      "char_offset": 33068,
+      "cited_reference": {
+        "number": 98,
+        "text": "Rigger, M., and Su, Z. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI’20) (Nov. 2020).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "This paper studies cross-system interaction failures in cloud systems, where reliability depends not only on each subsystem but on how independently developed systems interact. Analysing 11 such incidents at Google, Azure and AWS plus 120 failure cases across seven co-deployed open-source systems, the authors focus on discrepancies between interacting systems as the root cause, and demonstrate cross-system testing on the Spark-Hive data plane.",
+    "narrative": "One citation, noting that data-plane discrepancies of the kind studied also exist in traditional systems following POSIX or SQL standards, where SQLancer's work is among the references given.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3582016_3582053.json b/_data/papers/paper_doi_10_1145_3582016_3582053.json
new file mode 100644
index 0000000..25984c5
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3582016_3582053.json
@@ -0,0 +1,437 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:10Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3582016.3582053",
+    "title": "Finding Unstable Code via Compiler-Driven Differential Testing",
+    "authors": [
+      "Shaohua Li",
+      "Zhendong Su"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Architectural Support for Programming Languages and Operating Systems",
+    "doi": "10.1145/3582016.3582053",
+    "arxiv_id": null,
+    "s2_paper_id": "03e0d8d1839045a6fd25080bbfa945b2801a2f3f",
+    "url": "https://doi.org/10.1145/3582016.3582053",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3582016.3582053",
+      "retrieved_at": "2026-09-09T01:30:10Z",
+      "chars": 78329,
+      "content_sha256": "sha256:e503be6e6e73184469a3b17b03067aeff970d50ed7a747ce276fbf87120f57ed"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 14,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 3086
+      },
+      {
+        "number": "2",
+        "title": "ILLUSTRATIVE EXAMPLES",
+        "start": 10491
+      },
+      {
+        "number": "3",
+        "title": "FINDING UNSTABLE CODE",
+        "start": 17031
+      },
+      {
+        "number": "3.1",
+        "title": "Compiler-Driven Differential Testing",
+        "start": 17338
+      },
+      {
+        "number": "3.2",
+        "title": "CompDiff-AFL++",
+        "start": 23596
+      },
+      {
+        "number": "4",
+        "title": "EVALUATION",
+        "start": 27537
+      },
+      {
+        "number": "20.04.3",
+        "title": "LTS.",
+        "start": 28327
+      },
+      {
+        "number": "4.1",
+        "title": "Effectiveness of CompDiff in Benchmark",
+        "start": 29691
+      },
+      {
+        "number": "4.2",
+        "title": "Impact of Reducing #Compiler",
+        "start": 38498
+      },
+      {
+        "number": "4.3",
+        "title": "CompDiff-AFL++",
+        "start": 41841
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Austin Appleby. 2016. MurmurHash3. https://github.com/aappleby/smhasher/ wiki/MurmurHash3. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Cornelius Aschermann, Sergej Schumilo, Tim Blazytko, Robert Gawlik, and Thorsten Holz. 2019. REDQUEEN: Fuzzing with Input-to-State Correspondence.. InNDSS, Vol. 19. 1–15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Marcel Böhme, Valentin JM Manès, and Sang Kil Cha. 2020. Boosting fuzzer efficiency: An information theoretic perspective. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 678–689.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. 2017. Coveragebased greybox fuzzing as markov chain. IEEE Transactions on Software Engineering 45, 5 (2017), 489–506.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Marcel Böhme, Ezekiel O Soremekun, Sudipta Chattopadhyay, Emamurho Ugherughe, and Andreas Zeller. 2017. Where is the bug and how is it fixed? an experiment with practitioners. In Proceedings of the 2017 11th joint meeting on foundations of software engineering. 117–128.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Derek Bruening and Qin Zhao. 2011. Practical memory checking with Dr. Memory. In International Symposium on Code Generation and Optimization (CGO 2011). IEEE, 213–223.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Liming Chen and Algirdas Avizienis. 1978. N-version Programming: A Faulttolerance Approach to Reliability of Software Operation. In Proceedings of the 8th IEEE International Symposium on Fault-Tolerant Computing (FTCS-8), Vol. 1. 3–9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Yuting Chen, Ting Su, Chengnian Sun, Zhendong Su, and Jianjun Zhao. 2016. Coverage-directed differential testing of JVM implementations. In proceedings of the 37th ACMSIGPLAN Conference on Programming Language Design and Implementation. 85–99.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "JTC1/SC22/WG14 The C Standards Committee. 2018. ISO/IEC 9899:2018, Programming languages — C. https://www.open-std.org/jtc1/sc22/wg14/www/docs/ n2310.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "JTC1/SC22/WG21 The C++ Standards Committee. 2020. Standard for Programming Language C++. https://www.open-std.org/jtc1/sc22/wg21/docs/papers/ 2020/n4849.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Mila Dalla Preda, Roberto Giacobazzi, Arun Lakhotia, and Isabella Mastroeni. 2015. Abstract symbolic automata: Mixed syntactic/semantic similarity analysis of executables. In Proceedings of the 42nd Annual ACMSIGPLAN-SIGACT Symposium on Principles of Programming Languages. 329–341.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "CppCheck developers. 2022. A Tool for Static C/C++ Code Analysis. Retrieved October 15, 2022 from http://cppcheck.sourceforge.net/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "GCC developers. 2022. Options That Control Optimization. https://gcc.gnu.org/ onlinedocs/gcc/Optimize-Options.html. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "LLVM developers. 2022. Clangthe Clang C, C++, and Objective-C compiler. https://clang.llvm.org/docs/CommandGuide/clang.html. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "LLVM developers. 2022. UndefinedBehaviorSanitizer. https://clang.llvm.org/ docs/UndefinedBehaviorSanitizer.html. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Sushant Dinesh, Nathan Burow, Dongyan Xu, and Mathias Payer. 2020. Retrowrite: Statically instrumenting cots binaries for fuzzing and sanitization. In. IEEE, 1497–1511.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. 2020. {AFL++}: Combining Incremental Steps of Fuzzing Research. In 14th USENIX Workshop on Offensive Technologies (WOOT 20).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Patrice Godefroid, Daniel Lehmann, and Marina Polishchuk. 2020. Differential regression testing for REST APIs. In Proceedings of the 29th ACMSIGSOFT International Symposium on Software Testing and Analysis. 312–323.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Google. 2022. Honggfuzz. https://honggfuzz.dev/. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Google. 2022. OSS-Fuzz: Continuous Fuzzing for Open Source Software. https: //github.com/google/oss-fuzz. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "The Tcpdump Group. 2022. TCPDUMP. https://github.com/the-tcpdumpgroup/tcpdump. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Petr Hosek and Cristian Cadar. 2015. VARAN the Unbelievable: An Efficient N-version Execution Framework. In Proceedings of the Twentieth International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS’15). 339–353.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Heqing Huang, Peisen Yao, Rongxin Wu, Qingkai Shi, and Charles Zhang. 2020. Pangolin: Incremental hybrid fuzzing with polyhedral path abstraction. In. IEEE, 1613–1627.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Jaewon Hur, Suhwan Song, Dongup Kwon, Eunjin Baek, Jangwoo Kim, and Byoungyoung Lee. 2021. DIFUZZRTL: Differential Fuzz Testing to Find CPU Bugs. In. IEEE, 1286–1303.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Nasif Imtiaz, Brendan Murphy, and Laurie Williams. 2019. How do developers act on static analysis alerts? an empirical study of coverity usage. In 2019 IEEE 30th International Symposium on Software Reliability Engineering (ISSRE). IEEE, 323–333.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "lcamtuf. 2014. Fuzzing random programs without execve(). https://lcamtuf. blogspot.com/2014/10/fuzzing-binaries-without-execve.html. Accessed: March 7, 2022. 250 ASPLOS ’23, March 25–29, 2023, Vancouver, BC, Canada Shaohua Li and Zhendong Su",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler Validation via Equivalence modulo Inputs. In Proceedings of the 35th ACMSIGPLAN Conference on Programming Language Design and Implementation (Edinburgh, United Kingdom) (PLDI ’14). 216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Stephan Lipp, Sebastian Banescu, and Alexander Pretschner. 2022. An Empirical Study on the Effectiveness of Static C Code Analyzers for Vulnerability Detection. InProceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA’22), July 18–22, 2022, Virtual, South Korea.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Chenyang Lyu, Shouling Ji, Chao Zhang, Yuwei Li, Wei-Han Lee, Yu Song, and Raheem Beyah. 2019. {MOPT}: Optimized mutation scheduling for fuzzers. In 28th USENIX Security Symposium (USENIX Security 19). 1949–1966.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Michaël Marcozzi, Qiyi Tang, Alastair F Donaldson, and Cristian Cadar. 2019. Compiler fuzzing: How much does it matter? Proceedings of the ACM on Programming Languages 3, OOPSLA (2019), 1–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Meta. 2022. A Tool to Detect Bugs in Java and C/C++/Objective-c Code. Retrieved October 15, 2022 from https://fbinfer.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Microsoft. 2022. One Fuzz: A self-hosted Fuzzing-As-A-Service platform. https: //github.com/microsoft/onefuzz. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Nicholas Nethercote and Julian Seward. 2007. Valgrind: A Framework for Heavyweight Dynamic Binary Instrumentation (PLDI ’07). 89–100.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Shirin Nilizadeh, Yannic Noller, and Corina S Pasareanu. 2019. DifFuzz: differential fuzzing for side-channel analysis. In 2019 IEEE/ACM 41st International Conference on Software Engineering (ICSE). IEEE, 176–187.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "NIST. 2017. Juliet Test Suite for C/C++ 1.3. https://samate.nist.gov/SARD/testsuites/112.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Spencer Pearson, José Campos, René Just, Gordon Fraser, Rui Abreu, Michael D Ernst, Deric Pang, and Benjamin Keller. 2017. Evaluating and improving fault localization. In neering (ICSE). IEEE, 609–620.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Theofilos Petsios, Adrian Tang, Salvatore Stolfo, Angelos D Keromytis, and Suman Jana. 2017. Nezha: Efficient domain-independent differential testing. In. IEEE, 615–632.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Xiaolei Ren, Michael Ho, Jiang Ming, Yu Lei, and Li Li. 2021. Unleashing the hidden power of compiler optimization on binary code difference: An empirical study. In Proceedings of the 42nd ACMSIGPLAN International Conference on Programming Language Design and Implementation. 142–157.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference andSymposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 40,
+      "text": "Coverity Scan. 2022. Coverity Scan: Find and Fix Defects in Your Java, C/C++, C#, JavaScript, Ruby, or Python Open Source Project for Free. Retrieved October 15, 2022 from https://scan.coverity.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Mozilla Security. 2021. Fuzzdata. https://github.com/MozillaSecurity/fuzzdata. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitriy Vyukov. 2012. AddressSanitizer: A Fast Address Sanity Checker. In 2012 USENIX Annual Technical Conference (USENIX ATC 12). 309–318.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Evgeniy Stepanov and Konstantin Serebryany. 2015. MemorySanitizer: fast detector of uninitialized memory use in C++. In 2015 IEEE/ACM International Symposium on Code Generation and Optimization (CGO). IEEE, 46–55.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Nick Stephens, John Grosen, Christopher Salls, Andrew Dutcher, Ruoyu Wang, Jacopo Corbetta, Yan Shoshitaishvili, Christopher Kruegel, and Giovanni Vigna. 2016. Driller: Augmenting fuzzing through selective symbolic execution.. In NDSS, Vol. 16. 1–16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "The LLVM team. 2022. Honggfuzz. https://llvm.org/docs/LibFuzzer.html. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Kaushik Veeraraghavan, Peter M Chen, Jason Flinn, and Satish Narayanasamy. 2011. Detecting and surviving data races using complementary schedules. In Proceedings of the twenty-third ACM symposium on operating systems principles. 369–384.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Xi Wang, Haogang Chen, Alvin Cheung, Zhihao Jia, Nickolai Zeldovich, and M Frans Kaashoek. 2012. Undefined behavior: what happened to my code?. In Proceedings of the Asia-Pacific Workshop on Systems. 1–7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Xi Wang, Nickolai Zeldovich, M Frans Kaashoek, and Armando Solar-Lezama. 2013. Towards optimization-safe systems: Analyzing the impact of undefined behavior. In Proceedings of the Twenty-Fourth ACM Symposium on Operating Systems Principles. 260–275.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "W Eric Wong, Ruizhi Gao, Yihao Li, Rui Abreu, and Franz Wotawa. 2016. A survey on software fault localization. IEEE Transactions on Software Engineering 42, 8 (2016), 707–740.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Michal Zalewski. 2014. American fuzzy lop. https://lcamtuf.coredump.cx/afl/. Accessed: March 7, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Qian Zhang, Jiyuan Wang, and Miryung Kim. 2021. Heterofuzz: Fuzz testing to detect platform dependent divergence for heterogeneous applications. In Proceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 242–254. Received 2022-10-20; accepted 2023-01-19 251",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 39,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference andSymposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 39]",
+      "technique": "norec",
+      "sentence": "Researchers have leveraged differential testing to find semantic bugs across many types of programs, such as database management systems [ 39], Java Virtual Machine (JVM) implementations [ 8], REST APIs [ 18], and compilers [ 27].",
+      "context_before": "mplicity and effectiveness, coverage-guided greybox fuzzers such as AFL [ 50], LibFuzzer [ 45], and honggfuzz [ 19] are widely deployed and have been the fundamental tools for many other advanced fuzzers. To boost fuzzing performance, a lot of research efforts have been put into improving various aspects of fuzzing, such as seed scheduling [ 3,4], mutation strategies [ 2,29], and path explorations [ 23,44]. Our proposed CompDiff could serve as a plugand-play oracle for all these fuzzers to extend their capability from discovering security defects to finding semantic bugs. Differential testing.",
+      "context_after": "Incorporating different testing into fuzzing engines is also gaining more and more interest for discovering functional bugs. For example, HeteroFuzz [ 51] detects platform-dependent divergence for heterogeneous applications running on both CPU and FPGA. DifFuzz [34] discovers side-channel leakages by analyzing two executions on the same binary. DIFUZZRTL [ 24] finds CPU bugs by differentially comparing multiple CPU RTLs. It develops a novel register coverage for higher fuzzing efficiency and efficacy.Each of these differential fuzzing efforts focuses on one specific domain. In contrast, CompD",
+      "section": "4.3 CompDiff-AFL++",
+      "page": 12,
+      "char_offset": 62034,
+      "cited_reference": {
+        "number": 39,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference andSymposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "CompDiff finds unstable code -- code whose run-time semantics vary because of undefined behaviour, which compilers exploit by assuming it never happens. Sanitizers cover frequently occurring undefined behaviours, but detecting the rest remains hard. CompDiff rests on the observation that different compilers may produce semantically inconsistent binaries from unstable code, so it compares the outputs of differently compiled binaries on the same input. Integrated into AFL++ it found 78 new bugs across 23 open-source projects.",
+    "narrative": "SQLancer appears once, as a bibliography entry for non-optimizing reference engine construction in a discussion of differential approaches.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3588909.json b/_data/papers/paper_doi_10_1145_3588909.json
new file mode 100644
index 0000000..fb57ab0
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3588909.json
@@ -0,0 +1,940 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:07:15Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3588909",
+    "title": "Detecting Logic Bugs of Join Optimizations in DBMS",
+    "authors": [
+      "Xiu Tang",
+      "Sai Wu",
+      "Dongxiang Zhang",
+      "Fei Li",
+      "Gang Chen"
+    ],
+    "year": 2023,
+    "venue": "Proc. ACM Manag. Data",
+    "doi": "10.1145/3588909",
+    "arxiv_id": "2602.21955",
+    "s2_paper_id": "728c97e05665b2e966a1b17abc5d6d4eaf7e136f",
+    "url": "https://doi.org/10.1145/3588909",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2602.21955",
+      "retrieved_at": "2026-09-07T17:07:15Z",
+      "chars": 77776,
+      "content_sha256": "sha256:54b240932cf1e5d835aa37d78bf011b2a8761913d24420adb6eb762920e28c72"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 14,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2083
+      },
+      {
+        "number": "2",
+        "title": "Overview",
+        "start": 8625
+      },
+      {
+        "number": "2.1",
+        "title": "Problem Definition",
+        "start": 9202
+      },
+      {
+        "number": "2.2",
+        "title": "Scheme Overview",
+        "start": 10461
+      },
+      {
+        "number": "3.1",
+        "title": "Schema Normalization",
+        "start": 15983
+      },
+      {
+        "number": "3.2",
+        "title": "Noise Injection",
+        "start": 19639
+      },
+      {
+        "number": "3.3",
+        "title": "Join Query Generator",
+        "start": 26737
+      },
+      {
+        "number": "3.4",
+        "title": "Ground-truth Result Generation",
+        "start": 29753
+      },
+      {
+        "number": "5.1",
+        "title": "Overview and Showcase of Bug Reports",
+        "start": 43565
+      },
+      {
+        "number": "5.1.1",
+        "title": "Bugs in MySQL.A large portion of bugs in MySQL involves",
+        "start": 45325
+      },
+      {
+        "number": "5.1.2",
+        "title": "Bugs in MariaDB.Different from MySQL, many bugs were",
+        "start": 47807
+      },
+      {
+        "number": "5.1.3",
+        "title": "Bugs in TiDB.Merge join and join index are the main causes",
+        "start": 51480
+      },
+      {
+        "number": "5.1.4",
+        "title": "Bugs in X-DB.In our X-DB, most bugs are produced from",
+        "start": 55236
+      },
+      {
+        "number": "5.2",
+        "title": "Comparison with Existing Tools",
+        "start": 57346
+      },
+      {
+        "number": "5.3",
+        "title": "Ablation Studies",
+        "start": 60656
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Günes Aluç, Olaf Hartig, M. Tamer Özsu, and Khuzaima Daudjee. 2014. Diversified Stress Testing of RDF Data Management Systems. InISWC. Springer, 197–212.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Güneş Aluç, Olaf Hartig, M Tamer Özsu, and Khuzaima Daudjee. 2014. Diversified stress testing of RDF data management systems. InInternational Semantic Web Conference. 197–212.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Akhil Arora, Sakshi Sinha, Piyush Kumar, and Arnab Bhattacharya. 2018. HDIndex: Pushing the Scalability-Accuracy Boundary for Approximate kNN Search in High-Dimensional Spaces.Proc. VLDB Endow.11, 8 (2018), 906–919.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Guillaume Bagan, Angela Bonifati, Radu Ciucanu, George H. L. Fletcher, Aurélien Lemay, and Nicky Advokaat. 2017. gMark: Schema-Driven Generation of Graphs and Queries. InICDE. 63–64.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. 2007. A genetic approach for random testing of database systems. InVLDB. ACM, 1243–1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M. Tamer Özsu. 2007. QAGen: generating query-aware test databases. InSIGMOD. ACM, 341–352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Joachim Biskup. 1995. Achievements of Relational Database Schema Design Theory Revisited. InSemantics in Databases. Springer, 29–54.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Joachim Biskup, Umeshwar Dayal, and Philip A. Bernstein. 1979. Synthesizing Independent Database Schemas. InSIGMOD, Philip A. Bernstein (Ed.). ACM, 143–151.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Christian Bizer and Andreas Schultz. 2009. The berlin sparql benchmark.IJSWIS 5, 2 (2009), 1–24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Daniel Blum and Sara Cohen. 2011. Grr: generating random RDF. InExtended Semantic Web Conference. Springer, 16–30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Manuel Bodirsky. 2015. Graph homomorphisms and universal algebra course notes.TU Dresden(2015).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Robert Brummayer and Armin Biere. 2009. Fuzzing and delta-debugging SMT solvers. InProceedings of the 7th International Workshop on Satisfiability Modulo Theories. 1–5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Nicolas Bruno and Surajit Chaudhuri. 2005. Flexible Database Generators. In VLDB. ACM, 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. 2006. Generating Queries with Cardinality Constraints for DBMS Testing.IEEE Trans. Knowl. Data Eng.18, 12 (2006), 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Wei Cao, Yingqiang Zhang, Xinjun Yang, Feifei Li, Sheng Wang, Qingda Hu, Xuntao Cheng, Zongzhi Chen, Zhenjun Liu, Jing Fang, Bo Wang, Yuhui Wang, Haiqing Sun, Ze Yang, Zhushi Cheng, Sen Chen, Jian Wu, Wei Hu, Jianwei Zhao, Yusong Gao, Songlu Cai, Yunyang Zhang, and Jiawang Tong. 2021. PolarDB Serverless: A Cloud Native Database for Disaggregated Data Centers. InSIGMOD. ACM, 2477–2489.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Pascal Cuoq, Benjamin Monate, Anne Pacalet, Virgile Prevosto, John Regehr, Boris Yakobowski, and Xuejun Yang. 2012. Testing Static Analyzers with Randomly Generated Programs. InNASA Formal Methods-4th International Symposium, NFM. Springer, 120–125.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Angjela Davitkova, Damjan Gjurovski, and Sebastian Michel. 2022. LMKG: Learned Models for Cardinality Estimation in Knowledge Graphs. InEDBT. OpenProceedings.org, 2:169–2:182.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "DB-Engines. 2018. DB-Engines Ranking. [EB/OL]. https://db-engines.com/en/ ranking.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Jim Diederich and Jack Milton. 1988. New Methods and Fast Algorithms for Database Normalization.ACM Trans. Database Syst.13, 3 (1988), 339–365.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Chi Thang Duong, Dung Hoang, Hongzhi Yin, Matthias Weidlich, Quoc Viet Hung Nguyen, and Karl Aberer. 2021. Efficient Streaming Subgraph Isomorphism with Graph Neural Networks.VLDB14, 5 (2021), 730–742.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Orri Erling, Alex Averbuch, Josep Larriba-Pey, Hassan Chafi, Andrey Gubichev, Arnau Prat, Minh-Duc Pham, and Peter Boncz. 2015. The LDBC social network benchmark: Interactive workload. InSIGMOD. 619–630.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "M. R. Garey and David S. Johnson. 1979.Computers and Intractability: A Guide to the Theory of NP-Completeness. W. H. Freeman.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Jim Gray, Prakash Sundaresan, Susanne Englert, Kenneth Baclawski, and Peter J. Weinberger. 1994. Quickly Generating Billion-Record Synthetic Databases. In SIGMOD. ACM Press, 243–252.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Zhongxian Gu, Mohamed A Soliman, and Florian M Waas. 2012. Testing the accuracy of query optimizers. InProceedings of the Fifth International Workshop on Testing Database Systems. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Yuanbo Guo, Zhengxiang Pan, and Jeff Heflin. 2005. LUBM: A benchmark for OWL knowledge base systems.Journal of Web Semantics3, 2-3 (2005), 158–182.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Kenneth Houkjær, Kristian Torp, and Rico Wind. 2006. Simple and Realistic Data Generation. InVLDB. ACM, 1243–1246.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Dongxu Huang, Qi Liu, Qiu Cui, Zhuhe Fang, Xiaoyu Ma, Fei Xu, Li Shen, Liu Tang, Yuxing Zhou, Menglong Huang, et al .2020. TiDB: a Raft-based HTAP database.VLDB13, 12 (2020), 3072–3084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Yka Huhtala, Juha Kärkkäinen, Pasi Porkka, and Hannu Toivonen. 1999. TANE: An efficient algorithm for discovering functional and approximate dependencies.The computer journal42, 2 (1999), 100–111.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woon-Hak Kang. 2019. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems.VLDB13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Timotej Kapus and Cristian Cadar. 2017. Automatic testing of symbolic execution engines via program generation and differential testing. InASE. IEEE Computer Society, 590–600.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O. Laleye, and Sarfraz Khurshid. 2008. Query-Aware Test Generation Using a Relational Constraint Solver. InASE. IEEE Computer Society, 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. 2010. Automated SQL query generation for systematic testing of database engines. InASE. ACM, 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. InPLDI. ACM, 216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Eric Lo, Carsten Binnig, Donald Kossmann, M. Tamer Özsu, and Wing-Kai Hon. 2010. A framework for testing DBMS features.VLDB J.19, 2 (2010), 203–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "David Maier. 1983.The Theory of Relational Databases. Computer Science Press.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "MariaDB. 2022. MariaDB hints. [EB/OL]. https://mariadb.com/kb/en/optimizerswitch/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Mariadb. 2022. Mariadb Homepage. [EB/OL]. https://mariadb.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software.Digit. Tech. J.10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. 2008. Generating targeted queries for database testing. InSIGMOD. ACM, 499–510.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "MySQL. 2022. MySQL hints. [EB/OL]. https://dev.mysql.com/doc/refman/8.0/ en/optimizer-hints.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "MySQL. 2022. MySQL Homepage. [EB/OL]. https://www.mysql.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Stack Overflow. 2021. Developer Survey Results. [EB/OL]. https://insights. stackoverflow.com/survey/2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Thorsten Papenbrock and Felix Naumann. 2016. A Hybrid Approach to Functional Dependency Discovery. InSIGMOD, Fatma Özcan, Georgia Koutrika, and Sam Madden (Eds.). ACM, 821–833.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Thorsten Papenbrock and Felix Naumann. 2017. Data-driven Schema Normalization. InEDBT. OpenProceedings.org, 342–353.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Meikel Poess and John M. Stephens. 2004. Generating Thousand Benchmark Queries in Seconds. InVLDB. Morgan Kaufmann, 1045–1053.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Kim-Thomas Rehmann, Changyun Seo, Dongwon Hwang, Binh Than Truong, Alexander Boehm, and Dong Hun Lee. 2016. Performance Monitoring in SAPHANA’s Continuous Integration Process.SIGMETRICS Perform. Evaluation Rev. 43, 4 (2016), 43–52.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InACM Joint Meeting on ESEC and FSE. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 48,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. SQLancer. [EB/OL]. https://github.com/ sqlancer/sqlancer.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. InOSDI 20. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 51,
+      "text": "Michael Schmidt, Thomas Hornung, Georg Lausen, and Christoph Pinkel. 2009. SPˆ 2Bench: a SPARQL performance benchmark. InICDE. IEEE, 222–233.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Andreas Seltenreich. 2020. SQLSmith. [EB/OL]. https://github.com/anse1/ sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. InVLDB, Ashish Gupta, Oded Shmueli, and Jennifer Widom (Eds.). Morgan Kaufmann, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "TiDB. 2022. TiDB hints. [EB/OL]. https://docs.pingcap.com/tidb/v5.3/optimizerhints.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Manasi Vartak, Venkatesh Raghavan, and Elke A. Rundensteiner. 2010. QRelX: generating meaningful queries that provide cardinality assurance. InSIGMOD. ACM, 1215–1218.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Ziheng Wei and Sebastian Link. 2019. Embedded Functional Dependencies and Data-completeness Tailored Database Design.VLDB12, 11 (2019), 1458–1470.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Kesheng Wu, Ekow J. Otoo, and Arie Shoshani. 2002. Compressing Bitmap Indexes for Faster Search Operations. InSSDBM. IEEE Computer Society, 99–108.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Jiaqi Yan, Qiuye Jin, Shrainik Jain, Stratis D. Viglas, and Allison W. Lee. 2018. Snowtrail: Testing with Production Queries on a Cloud Database. InSIGMOD. ACM, 4:1–4:6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. InSIGPLAN, PLDI. ACM, 283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Yixing Yang, Yixiang Fang, Maria E. Orlowska, Wenjie Zhang, and Xuemin Lin. 2021. Efficient Bi-triangle Counting for Large Bipartite Networks.VLDB14, 6 (2021), 984–996.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Rex Ying, Zhaoyu Lou, Jiaxuan You, Chengtao Wen, Arquimedes Canedo, and Jure Leskovec. 2020. Neural Subgraph Matching.CoRRabs/2007.03092 (2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. InCCS. ACM, 955–970.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InACM Joint Meeting on ESEC and FSE. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 48,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. SQLancer. [EB/OL]. https://github.com/ sqlancer/sqlancer.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 50,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. InOSDI 20. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": "Pivoted Query Synthesis (PQS) has recently emerged as a promising way to detect logic bugs in DBMS [ 50].",
+      "context_before": "execution plan with the assistance of built-in cost models. The implementation errors in a query optimizer can ∗Sai Wu is the corresponding author. (a) MySQL’s incorrect hash join execution. (b) MySQL’s incorrect semi-join execution. Figure 1: Logic bug cases of join optimizations in MySQL. result in bugs, including crashes and logic bugs. Crashes are easier to detect as the system will halt immediately. Whereas, logic bugs are prone to be ignored, because they simply cause the DBMS to return incorrect result sets that are hard to detect. In this paper, we focus on detecting these silent bugs.",
+      "context_after": "Its core idea is to select a pivot row from a table and generate queries that fetch this row as the result. A logic bug is detected if the pivot row is not returned in any synthesized query. PQS is mainly designed to support selection queries in a single table and90%of its reported bugs are for queries involving only one table. There still exists substantial research gap regarding multi-table queries with different join algorithms and join structures, which are more error-prone than single-table queries. In Figure 1, we illustrate two logic bugs of MySQL for join queries. These two bugs can b",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3103,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS is mainly designed to support selection queries in a single table and90%of its reported bugs are for queries involving only one table.",
+      "context_before": "t in bugs, including crashes and logic bugs. Crashes are easier to detect as the system will halt immediately. Whereas, logic bugs are prone to be ignored, because they simply cause the DBMS to return incorrect result sets that are hard to detect. In this paper, we focus on detecting these silent bugs. Pivoted Query Synthesis (PQS) has recently emerged as a promising way to detect logic bugs in DBMS [ 50]. Its core idea is to select a pivot row from a table and generate queries that fetch this row as the result. A logic bug is detected if the pivot row is not returned in any synthesized query.",
+      "context_after": "There still exists substantial research gap regarding multi-table queries with different join algorithms and join structures, which are more error-prone than single-table queries. In Figure 1, we illustrate two logic bugs of MySQL for join queries. These two bugs can be detected by our proposed tool in this paper. Figure 1(a) demonstrates a logic bug of hash join in MySQL 8.0.18. In this example, the first query returns the correct result set because it is executed with the block nested loop join. However, when thearXiv:2602.21955v1 [cs.DB] 25 Feb 2026 Conference acronym ’XX, June 03–05, 2023",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3400,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3 Data-guided Schema and Query Generation (DSG) SQLancer [ 49] is a tool to automatically find logic bugs in the implementation of DBMS.",
+      "context_before": "unknown query space. For a query𝑞𝑖, TQS transforms the query with hint sets 𝑡𝑟𝑎𝑛𝑠_𝑞𝑖 to execute multiply different physical query plans (line 11). Finally, the result set of query 𝑡𝑟𝑎𝑛𝑠_𝑞𝑖is compared with the ground-truth 𝐺𝑇𝑞𝑖(line 14). If they are not consistent, a join optimization bug is detected (lines 15). TQS currently focuses on bug detection of equi-join queries. However, the idea of DSG and KQE can be extended to non-equal joins. The only challenge is how to generate and manage groundtruth results, whose sizes increase exponentially for non-equal joins. We leave it to our future work.",
+      "context_after": "It first creates a populated database with randomly generated tables. Afterwards, it randomly chooses SQL statements to create, modify, and delete data. Based on the randomly generated database, it adopts testing approaches such as Pivoted Query Synthesis (PQS) [ 50] to detect logic bugs. Note that SQLancer Conference acronym ’XX, June 03–05, 2023, Woodstock, NY Xiu Tang, et al. is designed for logic bug detection of single-table queries. We can extend it directly to multi-table queries by modifying its data and query generator, but many bugs are ignored in that case. For details, please refe",
+      "section": "2.2 Scheme Overview",
+      "page": 3,
+      "char_offset": 14591,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Based on the randomly generated database, it adopts testing approaches such as Pivoted Query Synthesis (PQS) [ 50] to detect logic bugs.",
+      "context_before": "s detected (lines 15). TQS currently focuses on bug detection of equi-join queries. However, the idea of DSG and KQE can be extended to non-equal joins. The only challenge is how to generate and manage groundtruth results, whose sizes increase exponentially for non-equal joins. We leave it to our future work. 3 Data-guided Schema and Query Generation (DSG) SQLancer [ 49] is a tool to automatically find logic bugs in the implementation of DBMS. It first creates a populated database with randomly generated tables. Afterwards, it randomly chooses SQL statements to create, modify, and delete data.",
+      "context_after": "Note that SQLancer Conference acronym ’XX, June 03–05, 2023, Woodstock, NY Xiu Tang, et al. is designed for logic bug detection of single-table queries. We can extend it directly to multi-table queries by modifying its data and query generator, but many bugs are ignored in that case. For details, please refer to the experimental section. To effectively support bug detection of multi-table join queries, we generate a test database as a wide table and leverage schema normalization to split the wide table into multiple tables with normal form. In addition, we propose an effective noise injection",
+      "section": "2.2 Scheme Overview",
+      "page": 3,
+      "char_offset": 14881,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Note that SQLancer Conference acronym ’XX, June 03–05, 2023, Woodstock, NY Xiu Tang, et al.",
+      "context_before": "on-equal joins. The only challenge is how to generate and manage groundtruth results, whose sizes increase exponentially for non-equal joins. We leave it to our future work. 3 Data-guided Schema and Query Generation (DSG) SQLancer [ 49] is a tool to automatically find logic bugs in the implementation of DBMS. It first creates a populated database with randomly generated tables. Afterwards, it randomly chooses SQL statements to create, modify, and delete data. Based on the randomly generated database, it adopts testing approaches such as Pivoted Query Synthesis (PQS) [ 50] to detect logic bugs.",
+      "context_after": "is designed for logic bug detection of single-table queries. We can extend it directly to multi-table queries by modifying its data and query generator, but many bugs are ignored in that case. For details, please refer to the experimental section. To effectively support bug detection of multi-table join queries, we generate a test database as a wide table and leverage schema normalization to split the wide table into multiple tables with normal form. In addition, we propose an effective noise injection techniques to increase the probability of causing logic bugs with inconsistent database sta",
+      "section": "2.2 Scheme Overview",
+      "page": 3,
+      "char_offset": 15018,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer is not designed to test multi-table queries.",
+      "context_before": "d TPCH data and the dataset from UCI machine learning repository2. However, since TPC-H dataset follows uniform distribution and has a simple schema, all reported bugs on TPC-H dataset have been covered by the UCI dataset. Therefore, we only show the results on the UCI data. As TQS runs, it continuously reports bugs and to be fair for all DBMSs, we only report the results for the first 24 hours. All DBMSs are run with default configurations and compilation options. Baseline approaches.We compare TQS with the SQLancer3 which is the start-of-the-art approach to detecting logic bugs in databases.",
+      "context_after": "However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one tables. Similar to the single table case, all queries and tuples are randomly generated. We use three methods in SQLancer as our baselines. The first one is PQS [ 50], which constructs queries to fetch a randomly selected tuple from a table. The tested DBMS may contain a bug if it fails to fetch that tuple. The second one is TLP [ 50], which decomposes a query into three partitioning queries, each of which computes its result on that tuple. The third one is NoRec [ 47], which",
+      "section": "3.4 Ground-truth Result Generation",
+      "page": 8,
+      "char_offset": 42213,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We use three methods in SQLancer as our baselines.",
+      "context_before": "the UCI data. As TQS runs, it continuously reports bugs and to be fair for all DBMSs, we only report the results for the first 24 hours. All DBMSs are run with default configurations and compilation options. Baseline approaches.We compare TQS with the SQLancer3 which is the start-of-the-art approach to detecting logic bugs in databases. SQLancer is not designed to test multi-table queries. However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one tables. Similar to the single table case, all queries and tuples are randomly generated.",
+      "context_after": "The first one is PQS [ 50], which constructs queries to fetch a randomly selected tuple from a table. The tested DBMS may contain a bug if it fails to fetch that tuple. The second one is TLP [ 50], which decomposes a query into three partitioning queries, each of which computes its result on that tuple. The third one is NoRec [ 47], which targets at logic bugs generated by the optimization process in DBMS. It performs a comparison between the results of randomly-generated queries and their rewritten ones. DBMS versions. Note that DBMS bugs will be fixed once reported to the community. In our",
+      "section": "3.4 Ground-truth Result Generation",
+      "page": 8,
+      "char_offset": 42475,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "The first one is PQS [ 50], which constructs queries to fetch a randomly selected tuple from a table.",
+      "context_before": "bugs and to be fair for all DBMSs, we only report the results for the first 24 hours. All DBMSs are run with default configurations and compilation options. Baseline approaches.We compare TQS with the SQLancer3 which is the start-of-the-art approach to detecting logic bugs in databases. SQLancer is not designed to test multi-table queries. However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one tables. Similar to the single table case, all queries and tuples are randomly generated. We use three methods in SQLancer as our baselines.",
+      "context_after": "The tested DBMS may contain a bug if it fails to fetch that tuple. The second one is TLP [ 50], which decomposes a query into three partitioning queries, each of which computes its result on that tuple. The third one is NoRec [ 47], which targets at logic bugs generated by the optimization process in DBMS. It performs a comparison between the results of randomly-generated queries and their rewritten ones. DBMS versions. Note that DBMS bugs will be fixed once reported to the community. In our experiments, we report the results of the latest released versions during our testing, namely, MySQL 8",
+      "section": "3.4 Ground-truth Result Generation",
+      "page": 8,
+      "char_offset": 42526,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The second one is TLP [ 50], which decomposes a query into three partitioning queries, each of which computes its result on that tuple.",
+      "context_before": "roaches.We compare TQS with the SQLancer3 which is the start-of-the-art approach to detecting logic bugs in databases. SQLancer is not designed to test multi-table queries. However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one tables. Similar to the single table case, all queries and tuples are randomly generated. We use three methods in SQLancer as our baselines. The first one is PQS [ 50], which constructs queries to fetch a randomly selected tuple from a table. The tested DBMS may contain a bug if it fails to fetch that tuple.",
+      "context_after": "The third one is NoRec [ 47], which targets at logic bugs generated by the optimization process in DBMS. It performs a comparison between the results of randomly-generated queries and their rewritten ones. DBMS versions. Note that DBMS bugs will be fixed once reported to the community. In our experiments, we report the results of the latest released versions during our testing, namely, MySQL 8.0.28, MariaDB 10.8.2, TiDB 5.4.0 and X-DB beta 8.0.18 respectively. Setup.All experiments are conducted on our in-house server equipped with Intel Xeon CPU E5-2682 (2.50GHz) 16 cores and 128GB memory. F",
+      "section": "3.4 Ground-truth Result Generation",
+      "page": 8,
+      "char_offset": 42695,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "The third one is NoRec [ 47], which targets at logic bugs generated by the optimization process in DBMS.",
+      "context_before": "esigned to test multi-table queries. However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one tables. Similar to the single table case, all queries and tuples are randomly generated. We use three methods in SQLancer as our baselines. The first one is PQS [ 50], which constructs queries to fetch a randomly selected tuple from a table. The tested DBMS may contain a bug if it fails to fetch that tuple. The second one is TLP [ 50], which decomposes a query into three partitioning queries, each of which computes its result on that tuple.",
+      "context_after": "It performs a comparison between the results of randomly-generated queries and their rewritten ones. DBMS versions. Note that DBMS bugs will be fixed once reported to the community. In our experiments, we report the results of the latest released versions during our testing, namely, MySQL 8.0.28, MariaDB 10.8.2, TiDB 5.4.0 and X-DB beta 8.0.18 respectively. Setup.All experiments are conducted on our in-house server equipped with Intel Xeon CPU E5-2682 (2.50GHz) 16 cores and 128GB memory. For our cloud-native database X-DB, we run an instance on our public cloud with similar computation capabi",
+      "section": "3.4 Ground-truth Result Generation",
+      "page": 8,
+      "char_offset": 42831,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "sqlancer",
+      "technique": null,
+      "sentence": "com/sqlancer/sqlancer Detecting Logic Bugs of Join Optimizations in DBMS Conference acronym ’XX, June 03–05, 2023, Woodstock, NY Table 3: We tested a diverse set of popular and emerging DBMS.",
+      "context_before": ", namely, MySQL 8.0.28, MariaDB 10.8.2, TiDB 5.4.0 and X-DB beta 8.0.18 respectively. Setup.All experiments are conducted on our in-house server equipped with Intel Xeon CPU E5-2682 (2.50GHz) 16 cores and 128GB memory. For our cloud-native database X-DB, we run an instance on our public cloud with similar computation capability to our in-house server. 5.1 Overview and Showcase of Bug Reports TQS has successfully detected 115 bugs from tested DBMSs within 24 hours, including 31 bugs in MySQL, 30 bugs in MariaDB, 31 bugs 2https://archive.ics.uci.edu/ml/datasets/KDD+Cup+1998+Data 3https://github.",
+      "context_after": "All numbers are the lasted as of July 2022. DBMSPopularity Rank LOCFirst ReleaseDBEnginesStack OverflowGithub Stars MySQL 2 1 8.0k 3.8M 1995 MariaDB 12 7 4.3k 3.6M 2009 TiDB 96 – 31.8k 0.8M 2017 in TiDB, and 23 bugs in X-DB respectively. However, some bugs may be caused by the same relational algebras. Hence, once finding a possible bug, it is essential to produce a minimal test case with C-Reduce4before reporting the bug, to save the DBMS developers’ time and effort. Then, we submit all bugs and the corresponding test cases to developer communities and received their positive feedbacks. Thro",
+      "section": "5.1 Overview and Showcase of Bug Reports",
+      "page": 8,
+      "char_offset": 43811,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "CREATE TABLEt 1 ( i d b i g i n t ( 6 4 )NOT NULLAUTO_INCREMENT, c o l 1varchar( 5 1 1 )DEFAULT NULL, PRIMARY KEY( i d ) ); CREATE TABLEt 2 ( i d b i g i n t ( 6 4 )NOT NULLAUTO_INCREMENT, c o l 1varchar( 5 1 1 )DEFAULT NULL, PRIMARY KEY( i d ) ); Detecting Logic Bugs of Join Optimizations in DBMS Conference acronym ’XX, June 03–05, 2023, Woodstock, NY 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560100200300400/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni0000004a/uni00000055/uni00000044/uni00000053/uni0000004b/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQSPQS TLP (a) MySQL diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560100200300400/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni0000004a/uni00000055/uni00000044/uni00000053/uni0000004b/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQS NoRec (b) MariaDB diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560100200300400/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni0000004a/uni00000055/uni00000044/uni00000053/uni0000004b/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQSTLP (c) TiDB diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560100200300400/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni0000004a/uni00000055/uni00000044/uni00000053/uni0000004b/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQSPQS TLP (d) X-DB diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560510152025/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 TQSPQS TLP (e) MySQL efficiency 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560510152025/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 TQS NoRec (f) MariaDB efficiency 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni00000056051015202530/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 TQSTLP (g) TiDB efficiency 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni0000005605101520/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 TQSPQS TLP (h) X-DB efficiency Figure 8: Comparison with existing tools in the query graph diversity and the efficiency of detecting bugs.",
+      "context_before": "−−−−−+ |NULL| |NULL| +−−−−−−+ t r a n s f o r m e d query: SETo p t i m i z e r _ s w i t c h ='j o i n _ c a c h e _ b k a = o f f'; SELECTt 3. c0FROMt 3RIGHT OUTER JOINt 4ONt 3. c1 = t 4. c1 JOINt 2ONt 2. c1 = t 4. c1CROSS JOINt 1; r e s u l t: Emptyset Listing 4: MariaDB’s incorrect index join execution. 5.1.3 Bugs in TiDB.Merge join and join index are the main causes of bugs in TiDB. Listing 5 shows an example query, where incorrect result set is obtained when hash join is converted to merge join (due to space constraints, the ground truth results and incorrect results are discarded here).",
+      "context_after": "CREATE TABLEt 3 ( i d b i g i n t ( 6 4 )NOT NULLAUTO_INCREMENT, c o l 1varchar( 5 1 1 )DEFAULT NULL, PRIMARY KEY( i d ) ); SELECT/∗+ me rge _j oi n ( t1, t2, t 3 )∗/ t 3. c o l 1FROM( t 1LEFT JOINt 2ONt 1. c o l 1 = t 2. c o l 1 )LEFT JOINt 3ONt 2. c o l 1 = t 3. c o l 1; SELECT/∗+ h a s h _ j o i n ( t1, t2, t 3 )∗/ t 3. c o l 1FROM( t 1LEFT JOINt 2ONt 1. c o l 1 = t 2. c o l 1 )LEFT JOINt 3ONt 2. c o l 1 = t 3. c o l 1; Listing 5: TiDB’s merge join executed incorrect result set which misses-0. We find that when TiDB executes the above merge join, intermediate data are mistakenly materializ",
+      "section": "5.1.3 Bugs in TiDB.Merge join and join index are the main causes",
+      "page": 10,
+      "char_offset": 51771,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ],
+      "text_is_letter_spaced": true
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Note that due to the compatibility problem, SQLancer implements different approaches on different databases (PQS and TLP on MySQL and X-DB; NoRec on MariaDB; TLP on TiDB).",
+      "context_before": "To p t i m i z e r _ s w i t c h ='m a t e r i a l i z a t i o n = o f f'; Conference acronym ’XX, June 03–05, 2023, Woodstock, NY Xiu Tang, et al. SELECT ALLt 1. c0FROMt 1RIGHT OUTER JOINt 0ONt 1. c0 = t 0. c0WHEREt 1. c0IN(SELECTt 0. c0FROMt 0WHERE( t 1. c0NOT IN(SELECTt 1. c0FROMt 1 ) ) = ( 1 )INt 1. c0 ); r e s u l t: +−−−−−−−−−−−+ | c0 | +−−−−−−−−−−−+ | 292269000 | +−−−−−−−−−−−+ Listing 7: X-DB’s incorrect hash join execution. 5.2 Comparison with Existing Tools We compare TQS with three state-of-the-art DBMS testing approaches in different aspects to show its effectiveness and efficiency.",
+      "context_after": "Figure 8 shows our comparison results. We adopt two metrics. The diversity of graphs shows the number of different isomorphic sets tested, and the bug type count represents the number of returned logic bug types. Query graph diversity.Figure 8(a), (b), (c) and (d) show the query graph diversity (in thousands of isomorphic sets) of MySQL, MariaDB, TiDB and X-DB, respectively. TQS significantly outperforms baseline approaches on testing diversity. This is because that (1) SQLancer approaches may generate random joins with empty results which are not usable for testing and (2) TQS adopts the KQE",
+      "section": "5.2 Comparison with Existing Tools",
+      "page": 12,
+      "char_offset": 57510,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This is because that (1) SQLancer approaches may generate random joins with empty results which are not usable for testing and (2) TQS adopts the KQE to avoid repeatedly testing the same query structure.",
+      "context_before": "compatibility problem, SQLancer implements different approaches on different databases (PQS and TLP on MySQL and X-DB; NoRec on MariaDB; TLP on TiDB). Figure 8 shows our comparison results. We adopt two metrics. The diversity of graphs shows the number of different isomorphic sets tested, and the bug type count represents the number of returned logic bug types. Query graph diversity.Figure 8(a), (b), (c) and (d) show the query graph diversity (in thousands of isomorphic sets) of MySQL, MariaDB, TiDB and X-DB, respectively. TQS significantly outperforms baseline approaches on testing diversity.",
+      "context_after": "Efficiency.Figure 8(e), (f), (g) and (h) show the testing efficiency of MySQL, MariaDB, TiDB and X-DB, respectively. Not surprisingly, we obtain a similar result as the diversity experiment. More query structures are being tested, more bugs can be discovered. 04812162024283236404448 /uni0000004b/uni00000052/uni00000058/uni00000055/uni0000005601020304050/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 bug count bug type Figure 9: Bug types vs bug counts on MySQL. In fact, we run the TQS for 48 hours and continue to find logic bugs for",
+      "section": "5.2 Comparison with Existing Tools",
+      "page": 12,
+      "char_offset": 58132,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 49] is a current stateof-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours.",
+      "context_before": "rectness of the system, have received less attention. Generation-based testings [24,32,34,39,46,58,62] have been adopted for extensive testings on DBMSs for purposes such as bug-finding and benchmarking. SQLSmith is a widely-used generation-based DBMS tester [ 52]. It synthesizes a schema from initial databases and generates limited types of queries, whose target is at the code coverage. Squirrel focuses on generating queries to detect memory corruption bugs [62]. All above random generators are mostly applied to detect crashing bugs, while our focus is the logic bug.Logic Bug Testing of DBMS.",
+      "context_after": "SQLancer proposes three approaches to detect logic bugs. PQS constructs queries to fetch a randomly selected tuple from a table [ 50]. TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 48]. NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [47]. SQLancer targets at single table queries and 90.0% of its bug reports include only one table. On the other hand, our TQS targets at detecting logic bugs of multi-table joins, which are more prone to bugs. Database Schema Norma",
+      "section": "5.3 Ablation Studies",
+      "page": 13,
+      "char_offset": 64967,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer proposes three approaches to detect logic bugs.",
+      "context_before": "extensive testings on DBMSs for purposes such as bug-finding and benchmarking. SQLSmith is a widely-used generation-based DBMS tester [ 52]. It synthesizes a schema from initial databases and generates limited types of queries, whose target is at the code coverage. Squirrel focuses on generating queries to detect memory corruption bugs [62]. All above random generators are mostly applied to detect crashing bugs, while our focus is the logic bug.Logic Bug Testing of DBMS.SQLancer [ 49] is a current stateof-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours.",
+      "context_after": "PQS constructs queries to fetch a randomly selected tuple from a table [ 50]. TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 48]. NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [47]. SQLancer targets at single table queries and 90.0% of its bug reports include only one table. On the other hand, our TQS targets at detecting logic bugs of multi-table joins, which are more prone to bugs. Database Schema Normalization.The well-known database design framework for rel",
+      "section": "5.3 Ablation Studies",
+      "page": 13,
+      "char_offset": 65092,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS constructs queries to fetch a randomly selected tuple from a table [ 50].",
+      "context_before": "ing and benchmarking. SQLSmith is a widely-used generation-based DBMS tester [ 52]. It synthesizes a schema from initial databases and generates limited types of queries, whose target is at the code coverage. Squirrel focuses on generating queries to detect memory corruption bugs [62]. All above random generators are mostly applied to detect crashing bugs, while our focus is the logic bug.Logic Bug Testing of DBMS.SQLancer [ 49] is a current stateof-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours. SQLancer proposes three approaches to detect logic bugs.",
+      "context_after": "TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 48]. NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [47]. SQLancer targets at single table queries and 90.0% of its bug reports include only one table. On the other hand, our TQS targets at detecting logic bugs of multi-table joins, which are more prone to bugs. Database Schema Normalization.The well-known database design framework for relational databases is centered around the notion of data redundancy [ 7,35]. Th",
+      "section": "5.3 Ablation Studies",
+      "page": 13,
+      "char_offset": 65149,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 48].",
+      "context_before": "52]. It synthesizes a schema from initial databases and generates limited types of queries, whose target is at the code coverage. Squirrel focuses on generating queries to detect memory corruption bugs [62]. All above random generators are mostly applied to detect crashing bugs, while our focus is the logic bug.Logic Bug Testing of DBMS.SQLancer [ 49] is a current stateof-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours. SQLancer proposes three approaches to detect logic bugs. PQS constructs queries to fetch a randomly selected tuple from a table [ 50].",
+      "context_after": "NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [47]. SQLancer targets at single table queries and 90.0% of its bug reports include only one table. On the other hand, our TQS targets at detecting logic bugs of multi-table joins, which are more prone to bugs. Database Schema Normalization.The well-known database design framework for relational databases is centered around the notion of data redundancy [ 7,35]. The redundant data value occurrences originate from functional dependencies (FDs) [ 8]. The data-driven normalization a",
+      "section": "5.3 Ablation Studies",
+      "page": 13,
+      "char_offset": 65227,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [47].",
+      "context_before": "ode coverage. Squirrel focuses on generating queries to detect memory corruption bugs [62]. All above random generators are mostly applied to detect crashing bugs, while our focus is the logic bug.Logic Bug Testing of DBMS.SQLancer [ 49] is a current stateof-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours. SQLancer proposes three approaches to detect logic bugs. PQS constructs queries to fetch a randomly selected tuple from a table [ 50]. TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 48].",
+      "context_after": "SQLancer targets at single table queries and 90.0% of its bug reports include only one table. On the other hand, our TQS targets at detecting logic bugs of multi-table joins, which are more prone to bugs. Database Schema Normalization.The well-known database design framework for relational databases is centered around the notion of data redundancy [ 7,35]. The redundant data value occurrences originate from functional dependencies (FDs) [ 8]. The data-driven normalization algorithms [ 19,44,56] can remove FDrelated redundancy effectively. Schema normalization has been well studied. In our app",
+      "section": "5.3 Ablation Studies",
+      "page": 13,
+      "char_offset": 65344,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer targets at single table queries and 90.",
+      "context_before": "re mostly applied to detect crashing bugs, while our focus is the logic bug.Logic Bug Testing of DBMS.SQLancer [ 49] is a current stateof-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours. SQLancer proposes three approaches to detect logic bugs. PQS constructs queries to fetch a randomly selected tuple from a table [ 50]. TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 48]. NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [47].",
+      "context_after": "0% of its bug reports include only one table. On the other hand, our TQS targets at detecting logic bugs of multi-table joins, which are more prone to bugs. Database Schema Normalization.The well-known database design framework for relational databases is centered around the notion of data redundancy [ 7,35]. The redundant data value occurrences originate from functional dependencies (FDs) [ 8]. The data-driven normalization algorithms [ 19,44,56] can remove FDrelated redundancy effectively. Schema normalization has been well studied. In our approach, we adopt previous database normalization a",
+      "section": "5.3 Ablation Studies",
+      "page": 13,
+      "char_offset": 65465,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M7"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:38:14Z",
+    "is_model_written": true,
+    "summary": "TQS targets logic bugs in join optimisation, which earlier oracles largely miss: PQS is designed for selection over a single table, and 90% of the bugs it reported involve one table only. TQS generates a schema and data together so that the correct result of a multi-table join is known by construction, then uses a knowledge-guided query enumerator to avoid re-testing equivalent query shapes. It found bugs in MySQL, MariaDB, TiDB and others.",
+    "narrative": "SQLancer is the work TQS is built against and the paper says so directly, calling it the current state-of-the-art tool for finding logic bugs in DBMSs and the most closely related work to its own. All three of its oracles serve as baselines -- PQS, TLP and NoREC -- and the paper is careful about the comparison's mechanics, noting that SQLancer supports different oracles on different systems and that its generated joins often return empty results and so cannot be used for testing. The gap TQS identifies is single-table scope: SQLancer is not designed to test multi-table queries, which is exactly where join-optimisation bugs live.",
+    "roles": {
+      "M1": "background",
+      "M2": "motivation",
+      "M3": "definition",
+      "M4": "definition",
+      "M5": "incidental",
+      "M6": "motivation",
+      "M7": "baseline",
+      "M8": "definition",
+      "M9": "definition",
+      "M10": "definition",
+      "M11": "incidental",
+      "M12": "incidental",
+      "M13": "result_comparison",
+      "M14": "result_comparison",
+      "M15": "state_of_the_art",
+      "M16": "definition",
+      "M17": "definition",
+      "M18": "definition",
+      "M19": "definition",
+      "M20": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The three oracles are run as baselines. Nothing states that TQS's own schema-and-data generation or its query enumerator is built on the SQLancer codebase."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "TQS constructs the schema and data so a join's correct result is known in advance, which is a different oracle from PQS, TLP or NoREC rather than a generalisation of one. Its relationship to them is that it covers the multi-table case they do not."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M7",
+          "M13",
+          "M14"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M7",
+            "sentence": "We use three methods in SQLancer as our baselines.",
+            "section": "3.4 Ground-truth Result Generation",
+            "page": 8
+          },
+          {
+            "mention_id": "M13",
+            "sentence": "Note that due to the compatibility problem, SQLancer implements different approaches on different databases (PQS and TLP on MySQL and X-DB; NoRec on MariaDB; TLP on TiDB).",
+            "section": "5.2 Comparison with Existing Tools",
+            "page": 12
+          },
+          {
+            "mention_id": "M14",
+            "sentence": "This is because that (1) SQLancer approaches may generate random joins with empty results which are not usable for testing and (2) TQS adopts the KQE to avoid repeatedly testing the same query structure.",
+            "section": "5.2 Comparison with Existing Tools",
+            "page": 12
+          }
+        ],
+        "reasoning": "M7 states that three methods in SQLancer are used as baselines, M13 records which oracle was run against which system, and M14 explains a difference in the measured outcome.",
+        "techniques": [
+          "pqs",
+          "tlp",
+          "norec"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M15"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M15",
+            "sentence": "SQLancer [ 49] is a current stateof-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours.",
+            "section": "5.3 Ablation Studies",
+            "page": 13
+          }
+        ],
+        "reasoning": "M15 calls SQLancer a current state-of-the-art tool in testing DBMSs for logic bugs and the work most closely related to this one."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3597503_3608133.json b/_data/papers/paper_doi_10_1145_3597503_3608133.json
new file mode 100644
index 0000000..4766997
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3597503_3608133.json
@@ -0,0 +1,404 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:33:10Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3608133",
+    "title": "EDEFuzz: A Web API Fuzzer for Excessive Data Exposures",
+    "authors": [
+      "Lianglu Pan",
+      "Shaanan Cohney",
+      "Toby C. Murray",
+      "Van-Thuan Pham"
+    ],
+    "year": 2023,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3608133",
+    "arxiv_id": "2301.09258",
+    "s2_paper_id": "d2bf26b7ec662c29c30e6ad4548ce53d360718f2",
+    "url": "https://doi.org/10.1145/3597503.3608133",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2301.09258",
+      "retrieved_at": "2026-09-08T14:33:10Z",
+      "chars": 71296,
+      "content_sha256": "sha256:3836cc133539dce41cc1124f2029e0464c1576f060fd693cc3fd7b53a72b1a28"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1785
+      },
+      {
+        "number": "1.1",
+        "title": "Research Ethics",
+        "start": 11322
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND",
+        "start": 12679
+      },
+      {
+        "number": "2.1",
+        "title": "Web APIs & Excessive Data Exposure",
+        "start": 12692
+      },
+      {
+        "number": "2.2",
+        "title": "Fuzzing",
+        "start": 14049
+      },
+      {
+        "number": "2.3",
+        "title": "Metamorphic Testing/Fuzzing",
+        "start": 15621
+      },
+      {
+        "number": "3",
+        "title": "MOTIVATING EXAMPLES",
+        "start": 17319
+      },
+      {
+        "number": "4",
+        "title": "OUR APPROACH",
+        "start": 23748
+      },
+      {
+        "number": "4.1",
+        "title": "Overview",
+        "start": 24439
+      },
+      {
+        "number": "4.2",
+        "title": "Simulated Server",
+        "start": 30706
+      },
+      {
+        "number": "4.3",
+        "title": "Mutation Engine",
+        "start": 32390
+      },
+      {
+        "number": "4.5",
+        "title": "Result Inspection",
+        "start": 39112
+      },
+      {
+        "number": "5",
+        "title": "EXPERIMENTAL EVALUATION",
+        "start": 40518
+      },
+      {
+        "number": "5.1",
+        "title": "Procedure",
+        "start": 44605
+      },
+      {
+        "number": "5.2",
+        "title": "Results",
+        "start": 46737
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "OWASP API Top 10: 2019, 2019. [Online]. Available: https://owasp.org/wwwproject-api-security/ (cit. on p. 1).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Microsoft announces new Project OneFuzz framework, 2020. [Online]. Available:https://www.microsoft.com/security/blog/2020/09/15/microsoft-onefuzzframework-open-source-developer-tool-fix-bugs/ (cit.on p. 1).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "E.T.Barr,M.Harman,P.McMinn,M.Shahbaz,andS.Yoo,Theoracleproblem in software testing: A survey, IEEE transactions on software engineering, vol.41, no. 5, 507–525, 2014 (cit. on p. 1).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "T.Y.Chen,S.C.Cheung,andS.M.Yiu “Metamorphictesting:Anewapproach for generating next test cases,” Tech. Rep. HKUST-CS98-01, 1998, arXiv preprint arXiv:2002.12543 (cit. on pp. 1,3).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Usingburptotestforsensitivedataexposureissues ,2022.[Online].Available: https://portswigger.net/support/using-burp-to-test-for-sensitive-data-exposureissues(cit. on pp. 2,4).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "V.Atlidakis,P.Godefroid,andM.Polishchuk Restler:Statefulrestapifuzzing, in (ICSE), IEEE, 2019, 748–758 (cit. on pp. 2,3,10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "R.Netravali,A.Sivaraman,S.Das, etal. Mahimahi:Accurate {record-andreplay }for{http},in2015USENIXAnnualTechnicalConference(USENIX ATC15), 2015, 417–429 (cit. on pp. 2,11).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "S. Ikeda, Massive optus data leak prompts new privacy rules in australia, CPO Magazine, 2022 (cit. on p. 2).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Api3:2019excessivedataexposure ,2019.[Online].Available: https://github. com/OWASP/API-Security/blob/master/2019/en/src/0xa3-excessive-dataexposure.md (cit. on p. 3).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "V.Atlidakis,R.Geambasu,P.Godefroid,M.Polishchuk,andB.Ray,Pythia: Grammar-basedfuzzingofrestapiswithcoverage-guidedfeedbackandlearningbased mutations, arXiv preprintarXiv:2005.11498 ,2020 (cit. on pp. 3,10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "V.Atlidakis,P.Godefroid,andM.Polishchuk Checkingsecurityproperties of cloud service rest apis, in Software Testing, Validation and Verification (ICST), IEEE, 2020, 387–397 (cit.on pp. 3,10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "OpenAPIInitiative Openapispecification ,Availableonline: https://swagger. io/resources/open-api/, 2020 (cit. on p. 3).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "V.J.M.Manès,H.Han,C.Han, etal.,Theart,science,andengineeringof fuzzing: A survey, TSE,2019 (cit. on pp. 3,10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Google clusterfuzz, 2022. [Online]. Available: https://google.github.io/ clusterfuzz/ (cit. on p. 3).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "P.He,C.Meister,andZ.Su Testingmachinetranslationvia referentialtransparency, in neering (ICSE), IEEE, 2021, 410–422 (cit. on p. 3).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "M. Rigger, and Z. Su Testing database engines via pivoted query synthesis, in 14thUSENIXSymposiumonOperatingSystemsDesignandImplementation (OSDI20), 2020, 667–682 (cit. on p. 3).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 17,
+      "text": "What is social engineering? 2022. [Online]. Available: https://www.kaspersky. com.au/resource-center/definitions/what-is-social-engineering (cit. on p. 4).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Webdriver ,2022.[Online].Available: https://www.selenium.dev/documentation/ webdriver/ (cit. on p. 5).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "American fuzzy lop. [Online]. Available: https://lcamtuf.coredump.cx/afl/ (cit.on p. 6).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "A.Zeller,Yesterday,myprogramworked.today,itdoesnot.why? ACMSIGSOFT Softwareengineering notes, vol.24, no. 6, 253–267, 1999 (cit. on p. 6).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "R. Wang, G. Xu, X. Zeng, X. Li, and Z. Feng, Tt-xss: A novel taint tracking baseddynamicdetectionframeworkfordomcross-sitescripting, Journalof Parallel and Distributed Computing, vol. 118, 100–106, 2018 (cit. on p. 10). ICSE 2024, April 14–20, 2024, Lisbon, Portugal Pan et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "F. Duchene, S. Rawat, J. -L. Richier, and R. Groz Kameleonfuzz: Evolutionary fuzzing for black-box xss detection, in Proceedings of the 4th ACM conference on Data and application security and privacy, 2014, 37–48 (cit. on p. 10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "O. v. Rooij, M. A. Charalambous, D. Kaizer, M. Papaevripides, and E. Athanasopoulos Webfuzz: Grey-box fuzzing for web applications, in European SymposiumonResearchinComputerSecurity ,Springer,2021,152–172(cit.on p.10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "U. Sarmah, D. Bhattacharyya, and J. K. Kalita, A survey of detection methods for xss attacks, Journal of Network and Computer Applications, vol. 118, 113– 143,2018 (cit. on p. 10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "A. Doupé, L. Cavedon, C. Kruegel, and G. Vigna Enemy of the state: A {state-aware }{black-box }web vulnerability scanner, in 21st USENIX Security Symposium (USENIX Security 12), 2012, 523–538 (cit. on p. 10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "M.C.Martin,andM.S.Lam Automaticgenerationofxssandsqlinjection attacks with goal-directed model checking. In USENIX Security symposium, 2008, 31–44 (cit. on p. 10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "E.Trickel,F.Pagani,C.Zhu, etal. Tossafaulttoyourwitcher:Applyinggreybox coverage-guided mutational fuzzing to detect sql and command injection vulnerabilities, in Computer Society, 2022, 116–133 (cit. on p. 10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Owasp top 10: 2021, 2021. [Online]. Available: https://owasp.org/Top10/ (cit. on p. 10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "W.Koch,A.Chaabane,M.Egele,W.Robertson,andE.Kirda Semi-automated discoveryofserver-basedinformationoversharingvulnerabilitiesinandroidapplications,in Proceedingsofthe26thACMSIGSOFTInternationalSymposium on Software Testing and Analysis, 2017, 147–157 (cit. on p. 10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "S. Segura, J. A. Parejo, J. Troya, and A. Ruiz-Cortés, Metamorphic testingof restful web apis, IEEE Transactions on Software Engineering, vol. 44, no. 11, 1083–1099,2017 (cit. on p. 10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "P.Godefroid,B. -Y.Huang,andM.Polishchuk Intelligentrestapidatafuzzing, inProceedingsofthe28thACMJointMeetingonEuropeanSoftwareEngineeringConferenceandSymposiumontheFoundationsofSoftwareEngineering, 2020, 725–736 (cit. on p. 10).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "S.Andrica,andG.Candea Warr:Atoolforhigh-fidelitywebapplicationrecord andreplay,in 2011IEEE/IFIP41stInternationalConferenceonDependable Systems & Networks (DSN), IEEE, 2011, 403–410 (cit. on p. 11).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "B.Burg,R.Bailey,A.J.Ko,andM.D.Ernst Interactiverecord/replayforweb application debugging, in Proceedings of the 26th annual ACM symposium on User interface software and technology, 2013, 473–484 (cit. on p. 11).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "S.Flesca,F.Furfaro,andE.Masciari Monitoringwebinformationchanges, inProceedings International Conference on Information Technology: Coding and Computing, IEEE, 2001, 421–425 (cit. on p. 11).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Y. Wang, D. J. DeWitt, and J. -Y. Cai X-diff: An effective change detection algorithm for xml documents, in Proceedings 19th international conference on data engineering (Cat. No. 03CH37405), IEEE, 2003, 519–530 (cit. on p. 11).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "I.Khoury,R.M.El-Mawas,O.El-Rawas,E.F.Mounayar,andH.Artail,An efficientwebpagechangedetectionsystembasedonanoptimizedhungarian algorithm, IEEETransactionsonKnowledgeandDataEngineering ,vol.19, no. 5, 599–613, 2007 (cit. on p. 11).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "M. Hammoudi, G. Rothermel, and A. Stocco Waterfall: An incremental approach for repairing record-replay tests of web applications, in Proceedings of the 2016 24th ACMSIGSOFT International Symposium on Foundations of Software Engineering, 2016, 751–762 (cit. on p. 11).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "F. Shao, R. Xu, W. Haque, etal. Webevo: Taming web application evolution via detecting semantic structure changes, in Proceedings of the 30th ACM SIGSOFTInternationalSymposiumonSoftwareTestingandAnalysis ,2021, 16–28 (cit. on p. 11).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "S.Schumilo,C.Aschermann,A.Jemmett,A.Abbasi,andT.Holz,Nyx-net: Networkfuzzingwithincrementalsnapshots, arXivpreprintarXiv:2111.03013, 2021 (cit. on p. 11).",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 16,
+      "text": "M. Rigger, and Z. Su Testing database engines via pivoted query synthesis, in 14thUSENIXSymposiumonOperatingSystemsDesignandImplementation (OSDI20), 2020, 667–682 (cit. on p. 3).",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 16]",
+      "technique": "pqs",
+      "sentence": "Pastworkhassuccessfullyidentifiedandusedmetamorphic relations to find bugs in a variety of systems, including Machine Translation software [ 15] and popular DBMSs [ 16].",
+      "context_before": "tion ) holds between their corresponding outputs. Consider the following toy example of metamorphic bug finding: wecantestafunctionthatreversesalistbytesting,foranarbitrary inputlist G,whetherreversingthereverseof Gyields Gitself;orfora functionthatcalculatesdistancebetweenapairofpoints,whether thedistancefrompoint 0to2isalwayssmallerorequaltothesum of the distances from 0to1and from 1to2, etc. Metamorphicrelationsarepropertiesthatmustnecessarilyhold with respect to the correct functioning of the SUT. In metamorphic testing, the violation of a metamorphic relation indicates a potential bug[4].",
+      "context_after": "3 MOTIVATING EXAMPLES Inthissection,wepresentaselectionofreal-world EDEsdetectedby ourtooltodemonstratetheirprevalenceandimplications,thechallenges in detecting them using prior approaches, and the motivation for our tool (which we comprehensively evaluate in Section 5 ). Vulnerability 1 - Locations and Contact Details EDEFuzz discoveredavulnerabilitywhiletestingthelivedeliverytrackingservice offered by Company-I, an Australian last-mile delivery service. The vulnerabilityhasbeenreportedandfixed.Asshownin Figure2,a customerreceivesauniquelinkonthedaythatanitemisonboard for delivery. The link o",
+      "section": "2.3 Metamorphic Testing/Fuzzing",
+      "page": 3,
+      "char_offset": 17149,
+      "cited_reference": {
+        "number": 16,
+        "text": "M. Rigger, and Z. Su Testing database engines via pivoted query synthesis, in 14thUSENIXSymposiumonOperatingSystemsDesignandImplementation (OSDI20), 2020, 667–682 (cit. on p. 3).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "EDEFuzz detects Excessive Data Exposure, where a web API sends clients far more data than they need, often over public channels. The authors note the problem has no explicit oracle -- nothing crashes -- and develop a metamorphic relation to supply one, building the first fuzzer for it. Across 69 targets from the Alexa Top-200 it found 33,365 potential leaks, and on eight websites it reached a 98.65% true positive rate.",
+    "narrative": "One citation, in a list of systems where metamorphic relations have been used successfully to find bugs.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3597503_3639112.json b/_data/papers/paper_doi_10_1145_3597503_3639112.json
new file mode 100644
index 0000000..985b6d3
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3597503_3639112.json
@@ -0,0 +1,1095 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:28Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3639112",
+    "title": "Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation",
+    "authors": [
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Mingzhe Wang",
+      "Chengnian Sun",
+      "Yu Jiang"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3639112",
+    "arxiv_id": null,
+    "s2_paper_id": "53f3c597ee4fe41383364fdaf42b13973ee1cb39",
+    "url": "https://doi.org/10.1145/3597503.3639112",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3597503.3639112",
+      "retrieved_at": "2026-09-09T01:31:28Z",
+      "chars": 74368,
+      "content_sha256": "sha256:603df18ff863b98d85cb164f18fbd72ebe04ea4b80d5b7be4a6c590fda2fa071"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 3078
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND ANDMOTIVATION",
+        "start": 10605
+      },
+      {
+        "number": "3",
+        "title": "Filter ....",
+        "start": 16095
+      },
+      {
+        "number": "3.1",
+        "title": "Definitions",
+        "start": 20418
+      },
+      {
+        "number": "3.2",
+        "title": "Plan Analysis",
+        "start": 21657
+      },
+      {
+        "number": "3.3",
+        "title": "Plan-Guided Equivalent Transformation",
+        "start": 24323
+      },
+      {
+        "number": "3.4",
+        "title": "Various Test Oracles Construction",
+        "start": 26854
+      },
+      {
+        "number": "4",
+        "title": "IMPLEMENTATION",
+        "start": 30816
+      },
+      {
+        "number": "5",
+        "title": "EVALUATION",
+        "start": 33530
+      },
+      {
+        "number": "5.1",
+        "title": "Evaluation Setup",
+        "start": 33980
+      },
+      {
+        "number": "2.25",
+        "title": "GHz, 128 cores, and 504 GiB of main memory. All DBMSs were",
+        "start": 34249
+      },
+      {
+        "number": "5.2",
+        "title": "DBMS Vulnerability Detection",
+        "start": 34590
+      },
+      {
+        "number": "5.3",
+        "title": "Comparison with Other Techniques",
+        "start": 40472
+      },
+      {
+        "number": "5.4",
+        "title": "Importance of Plan-Guided Algorithm",
+        "start": 46465
+      },
+      {
+        "number": "5.5",
+        "title": "Scalability of Mozi to Other Fuzzers",
+        "start": 49551
+      },
+      {
+        "number": "6",
+        "title": "DISCUSSION",
+        "start": 53538
+      },
+      {
+        "number": "7",
+        "title": "RELATED WORK",
+        "start": 55686
+      },
+      {
+        "number": "8",
+        "title": "CONCLUSION",
+        "start": 62038
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2023. Null Value Differences in RDBMS. https://www .linkedin.com/pulse/nullvalue-differences-rdbms-sumit-sengupta. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2024. SQL Language Reference–ANSI Standards. https://docs .oracle.com/ en/database/oracle/oracle-database/19/sqlrf/ANSI-Standards .html#GUIDF51EA195-0669-4DED-9D81-B7205AAC642F/. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In Proceedings of International Conference on Software Engineering (ICSE).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "Adam Bannister. 2021. SQLite patches use-after-free bug that left apps open to code execution, denial-of-service exploits. https://portswigger .net/dailyswig/sqlite-patches-use-after-free-bug-that-left-apps-open-to-codeexecution-denial-of-service-exploits. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Tsong Y Chen, Shing C Cheung, and Shiu Ming Yiu. 2020. Metamorphic testing: a new approach for generating next test cases. arXiv preprint arXiv:2002.12543 (2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Tsong Yueh Chen, Fei-Ching Kuo, Huai Liu, Pak-Lok Poon, Dave Towey, TH Tse, and Zhi Quan Zhou. 2018. Metamorphic testing: A review of challenges and opportunities. ACM Computing Surveys (CSUR) 51, 1 (2018), 1–27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Yuanliang Chen, Yu Jiang, Fuchen Ma, Jie Liang, Mingzhe Wang, Chijin Zhou, Xun Jiao, and Zhuo Su. 2019. EnFuzz: Ensemble Fuzzing with Seed Synchronization among Diverse Fuzzers. In USENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Catalin Cimpanu. 2019. Google Chrome impacted by new Magellan 2.0 vulnerabilities. https://www .zdnet.com/article/google-chrome-impacted-by-newmagellan-2-0-vulnerabilities/. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "ClickHouse 2024. ClickHouse Website. https://clickhouse .com/. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "PostgreSQL Community. 2014. PostgreSQL Bug List. https://www .postgresql.org/ list/pgsql-bugs/. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Carlos Coronel and Steven Morris. 2019. Database systems: design, implementation and management. Cengage learning.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Ziyu Cui, Wensheng Dou, Qianwang Dai, Jiansen Song, Wei Wang, Jun Wei, and Dan Ye. 2022. Differentially Testing Database Transactions for Fun and Profit. In37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Chris J Date. 1989. A Guide to the SQL Standard. Addison-Wesley Longman Publishing Co., Inc.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Songyun Duan, Vamsidhar Thummala, and Shivnath Babu. 2009. Tuning database configuration parameters with ituned. Proceedings of the VLDB Endowment 2, 1 (2009), 1246–1257. ICSE ’24, April 14–20, 2024, Lisbon, Portugal Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "R Elmasri, SB Navathe, R Elmasri, and SB Navathe. 2015. Fundamentals of Database Systems. In Advances in Databases and Information Systems. Springer, 139.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free DBMS fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "James R Groff, Paul N Weinberg, and Andrew J Oppel. 2002. SQL: the complete reference. Vol. 2. McGraw-Hill/Osborne.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "The PostgreSQL Global Development Group. 2023. PostgreSQL 14.10 Documentation. https://www .postgresql.org/files/documentation/pdf/14/postgresql-14A4.pdf. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "The PostgreSQL Global Development Group. 2023. Query Planning. https: //www.postgresql.org/docs/current/runtime-config-query .html. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "The PostgreSQL Global Development Group. 2024. EXPLAIN Statement of PostgreSQL. https://www .postgresql.org/docs/current/sql-explain .html. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "The PostgreSQL Global Development Group. 2024. Using EXPLAIN. https: //www.postgresql.org/docs/current/using-explain .html. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Kenneth Hoste and Lieven Eeckhout. 2008. Cole: compiler optimization level exploration. In Proceedings of the 6th annual IEEE/ACM international symposium on Code generation and optimization. 165–174.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Matthias Jarke and Jurgen Koch. 1984. Query optimization in database systems. ACM Computing surveys (CsUR) 16, 2 (1984), 111–152.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2020. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB). Tokyo, Japan.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. ACM Sigplan Notices 49, 6 (2014), 216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Guoliang Li, Xuanhe Zhou, Shifu Li, and Bo Gao. 2019. Qtune: A query-aware database tuning system with deep reinforcement learning. Proceedings of the VLDB Endowment 12, 12 (2019), 2118–2130.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequenceoriented DBMS fuzzing. In Proceedings of IEEE International Conference on Data Engineering (ICDE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jie Liang, Yu Jiang, Yuanliang Chen, Mingzhe Wang, Chijin Zhou, and Jiaguang Sun. 2018. PAFL: Extend Fuzzing Optimizations of Single Mode to Industrial Parallel Mode.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of {DBMS}with Coverage-based Guidance. In 31st USENIX Security Symposium (USENIX Security 22). 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "LibFuzzer 2024. LibFuzzer. https://www .llvm.org/docs/LibFuzzer .html. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Huai Liu, Fei-Ching Kuo, Dave Towey, and Tsong Yueh Chen. 2013. How effectively does metamorphic testing alleviate the oracle problem? IEEE Transactions on Software Engineering 40, 1 (2013), 4–22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. In Proceedings of the 44th International Conference on Software Engineering. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "MariaDB 2024. MariaDB. https://mariadb .org/. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "William M McKeeman. 1998. Differential testing for software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Jim Melton. 1996. Sql language summary. Acm Computing Surveys (CSUR) 28, 1 (1996), 141–143.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Barton P. Miller, Louis Fredriksen, and Bryan So. 1990. An Empirical Study of the Reliability of UNIX Utilities. Commun. ACM 33, 12 (Dec. 1990).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "MySQL 2014. MySQL 8.0 Reference Manual, Switchable Optimizations. https: //dev.mysql.com/doc/refman/8 .0/en/switchable-optimizations .html. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "MySQL 2024. Understanding the Query Execution Plan. https://dev .mysql.com/ doc/refman/8 .0/en/execution-plan-information .html. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Oracle. 2014. MySQL Bug List. https://bugs .mysql.com/. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Oracle. 2024. EXPLAIN Statement of MySQL. https://dev .mysql.com/doc/refman/ 8.0/en/explain.html. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Oracle 2024. MySQL. https://www .mysql.com/. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "PostgreSQL 2024. PostgreSQL. https://www .postgresql.org/. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger. 2024. Bugs found in Database Management Systems. https: //www.manuelrigger .at/dbms-bugs. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger. 2024. SQLancer Website. https://github .com/sqlancer/sqlancer. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 46,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 48,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 49,
+      "text": "Sergio Segura, Gordon Fraser, Ana B Sanchez, and Antonio Ruiz-Cortés. 2016. A survey on metamorphic testing. IEEE Transactions on software engineering 42, 9 (2016), 805–824.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https://github .com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitriy Vyukov. 2012. AddressSanitizer: A Fast Address Sanity Checker. In 2012 USENIX Annual Technical Conference, Boston, MA, USA, June 13-15, 2012, Gernot Heiser and Wilson C. Hsieh (Eds.). USENIX Association, 309–318. https://www .usenix.org/ conference/atc12/technical-sessions/presentation/serebryany",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, New York, USA. Morgan Kaufmann, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. In Proceedings of International Conference on Software Engineering (ICSE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Dana Van Aken, Andrew Pavlo, Geoffrey J Gordon, and Bohan Zhang. 2017. Automatic database management system tuning through large-scale machine learning. In Proceedings of the 2017 ACM international conference on management of data. 1009–1024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Dana Van Aken, Dongsheng Yang, Sebastien Brillard, Ari Fiorino, Bohan Zhang, Christian Bilien, and Andrew Pavlo. 2021. An inquiry into machine learningbased automatic configuration tuning services on real-world database management systems. Proceedings of the VLDB Endowment 14, 7 (2021), 1241–1253.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Mingzhe Wang, Jie Liang, Chijin Zhou, Zhiyong Wu, Xinyi Xu, and Yu Jiang. 2022. Odin: on-demand instrumentation with on-the-fly recompilation. In Proceedings of the 43rd ACMSIGPLAN International Conference on Programming Language Design and Implementation. 1010–1024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In 43rd IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice, ICSE (SEIP) 2021, Madrid, Spain, May 25-28, 2021. IEEE, 328–337. https://doi .org/10.1109/ICSE-SEIP52600 .2021.00042",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Wikipedia 2024. databases. https://en .wikipedia.org/wiki/Database. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. 2022. Unicorn: detect runtime errors in time-series databases with hybrid input synthesis. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis. 251–262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Michał Zalewski. 2017. american fuzzy lop. http://lcamtuf .coredump.cx/afl/. Accessed: January 18, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Xinyi Zhang, Hong Wu, Yang Li, Jian Tan, Feifei Li, and Bin Cui. 2022. Towards dynamic and safe configuration tuning for cloud databases. In Proceedings of the 2022 International Conference on Management of Data. 631–645.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In Proceedings of International Conference on Software Engineering (ICSE).",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger. 2024. Bugs found in Database Management Systems. https: //www.manuelrigger .at/dbms-bugs. Accessed: January 18, 2024.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger. 2024. SQLancer Website. https://github .com/sqlancer/sqlancer. Accessed: January 18, 2024.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 46,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 48,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 4,8,\n10,40,44]",
+      "technique": null,
+      "sentence": "It is crucial to ensure the correctness, reliability, and performance of DBMSs [ 4,8, 10,40,44].",
+      "context_before": "testing and debugging. KEYWORDS DBMS Testing, Configuration, Test Oracle ACM Reference Format: Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation. In Software Engineering (ICSE ’24), April 14–20, 2024, Lisbon, Portugal. ACM, New York, NY, USA, 12 pages. https://doi .org/10.1145/3597503 .3639112 1 INTRODUCTION Database Management Systems (DBMSs) serve as the backbone for various applications, ranging from simple web applications to complex, large-scale enterprise systems [ 11,15,58].",
+      "context_after": "Metamorphic testing is a prevalent approach to testing DBMS, which focuses on building metamorphic relations [5, 6, 32, 46,47,49]. These relations represent mathematical relationships between the input and output of a software system, which should hold even when the input undergoes certain changes or transformations. One common way of building metamorphic relations for DBMSs is by creating different SQL queries with equivalent semantics and comparing their outputs. For instance, consider the SQL queries “SELECT 1+1” and “SELECT 2.” Although they have different syntactic representations, their",
+      "section": "1 INTRODUCTION",
+      "page": 1,
+      "char_offset": 3269,
+      "cited_reference": {
+        "number": 44,
+        "text": "Manuel Rigger. 2024. Bugs found in Database Management Systems. https: //www.manuelrigger .at/dbms-bugs. Accessed: January 18, 2024.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[5, 6, 32,\n46,47,49]",
+      "technique": "norec",
+      "sentence": "Metamorphic testing is a prevalent approach to testing DBMS, which focuses on building metamorphic relations [5, 6, 32, 46,47,49].",
+      "context_before": "ie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation. In Software Engineering (ICSE ’24), April 14–20, 2024, Lisbon, Portugal. ACM, New York, NY, USA, 12 pages. https://doi .org/10.1145/3597503 .3639112 1 INTRODUCTION Database Management Systems (DBMSs) serve as the backbone for various applications, ranging from simple web applications to complex, large-scale enterprise systems [ 11,15,58]. It is crucial to ensure the correctness, reliability, and performance of DBMSs [ 4,8, 10,40,44].",
+      "context_after": "These relations represent mathematical relationships between the input and output of a software system, which should hold even when the input undergoes certain changes or transformations. One common way of building metamorphic relations for DBMSs is by creating different SQL queries with equivalent semantics and comparing their outputs. For instance, consider the SQL queries “SELECT 1+1” and “SELECT 2.” Although they have different syntactic representations, their semantics are equivalent, as both yield the same result. To identify metamorphic relations, a precise comprehension of the SQL spe",
+      "section": "1 INTRODUCTION",
+      "page": 1,
+      "char_offset": 3366,
+      "cited_reference": {
+        "number": 46,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "In 24-hour experiments, Mozi covers 64,973, 54,464, 43,236, 28,499, and 14,084 more branches, detects 25, 22, 21, 24, and 26 more bugs than PQS [ 48], NoREC [ 46], TLP1[47],Apollo [25], and Amoeba [33], respectively.",
+      "context_before": "e and configure the DBMS environment to test different scenarios and configurations.We conducted extensive experiments to evaluate the effectiveness of Mozi in identifying previously undiscovered bugs on four widely-used DBMSs, namely MySQL [ 42], MariaDB [ 34], Clickhouse [ 9], and PostgreSQL [ 43]. In the continuous testing of threeweeks, Mozi found a total of 101 previously unknown bugs, including 49 correctness and 52 performance bugs in four DBMSs. Among them, 90 bugs are confirmed and 57 bugs have been fixed. We also evaluate Mozi against some related state-of-the-art testing techniques.",
+      "context_after": "Additionally, we assess the significance of the plan-guided approach in Mozi and demonstrate its scalability by applying Mozi to other fuzzers for detecting correctness and performance bugs. In summary, our paper makes the following contributions: •We present Mozi, a framework for identifying DBMS bugs using configuration-based equivalent transformations, which complements existing methods such as metamorphic testing. Mozi aims to generate different DBMS instantiations that should produce equivalent results when executing the same SQL query. By designing test oracles based on the result compa",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 9045,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "1PQS, NoREC, and TLP are three correctness test oracles utilized in SQLancer.",
+      "context_before": "e settings and parameters that can adjust the behavior of database management systems. These configuration options can affect the performance, reliability, and security of the database system. Table 1 shows common categories of DBMS configurations. Table 1: Categories of DBMS Configurations Categor y Description Buffer cache Size and behavior of the buffer cache. Concurrency Contr ol of concurrency and locking. Memory allocation Memor y allocation and usage. Disk usage Contr ol of disk space usage. Query optimization Optimization of query execution. Security A ccess control and authentication.",
+      "context_after": "Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation ICSE ’24, April 14–20, 2024, Lisbon, Portugal The buffer cache configuration determines the size and behavior of the buffer cache used by the DBMS to store frequently accessed data in memory. The concurrency configuration controls the degree of concurrency and locking used to manage access to shared data. Memory allocation configuration specifies how memory is allocated and used by the DBMS. Disk usage configuration controls how disk space is allocated and used by the DBMS. The optimization configuration refers to t",
+      "section": "2 BACKGROUND ANDMOTIVATION",
+      "page": 2,
+      "char_offset": 11414,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, the NoREC [ 46] method of SQLancer detects the logic bugs of DBMS optimizer by constructing equivalent non-optimization SQL query which can not be optimized by DBMSs based on the existing SQL query.",
+      "context_before": "od to construct oracles to detect correctness bugs of DBMSs. A metamorphic relation is a mathematical relationship between the input change and output change of a software system. When the input changes, and the relationship no longer holds for the corresponding output change, it indicates the presence of an error. A common method is like EMI [ 26] to construct different, but equivalent variants of a query. More concretely, given a query 𝑄 and its semantic equivalent query collection 𝐶. The collection 𝐶 can be used to perform a comparison for any DBMS 𝐷: if𝐷(𝑄)≠ 𝐷(𝑄′)for some𝑄′∈𝐶, D has a bug.",
+      "context_after": "If the non-optimization SQL query gets inconsistent results from the existing SQL query, they detect a logic bug of the optimizer. However, finding metamorphic relationships for DBMSs is very difficult. Discovering an effective metamorphic relation needs a deep understanding of SQL specification but one relationship can usually focus on only one SQL feature. For instance, the TLP [ 47] method can only be tested for queries that use WHERE ,HAVING, GROUP BY, aggregation functions, and DISTINCT features. Besides, the relation may be highly customized for a specific DBMS, which means that adaptin",
+      "section": "2 BACKGROUND ANDMOTIVATION",
+      "page": 3,
+      "char_offset": 13123,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For instance, the TLP [ 47] method can only be tested for queries that use WHERE ,HAVING, GROUP BY, aggregation functions, and DISTINCT features.",
+      "context_before": ")for some𝑄′∈𝐶, D has a bug. For example, the NoREC [ 46] method of SQLancer detects the logic bugs of DBMS optimizer by constructing equivalent non-optimization SQL query which can not be optimized by DBMSs based on the existing SQL query. If the non-optimization SQL query gets inconsistent results from the existing SQL query, they detect a logic bug of the optimizer. However, finding metamorphic relationships for DBMSs is very difficult. Discovering an effective metamorphic relation needs a deep understanding of SQL specification but one relationship can usually focus on only one SQL feature.",
+      "context_after": "Besides, the relation may be highly customized for a specific DBMS, which means that adapting it to other DBMSs can be costly. Basic Idea of Mozi. The key idea of Mozi is to change DBMS itself to an equivalent one as a reference to verify the execution results. Specifically, rather than finding a mathematical relationship to build a different query, we choose to transform the DBMS into an equivalent one by changing configurations. Based on the transformation process and query execution results, we can build test oracles to find bugs. For example, when two equivalent DBMSs have different resul",
+      "section": "2 BACKGROUND ANDMOTIVATION",
+      "page": 3,
+      "char_offset": 13696,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We also used SQLancer [ 45] (with test oracles PQS [ 48], TLP [ 47], and NoREC [ 46]), Apollo [25], and Amoeba [33] to test these DBMSs, but they can only find a subset of these bugs (shown in Section 5.",
+      "context_before": "128 cores, and 504 GiB of main memory. All DBMSs were tested using docker containers that were downloaded directly from their websites. Each docker container is allocated with 5 CPU cores and 40 GiB of RAM. Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation ICSE ’24, April 14–20, 2024, Lisbon, Portugal 5.2 DBMS Vulnerability Detection Bug Statistics. We applied Mozi to test performance and correctness bugs in three weeks. The four evaluated DBMSs are widely used and well-tested. Nevertheless, Mozi performed well and finally found 101 previously unknown bugs in them.",
+      "context_after": "3). Table 2 shows the statistics of the bugs. Mozi reported a total of 49 correctness bugs and 52 performance bugs. We reported all the bugs to the corresponding DBMS developers. Among them, 46 correctness bugs and 44 performance bugs have been confirmed as previously unknown bugs. At the time of writing this paper, 57 bugs have been fixed, and we have received gratitude from the developers. In addition, 11 bugs are still being validated by the developers due to the complexity of DBMS. Table 2: Number of reported and confirmed bugs by Mozi in three weeks. Corr ectness Performance DBMS Rep orte",
+      "section": "5.2 DBMS Vulnerability Detection",
+      "page": 7,
+      "char_offset": 34858,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, SQLancer and Amoeba detect the bugs by constructing the equivalent SQL query, which does not change the configuration of DBMS.",
+      "context_before": "anning some records from table t1, which triggers the index_scan optimization of PostgreSQL. As shown, PostgreSQL takes 11.42 seconds to execute the SQL query. However, after disabling the index_scan optimization in Step 5, PostgreSQL only takes 0.534 seconds to execute the same SQL query. The performance decrease caused by using the index_scan optimization is significant, leading to a 20x slowdown compared to not using the optimization in execution. Why was the bug only discovered by Mozi? The bug involves DBMS configuration changes, which can hardly be detected by other DBMS testing methods.",
+      "context_after": "Therefore, this bug can not be discovered by them. Similarly, Apollo detects the regression performance bugs of DBMSs by comparing the response time of two different-version DBMSs with the same configuration, thus it cannot detect this kind of bug. In contrast, Mozi detected this bug ICSE ’24, April 14–20, 2024, Lisbon, Portugal Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang by constructing a test oracle for performance bug detection with configuration-based equivalent transformation. In summary, the results indicate that Mozi can detect previously unknown corre",
+      "section": "5.2 DBMS Vulnerability Detection",
+      "page": 7,
+      "char_offset": 39672,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, we compared Mozi𝑐𝑜𝑟against SQLancer using three logic test oracles, namely PQS [ 48], NoREC [ 46], and TLP [ 47] for correctness bug detection.",
+      "context_before": "acle for performance bug detection with configuration-based equivalent transformation. In summary, the results indicate that Mozi can detect previously unknown correctness and performance bugs, which adequately answers RQ1. 5.3 Comparison with Other Techniques We implement Mozi with correctness and performance test oracles into Mozi𝑐𝑜𝑟and Mozi𝑝𝑒𝑟to detect correctness bugs and performance bugs, respectively. Compared Techniques and Setup. To evaluate the effectiveness of our framework, Mozi, we conducted a comparison study between Mozi𝑐𝑜𝑟andMozi𝑝𝑒𝑟with the state-of-the-art DBMS testing methods.",
+      "context_after": "In the following text, we will refer to these three implementations of SQLancer by their respective oracle names. We also compared Mozi𝑝𝑒𝑟with Apollo [25] and Amoeba [33] for performance bug detection. Currently, Amoeba only supports PostgreSQL. We ran the testing tools on each DBMS for 24 hours and recorded the number of triggered bugs. All bugs were reported to the developers, and we used the confirmed number of bugs as the result. To ensure a fair comparison, we collected the generated queries from each testing method and dry-ran the queries to ensure uniform branch coverage. Results. Mozi",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 40848,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In the following text, we will refer to these three implementations of SQLancer by their respective oracle names.",
+      "context_before": "n correctness and performance bugs, which adequately answers RQ1. 5.3 Comparison with Other Techniques We implement Mozi with correctness and performance test oracles into Mozi𝑐𝑜𝑟and Mozi𝑝𝑒𝑟to detect correctness bugs and performance bugs, respectively. Compared Techniques and Setup. To evaluate the effectiveness of our framework, Mozi, we conducted a comparison study between Mozi𝑐𝑜𝑟andMozi𝑝𝑒𝑟with the state-of-the-art DBMS testing methods. Specifically, we compared Mozi𝑐𝑜𝑟against SQLancer using three logic test oracles, namely PQS [ 48], NoREC [ 46], and TLP [ 47] for correctness bug detection.",
+      "context_after": "We also compared Mozi𝑝𝑒𝑟with Apollo [25] and Amoeba [33] for performance bug detection. Currently, Amoeba only supports PostgreSQL. We ran the testing tools on each DBMS for 24 hours and recorded the number of triggered bugs. All bugs were reported to the developers, and we used the confirmed number of bugs as the result. To ensure a fair comparison, we collected the generated queries from each testing method and dry-ran the queries to ensure uniform branch coverage. Results. Mozi outperforms other testing methods in detecting both performance and correctness bugs. Table 3 displays the number",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 41006,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Compared to PQS, NoREC, and TLP, Mozi𝑐𝑜𝑟 found 25, 22, and 21 more bugs, respectively.",
+      "context_before": "ction. Currently, Amoeba only supports PostgreSQL. We ran the testing tools on each DBMS for 24 hours and recorded the number of triggered bugs. All bugs were reported to the developers, and we used the confirmed number of bugs as the result. To ensure a fair comparison, we collected the generated queries from each testing method and dry-ran the queries to ensure uniform branch coverage. Results. Mozi outperforms other testing methods in detecting both performance and correctness bugs. Table 3 displays the number of bugs detected by each method. In 24 hours, Mozi𝑐𝑜𝑟detected a total of 29 bugs.",
+      "context_after": "For performance bugs, Mozi𝑝𝑒𝑟discovered 27 bugs, while Apollo andAmoeba detected 3 and 1 bugs on the tested DBMSs, respectively. We investigated the bugs and found that the bugs identified by Mozi did not overlap with other techniques. This is because Mozi detects problems through the transformation of DBMS configurations, which is a process not employed by the other methods. On the other hand, Mozi did not find the bugs detected by other techniques. This disparity arises because PQS, NoREC, and TLP employ highly customized test oracles, with the latter two incorporating statement changes not",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 41802,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "This disparity arises because PQS, NoREC, and TLP employ highly customized test oracles, with the latter two incorporating statement changes not utilized in Mozi𝑐𝑜𝑟.",
+      "context_before": "h method. In 24 hours, Mozi𝑐𝑜𝑟detected a total of 29 bugs. Compared to PQS, NoREC, and TLP, Mozi𝑐𝑜𝑟 found 25, 22, and 21 more bugs, respectively. For performance bugs, Mozi𝑝𝑒𝑟discovered 27 bugs, while Apollo andAmoeba detected 3 and 1 bugs on the tested DBMSs, respectively. We investigated the bugs and found that the bugs identified by Mozi did not overlap with other techniques. This is because Mozi detects problems through the transformation of DBMS configurations, which is a process not employed by the other methods. On the other hand, Mozi did not find the bugs detected by other techniques.",
+      "context_after": "The test oracles of Amoeba and Apollo for identifying performance bugs differ from Mozi𝑝𝑒𝑟. The first one necessitates constructing an equivalent statement, while the other requires referencing the previous version. In contrast, Mozi utilizes DBMS transformation, resulting in a low likelihood of identifying common bugs. Table 3: The number of detected bugs by each DBMS testing method in 24 hours. Corr ectness Performance DBMSPQS NoREC TLP Mozi𝑐 𝑜𝑟Apollo Amoeba Mozi𝑝 𝑒𝑟 MySQL 1 2 2 9 1 – 11 MariaDB 2 3 2 10 0 – 13 Clickhouse 0 1 3 6 1 – 2 Postgr eSQL 1 1 1 4 1 1 1 T otal 4 7 8 29 3 1 27 Impr o",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 42344,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Corr ectness Performance DBMSPQS NoREC TLP Mozi𝑐 𝑜𝑟Apollo Amoeba Mozi𝑝 𝑒𝑟 MySQL 1 2 2 9 1 – 11 MariaDB 2 3 2 10 0 – 13 Clickhouse 0 1 3 6 1 – 2 Postgr eSQL 1 1 1 4 1 1 1 T otal 4 7 8 29 3 1 27 Impr ovement 25↑ 22↑ 21↑ - 24↑ 26↑ -Table 4 shows the number of branches covered by each method in 24 hours.",
+      "context_before": "bugs detected by other techniques. This disparity arises because PQS, NoREC, and TLP employ highly customized test oracles, with the latter two incorporating statement changes not utilized in Mozi𝑐𝑜𝑟. The test oracles of Amoeba and Apollo for identifying performance bugs differ from Mozi𝑝𝑒𝑟. The first one necessitates constructing an equivalent statement, while the other requires referencing the previous version. In contrast, Mozi utilizes DBMS transformation, resulting in a low likelihood of identifying common bugs. Table 3: The number of detected bugs by each DBMS testing method in 24 hours.",
+      "context_after": "It shows that Mozi covers more branches in 24 hours when compared to other testing methods. Specifically, compared to three correctness bug testing methods, Mozi𝑐𝑜𝑟covered a total of 64,973, 54,464, and 43,236 more branches than PQS, NoREC, and TLP, respectively. And compared to the performance bug testing fuzzer Apollo, Mozi𝑝𝑒𝑟covered a total of 28,499 more branches in MySQL, MariaDB, Clickhouse, and PostgreSQL. Compared to Amoeba, Mozi covered 14,084 more branches in PostgreSQL. Table 4: The number of covered branches by each DBMS testing method in 24 hours. Corr ectness Performance DBMSPQS",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 42910,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Specifically, compared to three correctness bug testing methods, Mozi𝑐𝑜𝑟covered a total of 64,973, 54,464, and 43,236 more branches than PQS, NoREC, and TLP, respectively.",
+      "context_before": "the previous version. In contrast, Mozi utilizes DBMS transformation, resulting in a low likelihood of identifying common bugs. Table 3: The number of detected bugs by each DBMS testing method in 24 hours. Corr ectness Performance DBMSPQS NoREC TLP Mozi𝑐 𝑜𝑟Apollo Amoeba Mozi𝑝 𝑒𝑟 MySQL 1 2 2 9 1 – 11 MariaDB 2 3 2 10 0 – 13 Clickhouse 0 1 3 6 1 – 2 Postgr eSQL 1 1 1 4 1 1 1 T otal 4 7 8 29 3 1 27 Impr ovement 25↑ 22↑ 21↑ - 24↑ 26↑ -Table 4 shows the number of branches covered by each method in 24 hours. It shows that Mozi covers more branches in 24 hours when compared to other testing methods.",
+      "context_after": "And compared to the performance bug testing fuzzer Apollo, Mozi𝑝𝑒𝑟covered a total of 28,499 more branches in MySQL, MariaDB, Clickhouse, and PostgreSQL. Compared to Amoeba, Mozi covered 14,084 more branches in PostgreSQL. Table 4: The number of covered branches by each DBMS testing method in 24 hours. Corr ectness Performance DBMSPQS NoREC TLP Mozi𝑐 𝑜𝑟Apollo Amoeba Mozi𝑝 𝑒𝑟 MySQL 50,294 52,984 60,274 75,293 70,923 – 74,836 MariaDB 40,154 42,938 45,792 54,982 46,365 – 55,387 Clickhouse 65,980 69,842 69,928 79,852 72,388 – 80,142 Postgr eSQL 59,811 60,984 61,982 71,085 63,748 56,844 70,928 T ot",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 43304,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Corr ectness Performance DBMSPQS NoREC TLP Mozi𝑐 𝑜𝑟Apollo Amoeba Mozi𝑝 𝑒𝑟 MySQL 50,294 52,984 60,274 75,293 70,923 – 74,836 MariaDB 40,154 42,938 45,792 54,982 46,365 – 55,387 Clickhouse 65,980 69,842 69,928 79,852 72,388 – 80,142 Postgr eSQL 59,811 60,984 61,982 71,085 63,748 56,844 70,928 T otal 216,239 226,748 237,976 281,212 252,794 56,844 281,293 Impr ovement 64,973↑54,464↑43,236↑ - 28,499↑14,084↑**The improvement is only for PostgreSQL since Amoeba only supports it among four DBMSs at the time of writing.",
+      "context_before": "vered by each method in 24 hours. It shows that Mozi covers more branches in 24 hours when compared to other testing methods. Specifically, compared to three correctness bug testing methods, Mozi𝑐𝑜𝑟covered a total of 64,973, 54,464, and 43,236 more branches than PQS, NoREC, and TLP, respectively. And compared to the performance bug testing fuzzer Apollo, Mozi𝑝𝑒𝑟covered a total of 28,499 more branches in MySQL, MariaDB, Clickhouse, and PostgreSQL. Compared to Amoeba, Mozi covered 14,084 more branches in PostgreSQL. Table 4: The number of covered branches by each DBMS testing method in 24 hours.",
+      "context_after": "The main reason for Mozi’s improvement in bug detection and branch coverage is its independence from specific SQL features in test oracle. This flexibility enables Mozi to test a broader range of functionalities within the DBMS. Specifically, PQS, NoREC, TLP, andAmoeba are constructed based on specific features to create test cases, which limits the SQL grammar they can support. For example, NoREC detects logic bugs in the DBMS optimizer by constructing equivalent optimized and unoptimized queries based on the optimizer rules. However, this approach only covers SQL grammar that conforms to th",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 43779,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Specifically, PQS, NoREC, TLP, andAmoeba are constructed based on specific features to create test cases, which limits the SQL grammar they can support.",
+      "context_before": "45,792 54,982 46,365 – 55,387 Clickhouse 65,980 69,842 69,928 79,852 72,388 – 80,142 Postgr eSQL 59,811 60,984 61,982 71,085 63,748 56,844 70,928 T otal 216,239 226,748 237,976 281,212 252,794 56,844 281,293 Impr ovement 64,973↑54,464↑43,236↑ - 28,499↑14,084↑**The improvement is only for PostgreSQL since Amoeba only supports it among four DBMSs at the time of writing. The main reason for Mozi’s improvement in bug detection and branch coverage is its independence from specific SQL features in test oracle. This flexibility enables Mozi to test a broader range of functionalities within the DBMS.",
+      "context_after": "For example, NoREC detects logic bugs in the DBMS optimizer by constructing equivalent optimized and unoptimized queries based on the optimizer rules. However, this approach only covers SQL grammar that conforms to the optimizer rules. In contrast, Mozi generates SQL queries without any limitations on the SQL grammar since it relies on the DBMS itself to evaluate query results. As a result, Mozi can support more SQL grammar and cover more branches compared to PQS, NoREC, and TLP, which explains why it can detect more correctness bugs in these DBMSs. Apollo generates queries by comparing the q",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 44525,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For example, NoREC detects logic bugs in the DBMS optimizer by constructing equivalent optimized and unoptimized queries based on the optimizer rules.",
+      "context_before": "216,239 226,748 237,976 281,212 252,794 56,844 281,293 Impr ovement 64,973↑54,464↑43,236↑ - 28,499↑14,084↑**The improvement is only for PostgreSQL since Amoeba only supports it among four DBMSs at the time of writing. The main reason for Mozi’s improvement in bug detection and branch coverage is its independence from specific SQL features in test oracle. This flexibility enables Mozi to test a broader range of functionalities within the DBMS. Specifically, PQS, NoREC, TLP, andAmoeba are constructed based on specific features to create test cases, which limits the SQL grammar they can support.",
+      "context_after": "However, this approach only covers SQL grammar that conforms to the optimizer rules. In contrast, Mozi generates SQL queries without any limitations on the SQL grammar since it relies on the DBMS itself to evaluate query results. As a result, Mozi can support more SQL grammar and cover more branches compared to PQS, NoREC, and TLP, which explains why it can detect more correctness bugs in these DBMSs. Apollo generates queries by comparing the query performance of two versions of the same DBMS, which is effective for detecting performance regression bugs. However, Apollo is limited by utilizin",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 44678,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "As a result, Mozi can support more SQL grammar and cover more branches compared to PQS, NoREC, and TLP, which explains why it can detect more correctness bugs in these DBMSs.",
+      "context_before": "s Mozi to test a broader range of functionalities within the DBMS. Specifically, PQS, NoREC, TLP, andAmoeba are constructed based on specific features to create test cases, which limits the SQL grammar they can support. For example, NoREC detects logic bugs in the DBMS optimizer by constructing equivalent optimized and unoptimized queries based on the optimizer rules. However, this approach only covers SQL grammar that conforms to the optimizer rules. In contrast, Mozi generates SQL queries without any limitations on the SQL grammar since it relies on the DBMS itself to evaluate query results.",
+      "context_after": "Apollo generates queries by comparing the query performance of two versions of the same DBMS, which is effective for detecting performance regression bugs. However, Apollo is limited by utilizing differential testing on current and previous DBMS versions. The bugs that exist in both versions will not be detected. In addition, this approach relies heavily on the differences between the two versions, which can limit the diversity of its generated queries. If the differences between versions are minimal, Apollo may not be able to generate a sufficient number of unique queries to fully test the d",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 8,
+      "char_offset": 45059,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "citation_marker",
+      "surface": "[ 3,16,24,28,30,50,57,59,62]",
+      "technique": "qpg",
+      "sentence": "Most fuzzers [ 3,16,24,28,30,50,57,59,62] focus on generating complex and valid SQL test cases to find the memory safety bugs.",
+      "context_before": "ontrols and permissions. By generating alternative DBMSs with varying security settings, we can thoroughly examine whether certain accesses violate the intended restrictions. 7 RELATED WORK In this section, we will first introduce DBMS Fuzzing, and then focus on works related to DBMS test oracle construction and configuration settings. DBMS Fuzzing. Fuzzing [ 7,29,31,37,60] has been applied to DBMSs in recent years and has found hundreds of bugs. These DBMS fuzzers generate SQL test cases continuously, feed them to the target DBMS, and check whether the DBMS works normally when executing them.",
+      "context_after": "They could be roughly divided into generationbased and mutation-based fuzzers. Generation-based fuzzers generate SQL queries based on predefined SQL generation models. SQLsmith [ 50] generates SELECT statements with complex structures according to its predefined AST models. Amoeba [33] leverages domain-specific design languages to generate SQL queries from scratch based on the schema of the database. Mutation-based fuzzers mutate existing SQL queries to produce new queries, which are always guided by coverage information. Sqirrel [62] designs an intermediate representation (IR) to mutate SQL",
+      "section": "7 RELATED WORK",
+      "page": 10,
+      "char_offset": 56111,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In Proceedings of International Conference on Software Engineering (ICSE).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "QPG [ 3] in SQLancer introduces a mutation technique to use DDL (e.",
+      "context_before": "which are always guided by coverage information. Sqirrel [62] designs an intermediate representation (IR) to mutate SQL test cases with code coverage guidance. DynSQL [ 24] establishes a mapping model between binary files and SQL query syntax trees. Lego [28] generates SQL statement sequences with abundant types by exploring SQL statements of different types and analyzing type affinities. Griffin[16] proposes metadata graphs of SQL statements to provide a grammar-free way for mutation of SQL test cases. Unicorn [59] utilizes hybrid input synthesis to generate queries with time-series elements.",
+      "context_after": "g., CREATE) and DML (e.g., INSERT) statements to change the database state. It saves those mutated cases that cover unseen query plans to explore more behaviors. SQLRight [ 30] separates SELECT and other statements for mutation, and finally concatenates them together. Unlike these works that generate complex SQL queries to test database management systems, Mozi takes a different approach. Instead, it is a framework that applies equivalent transformations to DBMSs. By using this technique, Mozi is able to test both correctness and performance bugs. Nevertheless, The SQL generation technique and",
+      "section": "7 RELATED WORK",
+      "page": 10,
+      "char_offset": 57318,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "citation_marker",
+      "surface": "[ 30,33,46–48,53]",
+      "technique": "norec",
+      "sentence": "Many works [ 30,33,46–48,53] detect errors in a DBMS by verifying whether the execution results of SQL statements conform to predefined rules.",
+      "context_before": "queries to test database management systems, Mozi takes a different approach. Instead, it is a framework that applies equivalent transformations to DBMSs. By using this technique, Mozi is able to test both correctness and performance bugs. Nevertheless, The SQL generation technique and DBMS equivalent transformation are orthogonal. As Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation ICSE ’24, April 14–20, 2024, Lisbon, Portugal a result, it is possible to adapt the SQL generation technique into Mozi to combine the advantages of both techniques. Metamorphic Testing.",
+      "context_after": "Amoeba [33] aims to find performance bugs in DBMS by constructing semantically equivalent query pairs and comparing the response times when the DBMS executes them. NOREC [ 46] and TLP [ 47] are two test oracles used in SQLancer. NoREC converts a SQL query to one that cannot be optimized by optimizers and compares the two’s execution results to find optimizer bugs. TLP employs the concept of partitioning, wherein a problem is identified by partitioning the original query into multiple more complex queries. The approach involves comparing the results of the original query with the composed resu",
+      "section": "7 RELATED WORK",
+      "page": 11,
+      "char_offset": 58300,
+      "cited_reference": {
+        "number": 46,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "NOREC [ 46] and TLP [ 47] are two test oracles used in SQLancer.",
+      "context_before": "sformation are orthogonal. As Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation ICSE ’24, April 14–20, 2024, Lisbon, Portugal a result, it is possible to adapt the SQL generation technique into Mozi to combine the advantages of both techniques. Metamorphic Testing. Many works [ 30,33,46–48,53] detect errors in a DBMS by verifying whether the execution results of SQL statements conform to predefined rules. Amoeba [33] aims to find performance bugs in DBMS by constructing semantically equivalent query pairs and comparing the response times when the DBMS executes them.",
+      "context_after": "NoREC converts a SQL query to one that cannot be optimized by optimizers and compares the two’s execution results to find optimizer bugs. TLP employs the concept of partitioning, wherein a problem is identified by partitioning the original query into multiple more complex queries. The approach involves comparing the results of the original query with the composed result of the partitioned queries to determine their consistency. DQE [ 53] constructs SQL statements that should access the same rows and verifies whether the rows fetched by these statements are the same during execution. The EMI (",
+      "section": "7 RELATED WORK",
+      "page": 11,
+      "char_offset": 58607,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC converts a SQL query to one that cannot be optimized by optimizers and compares the two’s execution results to find optimizer bugs.",
+      "context_before": "figuration-Based Equivalent Transformation ICSE ’24, April 14–20, 2024, Lisbon, Portugal a result, it is possible to adapt the SQL generation technique into Mozi to combine the advantages of both techniques. Metamorphic Testing. Many works [ 30,33,46–48,53] detect errors in a DBMS by verifying whether the execution results of SQL statements conform to predefined rules. Amoeba [33] aims to find performance bugs in DBMS by constructing semantically equivalent query pairs and comparing the response times when the DBMS executes them. NOREC [ 46] and TLP [ 47] are two test oracles used in SQLancer.",
+      "context_after": "TLP employs the concept of partitioning, wherein a problem is identified by partitioning the original query into multiple more complex queries. The approach involves comparing the results of the original query with the composed result of the partitioned queries to determine their consistency. DQE [ 53] constructs SQL statements that should access the same rows and verifies whether the rows fetched by these statements are the same during execution. The EMI (Equivalence Modulo Inputs) [ 26] can be seen as a special case of metamorphic testing. It aims to reduce the complexity of validating the",
+      "section": "7 RELATED WORK",
+      "page": 11,
+      "char_offset": 58672,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M24",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP employs the concept of partitioning, wherein a problem is identified by partitioning the original query into multiple more complex queries.",
+      "context_before": "ion technique into Mozi to combine the advantages of both techniques. Metamorphic Testing. Many works [ 30,33,46–48,53] detect errors in a DBMS by verifying whether the execution results of SQL statements conform to predefined rules. Amoeba [33] aims to find performance bugs in DBMS by constructing semantically equivalent query pairs and comparing the response times when the DBMS executes them. NOREC [ 46] and TLP [ 47] are two test oracles used in SQLancer. NoREC converts a SQL query to one that cannot be optimized by optimizers and compares the two’s execution results to find optimizer bugs.",
+      "context_after": "The approach involves comparing the results of the original query with the composed result of the partitioned queries to determine their consistency. DQE [ 53] constructs SQL statements that should access the same rows and verifies whether the rows fetched by these statements are the same during execution. The EMI (Equivalence Modulo Inputs) [ 26] can be seen as a special case of metamorphic testing. It aims to reduce the complexity of validating the correctness of compilers. It mainly focuses on generating the EMI variants (i.e., the equivalent programs under some test inputs) by randomly de",
+      "section": "7 RELATED WORK",
+      "page": 11,
+      "char_offset": 58810,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M3",
+        "M9",
+        "M14"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:14:29Z",
+    "is_model_written": true,
+    "summary": "Mozi finds DBMS bugs by comparing one system against itself under different configurations rather than by comparing semantically equivalent queries. Its motivation is that metamorphic testing needs a precise grasp of the SQL specification to build equivalent inputs, and the specification is vague and intricate enough to make modelling query semantics hard. Mozi analyses the query plan, changes configuration to produce an equivalent system, re-executes and compares: different results indicate correctness bugs, faster execution with optimisations closed indicates performance bugs. It found 101 previously unknown bugs across four DBMSs.",
+    "narrative": "SQLancer's three oracles are the point of contrast and the baseline: Mozi argues that TLP applies only to queries using particular clauses and that NoREC's equivalent-query construction is limited by how well SQL semantics can be modelled, then runs PQS, NoREC and TLP against itself and reports 21 to 25 more bugs than each.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "result_comparison",
+      "M4": "definition",
+      "M5": "definition",
+      "M6": "motivation",
+      "M7": "baseline",
+      "M8": "background",
+      "M9": "baseline",
+      "M10": "baseline",
+      "M11": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is run as a baseline with its three oracles; nothing shown says Mozi is built on it."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Configuration-based equivalent transformation replaces the equivalent-query idea rather than extending it."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M7",
+          "M9",
+          "M11"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M7",
+            "sentence": "We also used SQLancer [ 45] (with test oracles PQS [ 48], TLP [ 47], and NoREC [ 46]), Apollo [25], and Amoeba [33] to test these DBMSs, but they can only find a subset of these bugs (shown in Section 5.",
+            "section": "5.2 DBMS Vulnerability Detection",
+            "page": 7
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "Specifically, we compared Mozi𝑐𝑜𝑟against SQLancer using three logic test oracles, namely PQS [ 48], NoREC [ 46], and TLP [ 47] for correctness bug detection.",
+            "section": "5.3 Comparison with Other Techniques",
+            "page": 8
+          },
+          {
+            "mention_id": "M11",
+            "sentence": "Compared to PQS, NoREC, and TLP, Mozi𝑐𝑜𝑟 found 25, 22, and 21 more bugs, respectively.",
+            "section": "5.3 Comparison with Other Techniques",
+            "page": 8
+          }
+        ],
+        "reasoning": "Mozi is compared against SQLancer running PQS, NoREC and TLP, with branch coverage and bug counts reported against each.",
+        "techniques": [
+          "pqs",
+          "norec",
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The oracles are described by their limitations rather than as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3597503_3639200.json b/_data/papers/paper_doi_10_1145_3597503_3639200.json
new file mode 100644
index 0000000..5d13585
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3597503_3639200.json
@@ -0,0 +1,961 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:32:39Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3639200",
+    "title": "Testing Graph Database Systems via Equivalent Query Rewriting",
+    "authors": [
+      "Qiuyang Mang",
+      "Aoyang Fang",
+      "Boxi Yu",
+      "Hanfei Chen",
+      "Pinjia He"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3639200",
+    "arxiv_id": null,
+    "s2_paper_id": "29641e542dc3bcf96e4a8e6902b9e132a9c399da",
+    "url": "https://doi.org/10.1145/3597503.3639200",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3597503.3639200",
+      "retrieved_at": "2026-09-09T01:32:39Z",
+      "chars": 69243,
+      "content_sha256": "sha256:f71d2a4b915e89d0470f903ce0330419092a1cf6ef699618470f07e584106e81"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 3236
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND",
+        "start": 10918
+      },
+      {
+        "number": "2.1",
+        "title": "Labeled Property Graph in GDBMS",
+        "start": 10931
+      },
+      {
+        "number": "2.2",
+        "title": "Query Languages in GDBMS",
+        "start": 12046
+      },
+      {
+        "number": "2.3",
+        "title": "Query Rewriting",
+        "start": 12970
+      },
+      {
+        "number": "3",
+        "title": "APPROACH",
+        "start": 15482
+      },
+      {
+        "number": "3.2",
+        "title": "ASG Parsing",
+        "start": 19078
+      },
+      {
+        "number": "3.3",
+        "title": "Random Walk Covering",
+        "start": 20186
+      },
+      {
+        "number": "4",
+        "title": "IMPLEMENTATION",
+        "start": 25294
+      },
+      {
+        "number": "4.1",
+        "title": "Implementation Overview",
+        "start": 25311
+      },
+      {
+        "number": "4.2",
+        "title": "Graph Generation",
+        "start": 26680
+      },
+      {
+        "number": "4.3",
+        "title": "Base Query Generator",
+        "start": 27336
+      },
+      {
+        "number": "4.4",
+        "title": "Equivalent Query Rewriter and Validation",
+        "start": 28620
+      },
+      {
+        "number": "5",
+        "title": "EVALUATION",
+        "start": 31043
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Setting",
+        "start": 31866
+      },
+      {
+        "number": "5.2",
+        "title": "Detecting Previously Unknown Bugs",
+        "start": 33235
+      },
+      {
+        "number": "5.3",
+        "title": "Test Cases Analysis",
+        "start": 40910
+      },
+      {
+        "number": "5.4",
+        "title": "Analytical Comparison with Existing",
+        "start": 45490
+      },
+      {
+        "number": "5.5",
+        "title": "Empirical Comparison with Existing",
+        "start": 49707
+      },
+      {
+        "number": "6",
+        "title": "DISCUSSION",
+        "start": 53821
+      },
+      {
+        "number": "6.1",
+        "title": "Threats to Validity",
+        "start": 53834
+      },
+      {
+        "number": "7",
+        "title": "RELATED WORK",
+        "start": 55982
+      },
+      {
+        "number": "8",
+        "title": "CONCLUSION",
+        "start": 58387
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Muhammad Hilmi Asyrofi, Zhou Yang, and David Lo. 2021. Crossasr++: A modular differential testing framework for automatic speech recognition. In Proceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1575– 1579.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In (ICSE). 2060–2071. https://doi.org/10.1109/ICSE48619.2023.00174",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Nathan Bronson, Zach Amsden, George Cabrera, Prasad Chakka, Peter Dimov, Hui Ding, Jack Ferris, Anthony Giardullo, Sachin Kulkarni, Harry Li, Mark 9https://github.com/CUHK-Shenzhen-SE/GRev ICSE 2024, April 2024, Lisbon, Portugal Trovato and Tobin, et al. Marchukov, Dmitri Petrov, Lovro Puzar, Yee Jiun Song, and Venkat Venkataramani. 2013. TAO: Facebook’s Distributed Data Store for the Social Graph",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Yuting Chen, Ting Su, and Zhendong Su. 2019. Deep differential testing of JVM implementations. In Engineering (ICSE). IEEE, 1257–1268.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Yuting Chen, Ting Su, Chengnian Sun, Zhendong Su, and Jianjun Zhao. 2016. Coverage-directed differential testing of JVM implementations. In proceedings of the 37th ACMSIGPLAN Conference on Programming Language Design and Implementation. 85–99.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Dbrank. 2023. DB-Engines Ranking of Graph DBMS. https://db-engines.com/en/ ranking/graph+ dbms",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Haoran Peng, Chenyuan Yang, and Lingming Zhang. 2022. Fuzzing deep-learning libraries via large language models. arXiv preprint arXiv:2212.14834 (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Chenyuan Yang, Shizhuo Dylan Zhang, Shujing Yang, and Lingming Zhang. 2023. Large language models are edge-case fuzzers: Testing deep learning libraries via fuzzgpt. arXiv preprint arXiv:2304.02014 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Yinlin Deng, Chenyuan Yang, Anjiang Wei, and Lingming Zhang. 2022. Fuzzing deep-learning libraries via automated relational api inference. In Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 44–56.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Robert B Evans and Alberto Savoia. 2007. Differential testing: a new approach to change detection. In The 6th Joint Meeting on European software engineering conference and the ACMSIGSOFT Symposium on the Foundations of Software Engineering: Companion Papers. 549–552.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Alex Groce, Gerard Holzmann, and Rajeev Joshi. 2007. Randomized differential testing as a prelude to formal verification. In 29th International Conference on Software Engineering (ICSE’07). IEEE, 621–631.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Pinjia He, Clara Meister, and Zhendong Su. 2020. Structure-invariant testing for machine translation. In Engineering (ICSE). IEEE, 961–973.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting bugs in Cypher graph database engines. In Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "JanusGraph. 2023. JanusGraph. https://janusgraph.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhanga, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 16,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. ACM Sigplan Notices 49, 6 (2014), 216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Vu Le, Chengnian Sun, and Zhendong Su. 2015. Finding deep compiler bugs via guided stochastic program mutation. ACMSIGPLAN Notices 50, 10 (2015), 386–399.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Daniel Lehmann and Michael Pradel. 2018. Feedback-directed differential testing of interactive debuggers. In Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 610–620.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Mikael Lindvall, Dharmalingam Ganesan, Ragnar Árdal, and Robert E Wiegand. 2015. Metamorphic model-based testing applied on NASADAT–An experience report. In neering, Vol. 2. IEEE, 129–138.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Jiawei Liu, Jinkun Lin, Fabian Ruffy, Cheng Tan, Jinyang Li, Aurojit Panda, and Lingming Zhang. 2023. Nnsmith: Generating diverse and valid test cases for deep learning compilers. In Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2. 530–543.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. In Proceedings of the 44th International Conference on Software Engineering. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "William M McKeeman. 1998. Differential testing for software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "MemGraph. 2023. MemGraph. https://memgraph.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "NebulaGraph. 2023. NebulaGraph. https://www.nebula-graph.io/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Neo4j. 2023. Cypher Features Implemented by Neo4j. https://neo4j.com/docs/ cypher-manual/current/queries/basic/#find-connected-nodes",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Neo4j. 2023. Neo4J. https://neo4j.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "nGQL. 2023. nGQL. https://docs.nebula-graph.io/1.2.0/manual-EN/1.overview/1. concepts/2.nGQL-overview/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "openCypher. 2023. openCypher. https://opencypher.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Theofilos Petsios, Adrian Tang, Salvatore Stolfo, Angelos D Keromytis, and Suman Jana. 2017. Nezha: Efficient domain-independent differential testing. In. IEEE, 615–632.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Hamid Pirahesh, Joseph M. Hellerstein, and Waqar Hasan. 1992. Extensible/Rule Based Query Rewrite Optimization in Starburst. In Proceedings of the 1992 ACMSIGMOD International Conference on Management of Data (San Diego, California, USA) (SIGMOD ’92). Association for Computing Machinery, New York, NY, USA, 39–48. https://doi.org/10.1145/130283.130294",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Redis. 2023. RedisGraph. https://docs.redis.com/latest/stack/deprecated-features/ graph/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (nov 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis.. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), Vol. 20. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 34,
+      "text": "Daniel Ritter, Luigi Dell’Aquila, Andrii Lomakin, and Emanuele Tagliaferri. 2021. OrientDB: A NoSQL, Open Source MMDMS. In Proceedings of the The British International Conference on Databases 2021, London, United Kingdom, March 28, 2022 (CEUR Workshop Proceedings, Vol. 3163). CEUR-WS.org, 10–19.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Marko A. Rodriguez. 2015. The Gremlin Graph Traversal Machine and Language. CoRR abs/1508.03843 (2015). arXiv:1508.03843",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. In Engineering (ICSE). 2072–2084. https://doi.org/10.1109/ICSE48619.2023.00175",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Thodoris Sotiropoulos, Stefanos Chaliasos, Vaggelis Atlidakis, Dimitris Mitropoulos, and Diomidis Spinellis. 2021. Data-oriented differential testing of objectrelational mapping systems. In Software Engineering (ICSE). IEEE, 1535–1547.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Rainer Storn and Kenneth Price. 1997. Differential evolution–a simple and efficient heuristic for global optimization over continuous spaces. Journal of global optimization 11, 4 (1997), 341–359.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. 2016. Finding compiler bugs via live code mutation. In Proceedings of the 2016 ACMSIGPLAN international conference on object-oriented programming, systems, languages, and applications. 849–863.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "TigerGraph. 2023. TigerGraph. https://www.tigergraph.com",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "TinkerGraph. 2023. TinkerGraph. https://github.com/tinkerpop/blueprints/wiki/ tinkergraph",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Boxi Yu, Yiyan Hu, Qiuyang Mang, Wenhan Hu, and Pinjia He. 2023. Automated Testing and Improvement of Named Entity Recognition Systems. In Proceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (, San Francisco, CA, USA,) (ESEC/FSE 2023). Association for Computing Machinery, New York, NY, USA, 883–894. https://doi.org/10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Boxi Yu, Zhiqing Zhong, Xinran Qin, Jiayi Yao, Yuancheng Wang, and Pinjia He. 2022. Automated testing of image captioning systems. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis. 467–479.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis (Virtual, South Korea) (ISSTA 2022). Association for Computing Machinery, New York, NY, USA, 302–313. ",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In (ICSE). 2060–2071. https://doi.org/10.1109/ICSE48619.2023.00174",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "prints the DOI of the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 15,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhanga, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (nov 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis.. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), Vol. 20. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 2,\n32,33,36]",
+      "technique": "qpg",
+      "sentence": "Second, it is non-trivial to adapt RDBMS testing approaches [ 2, 32,33,36] to GDBMS because it incurs re-implementation under different database models and query languages.",
+      "context_before": "y languages, such as Neo4j’s Cypher [ 28], TinkerGraph’s Gremlin [ 35], and NebulaGraph’s nGQL [ 27]. Thus, existing differential testing tools, like GDsmith [ 13] and Grand [ 44], can only test the shared functionalities of GDBMS that adopt the same query language, such as Neo4j’s feature of label expressions [ 25]. Moreover, the adoption of different query languages can lead to extensive engineering effort when developers try to generalize existing tools to more query languages. We think this is the main reason why existing research focuses on at most two query languages: Cypher and Gremlin.",
+      "context_after": "Moreover, this kind of adaptation can only test the shared functionalities between RDBMS and GDBMS (e.g., expressions and predicates), leaving graph-related bugs underexplored. For example, GDBMeter [ 15], the only existing metamorphic testing approach for GDBMS, utilizes an RDBMS testing approach Ternery Logic Partitioning (TLP) [32], which partitions a query 𝑄into sub-queries 𝑄𝑃,𝑄¬𝑃,𝑄nullbased on three possible outcomes of the predicate 𝑃:true, false, and null. Thus, although GDBMeter found common bugs related to predicates handling, none of their reported bugs are graph-related. Third, per",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 6453,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In (ICSE). 2060–2071. https://doi.org/10.1109/ICSE48619.2023.00174",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "prints the DOI of the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For example, GDBMeter [ 15], the only existing metamorphic testing approach for GDBMS, utilizes an RDBMS testing approach Ternery Logic Partitioning (TLP) [32], which partitions a query 𝑄into sub-queries 𝑄𝑃,𝑄¬𝑃,𝑄nullbased on three possible outcomes of the predicate 𝑃:true, false, and null.",
+      "context_before": "rent query languages can lead to extensive engineering effort when developers try to generalize existing tools to more query languages. We think this is the main reason why existing research focuses on at most two query languages: Cypher and Gremlin. Second, it is non-trivial to adapt RDBMS testing approaches [ 2, 32,33,36] to GDBMS because it incurs re-implementation under different database models and query languages. Moreover, this kind of adaptation can only test the shared functionalities between RDBMS and GDBMS (e.g., expressions and predicates), leaving graph-related bugs underexplored.",
+      "context_after": "Thus, although GDBMeter found common bugs related to predicates handling, none of their reported bugs are graph-related. Third, performance bugs in GDBMS have not been carefully considered by existing approaches. Specifically, differential testing approaches, such as GDsmith [ 13] and Grand [ 44], fall short of detecting performance bugs because the diverse underlying architectures of different GDBMS can easily cause performance differences. The metamorphic testing approach, GDBMeter [ 15], is based on the relationship between the original query and three sub-queries, which do not exhibit cle",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 6803,
+      "found_by_all": [
+        "technique",
+        "citation_marker_project_authored",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 15]",
+      "technique": null,
+      "sentence": "The metamorphic testing approach, GDBMeter [ 15], is based on the relationship between the original query and three sub-queries, which do not exhibit clear performance relationships.",
+      "context_before": "Partitioning (TLP) [32], which partitions a query 𝑄into sub-queries 𝑄𝑃,𝑄¬𝑃,𝑄nullbased on three possible outcomes of the predicate 𝑃:true, false, and null. Thus, although GDBMeter found common bugs related to predicates handling, none of their reported bugs are graph-related. Third, performance bugs in GDBMS have not been carefully considered by existing approaches. Specifically, differential testing approaches, such as GDsmith [ 13] and Grand [ 44], fall short of detecting performance bugs because the diverse underlying architectures of different GDBMS can easily cause performance differences.",
+      "context_after": "Thus, it is difficult for existing approaches to find performance bugs. To tackle these challenges, this paper adapts Equivalent Query Rewriting (EQR) to GDBMS testing. For example, the first query 1https://github.com/RedisGraph/RedisGraph/issues/3093in Listing 1 was re-written to generate the second query as the equivalent query. If the returned results are different or a large difference was observed in the query time, a potential bug is found. To facilitate the realization of EQR in GDBMS, we propose a novel, widely applicable concept called Abstract Syntax Graph (ASG). The key insight is",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 7540,
+      "cited_reference": {
+        "number": 15,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhanga, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 13,15]",
+      "technique": null,
+      "sentence": "2 Graph Generation Unlike some testing techniques [ 13,15] that are specific to particular data, our EQR approach can be generally applied to any graph.",
+      "context_before": "the target graph database, potential bugs can be uncovered if they result in inconsistent systems behaviors. Graph Generator Base Query Generator Qbase Equivalent Query RewriterQ1 Q2 Qk.. . Inequivalent Result? Graph Database Equivalent Qu eries Base QuerySending queries to GDBMS and obtaining resultsFigure 4: An overview of GRev, which begins with the Graph Generator and Base Query Generator to produce a graph schema and base query, followed by the Equivalent Query Rewriter to generate several equivalent queries. A bug will be detected if these queries yield inconsistent system behaviors. 4.",
+      "context_after": "We can simply create a random LPG graph schema on the target graph database by executing the following steps. •First, create properties and assign them unique names and data types (e.g., int, float, and string). •Second, create labels for nodes and relationships, with each label corresponding to a set of randomly selected properties. •Third, create nodes and relationships, assigning them random labels and properties, and then assigning random values to their properties based on their data-types. 4.3 Base Query Generator To the best of our knowledge, existing tools for testing GDBMS are often",
+      "section": "4.2 Graph Generation",
+      "page": 5,
+      "char_offset": 26682,
+      "cited_reference": {
+        "number": 15,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhanga, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "Query Partitioning",
+      "technique": "tlp",
+      "sentence": "Similarly, GDBMeter [ 15], a tool based on Query Partitioning [ 32], requires different validation rules for different queries and may struggle to handle some widely-used clauses such as DISTINCT.",
+      "context_before": "string). •Second, create labels for nodes and relationships, with each label corresponding to a set of randomly selected properties. •Third, create nodes and relationships, assigning them random labels and properties, and then assigning random values to their properties based on their data-types. 4.3 Base Query Generator To the best of our knowledge, existing tools for testing GDBMS are often sensitive to the base query. For instance, differential testing tools like Grand [ 44] and GDsmith [ 13] are limited to using queries that rely on shared functionalities among the tested graph databases.",
+      "context_after": "This is because there can be potential duplicate elements across the sub-queries, and thus aggregating sub-queries with DISTINCT clauses may not produce results matching the original query. However, GRev is designed to be generally applicable to most base queries, with the only exception of the base queries that may produce uncertain results, such as those involving the random LIMIT clauses. Given this, we followed the grammar of GDBMS query languages to generate syntactically correct base queries. Furthermore, GRev can directly utilize any existing query generators, such as the generator of",
+      "section": "4.3 Base Query Generator",
+      "page": 5,
+      "char_offset": 27637,
+      "found_by_all": [
+        "technique",
+        "citation_marker_project_authored",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 15]",
+      "technique": null,
+      "sentence": ", GDsmith [ 13], Grand [ 44], and GDBMeter [ 15]).",
+      "context_before": "6], RedisGraph [ 31], MemGraph [ 23], TinkerGraph [ 40], and NebulaGraph [ 24]). Table 1 shows the basic information of the GDBMS selected, including DB-Engines Ranking [ 6]3, GitHub stars, and their initial release dates. Among these, Neo4j, RedisGraph, and MemGraph are the most popular opensource GDBMS that employ Cypher as their query language. TinkerGraph, developed by the creators of Gremlin [ 35], serves as the foundational library for several Gremlin-based GDBMS such as JanusGraph [ 14] and OrientDB [ 34]. Notably, these four GDBMS have undergone rigorous testing by existing works (e.g.",
+      "context_after": "Therefore, any previously unknown bugs we discovered on these databases highlight the proficiency of GRev. Additionally, we tested NebulaGraph, a widely-used GDBMS integrating its own query language, demonstrating the generalizability of GRev across multiple query languages. We tested the most recent versions of the GDBMS at the time of this work, which were Neo4j 5.6.0, RedisGraph 6.2.6-v7, MemGraph 2.8.0, TinkerGraph 3.6.0, and NebulaGraph 3.5.0. All experiments are conducted on a Linux (Ubuntu 22.04 LTS) workstation, equipped with a 13th Gen Intel(R) Core(TM) i5-13400 and 128 GB of memory.",
+      "section": "5.1 Experimental Setting",
+      "page": 6,
+      "char_offset": 32584,
+      "cited_reference": {
+        "number": 15,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhanga, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 15]",
+      "technique": null,
+      "sentence": "Notably, these bugs were rarely detected by the existing techniques, especially for GDBMeter [ 15], which reused the test oracles designed for RDBMS (i.",
+      "context_before": "s, we discovered a total of 22 previously unknown bugs, 15 already confirmed by the developers. Of the 22 reported bugs, 12 are logic bugs, 3 are performance bugs, 4 are missing exceptions, and 3 are unexpected exceptions. GDBMS Query Language Detected Confirmed Logic Bugs Performance Bugs Missing Exceptions Unexpected Exceptions Graph-related Ne o4j Cypher 3 2 0 0 2 1 1 Re disGraph Cypher 6 2 4 1 0 1 4 MemGraph Cypher 4 3 3 0 1 0 3 Tinker Graph Gremlin 3 2 2 0 1 0 2 NebulaGraph nGQL 6 6 3 2 0 1 4 T otal 22 15 12 3 4 3 14 rather than common functionalities shared with RDBMS (e.g., predicates).",
+      "context_after": "e., TLP [32]). Selected Interesting Bugs. To provide an insight into the variety of bugs that GRev is capable of identifying, we present a selection of interesting cases discovered by our approach. For brevity, we present simplified test cases (without showing graph data) for better understanding. Listing 2: MemGraph (logic bug, graph-related bug) - Returning unmatched data when executing MATCH clauses after OPTIONAL MATCH clauses. (1) MATCH (n0:L3)<-[ r0:T1]-(n1) OPTIONAL MATCH (n2:L2) < -[] -( n1:L1), ( n1:L1) -[] - >( n0) WITH *MATCH (n0) < -[] -( n1) -[] - >( n2)WITH *RETURN * --{n0: 1924",
+      "section": "5.2 Detecting Previously Unknown Bugs",
+      "page": 7,
+      "char_offset": 35433,
+      "cited_reference": {
+        "number": 15,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhanga, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", TLP [32]).",
+      "context_before": "ce bugs, 4 are missing exceptions, and 3 are unexpected exceptions. GDBMS Query Language Detected Confirmed Logic Bugs Performance Bugs Missing Exceptions Unexpected Exceptions Graph-related Ne o4j Cypher 3 2 0 0 2 1 1 Re disGraph Cypher 6 2 4 1 0 1 4 MemGraph Cypher 4 3 3 0 1 0 3 Tinker Graph Gremlin 3 2 2 0 1 0 2 NebulaGraph nGQL 6 6 3 2 0 1 4 T otal 22 15 12 3 4 3 14 rather than common functionalities shared with RDBMS (e.g., predicates). Notably, these bugs were rarely detected by the existing techniques, especially for GDBMeter [ 15], which reused the test oracles designed for RDBMS (i.e.",
+      "context_after": "Selected Interesting Bugs. To provide an insight into the variety of bugs that GRev is capable of identifying, we present a selection of interesting cases discovered by our approach. For brevity, we present simplified test cases (without showing graph data) for better understanding. Listing 2: MemGraph (logic bug, graph-related bug) - Returning unmatched data when executing MATCH clauses after OPTIONAL MATCH clauses. (1) MATCH (n0:L3)<-[ r0:T1]-(n1) OPTIONAL MATCH (n2:L2) < -[] -( n1:L1), ( n1:L1) -[] - >( n0) WITH *MATCH (n0) < -[] -( n1) -[] - >( n2)WITH *RETURN * --{n0: 19244, n1: 19230,",
+      "section": "5.2 Detecting Previously Unknown Bugs",
+      "page": 7,
+      "char_offset": 35588,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 15]",
+      "technique": null,
+      "sentence": "In contrast, existing techniques based on certain query transformations, such as GDBMeter [ 15], can only trigger 1 (from the base query) + 3 (from three sub-queries) = 4 distinct query plans.",
+      "context_before": "query plans for long base queries. The number of distinct query plans obtained for a base query increases as the number of generated equivalent queries increases. If we generate 3,000 equivalent queries for the 10 random base queries (300 each), we can obtain 178 distinct query plans (the rightmost pink bar in Fig. 6), averaging 17.8 distinct plans per base query. Similarly, we can obtain 1253 distinct query plans (the rightmost yellow bar in Fig. 6), averaging 125.3 distinct query plans per base query. Thus, we believe the equivalent queries generated by GRev can trigger distinct query plans.",
+      "context_after": "Moreover, differential testing tools like GDBMeter and Grand are restricted to triggering only one query plan in each target GDBMS for a given base query. Consequently, when executing differential testing on Testing Graph Database Systems via Equivalent Query Rewriting ICSE 2024, April 2024, Lisbon, Portugal Table 3: Comparing GRev with Existing Techniques T ool Approach Supp orted LanguagesAble to detect shared bugsAble to test unique featuresInsensitive to base queryDetected performance bugsDetected graph-related bugs GDsmith Differ ential Testing Cypher # # # # Grand Differential Testing G",
+      "section": "5.3 Test Cases Analysis",
+      "page": 8,
+      "char_offset": 43771,
+      "cited_reference": {
+        "number": 15,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhanga, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "Query Partitioning",
+      "technique": "tlp",
+      "sentence": "Consequently, when executing differential testing on Testing Graph Database Systems via Equivalent Query Rewriting ICSE 2024, April 2024, Lisbon, Portugal Table 3: Comparing GRev with Existing Techniques T ool Approach Supp orted LanguagesAble to detect shared bugsAble to test unique featuresInsensitive to base queryDetected performance bugsDetected graph-related bugs GDsmith Differ ential Testing Cypher # # # # Grand Differential Testing Gr emlin # # # # # GDBMeter Query Partitioning Cypher, Gremlin # # # GRe v Quer y Rewriting Cypher, Gremlin, nGQL G # three GDBMS simultaneously, these tools can only activate three distinct query plans for each base query.",
+      "context_before": "ns per base query. Similarly, we can obtain 1253 distinct query plans (the rightmost yellow bar in Fig. 6), averaging 125.3 distinct query plans per base query. Thus, we believe the equivalent queries generated by GRev can trigger distinct query plans. In contrast, existing techniques based on certain query transformations, such as GDBMeter [ 15], can only trigger 1 (from the base query) + 3 (from three sub-queries) = 4 distinct query plans. Moreover, differential testing tools like GDBMeter and Grand are restricted to triggering only one query plan in each target GDBMS for a given base query.",
+      "context_after": "To investigate RQ2 (c), we executed 1,250,000 test cases on RedisGraph [ 31] within a 12-hour window, where each test case includes one base query and one equivalent query. Among these test cases, (945/1,250,000) managed to trigger bugs, of which 936 are logic bugs and 9 are performance bugs. This result indicates that on average, GRev is able to identify one bug-triggering test case executing for every 1,322 test case executions. In our practical testing, it takes an average of one minute to identify a bug-triggering test case. In summary, the analysis of the test cases above demonstrates th",
+      "section": "5.3 Test Cases Analysis",
+      "page": 8,
+      "char_offset": 44119,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 15]",
+      "technique": null,
+      "sentence": ", GDsmith [ 13], Grand [ 44], and GDBMeter [ 15], that can detect bugs other than crashes.",
+      "context_before": "gs, of which 936 are logic bugs and 9 are performance bugs. This result indicates that on average, GRev is able to identify one bug-triggering test case executing for every 1,322 test case executions. In our practical testing, it takes an average of one minute to identify a bug-triggering test case. In summary, the analysis of the test cases above demonstrates the high effectiveness and efficiency of GRev in generating high-quality test cases and triggering bugs. 5.4 Analytical Comparison with Existing Techniques As shown in Table 3, there are three closest-related works in testing GDBMS, i.e.",
+      "context_after": "Specifically, GDsmith and Grand are based on Differential Testing [38] while GDBMeter is based on Query Partitioning [32] (i.e., partitioning query into multiple equivalent sub-queries). Notably, we can not compare GRev with RDBMS testing tools, such as SQLancer [ 33]. Although GRev and these tools both test the bugs in database systems, their target databases have fundamental differences, in terms of data structures (e.g., graph model vs. relational model), query mode (e.g., traversals vs. joins), and languages (e.g., Cypher [ 28], Gremlin [ 35], and nGQL [ 27] vs. SQL). Even for testing the",
+      "section": "5.4 Analytical Comparison with Existing",
+      "page": 9,
+      "char_offset": 45622,
+      "cited_reference": {
+        "number": 15,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhanga, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "Query Partitioning",
+      "technique": "tlp",
+      "sentence": "Specifically, GDsmith and Grand are based on Differential Testing [38] while GDBMeter is based on Query Partitioning [32] (i.",
+      "context_before": "average, GRev is able to identify one bug-triggering test case executing for every 1,322 test case executions. In our practical testing, it takes an average of one minute to identify a bug-triggering test case. In summary, the analysis of the test cases above demonstrates the high effectiveness and efficiency of GRev in generating high-quality test cases and triggering bugs. 5.4 Analytical Comparison with Existing Techniques As shown in Table 3, there are three closest-related works in testing GDBMS, i.e., GDsmith [ 13], Grand [ 44], and GDBMeter [ 15], that can detect bugs other than crashes.",
+      "context_after": "e., partitioning query into multiple equivalent sub-queries). Notably, we can not compare GRev with RDBMS testing tools, such as SQLancer [ 33]. Although GRev and these tools both test the bugs in database systems, their target databases have fundamental differences, in terms of data structures (e.g., graph model vs. relational model), query mode (e.g., traversals vs. joins), and languages (e.g., Cypher [ 28], Gremlin [ 35], and nGQL [ 27] vs. SQL). Even for testing the GDBMS that adopt the same query languages, different GDBMS can have their unique design or adopt some shared libraries with o",
+      "section": "5.4 Analytical Comparison with Existing",
+      "page": 9,
+      "char_offset": 45712,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Notably, we can not compare GRev with RDBMS testing tools, such as SQLancer [ 33].",
+      "context_before": "g-triggering test case. In summary, the analysis of the test cases above demonstrates the high effectiveness and efficiency of GRev in generating high-quality test cases and triggering bugs. 5.4 Analytical Comparison with Existing Techniques As shown in Table 3, there are three closest-related works in testing GDBMS, i.e., GDsmith [ 13], Grand [ 44], and GDBMeter [ 15], that can detect bugs other than crashes. Specifically, GDsmith and Grand are based on Differential Testing [38] while GDBMeter is based on Query Partitioning [32] (i.e., partitioning query into multiple equivalent sub-queries).",
+      "context_after": "Although GRev and these tools both test the bugs in database systems, their target databases have fundamental differences, in terms of data structures (e.g., graph model vs. relational model), query mode (e.g., traversals vs. joins), and languages (e.g., Cypher [ 28], Gremlin [ 35], and nGQL [ 27] vs. SQL). Even for testing the GDBMS that adopt the same query languages, different GDBMS can have their unique design or adopt some shared libraries with other GDBMS. In the first case, these differential testing tools are sensitive to queries or are prone to give false positives. For example, MemG",
+      "section": "5.4 Analytical Comparison with Existing",
+      "page": 9,
+      "char_offset": 45899,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "Query Partitioning",
+      "technique": "tlp",
+      "sentence": "Based on Query Partitioning, GDBMeter [ 15] leverages the test oracle called Ternary Logic Partitioning (TLP) to detect bugs in GDBMS.",
+      "context_before": "adopt some shared libraries with other GDBMS. In the first case, these differential testing tools are sensitive to queries or are prone to give false positives. For example, MemGraph [ 23] and Neo4j [ 26] will deliberately return different results for the same query MATCH ()-[r]-(), and thus GDsmith does not test such queries to avoid false alarms. In the second case, these tools are prone to give false negatives. In addition, GDsmith and Grand face challenges in detecting performance bugs by comparing performance discrepancies, as different target GDBMSs have varying underlying architectures.",
+      "context_after": "TLP [ 32] was initially introduced in testing RDBMS. It involves partitioning a query 𝑄into sub-queries 𝑄𝑃,𝑄¬𝑃,𝑄nullbased on three possible outcomes of the predicate 𝑃:true, false, and null. However, this partitioning approach is sensitive to query and query language. For example, where the query includes functionalities such as removing duplicate elements, aggregating the sub-queries may no longer produce an equivalent result to the original query.Listing 5: An example of GDBMeter’s missed bug-triggering test cases of the bug shown in Listing 2, where the aggregation of the results from the",
+      "section": "5.4 Analytical Comparison with Existing",
+      "page": 9,
+      "char_offset": 47004,
+      "found_by_all": [
+        "technique",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [ 32] was initially introduced in testing RDBMS.",
+      "context_before": "to give false positives. For example, MemGraph [ 23] and Neo4j [ 26] will deliberately return different results for the same query MATCH ()-[r]-(), and thus GDsmith does not test such queries to avoid false alarms. In the second case, these tools are prone to give false negatives. In addition, GDsmith and Grand face challenges in detecting performance bugs by comparing performance discrepancies, as different target GDBMSs have varying underlying architectures. Based on Query Partitioning, GDBMeter [ 15] leverages the test oracle called Ternary Logic Partitioning (TLP) to detect bugs in GDBMS.",
+      "context_after": "It involves partitioning a query 𝑄into sub-queries 𝑄𝑃,𝑄¬𝑃,𝑄nullbased on three possible outcomes of the predicate 𝑃:true, false, and null. However, this partitioning approach is sensitive to query and query language. For example, where the query includes functionalities such as removing duplicate elements, aggregating the sub-queries may no longer produce an equivalent result to the original query.Listing 5: An example of GDBMeter’s missed bug-triggering test cases of the bug shown in Listing 2, where the aggregation of the results from the three sub-queries still produces the same incorrect r",
+      "section": "5.4 Analytical Comparison with Existing",
+      "page": 9,
+      "char_offset": 47139,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "(TLP :𝑄𝑃)MATCH (n0:L3)<-[ r0:T1]-(n1) OPTIONAL MATCH (n0) < -[] -( n1:L1) -[] - >( n2:L2) WITH *MATCH (n0) < -[] -( n1) -[] - >( n2)WITH * WHERE (ID(n2) > ID(n0))RETURN * --{n0: 19244, n1: 19230, n2: 19328, r0: 85207} (TLP :𝑄¬𝑃)MATCH (n0:L3)<-[ r0:T1]-(n1) OPTIONAL MATCH (n0) < -[] -( n1:L1) -[] - >( n2:L2) WITH *MATCH (n0) < -[] -( n1) -[] - >( n2)WITH * WHERE NOT (ID(n2) > ID(n0))RETURN *--{} (TLP :𝑄null)MATCH (n0:L3)<-[ r0:T1]-(n1) OPTIONAL MATCH (n0) < -[] -( n1:L1) -[] - >( n2:L2) WITH *MATCH (n0) < -[] -( n1) -[] - >( n2)WITH * WHERE (ID(n2) > ID(n0))IS NULL RETURN * --{n0: 19317, n1: 19223, n2: null, r0: 85206} As a result, GDBMeter is unable to effectively test queries that involve DISTINCT clauses in Cypher [ 28] and dedup() clauses in Gremlin [ 35].",
+      "context_before": "ed on three possible outcomes of the predicate 𝑃:true, false, and null. However, this partitioning approach is sensitive to query and query language. For example, where the query includes functionalities such as removing duplicate elements, aggregating the sub-queries may no longer produce an equivalent result to the original query.Listing 5: An example of GDBMeter’s missed bug-triggering test cases of the bug shown in Listing 2, where the aggregation of the results from the three sub-queries still produces the same incorrect result as the original query, which leads to the bug being missed.",
+      "context_after": "Similarly to Grand [ 44] and GDsmith [ 13], GDBMeter also lacks general support for multiple query languages in GDBMS. For example, since Gremlin lacks steps to check for null values in expressions, GDBMeter can only test Gremlin-based GDBMS when working with graphs that do not contain null values. In addition, TLP can only be applied to filter clauses (e.g., WHERE ), and is not capable of effectively modifying query plans of the traversal process (e.g., MATCH ), resulting in GDBMeter facing challenges in finding traversal bugs. For example, as shown in Listing 5, the TLP approach fails to ge",
+      "section": "5.4 Analytical Comparison with Existing",
+      "page": 9,
+      "char_offset": 47859,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "In addition, TLP can only be applied to filter clauses (e.",
+      "context_before": "1:L1) -[] - >( n2:L2) WITH *MATCH (n0) < -[] -( n1) -[] - >( n2)WITH * WHERE (ID(n2) > ID(n0))IS NULL RETURN * --{n0: 19317, n1: 19223, n2: null, r0: 85206} As a result, GDBMeter is unable to effectively test queries that involve DISTINCT clauses in Cypher [ 28] and dedup() clauses in Gremlin [ 35]. Similarly to Grand [ 44] and GDsmith [ 13], GDBMeter also lacks general support for multiple query languages in GDBMS. For example, since Gremlin lacks steps to check for null values in expressions, GDBMeter can only test Gremlin-based GDBMS when working with graphs that do not contain null values.",
+      "context_after": "g., WHERE ), and is not capable of effectively modifying query plans of the traversal process (e.g., MATCH ), resulting in GDBMeter facing challenges in finding traversal bugs. For example, as shown in Listing 5, the TLP approach fails to generate bug-triggering test cases for the graphrelated bugs shown in Listing 2. Specifically, after conducting a thorough investigation of GDBMeter’s bug reports, we found that all of the logic bugs were triggered by queries that contain only a simple MATCH (n)clause or return a mathematical expression directly. This implies that the root causes of their det",
+      "section": "5.4 Analytical Comparison with Existing",
+      "page": 9,
+      "char_offset": 48928,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For example, as shown in Listing 5, the TLP approach fails to generate bug-triggering test cases for the graphrelated bugs shown in Listing 2.",
+      "context_before": "NCT clauses in Cypher [ 28] and dedup() clauses in Gremlin [ 35]. Similarly to Grand [ 44] and GDsmith [ 13], GDBMeter also lacks general support for multiple query languages in GDBMS. For example, since Gremlin lacks steps to check for null values in expressions, GDBMeter can only test Gremlin-based GDBMS when working with graphs that do not contain null values. In addition, TLP can only be applied to filter clauses (e.g., WHERE ), and is not capable of effectively modifying query plans of the traversal process (e.g., MATCH ), resulting in GDBMeter facing challenges in finding traversal bugs.",
+      "context_after": "Specifically, after conducting a thorough investigation of GDBMeter’s bug reports, we found that all of the logic bugs were triggered by queries that contain only a simple MATCH (n)clause or return a mathematical expression directly. This implies that the root causes of their detected logic bugs lie solely in the implementations of predicates and expressions, with no relation to the graph itself. 5.5 Empirical Comparison with Existing Techniques To better compare with previous works, we conducted two empirical experiments, each corresponding to code coverage, and bug ICSE 2024, April 2024, Li",
+      "section": "5.4 Analytical Comparison with Existing",
+      "page": 9,
+      "char_offset": 49163,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 15]",
+      "technique": null,
+      "sentence": "Based on metamorphic testing, GDBMeter [ 15] is the SOTA tool for detecting logic bugs in GDBMS.",
+      "context_before": "different query languages, provide test oracles for the single target GDBMS, and detect shared bugs across multiple GDBMS. In this work, our EQR approach addresses these challenges by providing a universal framework ASG and a test oracle based on metamorphic testing. Metamorphic Testing. Metamorphic testing has been successfully applied in various domains [ 12,19,42,43]. For example, Orion [ 16], Athena [ 17], and Hermes [ 39] introduce metamorphic testing techniques in compilers such as equivalent dead code mutation and equivalent live code mutation to identify logic bugs within the compiler.",
+      "context_after": "However, none of the graph-related bugs have been detected by it due to reusing TLP[ 32], a metamorphic relation for RDBMS. In this work, we propose a novel and more effective metamorphic testing method for GDBMS that can detect graph-related bugs and common bugs shared with RDBMS.Equivlanet Rewriting in RDBMS Testing. Both AMOEBA[ 21] and GRev are able to generate equivalent queries to trigger performance bugs in the database. AMOEBA supports 75 mutation rules defined by domain experts, which incurs domain knowledge and extensive effort in rule validation. However, our work does not need ext",
+      "section": "7 RELATED WORK",
+      "page": 11,
+      "char_offset": 57369,
+      "cited_reference": {
+        "number": 15,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhanga, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "However, none of the graph-related bugs have been detected by it due to reusing TLP[ 32], a metamorphic relation for RDBMS.",
+      "context_before": "gs across multiple GDBMS. In this work, our EQR approach addresses these challenges by providing a universal framework ASG and a test oracle based on metamorphic testing. Metamorphic Testing. Metamorphic testing has been successfully applied in various domains [ 12,19,42,43]. For example, Orion [ 16], Athena [ 17], and Hermes [ 39] introduce metamorphic testing techniques in compilers such as equivalent dead code mutation and equivalent live code mutation to identify logic bugs within the compiler. Based on metamorphic testing, GDBMeter [ 15] is the SOTA tool for detecting logic bugs in GDBMS.",
+      "context_after": "In this work, we propose a novel and more effective metamorphic testing method for GDBMS that can detect graph-related bugs and common bugs shared with RDBMS.Equivlanet Rewriting in RDBMS Testing. Both AMOEBA[ 21] and GRev are able to generate equivalent queries to trigger performance bugs in the database. AMOEBA supports 75 mutation rules defined by domain experts, which incurs domain knowledge and extensive effort in rule validation. However, our work does not need extra rules to define the transformation process. Our EQR only depends on the original ASG and uses RWC to automatically genera",
+      "section": "7 RELATED WORK",
+      "page": 11,
+      "char_offset": 57466,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M13"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:48:43Z",
+    "is_model_written": true,
+    "summary": "GRev tests graph database systems by rewriting a Cypher query into a form that must return the same result, then comparing the two. Its argument against the existing metamorphic approach is that partitioning a query into sub-queries yields only a handful of distinct query plans and cannot express graph-specific constructs, so graph-related bugs go undetected. GRev found bugs in Neo4j, RedisGraph and MemGraph that the existing tools had not.",
+    "narrative": "The paper works from Ternary Logic Partitioning throughout, but at one remove: its comparison target is GDBMeter, which carries TLP to graph databases, and the paper is precise that GDBMeter is based on Query Partitioning and reuses a test oracle designed for relational systems. That reuse is its critique -- it reports that none of the graph-related bugs were detected by GDBMeter for exactly that reason, that TLP applies only to filter clauses, and that partitioning yields four distinct query plans against GRev's many. It states outright that GRev cannot be compared with relational tools such as SQLancer, so SQLancer itself is not a baseline; TLP is, through GDBMeter.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "definition",
+      "M3": "background",
+      "M4": "background",
+      "M5": "motivation",
+      "M6": "baseline",
+      "M7": "result_comparison",
+      "M8": "definition",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "baseline",
+      "M12": "definition",
+      "M13": "incidental",
+      "M14": "definition",
+      "M15": "background",
+      "M16": "incidental",
+      "M17": "motivation",
+      "M18": "result_comparison",
+      "M19": "state_of_the_art",
+      "M20": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "GRev is compared with GDBMeter rather than built on it, and M13 states that GRev cannot be compared with relational tools such as SQLancer at all."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "GRev rewrites a query into an equivalent one rather than partitioning it, which the paper presents as an alternative to TLP rather than a generalisation. Its treatment of TLP is a critique of carrying a relational oracle to graphs."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M6",
+          "M11",
+          "M18",
+          "M20"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M6",
+            "sentence": ", GDsmith [ 13], Grand [ 44], and GDBMeter [ 15]).",
+            "section": "5.1 Experimental Setting",
+            "page": 6
+          },
+          {
+            "mention_id": "M11",
+            "sentence": ", GDsmith [ 13], Grand [ 44], and GDBMeter [ 15], that can detect bugs other than crashes.",
+            "section": "5.4 Analytical Comparison with Existing",
+            "page": 9
+          },
+          {
+            "mention_id": "M18",
+            "sentence": "For example, as shown in Listing 5, the TLP approach fails to generate bug-triggering test cases for the graphrelated bugs shown in Listing 2.",
+            "section": "5.4 Analytical Comparison with Existing",
+            "page": 9
+          },
+          {
+            "mention_id": "M20",
+            "sentence": "However, none of the graph-related bugs have been detected by it due to reusing TLP[ 32], a metamorphic relation for RDBMS.",
+            "section": "7 RELATED WORK",
+            "page": 11
+          }
+        ],
+        "reasoning": "M6 and M11 name GDBMeter among the tools GRev is evaluated against, and M18 and M20 report the outcome and attribute it to GDBMeter's reuse of TLP. The comparison reaches TLP through GDBMeter, which M12 and M14 identify as a Query Partitioning tool; M13 states SQLancer itself was not compared against.",
+        "techniques": [
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M19"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M19",
+            "sentence": "Based on metamorphic testing, GDBMeter [ 15] is the SOTA tool for detecting logic bugs in GDBMS.",
+            "section": "7 RELATED WORK",
+            "page": 11
+          }
+        ],
+        "reasoning": "M19 calls GDBMeter the SOTA tool for detecting logic bugs in graph database systems. The recognition attaches to the TLP-based tool rather than to SQLancer directly."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3597503_3639207.json b/_data/papers/paper_doi_10_1145_3597503_3639207.json
new file mode 100644
index 0000000..1266757
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3597503_3639207.json
@@ -0,0 +1,801 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:32:49Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3639207",
+    "title": "Understanding Transaction Bugs in Database Systems",
+    "authors": [
+      "Ziyu Cui",
+      "Wensheng Dou",
+      "Yu Gao",
+      "Dong Wang",
+      "Jiansen Song",
+      "Yingying Zheng",
+      "Tao Wang",
+      "Rui Yang",
+      "Kang Xu",
+      "Yixin Hu",
+      "Jun Wei",
+      "Tao Huang"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3639207",
+    "arxiv_id": null,
+    "s2_paper_id": "c8376be6e5d269467545de4edffe06bb2b74b1c7",
+    "url": "https://doi.org/10.1145/3597503.3639207",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3597503.3639207",
+      "retrieved_at": "2026-09-09T01:32:49Z",
+      "chars": 84068,
+      "content_sha256": "sha256:cc5188028d9450e9ebc31238ebd52c5c36e2d8c9106adec4800bdce671800513"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2108
+      },
+      {
+        "number": "2",
+        "title": "PRELIMINARIES",
+        "start": 9197
+      },
+      {
+        "number": "3",
+        "title": "METHODOLOGY",
+        "start": 13710
+      },
+      {
+        "number": "3.1",
+        "title": "Target DBMSs",
+        "start": 13724
+      },
+      {
+        "number": "3.2",
+        "title": "Collecting TXBugs",
+        "start": 14749
+      },
+      {
+        "number": "3.3",
+        "title": "Analyzing TXBugs",
+        "start": 17425
+      },
+      {
+        "number": "3.4",
+        "title": "Threats to Validity",
+        "start": 19703
+      },
+      {
+        "number": "3.5",
+        "title": "TXBug Dataset",
+        "start": 21629
+      },
+      {
+        "number": "4.1",
+        "title": "Initialization",
+        "start": 22537
+      },
+      {
+        "number": "6.1",
+        "title": "Failure Symptoms",
+        "start": 42091
+      },
+      {
+        "number": "7.1",
+        "title": "Unfixed TXBugs",
+        "start": 47781
+      },
+      {
+        "number": "8.1",
+        "title": "Transaction Verification",
+        "start": 50812
+      },
+      {
+        "number": "8.2",
+        "title": "Transaction Testing",
+        "start": 52401
+      },
+      {
+        "number": "9.1",
+        "title": "Detecting More Types of TXBugs",
+        "start": 55253
+      },
+      {
+        "number": "9.2",
+        "title": "Transaction Testing",
+        "start": 55927
+      },
+      {
+        "number": "9.3",
+        "title": "Transaction Verification",
+        "start": 59432
+      },
+      {
+        "number": "9.4",
+        "title": "Transaction Semantics",
+        "start": 60103
+      },
+      {
+        "number": "9.5",
+        "title": "TXBug Diagnosis",
+        "start": 61046
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2022. The ANSI isolation levels. https://renenyffenegger.ch/notes/development/ databases/SQL/transaction/isolation-level.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2022. Async Commit in TiDB. https://docs.pingcap.com/tidb/stable/release5.0.0#async-commit.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2022. Character Set Support in PostgreSQL. https://www.postgresql.org/docs/ current/multibyte.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2022. Character Sets, Collations, Unicode in MySQL. https://dev.mysql.com/doc/ refman/8.0/en/charset.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2022. CockroachDB. https://www.cockroachlabs.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2022. CockroachDB issues. https://github.com/cockroachdb/cockroach/issues/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2022. DB-Engines. https://db-engines.com/en/ranking.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2022. Gretchen: Offline serializability verification, in Clojure. https://jepsen.io/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2022. MariaDB. https://mariadb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2022. MariaDB JIRA issues. https://jira.mariadb.org/issues/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2022. MySQL. https://www.mysql.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2022. MySQL Bugs Home. https://bugs.mysql.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "2022. Pgsql-bugs. https://www.postgresql.org/list/pgsql-bugs/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "2022. PostgreSQL. https://www.postgresql.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "2022. SQLite. https://www.sqlite.org/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "2022. SQLite Ticket Main Menu. https://www.sqlite.org/src/reportlist.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "2022. SQLsmith. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "2022. TiDB issues. https://github.com/pingcap/tidb/issues/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "2022. TiDB, PingCAP. https://pingcap.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "2022. Transaction Isolation in PostgreSQL. https://www.postgresql.org/docs/14/ transaction-iso.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "2022. Transaction Isolation Levels in MySQL. https://dev.mysql.com/doc/refman/ 8.0/en/innodb-transaction-isolation-levels.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "2022. Transactions in TiDB. https://docs.pingcap.com/tidb/stable/transactionoverview.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "2023. JanusGraph. https://janusgraph.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "2023. Memcached. http://www.memcached.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "2023. Neo4j. https://neo4j.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "2023. OrientDB. https://orientdb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "2023. Redis. https://redis.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "2023. XA Transactions in MariaDB. https://mariadb.com/kb/en/xa-transactions/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "2023. XA Transactions in MySQL. https://dev.mysql.com/doc/refman/8.0/en/xa. html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Atul Adya. 1999. Weak Consistency: A Generalized Theory and Optimistic Implementations for Distributed Transactions. Ph. D. Dissertation. Massachusetts Institute of Technology.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Atul Adya, Barbara Liskov, and Patrick O’Neil. 2000. Generalized Isolation Level Definitions. In Proceedings of International Conference on Data Engineering (ICDE). 67–78.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Mohammed Alfatafta, Basil Alkhatib, Ahmed Alquraan, and Samer Al-Kiswany. 2020. Toward a Generic Fault Tolerance Technique for Partial Network Partitioning. In Proceedings of USENIX Conference on Operating Systems Design and Implementation (OSDI). 351–368.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Ahmed Alquraan, Hatem Takruri, Mohammed Alfatafta, and Samer Al-Kiswany. 2018. An Analysis of Network-Partitioning Failures in Cloud Systems. In Proceedings of USENIX Conference on Operating Systems Design and Implementation (OSDI). 51–68.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of International Conference on Software Engineering (ICSE). 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 35,
+      "text": "Peter Bailis, Aaron Davidson, Alan Fekete, Ali Ghodsi, Joseph M. Hellerstein, and Ion Stoica. 2013. Highly Available Transactions: Virtues and Limitations. Proceedings of the VLDB Endowment 7, 3 (2013), 181–192.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Peter Bailis, Alan Fekete, Ali Ghodsi, Joseph M. Hellerstein, and Ion Stoica. 2016. Scalable Atomic Visibility with RAMP Transactions. ACM Transactions on Database Systems 41, 3 (2016), 15:1–15:45.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Hal Berenson, Phil Bernstein, Jim Gray, Jim Melton, Elizabeth O’Neil, and Patrick O’Neil. 1995. A Critique of ANSI SQL Isolation Levels. In Proceedings of International Conference on Management of Data (SIGMOD). 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Philip A. Bernstein and Nathan Goodman. 1983. Multiversion Concurrency Control—Theory and Algorithms. ACM Transactions on Database Systems (TODS) 8, 4 (1983), 465–483. ICSE ’24, April 14–20, 2024, Lisbon, Portugal Ziyu Cui, et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Ranadeep Biswas and Constantin Enea. 2019. On the Complexity of Checking Transactional Consistency. In Proceedings of ACMSIGPLAN Conference on ObjectOriented Programming Systems, Languages, and Applications (OOPSLA). 165:1– 165:28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Lucas Brutschy, Dimitar Dimitrov, Peter Müller, and Martin Vechev. 2017. Serializability for Eventual Consistency: Criterion, Analysis, and Applications. In Proceedings of ACMSIGPLAN Symposium on Principles of Programming Languages (POPL). 458–472.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Andrea Cerone, Giovanni Bernardi, and Alexey Gotsman. 2015. A Framework for Transactional Consistency Models with Atomic Visibility. In Proceedings of International Conference on Concurrency Theory (CONCUR). 58–71.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Junjie Chen, Yihua Liang, Qingchao Shen, Jiajun Jiang, and Shuochuan Li. 2023. Toward Understanding Deep Learning Framework Bugs. ACM Trans. Softw. Eng. Methodol. 32, 6 (2023), 31 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Natacha Crooks, Youer Pu, Lorenzo Alvisi, and Allen Clement. 2017. Seeing is Believing: A Client-Centric Specification of Database Isolation. In Proceedings of the ACM Symposium on Principles of Distributed Computing (PODC). 73–82.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Ziyu Cui, Wensheng Dou, Qianwang Dai, Jiansen Song, Wei Wang, Jun Wei, and Dan Ye. 2022. Differentially Testing Database Transactions for Fun and Profit. In Proceedings of International Conference on Automated Software Engineering (ASE). 35:1–35:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Wensheng Dou, Ziyu Cui, Qianwang Dai, Jiansen Song, Dong Wang, Yu Gao, Wei Wang, Jun Wei, Lei Chen, Hanmo Wang, Hua Zhong, and Tao Huang. 2023. Detecting Isolation Bugs via Transaction Oracle Construction. In Proceedings of International Conference on Software Engineering (ICSE). 1123–1135.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Yu Gao, Wensheng Dou, Feng Qin, Chushu Gao, Dong Wang, Jun Wei, Ruirui Huang, Li Zhou, and Yongming Wu. 2018. An Empirical Study on Crash Recovery Bugs in Large-Scale Distributed Systems. In Proceedings of ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 539–550.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Hao Guan, Ying Xiao, Jiaying Li, Yepang Liu, and Guangdong Bai. 2023. A Comprehensive Study of Real-World Bugs in Machine Learning Model Optimization. In Proceedings of International Conference on Software Engineering (ICSE). 147–158.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Haryadi S. Gunawi, Mingzhe Hao, Tanakorn Leesatapornwongsa, Tiratat Patanaanake, Thanh Do, Jeffry Adityatama, Kurnia J. Eliazar, Agung Laksono, Jeffrey F. Lukman, Vincentius Martin, and Anang D. Satria. 2014. What Bugs Live in the Cloud? A Study of 3000+ Issues in Cloud Systems. In Proceedings of ACM Symposium on Cloud Computing (SOCC). 1–14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting Bugs in Cypher Graph Database Engines. In Proceedings of International Symposium on Software Testing and Analysis (ISSTA). 163–174.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Daniel Jackson and Craig A. Damon. 1996. Elements of Style: Analyzing a Software Design Feature with a Counterexample Detector. IEEE Transactions on Software Engineering 22, 7 (1996), 484–495.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. InProceedings of USENIX Security Symposium (USENIX Security). 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 397–417.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 53,
+      "text": "Guoliang Jin, Linhai Song, Xiaoming Shi, Joel Scherpelz, and Shan Lu. 2012. Understanding and Detecting Real-World Performance Bugs. In Proceedings of ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). 77–88.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. Proceedings of the VLDB Endowment (VLDB) 13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of International Symposium on Software Testing and Analysis (ISSTA). 140–149.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 56,
+      "text": "Kyle Kingsbury and Peter Alvaro. 2020. Elle: Inferring Isolation Anomalies from Experimental Observations. Proceedings of the VLDB Endowment 14, 3 (2020), 268–280.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Hsiang-Tsung Kung and John T Robinson. 1981. On Optimistic Methods for Concurrency Control. ACM Transactions on Database Systems (TODS) 6, 2 (1981), 213–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Tanakorn Leesatapornwongsa, Jeffrey F. Lukman, Shan Lu, and Haryadi S. Gunawi. 2016. TaxDC: A Taxonomy of Non-Deterministic Concurrency Bugs in Datacenter Distributed Systems. In Proceedings of International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS). 517–530.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-OrientedDBMS Fuzzing. In Proceedings of International Conference on Data Engineering (ICDE). 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of DBMS with Coverage-based Guidance. In Proceedings of USENIX Security Symposium (USENIX Security). 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic Detection of Performance Bugs in Database Systems using Equivalent Queries. In Proceedings of International Conference on Software Engineering (ICSE). 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Shan Lu, Soyeon Park, Eunsoo Seo, and Yuanyuan Zhou. 2008. Learning from Mistakes: A Comprehensive Study on Real World Concurrency Bug Characteristics. In Proceedings of International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS). 329–339.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "David Patrick Reed. 1978. Naming and Synchronization in a Decentralized Computer System. Technical Report.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 65,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 211:1–211:30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 66,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 67,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In Proceedings of International Conference on Very Large Data Bases (VLDB). 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. In Proceedings of International Conference on Software Engineering (ICSE). 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Xiaohui Song and Jane W-S Liu. 1990. Performance of Multiversion Concurrency Control Algorithms in Maintaining Temporal Consistency. In Proceedings of Annual International Computer Software and Applications Conference (COMPSAC). 132–139.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Chengnian Sun, Vu Le, Qirun Zhang, and Zhendong Su. 2016. Toward Understanding Compiler Bugs in GCC and LLVM. In Proceedings of International Symposium on Software Testing and Analysis (ISSTA). 294–305.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Cheng Tan, Changgeng Zhao, Shuai Mu, and Michael Walfish. 2020. Cobra: Making Transactional Key-Value Stores Verifiably Serializable. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 63–80.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Jie Wang, Wensheng Dou, Yu Gao, Chushu Gao, Feng Qin, Kang Yin, and Jun Wei. 2017. A Comprehensive Study on Real World Concurrency Bugs in Node.Js. InProceedings of International Conference on Automated Software Engineering (ASE). 520–531.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In Proceedings of International Conference on Software Engineering (ICSESEIP). 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Rui Yang, Yingying Zheng, Lei Tang, Wensheng Dou, Wei Wang, and Jun Wei. 2023. Randomized Differential Testing of RDF Stores. In Proceedings of International Conference on Software Engineering (ICSE Demo). 136–140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Xiangyao Yu, Andrew Pavlo, Daniel Sanchez, and Srinivas Devadas. 2016. TicToc: Time Traveling Optimistic Concurrency Control. In Proceedings of International Conference on Management of Data (SIGMOD). 1629–1642.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Yingying Zheng, Wensheng Dou, Lei Tang, Ziyu Cui, Jiansen Song, Ziyue Cheng, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2024. Differential Optimization Testing of Gremlin-Based Graph Database Systems. In Proceedings of IEEE International Conference on Software Testing, Verification and Validation (ICST).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing. In Proceedings of International Symposium on Software Testing and Analysis (ISSTA). 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of ACMSIGSAC Conference on Computer and Communications Security (CCS). 955–970. ADETAILED STUDY METHODOLOGY Due to the space limit, we cannot present the detailed process about our study methodology in Section ",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 34,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of International Conference on Software Engineering (ICSE). 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 52,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 397–417.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 55,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of International Symposium on Software Testing and Analysis (ISSTA). 140–149.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 64,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 65,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 211:1–211:30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 66,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 78,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of ACMSIGSAC Conference on Computer and Communications Security (CCS). 955–970. ADETAILED STUDY METHODOLOGY Due to the space limit, we cannot present the detailed process about our study methodology in Section 3. In this appendix, we provide some supplementary details about the TXBug collection and categorization process. Understanding Transaction Bugs in Database Systems ICSE ’24, April 14–20, 2024, Lisbon, Portugal A.1 Collecting TXBugs In our empirical study, we adopt the following process to identify relevant DBMSs and TXBugs. 1)To make our studied TXBugs representative, we select six welldeveloped and popular DBMSs as our target systems, i.e., MySQL [11] , PostgreSQL [ 14], SQLite [ 15], MariaDB [ 9], CockroachDB [5] , and TiDB [ 19]. These systems cover various types of DBMSs (i.e., traditional, embedded and distributed DBMSs), and a diverse set of transaction features (e.g., transaction types, isolation levels and concurrency control mechanisms). Among them, MySQL, PostgreSQL, SQLite and MariaDB are well-developed DBMSs and rank high in the DB-Engines Ranking (2nd, 4th, 10th and 13th, respectively). CockroachDB and TiDB are two popular distributed DBMSs in the open source community (ranked as 3rd and 5th among DBMSs in GitHub). 2)We manually collect issues that are potentially related to TXBugs from the above six DBMSs’ issue repositories. a)Our studied DBMSs usually evolve quickly. To keep our study results up-to-date, we only consider issues confirmed by DBMS developers in the last 5 years, i.e., from January 2018 to December 2022. b)Our target DBMSs contain a huge amount of issues, e.g., MySQL contains about 20,000 issues in the last 5 years. In these DBMSs, they usually do not label whether an issue is related to transaction bugs. It is also challenging and time-consuming to manually check all the reported issues. Therefore, we use the following keywords, i.e., “transaction”, “abort”, “commit”, “isolation level”, and their variations, to effectively collect potentially relevant issues in these DBMSs. This search returns us with 7,775 issues. 3)Note that the above search keywords are general and widely used in DBMSs. Although some issues contain our search keywords, they are unrelated to transaction processing mechanisms in fact. To exclude these issues that are not related to transaction processing mechanisms from our study, we manually inspect the bug description and developer comments for each retrieved issue, and check whether the issue is related to transaction processing mechanisms in DBMSs. We only keep the issues that are related to transaction processing mechanisms. In this step, we filter out 7,220 issues, and keep 555 issues for further investigation. 4)For each remaining issue, we carefully investigate its bug description, test cases, developer discussions and available fixing patches, and further rebuild its bug test case step by step, and use the following criteria to select TXBugs. a)We keep an issue as a TXBug if it needs at least one explicit transaction or XA transaction to trigger, or it needs at least two transactions to trigger. b)If a bug can be triggered by only one auto transaction, we do not consider it as a TXBug, since it usually belongs to transaction-unrelated bugs, e.g., logic bugs in single SELECT statements. c)To maintain the accuracy of our study results, we only keep TXBugs that we can completely understand. To maintain the accuracy of our study, each issue has been investigated by at least three authors. For each issue, its investigatorscarefully discuss it and reach a consensus. If we cannot reach a consensus about an issue after discussion, we also exclude it from our study. Finally, we obtain 140 TXBugs in our study. A.2 Categorizing TXBugs We adopt the open card sorting approach to build the categories for TXBugs’ triggering conditions, root causes, impacts and fixing. The basic categorization process is described as follows. 1)For each TXBug, we extract necessary information from its issue report, e.g., test cases, expected execution results, developer discussions, and available fixing patches, and then write down its test case step by step. All the information is cross-checked by at least three authors. 2)Each TXBug is labeled by at least three investigators (i.e., authors) independently. Each investigator tries to classify a TXBug into an existing category in a dimension (e.g., root cause and bug impact) according to the extracted information. If a bug cannot be classified into any existing category, the investigator creates a new category for it. 3)For each TXBug, its investigators discuss their categorization results and try to reach a consensus. If some investigators have disagreements about a TXBug’s categorization result, they reinvestigate it further until they reach a consensus. They also refine categories in this step if necessary. 4)To ensure correctness and consistency, step 3 and 4 are performed multiple rounds, until all investigators reach a consensus for all transaction bugs’ categorization results. Specifically, we apply the following process to categorize the studied transaction bugs. •Triggering conditions. Triggering conditions are usually hidden in a TXBug’s test case, e.g., the number of transactions and the types of the involved transactions. Each investigator extracts these triggering information from a bug’s test case, and classifies TXBugs according to these triggering information. •Root causes. For root causes, we try to answer the question “what transaction semantics does a transaction bug violate?”. We investigate root causes from a bug’s transaction test case, expected execution results and developers’ discussions. Initially, we create four categories related to ACID properties (i.e., atomicity violations, consistency violations, isolation violations and durability violations). During our investigation, we further reveal more categories, e.g., read-only constraint violations and statement correctness violations. These are surprising categories to us. We do not find any TXBugs for durability violations, so we delete this category in our study. •Impacts. For bug impacts, we try to answer the question “what consequences does a transaction bug cause?”. We investigate bug impacts from a bug’s transaction test case and developers’ discussions about the unexpected execution results of the test case. We also reproduce some TXBugs if possible, and observe their consequences. •Fixing. We extract necessary information from issue reports, and available fixing patches. Then, we classify TXBugs according to these information.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 8,39,44,45,52,56,71]",
+      "technique": null,
+      "sentence": "Some approaches have been proposed to verify and test transaction processing mechanisms in DBMSs [ 8,39,44,45,52,56,71].",
+      "context_before": "e., transaction bugs, which can cause DBMSs to violate their transaction semantics, e.g., their claimed ACID (Atomicity, Consistency, Isolation, and Durability) properties [ 1,35–37,40,41,43]. We call transaction bugs in DBMSs as TXBugs for short. TXBugs can cause severe consequences in DBMSs, and easily go unnoticed by DBMS developers. For example, we find that about a third of TXBugs can result in incorrect query results, database states and DBMS states. Considering that DBMSs are usually used to handle important data assets, it is critically important to better understand and detect TXBugs.",
+      "context_after": "Transaction verification approaches [ 8,39,56,71] verify whether transactions’ concurrent executions violate the claimed isolation constraints. These approaches can only utilize simple data structures, e.g., 𝑘𝑒𝑦−𝑣𝑎𝑙𝑢𝑒, and cannot support the rich and complex features in modern DBMSs, e.g., partition configurations for data sharding. For transaction testing, DT2[44] utilizes differential testing to detect TXBugs, which cannot handle DBMS-specific features. Troc [ 45] detects TXBugs by decoupling a pair of transactions into independent statements and building test oracles for transactions. Howe",
+      "section": "1 INTRODUCTION",
+      "page": 1,
+      "char_offset": 4652,
+      "cited_reference": {
+        "number": 52,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 397–417.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 64–66]",
+      "technique": "norec",
+      "sentence": ", logic bugs in single SELECT statements [ 64–66].",
+      "context_before": "92.9%) issues do not involve transaction processing mechanisms. We exclude these issues from our study. For each remaining issue, we carefully investigate its description, embedded test cases, developer discussions and available fixing patches, and further rebuild its bug test case step by step. We keep an issue as a TXBug if it needs at least one explicit transaction or XA transaction to trigger, or it needs at least two transactions to trigger. If a bug can be triggered by only one auto transaction, we do not consider it as a TXBug, since it usually belongs to transactionunrelated bugs, e.g.",
+      "context_after": "To maintain the accuracy of our study results, we only keep TXBugs that we can completely understand. We finally collect 140 TXBugs from our studied DBMSs. Table 2 shows the detailed statistics from their isolation levels and concurrency control mechanisms. We can see that these TXBugs cover various isolation levels and concurrency control mechanisms. 3.3 Analyzing TXBugs To answer our five research questions, we perform an in-depth analysis on these 140 TXBugs based on their issue descriptions, embedded test cases, developer discussions, and available fixing patches, and further assign them",
+      "section": "3.2 Collecting TXBugs",
+      "page": 3,
+      "char_offset": 17020,
+      "cited_reference": {
+        "number": 64,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 17,34,49,\n51,54,55,59–61,64–68,73,74,77,78]",
+      "technique": "qpg",
+      "sentence": "Although many approaches have been proposed for DBMS testing [ 17,34,49, 51,54,55,59–61,64–68,73,74,77,78], there are only few works for transaction testing in DBMSs [ 44,45].",
+      "context_before": "constraint and statement correctness under transactions. Existing verification approaches [ 8,39,56,71] mainly focus on detecting insufficient isolation violations, while DT2[44] and Troc [ 45] can further detect some other violations except insufficient isolation violations, e.g., consistency violations. However, these approaches cannot detect some types of our studied TXBugs, e.g., atomicity and read-only constraint violations. Our community urgently needs to invest more effort to combat more types of TXBugs. 9.2 Transaction Testing Software testing plays an important role in exposing bugs.",
+      "context_after": "Our study provides guidance that can be exploited by transaction testing from several aspects. Transaction test case generation. A well-designed transaction test case generation approach can greatly improve the effectiveness of exposing TXBugs. We find that transaction test cases that trigger TXBugs are extremely diverse. Transaction test cases should consider specific database properties (Finding 3), multiple types of transactions (Finding 4), and various types of SQL statements (Finding 5). However, existing approaches, e.g., Cobra [ 71] and Elle [56], can only generate transactions that us",
+      "section": "9.2 Transaction Testing",
+      "page": 10,
+      "char_offset": 56009,
+      "cited_reference": {
+        "number": 34,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of International Conference on Software Engineering (ICSE). 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[ 17,34,49,51,54,55,59–\n61,64–68,73,74,76–78]",
+      "technique": "qpg",
+      "sentence": "Some techniques have been proposed to combat logic and crash bugs in DBMSs [ 17,34,49,51,54,55,59– 61,64–68,73,74,76–78].",
+      "context_before": "inds of bugs, e.g., concurrency bugs [ 58,62,72], crash / network partition recovery bugs in distributed systems [ 32,33,46], compiler bugs in GCC and LLVM [ 70], bugs in deep learning systems [ 42,47] and bugs in cloud systems [ 48]. These empirical studies have motivated researchers to develop various techniques to combating related bugs. In our study, we adopt similar study methodologies to existing studies in the process of bug selection, bug analysis, and bug categorization. To the best of our knowledge, our work is the first comprehensive study on TXBugs in DBMSs. Bug detection in DBMSs.",
+      "context_after": "SQLsmith [ 17] detects DBMS bugs by generating random SQL statements. RAGS [ 67] utilizes differential testing to find bugs in DBMSs. SQLancer [ 64–66] detects logic bugs in single SELECT statements. Squirrel [ 78] presents a fuzzing framework to find bugs in DBMSs. Grand [ 77] and DOT [ 76] present differential testing approaches to find logic bugs in graph database systems (GDBs). These approaches mainly focus on detecting bugs in single SELECT statements, and cannot detect TXBugs in DBMSs. 11 CONCLUSION Incorrect designs and implementations in DBMSs’ transaction processing mechanisms can i",
+      "section": "9.5 TXBug Diagnosis",
+      "page": 11,
+      "char_offset": 62630,
+      "cited_reference": {
+        "number": 34,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of International Conference on Software Engineering (ICSE). 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 64–66] detects logic bugs in single SELECT statements.",
+      "context_before": "es have motivated researchers to develop various techniques to combating related bugs. In our study, we adopt similar study methodologies to existing studies in the process of bug selection, bug analysis, and bug categorization. To the best of our knowledge, our work is the first comprehensive study on TXBugs in DBMSs. Bug detection in DBMSs. Some techniques have been proposed to combat logic and crash bugs in DBMSs [ 17,34,49,51,54,55,59– 61,64–68,73,74,76–78]. SQLsmith [ 17] detects DBMS bugs by generating random SQL statements. RAGS [ 67] utilizes differential testing to find bugs in DBMSs.",
+      "context_after": "Squirrel [ 78] presents a fuzzing framework to find bugs in DBMSs. Grand [ 77] and DOT [ 76] present differential testing approaches to find logic bugs in graph database systems (GDBs). These approaches mainly focus on detecting bugs in single SELECT statements, and cannot detect TXBugs in DBMSs. 11 CONCLUSION Incorrect designs and implementations in DBMSs’ transaction processing mechanisms can introduce transaction bugs, which lead to severe consequences. We conduct the first in-depth study on 140 transaction bugs from six popular DBMSs. From our study, we obtain many interesting findings an",
+      "section": "9.5 TXBug Diagnosis",
+      "page": 11,
+      "char_offset": 62886,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 78]",
+      "technique": null,
+      "sentence": "Squirrel [ 78] presents a fuzzing framework to find bugs in DBMSs.",
+      "context_before": "bating related bugs. In our study, we adopt similar study methodologies to existing studies in the process of bug selection, bug analysis, and bug categorization. To the best of our knowledge, our work is the first comprehensive study on TXBugs in DBMSs. Bug detection in DBMSs. Some techniques have been proposed to combat logic and crash bugs in DBMSs [ 17,34,49,51,54,55,59– 61,64–68,73,74,76–78]. SQLsmith [ 17] detects DBMS bugs by generating random SQL statements. RAGS [ 67] utilizes differential testing to find bugs in DBMSs. SQLancer [ 64–66] detects logic bugs in single SELECT statements.",
+      "context_after": "Grand [ 77] and DOT [ 76] present differential testing approaches to find logic bugs in graph database systems (GDBs). These approaches mainly focus on detecting bugs in single SELECT statements, and cannot detect TXBugs in DBMSs. 11 CONCLUSION Incorrect designs and implementations in DBMSs’ transaction processing mechanisms can introduce transaction bugs, which lead to severe consequences. We conduct the first in-depth study on 140 transaction bugs from six popular DBMSs. From our study, we obtain many interesting findings and lessons. We believe that our study can be beneficial for DBMS and",
+      "section": "9.5 TXBug Diagnosis",
+      "page": 11,
+      "char_offset": 62952,
+      "cited_reference": {
+        "number": 78,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of ACMSIGSAC Conference on Computer and Communications Security (CCS). 955–970. ADETAILED STUDY METHODOLOGY Du",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:20:36Z",
+    "is_model_written": true,
+    "summary": "The first comprehensive study of transaction bugs in DBMSs, based on 140 bugs collected from MySQL, PostgreSQL, SQLite, MariaDB, CockroachDB and TiDB. The authors examine how the bugs manifest, their root causes, their impact -- incorrect database states and crashes among them -- and how they were fixed, arguing that understanding real transaction bugs is what effective detection techniques must be built on.",
+    "narrative": "SQLancer is cited as the work on logic bugs in single SELECT statements, which the authors distinguish from the transaction bugs they collect: the boundary that defines their dataset.",
+    "roles": {
+      "M1": "background",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3597503_3639210.json b/_data/papers/paper_doi_10_1145_3597503_3639210.json
new file mode 100644
index 0000000..1d3639a
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3597503_3639210.json
@@ -0,0 +1,569 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:34Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3639210",
+    "title": "Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer",
+    "authors": [
+      "Jingzhou Fu",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Yu Jiang"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3639210",
+    "arxiv_id": null,
+    "s2_paper_id": "368ae600b0d2ac9e8e228f2c84e32a270403d7fa",
+    "url": "https://doi.org/10.1145/3597503.3639210",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3597503.3639210",
+      "retrieved_at": "2026-09-09T01:31:34Z",
+      "chars": 69156,
+      "content_sha256": "sha256:4a8b655acf4a1a6b82b8061eca6b5c7af615bb5ee1f4c7b0a3248ddd37244d21"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 3434
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND ANDMOTIVATION",
+        "start": 10955
+      },
+      {
+        "number": "3",
+        "title": "DESIGN",
+        "start": 16643
+      },
+      {
+        "number": "3.1",
+        "title": "SchemaCapture",
+        "start": 16833
+      },
+      {
+        "number": "3.2",
+        "title": "SQLTransfer with LLM",
+        "start": 21379
+      },
+      {
+        "number": "3.3",
+        "title": "MutabilityRefinement",
+        "start": 26081
+      },
+      {
+        "number": "4",
+        "title": "IMPLEMENTATION",
+        "start": 30469
+      },
+      {
+        "number": "5",
+        "title": "EVALUATION",
+        "start": 31607
+      },
+      {
+        "number": "5.1",
+        "title": "EvaluationSetup",
+        "start": 32200
+      },
+      {
+        "number": "5.2",
+        "title": "DBMSVulnerabilities",
+        "start": 34185
+      },
+      {
+        "number": "5.3",
+        "title": "Overall Experiments",
+        "start": 38316
+      },
+      {
+        "number": "5.4",
+        "title": "Contributionof SQL Transfer",
+        "start": 47072
+      },
+      {
+        "number": "5.5",
+        "title": "Effectivenessof Mutability Refinement",
+        "start": 50826
+      },
+      {
+        "number": "6",
+        "title": "DISCUSSION",
+        "start": 53128
+      },
+      {
+        "number": "7",
+        "title": "RELATED WORK",
+        "start": 56160
+      },
+      {
+        "number": "8",
+        "title": "CONCLUSION",
+        "start": 59806
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "ANSI. 2022. ANSI Standard. https://www .ansi.org/. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "JinshengBaandManuelRigger.2023. Testingdatabaseenginesviaqueryplan guidance. In Proceedings of International Conference on Software Engineering (ICSE).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "MonetDBB.V.2023. MonetDBWebsite. https://www .monetdb.org. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "ClickHouse. 2021. Fuzzing: Practical Approaches in ClickHouse. https:// presentations .clickhouse .com/cpp_siberia_2021/index _en.html. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "DuckDB. 2023. DuckDB Fuzzers. https://github .com/duckdb/duckdb-fuzzer/ issues. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "DuckDB. 2023. DuckDB WebSite. https://www .duckdb.org/. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Zhiyu Fan, Xiang Gao, Martin Mirchev, Abhik Roychoudhury, and Shin Hwei Tan.2023. Automatedrepairofprogramsfromlargelanguagemodels.In. IEEE, 1469–1481.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. 2020. AFL++: Combining Incremental Steps of Fuzzing Research. In USENIX Workshop on Offensive Technologies (WOOT).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. In 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "ClickHouseInc.2023. ClickHouseWebsite. https://clickhouse .com. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Free Software Foundation Inc. 2023. Error Recovery. https://www .gnu.org/ software/bison/manual/html _node/Error-Recovery .html. Accessed:January4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Solid IT. 2022. DB-Engines Ranking. https://db-engines .com/en/ranking. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "SolidIT.2022. DB-EnginesRankingofRelationalDBMS. https://db-engines .com/ en/ranking/relational +dbms. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for DatabaseManagementSystemswithComplexandValidSQLQueryGeneration.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2020. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB). Tokyo, Japan.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "SungminKang,JuyeonYoon,andShinYoo.2023. Largelanguagemodelsarefewshottesters:Exploringllm-basedgeneralbugreproduction.In 2023IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2312–2323.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Caroline Lemieux, Jeevana Priya Inala, Shuvendu K Lahiri, and Siddhartha Sen. 2023. CODAMOSA: Escaping coverage plateaus in test generation with pretrainedlargelanguagemodels.In Internationalconferenceonsoftwareengineering (ICSE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "JohnRLevine,TonyMason,andDougBrown.1992. Lex&yacc. \"O’ReillyMedia, Inc.\".",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "JieLiang,YaoguangChen,ZhiyongWu,JingzhouFu,MingzheWang,YuJiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequenceoriented DBMS fuzzing. In Proceedings of IEEE International Conference on Data Engineering (ICDE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "YuLiang,SongLiu,andHongHu.2022. DetectingLogicalBugsof {DBMS}with Coverage-based Guidance. In 31st USENIX Security Symposium (USENIX Security 22). 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "XuweiLiu,WeiYou,ZhuoZhang,andXiangyuZhang.2022.TensileFuzz:facilitatingseedinputgenerationinfuzzingviastringconstraintsolving.In Proceedingsof the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis. 391–403.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "XinyuLiu,QiZhou,JoyArulraj,andAlessandroOrso.2022. AutomaticDetection of Performance Bugs in Database Systems using Equivalent Queries. (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "MySQL.2023. MySQLGithub. https://github .com/mysql/mysql-server. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "MySQL. 2023. The MySQL Test Suite. https://dev .mysql.com/doc/dev/mysqlserver/latest/PAGE_MYSQL_TEST_RUN_PL .html. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Shankara Pailoor, Andrew Aday, and Suman Jana. 2018. Moonshine: Optimiz-ing OS fuzzer seed selection with trace distillation. In 27th USENIX Security Symposium (USENIX Security 18). 729–743.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "JorgePérez,MarceloArenas,andClaudioGutierrez.2009. SemanticsandcomplexityofSPARQL. ACMTransactionsonDatabaseSystems(TODS) 34,3(2009), 1–45. ICSE ’24, April 14–20, 2024, Lisbon, Portugal Jingzhou Fu, Jie Liang, Zhiyong Wu, and Yu Jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "PostgreSQL. 2023. PostgreSQL GitHub. https://github .com/postgres/postgres. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "PostgreSQL. 2023. Regression Tests. https://www .postgresql .org/docs/current/ regress-run .html. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Nitarshan Rajkumar, Raymond Li, and Dzmitry Bahdanau. 2022. Evaluating the text-to-sql capabilities of large language models. arXiv preprint arXiv:2204.00498 (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "ManuelRiggerandZhendongSu.2020. FindingBugsinDatabaseSystemsvia Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (2020). https://doi .org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 33,
+      "text": "Baptiste Roziere, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, et al .2023. Code llama: Open foundation models for code. arXiv preprint arXiv:2308.12950 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https://github .com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "OpenLink Software. 2023. ChatGLM2 repository. https://github .com/THUDM/ ChatGLM2-6B/blob/main/README_EN .md. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "OpenLink Software. 2023. Virtuoso Open-Source Edition. https:// vos.openlinksw .com/owiki/wiki/VOS. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "JiansenSong,WenshengDou,ZiyuCui,QianwangDai,WeiWang,JunWei,Hua Zhong,andTaoHuang.2023. TestingDatabaseSystemsviaDifferentialQuery Execution. In Proceedings of International Conference on Software Engineering (ICSE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "SQLite.2023. HowSQLiteIsTested. https://www .sqlite.org/testing .html. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "SQLite. 2023. SQLite Github. https://github .com/sqlite/sqlite. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "The Clang Team. 2023. SanitizerCoverage. https://clang .llvm.org/docs/ SanitizerCoverage .html. Accessed: January 4, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "HugoTouvron,LouisMartin,KevinStone,PeterAlbert,AmjadAlmahairi,Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosa l e ,e ta l.2023. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. 2017. Skyfire: Data-driven seedgenerationforfuzzing.In 2017IEEESymposiumonSecurityandPrivacy(SP). IEEE, 579–594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "MingzheWang,JieLiang,YuanliangChen,YuJiang,XunJiao,HanLiu,Xibin Zhao, and Jiaguang Sun. 2018. SAFL: increasing and accelerating testing cov-erage with symbolic execution and guided fuzzing. In Proceedings of the 40th International Conference on Software Engineering: Companion Proceeedings, ICSE 2018, Gothenburg, Sweden. ACM, 61–64.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, andYuJiang.2021. IndustryPracticeofCoverage-GuidedEnterprise-LevelDBMSFuzzing.In 2021IEEE/ACM43rdInternationalConferenceonSoftwareEngineering: Software Engineering in Practice (ICSE-SEIP). IEEE, 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "ZhiyongWu,JieLiang,MingzheWang,ChijinZhou,andYuJiang.2022. Unicorn: detect runtime errors in time-series databases with hybrid input synthesis. In ISSTA’22:31stACMSIGSOFTInternationalSymposiumonSoftwareTestingand Analysis, Virtual Event, South Korea, July 18 - 22, 2022, Sukyoung Ryu and Yannis Smaragdakis (Eds.). ACM, 251–262. https://doi .org/10.1145/3533767 .3534364",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "WeiYou,XuweiLiu,ShiqingMa,DavidPerry,XiangyuZhang,andBinLiang. 2019. SLF:Fuzzingwithoutvalidseedinputs.In 2019IEEE/ACM41stInternational Conferenceon Software Engineering (ICSE). IEEE, 712–723.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Aohan Zeng, Xiao Liu, Zhengxiao Du, Zihan Wang, Hanyu Lai, Ming Ding, Zhuoyi Yang, Yifan Xu, Wendi Zheng, Xiao Xia, et al .2022. Glm-130b: An open bilingual pre-trained model. arXiv preprint arXiv:2210.02414 (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "RuiZhong,YonghengChen,HongHu,HangfanZhang,WenkeLee,andDinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "JinshengBaandManuelRigger.2023. Testingdatabaseenginesviaqueryplan guidance. In Proceedings of International Conference on Software Engineering (ICSE).",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg"
+      ]
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 31,
+      "text": "ManuelRiggerandZhendongSu.2020. FindingBugsinDatabaseSystemsvia Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (2020). https://doi .org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp"
+      ]
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 2,9,14,19,20,44,45,48]",
+      "technique": "qpg",
+      "sentence": "The mutation-based fuzzing techniques [ 2,9,14,19,20,44,45,48] have been successfully appliedto DBMSsforeffectivetestingandbugfinding.",
+      "context_before": "and Figure 7 on several state-of-the-art LLMs including Llama 2[41],ChatGLM2 [35,47], and CodeLlama [33], with the results listed in Table 8. As illustrated, the LLMs finetuned for code tasks (e.g. CodeLlama ) and the LLMs with more parameters(e.g. ChatGLM2-130B )producebetterresults.Webelieve thatSedarcanperformwellwiththesemodelsontheSQLtransfer task. Moreover, to enhance the performance, we can fine-tune the LLMs with the documentation, tutorials, and native test cases ofthe target DBMSs, especially for new DBMS features or newly developed DBMSs. 7 RELATED WORK Mutation-Based DBMS Fuzzing.",
+      "context_after": "Sqirrel[ 48] mutates SQL statements by translating SQL statements into theIntermediate Representation (IR) for SQL, mutating the IRs, and re-translating the mutated IRs into syntactically and semantically correct SQLstatements.Lego[ 19]proposesthetypeaffinityof SQL sequencesandemploysittoguidethegenerationof SQLstatements withabundant SQLtypesequences.Griffin[ 9]utilizesagrammarfreewaytomutate SQLstatementswithstatement reshuffleand metadata-basedsubstitution.SQLRight[ 20]employstestoraclesto mutation-basedfuzzerstodetectlogicbugs.Whiletheaboveare allcoverage-guidedapproaches,QPG[ 2]proposestheconc",
+      "section": "7 RELATED WORK",
+      "page": 10,
+      "char_offset": 56203,
+      "cited_reference": {
+        "number": 2,
+        "text": "JinshengBaandManuelRigger.2023. Testingdatabaseenginesviaqueryplan guidance. In Proceedings of International Conference on Software Engineering (ICSE).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Whiletheaboveare allcoverage-guidedapproaches,QPG[ 2]proposestheconceptof query plan guidance to guide the fuzzing other than code coverage.",
+      "context_before": "ve been successfully appliedto DBMSsforeffectivetestingandbugfinding.Sqirrel[ 48] mutates SQL statements by translating SQL statements into theIntermediate Representation (IR) for SQL, mutating the IRs, and re-translating the mutated IRs into syntactically and semantically correct SQLstatements.Lego[ 19]proposesthetypeaffinityof SQL sequencesandemploysittoguidethegenerationof SQLstatements withabundant SQLtypesequences.Griffin[ 9]utilizesagrammarfreewaytomutate SQLstatementswithstatement reshuffleand metadata-basedsubstitution.SQLRight[ 20]employstestoraclesto mutation-basedfuzzerstodetectlogicbugs.",
+      "context_after": "Generation-Based DBMS Testing. Some generation-based approaches[ 15,22,30–32,34,37]areproposedtofindspecifictypes of DBMSbugs.SQLsmith[ 34]generates SELECTstatementswith Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer ICSE ’24, April 14–20, 2024, Lisbon, Portugal complexstructuresfromapredefinedAbstractSyntaxTree(AST)to findcrashbugs.NoREC[ 30]isdesignedtodetectoptimizationbugs bygenerating SQLqueriesaswellastheequivalentonesthatcannotbeoptimized.PQS[ 32]generatesqueriesthatshouldfetchaspecific row, and indicates a bug triggered when the row is not included in r",
+      "section": "7 RELATED WORK",
+      "page": 10,
+      "char_offset": 56869,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 15,22,30–32,34,37]",
+      "technique": "norec",
+      "sentence": "Some generation-based approaches[ 15,22,30–32,34,37]areproposedtofindspecifictypes of DBMSbugs.",
+      "context_before": "IR) for SQL, mutating the IRs, and re-translating the mutated IRs into syntactically and semantically correct SQLstatements.Lego[ 19]proposesthetypeaffinityof SQL sequencesandemploysittoguidethegenerationof SQLstatements withabundant SQLtypesequences.Griffin[ 9]utilizesagrammarfreewaytomutate SQLstatementswithstatement reshuffleand metadata-basedsubstitution.SQLRight[ 20]employstestoraclesto mutation-basedfuzzerstodetectlogicbugs.Whiletheaboveare allcoverage-guidedapproaches,QPG[ 2]proposestheconceptof query plan guidance to guide the fuzzing other than code coverage. Generation-Based DBMS Testing.",
+      "context_after": "SQLsmith[ 34]generates SELECTstatementswith Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer ICSE ’24, April 14–20, 2024, Lisbon, Portugal complexstructuresfromapredefinedAbstractSyntaxTree(AST)to findcrashbugs.NoREC[ 30]isdesignedtodetectoptimizationbugs bygenerating SQLqueriesaswellastheequivalentonesthatcannotbeoptimized.PQS[ 32]generatesqueriesthatshouldfetchaspecific row, and indicates a bug triggered when the row is not included in resultsets.DQE[ 37]focusesonlogicbugsin UPDATEandDELETE queriesbyverifyingthatthesamerowsareaccessedwiththesame predicate. Amoe",
+      "section": "7 RELATED WORK",
+      "page": 10,
+      "char_offset": 57040,
+      "cited_reference": {
+        "number": 30,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC[ 30]isdesignedtodetectoptimizationbugs bygenerating SQLqueriesaswellastheequivalentonesthatcannotbeoptimized.",
+      "context_before": "basedsubstitution.SQLRight[ 20]employstestoraclesto mutation-basedfuzzerstodetectlogicbugs.Whiletheaboveare allcoverage-guidedapproaches,QPG[ 2]proposestheconceptof query plan guidance to guide the fuzzing other than code coverage. Generation-Based DBMS Testing. Some generation-based approaches[ 15,22,30–32,34,37]areproposedtofindspecifictypes of DBMSbugs.SQLsmith[ 34]generates SELECTstatementswith Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer ICSE ’24, April 14–20, 2024, Lisbon, Portugal complexstructuresfromapredefinedAbstractSyntaxTree(AST)to findcrashbugs.",
+      "context_after": "PQS[ 32]generatesqueriesthatshouldfetchaspecific row, and indicates a bug triggered when the row is not included in resultsets.DQE[ 37]focusesonlogicbugsin UPDATEandDELETE queriesbyverifyingthatthesamerowsareaccessedwiththesame predicate. Amoeba [ 22] discovers performance bugs by generating equivalent SQL queries and comparing their execution times. InitialSeedGeneration. Severalexistingworks[ 21,25,43,46] concentrate on generating high-quality seeds for general-purpose mutation-based fuzzing. SAFL [ 43] employs symbolic execution to generate qualified initial seeds that provide valuable expl",
+      "section": "7 RELATED WORK",
+      "page": 11,
+      "char_offset": 57378,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS[ 32]generatesqueriesthatshouldfetchaspecific row, and indicates a bug triggered when the row is not included in resultsets.",
+      "context_before": "erage-guidedapproaches,QPG[ 2]proposestheconceptof query plan guidance to guide the fuzzing other than code coverage. Generation-Based DBMS Testing. Some generation-based approaches[ 15,22,30–32,34,37]areproposedtofindspecifictypes of DBMSbugs.SQLsmith[ 34]generates SELECTstatementswith Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer ICSE ’24, April 14–20, 2024, Lisbon, Portugal complexstructuresfromapredefinedAbstractSyntaxTree(AST)to findcrashbugs.NoREC[ 30]isdesignedtodetectoptimizationbugs bygenerating SQLqueriesaswellastheequivalentonesthatcannotbeoptimized.",
+      "context_after": "DQE[ 37]focusesonlogicbugsin UPDATEandDELETE queriesbyverifyingthatthesamerowsareaccessedwiththesame predicate. Amoeba [ 22] discovers performance bugs by generating equivalent SQL queries and comparing their execution times. InitialSeedGeneration. Severalexistingworks[ 21,25,43,46] concentrate on generating high-quality seeds for general-purpose mutation-based fuzzing. SAFL [ 43] employs symbolic execution to generate qualified initial seeds that provide valuable explorationdirections. SLF [ 46] generates valid seeds by automatically identifyinginputvaliditychecksofprogramsandinputfieldsrelat",
+      "section": "7 RELATED WORK",
+      "page": 11,
+      "char_offset": 57492,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:20:36Z",
+    "is_model_written": true,
+    "summary": "Sedar produces high-quality initial seeds for DBMS fuzzing by transferring test cases from other DBMSs, addressing the problem that many systems lack comprehensive built-in test suites. The insight is that DBMSs share many features, so a seed reaching deep paths in one can be adapted for another. Sedar executes existing test cases in their original DBMS while capturing schema information, uses an LLM guided by that information to generate new cases, and comments out unparsable sections so fuzzers can mutate them. It found 70 new vulnerabilities.",
+    "narrative": "Three citations in related work describing SQLancer's oracles -- NoREC, PQS and query plan guidance -- as the generation-based approaches aimed at specific bug types, in contrast to Sedar's coverage-guided seed transfer.",
+    "roles": {
+      "M1": "definition",
+      "M2": "background",
+      "M3": "definition",
+      "M4": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3597503_3639212.json b/_data/papers/paper_doi_10_1145_3597503_3639212.json
new file mode 100644
index 0000000..b1f74e8
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3597503_3639212.json
@@ -0,0 +1,723 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:36:43Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3597503.3639212",
+    "title": "Combining Structured Static Code Information and Dynamic Symbolic Traces for Software Vulnerability Prediction",
+    "authors": [
+      "Huanting Wang",
+      "Zhanyong Tang",
+      "Shin Hwei Tan",
+      "Jie Wang",
+      "Yuzhe Liu",
+      "Hejun Fang",
+      "Chunwei Xia",
+      "Zheng Wang"
+    ],
+    "year": 2024,
+    "venue": "International Conference on Software Engineering",
+    "doi": "10.1145/3597503.3639212",
+    "arxiv_id": null,
+    "s2_paper_id": "a3f32348f15845519abf3b862fce1682001da1cd",
+    "url": "https://doi.org/10.1145/3597503.3639212",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3597503.3639212",
+      "retrieved_at": "2026-09-08T14:36:43Z",
+      "chars": 82896,
+      "content_sha256": "sha256:9e8e57ebe4fc9e0e2df8a5a06a86c78341822f24d253bd635ebf4ece438743fc"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2529
+      },
+      {
+        "number": "2",
+        "title": "MOTIVATION",
+        "start": 10712
+      },
+      {
+        "number": "3",
+        "title": "OUR APPROACH",
+        "start": 13573
+      },
+      {
+        "number": "3.1",
+        "title": "Overview of Concoction",
+        "start": 14636
+      },
+      {
+        "number": "3.2",
+        "title": "The Concoction Architecture",
+        "start": 15889
+      },
+      {
+        "number": "3.2.1",
+        "title": "Extractingstaticcodeinformation. Weuseaparserbuiltupon",
+        "start": 16944
+      },
+      {
+        "number": "3.2.2",
+        "title": "Extracting dynamic information. We use KLEE [ 17] to obtain",
+        "start": 19107
+      },
+      {
+        "number": "3.3",
+        "title": "Contrastive Pre-training",
+        "start": 19792
+      },
+      {
+        "number": "3.3.1",
+        "title": "Model inputs. We pair the source code text and flattened",
+        "start": 20283
+      },
+      {
+        "number": "3.3.3",
+        "title": "Trainingtheend-to-enddetectionmodel. Aftertrainingthe",
+        "start": 21979
+      },
+      {
+        "number": "3.4.1",
+        "title": "Overview of pathselection. Figure 5 showsthe workflow of",
+        "start": 23885
+      },
+      {
+        "number": "3.4.3",
+        "title": "Network structure. As our path selection component uses",
+        "start": 25736
+      },
+      {
+        "number": "3.4.6",
+        "title": "Symbolicexecutionforchosenpaths. WeextendedKLEE[ 17]",
+        "start": 29440
+      },
+      {
+        "number": "4",
+        "title": "EXPERIMENTAL SETUP",
+        "start": 32119
+      },
+      {
+        "number": "4.1",
+        "title": "Workloads",
+        "start": 32599
+      },
+      {
+        "number": "4.2",
+        "title": "Competing Baselines",
+        "start": 34907
+      },
+      {
+        "number": "4.3",
+        "title": "Evaluation Methodology",
+        "start": 37450
+      },
+      {
+        "number": "5",
+        "title": "EXPERIMENTAL RESULTS",
+        "start": 40339
+      },
+      {
+        "number": "5.1.1",
+        "title": "Vulnerabilitycount. Table3reportsthedistributionofour",
+        "start": 40732
+      },
+      {
+        "number": "5.1.2",
+        "title": "Vulnerability types. Table 4 categorizes the vulnerabilities",
+        "start": 41048
+      },
+      {
+        "number": "5.2",
+        "title": "Comparison on Open Datasets",
+        "start": 46266
+      },
+      {
+        "number": "5.2.2",
+        "title": "CVE dataset. As explained in Sec. 4.1, test samples in the",
+        "start": 48173
+      },
+      {
+        "number": "5.3",
+        "title": "Comparison on Known CVEs",
+        "start": 49049
+      },
+      {
+        "number": "5.3.1",
+        "title": "Vulnerabilitiesidentified. Table5demonstratesConcoction’s",
+        "start": 49631
+      },
+      {
+        "number": "5.3.2",
+        "title": "Case missed by baselines. Listing 4 shows CVE-2020-35523,",
+        "start": 51274
+      },
+      {
+        "number": "5.3.3",
+        "title": "CasemissedbyConcoction. List5shows CVE-2020-35523,",
+        "start": 51976
+      },
+      {
+        "number": "5.4",
+        "title": "Ablation Study",
+        "start": 52692
+      },
+      {
+        "number": "5.4.1",
+        "title": "DLmodelimplementationchoices. Weconductanablation",
+        "start": 52711
+      },
+      {
+        "number": "5.4.2",
+        "title": "Sensitivityanalysis. Totestthesensitivityof Concoction",
+        "start": 54173
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "[n.d.]. CodeQL, discover vulnerabilities with semantic code analysis engine. https://codeql.github.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "[n.d.]. CommonVulnerabilities and Exposures (CVE). https://cve.mitre.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "[n.d.]. CommonWeakness Enumeration. https://cwe.mitre.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "[n.d.]. Infer, a static program analyzer. https://fbinfer.com/docs/about-Infer",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "[n.d.]. Jenkins, open source automation server. https://www.jenkins.io/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "[n.d.]. Joern(Open-Source Code Querying Engine for C/C++.). https://joern.io/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "[n.d.]. OSS-Fuzz. https://github.com/google/oss-fuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Elena N Akimova, Alexander Yu Bersenev, Artem A Deikov, Konstantin S Kobylkin,AntonVKonygin,IlyaPMezentsev,andVladimirEMisilov.2021. A survey on software defect prediction using deep learning. Mathematics 9, 11 (2021), 1180.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "MiltiadisAllamanis,MarcBrockschmidt,andMahmoudKhademi.2018. Learning to Represent Programs with Graphs. In Proceedings of the ICLR.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Sanjeev Arora, Hrishikesh Khandeparkar, Mikhail Khodak, Orestis Plevrakis,and Nikunj Saunshi. 2019. A theoretical analysis of contrastive unsupervised representation learning. arXivpreprint arXiv:1902.09229 (2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Roberto Baldoni, Emilio Coppa, Daniele Cono D’elia, Camil Demetrescu, and IreneFinocchi.2018.Asurveyofsymbolicexecutiontechniques. ACMComputing Surveys (CSUR) 51,3 (2018), 1–39.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "MarcelBöhme,Van-ThuanPham,Manh-DungNguyen,andAbhikRoychoudhury.2017.Directedgreyboxfuzzing.In Proceedingsofthe2017ACMSIGSACConference on Computer and Communications Security. 2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "CarlaEBrodleyandMarkAFriedl.1999. Identifyingmislabeledtrainingdata. Journal of artificial intelligence research 11 (1999), 131–167.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "StefanBucur,VladUreche,CristianZamfir,andGeorgeCandea.2011. Parallel Symbolic Execution for Automated Real-World Software Testing. In Proceedings of the Sixth Conference on Computer Systems (Salzburg, Austria) (EuroSys ’11). Association for Computing Machinery, New York, NY, USA, 183–198. https: //doi.org/10.1145/1966445.1966463",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Frank Busse, Pritam Gharat, Cristian Cadar, and Alastair F Donaldson. 2022. Combining static analysis error traces with dynamic symbolic execution (experi-ence paper). In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis. 568–579. ICSE ’24, April 14–20,2024, Lisbon, Portugal HuantingWang, Zhanyong Tang∗, Shin Hwei Tan, Jie Wang, Yuzhe Liu, Hejun Fang, Chun",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "FrankBusse,MartinNowack,andCristianCadar.2020. Runningsymbolicexecution forever. In Proceedings of the 29th ACMSIGSOFT International Symposium on Software Testing and Analysis. 63–74.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Cristian Cadar, Daniel Dunbar, Dawson R Engler, et al .2008. Klee: unassisted and automatic generation of high-coverage tests for complex systems programs.. InOSDI, Vol. 8. 209–224.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Cristian Cadar and Koushik Sen. 2013. Symbolic execution for software testing: three decades later. Commun.ACM 56,2 (2013), 82–90.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Mathilde Caron, Ishan Misra, Julien Mairal, Priya Goyal, Piotr Bojanowski, and Armand Joulin. 2020. Unsupervised Learning of Visual Features by Contrasting Cluster Assignments. https://doi.org/10.48550/ARXIV.2006.09882",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "DavideCastelvecchi. 2016. Can we open the black boxof AI? Nature News 538, 7623 (2016), 20.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Saikat Chakraborty, Rahul Krishna, Yangruibo Ding, and Baishakhi Ray. 2021. Deep learning based vulnerability detection: Are we there yet. IEEE Transactions on Software Engineering (2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "TingChen,SimonKornblith,MohammadNorouzi,andGeoffreyHinton.2020. A simple frameworkfor contrastive learning ofvisual representations. In International conference on machine learning. PMLR, 1597–1607.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "XiaoCheng,GuanqinZhang,HaoyuWang,andYuleiSui.2022. Path-Sensitive Code Embedding via Contrastive Learning for Software Vulnerability Detection. InProceedings of the 31st ACMSIGSOFT International Symposium on Software TestingandAnalysis (Virtual,SouthKorea) (ISSTA2022).AssociationforComputingMachinery,NewYork,NY,USA,519–531. https://doi.org/10.1145/3533767. 3534371",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "KhushbooChitre,PiyusKedia,andRahulPurandare.2022. Theroadnottaken: exploring alias analysis based optimizations missed by the compiler. Proceedings of the ACM on Programming Languages 6, OOPSLA2 (2022), 786–810.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Nathan Cooper, Carlos Bernal-Cárdenas, Oscar Chaparro, Kevin Moran, and DenysPoshyvanyk.2021. Ittakestwototango:Combiningvisualandtextual informationfordetectingduplicatevideo-basedbugreports.In 2021IEEE/ACM 43rd International Conference on Software Engineering (ICSE). IEEE, 957–969.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Chris Cummins, Pavlos Petoumenos, Zheng Wang, and Hugh Leather. 2017. End-to-enddeeplearningofoptimizationheuristics.In 201726thInternational Conference on Parallel Architectures and Compilation Techniques (PACT). IEEE, 219–232.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Chris Cummins, Pavlos Petoumenos, Zheng Wang, and Hugh Leather. 2017. Synthesizingbenchmarksforpredictivemodeling.In 2017IEEE/ACMInternational Symposium on Code Generation and Optimization (CGO). IEEE, 86–99.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "LeonardoDeMouraandNikolajBjørner.2008. Z3:AnEfficientSMTSolver.In ProceedingsoftheTheoryandPractice ofSoftware,14thInternationalConference onToolsandAlgorithmsfortheConstructionandAnalysisofSystems (Budapest, Hungary) (TACAS’08/ETAPS’08). Springer-Verlag, Berlin, Heidelberg, 337–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Chunrong Fang, Zixi Liu, Yangyang Shi, Jeff Huang, and Qingkai Shi. 2020. Functionalcodeclonedetectionwithsyntaxandsemanticsfusionlearning.In Proceedingsofthe29thACMSIGSOFTInternationalSymposiumonSoftwareTesting and Analysis. 516–527.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Farzaneh S. Fard, Paul Hollensen, Stuart Mcilory, and Thomas Trappenberg. 2017. Impact of biased mislabeling on learning with deep networks. In 2017 International Joint Conference on Neural Networks (IJCNN). 2652–2657. https: //doi.org/10.1109/IJCNN.2017.7966180",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. 2020. AFL++: Combining Incremental Steps of Fuzzing Research. In Proceedings of the 14th USENIXConferenceonOffensiveTechnologies (WOOT’20).USENIXAssociation, USA, Article 10,1 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Benoît Frénay and Michel Verleysen. 2013. Classification in the presence of label noise:asurvey. IEEEtransactionsonneuralnetworksandlearningsystems 25,5 (2013), 845–869.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Michael Fu and Chakkrit Tantithamthavorn. 2022. Linevul: A transformer-based line-level vulnerability prediction. In Proceedings of the 19th International Conference on Mining Software Repositories. 608–620.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Tianyu Gao, Xingcheng Yao, and Danqi Chen. 2021. SimCSE: Simple Contrastive Learning of Sentence Embeddings. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, Online and Punta Cana, Dominican Republic, 6894–6910. https://doi.org/10.18653/v1/2021.emnlp-main.552",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "John Giorgi, Osvald Nitski, Bo Wang, and Gary Bader. 2020. Declutr: Deepcontrastive learning for unsupervised textual representations. arXiv preprint arXiv:2006.03659 (2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Ross Girshick, Jeff Donahue, Trevor Darrell, and Jitendra Malik. 2014. Rich featurehierarchiesforaccurateobjectdetectionandsemanticsegmentation.In Proceedings of the IEEE conference on computer vision and pattern recognition. 580–587.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "WeiweiGu,AdityaTandon,Yong-YeolAhn,andFilippoRadicchi.2021. Principled approach to the selection of the embedding dimension of networks. Nature Communications 12,1 (2021), 3772.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "NadeeshaanGunasingheandNipunaMarcus.2021. LanguageServerProtocoland Implementation. Springer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Daya Guo, Shuo Ren, Shuai Lu, Zhangyin Feng, Duyu Tang, Shujie Liu, Long Zhou, Nan Duan, Alexey Svyatkovskiy, Shengyu Fu, Michele Tufano, Shao Kun Deng, Colin B. Clement, Dawn Drain, Neel Sundaresan, Jian Yin, Daxin Jiang, andMingZhou.2021. GraphCodeBERT:Pre-trainingCodeRepresentationswithDataFlow.In 9thInternationalConferenceonLearningRepresentations,ICLR2021, VirtualEvent,Austria,May3-7,2021.Ope",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Hazim Hanif, Mohd Hairul Nizam Md Nasir, Mohd Faizal Ab Razak, Ahmad Firdaus, and Nor Badrul Anuar. 2021. The rise of software vulnerability: Taxonomy of software vulnerabilities detection and machine learning approaches. Journal of Network and Computer Applications 179 (2021),103009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "JingxuanHe,GishorSivanrupan,PetarTsankov,andMartinVechev.2021. Learning to Explore Paths for Symbolic Execution. In Proceedings of the 2021 ACMSIGSAC Conference on Computer and Communications Security. 2526–2540.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Kaiming He, Haoqi Fan, Yuxin Wu, Saining Xie, and Ross Girshick. 2020. Momentum contrast for unsupervised visual representation learning. In Proceedings ofthe IEEE/CVFconferenceoncomputervisionandpatternrecognition.9729–9738.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "David Hin, Andrey Kan, Huaming Chen, and M Ali Babar. 2022. LineVD: Statement-level vulnerability detection using graph neural networks. In Proceedings of the 19th International Conference on Mining Software Repositories. 596–607.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "MichaelHind,MichaelBurke, Paul Carini,and Jong-Deok Choi. 1999. Interproceduralpointeraliasanalysis. ACMTransactionsonProgrammingLanguagesand Systems (TOPLAS) 21,4 (1999), 848–894.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "MIrlbecketal .2015. Deconstructingdynamicsymbolicexecution. Dependable Software Systems Engineering 40,2015 (2015),26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "AshishJaiswal,AshwinRameshBabu,MohammadZakiZadeh,DebapriyaBanerjee, and Fillia Makedon. 2020. A survey on contrastive self-supervised learning. Technologies 9, 1 (2020), 2.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "PrannayKhosla,PiotrTeterwak,ChenWang,AaronSarna,YonglongTian,Phillip Isola,AaronMaschinot,CeLiu,andDilipKrishnan.2020. Supervisedcontrastive learning. Advances in Neural Information Processing Systems 33 (2020), 18661– 18673.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "SeulbaeKim,SeunghoonWoo,HeejoLee,andHakjooOh.2017.Vuddy:Ascalable approach for vulnerable code clone discovery. In Proceedings of the 2017 IEEE Symposiumon Security and Privacy (SP). IEEE, 595–614.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Sun Hyoung Kim, Dongrui Zeng, Cong Sun, and Gang Tan. 2022. Binpointer: towards precise, sound, and scalable binary-level pointer analysis. In Proceedings of the 31st ACMSIGPLAN International Conference on Compiler Construction. 169–180.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Chris Lattner.2008. LLVMandClang:Nextgeneration compilertechnology.In TheBSD conference, Vol. 5. 1–20.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Changsheng Li, Handong Ma, Zhao Kang, Ye Yuan, Xiao-Yu Zhang, and Guoren Wang. 2021. On deep unsupervised active learning. In Proceedings of the TwentyNinth International Conference on International Joint Conferences on Artificial Intelligence. 2626–2632.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "GuorenLi,HangZhang,JinmengZhou,WenboShen,YuleiSui,andZhiyunQian. 2023. A hybrid alias analysis and its application to global variable protection in the linux kernel. In 32nd USENIX Security Symposium (USENIX Security 23). 4211–4228.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Ning Li, Martin Shepperd, and Yuchen Guo. 2020. A systematic review of unsupervisedlearningtechniquesforsoftwaredefectprediction. Informationand Software Technology 122 (2020),106287.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "YiLi,ShaohuaWang,TienNNguyen,andSonVanNguyen.2019. Improvingbug detection via context-based code representation learning and attention-based neuralnetworks. ProceedingsoftheACMonProgrammingLanguages 3,OOPSLA (2019), 1–30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Zhuohua Li, Jincheng Wang, Mingshen Sun, and John CS Lui. 2021. MirChecker: detecting bugs in Rust programs via static analysis. In Proceedings of the 2021 ACMSIGSAC Conference on Computer and Communications Security. 2183–2196.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Zhen Li, Deqing Zou, Shouhuai Xu, Hai Jin, Hanchao Qi, and Jie Hu. 2016. VulPecker:anautomatedvulnerabilitydetectionsystembasedoncodesimilarity analysis. In Proceedings of the 32nd Annual Conference on Computer Security Applications. 201–213.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Zhen Li, Deqing Zou, Shouhuai Xu, Xinyu Ou, Hai Jin, Sujuan Wang, Zhijun Deng, and Yuyi Zhong. 2018. VulDeePecker: A Deep Learning-Based System for Vulnerability Detection. Proceedings of the NDSS (2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Guanjun Lin, Sheng Wen, Qing-Long Han, Jun Zhang, and Yang Xiang. 2020. Softwarevulnerabilitydetectionusingdeepneuralnetworks:asurvey. Proc.IEEE 108, 10 (2020), 1825–1848.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "GuanjunLin,JunZhang,Wei Luo,Lei Pan,Olivier DeVel,PaulMontague,and YangXiang.2019. Softwarevulnerabilitydiscoveryvialearningmulti-domain knowledge bases. Proceedings of the IEEE Transactions on Dependable and Secure Computing (2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Shuai Lu, Daya Guo, Shuo Ren, Junjie Huang, Alexey Svyatkovskiy, Ambrosio Blanco, Colin Clement, Dawn Drain, Daxin Jiang, Duyu Tang, et al .2021. CodeXGLUE: A Machine Learning Benchmark Dataset for Code UnderstandingandGeneration.In Thirty-fifthConferenceonNeuralInformationProcessing CombiningStructured Static Code Information and Dynamic Symbolic Traces for Software Vulnerability Prediction ICSE ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Kasper Luckow, Marko Dimjašević, Dimitra Giannakopoulou, Falk Howar, Malte Isberner,TemesghenKahsai,ZvonimirRakamarić,andVishwanathRaman.2016. JDart:adynamicsymbolicanalysisframework.In ToolsandAlgorithmsforthe Construction and Analysis of Systems: 22nd International Conference, TACAS 2016, Held as Part of the European Joint Conferences on Theory and Practice of Software, ETAPS 2016, Eindhoven, Th",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Andrew A Neath and Joseph E Cavanaugh. 2012. The Bayesian information criterion: background, derivation, and applications. Wiley Interdisciplinary Reviews: ComputationalStatistics 4, 2 (2012), 199–203.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Van-Anh Nguyen,DaiQuocNguyen,VanNguyen,TrungLe,QuanHungTran, and Dinh Phung. 2022. ReGVD: Revisiting Graph Neural Networks for VulnerabilityDetection.In Proceedingsofthe44thInternationalConferenceonSoftware EngineeringCompanion (ICSE ’22 Companion).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "NIST. [n.d.]. Software Assurance Reference Dataset Project. https://samate.nist. gov/SRD/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Hristina Palikarevaand CristianCadar. 2013. Multi-solversupport insymbolic execution. In Computer Aided Verification: 25th International Conference, CAV 2013, Saint Petersburg,Russia, July 13-19, 2013.Proceedings 25. Springer, 53–68.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "XiaojiangPeng,KaiWang,ZhaoyangZeng,QingLi,JianfeiYang,andYuQiao. 2020. Suppressing mislabeled data via grouping and self-attention. In Computer Vision–ECCV2020:16thEuropeanConference,Glasgow,UK,August23–28,2020, Proceedings, Part XVI 16. Springer, 786–802.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "HenningPerl,SergejDechand,MatthewSmith,DanielArp,FabianYamaguchi, Konrad Rieck, Sascha Fahl, and Yasemin Acar. 2015. Vccfinder: Finding potential vulnerabilities in open-source projects to assist code audits. In Proceedings of the 22nd ACMSIGSAC Conference on Computer and Communications Security. 426–437.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "GabrielPoesiaandFernandoMagnoQuintãoPereira.2020. Dynamicdispatch of context-sensitive optimizations. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Zichao Qi, Fan Long, Sara Achour, and Martin Rinard. 2015. An analysis of patch plausibility and correctness for generate-and-validate patch generation systems.In Proceedingsofthe2015InternationalSymposiumonSoftwareTesting and Analysis. 24–36.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Leslie Rice, Eric Wong, and Zico Kolter. 2020. Overfitting in adversarially robust deep learning. In Proceedings of the 37th International Conference on Machine Learning (Proceedings of Machine Learning Research, Vol. 119), Hal Daumé III and AartiSingh(Eds.).PMLR,8093–8104. https://proceedings.mlr.press/v119/rice20a. html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "ManuelRiggerandZhendongSu.2020. TestingDatabaseEnginesviaPivoted Query Synthesis.. In OSDI, Vol. 20. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 72,
+      "text": "Iqbal H Sarker. 2021. Machine learning: Algorithms, real-world applications and research directions. SN Computer Science 2, 3 (2021), 1–21.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Koushik Sen, Swaroop Kalasapur, Tasneem Brutch, and Simon Gibbs. 2013. Jalangi: A Selective Record-Replay and Dynamic Analysis Framework for JavaScript.In Proceedingsofthe20139thJointMeetingonFoundationsofSoftware Engineering (Saint Petersburg, Russia) (ESEC/FSE 2013). Association for ComputingMachinery,NewYork,NY,USA,488–498. https://doi.org/10.1145/2491411. 2491447",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "RidwanShariffdeen,YannicNoller,LarsGrunske,andAbhikRoychoudhury.2021. Concolic program repair. In Proceedings of the 42nd ACMSIGPLAN International Conference on Programming Language Design and Implementation. 390–405.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Ke Shi, Yang Lu, Jingfei Chang, and Zhen Wei. 2020. PathPair2Vec: An AST path pair-basedcoderepresentationmethodfordefectprediction. JournalofComputer Languages 59 (2020), 100979. https://doi.org/10.1016/j.cola.2020.100979",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Wilson L Taylor. 1953. “Cloze procedure”: A new tool for measuring readability. Journalism quarterly 30, 4 (1953), 415–433.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "David A Tomassi and Cindy Rubio-González. 2021. On the Real-World Effectiveness of Static Bug Detectors at Finding Null Pointer Exceptions. In 2021 36th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 292–303.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all youneed. Advances in neural information processing systems 30 (2017).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Huanting Wang, Zhanyong Tang, Cheng Zhang, Jiaqi Zhao, Chris Cummins, Hugh Leather, and Zheng Wang. 2022. Automating reinforcement learning architecturedesignforcodeoptimization.In Proceedingsofthe31stACMSIGPLAN InternationalConference on Compiler Construction. 129–143.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "H. Wang, G. Ye, Z. Tang, S. H. Tan, S. Huang, D. Fang, Y. Feng, L. Bian, and Z. Wang.2021. CombiningGraph-BasedLearningWithAutomatedDataCollection for Code Vulnerability Detection. IEEE Transactions on Information Forensics and Security16 (2021), 1943–1958. https://doi.org/10.1109/TIFS.2020.3044773",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "Ke Wang and Zhendong Su. 2020. Blended, Precise Semantic Program Embeddings.In Proceedingsofthe41stACMSIGPLANConferenceonProgramming LanguageDesignand Implementation (London,UK) (PLDI2020). Association for ComputingMachinery,NewYork,NY,USA,121–134. https://doi.org/10.1145/3385412.3385999",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "ZhengWangandMichaelO’Boyle.2018. Machinelearningincompileroptimization.Proc. IEEE 106,11 (2018), 1879–1901.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "GuixinYe,ZhanyongTang,ShinHweiTan,SongfangHuang,DingyiFang,Xiaoyang Sun, Lizhong Bian, Haibo Wang, and Zheng Wang. 2021. Automatedconformance testing for JavaScript engines via deep compiler fuzzing. In Proceedings of the 42nd ACMSIGPLAN International Conference on Programming Language Designand Implementation. 435–450.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "GuixinYe,ZhanyongTang,HuantingWang,DingyiFang,JianbinFang,SongfangHuang,andZhengWang.2020. Deepprogramstructuremodelingthroughmultirelationalgraph-basedlearning.In ProceedingsoftheACMInternationalconference on parallel architectures and compilation techniques. 111–123.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "ZhengranZeng,YuqunZhang,HaotianZhang,andLingmingZhang.2021. Deep just-in-timedefectprediction:howfararewe?.In Proceedingsofthe30thACM SIGSOFT International Symposium on Software Testing and Analysis. 427–438.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Dan Zhang, Qing-Guo Wang, Gang Feng, Yang Shi, and Athanasios V Vasilakos. 2021. Asurveyonattackdetection,estimationandcontrolofindustrialcyber– physical systems. ISA transactions 116 (2021),1–16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "Yu Zhao, Tingting Yu, Ting Su, Yang Liu, Wei Zheng, Jingzhi Zhang, and William GJ Halfond. 2019. Recdroid: automatically reproducing android application crashes from bug reports. In Software Engineering (ICSE). IEEE, 128–139.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "Yaqin Zhou, Shangqing Liu, Jingkai Siow, Xiaoning Du, and Yang Liu. 2019. Devign:Effectivevulnerabilityidentificationbylearningcomprehensiveprogram semanticsviagraphneuralnetworks. In ProceedingsoftheAdvancesinNeural InformationProcessing Systems. 10197–10207.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 71,
+      "text": "ManuelRiggerandZhendongSu.2020. TestingDatabaseEnginesviaPivoted Query Synthesis.. In OSDI, Vol. 20. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 69,71]",
+      "technique": "pqs",
+      "sentence": "These projects contain 35 CVEs reported by independent users, which were also used by prior work [ 69,71].",
+      "context_before": "n Sec. 5.1). This informationenhances precision by reducing the false positive rate and helps discover more vulnerabilities with a higher true positive rate than static information alone, resulting in a higher recall. Among thebaseline methods, LIGER performs best, but its F1 score is 10.4%lower than that of Concoction. This shows that Concoction strikes a better balance between false and negative positives, leveraging the advantages of structured static source code information. 5.3 Comparison on Known CVEs We compare Concoction to the baselines on three open-source projects listed in Table 2.",
+      "context_after": "WeapplyallmethodstofunctionsassociatedwithaCVEandusethe reported CVEs to compute evaluation metrics (Sec. 4.3). To ensure a fair comparison, we train all methods, including Concoction, onthesametrainingdataset,butweexcludetheseprojectsfrom the training data. For the dynamic methods listed in Sec. 4.2, we allocate 200 hoursof machine time for each project. 5.3.1 Vulnerabilitiesidentified. Table5demonstratesConcoction’s advantagesoverothermethodsinidentifyingvulnerabilitiesacross thethreeopen-sourceprojectsevaluated.Concoctionachieved 100% precision and 89% recall, correctly detecting 31 out of",
+      "section": "5.3 Comparison on Known CVEs",
+      "page": 9,
+      "char_offset": 49165,
+      "cited_reference": {
+        "number": 71,
+        "text": "ManuelRiggerandZhendongSu.2020. TestingDatabaseEnginesviaPivoted Query Synthesis.. In OSDI, Vol. 20. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "Concoction learns program representations for bug detection by combining static source code information with dynamic symbolic execution traces, addressing the limitation that existing deep-learning representations either miss precise semantics or do not scale. Unsupervised active learning picks a subset of important paths for trace collection, and a focused symbolic execution implementation keeps the cost down. Applied to C programs from 20 open-source projects it found 54 unique vulnerabilities and 37 new CVE IDs.",
+    "narrative": "One citation, noting that the CVEs used for evaluation were also used by prior work, SQLancer's among the references.",
+    "roles": {
+      "M1": "incidental"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3597926_3598046.json b/_data/papers/paper_doi_10_1145_3597926_3598046.json
new file mode 100644
index 0000000..8ea06cb
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3597926_3598046.json
@@ -0,0 +1,532 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:26:40Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3597926.3598046",
+    "title": "GDsmith: Detecting Bugs in Cypher Graph Database Engines",
+    "authors": [
+      "Ziyue Hua",
+      "Weisheng Lin",
+      "Luyao Ren",
+      "Zongyang Li",
+      "Lu Zhang",
+      "Wenpin Jiao",
+      "Tao Xie"
+    ],
+    "year": 2023,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3597926.3598046",
+    "arxiv_id": null,
+    "s2_paper_id": "6c755fc901d0b41a5d73c265f64a5aacf62e83b8",
+    "url": "https://doi.org/10.1145/3597926.3598046",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3597926.3598046",
+      "retrieved_at": "2026-09-10T15:26:40Z",
+      "chars": 68528,
+      "content_sha256": "sha256:461e59022159debff7e6bf5cc5dd56ff0dfe1eef565002f5cbff0982174c042a"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2767
+      },
+      {
+        "number": "2.1",
+        "title": "Graph Model in Cypher Databases",
+        "start": 11828
+      },
+      {
+        "number": "2.2",
+        "title": "The Cypher Query Language",
+        "start": 15281
+      },
+      {
+        "number": "2.3",
+        "title": "Differential Testing of Graph Database",
+        "start": 19829
+      },
+      {
+        "number": "2.4",
+        "title": "Motivation",
+        "start": 21088
+      },
+      {
+        "number": "3.1",
+        "title": "Schema and Property Graph Generation",
+        "start": 27168
+      },
+      {
+        "number": "3.2",
+        "title": "Query Generation",
+        "start": 28523
+      },
+      {
+        "number": "3.2.1",
+        "title": "Overview of the Query Generation Process. GDsmith includes",
+        "start": 29259
+      },
+      {
+        "number": "3.2.3",
+        "title": "Data-Guided Condition Generation. To generate queries with",
+        "start": 35063
+      },
+      {
+        "number": "3.3",
+        "title": "Bug Detection",
+        "start": 39759
+      },
+      {
+        "number": "4.1",
+        "title": "Evaluation Setup",
+        "start": 42291
+      },
+      {
+        "number": "4.1.1",
+        "title": "Subjects. We select three popular real-world graph database",
+        "start": 42312
+      },
+      {
+        "number": "4.1.2",
+        "title": "Implementation. We implement the GDsmith prototype with",
+        "start": 43271
+      },
+      {
+        "number": "4.1.3",
+        "title": "Baselines. There is no applicable baseline to which we can",
+        "start": 44288
+      },
+      {
+        "number": "4.1.4",
+        "title": "Metrics. To measure the effectiveness of GDsmith, we design",
+        "start": 45247
+      },
+      {
+        "number": "4.2",
+        "title": "RQ1/RQ2: Effectiveness",
+        "start": 47654
+      },
+      {
+        "number": "4.2.1",
+        "title": "Wrong-Result Bug Detection. Table 3 shows the 𝑁𝐷and",
+        "start": 50348
+      },
+      {
+        "number": "4.2.2",
+        "title": "Non-Empty-Result Query Ratio. We run the three approaches",
+        "start": 51537
+      },
+      {
+        "number": "4.3",
+        "title": "RQ3: Practicability",
+        "start": 52916
+      },
+      {
+        "number": "4.3.2",
+        "title": "Example Bugs. We next show the following two examples",
+        "start": 53819
+      },
+      {
+        "number": "4.4",
+        "title": "Threats to Validity",
+        "start": 56265
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Ibrahim Abdelaziz, Essam Mansour, Mourad Ouzzani, Ashraf Aboulnaga, and Panos Kalnis. 2017. Query Optimizations over Decentralized RDF Graphs. In Proceedings of the 33rd International Conference on Data Engineering. 139–142. https://doi.org/10.1109/ICDE.2017.59",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Junjie Chen, Wenxiang Hu, Dan Hao, Yingfei Xiong, Hongyu Zhang, Lu Zhang, and Bing Xie. 2016. An Empirical Comparison of Compiler Testing Techniques. In Proceedings of the 38th International Conference on Software Engineering. 180–190. https://doi.org/10.1145/2884781.2884878",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Peter Pin-Shan Chen. 1976. The Entity-Relationship Model—Toward a Unified View of Data. ACM Transactions on Database Systems (1976), 9–36. https: //doi.org/10.1145/320434.320440 173 ISSTA ’23, July 17–21, 2023, Seattle, WA, USA Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Tsong Yueh Chen, Shing Chi Cheung, and Siu Ming Yiu. 1998. Metamorphic Testing: a New Approach for Generating Next Test Cases. Technical Report HKUST-CS98-01 (1998).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Xinyue Chen, Chenglong Wang, and Alvin Cheung. 2020. Testing Query Execution Engines with Mutations. In Proceedings of the 8th International Workshop on Testing Database Systems. 6:1–6:5. https://doi.org/10.1145/3395032.3395322",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "The Apache Software Foundation. 2022. Cypher for Gremlin. https: //github.com/opencypher/cypher-for-gremlin/tree/master/tinkerpop/cyphergremlin-server-client",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "The Apache Software Foundation. 2022. Gremlin Query Language. https: //tinkerpop.apache.org/gremlin.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Martin Schuster, Petra Selmer, and Andrés Taylor. 2018. Formal Semantics of the Language Cypher. arXiv preprint arXiv:1802.09984 (2018). http://arxiv.org/abs/1802.09984",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and Andrés Taylor. 2018. Cypher: An Evolving Query Language for Property Graphs. InProceedings of the 2018 International Conference on Management of Data. 1433– 1445. https://doi.org/10.1145/3183713.3190657",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Bogdan Ghit, Nicolás Poggi, Josh Rosen, Reynold Xin, and Peter A. Boncz. 2020. SparkFuzz: Searching Correctness Regressions in Modern Query Engines. In Proceedings of the 8th International Workshop on Testing Database Systems. 1:1–1:6. https://doi.org/10.1145/3395032.3395327",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Lior Kogan. 2017. V1: A Visual Query Language for Property Graphs. arXiv preprint arXiv:1710.04470 (2017). http://arxiv.org/abs/1710.04470",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Takahiro Konno, Runhe Huang, Tao Ban, and Chuanhe Huang. 2017. Goods Recommendation Based on Retail Knowledge in a Neo4j Graph Database Combined with an Inference Mechanism Implemented in Jess. In 2017 IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Computed, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation. 1–8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software. Digital Technical Journal (1998), 100–107. http://www.hpl.hp.com/hpjournal/dtj/vol10num1/ vol10num1art9.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Memgraph. 2022. Memgraph: Frictionless, Innovative, Graph Applications. https: //memgraph.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Neo4j. 2022. The Fastest Path To Graph Productivity: Neo4j Graph Database. https://neo4j.com/product/neo4j-graph-database/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "The openCypher Implementers Group. 2022. Cypher Query Language Reference, Version 9. https://s3.amazonaws.com/artifacts.opencypher.org/openCypher9.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "RedisGraph. 2022. RedisGrapha Graph Database Module for Redis. https: //oss.redis.com/redisgraph/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Manuel Rigger. 2022. SQLancer: Detecting Logic Bugs in DBMS. https://github. com/sqlancer/sqlancer",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 19,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152. https://doi.org/10.1145/ 3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages (2020),211:1–211:30. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation. 667–682. https://www.usenix.org/conference/ osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "Andreas Seltenreich. 2022. Bug Squashing with SQLsmith. https://github.com/ anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Sudipta Sen, Akash Mehta, Runa Ganguli, and Soumya Sen. 2021. Recommendation of Influenced Products Using Association Rule Mining: Neo4j as a Case Study. SN Computer Science (2021), 74. https://doi.org/10.1007/s42979-021-00460-8",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In Proceedings of the 24rd International Conference on Very Large Data Bases. 618–622. http: //www.vldb.org/conf/1998/p618.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "solid IT gmbh. 2022. DB-Engines Ranking of Graph DBMS. https://db-engines. com/en/ranking/graph+ dbms",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Jian Wang, Ke Wang, Jing Li, Jianmin Jiang, Yanfei Wang, Jing Mei, and Shaochun Li. 2020. Accelerating Epidemiological Investigation Analysis by Using NLP and Knowledge Reasoning: A Case Study on COVID-19. In2020 American Medical Informatics Association Annual Symposium. 1258– 1267. https://knowledge.amia.org/72332-amia-1.4602255/t003-1.4606204/t0031.4606205/3417206-1.4606266/3415131-1.4606263",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In Proceedings of the 43rd International Conference on Software Engineering: Software Engineering in Practice. 328–337. https://doi.org/10.1109/ ICSE-SEIP52600.2021.00042",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Ran Wang, Zhengyi Yang, Wenjie Zhang, and Xuemin Lin. 2020. An Empirical Study on Recent Graph Database Systems. In Knowledge Science, Engineering and Management. 328–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Rui Yang, Yingying Zheng, Lei Tang, Wensheng Dou, Wei Wang, and Jun Wei. 2023. Randomized Differential Testing of RDF Stores. In Proceedings of the 45th International Conference on Software Engineering: Demonstrations. 136–140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding And Understanding Bugs in C Compilers. In Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation. 283–294. https://doi.org/10.1145/1993498.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Michal Zalewski. 2022. American Fuzzy Lop (2.52b). https://lcamtuf.coredump. cx/afl/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis. 302–313. https://doi.org/10.1145/3533767.3534409",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 955–970. https://doi.org/10. 1145/3372297.3417260 Received 2023-02-16; accepted 2023-05-03 174",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 18,
+      "text": "Manuel Rigger. 2022. SQLancer: Detecting Logic Bugs in DBMS. https://github. com/sqlancer/sqlancer",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 19,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152. https://doi.org/10.1145/ 3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages (2020),211:1–211:30. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 21,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation. 667–682. https://www.usenix.org/conference/ osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Its framework is derived from SQLancer [ 18] (which is a tool to automatically test relational database engines).",
+      "context_before": "tabase to use sparse matrices to represent the adjacency matrix in graphs and linear algebra to query a graph. (3) Memgraph [ 14] is a streaming graph application platform and leverages an in-memory-first architecture. We test the released versions of the Neo4j Community Edition from 3.5 to 4.4, the released versions of RedisGraph 2.8, and the released version 2.4 of the Memgraph Community Edition. All of these subjects are downloaded from their official repositories without any modification. 4.1.2 Implementation. We implement the GDsmith prototype with over 22K non-comment lines of Java code.",
+      "context_after": "GDsmith uses Neo4j Java Driver 4.1.1 to connect and interact with Neo4j and Memgraph, and uses JRedisGraph 2.5.1 to connect and interact with RedisGraph. The default length of clauses is set to nine. The default number of nodes in a graph is set to 128. The default maximum number of patterns in each MATCH clause is set to four. The default maximum depth of expression trees is set to two. Some graph database engines implement only a subset of the Cypher language. When conducting cross-engine differential testing, we configure GDsmith in advance so that all Cypher queries generated by GDsmith d",
+      "section": "4.1.2 Implementation. We implement the GDsmith prototype with",
+      "page": 8,
+      "char_offset": 43373,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [ 21] detects wrong-result bugs by checking whether a specific record is fetched correctly.",
+      "context_before": "l analysis. Ratel [ 27] is an enterprise-level fuzzer that improves the feedback precision, enhances the robustness of input generation, and performs an on-line investigation on the root cause of bugs with its industry-oriented design. The aforementioned approaches can detect only crashing bugs in relational database engines. To detect wrong-result bugs, RAGS [ 24] generates and executes SQL queries in multiple relational database engines, and meanwhile observes differences in the output sets. Any inconsistency among results indicates that at least one relational database engine contains bugs.",
+      "context_after": "NoREC [19] detects bugs in a relational database engine by applying a semantic-preserving transformation to a given SQL query to disable the engine’s optimizations and addresses PQS’ high implementation effort. TLP [ 20] derives multiple SQL queries thatcompute a partial result of the initial query. By using a composition operator, the partitions can be combined to yield the same result as the original query; if the result differs, a bug in the relational database engine has been detected. MutaSQL [ 5] generates test cases by mutating a SQL query over a database instance into a semantically e",
+      "section": "4.4 Threats to Validity",
+      "page": 11,
+      "char_offset": 58097,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [19] detects bugs in a relational database engine by applying a semantic-preserving transformation to a given SQL query to disable the engine’s optimizations and addresses PQS’ high implementation effort.",
+      "context_before": "nces the robustness of input generation, and performs an on-line investigation on the root cause of bugs with its industry-oriented design. The aforementioned approaches can detect only crashing bugs in relational database engines. To detect wrong-result bugs, RAGS [ 24] generates and executes SQL queries in multiple relational database engines, and meanwhile observes differences in the output sets. Any inconsistency among results indicates that at least one relational database engine contains bugs. PQS [ 21] detects wrong-result bugs by checking whether a specific record is fetched correctly.",
+      "context_after": "TLP [ 20] derives multiple SQL queries thatcompute a partial result of the initial query. By using a composition operator, the partitions can be combined to yield the same result as the original query; if the result differs, a bug in the relational database engine has been detected. MutaSQL [ 5] generates test cases by mutating a SQL query over a database instance into a semantically equivalent query mutant, and checks the results returned by the relational database engine under test. Compared with these approaches, GDsmith includes our skeletonbased completion technique to ensure that each r",
+      "section": "4.4 Threats to Validity",
+      "page": 11,
+      "char_offset": 58193,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [ 20] derives multiple SQL queries thatcompute a partial result of the initial query.",
+      "context_before": "al database engines. To detect wrong-result bugs, RAGS [ 24] generates and executes SQL queries in multiple relational database engines, and meanwhile observes differences in the output sets. Any inconsistency among results indicates that at least one relational database engine contains bugs. PQS [ 21] detects wrong-result bugs by checking whether a specific record is fetched correctly. NoREC [19] detects bugs in a relational database engine by applying a semantic-preserving transformation to a given SQL query to disable the engine’s optimizations and addresses PQS’ high implementation effort.",
+      "context_after": "By using a composition operator, the partitions can be combined to yield the same result as the original query; if the result differs, a bug in the relational database engine has been detected. MutaSQL [ 5] generates test cases by mutating a SQL query over a database instance into a semantically equivalent query mutant, and checks the results returned by the relational database engine under test. Compared with these approaches, GDsmith includes our skeletonbased completion technique to ensure that each randomly generated Cypher query satisfies the semantic requirements. GDsmith also includes",
+      "section": "4.4 Threats to Validity",
+      "page": 11,
+      "char_offset": 58404,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/ddaa2000/GDsmith",
+    "markers": [
+      "renamed_sqlancer_package",
+      "sqlancer_source_content_match"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/ddaa2000/GDsmith",
+        "source_type": "github_repository",
+        "excerpt": "# GDsmith",
+        "note": "Repository is named after GDsmith, the tool this paper says it built, and is about database testing.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/ddaa2000/GDsmith/blob/master/src/main/java/org/example/gdsmith/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.ALPHANUMERIC;",
+        "excerpt_is_verbatim": true,
+        "note": "src/main/java/org/example/gdsmith/Randomly.java is SQLancer's Randomly.java, with the package renamed to org.example.gdsmith (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+        "content_sha256": "sha256:d09fa9dddb2efa23e9184c0a8e2c71c39cdaffe6bde161da903434f62c4c4711",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "uses_infrastructure": [
+        "M1"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:20:36Z",
+    "is_model_written": true,
+    "summary": "GDsmith tests graph database engines that use Cypher. Its central difficulty is that random test generation struggles to produce property graphs and queries complex enough to return non-empty results, which is what wrong-result bugs require. GDsmith ensures every generated query meets the semantic requirements, and raises the chance of complex non-empty results with graph-guided generation of pattern combinations and data-guided generation of conditions. It detected bugs across three popular open-source graph engines.",
+    "narrative": "GDsmith says plainly that its framework is derived from SQLancer, carrying that tool's approach from relational engines to Cypher, and its artifact holds SQLancer's source under a renamed package. PQS, NoREC and TLP are summarised as the relational oracles that came before it.",
+    "roles": {
+      "M1": "reuse_implementation",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M1"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "Its framework is derived from SQLancer [ 18] (which is a tool to automatically test relational database engines).",
+            "section": "4.1.2 Implementation. We implement the GDsmith prototype with",
+            "page": 8
+          }
+        ],
+        "reasoning": "M1 states the framework is derived from SQLancer. The artifact carrying SQLancer source under a renamed package agrees.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3597926_3598052.json b/_data/papers/paper_doi_10_1145_3597926_3598052.json
new file mode 100644
index 0000000..0e00bb4
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3597926_3598052.json
@@ -0,0 +1,516 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:37:15Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3597926.3598052",
+    "title": "Dependency-Aware Metamorphic Testing of Datalog Engines",
+    "authors": [
+      "Muhammad Numair Mansur",
+      "Valentin Wüstholz",
+      "M. Christakis"
+    ],
+    "year": 2023,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3597926.3598052",
+    "arxiv_id": null,
+    "s2_paper_id": "df170dd0f26ec903ee0a8d0ad89305ec60a125c5",
+    "url": "https://doi.org/10.1145/3597926.3598052",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3597926.3598052",
+      "retrieved_at": "2026-09-08T14:37:15Z",
+      "chars": 67080,
+      "content_sha256": "sha256:c7239f3ae917c965ae41af24b0a89f7399bcfdeb319f53c2f856f4295acc6f4e"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1484
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND",
+        "start": 6822
+      },
+      {
+        "number": "2.1",
+        "title": "DatalogPrograms",
+        "start": 6930
+      },
+      {
+        "number": "2.2",
+        "title": "Precedence Graphs",
+        "start": 10093
+      },
+      {
+        "number": "3",
+        "title": "OVERVIEW",
+        "start": 11238
+      },
+      {
+        "number": "4",
+        "title": "GRAPH ANNOTATOR",
+        "start": 15379
+      },
+      {
+        "number": "5",
+        "title": "GRAPH TRANSFORMER",
+        "start": 23242
+      },
+      {
+        "number": "5.1",
+        "title": "GraphRewriteRules",
+        "start": 23464
+      },
+      {
+        "number": "5.2",
+        "title": "SpecifyingMetamorphicTransformations",
+        "start": 28839
+      },
+      {
+        "number": "5.3",
+        "title": "Example MetamorphicTransformations",
+        "start": 31954
+      },
+      {
+        "number": "6",
+        "title": "IMPLEMENTATION",
+        "start": 41349
+      },
+      {
+        "number": "7.1",
+        "title": "Setup",
+        "start": 45166
+      },
+      {
+        "number": "7.2",
+        "title": "Results",
+        "start": 46241
+      },
+      {
+        "number": "7.3",
+        "title": "Threatsto Validity",
+        "start": 53556
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Paolo Baldan, Andrea Corradini, and Barbara König. 2001. A Static Analysis Technique for Graph Transformation Systems. In CONCUR (LNCS, Vol. 2154). Springer, 381–395.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "PaoloBaldan andBarbara König. 2002. ApproximatingtheBehaviourof Graph TransformationSystems. In ICGT (LNCS, Vol.2505). Springer, 14–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Earl T. Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. 2015. TheOracle Problem inSoftware Testing: A Survey. TSE41(2015), 507–525. Issue 5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Aaron Bembenek, Michael Greenberg, and Stephen Chong. 2020. Formulog: Datalog for SMT-Based Static Analysis. In OOPSLA. ACM,141:1–141:31.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Dmitry Blotsky, Federico Mora, Murphy Berzish, Yunhui Zheng, Ifaz Kabir, and Vijay Ganesh. 2018. StringFuzz: A Fuzzer for String Solvers. In CAV (LNCS, Vol.10982). Springer, 45–51.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Martin Bravenboer and Yannis Smaragdakis. 2009. Strictly Declarative Speci/f_icationof Sophisticated Points-ToAnalyses. In OOPSLA. ACM,243–262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Neville Brent, Lexiand Grech, Si/f_is Lagouvardos, Bernhard Scholz, and Yannis Smaragdakis.2020. Ethainter:ASmartContractSecurityAnalyzerforComposite Vulnerabilities.In PLDI. ACM,454–469.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Danilo Bruschi, Lorenzo Martignoni, and Mattia Monga. 2006. Detecting SelfMutating Malware Using Control-Flow Graph Matching. In DIMVA (LNCS, Vol.4064). Springer, 129–143.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Alexandra Bugariu and Peter Müller. 2020. Automatically Testing String Solvers. InICSE. ACM,1459–1470.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Alexandra Bugariu, Valentin Wüstholz, Maria Christakis, and Peter Müller. 2018. Automatically Testing Implementations of Numerical Abstract Domains. In ASE. ACM,768–778.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "CristianCadarandAlastairF.Donaldson.2016. AnalysingtheProgramAnalyser. InICSE. ACM,765–768.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "W.K.Chan,S.C.Cheung,andKarlR.P.H.Leung.2005. TowardsaMetamorphic Testing Methodology for Service-Oriented Software Applications. In QSIC. IEEE Computer Society, 470–476.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "JunjieChen,JibeshPatra,MichaelPradel,YingfeiXiong,HongyuZhang,Dan Hao,andLuZhang.2020. ASurveyofCompilerTesting. Comput.Surv. 53(2020), 4:1–4:36. Issue 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "J.-L. Chen, F.-J. Wang, and Y.-L. Chen. 1997. An Object-Oriented Dependency Graphfor ProgramSlicing. In TOOLS. IEEE Computer Society, 121–130.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "TsongYuehChen,S.C.Cheung,andSiu-MingYiu.1998. MetamorphicTesting:A NewApproachforGeneratingNextTestCases. TechnicalReportHKUST–CS98–01. HKUST.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Distributed and Parallel Systems Group at the University of Innsbruck. 2012. Insieme Compiler. http://insieme-compiler.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "AlastairF.Donaldson,HuguesEvrard,AndreiLascu,andPaulThomson.2017. Automated Testing of Graphics Shader Compilers. PACMPL1 (2017), 93:1–93:29. Issue OOPSLA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Antonio Flores-Montoya and Eric M. Schulte. 2020. Datalog Disassembly. In Security. USENIX, 1075–1092.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Neville Grech, Michael Kong, Anton Jurisevic, Lexi Brent, Bernhard Scholz, and Yannis Smaragdakis. 2018. MadMax: Surviving Out-of-Gas Conditions in EthereumSmartContracts. PACMPL 2 (2018), 116:1–116:27. Issue OOPSLA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Sergio Greco and Cristian Molinaro. 2015. Datalog and Logic Databases. Morgan & Claypool.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Krystof Hoder, Nikolaj Bjørner, and Leonardo de Moura. 2011. /u1D707Z—An Efficient Engine for Fixed Points with Constraints. In CAV (LNCS, Vol. 6806). Springer, 457–462.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Susan Horwitz and Thomas W. Reps. 1992. The Use of Program Dependence Graphs in SoftwareEngineering.In ICSE. ACM,392–411.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Jiani Huang, Ziyang Li, Binghong Chen, Karan Samel, Mayur Naik, Le Song, andXujieSi.2021. Scallop:FromProbabilisticDeductiveDatabasestoScalable DifferentiableReasoning.In NeurIPS. 25134–25145.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Herbert Jordan, Bernhard Scholz, and Pavle Subotic. 2016. Soufflé: On Synthesis of ProgramAnalyzers. In CAV (LNCS,Vol.9780). Springer, 422–430.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "TimotejKapusandCristianCadar.2017.AutomaticTestingofSymbolicExecution EnginesviaProgramGenerationandDifferentialTesting.In ASE.IEEEComputer Society, 590–600.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Christian Klinger, Maria Christakis, and Valentin Wüstholz. 2019. Differentially TestingSoundnessandPrecisionofProgramAnalyzers.In ISSTA.ACM,239–250.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler Validation via EquivalenceModulo Inputs.In PLDI. ACM,216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "VuLe,ChengnianSun,andZhendongSu.2015. FindingDeepCompilerBugsvia Guided StochasticProgramMutation. In OOPSLA. ACM,386–399.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "ChristopherLidbury,AndreiLascu,NathanChong,andAlastairF.Donaldson. 2015. Many-CoreCompilerFuzzing. In PLDI. ACM,65–76.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "VladimirLifschitz.1988. OntheDeclarativeSemanticsofLogicProgramswith Negation. In FoundationsofDeductiveDatabasesandLogicProgramming .Morgan Kaufmann,177–192.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "JiaweiLiu, JinkunLin,Fabian Ruffy, ChengTan,JinyangLi,AurojitPanda, and LingmingZhang.2022. FindingDeep-LearningCompilationBugswithNNSmith. CoRRabs/2207.13066 (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Vsevolod Livinskii, Dmitry Babokin, and John Regehr. 2020. Random Testing for C and C++ Compilers with YARPGen. PACMPL4 (2020), 196:1–196:25. Issue OOPSLA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Magnus Madsen and Ondrej Lhoták. 2018. Safe and Sound Program Analysis with FLIX.In ISSTA. ACM,38–48.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Muhammad Numair Mansur, Maria Christakis, and Valentin Wüstholz. 2021. MetamorphicTesting of Datalog Engines. In ESEC/FSE. ACM,639–650.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Muhammad NumairMansur, Maria Christakis, Valentin Wüstholz, and Fuyuan Zhang. 2020. Detecting Critical Bugs in SMT Solvers Using Blackbox Mutational Fuzzing. In ESEC/FSE. ACM,701–712.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software. Digital Technical Journal10(1998), 100–107. Issue 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "John M. McQuillan. 1977. Graph Theory Applied to Optimal Connectivity in Computer Networks. Comput. Commun. Rev. 7 (1977), 13–41. Issue 2.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Mayur Naik, Alex Aiken, and John Whaley. 2006. Effective Static Race Detection for Java. In PLDI. ACM,308–319.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "MatteoPaltenghiand MichaelPradel.2022. Bugsin Quantum ComputingPlatforms: AnEmpiricalStudy. PACMPL 6 (2022), 1–27. Issue OOPSLA. 246 ISSTA ’23, July 17–21, 2023,Sea/t_tle, WA,USA MuhammadNumair Mansur,Valentin Wüstholz,andMaria Christakis",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Georgios A. Pavlopoulos, Maria Secrier, Charalampos N. Moschopoulos, Theodoros G. Soldatos, Sophia Kossida, Jan Aerts, Reinhard Schneider, and PantelisG.Bagos.2011.UsingGraphTheorytoAnalyzeBiologicalNetworks. BioData Min.4 (2011). Issue 10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Karl-HeinzPennemann.2008. DevelopmentofCorrectGraphTransformation Systems. In ICGT (LNCS,Vol.5214). Springer, 508–510.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "DetlefPlump.1995. OnTerminationofGraphRewriting.In WG (LNCS,Vol.1017). Springer, 88–100.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "ManuelRiggerandZhendongSu.2020. FindingBugsinDatabaseSystemsvia QueryPartitioning. PACMPL 4 (2020), 211:1–211:30. Issue OOPSLA.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "LeonidRyzhykandMihaiBudiu.2019. DifferentialDatalog.In Datalog (CEUR, Vol.2368). CEUR-WS.org, 56–67.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Arash Sahebolamri, Thomas Gilray, and Kristopher K. Micinski. 2022. Seamless DeductiveInferenceviaMacros.In CC. ACM,77–88.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Sergio Segura, Juan C. Alonso, Alberto Martin-Lopez, Amador Durán, Javier Troya,andAntonioRuiz-Cortés.2022. AutomatedGenerationofMetamorphic Relations for Query-Based Systems. In MET@ICSE. ACM,48–55.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Sergio Segura, Gordon Fraser, Ana B. Sánchez, and Antonio Ruiz Cortés. 2016. A SurveyonMetamorphicTesting. TSE42(2016), 805–824. Issue 9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "ChengnianSun,VuLe,andZhendongSu.2016. FindingandAnalyzingCompiler WarningDefects.In ICSE. ACM,203–213.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. 2016. Finding Compiler Bugs via Live CodeMutation. In OOPSLA. ACM,849–863.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Jubi Taneja, ZhengyangLiu, andJohn Regehr.2020. Testing Static Analysesfor Precisionand Soundness. In CGO. ACM,81–93.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Petar Tsankov, Andrei Marian Dan, Dana Drachsler-Cohen, Arthur Gervais, Florian Bünzli, and Martin T. Vechev. 2018. Securify: Practical Security Analysis of SmartContracts. In CCS. ACM,67–82.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Muhammad Usman, Wenxi Wang, and Sarfraz Khurshid. 2020. TestMC: TestingModelCountersUsingDifferentialandMetamorphicTesting.In ASE.IEEE Computer Society, 709–721.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "JohnWhaley,DzintarsAvots,MichaelCarbin,andMonica S.Lam.2005. Using Datalog with Binary Decision Diagrams for Program Analysis. In APLAS (LNCS, Vol.3780). Springer, 97–118.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2020. On the Unusual EffectivenessofType-AwareOperatorMutationsforTestingSMTSolvers. PACMPL 4 (2020), 193:1–193:25. Issue OOPSLA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2020. Validating SMT SolversviaSemantic Fusion.In PLDI. ACM,718–730.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Dongwei Xiao, Zhibo Liu, Yuanyuan Yuan, Qi Pang, and Shuai Wang. 2022. MetamorphicTestingofDeepLearningCompilers. Proc.ACMMeas.Anal.Comput. Syst.6 (2022), 15:1–15:28. Issue 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "XiaoyuanXie,JoshuaWingKeiHo,ChristianMurphy,GailE.Kaiser,BaowenXu, and Tsong Yueh Chen. 2009. Application of Metamorphic Testing to Supervised Classi/f_iers.In QSIC. IEEE Computer Society, 135–144.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "XuejunYang,Yang Chen,EricEide,andJohnRegehr.2011. FindingandUnderstanding Bugs in CCompilers. In PLDI. ACM,283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Peisen Yao, Heqing Huang, Wensheng Tang, Qingkai Shi, Rongxin Wu, and Charles Zhang. 2021. Fuzzing SMT Solvers via Two-Dimensional Input Space Exploration. In ISSTA. ACM,322–335.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Peisen Yao, Heqing Huang, Wensheng Tang, Qingkai Shi, Rongxin Wu, and Charles Zhang. 2021. Skeletal Approximation Enumeration for SMT Solver Testing.In ESEC/FSE. ACM,1141–1153.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "ChengyuZhang,TingSu,YichenYan,FuyuanZhang,GeguangPu,andZhendong Su. 2019. Finding and Understanding Bugs in Software Model Checkers. In ESEC/FSE. ACM,763–773.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "QirunZhang,ChengnianSun,andZhendongSu.2017. SkeletalProgramEnumerationfor Rigorous CompilerTesting.In PLDI. ACM,347–361.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Zhiquan Zhou, Shaowen Xiang, and Tsong Yueh Chen. 2016. Metamorphic Testing for Software Quality Assessment: A Study of Search Engines. TSE42 (2016), 264–284. Issue 3. Received 2023-02-16; accepted 2023-05-03 247",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 43,
+      "text": "ManuelRiggerandZhendongSu.2020. FindingBugsinDatabaseSystemsvia QueryPartitioning. PACMPL 4 (2020), 211:1–211:30. Issue OOPSLA.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 43,46,63]",
+      "technique": "tlp",
+      "sentence": "Metamorphictestinghasalsobeensuccessfully used to test a variety of software applications [ 12,27,47,57], includingotherquery-basedsystems[ 43,46,63].",
+      "context_before": "velystudiedinthegraphtheorycommunity.Next,wediscussthe three mostcloselyrelatedareas of work. Metamorphictesting. Metamorphictesting[ 15,47]isatechnique to effectively address the oracle problem in software testing. It works by constructing new test cases via mutating existing ones usingmetamorphicrelations.Theserelationsarethenusedtoinfer the output of the newly generated test cases. In previous work, we 245 Dependency-Aware MetamorphicTestingof DatalogEngines ISSTA ’23, July 17–21, 2023,Sea/t_tle, WA,USA presented the /f_irst metamorphic testing approach to detect bugs in Datalogengines[ 34].",
+      "context_after": "Testing compilers. Ensuring the correctness of compilers is acriticalareaofresearch,andsigni/f_icantefforthasbeendevoted to testing compilers [ 13,29,48,58]. Csmith [ 58] is a popular random program generator for C compilers that can generate large and complex programs. Le et al.[27] proposed a metamorphic testingapproachtogenerateequivalentprograms.Thetechnique inspired a number of related approaches [ 17,28,29,49,62] that wereusedtodetecthundredsofbugsinpopularandwidelyused compilers.YarpGen[ 32]isatechniqueforgeneratingexpressive programs without unde/f_ined behavior to test C and C++ compi",
+      "section": "7.3 Threatsto Validity",
+      "page": 11,
+      "char_offset": 55408,
+      "cited_reference": {
+        "number": 43,
+        "text": "ManuelRiggerandZhendongSu.2020. FindingBugsinDatabaseSystemsvia QueryPartitioning. PACMPL 4 (2020), 211:1–211:30. Issue OOPSLA.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:03:26Z",
+    "is_model_written": true,
+    "summary": "This work tests Datalog engines with dependency-aware metamorphic relations, taking account of how rules depend on one another so that a transformation's effect on the derived result stays predictable in the presence of recursion.",
+    "narrative": "The single mention places the work in context: metamorphic testing has been used successfully across a variety of software, including other query-based systems, which is where SQLancer's techniques are cited. The relation is one of shared method rather than of use, and the mention is reachable only through the citation marker.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3597926_3598068.json b/_data/papers/paper_doi_10_1145_3597926_3598068.json
new file mode 100644
index 0000000..4b73cf1
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3597926_3598068.json
@@ -0,0 +1,729 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:12:44Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3597926.3598068",
+    "title": "Exploring Missed Optimizations in WebAssembly Optimizers",
+    "authors": [
+      "Zhibo Liu",
+      "Dongwei Xiao",
+      "Zongjie Li",
+      "Shuai Wang",
+      "W. Meng"
+    ],
+    "year": 2023,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3597926.3598068",
+    "arxiv_id": null,
+    "s2_paper_id": "dc1cc47e14653eb9a834f44a39273c1a2e2de4ae",
+    "url": "https://doi.org/10.1145/3597926.3598068",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3597926.3598068",
+      "retrieved_at": "2026-09-07T17:12:44Z",
+      "chars": 85110,
+      "content_sha256": "sha256:1a636b6976b7364e9b1e92e0a26bc9a1d9e23d2d087e87354f1b2f971927058c"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2905
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND",
+        "start": 9202
+      },
+      {
+        "number": "3",
+        "title": "MOTIVATION",
+        "start": 15239
+      },
+      {
+        "number": "4",
+        "title": "APPROACH OVERVIEW",
+        "start": 19706
+      },
+      {
+        "number": "5",
+        "title": "TRACE CONSISTENCY CHECKS",
+        "start": 25535
+      },
+      {
+        "number": "7.1",
+        "title": "RQ1: Characteristics of Missed Opt.",
+        "start": 36513
+      },
+      {
+        "number": "7.2",
+        "title": "RQ2: Improvement Estimation",
+        "start": 53133
+      },
+      {
+        "number": "7.3",
+        "title": "RQ3: Lessons and Future Improvements",
+        "start": 59133
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2021. Binaryen vs LLVM as a compiler backend. https://github.com/ WebAssembly/binaryen/discussions/3886.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2022. AssemblyScript. https://github.com/AssemblyScript/assemblyscript.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2022. Awesome WebAssembly Languages. https://github.com/appcypher/ awesome-wasm-langs.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2022. Browsers support WebAssembly. https://caniuse.com/?search= WebAssembly.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2022. Ditwo Artifact. https://github.com/monkbai/wasm-testing.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2022. Ethereum WebAssembly. https://ewasm.readthedocs.io/en/mkdocs/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2022. ink! Documentation. https://use.ink/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2022. Terra Wasm Module. https://docs.terra.money/develop/modulespecifications/spec-wasm/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2022. TinyGo. https://github.com/tinygo-org/tinygo.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2022. wasm-bindgen. https://github.com/rustwasm/wasm-bindgen. 446 ISSTA ’23, July 17–21, 2023, Seattle, WA, USA Zhibo Liu, Dongwei Xiao, Zongjie Li, Shuai Wang, and Wei Meng",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2022. wasm3. https://github.com/wasm3/wasm3.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2022. WasmEdgeRuntime. https://github.com/WasmEdge/WasmEdge.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "2022. wasmer. https://github.com/wasmerio/wasmer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "2022. wasmi. https://github.com/paritytech/wasmi.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "2022. wasmtime. https://github.com/bytecodealliance/wasmtime.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Gogul Balakrishnan and Thomas Reps. 2010. Wysinwyx: What you see is not what you execute. ACM Transactions on Programming Languages and Systems (TOPLAS) 32, 6 (2010), 1–84.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Gogul Balakrishnan, Thomas Reps, David Melski, and Tim Teitelbaum. 2005. Wysinwyx: What you see is not what you execute. In Working Conference on Verified Software: Theories, Tools, and Experiments. Springer, 202–213.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Gergö Barany. 2018. Finding missed compiler optimizations by differential testing. InProceedings of the 27th international conference on compiler construction. 82–92.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "bytecodealliance. [n. d.]. Wasmtime Fuzzer. https://github.com/bytecodealliance/ wasmtime/tree/main/fuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Dhruva R Chakrabarti and Shin-Ming Liu. 2006. Inline analysis: Beyond selection heuristics. In International Symposium on Code Generation and Optimization (CGO’06). IEEE, 12–pp.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Yuting Chen, Ting Su, and Zhendong Su. 2019. Deep differential testing of JVM implementations. In Engineering (ICSE). IEEE, 1257–1268.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Yuting Chen, Ting Su, Chengnian Sun, Zhendong Su, and Jianjun Zhao. 2016. Coverage-directed Differential Testing of JVM Implementations. In PLDI.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Emscripten Contributors. 2022. Emscripten. https://github.com/emscriptencore/emscripten.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Grain core team. 2022. Grain Compiler. https://github.com/grain-lang/grain.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Clang Developers. 2022. Clang-Tidy-Extra Clang Tools 16.0.0git documentation. https://clang.llvm.org/extra/clang-tidy/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "LLVM Project Developers. 2022. llvm-dwarfdumpdump and verify DWARF debug information. https://llvm.org/docs/CommandGuide/llvm-dwarfdump. html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Sushant Dinesh, Nathan Burow, Dongyan Xu, and Mathias Payer. 2020. Retrowrite: Statically instrumenting cots binaries for fuzzing and sanitization. In. IEEE, 1497–1511.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Antonio Flores-Montoya and Eric Schulte. 2020. Datalog disassembly. In 29th USENIX Security Symposium (USENIX Security 20). 1075–1092.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "William Fu, Raymond Lin, and Daniel Inge. 2018. Taintassembly: Taint-based information flow control tracking for webassembly. arXiv preprint arXiv:1802.01050 (2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "David Goltzsche, Manuel Nieke, Thomas Knauth, and Rüdiger Kapitza. 2019. Acctee: A webassembly-based two-way sandbox for trusted resource accounting. InProceedings of the 20th International Middleware Conference. 123–135.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "WebAssembly Group. 2022. Binaryen. https://github.com/WebAssembly/ binaryen.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "WebAssembly Group. 2022. Binaryen Optimizations. https://github.com/ WebAssembly/binaryen#binaryen-optimizations.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "WebAssembly Group. 2022. Roadmap-WebAssembly. https://webassembly.org/ roadmap/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "WebAssembly Group. 2022. WABT: The WebAssembly Binary Toolkit. https: //github.com/WebAssembly/wabt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "WebAssembly Group. 2022. wasm-opt peephole optimizations. https://github. com/WebAssembly/binaryen/blob/main/src/passes/OptimizeInstructions.cpp.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "WebAssembly Community Group. 2022. Use Cases-WebAssembly. https: //webassembly.org/docs/use-cases/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Jianmin Guo, Yu Jiang, Yue Zhao, Quan Chen, and Jiaguang Sun. 2018. Dlfuzz: Differential fuzzing testing of deep learning systems. In Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 739–743.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Andreas Haas, Andreas Rossberg, Derek L Schuff, Ben L Titzer, Michael Holman, Dan Gohman, Luke Wagner, Alon Zakai, and JF Bastien. 2017. Bringing the web up to speed with WebAssembly. In Proceedings of the 38th ACMSIGPLAN Conference on Programming Language Design and Implementation. 185–200.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Yuko Hara, Hiroyuki Tomiyama, Shinya Honda, Hiroaki Takada, and Katsuya Ishii. 2008. Chstone: A benchmark program suite for practical c-based high-level synthesis. In. IEEE, 1192–1195.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Keno Haßler and Dominik Maier. 2021. WAFL: Binary-Only WebAssembly Fuzzing with Fast Snapshots. In Reversing and Offensive-oriented Trends Symposium. 23–30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Jingxuan He, Pesho Ivanov, Petar Tsankov, Veselin Raychev, and Martin Vechev. 2018. Debin: Predicting debug information in stripped binaries. In Proceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security. 1667–1680.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Aaron Hilbig, Daniel Lehmann, and Michael Pradel. 2021. An empirical study of real-world webassembly binaries: Security, languages, use cases. In Proceedings of the Web Conference 2021. 2696–2708.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Tweag I/O. 2022. Asterius: A Haskell to WebAssembly compiler. https://github. com/tweag/asterius.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Abhinav Jangda, Bobby Powers, Emery D Berger, and Arjun Guha. 2019. Not So Fast: Analyzing the Performance of WebAssembly vs. Native Code. In 2019 USENIX Annual Technical Conference (USENIX ATC 19). 107–120.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Timotej Kapus and Cristian Cadar. 2017. Automatic testing of symbolic execution engines via program generation and differential testing. In 2017 32nd IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 590–600.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Matthew Kolosick, Shravan Narayan, Conrad Watt, Michael LeMay, Deepak Garg, Ranjit Jhala, and Deian Stefan. 2021. Isolation Without Taxation: Near Zero Cost Transitions for SFI. arXiv preprint arXiv:2105.00033 (2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler Validation via Equivalence Modulo Inputs. In PLDI.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Vu Le, Chengnian Sun, and Zhendong Su. 2015. Finding Deep Compiler Bugs via Guided Stochastic Program Mutation. In OOPSLA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Daniel Lehmann, Johannes Kinder, and Michael Pradel. 2020. Everything Old is New Again: Binary Security of WebAssembly. In 29th USENIX Security Symposium (USENIX Security 20). 217–234.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Daniel Lehmann and Michael Pradel. 2019. Wasabi: A framework for dynamically analyzing webassembly. In Proceedings of the Twenty-Fourth International Conference on Architectural Support for Programming Languages and Operating Systems. 1045–1058.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Daniel Lehmann and Michael Pradel. 2022. Finding the dwarf: recovering precise types from WebAssembly binaries. In Proceedings of the 43rd ACMSIGPLAN International Conference on Programming Language Design and Implementation. 410–425.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Daniel Lehmann, Martin Toldam Torp, and Michael Pradel. 2021. Fuzzm: Finding Memory Bugs through Binary-Only Instrumentation and Fuzzing of WebAssembly.arXiv preprint arXiv:2110.15433 (2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Shaohua Li and Zhendong Su. 2023. Finding Unstable Code via Compiler-Driven Differential Testing. In Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 3. 238–251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Zhibo Liu and Shuai Wang. 2020. How far we have come: Testing decompilation correctness of C decompilers. In Proceedings of the 29th ACMSIGSOFT International Symposium on Software Testing and Analysis. 475–487.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Nuno P Lopes, David Menendez, Santosh Nagarakatte, and John Regehr. 2015. Provably correct peephole optimizations with alive. In Proceedings of the 36th ACMSIGPLAN Conference on Programming Language Design and Implementation. 22–32.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Chi-Keung Luk, Robert Cohn, Robert Muth, Harish Patil, Artur Klauser, Geoff Lowney, Steven Wallace, Vijay Janapa Reddi, and Kim Hazelwood. 2005. Pin: Building Customized Program Analysis Tools with Dynamic Instrumentation. InProceedings of the 2005 ACMSIGPLAN Conference on Programming Language Design and Implementation (Chicago, IL, USA) (PLDI ’05). ACM, 190–200.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Jämes Ménétrey, Marcelo Pasin, Pascal Felber, and Valerio Schiavoni. 2021. Twine: An embedded trusted runtime for webassembly. In 2021 IEEE 37th International Conference on Data Engineering (ICDE). IEEE, 205–216.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Eric Mullen, Daryl Zuniga, Zachary Tatlock, and Dan Grossman. 2016. Verified peephole optimizations for CompCert. In Proceedings of the 37th ACMSIGPLAN Conference on Programming Language Design and Implementation. 448–461.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Shravan Narayan, Craig Disselkoen, Tal Garfinkel, Nathan Froyd, Eric Rahm, Sorin Lerner, Hovav Shacham, and Deian Stefan. 2020. Retrofitting fine grain isolation in the Firefox renderer. In 29th USENIX Security Symposium (USENIX Security 20). 699–716.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Shravan Narayan, Craig Disselkoen, Daniel Moghimi, Sunjay Cauligi, Evan Johnson, Zhao Gang, Anjo Vahldiek-Oberwagner, Ravi Sahita, Hovav Shacham, Dean Tullsen, et al .2021. Swivel: Hardening WebAssembly against Spectre. In 30th USENIX Security Symposium (USENIX Security 21). 1433–1450.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Shravan Narayan, Tal Garfinkel, Sorin Lerner, Hovav Shacham, and Deian Stefan. 2019. Gobi: WebAssembly as a practical path to library sandboxing. arXiv preprint arXiv:1912.02285 (2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Senthil Padmanabhan and Pranav Jha. 2022. WebAssembly at eBay: A Real-World Use Case. https://tech.ebayinc.com/engineering/webassembly-at-ebay-a-realworld-use-case/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Roberto Paleari, Lorenzo Martignoni, Giampaolo Fresi Roglia, and Danilo Bruschi. 2010. N-version disassembly: differential testing of x86 disassemblers. In Proceedings of the 19th international symposium on Software testing and analysis. 265–274.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "G Paolini. 1994. Netscape and Sun announce JavaScript, the open cross-platform object scripting language for enterprise networks and the internet. Press Release]. Sun Microsystems, Inc (1994).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "John Regehr, Yang Chen, Pascal Cuoq, Eric Eide, Chucky Ellison, and Xuejun Yang. 2012. Test-case reduction for C compiler bugs. In Proceedings of the 33rd ACMSIGPLAN conference on Programming Language Design and Implementation. 335–346.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152. 447 Exploring Missed Optimizations in WebAssembly Optimizers ISSTA ’23, July 17–21, 2023, Seattle, WA, USA",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 67,
+      "text": "Alan Romano, Xinyue Liu, Yonghwi Kwon, and Weihang Wang. 2021. An Empirical Study of Bugs in WebAssembly Compilers. In 2021 36th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 42–54.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Alan Romano and Weihang Wang. 2020. Wasim: Understanding webassembly applications through classification. In 2020 35th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 1321–1325.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "The Rust and WebAssembly Working Group. 2022. Shrinking .wasm Code Size Rust and WebAssembly. https://rustwasm.github.io/docs/book/reference/codesize.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Andreas Sewe, Jannik Jochem, and Mira Mezini. 2011. Next in line, please! exploiting the indirect benefits of inlining by accurately predicting further inlining. InProceedings of the compilation of the co-located workshops on DSM’11, TMC’11, AGERE! 2011, AOOPES’11, NEAT’11, & VMIL’11. 317–328.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Simon Shillaker and Peter Pietzuch. 2020. Faasm: Lightweight isolation for efficient stateful serverless computing. In 2020 USENIX Annual Technical Conference (USENIX ATC 20). 419–433.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Donald R Slutz. 1998. Massive stochastic testing of SQL. In VLDB, Vol. 98. Citeseer, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Daniel Smilkov, Nikhil Thorat, and Ann Yuan. 2022. Introducing the WebAssembly backend for TensorFlow.js. https://blog.tensorflow.org/2020/03/introducingwebassembly-backend-for-tensorflow-js.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Quentin Stiévenart, David W Binkley, and Coen De Roover. 2022. Static StackPreserving Intra-Procedural Slicing of WebAssembly Binaries. In 2022 IEEE/ACM 44th International Conference on Software Engineering (ICSE). IEEE, 2031–2042.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Yulei Sui and Jingling Xue. 2016. SVF: interprocedural static value-flow analysis in LLVM. In Proceedings of the 25th international conference on compiler construction. 265–266.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. 2016. Finding Compiler Bugs via Live Code Mutation. In OOPSLA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Aron Szanto, Timothy Tamm, and Artidoro Pagnoni. 2018. Taint tracking for webassembly. arXiv preprint arXiv:1807.08349 (2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Leaning Technologies. 2022. An Enterprise-Grade C++ Compiler For The Web. https://leaningtech.com/cheerp/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Theodoros Theodoridis, Tobias Grosser, and Zhendong Su. 2022. Understanding and exploiting optimal function inlining. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems. 977–989.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems. 697–709.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 81,
+      "text": "Ruoyu Wang, Yan Shoshitaishvili, Antonio Bianchi, Aravind Machiry, John Grosen, Paul Grosen, Christopher Kruegel, and Giovanni Vigna. 2017. Ramblr: Making Reassembly Great Again. In NDSS.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Shuai Wang, Pei Wang, and Dinghao Wu. 2015. Reassembleable Disassembling. InUSENIX Sec.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Wenwen Wang. 2022. How Far We’ve Come–A Characterization Study of Standalone WebAssembly Runtimes. (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "wasmCloud Project Authors. 2022. WasmCloud. https://wasmcloud.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "wasmerio. [n. d.]. Wasmer Fuzz Testing. https://github.com/wasmerio/wasmer/ tree/master/fuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Elliott Wen and Gerald Weber. 2020. Wasmachine: Bring iot up to speed with a webassembly os. In and Communications Workshops (PerCom Workshops). IEEE, 1–4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "Yutian Yan, Tengfei Tu, Lijian Zhao, Yuchen Zhou, and Weihang Wang. 2021. Understanding the performance of webassembly applications. In Proceedings of the 21st ACM Internet Measurement Conference. 533–549.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and Understanding Bugs in C Compilers. In PLDI.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "Peng Zhao and José Nelson Amaral. 2003. To inline or not to inline? Enhanced inlining decisions. In International Workshop on Languages and Compilers for Parallel Computing. Springer, 405–419. Received 2023-02-16; accepted 2023-05-03 448",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 66,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152. 447 Exploring Missed Optimizations in WebAssembly Optimizers ISSTA ’23, July 17–21, 2023, Seattle, WA, USA",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 80,
+      "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems. 697–709.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 80]",
+      "technique": null,
+      "sentence": "Inspired by contemporary research on testing C compiler optimizations [ 80], we instead explore a differential testing setting, by treating native x86 executables fully optimized by modern C compilers as the “reference” to unveil MO.",
+      "context_before": "d optimizations (MO) of wasm optimizers. While this may be partially accomplished by reading the wasm optimizer documents and code, in practice the feasibility is limited by the complexity of the wasm optimizer as well as the nature of program optimizations: optimization opportunities may be subtle and certain optimizations are deemed “missed” only when processing specific code emitted by compiler frontends. In principle, deciding MO of wasm optimizers would require a “ground truth” (e.g., manually crafting some fully-optimized wasm executables) to compare with, which is challenging to obtain.",
+      "context_after": "This enables an automated, systematic, and scalable testing of wasm optimizers. Overall, we present Ditwo, adifferential testing framework for wasmoptimizers. Ditwo differentiates runtime behaviors of wasm binary code and its native x86 counterparts compiled from the same C code to uncover MO. The key technical challenge is to select proper “performance indicators” from the wasm runtime logs that are practically feasible to compare and uncover the neglect of various wasm optimization opportunities. To do so, Ditwo launches both wasm and native executables to log two indicators: global variabl",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 5692,
+      "cited_reference": {
+        "number": 80,
+        "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems. 697–709.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 79,80]",
+      "technique": null,
+      "sentence": "While recent research has conducted empirical studies to evaluate MO in GCC/Clang [ 79,80], they mainly focus on dead code elimination and code size optimization when compiling C code to x86 executables.",
+      "context_before": "Moreover, during the WW phase, both wasm -specific and classic compile-time optimizations can be fulfilled, whereas during the SW phase, the optimizations are mostly target-agnostic, potentially losing optimization opportunities and being less comprehensive than the later phase. In practice, however, it is hard to implement comprehensive optimization passes at the WW phase. As an assembly-like language, wasm lacks high-level abstraction compared with compiler IR. It becomes technically harder for the optimizer to collect context information from wasm code, as illustrated in R1–3; see Sec. 7.1.",
+      "context_after": "To systematically assess the status quo of WW optimizations, we detect missed optimizations in live code blocks that are likely executed in runtime. From this point, the issues we found should be more severe, impeding program performance if not properly handled. We aim to provide an up-to-date assessment of present WW optimizations and identify their MO. We will investigate and categorize root causes of the detected MO cases, and harvest lessons to serve as guidelines for better WW optimizations. 4 APPROACH OVERVIEW This section describes the challenges in detecting missed optimization opport",
+      "section": "3 MOTIVATION",
+      "page": 4,
+      "char_offset": 18999,
+      "cited_reference": {
+        "number": 80,
+        "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems. 697–709.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 79,80]",
+      "technique": null,
+      "sentence": "4 and argued by relevant works [ 79,80], GW and FC heavily depend on the whole optimization pipeline.",
+      "context_before": "encies, to detect MO. They are: Global Variable Write Consistency (GW) checks for redundant assignments (memory writes) to global variables. Function Call Consistency (FC) checks for function calls that the WW optimizer fails to inline. Clarification. Note that we do not aim for false negative-free testing, i.e., our trace consistency checks are not complete, and even if both GW and FC are satisfied, potential MO can still exist in wasm-opt -optimized binary. Ditwo uses these two checks, given their generally promising comprehensiveness and high implementation feasibility; as clarified in Sec.",
+      "context_after": "Thus, inconsistencies reflect missed GW/FC optimization opportunities, or other MO that have dependency with GW/FC. Overall, we believe through the lens of GW and FC, we shed light on optimizations to quantify how well wasm-opt optimizes wasm binary. Also, other consistency checks could be defined to capture finer-grained program runtime states, and we present relevant discussions in Sec. 8. We now elaborate on checking each indicator below. 5.1 Global Variable Write Consistency (GW) We assume that if a global variable write is optimized out by the mainstream C compiler, then it should be equ",
+      "section": "5 TRACE CONSISTENCY CHECKS",
+      "page": 5,
+      "char_offset": 26255,
+      "cited_reference": {
+        "number": 80,
+        "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems. 697–709.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[ 66,\n72]",
+      "technique": "norec",
+      "sentence": "Differential testing (DT) is used in different software domains, including databases [ 66, 72], Java Virtual Machines (JVMs) [ 21,22], symbolic execution engines [ 45], disassemblers [ 63], decompilers [ 54], and deep learning systems [37].",
+      "context_before": "ave noticed recent works launching empirical studies to characterize wasm compiler bugs and performance defects [ 67,87]. A recent research characterizes standalone wasm runtimes and advocates for effective and low-cost runtime wasm optimizations [ 83]. In addition to empirical studies, we have also noticed some community efforts launching fuzz testing toward wasm VMs [ 19,85].Ditwo aims to expose and explore more stealthy MO that can induce performance defects. To the best of our knowledge, this has not been comprehensively studied by previous works. Differential Testing for Systems Software.",
+      "context_after": "Beyond finding functionality bugs, recent work also explores locating missed optimization opportunities with DT to improve C compiler infrastructures [ 18,80].CompDiff [53] detects undefined behavior in C/C++ programs with compiler-driven DT. 10 CONCLUSION This study systematically investigated the hidden MO of wasm optimizers with Ditwo, a differential testing framework to crosscompare wasm executables and their x86 counterparts. Our study exposes a large number of MO. We analyze root causes of all uncovered MO, and outline the key takeaways from this study. This work may serve as a roadmap",
+      "section": "7.3 RQ3: Lessons and Future Improvements",
+      "page": 11,
+      "char_offset": 69061,
+      "cited_reference": {
+        "number": 66,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152. 447 Exp",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 18,80]",
+      "technique": null,
+      "sentence": "Beyond finding functionality bugs, recent work also explores locating missed optimization opportunities with DT to improve C compiler infrastructures [ 18,80].",
+      "context_before": "tions [ 83]. In addition to empirical studies, we have also noticed some community efforts launching fuzz testing toward wasm VMs [ 19,85].Ditwo aims to expose and explore more stealthy MO that can induce performance defects. To the best of our knowledge, this has not been comprehensively studied by previous works. Differential Testing for Systems Software. Differential testing (DT) is used in different software domains, including databases [ 66, 72], Java Virtual Machines (JVMs) [ 21,22], symbolic execution engines [ 45], disassemblers [ 63], decompilers [ 54], and deep learning systems [37].",
+      "context_after": "CompDiff [53] detects undefined behavior in C/C++ programs with compiler-driven DT. 10 CONCLUSION This study systematically investigated the hidden MO of wasm optimizers with Ditwo, a differential testing framework to crosscompare wasm executables and their x86 counterparts. Our study exposes a large number of MO. We analyze root causes of all uncovered MO, and outline the key takeaways from this study. This work may serve as a roadmap for researchers and users to improve wasm program performance with optimizers. ACKNOWLEDGMENTS We thank anonymous reviewers for their valuable feedback. The res",
+      "section": "7.3 RQ3: Lessons and Future Improvements",
+      "page": 11,
+      "char_offset": 69302,
+      "cited_reference": {
+        "number": 80,
+        "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems. 697–709.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:20:36Z",
+    "is_model_written": true,
+    "summary": "DITWO looks for optimisations that WebAssembly optimizers miss, motivated by findings that real wasm applications run slower than expected. It compiles a C program to both a native x86 executable and a wasm executable, then compares optimization indication traces -- global variable writes and function calls -- and treats divergence as a missed optimisation. Analysing wasm-opt it identified 1,293 inputs triggering missed optimisations and nine root causes.",
+    "narrative": "The citations that shape the paper are to work by a SQLancer author on C compiler optimisation testing, which DITWO says inspired its differential setting. SQLancer itself appears once, listing databases among the domains where differential testing is used.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3597926_3598130.json b/_data/papers/paper_doi_10_1145_3597926_3598130.json
new file mode 100644
index 0000000..85540e3
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3597926_3598130.json
@@ -0,0 +1,755 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:26:56Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3597926.3598130",
+    "title": "GrayC: Greybox Fuzzing of Compilers and Analysers for C",
+    "authors": [
+      "Karine Even-Mendoza",
+      "Arindam Sharma",
+      "Alastair F. Donaldson",
+      "Cristian Cadar"
+    ],
+    "year": 2023,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3597926.3598130",
+    "arxiv_id": null,
+    "s2_paper_id": "86a60acfe93752af1159dea65d01808134754467",
+    "url": "https://doi.org/10.1145/3597926.3598130",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3597926.3598130",
+      "retrieved_at": "2026-09-08T15:26:56Z",
+      "chars": 77498,
+      "content_sha256": "sha256:a8254799f9ab108b95ac3dad72360144384681c4dd21b51dcf4837c6f8d9f355"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 3191
+      },
+      {
+        "number": "2.1",
+        "title": "Compiler Bugs andProgramValidity",
+        "start": 13911
+      },
+      {
+        "number": "3.1",
+        "title": "CustomMutators",
+        "start": 18571
+      },
+      {
+        "number": "3.3",
+        "title": "Implementation Details",
+        "start": 29861
+      },
+      {
+        "number": "4.1",
+        "title": "ExperimentalSetup",
+        "start": 31537
+      },
+      {
+        "number": "4.2",
+        "title": "Bugs Found",
+        "start": 33893
+      },
+      {
+        "number": "4.3",
+        "title": "Compiler TestCaseContributions",
+        "start": 35747
+      },
+      {
+        "number": "5.1",
+        "title": "ExperimentalSetup",
+        "start": 36510
+      },
+      {
+        "number": "5.2",
+        "title": "Throughput andStatic ValidityRate",
+        "start": 41848
+      },
+      {
+        "number": "5.3",
+        "title": "Coverage",
+        "start": 44609
+      },
+      {
+        "number": "5.4",
+        "title": "Bug Finding",
+        "start": 48704
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "CorneliusAschermann,TommasoFrassetto,ThorstenHolz,PatrickJauernig, Ahmad-Reza Sadeghi, and Daniel Teuchert. 2019. NAUTILUS: Fishing for Deep BugswithGrammars.In Proc.ofthe26thNetworkandDistributedSystemSecurity Symposium (NDSS’19) (San Diego, CA, USA). https://doi.org/10.14722/ndss. 2019.23412",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "C-Reduce Bugclang delta (found as a by-product of fuzzing). Date Confirmed and Fixed Jan. 16, 2022. https://www.flux.utah.edu/listarchives/creducebugs/msg00555.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "C-Reduce Bugclang delta (found as a by-product of fuzzing). Date Confirmed and Fixed Jan. 4, 2022. https://www.flux.utah.edu/listarchives/creducebugs/msg00553.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "C-Reduce Bugclang delta (found as a by-product of fuzzing). Date Confirmed Jun. 7, 2021 and Fixed Jun. 20, 2021. https://www.flux.utah.edu/listarchives/ creduce-bugs/msg00537.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "C-Reduce Bugclang delta (found as a by-product of fuzzing). Date Confirmed November 2, 2022. https://www.flux.utah.edu/listarchives/creduce-bugs/ msg00563.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "C-ReduceBug-clangdelta(foundasaby-productoffuzzing).DateReportedDec. 17,2021. https://www.flux.utah.edu/listarchives/creduce-bugs/msg00551.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Cristian Cadar and Alastair Donaldson. 2016. Analysing the Program Analyser. InProc. of the 38th International Conference on SoftwareEngineering, New Ideas and EmergingResults (ICSENIER’16) (Austin,TX, USA).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "JunjieChen,JibeshPatra,MichaelPradel,YingfeiXiong,HongyuZhang,Dan Hao, and Lu Zhang. 2020. A Survey of Compiler Testing. Comput. Surveys 53, 1 (2020), 4:1–4:36. https://doi.org/10.1145/3363562",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "T.Y.Chen,S.C.Cheung,andS.M.Yiu.1998. Metamorphictesting:anewapproach for generating next test cases. Technical Report HKUST-CS98-01. Hong Kong Universityof Science and Technology.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "YutingChen,TingSu,ChengnianSun,ZhendongSu,andJianjunZhao.2016. Coverage-directeddifferentialtestingofJVMimplementations.In Proc.ofthe 1229 ISSTA ’23, July 17–21, 2023,Sea/t_tle, WA,USA KarineEven-Mendoza,Arindam Sharma,AlastairF. Donaldson, andCristian Cadar ConferenceonProgramingLanguageDesignandImplementation(PLDI’16) (Santa Barbara, CA, USA). https://doi.org/10.1145/2908080.2908095",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Yongheng Chen, Rui Zhong, Hong Hu, Hangfan Zhang, Yupeng Yang, Dinghao Wu, and Wenke Lee. 2022. One Engine to Fuzz ’em All: Generic Language Processor Testing with Semantic Validation. In Proc. of the IEEE Symposium on SecurityandPrivacy(IEEES&P’22) (SanFrancisco,CA,USA). https://doi.org/ 10.1109/SP40001.2021.00071",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Clang LibTooling 2023. LibTooling. https://clang.llvm.org/docs/LibTooling. html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "clangfuzzer [n.d.]. clang-fuzzer. https://github.com/llvm/llvm-project/tree/ main/clang/tools/clang-fuzzer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "clangprotofuzzer [n.d.]. clang-proto-fuzzer. https://llvm.org/docs/ FuzzingLLVM.html#clang-proto-fuzzer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "PascalCuoq,FlorentKirchner,NikolaiKosmatov,VirgilePrevosto,JulienSignoles,and BorisYakobowski.2012. Frama-C: Asoftwareanalysisperspective. InProc. of the 10th International Conference on Software Engineering and Formal Methods(SEFM’12) (Thessaloniki, Greece). https://doi.org/10.1007/s00165-0140326-7",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "PascalCuoq,BenjaminMonate,AnnePacalet,VirgilePrevosto,JohnRegehr, Boris Yakobowski, and Xuejun Yang. 2012. Testing StaticAnalyzers with RandomlyGeneratedPrograms.In Proc.ofthe4thInternationalConferenceonNASA Formal Methods(NFM’12) (Norfolk, VA, USA).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Alastair F. Donaldson, Hugues Evrard, Andrei Lascu, and Paul Thomson. 2017. Automated Testing of Graphics Shader Compilers. Proceedings of the ACM on Programming Languages (PACMPL) 1, OOPSLA (2017), 93:1–93:29. https: //doi.org/10.1145/3133917",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "KarineEven-Mendoza,ArindamSharma,AlastairF.Donaldson,andCristian Cadar.2023. ArtifactofGrayC:GreyboxFuzzingofCompilersandAnalysers for C.https://doi.org/10.5281/zenodo.7978251. Zenodo.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Frama-C Bug-Eva plugin. Date Confirmed Mar. 13, 2022 and Closed and fixed Jul. 11,2022. https://git.frama-c.com/pub/frama-c/-/issues/2595.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Frama-C Bug-Eva plugin. Date Confirmed May 10, 2022 and Fixed Jun. 10, 2022.https://git.frama-c.com/pub/frama-c/-/issues/2610.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Frama-CBug-Evaplugin.DateConfirmedNov.8,2021andFixedSept.15,2022. https://git.frama-c.com/pub/frama-c/-/issues/2585.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Frama-C Bug-Eva plugin, kernel, abstract interpretation. Date Confirmed and Fixed Jun. 10,2021. https://git.frama-c.com/pub/frama-c/-/issues/2563.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Frama-C Bug-Front-end. Date Confirmed Oct. 14, 2021 and Date Fixed Dec. 3, 2021.https://git.frama-c.com/pub/frama-c/-/issues/2576.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Frama-CBug-Front-end(foundasaby-productoffuzzing).DateConfirmed May 28, 2022 and Fixed Oct. 20, 2021. https://git.frama-c.com/pub/frama-c//issues/2559.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Frama-CBug-Front-end(foundasaby-productoffuzzing).DateConfirmed Sept. 14, 2021 and Fixed Jul. 11, 2022. https://git.frama-c.com/pub/frama-c//issues/2573.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Frama-CBug-Front-end(foundasaby-productoffuzzing).DateConfirmed Sept.16,2021andFixedOct.20,2021. https://git.frama-c.com/pub/frama-c//issues/2574.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Frama-CBug-Kernel.DateConfirmedApr.20,2021andFixedApr.30,2021. https://git.frama-c.com/pub/frama-c/-/issues/2551.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Frama-C Bug-Kernel. Date Confirmed Apr. 6, 2021 and Fixed Oct. 13, 2021. https://git.frama-c.com/pub/frama-c/-/issues/2550.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Frama-C Bugkernel. Date Confirmed Jan. 11, 2022 and Fixed Jul. 11, 2022. https://git.frama-c.com/pub/frama-c/-/issues/2592.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Frama-CBug-kernel,abstractinterpretation.DateConfirmedandFixedJan. 24,2022. https://git.frama-c.com/pub/frama-c/-/issues/2588.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Frama-C Bugkernel, abstract interpretation. Date Confirmed May 18, 2021 and Fixed May 21,2021. https://git.frama-c.com/pub/frama-c/-/issues/2556.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Frama-CBug-kernel,Front-end.DateConfirmedJan.10,2022andFixedFeb. 9,2022.https://git.frama-c.com/pub/frama-c/-/issues/2590.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Frama-CBug-Parsing,EVA-plugin.DateConfirmedMay18,2021andFixed May 21,2021. https://git.frama-c.com/pub/frama-c/-/issues/2555.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "GCC Bug. Date Reported Aug. 6, 2016. https://gcc.gnu.org/bugzilla/show_bug. cgi?id=72825.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "GCC Bug-Front-end. Date Confirmed Apr. 9, 2021 and Fixed Apr. 22, 2021. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "GCC Bug-Front-end. Date Confirmed Aug. 8, 2022 and Fixed Nov. 21, 2022. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106560.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "GCC Bugipa. Date Confirmed Dec. 23, 2021 and Fixed Apr. 20, 2022. https: //gcc.gnu.org/bugzilla/show_bug.cgi?id=103818.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "GCCBug-Middle-end.DateConfirmedDec.22,2021andFixedJan.24,2022. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103813.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "GCCBug-Middle-end.DateConfirmedMay02,2022andFixedMay27,2022. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104402.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "GCC Bug-Middle-end (reported independently before we found it). Date Confirmed Mar. 20, 2018 and Fixed Apr. 14, 2022. https://gcc.gnu.org/bugzilla/ show_bug.cgi?id=84964.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "GCC Bug-Middle-end (reported independently shortly before we found it). Date Confirmed Nov. 18, 2022 and Fixed Nov. 19, 2022. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103314.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "GCCBug-rtl-optimization,middle-end.DateReportedJun.9,2022andConfirmed May 17,2023. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105910.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "GCCBug-Treeoptimization.DateConfirmedandFixedApr.12,2022. https: //gcc.gnu.org/bugzilla/show_bug.cgi?id=105232.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "GCCBug-Treeoptimization.DateConfirmedandFixedJun.10,2022. https: //gcc.gnu.org/bugzilla/show_bug.cgi?id=107170.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "GCCBug-Treeoptimization.DateConfirmedDec.23,2021andFixedJan.5, 2022.https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103816.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "GCCBug-Tree-optimization(reportedindependentlybeforewefoundit).Date ConfirmedJul.27,2021andFixedMar.23,2022. https://gcc.gnu.org/bugzilla/ show_bug.cgi?id=101636.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "GCC Bug: incomplete program (several duplicate reports exist). Date Reported Aug.28,2022. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106764.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "GCC Bug (several related reports exist). Date Reported May 11, 2021, and Fixed 15November 2022. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100525.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "GitHub.2023. GitRepositoryofGrayC. https://github.com/srg-imperial/GrayC.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "GitHub.DateAccessedDecember31,2022. GitRepositoryofGrammarinator. https://github.com/renatahodovan/grammarinator.git.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "GitHub.DateAccessedMarch23,2022. GitRepositoryofgfauto. https://github. com/google/graphicsfuzz.git.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Google.2020. AFLdictionaries. https://github.com/google/AFL/blob/master/ dictionaries/README.dictionaries.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Alex Groce, Josie Holmes, Darko Marinov, August Shi, and Lingming Zhang. 2018. AnExtensible,Regular-Expression-BasedToolforMulti-languageMutant Generation.In Proc.ofthe40thInternationalConferenceonSoftwareEngineering (ICSE’18) (Gothenburg,Sweden). https://doi.org/10.1145/3183440.3183485",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "AlexGroce,GerardJ.Holzmann,andRajeevJoshi.2007.RandomizedDifferential Testing as a Prelude to Formal Verification. In Proc. of the 29th International ConferenceonSoftwareEngineering(ICSE’07) (Minneapolis,MN,USA). https: //doi.org/10.1109/ICSE.2007.68",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Alex Groce, Rijnard van Tonder, Goutamkumar Tulajappa Kalburgi, and Claire Le Goues. 2022. Making No-Fuss Compiler Fuzzing Effective. In Proc. of the31stInternationalConferenceonCompilerConstruction(CC’22) (Seoul,Korea). https://doi.org/10.1145/3497776.3517765",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "VladimirHerdt,DanielGroße,HoangM.Le,andRolfDrechsler.2019. Verifying Instruction Set Simulators using Coverage-guided Fuzzing*. InProc. of the 22nd Design,Automation&TestinEuropeConference&Exhibition(DATE’19) (Florence, Italy).IEEE,360–365. https://doi.org/10.23919/DATE.2019.8714912",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Yann Herklotz and John Wickerson. 2020. Finding and Understanding Bugs in FPGA Synthesis Tools. In Proc. of the 28th InternationalSymposium on FieldProgrammableGateArrays(FPGA’20) .ACM/SIGDA,277–287. https://doi.org/ 10.1145/3373087.3375310",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Renáta Hodován, Ákos Kiss, and Tibor Gyimóthy. 2018. Grammarinator: A Grammar-Based Open Source Fuzzer. In Proc. of the 9th ACMSIGSOFT International Workshop on Automating TEST Case Design, Selection, and Evaluation (ATEST’18) (Lake Buena Vista, FL, USA). https://doi.org/10.1145/3278186.3278193",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Christian Holler, Kim Herzig, and Andreas Zeller. 2012. Fuzzing with Code Fragments.In Proc.ofthe21stUSENIXSecuritySymposium(USENIXSecurity’12) (Bellevue, WA, USA).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "GrayCHomepage. Date Accessed May 23, 2022. https://srg.doc.ic.ac.uk/ projects/grayc/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Timotej Kapus and Cristian Cadar. 2017. Automatic Testing of Symbolic Execution Engines via Program Generation and Differential Testing. In Proc. of the 32nd IEEE International Conference on Automated Software Engineering (ASE’17) (Urbana-Champaign,IL, USA).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Fitsum Meshesha Kifetew, Roberto Tiella, and Paolo Tonella. 2017. Generating Valid Grammar-Based Test Inputs by Means of Genetic Programming and Annotated Grammars. Empirical Softw. Engg. 22, 2 (apr 2017), 928–961. https://doi.org/10.1007/s10664-015-9422-4",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Christian Klinger, Maria Christakis, and Valentin Wüstholz. 2019. Differentially Testing Soundness and Precision of Program Analyzers. In Proc. of the InternationalSymposiumonSoftwareTestingandAnalysis(ISSTA’19) (Beijing, China).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "ChrisLattnerandVikramAdve.2004. LLVM:ACompilationFrameworkfor LifelongProgramAnalysis&Transformation.In Proc.ofthe2ndInternational SymposiumonCodeGenerationandOptimization(CGO’04) (PaloAlto,CA,USA). https://doi.org/0.1109/CGO.2004.1281665",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler Validation via Equivalence Modulo Inputs. In Proc. of the Conference on Programing Language Design and Implementation (PLDI’14) (Edinburgh, UK). https://doi.org/10.1145/ 2594291.2594334",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "LibFuzzer 2022. http://llvm.org/docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Christopher Lidbury,Andrei Lascu,NathanChong,andAlastairF. Donaldson. 2015. Many-core compiler fuzzing. In Proc. of the Conference on Programing Language Design and Implementation (PLDI’15) (Portland, OR, USA). https: //doi.org/10.1145/2737924.2737986 1230 GrayC:GreyboxFuzzing of CompilersandAnalysers forC ISSTA ’23, July 17–21, 2023,Sea/t_tle, WA,USA",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Vsevolod Livinskii, Dmitry Babokin, and John Regehr. 2020. Random testing forCandC++compilerswith YARPGen.In Proc.ofthe ACMonProgramming Languages (OOPSLA’20) (Chicago, IL, USA). https://doi.org/10.1145/3428264",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "LLVMBug-Arrays.DateReportedJun.9,2021andClosedJan.7,2022. https: //github.com/llvm/llvm-project/issues/49983.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "LLVMBug-Clangcodegen(foundasaby-productoffuzzing).DateConfirmed Jan. 15,2022. https://github.com/llvm/llvm-project/issues/53105.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "LLVM Bug-Clang Front-end. Date Reported May 6, 2022. https://github.com/ llvm/llvm-project/issues/55312.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "LLVMBug-compiler-rt:ubsan(foundasaby-productoffuzzing).DateConfirmed Jan. 16,2022. https://github.com/llvm/llvm-project/issues/51421.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "LLVMBug-IR(foundasaby-productoffuzzing).DateReportedJul.5,2021. https://github.com/llvm/llvm-project/issues/50332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "LLVM Bug-Union declaration. Date Reported Jun. 10, 2021 and Closed Jan. 13, 2022.https://github.com/llvm/llvm-project/issues/49993.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "LLVM Project.DateAccessed July 21, 2022. libFuzzer – a libraryfor coverageguided fuzz testing. https://llvm.org/docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "LLVM website [n.d.]. LLVM website. http://llvm.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Michaël Marcozzi, Qiyi Tang, Alastair Donaldson, and Cristian Cadar. 2019. Compiler Fuzzing: How Much Does It Matter?. In Proc. of the ACM on ProgrammingLanguages(OOPSLA’19) (Athens,Greece). https://doi.org/10.1145/3360581",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "W. M. McKeeman. 1998. Differential testing for software. Digital Technical Journal10(1998), 100–107. Issue 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "MSVC Bug-CppCompiler, Front-end. Date Confirmed May 20, 2021. https: //developercommunity.visualstudio.com/t/internal-compiler-error-whencompiling-program-wit/1427557.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "MSVCBug-CppCompiler,Front-end.DateConfirmedMay20,2021andClosed Nov. 24, 2021. https://developercommunity.visualstudio.com/t/internalcompiler-error-when-compiling-program-wit/1427553.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "MSVC Bug-CppCompiler, Front-end. Date Confirmed May 20, 2021 and Fixed Nov. 9, 2021. https://developercommunity.visualstudio.com/t/syntacticallyinvalid-c-program-causes-microsoft-c/1427550.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Kazuhiro Nakamura andNagisaIshiura.2016. Random testing ofCcompilers based on test program generation by equivalence transformation. In 2016 IEEE AsiaPacificConferenceonCircuitsandSystems(APCCAS) .https://doi.org/10. 1109/APCCAS.2016.7804063",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Phabricator-LLVM.2020. RequestsD88931andD97686. https://reviews.llvm. org/D88931. Date Approved March3,2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Phabricator-LLVM. 2022. Request D118234. https://reviews.llvm.org/D118234. Date Approved October 11,2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "Phabricator-LLVM. 2023. Requests D142638 and D150857. https://reviews.llvm. org/D150857. Underreview: dateJanuary26,2023 (re-open:May 18,2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "John Regehr. 2020. The Saturation Effect in Fuzzing. https://blog.regehr.org/ archives/1796.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "JohnRegehr,YangChen,PascalCuoq,EricEide,ChuckyEllison,andXuejun Yang. 2012. Test-casereductionforCcompiler bugs.In Proc.oftheConference on Programing Language Design and Implementation (PLDI’12) (Beijing, China). https://doi.org/10.1145/2254064.2254104",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "LLVM Bug-Front-end (reported independently before we found it). Date Confirmed Jan. 26,2022. https://github.com/llvm/llvm-project/issues/49081.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "LLVM Bug-ASan (reported independently before we found it). Date Reported Feb. 20,2021. https://github.com/llvm/llvm-project/issues/48633.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "LLVMBug-Front-end(reportedindependentlybeforewefoundit).DateReported Jun. 26,2021. https://github.com/llvm/llvm-project/issues/50222.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "LLVMBug-Front-end(reportedindependentlybeforewefoundit).DateReported Nov. 12, 2015 and Fixed on early 2022. https://github.com/llvm/llvm-project/issues/25871.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "ManuelRiggerandZhendongSu.2020.DetectingOptimizationBugsinDatabase Engines via Non-optimizing Reference Engine Construction. In Proc. of the Joint Meeting of the European Software Engineering Conference and the ACM SymposiumontheFoundationsofSoftwareEngineering(ESEC/FSE’20) (Online). https://doi.org/10.1145/3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 93,
+      "text": "Sergio Segura, Gordon Fraser, Ana Sanchez, and Antonio Ruiz-Cortés. 2016. A SurveyonMetamorphicTesting. (2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "KostyaSerebryany. 2022. Personal communication.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitry Vyukov.2012. AddressSanitizer:AFastAddressSanity Checker.In Proc.ofthe 2012 USENIX Annual Technical Conference (USENIX ATC’12) (Boston, MA, USA). https://doi.org/10.5555/2342821.2342849",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "KostyaSerebryany,VitalyBuka,andMattMorehouse.2017. Structure-aware fuzzing for Clang and LLVM with libprotobuf-mutator. In 2017 USLLVM Developers’Meeting .https://llvm.org/devmtg/2017-10/slides/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "KostyaSerebryany,MaximLifantsev,KonstantinShtoyk,DougKwan,andPeter Hochschild.2021. SiliFuzz:FuzzingCPUsbyproxy. CoRRabs/2110.11519(2021). arXiv:2110.11519",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 98,
+      "text": "Evgeniy Stepanov and Konstantin Serebryany. 2015. MemorySanitizer: fast detectorofuninitializedmemoryuseinC++.In Proc.oftheInternationalSymposium on Code Generation and Optimization (CGO’15) (San Francisco, CA, USA). https://doi.org/10.1109/CGO.2015.7054186",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 99,
+      "text": "UBSan 2017. Undefined Behavior Sanitizer. https://clang.llvm.org/docs/ UndefinedBehaviorSanitizer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 100,
+      "text": "Phillip van Heerden, Moeketsi Raselimo, Konstantinos Sagonas, and Bernd Fischer. 2020. Grammar-Based Testing for Little Languages: An Experience Report with Student Compilers. In Proceedings of the 13th ACMSIGPLAN International Conference on Software Language Engineering (Virtual, USA) (SLE 2020). Association for Computing Machinery, New York, NY, USA, 253–269. https://doi.org/10.1145/3426425.3426",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 101,
+      "text": "Rijnard van Tonder and Claire Le Goues. 2019. Lightweight multi-language syntaxtransformationwithparserparsercombinators.In Proc.oftheConference on Programing Language Design and Implementation (PLDI’19) (Phoenix, AZ, USA).https://doi.org/10.1145/3314221.3314589",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 102,
+      "text": "RijnardvanTonderandAlexGroce.2022. MakingNo-FussCompilerFuzzing Effective: CC 2022 Artifact (0.1.0). https://doi.org/10.5281/zenodo.5982794. Zenodo.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 103,
+      "text": "JunjieWang,BihuanChen,LeiWei,andYangLiu.2019. Superion:GrammarAwareGreyboxFuzzing.In Proc.ofthe41stInternationalConferenceonSoftware Engineering(ICSE’19) (Montreal,Canada). https://doi.org/10.1109/ICSE.2019. 00081",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 104,
+      "text": "Wikipedia: GNU Compiler Collection. Date Accessed May 18, 2022. https: //en.wikipedia.org/wiki/GNU_Compiler_Collection#cite_note-loc-4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 105,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and Understanding Bugs in C Compilers. In Proc. of the Conference on Programing Language Design and Implementation (PLDI’11) (San Jose, CA, USA). https: //doi.org/10.1145/1993498.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 106,
+      "text": "Michal Zalewski. [n.d.]. Technical “whitepaper” for afl-fuzz. http://lcamtuf. coredump.cx/afl/technical_details.txt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 107,
+      "text": "Zig Software Foundation. Date Accessed September 1, 2022. Zig programming language. https://ziglang.org/. Received 2023-02-16; accepted 2023-05-03 1231",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 92,
+      "text": "ManuelRiggerandZhendongSu.2020.DetectingOptimizationBugsinDatabase Engines via Non-optimizing Reference Engine Construction. In Proc. of the Joint Meeting of the European Software Engineering Conference and the ACM SymposiumontheFoundationsofSoftwareEngineering(ESEC/FSE’20) (Online). https://doi.org/10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[92]",
+      "technique": "norec",
+      "sentence": "3598130 1 INTRODUCTION Over the last decade or so, randomised compiler testing, often termedcompiler fuzzing, has seen an explosion of interest, with compilerfuzzersleadingtothefindingand fixingof thousandsof bugs in C compilers such as Clang/ LLVMandGCC[65,82,105], aswell asincompilers forother languagessuchasOpenCL[ 67], OpenGL[ 17],SQL[92]and Verilog[ 57].",
+      "context_before": "doi.org/10.1145/3597926.3598130CCSCONCEPTS •Software and its engineering →Compilers ;Maintaining software;Software testing and debugging. KEYWORDS Greybox fuzzing, compilers, program analysers, code mutators, LibFuzzer,Clang,LLVM,GCC,MSVC,Frama-CACM Reference Format: KarineEven-Mendoza,ArindamSharma,AlastairF.Donaldson,andCristian Cadar. 2023. GrayC: Greybox Fuzzing of Compilers and Analysers for C. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software TestingandAnalysis(ISSTA’23),July17–21,2023,Seattle,WA,USA. ACM, NewYork, NY, USA, 13pages.https://doi.org/10.1145/3597926.",
+      "context_after": "Similar efforts have been proposedfortestingcodeanalysers,leadingtothediscoveryofbugs in popular frameworks such as model checkers, static analysers andsymbolic executors [ 16,61,63]. During roughly the same period, fuzzing has revolutionised the fieldofsoftwaretesting.However,mostcompilerfuzzersoperate very differently from mainstream general-purpose fuzzers, such as AFL [106]andLibFuzzer [75], which are coverage-directed and mutation-based .Takinginspirationfromgeneticalgorithms,such general-purposefuzzerssynthesisenewinputsbymutatingexisting ones, and use coverage feedback as a fitness functio",
+      "section": null,
+      "page": 1,
+      "char_offset": 3183,
+      "cited_reference": {
+        "number": 92,
+        "text": "ManuelRiggerandZhendongSu.2020.DetectingOptimizationBugsinDatabase Engines via Non-optimizing Reference Engine Construction. In Proc. of the Joint Meeting of the European Software Engineering Conference and the ACM SymposiumontheFoundationsofSoftwareEngineering(ESEC/FSE’20) (Online). https://doi.org",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[92]",
+      "technique": "norec",
+      "sentence": "[ 65,82,105]), but also on other languages such as OpenCL [ 67], OpenGL [ 17], SQL [92] and Verilog [57].",
+      "context_before": "surprisingthat Csmithdid not uncover any bugs during this controlled experiment. We note that Frama-C hasbeenextensivelytestedusing Csmithinthepast[ 16]. Grammarinator detectednobugs,probablyduetoitsextremely lowcompilationrateandthefactthematureahead-of-timecompilers’ front-ends have already been heavily tested. Similarly, GrayCFragments-Fuzzing ’spoorcoveragedelta(fromtheinitialcorpus) inboth LLVMandGCCcan explaintheseresults. 6 RELATED WORK Asdiscussedin§ 1,randomisedtestingtechniqueshavebeensuccessful in finding bugs in compilers for a range of languages, with a recent major focus on C (e.g.",
+      "context_after": "There has also been significant work on applying similar techniques to testingcode analysers(see e.g.[ 7,16,61,63]). Randomised compiler testing techniques mainly work by crosscheckingmultiplecompilers(e.g.[ 57,67,105],aformofdifferential testing [54,78], or checking expected equivalences between programs (e.g. [ 17,65]), a form of metamorphic testing [ 9,93]. We 1228 GrayC:GreyboxFuzzing of CompilersandAnalysers forC ISSTA ’23, July 17–21, 2023,Sea/t_tle, WA,USA refer the reader to a survey for an overview of state-of-the-art techniques [ 8] and toa recent paper for a discussionof the importan",
+      "section": "5.4 Bug Finding",
+      "page": 10,
+      "char_offset": 53438,
+      "cited_reference": {
+        "number": 92,
+        "text": "ManuelRiggerandZhendongSu.2020.DetectingOptimizationBugsinDatabase Engines via Non-optimizing Reference Engine Construction. In Proc. of the Joint Meeting of the European Software Engineering Conference and the ACM SymposiumontheFoundationsofSoftwareEngineering(ESEC/FSE’20) (Online). https://doi.org",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "GrayC brings coverage-directed, mutation-based greybox fuzzing to C compilers and analysers, which have become relatively resistant to the blackbox fuzzers used until now. The difficulty is that naive mutations produce programs that do not compile and so cannot reach deep bugs. GrayC introduces mutations targeting common C constructs and transforms fuzzed programs so they produce meaningful output, enabling differential testing. It found 30 confirmed bugs, 25 previously unknown.",
+    "narrative": "Two citations placing SQL among the languages, alongside OpenCL, OpenGL and Verilog, where randomised compiler-style testing has been applied.",
+    "roles": {
+      "M1": "background",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3605157_3605177.json b/_data/papers/paper_doi_10_1145_3605157_3605177.json
new file mode 100644
index 0000000..07ca336
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3605157_3605177.json
@@ -0,0 +1,474 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:07:12Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3605157.3605177",
+    "title": "Beyond the Coverage Plateau: A Comprehensive Study of Fuzz Blockers (Registered Report)",
+    "authors": [
+      "Wentao Gao",
+      "Van-Thuan Pham",
+      "Dongge Liu",
+      "Oliver Chang",
+      "Toby C. Murray",
+      "Benjamin I. P. Rubinstein"
+    ],
+    "year": 2023,
+    "venue": "FUZZING",
+    "doi": "10.1145/3605157.3605177",
+    "arxiv_id": null,
+    "s2_paper_id": "1b13b395ea1d7d0f23fc78df96ca430831408634",
+    "url": "https://doi.org/10.1145/3605157.3605177",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3605157.3605177",
+      "retrieved_at": "2026-09-07T17:07:12Z",
+      "chars": 50567,
+      "content_sha256": "sha256:ce31ab2a5fc1a2ba362ab6b3caa3fd20792027d8ca5d9ed5ef4bcd787bdddcfa"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 9,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2041
+      },
+      {
+        "number": "2.1",
+        "title": "Background of Coverage-Guided Greybox",
+        "start": 7197
+      },
+      {
+        "number": "2.2",
+        "title": "Introduction to Fuzz Introspector",
+        "start": 9272
+      },
+      {
+        "number": "2.3",
+        "title": "Motivating Example: A Challenging Fuzz",
+        "start": 10474
+      },
+      {
+        "number": "3.1",
+        "title": "Subject Selection",
+        "start": 13937
+      },
+      {
+        "number": "3.2",
+        "title": "Analysis Workflow",
+        "start": 18373
+      },
+      {
+        "number": "3.2.1",
+        "title": "Step-1. Understanding the subject program .It is worth",
+        "start": 18521
+      },
+      {
+        "number": "3.2.2",
+        "title": "Step-2. Identifying fuzz blockers .In this step, we use",
+        "start": 20274
+      },
+      {
+        "number": "3.2.3",
+        "title": "Step-3. Analyzing fuzz blockers .There could be several",
+        "start": 20800
+      },
+      {
+        "number": "3.2.4",
+        "title": "Step-4. Classifying fuzz blockers .We aim to classify the",
+        "start": 22855
+      },
+      {
+        "number": "4.1",
+        "title": "RQ1. Types of fuzz blockers",
+        "start": 26954
+      },
+      {
+        "number": "4.2",
+        "title": "RQ2. What makes them challenging?",
+        "start": 33968
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "[n. d.]. American Fuzzy Lop. https://lcamtuf.coredump.cx/afl/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "[n. d.]. Black Duck Open Hub. https://www.openhub.net/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "[n. d.]. ClusterFuzz. https://google.github.io/clusterfuzz/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "[n. d.]. Data Flow Sanitizer. https://clang.llvm.org/docs/DataFlowSanitizer.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "[n. d.]. Fuzz Introspector. https://github.com/ossf/fuzz-introspector",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "[n. d.]. Honggfuzz. https://github.com/google/honggfuzz",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "[n. d.]. Igraph The Network Analysis Package. https://igraph.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "[n. d.]. libFuzzer – a library for coverage-guided fuzz testing. https://github. com/llvm-mirror/llvm/blob/master/docs/LibFuzzer.rst",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "[n. d.]. libpng home page. http://www.libpng.org/pub/png/libpng.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "[n. d.]. libpng manual. http://www.libpng.org/pub/png/libpng-manual.txt",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "[n. d.]. libprotobuf-mutator. https://github.com/google/libprotobuf-mutator/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "[n. d.]. OpenSSL crytography and SSL/TLS toolket. https://www.openssl.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "[n. d.]. OSS-Fuzz: Continuous Fuzzing for Open Source Software. https://github. com/google/oss-fuzz",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "[n. d.]. Successful case studies of FuzzIntrospector. https://github.com/ossf/fuzzintrospector/blob/main/doc/CaseStudies.md",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Cornelius Aschermann, Tommaso Frassetto, Thorsten Holz, Patrick Jauernig, Ahmad-Reza Sadeghi, and Daniel Teuchert. 2019. NAUTILUS: Fishing for Deep Bugs with Grammars.. In NDSS.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Cornelius Aschermann, Sergej Schumilo, Ali Abbasi, and Thorsten Holz. 2020. Ijon: Exploring deep state spaces via fuzzing. In. IEEE, 1597–1612.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Cornelius Aschermann, Sergej Schumilo, Tim Blazytko, Robert Gawlik, and Thorsten Holz. 2019. REDQUEEN: Fuzzing with Input-to-State Correspondence.. InNDSS, Vol. 19. 1–15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jinsheng Ba, Marcel Böhme, Zahra Mirzamomen, and Abhik Roychoudhury. 2022. Stateful greybox fuzzing. In 31st USENIX Security Symposium (USENIX Security 22). 3255–3272.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Domagoj Babić, Stefan Bucur, Yaohui Chen, Franjo Ivančić, Tim King, Markus Kusano, Caroline Lemieux, László Szekeres, and Wei Wang. 2019. Fudge: fuzz driver generation at scale. In Proceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 975–985.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Tim Blazytko, Cornelius Aschermann, Moritz Schlögel, Ali Abbasi, Sergej Schumilo, Simon Wörner, and Thorsten Holz. 2019. GRIMOIRE: Synthesizing Structure while Fuzzing.. In USENIX Security Symposium, Vol. 19.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Marcel Böhme, Cristian Cadar, and Abhik Roychoudhury. 2021. Fuzzing: Challenges and Reflections. IEEE Softw. 38, 3 (2021), 79–86.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Marcel Böhme, Van-Thuan Pham, Manh-Dung Nguyen, and Abhik Roychoudhury. 2017. Directed greybox fuzzing. In Proceedings of the 2017 ACMSIGSAC Conference on Computer and Communications Security. 2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. 2016. CoverageBased Greybox Fuzzing as Markov Chain. IEEE Transactions on Software Engineering 45 (2016), 489–506.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Gemma Catolino, Fabio Palomba, Andy Zaidman, and Filomena Ferrucci. 2019. Not all bugs are the same: Understanding, characterizing, and classifying bug types. Journal of Systems and Software 152 (2019), 165–181.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Jaeseung Choi, Joonun Jang, Choongwoo Han, and Sang Kil Cha. 2019. Grey-box concolic testing on binary code. In on Software Engineering (ICSE). IEEE, 736–747.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Karine Even-Mendoza, Cristian Cadar, and Alastair F Donaldson. 2022. CsmithEdge: more effective compiler testing by handling undefined behaviour less conservatively. Empirical Software Engineering 27, 6 (2022), 129.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Andrea Fioraldi, Daniele Cono D’Elia, and Davide Balzarotti. 2021. The Use of Likely Invariants as Feedback for Fuzzers. In 30th USENIX Security Symposium (USENIX Security 21). USENIX Association, 2829–2846. https://www.usenix.org/ conference/usenixsecurity21/presentation/fioraldi",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. 2020. AFL++: Combining Incremental Steps of Fuzzing Research. In 14th USENIX Workshop on Offensive Technologies (WOOT 20). USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Shuitao Gan, Chao Zhang, Peng Chen, Bodong Zhao, Xiaojun Qin, Dong Wu, and Zuoning Chen. 2020. GREYONE: Data Flow Sensitive Fuzzing.. In USENIX Security Symposium. 2577–2594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Adrian Herrera, Hendra Gunadi, Shane Magrath, Michael Norrish, Mathias Payer, and Antony L Hosking. 2021. Seed selection for successful fuzzing. In Proceedings of the 30th ACMSIGSOFT International Symposium on Software Testing and Analysis. 230–243.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Renáta Hodován, Ákos Kiss, and Tibor Gyimóthy. 2018. Grammarinator: a grammar-based open source fuzzer. In Proceedings of the 9th ACMSIGSOFT international workshop on automating TEST case design, selection, and evaluation. 45–48.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Kyriakos K Ispoglou, Daniel Austin, Vishwath Mohan, and Mathias Payer. 2020. Fuzzgen: Automatic fuzzer generation. In Proceedings of the 29th USENIX Conference on Security Symposium. 2271–2287.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Jie Liang, Yu Jiang, Yuanliang Chen, Mingzhe Wang, Chijin Zhou, and Jiaguang Sun. 2018. Pafl: extend fuzzing optimizations of single mode to industrial parallel mode. In Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 809–814.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jie Liang, Mingzhe Wang, Yuanliang Chen, Yu Jiang, and Renwei Zhang. 2018. Fuzz testing in practice: Obstacles and solutions. In 2018 IEEE 25th International Conference on Software Analysis, Evolution and Reengineering (SANER). IEEE, 562–566.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Dongge Liu, Jonathan Metzman, Marcel Böhme, Oliver Chang, and Abhishek Arya. 2023. SBFT Tool Competition 2023–Fuzzing Track. arXiv preprint arXiv:2304.10070 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Valentin JM Manès, HyungSeok Han, Choongwoo Han, Sang Kil Cha, Manuel Egele, Edward J Schwartz, and Maverick Woo. 2019. The art, science, and engineering of fuzzing: A survey. IEEE Transactions on Software Engineering 47, 11 (2019), 2312–2331.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Jonathan Metzman, László Szekeres, Laurent Simon, Read Sprabery, and Abhishek Arya. 2021. Fuzzbench: an open fuzzer benchmarking platform and service. In Proceedings of the 29th ACM joint meeting on European software engineering conference and symposium on the foundations of software engineering. 1393–1403.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Sebastian Österlund, Elia Geretto, Andrea Jemmett, Emre Güler, Philipp Görz, Thorsten Holz, Cristiano Giuffrida, and Herbert Bos. 2021. Collabfuzz: A framework for collaborative fuzzing. In Proceedings of the 14th European Workshop on Systems Security. 1–7. 54 Beyond the Coverage Plateau: A Comprehensive Study of Fuzz Blockers (Registered Report) FUZZING ’23, July 17, 2023, Seattle, WA, USA",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Lianglu Pan, Shaanan Cohney, Toby Murray, and Van-Thuan Pham. 2023. Detecting Excessive Data Exposures in Web Server Responses with Metamorphic Fuzzing. arXiv preprint arXiv:2301.09258 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Jiwon Park, Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2021. Generative type-aware mutation for testing SMT solvers. Proceedings of the ACM on Programming Languages 5, OOPSLA (2021), 1–19.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Hui Peng and Mathias Payer. 2020. Usbfuzz: A framework for fuzzing USB drivers by device emulation. In Proceedings of the 29th USENIX Conference on Security Symposium. 2559–2575.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Van-Thuan Pham, Manh-Dung Nguyen, Quang-Trung Ta, Toby Murray, and Benjamin I.P. Rubinstein. 2021. Towards Systematic and Dynamic Task Allocation for Collaborative Parallel Fuzzing. In Proceedings of the 36th IEEE/ACM International Conference on Automated Software Engineering: NIER Track.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Van-Thuan Pham, Marcel Böhme, and Abhik Roychoudhury. 2020. AFLNet: a greybox fuzzer for network protocols. In on Software Testing, Validation and Verification (ICST). IEEE, 460–465.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Van-Thuan Pham, Marcel Böhme, Andrew E Santosa, Alexandru Răzvan Căciulescu, and Abhik Roychoudhury. 2019. Smart greybox fuzzing. IEEE Transactions on Software Engineering 47, 9 (2019), 1980–1997.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis.. In OSDI, Vol. 20. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 46,
+      "text": "Yang Song and Oscar Chaparro. 2020. BEE: a tool for structuring and analyzing bug reports. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1551–1555.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. 2019. Superion: Grammaraware greybox fuzzing. In ware Engineering (ICSE). IEEE, 724–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Qian Zhang, Jiyuan Wang, and Miryung Kim. 2021. Heterofuzz: Fuzz testing to detect platform dependent divergence for heterogeneous applications. In Proceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 242–254.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Zenong Zhang, George Klees, Eric Wang, Michael Hicks, and Shiyi Wei. 2023. Fuzzing Configurations of Program Options. ACM Transactions on Software Engineering and Methodology 32, 2 (2023), 1–21.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 955–970. 55",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 45,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis.. In OSDI, Vol. 20. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 39,45]",
+      "technique": "pqs",
+      "sentence": "These efforts have focused on enhancing fuzzing in areas such as feedback collection [ 16,25,27,30], corpus management [ 29], seed selection algorithms [ 22,23], input generation algorithms [ 15,20,31,44,47], and novel test oracle designs [ 39,45].",
+      "context_before": "be honored. For all other uses, contact the owner/author(s). FUZZING ’23, July 17, 2023, Seattle, WA, USA ©2023 Copyright held by the owner/author(s). ACMISBN 979-8-4007-0247-1/23/07. https://doi.org/10.1145/3605157.3605177vulnerability discovery. Popular CGF Fuzzers such as libFuzzer [8], AFL/AFL++ [1,28], and Honggfuzz [6] have discovered thousands of vulnerabilities in large real-world systems [3, 21, 36]. The state-of-the-art in fuzzing has seen significant advancements, with hundreds of research papers and dozens of tools being published to improve the technique in various aspects [ 36].",
+      "context_after": "Additionally, researchers have attempted to extend the applicability of fuzzing to challenging targets such as network protocols [18, 43], database systems [45, 50], SMT solvers [ 40], compilers [ 26], device drivers [ 41], and heterogeneous applications [ 48]. Another noteworthy research direction is parallel or distributed fuzzing [ 33,38,42], which aims to improve fuzzing efficiency by utilizing high-performance computing resources. Figure 1: SBFT’23 Fuzzing Competition [ 35] result of LibPNG. Mean branch coverage growth over time is reported. At least 20 trials/fuzzer, 23 hours per trial.",
+      "section": "1 INTRODUCTION",
+      "page": 1,
+      "char_offset": 3192,
+      "cited_reference": {
+        "number": 45,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis.. In OSDI, Vol. 20. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[45, 50]",
+      "technique": "pqs",
+      "sentence": "Additionally, researchers have attempted to extend the applicability of fuzzing to challenging targets such as network protocols [18, 43], database systems [45, 50], SMT solvers [ 40], compilers [ 26], device drivers [ 41], and heterogeneous applications [ 48].",
+      "context_before": "Popular CGF Fuzzers such as libFuzzer [8], AFL/AFL++ [1,28], and Honggfuzz [6] have discovered thousands of vulnerabilities in large real-world systems [3, 21, 36]. The state-of-the-art in fuzzing has seen significant advancements, with hundreds of research papers and dozens of tools being published to improve the technique in various aspects [ 36]. These efforts have focused on enhancing fuzzing in areas such as feedback collection [ 16,25,27,30], corpus management [ 29], seed selection algorithms [ 22,23], input generation algorithms [ 15,20,31,44,47], and novel test oracle designs [ 39,45].",
+      "context_after": "Another noteworthy research direction is parallel or distributed fuzzing [ 33,38,42], which aims to improve fuzzing efficiency by utilizing high-performance computing resources. Figure 1: SBFT’23 Fuzzing Competition [ 35] result of LibPNG. Mean branch coverage growth over time is reported. At least 20 trials/fuzzer, 23 hours per trial. Most fuzzers reach their plateau within 14 hours. Despite these advancements, the top-performing fuzzers continue to exhibit limitations as evidenced by results obtained from FuzzBench [37] and FuzzIntrospector [5]. As shown in Figure 1, all fuzzers participate",
+      "section": "1 INTRODUCTION",
+      "page": 1,
+      "char_offset": 3441,
+      "cited_reference": {
+        "number": 45,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis.. In OSDI, Vol. 20. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "A registered report studying fuzz blockers -- the obstacles that make coverage-guided greybox fuzzers plateau, which benchmarks show typically happens after around 12 hours. Using the FuzzIntrospector platform the authors analyse and categorise these blockers, finding that most top blockers are not directly related to the program input, which points to fuzz driver generation as the area needing better techniques.",
+    "narrative": "Two citations, placing DBMS fuzzing among the challenging targets researchers have extended fuzzing to.",
+    "roles": {
+      "M1": "background",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3611643_3613893.json b/_data/papers/paper_doi_10_1145_3611643_3613893.json
new file mode 100644
index 0000000..e326c2d
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3611643_3613893.json
@@ -0,0 +1,470 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:33:03Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3611643.3613893",
+    "title": "Adapting Performance Analytic Techniques in a Real-World Database-Centric System: An Industrial Experience Report",
+    "authors": [
+      "Lizhi Liao",
+      "Heng Li",
+      "Weiyi Shang",
+      "Catalin Sporea",
+      "Andrei Toma",
+      "S. Sajedi"
+    ],
+    "year": 2023,
+    "venue": "ESEC/SIGSOFT FSE",
+    "doi": "10.1145/3611643.3613893",
+    "arxiv_id": null,
+    "s2_paper_id": "b0b3eecc7e3953547ae24a909d5967240c19f373",
+    "url": "https://doi.org/10.1145/3611643.3613893",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3611643.3613893",
+      "retrieved_at": "2026-09-08T14:33:03Z",
+      "chars": 81027,
+      "content_sha256": "sha256:29bfef876402ca9d10288515893d28eaaa5ec1699e82fd59aa3853de2de2c374"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 3303
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND",
+        "start": 8559
+      },
+      {
+        "number": "3.1",
+        "title": "C1: Exposing the runtime measurement of",
+        "start": 15715
+      },
+      {
+        "number": "3.2",
+        "title": "C2: Exposing the source code from the",
+        "start": 23965
+      },
+      {
+        "number": "4.1",
+        "title": "P erformance a nomaly a nalysis within one",
+        "start": 32171
+      },
+      {
+        "number": "4.2",
+        "title": "P erformance r egression a nalysis between",
+        "start": 36503
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2014. CanaryRelease. Retrieved Mar. 7, 2023 from https://martinfowler.com/ bliki/CanaryRelease.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2017. sp_whoisactive SQL Server Monitoring Stored Procedure by Adam Machanic. Retrieved Jan. 23, 2023 from http://whoisactive.com",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2022. Apache Spark-Unified Engine for large-scale data analytics. Retrieved Mar. 15, 2023 from https://spark.apache.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2022. Apache Tomcat ®- Welcome! Retrieved Jan. 7, 2023 from https://tomcat. apache.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2022. typeperf|Microsoft. Retrieved Jan. 17, 2023 from https://learn.microsoft. com/en-us/windows-server/administration/windows-commands/typeperf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Juan Pablo Sandoval Alcocer, Alexandre Bergel, and Marco Tulio Valente. 2016. Learning from Source Code History to Identify Performance Failures. In Proceedings of the 7th ACM/SPEC International Conference on Performance Engineering, ICPE 2016, Delft, The Netherlands, March 12-16, 2016. ACM, 37–48.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Hamzeh Alimohammadi and Shengnan Nancy Chen. 2022. Performance evaluation of outlier detection techniques in production timeseries: A systematic review and meta-analysis. Expert Syst. Appl. 191 (2022), 116371.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Erik R. Altman, Matthew Arnold, Stephen Fink, and Nick Mitchell. 2010. Performance analysis of idle programs. In Proceedings of the 25th Annual ACMSIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications, OOPSLA 2010, October 17-21, 2010, Reno/Tahoe, Nevada, USA. ACM, 739–753.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "André B Bondi. 2015. Foundations of software and system performance engineering: process, performance modeling, requirements, testing, scalability, and practice. Pearson Education.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Leo Breiman. 2001. Random Forests. Mach. Learn. 45, 1 (2001), 5–32.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Boyuan Chen, Zhen Ming Jiang, Paul Matos, and Michael Lacaria. 2019. An Industrial Experience Report on Performance-Aware Refactoring on a Database-Centric Web Application. In 34th IEEE/ACM International Conference on Automated Software Engineering, ASE 2019, San Diego, CA, USA, November 11-15, 2019. IEEE, 653–664.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jinfu Chen and Weiyi Shang. 2017. An Exploratory Study of Performance Regression Introducing Code Changes. In Software Maintenance and Evolution, ICSME 2017, Shanghai, China, September 17-22, 2017. IEEE Computer Society, 341–352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Tse-Hsun Chen, Weiyi Shang, Ahmed E. Hassan, Mohamed N. Nasser, and Parminder Flora. 2016. Detecting problems in the database access code of large scale systems: an industrial experience report. In Proceedings of the 38th International Conference on Software Engineering, ICSE 2016, Austin, TX, USA, May 14-22, 2016 Companion Volume. ACM, 71–80.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Tse-Hsun Chen, Weiyi Shang, Zhen Ming Jiang, Ahmed E. Hassan, Mohamed N. Nasser, and Parminder Flora. 2014. Detecting performance anti-patterns for applications developed using object-relational mapping. In 36th International Conference on Software Engineering, ICSE ’14, Hyderabad, India-May 31 - June 07, 2014. ACM, 1001–1012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Tse-Hsun Chen, Weiyi Shang, Zhen Ming Jiang, Ahmed E. Hassan, Mohamed N. Nasser, and Parminder Flora. 2016. Finding and Evaluating the Performance Impact of Redundant Data Access for Applications that are Developed Using Object-Relational Mapping Frameworks. IEEE Trans. Software Eng. 42, 12 (2016), 1148–1161.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Tse-Hsun Chen, Mark D. Syer, Weiyi Shang, Zhen Ming Jiang, Ahmed E. Hassan, Mohamed N. Nasser, and Parminder Flora. 2017. Analytics-Driven Load Testing: An Industrial Experience Report on Load Testing of Large-Scale Systems. In 39th IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice Track, ICSE-SEIP 2017, Buenos Aires, Argentina, May 20-28, 2017. IEEE Compu",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "King Chun Foo, Zhen Ming Jiang, Bram Adams, Ahmed E. Hassan, Ying Zou, and Parminder Flora. 2010. Mining Performance Regression Testing Repositories for Automated Performance Analysis. In Proceedings of the 10th International Conference on Quality Software, QSIC 2010, Zhangjiajie, China, 14-15 July 2010. IEEE Computer Society, 32–41.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "King Chun Foo, Zhen Ming Jiang, Bram Adams, Ahmed E. Hassan, Ying Zou, and Parminder Flora. 2015. An Industrial Case Study on the Automated Detection of Performance Regressions in Heterogeneous Environments. In 37th IEEE/ACM International Conference on Software Engineering, ICSE 2015, Florence, Italy, May 16-24, 2015, Volume 2. IEEE Computer Society, 159–168.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "David A Freedman. 2009. Statistical models: theory and practice. cambridge university press.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Mark Grechanik, B. M. Mainul Hossain, and Ugo A. Buy. 2013. Testing DatabaseCentric Applications for Causes of Database Deadlocks. In Sixth IEEE International Conference on Software Testing, Verification and Validation, ICST 2013, Luxembourg, Luxembourg, March 18-22, 2013. IEEE Computer Society, 174–183.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Mark Grechanik, B. M. Mainul Hossain, Ugo A. Buy, and Haisheng Wang. 2013. Preventing database deadlocks in applications. In Joint Meeting of the European Software Engineering Conference and the ACMSIGSOFT Symposium on the Foundations of Software Engineering, ESEC/FSE’13, Saint Petersburg, Russian Federation, August 18-26, 2013. ACM, 356–366.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Wilhelm Hasselbring and André van Hoorn. 2020. Kieker: A monitoring framework for software engineering research. Softw. Impacts 5 (2020), 100019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Christoph Heger, Jens Happe, and Roozbeh Farahbod. 2013. Automated root cause isolation of performance regressions during software development. In ACM/SPEC International Conference on Performance Engineering, ICPE’13, Prague, Czech Republic-April 21 - 24, 2013. ACM, 27–38.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Zhen Ming Jiang, Ahmed E. Hassan, Gilbert Hamann, and Parminder Flora. 2008. Automatic identification of load testing problems. In 24th IEEE International Conference on Software Maintenance (ICSM 2008), September 28 - October 4, 2008, Beijing, China. IEEE Computer Society, 307–316.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Zhen Ming Jiang, Ahmed E. Hassan, Gilbert Hamann, and Parminder Flora. 2009. Automated performance analysis of load tests. In 25th IEEE International Conference on Software Maintenance (ICSM 2009), September 20-26, 2009, Edmonton, Alberta, Canada. IEEE Computer Society, 125–134.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Toon Koppelaars. 2004. A database-centric approach to j2ee application development. Oracle Development Tools Users Group (ODTUG) (2004).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Lizhi Liao, Jinfu Chen, Heng Li, Yi Zeng, Weiyi Shang, Catalin Sporea, Andrei Toma, and Sarah Sajedi. 2021. Locating Performance Regression Root Causes in the Field Operations of Web-based Systems: An Experience Report. IEEE Transactions on Software Engineering (2021), 1–1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Siyang Lu, BingBing Rao, Xiang Wei, Byung-Chul Tak, Long Wang, and Liqiang Wang. 2017. Log-based Abnormal Task Detection and Root Cause Analysis for Spark. In HI, USA, June 25-30, 2017. IEEE, 389–396.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Qi Luo, Denys Poshyvanyk, and Mark Grechanik. 2016. Mining performance regression inducing code changes in evolving software. In Proceedings of the 13th International Conference on Mining Software Repositories, MSR 2016, Austin, TX, USA, May 14-22, 2016. ACM, 25–36.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Yingjun Lyu, Sasha Volokh, William G. J. Halfond, and Omer Tripp. 2021. SAND: a static analysis approach for detecting SQL antipatterns. In ISSTA ’21: 30th ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, Denmark, July 11-17, 2021. ACM, 270–282.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Haroon Malik, Hadi Hemmati, and Ahmed E. Hassan. 2013. Automatic detection of performance deviations in the load testing of large scale systems. In 35th International Conference on Software Engineering, ICSE ’13, San Francisco, CA, USA, May 18-26, 2013. IEEE Computer Society, 1012–1021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Leonardo Mariani, Cristina Monni, Mauro Pezzè, Oliviero Riganelli, and Rui Xin. 2018. Localizing Faults in Cloud Systems. In 11th IEEE International Conference on Software Testing, Verification and Validation, ICST 2018, Västerås, Sweden, April 9-13, 2018. IEEE Computer Society, 262–273.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Loup Meurice, Csaba Nagy, and Anthony Cleve. 2016. Static analysis of dynamic database usage in java systems. In Advanced Information Systems Engineering: 28th International Conference, CAiSE 2016, Ljubljana, Slovenia, June 13-17, 2016. Proceedings 28. Springer, 491–506.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Thanh H. D. Nguyen. 2012. Using Control Charts for Detecting and Understanding Performance Regressions in Large Software. In Fifth IEEE International Conference on Software Testing, Verification and Validation, ICST 2012, Montreal, QC, Canada, April 17-21, 2012. IEEE Computer Society, 491–494.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Thanh H. D. Nguyen, Bram Adams, Zhen Ming Jiang, Ahmed E. Hassan, Mohamed N. Nasser, and Parminder Flora. 2011. Automated Verification of Load Tests Using Control Charts. In 18th Asia Pacific Software Engineering Conference, APSEC 2011, Ho Chi Minh, Vietnam, December 5-8, 2011. IEEE Computer Society, 282–289.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Thanh H. D. Nguyen, Bram Adams, Zhen Ming Jiang, Ahmed E. Hassan, Mohamed N. Nasser, and Parminder Flora. 2012. Automated detection of performance regressions using statistical process control techniques. In Third Joint WOSP/SIPEW International Conference on Performance Engineering, ICPE’12, Boston, MA, USA-April 22 - 25, 2012. ACM, 299–310.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Thanh H. D. Nguyen, Meiyappan Nagappan, Ahmed E. Hassan, Mohamed N. Nasser, and Parminder Flora. 2014. An industrial case study of automatically identifying performance regression-causes. In 11th Working Conference on Mining Software Repositories, MSR 2014, Proceedings, May 31 - June 1, 2014, Hyderabad, India. ACM, 232–241.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "George Papadimitriou, Cong Wang, Karan Vahi, Rafael Ferreira da Silva, Anirban Mandal, Zhengchun Liu, Rajiv Mayani, Mats Rynge, Mariam Kiran, Vickie E. Lynch, Rajkumar Kettimuthu, Ewa Deelman, Jeffrey S. Vetter, and Ian T. Foster. 2021. End-to-end online performance data capture and analysis for scientific workflows. Future Gener. Comput. Syst. 117 (2021), 387–400.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "John C. Platt. 1999. Fast Training of Support Vector Machines Using Sequential Minimal Optimization. MIT Press, Cambridge, MA, USA, 185–208.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "John W Ratcliff, David Metzener, et al .1988. Pattern matching: The gestalt approach. Dr. Dobb’s Journal 13, 7 (1988), 46.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "David Georg Reichelt, Stefan Kühne, and Wilhelm Hasselbring. 2019. PeASS: A Tool for Identifying Performance Changes at Code Level. In 34th IEEE/ACM International Conference on Automated Software Engineering, ASE 2019, San Diego, CA, USA, November 11-15, 2019. IEEE, 1146–1149.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In ESEC/FSE ’20: 28th 1865 ESEC/FSE ’23, December 3–9, 2023, San Francisco, CA, USA Lizhi Liao et al. ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020. ACM, 1140–115",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 45,
+      "text": "Peter J Rousseeuw and Mia Hubert. 2011. Robust statistics for outlier detection. Wiley interdisciplinary reviews: Data mining and knowledge discovery 1, 1 (2011), 73–79.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Weiyi Shang, Ahmed E. Hassan, Mohamed N. Nasser, and Parminder Flora. 2015. Automated Detection of Performance Regressions Using Regression Models on Clustered Performance Counters. In Proceedings of the 6th ACM/SPEC International Conference on Performance Engineering, Austin, TX, USA, January 31 February 4, 2015. ACM, 15–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Connie U. Smith. 2007. Introduction to Software Performance Engineering: Origins and Outstanding Problems. In Formal Methods for Performance Evaluation, 7th International School on Formal Methods for the Design of Computer, Communication, and Software Systems, SFM 2007, Bertinoro, Italy, May 28-June 2, 2007, Advanced Lectures (Lecture Notes in Computer Science, Vol. 4486). Springer, 395–428.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Connie U. Smith and Lloyd G. Williams. 2002. New Software Performance AntiPatterns: More Ways to Shoot Yourself in the Foot. In 28th International Computer Measurement Group Conference, December 8-13, 2002, Reno, Nevada, USA, Proceedings. Computer Measurement Group, 667–674.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Connie U. Smith and Lloyd G. Williams. 2003. More New Software Antipatterns: Even More Ways to Shoot Yourself in the Foot. In 29th International Computer Measurement Group Conference, December 7-12, 2003, Dallas, Texas, USA, Proceedings. Computer Measurement Group, 717–725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Connie U. Smith and Lloyd G. Williams. 2003. Software Performance Engineering. InUML for Real-Design of Embedded Real-Time Systems. Kluwer, 343–365.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "André van Hoorn, Jan Waller, and Wilhelm Hasselbring. 2012. Kieker: a framework for application performance monitoring and dynamic software analysis. InThird Joint WOSP/SIPEW International Conference on Performance Engineering, ICPE’12, Boston, MA, USA-April 22 - 25, 2012. ACM, 247–248.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "HP Vinutha, B Poornima, and BM Sagar. 2018. Detection of outliers using interquartile range technique from intrusion dataset. In Information and decision sciences. Springer, 511–518.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Frank Wilcoxon. 1992. Individual comparisons by ranking methods. In Breakthroughs in statistics. Springer, 196–202.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "C. Murray Woodside, Greg Franks, and Dorina C. Petriu. 2007. The Future of Software Performance Engineering. In International Conference on Software Engineering, ISCE 2007, Workshop on the Future of Software Engineering, FOSE 2007, May 23-25, 2007, Minneapolis, MN, USA. IEEE Computer Society, 171–187.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "PengCheng Xiong, Calton Pu, Xiaoyun Zhu, and Rean Griffith. 2013. vPerfGuard: an automated model-driven framework for application performance diagnosis in consolidated cloud environments. In ACM/SPEC International Conference on Performance Engineering, ICPE’13, Prague, Czech Republic-April 21 - 24, 2013. ACM, 271–282.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Cong Yan, Alvin Cheung, Junwen Yang, and Shan Lu. 2017. Understanding Database Performance Inefficiencies in Real-world Web Applications. In Proceedings of the 2017 ACM on Conference on Information and Knowledge Management, CIKM 2017, Singapore, November 06 - 10, 2017. ACM, 1299–1308.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Junwen Yang, Pranav Subramaniam, Shan Lu, Cong Yan, and Alvin Cheung. 2018. How notto structure your database-backed web applications: a study of performance bugs in the wild. In Proceedings of the 40th International Conference on Software Engineering, ICSE 2018, Gothenburg, Sweden, May 27 - June 03, 2018. ACM, 800–810.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Junwen Yang, Cong Yan, Pranav Subramaniam, Shan Lu, and Alvin Cheung. 2018. PowerStation: automatically detecting and fixing inefficiencies of databasebacked web applications in IDE. In Proceedings of the 2018 ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/SIGSOFT FSE 2018, Lake Buena Vista, FL, USA, November 04-09, 2018",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 42,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In ESEC/FSE ’20: 28th 1865 ESEC/FSE ’23, December 3–9, 2023, San Francisco, CA, USA Lizhi Liao et al. ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020. ACM, 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "Rigger and Su utilize multiple analysis techniques, such as query partitioning [ 43], pivoted query synthesis [ 44], or non-optimizing reference engine construction [ 42] to detect database logic bugs that are critical in databasecentric system quality but often go unnoticed by developers.",
+      "context_before": "stinct from them, we target a large-scale database-centric system from the industry (i.e., EA) with special attention to not only the runtime behavior and performance of the database but also its interactions with the surrounding applications. Improving database-centric system quality. Due to the widespread use of databases in practice, a great number of studies have been conducted to improve database-centric system quality. Grechanik et al. [ 20,21] develop approaches that utilize static and dynamic code analysis techniques to detect and prevent database deadlocks in database-centric systems.",
+      "context_after": "Prior studies also discuss and summarize many common performance issues (e.g., source code-level performance anti-patterns [ 48, 49] or ORM-level performance anti-patterns [ 56,57]) in databasecentric systems and suggest solutions to alleviate or eliminate the adverse performance impact. Based on these studies, extensive work has been proposed to detect performance anti-patterns, for example, Yan et al. [ 58] develop a RubyMine IDE plugin for Ruby on Rails systems by using static code analysis techniques to automatically detect ORM-level performance issues and provide the corresponding fixes.",
+      "section": "4.2 P erformance r egression a nalysis between",
+      "page": 10,
+      "char_offset": 60492,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec",
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "An industrial experience report on adapting performance analysis to a large-scale database-centric system. The authors argue that in such systems most business logic and calculation live in the database rather than the application, and that because those calculations are written in SQL, both the performance issues and their diagnosis differ from systems dominated by conventional programming languages.",
+    "narrative": "A single citation naming SQLancer's three oracles -- query partitioning, pivoted query synthesis and non-optimizing reference engine construction -- as analysis techniques Rigger and Su apply to databases.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3611643_3616286.json b/_data/papers/paper_doi_10_1145_3611643_3616286.json
new file mode 100644
index 0000000..79c4425
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3611643_3616286.json
@@ -0,0 +1,622 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:15Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3611643.3616286",
+    "title": "Property-Based Fuzzing for Finding Data Manipulation Errors in Android Apps",
+    "authors": [
+      "Jingling Sun",
+      "Ting Su",
+      "Jiayi Jiang",
+      "Jue Wang",
+      "G. Pu",
+      "Zhendong Su"
+    ],
+    "year": 2023,
+    "venue": "ESEC/SIGSOFT FSE",
+    "doi": "10.1145/3611643.3616286",
+    "arxiv_id": null,
+    "s2_paper_id": "8ea47327053957637ce8782a8b22aad19878dea1",
+    "url": "https://doi.org/10.1145/3611643.3616286",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3611643.3616286",
+      "retrieved_at": "2026-09-09T01:33:15Z",
+      "chars": 85782,
+      "content_sha256": "sha256:df2236c971be2c7c3174ddf2d5b9020cbb076b17ffb1a8c11a546cd4756a70ee"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 3719
+      },
+      {
+        "number": "2",
+        "title": "APPROACH",
+        "start": 13537
+      },
+      {
+        "number": "2.1",
+        "title": "High-Level Idea",
+        "start": 13651
+      },
+      {
+        "number": "2.2",
+        "title": "Challenges and Observations",
+        "start": 17411
+      },
+      {
+        "number": "2.3",
+        "title": "Instantiation of the High-Level Idea",
+        "start": 21028
+      },
+      {
+        "number": "3",
+        "title": "IMPLEMENTATION",
+        "start": 25188
+      },
+      {
+        "number": "3.1",
+        "title": "Input Generator",
+        "start": 25914
+      },
+      {
+        "number": "3.2",
+        "title": "Data Recorder",
+        "start": 29790
+      },
+      {
+        "number": "3.3",
+        "title": "Property Checker",
+        "start": 30243
+      },
+      {
+        "number": "3.4",
+        "title": "DMF Instantiator",
+        "start": 30717
+      },
+      {
+        "number": "4",
+        "title": "EVALUATION",
+        "start": 35371
+      },
+      {
+        "number": "4.1",
+        "title": "Evaluation Setup",
+        "start": 37112
+      },
+      {
+        "number": "5",
+        "title": "DISCUSSION",
+        "start": 57720
+      },
+      {
+        "number": "6",
+        "title": "RELATED WORK",
+        "start": 59737
+      },
+      {
+        "number": "7",
+        "title": "CONCLUSION",
+        "start": 65291
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Christoffer Quist Adamsen, Gianluca Mezzetti, and Anders Møller. 2015. Systematic execution of android test suites in adverse conditions. In Proceedings of the 2015 International Symposium on Software Testing and Analysis (ISSTA). 83–93. https://doi.org/10.1145/2771783.2771786",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Domenico Amalfitano, Anna Rita Fasolino, Porfirio Tramontana, Bryan Dzung Ta, and Atif M. Memon. 2015. MobiGUITAR: Automated Model-Based Testing of Mobile Apps. IEEE Software (2015), 53–59. https://doi.org/10.1109/MS.2014.55",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "AnkiDroid Team. 2022. AnkiDroid. Retrieved 2023-1 from https://github.com/ ankidroid/Anki-Android.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "AnyMemo Team. 2022. AnyMemo. Retrieved 2023-1 from https://github.com/ helloworld1/AnyMemo.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Young-Min Baek and Doo-Hwan Bae. 2016. Automated model-based Android GUI testing using multi-level GUI comparison criteria. In 2016 31st IEEE/ACM International Conference on Automated Software Engineering (ASE). 238–249. https: //doi.org/10.1145/2970276.2970313",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Earl T Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. 2014. The oracle problem in software testing: A survey. IEEE transactions on software engineering (TSE) (2014), 507–525. https://doi.org/10.1109/TSE.2014.2372785",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Farnaz Behrang and Alessandro Orso. 2019. Test migration between mobile apps with similar functionality. In 2019 34th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 54–65.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Benoît Joossen. 2018. Voice Memos “Epidemic Failure” and How to Avoid it. Retrieved 2023-1 from https://aeroquartet.com/wordpress/2019/03/05/voice-memosepidemic-failure-and-how-to-avoid-it/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Manfred Broy, Bengt Jonsson, Joost-Pieter Katoen, Martin Leucker, and Alexander Pretschner. 2005. Model-based testing of reactive systems: advanced lectures. Springer Science & Business Media.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Business of Apps. 2022. Android Statistics (2022). Retrieved 2023-1 from https: //www.businessofapps.com/data/android-statistics/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "CapCut Team. 2021. CapCut. Retrieved 2023-1 from https://lv.faceueditor.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Chunyang Chen, Ting Su, Guozhu Meng, Zhenchang Xing, and Yang Liu. 2018. From ui design image to gui skeleton: a neural machine translator to bootstrap mobile gui implementation. In Proceedings of the 40th International Conference on Software Engineering (ICSE). 665–676. https://doi.org/10.1145/3180155.3180240",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Tsong Y. Chen, Shing C. Cheung, and Shiu Ming Yiu. 2020. Metamorphic testing: a new approach for generating next test cases. Technical Report. HKUST-CS98-01, Hong Kong University of Science and Technology. https://doi.org/10.48550/ arXiv.2002.12543",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Shauvik Roy Choudhary, Alessandra Gorla, and Alessandro Orso. 2015. Automated Test Input Generation for Android: Are We There Yet? (E). In 30th IEEE/ACM International Conference on Automated Software Engineering (ASE). 429–440. https://doi.org/10.1109/ASE.2015.89",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Koen Claessen and John Hughes. 2000. QuickCheck: a lightweight tool for random testing of Haskell programs. In Proceedings of the Fifth ACMSIGPLAN International Conference on Functional Programming (ICFP). 268–279. https: //doi.org/10.1145/351240.351266",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Koen Claessen and John Hughes. 2000. QuickCheck: a lightweight tool for random testing of Haskell programs. In Proceedings of the fifth ACMSIGPLAN international conference on Functional programming (ICFP). 268–279. https: //doi.org/10.1145/357766.351266",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Pedro Costa, Ana CR Paiva, and Miguel Nabuco. 2014. Pattern based GUI testing for mobile applications. In Proceedings of the 9th International Conference on the Quality of Information and Communications Technology (QUATIC). 66–74. https://doi.org/10.1109/QUATIC.2014.16",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Guilherme de Cleva Farto and Andre Takeshi Endo. 2015. Evaluating the modelbased testing approach in the context of mobile applications. Electronic notes in Theoretical computer science (2015), 3–21. https://doi.org/10.1016/j.entcs.2015.05. 002",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Zhen Dong, Marcel Böhme, Lucia Cojocaru, and Abhik Roychoudhury. 2020. Time-travel testing of Android apps. In Proceedings of the 42nd International Conference on Software Engineering (ICSE). 1–12. https://doi.org/10.1145/3377811. 3380402",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Dawson R. Engler, David Yu Chen, and Andy Chou. 2001. Bugs as Deviant Behavior: A General Approach to Inferring Errors in Systems Code. In Proceedings of the 18th ACM Symposium on Operating System Principles (SOSP 2001). 57–72. https://doi.org/10.1145/502034.502041",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Lingling Fan, Ting Su, Sen Chen, Guozhu Meng, Yang Liu, Lihua Xu, Geguang Pu, and Zhendong Su. 2018. Large-scale analysis of framework-specific exceptions in Android apps. In Proceedings of the 40th International Conference on SoftwareEngineering (ICSE), Michel Chaudron, Ivica Crnkovic, Marsha Chechik, and Mark Harman (Eds.). 408–419. https://doi.org/10.1145/3180155.3180222",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Mattia Fazzini and Alessandro Orso. 2017. Automated cross-platform inconsistency detection for mobile apps. In 2017 32nd IEEE/ACM International Conference on Automated Software Engineering (ASE). 308–318. https://doi.org/10.1109/ASE. 2017.8115644",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "FeiShu Team. 2021. FeiShu. Retrieved 2023-1 from https://www.feishu.cn/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "GeeksforGeeks. 2020. MVC (Model View Controller) Architecture Pattern in Android. Retrieved 2023-1 from https://www.geeksforgeeks.org/mvc-model-viewcontroller-architecture-pattern-in-android-with-example/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Google. 2021. Google Play. Retrieved 2023-1 from https://play.google.com/store.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Tianxiao Gu, Chengnian Sun, Xiaoxing Ma, Chun Cao, Chang Xu, Yuan Yao, Qirun Zhang, Jian Lu, and Zhendong Su. 2019. Practical GUI testing of Android applications via model abstraction and refinement. In 2019 IEEE/ACM 41st International Conference on Software Engineering (ICSE). 269–280. https: //doi.org/10.1109/ICSE.2019.00042",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Wunan Guo, Zhen Dong, Liwei Shen, Wei Tian, Ting Su, and Xin Peng. 2022. Detecting and fixing data loss issues in Android apps. In ISSTA ’22: 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, South Korea, July 18 - 22, 2022. 605–616.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Wunan Guo, Zhen Dong, Liwei Shen, Wei Tian, Ting Su, and Xin Peng. 2022. Detecting and fixing data loss issues in Android apps. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 605–616. https://doi.org/10.1145/3533767.3534402",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Hacker News. 2013. Tell Facebook: There’s a severe bug when changing profile pics on the iOS app. Retrieved 2023-1 from https://news.ycombinator.com/item?id= 6456285.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Gang Hu, Linjie Zhu, and Junfeng Yang. 2018. AppFlow: using machine learning to synthesize robust, reusable UI tests. In Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 269–282. https://doi.org/10. 1145/3236024.3236055",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "John Hughes. 2020. How to Specify It!. In Trends in Functional Programming. Springer International Publishing, 58–83.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "k-9 Team. 2022. k-9. Retrieved 2023-1 from https://github.com/thundernest/k-9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Stefan Karlsson, Adnan Causevic, Daniel Sundmark, and Mårten Larsson. 2021. Model-based Automated Testing of Mobile Applications: An Industrial Case Study. In 14th IEEE International Conference on Software Testing, Verification and Validation Workshops (ICSTW 2021. 130–137.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Stefan Karlsson, Adnan Čaušević, and Daniel Sundmark. 2020. QuickREST: Property-based Test Generation of OpenAPI-Described RESTful APIs. In (ICST). 131–141. https://doi.org/10.1109/ICST46399.2020.00023",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Pingfan Kong, Li Li, Jun Gao, Kui Liu, Tegawendé F. Bissyandé, and Jacques Klein. 2019. Automated Testing of Android Apps: A Systematic Literature Review. IEEE Trans. Reliability 68, 1 (2019), 45–66. https://doi.org/10.1109/TR.2018.2865733",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Edmund SL Lam, Peilun Zhang, and Bor-Yuh Evan Chang. 2017. ChimpCheck: property-based randomized test generation for interactive apps. In Proceedings of the 2017 ACMSIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software (Onward!). 58–77. https://doi.org/ 10.1145/3133850.3133853",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Leonidas Lampropoulos, Michael Hicks, and Benjamin C. Pierce. 2019. Coverage guided, property based testing. Proc. ACM Program. Lang. OOPSLA (2019), 181:1– 181:29. https://doi.org/10.1145/3360607",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Yuanchun Li, Ziyue Yang, Yao Guo, and Xiangqun Chen. 2017. Droidbot: a lightweight ui-guided test input generator for android. In Proceedings of the (ICSE-C). 23–26. https://doi.org/10.1109/ICSE-C.2017.8",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Jun-Wei Lin, Reyhaneh Jabbarvand, and Sam Malek. 2019. Test transfer across mobile apps through semantic mapping. In 2019 34th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 42–53. https://doi. org/10.1109/ASE.2019.00015",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Ying-Dar Lin, José F. Rojas, Edward T.-H. Chu, and Yuan-Cheng Lai. 2014. On the Accuracy, Efficiency, and Reusability of Automated Test Oracles for Android Devices. IEEE Trans. Software Eng. (2014), 957–970. https://doi.org/10.1109/TSE. 2014.2331982",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Zhengwei Lv, Chao Peng, Zhao Zhang, Ting Su, Kai Liu, and Ping Yang. 2022. Fastbot2: Reusable Automated Model-based GUI Testing for Android Enhanced by Reinforcement Learning. In 37th IEEE/ACM International Conference on Automated Software Engineering (ASE). 135:1–135:5. https://doi.org/10.1145/3551349.3559505 1099 Property-Based Fuzzing for Finding Data Manipulation Errors in Android Apps ESEC/FS",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Ke Mao, Mark Harman, and Yue Jia. 2016. Sapienz: multi-objective automated testing for Android applications. In Proceedings of the 25th International Symposium on Software Testing and Analysis (ISSTA). 94–105. https://doi.org/10.1145/ 2931037.2931054",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Leonardo Mariani, Mauro Pezzè, and Daniele Zuddas. 2018. Augusto: Exploiting Popular Functionalities for the Generation of Semantic GUI Tests with Oracles. InProceedings of the 40th International Conference on Software Engineering (ICSE). 280–290. https://doi.org/10.1145/3180155.3180162",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Monkey Team. 2022. Android Monkey. Retrieved 2023-1 from https://developer. android.com/studio/test/monkey.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Liam O’Connor and Oskar Wickström. 2022. Quickstrom: property-based acceptance testing with LTL specifications. In Proceedings of the 43rd ACMSIGPLAN International Conference on Programming Language Design and Implementation (PLDI). 1025–1038. https://doi.org/10.1145/3519939.3523728",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Carlos Pacheco, Shuvendu K Lahiri, Michael D Ernst, and Thomas Ball. 2007. Feedback-directed random test generation. In 29th International Conference on Software Engineering (ICSE). 75–84. https://doi.org/10.1109/ICSE.2007.37",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Sumit Padhiyar and KC Sivaramakrishnan. 2021. ConFuzz: Coverage-guided property fuzzing for event-driven programs. In Practical Aspects of Declarative Languages: 23rd International Symposium, PADL 2021, Copenhagen, Denmark, January 18-19, 2021, Proceedings 23. 127–144. https://doi.org/10.1007/978-3-03067438-0_8",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Rohan Padhye, Caroline Lemieux, and Koushik Sen. 2019. JQF: coverageguided property-based testing in Java. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 398–401. https://doi.org/10.1145/3293882.3339002",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Minxue Pan, An Huang, Guoxin Wang, Tian Zhang, and Xuandong Li. 2020. Reinforcement Learning Based Curiosity-Driven Testing of Android Applications. InProceedings of the 29th ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 153–164. https://doi.org/10.1145/3395363.3397354",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Oliviero Riganelli, Simone Paolo Mottadelli, Claudio Rota, Daniela Micucci, and Leonardo Mariani. 2020. Data loss detector: automatically revealing data loss bugs in Android apps. In Proceedings of the 29th ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 141–152. https://doi.org/10. 1145/3395363.3397379",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. OOPSLA (2020), 211:1–211:30. https://doi.org/10.5281/zenodo.4032401",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 52,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI). USENIX Association, 667–682. https://doi.org/10.48550/ arXiv.2001.04174",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 53,
+      "text": "Ariel Rosenfeld, Odaya Kardashov, and Orel Zang. 2018. Automation of android applications functional testing using machine learning activities classification. InProceedings of the 5th International Conference on Mobile Software Engineering and Systems (MOBILESoft). 122–132. https://doi.org/10.1145/3197231.3197241",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Iflaah Salman, Ayse Tosun Misirli, and Natalia Juristo. 2015. Are students representatives of professionals in software engineering experiments?. In Proceedings of the (ICSE). 666–676. https://doi.org/10.5555/2818754.2818836",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "André Santos, Alcino Cunha, and Nuno Macedo. 2018. Property-based testing for the robot operating system. In Proceedings of the 9th ACMSIGSOFT International Workshop on Automating TEST Case Design, Selection, and Evaluation. 56–62.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "SimpleTask Team. 2022. SimpleTask. Retrieved 2023-1 from https://github.com/ mpcjanssen/simpletask-android.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Sixth Tone. 2019. E-Commerce App Loses ‘Tens of Millions’ From Coupon Glitches. Retrieved 2023-1 from https://www.sixthtone.com/news/1003483/e-commerceapp-loses-tens-of-millions-from-coupon-glitches.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "SkyTube Team. 2022. SkyTube. Retrieved 2023-1 from https://github.com/ SkyTubeTeam/SkyTube.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Ting Su, Lingling Fan, Sen Chen, Yang Liu, Lihua Xu, Geguang Pu, and Zhendong Su. 2022. Why My App Crashes? Understanding and Benchmarking FrameworkSpecific Exceptions of Android Apps. IEEE Trans. Software Eng. 48, 4 (2022), 1115–1137. https://doi.org/10.1109/TSE.2020.3013438",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Ting Su, Guozhu Meng, Yuting Chen, Ke Wu, Weiming Yang, Yao Yao, Geguang Pu, Yang Liu, and Zhendong Su. 2017. Guided, stochastic model-based GUI testing of Android apps. In Proceedings of the 2017 11th Joint Meeting on Foundations of Software Engineering (ESEC/FSE). 245–256. https://doi.org/10.1145/3106237. 3106298",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Ting Su, Jue Wang, and Zhendong Su. 2021. Benchmarking automated GUI testing for Android against real-world bugs. In 29th ACM Joint European SoftwareEngineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 119–130. https://doi.org/10.1145/3468264.3468620",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Ting Su, Yichen Yan, Jue Wang, Jingling Sun, Yiheng Xiong, Geguang Pu, Ke Wang, and Zhendong Su. 2021. Fully automated functional fuzzing of Android apps for detecting non-crashing logic bugs. Proceedings of the ACM on Programming Languages (OOPSLA) (2021), 1–31. https://doi.org/10.1145/3485533",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Jingling Sun, Ting Su, Junxin Li, Zhen Dong, Geguang Pu, Tao Xie, and Zhendong Su. 2021. Understanding and finding system setting-related defects in Android apps. In Proceedings of the 30th ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 204–215. https://doi.org/10.1145/3460319. 3464806",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Jingling Sun, Ting Su, Kai Liu, Chao Peng, Zhao Zhang, Geguang Pu, Tao Xie, and Zhendong Su. 2023. Characterizing and Finding System Setting-Related Defects in Android Apps. IEEE Trans. Software Eng. 49, 4 (2023), 2941–2963. https://doi.org/10.1109/TSE.2023.3236449",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Amaze Team. 2022. AmazeFileManager. Retrieved 2023-1 from https://github. com/TeamAmaze/AmazeFileManager.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Tiktok Team. 2021. Tiktok. Retrieved 2023-1 from https://www.tiktok.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Porfirio Tramontana, Domenico Amalfitano, Nicola Amatucci, and Anna Rita Fasolino. 2019. Automated functional testing of mobile applications: a systematic mapping study. Software Quality Journal 27, 1 (2019), 149–201. https://doi.org/ 10.1007/s11219-018-9418-6",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "David Travis. 2020. What user researchers ought to know about informed consent. Retrieved 2023-1 from https://userfocus.co.uk/articles/what_user_researchers_ ought_to_know_about_informed_consent.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "uiautomator2 Team. 2022. uiautomator2. Retrieved 2023-1 from https://github. com/openatx/uiautomator2.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Jue Wang, Yanyan Jiang, Ting Su, Shaohua Li, Chang Xu, Jian Lu, and Zhendong Su. 2022. Detecting non-crashing functional bugs in Android apps via deepstate differential analysis. In Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 434–446. https://doi.org/10.1145/3540250.3549170",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Jue Wang, Yanyan Jiang, Chang Xu, Chun Cao, Xiaoxing Ma, and Jian Lu. 2020. ComboDroid: Generating High-Quality Test Inputs for Android Apps via Use Case Combinations. In Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering (ICSE). 469–480. https://doi.org/10.1145/3377811.3380382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Wenyu Wang, Dengfeng Li, Wei Yang, Yurui Cao, Zhenwen Zhang, Yuetang Deng, and Tao Xie. 2018. An empirical study of Android test generation tools in industrial cases. In Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering (ASE). 738–748. https://doi.org/10.1145/3238147. 3240465",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "weditor Team. 2022. weditor. Retrieved 2023-1 from https://pypi.org/project/ weditor/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Wikipedia. 2022. Create, read, update and delete. Retrieved 2023-1 from https: //en.wikipedia.org/wiki/Create,_read,_update_and_delete.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Wikipedia Team. 2022. Wikipedia. Retrieved 2023-1 from https://github.com/ wikimedia/apps-android-wikipedia.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "WordPress Team. 2022. WordPress. Retrieved 2023-1 from https://github.com/ wordpress-mobile/WordPress-Android.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Yiheng Xiong, Mengqian Xu, Ting Su, Jingling Sun, Jue Wang, He Wen, Geguang Pu, Jifeng He, and Zhendong Su. 2023. An Empirical Study of Functional Bugs in Android Apps. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 1319–1331. https://doi.org/10.1145/ 3597926.3598138",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Razieh Nokhbeh Zaeem, Mukul R. Prasad, and Sarfraz Khurshid. 2014. Automated Generation of Oracles for Testing User-Interaction Features of Mobile Apps. In Proceedings of the International Conference on Software Testing, Verification and Validation (ICST). 183–192. https://doi.org/10.1109/ICST.2014.31",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Yixue Zhao, Justin Chen, Adriana Sejfia, Marcelo Schmitt Laser, Jie Zhang, Federica Sarro, Mark Harman, and Nenad Medvidovic. 2020. FrUITeR: a framework for evaluating UI test reuse. In Proceedings of the 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1190–1201. https://doi.org/10.1145/3368089.3409708",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Yu Zhao, Tingting Yu, Ting Su, Yang Liu, Wei Zheng, Jingzhi Zhang, and William GJ Halfond. 2019. Recdroid: automatically reproducing android application crashes from bug reports. In Software Engineering (ICSE). 128–139. https://doi.org/10.1109/ICSE.2019.00030 Received 2023-02-02; accepted 2023-07-27 1100",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 51,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. OOPSLA (2020), 211:1–211:30. https://doi.org/10.5281/zenodo.4032401",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 52,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI). USENIX Association, 667–682. https://doi.org/10.48550/ arXiv.2001.04174",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "prints the arXiv id of the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 51,52]",
+      "technique": "tlp",
+      "sentence": "Some work [ 51,52] uses metamorphic testing to find the CRUD errors in database management systems.",
+      "context_before": "as one UI test pattern to check its correctness. Mariani et al. [43] specify the CRUD operations of Web apps in Alloy specification language and build a tool Augusto to generate semantic tests with oracles. However, Augusto only tests one CRUD operation independently, and does not consider their combinations. Since a direct comparison is infeasible (as Augusto targets Web apps), Baseline A in our evaluation (see Section 4.4) can be viewed as a similar implementation of Augusto. Our approach which interleaves different DMFs of CRUD operations is shown to be much more effective than Baseline A.",
+      "context_after": "7 CONCLUSION We introduce a property-based fuzzing approach to effectively finding DMEs, which combines the idea of property-based testing and model-based properties. Given some type of app data, we interleave different DMFs and other possible events to generate diverse app states for validation. The realization of our idea, PBFDroid, has successfully discovered 30previously unknown bugs from 20popular Android apps. Among these bugs, 22are non-crashing DMEs, which cannot be found by the state-of-the-arts. So far, 19have been confirmed and 9fixed by the developers. The majority (20 out of 25) c",
+      "section": "6 RELATED WORK",
+      "page": 11,
+      "char_offset": 65190,
+      "cited_reference": {
+        "number": 51,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. OOPSLA (2020), 211:1–211:30. https://doi.org/10.5281/zenodo.4032401",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "PBFDroid applies property-based fuzzing to data manipulation errors in Android apps -- bugs in the create, read, update and delete operations that handle app-specific data, especially non-crashing logic ones. Given a type of app data it randomly interleaves the relevant operations with other events to explore diverse app states, characterising those functions as data-model-based properties and checking them against the UI. Across 20 real apps it found 30 previously unknown bugs.",
+    "narrative": "One citation in related work, noting that SQLancer's line of work uses metamorphic testing to find CRUD errors in database management systems -- the same class of error, in engines rather than apps.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3622819.json b/_data/papers/paper_doi_10_1145_3622819.json
new file mode 100644
index 0000000..a521f44
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3622819.json
@@ -0,0 +1,191 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:12:55Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3622819",
+    "title": "Towards Better Semantics Exploration for Browser Fuzzing",
+    "authors": [
+      "Chijin Zhou",
+      "Quan Zhang",
+      "Lihua Guo",
+      "Mingzhe Wang",
+      "Yu Jiang",
+      "Qing Liao",
+      "Zhiyong Wu",
+      "Shanshan Li",
+      "Bin Gu"
+    ],
+    "year": 2023,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3622819",
+    "arxiv_id": null,
+    "s2_paper_id": "c4a6a53c4c859e6d4c71ae122272a5e7bf91a163",
+    "url": "https://doi.org/10.1145/3622819",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3622819",
+      "retrieved_at": "2026-09-07T17:12:55Z",
+      "chars": 104549,
+      "content_sha256": "sha256:4eac6201c68f9631a856499405ce6b29dd551be1619f15e8590108c1649ebaef"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 28,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2194
+      },
+      {
+        "number": "2.1",
+        "title": "Grammar-BasedBrowser Fuzzing",
+        "start": 10380
+      },
+      {
+        "number": "2.2",
+        "title": "Challengesin Browser Fuzzing",
+        "start": 14712
+      },
+      {
+        "number": "2.3",
+        "title": "TowardsBe/t_ter SemanticsExploration",
+        "start": 22137
+      },
+      {
+        "number": "3.1",
+        "title": "GrammarExtraction",
+        "start": 24670
+      },
+      {
+        "number": "3.2",
+        "title": "SemanticsInference",
+        "start": 32522
+      },
+      {
+        "number": "3.3",
+        "title": "Input Generation",
+        "start": 46536
+      },
+      {
+        "number": "5.1",
+        "title": "BugFinding",
+        "start": 57038
+      },
+      {
+        "number": "5.2",
+        "title": "Comparisonto Other Browser Fuzzers",
+        "start": 66584
+      },
+      {
+        "number": "5.3",
+        "title": "StatisticsofGrammarExtraction",
+        "start": 77526
+      },
+      {
+        "number": "5.4",
+        "title": "Effectiveness ofGrammarExtraction and SemanticsInference",
+        "start": 78850
+      },
+      {
+        "number": "5.5",
+        "title": "OverheadofInput Generation",
+        "start": 81206
+      },
+      {
+        "number": "7.1",
+        "title": "Grammar-BasedFuzzing",
+        "start": 85448
+      },
+      {
+        "number": "7.2",
+        "title": "Browser Fuzzing",
+        "start": 87234
+      },
+      {
+        "number": "7.3",
+        "title": "Learn Program SemanticsforFuzzing",
+        "start": 89259
+      }
+    ]
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "author_year_citation",
+      "surface": "Rigger and Su 2020",
+      "technique": null,
+      "sentence": "2011], databases [ Rigger and Su 2020 ;Wang et al .",
+      "context_before": "seedinputs.Withauniformintermediaterepresentation,PolyGlot [Chen et al .2021] supports generating valid inputs for different testing targets. Zest [ Padhye etal.2019a]leveragescoveragefeedbacktoinferthetypeofvaluesthatarenotwell-definedin the grammar. These methods attempt to supplement semantic information that is not explicitly defined in the grammar, in an effort to increase the correctness of generated seeds. In addition to generic grammar-based fuzzers, there are also many custom fuzzers, enhanced with domain knowledge,forfuzzingspecificdomainsuchasC/C++compilers[ Livinskiietal .2020;Yangetal.",
+      "context_after": "2021b;Zhong et al .2020], and deep learning frameworks [ Liu et al.2023]. Different from the generic fuzzers and other domain’s fuzzers, SaGe only focuses ongenerating semantically correct browser inputs with browser domainknowledge. 7.2 Browser Fuzzing Among all browser fuzzers, Domato [ Google 2017 ] is recognized as the most successful fuzzers thatiswidelyadoptedintheindustry.Relyingonhand-writtengrammar,itgeneratesextensive Proc. ACM Program. Lang.,Vol. 7,No. OOPSLA2,Article 243. Publication date: October2023. TowardsBe/t_ter Semantics Explorationfor Browser Fuzzing 243:25 syntacticcorrecti",
+      "section": "7.1 Grammar-BasedFuzzing",
+      "page": 24,
+      "char_offset": 86949,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "SaGe improves browser fuzzing by replacing handwritten context-free grammars, which the authors say model browser semantics too crudely and yield inputs that cover only part of the semantics and are often semantically wrong. It extracts a rudimentary grammar from W3C standards and iteratively enriches it into a production-context sensitive grammar carrying semantic information. Over 24-hour campaigns on Chrome, Safari and Firefox it improved edge coverage by 6-278% and found 62 bugs.",
+    "narrative": "One author-year citation, in a list of domains where grammar-based fuzzing has been applied: compilers, databases, and deep learning frameworks.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3643779.json b/_data/papers/paper_doi_10_1145_3643779.json
new file mode 100644
index 0000000..5076c2a
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3643779.json
@@ -0,0 +1,161 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:12:10Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3643779",
+    "title": "DTD: Comprehensive and Scalable Testing for Debuggers",
+    "authors": [
+      "Hongyi Lu",
+      "Zhibo Liu",
+      "Shuai Wang",
+      "Fengwei Zhang"
+    ],
+    "year": 2024,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3643779",
+    "arxiv_id": null,
+    "s2_paper_id": "ebb7a876107f751507a0074cc8dbc28ded7115cf",
+    "url": "https://doi.org/10.1145/3643779",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3643779",
+      "retrieved_at": "2026-09-07T17:12:10Z",
+      "chars": 81717,
+      "content_sha256": "sha256:16d81359dc66b7639f66ae6c5467534c95fe6ecf16257ba365a0dbd505556ad1"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 22,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 4130
+      },
+      {
+        "number": "2",
+        "title": "PRELIMINARY",
+        "start": 10148
+      },
+      {
+        "number": "3.1",
+        "title": "Debugger Testing Requirements",
+        "start": 15603
+      },
+      {
+        "number": "3.2",
+        "title": "Review ofExisting Approaches",
+        "start": 18758
+      },
+      {
+        "number": "3.3",
+        "title": "DTDOverview",
+        "start": 23884
+      },
+      {
+        "number": "4.1",
+        "title": "C1: Trace Complexity",
+        "start": 33478
+      },
+      {
+        "number": "4.2",
+        "title": "C2: Randomness in Debugging CPrograms",
+        "start": 41487
+      },
+      {
+        "number": "6.1",
+        "title": "Case Studies",
+        "start": 54079
+      },
+      {
+        "number": "6.2",
+        "title": "Root Causes ofBugs",
+        "start": 59104
+      },
+      {
+        "number": "6.3",
+        "title": "Comparing DTDwithTheState-of-the-ArtWork—Debug2",
+        "start": 62884
+      }
+    ]
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "author_year_citation",
+      "surface": "Rigger and Su 2020",
+      "technique": null,
+      "sentence": "2019,2016;LiandSu2023; McKeeman 1998 ;Rigger and Su 2020 ;Theodoridiset al .",
+      "context_before": "in Fig. 1are very confusing. They can lead developers tobelievetheir code is wrongevenwhenit isnot. Worse still, debugger bugscan makeithardor impossibletoidentify actualsoftware bugs.Thispaperaimstouncover such debuggerdefects automatically and efficiently. We have noticed a line of research works [ Lehmann and Pradel 2018 ;Tolksdorf et al .2019] that focusontestingdebuggers.Furthermore,similarworks[ Assaianteetal .2023;DiLunaetal .2021;Li etal.2020]alsodetectcompiler bugsthatfailtocorrectlyattachdebuggingsymbols inexecutables. Overall,mostexistingworksfollowadifferentialtesting(DT)[ Chenetal .",
+      "context_after": "2022] paradigm, such that two debuggers aretestedbycheckingthe consistency oftheirretrievedprogramstates.Inconsistentlyretrieved program states indicate a bug inone of thetwodebuggers. Existingapproachesintestingdebuggers mainlyfocusonJavaScript(JS)debuggers,ascenario thatisgenerallydistinctfromtestingCdebuggers(Sec. 3.2).Despitethesuccessofthoseapproaches, webelievetheyfallshortintermsof scalability whenconsideringCdebuggers.JavaScriptprograms testedintheseworksareonlyofafewhundredlines,butreal-worldCprogramscaneasilyhave overtenthousandorevenmillionsoflinesofassemblycode.Inotherwords,anaivec",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 6234,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "DTD tests interactive debuggers, which can misretrieve or misinterpret program state and so mislead developers, and for which no scalable comprehensive correctness measurement existed. The authors propose testing criteria covering comprehensiveness of debug information and scalability, then build a differential framework that runs the same C executable through two mainstream debuggers and treats discrepancies as bugs. DTD found 13 bugs in the LLVM toolchain and 5 in the GNU toolchain.",
+    "narrative": "One author-year citation, among the differential-testing work the paper places itself alongside.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3643781.json b/_data/papers/paper_doi_10_1145_3643781.json
new file mode 100644
index 0000000..951ec6f
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3643781.json
@@ -0,0 +1,478 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:27Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3643781",
+    "title": "Metamorphic Testing of Secure Multi-party Computation (MPC) Compilers",
+    "authors": [
+      "Yichen Li",
+      "Dongwei Xiao",
+      "Zhibo Liu",
+      "Qi Pang",
+      "Shuai Wang"
+    ],
+    "year": 2024,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3643781",
+    "arxiv_id": null,
+    "s2_paper_id": "1fccf0ed3dbbaeedaf7e587b93bbe993ae638ed7",
+    "url": "https://doi.org/10.1145/3643781",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3643781",
+      "retrieved_at": "2026-09-09T01:31:27Z",
+      "chars": 88052,
+      "content_sha256": "sha256:c14124985f57e4276865617c8ae884f605d4308dd4c35a88e04922167cb8bccc"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 22,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 4054
+      },
+      {
+        "number": "2",
+        "title": "PRELIMINARY",
+        "start": 9502
+      },
+      {
+        "number": "2.2",
+        "title": "MPCProgram Compilation and Execution",
+        "start": 13476
+      },
+      {
+        "number": "3",
+        "title": "METHODOLOGY",
+        "start": 19119
+      },
+      {
+        "number": "3.1",
+        "title": "SeedGeneration",
+        "start": 22359
+      },
+      {
+        "number": "3.3",
+        "title": "MR /u1D451/u1D450/u1D453: Data/Control Flow-BasedMR",
+        "start": 25913
+      },
+      {
+        "number": "4",
+        "title": "IMPLEMENTATION",
+        "start": 42440
+      },
+      {
+        "number": "6.1",
+        "title": "Effectiveness and Efficiencyof MT-MPC",
+        "start": 49374
+      },
+      {
+        "number": "6.2",
+        "title": "Bugs forCompilation Failures",
+        "start": 56136
+      },
+      {
+        "number": "6.3",
+        "title": "Bugs forLogicErrors",
+        "start": 59242
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Donald Beaver. 1991. Foundations of secure interactive computing. In Annual International Cryptology Conference. Springer, 377–391. https://doi.org/10.1007/3-540-46766-1_31",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "LukasBernhard,TobiasScharnowski,MoritzSchloegel,TimBlazytko,andThorstenHolz.2022.JIT-Picking:Differential FuzzingofJavaScriptEngines.In Proceedingsofthe2022ACMSIGSACConferenceonComputerandCommunications Security (CCS ’22). Association for Computing Machinery, New York, NY, USA, 351–364. https://doi.org/10.1145/ 3548606.3560624",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Zvika Brakerski, Craig Gentry, and Vinod Vaikuntanathan. 2012. (Leveled) Fully Homomorphic Encryption without Bootstrapping. In Proceedings of the 3rd Innovations in Theoretical Computer Science Conference (Cambridge, Massachusetts) (ITCS ’12). Association for Computing Machinery, New York, NY, USA, 309–325. https: //doi.org/10.1145/2090236.2090262",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Stefanos Chaliasos, Thodoris Sotiropoulos, Diomidis Spinellis, Arthur Gervais, Benjamin Livshits, and Dimitris Mitropoulos.2022. FindingTyping Compiler Bugs.In Proceedings of the43rd ACMSIGPLAN International Conference onProgrammingLanguageDesign and Implementation (PLDI2022) .Associationfor ComputingMachinery,NewYork, NY, USA,183–198. https://doi.org/10.1145/3519939.3523427",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Wing Kwong Chan, Shing Chi Cheung, and Karl RPH Leung. 2007. A metamorphic testing approach for online testing ofservice-oriented softwareapplications. International Journal of Web Services Research(IJWSR) 4,2(2007), 61–81.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "NishanthChandran,DivyaGupta,AseemRastogi,RahulSharma,andShardulTripathi.2019. EzPC:Programmable and Efficient Secure Two-Party Computation for Machine Learning. In Privacy (EuroS&P). 496–511. https://doi.org/10.1109/EuroSP.2019.00043",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Junjie Chen, Jibesh Patra, Michael Pradel, Yingfei Xiong, Hongyu Zhang, Dan Hao, and Lu Zhang. 2020. A Survey of CompilerTesting. Comput.Surveys 53, 1(2020), 4:1–4:36. https://doi.org/10.1145/3363562",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Tsong YChen, Shing C Cheung, and ShiuMing Yiu. 2020. Metamorphictesting: a newapproach for generatingnext test cases. arXivpreprintarXiv:2002.12543 (2020).https://doi.org/10.48550/arXiv.2002.12543",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "TsongYuehChen,JoshuaWKHo,HuaiLiu,andXiaoyuanXie.2009. Aninnovativeapproachfortestingbioinformatics programs using metamorphic testing. BMC bioinformatics 10, 1 (2009), 1–12. https://doi.org/10.1186/1471-2105-10-24",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Tsong Yueh Chen, Fei-Ching Kuo, Huai Liu, Pak-Lok Poon, Dave Towey, T. H. Tse, and Zhi Quan Zhou. 2018. Metamorphic Testing: A Review of Challenges and Opportunities. ACM Comput. Surv. 51, 1, Article 4 (jan 2018), 27pages. https://doi.org/10.1145/3143561",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Tsong Yueh Chen, Hing Leung, and IK Mak. 2004. Adaptive random testing. In Annual Asian Computing Science Conference. Springer, 320–329. https://doi.org/10.1007/978-3-540-30502-6_23",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jung Hee Cheon, Andrey Kim, Miran Kim, and Yongsoo Song. 2017. Homomorphic encryption for arithmetic of approximatenumbers.In AdvancesinCryptology–ASIACRYPT2017:23rdInternationalConferenceonthe Theoryand Applications of Cryptology and Information Security, Hong Kong, China, December 3-7, 2017, Proceedings, Part I 23. Springer, 409–437. https://doi.org/10.1007/978-3-319-70694-8_15",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Sha/f_iulAzamChowdhury,SohilLalShrestha,TaylorT.Johnson,andChristophCsallner.2020. SLEMI:Equivalence modulo Input (EMI) Based Mutation of CPS Models for Finding Compiler Bugs in Simulink. In Proceedings of the ACM/IEEE42ndInternationalConferenceonSoftwareEngineering (ICSE’20) .AssociationforComputingMachinery, NewYork, NY, USA,335–346. https://doi.org/10.1145/3377811.3380381",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Ivan Damgård, Valerio Pastro, Nigel Smart, and Sarah Zakarias. 2012. Multiparty computation from somewhat homomorphicencryption.In AnnualCryptologyConference .Springer,643–662. https://doi.org/10.1007/978-3-64232009-5_38 Proc. ACM Softw. Eng., Vol. 1,No. FSE, Article 55. Publication date: July2024. MetamorphicTestingofSecure Multi-party Computation(MPC)Compilers 55:21",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "data61. 2016. tpmpc_tutorial.mpc. https://github.com/data61/MP-SPDZ/blob/master/Programs/Source/tpmpc_tutorial. mpc.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Daniel Demmler, Thomas Schneider, and Michael Zohner. 2015. ABY-A framework for efficient mixed-protocol secure two-party computation..In NDSS.https://doi.org/10.14722/ndss.2015.23113",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "KyleDewey,JaredRoesch,andBenHardekopf.2015.FuzzingtheRusttypecheckerusingCLP(T).In 201530thIEEE/ACM InternationalConferenceonAutomatedSoftwareEngineering(ASE) .IEEE,482–493. https://doi.org/10.1109/ASE.2015.65",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Samuel Groß, Simon Koch, Lukas Bernhard, Thorsten Holz, and Martin Johns. 2023. FUZZILLI: Fuzzing for JavaScript JIT Compiler Vulnerabilities. In Proceedings 2023 Network and Distributed System Security Symposium. Internet Society, San Diego, CA, USA. https://doi.org/10.14722/ndss.2023.24290",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Marcella Hastings, Brett Hemenway, Daniel Noble, and Steve Zdancewic. 2019. Sok: General purpose compilers for secure multi-party computation. In. IEEE, 1220–1237. https: //doi.org/10.1109/SP.2019.00028",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "MarcelKeller. 2020. MP-SPDZ: AVersatile Framework forMulti-PartyComputation. In Proceedingsof the 2020ACM SIGSACConferenceonComputer andCommunicationsSecurity .https://doi.org/10.1145/3372297.3417872",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "NishantKumar,MayankRathee,NishanthChandran,DivyaGupta,AseemRastogi,andRahulSharma.2020. Crypt/f_low: Secure tensor/f_low inference. In. IEEE, 336–353. https://doi.org/10. 1109/SP40000.2020.00092",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Vu Le,Mehrdad Afshari,andZhendong Su.2014. Compilervalidationviaequivalence moduloinputs. ACMSigplan Notices49, 6(2014), 216–226. https://doi.org/10.1145/2666356.2594334",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Vu Le,Chengnian Sun, and Zhendong Su. 2015. Finding deep compiler bugsvia guided stochastic programmutation. InProceedingsofthe2015ACMSIGPLANInternationalConferenceonObject-OrientedProgramming,Systems,Languages, andApplications. 386–399. https://doi.org/10.1145/2858965.2814319",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "CongLi,YanyanJiang,ChangXu,andZhendongSu.2023. ValidatingJITCompilersviaCompilationSpaceExploration. InProceedingsofthe29thSymposiumonOperatingSystemsPrinciples(SOSP’23) .AssociationforComputingMachinery, NewYork, NY, USA,66–79. https://doi.org/10.1145/3600006.3613140",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Yichen Li. 2023. MT-MPC. https://github.com/winnylyc/MT-MPC/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "ChangLiu,XiaoShaunWang,KartikNayak,YanHuang,andElaineShi.2015. Oblivm:Aprogrammingframeworkfor securecomputation.In 2015IEEESymposiumonSecurityandPrivacy .IEEE,359–376. https://doi.org/10.1109/SP.2015.29",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Zhibo Liu and Shuai Wang. 2020. How far we have come: Testing decompilation correctness of C decompilers. In Proceedings of the 29th ACMSIGSOFT International Symposium on Software Testing and Analysis. 475–487. https: //doi.org/10.1145/3395363.3397370",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Vsevolod Livinskii, Dmitry Babokin, and John Regehr. 2020. Random testing for C and C++ compilers with YARPGen. Proceedingsof the ACM onProgrammingLanguages 4,OOPSLA (2020), 1–25. https://doi.org/10.1145/3428264",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Pingchuan Ma, Shuai Wang, and Jin Liu. 2020. Metamorphic Testing and Certi/f_ied Mitigation of Fairness Violations in NLP Models. In Proceedings of the Twenty-Ninth International Joint Conference on Arti/f_icial Intelligence, IJCAI 2020, ChristianBessiere(Ed.). ijcai.org, 458–465. https://doi.org/10.24963/IJCAI.2020/64",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Microsoft.2019. EzPC (EasySecureMulti-partyComputation). https://github.com/mpc-msri/EzPC.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "PaymanMohasselandPeterRindal.2018.ABY3:Amixedprotocolframeworkformachinelearning.In Proceedingsofthe 2018ACMSIGSACconferenceoncomputerandcommunicationssecurity .35–52.https://doi.org/10.1145/3243734.3243760",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Alex Ozdemir, Fraser Brown, and Riad S Wahby. 2022. CirC: Compiler infrastructure for proof systems, software veri/f_ication, and more. In. IEEE, 2248–2266. https://doi.org/10.1109/ SP46214.2022.9833782",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Qi Pang, Yuanyuan Yuan, and Shuai Wang. 2024. MPCDiff: Testing and Repairing MPC-Hardened Deep Learning Models...In NDSS.https://doi.org/10.14722/ndss.2024.23380",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Deevashwer Rathee, Anwesh Bhattacharya, Divya Gupta, Rahul Sharma, and Dawn Song. 2023. Secure Floating-Point Training.In 32ndUSENIXSecuritySymposium(USENIXSecurity23) .USENIXAssociation,Anaheim, CA,6329–6346. https://www.usenix.org/conference/usenixsecurity23/presentation/rathee",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "DeevashwerRathee, MayankRathee, RahulKranti KiranGoli, DivyaGupta,RahulSharma,NishanthChandran,and Aseem Rastogi. 2021. Sirnn: A math library for secure rnn inference. In (SP). IEEE, 1003–1020. https://doi.org/10.1109/SP40001.2021.00086",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "DeevashwerRathee,MayankRathee,NishantKumar,NishanthChandran,DivyaGupta,AseemRastogi,andRahul Sharma.2020. Crypt/f_low2:Practical2-partysecureinference.In Proceedingsofthe2020ACMSIGSACConferenceon Computer andCommunicationsSecurity. 325–342. https://doi.org/10.1145/3372297.3417274",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engineconstruction.In Proceedingsofthe28thACMJointMeetingonEuropeanSoftwareEngineeringConferenceand Symposiumonthe Foundationsof Software Engineering. 1140–1152. https://doi.org/10.1145/3368089.3409710 Proc. ACM Softw. Eng., Vol. 1,No. FSE, Article 55. Publication date: July2024. 55:22",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM onProgrammingLanguages 4,OOPSLA (2020), 211:1–211:30. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "Andrei Sabelfeld and Andrew C Myers. 2003. Language-based information-/f_low security. IEEE Journal on selected areasincommunications 21, 1(2003), 5–19. https://doi.org/10.1109/JSAC.2002.806121",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Edward J Schwartz, Thanassis Avgerinos, and David Brumley. 2010. All you ever wanted to know about dynamic taint analysis and forward symbolic execution (but might have been afraid to ask). In privacy.https://doi.org/10.1109/SP.2010.26",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Sergio Segura, Gordon Fraser, Ana B Sanchez, and Antonio Ruiz-Cortés. 2016. A survey on metamorphic testing. IEEE Transactionsonsoftware engineering 42, 9(2016), 805–824. https://doi.org/10.1109/TSE.2016.2532875",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "AdiShamir.1979. Howtoshareasecret. Commun.ACM 22,11(1979),612–613. https://doi.org/10.1145/359168.359176",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "ChengnianSun,VuLe,andZhendongSu.2016. Findingcompilerbugsvialivecodemutation.In Proceedingsofthe2016 ACMSIGPLANinternational conference on object-oriented programming,systems, languages, andapplications .849–863. https://doi.org/10.1145/2983990.2984038",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Shuai Wang and Zhendong Su. 2020. Metamorphic Object Insertion for Testing Object Detection Systems. In 35th IEEE/ACM International Conference on Automated Software Engineering, ASE 2020, Melbourne, Australia, September 21-25,2020. IEEE, 1053–1065. https://doi.org/10.1145/3324884.3416584",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Xiao Wang, Alex J. Malozemoff, and Jonathan Katz. 2016. EMP-toolkit: Efficient MultiParty computation toolkit. https://github.com/emp-toolkit.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Xiao Wang, Samuel Ranellucci, and Jonathan Katz. 2017. Authenticated garbling and efficient maliciously secure two-partycomputation.In Proceedingsofthe2017ACMSIGSACconferenceoncomputerandcommunicationssecurity. 21–37.https://doi.org/10.1145/3133956.3134053",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Dongwei Xiao, Zhibo Liu, and ShuaiWang. 2023. Metamorphic shader fusionfortesting graphics shadercompilers. In. IEEE, 2400–2412. https://doi.org/10.1109/ ICSE48619.2023.00201",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Dongwei Xiao,Zhibo Liu, YuanyuanYuan,QiPang, andShuai Wang. 2022. Metamorphic testing of deep learning compilers. Proceedings of the ACM on Measurement and Analysis of Computing Systems 6, 1 (2022), 1–28. https: //doi.org/10.1145/3508035",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "KangYang,ChenkaiWeng,XiaoLan,JiangZhang,andXiaoWang.2020. Ferret:FastextensionforcorrelatedOTwith small communication. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 1607–1626. https://doi.org/10.1145/3372297.3417276",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In Proceedings of the 32nd ACMSIGPLAN conference on Programming language design and implementation. 283–294. https://doi.org/10.1145/1993316.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Andrew C Yao. 1982. Protocols for secure computations. In 23rd annual symposium on foundations of computer science (sfcs1982). IEEE, 160–164. https://doi.org/10.1109/SFCS.1982.38",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Andrew Chi-Chih Yao. 1986. How to generate and exchange secrets. In 27th annual symposium on foundations of computerscience(Sfcs 1986). IEEE, 162–167. https://doi.org/10.1109/SFCS.1986.25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Yuanyuan Yuan, Qi Pang, and Shuai Wang. 2022. Unveiling Hidden DNN Defects with Decision-Based Metamorphic Testing. In 37thIEEE/ACMInternationalConferenceon AutomatedSoftwareEngineering, ASE2022,Rochester,MI,USA, October 10-14,2022. ACM, 113:1–113:13. https://doi.org/10.1145/3551349.3561157",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "YuanyuanYuan,Shuai Wang, Mingyue Jiang, and TsongYuehChen. 2021. Perception Matters: Detecting Perception Failures of VQA ModelsUsing MetamorphicTesting. In IEEEConference on Computer Vision and Pattern Recognition, CVPR 2021, virtual, June 19-25, 2021. Computer Vision Foundation / IEEE, 16908–16917. https://doi.org/10.1109/ CVPR46437.2021.01663",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "AndreasZeller.1999. Yesterday,myprogramworked.Today,itdoesnot.Why? ACMSIGSOFTSoftwareengineering notes24, 6(1999), 253–267. https://doi.org/10.1145/318774.318946",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Google Project Zero. 2017. domato: DOMfuzzer. https://github.com/googleprojectzero/domato. Received2023-09-29;accepted2024-01-23 Proc. ACM Softw. Eng., Vol. 1,No. FSE, Article 55. Publication date: July2024.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engineconstruction.In Proceedingsofthe28thACMJointMeetingonEuropeanSoftwareEngineeringConferenceand Symposiumonthe Foundationsof Software Engineering. 1140–1152. https://doi.org/10.1145/3368089.3409710 Proc. ACM Softw. Eng., Vol. 1,No. FSE, Article 55. Publication date: July2024. 55:22 YichenLi,Dongwei Xiao, Zhibo Liu, Qi Pang, andShuaiWang",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM onProgrammingLanguages 4,OOPSLA (2020), 211:1–211:30. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[37,38]",
+      "technique": "norec",
+      "sentence": "MThasbeenappliedtoalargevarietyofsoftwaresystems,includingdatabase[37,38],AImodels[ 29,44,53,54],andCyberPhysicalSystems[ 13].",
+      "context_before": "ng topic in the community [ 7]. Existing research works have been focusing ontestingCcompilers[ 22,23,28,43,50]andhaveachievedgreatsuccessin/f_indingbugsinthe optimizationsof C compilers.Some other works aim to test the typingsystems of compilers [ 4,17]. TestingJust-in-Time(JIT)compilersisalsoreceivinggreaterattention[ 2,18,24].Tothebestof our knowledge, there is no prior work on testing MPC compilers, which are becoming increasingly importantin enhancing and protectingtheprivacy of userdata. MetamorphicTesting. Metamorphictesting(MT)worksbyassertinginvariantpropertiesonthe softwareundertest.",
+      "context_after": "MThasgainedincreasing attention in compiler testing, helping to uncover thousands of faults in industrial-strength compilers[5,9,10,22,23,43,47,48].OurcarefullycraftedMRsareintrinsictoMPCandarenotadequately addressed by conventional MT methods. On the other hand, our MRs are not constrained to a speci/f_ic MPCDSL,thusallowing usto test various MPC compilers. Testing MPC-DNN Models. We have also noticed a contemporary work over MPC-protected deep neuralnetwork(DNN)models[ 33].Differentfrom MT-MPC,Pangetal .[33]leveragesgradientbasedmethodstodetectpredictionerrorsof MPC-hardened DNNmodels,whilewefo",
+      "section": "6.3 Bugs forLogicErrors",
+      "page": 19,
+      "char_offset": 72637,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engineconstruction.In Proceedingsofthe28thACMJointMeetingonEuropeanSoftwareEngineeringConferenceand Symposiumonthe Foundationsof Software Engineering. 1140–1152. https://doi.org/10.1145/",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "MT-MPC applies metamorphic testing to compilers for secure multi-party computation, which translate high-level descriptions into low-level executables through several intermediate representations and many optimisations. The authors note no systematic understanding of these compilers' correctness existed. They propose three metamorphic relations tailored to MPC programs and check whether the compiler produces semantically equivalent executables. Across three industrial compilers they found 4,772 error-triggering inputs and traced thirteen bugs.",
+    "narrative": "One citation, in a list of the software systems metamorphic testing has been applied to: databases, AI models and cyber-physical systems.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3650212_3680311.json b/_data/papers/paper_doi_10_1145_3650212_3680311.json
new file mode 100644
index 0000000..fda708a
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3650212_3680311.json
@@ -0,0 +1,742 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:26:30Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3650212.3680311",
+    "title": "Testing Graph Database Systems with Graph-State Persistence Oracle",
+    "authors": [
+      "Shuang Liu",
+      "Junhao Lan",
+      "Xiaoning Du",
+      "Jiyuan Li",
+      "Wei Lu",
+      "Jiajun Jiang",
+      "Xiaoyong Du"
+    ],
+    "year": 2024,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3650212.3680311",
+    "arxiv_id": null,
+    "s2_paper_id": "ada7405e20ac861d632055bcb871ee4d57afc710",
+    "url": "https://doi.org/10.1145/3650212.3680311",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3650212.3680311",
+      "retrieved_at": "2026-09-08T15:26:30Z",
+      "chars": 70962,
+      "content_sha256": "sha256:9f0ec7acfb451549cf53f94569af9d267fd92cf25be46a194cb66321d1722a25"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3406
+      },
+      {
+        "number": "2",
+        "title": "Preliminary",
+        "start": 11174
+      },
+      {
+        "number": "2.1",
+        "title": "LabeledProperty GraphandCypher",
+        "start": 11188
+      },
+      {
+        "number": "2.2",
+        "title": "GraphIsomorphism",
+        "start": 14017
+      },
+      {
+        "number": "4.1",
+        "title": "QueryGeneration",
+        "start": 26589
+      },
+      {
+        "number": "4.2",
+        "title": "Mutation Rules",
+        "start": 28069
+      },
+      {
+        "number": "4.3",
+        "title": "Oracle",
+        "start": 31579
+      },
+      {
+        "number": "5.1",
+        "title": "EvaluationSetup",
+        "start": 39551
+      },
+      {
+        "number": "5.2",
+        "title": "RQ1: Ability on Detecting Unknown Bugs",
+        "start": 41958
+      },
+      {
+        "number": "5.4",
+        "title": "RQ3: Comparisonwith Baselines",
+        "start": 44514
+      },
+      {
+        "number": "5.4.1",
+        "title": "DetectedUniqueBugs. Table7showsthecomparisonresults",
+        "start": 45533
+      },
+      {
+        "number": "5.5",
+        "title": "Threatsto Validity",
+        "start": 52676
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2007.StructuredQueryLanguage. SpringerBerlinHeidelberg,Berlin,Heidelberg, 111–212. https://doi.org/10.1007/978-3-540-48399-1_4",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2024.03. AgensGraphOfficialWebsite. https://bitnine.net/agensgraph/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2024.03.CypherforgremlinPlugin. https://github.com/opencypher/cypher-forgremlin/tree/master/tinkerpop/cypher-gremlin-server-client.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2024.03.The DBRanking Website. https://db-engines.com/en/ranking/graph+ dbms/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2024.03.GraphDatabasemarketshare. https://www.marketsandmarkets.com/ Market-Reports/graph-database-market-126230231.html..",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2024.03. Graph isomorphism problem for labeled graphs, computer science. https://cs.stackexchange.com/questions/28714/graph-isomorphismproblem-for-labeled-graphs.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2024.03.Limitations of RedisGraph .https://github.com/RedisGraph/RedisGraph/ blob/master/docs/docs/known_limitations.md.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2024.03. MemGraph OfficialWebsite. https://memgraph.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2024.03.Neo4j Official Documents: Clauses in the Cypher Query Language .https: //neo4j.com/docs/cypher-manual/current/clauses/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2024.03. Neo4j Official Website. https://neo4j.com/product/neo4j-graphdatabase/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2024.03. RedisGraph Official Website. https://docs.redis.com/latest/stack/ deprecated-features/graph/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "AsimAnsari,SkanderEssegaier,andRajeevKohli.2000.Internetrecommendation systems. https://doi.org/10.1509/jmkr.37.3.363.18779",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Marcelo Arenas, Claudio Gutiérrez, and Juan F Sequeda. 2021. Querying in the age of graph databases and knowledge graphs. In Proceedings of the 2021 International Conference on Management of Data. 2821–2828. https://doi.org/10. 1145/3448016.3457545",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "L.P.Cordella,P.Foggia,C.Sansone,andM.Vento.2004. A(sub)graphisomorphismalgorithmformatchinglargegraphs. IEEETransactionsonPatternAnalysis andMachineIntelligence 26,10(2004),1367–1372. https://doi.org/10.1109/TPAMI. 2004.75",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "WenqiFan,YaoMa,QingLi,YuanHe,EricZhao,JiliangTang,andDaweiYin. 2019. Graph neural networks for social recommendation. In The world wide web conference. 417–426. https://doi.org/10.1145/3308558.3313488",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "PasqualeFoggia,CarloSansone,MarioVento,etal .2001. Aperformancecomparison of /f_ive algorithms for graph isomorphism. In Proceedings of the 3rd IAPRTC-15 Workshop on Graph-basedRepresentations inPatternRecognition. Citeseer, 188–199. https://src.acm.org/binaries/content/assets/src/2009/sara-voss.pdf 676 ISSTA ’24, September16–20, 2024,Vienna, Austria Shuang Liu, Junhao Lan,XiaoningDu,Jiyuan Li, We",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Martin Schuster, Petra Selmer, et al .2018. Formal semantics of the language cypher. arXiv preprint arXiv:1802.09984 (2018).https://dblp.org/rec/journals/corr/abs-1802-09984.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and AndrésTaylor.2018. Cypher:Anevolvingquerylanguageforpropertygraphs. InProceedings of the 2018 international conference on management of data. 1433– 1445.https://doi.org/10.1145/3183713.3190657",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Shu-Ming Hsieh, Chiun-Chieh Hsu, and Li-Fu Hsu. 2006. Efficient method to perform isomorphism testing of labeled graphs. In Computational Science and Its Applications-ICCSA2006:InternationalConference,Glasgow,UK,May8-11,2006, Proceedings,Part V6. Springer, 422–431. https://doi.org/10.1007/11751649_46",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting bugs in Cypher graph database engines. InProceedingsofthe32ndACMSIGSOFT InternationalSymposiumonSoftware Testingand Analysis. 163–174. https://doi.org/10.1145/3597926.3598046",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 531–542. https://doi.org/10.1145/3597503.3623307",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "Matteo Kamm. 2022. Testing GraphDatabases using Predicate Partitioning. Master’s thesis. ETHZurich. https://doi.org/10.1145/3597926.3598044",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "MAO Linfan.2016. Labeledgraph—Amathematical element. In/f_inite Study.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "ShuangLiu,JunhaoLan,XiaoningDu,JiyuanLi,WeiLu,JiajunJiang,andXiaoyong Du. 2024. Testing Graph Database Systems with Graph-State Persistence Oracle. Zenodo. https://doi.org/10.5281/zenodo.12670528",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "QiuyangMang,AoyangFang,BoxiYu,HanfeiChen,andPinjiaHe.2024. Testing Graph Database Systems via Equivalent Query Rewriting. (2024). https://doi. org/10.1145/3597503.3639200",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Neo4j. 2024.03. Recommendation Graph .https://github.com/neo4j-graphexamples/recommendations/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Neo4jofficial.2023.12. CypherQueryLanguage .https://neo4j.com/developer/ cypher/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Paul Resnick and Hal R Varian. 1997. Recommender systems. Commun. ACM 40, 3 (1997), 56–58. https://doi.org/10.1145/245108.245121",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and SymposiumontheFoundationsofSoftwareEngineering .1140–1152. https://doi. org/10.1145/3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via querypartitioning. ProceedingsoftheACMonProgrammingLanguages 4,OOPSLA(2020), 1–30. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation(OSDI 20). 667–682. https://doi.org/10.5555/3488766.3488804",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 32,
+      "text": "CarlosRRiveroandHasanMJamil.2017. Efficientandscalablelabeledsubgraph matchingusing SGMatch. KnowledgeandInformationSystems 51,1(2017),61–87. https://doi.org/10.1007/s10115-016-0968-2",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Ian Robinson, Jim Webber, and Emil Eifrem. 2015. Graph databases: new opportunities for connected data. \" O’Reilly Media, Inc.\". https://doi.org/10.5555/2846367",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "MarkoARodriguez.2015. Thegremlingraphtraversalmachineandlanguage (invited talk). In Proceedings of the 15th Symposium on Database Programming Languages. 1–10.https://doi.org/10.1145/2815072.2815073",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "MarkoA.Rodriguez.2015. TheGremlinGraphTraversalMachineandLanguage (InvitedTalk).In Proceedingsofthe15thSymposiumonDatabaseProgramming Languages (Pittsburgh,PA,USA) (DBPL2015) .AssociationforComputingMachinery, NewYork, NY, USA,1–10. https://doi.org/10.1145/2815072.2815073",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "ChristosTjortjis.2023. GraphDatabases:ApplicationsonSocial MediaAnalytics and SmartCities. CRC Press. https://doi.org/10.1201/9781003183532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "J. R. Ullmann. 1976. An Algorithm for Subgraph Isomorphism. J. ACM23, 1 (jan 1976),31–42. https://doi.org/10.1145/321921.321925",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "RongjingXiang,JenniferNeville,andMonicaRogati.2010. Modelingrelationshipstrengthinonlinesocialnetworks.In Proceedingsofthe19thinternational conference on World wide web. 981–990. https://doi.org/10.1145/1772690.1772790",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Chen Yang, Junjie Chen, Xingyu Fan, Jiajun Jiang, and Jun Sun. 2023. Silent Compiler Bug De-duplication via Three-Dimensional Analysis. In Proceedings of the32ndACMSIGSOFTInternationalSymposiumonSoftwareTestingandAnalysis (Seattle, WA, USA) (ISSTA 2023). Association for Computing Machinery, New York, NY, USA,677–689. https://doi.org/10.1145/3597926.3598087",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "JunZengy,XiangWang,JiahaoLiu,YinfangChen,ZhenkaiLiang,Tat-SengChua, and Zheng Leong Chua. 2022. Shadewatcher: Recommendation-guided cyber threat analysis using system audit records.In and Privacy (SP). IEEE,489–506. https://doi.org/10.1109/SP46214.2022.9833669",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "YingyingZheng,WenshengDou,YichengWang,ZhengQin,LeiTang,YuGao, DongWang,WeiWang,andJunWei.2022. FindingbugsinGremlin-basedgraph databasesystemsviarandomizeddifferentialtesting.In Proceedingsofthe31st ACMSIGSOFTInternationalSymposiumonSoftwareTestingandAnalysis .302–313. https://doi.org/10.1145/3533767.3534409",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "ZeyangZhuang,PenghuiLi,PingchuanMa,WeiMeng,andShuaiWang.2023. Testing Graph Database Systems via Graph-Aware Metamorphic Relations. Proceedings of the VLDB Endowment 17, 4 (2023), 836–848. https://doi.org/10.14778/ 3636218.3636236 Received 2024-04-12; accepted 2024-07-03 677",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 21,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 531–542. https://doi.org/10.1145/3597503.3623307",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and SymposiumontheFoundationsofSoftwareEngineering .1140–1152. https://doi. org/10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via querypartitioning. ProceedingsoftheACMonProgrammingLanguages 4,OOPSLA(2020), 1–30. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp"
+      ]
+    },
+    {
+      "number": 31,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation(OSDI 20). 667–682. https://doi.org/10.5555/3488766.3488804",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 20–22,41]",
+      "technique": null,
+      "sentence": "Recognizing this challenge, the software testing community has devoted increasing attention to uncovering bugsin DBMS, with a particular focus on GDBMSs [ 20–22,41].",
+      "context_before": "000organizations,includingindustryleaderssuchasAdobe,eBayandUBS[ 36] Similartootherdatabasesmanagementsystems(DBMSs),GDBM-Ssacceptuserqueriesasinputandreturnthecorrespondingquery results. These user queries convey requests to store, manipulate, or retrieve data from the databases. To ensure better reliability and efficiency,DBMSsdemandsoundandoptimizedimplementationsof transactionmanagement,indexingandstoragemanagement,query optimization,andconcurrencycontrol.Thecombinationofthese factors, coupled with the inherent complexity of database systems, renders their implementation susceptible to bugs.",
+      "context_after": "In order to reveal the deeply buried logic bugs in GDBMSs, one ofthebiggestchallengesliesindevisingpowerfultestoracles.As a countermeasure, differential testing [ 20,41] and metamorphic testing [21,22,25,42] play an crucial role in existing works on GDBMStesting.Theinsightbehinddifferentialtestingisthatdifferent GDBMS implementations should return the same results when dealingwithidenticaluserrequests.However,differentGDBMSs may adopt different graph query languages, such as Cypher [ 18] andGremlin[ 34],whichlimitstheapplicationofdifferentialtesting approachestoaspeci/f_icgroupofimplementationsthat",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4870,
+      "cited_reference": {
+        "number": 21,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 53",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[21,22,25,42]",
+      "technique": null,
+      "sentence": "As a countermeasure, differential testing [ 20,41] and metamorphic testing [21,22,25,42] play an crucial role in existing works on GDBMStesting.",
+      "context_before": "nsure better reliability and efficiency,DBMSsdemandsoundandoptimizedimplementationsof transactionmanagement,indexingandstoragemanagement,query optimization,andconcurrencycontrol.Thecombinationofthese factors, coupled with the inherent complexity of database systems, renders their implementation susceptible to bugs. Recognizing this challenge, the software testing community has devoted increasing attention to uncovering bugsin DBMS, with a particular focus on GDBMSs [ 20–22,41]. In order to reveal the deeply buried logic bugs in GDBMSs, one ofthebiggestchallengesliesindevisingpowerfultestoracles.",
+      "context_after": "Theinsightbehinddifferentialtestingisthatdifferent GDBMS implementations should return the same results when dealingwithidenticaluserrequests.However,differentGDBMSs may adopt different graph query languages, such as Cypher [ 18] andGremlin[ 34],whichlimitstheapplicationofdifferentialtesting approachestoaspeci/f_icgroupofimplementationsthatsupportthe samequery language.GDSmith[ 20]and Grand[ 41]are thestateof-the-art differential testing approaches for GDBMSs, respectively supporting Cypher and Gremlin. Another factor that in/f_luences theeffectivenessofdifferentialtestingistheabsenceofstandard speci/",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5156,
+      "cited_reference": {
+        "number": 21,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 53",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 21]",
+      "technique": null,
+      "sentence": "Failing to carefullyverifywhethertwoimplementationsadheretothesame speci/f_ication alsoresult infalse alarms [ 21].",
+      "context_before": "dealingwithidenticaluserrequests.However,differentGDBMSs may adopt different graph query languages, such as Cypher [ 18] andGremlin[ 34],whichlimitstheapplicationofdifferentialtesting approachestoaspeci/f_icgroupofimplementationsthatsupportthe samequery language.GDSmith[ 20]and Grand[ 41]are thestateof-the-art differential testing approaches for GDBMSs, respectively supporting Cypher and Gremlin. Another factor that in/f_luences theeffectivenessofdifferentialtestingistheabsenceofstandard speci/f_icationsforgraphquerylanguages,leadingtodifferentimplementation choices even for the same query language.",
+      "context_after": "In contrast, metamorphic testing approaches offer more precise oracles for testing GDBMSs, and a body of works surge in this line, including GDBMeter [ 22], GraphGenie [ 21], and GRev [ 25]. They have particularly focused on the data retrieval functionalityof GDBMSsanddesignedmetamorphicrelationsthatcapture whether and how the query results are affected when particular transformations are made to the matching conditions. For example, GRevproposesrepresentingthematchingpatternswithAbstract Syntax Graph (ASG) and developing an algorithm to extract equivalent matching conditions from the graph, whi",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6005,
+      "cited_reference": {
+        "number": 21,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 53",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 21]",
+      "technique": null,
+      "sentence": "In contrast, metamorphic testing approaches offer more precise oracles for testing GDBMSs, and a body of works surge in this line, including GDBMeter [ 22], GraphGenie [ 21], and GRev [ 25].",
+      "context_before": "18] andGremlin[ 34],whichlimitstheapplicationofdifferentialtesting approachestoaspeci/f_icgroupofimplementationsthatsupportthe samequery language.GDSmith[ 20]and Grand[ 41]are thestateof-the-art differential testing approaches for GDBMSs, respectively supporting Cypher and Gremlin. Another factor that in/f_luences theeffectivenessofdifferentialtestingistheabsenceofstandard speci/f_icationsforgraphquerylanguages,leadingtodifferentimplementation choices even for the same query language. Failing to carefullyverifywhethertwoimplementationsadheretothesame speci/f_ication alsoresult infalse alarms [ 21].",
+      "context_after": "They have particularly focused on the data retrieval functionalityof GDBMSsanddesignedmetamorphicrelationsthatcapture whether and how the query results are affected when particular transformations are made to the matching conditions. For example, GRevproposesrepresentingthematchingpatternswithAbstract Syntax Graph (ASG) and developing an algorithm to extract equivalent matching conditions from the graph, which describes exactly theoriginalpatternbutwithadifferentsetofconditions,thereby rewriting the queries. If the query results differ, a bug is caught. Additionally,GAMERA[ 42]proposesmetamorphic",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6121,
+      "cited_reference": {
+        "number": 21,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 53",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 21,25]",
+      "technique": null,
+      "sentence": "Note that different from existing approaches on testing GDBMS [ 21,25], where only identical result setsareregardedascorrect,ourmutationrulesmaygeneratequery pairswhichresultindifferentresultsets(e.",
+      "context_before": "Isomorphic. Query result correctness. Since the queries with writing clauses mayalsohavereturnresults,weneedalsocheckthecorrectness ofthequeryresultsaspartofouroracletoreducepotentialfalse negatives,i.e.,thecaseswheregraph-stateiscorrectlypreserved yetthereturnresultsareincorrect.ThisstepsimplycheckwhetherTable 3:Informationon The Tested GDBMS. GDBMS Rank GithubStars Init Release LoC Neo4j 1 12.4k 2007 1,304K RedisGraph-2.0k 2018 1,298K MemGraph 8 2.1k 2017 268K AgensGraph 30 1.3k 2016 1,510K the returned result sets satisfy the relations ( =,⊂,⊃) introduced by the corresponding mutation rule.",
+      "context_after": "g.,rules18-20inTable 2), making our oracle able to detect bugs that are not detectable by oracles ofexisting approaches. 5 Evaluations Weimplementourprototype GraspDB withover11Knon-comment lines of Java code. GraspDB uses Neo4j Java Driver 4.1.1 to connectand interactwith Neo4jand Memgraph,JRedisGraph2.5.1to connectandinteractwithRedisGraphandAgensGraphJavaDriver 1.4.2 to connect and interact with AgensGraph. All evaluations are conductedonacomputerwithInteli5-8400CPU,16GBofmemory and Windows 11 OS. We aim to address the following research questionsinour evaluation. •RQ1:CanGraspDB detectunkn",
+      "section": "4.3 Oracle",
+      "page": 7,
+      "char_offset": 38579,
+      "cited_reference": {
+        "number": 21,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 53",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[21]",
+      "technique": null,
+      "sentence": "We compare GraspDB with GDSmith [ 20] and GraphGenie[21].",
+      "context_before": "cessing capabilities.Memgraph[ 8]isanin-memorygraphdatabasedesignedfor real-time data analytics. It supports openCypher and is compatible withNeo4j.AgensGraph[ 2]isbasedonthepowerfulPostgreSQL RDBMS,andisoptimizedforhandlingcomplexconnectedgraph data and provides plenty of powerful database features essential to the enterprise database environment. We test the Neo4j CommunityEditionfromv5.6.0tov5.12.0,RedisGraphfromv2.10.9to v2.12.10,MemGraphCommunityEditionfromv2.7.0tov2.11.0and AgensGraphCommunityEdition2.13.1.Alltheseversionsarelatest duringour testingperiodfrom 2023.4to 2023.10. Baselines.",
+      "context_after": "GDSmithisthelatestdifferentialtestingapproachon graph databases that is capable of generating syntax and semantic correct Cypher queries. GraphGenie is an metamorphic testing 672 ISSTA ’24, September16–20, 2024,Vienna, Austria Shuang Liu, Junhao Lan,XiaoningDu,Jiyuan Li, We Lu,Jiajun Jiang,andXiaoyong Du Table 4:BugsDetected by GraspDB. GDBMS Detected Con/f_irmed Fixed Duplicate Writing-related GSO-related Neo4j 35 33 30 2 18 2 RedisGraph 15 12 6 1 6 4 MemGraph 17 13 7 0 7 4 AgensGraph 10 0 0 0 0 0 Total 77 58 43 3 31 10 Table 5:TypesofThe Con/f_irmed Bugs GDBMS Logic Error Crash Inconsistency",
+      "section": "5.1 EvaluationSetup",
+      "page": 7,
+      "char_offset": 40720,
+      "cited_reference": {
+        "number": 21,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 53",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 21]",
+      "technique": null,
+      "sentence": "GraphGeniehasshownstate-of-the-artperformanceindetectingbugs in GDBMS [ 21]and outperformsother metamorphic testingapproachessuchas GDBMeter[ 22].",
+      "context_before": "d Duplicate Writing-related GSO-related Neo4j 35 33 30 2 18 2 RedisGraph 15 12 6 1 6 4 MemGraph 17 13 7 0 7 4 AgensGraph 10 0 0 0 0 0 Total 77 58 43 3 31 10 Table 5:TypesofThe Con/f_irmed Bugs GDBMS Logic Error Crash Inconsistency Neo4j 2 29 0 2 RedisGraph 4 2 6 0 MemGraph 5 1 7 0 Total 11 32 13 2 Table 6:DistinctBugsDetected by Variants of GraspDB Method Neo4j RedisGraph MemGraph GraspDB 8 7 10 GraspDB-GSO 7 6 7 GraspDB-AWC 7 6 9 GraspDB-MWC 6 4 6 GraspDB-MRC 6 5 8 approach,whichconductsgraphpatterntransformationsguided by injective and surjective relations to generate comparable query pairs.",
+      "context_after": "Therefore,wechoose GDSmithandGraphGenieas baselinesfor comparison. 5.2 RQ1: Ability on Detecting Unknown Bugs Table4showthebugsthat GraspDB detectedonthefourtesting subjects. GraspDB detected 77 bugs during the testing period of around6months.Amongthosebugs,58arecon/f_irmedand43have been/f_ixed.Amongthecon/f_irmedbugs,31isdetectedbywriting clausesinthegeneratedqueries,10aredetectedbyourgraph-state persistent oracle andand13 viaour mutation rules. Table5showsthetypesofthecon/f_irmedbugs.Wecanobserved thatthemajorityofbugscauseerrorsorsystemcrashes,indicating thatthewriting-relatedfunctionalitie",
+      "section": "5.1 EvaluationSetup",
+      "page": 8,
+      "char_offset": 41747,
+      "cited_reference": {
+        "number": 21,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 53",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 21]",
+      "technique": null,
+      "sentence": "This is a common limitation for differential testing approaches andsimilar /f_indings is also reported by GraphGenie [ 21].",
+      "context_before": "null, null (a)ExamplefalsealarmbyGDSmithduetodifferentsemantic implementations on GDBMSs 1OPTIONAL MATCH (c)-[:PP]-() WHERE false RETURN count (c) 2−−Return Result: 0 3OPTIONAL MATCH (c)-[:PP]-() WHERE false RETURN count (*) 4−−Return Result: 1 (b) Example false alarm by GraphGenie due to nonequivalent semantics of Count(c) and Count(*) 1CREATE (),(); 2MATCH()UNWIND [1,1]asa WITH DISTINCT *RETURN a; 3−−Return Result: 1 4CREATE (),(); 5MATCHpath=() UNWIND [1,1]asa WITH DISTINCT *RETURN a; 6−−Return Result: 1 ,1 (c) False alarmof GraspDB due to addinga variable Figure 5:False alarm examples GDBMSs.",
+      "context_after": "Thislimitationleadstoheavymanualeffortsinanalyzingthelarge amountbugreportstoruleoutfalsepositives.AsshowninTable 7, 87 false alarms have been identi/f_ied from the 250 bug reports of GDsmith, making more than 1/3 of the reported bugs false positives.Allofthefalsealarmsarecausedbydifferentimplementation choices by different GDBMSs, Among which 70 are due to unde/f_inedbehaviorinCypherdocumentation,and17areduetodifferent numericalrepresentationformats.Onesuchfalsealarmexample isshowninFigure 5(a),whereNeo4jandMemGraphhavedifferent implementation choices for the return clause which returns aggregate",
+      "section": "5.4.1 DetectedUniqueBugs. Table7showsthecomparisonresults",
+      "page": 9,
+      "char_offset": 50308,
+      "cited_reference": {
+        "number": 21,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 53",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "GDBMeter [ 22] is the /f_irst metamorphic testing approach which applied the TLP [ 30] from relational DBMS to graph DBMS and found 40 unique, previously unknown bugs.",
+      "context_before": "theGremlinlanguage.Fortestcasegeneration,Grandusesamodelbased approach to generate valid Gremlin queries. GDsmith [ 20] is anotherdifferentialtestingapproachtotestGDBMSswhichsupporttheCypherquerylanguage,anotherpopularquerylanguage for GDBMS.Fortestcasegeneration,GDsmithusesskeletongenerationandcompletiontogeneratesemanticallyvalidCypherqueries. Themajordrawbackofdifferentialtestingisthatbugscannotbe detectedwhendifferentDBMSssufferfromthesameones.Furthermore,differentialtestingapproachessufferfromhighfalsealarm ratesduetodifferentimplementationchoicesofGDBMSs,ashas been discussedinshowninSection 5.4.",
+      "context_after": "GraphGenie [ 21] proposes injective and surjective Graph Query Transformation (GQT) to detect logic bugs. It leverages graph properties to generate follow-upqueriesbymutatinggraphquerypatternsanddetected 25unknownbugs.GRev[ 25]adaptsEquivalentQueryRewriting (EQR) to GDBMS queries by Random Walking on Abstract SyntaxGraph (ASG), an abstractiontheyproposedtorepresentquery paths.Gamera[ 42]developsthreeclassesofgraph-awaremetamorphic relations, i.e., elementary MRs, compound MRsand dynamic MRs, which directly applies to labeled property graphs, for testing GDBMS. Different from GraspDB, non of ex",
+      "section": "5.5 Threatsto Validity",
+      "page": 11,
+      "char_offset": 59081,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 21]",
+      "technique": null,
+      "sentence": "GraphGenie [ 21] proposes injective and surjective Graph Query Transformation (GQT) to detect logic bugs.",
+      "context_before": "whichsupporttheCypherquerylanguage,anotherpopularquerylanguage for GDBMS.Fortestcasegeneration,GDsmithusesskeletongenerationandcompletiontogeneratesemanticallyvalidCypherqueries. Themajordrawbackofdifferentialtestingisthatbugscannotbe detectedwhendifferentDBMSssufferfromthesameones.Furthermore,differentialtestingapproachessufferfromhighfalsealarm ratesduetodifferentimplementationchoicesofGDBMSs,ashas been discussedinshowninSection 5.4. GDBMeter [ 22] is the /f_irst metamorphic testing approach which applied the TLP [ 30] from relational DBMS to graph DBMS and found 40 unique, previously unknown bugs.",
+      "context_after": "It leverages graph properties to generate follow-upqueriesbymutatinggraphquerypatternsanddetected 25unknownbugs.GRev[ 25]adaptsEquivalentQueryRewriting (EQR) to GDBMS queries by Random Walking on Abstract SyntaxGraph (ASG), an abstractiontheyproposedtorepresentquery paths.Gamera[ 42]developsthreeclassesofgraph-awaremetamorphic relations, i.e., elementary MRs, compound MRsand dynamic MRs, which directly applies to labeled property graphs, for testing GDBMS. Different from GraspDB, non of existing metamorphic testingapproachesfocusondetectingbugscausedbywritingoperations, and the oracles in thei",
+      "section": "5.5 Threatsto Validity",
+      "page": 11,
+      "char_offset": 59249,
+      "cited_reference": {
+        "number": 21,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang,andManuel Rigger.2023. Detecting LogicBugsin GraphDatabaseManagementSystemsviaInjectiveandSurjectiveGraphQueryTransformation.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEngineering(ICSE) .IEEE Computer Society, 53",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "Ternary Logic Partitioning (TLP) [30], a metamorphic testing approach /f_irst proposed for testing RDBMSs,has alsobeenappliedtotest GDBMS[ 22].",
+      "context_before": "uivalentQueryRewriting (EQR) to GDBMS queries by Random Walking on Abstract SyntaxGraph (ASG), an abstractiontheyproposedtorepresentquery paths.Gamera[ 42]developsthreeclassesofgraph-awaremetamorphic relations, i.e., elementary MRs, compound MRsand dynamic MRs, which directly applies to labeled property graphs, for testing GDBMS. Different from GraspDB, non of existing metamorphic testingapproachesfocusondetectingbugscausedbywritingoperations, and the oracles in their approach are not able to detect labeled property graph changes, thus unable to capture those bugs. Metamorphic Testing of RDBMS.",
+      "context_after": "Thistool has been used to /f_ind 175 bugs in widely deployed RDBMSs. (PQS) isageneralandhighly-effectiveapproachto/f_indingbugsinDBMS. The core idea of PQS [ 31] is to automatically synthesize queries which guarantees to fetch a speci/f_ic, randomly selected row, called the pivot row. If the DBMS fails to fetch the pivot row, it likely causes a bug in the RDBMS. Non-optimizing Reference Engine Construction (NoREC) [ 29] is another widely-known metamorphic testing approach to test RDBMS. It compares the execution results ofagivenoptimizedquerywiththenon-optimizedversionthatthey rewrittenbased on",
+      "section": "5.5 Threatsto Validity",
+      "page": 11,
+      "char_offset": 60085,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "(PQS) isageneralandhighly-effectiveapproachto/f_indingbugsinDBMS.",
+      "context_before": "i.e., elementary MRs, compound MRsand dynamic MRs, which directly applies to labeled property graphs, for testing GDBMS. Different from GraspDB, non of existing metamorphic testingapproachesfocusondetectingbugscausedbywritingoperations, and the oracles in their approach are not able to detect labeled property graph changes, thus unable to capture those bugs. Metamorphic Testing of RDBMS. Ternary Logic Partitioning (TLP) [30], a metamorphic testing approach /f_irst proposed for testing RDBMSs,has alsobeenappliedtotest GDBMS[ 22].Thistool has been used to /f_ind 175 bugs in widely deployed RDBMSs.",
+      "context_after": "The core idea of PQS [ 31] is to automatically synthesize queries which guarantees to fetch a speci/f_ic, randomly selected row, called the pivot row. If the DBMS fails to fetch the pivot row, it likely causes a bug in the RDBMS. Non-optimizing Reference Engine Construction (NoREC) [ 29] is another widely-known metamorphic testing approach to test RDBMS. It compares the execution results ofagivenoptimizedquerywiththenon-optimizedversionthatthey rewrittenbased ontheoriginalquery, todetectoptimizationbugs in DBMS.The metamorphic testing approaches on RDBMSs share similar purposes and scenarios a",
+      "section": "5.5 Threatsto Validity",
+      "page": 11,
+      "char_offset": 60295,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "The core idea of PQS [ 31] is to automatically synthesize queries which guarantees to fetch a speci/f_ic, randomly selected row, called the pivot row.",
+      "context_before": "y applies to labeled property graphs, for testing GDBMS. Different from GraspDB, non of existing metamorphic testingapproachesfocusondetectingbugscausedbywritingoperations, and the oracles in their approach are not able to detect labeled property graph changes, thus unable to capture those bugs. Metamorphic Testing of RDBMS. Ternary Logic Partitioning (TLP) [30], a metamorphic testing approach /f_irst proposed for testing RDBMSs,has alsobeenappliedtotest GDBMS[ 22].Thistool has been used to /f_ind 175 bugs in widely deployed RDBMSs. (PQS) isageneralandhighly-effectiveapproachto/f_indingbugsinDBMS.",
+      "context_after": "If the DBMS fails to fetch the pivot row, it likely causes a bug in the RDBMS. Non-optimizing Reference Engine Construction (NoREC) [ 29] is another widely-known metamorphic testing approach to test RDBMS. It compares the execution results ofagivenoptimizedquerywiththenon-optimizedversionthatthey rewrittenbased ontheoriginalquery, todetectoptimizationbugs in DBMS.The metamorphic testing approaches on RDBMSs share similar purposes and scenarios as GDBMSs, and can potentially be applicable to GDBMSs. 8 Conclusion Inthispaper,wepresent GraspDB ,the/f_irstmetamorphictesting approachspeci/f_icallyd",
+      "section": "5.5 Threatsto Validity",
+      "page": 11,
+      "char_offset": 60360,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Non-optimizing Reference Engine Construction (NoREC) [ 29] is another widely-known metamorphic testing approach to test RDBMS.",
+      "context_before": "abeled property graph changes, thus unable to capture those bugs. Metamorphic Testing of RDBMS. Ternary Logic Partitioning (TLP) [30], a metamorphic testing approach /f_irst proposed for testing RDBMSs,has alsobeenappliedtotest GDBMS[ 22].Thistool has been used to /f_ind 175 bugs in widely deployed RDBMSs. (PQS) isageneralandhighly-effectiveapproachto/f_indingbugsinDBMS. The core idea of PQS [ 31] is to automatically synthesize queries which guarantees to fetch a speci/f_ic, randomly selected row, called the pivot row. If the DBMS fails to fetch the pivot row, it likely causes a bug in the RDBMS.",
+      "context_after": "It compares the execution results ofagivenoptimizedquerywiththenon-optimizedversionthatthey rewrittenbased ontheoriginalquery, todetectoptimizationbugs in DBMS.The metamorphic testing approaches on RDBMSs share similar purposes and scenarios as GDBMSs, and can potentially be applicable to GDBMSs. 8 Conclusion Inthispaper,wepresent GraspDB ,the/f_irstmetamorphictesting approachspeci/f_icallydesignedtoidentifybugsrelatedtowritingoperationsingraphdatabasesystems.Inparticular,wede/f_inetheconcepts of Labeled Property Graph Isomorphism (LPG-Isomorphism) and Labeled Property Subgraph Isomorphism (LP",
+      "section": "5.5 Threatsto Validity",
+      "page": 11,
+      "char_offset": 60590,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:58:40Z",
+    "is_model_written": true,
+    "summary": "GraspDB tests graph database systems with a graph-state persistence oracle: mutation rules transform a query in ways whose effect on the graph state is known, so the expected relationship between the two results is known even when the result sets differ. That distinguishes it from the existing metamorphic graph testers, whose oracles treat only identical result sets as correct and are therefore confined to transformations that preserve results exactly.",
+    "narrative": "TLP reaches this paper as the technique GDBMeter carried into graph databases -- the paper describes GDBMeter as the first metamorphic approach to apply TLP from relational DBMSs to graph DBMSs, finding 40 previously unknown bugs, and notes elsewhere that TLP was first proposed for relational systems and has since been applied to graph ones. PQS and NoREC are described alongside as the relational approaches, with PQS called general and highly effective. GraspDB's own baselines are GDsmith and GraphGenie rather than GDBMeter, on the stated grounds that GraphGenie has shown state-of-the-art performance and outperforms GDBMeter.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "motivation",
+      "M6": "baseline",
+      "M7": "background",
+      "M8": "background",
+      "M9": "definition",
+      "M10": "definition",
+      "M11": "definition",
+      "M12": "state_of_the_art",
+      "M13": "definition",
+      "M14": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Nothing describes reuse of SQLancer's code; GraspDB implements its own mutation rules and oracle."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The graph-state persistence oracle admits transformations whose result sets differ, which M5 gives as the departure from the existing approaches rather than a development of TLP."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M6 states the comparison is with GDsmith and GraphGenie, and M7 explains GDBMeter was left out because GraphGenie outperforms it. Neither TLP nor a TLP-based tool is run against GraspDB."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M12",
+            "sentence": "(PQS) isageneralandhighly-effectiveapproachto/f_indingbugsinDBMS.",
+            "section": "5.5 Threatsto Validity",
+            "page": 11
+          }
+        ],
+        "reasoning": "M12 calls PQS a general and highly-effective approach to finding bugs in DBMSs. The state-of-the-art claim in M7 attaches to GraphGenie, not to a SQLancer technique."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3650212_3680317.json b/_data/papers/paper_doi_10_1145_3650212_3680317.json
new file mode 100644
index 0000000..a836339
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3650212_3680317.json
@@ -0,0 +1,629 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:32:42Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3650212.3680317",
+    "title": "SQLess: Dialect-Agnostic SQL Query Simplification",
+    "authors": [
+      "Li Lin",
+      "Zongyin Hao",
+      "Chengpeng Wang",
+      "Zhuangda Wang",
+      "Rongxin Wu",
+      "Gang Fan"
+    ],
+    "year": 2024,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3650212.3680317",
+    "arxiv_id": null,
+    "s2_paper_id": "deb987f2a988b7585e7423ec2d48b5cd20f2cb6a",
+    "url": "https://doi.org/10.1145/3650212.3680317",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3650212.3680317",
+      "retrieved_at": "2026-09-08T14:32:42Z",
+      "chars": 75308,
+      "content_sha256": "sha256:329427759e974e6ca4077f0fa30c05ff7d367b583c1d55b767bddf6dd88e925b"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 12,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3068
+      },
+      {
+        "number": "2",
+        "title": "Motivation",
+        "start": 10808
+      },
+      {
+        "number": "2.1",
+        "title": "Importance of SQL Query Simplification",
+        "start": 11091
+      },
+      {
+        "number": "2.2",
+        "title": "Challenges of SQL Query Simplification",
+        "start": 13831
+      },
+      {
+        "number": "2.3",
+        "title": "Key Idea",
+        "start": 16948
+      },
+      {
+        "number": "3",
+        "title": "Approach",
+        "start": 21529
+      },
+      {
+        "number": "3.2",
+        "title": "Adaptive Parsing",
+        "start": 22791
+      },
+      {
+        "number": "3.2.2",
+        "title": "Dialect Builder. Dialect builder is a crucial step in adaptive",
+        "start": 25881
+      },
+      {
+        "number": "3.3",
+        "title": "Semantics-Sensitive Query Trimming",
+        "start": 28590
+      },
+      {
+        "number": "3.3.1",
+        "title": "Semantic Analysis. An effective query trimmer maintains",
+        "start": 29263
+      },
+      {
+        "number": "4",
+        "title": "Implementation",
+        "start": 38486
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 40226
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Setting",
+        "start": 40883
+      },
+      {
+        "number": "5.1.2",
+        "title": "Metrics. We use two metrics to assess SQLess: simplification",
+        "start": 41991
+      },
+      {
+        "number": "5.2",
+        "title": "Effectiveness and Efficiency",
+        "start": 44974
+      },
+      {
+        "number": "5.3",
+        "title": "Comparison with Existing Approaches",
+        "start": 50010
+      },
+      {
+        "number": "5.4",
+        "title": "Ablation Study",
+        "start": 53790
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 62004
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 62825
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 66679
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Alfred V. Aho, Monica S. Lam, Ravi Sethi, and Jeffrey D. Ullman. 2006. Compilers: Principles, Techniques, and Tools (2nd ed.). Pearson Education, Chapter 4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "ANSI. 2023. The SQL Standard-ISO/IEC 9075:2023 (ANSI X3.135). https:// blog.ansi.org/sql-standard-iso-iec-9075-2023-ansi-x3-135/. [Online; accessed 12-Dec-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Andrea Arcuri and Lionel C. Briand. 2011. A practical guide for using statistical tests to assess randomized algorithms in software engineering. In Proceedings of the 33rd International Conference on Software Engineering. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "J Asplund. 2011. Data Science revealed: Data-driven glimpse into the burgeoning new field.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M Tamer Özsu. 2007. QAGen: generating query-aware test databases. In Proceedings of the 2007 ACMSIGMOD international conference on Management of data. 341–352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free DBMS fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023. Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis. In 2023 USENIX Annual Technical Conference. 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "IECISO and NIEC. 2017. ISO/IEC. IEEE International Standard-Systems and software engineering–Vocabulary (2017), 1–541.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful fuzzing for database management systems with complex and valid SQL query generation. In Proceedings of USENIX Security Symposium. 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "JSQlParser. 2023. JSqlParser. https://jsqlparser.github.io/. [Online; accessed 23-Nov-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. Apollo: Automatic detection and diagnosis of performance regressions in database systems. Proceedings of the VLDB Endowment 13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Yong Lei and James H Andrews. 2005. Minimization of randomized unit test cases. In 16th IEEE International Symposium on Software Reliability Engineering. 10–pp.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of {DBMS}with Coverage-based Guidance. In 31st USENIX Security Symposium. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "libpgquery. 2023. libpg_query: PostgreSQL Parser Library. https://github.com/ pganalyze/libpg_query. [Online; accessed 23-Nov-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Mariadb. 2023. Mariadb Database. https://mariadb.org/. [Online; accessed Dec-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Ghassan Misherghi and Zhendong Su. 2006. HDD: hierarchical delta debugging. InProceedings of the 28th international conference on Software engineering. 142– 151.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. 2008. Generating targeted queries for database testing. In Proceedings of the 2008 ACMSIGMOD international conference on Management of data. 499–510.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "MySQL. 2023. MySQL Bug Reporting Guidelines. https://bugs.mysql.com/. [Online; accessed Dec-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "MySQL. 2023. MySQL Database. https://www.mysql.com/. [Online; accessed Dec-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "OceanBase. 2023. OceanBase Database. https://www.oceanbase.com/. [Online; accessed Dec-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Alessandro Orso, Shrinivas Joshi, Martin Burger, and Andreas Zeller. 2006. Isolating relevant component interactions with JINSI. In Proceedings of the 2006 international workshop on Dynamic systems analysis. 3–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Terence Parr. 2013. The Definitive ANTLR 4 Reference. Pragmatic Bookshelf, Chapter 9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "pincap. 2023. Parser-A MySQL Compatible SQL Parser. https://github.com/ pingcap/tidb/tree/master/parser. [Online; accessed January-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "pincap. 2023. TiDB Parser: The SQL Parser for TiDB. https://github.com/pingcap/ tidb/tree/master/parser. [Online; accessed Nov-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "PostgreSQL. 2023. PostgreSQL Bug Reporting Guidelines. https://www.postgresql. org/list/pgsql-bugs/. [Online; accessed Dec-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "PostgreSQL. 2023. PostgreSQL Database. https://www.postgresql.org/. [Online; accessed Dec-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "pquery. 2023. PQuery: Multithreaded SQL Tester / Reducer. https://github.com/ Percona-QA/pquery. [Online; accessed 23-Nov-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "John Regehr, Yang Chen, Pascal Cuoq, Eric Eide, Chucky Ellison, and Xuejun Yang. 2012. Test-case reduction for C compiler bugs. In Proceedings of the 33rd ACMSIGPLAN conference on Programming Language Design and Implementation. 335–346.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Manuel Rigger. 2023. SQLancer. https://github.com/sqlancer/sqlancer. [Online; accessed 23-Nov-2023].",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "Donald R Slutz. 1998. Massive stochastic testing of SQL. In VLDB, Vol. 98. Citeseer, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "SQLite. 2023. SQLite Database. https://sqlite.org/. [Online; accessed Dec-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "sqlparser. 2023. phpMyAdmin SQL Parser. https://github.com/phpmyadmin/sqlparser. [Online; accessed Nov-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "SQLParser. 2023. SQLParser. https://www.sqlparser.com/. [Online; accessed Nov-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "SQLSmith. 2023. SQLSmith. https://github.com/anse1/sqlsmith. [Online; accessed June-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "ANTLR Team. 2023. ANTLR (ANother Tool for Language Recognition). https: //www.ANTLR.org/. [Online; accessed 23-Nov-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "ANTLR Team. 2023. ANTLR DefaultErrorStrategy Class Documentation. https:// www.ANTLR.org/api/Java/org/ANTLR/v4/runtime/DefaultErrorStrategy.html. [Online; accessed 23-Nov-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "SQLess Team. 2023. SQLess. https://github.com/SQLess/AdaptSQLess.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "SQLess Team. 2023. SQLessBugReports. https://github.com/SQLess/SQLess_ Bugreports.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "SQLess Team. 2023. SQLessExample. https://github.com/SQLess/Examples/blob/ main/README.md",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "SQLite Team. 2023. SQLite Bug Reporting Guidelines. https://www.chiark. greenend.org.uk/~sgtatham/bugs.html. [Online; accessed Dec-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "The ANTLR Organization. 2023. ANTLR Grammars Repository. https://github. com/ANTLR/grammars-v4. [Online; accessed 23-Nov-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "TiDB. 2023. TiDB Database. https://www.pingcap.com/tidb/. [Online; accessed Dec-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Jiajie Wang, Puhan Zhang, Lei Zhang, Haowen Zhu, and Xiaojun Ye. 2013. A model-based fuzzing approach for DBMS. In 2013 8th International Conference on Communications and Networking in China. 426–431.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry practice of coverage-guided enterprise-level DBMS fuzzing. In Software Engineering in Practice. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "wiki. 2023. Comparison of Relational Database Management Systems. https://en.wikipedia.org/wiki/Comparison_of_relational_database_ management_systems. [Online; accessed Sep-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "wikibooks. 2023. SQL Dialects Reference/Introduction. https://en.wikibooks.org/ wiki/SQL_Dialects_Reference/Introduction. [Online; accessed Nov-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Michal Zalewski. 2021. American Fuzzy Lop (2.52b). http://lcamtuf.coredump.cx/ afl. Accessed: June 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Andreas Zeller. 1999. Yesterday, my program worked. Today, it does not. Why? ACMSIGSOFT Software engineering notes 24, 6 (1999), 253–267.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Andreas Zeller and Ralf Hildebrandt. 2002. Simplifying and isolating failureinducing input. IEEE Transactions on Software Engineering 28, 2 (2002), 183–200.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 955–970. Received 2024-04-12; accepted 2024-07-03 754",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 29,
+      "text": "Manuel Rigger. 2023. SQLancer. https://github.com/sqlancer/sqlancer. [Online; accessed 23-Nov-2023].",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 7,11,29,31]",
+      "technique": null,
+      "sentence": "Existing studies [ 7,11,29,31] on database bug detection merely integrate SQL query simplification into their workflows in a simplistic manner.",
+      "context_before": "inpointing the root cause of the issue. Figure 1 gives a real-world example of bug report in TiDB. This bug was quickly confirmed by the developer within one day, but it had been lasted for more than two months without any fix due to the lengthy test case. Until a simplified SQL query was provided, this bug then was quickly fixed within two days. This example indicates that SQL query simplification is critical for developers to diagnose the bugs in DBMSs. While many approaches concentrate on bug detection in DBMSs, most do not adequately consider the importance of the SQL query simplification.",
+      "context_after": "They employ the principles of Delta debugging [ 50] to minimize SQL queries on structured inputs such as Abstract Syntax Tree (AST). However, a common oversight in these studies is the neglect of interdependencies among different structural elements. Such dependency-insensive deletion approach without semantic checks can render the simplified SQL query nonexecutable. For instance, in a query like \"SELECT c1, c2 from t 743 ISSTA ’24, September 16–20, 2024, Vienna, Austria Li Lin, Zongyin Hao, Chengpeng Wang, Zhuangda Wang, Rongxin Wu, and Gang Fan Table 1: Comparison of SQL query simplificatio",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 4562,
+      "cited_reference": {
+        "number": 29,
+        "text": "Manuel Rigger. 2023. SQLancer. https://github.com/sqlancer/sqlancer. [Online; accessed 23-Nov-2023].",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For instance, in a query like \"SELECT c1, c2 from t 743 ISSTA ’24, September 16–20, 2024, Vienna, Austria Li Lin, Zongyin Hao, Chengpeng Wang, Zhuangda Wang, Rongxin Wu, and Gang Fan Table 1: Comparison of SQL query simplification tools Simplification Strategy Correctness Guarantee Extensibility Bug type Tool Expression Clause Subquery Column Syntactic Semantic Adaption to Crash Logical Simplification Simplification Simplification Simplification Correctness Correctness other DBMSs Bugs Bugs SQLancer [29] ✓ ✓××××× ✓ ✓ PINOLO [7] ✓ ✓×××××× ✓ RAGS [31] ✓ ✓××××× ✓× APOLLO [11] ✓ ✓ ✓ ✓ ✓×× ✓× IssueReport:#40015User:qaqcatzReportedDate:Dec 18, 2022Description:Bug Report: Logical bug.",
+      "context_before": "DBMSs, most do not adequately consider the importance of the SQL query simplification. Existing studies [ 7,11,29,31] on database bug detection merely integrate SQL query simplification into their workflows in a simplistic manner. They employ the principles of Delta debugging [ 50] to minimize SQL queries on structured inputs such as Abstract Syntax Tree (AST). However, a common oversight in these studies is the neglect of interdependencies among different structural elements. Such dependency-insensive deletion approach without semantic checks can render the simplified SQL query nonexecutable.",
+      "context_after": "Value changed after relaxing ON condition [v5.0.2, v6.4.0]Minimal reproduce step (Required):drop table if exists t; CREATE TABLE t (c1 FLOAT UNSIGNED); INSERT INTO t VALUES (47),(28.1237); SELECT (DATE_ADD(BIN(c1), INTERVAL 1 DAY_HOUR)) AS f1 FROM (SELECT 1 FROM t) AS t1 JOIN (SELECT c1 FROM t) AS t2 ON 1; --sql1 SELECT (DATE_ADD(BIN(c1), INTERVAL 1 DAY_HOUR)) AS f1 FROM (SELECT 1 FROM t) AS t1 JOIN (SELECT c1 FROM t) AS t2 ON (0 AND c1 != 1) IS FALSE; --sql2SeaRise commented on Feb 28SeaRise: “We need minor case.SELECT DATE_ADD(c1, INTERVAL 1 DAY_HOUR) from test.test; --sql1”Fix error handli",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 5076,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 7,29,31]",
+      "technique": null,
+      "sentence": "Expression simplification and clause simplification [ 7,29,31] are relatively easy to implement, but the scope for simplification is limited.",
+      "context_before": "ifying SQL queries. 2.2 Challenges of SQL Query Simplification Although SQL query simplification plays a crucial role in the detection of DBMS bugs, to date, there has yet to be a systematic study dedicated exclusively to the exploration of SQL query simplification. Existing work on database bug detection integrates SQL query simplification into their workflows, employing methods for SQL query simplification as outlined in Table 1. Table 1 provides a comparative overview of various SQL query simplification tools, detailing their strategies, correctness guarantee, DBMS extensibility, and so on.",
+      "context_after": "The elimination of operations within subqueries and removing unused columns [ 11] broaden the extent of the simplification,which enhances the simplification effectiveness. However, existing tools have yet to effectively address these two main challenges: adaptability to various DBMSs and preserving semantic correctness and bug-triggering capability after simplification. The first challenge involves achieving adaptability to various DBMSs, as existing efforts predominantly depend on the specific grammar of target DBMSs to simplify SQL queries. SQL dialects [ 47] represent subtle variations of",
+      "section": "2.2 Challenges of SQL Query Simplification",
+      "page": 3,
+      "char_offset": 14411,
+      "cited_reference": {
+        "number": 29,
+        "text": "Manuel Rigger. 2023. SQLancer. https://github.com/sqlancer/sqlancer. [Online; accessed 23-Nov-2023].",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As illustrated in Table 1, existing tools primarily deploy two approaches: •ClauseDelete: Tools such as SQLancer [29],RAGS [31], and PINOLO [7] employ a strategy of randomly deleting clauses from SQL queries.",
+      "context_before": "QLite 143 48.7 37.10% 90.32% 46.1 403 All 180 47.3 37.74% 90.32% 43.3 403 queries which is defined as the percentage reduction in the number of tokens. It is computed as follows: SimRatio(%) = Original Token Count −Reduced Token Count Original Token Count ×100 (1) •Simplification Time: This metric measures the temporal efficiency of the simplification process from initiation to completion. Shorter times are indicative of a more efficient simplification algorithm, which is critical for real-time database bug detection and maintenance tasks. 5.1.3 Baselines. We compare SQLess with two baselines.",
+      "context_after": "This approach is straightforward to implement and is limited to the parts of the query. •APOLLO: This tool [ 11] takes simplification further by allowing subquery simplifications and column deletions, thereby expanding the scope of the simplification strategy. Note that the above tools all use different parsers and a variety of error types. Since the query simplification strategies in these baselines are not implemented as an isolated component, and we have no way to run the original tools, to ensure a fair comparison, we have implemented these two baselines on the ANTLR MySQL grammar utilize",
+      "section": "5.1.2 Metrics. We use two metrics to assess SQLess: simplification",
+      "page": 8,
+      "char_offset": 43842,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[ 7,9,30]",
+      "technique": "tlp",
+      "sentence": "They typically suffer from a dilemma that the malformed, complex, and lengthy queries are more likely to trigger bugs but are less likely to 752 SQLess: Dialect-Agnostic SQL Query Simplification ISSTA ’24, September 16–20, 2024, Vienna, Austria be accepted by DBMS developers, which is highlighted by many prior studies [ 7,9,30].",
+      "context_before": "he confirmation or fixing of bugs that had been outstanding for an average of 246.3 days, all within an average span of just 0.7 days. Owing to the rapid resolution times by the developers of PostgreSQL and SQLite, SQLess can only simplify three bug reports for these DBMSs, focusing on issues that were still unconfirmed or unresolved. To track the status of the bugs we generated simplified queries, we published a list of these bugs in a public GitHub repository [39]. The potential users of SQLess primarily fall into two categories. First, the tool can greatly benefit designers of DBMS fuzzers.",
+      "context_after": "Second, SQLess can potentially benefit DBMS developers when they have difficulties in debugging bugs involving lengthy and complex queries, thereby reducing their workload. For instance, the bug report #32981 for MariaDB elicited appreciative feedback from an experienced developer: \"Thank you very much! Yes, it is quite useful.\" Similarly, in the bug report #1678 for OceanBase, the developer acknowledged the usefulness of the simplified query. Such positive feedback underscores the practical utility of SQLess and its significance in assisting developers to effectively address and fix bugs. Po",
+      "section": "5.4 Ablation Study",
+      "page": 10,
+      "char_offset": 60662,
+      "cited_reference": {
+        "number": 30,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[ 7,11,27,29,31]",
+      "technique": null,
+      "sentence": "Nevertheless, existing work has adapted the principles of Delta Debugging to simplify complex SQL queries that lead to certain behaviors, such as incorrect result sets or performance issues [ 7,11,27,29,31].",
+      "context_before": "e studies in this work, performing more thorough and comprehensive user studies for more debugging tasks is an important future work. 7 Related Work Test Case Minimization. Delta debugging [ 12,21,49,50] and Hierarchical Delta Debugging [ 16] target simplifying failing test cases to minimal ones that still produce failures. They can be utilized for batch debugging of independent SQL statement collections or lists of numbers. However, they cannot directly be used for SQL simplification because they require adherence to strict grammar rules that SQL’s complex and nested structures often violate.",
+      "context_after": "For example, Reducer [27] employs a two-step approach for minimizing queries: it sequentially removes each line and eliminates column names from SELECT statements. Besides, RAGS [ 31] eliminates terms within expressions and also removes WHERE and HAVING clauses. Later, APOLLO [11] improves the above approaches by adding several simplification strategies. Unfortunately, these tools can achieve effective simplification only if there are few dependencies between elements. An excessive amount of dependence yields output that is not minimized adequately. Furthermore, these tools expects structured",
+      "section": "7 Related Work",
+      "page": 11,
+      "char_offset": 63291,
+      "cited_reference": {
+        "number": 29,
+        "text": "Manuel Rigger. 2023. SQLancer. https://github.com/sqlancer/sqlancer. [Online; accessed 23-Nov-2023].",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For instance, SQLancer [ 29] constructs equivalent queries to ensure consistent results.",
+      "context_before": "h attention. Specifically, researchers have focused on detecting two types of bugs in DBMSs: crashes and logical bugs. To induce crashes in a DBMS, previous studies have primarily concentrated on automating query generation. On the one hand, generation-based methods employ well-defined rules to create valid DBMS queries [ 5,17,35,44], ensuring the syntactic validity of the queries. On the other hand, mutation-based approaches apply specific forms of mutators to SQL queries [ 6,48,51]. In addition to automatically generating SQL queries, detecting logical bugs requires an effective test oracle.",
+      "context_after": "Similarly, PINOLO [ 7] leverages the approximation relation as a specific type of metamorphic relation acting as the test oracle. Our approach, distinct from prior SQL testing studies, simplifies SQL queries using SQLess, providing concise, compelling evidence to address bugs, thanks to reduced query sizes. 8 Conclusion SQL query simplification reduces the length of a lengthy SQL query in different contexts, facilitating debugging and patching in DBMSs. This paper presents SQLess, a dialect-agnostic approach for simplifying SQL queries. SQLess effectively reduces the sizes of queries generate",
+      "section": "7 Related Work",
+      "page": 11,
+      "char_offset": 66280,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:24:46Z",
+    "is_model_written": true,
+    "summary": "SQLess simplifies bug-triggering SQL queries. The authors note that to reach deep bugs most testing techniques generate long, complex queries, leaving developers to debug them. Unlike previous simplifiers that depend on DBMS-specific grammar, SQLess uses an adaptive parser with error recovery and grammar expansion to handle different dialects, and performs semantics-sensitive trimming using alias and dependency analysis so the query still triggers the bug. On over 32,000 complex queries across six DBMSs it achieved an average simplification rate of 72.45%.",
+    "narrative": "SQLancer is the example of the problem SQLess addresses: it is named among the tools that integrate query simplification into their workflow only simplistically, deleting whole clauses, and its equivalent-query construction is what produces the long queries that then need reducing.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "background",
+      "M3": "motivation",
+      "M4": "motivation",
+      "M5": "motivation",
+      "M6": "background",
+      "M7": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3650212_3680318.json b/_data/papers/paper_doi_10_1145_3650212_3680318.json
new file mode 100644
index 0000000..3587133
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3650212_3680318.json
@@ -0,0 +1,1045 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:09:50Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3650212.3680318",
+    "title": "DBStorm: Generating Various Effective Workloads for Testing Isolation Levels",
+    "authors": [
+      "Keqiang Li",
+      "Siyang Weng",
+      "Lyu Ni",
+      "Chengcheng Yang",
+      "Rong Zhang",
+      "Xuan Zhou",
+      "Aoying Zhou"
+    ],
+    "year": 2024,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3650212.3680318",
+    "arxiv_id": null,
+    "s2_paper_id": "e54a5920715b453038a64e65a8c06fa4ce1199db",
+    "url": "https://doi.org/10.1145/3650212.3680318",
+    "also_indexed_as": [
+      "paper:s2:1b209b2da737fd26c50a2cb9116c73c9d1db904e"
+    ]
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3650212.3680318",
+      "retrieved_at": "2026-09-10T15:09:50Z",
+      "chars": 85543,
+      "content_sha256": "sha256:4ad138e395c55783f24cb10ec52a4146de7fe300a7c23ca996bed393be3211fa"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2918
+      },
+      {
+        "number": "2",
+        "title": "Preliminary",
+        "start": 11852
+      },
+      {
+        "number": "2.1",
+        "title": "Database and Workload",
+        "start": 11866
+      },
+      {
+        "number": "2.2",
+        "title": "General Concurrency Control Algorithm",
+        "start": 13876
+      },
+      {
+        "number": "2.3",
+        "title": "Problem Definition",
+        "start": 15776
+      },
+      {
+        "number": "3",
+        "title": "DBStorm Framework",
+        "start": 20805
+      },
+      {
+        "number": "4",
+        "title": "Operation Generation",
+        "start": 27828
+      },
+      {
+        "number": "4.1",
+        "title": "Mirroring the Data State",
+        "start": 27851
+      },
+      {
+        "number": "4.2",
+        "title": "Instantiating Access Parameters",
+        "start": 34596
+      },
+      {
+        "number": "5",
+        "title": "Transaction Generation",
+        "start": 39223
+      },
+      {
+        "number": "5.1",
+        "title": "Conflict Graph Isomorphism Mapping",
+        "start": 39248
+      },
+      {
+        "number": "5.2",
+        "title": "The Mutually Orthogonal Editing Methods",
+        "start": 42511
+      },
+      {
+        "number": "5.3",
+        "title": "The Hybrid Editing Method",
+        "start": 45225
+      },
+      {
+        "number": "6",
+        "title": "Anomaly Customization",
+        "start": 49521
+      },
+      {
+        "number": "7",
+        "title": "Experiments",
+        "start": 54985
+      },
+      {
+        "number": "7.1",
+        "title": "IL-related Code Coverage",
+        "start": 56226
+      },
+      {
+        "number": "7.2",
+        "title": "Effective Testing",
+        "start": 57967
+      },
+      {
+        "number": "7.3",
+        "title": "Anomaly-Sensitive Testing",
+        "start": 60394
+      },
+      {
+        "number": "7.4",
+        "title": "Bug Result",
+        "start": 63028
+      },
+      {
+        "number": "8",
+        "title": "Discussion",
+        "start": 65350
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2023. Carnegie Mellon Database Application Catalog. https://github.com/cmudb/cmdbac Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2023. InnoDB Transaction Isolation Levels. https://dev.mysql.com/doc/refman/ 8.0/en/innodb-transaction-isolation-levels.html Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2023. Jepsen. https://github.com/jepsen-io/jepsen Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2023. NuoDB. https://nuodb.com/ Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2023. Oracle Database. https://www.oracle.com/hk/database/technologies/ Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2023. SingleStore. https://www.singlestore.com/ Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2023. SQLite. https://www.sqlite.org/index.html. Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2023. SQLsmith. https://github.com/anse1/sqlsmith Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2023. TPC-C Benchmark. http://www.tpc.org/tpcc/ Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2023. TPC-E Benchmark. http://www.tpc.org/tpce/ Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2023. yugabyte DB. https://www.yugabyte.com/ Accessed: 2023-12-1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2024. Technical Report of DBStorm. https://github.com/DBHammer/DBStormmaterial/blob/main/tech-report.pdf Accessed: 2024-7-13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Atul Adya and Barbara H Liskov. 1999. Weak consistency: a generalized theory and optimistic implementations for distributed transactions. Ph. D. Dissertation. Massachusetts Institute of Technology. https://dl.acm.org/doi/10.5555/888672",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Mohammad Alomari, Michael Cahill, Alan Fekete, and Uwe Rohm. 2008. The cost of serializability on platforms that use snapshot isolation. In ICDE. 576–585. https://dl.acm.org/doi/10.1109/ICDE.2008.4497466",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Güneş Aluç, Olaf Hartig, M Tamer Özsu, and Khuzaima Daudjee. 2014. Diversified stress testing of RDF data management systems. In ISWC. 197–212. https: //dl.acm.org/doi/10.1007/978-3-319-11964-9_13",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Peter Alvaro and Kyle Kingsbury. 2020. Elle: Inferring isolation anomalies from experimental observations. In VLDB, Vol. 14. 268–280. https://dl.acm.org/doi/10. 14778/3430915.3430918",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Paul Ammann and Jeff Offutt. 2016. Introduction to software testing. Cambridge University Press. https://dl.acm.org/doi/10.5555/1355340",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Peter Armitage, Geoffrey Berry, and John Nigel Scott Matthews. 2008. Statistical methods in medical research. John Wiley & Sons.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In ICSE. 2060–2071. https://dl.acm.org/doi/10.1109/ICSE48619.2023. 00174",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Yunsheng Bai, Hao Ding, Ken Gu, Yizhou Sun, and Wei Wang. 2020. Learningbased efficient graph similarity computation via multi-scale convolutional set matching. In AAAI, Vol. 34. 3219–3226. https://doi.org/10.1609/aaai.v34i04.5720",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. 2007. A genetic approach for random testing of database systems. In VLDB. 1243–1251. https://dl.acm.org/doi/10.5555/1325851.1325993",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Hal Berenson, Phil Bernstein, Jim Gray, Jim Melton, Elizabeth O’Neil, and Patrick O’Neil. 1995. A critique of ANSI SQL isolation levels. In SIGMOD, Vol. 24. 1–10. https://dl.acm.org/doi/10.1145/223784.223785",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Souvik Bhattacherjee, Gang Liao, Michael Hicks, and Daniel J. Abadi. 2021. BullFrog: Online schema evolution via lazy evaluation. In SIGMOD. 194–206. https://dl.acm.org/doi/10.1145/3448016.3452842",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Ranadeep Biswas, Diptanshu Kakwani, Jyothi Vedurada, Constantin Enea, and Akash Lal. 2021. MonkeyDB: effectively testing correctness under weak isolation levels. In PACMPL, Vol. 5. 1–27. https://dl.acm.org/doi/10.1145/3485546",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "David B Blumenthal and Johann Gamper. 2020. On the exact computation of the graph edit distance. In Pattern Recognition Letters, Vol. 134. Elsevier, 46–57. https://dl.acm.org/doi/10.1016/j.patrec.2018.05.002",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Lixiang Chen, Ruihao Chen, Chengcheng Yang, Yuxing Han, Rong Zhang, Xuan Zhou, Peiquan Jin, and Weining Qian. 2023. Workload-aware log-structured merge key-value store for NVM-SSD hybrid storage. In ICDE. 2207–2219. https: //doi.org/10.1109/ICDE55515.2023.00171",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Computer, Business Equipment Manufacturers Association, et al .1986. American National Standard for Information Systems-Database Language SQL. In NY, American National Standards Institute. 27–28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Brian F Cooper, Adam Silberstein, Erwin Tam, Raghu Ramakrishnan, and Russell Sears. 2010. Benchmarking cloud serving systems with YCSB. In SoCC. 143–154. https://dl.acm.org/doi/10.1145/1807128.1807152",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Irving M Copi, Carl Cohen, and Victor Rodych. 2018. Introduction to logic. Routledge. https://dl.acm.org/doi/book/10.5555/2385432",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "James C Corbett, Jeffrey Dean, Michael Epstein, Andrew Fikes, Christopher Frost, Jeffrey John Furman, Sanjay Ghemawat, Andrey Gubarev, Christopher Heiser, Peter Hochschild, et al .2013. Spanner: Google’s globally distributed database. In TOCS, Vol. 31. 1–22. https://dl.acm.org/doi/pdf/10.1145/2491245",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Natacha Crooks, Youer Pu, Lorenzo Alvisi, and Allen Clement. 2017. Seeing is believing: A client-centric specification of database isolation. In PODC. 73–82. https://dl.acm.org/doi/10.1145/3087801.3087802",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Ziyu Cui, Wensheng Dou, Qianwang Dai, Jiansen Song, Wei Wang, Jun Wei, and Dan Ye. 2022. Differentially testing database transactions for fun and profit. In ASE. 1–12. https://dl.acm.org/doi/abs/10.1145/3551349.3556924",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Yuetang Deng, Phyllis Frankl, and Zhongqiang Chen. 2003. Testing database transaction concurrency. In ASE. 184–193. https://dl.acm.org/doi/10.1109/ASE. 2003.1240306",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Akon Dey, Alan Fekete, Raghunath Nambiar, and Uwe Röhm. 2014. YCSB+ T: Benchmarking web-scale transactional databases. In ICDE Workshop. 223–230. https://doi.org/10.1109/ICDEW.2014.6818330",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Djellel Eddine Difallah, Andrew Pavlo, Carlo Curino, and Philippe CudreMauroux. 2013. OLTP-bench: An extensible testbed for benchmarking relational databases. In VLDB, Vol. 7. 277–288. https://dl.acm.org/doi/10.14778/2732240. 2732246",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "W Dou, Z Cui, Q Dai, J Song, D Wang, Y Gao, W Wang, J Wei, L Chen, H Wang, et al.2023. Detecting isolation bugs via transaction oracle construction. In ICSE. 1123–1135. https://dl.acm.org/doi/10.1109/ICSE48619.2023.00101",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Alan Fekete, Dimitrios Liarokapis, Elizabeth O’Neil, Patrick O’Neil, and Dennis Shasha. 2005. Making snapshot isolation serializable. In TODS, Vol. 30. 492–528. https://dl.acm.org/doi/abs/10.1145/1071610.1071615",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Muon Ha and Yulia Shichkina. 2022. Translating a distributed relational database to a document database. DSE7, 2 (2022), 136–155. https://doi.org/10.1007/s41019022-00181-9",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting bugs in cypher graph database engines. In ISSTA. 163–174. https://dl.acm.org/doi/10.1145/3597926.3598046",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Dongxu Huang, Qi Liu, Qiu Cui, Zhuhe Fang, Xiaoyu Ma, Fei Xu, Li Shen, Liu Tang, Yuxing Zhou, Menglong Huang, et al .2020. TiDB: a raft-based HTAP database. In VLDB, Vol. 13. 3072–3084. https://dl.acm.org/doi/10.14778/3415478. 3415535",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via graph-based oracle construction. In OSDI. 397–417. https://www.usenix.org/conference/osdi23/presentation/jiang",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 42,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. Apollo: Automatic detection and diagnosis of performance regressions in database systems. In VLDB, Vol. 13. 57–70. https://dl.acm.org/doi/10.14778/3357377.3357382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing graph database engines via query partitioning. In ISSTA. 140–149. https://dl.acm. org/doi/abs/10.1145/3597926.3598044",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "Per-Åke Larson, Adrian Birka, Eric N Hanson, Weiyun Huang, Michal Nowakiewicz, and Vassilis Papadimos. 2015. Real-time analytical processing with SQL server. In VLDB, Vol. 8. 1740–1751. https://dl.acm.org/doi/10.14778/ 2824032.2824071",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Feifei Li. 2019. Cloud-native database systems at Alibaba: Opportunities and challenges. In VLDB, Vol. 12. 2263–2272. https://dl.acm.org/doi/abs/10.14778/ 3352063.3352141",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Guoliang Li, Xuanhe Zhou, Ji Sun, Xiang Yu, Yue Han, Lianyuan Jin, Wenbo Li, Tianqing Wang, and Shifu Li. 2021. Opengauss: An autonomous database system. In VLDB, Vol. 14. 3028–3042. https://dl.acm.org/doi/abs/10.14778/3476311. 3476380",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Keqiang Li, Siyang Weng, Peiyuan Liu, Lyu Ni, Chengcheng Yang, Rong Zhang, Xuan Zhou, Jianghang Lou, Gui Huang, Weining Qian, et al .2023. Leopard: A black-box approach for efficiently verifying various isolation levels. In ICDE. 722–735. https://doi.org/10.1109/ICDE55515.2023.00061",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Yujia Li, Chenjie Gu, Thomas Dullien, Oriol Vinyals, and Pushmeet Kohli. 2019. Graph matching networks for learning the similarity of graph structured objects. InICML. 3835–3845. http://proceedings.mlr.press/v97/li19d.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting logical bugs of DBMS with coverage-based guidance. In USENIX Security 22. 4309–4326. https://www.usenix. org/conference/usenixsecurity22/presentation/liang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Peiyuan Liu, Siyang Weng, Keqiang Li, Lyu Ni, Chengcheng Yang, Rong Zhang, Weining Qian, and Dian Qiao. 2024. Leopard: A general test suite for isolation level verification. In CIDR. https://www.cidrdb.org/cidr2024/papers/p44-liu.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. In ICSE. 225– 236. https://dl.acm.org/doi/abs/10.1145/3510003.3510093",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Yi Lu, Xiangyao Yu, Lei Cao, and Samuel Madden. 2020. Aria: a fast and practical deterministic OLTP database. In VLDB, Vol. 13. 2047–2060. https://dl.acm.org/ doi/10.14778/3407790.3407808",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Hao Luo, Mehedi Masud, and Hasan Ural. 2012. Detecting offline transaction concurrency problems.. In J. Softw., Vol. 7. 1855–1860. https://doi.org/10.4304/ jsw.7.8.1855-1860",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Dan R. K. Ports and Kevin Grittner. 2012. Serializable Snapshot Isolation in PostgreSQL. In VLDB, Vol. 5. 1850–1861. https://dl.acm.org/doi/10.14778/2367502. 2367523",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Dong Qiu, Bixin Li, and Zhendong Su. 2013. An empirical analysis of the coevolution of schema and code in database applications. In SIGSOFT. 125–135. https://dl.acm.org/doi/abs/10.1145/2491411.2491431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Luyi Qu, Yuming Li, Rong Zhang, Ting Chen, Ke Shu, Weining Qian, and Aoying Zhou. 2022. Application-oriented workload generation for transactional database performance evaluation. In ICDE. 420–432. https://doi.org/10.1109/ICDE53745. 766 DBStorm: Generating Various Effective Workloads for Testing Isolation Levels ISSTA ’24, September 16–20, 2024, Vienna, Austria 2022.00036",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Kia Rahmani, Kartik Nagar, Benjamin Delaware, and Suresh Jagannathan. 2019. CLOTHO: directed test generation for weakly consistent database systems. In PACMPL, Vol. 3. 1–28. https://dl.acm.org/doi/abs/10.1145/3360543",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Kim-Thomas Rehmann, Changyun Seo, Dongwon Hwang, Binh Than Truong, Alexander Boehm, and Dong Hun Lee. 2016. Performance monitoring in SAPHANA’s continuous integration process. In ACMSIGMETRICS Performance Evaluation Review, Vol. 43. 43–52. https://dl.acm.org/doi/abs/10.1145/2897356.2897362",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Kun Ren, Jose M Faleiro, and Daniel J Abadi. 2016. Design principles for scaling multi-core OLTP under high contention. In SIGMOD. 1583–1598. https://dl.acm. org/doi/abs/10.1145/2882903.2882958",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In OSDI. 667–682. https://dl.acm.org/doi/abs/10.1145/2882903. 2882958",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 61,
+      "text": "Jun Shao. 2003. Mathematical Statistics. In Springer Texts in Statistics.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Vishal Sikka, Franz Färber, Wolfgang Lehner, Sang Kyun Cha, Thomas Peh, and Christof Bornhövd. 2012. Efficient transaction processing in SAPHANA database: the end of a column store myth. In SIGMOD. 731–742. https://dl.acm.org/doi/ abs/10.1145/2213836.2213946",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In VLDB. 618–622. https://dl.acm.org/doi/10.5555/645924.671199",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Adriana Szekeres and Irene Zhang. 2018. Making consistency more consistent: A unified model for coherence, consistency and isolation. In PaPoC. 1–8. https: //dl.acm.org/doi/abs/10.1145/3194261.3194268",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Rebecca Taft, Irfan Sharif, Andrei Matei, Nathan VanBenschoten, Jordan Lewis, Tobias Grieger, Kai Niemi, Andy Woods, Anne Birzin, Raphael Poss, et al .2020. CockroachDB: The resilient geo-distributed SQL database. In SIGMOD. 1493–1509. https://dl.acm.org/doi/abs/10.1145/3318464.3386134",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Cheng Tan, Changgeng Zhao, Shuai Mu, and Michael Walfish. 2020. Cobra: Making transactional key-value stores verifiably serializable. In OSDI. 63–80. https://dl.acm.org/doi/abs/10.5555/3488766.3488770",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting logic bugs of join optimizations in DBMS. In SIGMOD, Vol. 1. 1–26. https: //dl.acm.org/doi/abs/10.1145/3588909",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Alexandre Verbitski, Anurag Gupta, Debanjan Saha, Murali Brahmadesam, Kamal Gupta, Raman Mittal, Sailesh Krishnamurthy, Sandor Maurice, Tengiz Kharatishvili, and Xiaofeng Bao. 2017. Amazon aurora: Design considerations for high throughput cloud-native relational databases. In SIGMOD. 1041–1052.https://dl.acm.org/doi/abs/10.1145/3035918.3056101",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry practice of coverage-guided enterprise-level DBMS fuzzing. In ICSE. 328–337. https://dl.acm.org/doi/abs/10.1109/ICSE-SEIP52600. 2021.00042",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Tianzheng Wang and Hideaki Kimura. 2016. Mostly-optimistic concurrency control for highly contended dynamic workloads on a thousand cores. In VLDB, Vol. 10. 49–60. https://dl.acm.org/doi/abs/10.14778/3015274.3015276",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Gerhard Weikum and Gottfried Vossen. 2001. Transactional information systems: theory, algorithms, and the practice of concurrency control and recovery. https: //dl.acm.org/doi/10.1145/604264.604275",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Zhenkun Yang, Chen Qian, Xuwang Teng, Fanyu Kong, Fusheng Han, and Quanqing Xu. 2023. LCL: A lock chain length-based distributed algorithm for deadlock detection and resolution. In ICDE. 151–163. https://doi.org/10.1109/ICDE55515. 2023.00019",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Zhenkun Yang, Chuanhui Yang, Fusheng Han, Mingqiang Zhuang, Bing Yang, Zhifeng Yang, Xiaojun Cheng, Yuzhong Zhao, Wenhui Shi, Huafeng Xi, et al. 2022. OceanBase: a 707 million tpmC distributed relational database system. In VLDB, Vol. 15. 3385–3397. https://dl.acm.org/doi/abs/10.14778/3554821.3554830",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Xiangyao Yu, Andrew Pavlo, Daniel Sanchez, and Srinivas Devadas. 2016. Tictoc: Time traveling optimistic concurrency control. In SIGMOD. 1629–1642. https: //dl.acm.org/doi/abs/10.1145/2882903.2882935",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Zhiping Zeng, Anthony KH Tung, Jianyong Wang, Jianhua Feng, and Lizhu Zhou. 2009. Comparing stars: On approximating graph edit distance. In VLDB, Vol. 2. 25–36. https://dl.acm.org/doi/abs/10.14778/1687627.1687631",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Mai Zheng, Joseph Tucek, Dachuan Huang, Feng Qin, Mark Lillibridge, Elizabeth S. Yang, Bill W. Zhao, and Shashank Singh. 2014. Torturing databases for fun and profit. In OSDI. 449–464. https://dl.acm.org/doi/10.5555/2685048.2685083",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding bugs in Gremlin-based graph database systems via randomized differential testing. In ISSTA. 302–313. https://dl.acm.org/doi/abs/10.1145/3533767.3534409",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing database management systems with language validity and coverage feedback. In CCS. 955–970. https://dl.acm.org/doi/abs/10. 1145/3372297.3417260 Received 2024-04-12; accepted 2024-07-03 767",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 19,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In ICSE. 2060–2071. https://dl.acm.org/doi/10.1109/ICSE48619.2023. 00174",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "prints the DOI of the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 41,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via graph-based oracle construction. In OSDI. 397–417. https://www.usenix.org/conference/osdi23/presentation/jiang",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 43,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing graph database engines via query partitioning. In ISSTA. 140–149. https://dl.acm. org/doi/abs/10.1145/3597926.3598044",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 60,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In OSDI. 667–682. https://dl.acm.org/doi/abs/10.1145/2882903. 2882958",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 60,63,67,78]",
+      "technique": "pqs",
+      "sentence": "Specifically, the query-oriented methods [ 60,63,67,78] generate carefully crafted queries to detect the bugs of query engines, thus they only trigger the code of query engine instead of the IL code.",
+      "context_before": "ays been tough work. The core challenges come from the requirements of: Black-box Testing (C1). As the source codes of numerous DBMSs (e.g., cloud DBMSs) are not visible [ 47,66], it imposes the 755 ISSTA ’24, September 16–20, 2024, Vienna, Austria Keqiang Li, Siyang Weng, Lyu Ni, Chengcheng Yang, Rong Zhang, Xuan Zhou, and Aoying Zhou requirement of generating workloads which trigger the IL code of a black-box DBMS. Though there are extensively studied in DBMS testing, none can generate a huge variety of IL relevant workloads. As a result, the IL test space cannot be comprehensively explored.",
+      "context_after": "The transactional-oriented methods [ 32,36,41] generate workloads where a small number of operations are serially executed, constructing sparse and small conflict graphs. Thus, they might miss bugs triggered by large and dense conflict graphs, since stress testing has been proved to be effective in finding some complex bugs [ 15]. Additionally, the widely used benchmarking [ 9,14,28] are designed for a specific scenario but could not cover more various scenarios. Effective Testing (C2). Most of the DBMSs that serve for OLTP applications are relational [ 1,35,38]. In general, the sources of di",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5562,
+      "cited_reference": {
+        "number": 60,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In OSDI. 667–682. https://dl.acm.org/doi/abs/10.1145/2882903. 2882958",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 32,36,41]",
+      "technique": null,
+      "sentence": "The transactional-oriented methods [ 32,36,41] generate workloads where a small number of operations are serially executed, constructing sparse and small conflict graphs.",
+      "context_before": "SSTA ’24, September 16–20, 2024, Vienna, Austria Keqiang Li, Siyang Weng, Lyu Ni, Chengcheng Yang, Rong Zhang, Xuan Zhou, and Aoying Zhou requirement of generating workloads which trigger the IL code of a black-box DBMS. Though there are extensively studied in DBMS testing, none can generate a huge variety of IL relevant workloads. As a result, the IL test space cannot be comprehensively explored. Specifically, the query-oriented methods [ 60,63,67,78] generate carefully crafted queries to detect the bugs of query engines, thus they only trigger the code of query engine instead of the IL code.",
+      "context_after": "Thus, they might miss bugs triggered by large and dense conflict graphs, since stress testing has been proved to be effective in finding some complex bugs [ 15]. Additionally, the widely used benchmarking [ 9,14,28] are designed for a specific scenario but could not cover more various scenarios. Effective Testing (C2). Most of the DBMSs that serve for OLTP applications are relational [ 1,35,38]. In general, the sources of diversity in relational model include the diversities of both SQL operations and the interleavings among concurrent SQL operations. Moreover, no matter how a SQL operation c",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5762,
+      "cited_reference": {
+        "number": 41,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via graph-based oracle construction. In OSDI. 397–417. https://www.usenix.org/conference/osdi23/presentation/jiang",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 32,36,41]",
+      "technique": null,
+      "sentence": "Moreover, others [ 32,36,41] support complex SQL operations and predicates but lacks guidance for avoiding redundant or ineffective workloads.",
+      "context_before": "ineffective workloads. The redundant workloads are characterized by repeating a similar partial order of record accesses, while the ineffective workloads appear when the expected conflict cannot be formed due to unreasonable scheduling (e.g., the record expected to be accessed by an operation is deleted by another parallel operation, and then it could not access its expected record or participate in any conflict formation). Some of existing work [ 3,16,66] has been designed to only generate simplified key-value operations, and they might omit some bugs triggered by more complex SQL operations.",
+      "context_after": "Anomaly-sensitive Testing (C3). DBMSs have various ILs, each with a distinctive definition to preclude specific non-serializable anomalies [ 13,22,27,64]. Although the effective testing could provide various scenarios for IL test, it could not help customize some specific non-serializable anomalies. Then, to effectively test various ILs, it imposes the requirement of generating anomaly-sensitive workloads which check whether an IL prohibits the anomalies as defined. However, customizing non-serializable anomalies is still vacant in existing researches [19, 24, 32, 33, 36, 39, 43, 53, 57, 77].",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7406,
+      "cited_reference": {
+        "number": 41,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via graph-based oracle construction. In OSDI. 397–417. https://www.usenix.org/conference/osdi23/presentation/jiang",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[19, 24, 32, 33, 36, 39, 43, 53, 57, 77]",
+      "technique": "qpg",
+      "sentence": "However, customizing non-serializable anomalies is still vacant in existing researches [19, 24, 32, 33, 36, 39, 43, 53, 57, 77].",
+      "context_before": "ers [ 32,36,41] support complex SQL operations and predicates but lacks guidance for avoiding redundant or ineffective workloads. Anomaly-sensitive Testing (C3). DBMSs have various ILs, each with a distinctive definition to preclude specific non-serializable anomalies [ 13,22,27,64]. Although the effective testing could provide various scenarios for IL test, it could not help customize some specific non-serializable anomalies. Then, to effectively test various ILs, it imposes the requirement of generating anomaly-sensitive workloads which check whether an IL prohibits the anomalies as defined.",
+      "context_after": "We propose DBStorm to address the challenges described above. After carefully investigating 15popular DBMSs [ 2,4–7,11,30,40, 44–46,54,62,68,73], we discover their IL implementations follow a general framework to manage and resolve the access conflicts. Specifically, it consists of two tasks: 1) build a conflict graph to manage access conflicts; 2) perform a verification policy to prevent the conflict graph from non-serializable anomalies. Then, to effectively explore the IL code logic of a black-box DBMS (i.e., C1), we should generate a workload with a large number of conflict graphs. To add",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 8020,
+      "cited_reference": {
+        "number": 19,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In ICSE. 2060–2071. https://dl.acm.org/doi/10.1109/ICSE48619.2023. 00174",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "prints the DOI of the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLancer ,DT2).",
+      "context_before": "ccess records, i.e., the access ratio 𝛼=1; and (4) non-serializable anomalies can be customized for any kind of IL. 3 DBStorm Framework The framework in Fig. 2 includes: 1) populating data into a database; 2) generating syntax/semantic correct operations that follow a specific distribution to access the database; 3) organizing these operations as transactions to create workloads with dissimilar conflict graphs; 4) customizing specific anomalies within the workloads. ①Preparing a Database Instance. We prepare a database instance with massive amount of data by following existing approaches (e.g.",
+      "context_after": "Specifically, we regard all database objects, i.e., tables, columns and keys, as the nodes of a topology graph whose edges indicate the dependencies among objects. For example, there is a dependency from a referencing table to the referenced one. Then, we create the referenced table first according to the dependency order. Moreover, we also elaborate some rules for records generation. For example, generate 𝑝𝑘values in an incremental manner, and generate𝑝𝑘values before 𝑓𝑘values. Note, 𝑛𝑘and𝑓𝑘values can be generated by following user-specified data distribution functions. ②Generating SQL Operat",
+      "section": "3 DBStorm Framework",
+      "page": 4,
+      "char_offset": 21289,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[ 8,32,36,60,63,78]",
+      "technique": "pqs",
+      "sentence": "SQL operation generation has been extensively studied in [ 8,32,36,60,63,78].",
+      "context_before": "ifically, we regard all database objects, i.e., tables, columns and keys, as the nodes of a topology graph whose edges indicate the dependencies among objects. For example, there is a dependency from a referencing table to the referenced one. Then, we create the referenced table first according to the dependency order. Moreover, we also elaborate some rules for records generation. For example, generate 𝑝𝑘values in an incremental manner, and generate𝑝𝑘values before 𝑓𝑘values. Note, 𝑛𝑘and𝑓𝑘values can be generated by following user-specified data distribution functions. ②Generating SQL Operations.",
+      "context_after": "As SQLancer [60] targets for the relational model, we leverage it to generate syntax/semantic correct SQL operations. Since SQLancer mainly generates serially executed operations, we revise it to generate parallel operations. However, due to the various semantic operations in relational model (e.g., delete ,insert ,update and select with complex predicates), the existence of a record is often ambiguous for a given operation when executing concurrent operations. This ambiguity would cause the operation to fail to access any record (e.g., no record satisfies the operation’s predicate). Since th",
+      "section": "3 DBStorm Framework",
+      "page": 4,
+      "char_offset": 21911,
+      "cited_reference": {
+        "number": 60,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In OSDI. 667–682. https://dl.acm.org/doi/abs/10.1145/2882903. 2882958",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As SQLancer [60] targets for the relational model, we leverage it to generate syntax/semantic correct SQL operations.",
+      "context_before": "he nodes of a topology graph whose edges indicate the dependencies among objects. For example, there is a dependency from a referencing table to the referenced one. Then, we create the referenced table first according to the dependency order. Moreover, we also elaborate some rules for records generation. For example, generate 𝑝𝑘values in an incremental manner, and generate𝑝𝑘values before 𝑓𝑘values. Note, 𝑛𝑘and𝑓𝑘values can be generated by following user-specified data distribution functions. ②Generating SQL Operations. SQL operation generation has been extensively studied in [ 8,32,36,60,63,78].",
+      "context_after": "Since SQLancer mainly generates serially executed operations, we revise it to generate parallel operations. However, due to the various semantic operations in relational model (e.g., delete ,insert ,update and select with complex predicates), the existence of a record is often ambiguous for a given operation when executing concurrent operations. This ambiguity would cause the operation to fail to access any record (e.g., no record satisfies the operation’s predicate). Since the precondition of conflict formation is that two operations actually access the same record, the operation’s failed ac",
+      "section": "3 DBStorm Framework",
+      "page": 4,
+      "char_offset": 21989,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Since SQLancer mainly generates serially executed operations, we revise it to generate parallel operations.",
+      "context_before": "rom a referencing table to the referenced one. Then, we create the referenced table first according to the dependency order. Moreover, we also elaborate some rules for records generation. For example, generate 𝑝𝑘values in an incremental manner, and generate𝑝𝑘values before 𝑓𝑘values. Note, 𝑛𝑘and𝑓𝑘values can be generated by following user-specified data distribution functions. ②Generating SQL Operations. SQL operation generation has been extensively studied in [ 8,32,36,60,63,78]. As SQLancer [60] targets for the relational model, we leverage it to generate syntax/semantic correct SQL operations.",
+      "context_after": "However, due to the various semantic operations in relational model (e.g., delete ,insert ,update and select with complex predicates), the existence of a record is often ambiguous for a given operation when executing concurrent operations. This ambiguity would cause the operation to fail to access any record (e.g., no record satisfies the operation’s predicate). Since the precondition of conflict formation is that two operations actually access the same record, the operation’s failed access cannot participate the conflict formation. Then, the operation is considered as ineffective. Thus, we s",
+      "section": "3 DBStorm Framework",
+      "page": 4,
+      "char_offset": 22107,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 32,36,41]",
+      "technique": null,
+      "sentence": "Previous studies [ 32,36,41] propose to sequentially execute all operations such that the order of execution can be effectively manifested, but they might overlook some parallelism-related bugs.",
+      "context_before": "between transactions and the edge density of the conflict graph. This is because all the three kinds of conflicts (i.e., 𝑟𝑤,𝑤𝑟and𝑤𝑤) are closely related to write operations in each transaction. Thus, two transactions with different read/write ratios cannot be isomorphically mapped. Note, the vertices in the conflict graph are connected by directed edges and the edge direction is determined by the execution order of conflict operations. However, the parallel execution of numerous operations imposes great challenges in specifying the deterministic execution order of conflicting operations [ 52].",
+      "context_after": "Since the undirected conflict graph could well depict the conflict relation between transactions, we consider the case of undirected graph in this paper. 5.2 The Mutually Orthogonal Editing Methods We propose to follow criteria C1andC2to produce a sequence of dissimilar graphs G0,···G𝑛−1. Specifically, we devise a historyindependent approach by editing the graph monotonically. Definition 3. The monotonicity of a graph editing method. Suppose a graph editing method continuously produces a sequence of graphsG0,···,G𝑛−1. It has a monotonicity property if∀𝑖,𝑗∈ [0,𝑛−1]and𝑖<𝑗, there must exist 𝐺𝐸𝐷(",
+      "section": "5.1 Conflict Graph Isomorphism Mapping",
+      "page": 7,
+      "char_offset": 42161,
+      "cited_reference": {
+        "number": 41,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via graph-based oracle construction. In OSDI. 397–417. https://www.usenix.org/conference/osdi23/presentation/jiang",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [60] and TQS [67] randomly generate queries and provide a test oracle to identify logical bugs.",
+      "context_before": "Storm effectively trigger IL codes? (§ 7.1) RQ2) Can DBStorm generate effective workloads? (§ 7.2) RQ3) Can DBStorm customize non-serializable anomalies? (§ 7.3) RQ4) Can DBStorm uniquely identify IL bugs in popular commercial DBMSs? (§ 7.4) Setup. We deploy four servers connected via 1 Gigabit Ethernet. Each server is equipped with 2 Intel Xeon Silver 4110 @ 2.1 GHz CPUs, 120 GB memory, and 4 TBHDD disk. We use the open-source relational DBMS PostgreSQL ( 𝑣12.7) to verify our design. Baselines. SQLsmith [8] and Squirrel [78] randomly generate SQL expressions to find memory-related crash bugs.",
+      "context_after": "We also compare with the commonly used classic benchmarks TPC-C [9],SmallBank [14] and YCSB [28] to show the superiority of DBStorm, although it is not a perfect likefor-like comparison. Jepsen [3],DT2[32] and Troc [36], similar to DBStorm, generate transactional workloads. However, Jepsen is based on a simple key-value model instead of a relational model; DT2andTroc could generate complex SQL operations but execute them serially without parallelism. 7.1 IL-related Code Coverage A higher code coverage implies a higher likelihood of detecting bugs [ 17]. We employ the commonly used line covera",
+      "section": "7 Experiments",
+      "page": 9,
+      "char_offset": 55665,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer ensures queries access at least one record, but it also issues queries in a single-thread way.",
+      "context_before": "ge Comparison 0.05.0K 10.0K 15.0K 20.0K 25.0K 30.0K0.860.880.900.920.940.960.981.00 # of Operations01020304050 (a) Zipfian 0.05.0K 10.0K 15.0K 20.0K 25.0k 30.0K0.860.880.900.920.940.960.981.00 # of Operations0.00.10.20.30.40.5 (b) Uniform Figure 7: Access Ratio 𝜶and Distribution Fitness 𝜷 run all tools for one hour as in Fig. 6. SQLsmith andSquirrel generate single-threaded queries with most of operations access no record. It greatly limits conflict formation having the lowest coverage. Benchmarks only have manually tailored workloads and lack the workload variety, resulting in low coverage.",
+      "context_after": "Jepsen is designed for a key-value model, and cannot trigger the complex IL logic in a relational model. Although DT2and Troc execute transactions in parallel, all operations are executed sequentially without contentions. In contrast, DBStorm guarantees to generate effective workloads, which are concurrently executed on the relational model and achieves the best coverage. Note, we cannot have a full coverage, because there are many codes for debugging purposes, only triggered by debugging tools. 7.2 Effective Testing Access Ratio 𝛼& Distribution Fitness 𝛽.We take𝛼and𝛽to evaluate the ratio of",
+      "section": "7.1 IL-related Code Coverage",
+      "page": 9,
+      "char_offset": 57361,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer ,Squrriel and SQLsmith sequentially execute transactions in a single thread, so it is impossible to exist a cycle in the conflict graph and we do not put them in Fig.",
+      "context_before": "We can generate diverse non-serializable anomalies. If concurrency control algorithms are enabled, it would prevent anomalies and no anomalies would be collected regardless of test tools, unless isolation bugs occur in PostgreSQL. Fortunately, PostgreSQL takes specific prevention actions for each kind of anomaly, the number of which can serve as a good metric to evaluate the effectiveness of customized workloads. We then compare with existing approach in both disabled and enabled concurrency control (CC) algorithms. All tools generate workloads for one hour under default settings. Disable CC.",
+      "context_after": "10. YCSB only contains key-value operations with each operation treated as a transaction. However, it hardly produces dirty-write ,dirtyread and write-skew anomalies. SmallBank aims to create writeskew anomalies to benchmark the serializable snapshot isolation Figure 10: Anomaly Formation Comparison Figure 11: Anomaly Prevention Action Comparison algorithm. However, our results indicate it hardly generates writeskew anomalies with cycle lengths more than 2. In TPC-C, both new-order and payment transactions modify the same record in thedistrict table. Thus, there is a high probability of lost-",
+      "section": "7.3 Anomaly-Sensitive Testing",
+      "page": 10,
+      "char_offset": 61023,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "764 DBStorm: Generating Various Effective Workloads for Testing Isolation Levels ISSTA ’24, September 16–20, 2024, Vienna, Austria Table 2: Bug Comparison Total/Confirmed/Fixed DBStorm TxCheck Jepsen DT2 Isolation 20/12/6 7/6/0 22/21/13 4/2/1 Non-isolation 13/12/10 49/46/18 33/33/20 12/10/0 Total/Confirmed/Fixed Troc SQLsmith SQLancer Squirrel Isolation 5/5/1 0/0/0 0/0/0 0/0/0 Non-isolation 7/7/0 248/236/183 449/445/415 63/63/52 They can be classified into improper lock behavior (7 bugs), read inconsistent version (13 bugs), violated transaction control statements (7 bugs), unexpected concurrency-related error messages(3 bugs), unexpected update results (2 bugs), and transaction performance degradation (1 bug).",
+      "context_before": "er, this result is basically consistent with the result of disabling the concurrency control algorithms. Note, YCSB andJepsen also trigger lots of prevention actions for the read-skew anomaly. This is because both of them involve numerous concurrent write operations and range read operations that cover more data than item-reads, which would increase the likelihood of inconsistent data reads. 7.4 Bug Result Bug Summary. DBStorm has revealed 33 bugs, distributed across several commercial DBMSs, i.e., MySQL (5 bugs), PostgreSQL (2 bugs), TiDB (20 bugs), Oceanbase (3 bugs), and OpenGauss (3 bugs).",
+      "context_after": "We actually reveal 31 dissimilar bugs. Two pairs of similar bugs in MySQL/TiDB are revealed. There are 24 confirmed and 16 fixed bugs. Two reasons for the 8 unfixed bugs: (1) Technical limitations: four bugs are challenging to be resolved due to complex code logic; (2) Priority: four bugs are delayed due to their low impact. There are 9 bugs reported but cannot be reproduced by simple cases due to the nondeterministic transaction interleavings. We will further discuss with the developers. Bug Comparison. In Table 2, we compare our isolation bug findings with others. SQLsmith /SQLancer /Squirr",
+      "section": "7.4 Bug Result",
+      "page": 10,
+      "char_offset": 63233,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLsmith /SQLancer /Squirrel reveal 248/449/63 bugs, but none of them is an isolation bug.",
+      "context_before": "rmance degradation (1 bug). We actually reveal 31 dissimilar bugs. Two pairs of similar bugs in MySQL/TiDB are revealed. There are 24 confirmed and 16 fixed bugs. Two reasons for the 8 unfixed bugs: (1) Technical limitations: four bugs are challenging to be resolved due to complex code logic; (2) Priority: four bugs are delayed due to their low impact. There are 9 bugs reported but cannot be reproduced by simple cases due to the nondeterministic transaction interleavings. We will further discuss with the developers. Bug Comparison. In Table 2, we compare our isolation bug findings with others.",
+      "context_after": "DBStorm /TxCheck /Jepsen / DT2/Troc discover 20/7/22/4/5 isolation bugs. Note, Jepsen began testing in 2013 when DBMSs were less mature than nowadays. Among the 22 isolation bugs discovered by Jepsen ,DBStorm can reproduce 16 ones in DBMSs under different versions. We fail to reproduce the other 6 bugs because they are triggered by injecting faults, e.g., network interruptions, into DBMSs. In addition, DBStorm still finds 20 isolation bugs that have not been found by Jepsen in modern DBMSs; it also finds 13 non-isolation bugs due to collateral effects of testing. To summarize, existing method",
+      "section": "7.4 Bug Result",
+      "page": 11,
+      "char_offset": 64527,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[41]",
+      "technique": null,
+      "sentence": "TxCheck [41] constructs two semanticallyequivalent transactions with complex SQL queries, and compares their results to find bugs in transaction processing, including those in the query optimizer, query execution engine, and IL implementations.",
+      "context_before": "ving single-table operations with equality predicates but misses SQL semantics-related bugs (e.g., multi-table joins) founded by DBStorm .Elle[16] is proposed as Jepsen ’s test oracle to infer IL bugs, but it only handles simple key-value transactions and fails with complex SQL transactions. Moreover, Jepsen also injects faults to uncover non-isolation bugs regarding consistency and availability, such as those in consensus protocols, disaster recovery, and failover mechanisms. Thus, most bugs found by Jepsen are non-isolation, including system availability bugs, meta-data bugs, and crash bugs.",
+      "context_after": "It uncovers 49 non-isolation bugs and 7 isolation bugs. It generates semantically-equivalent test cases via topological sorting of operation-level conflict graphs but lacks guidance for generating critical cycles in transaction-level conflict graphs, which are essential for adequately testing ILs. DT2[32] randomly generates transactions, deterministically executes the operations one by one on multiple different DBMSs, and compares their results to find discrepancies. It would be more likely to find compatibility issues among different DBMSs, and in fact, 92 compatibility issues have been foun",
+      "section": "8 Discussion",
+      "page": 11,
+      "char_offset": 67395,
+      "cited_reference": {
+        "number": 41,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via graph-based oracle construction. In OSDI. 397–417. https://www.usenix.org/conference/osdi23/presentation/jiang",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[ 21,49,60,63]",
+      "technique": "pqs",
+      "sentence": "The bugs that have been discovered can be classified into logical bugs [ 21,49,60,63], performance bugs [ 42,51,58], and crash bugs [ 8,69,78].",
+      "context_before": ", 92 compatibility issues have been found. Troc [36] also executes the operations one by one, but the difference is that it theoretically calculates the ideal execution results of a pair of transactions and compares them with the results returned from the actual DBMS. TxCheck ,DT2andTroc fail to generate multiple parallel operations with high data access contentions, which is critical for trigger the IL-related code. Thus, they omit the IL bugs found by DBStorm, even for the DBMS they have tested. Query-Oriented Testing. Query-oriented testing aims to generate various queries for DBMS testing.",
+      "context_after": "A logical bug makes a query return an incorrect result. SQLancer [60] creates a query that would fetch a target data to detect logical bug. A performance bug adversely affects the query performance of DBMS. Apollo [42] detects performance bugs by comparing a new DBMS with an old one. A crash bug causes a DBMS to abruptly stop running. Squirrel [78] generates abnormal queries to detect crash bugs. 10 Conclusion In this paper, we presented DBStorm, an efficient approach to generate effective workloads for IL tests. We provided theoretical proofs to illustrate DBStorm ’s effectiveness and effici",
+      "section": "8 Discussion",
+      "page": 11,
+      "char_offset": 68800,
+      "cited_reference": {
+        "number": 60,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In OSDI. 667–682. https://dl.acm.org/doi/abs/10.1145/2882903. 2882958",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [60] creates a query that would fetch a target data to detect logical bug.",
+      "context_before": "ns and compares them with the results returned from the actual DBMS. TxCheck ,DT2andTroc fail to generate multiple parallel operations with high data access contentions, which is critical for trigger the IL-related code. Thus, they omit the IL bugs found by DBStorm, even for the DBMS they have tested. Query-Oriented Testing. Query-oriented testing aims to generate various queries for DBMS testing. The bugs that have been discovered can be classified into logical bugs [ 21,49,60,63], performance bugs [ 42,51,58], and crash bugs [ 8,69,78]. A logical bug makes a query return an incorrect result.",
+      "context_after": "A performance bug adversely affects the query performance of DBMS. Apollo [42] detects performance bugs by comparing a new DBMS with an old one. A crash bug causes a DBMS to abruptly stop running. Squirrel [78] generates abnormal queries to detect crash bugs. 10 Conclusion In this paper, we presented DBStorm, an efficient approach to generate effective workloads for IL tests. We provided theoretical proofs to illustrate DBStorm ’s effectiveness and efficiency. The experiments show that DBStorm outperforms the existing work in efficiency and effectiveness, and we have found 33 bugs in popular",
+      "section": "8 Discussion",
+      "page": 11,
+      "char_offset": 69000,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:52:31Z",
+    "is_model_written": true,
+    "summary": "DBStorm generates workloads for testing isolation-level implementations. Its argument is that query-oriented tools only exercise the query engine, while existing transaction-oriented tools execute a few operations serially and so build sparse conflict graphs that never contain a cycle. DBStorm instead constructs workloads targeting chosen non-serializable anomalies, using conflict-graph isomorphism to avoid generating redundant ones. It found 20 isolation bugs and 13 others.",
+    "narrative": "SQLancer supplies DBStorm's SQL. The paper states that because SQLancer targets the relational model it is used to generate syntactically and semantically correct SQL operations, and that since SQLancer mainly produces serially executed operations they revised it to generate parallel ones. It is also one of the tools measured against, and the comparison makes the paper's point precisely: SQLsmith, SQLancer and Squirrel between them reveal 760 bugs of which none is an isolation bug, because they execute transactions sequentially in a single thread and so cannot produce a cycle in the conflict graph at all.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "motivation",
+      "M3": "motivation",
+      "M4": "motivation",
+      "M5": "reuse_component",
+      "M6": "background",
+      "M7": "reuse_component",
+      "M8": "reuse_component",
+      "M9": "background",
+      "M10": "baseline",
+      "M11": "result_comparison",
+      "M12": "result_comparison",
+      "M13": "result_comparison",
+      "M14": "result_comparison",
+      "M15": "background",
+      "M16": "background",
+      "M17": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M7",
+          "M8"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M7",
+            "sentence": "As SQLancer [60] targets for the relational model, we leverage it to generate syntax/semantic correct SQL operations.",
+            "section": "3 DBStorm Framework",
+            "page": 4
+          },
+          {
+            "mention_id": "M8",
+            "sentence": "Since SQLancer mainly generates serially executed operations, we revise it to generate parallel operations.",
+            "section": "3 DBStorm Framework",
+            "page": 4
+          }
+        ],
+        "reasoning": "M7 states DBStorm leverages SQLancer to generate syntactically and semantically correct SQL operations, and M8 that they revised it to emit parallel rather than serial operations. SQLancer is the operation generator inside DBStorm, not the framework it is built on -- the workload construction and conflict-graph analysis are the paper's own.",
+        "reuse_kind": "generator"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The revision described in M8 is to SQLancer's operation generation so it can emit parallel operations, not to any of its test oracles. DBStorm's oracle is isolation-level checking over conflict graphs."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M10",
+          "M12",
+          "M14"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M10",
+            "sentence": "SQLancer [60] and TQS [67] randomly generate queries and provide a test oracle to identify logical bugs.",
+            "section": "7 Experiments",
+            "page": 9
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "SQLancer ,Squrriel and SQLsmith sequentially execute transactions in a single thread, so it is impossible to exist a cycle in the conflict graph and we do not put them in Fig.",
+            "section": "7.3 Anomaly-Sensitive Testing",
+            "page": 10
+          },
+          {
+            "mention_id": "M14",
+            "sentence": "SQLsmith /SQLancer /Squirrel reveal 248/449/63 bugs, but none of them is an isolation bug.",
+            "section": "7.4 Bug Result",
+            "page": 11
+          }
+        ],
+        "reasoning": "M10 introduces SQLancer among the tools evaluated, M12 explains why it cannot produce a conflict-graph cycle, and M14 reports its bug count alongside SQLsmith's and Squirrel's, none of which is an isolation bug."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is described by what it generates and by the limits of single-threaded execution; no state-of-the-art claim is made."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3650212_3680392.json b/_data/papers/paper_doi_10_1145_3650212_3680392.json
new file mode 100644
index 0000000..9c8c54e
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3650212_3680392.json
@@ -0,0 +1,1010 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:32:41Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3650212.3680392",
+    "title": "Testing Gremlin-Based Graph Database Systems via Query Disassembling",
+    "authors": [
+      "Yingying Zheng",
+      "Wensheng Dou",
+      "Leile Tang",
+      "Ziyu Cui",
+      "Yu Gao",
+      "Jiansen Song",
+      "Liangying Xu",
+      "Jiaxin Zhu",
+      "Wei Wang",
+      "Jun Wei",
+      "Hua Zhong",
+      "Tao Huang"
+    ],
+    "year": 2024,
+    "venue": "International Symposium on Software Testing and Analysis",
+    "doi": "10.1145/3650212.3680392",
+    "arxiv_id": null,
+    "s2_paper_id": "dce7daf3d485e83ae503cfa60e7a0ccce994a764",
+    "url": "https://doi.org/10.1145/3650212.3680392",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3650212.3680392",
+      "retrieved_at": "2026-09-09T01:32:41Z",
+      "chars": 85333,
+      "content_sha256": "sha256:907c0d3e374eae6906a410f45055043dc523fdc13e7d94ec9a49baa45186b79d"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2725
+      },
+      {
+        "number": "2",
+        "title": "Preliminaries",
+        "start": 11361
+      },
+      {
+        "number": "2.1",
+        "title": "Labeled Property GraphModel",
+        "start": 11611
+      },
+      {
+        "number": "2.2",
+        "title": "Gremlin QueryLanguage",
+        "start": 14629
+      },
+      {
+        "number": "2.3",
+        "title": "Gremlin Traversal Model",
+        "start": 16280
+      },
+      {
+        "number": "2.4",
+        "title": "Gremlin QueryExecution",
+        "start": 18099
+      },
+      {
+        "number": "3",
+        "title": "Approach",
+        "start": 19397
+      },
+      {
+        "number": "3.1",
+        "title": "GraphDatabaseandQueryGeneration",
+        "start": 22655
+      },
+      {
+        "number": "3.2",
+        "title": "QueryDisassembling",
+        "start": 26041
+      },
+      {
+        "number": "3.3",
+        "title": "Atomic Traversal Execution",
+        "start": 31826
+      },
+      {
+        "number": "3.3.1",
+        "title": "Parameter Passing Strategy. In this strategy, we store the",
+        "start": 32958
+      },
+      {
+        "number": "3.3.2",
+        "title": "TemporaryIDTableStrategy. Inthisstrategy,westorethe",
+        "start": 35918
+      },
+      {
+        "number": "3.3.3",
+        "title": "Barrier Strategy. The core insight of this strategy is that the",
+        "start": 38954
+      },
+      {
+        "number": "4",
+        "title": "Evaluation",
+        "start": 40763
+      },
+      {
+        "number": "4.1",
+        "title": "Methodology",
+        "start": 41977
+      },
+      {
+        "number": "4.2",
+        "title": "Detected Bugs",
+        "start": 46707
+      },
+      {
+        "number": "4.2.1",
+        "title": "Bug Overview. We obtain 3,047 bug reportsin the six tested",
+        "start": 46725
+      },
+      {
+        "number": "4.2.2",
+        "title": "BugAnalysis. Weanalyzelogicbugsdetectedby QuDiinthe",
+        "start": 48916
+      },
+      {
+        "number": "4.2.3",
+        "title": "Execution Strategies Comparison. We design three execution",
+        "start": 53154
+      },
+      {
+        "number": "4.3",
+        "title": "Comparingwith ExistingApproaches",
+        "start": 55945
+      },
+      {
+        "number": "4.4",
+        "title": "SelectedInterestingBugs",
+        "start": 62500
+      },
+      {
+        "number": "5",
+        "title": "Discussion",
+        "start": 65610
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2024. AFL. https://github.com/google/AFL.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2024. Alloy. https://alloytools.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2024. DB-Engines Ranking of Graph DBMS. https://db-engines.com/en/ranking/ graph+ dbms.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2024. Gremlin QueryLanguage. https://tinkerpop.apache.org/gremlin.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2024. Gremlin TraversalStrategy. https://tinkerpop.apache.org/docs/3.5.2/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2024. HugeGraph. https://hugegraph.github.io/hugegraph-doc/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2024. Introducing the new Cypher Query Optimizer. https://neo4j.com/blog/ introducing-new-cypher-query-optimizer/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2024. JanusGraph. https://janusgraph.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2024. MariaDB. https://mariadb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2024. MySQL. https://www.mysql.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2024. NebulaGraphQueryLanguage(nGQL). https://docs.nebula-graph.io/2.0. 1/3.ngql-guide/1.nGQL-overview/1.overview/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2024. Neo4j. https://neo4j.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "2024. Neo4j-Gremlin. https://github.com/thinkaurelius/neo4j-gremlin-plugin.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "2024. TheNextGenerationMulti-ModelDatabaseSupportingGraphsKey/Value, Documents and Time-Series. https://arcadedb.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "2024. OpenSource,Distributed,Scalable,Lightning Fast. https://nebula-graph. io/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "2024. OrientDB. https://orientdb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "2024. PostgreSQL. https://www.postgresql.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "2024. SPARQL 1.1QueryLanguage. https://www.w3.org/TR/sparql11-query/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "2024. SQLsmith. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "2024.TestingGremlin-BasedGraphDatabaseSystemsviaQueryDisassembling. Retrieved July18,2024 from https://doi.org/10.5281/zenodo.12771889",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "2024. TiDB, PingCAP. https://pingcap.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "2024. TinkerGraph. https://github.com/tinkerpop/blueprints/wiki/tinkergraph.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "2024. TinkerPop. https://tinkerpop.apache.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "2024. What is openCypher? http://www.opencypher.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Renzo Angles, Juan L. Reutter, and Hannes Voigt. 2019. Graph Query Languages. InEncyclopediaofBigDataTechnologies ,SherifSakrandAlbertY.Zomaya(Eds.).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Marcelo Arenas, Claudio Gutiérrez, and Juan F. Sequeda. 2021. Querying in the Age of Graph Databases and Knowledge Graphs. In Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD). 2821–2828. https://doi.org/10.1145/3448016.3457545",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). 2060–2071. https://doi.org/10.1109/ICSE48619.2023.00174",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "Earl T. Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. 2015. TheOracleProbleminSoftwareTesting:ASurvey. IEEETrans.Software Eng.41,5 (2015), 507–525.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Sha/f_iul Azam Chowdhury, SoumikMohian, Sidharth Mehra, Siddhant Gawsane, Taylor T. Johnson, and Christoph Csallner. 2018. Automatically Finding Bugs in a Commercial Cyber-Physical System Development Tool Chain with SLforge. In ProceedingsofInternationalConferenceonSoftwareEngineering(ICSE) .981–992. https://doi.org/10.1145/3180155.3180231",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Ziyu Cui, Wensheng Dou, Qianwang Dai, Jiansen Song, Wei Wang, Jun Wei, and Dan Ye. 2022. Differentially Testing Database Transactions for Fun and Pro/f_it. In Proceedings of International Conference on Automated Software Engineering (ASE). 35:1–35:12. https://doi.org/10.1145/3551349.3556924",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "ZiyuCui,WenshengDou,YuGao,DongWang,JiansenSong,YingyingZheng, Tao Wang, Rui Yang, Kang Xu, Yixin Hu, Jun Wei, and Tao Huang. 2024. Understanding Transaction Bugs in Database Systems. In Proceedings of the 46th IEEE/ACM International Conference on Software Engineering (ICSE). 163:1–163:13. https://doi.org/10.1145/3597503.3639207",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Alin Deutsch. 2018. Querying Graph Databases with the GSQL Query Language. InProceedingsofSimpósio BrasileirodeBanco deDados(SBBD). 313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Alin Deutsch, Yu Xu, Mingxi Wu, and Victor E. Lee. 2020. Aggregation Support for Modern Graph Analytics in TigerGraph. In Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD). 377–392. https: //doi.org/10.1145/3318464.3386144",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Wensheng Dou, Ziyu Cui, Qianwang Dai, Jiansen Song, Dong Wang, Yu Gao, Wei Wang, Jun Wei, Lei Chen, Hanmo Wang, Hua Zhong, and Tao Huang. 2023. DetectingIsolation Bugs viaTransactionOracleConstruction. In Proceedings of IEEE/ACMInternationalConferenceonSoftwareEngineering(ICSE) .1123–1135. https://doi.org/10.1109/ICSE48619.2023.00101",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "OrriErling,AlexAverbuch,JosepLluísLarriba-Pey,HassanCha/f_i,AndreyGubichev, Arnau Prat-Pérez, Minh-Duc Pham, and Peter A. Boncz. 2015. The LDBC Social NetworkBenchmark: Interactive Workload. In Proceedings of ACM SIGMODInternationalConferenceonManagementofData(SIGMOD) .619–630. https://doi.org/10.1145/2723372.2742786",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and Andrés Taylor.2018. Cypher: AnEvolvingQueryLanguage forPropertyGraphs. InProceedings of ACMSIGMOD International Conference on Management of Data(SIGMOD). 1433–1445. https://doi.org/10.1145/3183713.3190657",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang,RongxinWu,andCharlesZhang.2023. Pinolo:DetectingLogicalBugs inDatabaseManagementSystemswithApproximateQuerySynthesis.In ProceedingsofUSENIXAnnual TechnicalConference (USENIXATC). 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "ZhenzhenHe,JiongYu,andBingleiGuo.2022. ExecutionTimePredictionfor CypherQueriesintheNeo4jDatabaseUsingaLearningApproach. Symmetry 14,1 (2022), 55.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, and Tao Xie. 2023. GDsmith: Detecting Bugs in Cypher Graph Database Engines. In Proceedings of International Symposium on Software Testing and Analysis (ISSTA). 163–174. https://doi.org/10.1145/3597926.3598046",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management SystemsviaInjectiveandSurjectiveGraphQueryTransformation.In Proceedings of the 46th IEEE/ACM International Conference on Software Engineering (ICSE). 46:1–46:12. https://doi.org/10.1109/ICST60714.2024.00012",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woon-Hak Kang. 2019. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. Proceedings of the VLDB Endowment (VLDB) 13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of ACMSIGSOFT InternationalSymposiumonSoftwareTestingandAnalysis(ISSTA) .https://doi. org/10.1145/3597926.3598044",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O. Laleye, and Sarfraz Khurshid. 2008. Query-Aware Test Generation Using a Relational Constraint Solver. In ProceedingsofIEEE/ACMInternationalConferenceonAutomatedSoftwareEngineering (ASE). 238–247. https://doi.org/10.1109/ASE.2008.34",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "ShadiAbdulKhalekandSarfrazKhurshid.2010. AutomatedSQLquerygenerationforsystematictestingofdatabaseengines.In InternationalConferenceon AutomatedSoftwareEngineering(ASE) .329–332. https://doi.org/10.1145/1858996. 1859063",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Baozhu Liu, Xin Wang, Pengkai Liu, Sizhuo Li, Qiang Fu, and Yunpeng Chai. 2021. UniKG: A Uni/f_ied Interoperable Knowledge Graph Database System. In ProceedingsofIEEEInternationalConferenceonDataEngineering(ICDE) .2681– 2684.https://doi.org/10.1109/ICDE51399.2021.00303",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "XinyuLiu,QiZhou,JoyArulraj,andAlessandroOrso.2022. AutomaticDetection of Performance Bugs in Database Systems Using Equivalent Queries. In ProceedingsofInternationalConferenceonSoftwareEngineering(ICSE) .225–236. https://doi.org/10.1145/3510003.3510093",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "QiuyangMang,AoyangFang,BoxiYu,HanfeiChen,andPinjiaHe.2024. Testing GraphDatabaseSystemsviaEquivalentQueryRewriting.In Proceedingsofthe 46thIEEE/ACMInternationalConferenceonSoftwareEngineering(ICSE) .143:1– 143:12.https://doi.org/10.1145/3597503.3639200",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Muhammad Numair Mansur, Maria Christakis, and Valentin Wüstholz. 2021. Metamorphic Testing of Datalog Engines. In Proceedings of ACM Joint European SoftwareEngineeringConferenceandSymposiumontheFoundationsofSoftware Engineering (ESEC/FSE). 639–650. https://doi.org/10.1145/3468264.3468573",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software. Digit. Tech. J. 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Yuxiang Ren, Hao Zhu, Jiawei Zhang, Peng Dai, and Liefeng Bo. 2021. EnsemFDet: An Ensemble Approach to Fraud Detection based on Bipartite Graph. InProceedingsofInternationalConferenceonDataEngineering(ICDE) .2039–2044. https://doi.org/10.1109/ICDE51399.2021.00197",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "ManuelRiggerandZhendongSu.2020. DetectingOptimizationBugsinDatabase EnginesviaNon-OptimizingReferenceEngineConstruction.In Proceedingsof ACM Joint European Software Engineering Conference and Symposium on the FoundationsofSoftwareEngineering(ESEC/FSE) .1140–1152. https://doi.org/10. 1145/3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 52,
+      "text": "ManuelRiggerandZhendongSu.2020. FindingBugsinDatabaseSystemsvia QueryPartitioning. 4,Article211(2020), 30pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 53,
+      "text": "ManuelRiggerandZhendongSu.2020. TestingDatabaseEnginesviaPivoted Query Synthesis. In Proceedings of USENIX Symposium on Operating Systems Designand Implementation(OSDI). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 54,
+      "text": "Marko A. Rodriguez. 2015. The Gremlin Graph Traversal Machine and Language (Invited Talk). In Proceedings of the Symposium on Database Programming Languages (DBPL). 1–10.https://doi.org/10.1145/2815072.2815073",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Donald S. Slutz. 1998. Massive Stochastic Testing of SQL. In Proceedings of InternationalConference onVeryLarge Data Bases (VLDB). 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential QueryExecution.In ProceedingsofIEEE/ACMInternationalConferenceonSoftware Engineering (ICSE). 2072–2084. https://doi.org/10.1109/ICSE48619.2023.00175",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "JiansenSong,WenshengDou,YuGao,ZiyuCui,YingyingZheng,DongWang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction. Proceedings of the VLDBEndowment (VLDB) (2024). 1706 TestingGremlin-BasedGraphDatabase Systems via/Q_uery Disassembling ISSTA ’24, September16–20, 2024,Vienna, Austria",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "RanWang,ZhengyiYang,WenjieZhang,andXueminLin.2020. AnEmpirical StudyonRecentGraphDatabaseSystems.In ProceedingsofInternationalConference on Knowledge Science, Engineering and Management (KSEM). 328–340. https://doi.org/10.1007/978-3-030-55130-8_29",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "MinWu,XingluYi,HuiYu,YuLiu,andYujueWang.2022. NebulaGraph:An Open Source Distributed GraphDatabase. CoRRabs/2206.07278 (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "RuiYang,YingyingZheng,LeiTang,WenshengDou,WeiWang,andJunWei. 2023. RandomizedDifferentialTestingofRDFStores.In ProceedingsofIEEE/ACM InternationalConferenceonSoftwareEngineering(ICSEDemo) .136–140. https: //doi.org/10.1109/ICSE-Companion58688.2023.00041",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and Understanding Bugs in C Compilers. In Proceedings of International Conference on Programming Language Design and Implementation (PLDI). 283–294. https://doi.org/10.1145/1993316.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "YuiYang,YingyingZheng,LeiTang,WenshengDou,WeiWang,andJunWei. 2023. RandomizedDifferentialTestingofRDFStores.In ProceedingsofInternational Conference on Software Engineering: Companion Proceedings (ICSE Companion). 136–140. https://doi.org/10.1109/ICSE-Companion58688.2023.00041",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "YingyingZheng,WenshengDou,LeiTang,ZiyuCui,JiansenSong,ZiyueCheng, WeiWang,JunWei,HuaZhong,andTaoHuang.2024. DifferentialOptimization TestingofGremlin-BasedGraphDatabaseSystems.In Proceedingsofthe17thIEEE InternationalConferenceonSoftwareTesting,Veri/f_icationandValidation(ICST). 25–36.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "YingyingZheng,WenshengDou,YichengWang,ZhengQin,LeiTang,YuGao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing. In Proceedings of International Symposium on Software Testing and Analysis (ISSTA). 302–313. https://doi.org/10.1145/3533767.3534409",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "ZeyangZhuang,PenghuiLi,PingchuanMa,WeiMeng,andShuaiWang.2023. Testing Graph Database Systems via Graph-Aware Metamorphic Relations. Proceedingsofthe VLDBEndowment (VLDB) 17,4 (2023), 836–848. 1707",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 27,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). 2060–2071. https://doi.org/10.1109/ICSE48619.2023.00174",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "prints the DOI of the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management SystemsviaInjectiveandSurjectiveGraphQueryTransformation.In Proceedings of the 46th IEEE/ACM International Conference on Software Engineering (ICSE). 46:1–46:12. https://doi.org/10.1109/ICST60714.2024.00012",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of ACMSIGSOFT InternationalSymposiumonSoftwareTestingandAnalysis(ISSTA) .https://doi. org/10.1145/3597926.3598044",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 51,
+      "text": "ManuelRiggerandZhendongSu.2020. DetectingOptimizationBugsinDatabase EnginesviaNon-OptimizingReferenceEngineConstruction.In Proceedingsof ACM Joint European Software Engineering Conference and Symposium on the FoundationsofSoftwareEngineering(ESEC/FSE) .1140–1152. https://doi.org/10. 1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec"
+      ]
+    },
+    {
+      "number": 52,
+      "text": "ManuelRiggerandZhendongSu.2020. FindingBugsinDatabaseSystemsvia QueryPartitioning. 4,Article211(2020), 30pages.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp"
+      ]
+    },
+    {
+      "number": 53,
+      "text": "ManuelRiggerandZhendongSu.2020. TestingDatabaseEnginesviaPivoted Query Synthesis. In Proceedings of USENIX Symposium on Operating Systems Designand Implementation(OSDI). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "GDBMeter[ 42] adopts Ternary LogicPartitioning (TLP)[52]toderiveagraphqueryintothreedisjointsub-queries based on a randomly generated predicate, and only focuses on detecting predicate-related bugs.",
+      "context_before": "entation for the assembly of multiple /f_iltering conditions, and /f_ixed it quicklyafter we submittedit. Logic bugs in GDBs are likely to go unnoticed by GDB developers,becausewelackaneffectivetestoracleforautomatictestingto verify whether a GDB behaves correctly for a given Gremlin query, thusdetectinglogicbugs.Someapproaches[ 39,62,64]adoptdifferential testing [ 49] to reveal discrepancies among multiple GDBs forthesamegraphdatabaseandgraphqueries.Thecleardrawback ofthistechniqueisthatitcannotdetectthesamebugsthatexist inalltarget GDBs and can onlybe usedto testcommon features intarget GDBs.",
+      "context_after": "Furthermore, some testing approaches [ 27,30,37,51–53,56,57] have proposed new test oracles to effectively /f_ind logic bugs in an individual relational database system. However, these approaches cannot be applied on GremlinbasedGDBsbecausethe procedural Gremlinquerylanguageadopts differentsyntaxesandquery patterns from declarative SQL. We observe that a complex Gremlin query can be disassembled into a sequence of atomic graph traversals (an atomic graph traversal can achieve one-step traversal in the graph database), which shares the equivalent execution semantics as the original query. Inspir",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6792,
+      "found_by_all": [
+        "technique",
+        "citation_marker_project_authored",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 27,30,37,51–53,56,57]",
+      "technique": "qpg",
+      "sentence": "Furthermore, some testing approaches [ 27,30,37,51–53,56,57] have proposed new test oracles to effectively /f_ind logic bugs in an individual relational database system.",
+      "context_before": "racleforautomatictestingto verify whether a GDB behaves correctly for a given Gremlin query, thusdetectinglogicbugs.Someapproaches[ 39,62,64]adoptdifferential testing [ 49] to reveal discrepancies among multiple GDBs forthesamegraphdatabaseandgraphqueries.Thecleardrawback ofthistechniqueisthatitcannotdetectthesamebugsthatexist inalltarget GDBs and can onlybe usedto testcommon features intarget GDBs.GDBMeter[ 42] adopts Ternary LogicPartitioning (TLP)[52]toderiveagraphqueryintothreedisjointsub-queries based on a randomly generated predicate, and only focuses on detecting predicate-related bugs.",
+      "context_after": "However, these approaches cannot be applied on GremlinbasedGDBsbecausethe procedural Gremlinquerylanguageadopts differentsyntaxesandquery patterns from declarative SQL. We observe that a complex Gremlin query can be disassembled into a sequence of atomic graph traversals (an atomic graph traversal can achieve one-step traversal in the graph database), which shares the equivalent execution semantics as the original query. Inspiredbythisobservation,wepropose QueryDisassembling (QuDi), aneffectivetestoracletoreveallogicbugsinindividualGremlinbasedGDBs.Speci/f_ically,givenaGremlinquery /u1D444,wedi",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6990,
+      "cited_reference": {
+        "number": 27,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). 2060–2071. https://doi.org/10.1109/ICSE48619.2023.00174",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "prints the DOI of the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 42]",
+      "technique": null,
+      "sentence": ",Grand[ 64],GDsmith[ 39],RD2[62], and GDBMeter [ 42], can /f_ind bugs in GDBs.",
+      "context_before": "er.However, TempIDisageneralstrategythat canbepotentiallyappliedtotestotherdatabasesystems(Section 5), and can also work with large graph data. Overall, since these three execution strategies construct and execute the disassembled query using different Gremlin features, they have different bug detection capabilities, and thus do not detect exactly the same bugs. Thus, our three executionstrategies can complementeachother. OurthreeexecutionstrategiescancoverdifferentGremlinfeatures and complementeach other,thusdetecting morelogicbugs. 4.3 Comparingwith ExistingApproaches Fourrelatedapproaches,i.e.",
+      "context_after": "Speci/f_ically, Grand, GDsmith, and RD2utilizedifferential testing [49] toreveal discrepancies among multiple target GDBs. GDBMeter utilizes Ternary Logic Partitioning (TLP), an invariant of query partitioning [52], toreveallogicbugsinatarget GDB.Therefore,wecompare QuDi withdifferentialtesting andquerypartitioning .Notethatwecannot compareQuDiwithrelationaldatabasetestingtools,e.g.,SQLancer [51–53],sinceQuDiandSQLancertargetdifferenttypesofdatabase systems, which utilize different query languages (Gremlin vs. SQL) anddata models (graph modelvs. relationalmodel). Comparison with differential testin",
+      "section": "4.3 Comparingwith ExistingApproaches",
+      "page": 9,
+      "char_offset": 56008,
+      "cited_reference": {
+        "number": 42,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of ACMSIGSOFT InternationalSymposiumonSoftwareTestingandAnalysis(ISSTA) .https://doi. org/10.1145/3597926.3598044",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "GDBMeter utilizes Ternary Logic Partitioning (TLP), an invariant of query partitioning [52], toreveallogicbugsinatarget GDB.",
+      "context_before": "t and execute the disassembled query using different Gremlin features, they have different bug detection capabilities, and thus do not detect exactly the same bugs. Thus, our three executionstrategies can complementeachother. OurthreeexecutionstrategiescancoverdifferentGremlinfeatures and complementeach other,thusdetecting morelogicbugs. 4.3 Comparingwith ExistingApproaches Fourrelatedapproaches,i.e.,Grand[ 64],GDsmith[ 39],RD2[62], and GDBMeter [ 42], can /f_ind bugs in GDBs. Speci/f_ically, Grand, GDsmith, and RD2utilizedifferential testing [49] toreveal discrepancies among multiple target GDBs.",
+      "context_after": "Therefore,wecompare QuDi withdifferentialtesting andquerypartitioning .Notethatwecannot compareQuDiwithrelationaldatabasetestingtools,e.g.,SQLancer [51–53],sinceQuDiandSQLancertargetdifferenttypesofdatabase systems, which utilize different query languages (Gremlin vs. SQL) anddata models (graph modelvs. relationalmodel). Comparison with differential testing. Differential testing (e.g.,Grand)requiresmorethanonetarget GDBasinput.IfaGremlinfeatureisnot supported byall target GDBs,differentialtesting willnotbeabletoapplyonthisfeature.Thus,differentialtesting canonlybeusedtotestcommonfeaturesinthetarget GDB",
+      "section": "4.3 Comparingwith ExistingApproaches",
+      "page": 9,
+      "char_offset": 56208,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ",SQLancer [51–53],sinceQuDiandSQLancertargetdifferenttypesofdatabase systems, which utilize different query languages (Gremlin vs.",
+      "context_before": "fferentGremlinfeatures and complementeach other,thusdetecting morelogicbugs. 4.3 Comparingwith ExistingApproaches Fourrelatedapproaches,i.e.,Grand[ 64],GDsmith[ 39],RD2[62], and GDBMeter [ 42], can /f_ind bugs in GDBs. Speci/f_ically, Grand, GDsmith, and RD2utilizedifferential testing [49] toreveal discrepancies among multiple target GDBs. GDBMeter utilizes Ternary Logic Partitioning (TLP), an invariant of query partitioning [52], toreveallogicbugsinatarget GDB.Therefore,wecompare QuDi withdifferentialtesting andquerypartitioning .Notethatwecannot compareQuDiwithrelationaldatabasetestingtools,e.g.",
+      "context_after": "SQL) anddata models (graph modelvs. relationalmodel). Comparison with differential testing. Differential testing (e.g.,Grand)requiresmorethanonetarget GDBasinput.IfaGremlinfeatureisnot supported byall target GDBs,differentialtesting willnotbeabletoapplyonthisfeature.Thus,differentialtesting canonlybeusedtotestcommonfeaturesinthetarget GDBs.We can encounter the following casesindifferential testing. •C1: At least two GDBs can return inconsistent results, then differential testingreports abug. •C2:AllGDBsreturnthesamewrongresults,thendifferential testingcan miss abug. •C3:AllGDBsreturnthesameresults.How",
+      "section": "4.3 Comparingwith ExistingApproaches",
+      "page": 9,
+      "char_offset": 56469,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "Comparison with query partitioning.",
+      "context_before": "ally, foranoriginalGremlinquerythattriggersabuginthese21bug reports,wedisassembleitintoatomicgraphtraversalsandverify whethertheycancomputedifferentqueryresultsastheoriginal query. If yes, we consider that QuDican detect the logic bug. We /f_indthat, QuDican/f_ind16(76%)logicbugsdetectedbyGrand. QuDi misses the remaining 5 bugs because their original queries and the corresponding disassembledqueriesreturnthesamewrongresults. This result indicates that QuDihas the capability to detect most logic bugsdetectedbyGrandwithoutsufferingfromGrand’sdrawbacks, e.g.,multiple GDBsas inputand false positives.",
+      "context_after": "In query partitioning (e.g., GDBMeter), a given query /u1D444can derive multiple disjoint subqueries (e.g., /u1D444′ /u1D45D=/u1D447/u1D445/u1D448/u1D438,/u1D444′ /u1D45D=/u1D439/u1D434/u1D43F/u1D446/u1D438, and/u1D444′ /u1D45D /u1D456/u1D460 /u1D441/u1D448/u1D43F/u1D43Fbased on a random predicate /u1D45Din TLP). These individual partitions are composed to obtain a result set /u1D445/u1D446/u1D444′, which should be equal to /u1D445/u1D446/u1D444. However,querypartitioningcannotprevent GDBoptimizationsfrom kickinginandrevealassemblyissuesin GDBs,sothatitcanhardly detect logic bugs that are cause",
+      "section": "4.3 Comparingwith ExistingApproaches",
+      "page": 9,
+      "char_offset": 59303,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "In query partitioning (e.",
+      "context_before": "riggersabuginthese21bug reports,wedisassembleitintoatomicgraphtraversalsandverify whethertheycancomputedifferentqueryresultsastheoriginal query. If yes, we consider that QuDican detect the logic bug. We /f_indthat, QuDican/f_ind16(76%)logicbugsdetectedbyGrand. QuDi misses the remaining 5 bugs because their original queries and the corresponding disassembledqueriesreturnthesamewrongresults. This result indicates that QuDihas the capability to detect most logic bugsdetectedbyGrandwithoutsufferingfromGrand’sdrawbacks, e.g.,multiple GDBsas inputand false positives. Comparison with query partitioning.",
+      "context_after": "g., GDBMeter), a given query /u1D444can derive multiple disjoint subqueries (e.g., /u1D444′ /u1D45D=/u1D447/u1D445/u1D448/u1D438,/u1D444′ /u1D45D=/u1D439/u1D434/u1D43F/u1D446/u1D438, and/u1D444′ /u1D45D /u1D456/u1D460 /u1D441/u1D448/u1D43F/u1D43Fbased on a random predicate /u1D45Din TLP). These individual partitions are composed to obtain a result set /u1D445/u1D446/u1D444′, which should be equal to /u1D445/u1D446/u1D444. However,querypartitioningcannotprevent GDBoptimizationsfrom kickinginandrevealassemblyissuesin GDBs,sothatitcanhardly detect logic bugs that are caused by incorrect implementat",
+      "section": "4.3 Comparingwith ExistingApproaches",
+      "page": 9,
+      "char_offset": 59339,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", /u1D444′ /u1D45D=/u1D447/u1D445/u1D448/u1D438,/u1D444′ /u1D45D=/u1D439/u1D434/u1D43F/u1D446/u1D438, and/u1D444′ /u1D45D /u1D456/u1D460 /u1D441/u1D448/u1D43F/u1D43Fbased on a random predicate /u1D45Din TLP).",
+      "context_before": "rentqueryresultsastheoriginal query. If yes, we consider that QuDican detect the logic bug. We /f_indthat, QuDican/f_ind16(76%)logicbugsdetectedbyGrand. QuDi misses the remaining 5 bugs because their original queries and the corresponding disassembledqueriesreturnthesamewrongresults. This result indicates that QuDihas the capability to detect most logic bugsdetectedbyGrandwithoutsufferingfromGrand’sdrawbacks, e.g.,multiple GDBsas inputand false positives. Comparison with query partitioning. In query partitioning (e.g., GDBMeter), a given query /u1D444can derive multiple disjoint subqueries (e.g.",
+      "context_after": "These individual partitions are composed to obtain a result set /u1D445/u1D446/u1D444′, which should be equal to /u1D445/u1D446/u1D444. However,querypartitioningcannotprevent GDBoptimizationsfrom kickinginandrevealassemblyissuesin GDBs,sothatitcanhardly detect logic bugs that are caused by incorrect implementations and optimizationsoftheassemblyofatomic graphtraversals,andour query disassembling iscomplementary to query partitioning. We /f_irst verify whether query partitioning in GDBMeter can detect the 25 logic bugs detected by QuDi. For each test case in our 25bugreports,wetrytoconstructdisj",
+      "section": "4.3 Comparingwith ExistingApproaches",
+      "page": 9,
+      "char_offset": 59446,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "However,querypartitioningcannotprevent GDBoptimizationsfrom kickinginandrevealassemblyissuesin GDBs,sothatitcanhardly detect logic bugs that are caused by incorrect implementations and optimizationsoftheassemblyofatomic graphtraversals,andour query disassembling iscomplementary to query partitioning.",
+      "context_before": "most logic bugsdetectedbyGrandwithoutsufferingfromGrand’sdrawbacks, e.g.,multiple GDBsas inputand false positives. Comparison with query partitioning. In query partitioning (e.g., GDBMeter), a given query /u1D444can derive multiple disjoint subqueries (e.g., /u1D444′ /u1D45D=/u1D447/u1D445/u1D448/u1D438,/u1D444′ /u1D45D=/u1D439/u1D434/u1D43F/u1D446/u1D438, and/u1D444′ /u1D45D /u1D456/u1D460 /u1D441/u1D448/u1D43F/u1D43Fbased on a random predicate /u1D45Din TLP). These individual partitions are composed to obtain a result set /u1D445/u1D446/u1D444′, which should be equal to /u1D445/u1D446/u1D444.",
+      "context_after": "We /f_irst verify whether query partitioning in GDBMeter can detect the 25 logic bugs detected by QuDi. For each test case in our 25bugreports,wetrytoconstructdisjointsub-queriesaccording tothe TLPoracleandcheckwhetherthesedisjointsub-queriescan reportaninconsistency.We/f_indthat,only6bugsdetectedby QuDi can be found by GDBMeter theoretically. Among these six bugs, 5 bugsareidenti/f_iedbyunexpectedexceptionsindisjointsub-queries and the remaining one is identi/f_ied by incorrect query results. The remaining19bugscouldnotbedetectedbyGDBMeterbecausethe original queries in these test cases can co",
+      "section": "4.3 Comparingwith ExistingApproaches",
+      "page": 9,
+      "char_offset": 59790,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "We /f_irst verify whether query partitioning in GDBMeter can detect the 25 logic bugs detected by QuDi.",
+      "context_before": "438,/u1D444′ /u1D45D=/u1D439/u1D434/u1D43F/u1D446/u1D438, and/u1D444′ /u1D45D /u1D456/u1D460 /u1D441/u1D448/u1D43F/u1D43Fbased on a random predicate /u1D45Din TLP). These individual partitions are composed to obtain a result set /u1D445/u1D446/u1D444′, which should be equal to /u1D445/u1D446/u1D444. However,querypartitioningcannotprevent GDBoptimizationsfrom kickinginandrevealassemblyissuesin GDBs,sothatitcanhardly detect logic bugs that are caused by incorrect implementations and optimizationsoftheassemblyofatomic graphtraversals,andour query disassembling iscomplementary to query partitioning.",
+      "context_after": "For each test case in our 25bugreports,wetrytoconstructdisjointsub-queriesaccording tothe TLPoracleandcheckwhetherthesedisjointsub-queriescan reportaninconsistency.We/f_indthat,only6bugsdetectedby QuDi can be found by GDBMeter theoretically. Among these six bugs, 5 bugsareidenti/f_iedbyunexpectedexceptionsindisjointsub-queries and the remaining one is identi/f_ied by incorrect query results. The remaining19bugscouldnotbedetectedbyGDBMeterbecausethe original queries in these test cases can compute the same wrong queryresultsastheunionqueryresultsoftheirdisjointsub-queries. We further verify whe",
+      "section": "4.3 Comparingwith ExistingApproaches",
+      "page": 9,
+      "char_offset": 60090,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": ", query partitioning and query disassembling).",
+      "context_before": "dby GDBMeterbecausethe original queries in these test cases can compute the same wrong queryresultsastheunionqueryresultsoftheirdisjointsub-queries. We further verify whether QuDican detect the 3 logic bugs reported by GDBMeter in JanusGraph, which adopt Gremlin as the querylanguage.Weignoretheother36bugsreportedbyGDBMeter since their test cases use Cypher queries, which we do not support fornow.Wetrytoruneachtestcaseinthese3bugreportsvia QuDi, and check whether QuDican reveal them. However, 2 of 3 bugs are internal errors that are triggered by query generation instead of the test oracles (e.g.",
+      "context_after": "1703 ISSTA ’24, September16–20, 2024,Vienna, Austria Y. Zheng,W. Dou, L.Tang,Z.Cui,Y. Gao, J. Song,L.Xu,J. Zhu, W. Wang,J. Wei, H.Zhong,andT. Huang 1v1 = g.addV( 'vL').next(); 2v2 = g.addV( 'vL').next(); 3e1 = g.addE( 'eL').from(v1).to(v2).next(); 4g.E(e1).property( 'p',newFloat(0.94461)); 5 6// original query 7g.E().has( 'p', 0.94461); -- e:{1} ✘ 8 9// disassembled queries 10g.E(); -- e:{1} 11g.E(1).has( 'p', 0.94461); -- {} ✔ Figure 9: JanusGraph inconsistently retrieves properties without explicit data type inJanusGraph#3200. 1v1 = g.addV( 'vL').property( 'p0', 1).next(); 2v2 = g.addV( 'vL",
+      "section": "4.3 Comparingwith ExistingApproaches",
+      "page": 9,
+      "char_offset": 61225,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[37,51,52,56,57]",
+      "technique": "norec",
+      "sentence": "Somemetamorphictestingapproaches [37,51,52,56,57]for/f_indingbugsinrelationaldatabasesystems have beenproposed.",
+      "context_before": "testing. DT2[30] utilizes differential testing to /f_ind transaction bugs in relational database systems. To test GDBs, Grand [ 64], GDsmith [39], and RD2[60] utilize differential testing to /f_ind logic bugs in multiple GDBs with Gremlin, Cypher, and SPARQL query languages,respectively.However,differentialtestingcannotprovide a testing oracle for a target GDB, and can miss logic bugs and encounter false positives dueto target GDBs’specialsemantics. Metamorphic testing. Metamorphic testing [ 48] aims to address the test oracle problem [ 28] by mutating test cases according tometamorphicrelations.",
+      "context_after": "For example,query partitioning[ 52]derives a given querytomultiple disjointsub-queries,in whichtheresult of the given query is the same as the combination result of disjoint sub-queries. NoREC [ 51] transforms an optimized SQL query toanon-optimized SQLquery,andcan/f_indoptimizationbugsin relational database systems. However, all the above approaches targetdatabasesystemswith declarative SQLquerylanguage.They cannot effectivelytestGremlin-based GDBs becauseGremlinand SQL have totallydifferentsyntaxesandquery patterns.TotestGDBs,GDBMeter[ 42]derivesagraphqueryintothree disjoint sub-queries by rando",
+      "section": "5 Discussion",
+      "page": 11,
+      "char_offset": 69555,
+      "cited_reference": {
+        "number": 51,
+        "text": "ManuelRiggerandZhendongSu.2020. DetectingOptimizationBugsinDatabase EnginesviaNon-OptimizingReferenceEngineConstruction.In Proceedingsof ACM Joint European Software Engineering Conference and Symposium on the FoundationsofSoftwareEngineering(ESEC/FSE) .1140–1152. https://doi.org/10. 1145/3368089.340",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "For example,query partitioning[ 52]derives a given querytomultiple disjointsub-queries,in whichtheresult of the given query is the same as the combination result of disjoint sub-queries.",
+      "context_before": "t GDBs, Grand [ 64], GDsmith [39], and RD2[60] utilize differential testing to /f_ind logic bugs in multiple GDBs with Gremlin, Cypher, and SPARQL query languages,respectively.However,differentialtestingcannotprovide a testing oracle for a target GDB, and can miss logic bugs and encounter false positives dueto target GDBs’specialsemantics. Metamorphic testing. Metamorphic testing [ 48] aims to address the test oracle problem [ 28] by mutating test cases according tometamorphicrelations.Somemetamorphictestingapproaches [37,51,52,56,57]for/f_indingbugsinrelationaldatabasesystems have beenproposed.",
+      "context_after": "NoREC [ 51] transforms an optimized SQL query toanon-optimized SQLquery,andcan/f_indoptimizationbugsin relational database systems. However, all the above approaches targetdatabasesystemswith declarative SQLquerylanguage.They cannot effectivelytestGremlin-based GDBs becauseGremlinand SQL have totallydifferentsyntaxesandquery patterns.TotestGDBs,GDBMeter[ 42]derivesagraphqueryintothree disjoint sub-queries by randomly generating a predicate, which mainly focuses on predicate-related bugs. Gamera [ 65] designs threetypesofgraph-awaremetamorphicrelations,whichcanbe usedtogeneratediverseandcomplexgra",
+      "section": "5 Discussion",
+      "page": 11,
+      "char_offset": 69666,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [ 51] transforms an optimized SQL query toanon-optimized SQLquery,andcan/f_indoptimizationbugsin relational database systems.",
+      "context_before": "erentialtestingcannotprovide a testing oracle for a target GDB, and can miss logic bugs and encounter false positives dueto target GDBs’specialsemantics. Metamorphic testing. Metamorphic testing [ 48] aims to address the test oracle problem [ 28] by mutating test cases according tometamorphicrelations.Somemetamorphictestingapproaches [37,51,52,56,57]for/f_indingbugsinrelationaldatabasesystems have beenproposed.For example,query partitioning[ 52]derives a given querytomultiple disjointsub-queries,in whichtheresult of the given query is the same as the combination result of disjoint sub-queries.",
+      "context_after": "However, all the above approaches targetdatabasesystemswith declarative SQLquerylanguage.They cannot effectivelytestGremlin-based GDBs becauseGremlinand SQL have totallydifferentsyntaxesandquery patterns.TotestGDBs,GDBMeter[ 42]derivesagraphqueryintothree disjoint sub-queries by randomly generating a predicate, which mainly focuses on predicate-related bugs. Gamera [ 65] designs threetypesofgraph-awaremetamorphicrelations,whichcanbe usedtogeneratediverseandcomplexgraphqueries,andthenreveal logic bugs in GDBs. Mang et al. [ 47] propose equivalent query rewriting (EQR), which rewrites a graph quer",
+      "section": "5 Discussion",
+      "page": 11,
+      "char_offset": 69852,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 42]",
+      "technique": null,
+      "sentence": "TotestGDBs,GDBMeter[ 42]derivesagraphqueryintothree disjoint sub-queries by randomly generating a predicate, which mainly focuses on predicate-related bugs.",
+      "context_before": "es [37,51,52,56,57]for/f_indingbugsinrelationaldatabasesystems have beenproposed.For example,query partitioning[ 52]derives a given querytomultiple disjointsub-queries,in whichtheresult of the given query is the same as the combination result of disjoint sub-queries. NoREC [ 51] transforms an optimized SQL query toanon-optimized SQLquery,andcan/f_indoptimizationbugsin relational database systems. However, all the above approaches targetdatabasesystemswith declarative SQLquerylanguage.They cannot effectivelytestGremlin-based GDBs becauseGremlinand SQL have totallydifferentsyntaxesandquery patterns.",
+      "context_after": "Gamera [ 65] designs threetypesofgraph-awaremetamorphicrelations,whichcanbe usedtogeneratediverseandcomplexgraphqueries,andthenreveal logic bugs in GDBs. Mang et al. [ 47] propose equivalent query rewriting (EQR), which rewrites a graph query into equivalent graph queries that trigger distinct query plans, to detect bugs in GDBs. Given a graph query /u1D444, GraphGenie [ 40] derives a mutated graphquerywhosequeryresultsetiseithersemanticallyequivalent totheresultsetof /u1D444orconstitutesasubsetorsupersetoftheresult set of/u1D444, dependingon the mutationapplied. DOT [ 63] candetect optimizat",
+      "section": "5 Discussion",
+      "page": 11,
+      "char_offset": 70185,
+      "cited_reference": {
+        "number": 42,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. In Proceedings of ACMSIGSOFT InternationalSymposiumonSoftwareTestingandAnalysis(ISSTA) .https://doi. org/10.1145/3597926.3598044",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 40]",
+      "technique": null,
+      "sentence": "Given a graph query /u1D444, GraphGenie [ 40] derives a mutated graphquerywhosequeryresultsetiseithersemanticallyequivalent totheresultsetof /u1D444orconstitutesasubsetorsupersetoftheresult set of/u1D444, dependingon the mutationapplied.",
+      "context_before": "They cannot effectivelytestGremlin-based GDBs becauseGremlinand SQL have totallydifferentsyntaxesandquery patterns.TotestGDBs,GDBMeter[ 42]derivesagraphqueryintothree disjoint sub-queries by randomly generating a predicate, which mainly focuses on predicate-related bugs. Gamera [ 65] designs threetypesofgraph-awaremetamorphicrelations,whichcanbe usedtogeneratediverseandcomplexgraphqueries,andthenreveal logic bugs in GDBs. Mang et al. [ 47] propose equivalent query rewriting (EQR), which rewrites a graph query into equivalent graph queries that trigger distinct query plans, to detect bugs in GDBs.",
+      "context_after": "DOT [ 63] candetect optimization bugs by testing a graph query with two different optimization con/f_igurations. However, these approaches cannot reveal assembly issuesin GDBs.Our approach proposesa newmetamorphic rule to detect logic bugs related to incorrect implementations andoptimizationsoftheassemblyofatomicgraphtraversals.Thus, our approach iscomplementary to theseapproaches. Othertestingapproachesofrelationaldatabasesystems. SQLsmith [ 19] can test relational database systems by randomly generating SQL queries. The generic fuzzing tools (e.g., AFL [ 1]) can also be used to test relation",
+      "section": "5 Discussion",
+      "page": 11,
+      "char_offset": 70673,
+      "cited_reference": {
+        "number": 40,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management SystemsviaInjectiveandSurjectiveGraphQueryTransformation.In Proceedings of the 46th IEEE/ACM International Conference on Software Engineering (ICSE). ",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS[ 53]teststhecorrectness ofrelationaldatabasesystemsbyrandomlyselectingapivotrow and generating random SQL queries that contain the selected row.",
+      "context_before": "aches cannot reveal assembly issuesin GDBs.Our approach proposesa newmetamorphic rule to detect logic bugs related to incorrect implementations andoptimizationsoftheassemblyofatomicgraphtraversals.Thus, our approach iscomplementary to theseapproaches. Othertestingapproachesofrelationaldatabasesystems. SQLsmith [ 19] can test relational database systems by randomly generating SQL queries. The generic fuzzing tools (e.g., AFL [ 1]) can also be used to test relational database systems. ADUSA [ 43, 44] uses a relational constraint solver, Alloy [ 2], to generate the expectedresultofagiven SQLquery.",
+      "context_after": "QPG [27] can improve testing efficiency by exploring a variety ofuniquequeryplans.Troc[ 34]detectstransactionbugs[ 31]by inferring the expected results of concurrent transaction executions. However,theseapproachescannotbedirectlyappliedtotest GDBs. 7 Conclusion Buggy implementations and optimizations of Gremlin-based graph database systems can introduce logic bugs, which can lead to severeconsequences,e.g.,incorrectqueryresults.Inthispaper,we proposequerydisassembling( QuDi)toreveallogicbugsingraph databasesystemsbydisassemblingacomplexGremlinqueryinto anequivalentatomicgraphtraversalsequence.We",
+      "section": "5 Discussion",
+      "page": 11,
+      "char_offset": 71644,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG [27] can improve testing efficiency by exploring a variety ofuniquequeryplans.",
+      "context_before": "ptimizationsoftheassemblyofatomicgraphtraversals.Thus, our approach iscomplementary to theseapproaches. Othertestingapproachesofrelationaldatabasesystems. SQLsmith [ 19] can test relational database systems by randomly generating SQL queries. The generic fuzzing tools (e.g., AFL [ 1]) can also be used to test relational database systems. ADUSA [ 43, 44] uses a relational constraint solver, Alloy [ 2], to generate the expectedresultofagiven SQLquery.PQS[ 53]teststhecorrectness ofrelationaldatabasesystemsbyrandomlyselectingapivotrow and generating random SQL queries that contain the selected row.",
+      "context_after": "Troc[ 34]detectstransactionbugs[ 31]by inferring the expected results of concurrent transaction executions. However,theseapproachescannotbedirectlyappliedtotest GDBs. 7 Conclusion Buggy implementations and optimizations of Gremlin-based graph database systems can introduce logic bugs, which can lead to severeconsequences,e.g.,incorrectqueryresults.Inthispaper,we proposequerydisassembling( QuDi)toreveallogicbugsingraph databasesystemsbydisassemblingacomplexGremlinqueryinto anequivalentatomicgraphtraversalsequence.Weevaluate QuDi onsixwidely-usedgraphdatabasesystems,andhavedetected25 uniquelogicb",
+      "section": "5 Discussion",
+      "page": 11,
+      "char_offset": 71792,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:50:48Z",
+    "is_model_written": true,
+    "summary": "QuDi tests Gremlin-based graph databases by disassembling a traversal into its atomic steps and checking that reassembling them reproduces the original result. The bugs it targets are in how a system implements and optimises the assembly of traversals, which a predicate-based oracle cannot reach because it varies the predicate rather than the traversal structure. It found 25 logic bugs across Gremlin-based systems.",
+    "narrative": "Query partitioning is the technique QuDi positions itself against, reached through GDBMeter, which the paper describes as adopting TLP to derive a graph query into three disjoint sub-queries and so focusing on predicate-related bugs. The comparison is run: the authors check whether GDBMeter's query partitioning can detect the 25 bugs QuDi found. The conclusion is complementarity rather than replacement -- query partitioning cannot stop the system's optimisations from kicking in and so cannot expose assembly issues, and the paper says the two are complementary. SQLancer itself is explicitly not compared against, since QuDi targets Gremlin and SQLancer targets SQL.",
+    "roles": {
+      "M1": "definition",
+      "M2": "background",
+      "M3": "baseline",
+      "M4": "definition",
+      "M5": "incidental",
+      "M6": "baseline",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "definition",
+      "M12": "background",
+      "M13": "definition",
+      "M14": "definition",
+      "M15": "definition",
+      "M16": "background",
+      "M17": "definition",
+      "M18": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M5 states QuDi and SQLancer target different database systems with different query languages, and that no comparison with SQLancer was run; nothing indicates reuse of its code either."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Query disassembling operates on the structure of a traversal rather than on a predicate, and M9 calls it complementary to query partitioning rather than a development of it."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M3",
+          "M6",
+          "M10"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": ",Grand[ 64],GDsmith[ 39],RD2[62], and GDBMeter [ 42], can /f_ind bugs in GDBs.",
+            "section": "4.3 Comparingwith ExistingApproaches",
+            "page": 9
+          },
+          {
+            "mention_id": "M6",
+            "sentence": "Comparison with query partitioning.",
+            "section": "4.3 Comparingwith ExistingApproaches",
+            "page": 9
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "We /f_irst verify whether query partitioning in GDBMeter can detect the 25 logic bugs detected by QuDi.",
+            "section": "4.3 Comparingwith ExistingApproaches",
+            "page": 9
+          }
+        ],
+        "reasoning": "M3 and M6 name GDBMeter among the tools QuDi is compared with, and M10 states the authors verified whether its query partitioning could detect the 25 bugs QuDi found. The comparison reaches TLP through GDBMeter, which M4 identifies as a TLP-based tool.",
+        "techniques": [
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The techniques are described by what they do and where they fall short; no state-of-the-art claim is made."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3689031_3696064.json b/_data/papers/paper_doi_10_1145_3689031_3696064.json
new file mode 100644
index 0000000..59887d8
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3689031_3696064.json
@@ -0,0 +1,884 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:11Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3689031.3696064",
+    "title": "Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs",
+    "authors": [
+      "Jingzhou Fu",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Yanyang Zhao",
+      "Shanshan Li",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "European Conference on Computer Systems",
+    "doi": "10.1145/3689031.3696064",
+    "arxiv_id": null,
+    "s2_paper_id": "46c6d9b57f9ef11a0f1dd4eae995265c3478194a",
+    "url": "https://doi.org/10.1145/3689031.3696064",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3689031.3696064",
+      "retrieved_at": "2026-09-09T01:31:11Z",
+      "chars": 90248,
+      "content_sha256": "sha256:deaa3ac8022767694a3d9572ea0bd82523ab1170e4f0d57354917f999ed795f5"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 16,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2894
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 11260
+      },
+      {
+        "number": "3",
+        "title": "Methodology",
+        "start": 17192
+      },
+      {
+        "number": "4",
+        "title": "General Findings of SQL Function Bugs",
+        "start": 21006
+      },
+      {
+        "number": "4.1",
+        "title": "Occurrence Stages",
+        "start": 21417
+      },
+      {
+        "number": "4.2",
+        "title": "Types of SQL Functions",
+        "start": 23272
+      },
+      {
+        "number": "4.3",
+        "title": "Prerequisite SQL Statements",
+        "start": 26844
+      },
+      {
+        "number": "5",
+        "title": "Root Causes of SQL Function Bugs",
+        "start": 28351
+      },
+      {
+        "number": "5.1",
+        "title": "Boundary Literal Values",
+        "start": 29907
+      },
+      {
+        "number": "5.2",
+        "title": "Boundary Type Castings",
+        "start": 33403
+      },
+      {
+        "number": "5.3",
+        "title": "Boundary Results of Nested Functions",
+        "start": 36450
+      },
+      {
+        "number": "5.4",
+        "title": "Other Root Causes",
+        "start": 39574
+      },
+      {
+        "number": "6",
+        "title": "Boundary Value Generation Patterns",
+        "start": 41320
+      },
+      {
+        "number": "7",
+        "title": "Pattern-Based Bug Detection",
+        "start": 48158
+      },
+      {
+        "number": "7.1",
+        "title": "Implementation",
+        "start": 48555
+      },
+      {
+        "number": "7.2",
+        "title": "Testing Setup",
+        "start": 53843
+      },
+      {
+        "number": "7.3",
+        "title": "Detected DBMS Vulnerabilities",
+        "start": 54426
+      },
+      {
+        "number": "7.4",
+        "title": "Bugs of Each Pattern",
+        "start": 57610
+      },
+      {
+        "number": "7.5",
+        "title": "Comparison with Other Testing Works",
+        "start": 64435
+      },
+      {
+        "number": "8",
+        "title": "Discussion",
+        "start": 68173
+      },
+      {
+        "number": "9",
+        "title": "Related Work",
+        "start": 71145
+      },
+      {
+        "number": "10",
+        "title": "Conclusion",
+        "start": 77608
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Khaled Alnawasreh, Patrizio Pelliccione, Zhenxiao Hao, Mårten Rånge, and Antonia Bertolino. 2017. Online robustness testing of distributed embedded systems: An industrial approach. In 2017 IEEE/ACM 39th International Conference on Software Engineering: Software Engineering in Practice Track (ICSE-SEIP). IEEE, 133–142.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Peng Chen and Hao Chen. 2018. Angora: Efficient fuzzing by principled search. In IEEE Symposium on Security and Privacy (S&P).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Yuanliang Chen, Fuchen Ma, Yuanhang Zhou, Ming Gu, Qing Liao, and Yu Jiang. 2024. Chronos: Finding Timeout Bugs in Practical Distributed Systems by Deep-Priority Fuzzing with Transient Delay. In IEEE Symposium on Security and Privacy (S&P). IEEE, 1939–1955.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Lori A. Clarke, Johnette Hassell, and Debra J. Richardson. 1982. A close look at domain testing. IEEE Transactions on Software Engineering 4 (1982), 380–390.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "MariaDB Corporation. 2024. MariaDB. https://mariadb .org/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "MariaDB Corporation. 2024. MariaDB’s JIRA. https://jira .mariadb .org. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "MariaDB Corporation. 2024. MDEV-11030. https://jira .mariadb .org/ browse/MDEV-11030. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "MariaDB Corporation. 2024. MDEV-14596. https://jira .mariadb .org/ browse/MDEV-14596. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "MariaDB Corporation. 2024. MDEV-23415. https://jira .mariadb .org/ browse/MDEV-23415. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "MariaDB Corporation. 2024. MDEV-32315. https://jira .mariadb .org/ browse/MDEV-32315. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "MariaDB Corporation. 2024. MDEV-8407. https://jira .mariadb .org/ browse/MDEV-8407. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Oracle Corporation. 2024. MySQL Bug System. https:// bugs .mysql .com/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Domenico Cotroneo, Domenico Di Leo, Francesco Fucci, and Roberto Natella. 2013. Sabrine: State-based robustness testing of operating systems. In 2013 28th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 125–135.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Domenico Cotroneo, Antonio Ken Iannillo, Roberto Natella, and Stefano Rosiello. 2019. Dependability assessment of the Android OS through fault injection. IEEE Transactions on Reliability 70, 1 (2019), 346–361.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Wenqian Deng, Jie Liang, Zhiyong Wu, Jigzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. CONI: Detecting Database Connector Bugs via State-Aware Test Case Generation. In Proceedings of the IEEE/ACM 47th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "DuckDB 2024. DuckDB Website. https://duckdb .org/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Gordon Fraser and Andrea Arcuri. 2011. Evosuite: automatic test suite generation for object-oriented software. In Proceedings of the 19th ACMSIGSOFT symposium and the 13th European conference on Foundations of software engineering. 416–419.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, and Yu Jiang. 2024. Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free DBMS fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Ying Fu, Zhiyong Wu, Yuanliang Zhang, Jie Liang, Jingzhou Fu, Yu Jiang, Shanshan Li, and Xiangke Liao. 2025. THANOS: DBMS Bug Detection via Storage Engine Rotation Based Differential Testing. In Proceedings of the IEEE/ACM 47th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "GitHub. 2024. A Bug of DecimalString Function. https://github .com/ ClickHouse/ClickHouse/issues/52407. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "GitHub. 2024. Decimal in MariaDB. https://mariadb .com/kb/en/ decimal/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "The PostgreSQL Global Development Group. 2024. Chapter 9. Functions and Operators. https://www .postgresql .org/docs/current/ functions .html. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "The PostgreSQL Global Development Group. 2024. PostgreSQL Bug Report Mailing List. https://www .postgresql .org/list/pgsql-bugs/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "The PostgreSQL Global Development Group. 2024. PostgreSQL Security Information. https://www .postgresql .org/support/security/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "The PostgreSQL Global Development Group. 2024. Unchecked JSON Input Can Crash the Server. https://www .postgresql .org/support/ security/CVE-2015-5289/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "The PostgreSQL Global Development Group. 2024. Unchecked Regex Can Crash the Server. https://www .postgresql .org/support/security/ CVE-2016-0773/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "The PostgreSQL Global Development Group. 2024. User-Defined Functions. https://www .postgresql .org/docs/current/xfunc .html. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Casidhe Hutchison, Milda Zizyte, Patrick E Lanigan, David Guttendorf, Michael Wagner, Claire Le Goues, and Philip Koopman. 2018. Robustness testing of autonomy software. In Proceedings of the 40th International Conference on Software Engineering: Software Engineering in Practice. 276–285.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "ClickHouse Inc. 2024. ClickHouse Website. https://clickhouse .com/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "ClickHouse Inc. 2024. Type Conversion Functions. https: //clickhouse .com/docs/en/sql-reference/functions/type-conversionfunctions. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2020. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB). Tokyo, Japan.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-oriented DBMS fuzzing. In Proceedings of IEEE International Conference on Data Engineering (ICDE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jie Liang, Mingzhe Wang, Chijin Zhou, Zhiyong Wu, Yu Jiang, Jianzhong Liu, Zhe Liu, and Jiaguang Sun. 2022. Pata: Fuzzing with path aware taint analysis. In Privacy (S&P). IEEE, 1–17.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Jie Liang, Mingzhe Wang, Chijin Zhou, Zhiyong Wu, Jianzhong Liu, and Yu Jiang. 2024. Dodrio: Parallelizing Taint Analysis Based Fuzzing via Redundancy-Free Scheduling. In Companion Proceedings of the 32nd ACM International Conference on the Foundations of Software Engineering. 244–254.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Yiyuan Bai, Qiang Zhang, and Yu Jiang. 2024. WingFuzz: Implementing Continuous Fuzzing for DBMSs. In 2024 USENIX Annual Technical Conference (USENIX ATC 24). 479–492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS Bugs via ConfigurationBased Equivalent Transformation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "LibFuzzer 2024. LibFuzzer. https://www .llvm .org/docs/LibFuzzer .html. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. In Proceedings of the 44th International Conference on Software Engineering. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "MonetDB 2024. The Database System to speed up your Analytical Jobs. https://www .monetdb .org/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Roberto Natella, Stefan Winter, Domenico Cotroneo, and Neeraj Suri. 2018. Analyzing the effects of bugs on software interfaces. IEEE 1075 EuroSys ’25, March 30–April 3, 2025, Rotterdam, Netherlands Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang Transactions on Software Engineering 46, 3 (2018), 280–301.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Openlink. 2024. Virtuoso Open-Source Edition. https:// vos .openlinksw .com/owiki/wiki/VOS. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Oracle. 2024. Chapter 6 Adding Functions to MySQL. https://dev .mysql .com/doc/extending-mysql/8 .3/en/addingfunctions .html. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Oracle. 2024. Functions and Operators. https://dev .mysql .com/doc/ refman/8 .0/en/functions .html. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Oracle 2024. MySQL. https://www .mysql .com/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Carlos Pacheco and Michael D Ernst. 2007. Randoop: feedback-directed random testing for Java. In Companion to the 22nd ACMSIGPLAN conference on Object-oriented programming systems and applications companion. 815–816.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Sebastian Poeplau and Aurélien Francillon. 2020. Symbolic execution with SymCC: Don’t interpret, compile!. In USENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "PostgreSQL 2024. PostgreSQL. https://www .postgresql .org/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 51,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 52,
+      "text": "Per Runeson. 2006. A survey of unit testing practices. IEEE software 23, 4 (2006), 22–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https://github .com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, New York, USA. Morgan Kaufmann, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Rick F Van Der Lans. 1989. The SQL standard: a complete guide reference. Prentice Hall International (UK) Ltd.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Mingzhe Wang, Jie Liang, Chijin Zhou, Zhiyong Wu, Jingzhou Fu, Zhuo Su, Qing Liao, Bin Gu, Bodong Wu, and Yu Jiang. 2024. Data Coverage for Guided Fuzzing. In 33rd USENIX Security Symposium (USENIX Security 24). 2511–2526.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of CoverageGuided Enterprise-Level DBMS Fuzzing. In 43rd IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice, ICSE (SEIP) 2021, Madrid, Spain, May 25-28, 2021. IEEE, 328– 337. https://doi .org/10 .1109/ICSE-SEIP52600 .2021 .00042",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Lee J White and Edward I Cohen. 1980. A domain strategy for computer program testing. IEEE transactions on software engineering 3 (1980), 247–257.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Wikipedia 2024. databases. https://en .wikipedia .org/wiki/Database. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. PUPPY: Finding Performance Degradation Bugs in DBMSs via LimitedOptimization Plan Construction. In Proceedings of the IEEE/ACM 47th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Insu Yun, Sangho Lee, Meng Xu, Yeongjin Jang, and Taesoo Kim. 2018. QSYM: A Practical Concolic Execution Engine Tailored for Hybrid Fuzzing. In USENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Michał Zalewski. 2017. american fuzzy lop. http: //lcamtuf .coredump .cx/afl/. Accessed: September 23, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020. 1076",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 50,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 51,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "86% more code branches of the SQL function components of DBMSs in 24 hours than Sqirrel, SQLancer, and SQLsmith, respectively.",
+      "context_before": "rom these three sources, we summarized 10 boundary-value-generation patterns of bug-inducing queries. Based on these findings, we implemented Soft, a testing tool that generates test cases containing built-in SQL function expressions, with arguments produced based on these patterns. To evaluate the performance of Soft, we apply Soft to test the SQL functions in seven widely-used DBMSs: PostgreSQL, MySQL, MariaDB, ClickHouse, MonetDB, DuckDB, and Virtuoso. Compared to existing DBMS testing tools, Soft triggers 984, 1567, and 181 more functions of these DBMSs, and covers 433.93%, 98.70%, and 19.",
+      "context_after": "Moreover, Soft detected 132 previously unknown bugs in these DBMSs, which have garnered the attention of the DBMS developers. For example, ClickHouse’s CTO quickly commented on the SQL function bug shown in Listing 1 when we reported it [ 21]: “We must fix it immediately or get rid of this function. ” The CTO then promptly submitted a pull request to remove the flawed function. Within two days, ClickHouse developers fixed the bug and resubmitted the SQL function’s code. In summary, this paper has the following contributions: 1062 Understanding and Detecting SQL Function Bugs EuroSys ’25, Marc",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 10025,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We also used the latest versions of Sqirrel [63], SQLsmith [ 53], and SQLancer in PQS mode [ 51] with their default configurations to test these DBMSs, but they did not find any SQL function bugs.",
+      "context_before": "Ubuntu 20.04 with an AMDEPYC 7742 Processor @ 2.25 GHz, 128 cores, and 504 GiB of main memory. We useSoft to test the latest version of 7 open-source DBMSs, namely PostgreSQL v16.1 [ 48], MySQL v8.3.0 [ 45], MariaDB v11.3.2 [ 5], ClickHouse v23.6.2.18 [ 30], MonetDB v11.47.11 [40], DuckDB v0.10.1 [ 16], and Virtuoso v7.2.12 [ 42], which are widely used in industry. All DBMSs were tested using docker containers that were downloaded directly from their websites, each with 5 CPU cores and 50 GiB of RAM. 7.3 Detected DBMS Vulnerabilities Soft successfully detected 132 vulnerabilities in two weeks.",
+      "context_after": "Table 4 shows Soft discovered 1, 16, 24, 6, 19, 21, and 45 bugs in PostgreSQL, MySQL, MariaDB,ClickHouse, MonetDB, DuckDB, and Virtuoso, respectively. It shows that the SQL function bugs found by Soft cover various types with different boundary-value-generation patterns. Specifically, Soft detected 56, 28, and 48 SQL function bugs in patterns of literal values, type castings, and nested functions, respectively. In particular, the bugs also cover different kinds of memory errors, including 61 null pointer dereferences, 29 segmentation violations, 12 heap buffer overflows, 4 global buffer overf",
+      "section": "7.3 Detected DBMS Vulnerabilities",
+      "page": 10,
+      "char_offset": 54520,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "5 Comparison with Other Testing Works To demonstrate the effectiveness of our methods, we compared Soft against three state-of-the-art DBMS testing tools, namely Sqirrel, SQLancer, and SQLsmith, which are widely used in the industry.",
+      "context_before": "ce involves a segmentation violation in MariaDB triggered by a combination of the ST_ASTEXT ,BOUNDARY, and INET6_ATON functions. The function INET6_ATON converts the IP address ‘255.255.255.255’ to its binary form, which is then passed to the BOUNDARY function and subsequently to ST_ASTEXT. This sequence of operations causes a segmentation violation, likely due to improper handling or invalid memory access during the type casting and boundary calculation processes, leading to a crash inMariaDB. This reveals vulnerabilities in the system’s handling of complex function chains and type castings. 7.",
+      "context_after": "Among the DBMSs we tested, Sqirrel supports PostgreSQL, MySQL, and MariaDB; SQLsmith supports PostgreSQL and MonetDB; while SQLancer supports PostgreSQL, MySQL, MariaDB, and ClickHouse. We evaluated these DBMS testing tools based on the number of covered SQL functions and code branches of built-in SQL function components in 24 hours. We also countered the unique SQL function bugs discovered by these tools in 24 hours. For a fair comparison, when we finished DBMS testing, we collected the queries generated by each DBMS tool and reran all the queries to uniform the branch coverage. Table 5. Num",
+      "section": "7.5 Comparison with Other Testing Works",
+      "page": 12,
+      "char_offset": 64437,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Among the DBMSs we tested, Sqirrel supports PostgreSQL, MySQL, and MariaDB; SQLsmith supports PostgreSQL and MonetDB; while SQLancer supports PostgreSQL, MySQL, MariaDB, and ClickHouse.",
+      "context_before": "to the BOUNDARY function and subsequently to ST_ASTEXT. This sequence of operations causes a segmentation violation, likely due to improper handling or invalid memory access during the type casting and boundary calculation processes, leading to a crash inMariaDB. This reveals vulnerabilities in the system’s handling of complex function chains and type castings. 7.5 Comparison with Other Testing Works To demonstrate the effectiveness of our methods, we compared Soft against three state-of-the-art DBMS testing tools, namely Sqirrel, SQLancer, and SQLsmith, which are widely used in the industry.",
+      "context_after": "We evaluated these DBMS testing tools based on the number of covered SQL functions and code branches of built-in SQL function components in 24 hours. We also countered the unique SQL function bugs discovered by these tools in 24 hours. For a fair comparison, when we finished DBMS testing, we collected the queries generated by each DBMS tool and reran all the queries to uniform the branch coverage. Table 5. Number of triggered built-in SQL functions by generated SQL statements in 24 hours. DBMS Sqirrel SQLancer SQLsmith Soft PostgreSQL 29 123 417 456 MySQL 23 35 – 323 MariaDB 22 20 – 279 Click",
+      "section": "7.5 Comparison with Other Testing Works",
+      "page": 12,
+      "char_offset": 64670,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "DBMS Sqirrel SQLancer SQLsmith Soft PostgreSQL 29 123 417 456 MySQL 23 35 – 323 MariaDB 22 20 – 279 ClickHouse – 24 – 711 MonetDB – – 29 171 Total 74 202 446 2,956 Increment* 984 1,567 181 – *Increments are calculated only for commonly supported DBMSs.",
+      "context_before": "smith supports PostgreSQL and MonetDB; while SQLancer supports PostgreSQL, MySQL, MariaDB, and ClickHouse. We evaluated these DBMS testing tools based on the number of covered SQL functions and code branches of built-in SQL function components in 24 hours. We also countered the unique SQL function bugs discovered by these tools in 24 hours. For a fair comparison, when we finished DBMS testing, we collected the queries generated by each DBMS tool and reran all the queries to uniform the branch coverage. Table 5. Number of triggered built-in SQL functions by generated SQL statements in 24 hours.",
+      "context_after": "Covered Functions and Related Code Branches. Tables 5 and 6 show the number of SQL functions and code branches of these SQL function modules covered by those DBMS testing tools over 24-hour testing. The results indicate that Soft outperformed other DBMS testing tools. Specifically, Soft covered 984, 1567, and 181 more SQL functions, as well as 433.93%, 98.70%, and 19.86% more branches in built-in SQL function components than Sqirrel, SQLancer, and SQLsmith, respectively. The main reason for the improvement ofSoft is the boundary-value-generation patterns utilized. These generation patterns en",
+      "section": "7.5 Comparison with Other Testing Works",
+      "page": 12,
+      "char_offset": 65350,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "86% more branches in built-in SQL function components than Sqirrel, SQLancer, and SQLsmith, respectively.",
+      "context_before": "QLsmith Soft PostgreSQL 29 123 417 456 MySQL 23 35 – 323 MariaDB 22 20 – 279 ClickHouse – 24 – 711 MonetDB – – 29 171 Total 74 202 446 2,956 Increment* 984 1,567 181 – *Increments are calculated only for commonly supported DBMSs. Covered Functions and Related Code Branches. Tables 5 and 6 show the number of SQL functions and code branches of these SQL function modules covered by those DBMS testing tools over 24-hour testing. The results indicate that Soft outperformed other DBMS testing tools. Specifically, Soft covered 984, 1567, and 181 more SQL functions, as well as 433.93%, 98.70%, and 19.",
+      "context_after": "The main reason for the improvement ofSoft is the boundary-value-generation patterns utilized. These generation patterns enable Soft to construct boundary argument values and function expressions for a wide range of SQL functions. These boundary arguments help trigger the deep logic when DBMS processes function expressions, leading to higher branch coverage. In contrast, other tools lack a universal method for generating function expressions and their boundary arguments. For example, SQLancer requires writing function models in Java code to support the generation of a new function, and it onl",
+      "section": "7.5 Comparison with Other Testing Works",
+      "page": 12,
+      "char_offset": 65974,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer requires writing function models in Java code to support the generation of a new function, and it only supports generating random values for SQL function arguments.",
+      "context_before": "3%, 98.70%, and 19.86% more branches in built-in SQL function components than Sqirrel, SQLancer, and SQLsmith, respectively. The main reason for the improvement ofSoft is the boundary-value-generation patterns utilized. These generation patterns enable Soft to construct boundary argument values and function expressions for a wide range of SQL functions. These boundary arguments help trigger the deep logic when DBMS processes function expressions, leading to higher branch coverage. In contrast, other tools lack a universal method for generating function expressions and their boundary arguments.",
+      "context_after": "The complexity of adaptation and the lack of boundary argument generation limit its ability on SQL function testing. 1072 Understanding and Detecting SQL Function Bugs EuroSys ’25, March 30–April 3, 2025, Rotterdam, Netherlands Table 6. Number of covered code branches of DBMSs’ builtin SQL function modules in 24 hours. DBMS Sqirrel SQLancer SQLsmith Soft PostgreSQL 2,106 6,106 11,768 13,334 MySQL 1,105 1,927 – 6,914 MariaDB 1,758 1,732 – 6,283 ClickHouse – 26,655 – 45,836 MonetDB – – 551 1,431 Total 4,969 36,420 12,319 73,798 Increment* 21,562 35,947 2,446 – *Increments are calculated only fo",
+      "section": "7.5 Comparison with Other Testing Works",
+      "page": 12,
+      "char_offset": 66555,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "DBMS Sqirrel SQLancer SQLsmith Soft PostgreSQL 2,106 6,106 11,768 13,334 MySQL 1,105 1,927 – 6,914 MariaDB 1,758 1,732 – 6,283 ClickHouse – 26,655 – 45,836 MonetDB – – 551 1,431 Total 4,969 36,420 12,319 73,798 Increment* 21,562 35,947 2,446 – *Increments are calculated only for commonly supported DBMSs.",
+      "context_before": "ls lack a universal method for generating function expressions and their boundary arguments. For example, SQLancer requires writing function models in Java code to support the generation of a new function, and it only supports generating random values for SQL function arguments. The complexity of adaptation and the lack of boundary argument generation limit its ability on SQL function testing. 1072 Understanding and Detecting SQL Function Bugs EuroSys ’25, March 30–April 3, 2025, Rotterdam, Netherlands Table 6. Number of covered code branches of DBMSs’ builtin SQL function modules in 24 hours.",
+      "context_after": "Triggered SQL Function Bugs. Sqirrel, SQLancer, and SQLsmith did not find any SQL function bugs in 24 hours. In contrast, Soft triggered 22 unique SQL function bugs within 24 hours, including 1, 5, 6, 3, and 7 bugs in PostgreSQL, MySQL, MariaDB, ClickHouse, and MonetDB, respectively. Unlike Sqirrel, SQLancer, and SQLsmith, our tool Soft specifically targets boundary values of SQL function arguments. By focusing on boundary literal values, type castings, and nested function return values, Soft is able to identify and test edge cases that are often the root cause of bugs. Besides, Soft utilizes",
+      "section": "7.5 Comparison with Other Testing Works",
+      "page": 13,
+      "char_offset": 67063,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Sqirrel, SQLancer, and SQLsmith did not find any SQL function bugs in 24 hours.",
+      "context_before": "rgument generation limit its ability on SQL function testing. 1072 Understanding and Detecting SQL Function Bugs EuroSys ’25, March 30–April 3, 2025, Rotterdam, Netherlands Table 6. Number of covered code branches of DBMSs’ builtin SQL function modules in 24 hours. DBMS Sqirrel SQLancer SQLsmith Soft PostgreSQL 2,106 6,106 11,768 13,334 MySQL 1,105 1,927 – 6,914 MariaDB 1,758 1,732 – 6,283 ClickHouse – 26,655 – 45,836 MonetDB – – 551 1,431 Total 4,969 36,420 12,319 73,798 Increment* 21,562 35,947 2,446 – *Increments are calculated only for commonly supported DBMSs. Triggered SQL Function Bugs.",
+      "context_after": "In contrast, Soft triggered 22 unique SQL function bugs within 24 hours, including 1, 5, 6, 3, and 7 bugs in PostgreSQL, MySQL, MariaDB, ClickHouse, and MonetDB, respectively. Unlike Sqirrel, SQLancer, and SQLsmith, our tool Soft specifically targets boundary values of SQL function arguments. By focusing on boundary literal values, type castings, and nested function return values, Soft is able to identify and test edge cases that are often the root cause of bugs. Besides, Soft utilizes a pattern-based generation technique derived from our study of 318 SQL function bugs. These patterns provide",
+      "section": "7.5 Comparison with Other Testing Works",
+      "page": 13,
+      "char_offset": 67398,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Unlike Sqirrel, SQLancer, and SQLsmith, our tool Soft specifically targets boundary values of SQL function arguments.",
+      "context_before": "24 hours. DBMS Sqirrel SQLancer SQLsmith Soft PostgreSQL 2,106 6,106 11,768 13,334 MySQL 1,105 1,927 – 6,914 MariaDB 1,758 1,732 – 6,283 ClickHouse – 26,655 – 45,836 MonetDB – – 551 1,431 Total 4,969 36,420 12,319 73,798 Increment* 21,562 35,947 2,446 – *Increments are calculated only for commonly supported DBMSs. Triggered SQL Function Bugs. Sqirrel, SQLancer, and SQLsmith did not find any SQL function bugs in 24 hours. In contrast, Soft triggered 22 unique SQL function bugs within 24 hours, including 1, 5, 6, 3, and 7 bugs in PostgreSQL, MySQL, MariaDB, ClickHouse, and MonetDB, respectively.",
+      "context_after": "By focusing on boundary literal values, type castings, and nested function return values, Soft is able to identify and test edge cases that are often the root cause of bugs. Besides, Soft utilizes a pattern-based generation technique derived from our study of 318 SQL function bugs. These patterns provide a structured methodology for creating test cases that are more likely to uncover hidden bugs. 8 Discussion UDF Bugs. This paper only exploits vulnerabilities in the built-in SQL functions. However, user-defined functions of DBMS can also be problematic to a large extent. In contrast to built-",
+      "section": "7.5 Comparison with Other Testing Works",
+      "page": 13,
+      "char_offset": 67654,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", TLP [50]) and transformation (e.",
+      "context_before": "er, we mainly focus on memory safety problems in built-in SQL functions. However, it is still possible for a SQL function to have correctness problems (or logic problems), where functions return incorrect results. To address these issues, we can extend the existing testing frameworks to construct test guidelines that include function expressions with equivalent semantics and perform differential testing. Equivalent semantic function expressions can be constructed by replacing intermediate clauses in existing nested structures, e.g., intermediate result replacement, where clause splitting (e.g.",
+      "context_after": "g., NoREC [49]). Reality of Found Bugs. In this paper, we found that the core cause of the built-in function problem is the poor handling of boundary values, so we test the built-in function bygenerating literals, type conversions, and nested structures to get different boundary values. In the real world, software systems often need to process a wide variety of input data, including those under various boundary conditions. These boundary conditions might be triggered due to user input, interactions with external systems, or internal data processing errors. Moreover, once these bugs occur, they",
+      "section": "8 Discussion",
+      "page": 13,
+      "char_offset": 69435,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": ", NoREC [49]).",
+      "context_before": "problems in built-in SQL functions. However, it is still possible for a SQL function to have correctness problems (or logic problems), where functions return incorrect results. To address these issues, we can extend the existing testing frameworks to construct test guidelines that include function expressions with equivalent semantics and perform differential testing. Equivalent semantic function expressions can be constructed by replacing intermediate clauses in existing nested structures, e.g., intermediate result replacement, where clause splitting (e.g., TLP [50]) and transformation (e.g.",
+      "context_after": "Reality of Found Bugs. In this paper, we found that the core cause of the built-in function problem is the poor handling of boundary values, so we test the built-in function bygenerating literals, type conversions, and nested structures to get different boundary values. In the real world, software systems often need to process a wide variety of input data, including those under various boundary conditions. These boundary conditions might be triggered due to user input, interactions with external systems, or internal data processing errors. Moreover, once these bugs occur, they can lead to app",
+      "section": "8 Discussion",
+      "page": 13,
+      "char_offset": 69471,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "citation_marker",
+      "surface": "[ 15,49–51,54]",
+      "technique": "norec",
+      "sentence": "DBMS correctness testing [ 15,49–51,54] aims to verify that the DBMS accurately executes queries.",
+      "context_before": "conditions under which the DBMS might unexpectedly stop working or crash. These works typically concentrate on generating valid SQL statements and then monitoring whether the DBMS under test crashes. For example, SQLsmith [ 53] is a generation-based fuzzer to generate SQL queries for testing. Sqirrel [63] proposes an IR-based mutation method to generate queries with syntactical and semantical correctness. Lego [33] uses typeaffinity analysis to generate queries with more SQL type sequences to explore the SQL state space. Griffin [19] proposes a grammar-free fuzzing method for query generation.",
+      "context_after": "The state-of-theart works focus on constructing effective logic testing oracles. The mainstream approach involves adopting metamorphic testing and differential testing. For example, PQS [ 51] detects whether the pivot row exists in the preset query results. NoREC [ 49] detects inconsistencies between query results before and after optimization. Mozi [ 37] utilizes equivalent configuration transformation to test whether the result matches. 1073 EuroSys ’25, March 30–April 3, 2025, Rotterdam, Netherlands Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang DBMS performanc",
+      "section": "9 Related Work",
+      "page": 13,
+      "char_offset": 72017,
+      "cited_reference": {
+        "number": 49,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "For example, PQS [ 51] detects whether the pivot row exists in the preset query results.",
+      "context_before": "e SQL queries for testing. Sqirrel [63] proposes an IR-based mutation method to generate queries with syntactical and semantical correctness. Lego [33] uses typeaffinity analysis to generate queries with more SQL type sequences to explore the SQL state space. Griffin [19] proposes a grammar-free fuzzing method for query generation. DBMS correctness testing [ 15,49–51,54] aims to verify that the DBMS accurately executes queries. The state-of-theart works focus on constructing effective logic testing oracles. The mainstream approach involves adopting metamorphic testing and differential testing.",
+      "context_after": "NoREC [ 49] detects inconsistencies between query results before and after optimization. Mozi [ 37] utilizes equivalent configuration transformation to test whether the result matches. 1073 EuroSys ’25, March 30–April 3, 2025, Rotterdam, Netherlands Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang DBMS performance testing [ 32,39,60] assesses the performance under different workloads and conditions. Apollo [32] utilizes performance regression testing to find errors in the evolution of code during development. Amoeba [39] generates semantically equivalent queries and",
+      "section": "9 Related Work",
+      "page": 13,
+      "char_offset": 72284,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [ 49] detects inconsistencies between query results before and after optimization.",
+      "context_before": "queries with syntactical and semantical correctness. Lego [33] uses typeaffinity analysis to generate queries with more SQL type sequences to explore the SQL state space. Griffin [19] proposes a grammar-free fuzzing method for query generation. DBMS correctness testing [ 15,49–51,54] aims to verify that the DBMS accurately executes queries. The state-of-theart works focus on constructing effective logic testing oracles. The mainstream approach involves adopting metamorphic testing and differential testing. For example, PQS [ 51] detects whether the pivot row exists in the preset query results.",
+      "context_after": "Mozi [ 37] utilizes equivalent configuration transformation to test whether the result matches. 1073 EuroSys ’25, March 30–April 3, 2025, Rotterdam, Netherlands Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang DBMS performance testing [ 32,39,60] assesses the performance under different workloads and conditions. Apollo [32] utilizes performance regression testing to find errors in the evolution of code during development. Amoeba [39] generates semantically equivalent queries and compares their response times to identify performance issues. Soft aims to find crash bu",
+      "section": "9 Related Work",
+      "page": 13,
+      "char_offset": 72373,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M3"
+      ],
+      "describes_as_state_of_the_art": [
+        "M3"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:56:28Z",
+    "is_model_written": true,
+    "summary": "Soft targets built-in SQL functions, testing them with boundary arguments -- extreme, empty and edge-case values -- rather than random ones. The insight is that function implementations fail at their argument boundaries, and that generators producing random values almost never land there. Across PostgreSQL, MySQL, MariaDB, ClickHouse and MonetDB it found hundreds of SQL function bugs and covered far more of the function code.",
+    "narrative": "SQLancer is one of three baselines, described as a state-of-the-art DBMS testing tool widely used in industry, and run in PQS mode with default settings. The result is stark and reported twice: over 24 hours SQLancer found no SQL function bugs at all, against Soft's 2,956, and Soft covered many times more branches in the function components. The paper's explanation is concrete rather than dismissive -- supporting a new function in SQLancer means writing a function model in Java, and it only generates random values for function arguments, so the boundary values these bugs need are not something it produces.",
+    "roles": {
+      "M1": "result_comparison",
+      "M2": "result_comparison",
+      "M3": "state_of_the_art",
+      "M4": "baseline",
+      "M5": "result_comparison",
+      "M6": "result_comparison",
+      "M7": "motivation",
+      "M8": "result_comparison",
+      "M9": "result_comparison",
+      "M10": "result_comparison",
+      "M11": "definition",
+      "M12": "definition",
+      "M13": "background",
+      "M14": "definition",
+      "M15": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is run as a baseline in PQS mode; Soft's boundary-argument generation is its own."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer oracle is extended. Soft changes which argument values are generated, and M7 gives that as precisely the difference from SQLancer's random values."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M2",
+          "M3",
+          "M5",
+          "M9"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "We also used the latest versions of Sqirrel [63], SQLsmith [ 53], and SQLancer in PQS mode [ 51] with their default configurations to test these DBMSs, but they did not find any SQL function bugs.",
+            "section": "7.3 Detected DBMS Vulnerabilities",
+            "page": 10
+          },
+          {
+            "mention_id": "M3",
+            "sentence": "5 Comparison with Other Testing Works To demonstrate the effectiveness of our methods, we compared Soft against three state-of-the-art DBMS testing tools, namely Sqirrel, SQLancer, and SQLsmith, which are widely used in the industry.",
+            "section": "7.5 Comparison with Other Testing Works",
+            "page": 12
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "DBMS Sqirrel SQLancer SQLsmith Soft PostgreSQL 29 123 417 456 MySQL 23 35 – 323 MariaDB 22 20 – 279 ClickHouse – 24 – 711 MonetDB – – 29 171 Total 74 202 446 2,956 Increment* 984 1,567 181 – *Increments are calculated only for commonly supported DBMSs.",
+            "section": "7.5 Comparison with Other Testing Works",
+            "page": 12
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "Sqirrel, SQLancer, and SQLsmith did not find any SQL function bugs in 24 hours.",
+            "section": "7.5 Comparison with Other Testing Works",
+            "page": 13
+          }
+        ],
+        "reasoning": "M3 states Soft was compared against SQLancer among three tools, M2 records it being run in PQS mode with default configurations, and M5 and M9 report the bug counts and that SQLancer found no SQL function bugs in 24 hours.",
+        "techniques": [
+          "pqs"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M3"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "5 Comparison with Other Testing Works To demonstrate the effectiveness of our methods, we compared Soft against three state-of-the-art DBMS testing tools, namely Sqirrel, SQLancer, and SQLsmith, which are widely used in the industry.",
+            "section": "7.5 Comparison with Other Testing Works",
+            "page": 12
+          }
+        ],
+        "reasoning": "M3 calls SQLancer one of three state-of-the-art DBMS testing tools, widely used in the industry."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3689491_3691821.json b/_data/papers/paper_doi_10_1145_3689491_3691821.json
new file mode 100644
index 0000000..f60d28e
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3689491_3691821.json
@@ -0,0 +1,247 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:07:03Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3689491.3691821",
+    "title": "Step-wise Execution of Data-Centric Systems",
+    "authors": [
+      "Chi Zhang"
+    ],
+    "year": 2024,
+    "venue": "SPLASH Companion",
+    "doi": "10.1145/3689491.3691821",
+    "arxiv_id": null,
+    "s2_paper_id": "c8bb406e87172ca7ccc8f009d0554308232c01e5",
+    "url": "https://doi.org/10.1145/3689491.3691821",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3689491.3691821",
+      "retrieved_at": "2026-09-07T17:07:03Z",
+      "chars": 15957,
+      "content_sha256": "sha256:12f24d4be4b7c151c5d93d31a93a0e6892509af45ec4fce6b0e49864322959ed"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 3,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Motivation",
+        "start": 1781
+      },
+      {
+        "number": "2",
+        "title": "Problem",
+        "start": 5762
+      },
+      {
+        "number": "3",
+        "title": "Approach",
+        "start": 7423
+      },
+      {
+        "number": "4",
+        "title": "Evaluation Methodology",
+        "start": 11287
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Mario Alviano, Wolfgang Faber, Gianluigi Greco, and Nicola Leone. 2012. Magic sets for disjunctive datalog programs. Artificial Intelligence 187 (2012), 156–192.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Matthias Brantner, Norman May, and Guido Moerkotte. 2006. Unnesting scalar SQL queries in the presence of disjunction. In 2007 IEEE 23rd International Conference on Data Engineering. IEEE, 46–55.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Martin Bravenboer and Yannis Smaragdakis. 2009. Strictly declarative specification of sophisticated points-to analyses. In Proceedings of the 24th Annual ACMSIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications, OOPSLA 2009, October 25-29, 2009, Orlando, Florida, USA, Shail Arora and Gary T. Leavens (Eds.). ACM, 243–262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Haoran Peng, Chenyuan Yang, and Lingming Zhang. 2023. Large Language Models Are Zero-Shot Fuzzers: Fuzzing Deep-Learning Libraries via Large Language Models. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (Seattle, WA, USA) (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 423–435.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2019, Minneapolis, MN, USA, June 2-7, 2019, Volume 1 (Long and Short Papers). Association",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Muhammad Numair Mansur, Maria Christakis, and Valentin Wüstholz. 2021. Metamorphic testing of Datalog engines. In Proceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 639–650.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 8,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 9,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting logic bugs of join optimizations in dbms. Proceedings of the ACM on Management of Data 1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Chi Zhang, Linzhang Wang, and Manuel Rigger. 2024. Finding CrossRule Optimization Bugs in Datalog Engines. Proc. ACM Program. Lang. 8, OOPSLA1, Article 98 (apr 2024), 27 pages. 21",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 7,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 8,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 10,
+      "text": "Chi Zhang, Linzhang Wang, and Manuel Rigger. 2024. Finding CrossRule Optimization Bugs in Datalog Engines. Proc. ACM Program. Lang. 8, OOPSLA1, Article 98 (apr 2024), 27 pages. 21",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [ 7], TLP [ 8], and TQS [ 9] are all stateof-the-art approaches for detecting logic bugs in DBMSs.",
+      "context_before": "the original and transformed test cases: equivalence, containment, and substitution. However, the containment and substitution relationships are loose, and queryFuzz detects bugs based on the monotonicity of Datalog programs without negation, which is a key language feature, leading to queryFuzz overlooked multiple logic bugs. Furthermore, queryFuzz generates the entire program at once for each test case, resulting in low test case validity. Therefore, a stricter test oracle supporting negation, combined with an efficient test case generation method, is needed to uncover more logic bugs. DBMS.",
+      "context_after": "However, none of these approaches can effectively test subqueries, and it is unclear how to support this capability. A test oracle that supports subqueries is necessary, as they are a critical feature in SQL. Deep Learning Library. Existing fuzzing approaches for Deep Learning libraries fall into two types [ 4]: API-level, which focuses on individual functions and lacks testing of function interactions, and model-level, which relies on differential testing but often misses logic bugs. An efficient test oracle is needed to address these gaps and detect logic bugs in Deep Learning libraries. 3",
+      "section": "2 Problem",
+      "page": 2,
+      "char_offset": 6720,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 10]",
+      "technique": null,
+      "sentence": "Fuzzing Datalog engines [ 10].",
+      "context_before": "an be used to detect this bug. However, we still face other challenges, such as generating more complex test cases that include conditional branches and loops. Conclusion. While we have successfully realized and evaluated the idea to Datalog engines and DBMSs, we plan to next apply it to Deep Learning libraries. We are still looking for further domains to which the idea could be applied. 4 Evaluation Methodology For this research, we evaluated three aspects to investigate the effectiveness of test oracle, the efficiency of test case generation, and compare with the state-of-the-art approaches.",
+      "context_after": "We evaluate our approach on four mature Datalog engines, and discovered a total of 30 bugs. Of these, 13 were logic bugs, while the remaining were crash and error bugs. Our approach can detect all bugs found by queryFuzz, a state-of-the-art approach. Our incremental test case generation approach can generate 1.17 ×to 31.02× as many valid test cases than the random method. Fuzzing DBMSs. We evaluated our approach on five mature and extensively-tested DBMSs and found 45 unique, previously unknown bugs in them. Out of these, 24 are unique logic bugs. Our manual analysis of the state-of-the-art ap",
+      "section": "4 Evaluation Methodology",
+      "page": 3,
+      "char_offset": 11496,
+      "cited_reference": {
+        "number": 10,
+        "text": "Chi Zhang, Linzhang Wang, and Manuel Rigger. 2024. Finding CrossRule Optimization Bugs in Datalog Engines. Proc. ACM Program. Lang. 8, OOPSLA1, Article 98 (apr 2024), 27 pages. 21",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:19:46Z",
+    "is_model_written": true,
+    "summary": "This paper proposes a black-box methodology for testing data-centric systems, motivated by the observation that existing methods suffer from ineffective oracles, insufficient coverage and inefficient test generation. Its key idea is to build the test case incrementally and use the intermediate results to construct a reference case, whose result must match the original's. Applied to Datalog engines and DBMSs it uncovered 75 unique bugs.",
+    "narrative": "NoREC and TLP are named as state-of-the-art approaches for detecting logic bugs in DBMSs, the line of work the step-wise reference construction is set against.",
+    "roles": {
+      "M1": "state_of_the_art",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M1"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "NoREC [ 7], TLP [ 8], and TQS [ 9] are all stateof-the-art approaches for detecting logic bugs in DBMSs.",
+            "section": "2 Problem",
+            "page": 2
+          }
+        ],
+        "reasoning": "M1 calls NoREC, TLP and TQS all state-of-the-art approaches for detecting logic bugs in DBMSs."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3689757.json b/_data/papers/paper_doi_10_1145_3689757.json
new file mode 100644
index 0000000..0d975ba
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3689757.json
@@ -0,0 +1,637 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:06:58Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3689757",
+    "title": "PolyJuice: Detecting Mis-compilation Bugs in Tensor Compilers with Equality Saturation Based Rewriting",
+    "authors": [
+      "Chijin Zhou",
+      "Bingzhou Qian",
+      "Gwihwan Go",
+      "Quan Zhang",
+      "Shanshan Li",
+      "Yu Jiang"
+    ],
+    "year": 2024,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3689757",
+    "arxiv_id": null,
+    "s2_paper_id": "9889059fce1930bbb29d7cbcc33e5fc1ed168f2a",
+    "url": "https://doi.org/10.1145/3689757",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3689757",
+      "retrieved_at": "2026-09-07T17:06:58Z",
+      "chars": 109508,
+      "content_sha256": "sha256:dc385935b82e4fff8265497f02d64ff9a486426e4e80992dc14da5732a6a9293"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 27,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2252
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 12764
+      },
+      {
+        "number": "2.1",
+        "title": "Tensor Program Compilation",
+        "start": 12777
+      },
+      {
+        "number": "2.2",
+        "title": "Equality Saturation",
+        "start": 14335
+      },
+      {
+        "number": "3",
+        "title": "Motivation",
+        "start": 16485
+      },
+      {
+        "number": "4",
+        "title": "Approach",
+        "start": 24014
+      },
+      {
+        "number": "4.1",
+        "title": "Graph Conversion",
+        "start": 26573
+      },
+      {
+        "number": "4.2",
+        "title": "Computation Expression Rewriting",
+        "start": 30999
+      },
+      {
+        "number": "4.3",
+        "title": "Complexity Aware Graph Extraction",
+        "start": 39668
+      },
+      {
+        "number": "4.4",
+        "title": "Testcase Generation",
+        "start": 43844
+      },
+      {
+        "number": "5",
+        "title": "Implementation",
+        "start": 45127
+      },
+      {
+        "number": "6",
+        "title": "Evaluation",
+        "start": 46822
+      },
+      {
+        "number": "6.1",
+        "title": "Mis-Compilation Bug Detection",
+        "start": 49013
+      },
+      {
+        "number": "6.2",
+        "title": "Comparative Study",
+        "start": 61561
+      },
+      {
+        "number": "6.3",
+        "title": "Effectiveness of Equality Saturation Based Rewriting",
+        "start": 64286
+      },
+      {
+        "number": "6.4",
+        "title": "Overhead During Testing",
+        "start": 75963
+      },
+      {
+        "number": "7",
+        "title": "Discussion",
+        "start": 77886
+      },
+      {
+        "number": "8",
+        "title": "Related Work",
+        "start": 83405
+      },
+      {
+        "number": "9",
+        "title": "Conclusion",
+        "start": 89596
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2024. PolyJuice: Detecting Mis-Compilation Bugs in Tensor Compilers with Equality Saturation Based Rewriting. Zenodo. https://doi.org/10.5281/zenodo.12671619",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Lukas Bernhard, Tobias Scharnowski, Moritz Schloegel, Tim Blazytko, and Thorsten Holz. 2022. JIT-Picking: Differential Fuzzing of JavaScript Engines. In 2022 ACMSIGSAC Conference on Computer and Communications Security, CCS 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Paul E Black. 1999. Algorithms and theory of computation handbook, \"Levenshtein distance\". CRC press. https: //www.nist.gov/dads/HTML/Levenshtein.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Junming Cao, Bihuan Chen, Chao Sun, Longjie Hu, Shuaihong Wu, and Xin Peng. 2022. Understanding performance problems in deep learning systems. In 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Tianqi Chen, Thierry Moreau, Ziheng Jiang, Lianmin Zheng, Eddie Q. Yan, Haichen Shen, Meghan Cowan, Leyuan Wang, Yuwei Hu, Luis Ceze, Carlos Guestrin, and Arvind Krishnamurthy. 2018. TVM: An Automated End-to-End Optimizing Compiler for Deep Learning. In 13th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Chenyuan Yang, Shizhuo Dylan Zhang, Shujing Yang, and Lingming Zhang. 2024. Large language models are edge-case generators: Crafting unusual programs for fuzzing deep learning libraries. In 46th IEEE/ACM International Conference on Software Engineering, ICSE 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Yaoyao Ding, Cody Hao Yu, Bojian Zheng, Yizhi Liu, Yida Wang, and Gennady Pekhimenko. 2023. Hidet: Task-Mapping Programming Paradigm for Deep Learning Tensor Programs. In 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2, ASPLOS 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. In 37th IEEE/ACM International Conference on Automated Software Engineering, ASE 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Jiazhen Gu, Xuchuan Luo, Yangfan Zhou, and Xin Wang. 2022. Muffin: Testing Deep Learning Libraries via Neural Architecture Fuzzing. In 44th IEEE/ACM 44th International Conference on Software Engineering, ICSE 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Hao Guan, Ying Xiao, Jiaying Li, Yepang Liu, and Guangdong Bai. 2023. A Comprehensive Study of Real-World Bugs in Machine Learning Model Optimization. In 45th IEEE/ACM International Conference on Software Engineering, ICSE 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Qianyu Guo, Xiaofei Xie, Yi Li, Xiaoyu Zhang, Yang Liu, Xiaohong Li, and Chao Shen. 2020. Audee: Automated Testing for Deep Learning Frameworks. In 35th IEEE/ACM International Conference on Automated Software Engineering, ASE 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "HyungSeok Han, DongHyeon Oh, and Sang Kil Cha. 2019. CodeAlchemist: Semantics-Aware Code Generation to Find Vulnerabilities in JavaScript Engines. In 26th Annual Network and Distributed System Security Symposium, NDSS 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "The IEEE and The Open Group. 2023. diffcompare two files. https://pubs.opengroup.org/onlinepubs/009604499/utilit ies/diff.html. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Zhihao Jia, Oded Padon, James Thomas, Todd Warszawski, Matei Zaharia, and Alex Aiken. 2019. TASO: optimizing deep learning computation with automatic generation of graph substitutions. In 27th ACM Symposium on Operating Systems Principles, SOSP 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Zhihao Jia, James Thomas, Todd Warszawski, Mingyu Gao, Matei Zaharia, and Alex Aiken. 2019. Optimizing DNN Computation with Relaxed Graph Substitutions. In Proceedings of Machine Learning and Systems 2019, MLSys 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Wookeun Jung, Thanh Tuan Dao, and Jaejin Lee. 2021. DeepCuts: a deep learning optimization framework for versatile GPU workloads. In 42nd ACMSIGPLAN International Conference on Programming Language Design and Implementation, PLDI 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Seulbae Kim, Meng Xu, Sanidhya Kashyap, Jungyeon Yoon, Wen Xu, and Taesoo Kim. 2019. Finding semantic bugs in file systems with an extensible fuzzing framework. In 27th ACM Symposium on Operating Systems Principles, SOSP 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Fredrik Kjolstad, Shoaib Kamil, Stephen Chou, David Lugato, and Saman P. Amarasinghe. 2017. The tensor algebra compiler. Proc. ACM Program. Lang. 1, OOPSLA (2017), 77:1–77:29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Thomas Koehler, Andrés Goens, Siddharth Bhat, Tobias Grosser, Phil Trinder, and Michel Steuwer. 2024. Guided Equality Saturation. Proc. ACM Program. Lang. 8, POPL (2024), 1727–1758.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Sameer Kulkarni and John Cavazos. 2012. Mitigating the compiler optimization phase-ordering problem using machine learning. In 27th Annual ACMSIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications, OOPSLA 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. In 35th ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Cong Li, Yanyan Jiang, Chang Xu, and Zhendong Su. 2023. Validating JIT Compilers via Compilation Space Exploration. In29th Symposium on Operating Systems Principles, SOSP 2023, Koblenz, Germany, October 23-26, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Cong Li, Yanyan Jiang, Chang Xu, and Zhendong Su. 2023. Validating JIT Compilers via Compilation Space Exploration. In29th Symposium on Operating Systems Principles, SOSP 2023. Proc. ACM Program. Lang., Vol. 8, No. OOPSLA2, Article 317. Publication date: October 2024. PolyJuice: Detecting Mis-compilation Bugs in Tensor Compilers with Equality Saturation Based Rewriting 317:25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Mingzhen Li, Yi Liu, Xiaoyan Liu, Qingxiao Sun, Xin You, Hailong Yang, Zhongzhi Luan, Lin Gan, Guangwen Yang, and Depei Qian. 2021. The Deep Learning Compiler: A Comprehensive Survey. IEEE Trans. Parallel Distributed Syst. 32, 3 (2021), 708–727.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Shaohua Li and Zhendong Su. 2023. Accelerating Fuzzing through Prefix-Guided Execution. Proc. ACM Program. Lang. 7, OOPSLA1 (2023), 1–27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation. In 46th IEEE/ACM International Conference on Software Engineering, ICSE 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Jiawei Liu, Jinkun Lin, Fabian Ruffy, Cheng Tan, Jinyang Li, Aurojit Panda, and Lingming Zhang. 2023. NNSmith: Generating Diverse and Valid Test Cases for Deep Learning Compilers. In 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2, ASPLOS 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Jiawei Liu, Yuxiang Wei, Sen Yang, Yinlin Deng, and Lingming Zhang. 2022. Coverage-guided tensor compiler fuzzing with joint IR-pass mutation. Proc. ACM Program. Lang. 6, OOPSLA1 (2022), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Weisi Luo, Dong Chai, Xiaoyue Run, Jiang Wang, Chunrong Fang, and Zhenyu Chen. 2021. Graph-based Fuzz Testing for Deep Learning Inference Engines. In 43rd IEEE/ACM International Conference on Software Engineering, ICSE 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Haoyang Ma, Qingchao Shen, Yongqiang Tian, Junjie Chen, and Shing-Chi Cheung. 2023. Fuzzing Deep Learning Compilers with HirGen. In 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "microsoft. 2023. ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator. https: //github.com/microsoft/onnxruntime. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Jayashree Mohan, Ashlie Martinez, Soujanya Ponnapalli, Pandian Raju, and Vijay Chidambaram. 2018. Finding Crash-Consistency Bugs with Bounded Black-Box Crash Testing. In 13th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Nvidia. 2023. NVIDIA TensorRT. https://developer.nvidia.com/tensorrt. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "open std. 2023. ISO/IEC 9899:TC3. https://www.open-std.org/jtc1/sc22/wg14/www/docs/n1256.pdf. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Anjali Pal, Brett Saiki, Ryan Tjoa, Cynthia Richey, Amy Zhu, Oliver Flatt, Max Willsey, Zachary Tatlock, and Chandrakana Nandi. 2023. Equality Saturation Theory Exploration à la Carte. Proc. ACM Program. Lang. 7, OOPSLA2 (2023), 1034–1062.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Pavel Panchekha, Alex Sanchez-Stern, James R. Wilcox, and Zachary Tatlock. 2015. Automatically improving accuracy for floating point expressions. In 36th ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Douglas M. Priest. 1992. On properties of floating point arithmetics: numerical stability and the cost of accurate computations. Ph. D. Dissertation. USA. UMI Order No. GAX93-30692.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "PyTorch. 2023. Accelerated CPU Inference with PyTorch Inductor using torch.compile. https://pytorch.org/blog/accel erated-cpu-inference/. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "PyTorch. 2023. Glow-Compiler for Neural Network hardware accelerators. https://github.com/pytorch/glow. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Jonathan Ragan-Kelley, Connelly Barnes, Andrew Adams, Sylvain Paris, Frédo Durand, and Saman P. Amarasinghe. 2013. Halide: a language and compiler for optimizing parallelism, locality, and recomputation in image processing pipelines. In 34th ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "James K. Reed, Zachary DeVito, Horace He, Ansley Ussery, and Jason Ansel. 2022. torch.fx: Practical Program Capture and Transformation for Deep Learning in Python. In Proceedings of Machine Learning and Systems 2022, MLSys 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Albert Reuther, Peter Michaleas, Michael Jones, Vijay Gadepally, Siddharth Samsi, and Jeremy Kepner. 2020. Survey of Machine Learning Accelerators. In.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "Junru Shao, Xiyou Zhou, Siyuan Feng, Bohan Hou, Ruihang Lai, Hongyi Jin, Wuwei Lin, Masahiro Masuda, Cody Hao Yu, and Tianqi Chen. 2022. Tensor Program Optimization with Probabilistic Programs. In NeurIPS.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Qingchao Shen, Haoyang Ma, Junjie Chen, Yongqiang Tian, Shing-Chi Cheung, and Xiang Chen. 2021. A comprehensive study of deep learning compiler bugs. In 29th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Hao Sun, Yiru Xu, Jianzhong Liu, Yuheng Shen, Nan Guan, and Yu Jiang. 2024. Finding Correctness Bugs in eBPF Verifier with Structured and Sanitized Program. In 19th European Conference on Computer Systems, EuroSys 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Ross Tate, Michael Stepp, Zachary Tatlock, and Sorin Lerner. 2009. Equality saturation: a new approach to optimization. In36th ACMSIGPLAN-SIGACT Symposium on Principles of Programming Languages, POPL 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Tensorflow. 2023. https://www.tensorflow.org/xla. https://www.tensorflow.org/xla. (visited on September 1, 2023). Proc. ACM Program. Lang., Vol. 8, No. OOPSLA2, Article 317. Publication date: October 2024. 317:26 Chijin Zhou, Bingzhou Qian, Gwihwan Go, Quan Zhang, Shanshan Li, and Yu Jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. In 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, ASPLOS 2022. ACM.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "tvm. 2023. PR: Fix a bug of iter map floormod(x,2) simplify. https://github.com/apache/tvm/pull/14571. (visited on September 1, 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Colin Unger, Zhihao Jia, Wei Wu, Sina Lin, Mandeep Baines, Carlos Efrain Quintero Narvaez, Vinay Ramakrishnaiah, Nirmal Prajapati, Patrick S. McCormick, Jamaludin Mohd-Yusof, Xi Luo, Dheevatsa Mudigere, Jongsoo Park, Misha Smelyanskiy, and Alex Aiken. 2022. Unity: Accelerating DNN Training Through Joint Optimization of Algebraic Transformations and Parallelization. In 16th USENIX Symposium on Oper",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Alexa VanHattum, Rachit Nigam, Vincent T. Lee, James Bornholt, and Adrian Sampson. 2021. Vectorization for digital signal processors via equality saturation. In 26th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, ASPLOS 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Haojie Wang, Jidong Zhai, Mingyu Gao, Zixuan Ma, Shizhi Tang, Liyan Zheng, Yuanzhi Li, Kaiyuan Rong, Yuanyong Chen, and Zhihao Jia. 2021. PET: Optimizing Tensor Programs with Partially Equivalent Transformations and Automated Corrections. In 15th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Jiannan Wang, Thibaud Lutellier, Shangshu Qian, Hung Viet Pham, and Lin Tan. 2022. EAGLE: Creating Equivalent Graphs to Test Deep Learning Libraries. In 44th IEEE/ACM 44th International Conference on Software Engineering, ICSE 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Mingzhe Wang, Jie Liang, Chijin Zhou, Yu Jiang, Rui Wang, Chengnian Sun, and Jiaguang Sun. 2021. RIFF: Reduced Instruction Footprint for Coverage-Guided Fuzzing. In 2021 USENIX Annual Technical Conference, USENIX ATC 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Mingzhe Wang, Jie Liang, Chijin Zhou, Zhiyong Wu, Jingzhou Fu, Zhuo Su, Qing Liao, Bin Gu, Bodong Wu, and Yu Jiang. 2024. Data Coverage for Guided Fuzzing. In 33rd USENIX Security Symposium, USENIX Security 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Mingzhe Wang, Jie Liang, Chijin Zhou, Zhiyong Wu, Xinyi Xu, and Yu Jiang. 2022. Odin: on-demand instrumentation with on-the-fly recompilation. In 43rd ACMSIGPLAN International Conference on Programming Language Design and Implementation, PLDI 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In 43rd IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice, ICSE (SEIP) 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Zan Wang, Ming Yan, Junjie Chen, Shuang Liu, and Dongdi Zhang. 2020. Deep learning library testing via effective model generation. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Max Willsey, Chandrakana Nandi, Yisu Remy Wang, Oliver Flatt, Zachary Tatlock, and Pavel Panchekha. 2021. egg: Fast and extensible equality saturation. Proc. ACM Program. Lang. 5, POPL (2021), 1–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Dongwei Xiao, Zhibo Liu, Yuanyuan Yuan, Qi Pang, and Shuai Wang. 2022. Metamorphic Testing of Deep Learning Compilers. Proc. ACM Meas. Anal. Comput. Syst. 6, 1 (2022), 15:1–15:28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Yichen Yang, Phitchaya Mangpo Phothilimthana, Yisu Remy Wang, Max Willsey, Sudip Roy, and Jacques Pienaar. 2021. Equality Saturation for Tensor Graph Superoptimization. In Proceedings of Machine Learning and Systems 2021, MLSys 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Chen Zhang, Lingxiao Ma, Jilong Xue, Yining Shi, Ziming Miao, Fan Yang, Jidong Zhai, Zhi Yang, and Mao Yang. 2023. Cocktailer: Analyzing and Optimizing Dynamic Control Flow in Deep Learning. In 17th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Yihong Zhang, Yisu Remy Wang, Oliver Flatt, David Cao, Philip Zucker, Eli Rosenthal, Zachary Tatlock, and Max Willsey. 2023. Better Together: Unifying Datalog and Equality Saturation. Proc. ACM Program. Lang. 7, PLDI (2023), 468–492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Jie Zhao and Peng Di. 2020. Optimizing the Memory Hierarchy by Compositing Automatic Transformations on Computations and Data. In 53rd Annual IEEE/ACM International Symposium on Microarchitecture, MICRO 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Jie Zhao, Xiong Gao, Ruijie Xia, Zhaochuang Zhang, Deshi Chen, Lei Chen, Renwei Zhang, Zhen Geng, Bin Cheng, and Xuefeng Jin. 2022. Apollo: Automatic Partition-based Operator Fusion through Layer by Layer Optimization. In Proceedings of Machine Learning and Systems 2022, MLSys 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Lianmin Zheng, Chengfan Jia, Minmin Sun, Zhao Wu, Cody Hao Yu, Ameer Haj-Ali, Yida Wang, Jun Yang, Danyang Zhuo, Koushik Sen, Joseph E. Gonzalez, and Ion Stoica. 2020. Ansor: Generating High-Performance Tensor Programs for Deep Learning. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020. USENIX Association. Proc. ACM Program. Lang., Vol. 8, No. OOPSLA2, Article 317",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Lianmin Zheng, Zhuohan Li, Hao Zhang, Yonghao Zhuang, Zhifeng Chen, Yanping Huang, Yida Wang, Yuanzhong Xu, Danyang Zhuo, Eric P. Xing, Joseph E. Gonzalez, and Ion Stoica. 2022. Alpa: Automating Interand Intra-Operator Parallelism for Distributed Deep Learning. In 16th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Liyan Zheng, Haojie Wang, Jidong Zhai, Muyan Hu, Zixuan Ma, Tuowei Wang, Shuhong Huang, Xupeng Miao, Shizhi Tang, Kezhao Huang, and Zhihao Jia. 2023. EINNET: Optimizing Tensor Programs with Derivation-Based Transformations. In 17th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Ningxin Zheng, Huiqiang Jiang, Quanlu Zhang, Zhenhua Han, Lingxiao Ma, Yuqing Yang, Fan Yang, Chengruidong Zhang, Lili Qiu, Mao Yang, and Lidong Zhou. 2023. PIT: Optimization of Dynamic Sparse Deep Learning Models via Permutation Invariant Transformation. In 29th Symposium on Operating Systems Principles, SOSP 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Chijin Zhou, Mingzhe Wang, Jie Liang, Zhe Liu, and Yu Jiang. 2020. Zeror: Speed Up Fuzzing with Coverage-sensitive Tracing and Scheduling. In 35th IEEE/ACM International Conference on Automated Software Engineering, ASE 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Chijin Zhou, Quan Zhang, Lihua Guo, Mingzhe Wang, Yu Jiang, Qing Liao, Zhiyong Wu, Shanshan Li, and Bin Gu. 2023. Towards Better Semantics Exploration for Browser Fuzzing. Proc. ACM Program. Lang. 7, OOPSLA2 (2023), 604–631.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Chijin Zhou, Quan Zhang, Bingzhou Qian, and Yu Jiang. 2025. Janus: Detecting Rendering Bugs in Web Browsers via Visual Delta Consistency. In 47th IEEE/ACM International Conference on Software Engineering, ICSE 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Chijin Zhou, Quan Zhang, Mingzhe Wang, Lihua Guo, Jie Liang, Zhe Liu, Mathias Payer, and Yu Jiang. 2022. Minerva: browser API fuzzing with dynamic mod-ref analysis. In 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Hongyu Zhu, Ruofan Wu, Yijia Diao, Shanbin Ke, Haoyu Li, Chen Zhang, Jilong Xue, Lingxiao Ma, Yuqing Xia, Wei Cui, Fan Yang, Mao Yang, Lidong Zhou, Asaf Cidon, and Gennady Pekhimenko. 2022. ROLLER: Fast and Efficient Tensor Compilation for Deep Learning. In 16th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2022. Received 2024-04-05; accepted 2024-08-18 Proc. ACM Program. L",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 49,
+      "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. In 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, ASPLOS 2022. ACM.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 21,22,49]",
+      "technique": null,
+      "sentence": "Traditional compiler testing techniques [ 21,22,49] often construct equivalent programs by inserting dead code, e.",
+      "context_before": "sors. These changes are supposed not to influence the outputs. If these two models, when passed to the same compiler in the same execution environment, produce different results for the same inputs, it indicates the presence of a mis-compilation bug in the compiler. This approach eliminates the need for a separate reference executor, thereby avoiding the false positives often caused by cross-platform numerical accuracy variations, enhancing the reliability of bug detection. Constructing equivalent tensor programs that can efficiently detect mis-compilation bugs is challenging. (1) Equivalence.",
+      "context_after": "g., if(false){...}, into a given program. However, this approach cannot be directly applicable to tensor programs. Tensor compilers primarily analyze the flow of tensors, and as such, ignore control-flow-wise dead code by design. An intuitive thought is to modify the tensors’ flow of a tensor program, but it is non-trivial to maintain programs’ equivalence after the modification. (2) Test Efficiency. The testing throughput for tensor compilers is much lower than for other applications (roughly 4 testcases per second [ 27] v.s. 10,000 testcases per second [ 72]). Therefore, each pair of equival",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 8459,
+      "cited_reference": {
+        "number": 49,
+        "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. In 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, ASPLOS 2022. ACM.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "For database engine testing, PQS [ 43] was introduced to generate queries that are designed to retrieve a specific data entity for verifying the integrity of the database engine; Mozi [ 8,26,58] constructs equivalent database engines through adjusting configurations to detect correctness bugs.",
+      "context_before": "-preserving code into Java programs to efficiently explore compilation space in order to detect crashes or mis-compilation bugs in JVM. Our work differs from these techniques in that we focus on tensor compilers and our proposed test oracle is based on tensor-compiler domain-specific knowledge, i.e., arithmetic and structural invariant instead of dead code insertion. New Oracle for Testing Other Systems. Since correctness bugs are often silent in complex systems, many research efforts are paid to combine domain knowledge to design new test oracle to detect correctness bugs in specific systems.",
+      "context_after": "For operating system testing, B3 [ 32] and Hydra [ 17] finds different test oracle such as formal specifications and crash consistencies to validate correctness of filesystems; BVF [ 46] leverages structured and sanitized Programs to verify the correctness of eBPF programs. For web browser testing, Janus [ 73–75] leverages rendering change statuses of two web pages to conduct differential testing to detect rendering logic bugs in web browsers. Our work serves as a new test oracle for tensor compilers, which can inspire the design of test oracles for other complex systems. Term Rewriting Syste",
+      "section": "8 Related Work",
+      "page": 23,
+      "char_offset": 88099,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:19:46Z",
+    "is_model_written": true,
+    "summary": "PolyJuice detects mis-compilation bugs in tensor compilers by constructing semantically equivalent computation graphs and checking the compiler treats them alike. The difficulty is producing equivalent graphs that exercise different optimisation logic, which the authors address with arithmetic and structural rewrite rules and an equality saturation framework that finds the most simplified and most complex equivalent graphs. Across five industrial and two academic compilers it found 84 non-crash mis-compilation bugs.",
+    "narrative": "One citation in related work describing PQS as generating queries designed to retrieve a specific data entity for verification -- the database counterpart to PolyJuice's equivalent-graph oracle.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3690631.json b/_data/papers/paper_doi_10_1145_3690631.json
new file mode 100644
index 0000000..acfee61
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3690631.json
@@ -0,0 +1,479 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:25:45Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3690631",
+    "title": "T-Rec: Fine-Grained Language-Agnostic Program Reduction Guided by Lexical Syntax",
+    "authors": [
+      "Zhenyang Xu",
+      "Yongqiang Tian",
+      "Mengxiao Zhang",
+      "Jiarui Zhang",
+      "Puzhuo Liu",
+      "Yu Jiang",
+      "Chengnian Sun"
+    ],
+    "year": 2024,
+    "venue": "ACM Transactions on Software Engineering and Methodology",
+    "doi": "10.1145/3690631",
+    "arxiv_id": null,
+    "s2_paper_id": "c7ca7db1f2ed7eafe81c58bf6732efb95bee0146",
+    "url": "https://doi.org/10.1145/3690631",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3690631",
+      "retrieved_at": "2026-09-08T15:25:45Z",
+      "chars": 106966,
+      "content_sha256": "sha256:c631a086f5730301b4d0cfc67116c226f94dbf8927c4290e20f1b9f6a23b9166"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 31,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3828
+      },
+      {
+        "number": "2",
+        "title": "Motivating Example",
+        "start": 14630
+      },
+      {
+        "number": "3",
+        "title": "Preliminaries",
+        "start": 18254
+      },
+      {
+        "number": "3.1",
+        "title": "Program Reduction",
+        "start": 18270
+      },
+      {
+        "number": "3.2",
+        "title": "Lexical Syntax",
+        "start": 25053
+      },
+      {
+        "number": "4",
+        "title": "Methodology",
+        "start": 26781
+      },
+      {
+        "number": "4.1",
+        "title": "Definition of Canonical Tokens",
+        "start": 28739
+      },
+      {
+        "number": "4.2",
+        "title": "Replacement-Based Token Canonicalization",
+        "start": 33017
+      },
+      {
+        "number": "4.3",
+        "title": "Deletion-Based Token Canonicalization",
+        "start": 37751
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 48184
+      },
+      {
+        "number": "5.1",
+        "title": "RQ1: Effectiveness in De-Duplication",
+        "start": 55693
+      },
+      {
+        "number": "5.2",
+        "title": "RQ2: Effectiveness in Reducing Byte Size",
+        "start": 62660
+      },
+      {
+        "number": "5.3",
+        "title": "RQ3: Effectiveness in Removing Tokens",
+        "start": 69944
+      },
+      {
+        "number": "5.4",
+        "title": "RQ4: Efficiency",
+        "start": 74181
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 80177
+      },
+      {
+        "number": "6.1",
+        "title": "Comparing to Perses with Lex Tree Extension",
+        "start": 80368
+      },
+      {
+        "number": "6.2",
+        "title": "Threats to Validity",
+        "start": 82569
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 84787
+      },
+      {
+        "number": "7.1",
+        "title": "Language-Agnostic Program Reducers",
+        "start": 84908
+      },
+      {
+        "number": "7.2",
+        "title": "Language-Specific Program Reducers",
+        "start": 90334
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 92679
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "ANTLR. 2017. The ANTLR Parser Generator. Retrieved September 20, 2022 from https://www.antlr.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Cornelius Aschermann, Tommaso Frassetto, Thorsten Holz, Patrick Jauernig, Ahmad-Reza Sadeghi, and Daniel Teuchert. 2019. NAUTILUS: Fishing for deep bugs with grammars. In Proceedings of the 26th Annual Network and DistributedSystemSecuritySymposium(NDSS’19). The Internet Society. Retrieved from https://www.ndss-symposium. org/ndss-paper/nautilus-fishing-for-deep-bugs-with-grammars/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Yang Chen, Alex Groce, Chaoqiang Zhang, Weng-Keen Wong, Xiaoli Fern, Eric Eide, and John Regehr. 2013. Taming compiler fuzzers. In Proceedings of the 2013 ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI ’13), 197–208.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Nathan Chong, Alastair Donaldson, Andrei Lascu, and Christopher Lidbury. 2015. Many-core compiler fuzzing. In Proceedingsofthe35thACMSIGPLANConferenceonProgrammingLanguageDesignandImplementation(PLDI’15), 65–76.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "CPython. 2022. Bug Report. Retrieved September 20, 2022 from https://github.com/python/cpython/issues/new? assignees={&}labels=type-bug{&}template=bug.md",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Alastair Donaldson and David MacIver. 2021. Test Case Reduction: Beyond Bugs. Retrieved May 29, 2023 from https://blog.sigplan.org/2021/05/25/test-case-reduction-beyond-bugs",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "GCC. 2020. A Guide to Testcase Reduction. Retrieved October 28, 2022 from https://gcc.gnu.org/wiki/A_guide_to_ testcase_reduction",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "GCC-Wiki. 2020. A Guide to Testcase Reduction. Retrieved September 20, 2022 from https://gcc.gnu.org/wiki/A_ guide_to_testcase_reduction",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "GolnazGharachorluandNickSumner.2023.Typebatchedprogramreduction.In Proceedingsofthe32ndACMSIGSOFT International Symposium on Software Testing and Analysis, 398–410.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Rahul Gopinath, Alexander Kampmann, Nikolas Havrikov, Ezekiel O. Soremekun, and Andreas Zeller. 2020a. Abstracting failure-inducing inputs. In Proceedingsofthe29thACMSIGSOFTInternationalSymposiumonSoftwareTesting and Analysis, 237–248.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Rahul Gopinath, Alexander Kampmann, Nikolas Havrikov, Ezekiel O. Soremekun, and Andreas Zeller. 2020b. The DDSET Github Repo. Retrieved June 20, 2024 from https://github.com/vrthra/ddset ACM Transactions on Software Engineering and Methodology, Vol. 34, No. 2, Article 34. Publication date: January 2025. 34:30 Z. Xu et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Alex Groce, Josie Holmes, and Kevin Kellar. 2017. One test to rule them all. In Proceedings of the 26th ACMSIGSOFT InternationalSymposiumonSoftwareTestingandAnalysis(ISSTA’17). ACM, New York, NY, USA, 1–11. DOI:https: //doi.org/10.1145/3092703.3092704",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "KihongHeo,WoosukLee,PardisPashakhanloo,andMayurNaik.2018.Effectiveprogramdebloatingviareinforcement learning. In Proceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security (CCS ’18). David Lie, Mohammad Mannan, Michael Backes, and XiaoFeng Wang (Eds.), ACM, 380–394. DOI:https://doi.org/10. 1145/3243734.3243838",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Renáta Hodován and Ákos Kiss. 2016. Modernizing hierarchical delta debugging. In Proceedingsofthe7thInternational Workshop on AutomatingTest Case Design, Selection, and Evaluation (A-TEST’16). ACM, New York, NY, USA, 31–37. DOI:https://doi.org/10.1145/2994291.2994296",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Renáta Hodován, Ákos Kiss, and Tibor Gyimóthy. 2017. Coarse hierarchical delta debugging. In Proceedingsofthe. IEEE Computer Society, 194–203. DOI:https://doi.org/10.1109/ICSME.2017.26",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Renáta Hodován, Ákos Kiss, and Tibor Gyimóthy. 2018. Grammarinator: A grammar-based open source fuzzer. In Proceedingsofthe9thACMSIGSOFTInternationalWorkshoponAutomatingTESTCaseDesign,Selection,andEvaluation (A-TEST ’18). ACM, New York, NY, USA, 45–48. DOI:https://doi.org/10.1145/3278186.3278193",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "JerryScript. 2022. Bug Report. Retrieved September 20, 2022 from https://github.com/jerryscript-project/jerryscript/ blob/master/.github/ISSUE_TEMPLATE/bug_report.md",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "JS Delta. 2017. JS Delta. Retrieved October 28, 2022 from https://github.com/wala/jsdelta",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Christian Gram Kalhauge and Jens Palsberg. 2019. Binary reduction of dependency graphs. In Proceedings of the ACM JointMeetingonEuropeanSoftwareEngineeringConferenceandSymposiumontheFoundationsofSoftwareEngineering (ESEC/SIGSOFT FSE ’19). Marlon Dumas, Dietmar Pfahl, Sven Apel, and Alessandra Russo (Eds.), ACM, 556–566. DOI: https://doi.org/10.1145/3338906.3338956",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Christian Gram Kalhauge and Jens Palsberg. 2021. Logical bytecode reduction. In Proceedingsofthe PLDI’21:42nd ACMSIGPLANInternationalConferenceonProgrammingLanguageDesignandImplementation. Stephen N. Freund and Eran Yahav (Eds.), ACM, 1003–1016. DOI:https://doi.org/10.1145/3453483.3454091",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Ákos Kiss, Renáta Hodován, and Tibor Gyimóthy. 2018. HDDr: A recursive variant of the hierarchical delta debugging algorithm. In Proceedings of the 9th ACMSIGSOFT International Workshop on Automating TEST Case Design, Selection, and Evaluation (A-TEST ’18). ACM, New York, NY, USA, 16–22. DOI:https://doi.org/10.1145/3278186.3278189",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Gereon Kremer, Aina Niemetz, and Mathias Preiner. 2021. dd SMT 2.0: Better delta debugging for the SMT-LIBv2 language and friends. In Proceedings of the Computer Aided Verification-33rd International Conference (CAV ’21). Alexandra Silva and K. Rustan M. Leino (Eds.), Part II (Lecture Notes in Computer Science, Vol. 12760), Springer, 231–242. DOI:https://doi.org/10.1007/978-3-030-81688-9_11",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. 216–226. DOI: https://doi.org/10.1145/2594291.2594334",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Cong Li, Yanyan Jiang, Chang Xu, and Zhendong Su. 2023. Validating JIT compilers via compilation space exploration. InProceedingsofthe29thSymposiumonOperatingSystemsPrinciples (SOSP’23). ACM, New York, NY, USA, 66–79. DOI:https://doi.org/10.1145/3600006.3613140",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Vsevolod Livinskii, Dmitry Babokin, and John Regehr. 2020. Random testing for C and C++ compilers with YARPGen. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "LLVM. 2022. How to Submit an LLVM Bug Report. Retrieved September 20, 2022 from https://llvm.org/docs/ HowToSubmitABug.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "David R. MacIver and Alastair F. Donaldson. 2020. Test-case reduction via test-case generation: Insights from the hypothesisreducer(toolinsightspaper).In Proceedingsofthe34thEuropeanConferenceonObject-OrientedProgramming (ECOOP ’20). Schloss-Dagstuhl-Leibniz Zentrum für Informatik.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "GhassanMisherghiandZhendongSu.2006.HDD:HierarchicalDeltaDebugging.In Proceedingsofthe28thInternational ConferenceonSoftwareEngineering(ICSE’06). Leon J. Osterweil, H. Dieter Rombach, and Mary Lou Soffa (Eds.), ACM, 142–151. DOI:https://doi.org/10.1145/1134285.1134307",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Aina Niemetz and Armin Biere. 2013. dd SMT: A delta debugger for the SMT-LIB v2 format. In Proceedings of the 11th International Workshop on Satisfiability Modulo Theories, SMT, 8–9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "John Regehr. 2019. Design and Evolution of C-Reduce. Retrieved January 20, 2024 from https://blog.regehr.org/ archives/1679",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "John Regehr, Yang Chen, Pascal Cuoq, Eric Eide, Chucky Ellison, and Xuejun Yang. 2012. Test-case reduction for C compiler bugs. In Proceedings of the 33rd ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI ’12). ACM, New York, NY, USA, 335–346. DOI:https://doi.org/10.1145/2254064.2254104",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedingsofthe28thACMJointMeetingonEuropeanSoftwareEngineeringConferenceand ACM Transactions on Software Engineering and Methodology, Vol. 34, No. 2, Article 34. Publication date: January 2025. T-Rec: Fine-Grained Language-AgnosticProgramReduction 34:31 Sympos",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 33,
+      "text": "Rust. 2024. Finding a Minimal, Standalone Example. Retrieved Januray 20, 2024 from https://rustc-dev-guide.rustlang.org/notification-groups/cleanup-crew.html#finding-a-minimal-standalone-example",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Prashast Srivastava and Mathias Payer. 2021. Gramatron: Effective grammar-aware fuzzing. In Proceedings of the 30th ACMSIGSOFT International Symposium on Software Testing and Analysis, 244–256.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. 2016. Finding compiler bugs via live code mutation. In Proceedingsofthe 2016 ACMSIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications (OOPSLA’16,partof SPLASH’16). Eelco Visser and Yannis Smaragdakis (Eds.), ACM, 849–863. DOI:https://doi.org/10. 1145/2983990.2984038",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Chengnian Sun, Yuanbo Li, Qirun Zhang, Tianxiao Gu, and Zhendong Su. 2018. Perses: Syntax-guided program reduction. In Proceedings of the 40th International Conference on Software Engineering (ICSE ’18). ACM, New York, NY, USA, 361–371. DOI:https://doi.org/10.1145/3180155.3180236",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Yongqiang Tian, Xueyan Zhang, Yiwen Dong, Zhenyang Xu, Mengxiao Zhang, Yu Jiang, Shing-Chi Cheung, and Chengnian Sun. 2023. On the caching schemes to speed up program reduction. ACM Transactions on Software Engineering and Methodology 33, 1, Article 17 (Nov 2023), 30 pages. DOI: https://doi.org/10.1145/3617172",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Dániel Vince. 2022. Iterating the minimizing Delta debugging algorithm. In Proceedings of the 13th International Workshop on Automating Test Case Design, Selection and Evaluation (A-TEST ’22). Ákos Kiss, Beatriz Marín, and Mehrdad Saadatmand (Eds.), ACM, 57–60. DOI:https://doi.org/10.1145/3548659.3561314",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Dániel Vince, Renáta Hodován, Daniella Bársony, and Ákos Kiss. 2021. Extending hierarchical delta debugging with hoisting. In Proceedingsofthe 2021IEEE/ACMInternationalConferenceonAutomationofSoftwareTest(AST). IEEE, 60–69.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Guancheng Wang, Ruobing Shen, Junjie Chen, Yingfei Xiong, and Lu Zhang. 2021. Probabilistic Delta debugging. In Proceedings of the ESEC/FSE ’21: 29th ACM Joint European Software Engineering Conference and Symposium on the FoundationsofSoftwareEngineering. Diomidis Spinellis, Georgios Gousios, Marsha Chechik, and Massimiliano Di Penta (Eds.), ACM, 881–892. DOI:https://doi.org/10.1145/3468264.346862",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Frank Wilcoxon. 1992. Individual Comparisons by Ranking Methods. Springer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "DominikWinterer,ChengyuZhang,andZhendongSu.2020.ValidatingSMTsolversviasemanticfusion.In Proceedings ofthe41stACMSIGPLANConferenceonProgrammingLanguageDesignandImplementation (PLDI’20). ACM, New York, NY, USA, 718–730. DOI:https://doi.org/10.1145/3385412.3385985",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Zhenyang Xu, Yongqiang Tian, Mengxiao Zhang, Gaosen Zhao, Yu Jiang, and Chengnian Sun. 2023. Pushing the limit of 1-minimality of language-agnostic program reduction. Proceedings of the ACM on Programming Languages 7, OOPSLA1, Article 97 (Apr 2023), 29 pages. DOI:https://doi.org/10.1145/3586049",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In Proceedingsofthe32ndACMSIGPLANConferenceonProgrammingLanguageDesignandImplementation(PLDI’11). ACM, New York, NY, USA, 283–294. DOI:https://doi.org/10.1145/1993498.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "A. Zeller and R. Hildebrandt. 2002. Simplifying and isolating failure-inducing input. IEEETransactionsonSoftware Engineering 28, 2 (2002), 183–200. DOI:https://doi.org/10.1109/32.988498",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Mengxiao Zhang, Yongqiang Tian, Zhenyang Xu, Yiwen Dong, Shin Hwei Tan, and Chengnian Sun. 2024. LPR: Large language models-aided program reduction. In Proceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis, 261–273.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "MengxiaoZhang,ZhenyangXu,YongqiangTian,YuJiang,andChengnianSun.2023.PPR:Pairwiseprogramreduction. InProceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 338–349. Received 8 February 2024; revised 5 July 2024; accepted 10 August 2024 ACM Transactions on Software Engineering and Methodology, Vol. 34, No. 2, Article 34. P",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedingsofthe28thACMJointMeetingonEuropeanSoftwareEngineeringConferenceand ACM Transactions on Software Engineering and Methodology, Vol. 34, No. 2, Article 34. Publication date: January 2025. T-Rec: Fine-Grained Language-AgnosticProgramReduction 34:31 SymposiumontheFoundationsofSoftwareEngineering (ESEC/FSE’20). ACM, New York, NY, USA, 1140–1152. DOI: https://doi.org/10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 7,23–26,32,33,42]",
+      "technique": "norec",
+      "sentence": "It is a widely recognized and powerful reducer commonly employed in compiler testing work and programming language communities [ 7,23–26,32,33,42].",
+      "context_before": "pecific knowledge, and thus they are more generalized. ACM Transactions on Software Engineering and Methodology, Vol. 34, No. 2, Article 34. Publication date: January 2025. T-Rec: Fine-Grained Language-AgnosticProgramReduction 34:3 C programs. One possible explanation is that when the inputs have strict restrictions imposed upon on each token by the semantics or the property (i.e., triggering a certain bug), the abstracting process may become infeasible, and thus the effectiveness of ddsetmay significantly deteriorate. C-Reduce [ 31] is another reducer that attempts to overcome the limitation.",
+      "context_after": "C-Reduce strives to reduce identifiers, strings, and numbers to their canonical forms (more details in Section 3.1.1). However, the approach used by C-Reduce is specifically optimized for reducing C/C++ programs. Although some parts of its approach are language-agnostic, their performance tends to be limited when the language is less similar to C/C++, and thus could benefit from further enhancements. Practical Benefits of Fine-Grained Reduction. Fine-grained reduction improves program reduction in two aspects. First, it can minimize tokens such as identifiers and literals to canonical tokens",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 8248,
+      "cited_reference": {
+        "number": 32,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedingsofthe28thACMJointMeetingonEuropeanSoftwareEngineeringConferenceand ACM Transactions on Software Engineering and Methodology, Vol. 34, No. 2, Article 34",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "T-Rec is a language-agnostic program reduction technique guided by lexical syntax. Program reduction removes bug-irrelevant code from a bug-triggering program, giving a smaller reproducer and making duplicates easier to tell apart. The authors observe that existing language-agnostic reducers treat tokens as atomic and so miss reduction opportunities inside them; T-Rec reduces within tokens, removing up to 65.52% more bytes than Perses on a multi-lingual benchmark.",
+    "narrative": "A single citation, among the compiler-testing and programming-language work that commonly employs program reducers.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3704870.json b/_data/papers/paper_doi_10_1145_3704870.json
new file mode 100644
index 0000000..411250a
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3704870.json
@@ -0,0 +1,148 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:01:46Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3704870",
+    "title": "The Decision Problem for Regular First Order Theories",
+    "authors": [
+      "Umang Mathur",
+      "David Mestel",
+      "Mahesh Viswanathan"
+    ],
+    "year": 2024,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3704870",
+    "arxiv_id": "2410.17185",
+    "s2_paper_id": "f5df824295f8a8bcf107f122c26590b761e8b9f9",
+    "url": "https://doi.org/10.1145/3704870",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2410.17185",
+      "retrieved_at": "2026-09-07T17:01:46Z",
+      "chars": 129551,
+      "content_sha256": "sha256:81d056dd7ea2140058197dc14c01d5ba1606d901c9651098ac8f94ae2e3a1252"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 31,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1557
+      },
+      {
+        "number": "1.1",
+        "title": "Our Contributions",
+        "start": 7536
+      },
+      {
+        "number": "3.1",
+        "title": "Firstorder logicand its fragments",
+        "start": 18396
+      },
+      {
+        "number": "3.3",
+        "title": "Regulartheories",
+        "start": 27011
+      },
+      {
+        "number": "5.1",
+        "title": "The EPR class",
+        "start": 43682
+      },
+      {
+        "number": "5.2",
+        "title": "Existential formulae with unaryfunctions",
+        "start": 50844
+      },
+      {
+        "number": "6.1",
+        "title": "Purely existential andpurely universal function-free theories",
+        "start": 60073
+      },
+      {
+        "number": "6.2",
+        "title": "Conjunctivesatisfiability of EUF formulae",
+        "start": 61646
+      },
+      {
+        "number": "7.1",
+        "title": "Coherentformulae",
+        "start": 76802
+      },
+      {
+        "number": "7.2",
+        "title": "Executions asformulae",
+        "start": 91629
+      },
+      {
+        "number": "A",
+        "title": "Let the acceptingstates of the A/u1D456be denoted by /u1D45E/u1D456.",
+        "start": 116482
+      }
+    ]
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:23:30Z",
+    "is_model_written": true,
+    "summary": "This paper studies the decision problem for regular first order theories, establishing decidability results for a class of logical theories. It is a theory paper rather than a systems one.",
+    "narrative": "No SQLancer mention could be extracted, and no bibliography parsed. The subject is mathematical logic rather than database testing, so the paper is most likely in the citing set through a shared citation rather than through any relationship to SQLancer.",
+    "roles": {},
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      },
+      "extends_technique": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      },
+      "compares_with": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "insufficient_evidence",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The extracted text contains no sentence naming SQLancer or one of its techniques, and no bibliography entry resolved to a SQLancer publication. With no mention to reason from, no relationship can be judged either way."
+      }
+    },
+    "disagreements": [],
+    "unresolved": [
+      "No bibliography parsed and no mention found. Given the subject, its presence in the citing set should be re-checked."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3708533.json b/_data/papers/paper_doi_10_1145_3708533.json
new file mode 100644
index 0000000..8592129
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3708533.json
@@ -0,0 +1,1286 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:36:56Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3708533",
+    "title": "Software Security Analysis in 2030 and Beyond: A Research Roadmap",
+    "authors": [
+      "Marcel Böhme",
+      "Eric Bodden",
+      "T. Bultan",
+      "Cristian Cadar",
+      "Yang Liu",
+      "Giuseppe Scanniello"
+    ],
+    "year": 2024,
+    "venue": "ACM Transactions on Software Engineering and Methodology",
+    "doi": "10.1145/3708533",
+    "arxiv_id": "2409.17844",
+    "s2_paper_id": "92276c9868f6488744e40c18835c76b8588a2d35",
+    "url": "https://doi.org/10.1145/3708533",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2409.17844",
+      "retrieved_at": "2026-09-08T14:36:56Z",
+      "chars": 109581,
+      "content_sha256": "sha256:846047b696b9539292af645459b085cf6831e5e0316701bfd59269c1adc4a663"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 24,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1940
+      },
+      {
+        "number": "2.1",
+        "title": "So/f_tware Analysis Techniques",
+        "start": 4283
+      },
+      {
+        "number": "2.2",
+        "title": "Detected Errors",
+        "start": 10605
+      },
+      {
+        "number": "3.1",
+        "title": "Overview: So/f_tware Systems ofthe Future",
+        "start": 12885
+      },
+      {
+        "number": "3.2",
+        "title": "Machine LearningandSecurity",
+        "start": 16148
+      },
+      {
+        "number": "3.3",
+        "title": "Security of Evolving So/f_tware",
+        "start": 24772
+      },
+      {
+        "number": "3.4",
+        "title": "Supply ChainSecurity",
+        "start": 26034
+      },
+      {
+        "number": "3.5",
+        "title": "Beyond Memory Safety",
+        "start": 42897
+      },
+      {
+        "number": "3.6",
+        "title": "Beyond Monolithic, Fully Virtual So/f_tware Systems",
+        "start": 49025
+      },
+      {
+        "number": "3.7",
+        "title": "Resilient Computing",
+        "start": 54877
+      },
+      {
+        "number": "3.8",
+        "title": "EmergentBehaviors andVulnerability Composition",
+        "start": 57564
+      },
+      {
+        "number": "3.9",
+        "title": "Security Education",
+        "start": 60609
+      },
+      {
+        "number": "3.10",
+        "title": "LawandPolicy Making",
+        "start": 62756
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "[n.d.].WhatIsSoftwareSupplyChainSecurityandHowDo esItWork?|Synopsys. https://www.synopsys.com/glossary/what-is-softwaresupply-chain-security.html. (Accessed on03/04/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2024.Bestpracticesfordependencymanagement|Google CloudBlog. https://cloud.google.com/blog/topics/developers-pra ctitioners/best-practices-dependency-management. (Accessed on03/14/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2024. Dependabot. https://github.com/dependabot. (Accessed on03/14/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2024. Home ·Sigstore. https://www.sigstore.dev/. (Accessed on 03/14/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2024. NVD-SWID. https://nvd.nist.gov/products/swid. (Accessed on 03/15/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2024. ossf/criticality_score:Givescriticalityscor eforanopensourceproject. https://github.com/ossf/criticality_score. (Accessedon03/14/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2024. ossf/package-manager-best-practices: Collect ion of security best practices for package managers. https://github.com/ossf/package-manager-best-practi ces/tree/main. (Accessed on 03/14/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2024. ossf/scorecard:OpenSSF Scorecard-Securityhea lthmetricsforOpenSource. https://github.com/ossf/scorecard. (Accessedon03/14/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2024. OWASP CycloneDXSoftware Bill of Materials(SBOM) Standard. https://cyclonedx.org/. (Accessed on03/15/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2024. OWASP Dependency-Check | OWASP Foundation. https://owasp.org/www-project-dependency-check/. (Accessed on 03/14/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2024. SecureSoftware Development Framework|CSRC. https://csrc.nist.gov/projects/ssdf. (Accessed on03/14/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2024. SLSA • Supply-chainLevelsfor SoftwareArtifact s.https://slsa.dev/. (Accessed on 03/14/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "2024. SPDX –LinuxFoundation Projects Site. https://spdx.dev/. (Accessed on03/15/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "2024. SPIFFE –SecureProduction Identity Frameworkfo rEveryone. https://spiffe.io/. (Accessed on03/14/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Martín Abadi, Mihai Budiu, Ulfar Erlingsson, and Jay Li gatti. 2009. Control-flow integrity principles, implement ations, and applications. ACM TransactionsonInformation and SystemSecurity (TISSEC) 13, 1(2009), 1–40.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Periklis Akritidis, Cristian Cadar, Costin Raiciu, Ma nuel Costa, and Miguel Castro. 2008. Preventing memory erro r exploits with WIT. In .IEEE, 263–277.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "aleak[n.d.].AnothermisconfiguredAmazonS3serverle aksdataof50,000Australians. https://www.scmagazineuk.com/another-misconfigured-a mazon\\-s3-server-leaks-data-of-50000-austra",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "MuathAlkhalaf,AbdulbakiAydin,andTevfikBultan.201 4. Semanticdifferentialrepairforinputvalidationandsan itization.In Proceedingsofthe 2014International Symposium on SoftwareTestingand Analy sis.225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Kevin Allix, Tegawendé F Bissyandé, Quentin Jérome, Ja cques Klein, Radu State, and Yves Le Traon. 2016. Empirical a ssessment of machine learning-based malware detectors for Android: Measuring t he gap between in-the-lab and in-the-wild validation scena rios.Empirical Software Engineering 21(2016), 183–211.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "MajedAlmansoori,JessicaLam,EliasFang,KieranMull igan,AdalbertGeraldSoosaiRaj,andRahulChatterjee.202 0.HowSecureareourComputer SystemsCourses?.In Proceedings of Conferenceon International ComputingEduca tionResearch .ACM,271–281.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Anastasios Andronidis and CristianCadar.2022. SnapF uzz: High-Throughput Fuzzing of Network Applications. In ACMSIGSOFT International Symposiumon Software Testingand Analysis (ISSTA 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Daniel Arp, Erwin Quiring, Feargus Pendlebury, Alexan der Warnecke, Fabio Pierazzi, Christian Wressnegger, Lore nzo Cavallaro, and Konrad Rieck.2022. Dosand Don’ts ofMachine LearninginComputerS ecurity.In Proc.of USENIXSecurity Symposium. Manuscriptsubmitted to ACM 18 M.Böhme, E.Bodden,T.Bultan,C.Cadar,Y.Liu,and G.Scan niello",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "StevenArztandEricBodden.2014. Reviser:efficientlyu pdating IDE-/IFDS-baseddata-flowanalysesinresponsetoi ncrementalprogramchanges. InProceedings of the 36thInternational Conferenceon Softwar eEngineering. 288–298.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "OwuraAsare,MeiyappanNagappan,and NAsokan. 2023. Is github’scopilot asbadashumans atintroducing vulnerabil itiesincode? Empirical SoftwareEngineering 28, 6(2023), 129.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "azureflaw[n.d.].MicrosoftAzurecloudvulnerability isthe‘worstyoucanimagine’. https://www.theverge.com/2021/8/27/22644161/microso ft-azure-database-vulnerabilty-chaosdb?f",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "John Backes, Pauline Bolignano, Byron Cook, Catherine Dodge, Andrew Gacek, Kasper Luckow, Neha Rungta, Oksana Tka chu, and Carsten Varming. 2018. Semantic-based Automated Reasoning for AWS Access Policies using SMT. In Proceedings of the 18th Conference on Formal Methods inComputer-Aided Design(FMCAD2018),Austin,Tex as,USA,October30 -November2, 2018 .1–9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "MichaelBackes,BorisKöpf, and AndreyRybalchenko. 20 09. Automatic Discoveryand Quantificationof Information L eaks.In30thIEEESymposiumonSecurity and Privacy (S&P2009),17-20May 2009,Oakla nd, California, USA. 141–153.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "DeepikaBadampudi, MichaelUnterkalmsteiner,and Ric ardoBritto. 2023. Modern code reviews–surveyof literatur eand practice. ACMTransactionsonSoftware Engineering and Methodology 32, 4(2023), 1–61.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Roberto Baldoni, Emilio Coppa, Daniele Cono D’elia, Ca mil Demetrescu, and Irene Finocchi. 2018. A surveyof symbol ic execution techniques. ACMComputingSurveys(CSUR) 51,3(2018), 1–39.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "DavideBalzarotti,MarcoCova,VikaFelmetsger,Nenad Jovanovic,EnginKirda,ChristopherKruegel,andGiovanni Vigna.2008.Saner:Composing staticanddynamicanalysistovalidatesanitizationinweb applications.In 2008IEEESymposiumonSecurityandPrivacy(sp2008) .IEEE,387–401.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "LingfengBao,XinXia,AhmedEHassan,andXiaohuYang.2 022. V-SZZ:automaticidentificationofversionrangesaffec tedby CVEvulnerabilities. InProceedings of the 44thInternational Conferenceon Softwar eEngineering. 2352–2364.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Mohamed Ben-Daya,ElkafiHassini,and ZiedBahroun.201 9. Internet of things and supplychainmanagement: aliterat urereview. International journal of production research 57, 15-16(2019), 4719–4742.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "MarcelBöhme.2022.StatisticalReasoningAboutProgr ams.InProceedingsofthe44thInternationalConferenceonSoftwar eEngineering (Pittsburgh, USA)(ICSE2022) .5pages. https://doi.org/10.1145/3510455.3512796",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Marcel Böhme, Cristian Cadar, and Abhik Roychoudhury. 2021. Fuzzing: Challenges and Reflections. IEEE Software 38, 3 (2021), 79–86. https://doi.org/10.1109/MS.2020.3016773",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Marcel Böhme, Van-Thuan Pham, Manh-Dung Nguyen, and Ab hik Roychoudhury. 2017. Directed GreyboxFuzzing. In Proceedings of the 2017 ACMSIGSACConference onComputer and CommunicationsSecur ity(Dallas,Texas,USA) (CCS’17).2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "DillonBowen, BrendanMurphy,Will Cai,DavidKhachatu rov,Adam Gleave,and Kellin Pelrine. 2024. Scaling Lawsfor DataPoisoning in LLMs. arXiv preprint arXiv:2408.02946 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "SergeyBratus, MichaelLocasto, Meredith Patterson, L enSassaman,and Anna Shubina. 2011. From BufferOverflowsto W eird Machines. ;login: (December2011).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Tegan Brennan, Nicolás Rosner, and Tevfik Bultan. 2020. JIT Leaks: Inducing Timing Side Channels through Just-In-T ime Compilation. In 8-21,2020 .IEEE, 1207–1222.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "BobbyRBruce,TianyiZhang,JaspreetArora,GuoqingHa rryXu,andMiryungKim.2020. Jshrink:In-depthinvestigat ionintodebloatingmodern javaapplications. In Proceedings of the 28th ACM joint meeting on european softwar e engineering conference and symposium on the foundations o f softwareengineering. 135–146.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "TevfikBultan, Fang Yu,Muath Alkhalaf,and AbdulbakiAy din. 2017. String Analysisfor Software Verificationand Security. Springer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Frank Busse, Martin Nowack, and Cristian Cadar. 2020. R unning Symbolic Execution Forever. In ACMSIGSOFT International Symposium on SoftwareTestingand Analysis(ISSTA 2020) (Virtual Event). 63–74.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Cristian Cadar, Daniel Dunbar, and Dawson Engler. 2008. KLEE: Unassisted and Automatic Generation of High-Covera ge Tests for Complex SystemsPrograms.In USENIXSymposium onOperating SystemsDesignand Implement ation (OSDI 2008) (SanDiego, CA,USA). 209–224.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "CristianCadar,Vijay Ganesh, Peter Pawlowski, David D ill, and Dawson Engler. 2006. EXE: Automatically Generatin g Inputs of Death. In ACM Conferenceon Computerand CommunicationsSecurity (CCS 20 06)(Alexandria,VA, USA). 322–335.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Cristian Cadar, Patrice Godefroid, Sarfraz Khurshid, Corina S Păsăreanu, Koushik Sen, Nikolai Tillmann, and Will em Visser. 2011. Symbolic execution for software testing in practice: preliminary as sessment. In Proceedings of the 33rd International Conference on Softwar e Engineering. 1066–1071.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "CristianCadarandMartinNowack.2020.KLEESymbolicE xecutionEnginein2019. InternationalJournalonSoftwareToolsforTechnologyTra nsfer (2020)23,6 (2020), 867–870. https://doi.org/10.1007/s10009-020-00570-3",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Cristian Cadar, Luís Pina, and John Regehr. 2015. Multi-Version Execution Defeats a Compiler-Bug-Based Backdoor. https://ccadar.blogspot.co.uk/2015/11/multi-version -execution-defeats.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Cristian Cadar, Daniel Schemmel, and Arindam Sharma. 2 023. Patch Specifications via Product Programs. In 2023 International Conference on FormalMethods inSoftware Engineering (FormaliSE2023) (Melbourne,Australia).39–43. https://doi.org/10.1109/FormaliSE58978.2023.00012",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Cristian Cadar and Koushik Sen. 2013. Symbolic Executi on for Software Testing: Three Decades Later. Communications of the Association for ComputingMachinery(CACM2013) 56, 2(2013), 82–90.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "CristianoCalcagno,DinoDistefano,JérémyDubreil,D ominikGabi,PieterHooimeijer,MartinoLuca,PeterO’Hear n,IrenePapakonstantinou,Jim Purbrick,and DulmaRodriguez. 2015. Movingfastwith softw areverification.In NASAFormalMethods Symposium .Springer,3–11. Manuscriptsubmitted to ACM SoftwareSecurityAnalysis in2030and Beyond: AResearch Ro admap 19",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "cancan2015. ryanb/cancan• GitHub. https://github.com/ryanb/cancan.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Anthony James Cartwright. 2023. The elephant in the roo m: cybersecurity in healthcare. Journal of Clinical Monitoring and Computing 37, 5 (2023), 1123–1132.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "SaikatChakraborty,RahulKrishna,YangruiboDing,an dBaishakhiRay.2021. Deeplearningbasedvulnerabilityde tection:Arewethereyet. IEEE TransactionsonSoftware Engineering (2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "S. Chakraborty, R. Krishna, Y. Ding, and B. Ray. 2022. De ep Learning Based Vulnerability Detection: Are We There Yet ?IEEE Transactions on SoftwareEngineering 48, 09(Sept. 2022),3280–3296. https://doi.org/10.1109/TSE.2021.3087402",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "YangChen,AndrewESantosa,AsankhayaSharma,andDavi dLo.2020.Automatedidentificationoflibrariesfromvulne rabilitydata.In Proceedings ofthe ACM/IEEE42nd International Conference on SoftwareE ngineering: Software Engineeringin Practice .90–99.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "ClangStatic Analyzer[n.d.]. ClangStatic Analyzer. https://clang-analyzer.llvm.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Lori A. Clarke. 1976. A system to generate test data and s ymbolically execute programs. IEEE Transactions on software engineering 3 (1976), 215–222.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "MichaelRClarksonand Fred B Schneider.2010. Hyperpro perties.Journal of ComputerSecurity 18,6 (2010), 1157–1210.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "JamesClause,WanchunLi,andAlessandroOrso.2007. Dy tan:agenericdynamictaintanalysisframework.In Proceedingsofthe2007international symposiumonSoftware testingand analysis. 196–206.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "CLion2024. CLionIDE. https://www.jetbrains.com/clion/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Domenico Cotroneo, Cristina Improta, Pietro Liguori, and Roberto Natella. 2024. Vulnerabilities in ai code gener ators: Exploring targeted data poisoning attacks.In Proceedings of the 32nd IEEE/ACMInternational Conference o nProgramComprehension .280–292.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "PatrickCousot.1996. Abstractinterpretation. ACMComputingSurveys(CSUR) 28,2 (1996), 324–328.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Coveritysoftware[n.d.]. Coveritysoftware. http://www.coverity.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "CrispanCowan,CaltonPu,DaveMaier,JonathanWalpole ,PeatBakke,SteveBeattie,AaronGrier,PerryWagle,QianZ hang,andHeatherHinton. 1998. Stackguard:automatic adaptivedetection and preven tionof buffer-overflowattacks..In USENIX security symposium ,Vol. 98. SanAntonio, TX,63–78.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Olivier Crameri, Nikola Knezevic, Dejan Kostic, Ricar do Bianchini, and Willy Zwaenepoel. 2007. Staged deploymen t in Mirage, an integrated softwareupgradetesting and distributionsystem. SIGOPS Oper.Syst.Rev. 41,6 (oct 2007),221–236. https://doi.org/10.1145/1323293.1294283",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "CWETOP 25Most DangerousSoftware Errors.2024. https://www.sans.org/top25-software-errors/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Ozren Dabic, Emad Aghajani, and Gabriele Bavota. 2021. Sampling projects in github for MSR studies. In 2021 IEEE/ACM 18th International Conferenceon MiningSoftware Repositories (MSR) .IEEE, 560–564.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "JiarunDai,YuanZhang,HailongXu,HaimingLyu,Zichen gWu,XinyuXing,andMinYang.2021. FacilitatingVulnerabi lityAssessmentthrough PoC Migration. In Proceedings of the 2021 ACMSIGSAC Conference on Computer and CommunicationsSecurity (Virtual Event, Republic of Korea) (CCS’21).Associationfor Computing Machinery,NewYork,NY, USA, 33 00–3317. https://doi.org/10.1145/3460120.3484594",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Andreas Dann, Henrik Plate, Ben Hermann, Serena Elisa P onta, and Eric Bodden. 2021. Identifying challenges for oss vulnerability scanners-a study& test suite. IEEETransactionson SoftwareEngineering 48, 9(2021), 3613–3625.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Robin David, Sébastien Bardin, Thanh Dinh Ta, Laurent M ounier, Josselin Feist, Marie-Laure Potet, and Jean-Yves M arion. 2016. BINSEC/SE: A dynamic symbolic execution toolkit for binary-level anal ysis. In Reengineering (SANER) ,Vol. 1. IEEE, 653–656.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "LeonardoDeMouraand NikolajBjørner.2011. Satisfiabi litymodulo theories:introductionand applications. Commun.ACM 54,9 (2011),69–77.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Alexandre Decan, Tom Mens, Ahmed Zerouali,and Coen De R oover. 2021. Back to the past–analysing backporting practi cesin package dependency networks. IEEETransactionson Software Engineering 48,10(2021), 4087–4099.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "DorothyE Denning. 1987. An intrusion-detection model .IEEETransactionsonsoftwareengineering 2(1987), 222–232.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "djleak[n.d.]. CloudLeak:WSJParentCompanyDowJones ExposedCustomerData. https://www.upguard.com/breaches/cloud-leak-dow-jon es.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "ThomasDullien.2011.Weirdmachines,exploitability ,andprovableunexploitability. http://www.dullien.net/thomas/weird-machines-exploi tability.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "William Eiers, Ganesh Sankaran, Albert Li, Emily O’Mah ony, Benjamin Prince, and Tevfik Bultan. 2022. Quantifying P ermissivenessof Access Control Policies. In 44th IEEE/ACM44th International Conference on Software En gineering, ICSE 2022,Pittsburgh,PA, USA, May 25-27,2022. ACM, 1805–1817.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "European Parliament. 2022. Proposal for a REGULATION OF THE EUROPEAN PARLIAMENT ANDOF THECOUNCIL on horizontal cybersecurity requirements for products with digital elements and amending Regulation (EU) 2019/10 20. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri =celex%3A52022PC0454",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "European Union Agency for Cybersecurity. 2020. Guidel ines for Securing the Internet of Things-ENISA. http://archive.md/2023.04.18-071548/https://www.eni sa.europa.eu/publications/guidelines-for-securing-t he-internet-of-things/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Yong Fang, Yongcheng Liu, Cheng Huang, and Liang Liu. 20 20. FastEmbed: Predicting vulnerability exploitation pos sibility based on ensemble machinelearning algorithm. Plosone15, 2(2020), e0228439.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Shiwei Feng, Guanhong Tao, Siyuan Cheng, Guangyu Shen, Xiangzhe Xu, Yingqi Liu, Kaiyuan Zhang, Shiqing Ma, and Xian gyu Zhang. 2023. Detectingbackdoorsinpre-trainedencoders.In ProceedingsoftheIEEE/CVFConferenceonComputerVisionan dPatternRecognition .16352–16362. Manuscriptsubmitted to ACM 20 M.Böhme, E.Bodden,T.Bultan,C.Cadar,Y.Liu,and G.Scan niello",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "K. Fisler,S. Krishnamurthi,L. A. Meyerovich,and M. C. Tschantz. 2005. Verificationand Change-Impact Analysis of Access-Control Policies. In Proceedings ofthe 27thInternational Conference onSoftwar e Engineering (ICSE05) .196–205.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "AlexanderFrömmgen,JacobAustin,PeterChoy,NimeshG helani,LeraKharatyan,GabrielaSurita,ElenaKhrapko,Pa scalLamblin,Pierre-Antoine Manzagol, Marcus Revaj, Maxim Tabachnyk, Daniel Tarlow, Ke vin Villela, Dan Zheng, Satish Chandra, and Petros Maniatis. 2024. Resolving Code Review Comments with Machine Learning. In inPractice(ICSE-SEIP).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Michael Fu and Chakkrit Tantithamthavorn. 2022. Linev ul: A transformer-based line-level vulnerability predict ion. InProceedings of the 19th International Conference on MiningSoftware Repositories .608–620.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Dr Bhargav Gangadhara. 2023. Optimizing cloud-based m anufacturing: a study on service and development models. International Journal of Science and Research(IJSR) 12,6 (2023), 2487–2491.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "JunGao,PingfanKong,LiLi,TegawendéFBissyandé,and JacquesKlein.2019. Negativeresultsonminingcrypto-api usagerulesinandroidapps. In2019IEEE/ACM16th International Conference on MiningSoft ware Repositories(MSR). IEEE, 388–398.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "GitHub2024. GitHubwebsite. https://github.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "PatriceGodefroid, MichaelY Levin,and DavidMolnar.2 012. SAGE: whiteboxfuzzingforsecuritytesting. Commun.ACM 55,3 (2012), 40–44.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "Amid Golmohammadi, Man Zhang, and Andrea Arcuri. 2023. Testing restful apis: A survey. ACM Transactions on Software Engineering and Methodology 33,1 (2023), 1–41.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "YaroslavGolubev,ViktorPoletansky,NikitaPovarov, andTimofeyBryksin.2021.Multi-thresholdtoken-basedco declonedetection.In Proceedings ofthe 28thIEEEInternational Conferenceon Software Analy sis,Evolution and Reengineering (SANER’21).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "Qingyuan Gong, Jiayun Zhang, Yang Chen, Qi Li, Yu Xiao, X in Wang, and Pan Hui. 2019. Detecting malicious accounts in o nline developer communities using deep learning. In Proceedings of the 28th ACM International Conference on Info rmation and Knowledge Management. 1251– 1260.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "Danielle Gonzalez, Thomas Zimmermann, Patrice Godefr oid, and Max Schaefer. 2021. Anomalicious: Automated Detec tion of Anomalous and Potentially Malicious Commits on GitHub. In Practice(ICSE-SEIP) .258–267. https://doi.org/10.1109/ICSE-SEIP52600.2021.00035",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "Jesus M Gonzalez-Barahona. 2020. Characterizing outd ateness with technical lag: An exploratory study. In Proceedings of the IEEE/ACM 42nd International Conference on SoftwareEngineering Worksho ps.735–741.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "Yacong Gu, LingyunYing, YingyuanPu, Xiao Hu, HuajunCh ai,RuiminWang, Xing Gao, and Haixin Duan.2023. Investigat ing packagerelated securitythreatsinsoftwareregistries.In 2023IEEESymposium onSecurity and Privacy (SP). IEEE, 1578–1595.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "Shangwei Guo, Chunlong Xie, JiweiLi, LingjuanLyu,and TianweiZhang. 2022. Threatsto pre-trainedlanguage model s: Surveyand taxonomy. arXiv preprint arXiv:2202.06862 (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "WenboGuo,ZhengziXu,ChengweiLiu,ChengHuang,YongF ang,andYangLiu.2023. AnEmpiricalStudyofMaliciousCode InPyPIEcosystem. In202338thIEEE/ACMInternational Conference on Automated S oftwareEngineering (ASE) .IEEE, 166–177.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "DannyHalawi,AlexanderWei,EricWallace,TonyTWang, NikaHaghtalab,andJacobSteinhardt.2024. CovertMalicio usFinetuning:Challenges inSafeguardingLLMAdaptation. arXiv preprintarXiv:2406.20053 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "SubirHalder,Amrita Ghosal, and MauroConti. 2020. Sec ure over-the-airsoftware updates in connected vehicles:A survey.Computer Networks 178 (2020), 107343.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "Hao He, Yulin Xu, Yixiao Ma,Yifei Xu, GuangtaiLiang,an d MinghuiZhou. 2021. A multi-metricranking approachfor li brarymigrationrecommendations. In 2021IEEEInternational Conference onSoftware Analysis,E volution and Reengineering (SANER) .IEEE, 72–83.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 98,
+      "text": "Petr Hosek and Cristian Cadar. 2013. Safe Software Upda tes via Multi-version Execution. In International Conference on Software Engineering (ICSE2013) (SanFrancisco,CA,USA). 612–621.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 99,
+      "text": "JinchangHu,LyuyeZhang,ChengweiLiu,SenYang,SongH uang,andYangLiu.2023. EmpiricalAnalysisofVulnerabili tiesLifeCycleinGolang Ecosystem. arXiv preprint arXiv:2401.00515 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 100,
+      "text": "Kaifeng Huang, Bihuan Chen, Congying Xu, Ying Wang, Bo wen Shi, Xin Peng, Yijian Wu, and Yang Liu. 2022. Characteriz ingusages, updates and risksofthird-partylibrariesinJavaprojects. EmpiricalSoftware Engineering 27,4 (2022), 90.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 101,
+      "text": "GrahamHughes and TevfikBultan. 2008. Automated verifi cationof accesscontrol policies using aSAT solver. STTT10,6 (2008), 503–520.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 102,
+      "text": "Yu-LiangHung and Shingo Takada.2020. CPPCD:A Token- Based Approach to Detecting Potential Clones. In IEEE 14th International Workshop onSoftware Clones (IWSC’20).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 103,
+      "text": "iam[n.d.]. AWS IAMPolicy Language. http://docs.aws.amazon.com/IAM/latest/UserGuide/acc ess_policies.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 104,
+      "text": "NasifImtiaz,SeaverThorn,andLaurieWilliams.2021. Acomparativestudyofvulnerabilityreportingbysoftwar ecompositionanalysistools.In Proceedings of the 15th ACM / IEEEInternational Symposium on Empirical Software Engineering and Measurement (ESEM) (Bari, Italy) (ESEM ’21). AssociationforComputing Machinery,New York,NY, USA, Art icle 5,11 pages. https://doi.org/10.1145/3475716.3475769",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 105,
+      "text": "Ling Jiang, Hengchen Yuan, Qiyi Tang, Sen Nie, Shi Wu, a nd Yuqun Zhang. 2023. Third-Party Library Dependency for La rge-Scale SCA in the C/C++Ecosystem: How FarAre We?. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testingand Analysis (<confloc>,<city>Seattle</city>,<state>WA</state>,<countr y>USA</country>,</conf-loc>) (ISSTA2023) .AssociationforComputingMachinery,New",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 106,
+      "text": "Wenxin Jiang, NicholasSynovic, Rohan Sethi, AryanIn darapu, Matt Hyatt, Taylor R Schorlemmer,George K Thiruvat hukal,and JamesC Davis. 2022.Anempiricalstudyofartifactsandsecurityrisksint hepre-trainedmodelsupplychain.In Proceedingsofthe2022ACMWorkshoponSoftware Supply Chain Offensive Researchand EcosystemDefenses .105–114.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 107,
+      "text": "Jirayus Jiarpakdee, Chakkrit Kla Tantithamthavorn, and John Grundy. 2021. Practitioners’ perceptions of the go als and visual explanations of defect prediction models. In Proceedings of the 2021 IEEE/ACM 18th International Confere nce on Mining Software Repositories (MSR 2021). IEEE, Institute of Electrical and Electronics Engineers, 432–44 3.https://doi.org/10.1109/MSR52588.2021.00055",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 108,
+      "text": "JoeBiden.2021. ExecutiveOrderonImprovingtheNation’sCybersecurity .https://www.whitehouse.gov/briefing-room/presidentia l-actions/2021/05/12/executive-order-on-improving-t he-nations",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 109,
+      "text": "Timotej Kapus, Oren Ish-Shalom, Shachar Itzhaky, Noa m Rinetzky, and Cristian Cadar. 2019. Computing Summaries o f String Loops in C for Better Testing and Refactoring. In ACMSIGPLAN Conference on Programming Language Design and Imp lementation (PLDI 2019) (Phoenix, AZ, USA). 874–888.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 110,
+      "text": "Tushar Khinvasara, Stephanie Ness, and Nikolaos Tzen ios. 2023. Risk Management in Medical Device Industry. J. Eng. Res. Rep 25, 8 (2023), 130–140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 111,
+      "text": "JamesCKing. 1976. Symbolicexecutionand programtes ting.Commun.ACM 19, 7(1976), 385–394.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 112,
+      "text": "George Klees, Andrew Ruef, Benji Cooper, Shiyi Wei, an d Michael Hicks. 2018. Evaluating fuzz testing. In Proceedings of the 2018 ACMSIGSAC conferenceon computerand communicationssecurity .2123–2138.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 113,
+      "text": "GerwinKlein,KevinElphinstone,GernotHeiser,June Andronick,DavidCock,PhilipDerrin,DhammikaElkaduwe,K aiEngelhardt,RafalKolanski, Michael Norrish,et al. 2009. seL4: Formal verificationof an OS kernel. In Proceedings of the ACMSIGOPS 22nd symposium on Operating syst ems principles. 207–220.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 114,
+      "text": "Thijs Klooster, Fatih Turkmen, Gerben Broenink, Rube n Ten Hove, and Marcel Böhme. 2023. Continuous Fuzzing: A Stu dy of the Effectiveness and Scalability of Fuzzing in CI/CD Pipelines. In 2023 IEEE/ACM International Workshop on Search-Based and F uzz Testing (SBFT). 25–32. https://doi.org/10.1109/SBFT59156.2023.00015",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 115,
+      "text": "Padmanabhan Krishnan, Cristina Cifuentes, Li Li, Teg awendé F Bissyandé, and Jacques Klein. 2023. Why Is Static Ap plication Security Testing Hardto Learn? IEEESecurity & Privacy 21, 5(2023), 68–72.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 116,
+      "text": "TomaszKuchtaandBartoszZator.2022. AutoOff-Target :EnablingThoroughandScalableTestingforComplexSoftwa reSystems.In Proceedings ofthe 37thIEEE/ACMInternational Conferenceon Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 117,
+      "text": "VolodymyrKuznetzov,LászlóSzekeres,MathiasPayer ,GeorgeCandea,RSekar,andDawnSong.2018. Code-pointeri ntegrity. In TheContinuing ArmsRace:Code-Reuse Attacksand Defenses .81–116.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 118,
+      "text": "VuLe,MehrdadAfshari,and Zhendong Su. 2014. Compile rvalidationviaequivalence modulo inputs. ,216–226 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 119,
+      "text": "Seongmin Lee and Marcel Böhme. 2023. Statistical Reac hability Analysis. In Proceedings of the 31st ACM Joint European Software Engineer ing Conferenceand Symposiumon the Foundationsof Software Eng ineering (ESEC/FSE2023) .12.https://doi.org/10.1145/3611643.3616268",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 120,
+      "text": "XavierLeroy,SandrineBlazy,DanielKästner,Bernha rdSchommer,MarkusPister,andChristianFerdinand.2016. CompCert-aformallyverified optimizingcompiler.In ERTS 2016:Embedded Real TimeSoftwareand Systems,8th Euro pean Congress.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 121,
+      "text": "GuodongLi,IndradeepGhosh,andSreerangaPRajan.20 11. KLOVER:Asymbolicexecutionandautomatictestgenerat iontoolforC++programs. InComputerAided Verification:23rdInternational Conferenc e, CAV2011,Snowbird,UT,USA,July 14-20,2011.Proceedings 23.Springer,609–615.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 122,
+      "text": "Li Li, Tegawendé F Bissyandé, Mike Papadakis, Siegfri ed Rasthofer, Alexandre Bartel, DamienOcteau,Jacques Kle in, and Le Traon. 2017. Static analysisof android apps:A systematicliteraturereview. Information and Software Technology 88(2017), 67–95.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 123,
+      "text": "MenghaoLi,WeiWang,PeiWang,ShuaiWang,DinghaoWu, JianLiu,RuiXue,andWeiHuo.2017. Libd:Scalableandpreci sethird-partylibrary detection inandroidmarkets.In 2017IEEE/ACM39th International Conference on SoftwareEn gineering (ICSE) .IEEE, 335–346.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 124,
+      "text": "ZongjieLi,ChaozhengWang,ShuaiWang,andCuiyunGao .2023. Protectingintellectualpropertyoflargelanguage model-basedcodegeneration apisviawatermarks.In Proceedings of the 2023ACMSIGSAC Conference onComputer and CommunicationsSecurity. 2336–2350.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 125,
+      "text": "ZhenLi,DeqingZou,ShouhuaiXu,XinyuOu,HaiJin,Suj uanWang,ZhijunDeng,andYuyiZhong.2018. VulDeePecker:A DeepLearning-Based Systemfor VulnerabilityDetection. In NDSS. TheInternet Society.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 126,
+      "text": "LibFuzzer2022. LibFuzzerwebsite. http://llvm.org/docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 127,
+      "text": "ChengweiLiu,SenChen,LinglingFan,BihuanChen,Yan gLiu,andXinPeng.2022. Demystifyingthevulnerabilitypr opagationanditsevolution viadependency treesinthe npm ecosystem.In Proceedings of the 44thInternational Conferenceon Softwar e Engineering .672–684.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 128,
+      "text": "Antonio López Martínez, Manuel Gil Pérez, and Antonio Ruiz-Martínez. 2023. A comprehensive review of the state-o f-the-art on security and privacyissuesinhealthcare. Comput.Surveys 55, 12(2023), 1–38.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 129,
+      "text": "EthirajanManavalanandKandasamyJayakrishna.2019. AreviewofInternet ofThings(IoT)embedded sustainables upplychainforindustry4.0 requirements. Computers& industrial engineering 127(2019), 925–953.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 130,
+      "text": "PaulDanMarinescuandCristianCadar.2013. KATCH:Hi gh-CoverageTestingofSoftwarePatches.In EuropeanSoftwareEngineeringConference /ACMSIGSOFTSymposium on the Foundationsof Software Engin eering(ESEC/FSE2013) (Saint Petersburg,Russia).235–245.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 131,
+      "text": "Ankur Maurya and Divya Kumar. 2020. Reliability of saf ety-critical systems: A state-of-the-art review. Quality and Reliability Engineering International 36,7 (2020), 2547–2568.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 132,
+      "text": "RuijieMeng,MartinMirchev,MarcelBöhme, and Abhik R oychoudhury.2024. LargeLanguageModel guided Protocol Fu zzing.InProceedings of theNetworkand DistributedSystemSecurity Symposium (NDS S’24).15 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 133,
+      "text": "BertrandMeyer.1992. Applying ’designbycontract’. IEEEComputer 25, 10(1992), 40–51. Manuscriptsubmitted to ACM 22 M.Böhme, E.Bodden,T.Bultan,C.Cadar,Y.Liu,and G.Scan niello",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 134,
+      "text": "Microsoft.[n.d.]. Copilot. https://copilot.microsoft.com/. (Accessed on03/04/2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 135,
+      "text": "Imanol Mugarza, Jose Luis Flores, and Jose Luis Monter o. 2020. Security issues and software updates management in the industrial internet of things (iiot) era. Sensors20, 24(2020), 7160.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 136,
+      "text": "Phil Muncaster. 2021. Global Security Skills Shortag e Falls to 2.7 Million Workers-Infosecurity Magazine. https://www.infosecurity-magazine.com/news/global-s ecurity-skills-shortage/. Online; last accessed12October2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 137,
+      "text": "Benjamin Barslev Nielsen, Martin Toldam Torp, and And ers Møller. 2021. Modular Call Graph Construction for Secur ity Scanning of Node.js Applications. In Proc.30thInternational Symposiumon Software Testingand A nalysis (ISSTA).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 138,
+      "text": "Sabato Nocera, Simone Romano, Massimiliano Di Penta, Rita Francese, and Giuseppe Scanniello. 2023. Software Bil l of Materials Adoption: A Mining Study from GitHub. In Procedings of International Conference on Software Mainten ance and Evolution. IEEE, 39–49. https://doi.org/10.1109/ICSME58846.2023.00016",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 139,
+      "text": "Sabato Nocera, Simone Romano, Rita Francese, and Gius eppe Scanniello. 2023. Training for Security: Planning the Use of a SAT in the Development Pipeline of Web Apps. In 45th IEEE/ACMInternational Conference on Software Engine ering: Software Engineering Education and Training, SEET@ICSE2023,Melbourne, Australia,May 14-20,2023 .40–45.https://doi.org/10.1109/ICSE-SEET58685.2023.00010",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 140,
+      "text": "Sabato Nocera, Simone Romano, Rita Francese, and Gius eppe Scanniello. 2024. Training for Security: Results from Using a SAT in the Development Pipeline of Web Apps. In 46th IEEE/ACMInternational Conference on Software Engine ering: Software Engineering Education and Training, SEET@ICSE2024,Lisbon,Portugal. 247–257. https://doi.org/doi.org/10.1145/3639474.3640073",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 141,
+      "text": "Philippe Ombredanne. 2020. Free and Open Source Softw are License Compliance: Tools for Software Composition Ana lysis.Computer 53, 10 (2020), 105–109. https://doi.org/10.1109/MC.2020.3011082",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 142,
+      "text": "OSS-Fuzz2024. https://github.com/google/oss-fuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 143,
+      "text": "Rohan Padhye, Caroline Lemieux, and Koushik Sen. 2019. JQF: coverage-guided property-based testing in Java. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA) (Beijing, China) (ISSTA 2019). 398–401. https://doi.org/10.1145/3293882.3339002",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 144,
+      "text": "Hristina Palikareva, Tomasz Kuchta, and Cristian Cad ar. 2016. Shadow of a Doubt: Testing for Divergences Between Software Versions. In International Conference on SoftwareEngineering (ICSE20 16)(Austin,TX, USA). 1181–1192.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 145,
+      "text": "FatemehKhoda Parast,ChandniSindhav,SeemaNikam,H adisehIzadiYekta,Kenneth BKent,andSaqibHakak.2022. Cl oudcomputingsecurity: A surveyof service-basedmodels. Computers&Security 114 (2022), 102580.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 146,
+      "text": "Corina S Păsăreanu and Neha Rungta. 2010. Symbolic Pat hFinder: symbolic execution of Java bytecode. In Proceedings of the 25th IEEE/ACM International Conference on Automated Software Engineeri ng.179–180.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 147,
+      "text": "IvanPashchenko,Duc-LyVu,andFabioMassacci.2020. Aqualitativestudyofdependency managementanditssecuri tyimplications.In Proceedingsofthe 2020ACMSIGSACconference oncomputer and commun icationssecurity .1513–1531.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 148,
+      "text": "Pankayaraj Pathmanathan, Souradip Chakraborty, Xia ngyu Liu, Yongyuan Liang, and Furong Huang. 2024. Is poisoni ng a real threat to LLM alignment? Maybemoreso thanyouthink. arXiv preprint arXiv:2406.12091 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 149,
+      "text": "Felix Pauck, Eric Bodden, and Heike Wehrheim. 2018. Do android taint analysistools keep their promises?. In Proceedings of the 2018 26th ACM JointMeeting on European Software EngineeringConference and Symposium on theFoundations of Software Engineering .331–341.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 150,
+      "text": "HammondPearce,BaleeghAhmad,BenjaminTan,Brendan Dolan-Gavitt,andRameshKarri.2022.AsleepattheKeyboar d?AssessingtheSecurity of GitHubCopilot’sCode Contributions.In SP. IEEE, 754–768.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 151,
+      "text": "HammondPearce,BaleeghAhmad,BenjaminTan,Brendan Dolan-Gavitt,andRameshKarri.2022. Asleepatthekeyboar d?assessingthesecurity of githubcopilot’s code contributions. In 2022IEEESymposiumon Security and Privacy (SP) .IEEE, 754–768.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 152,
+      "text": "NeilPerry,MeghaSrivastava,DeepakKumar,andDanBo neh.2023. DoUsersWriteMoreInsecureCodewithAIAssistan ts?.InProceedingsofthe 2023 ACMSIGSAC Conference on Computer and CommunicationsS ecurity (CCS ’23) .Association for Computing Machinery,New York, NY, USA, 2785–2799. https://doi.org/10.1145/3576915.3623157",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 153,
+      "text": "JonasPeters,DominikJanzing,andBernhardSchlkopf .2017.ElementsofCausalInference:FoundationsandLearningAlg orithms. TheMITPress.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 154,
+      "text": "Van-ThuanPham, Marcel Böhme, Andrew E. Santosa, Alex andru R. Căciulescu,and Abhik Roychoudhury. 2021. Smart Gr eyboxFuzzing. IEEE TransactionsonSoftware Engineering 47,9 (2021), 1980–1997. https://doi.org/10.1109/TSE.2019.2941681",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 155,
+      "text": "Quoc-SangPhan,LucasBang,CorinaS.Pasareanu,Pasq ualeMalacaria,andTevfikBultan. 2017. Synthesis ofAdapti veSide-Channel Attacks.In 30thIEEEComputer Security Foundations Symposium,CSF201 7,Santa Barbara, CA,USA,August21-25,2017 .IEEE ComputerSociety, 328–342.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 156,
+      "text": "LuísPina, Anastasios Andronidis, Michael Hicks, and CristianCadar.2019. Mvedsua: Higher Availability Dynami cSoftware Updates via MultiVersionExecution. In International Conference on Architectural Support for Prog rammingLanguages and Operating Systems(ASPLOS 2019) (Providence, RI, USA). 573–585.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 157,
+      "text": "Goran Piskachev, Lisa Nguyen Quang Do, and Eric Bodden. 2019. Codebase-adaptive detection of security-relevant methods. In Proceedings of the 28thACMSIGSOFT International Symposium on Software Te stingand Analysis .181–191.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 158,
+      "text": "GoranPiskachev,Ranjith Krishnamurthy,and EricBod den. 2021. Secucheck:Engineering configurabletaint analy sisforsoftwaredevelopers.In 2021IEEE21stInternational WorkingConference on SourceC ode Analysis and Manipulation (SCAM) .IEEE, 24–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 159,
+      "text": "SebastianPoeplauandAurélienFrancillon.2020. Sym bolicexecutionwithSymCC:Don’tinterpret,compile!.In 29thUSENIXSecuritySymposium (USENIXSecurity 20) .181–198.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 160,
+      "text": "ChromiumProject. 2021. Memorysafety. https://www.chromium.org/Home/chromium-security/mem ory-safety/ Manuscriptsubmitted to ACM SoftwareSecurityAnalysis in2030and Beyond: AResearch Ro admap 23",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 161,
+      "text": "pundit 2016. GitHubelabs/pundit:Minimalauthoriz ationthrought OO designand pureRubyclasses. https://github.com/elabs/pundit.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 162,
+      "text": "ManuelRiggerandZhendong Su.2020. TestingDatabase EnginesviaPivoted QuerySynthesis.In 14thUSENIXSymposiumonOperating Systems Designand Implementation (OSDI 20) .USENIX Association, 667–682. https://www.usenix.org/conference/osdi20/presentati on/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 163,
+      "text": "NiklasRisseand MarcelBöhme. 2023. LimitsofMachine LearningforAutomatic VulnerabilityDetection. arXiv: 2306.17193 [cs.CR]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 164,
+      "text": "YamanRoumani.2021. Patching zero-dayvulnerabilit ies:anempiricalanalysis. Journal ofCybersecurity 7,1 (2021), tyab023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 165,
+      "text": "Gustavo Sandoval, Hammond Pearce, Teo Nys, Ramesh Kar ri, Siddharth Garg, and Brendan Dolan-Gavitt. 2023. Lost at c: A user study on the securityimplicationsof largelanguagemodel code assista nts.In32nd USENIXSecurity Symposium (USENIXSecurity 23). 2205–2222.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 166,
+      "text": "Koushik Sen, Darko Marinov,and Gul Agha. 2005. CUTE: A concolic unit testing engine for C. ACMSIGSOFT Software Engineering Notes 30, 5 (2005), 263–272.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 167,
+      "text": "KostyaSerebryany.2017. OSS-Fuzz-Google’s continu ous fuzzingservicefor open sourcesoftware.USENIX Associ ation, Vancouver,BC.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 168,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Po tapenko, and Dmitriy Vyukov. 2012. AddressSanitizer: A fas t address sanity checker. In 2012USENIXannual technical conference (USENIXATC12) .309–318.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 169,
+      "text": "GeoffreySmith.2009. OntheFoundationsofQuantitati veInformationFlow.In FoundationsofSoftwareScienceandComputationalStructu res,12th International Conference, FOSSACS 2009,York,UK, March22 -29,2009.Proceedings. 288–302.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 170,
+      "text": "EliezioSoares,GustavoSizilio,JadsonSantos,Dani elAlencardaCosta,andUiráKulesza.2022. Theeffectsofcon tinuous integrationonsoftware development: asystematicliteraturereview. EmpiricalSoftware Engineering 27,3(2022), 78.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 171,
+      "text": "X. Song, Y. Wang, X. Cheng, G. Liang, W. Qianxiang, and Z. Zhu. 2024. Efficiently Trimmingthe Fat: Streamlining Softw are Dependencies with JavaReflectionandDependencyAnalysis.In 2024IEEE/ACM46thInternationalConferenceonSoftwareEn gineering(ICSE) .IEEEComputerSociety, LosAlamitos,CA, USA, 887–887. https://doi.ieeecomputersociety.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 172,
+      "text": "JohannesSpäth, KarimAli,andEricBodden. 2019. Cont ext-, flow-,andfield-sensitivedata-flowanalysisusingsyn chronized pushdownsystems. Proceedings ofthe ACMon ProgrammingLanguages 3, POPL (2019), 1–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 173,
+      "text": "Evgeniy Stepanov and Konstantin Serebryany. 2015. Me morySanitizer: fast detector of uninitialized memory use i n C++. In 2015 IEEE/ACM International Symposium on Code Generation and Optimizati on (CGO) .IEEE, 46–55.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 174,
+      "text": "JeffreyVanderStoep.2022.MemorySafeLanguagesinAn droid13. https://security.googleblog.com/2022/12/memory-safe -languages-in-android-13.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 175,
+      "text": "Jacob Stringer, Amjed Tahir, Kelly Blincoe, and Jens D ietrich. 2020. Technical lag of dependencies in major packa ge managers. In 2020 27th Asia-PacificSoftwareEngineering Conference (APSEC) .IEEE, 228–237.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 176,
+      "text": "KairanSun,ZhengziXu,ChengweiLiu,KaixuanLi,andY angLiu.2023.DemystifyingtheCompositionandCodeReusei nSoliditySmartContracts. InProceedingsofthe31stACMJointEuropeanSoftwareEngineer ingConferenceandSymposiumontheFoundationsofSoftware Engineering .796–807.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 177,
+      "text": "ZhensuSun, Xiaoning Du,Fu Song, MingzeNi,andLiLi.2 022. Coprotector:Protect open-sourcecode againstunauth orized trainingusagewith datapoisoning. In Proceedings of the ACMWebConference2022 .652–660.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 178,
+      "text": "Laszlo Szekeres, Mathias Payer, Tao Wei, and Dawn Song. 2013. Sok: Eternal war in memory. In. IEEE, 48–62.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 179,
+      "text": "Matthew Taylor, Ruturaj Vaidya, Drew Davidson, Loren zo De Carli, and Vaibhav Rastogi. 2020. Defending against pa ckage typosquatting. In Proceedings ofthe 14thInternational Conference Networkan d SystemSecurity (NSS 2020) .Springer,112–131.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 180,
+      "text": "KenThompson. 1984. Reflections on trustingtrust. Commun.ACM 27, 8(aug1984), 761–763. https://doi.org/10.1145/358198.358210",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 181,
+      "text": "Omer Tripp, Marco Pistoia, Stephen J Fink, Manu Sridha ran, and Omri Weisman. 2009. TAJ: effective taint analysis of web applications. ACM Sigplan Notices 44, 6(2009), 87–97.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 182,
+      "text": "Daniele Ucci, Leonardo Aniello, and Roberto Baldoni. 2019. Surveyof machine learning techniques for malwareana lysis.Computers & Security 81(2019), 123–147.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 183,
+      "text": "Celina G. Val, Michael A. Enescu, Sam Bayless, William Aiello, and Alan J. Hu. 2016. Precisely Measuring Quantitat ive Information Flow: 10K Linesof Code and Beyond.In 2016IEEEEuropean Symposium on Security and Privacy (EuroS& P).31–46.https://doi.org/10.1109/EuroSP.2016.15",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 184,
+      "text": "verizonleak [n.d.]. 14 MILLION Verizon subscribers’ details leak from crappily configured AWS S3 data store. https://www.theregister.co.uk/2017/07/12/14m_verizo n_customers_details_out/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 185,
+      "text": "GarimaVermaand Sandhya Adhikari.2020. Cloud comput ing securityissues:astakeholder’sperspective. SNComputer Science 1,6 (2020), 329.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 186,
+      "text": "VSCode 2024. VSCode IDE. https://code.visualstudio.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 187,
+      "text": "Duc-LyVu,IvanPashchenko,FabioMassacci,HenrikPl ate,andAntoninoSabetta.2020.Typosquattingandcombosq uattingattacksonthepython ecosystem.In 2020ieee european symposiumon securityand privacy worksh ops(euros&pw) .IEEE, 509–514.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 188,
+      "text": "Ying Wang, Peng Sun, Lin Pei, Yue Yu, Chang Xu, Shing-Ch i Cheung, Hai Yu, and Zhiliang Zhu. 2023. Plumber: Boosting t he propagation of vulnerabilityfixesinthe npm ecosystem. IEEETransactionsonSoftware Engineering (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 189,
+      "text": "Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosm a, Ed H. Chi, Quoc Le, and Denny Zhou. 2022. Chain of Thought Pr ompting Elicits Reasoning inLargeLanguageModels. CoRRabs/2201.11903(2022). arXiv: 2201.11903 https://arxiv.org/abs/2201.11903",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 190,
+      "text": "JonathanWoodruff,RobertNMWatson,DavidChisnall,S imon WMoore,JonathanAnderson,BrooksDavis,BenLaurie,P eter GNeumann,Robert Norton,and MichaelRoe. 2014. TheCHERI capabilitymodel: R evisiting RISCinanageof risk. ACMSIGARCHComputerArchitectureNews 42,3 (2014), 457–468.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 191,
+      "text": "Fangzhou Wu, Ning Zhang, Somesh Jha, Patrick McDaniel, and Chaowei Xiao. 2024. A new era in llm security: Exploring security concerns in real-worldllm-based systems. arXiv preprint arXiv:2402.18649 (2024). Manuscriptsubmitted to ACM 24 M.Böhme, E.Bodden,T.Bultan,C.Cadar,Y.Liu,and G.Scan niello",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 192,
+      "text": "Haoze Wu, Alex Ozdemir,Aleksandar Zeljić, Kyle Julia n, Ahmed Irfan, DivyaGopinath, Sadjad Fouladi, Guy Katz, Co rina Pasareanu,and Clark Barrett. 2020. Parallelization Techniques for Verifying N eural Networks. In 2020 Formal Methods in Computer Aided Design (FMCAD). 128–137. https://doi.org/10.34727/2020/isbn.978-3-85448-042- 6_20",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 193,
+      "text": "YuemingWu,ChengweiLiu,andYangLiu.2023. TheSoftw areGenomeProject:VenturetotheGenomicPathwaysofOpenS ourceSoftwareand ItsApplications. arXiv preprint arXiv:2311.09881 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 194,
+      "text": "YulunWu,ZeliangYu,MingWen,QiangLi,DeqingZou,an dHaiJin.2023. Understanding thethreatsofupstreamvulne rabilitiestodownstream projectsinthe mavenecosystem.In 2023IEEE/ACM45thInternational Conferenceon Software En gineering(ICSE). IEEE, 1046–1058.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 195,
+      "text": "XACML 2003. eXtensible Access Control Markup Languag e (XACML) Version 1.0. OASIS Standard. http://www.oasis-open.org/committees/tc_home.php?wg _abbrev=xacmlhttp://www.oasis-open.org/committees/t c_home.php?wg_abbrev=xacml.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 196,
+      "text": "GuoweiYang,CorinaS.Păsăreanu,andSarfrazKhurshi d.2012. Memoizedsymbolicexecution.In Proceedingsofthe2012InternationalSymposium on Software Testing and Analysis (Minneapolis, MN, USA) (ISSTA 2012). Association for Computing Machinery, New York, NY, USA, 14 4–154. https://doi.org/10.1145/2338965.2336771",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 197,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 201 1. Finding and understanding bugs in C compilers. In Proceedings of the 32nd ACM SIGPLANconferenceon Programminglanguage design and implem entation. 283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 198,
+      "text": "YifanYao, Jinhao Duan,Kaidi Xu,Yuanfang Cai,Zhibo S un, and Yue Zhang. 2024. A surveyonlarge language model (llm ) securityand privacy: Thegood, the bad,and the ugly. High-Confidence Computing (2024), 100211.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 199,
+      "text": "Jerin Yasmin, Yuan Tian, and Jinqiu Yang. 2020. A first l ook at the deprecation of RESTful APIs: An empirical study. I n2020 IEEE International Conferenceon Software Maintenance and Evolution (ICSME) .IEEE, 151–161.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 200,
+      "text": "ZhiyuanYu,YuhaoWu,NingZhang,ChenguangWang,Yevg eniyVorobeychik,andChaoweiXiao.2023. CODEIPPROMPT:in tellectualproperty infringement assessmentofcode languagemodels. In International Conferenceon Machine Learning .PMLR, 40373–40389.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 201,
+      "text": "NusratZahan,ThomasZimmermann,PatriceGodefroid, BrendanMurphy,ChandraMaddila,andLaurieWilliams.2022. Whatareweaklinksin the npm supply chain?. In Proceedings of the44th International Conference on Softwar eEngineering: Software Engineeringin Practice .331–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 202,
+      "text": "MichalZalewski.[n.d.]. Technical “whitepaper”for afl-fuzz.http://lcamtuf.coredump.cx/afl/technical_details.txt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 203,
+      "text": "XianZhan,LinglingFan,SenChen,FengWu,TianmingLi u,XiapuLuo,andYangLiu.2021. Atvhunter: Reliableversio ndetection ofthird-party librariesfor vulnerability identification in android appl ications. In. IEEE, 1695–1707.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 204,
+      "text": "LyuyeZhang, Chengwei Liu,Sen Chen, Zhengzi Xu, Lingl ing Fan, Lida Zhao, YiranZhang, and Yang Liu.2023. Mitigati ng persistence of opensourcevulnerabilitiesinmavenecosystem.In 202338thIEEE/ACMInternational Conferenceon Automated S oftwareEngineering (ASE) .IEEE, 191– 203.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 205,
+      "text": "Lyuye Zhang, Chengwei Liu, Zhengzi Xu, Sen Chen, Lingl ing Fan, Bihuan Chen, and Yang Liu. 2023. Has My Release Disob eyedSemantic Versioning? Static Detection Based on Semantic Differencing. I nProceedings of the 37th IEEE/ACM International Conference o n Automated Software Engineering (<conf-loc>, <city>Rochester</city>,<state>MI</state >,<country>USA</country>,</conf-loc>) (ASE’22).Associationfo",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 206,
+      "text": "Lyuye Zhang, Chengwei Liu, Zhengzi Xu, Sen Chen, Lingl ing Fan, Lida Zhao, Jiahui Wu, and Yang Liu. 2023. Compatible remediation on vulnerabilities from third-party libraries for java projects. In. IEEE, 2540–2552.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 207,
+      "text": "Yanan Zhang, Yuqiao Ning, Chao Ma, Longhai Yu, and Zhen Guo. 2023. Empirical Study for Open Source Libraries in Auto motive Software Systems. IEEEAccess (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 208,
+      "text": "Yuntong Zhang,HaifengRuan,ZhiyuFan,andAbhikRoyc houdhury.2024. AutoCodeRover:Autonomous ProgramImprov ement.arXiv preprint arXiv:2404.05427 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 209,
+      "text": "ZhengyanZhang,GuangxuanXiao,Yongwei Li,TianLv,F anchaoQi,ZhiyuanLiu,YashengWang,XinJiang,andMaosong Sun.2023. Redalarm forpre-trainedmodels: Universalvulnerabilityto neuron -level backdoorattacks. MachineIntelligence Research 20, 2(2023), 180–193.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 210,
+      "text": "JianZhao,ShenaoWang,YanjieZhao,XinyiHou,Kailon gWang,PeimingGao,YuanchaoZhang,ChenWei,andHaoyuWang .2024. ModelsAre Codes:TowardsMeasuringMaliciousCode Poisoning Attacks onPre-trained Model Hubs. arXiv preprint arXiv:2409.09368 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 211,
+      "text": "YaqinZhou,ShangqingLiu,JingkaiSiow,XiaoningDu, andYangLiu.2019. Devign:Effectivevulnerabilityidentifi cationbylearningcomprehensiveprogramsemanticsviagraphneuralnetworks. Advances in neural informationprocessingsystems 32(2019).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 212,
+      "text": "MarkusZimmermann,Cristian-AlexandruStaicu,CamT enny,andMichaelPradel.2019. Smallworldwith highrisks: A studyof securitythreats inthe npm ecosystem.In 28thUSENIXSecurity Symposium(USENIXSecurity 19) .995–1010. Manuscriptsubmitted to ACM",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 162,
+      "text": "ManuelRiggerandZhendong Su.2020. TestingDatabase EnginesviaPivoted QuerySynthesis.In 14thUSENIXSymposiumonOperating Systems Designand Implementation (OSDI 20) .USENIX Association, 667–682. https://www.usenix.org/conference/osdi20/presentati on/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 118,162,197]",
+      "technique": "pqs",
+      "sentence": "The technique has been appliedwithalotofsuccessinseveralareas,forinstancef oroptimizingcompilersanddatabasemanagement systems (DBMS), where it has found hundreds of critical bugs in matur e compiler and DBMS implementations [ 118,162,197].",
+      "context_before": "orego aspecification,yet here one specifiesvulnerability types,notprogram functionality. Fuzzing [34], at a high level, involves testing software with randomly g enerated inputs. There are three main forms offuzzing:black-box,grey-box,andwhite-boxfuzzing,wi ththelatteralsoknownasdynamicsymbolicexecutionand discussed separatelybelow.Black-boxfuzzingdoes notuse any knowledgeof theimplementation, norany execution feedbackotherthanwhatcanbeobservedbyrunningthesoftw areinablack-boxmanner.Atthesametime,itrequires domain knowledge to be effective, such as grammar for the inpu ts accepted by the program.",
+      "context_after": "Greybox fuzzing is the most widely used form of fuzzing for ge neral software systems, which is guided by the code coverage achieved by the generated inputs. It is typically c ombined with mutation-based input generation. Greybox fuzzingwaspioneeredby AFL[ 202]andnowtherearedozensofgreyboxfuzzersavailableforma nydifferentlanguages Manuscriptsubmitted to ACM SoftwareSecurityAnalysis in2030and Beyond: AResearch Ro admap 3 and domains. OSS-Fuzz [ 142] is a service provided by Google which runs several state-of -the-art greybox fuzzers on open-sourcesoftware.To date,ithas foundover 10Kvulnera bilit",
+      "section": "2.1 So/f_tware Analysis Techniques",
+      "page": 2,
+      "char_offset": 7092,
+      "cited_reference": {
+        "number": 162,
+        "text": "ManuelRiggerandZhendong Su.2020. TestingDatabase EnginesviaPivoted QuerySynthesis.In 14thUSENIXSymposiumonOperating Systems Designand Implementation (OSDI 20) .USENIX Association, 667–682. https://www.usenix.org/conference/osdi20/presentati on/rigger",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 162]",
+      "technique": "pqs",
+      "sentence": "Forinstance,differentialtestinghasbeenextremelysuccessfulinthecontextof compilertesting[ 197],whilevarioustypesofmetamorphic transformations have beeneffective fordatabasemanagemen t systems [ 162].",
+      "context_before": "and C++,mostattentionhasbeengiven tomemorysafetyerrors[ 178].Ontheonehand,techniquestargetingsucherrors havebeendesigned andadoptedinpractice,suchascompiler sanitizers [ 168,173],whichcanfindbugssuchasbuffer overflows and use-after-free errors. On the other hand, tech niques that can in principle find a broader class of bugs, suchasstaticanalysisandsymbolicexecution,havebeenpr imarilyappliedtofindmemorysafetyerrorsandinjection vulnerabilities, bothduetotheir importanceand their eas ily availableoracles[ 42,49,55,166]. Togobeyondmemorysafety,researchershaveexploiteddiffe rentialandmetamorphicoracles.",
+      "context_after": "Certaintypesoffunctionalbugsmayescapedifferentialand metamorphictesting.Specificationshavebeenforlong proposedas a waytowritecorrectsoftware, butadoptionhas beenlimited,particularlyinmainstream languages. Manuscriptsubmitted to ACM 4 M.Böhme, E.Bodden,T.Bultan,C.Cadar,Y.Liu,and G.Scann iello 3 CHALLENGES ANDOPPORTUNITIES Aswelookforward,wepredictasubstantialshiftinthe keychallenges forsecurityanalysis.Notonlywilloursoftware systems grow larger, moreinterconnected, moredistribute d,and evolve faster. We also expectto seea greater dependence on the software supply chain, a deeper integration wit",
+      "section": "2.2 Detected Errors",
+      "page": 3,
+      "char_offset": 11336,
+      "cited_reference": {
+        "number": 162,
+        "text": "ManuelRiggerandZhendong Su.2020. TestingDatabase EnginesviaPivoted QuerySynthesis.In 14thUSENIXSymposiumonOperating Systems Designand Implementation (OSDI 20) .USENIX Association, 667–682. https://www.usenix.org/conference/osdi20/presentati on/rigger",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "A roadmap article on software security analysis for the coming decade. It asks what vulnerabilities future systems will have and how to detect them, how to find bugs once the shallow ones are gone, and how to protect systems when not all flaws can be found -- in a setting where code is increasingly co-written by generative AI and dependencies span whole supply chains.",
+    "narrative": "Two citations, both to a paper by a SQLancer author rather than to SQLancer itself, placing DBMSs among the areas where the technique has been applied successfully and noting metamorphic transformations among the approaches discussed.",
+    "roles": {
+      "M1": "background",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3720504.json b/_data/papers/paper_doi_10_1145_3720504.json
new file mode 100644
index 0000000..d687b49
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3720504.json
@@ -0,0 +1,587 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:03Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3720504",
+    "title": "Checking Observational Correctness of Database Systems",
+    "authors": [
+      "Lauren Pick",
+      "Amanda Xu",
+      "Ankush Desai",
+      "S. Seshia",
+      "Aws Albarghouthi"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3720504",
+    "arxiv_id": null,
+    "s2_paper_id": "8e351e69fa979c4de3f4534211b50b4052a44218",
+    "url": "https://doi.org/10.1145/3720504",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3720504",
+      "retrieved_at": "2026-09-09T01:30:03Z",
+      "chars": 108916,
+      "content_sha256": "sha256:1de6578302f65a54c891e9317a8c4f6e8d051fbea77fb3c895df944d50060858"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 28,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2162
+      },
+      {
+        "number": "2",
+        "title": "Motivating Example",
+        "start": 10192
+      },
+      {
+        "number": "2.1",
+        "title": "Verifying Observational Correctness",
+        "start": 14892
+      },
+      {
+        "number": "3",
+        "title": "Observational Correctness",
+        "start": 23529
+      },
+      {
+        "number": "3.1",
+        "title": "States, Transactions, and Execution Traces",
+        "start": 23620
+      },
+      {
+        "number": "3.2",
+        "title": "Version-Aware TDL Semantics",
+        "start": 27151
+      },
+      {
+        "number": "3.3",
+        "title": "Semantic Correctness of Traces",
+        "start": 31661
+      },
+      {
+        "number": "3.4",
+        "title": "Isolation Correctness of Traces",
+        "start": 33122
+      },
+      {
+        "number": "3.5",
+        "title": "Observational Correctness",
+        "start": 34086
+      },
+      {
+        "number": "4.1",
+        "title": "Checking Approach",
+        "start": 37117
+      },
+      {
+        "number": "5.1",
+        "title": "BMC-Based Encoding",
+        "start": 39483
+      },
+      {
+        "number": "5.2",
+        "title": "Improved Encoding",
+        "start": 44046
+      },
+      {
+        "number": "5.3",
+        "title": "Partially-Symbolic State Representation",
+        "start": 45493
+      },
+      {
+        "number": "5.4",
+        "title": "Semantic Constraint Generation",
+        "start": 48110
+      },
+      {
+        "number": "6.1",
+        "title": "Instrumented VA-TDL Semantics",
+        "start": 56759
+      },
+      {
+        "number": "6.2",
+        "title": "Isolation Constraints on Execution Traces",
+        "start": 58577
+      },
+      {
+        "number": "6.3",
+        "title": "Symbolic Encoding of Isolation Constraints",
+        "start": 61099
+      },
+      {
+        "number": "6.4",
+        "title": "Isolation-Level-Aware Reads",
+        "start": 61994
+      },
+      {
+        "number": "7.1",
+        "title": "Heuristics",
+        "start": 65197
+      },
+      {
+        "number": "8.1",
+        "title": "RQ1: Impact of the Improved Encoding",
+        "start": 70930
+      },
+      {
+        "number": "8.2",
+        "title": "RQ2: Classes of Bugs Considered",
+        "start": 72836
+      },
+      {
+        "number": "8.3",
+        "title": "RQ3: Scaling Across Workloads",
+        "start": 78758
+      },
+      {
+        "number": "8.4",
+        "title": "RQ4: Impact of Heuristics",
+        "start": 85850
+      },
+      {
+        "number": "8.5",
+        "title": "Industrial Case Study",
+        "start": 87853
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2011. Big data in real time at Twitter. https://www.infoq.com/presentations/Big-Data-in-Real-Time-at-Twitter.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2017. RUBiS. https://projects.ow2.org/view/rubis/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2022. SQLsmith: a random sql query generator. https://github.com/anse1/sqlsmith. Accessed: 2023-11-15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2024. Jepsen. https://jepsen.io/. Accessed: 2024-04-05.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Atul Adya. 1999. Weak consistency: a generalized theory and optimistic implementations for distributed transactions. (1999).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Peter Alvaro and Kyle Kingsbury. 2020. Elle: Inferring Isolation Anomalies from Experimental Observations. Proc. VLDB Endow. 14, 3 (2020), 268–280. https://doi.org/10.5555/3430915.3442427",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Ranadeep Biswas and Constantin Enea. 2019. On the complexity of checking transactional consistency. Proc. ACM Program. Lang. 3, OOPSLA (2019), 165:1–165:28. https://doi.org/10.1145/3360591",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Ranadeep Biswas, Diptanshu Kakwani, Jyothi Vedurada, Constantin Enea, and Akash Lal. 2021. MonkeyDB: effectively testing correctness under weak isolation levels. Proc. ACM Program. Lang. 5, OOPSLA (2021), 1–27. https://doi.org/10. 1145/3485546",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Ahmed Bouajjani, Constantin Enea, Rachid Guerraoui, and Jad Hamza. 2017. On verifying causal consistency. In Proceedings of the 44th ACMSIGPLAN Symposium on Principles of Programming Languages, POPL 2017, Paris, France, January 18-20, 2017, Giuseppe Castagna and Andrew D. Gordon (Eds.). ACM, 626–638. https://doi.org/10.1145/3009837. 3009888",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Shumo Chu, Konstantin Weitz, Alvin Cheung, and Dan Suciu. 2017. HoTTSQL: proving query rewrites with univalent SQL semantics. In Proceedings of the 38th ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2017, Barcelona, Spain, June 18-23, 2017, Albert Cohen and Martin T. Vechev (Eds.). ACM, 510–524.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Edmund M. Clarke, Armin Biere, Richard Raimi, and Yunshan Zhu. 2001. Bounded Model Checking Using Satisfiability Solving. Formal Methods Syst. Des. 19, 1 (2001), 7–34. https://doi.org/10.1023/A:1011276507260",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Natacha Crooks, Youer Pu, Lorenzo Alvisi, and Allen Clement. 2017. Seeing is Believing: A Client-Centric Specification of Database Isolation. In Proceedings of the ACM Symposium on Principles of Distributed Computing, PODC 2017, Washington, DC, USA, July 25-27, 2017, Elad Michael Schiller and Alexander A. Schwarzmann (Eds.). ACM, 73–82. https://doi.org/10.1145/3087801.3087802",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Leonardo De Moura and Nikolaj Bjørner. 2008. Z3: An Efficient SMT Solver. In Proceedings of the Theory and Practice of Software, 14th International Conference on Tools and Algorithms for the Construction and Analysis of Systems (Budapest, Hungary) (TACAS’08/ETAPS’08). Springer-Verlag, Berlin, Heidelberg, 337–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Haoran Ding, Zhaoguo Wang, Yicun Yang, Dexin Zhang, Zhenglin Xu, Haibo Chen, Ruzica Piskac, and Jinyang Li. 2023. Proving Query Equivalence Using Linear Integer Arithmetic.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. In 37th IEEE/ACM International Conference on Automated Software Engineering, ASE 2022, Rochester, MI, USA, October 10-14, 2022. ACM, 49:1–49:12. https://doi.org/10.1145/3551349.3560431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Google Cloud. 2024. BigQuery: Multi-statement transactions. https://cloud.google.com/bigquery/docs/transactions. Accessed: 2024-04-02.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Jim Gray and Andreas Reuter. 1993. Transaction Processing: Concepts and Techniques. Morgan Kaufmann.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Thomas Haas, Roland Meyer, and Hernán Ponce de León. 2022. CAAT: consistency as a theory. Proc. ACM Program. Lang. 6, OOPSLA2 (2022), 114–144. https://doi.org/10.1145/3563292",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Fei He, Zhihang Sun, and Hongyu Fan. 2021. Satisfiability modulo ordering consistency theory for multi-threaded program verification. In PLDI ’21: 42nd ACMSIGPLAN International Conference on Programming Language Design and Implementation, Virtual Event, Canada, June 20-25, 2021, Stephen N. Freund and Eran Yahav (Eds.). ACM, 1264–1279. https://doi.org/10.1145/3453483.3454108",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Yang He, Pinhan Zhao, Xinyu Wang, and Yuepeng Wang. 2024. VeriEQL: Bounded Equivalence Verification for Complex SQL Queries with Integrity Constraints. Proc. ACM Program. Lang. 8, OOPSLA1 (2024), 1071–1099. https: Proc. ACM Program. Lang., Vol. 9, No. OOPSLA1, Article 139. Publication date: April 2025. Checking Observational Correctness of Database Systems 139:27 //doi.org/10.1145/3649849",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Kaile Huang, Si Liu, Zhenge Chen, Hengfeng Wei, David A. Basin, Haixiang Li, and Anqun Pan. 2023. Efficient Black-box Checking of Snapshot Isolation in Databases. Proc. VLDB Endow. 16, 6 (2023), 1264–1276. https://doi.org/10. 14778/3583140.3583145",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2023, Boston, MA, USA, July 10-12, 2023, Roxana Geambasu and Ed Nightingale (Eds.). USENIX Association, 397–417. https://www.usenix.org/conference/osdi23/presentation/jiang",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 23,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. 2010. Automated SQL query generation for systematic testing of database engines. In ASE 2010, 25th IEEE/ACM International Conference on Automated Software Engineering, Antwerp, Belgium, September 20-24, 2010, Charles Pecheur, Jamie Andrews, and Elisabetta Di Nitto (Eds.). ACM, 329–332. https://doi.org/ 10.1145/1858996.1859063",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. 2011. Systematic Testing of Database Engines Using a Relational Constraint Solver. In Fourth IEEE International Conference on Software Testing, Verification and Validation, ICST 2011, Berlin, Germany, March 21-25, 2011. IEEE Computer Society, 50–59. https://doi.org/10.1109/ICST.2011.21",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Keqiang Li, Siyang Weng, Peiyuan Liu, Lyu Ni, Chengcheng Yang, Rong Zhang, Xuan Zhou, Jianghang Lou, Gui Huang, Weining Qian, and Aoying Zhou. 2023. Leopard: A Black-Box Approach for Efficiently Verifying Various Isolation Levels. In 39th IEEE International Conference on Data Engineering, ICDE 2023, Anaheim, CA, USA, April 3-7, 2023. IEEE, 722–735. https://doi.org/10.1109/ICDE55515.2023.00061",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Yang Liu, Wei Chen, Yanhong A. Liu, and Jun Sun. 2009. Model Checking Linearizability via Refinement. In FM 2009: Formal Methods, Second World Congress, Eindhoven, The Netherlands, November 2-6, 2009. Proceedings (Lecture Notes in Computer Science, Vol. 5850), Ana Cavalcanti and Dennis Dams (Eds.). Springer, 321–337. https://doi.org/10.1007/9783-642-05089-3_21",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Kartik Nagar and Suresh Jagannathan. 2018. Automated Detection of Serializability Violations Under Weak Consistency. In29th International Conference on Concurrency Theory, CONCUR 2018, September 4-7, 2018, Beijing, China (LIPIcs, Vol. 118), Sven Schewe and Lijun Zhang (Eds.). Schloss Dagstuhl-Leibniz-Zentrum für Informatik, 41:1–41:18. https: //doi.org/10.4230/LIPICS.CONCUR.2018.41",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Lauren Pick, Amanda Xu, Ankush Desai, Sanjit Seshia, and Aws Albarghouthi. 2024. Checking Observational Correctness of Database Systems Artifact (Troubadour). https://doi.org/10.5281/zenodo.14918621",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "PostgreSQL Global Development Group. 2023. PostgreSQL. http://www.postgresql.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Shaz Qadeer. 2003. Verifying Sequential Consistency on Shared-Memory Multiprocessors by Model Checking. IEEE Trans. Parallel Distributed Syst. 14, 8 (2003), 730–741. https://doi.org/10.1109/TPDS.2003.1225053",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 32,
+      "text": "Arnab Sinha and Sharad Malik. 2010. Runtime checking of serializability in software transactional memory. In 24th IEEE International Symposium on Parallel and Distributed Processing, IPDPS 2010, Atlanta, Georgia, USA, 19-23 April 2010 - Conference Proceedings. IEEE, 1–12. https://doi.org/10.1109/IPDPS.2010.5470389",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Zhihang Sun, Hongyu Fan, and Fei He. 2022. Consistency-preserving propagation for SMT solving of concurrent program verification. Proc. ACM Program. Lang. 6, OOPSLA2 (2022), 929–956. https://doi.org/10.1145/3563321",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Cheng Tan, Changgeng Zhao, Shuai Mu, and Michael Walfish. 2020. Cobra: Making Transactional Key-Value Stores Verifiably Serializable. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 63–80. https://www.usenix.org/conference/osdi20/presentation/tan",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Transaction Processing Performance Council. 2010. TPC-C Benchmark Revision 5.11.0. https://www.tpc.org/tpcc/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Yuepeng Wang, Isil Dillig, Shuvendu K. Lahiri, and William R. Cook. 2018. Verifying equivalence of database-driven applications. Proc. ACM Program. Lang. 2, POPL (2018), 56:1–56:29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "ANSI X3.135-1992. November 1992. American National Standard for Information Systems – Database Language – SQL.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Yu Xia, Xiangyao Yu, Matthew Butrovich, Andrew Pavlo, and Srinivas Devadas. 2022. Litmus: Towards a Practical Database Management System with Verifiable ACID Properties and Transaction Correctness. In SIGMOD ’22: International Conference on Management of Data, Philadelphia, PA, USA, June 12 - 17, 2022, Zachary G. Ives, Angela Bonifati, and Amr El Abbadi (Eds.). ACM, 1478–1492. https://doi.org/10.1",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Rachid Zennou, Ranadeep Biswas, Ahmed Bouajjani, Constantin Enea, and Mohammed Erradi. 2022. Checking causal consistency of distributed databases. Computing 104, 10 (2022), 2181–2201. https://doi.org/10.1007/S00607-021-00911-3",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Jian Zhang, Ye Ji, Shuai Mu, and Cheng Tan. 2023. Viper: A Fast Snapshot Isolation Checker. In Proceedings of the Eighteenth European Conference on Computer Systems, EuroSys 2023, Rome, Italy, May 8-12, 2023, Giuseppe Antonio Di Luna, Leonardo Querzoni, Alexandra Fedorova, and Dushyanth Narayanan (Eds.). ACM, 654–671. https://doi.org/10. 1145/3552326.3567492 Proc. ACM Program. Lang., Vol. 9, No. O",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Shao Jie Zhang. 2011. Scalable automatic linearizability checking. In Proceedings of the 33rd International Conference on Software Engineering, ICSE 2011, Waikiki, Honolulu, HI, USA, May 21-28, 2011, Richard N. Taylor, Harald C. Gall, and Nenad Medvidovic (Eds.). ACM, 1185–1187. https://doi.org/10.1145/1985793.1986037",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event, USA, November 9-13, 2020, Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna (Eds.). ACM, 955–970. https://doi.org/10.1145/",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 22,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2023, Boston, MA, USA, July 10-12, 2023, Roxana Geambasu and Ed Nightingale (Eds.). USENIX Association, 397–417. https://www.usenix.org/conference/osdi23/presentation/jiang",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 31,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Troubadour can verify that a trace does not demonstrate any of the classes of bugs detected by DBMS fuzzer SQLancer for the fragment of SQL that Troubadour supports as well as detect the presence of such bugs.",
+      "context_before": "t A. Seshia, and Aws Albarghouthi is more variability in solver performance for the correct example. Solver heuristics lead to faster verification for certain values of 𝑘. Summary. Our improved encoding leads to a significant reduction in checking times for most example sizes and results in less variability in the solving times. While for correct examples, the baseline sometimes outperforms the improved encoding, the improved encoding consistently leads to solving times under two seconds, even where solving times exceed 40 minutes for the baseline. 8.2 RQ2: Classes of Bugs Considered Semantic.",
+      "context_after": "We used Troubadour to check correctness of queries by generating a transaction for the query and a correct observation for the transaction. We then ranTroubadour on the transaction and observation. It showed that all 11 of the considered examples were observationally correct in under a second total. We also ran the tool on the query and the original incorrect response observed by SQLancer, and Troubadour reported errors for all the examples in under a second total as well, demonstrating its ability to detect semantic bugs. We considered benchmarks based on MySQL queries 2, 3, 9, 15, 16, 18, a",
+      "section": "8.2 RQ2: Classes of Bugs Considered",
+      "page": 20,
+      "char_offset": 72881,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We also ran the tool on the query and the original incorrect response observed by SQLancer, and Troubadour reported errors for all the examples in under a second total as well, demonstrating its ability to detect semantic bugs.",
+      "context_before": "g times exceed 40 minutes for the baseline. 8.2 RQ2: Classes of Bugs Considered Semantic. Troubadour can verify that a trace does not demonstrate any of the classes of bugs detected by DBMS fuzzer SQLancer for the fragment of SQL that Troubadour supports as well as detect the presence of such bugs. We used Troubadour to check correctness of queries by generating a transaction for the query and a correct observation for the transaction. We then ranTroubadour on the transaction and observation. It showed that all 11 of the considered examples were observationally correct in under a second total.",
+      "context_after": "We considered benchmarks based on MySQL queries 2, 3, 9, 15, 16, 18, and 19 and TiDB queries 1, 3, 10, and 13 generated by SQLancer1, which detected real bugs in MySQL and TiDB DBMS implementations. Other queries used features (e.g., views) not currently supported by Troubadour. Isolation and data anomalies. Troubadour can also be used to ensure the absence of observable isolation level violations and detect anomalies as isolation checking tools can. Many tools are specialized for finding anomalies for specific isolation levels. For example, Cobra supports only anomalies for the serializable",
+      "section": "8.2 RQ2: Classes of Bugs Considered",
+      "page": 20,
+      "char_offset": 73392,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Testing tools like SQLancer and TxCheck can detect certain semantic bugs but clearly cannot perform verification of observational correctness [ 22, 31].",
+      "context_before": "aining a dirty write for isolation level RC), Troubadour reported that the log demonstrated a bug in under half a second for each log. When run on a log containing an anomaly allowed by the provided isolation level (e.g., a log containing a write skew for isolation level SI), it verified observational correctness in under half a second for each log. Observational correctness. While Troubadour can detect semanticsor isolation-related bugs in logs separately, its purpose is primarily to verify the absence of observable semantic or isolation correctness errors with respect to logged transactions.",
+      "context_after": "These tools further cannot catch bugs that require reasoning about isolation levels, such as the bug for the example in §2. Techniques that reason only about isolation levels similarly cannot show the absence of semantic bugs and may also miss errors because they lack information about which transactions write which values. Tools like Elle, Cobra ,Viper, and PolySI can only be applied in settings where transactions’ reads and writes are all known, since they construct dependency graphs from reads. They would 1Found here: https://www.manuelrigger.at/dbms-bugs/ Proc. ACM Program. Lang., Vol. 9,",
+      "section": "8.2 RQ2: Classes of Bugs Considered",
+      "page": 20,
+      "char_offset": 75779,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "ternary logic partitioning",
+      "technique": "tlp",
+      "sentence": "There has been much work on DBMS fuzzing for SQL using a variety of techniques [ 3,15,31,42], such as ternary logic partitioning and mutation-based fuzzing.",
+      "context_before": "aker consistency guarantees than considered by automated verifiers for consistency levels; and (2) we operate in a black-box setting without access to the state of the system, meaning that we must reason about semantics to infer whether there exists a sequence of states such that an execution is correct. Testing. The testing approaches for DBMSs that we discuss here generate test cases that expose bugs in the DBMS implementation, and do not test arbitrary transactions and client-side transaction logs for correctness. They therefore solve a much different problem than DBMS correctness auditing.",
+      "context_after": "These tools focus exclusively on semantic correctness of intra-transactional operations as they do not reason about isolation levels. ADUSA [ 23,24] uses constraint solving in Alloy to systematically test DBMSs. The TxCheck fuzzer employs some limited reasoning about isolation levels [ 22]. The MonkeyDB work not only tests for semantic bugs but also for isolation bugs in DBMSs [ 8]. It proposes operational semantics (based on previous work [ 7]) and handles a limited subset of SQL by compiling to key-value store Proc. ACM Program. Lang., Vol. 9, No. OOPSLA1, Article 139. Publication date: Apr",
+      "section": "8.5 Industrial Case Study",
+      "page": 24,
+      "char_offset": 92150,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 22]",
+      "technique": null,
+      "sentence": "The TxCheck fuzzer employs some limited reasoning about isolation levels [ 22].",
+      "context_before": "generate test cases that expose bugs in the DBMS implementation, and do not test arbitrary transactions and client-side transaction logs for correctness. They therefore solve a much different problem than DBMS correctness auditing. There has been much work on DBMS fuzzing for SQL using a variety of techniques [ 3,15,31,42], such as ternary logic partitioning and mutation-based fuzzing. These tools focus exclusively on semantic correctness of intra-transactional operations as they do not reason about isolation levels. ADUSA [ 23,24] uses constraint solving in Alloy to systematically test DBMSs.",
+      "context_after": "The MonkeyDB work not only tests for semantic bugs but also for isolation bugs in DBMSs [ 8]. It proposes operational semantics (based on previous work [ 7]) and handles a limited subset of SQL by compiling to key-value store Proc. ACM Program. Lang., Vol. 9, No. OOPSLA1, Article 139. Publication date: April 2025. Checking Observational Correctness of Database Systems 139:25 operations [ 8]. The semantics is used to test DBMS correctness with respect to a set of transactions for which all read values are observable (i.e., there is no read nondeterminism). Checking isolation levels. Most work",
+      "section": "8.5 Industrial Case Study",
+      "page": 24,
+      "char_offset": 92519,
+      "cited_reference": {
+        "number": 22,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2023, Boston, MA, USA, July 10-12, 2023, Roxana Geambasu and Ed Nightingale",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 22]",
+      "technique": null,
+      "sentence": "While these values may be gotten by instrumenting transactions [ 22], instrumentation implicitly relies on the semantic correctness of the DBMS and may not be possible for certain DBMSs [16].",
+      "context_before": "g problem: given an isolation level 𝐼and set of transactions Twith known reads and writes, does there exist an execution of Tthat meets constraints for 𝐼?This problem is similar to the problem of proving observational correctness with respect to a log, but importantly includes no reasoning about the semantics of the transactions. The assumption that allread and written values are observable either limits the kinds of transactions that can be considered (e.g., the Deposit andTransfer transactions from Fig. 2 could not be used as-is), requires a non-black-box DBMS, or leads to bugs being missed.",
+      "context_after": "Elle deduces dependencies between transactions based on read and written values, and then reasons possible transaction dependency graphs according to Adya’s formalism [ 5,6]. Related techniques do specialized checking of certain isolation levels [ 21,27,34,40], such as serializability and SI. Some of these further make the assumption that each value in the database is unique [ 21,34]. Several of these techniques rely on checking for graph properties, where this checking is encoded as a SAT or SMT problem. While we also reduce DBMS correctness auditing to SMT solving, the queries we pose are m",
+      "section": "8.5 Industrial Case Study",
+      "page": 25,
+      "char_offset": 93852,
+      "cited_reference": {
+        "number": 22,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2023, Boston, MA, USA, July 10-12, 2023, Roxana Geambasu and Ed Nightingale",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:21:29Z",
+    "is_model_written": true,
+    "summary": "This paper checks database systems as a client would, verifying both transaction semantics and isolation guarantees from observations of running transactions against a black-box system. Observational correctness holds if some correct execution under the given isolation level could have produced the observations. The technique rests on symbolic encodings discharged with an SMT solver. The tool, Troubadour, was applied to PostgreSQL and an industrial system under development, finding two new bugs.",
+    "narrative": "SQLancer supplies the evaluation's yardstick: Troubadour is shown to verify that a trace exhibits none of the classes of bug SQLancer detects for the SQL fragment it supports, and was run on the queries and incorrect responses SQLancer had observed, reporting errors for all of them. The paper is careful that testing tools like SQLancer detect bugs but cannot verify observational correctness.",
+    "roles": {
+      "M1": "result_comparison",
+      "M2": "result_comparison",
+      "M3": "background",
+      "M4": "definition",
+      "M5": "background",
+      "M6": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M1",
+          "M2"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "Troubadour can verify that a trace does not demonstrate any of the classes of bugs detected by DBMS fuzzer SQLancer for the fragment of SQL that Troubadour supports as well as detect the presence of such bugs.",
+            "section": "8.2 RQ2: Classes of Bugs Considered",
+            "page": 20
+          },
+          {
+            "mention_id": "M2",
+            "sentence": "We also ran the tool on the query and the original incorrect response observed by SQLancer, and Troubadour reported errors for all the examples in under a second total as well, demonstrating its ability to detect semantic bugs.",
+            "section": "8.2 RQ2: Classes of Bugs Considered",
+            "page": 20
+          }
+        ],
+        "reasoning": "M1 and M2 report running Troubadour on the bug classes and on the actual incorrect responses SQLancer found, and comparing what each detects. That is an empirical comparison of coverage of bug classes rather than a bug-count race, but it is a comparison."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3728908.json b/_data/papers/paper_doi_10_1145_3728908.json
new file mode 100644
index 0000000..5a343c5
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3728908.json
@@ -0,0 +1,990 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:50Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3728908",
+    "title": "QTRAN: Extending Metamorphic-Oracle Based Logical Bug Detection Techniques for Multiple-DBMS Dialect Support",
+    "authors": [
+      "Li Lin",
+      "Qinglin Zhu",
+      "Hongqiao Chen",
+      "Zhuangda Wang",
+      "Rongxin Wu",
+      "Xiaoheng Xie"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3728908",
+    "arxiv_id": null,
+    "s2_paper_id": "196fb1deaa2eda163dcfc0a7e0a7dec93228fea3",
+    "url": "https://doi.org/10.1145/3728908",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3728908",
+      "retrieved_at": "2026-09-09T01:30:50Z",
+      "chars": 85673,
+      "content_sha256": "sha256:2ced0a2090d0c5427802bba7679fe6bf69b0db9be87b76053074f4b235fbd67c"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 22,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 4444
+      },
+      {
+        "number": "2",
+        "title": "Background and Motivation",
+        "start": 14673
+      },
+      {
+        "number": "3",
+        "title": "Approach",
+        "start": 23611
+      },
+      {
+        "number": "3.1",
+        "title": "Transfer Phase",
+        "start": 27970
+      },
+      {
+        "number": "3.2",
+        "title": "Mutation Phase",
+        "start": 35149
+      },
+      {
+        "number": "4",
+        "title": "Evaluation",
+        "start": 39630
+      },
+      {
+        "number": "4.1",
+        "title": "Q1. Effectiveness and Efficiency",
+        "start": 43081
+      },
+      {
+        "number": "4.2",
+        "title": "Q2. Bug Detection",
+        "start": 49336
+      },
+      {
+        "number": "4.3",
+        "title": "Q3. Contributions of Dialect Mapping",
+        "start": 56944
+      },
+      {
+        "number": "4.4",
+        "title": "Q4. Contributions of Fine-tuning",
+        "start": 59468
+      },
+      {
+        "number": "5",
+        "title": "Discussion",
+        "start": 66128
+      },
+      {
+        "number": "6",
+        "title": "Related Work",
+        "start": 69419
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 74646
+      },
+      {
+        "number": "8",
+        "title": "Data Availability",
+        "start": 75277
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2015. SQLsmith. https://github.com/anse1/sqlsmith. Accessed: 2024-10-24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2022. ANSI Standard. https://www.ansi.org/. Accessed: 2024-10-24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2023. SQLancer. https://github.com/sqlancer/sqlancer. Accessed: 2024-10-24.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "2024. C to Go Translator. https://github.com/gotranspile/cxgo. Accessed: 2024-10-13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2024. C2Rust Transpiler. https://github.com/immunant/c2rust. Accessed: 2024-10-13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2024. Java 2 CSharp Translator for Eclipse. https://sourceforge.net/projects/j2cstranslator/. Accessed: 2024-10-13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2024. Sharpen-Automated Java to C# Conversion. https://github.com/mono/sharpen. Accessed: 2024-10-13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Atul Adya. 1999. Weak consistency: a generalized theory and optimistic implementations for distributed transactions. (1999).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. 2007. A genetic approach for random testing of database systems. In Proceedings of the 33rd international conference on Very large data bases. 1243–1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Tom B Brown. 2020. Language models are few-shot learners. arXiv preprint arXiv:2005.14165 (2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Aakanksha Chowdhery, Sharan Narang, Jacob Devlin, Maarten Bosma, Gaurav Mishra, Adam Roberts, Paul Barham, Hyung Won Chung, Charles Sutton, Sebastian Gehrmann, et al .2023. Palm: Scaling language modeling with pathways. Journal of Machine Learning Research 24, 240 (2023), 1–113.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, et al .2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168 (2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Ziyu Cui, Wensheng Dou, Qianwang Dai, Jiansen Song, Wei Wang, Jun Wei, and Dan Ye. 2022. Differentially testing database transactions for fun and profit. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Arghavan Moradi Dakhel, Amin Nikanjam, Vahid Majdinasab, Foutse Khomh, and Michel C Desmarais. 2024. Effective test generation using pre-trained large language models and mutation testing. Information and Software Technology 171 (2024), 107468.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Victor Dantas. 2023. Large Language Model Powered Test Case Generation for Software Applications. (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Haoran Peng, Chenyuan Yang, and Lingming Zhang. 2023. Large language models are zero-shot fuzzers: Fuzzing deep-learning libraries via large language models. In Proceedings of the 32nd ACMSIGSOFT international symposium on software testing and analysis. 423–435.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Chenyuan Yang, Shizhuo Dylan Zhang, Shujing Yang, and Lingming Zhang. 2023. Large language models are edge-case fuzzers: Testing deep learning libraries via fuzzgpt. arXiv preprint arXiv:2304.02014 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Lukas Eder. 2024. JOOQ: Fluent API for Typesafe SQL Query Construction and Translation. https://www.jooq.org/. Accessed: 2024-12-29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Daniela Florescu, Alon Levy, and Alberto Mendelzon. 1998. Database techniques for the World-Wide Web: A survey. ACM Sigmod Record 27, 3 (1998), 59–74.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023. Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis. In 2023 USENIX Annual Technical Conference (USENIX ATC 23). 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "William E Howden. 1978. Theoretical and empirical studies of program testing. IEEE Transactions on Software Engineering 4 (1978), 293–298.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Ali Reza Ibrahimzada, Yang Chen, Ryan Rong, and Reyhaneh Jabbarvand. 2023. Automated bug generation in the era of large language models. arXiv preprint arXiv:2310.02407 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Yu Jiang, Jie Liang, Fuchen Ma, Yuanliang Chen, Chijin Zhou, Yuheng Shen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Shanshan Li, et al .2024. When fuzzing meets llms: Challenges and opportunities. In Companion Proceedings of the 32nd ACM International Conference on the Foundations of Software Engineering. 492–496.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. {DynSQL}: Stateful Fuzzing for Database Management Systems with Complex and Valid {SQL}Query Generation. In 32nd USENIX Security Symposium (USENIX Security 23). 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting Logic Bugs in Database Engines via Equivalent Expression Transformation. In 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24). 821–835.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. Apollo: Automatic detection and diagnosis of performance regressions in database systems. Proceedings of the VLDB Endowment 13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Zhenwen Li and Tao Xie. 2024. Using LLM to select the right SQL Query from candidates. arXiv preprint arXiv:2401.02115 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting logical bugs of {DBMS}with coverage-based guidance. In 31st USENIX Security Symposium (USENIX Security 22). 4309–4326. Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA033. Publication date: July 2025. ISSTA033:22 Li Lin, Qinglin Zhu, Hongqiao Chen, Zhuangda Wang, Rongxin Wu, and Xiaoheng Xie",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of DBMS with Coverage-based Guidance. In 31st USENIX Security Symposium (USENIX Security 22). USENIX Association, Boston, MA, 4309–4326. https://www.usenix. org/conference/usenixsecurity22/presentation/liang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Li Lin, Zongyin Hao, Chengpeng Wang, Zhuangda Wang, Rongxin Wu, and Gang Fan. 2024. SQLess: Dialect-Agnostic SQL Query Simplification. In Proceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. 743–754.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Yunlong Lyu, Yuxuan Xie, Peng Chen, and Hao Chen. 2023. Prompt Fuzzing for Fuzz Driver Generation. arXiv preprint arXiv:2312.17677 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Toby Mao. 2024. SQLGlot: Python SQL Parser and Translator. https://github.com/tobymao/sqlglot. Accessed: 2024-12-29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "OpenAI. 2024. GPT-4 Turbo Fine-Tuning. https://openai.com/index/gpt-4o-fine-tuning/. Accessed: 2024-10-30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Terence Parr. 2013. The Definitive ANTLR 4 Reference. Pragmatic Bookshelf, Chapter 9.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "PingCap. 2023. go randgen. https://github.com/pingcap/go-randgen. [Online; accessed 2-October-2024].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Alec Radford. 2018. Improving language understanding by generative pre-training. (2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 40,
+      "text": "Manuel Rigger and Zhendong Su. 2022. Intramorphic testing: A new approach to the test oracle problem. In Proceedings of the 2022 ACMSIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software. 128–136.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "Disha Shrivastava, Hugo Larochelle, and Daniel Tarlow. 2023. Repository-level prompt generation for large language models of code. In International Conference on Machine Learning. PMLR, 31693–31715.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Donald R Slutz. 1998. Massive stochastic testing of SQL. In VLDB, Vol. 98. Citeseer, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Solid IT. 2024. DB-Engines Ranking. https://db-engines.com/en/ranking. Accessed: September 3, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing database systems via differential query execution. In Engineering (ICSE). IEEE, 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "QTRAN Team. 2024. QTRANBugReports. https://github.com/QTRANll/QTRAN-BUGLIST.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al .2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems 35 (2022), 24824–24837.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Chunqiu Steven Xia, Matteo Paltenghi, Jia Le Tian, Michael Pradel, and Lingming Zhang. 2024. Fuzz4all: Universal fuzzing with large language models. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Congying Xu, Songqiang Chen, Jiarong Wu, Shing-Chi Cheung, Valerio Terragni, Hengcheng Zhu, and Jialun Cao. 2024. MR-Adopt: Automatic Deduction of Input Transformation Function for Metamorphic Testing. arXiv preprint arXiv:2408.15815 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Guixin Ye, Zhanyong Tang, Shin Hwei Tan, Songfang Huang, Dingyi Fang, Xiaoyang Sun, Lizhong Bian, Haibo Wang, and Zheng Wang. 2021. Automated conformance testing for JavaScript engines via deep compiler fuzzing. In Proceedings of the 42nd ACMSIGPLAN international conference on programming language design and implementation. 435–450.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Cen Zhang, Mingqiang Bai, Yaowen Zheng, Yeting Li, Wei Ma, Xiaofei Xie, Yuekang Li, Limin Sun, and Yang Liu. 2023. Understanding large language model based fuzz driver generation. arXiv e-prints (2023), arXiv–2307.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Qinggang Zhang, Junnan Dong, Hao Chen, Wentao Li, Feiran Huang, and Xiao Huang. 2024. Structure guided large language model for sql generation. arXiv preprint arXiv:2402.13284 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Xuanhe Zhou, Chengliang Chai, Guoliang Li, and Ji Sun. 2020. Database meets artificial intelligence: A survey. IEEE Transactions on Knowledge and Data Engineering 34, 3 (2020), 1096–1116. Received 2024-10-31; accepted 2025-03-31 Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA033. Publication date: July 2025.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 3,
+      "text": "2023. SQLancer. https://github.com/sqlancer/sqlancer. Accessed: 2024-10-24.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 39,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "Manuel Rigger and Zhendong Su. 2022. Intramorphic testing: A new approach to the test oracle problem. In Proceedings of the 2022 ACMSIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software. 128–136.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 8,20,24,26,37–39]",
+      "technique": "norec",
+      "sentence": "However, the complex code logic and diverse functionalities of DBMSs often make them susceptible to bugs, especially logical bugs that result in incorrect result sets being returned without obvious symptoms [ 8,20,24,26,37–39].",
+      "context_before": "Zhu, Hongqiao Chen, Zhuangda Wang, Rongxin Wu, and Xiaoheng Xie Additional Key Words and Phrases: Database Testing, Metamorphic Testing, SQL Dialects ACM Reference Format: Li Lin, Qinglin Zhu, Hongqiao Chen, Zhuangda Wang, Rongxin Wu, and Xiaoheng Xie. 2025. QTRAN: Extending Metamorphic-Oracle Based Logical Bug Detection Techniques for Multiple-DBMS Dialect Support. Proc. ACM Softw. Eng. 2, ISSTA, Article ISSTA033 (July 2025), 22 pages. https://doi.org/10.1145/3728908 1 Introduction Database Management Systems (DBMSs) are crucial in applications such as online banking and e-commerce [ 19,52].",
+      "context_after": "Automated testing approaches use test oracles [ 21] to verify the correctness of DBMS results. Notably, metamorphic oracle, a widelyused method that constructs SQL statements maintaining either exact [ 25,28,29,37–39,44] or approximate equivalence [ 20] with the original query, has been recognized as state-of-the-art in DBMS testing for logical bug detection [ 20,38,40]. The detection of logical bugs is facilitated by comparing the results of these mutated queries with those of the original queries. Unlike traditional fuzzing [ 1,24], it is essential to ensure both syntactic and semantic corr",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4570,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 25,28,29,37–39,44]",
+      "technique": "norec",
+      "sentence": "Notably, metamorphic oracle, a widelyused method that constructs SQL statements maintaining either exact [ 25,28,29,37–39,44] or approximate equivalence [ 20] with the original query, has been recognized as state-of-the-art in DBMS testing for logical bug detection [ 20,38,40].",
+      "context_before": "Techniques for Multiple-DBMS Dialect Support. Proc. ACM Softw. Eng. 2, ISSTA, Article ISSTA033 (July 2025), 22 pages. https://doi.org/10.1145/3728908 1 Introduction Database Management Systems (DBMSs) are crucial in applications such as online banking and e-commerce [ 19,52]. However, the complex code logic and diverse functionalities of DBMSs often make them susceptible to bugs, especially logical bugs that result in incorrect result sets being returned without obvious symptoms [ 8,20,24,26,37–39]. Automated testing approaches use test oracles [ 21] to verify the correctness of DBMS results.",
+      "context_after": "The detection of logical bugs is facilitated by comparing the results of these mutated queries with those of the original queries. Unlike traditional fuzzing [ 1,24], it is essential to ensure both syntactic and semantic correctness, while strictly preserving the metamorphic relationship between the queries. We call this approach MOLT (Metamorphic- Oracle based Logical Bug Detection Technique). To generate SQL statement pairs suitable for MOLT s, including original queries andmutated queries, a general idea is to model the SQL grammar as an Abstract Syntax Tree (AST). This AST model forms the",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4893,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 20,38,40]",
+      "technique": "tlp",
+      "sentence": "Notably, the metamorphic testing based approach MOLT has been recognized to be state-of-the-art in DBMS testing for logical bug detection [ 20,38,40].",
+      "context_before": "of these SQL dialects are not compatible with one another, as similar features might be implemented with distinct grammar variations across different DBMSs. For example, while the character set comparison and sorting function is implemented as COLLATION in MySQL, it is implemented as COLLATE FOR in PostgreSQL. There are approximately 423 popular DBMS products [ 43], each differing in supported features, such as data types, functions, operators, etc. Metamorphic Testing in DBMSs. Recent years have witnessed tremendous efforts in resolving the test oracle for logical bug detection in the DBMSs.",
+      "context_after": "Metamorphic testing validates database systems by applying transformations to original queries to generate new, mutated queries. Let an original query be represented as 𝑄0, and Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA033. Publication date: July 2025. QTRAN: Extending Metamorphic-Oracle Based Logical Bug Detection Techniques... ISSTA033:5 a set of transformed queries {𝑄1,𝑄2,...,𝑄 𝑛}derived from applying metamorphic transformations. The testing process validates that for any test database 𝐷, the following relation holds: 𝐷(𝑄0)≈𝐷(𝑄𝑗),∀𝑗∈[1,𝑛] where≈represents either equivalence (",
+      "section": "2 Background and Motivation",
+      "page": 4,
+      "char_offset": 16045,
+      "cited_reference": {
+        "number": 38,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Table 1 shows the lines of extra code of SQLancer, SQRight, Pinolo and DQE to adapt for SQLite, PostgreSQL, MySQL, and MariaDB.",
+      "context_before": "ensure uniqueness in the results, thereby removing any potential duplicates. This specific alteration helps highlight discrepancies between the original and mutated query results, which, if present, may indicate a logical bug due to a violation of the defined approximation relation. Motivation. The dependency on grammar limits the adaptability of MOLT s. Specifically, no matter what generation model, SQL parser, or metamorphic mutator, they are all strongly dependent on the DBMS grammar, and adaptation to a new DBMS requires an amount of extra code, which is time-consuming and labor-intensive.",
+      "context_after": "This demonstrates that due to their dependency on grammar, all of these tools demand significant code expansion to accommodate different DBMSs. For example, SQLRight uses over 90,000 lines of code to build the generation model for PostgreSQL. In contrast, Pinolo simplifies its implementation by only supporting the common syntax shared between MySQL and MariaDB. Table 1. The lines of code of SQLancer, SQRight, Pinolo and DQE to adapt for SQLite, PostgreSQL, MySQL, and MariaDB. The numbers of lines of code are calculated from the official repositories of these tools. Due to the complexities in",
+      "section": "2 Background and Motivation",
+      "page": 5,
+      "char_offset": 18543,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The lines of code of SQLancer, SQRight, Pinolo and DQE to adapt for SQLite, PostgreSQL, MySQL, and MariaDB.",
+      "context_before": "tation to a new DBMS requires an amount of extra code, which is time-consuming and labor-intensive. Table 1 shows the lines of extra code of SQLancer, SQRight, Pinolo and DQE to adapt for SQLite, PostgreSQL, MySQL, and MariaDB. This demonstrates that due to their dependency on grammar, all of these tools demand significant code expansion to accommodate different DBMSs. For example, SQLRight uses over 90,000 lines of code to build the generation model for PostgreSQL. In contrast, Pinolo simplifies its implementation by only supporting the common syntax shared between MySQL and MariaDB. Table 1.",
+      "context_after": "The numbers of lines of code are calculated from the official repositories of these tools. Due to the complexities in adapting the grammar, SQLRight does not support MariaDB, and DQE does not support PostgreSQL. Notably, although Pinolo supports both MySQL and MariaDB, it is implemented on a grammar common to both. LoC SQLancer SQLRight Pinolo DQE SQLite 10,020 17,120 - 7,606 PostgreSQL 8,012 93,671 - MySQL 6,512 76,067 8,055 4,225 MariaDB 2,085 - 8,055 2,092 More importantly, there are 423 different DBMSs on the market [ 43], most of which use different grammar. Present-day leading MOLT s ar",
+      "section": "2 Background and Motivation",
+      "page": 5,
+      "char_offset": 19044,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "LoC SQLancer SQLRight Pinolo DQE SQLite 10,020 17,120 - 7,606 PostgreSQL 8,012 93,671 - MySQL 6,512 76,067 8,055 4,225 MariaDB 2,085 - 8,055 2,092 More importantly, there are 423 different DBMSs on the market [ 43], most of which use different grammar.",
+      "context_before": "to build the generation model for PostgreSQL. In contrast, Pinolo simplifies its implementation by only supporting the common syntax shared between MySQL and MariaDB. Table 1. The lines of code of SQLancer, SQRight, Pinolo and DQE to adapt for SQLite, PostgreSQL, MySQL, and MariaDB. The numbers of lines of code are calculated from the official repositories of these tools. Due to the complexities in adapting the grammar, SQLRight does not support MariaDB, and DQE does not support PostgreSQL. Notably, although Pinolo supports both MySQL and MariaDB, it is implemented on a grammar common to both.",
+      "context_after": "Present-day leading MOLT s are designed primarily for well-known DBMSs such as MySQL, PostgreSQL and MariaDB. If we want to adapt them to these different DBMSs, it will take too long and require too many human resources to be economically feasible. Consequently, there exists a pressing need for a technique that allows for the straightforward extension of various MOLT s across multiple DBMS environments. Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA033. Publication date: July 2025. ISSTA033:6 Li Lin, Qinglin Zhu, Hongqiao Chen, Zhuangda Wang, Rongxin Wu, and Xiaoheng Xie Challenges.",
+      "section": "2 Background and Motivation",
+      "page": 5,
+      "char_offset": 19469,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "In our evaluation, we selected four state-of-the-art MOLT s for extension: NoRec [ 37], TLP [38], Pinolo [20], and DQE [44].",
+      "context_before": "g help mutation for metamorphic testing? Test DBMSs. We focused on testing eight widely-used and large-scale open-source DBMSs: MySQL, MariaDB, TiDB, PostgreSQL, SQLite, MonetDB, DuckDB, and ClickHouse. Table 3 describes the detailed information of these databases. Table 3. The demographics of the DBMSs under test DBMS Version GitHub Stars DB-Engine First Release MySQL 8.0.39 8.6K 2 1995 MariaDB 11.5.2 4.6K 15 2009 TiDB 8.3.0 33.1K 77 2017 PostgreSQL 16.3 16.1K 4 1989 SQLite 3.45.1 6.5K 10 2000 MonetDB 11.5.3 375 135 2004 DuckDB 1.1.2 23.6K 57 2018 ClickHouse 24.9.2.42 37.2K 37 2016 Baselines.",
+      "context_after": "Below, we detail each of these tools: (1)NoRec [37]: This technique involves transferring predicates from the WHERE clause to the SELECT clause. A logical bug is identified if the results differ after this transformation. (2)TLP [38]: This method decomposes a single query into three separate queries, each isolated by its predicates. It asserts a logical bug if the collective results of these queries diverge from the original query’s outcome. (3)Pinolo [20]: Modifies query predicates to adjust constraints, expecting either a superset or subset of the original results, which serves to confirm t",
+      "section": "4 Evaluation",
+      "page": 11,
+      "char_offset": 40910,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "Below, we detail each of these tools: (1)NoRec [37]: This technique involves transferring predicates from the WHERE clause to the SELECT clause.",
+      "context_before": "s: MySQL, MariaDB, TiDB, PostgreSQL, SQLite, MonetDB, DuckDB, and ClickHouse. Table 3 describes the detailed information of these databases. Table 3. The demographics of the DBMSs under test DBMS Version GitHub Stars DB-Engine First Release MySQL 8.0.39 8.6K 2 1995 MariaDB 11.5.2 4.6K 15 2009 TiDB 8.3.0 33.1K 77 2017 PostgreSQL 16.3 16.1K 4 1989 SQLite 3.45.1 6.5K 10 2000 MonetDB 11.5.3 375 135 2004 DuckDB 1.1.2 23.6K 57 2018 ClickHouse 24.9.2.42 37.2K 37 2016 Baselines. In our evaluation, we selected four state-of-the-art MOLT s for extension: NoRec [ 37], TLP [38], Pinolo [20], and DQE [44].",
+      "context_after": "A logical bug is identified if the results differ after this transformation. (2)TLP [38]: This method decomposes a single query into three separate queries, each isolated by its predicates. It asserts a logical bug if the collective results of these queries diverge from the original query’s outcome. (3)Pinolo [20]: Modifies query predicates to adjust constraints, expecting either a superset or subset of the original results, which serves to confirm the logic’s soundness. (4)DQE [44]: This approach varies the types of SQL queries—such as SELECT ,UPDATE, and DELETE —while keeping the predicates",
+      "section": "4 Evaluation",
+      "page": 11,
+      "char_offset": 41035,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "(2)TLP [38]: This method decomposes a single query into three separate queries, each isolated by its predicates.",
+      "context_before": "gine First Release MySQL 8.0.39 8.6K 2 1995 MariaDB 11.5.2 4.6K 15 2009 TiDB 8.3.0 33.1K 77 2017 PostgreSQL 16.3 16.1K 4 1989 SQLite 3.45.1 6.5K 10 2000 MonetDB 11.5.3 375 135 2004 DuckDB 1.1.2 23.6K 57 2018 ClickHouse 24.9.2.42 37.2K 37 2016 Baselines. In our evaluation, we selected four state-of-the-art MOLT s for extension: NoRec [ 37], TLP [38], Pinolo [20], and DQE [44]. Below, we detail each of these tools: (1)NoRec [37]: This technique involves transferring predicates from the WHERE clause to the SELECT clause. A logical bug is identified if the results differ after this transformation.",
+      "context_after": "It asserts a logical bug if the collective results of these queries diverge from the original query’s outcome. (3)Pinolo [20]: Modifies query predicates to adjust constraints, expecting either a superset or subset of the original results, which serves to confirm the logic’s soundness. (4)DQE [44]: This approach varies the types of SQL queries—such as SELECT ,UPDATE, and DELETE —while keeping the predicates constant. It anticipates identical operations on rows and flags discrepancies as potential bugs. These MOLT s require an understanding of the SQL syntax for different DBMSs. Table 4 describ",
+      "section": "4 Evaluation",
+      "page": 11,
+      "char_offset": 41257,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Applicability of existing MOLT s for the selected DBMSs DBMS NoREC TLP Pinolo DQE MySQL ✗ ✓ ✓ ✓ MariaDB ✓ ✗ ✓ ✓ TiDB ✗ ✓ ✓ ✓ PostgreSQL ✓ ✗ ✗ ✗ SQLite ✓ ✓ ✗ ✓ MonetDB ✗ ✗ ✗ ✗ DuckDB ✗ ✓ ✗ ✗ ClickHouse ✗ ✗ ✗ ✗ Proc.",
+      "context_before": "er a superset or subset of the original results, which serves to confirm the logic’s soundness. (4)DQE [44]: This approach varies the types of SQL queries—such as SELECT ,UPDATE, and DELETE —while keeping the predicates constant. It anticipates identical operations on rows and flags discrepancies as potential bugs. These MOLT s require an understanding of the SQL syntax for different DBMSs. Table 4 describes the databases supported by these MOLT s. To avoid redundant testing, we employ QTRAN to extend these MOLT s and conduct tests on databases that are not natively supported by them. Table 4.",
+      "context_after": "ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA033. Publication date: July 2025. ISSTA033:12 Li Lin, Qinglin Zhu, Hongqiao Chen, Zhuangda Wang, Rongxin Wu, and Xiaoheng Xie Environment. We conduct the experiments on one server with 104-cores Intel(R) Xeon(R) Gold 6230R CPU @2.10GHz and 500 GB memory. The server operates under the Ubuntu 20.04 OS, with the 5.4.0-135-generic version of the Linux kernel. If not otherwise specified, our experiments are conducted by invoking OpenAI API (version gpt-4o-mini-2024-07-18), with the proposed prompt in Section 3. To harness the deterministic nature o",
+      "section": "4 Evaluation",
+      "page": 11,
+      "char_offset": 42161,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Specifically, NoREC uses SQLite, DQE uses TiDB, while TLP and Pinolo use MySQL.",
+      "context_before": "fectiveness and Efficiency InQ1, we aim to evaluate the effectiveness and efficiency of QTRAN in transferring SQL statement pairs from existing MOLT s across various DBMS environments. This is essential as it directly impacts the success of metamorphic testing. 4.1.1 Experimental Setup. We utilized QTRAN to extend four state-of-the-art MOLT s, denoted as QTRAN +NoREC, QTRAN +TLP, QTRAN +Pinolo, and QTRAN +DQE. These MOLT s were used to generate SQL statement pairs as source pairs for transfer. To generate the original SQL statement pairs, we selected the DBMS that each tool found the most bugs.",
+      "context_after": "Our focus was solely on databases not supported by the source MOLT s. In our experiments, a successfully transferred SQL statement pair is considered effective if it preserves syntactic and semantic correctness when executed on the target DBMSs, and fulfills predefined metamorphic relationships. We refer to such pairs as valid pairs. For SQL statements that fail during the transfer phase, we discard them immediately, foregoing the subsequent mutation phase. This strategy primarily impacts the efficiency of bug detection without causing false positives; we will discuss the impact of the transf",
+      "section": "4.1 Q1. Effectiveness and Efficiency",
+      "page": 12,
+      "char_offset": 43691,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "The source statement pair is derived from the NoREC oracle, which detects bugs by moving the predicate from the “ WHERE ” clause into the “ SUM” function, ensuring semantic consistency between the original and mutated statements.",
+      "context_before": "pers have confirmed or fixed. The Not a bug column indicates the bug reports that developers have not confirmed or will not fix. The Waiting column indicates the bug reports that are pending confirmation. DBMS Reported Confirmed Not a bug Waiting MySQL 2 1 1 0 MariaDB 9 9 0 0 TiDB 4 3 1 0 PostgreSQL 1 1 0 0 SQLite 1 0 1 0 MonetDB 1 0 1 0 DuckDB 2 1 1 0 ClickHouse 4 1 1 2 Total 24 16 6 2 4.2.3 Case Study. True Positive Bug. Figure 6 (b) shows a logical bug in DuckDB identified by QTRAN, triggered by the translation of the “ DATE ” function from SQLite to the “ CURRENT_DATE ” function in DuckDB.",
+      "context_after": "After the QTRAN transfer, however, the results of the original query and the mutated query in DuckDB diverged, indicating a logical bug. Figure 6 (c) shows the results of directly transferring SQL statement pairs using GPT-4o-mini, which resulted in execution failures. The lack of dialect-specific knowledge within LLMs led to the erroneous translation of the SQLite “ DATE ” function to the “ CURRENT_DATE ” function in DuckDB, causing the error “ Scalar Function with name date does not exist! ”. This highlights the superiority of QTRAN in translating SQL statement pairs, which better understan",
+      "section": "4.2 Q2. Bug Detection",
+      "page": 14,
+      "char_offset": 51472,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Sub-figure (a) demonstrates a SQL statement pair generated by NoREC in SQLite without errors.",
+      "context_before": "get Original Query SELECT COUNT(*) FROM v0 WHERE (DATE('now'), c0) != (CAST(NULL AS TEXT), '0’); --(duckdb.duckdb.CatalogException) Catalog Error: Scalar Function with name date does not exist! -- Target Mutated Query SELECT SUM(CASEWHEN (DATE('now'), c0) != (CAST(NULL AS TEXT), '0') THEN 1 ELSE 0 END) FROM v0; --(duckdb.duckdb.CatalogException) Catalog Error: Scalar Function with name date does not exist! (c) Result of Direct SQL Statement Pair Transfer in DuckDB Using GPT-4o-miniFailed! Fig. 6. True Positive Bug: A logical bug in DuckDB when transferring source SQL statement pair in SQLite.",
+      "context_after": "Sub-figure (b) demonstrates the target SQL statement pair in DuckDB transferred by QTRAN, resulting in an unexpected result due to a logical bug. Sub-figure (c) demonstrates the target SQL statement pair directly transferred using GPT-4o-mini, leading to translation failures due to limited understanding of the dialect difference between SQLite and DuckDB. The highlighted yellow background underscores the dialect differences between SQLite and DuckDB. kind of ill-formed;\" this inconsistency is due to different uses of implicit casting in the execution plans for =ANY and ≥ANY. The developers no",
+      "section": "4.2 Q2. Bug Detection",
+      "page": 15,
+      "char_offset": 55068,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": ", NoREC, TLP, Pinolo, and DQE) and then directly transfer both original and mutated queries to the target DBMS in transfer phase.",
+      "context_before": "uration, we implemented QTRAN-POM without using a customized model. We directly apply the same base model, GPT-4o-mini with carefully crafted prompts to mutate transferred original queries. No fine-tuning or additional metamorphic relationship knowledge is incorporated into the model. This baseline examines whether prompt engineering alone is sufficient for generating valid metamorphic pairs without disrupting the predefined relationships between queries. •QTRAN-DTM (Direct Transfer of Existing Mutated Queries). In this setup, we first mutate source original queries using existing MOLT s (e.g.",
+      "context_after": "This baseline tests if existing mutated queries can be effectively adapted to the target DBMS without requiring an LLM-driven mutation process during transfer. We selected 100 successfully transferred target original queries from each of the four baselines: QTRAN +NoREC, QTRAN +TLP, QTRAN +Pinolo, and QTRAN +DQE. We then performed mutations on these queries using QTRAN and two baselines, QTRAN-POM and QTRAN-DTM. We also employed the same evaluation method as in Q1. A valid SQL statement pair must ensure both syntactic and semantic correctness and fulfill predefined metamorphic relationships.",
+      "section": "4.4 Q4. Contributions of Fine-tuning",
+      "page": 17,
+      "char_offset": 61899,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[ 1,13,35,37,38,42]",
+      "technique": "norec",
+      "sentence": "However, existing studies [ 1,13,35,37,38,42] highlight that differential testing is limited in its applicability, as not all DBMSs share the same SQL grammar or operation semantics, despite supporting the core SQL syntax, with each DBMS forming its own dialect [ 42], thus restricting the generality of differential testing.",
+      "context_before": "itigate them, and our findings provide valuable insights into using LLMs for extending MOLT s for logical bug detection in various DBMSs. 6 Related Work Logical Bug Detection in DBMSs. Researchers have proposed several methods for obtaining the test oracle to detect logical bugs, which can be categorized into three types: differential oracle, oracle-guided synthesis oracle and metamorphic oracle [20]. The first category is the differential oracles. Differential oracles [ 9,13,42] detect logical bugs by comparing the execution results of different DBMSs (or different versions of the same DBMS).",
+      "context_after": "The purpose of QTRAN is to extend existing MOLT s to support various dialects, therefore differential oracles are not considered within the scope of this paper. The second category is the oracle-guided synthesis approach [ 39]. PQS synthesizes a query that guarantees to return a specific row using its manually implemented interpreter. If the tested DBMS fails to fetch the row, PQS identifies a logical bug. Although such methods are tailored to the specific DBMS grammar, they only require constructing original queries to check the results. Therefore, the transfer phase of QTRAN can easily exte",
+      "section": "6 Related Work",
+      "page": 19,
+      "char_offset": 69881,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[ 39]",
+      "technique": "pqs",
+      "sentence": "The second category is the oracle-guided synthesis approach [ 39].",
+      "context_before": "tect logical bugs by comparing the execution results of different DBMSs (or different versions of the same DBMS). However, existing studies [ 1,13,35,37,38,42] highlight that differential testing is limited in its applicability, as not all DBMSs share the same SQL grammar or operation semantics, despite supporting the core SQL syntax, with each DBMS forming its own dialect [ 42], thus restricting the generality of differential testing. The purpose of QTRAN is to extend existing MOLT s to support various dialects, therefore differential oracles are not considered within the scope of this paper.",
+      "context_after": "PQS synthesizes a query that guarantees to return a specific row using its manually implemented interpreter. If the tested DBMS fails to fetch the row, PQS identifies a logical bug. Although such methods are tailored to the specific DBMS grammar, they only require constructing original queries to check the results. Therefore, the transfer phase of QTRAN can easily extend this approach. The third category is the metamorphic oracles. To avoid the drawbacks of the differential oracle, many fuzzing methods [ 20,28,37,38] use the metamorphic oracle to detect logical bugs. NoREC [ 37] creates a que",
+      "section": "6 Related Work",
+      "page": 19,
+      "char_offset": 70368,
+      "cited_reference": {
+        "number": 39,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS synthesizes a query that guarantees to return a specific row using its manually implemented interpreter.",
+      "context_before": "BMSs (or different versions of the same DBMS). However, existing studies [ 1,13,35,37,38,42] highlight that differential testing is limited in its applicability, as not all DBMSs share the same SQL grammar or operation semantics, despite supporting the core SQL syntax, with each DBMS forming its own dialect [ 42], thus restricting the generality of differential testing. The purpose of QTRAN is to extend existing MOLT s to support various dialects, therefore differential oracles are not considered within the scope of this paper. The second category is the oracle-guided synthesis approach [ 39].",
+      "context_after": "If the tested DBMS fails to fetch the row, PQS identifies a logical bug. Although such methods are tailored to the specific DBMS grammar, they only require constructing original queries to check the results. Therefore, the transfer phase of QTRAN can easily extend this approach. The third category is the metamorphic oracles. To avoid the drawbacks of the differential oracle, many fuzzing methods [ 20,28,37,38] use the metamorphic oracle to detect logical bugs. NoREC [ 37] creates a query by transferring predicates from the WHERE clause to the SELECT clause; the discovery of a logical bug occu",
+      "section": "6 Related Work",
+      "page": 19,
+      "char_offset": 70435,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "If the tested DBMS fails to fetch the row, PQS identifies a logical bug.",
+      "context_before": "ifferential testing is limited in its applicability, as not all DBMSs share the same SQL grammar or operation semantics, despite supporting the core SQL syntax, with each DBMS forming its own dialect [ 42], thus restricting the generality of differential testing. The purpose of QTRAN is to extend existing MOLT s to support various dialects, therefore differential oracles are not considered within the scope of this paper. The second category is the oracle-guided synthesis approach [ 39]. PQS synthesizes a query that guarantees to return a specific row using its manually implemented interpreter.",
+      "context_after": "Although such methods are tailored to the specific DBMS grammar, they only require constructing original queries to check the results. Therefore, the transfer phase of QTRAN can easily extend this approach. The third category is the metamorphic oracles. To avoid the drawbacks of the differential oracle, many fuzzing methods [ 20,28,37,38] use the metamorphic oracle to detect logical bugs. NoREC [ 37] creates a query by transferring predicates from the WHERE clause to the SELECT clause; the discovery of a logical bug occurs if results vary post-movement. Similarly, TLP [ 38] breaks down a sing",
+      "section": "6 Related Work",
+      "page": 19,
+      "char_offset": 70544,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "citation_marker",
+      "surface": "[ 20,28,37,38]",
+      "technique": "norec",
+      "sentence": "To avoid the drawbacks of the differential oracle, many fuzzing methods [ 20,28,37,38] use the metamorphic oracle to detect logical bugs.",
+      "context_before": "rious dialects, therefore differential oracles are not considered within the scope of this paper. The second category is the oracle-guided synthesis approach [ 39]. PQS synthesizes a query that guarantees to return a specific row using its manually implemented interpreter. If the tested DBMS fails to fetch the row, PQS identifies a logical bug. Although such methods are tailored to the specific DBMS grammar, they only require constructing original queries to check the results. Therefore, the transfer phase of QTRAN can easily extend this approach. The third category is the metamorphic oracles.",
+      "context_after": "NoREC [ 37] creates a query by transferring predicates from the WHERE clause to the SELECT clause; the discovery of a logical bug occurs if results vary post-movement. Similarly, TLP [ 38] breaks down a single query into three, isolating each by its predicates, asserting a logical bug if the collective results diverge from the original query’s outcome. SQLancer [ 3] integrates the above techniques and has been deployed to test various DBMSs. DQE [ 44] varies query types—such as SELECT ,UPDATE, and DELETE —keeping predicates constant and anticipates identical row operations, flagging discrepan",
+      "section": "6 Related Work",
+      "page": 19,
+      "char_offset": 70871,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [ 37] creates a query by transferring predicates from the WHERE clause to the SELECT clause; the discovery of a logical bug occurs if results vary post-movement.",
+      "context_before": "synthesis approach [ 39]. PQS synthesizes a query that guarantees to return a specific row using its manually implemented interpreter. If the tested DBMS fails to fetch the row, PQS identifies a logical bug. Although such methods are tailored to the specific DBMS grammar, they only require constructing original queries to check the results. Therefore, the transfer phase of QTRAN can easily extend this approach. The third category is the metamorphic oracles. To avoid the drawbacks of the differential oracle, many fuzzing methods [ 20,28,37,38] use the metamorphic oracle to detect logical bugs.",
+      "context_after": "Similarly, TLP [ 38] breaks down a single query into three, isolating each by its predicates, asserting a logical bug if the collective results diverge from the original query’s outcome. SQLancer [ 3] integrates the above techniques and has been deployed to test various DBMSs. DQE [ 44] varies query types—such as SELECT ,UPDATE, and DELETE —keeping predicates constant and anticipates identical row operations, flagging discrepancies as bugs. Pinolo [ 20] modifies query predicates to adjust constraints, expecting Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA033. Publication date: July",
+      "section": "6 Related Work",
+      "page": 19,
+      "char_offset": 71009,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Similarly, TLP [ 38] breaks down a single query into three, isolating each by its predicates, asserting a logical bug if the collective results diverge from the original query’s outcome.",
+      "context_before": "h the row, PQS identifies a logical bug. Although such methods are tailored to the specific DBMS grammar, they only require constructing original queries to check the results. Therefore, the transfer phase of QTRAN can easily extend this approach. The third category is the metamorphic oracles. To avoid the drawbacks of the differential oracle, many fuzzing methods [ 20,28,37,38] use the metamorphic oracle to detect logical bugs. NoREC [ 37] creates a query by transferring predicates from the WHERE clause to the SELECT clause; the discovery of a logical bug occurs if results vary post-movement.",
+      "context_after": "SQLancer [ 3] integrates the above techniques and has been deployed to test various DBMSs. DQE [ 44] varies query types—such as SELECT ,UPDATE, and DELETE —keeping predicates constant and anticipates identical row operations, flagging discrepancies as bugs. Pinolo [ 20] modifies query predicates to adjust constraints, expecting Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA033. Publication date: July 2025. ISSTA033:20 Li Lin, Qinglin Zhu, Hongqiao Chen, Zhuangda Wang, Rongxin Wu, and Xiaoheng Xie either a superset or subset of the original results to confirm logic soundness. EET [ 2",
+      "section": "6 Related Work",
+      "page": 19,
+      "char_offset": 71177,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 3] integrates the above techniques and has been deployed to test various DBMSs.",
+      "context_before": "the transfer phase of QTRAN can easily extend this approach. The third category is the metamorphic oracles. To avoid the drawbacks of the differential oracle, many fuzzing methods [ 20,28,37,38] use the metamorphic oracle to detect logical bugs. NoREC [ 37] creates a query by transferring predicates from the WHERE clause to the SELECT clause; the discovery of a logical bug occurs if results vary post-movement. Similarly, TLP [ 38] breaks down a single query into three, isolating each by its predicates, asserting a logical bug if the collective results diverge from the original query’s outcome.",
+      "context_after": "DQE [ 44] varies query types—such as SELECT ,UPDATE, and DELETE —keeping predicates constant and anticipates identical row operations, flagging discrepancies as bugs. Pinolo [ 20] modifies query predicates to adjust constraints, expecting Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA033. Publication date: July 2025. ISSTA033:20 Li Lin, Qinglin Zhu, Hongqiao Chen, Zhuangda Wang, Rongxin Wu, and Xiaoheng Xie either a superset or subset of the original results to confirm logic soundness. EET [ 25] transforms these expressions into semantically equivalent ones to construct queries that",
+      "section": "6 Related Work",
+      "page": 19,
+      "char_offset": 71364,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M7"
+      ],
+      "describes_as_state_of_the_art": [
+        "M7"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:15:22Z",
+    "is_model_written": true,
+    "summary": "QTRAN extends existing metamorphic-oracle logic bug detection techniques to DBMSs they were never written for. The obstacle is that these techniques depend on a specific DBMS's grammar to build valid statement pairs, so only a few systems are supported and extending them takes considerable manual work -- the paper counts the lines of code each tool needs per system. QTRAN uses an LLM in two phases: a transfer phase that identifies dialect differences and uses SQL documentation to translate the original queries, and a mutation phase using a fine-tuned model to mutate them while preserving the metamorphic relation. Over 99% of transferred pairs satisfied the relations, and it found 24 logic bugs.",
+    "narrative": "SQLancer's oracles are precisely what QTRAN extends: NoREC and TLP are two of the four state-of-the-art techniques it takes and carries to eight DBMSs, and the paper tabulates which systems each currently supports to show the gap it closes. SQLancer is also its measure of the per-DBMS adaptation cost.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "state_of_the_art",
+      "M4": "motivation",
+      "M5": "motivation",
+      "M6": "motivation",
+      "M7": "state_of_the_art",
+      "M8": "definition",
+      "M9": "definition",
+      "M10": "motivation",
+      "M11": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "QTRAN takes statement pairs from existing implementations of these oracles, which implies running them, but no mention read says its own tool is built on SQLancer's codebase."
+      },
+      "extends_technique": {
+        "value": "yes",
+        "mention_ids": [
+          "M7",
+          "M8",
+          "M9"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M7",
+            "sentence": "In our evaluation, we selected four state-of-the-art MOLT s for extension: NoRec [ 37], TLP [38], Pinolo [20], and DQE [44].",
+            "section": "4 Evaluation",
+            "page": 11
+          },
+          {
+            "mention_id": "M8",
+            "sentence": "Below, we detail each of these tools: (1)NoRec [37]: This technique involves transferring predicates from the WHERE clause to the SELECT clause.",
+            "section": "4 Evaluation",
+            "page": 11
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "(2)TLP [38]: This method decomposes a single query into three separate queries, each isolated by its predicates.",
+            "section": "4 Evaluation",
+            "page": 11
+          }
+        ],
+        "reasoning": "M7 states four state-of-the-art metamorphic oracles were selected for extension, NoREC and TLP among them, and M8 and M9 describe each before it is carried to DBMSs it did not support. Extending them is the paper's contribution.",
+        "techniques": [
+          "norec",
+          "tlp"
+        ]
+      },
+      "compares_with": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The oracles are extended rather than run against QTRAN as competitors; the mentions read do not show a head-to-head comparison."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M3",
+          "M7"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "Notably, the metamorphic testing based approach MOLT has been recognized to be state-of-the-art in DBMS testing for logical bug detection [ 20,38,40].",
+            "section": "2 Background and Motivation",
+            "page": 4
+          },
+          {
+            "mention_id": "M7",
+            "sentence": "In our evaluation, we selected four state-of-the-art MOLT s for extension: NoRec [ 37], TLP [38], Pinolo [20], and DQE [44].",
+            "section": "4 Evaluation",
+            "page": 11
+          }
+        ],
+        "reasoning": "M3 says the metamorphic testing approach has been recognised as state-of-the-art in DBMS testing for logical bug detection, and M7 calls the four selected oracles state-of-the-art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3728953.json b/_data/papers/paper_doi_10_1145_3728953.json
new file mode 100644
index 0000000..7ad903e
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3728953.json
@@ -0,0 +1,920 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:24:34Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3728953",
+    "title": "Detecting Isolation Anomalies in Relational DBMSs",
+    "authors": [
+      "Rui Yang",
+      "Ziyu Cui",
+      "Wensheng Dou",
+      "Yu Gao",
+      "Jiansen Song",
+      "Xudong Xie",
+      "Jun Wei"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3728953",
+    "arxiv_id": null,
+    "s2_paper_id": "189d9b84961fe904388013b4a8f7448854a847c6",
+    "url": "https://doi.org/10.1145/3728953",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3728953",
+      "retrieved_at": "2026-09-08T15:24:34Z",
+      "chars": 94203,
+      "content_sha256": "sha256:ca349ef2551286709577940749bb238ce05ec11ec193a2976d7ea7179600e646"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 23,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 4019
+      },
+      {
+        "number": "2",
+        "title": "Motivation and Preliminaries",
+        "start": 13370
+      },
+      {
+        "number": "2.1",
+        "title": "Motivating Example",
+        "start": 13580
+      },
+      {
+        "number": "2.2",
+        "title": "Transactions and SQL Statements",
+        "start": 16189
+      },
+      {
+        "number": "2.3",
+        "title": "Transaction Execution Histories and Dependency Graphs",
+        "start": 19129
+      },
+      {
+        "number": "2.4",
+        "title": "Isolation Anomalies",
+        "start": 21815
+      },
+      {
+        "number": "3.1",
+        "title": "Database Generation",
+        "start": 28320
+      },
+      {
+        "number": "3.2",
+        "title": "Transaction Generation and SQL Statement Instrumentation",
+        "start": 32961
+      },
+      {
+        "number": "3.3",
+        "title": "Dependency Inference",
+        "start": 39498
+      },
+      {
+        "number": "3.4",
+        "title": "Isolation Anomaly Detection",
+        "start": 51338
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Methodology",
+        "start": 59329
+      },
+      {
+        "number": "5.2",
+        "title": "Isolation Anomaly Detection Results",
+        "start": 63549
+      },
+      {
+        "number": "5.3",
+        "title": "Analyzing Isolation Anomalies in Relational DBMSs",
+        "start": 65896
+      },
+      {
+        "number": "5.4",
+        "title": "Comparison with Existing Isolation Checkers",
+        "start": 70708
+      },
+      {
+        "number": "6.1",
+        "title": "Limitations",
+        "start": 72576
+      },
+      {
+        "number": "6.2",
+        "title": "Future Work",
+        "start": 74485
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2023. go-randgen. https://github.com/pingcap/go-randgen.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2024. The ANSI Isolation Levels. http://www.adp-gmbh.ch/ora/misc/isolation_level.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2024. CockroachDB. https://www.cockroachlabs.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2024. DB-Engines. https://db-engines.com/en/ranking.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2024. GitHub. https://github.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2024. Jepsen. https://github.com/jepsen-io/jepsen.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2024. MariaDB. https://mariadb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2024. MySQL. https://www.mysql.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2024. MySQL 8.0.34. https://jepsen.io/analyses/mysql-8.0.34?file=mysql-8.0.34.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2024. PostgreSQL. https://www.postgresql.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2024. PostgreSQL 12.3. https://jepsen.io/analyses/postgresql-12.3?file=postgresql-12.3.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2024. SQLancer. https://www.manuelrigger.at/dbms-bugs.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 13,
+      "text": "2024. SQLsmith. https://jepsen.io.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "2024. TiDB 2.1.7. https://jepsen.io/analyses/tidb-2.1.7?file=tidb-2.1.7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "2024. TiDB, PingCAP. https://pingcap.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Atul Adya. 1999. Weak Consistency: A Generalized Theory and Optimistic Implementations for Distributed Transactions. Ph. D. Dissertation. Massachusetts Institute of Technology.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Atul Adya, Barbara Liskov, and Patrick O’Neil. 2000. Generalized Isolation Level Definitions. In Proceedings of International Conference on Data Engineering (ICDE). 67–78.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 19,
+      "text": "Peter Bailis, Aaron Davidson, Alan Fekete, Ali Ghodsi, Joseph M. Hellerstein, and Ion Stoica. 2013. Highly Available Transactions: Virtues and Limitations. Proceedings of the VLDB Endowment (VLDB) 7, 3 (2013), 181–192.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Peter Bailis, Alan Fekete, Ali Ghodsi, Joseph M. Hellerstein, and Ion Stoica. 2016. Scalable Atomic Visibility with RAMP Transactions. ACM Transactions on Database Systems (TODS) 41, 3 (2016), 15:1–15:45.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Hal Berenson, Phil Bernstein, Jim Gray, Jim Melton, Elizabeth O’Neil, and Patrick O’Neil. 1995. A Critique of ANSI SQL Isolation Levels. In Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD). 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Philip A Bernstein, Vassos Hadzilacos, Nathan Goodman, et al .1987. Concurrency Control and Recovery in Database Systems. Addison-Wesley Longman Publishing Co., Inc.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M. Tamer Özsu. 2007. QAGen: Generating Query-Aware Test Databases. InProceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD). 341–352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Ranadeep Biswas and Constantin Enea. 2019. On the Complexity of Checking Transactional Consistency. In Proceedings of ACMSIGPLAN Conference on Object-Oriented Programming Systems, Languages, and Applications (OOPSLA). 165:1– 165:28.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Nicolas Bruno and Surajit Chaudhuri. 2005. Flexible Database Generators. In Proceedings of International Conference on Very Large Data Bases (VLDB). 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. 2006. Generating Queries with Cardinality Constraints for DBMS Testing. IEEE Transactions on Knowledge and Data Engineering (TKDE) 18, 12 (2006), 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Lucas Brutschy, Dimitar Dimitrov, Peter Müller, and Martin Vechev. 2017. Serializability for Eventual Consistency: Criterion, Analysis, and Applications. In Proceedings of ACMSIGPLAN Symposium on Principles of Programming Languages (POPL). 458–472.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Andrea Cerone, Giovanni Bernardi, and Alexey Gotsman. 2015. A Framework for Transactional Consistency Models with Atomic Visibility. In Proceedings of International Conference on Concurrency Theory (CONCUR). 58–71.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Donald D. Chamberlin and Raymond F. Boyce. 1974. SEQUEL: A Structured English Query Language. In Proceedings of ACMSIGFIDET Workshop on Data Description, Access and Control (SIGFIDET). 249–264.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "Edgar F. Codd. 1970. A Relational Model of Data for Large Shared Data Banks. Commun. ACM 13, 6 (1970), 377–387.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Ziyu Cui, Wensheng Dou, Qianwang Dai, Jiansen Song, Wei Wang, Jun Wei, and Dan Ye. 2022. Differentially Testing Database Transactions for Fun and Profit. In Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE). 35:1–35:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Ziyu Cui, Wensheng Dou, Yu Gao, Dong Wang, Jiansen Song, Yingying Zheng, Tao Wang, Rui Yang, Kang Xu, Yixin Hu, Jun Wei, and Tao Huang. 2024. Understanding Transaction Bugs in Database Systems. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). 163:1–163:13. Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA076. Publication date: July 2025. ISSTA076:22 Rui Yan",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Ziyu Cui, Wensheng Dou, Yu Gao, Rui Yang, Yingying Zheng, Jiansen Song, Yuan Feng, and Jun Wei. 2025. Simple Testing Can Expose Most Critical Transaction Bugs: Understanding and Detecting Write-Specific Serializability Violations in Database Systems. Proceedings of the VLDB Endowment (VLDB) (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Wensheng Dou, Ziyu Cui, Qianwang Dai, Jiansen Song, Dong Wang, Yu Gao, Wei Wang, Jun Wei, Lei Chen, Hanmo Wang, Hua Zhong, and Tao Huang. 2023. Detecting Isolation Bugs via Transaction Oracle Construction. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). 1123–1135.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Jim Gray. 1981. The Transaction Concept: Virtues and Limitations. In Proceedings of International Conference on Very Large Data Bases (VLDB). 144–154.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Jim Gray and Andreas Reuter. 1992. Transaction Processing: Concepts and Techniques. Elsevier.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Jim Gray, Prakash Sundaresan, Susanne Englert, Ken Baclawski, and Peter J. Weinberger. 1994. Quickly Generating Billion-Record Synthetic Databases. In Proceedings of ACMSIGMOD International Conference on Management of Data (SIGMOD). 243–252.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Theo Haerder and Andreas Reuter. 1983. Principles of Transaction-Oriented Database Recovery. ACM Computing Surveys (CSUR) 15, 4 (1983), 287–317.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Kenneth Houkjær, Kristian Torp, and Rico Wind. 2006. Simple and Realistic Data Generation. In Proceedings of International Conference on Very Large Data Bases (VLDB). 1243–1246.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting Bugs in Cypher Graph Database Engines. In Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 163–174.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Kaile Huang, Si Liu, Zhenge Chen, Hengfeng Wei, David Basin, Haixiang Li, and Anqun Pan. 2023. Efficient Black-box Checking of Snapshot Isolation in Databases. Proceedings of the VLDB Endowment (VLDB) 16, 6 (2023), 1264–1276.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). Article 46, 12 pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 397–417.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 45,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O Laleye, and Sarfraz Khurshid. 2008. Query-Aware Test Generation Using a Relational Constraint Solver. In Proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE). 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Kyle Kingsbury and Peter Alvaro. 2020. Elle: Inferring Isolation Anomalies from Experimental Observations. In Proceedings of the VLDB Endowment (VLDB), Vol. 14. 268–280.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Philip M Lewis, Arthur Bernstein, and Michael Kifer. 2002. Databases and Transaction Processing: An ApplicationOriented Approach. ACMSIGMOD Record 31, 1 (2002), 74–75.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Haixiang Li, Xiaoyan Li, Chang Liu, Xiaoyong Du, Wei Lu, and Anqun Pan. 2021. Systematic Definition and Classification of Data Anomalies in DBMS. Journal of Software 33, 3 (2021), 909–930.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Keqiang Li, Siyang Weng, Peiyuan Liu, Lyu Ni, Chengcheng Yang, Rong Zhang, Xuan Zhou, Jianghang Lou, Gui Huang, Weining Qian, and Aoying Zhou. 2023. Leopard: A Black-Box Approach for Efficiently Verifying Various Isolation Levels. In Proceedings of IEEE International Conference on Data Engineering (ICDE). 722–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Qiuyang Mang, Aoyang Fang, Boxi Yu, Hanfei Chen, and Pinjia He. 2024. Testing Graph Database Systems via Equivalent Query Rewriting. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). Article 143, 12 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 52,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 211:1–211:30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 53,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 54,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. In Proceedings of International Conference on Very Large Data Bases (VLDB). 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Jiansen Song, Wensheng Dou, Yu Gao, Ziyu Cui, Yingying Zheng, Dong Wang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction. Proceedings of the VLDB Endowment (VLDB) 17, 8 (2024), 1884–1897. Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA076. Publication date: July 2025. Detecting Isolation Anomalies in Relational",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Jiansen Song, Wensheng Dou, Yingying Zheng, Yu Gao, Ziyu Cui, Wei Wang, and Jun Wei. 2025. Detecting SchemaRelated Logic Bugs in Relational DBMSs via Equivalent Database Construction. Proceedings of the VLDB Endowment (VLDB) (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Cheng Tan, Changgeng Zhao, Shuai Mu, and Michael Walfish. 2020. Cobra: Making Transactional Key-Value Stores Verifiably Serializable. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 63–80.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Lei Tang, Wensheng Dou, Yingying Zheng, Lijie Xu, Wei Wang, Jun Wei, and Tao Huang. 2025. Proving Cypher Query Equivalence. In Proceedings of IEEE International Conference on Data Engineering (ICDE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Robert Tarjan. 1972. Depth-First Search and Linear Graph Algorithms. SIAM J. Comput. 1, 2 (1972), 146–160.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Rui Yang, Yingying Zheng, Lei Tang, Wensheng Dou, Wei Wang, and Jun Wei. 2023. Randomized Differential Testing of RDF Stores. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE Demo). 136–140.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Jian Zhang, Ye Ji, Shuai Mu, and Cheng Tan. 2023. Viper: A Fast Snapshot Isolation Checker. In Proceedings of European Conference on Computer Systems (EuroSys). 654–671.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Yingying Zheng, Wensheng Dou, Lei Tang, Ziyu Cui, Yu Gao, Jiansen Song, Liang Xu, Jiaxin Zhu, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2024. Testing Gremlin-Based Graph Database Systems via Query Disassembling. In Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 1695–1707.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Yingying Zheng, Wensheng Dou, Lei Tang, Ziyu Cui, Jiansen Song, Ziyue Cheng, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2024. Differential Optimization Testing of Gremlin-Based Graph Database Systems. In Proceedings of IEEE International Conference on Software Testing, Verification and Validation (ICST). 25–36.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing. In Proceedings of ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of ACMSIGSAC Conference on Computer and Communications Security (CCS). 58–71. Received 2025-02-20; accepted 2025-03-31 Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA076. Publication date: July 2025.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 12,
+      "text": "2024. SQLancer. https://www.manuelrigger.at/dbms-bugs.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 18,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 30,
+      "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 43,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). Article 46, 12 pages.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 44,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 397–417.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 51,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 52,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 211:1–211:30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 53,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 24,27,28,30,42,46,49,58,62]",
+      "technique": null,
+      "sentence": "Existing isolation checkers [ 24,27,28,30,42,46,49,58,62] mainly work on 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models, in which data is organized in a 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 format.",
+      "context_before": "mechanisms can undermine the isolation levels claimed by relational DBMSs, leading to isolation anomalies. Isolation anomalies can lead to serious consequences, e.g., incorrect query results and incorrect database states. In relational DBMSs, isolation anomalies usually involve relational data models and complex SQL operations. A recent study on 140 real-world transaction bugs collected from six relational DBMSs has found that 31.4% of these transaction bugs are caused by isolation anomalies, and 93.2%1of these isolation anomalies involve relational data models or complex SQL operations [33].",
+      "context_after": "These checkers analyze the execution history of transactions based on 𝑟𝑒𝑎𝑑(𝑘𝑒𝑦)and𝑤𝑟𝑖𝑡𝑒(𝑘𝑒𝑦,𝑣𝑎𝑙𝑢𝑒)operations, infer dependencies among transactions based on the read and written 𝑘𝑒𝑦s, and construct a transaction dependency graph to detect isolation anomalies. These checkers are limited to detecting isolation anomalies within𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models and the associated 𝑟𝑒𝑎𝑑(𝑘𝑒𝑦)and𝑤𝑟𝑖𝑡𝑒(𝑘𝑒𝑦,𝑣𝑎𝑙𝑢𝑒)operations. Consequently, existing isolation checkers designed for 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models cannot be easily adapted to relational DBMSs. In 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models, transaction dependencies",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6641,
+      "cited_reference": {
+        "number": 30,
+        "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 24,27,28,30,42,46,49,58,62]",
+      "technique": null,
+      "sentence": "Existing isolation checkers are designed specifically for 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models [ 24,27,28,30,42,46,49,58,62], and are not equipped to handle the complexities of relational data models in relational DBMSs.",
+      "context_before": "s involve complex SQL operations in relational DBMSs. Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA076. Publication date: July 2025. Detecting Isolation Anomalies in Relational DBMSs ISSTA076:3 •Relational data models. Unlike𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 data models that utilize 𝑘𝑒𝑦and𝑣𝑎𝑙𝑢𝑒 to manage data, the data models in relational DBMSs [ 31] manage data in tables and support complex data structures (e.g., multiple tables, multi-column primary keys, and foreign keys), and column constraints (e.g., unique constraints and data constraints). A relational table may contain no keys, compound keys, etc.",
+      "context_after": "•Flexible SQL operations. Relational DBMSs support different kinds of complex SQL operations, e.g., SELECT ,INSERT ,UPDATE ,DELETE, and REPLACE. Unlike the simple 𝑟𝑒𝑎𝑑(𝑘𝑒𝑦)and 𝑤𝑟𝑖𝑡𝑒(𝑘𝑒𝑦,𝑣𝑎𝑙𝑢𝑒)operations, SQL operations can support predicates, subqueries, compound queries, aggregate functions, etc. For example, a SELECT statement “ SELECT name, gender FROM student WHERE age < 20 ” retrieves the name and gender of students who are under 20. An UPDATE statement “ UPDATE student SET age = age+1 WHERE age < 20 ” updates the age by adding 1for students who are under 20. Existing isolation checkers [",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 8911,
+      "cited_reference": {
+        "number": 30,
+        "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 24,27,28,30,42,46,49,58,62]",
+      "technique": null,
+      "sentence": "Existing isolation checkers [ 24,27,28,30,42,46,49,58,62] are limited to supporting simple read and write operations based on 𝑘𝑒𝑦, i.",
+      "context_before": "ta models in relational DBMSs. •Flexible SQL operations. Relational DBMSs support different kinds of complex SQL operations, e.g., SELECT ,INSERT ,UPDATE ,DELETE, and REPLACE. Unlike the simple 𝑟𝑒𝑎𝑑(𝑘𝑒𝑦)and 𝑤𝑟𝑖𝑡𝑒(𝑘𝑒𝑦,𝑣𝑎𝑙𝑢𝑒)operations, SQL operations can support predicates, subqueries, compound queries, aggregate functions, etc. For example, a SELECT statement “ SELECT name, gender FROM student WHERE age < 20 ” retrieves the name and gender of students who are under 20. An UPDATE statement “ UPDATE student SET age = age+1 WHERE age < 20 ” updates the age by adding 1for students who are under 20.",
+      "context_after": "e.,𝑟𝑒𝑎𝑑(𝑘𝑒𝑦)and 𝑤𝑟𝑖𝑡𝑒(𝑘𝑒𝑦,𝑣𝑎𝑙𝑢𝑒). Thus, they cannot support such complex SQL operations in relational DBMSs. In this paper, we propose a novel black-box Isolation checker for Relational DBMSs ,IsoRel. The core idea of IsoRel is to add two auxiliary columns into each table and instrument SQL statements in an isolation-agnostic manner. By analyzing the execution history of the instrumented SQL statements, IsoRel can infer dependencies among transactions that involve relational data models and complex SQL operations. Specifically, we add an auxiliary column 𝑟𝑜𝑤𝐼𝑑 into each table and assign a uniq",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 9693,
+      "cited_reference": {
+        "number": 30,
+        "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 24,27,28,30,42,46,49,58,62]",
+      "technique": null,
+      "sentence": "For these 48 isolation anomalies, existing isolation checkers [ 24,27,28,30,42,46,49,58,62] can only identify 2 anomalies due to their incapability of handling relational data models and complex SQL operations.",
+      "context_before": "idely-used relational DBMSs, i.e., MySQL [ 8], PostgreSQL [ 10], MariaDB [ 7], CockroachDB [ 3], and TiDB [ 15]. We further evaluate IsoRel on all the isolation levels that these relational DBMSs support. In total, IsoRel has revealed 325 isolation anomalies. From these anomalies, we identify 48 unique isolation anomalies. Among the 48 isolation anomalies, 12 anomalies cause lost update, 12 anomalies cause read-write skew, and 24 anomalies cause write skew. These isolation anomalies violate the isolation levels defined by Adya [ 16,17], and have been confirmed by the respective DBMS community.",
+      "context_after": "This result demonstrates that IsoRel can effectively handle the intricacies of relational data models and complex SQL operations. Such advancements would mitigate the risks associated with isolation anomalies and enhance the reliability of relational DBMSs. In summary, we make the following contributions in this paper. 2Note that the instrumentation for DELETE and REPLACE statements may potentially affect concurrent transaction execution. However, based on our analysis of real-world relational DBMSs, we do not find such cases in real-world relational DBMSs. We will further explain this in Sect",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 11985,
+      "cited_reference": {
+        "number": 30,
+        "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 24,27,28,30,42,46,49,58,62]",
+      "technique": null,
+      "sentence": "However, existing isolation checkers [ 24,27,28,30,42,46,49,58,62] are limited to 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models and the associated 𝑟𝑒𝑎𝑑(𝑘𝑒𝑦)and𝑤𝑟𝑖𝑡𝑒(𝑘𝑒𝑦,𝑣𝑎𝑙𝑢𝑒) operations.",
+      "context_before": "infer that 𝑇2 read-write depends on 𝑇1through𝑟1(we userw(𝑟1)to denote read-write dependency through 𝑟1), and𝑇1write-write depends on 𝑇2through𝑟5(we useww(𝑟5)to denote write-write dependency through𝑟5). This causes a read-write and write-write dependency cycle between 𝑇1and𝑇2in the dependency graph that indicates read-write skew. This may cause incorrect database states. It is difficult to infer dependencies among transactions in relational DBMSs. In this example, tables 𝑡1and𝑡2are constructed without a 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 structure, and SQL statements involve complex SQL operations, i.e., range queries.",
+      "context_after": "They cannot identify transaction dependencies in such cases. For example, they cannot identify that 𝑠12retrieves𝑟1from its query result, because the query result of 𝑠12is 1, and both 𝑟1and𝑟2have the value of 1 for 𝑐2. They also cannot identify that 𝑠22updates𝑟1, since𝑡1is not a𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 structure and allows different rows to store the same value (e.g., 𝑟1and𝑟3have the same value of 2for𝑐2). Isolation checkers for 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models cannot determine which rows a transaction has updated by identifying unique values. As a result, they cannot infer that 𝑇2 read-write depends on 𝑇1through",
+      "section": "2.1 Motivating Example",
+      "page": 4,
+      "char_offset": 15300,
+      "cited_reference": {
+        "number": 30,
+        "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 16,17,30,42,\n46,49,58,62]",
+      "technique": null,
+      "sentence": "Existing works [ 16,17,30,42, 46,49,58,62] have proved that the appearance of different kinds of cycles in the dependency graph 𝐺indicates different kinds of isolation anomalies, e.",
+      "context_before": "𝑚𝑡𝑊in𝑇𝑗writes the version of row 𝑟with𝑤𝐿𝑖𝑠𝑡𝑟=[...,𝑇𝑘,𝑇𝑗]. No other statements can update row 𝑟between𝑆𝑡𝑚𝑡𝑅 and𝑆𝑡𝑚𝑡𝑊. Therefore, 𝑇𝑗read-write depends on 𝑇𝑖through row 𝑟. The above three lemmas prove that the transaction dependencies inferred by IsoRel are sound, and have no false positives. Therefore, the dependency graph 𝐺=(𝐶𝑇,𝑆𝑂,𝑊𝑅,𝑊𝑊,𝑅𝑊 ), in which 𝑆𝑂represents the session order on concurrent transactions 𝐶𝑇, and𝑊𝑅,𝑊𝑊 and𝑅𝑊represent the set of write-read dependencies, write-write dependencies and read-write dependencies among transactions 𝐶𝑇, is sound and does not contain wrong dependencies.",
+      "context_after": "g., a cycle caused by read-write and write-write dependencies through the same row in the dependency graph 𝐺indicates a lost update. In IsoRel, we adopt the Tarjan algorithm [ 60] to identify cycles in the dependency graph, which is sound and does not report false positives. Therefore, IsoRel is sound. 5 Evaluation We evaluate IsoRel on five widely-used relational DBMSs, i.e., MySQL [ 8], PostgreSQL [ 10], MariaDB [ 7], CockroachDB [ 3], and TiDB [ 15], and answer the following three research questions: •RQ1: How effective is IsoRel in detecting isolation anomalies in relational DBMSs? •RQ2: W",
+      "section": "3.4 Isolation Anomaly Detection",
+      "page": 15,
+      "char_offset": 58396,
+      "cited_reference": {
+        "number": 30,
+        "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 24,27,28,30,42,46,49,58,62]",
+      "technique": null,
+      "sentence": "4 Comparison with Existing Isolation Checkers Existing isolation checkers [ 24,27,28,30,42,46,49,58,62] mainly work on 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models, in which data is organized in a 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 format.",
+      "context_before": "this case, two transactions 𝑇1 and𝑇2write different rows in the same table. If the relational DBMS does not apply gap locks (e.g., TiDB does not apply gap locks), lock conflicts do not occur between 𝑇1and𝑇2. Figure 6 shows a write skew detected in MySQL under Repeatable Read. In this case, 𝑇1and 𝑇3are in the same session. 𝑇2rwdepends on 𝑇1through row 𝑟1, and𝑇1rwdepends on 𝑇2through row𝑟3. Thus, this isolation anomaly satisfies write skew defined by Berenson [ 21]. However, it is allowed to occur in MySQL, since row 𝑟3and𝑟1are in different tables, and no lock conflict occurs between𝑇1and𝑇2. 5.",
+      "context_after": "These checkers are limited to detecting isolation anomalies only within simple 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models and the associated 𝑟𝑒𝑎𝑑(𝑘𝑒𝑦)and 𝑤𝑟𝑖𝑡𝑒(𝑘𝑒𝑦,𝑣𝑎𝑙𝑢𝑒)operations. They cannot infer wr,wwandrwdependencies among transactions that involve complex SQL operations and relational data models in relational DBMSs, which IsoRel can infer. Therefore, we cannot directly compare IsoRel with existing checkers when these checkers are applied to relational DBMSs. Thus, we analyze whether three existing checkers (Elle [ 46], Cobra [58], PolySI [42]) can conceptually detect the 48 isolation anomalies revea",
+      "section": "5.4 Comparison with Existing Isolation Checkers",
+      "page": 18,
+      "char_offset": 70710,
+      "cited_reference": {
+        "number": 30,
+        "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 24,27,28,30,42,46,49,58,62]",
+      "technique": null,
+      "sentence": "Existing isolation anomaly detection algorithms [ 24,27,28,30,42,46,49,58,62] also aim to detect cycles in the dependency graph, among which there are some efficient anomaly detection algorithms, e.",
+      "context_before": "isolation anomalies detected in one relational DBMS. These templates can potentially lead to anomalies in other relational DBMSs. Thus, we can use these templates to speed up isolation anomaly detection in other relational DBMSs, and help converge more quickly. Extending IsoRel by adopting existing anomaly detection algorithms. In our work, building the dependency graph from the transaction execution history in relational DBMSs is our main contribution. We further adopt a simple algorithm (i.e., graph traversal by utilizing the Tarjan algorithm [ 60]) to detect cycles in the dependency graph.",
+      "context_after": "g., resolving uncertain dependencies through SMT solvers [ 42,58,62]. Therefore, once we build the dependency graph in relational DBMSs through our SQL statement instrumentation and dependency inference approach, we can potentially adopt existing anomaly detection algorithms (e.g., Cobra [ 58] and PolySI [ 42]) to detect isolation anomalies in relational DBMSs to improve detection efficiency. 7 Related Work Isolation level specification. ANSI [ 2] defines four isolation levels, i.e., Read Uncommitted ,Read Committed ,Repeatable Read, andSerializable, and the corresponding four isolation anomal",
+      "section": "6.2 Future Work",
+      "page": 19,
+      "char_offset": 75685,
+      "cited_reference": {
+        "number": 30,
+        "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[30]",
+      "technique": null,
+      "sentence": "Emme [30] is a white-box checker that retrieves transaction timestamps and recovers a version certificate to check isolation levels in DBMSs that use multi-version concurrency control timestamp ordering protocol to guarantee serializability.",
+      "context_before": "A076. Publication date: July 2025. ISSTA076:20 Rui Yang, Ziyu Cui, Wensheng Dou, Yu Gao, Jiansen Song, Xudong Xie, and Jun Wei models. Cobra [ 58] utilizes polygraph, which utilizes constraints to capture unknown dependencies, and uses SMT solvers to detect cycles. PolySI [ 42] proposes a generalized polygraph and the formal definition of Snapshot Isolation. It accelerates the checking of Snapshot Isolation in𝑘𝑒𝑦𝑣𝑎𝑙𝑢𝑒 -like data models through SMT solvers. Viper [ 62] proposes a BC-polygraph, and accelerates the checking of Snapshot Isolation in𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models through SMT solvers.",
+      "context_after": "However, these isolation checkers focus on 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models, and can not be applied to relational data models. Transaction testing. To improve the quality of transaction implementation, some transaction testing approaches have been proposed. DT2[32] executes the same group of transactions in multiple DBMSs and compares their execution results to detect transaction bugs. Troc [ 35] builds a transaction test oracle to determine whether a group of transactions executes correctly. TxCheck [44] constructs semantically equivalent test cases and compares their execution results to detect",
+      "section": "6.2 Future Work",
+      "page": 20,
+      "char_offset": 77910,
+      "cited_reference": {
+        "number": 30,
+        "text": "Jack Clark, Alastair F Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. In Proceedings of European Conference on Computer Systems (EuroSys). 754–768.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[44]",
+      "technique": null,
+      "sentence": "TxCheck [44] constructs semantically equivalent test cases and compares their execution results to detect transaction bugs.",
+      "context_before": "DBMSs that use multi-version concurrency control timestamp ordering protocol to guarantee serializability. However, these isolation checkers focus on 𝑘𝑒𝑦-𝑣𝑎𝑙𝑢𝑒 -like data models, and can not be applied to relational data models. Transaction testing. To improve the quality of transaction implementation, some transaction testing approaches have been proposed. DT2[32] executes the same group of transactions in multiple DBMSs and compares their execution results to detect transaction bugs. Troc [ 35] builds a transaction test oracle to determine whether a group of transactions executes correctly.",
+      "context_after": "WriteCheck [ 34] proposes write-specific serializability, and uses this property to detect write-specific serializability violations. However, these works cannot detect isolation anomalies under non-deterministic executions in relational DBMSs. DBMS testing. Many approaches have been proposed to test DBMSs. SQLsmith [ 13] detects crash bugs by executing randomly generated SQL statements. Rigger et al. propose SQLancer [ 12] and several approaches, e.g., PQS [ 53], TLP [ 52] and NoREC [ 51], to detect logic bugs in SELECT statements. QPG [ 18] utilizes query plans to guide DBMS testing. DQE [",
+      "section": "6.2 Future Work",
+      "page": 20,
+      "char_offset": 78645,
+      "cited_reference": {
+        "number": 44,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In Proceedings of USENIX Symposium on Operating Systems Design and Implementation (OSDI). 397–417.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "propose SQLancer [ 12] and several approaches, e.",
+      "context_before": "oracle to determine whether a group of transactions executes correctly. TxCheck [44] constructs semantically equivalent test cases and compares their execution results to detect transaction bugs. WriteCheck [ 34] proposes write-specific serializability, and uses this property to detect write-specific serializability violations. However, these works cannot detect isolation anomalies under non-deterministic executions in relational DBMSs. DBMS testing. Many approaches have been proposed to test DBMSs. SQLsmith [ 13] detects crash bugs by executing randomly generated SQL statements. Rigger et al.",
+      "context_after": "g., PQS [ 53], TLP [ 52] and NoREC [ 51], to detect logic bugs in SELECT statements. QPG [ 18] utilizes query plans to guide DBMS testing. DQE [ 55] executes SELECT ,UPDATE and DELETE statements with the same predicate to detect logic bugs. Radar [ 56] differentially tests the original database and raw database that wipes out the metadata to detect metadata-related logic bugs. DDLCheck [ 57] examines the behavior differences on the same database schema created by different DDL sequences to detect schema-related logic bugs. Grand [ 65], GDsmith [ 41] and RD2[61] adopt differential testing to det",
+      "section": "6.2 Future Work",
+      "page": 20,
+      "char_offset": 79174,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", PQS [ 53], TLP [ 52] and NoREC [ 51], to detect logic bugs in SELECT statements.",
+      "context_before": "executes correctly. TxCheck [44] constructs semantically equivalent test cases and compares their execution results to detect transaction bugs. WriteCheck [ 34] proposes write-specific serializability, and uses this property to detect write-specific serializability violations. However, these works cannot detect isolation anomalies under non-deterministic executions in relational DBMSs. DBMS testing. Many approaches have been proposed to test DBMSs. SQLsmith [ 13] detects crash bugs by executing randomly generated SQL statements. Rigger et al. propose SQLancer [ 12] and several approaches, e.g.",
+      "context_after": "QPG [ 18] utilizes query plans to guide DBMS testing. DQE [ 55] executes SELECT ,UPDATE and DELETE statements with the same predicate to detect logic bugs. Radar [ 56] differentially tests the original database and raw database that wipes out the metadata to detect metadata-related logic bugs. DDLCheck [ 57] examines the behavior differences on the same database schema created by different DDL sequences to detect schema-related logic bugs. Grand [ 65], GDsmith [ 41] and RD2[61] adopt differential testing to detect logic bugs in graph DBMSs. Qudi [ 63], GQT [ 43] and GRev [ 50] further utilize",
+      "section": "6.2 Future Work",
+      "page": 20,
+      "char_offset": 79226,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG [ 18] utilizes query plans to guide DBMS testing.",
+      "context_before": "compares their execution results to detect transaction bugs. WriteCheck [ 34] proposes write-specific serializability, and uses this property to detect write-specific serializability violations. However, these works cannot detect isolation anomalies under non-deterministic executions in relational DBMSs. DBMS testing. Many approaches have been proposed to test DBMSs. SQLsmith [ 13] detects crash bugs by executing randomly generated SQL statements. Rigger et al. propose SQLancer [ 12] and several approaches, e.g., PQS [ 53], TLP [ 52] and NoREC [ 51], to detect logic bugs in SELECT statements.",
+      "context_after": "DQE [ 55] executes SELECT ,UPDATE and DELETE statements with the same predicate to detect logic bugs. Radar [ 56] differentially tests the original database and raw database that wipes out the metadata to detect metadata-related logic bugs. DDLCheck [ 57] examines the behavior differences on the same database schema created by different DDL sequences to detect schema-related logic bugs. Grand [ 65], GDsmith [ 41] and RD2[61] adopt differential testing to detect logic bugs in graph DBMSs. Qudi [ 63], GQT [ 43] and GRev [ 50] further utilize equivalent queries [ 59] to detect logic bugs in graph",
+      "section": "6.2 Future Work",
+      "page": 20,
+      "char_offset": 79308,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 43]",
+      "technique": null,
+      "sentence": "Qudi [ 63], GQT [ 43] and GRev [ 50] further utilize equivalent queries [ 59] to detect logic bugs in graph DBMSs.",
+      "context_before": "oREC [ 51], to detect logic bugs in SELECT statements. QPG [ 18] utilizes query plans to guide DBMS testing. DQE [ 55] executes SELECT ,UPDATE and DELETE statements with the same predicate to detect logic bugs. Radar [ 56] differentially tests the original database and raw database that wipes out the metadata to detect metadata-related logic bugs. DDLCheck [ 57] examines the behavior differences on the same database schema created by different DDL sequences to detect schema-related logic bugs. Grand [ 65], GDsmith [ 41] and RD2[61] adopt differential testing to detect logic bugs in graph DBMSs.",
+      "context_after": "DOT [ 64] further detects optimization bugs in graph DBMSs. However, these works cannot detect isolation anomalies in relational DBMSs. 8 Conclusion Incorrect implementations of transaction mechanisms can cause isolation anomalies and undermine the isolation levels claimed by relational DBMSs. In this paper, we propose IsoRel, a black-box isolation checker for relational DBMSs, which can support relational data models and complex SQL operations. The core idea of IsoRel is to instrument SQL statements in an isolation-agnostic manner and extract the accessed rows of each SQL statement to track",
+      "section": "6.2 Future Work",
+      "page": 20,
+      "char_offset": 79854,
+      "cited_reference": {
+        "number": 43,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of IEEE/ACM International Conference on Software Engineering (ICSE). A",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:58:40Z",
+    "is_model_written": true,
+    "summary": "This work checks isolation levels in relational DBMSs directly, rather than through the key-value abstraction existing isolation checkers assume. Because those checkers model only read(key) and write(key,value), they cannot represent the SQL operations real workloads use: of 48 isolation anomalies the authors collect, existing checkers identify 2. The approach builds dependency graphs over relational operations and detects the cycles that indicate each anomaly class.",
+    "narrative": "SQLancer appears only in the future-work discussion, as background on what else DBMS testing covers: the paper notes that SQLancer and its approaches -- PQS, TLP and NoREC -- detect logic bugs in SELECT statements, that QPG uses query plans to guide testing, and that QuDi, GQT and GRev carry equivalent-query ideas to graph systems. That is a survey of the adjacent literature rather than a relationship: the paper's own comparison is against isolation checkers, and the anomaly work it builds on is Emme and TxCheck.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "background",
+      "M6": "background",
+      "M7": "background",
+      "M8": "background",
+      "M9": "definition",
+      "M10": "definition",
+      "M11": "definition",
+      "M12": "definition",
+      "M13": "definition",
+      "M14": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No reuse is described; the paper implements dependency-graph construction over relational operations."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Isolation-anomaly detection through dependency-graph cycles shares no mechanism with the SELECT-oriented oracles the paper cites."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M7 states the comparison is with existing isolation checkers. SQLancer's oracles are mentioned only in the future-work survey, with no run reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The mentions describe what each technique does; no standing is claimed for any of them."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3728965.json b/_data/papers/paper_doi_10_1145_3728965.json
new file mode 100644
index 0000000..6ed611f
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3728965.json
@@ -0,0 +1,781 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:24:27Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3728965",
+    "title": "DepState: Detecting Synchronization Failure Bugs in Distributed Database Management Systems",
+    "authors": [
+      "Cundi Fang",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Zhouyang Jia",
+      "Chun Huang",
+      "Yu Jiang",
+      "Shanshan Li"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3728965",
+    "arxiv_id": null,
+    "s2_paper_id": "810786dd896fe2eef0a129db6e3b115c95375130",
+    "url": "https://doi.org/10.1145/3728965",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3728965",
+      "retrieved_at": "2026-09-08T15:24:27Z",
+      "chars": 84249,
+      "content_sha256": "sha256:811218796936009b291d6e145761cf6cb98569e9fa867f83744569faba7a09d0"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 22,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3893
+      },
+      {
+        "number": "2",
+        "title": "Background and Motivation",
+        "start": 11041
+      },
+      {
+        "number": "2.1",
+        "title": "DDBMS Overview",
+        "start": 11069
+      },
+      {
+        "number": "2.2",
+        "title": "Synchronization Process in DDBMS",
+        "start": 12945
+      },
+      {
+        "number": "2.3",
+        "title": "An Example of Synchronization Failure Bug",
+        "start": 15915
+      },
+      {
+        "number": "3",
+        "title": "Design of DepState",
+        "start": 20041
+      },
+      {
+        "number": "3.1",
+        "title": "Sharding-Aware Dependent Data Generation",
+        "start": 21000
+      },
+      {
+        "number": "3.2",
+        "title": "Synchronization Sensitive Cluster Operation Sequence Exploration",
+        "start": 30680
+      },
+      {
+        "number": "4",
+        "title": "Implementation",
+        "start": 45333
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 47842
+      },
+      {
+        "number": "5.1",
+        "title": "Evaluation Setup",
+        "start": 48252
+      },
+      {
+        "number": "5.2",
+        "title": "Synchronization Failure Bug Detection",
+        "start": 53272
+      },
+      {
+        "number": "5.3",
+        "title": "Comparison with Other Techniques",
+        "start": 58538
+      },
+      {
+        "number": "5.4",
+        "title": "Effectiveness of Each Component",
+        "start": 63341
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 65219
+      },
+      {
+        "number": "6.1",
+        "title": "Generality and Flexibility of DepState",
+        "start": 65232
+      },
+      {
+        "number": "6.2",
+        "title": "Testing Non-relational DDBMSs",
+        "start": 66809
+      },
+      {
+        "number": "6.3",
+        "title": "Testing Other Distributed Systems",
+        "start": 68439
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 69490
+      },
+      {
+        "number": "7.1",
+        "title": "Distributed System Testing",
+        "start": 69505
+      },
+      {
+        "number": "7.2",
+        "title": "DBMS Testing",
+        "start": 71559
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 73370
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Raoof Altaher. 2024. Transparency Levels in Distributed Database Management System DDBMS. Preprints (April 2024). doi:10.20944/preprints202404.1327.v1",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Anse1. 2015. sqlsmith – A Random SQL Query Generator. https://github.com/anse1/sqlsmith A random SQL query generator.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Shannon Bradshaw, Eoin Brazil, and Kristina Chodorow. 2019. MongoDB: The Definitive Guide: Powerful and Scalable Data Storage. O’Reilly Media, Sebastopol, CA, USA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Michael Alan Chang, Bredan Tschaen, Theophilus Benson, and Laurent Vanbever. 2015. Chaos Monkey: Increasing SDN Reliability through Systematic Network Destruction. In Proceedings of the 2015 ACM Conference on Special Interest Group on Data Communication (London, United Kingdom) (SIGCOMM ’15). Association for Computing Machinery, New York, NY, USA, 371–372. doi:10.1145/2785956.2790038",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "MariaDB Community. 2024. MariaDB Galera Cluster. https://mariadb.com/kb/en/galera-cluster/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "James C. Corbett, Jeffrey Dean, Michael Epstein, Andrew Fikes, Christopher Frost, J. J. Furman, Sanjay Ghemawat, Andrey Gubarev, Christopher Heiser, Peter Hochschild, Wilson Hsieh, Sebastian Kanthak, Eugene Kogan, Hongyi Li, Alexander Lloyd, Sergey Melnik, David Mwaura, David Nagle, Sean Quinlan, Rajesh Rao, Lindsay Rolig, Yasushi Saito, Michal Szymaniak, Christopher Taylor, Ruth Wang, and Dale Wo",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Stephen Elliot. 2014. DevOps and the Cost of Downtime: Fortune 1000 Best Practice Metrics Quantified. International Data Corporation (IDC) (2014).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Paul Ezhilchelvan, Amjad Aldweesh, and Aad van Moorsel. 2018. Non-Blocking Two Phase Commit Using Blockchain. InProceedings of the 1st Workshop on Cryptocurrencies and Blockchains for Distributed Systems (Munich, Germany) (CryBlock’18). Association for Computing Machinery, New York, NY, USA, 36–41. doi:10.1145/3211933.3211940",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2023. Griffin: Grammar-Free DBMS Fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering (Rochester, MI, USA) (ASE ’22). Association for Computing Machinery, New York, NY, USA, Article 49, 12 pages. doi:10.1145/3551349.3560431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Ying Fu, Zhiyong Wu, Yuanliang Zhang, Jie Liang, Jingzhou Fu, Yu Jiang, Shanshan Li, and Xiangke Liao. 2024. THANOS: DBMS Bug Detection via Storage Engine Rotation Based Differential Testing. In Proceedings of the. IEEE Computer Society, 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jim Gray and Leslie Lamport. 2006. Consensus on Transaction Commit. ACM Transactions on Database Systems (TODS) 31, 1 (2006), 133–160. doi:10.1145/1188913.1188915",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Theo Haerder and Andreas Reuter. 1983. Principles of transaction-oriented database recovery. ACM Comput. Surv. 15, 4 (Dec. 1983), 287–317. doi:10.1145/289.291",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Haochen He, Erci Xu, Shanshan Li, Zhouyang Jia, Si Zheng, Yue Yu, Jun Ma, and Xiangke Liao. 2023. When Database Meets New Storage Devices: Understanding and Exposing Performance Mismatches via Configurations. Proc. VLDB Endow. 16, 7 (March 2023), 1712–1725. doi:10.14778/3587136.3587145",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Victor Heorhiadi, Shriram Rajagopalan, Hani Jamjoom, Michael K. Reiter, and Vyas Sekar. 2016. Gremlin: Systematic Resilience Testing of Microservices. In (ICDCS). 57–66. doi:10.1109/ICDCS.2016.11",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Shawn Hogan, Frazer Clement, and Jon Stephens. 2020. Bug #98526 Forced Shutdown at End of Process When Data Node Tries to Join Cluster. https://bugs.mysql.com/bug.php?id=98526",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Luke Jones. 2020. Google Explains Monday’s Google Cloud Outage That Took Down Gmail and YouTube. https://winbuzzer. com/2020/12/15/google-explains-mondays-google-cloud-outage-that-took-down-gmail-and-youtube-xcxwbn/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Kyle Kingsbury. 2023. Jepsen. https://jepsen.io/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Kyle Kingsbury and Peter Alvaro. 2020. Elle: Inferring Isolation Anomalies from Experimental Observations. arXiv preprint (2020). doi:10.48550/arXiv.2003.10554 arXiv:arXiv:2003.10554",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Jonathan Kirsch and Yair Amir. 2008. Paxos for System Builders: an overview. In Proceedings of the 2nd Workshop on Large-Scale Distributed Systems and Middleware (Yorktown Heights, New York, USA) (LADIS ’08). Association for Computing Machinery, New York, NY, USA, Article 3, 6 pages. doi:10.1145/1529974.1529979",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Leslie Lamport. 2019. The part-time parliament. Association for Computing Machinery, New York, NY, USA, 277–317. https://doi.org/10.1145/3335772.3335939",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Sami M. Lasassmeh and James M. Conrad. 2010. Time synchronization in wireless sensor networks: A survey. In Proceedings of the IEEE SoutheastCon 2010 (SoutheastCon). 242–245. doi:10.1109/SECON.2010.5453878",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-Oriented DBMS Fuzzing. In Engineering (ICDE). 668–681. doi:10.1109/ICDE55515.2023.00057",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation. In Proceedings of the IEEE/ACM 46th International Conference Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA088. Publication date: July 2025. ISSTA088:22 Cundi Fang, Jie Liang, Zhiyong Wu, Jingzhou Fu, Zhouyang Jia, Chun Huang, ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Tao Lyu, Liyi Zhang, Zhiyao Feng, Yueyang Pan, Yujie Ren, Meng Xu, Mathias Payer, and Sanidhya Kashyap. 2024. Monarch: A Fuzzing Framework for Distributed File Systems. In 2024 USENIX Annual Technical Conference (USENIX ATC 24). USENIX Association, Santa Clara, CA, 529–543. https://www.usenix.org/conference/atc24/presentation/lyu",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Fuchen Ma, Yuanliang Chen, Yuanhang Zhou, Jingxuan Sun, Zhuo Su, Yu Jiang, Jiaguang Sun, and Huizhong Li. 2023. Phoenix: Detect and Locate Resilience Issues in Blockchain via Context-Sensitive Chaos. In Proceedings of the 2023 ACMSIGSAC Conference on Computer and Communications Security (Copenhagen, Denmark) (CCS ’23). Association for Computing Machinery, New York, NY, USA, 1182–1196. doi:10.1145/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Ruijie Meng, George Pîrlea, Abhik Roychoudhury, and Ilya Sergey. 2023. Greybox Fuzzing of Distributed Systems (CCS ’23). Association for Computing Machinery, New York, NY, USA, 1615–1629. doi:10.1145/3576915.3623097",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Diego Ongaro and John Ousterhout. 2014. In search of an understandable consensus algorithm. In Proceedings of the 2014 USENIX Conference on USENIX Annual Technical Conference (Philadelphia, PA) (USENIX ATC’14). USENIX Association, USA, 305–320.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Oracle. 2024. MySQL InnoDB Cluster Documentation. https://dev.mysql.com/doc/refman/8.0/en/mysql-cluster.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Oracle. 2024. MySQL NDB Cluster Documentation. https://dev.mysql.com/doc/refman/8.0/en/mysql-cluster.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "M. Tamer Özsu and Patrick Valduriez. 1999. Principles of Distributed Database Systems. Springer, Berlin, Heidelberg, Germany.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "PingCAP. 2024. TiDB: An Open Source Distributed SQL Database. https://pingcap.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA) (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/33",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). USENIX Association, USA, Article 38, 16 pages. doi:10.5555/3488766.3488804",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 34,
+      "text": "Mark Roantree. 2002. Metadata Management in Federated Multimedia Systems. In Australasian Database Conference. 147–155.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Amanpreet Kaur Sandhu. 2022. Big data with cloud computing: Discussions and challenges. Big Data Mining and Analytics 5, 1 (2022), 32–40. doi:10.26599/BDMA.2021.9020016",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Fred B. Schneider. 1982. Synchronization in Distributed Programs. ACM Trans. Program. Lang. Syst. 4, 2 (April 1982), 125–148. doi:10.1145/357162.357163",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Redgate Software. 2024. DB-Engines Ranking. https://db-engines.com/en/ranking",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Siamak Solat. 2024. Sharding Distributed Databases: A Critical Review. arXiv preprint (2024). doi:10.48550/arXiv.2404. 04384 arXiv:arXiv:2404.04384",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Jason Warner. 2018. October 21 Post-Incident Analysis. https://github.blog/news-insights/company-news/oct21-postincident-analysis/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "M. Wiesmann, F. Pedone, A. Schiper, B. Kemme, and G. Alonso. 2000. Understanding replication in databases and distributed systems. In Proceedings 20th IEEE International Conference on Distributed Computing Systems. 464–474. doi:10.1109/ICDCS.2000.840959",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security (Virtual Event, USA) (CCS ’20). Association for Computing Machinery, New York, NY, USA, 955–970. doi:10.1145/3372297.3417260 Received 202",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 32,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA) (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). USENIX Association, USA, Article 38, 16 pages. doi:10.5555/3488766.3488804",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "28% more lines in synchronization-related functions than Jepsen, Mallory, SQLsmith, SQLancer, and Mozi in 24 hours, respectively.",
+      "context_before": ". DepState enhances synchronization testing by simulating the complexities of data sharding and dynamic cluster conditions. It establishes dependencies between tables across nodes and systematically introduces controlled variations in cluster states. We utilize DepState on four DDBMSs: MySQL NDB Cluster, MySQL InnoDB Cluster, MariaDB Galera Cluster, and TiDB Cluster, discovering 25 new bugs, with 13 confirmed. We compare DepState against state-of-the-art tools. DepState finds 14 more synchronization failure bugs and covers 6.13%-66.51%, 5.82%57.28%, 14.12%-83.30%, 36.81%-83.88%, and 43.24%-54.",
+      "context_after": "CCS Concepts: •Software and its engineering →Software testing and debugging ;•Information systems→Relational parallel and distributed DBMSs; Distributed data locking. Additional Key Words and Phrases: Distributed Database Management Systems; Synchronization Failure ACM Reference Format: Cundi Fang, Jie Liang, Zhiyong Wu, Jingzhou Fu, Zhouyang Jia, Chun Huang, Yu Jiang, and Shanshan Li. 2025. DepState: Detecting Synchronization Failure Bugs in Distributed Database Management Systems. Proc. ACM Softw. Eng. 2, ISSTA, Article ISSTA088 (July 2025), 22 pages. https://doi.org/10.1145/3728965 ∗Cundi",
+      "section": null,
+      "page": 1,
+      "char_offset": 1779,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In addition, we compare DepState with the state-of-the-art distributed system testing tool Jepsen and Mallory, and advanced database testing tools SQLsmith, SQLancer, and Mozi.",
+      "context_before": "r, we propose synchronization-sensitive cluster state sequence exploration to precisely modify cluster states during different synchronization phases. This methodology investigates novel test sequences that emerge at distinct synchronization intervals via mutation, leveraging coverage feedback to efficiently identify potential synchronization issues. We implement DepState and evaluate it on four widely-used DDBMSs: MySQL NDB Cluster, MySQL InnoDB Cluster, MariaDB Galera Cluster, and TiDB Cluster. DepState reports a total of 25synchronization failure bugs, with 13 anomalies have been confirmed.",
+      "context_after": "The 24-hour result shows that DepState found 14 more synchronization failure bugs, and covered 6.13%-66.51%, 5.82%-57.28%, 14.12%-83.30%, 36.81%-83.88%, and 43.24%-54.28% more lines in synchronization-related functions than Jepsen, Mallory, SQLsmith, SQLancer, and Mozi, respectively. In summary, we make the following contributions: •We identified that synchronization process in DDBMSs can be faulty and lead to serious issues such as service interruptions, but state-of-the-art testing techniques pay little attention to the synchronization failure bugs. •We designed and implemented a fuzzing fr",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 9871,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "28% more lines in synchronization-related functions than Jepsen, Mallory, SQLsmith, SQLancer, and Mozi, respectively.",
+      "context_before": "issues. We implement DepState and evaluate it on four widely-used DDBMSs: MySQL NDB Cluster, MySQL InnoDB Cluster, MariaDB Galera Cluster, and TiDB Cluster. DepState reports a total of 25synchronization failure bugs, with 13 anomalies have been confirmed. In addition, we compare DepState with the state-of-the-art distributed system testing tool Jepsen and Mallory, and advanced database testing tools SQLsmith, SQLancer, and Mozi. The 24-hour result shows that DepState found 14 more synchronization failure bugs, and covered 6.13%-66.51%, 5.82%-57.28%, 14.12%-83.30%, 36.81%-83.88%, and 43.24%-54.",
+      "context_after": "In summary, we make the following contributions: •We identified that synchronization process in DDBMSs can be faulty and lead to serious issues such as service interruptions, but state-of-the-art testing techniques pay little attention to the synchronization failure bugs. •We designed and implemented a fuzzing framework DepState, which combines dependencyaware sharding data generation and synchronization-sensitive cluster state sequence exploration to detect synchronization failure bugs. •We demonstrated that DepState improves synchronization code coverage compared to existing state-of-the-ar",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 10216,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To evaluate the effectiveness of DepState in testing DDBMSs, we compare it with five state-of-the-art tools: Jepsen [ 17],Mallory [26], SQLsmith [ 2], SQLancer [ 33], andMozi [23].",
+      "context_before": "te, we select four popular open-source DDBMSs for evaluation, namely MySQL NDB Cluster [ 29], MySQL InnoDB Cluster [ 28], MariaDB Galera Cluster [ 5], and TiDB Cluster [ 31]. They are all widely used opensource DDBMSs according to DB-Engine Ranking [ 37]. They support similar DBMS features (e.g., the relational data model) but use different distributed architecture implementations. The versions under evaluation are MySQL NDB Cluster v8.0.40, MySQL InnoDB Cluster v8.0.35, MariaDB Galera Cluster v10.6.20, and TiDB Cluster 8.0.11-TiDB-v9.0.0-alpha-100-g91706ec8df-dirty. 5.1.2 Compared Techniques.",
+      "context_after": "Jepsen [ 17] is a distributed system testing tool, which has been adapted to test DDBMSs in the industry. Mallory [26] is an enhancement built upon the foundation of Jepsen. SQLsmith, SQLancer, and Mozi are three state-of-the-art DBMS testing techniques, which have detected hundreds of bugs in practice. We adapted Jepsen’s default configuration to make it work on these DDBMSs, and Mallory was similarly configured. For SQLsmith, SQLancer, and Mozi, since they were designed for testing traditional DBMS, we used them on a single SQL server of the DDBMS as a baseline. We selected traditional data",
+      "section": "5.1 Evaluation Setup",
+      "page": 14,
+      "char_offset": 48945,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLsmith, SQLancer, and Mozi are three state-of-the-art DBMS testing techniques, which have detected hundreds of bugs in practice.",
+      "context_before": "architecture implementations. The versions under evaluation are MySQL NDB Cluster v8.0.40, MySQL InnoDB Cluster v8.0.35, MariaDB Galera Cluster v10.6.20, and TiDB Cluster 8.0.11-TiDB-v9.0.0-alpha-100-g91706ec8df-dirty. 5.1.2 Compared Techniques. To evaluate the effectiveness of DepState in testing DDBMSs, we compare it with five state-of-the-art tools: Jepsen [ 17],Mallory [26], SQLsmith [ 2], SQLancer [ 33], andMozi [23]. Jepsen [ 17] is a distributed system testing tool, which has been adapted to test DDBMSs in the industry. Mallory [26] is an enhancement built upon the foundation of Jepsen.",
+      "context_after": "We adapted Jepsen’s default configuration to make it work on these DDBMSs, and Mallory was similarly configured. For SQLsmith, SQLancer, and Mozi, since they were designed for testing traditional DBMS, we used them on a single SQL server of the DDBMS as a baseline. We selected traditional database testing tools, including SQLsmith, SQLancer, and Mozi, as benchmarks. By using them as baselines, we aim to demonstrate that traditional database testing tools are not directly applicable to distributed databases. 5.1.3 Basic Setup. All experiments were conducted on a 64-bit Ubuntu 22.04 machine wit",
+      "section": "5.1 Evaluation Setup",
+      "page": 14,
+      "char_offset": 49300,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For SQLsmith, SQLancer, and Mozi, since they were designed for testing traditional DBMS, we used them on a single SQL server of the DDBMS as a baseline.",
+      "context_before": ". To evaluate the effectiveness of DepState in testing DDBMSs, we compare it with five state-of-the-art tools: Jepsen [ 17],Mallory [26], SQLsmith [ 2], SQLancer [ 33], andMozi [23]. Jepsen [ 17] is a distributed system testing tool, which has been adapted to test DDBMSs in the industry. Mallory [26] is an enhancement built upon the foundation of Jepsen. SQLsmith, SQLancer, and Mozi are three state-of-the-art DBMS testing techniques, which have detected hundreds of bugs in practice. We adapted Jepsen’s default configuration to make it work on these DDBMSs, and Mallory was similarly configured.",
+      "context_after": "We selected traditional database testing tools, including SQLsmith, SQLancer, and Mozi, as benchmarks. By using them as baselines, we aim to demonstrate that traditional database testing tools are not directly applicable to distributed databases. 5.1.3 Basic Setup. All experiments were conducted on a 64-bit Ubuntu 22.04 machine with an AMDEPYC 7742 processor (128 cores @ 2.25 GHz) and 488 GiB of main memory. Each testing tool used its default configuration to ensure consistency throughout all evaluations. The DDBMS clusters employed a single setup with 2 replicas and 2 slices, without multi-c",
+      "section": "5.1 Evaluation Setup",
+      "page": 14,
+      "char_offset": 49544,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We selected traditional database testing tools, including SQLsmith, SQLancer, and Mozi, as benchmarks.",
+      "context_before": "SQLancer [ 33], andMozi [23]. Jepsen [ 17] is a distributed system testing tool, which has been adapted to test DDBMSs in the industry. Mallory [26] is an enhancement built upon the foundation of Jepsen. SQLsmith, SQLancer, and Mozi are three state-of-the-art DBMS testing techniques, which have detected hundreds of bugs in practice. We adapted Jepsen’s default configuration to make it work on these DDBMSs, and Mallory was similarly configured. For SQLsmith, SQLancer, and Mozi, since they were designed for testing traditional DBMS, we used them on a single SQL server of the DDBMS as a baseline.",
+      "context_after": "By using them as baselines, we aim to demonstrate that traditional database testing tools are not directly applicable to distributed databases. 5.1.3 Basic Setup. All experiments were conducted on a 64-bit Ubuntu 22.04 machine with an AMDEPYC 7742 processor (128 cores @ 2.25 GHz) and 488 GiB of main memory. Each testing tool used its default configuration to ensure consistency throughout all evaluations. The DDBMS clusters employed a single setup with 2 replicas and 2 slices, without multi-cluster failover. We ran each DDBMS with one testing tool for 24 hours, which is a commonly used time se",
+      "section": "5.1 Evaluation Setup",
+      "page": 14,
+      "char_offset": 49697,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3 Comparison with Other Techniques To evaluate the effectiveness of our DepState, we conducted comparison experiments between DepState with Jepsen, Mallory, SQLsmith, SQLancer, and Mozi.",
+      "context_before": "te ?Detecting this bug requires complex cross-node dependencies (from node5 tonode2 ) triggered by cluster architecture changes during synchronization. DepState addresses this via ShardinOg-Aware Data Generation and Synchronization-Sensitive Operation Sequences, injecting fine-grained behaviors to reveal intricate synchronization issues. DepState successfully detected this vulnerability within hours, highlighting its efficacy in capturing complex synchronization behaviors. This demonstrates DepState ’s capability in discovering previously unknown synchronization failure bugs, answering RQ1. 5.",
+      "context_after": "Jepsen and Mallory are stateof-the-art tools for testing distributed systems. SQLsmith, SQLancer and Mozi are three popular DBMS testing methods. These five tools support testing the tested DDBMSs, and we compare DepState to them using their default configurations. We ran the testing tools on each DDBMS for Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA088. Publication date: July 2025. DepState: Detecting Synchronization Failure Bugs in Distributed Database Management Systems ISSTA088:17 Table 3. Number of Unique Bugs Detected by Jepsen, Mallory, SQLsmith, SQLancer, Mozi, and DepSta",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 16,
+      "char_offset": 58540,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLsmith, SQLancer and Mozi are three popular DBMS testing methods.",
+      "context_before": "njecting fine-grained behaviors to reveal intricate synchronization issues. DepState successfully detected this vulnerability within hours, highlighting its efficacy in capturing complex synchronization behaviors. This demonstrates DepState ’s capability in discovering previously unknown synchronization failure bugs, answering RQ1. 5.3 Comparison with Other Techniques To evaluate the effectiveness of our DepState, we conducted comparison experiments between DepState with Jepsen, Mallory, SQLsmith, SQLancer, and Mozi. Jepsen and Mallory are stateof-the-art tools for testing distributed systems.",
+      "context_after": "These five tools support testing the tested DDBMSs, and we compare DepState to them using their default configurations. We ran the testing tools on each DDBMS for Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA088. Publication date: July 2025. DepState: Detecting Synchronization Failure Bugs in Distributed Database Management Systems ISSTA088:17 Table 3. Number of Unique Bugs Detected by Jepsen, Mallory, SQLsmith, SQLancer, Mozi, and DepState DDBMS Jepsen Mallory SQLsmith SQLancer Mozi DepState MySQL NDB Cluster 0 1 0 1 7 7 MySQL InnoDB Cluster 1 0 0 1 6 2 MariaDB Galera Cluster 1 1",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 16,
+      "char_offset": 58804,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Number of Unique Bugs Detected by Jepsen, Mallory, SQLsmith, SQLancer, Mozi, and DepState DDBMS Jepsen Mallory SQLsmith SQLancer Mozi DepState MySQL NDB Cluster 0 1 0 1 7 7 MySQL InnoDB Cluster 1 0 0 1 6 2 MariaDB Galera Cluster 1 1 0 1 0 2 TiDB Cluster 0 0 0 0 0 3 Total (Sync Failure Bugs) 2 (0) 2 (0) 0 (0) 3 (0) 13 (0) 14 (14) 24 hours, recording the number of detected bugs and covered synchronization-related functions as metrics.",
+      "context_before": "parison experiments between DepState with Jepsen, Mallory, SQLsmith, SQLancer, and Mozi. Jepsen and Mallory are stateof-the-art tools for testing distributed systems. SQLsmith, SQLancer and Mozi are three popular DBMS testing methods. These five tools support testing the tested DDBMSs, and we compare DepState to them using their default configurations. We ran the testing tools on each DDBMS for Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA088. Publication date: July 2025. DepState: Detecting Synchronization Failure Bugs in Distributed Database Management Systems ISSTA088:17 Table 3.",
+      "context_after": "These functions are implemented specifically for the synchronization process in DDBMSs. 5.3.1 Bug Statistics. DepState outperforms other state-of-the-art testing techniques in finding DDBMS bugs. Table 3 displays the number of bugs detected by each tool in 24 hours. Specifically, DepState detected a total of 7, 2, 2, and 3 synchronization failure bugs on MySQL NDB Cluster, MySQL InnoDB Cluster, MariaDB Galera Cluster, and TiDB Cluster, respectively. In comparison, Jepsen, Mallory, SQLancer, and Mozi reported 2, 2, 3, and 13 bugs, respectively, while SQLsmith found no bugs. We analyzed the roo",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 17,
+      "char_offset": 59238,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In comparison, Jepsen, Mallory, SQLancer, and Mozi reported 2, 2, 3, and 13 bugs, respectively, while SQLsmith found no bugs.",
+      "context_before": ") 2 (0) 2 (0) 0 (0) 3 (0) 13 (0) 14 (14) 24 hours, recording the number of detected bugs and covered synchronization-related functions as metrics. These functions are implemented specifically for the synchronization process in DDBMSs. 5.3.1 Bug Statistics. DepState outperforms other state-of-the-art testing techniques in finding DDBMS bugs. Table 3 displays the number of bugs detected by each tool in 24 hours. Specifically, DepState detected a total of 7, 2, 2, and 3 synchronization failure bugs on MySQL NDB Cluster, MySQL InnoDB Cluster, MariaDB Galera Cluster, and TiDB Cluster, respectively.",
+      "context_after": "We analyzed the root causes of the bugs detected by each tool. The 14 bugs found byDepState are all related to the synchronization process. In contrast, the bugs identified by other tools are transaction or logic bugs, with none involving synchronization functions in DDBMS. 5.3.2 Synchronization-Related Function Coverage. One of the main reasons that DepState discovered synchronization failure bugs is that DepState can cover more synchronization function logic and trigger more behaviors of DDBMSs compared to Jepsen, Mallory, SQLsmith, SQLancer, andMozi. Table 4 shows the number of synchroniza",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 17,
+      "char_offset": 60129,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "One of the main reasons that DepState discovered synchronization failure bugs is that DepState can cover more synchronization function logic and trigger more behaviors of DDBMSs compared to Jepsen, Mallory, SQLsmith, SQLancer, andMozi.",
+      "context_before": "tal of 7, 2, 2, and 3 synchronization failure bugs on MySQL NDB Cluster, MySQL InnoDB Cluster, MariaDB Galera Cluster, and TiDB Cluster, respectively. In comparison, Jepsen, Mallory, SQLancer, and Mozi reported 2, 2, 3, and 13 bugs, respectively, while SQLsmith found no bugs. We analyzed the root causes of the bugs detected by each tool. The 14 bugs found byDepState are all related to the synchronization process. In contrast, the bugs identified by other tools are transaction or logic bugs, with none involving synchronization functions in DDBMS. 5.3.2 Synchronization-Related Function Coverage.",
+      "context_after": "Table 4 shows the number of synchronization-related function lines covered by each technique in 24 hours. It demonstrates that DepState covers more synchronization-related function lines compared to Jepsen, Mallory, SQLsmith, SQLancer, and Mozi. Specifically, DepState covers 6.13%-66.51%, 5.82%-57.28%, 14.12%-83.30%, 36.81%-83.88%, and 43.24%-54.28% more synchronizationrelated function lines compared to Jepsen, Mallory, SQLsmith, SQLancer, and Mozi, respectively. The increased coverage of synchronization-related function lines is attributed to DepState ’s sharding-aware dependent data generat",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 17,
+      "char_offset": 60579,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "It demonstrates that DepState covers more synchronization-related function lines compared to Jepsen, Mallory, SQLsmith, SQLancer, and Mozi.",
+      "context_before": "e 14 bugs found byDepState are all related to the synchronization process. In contrast, the bugs identified by other tools are transaction or logic bugs, with none involving synchronization functions in DDBMS. 5.3.2 Synchronization-Related Function Coverage. One of the main reasons that DepState discovered synchronization failure bugs is that DepState can cover more synchronization function logic and trigger more behaviors of DDBMSs compared to Jepsen, Mallory, SQLsmith, SQLancer, andMozi. Table 4 shows the number of synchronization-related function lines covered by each technique in 24 hours.",
+      "context_after": "Specifically, DepState covers 6.13%-66.51%, 5.82%-57.28%, 14.12%-83.30%, 36.81%-83.88%, and 43.24%-54.28% more synchronizationrelated function lines compared to Jepsen, Mallory, SQLsmith, SQLancer, and Mozi, respectively. The increased coverage of synchronization-related function lines is attributed to DepState ’s sharding-aware dependent data generation and cluster state sequence exploration. Sharding-aware data generation aids DepState in producing dependent test data, while cluster state sequence exploration triggers more synchronization behaviors. In contrast, Jepsen covers fewer synchron",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 17,
+      "char_offset": 60921,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "28% more synchronizationrelated function lines compared to Jepsen, Mallory, SQLsmith, SQLancer, and Mozi, respectively.",
+      "context_before": "ction Coverage. One of the main reasons that DepState discovered synchronization failure bugs is that DepState can cover more synchronization function logic and trigger more behaviors of DDBMSs compared to Jepsen, Mallory, SQLsmith, SQLancer, andMozi. Table 4 shows the number of synchronization-related function lines covered by each technique in 24 hours. It demonstrates that DepState covers more synchronization-related function lines compared to Jepsen, Mallory, SQLsmith, SQLancer, and Mozi. Specifically, DepState covers 6.13%-66.51%, 5.82%-57.28%, 14.12%-83.30%, 36.81%-83.88%, and 43.24%-54.",
+      "context_after": "The increased coverage of synchronization-related function lines is attributed to DepState ’s sharding-aware dependent data generation and cluster state sequence exploration. Sharding-aware data generation aids DepState in producing dependent test data, while cluster state sequence exploration triggers more synchronization behaviors. In contrast, Jepsen covers fewer synchronization function lines because its simplified approach lacks fine-grained analysis, modeling, and creation of complex data scenarios. Its fault injection is limited to the system level, without addressing deeper synchroniz",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 17,
+      "char_offset": 61164,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Moreover, SQLsmith, SQLancer, and Mozi only generate SQL queries, hindering their ability to navigate the vast search space of distributed environments and detect cluster-state-induced bugs.",
+      "context_before": "overage of synchronization-related function lines is attributed to DepState ’s sharding-aware dependent data generation and cluster state sequence exploration. Sharding-aware data generation aids DepState in producing dependent test data, while cluster state sequence exploration triggers more synchronization behaviors. In contrast, Jepsen covers fewer synchronization function lines because its simplified approach lacks fine-grained analysis, modeling, and creation of complex data scenarios. Its fault injection is limited to the system level, without addressing deeper synchronization behaviors.",
+      "context_after": "In contrast, by using a designed data and mutation strategy, DepState discovered 14 synchronization failure bugs in 24 hours, showing DepState can exercise more synchronization functionality and find additional bugs, thus answering RQ2. Table 4. Lines of Synchronization-Related Functions Covered by Tested Tools in 24 Hours DDBMS Jepsen Mallory SQLsmith SQLancer Mozi DepState MySQL NDB Cluster 11,850 11,092 11,505 10,588 10,622 16,388 MySQL InnoDB Cluster 2,096 2,219 1,904 2,551 2,731 3,490 MariaDB Galera Cluster 1,241 1,257 1,147 1,185 1,191 1,706 TiDB Cluster 4,127 4,139 3,838 2,382 3,002 4,",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 17,
+      "char_offset": 61899,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Lines of Synchronization-Related Functions Covered by Tested Tools in 24 Hours DDBMS Jepsen Mallory SQLsmith SQLancer Mozi DepState MySQL NDB Cluster 11,850 11,092 11,505 10,588 10,622 16,388 MySQL InnoDB Cluster 2,096 2,219 1,904 2,551 2,731 3,490 MariaDB Galera Cluster 1,241 1,257 1,147 1,185 1,191 1,706 TiDB Cluster 4,127 4,139 3,838 2,382 3,002 4,380 Proc.",
+      "context_before": "nalysis, modeling, and creation of complex data scenarios. Its fault injection is limited to the system level, without addressing deeper synchronization behaviors. Moreover, SQLsmith, SQLancer, and Mozi only generate SQL queries, hindering their ability to navigate the vast search space of distributed environments and detect cluster-state-induced bugs. In contrast, by using a designed data and mutation strategy, DepState discovered 14 synchronization failure bugs in 24 hours, showing DepState can exercise more synchronization functionality and find additional bugs, thus answering RQ2. Table 4.",
+      "context_after": "ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA088. Publication date: July 2025. ISSTA088:18 Cundi Fang, Jie Liang, Zhiyong Wu, Jingzhou Fu, Zhouyang Jia, Chun Huang, Yu Jiang, and Shanshan Li Table 5. Number of Synchronization Failure Bugs and Synchronization-Related Function Lines Covered by DepState𝐷𝑎𝑡𝑎−,DepState𝑆𝑒𝑞−, and DepState in 24 Hours Number of Synchronization Failure Bugs Line Coverage of Sync-Related Functions DDBMS DepState𝐷𝑎𝑡𝑎−DepState𝑆𝑒𝑞−DepState DepState𝐷𝑎𝑡𝑎−DepState𝑆𝑒𝑞−DepState MySQL NDB Cluster 4 2 7 15,252 12,920 16,388 MySQL InnoDB Cluster 1 0 2 1,765 2,992 3,490 Maria",
+      "section": "5.3 Comparison with Other Techniques",
+      "page": 17,
+      "char_offset": 62336,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Furthermore, SQLancer [ 33] employs metamorphic testing methodologies [ 32] to detect logical bugs within DBMSs, proposing three distinct test oracles for the automatic detection of various correctness issues.",
+      "context_before": "system, primarily focused on the generation of complex SQL queries and the evaluation of their effects on system performance and correctness. Sqirrel [41] introduces mutation-based fuzzing to detect memory safety bugs, which combines SQL query generation with coverage feedback. LEGO [ 22] improves fuzzing by generating SQL sequences through combinatorial logic. It focuses on pairing SQL types to enhance branch coverage. Griffin [ 9] adopts a grammar-free approach. It uses metadata graphs to guide mutations, creating more valid SQL statements and finding errors missed by grammar-based fuzzers.",
+      "context_after": "Mozi [ 23] proposes a configuration-based equivalent transformations framework to detect bugs. Thanos [ 10] is testing against the database’s storage engine. As well as some more work on the configuration in the database [13]. However, existing DBMS testing tools are inadequate for DDBMSs due to their inability to effectively manage the complexities of distributed architectures, such as handling multiple nodes, shards, and intricate data dependencies. In contrast, DepState is specifically designed to detect problems in the synchronization process of DDBMSs. It first generates dependent data a",
+      "section": "7.2 DBMS Testing",
+      "page": 20,
+      "char_offset": 72353,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M2",
+        "M4",
+        "M8"
+      ],
+      "describes_as_state_of_the_art": [
+        "M4",
+        "M5"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:52:31Z",
+    "is_model_written": true,
+    "summary": "DepState looks for synchronization failure bugs in distributed DBMSs -- cases where replicas diverge or a node fails to converge after a state change. It drives clusters through dependency-aware state transitions rather than only issuing queries, so it reaches synchronization code that query-level testing never executes. On MySQL NDB Cluster, MySQL InnoDB Cluster, MariaDB Galera Cluster and TiDB Cluster it found 14 synchronization failure bugs, where the five existing tools found none.",
+    "narrative": "SQLancer is one of five baselines and is called a state-of-the-art DBMS testing technique. Because it was built for single-node systems, the authors ran it against one SQL server of each cluster. The comparison is reported in both bugs and coverage of synchronization-related functions, and the paper's explanation for the gap is structural rather than dismissive: SQLancer, SQLsmith and Mozi only generate SQL queries, which cannot navigate the space of cluster states, so all three of SQLancer's bugs were ordinary ones and none was a synchronization failure.",
+    "roles": {
+      "M1": "result_comparison",
+      "M2": "baseline",
+      "M3": "result_comparison",
+      "M4": "state_of_the_art",
+      "M5": "state_of_the_art",
+      "M6": "baseline",
+      "M7": "baseline",
+      "M8": "baseline",
+      "M9": "definition",
+      "M10": "result_comparison",
+      "M11": "result_comparison",
+      "M12": "result_comparison",
+      "M13": "result_comparison",
+      "M14": "result_comparison",
+      "M15": "result_comparison",
+      "M16": "result_comparison",
+      "M17": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is executed as a baseline on a single server of each cluster; DepState's own mechanism is dependency-aware state transitions across the cluster."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer oracle is carried forward. DepState's oracle is divergence between replicas, which the paper contrasts with query-level testing."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M10",
+          "M11"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "To evaluate the effectiveness of DepState in testing DDBMSs, we compare it with five state-of-the-art tools: Jepsen [ 17],Mallory [26], SQLsmith [ 2], SQLancer [ 33], andMozi [23].",
+            "section": "5.1 Evaluation Setup",
+            "page": 14
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "Number of Unique Bugs Detected by Jepsen, Mallory, SQLsmith, SQLancer, Mozi, and DepState DDBMS Jepsen Mallory SQLsmith SQLancer Mozi DepState MySQL NDB Cluster 0 1 0 1 7 7 MySQL InnoDB Cluster 1 0 0 1 6 2 MariaDB Galera Cluster 1 1 0 1 0 2 TiDB Cluster 0 0 0 0 0 3 Total (Sync Failure Bugs) 2 (0) 2 (0) 0 (0) 3 (0) 13 (0) 14 (14) 24 hours, recording the number of detected bugs and covered synchronization-related functions as metrics.",
+            "section": "5.3 Comparison with Other Techniques",
+            "page": 17
+          },
+          {
+            "mention_id": "M11",
+            "sentence": "In comparison, Jepsen, Mallory, SQLancer, and Mozi reported 2, 2, 3, and 13 bugs, respectively, while SQLsmith found no bugs.",
+            "section": "5.3 Comparison with Other Techniques",
+            "page": 17
+          }
+        ],
+        "reasoning": "M4 names SQLancer among the five tools DepState is compared with, M10 gives the per-cluster unique-bug table, and M11 reports SQLancer's three bugs against DepState's fourteen."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "To evaluate the effectiveness of DepState in testing DDBMSs, we compare it with five state-of-the-art tools: Jepsen [ 17],Mallory [26], SQLsmith [ 2], SQLancer [ 33], andMozi [23].",
+            "section": "5.1 Evaluation Setup",
+            "page": 14
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "SQLsmith, SQLancer, and Mozi are three state-of-the-art DBMS testing techniques, which have detected hundreds of bugs in practice.",
+            "section": "5.1 Evaluation Setup",
+            "page": 14
+          }
+        ],
+        "reasoning": "M4 calls the five comparison tools state-of-the-art, and M5 says specifically that SQLsmith, SQLancer and Mozi are three state-of-the-art DBMS testing techniques that have detected hundreds of bugs in practice."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3728973.json b/_data/papers/paper_doi_10_1145_3728973.json
new file mode 100644
index 0000000..62a466a
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3728973.json
@@ -0,0 +1,778 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:28Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3728973",
+    "title": "Hulk: Exploring Data-Sensitive Performance Anomalies in DBMSs via Data-Driven Analysis",
+    "authors": [
+      "Zhiyong Wu",
+      "Jie Liang",
+      "Jingzhou Fu",
+      "Mingzhe Wang",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Softw. Eng.",
+    "doi": "10.1145/3728973",
+    "arxiv_id": null,
+    "s2_paper_id": "9941e9dde900bcf3bbd359d4166157028ac29f38",
+    "url": "https://doi.org/10.1145/3728973",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3728973",
+      "retrieved_at": "2026-09-09T01:30:28Z",
+      "chars": 88416,
+      "content_sha256": "sha256:bac7266ad906da00dcf86bb7eb6e30d6642d3c31f5f6d2762859382719d730b4"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 22,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2206
+      },
+      {
+        "number": "2",
+        "title": "Data-Sensitive Performance Anomaly",
+        "start": 13030
+      },
+      {
+        "number": "3",
+        "title": "Design of Hulk",
+        "start": 25083
+      },
+      {
+        "number": "3.1",
+        "title": "Data-Sensitive Query Synthesis",
+        "start": 28044
+      },
+      {
+        "number": "3.2",
+        "title": "Performance Anomaly Analysis",
+        "start": 35115
+      },
+      {
+        "number": "4",
+        "title": "Implementation",
+        "start": 46504
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 49719
+      },
+      {
+        "number": "5.1",
+        "title": "Evaluation Setup",
+        "start": 50280
+      },
+      {
+        "number": "5.2",
+        "title": "Performance Anomaly Detection",
+        "start": 51586
+      },
+      {
+        "number": "5.3",
+        "title": "Comparison With Other Techniques",
+        "start": 61537
+      },
+      {
+        "number": "5.4",
+        "title": "Analysis of Hulk ’s Workloads",
+        "start": 64093
+      },
+      {
+        "number": "5.5",
+        "title": "Importance of Data-Sensitivity",
+        "start": 66516
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 69685
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 72761
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 77267
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "[n. d.]. databases. https://en.wikipedia.org/wiki/Database. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "[n. d.]. MariaDB. https://mariadb.org/. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "[n. d.]. MariaDB GitHub. https://github.com/MariaDB/server. Accessed: 2023-10-05.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "[n. d.]. MySQL. https://www.mysql.com/. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "[n. d.]. PostgreSQL. https://www.postgresql.org/. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "[n. d.]. PostgreSQL derived databases. https://wiki.postgresql.org/wiki/PostgreSQL_derived_databases. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "[n. d.]. SQLite and SQLite alternativesdatabases for the Mobile and IoT edge. https://objectbox.io/sqlite-alternatives/. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "AISWare. [n. d.]. AntDB Github. https://github.com/ADBSQL. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "AISWare. [n. d.]. AntDB Website. http://www.antdb.co//. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Abdullah Alsharif, Gregory M Kapfhammer, and Phil McMinn. 2018. DOMINO: Fast and effective test data generation for relational database schemas. In (ICST). IEEE, 12–22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 12,
+      "text": "Daniel Bartholomew. 2014. MariaDB cookbook. Packt Publishing Ltd.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Youssef Bassil. 2012. A comparative study on the performance of the Top DBMS systems. arXiv preprint arXiv:1205.2889 (2012).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M Tamer Özsu. 2007. QAGen: generating query-aware test databases. InProceedings of the 2007 ACMSIGMOD international conference on Management of data. 341–352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "BlackFriday. [n. d.]. Get Access to Black Friday Prices All Year. https://blackfriday.com/. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Avatar Magnus Blåudd. [n. d.]. Percona Website. https://www.percona.com/. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Nicolas Bruno and Surajit Chaudhuri. 2005. Flexible database generators. In Proceedings of the 31st international conference on Very large data bases. 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Ol,egs Čapligins and Andrejs Ermuiža. [n. d.]. MySQL Database Management System Forks Comparison and Usage. ([n. d.]).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "CL Philip Chen and Chun-Yang Zhang. 2014. Data-intensive applications, challenges, techniques and technologies: A survey on Big Data. Information sciences 275 (2014), 314–347.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Raj Chhikara. 1988. The inverse Gaussian distribution: theory: methodology, and applications. Vol. 95. CRC Press.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. In Conference on Automated Software Engineering (ASE’22).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Jim Gray, Prakash Sundaresan, Susanne Englert, Ken Baclawski, and Peter J Weinberger. 1994. Quickly generating billion-record synthetic databases. In Proceedings of the 1994 ACMSIGMOD international conference on Management of data. 243–252.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Kenneth Houkjær, Kristian Torp, and Rico Wind. 2006. Simple and realistic data generation. In Proceedings of the 32nd international conference on Very large data bases. 1243–1246.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Dongxu Huang, Qi Liu, Qiu Cui, Zhuhe Fang, Xiaoyu Ma, Fei Xu, Li Shen, Liu Tang, Yuxing Zhou, Menglong Huang, et al. 2020. TiDB: a Raft-based HTAP database. Proceedings of the VLDB Endowment 13, 12 (2020), 3072–3084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Toshihide Ibaraki and Tiko Kameda. 1984. On the optimal nesting order for computing n-relational joins. ACM Transactions on Database Systems (TODS) 9, 3 (1984), 482–502.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Yannis E Ioannidis. 1996. Query optimization. ACM Computing Surveys (CSUR) 28, 1 (1996), 121–123.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "ISO/IEC. [n. d.]. ISO/IEC 9075-1:2003. https://www.iso.org/standard/34132.html. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Solid IT. [n. d.]. DB-Engines Ranking. https://db-engines.com/en/ranking. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Matthias Jarke and Jurgen Koch. 1984. Query optimization in database systems. ACM Computing surveys (CsUR) 16, 2 (1984), 111–152.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2020. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB). Tokyo, Japan.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O Laleye, and Sarfraz Khurshid. 2008. Query-aware test generation using a relational constraint solver. In 2008 23rd IEEE/ACM International Conference on Automated Software Engineering. IEEE, 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Sang Gyu Kwak and Jong Hae Kim. 2017. Central limit theorem: the cornerstone of modern statistics. Korean journal of anesthesiology 70, 2 (2017), 144–156.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Hai Lan, Zhifeng Bao, and Yuwei Peng. 2021. A survey on advancing the dbms query optimizer: Cardinality estimation, cost model, and plan enumeration. Data Science and Engineering 6 (2021), 86–101.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jonathan Lewis and Thomas Kyte. 2006. Cost-based Oracle fundamentals. Springer. Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA096. Publication date: July 2025. ISSTA096:22 Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. [n. d.]. Sequence-Oriented DBMS Fuzzing. In (ICDE). IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS bugs via configuration-based equivalent transformation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic Detection of Performance Bugs in Database Systems using Equivalent Queries. (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Bruce Momjian. 2001. PostgreSQL: introduction and concepts. Vol. 192. Addison-Wesley New York.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Douglas C Montgomery and George C Runger. 2010. Applied statistics and probability for engineers. John wiley & sons.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Oracle. [n. d.]. MySQL Bug List. https://bugs.mysql.com/. Accessed:April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Oracle. [n. d.]. MySQL Explain Manual. https://dev.mysql.com/doc/refman/8.0/en/using-explain.html. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "PingCAP. [n. d.]. TiDB. https://github.com/pingcap/tidb. Accessed: April 24, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Raghu Ramakrishnan, Johannes Gehrke, and Johannes Gehrke. 2003. Database management systems. Vol. 3. McGrawHill New York.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Kim-Thomas Rehmann, Changyun Seo, Dongwon Hwang, Binh Than Truong, Alexander Boehm, and Dong Hun Lee. 2016. Performance monitoring in sap hana’s continuous integration process. ACMSIGMETRICS Performance Evaluation Review 43, 4 (2016), 43–52.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "M Rigger and Z Su. [n. d.]. Finding Bugs in Database Systems via Query Partitioning. PACMPL 4 (OOPSLA)(Nov 2020).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 46,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 48,
+      "text": "Murray Rosenblatt. 1956. A central limit theorem and a strong mixing condition. Proceedings of the national Academy of Sciences 42, 1 (1956), 43–47.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https: //github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitriy Vyukov. 2012. AddressSanitizer: A Fast Address Sanity Checker. In 2012 USENIX Annual Technical Conference, Boston, MA, USA, June 13-15, 2012, Gernot Heiser and Wilson C. Hsieh (Eds.). USENIX Association, 309–318. https://www.usenix.org/conference/atc12/technicalsessions/presentation/serebryany",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Rick F Van Der Lans. 1989. The SQL standard: a complete guide reference. Prentice Hall International (UK) Ltd.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In Engineering: Software Engineering in Practice (ICSE-SEIP). IEEE, 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Qingshuai Wang, Yuming Li, Rong Zhang, Ke Shu, Zhenjie Zhang, and Aoying Zhou. 2022. A Scalable Query-Aware Enormous Database Generator for Database Evaluation. IEEE Transactions on Knowledge and Data Engineering (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Zuozhi Wang, Kai Zeng, Botong Huang, Wei Chen, Xiaozong Cui, Bo Wang, Ji Liu, Liya Fan, Dachuan Qu, Zhenyu Hou, et al .2020. Tempura: A general cost based optimizer framework for incremental data processing (extended version). arXiv preprint arXiv:2009.13631 (2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Michael Widenius, David Axmark, and Kaj Arno. 2002. MySQL reference manual: documentation from the source. \" O’Reilly Media, Inc.\".",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2024. PUPPY: Finding Performance Degradation Bugs in DBMSs via Limited-Optimization Plan Construction. In Software Engineering (ICSE). IEEE Computer Society, 560–571.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. 2022. Unicorn: Detect Runtime Errors in Time-Series Databases With Hybrid Input Synthesis. In Symposium on Software Testing and Analysis (ISSTA’22).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Khaled Yagoub, Peter Belknap, Benoit Dageville, Karl Dias, Shantanu Joshi, and Hailing Yu. 2008. Oracle’s SQL Performance Analyzer. IEEE Data Eng. Bull. 31, 1 (2008), 51–58.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Jiaqi Yan, Qiuye Jin, Shrainik Jain, Stratis D Viglas, and Allison Lee. 2018. Snowtrail: Testing with production queries on a cloud database. In Proceedings of the Workshop on Testing Database Systems. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020. Received 2024-10-31; accepted 2025-03-31 Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA096. Publication date: July 2025.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 11,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    },
+    {
+      "number": 45,
+      "text": "M Rigger and Z Su. [n. d.]. Finding Bugs in Database Systems via Query Partitioning. PACMPL 4 (OOPSLA)(Nov 2020).",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 46,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[11]",
+      "technique": "cert",
+      "sentence": "Additionally, CERT1[11] tests cardinality estimation to find performance anomalies.",
+      "context_before": "s ISSTA096:3 However, conventional techniques test DBMS performance with fixed datasets, potentially ignoring data-sensitive performance anomalies. Specifically, conventional methods (e.g., system validation tests) compare the response time for predefined queries and data with empirical values as the baseline. Latest fuzzers like Amoeba [37] and APOLLO [ 30], also compare the response time of a query either with the time for a semantically equivalent query or in a previous version of the target DBMS with the fixed data. Consequently, they can hardly detect data-sensitive performance anomalies.",
+      "context_after": "However, it still cannot detect data-sensitive performance anomalies due to the fixed dataset. To efficiently detect data-sensitive performance anomalies caused by optimizers, the intuition is to execute queries with a changed dataset and check whether the DBMS generates a significantly costly plan. However, it is challenging to determine whether the generated plan has low performance for each combination of data and query because it requires accurately estimating cardinality and building the cost model for different datasets. It can be a costly and complex process, which is almost equivalent",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 9249,
+      "cited_reference": {
+        "number": 11,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[11]",
+      "technique": "cert",
+      "sentence": "In addition, we compare Hulk with the state-of-the-art DBMS validation tools in industry, including both DBMS performance testing tool APOLLO [ 30] and SQLancer𝐶𝐸𝑅𝑇[11], as well as DBMS fuzzing tools Sqirrel [60].",
+      "context_before": "omaly by checking if there is a significant difference in the uniform cost of plans and if the comparable DBMS can supply a better plan. The availability of a better plan confirms the performance cliff as a data-sensitive performance anomaly. We implemented the approach in Hulk and evaluated it on six widely-used DBMSs: MySQL, MariaDB, Percona, TiDB, AntDB, and PostgreSQL. Hulk reports a total of 135 anomalies, with 129 anomalies have been confirmed as new bugs, including 14 CVEs assigned. Among them, 94 are confirmed as data-sensitive performance anomalies, and 35 are confirmed as crash bugs.",
+      "context_after": "The 24-hour result shows that Hulk covered 17%, 21%, and 6% more branches, and found 50, 52, and 58 more bugs than APOLLO, SQLancer𝐶𝐸𝑅𝑇, and Sqirrel, respectively. Particularly, Hulk found 37 and 35 more performance anomalies than APOLLO and SQLancer𝐶𝐸𝑅𝑇 in 24 hours. In summary, our paper makes the following contributions: •We find the data sensitive performance anomalies of DBMSs are common and can do great damage to modern data-intensive applications, but the state-of-the-art testing techniques pay little attention to the problem. •We propose data-driven analysis to detect data sensitive pe",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 11449,
+      "cited_reference": {
+        "number": 11,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "1It is implemented in SQLancer and will be referred to as SQLancer𝐶𝐸𝑅𝑇hereafter.",
+      "context_before": "on to the problem. •We propose data-driven analysis to detect data sensitive performance anomalies, consisting of data-sensitive query synthesis and sampling-based performance estimation. It first estimates a range of reasonable response times for growth data volume to capture the performance cliff, and then validates the performance anomaly by plan cost comparison. •We implement our approach in Hulk and detect 135 anomalies in six widely-used DBMSs. Among them, 129 are confirmed as new bugs, including 14 CVEs assigned. Among them, 94 data-sensitive performance anomalies and 35 are crash bugs.",
+      "context_after": "Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA096. Publication date: July 2025. ISSTA096:4 Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang 2 Data-Sensitive Performance Anomaly Basic Concepts. Database Management Systems (DBMS) are software systems designed to store, retrieve, query, and manage data. Structured Query Language (SQL) is a declarative query language that mediates information exchange between users and the DBMS. Specifically, users describe their demands with it in a query. A clause is a built-in function that processes the database table and gives interme",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 12784,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[11]",
+      "technique": "cert",
+      "sentence": "Since only APOLLO [30] and SQLancer𝐶𝐸𝑅𝑇[11] are open-source tools, as a remedy, we still compared Hulk with other state-of-the-art SQL fuzzer ( Sqirrel [60]) to evaluate the capability to explore the state space on DBMSs.",
+      "context_before": "ies? •RQ2: How does Hulk perform compared to other related DBMS testing techniques? •RQ3: How effective of the workloads generated by Hulk? •RQ4: How important is the data-sensitivity of queries in finding performance anomalies? 5.1 Evaluation Setup Tested DBMSs. To evaluate the generality and efficiency of Hulk, we select six popular opensource DBMSs, namely MySQL [ 4,55], MariaDB [ 2,12], Percona [ 16], TiDB [ 24,42], AntDB [ 8,9], and PostgreSQL [5, 38], which are widely used in industry. Compared Techniques. We tried our best to compare Hulk with open-source DBMS performance testing tools.",
+      "context_after": "Basic Setup. We perform the experiments on a machine running 64-bit Ubuntu 20.04 with 128 cores (AMDEPYC 7742 Processor @ 2.25 GHz) and 488 GiB of main memory. All DBMS tested are run in docker containers and can be downloaded directly from their website. For quantitative comparisons, we run the docker containers for each DBMS experiment(including DBMS server Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA096. Publication date: July 2025. ISSTA096:14 Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang andHulk ) with 5 CPU cores and 32 GiB of main memory. To detect real-wor",
+      "section": "5.1 Evaluation Setup",
+      "page": 13,
+      "char_offset": 50651,
+      "cited_reference": {
+        "number": 11,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Number of confirmed performance anomalies and crashes, and covered branches on six DBMSs Performance Bugs Crashes Branches APOLLO 5 6 281,056 SQLancer 7 3 273,467 Sqirrel 0 9 312,713 Hulk 42 19 331,909 data-sensitive clauses on average (which will be discussed in Section 5.",
+      "context_before": "from the sampled historical data to identify anomalies. Consequently, Hulk detects more performance anomalies compared to APOLLO. Moreover, we also find that Hulk performs well in detecting crash bugs. Specifically, Hulk detects 10 more crash bugs when compared to Sqirrel. This can be explained by the design of data-sensitive query synthesis. Hulk generates amounts of complex queries that contain 40 Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA096. Publication date: July 2025. Hulk: Exploring Data-Sensitive Performance Anomalies in DBMSs via Data-Driven Analysis ISSTA096:17 Table 5.",
+      "context_after": "5). These data-sensitive queries trigger more complex behaviors of CBO in DBMSs with the growing data volume. It could be seen from the branches covered by each tool in 24 hours. The third column in Table 5 shows the number of branches covered for APOLLO, SQLancer𝐶𝐸𝑅𝑇,Sqirrel, and Hulk. We can see that Hulk covers 17%, 21%, and 6% more branches than APOLLO, SQLancer𝐶𝐸𝑅𝑇, and Sqirrel, respectively. The improvement of branches proves that the data-sensitive query generated by Hulk can trigger more behaviors of DBMSs compared to other tools. Therefore, Hulk detects more crash bugs compared to oth",
+      "section": "5.3 Comparison With Other Techniques",
+      "page": 17,
+      "char_offset": 63208,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To accommodate the majority of queries generated by popular DBMS fuzzers like Sqirrel and SQLancer, we set the minimum average clause number for evaluation to 5.",
+      "context_before": "so not always conducive to detecting performance anomalies. Specifically, only 5% of performance anomalies are triggered when the data volume is over 100,000 records. 5.5 Importance of Data-Sensitivity Section 3.1 highlights that a SQL query’s data sensitivity increases as the number of data-sensitive clauses it contains increases. To investigate the impact of data sensitivity for detecting performance anomalies, we conducted an experiment that involved running Hulk on 5 different DBMSs with data-sensitive query synthesis settings that varied average clause number (5, 10, 20, 30, 40, 50, 100).",
+      "context_after": "We then analyzed the number of performance bugs, crashes, and branches found by Hulk across different settings. Table 7 presents the number of bugs detected by Hulk when using data-sensitive query synthesis settings with varying clause numbers. The results demonstrate that the data-sensitivity of a query has a significant impact on Hulk ’s ability to detect anomalies. As the number of clauses in a query increases up to a certain point (i.e., when the clause number is less than 50), we observe an increase in the number of performance bugs and crash bugs detected by Hulk. This is because a quer",
+      "section": "5.5 Importance of Data-Sensitivity",
+      "page": 18,
+      "char_offset": 66949,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[ 36,47]",
+      "technique": "pqs",
+      "sentence": "Generally, current DBMS fuzzers are used to detect crash bugs, memory safety bugs with AddressSanitizer [ 50], or logic bugs with the test oracle [ 36,47].",
+      "context_before": "-sensitive performance anomalies and provides an oracle to find them by analyzing query plans to data variations. No matter whether it is a change in data volume or a change in optimizer parameters that causes performance anomalies, the plan-guided oracle could be used to determine both of them. Moreover, since the two types of anomalies are orthogonal, we can even change both the data and parameters to find more bugs for a given workload in the future. 7 Related Work DBMS Fuzzing. Fuzzing is one of the most popular testing techniques and it is natural to adapt fuzzing to the testing of DBMSs.",
+      "context_after": "SQLsmith [ 49] generates queries based on predefined rules in target DBMSs to detect crash bugs. It adapts very well to PostgreSQL and finds many bugs in it. However, it will require additional human efforts when testing other DBMSs. SQLancer [ 47] detects logic bugs of DBMSs by generating queries to fetch an existing row from databases. If the DBMS fails to fetch that, the likely cause is a logic bug. Its following works [ 45,46] also utilize the similar idea by generating semantically equivalent queries and comparing the results. Sqirrel [60],Ratel [52], Unicorn [ 57], and Griffin [ 21] int",
+      "section": "7 Related Work",
+      "page": 19,
+      "char_offset": 72903,
+      "cited_reference": {
+        "number": 47,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 47] detects logic bugs of DBMSs by generating queries to fetch an existing row from databases.",
+      "context_before": "nd parameters to find more bugs for a given workload in the future. 7 Related Work DBMS Fuzzing. Fuzzing is one of the most popular testing techniques and it is natural to adapt fuzzing to the testing of DBMSs. Generally, current DBMS fuzzers are used to detect crash bugs, memory safety bugs with AddressSanitizer [ 50], or logic bugs with the test oracle [ 36,47]. SQLsmith [ 49] generates queries based on predefined rules in target DBMSs to detect crash bugs. It adapts very well to PostgreSQL and finds many bugs in it. However, it will require additional human efforts when testing other DBMSs.",
+      "context_after": "If the DBMS fails to fetch that, the likely cause is a logic bug. Its following works [ 45,46] also utilize the similar idea by generating semantically equivalent queries and comparing the results. Sqirrel [60],Ratel [52], Unicorn [ 57], and Griffin [ 21] introduce coverage-feedback into the query mutation to cover more branches, and they use the AddressSanitizer [ 50] to detect the memory safety bugs. Lego [35] proposes sequence-oriented fuzzing to improve the code coverage by combining different SQL Type Sequences of the statements. Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA096",
+      "section": "7 Related Work",
+      "page": 19,
+      "char_offset": 73293,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker",
+      "surface": "[ 45,46]",
+      "technique": "tlp",
+      "sentence": "Its following works [ 45,46] also utilize the similar idea by generating semantically equivalent queries and comparing the results.",
+      "context_before": "adapt fuzzing to the testing of DBMSs. Generally, current DBMS fuzzers are used to detect crash bugs, memory safety bugs with AddressSanitizer [ 50], or logic bugs with the test oracle [ 36,47]. SQLsmith [ 49] generates queries based on predefined rules in target DBMSs to detect crash bugs. It adapts very well to PostgreSQL and finds many bugs in it. However, it will require additional human efforts when testing other DBMSs. SQLancer [ 47] detects logic bugs of DBMSs by generating queries to fetch an existing row from databases. If the DBMS fails to fetch that, the likely cause is a logic bug.",
+      "context_after": "Sqirrel [60],Ratel [52], Unicorn [ 57], and Griffin [ 21] introduce coverage-feedback into the query mutation to cover more branches, and they use the AddressSanitizer [ 50] to detect the memory safety bugs. Lego [35] proposes sequence-oriented fuzzing to improve the code coverage by combining different SQL Type Sequences of the statements. Proc. ACM Softw. Eng., Vol. 2, No. ISSTA, Article ISSTA096. Publication date: July 2025. ISSTA096:20 Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang Different from them, Hulk aims at data-sensitive performance anomalies. First, Hulk continu",
+      "section": "7 Related Work",
+      "page": 19,
+      "char_offset": 73465,
+      "cited_reference": {
+        "number": 45,
+        "text": "M Rigger and Z Su. [n. d.]. Finding Bugs in Database Systems via Query Partitioning. PACMPL 4 (OOPSLA)(Nov 2020).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "CERT [11] identifies performance issues stemming from unexpectedly estimated cardinalities, which indicate the projected number of rows returned by a query.",
+      "context_before": "izes sampling-based performance estimation to identify data-sensitive performance anomalies. Performance Testing. The traditional approach to detecting performance anomalies is using predefined test suites (i.e. workloads) and checking whether the result exceeds the baseline obtained from experience or empirical experiments [ 44,56,58,59]. Specifically, APOLLO [ 30] detects performance regression bugs by differential testing. It uses the response time in the older version of the target DBMS as a baseline. Amoeba [37] sets the baseline as the execution of equivalent queries on the target DBMSs.",
+      "context_after": "It transforms a query into a more constrained form, where the estimated cardinality should ideally be no greater than that of the original query. Any violation suggests a potential performance concern. Differently, Hulk focuses on performance anomalies related to the optimizer and variation of data volume. It changes the data volume on each query and checks whether the growth in data causes abnormal behaviors in the target comparable DBMSs. After identifying the performance cliffs, it validates the performance anomaly with the plan cost comparison. Data Generation in DBMS. Generating a suffic",
+      "section": "7 Related Work",
+      "page": 20,
+      "char_offset": 75055,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "uses_infrastructure": [
+        "M3"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:25:35Z",
+    "is_model_written": true,
+    "summary": "Hulk looks for performance anomalies that depend on the data rather than the query. The authors argue developers diagnose performance issues by intuition or by comparing execution time against a baseline DBMS, and both overlook how the dataset affects performance. Hulk instead watches performance as the dataset grows, estimating a reasonable response-time range for each data volume to locate performance cliffs, then checks whether a cliff deviates from expectation. Across six DBMSs it reported 135 anomalies, 129 confirmed as new bugs including 14 CVEs.",
+    "narrative": "CERT is the closest prior work and the baseline, and the paper notes it is implemented in SQLancer, referring to it throughout as SQLancer-CERT -- one of only two open-source tools available for the comparison. SQLancer's own oracles are described in related work as the logic-bug counterpart to Hulk's performance focus.",
+    "roles": {
+      "M1": "background",
+      "M2": "baseline",
+      "M3": "definition",
+      "M4": "baseline",
+      "M5": "result_comparison",
+      "M6": "background",
+      "M7": "background",
+      "M8": "definition",
+      "M9": "definition",
+      "M10": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M3, which a pattern fired on, says CERT is implemented in SQLancer -- a fact about the baseline, not about Hulk. Hulk itself is not built on SQLancer."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M2",
+          "M4",
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "In addition, we compare Hulk with the state-of-the-art DBMS validation tools in industry, including both DBMS performance testing tool APOLLO [ 30] and SQLancer𝐶𝐸𝑅𝑇[11], as well as DBMS fuzzing tools Sqirrel [60].",
+            "section": "1 Introduction",
+            "page": 3
+          },
+          {
+            "mention_id": "M4",
+            "sentence": "Since only APOLLO [30] and SQLancer𝐶𝐸𝑅𝑇[11] are open-source tools, as a remedy, we still compared Hulk with other state-of-the-art SQL fuzzer ( Sqirrel [60]) to evaluate the capability to explore the state space on DBMSs.",
+            "section": "5.1 Evaluation Setup",
+            "page": 13
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "Number of confirmed performance anomalies and crashes, and covered branches on six DBMSs Performance Bugs Crashes Branches APOLLO 5 6 281,056 SQLancer 7 3 273,467 Sqirrel 0 9 312,713 Hulk 42 19 331,909 data-sensitive clauses on average (which will be discussed in Section 5.",
+            "section": "5.3 Comparison With Other Techniques",
+            "page": 17
+          }
+        ],
+        "reasoning": "M4 states SQLancer-CERT is one of two open-source tools Hulk was compared against, and M5 reports the anomaly, crash and branch figures for each.",
+        "techniques": [
+          "cert"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M2"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "In addition, we compare Hulk with the state-of-the-art DBMS validation tools in industry, including both DBMS performance testing tool APOLLO [ 30] and SQLancer𝐶𝐸𝑅𝑇[11], as well as DBMS fuzzing tools Sqirrel [60].",
+            "section": "1 Introduction",
+            "page": 3
+          }
+        ],
+        "reasoning": "M2 calls the tools Hulk is compared with, CERT among them, the state-of-the-art DBMS validation tools in industry."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3729175.json b/_data/papers/paper_doi_10_1145_3729175.json
new file mode 100644
index 0000000..11937e1
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3729175.json
@@ -0,0 +1,164 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:09:34Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3729175",
+    "title": "Systems Correctness Practices at Amazon Web Services",
+    "authors": [
+      "Marc Brooker",
+      "A. Desai"
+    ],
+    "year": 2025,
+    "venue": "Communications of the ACM",
+    "doi": "10.1145/3729175",
+    "arxiv_id": null,
+    "s2_paper_id": "f9fb97518fae800c5bd88b573e3211307cd61b5d",
+    "url": "https://doi.org/10.1145/3729175",
+    "also_indexed_as": [
+      "paper:doi:10.1145/3712057"
+    ]
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3729175",
+      "retrieved_at": "2026-09-10T15:09:34Z",
+      "chars": 25458,
+      "content_sha256": "sha256:2e021fb5b3190121fc6b8405e871b09dfea313616429eeba8ed2902ebbfd3053"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 5,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Amazon S3 now delivers strong read-after-write",
+        "start": 22313
+      },
+      {
+        "number": "2",
+        "title": "Announcing general availability of AWS Fault Injection",
+        "start": 22469
+      },
+      {
+        "number": "3",
+        "title": "Announcing Amazon Aurora Limitless Database.",
+        "start": 22651
+      },
+      {
+        "number": "4",
+        "title": "Bornholt, J. et al. Using lightweight formal methods",
+        "start": 22757
+      },
+      {
+        "number": "5",
+        "title": "Bronson, N., Aghayev, A., Charapko, A., and Zhu,",
+        "start": 22983
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Amazon S3 now delivers strong read-after-write consistency automatically for all applications. Amazon Web Services (2020); https://tinyurl.com/25mwxugm.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Announcing general availability of AWS Fault Injection Simulator, a fully managed service to run controlled experiments. Amazon Web Services (2021); https://tinyurl.com/yhnr9nuj.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Announcing Amazon Aurora Limitless Database. Amazon Web Services (2023); https://tinyurl.com/28hwdrm8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Bornholt, J. et al. Using lightweight formal methods to validate a key-value storage node in Amazon S3. In Proceedings of the ACMSIGOPS 28th Symp. Operating Systems Principles (2021), 836–850; https://tinyurl.com/2yczbl4s.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Bronson, N., Aghayev, A., Charapko, A., and Zhu, T. Metastable failures in distributed systems. In Challenges and Opportunities for the Future Despite significant success in scaling formal and semi-formal testing meth ods across AWS over the past 15 years, several challenges persist, particularly in industrial adoption of formal meth ods. The primary barriers for formal methods tools include their",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "pivoted query synthesis",
+      "technique": "pqs",
+      "sentence": "Testing database engines via pivoted query synthesis.",
+      "context_before": "cIver, D.R. et al. Hypothesis: A new approach to property-based testing. J. Open Source Software 4, 43 (2019), 1,891; https://tinyurl.com/239azs4n. 20. Monteiro, F. and Roy, P. Using Kani to validate security boundaries in AWS Firecracker. Technical report, Amazon Web Services (2023); https://tinyurl.com/2cf6dub8. 21. Newcombe, C. et al. How Amazon Web Services uses formal methods. Commun. ACM 58, 4 (2015), 66–73; https://tinyurl.com/2au9fctm. 22. P Team. P: A programming language for formal specification of distributed systems (2024); https://github.com/p-org/P 23. Rigger, M. and Su, Z. 2020.",
+      "context_after": "In Proceedings of the 14th Usenix Symp. on Operating Systems Design and Implementation, Article 38 (2020), 667–682; https://tinyurl.com/2da6rn2e. 24. Rungta, N. Trillions of formally verified authorizations a day! Splash keynote (2024); https://tinyurl.com/24ufvhff. 25. Vogels, W. Diving deep on S3 consistency. All Things Distributed; https://tinyurl.com/2xo4xelw. Marc Brooker is a distinguished engineer at Amazon Web Services, Seattle, WA, USA. Ankush Desai is a principal applied scientist at Amazon Web Services, Seattle, WA, USA.challenges of formal methods in prac tice. Just as AI-assisted",
+      "section": null,
+      "page": 5,
+      "char_offset": 20679,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "An account of how Amazon Web Services applies formal and semi-formal methods to systems correctness, covering lightweight formal methods, deterministic simulation testing, and property-based approaches used across its services.",
+    "narrative": "The extracted page carries a bibliography entry for pivoted query synthesis. The passage crediting SQLancer with pioneering validation approaches used alongside AWS's own non-sharded reference oracle is in the article but did not survive extraction of this five-page excerpt, which begins mid-sentence and carries no title.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation index recorded a sentence describing AWS as validating against an oracle 'as well as other approaches to validation, like those pioneered by SQLancer', which suggests use in production, but that sentence is not among the mentions extracted from this PDF."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3729319.json b/_data/papers/paper_doi_10_1145_3729319.json
new file mode 100644
index 0000000..e88401b
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3729319.json
@@ -0,0 +1,545 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:26Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3729319",
+    "title": "Graphiti: Bridging Graph and Relational Database Queries",
+    "authors": [
+      "Yang He",
+      "Ruijie Fang",
+      "Işıl Dillig",
+      "Yuepeng Wang"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Program. Lang.",
+    "doi": "10.1145/3729319",
+    "arxiv_id": "2504.03182",
+    "s2_paper_id": "778b9fb9fe2b30e6071d406d3abbbb82491ad070",
+    "url": "https://doi.org/10.1145/3729319",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2504.03182",
+      "retrieved_at": "2026-09-09T01:30:26Z",
+      "chars": 144426,
+      "content_sha256": "sha256:6547dc3fb05fc14f9864c871f9c8fc3b06c7ad18503080f282b130f3ad069ab4"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 44,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1876
+      },
+      {
+        "number": "2",
+        "title": "Motivating Example",
+        "start": 13664
+      },
+      {
+        "number": "3",
+        "title": "Preliminaries",
+        "start": 25094
+      },
+      {
+        "number": "3.1",
+        "title": "Background on Graph Databases",
+        "start": 25110
+      },
+      {
+        "number": "3.3",
+        "title": "Relational Databases",
+        "start": 29954
+      },
+      {
+        "number": "4",
+        "title": "Problem Statement",
+        "start": 32732
+      },
+      {
+        "number": "4.2",
+        "title": "Equivalence Checking Problem",
+        "start": 36380
+      },
+      {
+        "number": "5",
+        "title": "Equivalence Checking Algorithm",
+        "start": 39367
+      },
+      {
+        "number": "5.1",
+        "title": "Induced Relational Schema and Standard Transformer Inference",
+        "start": 42121
+      },
+      {
+        "number": "5.2",
+        "title": "Syntax-Directed Transpilation",
+        "start": 45737
+      },
+      {
+        "number": "5.3",
+        "title": "Reduction to SQL Equivalence Checking",
+        "start": 56673
+      },
+      {
+        "number": "6",
+        "title": "Evaluation",
+        "start": 58936
+      },
+      {
+        "number": "6.1",
+        "title": "Evaluation of Graphiti with BMC backend",
+        "start": 62309
+      },
+      {
+        "number": "6.2",
+        "title": "Evaluation of Graphiti with Deductive Verifier",
+        "start": 66757
+      },
+      {
+        "number": "6.3",
+        "title": "Evaluation of Transpilation",
+        "start": 69589
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 72767
+      },
+      {
+        "number": "8",
+        "title": "Limitation",
+        "start": 78076
+      },
+      {
+        "number": "9",
+        "title": "Conclusion and Future Work",
+        "start": 78839
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "AGE. 2024. Apache AGE. https://age.apache.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Alfred V. Aho, Yehoshua Sagiv, and Jeffrey D. Ullman. 1979. Equivalences Among Relational Expressions. SIAM J. Comput. 8, 2 (1979), 218–246. https://doi.org/10.1137/0208017",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Amazon. 2024. What’s the Difference Between a Graph Database and a Relational Database? https://aws.amazon.com/ compare/the-difference-between-graph-and-relational-database/?nc1=h_ls.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Abdelkrim Boudaoud, Houari Mahfoud, and Azeddine Chikh. 2022. Towards a Complete Direct Mapping from Relational Databases to Property Graphs. In Proceedings of the International Conference on Model and Data Engineering (MEDI). 222–235. https://doi.org/10.1007/978-3-031-21595-7_16",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Ashok K. Chandra and Philip M. Merlin. 1977. Optimal Implementation of Conjunctive Queries in Relational Data Bases. InProceedings of the ACM Symposium on Theory of Computing (STOC). 77–90. https://doi.org/10.1145/800105.803397",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Bikash Chandra, Bhupesh Chawda, Biplab Kar, K. V. Maheshwara Reddy, Shetal Shah, and S. Sudarshan. 2015. Data generation for testing and grading SQL queries. VLDB J. 24, 6 (2015), 731–755. https://doi.org/10.1007/S00778-0150395-0",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Tsong Yueh Chen, S. C. Cheung, and Siu-Ming Yiu. 2020. Metamorphic testing: a new approach for generating next test cases. CoRR abs/2002.12543 (2020). arXiv:2002.12543 https://arxiv.org/abs/2002.12543",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Yanju Chen, Yuepeng Wang, Maruth Goyal, James Dong, Yu Feng, and Isil Dillig. 2022. Synthesis-powered optimization of smart contracts via data type refactoring. Proc. ACM Program. Lang. 6, OOPSLA2 (2022), 560–588. https://doi.org/ 10.1145/3563308",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Alvin Cheung, Armando Solar-Lezama, and Samuel Madden. 2013. Optimizing database-backed applications with query synthesis. In ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). ACM, 3–14. https://doi.org/10.1145/2491956.2462180",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Shumo Chu, Brendan Murphy, Jared Roesch, Alvin Cheung, and Dan Suciu. 2018. Axiomatic Foundations and Algorithms for Deciding Semantic Equivalences of SQL Queries. Proc. VLDB Endow. 11, 11 (2018), 1482–1495. https: //doi.org/10.14778/3236187.3236200",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Shumo Chu, Chenglong Wang, Konstantin Weitz, and Alvin Cheung. 2017. Cosette: An Automated Prover for SQL. In Proceedings of the Biennial Conference on Innovative Data Systems Research (CIDR). http://cidrdb.org/cidr2017/papers/ p51-chu-cidr17.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Shumo Chu, Konstantin Weitz, Alvin Cheung, and Dan Suciu. 2017. HoTTSQL: proving query rewrites with univalent SQL semantics. In Proceedings of the SIGPLAN Conference on Programming Language Design and Implementation (PLDI). 510–524. https://doi.org/10.1145/3062341.3062348",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Carlo Curino, Hyun Jin Moon, Alin Deutsch, and Carlo Zaniolo. 2013. Automating the database schema evolution process. VLDB J. 22, 1 (2013), 73–98. https://doi.org/10.1007/s00778-012-0302-x",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Carlo Curino, Hyun Jin Moon, and Carlo Zaniolo. 2008. Graceful database schema evolution: the PRISM workbench. Proc. VLDB Endow. 1, 1 (2008), 761–772. https://doi.org/10.14778/1453856.1453939",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Benjamin Delaware, Clément Pit-Claudel, Jason Gross, and Adam Chlipala. 2015. Fiat: Deductive Synthesis of Abstract Data Types in a Proof Assistant. In Proceedings of the ACMSIGPLAN-SIGACT Symposium on Principles of Programming Languages (POPL). 689–700. https://doi.org/10.1145/2676726.2677006",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Alin Deutsch, Nadime Francis, Alastair Green, Keith Hare, Bei Li, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Wim Martens, Jan Michels, Filip Murlak, Stefan Plantikow, Petra Selmer, Oskar van Rest, Hannes Voigt, Domagoj Vrgoc, Mingxi Wu, and Fred Zemke. 2022. Graph Pattern Matching in GQL and SQL/PGQ. In International Conference on Management of Data (SIGMOD). 2246–2258. https://doi.org/10.1",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Ronald Fagin, Laura M. Haas, Mauricio A. Hernández, Renée J. Miller, Lucian Popa, and Yannis Velegrakis. 2009. Clio: Schema Mapping Creation and Data Exchange. In Conceptual Modeling: Foundations and Applications-Essays in Honor of John Mylopoulos (Lecture Notes in Computer Science, Vol. 5600). 198–236. https://doi.org/10.1007/978-3-642-02463-4_12",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Yu Feng, Ruben Martins, Jacob Van Geffen, Isil Dillig, and Swarat Chaudhuri. 2017. Component-based synthesis of table consolidation and transformation tasks from examples. In Proceedings of the ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). 422–436. https://doi.org/10.1145/3062341.3062351",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and Andrés Taylor. 2018. Cypher: An Evolving Query Language for Property Graphs. In Proceedings of the 2018 International Conference on Management of Data (SIGMOD). 1433–1445. https://doi.org/10.1145/ 3183713.3190657",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Xi Ge, Quinton L. DuBose, and Emerson R. Murphy-Hill. 2012. Reconciling manual and automatic refactoring. In International Conference on Software Engineering (ICSE). 211–221. https://doi.org/10.1109/ICSE.2012.6227192",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Todd J. Green. 2011. Containment of Conjunctive Queries on Annotated Relations. Theory Comput. Syst. 49, 2 (2011), 429–459. https://doi.org/10.1007/S00224-011-9327-6 Proc. ACM Program. Lang., Vol. 1, No. 1, Article 1. Publication date: January 2025. 1:24 Yang He, Ruijie Fang, Işıl Dillig, and Yuepeng Wang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Harry Halpin and James Cheney. 2011. Dynamic Provenance for SPARQL Updates Using Named Graphs. In Workshop on the Theory and Practice of Provenance (TaPP). https://doi.org/10.1145/2567948.2577357",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Olaf Hartig and Jorge Pérez. 2018. Semantics and Complexity of GraphQL. In Proceedings of the World Wide Web Conference on World Wide Web (WWW). 1155–1164. https://doi.org/10.1145/3178876.3186014",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Yang He, Pinhan Zhao, Xinyu Wang, and Yuepeng Wang. 2024. VeriEQL: Bounded Equivalence Verification for Complex SQL Queries with Integrity Constraints. Proc. ACM Program. Lang. 8, OOPSLA1 (2024), 1071–1099. https: //doi.org/10.1145/3649849",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting Bugs in Cypher Graph Database Engines. In Proceedings of the SIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 163–174. https://doi.org/10.1145/3597926.3598046",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland HC Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the International Conference on Software Engineering (ICSE). https://doi.org/10.1145/3597503.3623307",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "Zhongjun Jin, Michael R. Anderson, Michael J. Cafarella, and H. V. Jagadish. 2017. Foofah: Transforming Data By Example. In Proceedings of the ACM International Conference on Management of Data (SIGMOD). 683–698. https: //doi.org/10.1145/3035918.3064034",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Shadaj Laddad, Conor Power, Mae Milano, Alvin Cheung, and Joseph M. Hellerstein. 2022. Katara: synthesizing CRDTs with verified lifting. Proc. ACM Program. Lang. 6, OOPSLA2 (2022), 1349–1377. https://doi.org/10.1145/3563336",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jerry Liang. 2025. openCypher Transpiler. https://github.com/microsoft/openCypherTranspiler.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Chunbin Lin, Benjamin Mandel, Yannis Papakonstantinou, and Matthias Springer. 2016. Fast In-Memory SQL Analytics on Relationships between Entities. CoRR abs/1602.00033 (2016). arXiv:1602.00033 http://arxiv.org/abs/1602.00033",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Qiuyang Mang, Aoyang Fang, Boxi Yu, Hanfei Chen, and Pinjia He. 2024. Testing Graph Database Systems via Equivalent Query Rewriting. In Proceedings of the International Conference on Software Engineering (ICSE). https: //doi.org/10.1145/3597503.3639200",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Ruben Martins, Jia Chen, Yanju Chen, Yu Feng, and Isil Dillig. 2019. Trinity: An Extensible Synthesis Framework for Data Science. Proc. VLDB Endow. 12, 12 (2019), 1914–1917. https://doi.org/10.14778/3352063.3352098",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software. Digit. Tech. J. 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Zhengjie Miao, Sudeepa Roy, and Jun Yang. 2019. Explaining Wrong Queries Using Small Examples. In Proceedings of the International Conference on Management of Data (SIGMOD). 503–520. https://doi.org/10.1145/3299869.3319866",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Renée J. Miller, Laura M. Haas, and Mauricio A. Hernández. 2000. Schema Mapping as Query Discovery. In Proceedings of the International Conference on Very Large Data Bases (VLDB). 77–88. http://www.vldb.org/conf/2000/P077.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Neo4j. 2024. Cypher for SQL Users. https://neo4j.com/docs/getting-started/cypher-intro/cypher-sql/ Accessed: 2024-11-07.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Neo4j. 2024. Neo4j Docs. https://neo4j.com/docs/getting-started/cypher-intro/cypher-sql/#_return_customers_ without_existing_orders.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Neo4j. 2024. Transition from Relational to Graph Database. https://neo4j.com/docs/getting-started/appendix/graphdbconcepts/graphdb-vs-rdbms.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "NLM. 2024. SemMedDB Database Details-Version 2.0. https://lhncbc.nlm.nih.gov/ii/tools/SemRep_SemMedDB_SKR/ dbinfo20.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Stack Overflow. 2024. How could i use this SQL on cypher(neo4j). https://stackoverflow.com/questions/43438214/howcould-i-use-this-sql-on-cypherneo4j.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Shankara Pailoor, Yuepeng Wang, and Isil Dillig. 2024. Semantic Code Refactoring for Abstract Data Types. Proc. ACM Program. Lang. 8, POPL (2024), 816–847. https://doi.org/10.1145/3632870",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Shankara Pailoor, Yuepeng Wang, Xinyu Wang, and Isil Dillig. 2021. Synthesizing data structure refinements from integrity constraints. In Proceedings of the International Conference on Programming Language Design and Implementation (PLDI). 574–587. https://doi.org/10.1145/3453483.3454063",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Kia Rahmani, Kartik Nagar, Benjamin Delaware, and Suresh Jagannathan. 2021. Repairing serializability bugs in distributed database programs via automated schema refactoring. In ACMSIGPLAN International Conference on Programming Language Design and Implementation (PLDI). 32–47. https://doi.org/10.1145/3453483.3454028",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 45,
+      "text": "George G. Robertson, Mary Czerwinski, and John E. Churchill. 2005. Visualization of mappings between schemas. In Proceedings of the Conference on Human Factors in Computing Systems (CHI). 431–439. https://doi.org/10.1145/1054972. 1055032",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Marko A. Rodriguez. 2015. The Gremlin graph traversal machine and language (invited talk). In Proceedings of the Symposium on Database Programming Languages (DBPL). 1–10. https://doi.org/10.1145/2815072.2815073 Proc. ACM Program. Lang., Vol. 1, No. 1, Article 1. Publication date: January 2025. Graphiti: Bridging Graph and Relational Database Queries 1:25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "StackOverflow. 2024. Comparison of relational databases and graph databases. https://stackoverflow.com/questions/ 13046442/comparison-of-relational-databases-and-graph-databases.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Kuzu Team. 2025. Kuzu. https://kuzudb.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Boris A Trakhtenbrot. 1950. Impossibility of an algorithm for the decision problem in finite classes. In Doklady Akademii Nauk SSSR 70. 569–572.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Oskar van Rest, Sungpack Hong, Jinha Kim, Xuming Meng, and Hassan Chafi. 2016. PGQL: a property graph query language. In Proceedings of the International Workshop on Graph Data Management Experiences and Systems (GRADES). 7. https://doi.org/10.1145/2960414.2960421",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Margus Veanes, Nikolai Tillmann, and Jonathan de Halleux. 2010. Qex: Symbolic SQL Query Explorer. In International Conference on Logic for Programming, Artificial Intelligence, and Reasoning (LPAR). 425–446. https://doi.org/10.1007/9783-642-17511-4_24",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Chad Vicknair, Michael Macias, Zhendong Zhao, Xiaofei Nan, Yixin Chen, and Dawn Wilkins. 2010. A comparison of a graph database and a relational database: a data provenance perspective. In Proceedings of the Annual Southeast Regional Conference. 42. https://doi.org/10.1145/1900008.1900067",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Chenglong Wang, Alvin Cheung, and Rastislav Bodík. 2017. Synthesizing highly expressive SQL queries from inputoutput examples. In Proceedings of the ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). 452–466. https://doi.org/10.1145/3062341.3062365",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Meng Wang, Jeremy Gibbons, Kazutaka Matsuda, and Zhenjiang Hu. 2013. Refactoring pattern matching. Sci. Comput. Program. 78, 11 (2013), 2216–2242. https://doi.org/10.1016/J.SCICO.2012.07.014",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Yuepeng Wang, Isil Dillig, Shuvendu K. Lahiri, and William R. Cook. 2018. Verifying equivalence of database-driven applications. Proc. ACM Program. Lang. 2, POPL (2018), 56:1–56:29. https://doi.org/10.1145/3158144",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Yuepeng Wang, James Dong, Rushi Shah, and Isil Dillig. 2019. Synthesizing database programs for schema refactoring. InProceedings of the ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). 286–300. https://doi.org/10.1145/3314221.3314588",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Yuepeng Wang, Rushi Shah, Abby Criswell, Rong Pan, and Isil Dillig. 2020. Data Migration using Datalog Program Synthesis. Proc. VLDB Endow. 13, 7 (2020), 1006–1019. https://doi.org/10.14778/3384345.3384350",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Navid Yaghmazadeh, Christian Klinger, Isil Dillig, and Swarat Chaudhuri. 2016. Synthesizing transformations on hierarchically structured data. In Proceedings of the ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI). 508–521. https://doi.org/10.1145/2908080.2908088",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Navid Yaghmazadeh, Xinyu Wang, and Isil Dillig. 2018. Automated Migration of Hierarchical Data to Relational Tables using Programming-by-Example. Proc. VLDB Endow. 11, 5 (2018), 580–593. https://doi.org/10.1145/3187009.3177735",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding bugs in Gremlin-based graph database systems via Randomized differential testing. In Proceedings of the SIGSOFT International Symposium on Software Testing and Analysis (ISSTA). 302–313. https://doi.org/10.1145/3533767. 3534409",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Qi Zhou, Joy Arulraj, Shamkant B. Navathe, William Harris, and Jinpeng Wu. 2022. SPES: A Symbolic Approach to Proving Query Equivalence Under Bag Semantics. In International Conference on Data Engineering (ICDE). 2735–2748. https://doi.org/10.1109/ICDE53745.2022.00250",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Qi Zhou, Joy Arulraj, Shamkant B. Navathe, William Harris, and Dong Xu. 2019. Automated Verification of Query Equivalence Using Satisfiability Modulo Theories. Proc. VLDB Endow. 12, 11 (2019), 1276–1288. https://doi.org/10. 14778/3342263.3342267 Proc. ACM Program. Lang., Vol. 1, No. 1, Article 1. Publication date: January 2025. 1:26 Yang He, Ruijie Fang, Işıl Dillig, and Yuepeng Wang A Semantics o",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 26,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland HC Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the International Conference on Software Engineering (ICSE). https://doi.org/10.1145/3597503.3623307",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 7,26,31,44]",
+      "technique": "tlp",
+      "sentence": "Work in this space includes differential testing [ 25,33,60] and metamorphic testing [ 7,26,31,44] to detect bugs in database management systems, mutation-based testing to grade programming assignments involving SQL queries [ 6], and provenance-based techniques for explaining wrong SQL queries [ 34].",
+      "context_before": "ntation of the Entity-Relationship Diagram (ER Diagram) of a relational database schema. In part due to this unifying insight, these graph query languages are both semantically and syntactically similar Proc. ACM Program. Lang., Vol. 1, No. 1, Article 1. Publication date: January 2025. 1:22 Yang He, Ruijie Fang, Işıl Dillig, and Yuepeng Wang to Cypher. Because of this similarity, we believe our proposed methodology can be adapted fairly easily to graph database query languages other than Cypher. Testing database queries. Another related line of related work focuses on testing database queries.",
+      "context_after": "In contrast, Graphiti transpiles graph database queries into equivalent SQL queries and uses existing automated reasoning tools for SQL equivalence checking. Thus, Graphiti is complementary and can benefit from advances in testing SQL queries. Transpiling Cypher queries. There are a few existing tools that can translate Cypher queries to queries in SQL-like languages [ 29,48]. The most relevant to this paper is OpenCypherTranspiler [29], which first transforms a Cypher query into a logical plan and renders it as a relational query. However, it supports only a limited subset of Cypher and lack",
+      "section": "7 Related Work",
+      "page": 22,
+      "char_offset": 76677,
+      "cited_reference": {
+        "number": 44,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30. https://doi.org/10.1145/3428279",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "Graphiti checks whether a Cypher graph query and a SQL relational query are equivalent. To make that question well-posed across two data models the authors introduce database transformers, mapping instances between the graph and relational representations, and define equivalence modulo a transformer. Verification reduces to checking two SQL queries by translating a subset of Cypher into SQL. Graphiti proved useful for verification and refutation, uncovering subtle bugs including some in Cypher tutorials and academic papers.",
+    "narrative": "One citation in related work, listing metamorphic testing for DBMS bug detection alongside differential testing, with work by the SQLancer authors among the references.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3731569_3764841.json b/_data/papers/paper_doi_10_1145_3731569_3764841.json
new file mode 100644
index 0000000..b2aad15
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3731569_3764841.json
@@ -0,0 +1,573 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:19Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3731569.3764841",
+    "title": "Fawkes: Finding Data Durability Bugs in DBMSs via Recovered Data State Verification",
+    "authors": [
+      "Zhiyong Wu",
+      "Jie Liang",
+      "Jingzhou Fu",
+      "Wenqian Deng",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "Symposium on Operating Systems Principles",
+    "doi": "10.1145/3731569.3764841",
+    "arxiv_id": null,
+    "s2_paper_id": "e8506c9f0e513b48630e0d8329746e4fc73a99af",
+    "url": "https://doi.org/10.1145/3731569.3764841",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3731569.3764841",
+      "retrieved_at": "2026-09-09T01:30:19Z",
+      "chars": 84265,
+      "content_sha256": "sha256:9fca9e49ebeee401968c057e0ff901a2d7a230768bdf0a0878e8f3ea63ec5a02"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Crash recovery to ensure durability. https://docs .oracle.com/ en-us/iaas/mysql-database/doc/crash-recovery .html/. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Data durability in dbms. https://en .wikipedia.org/wiki/ Durability _(database _systems). Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Mariadb. https://mariadb .org/. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Mysql. https://www .mysql.com/. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Official iotdb grammar file. https://github .com/apache/ iotdb/blob/858c8b8538ba5dd370ce9604e659bd0050303c58/ iotdb-core/antlr/src/main/antlr4/org/apache/iotdb/db/qp/ sql/IoTDBSqlParser .g4#L4. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Official mariadb grammar file. https://github .com/MariaDB/ server/blob/main/sql/sql _yacc.yy. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Official mysql grammar file. https://github .com/mysql/mysqlserver/blob/ff05628a530696bc6851ba6540ac250c7a059aa7/ sql/sql_yacc .yy. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Official postgresql sql grammar file. https://github .com/ postgres/postgres/blob/master/src/backend/parser/gram .y. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Postgresql. https://www .postgresql.org/. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Alvaro, P., Rosen, J., and Hellerstein, J. M. Lineage-driven fault injection. In Proceedings of the 2015 ACMSIGMOD International Conference on Management of Data, Melbourne, Victoria, Australia, May 31 - June 4, 2015 (2015), T. K. Sellis, S. B. Davidson, and Z. G. Ives, Eds., ACM, pp. 331–346.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Ba, J., and Rigger, M. Keep it simple: Testing databases via differential query plans. Proceedings of the ACM on Management of Data 2, 3 (2024), 1–26.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 12,
+      "text": "Campbell, L., and Majors, C. Database reliability engineering: designing and operating resilient database systems. \" O’Reilly Media, Inc.\", 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Commitee, P. Write-ahead logging (wal) in postgresql. https: //www.postgresql.org/docs/current/wal-intro .html. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Community, C. Cnosdb github, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Community, O. Opengemini github, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Community, P. Postgresql bug list. https: //www.postgresql.org/. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Community, P. Postgresql test, Today.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Company, G. Griddb: Open database supporting cyberphysical systems, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Fu, J., Liang, J., Wu, Z., Wang, M., and Jiang, Y. Griffin: Grammar-free dbms fuzzing. In Conference on Automated Software Engineering (ASE’22) (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Fu, J., Liang, J., Wu, Z., Zhao, Y., Li, S., and Jiang, Y. Understanding and detecting sql function bugs: Using simple boundary arguments to trigger hundreds of dbms bugs. In Proceedings of the Twentieth European Conference on Computer Systems (2025), pp. 1061–1076.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Gao, Y., Dou, W., Wang, D., Feng, W., Wei, J., Zhong, H., and Huang, T. Coverage guided fault injection for cloud systems. In Engineering (ICSE) (2023), IEEE, pp. 2211–2223.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Gunawi, H. S., Do, T., Joshi, P., Alvaro, P., Hellerstein, J. M., Arpaci-Dusseau, A. C., Arpaci-Dusseau, R. H., Sen, K., and Borthakur, D. FATE and DESTINI: A framework for cloud recovery testing. In Proceedings of the 8th USENIX Symposium on Networked Systems Design and Implementation, NSDI 2011, Boston, MA, USA, March 30 - April 1, 2011 (2011), D. G. Andersen and S. Ratnasamy, Eds., USENIX Assoc",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Hao, Z., Huang, Q., Wang, C., Wang, J., Zhang, Y., Wu, R., and Zhang, C. Pinolo: Detecting logical bugs in database management systems with approximate query synthesis. In 2023 USENIX Annual Technical Conference (USENIX ATC 23) (2023), pp. 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "IBM. Checkpoints. https://www .ibm.com/docs/en/informixservers/12.10?topic =recovery-checkpoints. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Jepsen. Jepsen. https://github .com/jepsen-io/jepsen, 2024. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Jiang, Z.-M., Liu, S., Rigger, M., and Su, Z. Detecting transactional bugs in database engines via {graph-based}oracle 682 SOSP ’25, October 13–16, 2025, Seoul, Republic of Korea Zhiyong Wu, Jie Liang*, Jingzhou Fu, Wenqian Deng, and Yu Jiang* construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23) (2023), pp. 397–417.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "Joshi, P., Gunawi, H. S., and Sen, K. PREFAIL: a programmable tool for multiple-failure injection. In Proceedings of the 26th Annual ACMSIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications, OOPSLA 2011, part of SPLASH 2011, Portland, OR, USA, October 22 - 27, 2011 (2011), C. V. Lopes and K. Fisher, Eds., ACM, pp. 171–188.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Jung, J., Hu, H., Arulraj, J., Kim, T., and Kang, W. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB) (Tokyo, Japan, Aug. 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Liang, J., Chen, Y., Wu, Z., Fu, J., Wang, M., Jiang, Y., Huang, X., Chen, T., Wang, J., and Li, J. Sequence-oriented dbms fuzzing. In neering (ICDE), IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Liang, J., Wu, Z., Fu, J., Bai, Y., Zhang, Q., and Jiang, Y. {WingFuzz}: Implementing continuous fuzzing for {DBMSs}. In2024 USENIX Annual Technical Conference (USENIX ATC 24) (2024), pp. 479–492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Liang, J., Wu, Z., Fu, J., Wang, M., Sun, C., and Jiang, Y. Mozi: Discovering dbms bugs via configuration-based equivalent transformation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (2024), pp. 1– 12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Liu, X., Zhou, Q., Arulraj, J., and Orso, A. Automatic detection of performance bugs in database systems using equivalent queries.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Ltd, R. G. S. Db-engines ranking of time series dbms. https: //db-engines .com/en/ranking/time +series+ dbms. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Lu, J., Li, F., Liu, C., Li, L., Feng, X., and Xue, J. Cloudraid: Detecting distributed concurrency bugs via log mining and enhancement. IEEE Trans. Software Eng. 48, 2 (2022), 662–677.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Lu, J., Liu, C., Li, L., Feng, X., Tan, F., Yang, J., and You, L.Crashtuner: detecting crash-recovery bugs in cloud systems via meta-info analysis. In Proceedings of the 27th ACM Symposium on Operating Systems Principles, SOSP 2019, Huntsville, ON, Canada, October 27-30, 2019 (2019), T. Brecht and C. Williamson, Eds., ACM, pp. 114–130.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "MariaDB. Innodb undo log. https://mariadb .com/kb/en/ innodb-undo-log/. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Meng, R., Pîrlea, G., Roychoudhury, A., and Sergey, I. Greybox fuzzing of distributed systems. In Proceedings of the 2023 ACMSIGSAC Conference on Computer and Communications Security (2023), pp. 1615–1629.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Musuvathi, M., Park, D. Y. W., Chou, A., Engler, D. R., and Dill, D. L. CMC: A pragmatic approach to model checking real code. In 5th Symposium on Operating System Design and Implementation (OSDI 2002), Boston, Massachusetts, USA, December 9-11, 2002 (2002), D. E. Culler and P. Druschel, Eds., USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "MySQL. Mysql 8.0: New lock free, scalable wal design. https://dev .mysql.com/blog-archive/mysql-8-0-newlock-free-scalable-wal-design/. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Chaos monkey. https://netflix .github.io/chaosmonkey/. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Oracle. Mysql bug list. https://bugs .mysql.com/, 1 2014. Accessed: September 17, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Oracle. Mysql test, Today.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Pillai, T. S., Chidambaram, V., Alagappan, R., Al-Kiswany, S., Arpaci-Dusseau, A. C., and Arpaci-Dusseau, R. H. All file systems are not created equal: On the complexity of crafting {Crash-Consistent}applications. In 11th USENIX Symposium on Operating Systems Design and Implementation (OSDI 14) (2014), pp. 433–448.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Post, G. V. Database management systems. PHI Learning Pvt. Limited, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Qiao, J. Apache iotdb: Database for internet of things, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Qiao, J. Apache iotdb github, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Rigger, M., and Su, Z. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (2020), pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 48,
+      "text": "Rigger, M., and Su, Z. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 49,
+      "text": "Rigger, M., and Su, Z. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20) (2020), pp. 667– 682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "Seltenreich, A., Tang, B., and Mullender, S. Sqlsmith: a random sql query generator.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Song, J., Dou, W., Cui, Z., Dai, Q., Wang, W., Wei, J., Zhong, H., and Huang, T. Testing database systems via differential query execution. In 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE) (2023), IEEE, pp. 2072– 2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Tang, X., Wu, S., Zhang, D., Li, F., and Chen, G. Detecting logic bugs of join optimizations in dbms. Proceedings of the ACM on Management of Data 1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "TaosData. Tdengine github, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "TaosData. Tdengine website, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Wang, M., Wu, Z., Xu, X., Liang, J., Zhou, C., Zhang, H., and Jiang, Y. Industry practice of coverage-guided enterprise-level dbms fuzzing. In on Software Engineering: Software Engineering in Practice (ICSESEIP) (2021), IEEE, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Wu, Z., Liang, J., Fu, J., Wang, M., and Jiang, Y. Puppy: Finding performance degradation bugs in dbmss via limitedoptimization plan construction. In 2025 IEEE/ACM 47th International Conference on Software Engineering (ICSE) (2024), IEEE Computer Society, pp. 560–571.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Wu, Z., Liang, J., Fu, J., Wang, M., and Jiang, Y. Hulk: Exploring data-sensitive performance anomalies in dbmss via data-driven analysis. Proceedings of the ACM on Software Engineering 2, ISSTA (2025), 2181–2202.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Wu, Z., Liang, J., Wang, M., Zhou, C., and Jiang, Y. Unicorn: Detect runtime errors in time-series databases with hybrid input synthesis. In Symposium on Software Testing and Analysis (ISSTA’22) (2022). 683 Fawkes: Finding Data Durability Bugs in DBMSs via Recovered Data State Verification SOSP ’25, October 13–16, 2025, Seoul, Republic of Korea",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Yang, J., Chen, T., Wu, M., Xu, Z., Liu, X., Lin, H., Yang, M., Long, F., Zhang, L., and Zhou, L. MODIST: transparent model checking of unmodified distributed systems. In Proceedings of the 6th USENIX Symposium on Networked Systems Design and Implementation, NSDI 2009, April 22-24, 2009, Boston, MA, USA (2009), J. Rexford and E. G. Sirer, Eds., USENIX Association, pp. 213–228.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Zhang, C., and Rigger, M. Constant optimization driven database system testing. Proceedings of the ACM on Management of Data 3, 1 (2025), 1–24.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 61,
+      "text": "Zheng, M., Tucek, J., Huang, D., Qin, F., Lillibridge, M., Yang, E. S., Zhao, B. W., and Singh, S. Torturing databases for fun and profit. In 11th USENIX Symposium on Operating Systems Design and Implementation (OSDI 14) (2014), pp. 449– 464.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Zheng, W., Tu, S., Kohler, E., and Liskov, B. Fast databases with fast durability and recovery through multicore parallelism. In 11th USENIX Symposium on Operating Systems Design and Implementation (OSDI 14) (2014), pp. 465–477.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Zhong, R., Chen, Y., Hu, H., Zhang, H., Lee, W., and Wu, D. Squirrel: Testing database management systems with language validity and coverage feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020 (2020). 684",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 11,
+      "text": "Ba, J., and Rigger, M. Keep it simple: Testing databases via differential query plans. Proceedings of the ACM on Management of Data 2, 3 (2024), 1–26.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "Fu, J., Liang, J., Wu, Z., Zhao, Y., Li, S., and Jiang, Y. Understanding and detecting sql function bugs: Using simple boundary arguments to trigger hundreds of dbms bugs. In Proceedings of the Twentieth European Conference on Computer Systems (2025), pp. 1061–1076.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 26,
+      "text": "Jiang, Z.-M., Liu, S., Rigger, M., and Su, Z. Detecting transactional bugs in database engines via {graph-based}oracle 682 SOSP ’25, October 13–16, 2025, Seoul, Republic of Korea Zhiyong Wu, Jie Liang*, Jingzhou Fu, Wenqian Deng, and Yu Jiang* construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23) (2023), pp. 397–417.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 47,
+      "text": "Rigger, M., and Su, Z. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (2020), pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 48,
+      "text": "Rigger, M., and Su, Z. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 49,
+      "text": "Rigger, M., and Su, Z. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20) (2020), pp. 667– 682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 60,
+      "text": "Zhang, C., and Rigger, M. Constant optimization driven database system testing. Proceedings of the ACM on Management of Data 3, 1 (2025), 1–24.",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLancer [47]) to detect SQL correctness bugs.",
+      "context_before": "ly targets DDBs in single-node DBMSs by injecting faults at filesystem or kernel-level code regions, its fault-triggering strategy and data graph validation can also benefit other DBMS testing tools. First, the injection and triggering strategies can also be applied to distributed DBMSs to enhance durability. Second, integrating Fawkes ’s functionality-guided workload generation with grammarbased SQL generators (e.g., Sqirrel [63]) would allow them to focus on durability-critical paths for triggering DDBs. Finally, the data graph validation could be adopted by metamorphic testing methods (e.g.",
+      "context_after": "Overhead of Fawkes .Fawkes continuously injects faults during the SQL execution and verifies the recovery DBMS state to detect DDBs with checkpoint-based data graph analysis, which may import overhead. We accessed Fawkes ’s overhead by disabling itscontinuous fault injection (Fawkes𝑖𝑛𝑗𝑒𝑐𝑡𝑖𝑜𝑛−) and data graph analysis (Fawkes𝑔𝑟𝑎𝑝ℎ−) separately, measuring cases executed and DDBs found over a 24-hour experiment on 8 DBMSs. Table 7: Overhead Evaluation of Fawkes in 24 hours. Fawkes Fawkes𝑖𝑛𝑗𝑒𝑐𝑡𝑖𝑜𝑛−Fawkes𝑔𝑟𝑎𝑝ℎ− Test Cases Executed 1,434,559 4,070,465 1,475,236 DDBs Found 12 0 3 Table 7 shows that",
+      "section": null,
+      "page": 12,
+      "char_offset": 66138,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 49,63,58,29,30,55,57,56,20,48]",
+      "technique": "pqs",
+      "sentence": ", crash, logic bug, and performance bug) [ 49,63,58,29,30,55,57,56,20,48].",
+      "context_before": "reasing the time per checkpoint) slows test case execution and thus reduces the pace of bug discovery. This occurs primarily because higher time per checkpoint leads to higher overhead for Fawkes to rectify the data graph upon DBMS crashes, consequently reducing the total number of executed test cases and detected DDBs. In the evaluation, we set the DBMS’s default tuning parameters for experiments. 8 Related Work DBMS Testing. Database systems are complex and are prone to various issues. To improve the reliability of DBMS, a variety of DBMS testing tools have been developed to detect bugs(e.g.",
+      "context_after": "For crash detection, SQLsmith[ 50] uses a generative approach to discover crash issues by generating a large number of test cases. Similarly, Sqirrel [63],Lego [29], Ratel [ 55], and Griffin [19] apply coverage feedback techniques to trigger crashes. Unicorn [58] designs time-series mutation to test time-series databases. For logic bug detection, SQLancer[ 49] adopts a metamorphic testing approach[ 47,48,11,60] to detect logical errors in DBMS. DQE[ 51] extends this idea by verifying whether different SQL queries with equivalent predicates access consistent row sets. Mozi [31] proposes a conf",
+      "section": null,
+      "page": 12,
+      "char_offset": 68409,
+      "cited_reference": {
+        "number": 49,
+        "text": "Rigger, M., and Su, Z. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20) (2020), pp. 667– 682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For logic bug detection, SQLancer[ 49] adopts a metamorphic testing approach[ 47,48,11,60] to detect logical errors in DBMS.",
+      "context_before": "ts. 8 Related Work DBMS Testing. Database systems are complex and are prone to various issues. To improve the reliability of DBMS, a variety of DBMS testing tools have been developed to detect bugs(e.g., crash, logic bug, and performance bug) [ 49,63,58,29,30,55,57,56,20,48]. For crash detection, SQLsmith[ 50] uses a generative approach to discover crash issues by generating a large number of test cases. Similarly, Sqirrel [63],Lego [29], Ratel [ 55], and Griffin [19] apply coverage feedback techniques to trigger crashes. Unicorn [58] designs time-series mutation to test time-series databases.",
+      "context_after": "DQE[ 51] extends this idea by verifying whether different SQL queries with equivalent predicates access consistent row sets. Mozi [31] proposes a configuration-based equivalence transformation framework that leverages DBMS-specific optimizations to expose hidden logic bugs. TQS[ 52] decomposes wide tables into smaller ones and synthesizes join queries, using the original table as a ground truth reference. TxCheck [ 26] constructs semantically equivalent test cases based on fine-grained statement-level dependencies in transactions to detect isolation bugs in DBMS. PINOLO [ 23] introduces a res",
+      "section": null,
+      "page": 12,
+      "char_offset": 68807,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 26]",
+      "technique": null,
+      "sentence": "TxCheck [ 26] constructs semantically equivalent test cases based on fine-grained statement-level dependencies in transactions to detect isolation bugs in DBMS.",
+      "context_before": "n [58] designs time-series mutation to test time-series databases. For logic bug detection, SQLancer[ 49] adopts a metamorphic testing approach[ 47,48,11,60] to detect logical errors in DBMS. DQE[ 51] extends this idea by verifying whether different SQL queries with equivalent predicates access consistent row sets. Mozi [31] proposes a configuration-based equivalence transformation framework that leverages DBMS-specific optimizations to expose hidden logic bugs. TQS[ 52] decomposes wide tables into smaller ones and synthesizes join queries, using the original table as a ground truth reference.",
+      "context_after": "PINOLO [ 23] introduces a result-set containment approach, generating queries whose results should be supersets or subsets of a reference query; discrepancies are used to identify violations of expected result relationships. Zheng et al. propose a prototype that detects potential violations of the ACID properties in DBMS by focusing on simulating power failures [ 61]. For performance bug detection, Apollo [28] utilizes differential testing on multiple versions to test regression testing. Amoeba [32] generates equivalent queries and compares their response time to find performance issues in DB",
+      "section": null,
+      "page": 12,
+      "char_offset": 69341,
+      "cited_reference": {
+        "number": 26,
+        "text": "Jiang, Z.-M., Liu, S., Rigger, M., and Su, Z. Detecting transactional bugs in database engines via {graph-based}oracle 682 SOSP ’25, October 13–16, 2025, Seoul, Republic of Korea Zhiyong Wu, Jie Liang*, Jingzhou Fu, Wenqian Deng, and Yu Jiang* construction. In 17th USENIX Symposium on Operating Syst",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:20:36Z",
+    "is_model_written": true,
+    "summary": "Fawkes studies and detects data durability bugs, where committed data does not survive faults such as power failures. The authors examine 43 such bugs across four DBMSs, finding they surface as data loss, inconsistency, log corruption or unavailability, usually stem from flawed durability and recovery mechanisms, and are triggered by faults during filesystem or kernel operations. Fawkes combines context-aware fault injection, functionality-guided triggering and checkpoint-based data graph verification, finding 48 previously unknown durability bugs.",
+    "narrative": "SQLancer is cited as the established means of detecting SQL correctness bugs through metamorphic testing -- the class of testing Fawkes distinguishes itself from, since durability bugs survive any query-level oracle.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "definition",
+      "M4": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3744916_3773102.json b/_data/papers/paper_doi_10_1145_3744916_3773102.json
new file mode 100644
index 0000000..9c04e06
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3744916_3773102.json
@@ -0,0 +1,781 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:06:15Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3744916.3773102",
+    "title": "Locus: Agentic Predicate Synthesis for Directed Fuzzing",
+    "authors": [
+      "Jie Zhu",
+      "Chihao Shen",
+      "Ziyang Li",
+      "Jiahao Yu",
+      "Yizheng Chen",
+      "Kexin Pei"
+    ],
+    "year": 2025,
+    "venue": "arXiv.org",
+    "doi": "10.1145/3744916.3773102",
+    "arxiv_id": "2508.21302",
+    "s2_paper_id": "1ea85ffff5bc84a288f30b95c2f2d44f2df5e633",
+    "url": "https://doi.org/10.1145/3744916.3773102",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2508.21302",
+      "retrieved_at": "2026-09-07T17:06:15Z",
+      "chars": 83566,
+      "content_sha256": "sha256:3a88df3032c171d938d1cad26200b6a0f29d01bd91f0649b0d13840e04ab72fc"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2925
+      },
+      {
+        "number": "2",
+        "title": "Overview",
+        "start": 10338
+      },
+      {
+        "number": "2.1",
+        "title": "Directed Fuzzing",
+        "start": 10542
+      },
+      {
+        "number": "2.2",
+        "title": "Motivating Example",
+        "start": 14347
+      },
+      {
+        "number": "3",
+        "title": "Methodology",
+        "start": 21999
+      },
+      {
+        "number": "3.1",
+        "title": "Formalization",
+        "start": 22712
+      },
+      {
+        "number": "3.2",
+        "title": "Agent Toolset",
+        "start": 26240
+      },
+      {
+        "number": "3.3",
+        "title": "Synthesis",
+        "start": 28901
+      },
+      {
+        "number": "4",
+        "title": "Evaluation",
+        "start": 35453
+      },
+      {
+        "number": "4.1",
+        "title": "Setup",
+        "start": 38593
+      },
+      {
+        "number": "4.2",
+        "title": "RQ1: Vulnerabilities Reproduction",
+        "start": 42193
+      },
+      {
+        "number": "4.3",
+        "title": "RQ2: Cost Analysis",
+        "start": 43080
+      },
+      {
+        "number": "4.4",
+        "title": "RQ3: Ablations",
+        "start": 45434
+      },
+      {
+        "number": "4.5",
+        "title": "RQ4: Detecting New Vulnerabilities",
+        "start": 48122
+      },
+      {
+        "number": "4.6",
+        "title": "Case Study",
+        "start": 50196
+      },
+      {
+        "number": "5",
+        "title": "Related Work",
+        "start": 57550
+      },
+      {
+        "number": "6",
+        "title": "Threats to Validity",
+        "start": 61052
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 62670
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Anthropic. 2024. Claude Code. https://claude.com/product/claude-code",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Cornelius Aschermann, Sergej Schumilo, et al .2020. Ijon: Exploring Deep State Spaces via Fuzzing. In. 1597–1612.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba, Marcel Böhme, et al .2022. Stateful Greybox Fuzzing. In31st USENIX Security Symposium (USENIX Security 22). 3255–3272.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Davide Balzarotti. 2021. The use of likely invariants as feedback for fuzzers. In 30th USENIX Security Symposium (USENIX Security 21).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Marcel Böhme, Bruno C d S Oliveira, and Abhik Roychoudhury. 2013. Regression tests to expose change interaction errors. InProceedings of the 2013 9th Joint Meeting on Foundations of Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Marcel Böhme, Van-Thuan Pham, et al .2017. Directed Greybox Fuzzing. In Proceedings of the 2017 ACM Sigsac Conference on Computer and Communications Security. 2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "David Brumley, Pongsin Poosankam, Dawn Song, and Jiang Zheng. 2008. Automatic patch-based exploit generation is possible: Techniques and implications. In.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Cristian Cadar, Daniel Dunbar, et al .2008. KLEE: Unassisted and Automatic Generation of High-Coverage Tests for Complex Systems Programs. In8th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2008, December 8-10, 2008, San Diego, California, USA, Proceedings, Richard Draves and Robbert van Renesse (Eds.). 209–224.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Sicong Cao, Biao He, et al .2023. Oddfuzz: Discovering java deserialization vulnerabilities via structure-aware directed greybox fuzzing. In2023 IEEE Symposium on Security and Privacy (SP). IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Saikat Chakraborty, Shuvendu K Lahiri, et al .2023. Ranking llm-generated loop invariants for program verification.arXiv preprint arXiv:2310.09342(2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Chuyang Chen, Brendan Dolan-Gavitt, and Zhiqiang Lin. 2025. ELFuzz: Efficient Input Generation via LLM-driven Synthesis Over Fuzzer Space.arXiv preprint arXiv:2506.10323(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Hongxu Chen, Yinxing Xue, et al .2018. Hawkeye: Towards a Desired Directed Grey-box Fuzzer. InProceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security. 2095–2108.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Peng Chen and Hao Chen. 2018. Angora: Efficient Fuzzing by Principled Search. In. 711–725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Tianyu Chen, Shuai Lu, et al .2024. Automated proof generation for rust code via self-evolution.arXiv preprint arXiv:2410.15756(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Yuanliang Chen, Yu Jiang, Fuchen Ma, Jie Liang, Mingzhe Wang, Chijin Zhou, Xun Jiao, and Zhuo Su. 2019. EnFuzz: Ensemble fuzzing with seed synchronization among diverse fuzzers. In28th USENIX Security Symposium (USENIX Security 19).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Maria Christakis, Peter Müller, and Valentin Wüstholz. 2016. Guiding dynamic symbolic execution toward unverified program executions. InProceedings of the 38th International Conference on Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Samuel Colvin. 2025.PydanticAI. https://ai.pydantic.dev/ Version 0.4.3.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "DARPA. 2024. DARPA AI Cyber Challenge. https://aicyberchallenge.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, et al .2023. Large Language Models Are ZeroShot Fuzzers: Fuzzing Deep-Learning Libraries via Large Language Models. In Proceedings of the 32nd ACM Sigsoft International Symposium on Software Testing and Analysis. 423–435.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Yinlin Deng, Chunqiu Steven Xia, Chenyuan Yang, Shizhuo Dylan Zhang, Shujing Yang, and Lingming Zhang. 2024. Large language models are edge-case generators: Crafting unusual programs for fuzzing deep learning libraries. In Proceedings of the 46th IEEE/ACM international conference on software engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Yangruibo Ding, Yanjun Fu, Omniyyah Ibrahim, Chawin Sitawarin, Xinyun Chen, Basel Alomair, David Wagner, Baishakhi Ray, and Yizheng Chen. 2024. Vulnerability detection with code language models: How far are we?arXiv preprint arXiv:2403.18624(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Yangruibo Ding, Yanjun Fu, Omniyyah Ibrahim, Chawin Sitawarin, Xinyun Chen, Basel Alomair, David Wagner, Baishakhi Ray, and Yizheng Chen. 2024. Vulnerability Detection with Code Language Models: How Far Are We?arXiv preprint arXiv:2403.18624(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Yangruibo Ding, Jinjun Peng, et al .2024. Semcoder: Training code language models with comprehensive semantics reasoning.Advances in Neural Information Processing Systems37 (2024), 60275–60308.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Yangruibo Ding, Benjamin Steenhoek, et al .2024. TRACED: Execution-aware Pre-training for Source Code. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Zhengjie Du, Yuekang Li, et al .2022. WindRanger: A Directed Greybox Fuzzer Driven by Deviation Basic Blocks. InProceedings of the 44th International Conference on Software Engineering. 2440–2451.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Rafael Dutra, Rahul Gopinath, and Andreas Zeller. 2023. Formatfuzzer: Effective fuzzing of binary file formats.ACM Transactions on Software Engineering and Methodology33, 2 (2023), 1–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Andrea Fioraldi, Daniele Cono D’Elia, et al .2021. The Use of Likely Invariants as Feedback for Fuzzers. In30th USENIX Security Symposium (USENIX Security 21). 2829–2846.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Andrea Fioraldi, Dominik Maier, et al .2020. AFL++: Combining Incremental Steps of Fuzzing Research. In14th Usenix Workshop on Offensive Technologies (Woot 20).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Code Intelligence GmbH and Contributors. 2020.Jazzer. https://github.com/ CodeIntelligenceTesting/jazzer",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Linyuan Gong, Sida Wang, Mostafa Elhoushi, and Alvin Cheung. 2024. Evaluation of llms on syntax-aware code fill-in-the-middle tasks.arXiv preprint arXiv:2403.04814(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Peter Goodman. 2025.Multiplier. https://github.com/trailofbits/multiplier Version 1705339.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Google and Contributors. 2020.Atheris. https://github.com/google/atheris",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Google DeepMind. 2024. Gemini 2.0 Flash. https://deepmind.google/ technologies/gemini/flash/. Accessed: 2025-03-29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Alex Gu, Baptiste Rozière, et al .2024. CRUXEval: A Benchmark for Code Reasoning, Understanding and Execution. InProceedings of the 41st International Conference on Machine Learning. 16568–16621.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Ahmad Hazimeh, Adrian Herrera, et al .2020. Magma: A Ground-Truth Fuzzing Benchmark. InProceedings of the ACM on Measurement and Analysis of Computing Systems, Vol. 4. 1–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Christian Holler, Kim Herzig, and Andreas Zeller. 2012. Fuzzing with code fragments. In21st USENIX Security Symposium (USENIX Security 12).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Jie Hu, Qian Zhang, and Heng Yin. 2023. Augmenting greybox fuzzing with generative ai.arXiv preprint arXiv:2306.06782(2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Heqing Huang et al .2022. BEACON: Directed Grey-Box Fuzzing with Provable Path Pruning. In. 36–50.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Heqing Huang, Peisen Yao, et al .2024. Titan: Efficient Multi-target Directed Greybox Fuzzing. In. 1849– 1864.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Heqing Huang, Anshunkang Zhou, et al .2024. Everything Is Good for Something: Counterexample-Guided Directed Fuzzing via Likely Invariant Inference. In. 1956–1973.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Kyriakos Ispoglou, Daniel Austin, Vishwath Mohan, and Mathias Payer. 2020. {FuzzGen}: Automatic fuzzer generation. In29th USENIX Security Symposium (USENIX Security 20).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Zongze Jiang, Ming Wen, Jialun Cao, Xuanhua Shi, and Hai Jin. 2024. Towards Understanding the Effectiveness of Large Language Models on Directed Test Input Generation. InProceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Carlos E Jimenez, John Yang, et al .2024. SWE-bench: Can Language Models Resolve Real-world Github Issues?. InThe Twelfth International Conference on Learning Representations.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Tae Eun Kim, Jaeseung Choi, et al .2023. DAFL: Directed Grey-Box Fuzzing Guided by Data Dependency. In32nd USENIX Security Symposium (USENIX Security 23). 4931–4948.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "George Klees, Andrew Ruef, et al .2018. Evaluating Fuzz Testing. InProceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security. 2123–2138.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Andrei Kozyrev, Gleb Solovev, Nikita Khramov, and Anton Podkopaev. 2024. CoqPilot, a plugin for LLM-based generation of proofs. InProceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Gwangmu Lee, Woochul Shim, et al .2021. Constraint-Guided Directed Greybox Fuzzing. In30th Usenix Security Symposium (Usenix Security 21). 3559–3576.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Haonan Li, Yu Hao, Yizhuo Zhai, and Zhiyun Qian. 2024. Enhancing static analysis for practical bug detection: An llm-integrated approach.Proceedings of the ACM on Programming Languages8, OOPSLA1 (2024), 474–499.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Haonan Li, Hang Zhang, Kexin Pei, and Zhiyun Qian. 2025. The Hitchhiker’s Guide to Program Analysis, Part II: Deep Thoughts by LLMs.arXiv preprint arXiv:2504.11711(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Penghui Li, Wei Meng, et al .2024. SDFuzz: Target States Driven Directed Fuzzing. In33rd Usenix Security Symposium (Usenix Security 24). 2441–2457.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Yuekang Li, Bihuan Chen, Mahinthan Chandramohan, Shang-Wei Lin, Yang Liu, and Alwen Tiu. 2017. Steelix: program-state based binary fuzzing. InProceedings of the 2017 11th joint meeting on foundations of software engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Ziyang Li, Saikat Dutta, and Mayur Naik. 2024. Llm-assisted static analysis for detecting security vulnerabilities.arXiv preprint arXiv:2405.17238(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Hongliang Liang, Lin Jiang, Lu Ai, and Jinyi Wei. 2020. Sequence directed hybrid fuzzing. In Evolution and Reengineering (SANER). IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "libarchive contributors. 2025.libarchive: Multi-format archive and compression library. https://www.libarchive.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "libming contributors. 2025.libming: SWF (Flash) file creation library. https: //www.libming.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Dongge Liu, Oliver Chang, et al .2024. OSS-fuzz-gen: Automated Fuzz Target Generation. Locus: Agentic Predicate Synthesis for Directed Fuzzing ICSE ’26, April 12–18, 2026, Rio de Janeiro, Brazil",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Jiawei Liu, Songrun Xie, Junhao Wang, Yuxiang Wei, Yifeng Ding, and Lingming Zhang. 2024. Evaluating language models for efficient code generation.arXiv preprint arXiv:2408.06450(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Zhe Liu et al .2024. Testing the Limits: Unusual Text Inputs Generation for Mobile App Crash Detection with Large Language Model. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Minghai Lu, Benjamin Delaware, and Tianyi Zhang. 2024. Proof automation with large language models. InProceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Changhua Luo, Wei Meng, et al .2023. SelectFuzz: Efficient Directed Fuzzing with Selective Path Exploration. In (Sp). 2693–2707.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Chenyang Lyu, Shouling Ji, et al .2019. MOPT: Optimized Mutation Scheduling for Fuzzers. In28th USENIX Security Symposium (USENIX Security 19). 1949– 1966.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Yunlong Lyu, Yuxuan Xie, et al .2024. Prompt Fuzzing for Fuzz Driver Generation. InProceedings of the 2024 on ACMSIGSAC Conference on Computer and Communications Security. 3793–3807.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "H. B. Mann and D. R. Whitney. 1947. On a Test of Whether One of Two Random Variables Is Stochastically Larger than the Other.Annals of Mathematical Statistics18, 1 (March 1947), 50–60.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Paul Dan Marinescu and Cristian Cadar. 2013. KATCH: High-coverage testing of software patches. InProceedings of the 2013 9th Joint Meeting on Foundations of Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Ruijie Meng, Zhen Dong, et al .2022. Linear-Time Temporal Logic Guided Greybox Fuzzing. InProceedings of the 44th International Conference on Software Engineering. 1343–1355.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Charalambos Mitropoulos et al .2023. Syntax-aware mutation for testing the solidity compiler. InEuropean Symposium on Research in Computer Security. Springer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Aniruddhan Murali, Noble Mathews, et al .2024. Fuzzslice: Pruning false positives in static analysis warnings through function-level fuzzing. InProceedings of the 46th IEEE/ACM International Conference on Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Manh-Dung Nguyen, Sébastien Bardin, et al .2020. Binary-level directed fuzzing for{use-after-free}vulnerabilities. In23rd International Symposium on Research in Attacks, Intrusions and Defenses (RAID 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Laurent Simon, and Hayawardh Vijayakumar. 2019. Fuzzfactory: domain-specific fuzzing with waypoints.Proceedings of the ACM on Programming Languages3, OOPSLA (2019), 1–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Jibesh Patra and Michael Pradel. 2016. Learning to fuzz: Applicationindependent fuzz testing with probabilistic, generative models of input data. TU Darmstadt, Department of Computer Science, Tech. Rep. TUD-CS-2016-14664 (2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Paul Gauthier. 2024. Aider, AI pair programming in your terminal. https: //aider.chat.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Hui Peng, Yan Shoshitaishvili, et al .2018. T-Fuzz: Fuzzing by Program Transformation. In. 697–710.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages(2020).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 74,
+      "text": "Niklas Risse and Marcel Böhme. 2024. Uncovering the limits of machine learning for automatic vulnerability detection. In33rd USENIX Security Symposium (USENIX Security 24).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Niklas Risse, Jing Liu, et al .2025. Top Score on the Wrong Exam: On Benchmarking in Machine Learning for Vulnerability Detection. InProceedings of the ACM on Software Engineering, Vol. 2. 388–410.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Pranab Sahoo, Prabhash Meharia, et al .2024. A comprehensive survey of hallucination in large language, image, video and audio foundation models. arXiv preprint arXiv:2405.09589(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitriy Vyukov. 2012.{AddressSanitizer}: A fast address sanity checker. In2012 USENIX annual technical conference (USENIX ATC 12). 309–318.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Abhishek Shah, Dongdong She, et al .2022. MC2: Rigorous and Efficient Directed Greybox Fuzzing. InProceedings of the 2022 ACMSIGSAC Conference on Computer and Communications Security. 2595–2609.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Dongdong She, Adam Storek, et al .2024. FOX: Coverage-guided Fuzzing as Online Stochastic Control. InProceedings of the 2024 on ACMSIGSAC Conference on Computer and Communications Security. 765–779.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Gabriel Sherman and Stefan Nagy. 2025. No Harness, No Problem: Oracleguided Harnessing for Auto-Generating CAPI Fuzzing Harnesses. InIEEE/ACM International Conference on Software Engineering (ICSE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "Wenxuan Shi, Yunhang Zhang, et al .2024. Harnessing Large Language Models for Seed Generation in Greybox Fuzzing.arXiv preprint arXiv:2411.18143(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Prashast Srivastava, Stefan Nagy, et al .2022. One Fuzz Doesn’t Fit All: Optimizing Directed Fuzzing via Target-tailored Program State Restriction. In Proceedings of the 38th Annual Computer Security Applications Conference. 388– 399.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Prashast Srivastava, Stefan Nagy, Matthew Hicks, Antonio Bianchi, and Mathias Payer. 2022. One fuzz doesn’t fit all: Optimizing directed fuzzing via targettailored program state restriction. InProceedings of the 38th Annual Computer Security Applications Conference.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Yulei Sui and Jingling Xue. 2016. SVF: Interprocedural Static Value-Flow Analysis in LLVM. InProceedings of the 25th International Conference on Compiler Construction. 265–266.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "Xin Tan, Yuan Zhang, et al .2023. SyzDirect: Directed Greybox Fuzzing for Linux Kernel. InProceedings of the 2023 ACMSIGSAC Conference on Computer and Communications Security. 1630–1644.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "{DeepSeek-AI}, Daya Guo, et al .2025. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning. arXiv:2501.12948 [cs]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "David Trabish, Andrea Mattavelli, et al .2018. Chopped Symbolic Execution. In Proceedings of the 40th International Conference on Software Engineering. 350– 360.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "Aaron Turner and Contributors. 2001.Tcpreplay. https://tcpreplay.app/ Replay captured network traffic. Accessed: 2025-11-26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "VideoLAN. 2025.VLC media player. https://www.videolan.org/vlc/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "Dmitry Vyukov. 2015.go-fuzz. https://github.com/dvyukov/go-fuzz",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "Chengpeng Wang, Wuqi Zhang, et al .2024. LLMDFA: Analyzing Dataflow in Code with Large Language Models. InAdvances in Neural Information Processing Systems, A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang (Eds.), Vol. 37. 131545–131574.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "Daimeng Wang, Zheng Zhang, et al .2021. SyzVegas: Beating Kernel Fuzzing Odds with Reinforcement Learning. In30th USENIX Security Symposium (USENIX Security 21). 2741–2758.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "Junjie Wang, Yuhan Ma, et al .2025. PatchFuzz: Patch Fuzzing for JavaScript Engines.arXiv preprint arXiv:2505.00289(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "Yanhao Wang, Xiangkun Jia, Yuwei Liu, Kyle Zeng, Tiffany Bao, Dinghao Wu, and Purui Su. 2020. Not All Coverage Measurements Are Equal: Fuzzing by Coverage Accounting for Input Prioritization.. InNDSS.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "Felix Weissberg, Jonas Möller, et al .2024. SoK: Where to Fuzz? Assessing Target Selection Methods in Directed Fuzzing. InProceedings of the 19th ACM Asia Conference on Computer and Communications Security. 1539–1553.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "Chunqiu Steven Xia, Matteo Paltenghi, et al .2024. Fuzz4ALL: Universal Fuzzing with Large Language Models. In on Software Engineering (ICSE). 1547–1559.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "Hanxiang Xu, Yanjie Zhao, et al .2025. Directed Greybox Fuzzing via Large Language Model. arXiv:2505.03425 [cs]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 98,
+      "text": "Chenyuan Yang, Yinlin Deng, et al .2024. WhiteFox: White-Box Compiler Fuzzing Empowered by Large Language Models. InObject-Oriented Programming, Systems, Languages, and Applications, Vol. 8. 709–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 99,
+      "text": "Chenyuan Yang, Zijie Zhao, et al .2025. KNighter: Transforming Static Analysis with LLM-Synthesized Checkers. arXiv:2503.09002 [cs]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 100,
+      "text": "Chenyuan Yang, Zijie Zhao, and Lingming Zhang. 2023. Kernelgpt: Enhanced kernel fuzzing via large language models.arXiv preprint arXiv:2401.00563(2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 101,
+      "text": "John Yang, Carlos E. Jimenez, et al .2024. SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering. InAdvances in Neural Information Processing Systems, A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang (Eds.), Vol. 37. 50528–50652.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 102,
+      "text": "Yupeng Yang, Shenglong Yao, Jizhou Chen, and Wenke Lee. 2025. Hybrid Language Processor Fuzzing via LLM-Based Constraint Solving. In34th USENIX Security Symposium (USENIX Security 25).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 103,
+      "text": "Shunyu Yao et al .2023. ReAct: Synergizing Reasoning and Acting in Language Models. InThe Eleventh International Conference on Learning Representations.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 104,
+      "text": "Zijun Yao, Yantao Liu, et al .2025. Are Reasoning Models More Prone to Hallucination?arXiv preprint arXiv:2505.23646(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 105,
+      "text": "Michal Zalewski. 2020. American Fuzzy Lop.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 106,
+      "text": "Cen Zhang, Yuekang Li, et al .2023. Automata-Guided Control-Flow-Sensitive Fuzz Driver Generation.. InUSENIX Security Symposium.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 107,
+      "text": "Cen Zhang, Yaowen Zheng, et al .2024. How effective are they? Exploring large language model based fuzz driver generation. InProceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 108,
+      "text": "Cen Zhang, Yaowen Zheng, et al .2024. How Effective Are They? Exploring Large Language Model Based Fuzz Driver Generation. InProceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. 1223–1235.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 109,
+      "text": "Hongxiang Zhang, Yuyang Rong, Yifeng He, and Hao Chen. 2024. Llamafuzz: Large language model enhanced greybox fuzzing.arXiv preprint arXiv:2406.07714(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 110,
+      "text": "Qiang Zhang et al .2024. ECG: Augmenting Embedded Operating System Fuzzing via LLM-Based Corpus Generation.IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems43, 11 (2024), 4238–4249.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 111,
+      "text": "Yuntong Zhang, Haifeng Ruan, et al .2024. AutoCodeRover: Autonomous Program Improvement. InProceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. 1592–1604.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 73,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages(2020).",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 2,26,36,44,45,66,73,80,\n106]",
+      "technique": "tlp",
+      "sentence": "However, such feedback is sometimes too sparse or indirect to reliably measure the progress, especially when there is a long chain ofimplicitpreconditions guarding the target program states [ 2,26,36,44,45,66,73,80, 106].",
+      "context_before": "et, it has found numerous impactful zero-day vulnerabilities [9, 53, 68, 78, 93]. DGF is challenging, as the target program states are often deeply nested in the program, while the search space introduced by the complexity of real-world software is prohibitively large. To speed up the search and schedule promising inputs, most existing techniques rely on metrics based on control flow proximity,e.g.,the distance to target location in the control flow graph, or heuristics based on the semantics of the branch predicates [ 13,51,78],e.g.,the target state if x==42 has the distance metric of |𝑥−42|.",
+      "context_after": "For example, triggering CVE-2018-13785 in libpng requires a PNG file to satisfy a precise sequence of preconditions,i.e.,valid signature, correct chunk layout, specific IHDR fields (e.g.,bit depth, color type), and a magic image width ( 0x55555555 ), to trigger an integer overflow [ 44], while the predicates to explicitly check these conditions are largely absent in the code to provide an incremental progress guidance. To capture the intricate feedback to improve search efficiency, more advanced approaches identify progress-capturing constraints in the program to drive execution towards satis",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 4181,
+      "cited_reference": {
+        "number": 73,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages(2020).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "Locus improves directed fuzzing, where the goal is an input reaching a specified program state. The authors argue branch distances do not precisely capture progress toward a deeply nested target and manual constraints do not generalise. Locus synthesises predicates marking semantically meaningful intermediate states as milestones, instrumented into the program to reject executions unlikely to reach the target. It sped up eight fuzzers by 41.6x on average and found nine unpatched bugs.",
+    "narrative": "One citation, in a discussion of feedback signals being too sparse or indirect to measure progress.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3749186.json b/_data/papers/paper_doi_10_1145_3749186.json
new file mode 100644
index 0000000..1587444
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3749186.json
@@ -0,0 +1,983 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:05Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3749186",
+    "title": "Testing Graph Databases with Synthesized Queries",
+    "authors": [
+      "Zijing Yin",
+      "Si Liu",
+      "David A. Basin"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Manag. Data",
+    "doi": "10.1145/3749186",
+    "arxiv_id": null,
+    "s2_paper_id": "00940ec18283156770edb798589bf89c07d7544c",
+    "url": "https://doi.org/10.1145/3749186",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3749186",
+      "retrieved_at": "2026-09-09T01:31:05Z",
+      "chars": 91902,
+      "content_sha256": "sha256:8dac5928ddcd8824db500a2514f3900c6311847bea186d63c91bd093fd1a5cbd"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 26,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1913
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 12170
+      },
+      {
+        "number": "2.1",
+        "title": "GDBs and Labeled Property Graphs",
+        "start": 12183
+      },
+      {
+        "number": "2.2",
+        "title": "The Cypher Query Language",
+        "start": 13526
+      },
+      {
+        "number": "3",
+        "title": "Graph Query Synthesis",
+        "start": 17284
+      },
+      {
+        "number": "3.1",
+        "title": "Overview",
+        "start": 17561
+      },
+      {
+        "number": "3.2",
+        "title": "Integrating Diverse Clauses",
+        "start": 21907
+      },
+      {
+        "number": "3.3",
+        "title": "Building Cross-Step Dependencies",
+        "start": 25932
+      },
+      {
+        "number": "3.4",
+        "title": "Mutating Search Patterns",
+        "start": 31525
+      },
+      {
+        "number": "3.5",
+        "title": "Generating Complex Expressions",
+        "start": 38402
+      },
+      {
+        "number": "4",
+        "title": "Implementation",
+        "start": 43164
+      },
+      {
+        "number": "5",
+        "title": "Experiments",
+        "start": 46764
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Setup",
+        "start": 47428
+      },
+      {
+        "number": "5.2",
+        "title": "Discovering New Bugs",
+        "start": 49435
+      },
+      {
+        "number": "5.3",
+        "title": "A Closer Look at GQS",
+        "start": 56239
+      },
+      {
+        "number": "5.4",
+        "title": "Comparison with State-of-the-Art",
+        "start": 61082
+      },
+      {
+        "number": "6",
+        "title": "Related Work",
+        "start": 72263
+      },
+      {
+        "number": "7",
+        "title": "Discussion and Concluding Remark",
+        "start": 75538
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "ISO/IEC 39075:2024. Accessed in July, 2024. Information technology — Database languages — GQL. https://www.iso. org/standard/76120.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Atul Adya. 1999.Weak consistency: a generalized theory and optimistic implementations for distributed transactions. Ph. D. Dissertation. Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "John W. Backus. 1959. The syntax and semantics of the proposed international algebraic language of the Zurich ACM-GAMM Conference. InInformation Processing, Proceedings of the 1st International Conference on Information Processing, UNESCO, Paris 15-20 June 1959. UNESCO (Paris), 125–131.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Peter Bailis, Aaron Davidson, Alan D. Fekete, Ali Ghodsi, Joseph M. Hellerstein, and Ion Stoica. 2013. Highly Available Transactions: Virtues and Limitations.Proc. VLDB Endow.7, 3 (2013), 181–192.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Maciej Besta, Robert Gerstenberger, Emanuel Peter, Marc Fischer, Michał Podstawski, Claude Barthels, Gustavo Alonso, and Torsten Hoefler. 2023. Demystifying Graph Databases: Analysis and Taxonomy of Data Organization, System Designs, and Graph Queries.ACM Comput. Surv.56, 2, Article 31 (sep 2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Ranadeep Biswas and Constantin Enea. 2019. On the complexity of checking transactional consistency.Proc. ACM Program. Lang.3, OOPSLA (2019), 165:1–165:28. 3, No. 4 (SIGMOD), Article 268. Publication date: September 2025. Testing Graph Databases with Synthesized Queries 268:25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Angela Bonifati, George H. L. Fletcher, Hannes Voigt, and Nikolay Yakovets. 2018.Querying Graphs. Morgan & Claypool Publishers.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Tsong Yueh Chen, S. C. Cheung, and Siu-Ming Yiu. 2020. Metamorphic Testing: A New Approach for Generating Next Test Cases.CoRRabs/2002.12543 (2020). arXiv:2002.12543 https://arxiv.org/abs/2002.12543",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Yang Chen, Alex Groce, Chaoqiang Zhang, Weng-Keen Wong, Xiaoli Z. Fern, Eric Eide, and John Regehr. 2013. Taming compiler fuzzers. InPLDI ’13, Hans-Juergen Boehm and Cormac Flanagan (Eds.). ACM, 197–208.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Cypher. Accessed in July, 2024. https://neo4j.com/docs/cypher-manual/current/introduction/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "FalkorDB. Accessed in July, 2024. https://www.falkordb.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Shabnam Ghasemirad, Si Liu, Christoph Sprenger, Luca Multazzu, and David Basin. 2025. VerIso: Verifiable Isolation Guarantees for Database Transactions.Proc. VLDB Endow.18, 5 (2025), 1362–1375.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Long Gu, Si Liu, Tiancheng Xing, Hengfeng Wei, Yuxing Chen, and David Basin. 2024. IsoVista: Black-box Checking Database Isolation Guarantees.Proc. VLDB Endow.17, 12 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023. Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis. InUSENIX ATC ’23. USENIX Association, 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting Bugs in Cypher Graph Database Engines. InISSTA ’23. ACM, 163–174.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Kaile Huang, Si Liu, Zhenge Chen, Hengfeng Wei, David A. Basin, Haixiang Li, and Anqun Pan. 2023. Efficient Black-box Checking of Snapshot Isolation in Databases.Proc. VLDB Endow.16, 6 (2023), 1264–1276.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "JanusGraph. Accessed in April, 2025. https://janusgraph.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. InICSE ’24. ACM, Article 46.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. InOSDI ’23. 397–417.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting Logic Bugs in Database Engines via Equivalent Expression Transformation. InOSDI ’24. USENIX Association, 821–835.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. InISSTA ’23. ACM, 140–149.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 23,
+      "text": "Kyle Kingsbury and Peter Alvaro. 2020. Elle: Inferring Isolation Anomalies from Experimental Observations.Proc. VLDB Endow.14, 3 (2020), 268–280.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Kùzu. Accessed in July, 2024. https://kuzudb.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Shunyang Li, Zhengyi Yang, Xianhang Zhang, Wenjie Zhang, and Xuemin Lin. 2021. SQL2Cypher: Automated Data and Query Migration from RDBMS to GDBMS. InWISE 2021. Springer-Verlag, 510–517.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Jerry Liang. Accessed in April, 2025. openCypher Transpiler. https://github.com/microsoft/openCypherTranspiler.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Cypher Parser Library. Accessed in July, 2024. libcypher-parser. https://github.com/cleishm/libcypher-parser",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Si Liu, Long Gu, Hengfeng Wei, and David Basin. 2024. Plume: Efficient and Complete Black-box Checking of Weak Isolation Levels.Proc. ACM Program. Lang.8, OOPSLA2 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Si Liu, Luca Multazzu, Hengfeng Wei, and David A. Basin. 2024. NOC-NOC: Towards Performance-optimal Distributed Transactions.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Si Liu, Peter Csaba Ölveczky, Min Zhang, Qi Wang, and José Meseguer. 2019. Automatic Analysis of Consistency Properties of Distributed Transaction Systems in Maude. InTACAS 2019 (LNCS, Vol. 11428). Springer, 40–57.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Rupak Majumdar and Filip Niksic. 2017. Why is random testing effective for partition tolerance bugs?Proc. ACM Program. Lang.2, POPL, Article 46 (Dec. 2017), 24 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Valerio Malenchino. Accessed in July, 2024. GQL is Here: Your Cypher Queries in a GQL World. https://neo4j.com/ blog/cypher-gql-world/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Qiuyang Mang, Aoyang Fang, Boxi Yu, Hanfei Chen, and Pinjia He. 2024. Testing Graph Database Systems via Equivalent Query Rewriting. InICSE ’24. ACM, Article 143, 12 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "William M. McKeeman. 1998. Differential Testing for Software.Digit. Tech. J.10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Memgraph. Accessed in July, 2024. https://memgraph.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Jonathan Mendelson, Aaditya Naik, Mukund Raghothaman, and Mayur Naik. 2021. GENSYNTH: Synthesizing Datalog Programs without Language Bias.AAAI’2135, 7 (May 2021), 6444–6453.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Peter Müller and Natarajan Shankar. 2021. The First Fifteen Years of the Verified Software Project. InTheories of Programming: The Life and Works of Tony Hoare. ACM Books, Vol. 39. ACM / Morgan & Claypool, 93–124.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Neo4j. Accessed in April, 2025. SQL to Cypher translation. https://neo4j.com/docs/jdbc-manual/current/sql2cypher/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Neo4j. Accessed in July, 2024. https://neo4j.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "openCypher. Accessed in July, 2024. https://opencypher.org/. 3, No. 4 (SIGMOD), Article 268. Publication date: September 2025. 268:26 Zijing Yin, Si Liu, & David Basin",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "openCypher. Accessed in July, 2024. Cypher for Gremlin. https://github.com/opencypher/cypher-for-gremlin.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "openCypher. Accessed in July, 2024. Cypher Query Language Reference, Version 9. https://s3.amazonaws.com/artifacts. opencypher.org/openCypher9.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Philip Rathle and Brad Bebee. Accessed in July, 2024. GQL: The ISO standard for graphs has arrived. https://aws. amazon.com/blogs/database/gql-the-iso-standard-for-graphs-has-arrived/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "RedisGraph. Accessed in July, 2024. https://github.com/RedisGraph/RedisGraph.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InESEC/FSE ’20. ACM, 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 46,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (nov 2020), 30 pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. InOSDI ’20. USENIX Association, 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 48,
+      "text": "Marko A. Rodriguez. 2015. The Gremlin graph traversal machine and language (invited talk). InProceedings of the 15th Symposium on Database Programming Languages (DBPL 2015). ACM, 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. InICSE ’23. IEEE Press, 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Jiansen Song, Wensheng Dou, Yu Gao, Ziyu Cui, Yingying Zheng, Dong Wang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction.Proc. VLDB Endow. 17, 8 (may 2024), 1884–1897.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Andrei Stefanescu, Daejun Park, Shijiao Yuwen, Yilong Li, and Grigore Rosu. 2016. Semantics-based program verifiers for all languages. InOOPSLA 2016. ACM, 74–91.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Keita Takenouchi, Takashi Ishio, Joji Okada, and Yuji Sakata. 2021. PATSQL: Efficient Synthesis of SQL Queries from Example Tables with Quick Inference of Projected Columns.Proc. VLDB Endow.14, 11 (2021), 1937–1949.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Cheng Tan, Changgeng Zhao, Shuai Mu, and Michael Walfish. 2020. COBRA: Making Transactional Key-Value Stores Verifiably Serializable(OSDI ’20). USENIX Association, Article 4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting Logic Bugs of Join Optimizations in DBMS.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Aalok Thakkar, Aaditya Naik, Nathaniel Sands, Rajeev Alur, Mayur Naik, and Mukund Raghothaman. 2021. Exampleguided synthesis of relational queries. InPLDI ’21. ACM, 1110–1125.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Chenglong Wang, Alvin Cheung, and Rastislav Bodík. 2017. Synthesizing highly expressive SQL queries from input-output examples. InPLDI ’17. ACM, 452–466.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Qianqian Wang, Chris Parnin, and Alessandro Orso. 2015. Evaluating the usefulness of IR-based fault localization techniques. InISSTA ’15. ACM, 1–11.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Hengfeng Wei, Jiang Xiao, Na Yang, Si Liu, Zijing Yin, Yuxing Chen, and Anqun Pan. 2025. Boosting End-to-End Database Isolation Checking via Mini-Transactions. InICDE 2025. IEEE Computer Society, 3998–4010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Jian Zhang, Ye Ji, Shuai Mu, and Cheng Tan. 2023. Viper: A Fast Snapshot Isolation Checker. InEuroSys ’23. ACM, 654–671.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Sai Zhang and Yuyin Sun. 2013. Automatically synthesizing SQL queries from input-output examples. InASE ’13. IEEE, 224–234.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding bugs in Gremlin-based graph database systems via randomized differential testing. InISSTA ’22. ACM, 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Zeyang Zhuang, Penghui Li, Pingchuan Ma, Wei Meng, and Shuai Wang. 2024. Testing Graph Database Systems via Graph-Aware Metamorphic Relations.Proc. VLDB Endow.17, 4 (mar 2024), 836–848. Received January 2025; revised April 2025; accepted May 2025 3, No. 4 (SIGMOD), Article 268. Publication date: September 2025.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 19,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. InICSE ’24. ACM, Article 46.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. InOSDI ’23. 397–417.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 22,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. InISSTA ’23. ACM, 140–149.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InESEC/FSE ’20. ACM, 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 46,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (nov 2020), 30 pages.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 47,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. InOSDI ’20. USENIX Association, 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[19, 47, 54, 62]",
+      "technique": "pqs",
+      "sentence": "Such logic bugs are widely recognized as being more challenging to detect than database crashes or exceptions [19, 47, 54, 62].",
+      "context_before": "operties attached to them (e.g., names for users and “since when” for friendships). This makes GDBs particularly well suited for applications such as recommender systems, social networks, and data mining [ 6], where intricate relationships and patterns in large datasets must be rapidly discovered and analyzed. Unfortunately, GDBs’ advanced features make them prone to logic bugs, where they silently produce incorrect query results. Figure 1 presents an example of a bug discovered in FalkorDB that existed for four years. FalkorDB returns an incorrect result that has a wrong value in the a3field.",
+      "context_after": "Authors’ Contact Information: Zijing Yin, ETH Zurich, Switzerland, zijing.yin@inf.ethz.ch; Si Liu, ETH Zurich, Switzerland, si.liu@inf.ethz.ch; David Basin, ETH Zurich, Switzerland, basin@inf.ethz.ch. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. ©2025 Copyright held by the owner/author(s). ACM 2836-6573/2025/9-ART268 https://doi.org/10.1145/3749186 3, No. 4 (SIGMOD), Article 268. Publication date: September 2025. 268:2 Zijing Yin, Si Liu, & David Basin Related Work on GDB Testing.A key challenge in automatically detecting l",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2794,
+      "cited_reference": {
+        "number": 47,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. InOSDI ’20. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 16,19,22,33,61,62]",
+      "technique": null,
+      "sentence": "Recent research has produced excellent testers [ 16,19,22,33,61,62] that attempt to address this challenge.",
+      "context_before": "Switzerland, si.liu@inf.ethz.ch; David Basin, ETH Zurich, Switzerland, basin@inf.ethz.ch. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. ©2025 Copyright held by the owner/author(s). ACM 2836-6573/2025/9-ART268 https://doi.org/10.1145/3749186 3, No. 4 (SIGMOD), Article 268. Publication date: September 2025. 268:2 Zijing Yin, Si Liu, & David Basin Related Work on GDB Testing.A key challenge in automatically detecting logic bugs is to develop an effective test oracle for determining whether a GDB correctly answers a given query.",
+      "context_after": "In line with the growing trend of using randomized testing to effectively uncover system-level bugs in general [ 23,31] and relational databases in particular [ 20,47,54], these tools all generate random graph data and queries to exercise the GDB under test. To determine correctness, they rely on either differential [ 34] or metamorphic [ 9] test oracles. A common limitation of these efforts is the absence of ground truth for their test oracles, resulting in both false positives (incorrectly identified bugs) and false negatives (overlooked bugs). Differential testers [ 16,61] execute the same",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3633,
+      "cited_reference": {
+        "number": 19,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. InICSE ’24. ACM, Article 46.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 20,47,54]",
+      "technique": "pqs",
+      "sentence": "In line with the growing trend of using randomized testing to effectively uncover system-level bugs in general [ 23,31] and relational databases in particular [ 20,47,54], these tools all generate random graph data and queries to exercise the GDB under test.",
+      "context_before": "nsed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. ©2025 Copyright held by the owner/author(s). ACM 2836-6573/2025/9-ART268 https://doi.org/10.1145/3749186 3, No. 4 (SIGMOD), Article 268. Publication date: September 2025. 268:2 Zijing Yin, Si Liu, & David Basin Related Work on GDB Testing.A key challenge in automatically detecting logic bugs is to develop an effective test oracle for determining whether a GDB correctly answers a given query. Recent research has produced excellent testers [ 16,19,22,33,61,62] that attempt to address this challenge.",
+      "context_after": "To determine correctness, they rely on either differential [ 34] or metamorphic [ 9] test oracles. A common limitation of these efforts is the absence of ground truth for their test oracles, resulting in both false positives (incorrectly identified bugs) and false negatives (overlooked bugs). Differential testers [ 16,61] execute the same query on different GDBs or on different versions of the same GDB. The individual execution results are compared with each other, and any discrepancy suggests a bug. Consequently, bugs that arise from libraries shared by multiple database implementations unde",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3741,
+      "cited_reference": {
+        "number": 47,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. InOSDI ’20. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[22, 62]",
+      "technique": null,
+      "sentence": "For instance, the Grand tool [ 61] was recently found to report a large number of false alarms [22, 62].",
+      "context_before": "gs) and false negatives (overlooked bugs). Differential testers [ 16,61] execute the same query on different GDBs or on different versions of the same GDB. The individual execution results are compared with each other, and any discrepancy suggests a bug. Consequently, bugs that arise from libraries shared by multiple database implementations under test may be missed, e.g., two query results are the same, but both are incorrect. In addition, false positives may occur when queries yield different yet expected results for different GDBs (due to their intended designs or implementations [ 35,39]).",
+      "context_after": "Metamorphic testers [ 19,22,33,62] avoid false positives by design. They rewrite the originally generated query into a set of semantically related queries and then check if their outputs maintain the same semantic relation, e.g., whether one result set is equivalent to another [ 33], is a subset of another [ 19], or is a union of multiple other result sets [ 22,62]. However, many bugs may still go unnoticed because these testers focus solely on the relationships between the answers to different queries, rather than establishing a ground truth for each individual query. For example, suppose th",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4852,
+      "cited_reference": {
+        "number": 22,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. InISSTA ’23. ACM, 140–149.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 19,22,33,62]",
+      "technique": null,
+      "sentence": "Metamorphic testers [ 19,22,33,62] avoid false positives by design.",
+      "context_before": "ent GDBs or on different versions of the same GDB. The individual execution results are compared with each other, and any discrepancy suggests a bug. Consequently, bugs that arise from libraries shared by multiple database implementations under test may be missed, e.g., two query results are the same, but both are incorrect. In addition, false positives may occur when queries yield different yet expected results for different GDBs (due to their intended designs or implementations [ 35,39]). For instance, the Grand tool [ 61] was recently found to report a large number of false alarms [22, 62].",
+      "context_after": "They rewrite the originally generated query into a set of semantically related queries and then check if their outputs maintain the same semantic relation, e.g., whether one result set is equivalent to another [ 33], is a subset of another [ 19], or is a union of multiple other result sets [ 22,62]. However, many bugs may still go unnoticed because these testers focus solely on the relationships between the answers to different queries, rather than establishing a ground truth for each individual query. For example, suppose the query𝑄′is derived from the query 𝑄based on the subset relation, an",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4957,
+      "cited_reference": {
+        "number": 19,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. InICSE ’24. ACM, Article 46.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 19]",
+      "technique": null,
+      "sentence": ", whether one result set is equivalent to another [ 33], is a subset of another [ 19], or is a union of multiple other result sets [ 22,62].",
+      "context_before": "entations under test may be missed, e.g., two query results are the same, but both are incorrect. In addition, false positives may occur when queries yield different yet expected results for different GDBs (due to their intended designs or implementations [ 35,39]). For instance, the Grand tool [ 61] was recently found to report a large number of false alarms [22, 62]. Metamorphic testers [ 19,22,33,62] avoid false positives by design. They rewrite the originally generated query into a set of semantically related queries and then check if their outputs maintain the same semantic relation, e.g.",
+      "context_after": "However, many bugs may still go unnoticed because these testers focus solely on the relationships between the answers to different queries, rather than establishing a ground truth for each individual query. For example, suppose the query𝑄′is derived from the query 𝑄based on the subset relation, and their query result sets are𝑅′and𝑅, respectively. While𝑅′⊆𝑅passes the test, both result sets could be incorrect. Additionally, metamorphic testers rely on oracles that may constrain the complexity of test queries, which may be needed to trigger subtle logic bugs. For instance, GDBMeter [ 22] bases i",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5186,
+      "cited_reference": {
+        "number": 19,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. InICSE ’24. ACM, Article 46.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 22]",
+      "technique": null,
+      "sentence": "For instance, GDBMeter [ 22] bases its test oracle on a three-valued logic, which can be used only to filter clauses like WHERE (in Cypher language [ 11]).",
+      "context_before": "multiple other result sets [ 22,62]. However, many bugs may still go unnoticed because these testers focus solely on the relationships between the answers to different queries, rather than establishing a ground truth for each individual query. For example, suppose the query𝑄′is derived from the query 𝑄based on the subset relation, and their query result sets are𝑅′and𝑅, respectively. While𝑅′⊆𝑅passes the test, both result sets could be incorrect. Additionally, metamorphic testers rely on oracles that may constrain the complexity of test queries, which may be needed to trigger subtle logic bugs.",
+      "context_after": "Consequently, those bugs not rooted in incorrect predicate handling are likely to be overlooked. Our Approach.We introduce GQS (Graph Query Synthesis), the first automated testing approach for identifying logic bugs in GDBs using an established ground truth. GQS begins by randomly generating a graph and selecting a set of properties associated with its elements, whose key-value pairs constitute theexpected result setserving as the ground truth. It then synthesizes acomplex query aimed at retrieving these values from the graph. When the GDB under test executes this query on the initially gener",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5889,
+      "cited_reference": {
+        "number": 22,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. InISSTA ’23. ACM, 140–149.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[ 16,47]",
+      "technique": "pqs",
+      "sentence": "Previous work [ 16,47] provides a good basis for generating complex expressions in general, likechar_length(‘abc’)+sqrt(round(1.",
+      "context_before": "ked. Note that for OPTIONAL MATCH clauses, the randomly extended pattern segments are also checked to ensure that the pattern uniquely matches one subgraph, thereby avoiding an increase in the multiplicity of the results. Finally, the expressions used in these predicates will be further substituted with more complicated ones that also achieve the same filtering, as described in Section 3.5. 3.5 Generating Complex Expressions While predicates with simple expressions, such as n’.id = 4, are sufficient to filter out undesired graph elements, more complex expressions help to extensively test GDBs.",
+      "context_after": "2)), that satisfy specific value constraints, such as evaluating to 4as in the right-hand side of the above predicate. GQS adapts the approach proposed in [16] to construct such expressions. Specifically, GQS first selects an expression template for the new expression. This template includes functions (e.g., sqrt(par) ), operator applications (e.g., par1 + par2 ), and string match expressions (e.g., par1 CONTAINS par2 ), wherepar1 andpar2 denote the parameters of the template. Such templates function as the skeleton of the expressions. Next, GQS converts the value constraint into respective su",
+      "section": "3.5 Generating Complex Expressions",
+      "page": 12,
+      "char_offset": 38608,
+      "cited_reference": {
+        "number": 47,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. InOSDI ’20. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 22]",
+      "technique": null,
+      "sentence": "•The metamorphic bug detectors, GDBMeter [ 22], Gamera [ 62], GQT [ 19], and GRev [ 33], each applying specifically designed query rewrite rules to identify discrepancies between the result of the original query and that of the rewritten query.",
+      "context_before": "lyze the variety of search patterns used and the depth of nested expressions involved in the bug-triggering test queries. Figure 14 presents the distribution of bugs based on the number of patterns used. Notably, two-thirds of the bugs are triggered by test queries containing more than three distinct patterns. Nested expressions also play a crucial role in triggering bugs. As shown in Figure 15, 83% of the bugs are associated with queries that contain more than five levels of nested expressions. 5.4 Comparison with State-of-the-Art We compare GQS with five state-of-the-art logic bug detectors.",
+      "context_after": "•The differential tester GDsmith [ 16], which detects logic bugs by comparing the results of the same query across multiple GDBs. 3, No. 4 (SIGMOD), Article 268. Publication date: September 2025. Testing Graph Databases with Synthesized Queries 268:19 5.4.1 Bug Latency.In line with common practice [ 21,33], we analyze the latencies of the bugs found by GQS, i.e., the time span between the earliest version of the GDB in which a bug was introduced and when it was first discovered by us. This analysis provides a perspective on GQS’s effectiveness in uncovering hidden bugs: if a bug exists in ve",
+      "section": "5.4 Comparison with State-of-the-Art",
+      "page": 18,
+      "char_offset": 61181,
+      "cited_reference": {
+        "number": 22,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. InISSTA ’23. ACM, 140–149.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 22]",
+      "technique": null,
+      "sentence": "Upon applying GDBMeter’s ternary logic partitioning [ 22] to the test query, all three partitioned queries (lines 5–13) also produce empty results.",
+      "context_before": "d GRev, we are able to quantify the number of missed bugs because GDBMeter’s metamorphic relations are straightforward to implement, and GRev provides direct interfaces for applying its rewrite rules to given queries. Out of 26 logic bugs detected by GQS, GDBMeter and GRev could only identify 11 and 3, respectively, even when using the bug-triggering test cases. Example 5.4.Figure 16 shows an example bug found by GQS that is missed by GDBMeter. When running Memgraph with the GQS-synthesized test query (lines 2–3), it produces an incorrect empty output, instead of the expected result (line 14).",
+      "context_after": "The union of these results matches the incorrect output (namely GDBMeter’s oracle), causing GDBMeter to mistakenly pass the test. The key factor triggering this bug lies in the WITH clause, which is not accounted for by GDBMeter’s metamorphic rules. 1// GQS-synthesized test query 2MATCH (n0)-[r0]->(n1) WITH r0 ... 3WHERE (\"1\"<>n0.k99) RETURN r0.id AS a0 4 5// GDBMeter-rewritten query #1 (NOT) 6MATCH (n0)-[r0]->(n1) WITH r0 ... 7WHERE NOT (\"1\"<>n0.k99) RETURN r0.id AS a0 8// GDBMeter-rewritten query #2 (IS NULL) 9MATCH (n0)-[r0]->(n1) WITH r0 ... 10WHERE (\"1\"<>n0.k99) IS NULL RETURN r0.id AS a",
+      "section": "5.4 Comparison with State-of-the-Art",
+      "page": 20,
+      "char_offset": 66496,
+      "cited_reference": {
+        "number": 22,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. InISSTA ’23. ACM, 140–149.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[1,2,3]",
+      "technique": "dqp",
+      "sentence": "1UNWIND [1,2,3] AS a0 2MATCH (n2 :L12)-[r1]-(n3) WHERE (((r1.",
+      "context_before": "e: September 2025. Testing Graph Databases with Synthesized Queries 268:21 relations and observe that these tools are likely to miss some bugs reported by our tool, as their root causes are unrelated to the relations. Example 5.5.Figure 17 illustrates a representative bug found in the latest version of FalkorDB, which is unlikely to be detected by Gamera or GQT. This bug’s root cause lies in FalkorDB’s mishandling of UNWIND ’s semantics, resulting in some records not being fetched. Gamera’s and GQT’s oracles are unlikely to identify this bug because their metamorphic rules do not cover UNWIND.",
+      "context_after": "id) = 13) AND ... 3RETURN a0 4// expected result: [{a0:1},{a0:2},{a0:3}]✓ 5// actual result: [{a0:1}]ὁB Fig. 17. A bug found by GQS in FalkorDB’s latest version that is unlikely to be detected by Gamera or GQT. False Alarms.For the differential tester GDsmith, feeding it with GQS-generated bug-triggering queries resulted in no missed bugs. This is expected since our tested GDBs do not share a substantial common codebase; therefore, the same bugs do not exist across these GDBs. However, we observe a significant number of false positives, which may limit GDsmith’s practical applicability. In par",
+      "section": "5.4 Comparison with State-of-the-Art",
+      "page": 21,
+      "char_offset": 68732,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[ 3,15,20,21,46,47,49,50,54]",
+      "technique": "dqp",
+      "sentence": "In recent years, a wide range of logic bug detectors have been developed for RDBs [ 3,15,20,21,46,47,49,50,54].",
+      "context_before": "ontrast, GQS restarts the database instance for each new graph to ensure a clean execution environment. This design choice enhances bug reproducibility and accelerates the detection of logic bugs, but may reduce the likelihood of revealing crash bugs that emerge only after prolonged execution. Given our focus on logic bug detection, we consider this trade-off reasonable. 6 Related Work We have already discussed the state-of-the-art approaches for finding GDB logic bugs in Section 1. In this section, we focus on related work in relational database (RDB) testing and query synthesis. RDB Testing.",
+      "context_after": "While RDBs and GDBs differ in their data models and query languages, many of their bug-detection techniques are grounded in similar principles—most notably, the use of randomized testing with randomly generated initial database states and queries. Moreover, test oracles such as differential testing and certain metamorphic testing strategies developed for RDBs can, in principle, be adapted to the GDB setting. GDBMeter illustrates this by adopting the metamorphic oracles from TLP [ 46], originally designed for SQL, which partitions queries using three-valued logic. However, such adaptations may",
+      "section": "6 Related Work",
+      "page": 22,
+      "char_offset": 72489,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "GDBMeter illustrates this by adopting the metamorphic oracles from TLP [ 46], originally designed for SQL, which partitions queries using three-valued logic.",
+      "context_before": "d work in relational database (RDB) testing and query synthesis. RDB Testing.In recent years, a wide range of logic bug detectors have been developed for RDBs [ 3,15,20,21,46,47,49,50,54]. While RDBs and GDBs differ in their data models and query languages, many of their bug-detection techniques are grounded in similar principles—most notably, the use of randomized testing with randomly generated initial database states and queries. Moreover, test oracles such as differential testing and certain metamorphic testing strategies developed for RDBs can, in principle, be adapted to the GDB setting.",
+      "context_after": "However, such adaptations may leave substantial parts of the graph-specific semantics untested. For example, key constructs in GDBs, such as paths, neighborhoods, and recursive traversals, do not map directly to the relational structures and operators in RDBs. Similar to our approach, PQS [ 47], Pinolo [ 15], and TQS [ 54] test RDBs by synthesizing SQL queries. However, PQS and Pinolo cannot establish the ground truth and, instead, rely on subset relationships as test oracles—an approach that doesnotguarantee the correctness of individual queries. TQS targets logic bugs in join optimizations",
+      "section": "6 Related Work",
+      "page": 22,
+      "char_offset": 73012,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Similar to our approach, PQS [ 47], Pinolo [ 15], and TQS [ 54] test RDBs by synthesizing SQL queries.",
+      "context_before": "ates and queries. Moreover, test oracles such as differential testing and certain metamorphic testing strategies developed for RDBs can, in principle, be adapted to the GDB setting. GDBMeter illustrates this by adopting the metamorphic oracles from TLP [ 46], originally designed for SQL, which partitions queries using three-valued logic. However, such adaptations may leave substantial parts of the graph-specific semantics untested. For example, key constructs in GDBs, such as paths, neighborhoods, and recursive traversals, do not map directly to the relational structures and operators in RDBs.",
+      "context_after": "However, PQS and Pinolo cannot establish the ground truth and, instead, rely on subset relationships as test oracles—an approach that doesnotguarantee the correctness of individual queries. TQS targets logic bugs in join optimizations by synthesizing join queries based on normal forms. This is a strategy specific to the relational model that does not naturally extend to graphs. While our stepwise synthesis approach is tailored to the semantic characteristics of Cypher queries, many SQL queries, particularly those involving sequences of joins or recursion, exhibit conceptually similar chain-li",
+      "section": "6 Related Work",
+      "page": 22,
+      "char_offset": 73431,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "However, PQS and Pinolo cannot establish the ground truth and, instead, rely on subset relationships as test oracles—an approach that doesnotguarantee the correctness of individual queries.",
+      "context_before": "trategies developed for RDBs can, in principle, be adapted to the GDB setting. GDBMeter illustrates this by adopting the metamorphic oracles from TLP [ 46], originally designed for SQL, which partitions queries using three-valued logic. However, such adaptations may leave substantial parts of the graph-specific semantics untested. For example, key constructs in GDBs, such as paths, neighborhoods, and recursive traversals, do not map directly to the relational structures and operators in RDBs. Similar to our approach, PQS [ 47], Pinolo [ 15], and TQS [ 54] test RDBs by synthesizing SQL queries.",
+      "context_after": "TQS targets logic bugs in join optimizations by synthesizing join queries based on normal forms. This is a strategy specific to the relational model that does not naturally extend to graphs. While our stepwise synthesis approach is tailored to the semantic characteristics of Cypher queries, many SQL queries, particularly those involving sequences of joins or recursion, exhibit conceptually similar chain-like structures. Hence, it should be possible to adapt our synthesis strategy to RDB testing as well. Query Synthesis.SQL and Datalog query synthesis has been extensively studied over the past",
+      "section": "6 Related Work",
+      "page": 22,
+      "char_offset": 73534,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[ 45]",
+      "technique": "norec",
+      "sentence": "268:24 Zijing Yin, Si Liu, & David Basin of ground truth on the expected execution time, where the cardinality estimation approach [ 45] designed for RDBs may provide a direction for future research.",
+      "context_before": "these additional features in future work. Beyond Logic Bugs.Graph query generation is fundamental to GDB testing, not only for detecting logic bugs but also for assessing other properties. As already demonstrated by our experiments, the complex queries we synthesized are effective in triggering GDB crashes or unexpected exceptions (10 in total across four GDBs). Additionally, our query synthesizer can serve as a basis for examining GDB performance issues like inefficient query processing. The challenge lies in the current lack 3, No. 4 (SIGMOD), Article 268. Publication date: September 2025.",
+      "context_after": "Recent years have also seen advances in black-box checking of isolation levels [ 14,17,23,28, 53,58,59], such as serializability, snapshot isolation, and weaker levels [ 5,29]. These efforts primarily focus on RDBs and their concurrency control mechanisms for transactions (assuming the correctness for individual queries). Our work can be naturally extended to address this problem for GDBs, e.g., checking serializability claimed by Neo4j or snapshot isolation claimed by Memgraph. This can be done by wrapping generated Cypher queries as transactions and utilizing existing oracles like Adya’s th",
+      "section": "7 Discussion and Concluding Remark",
+      "page": 23,
+      "char_offset": 78792,
+      "cited_reference": {
+        "number": 45,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InESEC/FSE ’20. ACM, 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:55:50Z",
+    "is_model_written": true,
+    "summary": "This work tests graph databases by synthesising queries whose correct result is known by construction, so the oracle does not depend on comparing two queries. That distinguishes it from the metamorphic graph testers, which detect only discrepancies between an original and a rewritten query and are therefore bounded by what their rewrite rules can express. It reports bugs in Neo4j, RedisGraph and Memgraph that the existing detectors do not find.",
+    "narrative": "TLP reaches this paper through GDBMeter, which the authors describe as adopting the metamorphic oracles from TLP, originally designed for SQL, and partitioning queries using three-valued logic. That inheritance is the limitation they identify: a three-valued oracle applies only to filter clauses such as WHERE, and they show a case where all three of GDBMeter's partitioned queries return empty results and so reveal nothing. GDBMeter is one of four metamorphic detectors compared against. PQS is discussed as the closest relational analogue -- it also synthesises queries -- but is distinguished on the grounds that it relies on a subset relationship rather than establishing ground truth.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "background",
+      "M6": "definition",
+      "M7": "motivation",
+      "M8": "background",
+      "M9": "baseline",
+      "M10": "result_comparison",
+      "M11": "incidental",
+      "M12": "background",
+      "M13": "definition",
+      "M14": "definition",
+      "M15": "motivation",
+      "M16": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The comparison tools are graph testers run as baselines; nothing describes reuse of the SQLancer codebase."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The oracle is ground truth established by construction, which M15 contrasts with the subset relationships PQS relies on. It is an alternative to those oracles rather than a development of one."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M9",
+          "M10"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M9",
+            "sentence": "•The metamorphic bug detectors, GDBMeter [ 22], Gamera [ 62], GQT [ 19], and GRev [ 33], each applying specifically designed query rewrite rules to identify discrepancies between the result of the original query and that of the rewritten query.",
+            "section": "5.4 Comparison with State-of-the-Art",
+            "page": 18
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "Upon applying GDBMeter’s ternary logic partitioning [ 22] to the test query, all three partitioned queries (lines 5–13) also produce empty results.",
+            "section": "5.4 Comparison with State-of-the-Art",
+            "page": 20
+          }
+        ],
+        "reasoning": "M9 lists GDBMeter among the metamorphic bug detectors compared against, and M10 works through what GDBMeter's ternary logic partitioning produces on a test query. M13 identifies GDBMeter's oracle as TLP's, carried over from SQL.",
+        "techniques": [
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The section is headed as a comparison with state-of-the-art tools, but no sentence attributes that standing to TLP, PQS or SQLancer specifically."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3757347_3759132.json b/_data/papers/paper_doi_10_1145_3757347_3759132.json
new file mode 100644
index 0000000..56e2e49
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3757347_3759132.json
@@ -0,0 +1,414 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:31:55Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3757347.3759132",
+    "title": "PANGOLIN: a Comprehensive Testing Framework for Configuration-Rich Key-Value Stores",
+    "authors": [
+      "Shaohua Duan",
+      "Sudarsun Kannan",
+      "Andrea C. Arpaci-Dusseau",
+      "Remzi H. Arpaci-Dusseau"
+    ],
+    "year": 2025,
+    "venue": "Annual Haifa Experimental Systems Conference",
+    "doi": "10.1145/3757347.3759132",
+    "arxiv_id": null,
+    "s2_paper_id": "34675858280c9b98a3df15fab99c590458e35afd",
+    "url": "https://doi.org/10.1145/3757347.3759132",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3757347.3759132",
+      "retrieved_at": "2026-09-08T14:31:55Z",
+      "chars": 70145,
+      "content_sha256": "sha256:c74bc00b82318aa369c57637716a2b40dbd26c2115ea6f2470f68477314b98ab"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 14,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2121
+      },
+      {
+        "number": "2.1",
+        "title": "Large and Complex Configurations",
+        "start": 10351
+      },
+      {
+        "number": "3.1",
+        "title": "Bug Reproduction",
+        "start": 14111
+      },
+      {
+        "number": "3.2",
+        "title": "Bug Pattern",
+        "start": 15683
+      },
+      {
+        "number": "4.1",
+        "title": "Fine-Grained Boundary",
+        "start": 19898
+      },
+      {
+        "number": "5.1",
+        "title": "Overview",
+        "start": 23966
+      },
+      {
+        "number": "5.2",
+        "title": "KVDB Command and Interpreter",
+        "start": 25365
+      },
+      {
+        "number": "5.3",
+        "title": "KVDB Grammar",
+        "start": 27480
+      },
+      {
+        "number": "5.4",
+        "title": "KVDB Executor and Checker",
+        "start": 29987
+      },
+      {
+        "number": "5.6",
+        "title": "Bounded Fuzzer",
+        "start": 33853
+      },
+      {
+        "number": "7.2",
+        "title": "Bug Finding",
+        "start": 43889
+      },
+      {
+        "number": "7.2.1",
+        "title": "Bug Case Studies. We specifically explore two typical",
+        "start": 46331
+      },
+      {
+        "number": "7.3",
+        "title": "Compute and Storage Overhead",
+        "start": 47971
+      },
+      {
+        "number": "7.4",
+        "title": "Code Coverage",
+        "start": 49054
+      },
+      {
+        "number": "7.4.2",
+        "title": "Improvement under different boundaries. The size and",
+        "start": 51389
+      },
+      {
+        "number": "7.4.3",
+        "title": "Improvement on other key-value stores. To verify the",
+        "start": 52202
+      },
+      {
+        "number": "7.5",
+        "title": "Bug Coverage",
+        "start": 53349
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Cornelius Aschermann, Tommaso Frassetto, Thorsten Holz, Patrick Jauernig, Ahmad-Reza Sadeghi,, and Daniel Teuchert. 2019. NAUTILUS: Fishing for Deep Bugs with Grammars. In Symposium on Network and Distributed System Security (NDSS).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Marcel Böhme, Van-Thuan Pham, Manh-Dung Nguyen, and Abhik Roychoudhury. 2017. Directed Greybox Fuzzing. In Proceedings ofthe 24th ACMSIGSAC Conference on Computer and Communications Security (CCS’17). Dallas, TX, 2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Mariano Ceccato, Cu D. Nguyen, Dennis Appelt, and Lionel C. Briand. 2016. SOFIA: An automated security oracle for black-box testing of SQL-injection vulnerabilities. In 2016 31st IEEE/ACM International Conference on Automated Software Engineering (ASE). 167–177.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Matteo Collina. [n.d.]. Graph database JS style for Node.js and the Browser. Built upon LevelUp and LevelDB. https://github.com/ levelgraph/levelgraph.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Dmitry Duplyakin, Robert Ricci, Aleksander Maricq, Gary Wong, Jonathon Duerig, Eric Eide, Leigh Stoller, Mike Hibler, David Johnson, Kirk Webb, Aditya Akella, Kuangching Wang, Glenn Ricart, Larry Landweber, Chip Elliott, Michael Zink, Emmanuel Cecchet, Snigdhaswin Kar, and Prabodh Mishra. 2019. The Design and Operation of CloudLab. In Proceedings of the USENIX Annual Technical Conference (ATC). 1–",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Facebook. [n.d.]. False detection of corruption after system crash due to race condition with WAL_sync and track_and_verify_wals_in_manifest. https://github.com/facebook/ rocksdb/pull/10185/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Facebook. [n.d.]. RocksDB. http://rocksdb.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. 2020. AFL++: Combining Incremental Steps of Fuzzing Research. In 14th USENIX Workshop on Offensive Technologies (WOOT 20).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Xinwei Fu, Wook-Hee Kim, Ajay Paddayuru Shreepathi, Mohannad Ismail, Sunny Wadkar, Dongyoon Lee, and Changwoo Min. 2021. Witcher: Systematic Crash Consistency Testing for Non-Volatile Memory Key-Value Stores. In Proceedings of the 28th Symposium on Operating Systems Principles (SOSP’21). 111–115.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Peter M. Goldstein and Mike Perham. [n.d.]. Dalli. https://github. com/petergoldstein/dalli.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Google. [n.d.]. LevelDB. https://github.com/google/leveldb.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Ethan Hamilton. [n.d.]. Rocksdb is eating the database world. https: //rockset.com/blog/rocksdb-is-eating-the-database-world/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jun He, Kan Wu, Sudarsun Kannan, Andrea Arpaci-Dusseau, and Remzi Arpaci-Dusseau. 2020. Read as Needed: Building WiSER, a Flash-Optimized Search Engine. In 18th USENIX Conference on File and Storage Technologies (FAST 20). 59–73.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Christian Holler, Kim Herzig, and Andreas Zeller. 2012. Fuzzing with code fragments. In Proceedings of USENIX Security Symposium. 445–458.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "John E. Hopcroft, Rajeev Motwani, and Jeffrey D. Ullman. 2006. Introduction to Automata Theory, Languages, and Computation (3.00 ed.). Pearson.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Shengtuo Hu. [n.d.]. A grammar-based custom mutator for AFL++. https://github.com/AFLplusplus/Grammar-Mutator.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Dae R. Jeong, Kyungtae Kim, Basavesh Shivakumar, Byoungyoung Lee, and Insik Shin. 2019. Razzer: Finding Kernel Race Bugs through Fuzzing. In Proceedings of the 40th IEEE Symposium on Security and Privacy (SP’19). San Francisco, CA, 754–768.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Matt Jibson. [n.d.]. SQLsmith: Randomized SQL Testing in CockroachDB. https://www.cockroachlabs.com/blog/sqlsmith-randomizedsql-testing/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "James Johnson. [n.d.]. gramfuzz is a grammar-based fuzzer that lets one define complex grammars to generate text and binary data formats. https://github.com/d0c-s4vage/gramfuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Mohd. Ehmer Khan. 2011. Different Approaches To Black box Testing Technique For Finding Errors. International Journal of Software Engineering & Applications 2 (10 2011), 31–40.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Seulbae Kim, Meng Xu, Sanidhya Kashyap, Jungyeon Yoon, Wen Xu, and Taesoo Kim. 2019. Finding Semantic Bugs in File Systems with an Extensible Fuzzing Framework. In Proceedings of the 27th Symposium 214 SYSTOR ’25, September 8–9, 2025, Virtual, Israel S. Duan et al. on Operating Systems Principles (SOSP’19). 147–161.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Emre Kycyman and Yi-Min Wang. 2004. Discovering correctness constraints for self-management of system configuration. Proceedings of the 1st International Conference on Autonomic Computing (ICAC’04). (2004), 28–35.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "LLVM. [n.d.]. libFuzzer. https://github.com/facebook/rocksdb/tree/ master/fuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Memcached.org. [n.d.]. Memcached. https://github.com/memcached/ memcached.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Jayashree Mohan, Ashlie Martinez, Soujanya Ponnapalli, Pandian Raju, and Vijay Chidambaram. 2018. Finding Crash-Consistency Bugs with Bounded Black-Box Crash Testing. In Proceedings of the 13th USENIX Conference on Operating Systems Design and Implementation (OSDI’18). 33–50.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Sarah Nadi, Thorsten Berger, Christian Kästner, and Krzysztof Czarnecki. 2014. Mining Configuration Constraints: Static Analyses and Empirical Results. In Proceedings of the 36th International Conference on Software Engineering (ICSE’14). 140–151.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Arjun Narayan and Peter Mattis. [n.d.]. Why we built cockroachdb on top of rocksdb. https://www.cockroachlabs.com/blog/cockroachdbon-rocksd/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Srinivas Nidhra1 and Jagruthi Dondeti. 2012. Black Box and White Box Testing Techniques-A Literature Review. International Journal of Embedded Systems and Applications 2 (06 2012), 29–50.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Joshua Pereyda. [n.d.]. Sulley: A pure-python fully automated and unattended fuzzing framework. https://github.com/OpenRCE/sulley.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Sanjay Rawat, Vivek Jain, Ashish Kumar, Lucian Cojocar, Cristiano Giuffrida, and Herbert Bos. 2017. VUzzer: Application-aware Evolutionary Fuzzing. In Proceedings of the 24th ACMSIGSAC Conference on Computer and Communications Security (CCS’17). Dallas, TX,2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). 33–50.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 32,
+      "text": "Salvatore Sanfilippo. [n.d.]. Redis. https://redis.io/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Andreas Seltenreich. 2020. SQLSmith. https://github.com/anse1/ sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Shan. [n.d.]. Ceph using Monitor key/value store. https://ceph.io/en/ news/blog/2015/ceph-using-monitor-keyvalue-store/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Spandan Veggalam, Sanjay Rawat, Istvan Haller, and Herbert Bos. 2016. Ifuzzer: An evolutionary interpreter fuzzer using genetic programming.. In Proceedings of European Symposium on Research in Computer Security. Springer, 581–601.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "VMware. [n.d.]. Concord Byzantine fault tolerant state machine replication library. https://github.com/vmware/concord-bft.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Tianyin Xu, Long Jin, Xuepeng Fan, Yuanyuan Zhou, Shankar Pasupathy, and Rukma Talwadker. 2015. Hey, you have given me too many knobs!: Understanding and dealing with over-designed configuration in system software.. In Proceedings of the Joint Meeting of the European Software Engineering Conference and the ACMSIGSOFT Symposium on the Foundations of Software Engineering (ESEC/FSE’15). 307–319.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Tianyin Xu, Jiaqi Zhang, Peng Huang, Jing Zheng, Tianwei Sheng, Ding Yuan, Yuanyuan Zhou, and Shankar Pasupathy. 2013. Do Not Blame Users for Misconfigurations. In Proceedings of the 24th Symposium on Operating Systems Principles (SOSP’13). 244–259.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Michał Zalewski. 2017. American Fuzzy Lop. https://github.com/ google/AFL. 215",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 31,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). 33–50.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 31]",
+      "technique": "pqs",
+      "sentence": "Black-box testing and fuzzing have been used to identify various types of bugs in File systems and DBMSs, including crash inconsistency bugs [ 21,25], logic 203 PANGOLIN: a Comprehensive Testing Framework for Configuration-Rich Key-Value Stores SYSTOR ’25, September 8–9, 2025, Virtual, Israel bugs [ 31], and correctness bugs [ 3].",
+      "context_before": "ack-box testing is a software testing method in which the functionalities of software applications are tested without having knowledge of internal code structure, implementation details, and internal code paths. Differed from black-box testing, modern fuzzing tools, such as AFL [ 8,39] and Nautilus [ 1], also employ light-weight instrumentation of the target program to collect some feedback about the generated inputs, which is used to guide the test case mutation process. More recently, leveraging grammar-based input generation greatly improved the performance of feedbackoriented fuzzers [ 1].",
+      "context_after": "Compared with handwritten tests, these bug-finding tools have the capability to generate interesting test cases, including corner cases that are difficult for humans to contemplate, under acceptable overhead and little human effort involved. In principle, black-box testing and fuzzing can be routinely applied for systematically testing configuration-rich key-value stores. Unfortunately, our study of historical bugs in five widely used key-value stores reveals that generalpurpose testing tools can only identify a few bugs. One main challenge is the large number of operations and configuration",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 9241,
+      "cited_reference": {
+        "number": 31,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). 33–50.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "Pangolin tests configuration-rich key-value stores. The authors first study historical bugs in five mature stores over eight years and find that most could be caught by systematically testing a small sequence of operations and configurations. Pangolin turns that into a bounded testing strategy across black-box and fuzzing procedures, finding 20 bugs and reproducing 443 historical ones in RocksDB, LevelDB, HyperlevelDB, BadgerDB and Redis.",
+    "narrative": "One citation, noting that black-box testing and fuzzing have been used to find bugs in file systems and DBMSs, SQLancer's work among the references.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3758316_3763249.json b/_data/papers/paper_doi_10_1145_3758316_3763249.json
new file mode 100644
index 0000000..c4b733d
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3758316_3763249.json
@@ -0,0 +1,243 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:24:50Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3758316.3763249",
+    "title": "LLM-Assisted Dialect-Agnostic SQL Query Parsing",
+    "authors": [
+      "Junwen An"
+    ],
+    "year": 2025,
+    "venue": "SPLASH Companion",
+    "doi": "10.1145/3758316.3763249",
+    "arxiv_id": null,
+    "s2_paper_id": "541b4674b5c71530b124a319901697eb713f6746",
+    "url": "https://doi.org/10.1145/3758316.3763249",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3758316.3763249",
+      "retrieved_at": "2026-09-08T15:24:50Z",
+      "chars": 14130,
+      "content_sha256": "sha256:151f360a226ec140dc69bc41e56d1ded6a124436293dff7dc7515c58bc4b4233"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 3,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1388
+      },
+      {
+        "number": "2",
+        "title": "Approach",
+        "start": 3994
+      },
+      {
+        "number": "3",
+        "title": "Evaluation",
+        "start": 7849
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "ANSI/ISO. 1992. SQL-92. https://www.contrib.andrew.cmu.edu/ ~shadow/sql/sql1992.txt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Bikash Chandra, Bhupesh Chawda, Biplab Kar, K. V. Reddy, Shetal Shah, and S. Sudarshan. 2015. Data generation for testing and grading SQL queries. The VLDB Journal (2015), 731–755. https://doi.org/10. 1007/s00778-015-0395-0",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Koen Claessen and John Hughes. 2000. QuickCheck: a lightweight tool for random testing of Haskell programs. In Proceedings of the Fifth ACMSIGPLAN International Conference on Functional Programming (ICFP ’00). 268–279. https://doi.org/10.1145/351240.351266",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Nils Anders Danielsson. 2013. Correct-by-construction pretty-printing. InProceedings of the 2013 ACMSIGPLAN Workshop on DependentlyTyped Programming (DTP ’13). https://doi.org/10.1145/2502409. 2502410",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Github Issue. 2025. Add Dremio dialect #4904. https://github.com/ tobymao/sqlglot/issues/4904.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Moshe Hazoom, Vibhor Malik, and Ben Bogin. 2021. Text-to-SQL in the Wild: A Naturally-Occurring Dataset Based on Stack Exchange Data. InProceedings of the 1st Workshop on Natural Language Processing for Programming (NLP4Prog 2021), Royi Lachmy, Ziyu Yao, Greg Durrett, Milos Gligoric, Junyi Jessy Li, Ray Mooney, Graham Neubig, Yu Su, Huan Sun, and Reut Tsarfaty (Eds.). 77–87. https://doi.org/10.186",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Zhuohang Jiang, Pangjing Wu, Ziran Liang, Peter Q. Chen, Xu Yuan, Ye Jia, Jiancheng Tu, Chen Li, Peter H. F. Ng, and Qing Li. 2025. HiBench: Benchmarking LLMs Capability on Hierarchical Structure Reasoning. arXiv:2503.00912 [cs.CL] https://arxiv.org/abs/2503.00912",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting Logic Bugs in Database Engines via Equivalent Expression Transformation. In 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24). USENIX Association, Santa Clara, CA, 821–835.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Bill Karwin. 2010. SQL Antipatterns: Avoiding the Pitfalls of Database Programming. Pragmatic Bookshelf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Li Lin, Zongyin Hao, Chengpeng Wang, Zhuangda Wang, Rongxin Wu, and Gang Fan. 2024. SQLess: Dialect-Agnostic SQL Query Simplification. In Proceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis. 743–754. https://doi.org/10.1145/ 3650212.3680317",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Thomas Neumann and Viktor Leis. 2024. A Critique of Modern SQL and a Proposal Towards a Simple and Expressive Query Language. In14th Conference on Innovative Data Systems Research, CIDR 2024, Chaminade, HI, USA, January 14-17, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "OpenAI. [n. d.]. GPT-4.1. https://platform.openai.com/docs/models/ gpt-4.1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "pglast. 2024. https://github.com/lelit/pglast.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (2020). https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "Stuart Russel and Peter Norvig. 2020. Artificial Intelligence: A Modern Approach, 4th edition.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jeff Shute, Shannon Bales, Matthew Brown, Jean-Daniel Browne, Brandon Dolphin, Romit Kudtarkar, Andrey Litvinov, Jingchi Ma, John Morcos, Michael Shen, David Wilhite, Xi Wu, and Lulan Yu. 2024. SQL Has Problems. We Can Fix Them: Pipe Syntax In SQL. Proc. VLDB Endow. (2024). https://doi.org/10.14778/3685800.3685826",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "SQLFluff. 2024. SQLFluff. https://docs.sqlfluff.com/en/stable/index. html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "SQLFluff. 2024. SQLFluff Rule Reference. https://docs.sqlfluff.com/en/ stable/reference/rules.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Chengnian Sun, Yuanbo Li, Qirun Zhang, Tianxiao Gu, and Zhendong Su. 2018. Perses: Syntax-Guided Program Reduction. In Proceedings of the 40th International Conference on Software Engineering. 361–371. https://doi.org/10.1145/3180155.3180236 Received 2025-07-07; accepted 2025-08-12 33",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 14,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (2020). https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 8,14]",
+      "technique": "tlp",
+      "sentence": "Many tasks, such as SQL linting [ 17], test case reduction [ 10], DBMS testing [ 8,14], and SQL grading [ 2], rely on query analysis and rewriting.",
+      "context_before": "s engineering →Parsers; Software maintenance tools. Keywords: SQL Dialect, Parser, Large Language Model ACM Reference Format: Junwen An. 2025. LLM-Assisted Dialect-Agnostic SQL Query Parsing. In Companion Proceedings of the 2025 ACMSIGPLAN International Conference on Systems, Programming, Languages, and Applications: Software for Humanity (SPLASH Companion ’25), October 12–18, 2025, Singapore, Singapore. ACM, New York, NY, USA, 3 pages. https://doi.org/10.1145/3758316.3763249 1 Introduction Relational Database Management Systems (RDBMS) are widely used, with SQL as the primary query interface.",
+      "context_after": "Formally, query rewriting transforms a query 𝑄into𝑄′to satisfy an objective 𝑂. Query analysis, which inspects queries without modifying them, is a subtask of rewriting, since most rewrites begin with analysis. Query rewriting tools typically parse queries into ASTs, analyze or rewrite them, and generate new queries. However, grammar-based parsers struggle with dialect-specific syntax, limiting cross-DBMS compatibility. LLMs are a promising This work is licensed under a Creative Commons Attribution 4.0 International License. SPLASH Companion ’25, Singapore, Singapore ©2025 Copyright held by th",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 1507,
+      "cited_reference": {
+        "number": 14,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (2020). https://doi.org/10.1145/3428279",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "prints the DOI of the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "SQLFlex parses SQL across dialects. Grammar-based parsers, on which query analysis and rewriting tools depend, fail on dialect-specific syntax, while large language models understand SQL but do not reliably produce accurate syntax trees. SQLFlex tries a grammar-based parser first and, when that fails, uses an LLM to split the query into smaller parsable pieces. It parsed 96.37% of queries across eight dialects.",
+    "narrative": "One citation, listing DBMS testing among the tasks that rely on query analysis, with SQLancer's work among the references.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3764583.json b/_data/papers/paper_doi_10_1145_3764583.json
new file mode 100644
index 0000000..2524fd7
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3764583.json
@@ -0,0 +1,1290 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:25:20Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3764583",
+    "title": "Unveiling Logic Bugs in SPJG Query Optimizations within DBMS",
+    "authors": [
+      "Xiu Tang",
+      "Shijie Yang",
+      "Sai Wu",
+      "Dongxiang Zhang",
+      "Wenchao Zhou",
+      "Feifei Li",
+      "Gang Chen"
+    ],
+    "year": 2025,
+    "venue": "ACM Transactions on Database Systems",
+    "doi": "10.1145/3764583",
+    "arxiv_id": null,
+    "s2_paper_id": "3374c834b786e3f37193ee037036ee45c3dc1620",
+    "url": "https://doi.org/10.1145/3764583",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3764583",
+      "retrieved_at": "2026-09-08T15:25:20Z",
+      "chars": 109275,
+      "content_sha256": "sha256:b41577ee35fcda52976d27430f0ca394a17c955228802dd0bde343143a96c0b7"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 35,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 4041
+      },
+      {
+        "number": "2",
+        "title": "Overview",
+        "start": 10141
+      },
+      {
+        "number": "2.1",
+        "title": "Problem Definition",
+        "start": 10742
+      },
+      {
+        "number": "2.2",
+        "title": "Scheme Overview",
+        "start": 12151
+      },
+      {
+        "number": "3.1",
+        "title": "Data Generation",
+        "start": 18907
+      },
+      {
+        "number": "3.2",
+        "title": "Noise Injection",
+        "start": 27526
+      },
+      {
+        "number": "3.3",
+        "title": "SPJG Query Generator",
+        "start": 36183
+      },
+      {
+        "number": "3.4",
+        "title": "Ground-truth Result Generation",
+        "start": 39890
+      },
+      {
+        "number": "4.1",
+        "title": "Data Cube Space",
+        "start": 47002
+      },
+      {
+        "number": "4.2",
+        "title": "Query Generation Space",
+        "start": 49454
+      },
+      {
+        "number": "5.1",
+        "title": "Overview and Showcase of Join Bug Reports",
+        "start": 60196
+      },
+      {
+        "number": "5.2",
+        "title": "Overview and Showcase of Grouped Aggregation Bug Reports",
+        "start": 72584
+      },
+      {
+        "number": "5.3",
+        "title": "Comparison with Existing Tools",
+        "start": 78105
+      },
+      {
+        "number": "5.4",
+        "title": "Ablation Studies",
+        "start": 88292
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Günes Aluç, Olaf Hartig, M. Tamer Özsu, and Khuzaima Daudjee. 2014. Diversified stress testing of RDF data management systems. In Proceedings of the International Semantic Web Conference. Springer, 197–212.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Güneş Aluç, Olaf Hartig, M. Tamer Özsu, and Khuzaima Daudjee. 2014. Diversified stress testing of RDF data management systems. In Proceedings of the International Semantic Web Conference. 197–212.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Akhil Arora, Sakshi Sinha, Piyush Kumar, and Arnab Bhattacharya. 2018. HD-index: Pushing the scalability-accuracy boundary for approximate kNN search in high-dimensional spaces. Proc. VLDB Endow. 11, 8 (2018), 906–919.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding performance issues in database systems through the lens of cardinality estimation. In ICSE. 133:1–133:13.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans. Proc.ACMManag. Data2, 3 (2024), 188.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Guillaume Bagan, Angela Bonifati, Radu Ciucanu, George H. L. Fletcher, Aurélien Lemay, and Nicky Advokaat. 2017. gMark: Schema-driven generation of graphs and queries. In ProceedingsoftheIEEETransactionsonKnowledgeand Data Engineering. 63–64.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. 2007. A genetic approach for random testing of database systems. In Proceedings of the 33rd International Conference on Very Large Data Bases. ACM, 1243–1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M. Tamer Özsu. 2007. QAGen: Generating query-aware test databases. InProceedings of the 2007 ACMSIGMOD International Conference on Management of Data. ACM, 341–352.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Joachim Biskup. 1995. Achievements of relational database schema design theory revisited. In Proceedings of the Semantics in Databases. Springer, 29–54.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Joachim Biskup, Umeshwar Dayal, and Philip A. Bernstein. 1979. Synthesizing independent database schemas. In Proceedings of the SIGMOD. Philip A. Bernstein (Ed.), ACM, 143–151.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Christian Bizer and Andreas Schultz. 2009. The berlin sparql benchmark. IJSWIS5, 2 (2009), 1–24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Daniel Blum and Sara Cohen. 2011. Grr: Generating random RDF. In Proceedings of the Extended Semantic Web Conference. Springer, 16–30.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Manuel Bodirsky. 2015. Graph homomorphisms and universal algebra course notes. TU Dresden (2015), 6–18.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Robert Brummayer and Armin Biere. 2009. Fuzzing and delta-debugging SMT solvers. In Proceedings of the 7th International Workshop on Satisfiability Modulo Theories. 1–5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Nicolas Bruno and Surajit Chaudhuri. 2005. Flexible database generators. In ProceedingsoftheVLDB. ACM, 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. 2006. Generating queries with cardinality constraints for DBMS testing.IEEETransactionson Knowledge and Data Engineering 18, 12 (2006), 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Wei Cao, Yingqiang Zhang, Xinjun Yang, Feifei Li, Sheng Wang, Qingda Hu, Xuntao Cheng, Zongzhi Chen, Zhenjun Liu, Jing Fang, Bo Wang, Yuhui Wang, Haiqing Sun, Ze Yang, Zhushi Cheng, Sen Chen, Jian Wu, Wei Hu, Jianwei Zhao, Yusong Gao, Songlu Cai, Yunyang Zhang, and Jiawang Tong. 2021. PolarDB serverless: A cloud native database for disaggregated data centers. In Proceedings of the SIGMOD. ACM, 247",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "PascalCuoq,BenjaminMonate,AnnePacalet,VirgilePrevosto,JohnRegehr,BorisYakobowski,andXuejunYang.2012. Testing static analyzers with randomly generated programs. In Proceedings of the NASA Formal Methods Symposium. Springer, 120–125.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Angjela Davitkova, Damjan Gjurovski, and Sebastian Michel. 2022. LMKG: Learned models for cardinality estimation in knowledge graphs. In EDBT. 2:169–2:182.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "DB-Engines.2018.DB-EnginesRanking.[EB/OL].(2018).RetrievedMay2024from https://db-engines.com/en/ranking",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Jim Diederich and Jack Milton. 1988. New methods and fast algorithms for database normalization. ACMTransactions on Database Systems 13, 3 (1988), 339–365.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Chi Thang Duong, Dung Hoang, Hongzhi Yin, Matthias Weidlich, Quoc Viet Hung Nguyen, and Karl Aberer. 2021. Efficient streaming subgraph isomorphism with graph neural networks. VLDB14, 5 (2021), 730–742.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Orri Erling, Alex Averbuch, Josep Larriba-Pey, Hassan Chafi, Andrey Gubichev, Arnau Prat, Minh-Duc Pham, and Peter Boncz. 2015. The LDBC social network benchmark: Interactive workload. In Proceedings of the 2015 ACMSIGMOD International Conference on Management of Data. 619–630.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "M. R. Garey and David S. Johnson. 1979. ComputersandIntractability:AGuidetotheTheoryofNP-Completeness. W. H. Freeman.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "JimGray,PrakashSundaresan,SusanneEnglert,KennethBaclawski,andPeterJ.Weinberger.1994.Quicklygenerating billion-record synthetic databases. In Proceedings of the 1994 ACMSIGMOD International Conference on Management of Data. ACM, 243–252. ACM Trans. Datab. Syst., Vol. 51, No. 1, Article 3. Publication date: September 2025. 3:34 X. Tang et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Zhongxian Gu, Mohamed A. Soliman, and Florian M. Waas. 2012. Testing the accuracy of query optimizers. In Proceedings of the 5th International Workshop on Testing Database Systems. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Yuanbo Guo, Zhengxiang Pan, and Jeff Heflin. 2005. LUBM: A benchmark for OWL knowledge base systems. Journal of Web Semantics 3, 2-3 (2005), 158–182.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Kenneth Houkjær, Kristian Torp, and Rico Wind. 2006. Simple and realistic data generation. In Proceedingsofthe32nd International Conference on Very Large Data Bases. ACM, 1243–1246.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Dongxu Huang, Qi Liu, Qiu Cui, Zhuhe Fang, Xiaoyu Ma, Fei Xu, Li Shen, Liu Tang, Yuxing Zhou, Menglong Huang, et al. 2020. TiDB: A raft-based HTAP database. VLDB13, 12 (2020), 3072–3084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Yka Huhtala, Juha Kärkkäinen, Pasi Porkka, and Hannu Toivonen. 1999. TANE: An efficient algorithm for discovering functional and approximate dependencies. The Computer Journal 42, 2 (1999), 100–111.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woon-Hak Kang. 2019. APOLLO: Automatic detection and diagnosis of performance regressions in database systems. VLDB13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Timotej Kapus and Cristian Cadar. 2017. Automatic testing of symbolic execution engines via program generation and differential testing. In Proceedingsofthe201732ndIEEE/ACMInternationalConferenceonAutomatedSoftware Engineering. IEEE Computer Society, 590–600.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O. Laleye, and Sarfraz Khurshid. 2008. Query-aware test generation using a relational constraint solver. In Proceedings of the 2008 23rd IEEE/ACM International Conference on Automated Software Engineering. IEEE Computer Society, 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. 2010. Automated SQL query generation for systematic testing of database engines. In Proceedings of the 25th IEEE/ACM International Conference on Automated Software Engineering. ACM, 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. In Proceedings of the ACM Sigplan Notices. ACM, 216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Eric Lo, Carsten Binnig, Donald Kossmann, M. Tamer Özsu, and Wing-Kai Hon. 2010. A framework for testing DBMS features.VLDB J. 19, 2 (2010), 203–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "David Maier. 1983. TheTheory of Relational Databases. Computer Science Press.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "MariaDB. 2022. MariaDB hints. [EB/OL]. (2022). Retrieved May 2024 from https://mariadb.com/kb/en/optimizerswitch/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Mariadb. 2022. Mariadb Homepage. [EB/OL]. (2022). Retrieved from https://mariadb.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "William M. McKeeman. 1998. Differential testing for software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. 2008. Generating targeted queries for database testing. In Proceedings of the 2008 ACMSIGMOD International Conference on Management of Data. ACM, 499–510.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "MySQL. 2022. MySQL hints. [EB/OL]. (2022). Retrieved May 2024 from https://dev.mysql.com/doc/refman/8.0/en/ optimizer-hints.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "MySQL. 2022. MySQL Homepage. [EB/OL]. (2022). Retrieved May 2024 from https://www.mysql.com",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Stack Overflow. 2021. Developer Survey Results. [EB/OL]. (2021). Retrieved May 2024 from https://insights. stackoverflow.com/survey/2021",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "ThorstenPapenbrockandFelixNaumann.2016.Ahybridapproachtofunctionaldependencydiscovery.In Proceedings of the 2016 International Conference on Management of Data. Fatma Özcan, Georgia Koutrika, and Sam Madden (Eds.), ACM, 821–833.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Thorsten Papenbrock and Felix Naumann. 2017. Data-driven schema normalization. In Proceedings of the EDBT. OpenProceedings.org, 342–353.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Meikel Poess and John M. Stephens. 2004. Generating thousand benchmark queries in seconds. In Proceedings of the VLDB. Morgan Kaufmann, 1045–1053.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Kim-Thomas Rehmann, Changyun Seo, Dongwon Hwang, Binh Than Truong, Alexander Boehm, and Dong Hun Lee. 2016. Performance monitoring in SAPHANA’s continuous integration process. SIGMETRICS Perform. Evaluation Rev. 43, 4 (2016), 43–52.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the ACM Joint Meeting on ESEC and FSE. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 50,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedingsofthe ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 51,
+      "text": "Manuel Rigger and Zhendong Su. 2020. SQLancer. [EB/OL]. (2020). Retrieved May 2024 from https://github.com/ sqlancer/sqlancer",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 52,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedingsofthe OSDI 20. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 53,
+      "text": "Michael Schmidt, Thomas Hornung, Georg Lausen, and Christoph Pinkel. 2009. SP^ 2Bench: ASPARQL performance benchmark. In Proceedings of the ICDE. IEEE, 222–233. ACM Trans. Datab. Syst., Vol. 51, No. 1, Article 3. Publication date: September 2025. Unveiling Logic Bugs in SPJG Query Optimizations within DBMS 3:35",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Andreas Seltenreich. 2020. SQLSmith. [EB/OL]. (2020). Retrieved May 2024 from https://github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Donald R. Slutz. 1998. Massive stochastic testing of SQL. In ProceedingsoftheVLDB. Ashish Gupta, Oded Shmueli, and Jennifer Widom (Eds.), Morgan Kaufmann, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting logic bugs of join optimizations in DBMS. DOI:https://doi.org/10.1145/3588909",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "TiDB. 2022. TiDB hints. [EB/OL]. (2022). Retrieved May 2024 from https://docs.pingcap.com/tidb/v5.3/optimizer-hints",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Manasi Vartak, Venkatesh Raghavan, and Elke A. Rundensteiner. 2010. QRelX: Generating meaningful queries that provide cardinality assurance. In Proceedings of the SIGMOD. ACM, 1215–1218.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Ziheng Wei and Sebastian Link. 2019. Embedded functional dependencies and data-completeness tailored database design.VLDB12, 11 (2019), 1458–1470.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Kesheng Wu, Ekow J. Otoo, and Arie Shoshani. 2002. Compressing bitmap indexes for faster search operations. In Proceedings of the SSDBM. IEEE Computer Society, 99–108.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Jiaqi Yan, Qiuye Jin, Shrainik Jain, Stratis D. Viglas, and Allison W. Lee. 2018. Snowtrail: Testing with production queries on a cloud database. In Proceedings of the SIGMOD. ACM, 4:1–4:6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In Proceedings of the SIGPLAN, PLDI. ACM, 283–294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Yixing Yang, Yixiang Fang, Maria E. Orlowska, Wenjie Zhang, and Xuemin Lin. 2021. Efficient bi-triangle counting for large bipartite networks. VLDB14, 6 (2021), 984–996.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Rex Ying, Zhaoyu Lou, Jiaxuan You, Chengtao Wen, Arquimedes Canedo, and Jure Leskovec. 2020. Neural subgraph matching. CoRRabs/2007.03092 (2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing database management systems with language validity and coverage feedback. In CCS. ACM, 955–970. Received 30 May 2024; revised 12 May 2025; accepted 30 July 2025 ACM Trans. Datab. Syst., Vol. 51, No. 1, Article 3. Publication date: September 2025.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding performance issues in database systems through the lens of cardinality estimation. In ICSE. 133:1–133:13.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans. Proc.ACMManag. Data2, 3 (2024), 188.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 49,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the ACM Joint Meeting on ESEC and FSE. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 50,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedingsofthe ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 51,
+      "text": "Manuel Rigger and Zhendong Su. 2020. SQLancer. [EB/OL]. (2020). Retrieved May 2024 from https://github.com/ sqlancer/sqlancer",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 52,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedingsofthe OSDI 20. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 51] has emerged as a promising tool for detecting logic bugs in DBMS through the generation and execution of random SQL queries, supplemented by differential testing.",
+      "context_before": "ions for query evaluation. The query optimizer is considered one of the most complex and important components in DBMS. It parses the input SQL queries and generates an efficient execution plan with the assistance of built-in cost models. The implementation errors in a query optimizer can lead to bugs, including crashes and logic bugs. Crashes are easier to detect as the system will halt immediately. Whereas, logic bugs are prone to being ignored, because they simply cause the DBMS to return incorrect result sets that are hard to detect. In this article, we focus on detecting these silent bugs.",
+      "context_after": "However, exclusive reliance on random query generation and differential testing introduces limitations. Random queries often fail to cover diverse scenarios, potentially overlooking critical bug points withinthe DBMS.Therefore,thereisanurgentneedfor toolstothoroughlytestthekeyerror-prone areas of the DBMS. In Figure 1, we illustrate two logic bugs encountered in databases, specifically pertaining to join and grouped aggregation queries. These bugs are detectable by the tool proposed in this article. In Figure1(a),alogicbuginhashjoinprocessingwithinMySQL8.0.18isdemonstrated,whereaninner hash jo",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4882,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3 Data-guided Schema and Query Generation (DSG) SQLancer [ 51] automatically detects logic bugs in the implementation of DBMS.",
+      "context_before": "6:𝐺(𝑉,𝐸,𝜋) = 𝑃𝑙𝑎𝑛𝐼𝑡𝑒𝑟𝑎𝑡𝑖𝑣𝑒(𝐺𝑠) 7:for𝑖 = 0to𝛾do 8:𝑂 = 𝑆ℎ𝑢𝑓𝑓𝑙𝑒(𝑉) 9:foreach𝑣𝑖∈ 𝑂do 10: 𝑊𝑣𝑖= 𝐴𝑑𝑎𝑝𝑡𝑖𝑣𝑒𝑅𝑎𝑛𝑑𝑜𝑚𝑊𝑎𝑙𝑘(𝐺,𝑣𝑖,𝑙,𝐺𝐼) 11: 𝐺𝐼 = 𝐺𝐼∪𝐸𝑚𝑏𝑒𝑑𝑑𝑖𝑛𝑔(𝑊𝑣𝑖) 12: 𝑞𝑖= 𝑄𝑢𝑒𝑟𝑦𝐺𝑒𝑛(𝑊𝑣𝑖) 13: 𝑡𝑟𝑎𝑛𝑠_𝑞𝑖= 𝐻𝑖𝑛𝑡𝐺𝑒𝑛(𝑞𝑖,𝐻,𝐺𝑡) 14: 𝐺𝑇𝑞𝑖= 𝑔𝑒𝑡𝐺𝑇(𝑞𝑖) 15: 𝑆𝑡𝑟𝑎𝑛𝑠_𝑞𝑖= 𝑅𝑒𝑠𝑢𝑙𝑡𝑆𝑒𝑡(𝑡𝑟𝑎𝑛𝑠 _𝑞𝑖) 16: if𝑆𝑡𝑟𝑎𝑛𝑠_𝑞𝑖≠ 𝐺𝑇𝑞𝑖then 17: 𝑏𝑢𝑔𝑠 = 𝑏𝑢𝑔𝑠 ∪𝑡𝑟𝑎𝑛𝑠 _𝑞𝑖 18:Return:𝑏𝑢𝑔𝑠,𝐺𝐼 TQS currently focuses on bug detection of equi-join queries. However, the idea of DSG and KQE canbeextendedtonon-equaljoins.Theonlychallengeishowtogenerateandmanageground-truth results, whose sizes increase exponentially for non-equal joins. We leave it to our future work.",
+      "context_after": "It first creates a populateddatabasewithrandomlygeneratedtables.Afterward,itrandomlychooses SQLstatements to create, modify, and delete data. Using the randomly generated database, it adopts differential testing approaches such as DifferentialQueryPlans (DQP) [5] to detect logic bugs. Note that SQLancer is originally designed for logic bug detection in random queries, with a particular focus on differential testing of complex expressions in single-table scenarios. Although its latest version has been extended to support multi-table join queries by updating the data and query generator, it stil",
+      "section": "2.2 Scheme Overview",
+      "page": 6,
+      "char_offset": 18038,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "Using the randomly generated database, it adopts differential testing approaches such as DifferentialQueryPlans (DQP) [5] to detect logic bugs.",
+      "context_before": "_𝑞𝑖≠ 𝐺𝑇𝑞𝑖then 17: 𝑏𝑢𝑔𝑠 = 𝑏𝑢𝑔𝑠 ∪𝑡𝑟𝑎𝑛𝑠 _𝑞𝑖 18:Return:𝑏𝑢𝑔𝑠,𝐺𝐼 TQS currently focuses on bug detection of equi-join queries. However, the idea of DSG and KQE canbeextendedtonon-equaljoins.Theonlychallengeishowtogenerateandmanageground-truth results, whose sizes increase exponentially for non-equal joins. We leave it to our future work. 3 Data-guided Schema and Query Generation (DSG) SQLancer [ 51] automatically detects logic bugs in the implementation of DBMS. It first creates a populateddatabasewithrandomlygeneratedtables.Afterward,itrandomlychooses SQLstatements to create, modify, and delete data.",
+      "context_after": "Note that SQLancer is originally designed for logic bug detection in random queries, with a particular focus on differential testing of complex expressions in single-table scenarios. Although its latest version has been extended to support multi-table join queries by updating the data and query generator, it still lacks support for testing aggregate queries, causing many related bugs to be missed. For details, please refer to the experimental section. 3.1 Data Generation To efficiently handle grouped aggregation queries, we create a test database in the form of a data cube. For multi-table jo",
+      "section": "2.2 Scheme Overview",
+      "page": 6,
+      "char_offset": 18306,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Note that SQLancer is originally designed for logic bug detection in random queries, with a particular focus on differential testing of complex expressions in single-table scenarios.",
+      "context_before": "and KQE canbeextendedtonon-equaljoins.Theonlychallengeishowtogenerateandmanageground-truth results, whose sizes increase exponentially for non-equal joins. We leave it to our future work. 3 Data-guided Schema and Query Generation (DSG) SQLancer [ 51] automatically detects logic bugs in the implementation of DBMS. It first creates a populateddatabasewithrandomlygeneratedtables.Afterward,itrandomlychooses SQLstatements to create, modify, and delete data. Using the randomly generated database, it adopts differential testing approaches such as DifferentialQueryPlans (DQP) [5] to detect logic bugs.",
+      "context_after": "Although its latest version has been extended to support multi-table join queries by updating the data and query generator, it still lacks support for testing aggregate queries, causing many related bugs to be missed. For details, please refer to the experimental section. 3.1 Data Generation To efficiently handle grouped aggregation queries, we create a test database in the form of a data cube. For multi-table join query bug detection, we transform the data cube into a wide table and apply schema normalization to decompose it into multiple normalized tables. In addition, we propose an effecti",
+      "section": "2.2 Scheme Overview",
+      "page": 6,
+      "char_offset": 18450,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We compare TQS with SQLancer,3which is the state-of-the-art approach to detecting logic bugs in databases.",
+      "context_before": "ntentionally change the physical plans [ 38,42,57]. We have evaluated TQS with both randomly generated TPC-H data and the dataset from UCI machine learning repository.2However, since the TPC-H dataset follows a uniform distribution and has a simple schema, all reported bugs on the TPC-H dataset have been covered by the UCI dataset. Therefore, we only show the results on the UCI data. As TQS runs, it continuously reports bugs and to be fair for all DBMSs, we only report the results for the first 24 hours. All DBMSs are run with default configurations and compilation options. Baselineapproaches.",
+      "context_after": "SQLancer is not designed to test multi-table queries. However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one table. Similar to the single table case, all queries and tuples are randomly generated. We use four methods in SQLancer as our baselines. The first one is PQS [ 52], which constructs queries to fetch a randomly selected tuple from a table. The tested DBMS may contain a bug if it fails to fetch that tuple. The second one is TLP [ 52], which decomposes a query into three partitioning queries, each of which computes its resul",
+      "section": "4.2 Query Generation Space",
+      "page": 18,
+      "char_offset": 58452,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer is not designed to test multi-table queries.",
+      "context_before": "H data and the dataset from UCI machine learning repository.2However, since the TPC-H dataset follows a uniform distribution and has a simple schema, all reported bugs on the TPC-H dataset have been covered by the UCI dataset. Therefore, we only show the results on the UCI data. As TQS runs, it continuously reports bugs and to be fair for all DBMSs, we only report the results for the first 24 hours. All DBMSs are run with default configurations and compilation options. Baselineapproaches. We compare TQS with SQLancer,3which is the state-of-the-art approach to detecting logic bugs in databases.",
+      "context_after": "However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one table. Similar to the single table case, all queries and tuples are randomly generated. We use four methods in SQLancer as our baselines. The first one is PQS [ 52], which constructs queries to fetch a randomly selected tuple from a table. The tested DBMS may contain a bug if it fails to fetch that tuple. The second one is TLP [ 52], which decomposes a query into three partitioning queries, each of which computes its result on that tuple. The third one is NoRec [ 49], which t",
+      "section": "4.2 Query Generation Space",
+      "page": 18,
+      "char_offset": 58559,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We use four methods in SQLancer as our baselines.",
+      "context_before": "s on the UCI data. As TQS runs, it continuously reports bugs and to be fair for all DBMSs, we only report the results for the first 24 hours. All DBMSs are run with default configurations and compilation options. Baselineapproaches. We compare TQS with SQLancer,3which is the state-of-the-art approach to detecting logic bugs in databases. SQLancer is not designed to test multi-table queries. However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one table. Similar to the single table case, all queries and tuples are randomly generated.",
+      "context_after": "The first one is PQS [ 52], which constructs queries to fetch a randomly selected tuple from a table. The tested DBMS may contain a bug if it fails to fetch that tuple. The second one is TLP [ 52], which decomposes a query into three partitioning queries, each of which computes its result on that tuple. The third one is NoRec [ 49], which targets logic bugs generated by the optimization process in DBMS. It performs a comparison between the results of randomly-generated queries and their rewritten ones. The fourth one is DQP [ 5], which enforces different query plans for the same query and val",
+      "section": "4.2 Query Generation Space",
+      "page": 18,
+      "char_offset": 58820,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "The first one is PQS [ 52], which constructs queries to fetch a randomly selected tuple from a table.",
+      "context_before": "ports bugs and to be fair for all DBMSs, we only report the results for the first 24 hours. All DBMSs are run with default configurations and compilation options. Baselineapproaches. We compare TQS with SQLancer,3which is the state-of-the-art approach to detecting logic bugs in databases. SQLancer is not designed to test multi-table queries. However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one table. Similar to the single table case, all queries and tuples are randomly generated. We use four methods in SQLancer as our baselines.",
+      "context_after": "The tested DBMS may contain a bug if it fails to fetch that tuple. The second one is TLP [ 52], which decomposes a query into three partitioning queries, each of which computes its result on that tuple. The third one is NoRec [ 49], which targets logic bugs generated by the optimization process in DBMS. It performs a comparison between the results of randomly-generated queries and their rewritten ones. The fourth one is DQP [ 5], which enforces different query plans for the same query and validates that the results are consistent. DBMS versions. Note that DBMS bugs will be fixed once reported",
+      "section": "4.2 Query Generation Space",
+      "page": 18,
+      "char_offset": 58870,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The second one is TLP [ 52], which decomposes a query into three partitioning queries, each of which computes its result on that tuple.",
+      "context_before": "neapproaches. We compare TQS with SQLancer,3which is the state-of-the-art approach to detecting logic bugs in databases. SQLancer is not designed to test multi-table queries. However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one table. Similar to the single table case, all queries and tuples are randomly generated. We use four methods in SQLancer as our baselines. The first one is PQS [ 52], which constructs queries to fetch a randomly selected tuple from a table. The tested DBMS may contain a bug if it fails to fetch that tuple.",
+      "context_after": "The third one is NoRec [ 49], which targets logic bugs generated by the optimization process in DBMS. It performs a comparison between the results of randomly-generated queries and their rewritten ones. The fourth one is DQP [ 5], which enforces different query plans for the same query and validates that the results are consistent. DBMS versions. Note that DBMS bugs will be fixed once reported to the community. In our experiments, we report the results of the join testing on MySQL 8.0.28, MariaDB 10.8.2, TiDB 5.4.0, and PolarDB beta 8.0.18, respectively. For aggregation testing, we report the",
+      "section": "4.2 Query Generation Space",
+      "page": 18,
+      "char_offset": 59039,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "The third one is NoRec [ 49], which targets logic bugs generated by the optimization process in DBMS.",
+      "context_before": "designed to test multi-table queries. However, it can be tailored for multi-table queries by artificially generating queries and tuples across more than one table. Similar to the single table case, all queries and tuples are randomly generated. We use four methods in SQLancer as our baselines. The first one is PQS [ 52], which constructs queries to fetch a randomly selected tuple from a table. The tested DBMS may contain a bug if it fails to fetch that tuple. The second one is TLP [ 52], which decomposes a query into three partitioning queries, each of which computes its result on that tuple.",
+      "context_after": "It performs a comparison between the results of randomly-generated queries and their rewritten ones. The fourth one is DQP [ 5], which enforces different query plans for the same query and validates that the results are consistent. DBMS versions. Note that DBMS bugs will be fixed once reported to the community. In our experiments, we report the results of the join testing on MySQL 8.0.28, MariaDB 10.8.2, TiDB 5.4.0, and PolarDB beta 8.0.18, respectively. For aggregation testing, we report the results of the latest released versions during our testing, namely, MySQL 8.1.0, MariaDB 11.3.2, TiDB",
+      "section": "4.2 Query Generation Space",
+      "page": 18,
+      "char_offset": 59175,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "The fourth one is DQP [ 5], which enforces different query plans for the same query and validates that the results are consistent.",
+      "context_before": "queries and tuples are randomly generated. We use four methods in SQLancer as our baselines. The first one is PQS [ 52], which constructs queries to fetch a randomly selected tuple from a table. The tested DBMS may contain a bug if it fails to fetch that tuple. The second one is TLP [ 52], which decomposes a query into three partitioning queries, each of which computes its result on that tuple. The third one is NoRec [ 49], which targets logic bugs generated by the optimization process in DBMS. It performs a comparison between the results of randomly-generated queries and their rewritten ones.",
+      "context_after": "DBMS versions. Note that DBMS bugs will be fixed once reported to the community. In our experiments, we report the results of the join testing on MySQL 8.0.28, MariaDB 10.8.2, TiDB 5.4.0, and PolarDB beta 8.0.18, respectively. For aggregation testing, we report the results of the latest released versions during our testing, namely, MySQL 8.1.0, MariaDB 11.3.2, TiDB 7.5.0, and PolarDB beta 8.5.0, respectively. Setup.All experiments are conducted on our in-house server equipped with Intel Xeon CPU E5-2682 (2.50 GHz) 16 cores and 128 GB memory. For our cloud-native database PolarDB, we run an in",
+      "section": "4.2 Query Generation Space",
+      "page": 18,
+      "char_offset": 59378,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "sqlancer",
+      "technique": null,
+      "sentence": "com/sqlancer/sqlancer 4http://embed.",
+      "context_before": "r. 5.1 Overview and Showcase of Join Bug Reports TQS has successfully detected 115 join bugs in tested DBMSs within 24 hours, including 31 bugs in MySQL, 30 bugs in MariaDB, 31 bugs in TiDB, and 23 bugs in PolarDB, respectively. However, some bugs may be caused by the same relational algebras. Hence, once finding a possible bug, it is essential to produce a minimal test case with C-Reduce4before reporting the bug, to save the DBMS developers’ time and effort. Then, we submit all bugs and the corresponding test cases to 2https://archive.ics.uci.edu/ml/datasets/KDD+Cup+1998+Data 3https://github.",
+      "context_after": "cs.utah.edu/creduce/ ACM Trans. Datab. Syst., Vol. 51, No. 1, Article 3. Publication date: September 2025. Unveiling Logic Bugs in SPJG Query Optimizations within DBMS 3:19 Table 5. Detected Join Bugs Database IDStatus Severity Description MySQL 8.0.281Fixed S1 (Critical) Semi-join gives wrong results. 2Fixed S2 (Serious) Incorrect inner hash join when using materialization strategy. 3Verified S2 (Serious) Incorrect semi-join execution results in unknown data. 4Verified S2 (Serious) Incorrect left hash join with subquery in condition. 5Verified S2 (Serious) Incorrect nested loop antijoin. 6Fix",
+      "section": "5.1 Overview and Showcase of Join Bug Reports",
+      "page": 18,
+      "char_offset": 60793,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560100200300400/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni0000004a/uni00000055/uni00000044/uni00000053/uni0000004b/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQSPQS TLPDQP (a) MySQL join diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560100200300400/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni0000004a/uni00000055/uni00000044/uni00000053/uni0000004b/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQS NoRec DQP (b) MariaDB join diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560100200300400/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni0000004a/uni00000055/uni00000044/uni00000053/uni0000004b/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQSTLP DQP (c) TiDB join diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560100200300400/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni0000004a/uni00000055/uni00000044/uni00000053/uni0000004b/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQSPQS TLPDQP (d) PolarDB join diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni00000056050100150200/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni00000046/uni00000058/uni00000045/uni00000048/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQSPQS TLPDQP (e) MySQL cube diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni00000056050100150200/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni00000046/uni00000058/uni00000045/uni00000048/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQS NoRec DQP (f)MariaDBcubediversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni00000056050100150200/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni00000046/uni00000058/uni00000045/uni00000048/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQSTLP DQP (g)TiDB cube diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni00000056050100150200/uni00000047/uni0000004c/uni00000059/uni00000048/uni00000055/uni00000056/uni00000048/uni00000003/uni00000046/uni00000058/uni00000045/uni00000048/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e TQSPQS TLPDQP (h) PolarDB cube diversity 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni000000560816243240485664/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000057/uni0000005c/uni00000053/uni00000048/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 TQSPQS TLPDQP (i) MySQL efficiency 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni00000056061218243036424854/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000057/uni0000005c/uni00000053/uni00000048/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 TQS NoRec DQP (j) MariaDB efficiency 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni00000056010203040506070/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000057/uni0000005c/uni00000053/uni00000048/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 TQSTLP DQP (k) TiDB efficiency 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni00000056051015202530354045/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000057/uni0000005c/uni00000053/uni00000048/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 TQSPQS TLPDQP (l) PolarDB efficiency Fig.",
+      "context_before": "00) ,( NULL); query: SELECT t0. c0FROMt2INNER JOIN t0ONt2. c0 = t0. c0 CROSS JOIN t1GROUP BY t0. c0; ground truth: +−−−−−−−−−+ | c0 | +−−−−−−−−−+ | 0^BtQYw | | | +−−−−−−−−−+ result: +−−−−−−+ | c0 | +−−−−−−+ | | +−−−−−−+ Listing 12. PolarDB’s incorrect grouping execution. 5.3 Comparison with Existing Tools 5.3.1 BugDetectionMetricsAnalysis. We compare TQS with four state-of-the-art DBMS logic bug testing approaches in different aspects to show its effectiveness and efficiency. Note that due ACM Trans. Datab. Syst., Vol. 51, No. 1, Article 3. Publication date: September 2025. 3:28 X. Tang et al.",
+      "context_after": "10. Comparisonwithexistingtoolsinthequerygraphdiversity,thequerycubediversityandtheefficiency of detecting logic bugs. to the compatibility problem, SQLancer implements different approaches on different databases (PQS, TLP and DQP on MySQL and PolarDB; NoRec and DQP on MariaDB; TLP and DQP on TiDB). Figure 10shows our comparison results. We adopt three metrics. The diversity of graphs indicates the number of distinct isomorphic sets tested, the diversity of cubes reflects the variety of cubes tested across different dimension and measure column combinations, and the bug type count represents t",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 28,
+      "char_offset": 78433,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "to the compatibility problem, SQLancer implements different approaches on different databases (PQS, TLP and DQP on MySQL and PolarDB; NoRec and DQP on MariaDB; TLP and DQP on TiDB).",
+      "context_before": "/uni0000004a/uni00000003/uni00000057/uni0000005c/uni00000053/uni00000048/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 TQSTLP DQP (k) TiDB efficiency 0 4 8 12 16 20 24 /uni0000004b/uni00000052/uni00000058/uni00000055/uni00000056051015202530354045/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000057/uni0000005c/uni00000053/uni00000048/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 TQSPQS TLPDQP (l) PolarDB efficiency Fig.10. Comparisonwithexistingtoolsinthequerygraphdiversity,thequerycubediversityandtheefficiency of detecting logic bugs.",
+      "context_after": "Figure 10shows our comparison results. We adopt three metrics. The diversity of graphs indicates the number of distinct isomorphic sets tested, the diversity of cubes reflects the variety of cubes tested across different dimension and measure column combinations, and the bug type count represents the number of returned logic bug types. Querygraphdiversity. Figure10(a), (b), (c), and (d) show the query graph diversity (in thousands of isomorphic sets) of MySQL, MariaDB, TiDB, and PolarDB, respectively. TQS significantly outperforms baseline approaches on testing diversity. This is because that",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 28,
+      "char_offset": 82471,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp",
+        "dqp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This is because that (1) SQLancer approaches may generate random joins with empty results which are not usable for testing and (2) TQS adopts the KQE to avoid repeatedly testing the same query structure.",
+      "context_before": "TLP and DQP on TiDB). Figure 10shows our comparison results. We adopt three metrics. The diversity of graphs indicates the number of distinct isomorphic sets tested, the diversity of cubes reflects the variety of cubes tested across different dimension and measure column combinations, and the bug type count represents the number of returned logic bug types. Querygraphdiversity. Figure10(a), (b), (c), and (d) show the query graph diversity (in thousands of isomorphic sets) of MySQL, MariaDB, TiDB, and PolarDB, respectively. TQS significantly outperforms baseline approaches on testing diversity.",
+      "context_after": "Querycubediversity. Figure10(e), (f), (g), and (h) show the query cube diversity of MySQL, MariaDB, TiDB, and PolarDB, respectively. TQS significantly outperforms baseline approaches on testing diversity. This is due to SQLancer occasionally generating random “Group By” clauses, which are often disorganized and invalid, rendering them ineffective for testing. Furthermore, TQS employs the KQE to avoid repetitive testing of the same query cube structure. Efficiency. Figure10(i), (j), (k), and (l) show the testing efficiency of MySQL, MariaDB, TiDB, and PolarDB, respectively. Not surprisingly, w",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 28,
+      "char_offset": 83232,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This is due to SQLancer occasionally generating random “Group By” clauses, which are often disorganized and invalid, rendering them ineffective for testing.",
+      "context_before": "d) show the query graph diversity (in thousands of isomorphic sets) of MySQL, MariaDB, TiDB, and PolarDB, respectively. TQS significantly outperforms baseline approaches on testing diversity. This is because that (1) SQLancer approaches may generate random joins with empty results which are not usable for testing and (2) TQS adopts the KQE to avoid repeatedly testing the same query structure. Querycubediversity. Figure10(e), (f), (g), and (h) show the query cube diversity of MySQL, MariaDB, TiDB, and PolarDB, respectively. TQS significantly outperforms baseline approaches on testing diversity.",
+      "context_after": "Furthermore, TQS employs the KQE to avoid repetitive testing of the same query cube structure. Efficiency. Figure10(i), (j), (k), and (l) show the testing efficiency of MySQL, MariaDB, TiDB, and PolarDB, respectively. Not surprisingly, we obtain a similar result as the diversity experiment. More query structures are being tested, more bugs can be discovered. In fact, we run the TQS for 48 hours and continue to find logic bugs for all DBMSs. But there are too many bugs and we have not submitted them to the developer communities for verification. So ACM Trans. Datab. Syst., Vol. 51, No. 1, Arti",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 28,
+      "char_offset": 83641,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Unique Logic Bugs Reported by TQS and SQLancer [ 5] DatabaseTQS SQLancer all bugs join bugs aggregation bugs all bugs join bugs aggregation bugs MySQL 13 9 4 5 4 0 MariaDB 8 5 3 7 6 0 TiDB 10 5 5 9 5 0 TOTAL 31 19 12 14 15 0 Table 8.",
+      "context_before": ". 11. Bug types vs. bug counts on MySQL. 1 2 3 4 5 /uni00000051/uni00000058/uni00000050/uni00000045/uni00000048/uni00000055/uni00000003/uni00000052/uni00000049/uni00000003/uni00000046/uni0000004f/uni0000004c/uni00000048/uni00000051/uni00000057/uni00000056025050075010001250150017502000/uni00000051/uni00000058/uni00000050/uni00000045/uni00000048/uni00000055/uni00000003/uni00000052/uni00000049/uni00000003/uni00000054/uni00000058/uni00000048/uni00000055/uni0000004c/uni00000048/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e number of queries Fig. 12. Effect of parallel search. Table 7.",
+      "context_after": "Bug Reproduction of Logic Join Bugs Reported by TQS and SQLancer [ 5] DatabaseTQS SQLancer all bugs non-reproduced reproduced reproduced all bugs non-reproduced MySQL 9 7 2 4 2 MariaDB 5 1 4 6 2 TiDB 5 3 2 5 3 TOTAL 19 11 8 15 7 we only show bug types of the first 24 hours. Figure 11illustrates that the number of bugs increases linearly with the testing time, while the number of bug types is not. This indicates that most bugs are caused by a small set of improperly implemented operators. InSection 4,webrieflyexplaintheideaofusing KQEtobuildadistributedcomputationframework to speed up the testi",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 29,
+      "char_offset": 85333,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Bug Reproduction of Logic Join Bugs Reported by TQS and SQLancer [ 5] DatabaseTQS SQLancer all bugs non-reproduced reproduced reproduced all bugs non-reproduced MySQL 9 7 2 4 2 MariaDB 5 1 4 6 2 TiDB 5 3 2 5 3 TOTAL 19 11 8 15 7 we only show bug types of the first 24 hours.",
+      "context_before": "i00000057/uni00000056025050075010001250150017502000/uni00000051/uni00000058/uni00000050/uni00000045/uni00000048/uni00000055/uni00000003/uni00000052/uni00000049/uni00000003/uni00000054/uni00000058/uni00000048/uni00000055/uni0000004c/uni00000048/uni00000056/uni00000003/uni00000012/uni00000003/uni0000004e number of queries Fig. 12. Effect of parallel search. Table 7. Unique Logic Bugs Reported by TQS and SQLancer [ 5] DatabaseTQS SQLancer all bugs join bugs aggregation bugs all bugs join bugs aggregation bugs MySQL 13 9 4 5 4 0 MariaDB 8 5 3 7 6 0 TiDB 10 5 5 9 5 0 TOTAL 31 19 12 14 15 0 Table 8.",
+      "context_after": "Figure 11illustrates that the number of bugs increases linearly with the testing time, while the number of bug types is not. This indicates that most bugs are caused by a small set of improperly implemented operators. InSection 4,webrieflyexplaintheideaofusing KQEtobuildadistributedcomputationframework to speed up the testing process. We show our results in Figure 12by deploying the framework on the different number of clients. We run our experiments on MySQL for 24 hours and the results show that using parallel computation can effectively facilitate the testing process. 5.3.2 In-depthAnalysis",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 29,
+      "char_offset": 85567,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Table7comparesthenumberofuniquebugsdetectedby TQS and SQLancer across three DBMSs: MySQL, MariaDB, and TiDB.",
+      "context_before": "ustrates that the number of bugs increases linearly with the testing time, while the number of bug types is not. This indicates that most bugs are caused by a small set of improperly implemented operators. InSection 4,webrieflyexplaintheideaofusing KQEtobuildadistributedcomputationframework to speed up the testing process. We show our results in Figure 12by deploying the framework on the different number of clients. We run our experiments on MySQL for 24 hours and the results show that using parallel computation can effectively facilitate the testing process. 5.3.2 In-depthAnalysisofUniqueBugs.",
+      "context_after": "As shown, TQS consistently detects more unique bugs than SQLancer. For example, TQS detects 13 unique bugs in MySQL compared to SQLancer’s 5. Notably, TQS also detects aggregation bugs, which SQLancer fails to identify, such as the 4 aggregation bugs in MySQL and 5 in TiDB. This highlights TQS’s superior ability to detect a wider range of unique bugs, particularly in aggregation queries. Furthermore, we present the comparison of join bugs detected by both TQS and SQLancer in Table8. Among these bugs, some are reproduced by both TQS and SQLancer. As shown in Table 9, indicating an overlap in t",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 29,
+      "char_offset": 86455,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As shown, TQS consistently detects more unique bugs than SQLancer.",
+      "context_before": "ot. This indicates that most bugs are caused by a small set of improperly implemented operators. InSection 4,webrieflyexplaintheideaofusing KQEtobuildadistributedcomputationframework to speed up the testing process. We show our results in Figure 12by deploying the framework on the different number of clients. We run our experiments on MySQL for 24 hours and the results show that using parallel computation can effectively facilitate the testing process. 5.3.2 In-depthAnalysisofUniqueBugs. Table7comparesthenumberofuniquebugsdetectedby TQS and SQLancer across three DBMSs: MySQL, MariaDB, and TiDB.",
+      "context_after": "For example, TQS detects 13 unique bugs in MySQL compared to SQLancer’s 5. Notably, TQS also detects aggregation bugs, which SQLancer fails to identify, such as the 4 aggregation bugs in MySQL and 5 in TiDB. This highlights TQS’s superior ability to detect a wider range of unique bugs, particularly in aggregation queries. Furthermore, we present the comparison of join bugs detected by both TQS and SQLancer in Table8. Among these bugs, some are reproduced by both TQS and SQLancer. As shown in Table 9, indicating an overlap in the join bugs detected by both tools. However, TQS also identifies s",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 29,
+      "char_offset": 86564,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, TQS detects 13 unique bugs in MySQL compared to SQLancer’s 5.",
+      "context_before": "operly implemented operators. InSection 4,webrieflyexplaintheideaofusing KQEtobuildadistributedcomputationframework to speed up the testing process. We show our results in Figure 12by deploying the framework on the different number of clients. We run our experiments on MySQL for 24 hours and the results show that using parallel computation can effectively facilitate the testing process. 5.3.2 In-depthAnalysisofUniqueBugs. Table7comparesthenumberofuniquebugsdetectedby TQS and SQLancer across three DBMSs: MySQL, MariaDB, and TiDB. As shown, TQS consistently detects more unique bugs than SQLancer.",
+      "context_after": "Notably, TQS also detects aggregation bugs, which SQLancer fails to identify, such as the 4 aggregation bugs in MySQL and 5 in TiDB. This highlights TQS’s superior ability to detect a wider range of unique bugs, particularly in aggregation queries. Furthermore, we present the comparison of join bugs detected by both TQS and SQLancer in Table8. Among these bugs, some are reproduced by both TQS and SQLancer. As shown in Table 9, indicating an overlap in the join bugs detected by both tools. However, TQS also identifies several bugs that SQLancer does not reproduce. This comparison highlights th",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 29,
+      "char_offset": 86631,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Notably, TQS also detects aggregation bugs, which SQLancer fails to identify, such as the 4 aggregation bugs in MySQL and 5 in TiDB.",
+      "context_before": "tobuildadistributedcomputationframework to speed up the testing process. We show our results in Figure 12by deploying the framework on the different number of clients. We run our experiments on MySQL for 24 hours and the results show that using parallel computation can effectively facilitate the testing process. 5.3.2 In-depthAnalysisofUniqueBugs. Table7comparesthenumberofuniquebugsdetectedby TQS and SQLancer across three DBMSs: MySQL, MariaDB, and TiDB. As shown, TQS consistently detects more unique bugs than SQLancer. For example, TQS detects 13 unique bugs in MySQL compared to SQLancer’s 5.",
+      "context_after": "This highlights TQS’s superior ability to detect a wider range of unique bugs, particularly in aggregation queries. Furthermore, we present the comparison of join bugs detected by both TQS and SQLancer in Table8. Among these bugs, some are reproduced by both TQS and SQLancer. As shown in Table 9, indicating an overlap in the join bugs detected by both tools. However, TQS also identifies several bugs that SQLancer does not reproduce. This comparison highlights that while both tools detect some of the same join bugs, TQS has a broader detection capability and is able to identify unique bugs tha",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 29,
+      "char_offset": 86706,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Furthermore, we present the comparison of join bugs detected by both TQS and SQLancer in Table8.",
+      "context_before": "llel computation can effectively facilitate the testing process. 5.3.2 In-depthAnalysisofUniqueBugs. Table7comparesthenumberofuniquebugsdetectedby TQS and SQLancer across three DBMSs: MySQL, MariaDB, and TiDB. As shown, TQS consistently detects more unique bugs than SQLancer. For example, TQS detects 13 unique bugs in MySQL compared to SQLancer’s 5. Notably, TQS also detects aggregation bugs, which SQLancer fails to identify, such as the 4 aggregation bugs in MySQL and 5 in TiDB. This highlights TQS’s superior ability to detect a wider range of unique bugs, particularly in aggregation queries.",
+      "context_after": "Among these bugs, some are reproduced by both TQS and SQLancer. As shown in Table 9, indicating an overlap in the join bugs detected by both tools. However, TQS also identifies several bugs that SQLancer does not reproduce. This comparison highlights that while both tools detect some of the same join bugs, TQS has a broader detection capability and is able to identify unique bugs that SQLancer fails to catch. ACM Trans. Datab. Syst., Vol. 51, No. 1, Article 3. Publication date: September 2025. 3:30 X. Tang et al. Table 9. The Reproduced Join Bug IDs Reported by TQS and SQLancer [ 5] Database",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 29,
+      "char_offset": 86955,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Among these bugs, some are reproduced by both TQS and SQLancer.",
+      "context_before": "gs. Table7comparesthenumberofuniquebugsdetectedby TQS and SQLancer across three DBMSs: MySQL, MariaDB, and TiDB. As shown, TQS consistently detects more unique bugs than SQLancer. For example, TQS detects 13 unique bugs in MySQL compared to SQLancer’s 5. Notably, TQS also detects aggregation bugs, which SQLancer fails to identify, such as the 4 aggregation bugs in MySQL and 5 in TiDB. This highlights TQS’s superior ability to detect a wider range of unique bugs, particularly in aggregation queries. Furthermore, we present the comparison of join bugs detected by both TQS and SQLancer in Table8.",
+      "context_after": "As shown in Table 9, indicating an overlap in the join bugs detected by both tools. However, TQS also identifies several bugs that SQLancer does not reproduce. This comparison highlights that while both tools detect some of the same join bugs, TQS has a broader detection capability and is able to identify unique bugs that SQLancer fails to catch. ACM Trans. Datab. Syst., Vol. 51, No. 1, Article 3. Publication date: September 2025. 3:30 X. Tang et al. Table 9. The Reproduced Join Bug IDs Reported by TQS and SQLancer [ 5] Database BugIDs of SQLancer BugIDs of TQS MySQL112,243 106,715 112,264 10",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 29,
+      "char_offset": 87052,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M25",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "However, TQS also identifies several bugs that SQLancer does not reproduce.",
+      "context_before": "more unique bugs than SQLancer. For example, TQS detects 13 unique bugs in MySQL compared to SQLancer’s 5. Notably, TQS also detects aggregation bugs, which SQLancer fails to identify, such as the 4 aggregation bugs in MySQL and 5 in TiDB. This highlights TQS’s superior ability to detect a wider range of unique bugs, particularly in aggregation queries. Furthermore, we present the comparison of join bugs detected by both TQS and SQLancer in Table8. Among these bugs, some are reproduced by both TQS and SQLancer. As shown in Table 9, indicating an overlap in the join bugs detected by both tools.",
+      "context_after": "This comparison highlights that while both tools detect some of the same join bugs, TQS has a broader detection capability and is able to identify unique bugs that SQLancer fails to catch. ACM Trans. Datab. Syst., Vol. 51, No. 1, Article 3. Publication date: September 2025. 3:30 X. Tang et al. Table 9. The Reproduced Join Bug IDs Reported by TQS and SQLancer [ 5] Database BugIDs of SQLancer BugIDs of TQS MySQL112,243 106,715 112,264 109,211 MariaDB32,105 28,215 32,106 28,216 32,108 29,695 32,186 28,214 TiDB46,580 33,039 47,345 33,042 Table 10. Ablation Test Over the Effect of Model Compositio",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 29,
+      "char_offset": 87200,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M26",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This comparison highlights that while both tools detect some of the same join bugs, TQS has a broader detection capability and is able to identify unique bugs that SQLancer fails to catch.",
+      "context_before": "ySQL compared to SQLancer’s 5. Notably, TQS also detects aggregation bugs, which SQLancer fails to identify, such as the 4 aggregation bugs in MySQL and 5 in TiDB. This highlights TQS’s superior ability to detect a wider range of unique bugs, particularly in aggregation queries. Furthermore, we present the comparison of join bugs detected by both TQS and SQLancer in Table8. Among these bugs, some are reproduced by both TQS and SQLancer. As shown in Table 9, indicating an overlap in the join bugs detected by both tools. However, TQS also identifies several bugs that SQLancer does not reproduce.",
+      "context_after": "ACM Trans. Datab. Syst., Vol. 51, No. 1, Article 3. Publication date: September 2025. 3:30 X. Tang et al. Table 9. The Reproduced Join Bug IDs Reported by TQS and SQLancer [ 5] Database BugIDs of SQLancer BugIDs of TQS MySQL112,243 106,715 112,264 109,211 MariaDB32,105 28,215 32,106 28,216 32,108 29,695 32,186 28,214 TiDB46,580 33,039 47,345 33,042 Table 10. Ablation Test Over the Effect of Model Composition DBMS Approach Query Graph Diversity Query Cube Diversity Bug Count MySQLTQS 460k 236k 63 𝑇𝑄𝑆!𝑁𝑜𝑖𝑠𝑒 460k 236k 24 𝑇𝑄𝑆!𝐺𝑇 460k 236k 41 𝑇𝑄𝑆!𝐾𝑄𝐸 228k 153k 26 MariaDBTQS 475k 220k 52 𝑇𝑄𝑆!𝑁𝑜𝑖𝑠𝑒",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 29,
+      "char_offset": 87276,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M27",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The Reproduced Join Bug IDs Reported by TQS and SQLancer [ 5] Database BugIDs of SQLancer BugIDs of TQS MySQL112,243 106,715 112,264 109,211 MariaDB32,105 28,215 32,106 28,216 32,108 29,695 32,186 28,214 TiDB46,580 33,039 47,345 33,042 Table 10.",
+      "context_before": "the comparison of join bugs detected by both TQS and SQLancer in Table8. Among these bugs, some are reproduced by both TQS and SQLancer. As shown in Table 9, indicating an overlap in the join bugs detected by both tools. However, TQS also identifies several bugs that SQLancer does not reproduce. This comparison highlights that while both tools detect some of the same join bugs, TQS has a broader detection capability and is able to identify unique bugs that SQLancer fails to catch. ACM Trans. Datab. Syst., Vol. 51, No. 1, Article 3. Publication date: September 2025. 3:30 X. Tang et al. Table 9.",
+      "context_after": "Ablation Test Over the Effect of Model Composition DBMS Approach Query Graph Diversity Query Cube Diversity Bug Count MySQLTQS 460k 236k 63 𝑇𝑄𝑆!𝑁𝑜𝑖𝑠𝑒 460k 236k 24 𝑇𝑄𝑆!𝐺𝑇 460k 236k 41 𝑇𝑄𝑆!𝐾𝑄𝐸 228k 153k 26 MariaDBTQS 475k 220k 52 𝑇𝑄𝑆!𝑁𝑜𝑖𝑠𝑒 475k 220k 30 𝑇𝑄𝑆!𝐺𝑇 475k 220k 36 𝑇𝑄𝑆!𝐾𝑄𝐸 234k 146k 24 TiDBTQS 462k 234k 68 𝑇𝑄𝑆!𝑁𝑜𝑖𝑠𝑒 462k 234k 40 𝑇𝑄𝑆!𝐺𝑇 462k 234k 42 𝑇𝑄𝑆!𝐾𝑄𝐸 231k 169k 38 PolarDBTQS 465k 236k 43 𝑇𝑄𝑆!𝑁𝑜𝑖𝑠𝑒 465k 236k 22 𝑇𝑄𝑆!𝐺𝑇 465k 236k 28 𝑇𝑄𝑆!𝐾𝑄𝐸 225k 156k 25 5.4 Ablation Studies In this section, we perform ablation experiments over some facets of TQS in order to better understand their role",
+      "section": "5.3 Comparison with Existing Tools",
+      "page": 30,
+      "char_offset": 87580,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "citation_marker",
+      "surface": "[ 5]",
+      "technique": "dqp",
+      "sentence": "𝐺𝑇isequivalenttoDQP,asoutlinedin[ 5].",
+      "context_before": "differential testing. As shown in Table 10, some bugscannotberevealedbydifferentialtesting,whileusingground-truthresults,wecansuccessfully identify them. Figure 1(b) illustrates an aggregation bug in MariaDB version 11.3.2. The query result remains consistent across different plans. However, the query result differs from the ground truth, indicating a logic bug. Moreover, in PolarDB, differential testing failed to detect 8 bugs, such as the hash join bug in version 8.0.18 shown in Listing 13, where identical results from different plans masked the underlying issue. Theverificationmethodof 𝑇𝑄𝑆!",
+      "context_after": "DQPadoptsdifferential testing, a strategy aimed at replicating bugs identified by TQS(join) [ 56], which involves ground truth verification for join queries. While DQP claims to have uncovered 26 previously unidentified bugs, our analysis shows that most of these bugs closely resemble those already found by TQS(join). Table8compares the join bugs detected by DQP with those from TQS(join), revealing that only 7 of the bugs are truly distinct. The other bugs stem from single-table queries with complex expressions, differing from the join-query scenarios targeted by TQS(join). KQE vs. No-KQE. KQE",
+      "section": "5.4 Ablation Studies",
+      "page": 31,
+      "char_offset": 90357,
+      "cited_reference": {
+        "number": 5,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans. Proc.ACMManag. Data2, 3 (2024), 188.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "While DQP claims to have uncovered 26 previously unidentified bugs, our analysis shows that most of these bugs closely resemble those already found by TQS(join).",
+      "context_before": "ug in MariaDB version 11.3.2. The query result remains consistent across different plans. However, the query result differs from the ground truth, indicating a logic bug. Moreover, in PolarDB, differential testing failed to detect 8 bugs, such as the hash join bug in version 8.0.18 shown in Listing 13, where identical results from different plans masked the underlying issue. Theverificationmethodof 𝑇𝑄𝑆!𝐺𝑇isequivalenttoDQP,asoutlinedin[ 5].DQPadoptsdifferential testing, a strategy aimed at replicating bugs identified by TQS(join) [ 56], which involves ground truth verification for join queries.",
+      "context_after": "Table8compares the join bugs detected by DQP with those from TQS(join), revealing that only 7 of the bugs are truly distinct. The other bugs stem from single-table queries with complex expressions, differing from the join-query scenarios targeted by TQS(join). KQE vs. No-KQE. KQE is designed to target error-prone areas, and generates a diverse range of query structures, thereby avoiding redundancy in query testing. In Table 10, we observe that TQS is superior to the 𝑇𝑄𝑆!𝐾𝑄𝐸on the four databases, indicating the effectiveness of applying KQE to generate queries. Note that because iterating over",
+      "section": "5.4 Ablation Studies",
+      "page": 31,
+      "char_offset": 90551,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "Table8compares the join bugs detected by DQP with those from TQS(join), revealing that only 7 of the bugs are truly distinct.",
+      "context_before": "gic bug. Moreover, in PolarDB, differential testing failed to detect 8 bugs, such as the hash join bug in version 8.0.18 shown in Listing 13, where identical results from different plans masked the underlying issue. Theverificationmethodof 𝑇𝑄𝑆!𝐺𝑇isequivalenttoDQP,asoutlinedin[ 5].DQPadoptsdifferential testing, a strategy aimed at replicating bugs identified by TQS(join) [ 56], which involves ground truth verification for join queries. While DQP claims to have uncovered 26 previously unidentified bugs, our analysis shows that most of these bugs closely resemble those already found by TQS(join).",
+      "context_after": "The other bugs stem from single-table queries with complex expressions, differing from the join-query scenarios targeted by TQS(join). KQE vs. No-KQE. KQE is designed to target error-prone areas, and generates a diverse range of query structures, thereby avoiding redundancy in query testing. In Table 10, we observe that TQS is superior to the 𝑇𝑄𝑆!𝐾𝑄𝐸on the four databases, indicating the effectiveness of applying KQE to generate queries. Note that because iterating over all possible isomorphic sets for a graph is an NP-complete problem, there is no way to perform exhaustive testing. The intuit",
+      "section": "5.4 Ablation Studies",
+      "page": 31,
+      "char_offset": 90713,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 51] is a current state-of-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours.",
+      "context_before": "Publication date: September 2025. 3:32 X. Tang et al. attention. Generation-based testings [ 26,34,36,41,48,61,65] have been adopted for extensive testings on DBMSs for purposes such as bug-finding and benchmarking. SQLSmith is a widely-used generation-based DBMS tester [ 54]. It synthesizes a schema from initial databases and generates limitedtypesofqueries,whosetargetisatthecodecoverage.Squirrelfocusesongeneratingqueries to detect memory corruption bugs [ 65]. All above random generators are mostly applied to detect crashing bugs, while our focus is the logic bug. Logic Bug Testing of DBMS.",
+      "context_after": "SQLancer proposes five approaches to detect logic bugs. PQS constructs queries to fetch a randomly selected tuple from a table [ 52]. TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 50]. NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [ 49]. CERT aims at finding performance issues through unexpected estimated cardinalities, which represent the estimated number of returned rows [4]. It derives a query to a more restricted query, whose estimated cardinality should b",
+      "section": "5.4 Ablation Studies",
+      "page": 32,
+      "char_offset": 93422,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M32",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer proposes five approaches to detect logic bugs.",
+      "context_before": "dopted for extensive testings on DBMSs for purposes such as bug-finding and benchmarking. SQLSmith is a widely-used generation-based DBMS tester [ 54]. It synthesizes a schema from initial databases and generates limitedtypesofqueries,whosetargetisatthecodecoverage.Squirrelfocusesongeneratingqueries to detect memory corruption bugs [ 65]. All above random generators are mostly applied to detect crashing bugs, while our focus is the logic bug. Logic Bug Testing of DBMS. SQLancer [ 51] is a current state-of-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours.",
+      "context_after": "PQS constructs queries to fetch a randomly selected tuple from a table [ 52]. TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 50]. NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [ 49]. CERT aims at finding performance issues through unexpected estimated cardinalities, which represent the estimated number of returned rows [4]. It derives a query to a more restricted query, whose estimated cardinality should be no more than that of the original query. DQP aims at f",
+      "section": "5.4 Ablation Studies",
+      "page": 32,
+      "char_offset": 93549,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M33",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS constructs queries to fetch a randomly selected tuple from a table [ 52].",
+      "context_before": "as bug-finding and benchmarking. SQLSmith is a widely-used generation-based DBMS tester [ 54]. It synthesizes a schema from initial databases and generates limitedtypesofqueries,whosetargetisatthecodecoverage.Squirrelfocusesongeneratingqueries to detect memory corruption bugs [ 65]. All above random generators are mostly applied to detect crashing bugs, while our focus is the logic bug. Logic Bug Testing of DBMS. SQLancer [ 51] is a current state-of-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours. SQLancer proposes five approaches to detect logic bugs.",
+      "context_after": "TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 50]. NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [ 49]. CERT aims at finding performance issues through unexpected estimated cardinalities, which represent the estimated number of returned rows [4]. It derives a query to a more restricted query, whose estimated cardinality should be no more than that of the original query. DQP aims at finding logic bugs in database systems by checking whether the query plans of t",
+      "section": "5.4 Ablation Studies",
+      "page": 32,
+      "char_offset": 93605,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 50].",
+      "context_before": "BMS tester [ 54]. It synthesizes a schema from initial databases and generates limitedtypesofqueries,whosetargetisatthecodecoverage.Squirrelfocusesongeneratingqueries to detect memory corruption bugs [ 65]. All above random generators are mostly applied to detect crashing bugs, while our focus is the logic bug. Logic Bug Testing of DBMS. SQLancer [ 51] is a current state-of-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours. SQLancer proposes five approaches to detect logic bugs. PQS constructs queries to fetch a randomly selected tuple from a table [ 52].",
+      "context_after": "NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [ 49]. CERT aims at finding performance issues through unexpected estimated cardinalities, which represent the estimated number of returned rows [4]. It derives a query to a more restricted query, whose estimated cardinality should be no more than that of the original query. DQP aims at finding logic bugs in database systems by checking whether the query plans of the same query perform consistently [ 5]. SQLancer uses differential testing as the core to test the logic bugs of dat",
+      "section": "5.4 Ablation Studies",
+      "page": 32,
+      "char_offset": 93683,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M35",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [ 49].",
+      "context_before": "hecodecoverage.Squirrelfocusesongeneratingqueries to detect memory corruption bugs [ 65]. All above random generators are mostly applied to detect crashing bugs, while our focus is the logic bug. Logic Bug Testing of DBMS. SQLancer [ 51] is a current state-of-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours. SQLancer proposes five approaches to detect logic bugs. PQS constructs queries to fetch a randomly selected tuple from a table [ 52]. TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 50].",
+      "context_after": "CERT aims at finding performance issues through unexpected estimated cardinalities, which represent the estimated number of returned rows [4]. It derives a query to a more restricted query, whose estimated cardinality should be no more than that of the original query. DQP aims at finding logic bugs in database systems by checking whether the query plans of the same query perform consistently [ 5]. SQLancer uses differential testing as the core to test the logic bugs of databases. However, our TQS focuses on multi-table testing and group testing, and is oriented toward testing more error-prone",
+      "section": "5.4 Ablation Studies",
+      "page": 32,
+      "char_offset": 93800,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M36",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "CERT aims at finding performance issues through unexpected estimated cardinalities, which represent the estimated number of returned rows [4].",
+      "context_before": "mostly applied to detect crashing bugs, while our focus is the logic bug. Logic Bug Testing of DBMS. SQLancer [ 51] is a current state-of-the-art tool in testing DBMS for logic bugs and is the most closely related work to ours. SQLancer proposes five approaches to detect logic bugs. PQS constructs queries to fetch a randomly selected tuple from a table [ 52]. TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 50]. NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [ 49].",
+      "context_after": "It derives a query to a more restricted query, whose estimated cardinality should be no more than that of the original query. DQP aims at finding logic bugs in database systems by checking whether the query plans of the same query perform consistently [ 5]. SQLancer uses differential testing as the core to test the logic bugs of databases. However, our TQS focuses on multi-table testing and group testing, and is oriented toward testing more error-prone points. In the previous work [ 56], TQS(join) was designed specifically to find logic bugs in join optimizations. TQS(join) is a promising app",
+      "section": "5.4 Ablation Studies",
+      "page": 32,
+      "char_offset": 93922,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M37",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "DQP aims at finding logic bugs in database systems by checking whether the query plans of the same query perform consistently [ 5].",
+      "context_before": "ct logic bugs. PQS constructs queries to fetch a randomly selected tuple from a table [ 52]. TLP decomposes a query into three partitioning queries, each of which computes its result on a selected tuple [ 50]. NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [ 49]. CERT aims at finding performance issues through unexpected estimated cardinalities, which represent the estimated number of returned rows [4]. It derives a query to a more restricted query, whose estimated cardinality should be no more than that of the original query.",
+      "context_after": "SQLancer uses differential testing as the core to test the logic bugs of databases. However, our TQS focuses on multi-table testing and group testing, and is oriented toward testing more error-prone points. In the previous work [ 56], TQS(join) was designed specifically to find logic bugs in join optimizations. TQS(join) is a promising approach that splits a given input table into several sub-tables and validates the results of the queries that join these sub-tables by retrieving the given table. However, in our approach, we extend its methods to SPJG queries by constructing a cube, which all",
+      "section": "5.4 Ablation Studies",
+      "page": 32,
+      "char_offset": 94191,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M38",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer uses differential testing as the core to test the logic bugs of databases.",
+      "context_before": "tioning queries, each of which computes its result on a selected tuple [ 50]. NoRec compares the results of randomly-generated optimized queries and rewritten queries that DBMS cannot optimize [ 49]. CERT aims at finding performance issues through unexpected estimated cardinalities, which represent the estimated number of returned rows [4]. It derives a query to a more restricted query, whose estimated cardinality should be no more than that of the original query. DQP aims at finding logic bugs in database systems by checking whether the query plans of the same query perform consistently [ 5].",
+      "context_after": "However, our TQS focuses on multi-table testing and group testing, and is oriented toward testing more error-prone points. In the previous work [ 56], TQS(join) was designed specifically to find logic bugs in join optimizations. TQS(join) is a promising approach that splits a given input table into several sub-tables and validates the results of the queries that join these sub-tables by retrieving the given table. However, in our approach, we extend its methods to SPJG queries by constructing a cube, which allows us to verify the correctness of aggregate operations. Additionally, we use prior",
+      "section": "5.4 Ablation Studies",
+      "page": 32,
+      "char_offset": 94323,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M5",
+        "M7",
+        "M20",
+        "M23"
+      ],
+      "describes_as_state_of_the_art": [
+        "M5",
+        "M31"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:12:40Z",
+    "is_model_written": true,
+    "summary": "TQS targets logic bugs in Select-Project-Join-Group By query optimisation, arguing that existing generation-based tools rely on random testing and under-exercise error-prone areas such as multi-table joins and grouped aggregation. Its Data-guided Schema and Query Generation solves the ground-truth problem for multi-table queries by deriving data from dimensionally aggregated data cubes, mapping it into a wide table, normalising that into a testing schema, and keeping a bitmap index to track expected results. Knowledge-guided Query Space Exploration avoids repeating the same searches. Across four DBMSs it found 226 bugs in 24 hours.",
+    "narrative": "SQLancer is the paper's reference point and its baseline: four of SQLancer's methods -- PQS, TLP, NoREC and DQP -- are the four baselines TQS is measured against, and the paper calls SQLancer the state-of-the-art approach for detecting logic bugs while arguing it is not designed for multi-table queries.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "definition",
+      "M4": "motivation",
+      "M5": "state_of_the_art",
+      "M6": "motivation",
+      "M7": "baseline",
+      "M8": "definition",
+      "M9": "definition",
+      "M10": "definition",
+      "M11": "definition",
+      "M12": "incidental",
+      "M13": "result_comparison",
+      "M14": "baseline"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Nothing shown says TQS is built on SQLancer's codebase; the paper's own generation is derived from data cubes, and SQLancer appears as the thing being measured against."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "TQS presents its own ground-truth construction for multi-table queries rather than generalising PQS, TLP or NoREC."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M5",
+          "M7",
+          "M14"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M5",
+            "sentence": "We compare TQS with SQLancer,3which is the state-of-the-art approach to detecting logic bugs in databases.",
+            "section": "4.2 Query Generation Space",
+            "page": 18
+          },
+          {
+            "mention_id": "M7",
+            "sentence": "We use four methods in SQLancer as our baselines.",
+            "section": "4.2 Query Generation Space",
+            "page": 18
+          },
+          {
+            "mention_id": "M14",
+            "sentence": "to the compatibility problem, SQLancer implements different approaches on different databases (PQS, TLP and DQP on MySQL and PolarDB; NoRec and DQP on MariaDB; TLP and DQP on TiDB).",
+            "section": "5.3 Comparison with Existing Tools",
+            "page": 28
+          }
+        ],
+        "reasoning": "M7 states four methods in SQLancer are used as the baselines, and M8 to M11 name them as PQS, TLP, NoREC and DQP.",
+        "techniques": [
+          "pqs",
+          "tlp",
+          "norec",
+          "dqp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M5"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M5",
+            "sentence": "We compare TQS with SQLancer,3which is the state-of-the-art approach to detecting logic bugs in databases.",
+            "section": "4.2 Query Generation Space",
+            "page": 18
+          }
+        ],
+        "reasoning": "M5 calls SQLancer the state-of-the-art approach to detecting logic bugs in databases."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3769779.json b/_data/papers/paper_doi_10_1145_3769779.json
new file mode 100644
index 0000000..649f1e9
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3769779.json
@@ -0,0 +1,943 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:30:16Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3769779",
+    "title": "Detecting Logic Bugs in DBMSs via Equivalent Data Construction",
+    "authors": [
+      "Wenqian Deng",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Yu Jiang"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Manag. Data",
+    "doi": "10.1145/3769779",
+    "arxiv_id": null,
+    "s2_paper_id": "7762da5097d2f7d3482670909ad1d61c216119d6",
+    "url": "https://doi.org/10.1145/3769779",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3769779",
+      "retrieved_at": "2026-09-09T01:30:16Z",
+      "chars": 90335,
+      "content_sha256": "sha256:ad87be389a15239be541592a4e9e39c136902887e195ea97e6cb594ccd6bcf1d"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 25,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2352
+      },
+      {
+        "number": "2",
+        "title": "Background and Motivation",
+        "start": 11858
+      },
+      {
+        "number": "2.1",
+        "title": "Data Types and Operations",
+        "start": 11886
+      },
+      {
+        "number": "2.2",
+        "title": "Limitations of Existing Approaches",
+        "start": 14290
+      },
+      {
+        "number": "2.3",
+        "title": "Definitions of Data Equivalence",
+        "start": 16463
+      },
+      {
+        "number": "3.1",
+        "title": "Overview of EDC",
+        "start": 20811
+      },
+      {
+        "number": "3.2",
+        "title": "Equivalent Data Construction",
+        "start": 22627
+      },
+      {
+        "number": "3.3",
+        "title": "Operation-Oriented Query Generation",
+        "start": 29746
+      },
+      {
+        "number": "5.1",
+        "title": "Evaluation Setup",
+        "start": 41048
+      },
+      {
+        "number": "5.2",
+        "title": "Detected Logic Bugs",
+        "start": 42687
+      },
+      {
+        "number": "5.3",
+        "title": "Comprehensive Bug Analysis",
+        "start": 45393
+      },
+      {
+        "number": "5.4",
+        "title": "Selected Bugs Found by EDC",
+        "start": 50956
+      },
+      {
+        "number": "5.5",
+        "title": "Comparison with Existing Techniques",
+        "start": 64231
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2024. ClickHouse Operators. https://clickhouse .com/docs/en/sql-reference/operators. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2024. MDEV-34123. https://jira.mariadb.org/browse/MDEV-34123. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2024. MySQL Functions and Operators. https://dev .mysql.com/doc/refman/8 .0/en/built-in-function-reference .html. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2024. Percona. https://github.com/percona/percona-server. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2025. ClickHouse. https://clickhouse.com/. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2025. EDC. https://github.com/THU-WingTecher/EDC. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2025. Integer Types (Exact Value). https://dev .mysql.com/doc/refman/8 .0/en/integer-types .html. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2025. MariaDB. https://mariadb.org/. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2025. MariaDB Data Types. https://mariadb.com/kb/en/data-types/. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2025. MySQL. https://www.mysql.com/. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2025. MySQL Data Types. https://dev .mysql.com/doc/refman/8 .4/en/data-types .html. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "2025. OceanBase. https://github.com/oceanbase/oceanbase. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "2025. PostgreSQL. https://www.postgresql.org/. Accessed: November 11, 2025. 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. 314:24 Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "2025. SQL Data Types. https://www.databasestar.com/sql-data-types/. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "2025. TiDB. https://github.com/pingcap/tidb. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Abdullah Alsharif, Gregory M Kapfhammer, and Phil McMinn. 2018. DOMINO: Fast and Effective Test Data Generation for Relational Database Schemas. In Validation (ICST). IEEE, 12–22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Daniel Bartholomew. 2014.MariaDB cookbook. Packt Publishing Ltd.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jeroen Castelein, Maurício Aniche, Mozhan Soltani, Annibale Panichella, and Arie Van Deursen. 2018. Search-Based Test Data Generation for SQL Queries. InProceedings of the 40th International Conference on Software Engineering. ACM, 1220–1230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Bikash Chandra, Bhupesh Chawda, Biplab Kar, K. V. Maheshwara Reddy, Shetal Shah, and S. Sudarshan. 2015. Data Generation for Testing and Grading SQL Queries. 24, 6 (2015).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2024. CONI: Detecting Database Connector Bugs via State-Aware Test Case Generation. In Engineering (ICSE). IEEE Computer Society, 26–37.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, and Yu Jiang. 2024. Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. In 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023. Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis. In2023 USENIX Annual Technical Conference (USENIX ATC 23). 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Dongxu Huang, Qi Liu, Qiu Cui, Zhuhe Fang, Xiaoyu Ma, Fei Xu, Li Shen, Liu Tang, Yuxing Zhou, Menglong Huang, et al. 2020. TiDB: a Raft-based HTAP database.Proceedings of the VLDB Endowment13, 12 (2020), 3072–3084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Solid IT. 2022. DB-Engines Ranking of Relational DBMS. https://db-engines .com/en/ranking/relational +dbms. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. In32nd USENIX Security Symposium. 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting Logic Bugs in Database Engines via Equivalent Expression Transformation. InProceedings of the 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O Laleye, and Sarfraz Khurshid. 2008. Query-aware test generation using a relational constraint solver. In2008 23rd IEEE/ACM International Conference on Automated Software Engineering. IEEE, 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-oriented DBMS Fuzzing. InProceedings of IEEE International Conference on Data Engineering (ICDE).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Yiyuan Bai, Qiang Zhang, and Yu Jiang. 2024. WingFuzz: Implementing Continuous Fuzzing for DBMSs. In2024 USENIX Annual Technical Conference. 479–492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting logical bugs of DBMS with coverage-based guidance. In31st USENIX Security Symposium (USENIX Security 22). 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Eric Lo, Carsten Binnig, Donald Kossmann, M Tamer Özsu, and Wing-Kai Hon. 2010. A framework for testing DBMS features.The VLDB Journal19 (2010), 203–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Roger C Lyndon. 1950. The representation of relational algebras.Annals of mathematics51, 3 (1950), 707–729.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "PingCAP. [n. d.]. TiDB. https://github.com/pingcap/tidb. Accessed: November 11, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Raghu Ramakrishnan and Johannes Gehrke. 2002.Database management systems. McGraw-Hill, Inc.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 40,
+      "text": "Anupam Sanghi, Raghav Sood, Dharmendra Singh, Jayant R Haritsa, and Srikanta Tirthapura. 2018. HYDRA: a dynamic big data regenerator.Proceedings of the VLDB Endowment11, 12 (2018), 1974–1977. 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. Detecting Logic Bugs in DBMSs via Equivalent Data Construction 314:25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: A Random SQL Query Generator. https: //github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Shetal Shah, S. Sudarshan, Suhas Kajbaje, Sandeep Patidar, Bhanu Pratap Gupta, and Devang Vira. 2011. Generating Test Data for Killing SQL Mutants: A Constraint-Based Approach. In Engineering. IEEE, 1175–1186.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Donald R Slutz. 1998. Massive Stochastic Testing of SQL. 98 (1998), 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing database systems via differential query execution. In Engineering (ICSE). IEEE, 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Jiansen Song, Wensheng Dou, Yu Gao, Ziyu Cui, Yingying Zheng, Dong Wang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction.Proceedings of the VLDB Endowment17, 8 (2024), 1884–1897.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting logic bugs of join optimizations in dbms.Proceedings of the ACM on Management of Data1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Qingshuai Wang, Hao Li, Zirui Hu, Rong Zhang, Chengcheng Yang, Peng Cai, Xuan Zhou, and Aoying Zhou. 2024. Mirage: Generating Enormous Databases for Complex Workloads. In Engineering (ICDE). IEEE, 3989–4001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Qingshuai Wang, Yuming Li, Rong Zhang, Ke Shu, Zhenjie Zhang, and Aoying Zhou. 2022. A scalable query-aware enormous database generator for database evaluation.IEEE Transactions on Knowledge and Data Engineering35, 5 (2022), 4395–4410.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Michael Widenius, David Axmark, and Kaj Arno. 2002.MySQL reference manual: documentation from the source. \" O’Reilly Media, Inc.\".",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2024. PUPPY: Finding Performance Degradation Bugs in DBMSs via Limited-Optimization Plan Construction. In Software Engineering (ICSE). IEEE Computer Society, 560–571.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. 2022. Unicorn: Detect Runtime Errors in Time-series Databases with Hybrid Input Synthesis. InISSTA ’22: 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, South Korea, July 18 - 22, 2022. ACM, 251–262.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Jingyi Yang, Peizhi Wu, Gao Cong, Tieying Zhang, and Xiao He. 2022. SAM: Database generation from query workloads with supervised autoregressive models. InProceedings of the 2022 International Conference on Management of Data. 1542–1555.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. InThe ACM Conference on Computer and Communications Security (CCS), 2020. Received April 2025; revised July 2025; accepted August 2025 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 39,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For instance, EET [ 27] targets expression-level transformations, TLP [ 37] partitions queries based on predicate logic, and NoREC [36] rewrites queries to disable DBMS optimizations.",
+      "context_before": "n, and optimization rules. For instance, data operations may involve implicit type conversions and varying treatments of different data types, as well as handling implicit coercions, type overflows, or precision loss. Thus, logic bugs related to data operations in DBMSs are more likely to arise not from individual operations in isolation, but from subtle inconsistencies that emerge when multiple operations and data types interact in complex ways. This complexity makes the implementation of data operations inherently error-prone. There have been several studies on detecting logic bugs in DBMSs.",
+      "context_after": "Radar [45] introduces metadata constraints to find potential bugs. While effective for exposing query planner and optimizer issues, such as incorrect rewrites, joins, or predicate logic, previous tools mainly focus on the structural aspects of query logic. They are not primarily designed to detect issues that stem from how data operations behave over different data types, such as errors caused by implicit type coercions, overflows, or precision loss. In this paper, we focus on a different class of correctness issues: logic bugs arising from data types and operations. These issues often occur",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4361,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", TLP [ 37], Radar [ 45], and EET [ 27].",
+      "context_before": "15], OceanBase [ 12] and ClickHouse [ 5]. Although these databases had undergone thorough testing,EDCstill reported a total of 54 bugs. Among these, 39 bugs have been confirmed by the developers of the corresponding DBMSs. The reported issues received active attention from the developer communities. For example, TiDB developers actively inquired about our approach and acknowledged its effectiveness, while MariaDB developers described our findings as counterintuitive, helping them uncover more issues related to data operations. We also evaluate EDCagainst 3 state-of-art testing techniques, i.e.",
+      "context_after": "In 24-hour experiments,EDCdetected 38 bugs across seven DBMSs, including 35 that were not reported by existing tools. This result underscores the distinct bug-detection scope of EDCand its potential to complement prior approaches. We make the following contributions: •We demonstrate that logic bugs arising from data operations, such as type conversions and precision errors, can be significant. However, these issues are often underrepresented in the findings of existing tools, which tend to focus primarily on query-level transformations. •We propose a novel approach EDCthat is designed to detect",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 10670,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 27,36–38,44,45]",
+      "technique": "norec",
+      "sentence": "2 Limitations of Existing Approaches Several approaches have been proposed to detect logic bugs in DBMSs [ 27,36–38,44,45], many of which operate at the query level and rely on various strategies to identify inconsistencies in query results.",
+      "context_before": "returning boolean results, such as comparisons ( >,<,=), null checks ( IS NULL ), and set-based operations ( IN,BETWEEN ), are critical for filtering; and (4) Aggregate operations, which summarize data across multiple rows, producing grouped or single-value results through functions like SUM,AVG, and MAX. The diversity of data types and operations serves as the cornerstone of a DBMS’s ability to support a broad spectrum of functionalities. By leveraging this combination, DBMSs effectively address diverse application requirements while ensuring performance, reliability, and user-friendliness. 2.",
+      "context_after": "Specifically, some approaches split or transform queries to expose discrepancies in the results (e.g., TLP [ 37], NoREC [ 36]), while others construct logically crafted SQL queries designed to return specific results in a controlled manner (e.g., PQS [ 38], DQE [ 44]). While query-level strategies have proven effective in many cases, their focus on high-level semantics or syntactic transformations means they may not specifically target the types of subtle bugs that can emerge from complex data operations or type interactions. Additionally, some techniques (e.g., EET [ 27]) apply rule-based ex",
+      "section": "2.2 Limitations of Existing Approaches",
+      "page": 4,
+      "char_offset": 14292,
+      "cited_reference": {
+        "number": 36,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": ", TLP [ 37], NoREC [ 36]), while others construct logically crafted SQL queries designed to return specific results in a controlled manner (e.",
+      "context_before": "tions serves as the cornerstone of a DBMS’s ability to support a broad spectrum of functionalities. By leveraging this combination, DBMSs effectively address diverse application requirements while ensuring performance, reliability, and user-friendliness. 2.2 Limitations of Existing Approaches Several approaches have been proposed to detect logic bugs in DBMSs [ 27,36–38,44,45], many of which operate at the query level and rely on various strategies to identify inconsistencies in query results. Specifically, some approaches split or transform queries to expose discrepancies in the results (e.g.",
+      "context_after": "g., PQS [ 38], DQE [ 44]). While query-level strategies have proven effective in many cases, their focus on high-level semantics or syntactic transformations means they may not specifically target the types of subtle bugs that can emerge from complex data operations or type interactions. Additionally, some techniques (e.g., EET [ 27]) apply rule-based expression rewriting, while others, such as Radar [ 45] focus on detecting metadatarelated logic bugs by comparing results between a database with metadata constraints and a raw database. While these methods do not rely directly on data semantics",
+      "section": "2.2 Limitations of Existing Approaches",
+      "page": 4,
+      "char_offset": 14635,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", PQS [ 38], DQE [ 44]).",
+      "context_before": "ively address diverse application requirements while ensuring performance, reliability, and user-friendliness. 2.2 Limitations of Existing Approaches Several approaches have been proposed to detect logic bugs in DBMSs [ 27,36–38,44,45], many of which operate at the query level and rely on various strategies to identify inconsistencies in query results. Specifically, some approaches split or transform queries to expose discrepancies in the results (e.g., TLP [ 37], NoREC [ 36]), while others construct logically crafted SQL queries designed to return specific results in a controlled manner (e.g.",
+      "context_after": "While query-level strategies have proven effective in many cases, their focus on high-level semantics or syntactic transformations means they may not specifically target the types of subtle bugs that can emerge from complex data operations or type interactions. Additionally, some techniques (e.g., EET [ 27]) apply rule-based expression rewriting, while others, such as Radar [ 45] focus on detecting metadatarelated logic bugs by comparing results between a database with metadata constraints and a raw database. While these methods do not rely directly on data semantics, they operate at a differ",
+      "section": "2.2 Limitations of Existing Approaches",
+      "page": 4,
+      "char_offset": 14779,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC aims to reveal bugs by transforming query semantics, but its approach is primarily focused on a specific pattern of predicate rewriting, which may limit its coverage.",
+      "context_before": "45] focus on detecting metadatarelated logic bugs by comparing results between a database with metadata constraints and a raw database. While these methods do not rely directly on data semantics, they operate at a different level of abstraction and are not primarily intended to capture data relationships. Notably, the bug illustrated in Figure 1 triggered by a query containing WHERE CONCAT(𝑎 1,𝑎2)is uniquely detected by our approach. EET operates at the syntactic level, and while its rule-based expression rewrites provide valuable insights, they may not fully capture deeper data relationships.",
+      "context_after": "Radar, which emphasizes metadata constraints, offers useful perspectives but does not manipulate data expressions directly, 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. Detecting Logic Bugs in DBMSs via Equivalent Data Construction 314:5 potentially missing deeper inconsistencies in data computation. Taken together, these approaches do not explicitly model equivalence across data types and operations, which may make it challenging to detect certain types of data-level logic bugs. 2.3 Definitions of Data Equivalence To illustrate our idea, we first present formal definitio",
+      "section": "2.2 Limitations of Existing Approaches",
+      "page": 4,
+      "char_offset": 15783,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[ 21,27,29,30,38,50]",
+      "technique": "pqs",
+      "sentence": "These DBMSs have been extensively tested by many existing approaches [ 21,27,29,30,38,50].",
+      "context_before": "approaches. 5.1 Evaluation Setup Tested DBMSs.To evaluate the generality and efficiency of EDC, we select seven popular opensource DBMSs, namely MySQL [ 10,49], MariaDB [ 8,17], Percona [ 4], PostgreSQL [ 13], TiDB [ 24,34], OceanBase [ 12], and ClickHouse [ 5], which are widely used in industry. Table 3 shows the specific versions of the tested DBMSs for the experiments. The type in Table 3 follows the prior work [ 45] and established classifications [ 25]. Specifically, we applied EDCto MySQL 9.0.1, MariaDB 11.4.3, Percona 8.4.2, PostgreSQL 16.1, TiDB 8.2, OceanBase 4.2.1, and ClickHouse 24.9.",
+      "context_after": "However, existing tools have recently identified significantly fewer bugs in these DBMSs. Consequently, discovering new bugs in these DBMSs has become highly challenging. Table 3. Target DBMSs: version, complexity, popularity, and diversity overview. DBMS Version LOC GitHub Stars Type MySQL 9.0.1 6.2M 10.8k Relational MariaDB 11.4.3 2.3M 5.6k Relational Percona 8.4.2 5.0M 1.2k Relational PostgreSQL 16.1 1.9M 18k Relational TiDB 8.2 1.5M 36k NewSQL OceanBase 4.2.1 7.5M 8.5k Distributed ClickHouse 24.9.1 2.2M 36k Analytical Experimental Setup.We perform the experiments on a machine running 64-b",
+      "section": "5.1 Evaluation Setup",
+      "page": 12,
+      "char_offset": 41636,
+      "cited_reference": {
+        "number": 38,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We compare EDCagainst 3 state-of-the-art open-source tools for finding logic bugs in DBMSs: TLP [ 37], EET [ 27], and Radar [ 45].",
+      "context_before": "e range of operation invocations and type conversions, enabling it to uncover edge cases that previous approaches fail to consider. Third, its use of equivalence-based transformations can influence query optimization paths, thereby exposing execution-level bugs that are otherwise hidden from conventional testing strategies. Through its data equivalence strategy,EDCis able to generate rich and diverse test cases that penetrate deep into the internal logic of DBMSs, ultimately enabling the discovery of subtle and previously unreported bugs. 5.5 Comparison with Existing Techniques Compared Tools.",
+      "context_after": "TLP [ 37] generates randomized databases and SQL queries to uncover incorrect query results, flagging logic bugs when expected rows are not retrieved. EET [ 27] detects logic bugs by performing semantics-preserving transformations on complex SQL queries to ensure the transformed queries yield the same results as the originals. Additionally, Radar [ 45] identifies logic bugs by comparing query results on databases with identical data but differing metadata constraints, such as NOT NULL. We ran the testing tools on each target DBMS for 24 hours and recorded the number of detected logic bugs as",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 19,
+      "char_offset": 64286,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [ 37] generates randomized databases and SQL queries to uncover incorrect query results, flagging logic bugs when expected rows are not retrieved.",
+      "context_before": "r. Third, its use of equivalence-based transformations can influence query optimization paths, thereby exposing execution-level bugs that are otherwise hidden from conventional testing strategies. Through its data equivalence strategy,EDCis able to generate rich and diverse test cases that penetrate deep into the internal logic of DBMSs, ultimately enabling the discovery of subtle and previously unreported bugs. 5.5 Comparison with Existing Techniques Compared Tools.We compare EDCagainst 3 state-of-the-art open-source tools for finding logic bugs in DBMSs: TLP [ 37], EET [ 27], and Radar [ 45].",
+      "context_after": "EET [ 27] detects logic bugs by performing semantics-preserving transformations on complex SQL queries to ensure the transformed queries yield the same results as the originals. Additionally, Radar [ 45] identifies logic bugs by comparing query results on databases with identical data but differing metadata constraints, such as NOT NULL. We ran the testing tools on each target DBMS for 24 hours and recorded the number of detected logic bugs as the metric. For TLP and Radar, we conducted experiments using the default configurations for each DBMS. For EET, we used the–ignore-crash option to byp",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 19,
+      "char_offset": 64415,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For TLP and Radar, we conducted experiments using the default configurations for each DBMS.",
+      "context_before": "generates randomized databases and SQL queries to uncover incorrect query results, flagging logic bugs when expected rows are not retrieved. EET [ 27] detects logic bugs by performing semantics-preserving transformations on complex SQL queries to ensure the transformed queries yield the same results as the originals. Additionally, Radar [ 45] identifies logic bugs by comparing query results on databases with identical data but differing metadata constraints, such as NOT NULL. We ran the testing tools on each target DBMS for 24 hours and recorded the number of detected logic bugs as the metric.",
+      "context_after": "For EET, we used the–ignore-crash option to bypass crash bugs, allowing EET to focus solely on detecting logic bugs. This setting ensures a fair comparison, as EDCis also designed specifically to target logic bugs rather than crashes. Results.Table 8 displays the number of logic bugs detected by each tool in 24 hours. During the evaluation,EDCfound 8, 6, 6, 2, 5, 6, and 7 bugs in MySQL, MariaDB, Percona, PostgreSQL, TiDB, OceanBase, and ClickHouse, respectively, while TLP detected 3 bugs in MySQL, 2 in Percona, 4 in TiDB, and none in others. Radar identified 1 bug in MySQL and Percona, 5 in Ti",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 19,
+      "char_offset": 65026,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "During the evaluation,EDCfound 8, 6, 6, 2, 5, 6, and 7 bugs in MySQL, MariaDB, Percona, PostgreSQL, TiDB, OceanBase, and ClickHouse, respectively, while TLP detected 3 bugs in MySQL, 2 in Percona, 4 in TiDB, and none in others.",
+      "context_before": "identical data but differing metadata constraints, such as NOT NULL. We ran the testing tools on each target DBMS for 24 hours and recorded the number of detected logic bugs as the metric. For TLP and Radar, we conducted experiments using the default configurations for each DBMS. For EET, we used the–ignore-crash option to bypass crash bugs, allowing EET to focus solely on detecting logic bugs. This setting ensures a fair comparison, as EDCis also designed specifically to target logic bugs rather than crashes. Results.Table 8 displays the number of logic bugs detected by each tool in 24 hours.",
+      "context_after": "Radar identified 1 bug in MySQL and Percona, 5 in TiDB, with no support for ClickHouse, OceanBase, and PostgreSQL, and no bugs detected in MariaDB. EET found 3 bugs in MySQL, 1 in PostgreSQL, no bugs in TiDB, and 2 in ClickHouse, but it does not support MariaDB, Percona, and OceanBase. Across the seven DBMSs,EDCdetected 38 unique bugs, while TLP, Radar, and EET identified 9, 7, and 5 bugs respectively, highlighting the complementary nature of their testing strategies. Furthermore, to better understand the relationship between the bugs, we investigated the bug reports generated by these 4 tool",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 19,
+      "char_offset": 65437,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Across the seven DBMSs,EDCdetected 38 unique bugs, while TLP, Radar, and EET identified 9, 7, and 5 bugs respectively, highlighting the complementary nature of their testing strategies.",
+      "context_before": "Results.Table 8 displays the number of logic bugs detected by each tool in 24 hours. During the evaluation,EDCfound 8, 6, 6, 2, 5, 6, and 7 bugs in MySQL, MariaDB, Percona, PostgreSQL, TiDB, OceanBase, and ClickHouse, respectively, while TLP detected 3 bugs in MySQL, 2 in Percona, 4 in TiDB, and none in others. Radar identified 1 bug in MySQL and Percona, 5 in TiDB, with no support for ClickHouse, OceanBase, and PostgreSQL, and no bugs detected in MariaDB. EET found 3 bugs in MySQL, 1 in PostgreSQL, no bugs in TiDB, and 2 in ClickHouse, but it does not support MariaDB, Percona, and OceanBase.",
+      "context_after": "Furthermore, to better understand the relationship between the bugs, we investigated the bug reports generated by these 4 tools. Figure 10 provides insight into the relationships between the bugs detected by EDCand the other tools. In our evaluation,EDCuniquely identified 40 bugs, demonstrating its capability to uncover a broad range of issues. Among these, 3 bugs overlap with TLP, while no overlap exists with Radar or EET. Analysis.EDCand TLP share 3 overlapping bugs, all related to SQL functions. This is because both tools utilize SQL functions in their testing methodologies. However,EDCdete",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 19,
+      "char_offset": 65952,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Among these, 3 bugs overlap with TLP, while no overlap exists with Radar or EET.",
+      "context_before": "ClickHouse, but it does not support MariaDB, Percona, and OceanBase. Across the seven DBMSs,EDCdetected 38 unique bugs, while TLP, Radar, and EET identified 9, 7, and 5 bugs respectively, highlighting the complementary nature of their testing strategies. Furthermore, to better understand the relationship between the bugs, we investigated the bug reports generated by these 4 tools. Figure 10 provides insight into the relationships between the bugs detected by EDCand the other tools. In our evaluation,EDCuniquely identified 40 bugs, demonstrating its capability to uncover a broad range of issues.",
+      "context_after": "Analysis.EDCand TLP share 3 overlapping bugs, all related to SQL functions. This is because both tools utilize SQL functions in their testing methodologies. However,EDCdetects significantly more bugs related to SQL functions and extends its coverage to other data operations, such as predicate operations, which TLP might miss. The 6 unique bugs detected by TLP involve JOIN operations, which are outside the scope of EDC, as it focuses on single-table queries. The bugs detected 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. 314:20 Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou F",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 19,
+      "char_offset": 66484,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "EDCand TLP share 3 overlapping bugs, all related to SQL functions.",
+      "context_before": ",EDCdetected 38 unique bugs, while TLP, Radar, and EET identified 9, 7, and 5 bugs respectively, highlighting the complementary nature of their testing strategies. Furthermore, to better understand the relationship between the bugs, we investigated the bug reports generated by these 4 tools. Figure 10 provides insight into the relationships between the bugs detected by EDCand the other tools. In our evaluation,EDCuniquely identified 40 bugs, demonstrating its capability to uncover a broad range of issues. Among these, 3 bugs overlap with TLP, while no overlap exists with Radar or EET. Analysis.",
+      "context_after": "This is because both tools utilize SQL functions in their testing methodologies. However,EDCdetects significantly more bugs related to SQL functions and extends its coverage to other data operations, such as predicate operations, which TLP might miss. The 6 unique bugs detected by TLP involve JOIN operations, which are outside the scope of EDC, as it focuses on single-table queries. The bugs detected 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. 314:20 Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang Table 8. Number of detected logic bugs in 24 hours. “-” indi",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 19,
+      "char_offset": 66575,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "However,EDCdetects significantly more bugs related to SQL functions and extends its coverage to other data operations, such as predicate operations, which TLP might miss.",
+      "context_before": "ting strategies. Furthermore, to better understand the relationship between the bugs, we investigated the bug reports generated by these 4 tools. Figure 10 provides insight into the relationships between the bugs detected by EDCand the other tools. In our evaluation,EDCuniquely identified 40 bugs, demonstrating its capability to uncover a broad range of issues. Among these, 3 bugs overlap with TLP, while no overlap exists with Radar or EET. Analysis.EDCand TLP share 3 overlapping bugs, all related to SQL functions. This is because both tools utilize SQL functions in their testing methodologies.",
+      "context_after": "The 6 unique bugs detected by TLP involve JOIN operations, which are outside the scope of EDC, as it focuses on single-table queries. The bugs detected 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. 314:20 Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang Table 8. Number of detected logic bugs in 24 hours. “-” indicates that the tool does not support the DBMS. DBMSEDCTLP Radar EET MySQL 8 3 1 3 MariaDB 6 - 0 Percona 6 2 1 PostgreSQL 2 0 - 1 TiDB 5 4 5 0 OceanBase 6 0 - ClickHouse 7 0 - 2 Total 40 9 7 6 by EDCand EET are entirely orthogonal. EET specializes in fin",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 19,
+      "char_offset": 66722,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The 6 unique bugs detected by TLP involve JOIN operations, which are outside the scope of EDC, as it focuses on single-table queries.",
+      "context_before": "t into the relationships between the bugs detected by EDCand the other tools. In our evaluation,EDCuniquely identified 40 bugs, demonstrating its capability to uncover a broad range of issues. Among these, 3 bugs overlap with TLP, while no overlap exists with Radar or EET. Analysis.EDCand TLP share 3 overlapping bugs, all related to SQL functions. This is because both tools utilize SQL functions in their testing methodologies. However,EDCdetects significantly more bugs related to SQL functions and extends its coverage to other data operations, such as predicate operations, which TLP might miss.",
+      "context_after": "The bugs detected 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. 314:20 Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang Table 8. Number of detected logic bugs in 24 hours. “-” indicates that the tool does not support the DBMS. DBMSEDCTLP Radar EET MySQL 8 3 1 3 MariaDB 6 - 0 Percona 6 2 1 PostgreSQL 2 0 - 1 TiDB 5 4 5 0 OceanBase 6 0 - ClickHouse 7 0 - 2 Total 40 9 7 6 by EDCand EET are entirely orthogonal. EET specializes in finding bugs associated with advanced SQL constructs like WITH clauses and window functions. In contrast,EDCfocuses on exploring the div",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 19,
+      "char_offset": 66893,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[ 21,23,27,36,37,39,43–46]",
+      "technique": "norec",
+      "sentence": "Many works have proposed various methods for detecting logic bugs in DBMSs [ 21,23,27,36,37,39,43–46].",
+      "context_before": "ased techniques,EDCdoes not aim for complete bug coverage. Its detection capabilities are scoped to inconsistencies that arise from violations of data equivalence. To mitigate the risk of false negatives within its target domain,EDCgenerates a wide range of operation-type combinations and expression structures to increase coverage.EDCcan serve as a complementary component within a broader DBMS testing framework. By combining EDCwith approaches that emphasize query semantics or constraint-based inference, users can achieve more comprehensive DBMS testing. 7 Related Work DBMS Logic Bug Detection.",
+      "context_after": "NoREC [ 36] works by generating a query with a predicate and then transforming the query to one that DBMSs cannot optimize. A logic bug is detected if the output changes after this transformation. TLP [ 37] breaks down the original query 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. 314:22 Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang into three separate subqueries by splitting the predicate. The combined results of these subqueries should align with the original query’s result; if not, a logic bug is reported. PQS [ 39] generates queries that are designed",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 21,
+      "char_offset": 73442,
+      "cited_reference": {
+        "number": 36,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [ 36] works by generating a query with a predicate and then transforming the query to one that DBMSs cannot optimize.",
+      "context_before": "consistencies that arise from violations of data equivalence. To mitigate the risk of false negatives within its target domain,EDCgenerates a wide range of operation-type combinations and expression structures to increase coverage.EDCcan serve as a complementary component within a broader DBMS testing framework. By combining EDCwith approaches that emphasize query semantics or constraint-based inference, users can achieve more comprehensive DBMS testing. 7 Related Work DBMS Logic Bug Detection.Many works have proposed various methods for detecting logic bugs in DBMSs [ 21,23,27,36,37,39,43–46].",
+      "context_after": "A logic bug is detected if the output changes after this transformation. TLP [ 37] breaks down the original query 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. 314:22 Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang into three separate subqueries by splitting the predicate. The combined results of these subqueries should align with the original query’s result; if not, a logic bug is reported. PQS [ 39] generates queries that are designed to always return a specific row by utilizing a custom interpreter. If the DBMS under test fails to retrieve this row, PQS i",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 21,
+      "char_offset": 73544,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [ 37] breaks down the original query 3, No.",
+      "context_before": "n structures to increase coverage.EDCcan serve as a complementary component within a broader DBMS testing framework. By combining EDCwith approaches that emphasize query semantics or constraint-based inference, users can achieve more comprehensive DBMS testing. 7 Related Work DBMS Logic Bug Detection.Many works have proposed various methods for detecting logic bugs in DBMSs [ 21,23,27,36,37,39,43–46]. NoREC [ 36] works by generating a query with a predicate and then transforming the query to one that DBMSs cannot optimize. A logic bug is detected if the output changes after this transformation.",
+      "context_after": "6 (SIGMOD), Article 314. Publication date: December 2025. 314:22 Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang into three separate subqueries by splitting the predicate. The combined results of these subqueries should align with the original query’s result; if not, a logic bug is reported. PQS [ 39] generates queries that are designed to always return a specific row by utilizing a custom interpreter. If the DBMS under test fails to retrieve this row, PQS identifies it as a logic bug. Sedar [ 21] summarizes a series of SQL function patterns and generates test cases based on th",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 21,
+      "char_offset": 73741,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [ 39] generates queries that are designed to always return a specific row by utilizing a custom interpreter.",
+      "context_before": "ogic bugs in DBMSs [ 21,23,27,36,37,39,43–46]. NoREC [ 36] works by generating a query with a predicate and then transforming the query to one that DBMSs cannot optimize. A logic bug is detected if the output changes after this transformation. TLP [ 37] breaks down the original query 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. 314:22 Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang into three separate subqueries by splitting the predicate. The combined results of these subqueries should align with the original query’s result; if not, a logic bug is reported.",
+      "context_after": "If the DBMS under test fails to retrieve this row, PQS identifies it as a logic bug. Sedar [ 21] summarizes a series of SQL function patterns and generates test cases based on that. The core idea of DQE [ 44] is to detect logic bugs of DBMSs by comparing whether different SQL queries with the same predicate access the same rows in the database. TQS [ 46] decomposes wide tables into smaller ones and generates custom queries with join operations, using the base table to derive ground truth results. PINOLO [ 23] synthesizes queries that produce supersets or subsets of the result set of a seed qu",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 22,
+      "char_offset": 74098,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "If the DBMS under test fails to retrieve this row, PQS identifies it as a logic bug.",
+      "context_before": "transforming the query to one that DBMSs cannot optimize. A logic bug is detected if the output changes after this transformation. TLP [ 37] breaks down the original query 3, No. 6 (SIGMOD), Article 314. Publication date: December 2025. 314:22 Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang into three separate subqueries by splitting the predicate. The combined results of these subqueries should align with the original query’s result; if not, a logic bug is reported. PQS [ 39] generates queries that are designed to always return a specific row by utilizing a custom interpreter.",
+      "context_after": "Sedar [ 21] summarizes a series of SQL function patterns and generates test cases based on that. The core idea of DQE [ 44] is to detect logic bugs of DBMSs by comparing whether different SQL queries with the same predicate access the same rows in the database. TQS [ 46] decomposes wide tables into smaller ones and generates custom queries with join operations, using the base table to derive ground truth results. PINOLO [ 23] synthesizes queries that produce supersets or subsets of the result set of a seed query and identifies bugs when the DBMS’s output violates the expected approximation re",
+      "section": "5.5 Comparison with Existing Techniques",
+      "page": 22,
+      "char_offset": 74211,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "describes_as_state_of_the_art": [
+        "M8"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:36:46Z",
+    "is_model_written": true,
+    "summary": "EDC detects logic bugs by constructing a second database whose data is equivalent to the first under the query being tested, so that the two must return the same result. Where earlier oracles transform the query and hold the data fixed, EDC transforms the data and holds the query fixed, which reaches bugs in how values are stored, converted and compared. Across MySQL, MariaDB, Percona, PostgreSQL, TiDB, OceanBase and ClickHouse it found 38 unique bugs.",
+    "narrative": "TLP is one of EDC's three baselines, named as a state-of-the-art open-source tool for finding logic bugs alongside EET and Radar. The comparison is reported in detail and is unusually candid about where each wins: EDC found 38 bugs to TLP's 9, they overlap on 3 -- all SQL-function bugs -- and the 6 bugs only TLP found involve JOINs, which are outside EDC's scope because it tests single-table queries. NoREC, TLP and PQS are each also described in the related-work section as the query-level approaches EDC is positioned against.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "definition",
+      "M5": "definition",
+      "M6": "motivation",
+      "M7": "background",
+      "M8": "state_of_the_art",
+      "M9": "definition",
+      "M10": "baseline",
+      "M11": "result_comparison",
+      "M12": "result_comparison",
+      "M13": "result_comparison",
+      "M14": "result_comparison",
+      "M15": "result_comparison",
+      "M16": "result_comparison",
+      "M17": "background",
+      "M18": "definition",
+      "M19": "definition",
+      "M20": "definition",
+      "M21": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "TLP is run as a baseline; nothing states that EDC's own implementation reuses the SQLancer codebase."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "EDC transforms the data rather than the query, which the paper presents as a different oracle from TLP and NoREC rather than a generalisation of either."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M8",
+          "M11",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M8",
+            "sentence": "We compare EDCagainst 3 state-of-the-art open-source tools for finding logic bugs in DBMSs: TLP [ 37], EET [ 27], and Radar [ 45].",
+            "section": "5.5 Comparison with Existing Techniques",
+            "page": 19
+          },
+          {
+            "mention_id": "M11",
+            "sentence": "During the evaluation,EDCfound 8, 6, 6, 2, 5, 6, and 7 bugs in MySQL, MariaDB, Percona, PostgreSQL, TiDB, OceanBase, and ClickHouse, respectively, while TLP detected 3 bugs in MySQL, 2 in Percona, 4 in TiDB, and none in others.",
+            "section": "5.5 Comparison with Existing Techniques",
+            "page": 19
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "Across the seven DBMSs,EDCdetected 38 unique bugs, while TLP, Radar, and EET identified 9, 7, and 5 bugs respectively, highlighting the complementary nature of their testing strategies.",
+            "section": "5.5 Comparison with Existing Techniques",
+            "page": 19
+          }
+        ],
+        "reasoning": "M8 names TLP as one of the three tools EDC is compared against, and M11 and M12 report per-DBMS and total bug counts for both.",
+        "techniques": [
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M8"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M8",
+            "sentence": "We compare EDCagainst 3 state-of-the-art open-source tools for finding logic bugs in DBMSs: TLP [ 37], EET [ 27], and Radar [ 45].",
+            "section": "5.5 Comparison with Existing Techniques",
+            "page": 19
+          }
+        ],
+        "reasoning": "M8 calls TLP one of three state-of-the-art open-source tools for finding logic bugs in DBMSs."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3769828.json b/_data/papers/paper_doi_10_1145_3769828.json
new file mode 100644
index 0000000..a49b370
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3769828.json
@@ -0,0 +1,1667 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:01Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3769828",
+    "title": "SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis",
+    "authors": [
+      "Jinhui Lai",
+      "Chi Zhang",
+      "Bingyan Li",
+      "Chenglin Liang",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Yu Jiang",
+      "Zichen Xu"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Manag. Data",
+    "doi": "10.1145/3769828",
+    "arxiv_id": null,
+    "s2_paper_id": "06d55988c1b284df5962886057f46df47fec17d7",
+    "url": "https://doi.org/10.1145/3769828",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3769828",
+      "retrieved_at": "2026-09-09T01:31:01Z",
+      "chars": 93587,
+      "content_sha256": "sha256:4f70236730d814ed990ffdea21080030124f214403797b772c320b3501be196c"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 24,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3385
+      },
+      {
+        "number": "2",
+        "title": "Preliminaries",
+        "start": 11027
+      },
+      {
+        "number": "2.1",
+        "title": "Join Operations",
+        "start": 11043
+      },
+      {
+        "number": "2.2",
+        "title": "Set Relations Under SELECT Queries",
+        "start": 15451
+      },
+      {
+        "number": "2.3",
+        "title": "Metamorphic Testing",
+        "start": 19191
+      },
+      {
+        "number": "3",
+        "title": "Approach",
+        "start": 20202
+      },
+      {
+        "number": "3.1",
+        "title": "Overview",
+        "start": 21042
+      },
+      {
+        "number": "3.2",
+        "title": "Set Relation Model of Joins",
+        "start": 23699
+      },
+      {
+        "number": "3.3",
+        "title": "Set Relation-driven Join Transformation",
+        "start": 29318
+      },
+      {
+        "number": "3.4",
+        "title": "Results Checking",
+        "start": 35766
+      },
+      {
+        "number": "4",
+        "title": "Implementation",
+        "start": 40190
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 45359
+      },
+      {
+        "number": "5.1",
+        "title": "Testing Setup",
+        "start": 45791
+      },
+      {
+        "number": "5.2",
+        "title": "New Bugs",
+        "start": 47001
+      },
+      {
+        "number": "5.3",
+        "title": "Bugs Detected via Each Set Relation",
+        "start": 52880
+      },
+      {
+        "number": "5.4",
+        "title": "Comparison With the State-of-the-Art",
+        "start": 58862
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 74344
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 78323
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 84779
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Bo Tang Andreas Seltenreich and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https://github. com/anse1/sqlsmith.git",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 133, 13 pages. doi:10.1145/3597503.3639076",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. doi:10.1145/3654991",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M. Tamer Özsu. 2007. QAGen: generating query-aware test databases. InProceedings of the 2007 ACMSIGMOD International Conference on Management of Data(Beijing, China)(SIGMOD ’07). Association for Computing Machinery, New York, NY, USA, 341–352. doi:10.1145/1247480.1247520",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Nicolas Bruno and Surajit Chaudhuri. 2005. Flexible database generators. InProceedings of the 31st International Conference on Very Large Data Bases(Trondheim, Norway)(VLDB ’05). VLDB Endowment, 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Tsong Yueh Chen, Fei-Ching Kuo, Huai Liu, Pak-Lok Poon, Dave Towey, TH Tse, and Zhi Quan Zhou. 2018. Metamorphic testing: A review of challenges and opportunities.ACM Computing Surveys (CSUR)51, 1 (2018), 1–27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "E. F. Codd. 1970. A relational model of data for large shared data banks.Commun. ACM13, 6 (jun 1970), 377–387. doi:10.1145/362384.362685",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "DuckDB. 2025. https://duckdb.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Jim Gray, Prakash Sundaresan, Susanne Englert, Ken Baclawski, and Peter J. Weinberger. 1994. Quickly generating billion-record synthetic databases. InProceedings of the 1994 ACMSIGMOD International Conference on Management of Data(Minneapolis, Minnesota, USA)(SIGMOD ’94). Association for Computing Machinery, New York, NY, USA, 243–252. doi:10.1145/191839.191886",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Zhongxian Gu, Mohamed A. Soliman, and Florian M. Waas. 2012. Testing the accuracy of query optimizers. In Proceedings of the Fifth International Workshop on Testing Database Systems(Scottsdale, Arizona)(DBTest ’12). Association for Computing Machinery, New York, NY, USA, Article 11, 6 pages. doi:10.1145/2304510.2304525",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023. Pinolo: Detecting logical bugs in database management systems with approximate query synthesis. In2023 USENIX Annual Technical Conference (USENIX ATC 23). 345–358. 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:24 Jinhui Lai et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. In32nd USENIX Security Symposium (USENIX Security 23). USENIX Association, Anaheim, CA, 4949–4965. https://www.usenix.org/conference/usenixsecurity23/presentation/jiang-zuming",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting logic bugs in database engines via equivalent expression transformation. InProceedings of the 18th USENIX Conference on Operating Systems Design and Implementation(Santa Clara, CA, USA)(OSDI’24). USENIX Association, USA, Article 44, 15 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. APOLLO: automatic detection and diagnosis of performance regressions in database systems.Proc. VLDB Endow.13, 1 (sep 2019), 57–70. doi:10.14778/3357377.3357382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O Laleye, and Sarfraz Khurshid. 2008. Query-Aware Test Generation Using a Relational Constraint Solver. InProceedings of the 23rd IEEE/ACM International Conference on Automated Software Engineering (ASE ’08). IEEE Computer Society, USA, 238–247. doi:10.1109/ASE.2008.34",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of DBMS with Coverage-based Guidance. In31st USENIX Security Symposium (USENIX Security 22). USENIX Association, Boston, MA, 4309–4326. https://www.usenix. org/conference/usenixsecurity22/presentation/liang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. InProceedings of the 44th International Conference on Software Engineering(Pittsburgh, Pennsylvania)(ICSE ’22). Association for Computing Machinery, New York, NY, USA, 225–236. doi:10.1145/3510003. 3510093",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "MariaDB. 2025. https://mariadb.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "William M McKeeman. 1998. Differential testing for software.Digital Technical Journal10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Priti Mishra and Margaret H. Eich. 1992. Join processing in relational databases.ACM Comput. Surv.24, 1 (March 1992), 63–113. doi:10.1145/128762.128764",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "MySQL. 2025. https://www.mysql.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "PostgreSQL. 2025. https://www.postgresql.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/33680",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 24,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (nov 2020), 30 pages. doi:10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, 667–682. https://www. usenix.org/conference/osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 26,
+      "text": "Donald R Slutz. 1998. Massive stochastic testing of SQL. InVLDB, Vol. 98. Citeseer, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. In Engineering (ICSE). 2072–2084. doi:10.1109/ICSE48619.2023.00175",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting logic bugs of join optimizations in dbms.Proceedings of the ACM on Management of Data1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "TiDB. 2025. https://www.pingcap.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation(San Jose, California, USA)(PLDI ’11). Association for Computing Machinery, New York, NY, USA, 283–294. doi:10.1145/1993498. 1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Michal Zalewski. 2015. AFL: American Fuzzy Lop. https://github.com/google/AFL",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages. doi:10.1145/3709674",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 33,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security(Virtual Event, USA)(CCS ’20). Association for Computing Machinery, New York, NY, USA, 955–970. Received April 2025; revised July 2025; acc",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 133, 13 pages. doi:10.1145/3597503.3639076",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert",
+        "prints the DOI of the paper introducing cert"
+      ]
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. doi:10.1145/3654991",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "prints the DOI of the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 23,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 24,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (nov 2020), 30 pages. doi:10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, 667–682. https://www. usenix.org/conference/osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 32,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages. doi:10.1145/3709674",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "prints the DOI of the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "TQS [ 28] and DQP [ 3] are two state-of-the-art testing approaches that target logic bugs in the optimization of join operations.",
+      "context_before": ". Based on the standard string comparison method, if one string is a prefix of the other, the shorter one is considered smaller than the longer one. Therefore, -9876543210 is considered smaller than-9876543210.0. We expect this full outer join to produce one row, as the condition t0.c0 < t1.c1 should evaluate to true. However, DuckDB returns an incorrect result for this query, where t0.c0 < t1.c1 is incorrectly evaluated as false on the boundary values. Then two rows of incorrect results from the respective tables are retained independently in the resulting joined output, as shown in Figure 1.",
+      "context_after": "Both approaches use the equivalence among alternative query execution plans as a test oracle to indicate the presence of logic bugs. By leveraging query hints and system variables, they manipulate the behavior of the query optimizer to ensure that the same query is executed using different plans. Any inconsistency in the query results may indicate the presence of logic bugs in join optimization. However, this strategy becomes less effective for detecting logic bugs in join operations when query hints or system variables fail to influence the optimizer’s behavior, or when the logic bugs reside",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5992,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "To detect the logic bug shown in Figure 1, we applied all query plan transformation techniques from both TQS and DQP to the bug-inducing query, i.",
+      "context_before": "ecution plans as a test oracle to indicate the presence of logic bugs. By leveraging query hints and system variables, they manipulate the behavior of the query optimizer to ensure that the same query is executed using different plans. Any inconsistency in the query results may indicate the presence of logic bugs in join optimization. However, this strategy becomes less effective for detecting logic bugs in join operations when query hints or system variables fail to influence the optimizer’s behavior, or when the logic bugs reside in parts of the join implementation unrelated to optimization.",
+      "context_after": "e., 𝑞1; however, no inconsistencies were observed. Therefore, neither approach was able to identify this bug. To tackle logic bugs in the implementation of join in DBMSs, we propose a testing approach namedSetRelationSynthesis (SRS). The key insight is that different join operations involve various join types, orders, and constraints, which are not mutually independent but instead follow specific set relations. SRS thoroughly tests join operations by transforming join operations within queries— enabling the execution of various code paths related to join processing in the DBMS—and then detects",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6785,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 23–25]",
+      "technique": "norec",
+      "sentence": "In step 1, we randomly generate the database state with syntax-rule-based generators, which is a common practice in existing works [ 23–25].",
+      "context_before": "involve various join types, orders, and constraints, which are not mutually independent but instead follow specific set relations. Specifically, SRS first generates an original query containing randomly constructed join operations, then derives one or two follow-up queries based on our set relation model. The results of these follow-up queries are expected to satisfy predefined set relations with the result of the original query. Any violation of these expected set relations signals a potential logic bug in the DBMS’s join implementation. 3.1 Overview Figure 2 illustrates the overview of SRS.",
+      "context_after": "In Step 2, the original query is constructed by randomly selecting the join type, the participating tables, and the join condition to populate join skeletons. In step 3, we transform the join operations in the original query based on a manually extracted set relation model of different join operations. This transformation results in one or two transformed queries, depending on the synthesis rules of the set relation model. In step 4, we check whether the results of the original query and the transformed query reflect the set relation defined by the set relation model of the corresponding join",
+      "section": "3.1 Overview",
+      "page": 6,
+      "char_offset": 21096,
+      "cited_reference": {
+        "number": 23,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Additionally, the influence of NULLvalues must be carefully considered, as addressed in TLP[24].",
+      "context_before": "set—for example, switching a condition from JOIN ON t0.c0 IS NULL to JOIN ON t0.c0 IS NOT NULL, which reverses the comparison logic and produces a complementary result. (4) Reducing the join set—for example, changing a condition from >=or<=to=, thereby narrowing the match criteria. These transformations allow flexible control over query behavior while preserving logical consistency. Notably, modifying join conditions is primarily suitable for inner joins, as other types of joins—such as outer joins— may produce overlapping result sets, making the effects of such modifications harder to predict.",
+      "context_after": "Algorithmic sketch of applying transformation.Algorithm 1 shows the procedure for applying transformations to an original query in order to generate the corresponding follow-up transformed queries. For each generated original query, we first determine the join type 𝑗𝑜𝑖𝑛_𝑡𝑦𝑝𝑒 used in the query using the GetJoinType function. To account for the fact that not all language features support every type of set relation, we extract the language features 𝑓𝑒𝑎𝑡𝑢𝑟𝑒𝑠 present in the original query using the GetFeaturesInQuery function. A detailed explanation is provided in Section 3.4. Next, we retrieve th",
+      "section": "3.3 Set Relation-driven Join Transformation",
+      "page": 9,
+      "char_offset": 33506,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[ 23,24]",
+      "technique": "norec",
+      "sentence": "4 Implementation We implemented SRS on top of SQLancer1, a DBMS testing framework designed for the random generation of database states and SQL queries, which also supports multiple test oracles [ 23,24].",
+      "context_before": "ER BY clauses, their generation poses no limitation, as these clauses are expected to preserve the same set relations. For the GROUP BY clause, only equivalent relations can be supported. For the LIMIT clause, only disjoint relations can be supported. For the SELECT clause, there is no limitation if it simply returns the original data without modification. However, for more complex cases, only equivalent and union relations can be supported. Finally, we identify potential logic bugs by detecting violations of the expected set relations among the results of the original and transformed queries.",
+      "context_after": "In total, SRS comprises 5,500 lines of code. This code is primarily used to generate more complex join operations, upon which transformations are applied based on the synthesis rules. Despite this, the approach remains simple and can be easily adapted to other testing frameworks, such as SQLsmith [1]. In this section, we present the technical details of SRS’s implementation. Database state generation.Following the standard pipeline in popular DBMS testing tools like SQLancer, we generate the database state first to improve semantic correctness by generating stateconforming queries. SRS applie",
+      "section": "3.4 Results Checking",
+      "page": 11,
+      "char_offset": 40189,
+      "cited_reference": {
+        "number": 23,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Following the standard pipeline in popular DBMS testing tools like SQLancer, we generate the database state first to improve semantic correctness by generating stateconforming queries.",
+      "context_before": "ntation We implemented SRS on top of SQLancer1, a DBMS testing framework designed for the random generation of database states and SQL queries, which also supports multiple test oracles [ 23,24]. In total, SRS comprises 5,500 lines of code. This code is primarily used to generate more complex join operations, upon which transformations are applied based on the synthesis rules. Despite this, the approach remains simple and can be easily adapted to other testing frameworks, such as SQLsmith [1]. In this section, we present the technical details of SRS’s implementation. Database state generation.",
+      "context_after": "SRS applies the automated, syntax-rule-based random generation approach of SQLancer to ensure that the database state exhibits sufficient diversity, thereby enabling thorough and effective bug detection. SRS also incorporates several strategies, including the generation of diverse schemas and data characteristics, such as boundary values. The details of these strategies are further discussed in Section 6. 1https://github.com/sqlancer/sqlancer 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:12 Jinhui Lai et al. Table 2. SQL grammar of join skeleton. Explicit Join explicit",
+      "section": "4 Implementation",
+      "page": 11,
+      "char_offset": 40799,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SRS applies the automated, syntax-rule-based random generation approach of SQLancer to ensure that the database state exhibits sufficient diversity, thereby enabling thorough and effective bug detection.",
+      "context_before": "s [ 23,24]. In total, SRS comprises 5,500 lines of code. This code is primarily used to generate more complex join operations, upon which transformations are applied based on the synthesis rules. Despite this, the approach remains simple and can be easily adapted to other testing frameworks, such as SQLsmith [1]. In this section, we present the technical details of SRS’s implementation. Database state generation.Following the standard pipeline in popular DBMS testing tools like SQLancer, we generate the database state first to improve semantic correctness by generating stateconforming queries.",
+      "context_after": "SRS also incorporates several strategies, including the generation of diverse schemas and data characteristics, such as boundary values. The details of these strategies are further discussed in Section 6. 1https://github.com/sqlancer/sqlancer 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:12 Jinhui Lai et al. Table 2. SQL grammar of join skeleton. Explicit Join explicit-join::=SELECT <column-list> FROM <table> <join-clause> column-list::=[<column> [, <column> ...]]* join-clause::=[<join-type> <table> ON <join-condition>]* join-condition::=<term> [(AND|OR) <term> ]* term",
+      "section": "4 Implementation",
+      "page": 11,
+      "char_offset": 40983,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "sqlancer",
+      "technique": null,
+      "sentence": "com/sqlancer/sqlancer 3, No.",
+      "context_before": "the standard pipeline in popular DBMS testing tools like SQLancer, we generate the database state first to improve semantic correctness by generating stateconforming queries. SRS applies the automated, syntax-rule-based random generation approach of SQLancer to ensure that the database state exhibits sufficient diversity, thereby enabling thorough and effective bug detection. SRS also incorporates several strategies, including the generation of diverse schemas and data characteristics, such as boundary values. The details of these strategies are further discussed in Section 6. 1https://github.",
+      "context_after": "6 (SIGMOD), Article 363. Publication date: December 2025. 363:12 Jinhui Lai et al. Table 2. SQL grammar of join skeleton. Explicit Join explicit-join::=SELECT <column-list> FROM <table> <join-clause> column-list::=[<column> [, <column> ...]]* join-clause::=[<join-type> <table> ON <join-condition>]* join-condition::=<term> [(AND|OR) <term> ]* term::=<column> <comparison-operator> (<column>|<value>) Implicit Join implicit-join::=SELECT <column-list> FROM <table-list> <join-clause> column-list::=[<column> [, <column> ...]]* table-list::=[<table> [, <table> ...]]* join-clause::=WHERE <join-condit",
+      "section": "4 Implementation",
+      "page": 11,
+      "char_offset": 41409,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For the generation of other language features, aside from considering the tables involved in joins, our approach remains consistent with SQLancer’s query generation strategy for NoREC test oracle [23].",
+      "context_before": "itself via LEFT JOIN. This also enables SRS to cover as many join types as possible, thereby testing all the set relations proposed in this paper. Third, additional query clauses are randomly generated based on the table information in join operations. Although not all language features can preserve or reflect the input set relations in their output, we continue to use a fully random, syntax-rule-based generation approach for the rest of the SQL query. This approach decouples query generation from the set relation model, making it easier to extend the model with additional rules in the future.",
+      "context_after": "2https://dev.mysql.com/doc/refman/8.0/en/semijoins.html 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis 363:13 Table 3. Tested DBMSs. DBMS Version Ranking Github Stars Categories MySQL 9.2.0 2 11.3k Traditional MariaDB 11.7.2 14 6k Traditional TiDB 8.5.1 73 38.3k Distributed PostgreSQL 17.5.1 4 18.1k Traditional DuckDB 1.2.0 44 28.4k Embedded 5 Evaluation We evaluate the effectiveness and efficiency of SRS by answering the following research questions: •Q1. New Bugs: How many unique and new",
+      "section": "4 Implementation",
+      "page": 12,
+      "char_offset": 44708,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker",
+      "surface": "[ 3,23–25]",
+      "technique": "dqp",
+      "sentence": "These DBMSs are also commonly used in the evaluation of previous works [ 3,23–25].",
+      "context_before": "ring the following research questions: •Q1. New Bugs: How many unique and new bugs in real-world DBMSs can SRS detect? •Q2. Bugs Detected via Each Set Relation: How effective is the set relation model in detecting logic bugs? •Q3. Comparison with the State-of-the-Art: Can SRS outperform existing state-of-the-art approaches for detecting logic bug in DBMSs? 5.1 Testing Setup Tested DBMSs.We consider five DBMSs in our evaluation, including MySQL, MariaDB, TiDB, PostgreSQL, and DuckDB. According to the DB-Engine’s Ranking3, and GitHub, these DBMSs are among the most popular and widely-used DBMSs.",
+      "context_after": "Therefore, any bugs identified by SRS may have been overlooked by existing approaches. We tested the latest stable release versions of each DBMS— MySQL (9.2.0), MariaDB (11.7.2), TiDB (8.5.1), PostgreSQL (17.5.1), and DuckDB (1.2.0). Since release versions are generally considered stable, any bugs identified in these DBMSs serve as strong evidence of the effectiveness of our approach. These DBMSs represent different categories. MySQL, MariaDB, and PostgreSQL are traditional, relational, and client-server DBMSs, which are widely used in industry. TiDB is a distributed relational DBMS designed",
+      "section": "5.1 Testing Setup",
+      "page": 13,
+      "char_offset": 46032,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. doi:10.1145/3654991",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "prints the DOI of the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[ 3,24]",
+      "technique": "dqp",
+      "sentence": "This aligns with observations from previous works [ 3,24], which limited our reporting of additional potential bugs to avoid duplication.",
+      "context_before": "g query—including its involved language features and the results of the join operation—also indicated that the root cause was independent of the join. This suggests that SRS can also detect logic bugs in the implementation of other language features. Bug status.All bugs have been confirmed by developers, with 12 already fixed and 33 identified as logic bugs. In our bug reports for MariaDB, PostgreSQL, and DuckDB, developers were highly responsive in addressing the issues, indicating that these bugs may pose significant risks. In contrast, MySQL and TiDB required more time to fix reported bugs.",
+      "context_after": "Bug importance.Most of the bugs found by SRS are critical and have a high likelihood of impacting real users. In MySQL, MariaDB, and TiDB, 27 bugs are classified as Serious, Major, or Moderate, highlighting their significant impact on the systems and underscoring the necessity for high-priority fixes. 5.3 Bugs Detected via Each Set Relation All 33 logic bugs were uncovered by SRS through the four types of set relations derived from our synthesis rules. Specifically, 14 bugs were uncovered via equivalent relations, 2 via intersecting relations, 4 via disjoint relations, and 13 via union relati",
+      "section": "5.2 New Bugs",
+      "page": 15,
+      "char_offset": 52438,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. doi:10.1145/3654991",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "prints the DOI of the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "We selected six state-of-the-art approaches for comparison: PQS [ 25], TLP [ 24], NoREC [ 23], Pinolo [ 11], EET [ 13], and DQP [ 3].",
+      "context_before": "er 24 hours. However, due to the implementation quality, a tool may not fully reflect the true bug-detection capability of its underlying approach. Therefore, we further compared the test oracles proposed by different approaches from a theoretical perspective. This evaluation focuses on the detection of logic bugs, and the results include only the logic bugs found, as SRS and the baseline methods are specifically designed to uncover logic bugs in join operations. Other types of bugs (e.g., runtime errors or assertion failures) can typically be detected by most existing testing tools. Baseline.",
+      "context_after": "Both DQP and TQS [ 28] are approaches designed to detect logic bugs in join operations by leveraging the equivalence among different query plans for a given query. These two approaches are the most closely aligned with the goal of SRS. We did not include TQS in our comparison because its source code is unavailable, and the authors informed us via email that they are currently unable to share it. DQP is a simpler alternative to TQS and has specifically studied TQS, concluding that it offers the same level of bug-finding effectiveness [ 13] as TQS. Therefore, we assume that comparing with DQP e",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 17,
+      "char_offset": 59768,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp",
+        "norec",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "Both DQP and TQS [ 28] are approaches designed to detect logic bugs in join operations by leveraging the equivalence among different query plans for a given query.",
+      "context_before": "ying approach. Therefore, we further compared the test oracles proposed by different approaches from a theoretical perspective. This evaluation focuses on the detection of logic bugs, and the results include only the logic bugs found, as SRS and the baseline methods are specifically designed to uncover logic bugs in join operations. Other types of bugs (e.g., runtime errors or assertion failures) can typically be detected by most existing testing tools. Baseline.We selected six state-of-the-art approaches for comparison: PQS [ 25], TLP [ 24], NoREC [ 23], Pinolo [ 11], EET [ 13], and DQP [ 3].",
+      "context_after": "These two approaches are the most closely aligned with the goal of SRS. We did not include TQS in our comparison because its source code is unavailable, and the authors informed us via email that they are currently unable to share it. DQP is a simpler alternative to TQS and has specifically studied TQS, concluding that it offers the same level of bug-finding effectiveness [ 13] as TQS. Therefore, we assume that comparing with DQP effectively represents a comparison with this class of approaches. PQS generates queries based on the database state such that the result is guaranteed to include sp",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 17,
+      "char_offset": 59901,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "DQP is a simpler alternative to TQS and has specifically studied TQS, concluding that it offers the same level of bug-finding effectiveness [ 13] as TQS.",
+      "context_before": "can typically be detected by most existing testing tools. Baseline.We selected six state-of-the-art approaches for comparison: PQS [ 25], TLP [ 24], NoREC [ 23], Pinolo [ 11], EET [ 13], and DQP [ 3]. Both DQP and TQS [ 28] are approaches designed to detect logic bugs in join operations by leveraging the equivalence among different query plans for a given query. These two approaches are the most closely aligned with the goal of SRS. We did not include TQS in our comparison because its source code is unavailable, and the authors informed us via email that they are currently unable to share it.",
+      "context_after": "Therefore, we assume that comparing with DQP effectively represents a comparison with this class of approaches. PQS generates queries based on the database state such that the result is guaranteed to include specific rows from the database. It supports testing of join operations. EET leverages tautologies and contradictions to construct equivalent queries, enabling the testing of any syntactic feature involving expressions. As a result, it also supports the testing of join operations. Both TLP and Pionlo leverage set-based relationships to detect logic bugs, but their transformation targets a",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 17,
+      "char_offset": 60300,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "Therefore, we assume that comparing with DQP effectively represents a comparison with this class of approaches.",
+      "context_before": "C [ 23], Pinolo [ 11], EET [ 13], and DQP [ 3]. Both DQP and TQS [ 28] are approaches designed to detect logic bugs in join operations by leveraging the equivalence among different query plans for a given query. These two approaches are the most closely aligned with the goal of SRS. We did not include TQS in our comparison because its source code is unavailable, and the authors informed us via email that they are currently unable to share it. DQP is a simpler alternative to TQS and has specifically studied TQS, concluding that it offers the same level of bug-finding effectiveness [ 13] as TQS.",
+      "context_after": "PQS generates queries based on the database state such that the result is guaranteed to include specific rows from the database. It supports testing of join operations. EET leverages tautologies and contradictions to construct equivalent queries, enabling the testing of any syntactic feature involving expressions. As a result, it also supports the testing of join operations. Both TLP and Pionlo leverage set-based relationships to detect logic bugs, but their transformation targets are expressions within query clauses. Both tools provide support for generating join operations. NoREC generates",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 17,
+      "char_offset": 60454,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS generates queries based on the database state such that the result is guaranteed to include specific rows from the database.",
+      "context_before": "ugs in join operations by leveraging the equivalence among different query plans for a given query. These two approaches are the most closely aligned with the goal of SRS. We did not include TQS in our comparison because its source code is unavailable, and the authors informed us via email that they are currently unable to share it. DQP is a simpler alternative to TQS and has specifically studied TQS, concluding that it offers the same level of bug-finding effectiveness [ 13] as TQS. Therefore, we assume that comparing with DQP effectively represents a comparison with this class of approaches.",
+      "context_after": "It supports testing of join operations. EET leverages tautologies and contradictions to construct equivalent queries, enabling the testing of any syntactic feature involving expressions. As a result, it also supports the testing of join operations. Both TLP and Pionlo leverage set-based relationships to detect logic bugs, but their transformation targets are expressions within query clauses. Both tools provide support for generating join operations. NoREC generates equivalent queries by exploiting the equivalence between the WHERE and SELECT clauses. While NoREC is not designed to specificall",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 17,
+      "char_offset": 60566,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Both TLP and Pionlo leverage set-based relationships to detect logic bugs, but their transformation targets are expressions within query clauses.",
+      "context_before": "specifically studied TQS, concluding that it offers the same level of bug-finding effectiveness [ 13] as TQS. Therefore, we assume that comparing with DQP effectively represents a comparison with this class of approaches. PQS generates queries based on the database state such that the result is guaranteed to include specific rows from the database. It supports testing of join operations. EET leverages tautologies and contradictions to construct equivalent queries, enabling the testing of any syntactic feature involving expressions. As a result, it also supports the testing of join operations.",
+      "context_after": "Both tools provide support for generating join operations. NoREC generates equivalent queries by exploiting the equivalence between the WHERE and SELECT clauses. While NoREC is not designed to specifically test join operations, it has previously detected join-related bugs. As such, we include a comparison with NoREC in our evaluation. 5.4.1 Comparison to Other DBMS Logic Bug Testing Tools.We run each tool on the latest version of each supported DBMS for a duration of 24 hours. Since the baseline approaches do not support all the DBMSs that SRS does, our comparison is limited to the DBMSs comm",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 17,
+      "char_offset": 60944,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC generates equivalent queries by exploiting the equivalence between the WHERE and SELECT clauses.",
+      "context_before": "ss of approaches. PQS generates queries based on the database state such that the result is guaranteed to include specific rows from the database. It supports testing of join operations. EET leverages tautologies and contradictions to construct equivalent queries, enabling the testing of any syntactic feature involving expressions. As a result, it also supports the testing of join operations. Both TLP and Pionlo leverage set-based relationships to detect logic bugs, but their transformation targets are expressions within query clauses. Both tools provide support for generating join operations.",
+      "context_after": "While NoREC is not designed to specifically test join operations, it has previously detected join-related bugs. As such, we include a comparison with NoREC in our evaluation. 5.4.1 Comparison to Other DBMS Logic Bug Testing Tools.We run each tool on the latest version of each supported DBMS for a duration of 24 hours. Since the baseline approaches do not support all the DBMSs that SRS does, our comparison is limited to the DBMSs commonly supported by both SRS and each baseline. To ensure fair comparisons, we allocated four threads for each DBMS test in our experiments, following common practi",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 17,
+      "char_offset": 61149,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "While NoREC is not designed to specifically test join operations, it has previously detected join-related bugs.",
+      "context_before": "to include specific rows from the database. It supports testing of join operations. EET leverages tautologies and contradictions to construct equivalent queries, enabling the testing of any syntactic feature involving expressions. As a result, it also supports the testing of join operations. Both TLP and Pionlo leverage set-based relationships to detect logic bugs, but their transformation targets are expressions within query clauses. Both tools provide support for generating join operations. NoREC generates equivalent queries by exploiting the equivalence between the WHERE and SELECT clauses.",
+      "context_after": "As such, we include a comparison with NoREC in our evaluation. 5.4.1 Comparison to Other DBMS Logic Bug Testing Tools.We run each tool on the latest version of each supported DBMS for a duration of 24 hours. Since the baseline approaches do not support all the DBMSs that SRS does, our comparison is limited to the DBMSs commonly supported by both SRS and each baseline. To ensure fair comparisons, we allocated four threads for each DBMS test in our experiments, following common practice in previous work [ 11]. In this experiment, we aim to investigate the effectiveness and efficiency of SRS in",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 17,
+      "char_offset": 61252,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "As such, we include a comparison with NoREC in our evaluation.",
+      "context_before": "d contradictions to construct equivalent queries, enabling the testing of any syntactic feature involving expressions. As a result, it also supports the testing of join operations. Both TLP and Pionlo leverage set-based relationships to detect logic bugs, but their transformation targets are expressions within query clauses. Both tools provide support for generating join operations. NoREC generates equivalent queries by exploiting the equivalence between the WHERE and SELECT clauses. While NoREC is not designed to specifically test join operations, it has previously detected join-related bugs.",
+      "context_after": "5.4.1 Comparison to Other DBMS Logic Bug Testing Tools.We run each tool on the latest version of each supported DBMS for a duration of 24 hours. Since the baseline approaches do not support all the DBMSs that SRS does, our comparison is limited to the DBMSs commonly supported by both SRS and each baseline. To ensure fair comparisons, we allocated four threads for each DBMS test in our experiments, following common practice in previous work [ 11]. In this experiment, we aim to investigate the effectiveness and efficiency of SRS in comparison to state-of-the-art approaches. Results of effective",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 17,
+      "char_offset": 61364,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Table 5 shows that SRS found 12, 9, 10, 20, 17, and 13 more bugs than PQS, NoREC, TLP, DQP, Pinolo, and EET, respectively.",
+      "context_before": "xperiments, following common practice in previous work [ 11]. In this experiment, we aim to investigate the effectiveness and efficiency of SRS in comparison to state-of-the-art approaches. Results of effectiveness.Table 5 presents the number of unique logic bugs identified by each approach on each DBMS in the duration of 24 hours. Since not all approaches target join-related logic bugs, we did not exclude non-join-related data in Table 5 to facilitate a fair comparison of the bug detection capabilities of different approaches. For example, the 3 bugs detected by Pinolo are unrelated to joins.",
+      "context_after": "Note that the increments are calculated based only on the DBMSs commonly supported by both approaches. The other approaches were less effective in our experiments because we tested the latest versions of the DBMSs. Therefore, their 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:18 Jinhui Lai et al. Table 5. Number of unique logic bugs triggered in 24 hours. DBMSPQS NoREC TLPDQP Pinolo EETSRS MySQL 0 - 4 0 2 310 MariaDB-0 - 0 1 -6 TiDB - - 2 0 0 04 PostgreSQL 0 0 0 - - 02 DuckDB-01- - -1 Total 0 0 7 0 3 323 Increment 12↑9↑10↑20↑17↑13↑“-” indicates that the tool does not",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 17,
+      "char_offset": 62417,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "DBMSPQS NoREC TLPDQP Pinolo EETSRS MySQL 0 - 4 0 2 310 MariaDB-0 - 0 1 -6 TiDB - - 2 0 0 04 PostgreSQL 0 0 0 - - 02 DuckDB-01- - -1 Total 0 0 7 0 3 323 Increment 12↑9↑10↑20↑17↑13↑“-” indicates that the tool does not support the DBMS.",
+      "context_before": "es of different approaches. For example, the 3 bugs detected by Pinolo are unrelated to joins. Table 5 shows that SRS found 12, 9, 10, 20, 17, and 13 more bugs than PQS, NoREC, TLP, DQP, Pinolo, and EET, respectively. Note that the increments are calculated based only on the DBMSs commonly supported by both approaches. The other approaches were less effective in our experiments because we tested the latest versions of the DBMSs. Therefore, their 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:18 Jinhui Lai et al. Table 5. Number of unique logic bugs triggered in 24 hours.",
+      "context_after": "“Increments” are calculated only for commonly supported DBMSs. /uni00000013/uni00000017/uni0000001b/uni00000014/uni00000015/uni00000014/uni00000019/uni00000015/uni00000013/uni00000015/uni00000017 /uni0000004b/uni00000052/uni00000058/uni00000055/uni00000056/uni00000013/uni00000015/uni00000017/uni00000019/uni0000001b/uni00000014/uni00000013/uni00000045/uni00000058/uni0000004a/uni00000003/uni00000046/uni00000052/uni00000058/uni00000051/uni00000057 /uni0000000b/uni00000044/uni0000000c/uni00000003/uni00000030/uni0000005c/uni00000036/uni00000034/uni0000002f/uni00000036/uni00000035/uni00000036 /uni0",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 18,
+      "char_offset": 62923,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We compared the code coverage achieved by these approaches, and the results showed that the test oracles implemented on SQLancer— PQS, NoREC, TLP, DQP, and SRS—achieved similar code coverage.",
+      "context_before": "fficiency.Figure 5 illustrates the logic bug detection progress over time for SRS and the baselines. Results from baseline approaches are excluded if they pertain to unsupported DBMSs or if the approaches failed to detect any bugs. At each time point, SRS consistently detects an equal or greater number of unique logic bugs compared to all baselines, highlighting its effectiveness in identifying bugs both faster and more comprehensively. Code coverage is a commonly used metric in testing; however, it is mainly influenced by the test case generation strategy, which is not the focus of this work.",
+      "context_after": "For instance, when testing MySQL for 24 hours, PQS, TLP, DQP, and SRS achieved line coverage rates of 23.35%, 23.78%, 22.37%, and 24.19%, respectively. This similarity arises because these methods adopt comparable test case generation strategies, and the generated queries exercise similar language features. 5.4.2 Comparison to Other DBMS Logic Bug Testing Oracles.We conducted a best-effort manual analysis on the bug-inducing test cases to determine whether the logic bugs identified by SRS, as shown in Table 4, could theoretically be detected by other approaches’ oracle, and whether the join-r",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 18,
+      "char_offset": 66742,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M24",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "For instance, when testing MySQL for 24 hours, PQS, TLP, DQP, and SRS achieved line coverage rates of 23.",
+      "context_before": "e approaches failed to detect any bugs. At each time point, SRS consistently detects an equal or greater number of unique logic bugs compared to all baselines, highlighting its effectiveness in identifying bugs both faster and more comprehensively. Code coverage is a commonly used metric in testing; however, it is mainly influenced by the test case generation strategy, which is not the focus of this work. We compared the code coverage achieved by these approaches, and the results showed that the test oracles implemented on SQLancer— PQS, NoREC, TLP, DQP, and SRS—achieved similar code coverage.",
+      "context_after": "35%, 23.78%, 22.37%, and 24.19%, respectively. This similarity arises because these methods adopt comparable test case generation strategies, and the generated queries exercise similar language features. 5.4.2 Comparison to Other DBMS Logic Bug Testing Oracles.We conducted a best-effort manual analysis on the bug-inducing test cases to determine whether the logic bugs identified by SRS, as shown in Table 4, could theoretically be detected by other approaches’ oracle, and whether the join-related logic bugs detected by other approaches could be identified by SRS’s oracle. A test oracle is valua",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 18,
+      "char_offset": 66934,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "citation_marker",
+      "surface": "[ 23]",
+      "technique": "norec",
+      "sentence": "And this is a widely used method for comparing different test oracles [ 23].",
+      "context_before": "ecause these methods adopt comparable test case generation strategies, and the generated queries exercise similar language features. 5.4.2 Comparison to Other DBMS Logic Bug Testing Oracles.We conducted a best-effort manual analysis on the bug-inducing test cases to determine whether the logic bugs identified by SRS, as shown in Table 4, could theoretically be detected by other approaches’ oracle, and whether the join-related logic bugs detected by other approaches could be identified by SRS’s oracle. A test oracle is valuable as long as it can detect logic bugs that other test oracles cannot.",
+      "context_after": "In this experiment, to determine whether the bugs detected by SRS could be identified by other approaches’ oracle, we selected TQS/DQP, TLP, and NoREC for comparison because their fixed transformation patterns allow for reliable manual validation. We excluded EET, Pinolo, and PQS due to their vast transformation spaces, 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis 363:19 15 2 11 1 TQS/DQP ’s oracle TLP ’s oracle NoREC ’s oracle Logic bugs found by SRS 33 Fig. 6. Results of applying othe",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 18,
+      "char_offset": 67711,
+      "cited_reference": {
+        "number": 23,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "In this experiment, to determine whether the bugs detected by SRS could be identified by other approaches’ oracle, we selected TQS/DQP, TLP, and NoREC for comparison because their fixed transformation patterns allow for reliable manual validation.",
+      "context_before": "e generated queries exercise similar language features. 5.4.2 Comparison to Other DBMS Logic Bug Testing Oracles.We conducted a best-effort manual analysis on the bug-inducing test cases to determine whether the logic bugs identified by SRS, as shown in Table 4, could theoretically be detected by other approaches’ oracle, and whether the join-related logic bugs detected by other approaches could be identified by SRS’s oracle. A test oracle is valuable as long as it can detect logic bugs that other test oracles cannot. And this is a widely used method for comparing different test oracles [ 23].",
+      "context_after": "We excluded EET, Pinolo, and PQS due to their vast transformation spaces, 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis 363:19 15 2 11 1 TQS/DQP ’s oracle TLP ’s oracle NoREC ’s oracle Logic bugs found by SRS 33 Fig. 6. Results of applying other oracles on logic bugs detected by SRS. Of the total 33 logic bugs detected by SRS, 13 were undetectable by other oracles. which make manual validation infeasible. Specifically, EET constructs equivalent queries by adding expressions that always e",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 18,
+      "char_offset": 67788,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "We excluded EET, Pinolo, and PQS due to their vast transformation spaces, 3, No.",
+      "context_before": "wn in Table 4, could theoretically be detected by other approaches’ oracle, and whether the join-related logic bugs detected by other approaches could be identified by SRS’s oracle. A test oracle is valuable as long as it can detect logic bugs that other test oracles cannot. And this is a widely used method for comparing different test oracles [ 23]. In this experiment, to determine whether the bugs detected by SRS could be identified by other approaches’ oracle, we selected TQS/DQP, TLP, and NoREC for comparison because their fixed transformation patterns allow for reliable manual validation.",
+      "context_after": "6 (SIGMOD), Article 363. Publication date: December 2025. SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis 363:19 15 2 11 1 TQS/DQP ’s oracle TLP ’s oracle NoREC ’s oracle Logic bugs found by SRS 33 Fig. 6. Results of applying other oracles on logic bugs detected by SRS. Of the total 33 logic bugs detected by SRS, 13 were undetectable by other oracles. which make manual validation infeasible. Specifically, EET constructs equivalent queries by adding expressions that always evaluate to TRUE or FALSE. However, the search space for such expressions is extreme",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 18,
+      "char_offset": 68036,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis 363:19 15 2 11 1 TQS/DQP ’s oracle TLP ’s oracle NoREC ’s oracle Logic bugs found by SRS 33 Fig.",
+      "context_before": "ches could be identified by SRS’s oracle. A test oracle is valuable as long as it can detect logic bugs that other test oracles cannot. And this is a widely used method for comparing different test oracles [ 23]. In this experiment, to determine whether the bugs detected by SRS could be identified by other approaches’ oracle, we selected TQS/DQP, TLP, and NoREC for comparison because their fixed transformation patterns allow for reliable manual validation. We excluded EET, Pinolo, and PQS due to their vast transformation spaces, 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025.",
+      "context_after": "6. Results of applying other oracles on logic bugs detected by SRS. Of the total 33 logic bugs detected by SRS, 13 were undetectable by other oracles. which make manual validation infeasible. Specifically, EET constructs equivalent queries by adding expressions that always evaluate to TRUE or FALSE. However, the search space for such expressions is extremely large, making manual generation infeasible. Similarly, Pinolo and PQS were excluded due to scalability issues: in Pinolo, the space of approximate queries is prohibitively large, and in PQS, there can be infinitely many possible pivoted qu",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 18,
+      "char_offset": 68176,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Similarly, Pinolo and PQS were excluded due to scalability issues: in Pinolo, the space of approximate queries is prohibitively large, and in PQS, there can be infinitely many possible pivoted queries for a given row.",
+      "context_before": "cember 2025. SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis 363:19 15 2 11 1 TQS/DQP ’s oracle TLP ’s oracle NoREC ’s oracle Logic bugs found by SRS 33 Fig. 6. Results of applying other oracles on logic bugs detected by SRS. Of the total 33 logic bugs detected by SRS, 13 were undetectable by other oracles. which make manual validation infeasible. Specifically, EET constructs equivalent queries by adding expressions that always evaluate to TRUE or FALSE. However, the search space for such expressions is extremely large, making manual generation infeasible.",
+      "context_after": "To determine whether SRS can detect join-related logic bugs identified by other approaches, we collected bug reports from baselines, extracted those involving join operations, and then applied SRS’s test oracle to these bug-inducing test cases to assess whether SRS could detect the bug. Results of other oracles on bugs detected by SRS.The results are presented in Figure 6. Out of the 33 logic bugs detected by SRS, 16 can be detected by TQS/DQP’s oracle; 4 can be detected by TLP’s oracle; 2 can be detected by NoREC’s oracle; and 13 are uniquely detected by SRS. For TQS/DQP’s oracle, we incorpo",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 68764,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "Out of the 33 logic bugs detected by SRS, 16 can be detected by TQS/DQP’s oracle; 4 can be detected by TLP’s oracle; 2 can be detected by NoREC’s oracle; and 13 are uniquely detected by SRS.",
+      "context_before": "sible. Similarly, Pinolo and PQS were excluded due to scalability issues: in Pinolo, the space of approximate queries is prohibitively large, and in PQS, there can be infinitely many possible pivoted queries for a given row. To determine whether SRS can detect join-related logic bugs identified by other approaches, we collected bug reports from baselines, extracted those involving join operations, and then applied SRS’s test oracle to these bug-inducing test cases to assess whether SRS could detect the bug. Results of other oracles on bugs detected by SRS.The results are presented in Figure 6.",
+      "context_after": "For TQS/DQP’s oracle, we incorporated query hints and relevant system variables on the buginducing test cases. The result shows that 17 bugs are undetectable by TQS/DQP’s oracle. When TQS/DQP fail to influence the optimizer’s behavior, or when the logic bugs lie in the join implementation unrelated to optimization, TQS/DQP is unable to detect certain issues. For the example shown in Listing 2, despite applying all optimizer switches5and query hints6of MySQL, the bug remained undetectable. The inconsistency between the intersection of a left outer join and a right outer join with an inner join",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 69358,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "For TQS/DQP’s oracle, we incorporated query hints and relevant system variables on the buginducing test cases.",
+      "context_before": "pivoted queries for a given row. To determine whether SRS can detect join-related logic bugs identified by other approaches, we collected bug reports from baselines, extracted those involving join operations, and then applied SRS’s test oracle to these bug-inducing test cases to assess whether SRS could detect the bug. Results of other oracles on bugs detected by SRS.The results are presented in Figure 6. Out of the 33 logic bugs detected by SRS, 16 can be detected by TQS/DQP’s oracle; 4 can be detected by TLP’s oracle; 2 can be detected by NoREC’s oracle; and 13 are uniquely detected by SRS.",
+      "context_after": "The result shows that 17 bugs are undetectable by TQS/DQP’s oracle. When TQS/DQP fail to influence the optimizer’s behavior, or when the logic bugs lie in the join implementation unrelated to optimization, TQS/DQP is unable to detect certain issues. For the example shown in Listing 2, despite applying all optimizer switches5and query hints6of MySQL, the bug remained undetectable. The inconsistency between the intersection of a left outer join and a right outer join with an inner join helps SRS detect this bug. Although the analysis shows that 16 out of the 33 logic bugs detected by SRS could",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 69549,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "The result shows that 17 bugs are undetectable by TQS/DQP’s oracle.",
+      "context_before": "er approaches, we collected bug reports from baselines, extracted those involving join operations, and then applied SRS’s test oracle to these bug-inducing test cases to assess whether SRS could detect the bug. Results of other oracles on bugs detected by SRS.The results are presented in Figure 6. Out of the 33 logic bugs detected by SRS, 16 can be detected by TQS/DQP’s oracle; 4 can be detected by TLP’s oracle; 2 can be detected by NoREC’s oracle; and 13 are uniquely detected by SRS. For TQS/DQP’s oracle, we incorporated query hints and relevant system variables on the buginducing test cases.",
+      "context_after": "When TQS/DQP fail to influence the optimizer’s behavior, or when the logic bugs lie in the join implementation unrelated to optimization, TQS/DQP is unable to detect certain issues. For the example shown in Listing 2, despite applying all optimizer switches5and query hints6of MySQL, the bug remained undetectable. The inconsistency between the intersection of a left outer join and a right outer join with an inner join helps SRS detect this bug. Although the analysis shows that 16 out of the 33 logic bugs detected by SRS could theoretically be detected by DQP’s oracle, the tool of DQP failed to",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 69660,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M33",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "When TQS/DQP fail to influence the optimizer’s behavior, or when the logic bugs lie in the join implementation unrelated to optimization, TQS/DQP is unable to detect certain issues.",
+      "context_before": "ose involving join operations, and then applied SRS’s test oracle to these bug-inducing test cases to assess whether SRS could detect the bug. Results of other oracles on bugs detected by SRS.The results are presented in Figure 6. Out of the 33 logic bugs detected by SRS, 16 can be detected by TQS/DQP’s oracle; 4 can be detected by TLP’s oracle; 2 can be detected by NoREC’s oracle; and 13 are uniquely detected by SRS. For TQS/DQP’s oracle, we incorporated query hints and relevant system variables on the buginducing test cases. The result shows that 17 bugs are undetectable by TQS/DQP’s oracle.",
+      "context_after": "For the example shown in Listing 2, despite applying all optimizer switches5and query hints6of MySQL, the bug remained undetectable. The inconsistency between the intersection of a left outer join and a right outer join with an inner join helps SRS detect this bug. Although the analysis shows that 16 out of the 33 logic bugs detected by SRS could theoretically be detected by DQP’s oracle, the tool of DQP failed to uncover any of them during 24 hours, as shown in Table 5. To determine why a gap exists between theoretical (Figure 6) and practical (Table 5) bug detection in DQP, we checked the j",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 69728,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "Although the analysis shows that 16 out of the 33 logic bugs detected by SRS could theoretically be detected by DQP’s oracle, the tool of DQP failed to uncover any of them during 24 hours, as shown in Table 5.",
+      "context_before": "ncorporated query hints and relevant system variables on the buginducing test cases. The result shows that 17 bugs are undetectable by TQS/DQP’s oracle. When TQS/DQP fail to influence the optimizer’s behavior, or when the logic bugs lie in the join implementation unrelated to optimization, TQS/DQP is unable to detect certain issues. For the example shown in Listing 2, despite applying all optimizer switches5and query hints6of MySQL, the bug remained undetectable. The inconsistency between the intersection of a left outer join and a right outer join with an inner join helps SRS detect this bug.",
+      "context_after": "To determine why a gap exists between theoretical (Figure 6) and practical (Table 5) bug detection in DQP, we checked the join queries generated by DQP’s tool, and found it supports limited join operations (i.e., cross/inner/outer/natural join). In contrast, SRS supports a broader range of join types, including semi joins and anti joins, even though these join types are not explicitly defined in the query—for example, as illustrated in Listing 4. For TLP’s oracle, we manually constructed three follow-up join queries based on the condition in the bug-inducing query, corresponding to the forms",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 70176,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M35",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "To determine why a gap exists between theoretical (Figure 6) and practical (Table 5) bug detection in DQP, we checked the join queries generated by DQP’s tool, and found it supports limited join operations (i.",
+      "context_before": "or when the logic bugs lie in the join implementation unrelated to optimization, TQS/DQP is unable to detect certain issues. For the example shown in Listing 2, despite applying all optimizer switches5and query hints6of MySQL, the bug remained undetectable. The inconsistency between the intersection of a left outer join and a right outer join with an inner join helps SRS detect this bug. Although the analysis shows that 16 out of the 33 logic bugs detected by SRS could theoretically be detected by DQP’s oracle, the tool of DQP failed to uncover any of them during 24 hours, as shown in Table 5.",
+      "context_after": "e., cross/inner/outer/natural join). In contrast, SRS supports a broader range of join types, including semi joins and anti joins, even though these join types are not explicitly defined in the query—for example, as illustrated in Listing 4. For TLP’s oracle, we manually constructed three follow-up join queries based on the condition in the bug-inducing query, corresponding to the forms p,NOT p, and p IS NULL. We then examined whether the union of the results from these three queries covered all relevant relations. Our analysis shows that four of the bugs detected by SRS can also be identified",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 70386,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M36",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For TLP’s oracle, we manually constructed three follow-up join queries based on the condition in the bug-inducing query, corresponding to the forms p,NOT p, and p IS NULL.",
+      "context_before": "detected by SRS could theoretically be detected by DQP’s oracle, the tool of DQP failed to uncover any of them during 24 hours, as shown in Table 5. To determine why a gap exists between theoretical (Figure 6) and practical (Table 5) bug detection in DQP, we checked the join queries generated by DQP’s tool, and found it supports limited join operations (i.e., cross/inner/outer/natural join). In contrast, SRS supports a broader range of join types, including semi joins and anti joins, even though these join types are not explicitly defined in the query—for example, as illustrated in Listing 4.",
+      "context_after": "We then examined whether the union of the results from these three queries covered all relevant relations. Our analysis shows that four of the bugs detected by SRS can also be identified by TLP’s oracle, whereas the remaining logic bugs cannot be detected by it. For NoREC’s oracle, we manually constructed two equivalent queries by leveraging the equivalence between the WHERE and SELECT clauses in the bug-inducing query. Our analysis shows that two of the bugs detected by SRS can also be identified by NoREC’s oracle, while the remaining logic bugs cannot be detected by it. Given that neither N",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 70837,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M37",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Our analysis shows that four of the bugs detected by SRS can also be identified by TLP’s oracle, whereas the remaining logic bugs cannot be detected by it.",
+      "context_before": "eries generated by DQP’s tool, and found it supports limited join operations (i.e., cross/inner/outer/natural join). In contrast, SRS supports a broader range of join types, including semi joins and anti joins, even though these join types are not explicitly defined in the query—for example, as illustrated in Listing 4. For TLP’s oracle, we manually constructed three follow-up join queries based on the condition in the bug-inducing query, corresponding to the forms p,NOT p, and p IS NULL. We then examined whether the union of the results from these three queries covered all relevant relations.",
+      "context_after": "For NoREC’s oracle, we manually constructed two equivalent queries by leveraging the equivalence between the WHERE and SELECT clauses in the bug-inducing query. Our analysis shows that two of the bugs detected by SRS can also be identified by NoREC’s oracle, while the remaining logic bugs cannot be detected by it. Given that neither NoREC nor TLP’s oracle is designed to test the implementation of join operations, it is unsurprising that they fail to detect most of these logic bugs. 5https://dev.mysql.com/doc/refman/8.4/en/switchable-optimizations.html 6https://dev.mysql.com/doc/refman/8.4/en/",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 71116,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M38",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For NoREC’s oracle, we manually constructed two equivalent queries by leveraging the equivalence between the WHERE and SELECT clauses in the bug-inducing query.",
+      "context_before": "ge of join types, including semi joins and anti joins, even though these join types are not explicitly defined in the query—for example, as illustrated in Listing 4. For TLP’s oracle, we manually constructed three follow-up join queries based on the condition in the bug-inducing query, corresponding to the forms p,NOT p, and p IS NULL. We then examined whether the union of the results from these three queries covered all relevant relations. Our analysis shows that four of the bugs detected by SRS can also be identified by TLP’s oracle, whereas the remaining logic bugs cannot be detected by it.",
+      "context_after": "Our analysis shows that two of the bugs detected by SRS can also be identified by NoREC’s oracle, while the remaining logic bugs cannot be detected by it. Given that neither NoREC nor TLP’s oracle is designed to test the implementation of join operations, it is unsurprising that they fail to detect most of these logic bugs. 5https://dev.mysql.com/doc/refman/8.4/en/switchable-optimizations.html 6https://dev.mysql.com/doc/refman/8.4/en/optimizer-hints.html 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:20 Jinhui Lai et al. Listing 5. Example of applying the SRS oracle to",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 71272,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M39",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Our analysis shows that two of the bugs detected by SRS can also be identified by NoREC’s oracle, while the remaining logic bugs cannot be detected by it.",
+      "context_before": "g 4. For TLP’s oracle, we manually constructed three follow-up join queries based on the condition in the bug-inducing query, corresponding to the forms p,NOT p, and p IS NULL. We then examined whether the union of the results from these three queries covered all relevant relations. Our analysis shows that four of the bugs detected by SRS can also be identified by TLP’s oracle, whereas the remaining logic bugs cannot be detected by it. For NoREC’s oracle, we manually constructed two equivalent queries by leveraging the equivalence between the WHERE and SELECT clauses in the bug-inducing query.",
+      "context_after": "Given that neither NoREC nor TLP’s oracle is designed to test the implementation of join operations, it is unsurprising that they fail to detect most of these logic bugs. 5https://dev.mysql.com/doc/refman/8.4/en/switchable-optimizations.html 6https://dev.mysql.com/doc/refman/8.4/en/optimizer-hints.html 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:20 Jinhui Lai et al. Listing 5. Example of applying the SRS oracle to a join-related bug from other tools’ bug list -- A bug-inducing case from PQS's bug list: https://www.manuelrigger.at/dbms-bugs/ CREATE TABLEt0(c0INT); CRE",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 71433,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M40",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Given that neither NoREC nor TLP’s oracle is designed to test the implementation of join operations, it is unsurprising that they fail to detect most of these logic bugs.",
+      "context_before": "NOT p, and p IS NULL. We then examined whether the union of the results from these three queries covered all relevant relations. Our analysis shows that four of the bugs detected by SRS can also be identified by TLP’s oracle, whereas the remaining logic bugs cannot be detected by it. For NoREC’s oracle, we manually constructed two equivalent queries by leveraging the equivalence between the WHERE and SELECT clauses in the bug-inducing query. Our analysis shows that two of the bugs detected by SRS can also be identified by NoREC’s oracle, while the remaining logic bugs cannot be detected by it.",
+      "context_after": "5https://dev.mysql.com/doc/refman/8.4/en/switchable-optimizations.html 6https://dev.mysql.com/doc/refman/8.4/en/optimizer-hints.html 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:20 Jinhui Lai et al. Listing 5. Example of applying the SRS oracle to a join-related bug from other tools’ bug list -- A bug-inducing case from PQS's bug list: https://www.manuelrigger.at/dbms-bugs/ CREATE TABLEt0(c0INT); CREATE TABLEt1(c0INT) ENGINE = MEMORY; INSERT INTOt0(c0)VALUES(0); INSERT INTOt1(c0)VALUES(-1); -- PQS's oracle: expected: row is fetched, actual: no row is fetched SELECT*FR",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 19,
+      "char_offset": 71588,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M41",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Example of applying the SRS oracle to a join-related bug from other tools’ bug list -- A bug-inducing case from PQS's bug list: https://www.",
+      "context_before": "LECT clauses in the bug-inducing query. Our analysis shows that two of the bugs detected by SRS can also be identified by NoREC’s oracle, while the remaining logic bugs cannot be detected by it. Given that neither NoREC nor TLP’s oracle is designed to test the implementation of join operations, it is unsurprising that they fail to detect most of these logic bugs. 5https://dev.mysql.com/doc/refman/8.4/en/switchable-optimizations.html 6https://dev.mysql.com/doc/refman/8.4/en/optimizer-hints.html 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:20 Jinhui Lai et al. Listing 5.",
+      "context_after": "manuelrigger.at/dbms-bugs/ CREATE TABLEt0(c0INT); CREATE TABLEt1(c0INT) ENGINE = MEMORY; INSERT INTOt0(c0)VALUES(0); INSERT INTOt1(c0)VALUES(-1); -- PQS's oracle: expected: row is fetched, actual: no row is fetched SELECT*FROMt0, t1WHERE(CAST(t1.c0ASUNSIGNED)) > (IFNULL(\"u\", t0.c0)); -- SRS's oracle: \"{0, -1} != {0, null} intersect {0, -1}\", violating rule R06 in Table 1 SELECT*FROMt0INNER JOINt1ON(CAST(t1.c0ASUNSIGNED)) <= (IFNULL(\"u\", t0.c0)); -- {0, -1} SELECT*FROMt0LEFT JOINt1ON(CAST(t1.c0ASUNSIGNED)) <= (IFNULL(\"u\", t0.c0)); -- {0, null} SELECT*FROMt0RIGHT JOINt1ON(CAST(t1.c0ASUNSIGNED))",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 20,
+      "char_offset": 71994,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M42",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "at/dbms-bugs/ CREATE TABLEt0(c0INT); CREATE TABLEt1(c0INT) ENGINE = MEMORY; INSERT INTOt0(c0)VALUES(0); INSERT INTOt1(c0)VALUES(-1); -- PQS's oracle: expected: row is fetched, actual: no row is fetched SELECT*FROMt0, t1WHERE(CAST(t1.",
+      "context_before": "ing logic bugs cannot be detected by it. Given that neither NoREC nor TLP’s oracle is designed to test the implementation of join operations, it is unsurprising that they fail to detect most of these logic bugs. 5https://dev.mysql.com/doc/refman/8.4/en/switchable-optimizations.html 6https://dev.mysql.com/doc/refman/8.4/en/optimizer-hints.html 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:20 Jinhui Lai et al. Listing 5. Example of applying the SRS oracle to a join-related bug from other tools’ bug list -- A bug-inducing case from PQS's bug list: https://www.manuelrigger.",
+      "context_after": "c0ASUNSIGNED)) > (IFNULL(\"u\", t0.c0)); -- SRS's oracle: \"{0, -1} != {0, null} intersect {0, -1}\", violating rule R06 in Table 1 SELECT*FROMt0INNER JOINt1ON(CAST(t1.c0ASUNSIGNED)) <= (IFNULL(\"u\", t0.c0)); -- {0, -1} SELECT*FROMt0LEFT JOINt1ON(CAST(t1.c0ASUNSIGNED)) <= (IFNULL(\"u\", t0.c0)); -- {0, null} SELECT*FROMt0RIGHT JOINt1ON(CAST(t1.c0ASUNSIGNED)) <= (IFNULL(\"u\", t0.c0)); -- {0, -1} Results of SRS oracle on bugs detected by other tools.We collected a total of 75 logic bugs involving join operations from the bug lists of baselines. We then applied the SRS’s oracle to these bug-inducing case",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 20,
+      "char_offset": 72148,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M43",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "We then applied the SRS’s oracle to these bug-inducing cases and successfully reproduced 69 of them, including 16 out of 18 in TQS, 14 out of 15 in DQP, 5 out of 6 in PQS, all 5 in NoREC, 16 out of 17 in TLP, 7 out of 8 in Pinolo, and all 6 in EET.",
+      "context_before": "d, actual: no row is fetched SELECT*FROMt0, t1WHERE(CAST(t1.c0ASUNSIGNED)) > (IFNULL(\"u\", t0.c0)); -- SRS's oracle: \"{0, -1} != {0, null} intersect {0, -1}\", violating rule R06 in Table 1 SELECT*FROMt0INNER JOINt1ON(CAST(t1.c0ASUNSIGNED)) <= (IFNULL(\"u\", t0.c0)); -- {0, -1} SELECT*FROMt0LEFT JOINt1ON(CAST(t1.c0ASUNSIGNED)) <= (IFNULL(\"u\", t0.c0)); -- {0, null} SELECT*FROMt0RIGHT JOINt1ON(CAST(t1.c0ASUNSIGNED)) <= (IFNULL(\"u\", t0.c0)); -- {0, -1} Results of SRS oracle on bugs detected by other tools.We collected a total of 75 logic bugs involving join operations from the bug lists of baselines.",
+      "context_after": "This high rate of detection demonstrates the effectiveness of the SRS oracle in capturing a wide range of join-related logic bugs. Listing 5 demonstrates the SRS oracle applied to a join-related bug from PQS’s bug list. The example shows a violation of rule R06 in Table 1, where the intersection of a left and right outer join’s results is not equivalent to the inner join’s result, thereby illustrating SRS’s effectiveness. The remaining six bugs that SRS could not detect involve queries that contain join operations but have root causes unrelated to the join logic itself. For example, in one su",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 20,
+      "char_offset": 72921,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp",
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M44",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Listing 5 demonstrates the SRS oracle applied to a join-related bug from PQS’s bug list.",
+      "context_before": "T JOINt1ON(CAST(t1.c0ASUNSIGNED)) <= (IFNULL(\"u\", t0.c0)); -- {0, -1} Results of SRS oracle on bugs detected by other tools.We collected a total of 75 logic bugs involving join operations from the bug lists of baselines. We then applied the SRS’s oracle to these bug-inducing cases and successfully reproduced 69 of them, including 16 out of 18 in TQS, 14 out of 15 in DQP, 5 out of 6 in PQS, all 5 in NoREC, 16 out of 17 in TLP, 7 out of 8 in Pinolo, and all 6 in EET. This high rate of detection demonstrates the effectiveness of the SRS oracle in capturing a wide range of join-related logic bugs.",
+      "context_after": "The example shows a violation of rule R06 in Table 1, where the intersection of a left and right outer join’s results is not equivalent to the inner join’s result, thereby illustrating SRS’s effectiveness. The remaining six bugs that SRS could not detect involve queries that contain join operations but have root causes unrelated to the join logic itself. For example, in one such case, the bug was caused by an incorrect result from the MAX() function used in the HAVING clause. Although the bug can be triggered when two tables are joined, its root cause is not related to the join itself, and th",
+      "section": "5.4 Comparison With the State-of-the-Art",
+      "page": 20,
+      "char_offset": 73301,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M45",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The implementation beyond SQLancer.",
+      "context_before": "ect these set relations, we can identify logic bugs in join operations and other language features. We received feedback from at least two logic bug reports indicating that the bug exists in language features beyond joins. Besides, if both query variants produce the same incorrect result, like all metamorphic testing tools, SRS may fail to detect certain logic bugs in a single test iteration. Nevertheless, our approach uses the set relation among join operations to construct oracle queries, which naturally perform cross-checking. Thus, SRS may expose these bugs across multiple test iterations.",
+      "context_after": "We made several modifications to SQLancer to enhance the diversity and effectiveness of the generated test cases. Specifically, we adjusted SQLancer to produce schemas with a wider range of data types (e.g., composite types) and to generate more boundary values, such as maximum and minimum values, nulls, and special characters. These boundary values are particularly effective for exposing latent bugs in DBMSs. For instance, one logic bug shown in Figure 1 could only be triggered by such boundary values, not by common inputs. To assess the impact of these enhancements, we conducted an ablation",
+      "section": "6 Discussion",
+      "page": 21,
+      "char_offset": 77500,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M46",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We made several modifications to SQLancer to enhance the diversity and effectiveness of the generated test cases.",
+      "context_before": "tify logic bugs in join operations and other language features. We received feedback from at least two logic bug reports indicating that the bug exists in language features beyond joins. Besides, if both query variants produce the same incorrect result, like all metamorphic testing tools, SRS may fail to detect certain logic bugs in a single test iteration. Nevertheless, our approach uses the set relation among join operations to construct oracle queries, which naturally perform cross-checking. Thus, SRS may expose these bugs across multiple test iterations. The implementation beyond SQLancer.",
+      "context_after": "Specifically, we adjusted SQLancer to produce schemas with a wider range of data types (e.g., composite types) and to generate more boundary values, such as maximum and minimum values, nulls, and special characters. These boundary values are particularly effective for exposing latent bugs in DBMSs. For instance, one logic bug shown in Figure 1 could only be triggered by such boundary values, not by common inputs. To assess the impact of these enhancements, we conducted an ablation study by disabling them. Within a 24-hour testing period, SRS without these strategies detected 3 fewer logic bug",
+      "section": "6 Discussion",
+      "page": 21,
+      "char_offset": 77536,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M47",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, we adjusted SQLancer to produce schemas with a wider range of data types (e.",
+      "context_before": "reports indicating that the bug exists in language features beyond joins. Besides, if both query variants produce the same incorrect result, like all metamorphic testing tools, SRS may fail to detect certain logic bugs in a single test iteration. Nevertheless, our approach uses the set relation among join operations to construct oracle queries, which naturally perform cross-checking. Thus, SRS may expose these bugs across multiple test iterations. The implementation beyond SQLancer.We made several modifications to SQLancer to enhance the diversity and effectiveness of the generated test cases.",
+      "context_after": "g., composite types) and to generate more boundary values, such as maximum and minimum values, nulls, and special characters. These boundary values are particularly effective for exposing latent bugs in DBMSs. For instance, one logic bug shown in Figure 1 could only be triggered by such boundary values, not by common inputs. To assess the impact of these enhancements, we conducted an ablation study by disabling them. Within a 24-hour testing period, SRS without these strategies detected 3 fewer logic bugs compared to the 23 bugs found by the full version, as shown in Table 5. 7 Related Work In",
+      "section": "6 Discussion",
+      "page": 21,
+      "char_offset": 77649,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M48",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Non-optimizing Reference Engine (NoREC) [ 23] assumes that a predicate should evaluate to the same value in both the WHERE and SELECT clauses, and leverages this assumption to construct a non-optimizable form of the query.",
+      "context_before": "t by common inputs. To assess the impact of these enhancements, we conducted an ablation study by disabling them. Within a 24-hour testing period, SRS without these strategies detected 3 fewer logic bugs compared to the 23 bugs found by the full version, as shown in Table 5. 7 Related Work In this section, we focus on several DBMS testing tasks and highlight how they differ from SRS. Metamorphic testing of DBMSs.Metamorphic testing [ 6] addresses the test oracle problem by generating new inputs based on an existing input-output pair, inferring the expected result through metamorphic relations.",
+      "context_after": "The original and transformed queries are expected to produce consistent results. Differential Query Execution (DQE) [ 27] adopts a similar idea to NoREC, assuming that a predicate should yield consistent results, regardless of whether it appears in a SELECT ,UPDATE, or DELETE statement. Equivalent Expression Transformation (EET) [ 13] introduces tautologies and contradictions into expressions, ensuring that the transformed query produces consistent results with the original one. Pinolo [ 11] transforms predicates to make them either more permissive or more restrictive, thereby generating over-",
+      "section": "7 Related Work",
+      "page": 21,
+      "char_offset": 78647,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M49",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Differential Query Execution (DQE) [ 27] adopts a similar idea to NoREC, assuming that a predicate should yield consistent results, regardless of whether it appears in a SELECT ,UPDATE, or DELETE statement.",
+      "context_before": "on, we focus on several DBMS testing tasks and highlight how they differ from SRS. Metamorphic testing of DBMSs.Metamorphic testing [ 6] addresses the test oracle problem by generating new inputs based on an existing input-output pair, inferring the expected result through metamorphic relations. Non-optimizing Reference Engine (NoREC) [ 23] assumes that a predicate should evaluate to the same value in both the WHERE and SELECT clauses, and leverages this assumption to construct a non-optimizable form of the query. The original and transformed queries are expected to produce consistent results.",
+      "context_after": "Equivalent Expression Transformation (EET) [ 13] introduces tautologies and contradictions into expressions, ensuring that the transformed query produces consistent results with the original one. Pinolo [ 11] transforms predicates to make them either more permissive or more restrictive, thereby generating over-approximations or underapproximations of the original query. CODDTest [ 32] leverages constant folding and propagation on expressions to produce equivalent queries. Ternary Logic Partitioning (TLP) [ 24] assumes that a predicate pcan evaluate to TRUE,FALSE, or NULL. Based on this, a que",
+      "section": "7 Related Work",
+      "page": 21,
+      "char_offset": 78951,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M50",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "CODDTest [ 32] leverages constant folding and propagation on expressions to produce equivalent queries.",
+      "context_before": "e consistent results. Differential Query Execution (DQE) [ 27] adopts a similar idea to NoREC, assuming that a predicate should yield consistent results, regardless of whether it appears in a SELECT ,UPDATE, or DELETE statement. Equivalent Expression Transformation (EET) [ 13] introduces tautologies and contradictions into expressions, ensuring that the transformed query produces consistent results with the original one. Pinolo [ 11] transforms predicates to make them either more permissive or more restrictive, thereby generating over-approximations or underapproximations of the original query.",
+      "context_after": "Ternary Logic Partitioning (TLP) [ 24] assumes that a predicate pcan evaluate to TRUE,FALSE, or NULL. Based on this, a query can be decomposed into three follow-up queries that return the results corresponding to p,NOT p, and p IS NULL, respectively. The union of these follow-up query results should be equivalent to the result of the original query. Although both TLP and SRS use the UNION 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:22 Jinhui Lai et al. set relation as their test oracle, TLP and SRS differ in target and methodology: (1) SRS focuses on issues related t",
+      "section": "7 Related Work",
+      "page": 21,
+      "char_offset": 79530,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M51",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "Ternary Logic Partitioning (TLP) [ 24] assumes that a predicate pcan evaluate to TRUE,FALSE, or NULL.",
+      "context_before": "that a predicate should yield consistent results, regardless of whether it appears in a SELECT ,UPDATE, or DELETE statement. Equivalent Expression Transformation (EET) [ 13] introduces tautologies and contradictions into expressions, ensuring that the transformed query produces consistent results with the original one. Pinolo [ 11] transforms predicates to make them either more permissive or more restrictive, thereby generating over-approximations or underapproximations of the original query. CODDTest [ 32] leverages constant folding and propagation on expressions to produce equivalent queries.",
+      "context_after": "Based on this, a query can be decomposed into three follow-up queries that return the results corresponding to p,NOT p, and p IS NULL, respectively. The union of these follow-up query results should be equivalent to the result of the original query. Although both TLP and SRS use the UNION 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:22 Jinhui Lai et al. set relation as their test oracle, TLP and SRS differ in target and methodology: (1) SRS focuses on issues related to the implementation bugs of join operations, while TLP targets bugs in predicate handling code via qu",
+      "section": "7 Related Work",
+      "page": 21,
+      "char_offset": 79634,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M52",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Although both TLP and SRS use the UNION 3, No.",
+      "context_before": "es to make them either more permissive or more restrictive, thereby generating over-approximations or underapproximations of the original query. CODDTest [ 32] leverages constant folding and propagation on expressions to produce equivalent queries. Ternary Logic Partitioning (TLP) [ 24] assumes that a predicate pcan evaluate to TRUE,FALSE, or NULL. Based on this, a query can be decomposed into three follow-up queries that return the results corresponding to p,NOT p, and p IS NULL, respectively. The union of these follow-up query results should be equivalent to the result of the original query.",
+      "context_after": "6 (SIGMOD), Article 363. Publication date: December 2025. 363:22 Jinhui Lai et al. set relation as their test oracle, TLP and SRS differ in target and methodology: (1) SRS focuses on issues related to the implementation bugs of join operations, while TLP targets bugs in predicate handling code via query decomposition. (2) SRS is a mathematically rigorous framework based on four core set relations (i.e., Equivalence, Intersection, Disjointness, Union) to formally characterize join behavior. TLP decomposes a query into 3 based on predicates with TRUE/FALSE/NULLresults. Transformed Query Synthesi",
+      "section": "7 Related Work",
+      "page": 21,
+      "char_offset": 79986,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M53",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "set relation as their test oracle, TLP and SRS differ in target and methodology: (1) SRS focuses on issues related to the implementation bugs of join operations, while TLP targets bugs in predicate handling code via query decomposition.",
+      "context_before": "iginal query. CODDTest [ 32] leverages constant folding and propagation on expressions to produce equivalent queries. Ternary Logic Partitioning (TLP) [ 24] assumes that a predicate pcan evaluate to TRUE,FALSE, or NULL. Based on this, a query can be decomposed into three follow-up queries that return the results corresponding to p,NOT p, and p IS NULL, respectively. The union of these follow-up query results should be equivalent to the result of the original query. Although both TLP and SRS use the UNION 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:22 Jinhui Lai et al.",
+      "context_after": "(2) SRS is a mathematically rigorous framework based on four core set relations (i.e., Equivalence, Intersection, Disjointness, Union) to formally characterize join behavior. TLP decomposes a query into 3 based on predicates with TRUE/FALSE/NULLresults. Transformed Query Synthesis (TQS) [ 28] and Differential Query Plan (DQP) [ 3] target logic bugs in join optimizations and leverage equivalence relationships among different query plans generated for the same query. SRS is orthogonal to TQS and DQP in targets and methods: (1) TQS and DQP target join optimization bugs, while SRS focuses on join",
+      "section": "7 Related Work",
+      "page": 22,
+      "char_offset": 80117,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M54",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP decomposes a query into 3 based on predicates with TRUE/FALSE/NULLresults.",
+      "context_before": "should be equivalent to the result of the original query. Although both TLP and SRS use the UNION 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:22 Jinhui Lai et al. set relation as their test oracle, TLP and SRS differ in target and methodology: (1) SRS focuses on issues related to the implementation bugs of join operations, while TLP targets bugs in predicate handling code via query decomposition. (2) SRS is a mathematically rigorous framework based on four core set relations (i.e., Equivalence, Intersection, Disjointness, Union) to formally characterize join behavior.",
+      "context_after": "Transformed Query Synthesis (TQS) [ 28] and Differential Query Plan (DQP) [ 3] target logic bugs in join optimizations and leverage equivalence relationships among different query plans generated for the same query. SRS is orthogonal to TQS and DQP in targets and methods: (1) TQS and DQP target join optimization bugs, while SRS focuses on join implementation bugs. Thus, SRS may also detect implementation bugs unrelated to optimization. (2) TQS and DQP construct test oracles by using query hints to execute different plans, while SRS generates test oracles by transforming the order, type, and c",
+      "section": "7 Related Work",
+      "page": 22,
+      "char_offset": 80529,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M55",
+      "found_by": "technique",
+      "surface": "Differential Query Plan",
+      "technique": "dqp",
+      "sentence": "Transformed Query Synthesis (TQS) [ 28] and Differential Query Plan (DQP) [ 3] target logic bugs in join optimizations and leverage equivalence relationships among different query plans generated for the same query.",
+      "context_before": "d SRS use the UNION 3, No. 6 (SIGMOD), Article 363. Publication date: December 2025. 363:22 Jinhui Lai et al. set relation as their test oracle, TLP and SRS differ in target and methodology: (1) SRS focuses on issues related to the implementation bugs of join operations, while TLP targets bugs in predicate handling code via query decomposition. (2) SRS is a mathematically rigorous framework based on four core set relations (i.e., Equivalence, Intersection, Disjointness, Union) to formally characterize join behavior. TLP decomposes a query into 3 based on predicates with TRUE/FALSE/NULLresults.",
+      "context_after": "SRS is orthogonal to TQS and DQP in targets and methods: (1) TQS and DQP target join optimization bugs, while SRS focuses on join implementation bugs. Thus, SRS may also detect implementation bugs unrelated to optimization. (2) TQS and DQP construct test oracles by using query hints to execute different plans, while SRS generates test oracles by transforming the order, type, and conditions of join operations to synthesize set relations. However, not all join operations can trigger the generation of multiple query plans. As a result, some join-related bugs detected by SRS cannot be uncovered b",
+      "section": "7 Related Work",
+      "page": 22,
+      "char_offset": 80607,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M56",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "SRS is orthogonal to TQS and DQP in targets and methods: (1) TQS and DQP target join optimization bugs, while SRS focuses on join implementation bugs.",
+      "context_before": "s related to the implementation bugs of join operations, while TLP targets bugs in predicate handling code via query decomposition. (2) SRS is a mathematically rigorous framework based on four core set relations (i.e., Equivalence, Intersection, Disjointness, Union) to formally characterize join behavior. TLP decomposes a query into 3 based on predicates with TRUE/FALSE/NULLresults. Transformed Query Synthesis (TQS) [ 28] and Differential Query Plan (DQP) [ 3] target logic bugs in join optimizations and leverage equivalence relationships among different query plans generated for the same query.",
+      "context_after": "Thus, SRS may also detect implementation bugs unrelated to optimization. (2) TQS and DQP construct test oracles by using query hints to execute different plans, while SRS generates test oracles by transforming the order, type, and conditions of join operations to synthesize set relations. However, not all join operations can trigger the generation of multiple query plans. As a result, some join-related bugs detected by SRS cannot be uncovered by these two approaches. Differential testing of DBMSs.Differential testing [ 19] involves executing the same input on multiple versions of a system or",
+      "section": "7 Related Work",
+      "page": 22,
+      "char_offset": 80823,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M57",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "(2) TQS and DQP construct test oracles by using query hints to execute different plans, while SRS generates test oracles by transforming the order, type, and conditions of join operations to synthesize set relations.",
+      "context_before": "alence, Intersection, Disjointness, Union) to formally characterize join behavior. TLP decomposes a query into 3 based on predicates with TRUE/FALSE/NULLresults. Transformed Query Synthesis (TQS) [ 28] and Differential Query Plan (DQP) [ 3] target logic bugs in join optimizations and leverage equivalence relationships among different query plans generated for the same query. SRS is orthogonal to TQS and DQP in targets and methods: (1) TQS and DQP target join optimization bugs, while SRS focuses on join implementation bugs. Thus, SRS may also detect implementation bugs unrelated to optimization.",
+      "context_after": "However, not all join operations can trigger the generation of multiple query plans. As a result, some join-related bugs detected by SRS cannot be uncovered by these two approaches. Differential testing of DBMSs.Differential testing [ 19] involves executing the same input on multiple versions of a system or on comparable systems, with the expectation that they produce consistent outputs. Any discrepancies observed may indicate the presence of logic bugs. RAGS [ 26] was one of the earliest applications of differential testing to DBMSs, identifying bugs by comparing query results across differe",
+      "section": "7 Related Work",
+      "page": 22,
+      "char_offset": 81047,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M58",
+      "found_by": "technique",
+      "surface": "Cardinality Estimation Restriction Testing",
+      "technique": "cert",
+      "sentence": "Cardinality Estimation Restriction Testing (CERT) [ 2] is designed to detect performance issues from the perspective of cardinality estimation.",
+      "context_before": "ues focus on identifying securityrelevant bugs, such as memory errors. For instance, SQLSmith [ 1] and DynSQL [ 12] use grammarbased approaches to generate test cases aimed at detecting memory errors. Inspired by greybox fuzzers like AFL [ 31], Squirrel [ 33] employs code coverage as a guide to uncover memory issues, intelligently generating inputs to explore more code paths. In addition to security issues, performance issues in DBMSs have also been a major focus of research. Amoeba [ 17] constructs semantically equivalent query pairs and compares their response times to find performance bugs.",
+      "context_after": "Given a query, CERT generates a more restrictive variant under the assumption that the cardinality estimator should predict a lower number of rows returned compared to the original query. Jung et al. [14] proposed the APOLLO system, which leverages differential testing to detect performance regression bugs in DBMSs. Database and SQL query generation.Multiple prior works have focused on automatically generating database states and SQL queries, serving as complementary approaches to our research. Gray et al. [9] introduced a set of techniques designed to efficiently generate billions of databas",
+      "section": "7 Related Work",
+      "page": 22,
+      "char_offset": 82870,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M59",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "Given a query, CERT generates a more restrictive variant under the assumption that the cardinality estimator should predict a lower number of rows returned compared to the original query.",
+      "context_before": "to generate test cases aimed at detecting memory errors. Inspired by greybox fuzzers like AFL [ 31], Squirrel [ 33] employs code coverage as a guide to uncover memory issues, intelligently generating inputs to explore more code paths. In addition to security issues, performance issues in DBMSs have also been a major focus of research. Amoeba [ 17] constructs semantically equivalent query pairs and compares their response times to find performance bugs. Cardinality Estimation Restriction Testing (CERT) [ 2] is designed to detect performance issues from the perspective of cardinality estimation.",
+      "context_after": "Jung et al. [14] proposed the APOLLO system, which leverages differential testing to detect performance regression bugs in DBMSs. Database and SQL query generation.Multiple prior works have focused on automatically generating database states and SQL queries, serving as complementary approaches to our research. Gray et al. [9] introduced a set of techniques designed to efficiently generate billions of database records. Data Generation Language (DGL) [ 5] is a domain-specific language for generating data exhibiting complex correlations within and across tables. Query-aware database state genera",
+      "section": "7 Related Work",
+      "page": 22,
+      "char_offset": 83014,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "uses_infrastructure": [
+        "M23"
+      ],
+      "compares_with": [
+        "M12",
+        "M20",
+        "M21",
+        "M23"
+      ],
+      "describes_as_state_of_the_art": [
+        "M1",
+        "M12"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:11:45Z",
+    "is_model_written": true,
+    "summary": "SRS detects logic bugs in how a DBMS implements joins rather than in how it optimises them. The authors argue that existing work perturbs query hints and system variables to change the optimizer's plan choice, which fails when those knobs do not influence the optimizer or when the fault lies in join implementation code unrelated to optimisation. SRS instead transforms a join query -- changing join types, orders and conditions -- so the original and transformed results must stand in a known set relation, and treats a violation as a bug. Across five extensively tested DBMSs it found 33 previously unknown bugs, all confirmed.",
+    "narrative": "SRS is implemented on SQLancer, generating its database state with SQLancer's syntax-rule-based random generation and keeping SQLancer's query generation strategy for features outside the join. SQLancer's oracles are then also its baselines: PQS, NoREC, TLP and DQP are four of the six approaches it is measured against, and it reports finding 9 to 12 more bugs than each.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M4": "definition",
+      "M6": "reuse_component",
+      "M7": "reuse_component",
+      "M8": "reuse_component",
+      "M9": "reuse_component",
+      "M12": "state_of_the_art",
+      "M13": "background",
+      "M14": "background",
+      "M15": "background",
+      "M16": "definition",
+      "M17": "definition",
+      "M18": "definition",
+      "M19": "background",
+      "M20": "baseline",
+      "M21": "result_comparison",
+      "M22": "result_comparison",
+      "M23": "reuse_implementation",
+      "M24": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M7",
+          "M9",
+          "M23"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M7",
+            "sentence": "SRS applies the automated, syntax-rule-based random generation approach of SQLancer to ensure that the database state exhibits sufficient diversity, thereby enabling thorough and effective bug detection.",
+            "section": "4 Implementation",
+            "page": 11
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "For the generation of other language features, aside from considering the tables involved in joins, our approach remains consistent with SQLancer’s query generation strategy for NoREC test oracle [23].",
+            "section": "4 Implementation",
+            "page": 12
+          },
+          {
+            "mention_id": "M23",
+            "sentence": "We compared the code coverage achieved by these approaches, and the results showed that the test oracles implemented on SQLancer— PQS, NoREC, TLP, DQP, and SRS—achieved similar code coverage.",
+            "section": "5.4 Comparison With the State-of-the-Art",
+            "page": 18
+          }
+        ],
+        "reasoning": "M7 states SRS applies SQLancer's automated random generation for the database state and M9 that its query generation follows SQLancer's strategy for other language features; M23 describes SRS itself as one of the test oracles implemented on SQLancer, alongside PQS, NoREC, TLP and DQP.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Set Relation Synthesis is presented as its own oracle over join transformations. TLP is discussed as a related set-based approach whose transformation targets differ, not as something SRS generalises."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M12",
+          "M20",
+          "M21",
+          "M24"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M12",
+            "sentence": "We selected six state-of-the-art approaches for comparison: PQS [ 25], TLP [ 24], NoREC [ 23], Pinolo [ 11], EET [ 13], and DQP [ 3].",
+            "section": "5.4 Comparison With the State-of-the-Art",
+            "page": 17
+          },
+          {
+            "mention_id": "M20",
+            "sentence": "As such, we include a comparison with NoREC in our evaluation.",
+            "section": "5.4 Comparison With the State-of-the-Art",
+            "page": 17
+          },
+          {
+            "mention_id": "M21",
+            "sentence": "Table 5 shows that SRS found 12, 9, 10, 20, 17, and 13 more bugs than PQS, NoREC, TLP, DQP, Pinolo, and EET, respectively.",
+            "section": "5.4 Comparison With the State-of-the-Art",
+            "page": 17
+          },
+          {
+            "mention_id": "M24",
+            "sentence": "For instance, when testing MySQL for 24 hours, PQS, TLP, DQP, and SRS achieved line coverage rates of 23.",
+            "section": "5.4 Comparison With the State-of-the-Art",
+            "page": 18
+          }
+        ],
+        "reasoning": "Four of the six baselines are SQLancer oracles, all run for 24 hours on the same systems, with bug counts and line coverage reported for each.",
+        "techniques": [
+          "pqs",
+          "norec",
+          "tlp",
+          "dqp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M1",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M1",
+            "sentence": "TQS [ 28] and DQP [ 3] are two state-of-the-art testing approaches that target logic bugs in the optimization of join operations.",
+            "section": "1 Introduction",
+            "page": 2
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "We selected six state-of-the-art approaches for comparison: PQS [ 25], TLP [ 24], NoREC [ 23], Pinolo [ 11], EET [ 13], and DQP [ 3].",
+            "section": "5.4 Comparison With the State-of-the-Art",
+            "page": 17
+          }
+        ],
+        "reasoning": "M12 calls the six approaches, PQS, NoREC and TLP among them, state-of-the-art; M1 says the same of DQP for join optimisation."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3769832.json b/_data/papers/paper_doi_10_1145_3769832.json
new file mode 100644
index 0000000..44c387c
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3769832.json
@@ -0,0 +1,404 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:03Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3769832",
+    "title": "Test Data Generation for Complex SQL Queries",
+    "authors": [
+      "Sunanda Somwase",
+      "Parismita Das",
+      "S. Sudarshan"
+    ],
+    "year": 2025,
+    "venue": "Proc. ACM Manag. Data",
+    "doi": "10.1145/3769832",
+    "arxiv_id": null,
+    "s2_paper_id": "d05ca28c4eaaf47e4757524b7638551ba0881cd6",
+    "url": "https://doi.org/10.1145/3769832",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3769832",
+      "retrieved_at": "2026-09-09T01:31:03Z",
+      "chars": 88400,
+      "content_sha256": "sha256:e8b3ae7a073b845f2e5c6ddb91f075ff55a48380c9b2c8306ace66cbf8cd290e"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 26,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1486
+      },
+      {
+        "number": "2",
+        "title": "Background and Solution Approach",
+        "start": 9920
+      },
+      {
+        "number": "3",
+        "title": "Overview of Our Contributions",
+        "start": 13163
+      },
+      {
+        "number": "3.1",
+        "title": "Multiset Representation",
+        "start": 13923
+      },
+      {
+        "number": "3.2",
+        "title": "Data Types",
+        "start": 16316
+      },
+      {
+        "number": "3.3",
+        "title": "Handling Integrity Constraints",
+        "start": 18435
+      },
+      {
+        "number": "3.4",
+        "title": "Result Table Approach",
+        "start": 19852
+      },
+      {
+        "number": "3.5",
+        "title": "Datasets to Kill Mutations",
+        "start": 22024
+      },
+      {
+        "number": "3.6",
+        "title": "Handling Correlated Queries",
+        "start": 22872
+      },
+      {
+        "number": "4",
+        "title": "Result Table Constraints",
+        "start": 23177
+      },
+      {
+        "number": "4.1",
+        "title": "Join Result Table",
+        "start": 24215
+      },
+      {
+        "number": "4.2",
+        "title": "Selection Result Table",
+        "start": 31188
+      },
+      {
+        "number": "4.3",
+        "title": "Aggregation Result Table",
+        "start": 32374
+      },
+      {
+        "number": "4.4",
+        "title": "Set Operation Result Table",
+        "start": 37691
+      },
+      {
+        "number": "5",
+        "title": "Dataset Generation",
+        "start": 39290
+      },
+      {
+        "number": "5.2",
+        "title": "Deciding Number of Tuples",
+        "start": 42584
+      },
+      {
+        "number": "5.3",
+        "title": "Handling String Types",
+        "start": 47073
+      },
+      {
+        "number": "6",
+        "title": "Handling Nested Subqueries",
+        "start": 53670
+      },
+      {
+        "number": "6.1",
+        "title": "Single-Level Simple Subqueries",
+        "start": 55046
+      },
+      {
+        "number": "6.2",
+        "title": "Single-Level Complex Subqueries",
+        "start": 58569
+      },
+      {
+        "number": "6.3",
+        "title": "Multi-Level Nesting of Subqueries",
+        "start": 61333
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 62382
+      },
+      {
+        "number": "8",
+        "title": "Performance Study",
+        "start": 67154
+      },
+      {
+        "number": "8.1",
+        "title": "Effectiveness in Killing Mutations",
+        "start": 69929
+      },
+      {
+        "number": "8.2",
+        "title": "Execution Time",
+        "start": 75016
+      },
+      {
+        "number": "9",
+        "title": "Conclusion and Future Work",
+        "start": 79717
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Bikash Chandra, Ananyo Banerjee, Udbhas Hazra, Mathew Joseph, and S. Sudarshan. 2019. Automated Grading of SQL Queries. InIEEE Intl. Conf. on Data Engineering, ICDE. IEEE, 1630–1633. https://doi.org/10.1109/ICDE.2019.00159",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Bikash Chandra, Bhupesh Chawda, Biplab Kar, K. V. Maheshwara Reddy, Shetal Shah, and S. Sudarshan. 2015. Data generation for testing and grading SQL queries.VLDB J.24, 6 (2015), 731–755. https://doi.org/10.1007/s00778-0150395-0",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Bikash Chandra, Bhupesh Chawda, Shetal Shah, S. Sudarshan, and Ankit Shah. 2013. Extending XData to kill SQL query mutants in the wild. InProcs. International Workshop on Testing Database Systems, DBTest 2013. ACM, 2:1–2:6. https://doi.org/10.1145/2479440.2479442",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Shumo Chu, Brendan Murphy, Jared Roesch, Alvin Cheung, and Dan Suciu. 2018. Axiomatic Foundations and Algorithms for Deciding Semantic Equivalences of SQL Queries.Proc. VLDB Endow.11, 11 (2018), 1482–1495. https: //doi.org/10.14778/3236187.3236200",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Claudio de la Riva, María José Suárez-Cabal, and Javier Tuya. 2010. Constraint-based test database generation for SQL queries. InProceedings of the 5th Workshop on Automation of Software Test (AST ’10). 67–74. https://doi.org/10.1145/ 1808266.1808276",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Leonardo Mendonça de Moura and Nikolaj S. Bjørner. 2008. Z3: An Efficient SMT Solver. InTools and Algorithms for the Construction and Analysis of Systems, TACAS (Lecture Notes in Computer Science), Vol. 4963. Springer, 337–340. https://doi.org/10.1007/978-3-540-78800-3_24",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Yuetang Deng, Phyllis Frankl, and David Chays. 2005. Testing database transactions with AGENDA. InProceedings of the 27th International Conference on Software Engineering(St. Louis, MO, USA)(ICSE ’05). Association for Computing Machinery, New York, NY, USA, 78–87. https://doi.org/10.1145/1062455.1062486",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Haoran Ding, Zhaoguo Wang, Yicun Yang, Dexin Zhang, Zhenglin Xu, Haibo Chen, Ruzica Piskac, and Jinyang Li. 2023. Proving Query Equivalence Using Linear Integer Arithmetic. https://doi.org/10.1145/3626768",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Mostafa Elhemali, César A. Galindo-Legaria, Torsten Grabs, and Milind Joshi. 2007. Execution strategies for SQL subqueries. InProcs. ACMSIGMOD. ACM, 993–1004. https://doi.org/10.1145/1247480.1247598",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Florian Haftmann, Donald Kossmann, and Alexander Kreutz1. 2005. Efficient Regression Tests for Database Applications. InProcs. of the Conference on Innovative Database Systems Research (CIDR).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Yang He, Pinhan Zhao, Xinyu Wang, and Yuepeng Wang. 2024. VeriEQL: Bounded Equivalence Verification for Complex SQL Queries with Integrity Constraints.Proc. ACM Program. Lang.8, OOPSLA1 (2024), 1071–1099. https: //doi.org/10.1145/3649849",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Fangyu Lei, Jixuan Chen, Yuxiao Ye, Ruisheng Cao, Dongchan Shin, Hongjin Su, Zhaoqing Suo, Hongcheng Gao, Wenjing Hu, Pengcheng Yin, Victor Zhong, Caiming Xiong, Ruoxi Sun, Qian Liu, Sida Wang, and Tao Yu. 2025. Spider 2.0: Evaluating Language Models on Real-World Enterprise Text-to-SQL Workflows. arXiv:2411.07763 [cs.CL] https://arxiv.org/abs/2411.07763",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jinyang Li, Binyuan Hui, Ge Qu, Jiaxi Yang, Binhua Li, Bowen Li, Bailin Wang, Bowen Qin, Ruiying Geng, Nan Huo, Xuanhe Zhou, Chenhao Ma, Guoliang Li, Kevin Chen-Chuan Chang, Fei Huang, Reynold Cheng, and Yongbin Li. 2023. Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLs. In NeurIPS 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Mudathir Mahgoub Yahia Mohamed, Andrew Reynolds, Cesare Tinelli, and Clark Barrett. 2024. Verifying SQL queries using theories of tables and relations. InProceedings of 25th Conference on Logic for Programming, Artificial Intelligence and Reasoning (EPiC Series in Computing), Nikolaj Bj{\\o}rner, Marijn Heule, and Andrei Voronkov (Eds.), Vol. 100. EasyChair, 445–463. https://doi.org/10.29007/rlt7",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Arpit Narechania, Adam Fourney, Bongshin Lee, and Gonzalo Ramos. 2021. DIY: Assessing the Correctness of Natural Language to SQL Systems. InProceedings of the 26th International Conference on Intelligent User Interfaces (IUI ’21). 597–607. https://doi.org/10.1145/3397481.3450667",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Thomas Neumann. 2025. Improving Unnesting of Complex Queries. InDatenbanksysteme für Business, Technologie und Web (BTW 2025). Gesellschaft für Informatik, Bonn, 25–47. https://doi.org/10.18420/BTW2025-01",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Thomas Neumann and Alfons Kemper. 2015. Unnesting Arbitrary Queries. InDatenbanksysteme für Business, Technologie und Web (BTW 2015). Gesellschaft für Informatik e.V., Bonn, 383–402.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Zheng Ning, Yuan Tian, Zheng Zhang, Tianyi Zhang, and Toby Jia-Jun Li. 2024. Insights into Natural Language Database Query Errors: from Attention Misalignment to User Handling Strategies.ACM Trans. Interact. Intell. Syst.14, 4, Article 25 (Dec. 2024), 32 pages. https://doi.org/10.1145/3650114",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020). 1140–1152. 3, No. 6 (SIGMOD), Article 367. Publication date: December 2025. 367:26 Sunanda Somwase, Par",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (Nov. 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Shetal Shah, S. Sudarshan, Suhas Kajbaje, Sandeep Patidar, Bhanu Pratap Gupta, and Devang Vira. 2011. Generating test data for killing SQL mutants: A constraint-based approach. InIEEE International Conference on Data Engineering, ICDE. IEEE Computer Society, 1175–1186. https://doi.org/10.1109/ICDE.2011.5767876",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Abraham Silberschatz, Henry F. Korth, and S. Sudarshan. 2020.Database system concepts(7 ed.). McGraw-Hill.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Javier Tuya, María José Suárez-Cabal, and Claudio de la Riva. 2007. Mutating database queries.Information and Software Technology49 (04 2007), 398–417. https://doi.org/10.1016/j.infsof.2006.06.009",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Margus Veanes, Jonathan de Halleux, Nikolai Tillmann, and Peli de Halleux. 2009.Qex: Symbolic SQL Query Explorer. Technical Report MSR-TR-2009-2015. https://www.microsoft.com/en-us/research/publication/qex-symbolic-sqlquery-explorer/ Updated January 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Margus Veanes, Nikolai Tillmann, and Jonathan de Halleux. 2010. Qex: Symbolic SQL Query Explorer. InLogic for Programming, Artificial Intelligence, and Reasoning, Edmund M. Clarke and Andrei Voronkov (Eds.). Springer Berlin Heidelberg, Berlin, Heidelberg, 425–446.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Qi Zhou, Joy Arulraj, Shamkant B. Navathe, William Harris, and Jinpeng Wu. 2022. SPES: A Symbolic Approach to Proving Query Equivalence Under Bag Semantics. InIEEE Intl. Conf. on Data Engineering, ICDE. IEEE, 2735–2748. https://doi.org/10.1109/ICDE53745.2022.00250 Received April 2025; revised July 2025; accepted August 2025 3, No. 6 (SIGMOD), Article 367. Publication date: December 2025.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 19,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020). 1140–1152. 3, No. 6 (SIGMOD), Article 367. Publication date: December 2025. 367:26 Sunanda Somwase, Parismita Das, and S. Sudarshan",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (Nov. 2020), 30 pages. https://doi.org/10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[19, 20]",
+      "technique": "norec",
+      "sentence": "[19, 20].",
+      "context_before": "includes (i) extensions to handle more SQL features such as LIMIT clause with ORDER BY, windowing functions, string and date manipulation functions that are common in data wrangling tasks, (ii) extensions to catch further types of mutations, and (iii) optimizing the generated constraints to handle larger queries and larger datasets. We also plan to extend our system to support checking of query equivalence. Another potential area for future work is to explore the use of data generation to improve the detection of bugs in database system implementations, following the approach of Rigger et al.",
+      "context_after": "Acknowledgements: We thank Chaitra Gurjar, Kartik Patel, and Kumaran Kartikeyan, for their contributions towards the implementation of some of the techniques described in this paper. We also thank all students who have contributed to the development of the XData system since the version described in [2], including: Bikash Chandra, Saurabh Chaturvedi, Ravi Shankar, Sai Balaji Polakampalli, Rahul Sharma, Pooja Gayakwad, and Deeksha Kasture. We also thank the referees for their detailed comments that have helped improve the presentation of the paper. 3, No. 6 (SIGMOD), Article 367. Publication",
+      "section": "9 Conclusion and Future Work",
+      "page": 24,
+      "char_offset": 80750,
+      "cited_reference": {
+        "number": 19,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020). 114",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "This paper generates sample data for testing SQL queries, a task with applications in testing analytics and application queries, grading student queries, and validating text-to-SQL output. Earlier work handled single-block and single-level nested queries but not more complex ones. The authors present an architecture aimed at complex queries, and also beat the state of the art on the narrower problem of demonstrating that two queries are not equivalent.",
+    "narrative": "A single citation in the conclusion, placing SQLancer's work among related database testing research.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3779212_3790244.json b/_data/papers/paper_doi_10_1145_3779212_3790244.json
new file mode 100644
index 0000000..e76b4cc
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3779212_3790244.json
@@ -0,0 +1,976 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:43Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3779212.3790244",
+    "title": "Understanding Query Optimization Bugs in Graph Database Systems",
+    "authors": [
+      "Yuyu Chen",
+      "Zhongxing Yu"
+    ],
+    "year": 2026,
+    "venue": "International Conference on Architectural Support for Programming Languages and Operating Systems",
+    "doi": "10.1145/3779212.3790244",
+    "arxiv_id": null,
+    "s2_paper_id": "fd1257a7345b8ddc8fe8ed0344f300577679daab",
+    "url": "https://doi.org/10.1145/3779212.3790244",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3779212.3790244",
+      "retrieved_at": "2026-09-09T01:29:43Z",
+      "chars": 101347,
+      "content_sha256": "sha256:819d7e1c5dbe69250327015322c1b9311cd99e14dd13132f1c9a0b95f258c927"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 18,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1669
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 6864
+      },
+      {
+        "number": "2.1",
+        "title": "Labeled Property Graph",
+        "start": 6877
+      },
+      {
+        "number": "2.2",
+        "title": "Cypher Graph Query Language",
+        "start": 8436
+      },
+      {
+        "number": "2.3",
+        "title": "Cypher Query Optimization",
+        "start": 11828
+      },
+      {
+        "number": "3",
+        "title": "Study Methodology",
+        "start": 15208
+      },
+      {
+        "number": "4",
+        "title": "Bug Cause Study",
+        "start": 19706
+      },
+      {
+        "number": "4.1",
+        "title": "Buggy query optimization and normalization",
+        "start": 20108
+      },
+      {
+        "number": "4.1.1",
+        "title": "Ignore specific legal query formats.",
+        "start": 20470
+      },
+      {
+        "number": "4.1.2",
+        "title": "Limited semantic analysis.",
+        "start": 22030
+      },
+      {
+        "number": "4.2",
+        "title": "Buggy query plan generation and selection",
+        "start": 23967
+      },
+      {
+        "number": "4.2.1",
+        "title": "Incomplete or incorrect query plan space.",
+        "start": 24155
+      },
+      {
+        "number": "4.2.2",
+        "title": "Inaccurate cost estimation.",
+        "start": 29957
+      },
+      {
+        "number": "4.2.3",
+        "title": "Defective plan space exploration algorithm.",
+        "start": 32139
+      },
+      {
+        "number": "4.3",
+        "title": "Buggy rule-based query plan transformation",
+        "start": 33848
+      },
+      {
+        "number": "4.3.1",
+        "title": "Incomplete precondition checking.",
+        "start": 34012
+      },
+      {
+        "number": "4.3.2",
+        "title": "Incorrect plan operator relation assumption.",
+        "start": 35020
+      },
+      {
+        "number": "4.4",
+        "title": "Discussion: Query optimization bugs in GDBMS",
+        "start": 37254
+      },
+      {
+        "number": "5.1",
+        "title": "Attributes of Bug-exposing LPGs",
+        "start": 39795
+      },
+      {
+        "number": "5.2",
+        "title": "Attributes of Bug-exposing Cypher Query",
+        "start": 41295
+      },
+      {
+        "number": "5.2.1",
+        "title": "Clause skeleton.",
+        "start": 42029
+      },
+      {
+        "number": "5.2.2",
+        "title": "Expression.",
+        "start": 44990
+      },
+      {
+        "number": "5.3",
+        "title": "Symptoms of Query Optimization Bugs",
+        "start": 47848
+      },
+      {
+        "number": "6.1",
+        "title": "Mistakes during bug fixing",
+        "start": 51517
+      },
+      {
+        "number": "6.2",
+        "title": "Fix Strategies",
+        "start": 53791
+      },
+      {
+        "number": "7.1",
+        "title": "Implementation",
+        "start": 56388
+      },
+      {
+        "number": "7.2",
+        "title": "Evaluation Setup",
+        "start": 60851
+      },
+      {
+        "number": "7.3",
+        "title": "Evaluation Results",
+        "start": 62820
+      },
+      {
+        "number": "7.4",
+        "title": "Comparison with State of the Art",
+        "start": 68902
+      },
+      {
+        "number": "7.4.1",
+        "title": "Study of Missed Bugs. We first show that our tool",
+        "start": 69122
+      },
+      {
+        "number": "7.5",
+        "title": "Discussion",
+        "start": 74852
+      },
+      {
+        "number": "7.5.1",
+        "title": "False Negative Study. To see how our tool performs",
+        "start": 74867
+      },
+      {
+        "number": "7.5.2",
+        "title": "Future Work.",
+        "start": 76213
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2025. cypher-for-gremlin. https://github.com/opencypher/cypher-forgremlin",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2025. DB-Engines Ranking of Graph DBMS. https://db-engines.com/ en/ranking/graph+ dbms",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2025. Graph Database Market-Forecast(2025 - 2031). https://www.industryarc.com/Research/Graph-Database-MarketResearch-500973",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2025.HugeGraph. https://hugegraph.apache.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2025.Kuzu. https://kuzudb.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2025.MemGraph. https://memgraph.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2025.Neo4j. https://neo4j.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2025. Neo4j Offical Documents: Clauses in the Cypher Query Language. https://neo4j.com/docs/cypher-manual/current/clauses/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2025. Neo4j Offical Documents: Operators in detail. https: //neo4j.com/docs/cypher-manual/current/planning-andtuning/operators/operators-detail/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2025.RedisGraph. https://dbdb.io/db/redisgraph",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In on Software Engineering (ICSE). 2060–2071. doi:10.1109/ICSE48619. 2023.00174",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 12,
+      "text": "Earl T. Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. 2015. The Oracle Problem in Software Testing: A Survey. IEEE Transactions on Software Engineering 41, 05 (May 2015), 507–525. doi:10.1109/TSE.2014.2372785",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Milind Chabbi and Murali Krishna Ramanathan. 2022. A study of real-world data races in Golang. In Proceedings of the 43rd ACMSIGPLAN International Conference on Programming Language Design and Implementation (San Diego, CA, USA) (PLDI 2022). Association for Computing Machinery, New York, NY, USA, 474–489. doi:10.1145/3519939.3523720",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Surajit Chaudhuri. 1998. An overview of query optimization in relational systems. In Proceedings of the Seventeenth ACMSIGACTSIGMOD-SIGART Symposium on Principles of Database Systems (Seattle, Washington, USA) (PODS ’98). Association for Computing Machinery, New York, NY, USA, 34–43. doi:10.1145/275487.275492",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Weidong Cui, Marcus Peinado, Sang Kil Cha, Yanick Fratantonio, and Vasileios P Kemerlis. 2016. Retracer: Triaging crashes by reverse execution from partial memory dumps. In Proceedings of the 38th International Conference on Software Engineering. 820–831.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Nurit Dor, Tal Lev-Ami, Shay Litvak, Mooly Sagiv, and Dror Weiss. 2008. Customization change impact analysis for erp professionals via program slicing. In Proceedings of the 2008 International Symposium on Software Testing and Analysis (Seattle, WA, USA) (ISSTA ’08). Association for Computing Machinery, New York, NY, USA, 97–108. doi:10.1145/1390630.1390644",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Nadime Francis, Alastair Green, Paolo Guagliardo, Leonid Libkin, Tobias Lindaaker, Victor Marsault, Stefan Plantikow, Mats Rydberg, Petra Selmer, and Andrés Taylor. 2018. Cypher: An Evolving Query Language for Property Graphs. In Proceedings of the 2018 International Conference on Management of Data (Houston, TX, USA) (SIGMOD ’18). Association for Computing Machinery, New York, NY, USA, 1433–1445.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting Bugs in Cypher Graph Database Engines. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis (Seattle, WA, USA) (ISSTA 2023). Association for Computing Machinery, New York, NY, USA, 163–174. doi:10.1145/3597926.3598046",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Yannis E. Ioannidis. 1996. Query optimization. ACM Comput. Surv. 28, 1 (March 1996), 121–123. doi:10.1145/234313.234367",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal) (ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 46, 1",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). USENIX Association, Boston, MA, 397– 417. https://www.usenix.org/conference/osdi23/presentation/jiang",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting Logic Bugs in Database Engines via Equivalent Expression Transformation. In 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24). USENIX Association, Santa Clara, CA, 821–835. https: //www.usenix.org/conference/osdi24/presentation/jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Guoliang Jin, Linhai Song, Xiaoming Shi, Joel Scherpelz, and Shan Lu. 2012. Understanding and detecting real-world performance bugs. SIGPLAN Not.47, 6 (June 2012), 77–88. doi:10.1145/2345156.2254075",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Siva Kesava Reddy Kakarla, Ryan Beckett, Todd Millstein, and George Varghese. 2022. SCALE: Automatically Finding RFC Compliance Bugs in DNS Nameservers. In 19th USENIX Symposium on Networked Systems Design and Implementation (NSDI 22). USENIX Association, Renton, WA, 307–323. https://www.usenix.org/conference/nsdi22/ presentation/kakarla",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "James C. King. 1976. Symbolic execution and program testing. Commun. ACM 19, 7 (July 1976), 385–394. doi:10.1145/360248.360252",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Takahiro Konno, Runhe Huang, Tao Ban, and Chuanhe Huang. 2017. Goods recommendation based on retail knowledge in a Neo4j graph database combined with an inference mechanism implemented in jess. In 2017 IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Computed, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation (Smar",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Claire Le Goues, Michael Pradel, and Abhik Roychoudhury. 2019. Automated program repair. Commun. ACM 62, 12 (Nov. 2019), 56–65. doi:10.1145/3318162",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Lantian Li, Yuyu Chen, Jingwen Wu, Yue Pan, and Zhongxing Yu. 2025. Understanding Inconsistent State Update Vulnerabilities in Smart Contracts. arXiv:2508.06192 [cs.SE] https://arxiv.org/abs/2508.06192",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Lantian Li, Yejian Liang, Zhihao Liu, and Zhongxing Yu. 2023. Understanding Solidity Event Logging Practices in the Wild. In Proceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (San Francisco, CA, USA) (ESEC/FSE 2023). Association for Computing Machinery, New York, NY, USA, 300–312. doi:10.1145/3611643.3616342",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering dbms bugs via configurationbased equivalent transformation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting logical bugs of {DBMS}with coverage-based guidance. In 31st USENIX Security Symposium (USENIX Security 22). 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Shuang Liu, Junhao Lan, Xiaoning Du, Jiyuan Li, Wei Lu, Jiajun Jiang, and Xiaoyong Du. 2024. Testing Graph Database Systems with GraphState Persistence Oracle. In Proceedings of the 33rd ACMSIGSOFT International Symposium on Software Testing and Analysis (Vienna, Austria) (ISSTA 2024). Association for Computing Machinery, New York, NY, USA, 666–677. doi:10.1145/3650212.3680311",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Shan Lu, Soyeon Park, Eunsoo Seo, and Yuanyuan Zhou. 2008. Learning from mistakes: a comprehensive study on real world concurrency bug characteristics. SIGARCH Comput. Archit. News 36, 1 (March 2008), 329–339. doi:10.1145/1353534.1346323 2174 Understanding Query Optimization Bugs in Graph Database Systems ASPLOS ’26, March 22–26, 2026, Pittsburgh, PA, USA",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Bingqing Lyu, Xiaoli Zhou, Longbin Lai, Yufan Yang, Yunkai Lou, Wenyuan Yu, Ying Zhang, and Jingren Zhou. 2025. A Modular GraphNative Query Optimization Framework. In Companion of the 2025 International Conference on Management of Data (Berlin, Germany) (SIGMOD/PODS ’25). Association for Computing Machinery, New York, NY, USA, 566–579. doi:10.1145/3722212.3724425",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting. doi:10.1145/3725300",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 36,
+      "text": "Qiuyang Mang, Aoyang Fang, Boxi Yu, Hanfei Chen, and Pinjia He. 2024. Testing Graph Database Systems via Equivalent Query Rewriting. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal) (ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 143, 12 pages. doi:10.1145/3597503.3639200",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Christopher Manning, Mihai Surdeanu, John Bauer, Jenny Finkel, Steven Bethard, and David McClosky. 2014. The Stanford CoreNLP Natural Language Processing Toolkit. In Proceedings of 52nd Annual Meeting of the Association for Computational Linguistics: System Demonstrations, Kalina Bontcheva and Jingbo Zhu (Eds.). Association for Computational Linguistics, Baltimore, Maryland, 55–60. doi:10.3115/v1/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Ryan Marcus, Parimarjan Negi, Hongzi Mao, Nesime Tatbul, Mohammad Alizadeh, and Tim Kraska. 2021. Bao: Making Learned Query Optimization Practical. In Proceedings of the 2021 International Conference on Management of Data (Virtual Event, China) (SIGMOD ’21). Association for Computing Machinery, New York, NY, USA, 1275–1288. doi:10.1145/3448016.3452838",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Ryan Marcus, Parimarjan Negi, Hongzi Mao, Chi Zhang, Mohammad Alizadeh, Tim Kraska, Olga Papaemmanouil, and Nesime Tatbul. 2019. Neo: a learned query optimizer. Proc. VLDB Endow. 12, 11 (July 2019), 1705–1718. doi:10.14778/3342263.3342644",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Manasij Mukherjee and John Regehr. 2024. Hydra: Generalizing Peephole Optimizations with Program Synthesis. Proc. ACM Program. Lang. 8, OOPSLA1, Article 120 (April 2024), 29 pages. doi:10.1145/3649837",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Phitchaya Mangpo Phothilimthana, Aditya Thakur, Rastislav Bodik, and Dinakar Dhurjati. 2016. Scaling up Superoptimization. In Proceedings of the Twenty-First International Conference on Architectural Support for Programming Languages and Operating Systems (Atlanta, Georgia, USA) (ASPLOS ’16). Association for Computing Machinery, New York, NY, USA, 297–310. doi:10.1145/2872362.2872387",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Andy Podgurski, David Leon, Patrick Francis, Wes Masri, Melinda Minch, Jiayang Sun, and Bin Wang. 2003. Automated support for classifying software failure reports. In Proceedings of the 25th International Conference on Software Engineering (Portland, Oregon) (ICSE ’03). IEEE Computer Society, USA, 465–475.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Boqin Qin, Yilun Chen, Zeming Yu, Linhai Song, and Yiying Zhang. 2020. Understanding memory and thread safety practices and issues in real-world Rust programs. In Proceedings of the 41st ACMSIGPLAN Conference on Programming Language Design and Implementation (London, UK) (PLDI 2020). Association for Computing Machinery, New York, NY, USA, 763–779. doi:10.1145/3385412.3386036",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Yuxiang Ren, Hao Zhu, Jiawei Zhang, Peng Dai, and Liefeng Bo. 2021. EnsemFDet: An Ensemble Approach to Fraud Detection based on Bipartite Graph. In Engineering (ICDE). 2039–2044. doi:10.1109/ICDE51399.2021.00197",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA) (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/ 33",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 46,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). USENIX Association, USA, Article 38, 16 pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 47,
+      "text": "Marko A. Rodriguez. 2015. The Gremlin graph traversal machine and language (invited talk). In Proceedings of the 15th Symposium on Database Programming Languages (Pittsburgh, PA, USA) (DBPL 2015). Association for Computing Machinery, New York, NY, USA, 1–10. doi:10.1145/2815072.2815073",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Sudipta Sen, Akash Mehta, Runa Ganguli, and Soumya Sen. 2021. Recommendation of Influenced Products Using Association Rule Mining: Neo4j as a Case Study. SN Comput. Sci. 2, 2 (Feb. 2021), 17 pages. doi:10.1007/s42979-021-00460-8",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Utsav Sethi, Haochen Pan, Shan Lu, Madanlal Musuvathi, and Suman Nath. 2022. Cancellation in Systems: An Empirical Study of Task Cancellation Patterns and Failures. In 16th USENIX Symposium on Operating Systems Design and Implementation (OSDI 22). USENIX Association, Carlsbad, CA, 127–141. https://www.usenix.org/conference/osdi22/ presentation/sethi",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Chandan Sharma, Pierre Genevès, Nils Gesbert, and Nabil Layaïda. 2025. Schema-Based Query Optimisation for Graph Databases. 2025), 29 pages. doi:10.1145/ 3709722",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Lin Tan, Ding Yuan, Gopal Krishna, and Yuanyuan Zhou. 2007. /*icomment: bugs or bad comments?*/. In Proceedings of Twenty-First ACMSIGOPS Symposium on Operating Systems Principles (Stevenson, Washington, USA) (SOSP ’07). Association for Computing Machinery, New York, NY, USA, 145–158. doi:10.1145/1294261.1294276",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting logic bugs of join optimizations in dbms. Proceedings of the ACM on Management of Data 1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Tengfei Tu, Xiaoyu Liu, Linhai Song, and Yiying Zhang. 2019. Understanding Real-World Concurrency Bugs in Go. In Proceedings of the Twenty-Fourth International Conference on Architectural Support for Programming Languages and Operating Systems (Providence, RI, USA) (ASPLOS ’19). Association for Computing Machinery, New York, NY, USA, 865–878. doi:10.1145/3297858.3304069",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Joseph Tucek, Shan Lu, Chengdu Huang, Spiros Xanthos, and Yuanyuan Zhou. 2007. Triage: diagnosing production run failures at the user’s site. In Proceedings of Twenty-First ACMSIGOPS Symposium on Operating Systems Principles (Stevenson, Washington, USA) (SOSP ’07). Association for Computing Machinery, New York, NY, USA, 131–144. doi:10.1145/1294261.1294275",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Simon Urli, Zhongxing Yu, Lionel Seinturier, and Martin Monperrus. 2018. How to design a program repair bot? insights from the repairnator project. In Proceedings of the 40th International Conference on Software Engineering: Software Engineering in Practice (Gothenburg, Sweden) (ICSE-SEIP ’18). Association for Computing Machinery, New York, NY, USA, 95–104. doi:10.1145/3183519.3183540",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "M. Utting and B. Legeard. 2010. Practical Model-Based Testing: A Tools Approach. Morgan Kaufmann. https://books.google.com.hk/books? id=8hAGtY4-oOoC",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Jiayi Wang, Chengliang Chai, Jiabin Liu, and Guoliang Li. 2021. FACE: a normalizing flow based cardinality estimator. Proc. VLDB Endow. 15, 1 (Sept. 2021), 72–84. doi:10.14778/3485450.3485458",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Chunqiu Steven Xia, Matteo Paltenghi, Jia Le Tian, Michael Pradel, and Lingming Zhang. 2024. Fuzz4ALL: Universal Fuzzing with Large Language Models. In on Software Engineering (ICSE). IEEE Computer Society, Los Alamitos, CA, USA, 1547–1559. https://doi.ieeecomputersociety.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Fabian Yamaguchi, Nico Golde, Daniel Arp, and Konrad Rieck. 2014. Modeling and Discovering Vulnerabilities with Code Property Graphs. In. 590–604. doi:10.1109/ SP.2014.44 2175 ASPLOS ’26, March 22–26, 2026, Pittsburgh, PA, USA Yuyu Chen and Zhongxing Yu",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation (San Jose, California, USA) (PLDI ’11). Association for Computing Machinery, New York, NY, USA, 283–294. doi:10.1145/1993498.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Zongheng Yang, Eric Liang, Amog Kamsetty, Chenggang Wu, Yan Duan, Xi Chen, Pieter Abbeel, Joseph M. Hellerstein, Sanjay Krishnan, and Ion Stoica. 2019. Deep unsupervised cardinality estimation. Proc. VLDB Endow. 13, 3 (Nov. 2019), 279–292. doi:10.14778/3368289.3368294",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Zijing Yin, Si Liu, and David Basin. 2025. Testing Graph Databases with Synthesized Queries. 2025), 26 pages. doi:10.1145/3749186",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Zuoning Yin, Xiao Ma, Jing Zheng, Yuanyuan Zhou, Lakshmi N. Bairavasundaram, and Shankar Pasupathy. 2011. An empirical study on configuration errors in commercial and open source systems. In Proceedings of the Twenty-Third ACM Symposium on Operating Systems Principles (Cascais, Portugal) (SOSP ’11). Association for Computing Machinery, New York, NY, USA, 159–172. doi:10.1145/2043556.2043572",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Zuoning Yin, Ding Yuan, Yuanyuan Zhou, Shankar Pasupathy, and Lakshmi Bairavasundaram. 2011. How do fixes become bugs?. In Proceedings of the 19th ACMSIGSOFT Symposium and the 13th European Conference on Foundations of Software Engineering (Szeged, Hungary) (ESEC/FSE ’11). Association for Computing Machinery, New York, NY, USA, 26–36. doi:10.1145/2025113.2025121",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Xiang Yu, Chengliang Chai, Guoliang Li, and Jiabin Liu. 2022. CostBased or Learning-Based? A Hybrid Query Optimizer for Query Plan Selection. Proc. VLDB Endow. 15, 13 (Sept. 2022), 3924–3936. doi:10. 14778/3565838.3565846",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Zhongxing Yu, Chenggang Bai, Lionel Seinturier, and Martin Monperrus. 2021. Characterizing the Usage, Evolution and Impact of Java Annotations in Practice. IEEE Transactions on Software Engineering 47, 5 (2021), 969–986. doi:10.1109/TSE.2019.2910516",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Zhongxing Yu, Matias Martinez, Zimin Chen, Tegawendé F. Bissyandé, and Martin Monperrus. 2023. Learning the Relation Between Code Features and Code Transforms With Structured Prediction. IEEE Transactions on Software Engineering 49, 7 (2023), 3872–3900. doi:10.1109/TSE.2023.3275380",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Ding Yuan, Yu Luo, Xin Zhuang, Guilherme Renna Rodrigues, Xu Zhao, Yongle Zhang, Pranay U. Jain, and Michael Stumm. 2014. Simple testing can prevent most critical failures: an analysis of production failures in distributed data-intensive systems. In Proceedings of the 11th USENIX Conference on Operating Systems Design and Implementation (Broomfield, CO) (OSDI’14). USENIX Association, USA, 249–265.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Ding Yuan, Soyeon Park, and Yuanyuan Zhou. 2012. Characterizing logging practices in open-source software. In 2012 34th International Conference on Software Engineering (ICSE). 102–112. doi:10.1109/ICSE. 2012.6227202",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Andreas Zeller and Ralf Hildebrandt. 2002. Simplifying and Isolating Failure-Inducing Input. IEEE Trans. Softw. Eng. 28, 2 (Feb. 2002), 183–200. doi:10.1109/32.988498",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Yongle Zhang, Junwen Yang, Zhuqi Jin, Utsav Sethi, Kirk Rodrigues, Shan Lu, and Ding Yuan. 2021. Understanding and Detecting Software Upgrade Failures in Distributed Systems. In Proceedings of the ACMSIGOPS 28th Symposium on Operating Systems Principles (Virtual Event, Germany) (SOSP ’21). Association for Computing Machinery, New York, NY, USA, 116–131. doi:10.1145/3477132.3483577",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding bugs in Gremlin-based graph database systems via Randomized differential testing. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis (Virtual, South Korea) (ISSTA 2022). Association for Computing Machinery, New York, NY, USA, 302–313. ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Xiaogang Zhu, Sheng Wen, Seyit Camtepe, and Yang Xiang. 2022. Fuzzing: A Survey for Roadmap. ACM Comput. Surv. 54, 11s, Article 230 (Sept. 2022), 36 pages. doi:10.1145/3512345",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Zeyang Zhuang, Penghui Li, Pingchuan Ma, Wei Meng, and Shuai Wang. 2023. Testing Graph Database Systems via Graph-Aware Metamorphic Relations. Proc. VLDB Endow. 17, 4 (Dec. 2023), 836–848. doi:10.14778/3636218.3636236 2176",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 11,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In on Software Engineering (ICSE). 2060–2071. doi:10.1109/ICSE48619. 2023.00174",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "prints the DOI of the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal) (ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 46, 12 pages. doi:10.1145/3597503.3623307",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 21,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). USENIX Association, Boston, MA, 397– 417. https://www.usenix.org/conference/osdi23/presentation/jiang",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 35,
+      "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting. doi:10.1145/3725300",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA) (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/ 3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 46,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). USENIX Association, USA, Article 38, 16 pages.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[18,32,35,36]",
+      "technique": null,
+      "sentence": "Indeed, recent years have witnessed a surge of research efforts for testing GDBMSs [18,32,35,36].",
+      "context_before": "ate efficient storing and querying graph data that consists of vertices and edges. Hence, recent years have witnessed an ever-growing usage of GDBMSs in various datadriven applications, such as recommendation systems [ 26, 48], fraud detection [ 44], and program analysis [ 59]. As stated by the industry report [ 3], the market of graph database is forecast to reach USD 88.9 billion by 2030. For instance, Neo4j [7], the most popular and widely adopted GDBMSs, has been adopted by over 1,000 enterprise customers. GDBMSs are complex software systems like compilers, and thus are not immune to bugs.",
+      "context_after": "and numerous bugs in industrial GDBMSs have been reported. Existing techniques focus on integrating test oracles to identify bugs, and can be divided into two main strands: differential testing (DT) and metamorphic testing (MT). For multiple implementations of a single specification, DT provides the same valid input to these implementations and observes differences in their executions to detect bugs. GDSmith [ 18] and Grand [ 72] are the earliest GDBMS testing techniques that are built on top of DT. MT detects bugs by first identifying relationships between multiple inputs and their correspon",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2684,
+      "cited_reference": {
+        "number": 35,
+        "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting. doi:10.1145/3725300",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 35]",
+      "technique": null,
+      "sentence": "GRev [36], GraspDB [ 32], and Gslicer [ 35] are the representative GDBMS testing techniques that are built on top of MT.",
+      "context_before": "ting test oracles to identify bugs, and can be divided into two main strands: differential testing (DT) and metamorphic testing (MT). For multiple implementations of a single specification, DT provides the same valid input to these implementations and observes differences in their executions to detect bugs. GDSmith [ 18] and Grand [ 72] are the earliest GDBMS testing techniques that are built on top of DT. MT detects bugs by first identifying relationships between multiple inputs and their corresponding outputs, and then checking whether the relationships are preserved for software executions.",
+      "context_after": "These existing techniques target all different kinds of bugs that arise in all stages of the GDBMS workflow. Although offering the potential advantage of comprehensive bug detection, they may lose coverage for detecting a specific kind of bug. Various recent studies have shown that dedicated techniques for specific types of bugs are surprisingly effective, e.g., techniques for detecting task cancellation-related bugs [ 49], update bugs [ 71], performance bugs [ 23], configuration bugs [ 63], and Go data races [ 13]. These dedicated techniques make use of the specific characteristics of the bu",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3478,
+      "cited_reference": {
+        "number": 35,
+        "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting. doi:10.1145/3725300",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 11,21,45,46]",
+      "technique": "qpg",
+      "sentence": "For optimization bugs due to inaccurate cost estimations or defective plan space exploration algorithms, previous works on DBMS testing [ 11,21,45,46] also frequently report them.",
+      "context_before": "le flow control (e.g., Cypher requires explicit declaration of variables imported from the outer scope into the subquery). While this again offers strong expressive power, it also introduces more complex scope semantics and execution models compared to SQL. Due to the significant differences in data storage formats and query languages for DBMSs and GDBMSs, the optimization bugs should have certain differences. Indeed, our findings 2 and 3 show that between one third and one half of the query optimization bugs for GDBMSs arise due to the complexities of graph representation and Cypher language.",
+      "context_after": "Thus, the query optimization bugs in GDBMSs and DBMSs also share similarities.5 Bug Manifestation Study This section discusses the findings about how the optimization bugs can be manifested, and these findings can guide the design of testing techniques. Indeed, our optimization bug exposing tool in Section 7is guided by the findings here. 5.1 Attributes of Bug-exposing LPGs A unique challenge faced by testing GDBMSs is that the test input consists of two parts: the LPG for storing the data and the queries for querying the graph. Thus, the input space is extremely large, particularly given tha",
+      "section": "4.4 Discussion: Query optimization bugs in GDBMS",
+      "page": 8,
+      "char_offset": 39273,
+      "cited_reference": {
+        "number": 11,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. In on Software Engineering (ICSE). 2060–2071. doi:10.1109/ICSE48619. 2023.00174",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "prints the DOI of the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[35]",
+      "technique": null,
+      "sentence": "However, most existing GDBMS testing techniques do not support the subquery clause kind at all, including GDSmith [ 18], GAMERA [ 75], GraspDB [ 32], GRev [ 36], Gslicer [35], and GQS [ 62]).",
+      "context_before": "clause is used by 32% of bug-exposing queries. More specifically, among those queries that use the subquery clause kind, some of them use subquery clauses in a nested way. For instance, Figure 9agives the bug-exposing query for the issue Neo4j-12760 and the query uses the CALL{} subquery in a nested way. For subqueries, the optimizer must balance both the efficiency of graph pattern matching and subquery dependencies, thus it is no surprising that issues arise easily. In particular, note that Cypher employs explicit scope partitioning and variable flow control, further exacerbating the issue.",
+      "context_after": "Although GraphGenie [ 20] supports the kind of subquery clause, it only supports the CALL{} clause. Thus, in principle, these techniques cannot generate queries with diverse subquery clauses, making them ineffective for exposing query optimization bugs. Finding 7: Around 32% of the bug-exposing queries for query optimization bugs use subquery clauses, and some of them use subquery clauses in a nested way. As existing GDBMS testing techniques offer no or limited capability for generating subquery clauses (particularly nested subquery clauses), future testing tools for exposing query optimizati",
+      "section": "5.2.1 Clause skeleton.",
+      "page": 9,
+      "char_offset": 44147,
+      "cited_reference": {
+        "number": 35,
+        "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting. doi:10.1145/3725300",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 20]",
+      "technique": null,
+      "sentence": "Although GraphGenie [ 20] supports the kind of subquery clause, it only supports the CALL{} clause.",
+      "context_before": "e 9agives the bug-exposing query for the issue Neo4j-12760 and the query uses the CALL{} subquery in a nested way. For subqueries, the optimizer must balance both the efficiency of graph pattern matching and subquery dependencies, thus it is no surprising that issues arise easily. In particular, note that Cypher employs explicit scope partitioning and variable flow control, further exacerbating the issue. However, most existing GDBMS testing techniques do not support the subquery clause kind at all, including GDSmith [ 18], GAMERA [ 75], GraspDB [ 32], GRev [ 36], Gslicer [35], and GQS [ 62]).",
+      "context_after": "Thus, in principle, these techniques cannot generate queries with diverse subquery clauses, making them ineffective for exposing query optimization bugs. Finding 7: Around 32% of the bug-exposing queries for query optimization bugs use subquery clauses, and some of them use subquery clauses in a nested way. As existing GDBMS testing techniques offer no or limited capability for generating subquery clauses (particularly nested subquery clauses), future testing tools for exposing query optimization bugs should be customized with this capability. 5.2.2 Expression. For the studied 102 optimizatio",
+      "section": "5.2.1 Clause skeleton.",
+      "page": 9,
+      "char_offset": 44339,
+      "cited_reference": {
+        "number": 20,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering ",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[ 21,45,46]",
+      "technique": "norec",
+      "sentence": "Like the typical methodology used to evaluate the effectiveness of database testing methods [ 21,45,46], the testing process is intermittent and we continuously improved the tool during the process.",
+      "context_before": "uzu is not evaluated as it has rarely been tested by existing tools. We evaluated the tool on the two GDBMSs with the latest available release versions. During our testing period, if a new version of the two tested GDBMSs is released, we set up the tool to test the updated versions. More specifically, we test Neo4j starting from version 5.25.1 and Memgraph starting from versionTable 5. Numbers of bugs found by the tool and their status (the column “Opt bugs” denotes the optimization bugs). GDBMS # Found # Confirmed # Fixed # Opt bugs Ne o4j 12 12 10 6 Memgraph 8 8 1 4 T otal 20 20 11 10 3.0.0.",
+      "context_after": "The evaluation was conducted on Ubuntu 22.04 with a 64-core Intel Xeon Gold 6226R processor running at 2.09GHz and 256GB of RAM. Overall, we generate 456.3k and 986.5k test inputs for Neo4j and Memgraph respectively, and the average test execution time is 513.3ms and 28.5ms for Neo4j and Memgraph respectively. For the Cypher query part of the test input, the average number of clauses per query is 23.1 and 22.6 for Neo4j and Memgraph respectively. For the LPG part of the test input, the average number of relationships per graph is 334.2 and 302.6 for Neo4j and Memgraph respectively. After dete",
+      "section": "7.2 Evaluation Setup",
+      "page": 12,
+      "char_offset": 61732,
+      "cited_reference": {
+        "number": 45,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA)",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 20]",
+      "technique": null,
+      "sentence": "4 Comparison with State of the Art We compare our tool with 7 state-of-the-art GDBMS testing techniques, including GDSmith [ 18], GAMERA [ 75], GraspDB [ 32], GRev [ 36], GraphGenie [ 20], GSlicer [ 35], and GQS [62].",
+      "context_before": "ed attribute to remain logically bound to the original variable n1. Afterward, the aggregation operation triggers a rebuild of the runtime variable layout. At this point, the planner attempts to access the cached attribute using the original variable name n1, but n1has become invalid in this context due to being renamed to alias0. The fix modifies the optimization logic of property caching, so that the planner can correctly associate cached properties with valid variables in the current scope when the scope changes, avoiding the generation of query operators that reference unusable symbols. 7.",
+      "context_after": "7.4.1 Study of Missed Bugs. We first show that our tool can find bugs missed by existing techniques. To achieve this, for each of the 20 detected bugs, we check whether the bug exists in the version(s) tested by existing techniques. If so, we deem that the existing techniques have missed the bug. This method is reasonable as all the 20 confirmed bugs are not marked as duplicates by developers and all existing techniques have conducted an extensive evaluation. Table 6shows the results. It can be seen that all existing techniques miss some bugs detected by our tool, and most of the missed bugs",
+      "section": "7.4 Comparison with State of the Art",
+      "page": 13,
+      "char_offset": 68904,
+      "cited_reference": {
+        "number": 20,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering ",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[ 35,46,62]",
+      "technique": "pqs",
+      "sentence": "Note that 24-Hour empirical testing is widely used by existing works to compare the effectiveness of different testing techniques [ 35,46,62].",
+      "context_before": "4.12 2.4.0 0 (0) 2 (2) 2 (2) GraphGenie 4.1.12–5.9.0 – 5 (2) – 5 (2) GAMERA 5.4.0 – 1 (1) – 1 (1) GRev 5.6.0 2.8.0 5 (2) 4 (4) 9 (6) GraspDB 5.6.0–5.12.0 2.7.0–2.11.0 5 (2) 4 (4) 9 (6) GSlicer 5.10 – 5 (2) – 5 (2) GQS5.18, 5.20, 5.21.22.13, 2.14.1, 2.15, 2.177 (4) 5 (4) 12 (8) 7.4.2 24-Hour Empirical Testing. To empirically compare our tool with other techniques, we run our tool and existing techniques against Neo4j 5.25.1 and Memgraph 3.0.0, which are their latest versions at the start of the tool evaluation. Each evaluation lasts for 24 hours, and the results are shown in Table 7and Table 8.",
+      "context_after": "For the differential testing technique GDSmith, we compare the results between Neo4j and Memgraph, and also use crashes and internal errors as oracles. Table 7. The results of 24-hour empirical testing on Neo4j. The number in the parentheses denotes the number of false reports. Compar ed Coverage Bug Detection T echniques Line Branch # Bug Reports # Unique Bugs GDSmith 18% 10% 465 (453) 1 GraphGenie 19% 10% 3 (3) 0 GAMERA 17% 9% 947 (837) 1 GRev 19% 11% 12 (12) 0 GraspDB 22% 13% 370 (370) 0 GSlicer 18% 10% 4 (4) 0 GQS 19% 11% 48 (46) 1 Our Tool 21% 13% 79 (2) 5 Table 8. The results of 24-hour",
+      "section": "7.4.1 Study of Missed Bugs. We first show that our tool",
+      "page": 14,
+      "char_offset": 71273,
+      "cited_reference": {
+        "number": 46,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). USENIX Association, USA, Article 38, 16 pages.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[32,35]",
+      "technique": null,
+      "sentence": "For the other metamorphic testing techniques, the false alarms are mainly due to the inaccurate transformation rules as also reported by previous studies [32,35], particularly due to the ignorance of constraints on graph entities and the inadequate consideration of special error cases (e.",
+      "context_before": "rage, 43 bug reports need to be analyzed in order to identify a unique bug for our tool. This number is lower than that of other techniques except GQS (whose number is 39), suggesting that our tool can detect bugs with low duplication rate and thus requires few manual efforts for bug report analysis. Third, all compared techniques (including our tool) suffer from false alarms, but our tool has the lowest false alarm rate. For the differential testing technique GDSmith, the main reason for false alarms is that Neo4j and Memgraph have different implementation choices for certain Cypher features.",
+      "context_after": "g., division-by-zero, exceeding the type’s value range, and incompatible type conversion. ). The observed false alarms for our tool are due to the violation of constraint dependencies for certain queries that involve the DELETE clause. Overall, our tool has a lower false alarm rate than other techniques and thus save time in manual checking of the bug reports. Finally, besides one bug detected by both our tool and GraspDB, the other 7 bugs detected by our tool are distinct from that detected by other techniques. This suggests that our tool and the compared techniques are complementary in impro",
+      "section": "7.4.1 Study of Missed Bugs. We first show that our tool",
+      "page": 14,
+      "char_offset": 73935,
+      "cited_reference": {
+        "number": 35,
+        "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting. doi:10.1145/3725300",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 18,35,36,75]",
+      "technique": null,
+      "sentence": "In particular, significant recent efforts have been devoted to testing GDBMSs [ 18,35,36,75].",
+      "context_before": "suffer from various issues issues such as long training time and high integration cost [ 38]. As for GDBMSs, Sharma et al. [50] propose a schema-based query optimization technique which leverages structural information defined in a graph schema to enhance query execution efficiency. Lyu et al. [34] design a modular graph-native query optimization framework, which addresses the challenges of optimizing complex graph pattern queries that combine pattern matching with relational operations. GDBMS Testing. Recent years have witnessed a surge of research efforts for testing DBMSs [ 22,30,31,52,73].",
+      "context_after": "Overall, existing testing efforts for GDBMSs focus on addressing the oracle issue and can bedivided into two main strands: differential testing (DT) and metamorphic testing (MT). GDsmith [ 18] and Grand [ 72] are techniques for Cypher-based and Gremlin-based GDBMSs respectively, and they both are based on DT. GRev [ 36], Gamera [ 75], GraspDB [ 32], and Gslicer [ 35] are representative techniques built on top of MT. The test input generation problem has been rarely touched by existing GDBMS testing efforts, and two representative techniques GDsmith [ 18] and GQS [ 62] involve using predefined",
+      "section": "7.5.2 Future Work.",
+      "page": 15,
+      "char_offset": 78395,
+      "cited_reference": {
+        "number": 35,
+        "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting. doi:10.1145/3725300",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 35]",
+      "technique": null,
+      "sentence": "GRev [ 36], Gamera [ 75], GraspDB [ 32], and Gslicer [ 35] are representative techniques built on top of MT.",
+      "context_before": "complex graph pattern queries that combine pattern matching with relational operations. GDBMS Testing. Recent years have witnessed a surge of research efforts for testing DBMSs [ 22,30,31,52,73]. In particular, significant recent efforts have been devoted to testing GDBMSs [ 18,35,36,75]. Overall, existing testing efforts for GDBMSs focus on addressing the oracle issue and can bedivided into two main strands: differential testing (DT) and metamorphic testing (MT). GDsmith [ 18] and Grand [ 72] are techniques for Cypher-based and Gremlin-based GDBMSs respectively, and they both are based on DT.",
+      "context_after": "The test input generation problem has been rarely touched by existing GDBMS testing efforts, and two representative techniques GDsmith [ 18] and GQS [ 62] involve using predefined rules and program synthesis to generate test inputs respectively. There exist numerous test input generation mechanisms in the literature, including fuzzing [ 74], symbolic execution [ 25], and model-based generation [ 56]. To improve the effectiveness, different mechanisms can further be combined. For instance, to find implementation errors in network protocol, SCALE builds an executable model of the protocol and t",
+      "section": "7.5.2 Future Work.",
+      "page": 15,
+      "char_offset": 78800,
+      "cited_reference": {
+        "number": 35,
+        "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting. doi:10.1145/3725300",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:26:20Z",
+    "is_model_written": true,
+    "summary": "A characteristic study of query optimization bugs in graph database systems, covering root causes, how the bugs manifest and how they are fixed, from which the authors draw ten findings. Building on the study they developed a testing tool aimed specifically at graph query optimization, which found 20 unique bugs, 10 of them optimization bugs.",
+    "narrative": "The comparison here is against the graph-database descendants of SQLancer's techniques rather than SQLancer itself: GDsmith, GAMERA, GRev, GraspDB, GraphGenie and Gslicer are the seven state-of-the-art techniques the tool is measured against, several of them built on metamorphic testing and several co-authored by SQLancer's authors. SQLancer's own work is cited for the cost-estimation class of optimization bug and for the 24-hour evaluation methodology.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "background",
+      "M6": "background",
+      "M7": "baseline",
+      "M8": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The seven tools compared against are graph-database testing techniques, some derived from SQLancer's oracles, but no mention read shows SQLancer itself being run."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3785021_3787993.json b/_data/papers/paper_doi_10_1145_3785021_3787993.json
new file mode 100644
index 0000000..3670af7
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3785021_3787993.json
@@ -0,0 +1,431 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:25:01Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3785021.3787993",
+    "title": "Reproducibility Report for ACM SIGMOD 2025 Paper: 'Constant Optimization Driven Database System Testing'",
+    "authors": [
+      "Yuvaraj Chesetti",
+      "Chi Zhang"
+    ],
+    "year": 2025,
+    "venue": null,
+    "doi": "10.1145/3785021.3787993",
+    "arxiv_id": null,
+    "s2_paper_id": null,
+    "url": "https://doi.org/10.1145/3785021.3787993",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3785021.3787993",
+      "retrieved_at": "2026-09-08T15:25:01Z",
+      "chars": 6439,
+      "content_sha256": "sha256:c26385cb3deded449a207cf63f2c5ea1f9148c0b2cb11489d02f5a1939b0ed84"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 2,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1218
+      },
+      {
+        "number": "4.1",
+        "title": "Artifacts and Experiments",
+        "start": 3455
+      },
+      {
+        "number": "4.2",
+        "title": "Reproducibility Process",
+        "start": 3983
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "[n. d.]. https://github.com/sqlancer/sqlancer.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 2,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages. doi:10.1145/3709674 58",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 1,
+      "text": "[n. d.]. https://github.com/sqlancer/sqlancer.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 2,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages. doi:10.1145/3709674 58",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "prints the DOI of the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "Constant Optimization Driven Database System Testing",
+      "technique": "coddtest",
+      "sentence": "Reproducibility Report for ACMSIGMOD 2025 Paper: “Constant Optimization Driven Database System Testing” Yuvaraj Chesetti chesetti.",
+      "context_before": "",
+      "context_after": "y@northeastern.edu Northeastern University Boston, USAChi Zhang zhangchi_seg@smail.nju.edu.cn Nanjing University Nanjing, China Abstract The authors proposed a new automated methodology for finding bugs in databases and provided scripts to run their bug-finding method against production databases. They also provided a list of bugs found through their approach which are currently filed in the respective database’s bug tracking system. We were able to successfully run the scripts and verify the bug artifacts provided by the authors. CCS Concepts •Information systems →Database query processing; •",
+      "section": null,
+      "page": 1,
+      "char_offset": 0,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "Constant Optimization Driven Database System Testing",
+      "technique": "coddtest",
+      "sentence": "Reproducibility Report for ACMSIGMOD 2025 Paper: “Constant Optimization Driven Database System Testing”.",
+      "context_before": "automated methodology for finding bugs in databases and provided scripts to run their bug-finding method against production databases. They also provided a list of bugs found through their approach which are currently filed in the respective database’s bug tracking system. We were able to successfully run the scripts and verify the bug artifacts provided by the authors. CCS Concepts •Information systems →Database query processing; •Software and its engineering →Software defect analysis. Keywords DBMSs testing; logic bugs; test oracle ACM Reference Format: Yuvaraj Chesetti and Chi Zhang. 2025.",
+      "context_after": "InReproducibility Reports of the 2025 International Conference on Management of Data (SIGMOD ARI Reports ’25), June 22–27, 2025, Berlin, Germany. ACM, New York, NY, USA, 2 pages. https://doi.org/10.1145/3785021.3787993 1 Introduction The original paper presented a new methodology for automated bug finding in production databases [ 2]. The authors provided scripts to reproduce the bug finding process that implemented their method. The scripts automatically set up a database and generate SQL queries over a long duration. Database instances and queries that crash the database are logged for repr",
+      "section": null,
+      "page": 1,
+      "char_offset": 893,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 2]",
+      "technique": "coddtest",
+      "sentence": "3787993 1 Introduction The original paper presented a new methodology for automated bug finding in production databases [ 2].",
+      "context_before": "ts and verify the bug artifacts provided by the authors. CCS Concepts •Information systems →Database query processing; •Software and its engineering →Software defect analysis. Keywords DBMSs testing; logic bugs; test oracle ACM Reference Format: Yuvaraj Chesetti and Chi Zhang. 2025. Reproducibility Report for ACMSIGMOD 2025 Paper: “Constant Optimization Driven Database System Testing”. InReproducibility Reports of the 2025 International Conference on Management of Data (SIGMOD ARI Reports ’25), June 22–27, 2025, Berlin, Germany. ACM, New York, NY, USA, 2 pages. https://doi.org/10.1145/3785021.",
+      "context_after": "The authors provided scripts to reproduce the bug finding process that implemented their method. The scripts automatically set up a database and generate SQL queries over a long duration. Database instances and queries that crash the database are logged for reproducibility and inspected to determine if they are valid bugs. The bug-finding scripts are also used to report efficiency metrics of their method. Additionally, the authors also provided a list of reported bugs found using their method while working on the paper. The original paper ran a large-scale evaluation across multiple databases",
+      "section": null,
+      "page": 1,
+      "char_offset": 1210,
+      "cited_reference": {
+        "number": 2,
+        "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages. doi:10.1145/3709674 58",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing coddtest",
+          "prints the DOI of the paper introducing coddtest",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "•The tool works by integrating their method with SQLancer [ 1], a tool to automatically test databases.",
+      "context_before": "cs similar to those reported by the paper. Further, the bugs reported by the authors were validated by database developers and are being worked on. This work is licensed under a Creative Commons Attribution 4.0 International License. SIGMOD ARI Reports ’25, Berlin, Germany ©2025 Copyright held by the owner/author(s). ACMISBN 979-8-4007-2375-9/2025/06 https://doi.org/10.1145/3785021.37879932 Submission The authors shared their code as a fig archive. •Fig archive: figshare.com/s/f601c272cebc253bd008 •The archive contains a README.md in the root directory with instructions on how to run the tool.",
+      "context_after": "The SQLancer github repo also acknowledges the author’s work as a supported testing method. •Apart from the core tool, the authors provide scripts to automate the bug-finding process for SQLite. The script takes the result of the SQLancer logs and attempts to find the earliest version of SQLite that had the bug. 3 Hardware and Software Environment The hardware used to run the reproducibility test is listed in Table 1. Table 1: Hardware & Software environment Repro Review (Chesetti) CPU Intel(R) Xeon(R) Gold 5218 CPU cores 64 GHz 2.30 RAM 754GB Storage 21.8TB HDD 4 Reproducibility Evaluation 4",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2752,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The SQLancer github repo also acknowledges the author’s work as a supported testing method.",
+      "context_before": "atabase developers and are being worked on. This work is licensed under a Creative Commons Attribution 4.0 International License. SIGMOD ARI Reports ’25, Berlin, Germany ©2025 Copyright held by the owner/author(s). ACMISBN 979-8-4007-2375-9/2025/06 https://doi.org/10.1145/3785021.37879932 Submission The authors shared their code as a fig archive. •Fig archive: figshare.com/s/f601c272cebc253bd008 •The archive contains a README.md in the root directory with instructions on how to run the tool. •The tool works by integrating their method with SQLancer [ 1], a tool to automatically test databases.",
+      "context_after": "•Apart from the core tool, the authors provide scripts to automate the bug-finding process for SQLite. The script takes the result of the SQLancer logs and attempts to find the earliest version of SQLite that had the bug. 3 Hardware and Software Environment The hardware used to run the reproducibility test is listed in Table 1. Table 1: Hardware & Software environment Repro Review (Chesetti) CPU Intel(R) Xeon(R) Gold 5218 CPU cores 64 GHz 2.30 RAM 754GB Storage 21.8TB HDD 4 Reproducibility Evaluation 4.1 Artifacts and Experiments The paper provides the following artifacts to support their cla",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2856,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The script takes the result of the SQLancer logs and attempts to find the earliest version of SQLite that had the bug.",
+      "context_before": "he owner/author(s). ACMISBN 979-8-4007-2375-9/2025/06 https://doi.org/10.1145/3785021.37879932 Submission The authors shared their code as a fig archive. •Fig archive: figshare.com/s/f601c272cebc253bd008 •The archive contains a README.md in the root directory with instructions on how to run the tool. •The tool works by integrating their method with SQLancer [ 1], a tool to automatically test databases. The SQLancer github repo also acknowledges the author’s work as a supported testing method. •Apart from the core tool, the authors provide scripts to automate the bug-finding process for SQLite.",
+      "context_after": "3 Hardware and Software Environment The hardware used to run the reproducibility test is listed in Table 1. Table 1: Hardware & Software environment Repro Review (Chesetti) CPU Intel(R) Xeon(R) Gold 5218 CPU cores 64 GHz 2.30 RAM 754GB Storage 21.8TB HDD 4 Reproducibility Evaluation 4.1 Artifacts and Experiments The paper provides the following artifacts to support their claims. (1)Bug report: A list of bugs filed against production databases by employing their methods. This list is provided in the README of the artifact. (2)Oracle comparision test A script that parses output logs from an SQL",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3051,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "(2)Oracle comparision test A script that parses output logs from an SQLancer run on SQLite3 from various oracles to verify that their method finds bugs that other methods did not.",
+      "context_before": "gs and attempts to find the earliest version of SQLite that had the bug. 3 Hardware and Software Environment The hardware used to run the reproducibility test is listed in Table 1. Table 1: Hardware & Software environment Repro Review (Chesetti) CPU Intel(R) Xeon(R) Gold 5218 CPU cores 64 GHz 2.30 RAM 754GB Storage 21.8TB HDD 4 Reproducibility Evaluation 4.1 Artifacts and Experiments The paper provides the following artifacts to support their claims. (1)Bug report: A list of bugs filed against production databases by employing their methods. This list is provided in the README of the artifact.",
+      "context_after": "(3)Efficiency test SQLancer commands in various configurations to measure the efficacy of their method. 4.2 Reproducibility Process The list of bugs filed by the authors is the main contribution of the paper. The bugs are filed against bug trackers of various databases 57 SIGMOD ARI Reports ’25, June 22–27, 2025, Berlin, Germany Yuvaraj Chesetti and Chi Zhang Method # of tests # successful tests # unsuccesful tests CODD 233,985k 780,488k 25,828k NoRec 658,042k 1,326,809k 47,170k Table 2: Number of tests conducted by CODD and NoREC (SQLite, TiDB, MySQL and CockroachDB). These bugs were found a",
+      "section": "4.1 Artifacts and Experiments",
+      "page": 1,
+      "char_offset": 3698,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "(3)Efficiency test SQLancer commands in various configurations to measure the efficacy of their method.",
+      "context_before": "Table 1: Hardware & Software environment Repro Review (Chesetti) CPU Intel(R) Xeon(R) Gold 5218 CPU cores 64 GHz 2.30 RAM 754GB Storage 21.8TB HDD 4 Reproducibility Evaluation 4.1 Artifacts and Experiments The paper provides the following artifacts to support their claims. (1)Bug report: A list of bugs filed against production databases by employing their methods. This list is provided in the README of the artifact. (2)Oracle comparision test A script that parses output logs from an SQLancer run on SQLite3 from various oracles to verify that their method finds bugs that other methods did not.",
+      "context_after": "4.2 Reproducibility Process The list of bugs filed by the authors is the main contribution of the paper. The bugs are filed against bug trackers of various databases 57 SIGMOD ARI Reports ’25, June 22–27, 2025, Berlin, Germany Yuvaraj Chesetti and Chi Zhang Method # of tests # successful tests # unsuccesful tests CODD 233,985k 780,488k 25,828k NoRec 658,042k 1,326,809k 47,170k Table 2: Number of tests conducted by CODD and NoREC (SQLite, TiDB, MySQL and CockroachDB). These bugs were found across multiple runs of their tool over the course of several months. For the oracle comparison test, we",
+      "section": "4.1 Artifacts and Experiments",
+      "page": 1,
+      "char_offset": 3878,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "NoRec",
+      "technique": "norec",
+      "sentence": "The bugs are filed against bug trackers of various databases 57 SIGMOD ARI Reports ’25, June 22–27, 2025, Berlin, Germany Yuvaraj Chesetti and Chi Zhang Method # of tests # successful tests # unsuccesful tests CODD 233,985k 780,488k 25,828k NoRec 658,042k 1,326,809k 47,170k Table 2: Number of tests conducted by CODD and NoREC (SQLite, TiDB, MySQL and CockroachDB).",
+      "context_before": "e paper provides the following artifacts to support their claims. (1)Bug report: A list of bugs filed against production databases by employing their methods. This list is provided in the README of the artifact. (2)Oracle comparision test A script that parses output logs from an SQLancer run on SQLite3 from various oracles to verify that their method finds bugs that other methods did not. (3)Efficiency test SQLancer commands in various configurations to measure the efficacy of their method. 4.2 Reproducibility Process The list of bugs filed by the authors is the main contribution of the paper.",
+      "context_after": "These bugs were found across multiple runs of their tool over the course of several months. For the oracle comparison test, we followed the instructions provided by the authors. We downloaded the artifacts provided by the author that contain a version of SQLancer that supports their method. The packages dependencies and build process was managed by maven and producers a runnable jar. We ran a scaled down version of test: the paper ran the test for 24 hours, while we ran the the test for 12 hours. We ran the scripts for CODDTest (the author’s method), NoREC and Query Partitioning. The efficien",
+      "section": "4.2 Reproducibility Process",
+      "page": 1,
+      "char_offset": 4087,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We downloaded the artifacts provided by the author that contain a version of SQLancer that supports their method.",
+      "context_before": "d by the authors is the main contribution of the paper. The bugs are filed against bug trackers of various databases 57 SIGMOD ARI Reports ’25, June 22–27, 2025, Berlin, Germany Yuvaraj Chesetti and Chi Zhang Method # of tests # successful tests # unsuccesful tests CODD 233,985k 780,488k 25,828k NoRec 658,042k 1,326,809k 47,170k Table 2: Number of tests conducted by CODD and NoREC (SQLite, TiDB, MySQL and CockroachDB). These bugs were found across multiple runs of their tool over the course of several months. For the oracle comparison test, we followed the instructions provided by the authors.",
+      "context_after": "The packages dependencies and build process was managed by maven and producers a runnable jar. We ran a scaled down version of test: the paper ran the test for 24 hours, while we ran the the test for 12 hours. We ran the scripts for CODDTest (the author’s method), NoREC and Query Partitioning. The efficiency test uses similar scripts but with flags to SQLancer to perform additional logging. The authors ran 24 hour tests across roughly 30 configurations. Similar to the oracle comparison test, we ran a subset of these tests. Specifically, we measured number of executed statements and number of",
+      "section": "4.2 Reproducibility Process",
+      "page": 2,
+      "char_offset": 4632,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "We ran the scripts for CODDTest (the author’s method), NoREC and Query Partitioning.",
+      "context_before": "7,170k Table 2: Number of tests conducted by CODD and NoREC (SQLite, TiDB, MySQL and CockroachDB). These bugs were found across multiple runs of their tool over the course of several months. For the oracle comparison test, we followed the instructions provided by the authors. We downloaded the artifacts provided by the author that contain a version of SQLancer that supports their method. The packages dependencies and build process was managed by maven and producers a runnable jar. We ran a scaled down version of test: the paper ran the test for 24 hours, while we ran the the test for 12 hours.",
+      "context_after": "The efficiency test uses similar scripts but with flags to SQLancer to perform additional logging. The authors ran 24 hour tests across roughly 30 configurations. Similar to the oracle comparison test, we ran a subset of these tests. Specifically, we measured number of executed statements and number of query plans generated (Table 3 in the paper).4.3 Results As the bug-finding process is non-deterministic, our results differ numerically from those reported in the original paper. Additionally, we ran tests for half the duration used by the authors. However, the overall trends and methodology a",
+      "section": "4.2 Reproducibility Process",
+      "page": 2,
+      "char_offset": 4956,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "coddtest",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The efficiency test uses similar scripts but with flags to SQLancer to perform additional logging.",
+      "context_before": "CockroachDB). These bugs were found across multiple runs of their tool over the course of several months. For the oracle comparison test, we followed the instructions provided by the authors. We downloaded the artifacts provided by the author that contain a version of SQLancer that supports their method. The packages dependencies and build process was managed by maven and producers a runnable jar. We ran a scaled down version of test: the paper ran the test for 24 hours, while we ran the the test for 12 hours. We ran the scripts for CODDTest (the author’s method), NoREC and Query Partitioning.",
+      "context_after": "The authors ran 24 hour tests across roughly 30 configurations. Similar to the oracle comparison test, we ran a subset of these tests. Specifically, we measured number of executed statements and number of query plans generated (Table 3 in the paper).4.3 Results As the bug-finding process is non-deterministic, our results differ numerically from those reported in the original paper. Additionally, we ran tests for half the duration used by the authors. However, the overall trends and methodology align with the original work. •All bugs identified through the authors’ provided list point to legit",
+      "section": "4.2 Reproducibility Process",
+      "page": 2,
+      "char_offset": 5041,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "•CODDTest reported 1297 bugs, NoRec reported 644 bugs, and query partitioning generated 8280 database logs.",
+      "context_before": "subset of these tests. Specifically, we measured number of executed statements and number of query plans generated (Table 3 in the paper).4.3 Results As the bug-finding process is non-deterministic, our results differ numerically from those reported in the original paper. Additionally, we ran tests for half the duration used by the authors. However, the overall trends and methodology align with the original work. •All bugs identified through the authors’ provided list point to legitimate issues, which we consider the strongest indicator of reproducibility as it demonstrates real-world impact.",
+      "context_after": "The authors performed additional deduplication work that we did not replicate. However, we verified that the bug reports generated by each method are legitimate. •Table 2 presents the efficiency test results. 5 Summary We successfully reproduced the core claims of the original paper. References [1] [n. d.]. https://github.com/sqlancer/sqlancer. [2]Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages. doi:10.1145/3709674 58",
+      "section": "4.2 Reproducibility Process",
+      "page": 2,
+      "char_offset": 5852,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "coddtest",
+        "norec",
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:59:32Z",
+    "is_model_written": true,
+    "summary": "An independent reproducibility report on the SIGMOD 2025 paper introducing CODDTest. The reviewer re-ran the authors' artifacts to check the claims: an oracle comparison verifying CODDTest finds bugs other oracles do not, an efficiency test measuring its throughput, and a script identifying the earliest SQLite version carrying each bug. The reported counts were CODDTest 1297 bugs, NoREC 644, with query partitioning producing 8280 database logs.",
+    "narrative": "This paper is a reproduction of SQLancer work rather than a citation of it. CODDTest is implemented inside SQLancer -- the report states the tool works by integrating the method with SQLancer, and that the SQLancer repository acknowledges the work as a supported testing method -- so reproducing it meant running a version of SQLancer that supports the method, from the authors' artifacts. The comparison it verifies is between three oracles in that one tool: CODDTest, NoREC and query partitioning, all executed through SQLancer against SQLite.",
+    "roles": {
+      "M1": "incidental",
+      "M2": "incidental",
+      "M3": "background",
+      "M4": "reuse_implementation",
+      "M5": "reuse_implementation",
+      "M6": "reuse_component",
+      "M7": "reuse_component",
+      "M8": "reuse_component",
+      "M9": "result_comparison",
+      "M10": "reuse_implementation",
+      "M11": "result_comparison",
+      "M12": "reuse_component",
+      "M13": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M5",
+          "M10"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "•The tool works by integrating their method with SQLancer [ 1], a tool to automatically test databases.",
+            "section": "1 Introduction",
+            "page": 1
+          },
+          {
+            "mention_id": "M5",
+            "sentence": "The SQLancer github repo also acknowledges the author’s work as a supported testing method.",
+            "section": "1 Introduction",
+            "page": 1
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "We downloaded the artifacts provided by the author that contain a version of SQLancer that supports their method.",
+            "section": "4.2 Reproducibility Process",
+            "page": 2
+          }
+        ],
+        "reasoning": "M4 states the tool under review works by integrating its method with SQLancer, M5 that the SQLancer repository acknowledges it as a supported testing method, and M10 that the reproduction ran a version of SQLancer supporting the method. The whole exercise is conducted through SQLancer.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The report reproduces CODDTest rather than proposing anything; the extension is the reproduced paper's, not this one's."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M11",
+          "M13"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M11",
+            "sentence": "We ran the scripts for CODDTest (the author’s method), NoREC and Query Partitioning.",
+            "section": "4.2 Reproducibility Process",
+            "page": 2
+          },
+          {
+            "mention_id": "M13",
+            "sentence": "•CODDTest reported 1297 bugs, NoRec reported 644 bugs, and query partitioning generated 8280 database logs.",
+            "section": "4.2 Reproducibility Process",
+            "page": 2
+          }
+        ],
+        "reasoning": "M11 records running the scripts for CODDTest, NoREC and query partitioning, and M13 gives the counts each reported. The comparison is between oracles within SQLancer, which is what the reproduction set out to verify.",
+        "techniques": [
+          "norec",
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The report is an evaluation of reproducibility and makes no claim about standing."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3786673.json b/_data/papers/paper_doi_10_1145_3786673.json
new file mode 100644
index 0000000..b537fb0
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3786673.json
@@ -0,0 +1,1085 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:28Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3786673",
+    "title": "One DBMS, Two Modes, and a Bunch of Bugs: Catching Logic Bugs in Distributed DBMSs via Differential Testing",
+    "authors": [
+      "Zikun Fu",
+      "Jiaju Bai",
+      "Hong-Bo Feng",
+      "Kang Chen"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": "10.1145/3786673",
+    "arxiv_id": null,
+    "s2_paper_id": "dc60906c576d85944e5ba7612dd4ce32d34de9af",
+    "url": "https://doi.org/10.1145/3786673",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3786673",
+      "retrieved_at": "2026-09-09T01:29:28Z",
+      "chars": 91748,
+      "content_sha256": "sha256:debb87726b9adbc4f0454b44000cbbb25fce20a5756bc6def0b1ef6639c521b3"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 24,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2190
+      },
+      {
+        "number": "2",
+        "title": "Background and Motivation",
+        "start": 14387
+      },
+      {
+        "number": "2.1",
+        "title": "Logic Bugs in Distributed DBMSs",
+        "start": 14415
+      },
+      {
+        "number": "2.2",
+        "title": "Limitations of Existing Approaches",
+        "start": 16285
+      },
+      {
+        "number": "2.3",
+        "title": "Motivation and Challenges",
+        "start": 21121
+      },
+      {
+        "number": "3",
+        "title": "DistSQL Design",
+        "start": 25884
+      },
+      {
+        "number": "3.1",
+        "title": "Architecture",
+        "start": 25901
+      },
+      {
+        "number": "3.2",
+        "title": "Testing Workflow",
+        "start": 26472
+      },
+      {
+        "number": "4",
+        "title": "Distributed Database State Mutation",
+        "start": 30006
+      },
+      {
+        "number": "5.1",
+        "title": "Overview",
+        "start": 34909
+      },
+      {
+        "number": "5.2",
+        "title": "Feedback Workflow",
+        "start": 36442
+      },
+      {
+        "number": "5.3",
+        "title": "Plan Patterns based on Operator Interactions",
+        "start": 38393
+      },
+      {
+        "number": "5.4",
+        "title": "Distributed Execution Score",
+        "start": 43785
+      },
+      {
+        "number": "5.5",
+        "title": "Query Plan Feature Extract Algorithm",
+        "start": 45214
+      },
+      {
+        "number": "5.6",
+        "title": "Feature Pool",
+        "start": 46963
+      },
+      {
+        "number": "7.1",
+        "title": "Experimental Setup",
+        "start": 51418
+      },
+      {
+        "number": "7.2",
+        "title": "Open-Source Distributed DBMS Testing",
+        "start": 53379
+      },
+      {
+        "number": "7.3",
+        "title": "Sensitivity Analysis",
+        "start": 58619
+      },
+      {
+        "number": "7.4",
+        "title": "Comparison",
+        "start": 61217
+      },
+      {
+        "number": "7.5",
+        "title": "Case Studies of the Found Bugs",
+        "start": 63649
+      },
+      {
+        "number": "7.6",
+        "title": "Production Study",
+        "start": 69306
+      },
+      {
+        "number": "7.7",
+        "title": "Other System Bugs",
+        "start": 69970
+      },
+      {
+        "number": "9.1",
+        "title": "Coverage-Guided Fuzzing",
+        "start": 71949
+      },
+      {
+        "number": "9.2",
+        "title": "DBMS Fuzzing",
+        "start": 72681
+      },
+      {
+        "number": "9.3",
+        "title": "Distributed System Fuzzing",
+        "start": 74463
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "anse1. 2015. sqlsmith: A random SQL query generator. https://github.com/anse1/sqlsmith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. 59:22 Zi-Xuan Fu, Jia-Ju Bai, Hong-Bo Feng, and Kang Chen",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. Proceedings of the ACM on Management of Data 2, 3 (2024), 1–26.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "Jia-Ju Bai, Yu-Ping Wang, Jie Yin, and Shi-Min Hu. 2016. Testing error handling code in device drivers using characteristic fault injection. In Proceedings of the 2016 USENIX Annual Technical Conference. 635–647.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Tim Blazytko, Cornelius Aschermann, Moritz Schlögel, Ali Abbasi, Sergej Schumilo, Simon Wörner, and Thorsten Holz. 2019. GRIMOIRE: synthesizing structure while fuzzing. In Proceedings of the 28th USENIX Security Symposium, Vol. 19.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Mohamed-Slim Bouguerra, Thierry Gautier, Denis Trystram, and Jean-Marc Vincent. 2010. A flexible checkpoint/restart model in distributed systems. In Proceedings of the 8th International Conference on Parallel Processing and Applied Mathematics (PPAM). 206–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Anton Burtsev, Prashanth Radhakrishnan, Mike Hibler, and Jay Lepreau. 2009. Transparent checkpoints of closed distributed systems in emulab. In Proceedings of the 4th European Conference on Computer Systems (EuroSys). 173–186.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Yuanliang Chen, Fuchen Ma, Yuanhang Zhou, Ming Gu, Qing Liao, and Yu Jiang. 2024. Chronos: Finding Timeout Bugs in Practical Distributed Systems by Deep-Priority Fuzzing with Transient Delay. In Security and Privacy (SP). 1939–1955. https://doi.org/10.1109/SP54263.2024.00109",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "ClickHouse. 2025. ClickHouse ®is a real-time analytics DBMS. https://github.com/ClickHouse/ClickHouse.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "clickhouse-use 2025. ClickHouse use cases. https://clickhouse.com/use-cases.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "cockroachdb. 2025. CockroachDB — the cloud native, distributed SQL database designed for high availability, effortless scale, and control over data placement. https://github.com/cockroachdb/cockroach.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Kai Cong, Li Lei, Zhenkun Yang, and Fei Xie. 2015. Automatic fault injection for driver robustness testing. In Proceedings of the 2015 International Symposium on Software Testing and Analysis (ISSTA). 361–372.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free DBMS fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Shuitao Gan, Chao Zhang, Xiaojun Qin, Xuwen Tu, Kang Li, Zhongyu Pei, and Zuoning Chen. 2018. CollAFL: path sensitive fuzzing. In Proceedings of the 39th IEEE Symposium on Security and Privacy. 679–696.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Yu Gao, Wensheng Dou, Dong Wang, Wenhan Feng, Jun Wei, Hua Zhong, and Tao Huang. 2023. Coverage guided fault injection for cloud systems. In. IEEE, 2211–2223.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jiawei Tyler Gu, Xudong Sun, Wentao Zhang, Yuxuan Jiang, Chen Wang, Mandana Vaziri, Owolabi Legunsen, and Tianyin Xu. 2023. Acto: Automatic end-to-end testing for operation correctness of cloud system management. In Proceedings of the 29th Symposium on Operating Systems Principles. 96–112.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Diwaker Gupta, Kashi Venkatesh Vishwanath, Marvin McNett, Amin Vahdat, Ken Yocum, Alex Snoeren, and Geoffrey M Voelker. 2011. DieCast: Testing distributed systems with an accurate scale model. ACM Transactions on Computer Systems (TOCS) 29, 2 (2011), 1–48.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023. Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis. In 2023 USENIX Annual Technical Conference (USENIX ATC 23). 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Adrian Herrera, Hendra Gunadi, Shane Magrath, Michael Norrish, Mathias Payer, and Antony L Hosking. 2021. Seed selection for successful fuzzing. In Proceedings of the 2021 International Symposium on Software Testing and Analysis (ISSTA). 230–243.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "jepsen io. 2018. Jepsen: A framework for distributed systems verification, with fault injection. https://github.com/jepsenio/jepsen.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, Kangjie Lu, and Shi-Min Hu. 2020. Fuzzing error handling code using context-sensitive software fault injection. In Proceedings of the 29th USENIX Security Symposium. 2595–2612.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. In 32nd USENIX Security Symposium (USENIX Security 23). 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via {graph-based}oracle construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). 397–417.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 24,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting Logic Bugs in Database Engines via Equivalent Expression Transformation. In 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24). 821–835.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Ahmed Khoumsi. 2002. A temporal approach for testing distributed systems. IEEE Transactions on Software Engineering 28, 11 (2002), 1085–1103.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Tanakorn Leesatapornwongsa, Mingzhe Hao, Pallavi Joshi, Jeffrey F Lukman, and Haryadi S Gunawi. 2014. {SAMC}:{Semantic-Aware}model checking for fast discovery of deep bugs in cloud systems. In 11th USENIX Symposium on Operating Systems Design and Implementation (OSDI 14). 399–414.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Junqiang Li, Senyi Li, Keyao Li, Falin Luo, Hongfang Yu, Shanshan Li, and Xiang Li. 2024. ECFuzz: Effective Configuration Fuzzing for Large-Scale Systems. In Proceedings of the 46th IEEE/ACM International Conference on Software 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. One DBMS, Two Modes, and a Bunch of Bugs: Catching Logic Bugs in Distributed DBMSs via Differential Testing ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-oriented DBMS fuzzing. In Engineering (ICDE). IEEE, 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS bugs via configuration-based equivalent transformation. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of {DBMS}with Coverage-based Guidance. In 31st USENIX Security Symposium (USENIX Security 22). 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Shuang Liu, Chenglin Tian, Jun Sun, Ruifeng Wang, Wei Lu, Yongxin Zhao, Yinxing Xue, Junjie Wang, and Xiaoyong Du. 2024. Semantic Conformance Testing of Relational DBMS. Proceedings of the VLDB Endowment 18, 3 (2024), 850–862.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. In Proceedings of the 44th International Conference on Software Engineering. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Chang Lou, Yuzhuo Jing, and Peng Huang. 2022. Demystifying and checking silent semantic violations in large distributed systems. In 16th USENIX Symposium on Operating Systems Design and Implementation (OSDI 22). 91–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jeffrey F Lukman, Huan Ke, Cesar A Stuardo, Riza O Suminto, Daniar H Kurniawan, Dikaimin Simon, Satria Priambada, Chen Tian, Feng Ye, Tanakorn Leesatapornwongsa, et al .2019. Flymc: Highly scalable testing of complex interleavings in distributed systems. In Proceedings of the Fourteenth EuroSys Conference 2019. 1–16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Tao Lyu, Liyi Zhang, Zhiyao Feng, Yueyang Pan, Yujie Ren, Meng Xu, Mathias Payer, and Sanidhya Kashyap. 2024. Monarch: A Fuzzing Framework for Distributed File Systems. In 2024 USENIX Annual Technical Conference (USENIX ATC 24). 529–543.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Ruijie Meng, George Pîrlea, Abhik Roychoudhury, and Ilya Sergey. 2023. Greybox Fuzzing of Distributed Systems. In Proceedings of the 2023 ACMSIGSAC Conference on Computer and Communications Security. 1615–1629.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "oceanbase. 2025. OceanBase is an enterprise distributed relational database with high availability, high performance, horizontal scalability, and compatibility with SQL standards. https://github.com/Oceanbase/oceanbase.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Mike Papadakis, and Yves Le Traon. 2019. Semantic fuzzing with Zest. InProceedings of the 2019 International Symposium on Software Testing and Analysis (ISSTA). 329–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Shankara Pailoor, Andrew Aday, and Suman Jana. 2018. MoonShine: optimizing OS fuzzer seed selection with trace distillation. In Proceedings of the 27th USENIX Security Symposium. 729–743.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Van-Thuan Pham, Marcel Böhme, Andrew E Santosa, Alexandru Răzvan Căciulescu, and Abhik Roychoudhury. 2019. Smart greybox fuzzing. IEEE Transactions on Software Engineering (TSE) 47, 9 (2019), 1980–1997.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "pingcap. 2025. TiDBthe open-source, cloud-native, distributed SQL database designed for modern applications. https://github.com/pingcap/tidb.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Alexandre Rebert, Sang Kil Cha, Thanassis Avgerinos, Jonathan Foote, David Warren, Gustavo Grieco, and David Brumley. 2014. Optimizing seed selection for fuzzing. In Proceedings of the 23rd USENIX Security Symposium. 861–875.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 46,
+      "text": "Upamanyu Sharma, Ralf Jung, Joseph Tassarotti, Frans Kaashoek, and Nickolai Zeldovich. 2023. Grove: A separationlogic library for verifying distributed systems. In Proceedings of the 29th Symposium on Operating Systems Principles. 113–129.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Donald R Slutz. 1998. Massive stochastic testing of SQL. In VLDB, Vol. 98. Citeseer, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing database systems via differential query execution. In Engineering (ICSE). IEEE, 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Jiansen Song, Wensheng Dou, Yu Gao, Ziyu Cui, Yingying Zheng, Dong Wang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction. Proceedings of the VLDB Endowment 17, 8 (2024), 1884–1897.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "sqlancer. 2020. Automated testing to find logic and performance bugs in database systems. https://github.com/ sqlancer/sqlancer.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 51,
+      "text": "Prashast Srivastava and Mathias Payer. 2021. Gramatron: effective grammar-aware fuzzing. In Proceedings of the 2021 International Symposium on Software Testing and Analysis (ISSTA). 244–256. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. 59:24 Zi-Xuan Fu, Jia-Ju Bai, Hong-Bo Feng, and Kang Chen",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Xudong Sun, Wenqing Luo, Jiawei Tyler Gu, Aishwarya Ganesan, Ramnatthan Alagappan, Michael Gasch, Lalith Suresh, and Tianyin Xu. 2022. Automatic reliability testing for cluster management controllers. In 16th USENIX Symposium on Operating Systems Design and Implementation (OSDI 22). 143–159.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting logic bugs of join optimizations in dbms. Proceedings of the ACM on Management of Data 1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Dong Wang, Wensheng Dou, Yu Gao, Chenao Wu, Jun Wei, and Tao Huang. 2023. Model checking guided testing for distributed systems. In Proceedings of the Eighteenth European Conference on Computer Systems. 127–143.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Haoze Wu, Jia Pan, and Peng Huang. 2024. Efficient exposure of partial failure bugs in distributed systems with inferred abstract states. In 21st USENIX Symposium on Networked Systems Design and Implementation (NSDI 24). 1267–1283.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Jianan Yao, Runzhou Tao, Ronghui Gu, Jason Nieh, Suman Jana, and Gabriel Ryan. 2021. {DistAI}:{Data-Driven} automated invariant learning for distributed protocols. In 15th USENIX symposium on operating systems design and implementation (OSDI 21). 405–421.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Ding Yuan, Yu Luo, Xin Zhuang, Guilherme Renna Rodrigues, Xu Zhao, Yongle Zhang, Pranay U Jain, and Michael Stumm. 2014. Simple testing can prevent most critical failures: An analysis of production failures in distributed {Data-Intensive}systems. In 11th USENIX Symposium on Operating Systems Design and Implementation (OSDI 14). 249–265.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Xinhao Yuan and Junfeng Yang. 2020. Effective concurrency testing for distributed systems. In Proceedings of the Twenty-Fifth International Conference on Architectural Support for Programming Languages and Operating Systems. 1141–1156.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "yugabyte. 2025. YugabyteDBthe cloud native distributed SQL database for mission-critical applications. https: //github.com/yugabyte/yugabyte-db.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "yugabyte-use 2025. What YugabyteDB Users Have to Say. https://www.yugabyte.com/about/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. Proceedings of the ACM on Management of Data 3, 1 (2025), 1–24.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 62,
+      "text": "Tony Nuda Zhang, Travis Hance, Manos Kapritsos, Tej Chajed, and Bryan Parno. 2024. Inductive invariants that spark joy: using invariant taxonomies to streamline distributed protocol proofs. In 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24). 837–853.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Yongle Zhang, Junwen Yang, Zhuqi Jin, Utsav Sethi, Kirk Rodrigues, Shan Lu, and Ding Yuan. 2021. Understanding and detecting software upgrade failures in distributed systems. In Proceedings of the ACMSIGOPS 28th Symposium on Operating Systems Principles. 116–131.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Zenong Zhang, George Klees, Eric Wang, Michael Hicks, and Shiyi Wei. 2023. Fuzzing configurations of program options. ACM Transactions on Software Engineering and Methodology 32, 2 (2023), 1–21.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security. 955–970. Received July 2025; revised October 2025; accepted November 2025 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. 59:22 Zi-Xuan Fu, Jia-Ju Bai, Hong-Bo Feng, and Kang Chen",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. Proceedings of the ACM on Management of Data 2, 3 (2024), 1–26.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 23,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via {graph-based}oracle construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). 397–417.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 50,
+      "text": "sqlancer. 2020. Automated testing to find logic and performance bugs in database systems. https://github.com/ sqlancer/sqlancer.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 61,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. Proceedings of the ACM on Management of Data 3, 1 (2025), 1–24.",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 2,18,24,44,45,48,61]",
+      "technique": "qpg",
+      "sentence": "Some approaches check the consistency of semantically equivalent predicates or expressions across multiple statements [ 2,18,24,44,45,48,61].",
+      "context_before": "ferent execution interleavings. Other approaches detect error handling bugs by injecting faults [ 15,26,35,52]. A few general purpose bug detection approaches [ 20,36] can find logic bugs in distributed DBMSs, but they rely on fixed SQL workloads. Because these tools are designed primarily to expose system level issues, they are less effective at identifying logic bugs that involve richer SQL semantics. (2) Bug detection for general DBMSs: Numerous approaches detect logic bugs in DBMSs by validating against predefined test oracles and can be applied to both centralized and distributed systems.",
+      "context_after": "Others verify the correctness of query optimization [ 43,53] or specific features such as indexes and transactions [ 3,23,30,49]. However, these approaches do not target distributed execution and are therefore less effective at identifying distributed logic bugs. To this end, we design and implement the DistSQL framework to detect logic bugs in distributed DBMSs. DistSQL rests on two observations. (1) Many distributed DBMSs support both centralized and distributed execution modes. (2) If the same query produces different results across modes, this indicates at least one bug in the database. B",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5480,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. 59:22 Zi-Xuan Fu, Jia-Ju Bai, Hong-Bo Feng, and Kang ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 43,53]",
+      "technique": "norec",
+      "sentence": "Others verify the correctness of query optimization [ 43,53] or specific features such as indexes and transactions [ 3,23,30,49].",
+      "context_before": "ction approaches [ 20,36] can find logic bugs in distributed DBMSs, but they rely on fixed SQL workloads. Because these tools are designed primarily to expose system level issues, they are less effective at identifying logic bugs that involve richer SQL semantics. (2) Bug detection for general DBMSs: Numerous approaches detect logic bugs in DBMSs by validating against predefined test oracles and can be applied to both centralized and distributed systems. Some approaches check the consistency of semantically equivalent predicates or expressions across multiple statements [ 2,18,24,44,45,48,61].",
+      "context_after": "However, these approaches do not target distributed execution and are therefore less effective at identifying distributed logic bugs. To this end, we design and implement the DistSQL framework to detect logic bugs in distributed DBMSs. DistSQL rests on two observations. (1) Many distributed DBMSs support both centralized and distributed execution modes. (2) If the same query produces different results across modes, this indicates at least one bug in the database. Because the centralized mode is easier to deploy and is often used for testing and debugging, it is generally better tested and mor",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5622,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "Approach Main Target Bug Type Feedback Black-Box Logic Checker Database State Mutation Morpheus [58] Distributed System Concurrency Conflict Analysis No-Oathkeeper [33] Distributed System Semantic None No-Chronos [8] Distributed System Timeout None No-Generic Jepsen [20] Distributed System Logic None Yes Fixed Rules Fixed Input Mallory [36] Distributed System Logic Timeline Summary No Fixed Rules Fixed Input RAGS [47] DBMS Logic None Yes Differential Generic Squirrel [65] DBMS Memory Branch Coverage No-Generic Lego [28] DBMS Memory Type-Affinity No-Generic CODDTest [61] DBMS Logic None Yes Metamorphic Generic EET [24] DBMS Logic None Yes Metamorphic Generic DQP [3] DBMS Logic None Yes Metamorphic Generic Mozi [29] DBMS Logic None Yes Differential Generic Radar [49] DBMS Logic None Yes Differential Generic SemConT [31] DBMS Logic Keyword and Rule Yes Verification Generic TQS [53] DBMS Logic Graph Similarity Yes Ground Truth Generic QPG [2] DBMS Logic Query Plan Text Yes Metamorphic Generic DistSQL (this work) Distributed DBMS Logic Query Plan Feature Yes Differential Favor Distributed Logic centralized execution as a reference, our approach finds distributed logic bugs often missed by existing techniques.",
+      "context_before": "n Table 1. Reported issue types across four distributed DBMSs. DBMS logic consistency memory assert settings status perf. funcmiss resource stats enhance others Total ClickHouse [9] 365 78 186 257 168 58 59 400 32 37 6 40 1686 CockroachDB [11] 89 138 70 63 128 46 82 133 39 96 42 71 997 TiDB [41] 409 136 186 61 96 66 83 198 41 166 14 77 1533 YugaByteDB [59] 131 306 198 104 99 94 105 139 60 102 68 87 1493 Total 994 658 640 485 491 264 329 870 172 401 130 275 5709 Ratio 17.4% 11.5% 11.2% 8.5% 8.6% 4.6% 5.8% 15.2% 3.0% 7.0% 2.3% 4.8% 100% Table 2. Comparison of state-of-the-art testing approaches.",
+      "context_after": "•We develop DistSQL, which integrates two techniques: (1) distributed diversity oriented database state mutation, and (2) distributed interaction guided execution space exploration. These techniques address two core challenges in testing distributed DBMSs by effectively triggering diverse distributed executions which can efficiently explore the explosive execution space. •We evaluate DistSQL on five widely used open source distributed DBMSs, TiDB [ 41], CockroachDB [ 11], YugabyteDB [ 59], ClickHouse [ 9], and OceanBase [ 37], as well as a widely deployed commercial DBMS. In total, DistSQL fi",
+      "section": "1 Introduction",
+      "page": 4,
+      "char_offset": 12173,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "coddtest",
+        "dqp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "EET [ 24], and CODDTest [ 61] find logic bugs by rewriting queries into another forms that are expected to produce the same results.",
+      "context_before": "ted logic tend to focus on specific distributed constraints and pay less attention to diverse SQL semantics, which in turn narrows the scope of distributed logic testing for distributed DBMSs. DBMS testing. RAGS [ 47] performs differential testing across DBMS implementations to find logic bugs. However, it is constrained by the limited set of features shared among DBMSs, which narrows the scope of its testing. Squirrel [ 65] and Lego [ 28] targets memory-related bugs, which are out of scope for this work. Several approaches aim to find logic bugs in DBMSs through unguided test case generation.",
+      "context_after": "However, the test cases are limited by the scope of these transformation rules. Mozi [ 29] and Radar [ 49] find logic bugs by executing the same statements under different internal options or table constraints and comparing results. DQP [ 3] compares executions with different query hints applied, where query plans act as local heuristics rather than guiding test case generation. These three methods tend to find bugs tied to specific features but does not test other aspects of database processing logic. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. 59:6 Zi-Xuan Fu, Jia-Ju Ba",
+      "section": "2.2 Limitations of Existing Approaches",
+      "page": 5,
+      "char_offset": 18579,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "DQP [ 3] compares executions with different query hints applied, where query plans act as local heuristics rather than guiding test case generation.",
+      "context_before": "DBMSs, which narrows the scope of its testing. Squirrel [ 65] and Lego [ 28] targets memory-related bugs, which are out of scope for this work. Several approaches aim to find logic bugs in DBMSs through unguided test case generation. EET [ 24], and CODDTest [ 61] find logic bugs by rewriting queries into another forms that are expected to produce the same results. However, the test cases are limited by the scope of these transformation rules. Mozi [ 29] and Radar [ 49] find logic bugs by executing the same statements under different internal options or table constraints and comparing results.",
+      "context_after": "These three methods tend to find bugs tied to specific features but does not test other aspects of database processing logic. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. 59:6 Zi-Xuan Fu, Jia-Ju Bai, Hong-Bo Feng, and Kang Chen Some approaches leverage specific scenario feedback to guide test case generation. SemConT [ 31] formalizes SQL semantics and verifies results against a formal model, using rule coverage to guide generation. However, its formal model struggles with features such as correlated subqueries and diverse expressions. TQS [ 53] splits a wide table into mul",
+      "section": "2.2 Limitations of Existing Approaches",
+      "page": 5,
+      "char_offset": 18945,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "QPG [ 2] improves SQLancer [ 50] by using query plan text as feedback to mutate the database state when no new plans are found, which focuses only on centralized database states and ignores distributed ones.",
+      "context_before": "Hong-Bo Feng, and Kang Chen Some approaches leverage specific scenario feedback to guide test case generation. SemConT [ 31] formalizes SQL semantics and verifies results against a formal model, using rule coverage to guide generation. However, its formal model struggles with features such as correlated subqueries and diverse expressions. TQS [ 53] splits a wide table into multiple smaller tables and joins them to test the correctness of join operations, guiding generation by test case similarity. Similarly, it cannot generate statements with subqueries as join operands or non-equality joins.",
+      "context_after": "Although this approach is effective under low-diversity workloads, its discriminative power decreases as SQL diversity increases. In our evaluation on TiDB, 5 million queries produced over 2.3 million unique plans (47.93%), revealing low feedback granularity. Consequently, QPG is less effective on distributed DBMSs, as its feedback signal becomes noisy and less informative. Overall, while these approaches have shown effectiveness in finding certain classes of logic bugs, their testing scope is inherently limited by their methodology. In addition, their database mutation strategies do not acco",
+      "section": "2.2 Limitations of Existing Approaches",
+      "page": 6,
+      "char_offset": 19914,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Consequently, QPG is less effective on distributed DBMSs, as its feedback signal becomes noisy and less informative.",
+      "context_before": "generation by test case similarity. Similarly, it cannot generate statements with subqueries as join operands or non-equality joins. QPG [ 2] improves SQLancer [ 50] by using query plan text as feedback to mutate the database state when no new plans are found, which focuses only on centralized database states and ignores distributed ones. Although this approach is effective under low-diversity workloads, its discriminative power decreases as SQL diversity increases. In our evaluation on TiDB, 5 million queries produced over 2.3 million unique plans (47.93%), revealing low feedback granularity.",
+      "context_after": "Overall, while these approaches have shown effectiveness in finding certain classes of logic bugs, their testing scope is inherently limited by their methodology. In addition, their database mutation strategies do not account for distributed database states, such as data distribution across nodes. Moreover, even when applied to distributed DBMSs, their focus on a single execution mode prevents them from capturing inconsistencies specifically introduced by distributed execution, and it also prevents them from developing specialized runtime feedback and mutation strategies, leading to missed di",
+      "section": "2.2 Limitations of Existing Approaches",
+      "page": 6,
+      "char_offset": 20382,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[ 1,22,45,47]",
+      "technique": "pqs",
+      "sentence": "We implement an abstract syntax tree (AST) based SQL generator [ 1,22,45,47] that constructs queries in a top down manner by first selecting the SQL statement type (e.",
+      "context_before": "DBMSs, we address the above challenges and develop DistSQL, a differential testing framework tailored for this purpose, as shown in Figure 2. DistSQL first randomly initializes two database instances with the same data, one in centralized mode and one in distributed mode. It then continuously generates queries and executes them under both modes for differential testing. If feedback indicates no new distributed interaction after several consecutive executions, DistSQL performs database state mutation. 3.2 Testing Workflow The testing workflow continuously generates and executes SQL statements.",
+      "context_after": "g., SELECT ,CREATE TABLE ) and then recursively generating its components (e.g., ORDER BY clauses, expressions). Generation is guided by seeds, where each seed is an integer sequence [ 22]. During generation, when there are 𝑛feasible choices, the next value 𝑠in the seed selects an option via 𝑐=𝑠mod𝑛. For example, when a function needs to be chosen from several options (e.g., sum,avg,min,max), we have𝑛=4. If𝑠=9, then𝑐=1, and thus theavgfunction (counting from 0) is selected. This enables reproducible and controllable mutation of test cases. Additionally, we reduce the likelihood of generating o",
+      "section": "3.2 Testing Workflow",
+      "page": 7,
+      "char_offset": 26565,
+      "cited_reference": {
+        "number": 45,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker",
+      "surface": "[ 2]",
+      "technique": "qpg",
+      "sentence": "One approach is to treat the entire query plan as a pattern [ 2].",
+      "context_before": "and promote the discovery of new execution paths. (§ 4). (S1) Seed mutation: To continue testing, a seed is selected from the seed pool and randomly mutated to generate a new SQL query. This enables the exploration of meaningful variations of previously successful queries, progressively driving the system into various distributed states. 5.3 Plan Patterns based on Operator Interactions A key problem here is to determine which subgraphs (patterns) of a query plan, or whether the entire query plan, should be used as the pattern in a query plan feature for feedback. Entire query plan as pattern.",
+      "context_after": "However, it is obviously overly sensitive, particularly under diverse SQL inputs (§ 2.2), where nearly half of the queries result in unique plans. In practice, many query plans appear different but are actually repetitive in nature. They share similar core structure and semantics, with only minor variations (e.g., a few operators replaced or reordered). This motivates the need for a more fine-grained and semantics-aware representation of execution logic. Basic operators as pattern. An alternative is to track which operators appear in the query plan. However, focusing on individual operators d",
+      "section": "5.3 Plan Patterns based on Operator Interactions",
+      "page": 10,
+      "char_offset": 38652,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. 59:22 Zi-Xuan Fu, Jia-Ju Bai, Hong-Bo Feng, and Kang ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "We compare DistSQL against SOTA database and distributed system testing tools, including Squirrel [ 65], QPG [ 2], and EET [ 24], as well as distributed system testing tools, including Jepsen [20] and Mallory [36].",
+      "context_before": "e components (such as data storage and parallel execution), and allocate an additional node to each remaining component (such as coordinators or auxiliary workers). The use of three nodes captures a critical transition from centralized to distributed execution, where the differences in behavior become significant. Beyond three nodes, the marginal increase in execution diversity tends to be smaller [ 57]. To simulate a distributed environment, each node is deployed in a separate Docker container. All containers are connected via a virtual subnet and run on the same physical machine. Comparison.",
+      "context_after": "(§ 7.4) Machine configuration. All experiments are conducted on a machine with an AMDEPYC 7J13 64-Core Processor (2.45 GHz), 512 GB of RAM, running Ubuntu 22.04. All systems are tested with their latest versions at evaluation time. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. One DBMS, Two Modes, and a Bunch of Bugs: Catching Logic Bugs in Distributed DBMSs via Differential Testing 59:15 7.2 Open-Source Distributed DBMS Testing Each DBMS is tested three times, with each test lasting one week. We manually analyze root causes using query plans and developer feedback. After i",
+      "section": "7.1 Experimental Setup",
+      "page": 14,
+      "char_offset": 52752,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "3 Sensitivity Analysis To evaluate the effectiveness of our feedback mechanism based on query plan feature, we implement three variants of DistSQL: (1) DistSQL Random, which generates random SQL queries without feedback, (2) DistSQL Text, which uses the query plan text-guided feedback identical to that of QPG [ 2], with its paper and code referenced, and (3) DistSQL JoinAware, identical to DistSQL except that it does not sort the children of join operators, as their order may affect the underlying logic.",
+      "context_before": "Z QMBOGFBUVSFT DistSQL EETQPG Mallory JepsenDistSQL Text Rand JoinAware Testing time (h) Testing time (h) Testing time (h) Testing time (h) Testing time (h) Fig. 4. Coverage growth. (a)–(b): Sensitivity experiments; (c): Comparison with existing approaches. bugs are rarely triggered in centralized mode with relatively lower concurrency but occur more frequently in distributed mode with higher concurrency. ❹Float: 7 of these bugs involve floatingpoint operations, showing that our approach effectively handles floating-point non-determinism, enabling diverse test cases without false positives. 7.",
+      "context_after": "We run a 24-hour experiment with these tools and DistSQL. Table 6 summarizes the number of query plan texts, features (Feat.) extracted by DistSQL, and two bug metrics: LBug (total logic bugs) and DLBug (logic bugs unique to the distributed mode). 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. One DBMS, Two Modes, and a Bunch of Bugs: Catching Logic Bugs in Distributed DBMSs via Differential Testing 59:17 Testing coverage. DistSQL achieves substantially broader testing coverage than DistSQL Random and DistSQL Text, discovering 83% and 49% more query plan texts, and extractin",
+      "section": "7.3 Sensitivity Analysis",
+      "page": 16,
+      "char_offset": 58621,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "4 Comparison We evaluate five open-source distributed DBMSs (Table 3) using four existing tools: Jepsen [ 20], Mallory [36], QPG [2], and EET [24], and compare their performance with DistSQL.",
+      "context_before": "Random andDistSQL Textdetect 15 and 10 fewer logic bugs, and 10 and 6 fewer DLBugs, respectively. These results demonstrate the effectiveness of query plan feature for capturing execution behavior, reducing redundancy, and enhancing both bug detection and testing coverage. The additional variant DistSQL JoinAware reports the same set of bugs as DistSQL. Although it may explore more join combinations, this does not necessarily reveal new bugs. Its uneven coverage across DBMSs indicates that it may sometimes overemphasize fine-grained differences and slightly reduce coverage in other aspects. 7.",
+      "context_after": "These tools are selected from existing open-source approaches to ensure a representative and meaningful comparison. Among distributed system testing approaches, Jepsen and Mallory are the most closely related to DBMS testing. Among recent DBMS testing tools, QPG employs feedback mechanism related to query plan, while EET achieves higher SQL diversity than many prior approaches [ 24]. Since none of the tools support all DBMSs, we extend them for compatibility. Each system is tested for 24 hours, with results summarized in Table 7. Testing coverage. Our system achieves significantly higher cove",
+      "section": "7.4 Comparison",
+      "page": 17,
+      "char_offset": 61219,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Among recent DBMS testing tools, QPG employs feedback mechanism related to query plan, while EET achieves higher SQL diversity than many prior approaches [ 24].",
+      "context_before": "necessarily reveal new bugs. Its uneven coverage across DBMSs indicates that it may sometimes overemphasize fine-grained differences and slightly reduce coverage in other aspects. 7.4 Comparison We evaluate five open-source distributed DBMSs (Table 3) using four existing tools: Jepsen [ 20], Mallory [36], QPG [2], and EET [24], and compare their performance with DistSQL. These tools are selected from existing open-source approaches to ensure a representative and meaningful comparison. Among distributed system testing approaches, Jepsen and Mallory are the most closely related to DBMS testing.",
+      "context_after": "Since none of the tools support all DBMSs, we extend them for compatibility. Each system is tested for 24 hours, with results summarized in Table 7. Testing coverage. Our system achieves significantly higher coverage, with query plan text coverage 197 times that of EET and query plan feature coverage 3.5 times that of EET. It also surpasses Jepsen, Mallory, and QPG by orders of magnitude in both metrics. These results underscore DistSQL ’s effectiveness in thoroughly exploring a wide range of execution paths within the vast execution space of distributed DBMSs. DistSQL ’s higher testing cover",
+      "section": "7.4 Comparison",
+      "page": 17,
+      "char_offset": 61636,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "It also surpasses Jepsen, Mallory, and QPG by orders of magnitude in both metrics.",
+      "context_before": "son. Among distributed system testing approaches, Jepsen and Mallory are the most closely related to DBMS testing. Among recent DBMS testing tools, QPG employs feedback mechanism related to query plan, while EET achieves higher SQL diversity than many prior approaches [ 24]. Since none of the tools support all DBMSs, we extend them for compatibility. Each system is tested for 24 hours, with results summarized in Table 7. Testing coverage. Our system achieves significantly higher coverage, with query plan text coverage 197 times that of EET and query plan feature coverage 3.5 times that of EET.",
+      "context_after": "These results underscore DistSQL ’s effectiveness in thoroughly exploring a wide range of execution paths within the vast execution space of distributed DBMSs. DistSQL ’s higher testing coverage comes from two primary factors. First, its distributed diversity oriented database state mutation actively mutates database states to trigger more execution paths. Second, its distributed interaction guided execution space exploration mechanism effectively identifies unique distributed logic, avoiding redundant test cases and enabling efficient exploration of the explosive execution space. Figure 4(c)",
+      "section": "7.4 Comparison",
+      "page": 17,
+      "char_offset": 62122,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "DistSQL identifies 28 more bugs than QPG and 31 more than EET.",
+      "context_before": "its distributed diversity oriented database state mutation actively mutates database states to trigger more execution paths. Second, its distributed interaction guided execution space exploration mechanism effectively identifies unique distributed logic, avoiding redundant test cases and enabling efficient exploration of the explosive execution space. Figure 4(c) illustrates the rapid growth of covered query plan features in early testing phases, stabilizing later. Across all stages, DistSQL maintains superior feature coverage. Bug detection. Jepsen and Mallory fail to find any new logic bugs.",
+      "context_after": "Notably, 16 of these bugs are specific to distributed logic and do not manifest in centralized mode, highlighting the importance of targeting distributed execution paths. While Jepsen and Mallory are effective at finding certain types of bugs, many of the bugs they found have already been fixed in current versions. In contrast, DistSQL is able to find additional logic bugs beyond those detected by existing approaches. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. 59:18 Zi-Xuan Fu, Jia-Ju Bai, Hong-Bo Feng, and Kang Chen 7.5 Case Studies of the Found Bugs This section highli",
+      "section": "7.4 Comparison",
+      "page": 17,
+      "char_offset": 63040,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[ 2,18,\n23,24,30,43–45,48,53,61]",
+      "technique": "qpg",
+      "sentence": "Other approaches [ 2,18, 23,24,30,43–45,48,53,61] aim to find logic bugs by validating the query processing logic in local DBMSs through test oracles, which typically compare the execution results of two logically equivalent or related SQL queries.",
+      "context_before": "strategies guided by memory accesses and unexplored branches. Zest [ 38] enhances input generation by transforming random-input generators into deterministic parametric generators, effectively mapping low-level mutations to high-level structural changes. However, these approaches struggle to cope with the complex semantics and execution behaviors of distributed DBMSs, and their feedback is often insufficient. 9.2 DBMS Fuzzing Fuzzing has shown strong potential in finding various types of bugs in DBMSs. Several approaches [ 1, 13,22,28,65] focus primarily on memory bugs and assertion failures.",
+      "context_after": "In addition, SemConT [ 31] introduces a formal model of the database to provide a ground truth for validation. Other approaches [ 3,29,49] adopt differential testing strategies, executing the same SQL statements under different runtime configurations or query hints to identify inconsistencies. Most of these approaches rely heavily on random testing, providing limited guidance during SQL generation and database state mutation. Some approaches incorporate testing-scenario-specific 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. One DBMS, Two Modes, and a Bunch of Bugs: Catching",
+      "section": "9.2 DBMS Fuzzing",
+      "page": 20,
+      "char_offset": 72867,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. 59:22 Zi-Xuan Fu, Jia-Ju Bai, Hong-Bo Feng, and Kang ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[ 3,29,49]",
+      "technique": "dqp",
+      "sentence": "Other approaches [ 3,29,49] adopt differential testing strategies, executing the same SQL statements under different runtime configurations or query hints to identify inconsistencies.",
+      "context_before": "uted DBMSs, and their feedback is often insufficient. 9.2 DBMS Fuzzing Fuzzing has shown strong potential in finding various types of bugs in DBMSs. Several approaches [ 1, 13,22,28,65] focus primarily on memory bugs and assertion failures. Other approaches [ 2,18, 23,24,30,43–45,48,53,61] aim to find logic bugs by validating the query processing logic in local DBMSs through test oracles, which typically compare the execution results of two logically equivalent or related SQL queries. In addition, SemConT [ 31] introduces a formal model of the database to provide a ground truth for validation.",
+      "context_after": "Most of these approaches rely heavily on random testing, providing limited guidance during SQL generation and database state mutation. Some approaches incorporate testing-scenario-specific 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. One DBMS, Two Modes, and a Bunch of Bugs: Catching Logic Bugs in Distributed DBMSs via Differential Testing 59:21 feedback signals, such as formal rule coverage in SemConT [ 31] and graph similarity of test cases in TQS [ 53], while others adopt more generic runtime feedback, such as branch coverage [ 30] or query plan text [ 2]. However, thes",
+      "section": "9.2 DBMS Fuzzing",
+      "page": 20,
+      "char_offset": 73227,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. Proceedings of the ACM on Management of Data 2, 3 (2024), 1–26.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "citation_marker",
+      "surface": "[ 2]",
+      "technique": "qpg",
+      "sentence": "One DBMS, Two Modes, and a Bunch of Bugs: Catching Logic Bugs in Distributed DBMSs via Differential Testing 59:21 feedback signals, such as formal rule coverage in SemConT [ 31] and graph similarity of test cases in TQS [ 53], while others adopt more generic runtime feedback, such as branch coverage [ 30] or query plan text [ 2].",
+      "context_before": "of two logically equivalent or related SQL queries. In addition, SemConT [ 31] introduces a formal model of the database to provide a ground truth for validation. Other approaches [ 3,29,49] adopt differential testing strategies, executing the same SQL statements under different runtime configurations or query hints to identify inconsistencies. Most of these approaches rely heavily on random testing, providing limited guidance during SQL generation and database state mutation. Some approaches incorporate testing-scenario-specific 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026.",
+      "context_after": "However, these feedback signals are often insufficient for thoroughly exercising the complex behaviors of distributed DBMSs. Consequently, existing approaches fall short when applied to distributed DBMSs, as they largely overlook the unique characteristics of distributed execution. To address these challenges, DistSQL is specifically designed to more effectively explore distributed behaviors and find logic bugs that manifest in distributed execution scenarios. 9.3 Distributed System Fuzzing Many approaches have been proposed to test generic distributed systems. Some adopt formal verification",
+      "section": "9.2 DBMS Fuzzing",
+      "page": 20,
+      "char_offset": 73665,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071. 4, No. 1 (SIGMOD), Article 59. Publication date: February 2026. 59:22 Zi-Xuan Fu, Jia-Ju Bai, Hong-Bo Feng, and Kang ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M10",
+        "M15"
+      ]
+    },
+    "suppressed": [
+      {
+        "mention_id": "M12",
+        "pattern": "compares_with",
+        "suppressed_because": "the sentence is framed as related work, so it describes somebody else's approach"
+      }
+    ]
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:50:48Z",
+    "is_model_written": true,
+    "summary": "DistSQL tests distributed DBMSs by running the same query in the system's distributed and single-node modes and comparing the results, so a difference between the two modes is a bug without needing to know the right answer. Its feedback signal is drawn from query plan features rather than plan text, which the paper argues is necessary because a distributed plan's text is noisy. Across five open-source distributed DBMSs it found substantially more bugs than the existing tools.",
+    "narrative": "QPG is DistSQL's closest comparison and its point of departure. The paper describes QPG as improving SQLancer by using query plan text as feedback to mutate the database state when no new plans appear, and argues this focuses on centralized database state and so becomes noisy and less informative on distributed systems. That claim is tested directly: one of DistSQL's own ablation variants uses query-plan-text feedback identical to QPG's, and QPG is also run as an external baseline alongside Jepsen, Mallory and EET, with DistSQL finding 28 more bugs. NoREC, TLP, PQS and CODDTest appear as the local-DBMS oracles the distributed setting is contrasted with.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "baseline",
+      "M4": "definition",
+      "M5": "definition",
+      "M6": "definition",
+      "M7": "motivation",
+      "M8": "background",
+      "M9": "definition",
+      "M10": "state_of_the_art",
+      "M11": "baseline",
+      "M12": "baseline",
+      "M13": "definition",
+      "M14": "result_comparison",
+      "M15": "result_comparison",
+      "M16": "background",
+      "M17": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M8 records an AST-based SQL generator of the authors' own. QPG is run as a baseline and reimplemented as an ablation variant, neither of which makes DistSQL built on SQLancer."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "DistSQL replaces plan-text feedback with plan-feature feedback, which M7 presents as a correction of QPG's behaviour on distributed systems rather than a generalisation of it, and its oracle is differential execution across two modes."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M10",
+          "M12",
+          "M15"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M10",
+            "sentence": "We compare DistSQL against SOTA database and distributed system testing tools, including Squirrel [ 65], QPG [ 2], and EET [ 24], as well as distributed system testing tools, including Jepsen [20] and Mallory [36].",
+            "section": "7.1 Experimental Setup",
+            "page": 14
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "4 Comparison We evaluate five open-source distributed DBMSs (Table 3) using four existing tools: Jepsen [ 20], Mallory [36], QPG [2], and EET [24], and compare their performance with DistSQL.",
+            "section": "7.4 Comparison",
+            "page": 17
+          },
+          {
+            "mention_id": "M15",
+            "sentence": "DistSQL identifies 28 more bugs than QPG and 31 more than EET.",
+            "section": "7.4 Comparison",
+            "page": 17
+          }
+        ],
+        "reasoning": "M10 lists QPG among the state-of-the-art tools DistSQL is compared against, M12 records it as one of four tools run on five distributed DBMSs, and M15 gives the result: 28 more bugs than QPG.",
+        "techniques": [
+          "qpg"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M10"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M10",
+            "sentence": "We compare DistSQL against SOTA database and distributed system testing tools, including Squirrel [ 65], QPG [ 2], and EET [ 24], as well as distributed system testing tools, including Jepsen [20] and Mallory [36].",
+            "section": "7.1 Experimental Setup",
+            "page": 14
+          }
+        ],
+        "reasoning": "M10 describes QPG as one of the SOTA database and distributed system testing tools DistSQL is measured against."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3786699.json b/_data/papers/paper_doi_10_1145_3786699.json
new file mode 100644
index 0000000..1deaa95
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3786699.json
@@ -0,0 +1,467 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:25:08Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3786699",
+    "title": "SQLBarber: A System Leveraging Large Language Models to Generate Customized and Realistic SQL Workloads",
+    "authors": [
+      "Jiale Lao",
+      "Immanuel Trummer"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": "10.1145/3786699",
+    "arxiv_id": "2507.06192",
+    "s2_paper_id": "995979576a9a43c33d87711cc7ec1757a5f1a280",
+    "url": "https://doi.org/10.1145/3786699",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2507.06192",
+      "retrieved_at": "2026-09-08T15:25:08Z",
+      "chars": 95814,
+      "content_sha256": "sha256:e8d0348c099aa8883158cf07572099c55e54ec9fa8b4a484fcb7127af915158c"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 29,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2199
+      },
+      {
+        "number": "2",
+        "title": "Related Work",
+        "start": 12065
+      },
+      {
+        "number": "3",
+        "title": "Problem Formulation",
+        "start": 16132
+      },
+      {
+        "number": "4",
+        "title": "System Overview",
+        "start": 24792
+      },
+      {
+        "number": "5",
+        "title": "Customized SQL Template Generator",
+        "start": 27681
+      },
+      {
+        "number": "6",
+        "title": "Cost-Aware Query Generator",
+        "start": 34263
+      },
+      {
+        "number": "6.1",
+        "title": "Template Profiling via Strategic Sampling",
+        "start": 34970
+      },
+      {
+        "number": "6.2",
+        "title": "Adaptive Template Refinement and Pruning",
+        "start": 37593
+      },
+      {
+        "number": "6.3",
+        "title": "Predicate Search via Bayesian Optimization",
+        "start": 46371
+      },
+      {
+        "number": "7",
+        "title": "Experimental Evaluation",
+        "start": 52817
+      },
+      {
+        "number": "7.1",
+        "title": "Experimental Setup",
+        "start": 52843
+      },
+      {
+        "number": "7.2",
+        "title": "Performance Study",
+        "start": 59340
+      },
+      {
+        "number": "7.3",
+        "title": "Scalability Study",
+        "start": 64956
+      },
+      {
+        "number": "7.4",
+        "title": "Ablation Study",
+        "start": 67483
+      },
+      {
+        "number": "7.5",
+        "title": "Further Analysis",
+        "start": 69652
+      },
+      {
+        "number": "8",
+        "title": "Conclusions and Outlook",
+        "start": 78110
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2020. A. Seltenreich. Sqlsmith. https://github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Yushi Bai, Xin Lv, Jiajie Zhang, Hongchang Lyu, Jiankai Tang, Zhidian Huang, Zhengxiao Du, Xiao Liu, Aohan Zeng, Lei Hou, Yuxiao Dong, Jie Tang, and Juanzi Li. 2024. LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding. InProceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Lun-Wei Ku, Andre Martins, and Vivek Srikum",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. 2006. Generating queries with cardinality constraints for dbms testing.IEEE Transactions on Knowledge and Data Engineering18, 12 (2006), 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Victor Giannakouris and Immanuel Trummer. 2024. Demonstrating lambda-Tune: Exploiting Large Language Models for Workload-Adaptive Database System Tuning. InCompanion of the 2024 International Conference on Management of Data(Santiago AA, Chile)(SIGMOD/PODS ’24). Association for Computing Machinery, New York, NY, USA, 508–511. doi:10.1145/3626246.3654751",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, and Ting Liu. 2025. A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions.ACM Trans. Inf. Syst.43, 2, Article 42 (Jan. 2025), 55 pages. doi:10.1145/ 3703155",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Xinmei Huang, Haoyang Li, Jing Zhang, Xinxin Zhao, Zhiming Yao, Yiyan Li, Tieying Zhang, Jianjun Chen, Hong Chen, and Cuiping Li. 2025. E2ETune: End-to-End Knob Tuning via Fine-tuned Generative Language Model. arXiv:2404.11581 [cs.AI] https://arxiv.org/abs/2404.11581",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Saehan Jo and Immanuel Trummer. 2024. ThalamusDB: Approximate Query Processing on Multi-Modal Data. doi:10.1145/3654989",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Skander Krid, Mihail Stoian, and Andreas Kipf. 2025. Redbench: A Benchmark Reflecting Real Workloads.arXiv preprint arXiv:2506.12488(2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Jiale Lao and Immanuel Trummer. 2025. Demonstrating SQLBarber: Leveraging Large Language Models to Generate Customized and Realistic SQL Workloads. InCompanion of the 2025 International Conference on Management of Data(Berlin, Germany)(SIGMOD/PODS ’25). Association for Computing Machinery, New York, NY, USA, 151–154. doi:10.1145/3722212.3725101",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jiale Lao, Yibo Wang, Yufei Li, Jianping Wang, Yunjia Zhang, Zhiyuan Cheng, Wanghu Chen, Mingjie Tang, and Jianguo Wang. 2024. GPTuner: A Manual-Reading Database Tuning System via GPT-Guided Bayesian Optimization. Proc. VLDB Endow.17, 8 (may 2024), 1939–1952. doi:10.14778/3659437.3659449",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jiale Lao, Yibo Wang, Yufei Li, Jianping Wang, Yunjia Zhang, Zhiyuan Cheng, Wanghu Chen, Mingjie Tang, and Jianguo Wang. 2025. GPTuner: An LLM-Based Database Tuning System.SIGMOD Rec.54, 1 (April 2025), 101–110. doi:10.1145/3733620.3733641",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jiale Lao, Yibo Wang, Yufei Li, Jianping Wang, Yunjia Zhang, Zhiyuan Cheng, Wanghu Chen, Yuanchun Zhou, Mingjie Tang, and Jianguo Wang. 2024. A Demonstration of GPTuner: AGPT-Based Manual-Reading Database Tuning System. InCompanion of the 2024 International Conference on Management of Data(Santiago, Chile)(SIGMOD ’24). Association for Computing Machinery, New York, NY, USA, 4 pages. doi:10.1145/36",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Jiale Lao, Andreas Zimmerer, Olga Ovcharenko, Tianji Cong, Matthew Russo, Gerardo Vitagliano, Michael Cochez, Fatma Özcan, Gautam Gupta, Thibaud Hottelier, H. V. Jagadish, Kris Kissel, Sebastian Schelter, Andreas Kipf, and Immanuel Trummer. 2025. SemBench: A Benchmark for Semantic Query Processing Engines. arXiv:2511.01716 [cs.DB] https://arxiv.org/abs/2511.01716",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2015. How good are query optimizers, really?Proc. VLDB Endow.9, 3 (Nov. 2015), 204–215. doi:10.14778/2850583.2850594",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Haoyang Li, Jing Zhang, Hanbing Liu, Ju Fan, Xiaokang Zhang, Jun Zhu, Renjie Wei, Hongyan Pan, Cuiping Li, and Hong Chen. 2024. CodeS: Towards Building Open-source Language Models for Text-to-SQL. doi:10.1145/3654930",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Yiyan Li, Haoyang Li, Zhao Pu, Jing Zhang, Xinyi Zhang, Tao Ji, Luming Sun, Cuiping Li, and Hong Chen. 2024. Is Large Language Model Good at Database Knob Tuning? A Comprehensive Experimental Evaluation. arXiv:2408.02213 [cs.DB] https://arxiv.org/abs/2408.02213",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Zhaodonghui Li, Haitao Yuan, Huiming Wang, Gao Cong, and Lidong Bing. 2024. LLM-R2: A Large Language Model Enhanced Rule-Based Rewrite System for Boosting Query Efficiency.Proc. VLDB Endow.18, 1 (Sept. 2024), 53–65. doi:10.14778/3696435.3696440",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Meiyu Lin, Haichuan Zhang, Jiale Lao, Renyuan Li, Yuanchun Zhou, Carl Yang, Yang Cao, and Mingjie Tang. 2025. ToxicSQL: Migrating SQL Injection Threats into Text-to-SQL Models via Backdoor Attack. arXiv:2503.05445 [cs.CR] 3, No. 6 (SIGMOD), Article XXX. Publication date: December 2025. XXX:26 Jiale Lao and Immanuel Trummer https://arxiv.org/abs/2503.05445",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Marius Lindauer, Katharina Eggensperger, Matthias Feurer, André Biedenkapp, Difan Deng, Carolin Benjamins, Tim Ruhkopf, René Sass, and Frank Hutter. 2022. SMAC3: A Versatile Bayesian Optimization Package for Hyperparameter Optimization.Journal of Machine Learning Research23, 54 (2022), 1–9. http://jmlr.org/papers/v23/21-0888.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Marius Lindauer, Katharina Eggensperger, Matthias Feurer, André Biedenkapp, Difan Deng, Carolin Benjamins, Tim Ruhkopf, René Sass, and Frank Hutter. 2022. SMAC3: A Versatile Bayesian Optimization Package for Hyperparameter Optimization.Journal of Machine Learning Research23, 54 (2022), 1–9. http://jmlr.org/papers/v23/21-0888.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Chunwei Liu, Matthew Russo, Michael Cafarella, Lei Cao, Peter Baile Chen, Zui Chen, Michael Franklin, Tim Kraska, Samuel Madden, Rana Shahout, et al .2025. Palimpzest: Optimizing ai-powered analytics with declarative query processing. InProceedings of the Conference on Innovative Database Research (CIDR). 2.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Wei-Liem Loh. 1996. On Latin hypercube sampling.The annals of statistics24, 5 (1996), 2058–2080.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. 2008. Generating targeted queries for database testing. In Proceedings of the 2008 ACMSIGMOD International Conference on Management of Data(Vancouver, Canada)(SIGMOD ’08). Association for Computing Machinery, New York, NY, USA, 499–510. doi:10.1145/1376616.1376668",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Parimarjan Negi, Laurent Bindschaedler, Mohammad Alizadeh, Tim Kraska, Jyoti Leeka, Anja Gruenheid, and Matteo Interlandi. 2023. Unshackling Database Benchmarking from Synthetic Workloads. In2023 IEEE 39th International Conference on Data Engineering (ICDE). 3659–3662. doi:10.1109/ICDE55515.2023.00292",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Victor M Panaretos and Yoav Zemel. 2019. Statistical aspects of Wasserstein distances.Annual review of statistics and its application6, 1 (2019), 405–431.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Victor M Panaretos and Yoav Zemel. 2019. Statistical aspects of Wasserstein distances.Annual review of statistics and its application6, 1 (2019), 405–431.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Tobias Schmidt, Viktor Leis, Peter Boncz, and Thomas Neumann. 2025. SQLStorm: Taking Database Benchmarking into the LLM Era.Proc. VLDB Endow.18, 11 (Sept. 2025), 4144–4157. doi:10.14778/3749646.3749683",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Yuyang Song, Hanxu Yan, Jiale Lao, Yibo Wang, Yufei Li, Yuanchun Zhou, Jianguo Wang, and Mingjie Tang. 2025. QUITE: A Query Rewrite System Beyond Rules with LLM Agents. arXiv:2506.07675 [cs.DB] https://arxiv.org/abs/2506.07675",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Transaction Processing Performance Council (TPC). 2014.TPC BenchmarkTMH (Decision Support) Standard Specification, Revision 2.17.1. Technical Report. Transaction Processing Performance Council, San Francisco, CA, USA. https: //www.tpc.org/tpch/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Immanuel Trummer. 2022. CodexDB: synthesizing code for query processing from natural language instructions using GPT-3 codex.Proc. VLDB Endow.15, 11 (July 2022), 2921–2928. doi:10.14778/3551793.3551841",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Immanuel Trummer. 2022. DB-BERT: A Database Tuning Tool That \"Reads the Manual\". InProceedings of the 2022 International Conference on Management of Data(Philadelphia, PA, USA)(SIGMOD ’22). Association for Computing Machinery, New York, NY, USA, 190–203. doi:10.1145/3514221.3517843",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Matthias Urban and Carsten Binnig. 2024. Demonstrating CAESURA: Language Models as Multi-Modal Query Planners. InCompanion of the 2024 International Conference on Management of Data(Santiago AA, Chile)(SIGMOD ’24). Association for Computing Machinery, New York, NY, USA, 472–475. doi:10.1145/3626246.3654732",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Alexander van Renen, Dominik Horn, Pascal Pfeil, Kapil Vaidya, Wenjian Dong, Murali Narayanaswamy, Zhengchun Liu, Gaurav Saxena, Andreas Kipf, and Tim Kraska. 2024. Why TPC is Not Enough: An Analysis of the Amazon Redshift Fleet.Proc. VLDB Endow.17, 11 (aug 2024), 3694–3706. doi:10.14778/3681954.3682031",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Alexander van Renen and Viktor Leis. 2023. Cloud Analytics Benchmark.Proc. VLDB Endow.16, 6 (Feb. 2023), 1413–1425. doi:10.14778/3583140.3583156",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Adrian Vogelsgesang, Michael Haubenschild, Jan Finis, Alfons Kemper, Viktor Leis, Tobias Muehlbauer, Thomas Neumann, and Manuel Then. 2018. Get Real: How Benchmarks Fail to Represent the Real World. InProceedings of the Workshop on Testing Database Systems(Houston, TX, USA)(DBTest ’18). Association for Computing Machinery, New York, NY, USA, Article 1, 6 pages. doi:10.1145/3209950.3209952",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Midhul Vuppalapati, Justin Miron, Rachit Agarwal, Dan Truong, Ashish Motivala, and Thierry Cruanes. 2020. Building An Elastic Query Engine on Disaggregated Storage. In17th USENIX Symposium on Networked Systems Design and Implementation (NSDI 20). USENIX Association, Santa Clara, CA, 449–462. https://www.usenix.org/conference/nsdi20/ presentation/vuppalapati",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Jiaqi Wang, Tianyi Li, Anni Wang, Xiaoze Liu, Lu Chen, Jie Chen, Jianye Liu, Junyang Wu, Feifei Li, and Yunjun Gao. 2023. Real-Time Workload Pattern Analysis for Large-Scale Cloud Databases.Proc. VLDB Endow.16, 12 (Aug. 2023), 3689–3701. doi:10.14778/3611540.3611557",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Qingshuai Wang, Yuming Li, Rong Zhang, Ke Shu, Zhenjie Zhang, and Aoying Zhou. 2023. A Scalable Query-Aware Enormous Database Generator for Database Evaluation.IEEE Trans. on Knowl. and Data Eng.35, 5 (May 2023), 4395–4410. doi:10.1109/TKDE.2022.3153651",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Ziwei Xu, Sanjay Jain, and Mohan Kankanhalli. 2025. Hallucination is Inevitable: An Innate Limitation of Large Language Models. arXiv:2401.11817 [cs.CL] https://arxiv.org/abs/2401.11817 3, No. 6 (SIGMOD), Article XXX. Publication date: December 2025. SQLBarber: A System Leveraging Large Language Models to Generate Customized and Realistic SQL Workloads XXX:27",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Jingyi Yang, Peizhi Wu, Gao Cong, Tieying Zhang, and Xiao He. 2022. SAM: Database Generation from Query Workloads with Supervised Autoregressive Models. InProceedings of the 2022 International Conference on Management of Data(Philadelphia, PA, USA)(SIGMOD ’22). Association for Computing Machinery, New York, NY, USA, 1542–1555. doi:10.1145/3514221.3526168",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Geoffrey X. Yu, Ziniu Wu, Ferdi Kossmann, Tianyu Li, Markos Markakis, Amadou Ngom, Samuel Madden, and Tim Kraska. 2024. Blueprinting the Cloud: Unifying and Automatically Optimizing Cloud Data Infrastructures with BRAD. Proc. VLDB Endow.17, 11 (July 2024), 3629–3643. doi:10.14778/3681954.3682026",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Lixi Zhang, Chengliang Chai, Xuanhe Zhou, and Guoliang Li. 2022. LearnedSQLGen: Constraint-aware SQL Generation using Reinforcement Learning. InProceedings of the 2022 International Conference on Management of Data(Philadelphia, PA, USA)(SIGMOD ’22). Association for Computing Machinery, New York, NY, USA, 945–958. doi:10.1145/3514221. 3526155 3, No. 6 (SIGMOD), Article XXX. Publication date: Decem",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLSmith [ 1], SQLancer [ 2], and SQLStorm [ 28] focus on generating diverse queries for testing and debugging.",
+      "context_before": "haracteristics are shared. These typically include: (1) properties of SQL templates (e.g., the number of accessed tables and the distributions of query operators such as joins and aggregations), and (2) cost distributions of instantiated queries (e.g., the distributions of cardinalities, execution times, and CPU usages). This raises an important research direction: transforming high-level workload information into realistic and executable benchmarks. Recent methods have been proposed for generating SQL queries, but they struggle to produce customized queries that satisfy realistic constraints.",
+      "context_after": "However, these methods do not allow Authors’ Contact Information: Jiale Lao, Cornell University, Ithaca, New York, USA, jiale@cs.cornell.edu; Immanuel Trummer, Cornell University, Ithaca, New York, USA, itrummer@cornell.edu. Request permissions from permissions@acm.org. ©2025 Copyright held by the owner/author(s). Publication rights licensed to ACM. ACM 2836-6573/2025/12-ARTXXX https://doi.org/10.1145/XXXXXXX 3, No. 6 (SIGMOD), Article XXX. Publication date: December 2025.arXiv:2507.06192v2 [cs.DB] 2 Dec 2025 XXX:2 Jiale Lao and Immanuel Trummer users to control SQL templates or query costs",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3093,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 1,2,28]",
+      "technique": "qpg",
+      "sentence": "SQLBarber relates to prior work on generating diverse SQL queries for testing and debugging [ 1,2,28].",
+      "context_before": "baselines. The remainder of this paper is organized as follows. We discuss related work in Section 2. We formally define the SQL workload generation problem in Section 3 and give an overview of SQLBarber in Section 4. Section 5 describes the customized SQL template generator, and Section 6 discusses the cost-aware query generator. We conduct extensive experiments to evaluate the performance, scalability, and robustness of SQLBarber with an ablation study, a cost analysis, and extensive further analysis in Section 7. We conclude in Section 8. 2 Related Work Diversity-Oriented Query Generation.",
+      "context_after": "SQLSmith [ 1] recursively and randomly expands 3, No. 6 (SIGMOD), Article XXX. Publication date: December 2025. XXX:4 Jiale Lao and Immanuel Trummer grammar rules while ensuring schema and type consistency. SQLancer [ 2] employs a feedbackguided test generation approach that aims to produce diverse query execution plans, based on the idea that exploring a wide range of plans helps reveal bugs. SQLStorm [ 28] directly prompts LLMs to generate SQL queries with a high temperature to increase diversity. While the aforementioned methods emphasize query diversity for testing without constraints on",
+      "section": "2 Related Work",
+      "page": 3,
+      "char_offset": 12116,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE). IEEE, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [ 2] employs a feedbackguided test generation approach that aims to produce diverse query execution plans, based on the idea that exploring a wide range of plans helps reveal bugs.",
+      "context_before": "-aware query generator. We conduct extensive experiments to evaluate the performance, scalability, and robustness of SQLBarber with an ablation study, a cost analysis, and extensive further analysis in Section 7. We conclude in Section 8. 2 Related Work Diversity-Oriented Query Generation.SQLBarber relates to prior work on generating diverse SQL queries for testing and debugging [ 1,2,28]. SQLSmith [ 1] recursively and randomly expands 3, No. 6 (SIGMOD), Article XXX. Publication date: December 2025. XXX:4 Jiale Lao and Immanuel Trummer grammar rules while ensuring schema and type consistency.",
+      "context_after": "SQLStorm [ 28] directly prompts LLMs to generate SQL queries with a high temperature to increase diversity. While the aforementioned methods emphasize query diversity for testing without constraints on SQL templates or query costs, SQLBarber integrates constraints on both templates and costs to turn high-level workload information into realistic, executable benchmarks. Constraint-Aware Query Generation.SQLBarber is also related to prior work on cost-constrained SQL query generation [ 4,24,43]. Prior work [ 4,24] uses the hill climbing algorithm to instantiate given SQL templates by adjusting",
+      "section": "2 Related Work",
+      "page": 4,
+      "char_offset": 12426,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:19:46Z",
+    "is_model_written": true,
+    "summary": "SQLBarber generates customised, realistic SQL workloads with large language models. Real queries are hard to obtain for privacy reasons and existing generators offer little customisation or ability to meet realistic constraints. SQLBarber provides a declarative interface, removes the need to write SQL templates by hand while accepting natural-language constraints, and adds a self-correction pipeline that profiles, refines and prunes templates by query cost. The authors open-source ten benchmarks built from Snowflake and Amazon Redshift statistics.",
+    "narrative": "SQLancer is cited as prior work on generating diverse queries for testing and debugging, with its query plan guidance described as feedback-guided generation aiming at diverse execution plans. SQLBarber's goal is realistic workloads rather than bug finding, so the two are related but not compared.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3798226.json b/_data/papers/paper_doi_10_1145_3798226.json
new file mode 100644
index 0000000..5f410df
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3798226.json
@@ -0,0 +1,589 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:26Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3798226",
+    "title": "Metamorphic Testing for Infrastructure-as-Code Engines",
+    "authors": [
+      "David Spielmann",
+      "George Zakhour",
+      "Dominik Arnold",
+      "Matteo Biagiola",
+      "Roland Meier",
+      "Guido Salvaneschi"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Programming Languages",
+    "doi": "10.1145/3798226",
+    "arxiv_id": null,
+    "s2_paper_id": "3a205ccef9fb1acd547d866fa36a98c549d88344",
+    "url": "https://doi.org/10.1145/3798226",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3798226",
+      "retrieved_at": "2026-09-09T01:29:26Z",
+      "chars": 113311,
+      "content_sha256": "sha256:02c06451199c9725a74d59421c3f0bc52d61d49aacd32d397bd105f08e2250b5"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 29,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2653
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 14076
+      },
+      {
+        "number": "2.1",
+        "title": "Metamorphic Testing",
+        "start": 14244
+      },
+      {
+        "number": "2.2",
+        "title": "Infrastructure as Code",
+        "start": 16829
+      },
+      {
+        "number": "2.3",
+        "title": "E-Graphs",
+        "start": 21134
+      },
+      {
+        "number": "4.1",
+        "title": "Design Goals and Key Insights",
+        "start": 29442
+      },
+      {
+        "number": "4.2",
+        "title": "Overview of EMIaC",
+        "start": 33710
+      },
+      {
+        "number": "4.3",
+        "title": "Test Input Generation",
+        "start": 34409
+      },
+      {
+        "number": "4.4",
+        "title": "Metamorphic Transformation via E-Graphs",
+        "start": 37991
+      },
+      {
+        "number": "4.5",
+        "title": "Test Execution and Oracles",
+        "start": 49330
+      },
+      {
+        "number": "4.6",
+        "title": "Formalism",
+        "start": 51004
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Jon Ayerdi, Valerio Terragni, Gunel Jahangirova, Aitor Arrieta, and Paolo Tonella. 2024. GenMorph: Automatically Generating Metamorphic Relations via Genetic Programming. IEEE Trans. Software Eng. 50, 7 (2024), 1888–1900. doi:10.1109/TSE.2024.3407840",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Haniel Barbosa, Clark Barrett, Martin Brain, Gereon Kremer, Hanna Lachnitt, Makai Mann, Abdalrhman Mohamed, Mudathir Mohamed, Aina Niemetz, Andres Nötzli, et al .2022. cvc5: A versatile and industrial-strength SMT solver. In International Conference on Tools and Algorithms for the Construction and Analysis of Systems. Springer, 415–442.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Earl T. Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. 2015. The Oracle Problem in Software Testing: A Survey. IEEE Trans. Software Eng. 41, 5 (2015), 507–525. doi:10.1109/TSE.2014.2372785",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Roberto Bruttomesso, Edgar Pek, Natasha Sharygina, and Aliaksei Tsitovich. 2010. The opensmt solver. In International Conference on TOOLS and Algorithms for the Construction and Analysis of Systems. Springer, 150–153.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Christoph Bühler, David Spielmann, Roland Meier, and Guido Salvaneschi. 2025. TerraDS: A Dataset for Terraform HCL Programs. In Proceedings of the IEEE/ACM 22nd International Conference on Mining Software Repositories (MSR) (Ottawa, Canada) (MSR ’25). IEEE, 5 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Cristian Cadar, Daniel Dunbar, and Dawson Engler. 2008. KLEE: unassisted and automatic generation of high-coverage tests for complex systems programs. In Proceedings of the 8th USENIX Conference on Operating Systems Design and Implementation (San Diego, California) (OSDI’08). USENIX Association, USA, 209–224.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "David Cao, Rose Kunkel, Chandrakana Nandi, Max Willsey, Zachary Tatlock, and Nadia Polikarpova. 2023. babble: Learning Better Abstractions with E-Graphs and Anti-unification. Proc. ACM Program. Lang. 7, POPL, Article 14 (Jan. 2023), 29 pages. doi:10.1145/3571207",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Nazanin Bayati Chaleshtari, Fabrizio Pastore, Arda Goknil, and Lionel C. Briand. 2023. Metamorphic Testing for Web System Security. IEEE Trans. Software Eng. 49, 6 (2023), 3430–3471. doi:10.1109/TSE.2023.3256322",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Checkov by Prisma Cloud. 2025. Checkov. https://www.checkov.io/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "T.Y. Chen, S.C. Cheung, and S.M. Yiu. 1998. Metamorphic testing: A new approach for generating next test cases. Technical Report HKUST-CS98-01. Department of Computer Science, The Hong Kong University of Science and Technology, Hong Kong. https://arxiv.org/abs/2002.12543 Technical Report, arXiv:2002.12543.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Henry Coles, Thomas Laurent, Christopher Henard, Mike Papadakis, and Anthony Ventresque. 2016. PIT: a practical mutation testing tool for Java (demo). In Proceedings of the 25th International Symposium on Software Testing and Analysis (Saarbrücken, Germany) (ISSTA 2016). Association for Computing Machinery, New York, NY, USA, 449–452. doi:10.1145/2931037.2948707",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Pulumi Corporation. 2023. Resources deleted for no apparent reason. GitHub issue #2593, pulumi/pulumi-aws. https://github.com/pulumi/pulumi-aws/issues/2593",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Progress Software Corporation. 2025. Chef Documentation. https://docs.chef.io Accessed: 2025-02-25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Samuel Coward, Theo Drane, and George A Constantinides. 2024. Constraint-Aware E-Graph Rewriting for Hardware Performance Optimization. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Leonardo De Moura and Nikolaj Bjørner. 2007. Efficient E-matching for SMT solvers. In Automated Deduction–CADE-21: 21st International Conference on Automated Deduction Bremen, Germany, July 17-20, 2007 Proceedings 21. Springer, 183–198.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Leonardo De Moura and Nikolaj Bjørner. 2008. Z3: an efficient SMT solver. In Proceedings of the Theory and Practice of Software, 14th International Conference on Tools and Algorithms for the Construction and Analysis of Systems (Budapest, Hungary) (TACAS’08/ETAPS’08). Springer-Verlag, Berlin, Heidelberg, 337–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "R.A. DeMillo, R.J. Lipton, and F.G. Sayward. 1978. Hints on Test Data Selection: Help for the Practicing Programmer. Computer 11, 4 (1978), 34–41. doi:10.1109/C-M.1978.218136",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "David Detlefs, Greg Nelson, and James B. Saxe. 2005. Simplify: a theorem prover for program checking. J. ACM 52, 3 (May 2005), 365–473. doi:10.1145/1066100.1066102 Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 118. Publication date: April 2026. Metamorphic Testing for Infrastructure-as-Code Engines 118:27",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Georgios-Petros Drosos, Thodoris Sotiropoulos, Georgios Alexopoulos, Dimitris Mitropoulos, and Zhendong Su. 2024. When Your Infrastructure Is a Buggy Program: Understanding Faults in Infrastructure as Code Ecosystems. Proc. ACM Program. Lang. 8, OOPSLA2, Article 359 (Oct. 2024), 31 pages. doi:10.1145/3689799",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Chris Fallin. 2023. ægraphs: Acyclic E-graphs for Efficient Optimization in a Production Compiler. In EGRAPHS Workshop. https://pldi23.sigplan.org/details/egraphs-2023-papers/2/-graphs-Acyclic-E-graphs-for-Efficient-Optimizationin-a-Production-Compiler Invited Talk.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Kate Fazzini. [n. d.]. A Technical Slip-up Exposes Cloud Collaboration Risks. https://www.wsj.com/articles/a-technicalslip-up-exposes-cloud-collaboration-risks-1497353313. Accessed: 2025-05-20.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Firefly. 2025. The State of IaC 2025. https://www.firefly.ai/state-of-iac-2025 Accessed: 2025-10-07.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Gordon Fraser and Andrea Arcuri. 2014. A Large-Scale Evaluation of Automated Unit Test Generation Using EvoSuite. ACM Trans. Softw. Eng. Methodol. 24, 2, Article 8 (Dec. 2014), 42 pages. doi:10.1145/2685612",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Bernard A. Galler and Michael J. Fisher. 1964. An improved equivalence algorithm. Commun. ACM 7, 5 (May 1964), 301–303. doi:10.1145/364099.364331",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Google. 2025. Go Coverage Profiling. https://go.dev/doc/build-cover. Accessed 2025-09-15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Google. 2025. The Go Programming Language. https://go.dev/. Accessed 2025-09-15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Mark Harman and Phil McMinn. 2009. A theoretical and empirical study of search-based testing: Local, global, and hybrid search. IEEE Transactions on Software Engineering 36, 2 (2009), 226–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Mark Harman, Jillian Ritchey, Inna Harper, Shubho Sengupta, Ke Mao, Abhishek Gulati, Christopher Foster, and Hervé Robert. 2025. Mutation-Guided LLM-based Test Generation at Meta. Association for Computing Machinery, New York, NY, USA, 180–191. https://doi.org/10.1145/3696630.3728544",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "HashiCorp. 2020. Cycle error when removing a resource along with create_before_destroy. https://github.com/ hashicorp/terraform/issues/26226. Accessed: 2025-09-26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "HashiCorp. 2021. Graph missing dependency from resource update to resource destroy. https://github.com/hashicorp/ terraform/issues/28150. Accessed: 2025-09-26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "HashiCorp. 2022. Resources disappearing from state after successful apply. GitHub issue #27671, hashicorp/terraformprovider-aws. https://github.com/hashicorp/terraform-provider-aws/issues/27671",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "HashiCorp. 2025. CDK for Terraform. https://developer.hashicorp.com/terraform/cdktf Accessed: 2025-02-25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "HashiCorp. 2025. Terraform Docs Overview. https://developer.hashicorp.com/terraform/docs Accessed: 2025-02-25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Mohammad Mehedi Hassan and Akond Rahman. 2022. As Code Testing: Characterizing Test Quality in Open Source Ansible Development. In. IEEE, 208–219. doi:10.1109/ICST53961.2022.00031",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Waldemar Hummer, Florian Rosenberg, Fábio Oliveira, and Tamar Eilam. 2013. Testing Idempotence for Infrastructure as Code. In Middleware 2013, David Eyers and Karsten Schwan (Eds.). Springer Berlin Heidelberg, Berlin, Heidelberg, 368–388.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Puppet Inc. 2025. Puppet Documentation. https://www.puppet.com/docs/index.html Accessed: 2025-02-25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "RedHat Inc. 2025. Ansible Community Documentation. https://docs.ansible.com Accessed: 2025-02-25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Laura Inozemtseva and Reid Holmes. 2014. Coverage is not strongly correlated with test suite effectiveness. In Proceedings of the 36th international conference on software engineering. 435–445.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Rajeev Joshi, Greg Nelson, and Keith Randall. 2002. Denali: a goal-directed superoptimizer. In Proceedings of the ACMSIGPLAN 2002 Conference on Programming Language Design and Implementation (Berlin, Germany) (PLDI ’02). Association for Computing Machinery, New York, NY, USA, 304–314. doi:10.1145/512529.512566",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Cole Kurashige, Ruyi Ji, Aditya Giridharan, Mark Barbone, Daniel Noor, Shachar Itzhaky, Ranjit Jhala, and Nadia Polikarpova. 2024. CCLemma: E-Graph Guided Lemma Discovery for Inductive Equational Proofs. Proc. ACM Program. Lang. 8, ICFP, Article 264 (Aug. 2024), 27 pages. doi:10.1145/3674653",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Leonidas Lampropoulos, Michael Hicks, and Benjamin C Pierce. 2019. Coverage guided, property based testing. Proceedings of the ACM on Programming Languages 3, OOPSLA (2019), 1–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. In Proceedings of the 35th ACMSIGPLAN Conference on Programming Language Design and Implementation (Edinburgh, United Kingdom) (PLDI ’14). Association for Computing Machinery, New York, NY, USA, 216–226. doi:10.1145/2594291.2594334",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Vsevolod Livinskii, Dmitry Babokin, and John Regehr. 2020. Random testing for C and C++ compilers with YARPGen. Proc. ACM Program. Lang. 4, OOPSLA, Article 196 (Nov. 2020), 25 pages. doi:10.1145/3428264",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Stephan Lukasczyk and Gordon Fraser. 2022. Pynguin: automated unit test generation for Python. In Proceedings of the ACM/IEEE 44th International Conference on Software Engineering: Companion Proceedings (Pittsburgh, Pennsylvania) (ICSE ’22). Association for Computing Machinery, New York, NY, USA, 168–172. doi:10.1145/3510454.3516829",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Kurt Mackie. [n. d.]. Configuration Glitch Caused Microsoft’s Jan. 25 Exchange Online Disruption. https://redmondmag. com/articles/2023/01/26/configuration-glitch-caused-microsoft-jan-25-exchange-online-disruption.aspx. Accessed: 2025-05-20. Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 118. Publication date: April 2026. 118:28 David Spielmann, George Zakhour, Dominik Arnold, Matteo Biag",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Muhammad Numair Mansur, Maria Christakis, and Valentin Wüstholz. 2021. Metamorphic testing of Datalog engines. InProceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 639–650.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Muhammad Numair Mansur, Valentin Wüstholz, and Maria Christakis. 2023. Dependency-Aware Metamorphic Testing of Datalog Engines. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, Seattle, WA, USA, July 17-21, 2023, René Just and Gordon Fraser (Eds.). ACM, 236–247. doi:10.1145/3597926.3598052",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Barton P. Miller, Lars Fredriksen, and Bryan So. 1990. An empirical study of the reliability of UNIX utilities. Commun. ACM 33, 12 (Dec. 1990), 32–44. doi:10.1145/96267.96279",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Chandrakana Nandi, Max Willsey, Adam Anderson, James R. Wilcox, Eva Darulova, Dan Grossman, and Zachary Tatlock. 2020. Synthesizing structured CAD models with equality saturation and inverse transformations. In Proceedings of the 41st ACMSIGPLAN Conference on Programming Language Design and Implementation (London, UK) (PLDI 2020). Association for Computing Machinery, New York, NY, USA, 31–44. doi:",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Charles Gregory Nelson. 1980. Techniques for program verification. Stanford University.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Ruben Opdebeeck, Ahmed Zerouali, and Coen De Roover. 2022. Smelly Variables in Ansible Infrastructure Code: Detection, Prevalence, and Lifetime. In Proceedings of the 19th International Conference on Mining Software Repositories (Pittsburgh, Pennsylvania) (MSR ’22). Association for Computing Machinery, New York, NY, USA, 61–72. doi:10.1145/ 3524842.3527964",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Ruben Opdebeeck, Ahmed Zerouali, and Coen De Roover. 2023. Control and Data Flow in Security Smell Detection for Infrastructure as Code: Is It Worth the Effort?. In Repositories (MSR). IEEE, Melbourne, Australia, 534–545. doi:10.1109/MSR59073.2023.00079",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "OpenTofu. 2025. OpenTofu Documentation. https://opentofu.org/docs/ Accessed: 2025-03-03.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Owain Parry, Gregory M. Kapfhammer, Michael Hilton, and Phil McMinn. 2022. A Survey of Flaky Tests. ACM Trans. Softw. Eng. Methodol. 31, 1 (2022), 17:1–17:74. doi:10.1145/3476105",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Goran Petrović, Marko Ivanković, Gordon Fraser, and René Just. 2022. Practical Mutation Testing at Scale: A view from Google. IEEE Transactions on Software Engineering 48, 10 (2022), 3900–3912. doi:10.1109/TSE.2021.3107634",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Kun Qiu, Zheng Zheng, Tsong Yueh Chen, and Pak-Lok Poon. 2022. Theoretical and Empirical Analyses of the Effectiveness of Metamorphic Relation Composition. IEEE Trans. Software Eng. 48, 3 (2022), 1001–1017. doi:10.1109/ TSE.2020.3009698",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Akond Rahman, Md Rayhanur Rahman, Chris Parnin, and Laurie Williams. 2021. Security Smells in Ansible and Chef Scripts: A Replication Study. ACM Trans. Softw. Eng. Methodol. 30, 1, Article 3 (jan 2021), 31 pages. doi:10.1145/3408897",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA) (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/33",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 59,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (Nov. 2020), 30 pages. doi:10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 60,
+      "text": "Rudi Schneider, Marcus Rossel, Amir Shaikhha, Andrés Goens, Thomas Kœhler, and Michel Steuwer. 2025. Slotted E-Graphs: First-Class Support for (Bound) Variables in E-Graphs. Proceedings of the ACM on Programming Languages 9, PLDI (2025), 1888–1910.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Sergio Segura, Gordon Fraser, Ana Belén Sánchez, and Antonio Ruiz Cortés. 2016. A Survey on Metamorphic Testing. IEEE Trans. Software Eng. 42, 9 (2016), 805–824. doi:10.1109/TSE.2016.2532875",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Amazon Web Services. 2025. AWSCDK. https://docs.aws.amazon.com/cdk/v2/guide/home.html Accessed: 2025-02-25.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Amazon Web Services. 2025. AWS CloudFormation. https://aws.amazon.com/cloudformation/ Accessed: 2025-10-08.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Eytan Singher and Shachar Itzhaky. 2024. Easter egg: Equality reasoning based on E-graphs with multiple assumptions. In2024 Formal Methods in Computer-Aided Design (FMCAD). IEEE, 70–83.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Daniel Sokolowski, David Spielmann, and Guido Salvaneschi. 2024. Automated Infrastructure as Code Program Testing. IEEE Trans. Software Eng. 50, 6 (2024), 1585–1599. doi:10.1109/TSE.2024.3393070",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Daniel Sokolowski, David Spielmann, and Guido Salvaneschi. 2024. The PIPr Dataset of Public Infrastructure as Code Programs. In Proceedings of the 21st International Conference on Mining Software Repositories (Lisbon, Portugal) (MSR ’24). Association for Computing Machinery, New York, NY, USA, 498–503. doi:10.1145/3643991.3644888",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Thodoris Sotiropoulos, Dimitris Mitropoulos, and Diomidis Spinellis. 2020. Practical fault detection in puppet programs. InProceedings of the ACM/IEEE 42nd International Conference on Software Engineering (Seoul, South Korea) (ICSE ’20). Association for Computing Machinery, New York, NY, USA, 26–37. doi:10.1145/3377811.3380384",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "David Spielmann, George Zakhour, Dominik Arnold, Matteo Biagiola, Roland Meier, and Guido Salvaneschi. 2026. Artifact for Metamorphic Testing for Infrastructure-as-Code Engines. doi:10.5281/zenodo.18755966 Software artifact; latest version. Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 118. Publication date: April 2026. Metamorphic Testing for Infrastructure-as-Code Engines 118:29",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "David Spielmann, George Zakhour, Dominik Arnold, Matteo Biagiola, Roland Meier, and Guido Salvaneschi. 2026. Artifact for Metamorphic Testing for Infrastructure-as-Code Engines. doi:10.5281/zenodo.18755967 Software artifact; version 1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Maolin Sun, Yibiao Yang, Jiangchang Wu, and Yuming Zhou. 2025. Validating SMT Rewriters via Rewrite Space Exploration Supported by Generative Equality Saturation. Proceedings of the ACM on Programming Languages 9, OOPSLA2 (2025), 1205–1231.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Ross Tate, Michael Stepp, Zachary Tatlock, and Sorin Lerner. 2009. Equality saturation: a new approach to optimization. SIGPLAN Not. 44, 1 (Jan. 2009), 264–276. doi:10.1145/1594834.1480915",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Pulumi Team. 2025. Pulumi Docs. https://www.pulumi.com/docs/ Accessed: 2025-02-24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Avito Tech. 2024. go-mutesting: Mutation testing for Go source code. https://github.com/avito-tech/go-mutesting. Accessed: 2025-08-05.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Vpn Monitor Research Team. 2019. Report: Travel Reservations Platform Leaks US Government Personnel Data. https://www.vpnmentor.com/blog/us-travel-military-leak/. Accessed: 2025-05-20.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Max Willsey, Chandrakana Nandi, Yisu Remy Wang, Oliver Flatt, Zachary Tatlock, and Pavel Panchekha. 2021. egg: Fast and extensible equality saturation. Proc. ACM Program. Lang. 5, POPL, Article 23 (Jan. 2021), 29 pages. doi:10.1145/3434304",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Xiaoyuan Xie, Joshua Ho, Christian Murphy, Gail Kaiser, Baowen Xu, and Tsong Yueh Chen. 2009. Application of Metamorphic Testing to Supervised Classifiers. In Proceedings of the 2009 Ninth International Conference on Quality Software (QSIC ’09). IEEE Computer Society, USA, 135–144. doi:10.1109/QSIC.2009.26",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "George Zakhour, Pascal Weisenburger, Jahrim Gabriele Cesario, and Guido Salvaneschi. 2025. Dis/Equality Graphs. Proc. ACM Program. Lang. 9, POPL, Article 77 (Jan. 2025), 24 pages. doi:10.1145/3704913 Received 2025-10-10; accepted 2026-02-17 Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 118. Publication date: April 2026.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 58,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA) (ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 59,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA, Article 211 (Nov. 2020), 30 pages. doi:10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[58,59]",
+      "technique": "norec",
+      "sentence": "Notably, Rigger and Su [58,59]applied MT to database management systems and uncovered numerous bugs in popular systems.",
+      "context_before": "and graphs derived from real-world programs. Finally, because test generation and e-graph exploration are randomized, individual runs may differ. To mitigate this, we repeat each experiment multiple times and report averages. The evaluation shows that the differences between runs are small. 8 Related Work Metamorphic Testing. MT has been successfully applied to many domains. Mansur et al. [46,47] applied MT to test Datalog engines, revealing several bugs in state-of-the-art engines. Moreover, MT has been applied to test compilers [ 42], machine learning applications [ 76], or web systems [ 8].",
+      "context_after": "Segura et al. [61] provide a comprehensive survey of MT techniques and applications. Sun et al. [70] validate SMT solver rewrite systems by exploring the rewrite space via e-graphs and equality saturation. Compared to prior work that tests programs or domain-specific systems, we define MRs directly over resource graphs expressed in an IR, which makes the approach general and engine-agnostic. Infrastructure as Code. IaCengines are commonly classified as either configuration-based or provision-based [19]. The former, such as Ansible [ 37], Puppet [ 36], and Chef [ 13], manage the configuration",
+      "section": "4.6 Formalism",
+      "page": 24,
+      "char_offset": 88018,
+      "cited_reference": {
+        "number": 58,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (Virtual Event, USA",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "EMIaC applies metamorphic testing to Infrastructure-as-Code engines such as Terraform, OpenTofu and Pulumi, which the authors note have received little attention compared with the correctness of IaC programs themselves. It defines metamorphic relations as graph-based transformations of IaC programs and checks invariants across executions, using e-graphs both as an input generator and as an equivalence oracle. Three of its test cases were merged into Terraform.",
+    "narrative": "One citation in the formalism section, noting that Rigger and Su applied metamorphic testing to database management systems and uncovered numerous bugs in popular systems -- the precedent EMIaC carries into a different class of engine.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3798232.json b/_data/papers/paper_doi_10_1145_3798232.json
new file mode 100644
index 0000000..9a6dbf9
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3798232.json
@@ -0,0 +1,829 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:28:58Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3798232",
+    "title": "Beacon: Detecting Broken Access Control Vulnerabilities in DBMSs via System Catalog Consistency Validation",
+    "authors": [
+      "Zongrui Peng",
+      "Jingzhou Fu",
+      "Zhiyong Wu",
+      "Jie Liang",
+      "Xiangdong Huang",
+      "Dalong Shi",
+      "Yu Jiang"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Programming Languages",
+    "doi": "10.1145/3798232",
+    "arxiv_id": null,
+    "s2_paper_id": "804bc6d1f6c1acde45f1fabcab8b0d3ed5f647c9",
+    "url": "https://doi.org/10.1145/3798232",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3798232",
+      "retrieved_at": "2026-09-09T01:28:58Z",
+      "chars": 111469,
+      "content_sha256": "sha256:30ff25fd949c413dffa49b2a4b0b400c2cf70005ebd485c0fcd70ad9d95f4614"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 29,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3675
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 12556
+      },
+      {
+        "number": "3",
+        "title": "Motivation",
+        "start": 18947
+      },
+      {
+        "number": "4.2",
+        "title": "Privilege-Aware Test Case Generation",
+        "start": 31046
+      },
+      {
+        "number": "4.3",
+        "title": "Unauthorized Object Identification",
+        "start": 40804
+      },
+      {
+        "number": "4.4",
+        "title": "Catalog Consistency Validation",
+        "start": 46215
+      },
+      {
+        "number": "6.1",
+        "title": "Evaluation Setup",
+        "start": 57858
+      },
+      {
+        "number": "6.2",
+        "title": "BAC Vulnerability Detection",
+        "start": 58859
+      },
+      {
+        "number": "6.3",
+        "title": "Selected Vulnerabilities Found byBeacon",
+        "start": 64500
+      },
+      {
+        "number": "6.4",
+        "title": "Design Choice Analysis",
+        "start": 70850
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Ahmed Anas, Salwa Elgamal, and Basheer Youssef. 2024. Survey on detecting and preventing web application broken access control attacks.International Journal of Electrical and Computer Engineering (IJECE)14, 1 (2024), 772–781. doi:10.11591/ijece.v14i1.pp772-781",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (ICSE). 1–13. doi:10.1145/3597503.3639076",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Elisa Bertino, Gabriel Ghinita, Ashish Kamra, et al .2011. Access control for databases: Concepts and systems. Foundations and Trends®in Databases3, 1–2 (2011), 1–148. doi:10.1561/9781601984173",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Elisa Bertino and Ravi Sandhu. 2005. Database security-concepts, approaches, and challenges.IEEE Transactions on Dependable and secure computing2, 1 (2005), 2–19. doi:10.1109/TDSC.2005.9",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Jianneng Cao, Barbara Carminati, Elena Ferrari, and Kian-Lee Tan. 2009. Acstream: Enforcing access control over data streams. In. IEEE, 1495–1498. doi:10 .1109/ ICDE.2009.25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "MariaDB Corporation. 2025. MariaDB. https://mariadb.org/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "The MITRE Corporation. 2026. CWE-284: Improper Access Control. https://cwe .mitre .org/data/definitions/284 .html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Dameng. 2025. DM8 - A New-Generation Large General Relational Database. https://en .dameng .com/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Dameng. 2025. Privileges for using ArcGIS with Dameng. https://pro .arcgis .com/en/pro-app/latest/help/data/databases/ privileges-dameng.htm. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, and Yu Jiang. 2025. Detecting Logic Bugs in DBMSs via Equivalent Data Construction.Proceedings of the ACM on Management of Data3, 6 (2025), 1–25. doi:10.1145/3769779",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Sabrina De Capitani Di Vimercati, Sara Foresti, Sushil Jajodia, Stefano Paraboschi, and Pierangela Samarati. 2007. Over-encryption: Management of access control evolution on outsourced data. InProceedings of the 33rd International Conference on Very Large Data Bases (VLDB). 123–134. https://dl.acm.org/doi/10.5555/1325851.1325869",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Docker. 2025. Docker Hub. https://hub.docker.com/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Joint Task Force. 2020.Security and privacy controls for information systems and organizations. Technical Report. National Institute of Standards and Technology. doi:10.6028/NIST.SP.800-53r5",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "The Apache Software Foundation. 2025. Apache CouchDB. https://couchdb.apache.org. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, and Yu Jiang. 2024. Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–12. doi:10.1145/3597503.3639210",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free DBMS fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering (ASE). 1–12. doi:10 .1145/ 3551349.3560431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs. InProceedings of the Twentieth European Conference on Computer Systems (EuroSys). 1061–1076. doi:10.1145/3689031.3696064",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 18,
+      "text": "Ying Fu, Zhiyong Wu, Yuanliang Zhang, Jie Liang, Jingzhou Fu, Yu Jiang, Shanshan Li, and Xiangke Liao. 2025. THANOS: DBMS Bug Detection via Storage Engine Rotation Based Differential Testing. InProceedings of the IEEE/ACM 47th International Conference on Software Engineering (ICSE). 1–12. doi:10.1109/ICSE55347.2025.00257",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "The PostgreSQL Global Development Group. 2025. PostgreSQL Documentation: Role Attributes. https:// www.postgresql.org/docs/current/role-attributes.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "The PostgreSQL Global Development Group. 2025. PostgreSQL Documentation: Row Security Policies. https:// www.postgresql.org/docs/current/ddl-rowsecurity.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Yang Hu, Wenxi Wang, Casen Hunger, Riley Wood, Sarfraz Khurshid, and Mohit Tiwari. 2021. ACHyb: A hybrid analysis approach to detect kernel access control vulnerabilities. InProceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 316–327. doi:10.1145/3468264.3468627",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Yongheng Huang, Chenghang Shi, Jie Lu, Haofeng Li, Haining Meng, and Lian Li. 2024. Detecting Broken Object-Level Authorization Vulnerabilities in Database-Backed Applications. InProceedings of the 2024 on ACMSIGSAC Conference on Computer and Communications Security (CCS). 2934–2948. doi:10.1145/3658644.3690227",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "ISO/IEC. 2023. Information technology — Database languages — SQL — Part 2: Foundation (ISO/IEC 9075-2:2023). https://www.iso.org/standard/76584.html. ISO/IEC Standard.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2020. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. InProceedings of the 46th International Conference on Very Large Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 124. Publication date: April 2026. Beacon: Detecting Broken Access Control Vulnerabilities in DBMSs via System Catalog ",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Jinhui Lai, Chi Zhang, Bingyan Li, Chenglin Liang, Jie Liang, Zhiyong Wu, Jingzhou Fu, Yu Jiang, and Zichen Xu. 2025. SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis.Proceedings of the ACM on Management of Data3, 6 (2025), 1–24. doi:10.1145/3769828",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Vu Le, Chengnian Sun, and Zhendong Su. 2015. Finding deep compiler bugs via guided stochastic program mutation. Acm Sigplan Notices50, 10 (2015), 386–399. doi:10.1145/2858965.2814319",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-oriented DBMS fuzzing. InProceedings of IEEE International Conference on Data Engineering (ICDE). 668–681. doi:10.1109/ICDE55515.2023.00057",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (ICSE). 1–12. doi:10.1145/3597503.3639112",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. InProceedings of the 44th International Conference on Software Engineering (ICSE). 225–236. doi:10.1145/3510003.3510093",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "ProxySQL LLC. 2025. ProxySQL: Firewall Whitelist. https://proxysql .com/documentation/Firewall-whitelist/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Jie Lu, Haofeng Li, Chen Liu, Lian Li, and Kun Cheng. 2022. Detecting missing-permission-check vulnerabilities in distributed cloud systems. InProceedings of the 2022 ACMSIGSAC Conference on Computer and Communications Security (CCS). 2145–2158. doi:10.1145/3548606.3560589",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Toby Mao. 2023. SQLGlot Website. https://sqlglot.com/sqlglot.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "MariaDB. 2025. MariaDB Community Audit. https://mariadb .com/docs/server/reference/plugins/mariadb-audit-plugin. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Trend Micro. 2026. About Trend Micro. https://www.trendmicro.com/en_us/about.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Trend Micro. 2026. Apache CouchDB Vulnerabilities Permit Monero Mining. https://www .trendmicro .com/en/ research/18/b/vulnerabilities-apache-couchdb-open-door-monero-miners.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "MITRE. 2025. CWE-200: Exposure of Sensitive Information to an Unauthorized Actor. https://cwe .mitre .org/data/ definitions/200.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "MITRE. 2025. Privilege Escalation. https://attack.mitre.org/tactics/TA0004/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "MonetDB 2025. The Database System to speed up your Analytical Jobs. https://www .monetdb .org/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Maliheh Monshizadeh, Prasad Naldurg, and VN Venkatakrishnan. 2014. Mace: Detecting privilege escalation vulnerabilities in web applications. InProceedings of the 2014 ACMSIGSAC Conference on Computer and Communications Security (CCS). 690–701. doi:10.1145/2660267.2660337",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Rimma V. Nehme, Hyo-Sang Lim, and Elisa Bertino. 2010. FENCE: Continuous access control enforcement in dynamic data stream environments. InProceedings of IEEE 26th International Conference on Data Engineering (ICDE). 940–943. doi:10.1109/ICDE.2010.5447899",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "National Vulnerability Database (NVD). 2025. CVE-2017-12635 Detail. https://nvd .nist.gov/vuln/detail/CVE-2017-12635. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "OceanBase. 2025. CREATE TRIGGER. https://en .oceanbase .com/docs/common-oceanbase-database10000000001974512. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "OceanBase. 2025. The Multi-cloud Distributed Database for Mission-critical Workloads at Any Scale. https: //en.oceanbase.com/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "OceanBase. 2025. OceanBase Security Response Protocol. https://en .oceanbase .com/security. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Oracle 2025. MySQL. https://www.mysql.com/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Oracle. 2025. MySQL Documentation: Access Control and Account Management. https://dev .mysql .com/doc/refman/ 9.2/en/access-control.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Oracle. 2025. MySQL Documentation: GRANT Statement. https://dev .mysql .com/doc/refman/9 .2/en/grant .html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Oracle. 2025. MySQL Documentation: INFORMATION_SCHEMA Introduction. https://dev .mysql .com/doc/refman/ 9.2/en/information-schema-introduction.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Oracle. 2025. MySQL Documentation: INFORMATION_SCHEMA Table Reference. https://dev .mysql .com/doc/refman/ 9.2/en/information-schema-table-reference.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Oracle. 2025. MySQL Documentation: MySQL Enterprise Firewall. https://dev .mysql .com/doc/refman/9 .2/en/ firewall.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Oracle. 2025. MySQL Documentation: Privileges Provided by MySQL. https://dev .mysql .com/doc/refman/9 .2/en/ privileges-provided.html. Accessed: Feb. 27, 2026. Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 124. Publication date: April 2026. 124:28 Zongrui Peng, Jingzhou Fu, Zhiyong Wu, Jie Liang, Xiangdong Huang, Dalong Shi, and Yu Jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Oracle. 2025. MySQL Documentation: Show Open Tables Statements. https://dev .mysql .com/doc/refman/9 .2/en/showopen-tables.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Oracle. 2025. MySQL Documentation: The INFORMATION_SCHEMA COLUMNS Table. https://dev .mysql .com/doc/ refman/9.2/en/information-schema-columns-table.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Oracle. 2025. MySQL Documentation: Using Roles. https://dev .mysql .com/doc/refman/9 .2/en/roles .html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Oracle. 2025. MySQL Enterprise Audit. https://www .mysql .com/products/enterprise/audit .html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Oracle. 2025. Oracle Database Firewall. https://www .oracle .com/technetwork/cn/products/database-firewall/overview/ index.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Oracle. 2025. Using Fine-Grained Auditing. https://www .oracle .com/technical-resources/articles/middleware/idmfga-otn.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Oracle. 2026. MySQL Documentation: ALTER TABLE Statement. https://dev .mysql .com/doc/refman/9 .2/en/altertable.html. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "OWASP. 2026. OWASP Foundation. https://owasp.org/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "OWASP. 2026. Testing for SQL Injection. https://owasp .org/www-project-web-security-testing-guide/latest/4Web_Application _Security _Testing/07-Input _Validation _Testing/05-Testing _for_SQL_Injection. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Andrei Paduroiu, Sungheun Wi, Yan Yan, Roni Burd, Ruhollah Farchtchi, and Giovanni Matteo Fumarola. 2024. Membrane-Safe and Performant Data Access Controls in Apache Spark in the Presence of Imperative Code.Proc. VLDB Endow.17, 12 (2024), 3813–3826. doi:10.14778/3685800.3685808",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Primal Pappachan, Roberto Yus, Sharad Mehrotra, and Johann-Christoph Freytag. 2020. Sieve: a middleware approach to scalable access control for database management systems.Proc. VLDB Endow.13, 12 (2020), 2424–2437. doi:10 .14778/ 3407790.3407835",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Primal Pappachan, Shufan Zhang, Xi He, and Sharad Mehrotra. 2022. Don’t be a tattle-tale: Preventing leakages through data dependencies on access control protected data.Proc. VLDB Endow.15, 11 (2022), 2437–2449. doi:10 .14778/ 3551793.3551805",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Zongrui Peng, Jingzhou Fu, Zhiyong Wu, Jie Liang, Xiangdong Huang, Dalong Shi, and Yu Jiang. 2026. Beacon: Detecting Broken Access Control Vulnerabilities in DBMSs via System Catalog Consistency Validation. doi:10 .5281/ zenodo.18669646",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "PingCAP. 2026. TiDB. https://github.com/pingcap/tidb. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "PortSwigger. 2025. Access control vulnerabilities and privilege escalation. https://portswigger .net/web-security/accesscontrol. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "PortSwigger. 2026. Examining the database in SQL injection attacks. https://portswigger .net/web-security/sqlinjection/examining-the-database. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "PortSwigger. 2026. PortSwigger. https://portswigger.net/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "PostgreSQL 2026. PostgreSQL. https://www.postgresql.org/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152. doi:10.1145/3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 71,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30. doi:10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 72,
+      "text": "Jerome H Saltzer and Michael D Schroeder. 1975. The protection of information in computer systems.Proc. IEEE63, 9 (1975), 1278–1308. doi:10.1109/PROC.1975.9939",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Ravi S Sandhu. 1998. Role-based access control. InAdvances in computers. Vol. 46. Elsevier, 237–286. doi:10 .1016/S00652458(08)60206-5",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https: //github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Abraham Silberschatz, Henry F Korth, Shashank Sudarshan, et al .2002.Database system concepts. Vol. 5. McGraw-Hill New York.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "StarRocks 2025. Fast, Fresh, and Flexible: Analytics With No Compromise. https://www .starrocks .io/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "StarRocks. 2025. Manage audit logs within StarRocks via AuditLoader. https://docs .starrocks .io/docs/administration/ management/audit_loader/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "StarRocks. 2025. StarRocks Documentation: Information Schema. https://docs .starrocks .io/docs/sql-reference/ information_schema/. Accessed: Feb. 27, 2026. Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 124. Publication date: April 2026. Beacon: Detecting Broken Access Control Vulnerabilities in DBMSs via System Catalog Consistency Validation 124:29",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "StarRocks. 2025. StarRocks Documentation: Privileges supported by StarRocks. https://docs .starrocks .io/docs/ administration/user_privs/privilege_item/. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "StarRocks. 2025. StarRocks Test Suite. https://github .com/StarRocks/starrocks/blob/main/test/sql/test _rbac/R/ test_view_privilege. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. 2016. Finding compiler bugs via live code mutation. InProceedings of the 2016 ACMSIGPLAN international conference on object-oriented programming, systems, languages, and applications. 849–863. doi:10.1145/3022671.2984038",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In43rd IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice, ICSE (SEIP) 2021, Madrid, Spain, May 25-28, 2021. IEEE, 328–337. doi:10 .1109/ICSESEIP52600.2021.00042",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Qihua Wang, Ting Yu, Ninghui Li, Jorge Lobo, Elisa Bertino, Keith Irwin, and Ji-Won Byun. 2007. On the correctness criteria of fine-grained access control in relational databases. InProceedings of the 33rd international conference on Very large data bases (VLDB). 555–566. https://dl.acm.org/doi/10.5555/1325851.1325915",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Wikipedia. 2024. Database. https://en.wikipedia.org/wiki/Database. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "Wikipedia. 2025. Data control language. https://en .wikipedia .org/wiki/Data _control _language. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Wikipedia. 2025. Data definition language. https://en .wikipedia .org/wiki/Data _definition _language. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "Wikipedia. 2025. Data manipulation language. https://en .wikipedia .org/wiki/Data _manipulation _language. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "Wikipedia. 2025. Data query language. https://en .wikipedia .org/wiki/Data _query _language. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "Wikipedia. 2025. Role-based access control. https://en .wikipedia .org/wiki/Role-based _access _control. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "Wikipedia. 2025. SQL. https://en.wikipedia.org/wiki/SQL. Accessed: Feb. 27, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. Hulk: Exploring Data-Sensitive Performance Anomalies in DBMSs via Data-Driven Analysis.Proceedings of the ACM on Software Engineering2, ISSTA (2025), 2181–2202. doi:10.1145/3728973",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. PUPPY: Finding Performance Degradation Bugs in DBMSs via Limited-Optimization Plan Construction. InProceedings of the IEEE/ACM 47th International Conference on Software Engineering (ICSE). 1–12. doi:10.1109/ICSE55347.2025.00045",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. 2022. Unicorn: detect runtime errors in time-series databases with hybrid input synthesis. InProceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis. 251–262. doi:10.1145/3533767.3534364",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "Tong Zhang, Wenbo Shen, Dongyoon Lee, Changhee Jung, Ahmed M Azab, and Ruowen Wang. 2019. PeX: A permission check analysis framework for linux kernel. InProceedings of the 28th USENIX Conference on Security Symposium (USENIX Security). 1205–1220.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. Squirrel: Testing Database Management Systems with Language Validity and Coverage Feedback. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security (CCS). 955–970. doi:10.1145/3372297.3417260 Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 124. Publication date: April 2026.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (ICSE). 1–13. doi:10.1145/3597503.3639076",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert",
+        "prints the DOI of the paper introducing cert"
+      ]
+    },
+    {
+      "number": 17,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs. InProceedings of the Twentieth European Conference on Computer Systems (EuroSys). 1061–1076. doi:10.1145/3689031.3696064",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "OceanBase. 2025. CREATE TRIGGER. https://en .oceanbase .com/docs/common-oceanbase-database10000000001974512. Accessed: Feb. 27, 2026.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 70,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–1152. doi:10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 71,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proceedings of the ACM on Programming Languages4, OOPSLA (2020), 1–30. doi:10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 42]",
+      "technique": null,
+      "sentence": "Specifically, the user creates a trigger [ 42] containing SELECT operations that require the privileges of another user’s table.",
+      "context_before": "sing from failures to enforce proper access control, are collectively known asBroken Access Control(BAC) [ 7] vulnerabilities. For instance, according to Trend Micro [ 34], a BAC vulnerability in Apache CouchDB [ 14] enabled hundreds of real-world attacks that abused victim machines in organizations for cryptocurrency mining, causing severe resource exhaustion and device slowdowns [35, 41]. Figure 1 illustrates a BAC vulnerability in OceanBase where a regular user (with only basic privileges to manage their own tables) can bypass access control to retrieve information in tables of other users.",
+      "context_after": "When the trigger is activated, OceanBase skips permission checks on the SELECTcommand inside, resulting in the unauthorized disclosure of data. The root cause of this BAC vulnerability lies in the implementation of OceanBase, which enforces only the CREATE TRIGGER privilege at creation time, without validating the privileges required for the operations inside the trigger’s definition. This vulnerability can be exploited to bypass access control during trigger execution. For instance, a regular user can embed SELECT ,UPDATE, or DELETE operations on another user’s table inside a trigger defined on",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5224,
+      "cited_reference": {
+        "number": 42,
+        "text": "OceanBase. 2025. CREATE TRIGGER. https://en .oceanbase .com/docs/common-oceanbase-database10000000001974512. Accessed: Feb. 27, 2026.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 26,70,71,81]",
+      "technique": "norec",
+      "sentence": "Another test case generation approach for detecting consistency issues is mutation-based testing, which has been widely used in compiler testing and optimization testing [ 26,70,71,81].",
+      "context_before": "during execution. However, this reflects the expected behavior of MySQL’s access control since the tablea2is a temporary table used solely for handling the alias in the SELECTcommand. Test Case Generation Approaches.In our approach, we adopt a bottom-up generation method to compose test cases from DDL, DML, DQL, and DCL commands, thereby covering a wide range of SQL operations and object types in a unified workflow. Moreover, by leveraging catalog consistency as the test oracle, our approach can detect BAC vulnerabilities without explicitly modeling the underlying privilege model of each DBMS.",
+      "context_after": "Specifically, mutation-based methods for consistency testing transform a test case into semantically equivalent variants through certain mutation strategies, and then check whether execution results remain consistent before and after mutation. By accumulating non-trivial mutations over multiple iterations, the resulting test inputs can become more complex, which helps exercise deeper execution paths. However, adapting these methods to BAC testing presents several challenges. For instance, the mutations must preserve the privilege-relevant behaviors under the DBMS privilege model, which requir",
+      "section": "6.4 Design Choice Analysis",
+      "page": 23,
+      "char_offset": 80819,
+      "cited_reference": {
+        "number": 70,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–115",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 10,15,18,25,28,70,71]",
+      "technique": "norec",
+      "sentence": "Logic bug testing [ 10,15,18,25,28,70,71] focuses on verifying the correctness of execution results.",
+      "context_before": "ntly of any fixed permission models. Instead of analyzing code paths,Beacondirectly detects inconsistencies by comparing the execution results of SQL commands with the permission information in system catalogs. This dynamic approach allowsBeaconto adapt to various flexible authorization scenarios. Additionally, verifying the consistency between system catalogs and actual operations ensures cross-DBMS compatibility. DBMS Testing.DBMSs are complex software systems and thus prone to problems. DBMS testing efforts can be categorized into three types: testing for logic, crash, and performance bugs.",
+      "context_after": "These studies typically concentrate on constructing logic test oracles. For example,NoREC[ 70] detects differences in result sets between optimizing and non-optimizing query versions, and TLP[ 71] detects whether composed results of partitioning queries are equivalent to the original query’s results. Crash bug testing [ 16,17,27,74,82,93,95] aims to identify circumstances under which DBMSs might crash. These approaches focus on generating valid SQL commands and monitoring DBMS crashes. For instance, SQLsmith [ 74] generates SQL commands based on syntax modeling, andSqirrel[ 95] uses coverage f",
+      "section": "6.4 Design Choice Analysis",
+      "page": 24,
+      "char_offset": 86668,
+      "cited_reference": {
+        "number": 70,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE). 1140–115",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For example,NoREC[ 70] detects differences in result sets between optimizing and non-optimizing query versions, and TLP[ 71] detects whether composed results of partitioning queries are equivalent to the original query’s results.",
+      "context_before": "ssion information in system catalogs. This dynamic approach allowsBeaconto adapt to various flexible authorization scenarios. Additionally, verifying the consistency between system catalogs and actual operations ensures cross-DBMS compatibility. DBMS Testing.DBMSs are complex software systems and thus prone to problems. DBMS testing efforts can be categorized into three types: testing for logic, crash, and performance bugs. Logic bug testing [ 10,15,18,25,28,70,71] focuses on verifying the correctness of execution results. These studies typically concentrate on constructing logic test oracles.",
+      "context_after": "Crash bug testing [ 16,17,27,74,82,93,95] aims to identify circumstances under which DBMSs might crash. These approaches focus on generating valid SQL commands and monitoring DBMS crashes. For instance, SQLsmith [ 74] generates SQL commands based on syntax modeling, andSqirrel[ 95] uses coverage feedback to guide the mutation of SQL commands. Performance bug testing [ 2,24,29,91,92] checks whether execution results are returned within reasonable time limits. For example,Apollo[24] compares the execution time of different DBMS versions. In contrast to traditional DBMS testing approaches,Beacon",
+      "section": "6.4 Design Choice Analysis",
+      "page": 24,
+      "char_offset": 86841,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 16,17,27,74,82,93,95]",
+      "technique": null,
+      "sentence": "Crash bug testing [ 16,17,27,74,82,93,95] aims to identify circumstances under which DBMSs might crash.",
+      "context_before": "S compatibility. DBMS Testing.DBMSs are complex software systems and thus prone to problems. DBMS testing efforts can be categorized into three types: testing for logic, crash, and performance bugs. Logic bug testing [ 10,15,18,25,28,70,71] focuses on verifying the correctness of execution results. These studies typically concentrate on constructing logic test oracles. For example,NoREC[ 70] detects differences in result sets between optimizing and non-optimizing query versions, and TLP[ 71] detects whether composed results of partitioning queries are equivalent to the original query’s results.",
+      "context_after": "These approaches focus on generating valid SQL commands and monitoring DBMS crashes. For instance, SQLsmith [ 74] generates SQL commands based on syntax modeling, andSqirrel[ 95] uses coverage feedback to guide the mutation of SQL commands. Performance bug testing [ 2,24,29,91,92] checks whether execution results are returned within reasonable time limits. For example,Apollo[24] compares the execution time of different DBMS versions. In contrast to traditional DBMS testing approaches,Beacondirectly targets BAC vulnerabilities. We leverage an observable invariant in DBMS catalogs: each user sh",
+      "section": "6.4 Design Choice Analysis",
+      "page": 24,
+      "char_offset": 87070,
+      "cited_reference": {
+        "number": 17,
+        "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs. InProceedings of the Twentieth European Conference on Computer Systems (EuroSys). 1061–1076. doi:10.1145",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[ 2,24,29,91,92]",
+      "technique": "cert",
+      "sentence": "Performance bug testing [ 2,24,29,91,92] checks whether execution results are returned within reasonable time limits.",
+      "context_before": "ucting logic test oracles. For example,NoREC[ 70] detects differences in result sets between optimizing and non-optimizing query versions, and TLP[ 71] detects whether composed results of partitioning queries are equivalent to the original query’s results. Crash bug testing [ 16,17,27,74,82,93,95] aims to identify circumstances under which DBMSs might crash. These approaches focus on generating valid SQL commands and monitoring DBMS crashes. For instance, SQLsmith [ 74] generates SQL commands based on syntax modeling, andSqirrel[ 95] uses coverage feedback to guide the mutation of SQL commands.",
+      "context_after": "For example,Apollo[24] compares the execution time of different DBMS versions. In contrast to traditional DBMS testing approaches,Beacondirectly targets BAC vulnerabilities. We leverage an observable invariant in DBMS catalogs: each user should only view and operate on the objects that the user is authorized to access. By correlating catalog visibility with the executability of privileged commands, our technique uncovers subtle BAC vulnerabilities that have evaded detection in prior testing frameworks. DBMS Access Control.Recent research on DBMS access control has primarily focused on develop",
+      "section": "6.4 Design Choice Analysis",
+      "page": 24,
+      "char_offset": 87415,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (ICSE). 1–13. doi:10.1145/3597503.3639076",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert",
+          "prints the DOI of the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:20:36Z",
+    "is_model_written": true,
+    "summary": "Beacon detects broken access control vulnerabilities in DBMSs by checking SQL operations against the system catalogs. Its premise is that catalog visibility should agree with a user's privileges: if an object is invisible to a user in the catalogs, that user should have no access to it, so any inconsistency indicates a privilege they should not have. Applied to eight DBMSs it uncovered 39 previously unknown vulnerabilities, all confirmed by vendors.",
+    "narrative": "Cited in the design discussion as the logic-bug testing line of work, with NoREC and TLP described by what each compares -- the correctness dimension Beacon sets beside its own access-control one.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "definition",
+      "M4": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3798245.json b/_data/papers/paper_doi_10_1145_3798245.json
new file mode 100644
index 0000000..f3bfa8c
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3798245.json
@@ -0,0 +1,793 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:24:12Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3798245",
+    "title": "Understanding and Finding JIT Compiler Performance Bugs",
+    "authors": [
+      "Zijian Yi",
+      "Chen Ding",
+      "August Shi",
+      "Milos Gligoric"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Programming Languages",
+    "doi": "10.1145/3798245",
+    "arxiv_id": "2603.06551",
+    "s2_paper_id": "c55ef302233eb0c1ccc6c9677eda03b16089b148",
+    "url": "https://doi.org/10.1145/3798245",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2603.06551",
+      "retrieved_at": "2026-09-08T15:24:12Z",
+      "chars": 107482,
+      "content_sha256": "sha256:47e9a68c78c9ddd30d0a0047506cce030db271d67fa8b6279596d6acbd116932"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 29,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2491
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 11095
+      },
+      {
+        "number": "2.1",
+        "title": "Overview of JIT Compilers",
+        "start": 11428
+      },
+      {
+        "number": "2.2",
+        "title": "Speculative Optimization",
+        "start": 14443
+      },
+      {
+        "number": "2.3",
+        "title": "Tiered Compilation",
+        "start": 17861
+      },
+      {
+        "number": "3",
+        "title": "Study Methodology",
+        "start": 20928
+      },
+      {
+        "number": "3.1",
+        "title": "Bug Collection",
+        "start": 21241
+      },
+      {
+        "number": "3.2",
+        "title": "Bug Filtering and Analysis",
+        "start": 24449
+      },
+      {
+        "number": "3.3",
+        "title": "Research Questions",
+        "start": 26438
+      },
+      {
+        "number": "3.4",
+        "title": "Threats to Validity",
+        "start": 27269
+      },
+      {
+        "number": "4",
+        "title": "Case Study",
+        "start": 28255
+      },
+      {
+        "number": "5",
+        "title": "Characteristics of Bugs",
+        "start": 36096
+      },
+      {
+        "number": "5.1",
+        "title": "RQ1: Input Artifacts",
+        "start": 36504
+      },
+      {
+        "number": "5.2",
+        "title": "RQ2: Symptoms",
+        "start": 39407
+      },
+      {
+        "number": "5.3",
+        "title": "RQ3: Root Causes",
+        "start": 43989
+      },
+      {
+        "number": "5.4",
+        "title": "Fixes",
+        "start": 48163
+      },
+      {
+        "number": "6.1",
+        "title": "Overview",
+        "start": 51050
+      },
+      {
+        "number": "6.2",
+        "title": "Algorithm",
+        "start": 54753
+      },
+      {
+        "number": "6.3",
+        "title": "Filtering False Positives and Duplicates",
+        "start": 59682
+      },
+      {
+        "number": "6.4",
+        "title": "Evaluation",
+        "start": 61273
+      },
+      {
+        "number": "6.5",
+        "title": "Discovered Bugs",
+        "start": 71118
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Nader Al Awar, Zijian Yi, George Biros, and Milos Gligoric. 2025. Speeding up the local C++ development cycle with header substitution. InInternational Symposium on Code Generation and Optimization. 704–717. doi:10.1145/3696443. 3708942",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "V Aho Alfred, S Lam Monica, and D Ullman Jeffrey. 2007.Compilers principles, techniques & tools. pearson Education. Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 137. Publication date: April 2026. 137:26 Zijian Yi, Cheng Ding, August Shi, and Milos Gligoric",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Matthew Arnold, Stephen J Fink, David Grove, Michael Hind, and Peter F Sweeney. 2005. A survey of adaptive optimization in virtual machines.Proc. IEEE93, 2 (2005), 449–466. doi:10.1109/JPROC.2004.840305",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Md Abul Kalam Azad, Nafees Iqbal, Foyzul Hassan, and Probir Roy. 2023. An empirical study of high performance computing (HPC) performance bugs. InInternational Conference on Mining Software Repositories. 194–206. doi:10.1109/ MSR59073.2023.00037",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13. doi:10.1145/3597503.3639076",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Gergö Barany. 2018. Finding missed compiler optimizations by differential testing. InInternational Conference on Compiler Construction. 82–92. doi:10.1145/3178372.3179521",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Edd Barrett, Carl Friedrich Bolz-Tereick, Rebecca Killick, Sarah Mount, and Laurence Tratt. 2017. Virtual machine warmup blows hot and cold.Proceedings of the ACM on Programming Languages1, OOPSLA (2017), 1–27. doi:10.1145/ 3133876",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Aurèle Barrière, Sandrine Blazy, Olivier Flückiger, David Pichardie, and Jan Vitek. 2021. Formally verified speculation and deoptimization in a JIT compiler.Proceedings of the ACM on Programming Languages5, POPL (2021), 1–26. doi:10.1145/3434327",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Aurèle Barrière, Sandrine Blazy, and David Pichardie. 2023. Formally verified native code generation in an effectful JIT: turning the CompCert backend into a formally verified JIT compiler.Proceedings of the ACM on Programming Languages7, POPL (2023), 249–277. doi:10.1145/3571202",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Matteo Basso, Aleksandar Prokopec, Andrea Rosà, and Walter Binder. 2023. Optimization-aware compiler-level event profiling.ACM Transactions on Programming Languages and Systems45, 2 (2023), 1–50. doi:10.1145/3591473",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Stephen M Blackburn, Robin Garner, Chris Hoffmann, Asjad M Khang, Kathryn S McKinley, Rotem Bentzur, Amer Diwan, Daniel Feinberg, Daniel Frampton, Samuel Z Guyer, et al .2006. The DaCapo benchmarks: Java benchmarking development and analysis. InInternational Conference on Object-Oriented Programming, Systems, Languages, and Applications. 169–190. doi:10.1145/1167515.1167488",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Humphrey Burchell, Octave Larose, and Stefan Marr. 2024. Towards realistic results for instrumentation-based profilers for JIT-compiled systems. InInternational Conference on Managed Programming Languages and Runtimes. 82–89. doi:10.1145/3679007.3685058",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Stefanos Chaliasos, Thodoris Sotiropoulos, Georgios-Petros Drosos, Charalambos Mitropoulos, Dimitris Mitropoulos, and Diomidis Spinellis. 2021. Well-typed programs can go wrong: A study of typing-related bugs in JVM compilers. Proceedings of the ACM on Programming Languages5, OOPSLA (2021), 1–30. doi:10.1145/3485500",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Stefanos Chaliasos, Thodoris Sotiropoulos, Diomidis Spinellis, Arthur Gervais, Benjamin Livshits, and Dimitris Mitropoulos. 2022. Finding typing compiler bugs. InProgramming Language Design and Implementation. 183–198. doi:10.1145/3519939.3523427",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Craig Chambers and David Ungar. 1989. Customization: Optimizing compiler technology for SELF, a dynamically-typed object-oriented programming language. InProgramming Language Design and Implementation. 146–160. doi:10.1145/ 73141.74831",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Junjie Chen, Yanwei Bai, Dan Hao, Yingfei Xiong, Hongyu Zhang, and Bing Xie. 2017. Learning to prioritize test programs for compiler testing. InInternational Conference on Software Engineering. 700–711. doi:10.1109/ICSE.2017.70",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Junjie Chen, Yanwei Bai, Dan Hao, Yingfei Xiong, Hongyu Zhang, Lu Zhang, and Bing Xie. 2016. Test case prioritization for compilers: A text-vector based approach. InInternational Conference on Software Testing, Verification and Validation. 266–277. doi:10.1109/ICST.2016.19",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Junjie Chen, Jibesh Patra, Michael Pradel, Yingfei Xiong, Hongyu Zhang, Dan Hao, and Lu Zhang. 2020. A survey of compiler testing.Comput. Surveys53, 1 (2020), 1–36. doi:10.1145/3363562",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Yuting Chen, Ting Su, Chengnian Sun, Zhendong Su, and Jianjun Zhao. 2016. Coverage-directed differential testing of JVM implementations. InProgramming Language Design and Implementation. 85–99. doi:10.1145/2908080.2908095",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Chromium. 2025. Octane 2.0 Benchmark Suite. Accessed Oct 9, 2025 from https://chromium.github.io/octane/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Keith D Cooper and Linda Torczon. 2022.Engineering a compiler. Morgan Kaufmann.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Timothy Cramer, Richard Friedman, Terrence Miller, David Seberger, Robert Wilson, and Mario Wolczko. 1997. Compiling Java just in time.IEEE Micro17, 3 (1997), 36–43. doi:10.1109/40.591653",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Jan de Mooij, Matthew Gaudet, Iain Ireland, Nathan Henderson, and J Nelson Amaral. 2023. CacheIR: The benefits of a structured representation for inline caches. InInternational Conference on Managed Programming Languages and Runtimes. 34–46. doi:10.1145/3617651.3622979",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Daniele Cono D’Elia and Camil Demetrescu. 2018. On-stack replacement, distilled. InProgramming Language Design and Implementation. 166–180. doi:10.1145/3296979.3192396",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "L Peter Deutsch and Allan M Schiffman. 1984. Efficient implementation of the Smalltalk-80 system. InSymposium on Principles of Programming Languages. 297–302. doi:10.1145/800017.800542 Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 137. Publication date: April 2026. Understanding and Finding JIT Compiler Performance Bugs 137:27",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Kyle Dewey, Jared Roesch, and Ben Hardekopf. 2015. Fuzzing the Rust typechecker using CLP (T). InInternational Conference on Automated Software Engineering. 482–493. doi:10.1109/ASE.2015.65",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Christian Dietrich, Valentin Rothberg, Ludwig Füracker, Andreas Ziegler, and Daniel Lohmann. 2017. {cHash}: Detection of redundant compilations via{AST}hashing. InUSENIX Annual Technical Conference. 527–538.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Yuyou Fan and John Regehr. 2024. High-Throughput, Formal-Methods-Assisted Fuzzing for LLVM. InInternational Symposium on Code Generation and Optimization. 349–358. doi:10.1109/CGO57630.2024.10444854",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Olivier Flückiger, Gabriel Scherer, Ming-Ho Yee, Aviral Goel, Amal Ahmed, and Jan Vitek. 2017. Correctness of speculative optimizations with dynamic deoptimization.Proceedings of the ACM on Programming Languages2, POPL (2017), 1–28. doi:10.1145/3158137",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Fengjuan Gao, Hongyu Chen, Yuewei Zhou, and Ke Wang. 2024. Shoot yourself in the foot–efficient code causes inefficiency in compiler optimizations. InInternational Conference on Automated Software Engineering. 1846–1857. doi:10.1145/3691620.3695548",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Andy Georges, Dries Buytaert, and Lieven Eeckhout. 2007. Statistically rigorous Java performance evaluation.ACMSIGPLAN Notices42, 10 (2007), 57–76. doi:10.1145/1297027.1297033",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Liang Gong, Michael Pradel, and Koushik Sen. 2015. JITProf: Pinpointing JIT-unfriendly JavaScript code. InInternational Symposium on the Foundations of Software Engineering. 357–368. doi:10.1145/2786805.2786831",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Google Corporation and/or its affiliates. 2025. Addressing Performance Regressions. Accessed March 11, 2025 from https://chromium.googlesource.com/chromium/src/+/master/docs/speed/addressing_performance_regressions.md.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Google Corporation and/or its affiliates. 2025. Chrome Issues for V8 Perf Regression. Accessed Oct 3, 2025 from https://issues.chromium.org/issues?q=%22%5BV8%20JavaScript%20Perf%5D%22.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Google Corporation and/or its affiliates. 2025. Chromium Issue Tracker. Accessed March 10, 2025 from https: //issues.chromium.org/issues.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Samuel Groß, Simon Koch, Lukas Bernhard, Thorsten Holz, and Martin Johns. 2023. FUZZILLI: Fuzzing for JavaScript JIT compiler vulnerabilities. InThe Symposium on Network and Distributed System Security. doi:10.14722/ndss.2023.24290",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Xue Han and Tingting Yu. 2016. An empirical study on performance bugs for highly configurable software systems. In International Symposium on Empirical Software Engineering and Measurement. 1–10. doi:10.1145/2961111.2962602",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Xue Han, Tingting Yu, and David Lo. 2018. Perflearner: Learning from bug reports to understand and generate performance test frames. InInternational Conference on Automated Software Engineering. 17–28. doi:10.1145/3238147. 3238204",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Haoxiang Jia, Ming Wen, Zifan Xie, Xiaochen Guo, Rongxin Wu, Maolin Sun, Kang Chen, and Hai Jin. 2023. Detecting JVMJIT compiler bugs via exploring two-dimensional input spaces. InInternational Conference on Software Engineering. 43–55. doi:10.1109/ICSE48619.2023.00016",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Shan Jiang, Pranoy Kovuri, David Tao, and Zhixun Tan. 2026. CASCADE: LLM-Powered JavaScript deobfuscator at Google. InInternational Conference on Software Engineering, Software Engineering in Practice. doi:10.1145/3786583. 3786873",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Shan Jiang, Chenguang Zhu, and Sarfraz Khurshid. 2026. OBsmith: LLM-Powered JavaScript obfuscator testing. In OOPSLA. doi:10.1145/3798204",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Guoliang Jin, Linhai Song, Xiaoming Shi, Joel Scherpelz, and Shan Lu. 2012. Understanding and detecting real-world performance bugs.ACMSIGPLAN Notices47, 6 (2012), 77–88. doi:10.1145/2345156.2254075",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. Apollo: Automatic detection and diagnosis of performance regressions in database systems.Proceedings of the VLDB Endowment13, 1 (2019), 57–70. doi:10.14778/ 3357377.3357382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Motohiro Kawahito, Hideaki Komatsu, and Toshio Nakatani. 2000. Effective null pointer check elimination utilizing hardware trap.ACMSIGPLAN Notices35, 11 (2000), 139–149. doi:10.1145/356989.357002",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Prasad A Kulkarni. 2011. JIT compilation policy for modern machines. InInternational Conference on Object-Oriented Programming, Systems, Languages, and Applications. 773–788. doi:10.1145/2048066.2048126",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs.ACM Sigplan Notices49, 6 (2014), 216–226. doi:10.1145/2666356.2594334",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Vu Le, Chengnian Sun, and Zhendong Su. 2015. Finding deep compiler bugs via guided stochastic program mutation. Acm Sigplan Notices50, 10 (2015), 386–399. doi:10.1145/2858965.2814319",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Caroline Lemieux, Rohan Padhye, Koushik Sen, and Dawn Song. 2018. Perffuzz: Automatically generating pathological inputs. InInternational Symposium on Software Testing and Analysis. 254–265. doi:10.1145/3213846.3213874",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Xavier Leroy. 2009. Formal verification of a realistic compiler.Commun. ACM52, 7 (2009), 107–115. doi:10.1145/ 1538788.1538814",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Ao Li, Rohan Padhye, and Vyas Sekar. 2025. SPIDER: Fuzzing for Stateful Performance Issues in the ONOS SoftwareDefined Network Controller. InInternational Conference on Software Testing, Verification and Validation. 1–12. doi:10. 1109/ICST62969.2025.10988999 Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 137. Publication date: April 2026. 137:28 Zijian Yi, Cheng Ding, August Shi, and Milo",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Cong Li, Yanyan Jiang, Chang Xu, and Zhendong Su. 2023. Validating JIT compilers via compilation space exploration. InSymposium on Operating Systems Principles. 66–79. doi:10.1145/3600006.3613140",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Penghui Li, Yinxi Liu, and Wei Meng. 2021. Understanding and detecting performance bugs in markdown compilers. InInternational Conference on Automated Software Engineering. 892–904. doi:10.1109/ASE51524.2021.9678611",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Shaohua Li, Theodoros Theodoridis, and Zhendong Su. 2024. Boosting compiler testing by injecting real-world code. Proceedings of the ACM on Programming Languages8, PLDI (2024), 223–245. doi:10.1145/3656386",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Zixi Liu, Yang Feng, Yunbo Ni, Shaohua Li, Xizhe Yin, Qingkai Shi, Baowen Xu, and Zhendong Su. 2025. An empirical study of Rust-specific bugs in the rustc compiler.Proceedings of the ACM on Programming Languages9, OOPSLA2 (2025), 3869–3896. doi:10.1145/3763800",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Zhibo Liu, Dongwei Xiao, Zongjie Li, Shuai Wang, and Wei Meng. 2023. Exploring missed optimizations in webassembly optimizers. InInternational Symposium on Software Testing and Analysis. 436–448. doi:10.1145/3597926.3598068",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Vsevolod Livinskii, Dmitry Babokin, and John Regehr. 2023. Fuzzing loop optimizations in compilers for C++ and dataparallel languages.Proceedings of the ACM on Programming Languages7, PLDI (2023), 1826–1847. doi:10.1145/3591295",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Michaël Marcozzi, Qiyi Tang, Alastair F Donaldson, and Cristian Cadar. 2019. Compiler fuzzing: How much does it matter?Proceedings of the ACM on Programming Languages3, OOPSLA (2019), 1–29. doi:10.1145/3360581",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "William M McKeeman. 1998. Differential testing for software.Digital Technical Journal10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Benedikt Meurer. 2024. An Introduction to Speculative Optimization in V8. Accessed March 2, 2025 from https: //benediktmeurer.de/2017/12/13/an-introduction-to-speculative-optimization-in-v8/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Tipp Moseley, Dirk Grunwald, and Ramesh Peri. 2009. Optiscope: Performance accountability for optimizing compilers. InInternational Symposium on Code Generation and Optimization. 254–264. doi:10.1109/CGO.2009.26",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Mozilla Corporation and/or its affiliates. 2025. Bugzilla. Accessed March 10, 2025 from https://bugzilla.mozilla.org/ home.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Mozilla Corporation and/or its affiliates. 2025. SpiderMonkey. Accessed Oct 6, 2025 from https://firefox-sourcedocs.mozilla.org/js/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Oracle Corporation and/or its affiliates. 2025. Graal Issue Tracker. Accessed March 10, 2025 from https://github.com/ oracle/graal/issues.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Oracle Corporation and/or its affiliates. 2025. HotSpot Compilation Policy. Accessed March 11, 2025 from https: //github.com/openjdk/jdk/blob/master/src/hotspot/share/compiler/compilationPolicy.hpp.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Oracle Corporation and/or its affiliates. 2025. HotSpot Glossary of Terms. Accessed March 10, 2025 from https: //openjdk.org/groups/hotspot/docs/HotSpotGlossary.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Oracle Corporation and/or its affiliates. 2025. JDK Bug System. Accessed March 10, 2025 from https://bugs.openjdk. java.net/browse/JDK.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Alberto Parravicini and Rene Mueller. 2021. The cost of speculation: Revisiting overheads in the V8 JavaScript engine. InIEEE International Symposium on Workload Characterization. 13–23. doi:10.1109/IISWC53511.2021.00013",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Andrej Pečimúth, David Leopoldseder, and Petr Tůma. 2023. Diagnosing Compiler Performance by Comparing Optimization Decisions. InInternational Conference on Managed Programming Languages and Runtimes. 47–61. doi:10. 1145/3617651.3622994",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Guillermo Polito, Stéphane Ducasse, and Pablo Tesone. 2022. Interpreter-guided differential JIT compiler unit testing. InProgramming Language Design and Implementation. 981–992. doi:10.1145/3519939.3523457",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Aleksandar Prokopec, Andrea Rosà, David Leopoldseder, Gilles Duboscq, Petr Tůma, Martin Studener, Lubomír Bulej, Yudi Zheng, Alex Villazón, Doug Simon, Thomas Würthinger, and Walter Binder. 2019. Renaissance: benchmarking suite for parallel applications on the JVM. InProgramming Language Design and Implementation. 31–47. doi:10.1145/ 3314221.3314637",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Python Enhancement Proposals. 2024. PEP 744 - JIT Compilation. Accessed March 2, 2025 from https://peps.python. org/pep-0744/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Mohammad R. Haghighat. 2018. Java Fuzzer. Accessed March 11, 2025 from https://github.com/AzulSystems/ JavaFuzzer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "John Regehr, Yang Chen, Pascal Cuoq, Eric Eide, Chucky Ellison, and Xuejun Yang. 2012. Test-case reduction for C compiler bugs. InProgramming Language Design and Implementation. 335–346. doi:10.1145/2345156.2254104",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Marija Selakovic and Michael Pradel. 2016. Performance issues and optimizations in JavaScript: an empirical study. In International Conference on Software Engineering. 61–72. doi:10.1145/2884781.2884829",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Naomi Smith, Abhishek Sharma, John Renner, David Thien, Fraser Brown, Hovav Shacham, Ranjit Jhala, and Deian Stefan. 2024. Icarus: Trustworthy just-in-time compilers with symbolic meta-execution. InSymposium on Operating Systems Principles. 473–487. doi:10.1145/3694715.3695949",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Gabriel Southern and Jose Renau. 2016. Overhead of deoptimization checks in the V8 JavaScript engine. InIEEE International Symposium on Workload Characterization. 1–10. doi:10.1109/IISWC.2016.7581268 Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 137. Publication date: April 2026. Understanding and Finding JIT Compiler Performance Bugs 137:29",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Chengnian Sun, Vu Le, Qirun Zhang, and Zhendong Su. 2016. Toward understanding compiler bugs in GCC and LLVM. InInternational Symposium on Software Testing and Analysis. 294–305. doi:10.1145/2931037.2931074",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Chengnian Sun, Yuanbo Li, Qirun Zhang, Tianxiao Gu, and Zhendong Su. 2018. Perses: Syntax-guided program reduction. InInternational Conference on Software Engineering. 361–371. doi:10.1145/3180155.3180236",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "The PyPy Team. 2025. PyPy. Accessed Oct 3, 2025 from https://pypy.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. InInternational Conference on Architectural Support for Programming Languages and Operating Systems. 697–709. doi:10.1145/3503222.3507764",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 81,
+      "text": "Theodoros Theodoridis and Zhendong Su. 2024. Refined input, degraded output: The counterintuitive world of compiler behavior. (2024), 671–691. doi:10.1145/3656404",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Saeid Tizpaz-Niari, Pavol Čern `y, and Ashutosh Trivedi. 2020. Detecting and understanding real-world differential performance bugs in machine learning libraries. InInternational Symposium on Software Testing and Analysis. 189–199. doi:10.1145/3395363.3404540",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Luca Della Toffola, Michael Pradel, and Thomas R Gross. 2018. Synthesizing programs that expose performance bottlenecks. InInternational Symposium on Code Generation and Optimization. 314–326. doi:10.1145/3168830",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Luca Traini, Federico Di Menna, and Vittorio Cortellessa. 2024. AI-driven Java performance testing: Balancing result quality with testing time. InInternational Conference on Automated Software Engineering. 443–454. doi:10.1145/3691620. 3695017",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "V8. 2021. Sparkplug — a non-optimizing JavaScript compiler. Accessed March 10, 2025 from https://v8.dev/blog/ sparkplug.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "V8. 2023. Maglev-V8’s Fastest Optimizing JIT. Accessed March 10, 2025 from https://v8.dev/blog/maglev.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "V8. 2025. TurboFan. Accessed March 10, 2025 from https://v8.dev/docs/turbofan.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "Liam Wachter, Julian Gremminger, Christian Wressnegger, Mathias Payer, and Flavio Toffalini. 2025. DUMPLING: Fine-grained differential JavaScript engine fuzzing. (2025). doi:10.14722/ndss.2025.241411",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "Junjie Wang, Zhiyi Zhang, Shuang Liu, Xiaoning Du, and Junjie Chen. 2023. FuzzJIT: Oracle-Enhanced fuzzing for JavaScript engine JIT compiler. InUSENIX Security Symposium. 1865–1882.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "Webkit. 2025. JetStream 2 Benchmark Suite. Accessed Oct 9, 2025 from https://browserbench.org/JetStream2.0/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "Wikipedia. 2025. Hardware performance counter. Accessed Oct 9, 2025 from https://en.wikipedia.org/wiki/Hardware_ performance_counter.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "Wai Kin Wong, Dongwei Xiao, Cheuk Tung Lai, Yiteng Peng, Daoyuan Wu, and Shuai Wang. 2025. Extraction and mutation at a high level: Template-based fuzzing for JavaScript engines.Proceedings of the ACM on Programming Languages9, OOPSLA2 (2025), 2898–2926. doi:10.1145/3763154",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In Programming Language Design and Implementation. 283–294. doi:10.1145/1993316.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "Shahed Zaman, Bram Adams, and Ahmed E Hassan. 2012. A qualitative study on performance bugs. InInternational Conference on Mining Software Repositories. 199–208. doi:10.1109/MSR.2012.6224281",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "Zhiqiang Zang, Nathaniel Wiatrek, Milos Gligoric, and August Shi. 2022. Compiler Testing using Template Java Programs. InInternational Conference on Automated Software Engineering. 23:1–23:13. doi:10.1145/3551349.3556958",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "Zhiqiang Zang, Fu-Yao Yu, Aditya Thimmaiah, August Shi, and Milos Gligoric. 2024. Java JIT Testing with Template Extraction. InInternational Symposium on the Foundations of Software Engineering. 1129–1151. doi:10.1145/3643777",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "Yingquan Zhao, Zan Wang, Junjie Chen, Mengdi Liu, Mingyuan Wu, Yuqun Zhang, and Lingming Zhang. 2022. History-driven test program synthesis for JVM testing. InInternational Conference on Software Engineering. 1133–1144. doi:10.1145/3510003.3510059",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 98,
+      "text": "Yudi Zheng, Lubomír Bulej, and Walter Binder. 2015. Accurate profiling in the presence of dynamic compilation.ACMSIGPLAN Notices50, 10 (2015), 433–450. doi:10.1145/2858965.2814281",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 99,
+      "text": "Yudi Zheng, Lubomír Bulej, and Walter Binder. 2017. An empirical study on deoptimization in the Graal compiler. In European Conference on Object-Oriented Programming. 30:1–30:30. doi:10.4230/LIPIcs.ECOOP.2017.30 Received 2025-10-10; accepted 2026-02-17 Proc. ACM Program. Lang., Vol. 10, No. OOPSLA1, Article 137. Publication date: April 2026.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13. doi:10.1145/3597503.3639076",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert",
+        "prints the DOI of the paper introducing cert"
+      ]
+    },
+    {
+      "number": 80,
+      "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. InInternational Conference on Architectural Support for Programming Languages and Operating Systems. 697–709. doi:10.1145/3503222.3507764",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 6,30,55,80,81]",
+      "technique": null,
+      "sentence": "There has also been prior work on identifyinglong compilationbugs in C [ 46] and Markdown [ 52] compilers and on reducing compilation time in C++ [ 1,27], as well as studies onhigh-order performance bugs that primarily target missed optimizations in AOT compilers [ 6,30,55,80,81].",
+      "context_before": "performance bugs in JIT compilers. Existing research on performance bugs [ 4,37,42,74,94] mostly focuses on performance bugs in application code that can be fixed by modifying the application source itself. In contrast, performance bugs in JIT compilers cannot be easily, sometimes not even possibly, resolved at the application level. While certain workarounds at the source-code level can make specific code patterns more amenable to JIT optimizations [ 32], it remains desirable to improve JIT compilers themselves to make them more flexible and robust, thereby reducing the burden on developers.",
+      "context_after": "One study that explores JIT compilers [ 68] largely treats them as AOT compilers, focusing solely on optimizationrelated issues. However, JIT compilers exhibit several unique aspects like profiling, speculation, and interactions with other runtime components that are not present in AOT compilers. These aspects can also introduce performance bugs, yet they remain unexplored by prior work. To bridge this gap in understanding JIT compiler performance bugs, we present the first indepth empirical study of JIT compiler performance bugs from real-world Java and JavaScript issue trackers. Our analysi",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6774,
+      "cited_reference": {
+        "number": 80,
+        "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. InInternational Conference on Architectural Support for Programming Languages and Operating Systems. 697–709. doi:10.1145/3503222.3507764",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 5,43]",
+      "technique": "cert",
+      "sentence": "Other domain-specific tools target specialized systems, such as for database engines [ 5,43] and for machine learning libraries [ 82].",
+      "context_before": "studies have also investigated performance bugs from various perspectives [ 4,37,74,94], but all primarily focus on application-level software rather than compiler infrastructures. In addition to empirical analyses, several automated techniques have been proposed to detect performance bugs. Fuzzing-based approaches such as JITProf [ 32], PerfFuzz [ 48], and others [ 38,50, 83] generate diverse test inputs to expose pathological performance behaviors. Li et al. [ 52] studied long compilationbugs in markdown compilers, emphasizing inefficiencies during compilation rather than runtime execution.",
+      "context_after": "Our work differs from these in that we focus on detecting performance bugs in JIT compilers themselves, which are not caused by the application source code but by the compiler’s optimization and code generation phases. Compiler testing. Extensive research has focused on testing functional correctness of both AOT and JIT compilers [ 18], spanning a wide range of languages such as C/C++ [ 28,46,47,53,56,93], Java [19, 39, 51, 72, 95–97], JavaScript [36, 88, 89, 92], and Smalltalk [69]. Zhao et al. [ 97] synthesized realistic test programs by mining code snippets from real-world repositories to",
+      "section": "6.5 Discovered Bugs",
+      "page": 24,
+      "char_offset": 79629,
+      "cited_reference": {
+        "number": 5,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering. 1–13. doi:10.1145/3597503.3639076",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert",
+          "prints the DOI of the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 80,81]",
+      "technique": null,
+      "sentence": "[ 80,81] introduced techniques using dead-code markers and refined information to assess how effectively compilers eliminate redundant computations or exploit precise data-flow facts.",
+      "context_before": "y extended to other semantics-preserving code transformation tools such as JavaScript obfuscators and deobfuscators [40, 41]. Unlike this prior work, we are the first to focus on understanding and detecting performance bugs in JIT compilers. Finding missed optimizations in compilers. Moseley et al. [ 60] investigated performance accountability in optimizing compilers, aiming to better attribute performance effects to specific optimization decisions. Barany [ 6] applied differential testing to identify missed optimizations by analyzing discrepancies in the generated binaries. Theodoridis et al.",
+      "context_after": "Liu et al. [ 55] analyzed missed optimizations in WebAssembly compilers, while Gao et al. [ 30] proposed a source-level transformation framework to detect compiler-induced performance differences. The techniques used in prior works to derive semantically equivalent programs could potentially be extended to testing of JIT compilers. Jitterycan naturally incorporate such similar program pairs by treating them as two distinct configurations. Pečimúth et al. [ 68] studied Java JIT compilers by representing optimizations as trees and comparing them via tree edit distance to identify behavioral dif",
+      "section": "6.5 Discovered Bugs",
+      "page": 24,
+      "char_offset": 81846,
+      "cited_reference": {
+        "number": 80,
+        "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. 2022. Finding missed optimizations through the lens of dead code elimination. InInternational Conference on Architectural Support for Programming Languages and Operating Systems. 697–709. doi:10.1145/3503222.3507764",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:19:46Z",
+    "is_model_written": true,
+    "summary": "The first study of performance bugs in just-in-time compilers, which prior automated work had not targeted, having focused on functional bugs. The authors manually analyse 191 bug reports across four Java and JavaScript JIT compilers to characterise triggers, manifestation patterns and root causes, then propose layered differential performance testing implemented as Jittery. It found 12 previously unknown performance bugs in Oracle HotSpot and Graal, 11 confirmed.",
+    "narrative": "The citations here are to papers by SQLancer's authors rather than to SQLancer itself, listing database engines among the specialised systems that domain-specific tools target.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3799227.json b/_data/papers/paper_doi_10_1145_3799227.json
new file mode 100644
index 0000000..3e2df5e
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3799227.json
@@ -0,0 +1,2562 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:22:49Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3799227",
+    "title": "A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Evaluation",
+    "authors": [
+      "Xiyue Gao",
+      "Zhuang Liu",
+      "Jiangtao Cui",
+      "Hui Li",
+      "Hui Zhang",
+      "Kewei Wei",
+      "Kankan Zhao"
+    ],
+    "year": 2026,
+    "venue": "ACM Computing Surveys",
+    "doi": "10.1145/3799227",
+    "arxiv_id": "2311.06728",
+    "s2_paper_id": "de2dbf2b041d2deaef7756f69ef5b3bdc5e599cc",
+    "url": "https://doi.org/10.1145/3799227",
+    "also_indexed_as": [
+      "paper:arxiv:2311.06728"
+    ]
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3799227",
+      "retrieved_at": "2026-09-10T15:22:49Z",
+      "chars": 117149,
+      "content_sha256": "sha256:a5eea109b0f9f907483324456c6571f7afc29ba17ef487f3704e0772dc99310f"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 36,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3831
+      },
+      {
+        "number": "2",
+        "title": "Preliminary",
+        "start": 13684
+      },
+      {
+        "number": "2.1",
+        "title": "Notations",
+        "start": 13833
+      },
+      {
+        "number": "2.2",
+        "title": "Basic Definitions",
+        "start": 14140
+      },
+      {
+        "number": "3",
+        "title": "Fuzzing Step-based Taxonomy",
+        "start": 19892
+      },
+      {
+        "number": "3.1",
+        "title": "Test Case Generator",
+        "start": 20949
+      },
+      {
+        "number": "3.2",
+        "title": "Oracle-based Comparator",
+        "start": 35383
+      },
+      {
+        "number": "3.3",
+        "title": "Execution Feedback",
+        "start": 44242
+      },
+      {
+        "number": "3.4",
+        "title": "Query Reducer",
+        "start": 49965
+      },
+      {
+        "number": "4",
+        "title": "DBMS Component-based Taxonomy",
+        "start": 51251
+      },
+      {
+        "number": "4.1",
+        "title": "Overall Fuzzing",
+        "start": 51947
+      },
+      {
+        "number": "4.2",
+        "title": "Transaction Testing",
+        "start": 60098
+      },
+      {
+        "number": "4.3",
+        "title": "Optimizer Testing",
+        "start": 62815
+      },
+      {
+        "number": "4.4",
+        "title": "Executor Testing",
+        "start": 64961
+      },
+      {
+        "number": "5",
+        "title": "Non-relational DBMS Fuzzing and New Techniques",
+        "start": 67013
+      },
+      {
+        "number": "5.3",
+        "title": "LLM in Fuzzing",
+        "start": 72917
+      },
+      {
+        "number": "6",
+        "title": "Benchmarks and Comparison",
+        "start": 73877
+      },
+      {
+        "number": "6.1",
+        "title": "Database Instances and Test Cases",
+        "start": 75194
+      },
+      {
+        "number": "6.2",
+        "title": "Open-source Fuzzers",
+        "start": 77387
+      },
+      {
+        "number": "6.3",
+        "title": "Detection Capabilities across Different Bug Types",
+        "start": 77679
+      },
+      {
+        "number": "6.4",
+        "title": "Evaluation Metrics",
+        "start": 78775
+      },
+      {
+        "number": "6.5",
+        "title": "Logic Bugs Detection Comparison",
+        "start": 81657
+      },
+      {
+        "number": "6.6",
+        "title": "Crash Detection Comparison",
+        "start": 87763
+      },
+      {
+        "number": "6.7",
+        "title": "Performance Bugs Detection Evaluation",
+        "start": 88605
+      },
+      {
+        "number": "7",
+        "title": "Conclusion and Challenges",
+        "start": 90694
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "1996. TPC-H. Retrieved March 20, 2026 from https://www.tpc.org",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2011. Data science revealed: A data-driven glimpse into the burgeoning new field. Retrieved March 20, 2026 from https://www.predictiveanalyticsworld.com/machinelearningtimes/august12/emc-data-science-study.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "2015. AFL: American Fuzzy Lop. Retrieved March 20, 2026 from http://lcamtuf.coredump.cx/afl/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "2015. SQLsmith. Retrieved March 20, 2026 from https://github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "2016. Honggfuzz. Retrieved March 20, 2026 from https://google.github.io/honggfuzz/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "2016. OSS-Fuzz: Continuous Fuzzing for Open Source Software. Retrieved March 20, 2026 from https://github.com/ google/oss-fuzz",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "2017. LibFuzzer-A Library For Coverage-guided Fuzz Testing. Retrieved March 20, 2026 from http://llvm.org/docs/ LibFuzzer.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "2021. CockroachDB Performance Bug Reports. Retrieved March 20, 2026 from https://github.com/cockroachdb/ cockroach/issues?q=performance",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "2021. PostgreSQL Performance Bug Reports. Retrieved March 20, 2026 from https://www.postgresql.org/search/?m= 1&q=performance&l=8&d=-1&s=r ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:32 X. Gao et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "2021. SCOTT schema. Retrieved March 20, 2026 from https://www.orafaq.com/wiki/SCOTT",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "2023. MySQL Bugs Home. Retrieved March 20, 2026 from https://bugs.mysql.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. 2010. Automated SQL query generation for systematic testing of database engines. In Proceedings of the International Conferenceon Automated SoftwareEngineering. ACM, 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In Proceedings of the International Conference on Software Engineering. ACM, 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. In Proceedings of the International Conference on Software Engineering. 1–13.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans. Proceedingsofthe ACM on Management of Data 2, 3 (2024), 1–26.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 16,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. 2007. A genetic approach for random testing of database systems. In Proceedings of the International Conference on Very Large Data Bases. VLDB Endowment, 1243–1251.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "B. Beizer. 1992. Software testing techniques. Software Testing Verification and Reliability 2, 4 (1992), 215–216.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Donald A. Berry and Bert Fristedt. 1985. Bandit problems: Sequential allocation of experiments (Monographs on statistics and applied probability). London: Chapman and Hall 5, 71-87 (1985), 7–7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Tim Blazytko, Matt Bishop, Cornelius Aschermann, Justin Cappos, Moritz Schlögel, Nadia Korshun, Ali Abbasi, Marco Schweighauser, Sebastian Schinzel, Sergej Schumilo, et al .2019. GRIMOIRE: Synthesizing structure while fuzzing. In Proceedings of the USENIX Security Symposium. USENIX Association, 1985–2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Nicolas Bruno and Surajit Chaudhuri. 2005. Flexible database generators. In ProceedingsoftheInternationalConference on Very Large Data Bases. VLDB Endowment, 1097–1107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "N. Bruno, S. Chaudhuri, and D. Thomas. 2006. Generating queries with cardinality constraints for DBMS testing. IEEE Transactions on Knowledge and Data Engineering 18, 12 (2006), 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Marcel Böhme, Van-Thuan Pham, Manh-Dung Nguyen, and Abhik Roychoudhury. 2017. Directed greybox fuzzing. In Proceedings of the Conference on Computer and Communications Security. ACM, 2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. 2016. Coverage-based greybox fuzzing as markov chain. InProceedings of the Conference on Computer and Communications Security. ACM, 1032–1043.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Tsong Y. Chen, Shing C. Cheung, and Shiu Ming Yiu. 2020. Metamorphic testing: A new approach for generating next test cases. arXiv:2002.12543. Retrieved from https://arxiv.org/abs/2002.12543",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Xinyue Chen, Chenglong Wang, and Alvin Cheung. 2020. Testing query execution engines with mutations. In Proceedings of the Workshop on Testing Database Systems. ACM, 1–5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Shumo Chu, Chenglong Wang, Konstantin Weitz, and Alvin Cheung. 2017. Cosette: An automated prover for SQL. In Proceedings of the Conference on Innovative Data Systems Research.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Ziyu Cui, Wensheng Dou, Qianwang Dai, Jiansen Song, Wei Wang, Jun Wei, and Dan Ye. 2022. Differentially testing database transactions for fun and profit. In Proceedings of the International Conference on Automated Software Engineering. ACM, 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Ziyu Cui, Wensheng Dou, Yu Gao, Rui Yang, Yingying Zheng, Jiansen Song, Yuan Feng, and Jun Wei. 2025. Simple testing can expose most critical transaction bugs: Understanding and detecting write-specific serializability violations in database systems. In Proceedings of the VLDB Endowment.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Wenqian Deng, Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. CONI: Detecting database connector bugs via state-aware test case generation. In Proceedings of the International Conference on Software Engineering. 667–678.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Wenjing Deng, Qiuyang Mang, Chengyu Zhang, and Manuel Rigger. 2024. Finding logic bugs in spatial database engines via affine equivalent inputs. Proceedings of the ACM on Management of Data 2, 6 (2024), 1–26.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "Wensheng Dou, Ziyu Cui, Qianwang Dai, Jiansen Song, Dong Wang, Yu Gao, Wei Wang, Jun Wei, Lei Chen, Hanmo Wang, et al .2023. Detecting isolation bugs via transaction oracle construction. In Proceedings of the International Conference on Software Engineering. IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Cundi Fang, Jie Liang, Zhiyong Wu, Jingzhou Fu, Zhouyang Jia, Chun Huang, Yu Jiang, and Shanshan Li. 2025. DepState: Detecting synchronization failure bugs in distributed database management systems. Proceedingsofthe ACM on Software Engineering 2, ISSTA (2025), 2001–2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-free DBMS fuzzing. In Proceedings of the International Conference on Automated Software Engineering. ACM, 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and detecting sql function bugs: Using simple boundary arguments to trigger hundreds of DBMS bugs. In Proceedings of the European Conference on Computer Systems. 1061–1076.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 35,
+      "text": "César A. Galindo-Legaria, Stefano Stefani, and Florian Waas. 2004. Query processing for SQL updates. In Proceedings of the International Conference on Management of Data. ACM, 844–849. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:33",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Xiyue Gao, Zhuang Liu, Yiran Shen, Hui Li, Yingfan Liu, Hongjun Xiao, Yanguo Peng, and Jiangtao Cui. 2025. Fucci: Database transaction fuzzing via random conflict construction and multilevel constraint solving. Proceedingsofthe VLDB Endowment 18, 6 (2025), 1879–1891.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Leo Giakoumakis and César A Galindo-Legaria. 2008. Testing SQL server’s query optimizer: Challenges, techniques and experiences. Data Engineering Bulletin 31, 1 (2008), 36–43.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Patrice Godefroid, Adam Kiezun, and Michael Y. Levin. 2008. Grammar-based whitebox fuzzing. In Proceedings of the Conference on Programming Language Design and Implementation. ACM, 206–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Eugene Goldberg and Yakov Novikov. 2007. BerkMin: A fast and robust SAT-solver. Discrete Applied Mathematics 155, 12 (2007), 1549–1561.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Goetz Graefe. 1993. Query evaluation techniques for large databases. ACM Computing Surveys 25, 2 (1993), 73–169.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Goetz Graefe. 1995. The cascades framework for query optimization. Data Engineering Bulletin 18, 3 (1995), 19–29.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Jim Gray, Prakash Sundaresan, Susanne Englert, Ken Baclawski, and Peter J. Weinberger. 1994. Quickly generating billion-record synthetic databases. In Proceedings of the International Conference on Management of Data. ACM, 243–252.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Zhongxian Gu, Mohamed A. Soliman, and Florian M. Waas. 2012. Testing the accuracy of query optimizers. In Proceedings of the Workshop on Testing Database Systems. ACM, 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023.Pinolo:Detectinglogicalbugsindatabasemanagementsystemswithapproximatequerysynthesis.In Proceedings of the USENIX Annual Technical Conference. 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Kenneth Houkjær, Kristian Torp, and Rico Wind. 2006. Simple and realistic data generation. In Proceedings of the International Conference on Very Large Data Bases. VLDB Endowment, 1243–1246.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "William E. Howden. 1978. Theoretical and empirical studies of program testing. IEEE Transactions on Software Engineering SE-4, 4 (1978), 293–298.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Jiaxin Hu and Rongxin Wu. 2025. SQLaw: Detecting bugs in GPU database management systems via rule-based differential execution. IEEE Transactions on Software Engineering 51, 7 (2025), 2144–2160.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Toshihide Ibaraki and Tiko Kameda. 1984. On the optimal nesting order for computing N-Relational joins. ACM Transactions on Database Systems 9, 3 (1984), 482–502.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Daniel Jackson. 2002. Alloy: A lightweight object modelling notation. ACM Transactions on Software Engineering and Methodology 11, 2 (2002), 256–290.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Yu Jiang, Jie Liang, Fuchen Ma, Yuanliang Chen, Chijin Zhou, Yuheng Shen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Shanshan Li, et al .2024. When fuzzing meets llms: Challenges and opportunities. In ProceedingsoftheACM International Conference on the Foundations of Software Engineering. 492–496.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In Proceedings of the International Conference on Software Engineering. 1–12.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 52,
+      "text": "Yuancheng Jiang, Jianing Wang, Chuqi Zhang, Roland Yap, Zhenkai Liang, and Manuel Rigger. 2025. Enhanced differential testing in emerging database systems. arXiv:2501.01236. Retrieved from https://arxiv.org/abs/2501.01236",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 53,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful fuzzing for database management systems with complex and valid SQL query generation. In ProceedingsoftheUSENIXSecuritySymposium. USENIX Association, 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via{graph-based }oracle construction. In Proceedings of the USENIX Symposium on Operating Systems Design and Implementation. 397–417.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 55,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting logic bugs in database engines via equivalent expression transformation. In Proceedings of the USENIX Symposium on Operating Systems Design and Implementation. 821–835.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "JinhoJung,HongHu,JoyArulraj,TaesooKim,andWoonhakKang.2019.APOLLO:Automaticdetectionanddiagnosis of performance regressions in database systems. Very Large Data Bases Endowment 13, 1 (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing graph database engines via query partitioning. In Proceedings of the ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 58,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O. Laleye, and Sarfraz Khurshid. 2008. Query-aware test generation using a relational constraint solver. In Proceedings of the International Conference on Automated Software Engineering. IEEE, 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "SarfrazKhurshidandDarkoMarinov.2004.TestEra:Specification-basedtestingofJavaprogramsusingSAT. Automated Software Engineering 11, 4 (2004), 403–434.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Donghun Lee, Sang K. Cha, and Arthur H. Lee. 2011. A performance anomaly detection and analysis framework for DBMS development. IEEE Transactions on Knowledge and Data Engineering 24, 8 (2011), 1345–1360. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:34 X. Gao et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2015. How good are query optimizers, really? Very Large Data Bases Endowment 9, 3 (2015), 204–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Caroline Lemieux and Koushik Sen. 2018. FairFuzz: A targeted mutation strategy for increasing greybox fuzz testing coverage. In Proceedings of the International Conference on Automated Software Engineering. ACM, 475–485.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Jiaqi Li, Ke Wang, Yaoguang Chen, Yajin Zhou, Lei Wu, and Jiashui Wang. 2023. Detecting dbms bugs with contextsensitive instantiation and multi-plan execution. arXiv:2312.04941. Retrieved from https://arxiv.org/abs/2312.04941",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Jiaqi Li, Yajin Zhou, and Lei Wu. 2024. Improving multitasking DBMS fuzzing With more accurate coverage and testcase trimming. IEEE Transactions on Dependable and Secure Computing 22, 3 (2024), 2756–2770.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Zhan Li, Olga Papaemmanouil, and Mitch Cherniack. 2016. Optmark: A Toolkit for Benchmarking Query Optimizers. InProceedings of the Information and Knowledge Management. ACM, 2155–2160.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-oriented DBMS fuzzing. In ProceedingsoftheInternationalConferenceonData Engineering. IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering dbms bugs via configuration-based equivalent transformation. In ProceedingsoftheInternationalConferenceonSoftware Engineering. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting logical bugs of DBMS with coverage-based guidance. In Proceedings of the USENIX Security Symposium. USENIX Association, 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Li Lin, Qinglin Zhu, Hongqiao Chen, Zhuangda Wang, Rongxin Wu, and Xiaoheng Xie. 2025. QTRAN: Extending metamorphic-oracle based logical bug detection techniques for multiple-DBMS dialect support. Proceedingsofthe ACM on Software Engineering 2, ISSTA (2025), 731–752.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting bugs in cypher graph database engines. In ProceedingsoftheInternationalSymposiumonSoftwareTestingandAnalysis. 163–174.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Shuang Liu, Junhao Lan, Xiaoning Du, Jiyuan Li, Wei Lu, Jiajun Jiang, and Xiaoyong Du. 2024. Testing graph database systems with graph-state persistence oracle. In Proceedings of the ACMSIGSOFT International Symposium on Software Testing and Analysis. 666–677.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systemsusingequivalentqueries.In ProceedingsoftheInternationalConferenceonSoftwareEngineering .ACM,225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Eric Lo, Carsten Binnig, Donald Kossmann, M. Tamer Özsu, and Wing-Kai Hon. 2010. A framework for testing DBMS features.Very Large Data Bases Endowment 19, 2 (2010), 203–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Ce Lyu, Minghao Zhao, Yanhao Wang, and Liang Jie. 2025. LLM-based dynamic differential testing for database connectors with reinforcement learning-guided prompt election. In ProceedingsoftheInternationalConferenceon Automated Software Engineering. ACM, 3814–3818.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding logic bugs in graph-processing systems via graph-cutting. Proceedings of the ACM on Management of Data 3, 3 (2025), 1–27.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 76,
+      "text": "Qiuyang Mang, Aoyang Fang, Boxi Yu, Hanfei Chen, and Pinjia He. 2024. Testing graph database systems via equivalent query rewriting. In Proceedings of the International Conference on Software Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Darko Marinov. 2010. Session Details: Technical Session 8: Concurrency and differential testing. In Proceedings of the International Symposium on Software Testing and Analysis. ACM.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "William M. McKeeman. 1998. Differential testing for software. Digital Technical Journal 10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Phil Mcminn, Chris J. Wright, and Gregory M. Kapfhammer. 2015. The effectiveness of test coverage criteria for relational database schema integrity constraints. ACMTransactionsonSoftwareEngineeringandMethodology 25, 1 (2015), 1–49.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "KaimingMi,ChunxiZhang,WeiningQian,andRongZhang.2021.Artemis:Anautomatictestsuitegeneratorforlarge scale OLAP database. In Proceedings of the Benchmarking, Measuring, and Optimizing: 3rd BenchCouncil International Symposium. Springer, 74–89.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "Barton P. Miller, Lars Fredriksen, and Bryan So. 1990. An empirical study of the reliability of UNIX utilities. Communications of the ACM 33, 12 (1990), 32–44.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. 2008. Generating targeted queries for database testing. In Proceedings of the International Conference on Management of Data. ACM, 499–510.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Jayashree Mohan, Ashlie Martinez, Soujanya Ponnapalli, Pandian Raju, and Vijay Chidambaram. 2018. Finding crash-consistency bugs with bounded black-box crash testing. In Proceedings of the USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 33–50.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "MySQL. 2023. The MySQL Test Framework. Retrieved March 20, 2026 from https://dev.mysql.com/doc/dev/mysqlserver/latest/PAGE_MYSQL_TEST_RUN.html ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:35",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "Andrea Neufeld, Guido Moerkotte, and Peter C Loekemann. 1993. Generating consistent test data: Restricting the search space by A generator formula. Very Large Data Bases Endowment 2, 2 (1993), 173–213.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Carlos Pacheco, Shuvendu K. Lahiri, Michael D. Ernst, and Thomas Ball. 2007. Feedback-directed random test generation. In Proceedings of the International Conference on Software Engineering. IEEE, 75–84.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "PingCap. 2023. go randgen. Retrieved March 20, 2026 from https://github.com/pingcap/go-randgen",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "Meikel Poess and John M. Stephens. 2004. Generating thousand benchmark queries in seconds. In Proceedings of the International Conference on Very Large Data Bases. VLDB Endowment, 1045–1053.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "PostgreSQL. 2023. Regression Tests. Retrieved March 20, 2026 from https://www.postgresql.org/docs/current/regress. html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 91,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 92,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedingsofthe USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 93,
+      "text": "ManuelRiggerandZhendongSu.2022.Intramorphictesting:Anewapproachtothetestoracleproblem.In Proceedings oftheInternationalSymposiumonNewIdeas,NewParadigms,andReflectionsonProgrammingandSoftware. ACM, 128–136.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 94,
+      "text": "Shetal Shah, S. Sudarshan, Suhas Kajbaje, Sandeep Patidar, Bhanu Pratap Gupta, and Devang Vira. 2011. Generating test data for killing SQL mutants: A constraint-based approach. In Proceedings of the International Conference on Data Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "Donald R. Slutz. 1998. Massive stochastic testing of SQL. In Proceedings of the International Conference on Very Large Data Bases. VLDB Endowment, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing database systems via differential query execution. In ProceedingsoftheInternationalConferenceonSoftware Engineering.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "Jiansen Song, Wensheng Dou, Yu Gao, Ziyu Cui, Yingying Zheng, Dong Wang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting metadata-related logic bugs in database systems via raw database construction. Proceedingsofthe VLDB Endowment 17, 8 (2024), 1884–1897.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 98,
+      "text": "Jiansen Song, Wensheng Dou, Yingying Zheng, Yu Gao, Ziyu Cui, Wei Wang, and Jun Wei. 2025. Detecting schemarelated logic bugs in relational dbmss via equivalent database construction. Proceedings of the VLDB Endowment 18, 7 (2025) 2281–2294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 99,
+      "text": "SQLite. 2023. How SQLite Is Tested. Retrieved March 20, 2026 from https://sqlite.org/testing.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 100,
+      "text": "Keith Stobie. 2005. Too darned big to test: Testing large systems is a daunting task, but there are steps we can take to ease the pain. Queue3, 1 (2005), 30–37.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 101,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting logic bugs of join optimizations in DBMS.International Conference on Management of Data 1, 1 (2023), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 102,
+      "text": "PatrickValduriezandScottDanforth.1992.FunctionalSQL(FSQL),an SQLupward-compatibledatabaseprogramming language. InformationSciences 62, 3 (1992), 183–203.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 103,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. 2019. Superion: Grammar-aware greybox fuzzing. In Proceedingsof the International Conference on Software Engineering. IEEE, 724–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 104,
+      "text": "MingzheWang,ZhiyongWu,XinyiXu,JieLiang,ChijinZhou,HuafengZhang,andYuJiang.2021.Industrypracticeof coverage-guided enterprise-level DBMS fuzzing. In ProceedingsoftheInternationalConferenceonSoftwareEngineering: Software Engineering in Practice. IEEE, 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 105,
+      "text": "Shihao Wen, Peng Jia, Pin Yang, and Chi Hu. 2023. Squill: Testing DBMS with correctness feedback and accurate instantiation. Applied Sciences 13, 4 (2023), 2519.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 106,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. Puppy: Finding performance degradation bugs in dbmss via limited-optimization plan construction. In Proceedings of the International Conference on Software Engineering. 679–690.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 107,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. 2022. Unicorn: Detect runtime errors in time-series databases with hybrid input synthesis. In Proceedings of the International Symposium on Software Testing and Analysis.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 108,
+      "text": "Jiaqi Yan, Qiuye Jin, Shrainik Jain, Stratis D. Viglas, and Allison Lee. 2018. Snowtrail: Testing with production queries on a cloud database. In Proceedings of the Workshop on Testing Database Systems. ACM, 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 109,
+      "text": "Rui Yang, Yingying Zheng, Lei Tang, Wensheng Dou, Wei Wang, and Jun Wei. 2023. Randomized differential testing of RDF stores. In International Conference on Software Engineering: Companion Proceedings. IEEE, 136–140. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:36 X. Gao et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 110,
+      "text": "Yupeng Yang, Yongheng Chen, Rui Zhong, Jizhou Chen, and Wenke Lee. 2024. Towards generic database management system fuzzing. In Proceedings of the USENIX Security Symposium. 901–918.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 111,
+      "text": "Zijing Yin, Si Liu, and David Basin. 2025. Testing graph databases with synthesized queries. Proceedings of the ACM on Management of Data 3, 4 (2025), 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 112,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant optimization driven database system testing. ProceedingsoftheACM on Management of Data 3, 1 (2025), 1–24.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 113,
+      "text": "Yushan Zhang, Peisen Yao, Rongxin Wu, and Charles Zhang. 2021. Duplicate-sensitivity guided transformation synthesis for DBMS correctness bug detection. arXiv:2107.03660. Retrieved from https://arxiv.org/abs/2107.03660",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 114,
+      "text": "Yingying Zheng, Wensheng Dou, Lei Tang, Ziyu Cui, Yu Gao, Jiansen Song, Liang Xu, Jiaxin Zhu, Wei Wang, Jun Wei, et al .2024. Testing gremlin-based graph database systems via query disassembling. In ProceedingsoftheACM SIGSOFT International Symposium on Software Testing and Analysis. 1695–1707.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 115,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding bugs in gremlin-based graph database systems via randomized differential testing. In Proceedings of the International Symposium on Software Testing and Analysis. ACM, 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 116,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the Conference on Computer and Communications Security. ACM, 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 117,
+      "text": "SuyangZhongandManuelRigger.2025.TestingDatabaseSystemswithLargeLanguageModelSynthesizedFragments. arXiv:2505.02012. Retrieved from https://arxiv.org/abs/2505.02012",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 118,
+      "text": "Zenghui Zhou, Zheng Zheng, Tsong Yueh Chen, Jinyi Zhou, and Kun Qiu. 2021. Follow-up test cases are better than source test cases in metamorphic testing: A preliminary study. In ProceedingsoftheInternationalWorkshopon Metamorphic Testing. IEEE, 69–74.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 119,
+      "text": "Zeyang Zhuang, Penghui Li, Pingchuan Ma, Wei Meng, and Shuai Wang. 2023. Testing graph database systems via graph-aware metamorphic relations. Proceedings of the VLDB Endowment 17, 4 (2023), 836–848. Received 17 November 2023; revised 20 December 2025; accepted 10 February 2026 ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 13,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In Proceedings of the International Conference on Software Engineering. ACM, 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 14,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. In Proceedings of the International Conference on Software Engineering. 1–13.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    },
+    {
+      "number": 15,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans. Proceedingsofthe ACM on Management of Data 2, 3 (2024), 1–26.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 30,
+      "text": "Wenjing Deng, Qiuyang Mang, Chengyu Zhang, and Manuel Rigger. 2024. Finding logic bugs in spatial database engines via affine equivalent inputs. Proceedings of the ACM on Management of Data 2, 6 (2024), 1–26.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 34,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and detecting sql function bugs: Using simple boundary arguments to trigger hundreds of DBMS bugs. In Proceedings of the European Conference on Computer Systems. 1061–1076.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 51,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland H. C. Yap, Zhenkai Liang, and Manuel Rigger. 2024. Detecting logic bugs in graph database management systems via injective and surjective graph query transformation. In Proceedings of the International Conference on Software Engineering. 1–12.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 52,
+      "text": "Yuancheng Jiang, Jianing Wang, Chuqi Zhang, Roland Yap, Zhenkai Liang, and Manuel Rigger. 2025. Enhanced differential testing in emerging database systems. arXiv:2501.01236. Retrieved from https://arxiv.org/abs/2501.01236",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 54,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via{graph-based }oracle construction. In Proceedings of the USENIX Symposium on Operating Systems Design and Implementation. 397–417.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 57,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing graph database engines via query partitioning. In Proceedings of the ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 75,
+      "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding logic bugs in graph-processing systems via graph-cutting. Proceedings of the ACM on Management of Data 3, 3 (2025), 1–27.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 90,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 91,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 92,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedingsofthe USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 93,
+      "text": "ManuelRiggerandZhendongSu.2022.Intramorphictesting:Anewapproachtothetestoracleproblem.In Proceedings oftheInternationalSymposiumonNewIdeas,NewParadigms,andReflectionsonProgrammingandSoftware. ACM, 128–136.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 112,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant optimization driven database system testing. ProceedingsoftheACM on Management of Data 3, 1 (2025), 1–24.",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 117,
+      "text": "SuyangZhongandManuelRigger.2025.TestingDatabaseSystemswithLargeLanguageModelSynthesizedFragments. arXiv:2505.02012. Retrieved from https://arxiv.org/abs/2505.02012",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "prints the arXiv id of the paper introducing shqvel"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "proposed several new fuzzing methods, including NoREC [90], TLP [91], and PQS [ 92].",
+      "context_before": "straint-solving testing [ 12] was proposed as another way to obtain the test oracle in DBMS fuzzing, which provides the ground truth for the test cases. In 2015, SQLsmith [ 4] was released, which still generates statements based on predefined rules and can also connect to reference databases to serve as test oracles. Thanks to the open-source factor, compared to the SQL Server fuzzing tool, SQLsmith has rapidly attracted attention from both academia and industry. In 2020,Squirrel[ 116]firstintroducedexecutionfeedbacktoimprovecodecoverageincrashdetection fuzzing. In the same year, Rigger et al.",
+      "context_after": "NoREC and TLP apply metamorphic testing to the database domain, enabling fuzzing on a single DBMS. PQS is a constraint-solving testing method, and its oracle only requires that the execution result contain one row of the solving outcome, thereby accelerating the solving process compared to the former approach, which requires matching all the rows. In 2022, Liang et al. introduced SQLRight [ 68], a metamorphic testing method combined with execution feedback, and it is scalable to allow new oracles. In 2023, Tang et al. presented a constraint-solving testing method, TQS [ 101], which incorporat",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6438,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC and TLP apply metamorphic testing to the database domain, enabling fuzzing on a single DBMS.",
+      "context_before": "n DBMS fuzzing, which provides the ground truth for the test cases. In 2015, SQLsmith [ 4] was released, which still generates statements based on predefined rules and can also connect to reference databases to serve as test oracles. Thanks to the open-source factor, compared to the SQL Server fuzzing tool, SQLsmith has rapidly attracted attention from both academia and industry. In 2020,Squirrel[ 116]firstintroducedexecutionfeedbacktoimprovecodecoverageincrashdetection fuzzing. In the same year, Rigger et al. proposed several new fuzzing methods, including NoREC [90], TLP [91], and PQS [ 92].",
+      "context_after": "PQS is a constraint-solving testing method, and its oracle only requires that the execution result contain one row of the solving outcome, thereby accelerating the solving process compared to the former approach, which requires matching all the rows. In 2022, Liang et al. introduced SQLRight [ 68], a metamorphic testing method combined with execution feedback, and it is scalable to allow new oracles. In 2023, Tang et al. presented a constraint-solving testing method, TQS [ 101], which incorporated subgraph isomorphism search into DBMS fuzzing, significantly enhancing bug detection efficiency.",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6523,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS is a constraint-solving testing method, and its oracle only requires that the execution result contain one row of the solving outcome, thereby accelerating the solving process compared to the former approach, which requires matching all the rows.",
+      "context_before": "ased, which still generates statements based on predefined rules and can also connect to reference databases to serve as test oracles. Thanks to the open-source factor, compared to the SQL Server fuzzing tool, SQLsmith has rapidly attracted attention from both academia and industry. In 2020,Squirrel[ 116]firstintroducedexecutionfeedbacktoimprovecodecoverageincrashdetection fuzzing. In the same year, Rigger et al. proposed several new fuzzing methods, including NoREC [90], TLP [91], and PQS [ 92]. NoREC and TLP apply metamorphic testing to the database domain, enabling fuzzing on a single DBMS.",
+      "context_after": "In 2022, Liang et al. introduced SQLRight [ 68], a metamorphic testing method combined with execution feedback, and it is scalable to allow new oracles. In 2023, Tang et al. presented a constraint-solving testing method, TQS [ 101], which incorporated subgraph isomorphism search into DBMS fuzzing, significantly enhancing bug detection efficiency. In 2024, Ba et al. and Liang et al. respectively proposed DQP [ 15] and Mozi [ 67], further extending metamorphic testing to different execution paths of the same query. DBMS fuzzing techniques can be classified from different perspectives. —One clas",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6622,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "respectively proposed DQP [ 15] and Mozi [ 67], further extending metamorphic testing to different execution paths of the same query.",
+      "context_before": "method, and its oracle only requires that the execution result contain one row of the solving outcome, thereby accelerating the solving process compared to the former approach, which requires matching all the rows. In 2022, Liang et al. introduced SQLRight [ 68], a metamorphic testing method combined with execution feedback, and it is scalable to allow new oracles. In 2023, Tang et al. presented a constraint-solving testing method, TQS [ 101], which incorporated subgraph isomorphism search into DBMS fuzzing, significantly enhancing bug detection efficiency. In 2024, Ba et al. and Liang et al.",
+      "context_after": "DBMS fuzzing techniques can be classified from different perspectives. —One classification approach is based on the way expected results are obtained, as shown in Figure1: differential testing, metamorphic testing, and constraint-solving. Differential testing ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:3 Applied metamorphic testing to DBMS (NOREC [91] and TLP [92]) The concept of test oracle (Howden et al. [47]) 1978 The concept of metamorphic testing (Chen et al. [24]) First applied fuzzing to",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7258,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "A Comprehensive Survey on Database Management System Fuzzing 268:3 Applied metamorphic testing to DBMS (NOREC [91] and TLP [92]) The concept of test oracle (Howden et al.",
+      "context_before": "aph isomorphism search into DBMS fuzzing, significantly enhancing bug detection efficiency. In 2024, Ba et al. and Liang et al. respectively proposed DQP [ 15] and Mozi [ 67], further extending metamorphic testing to different execution paths of the same query. DBMS fuzzing techniques can be classified from different perspectives. —One classification approach is based on the way expected results are obtained, as shown in Figure1: differential testing, metamorphic testing, and constraint-solving. Differential testing ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026.",
+      "context_after": "[47]) 1978 The concept of metamorphic testing (Chen et al. [24]) First applied fuzzing to DBMS (SQL Server [96]) 1998 Mutation-based generator GARan (Bati et al. [16]) 2007 2010 Constraint solving testing for DBMS (Khalek et al. [12]) Open-source rule-based statement generator (SQLSmith [4]) 2015 Execution feedback for crash detection (Squirrel [117]) 2020 Accelerated single-row DBMS constraint solving (PQS [93]) Sub-graph isomorphism search for constraint solving (TQS [102]) 2023 Differential Testing Metamorphic Testing Constraint-solving Testing Execution feedback for metamorphic testing (S",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7731,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "[12]) Open-source rule-based statement generator (SQLSmith [4]) 2015 Execution feedback for crash detection (Squirrel [117]) 2020 Accelerated single-row DBMS constraint solving (PQS [93]) Sub-graph isomorphism search for constraint solving (TQS [102]) 2023 Differential Testing Metamorphic Testing Constraint-solving Testing Execution feedback for metamorphic testing (SQLRight[69]) 2022 Others The concept of differential testing (McKeeman et al.",
+      "context_before": "re obtained, as shown in Figure1: differential testing, metamorphic testing, and constraint-solving. Differential testing ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:3 Applied metamorphic testing to DBMS (NOREC [91] and TLP [92]) The concept of test oracle (Howden et al. [47]) 1978 The concept of metamorphic testing (Chen et al. [24]) First applied fuzzing to DBMS (SQL Server [96]) 1998 Mutation-based generator GARan (Bati et al. [16]) 2007 2010 Constraint solving testing for DBMS (Khalek et al.",
+      "context_after": "[79]) 2024 Execution Path Manipulation for metamorphic testing (DQP[15] and Mozi[68]) Fig. 1. Main milestones of DBMS fuzzing. compares output differences across databases [ 77], mutation testing modifies input samples to assess the tolerance of the DBMS to exceptional inputs [ 24,26,62], and constraint-solving utilizes solvers for benchmark ground truths [ 21,58,79,94]. —FuzzingtechniquescanalsobecategorizedbythetransparencyoftheDBMSattesttime:blackbox, gray-box, and white-box [ 19]. Black-box testing assesses the behavior and functionality of DBMS solely through input and output observation",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 8131,
+      "found_by_all": [
+        "technique",
+        "citation_marker",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "[79]) 2024 Execution Path Manipulation for metamorphic testing (DQP[15] and Mozi[68]) Fig.",
+      "context_before": "ed fuzzing to DBMS (SQL Server [96]) 1998 Mutation-based generator GARan (Bati et al. [16]) 2007 2010 Constraint solving testing for DBMS (Khalek et al. [12]) Open-source rule-based statement generator (SQLSmith [4]) 2015 Execution feedback for crash detection (Squirrel [117]) 2020 Accelerated single-row DBMS constraint solving (PQS [93]) Sub-graph isomorphism search for constraint solving (TQS [102]) 2023 Differential Testing Metamorphic Testing Constraint-solving Testing Execution feedback for metamorphic testing (SQLRight[69]) 2022 Others The concept of differential testing (McKeeman et al.",
+      "context_after": "1. Main milestones of DBMS fuzzing. compares output differences across databases [ 77], mutation testing modifies input samples to assess the tolerance of the DBMS to exceptional inputs [ 24,26,62], and constraint-solving utilizes solvers for benchmark ground truths [ 21,58,79,94]. —FuzzingtechniquescanalsobecategorizedbythetransparencyoftheDBMSattesttime:blackbox, gray-box, and white-box [ 19]. Black-box testing assesses the behavior and functionality of DBMS solely through input and output observations [ 83]. Gray-box testing collects state information to generate targeted test cases [ 22,2",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 8579,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[ 35,37,40,43,92]",
+      "technique": "pqs",
+      "sentence": "Performance bugs manifest themselves as significant differences in the execution time of the same query in different versions of DBMS, or large performance gaps between a query and its equivalent [ 35,37,40,43,92].",
+      "context_before": "is worth noting that adding an ‘s’ to any notation indicates multiple instances of that item. For example, ‘Ss’ denotes multiple statements. 2.2 Basic Definitions Database bugs encompass three main types: crashes, logic bugs, and performance bugs [ 73,81]. Crashes occur when program errors or memory leaks interrupt the database while queries are executed. Compared to other types of bugs, crashes are relatively easier to identify. Logic bugs occur when the execution results deviate from the expected ones or violate common sense in the DBMS. Such bugs are usually hidden and difficult to detect.",
+      "context_after": "However, it is important to note that not all of these differences are considered to be actual bugs. Because determining the optimal execution plan is inherently an NP-hard problem [ 48], some databases practically choose suboptimal plans to speed up the optimization process. We consider performance ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:5 Fig. 2. Illustration of the differences between black-box fuzzing, grey-box fuzzing, and white-box fuzzing. Theblackarrowpartrepresentstheprocessofblack",
+      "section": "2.2 Basic Definitions",
+      "page": 4,
+      "char_offset": 14598,
+      "cited_reference": {
+        "number": 92,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedingsofthe USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 93]",
+      "technique": null,
+      "sentence": "For example, intramorphic testing [ 93] works by replacing a specific component in the system to get another version and then testing between the two versions to detect bugs in that component.",
+      "context_before": "DBMS, as shown in Figure 2. Black-box fuzzing treats the DBMS as a black box, where a series of SQL statements or transactions are inputted, and a series of execution results or execution plans are obtained accordingly. The bugs are then determined by validating the results and plans. Grey-box fuzzing collects state information from the DBMS, such as code branch coverage and status, to guide statement generation and improve the efficiency of DBMS testing. By comparison, white-box fuzzing requires testing with an understanding of the internal structure and implementation mechanisms of the DBMS.",
+      "context_after": "However, white-box fuzzing is currently limited to theoretical research on general systems, and there is no white-box fuzzing tool for DBMS, which is also the focus of future research. A testing oracle refers to a mechanism to obtain the expected results of test cases. DBMS fuzzing can also be divided into differential testing, metamorphic testing, and constraint-solving testing according to different oracles. Figure 3shows the difference between them. Their formal definitions are shown in Definition 2.1,2.2, and2.3. Definition2.1 (DifferentialTesting). Differential testing detects bugs by co",
+      "section": "2.2 Basic Definitions",
+      "page": 5,
+      "char_offset": 17058,
+      "cited_reference": {
+        "number": 93,
+        "text": "ManuelRiggerandZhendongSu.2022.Intramorphictesting:Anewapproachtothetestoracleproblem.In Proceedings oftheInternationalSymposiumonNewIdeas,NewParadigms,andReflectionsonProgrammingandSoftware. ACM, 128–136.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Comparison of Test Case Generators Fuzzer YearGenerator TypeGenerator StrategyFeedbackDatabase Instance RAGS[95]1998 Generation-based AST Model (Static Configuration) No Feedback (Black-Box)Existing DatabasesSQLsmith[ 4]2015 APOLLO[ 56]2019 AST Model (Dynamic Configuration) AMOEBA[ 72]2022 Go-Randgen[ 87]2019 AST Model (Static Configuration) Random DatabasesSQLancer[ 92]2020 Artemis[ 80]2021 DT2[27]2022 DQE[96]2023 Troc[31]2023 Fucci[36]2025 TxCheck[ 54]2023 EET[55]2024 PUPPY[ 106]2025 DDLCheck[ 98]2025 TQS[101]2023AST Model (Dynamic Configuration) ADUSA[ 12]2010 Alloy Model GARan[ 16]2007 Mutation-based SQL Structure MutationFeedback from internal status (Grey-Box)Existing Databases Squirrel[ 116]2020 Random DatabasesSquill[105]2023 SQLRight[ 68]2022 DynSQL[ 53]2023 Tuzz[64]2024 Eqsql[113]2021 No Feedback (Black-Box)MutaSQL[ 25]2020 Kangaroo[ 63]2023 SOFT[34]2025 LEGO[66]2023SQL Sequence MutationFeedback from internal status (Grey-Box) Griffin[33]2022No Feedback (Black-Box) QPG[13]2023DBMS State MutationFeedback from external interface (Black-Box) Generation-basedmethodsstartfromscratchtocreatetestcaseswithoutanyinitialinput.",
+      "context_before": "on execution information from the current test case [ 86]. The query reducer module is used to reduce the difficulty of triggering test cases to facilitate the localization and resolution of bugs by database developers [ 100]. 3.1 Test Case Generator The test case generator is the first step in DBMS fuzzing and is used to automatically generate test cases. There are primarily two types of implementation: generation-based and mutation-based. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:7 Table 2.",
+      "context_after": "Ittypically relies on a specific SQL syntax model [ 102] to randomly generate constants and expressions, aiming to enumerate potential test cases that satisfy specific criteria or cover specific aspects of the DBMS. On the other hand, mutation-based methods begin with existing seed queries and apply various modifications or mutations to generate new test cases [ 94,118]. In Table 2, we gain insight into the generation strategies of existing fuzzers and categorize them. Specifically, we compare them in terms of database instance, generation type, strategy, and feedback. 3.1.1 Database Instance.",
+      "section": "3.1 Test Case Generator",
+      "page": 7,
+      "char_offset": 21321,
+      "found_by_all": [
+        "technique",
+        "citation_marker",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "qpg",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[12,14,15,27,31,33,54,80,87,91,92,96,101,105,106,116]",
+      "technique": "cert",
+      "sentence": "Ontheotherhand,fuzzersbasedonrandomdatabases [12,14,15,27,31,33,54,80,87,91,92,96,101,105,106,116] create random database instances from scratch.",
+      "context_before": "tabases. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:8 X. Gao et al. Fig. 5. Working process of generation-based generators based on AST model. Existing databases come from benchmark tests such as TPC-H [ 1] or real datasets such as SCOTT [10]. Fuzzers based on existing databases [ 4,16,56,72,95] first acquire the schema of the target databaseandthengeneratesyntacticallycorrectstatementsbasedonthatschema.However,asboth the schema and the data are fixed, the generated statements cannot effectively explore the entire queryspace,leadingtoincompletedetection.",
+      "context_after": "The random generation of databases has been widely explored [ 20,42,45,85]. A common practice is to first create tables, indexes, and views randomly and then populate them with data using INSERT, UPDATE, and DELETE statements [ 90]. The advantage of using randomly generated databases is their potential to thoroughly explore the query space. When coupled with execution feedback, these solutions can effectively explore the search space. 3.1.2 Generator Type and Strategy. After acquiring the database instance, we can proceed to generate test cases using either a generation-based or a mutation-ba",
+      "section": "3.1 Test Case Generator",
+      "page": 8,
+      "char_offset": 23825,
+      "cited_reference": {
+        "number": 14,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. In Proceedings of the International Conference on Software Engineering. 1–13.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[ 90]",
+      "technique": "norec",
+      "sentence": "A common practice is to first create tables, indexes, and views randomly and then populate them with data using INSERT, UPDATE, and DELETE statements [ 90].",
+      "context_before": "tests such as TPC-H [ 1] or real datasets such as SCOTT [10]. Fuzzers based on existing databases [ 4,16,56,72,95] first acquire the schema of the target databaseandthengeneratesyntacticallycorrectstatementsbasedonthatschema.However,asboth the schema and the data are fixed, the generated statements cannot effectively explore the entire queryspace,leadingtoincompletedetection.Ontheotherhand,fuzzersbasedonrandomdatabases [12,14,15,27,31,33,54,80,87,91,92,96,101,105,106,116] create random database instances from scratch. The random generation of databases has been widely explored [ 20,42,45,85].",
+      "context_after": "The advantage of using randomly generated databases is their potential to thoroughly explore the query space. When coupled with execution feedback, these solutions can effectively explore the search space. 3.1.2 Generator Type and Strategy. After acquiring the database instance, we can proceed to generate test cases using either a generation-based or a mutation-based approach. Among them, the generation-based approach consists primarily of two strategies: Abstract Syntax Tree (AST) model and Alloy model [ 49]. The general process of the AST model is depicted in Figure 5. First, we need an AST",
+      "section": "3.1 Test Case Generator",
+      "page": 8,
+      "char_offset": 24046,
+      "cited_reference": {
+        "number": 90,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "citation_marker",
+      "surface": "[ 27,31,\n90–92,96]",
+      "technique": "norec",
+      "sentence": "Without this limitation, numerous complex SQL statements would be generated, which may expand the search space but reduce the overall efficiency of bug detection [ 27,31, 90–92,96].",
+      "context_before": "JoinedTables: This parameter helps constrain the complexity of join relationships, preventing the generation of large-scale multi-table join statements and enhancing the efficiency of statement generation [ 87,95]. —Limiting the Maximum Number of Rows Returned: This parameter restricts the execution time and resource consumption of the test cases by adding a LIMITclause to the original statement [ 87,95]. —Limitingthe Maximum Depth of the AST: This parameter is set to limit the generation time of test cases, as the generation time tends to increase exponentially with the increase in AST depth.",
+      "context_after": "However, static configurations require manual adjustments based on specific requirements. This process can be time-consuming, and manually setting parameters may not sufficiently explore the state space in a targeted manner. Dynamic configurations of the AST model adjust generator parameters based on previous execution results, which include: —Dynamic Probability Table: The dynamic probability table stores the dynamic probability ofvarious SQLclauses.IfanSQLquerytriggersabug,theprobabilityoftheclausesappearing in that SQL query increases. This ensures that the generator explores the input spac",
+      "section": "3.1 Test Case Generator",
+      "page": 9,
+      "char_offset": 27035,
+      "cited_reference": {
+        "number": 90,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[16,25,34,53,63,64,68,105,113,116]",
+      "technique": null,
+      "sentence": "Mutation-based generators can be divided into three major categories: SQL structure mutation [16,25,34,53,63,64,68,105,113,116], SQL sequence mutation [ 33,66], and DBMS state mutation [13].",
+      "context_before": ". Alloy Syntax Modelsigid, name, student ID, course ID, grade extends FieldName {} sigstudents, grades extends TableName {} factselect_field_in_table { allf: Select.fields.field | somet: From.tables | fint.fields } factwhere_field_in_table{ allf: Where.(...).field | somet: From.tables | fint.fields } factunique_select_terms { alla, b: SELECT.fields | ( a.field = b.field and a.agg = b.agg ) => a=b } Listing 2. Alloy Data Model names in select clause and where clause. Therefore, a generator based on the Alloy model [ 12] can achieve a higher level of semantic correctness compared to the AST model.",
+      "context_after": "Figure 6illustrates the differences between three mutation strategies. SQL structure mutation generates new test cases by modifying the structure or expression of statement, and it is the most common type of mutation. SQL sequence mutation generates new test cases by reshuffling or deleting queries from the original test cases. Its advantage is that it does not rely on SQL syntax, which reduces the adaptation cost for new DBMSs. DBMS state mutation uses DDL operations to create indexes or modify table definitions to generate new database schemas. In addition, it combines old SQL statements wi",
+      "section": "3.1 Test Case Generator",
+      "page": 10,
+      "char_offset": 31239,
+      "cited_reference": {
+        "number": 34,
+        "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and detecting sql function bugs: Using simple boundary arguments to trigger hundreds of DBMS bugs. In Proceedings of the European Conference on Computer Systems. 1061–1076.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[ 13]",
+      "technique": "qpg",
+      "sentence": "Some generators of black-box fuzzers [ 13] obtain feedback such as the validity of query plans or test cases from the query or query plan interface provided by the DBMS, guiding the subsequent generation process.",
+      "context_before": "lectStmt. —Insertion: Insertion involves creating a new IR node and inserting it into an existing IR node. For example, a ‘Table’ can be inserted into a ‘FromClause’. —Replacement: Replacement can be achieved by modifying the type of an IR node, such as replacing column ‘x’ with ‘count(x)’. —Deletion: Deletion means removing at least one child node from an IR node. 3.1.3 Execution Feedback. Most generators do not take advantage of feedback information and treat the target DBMS as a black box, which accepts SQL statements or transactions as input and output execution results or execution plans.",
+      "context_after": "This kind of feedback is essentially different from the feedback from gray-box fuzzers. The gray-box fuzzer generators [16,53,66,68,105,116] collect the internal status of the DBMS, such as collecting code coverage to guide the generation of test cases that maximize code coverage. In contrast, the feedback of the black-box fuzzers relies only upon the external interface of the DBMS without any knowledge of the source code. Black-box fuzzers are easier to implement and can be used to test any commercial DBMSs. However, it is essentially a random exploration of the entire input space. Gray-box",
+      "section": "3.1 Test Case Generator",
+      "page": 11,
+      "char_offset": 34344,
+      "cited_reference": {
+        "number": 13,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In Proceedings of the International Conference on Software Engineering. ACM, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[33,34,53,66,105,116]",
+      "technique": null,
+      "sentence": "Fuzzing methods based on crash oracle [33,34,53,66,105,116] can only detect crashes.",
+      "context_before": "ecution results on the target DBMS. However, for fuzzing, it is not a trivial task to obtain the expected results of random test cases.Oraclereferstothemeanstoobtaintheexpectedresults[ 46].Existingoraclescanbeclassified into four types: crash oracle, differential oracle, metamorphic oracle, and constraint-solving oracle. Table3presents the classification of fuzzers along with their respective oracle implementation methods and applicable domains. 3.2.1 Crash Oracle. The crash oracle, also called a no-oracle, determines the occurrence of a crash based on the running state of the target database.",
+      "context_after": "After executing test cases, bugs are determined based on whether the DBMS stops running. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:12 X. Gao et al. Table 3. Comparison of Oracles Fuzzer Oracle Type Feature Test Scope Squirrel[ 116] Crash Database Crashes Crash BugsSquill[105] Griffin[33] LEGO[66] DynSQL[ 53] SOFT[34] RAGS[95] DifferentialDifferent DBMSs Logic BugsSQLsmith[ 4] Go-Randgen[ 87] GARan[ 16] DT2[27] Radar[97]Different Database InstancesDDLCheck[ 98] APOLLO[ 56] Different Versions of the Same DBMS Performance Bugs AMOEBA[ 72] MetamorphicState",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 11,
+      "char_offset": 36168,
+      "cited_reference": {
+        "number": 34,
+        "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and detecting sql function bugs: Using simple boundary arguments to trigger hundreds of DBMS bugs. In Proceedings of the European Conference on Computer Systems. 1061–1076.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Comparison of Oracles Fuzzer Oracle Type Feature Test Scope Squirrel[ 116] Crash Database Crashes Crash BugsSquill[105] Griffin[33] LEGO[66] DynSQL[ 53] SOFT[34] RAGS[95] DifferentialDifferent DBMSs Logic BugsSQLsmith[ 4] Go-Randgen[ 87] GARan[ 16] DT2[27] Radar[97]Different Database InstancesDDLCheck[ 98] APOLLO[ 56] Different Versions of the Same DBMS Performance Bugs AMOEBA[ 72] MetamorphicStatement RewritingCERT[14] MutaSQL[ 25] Logic BugsEqsql[113] NoREC[ 90] SQLRight[ 68] PINOLO[ 44] EET[55] CODDTest[ 112] PUPPY[ 106] Execution Path ManipulationPerformance Bugs Mozi[67] Performance and Logic Bugs Kangaroo[ 63] Logic and Crash Bugs DQP[15] Logic BugsDQE[96] Statement Type Transformation TLP[91] Query Partition Troc[31]Transaction SplittingTxCheck[ 54] ADUSA[ 12] Constraint-solvingForward Solving(SAT Solver)Artemis[ 80] PQS[92] Backward Solving TQS[101] Forward Solving(Logical Solver) WriteCheck[ 28] Fucci[36] 3.",
+      "context_before": "oracle, and constraint-solving oracle. Table3presents the classification of fuzzers along with their respective oracle implementation methods and applicable domains. 3.2.1 Crash Oracle. The crash oracle, also called a no-oracle, determines the occurrence of a crash based on the running state of the target database. Fuzzing methods based on crash oracle [33,34,53,66,105,116] can only detect crashes. After executing test cases, bugs are determined based on whether the DBMS stops running. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:12 X. Gao et al. Table 3.",
+      "context_after": "2.2 Differential Oracle. Differential oracles detect logic bugs or performance bugs by comparing execution results between different DBMSs [ 4,16,27,87,95], different versions of the same DBMS [ 56], or the same DBMS operating under different constructions [ 97,98]. It tries to get expected results from another database instance or DBMS. The advantage of differential oracle ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:13 (a) Constraint Rewriting (b) Structure Rewriting (c) Expression Rewriting",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 12,
+      "char_offset": 36451,
+      "found_by_all": [
+        "technique",
+        "citation_marker_project_authored",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "coddtest",
+        "dqp",
+        "tlp",
+        "pqs",
+        "cert"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "citation_marker",
+      "surface": "[ 15,25,54,55,68,72,90,\n91,96,106,112,113]",
+      "technique": "dqp",
+      "sentence": "Most relations are equivalence-based [ 15,25,54,55,68,72,90, 91,96,106,112,113], while only a small subset are non-equivalent [ 14,44].",
+      "context_before": "g. 8. Example of statement rewritingstrategies. testing is that it is relatively simple. Nevertheless, it cannot detect common bugs between the target DBMS and the referenced one. 3.2.3 Metamorphic Oracle. To avoid the drawbacks of the differential oracle, many fuzzing methods use the metamorphic oracle to detect logic bugs or performance bugs. Metamorphic oracle applies specific transformations on the original statements to construct new statements [ 24]. The execution results of the original and transformed statements are then checked to see if they satisfy a predefined metamorphic relation.",
+      "context_after": "There are several ways to construct metamorphic oracles: —StatementRewriting: Statement Rewriting refers to the process of modifying a statement according to specified rules, so that the final semantics remain unchanged or change as expected. Common methods of rewriting statements include constraint rewriting, structure rewriting, and expression rewriting, as shown in Figure 8. Constraint rewriting refers to modifying the constraints of a table, such as creating an index. Obviously, the execution result of the same SELECT statement should remain unchanged after creating the index. Structure r",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 13,
+      "char_offset": 38584,
+      "cited_reference": {
+        "number": 15,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans. Proceedingsofthe ACM on Management of Data 2, 3 (2024), 1–26.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp",
+        "cert"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "citation_marker",
+      "surface": "[ 14,25,44,72,90,112,113]",
+      "technique": "cert",
+      "sentence": "Expression Rewriting refers to the process of rewriting logical or arithmetic expressions, such as adding predicates that always hold true to the WHERE clause or changing the string comparison operator from ‘=’ to ‘LIKE’, which will affect the result of the expression as expected [ 14,25,44,72,90,112,113].",
+      "context_before": "fers to the process of modifying a statement according to specified rules, so that the final semantics remain unchanged or change as expected. Common methods of rewriting statements include constraint rewriting, structure rewriting, and expression rewriting, as shown in Figure 8. Constraint rewriting refers to modifying the constraints of a table, such as creating an index. Obviously, the execution result of the same SELECT statement should remain unchanged after creating the index. Structure rewriting refers to adding new clauses such as JOIN,DISTINCT, and LIMITbased on the original statement.",
+      "context_after": "—Execution Path Manipulation: Execution path manipulation is a technique that identifies bugs by manipulating the execution plan of a query and observing whether these manipulations produce divergent results or performance anomalies. For instance, query hints or system variables are employed to induce the same query to be executed using different plans, detecting performance or logic bugs [ 15,67]. Similarly, a sequence of system variable operations can be executed prior to the query to trigger the optimizer to select varied execution plans, which is used to identify performance bugs [ 106].",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 13,
+      "char_offset": 39421,
+      "cited_reference": {
+        "number": 14,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Cert: Finding performance issues in database systems through the lens of cardinality estimation. In Proceedings of the International Conference on Software Engineering. 1–13.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "citation_marker",
+      "surface": "[ 15,67]",
+      "technique": "dqp",
+      "sentence": "For instance, query hints or system variables are employed to induce the same query to be executed using different plans, detecting performance or logic bugs [ 15,67].",
+      "context_before": "as JOIN,DISTINCT, and LIMITbased on the original statement. Expression Rewriting refers to the process of rewriting logical or arithmetic expressions, such as adding predicates that always hold true to the WHERE clause or changing the string comparison operator from ‘=’ to ‘LIKE’, which will affect the result of the expression as expected [ 14,25,44,72,90,112,113]. —Execution Path Manipulation: Execution path manipulation is a technique that identifies bugs by manipulating the execution plan of a query and observing whether these manipulations produce divergent results or performance anomalies.",
+      "context_after": "Similarly, a sequence of system variable operations can be executed prior to the query to trigger the optimizer to select varied execution plans, which is used to identify performance bugs [ 106]. In addition, some research [63] modifies the DBMS source code to force the same query to execute under multiple plans. —StatementTypeTransformation: Statement type transformation involves constructing UPDATE and DELETE statements that correspond to a given SELECT statement. For example, a row updated by an UPDATE statement with a predicate 𝑝should also be fetched by a SELECT statement with the same p",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 13,
+      "char_offset": 39963,
+      "cited_reference": {
+        "number": 15,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep it simple: Testing databases via differential query plans. Proceedingsofthe ACM on Management of Data 2, 3 (2024), 1–26.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "citation_marker",
+      "surface": "[ 91]",
+      "technique": "tlp",
+      "sentence": "—Query Partition: Query partition refers to dividing an original SQL query into multiple partitions, and the merged results of the partitioned queries should be consistent with the results of the original query [ 91].",
+      "context_before": "to a given SELECT statement. For example, a row updated by an UPDATE statement with a predicate 𝑝should also be fetched by a SELECT statement with the same predicate 𝑝[96]. By adding tagging columns ‘rowid’ and ‘updated’ to the table, the SELECT statements can return ‘rowid’ of the affected rows, as shown in Figure 9. ForDELETE statements, the affected rows can be determined by calculating the difference between the set of ‘rowid’ before and after deletion. Similarly, for UPDATE statements, the affected rows can be obtained by querying the ‘rowid’ where the ‘updated’ column is marked as true.",
+      "context_after": "For example, a query ‘ SELECT * FROM t1 ’ can be ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:14 X. Gao et al. Fig. 9. Example of transformation of statement type. partitioned into three queries: ‘ SELECT * FROM t1 WHERE c1<10 ’, ‘SELECT * FROM t1 WHERE c1>=10 ’, and ‘SELECT * FROM t1 WHERE c1 IS NULL ’; —Transaction Splitting: Transaction splitting refers to splitting transactions into statements and then, by building a multiple version chain of data outside the database [ 31] or reorganizing the statement execution order [ 54], allowing statements withi",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 13,
+      "char_offset": 41173,
+      "cited_reference": {
+        "number": 91,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Programming Languages 4, OOPSLA (2020), 1–30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 54]",
+      "technique": null,
+      "sentence": "partitioned into three queries: ‘ SELECT * FROM t1 WHERE c1<10 ’, ‘SELECT * FROM t1 WHERE c1>=10 ’, and ‘SELECT * FROM t1 WHERE c1 IS NULL ’; —Transaction Splitting: Transaction splitting refers to splitting transactions into statements and then, by building a multiple version chain of data outside the database [ 31] or reorganizing the statement execution order [ 54], allowing statements within the transaction to run in non-transaction mode.",
+      "context_before": "set of ‘rowid’ before and after deletion. Similarly, for UPDATE statements, the affected rows can be obtained by querying the ‘rowid’ where the ‘updated’ column is marked as true. —Query Partition: Query partition refers to dividing an original SQL query into multiple partitions, and the merged results of the partitioned queries should be consistent with the results of the original query [ 91]. For example, a query ‘ SELECT * FROM t1 ’ can be ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:14 X. Gao et al. Fig. 9. Example of transformation of statement type.",
+      "context_after": "By comparing the execution results before and after transaction splitting, it is possible to determine whether there exist any logic bugs in the transaction management module of the DBMS. 3.2.4 Constraint-solvingOracle. Due to the fact that not all test queries can obtain test oracles by constructing equivalent queries, some fuzzers use constraint-solving to obtain the expected result. Constraint-solving fuzzing methods [ 12,80,92,101] generally rely on forward or backward solving to derive the ground truth of the execution results. —ForwardSolving: Forward solving employs an external solver",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 14,
+      "char_offset": 41593,
+      "cited_reference": {
+        "number": 54,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via{graph-based }oracle construction. In Proceedings of the USENIX Symposium on Operating Systems Design and Implementation. 397–417.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "citation_marker",
+      "surface": "[ 12,80,92,101]",
+      "technique": "pqs",
+      "sentence": "Constraint-solving fuzzing methods [ 12,80,92,101] generally rely on forward or backward solving to derive the ground truth of the execution results.",
+      "context_before": "s and then, by building a multiple version chain of data outside the database [ 31] or reorganizing the statement execution order [ 54], allowing statements within the transaction to run in non-transaction mode. By comparing the execution results before and after transaction splitting, it is possible to determine whether there exist any logic bugs in the transaction management module of the DBMS. 3.2.4 Constraint-solvingOracle. Due to the fact that not all test queries can obtain test oracles by constructing equivalent queries, some fuzzers use constraint-solving to obtain the expected result.",
+      "context_after": "—ForwardSolving: Forward solving employs an external solver to evaluate each tuple against predicate constraints in order to obtain the ground truth for the results of the statement execution. While traditional approaches utilize generic SAT solvers for this task solver [39,59], recent methods, such as TQS [ 101], also use a logical external solver as a substitute for the SAT solver to derive the ground truth results based on the original test cases through logical operations or theoretical deduction [ 28,36,101]. —Backward Solving: Backward solving entails initially selecting some tuples as",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 14,
+      "char_offset": 42429,
+      "cited_reference": {
+        "number": 92,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedingsofthe USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M24",
+      "found_by": "citation_marker",
+      "surface": "[ 92]",
+      "technique": "pqs",
+      "sentence": "—Backward Solving: Backward solving entails initially selecting some tuples as ground truth at random and then using an SAT solver to work backward and obtain a statement whose execution results include these tuples [ 92].",
+      "context_before": "forward or backward solving to derive the ground truth of the execution results. —ForwardSolving: Forward solving employs an external solver to evaluate each tuple against predicate constraints in order to obtain the ground truth for the results of the statement execution. While traditional approaches utilize generic SAT solvers for this task solver [39,59], recent methods, such as TQS [ 101], also use a logical external solver as a substitute for the SAT solver to derive the ground truth results based on the original test cases through logical operations or theoretical deduction [ 28,36,101].",
+      "context_after": "In all, each oracle has its own applicable scenarios. Fuzzing based on crash oracles can only detect crash-related issues such as program errors and memory leaks, which limits their application scenarios.Fuzzingbasedondifferentialoraclescanonlydetectcommonfunctionalmodulesbetween differentdatabasesandcannotidentifylogicorperformancebugssharedbythereferenceddatabase and the test one. Fuzzing based on metamorphic oracles does not require a reference database, but suffers from a limited test range. Although constrained solvers are not subject to testing scope limitations, they exhibit lower solv",
+      "section": "3.2 Oracle-based Comparator",
+      "page": 14,
+      "char_offset": 43099,
+      "cited_reference": {
+        "number": 92,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedingsofthe USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Comparison of Fuzzer Execution Feedbacks Fuzzer Metamorphic Validation Coverage Query Plan Syntax and Semantics Error APOLLO[ 56] ✓ Squirrel[ 116] ✓ LEGO[66] ✓ SQLRight[ 68] ✓ QPG[13] ✓ AMOEBA[ 72]✓ ✓ GARan[ 16] ✓ DynSQL[ 53] ✓ ✓ Squill[105] ✓ ✓ ✓ include seven types: metamorphic feedback, validation feedback, coverage feedback, query plan feedback, syntax error feedback, semantics error feedback, and fitness feedback.",
+      "context_before": "ugs in join optimizations. 3.3 Execution Feedback Execution feedback refers to the process of guiding test case generation based on execution results or database status. Execution feedback is not limited to the dynamic configuration of generators, but can also adjust mutation strategies and search directions, typically improving the efficiency of fuzz testing. Table 4summarizes the feedback mechanisms used by existing fuzzers, which usually ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:15 Table 4.",
+      "context_after": "3.3.1 Metamorphic Feedback. Metamorphic feedback [ 72] refers to the use of feedback information to generate more statements that satisfy the conditions for the use of metamorphic oracles. This feedback is often combined with generation-based generators, where the generation process is influenced by the success of equivalent transformations. Specifically, in fuzzers using metamorphic feedback, statements are generated based on a probability table. If a SQL query can employ some of the statement rewriting strategies shown in Figure 8, the subsequent occurrence probabilities of LIMIT, GROUP BY,",
+      "section": "3.3 Execution Feedback",
+      "page": 15,
+      "char_offset": 44815,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "citation_marker",
+      "surface": "[ 13]",
+      "technique": "qpg",
+      "sentence": "Query plan feedback [ 13] uses the emergence of new query plans to guide subsequent query generation.",
+      "context_before": "ation before and after executing each test case. If a mutation query leads to an increase in code coverage, it will be added to the seed queue. Additionally, LEGO [ 66] analyzes the type sequences of statements that increase the coverage of the code, and then the mutation processes will generate test cases containing the target type sequences. 3.3.4 QueryPlanFeedback. The goal of query plan feedback is for the generator to maximize the number of query plans. A query plan represents the execution path of an SQL query in a DBMS, with different query plans corresponding to distinct code branches.",
+      "context_after": "Specifically, it models the objective of maximizing query plan quantity as a Multi-Armed Bandit (MAB) problem [ 18]. The generation of new query plans corresponds to exploitation, while exploring new mutation operations corresponds to exploration. When considering both exploitation and exploration, more unique query plans can be generated, increasing the possibility of discovering logic bugs. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:16 X. Gao et al. Table 5. Comparison of Fuzzer Query Reducers FuzzerReduce ExpressionDelete ClauseDelete SubqueryDelete",
+      "section": "3.3 Execution Feedback",
+      "page": 15,
+      "char_offset": 47542,
+      "cited_reference": {
+        "number": 13,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing database engines via query plan guidance. In Proceedings of the International Conference on Software Engineering. ACM, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Comparison of Fuzzer Query Reducers FuzzerReduce ExpressionDelete ClauseDelete SubqueryDelete IR NodeSemantics Preservation RAGS[95] ✓ ✓ SQLancer[ 92]✓ ✓ MutaSQL[ 25]✓ ✓ GARan[ 16] ✓ ✓ SQLRight[ 68] ✓ ✓ APOLLO[ 56]✓ ✓ ✓ ✓ DynSQL[ 53]✓ ✓ ✓ ✓ 3.",
+      "context_before": "plan feedback [ 13] uses the emergence of new query plans to guide subsequent query generation. Specifically, it models the objective of maximizing query plan quantity as a Multi-Armed Bandit (MAB) problem [ 18]. The generation of new query plans corresponds to exploitation, while exploring new mutation operations corresponds to exploration. When considering both exploitation and exploration, more unique query plans can be generated, increasing the possibility of discovering logic bugs. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:16 X. Gao et al. Table 5.",
+      "context_after": "3.5 SyntaxErrorandSemanticsErrorFeedback. Some generators [ 105] consider statements with semantic and syntactic errors to be valuable because they may activate certain unexplored SQL features. Therefore, these statements are fed back into the seed queue, but given lower priority, rather than being discarded entirely. Some other generators [ 53] advocate that although syntax errors and semantic errors can improve code coverage, they do not help detect logic bugs. This is because, even though statements with syntax and semantic errors can trigger boundary branches in the code, they will fail du",
+      "section": "3.3 Execution Feedback",
+      "page": 16,
+      "char_offset": 48149,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "citation_marker",
+      "surface": "[ 16,25,92,95]",
+      "technique": "pqs",
+      "sentence": "Reducing expressions [ 16,25,92,95] and deleting clauses refer to the process of reducing queries by simplifying WHERE clauses, arithmetic expressions, or logical expressions.",
+      "context_before": "n of test cases that are more suitable for oracles. Syntax and semantic error feedback play a minor role because test cases with syntax errors do not help detect the underlying bugs. 3.4 Query Reducer Query reducers are designed to simplify and summarize complex queries, providing developers with minimal human-readable queries for debugging. Query reduction strategies can be classified into four types: reduce expression, delete clause, delete subquery, and delete IR node. Table 5shows the query reduction strategies used by each fuzzer and also indicates whether the reducer preserves semantics.",
+      "context_after": "They are relatively easy to implement, but cannot guarantee semantic correctness after simplification, resulting in a decrease in reduction efficiency. The removal of subquery operations [ 53,56] extends the scope of simplification to subqueries, leading to a substantial improvement in the efficiency of simplification. Delete IR node operations are used in mutation-based fuzzers based on IR [ 68] mutation. They treat query simplification as a mutation operation on SQL queries, ensuring semantic correctness by deleting specific IR nodes. When considering the dependency between deleted expressi",
+      "section": "3.4 Query Reducer",
+      "page": 16,
+      "char_offset": 50382,
+      "cited_reference": {
+        "number": 92,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In Proceedingsofthe USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [92] begins by randomly generating a pivot row as the ground truth and then constructs an SQL query that includes this pivot row in its result.",
+      "context_before": ". For performance bugs, conducting component fuzzing alone on a single component is often insufficient for detection, as performance disparities typically result from the cumulative impact of multiple components. For crashes, fuzzers basically run against the entire DBMS. While it is possible to perform performance bug and crash detection targeting specific components, such testing is not only time-consuming but also provides only localized results, failing to capture the overall state of the entire DBMS. Therefore, there is little research effort in this area. 4.1.1 OverallFuzzingonLogicBugs.",
+      "context_after": "By checking whether the result of the executed query in the test database contains the pivot row, logic bugs can be detected. Figure 11illustrates the entire PQS pipeline. The main challenge of PQS is how to generate an SQL query whose execution result contains a known pivot row. The solution to PQS is to first create predicates randomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False in",
+      "section": "4.1 Overall Fuzzing",
+      "page": 17,
+      "char_offset": 52920,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Figure 11illustrates the entire PQS pipeline.",
+      "context_before": "hile it is possible to perform performance bug and crash detection targeting specific components, such testing is not only time-consuming but also provides only localized results, failing to capture the overall state of the entire DBMS. Therefore, there is little research effort in this area. 4.1.1 OverallFuzzingonLogicBugs. PQS [92] begins by randomly generating a pivot row as the ground truth and then constructs an SQL query that includes this pivot row in its result. By checking whether the result of the executed query in the test database contains the pivot row, logic bugs can be detected.",
+      "context_after": "The main challenge of PQS is how to generate an SQL query whose execution result contains a known pivot row. The solution to PQS is to first create predicates randomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOTbefore the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding IS",
+      "section": "4.1 Overall Fuzzing",
+      "page": 17,
+      "char_offset": 53194,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "The main challenge of PQS is how to generate an SQL query whose execution result contains a known pivot row.",
+      "context_before": "and crash detection targeting specific components, such testing is not only time-consuming but also provides only localized results, failing to capture the overall state of the entire DBMS. Therefore, there is little research effort in this area. 4.1.1 OverallFuzzingonLogicBugs. PQS [92] begins by randomly generating a pivot row as the ground truth and then constructs an SQL query that includes this pivot row in its result. By checking whether the result of the executed query in the test database contains the pivot row, logic bugs can be detected. Figure 11illustrates the entire PQS pipeline.",
+      "context_after": "The solution to PQS is to first create predicates randomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOTbefore the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding ISNULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can p",
+      "section": "4.1 Overall Fuzzing",
+      "page": 17,
+      "char_offset": 53240,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "The solution to PQS is to first create predicates randomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions.",
+      "context_before": "only localized results, failing to capture the overall state of the entire DBMS. Therefore, there is little research effort in this area. 4.1.1 OverallFuzzingonLogicBugs. PQS [92] begins by randomly generating a pivot row as the ground truth and then constructs an SQL query that includes this pivot row in its result. By checking whether the result of the executed query in the test database contains the pivot row, logic bugs can be detected. Figure 11illustrates the entire PQS pipeline. The main challenge of PQS is how to generate an SQL query whose execution result contains a known pivot row.",
+      "context_after": "The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOTbefore the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding ISNULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried. The core idea of TLP [ 91] is that the result of the predicate evaluation always falls within the values of T",
+      "section": "4.1 Overall Fuzzing",
+      "page": 17,
+      "char_offset": 53349,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M33",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried.",
+      "context_before": "to generate an SQL query whose execution result contains a known pivot row. The solution to PQS is to first create predicates randomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOTbefore the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding ISNULL after the predicate can help.",
+      "context_after": "The core idea of TLP [ 91] is that the result of the predicate evaluation always falls within the values of True, False, and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝 𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:18 X. Gao et al. Fig. 11. Pipeline of PQS. Fig. 12. Pipeline of TLP. Fig. 13. Pipeline of QPG. to the union of the results of the three partitioned queries. Figure 12illustrates the main process of TLP. Using a generator si",
+      "section": "4.1 Overall Fuzzing",
+      "page": 17,
+      "char_offset": 53874,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "The core idea of TLP [ 91] is that the result of the predicate evaluation always falls within the values of True, False, and NULL.",
+      "context_before": "andomly and then use the AST interpreter to evaluate whether the pivot row satisfies the predicate conditions. The evaluation result can be True, False, or NULL. True indicates that the pivot row can be queried through the predicate, requiring no adjustment. False indicates that the pivot row cannot be queried, so adding NOTbefore the predicate rectifies it. NULL indicates that the result of the predicate query is unknown, so adding ISNULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried.",
+      "context_after": "Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝 𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:18 X. Gao et al. Fig. 11. Pipeline of PQS. Fig. 12. Pipeline of TLP. Fig. 13. Pipeline of QPG. to the union of the results of the three partitioned queries. Figure 12illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for resul",
+      "section": "4.1 Overall Fuzzing",
+      "page": 17,
+      "char_offset": 54001,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M35",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Pipeline of PQS.",
+      "context_before": "dicate query is unknown, so adding ISNULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried. The core idea of TLP [ 91] is that the result of the predicate evaluation always falls within the values of True, False, and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝 𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:18 X. Gao et al. Fig. 11.",
+      "context_after": "Fig. 12. Pipeline of TLP. Fig. 13. Pipeline of QPG. to the union of the results of the three partitioned queries. Figure 12illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [13] mutates the state of the database to generate more unique query plans, as shown in Figure13. QPG implements a generation-based generator based on PQS",
+      "section": "4.1 Overall Fuzzing",
+      "page": 18,
+      "char_offset": 54403,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M36",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Pipeline of TLP.",
+      "context_before": "o adding ISNULL after the predicate can help. By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried. The core idea of TLP [ 91] is that the result of the predicate evaluation always falls within the values of True, False, and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝 𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:18 X. Gao et al. Fig. 11. Pipeline of PQS. Fig. 12.",
+      "context_after": "Fig. 13. Pipeline of QPG. to the union of the results of the three partitioned queries. Figure 12illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [13] mutates the state of the database to generate more unique query plans, as shown in Figure13. QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also",
+      "section": "4.1 Overall Fuzzing",
+      "page": 18,
+      "char_offset": 54429,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M37",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Pipeline of QPG.",
+      "context_before": "predicate can help. By employing the above method, PQS ensures that any random predicate can produce a satisfactory SQL query after being queried. The core idea of TLP [ 91] is that the result of the predicate evaluation always falls within the values of True, False, and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝 𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:18 X. Gao et al. Fig. 11. Pipeline of PQS. Fig. 12. Pipeline of TLP. Fig. 13.",
+      "context_after": "to the union of the results of the three partitioned queries. Figure 12illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [13] mutates the state of the database to generate more unique query plans, as shown in Figure13. QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also utilizes TLP and NoREC to",
+      "section": "4.1 Overall Fuzzing",
+      "page": 18,
+      "char_offset": 54455,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M38",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Figure 12illustrates the main process of TLP.",
+      "context_before": "predicate can produce a satisfactory SQL query after being queried. The core idea of TLP [ 91] is that the result of the predicate evaluation always falls within the values of True, False, and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝 𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:18 X. Gao et al. Fig. 11. Pipeline of PQS. Fig. 12. Pipeline of TLP. Fig. 13. Pipeline of QPG. to the union of the results of the three partitioned queries.",
+      "context_after": "Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [13] mutates the state of the database to generate more unique query plans, as shown in Figure13. QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and ob",
+      "section": "4.1 Overall Fuzzing",
+      "page": 18,
+      "char_offset": 54534,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M39",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs.",
+      "context_before": "after being queried. The core idea of TLP [ 91] is that the result of the predicate evaluation always falls within the values of True, False, and NULL. Therefore, an original query 𝑄can be decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝 𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:18 X. Gao et al. Fig. 11. Pipeline of PQS. Fig. 12. Pipeline of TLP. Fig. 13. Pipeline of QPG. to the union of the results of the three partitioned queries. Figure 12illustrates the main process of TLP.",
+      "context_after": "Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [13] mutates the state of the database to generate more unique query plans, as shown in Figure13. QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and obtain different query plans for the same query. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necess",
+      "section": "4.1 Overall Fuzzing",
+      "page": 18,
+      "char_offset": 54580,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M40",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Neither PQS nor TLP adopts feedback but performs a random search throughout the state space.",
+      "context_before": "e decomposed into three partitioned queries: 𝑄′ 𝑝,𝑄′ ¬𝑝,𝑄′ 𝑝 𝐼𝑆𝑁𝑈𝐿𝐿. The expected result of the original query should be equal ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:18 X. Gao et al. Fig. 11. Pipeline of PQS. Fig. 12. Pipeline of TLP. Fig. 13. Pipeline of QPG. to the union of the results of the three partitioned queries. Figure 12illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs.",
+      "context_after": "QPG [13] mutates the state of the database to generate more unique query plans, as shown in Figure13. QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and obtain different query plans for the same query. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as an MAB [ 18] problem. A",
+      "section": "4.1 Overall Fuzzing",
+      "page": 18,
+      "char_offset": 54768,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M41",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG [13] mutates the state of the database to generate more unique query plans, as shown in Figure13.",
+      "context_before": "he original query should be equal ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:18 X. Gao et al. Fig. 11. Pipeline of PQS. Fig. 12. Pipeline of TLP. Fig. 13. Pipeline of QPG. to the union of the results of the three partitioned queries. Figure 12illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space.",
+      "context_after": "QPG implements a generation-based generator based on PQS, TLP, and NoREC. QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and obtain different query plans for the same query. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as an MAB [ 18] problem. As shown in Equation ( 1),𝜇𝑖(𝑡)represents the benefit of the 𝑖th mutation operator at time 𝑡, there is",
+      "section": "4.1 Overall Fuzzing",
+      "page": 18,
+      "char_offset": 54861,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M42",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG implements a generation-based generator based on PQS, TLP, and NoREC.",
+      "context_before": "pril 2026. 268:18 X. Gao et al. Fig. 11. Pipeline of PQS. Fig. 12. Pipeline of TLP. Fig. 13. Pipeline of QPG. to the union of the results of the three partitioned queries. Figure 12illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [13] mutates the state of the database to generate more unique query plans, as shown in Figure13.",
+      "context_after": "QPG also utilizes TLP and NoREC to validate query execution results. DDL and DML statements are used to alter the state of the database and obtain different query plans for the same query. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as an MAB [ 18] problem. As shown in Equation ( 1),𝜇𝑖(𝑡)represents the benefit of the 𝑖th mutation operator at time 𝑡, there is a probability of 𝑝for random selection, and a probability of 1−𝑝to opt for",
+      "section": "4.1 Overall Fuzzing",
+      "page": 18,
+      "char_offset": 54963,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg",
+        "pqs",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M43",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "QPG also utilizes TLP and NoREC to validate query execution results.",
+      "context_before": "e of TLP. Fig. 13. Pipeline of QPG. to the union of the results of the three partitioned queries. Figure 12illustrates the main process of TLP. Using a generator similar to PQS, the original query is split into three equivalent partitioned queries, and metamorphic oracles are employed for result verification to detect logic bugs. Neither PQS nor TLP adopts feedback but performs a random search throughout the state space. QPG [13] mutates the state of the database to generate more unique query plans, as shown in Figure13. QPG implements a generation-based generator based on PQS, TLP, and NoREC.",
+      "context_after": "DDL and DML statements are used to alter the state of the database and obtain different query plans for the same query. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as an MAB [ 18] problem. As shown in Equation ( 1),𝜇𝑖(𝑡)represents the benefit of the 𝑖th mutation operator at time 𝑡, there is a probability of 𝑝for random selection, and a probability of 1−𝑝to opt for the mutation operator with the highest benefit. Since different quer",
+      "section": "4.1 Overall Fuzzing",
+      "page": 18,
+      "char_offset": 55037,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg",
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M44",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Since different query plans represent different execution paths, QPG can enhance the code coverage of DBMS.",
+      "context_before": "ry execution results. DDL and DML statements are used to alter the state of the database and obtain different query plans for the same query. Because different mutation operations contribute differently to generating new query plans, a decision among all mutation operators is necessary. The decision-making process for mutation operators is modeled as an MAB [ 18] problem. As shown in Equation ( 1),𝜇𝑖(𝑡)represents the benefit of the 𝑖th mutation operator at time 𝑡, there is a probability of 𝑝for random selection, and a probability of 1−𝑝to opt for the mutation operator with the highest benefit.",
+      "context_after": "𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟(𝑡) = {𝑎𝑟𝑔𝑚𝑎𝑥 𝑖=1...𝑘(𝜇𝑖(𝑡)) (1−𝑝) 𝑟𝑎𝑛𝑑𝑜𝑚(𝑘) (𝑝). (1) ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:19 Fig. 14. Pipeline of Radar. Fig. 15. Pipeline of AMOEBA. Radar [97] constructs a database instance 𝐷1that contains the same data as a generated database 𝐷0in the same DBMS, and it checks for logic bugs by comparing the differences in SQL execution results between them. As shown in Figure 14, it first generates a database instance 𝐷0in the target DBMS by randomly constructing and executin",
+      "section": "4.1 Overall Fuzzing",
+      "page": 18,
+      "char_offset": 55685,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M45",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Pipeline of NoREC.",
+      "context_before": "strates superior detection efficiency and covers a broader exploration space for identifying isolation anomalies. 4.3 Optimizer Testing The optimizer is one of the most complex components of a DBMS. It analyzes different execution plans of a query and selects the best query plan by considering indexes, association conditions, data statistics, and other factors to minimize query time and resource consumption. However, implementation errors in the optimizer can lead to serious logic bugs. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:22 X. Gao et al. Fig. 20.",
+      "context_after": "Fig. 21. Pipeline of TQS. TheNoRECpipeline[ 90]isillustratedinFigure 20.NoRECusesanASTmodel-basedgeneratorto generatequerieswith WHERE clausesandobtainstheunoptimizedquerybymovingtheconditions from the WHERE clause to the SELECT clause. The presence of optimizer logic bugs is determined by comparing the number of rows returned by the original query and the number of ‘True’ returned by the unoptimized query. The pipeline of TQS [ 101] is illustrated in Figure 21. Firstly, TQS splits a wide table into multiple small ones. On the one hand, it constructs a schema graph based on the foreign key re",
+      "section": "4.3 Optimizer Testing",
+      "page": 22,
+      "char_offset": 63301,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M46",
+      "found_by": "citation_marker",
+      "surface": "[ 90]",
+      "technique": "norec",
+      "sentence": "TheNoRECpipeline[ 90]isillustratedinFigure 20.",
+      "context_before": "ers a broader exploration space for identifying isolation anomalies. 4.3 Optimizer Testing The optimizer is one of the most complex components of a DBMS. It analyzes different execution plans of a query and selects the best query plan by considering indexes, association conditions, data statistics, and other factors to minimize query time and resource consumption. However, implementation errors in the optimizer can lead to serious logic bugs. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:22 X. Gao et al. Fig. 20. Pipeline of NoREC. Fig. 21. Pipeline of TQS.",
+      "context_after": "NoRECusesanASTmodel-basedgeneratorto generatequerieswith WHERE clausesandobtainstheunoptimizedquerybymovingtheconditions from the WHERE clause to the SELECT clause. The presence of optimizer logic bugs is determined by comparing the number of rows returned by the original query and the number of ‘True’ returned by the unoptimized query. The pipeline of TQS [ 101] is illustrated in Figure 21. Firstly, TQS splits a wide table into multiple small ones. On the one hand, it constructs a schema graph based on the foreign key relationship. By performing random walks on the schema graph, table sequenc",
+      "section": "4.3 Optimizer Testing",
+      "page": 22,
+      "char_offset": 63346,
+      "cited_reference": {
+        "number": 90,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. ACM.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M47",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "Pipeline of DQP.",
+      "context_before": "ema graph are updated. This reduces the likelihood of generating duplicate queries and enhances the efficiency of detecting logic bugs. 4.4 Executor Testing The executor is responsible for translating a query plan into a combination of physical operators and executing them to obtain the query result. These physical operators are specific instructions for performing query operations, such as Index Scan, Hash Join, Sort, and so on. During the ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:23 Fig. 22.",
+      "context_after": "development process of DBMS, any modification of the operator must ensure its correctness in various combinations. Therefore, fuzzing needs to be performed under different query plans that contain various combinations of physical operators. GARan [ 16] utilizes a mutation-based generator with genetic algorithms to focus on checking executor logic bugs. By setting a fitness function that evaluates the quality of mutated SQL statements, GARan can generate statements that produce specific physical operators, such as Hash Join. Then, based on the idea of differential testing, it compares the exec",
+      "section": "4.4 Executor Testing",
+      "page": 23,
+      "char_offset": 65425,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M48",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "DQP [15] detects logic bugs by enforcing the DBMS to utilize distinct execution plans, 𝑃0and𝑃1, for the same query 𝑄and identifying inconsistencies in their execution results.",
+      "context_before": "perator must ensure its correctness in various combinations. Therefore, fuzzing needs to be performed under different query plans that contain various combinations of physical operators. GARan [ 16] utilizes a mutation-based generator with genetic algorithms to focus on checking executor logic bugs. By setting a fitness function that evaluates the quality of mutated SQL statements, GARan can generate statements that produce specific physical operators, such as Hash Join. Then, based on the idea of differential testing, it compares the execution results of the same statement on different DBMSs.",
+      "context_after": "In Figure 22,𝐷(𝑃0) and𝐷(𝑃1)denote the execution outcomes of query 𝑄on DBMS 𝐷utilizing the corresponding plans. The approach is generator-agnostic and influences plan selection via two primary mechanisms. The first mechanism involves injecting optimization-related hints directly into the generated query. These hints instruct the optimizer to alter specific components or strategies within the execution plan. The second mechanism alters system variables to modify optimization options, effectively enabling or disabling specific methods that the optimizer can use during plan generation. Finally, D",
+      "section": "4.4 Executor Testing",
+      "page": 23,
+      "char_offset": 66097,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M49",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "Finally, DQP compares execution results under these plan manipulations with those from the default execution to determine whether a logic bug exists.",
+      "context_before": "on results. In Figure 22,𝐷(𝑃0) and𝐷(𝑃1)denote the execution outcomes of query 𝑄on DBMS 𝐷utilizing the corresponding plans. The approach is generator-agnostic and influences plan selection via two primary mechanisms. The first mechanism involves injecting optimization-related hints directly into the generated query. These hints instruct the optimizer to alter specific components or strategies within the execution plan. The second mechanism alters system variables to modify optimization options, effectively enabling or disabling specific methods that the optimizer can use during plan generation.",
+      "context_after": "5 Non-relational DBMS Fuzzing and New Techniques BeyondfuzzingforrelationalDBMSs,recentworkhasexpandedtonon-relational DBMSs,including graph, vector, spatial, distributed, and GPU databases. In addition, some studies leverage Large Language Models (LLMs) to enhance fuzzing. This section summarizes these emerging efforts. 5.1 Graph Database Management System (GDBMS) Fuzzing GDBMSs are non-relational systems for managing and querying highly interconnected data, typically represented using two paradigms: the Property Graph model and the Resource Description Framework (RDF) model. The widely adopt",
+      "section": "4.4 Executor Testing",
+      "page": 23,
+      "char_offset": 66862,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M50",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 57]",
+      "technique": null,
+      "sentence": "Comparison of GDBMS Fuzzing Fuzzer YearGDBMS Type Language Oracle Feature Grand[115]2022PropertyGremlin Differential Different GDBMSs GDsmith[ 70]2023 Cypher RD2[109]2023 RDFSPARQL GDBMeter[ 57]2023 PropertyGremlin and Cypher MetamorphicQuery Partition Gamera[ 119]2023Statement Rewriting, Data Manipulation GraphGenie[ 51]2024 Statement Rewriting GRev[76]2024Cypher, Gremlin, nGQL GraspDB[ 71]2024 Cypher QuDi[114]2024 Gremlin Query Splitting GQS[111]2025CypherConstraint-solving Backward Solving Gslicer[75]2025 Differential Different Graphs queries are typically generated by first creating a query skeleton and then recursively populating it with patterns and expressions.",
+      "context_before": "of graph database systems. Table 6summarizes existing GDBMS fuzzers and their implementation characteristics. As pioneering fuzzers for GDBMSs, Grand [ 115] and GDSmith [ 70] established representative pipelines for database instance and query generation, which most subsequent fuzzers have built upon. For Gremlin, query generator models callable APIs, defining rules for random traversal, and synthesizing the final query by constructing valid method invocation chains. In contrast, Cypher ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:24 X. Gao et al. Table 6.",
+      "context_after": "RD2 [ 109] further provides an implementation for fuzzing the SPARQL query language under the RDF model. More recent studies—including GDBMeter [ 57], Gamera [ 119], GraphGenie [ 51], GRev [ 76], GraspDB [ 71], QuDi [ 114], GQS [ 111], and Gslicer [ 75]—focus on in-depth exploration of test oracles. While these oracles share conceptual similarities with those used in relational DBMS fuzzing, they are specifically tailored to the unique characteristics and challenges of GDBMSs. 5.2 Fuzzing for Other DBMS Types Unicorn [ 107] targetsTime-Series Database Management Systems (TSDBMSs ) using a hyb",
+      "section": "5 Non-relational DBMS Fuzzing and New Techniques",
+      "page": 24,
+      "char_offset": 68751,
+      "cited_reference": {
+        "number": 57,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing graph database engines via query partitioning. In Proceedings of the ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M51",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 57]",
+      "technique": null,
+      "sentence": "More recent studies—including GDBMeter [ 57], Gamera [ 119], GraphGenie [ 51], GRev [ 76], GraspDB [ 71], QuDi [ 114], GQS [ 111], and Gslicer [ 75]—focus on in-depth exploration of test oracles.",
+      "context_before": "GDBMeter[ 57]2023 PropertyGremlin and Cypher MetamorphicQuery Partition Gamera[ 119]2023Statement Rewriting, Data Manipulation GraphGenie[ 51]2024 Statement Rewriting GRev[76]2024Cypher, Gremlin, nGQL GraspDB[ 71]2024 Cypher QuDi[114]2024 Gremlin Query Splitting GQS[111]2025CypherConstraint-solving Backward Solving Gslicer[75]2025 Differential Different Graphs queries are typically generated by first creating a query skeleton and then recursively populating it with patterns and expressions. RD2 [ 109] further provides an implementation for fuzzing the SPARQL query language under the RDF model.",
+      "context_after": "While these oracles share conceptual similarities with those used in relational DBMS fuzzing, they are specifically tailored to the unique characteristics and challenges of GDBMSs. 5.2 Fuzzing for Other DBMS Types Unicorn [ 107] targetsTime-Series Database Management Systems (TSDBMSs ) using a hybrid generator that produces valid queries by mutating time-series–specific AST nodes. It injects an agent into each TSDB process to intercept runtime exceptions silently suppressed by built-in exception handling and report them to an analyzer, thereby exposing otherwise hidden crashes. Spatter [ 30]",
+      "section": "5 Non-relational DBMS Fuzzing and New Techniques",
+      "page": 24,
+      "char_offset": 69533,
+      "cited_reference": {
+        "number": 57,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing graph database engines via query partitioning. In Proceedings of the ACMSIGSOFT International Symposium on Software Testing and Analysis.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M52",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 30]",
+      "technique": null,
+      "sentence": "Spatter [ 30] focuses on SpatialDatabaseManagementSystems (SDBMSs ) and employs affine equivalent inputs as its test oracle.",
+      "context_before": "f test oracles. While these oracles share conceptual similarities with those used in relational DBMS fuzzing, they are specifically tailored to the unique characteristics and challenges of GDBMSs. 5.2 Fuzzing for Other DBMS Types Unicorn [ 107] targetsTime-Series Database Management Systems (TSDBMSs ) using a hybrid generator that produces valid queries by mutating time-series–specific AST nodes. It injects an agent into each TSDB process to intercept runtime exceptions silently suppressed by built-in exception handling and report them to an analyzer, thereby exposing otherwise hidden crashes.",
+      "context_after": "It first generates spatial instance from random geometric shape and applies predefined derivative strategies to create topologically related geometries. By further constructing affine-transformed database instances (e.g., via rotation, scaling, or translation) and comparing query results across equivalent instances, Spatter detects spatial logic bugs. SQLaw [ 47] is a fuzzer for GPU DBMSs that detects bugs by comparing the execution results of thesamequeryon GPUandCPUbackends.ToaddressthelimitedcoverageofGPUexecutionpaths, SQLaw adopts a mining-based approach. It generates raw data and queries",
+      "section": "5 Non-relational DBMS Fuzzing and New Techniques",
+      "page": 24,
+      "char_offset": 70314,
+      "cited_reference": {
+        "number": 30,
+        "text": "Wenjing Deng, Qiuyang Mang, Chengyu Zhang, and Manuel Rigger. 2024. Finding logic bugs in spatial database engines via affine equivalent inputs. Proceedings of the ACM on Management of Data 2, 6 (2024), 1–26.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M53",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 52]",
+      "technique": null,
+      "sentence": "A Comprehensive Survey on Database Management System Fuzzing 268:25 Additionally, SQLxDiff [ 52] enables differential testing for emerging systems—including TSDBMSs, DDBMSs, and streaming DBMSs—by mapping their queries to equivalent constructs in PostgreSQL.",
+      "context_before": "Ss). It constructs sharded database instances with explicit cross-partition dependencies and focuses on generating operation sequences that directly affect cluster state. By monitoring synchronization-related code coverage, DepState identifies precise timing windows during which cluster state changes occur and strategically injects cluster-modifying operations within these windows. Coverage feedback from these executions is then used to guide the exploration of a broader range of state-affecting operation sequences. ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026.",
+      "context_after": "BUZZBEE [ 110] extends fuzzing to non-relational systems such as key-value, graph, and document DBMSs. Inspired by Intermediate Representation techniques, BUZZBEE adapts ASTs to uniformly model heterogeneous syntaxes and enforces semantic validity through annotationguided constraints resolved by a custom parser. Coni [ 29] further broadens fuzzing to database connectors by modeling connector functions as state machines and performing differential testing across different connectors using generated operation sequences. 5.3 LLM in Fuzzing Recent studies integrate LLMs into DBMS fuzzing generato",
+      "section": "5 Non-relational DBMS Fuzzing and New Techniques",
+      "page": 24,
+      "char_offset": 72133,
+      "cited_reference": {
+        "number": 52,
+        "text": "Yuancheng Jiang, Jianing Wang, Chuqi Zhang, Roland Yap, Zhenkai Liang, and Manuel Rigger. 2025. Enhanced differential testing in emerging database systems. arXiv:2501.01236. Retrieved from https://arxiv.org/abs/2501.01236",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M54",
+      "found_by": "name",
+      "surface": "ShQveL",
+      "technique": null,
+      "sentence": "Similarly, ShQveL [ 117] adopts an LLM-guided approach for multi-dialect testing by decoupling structural consistency from dialect-specific features, using SQL sketchesand LLM-basedfragmentsynthesistoinjectcomplexdialectconstructs.",
+      "context_before": "esting across different connectors using generated operation sequences. 5.3 LLM in Fuzzing Recent studies integrate LLMs into DBMS fuzzing generators and oracles. LLMs have been explored for tasks such as database dialect translation, schema-aware statement mutation, surrogate oracle construction based on learned bug patterns, and execution-log analysis for latent bug detection [ 50]. Forinstance,QTRAN[ 69]proposesatwo-phaseframeworktoimprovecross-dialecttestportability, combining RAG-enhanced SQL translation with a fine-tuned LLM that performs dialect-aware mutation in the target environment.",
+      "context_after": "Additionally,a Reinforcement Learning (RL)-guided LLM framework [ 74] iteratively optimizes prompt templates for test generation based on oracle feedback from discovered bugs or coverage improvements. 6 Benchmarks and Comparison As can be seen from the previous sections, there are quite a number of DBMS fuzzers and their structures are quite complex. In order to objectively evaluate the performance of each fuzzer, we present a comprehensive open-source toolkit, OpenDBFuzz,1which integrates five relational DBMS (SQLite, MySQL, TiDB, CockroachDB, and PostgreSQL) and eight fuzzing tools (SQLancer",
+      "section": "5.3 LLM in Fuzzing",
+      "page": 25,
+      "char_offset": 73445,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M55",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In order to objectively evaluate the performance of each fuzzer, we present a comprehensive open-source toolkit, OpenDBFuzz,1which integrates five relational DBMS (SQLite, MySQL, TiDB, CockroachDB, and PostgreSQL) and eight fuzzing tools (SQLancer, DQE, AMOEBA, SQLRight, Squirrel, Troc, APOLLO, and Radar), and provides entry function, interactive interface, runtime configuration files, and detailed command prompts.",
+      "context_before": "rget environment. Similarly, ShQveL [ 117] adopts an LLM-guided approach for multi-dialect testing by decoupling structural consistency from dialect-specific features, using SQL sketchesand LLM-basedfragmentsynthesistoinjectcomplexdialectconstructs.Additionally,a Reinforcement Learning (RL)-guided LLM framework [ 74] iteratively optimizes prompt templates for test generation based on oracle feedback from discovered bugs or coverage improvements. 6 Benchmarks and Comparison As can be seen from the previous sections, there are quite a number of DBMS fuzzers and their structures are quite complex.",
+      "context_after": "Based on this toolkit, we conduct fair experimental comparisons using the same configurations and focus on answering the following questions: Q1:(Between fuzzers) For each bug type, which fuzzer performs best under a unified testing environment? Q2:(Fuzzers themselves) Which bug types can each fuzzer detect, and do they exhibit the advantages they claim? Q3:(Inside fuzzers) Which module approach performs best among the various steps of fuzzing? Therefore, in this section, we first introduce the database instances and the open-source fuzzers, followed by a classification of the bugs they disco",
+      "section": "6 Benchmarks and Comparison",
+      "page": 25,
+      "char_offset": 74028,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M56",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SourceCode of Related Fuzzers Fuzzer Supported DBMS Link SQLancer[ 92] (PQS[92], NoREC[ 90], TLP[91], QPG[13], CERT[14], DQP[15], CODDTest[ 112])SQLite, MySQL, TiDB, MariaDB CockroachDB, DuckDB, OceanBasehttps://github.",
+      "context_before": "nerate database instances, while others, such as SQLsmith [ 4], require existing database instances. For the latter experiments, we use TPC-H as the database instance. The mutation-based fuzzers generate diverse test cases by mutating seed statements. We use official unit test cases provided by DBMS vendors as our seed corpus, including SQLite TCL test scripts [ 99], PostgreSQL official test infrastructure [ 89], and MySQL unit test samples [ 84]. 1https://github.com/Reverie4u/OpenDBFuzz ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:26 X. Gao et al. Table7.",
+      "context_after": "com/sqlancer/sqlancer DQE[96]SQLite, MySQL, MariaDB TiDB, CockroachDBhttps://github.com/tcse-iscas/dqetool SQLRight[ 68] SQLite, PostgreSQL, MySQL https://github.com/psu-security-universe/sqlright SQLsmith[ 4] SQLite, MonetDB, PostgreSQL https://github.com/anse1/sqlsmith Go-Randgen[ 87] MySQL, TiDB https://github.com/pingcap/go-randgen Squirrel[ 116]SQLite, PostgreSQL, MySQL, MariaDB https://github.com/s3team/Squirrel DT2[27] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/DT2 Troc[31] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc APOLLO[ 56] SQLite, PostgreSQL https://github.com/",
+      "section": "6.1 Database Instances and Test Cases",
+      "page": 26,
+      "char_offset": 75886,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp",
+        "coddtest",
+        "cert"
+      ]
+    },
+    {
+      "id": "M57",
+      "found_by": "name",
+      "surface": "sqlancer",
+      "technique": null,
+      "sentence": "com/sqlancer/sqlancer DQE[96]SQLite, MySQL, MariaDB TiDB, CockroachDBhttps://github.",
+      "context_before": "es by mutating seed statements. We use official unit test cases provided by DBMS vendors as our seed corpus, including SQLite TCL test scripts [ 99], PostgreSQL official test infrastructure [ 89], and MySQL unit test samples [ 84]. 1https://github.com/Reverie4u/OpenDBFuzz ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:26 X. Gao et al. Table7. SourceCode of Related Fuzzers Fuzzer Supported DBMS Link SQLancer[ 92] (PQS[92], NoREC[ 90], TLP[91], QPG[13], CERT[14], DQP[15], CODDTest[ 112])SQLite, MySQL, TiDB, MariaDB CockroachDB, DuckDB, OceanBasehttps://github.",
+      "context_after": "com/tcse-iscas/dqetool SQLRight[ 68] SQLite, PostgreSQL, MySQL https://github.com/psu-security-universe/sqlright SQLsmith[ 4] SQLite, MonetDB, PostgreSQL https://github.com/anse1/sqlsmith Go-Randgen[ 87] MySQL, TiDB https://github.com/pingcap/go-randgen Squirrel[ 116]SQLite, PostgreSQL, MySQL, MariaDB https://github.com/s3team/Squirrel DT2[27] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/DT2 Troc[31] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc APOLLO[ 56] SQLite, PostgreSQL https://github.com/sslab-gatech/apollo AMOEBA[ 72] PostgreSQL, CockroachDB https://bit.ly/3I995jL TxChe",
+      "section": "6.1 Database Instances and Test Cases",
+      "page": 26,
+      "char_offset": 76106,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M58",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 54]",
+      "technique": null,
+      "sentence": "ly/3I995jL TxCheck[ 54] TiDB, MySQL, MariaDB https://github.",
+      "context_before": "Bhttps://github.com/tcse-iscas/dqetool SQLRight[ 68] SQLite, PostgreSQL, MySQL https://github.com/psu-security-universe/sqlright SQLsmith[ 4] SQLite, MonetDB, PostgreSQL https://github.com/anse1/sqlsmith Go-Randgen[ 87] MySQL, TiDB https://github.com/pingcap/go-randgen Squirrel[ 116]SQLite, PostgreSQL, MySQL, MariaDB https://github.com/s3team/Squirrel DT2[27] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/DT2 Troc[31] MySQL, MariaDB, TiDB https://github.com/tcse-iscas/Troc APOLLO[ 56] SQLite, PostgreSQL https://github.com/sslab-gatech/apollo AMOEBA[ 72] PostgreSQL, CockroachDB https://bit.",
+      "context_after": "com/JZuming/TxCheck WriteCheck[ 28]MySQL, PostgreSQL, SQLite MariaDB, CockroachDB, TiDBhttps://github.com/tcse-iscas/WriteCheck PINOLO[ 44] MySQL, MariaDB, TiDB, OceanBase https://github.com/qaqcatz/impomysql.git EET[55]MySQL, PostgreSQL, SQLite ClickHouse, TiDBhttps://github.com/JZuming/EET Radar[97]SQLite, MySQL, MariaDB CockroachDB, TiDBhttps://github.com/tcse-iscas/radar DDLCheck[ 98]SQLite, MySQL, TiDB PostgreSQL, MariaDB, CockroachDBhttps://figshare.com/s/e222bd56c4925375ac34 Fucci[36] MySQL, TiDB, MariaDB https://github.com/Reverie4u/Fucci 6.2 Open-source Fuzzers Source code is crucial",
+      "section": "6.1 Database Instances and Test Cases",
+      "page": 26,
+      "char_offset": 76774,
+      "cited_reference": {
+        "number": 54,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting transactional bugs in database engines via{graph-based }oracle construction. In Proceedings of the USENIX Symposium on Operating Systems Design and Implementation. 397–417.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M59",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "State & Resource MutaSQL ✓ ✓ ✓ Esql ✓ ✓ ✓ ✓ ✓ ADUSA ✓ NoREC ✓ ✓ ✓ ✓ ✓ SQLRight ✓ ✓ ✓ ✓ ✓ DQE ✓ ✓ ✓ ✓ TLP ✓ ✓ ✓ ✓ ✓ TQS ✓ ✓ ✓ ✓ ✓ PINOLO ✓ ✓ ✓ ✓ EET ✓ ✓ ✓ ✓ ✓ DQP ✓ ✓ ✓ ✓ ✓ ✓ CODD ✓ ✓ Radar ✓ ✓ ✓ ✓ ✓ DDLcheck ✓ ✓ ✓ Kangaroo ✓ ✓ ✓ PQS ✓ ✓ ✓ ✓ ✓ ✓ Mozi ✓ ✓ SQLsmith ✓ ✓ ✓ ✓ ✓ DT2 ✓ Troc ✓ Fucci ✓ TxCheck ✓ ✓ WriteCheck ✓ ✓ ✓(b) Crash Bugs Detection Fuzzer Buffer OverflowMemory LifecycleIllegal MemoryResource Exh.",
+      "context_before": "choices and targeted objectives. The complete classification criteria and the sources of the bug list are detailed in the Appendix A, and the full list is released as part of OpenDBFuzz. 6.4 Evaluation Metrics We adopt the standard metrics for evaluating fuzzing performance: ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:27 Table 8. Overview of Fuzzer Detection Capabilities Across Different Bug Types (a) Logic Bugs Detection Fuzzer Interface & Pars.Type & Seman.Query Opt.Exec. EngineFeature Inter.",
+      "context_after": "Internal LogicArith. Errors squirrel ✓ ✓ ✓ ✓ ✓ Squill ✓ ✓ ✓ ✓ Griffin ✓ ✓ ✓ ✓ LEGO ✓ ✓ ✓ ✓ DynSQL ✓ ✓ ✓ ✓ ✓ SOFT ✓ ✓ ✓ ✓ ✓ ✓ SQLRight ✓ ✓ ✓ SQLsmith ✓ ✓ ✓ ✓ ✓ ✓ TLP ✓ NoREC ✓ DQP ✓ ✓ CODD ✓ DDLcheck ✓ ✓ ✓ ✓ ✓ Kangaroo ✓ ✓ ✓ PQS ✓ ✓ Radar ✓ ✓ EET ✓ ✓ ✓ ✓ DQE ✓ TxCheck ✓ ✓ ✓ ✓ ✓ ✓ WriteCheck ✓ (c) Performance Bugs Detection Fuzzer Query RewritingPlan SelectionExecution Loss AMOEBA ✓ ✓ APOLLO ✓ ✓ ✓ CERT ✓ ✓ PUPPY ✓ ✓ Mozi ✓ ✓(d) Transactional Bugs Detection Fuzzer Lock ManagementMVCC VisibilityInvalidState TransitionRecovery Rollback DT2 ✓ ✓ ✓ TRoc ✓ ✓ TxCheck ✓ ✓ WriteCheck ✓ ✓ ✓ Fucci ✓ ✓ Esql",
+      "section": "6.4 Evaluation Metrics",
+      "page": 27,
+      "char_offset": 79187,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "dqp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M60",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Errors squirrel ✓ ✓ ✓ ✓ ✓ Squill ✓ ✓ ✓ ✓ Griffin ✓ ✓ ✓ ✓ LEGO ✓ ✓ ✓ ✓ DynSQL ✓ ✓ ✓ ✓ ✓ SOFT ✓ ✓ ✓ ✓ ✓ ✓ SQLRight ✓ ✓ ✓ SQLsmith ✓ ✓ ✓ ✓ ✓ ✓ TLP ✓ NoREC ✓ DQP ✓ ✓ CODD ✓ DDLcheck ✓ ✓ ✓ ✓ ✓ Kangaroo ✓ ✓ ✓ PQS ✓ ✓ Radar ✓ ✓ EET ✓ ✓ ✓ ✓ DQE ✓ TxCheck ✓ ✓ ✓ ✓ ✓ ✓ WriteCheck ✓ (c) Performance Bugs Detection Fuzzer Query RewritingPlan SelectionExecution Loss AMOEBA ✓ ✓ APOLLO ✓ ✓ ✓ CERT ✓ ✓ PUPPY ✓ ✓ Mozi ✓ ✓(d) Transactional Bugs Detection Fuzzer Lock ManagementMVCC VisibilityInvalidState TransitionRecovery Rollback DT2 ✓ ✓ ✓ TRoc ✓ ✓ TxCheck ✓ ✓ WriteCheck ✓ ✓ ✓ Fucci ✓ ✓ Esql ✓ PQS ✓ —Numberofbugs: The number of bugs can directly quantify the bug detection capabilities of different fuzzers.",
+      "context_before": "Overview of Fuzzer Detection Capabilities Across Different Bug Types (a) Logic Bugs Detection Fuzzer Interface & Pars.Type & Seman.Query Opt.Exec. EngineFeature Inter.State & Resource MutaSQL ✓ ✓ ✓ Esql ✓ ✓ ✓ ✓ ✓ ADUSA ✓ NoREC ✓ ✓ ✓ ✓ ✓ SQLRight ✓ ✓ ✓ ✓ ✓ DQE ✓ ✓ ✓ ✓ TLP ✓ ✓ ✓ ✓ ✓ TQS ✓ ✓ ✓ ✓ ✓ PINOLO ✓ ✓ ✓ ✓ EET ✓ ✓ ✓ ✓ ✓ DQP ✓ ✓ ✓ ✓ ✓ ✓ CODD ✓ ✓ Radar ✓ ✓ ✓ ✓ ✓ DDLcheck ✓ ✓ ✓ Kangaroo ✓ ✓ ✓ PQS ✓ ✓ ✓ ✓ ✓ ✓ Mozi ✓ ✓ SQLsmith ✓ ✓ ✓ ✓ ✓ DT2 ✓ Troc ✓ Fucci ✓ TxCheck ✓ ✓ WriteCheck ✓ ✓ ✓(b) Crash Bugs Detection Fuzzer Buffer OverflowMemory LifecycleIllegal MemoryResource Exh.Internal LogicArith.",
+      "context_after": "Some studies submit the detected bugs to the open-source database community to obtain the number of community confirmed bugs. However, to make comparative experiments more discriminative, we need to conduct subsequent experiments on older versions of the DBMSs, so this method cannot be used. To avoid overrating methods that output highly redundant bugs, we used expert filtering to remove the obviously repetitive bugs, thereby improving the rationality of this metric. —Validity: Validity measures the proportion of test statements that are syntactically and semantically correct. Only valid stat",
+      "section": "6.4 Evaluation Metrics",
+      "page": 27,
+      "char_offset": 79619,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec",
+        "dqp",
+        "pqs",
+        "cert"
+      ]
+    },
+    {
+      "id": "M61",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "including PQS, NoREC, TLP, QPG, DQP, and DQE.",
+      "context_before": "odules and then tested the feedback module independently. For logic bug detection, we conducted comparative experiments on six fuzzers, ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:28 X. Gao et al. (a)Number of bugs (SQLite) (b)Number of bugs (MySQL) (c)Number of bugs (TiDB) (d)Number of bugs (CockroachDB) (e)Validity(SQLite) (f)Validity (MySQL) (g)Validity (TiDB) (h)Validity (CockroachDB) (i)Valid Cases/s (SQLite) (j)Valid Cases/s (MySQL) (k)Valid Cases/s (TiDB) (l)Valid Cases/s (CockroachDB) Fig. 23. Comparison of oracles for detecting logic bugs.",
+      "context_after": "Among them, TLP and QPG only differ in feedback, and the other modules are consistent. Experiments were conducted on four popular DBMSs: MySQL 8.0.39, SQLite 3.47.2, TiDB 8.4.0, and CockroachDB 25.1.0. The number of bugs, the validity, and the valid cases per second are used as evaluation metrics. We conducted a 240-minute evaluation, with results shown in Figure 23. The results reveal substantial differences across methods and databases in terms of the number of bugs, validity ratios, and the throughput of valid cases per second, largely attributable to differences in statement generation st",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 28,
+      "char_offset": 82441,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp",
+        "qpg",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M62",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Among them, TLP and QPG only differ in feedback, and the other modules are consistent.",
+      "context_before": "ependently. For logic bug detection, we conducted comparative experiments on six fuzzers, ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:28 X. Gao et al. (a)Number of bugs (SQLite) (b)Number of bugs (MySQL) (c)Number of bugs (TiDB) (d)Number of bugs (CockroachDB) (e)Validity(SQLite) (f)Validity (MySQL) (g)Validity (TiDB) (h)Validity (CockroachDB) (i)Valid Cases/s (SQLite) (j)Valid Cases/s (MySQL) (k)Valid Cases/s (TiDB) (l)Valid Cases/s (CockroachDB) Fig. 23. Comparison of oracles for detecting logic bugs. including PQS, NoREC, TLP, QPG, DQP, and DQE.",
+      "context_after": "Experiments were conducted on four popular DBMSs: MySQL 8.0.39, SQLite 3.47.2, TiDB 8.4.0, and CockroachDB 25.1.0. The number of bugs, the validity, and the valid cases per second are used as evaluation metrics. We conducted a 240-minute evaluation, with results shown in Figure 23. The results reveal substantial differences across methods and databases in terms of the number of bugs, validity ratios, and the throughput of valid cases per second, largely attributable to differences in statement generation strategies and oracle designs. TLP and QPG generally report more bugs and achieve higher",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 28,
+      "char_offset": 82487,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M63",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP and QPG generally report more bugs and achieve higher throughput, but at the cost of lower query validity, owing to their efficient metamorphic transformationsandbroadexplorationofqueryplans.",
+      "context_before": "y differ in feedback, and the other modules are consistent. Experiments were conducted on four popular DBMSs: MySQL 8.0.39, SQLite 3.47.2, TiDB 8.4.0, and CockroachDB 25.1.0. The number of bugs, the validity, and the valid cases per second are used as evaluation metrics. We conducted a 240-minute evaluation, with results shown in Figure 23. The results reveal substantial differences across methods and databases in terms of the number of bugs, validity ratios, and the throughput of valid cases per second, largely attributable to differences in statement generation strategies and oracle designs.",
+      "context_after": "Incontrast,DQEexhibitsthelowestthroughput due to frequent failures during database instance construction; however, it still reports many bugs because it aggressively treats warning inconsistencies across SELECT, UPDATE, and DELETE statements as bugs. Finally, DQP and PQS report fewer bugs, primarily due to their stricter and more conservative oracle validation, which reduces false positives. As shown in Figure 23(a)–(d), TiDB reports the largest number of bugs, reaching 381. This is partly due to issues inherent to TiDB itself and partly related to the relatively higher false positive ACM Comp",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 28,
+      "char_offset": 83115,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M64",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "Finally, DQP and PQS report fewer bugs, primarily due to their stricter and more conservative oracle validation, which reduces false positives.",
+      "context_before": "validity ratios, and the throughput of valid cases per second, largely attributable to differences in statement generation strategies and oracle designs. TLP and QPG generally report more bugs and achieve higher throughput, but at the cost of lower query validity, owing to their efficient metamorphic transformationsandbroadexplorationofqueryplans.Incontrast,DQEexhibitsthelowestthroughput due to frequent failures during database instance construction; however, it still reports many bugs because it aggressively treats warning inconsistencies across SELECT, UPDATE, and DELETE statements as bugs.",
+      "context_after": "As shown in Figure 23(a)–(d), TiDB reports the largest number of bugs, reaching 381. This is partly due to issues inherent to TiDB itself and partly related to the relatively higher false positive ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:29 (a)Number of bugs (SQLite) (b) Validity (SQLite) (c) Valid Cases/s (SQLite) Fig. 24. Comparison of feedback ways for detecting logic bugs. rates of DQE, TLP, and QPG on this platform. In contrast, CockroachDB reports the fewest bugs, with at most six. H",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 28,
+      "char_offset": 83561,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "dqp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M65",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "rates of DQE, TLP, and QPG on this platform.",
+      "context_before": "fewer bugs, primarily due to their stricter and more conservative oracle validation, which reduces false positives. As shown in Figure 23(a)–(d), TiDB reports the largest number of bugs, reaching 381. This is partly due to issues inherent to TiDB itself and partly related to the relatively higher false positive ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:29 (a)Number of bugs (SQLite) (b) Validity (SQLite) (c) Valid Cases/s (SQLite) Fig. 24. Comparison of feedback ways for detecting logic bugs.",
+      "context_after": "In contrast, CockroachDB reports the fewest bugs, with at most six. However, this does not imply that CockroachDB is more stable; rather, the result is largely due to compatibility issues with DQE and NoREC, which frequently cause failures during database instance construction and significantly reduce the number of valid queries. Regarding the validity shown in Figures 23(e)–(h), all fuzzers achieve the highest query validity on MySQL, approaching 100%. This is mainly because MySQL adopts a relatively standard SQL dialect, allowing the generators of different methods to adapt well. For most f",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 29,
+      "char_offset": 84190,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M66",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "However, this does not imply that CockroachDB is more stable; rather, the result is largely due to compatibility issues with DQE and NoREC, which frequently cause failures during database instance construction and significantly reduce the number of valid queries.",
+      "context_before": "s. As shown in Figure 23(a)–(d), TiDB reports the largest number of bugs, reaching 381. This is partly due to issues inherent to TiDB itself and partly related to the relatively higher false positive ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. A Comprehensive Survey on Database Management System Fuzzing 268:29 (a)Number of bugs (SQLite) (b) Validity (SQLite) (c) Valid Cases/s (SQLite) Fig. 24. Comparison of feedback ways for detecting logic bugs. rates of DQE, TLP, and QPG on this platform. In contrast, CockroachDB reports the fewest bugs, with at most six.",
+      "context_after": "Regarding the validity shown in Figures 23(e)–(h), all fuzzers achieve the highest query validity on MySQL, approaching 100%. This is mainly because MySQL adopts a relatively standard SQL dialect, allowing the generators of different methods to adapt well. For most fuzzers, regardless of the underlying DBMS, validity remains stable between 80% and 100% throughout the evaluation, indicating that existing generators are generally robust, though there is still room for improvement. Since validity is computed cumulatively, the metric gradually stabilizes as execution proceeds. One exception is No",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 29,
+      "char_offset": 84303,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M67",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "One exception is NoREC on CockroachDB, where validity is noticeably lower due to frequent errors caused by version incompatibilities.",
+      "context_before": "er of valid queries. Regarding the validity shown in Figures 23(e)–(h), all fuzzers achieve the highest query validity on MySQL, approaching 100%. This is mainly because MySQL adopts a relatively standard SQL dialect, allowing the generators of different methods to adapt well. For most fuzzers, regardless of the underlying DBMS, validity remains stable between 80% and 100% throughout the evaluation, indicating that existing generators are generally robust, though there is still room for improvement. Since validity is computed cumulatively, the metric gradually stabilizes as execution proceeds.",
+      "context_after": "Regarding the valid cases per second metric shown in Figures 23(i)–(l), all methods achieve substantially higher throughput on SQLite, reaching up to 69,642 valid cases per second. Isolated tests indicate that query execution speed itself is not the primary bottleneck for achieving high throughput. Since all fuzzers connect via JDBC, SQLite’s embedded architecture and extremely low system overhead provide a significant advantage in connection handling. DQE and NoREC achieve only about one case per second on CockroachDB because of compatibility. In contrast, PQS exhibits high throughput becaus",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 29,
+      "char_offset": 85147,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M68",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "DQE and NoREC achieve only about one case per second on CockroachDB because of compatibility.",
+      "context_before": "proceeds. One exception is NoREC on CockroachDB, where validity is noticeably lower due to frequent errors caused by version incompatibilities. Regarding the valid cases per second metric shown in Figures 23(i)–(l), all methods achieve substantially higher throughput on SQLite, reaching up to 69,642 valid cases per second. Isolated tests indicate that query execution speed itself is not the primary bottleneck for achieving high throughput. Since all fuzzers connect via JDBC, SQLite’s embedded architecture and extremely low system overhead provide a significant advantage in connection handling.",
+      "context_after": "In contrast, PQS exhibits high throughput because its oracle adopts a partial validation strategy, checking only row-count consistency rather than full result equivalence. The feedback mechanisms currently applied to logic bug detection mainly include coverage feedback and query plan feedback. We conducted comparative experiments using TLP, QPG, and SQLRight on another SQLite version (3.28.0). These three methods employ the same oracle: TLP does not incorporate feedback, QPG adopts query plan feedback, and SQLRight utilizes coverage feedback. In addition, TLP and QPG share the same generation",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 29,
+      "char_offset": 85738,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M69",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "In contrast, PQS exhibits high throughput because its oracle adopts a partial validation strategy, checking only row-count consistency rather than full result equivalence.",
+      "context_before": "quent errors caused by version incompatibilities. Regarding the valid cases per second metric shown in Figures 23(i)–(l), all methods achieve substantially higher throughput on SQLite, reaching up to 69,642 valid cases per second. Isolated tests indicate that query execution speed itself is not the primary bottleneck for achieving high throughput. Since all fuzzers connect via JDBC, SQLite’s embedded architecture and extremely low system overhead provide a significant advantage in connection handling. DQE and NoREC achieve only about one case per second on CockroachDB because of compatibility.",
+      "context_after": "The feedback mechanisms currently applied to logic bug detection mainly include coverage feedback and query plan feedback. We conducted comparative experiments using TLP, QPG, and SQLRight on another SQLite version (3.28.0). These three methods employ the same oracle: TLP does not incorporate feedback, QPG adopts query plan feedback, and SQLRight utilizes coverage feedback. In addition, TLP and QPG share the same generation-based statement generator. As shown in Figure 24, the results indicate that fuzzing is highly sensitive to the SQLite version, leading to substantial differences across ve",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 29,
+      "char_offset": 85832,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M70",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We conducted comparative experiments using TLP, QPG, and SQLRight on another SQLite version (3.",
+      "context_before": "the primary bottleneck for achieving high throughput. Since all fuzzers connect via JDBC, SQLite’s embedded architecture and extremely low system overhead provide a significant advantage in connection handling. DQE and NoREC achieve only about one case per second on CockroachDB because of compatibility. In contrast, PQS exhibits high throughput because its oracle adopts a partial validation strategy, checking only row-count consistency rather than full result equivalence. The feedback mechanisms currently applied to logic bug detection mainly include coverage feedback and query plan feedback.",
+      "context_after": "28.0). These three methods employ the same oracle: TLP does not incorporate feedback, QPG adopts query plan feedback, and SQLRight utilizes coverage feedback. In addition, TLP and QPG share the same generation-based statement generator. As shown in Figure 24, the results indicate that fuzzing is highly sensitive to the SQLite version, leading to substantial differences across versions. Mismatches between the fuzzer adaptation and the tested DBMS may introduce syntax incompatibilities and even result in a large number of false positives. Within 240 minutes, QPG detects bugs more efficiently tha",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 29,
+      "char_offset": 86127,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M71",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "These three methods employ the same oracle: TLP does not incorporate feedback, QPG adopts query plan feedback, and SQLRight utilizes coverage feedback.",
+      "context_before": "bedded architecture and extremely low system overhead provide a significant advantage in connection handling. DQE and NoREC achieve only about one case per second on CockroachDB because of compatibility. In contrast, PQS exhibits high throughput because its oracle adopts a partial validation strategy, checking only row-count consistency rather than full result equivalence. The feedback mechanisms currently applied to logic bug detection mainly include coverage feedback and query plan feedback. We conducted comparative experiments using TLP, QPG, and SQLRight on another SQLite version (3.28.0).",
+      "context_after": "In addition, TLP and QPG share the same generation-based statement generator. As shown in Figure 24, the results indicate that fuzzing is highly sensitive to the SQLite version, leading to substantial differences across versions. Mismatches between the fuzzer adaptation and the tested DBMS may introduce syntax incompatibilities and even result in a large number of false positives. Within 240 minutes, QPG detects bugs more efficiently than TLP, and the number of detected bugs gradually converges in the later stage, demonstrating that query plan feedback effectively improves the efficiency of l",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 29,
+      "char_offset": 86229,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M72",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "In addition, TLP and QPG share the same generation-based statement generator.",
+      "context_before": "per second on CockroachDB because of compatibility. In contrast, PQS exhibits high throughput because its oracle adopts a partial validation strategy, checking only row-count consistency rather than full result equivalence. The feedback mechanisms currently applied to logic bug detection mainly include coverage feedback and query plan feedback. We conducted comparative experiments using TLP, QPG, and SQLRight on another SQLite version (3.28.0). These three methods employ the same oracle: TLP does not incorporate feedback, QPG adopts query plan feedback, and SQLRight utilizes coverage feedback.",
+      "context_after": "As shown in Figure 24, the results indicate that fuzzing is highly sensitive to the SQLite version, leading to substantial differences across versions. Mismatches between the fuzzer adaptation and the tested DBMS may introduce syntax incompatibilities and even result in a large number of false positives. Within 240 minutes, QPG detects bugs more efficiently than TLP, and the number of detected bugs gradually converges in the later stage, demonstrating that query plan feedback effectively improves the efficiency of logic bug detection. In contrast, SQLRight detects only five bugs; this outcome",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 29,
+      "char_offset": 86381,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M73",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "Within 240 minutes, QPG detects bugs more efficiently than TLP, and the number of detected bugs gradually converges in the later stage, demonstrating that query plan feedback effectively improves the efficiency of logic bug detection.",
+      "context_before": "using TLP, QPG, and SQLRight on another SQLite version (3.28.0). These three methods employ the same oracle: TLP does not incorporate feedback, QPG adopts query plan feedback, and SQLRight utilizes coverage feedback. In addition, TLP and QPG share the same generation-based statement generator. As shown in Figure 24, the results indicate that fuzzing is highly sensitive to the SQLite version, leading to substantial differences across versions. Mismatches between the fuzzer adaptation and the tested DBMS may introduce syntax incompatibilities and even result in a large number of false positives.",
+      "context_after": "In contrast, SQLRight detects only five bugs; this outcome cannot be solely attributed to coverage feedback, but is largely due to the low efficiency and poor validity of its mutation-based generator. The impact of coverage feedback on ACM Comput. Surv., Vol. 58, No. 10, Article 268. Publication date: April 2026. 268:30 X. Gao et al. (a) Number of Crashes (SQLite) (b) New Edges (SQLite) (c) Coverage (SQLite) Fig. 25. Comparison of fuzzers for detecting crashes (a) Threshold (Apollo) (b)Threshold(AMOEBA) (c)Feedback (AMOEBA) Fig. 26. Evaluation of fuzzers for detecting performance bugs log",
+      "section": "6.5 Logic Bugs Detection Comparison",
+      "page": 29,
+      "char_offset": 86765,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg",
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M73"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:11:05Z",
+    "is_model_written": true,
+    "summary": "A survey of fuzzing for database management systems, arguing that DBMSs differ from general-purpose software in internal structure, input and output, and testing objectives, so they need strategies that existing fuzzing surveys do not cover. It defines a general fuzzing procedure, decomposes existing methods against it, and classifies them again by testing objective across DBMS components. It also presents an open-source toolkit, OpenDBFuzz, and uses it to compare the surveyed methods experimentally.",
+    "narrative": "SQLancer's oracles are among the principal subjects of the survey: PQS, NoREC, TLP and QPG each get their own pipeline description, and the experimental section runs them against one another, reporting for instance that QPG detects bugs more efficiently than TLP within 240 minutes.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "background",
+      "M5": "background",
+      "M6": "background",
+      "M7": "background",
+      "M9": "background",
+      "M15": "background",
+      "M23": "background",
+      "M25": "background",
+      "M27": "definition",
+      "M28": "definition",
+      "M29": "definition",
+      "M30": "definition",
+      "M31": "definition",
+      "M32": "definition",
+      "M33": "definition",
+      "M34": "definition",
+      "M35": "definition",
+      "M71": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The survey describes SQLancer's methods and runs them through its own toolkit; nothing in the mentions says its toolkit is built on SQLancer's codebase."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "A survey classifies and compares; it does not claim to extend any of the techniques it reviews."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M71"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M71",
+            "sentence": "Within 240 minutes, QPG detects bugs more efficiently than TLP, and the number of detected bugs gradually converges in the later stage, demonstrating that query plan feedback effectively improves the efficiency of logic bug detection.",
+            "section": "6.5 Logic Bugs Detection Comparison",
+            "page": 29
+          }
+        ],
+        "reasoning": "The survey's experimental section runs SQLancer's oracles against one another and reports the outcome, comparing QPG's bug-detection efficiency with TLP's over 240 minutes. That is an empirical comparison rather than a description.",
+        "techniques": [
+          "tlp",
+          "qpg"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The survey treats SQLancer's oracles as the central methods of the field and devotes several pipelines to them, but no mention shown calls them the state of the art in those words."
+      }
+    },
+    "disagreements": [],
+    "unresolved": [
+      "39 further mentions were omitted from this task; the survey mentions SQLancer's techniques throughout, so the roles recorded here cover the mentions read rather than all of them."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3802034.json b/_data/papers/paper_doi_10_1145_3802034.json
new file mode 100644
index 0000000..2210898
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3802034.json
@@ -0,0 +1,1048 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:03:29Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3802034",
+    "title": "DBugScribe: Automatic Database Bug Reproduction from Community Reports",
+    "authors": [
+      "Suyang Zhong",
+      "Mo Sha",
+      "Sheng Wang",
+      "Fangyuan Zhou",
+      "Feifei Li",
+      "Kian-Lee Tan"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": "10.1145/3802034",
+    "arxiv_id": null,
+    "s2_paper_id": "8f02d1ed0dfacfff322cccdddf971240b3c8136d",
+    "url": "https://doi.org/10.1145/3802034",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3802034",
+      "retrieved_at": "2026-09-10T15:03:29Z",
+      "chars": 97615,
+      "content_sha256": "sha256:8309339282d6a8a73415144eb5cd6677c4303040bdd17ad922bf2d08895b487c"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 26,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3474
+      },
+      {
+        "number": "2",
+        "title": "Preliminaries",
+        "start": 11506
+      },
+      {
+        "number": "2.1",
+        "title": "Problem Formulation",
+        "start": 11522
+      },
+      {
+        "number": "2.2",
+        "title": "Scope and Assumptions",
+        "start": 15194
+      },
+      {
+        "number": "3",
+        "title": "The DBugScribe Framework",
+        "start": 16319
+      },
+      {
+        "number": "3.1",
+        "title": "Design Philosophy",
+        "start": 16346
+      },
+      {
+        "number": "3.2",
+        "title": "System Architecture",
+        "start": 19717
+      },
+      {
+        "number": "3.3",
+        "title": "DBugScribe DSL: DBMS Bug Scenarios",
+        "start": 21484
+      },
+      {
+        "number": "3.4",
+        "title": "LLM-Driven Synthesis Pipeline",
+        "start": 32378
+      },
+      {
+        "number": "3.5",
+        "title": "Execution and Validation Engine",
+        "start": 39120
+      },
+      {
+        "number": "4",
+        "title": "Experimental Evaluation",
+        "start": 44028
+      },
+      {
+        "number": "4.1",
+        "title": "Dataset Construction and Curation",
+        "start": 44934
+      },
+      {
+        "number": "4.2",
+        "title": "Reproduction Effectiveness Analysis",
+        "start": 49204
+      },
+      {
+        "number": "4.3",
+        "title": "Fine-Grained Performance Analysis",
+        "start": 56560
+      },
+      {
+        "number": "4.4",
+        "title": "Extended Applications and Discoveries",
+        "start": 60808
+      },
+      {
+        "number": "5",
+        "title": "Discussion",
+        "start": 69968
+      },
+      {
+        "number": "6",
+        "title": "Related Work",
+        "start": 73067
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 78129
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Abdulaziz Alaboudi and Thomas D. LaToza. 2023. What constitutes debugging? An exploratory study of debugging episodes.Empir. Softw. Eng.28, 5 (2023), 117.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Arvind Arasu, Raghav Kaushik, and Jian Li. 2011. Data generation using declarative constraints. InProceedings of the 2011 ACMSIGMOD International Conference on Management of Data(Athens, Greece)(SIGMOD ’11). Association for Computing Machinery, New York, NY, USA, 685–696.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. InProceedings of the 45th International Conference on Software Engineering(Melbourne, Victoria, Australia)(ICSE ’23). IEEE Press, 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering(Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 133, 13 pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Earl T. Barr, Mark Harman, Yue Jia, Alexandru Marginean, and Justyna Petke. 2015. Automated Software Transplantation. InProceedings of the 2015 International Symposium on Software Testing and Analysis(Baltimore, MD, USA)(ISSTA 2015). Association for Computing Machinery, New York, NY, USA, 257–269.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Kaiwen Chen, Yueting Chen, Nick Koudas, and Xiaohui Yu. 2025. Reliable Text-to-SQL with Adaptive Abstention.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Sibei Chen, Ju Fan, Bin Wu, Nan Tang, Chao Deng, Pengyi Wang, Ye Li, Jian Tan, Feifei Li, Jingren Zhou, and Xiaoyong Du. 2025. Automatic Database Configuration Debugging using Retrieval-Augmented Language Models. 2025), 27 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Jack Clark, Alastair F. Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. InProceedings of the Nineteenth European Conference on Computer Systems(Athens, Greece)(EuroSys ’24). Association for Computing Machinery, New York, NY, USA, 754–768.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 10,
+      "text": "Ziyu Cui, Wensheng Dou, Yu Gao, Rui Yang, Yingying Zheng, Jiansen Song, Yuan Feng, and Jun Wei. 2025. Simple Testing Can Expose Most Critical Transaction Bugs: Understanding and Detecting Write-Specific Serializability Violations in Database Systems.Proc. VLDB Endow.18, 8 (Sept. 2025), 2547–2560.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Wenjing Deng, Qiuyang Mang, Chengyu Zhang, and Manuel Rigger. 2024. Finding Logic Bugs in Spatial Database Engines via Affine Equivalent Inputs. 2024), 26 pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 12,
+      "text": "Han Fu, Chang Liu, Bin Wu, Feifei Li, Jian Tan, and Jianling Sun. 2023. CatSQL: Towards Real World Natural Language to SQL Applications.Proc. VLDB Endow.16, 6 (Feb. 2023), 1534–1547.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs. InProceedings of the Twentieth European Conference on Computer Systems(Rotterdam, Netherlands)(EuroSys ’25). Association for Computing Machinery, New York, NY, USA, 1061–1076.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "Dawei Gao, Haibin Wang, Yaliang Li, Xiuyu Sun, Yichen Qian, Bolin Ding, and Jingren Zhou. 2024. Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluation.Proc. VLDB Endow.17, 5 (Jan. 2024), 1132–1145.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Xiyue Gao, Zhuang Liu, Yiran Shen, Hui Li, Yingfan Liu, Hongjun Xiao, Yanguo Peng, and Jiangtao Cui. 2025. Fucci: Database Transaction Fuzzing via Random Conflict Construction and Multilevel Constraint Solving.Proc. VLDB Endow. 18, 6 (Aug. 2025), 1879–1891.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Victor Giannakouris and Immanuel Trummer. 2025. 𝜆-Tune: Harnessing Large Language Models for Automated Database System Tuning. 2025), 26 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Yue Gong, Chuan Lei, Xiao Qin, Kapil Vaidya, Balakrishnan Narayanaswamy, and Tim Kraska. 2025. SQLens: An End-to-End Framework for Error Detection and Correction in Text-to-SQL. InAdvances in Neural Information Processing Systems (NeurIPS).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023. Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis. In2023 USENIX Annual Technical Conference (USENIX ATC 23). 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Zijin Hong, Zheng Yuan, Qinggang Zhang, Hao Chen, Junnan Dong, Feiran Huang, and Xiao Huang. 2025. NextGeneration Database Interfaces: A Survey of LLM-based Text-to-SQL.IEEE Transactions on Knowledge and Data Engineering37, 12 (2025), 7328–7345.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. In32nd USENIX Security Symposium (USENIX Security 23). USENIX Association, Anaheim, CA, 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). USENIX Association, Boston, MA, 397–417. 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026. 157:24 Suyang Zhong et al.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting Logic Bugs in Database Engines via Equivalent Expression Transformation. In18th USENIX Symposium on Operating Systems Design and Implementation (OSDI 24). USENIX Association, Santa Clara, CA, 821–835.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Guoliang Jin, Linhai Song, Xiaoming Shi, Joel Scherpelz, and Shan Lu. 2012. Understanding and detecting real-world performance bugs. InACM SIGPLAN Conference on Programming Language Design and Implementation, PLDI ’12, Beijing, China-June 11 - 16, 2012. ACM, 77–88.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. APOLLO: automatic detection and diagnosis of performance regressions in database systems.Proc. VLDB Endow.13, 1 (Sept. 2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Sungmin Kang, Juyeon Yoon, and Shin Yoo. 2023. Large Language Models are Few-Shot Testers: Exploring LLM-Based General Bug Reproduction. InProceedings of the 45th International Conference on Software Engineering(Melbourne, Victoria, Australia)(ICSE ’23). IEEE Press, 2312–2323.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Jiale Lao, Yibo Wang, Yufei Li, Jianping Wang, Yunjia Zhang, Zhiyuan Cheng, Wanghu Chen, Mingjie Tang, and Jianguo Wang. 2024. GPTuner: A Manual-Reading Database Tuning System via GPT-Guided Bayesian Optimization. Proc. VLDB Endow.17, 8 (April 2024), 1939–1952.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Tanakorn Leesatapornwongsa, Xiang Ren, and Suman Nath. 2022. FlakeRepro: automated and efficient reproduction of concurrency-related flaky tests. InProceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Singapore, Singapore)(ESEC/FSE 2022). Association for Computing Machinery, New York, NY, USA, 1509–1520.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Haoyang Li, Jing Zhang, Cuiping Li, and Hong Chen. 2023. RESDSQL: Decoupling Schema Linking and Skeleton Parsing for Text-to-SQL. InProceedings of the 37th AAAI Conference on Artificial Intelligence (AAAI). AAAI Press, 13067–13075.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jinyang Li, Binyuan Hui, Ge Qu, Jiaxi Yang, Binhua Li, Bowen Li, Bailin Wang, Bowen Qin, Ruiying Geng, Nan Huo, Xuanhe Zhou, Chenhao Ma, Guoliang Li, Kevin C.C. Chang, Fei Huang, Reynold Cheng, and Yongbin Li. 2023. Can LLM already serve as a database interface? a big bench for large-scale database grounded text-to-SQLs. InProceedings of the 37th International Conference on Neural Information Proc",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Zhaodonghui Li, Haitao Yuan, Huiming Wang, Gao Cong, and Lidong Bing. 2024. LLM-R2: A Large Language Model Enhanced Rule-based Rewrite System for Boosting Query Efficiency.Proc. VLDB Endow.18, 1 (2024), 53–65.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-Oriented DBMS Fuzzing. In Engineering (ICDE). 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Yiyuan Bai, Qiang Zhang, and Yu Jiang. 2024. WingFuzz: Implementing Continuous Fuzzing for DBMSs. In2024 USENIX Annual Technical Conference (USENIX ATC 24). USENIX Association, Santa Clara, CA, 479–492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic Detection of Performance Bugs in Database Systems Using Equivalent Queries. InProceedings of the 44th International Conference on Software Engineering (Pittsburgh, Pennsylvania)(ICSE ’22). Association for Computing Machinery, New York, NY, USA, 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Yuyu Luo, Guoliang Li, Ju Fan, Chengliang Chai, and Nan Tang. 2025. Natural Language to SQL: State of the Art and Open Problems.Proc. VLDB Endow.18, 12 (2025), 5466–5471.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Kyle Luoma and Arun Kumar. 2025. SNAILS: Schema Naming Assessments for Improved LLM-Based SQL Inference.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "Niels Mündler, Mark Müller, Jingxuan He, and Martin Vechev. 2024. SWT-bench: Testing and validating real-world bug-fixes with code agents.Advances in Neural Information Processing Systems37 (2024), 81857–81887.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Chris Parnin and Alessandro Orso. 2011. Are automated debugging techniques actually helping programmers?. In Proceedings of the 20th International Symposium on Software Testing and Analysis, ISSTA 2011, Toronto, ON, Canada, July 17-21, 2011. ACM, 199–209.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Mohammad Masudur Rahman, Foutse Khomh, and Marco Castelluccio. 2022. Works for Me! Cannot Reproduce-A Large Scale Empirical Study of Non-reproducible Bugs.Empir. Softw. Eng.27, 5 (2022), 111.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Shanto Rahman, Saikat Dutta, and August Shi. 2025. Understanding and Improving Flaky Test Classification.Proc. ACM Program. Lang.9, OOPSLA2, Article 320 (Oct. 2025), 27 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. 4, No. 3 (SIGMOD)",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (nov 2020), 30 pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "Anupam Sanghi, Raghav Sood, Jayant Haritsa, and Srikanta Tirthapura. 2018. Scalable and dynamic regeneration of big data volumes. (2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Torsten Scholak, Nathan Schucher, and Dzmitry Bahdanau. 2021. PICARD: Parsing Incrementally for Constrained Auto-Regressive Decoding from Language Models. InProceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP). ACL, 9895–9901.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Andreas Seltenreich. 2022. Sqlsmith. https://github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Peter Shaw, Ming-Wei Chang, Panupong Pasupat, and Kristina Toutanova. 2021. Compositional Generalization and Natural Language Variation: Can a Semantic Parsing Approach Handle Both?. InProceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (ACL-IJCNLP). ACL, 922–938.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Jiansen Song, Wensheng Dou, Yu Gao, Ziyu Cui, Yingying Zheng, Dong Wang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction.Proc. VLDB Endow. 17, 8 (may 2024), 1884–1897.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Jiansen Song, Wensheng Dou, Yingying Zheng, Yu Gao, Ziyu Cui, Wei Wang, and Jun Wei. 2025. Detecting SchemaRelated Logic Bugs in Relational DBMSs via Equivalent Database Construction.Proc. VLDB Endow.18, 7 (Aug. 2025), 2281–2294.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Ruoxi Sun, Sercan Ö. Arik, Alexandre Muzio, Lesly Miculicich, Satya Kesav Gundabathula, Pengcheng Yin, Hanjun Dai, Hootan Nakhost, Rajarishi Sinha, Zifeng Wang, and Tomas Pfister. 2024. SQL-PaLM: Improved large language model adaptation for Text-to-SQL.Trans. Mach. Learn. Res.2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting Logic Bugs of Join Optimizations in DBMS.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Bailin Wang, Richard Shin, Xiaodong Liu, Oleksandr Polozov, and Matthew Richardson. 2020. RAT-SQL: Relation-Aware Schema Encoding and Linking for Text-to-SQL Parsers. InProceedings of the 58th Annual Meeting of the Association for Computational Linguistics (ACL). ACL, 7567–7578.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Xinchen Wang, Pengfei Gao, Xiangxin Meng, Chao Peng, Ruida Hu, Yun Lin, and Cuiyun Gao. 2025. AEGIS: An Agent-based Framework for Bug Reproduction from Issue Descriptions. InProceedings of the 33rd ACM International Conference on the Foundations of Software Engineering(Clarion Hotel Trondheim, Trondheim, Norway)(FSE Companion ’25). Association for Computing Machinery, New York, NY, USA, 331–342.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Wenqian Deng, and Yu Jiang. 2025. DDLumos: Understanding and Detecting Atomic DDL Bugs in DBMSs. In2025 USENIX Annual Technical Conference (USENIX ATC 25). 1327–1341.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Wenqian Deng, and Yu Jiang. 2025. Fawkes: Finding Data Durability Bugs in DBMSs via Recovered Data State Verification. InProceedings of the ACMSIGOPS 31st Symposium on Operating Systems Principles(Lotte Hotel World, Seoul, Republic of Korea)(SOSP ’25). Association for Computing Machinery, New York, NY, USA, 670–684.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "John Yang, Carlos Jimenez, Alexander Wettig, Kilian Lieret, Shunyu Yao, Karthik Narasimhan, and Ofir Press. 2024. Swe-agent: Agent-computer interfaces enable automated software engineering.Advances in Neural Information Processing Systems37 (2024), 50528–50652.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Jingyi Yang, Peizhi Wu, Gao Cong, Tieying Zhang, and Xiao He. 2022. SAM: Database Generation from Query Workloads with Supervised Autoregressive Models. InProceedings of the 2022 International Conference on Management of Data(Philadelphia, PA, USA)(SIGMOD ’22). Association for Computing Machinery, New York, NY, USA, 1542–1555.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Xiao Yang, Mo Sha, Yiran Li, Suyang Zhong, Sheng Wang, Fangyuan Zhou, and Feifei Li. 2026. SQLens: Continuous Code-to-SQL Visibility in the Wild. InProceedings of the 2026 ACMSIGMOD International Conference on Management of Data(Bengaluru, India)(SIGMOD ’26).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Yicun Yang, Zhaoguo Wang, Yu Xia, Zhuoran Wei, Haoran Ding, Ruzica Piskac, Haibo Chen, and Jinyang Li. 2025. Automated Validating and Fixing of Text-to-SQL Translation with Execution Consistency.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Tao Yu, Rui Zhang, Kai Yang, Michihiro Yasunaga, Dongxu Wang, Zifan Li, James Ma, Irene Li, Qingning Yao, Shanelle Roman, Zilin Zhang, and Dragomir Radev. 2018. Spider: A Large-Scale Human-Labeled Dataset for Complex and Cross-Domain Semantic Parsing and Text-to-SQL Task. InProceedings of the 2018 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguis",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Tao Yu, Rui Zhang, Michihiro Yasunaga, Yi Chern Tan, Xi Victoria Lin, Suyi Li, Heyang Er, Irene Li, Bo Pang, Tao Chen, Emily Ji, Shreya Dixit, David Proctor, Sungrok Shim, Jonathan Kraft, Vincent Zhang, Caiming Xiong, Richard Socher, and Dragomir Radev. 2019. SParC: Cross-Domain Semantic Parsing in Context. InProceedings of the 57th 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026. 157:",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Chao Zhang, Yuren Mao, Yijiang Fan, Yu Mi, Yunjun Gao, Lu Chen, Dongfang Lou, and Jinshu Lin. 2024. FinSQL: Model-Agnostic LLMs-based Text-to-SQL Framework for Financial Analysis. InCompanion of the 2024 International Conference on Management of Data(Santiago AA, Chile)(SIGMOD ’24). Association for Computing Machinery, New York, NY, USA, 93–105.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 64,
+      "text": "Xinyi Zhang, Tiantian Chen, Zhentao Han, Zhaoyan Hong, Wei Lu, Sheng Wang, Mo Sha, Anni Wang, Shuang Liu, Yakun Zhang, Feifei Li, and Xiaoyong Du. 2026. Why Database Manuals Are Not Enough: Efficient and Reliable Configuration Tuning for DBMSs via Code-Driven LLM Agents.Proc. VLDB Endow.19, 6 (2026), 1358–1371.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric Xing, et al .2023. Judging llm-as-a-judge with mt-bench and chatbot arena.Advances in neural information processing systems36 (2023), 46595–46623.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security(Virtual Event, USA)(CCS ’20). Association for Computing Machinery, New York, NY, USA, 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Suyang Zhong and Manuel Rigger. 2024. Understanding and Reusing Test Suites Across Database Systems. 2024), 26 pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 68,
+      "text": "Suyang Zhong and Manuel Rigger. 2026. Scaling Automated Database System Testing. InProceedings of the 31st ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2(USA) (ASPLOS ’26). Association for Computing Machinery, New York, NY, USA, 1677–1692.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 69,
+      "text": "Victor Zhong, Caiming Xiong, and Richard Socher. 2017. Seq2SQL: Generating Structured Queries from Natural Language using Reinforcement Learning.arXiv preprint arXiv:1709.00103(2017).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Wei Zhou, Yuyang Gao, Xuanhe Zhou, and Guoliang Li. 2025. Cracking SQL Barriers: An LLM-based Dialect Translation System.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Zeyang Zhuang, Penghui Li, Pingchuan Ma, Wei Meng, and Shuai Wang. 2023. Testing Graph Database Systems via Graph-Aware Metamorphic Relations.Proc. VLDB Endow.17, 4 (Dec. 2023), 836–848.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Thomas Zimmermann, Rahul Premraj, Nicolas Bettenburg, Sascha Just, Adrian Schröter, and Cathrin Weiss. 2010. What Makes a Good Bug Report?IEEE Trans. Software Eng.36, 5 (2010), 618–643. Received October 2025; revised January 2026; accepted February 2026 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 3,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. InProceedings of the 45th International Conference on Software Engineering(Melbourne, Victoria, Australia)(ICSE ’23). IEEE Press, 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 4,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering(Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 133, 13 pages.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert"
+      ]
+    },
+    {
+      "number": 5,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 9,
+      "text": "Jack Clark, Alastair F. Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. InProceedings of the Nineteenth European Conference on Computer Systems(Athens, Greece)(EuroSys ’24). Association for Computing Machinery, New York, NY, USA, 754–768.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 11,
+      "text": "Wenjing Deng, Qiuyang Mang, Chengyu Zhang, and Manuel Rigger. 2024. Finding Logic Bugs in Spatial Database Engines via Affine Equivalent Inputs. 2024), 26 pages.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 13,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs. InProceedings of the Twentieth European Conference on Computer Systems(Rotterdam, Netherlands)(EuroSys ’25). Association for Computing Machinery, New York, NY, USA, 1061–1076.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 21,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23). USENIX Association, Boston, MA, 397–417. 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026. 157:24 Suyang Zhong et al.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 36,
+      "text": "Qiuyang Mang, Jinsheng Ba, Pinjia He, and Manuel Rigger. 2025. Finding Logic Bugs in Graph-processing Systems via Graph-cutting.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 41,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026. DBugScribe: Automatic Database Bug Reproduction from Community Reports 157:25",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (nov 2020), 30 pages.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 63,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages.",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 67,
+      "text": "Suyang Zhong and Manuel Rigger. 2024. Understanding and Reusing Test Suites Across Database Systems. 2024), 26 pages.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 68,
+      "text": "Suyang Zhong and Manuel Rigger. 2026. Scaling Automated Database System Testing. InProceedings of the 31st ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2(USA) (ASPLOS ’26). Association for Computing Machinery, New York, NY, USA, 1677–1692.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing sqlancer_pp"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 3,41–43,48,49,63]",
+      "technique": "qpg",
+      "sentence": "This inherent complexity makes them prone to various categories of defects, including logic bugs [ 3,41–43,48,49,63] and crash bugs [ 20,46,66].",
+      "context_before": "hina; Feifei Li, lifeifei@alibaba-inc.com, Alibaba Cloud, China; Kian-Lee Tan, dcstankl@nus.edu.sg, National University of Singapore, Singapore. This work is licensed under a Creative Commons Attribution 4.0 International License. ©2026 Copyright held by the owner/author(s). ACM 2836-6573/2026/6-ART157 https://doi.org/10.1145/3802034 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026. 157:2 Suyang Zhong et al. 1 Introduction Database management systems (DBMSs) are among the most complex and critical software systems in operation today, with codebases often spanning millions of lines.",
+      "context_after": "Despite decades of research and engineering effort, mainstream DBMSs continue to receive large volumes of bug reports every year. However, the process of handling these reports remains highly inefficient. Empirical studies show that developers spend nearly half of their overall programming time on debugging, with a non-trivial fraction of this effort dedicated specifically to reproducing and validating reported issues [ 1,38]. Analyses of large-scale software repositories further indicate that a significant portion of reported bugs cannot be reliably reproduced [ 23,39,72], either because the",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3650,
+      "cited_reference": {
+        "number": 3,
+        "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. InProceedings of the 45th International Conference on Software Engineering(Melbourne, Victoria, Australia)(ICSE ’23). IEEE Press, 2060–2071.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing qpg",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 10,13,54]",
+      "technique": null,
+      "sentence": "While recent bug-detection tools analyze large corpora of DBMS bugs [ 10,13,54], they primarily focus on manual categorization and statistical characterization rather than automated reproduction.",
+      "context_before": "nterpret each report, reconstruct the necessary database environment, and attempt reproduction through labor-intensive trial and error. Given these challenges, one might look to automated bug reproduction techniques for relief. However, existing automated bug reproduction frameworks focus on general-purpose software with well-defined APIs and stateless execution contexts,e.g., LIBRO [ 25], AEGIS [ 53], and SWEagent [ 56]. These approaches cannot handle the distinctive challenges of database systems, where failures depend on intricate dependencies among schema design and configuration settings.",
+      "context_after": "This gap highlights the need for a systematic framework to translate bug descriptions into structured and automated reproduction pipelines. Fortunately, recent advances in natural language processing offer a promising path to bridge this gap. Large Language Models (LLMs) have demonstrated the ability to parse unstructured text and translate informal descriptions into formal representations,e.g., through Text-to-SQL systems [ 14, 19,29,60]. We leverage this capability for a novel purpose: applying it to debugging contexts, translating bug behavior descriptions into candidate executable queries",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6106,
+      "cited_reference": {
+        "number": 13,
+        "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs. InProceedings of the Twentieth European Conference on Computer Systems(Rotterdam, Netherlands)(EuroSys ’",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 67]",
+      "technique": null,
+      "sentence": "Prior work [ 67] has demonstrated the potential for cross-DBMS test reuse.",
+      "context_before": "ngs can affect string comparison results. Reproduction therefore requires precise environmental specification and isolation. (4) Cross-system portability: General software test cases rarely transfer across systems because of incompatible APIs and substantially different software architectures. Database bugs, by contrast, often stem from shared SQL semantics despite dialect-level differences, making cross-DBMS reproduction possible. Moreover, many DBMSs share similar underlying architectures and optimization strategies (e.g., MySQL and MariaDB), allowing related bugs to manifest across systems.",
+      "context_after": "Nevertheless, automatically adapting real-world bug scenarios to different DBMSs remains an open challenge. These four characteristics make database bug reproduction a distinct problem requiring specialized treatment beyond general-purpose techniques. Our framework addresses these challenges through formal scenario specification, automated synthesis, containerized execution, and cross DBMS abstraction. Section 3 presents the complete architecture and end-to-end pipeline. 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026. DBugScribe: Automatic Database Bug Reproduction from Community",
+      "section": "2.1 Problem Formulation",
+      "page": 4,
+      "char_offset": 14504,
+      "cited_reference": {
+        "number": 67,
+        "text": "Suyang Zhong and Manuel Rigger. 2024. Understanding and Reusing Test Suites Across Database Systems. 2024), 26 pages.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 67]",
+      "technique": null,
+      "sentence": "SQuaLity [ 67] represents an initial attempt to address the test reuse problem; however, its ability is limited, as it is hard to automatically 4, No.",
+      "context_before": "synthesis through few-shot learning and template reuse. 3.3 DBugScribe DSL: DBMS Bug Scenarios We now detail the DBugScribeDSL that serves as the structured intermediate representation (Rationale 1), establishing formal foundations for bug scenario specification. 3.3.1 Motivation and Design Goals.Existing representations for database testing occupy two extremes. SQL test suites (e.g., MySQL regression tests) provide concrete cases but lack structured metadata: schema dependencies are implicit, oracles are embedded in expected output files, and cross-DBMS portability requires manual adaptation.",
+      "context_after": "3 (SIGMOD), Article 157. Publication date: June 2026. DBugScribe: Automatic Database Bug Reproduction from Community Reports 157:7 expand the test cases. Automated testing tools (e.g., SQLsmith [ 46] and SQLancer [ 41–43]) provide automation test generation in bug discovery, but the generated scenarios are transient. Once a bug is detected, it is often difficult to deterministically reproduce the generation paths. OurDBugScribeDSL navigates this trade-off throughdeclarative specifications with formal semantics. Three design goals guide our approach, with measurable success criteria:(1) Expres",
+      "section": "3.3 DBugScribe DSL: DBMS Bug Scenarios",
+      "page": 6,
+      "char_offset": 22027,
+      "cited_reference": {
+        "number": 67,
+        "text": "Suyang Zhong and Manuel Rigger. 2024. Understanding and Reusing Test Suites Across Database Systems. 2024), 26 pages.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLsmith [ 46] and SQLancer [ 41–43]) provide automation test generation in bug discovery, but the generated scenarios are transient.",
+      "context_before": "testing occupy two extremes. SQL test suites (e.g., MySQL regression tests) provide concrete cases but lack structured metadata: schema dependencies are implicit, oracles are embedded in expected output files, and cross-DBMS portability requires manual adaptation. SQuaLity [ 67] represents an initial attempt to address the test reuse problem; however, its ability is limited, as it is hard to automatically 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026. DBugScribe: Automatic Database Bug Reproduction from Community Reports 157:7 expand the test cases. Automated testing tools (e.g.",
+      "context_after": "Once a bug is detected, it is often difficult to deterministically reproduce the generation paths. OurDBugScribeDSL navigates this trade-off throughdeclarative specifications with formal semantics. Three design goals guide our approach, with measurable success criteria:(1) Expressiveness: our DBugScribeDSL covers schema definitions, data state, operation sequences, configuration parameters, and oracle predicates, thereby capturing all five components of the formal bug scenario representation.(2) Simplicity: a compact YAML-based syntax maintains human readability while enabling automated parsin",
+      "section": "3.3 DBugScribe DSL: DBMS Bug Scenarios",
+      "page": 7,
+      "char_offset": 22363,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "DBugScribe: Automatic Database Bug Reproduction from Community Reports 157:13 TiDB MySQLMariaDBCombdb2SQLite Clickhouse CockroachDBPostgreSQLDuckDBPolarDB Oceanbase0153045607590Reported bugs78 6 1 3 1321 1 1 LEGO[31] (ICDE ’23) Pinolo[18] (ATC ’23) TQS[51] (SIGMOD ’23) DQP[5] (SIGMOD ’24) EET[22] (OSDI ’24) Radar[48] (VLDB ’24) DDLCheck[49] (VLDB ’25) CODDTest[63] (SIGMOD ’25) Fig.",
+      "context_before": "calls to prevent data leakage. For DBMS execution environments, we use Dockerized instances for MySQL and MariaDB, and TiUP3, TiDB’s package manager, for TiDB. All target DBMS versions are pre-installed before the reproduction experiments. 4.1 Dataset Construction and Curation To evaluate DBugScribe, we constructedBugBench, a curated dataset of bugs from three relational DBMSs: MySQL, TiDB, and MariaDB. Our methodology for constructingBugBenchwas guided by 2https://openai.com/gpt-5 3https://docs.pingcap.com/tidb/stable/tiup-overview/ 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026.",
+      "context_after": "4. Testing targets of recent DBMS testing tools. Table 1. Bug Reproduction Dataset Statistics across DBMSs DBMS Collected Reported Conf. Fixed Dupl. FP N/A MySQL 62 91 53 6 3 0 29 TiDB 91 94 53 34 4 3 0 MariaDB 65 84 20 12 33 2 17 Total218 269 127 51 40 5 46 the principle of collecting bugs that are both recent and significant enough to be of interest to the database research community. Selection Criteria.Rather than randomly sampling from public trackers, we source bugs from recent state-of-the-art database testing tools, which helps ensure non-triviality and prior vetting by researchers. We",
+      "section": "4.1 Dataset Construction and Curation",
+      "page": 12,
+      "char_offset": 45294,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp",
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[ 5,18,22,\n31,48,49,51,63]",
+      "technique": "dqp",
+      "sentence": "Applying the above criteria, we identified in total eight recent tools [ 5,18,22, 31,48,49,51,63] that reported confirmed bugs in widely used open-source relational DBMSs.",
+      "context_before": "ligning with the database community; and (iii) our selection spans three reporting platforms (e.g., GitHub, Jira, and MySQL forums) and eight DBMS testing tools, capturing diverse reporting formats for evaluating DBugScribe. We include tools that (1) were published at top-tier database (SIGMOD, VLDB, ICDE) or systems (OSDI, SOSP, ATC, EuroSys) venues in the past three years, (2) provide publicly accessible reports with explicit developer confirmation (excluding tools without public links or identifiers), and (3) target relational DBMSs (excluding NoSQL systems such as graph DBMSs). Tool Survey.",
+      "context_after": "Among surveyed papers, we excluded three tools [ 11,36,71] that aim for NoSQL systems (e.g., Graph DBMSs), three tools [ 9,10,15] that aim for detecting transactional anomalies, and four tools [ 13,32,54,55] without publicly available links or artifacts related to the bug reports. Fig. 4 shows each tool’s targets and the number of reported bugs. MySQL, TiDB, and MariaDB are the most frequently targeted systems, each tested by at least six tools and averaging 86 reported bugs. As noted above, the resulting reports span multiple bug-tracking platforms and reporting styles, providing a diverse b",
+      "section": "4.1 Dataset Construction and Curation",
+      "page": 13,
+      "char_offset": 47126,
+      "cited_reference": {
+        "number": 5,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 11,36,71]",
+      "technique": null,
+      "sentence": "Among surveyed papers, we excluded three tools [ 11,36,71] that aim for NoSQL systems (e.",
+      "context_before": "diverse reporting formats for evaluating DBugScribe. We include tools that (1) were published at top-tier database (SIGMOD, VLDB, ICDE) or systems (OSDI, SOSP, ATC, EuroSys) venues in the past three years, (2) provide publicly accessible reports with explicit developer confirmation (excluding tools without public links or identifiers), and (3) target relational DBMSs (excluding NoSQL systems such as graph DBMSs). Tool Survey.Applying the above criteria, we identified in total eight recent tools [ 5,18,22, 31,48,49,51,63] that reported confirmed bugs in widely used open-source relational DBMSs.",
+      "context_after": "g., Graph DBMSs), three tools [ 9,10,15] that aim for detecting transactional anomalies, and four tools [ 13,32,54,55] without publicly available links or artifacts related to the bug reports. Fig. 4 shows each tool’s targets and the number of reported bugs. MySQL, TiDB, and MariaDB are the most frequently targeted systems, each tested by at least six tools and averaging 86 reported bugs. As noted above, the resulting reports span multiple bug-tracking platforms and reporting styles, providing a diverse basis for evaluating DBugScribe. We therefore selected these three DBMSs for constructingBug",
+      "section": "4.1 Dataset Construction and Curation",
+      "page": 13,
+      "char_offset": 47297,
+      "cited_reference": {
+        "number": 11,
+        "text": "Wenjing Deng, Qiuyang Mang, Chengyu Zhang, and Manuel Rigger. 2024. Finding Logic Bugs in Spatial Database Engines via Affine Equivalent Inputs. 2024), 26 pages.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 9,10,15]",
+      "technique": null,
+      "sentence": ", Graph DBMSs), three tools [ 9,10,15] that aim for detecting transactional anomalies, and four tools [ 13,32,54,55] without publicly available links or artifacts related to the bug reports.",
+      "context_before": "d at top-tier database (SIGMOD, VLDB, ICDE) or systems (OSDI, SOSP, ATC, EuroSys) venues in the past three years, (2) provide publicly accessible reports with explicit developer confirmation (excluding tools without public links or identifiers), and (3) target relational DBMSs (excluding NoSQL systems such as graph DBMSs). Tool Survey.Applying the above criteria, we identified in total eight recent tools [ 5,18,22, 31,48,49,51,63] that reported confirmed bugs in widely used open-source relational DBMSs. Among surveyed papers, we excluded three tools [ 11,36,71] that aim for NoSQL systems (e.g.",
+      "context_after": "Fig. 4 shows each tool’s targets and the number of reported bugs. MySQL, TiDB, and MariaDB are the most frequently targeted systems, each tested by at least six tools and averaging 86 reported bugs. As noted above, the resulting reports span multiple bug-tracking platforms and reporting styles, providing a diverse basis for evaluating DBugScribe. We therefore selected these three DBMSs for constructingBugBench. 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026. 157:14 Suyang Zhong et al. Table 2. Bug Reproduction Performance across Methods Method Success Rate (%)F→P (%) Time (s) Ove",
+      "section": "4.1 Dataset Construction and Curation",
+      "page": 13,
+      "char_offset": 47389,
+      "cited_reference": {
+        "number": 9,
+        "text": "Jack Clark, Alastair F. Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. InProceedings of the Nineteenth European Conference on Computer Systems(Athens, Greece)(EuroSys ’24). Association for Computing Ma",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "This bug was found by DQP [ 5] and marked as critical by developers.",
+      "context_before": "-5 mini), while Stages 2 and 4 use the full GPT-5 model, which explains the lower parsing cost relative to instantiation and refinement. 4.4 Extended Applications and Discoveries By capturing bugs as persistent, executable DBugScribeDSL scenarios rather than ephemeral manual steps,DBugScribeenables systematic quality assurance capabilities and cross-system bug discovery previously infeasible with manual reproduction. 4.4.1 Flaky Bug Reproduction.To demonstrate DBugScribe’s ability to handle complex and even non-deterministic bugs, we present a case study of a flaky bug from TiDB5(see Fig. 7(a)).",
+      "context_after": "The original bug report provides 5https://github.com/pingcap/tidb/issues/46580 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026. 157:18 Suyang Zhong et al. (a) Ambiguous user report ### Minimal reproduce step CREATE TABLE t0(c0 INT); ... SELECT /*+ MERGE_JOIN(t1, t0, v0)*/v0.c2, t1.c0 FROM v0, t0... ### What did you see instead Executing the above test case multiple times and the results are not the same. Sometimes it returns 4 rows. Sometimes it returns empty: {}. (b) DBugScribe DSL # ... operations: - step_id: op_run1 sql_query: SELECT /*+ MERGE_JOIN(t1, t0, v0)*/ ...; - step_id",
+      "section": "4.4 Extended Applications and Discoveries",
+      "page": 17,
+      "char_offset": 61271,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 13,54]",
+      "technique": null,
+      "sentence": "Recent work [ 13,54] highlights the benefits of mining historical bug patterns to inform automated testing tools.",
+      "context_before": "been a long-standing challenge in general software research [ 27,40] and DBugScribeuses a best-effort strategy based on bounded repetitions and cross-run comparisons: it targets flaky behaviors that manifest with non-trivial probability within a practical repetition budget, and it is not intended to cover extremely rare flakes (e.g., once in 1K or 1M runs). DBugScribecan reproduce certain flaky bugs from ambiguous descriptions by synthesizing bounded-repetition scenarios. 4.4.2 Automatic Bug Analysis.DBugScribeenables automated bug analysis through its structured, executable DSL representation.",
+      "context_after": "DBugScribefacilitates this direction by providing standardized, machine-readable bug scenarios that can be systematically analyzed. For example, 4, No. 3 (SIGMOD), Article 157. Publication date: June 2026. DBugScribe: Automatic Database Bug Reproduction from Community Reports 157:19 Table 6. SQL Function Usage by Bug Type Type Bugs # Funcs Top Functions (# occurrence) Logic 105 39cast(10),date_add(8),date_sub(4) Error 18 6cast(5),reverse(2),acos(1) Crash 36 29strcmp(2),hex(2),from_days(2) given bugs successfully reproduced in the DBugScribeDSL, one can prompt an LLM: “Implement a script that i",
+      "section": "4.4 Extended Applications and Discoveries",
+      "page": 18,
+      "char_offset": 63655,
+      "cited_reference": {
+        "number": 13,
+        "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs. InProceedings of the Twentieth European Conference on Computer Systems(Rotterdam, Netherlands)(EuroSys ’",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[ 67,68]",
+      "technique": null,
+      "sentence": "Key challenges include differences in supported statements, functions, operators, and type systems [ 67,68].",
+      "context_before": "atches, to converge to stable and diagnostic reproductions. Cross-dialect portability.Our current cross-DBMS bug discovery experiment (Section 4.4.5) is limited to MySQL-compatible systems, including MySQL, MariaDB, and TiDB, where modifying only the environment fields in setup allowed more than half of the bug-inducing cases to execute successfully without error. Reusing bug-inducing cases across different SQL dialect families is feasible but non-trivial. It requires rewriting the sql_query strings in schema_data_steps and operations, and validating or adapting oracles when semantics diverge.",
+      "context_after": "However, reuse within a dialect family is often practical, since these systems typically share core semantics and differ mainly in dialect-specific features. For example, analogous to the MySQL family,DBugScribe can likely reuse bugs within the PostgreSQL family (e.g., PostgreSQL, DuckDB, and CockroachDB) and execute them across systems. Reproduction benefits.We identify the below actionable insights. First,DBugScribecan provide earlier, machine-checkable reproduction artifacts for new reports. Second,DBugScribecan support triage and confirmation workflows by providing ready-to-run scenarios",
+      "section": "5 Discussion",
+      "page": 21,
+      "char_offset": 71934,
+      "cited_reference": {
+        "number": 68,
+        "text": "Suyang Zhong and Manuel Rigger. 2026. Scaling Automated Database System Testing. InProceedings of the 31st ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2(USA) (ASPLOS ’26). Association for Computing Machinery, New York, NY, USA, 1677–1",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing sqlancer_pp"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", PQS [ 43], TLP [ 42], and NoREC [ 41], with recent work further specializing in optimizer-centric settings [ 3,5,22,48,49, 51,63].",
+      "context_before": "tructured DBugScribe DSL representation provides a natural substrate for duplicate analysis (e.g., clustering scenarios by structural similarity), which may support future duplicate detection and triage. Third,DBugScribe enables large-scale studies of bug characteristics and detection tools by automating reproduction. 6 Related Work Database Bug Detection.Automated DBMS testing has progressed from random query generation [ 46] to techniques that leverage language-validity feedback [ 66] and stateful fuzzing [ 20] for crash detection. Logic bugs are commonly exposed via oracle-based testing,e.g.",
+      "context_after": "Transactional correctness relies on graph-based oracles and isolation checking [ 9,10,21], while performance issues are targeted via regression and cardinality-estimation testing [ 4,24,33]. In contrast,DBugScribefocuses on reproducing already-reported bugs by extracting and synthesizing complete, executable scenarios from unstructured natural-language reports. Database Bug Studies.Large-scale studies of DBMS bugs provide empirical guidance for testing and debugging. For example, Fu et al. [ 13] analyze built-in SQL function bugs, Wu et al. [ 54] study atomic DDL bugs under concurrent schema",
+      "section": "6 Related Work",
+      "page": 21,
+      "char_offset": 73348,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "tlp",
+        "norec",
+        "qpg"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 9,10,21]",
+      "technique": null,
+      "sentence": "Transactional correctness relies on graph-based oracles and isolation checking [ 9,10,21], while performance issues are targeted via regression and cardinality-estimation testing [ 4,24,33].",
+      "context_before": "similarity), which may support future duplicate detection and triage. Third,DBugScribe enables large-scale studies of bug characteristics and detection tools by automating reproduction. 6 Related Work Database Bug Detection.Automated DBMS testing has progressed from random query generation [ 46] to techniques that leverage language-validity feedback [ 66] and stateful fuzzing [ 20] for crash detection. Logic bugs are commonly exposed via oracle-based testing,e.g., PQS [ 43], TLP [ 42], and NoREC [ 41], with recent work further specializing in optimizer-centric settings [ 3,5,22,48,49, 51,63].",
+      "context_after": "In contrast,DBugScribefocuses on reproducing already-reported bugs by extracting and synthesizing complete, executable scenarios from unstructured natural-language reports. Database Bug Studies.Large-scale studies of DBMS bugs provide empirical guidance for testing and debugging. For example, Fu et al. [ 13] analyze built-in SQL function bugs, Wu et al. [ 54] study atomic DDL bugs under concurrent schema changes, and Cui et al. [ 10] characterize transaction bugs via write-specific serializability violations. These efforts emphasize characterization and detection, whereas DBugScribetranslates",
+      "section": "6 Related Work",
+      "page": 21,
+      "char_offset": 73480,
+      "cited_reference": {
+        "number": 9,
+        "text": "Jack Clark, Alastair F. Donaldson, John Wickerson, and Manuel Rigger. 2024. Validating Database System Isolation Level Implementations with Version Certificate Recovery. InProceedings of the Nineteenth European Conference on Computer Systems(Athens, Greece)(EuroSys ’24). Association for Computing Ma",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 13]",
+      "technique": null,
+      "sentence": "[ 13] analyze built-in SQL function bugs, Wu et al.",
+      "context_before": "NoREC [ 41], with recent work further specializing in optimizer-centric settings [ 3,5,22,48,49, 51,63]. Transactional correctness relies on graph-based oracles and isolation checking [ 9,10,21], while performance issues are targeted via regression and cardinality-estimation testing [ 4,24,33]. In contrast,DBugScribefocuses on reproducing already-reported bugs by extracting and synthesizing complete, executable scenarios from unstructured natural-language reports. Database Bug Studies.Large-scale studies of DBMS bugs provide empirical guidance for testing and debugging. For example, Fu et al.",
+      "context_after": "[ 54] study atomic DDL bugs under concurrent schema changes, and Cui et al. [ 10] characterize transaction bugs via write-specific serializability violations. These efforts emphasize characterization and detection, whereas DBugScribetranslates unstructured reports into executable scenarios, enabling systematic validation and accumulating structured artifacts for downstream analysis. Automated Bug Reproduction.Automated reproduction has been studied extensively in general software [ 6], with recent LLM-based systems such as LIBRO [ 25], AEGIS [ 53], and SWT-Bench [ 37]. Complementarily, bug rep",
+      "section": "6 Related Work",
+      "page": 21,
+      "char_offset": 73975,
+      "cited_reference": {
+        "number": 13,
+        "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Yanyang Zhao, Shanshan Li, and Yu Jiang. 2025. Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs. InProceedings of the Twentieth European Conference on Computer Systems(Rotterdam, Netherlands)(EuroSys ’",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:57:35Z",
+    "is_model_written": true,
+    "summary": "DBugScribe reproduces database bugs automatically from the reports people file. Bug reports carry the setup, the query and the symptom in prose and fragments rather than as a runnable script, so reproducing one is manual work; DBugScribe parses a report into a domain-specific language describing the scenario and executes it. Its dataset is drawn from bugs that eight recent testing tools reported in widely used open-source relational DBMSs.",
+    "narrative": "SQLancer appears in two roles, neither of them reuse. As background it is one of the automated generators whose test scenarios the paper calls transient -- the contrast that motivates working from durable community reports instead. More substantially, tools descended from SQLancer supply the evaluation corpus: DQP and CODDTest are among the eight tools whose confirmed bug reports DBugScribe reproduces, and one reproduced bug is noted as found by DQP and marked critical by developers. PQS, TLP and NoREC are cited as the oracle line that recent optimizer-centric work specialises.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "motivation",
+      "M6": "incidental",
+      "M7": "background",
+      "M8": "incidental",
+      "M9": "incidental",
+      "M10": "incidental",
+      "M11": "background",
+      "M12": "background",
+      "M13": "definition",
+      "M14": "background",
+      "M15": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "What DBugScribe consumes is the bug reports these tools produced, as evaluation data. That is not reuse of SQLancer's code, its generator or its workload."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The contribution is reproducing reported bugs from prose, which shares no mechanism with any SQLancer oracle."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The tools supply the corpus of bugs to reproduce rather than serving as baselines; no run of SQLancer or its oracles is reported against DBugScribe."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is described as providing automated test generation whose scenarios are transient, which is a limitation rather than a standing claim."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3802053.json b/_data/papers/paper_doi_10_1145_3802053.json
new file mode 100644
index 0000000..e6c95ed
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3802053.json
@@ -0,0 +1,1064 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:12Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3802053",
+    "title": "EPSC: Testing Database Management Systems via Equivalent Prepared Statement Construction",
+    "authors": [
+      "Chi Zhang",
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Dalong Shi",
+      "Linzhang Wang",
+      "Yu Jiang"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": "10.1145/3802053",
+    "arxiv_id": null,
+    "s2_paper_id": "f42307b2aae0cdc8c3fae609e844622539b26baa",
+    "url": "https://doi.org/10.1145/3802053",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3802053",
+      "retrieved_at": "2026-09-09T01:29:12Z",
+      "chars": 93719,
+      "content_sha256": "sha256:d89e91ff5491d3fcc8dab7d46ff331d7cdd9ee8134595f3a6a5ad6cb5c0c2cdb"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 25,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3385
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 13601
+      },
+      {
+        "number": "3",
+        "title": "Approach",
+        "start": 18866
+      },
+      {
+        "number": "3.1",
+        "title": "Approach Overview",
+        "start": 19582
+      },
+      {
+        "number": "3.2",
+        "title": "Statement Generation",
+        "start": 23479
+      },
+      {
+        "number": "3.3",
+        "title": "Error Validation",
+        "start": 26566
+      },
+      {
+        "number": "3.4",
+        "title": "Result Comparison",
+        "start": 37914
+      },
+      {
+        "number": "4",
+        "title": "Evaluation",
+        "start": 40105
+      },
+      {
+        "number": "4.1",
+        "title": "Effectiveness",
+        "start": 43356
+      },
+      {
+        "number": "4.2",
+        "title": "Selected Bugs",
+        "start": 48276
+      },
+      {
+        "number": "4.3",
+        "title": "Compared with Other Approaches",
+        "start": 62796
+      },
+      {
+        "number": "4.4",
+        "title": "Ablation Study",
+        "start": 68024
+      },
+      {
+        "number": "5",
+        "title": "Discussion",
+        "start": 73492
+      },
+      {
+        "number": "6",
+        "title": "Related Work",
+        "start": 76077
+      },
+      {
+        "number": "7",
+        "title": "Conclusion",
+        "start": 83667
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. InProceedings of the 45th International Conference on Software Engineering(Melbourne, Victoria, Australia)(ICSE ’23). IEEE Press, 2060–2071. doi:10.1109/ICSE48619.2023.00174",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. doi:10.1145/3654991",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Carsten Binnig, Donald Kossmann, Eric Lo, and M. Tamer Özsu. 2007. QAGen: generating query-aware test databases. InProceedings of the 2007 ACMSIGMOD International Conference on Management of Data(Beijing, China)(SIGMOD ’07). Association for Computing Machinery, New York, NY, USA, 341–352. doi:10.1145/1247480.1247520",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Prithvi Bisht, A. Prasad Sistla, and V. N. Venkatakrishnan. 2010. TAPS: automatically preparing safe SQL queries. In Proceedings of the 17th ACM Conference on Computer and Communications Security(Chicago, Illinois, USA)(CCS ’10). Association for Computing Machinery, New York, NY, USA, 645–647. doi:10.1145/1866307.1866384",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Pedro Bizarro, Nicolas Bruno, and David J. DeWitt. 2009. Progressive Parametric Query Optimization.IEEE Trans. on Knowl. and Data Eng.21, 4 (April 2009), 582–594. doi:10.1109/TKDE.2008.160",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Surajit Chaudhuri, Hongrae Lee, and Vivek R. Narasayya. 2010. Variance aware optimization of parameterized queries. InProceedings of the 2010 ACMSIGMOD International Conference on Management of Data(Indianapolis, Indiana, USA) (SIGMOD ’10). Association for Computing Machinery, New York, NY, USA, 531–542. doi:10.1145/1807167.1807226",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Tsong Y Chen, Shing C Cheung, and Shiu Ming Yiu. 2020. Metamorphic testing: a new approach for generating next test cases.arXiv preprint arXiv:2002.12543(2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Anshuman Dutt, Vivek Narasayya, and Surajit Chaudhuri. 2017. Leveraging Re-costing for Online Optimization of Parameterized Queries with Guarantees. InProceedings of the 2017 ACM International Conference on Management of Data(Chicago, Illinois, USA)(SIGMOD ’17). Association for Computing Machinery, New York, NY, USA, 1539–1554. doi:10.1145/3035918.3064040",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, and Yu Jiang. 2024. Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 146, 12 pages. doi:10.1145/3597503.3639210",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2023. Griffin: Grammar-Free DBMS Fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering(Rochester, MI, USA) (ASE ’22). Association for Computing Machinery, New York, NY, USA, Article 49, 12 pages. doi:10.1145/3551349.3560431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Kevin P Gaffney, Martin Prammer, Larry Brasfield, D Richard Hipp, Dan Kennedy, and Jignesh M Patel. 2022. SQLite: past, present, and future.Proceedings of the VLDB Endowment15, 12 (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Ahmad Ghazal, Dawit Seid, Bhashyam Ramesh, Alain Crolotte, Manjula Koppuravuri, and Vinod G. 2009. Dynamic plan generation for parameterized queries. InProceedings of the 2009 ACMSIGMOD International Conference on Management of Data(Providence, Rhode Island, USA)(SIGMOD ’09). Association for Computing Machinery, New York, NY, USA, 909–916. doi:10.1145/1559845.1559946",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023. Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis. In 2023 USENIX Annual Technical Conference (USENIX ATC 23). USENIX Association, Boston, MA, 345–358. https: //www.usenix.org/conference/atc23/presentation/hao",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Dongxu Huang, Qi Liu, Qiu Cui, Zhuhe Fang, Xiaoyu Ma, Fei Xu, Li Shen, Liu Tang, Yuxing Zhou, Menglong Huang, et al. 2020. TiDB: a Raft-based HTAP database.Proceedings of the VLDB Endowment13, 12 (2020), 3072–3084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting logic bugs in database engines via equivalent expression transformation. InProceedings of the 18th USENIX Conference on Operating Systems Design and Implementation(Santa Clara, CA, USA)(OSDI’24). USENIX Association, USA, Article 44, 15 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "George Klees, Andrew Ruef, Benji Cooper, Shiyi Wei, and Michael Hicks. 2018. Evaluating fuzz testing. InProceedings of the 2018 ACMSIGSAC conference on computer and communications security. 2123–2138.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering(Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 135, 12 pages. doi:10.1145/3597503.3639112 4, No. 3 (SIGMOD",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of DBMS with Coverage-based Guidance. InProceedings of the 31st USENIX Security Symposium (USENIX 2022). Boston, MA.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "MariaDB 2024. MariaDB. https://mariadb.org/. Accessed: March 19, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "William M McKeeman. 1998. Differential testing for software.Digital Technical Journal10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Bruce Momjian. 2001.PostgreSQL: introduction and concepts. Vol. 192. Addison-Wesley New York.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "MySQL 2024. MySQL. https://www.mysql.com/. Accessed: March 19, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Mark Raasveldt and Hannes Mühleisen. 2019. Duckdb: an embeddable analytical database. InProceedings of the 2019 international conference on management of data. 1981–1984.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Karthik Ramachandra, Kwanghyun Park, K. Venkatesh Emani, Alan Halverson, César Galindo-Legaria, and Conor Cunningham. 2017. Froid: optimization of imperative programs in a relational database.Proc. VLDB Endow.11, 4 (Dec. 2017), 432–444. doi:10.1145/3186728.3164140",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/33680",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (Nov. 2020), 30 pages. doi:10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, 667–682. https://www. usenix.org/conference/osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 28,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2018. SQLsmith: a random SQL query generator. https: //github.com/anse1/sqlsmith",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Donald R. Slutz. 1998. Massive Stochastic Testing of SQL. InProceedings of the 24rd International Conference on Very Large Data Bases (VLDB ’98). Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Donald R Slutz. 1998. Massive stochastic testing of SQL. InVLDB, Vol. 98. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. InProceedings of the 45th International Conference on Software Engineering(Melbourne, Victoria, Australia)(ICSE ’23). IEEE Press, 2072–2084. doi:10.1109/ICSE48619.2023.00175",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Rebecca Taft, Irfan Sharif, Andrei Matei, Nathan VanBenschoten, Jordan Lewis, Tobias Grieger, Kai Niemi, Andy Woods, Anne Birzin, Raphael Poss, Paul Bardea, Amruta Ranade, Ben Darnell, Bram Gruneir, Justin Jaffray, Lucy Zhang, and Peter Mattis. 2020. CockroachDB: The Resilient Geo-Distributed SQL Database. InProceedings of the 2020 ACMSIGMOD International Conference on Management of Data(Portland,",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting Logic Bugs of Join Optimizations in DBMS. doi:10.1145/3588909",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Kapil Vaidya, Anshuman Dutt, Vivek Narasayya, and Surajit Chaudhuri. 2021. Leveraging query logs and machine learning for parametric query optimization.Proc. VLDB Endow.15, 3 (Nov. 2021), 401–413. doi:10.14778/3494124.3494126",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages. doi:10.1145/3709674 Received October 2025; revised January 2026; accepted February 2026 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026.",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 1,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. InProceedings of the 45th International Conference on Software Engineering(Melbourne, Victoria, Australia)(ICSE ’23). IEEE Press, 2060–2071. doi:10.1109/ICSE48619.2023.00174",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "prints the DOI of the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. doi:10.1145/3654991",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "prints the DOI of the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 25,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 26,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (Nov. 2020), 30 pages. doi:10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 27,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, 667–682. https://www. usenix.org/conference/osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 35,
+      "text": "Chi Zhang and Manuel Rigger. 2025. Constant Optimization Driven Database System Testing. 2025), 24 pages. doi:10.1145/3709674 Received October 2025; revised January 2026; accepted February 2026 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026.",
+      "technique": "coddtest",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing coddtest",
+        "prints the DOI of the paper introducing coddtest",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 25]",
+      "technique": "norec",
+      "sentence": "These bugs can lead to system crashes, produce unexpected errors, or more subtly, to logic bugs [ 25], which cause the DBMS to silently produce incorrect results, making them hard for developers and users to notice.",
+      "context_before": "26. 176:2 Chi Zhang et al. 1 Introduction Database Management Systems (DBMSs) serve as critical software infrastructure, storing and managing data for a wide range of essential applications. Structured Query Language (SQL) statements serve as inputs to DBMSs, enabling users to specify the operations to be performed on data. DBMSs parse SQL statements, choose an appropriate execution plan, perform the corresponding operations, and return the results. However, given the complexity of DBMS implementations, bugs are likely to arise during the parsing, optimization, and execution of SQL statements.",
+      "context_after": "Recently, various approaches have been proposed for detecting logic bugs in DBMSs. Logic bugs, unlike other types of bugs, seldom cause noticeable system anomalies, making them difficult to detect. Consequently, detecting logic bugs has become a major focus in DBMS testing research. Most existing approaches rely on metamorphic testing to construct effective test oracles. Metamorphic relations form the core of metamorphic testing. Existing approaches leverage a variety of equivalence relations in DBMSs as metamorphic relations to determine whether logic bugs are triggered, including, but not l",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3958,
+      "cited_reference": {
+        "number": 25,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 25,31]",
+      "technique": "norec",
+      "sentence": "Existing approaches leverage a variety of equivalence relations in DBMSs as metamorphic relations to determine whether logic bugs are triggered, including, but not limited to, equivalence across different systems [ 29], clauses [ 25,31], expressions [ 15,26,35], and configurations [ 2,17].",
+      "context_before": "expected errors, or more subtly, to logic bugs [ 25], which cause the DBMS to silently produce incorrect results, making them hard for developers and users to notice. Recently, various approaches have been proposed for detecting logic bugs in DBMSs. Logic bugs, unlike other types of bugs, seldom cause noticeable system anomalies, making them difficult to detect. Consequently, detecting logic bugs has become a major focus in DBMS testing research. Most existing approaches rely on metamorphic testing to construct effective test oracles. Metamorphic relations form the core of metamorphic testing.",
+      "context_after": "While these methods have successfully detected a large number of logic bugs in widely used DBMSs, they have concentrated exclusively on bugs in ordinary statements, neglecting another critical language feature—prepared statements. Prepared statements were introduced as a specialized language feature to enhance the efficiency of repeated executions, as production environments commonly involve numerous statements that are nearly identical except for their literal values. In contrast to ordinary statements—which are executed directly—prepared statements allow literals to be specified as paramete",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4608,
+      "cited_reference": {
+        "number": 25,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Due to the large number of abnormal test cases generated and the repeated triggering of certain common errors, existing testing methods, such as SQLancer, ignore these errors and treat them as normal behavior.",
+      "context_before": "the prepared statement path. By following execution paths distinct from those of ordinary statements, prepared statements can reveal logic bugs when the two statements produce different results. Furthermore, we observed that DBMSs exhibit another type of logic bug, in which a statement that should have triggered an error is executed and produces a result. Detecting this type of logic bug is challenging. Fuzzing typically generates a large number of normal and abnormal test cases and feeds them into the DBMS under test, determining the presence of bugs based on the system’s behavior or output.",
+      "context_after": "1 However, the erroneous parts may exhibit inconsistent behavior between equivalent ordinary and prepared statements. In DBMSs, certain optimizations, such as short-circuit evaluation, may cause parts of a statement’s expressions to be skipped during execution. If an erroneous part is skipped in an ordinary statement but not in a prepared statement—because the bound parameters are unknown and the optimization cannot be applied—the two equivalent statements may exhibit inconsistent behavior, with one executing successfully and returning a result while the other triggers an error. This inconsist",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 10535,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "sqlancer",
+      "technique": null,
+      "sentence": "com/sqlancer/sqlancer/blob/main/src/sqlancer/common/query/ExpectedErrors.",
+      "context_before": "and the optimization cannot be applied—the two equivalent statements may exhibit inconsistent behavior, with one executing successfully and returning a result while the other triggers an error. This inconsistency requires us to determine whether the correctly executed statement skipped the erroneous part or whether the erroneous part was executed incorrectly, producing an incorrect result—that is, a logic bug. Treating inconsistent behavior in equivalent statements as a logic bug in the correctly executed statement without careful analysis can lead to a lot of false positives. 1https://github.",
+      "context_after": "java 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026. 176:4 Chi Zhang et al. To mitigate false positives arising from the skipping of erroneous parts in statements due to DBMS optimizations—leading to behavior inconsistencies—we introduce an error validation mechanism. Specifically, we traverse and force the evaluation of each expression in the statement without error. If an expression produces the same error as in the other statement, this indicates that the erroneous part was skipped, and the error is considered expected. Otherwise, the correctly executed statement may have exec",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 11737,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[ 25]",
+      "technique": "norec",
+      "sentence": "Logic bugs [ 25] are among the most common types of bugs in DBMSs and cause the system to produce incorrect results silently for certain statements.",
+      "context_before": "y statements, such as short-circuit evaluation and constant folding [ 24], cannot be performed. Prepared statements may use distinct mechanisms for query plan generation [ 12] and selection [ 6]. Additionally, during execution, variables must be parsed in the EXECUTE statement, and SQL injection prevention must be enforced. Consequently, prepared statements may follow execution paths that differ from those of ordinary statements throughout parsing, compilation, optimization, and execution. DBMS logic bugs.DBMSs, due to their complex implementations and optimizations, are prone to various bugs.",
+      "context_after": "Because their occurrence does not produce obvious system behavior, they are difficult for developers and users to detect. These bugs are critical for DBMSs, as they can occur in statements such as SELECT, leading users to obtain incorrect results, or in statements such as INSERT ,DELETE, and UPDATE, which may construct incorrect database states and consequently affect all subsequent queries. Furthermore, such bugs may cause inconsistencies between prepared and ordinary statements (e.g., the ordinary statement produces the correct result, whereas the prepared one does not). This can mislead de",
+      "section": "2 Background",
+      "page": 4,
+      "char_offset": 15066,
+      "cited_reference": {
+        "number": 25,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We reused the statement generators from SQLancer [ 27] for statement generation, which are manually written and rule-based.",
+      "context_before": "yFROM'DELETE FROM t0 WHERE t0.c0 NOTLIKE(CASEWHEN false THEN? ELSE t0.c0 END)'; EXECUTEprepare_queryUSING@c; SELECT*FROMt0; -- empty resultὁB 4 Evaluation We implemented EPSC and evaluated its effectiveness through the following research questions: Q1:How effective is EPSC at discovering new bugs? Q2:How effective and efficient is EPSC compared with other test oracles? Q3: How does the error validation component of EPSC affect false positives and testing performance? Implementation.EPSC is implemented in 40K lines of Java code, providing a framework for testing DBMSs using prepared statements.",
+      "context_after": "In addition, we extend the generator to support richer JOIN constructions, including self-joins and joins with subqueries. We further add support for stored procedures in PostgreSQL, MySQL, and MariaDB, as well as prepared transactions in PostgreSQL. For generating prepared statements, we first filter out statements that can be executed as prepared statements. We then traverse their ASTs, randomly select literals, and replace them with placeholders supported by the target DBMS. Corresponding EXECUTE statements are also generated to run the prepared statements. Although DBMSs support various t",
+      "section": "4 Evaluation",
+      "page": 11,
+      "char_offset": 40563,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[ 2,25–27,31,35]",
+      "technique": "dqp",
+      "sentence": "Moreover, all these DBMSs have been tested by existing works on detecting logic bugs in DBMSs [ 2,25–27,31,35].",
+      "context_before": "Methodology.We evaluated the effectiveness of EPSC by testing seven widely-used and mature DBMSs, including MySQL [ 22], MariaDB [ 19], TiDB [ 14], PostgreSQL [ 21], CockroachDB [ 32], SQLite3 [ 11], and DuckDB [ 23]. We selected these DBMSs for two main reasons. First, they are highly popular and mature, and have been extensively tested by existing methods. Among these DBMSs, two rank in the top 10 (i,e,. MySQL and PostgreSQL) and four in the top 20 (i.e., SQLite3 and MariaDB) in terms of overall DBMS popularity according to the DB-Engines Ranking4, with all seven ranking within the top 100.",
+      "context_after": "The high popularity and extensive testing history of these DBMSs indicate a lower likelihood of undiscovered bugs, which demonstrates the effectiveness of EPSC. Second, they represent different types of DBMSs, demonstrating that EPSC can discover bugs across diverse DBMS architectures. MySQL, MariaDB, and PostgreSQL represent traditional client–server relational databases, CockroachDB and TiDB represent distributed relational databases, while SQLite3 and DuckDB represent embedded relational databases. We selected the latest release version of each DBMS as the testing target, since newer relea",
+      "section": "4.1 Effectiveness",
+      "page": 12,
+      "char_offset": 43973,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. doi:10.1145/3654991",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "prints the DOI of the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[25, 26, 35]",
+      "technique": "norec",
+      "sentence": "This is a standard practice for evaluating the effectiveness of logic bug detection methods [25, 26, 35].",
+      "context_before": "rchitectures. MySQL, MariaDB, and PostgreSQL represent traditional client–server relational databases, CockroachDB and TiDB represent distributed relational databases, while SQLite3 and DuckDB represent embedded relational databases. We selected the latest release version of each DBMS as the testing target, since newer releases tend to be more stable and robust. In particular, we tested MySQL 9.4.0, MariaDB 12.2.0, TiDB 8.5.3, PostgreSQL 18.0, CockroachDB 25.4.0, SQLite3 3.50.4, and DuckDB 1.4.0. We intermittently ran EPSC over a period of three months, during which we were also developing it.",
+      "context_after": "Results.Table 1 summarizes the number and status of bugs found by EPSC. In total, EPSC discovered 52 previously unknown bugs, including 28 logic bugs and 24 unexpected errors. Among these 52 bugs, 50 have been confirmed, of which 13 have been fixed, and 2 remain under analysis by the developers. These results are highly encouraging, given that the tested DBMSs have already been extensively tested by multiple prior works. Bug root causes.Among the 52 bugs detected by EPSC, 40 are found in prepared statements. Including 20 prepared statement logic bugs (8 type-related, 6 involving special chara",
+      "section": "4.1 Effectiveness",
+      "page": 12,
+      "char_offset": 44959,
+      "cited_reference": {
+        "number": 25,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "EPSC: Testing Database Management Systems via Equivalent Prepared Statement Construction 176:17 of prepared INSERT statements, it is beyond the reach of NoREC [ 25] and TLP [ 26], which target SELECT queries, and also DQE [ 31], whose detection of INSERT -related logic bugs is limited to WHERE-clause conditions.",
+      "context_before": "dation reveals that the two test cases return different results. We reported the issue to the PostgreSQL developers, who responded that in the ordinary statement, the division by zero occurs before c0is incremented, so its first increment takes place during the second insertion, giving it a value of 1. In the prepared statement, the underlying sequence for c0advances before the error and is not rolled back, causing a second increment on the second insertion; thus, c0takes the value 2. As this logic bug arises from the error handling 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026.",
+      "context_after": "Listing 13. A logic bug found in SQLite3 caused by inconsistent encoding. -- The test case executed in the original database PRAGMA encoding ='UTF-16'; CREATE TABLEt0 (c0 TEXTCHECK(c0 ISTRUE)); INSERT INTOt0(c0)VALUES(x'310a'); -- CHECK constraint failed: c0 IS TRUE✔ -- The test case executed in the reference database PRAGMA encoding ='UTF-16'; CREATE TABLEt0 (c0 TEXTCHECK(c0 ISTRUE)); .parameterset$1 x'310a' INSERT INTOt0(c0)VALUES($1); --ὁB SELECTc0, (c0 ISTRUE)FROMt0; -- { |0} The bugs discussed above were all discovered by observing inconsistencies in SELECT query results (i.e., detected",
+      "section": "4.2 Selected Bugs",
+      "page": 16,
+      "char_offset": 58050,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In SQLancer, the expression generator for DuckDB does not strictly validate the return types of generated expressions.",
+      "context_before": "SELECTLENGTH(?)FROMt0GROUP BY NULL; EXECUTEprepare_query(NULL); -- {NULL}✔ Unexpected syntax or semantic errors.Listing 14 shows an unexpected semantic error we found in DuckDB. In the ordinary statement, this error is triggered when a NULL value is used both in the GROUP BY clause and as the argument of the length function. The NULL value is implicitly 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026. 176:18 Chi Zhang et al. assigned an INTEGER type, which causes the error. In contrast, in the prepared statement, the length function correctly returns NULL when its argument is NULL.",
+      "context_after": "As a result, it can easily produce a non-string argument for the length function during random expression generation. Consequently, such errors are treated as expected semantic errors and thus ignored. In contrast, our approach revealed this issue by detecting the inconsistent behavior between the ordinary and prepared statements. Listing 15. An unexpected error occurs in TiDB when a bound parameter is used as the grouping key. CREATE TABLEt0(c0DECIMAL); INSERTIGNOREINTOt0VALUES(1); -- The ordinary statement SELECT2FROMt0WHEREt0.c0GROUP BY1; -- {2}✔ -- The prepared statement SET@a = 2; PREPAR",
+      "section": "4.2 Selected Bugs",
+      "page": 18,
+      "char_offset": 61238,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For our evaluation, we chose NoREC [ 25], TLP [ 26], DQE [ 31], and CODDTest [ 35] as baseline approaches.",
+      "context_before": "3.30.0 for 24 hours, gathering data on the number of unique bugs triggered by each tool, bugs detected exclusively by a given tool, code coverage, and the number of unique query plans. We selected an older version of SQLite3 because multiple bugs had been reported and fixed in this version by previous works. This allows us to determine whether two bug reports correspond to the same underlying bug by checking their respective fixing commits. In this experiment, we identified the fixing commit of each bug using bisect, a binary search tool supported by multiple source control systems. Baselines.",
+      "context_after": "We selected these methods for several reasons. First, they have each successfully uncovered multiple bugs in widely used DBMSs, demonstrating their practical effectiveness. Second, they represent different directions in constructing testing oracles for logic bug detection. Specifically, NoREC and DQE represent the class of approaches that detect logic bugs by leveraging clause equivalence; TLP represents the set-relation–based oracle design. A similar work, Pinolo [ 13], also adopts this principle, but it does not support testing SQLite3, preventing a direct comparison. CODDTest represents th",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 18,
+      "char_offset": 63584,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Specifically, NoREC and DQE represent the class of approaches that detect logic bugs by leveraging clause equivalence; TLP represents the set-relation–based oracle design.",
+      "context_before": "same underlying bug by checking their respective fixing commits. In this experiment, we identified the fixing commit of each bug using bisect, a binary search tool supported by multiple source control systems. Baselines.For our evaluation, we chose NoREC [ 25], TLP [ 26], DQE [ 31], and CODDTest [ 35] as baseline approaches. We selected these methods for several reasons. First, they have each successfully uncovered multiple bugs in widely used DBMSs, demonstrating their practical effectiveness. Second, they represent different directions in constructing testing oracles for logic bug detection.",
+      "context_after": "A similar work, Pinolo [ 13], also adopts this principle, but it does not support testing SQLite3, preventing a direct comparison. CODDTest represents the class of approaches that construct oracles by performing expression-equivalence transformations; a similar work, EET [ 15], adopts the same principle but cannot be run continuously—it terminates immediately after detecting a bug—making it unsuitable for long-running experiments. Another direction of oracle construction involves changing 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026. EPSC: Testing Database Management Systems v",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 18,
+      "char_offset": 63964,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "CODDTest represents the class of approaches that construct oracles by performing expression-equivalence transformations; a similar work, EET [ 15], adopts the same principle but cannot be run continuously—it terminates immediately after detecting a bug—making it unsuitable for long-running experiments.",
+      "context_before": "as baseline approaches. We selected these methods for several reasons. First, they have each successfully uncovered multiple bugs in widely used DBMSs, demonstrating their practical effectiveness. Second, they represent different directions in constructing testing oracles for logic bug detection. Specifically, NoREC and DQE represent the class of approaches that detect logic bugs by leveraging clause equivalence; TLP represents the set-relation–based oracle design. A similar work, Pinolo [ 13], also adopts this principle, but it does not support testing SQLite3, preventing a direct comparison.",
+      "context_after": "Another direction of oracle construction involves changing 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026. EPSC: Testing Database Management Systems via Equivalent Prepared Statement Construction 176:19 configurations or optimizer hints to alter query execution paths for bug detection, as exemplified by Mozi [ 17], DQP [ 2], and TQS [ 33]. However, these tools do not support SQLite3 testing, and most of them mainly target join-related bugs, whereas EPSC is not limited to any specific syntactic feature. Finally, all five tools directly reuse SQLancer’s test-case generation code,",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 18,
+      "char_offset": 64267,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "DQP",
+      "technique": "dqp",
+      "sentence": "EPSC: Testing Database Management Systems via Equivalent Prepared Statement Construction 176:19 configurations or optimizer hints to alter query execution paths for bug detection, as exemplified by Mozi [ 17], DQP [ 2], and TQS [ 33].",
+      "context_before": "esents the set-relation–based oracle design. A similar work, Pinolo [ 13], also adopts this principle, but it does not support testing SQLite3, preventing a direct comparison. CODDTest represents the class of approaches that construct oracles by performing expression-equivalence transformations; a similar work, EET [ 15], adopts the same principle but cannot be run continuously—it terminates immediately after detecting a bug—making it unsuitable for long-running experiments. Another direction of oracle construction involves changing 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026.",
+      "context_after": "However, these tools do not support SQLite3 testing, and most of them mainly target join-related bugs, whereas EPSC is not limited to any specific syntactic feature. Finally, all five tools directly reuse SQLancer’s test-case generation code, ensuring that the comparison of bug-detection capabilities focuses solely on the effectiveness of their testing oracles, while eliminating inconsistencies arising from differences in test-case generation implementations. Results.The comparison results are shown in Table 2. Overall, although EPSC did not find the largest total number of unique bugs—its co",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 18,
+      "char_offset": 64692,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Finally, all five tools directly reuse SQLancer’s test-case generation code, ensuring that the comparison of bug-detection capabilities focuses solely on the effectiveness of their testing oracles, while eliminating inconsistencies arising from differences in test-case generation implementations.",
+      "context_before": "ately after detecting a bug—making it unsuitable for long-running experiments. Another direction of oracle construction involves changing 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026. EPSC: Testing Database Management Systems via Equivalent Prepared Statement Construction 176:19 configurations or optimizer hints to alter query execution paths for bug detection, as exemplified by Mozi [ 17], DQP [ 2], and TQS [ 33]. However, these tools do not support SQLite3 testing, and most of them mainly target join-related bugs, whereas EPSC is not limited to any specific syntactic feature.",
+      "context_after": "Results.The comparison results are shown in Table 2. Overall, although EPSC did not find the largest total number of unique bugs—its count is smaller than that of NoREC, TLP, and CODDTest, but higher than DQE—it ranks first in terms of the number of bugs that were exclusively detected by a single tool, tying with TLP and outperforming NoREC, CODDTest, and DQE. Figure 2 presents the overlap among the bugs identified by different testing methods. To avoid false negatives caused by their lack of support for prepared statements, we applied the baseline test oracles to the buginducing test cases o",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 19,
+      "char_offset": 65093,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Overall, although EPSC did not find the largest total number of unique bugs—its count is smaller than that of NoREC, TLP, and CODDTest, but higher than DQE—it ranks first in terms of the number of bugs that were exclusively detected by a single tool, tying with TLP and outperforming NoREC, CODDTest, and DQE.",
+      "context_before": "ion paths for bug detection, as exemplified by Mozi [ 17], DQP [ 2], and TQS [ 33]. However, these tools do not support SQLite3 testing, and most of them mainly target join-related bugs, whereas EPSC is not limited to any specific syntactic feature. Finally, all five tools directly reuse SQLancer’s test-case generation code, ensuring that the comparison of bug-detection capabilities focuses solely on the effectiveness of their testing oracles, while eliminating inconsistencies arising from differences in test-case generation implementations. Results.The comparison results are shown in Table 2.",
+      "context_after": "Figure 2 presents the overlap among the bugs identified by different testing methods. To avoid false negatives caused by their lack of support for prepared statements, we applied the baseline test oracles to the buginducing test cases of the 6 logic bugs only found by EPSC to examine whether they could detect the same inconsistencies. The results show that one of the six bugs can be detected by NoREC, TLP, and CODDTest, while the remaining five bugs cannot be detected by any of the baseline oracles. We then checked the fix commits corresponding to these five bugs and found that four of them a",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 19,
+      "char_offset": 65444,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "The results show that one of the six bugs can be detected by NoREC, TLP, and CODDTest, while the remaining five bugs cannot be detected by any of the baseline oracles.",
+      "context_before": "t total number of unique bugs—its count is smaller than that of NoREC, TLP, and CODDTest, but higher than DQE—it ranks first in terms of the number of bugs that were exclusively detected by a single tool, tying with TLP and outperforming NoREC, CODDTest, and DQE. Figure 2 presents the overlap among the bugs identified by different testing methods. To avoid false negatives caused by their lack of support for prepared statements, we applied the baseline test oracles to the buginducing test cases of the 6 logic bugs only found by EPSC to examine whether they could detect the same inconsistencies.",
+      "context_after": "We then checked the fix commits corresponding to these five bugs and found that four of them are related to prepared statements. This result highlights the unique value of EPSC, demonstrating its effectiveness in discovering logic bugs that cannot be detected by existing methods. Table 2. Effectiveness and efficiency of EPSC compared to state-of-the-art test oracles Test oracles # of unique bugs # of oracle-exclusive bugs Code coverage # of unique query plans NoREC 23 4 63.36% 172,812 TLP 24 6 63.86% 137,755 DQE 3 0 45.07% 475 CODDTest 21 2 62.33% 2,577,542 EPSC 8 6 69.06% 2,709,700 Fig. 2. O",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 19,
+      "char_offset": 66091,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Effectiveness and efficiency of EPSC compared to state-of-the-art test oracles Test oracles # of unique bugs # of oracle-exclusive bugs Code coverage # of unique query plans NoREC 23 4 63.",
+      "context_before": "st oracles to the buginducing test cases of the 6 logic bugs only found by EPSC to examine whether they could detect the same inconsistencies. The results show that one of the six bugs can be detected by NoREC, TLP, and CODDTest, while the remaining five bugs cannot be detected by any of the baseline oracles. We then checked the fix commits corresponding to these five bugs and found that four of them are related to prepared statements. This result highlights the unique value of EPSC, demonstrating its effectiveness in discovering logic bugs that cannot be detected by existing methods. Table 2.",
+      "context_after": "36% 172,812 TLP 24 6 63.86% 137,755 DQE 3 0 45.07% 475 CODDTest 21 2 62.33% 2,577,542 EPSC 8 6 69.06% 2,709,700 Fig. 2. Overlap among bugs detected by different methods. Under identical statement generation settings, EPSC achieved higher code coverage across all tested DBMSs compared with other methods, improving coverage by up to 5.2% over the next best approach. This improvement primarily stems from its support for executing prepared statements, demonstrating that our approach effectively enables equivalent statements to exercise distinct execution paths within the DBMS. Furthermore, EPSC tr",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 19,
+      "char_offset": 66549,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "36% 172,812 TLP 24 6 63.",
+      "context_before": "be detected by NoREC, TLP, and CODDTest, while the remaining five bugs cannot be detected by any of the baseline oracles. We then checked the fix commits corresponding to these five bugs and found that four of them are related to prepared statements. This result highlights the unique value of EPSC, demonstrating its effectiveness in discovering logic bugs that cannot be detected by existing methods. Table 2. Effectiveness and efficiency of EPSC compared to state-of-the-art test oracles Test oracles # of unique bugs # of oracle-exclusive bugs Code coverage # of unique query plans NoREC 23 4 63.",
+      "context_after": "86% 137,755 DQE 3 0 45.07% 475 CODDTest 21 2 62.33% 2,577,542 EPSC 8 6 69.06% 2,709,700 Fig. 2. Overlap among bugs detected by different methods. Under identical statement generation settings, EPSC achieved higher code coverage across all tested DBMSs compared with other methods, improving coverage by up to 5.2% over the next best approach. This improvement primarily stems from its support for executing prepared statements, demonstrating that our approach effectively enables equivalent statements to exercise distinct execution paths within the DBMS. Furthermore, EPSC triggered the largest numb",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 19,
+      "char_offset": 66738,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "07% 475 CODDTest 21 2 62.",
+      "context_before": "the remaining five bugs cannot be detected by any of the baseline oracles. We then checked the fix commits corresponding to these five bugs and found that four of them are related to prepared statements. This result highlights the unique value of EPSC, demonstrating its effectiveness in discovering logic bugs that cannot be detected by existing methods. Table 2. Effectiveness and efficiency of EPSC compared to state-of-the-art test oracles Test oracles # of unique bugs # of oracle-exclusive bugs Code coverage # of unique query plans NoREC 23 4 63.36% 172,812 TLP 24 6 63.86% 137,755 DQE 3 0 45.",
+      "context_after": "33% 2,577,542 EPSC 8 6 69.06% 2,709,700 Fig. 2. Overlap among bugs detected by different methods. Under identical statement generation settings, EPSC achieved higher code coverage across all tested DBMSs compared with other methods, improving coverage by up to 5.2% over the next best approach. This improvement primarily stems from its support for executing prepared statements, demonstrating that our approach effectively enables equivalent statements to exercise distinct execution paths within the DBMS. Furthermore, EPSC triggered the largest number of unique 4, No. 3 (SIGMOD), Article 176. Pu",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 19,
+      "char_offset": 66785,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "This is largely attributed to its support for subqueries—similar to CODDTest—which increases query complexity and, consequently, plan diversity compared with NoREC and TLP.",
+      "context_before": "ugs detected by different methods. Under identical statement generation settings, EPSC achieved higher code coverage across all tested DBMSs compared with other methods, improving coverage by up to 5.2% over the next best approach. This improvement primarily stems from its support for executing prepared statements, demonstrating that our approach effectively enables equivalent statements to exercise distinct execution paths within the DBMS. Furthermore, EPSC triggered the largest number of unique 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026. 176:20 Chi Zhang et al. query plans.",
+      "context_after": "In addition, prepared statements were also able to trigger query plans that ordinary statements could not. During the 24-hour experiment, 2,319 unique query plans were observed exclusively from prepared statements. In summary, EPSC is capable of exploring previously uncovered code regions and execution paths, and of revealing logic bugs that existing methods fail to detect. 4.4 Ablation Study In our approach, we propose an error validation strategy to determine whether an logic bug occurs when inconsistent behavior is observed between equivalent ordinary statements and prepared statements. Th",
+      "section": "4.3 Compared with Other Approaches",
+      "page": 20,
+      "char_offset": 67473,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "coddtest",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This difference arises because we directly adopt SQLancer’s query generators, which are implemented separately for each DBMS and therefore produce SQL statements with different proportions of erroneous.",
+      "context_before": "repared Statement Construction 176:21 compared with EPSC is relatively small. The generator of MariaDB is relatively simpler, and fewer abnormal test cases were produced. As a result, EPSC-generated fewer alarms, and the performance impact of the aforementioned process was smaller than the overhead introduced by error validation. In SQLite3 and DuckDB, connecting to a database instance is much faster, which makes EPSC-less affected by this issue. In this evaluation, although MySQL, MariaDB, and TiDB support largely similar SQL syntax, EPSC and EPSC-exhibit different results across these DBMSs.",
+      "context_after": "Nevertheless, the overall trends of the results remain consistent. To verify that all filtered reports by EPSC are indeed false positives, we deduplicated all error messages reported by EPSC-and conducted a manual analysis. This analysis revealed one genuine unexpected error. The error is skipped in the ordinary statement due to a false WHERE condition but is triggered in the prepared statement. During error validation, the expression is forcibly executed and triggers the same error, causing EPSC to classify it as a false positive. This error lies outside the detection scope of EPSC, as it is",
+      "section": "4.4 Ablation Study",
+      "page": 21,
+      "char_offset": 72201,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [ 25] leverages the equivalence between expressions in the SELECT and WHERE clauses, assuming that an expression should evaluate to the same result in both contexts.",
+      "context_before": "ferent DBMSs or across versions of the same system. However, differential testing faces challenges in detecting logic bugs. On the one hand, SQL dialects differ across DBMSs, and differential testing has limited ability to handle system-specific features. On the other hand, logic bugs may arise within different versions of the same DBMS, making it difficult to uncover them by simply comparing results across versions. Metamorphic testing [ 7] generates follow-up test cases from existing ones, expecting the outcomes of the original and follow-up cases to satisfy predefined metamorphic relations.",
+      "context_after": "DQE [ 31] extends NoREC by asserting that the same expression should evaluate consistently in the WHERE clause of SELECT, UPDATE, and DELETE statements. TLP [ 26] is based on the fact that an expression can evaluate to TRUE ,FALSE, or NULL. It constructs three queries using p,NOT p, and p IS NULL as predicates, and asserts that their union should equal the complete set evaluated by the expression. Pinolo [ 13] similarly leverages set relationships to detect logic bugs by constructing looser or stricter predicates, such that the transformed query’s result should either contain or be a subset of",
+      "section": "6 Related Work",
+      "page": 22,
+      "char_offset": 77225,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M24",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "DQE [ 31] extends NoREC by asserting that the same expression should evaluate consistently in the WHERE clause of SELECT, UPDATE, and DELETE statements.",
+      "context_before": "Ss, and differential testing has limited ability to handle system-specific features. On the other hand, logic bugs may arise within different versions of the same DBMS, making it difficult to uncover them by simply comparing results across versions. Metamorphic testing [ 7] generates follow-up test cases from existing ones, expecting the outcomes of the original and follow-up cases to satisfy predefined metamorphic relations. NoREC [ 25] leverages the equivalence between expressions in the SELECT and WHERE clauses, assuming that an expression should evaluate to the same result in both contexts.",
+      "context_after": "TLP [ 26] is based on the fact that an expression can evaluate to TRUE ,FALSE, or NULL. It constructs three queries using p,NOT p, and p IS NULL as predicates, and asserts that their union should equal the complete set evaluated by the expression. Pinolo [ 13] similarly leverages set relationships to detect logic bugs by constructing looser or stricter predicates, such that the transformed query’s result should either contain or be a subset of the original query’s result. TQS [ 33] and Differential Query Plans (DQP) [ 2] detect logic bugs in join optimization by providing hints or modifying DB",
+      "section": "6 Related Work",
+      "page": 22,
+      "char_offset": 77396,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [ 26] is based on the fact that an expression can evaluate to TRUE ,FALSE, or NULL.",
+      "context_before": "the same DBMS, making it difficult to uncover them by simply comparing results across versions. Metamorphic testing [ 7] generates follow-up test cases from existing ones, expecting the outcomes of the original and follow-up cases to satisfy predefined metamorphic relations. NoREC [ 25] leverages the equivalence between expressions in the SELECT and WHERE clauses, assuming that an expression should evaluate to the same result in both contexts. DQE [ 31] extends NoREC by asserting that the same expression should evaluate consistently in the WHERE clause of SELECT, UPDATE, and DELETE statements.",
+      "context_after": "It constructs three queries using p,NOT p, and p IS NULL as predicates, and asserts that their union should equal the complete set evaluated by the expression. Pinolo [ 13] similarly leverages set relationships to detect logic bugs by constructing looser or stricter predicates, such that the transformed query’s result should either contain or be a subset of the original query’s result. TQS [ 33] and Differential Query Plans (DQP) [ 2] detect logic bugs in join optimization by providing hints or modifying DBMS configurations to induce different query plans during query execution. Both EET [ 15",
+      "section": "6 Related Work",
+      "page": 22,
+      "char_offset": 77549,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "Differential Query Plans",
+      "technique": "dqp",
+      "sentence": "TQS [ 33] and Differential Query Plans (DQP) [ 2] detect logic bugs in join optimization by providing hints or modifying DBMS configurations to induce different query plans during query execution.",
+      "context_before": "sserting that the same expression should evaluate consistently in the WHERE clause of SELECT, UPDATE, and DELETE statements. TLP [ 26] is based on the fact that an expression can evaluate to TRUE ,FALSE, or NULL. It constructs three queries using p,NOT p, and p IS NULL as predicates, and asserts that their union should equal the complete set evaluated by the expression. Pinolo [ 13] similarly leverages set relationships to detect logic bugs by constructing looser or stricter predicates, such that the transformed query’s result should either contain or be a subset of the original query’s result.",
+      "context_after": "Both EET [ 15] and CODDTest [ 35] transform expressions to construct equivalent queries. EET generates more complex but equivalent expressions by introducing subexpressions that always evaluate to TRUE or FALSE, while CODDTest creates simpler expressions by applying constant folding and propagation to replace parts of the original expression with equivalent constants. Essentially, EPSC can also be regarded as a form of metamorphic testing. It constructs prepared statements that alter the parsing, compilation, optimization, and execution paths of their ordinary counterparts, while expecting bot",
+      "section": "6 Related Work",
+      "page": 22,
+      "char_offset": 78025,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "Both EET [ 15] and CODDTest [ 35] transform expressions to construct equivalent queries.",
+      "context_before": "FALSE, or NULL. It constructs three queries using p,NOT p, and p IS NULL as predicates, and asserts that their union should equal the complete set evaluated by the expression. Pinolo [ 13] similarly leverages set relationships to detect logic bugs by constructing looser or stricter predicates, such that the transformed query’s result should either contain or be a subset of the original query’s result. TQS [ 33] and Differential Query Plans (DQP) [ 2] detect logic bugs in join optimization by providing hints or modifying DBMS configurations to induce different query plans during query execution.",
+      "context_after": "EET generates more complex but equivalent expressions by introducing subexpressions that always evaluate to TRUE or FALSE, while CODDTest creates simpler expressions by applying constant folding and propagation to replace parts of the original expression with equivalent constants. Essentially, EPSC can also be regarded as a form of metamorphic testing. It constructs prepared statements that alter the parsing, compilation, optimization, and execution paths of their ordinary counterparts, while expecting both forms to produce identical results. Compared with differential testing approaches, EPSC",
+      "section": "6 Related Work",
+      "page": 22,
+      "char_offset": 78222,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "technique",
+      "surface": "CODDTest",
+      "technique": "coddtest",
+      "sentence": "EET generates more complex but equivalent expressions by introducing subexpressions that always evaluate to TRUE or FALSE, while CODDTest creates simpler expressions by applying constant folding and propagation to replace parts of the original expression with equivalent constants.",
+      "context_before": "d asserts that their union should equal the complete set evaluated by the expression. Pinolo [ 13] similarly leverages set relationships to detect logic bugs by constructing looser or stricter predicates, such that the transformed query’s result should either contain or be a subset of the original query’s result. TQS [ 33] and Differential Query Plans (DQP) [ 2] detect logic bugs in join optimization by providing hints or modifying DBMS configurations to induce different query plans during query execution. Both EET [ 15] and CODDTest [ 35] transform expressions to construct equivalent queries.",
+      "context_after": "Essentially, EPSC can also be regarded as a form of metamorphic testing. It constructs prepared statements that alter the parsing, compilation, optimization, and execution paths of their ordinary counterparts, while expecting both forms to produce identical results. Compared with differential testing approaches, EPSC is capable of detecting logic bugs that reside in DBMS-specific language features as well as those hidden across multiple versions of the same system. Compared with other metamorphic testing approaches, EPSC can be regarded as orthogonal to existing methods, as it effectively det",
+      "section": "6 Related Work",
+      "page": 22,
+      "char_offset": 78311,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Generation-based approaches produce syntactically and semantically valid test cases based on the grammar and features supported by the target DBMSs, with SQLSmith [ 28] and SQLancer [ 27] being representative examples.",
+      "context_before": "e exposed by existing approaches. 4, No. 3 (SIGMOD), Article 176. Publication date: June 2026. EPSC: Testing Database Management Systems via Equivalent Prepared Statement Construction 176:23 DBMS test case generation.Numerous studies have focused on improving test case generation to enable fuzzing to cover more code regions and detect bugs more efficiently. These techniques can be considered complementary to EPSC, as they help improve the efficiency and effectiveness of bug triggering. Test case generation methods can be broadly categorized into generation-based and mutation-based approaches.",
+      "context_after": "Query Plan Guidance (QPG) [ 1] was proposed to mitigate redundancy in randomly generated test cases and improve resource efficiency. It evaluates the sufficiency of testing for a given database state by analyzing the query plans triggered by executed queries. Once no additional queries can generate new query plans, the database state is mutated to further enhance testing coverage. QAGen [ 3] takes a query along with a set of constraints and constructs a database that satisfies the expected query results by integrating conventional query processing techniques with symbolic execution. Mutation-",
+      "section": "6 Related Work",
+      "page": 23,
+      "char_offset": 80703,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "technique",
+      "surface": "Query Plan Guidance",
+      "technique": "qpg",
+      "sentence": "Query Plan Guidance (QPG) [ 1] was proposed to mitigate redundancy in randomly generated test cases and improve resource efficiency.",
+      "context_before": "umerous studies have focused on improving test case generation to enable fuzzing to cover more code regions and detect bugs more efficiently. These techniques can be considered complementary to EPSC, as they help improve the efficiency and effectiveness of bug triggering. Test case generation methods can be broadly categorized into generation-based and mutation-based approaches. Generation-based approaches produce syntactically and semantically valid test cases based on the grammar and features supported by the target DBMSs, with SQLSmith [ 28] and SQLancer [ 27] being representative examples.",
+      "context_after": "It evaluates the sufficiency of testing for a given database state by analyzing the query plans triggered by executed queries. Once no additional queries can generate new query plans, the database state is mutated to further enhance testing coverage. QAGen [ 3] takes a query along with a set of constraints and constructs a database that satisfies the expected query results by integrating conventional query processing techniques with symbolic execution. Mutation-based methods use existing test cases as seeds and derive new test cases through systematic transformations of the originals. SQLRigh",
+      "section": "6 Related Work",
+      "page": 23,
+      "char_offset": 80922,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M11"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:13:36Z",
+    "is_model_written": true,
+    "summary": "EPSC tests DBMSs by exploiting the fact that a statement can be written both as an ordinary SQL statement and as a prepared statement with literals bound as parameters. The two forms should behave identically, so any difference between them indicates a bug. The authors note that prepared statements, though widely used in production for performance and against SQL injection, have received far less testing attention. Applied to seven mature DBMSs, EPSC found 49 unique bugs, 31 confirmed and 10 fixed.",
+    "narrative": "EPSC reuses SQLancer's statement generators to produce the statements it then rewrites into prepared form, and takes four oracles as baselines, two of which -- NoREC and TLP -- are SQLancer's, alongside CODDTest, which also came out of the project.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "motivation",
+      "M4": "reuse_component",
+      "M5": "definition",
+      "M6": "reuse_component",
+      "M7": "background",
+      "M8": "background",
+      "M9": "definition",
+      "M10": "incidental",
+      "M11": "baseline",
+      "M12": "definition",
+      "M13": "definition",
+      "M14": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M6"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M6",
+            "sentence": "We reused the statement generators from SQLancer [ 27] for statement generation, which are manually written and rule-based.",
+            "section": "4 Evaluation",
+            "page": 11
+          }
+        ],
+        "reasoning": "M6 states the statement generators were reused from SQLancer for statement generation; the oracle itself is the paper's own.",
+        "reuse_kind": "generator"
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Equivalent prepared statement construction is a new metamorphic relation rather than a generalisation of NoREC or TLP, which the paper classes as clause-equivalence and set-relation approaches respectively."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M11",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M11",
+            "sentence": "For our evaluation, we chose NoREC [ 25], TLP [ 26], DQE [ 31], and CODDTest [ 35] as baseline approaches.",
+            "section": "4.3 Compared with Other Approaches",
+            "page": 18
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "Specifically, NoREC and DQE represent the class of approaches that detect logic bugs by leveraging clause equivalence; TLP represents the set-relation–based oracle design.",
+            "section": "4.3 Compared with Other Approaches",
+            "page": 18
+          }
+        ],
+        "reasoning": "M11 names NoREC, TLP, DQE and CODDTest as the baselines, and M12 explains which class each represents.",
+        "techniques": [
+          "norec",
+          "tlp",
+          "coddtest"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The baselines are described by the class of equivalence they exploit rather than as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3802061.json b/_data/papers/paper_doi_10_1145_3802061.json
new file mode 100644
index 0000000..fb8c839
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3802061.json
@@ -0,0 +1,857 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:15Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3802061",
+    "title": "Finding Missed Optimizations in DBMSs through Unbalanced Short-circuit Query Construction",
+    "authors": [
+      "Jinhui Lai",
+      "Chi Zhang",
+      "Jie Liang",
+      "Zihao Zeng",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Chijin Zhou",
+      "Shuai Ma",
+      "Yu Jiang",
+      "Zichen Xu"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Management of Data",
+    "doi": "10.1145/3802061",
+    "arxiv_id": null,
+    "s2_paper_id": "96bb90109a01a144448d9bf9b7b58f3e4c440eff",
+    "url": "https://doi.org/10.1145/3802061",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3802061",
+      "retrieved_at": "2026-09-09T01:29:15Z",
+      "chars": 90549,
+      "content_sha256": "sha256:325cd5949963bc3e702fc0d2d85d8f0362afc33fc090aa8f68cb717e53c1a4b2"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 24,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3788
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 11755
+      },
+      {
+        "number": "3",
+        "title": "Design of SCor",
+        "start": 15158
+      },
+      {
+        "number": "3.1",
+        "title": "Representative Short-Circuit Patterns",
+        "start": 17421
+      },
+      {
+        "number": "3.2",
+        "title": "Skeleton Construction and Transformation",
+        "start": 26911
+      },
+      {
+        "number": "3.3",
+        "title": "Query Generation and Mutation",
+        "start": 32607
+      },
+      {
+        "number": "3.4",
+        "title": "Performance Checking",
+        "start": 37390
+      },
+      {
+        "number": "4",
+        "title": "Implementation",
+        "start": 41641
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 44038
+      },
+      {
+        "number": "5.1",
+        "title": "Testing Setup",
+        "start": 44581
+      },
+      {
+        "number": "5.2",
+        "title": "Detection Capability",
+        "start": 47749
+      },
+      {
+        "number": "5.3",
+        "title": "Pattern Analysis",
+        "start": 55630
+      },
+      {
+        "number": "5.4",
+        "title": "Comparative Evaluation",
+        "start": 66556
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 71077
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 74124
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 78214
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 133, 13 pages. doi:10.1145/3597503.3639076",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 2,
+      "text": "Stefan Brass and Christian Goldberg. 2006. Semantic errors in SQL queries: A quite complete list.Journal of Systems and Software79, 5 (2006), 630–644. doi:10.1016/j.jss.2005.06.028 Quality Software.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Luis Gravano. 2002. Top-k selection queries over relational databases: Mapping strategies and performance evaluation.ACM Transactions on Database Systems (TODS)27, 2 (2002), 153–187.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "S. Ceri and G. Gottlob. 1985. Translating SQL Into Relational Algebra: Optimization, Semantics, and Equivalence of SQL Queries.IEEE Transactions on Software EngineeringSE-11, 4 (1985), 324–345. doi:10.1109/TSE.1985.232223",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Stefano Ceri and Georg Gottlob. 2006. Translating SQL into relational algebra: Optimization, semantics, and equivalence of SQL queries.IEEE Transactions on software engineering4 (2006), 324–345.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "ClickHouse. 2026. https://clickhouse.com/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "CockroachDB. 2026. https://www.cockroachlabs.com/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "CockroachDB. 2026. Short-circuit Evaluation in COALESCE() Function. https://www .cockroachlabs .com/docs/stable/ functions-and-operators Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "CockroachDB. 2026. Short-Circuit in Cross Join with Limit 1. https://github .com/cockroachdb/cockroach/issues/148145 Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "DB-Engines. 2026. The DB-Engines Ranking ranks database management systems according to their popularity. https://db-engines.com/en/ranking Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Weimin Du, Ravi Krishnamurthy, and Ming-Chien Shan. 1992. Query Optimization in a Heterogeneous DBMS. In Proceedings of the 18th International Conference on Very Large Data Bases (VLDB ’92). Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 277–291.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "DuckDB. 2026. https://duckdb.org/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Andrew Eisenberg and Jim Melton. 1999. SQL: 1999, formerly known as SQL3.SIGMOD Rec.28, 1 (March 1999), 131–138. doi:10.1145/309844.310075",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Grant Fritchey. 2012. Execution Plan Cache Analysis. InSQL Server 2012 Query Performance Tuning. Springer, 241–279.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2023. Griffin: Grammar-Free DBMS Fuzzing. In Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering(Rochester, MI, USA)(ASE ’22). Association for Computing Machinery, New York, NY, USA, Article 49, 12 pages. doi:10.1145/3551349.3560431",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Piyush Goel and Bala Iyer. 1996. SQL query optimization: reordering for a general class of queries(SIGMOD ’96). Association for Computing Machinery, New York, NY, USA, 47–56. doi:10.1145/233269.233318",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Zuming Jiang, Jiaju Bai, and et al. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. In32nd USENIX Security Symposium (USENIX Security 23). USENIX Association, Anaheim, CA, 4949–4965. https://www.usenix.org/conference/usenixsecurity23/presentation/jiang-zu-ming",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. 2019. APOLLO: automatic detection and diagnosis of performance regressions in database systems.Proc. VLDB Endow.13, 1 (sep 2019), 57–70. doi:10 .14778/3357377 .3357382",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Jinhui Lai, Chi Zhang, Bingyan Li, Chenglin Liang, Jie Liang, Zhiyong Wu, Jingzhou Fu, Yu Jiang, and Zichen Xu. 2025. SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis. 2025), 24 pages. doi:10.1145/3769828",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Hai Lan, Zhifeng Bao, and Yuwei Peng. 2021. A survey on advancing the dbms query optimizer: Cardinality estimation, cost model, and plan enumeration.Data Science and Engineering6, 1 (2021), 86–101.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2015. How good are query optimizers, really?Proceedings of the VLDB Endowment9, 3 (2015), 204–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. Sequence-Oriented DBMS Fuzzing. In Engineering (ICDE). 668–681. doi:10.1109/ICDE55515.2023.00057",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Jie Liang, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Chengnian Sun, and Yu Jiang. 2024. Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering(Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 135, 12 pages. doi:10.1145/3597503.3639112",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. 2022. Automatic detection of performance bugs in database systems using equivalent queries. InProceedings of the 44th International Conference on Software Engineering. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "MariaDB. 2026. https://mariadb.org/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "MariaDB. 2026. FirstMatch Strategy. https://mariadb .com/docs/server/ha-and-performance/optimization-andtuning/query-optimizations/optimization-strategies/firstmatch-strategy Accessed: March 25, 2026. 4, No. 3 (SIGMOD), Article 184. Publication date: June 2026. 184:24 Jinhui Lai et al.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Songsong Mo, Yile Chen, Hao Wang, Gao Cong, and Zhifeng Bao. 2023. Lemo: A Cache-Enhanced Learned Optimizer for Concurrent Queries. 1, 4, Article 247 (Dec. 2023), 26 pages. doi:10.1145/3626734",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "MySQL. 2026. https://www.mysql.com/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "MySQL. 2026. Limit Optimization in MySQL. https://dev .mysql.com/doc/refman/8 .4/en/limit-optimization .html Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Neo4j. 2026. https://neo4j.com/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "OceanBase. 2026. https://www.oceanbase.com/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "The openCypher Implementers Group. 2026. Cypher Query Language Reference. https://s3 .amazonaws.com/ artifacts.opencypher.org/openCypher9.pdf Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "OpenGauss. 2026. https://opengauss.org/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Oracle. 2026. https://www.oracle.com/database/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Oracle. 2026. The CASE Expression in Oracle. https://docs .oracle.com/en/database/oracle/oracle-database/19/sqlrf/ CASE-Expressions.html Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Oracle. 2026. Short-circuit Evaluation in Logical Operators. https://docs .oracle.com/cd/E19253-01/817-6223/chptypeopexpr-6/index.html Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "PingCAP. 2026. The EXPLAIN statement in TiDB. https://docs .pingcap.com/tidb/stable/sql-statement-explain/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Alban Ponse and Daan JC Staudt. 2018. An independent axiomatisation for free short-circuit logic.Journal of Applied Non-Classical Logics28, 1 (2018), 35–71.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "PostgreSQL. 2026. https://www.postgresql.org/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "PostgreSQL. 2026. CREATE TABLE AS SELECT (CTAS) Statements in PostgreSQL. https://www .postgresql.org/docs/ current/sql-createtableas.html Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "PostgreSQL. 2026. Set Returning Functions. https://www.postgresql.org/docs/current/functions-srf.html Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/33680",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (nov 2020), 30 pages. doi:10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, 667–682. https:// www.usenix.org/conference/osdi20/presentation/rigger",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 45,
+      "text": "Tobias Schmidt, Viktor Leis, Peter Boncz, and Thomas Neumann. 2025. SQLStorm: Taking Database Benchmarking into the LLM Era.Proceedings of the VLDB Endowment18, 11 (2025), 4144–4157.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Andreas Seltenreich, Bo Tang, and et al. 2015. https://github.com/anse1/sqlsmith.git Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "SQLite. 2026. https://sqlite.org/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "SQLite. 2026. IF() Function in SQLite. https://sqlite.org/lang_corefunc.html Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "SQLite. 2026. SQLite Fix Example. https://sqlite.org/src/info/e33da6d5dc964db8 Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "TiDB. 2026. https://www.pingcap.com/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "TPC. 2026. TPC-DS Benchmark. https://www.tpc.org/tpcds/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "TPC. 2026. TPC-H Benchmark. https://www.tpc.org/tpch/ Accessed: March 25, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In Engineering: Software Engineering in Practice (ICSE-SEIP). 328–337. doi:10.1109/ICSE-SEIP52600.2021.00042",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2024. PUPPY: Finding Performance Degradation Bugs in DBMSs via Limited-Optimization Plan Construction. In Software Engineering (ICSE). IEEE Computer Society, 560–571.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Zhiyong Wu, Jie Liang, Jingzhou Fu, Mingzhe Wang, and Yu Jiang. 2025. Hulk: Exploring Data-Sensitive Performance Anomalies in DBMSs via Data-Driven Analysis.Proceedings of the ACM on Software Engineering2, ISSTA (2025), 2181–2202.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. InProceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security(Virtual Event, USA)(CCS ’20). Association for Computing Machinery, New York, NY, USA, 955–970. Received October 2025; revised January 2026",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 1,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, USA, Article 133, 13 pages. doi:10.1145/3597503.3639076",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert",
+        "prints the DOI of the paper introducing cert"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(ESEC/FSE 2020). Association for Computing Machinery, New York, NY, USA, 1140–1152. doi:10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (nov 2020), 30 pages. doi:10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. In14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20). USENIX Association, 667–682. https:// www.usenix.org/conference/osdi20/presentation/rigger",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 1,18,23,24,54,55]",
+      "technique": "cert",
+      "sentence": "Some automated testing approaches have been widely adopted to detect performance issues for DBMSs [ 1,18,23,24,54,55].",
+      "context_before": "isms.The internal optimization strategies adopted by commercial DBMSs (e.g., Oracle [ 34]) are typically undisclosed. As a result, it is difficult to determine which optimizations are actually applied during query execution or whether certain optimization opportunities were intentionally omitted. (3)The Absence of a Reliable Performance Oracle.DBMS performance testing lacks a reliable oracle that specifies the reasonable execution time for a given query 𝑄and database 𝐷. Without such ground truth, it is challenging to distinguish a genuine missed optimization from an inherently expensive query.",
+      "context_after": "HULK [ 55] and APOLLO [ 18] detect performance regressions by comparing query execution across different systems or versions. MOZI [ 23] and PUPPY [ 54] vary optimizer configurations to trigger different execution plans and reveal performance degradation. CERT [ 1] focuses on cardinality estimation by deriving restrictive queries to identify estimation errors that may affect optimization. While effective in uncovering issues in implemented optimizations, none of these approaches explicitly targets missed optimization opportunities, where the optimizer fails to choose an execution strategy tha",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 6796,
+      "cited_reference": {
+        "number": 1,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, U",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert",
+          "prints the DOI of the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "CERT [ 1] focuses on cardinality estimation by deriving restrictive queries to identify estimation errors that may affect optimization.",
+      "context_before": "s a reliable oracle that specifies the reasonable execution time for a given query 𝑄and database 𝐷. Without such ground truth, it is challenging to distinguish a genuine missed optimization from an inherently expensive query. Some automated testing approaches have been widely adopted to detect performance issues for DBMSs [ 1,18,23,24,54,55]. HULK [ 55] and APOLLO [ 18] detect performance regressions by comparing query execution across different systems or versions. MOZI [ 23] and PUPPY [ 54] vary optimizer configurations to trigger different execution plans and reveal performance degradation.",
+      "context_after": "While effective in uncovering issues in implemented optimizations, none of these approaches explicitly targets missed optimization opportunities, where the optimizer fails to choose an execution strategy that would yield more efficient query evaluation. 4, No. 3 (SIGMOD), Article 184. Publication date: June 2026. Finding Missed Optimizations in DBMSs through Unbalanced Short-Circuit Query Construction 184:3 CREATE TABLE t1(c1 NUMBER(10)); INSERT INTO t1 SELECT LEVEL FROM dual CONNECT BYLEVEL <= 1000000000; 1. Generate table t1 and insert t1 with one billion rows 2. Generate two queries and e",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 7171,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 1,18,23,24,54,55]",
+      "technique": "cert",
+      "sentence": "Detecting these missed optimizations remains challenging, as existing approaches [ 1,18,23,24,54,55] typically observe performance anomalies without revealing where the optimizer fails.",
+      "context_before": "ften reorder operations to minimize estimated costs. For instance, a predicate written as 𝑒𝑥𝑝𝑟 1OR𝑒𝑥𝑝𝑟 2may be executed as 𝑒𝑥𝑝𝑟 2 OR𝑒𝑥𝑝𝑟 1depending on cost estimates, illustrating how optimization decisions dictate the actual execution steps of a query. Despite these sophisticated optimization techniques, DBMSs can still fail to select the most efficient execution plan for certain queries. Such missed optimizations occur when the optimizer does not apply the most cost-effective execution strategy, resulting in unnecessary computation, degraded performance, and inefficient resource utilization.",
+      "context_after": "Short-Circuit Evaluation.Short-circuit evaluation in DBMSs halts query execution once the final result is determined. For example, in 𝑒𝑥𝑝𝑟 1OR𝑒𝑥𝑝𝑟 2, if𝑒𝑥𝑝𝑟 1evaluates to TRUE,𝑒𝑥𝑝𝑟 2’s execution becomes unnecessary. DBMSs primarily implement three types of short-circuiting: (1) Compile-time short-circuit, where syntactic constructs (e.g., WHERE FALSE [36] or LIMIT 0 [29]) in a query enable the optimizer to infer a constant result or prune entire execution branches during query compilation. (2) Runtime short-circuit, where execution operators (e.g., semi-join [ 26] or LIMIT n) can stop processin",
+      "section": "2 Background",
+      "page": 4,
+      "char_offset": 12812,
+      "cited_reference": {
+        "number": 1,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, U",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert",
+          "prints the DOI of the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[ 1]",
+      "technique": "cert",
+      "sentence": "Metamorphic testing [ 1] addresses this by constructing oracles through relations between inputs and outputs.",
+      "context_before": "n stop processing once a termination condition is met, with the decision made dynamically during execution. (3) Statistics-aware short-circuit, where the optimizer leverages metadata such as table emptiness, selectivity estimates, or design choices like plan cache [ 27] to choose plans that are expected to exploit short-circuiting opportunities at runtime. Metamorphic testing.A reliable test oracle is essential for identifying missed optimizations in DBMSs, as it defines the expected performance of a query and allows distinguishing genuine optimization misses from inherently expensive queries.",
+      "context_after": "For a given input 𝐼producing output𝑂, the method derives a transformed input 𝐼′(with output 𝑂′) and verifies whether a predefined relation holds between 𝑂and𝑂′. SCor adapts this approach with the key insight thatif a query’s result can be determined from partial execution, yet the DBMS executes the query entirely, this indicates a missed optimization.We leverage short-circuit evaluation principles to construct such metamorphic relations. For example, given 𝑄1:SELECT (SELECT MIN(t1.c1) FROM t1) > 0 OR TRUE wheret1is a large table, we create 𝑄2by replacing t1with a small table or empty table t0",
+      "section": "2 Background",
+      "page": 4,
+      "char_offset": 14182,
+      "cited_reference": {
+        "number": 1,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. CERT: Finding Performance Issues in Database Systems Through the Lens of Cardinality Estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (Lisbon, Portugal)(ICSE ’24). Association for Computing Machinery, New York, NY, U",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert",
+          "prints the DOI of the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Finding Missed Optimizations in DBMSs through Unbalanced Short-Circuit Query Construction 184:13 Given the wide array of existing data generation methods—such as testing benchmarks, TPCH [52], automated testing tools like SQLancer [ 42–44], and built-in DBMS functions [ 41]—SCor does not reimplement data generation.",
+      "context_before": "construct unbalanced short-circuit queries, SCor creates tables with varying data sizes, including several empty tables ($ 𝑒𝑚𝑝𝑡𝑦𝑇𝑎𝑏𝑙𝑒 ), small tables ($ 𝑠𝑚𝑎𝑙𝑙𝑇𝑎𝑏𝑙𝑒 ) containing a limited number of records (e.g., 1-100 rows), and large tables ($ 𝑙𝑎𝑟𝑔𝑒𝑇𝑎𝑏𝑙𝑒 ) with a substantial volume of data (e.g., millions or billions of rows). This enables the generation of both high-cost expressions ($ ℎ𝑖𝑔ℎ𝐶𝑜𝑠𝑡𝐸𝑥𝑝𝑟 ), such as queries executed on large tables, and low-cost expressions ($𝑙𝑜𝑤𝐶𝑜𝑠𝑡𝐸𝑥𝑝𝑟), such as those applied to small or empty tables. 4, No. 3 (SIGMOD), Article 184. Publication date: June 2026.",
+      "context_after": "Instead, it leverages these established approaches. For instance, in PostgreSQL, the generate_series() function can be invoked to produce sequences of varying sizes, facilitating the creation of tables with diverse data volumes. Query Generation. SCor instantiates the query skeletons in Table 1 by generating concrete SQL expressions and statements based on the initial database state. To ensure syntactic correctness, we employ a grammar-based approach that encodes SQL-99 [ 13] grammar into production rules. Through random walks over the resulting grammar tree, the system generates diverse vali",
+      "section": "4 Implementation",
+      "page": 12,
+      "char_offset": 42453,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "We compared SCor with five state-of-the-art DBMS performance bug detection tools, including: APOLLO [ 18], AMOEBA [ 24], and CERT [ 1], Puppy [ 54], and HULK [ 55].",
+      "context_before": "1(c1 INT8); -- Insert into t1 with 10 million rows -- Pattern 5.1: emptyTable JOIN anyTable SELECT*FROMt1RIGHT JOIN(SELECT*FROMt1 WHERE FALSE)ASeONe.c1 = t1.c1; -- 0.001s SELECT*FROM(SELECT*FROMt1 WHERE FALSE)ASeLEFT JOINt1ONe.c1 = t1.c1; -- 3.975s -- Pattern 5.2: emptySet INTERSECT anySet SELECT*FROMt1 WHERE FALSEINTERSECT SELECT*FROMt1; -- 0.001s SELECT*FROMt1INTERSECT SELECT*FROMt1 WHERE FALSE; -- 18.764s 5.4 Comparative Evaluation To illustrate the effectiveness and efficiency of our tools, we conducted a comparative evaluation against existing DBMS performance bug testing tools. Baseline.",
+      "context_after": "APOLLO detects performance regression bugs by comparing the execution time of the same query across different DBMS versions. AMOEBA detects performance bugs by comparing execution times of semantically equivalent queries. CERT tackles performance bugs through cardinality estimation analysis. Puppy detects the performance bug by changing the optimization configurations. HULK detects the data-sensitive performance anomalies by changing the data volume. Effectiveness of SCor Comparison to Baseline.In this experiment, we run each tool on the latest version of each supported DBMS for a duration of",
+      "section": "5.4 Comparative Evaluation",
+      "page": 19,
+      "char_offset": 66744,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "CERT tackles performance bugs through cardinality estimation analysis.",
+      "context_before": "1 WHERE FALSE; -- 18.764s 5.4 Comparative Evaluation To illustrate the effectiveness and efficiency of our tools, we conducted a comparative evaluation against existing DBMS performance bug testing tools. Baseline.We compared SCor with five state-of-the-art DBMS performance bug detection tools, including: APOLLO [ 18], AMOEBA [ 24], and CERT [ 1], Puppy [ 54], and HULK [ 55]. APOLLO detects performance regression bugs by comparing the execution time of the same query across different DBMS versions. AMOEBA detects performance bugs by comparing execution times of semantically equivalent queries.",
+      "context_after": "Puppy detects the performance bug by changing the optimization configurations. HULK detects the data-sensitive performance anomalies by changing the data volume. Effectiveness of SCor Comparison to Baseline.In this experiment, we run each tool on the latest version of each supported DBMS for a duration of 24 hours. Since the baseline approaches do not support all the DBMSs that SCor does, our comparison focuses on the DBMSs commonly supported by both SCor and each baseline. Table 4 presents the number of unique performance bugs identified by each approach on each DBMS. The results show that S",
+      "section": "5.4 Comparative Evaluation",
+      "page": 19,
+      "char_offset": 67130,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "The results show that SCor found 24, 8, 36, 8, and 4 more bugs than APOLLO, AMOEBA, CERT, Puppy, and HULK, respectively.",
+      "context_before": "ity estimation analysis. Puppy detects the performance bug by changing the optimization configurations. HULK detects the data-sensitive performance anomalies by changing the data volume. Effectiveness of SCor Comparison to Baseline.In this experiment, we run each tool on the latest version of each supported DBMS for a duration of 24 hours. Since the baseline approaches do not support all the DBMSs that SCor does, our comparison focuses on the DBMSs commonly supported by both SCor and each baseline. Table 4 presents the number of unique performance bugs identified by each approach on each DBMS.",
+      "context_after": "Note that the increments are calculated based only on the DBMSs commonly supported by both approaches. The other approaches were less effective in our experiments because we tested the latest versions of the DBMSs, where previously known bug patterns have largely been addressed. This reduced effectiveness suggests that discovering additional bugs has become more challenging for the baseline, and the bugs detected by our approach are likely beyond the reach of existing approaches. Furthermore, our analysis reveals that the missed optimizations identified by SCor show limited overlap with only",
+      "section": "5.4 Comparative Evaluation",
+      "page": 19,
+      "char_offset": 67777,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "DBMSAPOLLO AMOEBA CERT Puppy HULK SCor PostgreSQL 0 0 - 2 2 3 MySQL - - 3 6 10 13 CockroachDB-1 2 - - 6 TiDB - - 4 - - 26 SQLite 0 - - - - 21 Total 0 1 9 8 12 69 Increment 24↑8↑36↑8↑4↑data-sensitive performance anomalies through data changes, while SCor employs unbalanced shortcircuit queries.",
+      "context_before": "verlap) in common with HULK and no overlap with other baseline tools. This partial alignment with HULK is expected, as both approaches leverage data volume variation. However, their testing strategies differ: HULK detects 4, No. 3 (SIGMOD), Article 184. Publication date: June 2026. 184:20 Jinhui Lai et al. Table 4. Number of unique performance bugs detected by each testing tool in 24 hours. This comparison focuses on the DBMSs commonly supported by both SCor and each baseline. “-” indicates that the tool does not support the DBMS. “Increments” are calculated only for commonly supported DBMSs.",
+      "context_after": "The limited overlap with HULK confirms that SCor’s pattern-based generation strategy uncovers different optimization opportunities, systematically constructing queries with short-circuit patterns that HULK does not target. The non-overlap with other tools, except HULK, stems from fundamental differences in both target and methodology. (1) Target: Existing baselines primarily target performance degradation in implemented optimizations, whereas SCor specifically detects missed optimization. Specifically, APOLLO targets performance regression bugs, AMOEBA focuses on performance differences among",
+      "section": "5.4 Comparative Evaluation",
+      "page": 20,
+      "char_offset": 69141,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "Specifically, APOLLO targets performance regression bugs, AMOEBA focuses on performance differences among semantically equivalent queries, CERT addresses cardinality estimation issues, PUPPY identifies configuration-related problems, and HULK detects data-sensitive anomalies.",
+      "context_before": "-sensitive performance anomalies through data changes, while SCor employs unbalanced shortcircuit queries. The limited overlap with HULK confirms that SCor’s pattern-based generation strategy uncovers different optimization opportunities, systematically constructing queries with short-circuit patterns that HULK does not target. The non-overlap with other tools, except HULK, stems from fundamental differences in both target and methodology. (1) Target: Existing baselines primarily target performance degradation in implemented optimizations, whereas SCor specifically detects missed optimization.",
+      "context_after": "(2) Methodology: As shown in Table 3, 147 missed optimizations found by SCor trace back to initial releases, indicating their long-standing nature. This makes them undetectable by APOLLO, which relies on cross-version comparison of identical queries. Similarly, AMOEBA may fail when semantically equivalent queries exhibit similarly severe performance degradation. CERT cannot detect opportunities that persist despite correct cardinality estimates. For example, in Listing 2, both queries have accurate cardinality estimates, yet the absence of short-circuit evaluation represents a distinct optimi",
+      "section": "5.4 Comparative Evaluation",
+      "page": 20,
+      "char_offset": 69930,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "CERT cannot detect opportunities that persist despite correct cardinality estimates.",
+      "context_before": "regression bugs, AMOEBA focuses on performance differences among semantically equivalent queries, CERT addresses cardinality estimation issues, PUPPY identifies configuration-related problems, and HULK detects data-sensitive anomalies. (2) Methodology: As shown in Table 3, 147 missed optimizations found by SCor trace back to initial releases, indicating their long-standing nature. This makes them undetectable by APOLLO, which relies on cross-version comparison of identical queries. Similarly, AMOEBA may fail when semantically equivalent queries exhibit similarly severe performance degradation.",
+      "context_after": "For example, in Listing 2, both queries have accurate cardinality estimates, yet the absence of short-circuit evaluation represents a distinct optimization gap beyond cardinality concerns. PUPPY identifies performance issues by modifying DBMS optimization configurations, which primarily reveals scenarios where existing optimizations perform poorly, but cannot detect missed optimizations that were never implemented. 6 Discussion SCor Scope.SCor systematically uncovers missed optimizations through unbalanced short-circuit query construction. By leveraging the prevalence of short-circuit pattern",
+      "section": "5.4 Comparative Evaluation",
+      "page": 20,
+      "char_offset": 70572,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "CERT",
+      "technique": "cert",
+      "sentence": "CERT [ 1] tackles performance bugs through cardinality estimation analysis.",
+      "context_before": "s. Third, the methodology of constructing unbalanced short-circuit queries could be systematically generalized to other DBMS paradigms, such as graph DBMSs, to further validate and broaden the applicability of SCor. 7 Related Work DBMS Performance Bug Testing.The most related strand of research is on automatic detection of performance issues in DBMSs. APOLLO [ 18] detects performance regression bugs by comparing the execution time of the same query across different DBMS versions. AMOEBA [ 24] identifies unexpected slowdowns by contrasting the execution times of semantically equivalent queries.",
+      "context_after": "MOZI [ 23] detects performance bugs by leveraging variations in configuration-related configurations. PUPPY [ 54] detects performance degradation bugs through limited optimization plan construction. HULK [ 55] detects data-sensitive performance anomalies via data-driven analysis. As we mentioned in Section 5.4, SCor differs from these testing works in both target and methodology. Overall, existing approaches primarily focus on identifying performance degradations caused by faulty or suboptimal implementations of optimization, whereas SCor is designed to expose optimization opportunities that",
+      "section": "7 Related Work",
+      "page": 21,
+      "char_offset": 74508,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "citation_marker",
+      "surface": "[ 19,22,42–44,46,56]",
+      "technique": "norec",
+      "sentence": "The recent application of fuzzing to DBMSs has successfully uncovered hundreds of bugs [ 19,22,42–44,46,56].",
+      "context_before": "QL workloads automatically. SCor complements these benchmarks by addressing an orthogonal problem. While traditional benchmarks primarily detect performance regression bugs using user-relevant workloads, SCor 4, No. 3 (SIGMOD), Article 184. Publication date: June 2026. 184:22 Jinhui Lai et al. exposes missed optimization opportunities that standard benchmarks rarely reveal, by constructing unbalanced short-circuit queries, even on previously less-seen workloads. DBMS Fuzzing.Besides testing DBMS performance, automated testing approaches have also been developed to uncover other types of bugs.",
+      "context_after": "A typical DBMS fuzzer operates by continuously generating SQL test cases, executing them on the target system, and monitoring for incorrect behavior. These approaches are broadly categorized as either generation-based or mutation-based fuzzing. Generation-based fuzzers construct SQL queries from scratch using predefined syntax or grammar models. For example, SQLSmith [ 46] generates queries with built-in AST generation rules to expose memory errors. Similarly, SQLancer uses a generation-based approach guided by its test oracles (PQS [ 44], NoREC [ 42], TLP [ 43], and SRS [ 19]) to detect logi",
+      "section": "7 Related Work",
+      "page": 22,
+      "char_offset": 76809,
+      "cited_reference": {
+        "number": 42,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Similarly, SQLancer uses a generation-based approach guided by its test oracles (PQS [ 44], NoREC [ 42], TLP [ 43], and SRS [ 19]) to detect logic bugs.",
+      "context_before": "loped to uncover other types of bugs. The recent application of fuzzing to DBMSs has successfully uncovered hundreds of bugs [ 19,22,42–44,46,56]. A typical DBMS fuzzer operates by continuously generating SQL test cases, executing them on the target system, and monitoring for incorrect behavior. These approaches are broadly categorized as either generation-based or mutation-based fuzzing. Generation-based fuzzers construct SQL queries from scratch using predefined syntax or grammar models. For example, SQLSmith [ 46] generates queries with built-in AST generation rules to expose memory errors.",
+      "context_after": "In contrast, mutation-based fuzzers generate new queries by mutating existing ones. SQUIRREL [ 56], LEGO [ 22], and Ratel [ 53] perform mutations based on a query’s AST structure. DynSQL [ 17] incrementally constructs complex queries by leveraging state information. GRIFFIN [ 15] adopts a grammar-free approach that reshuffles statements from different queries and restores semantic correctness by tracking the database schema. SCor could be regarded as a generation-based fuzzer. Different from other generation-based fuzzers, SCor generates SQL queries that incorporate a diverse range of short-c",
+      "section": "7 Related Work",
+      "page": 22,
+      "char_offset": 77372,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M6"
+      ],
+      "describes_as_state_of_the_art": [
+        "M6"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:58:40Z",
+    "is_model_written": true,
+    "summary": "SCor finds missed optimizations by constructing unbalanced short-circuit queries: a predicate is written so that one branch could be short-circuited away, and if the optimizer fails to exploit that the query does measurably more work than it needs to. Unlike approaches that observe a performance anomaly and leave its cause open, this points at the specific optimization the optimizer missed. It found 69 issues across five DBMSs.",
+    "narrative": "CERT is SCor's most direct comparison, one of five performance-bug detection tools it is measured against and described as tackling performance bugs through cardinality estimation analysis. SCor found 36 more bugs than CERT, and the paper gives a precise account of the difference in reach: CERT cannot detect optimization opportunities that persist despite correct cardinality estimates, which is exactly the class short-circuit construction targets. SQLancer also appears as a data-generation source the paper declines to reimplement, listing it among the existing methods available for producing test data.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "motivation",
+      "M4": "background",
+      "M5": "incidental",
+      "M6": "state_of_the_art",
+      "M7": "definition",
+      "M8": "result_comparison",
+      "M9": "result_comparison",
+      "M10": "definition",
+      "M11": "motivation",
+      "M12": "definition",
+      "M13": "background",
+      "M14": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M5 lists SQLancer among existing data generation methods as the reason SCor does not reimplement data generation, but does not say SCor uses it. Which generator supplies its data is left open, so this falls short of a statement of reuse."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Unbalanced short-circuit construction is a different mechanism from cardinality-estimation restriction, and M11 sets out the class of issues CERT cannot reach as the motivation."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M6",
+          "M8",
+          "M9"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M6",
+            "sentence": "We compared SCor with five state-of-the-art DBMS performance bug detection tools, including: APOLLO [ 18], AMOEBA [ 24], and CERT [ 1], Puppy [ 54], and HULK [ 55].",
+            "section": "5.4 Comparative Evaluation",
+            "page": 19
+          },
+          {
+            "mention_id": "M8",
+            "sentence": "The results show that SCor found 24, 8, 36, 8, and 4 more bugs than APOLLO, AMOEBA, CERT, Puppy, and HULK, respectively.",
+            "section": "5.4 Comparative Evaluation",
+            "page": 19
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "DBMSAPOLLO AMOEBA CERT Puppy HULK SCor PostgreSQL 0 0 - 2 2 3 MySQL - - 3 6 10 13 CockroachDB-1 2 - - 6 TiDB - - 4 - - 26 SQLite 0 - - - - 21 Total 0 1 9 8 12 69 Increment 24↑8↑36↑8↑4↑data-sensitive performance anomalies through data changes, while SCor employs unbalanced shortcircuit queries.",
+            "section": "5.4 Comparative Evaluation",
+            "page": 20
+          }
+        ],
+        "reasoning": "M6 names CERT among the five tools SCor is compared with, M8 reports SCor finding 36 more bugs than CERT, and M9 gives the per-DBMS table.",
+        "techniques": [
+          "cert"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M6"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M6",
+            "sentence": "We compared SCor with five state-of-the-art DBMS performance bug detection tools, including: APOLLO [ 18], AMOEBA [ 24], and CERT [ 1], Puppy [ 54], and HULK [ 55].",
+            "section": "5.4 Comparative Evaluation",
+            "page": 19
+          }
+        ],
+        "reasoning": "M6 describes the five comparison tools, CERT among them, as state-of-the-art DBMS performance bug detection tools."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3803437_3806090.json b/_data/papers/paper_doi_10_1145_3803437_3806090.json
new file mode 100644
index 0000000..3db4448
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3803437_3806090.json
@@ -0,0 +1,286 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:45Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3803437.3806090",
+    "title": "Why Property-Based Testing is Necessary for Data Intensive Scalable Computing",
+    "authors": [
+      "Yaoxuan Wu",
+      "I. Lee",
+      "Ahmad Humayun",
+      "Muhammad Ali Gulzar",
+      "Miryung Kim"
+    ],
+    "year": 2026,
+    "venue": "SIGSOFT FSE Companion",
+    "doi": "10.1145/3803437.3806090",
+    "arxiv_id": null,
+    "s2_paper_id": "ba02b6d2cf64ddf05e51681950d9094399769829",
+    "url": "https://doi.org/10.1145/3803437.3806090",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3803437.3806090",
+      "retrieved_at": "2026-09-09T01:29:45Z",
+      "chars": 11698,
+      "content_sha256": "sha256:01cdc3f91e34cd72c0714b52bf6186f546ab4abe2259e78f15585f9b2c7e8547"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 2,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "H. Goldstein, J. W. Cutler, D. Dickstein, B. C. Pierce, and A. Head, “Property-based testing in practice,” in ICSE, 2024, pp. 1–13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“SPARK-49000,” https://issues.apache.org/jira/browse/SPARK-49000.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“SPARK-33726,” https://issues.apache.org/jira/browse/SPARK-33726.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Q. Zhang, J. Wang, M. A. Gulzar, R. Padhye, and M. Kim, “BigFuzz: Efficient fuzz testing for data analytics using framework abstraction,” in ASE, 2020, pp. 722–733.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "M. A. Gulzar, S. Mardani, M. Musuvathi, and M. Kim, “White-box testing of big data analytics with complex user-defined functions,” in ESEC/FSE, 2019, pp. 290–301.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "A. Humayun, Y. Wu, M. Kim, and M. A. Gulzar, “NaturalFuzz: Natural input generation for big data analytics,” in ASE, 2023, pp. 1592–1603.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Y. Wu, A. Humayun, M. A. Gulzar, and M. Kim, “Natural symbolic executionbased testing for big data analytics,” Proc. ACM Softw. Eng., vol. 1, no. FSE, pp. 2677–2700, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "A. Riesco and J. Rodríguez-Hortalá, “Property-based testing for Spark Streaming,” arXiv preprint arXiv:1812.11838, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "C. V. Espinosa, E. Martin-Martin, A. Riesco, and J. Rodriguez-Hortala, “FlinkCheck: Property-based testing for Apache Flink,” IEEE Access, vol. 7, pp. 150 369–150 382, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "M. E. Kroner, “Achilles’ SPEar: Using metamorphic testing to find bugs in stream processing engines,” in BTW 2025, 2025, pp. 1031–1042.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "K. Kallas, F. Niksic, C. Stanford, and R. Alur, “DiffStream: differential output testing for stream processing programs,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 1–29, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "“SQLancer,” https://github.com/sqlancer/sqlancer.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 13,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "——, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in ESEC/FSE, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "——, “Testing database engines via pivoted query synthesis,” in OSDI, 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 16,
+      "text": "“SPARK-39854,” https://issues.apache.org/jira/browse/SPARK-39854.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "“Data Types: NaN Semantics, ” https://downloads.apache.org/spark/docs/3.0.0/sqlref-datatypes.html. 1727",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 12,
+      "text": "“SQLancer,” https://github.com/sqlancer/sqlancer.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 13,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 14,
+      "text": "——, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in ESEC/FSE, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 15,
+      "text": "——, “Testing database engines via pivoted query synthesis,” in OSDI, 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer integrates multiple property oracles and has uncovered many previously unknown DBMS logic bugs [ 12].",
+      "context_before": "Work. In DISC testing, SSCheck [8] and FlinkCheck [9] focus on temporal properties over streams, leaving optimizer-centric behaviors largely unexplored. Achilles’ SPEar [ 10] uses metamorphic testing to check stream processing engines, but its generated workloads cover only a narrow operator subset (filter/map/aggregation/windowed aggregation). DiffStream [ 11] provides differential testing with a stream-equivalence oracle, but it does not automatically generate equivalent program pairs beyond disabling parallelism. In contrast, DBMS testing has shown the effectiveness of oraclebased testing.",
+      "context_after": "Representative examples include query partitioning, semanticspreserving query rewriting, and pivoted query synthesis. [13–15] 1726 FSE Companion ’26, July 05–09, 2026, Montreal, QC, Canada Wu et al. Table 1: Example property contracts for DISC frameworks, organized by shared templates. What is tested Per-group recomposition for aggregation operators PerGroupRecompose(sum). Group by a key column, apply sum (𝑦)per group, recompose by sum, and match the global sum (𝑦). PerGroupRecompose(max). Group by a key column, apply max (𝑦)per group, recompose by max, and match the global max (𝑦). Cardinali",
+      "section": null,
+      "page": 1,
+      "char_offset": 4532,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "Representative examples include query partitioning, semanticspreserving query rewriting, and pivoted query synthesis.",
+      "context_before": "izer-centric behaviors largely unexplored. Achilles’ SPEar [ 10] uses metamorphic testing to check stream processing engines, but its generated workloads cover only a narrow operator subset (filter/map/aggregation/windowed aggregation). DiffStream [ 11] provides differential testing with a stream-equivalence oracle, but it does not automatically generate equivalent program pairs beyond disabling parallelism. In contrast, DBMS testing has shown the effectiveness of oraclebased testing. SQLancer integrates multiple property oracles and has uncovered many previously unknown DBMS logic bugs [ 12].",
+      "context_after": "[13–15] 1726 FSE Companion ’26, July 05–09, 2026, Montreal, QC, Canada Wu et al. Table 1: Example property contracts for DISC frameworks, organized by shared templates. What is tested Per-group recomposition for aggregation operators PerGroupRecompose(sum). Group by a key column, apply sum (𝑦)per group, recompose by sum, and match the global sum (𝑦). PerGroupRecompose(max). Group by a key column, apply max (𝑦)per group, recompose by max, and match the global max (𝑦). Cardinality relations for DataFrame operators Cardinality(union). Apply union to two inputs and validate that output cardinalit",
+      "section": null,
+      "page": 1,
+      "char_offset": 4643,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[13–15]",
+      "technique": "tlp",
+      "sentence": "[13–15] 1726 FSE Companion ’26, July 05–09, 2026, Montreal, QC, Canada Wu et al.",
+      "context_before": "ngines, but its generated workloads cover only a narrow operator subset (filter/map/aggregation/windowed aggregation). DiffStream [ 11] provides differential testing with a stream-equivalence oracle, but it does not automatically generate equivalent program pairs beyond disabling parallelism. In contrast, DBMS testing has shown the effectiveness of oraclebased testing. SQLancer integrates multiple property oracles and has uncovered many previously unknown DBMS logic bugs [ 12]. Representative examples include query partitioning, semanticspreserving query rewriting, and pivoted query synthesis.",
+      "context_after": "Table 1: Example property contracts for DISC frameworks, organized by shared templates. What is tested Per-group recomposition for aggregation operators PerGroupRecompose(sum). Group by a key column, apply sum (𝑦)per group, recompose by sum, and match the global sum (𝑦). PerGroupRecompose(max). Group by a key column, apply max (𝑦)per group, recompose by max, and match the global max (𝑦). Cardinality relations for DataFrame operators Cardinality(union). Apply union to two inputs and validate that output cardinality matches the intended union semantics. Cardinality(filter). Apply filter with a",
+      "section": null,
+      "page": 1,
+      "char_offset": 4761,
+      "cited_reference": {
+        "number": 13,
+        "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:19:46Z",
+    "is_model_written": true,
+    "summary": "A short position paper arguing that data-intensive scalable computing frameworks such as Spark, Flink, Beam and Dask need property-based testing. Its premise is that framework bugs are common and frequently show up as silent wrong results rather than crashes, and that property-based testing has succeeded elsewhere by using semantic contracts as oracles where ground truth is hard to obtain. The authors call for a general, extensible framework with reusable property templates.",
+    "narrative": "SQLancer is the paper's central example of the approach working: it is described as integrating multiple property oracles and having uncovered many previously unknown DBMS logic bugs, with query partitioning, semantics-preserving rewriting and pivoted query synthesis named as the representative examples.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3808109.json b/_data/papers/paper_doi_10_1145_3808109.json
new file mode 100644
index 0000000..66db41c
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3808109.json
@@ -0,0 +1,494 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:05:11Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3808109",
+    "title": "Eidolon: Perform Noise-Aware Fuzzing on FHE Libraries via Equivalence Expression Transformation",
+    "authors": [
+      "Zhensheng Xian",
+      "Zhen Yan",
+      "Yuanliang Chen",
+      "Xuelian Cao",
+      "Fuchen Ma",
+      "Dalong Shi",
+      "Yu Jiang"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Software Engineering",
+    "doi": "10.1145/3808109",
+    "arxiv_id": null,
+    "s2_paper_id": "2dd21a6a9998e76f0247f28e7d30938c66112b8a",
+    "url": "https://doi.org/10.1145/3808109",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3808109",
+      "retrieved_at": "2026-09-07T17:05:11Z",
+      "chars": 80638,
+      "content_sha256": "sha256:03b4dab181827a6305851d98eb60684aab0f9b6ff29e9cb7b6d4ab5f15e382cf"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 22,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2349
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 9815
+      },
+      {
+        "number": "2.1",
+        "title": "Fully Homomorphic Encryption Concept",
+        "start": 9828
+      },
+      {
+        "number": "2.2",
+        "title": "Noise Mechanism in FHE",
+        "start": 12969
+      },
+      {
+        "number": "3",
+        "title": "Motivation",
+        "start": 15177
+      },
+      {
+        "number": "4",
+        "title": "Design",
+        "start": 21412
+      },
+      {
+        "number": "4.1",
+        "title": "Eidolon Workflow",
+        "start": 22544
+      },
+      {
+        "number": "4.2",
+        "title": "Test Case Construction",
+        "start": 26180
+      },
+      {
+        "number": "4.3",
+        "title": "Noise-Aware Fuzzing",
+        "start": 33925
+      },
+      {
+        "number": "4.4",
+        "title": "Bug Detection and Analysis",
+        "start": 39131
+      },
+      {
+        "number": "5",
+        "title": "Implementation",
+        "start": 40871
+      },
+      {
+        "number": "6",
+        "title": "Evaluation",
+        "start": 46107
+      },
+      {
+        "number": "6.1",
+        "title": "Bugs in FHE Libraries",
+        "start": 47402
+      },
+      {
+        "number": "6.2",
+        "title": "Effectiveness of Noise-Aware Exploration",
+        "start": 54361
+      },
+      {
+        "number": "6.3",
+        "title": "Comparison with Other Fuzzers",
+        "start": 57959
+      },
+      {
+        "number": "7",
+        "title": "Discussion",
+        "start": 62522
+      },
+      {
+        "number": "8",
+        "title": "Related Work",
+        "start": 66871
+      },
+      {
+        "number": "8.2",
+        "title": "Metamorphic Testing",
+        "start": 67860
+      },
+      {
+        "number": "8.3",
+        "title": "Fuzzing",
+        "start": 68587
+      },
+      {
+        "number": "9",
+        "title": "Conclusion",
+        "start": 69458
+      },
+      {
+        "number": "10",
+        "title": "Data Availability",
+        "start": 70692
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Jonathan Anastasia and Derek Ho. 2024.Stepping through the looking glass of privacy-enhancing technologies. Mastercard. https://www.mastercard.com/news/perspectives/2024/stepping-through-the-looking-glass-of-privacyenhancing-technologies/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Apple. 2024. Combining Machine Learning and Homomorphic Encryption in the Apple Ecosystem. https: //machinelearning.apple.com/research/homomorphic-encryption",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Andrea Arcuri and Lionel Briand. 2011. A practical guide for using statistical tests to assess randomized algorithms in software engineering. InProceedings of the 33rd international conference on software engineering. 1–10.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Jean-Philippe Aumasson and Yolan Romailler. 2017. Automated testing of crypto software using differential fuzzing. Black Hat USA7 (2017), 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Ahmad Al Badawi, Andreea Alexandru, Jack Bates, Flavio Bergamaschi, David Bruce Cousins, Saroja Erabelli, Nicholas Genise, Shai Halevi, Hamish Hunt, Andrey Kim, Yongwoo Lee, Zeyu Liu, Daniele Micciancio, Carlo Pascoe, Yuriy Polyakov, Ian Quah, Saraswathy R.V., Kurt Rohloff, Jonathan Saylor, Dmitriy Suponitsky, Matthew Triplett, Vinod Vaikuntanathan, and Vincent Zucca. 2022. OpenFHE: Open-Source Fu",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Marcelo Blatt, Alexander Gusev, Yuriy Polyakov, Kurt Rohloff, and Vinod Vaikuntanathan. 2020. Optimized homomorphic encryption solution for secure genome-wide association studies.BMC Medical Genomics13, Suppl 7 (2020), 83.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Zvika Brakerski. 2012. Fully homomorphic encryption without modulus switching from classical GapSVP. InAnnual cryptology conference. Springer, 868–886.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Zvika Brakerski, Craig Gentry, and Vinod Vaikuntanathan. 2014. (Leveled) fully homomorphic encryption without bootstrapping.ACM Transactions on Computation Theory (TOCT)6, 3 (2014), 1–36.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Tsong Yueh Chen, Fei-Ching Kuo, Huai Liu, Pak-Lok Poon, Dave Towey, T. H. Tse, and Zhi Quan Zhou. 2018. Metamorphic Testing: A Review of Challenges and Opportunities.ACM Comput. Surv.51, 1, Article 4 (Jan. 2018), 27 pages. doi:10.1145/3143561",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Tsong Yueh Chen, Fei-Ching Kuo, Wenjuan Ma, Willy Susilo, Dave Towey, Jeffrey Voas, and Zhi Quan Zhou. 2016. Metamorphic Testing for Cybersecurity.Computer49, 6 (June 2016), 48–55. doi:10.1109/MC.2016.176",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jung Hee Cheon, Kyoohyung Han, Andrey Kim, Miran Kim, and Yongsoo Song. 2018. Bootstrapping for approximate homomorphic encryption. InAnnual International Conference on the Theory and Applications of Cryptographic Techniques. Springer, 360–384.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jung Hee Cheon, Kyoohyung Han, Andrey Kim, Miran Kim, and Yongsoo Song. 2018. A full RNS variant of approximate homomorphic encryption. InInternational Conference on Selected Areas in Cryptography. Springer, 347–368.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jung Hee Cheon, Andrey Kim, Miran Kim, and Yongsoo Song. 2017. Homomorphic encryption for arithmetic of approximate numbers. InInternational conference on the theory and application of cryptology and information security. Springer, 409–437.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Ilaria Chillotti, Nicolas Gama, Mariya Georgieva, and Malika Izabachène. 2020. TFHE: fast fully homomorphic encryption over the torus.Journal of Cryptology33, 1 (2020), 34–91.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Wikipedia contributors. 2025. Horner’s method. https://en.wikipedia.org/wiki/Horner%27s_method Accessed: 2025-09-12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Ana Costache and Nigel P Smart. 2016. Which ring based somewhat homomorphic encryption scheme is best?. In Cryptographers’ Track at the RSA Conference. Springer, 325–340.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Junfeng Fan and Frederik Vercauteren. 2012. Somewhat practical fully homomorphic encryption.Cryptology ePrint Archive(2012).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. 2020. AFL++: Combining Incremental Steps of Fuzzing Research. In14th USENIX Workshop on Offensive Technologies (WOOT 20). USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Craig Gentry. 2009.A fully homomorphic encryption scheme. Stanford university. Proc. ACM Softw. Eng., Vol. 3, No. FSE, Article FSE102. Publication date: July 2026. FSE102:22 Zhensheng Xian, Zhen Yan, Yuanliang Chen, Xuelian Cao, Fuchen Ma, Dalong Shi, and Yu Jiang",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Craig Gentry. 2010. Computing arbitrary functions of encrypted data.Commun. ACM53, 3 (2010), 97–105.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Craig Gentry, Shai Halevi, and Nigel P Smart. 2012. Homomorphic evaluation of the AES circuit. InAnnual Cryptology Conference. Springer, 850–867.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Craig Gentry, Amit Sahai, and Brent Waters. 2013. Homomorphic encryption from learning with errors: Conceptuallysimpler, asymptotically-faster, attribute-based. InAnnual cryptology conference. Springer, 75–92.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Google. 2015. American Fuzzy Lop. https://github.com/google/AFL",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Google. 2019. Project Wycheproof. https://github.com/google/wycheproof.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Google. 2021. AddressSanitizer. https://clang.llvm.org/docs/AddressSanitizer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Google. 2021. UndefinedBehaviorSanitizer. https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Shai Halevi and Victor Shoup. 2021. Bootstrapping for helib.Journal of Cryptology34, 1 (2021), 7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Jia Cheng Han and Zhi Quan Zhou. 2020. Metamorphic Fuzz Testing of Autonomous Vehicles. InProceedings of the IEEE/ACM 42nd International Conference on Software Engineering Workshops(Seoul, Republic of Korea)(ICSEW’20). Association for Computing Machinery, New York, NY, USA, 380–385. doi:10.1145/3387940.3392252",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "HElib. 2023. HElib release v2.3.0. https://github.com/homenc/HElib/releases/tag/v2.3.0",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "homenc. 2025. HElib. https://github.com/homenc/HElib. Accessed: 2025-08-19.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. 2014. Compiler validation via equivalence modulo inputs. InProceedings of the 35th ACMSIGPLAN Conference on Programming Language Design and Implementation(Edinburgh, United Kingdom) (PLDI ’14). Association for Computing Machinery, New York, NY, USA, 216–226. doi:10.1145/2594291.2594334",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "LLVM. 2025. LibFuzzer. https://llvm.org/docs/LibFuzzer.html Accessed: 2025-09-09.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Microsoft. 2025. SymCrypt. https://github.com/microsoft/SymCrypt Accessed: 2025-09-11.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Charlie Miller, Zachary NJ Peterson, et al .2007. Analysis of mutation and generation-based fuzzing.Independent Security Evaluators, Tech. Rep4 (2007).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Mozilla Security Team. 2025. Peach Fuzzer. https://github.com/MozillaSecurity/peach Accessed: 2025-09-09.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Yannic Noller and Saeid Tizpaz-Niari. 2021. QFuzz: quantitative fuzzing for side channels. InProceedings of the 30th ACMSIGSOFT International Symposium on Software Testing and Analysis(Virtual, Denmark)(ISSTA 2021). Association for Computing Machinery, New York, NY, USA, 257–269. doi:10.1145/3460319.3464817",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "OpenFHE. 2024. OpenFHE release v1.2.1. https://github.com/openfheorg/openfhe-development/releases/tag/v1.2.1",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "OpenSSL. 2025. OpenSSL. https://github.com/openssl/openssl Accessed: 2025-09-11.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Yiteng Peng, Daoyuan Wu, Zhibo Liu, Dongwei Xiao, Zhenlan Ji, Juergen Rahmel, and Shuai Wang. 2025. Testing and understanding deviation behaviors in fhe-hardened machine learning models. InIEEE Computer Society. 644–644.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Perforce. 2025. Klocwork. https://www.perforce.com/products/klocwork",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. InProceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). USENIX Association, USA, Article 38, 16 pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 42,
+      "text": "SEAL 2023. Microsoft SEAL (release 4.1). https://github.com/Microsoft/SEAL. Microsoft Research, Redmond, WA..",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "SEAL. 2024. SEAL release v4.1.2. https://github.com/microsoft/SEAL/releases/tag/v4.1.2",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Ross Tate, Michael Stepp, Zachary Tatlock, and Sorin Lerner. 2009. Equality saturation: a new approach to optimization. InProceedings of the 36th annual ACMSIGPLAN-SIGACT symposium on Principles of programming languages. 264–276.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "OpenFHE Development Team. 2025. Addition for CKKSFLEXIBLE* modes at multiplicative depth = 0 returns incorrect results for batch size = 1 or 2. https://github.com/openfheorg/openfhe-development/issues/728 Accessed: 2025-08-19.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "TFHE. 2017. TFHE release v1.0.1. https://github.com/tfhe/tfhe/releases/tag/v1.0.1, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "András Vargha and Harold D Delaney. 2000. A critique and improvement of the CL common language effect size statistics of McGraw and Wong.Journal of Educational and Behavioral Statistics25, 2 (2000), 101–132.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Mayank Varia, Sophia Yakoubov, and Yang Yang. 2015. HEtest: A homomorphic encryption testing framework. In International Conference on Financial Cryptography and Data Security. Springer, 213–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Guido Vranken. 2025. Cryptofuzz. https://github.com/guidovranken/cryptofuzz Accessed: 2025-08-19.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Wikipedia contributors. 2025. Homomorphic encryption. https://en.wikipedia.org/wiki/Homomorphic_encryption Accessed: 2025-08-19.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Chijin Zhou, Bingzhou Qian, Gwihwan Go, Quan Zhang, Shanshan Li, and Yu Jiang. 2024. Polyjuice: Detecting miscompilation bugs in tensor compilers with equality saturation based rewriting.Proceedings of the ACM on Programming Languages8, OOPSLA2 (2024), 1309–1335.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Yuanhang Zhou, Fuchen Ma, Yuanliang Chen, Meng Ren, and Yu Jiang. 2023. CLFuzz: Vulnerability Detection of Cryptographic Algorithm&nbsp;Implementation via Semantic-aware Fuzzing.ACM Trans. Softw. Eng. Methodol.33, 2, Article 45 (Dec. 2023), 28 pages. doi:10.1145/3628160 Received 2026-02-24; accepted 2026-03-24 Proc. ACM Softw. Eng., Vol. 3, No. FSE, Article FSE102. Publication date: July 2026.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 41,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. InProceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). USENIX Association, USA, Article 38, 16 pages.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For instance, metamorphic fuzz testing [ 28] combines MT with fuzzing to detect faults in autonomous driving systems, QFuzz [ 36] uses execution costs for side-channel leakage quantification, and SQLancer [ 41] employs a containment oracle for DBMS logic bugs.",
+      "context_before": "mations are generally unconstrained, Eidolon must incorporate noise-awareness to ensure that equivalent transformations remain computationally valid within the strict noise of FHE. 8.3 Fuzzing Fuzzing is an effective technique for uncovering implementation bugs [ 34]. Generation-based tools such as Peach [ 35] construct inputs from grammars to ensure syntactic validity, while mutationbased tools such as AFL [ 23], AFL++[ 18], and LibFuzzer[ 32] use coverage-guided feedback to discover boundary conditions. Recent techniques combine specialized feedback signals or oracles in the fuzzing process.",
+      "context_after": "Eidolon distinguishes itself by utilizing residual noise as feedback and Equivalence Expression Transformation as its oracle to detect bugs in the noise-constrained FHE computational space. 9 Conclusion In this paper, we propose Eidolon, a noise-aware fuzzer for the vulnerability detection of FHE libraries. Eidolon first leverages noise as a feedback signal to guide the mutation of arithmetic expressions, enabling systematic exploration of deeper computational paths within the available noise. It then introduces an Equivalence Expression Transformation technique that generates mathematically",
+      "section": "8.3 Fuzzing",
+      "page": 20,
+      "char_offset": 69006,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:18:50Z",
+    "is_model_written": true,
+    "summary": "Eidolon is a fuzzer for fully homomorphic encryption libraries. Existing cryptographic testing tools generate structured inputs suited to cryptographic algorithms and are unaware of FHE noise management, so they fail to exercise computation on encrypted data. Eidolon directs mutations at arithmetic expressions within the noise budget and uses Equivalence Expression Transformation as its oracle, rewriting an expression into mathematically identical but structurally different forms. It found 20 previously unknown bugs, 12 assigned CVEs.",
+    "narrative": "One citation, in a survey of where metamorphic testing has been combined with fuzzing.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3810991_3811632.json b/_data/papers/paper_doi_10_1145_3810991_3811632.json
new file mode 100644
index 0000000..c55395b
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3810991_3811632.json
@@ -0,0 +1,1318 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T16:59:46Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3810991.3811632",
+    "title": "DIRT: Database-Integrated Random Testing",
+    "authors": [
+      "A. Keles",
+      "Ethan Chou",
+      "Harrison Goldstein",
+      "Leonidas Lampropoulos"
+    ],
+    "year": 2026,
+    "venue": "DBTest@SIGMOD",
+    "doi": "10.1145/3810991.3811632",
+    "arxiv_id": "2604.16373",
+    "s2_paper_id": "7f0ecde1aadab94fe2464b9a8e4183e68159d931",
+    "url": "https://doi.org/10.1145/3810991.3811632",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2604.16373",
+      "retrieved_at": "2026-09-07T16:59:46Z",
+      "chars": 42097,
+      "content_sha256": "sha256:9178e442303b16928be908a52992b6e82f0b2684eee07bcc15561716b03abd7a"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 8,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1577
+      },
+      {
+        "number": "2",
+        "title": "Background and Related Work",
+        "start": 4645
+      },
+      {
+        "number": "2.1",
+        "title": "Fuzzing and Property-Based Testing",
+        "start": 4864
+      },
+      {
+        "number": "2.2",
+        "title": "SQLancer",
+        "start": 6585
+      },
+      {
+        "number": "3",
+        "title": "Specifying Correctness Oracles",
+        "start": 8376
+      },
+      {
+        "number": "3.1",
+        "title": "Definitions of Oracles in DIRT",
+        "start": 13052
+      },
+      {
+        "number": "3.2",
+        "title": "Query Generation",
+        "start": 15312
+      },
+      {
+        "number": "4",
+        "title": "Evaluation",
+        "start": 18709
+      },
+      {
+        "number": "4.1",
+        "title": "RQ1: DoesDIRTfind bugs in Turso?",
+        "start": 18906
+      },
+      {
+        "number": "4.2",
+        "title": "RQ2: How does DIRTcompare to SQLancer?",
+        "start": 25212
+      },
+      {
+        "number": "5",
+        "title": "Discussion",
+        "start": 27139
+      },
+      {
+        "number": "6",
+        "title": "Conclusion and Future Work",
+        "start": 32929
+      }
+    ]
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Traditional testing tools like SQLancer and SQLSmith are highly effective for mature databases, but they struggle with high false positive rates and low actionability when applied to evolving systems.",
+      "context_before": "DIRT: Database-Integrated Random Testing Alperen Keles University of Maryland, College Park College Park, USA akeles@umd.eduEthan Chou University of Maryland, College Park College Park, USA echou1@umd.edu Harrison Goldstein University of Buffalo Buffalo, USA hgoldste@buffalo.eduLeonidas Lampropoulos University of Maryland, College Park College Park, USA leonidas@umd.edu Abstract Database management systems (DBMSs) are notoriously complex, making them difficult to test effectively, especially during early development when many features are incomplete.",
+      "context_after": "We present DIRT, a paradigm designed specifically for testing databases during development, which integrates a testing framework directly into the DBMS, enabling the random testing process to evolve in tandem with the system and reducing false positives by construction. We introduce generation actions, an abstraction for allowing database developers rather than testing experts to specify correctness properties. We evaluate DIRTon Turso, an actively developed SQLite-compatible OLTP engine, and show that it finds 23 unique, confirmed bugs–significantly outperforming off-the-shelf SQLancer varian",
+      "section": null,
+      "page": 1,
+      "char_offset": 556,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We evaluate DIRTon Turso, an actively developed SQLite-compatible OLTP engine, and show that it finds 23 unique, confirmed bugs–significantly outperforming off-the-shelf SQLancer variants in terms of true positive rate and usefulness of bug reports.",
+      "context_before": "ting tools like SQLancer and SQLSmith are highly effective for mature databases, but they struggle with high false positive rates and low actionability when applied to evolving systems. We present DIRT, a paradigm designed specifically for testing databases during development, which integrates a testing framework directly into the DBMS, enabling the random testing process to evolve in tandem with the system and reducing false positives by construction. We introduce generation actions, an abstraction for allowing database developers rather than testing experts to specify correctness properties.",
+      "context_after": "Our results demonstrate that embedding testing infrastructure within the DBMS can dramatically improve its effectiveness and usability during development. 1 Introduction Database Management Systems (DBMSs) are complex systems, and inevitably, such complexity leads to bugs. To discover bugs, DBMSs have historically used random testing, with approaches ranging from binary fuzzing with AFL [ 6] to structured query generation in SQLSmith [ 24], and to the powerful test oracles of SQLancer [ 22]. Within prior work, SQLancer is distinguished by using specialized test oracles that reveal logic bugs",
+      "section": null,
+      "page": 1,
+      "char_offset": 1172,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "To discover bugs, DBMSs have historically used random testing, with approaches ranging from binary fuzzing with AFL [ 6] to structured query generation in SQLSmith [ 24], and to the powerful test oracles of SQLancer [ 22].",
+      "context_before": "ase developers rather than testing experts to specify correctness properties. We evaluate DIRTon Turso, an actively developed SQLite-compatible OLTP engine, and show that it finds 23 unique, confirmed bugs–significantly outperforming off-the-shelf SQLancer variants in terms of true positive rate and usefulness of bug reports. Our results demonstrate that embedding testing infrastructure within the DBMS can dramatically improve its effectiveness and usability during development. 1 Introduction Database Management Systems (DBMSs) are complex systems, and inevitably, such complexity leads to bugs.",
+      "context_after": "Within prior work, SQLancer is distinguished by using specialized test oracles that reveal logic bugs in DBMSs, and it has proven extremely successful at finding bugs in mature databases. Unfortunately, off-the-shelf testing frameworks are not as effective at finding actionable bugs in DBMSs that are actively being developed and have many missing features. When the space of randomly generated inputs is larger than the system under test supports, many of the generated tests lie in the unimplemented portion of the system, registering as false positives. Naturally, such false counterexamples are",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 1695,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Within prior work, SQLancer is distinguished by using specialized test oracles that reveal logic bugs in DBMSs, and it has proven extremely successful at finding bugs in mature databases.",
+      "context_before": "erforming off-the-shelf SQLancer variants in terms of true positive rate and usefulness of bug reports. Our results demonstrate that embedding testing infrastructure within the DBMS can dramatically improve its effectiveness and usability during development. 1 Introduction Database Management Systems (DBMSs) are complex systems, and inevitably, such complexity leads to bugs. To discover bugs, DBMSs have historically used random testing, with approaches ranging from binary fuzzing with AFL [ 6] to structured query generation in SQLSmith [ 24], and to the powerful test oracles of SQLancer [ 22].",
+      "context_after": "Unfortunately, off-the-shelf testing frameworks are not as effective at finding actionable bugs in DBMSs that are actively being developed and have many missing features. When the space of randomly generated inputs is larger than the system under test supports, many of the generated tests lie in the unimplemented portion of the system, registering as false positives. Naturally, such false counterexamples are of no help to developers of DBMSs. One way to alleviate this issue is to tailor the testing framework to the DBMS under test, ensuring that the testing framework only tests implemented fe",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 1918,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "5% with only one new bug, and running SQLancerTurso, an existing independently developed SQLancer integration for Turso, without modifications led to a false positive rate of 58% with six newly discovered bugs.",
+      "context_before": "e this tight integration to empower developers to write their own oracles by developing a domain-specific language (DSL) for expressing properties customized to particular features being developed. We study this problem in the context of Turso, an open-source SQLite3-compatible OLTP database engine in active development. With an average of more than 30 commits a day, Turso is a rapidly evolving target for testing.DIRThelped find 23 confirmed bugs, all of which were subsequently fixed. In comparison, running SQLancerSQLite3 with minimal modifications in Turso led to a false positive rate of 96.",
+      "context_after": "In summary, our contributions are as follows: (1)We propose DIRT, a paradigm for testing databases during development which integrates a testing framework within the database and provides convenient and flexible abstractions to developers. (2)We implement DIRTfor Turso, an open source SQLite3compatible OLTP database under active development. (3)We evaluate the effectiveness of DIRTby finding actionable bugs in Turso: we found 23 unique confirmed bugs with very few false positives compared to the state-of-the-art. 2 Background and Related Work We begin by providing background: first on coverage-g",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3918,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "2 Background and Related Work We begin by providing background: first on coverage-guided fuzzing and property-based testing at a high level, and then focusing on SQLancer, the prominent framework for testing databases.",
+      "context_before": "ut modifications led to a false positive rate of 58% with six newly discovered bugs. In summary, our contributions are as follows: (1)We propose DIRT, a paradigm for testing databases during development which integrates a testing framework within the database and provides convenient and flexible abstractions to developers. (2)We implement DIRTfor Turso, an open source SQLite3compatible OLTP database under active development. (3)We evaluate the effectiveness of DIRTby finding actionable bugs in Turso: we found 23 unique confirmed bugs with very few false positives compared to the state-of-the-art.",
+      "context_after": "2.1 Fuzzing and Property-Based Testing In Fuzz Testing [ 25], the primary focus is on smart and effective test-case generation that can work out of the box with minimal user inputs. The key idea is to leverageruntime feedback, usually in the form of branch coverage [ 4,6], though not always [ 19], to keep track of inputs that exhibit interesting behavior (e.g. uncovered new paths) and then mutate them in the hopes of discovering evenarXiv:2604.16373v1 [cs.DB] 23 Mar 2026 Conference’17, July 2017, Washington, DC, USA Alperen Keles, Ethan Chou, Harrison Goldstein, and Leonidas Lampropoulos more",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 4644,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Such approaches have been tried recently with great success [ 14,18], and in the context of database testing, both SQLancer and DIRTfollow such a hybrid viewpoint.",
+      "context_before": "wering users to write their own oracles [ 5,12]. PBT frameworks usually offer DSLs for expressing oracles in the form of universally quantified executable predicates, as well as infrastructure for writinggeneratorsprograms that generate test inputs to test such predicates [ 5,15]. Although traditionally these communities have been mostly distinct, the lines between them are increasingly blurred. In fact, fuzzing can be viewed as an instance of PBT where the property is fixed (e.g. the program does not crash) and PBT can be improved by extending its generation with coverage-guided capabilities.",
+      "context_after": "2.2 SQLancer SQLancer is a multi-year research project by Rigger et. al that set the bar for automated random testing of databases since its inception in 2020. It hosts an extensible core with adapters for many production DBMSs including but not limited to ClickHouse, Apache Datafusion, MySQL, PostgreSQL, and SQLite. Within the last five years, SQLancer has not only grown with respect to the breadth of databases it supports, but it has also widened its arsenal of oracles. It started with Pivoted Query Synthesis (PQS) [ 23], a rather \"simple\" containment property over databases that has found",
+      "section": "2.1 Fuzzing and Property-Based Testing",
+      "page": 2,
+      "char_offset": 6421,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "2 SQLancer SQLancer is a multi-year research project by Rigger et.",
+      "context_before": "as well as infrastructure for writinggeneratorsprograms that generate test inputs to test such predicates [ 5,15]. Although traditionally these communities have been mostly distinct, the lines between them are increasingly blurred. In fact, fuzzing can be viewed as an instance of PBT where the property is fixed (e.g. the program does not crash) and PBT can be improved by extending its generation with coverage-guided capabilities. Such approaches have been tried recently with great success [ 14,18], and in the context of database testing, both SQLancer and DIRTfollow such a hybrid viewpoint. 2.",
+      "context_after": "al that set the bar for automated random testing of databases since its inception in 2020. It hosts an extensible core with adapters for many production DBMSs including but not limited to ClickHouse, Apache Datafusion, MySQL, PostgreSQL, and SQLite. Within the last five years, SQLancer has not only grown with respect to the breadth of databases it supports, but it has also widened its arsenal of oracles. It started with Pivoted Query Synthesis (PQS) [ 23], a rather \"simple\" containment property over databases that has found at least 121 unique logic bugs in production databases. Two metamorph",
+      "section": "2.2 SQLancer",
+      "page": 2,
+      "char_offset": 6587,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Within the last five years, SQLancer has not only grown with respect to the breadth of databases it supports, but it has also widened its arsenal of oracles.",
+      "context_before": ".g. the program does not crash) and PBT can be improved by extending its generation with coverage-guided capabilities. Such approaches have been tried recently with great success [ 14,18], and in the context of database testing, both SQLancer and DIRTfollow such a hybrid viewpoint. 2.2 SQLancer SQLancer is a multi-year research project by Rigger et. al that set the bar for automated random testing of databases since its inception in 2020. It hosts an extensible core with adapters for many production DBMSs including but not limited to ClickHouse, Apache Datafusion, MySQL, PostgreSQL, and SQLite.",
+      "context_after": "It started with Pivoted Query Synthesis (PQS) [ 23], a rather \"simple\" containment property over databases that has found at least 121 unique logic bugs in production databases. Two metamorphic oracles followed: Non-Optimizing Reference Engine Construction (NoREC) [ 20], which found 51 optimization bugs, and Ternary Logic Partitioning (TLP) [21], which discovered 77 novel logic bugs. SQLancer currently supports Query Plan Guidance (QPG) [ 1] for feedback-guided generation, Cardinality Estimation Restriction Testing (CERT) [ 2] for finding performance bugs in DBMSs, Differential Query Plans (D",
+      "section": "2.2 SQLancer",
+      "page": 2,
+      "char_offset": 6903,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": "It started with Pivoted Query Synthesis (PQS) [ 23], a rather \"simple\" containment property over databases that has found at least 121 unique logic bugs in production databases.",
+      "context_before": "y with great success [ 14,18], and in the context of database testing, both SQLancer and DIRTfollow such a hybrid viewpoint. 2.2 SQLancer SQLancer is a multi-year research project by Rigger et. al that set the bar for automated random testing of databases since its inception in 2020. It hosts an extensible core with adapters for many production DBMSs including but not limited to ClickHouse, Apache Datafusion, MySQL, PostgreSQL, and SQLite. Within the last five years, SQLancer has not only grown with respect to the breadth of databases it supports, but it has also widened its arsenal of oracles.",
+      "context_after": "Two metamorphic oracles followed: Non-Optimizing Reference Engine Construction (NoREC) [ 20], which found 51 optimization bugs, and Ternary Logic Partitioning (TLP) [21], which discovered 77 novel logic bugs. SQLancer currently supports Query Plan Guidance (QPG) [ 1] for feedback-guided generation, Cardinality Estimation Restriction Testing (CERT) [ 2] for finding performance bugs in DBMSs, Differential Query Plans (DQP) [ 3] for detecting bugs in join optimizations, and Constant Optimization Driven Database System Testing (CODDTest) [ 26] for finding logic bugs in optimizations. As SQLancer",
+      "section": "2.2 SQLancer",
+      "page": 2,
+      "char_offset": 7061,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "Non-Optimizing Reference Engine Construction",
+      "technique": "norec",
+      "sentence": "Two metamorphic oracles followed: Non-Optimizing Reference Engine Construction (NoREC) [ 20], which found 51 optimization bugs, and Ternary Logic Partitioning (TLP) [21], which discovered 77 novel logic bugs.",
+      "context_before": "by Rigger et. al that set the bar for automated random testing of databases since its inception in 2020. It hosts an extensible core with adapters for many production DBMSs including but not limited to ClickHouse, Apache Datafusion, MySQL, PostgreSQL, and SQLite. Within the last five years, SQLancer has not only grown with respect to the breadth of databases it supports, but it has also widened its arsenal of oracles. It started with Pivoted Query Synthesis (PQS) [ 23], a rather \"simple\" containment property over databases that has found at least 121 unique logic bugs in production databases.",
+      "context_after": "SQLancer currently supports Query Plan Guidance (QPG) [ 1] for feedback-guided generation, Cardinality Estimation Restriction Testing (CERT) [ 2] for finding performance bugs in DBMSs, Differential Query Plans (DQP) [ 3] for detecting bugs in join optimizations, and Constant Optimization Driven Database System Testing (CODDTest) [ 26] for finding logic bugs in optimizations. As SQLancer focuses on testing large classes of behaviors across a variety of databases, each oracle amounts to a significant research contribution in a new research paper. Integrating a new DBMS to SQLancer is a time-con",
+      "section": "2.2 SQLancer",
+      "page": 2,
+      "char_offset": 7239,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer currently supports Query Plan Guidance (QPG) [ 1] for feedback-guided generation, Cardinality Estimation Restriction Testing (CERT) [ 2] for finding performance bugs in DBMSs, Differential Query Plans (DQP) [ 3] for detecting bugs in join optimizations, and Constant Optimization Driven Database System Testing (CODDTest) [ 26] for finding logic bugs in optimizations.",
+      "context_before": "ouse, Apache Datafusion, MySQL, PostgreSQL, and SQLite. Within the last five years, SQLancer has not only grown with respect to the breadth of databases it supports, but it has also widened its arsenal of oracles. It started with Pivoted Query Synthesis (PQS) [ 23], a rather \"simple\" containment property over databases that has found at least 121 unique logic bugs in production databases. Two metamorphic oracles followed: Non-Optimizing Reference Engine Construction (NoREC) [ 20], which found 51 optimization bugs, and Ternary Logic Partitioning (TLP) [21], which discovered 77 novel logic bugs.",
+      "context_after": "As SQLancer focuses on testing large classes of behaviors across a variety of databases, each oracle amounts to a significant research contribution in a new research paper. Integrating a new DBMS to SQLancer is a time-consuming process. At a minimum, SQLancer integration requires implementing AST connectors, generation APIs for the relevant queries, and the oracles to use for detecting bugs. SQLancer also has a notion ofexpected errors, bugs that the users can deem as expected, so SQLancer does not report them as errors and continues testing. 3 Specifying Correctness Oracles Rather than expec",
+      "section": "2.2 SQLancer",
+      "page": 2,
+      "char_offset": 7448,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "qpg",
+        "cert",
+        "dqp",
+        "coddtest"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As SQLancer focuses on testing large classes of behaviors across a variety of databases, each oracle amounts to a significant research contribution in a new research paper.",
+      "context_before": "on databases. Two metamorphic oracles followed: Non-Optimizing Reference Engine Construction (NoREC) [ 20], which found 51 optimization bugs, and Ternary Logic Partitioning (TLP) [21], which discovered 77 novel logic bugs. SQLancer currently supports Query Plan Guidance (QPG) [ 1] for feedback-guided generation, Cardinality Estimation Restriction Testing (CERT) [ 2] for finding performance bugs in DBMSs, Differential Query Plans (DQP) [ 3] for detecting bugs in join optimizations, and Constant Optimization Driven Database System Testing (CODDTest) [ 26] for finding logic bugs in optimizations.",
+      "context_after": "Integrating a new DBMS to SQLancer is a time-consuming process. At a minimum, SQLancer integration requires implementing AST connectors, generation APIs for the relevant queries, and the oracles to use for detecting bugs. SQLancer also has a notion ofexpected errors, bugs that the users can deem as expected, so SQLancer does not report them as errors and continues testing. 3 Specifying Correctness Oracles Rather than expecting database developers to modify SQLancer to suit their particular needs, we instead wanted to offer them flexible and extensible abstractions so that they test their data",
+      "section": "2.2 SQLancer",
+      "page": 2,
+      "char_offset": 7826,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Integrating a new DBMS to SQLancer is a time-consuming process.",
+      "context_before": "(TLP) [21], which discovered 77 novel logic bugs. SQLancer currently supports Query Plan Guidance (QPG) [ 1] for feedback-guided generation, Cardinality Estimation Restriction Testing (CERT) [ 2] for finding performance bugs in DBMSs, Differential Query Plans (DQP) [ 3] for detecting bugs in join optimizations, and Constant Optimization Driven Database System Testing (CODDTest) [ 26] for finding logic bugs in optimizations. As SQLancer focuses on testing large classes of behaviors across a variety of databases, each oracle amounts to a significant research contribution in a new research paper.",
+      "context_after": "At a minimum, SQLancer integration requires implementing AST connectors, generation APIs for the relevant queries, and the oracles to use for detecting bugs. SQLancer also has a notion ofexpected errors, bugs that the users can deem as expected, so SQLancer does not report them as errors and continues testing. 3 Specifying Correctness Oracles Rather than expecting database developers to modify SQLancer to suit their particular needs, we instead wanted to offer them flexible and extensible abstractions so that they test their database throughout its development.DIRTwas created to provide such",
+      "section": "2.2 SQLancer",
+      "page": 2,
+      "char_offset": 7999,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "At a minimum, SQLancer integration requires implementing AST connectors, generation APIs for the relevant queries, and the oracles to use for detecting bugs.",
+      "context_before": "ntly supports Query Plan Guidance (QPG) [ 1] for feedback-guided generation, Cardinality Estimation Restriction Testing (CERT) [ 2] for finding performance bugs in DBMSs, Differential Query Plans (DQP) [ 3] for detecting bugs in join optimizations, and Constant Optimization Driven Database System Testing (CODDTest) [ 26] for finding logic bugs in optimizations. As SQLancer focuses on testing large classes of behaviors across a variety of databases, each oracle amounts to a significant research contribution in a new research paper. Integrating a new DBMS to SQLancer is a time-consuming process.",
+      "context_after": "SQLancer also has a notion ofexpected errors, bugs that the users can deem as expected, so SQLancer does not report them as errors and continues testing. 3 Specifying Correctness Oracles Rather than expecting database developers to modify SQLancer to suit their particular needs, we instead wanted to offer them flexible and extensible abstractions so that they test their database throughout its development.DIRTwas created to provide such random testing infrastructure that evolves alongside the DBMS and produces actionable bug reports for the database developers. Thisevolution involves not only",
+      "section": "2.2 SQLancer",
+      "page": 2,
+      "char_offset": 8063,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer also has a notion ofexpected errors, bugs that the users can deem as expected, so SQLancer does not report them as errors and continues testing.",
+      "context_before": "gs in DBMSs, Differential Query Plans (DQP) [ 3] for detecting bugs in join optimizations, and Constant Optimization Driven Database System Testing (CODDTest) [ 26] for finding logic bugs in optimizations. As SQLancer focuses on testing large classes of behaviors across a variety of databases, each oracle amounts to a significant research contribution in a new research paper. Integrating a new DBMS to SQLancer is a time-consuming process. At a minimum, SQLancer integration requires implementing AST connectors, generation APIs for the relevant queries, and the oracles to use for detecting bugs.",
+      "context_after": "3 Specifying Correctness Oracles Rather than expecting database developers to modify SQLancer to suit their particular needs, we instead wanted to offer them flexible and extensible abstractions so that they test their database throughout its development.DIRTwas created to provide such random testing infrastructure that evolves alongside the DBMS and produces actionable bug reports for the database developers. Thisevolution involves not only tailoring the input space of SQL generation to the currently implemented portion of Turso for reducing false positives, but also tailoring the correctnes",
+      "section": "2.2 SQLancer",
+      "page": 2,
+      "char_offset": 8221,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3 Specifying Correctness Oracles Rather than expecting database developers to modify SQLancer to suit their particular needs, we instead wanted to offer them flexible and extensible abstractions so that they test their database throughout its development.",
+      "context_before": "est) [ 26] for finding logic bugs in optimizations. As SQLancer focuses on testing large classes of behaviors across a variety of databases, each oracle amounts to a significant research contribution in a new research paper. Integrating a new DBMS to SQLancer is a time-consuming process. At a minimum, SQLancer integration requires implementing AST connectors, generation APIs for the relevant queries, and the oracles to use for detecting bugs. SQLancer also has a notion ofexpected errors, bugs that the users can deem as expected, so SQLancer does not report them as errors and continues testing.",
+      "context_after": "DIRTwas created to provide such random testing infrastructure that evolves alongside the DBMS and produces actionable bug reports for the database developers. Thisevolution involves not only tailoring the input space of SQL generation to the currently implemented portion of Turso for reducing false positives, but also tailoring the correctness oracles to new features as they are added, decreasing false negatives. In this section, we introducegeneration actions, an imperative formulation for properties that do not just describewhatthe property tests, buthowto test it. We begin by using a very s",
+      "section": "2.2 SQLancer",
+      "page": 2,
+      "char_offset": 8375,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": "assert(r in RS) Figure 1: Pivoted Query Synthesis as a universally quantified property (left) and as a generation action (right).",
+      "context_before": ", USA 𝑑𝑏:𝐷𝑎𝑡𝑎𝑏𝑎𝑠𝑒 𝑡1,𝑡2...𝑡𝑛:𝑇𝑎𝑏𝑙𝑒 𝑐1,𝑐2...𝑐𝑛:𝐶𝑜𝑙𝑢𝑚𝑛 𝑝1,𝑝2...𝑝𝑛:𝐸𝑥𝑝𝑟𝑒𝑠𝑠𝑖𝑜𝑛 𝑟:𝑅𝑜𝑤 ∀𝑑𝑏,𝑡 1,𝑡2...𝑡𝑛,𝑐1,𝑐2...𝑐𝑛,𝑝1,𝑝2...𝑝𝑛,𝑟. 𝑡1,𝑡2...𝑡𝑛∈𝑑𝑏 ∧𝑐 1∈𝑡 1∧𝑐 2∈𝑡 2...∧𝑐 𝑛∈𝑡𝑛 ∧𝑟.𝑐 1∈𝑡 1.𝑐1∧𝑟.𝑐 2∈𝑡 2.𝑐2...∧𝑟.𝑐 𝑛∈𝑡𝑛.𝑐𝑛 ∧𝑝 1(𝑟)=𝑇𝑅𝑈𝐸∧𝑝 2(𝑟)=𝑇𝑅𝑈𝐸...∧𝑝 𝑛(𝑟)=𝑇𝑅𝑈𝐸 =⇒𝑟∈𝑆𝐸𝐿𝐸𝐶𝑇∗𝐹𝑅𝑂𝑀 𝑡 1.𝑐1...𝑡𝑛.𝑐𝑛𝑊𝐻𝐸𝑅𝐸𝑝 1...𝐴𝑁𝐷 𝑝 𝑛gen property db: (t1, t2)←(pick db.tables, pick db.tables) (c1, c2)←(pick t1.columns, pick t2.columns) (r1, r2)←(gen row t1, gen row t2) r := (r1.c1, r2.c2)! INSERT INTO t1 VALUES r1! INSERT INTO t2 VALUES r2 p1←gen expression (t1, r1) p2←gen expression (t2, r2)! RS := SELECT r FROM t1, t2 WHERE p1 AND p2!",
+      "context_after": "DIRTas part of the DSL for generation actions, as exemplified by the header initialization bug studied in §4.1. 3.1 Definitions of Oracles in DIRT In this subsection, we go over definitions of five different oracles written as sequences of generation actions, three of which are reimplementations of existing oracles in SQLancer. We start by defining the first of the three SQLancer oracles we implemented, Pivoted Query Synthesis (PQS) [ 23], as a universally quantified proposition in Fig. 1. PQS states that given a set of tables in the database, SELECT ing for a row constructed from the contents",
+      "section": "3 Specifying Correctness Oracles",
+      "page": 3,
+      "char_offset": 12809,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "1 Definitions of Oracles in DIRT In this subsection, we go over definitions of five different oracles written as sequences of generation actions, three of which are reimplementations of existing oracles in SQLancer.",
+      "context_before": "=⇒𝑟∈𝑆𝐸𝐿𝐸𝐶𝑇∗𝐹𝑅𝑂𝑀 𝑡 1.𝑐1...𝑡𝑛.𝑐𝑛𝑊𝐻𝐸𝑅𝐸𝑝 1...𝐴𝑁𝐷 𝑝 𝑛gen property db: (t1, t2)←(pick db.tables, pick db.tables) (c1, c2)←(pick t1.columns, pick t2.columns) (r1, r2)←(gen row t1, gen row t2) r := (r1.c1, r2.c2)! INSERT INTO t1 VALUES r1! INSERT INTO t2 VALUES r2 p1←gen expression (t1, r1) p2←gen expression (t2, r2)! RS := SELECT r FROM t1, t2 WHERE p1 AND p2! assert(r in RS) Figure 1: Pivoted Query Synthesis as a universally quantified property (left) and as a generation action (right). DIRTas part of the DSL for generation actions, as exemplified by the header initialization bug studied in §4.1. 3.",
+      "context_after": "We start by defining the first of the three SQLancer oracles we implemented, Pivoted Query Synthesis (PQS) [ 23], as a universally quantified proposition in Fig. 1. PQS states that given a set of tables in the database, SELECT ing for a row constructed from the contents of those tables should contain the row. Fig. 1 shows two formulations of PQS side by side, the propositional formulation with universally quantified variables that define what PQS is on the left and a GA formulation of PQS with two tables/columns as generation actions that definehow PQS is tested on the right, modeling the imple",
+      "section": "3.1 Definitions of Oracles in DIRT",
+      "page": 3,
+      "char_offset": 13054,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We start by defining the first of the three SQLancer oracles we implemented, Pivoted Query Synthesis (PQS) [ 23], as a universally quantified proposition in Fig.",
+      "context_before": "NTO t1 VALUES r1! INSERT INTO t2 VALUES r2 p1←gen expression (t1, r1) p2←gen expression (t2, r2)! RS := SELECT r FROM t1, t2 WHERE p1 AND p2! assert(r in RS) Figure 1: Pivoted Query Synthesis as a universally quantified property (left) and as a generation action (right). DIRTas part of the DSL for generation actions, as exemplified by the header initialization bug studied in §4.1. 3.1 Definitions of Oracles in DIRT In this subsection, we go over definitions of five different oracles written as sequences of generation actions, three of which are reimplementations of existing oracles in SQLancer.",
+      "context_after": "1. PQS states that given a set of tables in the database, SELECT ing for a row constructed from the contents of those tables should contain the row. Fig. 1 shows two formulations of PQS side by side, the propositional formulation with universally quantified variables that define what PQS is on the left and a GA formulation of PQS with two tables/columns as generation actions that definehow PQS is tested on the right, modeling the implementation of PQS that tests Turso today. The propositional formulation quantifies over a database and a sequence of tables, columns, and expressions, under the co",
+      "section": "3.1 Definitions of Oracles in DIRT",
+      "page": 3,
+      "char_offset": 13268,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS states that given a set of tables in the database, SELECT ing for a row constructed from the contents of those tables should contain the row.",
+      "context_before": "1: Pivoted Query Synthesis as a universally quantified property (left) and as a generation action (right). DIRTas part of the DSL for generation actions, as exemplified by the header initialization bug studied in §4.1. 3.1 Definitions of Oracles in DIRT In this subsection, we go over definitions of five different oracles written as sequences of generation actions, three of which are reimplementations of existing oracles in SQLancer. We start by defining the first of the three SQLancer oracles we implemented, Pivoted Query Synthesis (PQS) [ 23], as a universally quantified proposition in Fig. 1.",
+      "context_after": "Fig. 1 shows two formulations of PQS side by side, the propositional formulation with universally quantified variables that define what PQS is on the left and a GA formulation of PQS with two tables/columns as generation actions that definehow PQS is tested on the right, modeling the implementation of PQS that tests Turso today. The propositional formulation quantifies over a database and a sequence of tables, columns, and expressions, under the constraint that the expressions will return 𝑇𝑅𝑈𝐸 when tested against a row 𝑟. The corresponding GA closely follows along, but ensures these constraints",
+      "section": "3.1 Definitions of Oracles in DIRT",
+      "page": 3,
+      "char_offset": 13433,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "1 shows two formulations of PQS side by side, the propositional formulation with universally quantified variables that define what PQS is on the left and a GA formulation of PQS with two tables/columns as generation actions that definehow PQS is tested on the right, modeling the implementation of PQS that tests Turso today.",
+      "context_before": "s, as exemplified by the header initialization bug studied in §4.1. 3.1 Definitions of Oracles in DIRT In this subsection, we go over definitions of five different oracles written as sequences of generation actions, three of which are reimplementations of existing oracles in SQLancer. We start by defining the first of the three SQLancer oracles we implemented, Pivoted Query Synthesis (PQS) [ 23], as a universally quantified proposition in Fig. 1. PQS states that given a set of tables in the database, SELECT ing for a row constructed from the contents of those tables should contain the row. Fig.",
+      "context_after": "The propositional formulation quantifies over a database and a sequence of tables, columns, and expressions, under the constraint that the expressions will return 𝑇𝑅𝑈𝐸 when tested against a row 𝑟. The corresponding GA closely follows along, but ensures these constraints are satisfied by construction. We start by picking tables t1, t2 from the state db.tables, followed by picking columns from the respective table. gen row t and gen expression (t, r) are dependent generation primitives, where the former generates a row based on the table t, and the latter generates an expression that will evalu",
+      "section": "3.1 Definitions of Oracles in DIRT",
+      "page": 3,
+      "char_offset": 13584,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "2, we provide definitions of SQLancer oracles as GAs in addition to the other oracles we implemented for DIRT.",
+      "context_before": "s over a database and a sequence of tables, columns, and expressions, under the constraint that the expressions will return 𝑇𝑅𝑈𝐸 when tested against a row 𝑟. The corresponding GA closely follows along, but ensures these constraints are satisfied by construction. We start by picking tables t1, t2 from the state db.tables, followed by picking columns from the respective table. gen row t and gen expression (t, r) are dependent generation primitives, where the former generates a row based on the table t, and the latter generates an expression that will evaluate to TRUEfor rowrof the tablet. In Fig.",
+      "context_after": "Fig. 2a demonstrates how to write Non-Optimizing Reference Engine Construction Generation (NoREC) [ 20] as a GA. Fig. 2b presents the GA for WHERE Extended case of Ternary Logic Partitioning (TLP) [ 21] oracle from SQLancer. We can express other properties that are not present in SQLancer, including those fundamental to key-value stores such asDeleted rows should not be in the tablepresented in Fig. 2c. Properties, depending on how they are constructed, mightsometimes be invalid. In Fig. 2d, applying WHERE or UNION ALL operation for composing the results of multiple WHERE queries require both s",
+      "section": "3.1 Definitions of Oracles in DIRT",
+      "page": 3,
+      "char_offset": 14548,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "technique",
+      "surface": "Non-Optimizing Reference Engine Construction",
+      "technique": "norec",
+      "sentence": "2a demonstrates how to write Non-Optimizing Reference Engine Construction Generation (NoREC) [ 20] as a GA.",
+      "context_before": "l return 𝑇𝑅𝑈𝐸 when tested against a row 𝑟. The corresponding GA closely follows along, but ensures these constraints are satisfied by construction. We start by picking tables t1, t2 from the state db.tables, followed by picking columns from the respective table. gen row t and gen expression (t, r) are dependent generation primitives, where the former generates a row based on the table t, and the latter generates an expression that will evaluate to TRUEfor rowrof the tablet. In Fig. 2, we provide definitions of SQLancer oracles as GAs in addition to the other oracles we implemented for DIRT. Fig.",
+      "context_after": "Fig. 2b presents the GA for WHERE Extended case of Ternary Logic Partitioning (TLP) [ 21] oracle from SQLancer. We can express other properties that are not present in SQLancer, including those fundamental to key-value stores such asDeleted rows should not be in the tablepresented in Fig. 2c. Properties, depending on how they are constructed, mightsometimes be invalid. In Fig. 2d, applying WHERE or UNION ALL operation for composing the results of multiple WHERE queries require both sides of the WHEREs to have the same number of columns. 3.2 Query Generation We designed our query generation algor",
+      "section": "3.1 Definitions of Oracles in DIRT",
+      "page": 3,
+      "char_offset": 14663,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "2b presents the GA for WHERE Extended case of Ternary Logic Partitioning (TLP) [ 21] oracle from SQLancer.",
+      "context_before": "nts are satisfied by construction. We start by picking tables t1, t2 from the state db.tables, followed by picking columns from the respective table. gen row t and gen expression (t, r) are dependent generation primitives, where the former generates a row based on the table t, and the latter generates an expression that will evaluate to TRUEfor rowrof the tablet. In Fig. 2, we provide definitions of SQLancer oracles as GAs in addition to the other oracles we implemented for DIRT. Fig. 2a demonstrates how to write Non-Optimizing Reference Engine Construction Generation (NoREC) [ 20] as a GA. Fig.",
+      "context_after": "We can express other properties that are not present in SQLancer, including those fundamental to key-value stores such asDeleted rows should not be in the tablepresented in Fig. 2c. Properties, depending on how they are constructed, mightsometimes be invalid. In Fig. 2d, applying WHERE or UNION ALL operation for composing the results of multiple WHERE queries require both sides of the WHEREs to have the same number of columns. 3.2 Query Generation We designed our query generation algorithm with three goals in mind. First, we wanted each automatically generated database interaction to respect da",
+      "section": "3.1 Definitions of Oracles in DIRT",
+      "page": 3,
+      "char_offset": 14776,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We can express other properties that are not present in SQLancer, including those fundamental to key-value stores such asDeleted rows should not be in the tablepresented in Fig.",
+      "context_before": "picking columns from the respective table. gen row t and gen expression (t, r) are dependent generation primitives, where the former generates a row based on the table t, and the latter generates an expression that will evaluate to TRUEfor rowrof the tablet. In Fig. 2, we provide definitions of SQLancer oracles as GAs in addition to the other oracles we implemented for DIRT. Fig. 2a demonstrates how to write Non-Optimizing Reference Engine Construction Generation (NoREC) [ 20] as a GA. Fig. 2b presents the GA for WHERE Extended case of Ternary Logic Partitioning (TLP) [ 21] oracle from SQLancer.",
+      "context_after": "2c. Properties, depending on how they are constructed, mightsometimes be invalid. In Fig. 2d, applying WHERE or UNION ALL operation for composing the results of multiple WHERE queries require both sides of the WHEREs to have the same number of columns. 3.2 Query Generation We designed our query generation algorithm with three goals in mind. First, we wanted each automatically generated database interaction to respect database state (e.g. we should not select from a table that does not exist unless a user-written generation action explicitly calls for it). Second, we did not want to rely on quer",
+      "section": "3.1 Definitions of Oracles in DIRT",
+      "page": 3,
+      "char_offset": 14882,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "count(1)) (a) Non-Optimizing Reference Engine Construction (NoREC)gen property db: t←pick db.",
+      "context_before": "(𝑠𝑡𝑎𝑡𝑒)⊲Update the shadow state 𝑖𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛𝑠.𝑝𝑢𝑠ℎ(𝑖) 𝑁←𝑁−1 end while Database developers can specify three parameters, 𝑅,𝑊, and𝐶, describing the proportion of read, write, and create instructions that should appear in the generated interactions. The algorithm iteratively generates one or more interactions compatible with the Conference’17, July 2017, Washington, DC, USA Alperen Keles, Ethan Chou, Harrison Goldstein, and Leonidas Lampropoulos gen property db: t←pick db.tables r←pick t.rows p←gen expression (t, r)! RS1 = SELECT * FROM t WHERE p! RS2 = SELECT p FROM t! assert(RS1.length() == RS2.",
+      "context_after": "tables p←gen expression (t) p'←gen expression! RS1 = SELECT * FROM t WHERE p! RS2 = SELECT * FROM t WHERE p AND p'UNION ALL SELECT * FROM t WHERE p AND (NOT p') UNION ALL SELECT * FROM t WHERE p AND (p'is NULL)! assert(RS1.length() == RS2.count(1)) (b) Ternary Logic Partitioning (TLP) WHERE Extended gen property db: t←pick db.tables r←pick t.rows p←gen expression (t, r)! DELETE * FROM t WHERE P! RS = SELECT * FROM t WHERE p! assert(r not in RS) (c) \"Deletes rows should not be in the table\"gen property db: s1←gen SELECT db.tables s2←gen SELECT db.tables! RS3 = s1 UNION ALL s2! assume(s1.first()",
+      "section": "3.2 Query Generation",
+      "page": 4,
+      "char_offset": 17207,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "count(1)) (b) Ternary Logic Partitioning (TLP) WHERE Extended gen property db: t←pick db.",
+      "context_before": "rence’17, July 2017, Washington, DC, USA Alperen Keles, Ethan Chou, Harrison Goldstein, and Leonidas Lampropoulos gen property db: t←pick db.tables r←pick t.rows p←gen expression (t, r)! RS1 = SELECT * FROM t WHERE p! RS2 = SELECT p FROM t! assert(RS1.length() == RS2.count(1)) (a) Non-Optimizing Reference Engine Construction (NoREC)gen property db: t←pick db.tables p←gen expression (t) p'←gen expression! RS1 = SELECT * FROM t WHERE p! RS2 = SELECT * FROM t WHERE p AND p'UNION ALL SELECT * FROM t WHERE p AND (NOT p') UNION ALL SELECT * FROM t WHERE p AND (p'is NULL)! assert(RS1.length() == RS2.",
+      "context_after": "tables r←pick t.rows p←gen expression (t, r)! DELETE * FROM t WHERE P! RS = SELECT * FROM t WHERE p! assert(r not in RS) (c) \"Deletes rows should not be in the table\"gen property db: s1←gen SELECT db.tables s2←gen SELECT db.tables! RS3 = s1 UNION ALL s2! assume(s1.first().length() == s1.first().length())! assert(RS1.length() + RS2.length() == RS3.length()) (d) \"UNION ALL preserves cardinality\" Figure 2: Definitions of Oracles as Generation Actions current shadow state, updates the state to account for the new interactions, and repeats until enough interactions have been generated. SQLancer doe",
+      "section": "3.2 Query Generation",
+      "page": 4,
+      "char_offset": 17539,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer does not keep a separate state [ 23], but instead uses the database APIs for querying the current state, such as the table names in sqlite_master, due to the implementation effort for the shadow model.",
+      "context_before": "b: t←pick db.tables r←pick t.rows p←gen expression (t, r)! DELETE * FROM t WHERE P! RS = SELECT * FROM t WHERE p! assert(r not in RS) (c) \"Deletes rows should not be in the table\"gen property db: s1←gen SELECT db.tables s2←gen SELECT db.tables! RS3 = s1 UNION ALL s2! assume(s1.first().length() == s1.first().length())! assert(RS1.length() + RS2.length() == RS3.length()) (d) \"UNION ALL preserves cardinality\" Figure 2: Definitions of Oracles as Generation Actions current shadow state, updates the state to account for the new interactions, and repeats until enough interactions have been generated.",
+      "context_after": "We opted for the shadow state because it allows for complex reasoning over the database state for constructing arbitrary queries and properties, and because it gives us a canonical property over the database state: the shadow state is identical to the database at any given moment. 4 Evaluation In order to evaluate the effectiveness of DIRT, we explored the answers to the following questions: •RQ1: DoesDIRTfind bugs in Turso? •RQ2: How does DIRTcompare to the state-of-the-art? 4.1 RQ1: DoesDIRTfind bugs in Turso? We have collected over 23 confirmed and fixed unique bugs found by DIRTin Turso over",
+      "section": "3.2 Query Generation",
+      "page": 4,
+      "char_offset": 18215,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M30",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "The bug was discovered as a result of our the PQS implementation in Turso augmented with validity preserving queries between the INSERT and SELECT statements.",
+      "context_before": "ite uses [ 10]). The bytecode compilation is error-prone, as small changes to the compiled SQL expression can greatly affect the generated bytecode. The bug stemmed from an error in the bytecode compilation of DELETE, specifically in the case of constant expressions in the WHERE terms. Constant expressions can be compiled to run before the main execution loop of the query, but instead the compilation step just skipped them. This caused DELETE with a constant that evaluates to FALSEto be incorrectly executed. This bug was automatically reported by DIRTrunning in Turso CI and fixed within the week.",
+      "context_after": "The random tester added a naive DELETE operation that should not have affected the results of the SELECT, keeping the containment property intact, but the assertion that the inserted row should be in the result of the SELECTfailed, triggering the bug report. Case 2: Interior node replacement caused self-reference when depth exceeded 2.The second bug manifested in the balancing logic of the B-tree implementation of Turso, but only because the generated interactions were paced to match the state of the Turso development. When Turso added database indexes as an experimental feature, we extended g",
+      "section": "4.1 RQ1: DoesDIRTfind bugs in Turso?",
+      "page": 4,
+      "char_offset": 20721,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "This resulted in triggering a crash failure in the core B-tree data structure that could result in data DIRT: Database-Integrated Random Testing Conference’17, July 2017, Washington, DC, USA Bug Id Description Oracle Module 466 TRUE not accepted as catch-all predicate No Error Query Compiler 548 Infinite loop when checkpointing on Linux No Infinite Loop I/O Subsystem 629 Query Optimizer broke TRUE in predicates No Panic Query Optimizer 662 SELECT with nested Boolean expressions sometimes gave no results PQS Query Optimizer 681 Storage Engine (B-tree) insert caused subtract with overflow No Panic Storage Engine (B-tree) 682 Faulty recursive binop logic caused SELECT to miss rows PQS Query Compiler 924 Storage Engine (B-tree) balancing caused page corruption when deleting No Panic Storage Engine (B-tree) 1040 LIKE operator did not work for non-text values No Panic Query Executor 1203 Storage Engine (B-tree) balancing error No Panic Storage Engine (B-tree) 1629 Storage Engine (B-tree) cell updates caused infinite loop in UPDATE No Infinite Loop Storage Engine (B-tree) 1734 DELETE did not emit conditional jumps if WHERE term was constant Delete-Select Query Compiler 1815 Storage Engine (B-tree) failed to balance when insert caused cell overflow No Panic Storage Engine (B-tree) 1818 Always read DB header and schema from file instead of memory page No Panic Page Manager 1975 Storage Engine (B-tree) expected table or index leaf page No Panic Storage Engine (B-tree) 1991 Use after free when validating B-tree balance No Panic Storage Engine (B-tree) 2024 SELECT .",
+      "context_before": "failed, triggering the bug report. Case 2: Interior node replacement caused self-reference when depth exceeded 2.The second bug manifested in the balancing logic of the B-tree implementation of Turso, but only because the generated interactions were paced to match the state of the Turso development. When Turso added database indexes as an experimental feature, we extended generation to support statements such as SELECT DISTINCT, CREATE INDEX, and compound operators such as UNION or UNION ALL to the space of generated inputs, and started generating larger tables to take advantage of the indexes.",
+      "context_after": ".. LIMIT resulted in different rows from SQLite Differential Query Executor 2026 UPDATE then SELECT resulted in different rows from SQLite Differential Query Executor 2047 Overflow cell with divider cell was not found due to faulty validation No Panic Storage Engine (B-tree) 2074 SELECT hung with long text, CacheFull error No Panic Page Manager 2075 Large table with 128 columns handled incorrectly No Panic Page Manager 2088 Incorrect record header size calculation No Panic Page Manager 2106 Interior node replacement caused self-reference when depth exceeded 2 No Panic Storage Engine (B-tree) 2",
+      "section": "4.1 RQ1: DoesDIRTfind bugs in Turso?",
+      "page": 4,
+      "char_offset": 21704,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "2 RQ2: How does DIRTcompare to SQLancer?",
+      "context_before": "is the result of such a situation, where metadata like database size and schema is out-of-date at the time of reading, leading to incorrect operations.DIRTfound this bug through an assertion failure in the freelist structure in the database header. As a result, the developer reporting the original bug has written the first example of aregression property, added a new fault primitive ReopenDatabase that closes down existing connections with the database and reopens them later, enabling the detection of any future bugs that might be a result of an error in the persistent state logic of Turso. 4.",
+      "context_after": "We answer RQ2 by comparing the rate of true positives and false positives reports from three different random testing configurations:DIRT, the default SQLite3 integration of SQLancer with pragmas disabled (denoted SQLancer-SQLite), and the fork of a work-in-progress Turso integration of SQLancer by a Turso contributor (denoted SQLancer-Turso). Pragmas had to be removed for SQLancer-SQLite because most of them were unimplemented in Turso, causing all reported bugs to be false positives in our testing. We define an actionable bug report from the tools as a sequence of SQL statements thatwould h",
+      "section": "4.2 RQ2: How does DIRTcompare to SQLancer?",
+      "page": 5,
+      "char_offset": 25214,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M33",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We answer RQ2 by comparing the rate of true positives and false positives reports from three different random testing configurations:DIRT, the default SQLite3 integration of SQLancer with pragmas disabled (denoted SQLancer-SQLite), and the fork of a work-in-progress Turso integration of SQLancer by a Turso contributor (denoted SQLancer-Turso).",
+      "context_before": "e metadata like database size and schema is out-of-date at the time of reading, leading to incorrect operations.DIRTfound this bug through an assertion failure in the freelist structure in the database header. As a result, the developer reporting the original bug has written the first example of aregression property, added a new fault primitive ReopenDatabase that closes down existing connections with the database and reopens them later, enabling the detection of any future bugs that might be a result of an error in the persistent state logic of Turso. 4.2 RQ2: How does DIRTcompare to SQLancer?",
+      "context_after": "Pragmas had to be removed for SQLancer-SQLite because most of them were unimplemented in Turso, causing all reported bugs to be false positives in our testing. We define an actionable bug report from the tools as a sequence of SQL statements thatwould have beensubmitted and confirmed as abug at the time of its original report. We used the 23 bug reports we analyzed as the ground truth, finding commits which are known to have bugs. The bugs span over 20 Turso commits, 6 of which do not provide Java bindings necessary for connecting with SQLancer. We ran all three tools on the remaining 14 comm",
+      "section": "4.2 RQ2: How does DIRTcompare to SQLancer?",
+      "page": 5,
+      "char_offset": 25253,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M34",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Pragmas had to be removed for SQLancer-SQLite because most of them were unimplemented in Turso, causing all reported bugs to be false positives in our testing.",
+      "context_before": "ReopenDatabase that closes down existing connections with the database and reopens them later, enabling the detection of any future bugs that might be a result of an error in the persistent state logic of Turso. 4.2 RQ2: How does DIRTcompare to SQLancer? We answer RQ2 by comparing the rate of true positives and false positives reports from three different random testing configurations:DIRT, the default SQLite3 integration of SQLancer with pragmas disabled (denoted SQLancer-SQLite), and the fork of a work-in-progress Turso integration of SQLancer by a Turso contributor (denoted SQLancer-Turso).",
+      "context_after": "We define an actionable bug report from the tools as a sequence of SQL statements thatwould have beensubmitted and confirmed as abug at the time of its original report. We used the 23 bug reports we analyzed as the ground truth, finding commits which are known to have bugs. The bugs span over 20 Turso commits, 6 of which do not provide Java bindings necessary for connecting with SQLancer. We ran all three tools on the remaining 14 commits in the commit history. We analyzed the results of each run and separated them into three bins:True Positivesthat would have been submitted and confirmed as",
+      "section": "4.2 RQ2: How does DIRTcompare to SQLancer?",
+      "page": 5,
+      "char_offset": 25599,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M35",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The bugs span over 20 Turso commits, 6 of which do not provide Java bindings necessary for connecting with SQLancer.",
+      "context_before": "er with pragmas disabled (denoted SQLancer-SQLite), and the fork of a work-in-progress Turso integration of SQLancer by a Turso contributor (denoted SQLancer-Turso). Pragmas had to be removed for SQLancer-SQLite because most of them were unimplemented in Turso, causing all reported bugs to be false positives in our testing. We define an actionable bug report from the tools as a sequence of SQL statements thatwould have beensubmitted and confirmed as abug at the time of its original report. We used the 23 bug reports we analyzed as the ground truth, finding commits which are known to have bugs.",
+      "context_after": "We ran all three tools on the remaining 14 commits in the commit history. We analyzed the results of each run and separated them into three bins:True Positivesthat would have been submitted and confirmed as bugs,False Positivesthat are reported as failures by the tools but do not constitute bugs as they are unimplemented features, andNo Bugsthat did not report a failure, provided in the stacked bar chart in Fig. 3. SQLancer-SQLite achieves a true positive rate below 10% across all 14 commits. SQLancer-Turso performs better, with a 42% true positive rate, but still yields many more false posit",
+      "section": "4.2 RQ2: How does DIRTcompare to SQLancer?",
+      "page": 5,
+      "char_offset": 26034,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M36",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer-SQLite achieves a true positive rate below 10% across all 14 commits.",
+      "context_before": "the ground truth, finding commits which are known to have bugs. The bugs span over 20 Turso commits, 6 of which do not provide Java bindings necessary for connecting with SQLancer. We ran all three tools on the remaining 14 commits in the commit history. We analyzed the results of each run and separated them into three bins:True Positivesthat would have been submitted and confirmed as bugs,False Positivesthat are reported as failures by the tools but do not constitute bugs as they are unimplemented features, andNo Bugsthat did not report a failure, provided in the stacked bar chart in Fig. 3.",
+      "context_after": "SQLancer-Turso performs better, with a 42% true positive rate, but still yields many more false positives than DIRT(58% vs. under 1%).DIRTalso finds substantially more actionable bugs: 25 unique bugs, compared with six found by the two SQLancer configurations combined (§A.1). This does not mean that DIRThas better generators or oracles than SQLancer. Rather, it shows that for rapidly evolving databases with many missing features, database-integrated testing yields more actionable bugs. 5 Discussion DIRTis a paradigm for testing databases during development, even though we use DIRTthe paradigm an",
+      "section": "4.2 RQ2: How does DIRTcompare to SQLancer?",
+      "page": 5,
+      "char_offset": 26570,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M37",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer-Turso performs better, with a 42% true positive rate, but still yields many more false positives than DIRT(58% vs.",
+      "context_before": "over 20 Turso commits, 6 of which do not provide Java bindings necessary for connecting with SQLancer. We ran all three tools on the remaining 14 commits in the commit history. We analyzed the results of each run and separated them into three bins:True Positivesthat would have been submitted and confirmed as bugs,False Positivesthat are reported as failures by the tools but do not constitute bugs as they are unimplemented features, andNo Bugsthat did not report a failure, provided in the stacked bar chart in Fig. 3. SQLancer-SQLite achieves a true positive rate below 10% across all 14 commits.",
+      "context_after": "under 1%).DIRTalso finds substantially more actionable bugs: 25 unique bugs, compared with six found by the two SQLancer configurations combined (§A.1). This does not mean that DIRThas better generators or oracles than SQLancer. Rather, it shows that for rapidly evolving databases with many missing features, database-integrated testing yields more actionable bugs. 5 Discussion DIRTis a paradigm for testing databases during development, even though we use DIRTthe paradigm and DIRTfor Turso interchangeably throughout the paper. We compare the performance of DIRT for Turso against SQLancer with mini",
+      "section": "4.2 RQ2: How does DIRTcompare to SQLancer?",
+      "page": 5,
+      "char_offset": 26649,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M38",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "DIRTalso finds substantially more actionable bugs: 25 unique bugs, compared with six found by the two SQLancer configurations combined (§A.",
+      "context_before": "emaining 14 commits in the commit history. We analyzed the results of each run and separated them into three bins:True Positivesthat would have been submitted and confirmed as bugs,False Positivesthat are reported as failures by the tools but do not constitute bugs as they are unimplemented features, andNo Bugsthat did not report a failure, provided in the stacked bar chart in Fig. 3. SQLancer-SQLite achieves a true positive rate below 10% across all 14 commits. SQLancer-Turso performs better, with a 42% true positive rate, but still yields many more false positives than DIRT(58% vs. under 1%).",
+      "context_after": "1). This does not mean that DIRThas better generators or oracles than SQLancer. Rather, it shows that for rapidly evolving databases with many missing features, database-integrated testing yields more actionable bugs. 5 Discussion DIRTis a paradigm for testing databases during development, even though we use DIRTthe paradigm and DIRTfor Turso interchangeably throughout the paper. We compare the performance of DIRT for Turso against SQLancer with minimal modifications, which begets some questions worth discussing. Why not integrate SQLancer to the development process instead of developing a random",
+      "section": "4.2 RQ2: How does DIRTcompare to SQLancer?",
+      "page": 5,
+      "char_offset": 26783,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M39",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This does not mean that DIRThas better generators or oracles than SQLancer.",
+      "context_before": "e been submitted and confirmed as bugs,False Positivesthat are reported as failures by the tools but do not constitute bugs as they are unimplemented features, andNo Bugsthat did not report a failure, provided in the stacked bar chart in Fig. 3. SQLancer-SQLite achieves a true positive rate below 10% across all 14 commits. SQLancer-Turso performs better, with a 42% true positive rate, but still yields many more false positives than DIRT(58% vs. under 1%).DIRTalso finds substantially more actionable bugs: 25 unique bugs, compared with six found by the two SQLancer configurations combined (§A.1).",
+      "context_after": "Rather, it shows that for rapidly evolving databases with many missing features, database-integrated testing yields more actionable bugs. 5 Discussion DIRTis a paradigm for testing databases during development, even though we use DIRTthe paradigm and DIRTfor Turso interchangeably throughout the paper. We compare the performance of DIRT for Turso against SQLancer with minimal modifications, which begets some questions worth discussing. Why not integrate SQLancer to the development process instead of developing a random testing framework from scratch?As we briefly discussed earlier, SQLancer is no",
+      "section": "4.2 RQ2: How does DIRTcompare to SQLancer?",
+      "page": 5,
+      "char_offset": 26925,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M40",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We compare the performance of DIRT for Turso against SQLancer with minimal modifications, which begets some questions worth discussing.",
+      "context_before": "itive rate, but still yields many more false positives than DIRT(58% vs. under 1%).DIRTalso finds substantially more actionable bugs: 25 unique bugs, compared with six found by the two SQLancer configurations combined (§A.1). This does not mean that DIRThas better generators or oracles than SQLancer. Rather, it shows that for rapidly evolving databases with many missing features, database-integrated testing yields more actionable bugs. 5 Discussion DIRTis a paradigm for testing databases during development, even though we use DIRTthe paradigm and DIRTfor Turso interchangeably throughout the paper.",
+      "context_after": "Why not integrate SQLancer to the development process instead of developing a random testing framework from scratch?As we briefly discussed earlier, SQLancer is not designed for the constant evolution of the random testing infrastructure along with the project. It is a project with its own trajectory, development, new algorithms, Conference’17, July 2017, Washington, DC, USA Alperen Keles, Ethan Chou, Harrison Goldstein, and Leonidas Lampropoulos Commit 0%10%20%30%40%50%60%70%80%90%100%% of Tests09b341c6a62313c30e45324273ae3e489606b967731e675b868ac0994262b0971e2d8dLimbo SQLancer (Modified) SQ",
+      "section": "5 Discussion",
+      "page": 5,
+      "char_offset": 27301,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M41",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Why not integrate SQLancer to the development process instead of developing a random testing framework from scratch?",
+      "context_before": "unique bugs, compared with six found by the two SQLancer configurations combined (§A.1). This does not mean that DIRThas better generators or oracles than SQLancer. Rather, it shows that for rapidly evolving databases with many missing features, database-integrated testing yields more actionable bugs. 5 Discussion DIRTis a paradigm for testing databases during development, even though we use DIRTthe paradigm and DIRTfor Turso interchangeably throughout the paper. We compare the performance of DIRT for Turso against SQLancer with minimal modifications, which begets some questions worth discussing.",
+      "context_after": "As we briefly discussed earlier, SQLancer is not designed for the constant evolution of the random testing infrastructure along with the project. It is a project with its own trajectory, development, new algorithms, Conference’17, July 2017, Washington, DC, USA Alperen Keles, Ethan Chou, Harrison Goldstein, and Leonidas Lampropoulos Commit 0%10%20%30%40%50%60%70%80%90%100%% of Tests09b341c6a62313c30e45324273ae3e489606b967731e675b868ac0994262b0971e2d8dLimbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo S",
+      "section": "5 Discussion",
+      "page": 5,
+      "char_offset": 27436,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M42",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "As we briefly discussed earlier, SQLancer is not designed for the constant evolution of the random testing infrastructure along with the project.",
+      "context_before": "has better generators or oracles than SQLancer. Rather, it shows that for rapidly evolving databases with many missing features, database-integrated testing yields more actionable bugs. 5 Discussion DIRTis a paradigm for testing databases during development, even though we use DIRTthe paradigm and DIRTfor Turso interchangeably throughout the paper. We compare the performance of DIRT for Turso against SQLancer with minimal modifications, which begets some questions worth discussing. Why not integrate SQLancer to the development process instead of developing a random testing framework from scratch?",
+      "context_after": "It is a project with its own trajectory, development, new algorithms, Conference’17, July 2017, Washington, DC, USA Alperen Keles, Ethan Chou, Harrison Goldstein, and Leonidas Lampropoulos Commit 0%10%20%30%40%50%60%70%80%90%100%% of Tests09b341c6a62313c30e45324273ae3e489606b967731e675b868ac0994262b0971e2d8dLimbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer",
+      "section": "5 Discussion",
+      "page": 5,
+      "char_offset": 27553,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M43",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "It is a project with its own trajectory, development, new algorithms, Conference’17, July 2017, Washington, DC, USA Alperen Keles, Ethan Chou, Harrison Goldstein, and Leonidas Lampropoulos Commit 0%10%20%30%40%50%60%70%80%90%100%% of Tests09b341c6a62313c30e45324273ae3e489606b967731e675b868ac0994262b0971e2d8dLimbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original)True Positive False Positive No Bugclass Figure 3: Comparison of True Positives, False Positives and No Bugs for DIRT, SQLancer-SQLite, SQLancer-Turso generators, and oracles.",
+      "context_before": "ted testing yields more actionable bugs. 5 Discussion DIRTis a paradigm for testing databases during development, even though we use DIRTthe paradigm and DIRTfor Turso interchangeably throughout the paper. We compare the performance of DIRT for Turso against SQLancer with minimal modifications, which begets some questions worth discussing. Why not integrate SQLancer to the development process instead of developing a random testing framework from scratch?As we briefly discussed earlier, SQLancer is not designed for the constant evolution of the random testing infrastructure along with the project.",
+      "context_after": "DIRTaims to empower the database developers themselves by giving them a mechanism to test the database, instead of giving them bugs to solve. There are additional practical barriers; Java bindings for Turso are not complete at the moment, so we had to fix bugs in the SQLancer-Turso integration implemented by one of the Turso developers that relied on incomplete features that silently failed, panics in Turso caused SQLancer to terminate, so it was not possible to minimize the counterexamples with SQLancer reducers without further changes. Given that SQLancer supports expected errors for reducin",
+      "section": "5 Discussion",
+      "page": 5,
+      "char_offset": 27698,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M44",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "There are additional practical barriers; Java bindings for Turso are not complete at the moment, so we had to fix bugs in the SQLancer-Turso integration implemented by one of the Turso developers that relied on incomplete features that silently failed, panics in Turso caused SQLancer to terminate, so it was not possible to minimize the counterexamples with SQLancer reducers without further changes.",
+      "context_before": "Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original) Limbo SQLancer (Modified) SQLancer (Original)True Positive False Positive No Bugclass Figure 3: Comparison of True Positives, False Positives and No Bugs for DIRT, SQLancer-SQLite, SQLancer-Turso generators, and oracles.DIRTaims to empower the database developers themselves by giving them a mechanism to test the database, instead of giving them bugs to solve.",
+      "context_after": "Given that SQLancer supports expected errors for reducing false positives, could we use expected errors for known bugs and false positives we found to discover more bugs?We could use expected errors, and we would have discovered more bugs in the process. In terms of any static evaluation target, tweaking SQLancer by progressively finding more bugs as found ones are marked as expected, it is always possible to make SQLancer find more bugs. The point, however, is that constant modification is not the expected and supported mode of operation when using SQLancer. Although it is phenomenal for off",
+      "section": "5 Discussion",
+      "page": 6,
+      "char_offset": 28966,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M45",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Given that SQLancer supports expected errors for reducing false positives, could we use expected errors for known bugs and false positives we found to discover more bugs?",
+      "context_before": "SQLancer-SQLite, SQLancer-Turso generators, and oracles.DIRTaims to empower the database developers themselves by giving them a mechanism to test the database, instead of giving them bugs to solve. There are additional practical barriers; Java bindings for Turso are not complete at the moment, so we had to fix bugs in the SQLancer-Turso integration implemented by one of the Turso developers that relied on incomplete features that silently failed, panics in Turso caused SQLancer to terminate, so it was not possible to minimize the counterexamples with SQLancer reducers without further changes.",
+      "context_after": "We could use expected errors, and we would have discovered more bugs in the process. In terms of any static evaluation target, tweaking SQLancer by progressively finding more bugs as found ones are marked as expected, it is always possible to make SQLancer find more bugs. The point, however, is that constant modification is not the expected and supported mode of operation when using SQLancer. Although it is phenomenal for off-the-shelf usage, its capabilities unfortunately act as a disadvantage against its use in actively developed projects with many missing features. We have shown that a proj",
+      "section": "5 Discussion",
+      "page": 6,
+      "char_offset": 29368,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M46",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In terms of any static evaluation target, tweaking SQLancer by progressively finding more bugs as found ones are marked as expected, it is always possible to make SQLancer find more bugs.",
+      "context_before": "or Turso are not complete at the moment, so we had to fix bugs in the SQLancer-Turso integration implemented by one of the Turso developers that relied on incomplete features that silently failed, panics in Turso caused SQLancer to terminate, so it was not possible to minimize the counterexamples with SQLancer reducers without further changes. Given that SQLancer supports expected errors for reducing false positives, could we use expected errors for known bugs and false positives we found to discover more bugs?We could use expected errors, and we would have discovered more bugs in the process.",
+      "context_after": "The point, however, is that constant modification is not the expected and supported mode of operation when using SQLancer. Although it is phenomenal for off-the-shelf usage, its capabilities unfortunately act as a disadvantage against its use in actively developed projects with many missing features. We have shown that a project with a much smaller scope can adopt the ideas of input generation and oracles in SQLancer as well as other related work on database testing such as Apollo [ 13] or Thanos [ 8] and demonstrated that database developers can turn their domain expertise into writing prope",
+      "section": "5 Discussion",
+      "page": 6,
+      "char_offset": 29623,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M47",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The point, however, is that constant modification is not the expected and supported mode of operation when using SQLancer.",
+      "context_before": "failed, panics in Turso caused SQLancer to terminate, so it was not possible to minimize the counterexamples with SQLancer reducers without further changes. Given that SQLancer supports expected errors for reducing false positives, could we use expected errors for known bugs and false positives we found to discover more bugs?We could use expected errors, and we would have discovered more bugs in the process. In terms of any static evaluation target, tweaking SQLancer by progressively finding more bugs as found ones are marked as expected, it is always possible to make SQLancer find more bugs.",
+      "context_after": "Although it is phenomenal for off-the-shelf usage, its capabilities unfortunately act as a disadvantage against its use in actively developed projects with many missing features. We have shown that a project with a much smaller scope can adopt the ideas of input generation and oracles in SQLancer as well as other related work on database testing such as Apollo [ 13] or Thanos [ 8] and demonstrated that database developers can turn their domain expertise into writing properties as presented in the third case study in §4.1. We also observed that many crash failures are the result of inline asse",
+      "section": "5 Discussion",
+      "page": 6,
+      "char_offset": 29811,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M48",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We have shown that a project with a much smaller scope can adopt the ideas of input generation and oracles in SQLancer as well as other related work on database testing such as Apollo [ 13] or Thanos [ 8] and demonstrated that database developers can turn their domain expertise into writing properties as presented in the third case study in §4.",
+      "context_before": "und to discover more bugs?We could use expected errors, and we would have discovered more bugs in the process. In terms of any static evaluation target, tweaking SQLancer by progressively finding more bugs as found ones are marked as expected, it is always possible to make SQLancer find more bugs. The point, however, is that constant modification is not the expected and supported mode of operation when using SQLancer. Although it is phenomenal for off-the-shelf usage, its capabilities unfortunately act as a disadvantage against its use in actively developed projects with many missing features.",
+      "context_after": "1. We also observed that many crash failures are the result of inline assertions in Turso. These crashes imply logic bugs as they contradict invariants specified by the developers, blurring the distinction between crash failures and logic bugs in the literature. Such assertions do not replace SQLancer oracles or Turso properties because they do not have the ability to follow values through execution in a holistic way as properties do, so they can only reason about local invariants. Role in Turso.As a last point of discussion, we would like to clarify our role in the implementation of DIRTin Tur",
+      "section": "5 Discussion",
+      "page": 6,
+      "char_offset": 30113,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M49",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Such assertions do not replace SQLancer oracles or Turso properties because they do not have the ability to follow values through execution in a holistic way as properties do, so they can only reason about local invariants.",
+      "context_before": "shown that a project with a much smaller scope can adopt the ideas of input generation and oracles in SQLancer as well as other related work on database testing such as Apollo [ 13] or Thanos [ 8] and demonstrated that database developers can turn their domain expertise into writing properties as presented in the third case study in §4.1. We also observed that many crash failures are the result of inline assertions in Turso. These crashes imply logic bugs as they contradict invariants specified by the developers, blurring the distinction between crash failures and logic bugs in the literature.",
+      "context_after": "Role in Turso.As a last point of discussion, we would like to clarify our role in the implementation of DIRTin Turso, in order tofully credit the open source contributors for their work. At the time we started working on Turso, it had a small, unstructured random testing infrastructure that overwhelmingly focused on being able to support Deterministic Simulation Testing [ 28]. I/O was implemented in a way that could be easily simulated, which allowed for FAULT s to be integrated in the generated interactions. We have taken a role as an external open source contributor to the project, gradually",
+      "section": "5 Discussion",
+      "page": 6,
+      "char_offset": 30722,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M50",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The SQLancer-Turso integration was almost entirely developed by one contributor, which we have used in our evaluations with small changes to their code.",
+      "context_before": "e as an external open source contributor to the project, gradually proposing improvements to random testing infrastructure such as the Generation Actions DSL, most of the existing properties, stateful random generation, shrinking, additional oracles such as differential testing against SQLite and determinism checking, and have implemented such proposals. The infrastructure has grown outside of our control at times, producing regressions through the development, along with many improvements by the maintainers and contributors to the project that contributed to the list of bugs found in Table 1.",
+      "context_after": "Related work on adaptive DBMS testing.Although we compared DIRTmainly against SQLancer as the basis of our oracles, other systems such as SQLRight [ 17], Griffin [ 7], and SQLancer++ [ 27] also adapt testing to DBMS behavior. However, these systems primarily approach adaptation from the perspective of an external testing framework, with evaluations centered on broad effectiveness across multiple databases and minimal manual intervention. In contrast,DIRTstudies adaptation as a database-integrated process: the testing infrastructure evolves together with a single DBMS under development, and de",
+      "section": "5 Discussion",
+      "page": 6,
+      "char_offset": 32080,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M51",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Although we compared DIRTmainly against SQLancer as the basis of our oracles, other systems such as SQLRight [ 17], Griffin [ 7], and SQLancer++ [ 27] also adapt testing to DBMS behavior.",
+      "context_before": "stateful random generation, shrinking, additional oracles such as differential testing against SQLite and determinism checking, and have implemented such proposals. The infrastructure has grown outside of our control at times, producing regressions through the development, along with many improvements by the maintainers and contributors to the project that contributed to the list of bugs found in Table 1. The SQLancer-Turso integration was almost entirely developed by one contributor, which we have used in our evaluations with small changes to their code. Related work on adaptive DBMS testing.",
+      "context_after": "However, these systems primarily approach adaptation from the perspective of an external testing framework, with evaluations centered on broad effectiveness across multiple databases and minimal manual intervention. In contrast,DIRTstudies adaptation as a database-integrated process: the testing infrastructure evolves together with a single DBMS under development, and developers directly shape generators, properties, and fault models as the implementation changes. 6 Conclusion and Future Work In this paper, we introduced DIRT, a new paradigm and concrete implementation for testing databases du",
+      "section": "5 Discussion",
+      "page": 6,
+      "char_offset": 32272,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M52",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "By tightly integrating the testing infrastructure with the DBMS itself,DIRT avoids many of the limitations that off-the-shelf testing frameworks DIRT: Database-Integrated Random Testing Conference’17, July 2017, Washington, DC, USA like SQLancer face when applied to rapidly evolving systems with incomplete feature sets.",
+      "context_before": "ystems primarily approach adaptation from the perspective of an external testing framework, with evaluations centered on broad effectiveness across multiple databases and minimal manual intervention. In contrast,DIRTstudies adaptation as a database-integrated process: the testing infrastructure evolves together with a single DBMS under development, and developers directly shape generators, properties, and fault models as the implementation changes. 6 Conclusion and Future Work In this paper, we introduced DIRT, a new paradigm and concrete implementation for testing databases during development.",
+      "context_after": "Our framework provides developers with tools to express custom correctness properties via generation actions and ensures state-aware query generation. We implemented DIRTfor Turso and found 23 confirmed bugs with minimal false positives. In the same setting, SQLancer-based baselines produced substantially more false positives and fewer actionable reports. Our experience shows that when databases are still under heavy development, flexible, integrated testing approaches that evolve alongside the system are significantly more productive and developer-friendly than external, comprehensive tools.",
+      "section": "6 Conclusion and Future Work",
+      "page": 6,
+      "char_offset": 33076,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M53",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In the same setting, SQLancer-based baselines produced substantially more false positives and fewer actionable reports.",
+      "context_before": "or testing databases during development. By tightly integrating the testing infrastructure with the DBMS itself,DIRT avoids many of the limitations that off-the-shelf testing frameworks DIRT: Database-Integrated Random Testing Conference’17, July 2017, Washington, DC, USA like SQLancer face when applied to rapidly evolving systems with incomplete feature sets. Our framework provides developers with tools to express custom correctness properties via generation actions and ensures state-aware query generation. We implemented DIRTfor Turso and found 23 confirmed bugs with minimal false positives.",
+      "context_after": "Our experience shows that when databases are still under heavy development, flexible, integrated testing approaches that evolve alongside the system are significantly more productive and developer-friendly than external, comprehensive tools. Moreover, our approach fosters collaboration between testing infrastructure and DBMS development by giving developers intuitive ways to encode domain-specific knowledge as properties. DIRTbridges the gap between random testing and practical debugging by empowering developers to guide, understand, and act on test results. We believe this paradigm has broad",
+      "section": "6 Conclusion and Future Work",
+      "page": 7,
+      "char_offset": 33636,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M54",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "1 Bugs Found by SQLancer •(UPDATE t SET (c0, c0)=(0, 0) ): This expression results in a \"Column specified more than once\" error that has not been fixed at the time of our submission.",
+      "context_before": "lve alongside the system are significantly more productive and developer-friendly than external, comprehensive tools. Moreover, our approach fosters collaboration between testing infrastructure and DBMS development by giving developers intuitive ways to encode domain-specific knowledge as properties. DIRTbridges the gap between random testing and practical debugging by empowering developers to guide, understand, and act on test results. We believe this paradigm has broad applicability beyond Turso and offers a promising path forward for testing other evolving and complex systems. A Appendix A.",
+      "context_after": "•(SELECT (0x0) ): There was a parsing error for hexadecimals that resulted in an \"invalid float literal\" error that has since been fixed before our experiments with SQLancer. •(SELECT * FROM t WHERE c0 GLOB c0 ): There was a logic bug in the GLOB that panicked when the values passed were not TEXT. Executing this statement after inserting a NULL triggered the bug, which was fixed soon after we reported it. •(INSERT INTO t VALUES (\"a\") ): Using double quotes for string literals is discouraged, and SQLite3 even provides a runtime flag for disabling it. We reported this bug as Turso panicked for",
+      "section": "6 Conclusion and Future Work",
+      "page": 7,
+      "char_offset": 34481,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M55",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "•(SELECT (0x0) ): There was a parsing error for hexadecimals that resulted in an \"invalid float literal\" error that has since been fixed before our experiments with SQLancer.",
+      "context_before": "rastructure and DBMS development by giving developers intuitive ways to encode domain-specific knowledge as properties. DIRTbridges the gap between random testing and practical debugging by empowering developers to guide, understand, and act on test results. We believe this paradigm has broad applicability beyond Turso and offers a promising path forward for testing other evolving and complex systems. A Appendix A.1 Bugs Found by SQLancer •(UPDATE t SET (c0, c0)=(0, 0) ): This expression results in a \"Column specified more than once\" error that has not been fixed at the time of our submission.",
+      "context_after": "•(SELECT * FROM t WHERE c0 GLOB c0 ): There was a logic bug in the GLOB that panicked when the values passed were not TEXT. Executing this statement after inserting a NULL triggered the bug, which was fixed soon after we reported it. •(INSERT INTO t VALUES (\"a\") ): Using double quotes for string literals is discouraged, and SQLite3 even provides a runtime flag for disabling it. We reported this bug as Turso panicked for the provided statement. •(INSERT INTO t VALUES ((0 BETWEEN 0 AND 0)), (0) ): Constant BETWEEN expressions were not rewritten when used within INSERT, but the query compiler ex",
+      "section": "6 Conclusion and Future Work",
+      "page": 7,
+      "char_offset": 34663,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M56",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer discovered the bug when the initial table thad a NOT NULL clause for the column c0, which promptly failed after executing the statement.",
+      "context_before": "NTO t VALUES (\"a\") ): Using double quotes for string literals is discouraged, and SQLite3 even provides a runtime flag for disabling it. We reported this bug as Turso panicked for the provided statement. •(INSERT INTO t VALUES ((0 BETWEEN 0 AND 0)), (0) ): Constant BETWEEN expressions were not rewritten when used within INSERT, but the query compiler expected them to be rewritten. We reported this bug as Turso panicked for the provided statement. •(INSERT INTO t(c2, c0) VALUES (0, 0), (0, 0) ): There was a bug in calculating the column indexes when inserting values to a table in reverse order.",
+      "context_after": "We reported this bug after manually inspecting the result of the failure. In all six cases, we see the benefit of generating the entire possible input space for SQL dialect of SQLite, which DIRTdoes not, hence missing such bugs. Our evaluation demonstrates the benefit of such comprehensive generation, but also shows that without tailoring to the database, the produced bug reports will be overwhelmingly dominated by false positives and reproductions of existing bugs yet to be fixed.References [1]Ba, J., and Rigger, M.Testing database engines via query plan guidance. In Proceedings of the 45th I",
+      "section": "6 Conclusion and Future Work",
+      "page": 7,
+      "char_offset": 35683,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M57",
+      "found_by": "technique",
+      "surface": "query plan guidance",
+      "technique": "qpg",
+      "sentence": "Testing database engines via query plan guidance.",
+      "context_before": "clause for the column c0, which promptly failed after executing the statement. We reported this bug after manually inspecting the result of the failure. In all six cases, we see the benefit of generating the entire possible input space for SQL dialect of SQLite, which DIRTdoes not, hence missing such bugs. Our evaluation demonstrates the benefit of such comprehensive generation, but also shows that without tailoring to the database, the produced bug reports will be overwhelmingly dominated by false positives and reproductions of existing bugs yet to be fixed.References [1]Ba, J., and Rigger, M.",
+      "context_after": "In Proceedings of the 45th International Conference on Software Engineering(2023), ICSE ’23, IEEE Press, p. 2060–2071. [2]Ba, J., and Rigger, M.Cert: Finding performance issues in database systems through the lens of cardinality estimation. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering(New York, NY, USA, 2024), ICSE ’24, Association for Computing Machinery. [3]Ba, J., and Rigger, M.Keep it simple: Testing databases via differential query plans. [4]Böhme, M., Pham, V.-T., and Roychoudhury, A.Coverage-based greybox fuzzing as markov chain. InProceedings of",
+      "section": "6 Conclusion and Future Work",
+      "page": 7,
+      "char_offset": 36351,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M58",
+      "found_by": "technique",
+      "surface": "Cert",
+      "technique": "cert",
+      "sentence": "Cert: Finding performance issues in database systems through the lens of cardinality estimation.",
+      "context_before": "enerating the entire possible input space for SQL dialect of SQLite, which DIRTdoes not, hence missing such bugs. Our evaluation demonstrates the benefit of such comprehensive generation, but also shows that without tailoring to the database, the produced bug reports will be overwhelmingly dominated by false positives and reproductions of existing bugs yet to be fixed.References [1]Ba, J., and Rigger, M.Testing database engines via query plan guidance. In Proceedings of the 45th International Conference on Software Engineering(2023), ICSE ’23, IEEE Press, p. 2060–2071. [2]Ba, J., and Rigger, M.",
+      "context_after": "InProceedings of the IEEE/ACM 46th International Conference on Software Engineering(New York, NY, USA, 2024), ICSE ’24, Association for Computing Machinery. [3]Ba, J., and Rigger, M.Keep it simple: Testing databases via differential query plans. [4]Böhme, M., Pham, V.-T., and Roychoudhury, A.Coverage-based greybox fuzzing as markov chain. InProceedings of the 2016 ACMSIGSAC Conference on Computer and Communications Security(New York, NY, USA, 2016), CCS ’16, Association for Computing Machinery, p. 1032–1043. [5]Claessen, K., and Hughes, J.Quickcheck: a lightweight tool for random testing of h",
+      "section": "6 Conclusion and Future Work",
+      "page": 7,
+      "char_offset": 36545,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M59",
+      "found_by": "technique",
+      "surface": "non-optimizing reference engine construction",
+      "technique": "norec",
+      "sentence": "Detecting optimization bugs in database engines via non-optimizing reference engine construction.",
+      "context_before": "f DBMS with Coveragebased Guidance. InProceedings of the 31st USENIX Security Symposium (USENIX 2022)(Boston, MA, aug 2022). [18] Padhye, R., Lemieux, C., Sen, K., Papadakis, M., and Le Traon, Y.Semantic fuzzing with zest. InProceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis(New York, NY, USA, 2019), ISSTA 2019, Association for Computing Machinery, p. 329–340. [19] Padhye, R., Lemieux, C., Sen, K., Simon, L., and Vijayakumar, H.Fuzzfactory: domain-specific fuzzing with waypoints.Proc. ACM Program. Lang. 3, OOPSLA (Oct. 2019). [20] Rigger, M., and Su, Z.",
+      "context_after": "InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(New York, NY, USA, 2020), ESEC/FSE 2020, Association for Computing Machinery, p. 1140–1152. [21] Rigger, M., and Su, Z.Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (Nov. 2020). [22] Rigger, M., and Su, Z.Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20)(Nov. 2020), USENIX Association, pp. 667–682. [23] Rigger, M., and S",
+      "section": "6 Conclusion and Future Work",
+      "page": 7,
+      "char_offset": 40184,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M60",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "Finding bugs in database systems via query partitioning.",
+      "context_before": "tion for Computing Machinery, p. 329–340. [19] Padhye, R., Lemieux, C., Sen, K., Simon, L., and Vijayakumar, H.Fuzzfactory: domain-specific fuzzing with waypoints.Proc. ACM Program. Lang. 3, OOPSLA (Oct. 2019). [20] Rigger, M., and Su, Z.Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(New York, NY, USA, 2020), ESEC/FSE 2020, Association for Computing Machinery, p. 1140–1152. [21] Rigger, M., and Su, Z.",
+      "context_after": "Proc. ACM Program. Lang. 4, OOPSLA (Nov. 2020). [22] Rigger, M., and Su, Z.Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20)(Nov. 2020), USENIX Association, pp. 667–682. [23] Rigger, M., and Su, Z.Testing database engines via pivoted query synthesis. InProceedings of the 14th USENIX Conference on Operating Systems Design and Implementation(USA, 2020), OSDI’20, USENIX Association. [24]Seltenreich, A.Sqlsmith. https://github.com/anse1/sqlsmith, 2015. [25] Zeller, A., Gopinath, R., Böhme, M., Fraser, G., and Ho",
+      "section": "6 Conclusion and Future Work",
+      "page": 7,
+      "char_offset": 40546,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M61",
+      "found_by": "technique",
+      "surface": "pivoted query synthesis",
+      "technique": "pqs",
+      "sentence": "Testing database engines via pivoted query synthesis.",
+      "context_before": "pecific fuzzing with waypoints.Proc. ACM Program. Lang. 3, OOPSLA (Oct. 2019). [20] Rigger, M., and Su, Z.Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(New York, NY, USA, 2020), ESEC/FSE 2020, Association for Computing Machinery, p. 1140–1152. [21] Rigger, M., and Su, Z.Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (Nov. 2020). [22] Rigger, M., and Su, Z.",
+      "context_after": "In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20)(Nov. 2020), USENIX Association, pp. 667–682. [23] Rigger, M., and Su, Z.Testing database engines via pivoted query synthesis. InProceedings of the 14th USENIX Conference on Operating Systems Design and Implementation(USA, 2020), OSDI’20, USENIX Association. [24]Seltenreich, A.Sqlsmith. https://github.com/anse1/sqlsmith, 2015. [25] Zeller, A., Gopinath, R., Böhme, M., Fraser, G., and Holler, C.The Fuzzing Book. CISPA Helmholtz Center for Information Security, 2024. Retrieved 202407-01 16:50:18+02:00. Conference’1",
+      "section": "6 Conclusion and Future Work",
+      "page": 7,
+      "char_offset": 40678,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M62",
+      "found_by": "technique",
+      "surface": "pivoted query synthesis",
+      "technique": "pqs",
+      "sentence": "Testing database engines via pivoted query synthesis.",
+      "context_before": "oceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(New York, NY, USA, 2020), ESEC/FSE 2020, Association for Computing Machinery, p. 1140–1152. [21] Rigger, M., and Su, Z.Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (Nov. 2020). [22] Rigger, M., and Su, Z.Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20)(Nov. 2020), USENIX Association, pp. 667–682. [23] Rigger, M., and Su, Z.",
+      "context_after": "InProceedings of the 14th USENIX Conference on Operating Systems Design and Implementation(USA, 2020), OSDI’20, USENIX Association. [24]Seltenreich, A.Sqlsmith. https://github.com/anse1/sqlsmith, 2015. [25] Zeller, A., Gopinath, R., Böhme, M., Fraser, G., and Holler, C.The Fuzzing Book. CISPA Helmholtz Center for Information Security, 2024. Retrieved 202407-01 16:50:18+02:00. Conference’17, July 2017, Washington, DC, USA Alperen Keles, Ethan Chou, Harrison Goldstein, and Leonidas Lampropoulos [26] Zhang, C., and Rigger, M.Constant optimization driven database system testing. 2025). [27]Zhong",
+      "section": "6 Conclusion and Future Work",
+      "page": 7,
+      "char_offset": 40886,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M63",
+      "found_by": "technique",
+      "surface": "Constant optimization driven database system testing",
+      "technique": "coddtest",
+      "sentence": "Constant optimization driven database system testing.",
+      "context_before": "er, M., and Su, Z.Testing database engines via pivoted query synthesis. InProceedings of the 14th USENIX Conference on Operating Systems Design and Implementation(USA, 2020), OSDI’20, USENIX Association. [24]Seltenreich, A.Sqlsmith. https://github.com/anse1/sqlsmith, 2015. [25] Zeller, A., Gopinath, R., Böhme, M., Fraser, G., and Holler, C.The Fuzzing Book. CISPA Helmholtz Center for Information Security, 2024. Retrieved 202407-01 16:50:18+02:00. Conference’17, July 2017, Washington, DC, USA Alperen Keles, Ethan Chou, Harrison Goldstein, and Leonidas Lampropoulos [26] Zhang, C., and Rigger, M.",
+      "context_after": "2025). [27]Zhong, S., and Rigger, M.Scaling automated database system testing, 2026. [28] Zhou, J., Xu, M., Shraer, A., Namasivayam, B., Miller, A., Tschannen, E., Atherton, S., Beamon, A. J., Sears, R., Leach, J., Rosenthal, D., Dong, X.,Wilson, W., Collins, B., Scherer, D., Grieser, A., Liu, Y., Moore, A., Muppana, B., Su, X., and Yadav, V.Foundationdb: A distributed unbundled transactional key value store. InProceedings of the 2021 International Conference on Management of Data(New York, NY, USA, 2021), SIGMOD ’21, Association for Computing Machinery, p. 2653–2666.",
+      "section": "6 Conclusion and Future Work",
+      "page": 8,
+      "char_offset": 41468,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "coddtest"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M2",
+        "M39",
+        "M40",
+        "M51"
+      ]
+    },
+    "suppressed": [
+      {
+        "mention_id": "M6",
+        "pattern": "uses_infrastructure",
+        "suppressed_because": "the sentence is framed as related work, so it describes somebody else's approach"
+      },
+      {
+        "mention_id": "M48",
+        "pattern": "uses_infrastructure",
+        "suppressed_because": "the sentence is framed as related work, so it describes somebody else's approach"
+      }
+    ]
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:11:05Z",
+    "is_model_written": true,
+    "summary": "DIRT argues that testing tools built for mature databases fit early-stage systems badly, and integrates the testing framework into the DBMS itself so that random testing evolves with the system and false positives are reduced by construction. It introduces generation actions, an abstraction that lets database developers rather than testing experts state correctness properties, and defines five oracles as sequences of them. Evaluated on Turso, an actively developed SQLite-compatible engine, it found 23 unique confirmed bugs.",
+    "narrative": "DIRT takes SQLancer as both its point of departure and its baseline: three of its five oracles are reimplementations of SQLancer's, expressed as generation actions, and it explicitly declines to modify SQLancer itself, arguing that integrating a new DBMS into it is too laborious for a system under active development. It is careful to say that outperforming SQLancer on Turso does not mean it has better generators or oracles.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "result_comparison",
+      "M3": "background",
+      "M4": "background",
+      "M5": "motivation",
+      "M6": "background",
+      "M7": "background",
+      "M8": "background",
+      "M9": "background",
+      "M12": "background",
+      "M13": "background",
+      "M14": "motivation",
+      "M15": "motivation",
+      "M16": "background",
+      "M17": "motivation",
+      "M19": "extension",
+      "M20": "extension",
+      "M23": "extension",
+      "M25": "extension",
+      "M26": "extension",
+      "M39": "result_comparison",
+      "M40": "baseline",
+      "M51": "baseline"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The opposite of reuse is the paper's argument: M17 says that rather than expecting developers to modify SQLancer, DIRT offers its own framework, and M14 and M15 give integrating a DBMS into SQLancer as the cost being avoided. The oracles are reimplemented, not imported."
+      },
+      "extends_technique": {
+        "value": "yes",
+        "mention_ids": [
+          "M19",
+          "M20",
+          "M25",
+          "M26"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M19",
+            "sentence": "1 Definitions of Oracles in DIRT In this subsection, we go over definitions of five different oracles written as sequences of generation actions, three of which are reimplementations of existing oracles in SQLancer.",
+            "section": "3.1 Definitions of Oracles in DIRT",
+            "page": 3
+          },
+          {
+            "mention_id": "M20",
+            "sentence": "We start by defining the first of the three SQLancer oracles we implemented, Pivoted Query Synthesis (PQS) [ 23], as a universally quantified proposition in Fig.",
+            "section": "3.1 Definitions of Oracles in DIRT",
+            "page": 3
+          },
+          {
+            "mention_id": "M25",
+            "sentence": "2b presents the GA for WHERE Extended case of Ternary Logic Partitioning (TLP) [ 21] oracle from SQLancer.",
+            "section": "3.1 Definitions of Oracles in DIRT",
+            "page": 3
+          },
+          {
+            "mention_id": "M26",
+            "sentence": "We can express other properties that are not present in SQLancer, including those fundamental to key-value stores such asDeleted rows should not be in the tablepresented in Fig.",
+            "section": "3.1 Definitions of Oracles in DIRT",
+            "page": 3
+          }
+        ],
+        "reasoning": "Three SQLancer oracles, PQS and TLP among them, are re-expressed as sequences of generation actions, and the paper then uses that abstraction to state properties SQLancer cannot express. Generalising the oracles into a new form is part of what the paper contributes, not a baseline it runs.",
+        "techniques": [
+          "pqs",
+          "tlp"
+        ]
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M2",
+          "M39",
+          "M40"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "We evaluate DIRTon Turso, an actively developed SQLite-compatible OLTP engine, and show that it finds 23 unique, confirmed bugs–significantly outperforming off-the-shelf SQLancer variants in terms of true positive rate and usefulness of bug reports.",
+            "section": null,
+            "page": 1
+          },
+          {
+            "mention_id": "M39",
+            "sentence": "This does not mean that DIRThas better generators or oracles than SQLancer.",
+            "section": "4.2 RQ2: How does DIRTcompare to SQLancer?",
+            "page": 5
+          },
+          {
+            "mention_id": "M40",
+            "sentence": "We compare the performance of DIRT for Turso against SQLancer with minimal modifications, which begets some questions worth discussing.",
+            "section": "5 Discussion",
+            "page": 5
+          }
+        ],
+        "reasoning": "DIRT is evaluated against off-the-shelf SQLancer variants on Turso and reports a better true positive rate and more useful reports, while noting the comparison does not show its generators or oracles are better."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M4"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "Within prior work, SQLancer is distinguished by using specialized test oracles that reveal logic bugs in DBMSs, and it has proven extremely successful at finding bugs in mature databases.",
+            "section": "1 Introduction",
+            "page": 1
+          }
+        ],
+        "reasoning": "M4 singles SQLancer out among prior work for its specialised oracles and says it has proven extremely successful at finding bugs."
+      }
+    },
+    "disagreements": [
+      "Two uses_infrastructure patterns fired on M6 and M48 and were suppressed as related work. That suppression was right for a different reason: the paper reimplements SQLancer's oracles rather than reusing its code, and says so."
+    ],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3810991_3811634.json b/_data/papers/paper_doi_10_1145_3810991_3811634.json
new file mode 100644
index 0000000..3f9c8e9
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3810991_3811634.json
@@ -0,0 +1,382 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:04:56Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3810991.3811634",
+    "title": "Cloud Analytics Benchmarking: Where We Are, What Is Missing, and Where We Should Go",
+    "authors": [
+      "Michail Georgoulakis Misegiannis",
+      "Till Steinert"
+    ],
+    "year": 2026,
+    "venue": "DBTest@SIGMOD",
+    "doi": "10.1145/3810991.3811634",
+    "arxiv_id": null,
+    "s2_paper_id": "1ee6ae0a6a863da15be2b876cc1493dd2c6acde5",
+    "url": "https://doi.org/10.1145/3810991.3811634",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3810991.3811634",
+      "retrieved_at": "2026-09-07T17:04:56Z",
+      "chars": 47540,
+      "content_sha256": "sha256:242956d4dc36358c680718301d625ab8988909cca24c706bdd75c64a7d1bc4c2"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 8,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1305
+      },
+      {
+        "number": "2",
+        "title": "Background and Related Work",
+        "start": 4975
+      },
+      {
+        "number": "2.1",
+        "title": "Telemetry-based Workload Traces",
+        "start": 5565
+      },
+      {
+        "number": "2.2",
+        "title": "Telemetry-based Query Synthesis",
+        "start": 7048
+      },
+      {
+        "number": "3",
+        "title": "Challenges of Telemetry-based Synthesis",
+        "start": 9889
+      },
+      {
+        "number": "3.1",
+        "title": "System-dependent Challenges",
+        "start": 10777
+      },
+      {
+        "number": "3.2",
+        "title": "Execution Context Challenges",
+        "start": 15324
+      },
+      {
+        "number": "3.3",
+        "title": "Structure Reconstruction Challenges",
+        "start": 18917
+      },
+      {
+        "number": "4",
+        "title": "Two Paths to Realistic Benchmarks",
+        "start": 26234
+      },
+      {
+        "number": "4.2",
+        "title": "Vision of Use-Case-Centric Query Synthesis",
+        "start": 29248
+      },
+      {
+        "number": "5",
+        "title": "Conclusion",
+        "start": 35748
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2026. Substrait: Cross-Language Serialization for Relational Algebra. https: //substrait.io. Accessed: 2026-04-24.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Google BigQuery. 2026. Bitcoin Blockchain Historical Data — kaggle.com. https:// www.kaggle.com/datasets/bigquery/bitcoin-blockchain. [Accessed 21-03-2026].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "dbt labs. [n. d.]. GitHubdbt-labs/jaffle-shop: An open source sandbox project exploring dbt workflows via a fictional sandwich shop’s data. — github.com. https://github.com/dbt-labs/jaffle-shop. [Accessed 19-03-2026].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Shaleen Deep, Anja Gruenheid, Kruthi Nagaraj, Hiro Naito, Jeffrey F. Naughton, and Stratis Viglas. 2022. DIAMETRICS: benchmarking query engines at scale. Commun. ACM 65, 12 (2022), 105–112.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Bailu Ding, Surajit Chaudhuri, Johannes Gehrke, and Vivek R. Narasayya. 2021. DSB: A Decision Support Benchmark for Workload-Driven and Traditional Database Systems. Proc. VLDB Endow. 14, 13 (2021), 3376–3388.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Dominik Durner, Lennart Espe, Jana Giceva, and Anja Gruenheid. 2024. TracEx: Understanding and Analyzing Database Traces. In CIDR. www.cidrdb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Pascal Ginter and Viktor Leis. 2026. Active Data Lakes: Regaining Physical Data Independence Without Losing Interoperability. Proc. VLDB Endow. (2026).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Tobias Götz, Daniel Ritter, and Jana Giceva. 2025. LakeVilla: A Modular and Non-Invasive Toolbox for Lakehouse Transactions. arXiv:2504.20768 [cs.DB] https://arxiv.org/abs/2504.20768",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Benjamin Hilprecht and Carsten Binnig. 2022. Zero-Shot Cost Models for Out-ofthe-box Learned Cost Prediction. Proc. VLDB Endow. 15, 11 (2022), 2361–2374.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Yaroslav Isaienkov. 2025. Coffee Sales. doi:10.34740/KAGGLE/DSV/11159944",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Tengjun Jin, Yuxuan Zhu, and Daniel Kang. 2025. ELT-Bench: An End-to-End Benchmark for Evaluating AI Agents on ELT Pipelines. Proc. VLDB Endow. 19, 2 (2025), 84–98.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Skander Krid, Mihail Stoian, and Andreas Kipf. 2025. Redbench: A Benchmark Reflecting Real Workloads. CoRR abs/2506.12488 (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Jiale Lao and Immanuel Trummer. 2025. Demonstrating SQLBarber: Leveraging Large Language Models to Generate Customized and Realistic SQL Workloads. InSIGMOD Conference Companion. ACM, 151–154.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Jiale Lao and Immanuel Trummer. 2025. SQLBarber: A System Leveraging Large Language Models to Generate Customized and Realistic SQL Workloads. CoRR abs/2507.06192 (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2015. How Good Are Query Optimizers, Really? Proc. VLDB Endow. 9, 3 (2015), 204–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Shu Liu, Soujanya Ponnapalli, Shreya Shankar, Sepanta Zeighami, Alan Zhu, Shubham Agarwal, Ruiqi Chen, Samion Suwito, Shuo Yuan, Ion Stoica, Matei Zaharia, Alvin Cheung, Natacha Crooks, Joseph Gonzalez, and Aditya G. Parameswaran. 2026. Supporting Our AI Overlords: Redesigning Data Systems to be Agent-First. InCIDR. www.cidrdb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Chengzhi Lu, Kejiang Ye, Guoyao Xu, Cheng-Zhong Xu, and Tongxin Bai. 2017. Imbalance in the cloud: An analysis on Alibaba cluster trace. In IEEE BigData. IEEE Computer Society, 2884–2892.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Michail Georgoulakis Misegiannis, Daniel Ritter, Viktor Leis, and Jana Giceva. 2025. CloudGlide: Deconstructing the Landscape of Cloud-Based Analytics. Proc. VLDB Endow. 18, 13 (2025), 5638–5651.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "NFL. 2026. NFL Big Data Bowl 2026 - Prediction — kaggle.com. https://www. kaggle.com/competitions/nfl-big-data-bowl-2026-prediction/data. [Accessed 21-03-2026].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Mosha Pasumansky and Benjamin Wagner. 2022. Assembling a Query Engine From Spare Parts. In CDMS@VLDB.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 22,
+      "text": "Tobias Schmidt, Andreas Kipf, Dominik Horn, Gaurav Saxena, and Tim Kraska. 2024. Predicate Caching: Query-Driven Secondary Indexing for Cloud Data Warehouses. In SIGMOD Conference Companion. ACM, 347–359.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Tobias Schmidt, Viktor Leis, Peter Boncz, and Thomas Neumann. 2025. SQLStorm: Taking Database Benchmarking into the LLM Era. Proc. VLDB Endow. 18, 11 (2025), 4144–4157.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Robert Schulze, Tom Schreiber, Ilya Yatsishin, Ryadh Dahimene, and Alexey Milovidov. 2024. ClickHouse-Lightning Fast Analytics for Everyone. Proc. VLDB Endow. 17, 12 (2024), 3731–3744.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Vijaya Shree Raja Sekaran. 2019. Food Preferences — kaggle.com. https://www. kaggle.com/datasets/vijayashreer/food-preferences. [Accessed 21-03-2026].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. 2015. SQLsmith. https: //github.com/anse1/sqlsmith. Accessed: 2026-03-12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Till Steinert, Maximilian Kuschewski, and Viktor Leis. 2026. Cloudspecs: Cloud Hardware Evolution Through the Looking Glass. In CIDR. www.cidrdb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Murray Stokely, Neel Nadgir, Jack Peele, and Orestis Kostakis. 2025. Shaved Ice: Optimal Compute Resource Commitments for Dynamic Multi-Cloud Workloads. InICPE. 124–135.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Michael Stonebraker. 2014. Why the “Data Lake” is Really a “Data Swamp”. BLOG@CACM. [Accessed 21-03-2026].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Jan Vincent Szlang, Sebastian Breß, Sebastian Cattes, Jonathan Dees, Florian Funke, Max Heimel, Michel Oleynik, Ismail Oukid, and Tobias Maltenberger. 2025. Workload Insights From the Snowflake Data Cloud: What Do Production Analytic Queries Really Look Like? Proc. VLDB Endow. 18, 12 (2025), 5126–5138.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Alexander van Renen, Dominik Horn, Pascal Pfeil, Kapil Vaidya, Wenjian Dong, Murali Narayanaswamy, Zhengchun Liu, Gaurav Saxena, Andreas Kipf, and Tim Kraska. 2024. Why TPC Is Not Enough: An Analysis of the Amazon Redshift Fleet. Proc. VLDB Endow. 17, 11 (2024), 3694–3706.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Alexander van Renen and Viktor Leis. 2023. Cloud Analytics Benchmark. Proc. VLDB Endow. 16, 6 (2023), 1413–1425.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Midhul Vuppalapati, Justin Miron, Rachit Agarwal, Dan Truong, Ashish Motivala, and Thierry Cruanes. 2020. Building An Elastic Query Engine on Disaggregated Storage. In NSDI. USENIX Association, 449–462.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Chengcheng Wan, Yiwen Zhu, Joyce Cahoon, Wenjing Wang, Katherine Lin, Sean Liu, Raymond Truong, Neetu Singh, Alexandra M. Ciortea, Konstantinos Karanasos, and Subru Krishnan. 2023. Stitcher: Learned Workload Synthesis from Historical Performance Footprints. In EDBT. OpenProceedings.org, 417–423.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Zhengle Wang, Yanfei Zhang, and Chunwei Liu. 2026. ResQ: Realistic Performance-Aware Query Generation. arXiv preprint arXiv:2602.02999 (2026).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Johannes Wehrstein, Roman Heinrich, Mihail Stoian, Skander Krid, Martin Stemmer, Andreas Kipf, Carsten Binnig, and Muhammad El-Hindi. 2025. Redbench: Workload Synthesis From Cloud Traces. CoRR abs/2511.13059 (2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Jiaqi Yan, Qiuye Jin, Shrainik Jain, Stratis D. Viglas, and Allison W. Lee. 2018. Snowtrail: Testing with Production Queries on a Cloud Database. In DBTest@SIGMOD. ACM, 4:1–4:6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Matei Zaharia, Ali Ghodsi, Reynold Xin, and Michael Armbrust. 2021. Lakehouse: A New Generation of Open Platforms that Unify Data Warehousing and Advanced Analytics. In CIDR. www.cidrdb.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Yan Zhou, Chunwei Liu, Bhuvan Urgaonkar, Zhengle Wang, Magnus Mueller, Chao Zhang, Songyue Zhang, Pascal Pfeil, Dominik Horn, Zhengchun Liu, Davide Pagano, Tim Kraska, Samuel Madden, and Ju Fan. 2025. PBench: Workload Synthesizer with Real Statistics for Cloud Analytics Benchmarking. Proc. VLDB Endow. 18, 11 (2025), 3883–3895. ASQL Structural Features Table 1 lists the 27 structural features extra",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 21,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 21,37]",
+      "technique": "tlp",
+      "sentence": "Related work also considers database testing [ 21,37].",
+      "context_before": "queries are transparent by construction rather than reverse-engineered. As a proof of concept (Section 4.2), we show that open-source tools such as dbt (ELT) and Apache Superset (BI) can generate workloads with characteristics found in real traces, such as stringand join-heavy queries, while exposing the actual SQL rather than only telemetry. We illustrate this with an e-commerce revenue analysis based on the dbt Jaffle Shop project [3]. 2 Background and Related Work With continuously changing workloads and requirements, there is a rich body of work on advancing database benchmarking [ 4, 5].",
+      "context_after": "In the following, we focus on workload traces and synthesis approaches most relevant to benchmark design. Other workload traces [ 17], trace-analysis tools [ 6], and query generation methods [ 23,26] are related, but orthogonal to our focus. 38 DBTest ’26, May 31-June 05, 2026, Bengaluru, India Michail Georgoulakis Misegiannis, Till Steinert, Viktor Leis & Muhammad El-Hindi 2.1 Telemetry-based Workload Traces To date, two major cloud database providers have published telemetry workload traces that contain operational measurements of system state during query execution. Although these traces r",
+      "section": "2 Background and Related Work",
+      "page": 1,
+      "char_offset": 5132,
+      "cited_reference": {
+        "number": 21,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang. 4, OOPSLA (2020), 211:1–211:30.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "A position paper on benchmarking cloud analytics. It observes that vendor workload traces carry execution telemetry but omit SQL text for privacy, and asks whether realistic workloads can be synthesised from telemetry alone. The authors argue this works at the workload level but not the query level, since telemetry does not uniquely determine query structure, and propose enriching traces with transferable features and building use-case-centric benchmarks instead.",
+    "narrative": "One citation, listing database testing among the related work.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3810991_3811637.json b/_data/papers/paper_doi_10_1145_3810991_3811637.json
new file mode 100644
index 0000000..69c6957
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3810991_3811637.json
@@ -0,0 +1,355 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:22:55Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3810991.3811637",
+    "title": "Boosting DBMS Test Coverage via LLM-Driven SQL Generation",
+    "authors": [
+      "Eslam Abdelkarim",
+      "Carsten Binnig",
+      "Anupam Sanghi"
+    ],
+    "year": 2026,
+    "venue": "DBTest@SIGMOD",
+    "doi": "10.1145/3810991.3811637",
+    "arxiv_id": null,
+    "s2_paper_id": "bc956d15e20b65ccf568bf75cf6b3c5689d8f9fa",
+    "url": "https://doi.org/10.1145/3810991.3811637",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.1145/3810991.3811637",
+      "retrieved_at": "2026-09-10T15:22:55Z",
+      "chars": 24720,
+      "content_sha256": "sha256:8fc69fc4eb0136cf0b800cf50be1878a87658439e91ea25d7db7241ee1679516"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 5,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1769
+      },
+      {
+        "number": "2",
+        "title": "System Architecture",
+        "start": 6921
+      },
+      {
+        "number": "2.1",
+        "title": "Coverage Measurement",
+        "start": 7318
+      },
+      {
+        "number": "2.3",
+        "title": "Context Construction",
+        "start": 8727
+      },
+      {
+        "number": "2.4",
+        "title": "Query Generation",
+        "start": 10332
+      },
+      {
+        "number": "2.5",
+        "title": "Feedback",
+        "start": 11378
+      },
+      {
+        "number": "3",
+        "title": "Evaluation",
+        "start": 12383
+      },
+      {
+        "number": "3.1",
+        "title": "Experimental Setup",
+        "start": 12753
+      },
+      {
+        "number": "3.2",
+        "title": "Coverage Improvement",
+        "start": 15561
+      },
+      {
+        "number": "3.3",
+        "title": "Ablation Study",
+        "start": 17812
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Yu Liang, Song Liu, and Hong Hu. 2022. Detecting Logical Bugs of DBMS with Coverage-based Guidance. In31st USENIX Security Symposium (USENIX Security’22). USENIX Association, 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "PostgreSQL Global Development Group. 2026. PostgreSQL Documentation: Regression Tests. https://www.postgresql.org/docs/current/regress.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Manuel Rigger. 2022. SQLancer. https://github.com/sqlancer/sqlancer. Accessed 2026-02-25.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020). ACM, 1140–1152. doi:10.1145/3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (2020), 30 pages. doi:10.1145/3428279",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 6,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. InProceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). USENIX Association, Article 38, 16 pages.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 7,
+      "text": "Sujoy Roy Chowdhury, Giriprasad Sridhara, A K Raghavan, Joy Bose, Sourav Mazumdar, Hamender Singh, Srinivasan Bajji Sugumaran, and Ricardo Britto.2025. Static Program Analysis Guided LLM Based Unit Test Generation. In Proceedings of the 8th International Conference on Data Science and Management of Data (CODS-COMAD ’24). ACM, 279–283. doi:10.1145/3703323.3703742",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Tobias Schmidt, Viktor Leis, Peter Boncz, and Thomas Neumann. 2025. SQLStorm: Taking Database Benchmarking into the LLM Era.PVLDB18, 11 (2025), 4144–4157. doi:10.14778/3749646.3749683",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Transaction Processing Performance Council. 2022. TPC BenchmarkTMH Standard Specification. https://www.tpc.org/tpch/ Version 3.0.1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Transaction Processing Performance Council. 2024. TPC BenchmarkTMDS Standard Specification. https://www.tpc.org/tpcds/ Version 4.0.0.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. 2021. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. InProceedings of the 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP ’21). IEEE Press, 328–337. doi:10.1109/ICSE-SEIP52600.2021.00042",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Jin Wei, Ping Chen, Kangjie Lu, Jun Dai, and Xiaoyan Sun. 2026. SQLaser: Detecting database management system (DBMS) logic bugs with clause-guided fuzzing. Journal of Computer Security34, 1 (2026), 3–28. doi:10.1177/0926227X251370258",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments. arXiv:2505.02012 [cs.SE] 53",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 3,
+      "text": "Manuel Rigger. 2022. SQLancer. https://github.com/sqlancer/sqlancer. Accessed 2026-02-25.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 4,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2020). ACM, 1140–1152. doi:10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 5,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding bugs in database systems via query partitioning.Proc. ACM Program. Lang.4, OOPSLA, Article 211 (2020), 30 pages. doi:10.1145/3428279",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 6,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing database engines via pivoted query synthesis. InProceedings of the 14th USENIX Conference on Operating Systems Design and Implementation (OSDI’20). USENIX Association, Article 38, 16 pages.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 13,
+      "text": "Suyang Zhong and Manuel Rigger. 2025. Testing Database Systems with Large Language Model Synthesized Fragments. arXiv:2505.02012 [cs.SE] 53",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "prints the arXiv id of the paper introducing shqvel"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQL fuzzing tools such as SQLancer [3, 6] automatically generate syntactically valid queries and use logic-testing oracles (e.",
+      "context_before": "play a central role in validating correctness and stability across a wide range of features. For example, the PostgreSQL regression tests [ 2] provide a comprehensive set of SQL tests for the system’s core functionality, reaching over 80% line coverage. Standard benchmark suites such as TPC-H [ 9] and TPC-DS [ 10] provide widely adopted, carefully designed workloads for evaluating and comparing DBMS performance under realistic decision-support scenarios. In addition to such manually curated suites, a variety of automated approaches have been proposed for generating and executing SQL workloads.",
+      "context_after": "g., Pivoted Query Synthesis) to uncover logic bugs in DBMSs. Coverage-guided fuzzers such as RATEL [ 11] and SQLRight [ 1] incorporate code coverage feedback to iteratively refine and prioritize SQL inputs, facilitating effective exploration of diverse execution paths in database systems. More recently, LLM-based systems such as SQLStorm [ 8] leverage semantic understanding to generate diverse, realistic SQL workloads at scale, showing that LLMs can efficiently synthesize large corpora of queries that reflect real-world patterns. Existing Approaches are not Enough.While the aforementioned appr",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3367,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": ", Pivoted Query Synthesis) to uncover logic bugs in DBMSs.",
+      "context_before": "on tests [ 2] provide a comprehensive set of SQL tests for the system’s core functionality, reaching over 80% line coverage. Standard benchmark suites such as TPC-H [ 9] and TPC-DS [ 10] provide widely adopted, carefully designed workloads for evaluating and comparing DBMS performance under realistic decision-support scenarios. In addition to such manually curated suites, a variety of automated approaches have been proposed for generating and executing SQL workloads. SQL fuzzing tools such as SQLancer [3, 6] automatically generate syntactically valid queries and use logic-testing oracles (e.g.",
+      "context_after": "Coverage-guided fuzzers such as RATEL [ 11] and SQLRight [ 1] incorporate code coverage feedback to iteratively refine and prioritize SQL inputs, facilitating effective exploration of diverse execution paths in database systems. More recently, LLM-based systems such as SQLStorm [ 8] leverage semantic understanding to generate diverse, realistic SQL workloads at scale, showing that LLMs can efficiently synthesize large corpora of queries that reflect real-world patterns. Existing Approaches are not Enough.While the aforementioned approaches are useful, they face the following challenges. (a) L",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 3496,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Logic-testing tools such as SQLancer [ 3,6] generate queries and pair them with metamorphic or partitioning-based oracles (e.",
+      "context_before": "tions. Overall, each component contributes to clear coverage gains. Figure 3: Line coverage progress over iterations for each ablation configuration on the PostgreSQL executor (TPC-H). The full system achieves the highest coverage, while removing individual components shows their relative contribution.4 Related Work Two active research directions are particularly relevant to our work: coverage-guided fuzzing for database systems and LLM-based SQL generation for testing. We briefly survey representative systems in each area and position Quover relative to them. Fuzzing-based coverage targeting.",
+      "context_after": "g., Pivoted Query Synthesis (PQS) [ 6], NoREC [ 4], and Ternary Logic Partitioning (TLP) [ 5]) to uncover logic bugs and exercise diverse execution paths in DBMSs, but they do not explicitly optimize for exercising uncovered code regions. Coverage-guided fuzzers such as SQLRight [ 1], RATEL [ 11], and SQLaser [ 12] steer SQL generation using code coverage feedback and validity oracles, increasing bug-finding effectiveness over pure differential testing. Across these systems, coverage is used as a mutation signal: they lack semantic understanding of what an uncovered function computes and there",
+      "section": "3.3 Ablation Study",
+      "page": 4,
+      "char_offset": 18960,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": ", Pivoted Query Synthesis (PQS) [ 6], NoREC [ 4], and Ternary Logic Partitioning (TLP) [ 5]) to uncover logic bugs and exercise diverse execution paths in DBMSs, but they do not explicitly optimize for exercising uncovered code regions.",
+      "context_before": "blation configuration on the PostgreSQL executor (TPC-H). The full system achieves the highest coverage, while removing individual components shows their relative contribution.4 Related Work Two active research directions are particularly relevant to our work: coverage-guided fuzzing for database systems and LLM-based SQL generation for testing. We briefly survey representative systems in each area and position Quover relative to them. Fuzzing-based coverage targeting.Logic-testing tools such as SQLancer [ 3,6] generate queries and pair them with metamorphic or partitioning-based oracles (e.g.",
+      "context_after": "Coverage-guided fuzzers such as SQLRight [ 1], RATEL [ 11], and SQLaser [ 12] steer SQL generation using code coverage feedback and validity oracles, increasing bug-finding effectiveness over pure differential testing. Across these systems, coverage is used as a mutation signal: they lack semantic understanding of what an uncovered function computes and therefore cannot construct queries that target specific internal code paths. Quover instead provides the LLM with function descriptions and code snippets to enable semantics-aware targeting of uncovered regions. LLM-based coverage targeting.Se",
+      "section": "3.3 Ablation Study",
+      "page": 4,
+      "char_offset": 19087,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "ShQveL",
+      "technique": null,
+      "sentence": "ShQveL [ 13] integrates LLM-generated SQL fragments into an existing generator via “SQL sketching”, validating successful fragments and reusing them to increase feature coverage and uncover new DBMS bugs.",
+      "context_before": "an uncovered function computes and therefore cannot construct queries that target specific internal code paths. Quover instead provides the LLM with function descriptions and code snippets to enable semantics-aware targeting of uncovered regions. LLM-based coverage targeting.Several systems leverage Large Language Models (LLMs) to generate SQL workloads and tests. SQLStorm [ 8] uses GPT-based models and carefully engineered prompts to synthesize diverse analytical SQL queries for multiple DBMSs, focusing on workload diversity and realistic benchmarking rather than explicit coverage objectives.",
+      "context_after": "Beyond databases, LLM-based test and data generation [ 7] has been shown to improve coverage in general software testing.Quover combines fine-grained coverage feedback with LLM-driven SQL generation in a closed loop, explicitly targeting uncovered DBMS functions without relying on external query generators. 5 Conclusion This paper presents Quover, an automated approach for improving DBMS test coverage through LLM-generated SQL queries. Quover demonstrates practical effectiveness, achieving coverage gains on PostgreSQL and DuckDB. The iterative, coverage-guided methodology successfully leverag",
+      "section": "3.3 Ablation Study",
+      "page": 4,
+      "char_offset": 20245,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:21:29Z",
+    "is_model_written": true,
+    "summary": "Quover uses a large language model to generate SQL aimed at raising code coverage in DBMS testing. It works iteratively: it identifies functions in the DBMS source not yet covered, prompts the model with context including function descriptions and code snippets, generates targeted queries, and validates their effect. The authors report over 57% coverage within a few hours, a 14% improvement over the automated systems they compare with.",
+    "narrative": "SQLancer is the reference point for what coverage-oriented generation is being contrasted with: it is described as generating syntactically valid queries paired with logic-testing oracles such as PQS, NoREC and TLP. ShQveL, from the same project, is also cited for integrating LLM-generated fragments into an existing generator.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "background",
+      "M4": "definition",
+      "M5": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1145_3828685.json b/_data/papers/paper_doi_10_1145_3828685.json
new file mode 100644
index 0000000..aa1f728
--- /dev/null
+++ b/_data/papers/paper_doi_10_1145_3828685.json
@@ -0,0 +1,476 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:24:04Z",
+  "paper": {
+    "id": "paper:doi:10.1145/3828685",
+    "title": "Programmable Property-Based Testing",
+    "authors": [
+      "A. Keles",
+      "Justine Frank",
+      "Ceren Mert",
+      "Harrison Goldstein",
+      "Leonidas Lampropoulos"
+    ],
+    "year": 2026,
+    "venue": "Proceedings of the ACM on Programming Languages",
+    "doi": "10.1145/3828685",
+    "arxiv_id": "2602.18545",
+    "s2_paper_id": "1b21dbf4dfd10358f77bf5ba3b9ba39c81e18b25",
+    "url": "https://doi.org/10.1145/3828685",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2602.18545",
+      "retrieved_at": "2026-09-08T15:24:04Z",
+      "chars": 97220,
+      "content_sha256": "sha256:18f9d2692798274f96304f464e9dbafa2f863d4d06b3579434b5ad8e14387231"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 32,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1587
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND: PROPERTY RUNNERS",
+        "start": 9212
+      },
+      {
+        "number": "2.1",
+        "title": "Simple Generational Property Runner",
+        "start": 9930
+      },
+      {
+        "number": "2.2",
+        "title": "Integrated Shrinking Property Runner",
+        "start": 11562
+      },
+      {
+        "number": "2.5",
+        "title": "Combinatorial Property Runner",
+        "start": 17983
+      },
+      {
+        "number": "2.6",
+        "title": "Parallel Property Runner",
+        "start": 18680
+      },
+      {
+        "number": "2.7",
+        "title": "Discussion",
+        "start": 20538
+      },
+      {
+        "number": "3",
+        "title": "PROPERTY LANGUAGES",
+        "start": 21190
+      },
+      {
+        "number": "3.1",
+        "title": "Background Terminology: Deep and Shallow Embeddings",
+        "start": 22048
+      },
+      {
+        "number": "3.2",
+        "title": "Status Quo: A Shallow Property Language",
+        "start": 22913
+      },
+      {
+        "number": "3.3",
+        "title": "Hypothetical: A Deep Property Language",
+        "start": 25626
+      },
+      {
+        "number": "3.4",
+        "title": "Proposal: A Mixed Property Language",
+        "start": 26364
+      },
+      {
+        "number": "6.1",
+        "title": "Property Runners with DBAS",
+        "start": 44641
+      },
+      {
+        "number": "6.2",
+        "title": "Comparison of Deep and Shallow Embeddings",
+        "start": 53268
+      },
+      {
+        "number": "6.3",
+        "title": "DBAS-powered Experiments",
+        "start": 57119
+      },
+      {
+        "number": "6.4",
+        "title": "Comparison of Integrated Shrinking with External Shrinking",
+        "start": 64086
+      },
+      {
+        "number": "6.5",
+        "title": "Parallelizing Property-Based Testing",
+        "start": 67768
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. InThe 45th International Conference on Software Engineering (ICSE’23).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 2,
+      "text": "Marcel Böhme, Valentin J. M. Manès, and Sang Kil Cha. 2020. Boosting fuzzer efficiency: an information theoretic perspective. InESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, Prem Devanbu, Myra B. Cohen, and Thomas Zimmermann (Eds.). ACM, 678–689. https://doi.org/10.1145/336808",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. 2019. Coverage-Based Greybox Fuzzing as Markov Chain. IEEE Trans. Software Eng.45, 5 (2019), 489–506. https://doi.org/10.1109/TSE.2017.2785841",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Marcel Böhme, Van-Thuan Pham, Manh-Dung Nguyen, and Abhik Roychoudhury. 2017. Directed Greybox Fuzzing. In Proceedings of the 2017 ACMSIGSAC Conference on Computer and Communications Security(Dallas, Texas, USA)(CCS ’17). Association for Computing Machinery, New York, NY, USA, 2329–2344. https://doi.org/10.1145/3133956.3134020",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Richard J. Boulton, Andy Gordon, Michael J. C. Gordon, John Harrison, John Herbert, and John Van Tassel. 1992. Experience with Embedding Hardware Description Languages in HOL. InProceedings of the IFIP TC10/WG 10.2 International Conference on Theorem Provers in Circuit Design: Theory, Practice and Experience(proceedings of the ifip tc10/wg 10.2 international conference on theorem provers in circui",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Cameron Bytheway. [n. d.].Bolero: A fuzzing and property testing front-end framework for Rust.https://github.com/ camshaft/bolero",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Jacques Carrete, Oleg Kiselyov, and Chung-Chieh Shan. 2009. Finally tagless, partially evaluated: Tagless staged interpreters for simpler typed languages.Journal of Functional Programming19, 5 (2009), 509–543. https://doi.org/10. 1017/S0956796809007205 Programmable Property-Based Testing",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Jinfu Chen, Shengran Wang, Saihua Cai, Chi Zhang, Haibo Chen, Jingyi Chen, and Jianming Zhang. 2023. A Novel Coverage-guided Greybox Fuzzing based on Power Schedule Optimization with Time Complexity. InProceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering(Rochester, MI, USA)(ASE ’22). Association for Computing Machinery, New York, NY, USA, Article 172, 5 pages",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Adam Chlipala. 2008. Parametric Higher-order Abstract Syntax for Mechanized Semantics. InProceedings of the 13th ACMSIGPLAN International Conference on Functional Programming(Victoria, BC, Canada)(ICFP ’08). ACM, New York, NY, USA, 143–156. https://doi.org/10.1145/1411204.1411226",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Koen Claessen and John Hughes. 2000. QuickCheck: a lightweight tool for random testing of Haskell programs. In Proceedings of the 5th ACMSIGPLAN International Conference on Functional Programming. ACM, 268–279. https: //doi.org/10.1145/351240.351266",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Koen Claessen and Michał Pałka. 2013. Splittable pseudorandom number generators using cryptographic hashing. In ACMSIGPLAN Symposium on Haskell. ACM, 47–58. http://publications.lib.chalmers.se/records/fulltext/183348/local_ 183348.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Edsko de Vries. 2023. falsify: Internal Shrinking Reimagined for Haskell. InProceedings of the 16th ACMSIGPLAN International Haskell Symposium(Seattle, WA, USA)(Haskell 2023). Association for Computing Machinery, New York, NY, USA, 97–109. https://doi.org/10.1145/3609026.3609733",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Stephen Dolan. 2017. Property Fuzzing for OCaml. https://github.com/stedolan/crowbar.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Andrea Fioraldi, Dominik Maier, Heiko Eißfeldt, and Marc Heuse. 2020. AFL++: combining incremental steps of fuzzing research. InProceedings of the 14th USENIX Conference on Offensive Technologies (WOOT’20). USENIX Association, USA, Article 10, 1 pages.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Harrison Goldstein, Joseph W. Cutler, Daniel Dickstein, Benjamin C. Pierce, and Andrew Head. 2024. Property-Based Testing in Practice. InProceedings of the IEEE/ACM 46th International Conference on Software Engineering (ICSE ’24, Vol. 187). Association for Computing Machinery, Lisbon, Portugal, 1–13. https://doi.org/10.1145/3597503.3639581",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Harrison Goldstein, John Hughes, Leonidas Lampropoulos, and Benjamin C. Pierce. 2021. Do Judge a Test by its Cover: Combining Combinatorial and Property-Based Testing. InProceedings of the European Symposium on Programming (ESOP) (Lecture Notes in Computer Science). https://doi.org/10.1007/978-3-030-72019-3_10",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Harrison Goldstein and Benjamin C. Pierce. 2022. Parsing Randomness.Proceedings of the ACM on Programming Languages6, OOPSLA2 (Oct. 2022), 128:89–128:113. https://doi.org/10.1145/3563291",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Alex Groce, Chaoqiang Zhang, Eric Eide, Yang Chen, and John Regehr. 2012. Swarm Testing. InProceedings of the 2012 International Symposium on Software Testing and Analysis(Minneapolis, MN, USA)(ISSTA 2012). ACM, New York, NY, USA, 78–88. https://doi.org/10.1145/2338965.2336763",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Zac Hatfield-Dodds and contributors. [n. d.].HypoFuzz: Adaptive fuzzing of Hypothesis tests. https://github.com/ZacHD/hypofuzz",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Paul Hudak. 1996. Building domain-specific embedded languages.Comput. Surveys28, 4es (Dec. 1996), 196. https: //doi.org/10.1145/242224.242477",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Robert Krook, Nicholas Smallbone, Bo Joel Svensson, and Koen Claessen. 2024. QuickerCheck: Implementing and Evaluating a Parallel Run-Time for QuickCheck. arXiv:2404.16062 [cs.PL] https://arxiv.org/abs/2404.16062",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Leonidas Lampropoulos. 2018.Random Testing for Language Design. Ph. D. Dissertation. University of Pennsylvania.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Leonidas Lampropoulos, Michael Hicks, and Benjamin C. Pierce. 2019. Coverage Guided, Property Based Testing. In Proceedings of the ACM Conference on Object-Oriented Programming Languages, Systems, and Applications (OOPSLA). https://doi.org/10.1145/3360607",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Leonidas Lampropoulos, Zoe Paraskevopoulou, and Benjamin C. Pierce. 2018. Generating Good Generators for Inductive Relations. InProceedings of the ACM Conference on Principles of Programming Languages (POPL). https: //doi.org/10.1145/3158133",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Caroline Lemieux, Rohan Padhye, Koushik Sen, and Dawn Song. 2018. PerfFuzz: automatically generating pathological inputs. InProceedings of the 27th ACMSIGSOFT International Symposium on Software Testing and Analysis(Amsterdam, Netherlands)(ISSTA 2018). Association for Computing Machinery, New York, NY, USA, 254–265. https://doi.org/10. 1145/3213846.3213874",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Caroline Lemieux and Koushik Sen. 2018. FairFuzz: a targeted mutation strategy for increasing greybox fuzz testing coverage. InProceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering(Montpellier, France)(ASE ’18). Association for Computing Machinery, New York, NY, USA, 475–485. https://doi.org/10.1145/ 3238147.3238176",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Yao Li and Stephanie Weirich. 2022. Program Adverbs and Tlön Embeddings.Proceedings of the ACM on Programming Languages6. https://doi.org/10.1145/3547632",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Yao Li and Stephanie Weirich. 2022. Program adverbs and Tlön embeddings.Proc. ACM Program. Lang.6, ICFP (2022), 312–342. https://doi.org/10.1145/3547632 Alperen Keles, Justine Frank, Ceren Mert, Harrison Goldstein, and Leonidas Lampropoulos",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Andreas Löscher and Konstantinos Sagonas. 2017. Targeted Property-Based Testing. InProceedings of the 26th ACMSIGSOFT International Symposium on Software Testing and Analysis. ACM, 46–56. https://doi.org/10.1145/3092703. 3092711",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "David Maciver and Alastair F. Donaldson. 2020. Test-Case Reduction via Test-Case Generation: Insights from the Hypothesis Reducer (Tool Insights Paper). In34th European Conference on Object-Oriented Programming (LIPIcs, Vol. 166). Schloss Dagstuhl-Leibniz-Zentrum für Informatik, 13:1–13:27. https://doi.org/10.4230/LIPIcs.ECOOP.2020.13",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "David R. MacIver. 2016. Hypothesis: Property-Based Testing for Python. https://hypothesis.works/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Kazutaka Matsuda, Samantha Frohlich, Meng Wang, and Nicolas Wu. 2023. Embedding by Unembedding.Proc. ACM Program. Lang.7, ICFP, Article 189 (Aug. 2023), 47 pages. https://doi.org/10.1145/3607830",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Segev Elazar Mittelman, Alvin Resnick, Ivan Perez, Alwyn Goodloe, and Leonidas Lampropoulos. 2023. Don’t Go Down the Rabbit Hole: Reprioritizing Enumeration for Property-Based Testing. InProceedings of the ACMSIGPLAN International Symposium on Haskell.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Rohan Padhye, Caroline Lemieux, and Koushik Sen. 2019. JQF: coverage-guided property-based testing in Java. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis(Beijing, China)(ISSTA 2019). Association for Computing Machinery, New York, NY, USA, 398–401. https://doi.org/10.1145/3293882.3339002",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Mike Papadakis, and Yves Le Traon. 2019. Semantic Fuzzing with Zest. InProceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis. ACM, 329–340. https://doi.org/10.1145/3293882.3330576",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Mike Papadakis, and Yves Le Traon. 2019. Validity fuzzing and parametric generators for effective random testing. InProceedings of the 41st International Conference on Software Engineering: Companion Proceedings, ICSE 2019, Montreal, QC, Canada, May 25-31, 2019, Joanne M. Atlee, Tevfik Bultan, and Jon Whittle (Eds.). IEEE / ACM, 266–267. https://doi.org",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Laurent Simon, and Hayawardh Vijayakumar. 2019. FuzzFactory: domain-specific fuzzing with waypoints.Replication Package for \"FuzzFactory: Domain-Specific Fuzzing with Waypoints\" 3, OOPSLA (Oct. 2019), 174:1–174:29. https://doi.org/10.1145/3360600",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Zoe Paraskevopoulou, Catalin Hritcu, Maxime Denes, Leonidas Lampropoulos, and Benjamin C. Pierce. 2015. Foundational Property-Based Testing. In6th International Conference on Interactive Theorem Proving (ITP). https: //doi.org/10.1007/978-3-319-22102-1_22",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Jiwon Park, Dominik Winterer, Chengyu Zhang, and Zhendong Su. 2021. Generative type-aware mutation for testing SMT solvers.Proc. ACM Program. Lang.5, OOPSLA (2021), 1–19. https://doi.org/10.1145/3485529",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Theofilos Petsios, Jason Zhao, Angelos D. Keromytis, and Suman Jana. 2017. SlowFuzz: Automated Domain-Independent Detection of Algorithmic Complexity Vulnerabilities. InProceedings of the 2017 ACMSIGSAC Conference on Computer and Communications Security(Dallas, Texas, USA)(CCS ’17). Association for Computing Machinery, New York, NY, USA, 2155–2168. https://doi.org/10.1145/3133956.3134073",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Bogdan Popa. 2020. RackCheck: Property-Based Testing for Racket. https://docs.racket-lang.org/rackcheck/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Jacob Prinz, Alex Kavvos, and Leonidas Lampropoulos. 2022. Deeper Shallow Embeddings. In13th International Conference on Interactive Theorem Proving (ITP) (Lecture Notes in Computer Science). https://doi.org/10.4230/LIPIcs.ITP. 2022.28",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "John Regehr, Yang Chen, Pascal Cuoq, Eric Eide, Chucky Ellison, and Xuejun Yang. 2012. Test-case reduction for C compiler bugs. InACM SIGPLAN Conference on Programming Language Design and Implementation, PLDI ’12, Beijing, China-June 11 - 16, 2012. 335–346. https://doi.org/10.1145/2254064.2254104",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Jessica Shi, Alperen Keles, Harrison Goldstein, Benjamin C. Pierce, and Leonidas Lampropoulos. 2023. Etna: An Evaluation Platform for Property-Based Testing (Experience Report).Proc. ACM Program. Lang.7, ICFP, Article 218 (aug 2023), 17 pages. https://doi.org/10.1145/3607860",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "SQLancer contributors. [n. d.].SQLancer: Automated testing to find logic and performance bugs in database systems. https://github.com/sqlancer/sqlancer",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 46,
+      "text": "Jacob Stanley. 2019. Hedgehog: Release with Confidence. https://hackage.haskell.org/package/hedgehog/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Jingling Sun, Ting Su, Jiayi Jiang, Jue Wang, Geguang Pu, and Zhendong Su. 2023. Property-Based Fuzzing for Finding Data Manipulation Errors in Android Apps. InProceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2023, San Francisco, CA, USA, December 3-9, 2023, Satish Chandra, Kelly Blincoe, and Paolo Tonella",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Li-yao Xia, Yannick Zakowski, Paul He, Chung-Kil Hur, Gregory Malecha, Benjamin C. Pierce, and Steve Zdancewic. 2020. Interaction trees: representing recursive and impure programs in Coq.Proc. ACM Program. Lang.4, POPL (2020), 51:1–51:32. https://doi.org/10.1145/3371119",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. 2011. Finding and understanding bugs in C compilers. In Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation. ACM, Programmable Property-Based Testing 283–294. https://doi.org/10.1145/1993498.1993532",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Brent A. Yorgey, Stephanie Weirich, Julien Cretin, Simon Peyton Jones, Dimitrios Vytiniotis, and José Pedro Magalhães. 2012. Giving Haskell a promotion. InProceedings of the 8th ACMSIGPLAN Workshop on Types in Language Design and Implementation(Philadelphia, Pennsylvania, USA)(TLDI ’12). Association for Computing Machinery, New York, NY, USA, 53–66. https://doi.org/10.1145/2103786.2103795",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "A. Zeller and R. Hildebrandt. 2002. Simplifying and isolating failure-inducing input.IEEE Transactions on Software Engineering28, 2 (2002), 183–200. http://pag-www.gtisc.gatech.edu/courses/common/zeller-tse02.pdf Alperen Keles, Justine Frank, Ceren Mert, Harrison Goldstein, and Leonidas Lampropoulos AAPPENDIX A.1 Custom-Feedback Guided (Targeted) Property Runner In recent years, PBT tools and muta",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 1,
+      "text": "Jinsheng Ba and Manuel Rigger. 2023. Testing Database Engines via Query Plan Guidance. InThe 45th International Conference on Software Engineering (ICSE’23).",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 45,
+      "text": "SQLancer contributors. [n. d.].SQLancer: Automated testing to find logic and performance bugs in database systems. https://github.com/sqlancer/sqlancer",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The literature is rife with domain-specific metrics that can be used to produce interesting inputs: in SlowFuzz [ 40] and PerfFuzz [ 25], the goal of the testing campaign is to discover algorithmic complexity vulnerabilities that could lead to Denial of Service attacks in systems, which is achieved by maximizing instruction counts and execution path lengths; in SQLancer [ 45], the goal is to find bugs in SQL engines, achieved by maximizing the diversity of the generated query plans [ 1].",
+      "context_before": "tein, and Leonidas Lampropoulos instead of experimenting with new options in the new higher-level setting; in part because each option in such an experimentation would effectively require a new framework. Situations like this are precisely what a programmable and flexible runner framework addresses; in § 6.3.1, we carry out an extensive case study to explore what effect different search strategies and seed pool experimentations have in testing. 2.4 Custom-Feedback Guided (Targeted) Property Runner Coverage information is not the only form of feedback that can be used to guide input generation.",
+      "context_after": "It is also possible to consolidate such different feedback forms under a single compositional design, as demonstrated by FuzzFactory [ 37] and Target [ 29], which allows users to define and combine different feedback functions for guiding the generation of inputs. Fig. 6 depicts an abstract representation of this last targeted property runner, where the feedback is a separate function that is computed over the input instead of a value obtained via instrumentation. Targeting Loop Shrinking Loop G MSeed Pool FCcheck e feedback e gen/mutate fS Ccheck e’ shrink e’ | eshrink ePprint e/e’ Fig. 6. A",
+      "section": "2.2 Integrated Shrinking Property Runner",
+      "page": 6,
+      "char_offset": 16538,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "This paper proposes a deeper language for expressing properties in property-based testing. Properties in modern frameworks are written in shallowly embedded DSLs and tightly coupled to how they are run, so users are limited to whatever configuration the framework authors anticipated. The alternative, a mixed embedding called deferred binding abstract syntax, reifies a property as a data structure and separates it from the runner. The authors implement it in Rocq and Racket and prototype a variety of property runners.",
+    "narrative": "One citation, in a discussion of domain-specific metrics used to produce interesting inputs, alongside SlowFuzz and PerfFuzz.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_1177_0926227x251370258.json b/_data/papers/paper_doi_10_1177_0926227x251370258.json
new file mode 100644
index 0000000..465b8db
--- /dev/null
+++ b/_data/papers/paper_doi_10_1177_0926227x251370258.json
@@ -0,0 +1,1314 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:09:29Z",
+  "paper": {
+    "id": "paper:doi:10.1177/0926227x251370258",
+    "title": "SQLaser: Detecting database management system (DBMS) logic bugs with clause-guided fuzzing",
+    "authors": [
+      "Jin Wei",
+      "Ping Chen",
+      "Kangjie Lu",
+      "Jun Dai",
+      "Xiaoyan Sun"
+    ],
+    "year": 2025,
+    "venue": "Journal of computing and security",
+    "doi": "10.1177/0926227x251370258",
+    "arxiv_id": "2407.04294",
+    "s2_paper_id": "71f038fd6f135f0bc91c952de438cd717d675110",
+    "url": "https://doi.org/10.1177/0926227x251370258",
+    "also_indexed_as": [
+      "paper:arxiv:2407.04294"
+    ]
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2407.04294",
+      "retrieved_at": "2026-09-10T15:09:29Z",
+      "chars": 84025,
+      "content_sha256": "sha256:ec2e1d52bd7adfd51eec896ea0182261936dec917b41b9cf6734e1b4633a31b7"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 25,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2717
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 12926
+      },
+      {
+        "number": "2.1",
+        "title": "Logic Bugs and Threat Model",
+        "start": 12940
+      },
+      {
+        "number": "2.2",
+        "title": "Logic Bug Testing Oracles and Differential Testing",
+        "start": 15417
+      },
+      {
+        "number": "2.3",
+        "title": "Coverage-guided Fuzzing and Directed Fuzzing",
+        "start": 18375
+      },
+      {
+        "number": "3",
+        "title": "Analysis of Existing Logic Bug Patterns",
+        "start": 20905
+      },
+      {
+        "number": "4",
+        "title": "SQLaser",
+        "start": 26096
+      },
+      {
+        "number": "4.1",
+        "title": "Logic-Bug Pattern Modeling",
+        "start": 31622
+      },
+      {
+        "number": "4.2",
+        "title": "Call Chain Distance Calculation",
+        "start": 38671
+      },
+      {
+        "number": "4.3",
+        "title": "Mutation based on Seed Energy and Testing Oracles",
+        "start": 43178
+      },
+      {
+        "number": "4.3.1",
+        "title": "Mutation based on Seed Energy",
+        "start": 43427
+      },
+      {
+        "number": "4.3.2",
+        "title": "Testing Oracles",
+        "start": 44095
+      },
+      {
+        "number": "5",
+        "title": "Implementation",
+        "start": 45399
+      },
+      {
+        "number": "6",
+        "title": "Evaluation",
+        "start": 46922
+      },
+      {
+        "number": "20.04.5",
+        "title": "LTS operating system.",
+        "start": 47678
+      },
+      {
+        "number": "6.1",
+        "title": "Finding Bugs in DBMSs",
+        "start": 49159
+      },
+      {
+        "number": "6.2",
+        "title": "Comparison with Existing Tools",
+        "start": 56614
+      },
+      {
+        "number": "6.2.1",
+        "title": "Performance",
+        "start": 56964
+      },
+      {
+        "number": "6.2.2",
+        "title": "Code Coverage",
+        "start": 59193
+      },
+      {
+        "number": "6.2.3",
+        "title": "Seed Distance Distribution",
+        "start": 62213
+      },
+      {
+        "number": "6.3",
+        "title": "Trimmed Call Chains vs. Complete Call Chains",
+        "start": 63631
+      },
+      {
+        "number": "7",
+        "title": "Discussion",
+        "start": 65923
+      },
+      {
+        "number": "7.1",
+        "title": "Completeness of Testing Oracles",
+        "start": 66105
+      },
+      {
+        "number": "7.2",
+        "title": "Completeness of SQL Clause Collections",
+        "start": 67166
+      },
+      {
+        "number": "8",
+        "title": "Related Work",
+        "start": 67948
+      },
+      {
+        "number": "9",
+        "title": "Conclusion",
+        "start": 70130
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "K.S. Abdul and S. Khurshid, Automated SQL query generation for systematic testing of database engines, in: Proceedings of the 25th IEEE/ACM International Conference on Automated Software Engineering, 2010, pp. 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "C. Aschermann, S. Schumilo, T. Blazytko, R. Gawlik and T. Holz, REDQUEEN: Fuzzing with Input-to-State Correspondence., in: Proceedings of the 26th Annual Network and Distributed System Security Symposium (NDSS), V ol. 19, 2019, pp. 1–15.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "T. Blazytko, M. Bishop, C. Aschermann, J. Cappos, M. Schlögel, N. Korshun, A. Abbasi, M. Schweighauser, S. Schinzel, S. Schumilo et al., {GRIMOIRE }: Synthesizing structure while fuzzing, in: Proceedings of the 28th USENIX Security Symposium (USENIX Security 19), 2019, pp. 1985–2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "M. Böhme, V .-T. Pham, M.-D. Nguyen and A. Roychoudhury, Directed greybox fuzzing, in: Proceedings of the 2017 ACMSIGSAC conference on computer and communications security, 2017, pp. 2329–2344.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "H. Chen, Y. Xue, Y. Li, B. Chen, X. Xie, X. Wu and Y. Liu, Hawkeye: Towards a desired directed grey-box fuzzer, in: Proceedings of the 2018 ACMSIGSAC conference on computer and communications security, 2018, pp. 2095–2108. Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing 23",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "P. Chen and H. Chen, Angora: Efficient fuzzing by principled search, in: Proceedings of the Security and Privacy (SP), IEEE, 2018, pp. 711–725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Y. Chen, R. Zhong, H. Hu, H. Zhang, Y. Yang, D. Wu and W. Lee, One engine to fuzz’em all: Generic language processor testing with semantic validation, in: Proceedings of the pp. 642–658.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "J. Choi, K. Kim, D. Lee and S.K. Cha, NTFuzz: Enabling type-aware kernel fuzzing on windows with static binary analysis, in: Proceedings of the. 677–693.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Z.Y. Ding and C.L. Goues, An empirical study of oss-fuzz bugs, in: Proceedings of the 2021 IEEE/ACM 18th International Conference on Mining Software Repositories (MSR), IEEE, 2021, pp. 131–142.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "S.T. Dinh, H. Cho, K. Martin, A. Oest, K. Zeng, A. Kapravelos, G.-J. Ahn, T. Bao, R. Wang, A. Doupé et al., Favocado: Fuzzing the Binding Code of JavaScript Engines Using Semantically Correct Test Cases., in: Proceedings of the 28th Annual Network and Distributed System Security Symposium (NDSS), 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "X. Du, B. Chen, Y. Li, J. Guo, Y. Zhou, Y. Liu and Y. Jiang, Leopard: Identifying vulnerable code for vulnerability assessment through program metrics, in: Proceedings of the Engineering (ICSE), IEEE, 2019, pp. 60–71.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Z. Du, Y. Li, Y. Liu and B. Mao, WindRanger: a directed greybox fuzzer driven by deviation basic blocks, in: Proceedings of the 44th International Conference on Software Engineering, 2022, pp. 2440–2451.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "J.K. Fichte, M. Hecher, P. Thier and S. Woltran, Exploiting database management systems and treewidth for counting, Theory and Practice of Logic Programming 22(1) (2022), 128–157.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "P. Foundation, TiDB, 2023. https://www.pingcap.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "S. Gan, C. Zhang, P. Chen, B. Zhao, X. Qin, D. Wu and Z. Chen, GREYONE: Data flow sensitive fuzzing, in: Proceedings of the 29th USENIX security symposium (USENIX Security 20), 2020, pp. 2577–2594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "B. Ghit, N. Poggi, J. Rosen, R. Xin and P. Boncz, SparkFuzz: Searching correctness regressions in modern query engines, in:Proceedings of the workshop on Testing Database Systems, 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Google, ClusterFuzz, 2023. https://google.github.io/clusterfuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Google, Honggfuzz, 2023. https://google.github.io/honggfuzz/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "P.D. Group, PostgreSQL, 2023. https://www.postgresql.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "H. Han, D. Oh and S.K. Cha, CodeAlchemist: Semantics-Aware Code Generation to Find Vulnerabilities in JavaScript Engines., in: Proceedings of the 26th Network and Distributed System Security Symposium (NDSS), 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Intel, PIN, 2023. https://www.intel.cn/content/www/cn/zh/developer/articles/tool/pin-a-dynamic-binary-instrumentation-tool. html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "B. Jiang, Y. Liu and W.K. Chan, Contractfuzzer: Fuzzing smart contracts for vulnerability detection, in: Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering, 2018, pp. 259–269.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim and W. Kang, Apollo: Automatic detection and diagnosis of performance regressions in database systems, Proceedings of the 46th International Conference on Very Large Data Bases (VLDB) 13(1) (2019), 57–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "K. Kim, D.R. Jeong, C.H. Kim, Y. Jang, I. Shin and B. Lee, HFL: Hybrid Fuzzing on the Linux Kernel., in: Proceedings of the 27th Annual Network and Distributed System Security Symposium (NDSS), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "G. Lee, W. Shim and B. Lee, Constraint-guided directed greybox fuzzing, in: Proceedings of the 30th USENIX Security Symposium (USENIX Security 21), 2021, pp. 3559–3576.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "C. Lemieux and K. Sen, Fairfuzz: A targeted mutation strategy for increasing greybox fuzz testing coverage, in: Proceedings of the 33rd ACM/IEEE international conference on automated software engineering, 2018, pp. 475–485.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Y. Li, B. Chen, M. Chandramohan, S.-W. Lin, Y. Liu and A. Tiu, Steelix: program-state based binary fuzzing, in: Proceedings of the 2017 11th joint meeting on foundations of software engineering, 2017, pp. 627–637.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Y. Liang, S. Liu and H. Hu, Detecting Logical Bugs of {DBMS }with Coverage-based Guidance, in: Proceedings of the 31st USENIX Security Symposium (USENIX Security 22), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "X. Liu, Q. Zhou, J. Arulraj and A. Orso, Automated performance bug detection in database systems, arXiv preprint arXiv:2105.10016 (2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "LLVM, LibFuzzer: A Library For Coverage-guided Fuzz Testing, 2023. http://llvm.org/docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "E. Lo, C. Binnig, D. Kossmann, M. Tamer Özsu and W.-K. Hon, A framework for testing DBMS features, The VLDB Journal 19(2010), 203–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "W.M. McKeeman, Differential testing for software, Digital Technical Journal 10(1) (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "MySQL, MySQL Customers, 2023. https://www.mysql.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "T.D. Nguyen, L.H. Pham, J. Sun, Y. Lin and Q.T. Minh, sfuzz: An efficient adaptive fuzzer for solidity smart contracts, in:Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering, 2020, pp. 778–788.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "S. Österlund, K. Razavi, H. Bos and C. Giuffrida, {ParmeSan }: Sanitizer-guided greybox fuzzing, in: Proceedings of the 29th USENIX Security Symposium (USENIX Security 20), 2020, pp. 2289–2306. 24 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "S. Pailoor, A. Aday and S. Jana, {MoonShine }: Optimizing {OS}fuzzer seed selection with trace distillation, in: Proceedings of the 27th USENIX Security Symposium (USENIX Security 18), 2018, pp. 729–743.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "S. Park, W. Xu, I. Yun, D. Jang and T. Kim, Fuzzing javascript engines with aspect-preserving mutation, in: Proceedings of the. 1629–1642.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "M. Rigger and Z. Su, Detecting optimization bugs in database engines via non-optimizing reference engine construction, in:Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "M. Rigger and Z. Su, Finding bugs in database systems via query partitioning, in: Proceedings of the ACM on Programming Languages, V ol. 4, ACM New York, NY, USA, 2020, pp. 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, Testing database engines via pivoted query synthesis, in: Proceedings of 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "K. Serebryany, Sanitize, fuzz, and harden your C++ code, San Francisco, CA (2016).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "D.R. Slutz, Massive stochastic testing of SQL, in: VLDB, V ol. 98, Citeseer, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "SQLite, SQLite, 2023. https://sqlite.org/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "SQLite, SQLite bug fix, 2023. https://www.sqlite.org/src/info/5351e920f489562f.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "N. Stephens, J. Grosen, C. Salls, A. Dutcher, R. Wang, J. Corbetta, Y. Shoshitaishvili, C. Kruegel and G. Vigna, Driller: Augmenting fuzzing through selective symbolic execution., in: NDSS, V ol. 16, 2016, pp. 1–16.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang and Y. Jiang, Industry practice of coverage-guided enterpriselevel DBMS fuzzing, in: Proceedings of the Software Engineering in Practice (ICSE-SEIP), IEEE, 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "wikipedia, ASCII, 2023. https://zh.wikipedia.org/wiki/ASCII.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "wikipedia, coredump, 2023. https://en.wikipedia.org/wiki/Core_dump.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "V. Wüstholz and M. Christakis, Harvey: A greybox fuzzer for smart contracts, in: Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1398–1409.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "M. Xu, S. Kashyap, H. Zhao and T. Kim, Krace: Data race fuzzing for kernel file systems, in: Proceedings of the. 1643–1660.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "P. Xu, Y. Wang, H. Hu and P. Su, COOPER: Testing the Binding Code of Scripting Languages with Cooperative Mutation., in:Proceedings of the 28th Annual Network and Distributed System Security Symposium (NDSS), 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "W. Xu, S. Park and T. Kim, Freedom: Engineering a state-of-the-art dom fuzzer, in: Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, 2020, pp. 971–986.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "W. Xu, H. Moon, S. Kashyap, P.-N. Tseng and T. Kim, Fuzzing file systems via two-dimensional input space exploration, in:Proceedings of the. 818–834.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "I. Yun, S. Lee, M. Xu, Y. Jang and T. Kim, QSYM: A practical concolic execution engine tailored for hybrid fuzzing, in: Proceedings of the 27th USENIX Security Symposium (USENIX Security 18), 2018, pp. 745–761.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "M. Zalewski, American Fuzzy Lop, 2023. https://github.com/google/AFL.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee and D. Wu, Squirrel: Testing database management systems with language validity and coverage feedback, in: Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "P. Zong, T. Lv, D. Wang, Z. Deng, R. Liang and K. Chen, {FuzzGuard }: Filtering out unreachable inputs in directed grey-box fuzzing through deep learning, in: Proceedings of the 29th USENIX security symposium (USENIX security 20), 2020, pp. 2255–2269.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Bug CVE-2012-2081. https://nvd.nist.gov/vuln/detail/CVE-2012-2081.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Bug CVE-2012-2081. https://nvd.nist.gov/vuln/detail/CVE-2014-4987.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Bugs Found in Database Management Systems. https://www.manuelrigger.at/dbms-bugs/.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 61,
+      "text": "LLVM pass. https://llvm.org/docs/WritingAnLLVMPass.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "MySQL bug 95908. https://bugs.mysql.com/bug.php?id=95908.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "MySQL bug 95926. https://bugs.mysql.com/bug.php?id=95926.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "MySQL bug 95927. https://bugs.mysql.com/bug.php?id=95927.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "MySQL bug 95937. https://bugs.mysql.com/bug.php?id=95937.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "MySQL bug 95954. https://bugs.mysql.com/bug.php?id=95954.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "MySQL bug 95975. https://bugs.mysql.com/bug.php?id=95975.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "MySQL bug 95983. https://bugs.mysql.com/bug.php?id=95983.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "MySQL bug 96012. https://bugs.mysql.com/bug.php?id=96012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "MySQL bug 99122. https://bugs.mysql.com/bug.php?id=99122.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "SQLancer. https://github.com/sqlancer/sqlancer.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 72,
+      "text": "SQLite bug 16252d7. https://www.sqlite.org/src/info/16252d7.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "SQLite bug 1685610e. https://www.sqlite.org/src/info/1685610e. Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing 25",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "SQLite bug 1b1dd4d4. https://www.sqlite.org/src/info/1b1dd4d4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "SQLite bug 2363a14. https://www.sqlite.org/src/info/2363a14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "SQLite bug 5351e920. https://www.sqlite.org/src/info/5351e920.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "SQLite bug 54110870. https://www.sqlite.org/src/info/54110870.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "SQLite bug 5c6146b5. https://www.sqlite.org/src/info/5c6146b5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "SQLite bug 659c551d. https://www.sqlite.org/src/info/659c551d.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "SQLite bug 6ac0f822. https://www.sqlite.org/src/info/6ac0f822.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "SQLite bug 86fa0087. https://www.sqlite.org/src/info/86fa0087.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "SQLite bug 9c8c1092. https://www.sqlite.org/src/info/9c8c1092.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "SQLite bug c0c90961. https://www.sqlite.org/src/info/c0c90961.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "SQLite bug db9acef1. https://www.sqlite.org/src/info/db9acef1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "SQLite bug eb40248. https://www.sqlite.org/src/info/eb40248.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "SQLite bug ebe4845c. https://www.sqlite.org/src/info/ebe4845c.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "SQLite bug f65c929. https://www.sqlite.org/src/info/f65c929.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "SQLite bug f898d04c. https://www.sqlite.org/src/info/f898d04c.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "SQLite bug f9c6426. https://www.sqlite.org/src/info/f9c6426.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "SQLite bug faaaae49. https://www.sqlite.org/src/info/faaaae49.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "TiDB bug 15725. https://github.com/pingcap/tidb/issues/15725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "TiDB bug 15733. https://github.com/pingcap/tidb/issues/15733.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "TiDB bug 15789. https://github.com/pingcap/tidb/issues/15789.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "TiDB bug 15846. https://github.com/pingcap/tidb/issues/15846.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "TiDB bug 15986. https://github.com/pingcap/tidb/issues/15986.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "TiDB bug 15994. https://github.com/pingcap/tidb/issues/15994.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "TiDB bug 17814. https://github.com/pingcap/tidb/issues/17814.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 38,
+      "text": "M. Rigger and Z. Su, Detecting optimization bugs in database engines via non-optimizing reference engine construction, in:Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 39,
+      "text": "M. Rigger and Z. Su, Finding bugs in database systems via query partitioning, in: Proceedings of the ACM on Programming Languages, V ol. 4, ACM New York, NY, USA, 2020, pp. 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, Testing database engines via pivoted query synthesis, in: Proceedings of 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 60,
+      "text": "Bugs Found in Database Management Systems. https://www.manuelrigger.at/dbms-bugs/.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 71,
+      "text": "SQLancer. https://github.com/sqlancer/sqlancer.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[28, 38–40]",
+      "technique": "norec",
+      "sentence": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing code and are particularly common, can cause a variety of serious issues, such as incorrect query results, exposure of sensitive data, unauthorized access, and data corruption [28, 38–40].",
+      "context_before": "ms, Logic Bug detection, Directed Fuzzing 1. Introduction Database Management Systems (DBMSs) [13] are widely used as centralized storage and management systems for data. For instance, SQLite [43] is used on around 3.5 billion smartphones worldwide. Given their critical role of DBMSs in modern data-driven applications, logic bugs [28] within DBMSs can have severe consequences. These bugs, which are due to implementation errors in the DBMSs’ *Corresponding author. E-mail: pchen@fudan.edu.cn. 0926-227X/$35.00 © 0 – IOS Press. All rights reserved.arXiv:2407.04294v1 [cs.CR] 5 Jul 2024 2 Wei et al.",
+      "context_after": "Due to the non-crashing nature of most logic bugs, detecting them can be challenging, and researchers have invested significant efforts in this area. Rigger et al. propose oracles to detect incorrectness of SQL query results. For example, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] and the Ternary Logic Partitioning (TLP) oracle [39] are two oracles employing the concept of differential testing [32]. They transform the original SQL query into an equivalent form, and if discrepancies arise between the results of the original query and the transformed query, potential l",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3275,
+      "cited_reference": {
+        "number": 38,
+        "text": "M. Rigger and Z. Su, Detecting optimization bugs in database engines via non-optimizing reference engine construction, in:Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "Non-Optimizing Reference Engine Construction",
+      "technique": "norec",
+      "sentence": "For example, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] and the Ternary Logic Partitioning (TLP) oracle [39] are two oracles employing the concept of differential testing [32].",
+      "context_before": "hen@fudan.edu.cn. 0926-227X/$35.00 © 0 – IOS Press. All rights reserved.arXiv:2407.04294v1 [cs.CR] 5 Jul 2024 2 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing code and are particularly common, can cause a variety of serious issues, such as incorrect query results, exposure of sensitive data, unauthorized access, and data corruption [28, 38–40]. Due to the non-crashing nature of most logic bugs, detecting them can be challenging, and researchers have invested significant efforts in this area. Rigger et al. propose oracles to detect incorrectness of SQL query results.",
+      "context_after": "They transform the original SQL query into an equivalent form, and if discrepancies arise between the results of the original query and the transformed query, potential logic bugs may be indicated. Pivoted Query Synthesis (PQS) oracle [40] does not employ the concept of differential testing; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row. Failure to retrieve the pivot row indicates a potential bug in the system. SQLancer [60, 71] that implements these oracles discovers logic bugs in various DBMSs. However, as SQLancer generates",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3753,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": "Pivoted Query Synthesis (PQS) oracle [40] does not employ the concept of differential testing; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row.",
+      "context_before": "ture of most logic bugs, detecting them can be challenging, and researchers have invested significant efforts in this area. Rigger et al. propose oracles to detect incorrectness of SQL query results. For example, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] and the Ternary Logic Partitioning (TLP) oracle [39] are two oracles employing the concept of differential testing [32]. They transform the original SQL query into an equivalent form, and if discrepancies arise between the results of the original query and the transformed query, potential logic bugs may be indicated.",
+      "context_after": "Failure to retrieve the pivot row indicates a potential bug in the system. SQLancer [60, 71] that implements these oracles discovers logic bugs in various DBMSs. However, as SQLancer generates SQL queries based on specific rules, it may constrain the exploration of broader code paths. Therefore, Liang et al. introduce SQLRight [28], aiming to generate high-quality query statements with validity-oriented mutations to uncover more logic bugs. Specifically, SQLRight employs coverage-guided fuzzing [6, 15, 18, 27, 30, 41, 45, 54, 55] to achieve higher code coverage in code paths. 1CREATE TABLE t0",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4154,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [60, 71] that implements these oracles discovers logic bugs in various DBMSs.",
+      "context_before": "nd the Ternary Logic Partitioning (TLP) oracle [39] are two oracles employing the concept of differential testing [32]. They transform the original SQL query into an equivalent form, and if discrepancies arise between the results of the original query and the transformed query, potential logic bugs may be indicated. Pivoted Query Synthesis (PQS) oracle [40] does not employ the concept of differential testing; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row. Failure to retrieve the pivot row indicates a potential bug in the system.",
+      "context_after": "However, as SQLancer generates SQL queries based on specific rules, it may constrain the exploration of broader code paths. Therefore, Liang et al. introduce SQLRight [28], aiming to generate high-quality query statements with validity-oriented mutations to uncover more logic bugs. Specifically, SQLRight employs coverage-guided fuzzing [6, 15, 18, 27, 30, 41, 45, 54, 55] to achieve higher code coverage in code paths. 1CREATE TABLE t0(c0 COLLATE NOCASE, c1); 2CREATE INDEX i0 ON t0(c0) WHERE c0 >= c1; 3INSERT INTO t0 VALUES(’a’, ’B’); 4SELECT *FROM t0 WHERE t0.c1 <= t0.c0; 5-- output: {}, expec",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4437,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "However, as SQLancer generates SQL queries based on specific rules, it may constrain the exploration of broader code paths.",
+      "context_before": "pt of differential testing [32]. They transform the original SQL query into an equivalent form, and if discrepancies arise between the results of the original query and the transformed query, potential logic bugs may be indicated. Pivoted Query Synthesis (PQS) oracle [40] does not employ the concept of differential testing; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row. Failure to retrieve the pivot row indicates a potential bug in the system. SQLancer [60, 71] that implements these oracles discovers logic bugs in various DBMSs.",
+      "context_after": "Therefore, Liang et al. introduce SQLRight [28], aiming to generate high-quality query statements with validity-oriented mutations to uncover more logic bugs. Specifically, SQLRight employs coverage-guided fuzzing [6, 15, 18, 27, 30, 41, 45, 54, 55] to achieve higher code coverage in code paths. 1CREATE TABLE t0(c0 COLLATE NOCASE, c1); 2CREATE INDEX i0 ON t0(c0) WHERE c0 >= c1; 3INSERT INTO t0 VALUES(’a’, ’B’); 4SELECT *FROM t0 WHERE t0.c1 <= t0.c0; 5-- output: {}, expected: {a|B} Code 1: An example logic bug in SQLite. Despite substantial efforts dedicated to detecting logic bugs, we find so",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4524,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "technique",
+      "surface": "Non-Optimizing Reference Engine Construction",
+      "technique": "norec",
+      "sentence": "For instance, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] translates an optimized query containing a WHERE clause into an unoptimized version with identical semantics.",
+      "context_before": "are found among the results produced by these systems, it indicates the presence of at least one system containing a bug. However, constructing an oracle that covers all DBMS logic bugs is challenging. Additionally, due to various languages and extensions, popular DBMSs share only a small subset of functionalities, making cross-DBMS validation incapable of detecting DBMS-specific bugs [42]. Researchers, including Manuel Rigger et al., have made efforts in this direction. They build functionally equivalent SQL queries for a single DBMS and check if these queries could produce the same results.",
+      "context_after": "By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug. Ternary Logic Partitioning (TLP) oracle [39] composes several sub-queries to collectively achieve the semantics of the original query. If the composed query does not yield the same result as the original query, it suggests the possibility of a logic bug. SQLRight [28] also supplements their approach with some oracles. The INDEX oracle adds various CREATE INDEX clauses to the query set to test the impact of INDEX usage on query execution. The ROWID oracle inserts WITHOUT ROWID clause",
+      "section": "2.2 Logic Bug Testing Oracles and Differential Testing",
+      "page": 5,
+      "char_offset": 16820,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "Ternary Logic Partitioning (TLP) oracle [39] composes several sub-queries to collectively achieve the semantics of the original query.",
+      "context_before": "ctionalities, making cross-DBMS validation incapable of detecting DBMS-specific bugs [42]. Researchers, including Manuel Rigger et al., have made efforts in this direction. They build functionally equivalent SQL queries for a single DBMS and check if these queries could produce the same results. For instance, the Non-Optimizing Reference Engine Construction (NoREC) oracle [38] translates an optimized query containing a WHERE clause into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug.",
+      "context_after": "If the composed query does not yield the same result as the original query, it suggests the possibility of a logic bug. SQLRight [28] also supplements their approach with some oracles. The INDEX oracle adds various CREATE INDEX clauses to the query set to test the impact of INDEX usage on query execution. The ROWID oracle inserts WITHOUT ROWID clause to observe its effect on query execution. If the results of the queries with the inserted INDEX or WITHOUT ROWID clause differ from those without these insertions, it indicates a potential logic bug. And the LIKELY oracle checks if the original q",
+      "section": "2.2 Logic Bug Testing Oracles and Differential Testing",
+      "page": 5,
+      "char_offset": 17124,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": "It is worth noting that the Pivoted Query Synthesis (PQS) oracles [40] proposed by Manuel Rigger et al.",
+      "context_before": "ic bug. SQLRight [28] also supplements their approach with some oracles. The INDEX oracle adds various CREATE INDEX clauses to the query set to test the impact of INDEX usage on query execution. The ROWID oracle inserts WITHOUT ROWID clause to observe its effect on query execution. If the results of the queries with the inserted INDEX or WITHOUT ROWID clause differ from those without these insertions, it indicates a potential logic bug. And the LIKELY oracle checks if the original queries include the LIKELY clause. If they don’t, it adds the LIKELY clause to see how it impacts query execution.",
+      "context_after": "does not use the differential testing method; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row. If the DBMS fails to fetch the pivot row, the likely cause is a bug in the DBMS. As of now, we have not deployed PQS oracles into SQLaser. 2.3. Coverage-guided Fuzzing and Directed Fuzzing Fuzzing [9] is a software testing technique that involves providing invalid, unexpected, or random data as inputs to a computer program. The goal is to discover vulnerabilities or bugs by observing how the program reacts to these inputs [18, 30, 41, 5",
+      "section": "2.2 Logic Bug Testing Oracles and Differential Testing",
+      "page": 5,
+      "char_offset": 17972,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "As of now, we have not deployed PQS oracles into SQLaser.",
+      "context_before": "OUT ROWID clause differ from those without these insertions, it indicates a potential logic bug. And the LIKELY oracle checks if the original queries include the LIKELY clause. If they don’t, it adds the LIKELY clause to see how it impacts query execution. It is worth noting that the Pivoted Query Synthesis (PQS) oracles [40] proposed by Manuel Rigger et al. does not use the differential testing method; instead, it automatically generates queries for which they ensure fetching a specific row, called the pivot row. If the DBMS fails to fetch the pivot row, the likely cause is a bug in the DBMS.",
+      "context_after": "2.3. Coverage-guided Fuzzing and Directed Fuzzing Fuzzing [9] is a software testing technique that involves providing invalid, unexpected, or random data as inputs to a computer program. The goal is to discover vulnerabilities or bugs by observing how the program reacts to these inputs [18, 30, 41, 55]. Coverage-guided fuzzing is a specific type of fuzz testing that utilizes code coverage feedback from the program’s execution to guide the generation of new test 6 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing inputs. The key idea is to measure the code coverage ach",
+      "section": "2.2 Logic Bug Testing Oracles and Differential Testing",
+      "page": 5,
+      "char_offset": 18316,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[28, 60]",
+      "technique": null,
+      "sentence": "In particular, we analyze a total of 144 existing logic bugs, including 102 in SQLite, 15 in MySQL, 1 in PostgreSQL, and 26 in TiDB [28, 60].",
+      "context_before": "erability or when testing specific features of the program. In this paper, we employ directed fuzzing to deliberately detect logic bugs, rather than aimlessly testing code paths. Specifically, we target logic bug patterns, which denote paths in the source code that are highly prone to triggering logic bugs. We allocate a substantial portion of the time budget to explore these targeted paths, with the objective of uncovering more bugs. 3. Analysis of Existing Logic Bug Patterns To develop an effective DBMS fuzzer for logic bugs, we conduct the first systematic study against existing logic bugs.",
+      "context_after": "This analysis covers most logic bugs, and we do not investigate logic bugs that result in system crashes because we focus on logic bugs that do not lead to observable system failures (such as crashes). Based on the semantics of SQL clause, we classify SQL clauses related to these logic bugs into five categories: table element/schema, data processing clause, conditional expressions, special keyword, and query optimization function, as shown in Table 1. Below introduces each category: •Table element/schema: This category involves elements or schemas in the created tables, such as column indexes",
+      "section": "3 Analysis of Existing Logic Bug Patterns",
+      "page": 6,
+      "char_offset": 21066,
+      "cited_reference": {
+        "number": 60,
+        "text": "Bugs Found in Database Management Systems. https://www.manuelrigger.at/dbms-bugs/.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "In the case of PostgreSQL, only one logic bug was found, specifically related to the PRIMARY KEY and GROUP BY clauses [71].",
+      "context_before": "WID, and NOCASE. We obtain the SQL clause combination for the remaining logic bugs using a similar approach. As for MySQL, the first SQL clause combination pertains to column values, primarily including errors induced when handling floating-point numbers, such as a query whose predicate involves floating-point numbers yielding an incorrect result (bug: 99122 [70]). Additionally, some logic bugs are caused by query optimization functions such as ANY, LIKE, etc (bug: 95927 [64])). Incorrect results can also be caused by conditional expressions such as IFFULL, IF(FALSE), and IN (bug: 95926 [63]).",
+      "context_after": "For TiDB, it can be observed that certain types of columns such as special characters (like double negation, etc.) or specific data types (like floats/boolean, etc.) are more prone to triggering logic bugs in TiDB (bug: 15725 [91]). Additionally, certain SQL clauses such as conditional expressions (bug: 15733 [92]), data processing functions (bug: 15986 [95]), and special keywords (bugs: 15789 [93], 15846 [94], 15994 [96], 17814 [97]) also contribute to triggering logic bugs in TiDB. 4. SQLaser In this section, we present our design and implementation for SQLaser, which enables fuzzing to con",
+      "section": "3 Analysis of Existing Logic Bug Patterns",
+      "page": 7,
+      "char_offset": 25482,
+      "cited_reference": {
+        "number": 71,
+        "text": "SQLancer. https://github.com/sqlancer/sqlancer.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "DBMS SQL-level Bug Pattern (SQL Clause Combination) Number Bug Example SQLite [43]INDEX, PRIMARY KEY, WITHOUT ROWID, NOCASE 1 1b1dd4d4 [74] partial INDEX, LIKELY, ISFAIL 11 5351e920 [76] WITHOUT ROWID, PRIMARY KEYDESC 2 f65c929 [87] column value, CAST, LIKELY, UNLIKELY, GLOB 26 f9c6426 [89] column value, MIN 3 faaaae49 [90] column value, CAST 8 c0c90961 [83] column value, ROUND 1 db9acef1 [84] CAST, EXISTS 3 16252d7 [72] INSERT ORFAIL 1 659c551d [79] RTRIM, PRIMARY KEY, WITHOUT ROWID 1 86fa0087 [81] DISTINCT, ORDER BY 3 6ac0f822 [80] PRAGMA 3 ebe4845c [86] ALTER TABLE, column value 4 1685610e [73] rtree, (COUNT,CAST) 7 f898d04c [88] VIEW, INDEX 11 9c8c1092 [82] IN, ORDER BY 7 eb40248 [85] Others 10 54110870 [77] MySQL [33]column values 4 99122 [70] ANY 1 [28] LIKEESCAPE,XOR 1 95927 [64] BIGINT UNSIGNED, IFNULL 2 95954 [66] “<=>” comparison 1 95908 [62] IF(FALSE) 1 95926 [63] IN operator 1 95975 [67] &, <, and AND 1 95983 [68] GREATEST 1 96012 [62] others 2 95937 [65] PostgreSQL [19] PRIMARY KEY, GROUP BY 1 [71] TiDB [14]columns 11 15725 [91] CAST, IsTrue/IsFalse 2 15733 [92] CHAR() 1 15986 [95] collation 2 15789 [93] JOIN 6 15846 [94] USE_INDEX_MERGE 1 15994 [96] Others 3 17814 [97] bug pattern information as target sites for directed fuzzing, and creating a tool with the capability to efficiently reach predefined target sites.",
+      "context_before": "(bugs: 15789 [93], 15846 [94], 15994 [96], 17814 [97]) also contribute to triggering logic bugs in TiDB. 4. SQLaser In this section, we present our design and implementation for SQLaser, which enables fuzzing to concentrate on error-prone function chains. These function chains are the implementation of SQL-level bug patterns, at the source code level. Specifically, our design achieves two key objectives: incorporating 8 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing Table 2 Analysis of the SQL-level bug patterns of logic bugs for SQLite, MySQL, PostgreSQL and TiDB.",
+      "context_after": "Overview. As shown in Fig. 1, SQLaser consists of two components: 1) logic-bug pattern modeling, and 2) clause-guided fuzzing. Moreover, following established practice of existing directed fuzzing tools [4, 12], SQLaser instruments edge coverage information and basic block distance information into the binary code. Such information is obtained through an approach akin to WindRanger [12], with further intricate details omitted in this paper. In logic-bug pattern modeling phase, SQLaser aims to instrument the call chains of bug pattern information into binary code, and these call chains manifes",
+      "section": "4 SQLaser",
+      "page": 8,
+      "char_offset": 26590,
+      "cited_reference": {
+        "number": 71,
+        "text": "SQLancer. https://github.com/sqlancer/sqlancer.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "The five oracles include: 1) the NoREC oracle, which rewrites an optimized query containing a WHERE clause to a SELECT query statement without the WHERE optimization; 2) the TLP oracle, which partitions a given query into multiple equivalent queries, and their results can be combined to obtain the same results as the original query; 3) the INDEX oracle, which 14 Wei et al.",
+      "context_before": ". Specifically, SQLaser utilizes Equation 8 to calculate the distance between the seed and the target. A shorter distance results in a higher energy allocation for the seed, as expressed in Equation 12. energy =dCallChain−1(12) 4.3.2. Testing Oracles We utilize five testing oracles to analyze the outcomes of seed execution, similar to SQLRight [28]. These oracles can transform the original queries into variant forms with different syntactic structures but equivalent functionality. By comparing the execution results of the original queries and their variants, we can detect potential logic bugs.",
+      "context_after": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing introduces various CREATE INDEX clauses into the given query set, testing the impact of INDEX usage on the query execution; and 4) the ROWID oracle, which examines whether the original queries contain the WITHOUT ROWID clause. If it is not present, the WITHOUT ROWID clause is inserted to observe its effect on the query execution. 5) the LIKELY oracle checks if the original queries include the LIKELY clause. If they don’t, it adds the LIKELY clause to see how it impacts query execution. 5. Implementation In this section, we introd",
+      "section": "4.3.2 Testing Oracles",
+      "page": 13,
+      "char_offset": 44467,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "For instance, some bugs are identified by the PQS oracle [40] proposed by Manuel Rigger et al.",
+      "context_before": "y of the bug patterns. However, there are still some bug patterns where we do not find bugs, possibly due to the limited duration of our fuzzing runs. It is expected that extending the runtime will lead to the discovery of more bugs, and this is outlined as part of our future work. Furthermore, we note that in the case of MySQL, we find bugs in 4 out of the 10 patterns we instrument, indicating that we do not cover the majority of the bug patterns. We attribute this to the incompleteness of testing oracles supported by SQLaser, which may not be comprehensive enough to detect all types of bugs.",
+      "context_after": ", which is not yet integrated into SQLaser (as discussed in Section 7.1). Additionally, 18 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing out of the 15 existing bugs we summarize for MySQL, 13 are detected by the PQS oracle, explaining the comparatively lower number of bug found by SQLaser for MySQL. 6.2. Comparison with Existing Tools For evaluating SQLaser, we also ran SQLRight [28] and WindRanger [12], which represent an instance of coverage-guided fuzzer and single-function-target directed fuzzer, respectively. In this section, we delve into a comparison of the",
+      "section": "6.1 Finding Bugs in DBMSs",
+      "page": 17,
+      "char_offset": 56189,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing out of the 15 existing bugs we summarize for MySQL, 13 are detected by the PQS oracle, explaining the comparatively lower number of bug found by SQLaser for MySQL.",
+      "context_before": "l lead to the discovery of more bugs, and this is outlined as part of our future work. Furthermore, we note that in the case of MySQL, we find bugs in 4 out of the 10 patterns we instrument, indicating that we do not cover the majority of the bug patterns. We attribute this to the incompleteness of testing oracles supported by SQLaser, which may not be comprehensive enough to detect all types of bugs. For instance, some bugs are identified by the PQS oracle [40] proposed by Manuel Rigger et al., which is not yet integrated into SQLaser (as discussed in Section 7.1). Additionally, 18 Wei et al.",
+      "context_after": "6.2. Comparison with Existing Tools For evaluating SQLaser, we also ran SQLRight [28] and WindRanger [12], which represent an instance of coverage-guided fuzzer and single-function-target directed fuzzer, respectively. In this section, we delve into a comparison of the performance, code coverage and seed distance distribution of the three fuzzers. 6.2.1. Performance 0123456789103035...60 TTE(day)024681012Bug Number SQLite SQLaser SQLRight WindRanger 012345678910...30 TTE(day)0123456Bug Number MySQL SQLaser SQLRight WindRanger 012345678910...30 TTE(day)01234Bug Number TiDB SQLaser SQLRight Win",
+      "section": "6.1 Finding Bugs in DBMSs",
+      "page": 18,
+      "char_offset": 56385,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For NoREC and TLP oracles, they are implemented Wei et al.",
+      "context_before": "L and TiDB. Due to the fact that SQLRight and SQLaser do not find bugs in PostgreSQL, we are unable to gather TTE statistics for PostgreSQL. Through our evaluation, we find that both SQLRight and WindRanger take longer to discover bugs or cover target call chains due to their lack of explicit targeting of the bug pattern information. In contrast, SQLaser achieves a significant decrease in bug discovery time, approximately 60%. 6.2.2. Code Coverage As depicted in Fig. 3, Fig. 4 and Fig. 5, we record the code coverage of SQLaser, SQLRight, and WindRanger when testing DBMSs using testing oracles.",
+      "context_after": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing 19 012345678910...60 Day051015Code Coverage (%) SQLite SQLaser SQLRight WindRanger 012345678910...30 Day0204060Code Coverage (%) MySQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) PostgreSQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) TiDB SQLaser SQLRight WindRanger Fig. 3. Code coverage of SQLaser, SQLRight and WindRanger for NoREC oracle. 012345678910...60 Day051015Code Coverage (%) SQLite SQLaser SQLRight WindRanger 012345678910...30 Day0204060Code Coverage (%) MySQL SQ",
+      "section": "6.2.2 Code Coverage",
+      "page": 18,
+      "char_offset": 59362,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Code coverage of SQLaser, SQLRight and WindRanger for NoREC oracle.",
+      "context_before": "epicted in Fig. 3, Fig. 4 and Fig. 5, we record the code coverage of SQLaser, SQLRight, and WindRanger when testing DBMSs using testing oracles. For NoREC and TLP oracles, they are implemented Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing 19 012345678910...60 Day051015Code Coverage (%) SQLite SQLaser SQLRight WindRanger 012345678910...30 Day0204060Code Coverage (%) MySQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) PostgreSQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) TiDB SQLaser SQLRight WindRanger Fig. 3.",
+      "context_after": "012345678910...60 Day051015Code Coverage (%) SQLite SQLaser SQLRight WindRanger 012345678910...30 Day0204060Code Coverage (%) MySQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) PostgreSQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) TiDB SQLaser SQLRight WindRanger Fig. 4. Code coverage of SQLaser, SQLRight and WindRanger to trigger bugs for TLP oracle. 012345678910...60 Day051015Code Coverage (%) ROWID oracle SQLaser SQLRight WindRanger 012345678910...60 Day051015Code Coverage (%) LIKELY oracle SQLaser SQLRight WindRanger 012345678910..",
+      "section": "6.2.2 Code Coverage",
+      "page": 19,
+      "char_offset": 59818,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Code coverage of SQLaser, SQLRight and WindRanger to trigger bugs for TLP oracle.",
+      "context_before": "ySQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) PostgreSQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) TiDB SQLaser SQLRight WindRanger Fig. 3. Code coverage of SQLaser, SQLRight and WindRanger for NoREC oracle. 012345678910...60 Day051015Code Coverage (%) SQLite SQLaser SQLRight WindRanger 012345678910...30 Day0204060Code Coverage (%) MySQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) PostgreSQL SQLaser SQLRight WindRanger 012345678910...30 Day051015Code Coverage (%) TiDB SQLaser SQLRight WindRanger Fig. 4.",
+      "context_after": "012345678910...60 Day051015Code Coverage (%) ROWID oracle SQLaser SQLRight WindRanger 012345678910...60 Day051015Code Coverage (%) LIKELY oracle SQLaser SQLRight WindRanger 012345678910...60 Day051015Code Coverage (%) INDEX oracle SQLaser SQLRight WindRanger Fig. 5. Code coverage of SQLaser, SQLRight and WindRanger to trigger bugs in SQLite for ROWID, LIKELY and INDEX oracle. in tested four DBMSs. For ROWID, LIKELY, and INDEX oracles, they are only implemented in SQLite for now, and not in other DBMSs yet. According to Fig. 3, Fig. 4 and Fig. 5, we can see that SQLaser has a lower code covera",
+      "section": "6.2.2 Code Coverage",
+      "page": 19,
+      "char_offset": 60216,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Currently, SQLaser only supports differential testingbased oracles and does not accommodate other types of oracles, such as PQS oracle [40], which detects logic bugs through non-differential testing.",
+      "context_before": "the discovery of more bugs, since the fuzzer explores a wider range of possibilities, even if some of them are false positives. 7. Discussion In this section, we discuss two issues: the completeness of the testing oracles employed in this paper, and the completeness of the SQL clause collections we consider. 7.1. Completeness of Testing Oracles In this paper, we employ five oracles to test logic bugs. Specifically, we validate the existence of logic bugs by converting SQL query statements into functionally equivalent yet different queries and comparing the output results of these two queries.",
+      "context_after": "However, in the process of statistically analyzing existing logic bugs, we include the bugs detected by PQS oracles and instrument their call chains. As SQLaser does not currently support PQS oracles, this may result in its inability to identify logic bugs detectable solely by PQS oracles. Additionally, to the best of our knowledge, existing research has not extensively investigated the completeness of these five oracles (NoREC, TLP, INDEX, ROWID and LIKELY) in addressing logic bugs in real-world scenarios. We will undertake research on this matter in future work. 7.2. Completeness of SQL Cla",
+      "section": "7.1 Completeness of Testing Oracles",
+      "page": 21,
+      "char_offset": 66394,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "However, in the process of statistically analyzing existing logic bugs, we include the bugs detected by PQS oracles and instrument their call chains.",
+      "context_before": "of the testing oracles employed in this paper, and the completeness of the SQL clause collections we consider. 7.1. Completeness of Testing Oracles In this paper, we employ five oracles to test logic bugs. Specifically, we validate the existence of logic bugs by converting SQL query statements into functionally equivalent yet different queries and comparing the output results of these two queries. Currently, SQLaser only supports differential testingbased oracles and does not accommodate other types of oracles, such as PQS oracle [40], which detects logic bugs through non-differential testing.",
+      "context_after": "As SQLaser does not currently support PQS oracles, this may result in its inability to identify logic bugs detectable solely by PQS oracles. Additionally, to the best of our knowledge, existing research has not extensively investigated the completeness of these five oracles (NoREC, TLP, INDEX, ROWID and LIKELY) in addressing logic bugs in real-world scenarios. We will undertake research on this matter in future work. 7.2. Completeness of SQL Clause Collections We systematically analyze logic bugs extracted from state-of-the-art studies, identifying common error-prone sequences of functions. W",
+      "section": "7.1 Completeness of Testing Oracles",
+      "page": 21,
+      "char_offset": 66594,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "As SQLaser does not currently support PQS oracles, this may result in its inability to identify logic bugs detectable solely by PQS oracles.",
+      "context_before": "this paper, we employ five oracles to test logic bugs. Specifically, we validate the existence of logic bugs by converting SQL query statements into functionally equivalent yet different queries and comparing the output results of these two queries. Currently, SQLaser only supports differential testingbased oracles and does not accommodate other types of oracles, such as PQS oracle [40], which detects logic bugs through non-differential testing. However, in the process of statistically analyzing existing logic bugs, we include the bugs detected by PQS oracles and instrument their call chains.",
+      "context_after": "Additionally, to the best of our knowledge, existing research has not extensively investigated the completeness of these five oracles (NoREC, TLP, INDEX, ROWID and LIKELY) in addressing logic bugs in real-world scenarios. We will undertake research on this matter in future work. 7.2. Completeness of SQL Clause Collections We systematically analyze logic bugs extracted from state-of-the-art studies, identifying common error-prone sequences of functions. We have made every effort to comprehensively gather existing logic bugs, and these resultant collections effectively represent prevalent logic",
+      "section": "7.1 Completeness of Testing Oracles",
+      "page": 21,
+      "char_offset": 66744,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Additionally, to the best of our knowledge, existing research has not extensively investigated the completeness of these five oracles (NoREC, TLP, INDEX, ROWID and LIKELY) in addressing logic bugs in real-world scenarios.",
+      "context_before": "nts into functionally equivalent yet different queries and comparing the output results of these two queries. Currently, SQLaser only supports differential testingbased oracles and does not accommodate other types of oracles, such as PQS oracle [40], which detects logic bugs through non-differential testing. However, in the process of statistically analyzing existing logic bugs, we include the bugs detected by PQS oracles and instrument their call chains. As SQLaser does not currently support PQS oracles, this may result in its inability to identify logic bugs detectable solely by PQS oracles.",
+      "context_after": "We will undertake research on this matter in future work. 7.2. Completeness of SQL Clause Collections We systematically analyze logic bugs extracted from state-of-the-art studies, identifying common error-prone sequences of functions. We have made every effort to comprehensively gather existing logic bugs, and these resultant collections effectively represent prevalent logic bugs in the field. To our best knowledge, there is currently no theory or standard proposed in existing research to classify SQL clauses. We are the first to strive to investigate the types of SQL clause of existing logic",
+      "section": "7.1 Completeness of Testing Oracles",
+      "page": 21,
+      "char_offset": 66885,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [71] is a rulebased tool that successfully identifies numerous logic bugs.",
+      "context_before": "ollections effectively represent prevalent logic bugs in the field. To our best knowledge, there is currently no theory or standard proposed in existing research to classify SQL clauses. We are the first to strive to investigate the types of SQL clause of existing logic bugs, and leverage the discovered insights for detecting bugs in DBMS. However, we acknowledge that ensuring this classification covers all SQL clause might be a challenge, and addressing this limitation is a direction for our future work. 8. Related Work In this section, we discuss techniques for detecting logic bugs in DBMSs.",
+      "context_after": "It employs three distinct methods for detecting logic bugs: PQS [40], NoREC [38], and TLP [39]. PQS focuses on generating queries that fetch a 22 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing randomly selected row called the pivot row. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS. NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presenc",
+      "section": "8 Related Work",
+      "page": 21,
+      "char_offset": 68037,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "It employs three distinct methods for detecting logic bugs: PQS [40], NoREC [38], and TLP [39].",
+      "context_before": "ledge, there is currently no theory or standard proposed in existing research to classify SQL clauses. We are the first to strive to investigate the types of SQL clause of existing logic bugs, and leverage the discovered insights for detecting bugs in DBMS. However, we acknowledge that ensuring this classification covers all SQL clause might be a challenge, and addressing this limitation is a direction for our future work. 8. Related Work In this section, we discuss techniques for detecting logic bugs in DBMSs. SQLancer [71] is a rulebased tool that successfully identifies numerous logic bugs.",
+      "context_after": "PQS focuses on generating queries that fetch a 22 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing randomly selected row called the pivot row. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS. NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug. Lastly, TLP partitions a query into several sub-queries, which are composed to",
+      "section": "8 Related Work",
+      "page": 21,
+      "char_offset": 68121,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS focuses on generating queries that fetch a 22 Wei et al.",
+      "context_before": "auses. We are the first to strive to investigate the types of SQL clause of existing logic bugs, and leverage the discovered insights for detecting bugs in DBMS. However, we acknowledge that ensuring this classification covers all SQL clause might be a challenge, and addressing this limitation is a direction for our future work. 8. Related Work In this section, we discuss techniques for detecting logic bugs in DBMSs. SQLancer [71] is a rulebased tool that successfully identifies numerous logic bugs. It employs three distinct methods for detecting logic bugs: PQS [40], NoREC [38], and TLP [39].",
+      "context_after": "/ SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing randomly selected row called the pivot row. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS. NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug. Lastly, TLP partitions a query into several sub-queries, which are composed to have the same semantics as the original query. If the compos",
+      "section": "8 Related Work",
+      "page": 21,
+      "char_offset": 68217,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics.",
+      "context_before": "direction for our future work. 8. Related Work In this section, we discuss techniques for detecting logic bugs in DBMSs. SQLancer [71] is a rulebased tool that successfully identifies numerous logic bugs. It employs three distinct methods for detecting logic bugs: PQS [40], NoREC [38], and TLP [39]. PQS focuses on generating queries that fetch a 22 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing randomly selected row called the pivot row. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS.",
+      "context_after": "By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug. Lastly, TLP partitions a query into several sub-queries, which are composed to have the same semantics as the original query. If the composed query does not yield the same result as the original query, it suggests the possibility of a logic bug. While SQLancer relies on a rule-based generator to detect logic bugs, limiting its exploration of input samples. Liang et al. [28] propose SQLRight, a methodology for logic bug exploration based on coverage-based fuzzing. SQLRight successfull",
+      "section": "8 Related Work",
+      "page": 22,
+      "char_offset": 68517,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Lastly, TLP partitions a query into several sub-queries, which are composed to have the same semantics as the original query.",
+      "context_before": "distinct methods for detecting logic bugs: PQS [40], NoREC [38], and TLP [39]. PQS focuses on generating queries that fetch a 22 Wei et al. / SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing randomly selected row called the pivot row. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS. NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug.",
+      "context_after": "If the composed query does not yield the same result as the original query, it suggests the possibility of a logic bug. While SQLancer relies on a rule-based generator to detect logic bugs, limiting its exploration of input samples. Liang et al. [28] propose SQLRight, a methodology for logic bug exploration based on coverage-based fuzzing. SQLRight successfully detects logic bugs in SQLite and MySQL. In their approach, all sample queries are added to a queue, and the fuzzer selects a query to mutate. If the mutated query triggers a new execution path, it is added to the queue. These mutated q",
+      "section": "8 Related Work",
+      "page": 22,
+      "char_offset": 68738,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "While SQLancer relies on a rule-based generator to detect logic bugs, limiting its exploration of input samples.",
+      "context_before": "w. If the DBMS fails to retrieve the pivot row based on this random expression, it indicates the presence of a logic bug in the DBMS. NoREC, on the other hand, translates an optimized query into an unoptimized version with identical semantics. By comparing the results obtained from both queries, inconsistencies can indicate the presence of a logic bug. Lastly, TLP partitions a query into several sub-queries, which are composed to have the same semantics as the original query. If the composed query does not yield the same result as the original query, it suggests the possibility of a logic bug.",
+      "context_after": "Liang et al. [28] propose SQLRight, a methodology for logic bug exploration based on coverage-based fuzzing. SQLRight successfully detects logic bugs in SQLite and MySQL. In their approach, all sample queries are added to a queue, and the fuzzer selects a query to mutate. If the mutated query triggers a new execution path, it is added to the queue. These mutated queries are then tested to identify any unexpected behaviors. However, SQLRight’s coverage-based fuzzing approach lacks high effectiveness in reaching vulnerable target sites due to the equal evaluation of all seeds. In this paper, we",
+      "section": "8 Related Work",
+      "page": 22,
+      "char_offset": 68984,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:14:29Z",
+    "is_model_written": true,
+    "summary": "SQLaser is a directed fuzzer guided by SQL clauses. The authors argue rule-based detection is limited by how hard rules are to specify, while coverage-guided fuzzing explores many paths unlikely to hold logic bugs. From examining existing non-crashing logic bugs across four DBMSs they derive 35 patterns, expressed as clause combinations backed by function sequences, and model these as error-prone function chains driven by a directed fuzzer with a new path-to-path distance mechanism. On SQLite, MySQL, PostgreSQL and TiDB it cut detection time by about 60% and found 22 bugs, four zero-day.",
+    "narrative": "SQLaser adopts SQLancer's oracles rather than competing with them: NoREC and TLP are described as the oracles it works with, PQS is noted as not yet deployed in it, and its argument is that SQLancer's rule-based generation constrains the code paths reached. Several of the bug patterns it studies come from SQLancer's own reported bugs.",
+    "roles": {
+      "M1": "background",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "background",
+      "M5": "motivation",
+      "M6": "definition",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "reuse_component",
+      "M10": "incidental",
+      "M11": "incidental"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M9 says PQS oracles have not yet been deployed into SQLaser, which implies NoREC and TLP have been, but no mention read states that the codebase itself is reused rather than the oracles reimplemented."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The contribution is clause-guided directed fuzzing; the oracles are used as they are."
+      },
+      "compares_with": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer's oracles are used within SQLaser rather than run against it in the mentions read."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The oracles are described by what they do rather than as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": [
+      "Whether NoREC and TLP are reimplemented or taken from SQLancer's code is not stated in the mentions read."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_14711_thesis_991012980220103412.json b/_data/papers/paper_doi_10_14711_thesis_991012980220103412.json
new file mode 100644
index 0000000..6d98cd1
--- /dev/null
+++ b/_data/papers/paper_doi_10_14711_thesis_991012980220103412.json
@@ -0,0 +1,654 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:43Z",
+  "paper": {
+    "id": "paper:doi:10.14711/thesis-991012980220103412",
+    "title": "Duplicate-sensitivity Guided Transformation Synthesis for DBMS Correctness Bug Detection",
+    "authors": [
+      "Yushan Zhang",
+      "Peisen Yao",
+      "Rongxin Wu",
+      "Charles Zhang"
+    ],
+    "year": 2021,
+    "venue": "arXiv.org",
+    "doi": "10.14711/thesis-991012980220103412",
+    "arxiv_id": "2107.03660",
+    "s2_paper_id": "e27baad105b7fba74658580ed2fb7439328b053c",
+    "url": "https://doi.org/10.14711/thesis-991012980220103412",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2107.03660",
+      "retrieved_at": "2026-09-09T01:33:43Z",
+      "chars": 55842,
+      "content_sha256": "sha256:7e8445653b2e37fb9ec98e07f48820f1bf8a62b9ca693c3628a9f89e0ee0e233"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 11,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2026
+      },
+      {
+        "number": "II",
+        "title": "ILLUSTRATIVE EXAMPLES",
+        "start": 7199
+      },
+      {
+        "number": "III",
+        "title": "CATEGORY 1: SELECT-FROM-WHERE",
+        "start": 8254
+      },
+      {
+        "number": "IV",
+        "title": "CATEGORY 2: SELECT-FROM-GROUP BY",
+        "start": 9555
+      },
+      {
+        "number": "V",
+        "title": "PRELIMINARIES",
+        "start": 11685
+      },
+      {
+        "number": "A",
+        "title": "SQL Syntax and Relational Model",
+        "start": 12029
+      },
+      {
+        "number": "B",
+        "title": "Aggregations",
+        "start": 14544
+      },
+      {
+        "number": "C",
+        "title": "Problem Statement",
+        "start": 15362
+      },
+      {
+        "number": "VI",
+        "title": "APPROACH",
+        "start": 16464
+      },
+      {
+        "number": "B",
+        "title": "Sensitivity-guided Transformation Synthesis",
+        "start": 21190
+      },
+      {
+        "number": "C",
+        "title": "Detecting Bugs with Eqsql",
+        "start": 24647
+      },
+      {
+        "number": "VII",
+        "title": "IMPLEMENTATION",
+        "start": 27402
+      },
+      {
+        "number": "VIII",
+        "title": "EVALUATION",
+        "start": 30710
+      },
+      {
+        "number": "A",
+        "title": "Evaluation Setup",
+        "start": 31699
+      },
+      {
+        "number": "B",
+        "title": "RQ1: Efficacy in Detection of Correctness Bugs",
+        "start": 35062
+      },
+      {
+        "number": "C",
+        "title": "RQ2: Significance of the Bugs",
+        "start": 38776
+      },
+      {
+        "number": "D",
+        "title": "RQ3: Coverage Improvement",
+        "start": 40542
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, Nov. 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 2,
+      "text": "How sqlite is tested. [Online]. Available: https://www.sqlite.org/testing. html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "The mysql test framework. [Online]. Available: https://dev.mysql.com/ doc/dev/mysql-server/8.0.21/PAGEMYSQL TESTRUN.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "A. Causevic, D. Sundmark, and S. Punnekkat, “Factors limiting industrial adoption of test driven development: A systematic review,” in 2011 Fourth IEEE International Conference on Software Testing, Verification and Validation. IEEE, 2011, pp. 337–346.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "L. Giakoumakis and C. A. Galindo-Legaria, “Testing sql server’s query optimizer: Challenges, techniques and experiences.” IEEE Computer Society Technical Committee on Data Engineerin, vol. 31, no. 1, pp. 36–43, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "S. A. Khalek, B. Elkarablieh, Y. O. Laleye, and S. Khurshid, “Queryaware test generation using a relational constraint solver,” in 23rd IEEE/ACM International Conference on Automated Software Engineering. IEEE, 2008, pp. 238–247.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "M. Veanes, N. Tillmann, and J. De Halleux, “Qex: Symbolic sql query explorer,” in International Conference on Logic for Programming Artificial Intelligence and Reasoning. Springer, 2010, pp. 425–446.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 9,
+      "text": "X. Chen, C. Wang, and A. Cheung, “Testing query execution engines with mutations,” in Proceedings of the Workshop on Testing Database Systems. ACM, jun 2020, pp. 1–5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 11,
+      "text": "Q. Zhou, J. Arulraj, S. Navathe, W. Harris, and D. Xu, “Automated verification of query equivalence using satisfiability modulo theories,” Proceedings of the VLDB Endowment, vol. 12, no. 11, pp. 1276–1288, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Mysql. [Online]. Available: https://www.mysql.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Pingcap: Tidb, sql at scale. [Online]. Available: https://pingcap.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Cloud database cynosdb (chinese). [Online]. Available: https://cloud. tencent.com/product/cynosdb",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "S. Chaudhuri, “An overview of query optimization in relational systems,” inProceedings of the seventeenth ACMSIGACT-SIGMOD-SIGART symposium on Principles of database systems, 1998, pp. 34–43.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "H. G. Elmongui, V. Narasayya, and R. Ramamurthy, “A framework for testing query transformation rules,” in Proceedings of the 2009 ACMSIGMOD International Conference on Management of data, 2009, pp. 257–268.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "M. Elhemali and L. Giakoumakis, “Unit-testing query transformation rules,” in Proceedings of the 1st International Workshop on Testing Database Systems, ser. DBTest ’08. New York, NY, USA: Association for Computing Machinery, 2008. [Online]. Available: https://doi.org/10.1145/1385269.1385273",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "U. Dayal, N. Goodman, and R. H. Katz, “An extended relational algebra with control over duplicate elimination,” in Proceedings of the 1st ACMSIGACT-SIGMOD symposium on Principles of database systems. ACM, 1982, pp. 117–123.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "A formal semantics of SQL queries, its validation, and applications, vol. 11, no. 1. VLDB Endowment, sep 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "L. Libkin, “Sql’s three-valued logic and certain answers,” ACM Trans. Database Syst., vol. 41, no. 1, Mar. 2016. [Online]. Available: https://doi.org/10.1145/2877206",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "L. Hella, L. Libkin, J. Nurmonen, and L. Wong, “Logics with aggregate operators,” in Proceedings of the 14th Symposium on Logic in Computer Science. IEEE, 1999, pp. 35–44.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “Squirrel: Testing database management systems with language validity and coverage feedback,” in Proceedings of the 2020 ACMSIGSAC Conference on Computer and Communications Security, 2020, pp. 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "A. Silberschatz, H. F. Korth, and S. Sudarshan, “Databasesystem concepts, seventh edition,” 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Golang. [Online]. Available: https://golang.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Pingcap. [Online]. Available: https://github.com/pingcap/go-randgen",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Grammar file (bison 3.7.1). [Online]. Available: https://www.gnu.org/ software/bison/manual/html node/Grammar-File.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Postgresql php generatorfree postgres php generator by sql maestro group. [Online]. Available: https://www.sqlmaestro.com/ products/postgresql/phpgenerator/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "N. Hollingum, “On the practice and application of context-free language reachability,” 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "xwb1989/sqlparser: Sql parser implemented in go. [Online]. Available: https://github.com/xwb1989/sqlparser",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Db-engines rankingpopularity ranking of relational dbms. [Online]. Available: https://db-engines.com/en/ranking/relational+ dbms",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Mysql :: Mysql 8.0 reference manual :: 1.3 what is new in mysql 8.0. [Online]. Available: https://dev.mysql.com/doc/refman/8.0/ en/mysql-nutshell.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "D. R. Slutz, “Massive stochastic testing of sql,” in VLDB, vol. 98. Citeseer, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Mysql bugs. [Online]. Available: https://bugs.mysql.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Gcov (using the gnu compiler collection (gcc)). [Online]. Available: https://gcc.gnu.org/onlinedocs/gcc/Gcov.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Use mod result (float number) in having/where is not considered as true. [Online]. Available: https://bugs.mysql.com/bug.php?id=100270",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Mariadb foundationmariadb.org. [Online]. Available: https://mariadb. org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Mysql bugs: #99971: colmn name from outer query cannot be resolved in the subquery. [Online]. Available: https://bugs.mysql.com/bug.php? id=99971",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "S. Abdul Khalek and S. Khurshid, “Automated sql query generation for systematic testing of database engines,” in Proceedings of the IEEE/ACM international conference on Automated software engineering, 2010, pp. 329–332.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Alloy. [Online]. Available: https://alloytools.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "M. Veanes, P. Grigorenko, P. De Halleux, and N. Tillmann, “Symbolic query exploration,” in International Conference on Formal Engineering Methods. Springer, 2009, pp. 49–68.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "B. Ghit, N. Poggi, J. Rosen, R. Xin, and P. Boncz, “Sparkfuzz: searching correctness regressions in modern query engines,” in Proceedings of the workshop on Testing Database Systems. ACM, jun 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Postgresql: Documentation: 13: Chapter 8. data types. [Online]. Available: https://www.postgresql.org/docs/current/datatype.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "T. Y. Chen, S. C. Cheung, and S. M. Yiu, “Metamorphic testing: A new approach for generating next test cases,” 2020.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 1,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, Nov. 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 8,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 10,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[1]",
+      "technique": "pqs",
+      "sentence": "Unexpectedly, researchers could still find hundreds of queries where the popular and production-level DBMSs return wrong result sets [1], though these systems have been extensively tested during their development [2], [3].",
+      "context_before": "quivalence. After candidate synthesis, we keep the mutant query which is equivalent to the given one by using a query equivalent checker. Furthermore, we have implemented our idea in a tool Eqsql and used it to test the production-level DBMSs. In two months, we detected in total 30 newly confirmed and unique bugs in MySQL, TiDB and CynosDB. Index Terms —DBMS, database system, SQL, testing I. INTRODUCTION Database Management Systems (DBMSs) are widely used in modern industries. The users expect the DBMS to retrieve correct data records and report correct analysis results on the giving databases.",
+      "context_after": "To validate the correctness of different DBMS components (e.g. the optimizer), the developers usually manually write test cases and provide the expected result for a query. However, the quality of test cases is largely limited by developers’ testingskills [4] and the hand-written test cases only cover a very small part of the possible input query space [5], leading to insufficient testing of the DBMS. In what follows, we refer to a test case for the purpose of DBMS correctness testing as a test query and an expected result set. Recent studies [6]–[10] have proposed several approaches to better",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2235,
+      "cited_reference": {
+        "number": 1,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, Nov. 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[6]–[10]",
+      "technique": "norec",
+      "sentence": "Recent studies [6]–[10] have proposed several approaches to better explore the query space.",
+      "context_before": "ems have been extensively tested during their development [2], [3]. To validate the correctness of different DBMS components (e.g. the optimizer), the developers usually manually write test cases and provide the expected result for a query. However, the quality of test cases is largely limited by developers’ testingskills [4] and the hand-written test cases only cover a very small part of the possible input query space [5], leading to insufficient testing of the DBMS. In what follows, we refer to a test case for the purpose of DBMS correctness testing as a test query and an expected result set.",
+      "context_after": "They all use automated random query generation to enlarge the explored query space. However, the key challenge is the test oracle problem, i.e., how to find the expected result set of a given query. To address the problem, one category of approaches such as ADUSA [6] and Qex [7] convert the query to Satisfiability Modulo Theory (SMT) constraints and use a constraint solver to deduce the results. However, they cannot work on special data types (e.g. varchar) in the DBMS and big tables because of the performance issue of the solver. Another category of approaches identify a correctness bug by com",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 2990,
+      "cited_reference": {
+        "number": 8,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For example, TLP [10] first generates a query Qand then uses a predefined transformation to on Qto obtain a new equivalent queryQ′.",
+      "context_before": "find the expected result set of a given query. To address the problem, one category of approaches such as ADUSA [6] and Qex [7] convert the query to Satisfiability Modulo Theory (SMT) constraints and use a constraint solver to deduce the results. However, they cannot work on special data types (e.g. varchar) in the DBMS and big tables because of the performance issue of the solver. Another category of approaches identify a correctness bug by comparing the result set of a pair of equivalent queries. However, existing approaches can only use limited rules/transformations to derive the query pairs.",
+      "context_after": "Next, it checks whether the DBMS generates the same result under the input of QandQ′. However, since some correctness bugs can only be exposed under a certain query pair and the predefined transformation may fail to generate such query pair, these approaches would lose the chances to detect such bugs. To be more specific, the transformation rule of the existing studies [1], [8], [10] presumes that the generated query pair should explicitly include a “where” predicate (e.g. where a != 0) and the bug was triggered by the inconsistent interpretations of “where” predicates. However, as shown in Fig",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 3835,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[1]",
+      "technique": "pqs",
+      "sentence": "To be more specific, the transformation rule of the existing studies [1], [8], [10] presumes that the generated query pair should explicitly include a “where” predicate (e.",
+      "context_before": "ness bug by comparing the result set of a pair of equivalent queries. However, existing approaches can only use limited rules/transformations to derive the query pairs. For example, TLP [10] first generates a query Qand then uses a predefined transformation to on Qto obtain a new equivalent queryQ′. Next, it checks whether the DBMS generates the same result under the input of QandQ′. However, since some correctness bugs can only be exposed under a certain query pair and the predefined transformation may fail to generate such query pair, these approaches would lose the chances to detect such bugs.",
+      "context_after": "g. where a != 0) and the bug was triggered by the inconsistent interpretations of “where” predicates. However, as shown in Figure 1, this bug is exposed by two equivalent queries QA (Fig 6b) and QB(Fig 6c) in which there are no “where” predicates. Essentially, this bug is caused by an inconsistent representation of the float-point value 0.001.QAandQBarearXiv:2107.03660v1 [cs.SE] 8 Jul 2021 expected to return the same value on the given table t0, but they have a different value for 0.001 as shown in the result column. The key point in identifying such a correctness bug in Figure 1 is to compare",
+      "section": "I INTRODUCTION",
+      "page": 1,
+      "char_offset": 4267,
+      "cited_reference": {
+        "number": 1,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation. USENIX Association, Nov. 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Previous work NoREC [8] can only detect such bugs because it only manipulates the “where” predicate to create an unoptimized query mutant.",
+      "context_before": "responding seed query and the generated query mutant. The gray color highlights where the transformation takes place. III. CATEGORY 1: SELECT-FROM-WHERE This category is one of the most frequent queries in the DBMS, which asks the system to filter out data records not satisfying the “where” predicate. These queries will go through complex equivalent transformations during the optimization phase [15], potentially resulting in a nonequivalent final result set due to implementation bugs [16], [17]. However, such bugs are hard to be captured due to the large amount of optimization combinations [16].",
+      "context_after": "For queries in this category, there are many candidate transformations to apply. In this example, our approach inserts additional operators to the seed query. By adding DISTINCT to SELECT a in (Figure 2), it generated QA. And by adding groups with GROUP BY, it gets QB. Both inserted operators remove duplicates of an avalue and keep only one copy of each different avalue. Because adding duplicates for an a value or remove some duplicates will not change the result for DISTINCT and GROUP BY, they are duplicate insensitive. The operators de-duplicate the input tuples and output results with one copy",
+      "section": "III CATEGORY 1: SELECT-FROM-WHERE",
+      "page": 2,
+      "char_offset": 8736,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Thus, the only available tool is SQLancer [1], [8], [10].",
+      "context_before": "ecause CDB is not open-source and it is a cloud product, we set up the test environment with a Tencent VPS and one CDB instance compatible with MySQL 5.7 and tested the database server on the VPS. 1Lines of code. 2Provided by the developer. TABLE IV: Confirmed bugs in the tested DBMSs. Name Value3Error4Total Fixed MySQL 12 2 14 0 TiDB 6 7 13 12 CynosDB 2 1 3 0 TOTAL 20 10 30 12 Baselines. We have to admit that a nuts-and-bolts comparison among the related tools is hard to achieve. For the early studies [6], [7], [32], the tools are not available. More recent tool MutaSQL [9] is not open source.",
+      "context_after": "We omitted comparison with SQLancer for three reasons. Firstly, both tools support SQL grammar in a different way which makes it hard to compare fairly. More importantly, it doesn’t take seed input and ends in log explosion when executing queries. Additionally, we designed our evaluation to run on new versions of the tested database systems, which is manual-extensive and objective to run both tools and compare the bugs. Although we did not design a separate experiment, we did evaluate both tools on coverage. Our experiment result showed that both tools achieved almost the same coverage when r",
+      "section": "A Evaluation Setup",
+      "page": 8,
+      "char_offset": 34170,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We omitted comparison with SQLancer for three reasons.",
+      "context_before": "e set up the test environment with a Tencent VPS and one CDB instance compatible with MySQL 5.7 and tested the database server on the VPS. 1Lines of code. 2Provided by the developer. TABLE IV: Confirmed bugs in the tested DBMSs. Name Value3Error4Total Fixed MySQL 12 2 14 0 TiDB 6 7 13 12 CynosDB 2 1 3 0 TOTAL 20 10 30 12 Baselines. We have to admit that a nuts-and-bolts comparison among the related tools is hard to achieve. For the early studies [6], [7], [32], the tools are not available. More recent tool MutaSQL [9] is not open source. Thus, the only available tool is SQLancer [1], [8], [10].",
+      "context_after": "Firstly, both tools support SQL grammar in a different way which makes it hard to compare fairly. More importantly, it doesn’t take seed input and ends in log explosion when executing queries. Additionally, we designed our evaluation to run on new versions of the tested database systems, which is manual-extensive and objective to run both tools and compare the bugs. Although we did not design a separate experiment, we did evaluate both tools on coverage. Our experiment result showed that both tools achieved almost the same coverage when running for 12 hours by controlling the available parame",
+      "section": "A Evaluation Setup",
+      "page": 8,
+      "char_offset": 34228,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "More recently, PQS [1] synthesizes queries instead of solving constraints.",
+      "context_before": "ver, it knows which concrete values can satisfy the constraint and should be returned by the original query. It then inserts the value to the DBMS as data and checks if executing the query will return the same value. However, due to the incomplete model, ADUSA can only convert query with limited data types to Alloy specification. For example, it cannot convert a DBMS varchar field into the Alloy language. Qex [7], [40] is another symbolic query explorer which models more SQL features, but its practical adoption is hindered by high overhead in symbolically processing big tables and complex joins.",
+      "context_after": "It implements an expression generator, which takes a table row from the database and generates an expression that is evaluated to true on the row. Then it runs the query in the DBMS and checks if the row is contained in a query using the generated expression as the “where” condition. Though PQS can generate the test oracle more practically than the previous work, it can only reveal bugs with a symptom of missing the picked row. Compared to all the previous works, our idea is more generally applicable to the DBMS, and it can detect bugs having different symptoms (e.g., “order by” has wrong ord",
+      "section": "D RQ3: Coverage Improvement",
+      "page": 10,
+      "char_offset": 46617,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "Though PQS can generate the test oracle more practically than the previous work, it can only reveal bugs with a symptom of missing the picked row.",
+      "context_before": "t a DBMS varchar field into the Alloy language. Qex [7], [40] is another symbolic query explorer which models more SQL features, but its practical adoption is hindered by high overhead in symbolically processing big tables and complex joins. More recently, PQS [1] synthesizes queries instead of solving constraints. It implements an expression generator, which takes a table row from the database and generates an expression that is evaluated to true on the row. Then it runs the query in the DBMS and checks if the row is contained in a query using the generated expression as the “where” condition.",
+      "context_after": "Compared to all the previous works, our idea is more generally applicable to the DBMS, and it can detect bugs having different symptoms (e.g., “order by” has wrong ordering).Differential Testing. Differential testing is a common approach to decide which result set is incorrect. RAGS [32] and SparkFuzz [41] are in this category. They compare the result sets of one query from several chosen DBMSs, which are expected to return the same result set. The SQL Server team proposed a self-differential approach [16] for their query optimization rules, which compares results of the same query with turni",
+      "section": "D RQ3: Coverage Improvement",
+      "page": 10,
+      "char_offset": 46977,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Similarly, NoREC [8] constructs a non-optimizing version of a query and compares the result.",
+      "context_before": "e result set. The SQL Server team proposed a self-differential approach [16] for their query optimization rules, which compares results of the same query with turning on and off the optimization. However, differential testing cannot be used to test unique extensions such as PostgreSQL’s data type path to save a geometric path on a plane [42]. Metamorphic Testing. The key idea of metamorphic testing is to detect output violations of a bunch of related test cases [43]. In the context of our problem, it means to derive from the existing queries to obtain query mutants that have known result sets.",
+      "context_after": "However, they are only limited to testing the optimization, while our approach generally tests the DBMS. MutaSQL [9] takes a given equivalent transformation and construct a pair of equivalent queries and reports a bug when they return different result sets. In comparison, our approach use guided synthesis to discover new equivalent transformations, not relying on developer knowledge to provide them. XII. CONCLUSION In this paper, we presented a fully automated approach to generate equivalent query pairs, namely Eqsql, which can detect correctness bugs in the DBMS. Eqsql’s main contribution is",
+      "section": "D RQ3: Coverage Improvement",
+      "page": 10,
+      "char_offset": 48160,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M7"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:25:35Z",
+    "is_model_written": true,
+    "summary": "This work automates the discovery of query transformations for DBMS correctness testing. Existing transformation-based approaches generate an equivalent query pair and check that the system returns the same results, but rely on transformations written by hand, which limits the input space they explore. The proposed method synthesises many candidate transformations by mutating a query while preserving its duplicate sensitivity -- a necessary condition for equivalence -- then filters out the non-equivalent ones with a query equivalence checker. The tool, Eqsql, found 30 confirmed unique bugs in MySQL, TiDB and CynosDB in two months.",
+    "narrative": "PQS, NoREC and TLP are the hand-written transformations the paper sets out to synthesise automatically, and it argues each presumes an explicitly equivalent query pair. SQLancer is named as the only available comparable tool, and the authors then state they omitted the comparison, giving three reasons.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "background",
+      "M3": "definition",
+      "M4": "motivation",
+      "M5": "motivation",
+      "M6": "background",
+      "M7": "background",
+      "M8": "definition",
+      "M9": "motivation",
+      "M10": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The synthesised transformations occupy the same space as NoREC's and TLP's and the paper positions itself as automating what they do by hand, but no mention read claims to extend a named technique."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M7 says the comparison with SQLancer was omitted, giving three reasons. A pattern fired on that sentence, which records the absence of a comparison rather than one."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_14722_ndss_2025_230530.json b/_data/papers/paper_doi_10_14722_ndss_2025_230530.json
new file mode 100644
index 0000000..1369ee4
--- /dev/null
+++ b/_data/papers/paper_doi_10_14722_ndss_2025_230530.json
@@ -0,0 +1,854 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:18:16Z",
+  "paper": {
+    "id": "paper:doi:10.14722/ndss.2025.230530",
+    "title": "MTZK: Testing and Exploring Bugs in Zero-Knowledge (ZK) Compilers",
+    "authors": [
+      "Dongwei Xiao",
+      "Zhibo Liu",
+      "Yiteng Peng",
+      "Shuai Wang"
+    ],
+    "year": 2025,
+    "venue": "Network and Distributed System Security Symposium",
+    "doi": "10.14722/ndss.2025.230530",
+    "arxiv_id": null,
+    "s2_paper_id": "654064b6d2b49ac7d66bc4a8b296a4806221ab66",
+    "url": "https://doi.org/10.14722/ndss.2025.230530",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.14722/ndss.2025.230530",
+      "retrieved_at": "2026-09-08T15:18:16Z",
+      "chars": 112759,
+      "content_sha256": "sha256:1d919cdaf4f72e426b6b173e9178948714154e3ffb3c6cb4561bb678bd930c65"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 18,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 1651
+      },
+      {
+        "number": "II",
+        "title": "PRELIMINARY",
+        "start": 7305
+      },
+      {
+        "number": "III",
+        "title": "MOTIVATION",
+        "start": 15321
+      },
+      {
+        "number": "A",
+        "title": "Study Scope",
+        "start": 19783
+      },
+      {
+        "number": "C",
+        "title": "Also, since the number of constraints in EQC significantly",
+        "start": 23226
+      },
+      {
+        "number": "C",
+        "title": "Usage Scenario and Clarification",
+        "start": 24735
+      },
+      {
+        "number": "A",
+        "title": "IR Design",
+        "start": 33455
+      },
+      {
+        "number": "B",
+        "title": "Program Generation",
+        "start": 35685
+      },
+      {
+        "number": "C",
+        "title": "MRSIM: Satisfiability-Invariant Mutation",
+        "start": 40099
+      },
+      {
+        "number": "D",
+        "title": "MRIV M: Information Visibility Mutation",
+        "start": 45013
+      },
+      {
+        "number": "A",
+        "title": "RQ1: Testing Efficiency and Effectiveness",
+        "start": 56170
+      },
+      {
+        "number": "B",
+        "title": "RQ2: Bug Characteristics and Root Causes",
+        "start": 64043
+      },
+      {
+        "number": "C",
+        "title": "RQ3: Effectiveness of Each Testing Stage",
+        "start": 71691
+      },
+      {
+        "number": "D",
+        "title": "RQ4: Security Implications of Bugs",
+        "start": 74302
+      },
+      {
+        "number": "A",
+        "title": "Comparison of Bugs with Existing Works",
+        "start": 111244
+      },
+      {
+        "number": "B",
+        "title": "Security, Privacy, and Ethical Concerns",
+        "start": 112027
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Lifted A. Aztec network raises $100m in series b funding. https: //nfthorizon.io/aztec-network-raises-100m-in-series-b-funding/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "aharshbe. Token-mining application in aleo. https://github.com/AleoHQ/ workshop/blob/master/token/src/main.leo, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Aleo. Aleofully private applications. https://aleo.org/, 2023. 6This conceptually correlates to the classic backdoor attack noted in Ken Thompson’s Turning award lecture [ 106] — “Reflections on Trusting Trust.”",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Aleo. Leo — zero-knowledge programming language. https://leo-lang. org/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Robert Andrew. 21st century technologies: Zero-knowledge proofs. https://citylife.capetown/uncategorized/21st-century-technologies-zeroknowledge-proofs/31487/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Sanjeev Arora and Boaz Barak. Computational complexity: a modern approach. Cambridge University Press, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Aztec. $50,000 usd bug bounty. https://twitter.com/aztecnetwork/status/ 1453773879898378241, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Aztec. Disclosure of recent vulnerabilities. https://hackmd.io/ @aztec-network/disclosure-of-recent-vulnerabilities, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Aztec. Aztecethereum, encrypted. https://aztec.network/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Jos´e Bacelar Almeida, Manuel Barbosa, Endre Bangerter, Gilles Barthe, Stephan Krenn, and Santiago Zanella B ´eguelin. Full proof cryptography: Verifiable compilation of efficient zero-knowledge protocols. InProceedings of the 2012 ACM Conference on Computer and Communications Security, CCS ’12, page 488–500, New York, NY, USA, 2012. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Eli Ben-Sasson, Iddo Bentov, Yinon Horesh, and Michael Riabzev. Scalable, transparent, and post-quantum secure computational integrity. Cryptology ePrint Archive, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Lukas Bernhard, Tobias Scharnowski, Moritz Schloegel, Tim Blazytko, and Thorsten Holz. JIT-Picking: Differential Fuzzing of JavaScript Engines. In Proceedings of the 2022 ACMSIGSAC Conference on Computer and Communications Security, CCS ’22, pages 351–364, New York, NY, USA, 2022. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "EY Blockchain. Nightfall. https://github.com/EYBlockchain/nightfall, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "EY Blockchain. Nightfall3. https://github.com/EYBlockchain/ nightfall 3, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Sean Bowe, Jack Grigg, and Daira Hopwood. Recursive proof composition without a trusted setup. Cryptology ePrint Archive, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Alexander Bulekov, Bandan Das, Stefan Hajnoczi, and Manuel Egele. No grammar, no problem: Towards fuzzing the linux kernel without system-call descriptions. In Network and Distributed System Security (NDSS) Symposium, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Cristian Cadar, Daniel Dunbar, Dawson R Engler, et al. KLEE: unassisted and automatic generation of high-coverage tests for complex systems programs. In OSDI, volume 8, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Cairo. Cairolinear types. https://docs.cairo-lang.org/language constructs/linear-types.html, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Cairo. Cairo repo. https://github.com/starkware-libs/cairo/tree/main, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Cairo. cairo/docs/security.md at main ·starkware-libs/cairo ·github. https://github.com/starkware-libs/cairo/blob/main/docs/ SECURITY.md, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Cario. Cairoa stark-based turing-complete language for writing provable programs on blockchain. https://www.cairo-lang.org/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Chainlink. Chainlink — zero knowledge proofs. https://blog.chain.link/zero-knowledge-proof-projects/, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Chainlink. What are zk-rollups (zero-knowledge rollups)? https://blog.chain.link/zero-knowledge-rollup/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Kyle Charbonnet. Missing checks to ensure ’zero’ is less than snark scalar field. https://github.com/privacy-scaling-explorations/zk-kit/ issues/23, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Tsong Y Chen, Shing C Cheung, and Shiu Ming Yiu. Metamorphic testing: a new approach for generating next test cases. Technical report, Technical Report HKUST-CS98-01, Department of Computer Science, Hong Kong. .. , 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Tsong Y Chen, Shing C Cheung, and Shiu Ming Yiu. Metamorphic testing: a new approach for generating next test cases. arXiv preprint arXiv:2002.12543, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Alessandro Chiesa, Yuncong Hu, Mary Maller, Pratyush Mishra, Psi Vesely, and Nicholas Ward. Marlin: Preprocessing zksnarks with universal and updatable srs. Cryptology ePrint Archive, Paper 2019/1047, 2019. https://eprint.iacr.org/2019/1047.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Collin Chin, Howard Wu, Raymond Chu, Alessandro Coglio, Eric McCarthy, and Eric Smith. Leo: A programming language for formally verified, zero-knowledge applications. Cryptology ePrint Archive, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Jaeseung Choi, Kangsu Kim, Daejin Lee, and Sang Kil Cha. Ntfuzz: Enabling type-aware kernel fuzzing on windows with static binary analysis. In pages 677–693. IEEE, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "circomspect. Circomspect. https://github.com/trailofbits/circomspect, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Alessandro Coglio, Eric McCarthy, Eric Smith, Collin Chin, Pranav Gaddamadugu, and Michel Dellepere. Compositional formal verification of zero-knowledge circuits. Cryptology ePrint Archive, Paper 2023/1278, 2023. https://eprint.iacr.org/2023/1278.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Alastair F Donaldson, Hugues Evrard, Andrei Lascu, and Paul Thomson. Automated testing of graphics shader compilers. Proceedings of the ACM on Programming Languages, 1(OOPSLA):1–29, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Emine Dumlu, Cemal Yilmaz, Myra B Cohen, and Adam Porter. Feedback driven adaptive combinatorial testing. In Proceedings of the 2011 international symposium on software testing and analysis, pages 243–253, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Jacob Eberhardt and Stefan Tai. Zokratesscalable privacy-preserving off-chain computations. In on Internet of Things (iThings) and IEEE Green Computing and Communications (GreenCom) and IEEE Cyber, Physical and Social Computing (CPSCom) and IEEE Smart Data (SmartData), pages 1084– 1091, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "ecne. Ecne. https://github.com/franklynwang/EcneProject, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Jens Ernstberger, Stefanos Chaliasos, George Kadianakis, Sebastian Steinhorst, Philipp Jovanovic, Arthur Gervais, Benjamin Livshits, and Michele Orr `u. zk-bench: A toolset for comparative evaluation and performance benchmarking of snarks. Cryptology ePrint Archive, Paper 2023/1503, 2023. https://eprint.iacr.org/2023/1503.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Ethereum. Ethereum — zero knowledge proofs. https://ethereum.org/en/zero-knowledge-proofs/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Paul Fiterau-Brostean, Bengt Jonsson, Robert Merget, Joeri De Ruiter, Konstantinos Sagonas, and Juraj Somorovsky. Analysis of {DTLS } implementations using protocol state fuzzing. In 29th USENIX Security Symposium (USENIX Security 20), pages 2523–2540, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Marius Fleischer, Dipanjan Das, Priyanka Bose, Weiheng Bai, Kangjie Lu, Mathias Payer, Christopher Kruegel, and Giovanni Vigna. {ACTOR }:{Action-Guided }kernel fuzzing. In 32nd USENIX Security Symposium (USENIX Security 23), pages 5003–5020, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Anoma Foundation. Zkp (circuit) compiler shootout. https://github. com/anoma/zkp-compiler-shootout, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "C´edric Fournet, Chantal Keller, and Vincent Laporte. A certified compiler for verifiable computing. In 2016 IEEE 29th Computer Security Foundations Symposium (CSF), pages 268–280, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Tim Freeman and Frank Pfenning. Refinement types for ml. In Proceedings of the ACMSIGPLAN 1991 conference on Programming language design and implementation, pages 268–277, 1991.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Andres Freund. [oss-security] backdoor in upstream xz/liblzma leading to ssh server compromise. https://lwn.net/ml/oss-security/ 20240329155126.kjjfduxw2yrlxgzm@awork3.anarazel.de/, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Ariel Gabizon, Zachary J. Williamson, and Oana Ciobotaru. Plonk: Permutations over lagrange-bases for oecumenical noninteractive arguments of knowledge. Cryptology ePrint Archive, Paper 2019/953, 2019. https://eprint.iacr.org/2019/953.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Oliver Gale. Zero-knowledge proofs: Use-cases for businesses. https://www.forbes.com/sites/forbesbusinesscouncil/2023/04/19/ zero-knowledge-proofs-use-cases-for-businesses/?sh=1e7f6e5c58b8, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Xinyang Ge, Ben Niu, Robert Brotzman, Yaohui Chen, HyungSeok Han, Patrice Godefroid, and Weidong Cui. Hyperfuzzer: An efficient hybrid fuzzer for virtual cpus. In Proceedings of the 2021 ACMSIGSAC Conference on Computer and Communications Security, pages 366–378, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Lior Goldberg, Shahar Papini, and Michael Riabzev. Cairo – a turingcomplete stark-friendly cpu architecture. Cryptology ePrint Archive, Paper 2021/1063, 2021. https://eprint.iacr.org/2021/1063.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "S Goldwasser, S Micali, and C Rackoff. The knowledge complexity of interactive proof-systems. In Proceedings of the Seventeenth Annual ACM Symposium on Theory of Computing, STOC ’85, page 291–304, New York, NY, USA, 1985. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Nihal R Gowravaram. Zero Knowledge Proofs and Applications to Financial Regulation. PhD thesis, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Samuel Groß, Simon Koch, Lukas Bernhard, Thorsten Holz, and Martin Johns. Fuzzilli: Fuzzing for javascript jit compiler vulnerabilities. In NDSS, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Jens Groth. On the size of pairing-based non-interactive arguments. InAdvances in Cryptology–EUROCRYPT 2016: 35th Annual International Conference on the Theory and Applications of Cryptographic Techniques, Vienna, Austria, May 8-12, 2016, Proceedings, Part II 35, pages 305–326. Springer, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Jens Groth and Mary Maller. Snarky signatures: minimal signatures of knowledge from simulation-extractable snarks. Cryptology ePrint Archive, Paper 2017/540, 2017. https://eprint.iacr. org/2017/540.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Samuel Groß. A case study of javascriptcore and cve-2016-4622. http://www.phrack.org/issues/70/3.html#article, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Samuel Groß. Exploiting logic bugs in javascript jit engines. http: //www.phrack.org/issues/70/9.html#article, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Alyssa Hertig. Zcash team reveals it fixed a catastrophic coin counterfeiting bug. https://www.coindesk.com/markets/2019/02/05/ zcash-team-reveals-it-fixed-a-catastrophic-coin-counterfeiting-bug/, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Jaewon Hur, Suhwan Song, Dongup Kwon, Eunjin Baek, Jangwoo Kim, and Byoungyoung Lee. Difuzzrtl: Differential fuzz testing to find cpu bugs. In pages 1286–1303. IEEE, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Immunefi. Polygon double-spend bugfix review — $2m bounty. https://medium.com/immunefi/ polygon-double-spend-bug-fix-postmortem-2m-bounty-5a1db09db7f1, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Haoxiang Jia, Ming Wen, Zifan Xie, Xiaochen Guo, Rongxin Wu, Maolin Sun, Kang Chen, and Hai Jin. Detecting JVMJIT Compiler Bugs via Exploring Two-Dimensional Input Spaces. In 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE), pages 43–55, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. {DynSQL }: Stateful fuzzing for database management systems with complex and valid {SQL }query generation. In 32nd USENIX Security Symposium (USENIX Security 23), pages 4949–4965, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. Detecting transactional bugs in database engines via {Graph-Based }oracle construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23), pages 397–417, 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 61,
+      "text": "Guoliang Jin, Linhai Song, Xiaoming Shi, Joel Scherpelz, and Shan Lu. Understanding and detecting real-world performance bugs. ACMSIGPLAN Notices, 47(6):77–88, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Swihart Josh, Winston Benjamin, and Sean Bowe. Zcash counterfeiting vulnerability successfully remediated. https://electriccoin.co/blog/ zcash-counterfeiting-vulnerability-successfully-remediated/, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Jurarpereurs. Proving phase fails for operations on negative literals. https://github.com/noir-lang/noir/issues/2098, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "kevaundray, jfecher, and French Tom. Noir. https://noir-lang.org/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Kyungtae Kim, Dae R Jeong, Chung Hwan Kim, Yeongjin Jang, Insik Shin, and Byoungyoung Lee. Hfl: Hybrid fuzzing on the linux kernel. InNDSS, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "George Kushnir. Visibility declarations inside structs. https://github. com/noir-lang/noir/issues/2135, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. Compiler validation via equivalence modulo inputs. ACM Sigplan Notices, 49(6):216–226, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Vu Le, Chengnian Sun, and Zhendong Su. Finding deep compiler bugs via guided stochastic program mutation. ACMSIGPLAN Notices, 50(10):386–399, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Vu Le, Chengnian Sun, and Zhendong Su. Randomized stress-testing of link-time optimizers. In Proceedings of the 2015 International Symposium on Software Testing and Analysis, pages 327–337, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Bastien Lecoeur, Hasan Mohsin, and Alastair F Donaldson. Program reconditioning: Avoiding undefined behaviour when finding and reducing compiler bugs. Proceedings of the ACM on Programming Languages, 7(PLDI):1801–1825, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Daniel Lehmann and Michael Pradel. Feedback-directed differential testing of interactive debuggers. In Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 610–620, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Cong Li, Yanyan Jiang, Chang Xu, and Zhendong Su. Validating jit compilers via compilation space exploration. In Proceedings of the 29th Symposium on Operating Systems Principles, SOSP ’23, page 66–79, New York, NY, USA, 2023. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Shaohua Li, Theodoros Theodoridis, and Zhendong Su. Boosting compiler testing by injecting real-world code. Proc. ACM Program. Lang., 8(PLDI), jun 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Yichen Li, Dongwei Xiao, Zhibo Liu, Qi Pang, and Shuai Wang. Metamorphic testing of secure multi-party computation (mpc) compilers. Proc. ACM Softw. Eng., 1(FSE), jul 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Junrui Liu, Ian Kretz, Hanzhi Liu, Bryan Tan, Jonathan Wang, Yi Sun, Luke Pearson, Anders Miltner, I s¸ıl Dillig, and Yu Feng. Certifying zero-knowledge circuits with refinement types. Cryptology ePrint Archive, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Zhibo Liu, Dongwei Xiao, Zongjie Li, Shuai Wang, and Wei Meng. Exploring missed optimizations in webassembly optimizers. In Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, page 436–448, New York, NY, USA, 2023. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "Vsevolod Livinskii, Dmitry Babokin, and John Regehr. Random testing for C and C++ compilers with YARPGen. Proceedings of the ACM on Programming Languages, 4(OOPSLA):196:1–196:25, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Vsevolod Livinskii, Dmitry Babokin, and John Regehr. Fuzzing Loop Optimizations in Compilers for C++ and Data-Parallel Languages. Proceedings of the ACM on Programming Languages, 7(PLDI):181:1826– 181:1847, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 79,
+      "text": "Haoyang Ma, Qingchao Shen, Yongqiang Tian, Junjie Chen, and Shing-Chi Cheung. Fuzzing deep learning compilers with hirgen. InProceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, page 248–260, New York, NY, USA, 2023. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Ceyhun Onur and Arda Yurdakul. Electanon: A blockchain-based, anonymous, robust and scalable ranked-choice voting protocol. arXiv preprint arXiv:2204.00057, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "Alex Ozdemir, Fraser Brown, and Riad S Wahby. Circ: Compiler infrastructure for proof systems, software verification, and more. In. IEEE, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Alex Ozdemir, Riad S. Wahby, Fraser Brown, and Clark Barrett. Bounded verification for finite-field-blasting (in a compiler for zero knowledge proofs). Cryptology ePrint Archive, Paper 2023/778, 2023. https://eprint.iacr.org/2023/778.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Shankara Pailoor, Yanju Chen, Franklyn Wang, Clara Rodr ´ıguez, Jacob Van Geffen, Jason Morton, Michael Chu, Brian Gu, Yu Feng, and Is¸ıl Dillig. Automated detection of under-constrained circuits in zero-knowledge proofs. Proceedings of the ACM on Programming Languages, 7(PLDI):1510–1532, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Qi Pang, Yuanyuan Yuan, and Shuai Wang. Mdpfuzz: testing models solving markov decision processes. In Proceedings of the 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2022, page 378–390, New York, NY, USA, 2022. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "Kexin Pei, Yinzhi Cao, Junfeng Yang, and Suman Jana. DeepXplore: Automated whitebox testing of deep learning systems. In Proceedings of the 26th Symposium on Operating Systems Principles, SOSP ’17, pages 1–18, New York, NY, USA, 2017. ACM.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Hui Peng, Zhihao Yao, Ardalan Amiri Sani, Dave Jing Tian, and Mathias Payer. {GLeeFuzz }: Fuzzing {WebGL }through error message guided mutation. In 32nd USENIX Security Symposium (USENIX Security 23), pages 1883–1899, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "picus. Picus. https://github.com/Veridise/Picus, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "Polygon. Web3, aggregated. https://polygon.technology/, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140–1152, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 90,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pages 667–682, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 91,
+      "text": "Yuyang Sang, Ning Luo, Samuel Judson, Ben Chaimberg, Timos Antonopoulos, Xiao Wang, Ruzica Piskac, and Zhong Shao. Ou: Automating the parallelization of zero-knowledge protocols. Cryptology ePrint Archive, Paper 2023/657, 2023. https://eprint.iacr.org/2023/657.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "Qingchao Shen, Haoyang Ma, Junjie Chen, Yongqiang Tian, Shing-Chi Cheung, and Xiang Chen. A comprehensive study of deep learning compiler bugs. In Proceedings of the 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2021, page 968–980, New York, NY, USA, 2021. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "shuklaayush. Casting negative literals to unsigned doesn’t work. https: //github.com/noir-lang/noir/issues/2045, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "Juraj Somorovsky. Systematic fuzzing and testing of tls libraries. In Proceedings of the 2016 ACMSIGSAC conference on computer and communications security, pages 1492–1504, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "Fatemeh Heidari Soureshjani, Mathias Hall-Andersen, MohammadMahdi Jahanara, Jeffrey Kam, Jan Gorzny, and Mohsen Ahmadvand. Automated analysis of halo2 circuits. Cryptology ePrint Archive, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "StarkWare. Starknet. https://starkware.co/starknet/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "stealdrop. Stealthdrop: Anonymous airdrops using zk proofs. https: //github.com/stealthdrop/stealthdrop/blob/main/README.md, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 98,
+      "text": "Chengnian Sun, Vu Le, and Zhendong Su. Finding compiler bugs via live code mutation. In Proceedings of the 2016 ACMSIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications, pages 849–863, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 99,
+      "text": "Yoshiki Takashima, Ruben Martins, Limin Jia, and Corina S. P ˘as˘areanu. Syrust: Automatic testing of rust libraries with semantic-aware program synthesis. In Proceedings of the 42nd ACMSIGPLAN International Conference on Programming Language Design and Implementation, PLDI 2021, page 899–913, New York, NY, USA, 2021. Association for Computing Machinery.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 100,
+      "text": "Xin Tan, Yuan Zhang, Jiadong Lu, Xin Xiong, Zhuang Liu, and Min Yang. Syzdirect: Directed greybox fuzzing for linux kernel. In Proceedings of the 2023 ACMSIGSAC Conference on Computer and Communications Security, pages 1630–1644, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 101,
+      "text": "Remix Team. Remixethereum ide and community. https:// remix-project.org/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 102,
+      "text": "Theodoros Theodoridis, Tobias Grosser, and Zhendong Su. Understanding and exploiting optimal function inlining. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, pages 977–989, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 103,
+      "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. Finding missed optimizations through the lens of dead code elimination. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, pages 697–709, 2022.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 104,
+      "text": "Schaeffer Thibaut, Macesic Darko, Jacob Eberhardt, Stefan Tai, and Kuhnert Dennis. Zokratesa toolbox for zksnarks on ethereum. https://zokrates.github.io/, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 105,
+      "text": "Schaeffer Thibaut, Macesic Darko, Jacob Eberhardt, Stefan Tai, and Kuhnert Dennis. Zokratesgithub. https://github.com/Zokrates/ ZoKrates, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 106,
+      "text": "Ken Thompson. Reflections on trusting trust. Communications of the ACM, 27(8):761–763, 1984.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 107,
+      "text": "Michael Walfish and Andrew J Blumberg. Verifying computations without reexecuting them. Communications of the ACM, 58(2):74–84, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 108,
+      "text": "Hongbo Wen, Jon Stephens, Yanju Chen, Kostas Ferles, Shankara Pailoor, Kyle Charbonnet, Isil Dillig, and Yu Feng. Practical security analysis of zero-knowledge proof circuits. Cryptology ePrint Archive, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 109,
+      "text": "Hongwei Xi and Frank Pfenning. Dependent types in practical programming. In Proceedings of the 26th ACMSIGPLAN-SIGACT symposium on Principles of programming languages, pages 214–227, 1999.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 110,
+      "text": "Dongwei Xiao, Zhibo Liu, Yiteng Peng, and Shuai Wang. Research artifacts for mtzk. https://sites.google.com/view/mtzk, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 111,
+      "text": "Dongwei Xiao, Zhibo Liu, and Shuai Wang. Metamorphic shader fusion for testing graphics shader compilers. In 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE), pages 2400–2412, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 112,
+      "text": "Dongwei Xiao, Zhibo Liu, and Shuai Wang. Phyfu: Fuzzing modern physics simulation engines. In 2023 38th IEEE/ACM International Conference on Automated Software Engineering (ASE), pages 1579– 1591, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 113,
+      "text": "Dongwei Xiao, Zhibo Liu, Yuanyuan Yuan, Qi Pang, and Shuai Wang. Metamorphic testing of deep learning compilers. Proceedings of the ACM on Measurement and Analysis of Computing Systems, 6(1):1–28, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 114,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. Finding and understanding bugs in c compilers. In Proceedings of the 32nd ACMSIGPLAN conference on Programming language design and implementation, pages 283–294, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 115,
+      "text": "Michał Zalewski. American Fuzzy Lop (AFL). https://lcamtuf. coredump.cx/afl/, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 116,
+      "text": "Shahed Zaman, Bram Adams, and Ahmed E Hassan. A qualitative study on performance bugs. In 2012 9th IEEE working conference on mining software repositories (MSR), pages 199–208. IEEE, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 117,
+      "text": "zcash. Zcash: Privacy-protecting digital currency. https://z.cash/, 2023. TABLE XIV: Bug comparison with existing works. Existing Work #Compiler Bugs #Overlapping with MTZKCODA [75] 0 NAZKAP [108] 0 NA QED2 [83] 0 NA TABLE XV: Invalid program causes and their corresponding error message patterns. Invalid Reason Error Message Pattern Divide-by-zero “divisor of zero” Modulo-by-zero “remainder of zer",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 60,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. Detecting transactional bugs in database engines via {Graph-Based }oracle construction. In 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI 23), pages 397–417, 2023.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 89,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140–1152, 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 90,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), pages 667–682, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 103,
+      "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. Finding missed optimizations through the lens of dead code elimination. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, pages 697–709, 2022.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 89]",
+      "technique": "norec",
+      "sentence": "MTZK shares a similar testing-based approach with quality assurance tools for critical systems, such as CPUs [ 56], [46], databases [ 89], [59], [60], [90], and operating systems [ 65], [16], to detect errors instead of proving their absence.",
+      "context_before": "the compilation and optimization routines they employ. The black-box view enables us to test any ZK compiler without huge engineering efforts to adapt our tool to the compiler’s internal structure. Such a setting is also aligned with many existing testing tools for software compilers [ 114], [98]. We leave the exploration of white-box testing (e.g., relying on certain coverage criteria in the compiler codebase) for future work. Testing vs. Verification. To the best of our knowledge, MTZK is the first testing approach toward ZK compilers, addressing a timely demand of our community and industry.",
+      "context_after": "MTZK cannot offer a formal guarantee: when MTZK has no findings, we cannot conclude that a ZK compiler is free from bugs. Nevertheless, the testing approach delivered by MTZK is precise and speedy, with almost no false positives. More importantly, as a testing tool, MTZK provides defect-triggering inputs, which enable developers to debug and fix the uncovered bugs (see Section VII for details). In contrast, we notice emerging research that aims to formally verify the correctness of ZK compilers [ 41], [82], [28] (details in Section IX). As a common and textbook- level concept, testing and verifi",
+      "section": "C Usage Scenario and Clarification",
+      "page": 5,
+      "char_offset": 30215,
+      "cited_reference": {
+        "number": 89,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140–1152, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker",
+        "citation_marker_project_authored"
+      ],
+      "techniques": [
+        "norec",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[103]",
+      "technique": null,
+      "sentence": "Besides discovering correctness issues, testing is also applied to identify performance issues in compilers [ 76], [102], [103].",
+      "context_before": "ranslating programs written in high-level language to lowlevel executables. Traditional compilers like GCC, LLVM have been extensively tested, with thousands of bugs uncovered to date [ 114], [73], [98], [69], [67], [77], [78]. Just-in-time (JIT) compilers such as JVM and JavaScript engines have also been the focus of the research community [ 58], [12], [72], [50]. Other compilers like shader compilers, deep learning compilers, and Multi-Party Computation (MPC) compilers are fruitfully tested using metamorphic testing or differential testing as well [ 32], [113], [111], [70], [79], [92], [74].",
+      "context_after": "Our work isthe first to systematically test ZK compilers using metamorphic testing, with a focus on uncovering logic bugs. Quality Assurance for ZK. Our work focuses on testing the correctness of ZK compilers. Another line of work focuses on finding bugs in ZK application programs. Circomspect [ 30] is a static analyzer for linting common buggy patterns, such as unused variables, in Circom programs. Picus [ 87], Ecne [ 35], QED2[83], and Soureshjani et al.’s work [ 95] aim at discovering under-constrained circuits in ZKP programs. Wen et al. [108] study common vulnerability patterns in ZKP circ",
+      "section": "D RQ4: Security Implications of Bugs",
+      "page": 15,
+      "char_offset": 85974,
+      "cited_reference": {
+        "number": 103,
+        "text": "Theodoros Theodoridis, Manuel Rigger, and Zhendong Su. Finding missed optimizations through the lens of dead code elimination. In Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, pages 697–709, 2022.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:02:36Z",
+    "is_model_written": true,
+    "summary": "MTZK tests zero-knowledge compilers, which translate a high-level program into an arithmetic circuit. A miscompilation there is a soundness or completeness flaw with security consequences, since a wrong circuit can accept a false proof. MTZK applies metamorphic testing to find such bugs.",
+    "narrative": "Database testing is cited as one of the domains whose testing-based quality assurance MTZK's approach resembles -- finding errors rather than proving their absence -- alongside CPUs and operating systems. Both mentions are reachable only through citation markers; SQLancer is never named, and nothing of it is used or measured against.",
+    "roles": {
+      "M1": "background",
+      "M2": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_14722_ndss_2026_240198.json b/_data/papers/paper_doi_10_14722_ndss_2026_240198.json
new file mode 100644
index 0000000..58ec276
--- /dev/null
+++ b/_data/papers/paper_doi_10_14722_ndss_2026_240198.json
@@ -0,0 +1,988 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:29:09Z",
+  "paper": {
+    "id": "paper:doi:10.14722/ndss.2026.240198",
+    "title": "Efficiently Detecting DBMS Bugs through Bottom-up Syntax-based SQL Generation",
+    "authors": [
+      "Yu Liang",
+      "Peng Liu"
+    ],
+    "year": 2026,
+    "venue": "Network and Distributed System Security Symposium",
+    "doi": "10.14722/ndss.2026.240198",
+    "arxiv_id": null,
+    "s2_paper_id": "dcc3bf3a8219e17b5990057d6185f7007da89f88",
+    "url": "https://doi.org/10.14722/ndss.2026.240198",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.14722/ndss.2026.240198",
+      "retrieved_at": "2026-09-09T01:29:09Z",
+      "chars": 89669,
+      "content_sha256": "sha256:7e38b2d5b09bbf0853746da7fa660d16ba97eaa17db1c30d613c31373ebda485"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 16,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“PostgreSQL Clients,” https://www.postgresql.org/about/, 2025, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“MySQL Customers,” https://www.mysql.com/customers/, 2025, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“CockroachDB Customers,” https://www.cockroachlabs.com/customers/, 2025, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "“Awesome DuckDB,” https://github.com/davidgasquez/awesome-duckdb, 2025, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "“Starbucks: A Case Study on the Impact of ERP Systems,” https: //pangrow.com/blog/business-automation/starbucks-erp-case-study/, 2024, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Y. Yusuf, A. Gunasekaran, and M. S. Abthorpe, “Enterprise Information Systems Project Implementation: A Case Study of ERP in Rolls-Royce,” International journal of production economics, vol. 87, no. 3, pp. 251– 266, 2004.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "M. Ali and L. Miller, “ERP System Implementation in Large Enterprises– A Systematic Literature Review,” Journal of enterprise information management, vol. 30, no. 4, pp. 666–692, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "S. V. Grabski, S. A. Leech, and P. J. Schmidt, “A Review of ERP Research: A Future Agenda for Accounting Information Systems,” Journal of information systems, vol. 25, no. 1, pp. 37–78, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "I.-Y. Song and K.-Y. Whang, “Database Design for Real-world Ecommerce Systems,” IEEE Data Eng. Bull., vol. 23, no. 1, pp. 23–28, 2000.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "S. DeFazio, R. Krishnan, J. Srinivasan, and S. Zeldin, “The Importance of Extensible Database Systems for E-commerce,” in Proceedings 17th International Conference on Data Engineering. IEEE, 2001, pp. 63–70.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "F. Y. Ahmed, R. Sreejith, and M. I. Abdullah, “Enhancement of ECommerce Database System During the COVID-19 Pandemic,” in Electronics (ISCAIE). IEEE, 2021, pp. 174–179.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "L. Lim and M. Wang, “Managing E-commerce Catalogs in a DBMS with Native XML Support,” in IEEE International Conference on e-Business Engineering (ICEBE’05). IEEE, 2005, pp. 564–571.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "H. Wang, “Online banking: A case study for dynamic database-driven client/server system,” Ph.D. dissertation, Concordia University, 2000.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "P. Rai, A. Singh, T. Sharma, and N. Sreenarayanan, “Banking Management System—The Web Application Way,” in Applications of Computational Methods in Manufacturing and Product Design: Select Proceedings of IPDIMS 2020. Springer, 2022, pp. 581–590.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "A. Ayman, M. Sherif et al., “Review of Online Banking Systems: Innovations and Challenges,” Advanced Sciences and Technology Journal, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "H. Upadhyay, R. Wadhawan, and D. Gupta, “Integrating Multiple Security Features in and Online Banking Platform,” 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting Logical Bugs of DBMS with Coverage-based Guidance,” in 31st USENIX Security Symposium (USENIX Security 22), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "J. Liang, Z. Wu, J. Fu, M. Wang, C. Sun, and Y. Jiang, “Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation,” in Proceedings of the IEEE/ACM 46th International Conference on Software Engineering, 2024, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Y. Fu, Z. Wu, Y. Zhang, J. Liang, J. Fu, Y. Jiang, S. Li, and X. Liao, “THANOS: DBMS Bug Detection via Storage Engine Rotation Based Differential Testing,” in on Software Engineering (ICSE). IEEE Computer Society, 2024, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "M. Rigger, “Bugs Found in Database Management Systems,” https: //www.manuelrigger.at/dbms-bugs/, (visited in April 2025).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "“The Knight Capital Disaster-Speculative Branches,” https://specbranch. com/posts/knight-capital/, 2012, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Binder, Matt, “Apple Explains Why Deleted Photos were being Restored on some iPhones,” https://mashable.com/article/ apple-iphone-deleted-photos-restored-explained, 2024, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Federal Trade Commission, “Equifax Data Breach Settlement,” https: //www.ftc.gov/enforcement/refunds/equifax-data-breach-settlement, 2019, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Nick Baker, “NYSE Fined $14 Million by SEC for Series of Rule Violations,” https://www.bloomberg.com/news/articles/2018-03-06/ nyse-fined-14-million-by-sec-for-a-series-of-rule-violations, 2018, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "C. Mishra, N. Koudas, and C. Zuzarte, “Generating Targeted Queries for Database Testing,” in Proceedings of the 2008 ACMSIGMOD International Conference on Management of Data (SIGMOD), New York, NY, USA, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "J. Wang, P. Zhang, L. Zhang, H. Zhu, and X. Ye, “A Model-based Fuzzing Approach for DBMS,” in 2013 8th International Conference on Communications and Networking in China (CHINACOM). IEEE, 2013, pp. 426–431.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "E. Lo, C. Binnig, D. Kossmann, M. Tamer Özsu, and W.-K. Hon, “A Framework for Testing DBMS Features,” The VLDB Journal, vol. 19, no. 2, pp. 203–230, Apr. 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "B. Chandra, B. Chawda, B. Kar, K. V. M. Reddy, S. Shah, and S. Sudarshan, “Data Generation for Testing and Grading SQL Queries,” The VLDB Journal, vol. 24, no. 6, Aug 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "G. Li, X. Zhou, S. Li, and B. Gao, “QTune: A Query-aware Database Tuning System with Deep Reinforcement Learning,” Proceedings of the VLDB Endowment, vol. 12, no. 12, pp. 2118–2130, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger, “SQLancer,” https://github.com/sqlancer/sqlancer, (visited in April 2025).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "“SQLSmith,” https://github.com/anse1/sqlsmith, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "E. Soremekun, E. Pavese, N. Havrikov, L. Grunske, and A. Zeller, “Inputs From Hell,” IEEE Transactions on Software Engineering, vol. 48, no. 4, pp. 1138–1153, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "C. Aschermann, T. Frassetto, T. Holz, P. Jauernig, A.-R. Sadeghi, and D. Teuchert, “NAUTILUS: Fishing for Deep Bugs with Grammars,” in Proceedings of the 26th Annual Network and Distributed System Security Symposium (NDSS), San Diego, CA, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "L. Fernandez, G. Karlsson, and D. Hübinette, “A Framework for Feedback-enabled Blackbox Fuzzing Using Context-free Grammars,” 2022, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems,” in Proceedings of the 46th International Conference on Very Large Data Bases (VLDB), Tokyo, Japan, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "P. Srivastava and M. Payer, “Gramatron: Effective grammar-aware fuzzing,” in Proceedings of the 30th acm sigsoft international symposium on software testing and analysis, 2021, pp. 244–256.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Y. Liang and H. Hu, “Parser knows best: Testing dbms with coverageguided grammar-rule traversal,” arXiv preprint arXiv:2503.03893, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "J. Patra and M. Pradel, “Learning to Fuzz: Application-independent Fuzz Testing with Probabilistic, Generative Models of Input Data,” TU Darmstadt, Department of Computer Science, Tech. Rep. TUD-CS-201614664, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "R. Zhong, Y. Chen, H. Hu, H. Zhang, W. Lee, and D. Wu, “SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback,” in Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, Nov. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "“The Lemon LALR(1) Parser Generator,” https://www.sqlite.org/lemon. html, 2025, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "S. C. Johnson and R. Sethi, “Yacc: a parser generator,” UNIX Vol. II: research system, pp. 347–374, 1990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "T. Parr and K. Fisher, “Ll (*) the foundation of the antlr parser generator,” ACM Sigplan Notices, vol. 46, no. 6, pp. 425–436, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "T. J. Parr and R. W. Quong, “Antlr: A predicated-ll (k) parser generator,” Software: Practice and Experience, vol. 25, no. 7, pp. 789–810, 1995.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "“GNU Bison: Languages and Context-Free Grammars,” https://www.gnu. org/software/bison/manual/html_node/Language-and-Grammar.html, 2023, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "“ANTLR: Parser Rules,” https://github.com/antlr/antlr4/blob/master/doc/ parser-rules.md, 2023, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "“DuckDB Github Repository,” https://github.com/duckdb/duckdb.git, 2025, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "“MySQL Parser,” https://github.com/mysql/mysql-server/blob/trunk/sql/ sql_yacc.yy, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "“MariaDB Parser,” https://github.com/MariaDB/server/blob/11.3/sql/sql_ yacc.yy, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "“PostgreSQL: The Parser Stage,” https://www.postgresql.org/docs/current/ parser-stage.html, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "“CockroachDB Parser,” https://github.com/cockroachdb/cockroach/blob/ master/pkg/sql/parser/sql.y, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "“Squirrel Github Repository,” https://github.com/s3team/Squirrel, 2022, (visited in April 2025). 14",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "J. Ba and M. Rigger, “Testing Database Engines via Query Plan Guidance,” inProceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 53,
+      "text": "——, “Keep it simple: Testing databases via differential query plans,” 2, no. 3, May 2024. [Online]. Available: https://doi.org/10.1145/3654991",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 54,
+      "text": "M. Rigger and Z. Su, “Testing Database Engines via Pivoted Query Synthesis,” in Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), Virtual, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 55,
+      "text": "——, “Finding Bugs in Database Systems via Query Partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 56,
+      "text": "——, “Detecting Optimization Bugs in Database Engines via Nonoptimizing Reference Engine Construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 57,
+      "text": "M. Jibson, “SQLsmith: Randomized SQL Testing in CockroachDB,” https: //www.cockroachlabs.com/blog/sqlsmith-randomized-sql-testing/, 2019, (visited in April 2025).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "E. Pavese, E. Soremekun, N. Havrikov, L. Grunske, and A. Zeller, “Inputs From Hell: Generating Uncommon Inputs From Common Samples,” arXiv preprint arXiv:1812.07525, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "S. Deb, K. Jain, R. Van Tonder, C. Le Goues, and A. Groce, “Syntax Is All You Need: A Universal-Language Approach to Mutant Generation,” Proceedings of the ACM on Software Engineering, vol. 1, no. FSE, pp. 654–674, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "P. Godefroid, A. Kiezun, and M. Y. Levin, “Grammar-based Whitebox Fuzzing,” in Proceedings of the 29th ACMSIGPLAN conference on programming language design and implementation, 2008, pp. 206–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "S. Sargsyan, S. Kurmangaleev, M. Mehrabyan, M. Mishechkin, T. Ghukasyan, and S. Asryan, “Grammar-based Fuzzing,” in 2018 Ivannikov Memorial Workshop (IVMEM). IEEE, 2018, pp. 32–35.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Google, “Syzkaller-Kernel Fuzzer,” https://github.com/google/syzkaller, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "S. Tang, S. Liu, J. Wang, and X. Zhang, “An Empirical Study on ASTlevel Mutation-based Fuzzing Techniques for JavaScript Engines,” in Proceedings of the 14th Asia-Pacific Symposium on Internetware, 2023, pp. 216–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "J. Wang, B. Chen, L. Wei, and Y. Liu, “Superion: Grammar-aware Greybox Fuzzing,” in on Software Engineering (ICSE). IEEE, 2019, pp. 724–735.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "R. Hodován, Á. Kiss, and T. Gyimóthy, “Grammarinator: A Grammarbased Open Source Fuzzer,” in Proceedings of the 9th ACMSIGSOFT international workshop on automating TEST case design, selection, and evaluation, 2018, pp. 45–48.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Y. Chen, R. Zhong, H. Hu, H. Zhang, Y. Yang, D. Wu, and W. Lee, “One Engine to Fuzz’em All: Generic Language Processor Testing with Semantic Validation,” in Proceedings of the 42nd IEEE Symposium on Security and Privacy (Oakland), Virtual, May 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "S. Abdul Khalek and S. Khurshid, “Automated SQL Query Generation for Systematic Testing of Database Engines,” in Proceedings of the IEEE/ACM International Conference on Automated Software Engineering (ASE), 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "M. Marcozzi, W. Vanhoof, and J.-L. Hainaut, “Test Input Generation for Database Programs Using Relational Constraints,” in Proceedings of the Fifth International Workshop on Testing Database Systems (DBTest), Scottsdale, Arizona, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang, and Y. Jiang, “Industry Practice of Coverage-guided Enterprise-level DBMS Fuzzing,” in Software Engineering in Practice (ICSE-SEIP), 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "J. Liang, Y. Chen, Z. Wu, J. Fu, M. Wang, Y. Jiang, X. Huang, T. Chen, J. Wang, and J. Li, “Sequence-oriented DBMS Fuzzing,” in Proceedings of IEEE International Conference on Data Engineering (ICDE), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "J. Liang, Z. Wu, J. Fu, Y. Bai, Q. Zhang, and Y. Jiang, “WingFuzz: Implementing Continuous Fuzzing for DBMSs,” in 2024 USENIX Annual Technical Conference (USENIX ATC 24), 2024, pp. 479–492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Z.-M. Jiang, J.-J. Bai, and Z. Su, “DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation,” inProceedings of USENIX Security Symposium (USENIX Security), 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Y. Yang, Y. Chen, R. Zhong, J. Chen, and W. Lee, “Towards generic database management system fuzzing,” in 33rd USENIX Security Symposium (USENIX Security 24), 2024, pp. 901–918.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "S. Jiang, Y. Zhang, J. Li, H. Yu, L. Luo, and G. Sun, “A Survey of Network Protocol Fuzzing: Model, Techniques and Directions,” arXiv preprint arXiv:2402.17394, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "S. Park, W. Xu, I. Yun, D. Jang, and T. Kim, “Fuzzing JavaScript Engines with Aspect-preserving Mutation,” in Proceedings of the 41st IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "M. Raselimo, J. Taljaard, and B. Fischer, “Breaking Parsers: Mutationbased Generation of Programs with Guaranteed Syntax Errors,” in Proceedings of the 12th ACMSIGPLAN International Conference on Software Language Engineering, 2019, pp. 83–87. APPENDIX A. PoC Simplification Algorithm When the DBMS crashes or throws fatal error indicating memory or data corruption, SQLBull uses the PoC Simplificati",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 20,
+      "text": "M. Rigger, “Bugs Found in Database Management Systems,” https: //www.manuelrigger.at/dbms-bugs/, (visited in April 2025).",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger, “SQLancer,” https://github.com/sqlancer/sqlancer, (visited in April 2025).",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 52,
+      "text": "J. Ba and M. Rigger, “Testing Database Engines via Query Plan Guidance,” inProceedings of International Conference on Software Engineering (ICSE), 2023.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 53,
+      "text": "——, “Keep it simple: Testing databases via differential query plans,” 2, no. 3, May 2024. [Online]. Available: https://doi.org/10.1145/3654991",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "prints the DOI of the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 54,
+      "text": "M. Rigger and Z. Su, “Testing Database Engines via Pivoted Query Synthesis,” in Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), Virtual, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 55,
+      "text": "——, “Finding Bugs in Database Systems via Query Partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 56,
+      "text": "——, “Detecting Optimization Bugs in Database Engines via Nonoptimizing Reference Engine Construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 76,
+      "text": "M. Raselimo, J. Taljaard, and B. Fischer, “Breaking Parsers: Mutationbased Generation of Programs with Guaranteed Syntax Errors,” in Proceedings of the 12th ACMSIGPLAN International Conference on Software Language Engineering, 2019, pp. 83–87. APPENDIX A. PoC Simplification Algorithm When the DBMS crashes or throws fatal error indicating memory or data corruption, SQLBull uses the PoC Simplification algorithm shown in Algorithm 3 to simplify the generated SQL query sequence and save the simplified queries as bug PoC. Algorithm 3: PoC Simplification 1:function PoCSimplification( stmt _seq) 2:simpl _stmt _seq← ∅ 3:for all stmt∈stmt _seq do 4: reset_database() 5: for all stmt′∈stmt do 6: ifstmt ==stmt′then 7: continue 8: end if 9: execute( stmt′) 10: end for 11: ifnot database_crash() then 12: simpl _stmt _seq←simpl _stmt _seq+ stmt 13: end if 14: end for 15: for all stmt∈simpl _stmt _seq do 16: for all tree _node∈stmt. get_tree_nodes() do 17: reset_database() 18: for all stmt′∈simpl _stmt _seq do 19: ifstmt ==stmt′then 20: stmt′←stmt′.copy().remove( tree _node ) 21: end if 22: execute( stmt′) 23: end for 24: ifnot database_crash() then 25: stmt←stmt .remove( tree _node ) 26: end if 27: end for 28: end for 29: return simpl _stmt _seq 30: end function B. Syntax-based Testing Tools on Other Systems The technique of syntax-based testing has also been widely adopted in other systems, such as programming systems (e.g., JavaScript engines) [63] and network protocol implementations [74] . Specifically, [34] proposes a syntax-based fuzzing framework and applies the learned grammar to generate Link Layer Discovery Protocol (LLDP). DIE applies type-based AST mutation extracted from the JavaScript grammar code while preserving the context-rich parts of the input based on the intuition of the JavaScript programs [75] . NAUTILUS demonstrates that code coverage feedback is beneficial for guiding syntax-based input generation [33] . [76] proposes a syntax-based fuzzing tool that intentionally introduces slight syntactic errors to the generated inputs to expose faulty parser logic. Many of these previous works formulate the task of 15 syntax-based input generation as a Multi-arm Bandit (MAB) problem [33] , [34] , [58] . However, they all employ Top-down generation, which is not suitable for deep bug discovery in DBMS. To prioritize feature-rich SQL grammar that is more likely to expose bugs, some existing works [32] , [38] , [58] prioritize the grammar rules used to construct the previous failures. However, the previous failure inputs may not contain the deep SQL grammar we want to explore in this paper. Furthermore, not all grammar used in the previous failure is useful for triggering bugs. By blindly prioritizing all the grammar distributed in these inputs without knowing which grammar is critical for bug triggering, previous failure inputs are less efficient for preserving the deep SQL grammar to uncover new deep bugs. 16",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[17]–[20]",
+      "technique": null,
+      "sentence": "However, the high complexity of DBMSs makes them prone to bugs [17]–[20].",
+      "context_before": "diverse SQL queries for fuzzing. A prototype tool,SQLBull, adopts the Bottom-up generation technique for fuzzing. In the evaluation, SQLBullfound 63 zero-day bugs from 5 well-tested DBMSs: MySQL,MariaDB,CockroachDB,DuckDB, and PostgreSQL. It outperforms all existing tools in both bug-finding and code coverage. The evaluation results verify the effectiveness of the Bottom-up generation technique. I.INTRODUCTION Database Management Systems (DBMSs) power today’s internet applications [1]–[4], such as enterprise resource planning (ERP) [5]–[8], E-commerce [9]–[12], and online banking [ 13]– [16].",
+      "context_after": "ADBMS bug triggered in real-world applications can lead to severe consequences, such as money loss [21] or data breach [22], [23]. For example, in June 2015, the New York Stock Exchange (NYSE) froze stock trading for almost 4 hours because of a database glitch. This glitch caused the NYSE to lose at least $14 million [ 24]. Accordingly, proactively identifying DBMS bugs through automatic software testing (and other viable means) is a critical aspect of securing DBMS software development [ 25]–[29]. In the DBMS automatic testing domain, it is a well-known practice to generate SQL testing queri",
+      "section": null,
+      "page": 1,
+      "char_offset": 1892,
+      "cited_reference": {
+        "number": 20,
+        "text": "M. Rigger, “Bugs Found in Database Management Systems,” https: //www.manuelrigger.at/dbms-bugs/, (visited in April 2025).",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[30]",
+      "technique": null,
+      "sentence": ", syntax-based) [ 17], [30], [31].",
+      "context_before": "lications can lead to severe consequences, such as money loss [21] or data breach [22], [23]. For example, in June 2015, the New York Stock Exchange (NYSE) froze stock trading for almost 4 hours because of a database glitch. This glitch caused the NYSE to lose at least $14 million [ 24]. Accordingly, proactively identifying DBMS bugs through automatic software testing (and other viable means) is a critical aspect of securing DBMS software development [ 25]–[29]. In the DBMS automatic testing domain, it is a well-known practice to generate SQL testing queries by following SQLgrammar rules (i.e.",
+      "context_after": "The primary benefit is that the generated test cases have a higher chance of being syntactically correct. These test cases are more likely to pass the sanity checks conducted by the DBMS parser and subsequently trigger the execution of the internal DBMS program code. Therefore, the syntax-based testers achieve a higher code coverage and uncover more bugs than nonsyntax-based ones. Because of this benefit, the DBMS testing community has developed a series of syntax-based testing tools to examine the DBMS program code. For example, SQLancer[30] is a representative syntax-based SQL testing platf",
+      "section": null,
+      "page": 1,
+      "char_offset": 2604,
+      "cited_reference": {
+        "number": 30,
+        "text": "Manuel Rigger, “SQLancer,” https://github.com/sqlancer/sqlancer, (visited in April 2025).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer[30] is a representative syntax-based SQL testing platform that adopts hand-crafted SQL templates to generate testing SQL queries.",
+      "context_before": "queries by following SQLgrammar rules (i.e., syntax-based) [ 17], [30], [31]. The primary benefit is that the generated test cases have a higher chance of being syntactically correct. These test cases are more likely to pass the sanity checks conducted by the DBMS parser and subsequently trigger the execution of the internal DBMS program code. Therefore, the syntax-based testers achieve a higher code coverage and uncover more bugs than nonsyntax-based ones. Because of this benefit, the DBMS testing community has developed a series of syntax-based testing tools to examine the DBMS program code.",
+      "context_after": "All the existing syntax-based testing tools adopt the Topdown (TD) generation method. To generate a SQL query (syntax tree), aTDgenerator starts its grammar exploration from the rootsymbol, where the root is always the topmost node in the syntax tree. Then, it forward traverses the grammar, constructing a syntax tree with all the grammar rules it has explored, and only stops the tree construction when no more grammar rules can be used further to expand the leaf nodes of the syntax tree. A more detailed illustration of the TDgeneration method is shown in§II-B. However, the TDgenerator tends to",
+      "section": null,
+      "page": 1,
+      "char_offset": 3161,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This method is more commonly used by syntax-based testing tools that rely on handwritten SQL templates, such as SQLancer [30] and SQLsmith [31].",
+      "context_before": "mmon among theNetwork and Distributed System Security (NDSS) Symposium 202623-27 February 2026, San Diego, CA, USAISBN 979-8-9919276-8-0https://dx.doi.org/10.14722/ndss.2026.240198www.ndss-symposium.org rules in the shallow layers. They further increase the difficulty for the TDgenerator in exploring the deeper grammar space. Several existing works attempt to address the above issues [32]–[34]. For example, some TDgenerators allocate pre-defined exploration probabilities to different grammar rules, putting more weight on certain grammar rules that are known to lead to existing bugs [27], [35].",
+      "context_after": "Another method is to rewrite the grammar rules by uplifting some feature-rich grammar into the shallower layers for more frequent exploration [36]. However, both methods are manual effort-intensive to implement and are not scalable to the large number of grammar rules defined by the DBMSs. They demand the testers to have deep domain knowledge of the DBMS SQL grammar to identify the feature-rich SQL grammar and apply the necessary grammar annotation. Moreover, these methods are not adaptable in adjusting their weights when testing the ever-changing SQL dialects implemented by different DBMSs.",
+      "section": null,
+      "page": 2,
+      "char_offset": 5799,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For example, SQLancer uses Java to implement its SQL templates, while SQLsmith uses C++.",
+      "context_before": "hen introduce a motivating example in § II-B to demonstrate a TDgenerator’s inefficiency and a BUgenerator’s effectiveness in detecting DBMS bugs. A. Background of Syntax-based DBMS Testing To improve the quality of the generated test cases, the fuzzing community adopts the syntax-based fuzzing strategy to generate syntactically correct test cases. They can be classified into two main categories: those based on handwritten SQL templates and those based on DBMSs’ built-in grammar. The former uses hand-crafted SQL templates implemented in custom programming languages to output SQL query strings.",
+      "context_after": "The second category of tools borrows the built-in SQL grammar of a DBMS to guide the test case generation. The parser of a DBMS takes the grammar as input and transforms the input SQL query strings into syntax trees. Instead of matching the grammar rules to the provided seed (test case) queries, a syntax-based generator chooses one arbitrary grammar rule to execute for each parsed tree node, which eventually constructs a syntax tree from all the executed grammar rules. The generated syntax tree is then transformed into a syntactically correct SQL query for execution. A more detailed descripti",
+      "section": null,
+      "page": 2,
+      "char_offset": 10498,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The most advanced tool in this category is SQLancer.",
+      "context_before": "identifies which SQL grammar expands the symbols listed in the table and backtracks the identified grammar to the root to generate syntax paths. all SQL statements, to ensure the comparisons between SQLBull and the other tools are fair, we configured SQLBull to only generate SQL statements that are commonly supported by all baselines, specifically, CREATE ,INSERT ,ALTER ,DELETE ,UPDATE, and SELECT statements. We note this configuration as SQLBull M. We include existing works from three main categories. The first category of tools generates testing SQL queries based on handwritten SQL templates.",
+      "context_after": "Although SQLancer was first introduced in 2022, it has become the most popular platform for implementing the latest SQL testing techniques [52]–[56]. In this evaluation, we include two of the latest and most advanced testing techniques implemented in SQLancer which are capable of detecting DBMS memory errors, i.e., SQLancer +QPG[52] and SQLancer +DQP[53]. We compare SQLBull with SQLancer +QPG onCockroachDB, and SQLancer +DQPonMySQL and MariaDB. In addition, we include SQLsmith Cas an SQL template-based testing tool maintained by the CockroachDB developer team and evaluate it on CockroachDB [5",
+      "section": null,
+      "page": 8,
+      "char_offset": 41922,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Although SQLancer was first introduced in 2022, it has become the most popular platform for implementing the latest SQL testing techniques [52]–[56].",
+      "context_before": "ed in the table and backtracks the identified grammar to the root to generate syntax paths. all SQL statements, to ensure the comparisons between SQLBull and the other tools are fair, we configured SQLBull to only generate SQL statements that are commonly supported by all baselines, specifically, CREATE ,INSERT ,ALTER ,DELETE ,UPDATE, and SELECT statements. We note this configuration as SQLBull M. We include existing works from three main categories. The first category of tools generates testing SQL queries based on handwritten SQL templates. The most advanced tool in this category is SQLancer.",
+      "context_after": "In this evaluation, we include two of the latest and most advanced testing techniques implemented in SQLancer which are capable of detecting DBMS memory errors, i.e., SQLancer +QPG[52] and SQLancer +DQP[53]. We compare SQLBull with SQLancer +QPG onCockroachDB, and SQLancer +DQPonMySQL and MariaDB. In addition, we include SQLsmith Cas an SQL template-based testing tool maintained by the CockroachDB developer team and evaluate it on CockroachDB [57]. The second category of testing baselines is syntax-based fuzzing tools that directly learn the SQL grammar from the DBMS grammar code. We include",
+      "section": null,
+      "page": 8,
+      "char_offset": 41975,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In this evaluation, we include two of the latest and most advanced testing techniques implemented in SQLancer which are capable of detecting DBMS memory errors, i.",
+      "context_before": "ull and the other tools are fair, we configured SQLBull to only generate SQL statements that are commonly supported by all baselines, specifically, CREATE ,INSERT ,ALTER ,DELETE ,UPDATE, and SELECT statements. We note this configuration as SQLBull M. We include existing works from three main categories. The first category of tools generates testing SQL queries based on handwritten SQL templates. The most advanced tool in this category is SQLancer. Although SQLancer was first introduced in 2022, it has become the most popular platform for implementing the latest SQL testing techniques [52]–[56].",
+      "context_after": "e., SQLancer +QPG[52] and SQLancer +DQP[53]. We compare SQLBull with SQLancer +QPG onCockroachDB, and SQLancer +DQPonMySQL and MariaDB. In addition, we include SQLsmith Cas an SQL template-based testing tool maintained by the CockroachDB developer team and evaluate it on CockroachDB [57]. The second category of testing baselines is syntax-based fuzzing tools that directly learn the SQL grammar from the DBMS grammar code. We include Squirrel in this category, which has been the default choice by different DBMS vendors to fuzz their products [39]. Furthermore, we implemented a TDsyntax-based gen",
+      "section": null,
+      "page": 8,
+      "char_offset": 42125,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", SQLancer +QPG[52] and SQLancer +DQP[53].",
+      "context_before": "TER ,DELETE ,UPDATE, and SELECT statements. We note this configuration as SQLBull M. We include existing works from three main categories. The first category of tools generates testing SQL queries based on handwritten SQL templates. The most advanced tool in this category is SQLancer. Although SQLancer was first introduced in 2022, it has become the most popular platform for implementing the latest SQL testing techniques [52]–[56]. In this evaluation, we include two of the latest and most advanced testing techniques implemented in SQLancer which are capable of detecting DBMS memory errors, i.e.",
+      "context_after": "We compare SQLBull with SQLancer +QPG onCockroachDB, and SQLancer +DQPonMySQL and MariaDB. In addition, we include SQLsmith Cas an SQL template-based testing tool maintained by the CockroachDB developer team and evaluate it on CockroachDB [57]. The second category of testing baselines is syntax-based fuzzing tools that directly learn the SQL grammar from the DBMS grammar code. We include Squirrel in this category, which has been the default choice by different DBMS vendors to fuzz their products [39]. Furthermore, we implemented a TDsyntax-based generation tool that utilizes the ϵ-greedy meth",
+      "section": null,
+      "page": 8,
+      "char_offset": 42291,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg",
+        "dqp"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We compare SQLBull with SQLancer +QPG onCockroachDB, and SQLancer +DQPonMySQL and MariaDB.",
+      "context_before": "We note this configuration as SQLBull M. We include existing works from three main categories. The first category of tools generates testing SQL queries based on handwritten SQL templates. The most advanced tool in this category is SQLancer. Although SQLancer was first introduced in 2022, it has become the most popular platform for implementing the latest SQL testing techniques [52]–[56]. In this evaluation, we include two of the latest and most advanced testing techniques implemented in SQLancer which are capable of detecting DBMS memory errors, i.e., SQLancer +QPG[52] and SQLancer +DQP[53].",
+      "context_after": "In addition, we include SQLsmith Cas an SQL template-based testing tool maintained by the CockroachDB developer team and evaluate it on CockroachDB [57]. The second category of testing baselines is syntax-based fuzzing tools that directly learn the SQL grammar from the DBMS grammar code. We include Squirrel in this category, which has been the default choice by different DBMS vendors to fuzz their products [39]. Furthermore, we implemented a TDsyntax-based generation tool that utilizes the ϵ-greedy methodology to decide which grammar to explore from each symbol handling. This tool formulates",
+      "section": null,
+      "page": 8,
+      "char_offset": 42333,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For completeness, we also include the statistics of these SQL features from the queries generated by other testing tools, including mutation-based fuzzer Squirrel [39] and templatebased SQL generator SQLancer [30].",
+      "context_before": "selected based on the following criteria: 1) They are implemented on top of the SQL expression grammar rules, which matches the configuration we provided to SQLBull for the BUgrammar exploration; 2) They are feature-rich and are commonly seen in the bug-triggering PoCs. We do not claim the selected SQL features represent all the interesting features that could potentially trigger bugs. However, the statistics of these selected SQL features indicate the effectiveness of SQLBull BUgrammar exploration, especially when compared with the results from TDgeneration under similar experimental settings.",
+      "context_after": "These statistics infer the grammar space exploration priorities for these testing tools. All evaluations were conducted on a machine with Intel(R) Core(TM) i5-9600KF 6 cores CPU and 48GB of RAM. Each tool was run on a single CPU core and evaluated for 24 hours. We repeated each evaluation 10 times and report only the average result. We chose the latest version of the DBMSs available at the time of the evaluation, i.e., the evaluations were run upon MySQL on version 8.0.40 ,MariaDB on version 11.7.2 ,CockroachDB on version v24.3.1 ,DuckDB on version v1.1.3, and PostgreSQL on version REL_17_4.",
+      "section": null,
+      "page": 8,
+      "char_offset": 46582,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer +QPG and SQLancer +DQP are the latest and most advanced DBMS testing techniques implemented on top of SQLancer.",
+      "context_before": "is worth noting thatSquirrel could perform better if we provide an even more comprehensive testing seed that covers more SQL grammar. However, we observed that gathering a comprehensive testing corpus is also challenging and requires a lot of manual work. In comparison, SQLBull does not require any input seed. The DBMS user only needs to configure the SQL features to be tested, and SQLBull will automatically explore the SQL features with diverse generated queries. Overall, SQLBull and SQLBull Mcan find more bugs than Squirrel by more thoroughly exploring the feature-rich grammar of the DBMSs.",
+      "context_after": "Both tools utilize the same SQL templates from SQLancer to generate SQL queries but leverage different testing strategies to explore the grammar space of the DBMSs. Interestingly, both techniques achieve a near-perfect query correctness rate close to 99.9%. Especially for SQLancer +DQP, it is designed to execute the same valid SQL query multiple times with different optimization settings to understand the impact of the query optimization on the query execution plan, which leads to a near-perfect query correctness rate. However, the handwritten SQL templates limit the flexibility of both tools",
+      "section": null,
+      "page": 9,
+      "char_offset": 52333,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Both tools utilize the same SQL templates from SQLancer to generate SQL queries but leverage different testing strategies to explore the grammar space of the DBMSs.",
+      "context_before": "SQL grammar. However, we observed that gathering a comprehensive testing corpus is also challenging and requires a lot of manual work. In comparison, SQLBull does not require any input seed. The DBMS user only needs to configure the SQL features to be tested, and SQLBull will automatically explore the SQL features with diverse generated queries. Overall, SQLBull and SQLBull Mcan find more bugs than Squirrel by more thoroughly exploring the feature-rich grammar of the DBMSs. SQLancer +QPG and SQLancer +DQP are the latest and most advanced DBMS testing techniques implemented on top of SQLancer.",
+      "context_after": "Interestingly, both techniques achieve a near-perfect query correctness rate close to 99.9%. Especially for SQLancer +DQP, it is designed to execute the same valid SQL query multiple times with different optimization settings to understand the impact of the query optimization on the query execution plan, which leads to a near-perfect query correctness rate. However, the handwritten SQL templates limit the flexibility of both tools in exploring the vast grammar space of the DBMSs. Although they achieve higher query correctness rates than SQLBull in Figure 8c, Figure 8f, and Figure 8i, they fal",
+      "section": null,
+      "page": 9,
+      "char_offset": 52454,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Especially for SQLancer +DQP, it is designed to execute the same valid SQL query multiple times with different optimization settings to understand the impact of the query optimization on the query execution plan, which leads to a near-perfect query correctness rate.",
+      "context_before": "d, and SQLBull will automatically explore the SQL features with diverse generated queries. Overall, SQLBull and SQLBull Mcan find more bugs than Squirrel by more thoroughly exploring the feature-rich grammar of the DBMSs. SQLancer +QPG and SQLancer +DQP are the latest and most advanced DBMS testing techniques implemented on top of SQLancer. Both tools utilize the same SQL templates from SQLancer to generate SQL queries but leverage different testing strategies to explore the grammar space of the DBMSs. Interestingly, both techniques achieve a near-perfect query correctness rate close to 99.9%.",
+      "context_after": "However, the handwritten SQL templates limit the flexibility of both tools in exploring the vast grammar space of the DBMSs. Although they achieve higher query correctness rates than SQLBull in Figure 8c, Figure 8f, and Figure 8i, they fall behind by achieving a lower code coverage in Figure 8b, Figure 8e, and Figure 8h because the generated queries are less diverse. Both techniques fail to detect as many bugs as SQLBull and SQLBull Min our repeated 24-hour experiments. SQLsmith Cis yet another handwritten SQL template-based fuzzing tool, mimicking the original SQL syntax implemented in Cockr",
+      "section": null,
+      "page": 9,
+      "char_offset": 52712,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "4% SQLancer 3.",
+      "context_before": "00%) 2 (66.6%) Total 31 (49.2%) 13 (20.6%) 17 (30.0%) 12 (19.0%) 44 (69.8%) TABLE III: Statistics of SQL features under survey upon all the zero-day bugs detected by SQLBull. DBMS ToolsSQL Features Under Survey SubqueryGROUP BY, ORDER BY, HAVING, WINDOW, OVERJOINConditional SQL Expressions MySQLSQLBull 69.5% 51.1% 21.8% 34.1% SQLBull M 69.8% 57.7% 18.0% 64.1% SQLBull TD 41.3% 13.2% 1.3% 0.38% SQLBull MAB 47.4% 42.8% 18.5% 3.9% SQLBull NE 46.0% 40.8% 4.57% 12.6% Squirrel 19.3% 19.5% 13.1% 5.6% AFL 0.00% 0.00% 0.00% 0.00% CockroachDBSQLBull 46.9% 33.0% 39.2% 39.0% SQLBull M 55.6% 57.1% 26.5% 33.",
+      "context_after": "6% 18.9% 5.3% 7.2% SQLsmith 12.8% 62.6% 9.8% 3.6% TABLE IV: Statistics of SQL features under survey upon different tools. The statistics are calculated based on all the SQL queries generated from each testing tool running for 24 hours. We provide 0% statistics for AFLto demonstrate that it does not effectively explore these SQL features by mutation, but simply repeats the features presented in the seeds. that SQLBull MABtends to generate correct but simple queries, such as “(((TABLE t0)))”, where the repeated parenthesized expression is a recursive expression to hold the “ TABLE ” subexpressio",
+      "section": null,
+      "page": 12,
+      "char_offset": 62957,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[30]",
+      "technique": null,
+      "sentence": "We first discuss the existing syntax-based testing tools for the DBMS testing community [30], [39], [52] in § VI-A and § VI-B.",
+      "context_before": "th JOIN and Conditional SQL Expressions, limiting their bugfinding effectiveness. SQLBull and SQLBull M, instead, can more thoroughly explore all the SQL features, so they can detect more bugs. SQLBull NEexposes fewer SQL features than SQLBull, because it is not configured to target the SQL expression grammar rules, on which all SQL features under survey are defined. AFLhardly generates any valid SQL queries, and thus does not explore any of the SQL features under survey. VI. RELATED WORKS Syntax-based automatic testing has been widely adopted to find bugs in various software systems [59]–[62].",
+      "context_after": "We extend the scope of syntax-based testing tools to other securityoriented testing domains [63]–[66] in Appendix B. A. SQL Template-based DBMS Testing Tools Certain existing DBMS testing tools rely on handwritten SQL templates to generate the SQL queries. These templates are implemented with custom programming languages, such as C/C++ [31], [35], Java [27], [30], and languages designed for SAT solvers [67], [68]. The most well-known template-based query testing framework is SQLsmith [31]. The developers of CockroachDB recently ported SQLsmith to support their own DBMS [57], which is referred",
+      "section": null,
+      "page": 13,
+      "char_offset": 68797,
+      "cited_reference": {
+        "number": 30,
+        "text": "Manuel Rigger, “SQLancer,” https://github.com/sqlancer/sqlancer, (visited in April 2025).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[30]",
+      "technique": null,
+      "sentence": "These templates are implemented with custom programming languages, such as C/C++ [31], [35], Java [27], [30], and languages designed for SAT solvers [67], [68].",
+      "context_before": "ates any valid SQL queries, and thus does not explore any of the SQL features under survey. VI. RELATED WORKS Syntax-based automatic testing has been widely adopted to find bugs in various software systems [59]–[62]. We first discuss the existing syntax-based testing tools for the DBMS testing community [30], [39], [52] in § VI-A and § VI-B. We extend the scope of syntax-based testing tools to other securityoriented testing domains [63]–[66] in Appendix B. A. SQL Template-based DBMS Testing Tools Certain existing DBMS testing tools rely on handwritten SQL templates to generate the SQL queries.",
+      "context_after": "The most well-known template-based query testing framework is SQLsmith [31]. The developers of CockroachDB recently ported SQLsmith to support their own DBMS [57], which is referred to as SQLsmith Cin this paper. SQLancer [30] is another template-based DBMS testing tool that detects DBMS memory and logic bugs [54]–[56]. Its latest improvement, SQLancer +QPG[52] and SQLancer +DQP[53], leverage the DBMS query plan and optimization hints to guide the query generation. All of the SQL template-based DBMS testing tools suffer from the same limitation: their hand-crafted SQL templates are limited in",
+      "section": null,
+      "page": 13,
+      "char_offset": 69181,
+      "cited_reference": {
+        "number": 30,
+        "text": "Manuel Rigger, “SQLancer,” https://github.com/sqlancer/sqlancer, (visited in April 2025).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [30] is another template-based DBMS testing tool that detects DBMS memory and logic bugs [54]–[56].",
+      "context_before": "based testing tools to other securityoriented testing domains [63]–[66] in Appendix B. A. SQL Template-based DBMS Testing Tools Certain existing DBMS testing tools rely on handwritten SQL templates to generate the SQL queries. These templates are implemented with custom programming languages, such as C/C++ [31], [35], Java [27], [30], and languages designed for SAT solvers [67], [68]. The most well-known template-based query testing framework is SQLsmith [31]. The developers of CockroachDB recently ported SQLsmith to support their own DBMS [57], which is referred to as SQLsmith Cin this paper.",
+      "context_after": "Its latest improvement, SQLancer +QPG[52] and SQLancer +DQP[53], leverage the DBMS query plan and optimization hints to guide the query generation. All of the SQL template-based DBMS testing tools suffer from the same limitation: their hand-crafted SQL templates are limited in covering the complete set of SQL grammar implemented by the DBMSs. B. Grammar Code-based DBMS Testing Tools Several DBMS testing frameworks utilize the DBMS builtin grammar code to generate syntactically valid SQL queries. Squirrel is one of the state-of-the-art syntax-based DBMS fuzzing tools in this category [39]. Ado",
+      "section": null,
+      "page": 13,
+      "char_offset": 69555,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Its latest improvement, SQLancer +QPG[52] and SQLancer +DQP[53], leverage the DBMS query plan and optimization hints to guide the query generation.",
+      "context_before": "DBMS Testing Tools Certain existing DBMS testing tools rely on handwritten SQL templates to generate the SQL queries. These templates are implemented with custom programming languages, such as C/C++ [31], [35], Java [27], [30], and languages designed for SAT solvers [67], [68]. The most well-known template-based query testing framework is SQLsmith [31]. The developers of CockroachDB recently ported SQLsmith to support their own DBMS [57], which is referred to as SQLsmith Cin this paper. SQLancer [30] is another template-based DBMS testing tool that detects DBMS memory and logic bugs [54]–[56].",
+      "context_after": "All of the SQL template-based DBMS testing tools suffer from the same limitation: their hand-crafted SQL templates are limited in covering the complete set of SQL grammar implemented by the DBMSs. B. Grammar Code-based DBMS Testing Tools Several DBMS testing frameworks utilize the DBMS builtin grammar code to generate syntactically valid SQL queries. Squirrel is one of the state-of-the-art syntax-based DBMS fuzzing tools in this category [39]. Adopting the Top-down generation methodology, Squirrel parses the DBMS grammar into its internal representation (IR) and employs type-based IR mutation",
+      "section": null,
+      "page": 13,
+      "char_offset": 69664,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg",
+        "dqp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "uses_infrastructure": [
+        "M8"
+      ],
+      "compares_with": [
+        "M10"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:48:43Z",
+    "is_model_written": true,
+    "summary": "SQLBull generates SQL bottom-up from a DBMS's own grammar rather than from hand-written templates, on the argument that templates only cover the features whoever wrote them thought to include. Building queries upwards from grammar productions reaches syntax the template-based generators never emit, which is where a class of memory errors lives. It is evaluated on CockroachDB, MySQL and MariaDB.",
+    "narrative": "SQLancer is the paper's reference point for the template-based approach it is arguing against, and it is unusually direct about SQLancer's standing: the most advanced tool in the category, and the most popular platform for implementing the latest SQL testing techniques. Its baselines are two of those techniques rather than SQLancer's own oracles -- SQLancer+QPG on CockroachDB and SQLancer+DQP on MySQL and MariaDB -- and the paper notes both are built on top of SQLancer and use its SQL templates. The critique is that those hand-written Java templates bound which SQL features can appear at all, which is the limitation the grammar-based generation is meant to remove.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "definition",
+      "M4": "motivation",
+      "M5": "motivation",
+      "M6": "state_of_the_art",
+      "M7": "state_of_the_art",
+      "M8": "baseline",
+      "M9": "baseline",
+      "M10": "baseline",
+      "M11": "result_comparison",
+      "M12": "state_of_the_art",
+      "M13": "definition",
+      "M14": "definition",
+      "M15": "result_comparison",
+      "M16": "background",
+      "M17": "background",
+      "M18": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer+QPG and SQLancer+DQP are run as baselines. M12 and M13 describe those tools as built on SQLancer and using its templates, but that is a property of the baselines, not of SQLBull, whose generator works from the DBMS grammar instead."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLBull replaces template-based generation with bottom-up grammar-based generation. No SQLancer oracle is carried forward or generalised."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M8",
+          "M9",
+          "M10",
+          "M11"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M8",
+            "sentence": "In this evaluation, we include two of the latest and most advanced testing techniques implemented in SQLancer which are capable of detecting DBMS memory errors, i.",
+            "section": null,
+            "page": 8
+          },
+          {
+            "mention_id": "M9",
+            "sentence": ", SQLancer +QPG[52] and SQLancer +DQP[53].",
+            "section": null,
+            "page": 8
+          },
+          {
+            "mention_id": "M10",
+            "sentence": "We compare SQLBull with SQLancer +QPG onCockroachDB, and SQLancer +DQPonMySQL and MariaDB.",
+            "section": null,
+            "page": 8
+          },
+          {
+            "mention_id": "M11",
+            "sentence": "For completeness, we also include the statistics of these SQL features from the queries generated by other testing tools, including mutation-based fuzzer Squirrel [39] and templatebased SQL generator SQLancer [30].",
+            "section": null,
+            "page": 8
+          }
+        ],
+        "reasoning": "M8 and M9 name SQLancer+QPG and SQLancer+DQP as the techniques included in the evaluation, M10 records which was run against which system, and M11 reports SQLancer's generated queries in the feature statistics.",
+        "techniques": [
+          "qpg",
+          "dqp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M6",
+          "M7",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M6",
+            "sentence": "The most advanced tool in this category is SQLancer.",
+            "section": null,
+            "page": 8
+          },
+          {
+            "mention_id": "M7",
+            "sentence": "Although SQLancer was first introduced in 2022, it has become the most popular platform for implementing the latest SQL testing techniques [52]–[56].",
+            "section": null,
+            "page": 8
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "SQLancer +QPG and SQLancer +DQP are the latest and most advanced DBMS testing techniques implemented on top of SQLancer.",
+            "section": null,
+            "page": 9
+          }
+        ],
+        "reasoning": "M6 calls SQLancer the most advanced tool in its category, M7 the most popular platform for implementing the latest SQL testing techniques, and M12 describes QPG and DQP as the latest and most advanced DBMS testing techniques, implemented on top of it."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_14778_3712221_3712247.json b/_data/papers/paper_doi_10_14778_3712221_3712247.json
new file mode 100644
index 0000000..0eccaa1
--- /dev/null
+++ b/_data/papers/paper_doi_10_14778_3712221_3712247.json
@@ -0,0 +1,1410 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:10:02Z",
+  "paper": {
+    "id": "paper:doi:10.14778/3712221.3712247",
+    "title": "Semantic Conformance Testing of Relational DBMS",
+    "authors": [
+      "Shuang Liu",
+      "Chenglin Tian",
+      "Jun Sun",
+      "Ruifeng Wang",
+      "Wei Lu",
+      "Yongxin Zhao",
+      "Yinxing Xue",
+      "Junjie Wang",
+      "Xiaoyong Du"
+    ],
+    "year": 2024,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3712221.3712247",
+    "arxiv_id": "2406.09469",
+    "s2_paper_id": "bb1da01b4752d31f2f96f3abcc5502b58bedbc4e",
+    "url": "https://doi.org/10.14778/3712221.3712247",
+    "also_indexed_as": [
+      "paper:arxiv:2406.09469"
+    ]
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2406.09469",
+      "retrieved_at": "2026-09-10T15:10:02Z",
+      "chars": 97306,
+      "content_sha256": "sha256:27ebd6cab84ea1e46f5013d5a504231dc3986cfdf151d8e4edec4b96c231824d"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 21,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "I",
+        "title": "INTRODUCTION",
+        "start": 2287
+      },
+      {
+        "number": "II",
+        "title": "PRELIMINARY",
+        "start": 9638
+      },
+      {
+        "number": "A",
+        "title": "Prolog",
+        "start": 9654
+      },
+      {
+        "number": "III",
+        "title": "FORMAL SEMANTICS OF SQL",
+        "start": 18899
+      },
+      {
+        "number": "A",
+        "title": "Basic symbol definition",
+        "start": 27291
+      },
+      {
+        "number": "B",
+        "title": "Semantics of Expressions",
+        "start": 28352
+      },
+      {
+        "number": "C",
+        "title": "Semantics of SQL keywords",
+        "start": 30947
+      },
+      {
+        "number": "IV",
+        "title": "CONFORMANCE TESTING",
+        "start": 32971
+      },
+      {
+        "number": "A",
+        "title": "Overview of our approach",
+        "start": 32995
+      },
+      {
+        "number": "B",
+        "title": "Coverage guided test case generation",
+        "start": 34590
+      },
+      {
+        "number": "C",
+        "title": "Prolog implementation of SQL formal semantics",
+        "start": 44139
+      },
+      {
+        "number": "D",
+        "title": "Result comparison",
+        "start": 50393
+      },
+      {
+        "number": "E",
+        "title": "Discussion on Extensibility",
+        "start": 51954
+      },
+      {
+        "number": "V",
+        "title": "EVALUATION",
+        "start": 54045
+      },
+      {
+        "number": "A",
+        "title": "Experiment setup",
+        "start": 54602
+      },
+      {
+        "number": "B",
+        "title": "Experiment results",
+        "start": 58389
+      },
+      {
+        "number": "C",
+        "title": "Threats to Validity",
+        "start": 79583
+      },
+      {
+        "number": "VI",
+        "title": "RELATED WORK",
+        "start": 82290
+      },
+      {
+        "number": "VII",
+        "title": "CONCLUSION",
+        "start": 87912
+      },
+      {
+        "number": "J",
+        "title": "Wang, and X. Du, “Conformance testing of relational dbms against",
+        "start": 93944
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "“Amazon,” https://www.amazon.com/, 1995, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "“ebay,” https://www.ebay.com/, 1995, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "“Mysql,” https://www.mysql.com, 1995, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "“Booking,” https://www.booking.com/, 1996, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "“Postgresql,” https://www.postgresql.org, 1996, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "“Sqlite,” https://www.sqlite.org/index.html, 2000, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "“Facebook,” https://www.facebook.com/, 2004, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "“Jsqlparser,” https://github.com/JSQLParser/JSqlParser, 2011, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "“International organization for standardization. (2016). iso/iec 90752:2016: Information technology – database languages – sql/foundation,” 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "“Iso/iec 9075-2:2016,” https://www.iso.org/standard/63555.html, 2016, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "“Oceanbase,” https://en.oceanbase.com/, 2016, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "“Tidb,” https://www.pingcap.com/tidb/, 2016, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "“Duckdb,” https://duckdb.org, 2019, accessed on November 10, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "“Sqlancer,” https://github.com/sqlancer/sqlancer, 2019, accessed on November 10, 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "“An error occurred when the cast function converted the numerical value in the form of scientific notation,” https://sqlite.org/forum/forumpost/ 3f085531bf, 2022, accessed on November 11, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "V. Benzaken and E. Contejean, “A coq mechanised formal semantics for realistic sql queries: formally reconciling sql and bag relational algebra,” inProceedings of the 8th ACMSIGPLAN International Conference on Certified Programs and Proofs, 2019, pp. 249–261.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "J. Castelein, M. Aniche, M. Soltani, A. Panichella, and A. van Deursen, “Search-based test data generation for sql queries,” in Proceedings of the 40th international conference on software engineering, 2018, pp. 1220–1230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "S. Ceri and G. Gottlob, “Translating sql into relational algebra: Optimization, semantics, and equivalence of sql queries,” IEEE Transactions on software engineering, no. 4, pp. 324–345, 1985.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "T. Y. Chen, S. C. Cheung, and S. Yiu, “Metamorphic testing: A new approach for generating next test cases,” CoRR, vol. abs/2002.12543, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "T. Y. Chen, F. Kuo, H. Liu, P. Poon, D. Towey, T. H. Tse, and Z. Q. Zhou, “Metamorphic testing: A review of challenges and opportunities,” ACM Comput. Surv., vol. 51, no. 1, pp. 4:1–4:27, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "X. Chen, C. Wang, and A. Cheung, “Testing query execution engines with mutations,” in Proceedings of the workshop on Testing Database Systems, 2020, pp. 1–5.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "J. Cheney and W. Ricciotti, “Comprehending nulls,” in The 18th International Symposium on Database Programming Languages, 2021, pp. 3–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "H. R. Chinaei, “An ordered bag semantics for sql,” Master’s thesis, University of Waterloo, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "S. Chu, C. Wang, K. Weitz, and A. Cheung, “Cosette: An automated prover for sql.” in CIDR, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "S. Chu, K. Weitz, A. Cheung, and D. Suciu, “Hottsql: Proving query rewrites with univalent sql semantics,” ACMSIGPLAN Notices, vol. 52, no. 6, pp. 510–524, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "W. F. Clocksin and C. S. Mellish, Programming in PROLOG. Springer Science & Business Media, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "J. Fu, J. Liang, Z. Wu, M. Wang, and Y. Jiang, “Griffin: Grammarfree dbms fuzzing,” in Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering, 2022, pp. 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "B. Ghit, N. Poggi, J. Rosen, R. Xin, and P. Boncz, “Sparkfuzz: Searching correctness regressions in modern query engines,” in Proceedings of the workshop on Testing Database Systems, 2020, pp. 1–6.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "P. Guagliardo and L. Libkin, “A formal semantics of sql queries, its validation, and applications,” Proceedings of the VLDB Endowment, vol. 11, no. 1, pp. 27–39, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "International Organization for Standardization, Information technology – Database languages – SQL – Part 2: Foundation (SQL/Foundation), 2016, no. 9075-2:2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "——, Information technology – Database languages – SQL – Part 2: Foundation (SQL/Foundation), 2016, no. 9075-2:2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "J. Jiao, S. Kan, S.-W. Lin, D. Sanan, Y. Liu, and J. Sun, “Semantic understanding of smart contracts: Executable operational semantics of solidity,” in. IEEE, 2020, pp. 1695–1712.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "J. Jung, H. Hu, J. Arulraj, T. Kim, and W. Kang, “Apollo: Automatic detection and diagnosis of performance regressions in database systems,” Proceedings of the VLDB Endowment, vol. 13, no. 1, pp. 57–70, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Y. Liang, S. Liu, and H. Hu, “Detecting logical bugs of {DBMS } with coverage-based guidance,” in 31st USENIX Security Symposium (USENIX Security 22), 2022, pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "S. Liu, C. Tian, J. Sun, R. Wang, W. Lu, Y. Zhao, Y. Xue, J. Wang, and X. Du, “Conformance testing of relational dbms against sql specifications (technical report),” https://github.com/DBMSTesting/ Technical-report, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "G. Malecha, G. Morrisett, A. Shinnar, and R. Wisnesky, “Toward a verified relational database management system,” in Proceedings of the 37th annual ACMSIGPLAN-SIGACT symposium on Principles of programming languages, 2010, pp. 237–248.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "M. Mitzenmacher and E. Upfal, Probability and Computing: Randomization and Probabilistic Techniques in Algorithms and Data Analysis, 2nd ed. USA: Cambridge University Press, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "M. Negri, G. Pelagatti, and L. Sbattella, “Formal semantics of sql queries,” ACM Transactions on Database Systems (TODS), vol. 16, no. 3, pp. 513–534, 1991.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "W. Ricciotti and J. Cheney, “A formalization of sql with nulls,” Journal of Automated Reasoning, vol. 66, no. 4, pp. 989–1030, 2022. 21",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 42,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "G. Ros ,u and T. F. S ,erb˘anut˘a, “An overview of the k semantic framework,” The Journal of Logic and Algebraic Programming, vol. 79, no. 6, pp. 397–434, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "R. Schumi and J. Sun, “Exais: executable ai semantics,” in Proceedings of the 44th International Conference on Software Engineering, 2022, pp. 859–870.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "S. Segura, G. Fraser, A. B. S ´anchez, and A. R. Cort ´es, “A survey on metamorphic testing,” IEEE Trans. Software Eng., vol. 42, no. 9, pp. 805–824, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "A. Seltenreich, B. Tang, and S. Mullender, “Sqlsmith,” 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "D. R. Slutz, “Massive stochastic testing of sql,” in VLDB, vol. 98. Citeseer, 1998, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "J. Van den Bussche and S. Vansummeren, “Translating sql into the relational algebra,” Course notes, Hasselt University and Universit ´e Libre de Bruxelles, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "M. Veanes, N. Tillmann, and J. De Halleux, “Qex: Symbolic sql query explorer,” in Logic for Programming, Artificial Intelligence, and Reasoning: 16th International Conference, LPAR-16, Dakar, Senegal, April 25–May 1, 2010, Revised Selected Papers 16. Springer, 2010, pp. 425–446.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "M. Wang, Z. Wu, X. Xu, J. Liang, C. Zhou, H. Zhang, and Y. Jiang, “Industry practice of coverage-guided enterprise-level dbms fuzzing,” in ing: Software Engineering in Practice (ICSE-SEIP). IEEE, 2021, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Q. Zhou, J. Arulraj, S. Navathe, W. Harris, and J. Wu, “A symbolic approach to proving query equivalence under bag semantics,” arXiv preprint arXiv:2004.00481, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Z. Zhou, S. Xiang, and T. Y. Chen, “Metamorphic testing for software quality assessment: A study of search engines,” IEEE Trans. Software Eng., vol. 42, no. 3, pp. 264–284, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "X. Zhu, S. Wen, S. Camtepe, and Y. Xiang, “Fuzzing: a survey for roadmap,” ACM Computing Surveys (CSUR), vol. 54, no. 11s, pp. 1–36, 2022.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 14,
+      "text": "“Sqlancer,” https://github.com/sqlancer/sqlancer, 2019, accessed on November 10, 2023.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 41,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proceedings of the ACM on Programming Languages, vol. 4, no. OOPSLA, pp. 1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20), 2020, pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[14]",
+      "technique": null,
+      "sentence": "The serious impacts of bugs in RDBMS have been discussed by various existing studies [14], [34], [40], [41], and thus it is critical to detect those bugs.",
+      "context_before": "uage for relational databases, and its specification is formally documented in ISO/IEC 9075:2016 [10]. As the parser and executor of SQL queries, an RDBMS should conform to the SQL specification to ensure the correct implementation of the semantics. As of 2023, there are more than 416 different implementations of relational databases, yet many of these implementations deviate from the specification [18], [47], [53]. Bugs have also been reported due to violations of the SQL specifications [15], which may potentially lead to data integrity issues or even security vulnerabilities in the database.",
+      "context_after": "As the golden standard for correct SQL behavior, the specification of SQL should be clearly described, and an RDBMS should conform to the SQL specification. Inconsistencies between RDBMS implementations and the specification can lead to unexpected results. Figure 1 presents two motivating examples, one bug and one inconsistency that our approach detected. Figure 1a is a SQL query that triggers a bug in TiDB version 6.6.01, which occurs when performing a bitwise operation on negative numbers, which are by default signed numbers. The root cause is that TiDB incorrectly represents the result of",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 3156,
+      "cited_reference": {
+        "number": 14,
+        "text": "“Sqlancer,” https://github.com/sqlancer/sqlancer, 2019, accessed on November 10, 2023.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[40]–[42]",
+      "technique": "norec",
+      "sentence": "However, existing approaches on testing RDBMSs either use different RDBMSs as the test oracle [47] or propose metamorphic relations [34], [40]–[42].",
+      "context_before": "ific data type in SQL representing unknown data. This inconsistency poses substantial challenges for users of database systems. When transitioning between databases and leveraging features with inconsistent implementations, users may encounter unexpected outcomes. From the motivational examples, we observe that failing to respect the SQL specification or unclearly documented specifications can result in bugs or inconsistent implementations across different RDBMSs, potentially confusing users. Therefore, it is critical to test the conformance of RDBMS implementations with the SQL specification.",
+      "context_after": "None of those approaches consider testing the conformance of RDBMS implementations with SQL specifications. Consequently, they are only scratching the surface in evaluating the correctness of RDBMS. To address the issue of automatic conformance testing between SQL specifications and RDBMS implementations, two main challenges arise. Firstly, the SQL specification is written in natural language, which is not directly executable. Secondly, it is challenging to generate test queries that comprehensively 3 cover all aspects defined in the SQL specification, including descriptions of keywords and p",
+      "section": "I INTRODUCTION",
+      "page": 2,
+      "char_offset": 6680,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Here, we 10 enhance the syntax-guided generation method, SQLancer [14], with coverage-guided test case generation.",
+      "context_before": "g 7, we first introduced the combination rules of different types of keyword semantics (semantic rule 102), and then we introduce the composite semantics based on those rules (semantic rules 103-104). IV. CONFORMANCE TESTING A. Overview of our approach Figure 8 illustrates the overview of our conformance testing approach, which consists of four components. Initially, we define the formal semantics of SQL, as detailed in Section III. Next, we implement the SQL formal semantics in Prolog, which serves as an oracle for conformance testing. The third component is dedicated to test case generation.",
+      "context_after": "This enhancement is based on three coverage criteria, i.e., keyword coverage, rule coverage, and composite rule coverage, which we proposed based on the formal semantics we defined. The final component is dedicated to query results comparison, wherein the query results of the tested RDBMS and our Prolog implementation are compared to identify conformance issues. Conformance issues in our work refer to two types of issues, i.e., bugs or inconsistencies. Both issues are due to violating the SQL semantics defined in the SQL specification, or unclear or missing descriptions in the SQL specificati",
+      "section": "A Overview of our approach",
+      "page": 9,
+      "char_offset": 33370,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Coverage guided test case generation The state-of-the-art practice in test case generation involves randomly generating SQL queries guided by the syntax of SQL, among which SQLancer [14] stands out as one of the most effective tools of this kind.",
+      "context_before": "s defined in the SQL specification, or unclear or missing descriptions in the SQL specification. Bugs are defects that are confirmed by developers. Inconsistencies refer to inconsistent result produced by the tested RDBMS and SEMCONT. Inconsistencies are also confirmed by developers, yet they perceive them as deliberate design choices rather than bugs. We report these inconsistencies because various RDBMSs make differing design decisions, leading to varied query results that may potentially perplex users. This also underscore the importance of a comprehensively documented SQL specification. B.",
+      "context_after": "It is tailored to the syntactical structures of various RDBMSs. SQLancer considers database objects, such as tables, views, and indexes, as well as keywords and functionalities within query statements. Throughout the generation process, SQLancer maintains a set of keywords and functionalities, from which it randomly selects keywords to incorporate into the test cases, subject to syntactic rules of SQL (so that they remain syntactically valid). However, this generation process is entirely random and does not consider coverage of the SQL semantics. As a result, certain aspects of the semantics",
+      "section": "B Coverage guided test case generation",
+      "page": 10,
+      "char_offset": 34592,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer considers database objects, such as tables, views, and indexes, as well as keywords and functionalities within query statements.",
+      "context_before": "deliberate design choices rather than bugs. We report these inconsistencies because various RDBMSs make differing design decisions, leading to varied query results that may potentially perplex users. This also underscore the importance of a comprehensively documented SQL specification. B. Coverage guided test case generation The state-of-the-art practice in test case generation involves randomly generating SQL queries guided by the syntax of SQL, among which SQLancer [14] stands out as one of the most effective tools of this kind. It is tailored to the syntactical structures of various RDBMSs.",
+      "context_after": "Throughout the generation process, SQLancer maintains a set of keywords and functionalities, from which it randomly selects keywords to incorporate into the test cases, subject to syntactic rules of SQL (so that they remain syntactically valid). However, this generation process is entirely random and does not consider coverage of the SQL semantics. As a result, certain aspects of the semantics may never be tested. To address this problem, we propose three coverage criteria, i.e., keyword coverage, rule coverage, and composite rule coverage, based on the formal semantics we defined. Each cover",
+      "section": "B Coverage guided test case generation",
+      "page": 10,
+      "char_offset": 34903,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Throughout the generation process, SQLancer maintains a set of keywords and functionalities, from which it randomly selects keywords to incorporate into the test cases, subject to syntactic rules of SQL (so that they remain syntactically valid).",
+      "context_before": "g to varied query results that may potentially perplex users. This also underscore the importance of a comprehensively documented SQL specification. B. Coverage guided test case generation The state-of-the-art practice in test case generation involves randomly generating SQL queries guided by the syntax of SQL, among which SQLancer [14] stands out as one of the most effective tools of this kind. It is tailored to the syntactical structures of various RDBMSs. SQLancer considers database objects, such as tables, views, and indexes, as well as keywords and functionalities within query statements.",
+      "context_after": "However, this generation process is entirely random and does not consider coverage of the SQL semantics. As a result, certain aspects of the semantics may never be tested. To address this problem, we propose three coverage criteria, i.e., keyword coverage, rule coverage, and composite rule coverage, based on the formal semantics we defined. Each coverage criterion defines coverage at a different granularity level. Keyword coverage simply assesses whether each individual keyword in the SQL specification is covered. Rule coverage goes a step further by calculating whether each semantic rule, wh",
+      "section": "B Coverage guided test case generation",
+      "page": 10,
+      "char_offset": 35041,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We adopt SQLancer to randomly generatea large number of SQL statements (line 2), which serves as the seed pool of our query generation algorithm.",
+      "context_before": "and match the rules based on the parameter or input of that keyword. In this case, the SELECT rule that takes a column list as input and the FROM rule that takes one table as input are matched to form the composite rule covered by the given query. The time complexity of Algorithm 3 is O(n) with n being the number of operators in the given SQL query. Coverage-guided query generation. Algorithm 2 describes the process of query generation guided by coverage. We set the initial coverage to be 0 and the declare covered set (coveredSet ), which records the covered keywords, rules, or combined rules.",
+      "context_after": "The algorithm begins with randomly selecting an SQL statement from the seed pool (line 6), and mutates the query based on the mutation rules we proposed. Then we calculate coverage of the mutated query (line 8). We keep this mutation process (line 5-9) until the mutated query increases the overall coverage. Then the mutated query is added into the seed pool for future test case generation. The mutated query is executed to explore potential inconsistencies (line 11). The process terminates upon timeout. Function CalculateCoverage calculates the coverage of the given query. It first checks whet",
+      "section": "B Coverage guided test case generation",
+      "page": 11,
+      "char_offset": 41312,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[40]",
+      "technique": "norec",
+      "sentence": "It is important to note that we used the latest release of each RDBMS, which has been extensively tested by existing approaches [40], [41].",
+      "context_before": "e conducted all experiments on a server with two Intel(R) Xeon(R) Platinum 8260 CPUs at 2.30 GHz and 502 GB of memory, running Ubuntu 18.04.6 LTS. The SQL formal semantics were implemented in Prolog, while the SQL query generation program was developed in Java. We ran the experiments using Java version 11.0.15.1. Target RDBMS. We selected six popular and widely used RDBMSs, each offering a range of distinct features and application scenarios, to demonstrate the effectiveness of our approach. The statistics of these RDBMSs, as obtained from their opensource repositories, are shown in Table III.",
+      "context_after": "MySQL [3] and PostgreSQL [5] are the two most popular open-source database management systems. SQLite [6] and DuckDB [13] are both embedded DBMSs, running within the process of other applications. TiDB [12] and OceanBase [11] are popular distributed RDBMSs. 14 TABLE IV: Bugs and inconsistencies detected by SEMCONT SNID Target Type Reason Status 1 109146 MySQL Bug missing spec duplicate 2 109837 MySQL Bug missing spec comfirmed 3 109842 MySQL Bug missing spec comfirmed 4 109149 MySQL Bug missing spec comfirmed 5 110438 MySQL Bug violate spec comfirmed 6 109147 MySQL Inconsistency missing spec",
+      "section": "A Experiment setup",
+      "page": 13,
+      "char_offset": 55223,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+      "context_before": "irmed 21 42378 TiDB Bug violate spec comfirmed 22 42379 TiDB Bug violate spec comfirmed 23 42377 TiDB Bug violate spec comfirmed 24 42773 TiDB Inconsistency missing spec comfirmed 25 40995 TiDB Inconsistency missing spec comfirmed 26 7e03a4420a SQLite Bug missing spec comfirmed 27 3f085531bf SQLite Bug missing spec comfirmed 28 6e4d3e389e SQLite Bug violate spec comfirmed 29 411bce39d0 SQLite Inconsistency missing spec comfirmed 30 6804 DuckDB Bug violate spec fixed 31 2104 OceanBase Inconsistency missing spec confirmed 32 2105 OceanBase Inconsistency missing spec confirmed Compared baselines.",
+      "context_after": "NoREC constructs two semantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results. TLP, on the other hand, partitions the conditional expression of the original query into three segments, corresponding to the three possible results, i.e., TRUE, FALSE, and NULL, of the conditional expression. It then compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 57222,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC constructs two semantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results.",
+      "context_before": "rmed 24 42773 TiDB Inconsistency missing spec comfirmed 25 40995 TiDB Inconsistency missing spec comfirmed 26 7e03a4420a SQLite Bug missing spec comfirmed 27 3f085531bf SQLite Bug missing spec comfirmed 28 6e4d3e389e SQLite Bug violate spec comfirmed 29 411bce39d0 SQLite Inconsistency missing spec comfirmed 30 6804 DuckDB Bug violate spec fixed 31 2104 OceanBase Inconsistency missing spec confirmed 32 2105 OceanBase Inconsistency missing spec confirmed Compared baselines. We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+      "context_after": "TLP, on the other hand, partitions the conditional expression of the original query into three segments, corresponding to the three possible results, i.e., TRUE, FALSE, and NULL, of the conditional expression. It then compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness in RDBMS bug detection. Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34]. SQLancer adopts a generative approach for query generation",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 57346,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP, on the other hand, partitions the conditional expression of the original query into three segments, corresponding to the three possible results, i.",
+      "context_before": "27 3f085531bf SQLite Bug missing spec comfirmed 28 6e4d3e389e SQLite Bug violate spec comfirmed 29 411bce39d0 SQLite Inconsistency missing spec comfirmed 30 6804 DuckDB Bug violate spec fixed 31 2104 OceanBase Inconsistency missing spec confirmed 32 2105 OceanBase Inconsistency missing spec confirmed Compared baselines. We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS. NoREC constructs two semantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results.",
+      "context_after": "e., TRUE, FALSE, and NULL, of the conditional expression. It then compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness in RDBMS bug detection. Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34]. SQLancer adopts a generative approach for query generation and SQLRight adopts a mutation-based approach for generating queries. Therefore, in our experiment, we have four combined settings (concerning the oracl",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 57500,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34].",
+      "context_before": "mantically equivalent queries, one triggers the optimization and the other does not, executes the queries and compare the results. TLP, on the other hand, partitions the conditional expression of the original query into three segments, corresponding to the three possible results, i.e., TRUE, FALSE, and NULL, of the conditional expression. It then compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness in RDBMS bug detection.",
+      "context_after": "SQLancer adopts a generative approach for query generation and SQLRight adopts a mutation-based approach for generating queries. Therefore, in our experiment, we have four combined settings (concerning the oracle and query generation method) for the compared baselines, i.e., NoREC (SQLancer), NoREC (SQLRight), TLP (SQLancer) and TLP (SQLRight). B. Experiment results RQ1: Bugs and inconsistencies. We ran SEMCONT on four RDBMSs for a period of 3 months and reported the detected issues to the corresponding developer communities. Table IV shows the details of the confirmed bugs and inconsistencie",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 57970,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer adopts a generative approach for query generation and SQLRight adopts a mutation-based approach for generating queries.",
+      "context_before": "her does not, executes the queries and compare the results. TLP, on the other hand, partitions the conditional expression of the original query into three segments, corresponding to the three possible results, i.e., TRUE, FALSE, and NULL, of the conditional expression. It then compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness in RDBMS bug detection. Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34].",
+      "context_after": "Therefore, in our experiment, we have four combined settings (concerning the oracle and query generation method) for the compared baselines, i.e., NoREC (SQLancer), NoREC (SQLRight), TLP (SQLancer) and TLP (SQLRight). B. Experiment results RQ1: Bugs and inconsistencies. We ran SEMCONT on four RDBMSs for a period of 3 months and reported the detected issues to the corresponding developer communities. Table IV shows the details of the confirmed bugs and inconsistencies in four RDBMSs detected by SEMCONT. We have submitted 32 issues and 19 of them are confirmed by the developers as bugs. Out of",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 58041,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", NoREC (SQLancer), NoREC (SQLRight), TLP (SQLancer) and TLP (SQLRight).",
+      "context_before": "en compares the union of the result sets from executing the three queries with the three segments each, with the result set of the original statement, expecting them to be identical. Both approaches have demonstrated effectiveness in RDBMS bug detection. Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34]. SQLancer adopts a generative approach for query generation and SQLRight adopts a mutation-based approach for generating queries. Therefore, in our experiment, we have four combined settings (concerning the oracle and query generation method) for the compared baselines, i.e.",
+      "context_after": "B. Experiment results RQ1: Bugs and inconsistencies. We ran SEMCONT on four RDBMSs for a period of 3 months and reported the detected issues to the corresponding developer communities. Table IV shows the details of the confirmed bugs and inconsistencies in four RDBMSs detected by SEMCONT. We have submitted 32 issues and 19 of them are confirmed by the developers as bugs. Out of the issues identified, 23 are related to scalarexpressions and 9 to other keywords, including joins and various relational operators. Our primary objective is to detect inconsistencies between RDBMS implementations and",
+      "section": "A Experiment setup",
+      "page": 14,
+      "char_offset": 58316,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[40]–\n[42]",
+      "technique": "norec",
+      "sentence": "They are also under-specified in the SQL standard and insufficiently tested by existing approaches [34], [40]– [42].",
+      "context_before": "es and 19 of them are confirmed by the developers as bugs. Out of the issues identified, 23 are related to scalarexpressions and 9 to other keywords, including joins and various relational operators. Our primary objective is to detect inconsistencies between RDBMS implementations and the SQL specification by generating test cases that achieve high coverage across different SQL keywords. Our implementation focuses on scalar expressions, query expressions, and predicates. Among these, scalar expressions are the most complex, as they often involve combinations of multiple keywords and subqueries.",
+      "context_after": "In contrast, query expressions and predicates are clearly defined in the SQL specification, leading to fewer ambiguities across RDBMSs. Moreover, they have been extensively tested by prior research [34], [40]–[42], making it more challenging to uncover new inconsistencies. Among the confirmed bugs, 1 has been reported previously and 1 has been fixed. The remaining 13 issues are confirmed as inconsistencies, and the developers claim that they were their design choices. Among the 13 inconsistencies, 2 of them are due to violation of the SQL specification and 11 of them are due to unclear descri",
+      "section": "B Experiment results",
+      "page": 14,
+      "char_offset": 59304,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[40]–[42]",
+      "technique": "norec",
+      "sentence": "Moreover, they have been extensively tested by prior research [34], [40]–[42], making it more challenging to uncover new inconsistencies.",
+      "context_before": "tween RDBMS implementations and the SQL specification by generating test cases that achieve high coverage across different SQL keywords. Our implementation focuses on scalar expressions, query expressions, and predicates. Among these, scalar expressions are the most complex, as they often involve combinations of multiple keywords and subqueries. They are also under-specified in the SQL standard and insufficiently tested by existing approaches [34], [40]– [42]. In contrast, query expressions and predicates are clearly defined in the SQL specification, leading to fewer ambiguities across RDBMSs.",
+      "context_after": "Among the confirmed bugs, 1 has been reported previously and 1 has been fixed. The remaining 13 issues are confirmed as inconsistencies, and the developers claim that they were their design choices. Among the 13 inconsistencies, 2 of them are due to violation of the SQL specification and 11 of them are due to unclear descriptions in the SQL specification. Developers of different RDBMSs could have different interpretations on the SQL specification, and thus design and implement their RDBMS differently. Among the nine inconsistencies arising from unclear standard descriptions, eight inconsisten",
+      "section": "B Experiment results",
+      "page": 14,
+      "char_offset": 59557,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[40]–[42]",
+      "technique": "norec",
+      "sentence": "It is noteworthy that all four databases have been extensively tested by existing methods [40]–[42], yet SEMCONTis still able to detect bugs that were not detected by those approaches.",
+      "context_before": "SQL specification and 11 of them are due to unclear descriptions in the SQL specification. Developers of different RDBMSs could have different interpretations on the SQL specification, and thus design and implement their RDBMS differently. Among the nine inconsistencies arising from unclear standard descriptions, eight inconsistencies were detected in both MySQL (109147, 109148, 109836, 109845, 109962, 110711) and TiDB (42773, 40995). The queries triggering the inconsistencies have the same results when executed in MySQL, TiDB, and MariaDB databases, and are different from that of PostgreSQL.",
+      "context_after": "By a careful inspection on the detected bugs, we find that most of them indeed violate the SQL specification. One of the most representative bugs is related to the mishandling of NULL operands in keyword operations. According to the SQL specification, “ If the value of one or more, <string value expression >s,<datetime value expression >s,<interval value expression >s, and <collection value expression >s that are simply contained in a <numeric value function >is the NULL value, then the result of the <numeric value function >is the NULL value ” [30]. However, MySQL violates the specification b",
+      "section": "B Experiment results",
+      "page": 14,
+      "char_offset": 60562,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In Figure 9, SQLancer+ keyword syntax represents the setting of adding keywords and the corresponding generation rules which were not supported by SQLancer.",
+      "context_before": "onsistencies are either due to RDBMS violating the SQL specification, or missing or unclear SQL specification. RQ2: Effectiveness of the coverage criteria? We measure the effectiveness of the proposed coverage criteria in two aspects, i.e., whether they are effective in guiding generating test cases that achieve higher coverage, and whether they are effective in guiding generating test cases that uncover unknown bugs or inconsistencies. The experiment results on keyword coverage, rule coverage and composite rule coverage improvement are shown in Figure 9, Figure 10 and Figure 11, respectively.",
+      "context_after": "SQLancer+ keyword syntax greatly improved the keyword coverage for all four databases. Notably, within the first 1500 SQL statements, over 80% of the keywords were covered on all four databases, with SQLite achieving an impressive keyword coverage of 99%. Keyword coverage guided query generation (SQLancer+ keyword coverage) further improves the keyword coverage, and achieved 100% keyword coverage within the first 200 generate queries for all databases, demonstrating the effectiveness of our keyword-guided query generation method. Figure 10 shows the results on rule coverage, which show simila",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 65358,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer+ keyword syntax greatly improved the keyword coverage for all four databases.",
+      "context_before": "We measure the effectiveness of the proposed coverage criteria in two aspects, i.e., whether they are effective in guiding generating test cases that achieve higher coverage, and whether they are effective in guiding generating test cases that uncover unknown bugs or inconsistencies. The experiment results on keyword coverage, rule coverage and composite rule coverage improvement are shown in Figure 9, Figure 10 and Figure 11, respectively.In Figure 9, SQLancer+ keyword syntax represents the setting of adding keywords and the corresponding generation rules which were not supported by SQLancer.",
+      "context_after": "Notably, within the first 1500 SQL statements, over 80% of the keywords were covered on all four databases, with SQLite achieving an impressive keyword coverage of 99%. Keyword coverage guided query generation (SQLancer+ keyword coverage) further improves the keyword coverage, and achieved 100% keyword coverage within the first 200 generate queries for all databases, demonstrating the effectiveness of our keyword-guided query generation method. Figure 10 shows the results on rule coverage, which show similar trend with that on keyword coverage. Due to the limited support of SQL features, e.g.",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 65514,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Keyword coverage guided query generation (SQLancer+ keyword coverage) further improves the keyword coverage, and achieved 100% keyword coverage within the first 200 generate queries for all databases, demonstrating the effectiveness of our keyword-guided query generation method.",
+      "context_before": "own bugs or inconsistencies. The experiment results on keyword coverage, rule coverage and composite rule coverage improvement are shown in Figure 9, Figure 10 and Figure 11, respectively.In Figure 9, SQLancer+ keyword syntax represents the setting of adding keywords and the corresponding generation rules which were not supported by SQLancer. SQLancer+ keyword syntax greatly improved the keyword coverage for all four databases. Notably, within the first 1500 SQL statements, over 80% of the keywords were covered on all four databases, with SQLite achieving an impressive keyword coverage of 99%.",
+      "context_after": "Figure 10 shows the results on rule coverage, which show similar trend with that on keyword coverage. Due to the limited support of SQL features, e.g., data types, by SQLancer, especially for DBMS such as DuckDB and TiDB, relying only on SQLancer achieves low rule coverage, as shown in Figure 10. Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax. We can observe that adding those missing features improves the rule coverage, especially for DuckDB and TiDB. Rule coverage-guided query generation (SQLancer+ rul",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 65770,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": ", data types, by SQLancer, especially for DBMS such as DuckDB and TiDB, relying only on SQLancer achieves low rule coverage, as shown in Figure 10.",
+      "context_before": "Notably, within the first 1500 SQL statements, over 80% of the keywords were covered on all four databases, with SQLite achieving an impressive keyword coverage of 99%. Keyword coverage guided query generation (SQLancer+ keyword coverage) further improves the keyword coverage, and achieved 100% keyword coverage within the first 200 generate queries for all databases, demonstrating the effectiveness of our keyword-guided query generation method. Figure 10 shows the results on rule coverage, which show similar trend with that on keyword coverage. Due to the limited support of SQL features, e.g.",
+      "context_after": "Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax. We can observe that adding those missing features improves the rule coverage, especially for DuckDB and TiDB. Rule coverage-guided query generation (SQLancer+ rule coverage) achieves the highest rule coverage with the fewest number of queries. The results indicate the effectiveness of our rule coverage-guided query generation algorithm. Figure 11 depicts the improvements in composite rule coverage by the generated queries for the four databases. With SQLan",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 66201,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M22",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax.",
+      "context_before": "yword coverage of 99%. Keyword coverage guided query generation (SQLancer+ keyword coverage) further improves the keyword coverage, and achieved 100% keyword coverage within the first 200 generate queries for all databases, demonstrating the effectiveness of our keyword-guided query generation method. Figure 10 shows the results on rule coverage, which show similar trend with that on keyword coverage. Due to the limited support of SQL features, e.g., data types, by SQLancer, especially for DBMS such as DuckDB and TiDB, relying only on SQLancer achieves low rule coverage, as shown in Figure 10.",
+      "context_after": "We can observe that adding those missing features improves the rule coverage, especially for DuckDB and TiDB. Rule coverage-guided query generation (SQLancer+ rule coverage) achieves the highest rule coverage with the fewest number of queries. The results indicate the effectiveness of our rule coverage-guided query generation algorithm. Figure 11 depicts the improvements in composite rule coverage by the generated queries for the four databases. With SQLancer, which conducts random query generation, we observed that the increase in composite rule coverage tends to plateau after generating 60",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 66348,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Rule coverage-guided query generation (SQLancer+ rule coverage) achieves the highest rule coverage with the fewest number of queries.",
+      "context_before": "veness of our keyword-guided query generation method. Figure 10 shows the results on rule coverage, which show similar trend with that on keyword coverage. Due to the limited support of SQL features, e.g., data types, by SQLancer, especially for DBMS such as DuckDB and TiDB, relying only on SQLancer achieves low rule coverage, as shown in Figure 10. Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax. We can observe that adding those missing features improves the rule coverage, especially for DuckDB and TiDB.",
+      "context_after": "The results indicate the effectiveness of our rule coverage-guided query generation algorithm. Figure 11 depicts the improvements in composite rule coverage by the generated queries for the four databases. With SQLancer, which conducts random query generation, we observed that the increase in composite rule coverage tends to plateau after generating 60 million data points. At this stage, MySQL, SQLite, DuckDB and OceanBase each achieved a composite rule coverage of around 70% and TiDB 50%. With the introduction of composite rule coverage guidance, all four databases were able to reach 100% co",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 66597,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "With SQLancer, which conducts random query generation, we observed that the increase in composite rule coverage tends to plateau after generating 60 million data points.",
+      "context_before": "Figure 10. Therefore, we add those missing features in SQLancer for the corresponding DBMS query generation and refer this as SQLancer + rule syntax. We can observe that adding those missing features improves the rule coverage, especially for DuckDB and TiDB. Rule coverage-guided query generation (SQLancer+ rule coverage) achieves the highest rule coverage with the fewest number of queries. The results indicate the effectiveness of our rule coverage-guided query generation algorithm. Figure 11 depicts the improvements in composite rule coverage by the generated queries for the four databases.",
+      "context_after": "At this stage, MySQL, SQLite, DuckDB and OceanBase each achieved a composite rule coverage of around 70% and TiDB 50%. With the introduction of composite rule coverage guidance, all four databases were able to reach 100% composite rule coverage after generating 20 million queries (our Prolog implementation has a total of 19 million composite rules). The results indicate the effectiveness of our composite rule coverage-guided query generation algorithm. To verify the effectiveness of the coverage-guided query generation algorithm in assisting detecting bugs and inconsistencies in relational DB",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 66937,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M25",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Note that to conduct fair comparisons, we improved SQLancer by incorporating all keywords supported by our semantics and related generation rules in SEMCONT.",
+      "context_before": "uided query generation algorithm. To verify the effectiveness of the coverage-guided query generation algorithm in assisting detecting bugs and inconsistencies in relational DBMS, we conducted an ablation study of SEMCONT with and without coverage guidance. Table V] presents the experimental results obtained from testing four databases over a period of 6 hours. We record the number of bugs and inconsistencies detected on the four settings, i.e., SEMCONT without coverage guidance, and SEMCONT with three coverage guidance. We also report the time taken to discover the first bug or inconsistency.",
+      "context_after": "The experimental results indicate that within a 6-hour timeframe, all three coverage metrics successfully assist detecting more bugs and inconsistencies compared with random generation. Composite rule coverage is the most effective among all three coverage metrics. In terms of the time taken to detect the first bug or inconsistency, all three coverage guidance algorithm are faster than SEMCONT with 16 (a) MySQL (b) TiDB (c) SQLite (d) DuckDB (e) OceanBase (f) PostgreSQL Fig. 9: The keyword coverage increment (y-axis) with the number of queries (x-axis) (a) MySQL (b) TiDB (c) SQLite (d",
+      "section": "B Experiment results",
+      "page": 15,
+      "char_offset": 68131,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "We compare SEMCONT with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS.",
+      "context_before": "5 2 7.46 7 2 8.62 SQLite 1 1 20.50 2 1 17.83 3 1 14.33 3 1 13.05 DuckDB 1 0 29.37 1 0 27.32 1 0 31.68 1 0 26.29 OceanBase 0 2 13.25 0 2 13.37 0 2 7.92 0 2 9.13 random query generation. In particular, keyword coverage, rule coverage and composite rule coverage are 5.22%, 24.19% and 21.41% faster than random query generation. Answer to RQ2: The three coverage criteria all improve the query generating process, triggering more bugs and inconsistencies with faster speed. Composite rule coverage achieves the most significant improvement. RQ3: How does SEMCONT perform compared to baseline approaches?",
+      "context_after": "For both approaches, we adopt SQLancer [14] and SQLRight [34] for query generation. Notably, SQLancer does not support the NoREC oracle for MySQL and TiDB, while SQLRight does not support TiDB, DuckDB and OceanBase. Therefore, we excluded these specific scenarios from our experiments. We ran the comparedtools for a period of 6 hours and report the results in Table VI. The experimental results show that both SQLancer and SQLRight using the NoREC as the oracle were unable to detect new bugs or inconsistencies. The TLP oracle with SQLancer for query generation detected 4 bugs in three databases,",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 70163,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For both approaches, we adopt SQLancer [14] and SQLRight [34] for query generation.",
+      "context_before": "2 7.92 0 2 9.13 random query generation. In particular, keyword coverage, rule coverage and composite rule coverage are 5.22%, 24.19% and 21.41% faster than random query generation. Answer to RQ2: The three coverage criteria all improve the query generating process, triggering more bugs and inconsistencies with faster speed. Composite rule coverage achieves the most significant improvement. RQ3: How does SEMCONT perform compared to baseline approaches? We compare SEMCONT with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS.",
+      "context_after": "Notably, SQLancer does not support the NoREC oracle for MySQL and TiDB, while SQLRight does not support TiDB, DuckDB and OceanBase. Therefore, we excluded these specific scenarios from our experiments. We ran the comparedtools for a period of 6 hours and report the results in Table VI. The experimental results show that both SQLancer and SQLRight using the NoREC as the oracle were unable to detect new bugs or inconsistencies. The TLP oracle with SQLancer for query generation detected 4 bugs in three databases, and with SQLRight for query generation detectd 1 bug in MySQL. SEMCONT outperformed",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 70306,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M28",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Notably, SQLancer does not support the NoREC oracle for MySQL and TiDB, while SQLRight does not support TiDB, DuckDB and OceanBase.",
+      "context_before": "rage and composite rule coverage are 5.22%, 24.19% and 21.41% faster than random query generation. Answer to RQ2: The three coverage criteria all improve the query generating process, triggering more bugs and inconsistencies with faster speed. Composite rule coverage achieves the most significant improvement. RQ3: How does SEMCONT perform compared to baseline approaches? We compare SEMCONT with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS. For both approaches, we adopt SQLancer [14] and SQLRight [34] for query generation.",
+      "context_after": "Therefore, we excluded these specific scenarios from our experiments. We ran the comparedtools for a period of 6 hours and report the results in Table VI. The experimental results show that both SQLancer and SQLRight using the NoREC as the oracle were unable to detect new bugs or inconsistencies. The TLP oracle with SQLancer for query generation detected 4 bugs in three databases, and with SQLRight for query generation detectd 1 bug in MySQL. SEMCONT outperformed the compared approaches and detected 16 bugs and 9 inconsistencies in the four databases. The reason is that existing approaches do",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 70390,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The experimental results show that both SQLancer and SQLRight using the NoREC as the oracle were unable to detect new bugs or inconsistencies.",
+      "context_before": "ignificant improvement. RQ3: How does SEMCONT perform compared to baseline approaches? We compare SEMCONT with two state-of-the-art approaches TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS. For both approaches, we adopt SQLancer [14] and SQLRight [34] for query generation. Notably, SQLancer does not support the NoREC oracle for MySQL and TiDB, while SQLRight does not support TiDB, DuckDB and OceanBase. Therefore, we excluded these specific scenarios from our experiments. We ran the comparedtools for a period of 6 hours and report the results in Table VI.",
+      "context_after": "The TLP oracle with SQLancer for query generation detected 4 bugs in three databases, and with SQLRight for query generation detectd 1 bug in MySQL. SEMCONT outperformed the compared approaches and detected 16 bugs and 9 inconsistencies in the four databases. The reason is that existing approaches do not consider the SQL specification and thus fail to find bugs that violated the SQL specification. For instance, One bug (109842) we detected in MySQL is related to the MOD function. When applied to negative numbers, MySQL incorrectly represents the result as-0. Both TLP and NoREC failed to detec",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 70677,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The TLP oracle with SQLancer for query generation detected 4 bugs in three databases, and with SQLRight for query generation detectd 1 bug in MySQL.",
+      "context_before": "TLP [40] and NoREC [41], which are metamorphic testing approaches for relational DBMS. For both approaches, we adopt SQLancer [14] and SQLRight [34] for query generation. Notably, SQLancer does not support the NoREC oracle for MySQL and TiDB, while SQLRight does not support TiDB, DuckDB and OceanBase. Therefore, we excluded these specific scenarios from our experiments. We ran the comparedtools for a period of 6 hours and report the results in Table VI. The experimental results show that both SQLancer and SQLRight using the NoREC as the oracle were unable to detect new bugs or inconsistencies.",
+      "context_after": "SEMCONT outperformed the compared approaches and detected 16 bugs and 9 inconsistencies in the four databases. The reason is that existing approaches do not consider the SQL specification and thus fail to find bugs that violated the SQL specification. For instance, One bug (109842) we detected in MySQL is related to the MOD function. When applied to negative numbers, MySQL incorrectly represents the result as-0. Both TLP and NoREC failed to detect this bug, even after successfully generated the bug triggering query. On average, our tool finds a bug in 67 minutes using 19,381 test cases, compa",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 70820,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Both TLP and NoREC failed to detect this bug, even after successfully generated the bug triggering query.",
+      "context_before": "detect new bugs or inconsistencies. The TLP oracle with SQLancer for query generation detected 4 bugs in three databases, and with SQLRight for query generation detectd 1 bug in MySQL. SEMCONT outperformed the compared approaches and detected 16 bugs and 9 inconsistencies in the four databases. The reason is that existing approaches do not consider the SQL specification and thus fail to find bugs that violated the SQL specification. For instance, One bug (109842) we detected in MySQL is related to the MOD function. When applied to negative numbers, MySQL incorrectly represents the result as-0.",
+      "context_after": "On average, our tool finds a bug in 67 minutes using 19,381 test cases, compared to 300 minutes and 1.3 million test cases for SQLancer, and 30 hours and 8.4 million test cases for SQLRight. Our approach finds more bugs/inconsistencies 17 TABLE VI: The bugs and inconsistencies detected by SQLancer, SQLRight and SEMCONT in 6h DBMSTLP (SQLancer) NoREC (SQLancer) TLP (SQLRight) NoREC (SQLRight) SEMCONT Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies MySQL 1 0 - - 1 0 0 0 5 6 TiDB 2 0 - - - - - - 7 2 SQLite 0 1 0 0 0 0 0 0 3 1 DuckDB 1 0 0",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 71385,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "3 million test cases for SQLancer, and 30 hours and 8.",
+      "context_before": "compared approaches and detected 16 bugs and 9 inconsistencies in the four databases. The reason is that existing approaches do not consider the SQL specification and thus fail to find bugs that violated the SQL specification. For instance, One bug (109842) we detected in MySQL is related to the MOD function. When applied to negative numbers, MySQL incorrectly represents the result as-0. Both TLP and NoREC failed to detect this bug, even after successfully generated the bug triggering query. On average, our tool finds a bug in 67 minutes using 19,381 test cases, compared to 300 minutes and 1.",
+      "context_after": "4 million test cases for SQLRight. Our approach finds more bugs/inconsistencies 17 TABLE VI: The bugs and inconsistencies detected by SQLancer, SQLRight and SEMCONT in 6h DBMSTLP (SQLancer) NoREC (SQLancer) TLP (SQLRight) NoREC (SQLRight) SEMCONT Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies MySQL 1 0 - - 1 0 0 0 5 6 TiDB 2 0 - - - - - - 7 2 SQLite 0 1 0 0 0 0 0 0 3 1 DuckDB 1 0 0 0 - - - - 1 0 OceanBase 0 1 0 0 - - - - 0 2 with fewer test cases, demonstrating its effectiveness and efficiency in detecting bugs and inconsistencies that",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 71594,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M33",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Our approach finds more bugs/inconsistencies 17 TABLE VI: The bugs and inconsistencies detected by SQLancer, SQLRight and SEMCONT in 6h DBMSTLP (SQLancer) NoREC (SQLancer) TLP (SQLRight) NoREC (SQLRight) SEMCONT Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies MySQL 1 0 - - 1 0 0 0 5 6 TiDB 2 0 - - - - - - 7 2 SQLite 0 1 0 0 0 0 0 0 3 1 DuckDB 1 0 0 0 - - - - 1 0 OceanBase 0 1 0 0 - - - - 0 2 with fewer test cases, demonstrating its effectiveness and efficiency in detecting bugs and inconsistencies that violating SQL specifications.",
+      "context_before": "he reason is that existing approaches do not consider the SQL specification and thus fail to find bugs that violated the SQL specification. For instance, One bug (109842) we detected in MySQL is related to the MOD function. When applied to negative numbers, MySQL incorrectly represents the result as-0. Both TLP and NoREC failed to detect this bug, even after successfully generated the bug triggering query. On average, our tool finds a bug in 67 minutes using 19,381 test cases, compared to 300 minutes and 1.3 million test cases for SQLancer, and 30 hours and 8.4 million test cases for SQLRight.",
+      "context_after": "(a) MySQL (b) TiDB (c) SQLite (d) DuckDB (e) OceanBase (f) PostgreSQL Fig. 12: The memory consumption of SEMCONT with SQLancer. Memory Consumption. Figure 12 and Figure 13 shows the memory usage of SEMCONT and SQLancer during a 6-hour test on six databases. Memory usage is mainly influenced by query generation, execution, and result comparison, with query generation being the most memory-intensive. We report stabilized memory consumption, with detailed time-based changes provided in our technical report [35]. Results indicate that SEMCONT’s memory usage is comparable to the baseline. Memor",
+      "section": "B Experiment results",
+      "page": 16,
+      "char_offset": 71682,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "12: The memory consumption of SEMCONT with SQLancer.",
+      "context_before": "nconsistencies detected by SQLancer, SQLRight and SEMCONT in 6h DBMSTLP (SQLancer) NoREC (SQLancer) TLP (SQLRight) NoREC (SQLRight) SEMCONT Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies MySQL 1 0 - - 1 0 0 0 5 6 TiDB 2 0 - - - - - - 7 2 SQLite 0 1 0 0 0 0 0 0 3 1 DuckDB 1 0 0 0 - - - - 1 0 OceanBase 0 1 0 0 - - - - 0 2 with fewer test cases, demonstrating its effectiveness and efficiency in detecting bugs and inconsistencies that violating SQL specifications. (a) MySQL (b) TiDB (c) SQLite (d) DuckDB (e) OceanBase (f) PostgreSQL Fig.",
+      "context_after": "Memory Consumption. Figure 12 and Figure 13 shows the memory usage of SEMCONT and SQLancer during a 6-hour test on six databases. Memory usage is mainly influenced by query generation, execution, and result comparison, with query generation being the most memory-intensive. We report stabilized memory consumption, with detailed time-based changes provided in our technical report [35]. Results indicate that SEMCONT’s memory usage is comparable to the baseline. Memory consumption is low because we use small tables (a few hundred records). We do not test concurrent query execution, resulting in s",
+      "section": "B Experiment results",
+      "page": 17,
+      "char_offset": 72355,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M35",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Figure 12 and Figure 13 shows the memory usage of SEMCONT and SQLancer during a 6-hour test on six databases.",
+      "context_before": "SQLancer) NoREC (SQLancer) TLP (SQLRight) NoREC (SQLRight) SEMCONT Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies Bugs Inconsistencies MySQL 1 0 - - 1 0 0 0 5 6 TiDB 2 0 - - - - - - 7 2 SQLite 0 1 0 0 0 0 0 0 3 1 DuckDB 1 0 0 0 - - - - 1 0 OceanBase 0 1 0 0 - - - - 0 2 with fewer test cases, demonstrating its effectiveness and efficiency in detecting bugs and inconsistencies that violating SQL specifications. (a) MySQL (b) TiDB (c) SQLite (d) DuckDB (e) OceanBase (f) PostgreSQL Fig. 12: The memory consumption of SEMCONT with SQLancer. Memory Consumption.",
+      "context_after": "Memory usage is mainly influenced by query generation, execution, and result comparison, with query generation being the most memory-intensive. We report stabilized memory consumption, with detailed time-based changes provided in our technical report [35]. Results indicate that SEMCONT’s memory usage is comparable to the baseline. Memory consumption is low because we use small tables (a few hundred records). We do not test concurrent query execution, resulting in stable memory usage. Our goal is to detect compliance bugs in RDBMSs, not to evaluate performance. SEMCONT is designed for offline",
+      "section": "B Experiment results",
+      "page": 17,
+      "char_offset": 72428,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M36",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "13: Memory consumption of SEMCONT and SQLancer 1SELECT MOD( ’ −12 ’ ,-4); 2−−e x p e c t e d: 0 \", a c t u a l :−0 % Fig.",
+      "context_before": "t concurrent query execution, resulting in stable memory usage. Our goal is to detect compliance bugs in RDBMSs, not to evaluate performance. SEMCONT is designed for offline testing during RDBMS development, and its overhead is manageable. SEMCONT shows very low memory consumption and it requires far less test cases to detect a compliance issue, demonstrating its efficiency. Answer to RQ3: SEMCONT detected more bugs and inconsistencies than the compared baseline approaches. Baseline approaches fail to detect those bugs since they do not refer to SQL specification in their testing process. Fig.",
+      "context_after": "14: A bug in MySQL 8.0.29 RQ4: Case study. SEMCONT has identified 19 bugs and 13 inconsistencies, which arise from two reasons, i.e., (1) DBMS implementations are not consistent with SQL specification and (2) unclear or missing description in the SQL specification. These problems have resulted in variations in the specific implementations across different RDBMSs, consequently affecting the user experience. A bug due to missing specifications. Figure 14 is a bug we detected in MySQL 8.0.29. The query conducts MOD function with the string type as the first parameter. The expected result of the",
+      "section": "B Experiment results",
+      "page": 17,
+      "char_offset": 73564,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": [],
+      "text_is_letter_spaced": true
+    },
+    {
+      "id": "M37",
+      "found_by": "citation_marker",
+      "surface": "[40]–[42]",
+      "technique": "norec",
+      "sentence": "Numerous testing methods involve constructing a test oracle by proposing a variety of metamorphic relations [21], [28], [40]–[42].",
+      "context_before": "rammar-based mutations. Furthermore, it boosts the accuracy of feedback through the use of binary coverage linking and bijective block mapping, thereby enhancing the overall performance of RDBMS testing. SparkFuzz [28] introduces a fuzzing-based method that utilizes the query results from a reference database as test oracles. The effectiveness of these differential testing methods is limited by the shared functionalities and syntax supported across the databases under test. Moreover, they may yield false positives due to the varied implementation choices inherent in different database systems.",
+      "context_after": "MUTASQL [21] and Eqsql [17] are tools that construct test cases by defining mutation rules. These rules are used to generate or synthesize SQL query statements that are functionally equivalent to the original ones. In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42]. PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data. The design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 83989,
+      "cited_reference": {
+        "number": 40,
+        "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non-optimizing reference engine construction,” in Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 2020, pp. 1140–1152.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M38",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42].",
+      "context_before": "of these differential testing methods is limited by the shared functionalities and syntax supported across the databases under test. Moreover, they may yield false positives due to the varied implementation choices inherent in different database systems. Numerous testing methods involve constructing a test oracle by proposing a variety of metamorphic relations [21], [28], [40]–[42]. MUTASQL [21] and Eqsql [17] are tools that construct test cases by defining mutation rules. These rules are used to generate or synthesize SQL query statements that are functionally equivalent to the original ones.",
+      "context_after": "PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data. The design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by verifying whether the returned result includes the specific row of data. NoREC [40] transforms an optimized SQL query into an equivalent non-optimized version and then compares the execution results of both. TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query. If the results of executing the original q",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 84335,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M39",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data.",
+      "context_before": "sitives due to the varied implementation choices inherent in different database systems. Numerous testing methods involve constructing a test oracle by proposing a variety of metamorphic relations [21], [28], [40]–[42]. MUTASQL [21] and Eqsql [17] are tools that construct test cases by defining mutation rules. These rules are used to generate or synthesize SQL query statements that are functionally equivalent to the original ones. In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42].",
+      "context_after": "The design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by verifying whether the returned result includes the specific row of data. NoREC [40] transforms an optimized SQL query into an equivalent non-optimized version and then compares the execution results of both. TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query. If the results of executing the original query differ from those obtained from the three divided queries, it likely indicates the presence of logical err",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 84501,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M40",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [40] transforms an optimized SQL query into an equivalent non-optimized version and then compares the execution results of both.",
+      "context_before": "ning mutation rules. These rules are used to generate or synthesize SQL query statements that are functionally equivalent to the original ones. In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42]. PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data. The design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by verifying whether the returned result includes the specific row of data.",
+      "context_after": "TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query. If the results of executing the original query differ from those obtained from the three divided queries, it likely indicates the presence of logical errors. SQLRight [34] focuses on enhancing the semantic correctness of generated SQL queries and adopts the oracles proposed by PQS [42], NoREC [40], and TLP [41]. GRIFFIN [27] executes mutation testing within the grammatical boundaries of SQL language, transforming data into metadata for this purpose. While metamorp",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 84792,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M41",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query.",
+      "context_before": "al ones. In recent years, SQLancer [14] has emerged as the most effective black-box fuzz testing tool, distinguished by its adoption of three complementary oracles [40]–[42]. PQS [42] operates by first selecting a row of data, and then synthesizing a query based on this selected data. The design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by verifying whether the returned result includes the specific row of data. NoREC [40] transforms an optimized SQL query into an equivalent non-optimized version and then compares the execution results of both.",
+      "context_after": "If the results of executing the original query differ from those obtained from the three divided queries, it likely indicates the presence of logical errors. SQLRight [34] focuses on enhancing the semantic correctness of generated SQL queries and adopts the oracles proposed by PQS [42], NoREC [40], and TLP [41]. GRIFFIN [27] executes mutation testing within the grammatical boundaries of SQL language, transforming data into metadata for this purpose. While metamorphic testing approaches address syntax differences arising from various database implementations, consistently returned results do n",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 84927,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M42",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "SQLRight [34] focuses on enhancing the semantic correctness of generated SQL queries and adopts the oracles proposed by PQS [42], NoREC [40], and TLP [41].",
+      "context_before": "design of the query is such that it must return the initially chosen row. Thisapproach detects bugs by verifying whether the returned result includes the specific row of data. NoREC [40] transforms an optimized SQL query into an equivalent non-optimized version and then compares the execution results of both. TLP [41] divides a SQL query into three separate SQL statements that collectively retain the same semantics as the original query. If the results of executing the original query differ from those obtained from the three divided queries, it likely indicates the presence of logical errors.",
+      "context_after": "GRIFFIN [27] executes mutation testing within the grammatical boundaries of SQL language, transforming data into metadata for this purpose. While metamorphic testing approaches address syntax differences arising from various database implementations, consistently returned results do not always guarantee the absence of bugs. Furthermore, these methods fall short in detecting bugs caused by violations of SQL specifications. Formal semantics for SQL. There have been several approaches [18], [24], [25], [36], [38], [48], [49] that made attempts to formalize the semantics of SQL. Chinaei [23] was",
+      "section": "VI RELATED WORK",
+      "page": 19,
+      "char_offset": 85216,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "uses_infrastructure": [
+        "M12"
+      ],
+      "compares_with": [
+        "M9",
+        "M26"
+      ],
+      "describes_as_state_of_the_art": [
+        "M4",
+        "M9",
+        "M26",
+        "M38"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:12:40Z",
+    "is_model_written": true,
+    "summary": "This paper introduces semantic conformance testing for relational DBMSs -- checking a system against the semantics of SQL rather than against another system. Two obstacles are identified: the standard is written in ambiguous natural language and is not executable, and it is hard to generate queries covering all its keywords and parameters. The authors formally define SQL's denotational semantics and implement them in Prolog as an executable reference, then add coverage criteria over those semantics and a coverage-guided generation algorithm. Across six widely used RDBMSs it uncovered 19 bugs and 13 inconsistencies, all confirmed.",
+    "narrative": "SEMCONT is built on SQLancer, whose syntax-guided generation it enhances with coverage guidance and uses as the seed pool for its own query generation. TLP and NoREC are then its baselines, described as state-of-the-art metamorphic testing methods and run in both their SQLancer and SQLRight implementations.",
+    "roles": {
+      "M3": "reuse_implementation",
+      "M4": "state_of_the_art",
+      "M5": "definition",
+      "M6": "definition",
+      "M7": "reuse_component",
+      "M9": "state_of_the_art",
+      "M10": "definition",
+      "M11": "definition",
+      "M12": "baseline",
+      "M13": "background",
+      "M14": "baseline",
+      "M18": "extension",
+      "M19": "result_comparison",
+      "M20": "result_comparison"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M3",
+          "M7",
+          "M12"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M3",
+            "sentence": "Here, we 10 enhance the syntax-guided generation method, SQLancer [14], with coverage-guided test case generation.",
+            "section": "A Overview of our approach",
+            "page": 9
+          },
+          {
+            "mention_id": "M7",
+            "sentence": "We adopt SQLancer to randomly generatea large number of SQL statements (line 2), which serves as the seed pool of our query generation algorithm.",
+            "section": "B Coverage guided test case generation",
+            "page": 11
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34].",
+            "section": "A Experiment setup",
+            "page": 14
+          }
+        ],
+        "reasoning": "M3 says the approach enhances SQLancer's syntax-guided generation with coverage guidance, and M7 that SQLancer is adopted to generate the seed pool the query generation works from.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "uncertain",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M18 describes adding keywords and generation rules SQLancer did not support, which extends the tool's generator rather than one of the named oracles. Nothing shown claims to extend PQS, NoREC or TLP."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M9",
+          "M12",
+          "M14"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M9",
+            "sentence": "We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+            "section": "A Experiment setup",
+            "page": 14
+          },
+          {
+            "mention_id": "M12",
+            "sentence": "Both NoREC and TLP are implemented in SQLancer [14] and SQLRight [34].",
+            "section": "A Experiment setup",
+            "page": 14
+          },
+          {
+            "mention_id": "M14",
+            "sentence": ", NoREC (SQLancer), NoREC (SQLRight), TLP (SQLancer) and TLP (SQLRight).",
+            "section": "A Experiment setup",
+            "page": 14
+          }
+        ],
+        "reasoning": "TLP and NoREC are the baselines, run in both their SQLancer and SQLRight implementations.",
+        "techniques": [
+          "tlp",
+          "norec"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M4",
+          "M9"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M4",
+            "sentence": "Coverage guided test case generation The state-of-the-art practice in test case generation involves randomly generating SQL queries guided by the syntax of SQL, among which SQLancer [14] stands out as one of the most effective tools of this kind.",
+            "section": "B Coverage guided test case generation",
+            "page": 10
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "We compared SEMCONT with TLP [41] and NoREC [40], which are state-of-the-art metamorphic testing methods for testing RDBMS.",
+            "section": "A Experiment setup",
+            "page": 14
+          }
+        ],
+        "reasoning": "M9 calls TLP and NoREC state-of-the-art metamorphic testing methods for RDBMSs, and M4 describes syntax-guided random generation as the state-of-the-art practice."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_14778_3749646_3749661.json b/_data/papers/paper_doi_10_14778_3749646_3749661.json
new file mode 100644
index 0000000..9d55a5f
--- /dev/null
+++ b/_data/papers/paper_doi_10_14778_3749646_3749661.json
@@ -0,0 +1,424 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:31:57Z",
+  "paper": {
+    "id": "paper:doi:10.14778/3749646.3749661",
+    "title": "PBench: Workload Synthesizer with Real Statistics for Cloud Analytics Benchmarking",
+    "authors": [
+      "Yan Zhou",
+      "Chunwei Liu",
+      "B. Urgaonkar",
+      "Zhengle Wang",
+      "M. Mueller",
+      "Chao Zhang",
+      "Songyue Zhang",
+      "Pascal Pfeil",
+      "Dominik Horn",
+      "Zhengchun Liu",
+      "Davide Pagano",
+      "Tim Kraska",
+      "Samuel Madden",
+      "Ju Fan"
+    ],
+    "year": 2025,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3749646.3749661",
+    "arxiv_id": "2506.16379",
+    "s2_paper_id": "f174394f4819f1387b1405ee41f9e7219b325573",
+    "url": "https://doi.org/10.14778/3749646.3749661",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2506.16379",
+      "retrieved_at": "2026-09-08T14:31:57Z",
+      "chars": 99344,
+      "content_sha256": "sha256:8b817b120937b7a8a506f7590be14105bfec896edebca37bacd3d117961d6e34"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 3059
+      },
+      {
+        "number": "2",
+        "title": "PRELIMINARIES",
+        "start": 12734
+      },
+      {
+        "number": "2.1",
+        "title": "Cloud Workload Traces",
+        "start": 13043
+      },
+      {
+        "number": "2.2",
+        "title": "Problem Formalization",
+        "start": 18430
+      },
+      {
+        "number": "3.1",
+        "title": "Overall Framework of PBench",
+        "start": 23797
+      },
+      {
+        "number": "3.1.2",
+        "title": "Two-Phase Workload Synthesis. Given a target workload",
+        "start": 25804
+      },
+      {
+        "number": "3.2",
+        "title": "Synthesizer Data Preparation",
+        "start": 28479
+      },
+      {
+        "number": "3.2.1",
+        "title": "Workload Trace Preparation. The statistics collected from",
+        "start": 28681
+      },
+      {
+        "number": "3.2.2",
+        "title": "Workload Component Preparation. To ensure a diverse set",
+        "start": 30610
+      },
+      {
+        "number": "4.1",
+        "title": "Workload Component Selection Phase",
+        "start": 32143
+      },
+      {
+        "number": "4.1.1",
+        "title": "Optimization Objective. To synthesize a workload with real",
+        "start": 32182
+      },
+      {
+        "number": "4.1.2",
+        "title": "Programming Constraints. To obtain a synthetic workload",
+        "start": 33236
+      },
+      {
+        "number": "4.2",
+        "title": "Timestamp Assignment Phase",
+        "start": 35444
+      },
+      {
+        "number": "4.2.2",
+        "title": "Simulated Annealing. To achieve a fine-grained generation,",
+        "start": 60036
+      },
+      {
+        "number": "4.3",
+        "title": "Discussion: Query Level Matching",
+        "start": 61812
+      },
+      {
+        "number": "6.1",
+        "title": "Experiment Setup",
+        "start": 71038
+      },
+      {
+        "number": "6.3",
+        "title": "Ablation Study of PBench",
+        "start": 80191
+      },
+      {
+        "number": "6.4",
+        "title": "Evaluation of PBench on Redset",
+        "start": 82805
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Nikos Armenatzoglou, Sanuj Basu, Naga Bhanoori, Mengchu Cai, Naresh Chainani, Kiran Chinta, Venkatraman Govindaraju, Todd J Green, Monish Gupta, Sebastian Hillig, et al .2022. Amazon Redshift re-invented. In Proceedings of the 2022 International Conference on Management of Data. 2205–2217.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. Proceedings of the ACM on Management of Data 2, 3 (2024), 1–26.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Peter Boncz, Angelos-Christos Anatiotis, and Steffen Kläbe. 2018. JCC-H: Adding Join Crossing Correlations with Skew to TPC-H. In Performance Evaluation and Benchmarking for the Analytics Era, Raghunath Nambiar and Meikel Poess (Eds.). Springer International Publishing, Cham, 103–119.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Peter Boncz, Thomas Neumann, and Orri Erling. 2013. TPC-H analyzed: Hidden messages and lessons learned from an influential benchmark. In Technology Conference on Performance Evaluation and Benchmarking. Springer, 61–76.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Nicolas Bruno, Surajit Chaudhuri, and Dilys Thomas. 2006. Generating queries with cardinality constraints for dbms testing. IEEE Transactions on Knowledge and Data Engineering 18, 12 (2006), 1721–1725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Brian F Cooper, Adam Silberstein, Erwin Tam, Raghu Ramakrishnan, and Russell Sears. 2010. Benchmarking cloud serving systems with YCSB. In Proceedings of the 1st ACM symposium on Cloud computing. 143–154.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Benoit Dageville, Thierry Cruanes, Marcin Zukowski, Vadim Antonov, Artin Avanes, Jon Bock, Jonathan Claybaugh, Daniel Engovatov, Martin Hentschel, Jiansheng Huang, et al .2016. The snowflake elastic data warehouse. In Proceedings of the 2016 International Conference on Management of Data. 215–226.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Shaleen Deep, Anja Gruenheid, Kruthi Nagaraj, Hiro Naito, Jeff Naughton, and Stratis Viglas. 2021. Diametrics: benchmarking query engines at scale. ACMSIGMOD Record 50, 1 (2021), 24–31.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Akon Dey, Alan Fekete, Raghunath Nambiar, and Uwe Röhm. 2014. YCSB+ T: Benchmarking web-scale transactional databases. In 2014 IEEE 30th International Conference on Data Engineering Workshops. IEEE, 223–230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Haowen Dong, Chao Zhang, Guoliang Li, and Huanchen Zhang. 2024. CloudNative Databases: A Survey. IEEE Transactions on Knowledge and Data Engineering(2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Markus Dreseler, Martin Boissier, Tilmann Rabl, and Matthias Uflacker. 2020. Quantifying TPC-H choke points and their optimizations. Proceedings of the VLDB Endowment 13, 8 (2020), 1206–1220.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "John Forrest and Robin Lougee-Heimer. [n.d.]. CBC User Guide. https://www. coin-or.org/Cbc/. 2024/12/13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Peter I Frazier. 2018. A tutorial on Bayesian optimization. arXiv preprint arXiv:1807.02811 (2018).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Jacob R Gardner, Matt J Kusner, Zhixiang Eddie Xu, Kilian Q Weinberger, and John P Cunningham. 2014. Bayesian optimization with inequality constraints.. InICML, Vol. 2014. 937–945.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "IMDb. 2024. IMDb Non-Commercial Datasets. https://developer.imdb.com/noncommercial-datasets/. 2024/12/13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2015. How good are query optimizers, really? Proceedings of the VLDB Endowment 9, 3 (2015), 204–215.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Viktor Leis and Maximilian Kuschewski. 2021. Towards cost-optimal query processing in the cloud. Proceedings of the VLDB Endowment 14, 9 (2021), 1606– 1612.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Guoliang Li, Haowen Dong, and Chao Zhang. 2022. Cloud databases: New techniques, challenges, and opportunities. Proceedings of the VLDB Endowment 15, 12 (2022), 3758–3761.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Guoliang Li, Xuanhe Zhou, and Lei Cao. 2021. Machine learning for databases. InProceedings of the First International Conference on AI-ML Systems. 1–2.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Zhaodonghui Li, Haitao Yuan, Huiming Wang, Gao Cong, and Lidong Bing. 2024. LLM-R2: A Large Language Model Enhanced Rule-based Rewrite System for Boosting Query Efficiency. arXiv preprint arXiv:2404.12872 (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Microsoft. 2024. DSB benchmark. https://github.com/microsoft/dsb. 2024/12/13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. 2008. Generating targeted queries for database testing. In Proceedings of the 2008 ACMSIGMOD international conference on Management of data. 499–510.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Magnus Müller, Lucas Woltmann, and Wolfgang Lehner. 2023. Enhanced Featurization of Queries with Mixed Combinations of Predicates for ML-based Cardinality Estimation.. In EDBT. 273–284.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Fernando Nogueira. 2014–. Bayesian Optimization: Open source constrained global optimization tool for Python. https://github.com/bayesian-optimization/ BayesianOptimization",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Meikel Poess, Raghunath Othayoth Nambiar, and David Walrath. 2007. Why You Should Run TPC-DS: A Workload Analysis.. In VLDB, Vol. 7. 1138–1149.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Prometheus. 2024. Prometheus Github Page. https://github.com/prometheus. 2024/12/13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Gaurav Saxena, Mohammad Rahman, Naresh Chainani, Chunbin Lin, George Caragea, Fahim Chowdhury, Ryan Marcus, Tim Kraska, Ippokratis Pandis, and Balakrishnan (Murali) Narayanaswamy. 2023. Auto-WLM: Machine Learning Enhanced Workload Management in Amazon Redshift. In Companion of the 2023 International Conference on Management of Data (Seattle, WA, USA) (SIGMOD ’23). Association for Computing Machine",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Dana Van Aken, Andrew Pavlo, Geoffrey J Gordon, and Bohan Zhang. 2017. Automatic database management system tuning through large-scale machine learning. In Proceedings of the 2017 ACM international conference on management of data. 1009–1024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Alexander van Renen, Dominik Horn, Pascal Pfeil, Kapil Vaidya, Wenjian Dong, Murali Narayanaswamy, Zhengchun Liu, Gaurav Saxena, Andreas Kipf, and Tim Kraska. 2024. Why TPC is not enough: An analysis of the Amazon Redshift fleet. Proceedings of the VLDB Endowment 17, 11 (2024), 3694–3706.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Alexander Van Renen and Viktor Leis. 2023. Cloud analytics benchmark. Proceedings of the VLDB Endowment 16, 6 (2023), 1413–1425.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Midhul Vuppalapati, Justin Miron, Rachit Agarwal, Dan Truong, Ashish Motivala, and Thierry Cruanes. 2020. Building an elastic query engine on disaggregated storage. In 17th USENIX Symposium on Networked Systems Design and Implementation (NSDI 20). 449–462.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Richard J. Wagner. 2024. Python module for simulated annealing. https://github. com/perrygeo/simanneal. 2024/12/13.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Chengcheng Wan, Yiwen Zhu, Joyce Cahoon, Wenjing Wang, Katherine Lin, Sean Liu, Raymond Truong, Neetu Singh, Alexandra M. Ciortea, Konstantinos Karanasos, and Subru Krishnan. 2023. Stitcher: Learned Workload Synthesis from Historical Performance Footprints. In EDBT. OpenProceedings.org, 417–423. https://doi.org/10.48786/EDBT.2023.33",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Shaoyi Yin, Franck Morvan, Jorge Martinez-Gil, and Abdelkader Hameurlain. 2023. MTD-DS: an SLA-aware Decision Support Benchmark for Multi-tenant Parallel DBMSs. Ph.D. Dissertation. IRIT-Institut de Recherche en Informatique de Toulouse.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Chao Zhang, Guoliang Li, and Tao Lv. 2024. HyBench: A New Benchmark for HTAP Databases. Proceedings of the VLDB Endowment 17, 5 (2024), 939–951.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Chao Zhang, Guoliang Li, Jintao Zhang, Xinning Zhang, and Jianhua Feng. 2024. HTAP Databases: A Survey. IEEE Transactions on Knowledge and Data Engineering (2024).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Huanchen Zhang, Yihao Liu, and Jiaqi Yan. 2023. Cost-Intelligent Data Analytics in the Cloud. arXiv preprint arXiv:2308.09569 (2023).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Lixi Zhang, Chengliang Chai, Xuanhe Zhou, and Guoliang Li. 2022. Learnedsqlgen: Constraint-aware sql generation using reinforcement learning. In Proceedings of the 2022 International Conference on Management of Data. 945–958. 13",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. Proceedings of the ACM on Management of Data 2, 3 (2024), 1–26.",
+      "technique": "dqp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing dqp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 2]",
+      "technique": "dqp",
+      "sentence": "Then, the generated synthetic workload offers an effective solution for database benchmarking [ 8,35,36], performance tuning [ 28], and bug detection [ 2], allowing database developers to perform realistic evaluations while preserving customer privacy and avoiding exposure of sensitive workloads or data.",
+      "context_before": "ith real statistics. Given a real workload trace where original SQL queries and underlying databases are unavailable, the problem is to generate a synthetic workload that, when executed, produces statistics 3 closely approximate the target workload’s performance metrics and operator distributions. In particular, given that real workload traces lack the original SQL queries and their underlying databases, this work focuses on constructing synthetic workloads by combining workload components, i.e., SQL queries and their corresponding databases, from existing benchmarks, such as TPC-H and TPC-DS.",
+      "context_after": "This paper considers two settings to workload synthesis: querylevel andwindow-level. The query-level setting synthesizes a workload by generating a corresponding query for each original query, preserving the number of queries and their arrival times. However, this method may introduce significant computational overhead and higher approximation errors when identifying suitable queries for synthesis. In contrast, the window-level setting aims to approximate the aggregated statistics of all customer queries within a given time window, offering greater flexibility and reducing overall errors comp",
+      "section": "2.2 Problem Formalization",
+      "page": 4,
+      "char_offset": 19112,
+      "cited_reference": {
+        "number": 2,
+        "text": "Jinsheng Ba and Manuel Rigger. 2024. Keep It Simple: Testing Databases via Differential Query Plans. Proceedings of the ACM on Management of Data 2, 3 (2024), 1–26.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing dqp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "dqp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "PBench synthesises analytics workloads that reproduce the execution statistics of real cloud workloads. Standard benchmarks use fixed query patterns and miss production behaviour, while released traces lack the queries and databases needed to be benchmarks. PBench selects and combines components from existing benchmarks and augments them with new ones, using multi-objective optimisation, timestamp assignment for temporal dynamics, and LLM-generated components. It cuts approximation error by up to 6x.",
+    "narrative": "One citation, listing database benchmarking among the uses a synthetic workload serves, with SQLancer's work among the references.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_14778_3797919_3797928.json b/_data/papers/paper_doi_10_14778_3797919_3797928.json
new file mode 100644
index 0000000..fef020c
--- /dev/null
+++ b/_data/papers/paper_doi_10_14778_3797919_3797928.json
@@ -0,0 +1,885 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:31:24Z",
+  "paper": {
+    "id": "paper:doi:10.14778/3797919.3797928",
+    "title": "Dinkel: State-Aware and Granular Framework for Validating Graph Databases",
+    "authors": [
+      "C. Wüst",
+      "Zu-Ming Jiang",
+      "Zhendong Su"
+    ],
+    "year": 2024,
+    "venue": "Proceedings of the VLDB Endowment",
+    "doi": "10.14778/3797919.3797928",
+    "arxiv_id": "2408.07525",
+    "s2_paper_id": "10be4636a8900a45571f1744ee50430a4e125485",
+    "url": "https://doi.org/10.14778/3797919.3797928",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2408.07525",
+      "retrieved_at": "2026-09-09T01:31:24Z",
+      "chars": 75747,
+      "content_sha256": "sha256:b3097640feec05ec2ca8efb7aba4bbc4f811e1eb948f942c6f4cd2e7a8b28607"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 13,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 1792
+      },
+      {
+        "number": "2",
+        "title": "BACKGROUND",
+        "start": 14675
+      },
+      {
+        "number": "3",
+        "title": "APPROACH",
+        "start": 19031
+      },
+      {
+        "number": "3.1",
+        "title": "State-Aware Query Generation",
+        "start": 19212
+      },
+      {
+        "number": "3.1.2",
+        "title": "Query Generation.Our generation is based on a key insight",
+        "start": 21905
+      },
+      {
+        "number": "3.2",
+        "title": "Fine-Grained Query Transformation",
+        "start": 27024
+      },
+      {
+        "number": "4",
+        "title": "IMPLEMENTATION",
+        "start": 33020
+      },
+      {
+        "number": "5",
+        "title": "EVALUATION",
+        "start": 36569
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Setup",
+        "start": 36966
+      },
+      {
+        "number": "5.2",
+        "title": "Bug Detection",
+        "start": 38092
+      },
+      {
+        "number": "5.3",
+        "title": "Query Generation",
+        "start": 45810
+      },
+      {
+        "number": "5.4",
+        "title": "Importance of Key Techniques",
+        "start": 49485
+      },
+      {
+        "number": "5.5",
+        "title": "Comparison",
+        "start": 54649
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Address Sanitizer. 2019. https://clang.llvm.org/docs/AddressSanitizer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "American Fuzzy Lop. 2021. https://github.com/google/AFL.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Vaggelis Atlidakis, Patrice Godefroid, and Marina Polishchuk. 2019. RESTler: Stateful REST API Fuzzing. InProceedings of the 41st International Conference on Software Engineering (ICSE). 748–758.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Jingji Chen and Xuehai Qian. 2023. DecoMine: A Compilation-Based Graph Pattern Mining System with Pattern Decomposition. InProceedings of the 28th International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS). 47–61.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Peng Chen and Hao Chen. 2018. Angora: Efficient Fuzzing by Principled Search. InProceedings of the 2018 Symposium on Security and Privacy (S&P). 711–725.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Tsong Y Chen, Shing C Cheung, and Shiu Ming Yiu. 1998.Metamorphic testing: a new approach for generating next test cases. Technical Report. HKUST-CS98-01. Department of Computer Science, HKUST.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Creating a Modern Solution to Finding What’s Lost with Neo4j. 2024. https: //neo4j.com/case-studies/notlost.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Cypher Query Language Reference. 2022. https://s3.amazonaws.com/artifacts. opencypher.org/openCypher9.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "DB-Engines Ranking. 2024. https://db-engines.com/en/ranking.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "FalkorDB. A super fast Graph Database uses GraphBLAS under the hood for its sparse adjacency matrix graph representation. 2024. https://github.com/Falko rDB/FalkorDB/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. 2022. Griffin: Grammar-Free DBMS Fuzzing. InProceedings of the 37th International Conference on Automated Software Engineering (ASE). 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Zongyin Hao, Quanfeng Huang, Chengpeng Wang, Jianfeng Wang, Yushan Zhang, Rongxin Wu, and Charles Zhang. 2023. Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis. InProceedings of the 2023 USENIX Annual Technical Conference (ATC). 345–358.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Ziyue Hua, Wei Lin, Luyao Ren, Zongyang Li, Lu Zhang, Wenpin Jiao, and Tao Xie. 2023. GDsmith: Detecting Bugs in Cypher Graph Database Engines. In Proceedings of the 2023 International Symposium on Software Testing and Analysis (ISSTA). 163–174.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICSE). 531–542.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 15,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, Kangjie Lu, and Shi-Min Hu. 2020. Fuzzing Error Handling Code using Context-Sensitive Software Fault Injection. InProceedings of the 29th USENIX Security Symposium. 2595–2612.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. 2023. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. InProceedings of the 32nd USENIX Security Symposium. 4949–4965.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In Proceedings of the 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI). 397–417.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 18,
+      "text": "Zu-Ming Jiang and Zhendong Su. 2024. Detecting Logic Bugs in Database Engines via Equivalent Expression Transformation. InProceedings of the 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI). 821–835.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, Kangjie Lu, and Shi-Min Hu. 2022. Context-Sensitive and Directional Concurrency Fuzzing for Data-Race Detection. InProceedings of the 29th Network and Distributed System Security Symposium (NDSS).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. InProceedings of the 2023 International Symposium on Software Testing and Analysis (ISSTA). 140–149.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Kyungtae Kim, Taegyu Kim, Ertza Warraich, Byoungyoung Lee, Kevin RB Butler, Antonio Bianchi, and Dave Jing Tian. 2022. FuzzUSB: Hybrid Stateful Fuzzing of USB Gadget Stacks. InProceedings of the 2022 International Symposium on Security and Privacy (S&P). 2212–2229.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. 2023. SequenceOriented DBMS Fuzzing. InProceedings of the 2023 International Conference on Data Engineering (ICDE). 668–681.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Chenyang Lyu, Shouling Ji, Chao Zhang, Yuwei Li, Wei-Han Lee, Yu Song, and Raheem Beyah. 2019. MOPT: Optimized Mutation Scheduling for Fuzzers. In Proceedings of the 28th USENIX Security Symposium. 1949–1966.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Fuchen Ma, Yuanliang Chen, Meng Ren, Yuanhang Zhou, Yu Jiang, Ting Chen, Huizhong Li, and Jiaguang Sun. 2023. LOKI: State-Aware Fuzzing Framework for the Implementation of Blockchain Consensus Protocols. InProceedings of the 30th Network and Distributed System Security Symposium (NDSS).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "David Maciver and Alastair F. Donaldson. 2020. Test-Case Reduction via TestCase Generation: Insights from the Hypothesis Reducer. InProceedings of the 34th European Conference on Object-Oriented Programming (ECOOP). 13:1–13:27.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Qiuyang Mang, Aoyang Fang, Boxi Yu, Hanfei Chen, and Pinjia He. 2024. Testing Graph Database Systems via Equivalent Query Rewriting. InProceedings of the 46th International Conference on Software Engineering (ICSE). 1–12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Qiuyang Mang, Aoyang Fang, Boxi Yu, Hanfei Chen, and Pinjia He. 2024. Testing Graph Database Systems via Equivalent Query Rewriting. InProceedings of the 46th International Conference on Software Engineering (ICSE). 143:1–143:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "William M McKeeman. 1998. Differential testing for software.Digital Technical Journal10, 1 (1998), 100–107.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Memgraph. Open-source graph database, tuned for dynamic analytics environments. Easy to adopt, scale and own. 2024. https://github.com/memgraph/me mgraph.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "NBC News Analyzes Hundreds of Thousands of Russian Troll Tweets Using Neo4j. 2021. https://go.neo4j.com/rs/710-RRC-335/images/Neo4j-case-study NBC-News-EN-US.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Neo4j and Generative AI 2023. https://neo4j.com/generativeai/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Neo4j. Graphs for Everyone. 2024. https://github.com/neo4j/neo4j.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Novartis Captures the Latest Biological Knowledge for Drug Discovery. 2021. https://go.neo4j.com/rs/710-RRC-335/images/Neo4j-case-study-Novartis-ENUS.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "openCypher. 2024. https://opencypher.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "RedisGraph. A graph database as a Redis module. 2023. https://github.com/Red isGraph/RedisGraph.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "John Regehr, Yang Chen, Pascal Cuoq, Eric Eide, Chucky Ellison, and Xuejun Yang. 2012. Test-Case Reduction for C Compiler Bugs. InProceedings of the 2012 International Conference on Programming Language Design and Implementation (PLDI). 335–346.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESE/FSE). 1140–1152.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. InProceedings of the 2020 International Conference on Object Oriented Programming Systems Languages and Applications (OOPSLA). 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 39,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. InProceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI). 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 40,
+      "text": "Jiansen Song, Wensheng Dou, Ziyu Cui, Qianwang Dai, Wei Wang, Jun Wei, Hua Zhong, and Tao Huang. 2023. Testing Database Systems via Differential Query Execution. InProceedings of the 45th International Conference on Software Engineering (ICSE). 2072–2084.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "SQL standard. 1992. https://www.contrib.andrew.cmu.edu/~shadow/sql/sql1992. txt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "SQLsmith: A random SQL query generator. 2023. https://github.com/anse1/sqls mith.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Xiu Tang, Sai Wu, Dongxiang Zhang, Feifei Li, and Gang Chen. 2023. Detecting Logic Bugs of Join Optimizations in DBMS. InProceedings of the 2023 International Conference on Management of Data (SIGMOD). 1–26.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Who Uses Neo4j. 2024. https://neo4j.com/who-uses-neo4j.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Chengshuo Xu, Keval Vora, and Rajiv Gupta. 2019. Pnp: Pruning and prediction for point-to-point iterative graph analytics. InProceedings of the 24th International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS). 587–600.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Xizhe Yin, Zhijia Zhao, and Rajiv Gupta. 2023. Glign: Taming misaligned graph traversals in concurrent graph processing. InProceedings of the 28th International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS). 78–92.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Insu Yun, Sangho Lee, Meng Xu, Yeongjin Jang, and Taesoo Kim. 2018. QSYM: A Practical Concolic Execution Engine Tailored for Hybrid Fuzzing. InProceedings of the 27th USENIX Security Symposium. 745–761.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Bodong Zhao, Zheming Li, Shisong Qin, Zheyu Ma, Ming Yuan, Wenyu Zhu, Zhihong Tian, and Chao Zhang. 2022. StateFuzz: System Call-Based State-Aware Linux Driver Fuzzing. InProceedings of the 31st USENIX Security Symposium. 3273–3289.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. 2022. Finding Bugs in Gremlin-Based Graph Database Systems via Randomized Differential Testing. InProceedings of the 2022 International Symposium on Software Testing and Analysis (ISSTA). 302–313.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. 2020. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. InProceedings of the 2020 International Conference on Computer and Communications Security (CCS). 955–970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Zeyang Zhuang, Penghui Li, Pingchuan Ma, Wei Meng, and Shuai Wang. 2023. Testing Graph Database Systems via Graph-Aware Metamorphic Relations.Proceedings of the 50th International Conference on Very Large Databases (VLDB) (2023), 836–848.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 14,
+      "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICSE). 531–542.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 17,
+      "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In Proceedings of the 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI). 397–417.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 20,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. InProceedings of the 2023 International Symposium on Software Testing and Analysis (ISSTA). 140–149.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESE/FSE). 1140–1152.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 38,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Finding Bugs in Database Systems via Query Partitioning. InProceedings of the 2020 International Conference on Object Oriented Programming Systems Languages and Applications (OOPSLA). 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 39,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Testing Database Engines via Pivoted Query Synthesis. InProceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI). 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 13,14,20,51]",
+      "technique": null,
+      "sentence": "To improve GDBMS reliability, testing approaches [ 13,14,20,51] have been proposed to find bugs by generating queries in Cypher, the most widely-adopted graph query language [ 34].",
+      "context_before": "ems (GDBMSs) are crucial for modern interconnected, data-driven computer software [ 4,7,30, 31,33,45,46]. Notably, 75% of the Fortune 100 and all of North America’s top 20 banks have adopted the currently most popular GDBMS, Neo4j [ 9,44]. While rapidly evolving and becoming more complex (e.g., Neo4j has 1.2M LOC), GDBMSs inevitably suffer from bugs introduced during their development. These bugs are critical as they can corrupt the GDBMSs and lead to serious consequences. One of the most critical types of bugs are logic bugs, which can silently cause GDBMSs to produce incorrect query results.",
+      "context_after": "As large-scale, complex systems, GDBMSs are difficult to test effectively due to two key challenges:query generationandtest-oracle construction. To cover a broad spectrum of functionalities and deep system logic, approaches need to generate complex queries involving various Cypher language features and complicated data dependencies. To ∗Both authors contributed equally to this research.identify bugs, especially logic bugs that cause GDBMSs to silently produce incorrect results for queries, testing approaches require test oracles that can be generally applied to validate arbitrary queries. How",
+      "section": "1 INTRODUCTION",
+      "page": 1,
+      "char_offset": 2437,
+      "cited_reference": {
+        "number": 14,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICS",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 16–18]",
+      "technique": null,
+      "sentence": "In DBMSs, many bugs can only be triggered by complex queries [ 16–18].",
+      "context_before": "llenges remain unsolved. All existing approaches focus on proposing new test oracles, but struggle to systematically generate complex and valid queries, leaving much GDBMS logic and code inadequately tested or completely unexercised. Although existing approaches focus on test oracles to identify logic bugs, none of the proposed oracles can generally be applied to validate arbitrary queries. As a result, many logic bugs in GDBMSs get missed, even though they can be triggered by existing queries. The following discusses the details of these two challenges. Challenge 1:Generating Complex Queries.",
+      "context_after": "A complex Cypher query can involve various clause combinations and complicated data dependencies. Generating such complex queries is challenging because the clauses invoked by Cypher queries can change the graph states of the manipulated graph database during their execution. Such state changes are visible in subsequent clauses. Moreover, the data and variables used by Cypher queries have different scopes, depending on the clauses they are involved in. Due to the lack of systematic modeling of these Cypher language features, existing approaches cannot effectively generate complex queries, and",
+      "section": "1 INTRODUCTION",
+      "page": 1,
+      "char_offset": 3841,
+      "cited_reference": {
+        "number": 17,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In Proceedings of the 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI). 397–417.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 14]",
+      "technique": null,
+      "sentence": "For example, GraphGenie [ 14] requires the predicate in MATCH clauses to satisfy the preconditions of their transformation rules (e.",
+      "context_before": "H (n0 IN[]| FOREACH (xIN[]| SET x={})) RETURN 0ASn1 UNION CALL{ FOREACH(n2IN[] | FOREACH(n3IN[] | MERGE(:A {u:0}))) }MATCH(y) RETURNyASn1 UNION CREATE() RETURN0ASn1 Figure 2: Queries exposing a logic bug in Neo4j. 26,51]. Given a generated Cypher query, they transform it into another query and check whether the results of the transformed query and the original query follow specified relationships. However, these approaches cannot be generally applied to arbitrary queries because they require their generated queries to follow specific query patterns so that their transformations can be applied.",
+      "context_after": "g., one rule requires the predicates to involve a cyclic graph). Moreover, GraphGenie can transform and validate only queries that contain a single MATCH clause followed by a RETURN. GRev [ 26] integrates an abstract syntax graph (ASG) for interpreting the predicate in MATCH statements and can thus transform the predicate in more diverse ways and with fewer restrictions. However, the predicate modeling in GRev can only be applied to MATCH statements, while queries involving other common clauses like MERGE ,UNWIND, and FOREACH remain unsupported and thus untested. Such inapplicabilities arise f",
+      "section": "1 INTRODUCTION",
+      "page": 2,
+      "char_offset": 7107,
+      "cited_reference": {
+        "number": 14,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICS",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 14]",
+      "technique": null,
+      "sentence": "The most widely used graph model is thelabeled property graph model[ 14], which stores interconnected data using nodes connected via relationships (i.",
+      "context_before": "queries by manipulating their clauses and expressions. •Practical realization:We realize a fully automatic testing framework, Dinkel, to find bugs in GDBMSs by generating and validating complex Cypher queries. •Promising results:We evaluated Dinkel on three popular opensource GDBMSs, namely Neo4j, RedisGraph and Memgraph. Dinkel found 127 unique and new bugs, among which 33 are logic bugs. So far, 113 unique bugs have been confirmed, and 84 fixed. These results demonstrate the effectiveness ofDinkel. 2 BACKGROUND Graph Database Management Systems. GDBMSs utilize graph models to represent data.",
+      "context_after": "e., directed edges between nodes) [ 34]. Nodes and relationships, generally referred to asgraph entities, can be associated with labels and properties. Labels are used to group and classify elements, whereas properties are made up of key-value pairs, providing attribute information. Cypher Language.Cypher is the most widely adopted query language for property graph databases [ 34] and was proposed by Neo4j. It is designed as a declarative query language, which allows users to specify the required data without realizing the detailed procedures. The general way to specify data in Cypher is to co",
+      "section": "2 BACKGROUND",
+      "page": 3,
+      "char_offset": 14769,
+      "cited_reference": {
+        "number": 14,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICS",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 13,14,20,27,51]",
+      "technique": null,
+      "sentence": "As shown in Table 4, Dinkel supports more Cypher clauses than existing approaches [ 13,14,20,27,51].",
+      "context_before": "pported Clauses in Query Generation. Dinkel supports all non-administrative clauses in the tested GDBMSs. For example, in Neo4j, 21 out of the total 25 clauses are non-administrative and thereby supported. Administrative clauses, such as SHOWFUNCTIONS and USE, are not supported because they are not related to query processing logic. For example, SHOWFUNCTIONS is designed to list all available functions and cannot be combined with other clauses. USE is used to switch the manipulated databases, which is not necessary as all databases are empty on initialization and they function in the same way.",
+      "context_after": "None of the existing approaches support FOREACH clauses, because this clause complicates control flow and can significantly modify graph states during query execution. UNION ,EXISTS, and COUNT are not supported either, as these clauses invoke subqueries that can inherit and change the graph states of the main query. Lacking a systematic model for handling graph state changes, existing approaches are limited in generating complex queries using these clauses. Moreover, existing approaches support validation for only queries following specific patterns, while many clauses complicate the semantic",
+      "section": "4 IMPLEMENTATION",
+      "page": 7,
+      "char_offset": 34250,
+      "cited_reference": {
+        "number": 14,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICS",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 13,14,20,27,51]",
+      "technique": null,
+      "sentence": "1 Experimental Setup We evaluated Dinkel on Neo4j [ 32], RedisGraph [ 35], and Memgraph [ 29], which are popular and extensively tested by existing approaches [ 13,14,20,27,51].",
+      "context_before": "f some clauses are successfully deleted or replaced in one try, Dinkel will restart the reduction process for the reduced queries. The process stops when no clause in the query can be further reduced. 5 EVALUATION Our evaluation aims to answer the following questions: Q1CanDinkel find real bugs in widely used and extensively tested GDBMSs? (Section 5.2) Q2How complex and valid are the queries generated by Dinkel? (Section 5.3) Q3How do different techniques contribute to the effectiveness of Dinkel? (Section 5.4) Q4CanDinkel outperform state-of-the-art GDBMS testing approaches? (Section 5.5) 5.",
+      "context_after": "We evaluate Dinkel on the latest GDBMS versions. During testing, if the tested GDBMSs are updated (e.g., a new version is released), we set up new Dinkel instances to test the updated versions. Specifically, we test Neo4j from version 5.6.0, RedisGraph from 2.12.0, and Memgraph from 2.7.0. RedisGraph is no longer maintained after 2.12.10, and thus we move to test its successor, FalkorDB [ 10], from its first release version 4.0.0. We count the bugs of RedisGraph and FalkorDB together as they share the majority of their code base. To show its effectiveness for black-box testing GDBMSs, we eval",
+      "section": "5.1 Experimental Setup",
+      "page": 7,
+      "char_offset": 36968,
+      "cited_reference": {
+        "number": 14,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICS",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[17, 25, 36]",
+      "technique": null,
+      "sentence": "2, because all the bugtriggering queries are reduced, while the generated queries inevitably contain many redundant parts [17, 25, 36].",
+      "context_before": "se Dep. Size Neo4j 318k 89.90% 40.97 26.77 1394.69 RedisGraph 1’078k 93.90% 32.71 30.59 1071.58 Memgraph 570k 83.27% 29.95 21.71 993.63 Average655k 89.02% 34.54 26.36 1153.3 dependencies is 26.36. Because of the large number of clauses and dependencies, the queries generated by Dinkel are typically large. The average query size is 1153.3 bytes. These results demonstrate thatDinkel can effectively generate complex queries, which are large and contain multiple clauses with complicated data dependencies. The generated queries are more complex than the bugtriggering queries discussed in Section 5.",
+      "context_after": "Query Validity. Among the 655k queries generated by Dinkel, 583k are valid. The percentage of valid queries is 89.02%. The result demonstrates that Dinkel can keep a high validity rate even when generating complex queries. We investigated the invalid queries generated by Dinkel and found that they were mainly caused by illegal arithmetic operations. Some arithmetic operations require the operands to satisfy some constraints (e.g., the divisor in a division must be non-zero). However, Dinkel cannot track the value of each operator in queries because the value may depend on complex expressions,",
+      "section": "5.3 Query Generation",
+      "page": 9,
+      "char_offset": 47276,
+      "cited_reference": {
+        "number": 17,
+        "text": "Zu-Ming Jiang, Si Liu, Manuel Rigger, and Zhendong Su. 2023. Detecting Transactional Bugs in Database Engines via Graph-Based Oracle Construction. In Proceedings of the 17th USENIX Symposium on Operating Systems Design and Implementation (OSDI). 397–417.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 14,20]",
+      "technique": null,
+      "sentence": "We believe the current throughput is practical considering (1) GDBMS testing typically lasts for several months [ 14,20], and thus a sufficient number of test cases can be executed; and (2) setting up multiple Dinkel instances can significantly improve the test efficiency.",
+      "context_before": "ong Su Table 9: Bug-triggering queries related to QC: only query context; GS: only graph schema; Both: both graph states. The number in the parentheses is for the logic bugs Version Found QCGS Both Neo4j 51 (13) 23 (3) 3 (0) 5 (2) RedisGraph 62 (18) 34 (7) 1 (1) 4 (0) Memgraph 14 (2) 5 (0) 1 (0) 0 (0) Total127 (33) 72 (10) 5 (1) 9 (2) transformation, and execution. We found that for Neo4j, 74.75% of CPU time was used to execute the generated queries, which are complex and can be time-consuming. This result indicates that the test throughput is dominated by the performance of the tested GDBMSs.",
+      "context_after": "5.4 Importance of Key Techniques To show the importance of our techniques, we analyze how they contribute to the 127 bugs found by Dinkel. The analysis consists of two parts. The first focuses on the query context and graph schema in query generation, while the second assesses the transformation strategies (4 clause-level ones and 2 expression-level ones) used in test-oracle construction. We analyzed the 127 bug-triggering queries and checked whether their generations depend on specific graph states. Specifically, we extracted the data dependencies contained by each query and check whether at",
+      "section": "5.3 Query Generation",
+      "page": 10,
+      "char_offset": 49210,
+      "cited_reference": {
+        "number": 14,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICS",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 13,14,20,27,51]",
+      "technique": null,
+      "sentence": "This comparison is reasonable and objective because: (1) none of the bugs found by Dinkel are marked as duplicated by developers, meaning that no approach found these bugs until Dinkel found them; and (2) all existing approaches have extensively tested Neo4j and RedisGraph [ 13,14,20,27,51], meaning that in these two GDBMSs, no approach found the long-latent bugs found by Dinkel during their evaluation.",
+      "context_before": "bugs DBMS FoundBug-involved year Longest latency < 2024 < 2023 Neo4j 51 (13) 40 (9) 27 (7) 2016 RedisGraph 62 (18) 52 (15) 20 (3) 2020 Memgraph 14 (2) 9 (2) 4 (0) 2021 Total127 (33) 101 (26) 51 (10) 2016 5.5 Comparison Bug Latency Study.To demonstrate that Dinkel can find bugs missed by existing approaches, we investigate the latencies of bugs found by Dinkel. For each bug, we check whether its bug-inducing commit was created before the years when existing approaches were published. If Dinkel finds some long-latent bugs, we can conclude that Dinkel can find bugs missed by existing approaches.",
+      "context_after": "Some existing approaches do not test Memgraph, but we still include it to show the effectiveness ofDinkel. GDsmith [ 13], GDBMeter [ 20], GraphGenie [ 14], and GAMERA [ 51] were published in 2023, and GRev [ 27] was published in 2024. Therefore, for each bug, we checked whether its bug-inducing commit was created before 2024 (i.e., in 2023 or earlier) and 2023. As shown in Table 11, among the 113 bugs in Neo4j and RedisGraph, 47 already existed before 2023, among which 10 are logic bugs. It indicates that all existing approaches failed to find these 47 bugs in their extensive evaluation. In a",
+      "section": "5.5 Comparison",
+      "page": 11,
+      "char_offset": 55044,
+      "cited_reference": {
+        "number": 14,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICS",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 20]",
+      "technique": null,
+      "sentence": "GDsmith [ 13], GDBMeter [ 20], GraphGenie [ 14], and GAMERA [ 51] were published in 2023, and GRev [ 27] was published in 2024.",
+      "context_before": "-latent bugs, we can conclude that Dinkel can find bugs missed by existing approaches. This comparison is reasonable and objective because: (1) none of the bugs found by Dinkel are marked as duplicated by developers, meaning that no approach found these bugs until Dinkel found them; and (2) all existing approaches have extensively tested Neo4j and RedisGraph [ 13,14,20,27,51], meaning that in these two GDBMSs, no approach found the long-latent bugs found by Dinkel during their evaluation. Some existing approaches do not test Memgraph, but we still include it to show the effectiveness ofDinkel.",
+      "context_after": "Therefore, for each bug, we checked whether its bug-inducing commit was created before 2024 (i.e., in 2023 or earlier) and 2023. As shown in Table 11, among the 113 bugs in Neo4j and RedisGraph, 47 already existed before 2023, among which 10 are logic bugs. It indicates that all existing approaches failed to find these 47 bugs in their extensive evaluation. In addition, 45 bugs (24 logic bugs) were introduced to Neo4j and RedisGraph between 2023 and 2024, while GRev, which was proposed after 2024, cannot find these bugs. In addition, GDsmith and GRev support Memgraph, but neither can find the",
+      "section": "5.5 Comparison",
+      "page": 11,
+      "char_offset": 55558,
+      "cited_reference": {
+        "number": 20,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. 2023. Testing Graph Database Engines via Query Partitioning. InProceedings of the 2023 International Symposium on Software Testing and Analysis (ISSTA). 140–149.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 13,14,20,27,49,51]",
+      "technique": null,
+      "sentence": "GDBMS testing is an emerging research field, where several approaches [ 13,14,20,27,49,51] have been proposed.",
+      "context_before": "ound in Neo4j and 7 in RedisGraph. In Neo4j, GDBMeter, GRev, and GAMERA found 1 bug. In RedisGraph, GRev and GDBMeter found 1 bug. Both of these bugs were also identified by Dinkel. For the 43 bugs missed by existing approaches, their bug-triggering queries contain either complicated data dependencies or advanced Cypher features that are not supported by existing approaches. The bugs shown in Figure 10 and Figure 11 are two of the 43 bugs. These results demonstrate that Dinkel can find more bugs than existing approaches by generating and validating complex queries. 6 RELATED WORKGDBMS Testing.",
+      "context_after": "All these approaches aim to construct test oracles to identify logic bugs. Grand [ 49] targets GDBMSs using the Gremlin query language and detects logic bugs using differential testing [ 28]. Several approaches [ 13,14,20,27,51] support GDBMSs using Cypher. Similar to Grand, GDsmith [ 13] also uses differential testing to find logic bugs. GDBMeter [ 20] leverages query partitioning [ 38], which decomposes the predicate of a query and checks if the queries with the decomposed predicates produce consistent results with the original query. GraphGenie [ 14] modifies the graph patterns used in a q",
+      "section": "5.5 Comparison",
+      "page": 12,
+      "char_offset": 60474,
+      "cited_reference": {
+        "number": 14,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICS",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 13,14,20,27,51]",
+      "technique": null,
+      "sentence": "Several approaches [ 13,14,20,27,51] support GDBMSs using Cypher.",
+      "context_before": "or advanced Cypher features that are not supported by existing approaches. The bugs shown in Figure 10 and Figure 11 are two of the 43 bugs. These results demonstrate that Dinkel can find more bugs than existing approaches by generating and validating complex queries. 6 RELATED WORKGDBMS Testing.GDBMS testing is an emerging research field, where several approaches [ 13,14,20,27,49,51] have been proposed. All these approaches aim to construct test oracles to identify logic bugs. Grand [ 49] targets GDBMSs using the Gremlin query language and detects logic bugs using differential testing [ 28].",
+      "context_after": "Similar to Grand, GDsmith [ 13] also uses differential testing to find logic bugs. GDBMeter [ 20] leverages query partitioning [ 38], which decomposes the predicate of a query and checks if the queries with the decomposed predicates produce consistent results with the original query. GraphGenie [ 14] modifies the graph patterns used in a query and checks if the query with modified graph patterns satisfies the expected relationship (i.e., equivalence, subset, or superset) with the original queries. GAMERA [ 51] extracts metamorphic relations [ 6] from the manipulated graph data and checks whet",
+      "section": "5.5 Comparison",
+      "page": 12,
+      "char_offset": 60776,
+      "cited_reference": {
+        "number": 14,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICS",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "query partitioning",
+      "technique": "tlp",
+      "sentence": "GDBMeter [ 20] leverages query partitioning [ 38], which decomposes the predicate of a query and checks if the queries with the decomposed predicates produce consistent results with the original query.",
+      "context_before": "esults demonstrate that Dinkel can find more bugs than existing approaches by generating and validating complex queries. 6 RELATED WORKGDBMS Testing.GDBMS testing is an emerging research field, where several approaches [ 13,14,20,27,49,51] have been proposed. All these approaches aim to construct test oracles to identify logic bugs. Grand [ 49] targets GDBMSs using the Gremlin query language and detects logic bugs using differential testing [ 28]. Several approaches [ 13,14,20,27,51] support GDBMSs using Cypher. Similar to Grand, GDsmith [ 13] also uses differential testing to find logic bugs.",
+      "context_after": "GraphGenie [ 14] modifies the graph patterns used in a query and checks if the query with modified graph patterns satisfies the expected relationship (i.e., equivalence, subset, or superset) with the original queries. GAMERA [ 51] extracts metamorphic relations [ 6] from the manipulated graph data and checks whether the outputs of generated queries satisfy these relations. Dinkel is different from all existing approaches in two key aspects. First, Dinkel significantly improves query generation, the core foundation of GDBMS testing. Second, while existing approaches support correctness testing",
+      "section": "5.5 Comparison",
+      "page": 12,
+      "char_offset": 60925,
+      "found_by_all": [
+        "technique",
+        "citation_marker_project_authored",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 14]",
+      "technique": null,
+      "sentence": "GraphGenie [ 14] modifies the graph patterns used in a query and checks if the query with modified graph patterns satisfies the expected relationship (i.",
+      "context_before": "veral approaches [ 13,14,20,27,49,51] have been proposed. All these approaches aim to construct test oracles to identify logic bugs. Grand [ 49] targets GDBMSs using the Gremlin query language and detects logic bugs using differential testing [ 28]. Several approaches [ 13,14,20,27,51] support GDBMSs using Cypher. Similar to Grand, GDsmith [ 13] also uses differential testing to find logic bugs. GDBMeter [ 20] leverages query partitioning [ 38], which decomposes the predicate of a query and checks if the queries with the decomposed predicates produce consistent results with the original query.",
+      "context_after": "e., equivalence, subset, or superset) with the original queries. GAMERA [ 51] extracts metamorphic relations [ 6] from the manipulated graph data and checks whether the outputs of generated queries satisfy these relations. Dinkel is different from all existing approaches in two key aspects. First, Dinkel significantly improves query generation, the core foundation of GDBMS testing. Second, while existing approaches support correctness testing for only queries following specific patterns, Dinkel generally tackles the problems of testoracle construction, enabling validation for arbitrary queries",
+      "section": "5.5 Comparison",
+      "page": 12,
+      "char_offset": 61127,
+      "cited_reference": {
+        "number": 14,
+        "text": "Yuancheng Jiang, Jiahao Liu, Jinsheng Ba, Roland Hock Chuan Yap, Zhenkai Liang, and Manuel Rigger. 2023. Detecting Logic Bugs in Graph Database Management Systems via Injective and Surjective Graph Query Transformation. In Proceedings of the 46th International Conference on Software Engineering (ICS",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[ 12,18,37–40,43]",
+      "technique": "norec",
+      "sentence": "Compared to GDBMS testing, testing relational database management systems (RDBMSs) is more mature, where both query generation [ 11,16,22,42,50] and test-oracle construction [ 12,18,37–40,43] are well-researched.",
+      "context_before": "subset, or superset) with the original queries. GAMERA [ 51] extracts metamorphic relations [ 6] from the manipulated graph data and checks whether the outputs of generated queries satisfy these relations. Dinkel is different from all existing approaches in two key aspects. First, Dinkel significantly improves query generation, the core foundation of GDBMS testing. Second, while existing approaches support correctness testing for only queries following specific patterns, Dinkel generally tackles the problems of testoracle construction, enabling validation for arbitrary queries. RDBMS Testing.",
+      "context_after": "SQLsmith [ 42] embeds the SQL grammar [ 41] and can generate complex SQL statements.DynSQL [ 16] incrementally generates SQL statements for a query by querying the latest DBMS schema, which enables DynSQL to decouple the generation of each statement and helps it generate queries containing multiple complex statements. To find logic bugs in RDBMSs, PQS [ 39] synthesizes customized queries that fetch specific rows of tables. If the tested RDBMS fails to fetch the rows, a logic bug is identified. Pinolo [ 12] modifies the predicate of a query, where the modified query should produce the subset/s",
+      "section": "5.5 Comparison",
+      "page": 12,
+      "char_offset": 61897,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESE/FSE). 1140–1152",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "To find logic bugs in RDBMSs, PQS [ 39] synthesizes customized queries that fetch specific rows of tables.",
+      "context_before": "struction, enabling validation for arbitrary queries. RDBMS Testing.Compared to GDBMS testing, testing relational database management systems (RDBMSs) is more mature, where both query generation [ 11,16,22,42,50] and test-oracle construction [ 12,18,37–40,43] are well-researched. SQLsmith [ 42] embeds the SQL grammar [ 41] and can generate complex SQL statements.DynSQL [ 16] incrementally generates SQL statements for a query by querying the latest DBMS schema, which enables DynSQL to decouple the generation of each statement and helps it generate queries containing multiple complex statements.",
+      "context_after": "If the tested RDBMS fails to fetch the rows, a logic bug is identified. Pinolo [ 12] modifies the predicate of a query, where the modified query should produce the subset/superset of the results of the original query. Otherwise, Pinolo identifies a logic bug. EET [ 18] transforms the expressions of queries in a semantic-preserving manner and checks whether the queries with the transformed expressions produce the same results as the original queries. EET identifies a bug if their results are not the same. Due to the separation of querying and data manipulation in the SQL language, EET cannot p",
+      "section": "5.5 Comparison",
+      "page": 12,
+      "char_offset": 62429,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:55:50Z",
+    "is_model_written": true,
+    "summary": "Dinkel generates Cypher queries for graph database systems while tracking the state a query builds up, so that later clauses can refer correctly to variables bound earlier. That state-awareness lets it emit far more of the Cypher language than the existing generators, which restrict themselves to shapes their transformation rules can handle. On Neo4j, RedisGraph and Memgraph it found bugs none of the existing approaches had reported.",
+    "narrative": "SQLancer's techniques appear as the relational background against which graph testing is measured. GDBMeter is one of the existing approaches Dinkel is compared with, and the paper identifies its oracle as query partitioning -- decomposing a query's predicate and checking the decomposed queries agree with the original. PQS is cited as the relational counterpart that synthesises queries fetching specific rows. The paper's own contribution is generation rather than an oracle, and it observes that relational testing is the more mature field, with both query generation and test-oracle construction well researched there.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "result_comparison",
+      "M6": "background",
+      "M7": "background",
+      "M8": "background",
+      "M9": "result_comparison",
+      "M10": "background",
+      "M11": "background",
+      "M12": "background",
+      "M13": "definition",
+      "M14": "definition",
+      "M15": "background",
+      "M16": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Dinkel implements its own state-aware Cypher generator; nothing describes reuse of SQLancer's code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Dinkel contributes query generation rather than an oracle, and no SQLancer technique is carried forward or generalised."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M5",
+          "M9"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M5",
+            "sentence": "As shown in Table 4, Dinkel supports more Cypher clauses than existing approaches [ 13,14,20,27,51].",
+            "section": "4 IMPLEMENTATION",
+            "page": 7
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "This comparison is reasonable and objective because: (1) none of the bugs found by Dinkel are marked as duplicated by developers, meaning that no approach found these bugs until Dinkel found them; and (2) all existing approaches have extensively tested Neo4j and RedisGraph [ 13,14,20,27,51], meaning that in these two GDBMSs, no approach found the long-latent bugs found by Dinkel during their evaluation.",
+            "section": "5.5 Comparison",
+            "page": 11
+          }
+        ],
+        "reasoning": "M5 compares Dinkel's Cypher clause coverage against the existing approaches including GDBMeter, and M9 argues the bug comparison is fair because all of them have extensively tested the same systems. M13 identifies GDBMeter's oracle as query partitioning. The comparison is with the tool rather than a direct run of TLP.",
+        "techniques": [
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Relational testing is called the more mature field, which is not a claim that any particular technique is the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_14778_3828612_3828639.json b/_data/papers/paper_doi_10_14778_3828612_3828639.json
new file mode 100644
index 0000000..1b4063f
--- /dev/null
+++ b/_data/papers/paper_doi_10_14778_3828612_3828639.json
@@ -0,0 +1,705 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:24:08Z",
+  "paper": {
+    "id": "paper:doi:10.14778/3828612.3828639",
+    "title": "ReSequel: Robust LLM-assisted Query Rewriting and Optimization using Templatization and Sampling",
+    "authors": [
+      "Saeed Fathollahzadeh",
+      "Essam Mansour",
+      "Matthias Boehm"
+    ],
+    "year": 2026,
+    "venue": "arXiv.org",
+    "doi": "10.14778/3828612.3828639",
+    "arxiv_id": "2606.20853",
+    "s2_paper_id": "b368cff4b76de570c592a1d84bf021214e3b3831",
+    "url": "https://doi.org/10.14778/3828612.3828639",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "arxiv",
+      "url": "https://arxiv.org/pdf/2606.20853",
+      "retrieved_at": "2026-09-08T15:24:08Z",
+      "chars": 103202,
+      "content_sha256": "sha256:eee25747223c510428f570e5b866d50d578721c0ec72936f2dd721155b468d75"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 14,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "INTRODUCTION",
+        "start": 2255
+      },
+      {
+        "number": "2.1",
+        "title": "Rule-based Query Rewriting",
+        "start": 14106
+      },
+      {
+        "number": "2.2",
+        "title": "Messy Query Characteristics",
+        "start": 15609
+      },
+      {
+        "number": "2.3",
+        "title": "Problem Formulation",
+        "start": 21132
+      },
+      {
+        "number": "3.1",
+        "title": "Query, Metadata, and Rewrite Hints",
+        "start": 23474
+      },
+      {
+        "number": "3.2",
+        "title": "Online Query Processing",
+        "start": 26318
+      },
+      {
+        "number": "3.3",
+        "title": "Offline Query Rewriting",
+        "start": 28560
+      },
+      {
+        "number": "6.1",
+        "title": "Overall Template Rewriting",
+        "start": 46645
+      },
+      {
+        "number": "6.2",
+        "title": "Prompt Construction",
+        "start": 49197
+      },
+      {
+        "number": "6.3",
+        "title": "Query Reconstruction and Caching",
+        "start": 52524
+      },
+      {
+        "number": "6.4",
+        "title": "System Limitations",
+        "start": 53945
+      },
+      {
+        "number": "7.1",
+        "title": "Experimental Setup",
+        "start": 54542
+      },
+      {
+        "number": "7.2",
+        "title": "End-to-End Experiments",
+        "start": 60987
+      },
+      {
+        "number": "7.3",
+        "title": "Micro Benchmarks",
+        "start": 68656
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "2025. Apache Calcite is a dynamic data management framework. https: //calcite.apache.org/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "2025. Groq Cloud. https://console.groq.com/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Peter Akioyamen, Zixuan Yi, and Ryan Marcus. 2024. The Unreasonable Effectiveness of LLMs for Query Optimization.NeurIPS(2024). https: //doi.org/10.48550/ARXIV.2411.02862",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Morton M. Astrahan et al .1976. System R: Relational Approach to Database Management.ACM Trans. Database Syst.1, 2 (1976), 97–137. https://doi.org/10. 1145/320455.320457",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Qiushi Bai, Sadeem Alsudais, and Chen Li. 2023. QueryBooster: Improving SQL Performance Using Middleware Services for Human-Centered Query Rewriting. PVLDB16, 11 (2023), 2911–2924. https://doi.org/10.14778/3611479.3611497",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Eirik Bakke and David R. Karger. 2016. Expressive Query Construction through Direct Manipulation of Nested Relational Results. InSIGMOD. 1377–1392. https: //doi.org/10.1145/2882903.2915210",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Zhuowei Bao, Benny Kimelfeld, and Yunyao Li. 2012. Automatic suggestion of query-rewrite rules for enterprise search. InSIGIR. 591–600. https://doi.org/10. 1145/2348283.2348363",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Alexander Baumstark, Muhammad Attahir Jibril, and Kai-Uwe Sattler. 2023. Adaptive query compilation in graph databases.Distributed Parallel Databases 41, 3 (2023), 359–386. https://doi.org/10.1007/S10619-023-07430-4",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Edmon Begoli, Jesús Camacho-Rodríguez, Julian Hyde, Michael J. Mior, and Daniel Lemire. 2018. Apache Calcite: A Foundational Framework for Optimized Query Processing Over Heterogeneous Data Sources. InSIGMOD. 221–230. https://doi.org/10.1145/3183713.3190662",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Hal Berenson, Philip A. Bernstein, Jim Gray, Jim Melton, Elizabeth J. O’Neil, and Patrick E. O’Neil. 1995. A Critique of ANSI SQL Isolation Levels. InSIGMOD. 1–10. https://doi.org/10.1145/223784.223785",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Altan Birler, Alfons Kemper, and Thomas Neumann. 2024. Robust Join Processing with Diamond Hardened Joins.PVLDB17, 11 (2024), 3215–3228. https://doi.org/10.14778/3681954.3681995",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Matthias Böhm, Douglas R. Burdick, Alexandre V. Evfimievski, Berthold Reinwald, Frederick R. Reiss, Prithviraj Sen, Shirish Tatikonda, and Yuanyuan Tian. 2014. SystemML’s Optimizer: Plan Generation for Large-Scale Machine Learning Programs.Data Eng. Bull.37, 3 (2014), 52–62. http://sites.computer.org/ debull/A14sept/p52.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Nicolas Bruno and Surajit Chaudhuri. 2006. To Tune or not to Tune? A Lightweight Physical Design Alerter. InVLDB. 499–510. http://dl.acm.org/citation. cfm?id=1164171",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Sunil Chakkappen, Shreya Kunjibettu, Daniel Mcgreer, Masoomeh Kishi, Hong Su, Mohamed Ziauddin, Mohamed Zaït, Zhan Li, and Yuying Zhang. 2025. Automatic Indexing in Oracle.PVLDB18, 12 (2025), 4924–4937. https://doi.org/ 10.14778/3750601.3750616",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Donald D. Chamberlin, Morton M. Astrahan, Kapali P. Eswaran, Patricia P. Griffiths, Raymond A. Lorie, James W. Mehl, Phyllis Reisner, and Bradford W. Wade. 1976. SEQUEL 2: A Unified Approach to Data Definition, Manipulation, and Control.IBM J. Res. Dev.20, 6 (1976), 560–575. https://doi.org/10.1147/RD. 206.0560",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Donald D. Chamberlin and Raymond F. Boyce. 1974. SEQUEL: A Structured English Query Language. InSIGMOD. 249–264. https://doi.org/10.1145/800296. 811515",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Konstantinos Chasialis, Yannis Foufoulas, Alkis Simitsis, and Yannis E. Ioannidis. 2026. Optimizing UDF Queries in SQL Data Engines. InEDBT. https://doi.org/ 10.48786/EDBT.2026.21",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Surajit Chaudhuri and Kyuseok Shim. 1994. Including Group-By in Query Optimization. InVLDB. 354–366. http://www.vldb.org/conf/1994/P354.PDF",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Transaction Processing Council. 2025. https://www.tpc.org/tpch.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Bailu Ding, Surajit Chaudhuri, Johannes Gehrke, and Vivek Narasayya. 2021. DSB: A decision support benchmark for workload-driven and traditional database systems.PVLDB14, 13 (2021), 3376–3388. https://doi.org/10.14778/3484224. 3484234",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Haoran Ding, Zhaoguo Wang, Yicun Yang, Dexin Zhang, Zhenglin Xu, Haibo Chen, Ruzica Piskac, and Jinyang Li. 2023. Proving Query Equivalence Using Linear Integer Arithmetic. https://doi.org/10.1145/3626768",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Jens Dittrich and Joris Nix. 2020. The Case for Deep Query Optimisation. In CIDR. http://cidrdb.org/cidr2020/papers/p3-dittrich-cidr20.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Jens Dittrich, Joris Nix, and Christian Schön. 2021. The next 50 Years in Database Indexing or: The Case for Automatically Generated Index Structures.PVLDB 15, 3 (2021), 527–540. https://doi.org/10.14778/3494124.3494136",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Rui Dong, Jie Liu, Yuxuan Zhu, Cong Yan, Barzan Mozafari, and Xinyu Wang. 2023. SlabCity: Whole-Query Optimization Using Program Synthesis.PVLDB 16, 11 (2023), 3151–3164. https://doi.org/10.14778/3611479.3611515",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Markus Dreseler, Martin Boissier, Tilmann Rabl, and Matthias Uflacker. 2020. Quantifying TPC-H Choke Points and Their Optimizations.VLDB13, 8 (2020), 1206–1220. https://doi.org/10.14778/3389133.3389138",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Tarek Elgamal, Shangyu Luo, Matthias Boehm, Alexandre V. Evfimievski, Shirish Tatikonda, Berthold Reinwald, and Prithviraj Sen. 2017. SPOOF: Sum-Product Optimization and Operator Fusion for Large-Scale Machine Learning. InCIDR. http://cidrdb.org/cidr2017/papers/p3-elgamal-cidr17.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Jingzhi Fang, Yanyan Shen, Yue Wang, and Lei Chen. 2020. Optimizing DNN Computation Graph using Graph Substitutions.PVLDB13, 11 (2020), 2734–2746. http://www.vldb.org/pvldb/vol13/p2734-fang.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Amela Fejza, Pierre Genevès, and Nabil Layaïda. 2024. Efficient Enumeration of Recursive Plans in Transformation-based Query Optimizers.PVLDB17, 11 (2024), 3095–3108. https://doi.org/10.14778/3681954.3681986",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Philipp Fent, Altan Birler, and Thomas Neumann. 2023. Practical planning and execution of groupjoin and nested aggregates.VLDB J.32, 6 (2023), 1165–1190. https://doi.org/10.1007/S00778-022-00765-X",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Sofoklis Floratos, Ahmad Ghazal, Jason Sun, Jianjun Chen, and Xiaodong Zhang. 2021. DBSpinner: Making a Case for Iterative Processing in Databases. InICDE. 2399–2410. https://doi.org/10.1109/ICDE51399.2021.00273",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Yannis Foufoulas and Alkis Simitsis. 2023. Efficient Execution of User-Defined Functions in SQL Queries.VLDB16, 12 (2023), 3874–3877. https://doi.org/10. 14778/3611540.3611574",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Michael J. Freitag, Maximilian Bandle, Tobias Schmidt, Alfons Kemper, and Thomas Neumann. 2020. Adopting Worst-Case Optimal Joins in Relational Database Systems.VLDB13, 11 (2020), 1891–1904. http://www.vldb.org/pvldb/ vol13/p1891-freitag.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Jonathan Fürst, Catherine Kosten, Farhad Nooralahzadeh, Yi Zhang, and Kurt Stockinger. 2025. Evaluating the Data Model Robustness of Text-to-SQL Systems Based on Real User Queries. InEDBT. https://doi.org/10.48786/EDBT.2025.13",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Rainer Gemulla, Philipp Rösch, and Wolfgang Lehner. 2008. Linked Bernoulli Synopses: Sampling along Foreign Keys. InSSDBM, Vol. 5069. 6–23. https: //doi.org/10.1007/978-3-540-69497-7_4",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Parke Godfrey, Jarek Gryz, Andrzej Hoppe, Wenbin Ma, and Calisto Zuzarte. 2009. Query Rewrites with Views for XML in DB2. InICDE. 1339–1350. https: //doi.org/10.1109/ICDE.2009.131",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Google. 2025. https://gemini.google.com.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Goetz Graefe. 1995. The Cascades Framework for Query Optimization.Data Eng. Bull.18, 3 (1995), 19–29. http://sites.computer.org/debull/95SEP-CD.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Luca Gretscher and Jens Dittrich. 2025. How to Optimize SQL Queries? A Comparison Between Split, Holistic, and Hybrid Approaches.PVLDB18, 11 (2025), 3910–3922. https://doi.org/10.14778/3749646.3749663",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "CWI Database Architectures Group. 2025. https://github.com/cwida/public_bi_ benchmark.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Immanuel Haffner and Jens Dittrich. 2023. A simplified Architecture for Fast, Adaptive Compilation and Execution of SQL Queries. InEDBT. https://doi.org/ 10.48786/EDBT.2023.01",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Yuxing Han et al .2021. Cardinality Estimation in DBMS: A Comprehensive Benchmark Evaluation.PVLDB15, 4 (2021), 752–765. https://doi.org/10.14778/ 3503585.3503586",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Cinda Heeren, H. V. Jagadish, and Leonard Pitt. 2003. Optimal indexing using near-minimal space. InPODS. 244–251. https://doi.org/10.1145/773153.773177",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Gerald Held, Michael Stonebraker, and Eugene Wong. 1975. INGRES: A Relational Data Base System. InAFIPS, Vol. 44. 409–416. https://doi.org/10.1145/ 1499949.1500029",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Moshik Hershcovitch, Artem Khyzha, Daniel G. Waddington, and Adam Morrison. 2022. Elastic Indexes: Dynamic Space vs. Query Efficiency Tuning for In-Memory Database Indexing. InEDBT. https://doi.org/10.48786/EDBT.2022.18",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Zezhou Huang, Rathijit Sen, Jiaxiang Liu, and Eugene Wu. 2023. JoinBoost: Grow Trees Over Normalized Data Using Only SQL.PVLDB16, 11 (2023), 3071–3084. https://doi.org/10.14778/3611479.3611509",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Zhihao Jia, Oded Padon, James Thomas, Todd Warszawski, Matei Zaharia, and Alex Aiken. 2019. TASO: optimizing deep learning computation with automatic generation of graph substitutions. InSOSP. 47–62. https://doi.org/10.1145/ 3341301.3359630",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Zhihao Jia, James Thomas, Todd Warszawski, Mingyu Gao, Matei Zaharia, and Alex Aiken. 2019. Optimizing DNN Computation with Relaxed Graph Substitutions. InMLSys. https://proceedings.mlsys.org/paper_files/paper/2019/ hash/",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "David Justen, Daniel Ritter, Campbell Fraser, Andrew Lamb, Nga Tran, Allison Lee, Thomas Bodner, Mhd Yamen Haddad, Steffen Zeuch, Volker Markl, and Matthias Boehm. 2024. POLAR: Adaptive and Non-invasive Join Order Selection via Plans of Least Resistance.PVLDB17, 6 (2024), 1350–1363. https://doi.org/ 10.14778/3648160.3648175",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "George Katsogiannis-Meimarakis and Georgia Koutrika. 2021. A Deep Dive into Deep Learning Approaches for Text-to-SQL Systems. InSIGMOD. 2846–2851. https://doi.org/10.1145/3448016.3457543",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Michael S. Kester, Manos Athanassoulis, and Stratos Idreos. 2017. Access Path Selection in Main-Memory Optimized Data Systems: Should I Scan or Should I Probe?. InSIGMOD. 715–730. https://doi.org/10.1145/3035918.3064049",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Steffen Kläbe and Kai-Uwe Sattler. 2023. Patched Multi-Key Partitioning for Robust Query Performance. InEDBT. https://doi.org/10.48786/EDBT.2023.26",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Jan Kossmann, Thorsten Papenbrock, and Felix Naumann. 2023. Correction to: Data dependencies for query optimization: a survey.VLDB J.32, 2 (2023), 471. https://doi.org/10.1007/S00778-021-00710-4",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Muralidhar Krishnaprasad, Zhen Hua Liu, Anand Manikutty, James W. Warner, Vikas Arora, and Susan Kotsovolos. 2004. Query Rewrite for XML in Oracle XMLDB. InVLDB. 1122–1133. https://doi.org/10.1016/B978-012088469-8.50098-X",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Viktor Leis, Andrey Gubichev, Atanas Mirchev, Peter A. Boncz, Alfons Kemper, and Thomas Neumann. 2015. How Good Are Query Optimizers, Really?PVLDB 9, 3 (2015), 204–215. https://doi.org/10.14778/2850583.2850594",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Viktor Leis, Kan Kundhikanjana, Alfons Kemper, and Thomas Neumann. 2015. Efficient Processing of Window Functions in Analytical SQL Queries.PVLDB8, 10 (2015), 1058–1069. https://doi.org/10.14778/2794367.2794375",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Viktor Leis, Bernhard Radke, Andrey Gubichev, Alfons Kemper, and Thomas Neumann. 2017. Cardinality Estimation Done Right: Index-Based Join Sampling. InCIDR. http://cidrdb.org/cidr2017/papers/p9-leis-cidr17.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Viktor Leis, Bernhard Radke, Andrey Gubichev, Atanas Mirchev, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2018. Query optimization through the looking glass, and what we found running the Join Order Benchmark.VLDB J. 27, 5 (2018), 643–668. https://doi.org/10.1007/S00778-017-0480-7",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Zhaodonghui Li, Haitao Yuan, Huiming Wang, Gao Cong, and Lidong Bing. 2024. LLM-R2: A Large Language Model Enhanced Rule-Based Rewrite System for Boosting Query Efficiency.PVLDB18, 1 (2024), 53–65. https://doi.org/10. 14778/3696435.3696440",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "Daniel Lindner, Daniel Ritter, and Felix Naumann. 2024. Enabling Data Dependency-based Query Optimization.CoRRabs/2406.06886 (2024). https: //doi.org/10.48550/ARXIV.2406.06886",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Jie Liu and Barzan Mozafari. 2026. GenRewrite: Query Rewriting via Large Language Models.SIGMOD4, 1 (2026), 1–26. https://dl.acm.org/doi/pdf/10. 1145/3786684",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Ester Livshits, Rina Kochirgan, Segev Tsur, Ihab F. Ilyas, Benny Kimelfeld, and Sudeepa Roy. 2021. Properties of Inconsistency Measures for Databases. In SIGMOD. 1182–1194. https://doi.org/10.1145/3448016.3457310",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Guy Lohman. 2002. The DB2 Universal Database Optimizer. Guest Lecture. https://cs.uwaterloo.ca/~ilyas/CS448W14/ibm.pdf.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Guy Lohman. 2014. Is Query Optimization a “Solved” Problem? ACMSIGMOD Blog. https://wp.sigmod.org/?p=1075.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Toby Mao. 2024. SQLGlot. GitHub repository. Retrieved April 29, 2026 from https://github.com/tobymao/sqlglot.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Volker Markl, Vijayshankar Raman, David E. Simmen, Guy M. Lohman, and Hamid Pirahesh. 2004. Robust Query Processing through Progressive Optimization. InSIGMOD. 659–670. https://doi.org/10.1145/1007568.1007642",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Guido Moerkotte. 2025. Building Query Compilers. https://pi3.informatik.unimannheim.de/~moer/querycompiler.pdf Last Accessed: October 03, 2025.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Guido Moerkotte and Thomas Neumann. 2011. Accelerating Queries with Group-By and Join by Groupjoin.PVLDB4, 11 (2011), 843–851. http://www. vldb.org/pvldb/vol4/p843-moerkotte.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Magnus Müller, Lucas Woltmann, and Wolfgang Lehner. 2023. Enhanced Featurization of Queries with Mixed Combinations of Predicates for ML-based Cardinality Estimation. InEDBT. https://doi.org/10.48786/EDBT.2023.22",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Parimarjan Negi, Ryan Marcus, Andreas Kipf, Hongzi Mao, Nesime Tatbul, Tim Kraska, and Mohammad Alizadeh. 2021. Flow-Loss: Learning Cardinality Estimates That Matter.PVLDB14, 11 (jul 2021), 2019–2032. https://doi.org/10. 14778/3476249.3476259",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Thomas Neumann and Alfons Kemper. 2015. Unnesting Arbitrary Queries. In BTW, Vol. P-241. 383–402. https://dl.gi.de/handle/20.500.12116/2418",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Thomas Neumann and Viktor Leis. 2024. A Critique of Modern SQL and a Proposal Towards a Simple and Expressive Query Language. InCIDR. https: //www.cidrdb.org/cidr2024/papers/p48-neumann.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 72,
+      "text": "Oleksii Vasyliev. 2025. https://pgtune.leopard.in.ua.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 73,
+      "text": "Amy Ousterhout, Steven McCanne, Henri Dubois-Ferrière, Silvery Fu, Sylvia Ratnasamy, and Noah Treuhaft. 2023. Zed: Leveraging Data Types to Process Eclectic Data. InCIDR. https://www.cidrdb.org/cidr2023/papers/p52ousterhout.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 74,
+      "text": "Simone Papicchio, Paolo Papotti, and Luca Cagliero. 2023. QATCH: Benchmarking SQL-centric tasks with Table Representation Learning Models on Your Data. InNeurIPS. http://papers.nips.cc/paper_ files/paper/2023/hash/62a24b69b820d30e9e5ad4f15ff7bf72-AbstractDatasets_and_Benchmarks.html",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 75,
+      "text": "Hamid Pirahesh, Joseph M. Hellerstein, and Waqar Hasan. 1992. Extensible/Rule Based Query Rewrite Optimization in Starburst. InSIGMOD. 39–48. https: //doi.org/10.1145/130283.130294",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 76,
+      "text": "Hamid Pirahesh, T. Y. Cliff Leung, and Waqar Hasan. 1997. A Rule Engine for Query Transformation in Starburst and IBM DB2 C/SDBMS. InICDE. 391–400. https://doi.org/10.1109/ICDE.1997.581945",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 77,
+      "text": "PostgreSQL. 2026. CREATE CAST. https://www.postgresql.org/docs/current/ sql-createcast.html Last Accessed: January 01, 2026.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 78,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InESEC/FSE. 1140– 1152. https://doi.org/10.1145/3368089.3409710",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 79,
+      "text": "Dario Satriani, Enzo Veltri, Donatello Santoro, Sara Rosato, Simone Varriale, and Paolo Papotti. 2025. Logical and Physical Optimizations for SQL Query Execution over Large Language Models. https://doi.org/10.1145/3725411",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 80,
+      "text": "Tobias Schmidt, Philipp Fent, and Thomas Neumann. 2022. Efficiently Compiling Dynamic Code for Adaptive Query Processing. 11–22. http://www.admsconf.org/2022-camera-ready/ADMS22_schmidt.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 81,
+      "text": "Tobias Schmidt, Viktor Leis, Peter A. Boncz, and Thomas Neumann. 2025. SQLStorm: Taking Database Benchmarking into the LLM Era.PVLDB18, 11 (2025), 4144–4157. https://www.vldb.org/pvldb/vol18/p4144-schmidt.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 82,
+      "text": "Pranjal Shankhdhar, Feilong Liu, Jay Narale, James Sun, Rebecca Schlussel, and Lyublena Antova. 2024. Presto’s History-based Query Optimizer.PVLDB17, 12 (2024), 4077–4089. https://doi.org/10.14778/3685800.3685828",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 83,
+      "text": "Moritz Sichert and Thomas Neumann. 2022. User-Defined Operators: Efficiently Integrating Custom Algorithms into Modern Databases.PVLDB15, 5 (2022), 1119–1131. https://doi.org/10.14778/3510397.3510408",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 84,
+      "text": "Tarique Siddiqui and Wentao Wu. 2023. ML-Powered Index Tuning: An Overview of Recent Progress and Open Challenges.SIGMOD Rec.52, 4 (2023), 19–30. https://doi.org/10.1145/3641832.3641836",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 85,
+      "text": "Jiansen Song, Wensheng Dou, Yu Gao, Ziyu Cui, Yingying Zheng, Dong Wang, Wei Wang, Jun Wei, and Tao Huang. 2024. Detecting Metadata-Related Logic Bugs in Database Systems via Raw Database Construction.PVLDB17, 8 (2024), 1884–1897. https://doi.org/10.14778/3659437.3659445",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 86,
+      "text": "Zhaoyan Sun, Xuanhe Zhou, Guoliang Li, Xiang Yu, Jianhua Feng, and Yong Zhang. 2025. R-Bot: An LLM-based Query Rewrite System.PVLDB18, 12 (2025), 5031–5044. https://www.vldb.org/pvldb/vol18/p5031-li.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 87,
+      "text": "Jie Tan, Kangfei Zhao, Rui Li, Jeffrey Xu Yu, Chengzhi Piao, Hong Cheng, Helen Meng, Deli Zhao, and Yu Rong. 2025. Can Large Language Models Be Query Optimizer for Relational Databases?arXiv preprint arXiv:2502.05562(2025). https://arxiv.org/pdf/2502.05562",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 88,
+      "text": "Nikolaos Tziavelis, Wolfgang Gatterbauer, and Mirek Riedewald. 2020. Optimal Join Algorithms Meet Top-k. InSIGMOD. 2659–2665. https://doi.org/10.1145/ 3318464.3383132",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 89,
+      "text": "Adrian Vogelsgesang, Tobias Mühlbauer, Viktor Leis, Thomas Neumann, and Alfons Kemper. 2019. Domain Query Optimization: Adapting the GeneralPurpose Database System Hyper for Tableau Workloads. InBTW (LNI), Vol. P289. 313–333. https://doi.org/10.18420/BTW2019-19",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 90,
+      "text": "Yisu Remy Wang, Shana Hutchison, Dan Suciu, Bill Howe, and Jonathan Leang. 2020. SPORES: Sum-Product Optimization via Relational Equality Saturation for Large Scale Linear Algebra.PVLDB13, 11 (2020), 1919–1932. http://www. vldb.org/pvldb/vol13/p1919-wang.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 91,
+      "text": "Zijia Wang, Haoran Liu, Chen Lin, Zhifeng Bao, Guoliang Li, and Tianqing Wang. 2024. Leveraging Dynamic and Heterogeneous Workload Knowledge to Boost the Performance of Index Advisors.PVLDB17, 7 (2024), 1642–1654. https://doi.org/10.14778/3654621.3654631",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 92,
+      "text": "Zhaoguo Wang, Zhou Zhou, Yicun Yang, Haoran Ding, Gansen Hu, Ding Ding, Chuzhe Tang, Haibo Chen, and Jinyang Li. 2022. WeTune: Automatic Discovery and Verification of Query Rewrite Rules. InSIGMOD. 94–107. https://doi.org/ 10.1145/3514221.3526125",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 93,
+      "text": "Marianne Winslett. 2003. Interview with Pat Selinger.SIGMOD Rec.32, 4 (2003), 93–103. http://www.acm.org/sigmod/record/issues/0312/17.selingerinterview.pdf",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 94,
+      "text": "Weipeng P. Yan and Per-Åke Larson. 1995. Eager Aggregation and Lazy Aggregation. InVLDB. 345–357. http://www.vldb.org/conf/1995/P345.PDF",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 95,
+      "text": "Zhengtong Yan, Valter Uotila, and Jiaheng Lu. 2023. Join Order Selection with Deep Reinforcement Learning: Fundamentals, Techniques, and Challenges. PVLDB16, 12 (2023), 3882–3885. https://doi.org/10.14778/3611540.3611576",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 96,
+      "text": "Weijie Zhao, Florin Rusu, Bin Dong, and Kesheng Wu. 2016. Similarity Join over Array Data. 2007–2022. https://doi.org/10.1145/2882903.2915247",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 97,
+      "text": "Wenchao Zhou, Yifan Cai, Yanqing Peng, Sheng Wang, Ke Ma, and Feifei Li. 2021. VeriDB: An SGX-based Verifiable Database. InSIGMOD. 2182–2194. https://doi.org/10.1145/3448016.3457308",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 98,
+      "text": "Wei Zhou, Chen Lin, Xuanhe Zhou, Guoliang Li, and Tianqing Wang. 2024. TRAP: Tailored Robustness Assessment for Index Advisors via Adversarial Perturbation. InICDE. 42–55. https://doi.org/10.1109/ICDE60146.2024.00011",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 99,
+      "text": "Xuanhe Zhou, Guoliang Li, Chengliang Chai, and Jianhua Feng. 2021. A learned query rewrite system using Monte Carlo tree search.PVLDB15, 1 (2021), 46–58. https://doi.org/10.14778/3485450.3485456",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 100,
+      "text": "Xuanhe Zhou, Zhaoyan Sun, and Guoliang Li. 2024. Db-gpt: Large language model meets database.Data Science and Engineering9, 1 (2024), 102–111. https://doi.org/10.1007/s41019-023-00235-6",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 78,
+      "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InESEC/FSE. 1140– 1152. https://doi.org/10.1145/3368089.3409710",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 78]",
+      "technique": "norec",
+      "sentence": "Beyond these methods, prior work on SQL testing [ 78] and text-to-SQL generation [ 74] compare query results directly.",
+      "context_before": "including group-join [ 67], which captures common rewrite patterns combining joins and group-by aggregation on the same key. In contrast, we perform holistic rewrites over join orders and related transformations on individual query instances. Automatic Query Verification:Verifying rewritten queries is essential for correctness. Traditional systems define equivalence using source-target patterns [ 66] and apply them iteratively. Automatic rewrite discovery systems also rely on verification [ 92]. SQLSolver [ 21] determines query equivalence using linear integer arithmetic and existing solvers.",
+      "context_after": "ReSequel follows this result-oriented approach, comparing outputs across rewritten templates on samples. 9 CONCLUSIONS We introduced ReSequel, an end-to-end query rewriting system that operates on top of existing DBMSs and uses LLMs to optimize groups of related queries. ReSequel combines template-level rewriting with verification and Top-1 query selection. We draw three conclusions. First, despite decades of work on query rewriting, large rule sets and their interactions remain brittle. Phase ordering, limited rewrite budgets, and strict pattern matching prevent DBMSs from fully optimizing m",
+      "section": "7.3 Micro Benchmarks",
+      "page": 12,
+      "char_offset": 82172,
+      "cited_reference": {
+        "number": 78,
+        "text": "Manuel Rigger and Zhendong Su. 2020. Detecting optimization bugs in database engines via non-optimizing reference engine construction. InESEC/FSE. 1140– 1152. https://doi.org/10.1145/3368089.3409710",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "prints the DOI of the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:17:50Z",
+    "is_model_written": true,
+    "summary": "ReSequel is an outer optimisation layer that rewrites SQL with a large language model on top of an existing DBMS. It uses catalog and statistical metadata to infer template-specific rules that steer the model, then generates, verifies and ranks candidate rewrites on sampled data to keep results correct while improving runtime. Across eight benchmarks and three DBMSs it reports workload-level speedups of up to 16x over the native systems.",
+    "narrative": "One citation, noting that prior work on SQL testing compares query results directly, in a discussion of how rewrites are verified.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_22399_ijcesen_5462.json b/_data/papers/paper_doi_10_22399_ijcesen_5462.json
new file mode 100644
index 0000000..f3d214d
--- /dev/null
+++ b/_data/papers/paper_doi_10_22399_ijcesen_5462.json
@@ -0,0 +1,475 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:30Z",
+  "paper": {
+    "id": "paper:doi:10.22399/ijcesen.5462",
+    "title": "Benchmarking Autonomy: A Taxonomy of Human-in-the-Loop Checkpoints for AI-Generated Transformation Code",
+    "authors": [
+      "Jagan Ankam"
+    ],
+    "year": 2022,
+    "venue": "International Journal of Computational and Experimental Science and Engineering",
+    "doi": "10.22399/ijcesen.5462",
+    "arxiv_id": null,
+    "s2_paper_id": "7cba14d41c15d68f708b0c94c47772053b982465",
+    "url": "https://doi.org/10.22399/ijcesen.5462",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.22399/ijcesen.5462",
+      "retrieved_at": "2026-09-09T01:33:30Z",
+      "chars": 105522,
+      "content_sha256": "sha256:c5c9ac704e83e421618c3fa6f0de101a033ea2fe458b5999aa1f7787a8cdb2b1"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 24,
+    "has_outline": false,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 3141
+      },
+      {
+        "number": "2",
+        "title": "A controlled pre -production benchmark of 12",
+        "start": 11760
+      },
+      {
+        "number": "3",
+        "title": "A task -level comparison of schema checks,",
+        "start": 11988
+      },
+      {
+        "number": "4",
+        "title": "RQ4: How can hum an review be positioned as a",
+        "start": 20211
+      },
+      {
+        "number": "5",
+        "title": "Results",
+        "start": 51593
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 67304
+      },
+      {
+        "number": "7",
+        "title": "Use Level 4 release controls: canary runs,",
+        "start": 83777
+      },
+      {
+        "number": "J",
+        "title": "Comput. Exp. Sci. Eng., vol. 8, no. 3, pp. 113 –",
+        "start": 105423
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "R. Y. Wang and D. M. Strong, “Beyond accuracy: What data quality means to data consumers,” J. Manage. Inf. Syst., vol. 12, no. 4, pp. 5 –33, 1996, doi: 10.1080/07421222.1996.11518099.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "L. L. Pipino, Y. W. Lee, and R. Y. Wang, “Data quality assessment,” Commun. ACM, vol. 45, no. 4, pp. 211 –218, 2002, doi: 10.1145/505248.506010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "C. Batini, C. Cappiello, C. Francalanci, and A. Maurino, “Methodologies for data quality assessment and improvement,” ACM Comput. Surv., vol. 41, no. 3, Art. no. 16, 2009, doi: 10.1145/1541880.1541883.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "D. M. Strong, Y. W. Lee, and R. Y. Wang, “Data quality in context,” Commun. ACM, vol. 40, no. 5, pp. 103 –110, 1997, doi: 10.1145/253769.253804.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "D. Sculley et al., “Hidden technical debt in machine learning systems,” in Advances in Neural Information Processing Systems 28, 2015, pp. 2503 –2511, doi: 10.5555/2969442.2969519.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "E. Breck, S. Cai, E. Nielsen, M. Salib, and D. Sculley, “The ML test score: A rubric for ML production readiness and technical debt reduction,” in Proc. IEEE Int. Conf. Big Data, 2017, pp. 1123 – 1132, doi: 10.1109/BigData.2017.8258038.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "D. Baylor et al., “TFX: A TensorFlow -based production -scale machine learning platform,” in Proc. 23rd ACMSIGKDD Int. Conf. Knowl. Discovery Data Mining, 2017, pp. 1387 –1395, doi: 10.1145/3097983.3098021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "S. Schelter et al., “Automating large -scale data quality verification,” Proc. VLDB Endow., vol. 11, no. 12, pp. 1781 –1794, 2018, doi: 10.14778/322986 3.3229867.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "S. Schelter et al., “Unit testing data with Deequ,” in Proc. 2019 Int. Conf. Manage. Data, 2019, pp. 1993 –1996, doi: 10.1145/3299869.3320210.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "L. E. Lwakatare, E. Range, I. Crnk ovic, and J. Bosch, “On the experiences of adopting automated data validation in an industrial machine learning project,” in Proc. IEEE/ACMICSE-SEIP, 2021, pp. 248 –257, doi: 10.1109/ICSE SEIP52600.2021.00034.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "E. Caveness et al., “TensorFlow Data Validation: Data analysis and validation in continuous ML pipelines,” in Proc. 2020 ACMSIGMOD Int. Conf. Manage. Data, 2020, pp. 2793 –2796, doi: 10.1145/3318464.3384707.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "J. Song and Y. He, “Auto-Validate: Unsupervised data validation using data -domain patterns inferred from data lakes,” in Proc. 2021 ACMSIGMOD Int. Conf. Manage. Data, 2021, pp. 1678 –1691, doi: 10.1145/3448016.3457250.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "E. T. Barr, M. Harman, P. McMinn, M. Shahbaz, and S. Yoo, “The oracle problem in software testing: A survey,” IEEE Trans. Softw. Eng., vol. 41, no. 5, pp. 507 –525, 2015, doi: 10.1109/TSE.2014.2372785.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Y. Jia and M. Harman, “An analysis and survey of the development of mutation testing,” IEEE Trans. Softw. Eng., vol. 37, no. 5, pp. 649 –678, 2011, doi: 10.1109/TSE.2010.62.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "R. Just, D. Jalali, and M. D. Ernst, “Defects4J: A database of existing faults to enable controlled testing studies for Java programs,” in Proc. Int. Symp. Softw. Testing Anal., 2014, pp. 437 –440, doi: 10.1145/2610384.2628055.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "T. Y. Chen et al., “Metamorphic testing: A review of challenges and opportunities,” ACM Comput. Surv., vol. 51, no. 1, Art. no. 4, 2018, doi: 10.1145/3143561.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proc. 14th USENIX Symp. Oper. Syst. Design Implement., 2020, pp. 667–682, doi: 10.48550/arXiv.2001.04174.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 18,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non -optimizing reference engine construction,” in Proc. 28th ACM ESEC/FSE, 2020, pp. 1140 –1152, doi: 10.1145/3368089.3409710.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 19,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, OOPSLA, Art. no. 211, 2020, doi: 10.1145/3428279.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 20,
+      "text": "R. Zhong et al., “SQUIRREL: Testing database management systems with language validity and coverage feedback,” in Proc. ACMCCS, 2020, pp. 955–967, doi: 10.1145/3372297.3417260.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "V. S. P. Dintyala, A. Narechania, and J. Arulraj, “SQLCheck: Automated detection and diagnosis of SQL anti -patterns,” in Proc. 2020 ACMSIGMOD, 2020, pp. 2331 –2345, doi: 10.1145/3318464.338975 4.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Q. Zhang, J. Wang, M. A. Gulzar, R. Padhye, and M. Kim, “BigFuzz: Efficient fuzz testing for data analytics using framework abstraction,” in Proc. 35th IEEE/ACMASE, 2020, doi: 10.1145/3324 884.3416641.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "K. Kallas, F. Niksic, C. Stanford, and R. Alur, “DiffStream: Differential output testing for stream processing programs,” Proc. ACM Program. Lang., vol. 4, OOPSLA, Art. no. 153, 2020, doi: 10.1145/3428221.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "M. A. Gulzar, S. Mardani, M. Musuvathi, and M. Kim, “White -box testing of big data analytics with complex user -defined functions,” in Proc. 27th ACMESEC/FSE, 2019, pp. 290 –301, doi: 10.1145/3338906.3338953. Jagan Ankam / IJCESEN 8-3(2022)124-147 147",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Z. Feng et al., “CodeBERT: A pre -trained model for programming and natural languages,” in Findings of EMNLP, 2020, pp. 1536 –1547, doi: 10.18653/v 1/2020.findings -emnlp.139.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "D. Guo et al., “GraphCodeBERT: Pre -training code representations with data flow,” in Proc. ICLR, 2021, doi: 10.48550/arXiv.2009.08366.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "W. U. Ahmad, S. Chakra borty, B. Ray, and K. -W. Chang, “Unified pre -training for program understanding and generation,” in Proc. NAACL HLT, 2021, pp. 2655 –2668, doi: 10.18653/v1/2021.naacl -main.211.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "M. Chen et al., “Evaluating large language models trained on code,” 2021, doi: 10.48550/arXiv.2107.03374.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "J. Austin et al., “Program synthesis with large language models,” 2021, doi: 10.48550/arXiv.2108.07732.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "J. A. Prenner and R. Robbes, “Automatic program repair with OpenAI Codex: Evaluating QuixBugs,” 2021, doi: 10.48550/ar Xiv.2111.03922.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "S. Amershi et al., “Guidelines for human-AI interaction,” in Proc. 2019 CHI Conf. Human Factors Comput. Syst., 2019, pp. 1 –13, doi: 10.1145/3290605.3300233.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "X. Wu, L. Xia o, Y. Sun, J. Zhang, T. Ma, and L. He, “A survey of human -in-the-loop for machine learning,” 2021, doi: 10.48550/arXiv.2108.00941.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "A. Crisan and B. Fiore-Gartland, “Fits and starts: Enterpri se use of AutoML and the role of humans in the loop,” in Proc. 2021 CHI, 2021, doi: 10.1145/3411764.3445775.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "N. Sambasivan et al., “Everyone wants to do the model work, not the data work: Data cascades in high-stakes AI,” in Proc. 2021 CHI, 2021, doi: 10.1145/3411764.3445518.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "C. Wohlin et al., Experimentation in Software Engineering. Berlin, Germany: Springer, 2012, doi: 10.1007/978 -3-642-29044 -2.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "B. A. Kitchenham et al., “Preliminary guidelines for empirical research in software engineering,” IEEE Trans. Softw. Eng., vol. 28, no. 8, pp. 721 –734, 2002, doi: 10.1109/TSE.2002.1027796.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "B. Fitzgerald and K. -J. Stol, “Continuous software engineering: A roadmap and agenda,” J. Syst. Softw., vol. 123, pp. 176 –189, 2017, doi: 10.1016/j.jss.2015.06.063.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "R. Gollapudi, “Risk -controlled near -zero-downtime Oracle database migration using GoldenGate,” Int. J. Comput. Exp. Sci. Eng., vol. 8, no. 3, pp. 113 – 123, 2022, doi: 10.22399/ijcesen.5382.",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 17,
+      "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proc. 14th USENIX Symp. Oper. Syst. Design Implement., 2020, pp. 667–682, doi: 10.48550/arXiv.2001.04174.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "prints the arXiv id of the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 18,
+      "text": "M. Rigger and Z. Su, “Detecting optimization bugs in database engines via non -optimizing reference engine construction,” in Proc. 28th ACM ESEC/FSE, 2020, pp. 1140 –1152, doi: 10.1145/3368089.3409710.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "prints the DOI of the paper introducing norec"
+      ]
+    },
+    {
+      "number": 19,
+      "text": "M. Rigger and Z. Su, “Finding bugs in database systems via query partitioning,” Proc. ACM Program. Lang., vol. 4, OOPSLA, Art. no. 211, 2020, doi: 10.1145/3428279.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "prints the DOI of the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[17] –[21]",
+      "technique": "pqs",
+      "sentence": "Database testing research develops oracles for defects in database engines rather than defects in applicati on-level transformation intent [17] –[21].",
+      "context_before": "endor model ranking, or proof that human reviewers are infallible. Those questions require a larger corpus, preserved model -generation logs, rep eated samples, blinded adjudication, and multiple organizations. Research Gap and Contributions Research on data quality, software testing, machine -learning operations, and human-AI interaction supplies many of the components required to govern generated t ransformations, but Jagan Ankam / IJCESEN 8-3(2022)124-147 126 the components are usually studied separately. Data -quality systems focus on observed datasets and recurring anomalies [8] –[12].",
+      "context_after": "Big -data testing work generates inputs and exposes faults in data intensive programs [22] –[24]. Code -generation research measures functional correctness on programming benchmarks [28], [29], while human in-the-loop research characterizes when human feedback and oversight can improve automated systems [31] –[34]. The literature does not provide a unified, task -level benchmark that compares structural, statistical, differential, and contextual checkpoints for AI -genera ted SQL, Spark, and dbt style transformations. A second gap concerns the granularity of human involvement. Human review is",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 9712,
+      "cited_reference": {
+        "number": 17,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proc. 14th USENIX Symp. Oper. Syst. Design Implement., 2020, pp. 667–682, doi: 10.48550/arXiv.2001.04174.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "prints the arXiv id of the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "technique",
+      "surface": "Pivoted Query Synthesis",
+      "technique": "pqs",
+      "sentence": "Pivoted Query Synthesis constructs queries expected to retrieve a selected row [17].",
+      "context_before": "suite distinguishes the mutated program from the reference [14]. Defects4J demonstrated the value of curated, reproducible fault corpora for empirical software -engineering studies [15]. Metamorphic testing reduces dependence on exact expected outputs by checking relations that should hold across transformed inputs or executions [16]. These ideas directly motivate the benchmark design in this article: defects are intentionally seeded, and checkpoints are evaluated by unique defect detection rather than by assertion count. Database -system testing provides strong examples of behavioral oracles.",
+      "context_after": "NoREC compare s an optimizable query with an equivalent form intended to suppress optimization [18]. Ternary Logic Partitioning checks relations among partitioned query results [19]. SQUIRREL combines language validity with coverage feedback to generate meaningful DBMS inputs [20], while SQLCheck detects and diagnoses SQL anti -patterns using query and data analysis [21]. Although these approaches primarily test database engines or query quality, they demonstrate that equivalent or reference executions can reveal logic de fects that do not crash the system. For data -intensive application cod",
+      "section": "3 A task -level comparison of schema checks,",
+      "page": 4,
+      "char_offset": 16214,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC compare s an optimizable query with an equivalent form intended to suppress optimization [18].",
+      "context_before": "ed the value of curated, reproducible fault corpora for empirical software -engineering studies [15]. Metamorphic testing reduces dependence on exact expected outputs by checking relations that should hold across transformed inputs or executions [16]. These ideas directly motivate the benchmark design in this article: defects are intentionally seeded, and checkpoints are evaluated by unique defect detection rather than by assertion count. Database -system testing provides strong examples of behavioral oracles. Pivoted Query Synthesis constructs queries expected to retrieve a selected row [17].",
+      "context_after": "Ternary Logic Partitioning checks relations among partitioned query results [19]. SQUIRREL combines language validity with coverage feedback to generate meaningful DBMS inputs [20], while SQLCheck detects and diagnoses SQL anti -patterns using query and data analysis [21]. Although these approaches primarily test database engines or query quality, they demonstrate that equivalent or reference executions can reveal logic de fects that do not crash the system. For data -intensive application code, BigFuzz uses framework abstraction and schema -aware mutations to test Spark programs efficiently",
+      "section": "3 A task -level comparison of schema checks,",
+      "page": 4,
+      "char_offset": 16299,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "technique",
+      "surface": "Ternary Logic Partitioning",
+      "technique": "tlp",
+      "sentence": "Ternary Logic Partitioning checks relations among partitioned query results [19].",
+      "context_before": "Metamorphic testing reduces dependence on exact expected outputs by checking relations that should hold across transformed inputs or executions [16]. These ideas directly motivate the benchmark design in this article: defects are intentionally seeded, and checkpoints are evaluated by unique defect detection rather than by assertion count. Database -system testing provides strong examples of behavioral oracles. Pivoted Query Synthesis constructs queries expected to retrieve a selected row [17]. NoREC compare s an optimizable query with an equivalent form intended to suppress optimization [18].",
+      "context_after": "SQUIRREL combines language validity with coverage feedback to generate meaningful DBMS inputs [20], while SQLCheck detects and diagnoses SQL anti -patterns using query and data analysis [21]. Although these approaches primarily test database engines or query quality, they demonstrate that equivalent or reference executions can reveal logic de fects that do not crash the system. For data -intensive application code, BigFuzz uses framework abstraction and schema -aware mutations to test Spark programs efficiently [22]. DiffStream compares stream -processing implementations for differential outp",
+      "section": "3 A task -level comparison of schema checks,",
+      "page": 4,
+      "char_offset": 16400,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[17]",
+      "technique": "pqs",
+      "sentence": "Database differential testing [17], [18], [19] Equivalent or partitioned query results Yes Adjudicates divergent executions Targets DBMS implementation rather than business intent.",
+      "context_before": "142 Stream Representative work Primary evidence Silent semantic errors Human role Limitation for generated transformations data validation constraints, and learned patterns and reviews exceptions defects may remain inside learned bounds. Software test oracles [13], [16] Expected output, specifications, and metamorphic relations Yes Defines or validates the oracle Oracle construction can be costly and specification -dependent. Controlled fault benchmarks [14], [15] Seeded or curated faults and test -suite kills Yes Validates fault realism Not specialized for SQL, Spark, or dbt transformations.",
+      "context_after": "Data -intensive program testing [22], [23], [24] Generated inputs and differential outputs Yes Interprets specifications and results Framework -specific setup; limited business semantics. AI code generation [25], [26], [27], [28], [29], [30] Benchmark tests and functional correctness Yes Prompts, repairs, and accepts artifacts Limited evidence on production transformation checkpoints. Human-AI interaction [31], [32], [33], [34] Calibrated f eedback and accountable oversight Yes Intervenes under uncertainty Does not quantify checkpoint -specific defect coverage. Figure 3. Artifact -centric e",
+      "section": "7 Use Level 4 release controls: canary runs,",
+      "page": 19,
+      "char_offset": 90671,
+      "cited_reference": {
+        "number": 17,
+        "text": "M. Rigger and Z. Su, “Testing database engines via pivoted query synthesis,” in Proc. 14th USENIX Symp. Oper. Syst. Design Implement., 2020, pp. 667–682, doi: 10.48550/arXiv.2001.04174.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "prints the arXiv id of the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:20:36Z",
+    "is_model_written": true,
+    "summary": "A benchmark and taxonomy for human review of AI-generated data transformations. The premise is that generated SQL, Spark and dbt code can compile, run and preserve the expected schema while silently changing what downstream data means. Across 12 transformation tasks the authors evaluate three automated checkpoint families, finding that comparison against a trusted baseline catches most defects while schema checks and data-quality rules catch few, and that one residual defect is found only by reviewing logic and business intent together.",
+    "narrative": "All three SQLancer oracles are described by name as the database-testing precedent: PQS constructing queries expected to retrieve a selected row, NoREC comparing an optimizable query with a non-optimizable equivalent, TLP checking relations among partitioned results. The paper distinguishes them as targeting engine defects rather than the application-level transformations it studies.",
+    "roles": {
+      "M1": "motivation",
+      "M2": "definition",
+      "M3": "definition",
+      "M4": "definition",
+      "M5": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_3390_app13042519.json b/_data/papers/paper_doi_10_3390_app13042519.json
new file mode 100644
index 0000000..2704c36
--- /dev/null
+++ b/_data/papers/paper_doi_10_3390_app13042519.json
@@ -0,0 +1,522 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:19Z",
+  "paper": {
+    "id": "paper:doi:10.3390/app13042519",
+    "title": "Squill: Testing DBMS with Correctness Feedback and Accurate Instantiation",
+    "authors": [
+      "Shihao Wen",
+      "Peng Jia",
+      "Pin Yang",
+      "Chi Hu"
+    ],
+    "year": 2023,
+    "venue": "Applied Sciences",
+    "doi": "10.3390/app13042519",
+    "arxiv_id": null,
+    "s2_paper_id": "b3692c4bdaaff8e7e0f9fdf68bd810d3d8275ebd",
+    "url": "https://doi.org/10.3390/app13042519",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.3390/app13042519",
+      "retrieved_at": "2026-09-09T01:33:19Z",
+      "chars": 77794,
+      "content_sha256": "sha256:d1c845f7f539c88906bf8c016618462c972020430be23df4cd6f13b31ad8c4e0"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 23,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2460
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 10761
+      },
+      {
+        "number": "2.1",
+        "title": "Overview of Squirrel",
+        "start": 11035
+      },
+      {
+        "number": "2.1.1",
+        "title": "Mutation of Squirrel",
+        "start": 12046
+      },
+      {
+        "number": "2.1.2",
+        "title": "Instantiation of Squirrel",
+        "start": 13136
+      },
+      {
+        "number": "2.2",
+        "title": "Motivation",
+        "start": 14988
+      },
+      {
+        "number": "2.2.1",
+        "title": "Correctness Feedback",
+        "start": 15004
+      },
+      {
+        "number": "2.2.2",
+        "title": "Limitation of Squirrel’s Instantiation",
+        "start": 19229
+      },
+      {
+        "number": "3",
+        "title": "Design of Squill",
+        "start": 22210
+      },
+      {
+        "number": "3.1",
+        "title": "Correctness-Guided Mutation",
+        "start": 23824
+      },
+      {
+        "number": "3.1.1",
+        "title": "Correctness-Focused Seed Selection",
+        "start": 24114
+      },
+      {
+        "number": "3.1.2",
+        "title": "Mutation with Interesting Material Library",
+        "start": 27110
+      },
+      {
+        "number": "3.2",
+        "title": "Semantics-Aware Instantiation",
+        "start": 30467
+      },
+      {
+        "number": "3.2.1",
+        "title": "Instantiation of Simple Semantics",
+        "start": 32895
+      },
+      {
+        "number": "3.2.2",
+        "title": "Instantiation of Complicated Semantics",
+        "start": 34343
+      },
+      {
+        "number": "4",
+        "title": "Implementation",
+        "start": 41662
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 42580
+      },
+      {
+        "number": "5.1",
+        "title": "DBMS Bugs",
+        "start": 44088
+      },
+      {
+        "number": "5.2",
+        "title": "Comparison with Existing Tools",
+        "start": 49831
+      },
+      {
+        "number": "5.3",
+        "title": "Contribution of Correctness-Guided Mutation",
+        "start": 53847
+      },
+      {
+        "number": "5.4",
+        "title": "Contribution of Semantics-Aware Instantiation",
+        "start": 57025
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 60405
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 62803
+      },
+      {
+        "number": "8",
+        "title": "Conclusions",
+        "start": 65932
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Seltenreich, A. SQLSmith. Available online: https://github.com/anse1/sqlsmith (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Rigger, M.; Su, Z. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2020, Virtual Event, 8–13 November 2020; Association for Computing Machinery: New York, NY, USA, 2020; pp. 1140–1152. [CrossRef",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 3,
+      "text": "Rigger, M.; Su, Z. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation, OSDI’20, Virtual, 4–6 November 2020; USENIX Association: Berkeley, CA, USA, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 4,
+      "text": "Rigger, M.; Su, Z. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 2020 ,4, 1–30. [CrossRef]",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 5,
+      "text": "Wikipedia. Fuzzing. Available online: https://en.wikipedia.org/wiki/Fuzzing (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Zalewski, M. American Fuzzy Lop. Available online: https://github.com/google/AFL (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Consortium, S. How SQLite Is Tested. Available online: https://www.sqlite.org/testing.html (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "LLVM. LibFuzzer. Available online: https://www.llvm.org/docs/LibFuzzer.html (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Google. Honggfuzz. Available online: https://github.com/google/honggfuzz (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Fioraldi, A.; Maier, D.; Eißfeldt, H.; Heuse, M., AFL++: Combining Incremental Steps of Fuzzing Research. In Proceedings of the 14th USENIX Conference on Offensive Technologies, Berkeley, CA, USA, 11 August 2020; USENIX Association: Berkeley, CA, USA, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Zhong, R.; Chen, Y.; Hu, H.; Zhang, H.; Lee, W.; Wu, D. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event, 9–13 November 2020; Ligatti, J., Ou, X., Katz, J., Vigna, G., Eds.; ACM: New York, NY, USA, 2020; pp. 955–970.. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Oracle. MySQL Server. Available online: https://github.com/mysql/mysql-server (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Foundation, M. MariaDB. Available online: https://github.com/MariaDB/server (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Group, A. OceanBase. Available online: https://github.com/oceanbase/oceanbase (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Lemieux, C.; Sen, K. FairFuzz: A Targeted Mutation Strategy for Increasing Greybox Fuzz Testing Coverage. In Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering, ASE 2018, Montpellier, France, 3–7 September 2018; Association for Computing Machinery: New York, NY, USA, 2018; pp. 475–485. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Wang, Y.; Jia, X.; Liu, Y.; Zeng, K.; Bao, T.; Wu, D.; Su, P. Not All Coverage Measurements Are Equal: Fuzzing by Coverage Accounting for Input Prioritization. In Proceedings of the 27th Annual Network and Distributed System Security Symposium, NDSS 2020, San Diego, CA, USA, 23–26 February 2020; The Internet Society: Washington, DC, USA, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Yue, T.; Wang, P .; Tang, Y.; Wang, E.; Yu, B.; Lu, K.; Zhou, X. EcoFuzz: Adaptive Energy-Saving Greybox Fuzzing as a Variant of the Adversarial Multi-Armed Bandit. In Proceedings of the 29th USENIX Conference on Security Symposium, SEC’20, San Diego, CA, USA, 12–14 August 2020; USENIX Association: Berkeley, CA, USA, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Liang, Y.; Liu, S.; Hu, H. Detecting Logical Bugs of DBMS with Coverage-based Guidance. In Proceedings of the 31st USENIX Security Symposium (USENIX Security 22), Boston, MA, USA, 10–12 August 2022; USENIX Association: Boston, MA, USA, 2022; pp. 4309–4326.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Google. AddressSanitizer. Available online: https://github.com/google/sanitizers/wiki/AddressSanitizer (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Foundation, F.S. GDB: The GNU Project Debugger. Available online: http://www.sourceware.org/gdb/ (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Klees, G.; Ruef, A.; Cooper, B.; Wei, S.; Hicks, M. Evaluating Fuzz Testing. In Proceedings of the 2018 ACMSIGSAC Conference on Computer and Communications Security, CCS ’18, Toronto, ON, Canada, 15–19 October 2018; Association for Computing Machinery: New York, NY, USA, 2018; pp. 2123–2138. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Yan, L.; Ahmad, M.W.; Jawarneh, M.; Shabaz, M.; Raffik, R.; Kishore, K.H.; Azeem, I. Single-Input Single-Output System with Multiple Time Delay PID Control Methods for UAV Cluster Multiagent Systems. Secur. Commun. Netw. 2022 ,2022, 3935143. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Gao, H.; Kareem, A.; Jawarneh, M.; Ofori, I.; Raffik, R.; Kishore, K.H. Metaheuristics Based Modeling and Simulation Analysis of New Integrated Mechanized Operation Solution and Position Servo System. Math. Probl. Eng. 2022 ,2022, 1466775. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Aschermann, C.; Frassetto, T.; Holz, T.; Jauernig, P .; Sadeghi, A.; Teuchert, D. NAUTILUS: Fishing for Deep Bugs with Grammars. In Proceedings of the 26th Annual Network and Distributed System Security Symposium, NDSS 2019, San Diego, CA, USA, 24–27 February 2019; The Internet Society: Washington, DC, USA, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Slutz, D. Massive Stochastic Testing of SQL; Technical Report MSR-TR-98-21; Publisher: Burlington, MA, USA, 1998. Appl. Sci. 2023 ,13, 2519 23 of 23",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Ghit, B.; Poggi, N.; Rosen, J.; Xin, R.; Boncz, P. SparkFuzz: Searching Correctness Regressions in Modern Query Engines. In Proceedings of the Workshop on Testing Database Systems, DBTest ’20, Portland, OR, USA, 19 June 2020; Association for Computing Machinery: New York, NY, USA, 2020. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Liu, X.; Zhou, Q.; Arulraj, J.; Orso, A. Automatic Detection of Performance Bugs in Database Systems Using Equivalent Queries. In Proceedings of the 44th International Conference on Software Engineering, ICSE ’22, Pittsburgh, PA, USA, 25–27 May 2022; Association for Computing Machinery: New York, NY, USA, 2022; pp. 225–236. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Gan, S.; Zhang, C.; Qin, X.; Tu, X.; Li, K.; Pei, Z.; Chen, Z. CollAFL: Path Sensitive Fuzzing. In Proceedings of the 2018 IEEE Symposium on Security and Privacy (SP), San Francisco, CA, USA, 20–24 May 2018; pp. 679–696. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Manès, V .J.M.; Kim, S.; Cha, S.K. Ankou: Guiding Grey-Box Fuzzing towards Combinatorial Difference. In Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering, ICSE ’20, Seoul, Republic of Korea, 27 June–19 July 2020; Association for Computing Machinery: New York, NY, USA, 2020; pp. 1024–1036. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Lyu, C.; Ji, S.; Zhang, C.; Li, Y.; Lee, W.H.; Song, Y.; Beyah, R. MOPT: Optimized Mutation Scheduling for Fuzzers. In Proceedings of the 28th USENIX Conference on Security Symposium, SEC’19, Berkeley, CA, USA, 14–16 August 2019; USENIX Association: Berkeley, CA, USA, 2019; pp. 1949–1966.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Zhou, C.; Wang, M.; Liang, J.; Liu, Z.; Jiang, Y. Zeror: Speed up Fuzzing with Coverage-Sensitive Tracing and Scheduling. In Proceedings of the 35th IEEE/ACM International Conference on Automated Software Engineering, ASE ’20, Melbourne, Australia, 21–25 September 2020; Association for Computing Machinery: New York, NY, USA, 2020; pp. 858–870. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Aschermann, C.; Schumilo, S.; Blazytko, T.; Gawlik, R.; Holz, T. REDQUEEN: Fuzzing with Input-to-State Correspondence. In Proceedings of the 26th Annual Network and Distributed System Security Symposium, NDSS 2019, San Diego, CA, USA, 24–27 February 2019; The Internet Society: Washington, DC, USA, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Chen, P .; Chen, H. Angora: Efficient Fuzzing by Principled Search. In Proceedings of the Privacy (SP), San Francisco, CA, USA, 20–24 May 2018; pp. 711–725. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Park, S.; Xu, W.; Yun, I.; Jang, D.; Kim, T. Fuzzing JavaScript Engines with Aspect-preserving Mutation. In Proceedings of the May 2020; pp. 1629–1642. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Yun, I.; Lee, S.; Xu, M.; Jang, Y.; Kim, T. QSYM: A Practical Concolic Execution Engine Tailored for Hybrid Fuzzing. In Proceedings of the 27th USENIX Security Symposium (USENIX Security 18), Vancouver, BC, Canada, 16–18 August 2017; USENIX Association: Baltimore, MD, USA, 2018; pp. 745–761.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Pham, V .T.; Böhme, M.; Santosa, A.E.; C˘ aciulescu, A.R.; Roychoudhury, A. Smart Greybox Fuzzing. IEEE Trans. Softw. Eng. 2021, 47, 1980–1997. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Wüstholz, V .; Christakis, M. Harvey: A Greybox Fuzzer for Smart Contracts. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2020, Virtual Event, 8–13 November 2020; Association for Computing Machinery: New York, NY, USA, 2020; pp. 1398–1409. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Fioraldi, A.; Maier, D.; Zhang, D.; Balzarotti, D. LibAFL: A framework to build modular and reusable fuzzers. In Proceedings of the CCS 2022, 29th ACM Conference on Computer and Communications Security, Los Angeles, CA, USA, 7–11 November 2022; ACM: New York, NY, USA, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Padhye, R.; Lemieux, C.; Sen, K.; Papadakis, M.; Le Traon, Y. Semantic Fuzzing with Zest. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2019, Beijing, China, 15–19 July 2019; Association for Computing Machinery: New York, NY, USA, 2019; pp. 329–340. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Blazytko, T.; Aschermann, C.; Schlögel, M.; Abbasi, A.; Schumilo, S.; Wörner, S.; Holz, T. GRIMOIRE: Synthesizing Structure while Fuzzing. In Proceedings of the 28th USENIX Security Symposium (USENIX Security 19), Santa Clara, CA, USA, 14–16 August 2019; USENIX Association: Santa Clara, CA, USA, 2019; pp. 1985–2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Foundation, F.S. Gnu Bison. Available online: https://www.gnu.org/software/bison (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Paxson, V. Flex. Available online: https://github.com/westes/flex (accessed on 22 November 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Wang, M.; Wu, Z.; Xu, X.; Liang, J.; Zhou, C.; Zhang, H.; Jiang, Y. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In Proceedings of the 43rd International Conference on Software Engineering: Software Engineering in Practice, ICSE-SEIP ’21, Madrid, Spain, 22–30 May 2021; IEEE Press: Hoboken, NJ, USA, 2021; pp. 328–337. [CrossRef] Disclaimer/Publisher’s Note: The statements, op",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 2,
+      "text": "Rigger, M.; Su, Z. Detecting Optimization Bugs in Database Engines via Non-Optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ESEC/FSE 2020, Virtual Event, 8–13 November 2020; Association for Computing Machinery: New York, NY, USA, 2020; pp. 1140–1152. [CrossRef]",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 3,
+      "text": "Rigger, M.; Su, Z. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation, OSDI’20, Virtual, 4–6 November 2020; USENIX Association: Berkeley, CA, USA, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 4,
+      "text": "Rigger, M.; Su, Z. Finding Bugs in Database Systems via Query Partitioning. Proc. ACM Program. Lang. 2020 ,4, 1–30. [CrossRef]",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Black-box fuzzing, or generation-based fuzzing, has been extensively used in finding DBMS bugs, such as SQLsmith [1] and SQLancer [2–4].",
+      "context_before": "overy. Keywords: coverage-based grey-box fuzzing; database testing; vulnerability 1. Introduction Database management systems (DBMSs) are widely used worldwide as the core of modern information systems. Like other complicated computer applications, the security and reliability of DBMSs face severe challenges. Malicious attacks on DBMSs, such as remote code execution or denial of service, will seriously harm the information system. Therefore, it is of great significance to efficiently detect DBMS vulnerabilities to improve their robustness and the security of the information system built on them.",
+      "context_after": "Security researchers have found a considerable number of bugs using this technique. A black-box fuzzer treats the program as a black box and is unaware of internal program structure [ 5]. It randomly generates a large number of SQL statements and executes them in the DBMS. The current input is saved for subsequent analysis when unexpected behavior occurs, such as a crash. The disadvantage of black-box fuzzing has been thoroughly discussed by the academic circle, which is inefficiency. Since the generation of SQL statements is entirely random, considering the complexity of the DBMS, most of the",
+      "section": "1 Introduction",
+      "page": 1,
+      "char_offset": 2978,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We did not compare Squill with SQLRight [ 18] and SQLancer, because their target is the logic bug of DBMSs, while Squill, like Squirrel, focuses on the memory error of DBMSs.",
+      "context_before": "was designed to answer the following questions: Q1. Can Squill detect bugs from well-tested DBMSs? (Section 5.1). Q2. Can Squill perform better than existing tools? (Section 5.2). Q3. How does correctness-guided mutation help fuzzing? (Section 5.3). Q4. What is the contribution of semantics-aware instantiation? (Section 5.4). We selected three popular real-world DBMSs for evaluation, including MySQL, MariaDB, and OceanBase. We mainly compared Squill with Squirrel, as Squirrel had been shown to outperform other mutation-based fuzzers, such as AFL, and generation-based fuzzers, such as SQLsmith.",
+      "context_after": "We perform the experiments on three computers with Ubuntu 18.04 system, Intel(R) Core(TM) i7-10700 (2.90 GHz) CPU, and 32 GB memory. We used the llvm mode of AFL to instrument the DBMS. Because of the large codebase of DBMSs, we set the bitmap size to 256 K and used a 20% ratio instrumentation. The DBMS versions in the experiment are all the latest, including MySQL 8.0.29, MariaDB 10.10.0, and OceanBase 3.1.4. In the experiments, due to resource bottleneck, we ran one DBMS and a fuzzer on each machine for 24 h at a time and repeated three times. Squill and Squirrel used the same seed and init",
+      "section": "5 Evaluation",
+      "page": 14,
+      "char_offset": 43285,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "Sqlancer",
+      "technique": null,
+      "sentence": "Sqlancer [ 2–4] constructs different SQL statements of functionally equivalent through several different patterns and inputs them into the same DBMS.",
+      "context_before": "SQL statements into the DBMS to trigger abnormal behaviors (usually crashes) of the DBMS is one method of black-box DBMS fuzzing. Sqlsmith [ 1] is a representative of this kind of black-box DBMS fuzzer. Based on AST, it randomly generates SQL query statements for the initial database through a series of highly customized rules. In addition, differential testing is another standard method used to detect DBMS vulnerabilities in black-box DBMS fuzzing. Rags [ 25] and Sparkfuzz [ 26] send the same SQL query to different DBMSs and detect correctness bugs by comparing the differences in the results.",
+      "context_after": "If the results are different, the DBMS might have a logical bug. Similarly, AMOEBA [ 27] constructs query pairs that are semantically equivalent to each other and then compares their response time on the same database system to detect performance bugs. The main difference between Squill and the works above is that Squill is a grey-box fuzzer with feedback like code coverage. Compared with blind fuzzing, fuzzing with feedback can comprehensively explore program states and trigger the deep logic of DBMSs. Grey-box DBMS Fuzzing. In recent years, grey-box or mutation-based fuzzing has shown its e",
+      "section": "7 Related Work",
+      "page": 21,
+      "char_offset": 63727,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M2"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:52Z",
+    "is_model_written": true,
+    "summary": "Squill fuzzes DBMSs for memory errors, using correctness feedback to keep generated statements valid and accurate instantiation to fill in values that satisfy the schema. Valid statements reach deeper into the engine, which is where memory errors are found.",
+    "narrative": "SQLancer is named as a leading example of generation-based DBMS fuzzing and described as constructing functionally equivalent SQL statements through several patterns and feeding them to the same DBMS. The paper is explicit that it does not compare against it: SQLancer and SQLRight target logic bugs, while Squill, like Squirrel, targets memory errors, so the two are measured on different things.",
+    "roles": {
+      "M1": "background",
+      "M2": "incidental",
+      "M3": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M2 states directly that Squill was not compared with SQLancer, because SQLancer targets logic bugs while Squill targets memory errors."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_3390_electronics14193910.json b/_data/papers/paper_doi_10_3390_electronics14193910.json
new file mode 100644
index 0000000..1a0ed86
--- /dev/null
+++ b/_data/papers/paper_doi_10_3390_electronics14193910.json
@@ -0,0 +1,401 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T17:13:51Z",
+  "paper": {
+    "id": "paper:doi:10.3390/electronics14193910",
+    "title": "Nabil: A Text-to-SQL Model Based on Brain-Inspired Computing Techniques and Large Language Modeling",
+    "authors": [
+      "Feng Zhou",
+      "Shijing Hu",
+      "Xiaozheng Du",
+      "Nan Li",
+      "Tongming Zhou",
+      "Yanni Zhao",
+      "Sitong Shang",
+      "Xufeng Ling",
+      "Huaizhong Zhu"
+    ],
+    "year": 2025,
+    "venue": "Electronics",
+    "doi": "10.3390/electronics14193910",
+    "arxiv_id": null,
+    "s2_paper_id": "aeaf7936aa6c16580244cb4f315b3387d0d6a7e2",
+    "url": "https://doi.org/10.3390/electronics14193910",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.3390/electronics14193910",
+      "retrieved_at": "2026-09-08T17:13:51Z",
+      "chars": 57889,
+      "content_sha256": "sha256:8c562cfde99b9b9191ddca57d0251504f83e578eaf3eca9b740bd24567c0abb4"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 18,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2930
+      },
+      {
+        "number": "2",
+        "title": "Related Works",
+        "start": 9091
+      },
+      {
+        "number": "3",
+        "title": "Methodology",
+        "start": 20443
+      },
+      {
+        "number": "3.1",
+        "title": "Preliminaries",
+        "start": 20715
+      },
+      {
+        "number": "3.2",
+        "title": "Nabil Model",
+        "start": 23952
+      },
+      {
+        "number": "3.3",
+        "title": "Nabil Model Algorithm",
+        "start": 28868
+      },
+      {
+        "number": "3.4",
+        "title": "Nabil Model Normalization Module Algorithm",
+        "start": 30996
+      },
+      {
+        "number": "3.5",
+        "title": "Nabil Model Syntax Tree Abstraction Process",
+        "start": 33497
+      },
+      {
+        "number": "4",
+        "title": "Experimental Results and Analysis",
+        "start": 35003
+      },
+      {
+        "number": "4.1",
+        "title": "Datasets",
+        "start": 36285
+      },
+      {
+        "number": "4.2",
+        "title": "Evaluation Metrics",
+        "start": 37830
+      },
+      {
+        "number": "4.3",
+        "title": "Performance Comparison Experiment",
+        "start": 39010
+      },
+      {
+        "number": "4.4",
+        "title": "Ablation Comparison Experiment",
+        "start": 40712
+      },
+      {
+        "number": "4.5",
+        "title": "Large Language Model Comparison Experiment",
+        "start": 43456
+      },
+      {
+        "number": "5",
+        "title": "Conclusions",
+        "start": 46687
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Wang, J.; Liu, S.; Chen, Z.; Shen, T.; Wang, Y.; Yin, R.; Liu, H.; Liu, C.; Shen, C. Ultrasensitive electrospinning fibrous strain sensor with synergistic conductive network for human motion monitoring and human-computer interaction. J. Mater. Sci. Technol. 2025, 213, 213–222. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Meng, Q.; Yan, Z.; Abbas, J.; Shankar, A.; Subramanian, M. Human–computer interaction and digital literacy promote educational learning in pre-school children: Mediating role of psychological resilience for kids’ mental well-being and school readiness. Int. J. Hum.-Comput. Interact. 2025 ,41, 16–30. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Mehonic, A.; Kenyon, A.J. Brain-inspired computing needs a master plan. Nature 2022 ,604, 255–260. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Guo, J.; Zhan, Z.; Gao, Y.; Xiao, Y.; Lou, J.G.; Liu, T.; Zhang, D. Towards complex text-to-sql in cross-domain database with intermediate representation. arXiv 2019, arXiv:1905.08205.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Sen, J.; Lei, C.; Quamar, A.; Özcan, F.; Efthymiou, V .; Dalmia, A.; Stager, G.; Mittal, A.; Saha, D.; Sankaranarayanan, K. Athena++ natural language querying for complex nested sql queries. Proc. VLDB Endow. 2020 ,13, 2747–2759. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Liu, J.; Cui, Q.; Cao, H.; Shi, T.; Zhou, M. Auto-conversion from Natural Language to Structured Query Language using Neural Networks Embedded with Pre-training and Fine-tuning Mechanism. In Proceedings of the 2020 Chinese Automation Congress (CAC), Shanghai, China, 6–8 November 2020; IEEE: New York City, NY, USA; pp. 6651–6654.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Marcus, R.; Negi, P .; Mao, H.; Tatbul, N.; Alizadeh, M.; Kraska, T. Bao: Making learned query optimization practical. In Proceedings of the 2021 International Conference on Management of Data, Xi’an, China, 20–25 June 2021; pp. 1275–1288.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Cao, R.; Chen, L.; Chen, Z.; Zhao, Y.; Zhu, S.; Yu, K. LGESQL: Line graph enhanced text-to-SQL model with mixed local and non-local relations. arXiv 2021, arXiv:2106.01093.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Sioulas, P .; Ailamaki, A. Scalable multi-query execution using reinforcement learning. In Proceedings of the 2021 International Conference on Management of Data, Xi’an, China, 20–25 June 2021; pp. 1651–1663.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Ahkouk, K.; Machkour, M.; Ennaji, M. Data agnostic RoBERTa-based natural language to SQL query generation. In Proceedings of the IEEE 6th International Conference for Convergence in Technology (I2CT), Pune, India, 2–4 April 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Zhao, C.; Su, Y.; Pauls, A.; Platanios, E.A. Bridging the generalization gap in text-to-SQL parsing with schema expansion. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Dublin, Ireland, 22–27 May 2022; pp. 5568–5578.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Hui, B.; Geng, R.; Wang, L.; Qin, B.; Li, B.; Sun, J.; Li, Y. S2SQL: Injecting Syntax to Question-Schema Interaction Graph Encoder for Text-to-SQL Parsers. arXiv 2022, arXiv:2203.06958.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Yu, X.; Chai, C.; Li, G.; Liu, J. Cost-based or learning-based? A hybrid query optimizer for query plan selection. Proc. VLDB Endow. 2022 ,15, 3924–3936. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Gan, Y.; Chen, X.; Huang, Q.; Purver, M. Measuring and improving compositional generalization in text-to-sql via component alignment. arXiv 2022, arXiv:2205.02054.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Fu, H.; Liu, C.; Wu, B.; Li, F.; Tan, J.; Sun, J. Catsql: Towards real world natural language to sql applications. Proc. VLDB Endow. 2023 ,16, 1534–1547. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Chen, Z.; Chen, S.; White, M.; Mooney, R.; Payani, A.; Srinivasa, J.; Su, Y.; Sun, H. Text-to-SQL error correction with language models of code. arXiv 2023, arXiv:2305.13073.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Gu, Z.; Fan, J.; Tang, N.; Cao, L.; Jia, B.; Madden, S.; Du, X. Few-shot text-to-sql translation using structure and content prompt learning. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Giaquinto, R.; Zhang, D.; Kleiner, B.; Li, Y.; Tan, M.; Bhatia, P .; Nallapati, R.; Ma, X. Multitask pretraining with structured knowledge for text-to-SQL generation. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Toronto, ON, Canada, 9–14 July 2023; pp. 11067–11083.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Lee, K.; Dutt, A.; Narasayya, V .; Chaudhuri, S. Analyzing the impact of cardinality estimation on execution plans in microsoft SQL server. Proc. VLDB Endow. 2023 ,16, 2871–2883. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Ba, J.; Rigger, M. Testing database engines via query plan guidance. In Proceedings of the 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE), Melbourne, Australia, 14–20 May 2023; IEEE: New York City, NY, USA; pp. 2060–2071.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 21,
+      "text": "Chen, T.; Gao, J.; Chen, H.; Tu, Y. Loger: A learned optimizer towards generating efficient and robust query execution plans. Proc. VLDB Endow. 2023 ,16, 1777–1789. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Li, R.; Zhao, K.; Yu, J.X.; Wang, G. CardOOD: Robust Query-driven Cardinality Estimation under Out-of-Distribution. arXiv 2024, arXiv:2412.05864.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Fan, Y.; Ren, T.; Huang, C.; He, Z.; Wang, X.S. Grounding Natural Language to SQL Translation with Data-Based Self-Explanations. arXiv 2024, arXiv:2411.02948. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Fan, J.; Gu, Z.; Zhang, S.; Zhang, Y.; Chen, Z.; Cao, L.; Li, G.; Madden, S.; Du, X.; Tang, N. Combining small language models and large language models for zero-shot NL2SQL. Proc. VLDB Endow. 2024 ,17, 2750–2763. [CrossRef] Electronics 2025 ,14, 3910 18 of 18",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Liu, C.; Liao, W.; Xu, Z. Research on natural language query to SQL method with fused table structure. In Proceedings of the 2024 5th International Conference on Computer Engineering and Application (ICCEA), Hangzhou, China, 12–14 April 2024; IEEE: New York City, NY, USA; pp. 564–567.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Kim, H.; Jeon, T.; Choi, S.; Choi, S.; Cho, H. FLEX: Expert-level False-Less EXecution Metric for Reliable Text-to-SQL Benchmark. arXiv 2024, arXiv:2409.19014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Mao, W.; Wang, R.; Guo, J.; Zeng, J.; Gao, C.; Han, P .; Liu, C. Enhancing Text-to-SQL Parsing through Question Rewriting and Execution-Guided Refinement. In Proceedings of the Findings of the Association for Computational Linguistics ACL 2024, Bangkok, Thailand, 11–16 August 2024; pp. 2009–2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Xie, X.; Xu, G.; Zhao, L.; Guo, R. OpenSearch-SQL: Enhancing Text-to-SQL with Dynamic Few-shot and Consistency Alignment. arXiv 2025, arXiv:2502.14913. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Chen, K.; Chen, Y.; Koudas, N.; Yu, X. Reliable Text-to-SQL with Adaptive Abstention. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Castelein, J.; Aniche, M.; Soltani, M.; Panichella, A.; van Deursen, A. Search-based test data generation for SQL queries. In Proceedings of the 40th International Conference on Software Engineering, Gothenburg, Sweden, 27 May–3 June 2018; pp. 1220–1230.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Chu, S.; Murphy, B.; Roesch, J.; Cheung, A.; Suciu, D. Axiomatic foundations and algorithms for deciding semantic equivalences of SQL queries. arXiv 2018, arXiv:1802.02229. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Zhou, Q.; Arulraj, J.; Navathe, S.; Harris, W.; Xu, D. Automated verification of query equivalence using satisfiability modulo theories. Proc. VLDB Endow. 2019 ,12, 1276–1288. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Zhou, Q.; Arulraj, J.; Navathe, S.B.; Harris, W.; Wu, J. SPES: A symbolic approach to proving query equivalence under bag semantics. In Proceedings of the 9–12 May 2022; IEEE: New York City, NY, USA; pp. 2735–2748.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Wang, S.; Pan, S.; Cheung, A. QED: A Powerful Query Equivalence Decider for SQL. Proc. VLDB Endow. 2024 ,17, 3602–3614. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "He, Y.; Zhao, P .; Wang, X.; Wang, Y. VeriEQL: Bounded Equivalence Verification for Complex SQL Queries with Integrity Constraints. Proc. ACM Program. Lang. 2024 ,8, 1071–1099. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Zhang, Y.; Qu, P .; Ji, Y.; Zhang, W.; Gao, G.; Wang, G.; Song, S.; Li, G.; Chen, W.; Zheng, W.; et al. A system hierarchy for brain-inspired computing. Nature 2020 ,586, 378–384. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Du, X.; Hu, S.; Zhou, F.; Wang, C.; Nguyen, B.M. FI-NL2PY2SQL: Financial Industry NL2SQL Innovation Model Based on Python and Large Language Model. Future Internet 2025 ,17, 12. [CrossRef]",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Du, X.; Guo, X.; Zhou, F.; Gu, M.; Lu, Z.; Wang, C. FinDS2: A Novel Data Synthesis System for Fintech Product Risks. In Proceedings of the China, 28–30 June 2024; IEEE: New York City, NY, USA; pp. 73–78.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Li, J.; Hui, B.; Qu, G.; Yang, J.; Li, B.; Li, B.; Wang, B.; Qin, B.; Geng, R.; Huo, N.; et al. Can llm already serve as a database interface? a big bench for large-scale database grounded text-to-sqls. Adv. Neural Inf. Process. Syst. 2023 ,36, 42330–42357. Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and ",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 20,
+      "text": "Ba, J.; Rigger, M. Testing database engines via query plan guidance. In Proceedings of the 2023 IEEE/ACM 45th International Conference on Software Engineering (ICSE), Melbourne, Australia, 14–20 May 2023; IEEE: New York City, NY, USA; pp. 2060–2071.",
+      "technique": "qpg",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing qpg",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "technique",
+      "surface": "QPG",
+      "technique": "qpg",
+      "sentence": "proposed query plan guidance (QPG) to fully automatically test errors in database systems and applied it to three database systems: SQLite, TiDB, and CockroachDB [ 20].",
+      "context_before": "diting model was proposed. Gu, Z. et al. utilized the Spider dataset to enhance the generalization ability of text-to-SQL translation in cases involving limited text and proposed the SC-Prompt framework [ 17]. Giaquinto, R. et al. used a pre-training dataset containing SQL code, text, and tables to achieve the goal of improving model performance in the Text-to-SQL generation task [ 18]. Lee, K. et al. used Microsoft SQL Server to evaluate the impact of cardinality estimation on complex SQL queries, studying queries such as aggregation, subqueries, grouping, and outer joins [ 19]. Ba, J. et al.",
+      "context_after": "Chen T. et al. proposed the LOGER optimizer to generate robust and efficient query execution plans using the Join Order Benchmark (JOB), TPC-DS, and Stack Overflow datasets [ 21]. The experimental results show that LOGER outperforms existing learning query optimizers and is 2.07 times faster than PostgreSQL on the JOB dataset. Li R. et al. proposed the CardOOD general learning framework [ 22], which focuses on offline training algorithms to improve the out-of-distribution (OOD) problem of querydriven estimators. The framework extends traditional robust learning and transfer learning technique",
+      "section": "2 Related Works",
+      "page": 4,
+      "char_offset": 13209,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "qpg"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:03:26Z",
+    "is_model_written": true,
+    "summary": "Nabil is a text-to-SQL model combining brain-inspired computing techniques with large language modeling, translating natural-language questions into SQL queries. Its related-work section surveys work on database systems, including automated testing.",
+    "narrative": "QPG is described once in related work: the paper notes that query plan guidance was proposed to test database systems fully automatically and was applied to SQLite, TiDB and CockroachDB. The subject of this paper is query generation from natural language rather than testing, so the citation is background and nothing of SQLancer's is used.",
+    "roles": {
+      "M1": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_53799_zb6b1375.json b/_data/papers/paper_doi_10_53799_zb6b1375.json
new file mode 100644
index 0000000..06424dd
--- /dev/null
+++ b/_data/papers/paper_doi_10_53799_zb6b1375.json
@@ -0,0 +1,352 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:24:02Z",
+  "paper": {
+    "id": "paper:doi:10.53799/zb6b1375",
+    "title": "Optimizing Database System Performance: Design and Query Optimization Strategies",
+    "authors": [
+      "Juanda Hakim Lubis",
+      "I. Jaya",
+      "Fajrul Malik Aminullah Napitupulu",
+      "H. Mawengkang"
+    ],
+    "year": 2026,
+    "venue": "AIUB Journal of Science and Engineering (AJSE)",
+    "doi": "10.53799/zb6b1375",
+    "arxiv_id": null,
+    "s2_paper_id": "bdaced332423f73da24f03a45785321328579190",
+    "url": "https://doi.org/10.53799/zb6b1375",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.53799/zb6b1375",
+      "retrieved_at": "2026-09-08T15:24:02Z",
+      "chars": 67176,
+      "content_sha256": "sha256:0f768fe9b3a509e58cd7f130fac7d2d3f06208f0ffb25b54f1d871d28109173f"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": false,
+    "sections": []
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "P. G. Selinger, M. M. Astrahan, D. D. Chamberlin, R. A. Lor ie, and T. G. Price, “Access path selection in a relational database management system,” in Proc. ACMSIGMOD Int. C onf. Manage. Data, 1979, pp. 23–34, doi: 10.1145/582095.582099.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "S. Chaudhuri, “An overview of query optimization in relational systems,” in P roc. ACMSIGACT-SIGMOD-SIGART Symp. Principles Database Syst. (PODS), 1998, pp. 34 –43, doi: 10.1145/275487.275492.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "G. Graefe, “Query evaluation techniques for large databases,” ACM Comput. Surv., vol. 25, no. 2, pp. 73 –169, 1993, doi: 10.1145/152610.152611.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "V. Poosala, Y. E. Ioannidis, P. J. Haas, and E. J. Shekita, “Improved histograms for selectivity estimation of ra nge predicates,” in Proc. ACM SIGMOD Int. Conf. Manage. Data, 1996, pp. 294 –305, doi: 10.1145/233269.233342.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "V. Leis, A. Gubichev, A. Mirchev, P. Boncz, A. Kemper, and T. Neumann, “How good are query optimizers, really?” Proc. VLDB Endow., vol. 9, no. 3, p p. 204 –215, 2015, doi: 10.14778/2850583.2850594.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "H. Köhler and S. Link, “SQL schema design: Foundations, normal forms, and normalization,” in Proc. ACMSIGMOD Int. Conf. Manage. Data, 2016, pp. 267 –282, doi: 10.1145/2882903.2915239.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "G. L. Sanders and S. K. Shin, “Denormalization effects on performance of RDBMS,” in Proc. 34th Hawaii Int. Conf. Syst. Sci., 2001, doi: 10.1109/HICSS.2001.926306.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "S. K. Shin and G. L. Sanders, “Denormalization strategies for data retrieval from data warehouses,” Decis. Support S yst., vol. 42, no. 1, pp. 267–282, 2006, doi: 10.1016/j.dss.2004.12.004.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "D. Milićev, “Hyper -relations: A model for denormalization of transac tional relational databases,” IEEE Trans. Knowl. Data Eng., vol. 35, no. 4, pp. 3979 –3990, Apr. 2023, doi: 10.1109/ TKDE.2021.3124134.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "D. Lindner, D. Ritter, and F. Naumann, “Enabling data dependency based query optimization,” Proc. VLDB Endow., vol. 17, no. 6, pp. 1025 –1037, 2024, doi: 10.48550/arXiv.2406.06886.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "T. Taipalus, “On the effects of logical database design o n database size, query complexity, query performance, and energy consumption,” arXiv preprint arXiv:2501.07449, 2025. [Online]. Available: https://arxiv.org/abs/2501.07449",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "M. Fotache, M. I. Cluci, T. Taipalu s, and G. Talaba, “The effects of database normalization on decision support system performance,” Inf. Syst., vol. 136, Art. no. 102636, Feb. 2026, doi: 10.1016/j.is.2025.102636.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "J. Ba and M. Rigger, “CERT: Finding performance issues in database systems th rough the lens of cardinality estimation,” in Proc. ACM SIGMOD Int. Conf. Manage. Data, 2023, pp. 1600 –1612, doi: 10.1145/3597503.3639076.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 14,
+      "text": "J. Lao, Y. Wang, Y. Li, J. Wang, Y. Zhang, Z. Cheng, W. Chen, M. Tang, and J. Wang, “GPTuner: A manual -reading databa se tuning system via GPT -guided Bayesian optimization,” arXiv preprint arXiv:2311.03157, 2023. [Online]. Available: https://arxiv.org/abs/2311.03157",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "S.-W. Lee, B. Moon, C. Park, J. -M. Kim, and S. -W. Kim, “A case for flash memory SSD in enterprise database applications,” in Proc. ACM SIGMOD Int. Conf. Manage. Data, 2008, pp. 1075 –1086, doi: 10.114 5/1376616.1376723.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "S.-W. Lee, B. Moon, and C. Park, “Advances in flash memory SSD technology for enterprise databas e applications,” in Proc. ACM SIGMOD Int. Conf. Manage. Data, 2009, pp. 863 –870, doi: 10.1145/1559845.1559937.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "D. Bausch, I. Petrov, and A. B uchmann, “On the performance of database query processing algorithms on flash solid state disks,” in Proc. 2011 Dat abase and Expert Systems Applications (DEXA) Int. Workshops, Toulouse, France, 2011, pp. 139 –144, doi: 10.1109/DEXA.2011.60.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Q. Xu, H. Siyamw ala, M. Ghosh, T. Suri, M. Awasthi, Z. Guz, A. Shayesteh, and V. Balakrishnan, “Performance analysis of NVMe SSDs and their implication on real world databases,” in Proc. 8th ACM Int. Syst. Storage Conf. (SYSTOR), 2015, pp. 6:1 –6:11, doi: 10.1145/2757667.2 757684.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "A. Fevgas, L. Akritidis, P. Bozanis, and Y. Manolopoulos, “Indexing in flash storage devices: A survey on c hallenges, current approaches, and future trends,” VLDB J., vol. 29, no. 1, pp. 273 –311, 2020, doi: 10.1007/s00778 -019-00559 -8.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "G. Graefe, S. Harizopoulos, H. A. Kuno, M. A. Shah, D. Tsirogiannis, and J. L. Wiener, “Designing database operators for flash -enabled memory hierarchies,” IEEE Data Eng. Bull., vol. 33, no. 4, pp. 21 –27, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "S. A. M. Tipan and R. B. Ricafort, “A performance analysis: Evaluating relational database system on HDD vs. SSD storage,” Int. J. Latest Technol. Eng. Manag. Appl. Sci., vo l. 15, no. 4, pp. 548 –558, Apr. 2026, doi: 10.51583/IJLTEMAS.2026.150400049.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "D. A. Patterson, G. Gibson, and R. H. Katz, “A case for redundan t arrays of inexpensive disks (RAID),” in Proc. ACMSIGMOD Int. Conf. Manage. Data, 1988, pp. 109 –116, doi: 10.1145 /50202.50214.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "P. M. Chen, E. K. Lee, G. A. Gibson, R. H. Katz, and D. A. Patterson, “RAID: High -performance, reliable secondary storage,” ACM Comput. Surv., vol. 26, no. 2, pp. 145 –185, 1994, doi: 10.1145/176979.176981.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "P. M. Chen, E. K. Lee, G. A. Gibson, R. H. Katz, and D. A. Patterson, “Performance and design evaluation of the RAID-II storage server,” Distrib. Parallel Databases, vol. 2, no. 3, pp. 243 –260, 1994, doi: 10.1007/BF01266330.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "S. Manegold, P. Boncz, and M. Kersten, “Optimizing database archite cture for the new bottleneck: Memory access,” VLDB J., vol. 9, no. 3, pp. 231 –246, 2000, doi: 10.1007/s007780000031.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "R. Ramakrishnan and J. G ehrke, Database Management Systems, 3rd ed. New York, NY, USA: McGraw-Hill, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "H. Garcia-Molina, J. D. Ullman, a nd J. Widom, Database Systems: The Complete Book, 2nd ed. Upper Saddle River, NJ, USA: Pearson, 2008. Juanda Hakim Lubis is currently purs uing a doctoral program in Computer Science at the University of North Sumatra. He completed his bachelor's degree in informatics engineering at Institut Teknologi Telkom Bandung in 2010 and his master's degree in Informatics E",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 13,
+      "text": "J. Ba and M. Rigger, “CERT: Finding performance issues in database systems th rough the lens of cardinality estimation,” in Proc. ACM SIGMOD Int. Conf. Manage. Data, 2023, pp. 1600 –1612, doi: 10.1145/3597503.3639076.",
+      "technique": "cert",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing cert",
+        "prints the DOI of the paper introducing cert"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[13]",
+      "technique": "cert",
+      "sentence": "Recent studies have further shown that additional dependency information and cardinality -estimation quality can significantly affect query plan efficiency and execution behavior [10], [13].",
+      "context_before": "be evaluated not only from a conceptual and integrity -oriented perspective, but also from an empirical performance perspective. The performance consequences of relational design become especially important in the context o f query optimization. Cost-based optimizers compare alternative access paths, join orders, and execution strategies using estimated resource cost and data -distribution information [1 -5]. Schema characteristics such as primary keys, foreign keys, table deco mposition, and relation dependencies influence the optimizer’s ability to select efficient plans [1], [2], [4], [10].",
+      "context_after": "Therefore, two schemas implementing the same business domain may lead to substantially different optimizer decisions and query runtimes, even when the logical meaning of the workload is similar. RAIUB JOURNAL OFSCIENCE ANDENGINEERING ISSN: 1608 - 3679 (print) 2520 - 4890 (Online) https://doi.org/10.53799/zb6b1375Published in AJSE, Vol: 25, Issue: 1 Received on 5th April 2025 Revised on 16th August 2026 Published on 31st August 2026 A Controlled Evaluation of Relational Database Models for Query Performance across DAS, RAID-0, and RAID-1 Storage Configurations Juanda Hakim Lubis, Ivan Jaya, Fa",
+      "section": null,
+      "page": 1,
+      "char_offset": 4731,
+      "cited_reference": {
+        "number": 13,
+        "text": "J. Ba and M. Rigger, “CERT: Finding performance issues in database systems th rough the lens of cardinality estimation,” in Proc. ACM SIGMOD Int. Conf. Manage. Data, 2023, pp. 1600 –1612, doi: 10.1145/3597503.3639076.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert",
+          "prints the DOI of the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[13]",
+      "technique": "cert",
+      "sentence": "Existing studies tend to focus either on general normalization strategies [6-12] or on optimizer behavior and cost estimation [10], [13], leaving a gap in understanding how alternative r elational design choices influence practical DBMS performance under a shared workload.",
+      "context_before": "Controlled Evaluation of Relational Database Models for Query Performance across DAS, RAID-0, and RAID-1 Storage Configurations Juanda Hakim Lubis, Ivan Jaya, Fajrul Malik Aminullah Napitupulu, and Herman Mawengkang AJSE Volume 25, Issue 1, Page 19 - 30 Page 19 Although prior literature has examined normalization, denormalization, and query optimization extensively, fewer studies directly compare alternative ERD -level relational designs for the same application domain and evaluate their performance across diverse query ca tegories using both response time and optimizer -estimated query cost.",
+      "context_after": "This study addresses that gap by empirically comparing three relational database designs: ERD-1, ERD-2, and ERD-3, constructed for the same inventory -oriented applicatio n domain. The models are evaluated across multiple data volumes and seven categories of queries: simple queries, aggregate queries, inner joins, outer joins, subqueries, correlated subqueries, and complex queries. Performance is assessed using two compleme ntary indicators: observed query response time and cost -based optimizer output. The contribution of this study is threefold. First, it provides an empirical comparison of",
+      "section": null,
+      "page": 2,
+      "char_offset": 5961,
+      "cited_reference": {
+        "number": 13,
+        "text": "J. Ba and M. Rigger, “CERT: Finding performance issues in database systems th rough the lens of cardinality estimation,” in Proc. ACM SIGMOD Int. Conf. Manage. Data, 2023, pp. 1600 –1612, doi: 10.1145/3597503.3639076.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert",
+          "prints the DOI of the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[13]",
+      "technique": "cert",
+      "sentence": "Ba and Rigger furt her confirmed that cardinality -estimation errors remain a meaningful source of DBMS performance issues across mature database systems [13].",
+      "context_before": "r summarized the major components of relational query optimization, including query rewriting, access -path selection, and join optimization [2]. Graefe further reviewed the physical execution techniques used for large database workloads [3]. The quality of optimizer decisions depends heavily on the accuracy of statistics and selectivity estimation. Poosala et al. demonstrated the importance of improved histograms for estimating predicate selectivity [4]. Leis et al. later showed that query optimizers can suffer significant performance degradation when cardinality estimates are inaccurate [5].",
+      "context_after": "Schema design is directly connected to these optimization concerns. Primary -key and foreign -key structures, entity decomposition, join relationships, and available dependencies influence the optimizer’s ability to select efficient plans. Lindner et al. showed that additional data dependencies can improve query optimization and reduce execution time beyond what is possib le using only conventional key constraints [10]. This finding supports the broader argument that relational structure is not merely a modeling concern, but also a factor that shapes optimizer behavior and query efficiency. D",
+      "section": null,
+      "page": 2,
+      "char_offset": 10775,
+      "cited_reference": {
+        "number": 13,
+        "text": "J. Ba and M. Rigger, “CERT: Finding performance issues in database systems th rough the lens of cardinality estimation,” in Proc. ACM SIGMOD Int. Conf. Manage. Data, 2023, pp. 1600 –1612, doi: 10.1145/3597503.3639076.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert",
+          "prints the DOI of the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[13]",
+      "technique": "cert",
+      "sentence": "Research Gap Prior studies have established that relational schema design influences query performa nce [7 -12] and that optimizer effectiveness depends on access paths, cardinality estimation, AJSE Volume 25, Issue 1, Page 19 - 30 Page 20 and structural dependencies [1 -5], [10], [13].",
+      "context_before": "roader argument that relational structure is not merely a modeling concern, but also a factor that shapes optimizer behavior and query efficiency. Despite extensive research on que ry optimization, most studies focus on optimizer algorithms, statistics, or estimation quality rather than on how alternative ERD -level designs alter optimizer -reported query cost under equivalent application semantics. The present study contributes to thi s area by examining how different relational design models affect both observed response time and optimizer -estimated cost across multiple query categories. C.",
+      "context_after": "However, these two research strands are often investigated separately. There remains a need for empirical studies th at connect database design alternatives with query execution behavior in a unified experimental setting. Specifically, relatively few works compare several ERD level relational designs for the same domain while simultaneously evaluating: •response time acro ss multiple query categories; •cost-based optimizer outputs; •and the relative behavior of alternative relational structures under increasing data volume. This study addresses that gap by comparing three relational database",
+      "section": null,
+      "page": 2,
+      "char_offset": 11987,
+      "cited_reference": {
+        "number": 13,
+        "text": "J. Ba and M. Rigger, “CERT: Finding performance issues in database systems th rough the lens of cardinality estimation,” in Proc. ACM SIGMOD Int. Conf. Manage. Data, 2023, pp. 1600 –1612, doi: 10.1145/3597503.3639076.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing cert",
+          "prints the DOI of the paper introducing cert"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "cert"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:04Z",
+    "is_model_written": true,
+    "summary": "A controlled evaluation of how relational schema design affects query performance. Three alternative designs for one inventory domain are measured across three storage configurations, seven query categories and four data volumes, using both measured response time and the optimizer's own cost estimates on Oracle 11gR2.",
+    "narrative": "The connection is a single citation, to CERT. The paper cites Ba and Rigger as having confirmed that cardinality-estimation errors remain a meaningful source of DBMS performance issues across mature database systems, and uses that alongside other work to frame its research gap -- that optimizer effectiveness is known to depend on cardinality estimation and access paths, while the influence of schema design choices under a shared workload is less well understood. SQLancer itself is never named, and none of the four mentions is reachable except through the citation marker.",
+    "roles": {
+      "M1": "background",
+      "M2": "motivation",
+      "M3": "background",
+      "M4": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_doi_10_7717_peerj_cs_1592.json b/_data/papers/paper_doi_10_7717_peerj_cs_1592.json
new file mode 100644
index 0000000..bac5306
--- /dev/null
+++ b/_data/papers/paper_doi_10_7717_peerj_cs_1592.json
@@ -0,0 +1,148 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T15:20:23Z",
+  "paper": {
+    "id": "paper:doi:10.7717/peerj-cs.1592",
+    "title": "DAFuzz: data-aware fuzzing of in-memory data stores",
+    "authors": [
+      "Yingpei Zeng",
+      "Fengmin Zhu",
+      "Siyi Zhang",
+      "Yu Yang",
+      "Siyu Yi",
+      "Yufan Pan",
+      "Guojie Xie",
+      "Ting Wu"
+    ],
+    "year": 2023,
+    "venue": "PeerJ Computer Science",
+    "doi": "10.7717/peerj-cs.1592",
+    "arxiv_id": null,
+    "s2_paper_id": "a3f904ba97221a2e65f6ac9e34f3403e287e0a49",
+    "url": "https://doi.org/10.7717/peerj-cs.1592",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://doi.org/10.7717/peerj-cs.1592",
+      "retrieved_at": "2026-09-08T15:20:23Z",
+      "chars": 68956,
+      "content_sha256": "sha256:3c6a39098e20c89804ed4ae6443fe3b8bd2dd60be28dd1710900bd4b4569dfb9"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 26,
+    "has_outline": true,
+    "sections": []
+  },
+  "references": [],
+  "sqlancer_references": [],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "author_year_citation",
+      "surface": "Rigger, 2023",
+      "technique": null,
+      "sentence": "1592 20/26 related program is SQL database, and different special black-box fuzzers ( Seltenreich, Tang & Mullender, 2022 ;Guo, 2017 ;Rigger, 2023 ;Rigger & Su, 2020 ) and grey-box fuzzers (Zhong et al.",
+      "context_before": "ually could use data variables without de fining them first. However, different from the existing work, DAFuzz also generates data-aware and semantics-aware new inputs, which is to ensure that different data items are correctly referred to in the inputs. Data-related program fuzzing There are researchers focusing on fuzzing data-related programs. For in-memory data store, Google security researchers fuzzed Redis but only used existing ordinary fuzzers like AFL ( Google Information Security Engineering Team, 2020 ). Another kind of dataZeng et al. (2023), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.",
+      "context_after": ", 2020 ;Wang et al., 2021 ;Liang, Liu & Hu, 2022 ) have been developed. At first, fuzzers mainly focus on keeping the generated or mutated SQL statements syntactically valid ( Seltenreich, Tang & Mullender, 2022 ;Guo, 2017 ). Recently, fuzzers also try to ensure the statements are semantically valid ( Rigger, 2023 ;Rigger & Su, 2020 ;Zhong et al., 2020; Wang et al., 2021 ;Liang, Liu & Hu, 2022 ). For example, they ensure that the used tables are created first and the mentioned columns still exist ( Zhong et al., 2020 ;Liang, Liu & Hu, 2022). Several recent pieces of research focus on detecting l",
+      "section": null,
+      "page": 20,
+      "char_offset": 52803,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "author_year_citation",
+      "surface": "Rigger, 2023",
+      "technique": null,
+      "sentence": "Recently, fuzzers also try to ensure the statements are semantically valid ( Rigger, 2023 ;Rigger & Su, 2020 ;Zhong et al.",
+      "context_before": "d existing ordinary fuzzers like AFL ( Google Information Security Engineering Team, 2020 ). Another kind of dataZeng et al. (2023), PeerJ Comput. Sci., DOI 10.7717/peerj-cs.1592 20/26 related program is SQL database, and different special black-box fuzzers ( Seltenreich, Tang & Mullender, 2022 ;Guo, 2017 ;Rigger, 2023 ;Rigger & Su, 2020 ) and grey-box fuzzers (Zhong et al., 2020 ;Wang et al., 2021 ;Liang, Liu & Hu, 2022 ) have been developed. At first, fuzzers mainly focus on keeping the generated or mutated SQL statements syntactically valid ( Seltenreich, Tang & Mullender, 2022 ;Guo, 2017 ).",
+      "context_after": ", 2020; Wang et al., 2021 ;Liang, Liu & Hu, 2022 ). For example, they ensure that the used tables are created first and the mentioned columns still exist ( Zhong et al., 2020 ;Liang, Liu & Hu, 2022). Several recent pieces of research focus on detecting logic bugs but not traditional crashes or assert failures ( Rigger, 2023 ;Rigger & Su, 2020 ;Liang, Liu & Hu, 2022 ). However, all these previous works do not intentionally create different kinds of data as DAFuzz does for providing the required data, and all these fuzzers are tightly bound to the SQL language and cannot work with in-memory data",
+      "section": null,
+      "page": 21,
+      "char_offset": 53229,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "author_year_citation",
+      "surface": "Rigger, 2023",
+      "technique": null,
+      "sentence": "Several recent pieces of research focus on detecting logic bugs but not traditional crashes or assert failures ( Rigger, 2023 ;Rigger & Su, 2020 ;Liang, Liu & Hu, 2022 ).",
+      "context_before": ";Rigger & Su, 2020 ) and grey-box fuzzers (Zhong et al., 2020 ;Wang et al., 2021 ;Liang, Liu & Hu, 2022 ) have been developed. At first, fuzzers mainly focus on keeping the generated or mutated SQL statements syntactically valid ( Seltenreich, Tang & Mullender, 2022 ;Guo, 2017 ). Recently, fuzzers also try to ensure the statements are semantically valid ( Rigger, 2023 ;Rigger & Su, 2020 ;Zhong et al., 2020; Wang et al., 2021 ;Liang, Liu & Hu, 2022 ). For example, they ensure that the used tables are created first and the mentioned columns still exist ( Zhong et al., 2020 ;Liang, Liu & Hu, 2022).",
+      "context_after": "However, all these previous works do not intentionally create different kinds of data as DAFuzz does for providing the required data, and all these fuzzers are tightly bound to the SQL language and cannot work with in-memory data stores that use other languages. CONCLUSION To exercise the code paths of in-memory data stores that require different data, we presented a new fuzzing approach DAFuzz. DAFuzz could not only generate inputs that are syntactically and semantically valid but also use different data correctly. In addition, DAFuzz adopts the state-of-the-art tree-based mutation method as",
+      "section": null,
+      "page": 21,
+      "char_offset": 53550,
+      "found_by_all": [
+        "author_year_citation"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T17:01:52Z",
+    "is_model_written": true,
+    "summary": "DAFuzz fuzzes in-memory data stores such as Redis, whose commands operate on typed data structures. Feeding a command arguments of the wrong type gets it rejected immediately, so DAFuzz tracks the types of live data and generates commands whose arguments match, reaching code that type-blind fuzzing does not.",
+    "narrative": "SQLancer appears in the related-work survey in the ACM author-year style, so none of the three mentions names it directly -- they are reachable only through the citation. It is grouped with SQLsmith as a black-box fuzzer for SQL databases, cited again as an example of fuzzers that ensure statements are semantically valid, which is the property DAFuzz pursues for data-structure commands, and a third time as work targeting logic bugs rather than crashes or assertion failures.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited as related work; no reuse of its code, generator or workload is described."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer or one of its oracles is reported."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The text does not describe SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json b/_data/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json
new file mode 100644
index 0000000..8325efe
--- /dev/null
+++ b/_data/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.json
@@ -0,0 +1,821 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:32:53Z",
+  "paper": {
+    "id": "paper:s2:295c629691d5654def2d9d85f72c756c67c68583",
+    "title": "WingFuzz: Implementing Continuous Fuzzing for DBMSs",
+    "authors": [
+      "Jie Liang",
+      "Zhiyong Wu",
+      "Jingzhou Fu",
+      "Yiyuan Bai",
+      "Qiang Zhang",
+      "Yu Jiang"
+    ],
+    "year": 2024,
+    "venue": "USENIX Annual Technical Conference",
+    "doi": null,
+    "arxiv_id": null,
+    "s2_paper_id": "295c629691d5654def2d9d85f72c756c67c68583",
+    "url": "https://www.semanticscholar.org/paper/295c629691d5654def2d9d85f72c756c67c68583",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "usenix",
+      "url": "https://www.usenix.org/system/files/atc24-liang.pdf",
+      "retrieved_at": "2026-09-09T01:32:53Z",
+      "chars": 71581,
+      "content_sha256": "sha256:7ab2167fab543ceecf5bbc8593fe1cd52032e59417271193e03ae1395ad3b187"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2381
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 8313
+      },
+      {
+        "number": "3",
+        "title": "Obstacles in Continuous DBMS Fuzzing",
+        "start": 9525
+      },
+      {
+        "number": "3.1",
+        "title": "Step 1: Customize Query Generator",
+        "start": 9838
+      },
+      {
+        "number": "3.2",
+        "title": "Step 2: Fuzz Evolving Codebase",
+        "start": 12184
+      },
+      {
+        "number": "3.3",
+        "title": "Step 3: Detect and Analyze Anomalies",
+        "start": 13903
+      },
+      {
+        "number": "4",
+        "title": "Solutions of WINGFUZZ",
+        "start": 17266
+      },
+      {
+        "number": "4.1",
+        "title": "Specification-Based Mutator Construction",
+        "start": 17787
+      },
+      {
+        "number": "4.2",
+        "title": "Corpus-Driven Evolving Code Fuzzing",
+        "start": 21110
+      },
+      {
+        "number": "4.3",
+        "title": "Noise-Resilient Anomaly Assessment",
+        "start": 26666
+      },
+      {
+        "number": "5",
+        "title": "Implementation",
+        "start": 28865
+      },
+      {
+        "number": "6",
+        "title": "Evaluation",
+        "start": 31095
+      },
+      {
+        "number": "6.1",
+        "title": "Compared with Existing Fuzzers",
+        "start": 31391
+      },
+      {
+        "number": "6.2",
+        "title": "Contributions of Each Component",
+        "start": 36669
+      },
+      {
+        "number": "6.3",
+        "title": "Practice of Deploying WINGFUZZ",
+        "start": 38537
+      },
+      {
+        "number": "6.3.1",
+        "title": "Process of Deployment",
+        "start": 38738
+      },
+      {
+        "number": "6.3.2",
+        "title": "DBMS Vulnerability Results",
+        "start": 42088
+      },
+      {
+        "number": "6.3.3",
+        "title": "Case Studies",
+        "start": 44272
+      },
+      {
+        "number": "7",
+        "title": "Lessons Learned",
+        "start": 51931
+      },
+      {
+        "number": "8",
+        "title": "Related Work",
+        "start": 55486
+      },
+      {
+        "number": "8.1",
+        "title": "Continuous Fuzzing",
+        "start": 55501
+      },
+      {
+        "number": "8.2",
+        "title": "Generation Based DBMS Fuzzing",
+        "start": 57442
+      },
+      {
+        "number": "8.3",
+        "title": "Mutation Based DBMS Fuzzing",
+        "start": 58627
+      },
+      {
+        "number": "9",
+        "title": "Conclusion",
+        "start": 60454
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Bugs found in database management systems. https:// www .manuelrigger .at/dbms-bugs. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 2,
+      "text": "Clickhouse syntax. https://clickhouse .com/docs/en/sqlreference/syntax. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Continuous integration checks. https://clickhouse .com/docs/en/ development/continuous-integration. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "databases. https://en .wikipedia .org/wiki/Database. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Google chrome impacted by new magellan 2.0 vulnerabilities. https://www .zdnet .com/article/google-chrome-impactedby-new-magellan-2-0-vulnerabilities/. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Postgresql 14beta3 documentation. https://www .postgresql .org/ files/documentation/pdf/14/postgresql-14-A4 .pdf. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Postgresql’ grammar description file. https://github .com/ postgres/postgres/blob/master/src/backend/parser/ gram .y. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Sql commands. https://www .postgresql .org/docs/13/sqlcommands .html. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Sqlite patches use-after-free bug that left apps open to code execution, denial-of-service exploits. https://portswigger .net/ daily-swig/sqlite-patches-use-after-free-bug-thatleft-apps-open-to-code-execution-denial-of-serviceexploits. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Sqlsmith score list. https://github .com/anse1/sqlsmith/ wiki#score-list. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "BÖHME, M., PHAM, V.-T., NGUYEN, M.-D., ANDROYCHOUDHURY, A.Directed greybox fuzzing. In ACM Conference on Computer and Communications Security (CCS) (2017).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "CHEN, H., XUE, Y., L I, Y., CHEN, B., XIE, X., W U, X., ANDLIU, Y. Hawkeye: Towards a desired directed grey-box fuzzer. In Proceedings of the 2018 ACMSIGSAC conference on computer and communications security (2018), pp. 2095–2108.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Query billions of rows in milliseconds. https://clickhouse .com/, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Clickhouse server 23.7.1.857 terminated by sigabrt through create table and select stmts. https://github .com/ClickHouse/ClickHouse/ issues/52049, 7 2023. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "CORONEL, C., ANDMORRIS, S. Database systems: design, implementation and management. Cengage learning, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Damengdb. https://en .dameng .com/, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "DUVALL, P. M., MATYAS, S., ANDGLOVER, A. Continuous integration: improving software quality and reducing risk. Pearson Education, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "FOWLER, M., ANDFOEMMEL, M. Continuous integration, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "FU, J., LIANG, J., W U, Z., ANDJIANG, Y. Sedar: Obtaining highquality seeds for DBMS fuzzing via cross-dbms SQL transfer. In Proceedings of the 46th IEEE/ACM International Conference on Software Engineering, ICSE 2024, Lisbon, Portugal, April 14-20, 2024 (2024), ACM, pp. 146:1–146:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "FU, J., LIANG, J., W U, Z., WANG, M., ANDJIANG, Y. Griffin: Grammar-free dbms fuzzing. In Conference on Automated Software Engineering (ASE’22) (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "HIPP, R. Potential database corruption in sqlite versions 3.35.0 through 3.37.1. https://sqlite .org/forum/forumpost/ ac381d64d804407e?raw, 1 2022. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Security oriented fuzzer with powerful analysis options. https:// github .com/google/honggfuzz. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "JONATHAN METZMAN, G. O. S. S. T. Clusterfuzzlite: Continuous fuzzing for all. https://github .com/google/clusterfuzzlite, 11 2021. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "JUNG, J., H U, H., ARULRAJ, J., KIM, T., ANDKANG, W. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems (to appear). In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB) (Tokyo, Japan, Aug. 2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "KLOOSTER, T., TURKMEN, F., BROENINK, G., HOVE, R. T., AND BÖHME, M. Continuous fuzzing: A study of the effectiveness and scalability of fuzzing in ci/cd pipelines. In 2023 IEEE/ACM International Workshop on Search-Based and Fuzz Testing (SBFT) (2023), pp. 25–32.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "KUMAR, V., ANDSON, S. H. Database recovery, vol. 12. Springer Science & Business Media, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "LIANG, J., CHEN, Y., W U, Z., F U, J., WANG, M., JIANG, Y., HUANG, X., CHEN, T., WANG, J., ANDLI, J.Sequence-oriented dbms fuzzing. In IEEE.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "LIANG, J., WANG, M., CHEN, Y., JIANG, Y., ANDZHANG, R. Fuzz testing in practice: Obstacles and solutions. In 2018 IEEE 25th International Conference on Software Analysis, Evolution and Reengineering (SANER) (Los Alamitos, CA, USA, mar 2018), IEEE Computer Society, pp. 562–566.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "LIANG, J., WANG, M., ZHOU, C., W U, Z., JIANG, Y., LIU, J., LIU, Z., ANDSUN, J. Pata: Fuzzing with path aware taint analysis. In 2022. IEEE Computer Society, Los Alamitos, CA, USA (2022), pp. 154–170.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "LIANG, J., W U, Z., F U, J., WANG, M., SUN, C., ANDJIANG, Y. Mozi: Discovering DBMS bugs via configuration-based equivalent transformation. In Proceedings of the 46th IEEE/ACM International Conference on Software Engineering, ICSE 2024, Lisbon, Portugal, April 14-20, 2024 (2024), ACM, pp. 135:1–135:12.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Libfuzzer. https://www .llvm .org/docs/LibFuzzer .html. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "LIU, X., ZHOU, Q., ARULRAJ, J., ANDORSO, A. Automatic detection of performance bugs in database systems using equivalent queries. InProceedings of the 44th International Conference on Software Engineering (2022), pp. 225–236.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "MANÈS, V. J., HAN, H., HAN, C., CHA, S. K., EGELE, M., SCHWARTZ, E. J., ANDWOO, M. The art, science, and engineering of fuzzing: A survey. IEEE Transactions on Software Engineering 47, 11 (2019), 2312–2331.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Mariadb. https://mariadb .org/, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "MILLER, B. P., FREDRIKSEN, L., ANDSO, B.An Empirical Study of the Reliability of UNIX Utilities. Commun. ACM 33, 12 (Dec. 1990).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "The database system to speed up your analytical jobs. https:// www .monetdb .org/, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Mysql. https://www .mysql .com/, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Polardb. https://www .alibabacloud .com/product/polardb, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Postgresql. https://www .postgresql .org/, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "RIGGER, M., ANDSU, Z.Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (2020), pp. 1140–1152. USENIX Association 2024 USENIX Annual Technical Conference 491",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 41,
+      "text": "RIGGER, M., ANDSU, Z.Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 42,
+      "text": "RIGGER, M., ANDSU, Z.Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20) (2020), pp. 667–682.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "SELTENREICH, A., TANG, B., ANDMULLENDER, S. Sqlsmith: a random sql query generator.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "SEREBRYANY, K. OSS-Fuzzgoogle’s continuous fuzzing service for open source software. USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "SHAHIN, M., BABAR, M. A., ANDZHU, L.Continuous integration, delivery and deployment: a systematic review on approaches, tools, challenges and practices. IEEE access 5 (2017), 3909–3943.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "SLUTZ, D. R. Massive stochastic testing of SQL. In VLDB’98, Proceedings of 24rd International Conference on Very Large Data Bases, New York, USA, Morgan Kaufmann, pp. 618–622.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Sqlite home page. https://www .sqlite .org/index .html, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Embrace industry 4.0 with tdengine — the next generation data historian. https://tdengine .com/, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Tendb cluster. https://github .com/Tencent/TenDBClusterTenDB, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "TRAN, S. Fuzzing collaboration with wingfuzzpart 1. https:// clickhouse .com/blog/fuzzing-wingfuzz, 8 2023. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "VANDERLANS, R. F. The SQL standard: a complete guide reference. Prentice Hall International (UK) Ltd., 1989.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Vastbase. https://www .vastdata .com .cn/, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "VERHOFSTAD, J. S. Recovery techniques for database systems. ACM Computing Surveys (CSUR) 10, 2 (1978), 167–195.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "WANG, M., W U, Z., X U, X., LIANG, J., ZHOU, C., ZHANG, H., ANDJIANG, Y.Industry practice of coverage-guided enterprise-level dbms fuzzing. In Software Engineering: Software Engineering in Practice (ICSE-SEIP) (2021), IEEE, pp. 328–337.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "WIDENIUS, M., AXMARK, D., ANDARNO, K. MySQL reference manual: documentation from the source. \" O’Reilly Media, Inc.\", 2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "WU, Z., LIANG, J., WANG, M., ZHOU, C., ANDJIANG, Y.Unicorn: Detect runtime errors in time-series databases with hybrid input synthesis. In Symposium on Software Testing and Analysis (ISSTA’22) (2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Yashandb. https://www .yashandb .com/, 1 2024. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "YOO, H., HONG, J., BADER, L., HWANG, D. W., ANDHONG, S.Improving configurability of unit-level continuous fuzzing: An industrial case study with sap hana. In 2021 36th IEEE/ACM International Conference on Automated Software Engineering (ASE) (2021), pp. 1101– 1105.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "ZALEWSKI, M. american fuzzy lop. http://lcamtuf .coredump .cx/ afl/. Accessed: June 7, 2024.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "ZHANG, J., CUI, Z., CHEN, X., YANG, H., ZHENG, L., ANDLIU, J. Cidfuzz: Fuzz testing for continuous integration. IET Software 17, 3 (apr 2023), 301–315.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "ZHONG, R., CHEN, Y., H U, H., ZHANG, H., LEE, W., ANDWU, D. Squirrel: Testing database management systems with language validity and coverage feedback. In The ACM Conference on Computer and Communications Security (CCS), 2020 (2020).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "ZHU, X., AND BÖHME, M. Regression greybox fuzzing. In Proceedings of the 2021 ACMSIGSAC Conference on Computer and Communications Security (New York, NY, USA, 2021), CCS ’21, Association for Computing Machinery, p. 2169–2182.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "ZHU, X., WEN, S., CAMTEPE, S., ANDXIANG, Y.Fuzzing: a survey for roadmap. ACM Computing Surveys (CSUR) 54, 11s (2022), 1–36. 492 2024 USENIX Annual Technical Conference USENIX Association",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 1,
+      "text": "Bugs found in database management systems. https:// www .manuelrigger .at/dbms-bugs. Accessed: June 7, 2024.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 40,
+      "text": "RIGGER, M., ANDSU, Z.Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (2020), pp. 1140–1152. USENIX Association 2024 USENIX Annual Technical Conference 491",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 41,
+      "text": "RIGGER, M., ANDSU, Z.Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages 4, OOPSLA (2020), 1–30.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "RIGGER, M., ANDSU, Z.Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20) (2020), pp. 667–682.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[1, 5, 9]",
+      "technique": null,
+      "sentence": "Consequently, the resilience and security of DBMSs have become paramount considerations for organizations and corporations [1, 5, 9].",
+      "context_before": "” and WINGFUZZ has been successfully integrated into its development process. 1 Introduction Modern software applications rely heavily on database management systems (DBMSs) to efficiently store and manage data [4, 15]. Due to the complexity of DBMSs and the rapid iteration of versions, bugs are inevitably present. Once these ∗Jie Liang and Zhiyong Wu contributed equally to this work. †Yu Jiang is the corresponding author.bugs are exploited by attackers, it may lead to a series of problems such as information leakage, system crashes, and disruptions to the operation of upper-tier applications.",
+      "context_after": "Both the industry and academia have developed many methods for detecting issues in DBMSs [24, 32, 42, 46]. Among them, fuzzing [33, 35, 63] is an effective way to assess the security of DBMSs [43, 54, 61], revealing lots of vulnerabilities. Its basic process is to generate a stream of malformed inputs and monitor the exception behaviors [28, 29, 31, 59]. To test a DBMS, fuzzers typically model the specific SQL specification [51] for each DBMS to generate structured SQL inputs that are both syntactically and semantically correct. For example, SQLsmith [43] models the grammar of PostgreSQL [39]",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 2903,
+      "cited_reference": {
+        "number": 1,
+        "text": "Bugs found in database management systems. https:// www .manuelrigger .at/dbms-bugs. Accessed: June 7, 2024.",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[24, 32, 42, 46]",
+      "technique": "pqs",
+      "sentence": "Both the industry and academia have developed many methods for detecting issues in DBMSs [24, 32, 42, 46].",
+      "context_before": "on database management systems (DBMSs) to efficiently store and manage data [4, 15]. Due to the complexity of DBMSs and the rapid iteration of versions, bugs are inevitably present. Once these ∗Jie Liang and Zhiyong Wu contributed equally to this work. †Yu Jiang is the corresponding author.bugs are exploited by attackers, it may lead to a series of problems such as information leakage, system crashes, and disruptions to the operation of upper-tier applications. Consequently, the resilience and security of DBMSs have become paramount considerations for organizations and corporations [1, 5, 9].",
+      "context_after": "Among them, fuzzing [33, 35, 63] is an effective way to assess the security of DBMSs [43, 54, 61], revealing lots of vulnerabilities. Its basic process is to generate a stream of malformed inputs and monitor the exception behaviors [28, 29, 31, 59]. To test a DBMS, fuzzers typically model the specific SQL specification [51] for each DBMS to generate structured SQL inputs that are both syntactically and semantically correct. For example, SQLsmith [43] models the grammar of PostgreSQL [39] as an abstract syntax tree (AST) to generate queries, and it has found about 80 bugs in PostgreSQL since 2",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3037,
+      "cited_reference": {
+        "number": 42,
+        "text": "RIGGER, M., ANDSU, Z.Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation OSDI 20) (2020), pp. 667–682.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "1 Compared with Existing Fuzzers To show the effectiveness of WINGFUZZ, we compared it with three state-of-the-art fuzzers, including conventional mutation-based fuzzer SQUIRREL as well as generated-based fuzzer SQLancer and SQLsmith, which are widely used in the industry to test DBMSs.",
+      "context_before": "e anomaly signals, it means a potential bug is triggered, and we call the API of libunwind-ptrace to retrieve the call stack from the process. On the other hand, we trace the clone(), fork(), vfork(), and exec() of the process to catch all child processes and threads, monitoring them comprehensively. 6 Evaluation In this section, we begin by comparing WINGFUZZwith other fuzzers. Subsequently, we demonstrate the contributions of each component in WINGFUZZ by gradually adding them. Moreover, we present the practice of deploying WINGFUZZ on 12 popular or enterprise-level DBMSs like ClickHouse. 6.",
+      "context_after": "Tested DBMSs and Metrics. We chose four open-source DBMSs as tested DBMSs, namely PostgreSQL, MySQL, MariaDB, and PolarDB, which are widely used in industry and academic research. They were selected because all the compared fuzzers support testing them. Since other fuzzers do not support fuzzing evolving code, we used the latest version of four DBMSs as the target. We evaluated fuzzers using two metrics, namely branches covered and unique bugs triggered. The unique bugs were identified by comparing the call stack and manual analysis. For a fair comparison, when we finished fuzzing, we collect",
+      "section": "6.1 Compared with Existing Fuzzers",
+      "page": 8,
+      "char_offset": 31393,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "They show that WINGFUZZ covers a total of 211620, 197059, 132397 more branches than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+      "context_before": "8 cores (AMDEPYC 7742 Processor @ 2.25 GHz) and 504 GiB of main memory. All DBMS test instances run in Docker containers, each containing a single DBMS service and a DBMS test tool. The test tools operate with default configurations, such as instrumentation methods and initial seed corpus. We run each DBMS test instance for 24 hours, a commonly used time frame for comparative evaluation. For quantitative comparisons, we run the docker containers for each DBMS experiment with 5 CPU cores and 32 GiB of main memory. Branch Coverage. Table 1 shows the branches covered by those fuzzers in 24 hours.",
+      "context_after": "Table 1: Number of branches covered in 24 hours. DBMS SQLancer SQLsmith SQUIRREL WINGFUZZ PostgreSQL 61,420 66,821 63,742 84,954 MySQL 69,182 67,291 77,829 104,281 MariaDB 60,031 69,281 91,932 129,832 PolarDB 56,928 58,729 93,281 140,114 Total 247,561 262,122 326,784 459,181 Increment 211,620 197,059 132,397 – The main reason contributing to the enhanced coverage is that WINGFUZZ supports more SQL grammar of each DBMS than other fuzzers. Specifically, SQLancer generates queries based on custom pattern rules, limiting its support to the SQL grammar associated with the defined test oracle. Simil",
+      "section": "6.1 Compared with Existing Fuzzers",
+      "page": 8,
+      "char_offset": 33065,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "DBMS SQLancer SQLsmith SQUIRREL WINGFUZZ PostgreSQL 61,420 66,821 63,742 84,954 MySQL 69,182 67,291 77,829 104,281 MariaDB 60,031 69,281 91,932 129,832 PolarDB 56,928 58,729 93,281 140,114 Total 247,561 262,122 326,784 459,181 Increment 211,620 197,059 132,397 – The main reason contributing to the enhanced coverage is that WINGFUZZ supports more SQL grammar of each DBMS than other fuzzers.",
+      "context_before": "The test tools operate with default configurations, such as instrumentation methods and initial seed corpus. We run each DBMS test instance for 24 hours, a commonly used time frame for comparative evaluation. For quantitative comparisons, we run the docker containers for each DBMS experiment with 5 CPU cores and 32 GiB of main memory. Branch Coverage. Table 1 shows the branches covered by those fuzzers in 24 hours. They show that WINGFUZZ covers a total of 211620, 197059, 132397 more branches than SQLancer, SQLsmith, and SQUIRREL, respectively. Table 1: Number of branches covered in 24 hours.",
+      "context_after": "Specifically, SQLancer generates queries based on custom pattern rules, limiting its support to the SQL grammar associated with the defined test oracle. Similarly, SQLsmith focuses on the AST model of the DBMS’s SELECT statement, predominantly generating SELECT statements for testing. As a result, it lacks exploration of other DBMS functionalities. Benefiting from feedback guidance and the data dependency graph, SQUIRREL outperforms both SQLancer and SQLsmith in these DBMSs. Nevertheless, it still suffers from the limited support for the SQL grammar of the target DBMSs. For example, SQUIRREL",
+      "section": "6.1 Compared with Existing Fuzzers",
+      "page": 8,
+      "char_offset": 33246,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, SQLancer generates queries based on custom pattern rules, limiting its support to the SQL grammar associated with the defined test oracle.",
+      "context_before": "those fuzzers in 24 hours. They show that WINGFUZZ covers a total of 211620, 197059, 132397 more branches than SQLancer, SQLsmith, and SQUIRREL, respectively. Table 1: Number of branches covered in 24 hours. DBMS SQLancer SQLsmith SQUIRREL WINGFUZZ PostgreSQL 61,420 66,821 63,742 84,954 MySQL 69,182 67,291 77,829 104,281 MariaDB 60,031 69,281 91,932 129,832 PolarDB 56,928 58,729 93,281 140,114 Total 247,561 262,122 326,784 459,181 Increment 211,620 197,059 132,397 – The main reason contributing to the enhanced coverage is that WINGFUZZ supports more SQL grammar of each DBMS than other fuzzers.",
+      "context_after": "Similarly, SQLsmith focuses on the AST model of the DBMS’s SELECT statement, predominantly generating SELECT statements for testing. As a result, it lacks exploration of other DBMS functionalities. Benefiting from feedback guidance and the data dependency graph, SQUIRREL outperforms both SQLancer and SQLsmith in these DBMSs. Nevertheless, it still suffers from the limited support for the SQL grammar of the target DBMSs. For example, SQUIRREL currently supports only a subset of 165 grammar rules for MySQL, whereas the parser of MySQL encompasses 695 grammar rules. In contrast, WINGFUZZ adopts",
+      "section": "6.1 Compared with Existing Fuzzers",
+      "page": 8,
+      "char_offset": 33638,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Benefiting from feedback guidance and the data dependency graph, SQUIRREL outperforms both SQLancer and SQLsmith in these DBMSs.",
+      "context_before": "2 129,832 PolarDB 56,928 58,729 93,281 140,114 Total 247,561 262,122 326,784 459,181 Increment 211,620 197,059 132,397 – The main reason contributing to the enhanced coverage is that WINGFUZZ supports more SQL grammar of each DBMS than other fuzzers. Specifically, SQLancer generates queries based on custom pattern rules, limiting its support to the SQL grammar associated with the defined test oracle. Similarly, SQLsmith focuses on the AST model of the DBMS’s SELECT statement, predominantly generating SELECT statements for testing. As a result, it lacks exploration of other DBMS functionalities.",
+      "context_after": "Nevertheless, it still suffers from the limited support for the SQL grammar of the target DBMSs. For example, SQUIRREL currently supports only a subset of 165 grammar rules for MySQL, whereas the parser of MySQL encompasses 695 grammar rules. In contrast, WINGFUZZ adopts a specification-based mutator construction approach to maximize support for the entire SQL grammar of the target DBMSs. The process begins by extracting the complete grammar rules and tokens from the specification of the target DBMSs, which are then translated into uniform intermediate grammar rules. Subsequently, WINGFUZZ ge",
+      "section": "6.1 Compared with Existing Fuzzers",
+      "page": 8,
+      "char_offset": 33989,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Specifically, WINGFUZZ finds a total of 25, 24, 21 more unique bugs than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+      "context_before": "on, which could help maintain the syntax and semantic correctness of the SQL queries during mutation. More comprehensive support for SQL grammar allows WINGFUZZ to effectively trigger more complex logic within the target DBMS. For instance, WINGFUZZaccommodates all 695 SQL USENIX Association 2024 USENIX Annual Technical Conference 485 grammar rules of MySQL, ultimately leading to the discovery of 26452 more branches than SQUIRREL on MySQL. Unique Bugs. Besides branch coverage, WINGFUZZ also triggered more bugs than other fuzzers in 24 hours. Table 2 shows the number of bugs found by each tool.",
+      "context_after": "Table 2: Number of reproduced bugs triggered in 24 hours. DBMS SQLancer SQLsmith SQUIRREL WINGFUZZ PostgreSQL 0 1 1 2 MySQL 1 1 1 7 MariaDB 0 0 2 7 PolarDB 1 1 2 11 Total 2 3 6 27 Increment 25 24 21 – The enhancement in bug triggering primarily stems from two factors. Firstly, there is an improvement in coverage. The increased branch coverage implies that WINGFUZZ can encompass unique functionalities within the target DBMS, where potential bugs might be located. Secondly, WINGFUZZ utilizes a noise-resilient anomaly assessment. Unlike other fuzzers that may ignore anomalies due to the noise br",
+      "section": "6.1 Compared with Existing Fuzzers",
+      "page": 9,
+      "char_offset": 35478,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "DBMS SQLancer SQLsmith SQUIRREL WINGFUZZ PostgreSQL 0 1 1 2 MySQL 1 1 1 7 MariaDB 0 0 2 7 PolarDB 1 1 2 11 Total 2 3 6 27 Increment 25 24 21 – The enhancement in bug triggering primarily stems from two factors.",
+      "context_before": "gger more complex logic within the target DBMS. For instance, WINGFUZZaccommodates all 695 SQL USENIX Association 2024 USENIX Annual Technical Conference 485 grammar rules of MySQL, ultimately leading to the discovery of 26452 more branches than SQUIRREL on MySQL. Unique Bugs. Besides branch coverage, WINGFUZZ also triggered more bugs than other fuzzers in 24 hours. Table 2 shows the number of bugs found by each tool. Specifically, WINGFUZZ finds a total of 25, 24, 21 more unique bugs than SQLancer, SQLsmith, and SQUIRREL, respectively. Table 2: Number of reproduced bugs triggered in 24 hours.",
+      "context_after": "Firstly, there is an improvement in coverage. The increased branch coverage implies that WINGFUZZ can encompass unique functionalities within the target DBMS, where potential bugs might be located. Secondly, WINGFUZZ utilizes a noise-resilient anomaly assessment. Unlike other fuzzers that may ignore anomalies due to the noise brought by recovery mechanisms, WINGFUZZ directly monitors the exceptional signals from each thread of the target DBMS. Moreover, during the bug reproduction process, we observed that many bugs from other fuzzers cannot be triggered using their recorded inputs, due to th",
+      "section": "6.1 Compared with Existing Fuzzers",
+      "page": 9,
+      "char_offset": 35657,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [40 –42] designs three test oracles to detect logic errors and generates queries following the oracles.",
+      "context_before": "he latest version and provides the generated corpus as a valuable source for rapid commit testing. Additionally, WINGFUZZ considers the important DBMS characteristics, utilizing customized mutators and conducting noise-resilient anomaly assessments to ensure a thorough and accurate bug finding. 8.2 Generation Based DBMS Fuzzing Generation-based fuzzers generate SQL test cases following the pre-defined generating rules. For example, SQLsmith [43] models SQL specifications into an AST model and generates amounts of queries. It checks bugs by monitoring if they cause disconnections to the server.",
+      "context_after": "For example, its NOREC [40] oracle requires constructing a SQL query with WHERE and JOIN clauses. Then it translates the query to an equivalent one by moving the condition in the WHERE clause after the SELECT .MOZI[30] utilize configuration-based equivalent transformation to find logic and performance bugs. APOLLO [24] utilizes SQLsmith as the generator to construct SQL queries and perform regression testing between different DBMS versions to detect performance anomalies. Different from them, WINGFUZZ is a mutation-based fuzzing framework. It leverages coverage to guide the query generation pr",
+      "section": "8.2 Generation Based DBMS Fuzzing",
+      "page": 13,
+      "char_offset": 57746,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "For example, its NOREC [40] oracle requires constructing a SQL query with WHERE and JOIN clauses.",
+      "context_before": "WINGFUZZ considers the important DBMS characteristics, utilizing customized mutators and conducting noise-resilient anomaly assessments to ensure a thorough and accurate bug finding. 8.2 Generation Based DBMS Fuzzing Generation-based fuzzers generate SQL test cases following the pre-defined generating rules. For example, SQLsmith [43] models SQL specifications into an AST model and generates amounts of queries. It checks bugs by monitoring if they cause disconnections to the server. SQLancer [40 –42] designs three test oracles to detect logic errors and generates queries following the oracles.",
+      "context_after": "Then it translates the query to an equivalent one by moving the condition in the WHERE clause after the SELECT .MOZI[30] utilize configuration-based equivalent transformation to find logic and performance bugs. APOLLO [24] utilizes SQLsmith as the generator to construct SQL queries and perform regression testing between different DBMS versions to detect performance anomalies. Different from them, WINGFUZZ is a mutation-based fuzzing framework. It leverages coverage to guide the query generation process. In contrast to the mentioned fuzzers, WINGFUZZ automatically adapts the grammar of a new D",
+      "section": "8.2 Generation Based DBMS Fuzzing",
+      "page": 13,
+      "char_offset": 57859,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M3",
+        "M4",
+        "M8"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T14:25:35Z",
+    "is_model_written": true,
+    "summary": "WingFuzz implements continuous fuzzing for DBMSs, addressing the problem that a fuzzer must keep working as the system under test changes. It combines mutation with feedback guidance and a memory-sanitisation layer, and reports finding 27 unique bugs across PostgreSQL, MySQL, MariaDB and PolarDB during continuous operation.",
+    "narrative": "SQLancer is one of three state-of-the-art fuzzers WingFuzz is measured against, with branch counts and bug counts reported per DBMS -- 211,620 more branches and 25 more unique bugs. The paper attributes SQLancer's lower coverage to its generating queries from custom pattern rules tied to the grammar its oracles need, and describes its three oracles, noting NoREC requires a query with WHERE and JOIN clauses.",
+    "roles": {
+      "M1": "background",
+      "M2": "state_of_the_art",
+      "M3": "result_comparison",
+      "M4": "result_comparison",
+      "M5": "motivation",
+      "M6": "result_comparison",
+      "M7": "result_comparison",
+      "M8": "result_comparison",
+      "M9": "definition",
+      "M10": "definition"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M2",
+          "M3",
+          "M7"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "1 Compared with Existing Fuzzers To show the effectiveness of WINGFUZZ, we compared it with three state-of-the-art fuzzers, including conventional mutation-based fuzzer SQUIRREL as well as generated-based fuzzer SQLancer and SQLsmith, which are widely used in the industry to test DBMSs.",
+            "section": "6.1 Compared with Existing Fuzzers",
+            "page": 8
+          },
+          {
+            "mention_id": "M3",
+            "sentence": "They show that WINGFUZZ covers a total of 211620, 197059, 132397 more branches than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+            "section": "6.1 Compared with Existing Fuzzers",
+            "page": 8
+          },
+          {
+            "mention_id": "M7",
+            "sentence": "Specifically, WINGFUZZ finds a total of 25, 24, 21 more unique bugs than SQLancer, SQLsmith, and SQUIRREL, respectively.",
+            "section": "6.1 Compared with Existing Fuzzers",
+            "page": 9
+          }
+        ],
+        "reasoning": "M2 names SQLancer among the three fuzzers compared, and M3 and M7 give the results: 211,620 more branches and 25 more unique bugs than SQLancer."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M2"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M2",
+            "sentence": "1 Compared with Existing Fuzzers To show the effectiveness of WINGFUZZ, we compared it with three state-of-the-art fuzzers, including conventional mutation-based fuzzer SQUIRREL as well as generated-based fuzzer SQLancer and SQLsmith, which are widely used in the industry to test DBMSs.",
+            "section": "6.1 Compared with Existing Fuzzers",
+            "page": 8
+          }
+        ],
+        "reasoning": "M2 calls the three compared fuzzers, SQLancer among them, state-of-the-art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.json b/_data/papers/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.json
new file mode 100644
index 0000000..20a682e
--- /dev/null
+++ b/_data/papers/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.json
@@ -0,0 +1,1065 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:05Z",
+  "paper": {
+    "id": "paper:s2:3a7373bd891702ae93d7e058241a7e8be25e89eb",
+    "title": "DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation",
+    "authors": [
+      "Zu-Ming Jiang",
+      "Jia-Ju Bai",
+      "Zhendong Su"
+    ],
+    "year": 2023,
+    "venue": "USENIX Security Symposium",
+    "doi": null,
+    "arxiv_id": null,
+    "s2_paper_id": "3a7373bd891702ae93d7e058241a7e8be25e89eb",
+    "url": "https://www.semanticscholar.org/paper/3a7373bd891702ae93d7e058241a7e8be25e89eb",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "usenix",
+      "url": "https://www.usenix.org/system/files/usenixsecurity23-jiang-zu-ming.pdf",
+      "retrieved_at": "2026-09-09T01:33:05Z",
+      "chars": 79652,
+      "content_sha256": "sha256:19e7312a28d4011ce248dabb79990df074a027421ef2547fa06d1e8c6c90590c"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 18,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2208
+      },
+      {
+        "number": "2",
+        "title": "Background and Motivation",
+        "start": 8988
+      },
+      {
+        "number": "3",
+        "title": "Stateful DBMS Fuzzing",
+        "start": 15154
+      },
+      {
+        "number": "3.1",
+        "title": "Dynamic Query Interaction",
+        "start": 16807
+      },
+      {
+        "number": "3.2",
+        "title": "Error Feedback",
+        "start": 25888
+      },
+      {
+        "number": "4",
+        "title": "Framework and Implementation",
+        "start": 27924
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 31705
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Setup",
+        "start": 32238
+      },
+      {
+        "number": "5.2",
+        "title": "Runtime Testing",
+        "start": 32837
+      },
+      {
+        "number": "5.3",
+        "title": "Security Impact",
+        "start": 39571
+      },
+      {
+        "number": "5.4",
+        "title": "Sensitivity Analysis",
+        "start": 44203
+      },
+      {
+        "number": "5.5",
+        "title": "Comparison to Existing DBMS Fuzzers",
+        "start": 50999
+      },
+      {
+        "number": "6",
+        "title": "Limitations and Future Work",
+        "start": 57509
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 60869
+      },
+      {
+        "number": "7.1",
+        "title": "DBMS Testing",
+        "start": 60884
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 65502
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "B. Acohido. Small banks and credit union attack set for tuesday, 2013. https://www.usatoday.com/story /cybertruth/2013/05/06/ddos-denial-of-ser vice-small-business-cybersecurity-privacy /2139349/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "American fuzzy lop. https://github.com/google/ AFL.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "ASan: address sanitizer. https://github.com/goo gle/sanitizers/wiki/AddressSanitizer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. Coverage-based greybox fuzzing as markov chain. InProceedings of the 23rd International Conference on Computer and Communications Security (CCS), pages 1032–1043, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "E. V. Buskirk. Facebook confirms denial-of-service attack, 2009. https://www.wired.com/2009/08/ facebook-apparently-attacked-in-addition-t o-twitter/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Peng Chen and Hao Chen. Angora: efficient fuzzing by principled search. In Proceedings of the 2018 Symposium on Security and Privacy (S&P), pages 711–725, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "C. Cimpanu. Google chrome impacted by new magellan 2.0 vulnerabilities, 2019. https://www.zdnet.com/ article/google-chrome-impacted-by-new-mag ellan-2-0-vulnerabilities/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Clang: a llvm-based compiler for C/C++ programs. https://clang.llvm.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "ClickHouse. https://clickhouse.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "CLOC: count lines of code. https://cloc.sourcef orge.net/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Chris Cummins, Pavlos Petoumenos, Alastair Murray, and Hugh Leather. Compiler fuzzing through deep learning. In Proceedings of the 27th International Symposium on Software Testing and Analysis (ISSTA), pages 95–105, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "DB-Engines Ranking. https://db-engines.com/e n/ranking.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "UM Fayyad. Data science revealed: A data-driven glimpse into the burgeoning new field, 2011. https: //embed.cs.utah.edu/creduce/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "L. Franceschi-Bicchierai. Hacker tries to sell 427 milllion stolen myspace passwords for $2,800, 2016. https://www.vice.com/en/article/pgkk8v /427-million-myspace-passwords-emails-dat a-breach.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Shuitao Gan, Chao Zhang, Peng Chen, Bodong Zhao, Xiaojun Qin, Dong Wu, and Zuoning Chen. GREYONE: data flow sensitive fuzzing. In Proceedings of the 29th USENIX Security Symposium, pages 2577–2594, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Patrice Godefroid, Hila Peleg, and Rishabh Singh. Learn&Fuzz: machine learning for input fuzzing. In Proceedings of the 32nd International Conference on Automated Software Engineering (ASE), pages 50–59, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, Kangjie Lu, and Shi-Min Hu. Fuzzing error handling code using context-sensitive software fault injection. In Proceedings of the 29th USENIX Security Symposium, pages 2595–2612, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. APOLLO: automatic detection and diagnosis of performance regressions in database systems. InProceedings of the 46th International Conference on Very Large Data Bases (VLDB), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Shadi Abdul Khalek, Bassem Elkarablieh, Yai O. Laleye, and Sarfraz Khurshid. Query-aware test generation using a relational constraint solver. In Proceedings of the 23rd International Conference on Automated Software Engineering (ASE), pages 238–247, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. Automated SQL query generation for systematic testing of database engines. In Proceedings of the 2010 International Conference on Automated Software Engineering (ASE), pages 329–332, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "George Klees, Andrew Ruef, Benji Cooper, Shiyi Wei, and Michael Hicks. Evaluating fuzz testing. In Proceedings of the 2018 International Conference on Computer and Communications Security (CCS), pages 2123–2138, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Doug Laney et al. 3d data management: Controlling data volume, velocity and variety. META group research note, 6(70):1, 2001. 4962 32nd USENIX Security Symposium USENIX Association",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Guoliang Li, Xuanhe Zhou, Shifu Li, and Bo Gao. QTune: a query-aware database tuning system with deep reinforcement learning. Proceedings of the VLDB Endowment, 12(12):2118–2130, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "libFuzzera library for coverage-guided fuzz testing. https://llvm.org/docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. Automated performance bug detection in database systems. arXiv preprint arXiv:2105.10016, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Chenyang Lyu, Shouling Ji, Chao Zhang, Yuwei Li, WeiHan Lee, Yu Song, and Raheem Beyah. MOPT: optimized mutation scheduling for fuzzers. In Proceedings of the 28th USENIX Security Symposium, pages 1949– 1966, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "MariaDB. https://www.mariadb.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "MonetDB. https://www.monetdb.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "MySQL. https://www.mysql.com/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Mike Papadakis, and Yves Le Traon. Semantic fuzzing with zest. In Proceedings of the 28th International Symposium on Software Testing and Analysis (ISSTA), pages 329–340, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "PostgreSQL. https://www.postgresql.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Sameer Reddy, Caroline Lemieux, Rohan Padhye, and Koushik Sen. Quickly generating diverse valid test inputs with reinforcement learning. In Proceedings of the 42nd International Conference on Software Engineering (ICSE), pages 1410–1421, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "John Regehr, Yang Chen, Pascal Cuoq, Eric Eide, Chucky Ellison, and Xuejun Yang. Test-case reduction for C compiler bugs. In Proceedings of the 2012 International Conference on Programming Language Design and Implementation, pages 335–346, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Kim-Thomas Rehmann, Changyun Seo, Dongwon Hwang, Binh Than Truong, Alexander Boehm, and Dong Hun Lee. Performance monitoring in sap hana’s continuous integration process. ACMSIGMETRICS Performance Evaluation Review, 43(4):43–52, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESE/FSE), pages 1140–1152, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA):1– 30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), pages 667–682, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Avi Silberschatz, Henry F. Korth, and S. Sudarshan. Database System Concepts, Seventh Edition. McGrawHill Book Company, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Donald R. Slutz. Massive stochastic testing of SQL. In Proceedings of 24rd International Conference on Very Large Data Bases (VLDB), pages 618–622, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Database Language SQL, 1992. http://www.contri b.andrew.cmu.edu/~shadow/sql/sql1992.txt.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "SQLancer. https://github.com/sqlancer/sqla ncer.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 42,
+      "text": "SQLite. https://www.sqlite.org/index.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "SQLsmith. https://github.com/anse1/sqlsmi th.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Michael Stonebraker, Sam Madden, and Pradeep Dubey. Intel\" big data\" science and technology center vision and execution plan. ACMSIGMOD Record, 42(1):44–49, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. Industry practice of coverage-guided enterprise-level DBMS fuzzing. InProceedings of the 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSESEIP), pages 328–337, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Yahoo says all three billion accounts hacked in 2013 data theft, 2017. https://www.reuters.com/arti cle/us-yahoo-cyber/yahoo-says-all-three-b illion-accounts-hacked-in-2013-data-theft -id USKCN1C82O1.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Insu Yun, Sangho Lee, Meng Xu, Yeongjin Jang, and Taesoo Kim. QSYM: a practical concolic execution engine tailored for hybrid fuzzing. In Proceedings of the 27th USENIX Security Symposium, pages 745–761, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Association 32nd USENIX Security Symposium 4963 A Examples of Bug-Triggering Queries Listing 1-6 show 6 ",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 35,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESE/FSE), pages 1140–1152, 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA):1– 30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), pages 667–682, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 41,
+      "text": "SQLancer. https://github.com/sqlancer/sqla ncer.",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 48,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Association 32nd USENIX Security Symposium 4963 A Examples of Bug-Triggering Queries Listing 1-6 show 6 generated queries that trigger 6 bugs, respectively. Note that many bugs found by DynSQL have not been fixed yet, and related developers (e.g. developers in MySQL) hope that we do not publish the malicious queries triggering unfixed bugs, in order to protect their customers. Considering their concerns, we choose only the fixed bugs and show their corresponding queries. These 6 selected bugs are missed by both SQUIRREL and SQLsmith, and the queries have been simplified by us and related developers. 1CREATE TABLE t1 (i1 INT); 2INSERT INTO t1 VALUES (1),(2),(3); 3CREATE VIEW v1 AS 4 SELECT t1.i1 5 FROM ( 6 t1 a JOIN t1 ON ( 7 t1.i1 = ( 8 SELECT t1.i1 9 FROM t1 b))); 10SELECT 1 11FROM ( 12 SELECT count (SELECT i1 FROM v1) 13 FROM v1 14 ) dt1; Listing 1: Generated query that crashes MariaDB 5.5-10.5 1CREATE TABLE t1 (i1 INTPRIMARY KEY); 2INSERT INTO t1 VALUES (62),(66); 3CREATE TABLE t2 (i1 INT); 4SELECT 1 5FROM t1 6WHERE t1.i1 = ( 7 SELECT t1.i1 8 FROM t2 9 UNION 10 SELECT dt1.i1 11 FROM (t1 AS dt1) 12 WINDOW w1 AS (PARTITION BY t1.i1) 13 LIMIT 1 14 ); Listing 2: Generated query that crashes MariaDB 10.2-10.5 1CREATE TABLE t1 (a INT NOT NULLPRIMARY KEY); 2INSERT INTO t1 VALUES (0),(4),(31); 3CREATE TABLE t2 (i INT); 4DELETE FROM t1 5WHERE t1.a = ( 6 SELECT t1.a 7 FROM t2 8 UNION 9 SELECT DISTINCT 52 10 FROM t2 r 11 WHERE t1.a = t1.a 12 ); Listing 3: Generated query that crashes MariaDB 10.2-10.5 1CREATE TABLE t1 (id int); 2CREATE VIEW v1 AS 3 SELECT 4 b AS a, 5 b AS b 6 FROM (7 SELECT id AS b 8 FROM t1 9 ) AS dt 10 ORDER BY a,b; 11WITH cte AS ( 12 SELECT dt.b 13 FROM ( 14 (SELECT 11 AS b FROM v1) dt 15 JOIN v1 16 ON 1) 17 ) 18SELECT 5; Listing 4: Generated query that crashes MariaDB 10.2-10.5 1CREATE TABLE t1 (a1 TEXT) engine=myisam; 2SELECT c1 FROM ( 3 SELECT DISTINCT 4 t1.a1 AS c1, 5 t1.a1 AS c2, 6 t1.a1 AS c3, 7 ... 8 t1.a1 AS c2591, 9 t1.a1 AS c2592 10 FROM t1 11 ) dt; Listing 5: Generated query that crashes MariaDB 10.2-10.5 1CREATE TABLE t1 ( 2 t1_c0 INTEGER, 3 t1_c1 TEXT, 4 t1_c2 INTEGER, 5 PRIMARY KEY (t1_c0)); 6ALTER TABLE t1 RENAME COLUMN t1_c2 TO t1_c3; 7WITH cte_0 AS ( 8 SELECT 9 ref_0.t1_c1 AS c0 10 FROM 11 t1 AS ref_0 12 GROUP BY ref_0.t1_c1) 13SELECT 14 ref_2.c0 AS c0 15FROM ( 16 (t1 AS ref_1 INNER JOIN cte_0 AS ref_2 17 ON (( 18 SELECT c0 19 FROM cte_0 20 ORDER BY c0 21 LIMIT 1 OFFSET 5 22 ) IS NULL)) 23 INNER JOIN ( 24 SELECT 25 ref_3.c0 AS c0 26 FROM 27 cte_0 AS ref_3 28 WHERE EXISTS ( 29 SELECT 30 (SELECT max(t1_c0) FROM t1) AS c0 31 FROM 32 cte_0 AS ref_4 33 ) AND 0<>0 34 ORDER BY c0 ASCLIMIT 86 35 ) AS subq_0 36 ON ((ref_1.t1_c3 >= ref_1.t1_c3) 37 AND (ref_1.t1_c3 <= ref_1.t1_c0))); Listing 6: Generated query that crashes MySQL 8.0 4964 32nd USENIX Security Symposium USENIX Association BCVE details We applied for CVE IDs of the bugs found by DynSQL, and 19 were assigned, including 7 for MySQL bugs and 12 for MariaDB bugs. Among them, CVE-2021-46667 and CVE-2021-46669 are considered to have high security impact. Table 8 shows the details of the 19 assigned CVEs. The column \"File location\" shows the file name and line number where the bug is triggered, and the column \"Exploitation\" shows the possible attack by exploiting the bug. Table 8: The details of 19 assgined CVEs DBMS Bug type File location Exploitation CVEID MySQL Null-pointer dereference MySQL/sql/item_subselect.cc:799 Denial-of-service CVE-2021-2357 MySQL Null-pointer dereference MySQL/sql/sql_optimizer.cc:8881 Denial-of-service CVE-2021-2425 MySQL Null-pointer dereference MySQL/storage/innobase/dict/dict0dd.cc:4184 Denial-of-service CVE-2021-2426 MySQL Null-pointer dereference MySQL/strings/ctype-utf8.cc:5603 Denial-of-service CVE-2021-2427 MySQL Null-pointer dereference MySQL/sql/item_subselect.cc:660 Denial-of-service CVE-2021-35628 MySQL Null-pointer dereference MySQL/sql/sql_derived.cc:182 Denial-of-service CVE-2021-35635 MySQL Heap-buffer overflow MySQL/sql/sql_optimizer.cc:4231 Data leakage CVE-2022-21438 MariaDB Null-pointer dereference MariaDB/sql/sql_select.cc:25122 Denial-of-service CVE-2021-46657 MariaDB Null-pointer dereference MariaDB/sql/field_conv.cc:204 Denial-of-service CVE-2021-46658 MariaDB Null-pointer dereference MariaDB/sql/sql_lex.cc:2502 Denial-of-service CVE-2021-46659 MariaDB Null-pointer dereference MariaDB/sql/sql_base.cc:6013 Denial-of-service CVE-2021-46661 MariaDB Use-after-free MariaDB/sql/item.cc:7956 Data leakage CVE-2021-46662 MariaDB Null-pointer dereference MariaDB/storage/maria/ha_maria.cc:2656 Denial-of-service CVE-2021-46663 MariaDB Assertion failure MariaDB/sql/sql_select.cc:18576 Denial-of-service CVE-2021-46664 MariaDB Null-pointer dereference MariaDB/sql/sql_select.cc:18647 Denial-of-service CVE-2021-46665 MariaDB Null-pointer dereference MariaDB/sql/item.cc:3333 Denial-of-service CVE-2021-46666 MariaDB Integer overflow MariaDB/sql/sql_lex.cc:3521 Remote code execution CVE-2021-46667 MariaDB Null-pointer dereference MariaDB/storage/maria/ha_maria.cc:2782 Denial-of-service CVE-2021-46668 MariaDB Use-after-free MariaDB/sql/sql_class.cc: 2914 Privilege escalation CVE-2021-46669 USENIX Association 32nd USENIX Security Symposium 4965",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[18, 43, 45,\n48]",
+      "technique": null,
+      "sentence": "Fuzzing is a promising technique for bug detection [2,4,15, 17, 26, 30, 47], and it is applied to testing DBMSs [18, 43, 45, 48] by generating SQL (Structured Query Language) queries that contain a series of SQL statements [48].",
+      "context_before": "uation, DynSQL outperforms other state-of-the-art DBMS fuzzers, achieving 41% higher code coverage and finding many bugs missed by other fuzzers. 1 Introduction Database management systems (DBMSs) play essential roles in modern data-intensive applications [13, 22, 44], providing fundamental functionalities of data storage and management. Due to the large code size and complex logic of DBMSs, bugs are inevitably introduced during development and maintenance. By exploiting DBMS bugs, attackers can introduce malicious threats of paralyzing the system [1, 5, 7] or even hacking secret data [14, 46].",
+      "context_after": "Specifically, some fuzzers [18, 43] utilize well-defined rules to randomly generate SQL queries, feed these queries to target DBMSs, and check whether bugs are triggered. To improve bug detection in DBMSs, several approaches [45, 48] further involve feedback mechanisms. After the execution of each test case, they collect the runtime information (e.g. code coverage) of target DBMSs and check whether interesting behavior (e.g. covering new branches) occurs. If so, the test case will be stored as a seed for later test-case generation. However, existing DBMS fuzzers are still limited in generating",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 2663,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[45, 48]",
+      "technique": null,
+      "sentence": "To improve bug detection in DBMSs, several approaches [45, 48] further involve feedback mechanisms.",
+      "context_before": "s are inevitably introduced during development and maintenance. By exploiting DBMS bugs, attackers can introduce malicious threats of paralyzing the system [1, 5, 7] or even hacking secret data [14, 46].Fuzzing is a promising technique for bug detection [2,4,15, 17, 26, 30, 47], and it is applied to testing DBMSs [18, 43, 45, 48] by generating SQL (Structured Query Language) queries that contain a series of SQL statements [48]. Specifically, some fuzzers [18, 43] utilize well-defined rules to randomly generate SQL queries, feed these queries to target DBMSs, and check whether bugs are triggered.",
+      "context_after": "After the execution of each test case, they collect the runtime information (e.g. code coverage) of target DBMSs and check whether interesting behavior (e.g. covering new branches) occurs. If so, the test case will be stored as a seed for later test-case generation. However, existing DBMS fuzzers are still limited in generating complex and valid queries to find deep bugs in DBMSs. In general, a complex query contains multiple SQL statements that involve various SQL features (e.g. multi-level nested subqueries), while a valid query satisfies the dependencies among its statements (e.g., a subsequ",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3060,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[48]",
+      "technique": null,
+      "sentence": "For example, SQLsmith [43] generates only one statement in each query, avoiding the analysis of dependencies among statements, which sacrifices complexity for validity; SQUIRREL [48] uses an intermediate representation (IR) model to infer dependencies and generate queries that contain multiple statements, but it produces over 50% invalid queries and tends to generate simple statements.",
+      "context_before": "as a seed for later test-case generation. However, existing DBMS fuzzers are still limited in generating complex and valid queries to find deep bugs in DBMSs. In general, a complex query contains multiple SQL statements that involve various SQL features (e.g. multi-level nested subqueries), while a valid query satisfies the dependencies among its statements (e.g., a subsequent statement references the elements defined in an earlier statement) and guarantees both syntactic and semantic correctness. Existing DBMS fuzzers always make a trade-off between complexity and validity of generated queries.",
+      "context_after": "The contradiction between query complexity and query validity occurs because existing DBMS fuzzers heavily rely on their predefined grammar models and fixed knowledge about DBMSs, but do not capture runtime state information. They either neglect state changes (e.g. SQLsmith [43]), or statically infer the corresponding states (e.g. SQUIRREL [48]) but suffer from soundness and completeness issues. Without accurate state information, these fuzzers tend to build incorrect dependencies among statements or misuse SQL features, causing many invalid queries to be generated. To generate valid test USENI",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3985,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[48]",
+      "technique": null,
+      "sentence": "SQUIRREL [48]) but suffer from soundness and completeness issues.",
+      "context_before": "tements, which sacrifices complexity for validity; SQUIRREL [48] uses an intermediate representation (IR) model to infer dependencies and generate queries that contain multiple statements, but it produces over 50% invalid queries and tends to generate simple statements. The contradiction between query complexity and query validity occurs because existing DBMS fuzzers heavily rely on their predefined grammar models and fixed knowledge about DBMSs, but do not capture runtime state information. They either neglect state changes (e.g. SQLsmith [43]), or statically infer the corresponding states (e.g.",
+      "context_after": "Without accurate state information, these fuzzers tend to build incorrect dependencies among statements or misuse SQL features, causing many invalid queries to be generated. To generate valid test USENIX Association 32nd USENIX Security Symposium 4949 Table 1: Conceptual comparison of DBMS fuzzers Features SQLsmith SQUIRREL DynSQL Stateful Fuzzing None Partial Full Query Generation Static Static Dynamic Program Feedback None Code Cov Code Cov+Error Query Validity High Middle High Statement Number One Multiple Multiple Statement Complexity High Low High cases, these fuzzers have to limit the c",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4704,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[48]",
+      "technique": null,
+      "sentence": "Instead, they either generate only one complex statement in each query to avoid the analysis of state changes [18, 43], or combine multiple relatively simple statements where the state changes can be easily inferred [48].",
+      "context_before": "hat cause DBMS state changes. Fuzzers need to capture such changes so that the subsequent statements can correctly reference the elements built by the earlier statements. Second, these statements utilize various SQL features, such as sub SELECT statement, CTE,CROSS JOIN, etc, which make it hard for fuzzers to accurately infer possible state changes.Limitations of existing DBMS fuzzers. Existing DBMS fuzzers are limited in generating complex and valid SQL queries (e.g. the malicious query in Figure 1), because they cannot accurately capture the DBMS state changes caused by generated statements.",
+      "context_after": "SQLsmith [43], a popular grammar-based DBMS fuzzer, can generate complex SQL statements using its well-defined abstract syntax tree (AST) rules; but it is stateless without considering state changes caused by its generated statements, as a result of which it cannot build the dependencies among multiple statements and thus generates only one statement in each query. Using an intermediate representation (IR) to maintain query structures, SQUIRREL [48] is aware of state changes caused by its generated statements. It considers various SQL features and maintains the scopes and lives of multiple var",
+      "section": "2 Background and Motivation",
+      "page": 4,
+      "char_offset": 13753,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[48]",
+      "technique": null,
+      "sentence": "Using an intermediate representation (IR) to maintain query structures, SQUIRREL [48] is aware of state changes caused by its generated statements.",
+      "context_before": "statements. Instead, they either generate only one complex statement in each query to avoid the analysis of state changes [18, 43], or combine multiple relatively simple statements where the state changes can be easily inferred [48]. SQLsmith [43], a popular grammar-based DBMS fuzzer, can generate complex SQL statements using its well-defined abstract syntax tree (AST) rules; but it is stateless without considering state changes caused by its generated statements, as a result of which it cannot build the dependencies among multiple statements and thus generates only one statement in each query.",
+      "context_after": "It considers various SQL features and maintains the scopes and lives of multiple variables in statements, causing its IR mechanism to be complicated and error-prone when inferring state changes caused by complex statements. To mitigate this impact, SQUIRREL tends to generate simple statements in queries. Even so, SQUIRREL still generates over 50% invalid queries [48]. Without accurate DBMS state information, existing fuzzers are limited in generating complex and valid queries to extensively test DBMSs, causing many deep bugs to still exist. Therefore, proposing a practical solution to address",
+      "section": "2 Background and Motivation",
+      "page": 4,
+      "char_offset": 14342,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[48]",
+      "technique": null,
+      "sentence": "Even so, SQUIRREL still generates over 50% invalid queries [48].",
+      "context_before": "d statements, as a result of which it cannot build the dependencies among multiple statements and thus generates only one statement in each query. Using an intermediate representation (IR) to maintain query structures, SQUIRREL [48] is aware of state changes caused by its generated statements. It considers various SQL features and maintains the scopes and lives of multiple variables in statements, causing its IR mechanism to be complicated and error-prone when inferring state changes caused by complex statements. To mitigate this impact, SQUIRREL tends to generate simple statements in queries.",
+      "context_after": "Without accurate DBMS state information, existing fuzzers are limited in generating complex and valid queries to extensively test DBMSs, causing many deep bugs to still exist. Therefore, proposing a practical solution to address these limitations is necessary and important for DBMS fuzzing. 3 Stateful DBMS Fuzzing DBMSs process each statement of queries in sequence. After each statement is executed, the content of manipulated databases and the status of DBMSs are dynamically changed. In the interval between two statement execution, DBMSs record their latest state information, which accurately",
+      "section": "2 Background and Motivation",
+      "page": 4,
+      "char_offset": 14796,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[39, 41, 43]",
+      "technique": null,
+      "sentence": "Typically, AST-based tools [39, 41, 43] generate random SQL statements according to their random seed (e.",
+      "context_before": "ler returns the generated query, the code coverage covof the DBMS execution, and the final status of the query processing. Translator. Receiving the parameters from Scheduler, Translator first extracts the fresh part of file, which has not been read yet, into tmp_file .Translator uses an internal SQL statement generator StmtGenerator to generate a statement according to the provided schema andtmp_file. Finally, Translator returns the generated statement stmtand updates the number of bytes that have been read by StmtGenerator. The internal StmtGenerator deploys AST models to generate SQL statements.",
+      "context_after": "g. system clock), which makes the generation inefficient and difficult to guide [30, 32]. In contrast to these tools, StmtGenerator uses tmp_file as its random seed, which means that when StmtGenerator traverses its AST tree to generate SQL statements, it decides which paths should be chosen according to the value read from tmp_file. Specifically, it makes a decision by calculating the result of vmod n, where vis the value read from the input file and nis the number of available choices. In this way, StmtGenerator deterministically generates SQL statements according to the provided input file fileand",
+      "section": "3.1 Dynamic Query Interaction",
+      "page": 5,
+      "char_offset": 20466,
+      "cited_reference": {
+        "number": 41,
+        "text": "SQLancer. https://github.com/sqlancer/sqla ncer.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Similar to SQLancer [37], Scheduler also checks whether the target DBMS reports any error.",
+      "context_before": "able candidates (i.e. table t1and view v1),StmtGenerator reads the file and gets value 9. According to the calculated result of 9 mod 2, it uses the second candidate (i.e. view v1). The subsequent generation process follows a similar procedure. Status checking .Scheduler checks the execution status after each statement is fed to the target DBMS. Specifically, Scheduler checks whether any crash of the DBMS or its manipulated databases is triggered. If so, it reports the crash with the query, including the crash-triggering statement and the earlier statements generated in the previous interaction.",
+      "context_after": "If so, it performs a further check and reports an abnormal error if the error is not a syntactic or semantic error. For example, \"Subquery result missing\" in MonetDB is an abnormal error that indicates the DBMS has lost the data of calculated results. These checks enable our dynamic query interaction to report suspicious bugs that make the target DBMS alert but do not cause it to crash directly. Note that if any crash or any error is triggered, Scheduler will terminate the interaction loop because the target DBMS has entered a problematic state. Example. Figure 4 illustrates how our dynamic q",
+      "section": "3.1 Dynamic Query Interaction",
+      "page": 6,
+      "char_offset": 23015,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker",
+      "surface": "[37, 39]",
+      "technique": "pqs",
+      "sentence": "Because many DBMSs use their own SQL dialects and the common core of their SQL features is small [37, 39], it is difficult to use one grammar template to test all DBMSs effectively.",
+      "context_before": "up the testing process. Specifically, DynSQL needs interfaces to start the DBMS, connect to the DBMS, stop the DBMS, send SQL statements, get results of statement processing, and query database schema. DBMSs often have well-defined interfaces for these operations, so it is convenient for users to adopt DynSQL in different DBMSs. In our experience, it took one of our authors less than one hour to make DynSQL support a DBMS. SQL statement generation. We implement our AST-based statement generator referring to SQLsmith [43], and additionally support some SQL features, such as GROUP BY ,UNION, etc.",
+      "context_after": "To address this problem, we fix the general parts of the supported SQL features according to SQL standard [40] and make other parts optional. When testing a specific DBMS, we enable the optional SQL features supported by this DBMS according to its official documents. Bug detection .DynSQL uses ASan [3] as its default checker to detect critical memory bugs. In addition, DynSQL analyzes the abnormal errors collected in dynamic query interaction (Section 3.1) to detect bugs that lead DBMSs to report strange error messages. Query minimization. To reproduce and locate DBMS bugs more conveniently, we",
+      "section": "4 Framework and Implementation",
+      "page": 7,
+      "char_offset": 30327,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), pages 667–682, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "citation_marker",
+      "surface": "[18, 24, 41, 43, 48]",
+      "technique": null,
+      "sentence": "We choose these DBMSs because they are widely used according to DB-Engines Ranking [12] and extensively tested [18, 24, 41, 43, 48].",
+      "context_before": "e following questions: Q1CanDynSQL find bugs in real-world DBMSs by generating complex and valid queries? (Section 5.2) Q2How about the security impact of the bugs found by DynSQL? (Section 5.3) Q3How do dynamic query interaction and error feedback contribute to DynSQL in DBMS fuzzing? (Section 5.4) Q4Can DynSQL outperform other state-of-the-art DBMS fuzzers? (Section 5.5) 5.1 Experimental Setup We evaluate DynSQL on 6 open-source and widely-used DBMSs of the latest versions as of our evaluation, including SQLite [42], MySQL [29], MariaDB [27], PostgreSQL [31], MonetDB [28], and ClickHouse [9].",
+      "context_after": "The basic information of these DBMSs is listed in Table 2 (The lines of the source code are counted by CLOC [10]). We run the evaluation on a regular PC with eight Intel processors and 16GB physical memory, and the OS used is Ubuntu 18.04. 5.2 Runtime Testing Following the evaluation setup of SQUIRREL [48] and the recommendations of Klees et al. [21], we use DynSQL to fuzz each target DBMS five times and calculate the average to get sound results. We use 24 hours as the fuzzing timeout because we observe that the branch coverage and the found bugs of 6 target DBMSs converge and hardly change a",
+      "section": "5.1 Experimental Setup",
+      "page": 8,
+      "char_offset": 32463,
+      "cited_reference": {
+        "number": 41,
+        "text": "SQLancer. https://github.com/sqlancer/sqla ncer.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[48]",
+      "technique": null,
+      "sentence": "2 Runtime Testing Following the evaluation setup of SQUIRREL [48] and the recommendations of Klees et al.",
+      "context_before": ".1 Experimental Setup We evaluate DynSQL on 6 open-source and widely-used DBMSs of the latest versions as of our evaluation, including SQLite [42], MySQL [29], MariaDB [27], PostgreSQL [31], MonetDB [28], and ClickHouse [9]. We choose these DBMSs because they are widely used according to DB-Engines Ranking [12] and extensively tested [18, 24, 41, 43, 48]. The basic information of these DBMSs is listed in Table 2 (The lines of the source code are counted by CLOC [10]). We run the evaluation on a regular PC with eight Intel processors and 16GB physical memory, and the OS used is Ubuntu 18.04. 5.",
+      "context_after": "[21], we use DynSQL to fuzz each target DBMS five times and calculate the average to get sound results. We use 24 hours as the fuzzing timeout because we observe that the branch coverage and the found bugs of 6 target DBMSs converge and hardly change after 24 hours, which is consistent with SQUIRREL. Table 3 shows the results of runtime testing. The columns \"Found\", \"Confirmed\" and \"Fixed\" show the number of bugs that are found by DynSQL, confirmed and fixed by developers, respectively. The columns \"Statement\" and \"Query\" show the numbers of SQL statements and queries, respectively, which are val",
+      "section": "5.2 Runtime Testing",
+      "page": 8,
+      "char_offset": 32839,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[48]",
+      "technique": null,
+      "sentence": "5 Comparison to Existing DBMS Fuzzers We compare DynSQL to two state-of-the-art DBMS fuzzers, SQLsmith [43] and SQUIRREL [48].",
+      "context_before": "g error feedback. Indeed, error feedback can help the fuzzer generate more valid queries to detect more bugs. However, error feedback cannot improve query complexity, so DynSQL !DQIandDynSQL!DQI !EF find only the bugs triggering within two statements and miss the bugs triggering by over two statements (e.g. the bug in Figure 1). In comparison, dynamic query interaction leverages DBMS state information to improve both query complexity and query validity, and thus DynSQL !EFfinds 20 bugs missed by DynSQL!DQI !EF. By further enabling error feedback in DynSQL !EF, DynSQL additionally finds 7 bugs. 5.",
+      "context_after": "SQLancer [35 –37] is also a well-known DBMS testing tool, but it mainly focuses on test oracles, which require test cases with specific patterns to 4958 32nd USENIX Security Symposium USENIX Association Table 7: Results of comparison DBMSSQLsmith SQUIRREL DynSQL Statement Query Bug Statement Query Bug Statement Query Bug SQLite 265K/267K 265K/267K 1 31M/45M 2.4M/9.8M 1 279K/286K 24K/30K 4 MySQL 100K/102K 100K/102K 3 506K/854K 17K/171K 3 91K/96K 9.4K/13K 12 MariaDB 148K/152K 148K/152K 3 245K/392K 425/78K 2 170K/175K 17K/21K 13 PostgreSQL 192K/197K 192K/197K 0 8M/10M 35K/560K 0 154K/160K 14K/18K",
+      "section": "5.5 Comparison to Existing DBMS Fuzzers",
+      "page": 11,
+      "char_offset": 51001,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [35 –37] is also a well-known DBMS testing tool, but it mainly focuses on test oracles, which require test cases with specific patterns to 4958 32nd USENIX Security Symposium USENIX Association Table 7: Results of comparison DBMSSQLsmith SQUIRREL DynSQL Statement Query Bug Statement Query Bug Statement Query Bug SQLite 265K/267K 265K/267K 1 31M/45M 2.",
+      "context_before": "eedback cannot improve query complexity, so DynSQL !DQIandDynSQL!DQI !EF find only the bugs triggering within two statements and miss the bugs triggering by over two statements (e.g. the bug in Figure 1). In comparison, dynamic query interaction leverages DBMS state information to improve both query complexity and query validity, and thus DynSQL !EFfinds 20 bugs missed by DynSQL!DQI !EF. By further enabling error feedback in DynSQL !EF, DynSQL additionally finds 7 bugs. 5.5 Comparison to Existing DBMS Fuzzers We compare DynSQL to two state-of-the-art DBMS fuzzers, SQLsmith [43] and SQUIRREL [48].",
+      "context_after": "4M/9.8M 1 279K/286K 24K/30K 4 MySQL 100K/102K 100K/102K 3 506K/854K 17K/171K 3 91K/96K 9.4K/13K 12 MariaDB 148K/152K 148K/152K 3 245K/392K 425/78K 2 170K/175K 17K/21K 13 PostgreSQL 192K/197K 192K/197K 0 8M/10M 35K/560K 0 154K/160K 14K/18K 0 Total 705K/718K 705K/718K 7 40M/56M 2.5M/11M 6 695K/716K 64K/82K 29 015K30K45K60K 0h 4h 8h 12h 16h 20h 24hCovered branches TimeSQLite 0150K300K450K600K 0h 4h 8h 12h 16h 20h 24hCovered branches TimeMySQL 090K180K270K360K 0h 4h 8h 12h 16h 20h 24hCovered branches TimeMariaDB 040K80K120K160K 0h 4h 8h 12h 16h 20h 24hCovered branches TimePostgreSQL 𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒",
+      "section": "5.5 Comparison to Existing DBMS Fuzzers",
+      "page": 11,
+      "char_offset": 51127,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Considering DynSQL and SQLancer are designed for different research problems, we do not conduct the comparison experiment to SQLancer.",
+      "context_before": "/160K 14K/18K 0 Total 705K/718K 705K/718K 7 40M/56M 2.5M/11M 6 695K/716K 64K/82K 29 015K30K45K60K 0h 4h 8h 12h 16h 20h 24hCovered branches TimeSQLite 0150K300K450K600K 0h 4h 8h 12h 16h 20h 24hCovered branches TimeMySQL 090K180K270K360K 0h 4h 8h 12h 16h 20h 24hCovered branches TimeMariaDB 040K80K120K160K 0h 4h 8h 12h 16h 20h 24hCovered branches TimePostgreSQL 𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒𝐒 𝐒𝐒𝐒𝐒𝐒𝐒𝐬𝐬𝐬𝐬𝐬𝐬𝐬𝐬𝐬𝐬 𝐃𝐃𝐃𝐃𝐃𝐃𝐒𝐒𝐒𝐒𝐒𝐒 Figure 13: Code coverage of DynSQL and the other DBMS fuzzers. find logic bugs in DBMSs, while DynSQL aims at generating complex and valid queries to detect common bugs, especially memory bugs.",
+      "context_after": "We exploit DynSQL, SQUIRREL, and SQLsmith to test SQLite, MySQL, MariaDB, and PostgreSQL, because SQUIRREL supports only these DBMSs and applying it to other DBMSs requires significant modification of the SQUIRREL implementation [45]. Besides, SQLsmith supports only SQLite, PostgreSQL, and MonetDB originally. To perform better comparisons, we extend SQLsmith to support MySQL and MariaDB via little modification of its code. For each test case, we randomly generate a database with tables for SQLsmith because it requires an available database to begin its test. We use each fuzzer to test each DBMS",
+      "section": "5.5 Comparison to Existing DBMS Fuzzers",
+      "page": 12,
+      "char_offset": 52314,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[35 –37]",
+      "technique": "norec",
+      "sentence": "Several existing DBMS testing works [35 –37] use special oracles to detect such bugs.",
+      "context_before": "s. Other kinds of bugs. DynSQL mainly detects memory bugs with existing sanitizers like ASan, and it can additionally find semantic bugs leading to strange error messages. However, many semantic bugs silently affect the execution of DBMSs, and cause no obvious problems. For example, logic bugs do not cause any memory problem or strange error message, but make DBMSs return incorrect results. Performance bugs do not crash DBMSs directly, but often slow down its execution. Detecting these bugs is challenging because there is no general test oracle for checking whether the bugs have been triggered.",
+      "context_after": "But these oracles require fixed patterns of test cases, which limits their scalability. In the future, we will improve the detection of semantic bugs in DynSQL by referring to these approaches. AST rules construction. DynSQL uses an AST-based generator to produce SQL statements in each query. At present, we construct the AST rules based on our domain knowledge. Specifically, we construct general AST rules according to SQL-92 standard [40] and additionally write specific rules 4960 32nd USENIX Security Symposium USENIX Association for each supported DBMS according to its official documents. Users",
+      "section": "6 Limitations and Future Work",
+      "page": 13,
+      "char_offset": 59814,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESE/FSE), pages 1140–115",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[18,23,25,34 –37]",
+      "technique": "norec",
+      "sentence": "They either discover specific kinds of bugs [18,23,25,34 –37], or detect common bugs using general techniques [19, 20, 39].",
+      "context_before": "s 4960 32nd USENIX Security Symposium USENIX Association for each supported DBMS according to its official documents. Users can enable just our general AST rules to test a new DBMS. However, to thoroughly test the code about DBMSspecific usages, users may need to write additional AST rules to enable their unique SQL features. To reduce such manual effort, inspired by existing work [11, 16], we plan to adopt machine learning techniques to automatically extract AST rules from valid queries. 7 Related Work 7.1 DBMS Testing Some approaches are proposed to check the reliability and security of DBMSs.",
+      "context_after": "DBMS testing for specific bugs. SQLancer [41] is designed to detect logic bugs in DBMSs, and it integrates several novel approaches [35 –37]. PQS [37] can generate queries that require the target DBMS to return a result set where a specific row should be included. If the DBMS fails to fetch the row, it indicates a logic bug has been triggered. NoREC [35] is a metamorphic testing approach. It translates a query that can be optimized by the target DBMS to a query that cannot be optimized effectively. NoREC identifies a bug when these two queries make the DBMS return different results. To detect pe",
+      "section": "7.1 DBMS Testing",
+      "page": 14,
+      "char_offset": 60977,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESE/FSE), pages 1140–115",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer [41] is designed to detect logic bugs in DBMSs, and it integrates several novel approaches [35 –37].",
+      "context_before": "rules to test a new DBMS. However, to thoroughly test the code about DBMSspecific usages, users may need to write additional AST rules to enable their unique SQL features. To reduce such manual effort, inspired by existing work [11, 16], we plan to adopt machine learning techniques to automatically extract AST rules from valid queries. 7 Related Work 7.1 DBMS Testing Some approaches are proposed to check the reliability and security of DBMSs. They either discover specific kinds of bugs [18,23,25,34 –37], or detect common bugs using general techniques [19, 20, 39]. DBMS testing for specific bugs.",
+      "context_after": "PQS [37] can generate queries that require the target DBMS to return a result set where a specific row should be included. If the DBMS fails to fetch the row, it indicates a logic bug has been triggered. NoREC [35] is a metamorphic testing approach. It translates a query that can be optimized by the target DBMS to a query that cannot be optimized effectively. NoREC identifies a bug when these two queries make the DBMS return different results. To detect performance bugs, AMOEBA [25] transfers the given queries to semantically equivalent queries and then checks whether these queries result in pe",
+      "section": "7.1 DBMS Testing",
+      "page": 14,
+      "char_offset": 61131,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "PQS [37] can generate queries that require the target DBMS to return a result set where a specific row should be included.",
+      "context_before": "ite additional AST rules to enable their unique SQL features. To reduce such manual effort, inspired by existing work [11, 16], we plan to adopt machine learning techniques to automatically extract AST rules from valid queries. 7 Related Work 7.1 DBMS Testing Some approaches are proposed to check the reliability and security of DBMSs. They either discover specific kinds of bugs [18,23,25,34 –37], or detect common bugs using general techniques [19, 20, 39]. DBMS testing for specific bugs. SQLancer [41] is designed to detect logic bugs in DBMSs, and it integrates several novel approaches [35 –37].",
+      "context_after": "If the DBMS fails to fetch the row, it indicates a logic bug has been triggered. NoREC [35] is a metamorphic testing approach. It translates a query that can be optimized by the target DBMS to a query that cannot be optimized effectively. NoREC identifies a bug when these two queries make the DBMS return different results. To detect performance bugs, AMOEBA [25] transfers the given queries to semantically equivalent queries and then checks whether these queries result in performance differences. To trigger specific kinds of DBMS bugs, these approaches generate test cases with specific patterns,",
+      "section": "7.1 DBMS Testing",
+      "page": 14,
+      "char_offset": 61241,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC [35] is a metamorphic testing approach.",
+      "context_before": "ules from valid queries. 7 Related Work 7.1 DBMS Testing Some approaches are proposed to check the reliability and security of DBMSs. They either discover specific kinds of bugs [18,23,25,34 –37], or detect common bugs using general techniques [19, 20, 39]. DBMS testing for specific bugs. SQLancer [41] is designed to detect logic bugs in DBMSs, and it integrates several novel approaches [35 –37]. PQS [37] can generate queries that require the target DBMS to return a result set where a specific row should be included. If the DBMS fails to fetch the row, it indicates a logic bug has been triggered.",
+      "context_after": "It translates a query that can be optimized by the target DBMS to a query that cannot be optimized effectively. NoREC identifies a bug when these two queries make the DBMS return different results. To detect performance bugs, AMOEBA [25] transfers the given queries to semantically equivalent queries and then checks whether these queries result in performance differences. To trigger specific kinds of DBMS bugs, these approaches generate test cases with specific patterns, limiting their extensibility to detect other kinds of bugs. In contrast to these approaches, DynSQL is designed to detect commo",
+      "section": "7.1 DBMS Testing",
+      "page": 14,
+      "char_offset": 61444,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "NoREC identifies a bug when these two queries make the DBMS return different results.",
+      "context_before": "cific kinds of bugs [18,23,25,34 –37], or detect common bugs using general techniques [19, 20, 39]. DBMS testing for specific bugs. SQLancer [41] is designed to detect logic bugs in DBMSs, and it integrates several novel approaches [35 –37]. PQS [37] can generate queries that require the target DBMS to return a result set where a specific row should be included. If the DBMS fails to fetch the row, it indicates a logic bug has been triggered. NoREC [35] is a metamorphic testing approach. It translates a query that can be optimized by the target DBMS to a query that cannot be optimized effectively.",
+      "context_after": "To detect performance bugs, AMOEBA [25] transfers the given queries to semantically equivalent queries and then checks whether these queries result in performance differences. To trigger specific kinds of DBMS bugs, these approaches generate test cases with specific patterns, limiting their extensibility to detect other kinds of bugs. In contrast to these approaches, DynSQL is designed to detect common bugs in DBMSs and does not limit the patterns of test cases. DBMS testing for common bugs. RAGS [39] uses differential testing to detect bugs in DBMSs. It randomly generates SQL statements and th",
+      "section": "7.1 DBMS Testing",
+      "page": 14,
+      "char_offset": 61602,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "citation_marker",
+      "surface": "[37, 39]",
+      "technique": "pqs",
+      "sentence": "However, this approach is limited because the common part of supported SQL features in different DBMSs is small [37, 39].",
+      "context_before": "performance differences. To trigger specific kinds of DBMS bugs, these approaches generate test cases with specific patterns, limiting their extensibility to detect other kinds of bugs. In contrast to these approaches, DynSQL is designed to detect common bugs in DBMSs and does not limit the patterns of test cases. DBMS testing for common bugs. RAGS [39] uses differential testing to detect bugs in DBMSs. It randomly generates SQL statements and then feeds them to several DBMSs whose databases are the same. RAGS reports bugs if the execution status or returned results of these DBMSs are different.",
+      "context_after": "Some approaches [19, 20] convert SQL statement generation into SAT problems and generate valid statements to test DBMSs by solving syntactic and semantic constraints of the SQL language. However, these approaches cannot infer state changes caused by generated statements and thus can generate only one statement in each query. In contrast, DynSQL can generate queries with multiple statements by capturing the latest DBMS states before each statement generation. In addition, these approaches perform random test-case generation, or exhaustively enumerate all possible test cases. Comparatively, Dyn",
+      "section": "7.1 DBMS Testing",
+      "page": 14,
+      "char_offset": 62439,
+      "cited_reference": {
+        "number": 37,
+        "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), pages 667–682, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[48]",
+      "technique": null,
+      "sentence": "However, existing work [48] proves that general fuzzers cannot effectively test DBMSs.",
+      "context_before": "test-case generation and integrates various mutation strategies and engineering techniques to improve its efficiency. To cover more branches, Angora [6] first uses taint analysis to track specific input bytes that influence the control flows, and then it leverages gradient descent to quickly search for the suitable values of these bytes that satisfy the path constraints. To find more bugs, QSYM [47] adopts symbolic execution in fuzzing, and further uses dynamic binary translation to integrate symbolic emulation into the native execution, which dramatically reduces the overhead of symbolic execution.",
+      "context_after": "These fuzzers fail to involve any SQL knowledge and AST rules, and thus generate many invalid queries violating syntactic or semantic checks. DBMS fuzzing. To test DBMSs more effectively, several approaches [18, 43, 45, 48] combine fuzzing with grammarbased generation techniques. SQLsmith [43], a state-of-the-art DBMS fuzzer, uses its embedded AST rules to randomly generate SQL queries. However, each query generated by SQLsmith contains only one statement, because it is not aware of the state changes caused by generated statements. To generate queries with multiple statements, SQUIRREL [48] u",
+      "section": "7.1 DBMS Testing",
+      "page": 14,
+      "char_offset": 64102,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M24",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[18, 43, 45, 48]",
+      "technique": null,
+      "sentence": "To test DBMSs more effectively, several approaches [18, 43, 45, 48] combine fuzzing with grammarbased generation techniques.",
+      "context_before": "hen it leverages gradient descent to quickly search for the suitable values of these bytes that satisfy the path constraints. To find more bugs, QSYM [47] adopts symbolic execution in fuzzing, and further uses dynamic binary translation to integrate symbolic emulation into the native execution, which dramatically reduces the overhead of symbolic execution. However, existing work [48] proves that general fuzzers cannot effectively test DBMSs. These fuzzers fail to involve any SQL knowledge and AST rules, and thus generate many invalid queries violating syntactic or semantic checks. DBMS fuzzing.",
+      "context_after": "SQLsmith [43], a state-of-the-art DBMS fuzzer, uses its embedded AST rules to randomly generate SQL queries. However, each query generated by SQLsmith contains only one statement, because it is not aware of the state changes caused by generated statements. To generate queries with multiple statements, SQUIRREL [48] uses a new intermediate representation (IR) to model SQL queries and statically infers DBMS state changes caused by the generated statements. However, its static inference is inaccurate without runtime information, as a result of which it still generates over 50% invalid queries in",
+      "section": "7.1 DBMS Testing",
+      "page": 14,
+      "char_offset": 64345,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M25",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[48]",
+      "technique": null,
+      "sentence": "To generate queries with multiple statements, SQUIRREL [48] uses a new intermediate representation (IR) to model SQL queries and statically infers DBMS state changes caused by the generated statements.",
+      "context_before": "48] proves that general fuzzers cannot effectively test DBMSs. These fuzzers fail to involve any SQL knowledge and AST rules, and thus generate many invalid queries violating syntactic or semantic checks. DBMS fuzzing. To test DBMSs more effectively, several approaches [18, 43, 45, 48] combine fuzzing with grammarbased generation techniques. SQLsmith [43], a state-of-the-art DBMS fuzzer, uses its embedded AST rules to randomly generate SQL queries. However, each query generated by SQLsmith contains only one statement, because it is not aware of the state changes caused by generated statements.",
+      "context_after": "However, its static inference is inaccurate without runtime information, as a result of which it still generates over 50% invalid queries in the evaluation. These fuzzers are limited in generating complex and valid queries, as they fail to consider state changes or infer accurate state information. In contrast, DynSQL performs dynamic query interaction to capture accurate state information, including the latest database schema and the status of statement processing. In this way, DynSQL can effectively generate complex and valid queries to detect deep bugs in DBMSs. 8 Conclusion In this paper,",
+      "section": "7.1 DBMS Testing",
+      "page": 14,
+      "char_offset": 64727,
+      "cited_reference": {
+        "number": 48,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In Proceedings of the 2020 International Conference on Computer and Communications Security (CCS), pages 955–970, 2020. USENIX Asso",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M15"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:16:44Z",
+    "is_model_written": true,
+    "summary": "DynSQL fuzzes DBMSs with multi-statement SQL queries, targeting the trade-off that earlier generators face between query complexity and validity: single-statement tools sidestep inter-statement dependencies, while SQUIRREL's static intermediate representation still yields over 50% invalid queries. DynSQL instead builds a query one statement at a time, querying the live DBMS state after each, so dependencies are resolved dynamically. It found 40 bugs in six DBMSs, 38 confirmed and 21 fixed.",
+    "narrative": "SQLancer appears as related work rather than as a baseline, and the paper is explicit about why: it states that DynSQL and SQLancer are designed for different research problems and that no comparison experiment against SQLancer was run. Its baselines are SQLsmith and SQUIRREL. SQLancer is described as a well-known logic-bug detector integrating PQS, NoREC and TLP, each summarised in the related-work section, and one design point is credited to it -- DynSQL's scheduler checks whether the DBMS reports an error in the way SQLancer does. The paper also cites SQLancer's own observation that the common core of SQL features across dialects is small, using it to justify not relying on a single grammar template.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "background",
+      "M6": "background",
+      "M7": "background",
+      "M8": "background",
+      "M9": "definition",
+      "M10": "motivation",
+      "M11": "background",
+      "M12": "background",
+      "M13": "background",
+      "M14": "definition",
+      "M15": "incidental",
+      "M16": "background",
+      "M17": "background",
+      "M18": "definition",
+      "M19": "definition",
+      "M20": "definition",
+      "M21": "definition",
+      "M22": "motivation",
+      "M23": "background",
+      "M24": "background",
+      "M25": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited and compared against, not reused; nothing describes building on its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No SQLancer technique is extended; the work takes a different approach."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "M15 states outright that the authors do not conduct a comparison experiment against SQLancer, on the grounds that the two address different research problems. The evaluated baselines are SQLsmith and SQUIRREL."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is called a well-known DBMS testing tool in M14, which is recognition but not a claim that it is the state of the art; the paper reserves that description for the tools it measures against."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json b/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json
new file mode 100644
index 0000000..bbc2c90
--- /dev/null
+++ b/_data/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.json
@@ -0,0 +1,2384 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-10T15:26:59Z",
+  "paper": {
+    "id": "paper:s2:77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+    "title": "Detecting Logical Bugs of DBMS with Coverage-based Guidance",
+    "authors": [
+      "Yu Liang",
+      "Song Liu",
+      "Hong Hu"
+    ],
+    "year": 2022,
+    "venue": "USENIX Security Symposium",
+    "doi": null,
+    "arxiv_id": null,
+    "s2_paper_id": "77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+    "url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://www.semanticscholar.org/paper/77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42",
+      "retrieved_at": "2026-09-10T15:26:59Z",
+      "chars": 99500,
+      "content_sha256": "sha256:240fd50f196464c43d064ccad42fd1b0a11a48be703fc229ffc095b97a36a5df"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 19,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1867
+      },
+      {
+        "number": "2",
+        "title": "Background & Challenges",
+        "start": 10241
+      },
+      {
+        "number": "2.1",
+        "title": "An Example Logical Bug",
+        "start": 10267
+      },
+      {
+        "number": "2.3",
+        "title": "Coverage-guided Testing",
+        "start": 14871
+      },
+      {
+        "number": "2.4",
+        "title": "Challenges of Fuzzing Logical Bugs",
+        "start": 17339
+      },
+      {
+        "number": "3.1",
+        "title": "Validity-oriented Query Generation",
+        "start": 20960
+      },
+      {
+        "number": "3.1.1",
+        "title": "Cooperative Mutation",
+        "start": 21644
+      },
+      {
+        "number": "3.1.2",
+        "title": "Dedicated Parsing",
+        "start": 24089
+      },
+      {
+        "number": "3.1.3",
+        "title": "Context-based IR Instantiation",
+        "start": 25835
+      },
+      {
+        "number": "3.1.4",
+        "title": "Non-determinism Mitigation",
+        "start": 28840
+      },
+      {
+        "number": "5.1",
+        "title": "DBMS Logical Bugs",
+        "start": 43650
+      },
+      {
+        "number": "5.4",
+        "title": "Contribution of Validity",
+        "start": 60268
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Shadi Abdul Khalek and Sarfraz Khurshid. Automated SQL Query Generation for Systematic Testing of Database Engines. In Proceedings of the IEEE/ACM International Conference on Automated Software Engineering (ASE), 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Alloy. Alloy-Documentation of Alloy SAT solver. https:// alloytools.org/documentation.html. (visited in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Cornelius Aschermann, Tommaso Frassetto, Thorsten Holz, Patrick Jauernig, Ahmad-Reza Sadeghi, and Daniel Teuchert. NAUTILUS: Fishing for Deep Bugs with Grammars. In Proceedings of the 26th Annual Network and Distributed System Security Symposium (NDSS), San Diego, CA, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Cornelius Aschermann, Sergej Schumilo, Tim Blazytko, Robert Gawlik, and Thorsten Holz. REDQUEEN: Fuzzing with Input-to-State Correspondence. In Proceedings of the 26th Annual Network and Distributed System Security Symposium (NDSS), San Diego, CA, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Hardik Bati, Leo Giakoumakis, Steve Herbert, and Aleksandras Surna. A Genetic Approach for Random Testing of Database Systems. In Proceedings of the 33rd International Conference on Very Large Data Bases (VLDB), Vienna, Austria, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Tim Blazytko, Cornelius Aschermann, Moritz Schlögel, Ali Abbasi, Sergej Schumilo, Simon Wörner, and Thorsten Holz. GRIMOIRE: Synthesizing Structure while Fuzzing. In Proceedings of the 28th USENIX Security Symposium (USENIX Security), Santa Clara, CA, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Bikash Chandra, Bhupesh Chawda, Biplab Kar, K. V. Maheshwara Reddy, Shetal Shah, and S. Sudarshan. Data Generation for Testing and Grading SQL Queries. The VLDB Journal, 24(6), Aug 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Peng Chen and Hao Chen. Angora: Efficient Fuzzing By Principled Search. In Proceedings of the 39th IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Yaohui Chen, Peng Li, Jun Xu, Shengjian Guo, Rundong Zhou, Yulong Zhang, Long Lu, et al. Savior: Towards bug-driven hybrid testing. InProceedings of the 41st IEEE Symposium on Security and Privacy (Oakland), Virtual, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Yongheng Chen, Rui Zhong, Hong Hu, Hangfan Zhang, Yupeng Yang, Dinghao Wu, and Wenke Lee. One Engine to Fuzz ’em All: Generic Language Processor Testing with Semantic Validation. In Proceedings of the 42nd IEEE Symposium on Security and Privacy (Oakland), Virtual, May 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Jaeseung Choi, Kangsu Kim, Daejin Lee, and Sang Kil Cha. NTFUZZ: Enabling Type-Aware Kernel Fuzzing on Windows with Static Binary Analysis. In Proceedings of the 42nd IEEE Symposium on Security and Privacy (Oakland), Virtual, May 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Sung Ta Dinh, Haehyun Cho, Kyle Martin, Adam Oest, Kyle Zeng, Alexandros Kapravelos, Gail-Joon Ahn, Tiffany Bao, Ruoyu Wang, Adam Doupé, et al. Favocado: Fuzzing the Binding Code of JavaScript Engines Using Semantically Correct Test Cases. In Proceedings of the 28th Annual Network and Distributed System Security Symposium (NDSS), Virtual, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Andrea Fioraldi, Daniele Cono D’Elia, and Leonardo Querzoni. Fuzzing Binaries for Memory Safety Errors with QASan. In Proceedings of 2020 IEEE Secure Development (SecDev), Virtual, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Free Software Foundation. Gnu bison, 2014. https://www.gnu. org/software/bison/ .",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Shuitao Gan, Chao Zhang, Xiaojun Qin, Xuwen Tu, Kang Li, Zhongyu Pei, and Zuoning Chen. CollAFL: Path Sensitive Fuzzing. In Proceedings of the 39th IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Bogdan Ghit, Nicolas Poggi, Josh Rosen, Reynold Xin, and Peter Boncz. SparkFuzz: Searching Correctness Regressions in Modern Query Engines. In Proceedings of the Workshop on Testing Database Systems (DBTest), Portland, Oregon, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Google. ClusterFuzz. https://google.github.io/clusterfuzz. (visited in June, 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Google. Honggfuzz. https://google.github.io/honggfuzz/. (visited in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "HyungSeok Han, DongHyeon Oh, and Sang Kil Cha. CodeAlchemist: Semantics-Aware Code Generation to Find Vulnerabilities in JavaScript Engines. In Proceedings of the 26th Network and Distributed System Security Symposium (NDSS), San Diego, CA, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Bo Jiang, Ye Liu, and W. K. Chan. ContractFuzzer: Fuzzing Smart Contracts for Vulnerability Detection. In Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering (ASE), 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Jinho Jung, Hong Hu, Joy Arulraj, Taesoo Kim, and Woonhak Kang. APOLLO: Automatic Detection and Diagnosis of Performance Regressions in Database Systems. In Proceedings of the 46th International Conference on Very Large Data Bases (VLDB), Tokyo, Japan, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Kyungtae Kim, Dae R Jeong, Chung Hwan Kim, Yeongjin Jang, Insik Shin, and Byoungyoung Lee. HFL: Hybrid Fuzzing on the Linux Kernel. In Proceedings of the 27th Annual Network and Distributed System Security Symposium (NDSS), San Diego, CA, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Doug Laney. 3-D Data Management: Controlling Data V olume, Velocity and Variety. Technical report, February 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Caroline Lemieux and Koushik Sen. Fairfuzz: A Targeted Mutation Strategy for Increasing Greybox Fuzz Testing Coverage. In Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering (ASE), 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Guoliang Li, Xuanhe Zhou, Shifu Li, and Bo Gao. QTune: A QueryAware Database Tuning System with Deep Reinforcement Learning. Proceedings of the VLDB Endowment, 12(12):2118–2130, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Xinyu Liu, Qi Zhou, Joy Arulraj, and Alessandro Orso. Automated Performance Bug Detection in Database Systems. arXiv preprint arXiv:2105.10016, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "LLVM. LibFuzzer-A Library For Coverage-guided Fuzz Testing. http://llvm.org/docs/LibFuzzer.html. (visited in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "LLVM. Undefined Behavior Sanitizer (UBSan). https://clang. llvm.org/docs/UndefinedBehaviorSanitizer.html. (visted in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Eric Lo, Carsten Binnig, Donald Kossmann, M. Tamer Özsu, and WingKai Hon. A Framework for Testing DBMS Features. The VLDB Journal, 19(2):203–230, April 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Manuel Rigger. SQLancer. https://github.com/sqlancer/ sqlancer. (visited in June 2021).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 31,
+      "text": "Michaël Marcozzi, Wim Vanhoof, and Jean-Luc Hainaut. Test Input Generation for Database Programs Using Relational Constraints. In Proceedings of the Fifth International Workshop on Testing Database Systems (DBTest), Scottsdale, Arizona, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Barton P. Miller, Louis Fredriksen, and Bryan So. An Empirical Study of the Reliability of UNIX Utilities. Communications of the ACM, 33(12):32–44, December 1990.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Chaitanya Mishra, Nick Koudas, and Calisto Zuzarte. Generating Targeted Queries for Database Testing. In Proceedings of the 2008 ACMSIGMOD International Conference on Management of Data (SIGMOD), New York, NY, USA, 2008. 4322 31st USENIX Security Symposium USENIX Association",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "MySQL. MySQL Customers. https://www.mysql.com/ customers/. (visited in Jan 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Tai D Nguyen, Long H Pham, Jun Sun, Yun Lin, and Quang Tran Minh. sfuzz: An Efficient Adaptive Fuzzer for Solidity Smart Contracts. InProceedings of the ACM/IEEE 42nd International Conference on Software Engineering (ICSE), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Rohan Padhye, Caroline Lemieux, Koushik Sen, Mike Papadakis, and Yves Le Traon. Semantic fuzzing with zest. In Proceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis (ISSTA), 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Shankara Pailoor, Andrew Aday, and Suman Jana. MoonShine: Optimizing OS Fuzzer Seed Selection with Trace Distillation. In Proceedings of the 27th USENIX Security Symposium (USENIX Security), Baltimore, MD, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Soyeon Park, Wen Xu, Insu Yun, Daehee Jang, and Taesoo Kim. Fuzzing JavaScript Engines with Aspect-preserving Mutation. In Proceedings of the 41st IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "PostgreSQL. Aboug PostgreSQL. https://www.postgresql.org/ about/. (visited in Jan 2022).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Kim-Thomas Rehmann, Changyun Seo, Dongwon Hwang, Binh Truong, Alexander Böhm, and Donghun Lee. Performance Monitoring in SAPHANA’s Continuous Integration Process. ACMSIGMETRICS Performance Evaluation Review, 43:43–52, 02 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Richard Hipp. What is Fossil? https://www2.fossil-scm.org/ home/doc/trunk/www/index.wiki. (visited on Jan 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger. Bugs Found in Database Management Systems. https: //www.manuelrigger.at/dbms-bugs/. (visited in June 2021).",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger and Zhendong Su. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA):1–30, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger and Zhendong Su. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), Virtual, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 46,
+      "text": "Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitry Vyukov. AddressSanitizer: A Fast Address Sanity Checker. In Proceedings of the 2012 USENIX Annual Technical Conference (ATC), Boston, MA, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Konstantin Serebryany and Timur Iskhodzhanov. ThreadSanitizer: Data Race Detection in Practice. In Proceedings of the workshop on binary instrumentation and applications, pages 62–71, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Kostya Serebryany. Sanitize, Fuzz, and Harden Your C++ Code. San Francisco, CA, 2016. USENIX Association.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Donald R Slutz. Massive Stochastic Testing of SQL. In VLDB, volume 98, pages 618–622. Citeseer, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "SQLite. Datatypes In SQLite. https://www.sqlite.org/ datatype3.html. (visited in Jan 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "SQLite. Most Widely Deployed and Used Database Engine. https: //www.sqlite.org/mostdeployed.html. (visited in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "SQLite Team. sqllogictest Documentation. https://www.sqlite. org/sqllogictest/doc/trunk/about.wiki. (visited in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Nick Stephens, John Grosen, Christopher Salls, Andrew Dutcher, Ruoyu Wang, Jacopo Corbetta, Yan Shoshitaishvili, Christopher Kruegel, and Giovanni Vigna. Driller: Augmenting Fuzzing Through Selective Symbolic Execution. In Proceedings of the 23rd Annual Network and Distributed System Security Symposium (NDSS), San Diego, CA, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Michael Stonebraker, Sam Madden, and Pradeep Dubey. Intel \"Big Data\" Science and Technology Center Vision and Execution Plan. ACMSIGMOD Record, 42(1):44–49, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Well-known Users of SQLite. https://www.sqlite.org/famous. html. (visited in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "MySQL Customers. https://www.mysql.com/customers/. (visited in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "PostgreSQL Clients. https://wiki.postgresql.org/wiki/ PostgreSQL_Clients. (visited in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "SQLSmith. https://github.com/anse1/sqlsmith. (visited in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "J. Wang, P. Zhang, L. Zhang, H. Zhu, and X. Ye. A Model-based Fuzzing Approach for DBMS. In Proceedings of the 8th International Conference on Communications and Networking in China (CHINACOM), Aug 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In 2021 IEEE/ACM 43rd International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP), 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Gary Wassermann and Zhendong Su. Sound and Precise Analysis of Web Applications for Injection Vulnerabilities. In Proceedings of the 28th ACMSIGPLAN Conference on Programming Language Design and Implementation (PLDI), New York, NY, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Valentin Wüstholz and Maria Christakis. Harvey: A Greybox Fuzzer for Smart Contracts. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (FSE), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Meng Xu, Sanidhya Kashyap, Hanqing Zhao, and Taesoo Kim. Krace: Data Race Fuzzing for Kernel File Systems. In Proceedings of the 41st IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Peng Xu, Yanhao Wang, Hong Hu, and Purui Su. Cooper: Testing the Binding Code of Scripting Languages with Cooperative Mutation. InProceedings of the 29th Annual Network and Distributed System Security Symposium (NDSS 2022), San Diego, CA, apr 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Wen Xu, Hyungon Moon, Sanidhya Kashyap, Po-Ning Tseng, and Taesoo Kim. Fuzzing File Systems via Two-Dimensional Input Space Exploration. In Proceedings of the 40th IEEE Symposium on Security and Privacy (Oakland), San Francisco, CA, May 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Wen Xu, Soyeon Park, and Taesoo Kim. FREEDOM: Engineering a State-of-the-Art DOM Fuzzer. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, FL, November 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 67,
+      "text": "Khaled Yagoub, Peter Belknap, Benoit Dageville, Karl Dias, Shantanu Joshi, and Hailing Yu. Oracle’s SQL Performance Analyzer. Bulletin of the IEEE Computer Society Technical Committee on Data Engineering, 31(1), 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 68,
+      "text": "Jiaqi Yan, Qiuye Jin, Shrainik Jain, Stratis D. Viglas, and Allison Lee. Snowtrail: Testing with Production Queries on a Cloud Database. In Proceedings of the Workshop on Testing Database Systems (DBTest), New York, NY, USA, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 69,
+      "text": "Insu Yun, Sangho Lee, Meng Xu, Yeongjin Jang, and Taesoo Kim. QSYM: A Practical Concolic Execution Engine Tailored for Hybrid Fuzzing. In Proceedings of the 27th USENIX Security Symposium (USENIX Security), Baltimore, MD, August 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 70,
+      "text": "Michal Zalewski. American Fuzzy Lop (2.52b). http://lcamtuf. coredump.cx/afl. (visited in June 2021).",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 71,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Association 31st USENIX Security Symposium 4323 A More Case Studies on Detected Bugs 01 CREATE TABLE v0 (v",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 30,
+      "text": "Manuel Rigger. SQLancer. https://github.com/sqlancer/ sqlancer. (visited in June 2021).",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger. Bugs Found in Database Management Systems. https: //www.manuelrigger.at/dbms-bugs/. (visited in June 2021).",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 44,
+      "text": "Manuel Rigger and Zhendong Su. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA):1–30, 2020.",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 45,
+      "text": "Manuel Rigger and Zhendong Su. Testing Database Engines via Pivoted Query Synthesis. In Proceedings of the 14th USENIX Symposium on Operating Systems Design and Implementation (OSDI), Virtual, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 71,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Association 31st USENIX Security Symposium 4323 A More Case Studies on Detected Bugs 01 CREATE TABLE v0 (v1 PRIMARY KEY, v2, v3); 02 INSERT INTO v0 VALUES ( '111','222','333'); 03 CREATE TABLE v4 (v5 PRIMARY KEY); 04 INSERT INTO v4 VALUES ( '0'); 05 SELECT * FROM v4 JOIN v0 ON likely(v3=v1) AND v3= '111'; 06 -- output: 0|111|222|333 07 -- expect: (empty) Listing 12: A logical bug of SQLite that leaks data due to incorrect constant propagation. The SELECT statement should not return any rows, but the buggy SQLite returns one row. Data Leakage Bug. SQL statements in Listing 12 trigger logical bug 2 in Table 2, which returns extra rows. The first two statements create table v0and inserts one row ‘111’,‘222’,‘333’, while the following two statements create table v4with one row ‘0’. The SELECT statement joins rows if they satisfy the conditions v3=v1 andv3=‘111’. Function likely() provides a hint about the comparison result, and by design will not change the results. Since v0has no rows satisfying the conditions, SQLite should return nothing. However, due to the incorrect constant propagation, the conditions are incorrectly optimized to v1=‘111’, and SQLite returns row 0|111|222|333. From the security perspective, attackers may exploit this bug to steal information, if they can trigger the DBMS to conduct queries similar to Listing 12. In a worse scenario, if a system relies on such SELECT for authentication, e.g., searching for matched user name and password, attackers may get system access without valid credentials. Our bisector locates the buggy commit introduced in July 2018, which means the vulnerability has been there for several years. SQLRight can detect this bug using NoREC and TLPwithin six hours. Considering the simplicity of these queries, real-world uses of SQLite may have triggered this bug. 01 CREATE TABLE v0 (v1 INTEGER PRIMARY KEY) WITHOUT ROWID; 02 INSERT INTO v0 VALUES (10); 03 ALTER TABLE v0 ADD v2 INT; 04 SELECT * FROM v0 WHERE v1=20 OR (v1=10 AND v2=10); 05 -- output: 10|NULL 06 -- expect: (empty) Listing 13: A logical bug of SQLite that leaks data due to incorrect primary key. We expect the SELECT statement returns no rows, but the buggy SQLite provides one extra row. Data Leakage due to Primary Key. Listing 13 shows queries that trigger logical bug 8 in Table 2, which returns extra rows. The first statement creates table v0with one column v1, where v1has the INTEGER type and is the primary key. The second statement inserts one row 10into this table. The third statement adds another column v2of type INTinto v0. By default, all rows in v0will have a NULL value in column v2. The last SELECT statement searches in table v0for rows where v1=20 or both v1andv2have the value of 10. Since table v0 only has one row 10|NULL, we expect the DBMS returns no rows. However, it returns 10|NULL, even though the content does not satisfy the conditions in WHERE .SQLRight can detect this bug by oracle TLPand oracle Rowid. Regardless of the oracle, to trigger the bug the query should create a WITHOUT ROWID table. In an ordinary table (a table created without WITHOUT ROWID ),SQLite creates a hidden column rowid and uses it as the real primary key. The PRIMARY KEY column is merely implemented as a UNIQUE INDEX that redirects the search back to rowid. However, in a WITHOUT ROWID table the extra column rowid is disabled. SQLite implements the PRIMARY KEY columns as the unique id of each row, and directly uses them to narrow down the search. Without the extra rowid column, SQLite can improve the speed of fetching data and reduce the storage size. In the first statement of Listing 13, v0is aWITHOUT ROWID table. SQLite treated column v1as the real primary key and stored it in a sorted manner. In the third statement, SQLite adds column v2and stores it without sorting. The unsorted v2 should not be used to narrow the search. However, the query planner of SQLite does not take this exception into account and uses v2to traverse the table. Therefore, SQLite returns the extra row to the query. We reported the bug to SQLite developers, and they fix it by not considering the unsorted column when constructing the query plan. 01 CREATE TABLE v0 (v2 INT, v1 INT); 02 INSERT INTO v0 VALUES (10, 10); 03 CREATE INDEX v4 ON v0 (v1) WHERE v2>NULL; 04 SELECT * FROM v0 WHERE v2 IS NOT NULL; 05 -- output: (empty) 06 -- expect: 10|10 Listing 14: A logical bug of SQLite due to futile index. TheSELECT statement should return one row, but SQLite returns nothing as it incorrectly uses the index v4. Missing Data due to Futile Index. Listing 14 shows the query set that triggers logical bug 11 in Table 2, which drops expected rows from the result. The first two statements of this query create table v0and insert one row 10|10. The third statement creates a partial index v4on column v1when v2>NULL. Since the comparison with NULL always returns NULL, no rows are added into index v4. The last statement searches in table v0and aims to find rows where v2is not NULL. Since the table only has one row 10|10, we expect the query returns one row. However, the buggy version of SQLite does not return any rows. The reason is that SQLite confuses >NULL with IS NOT NULL, and thus uses the empty index v4to handle the SELECT statement. SQLRight can detect this bug using oracle NoREC and oracle Index. After our report, the developers have fixed this bug by resolving the confusion. 4324 31st USENIX Security Symposium USENIX Association 0 12 24 36 48 60 7201234Unique Bug (a)SQLite logical bugs 0 12 24 36 48 60 7201234Unique Bug (b)MySQL logcial bugs 0 12 24 36 48 60 722025303540Coverage (k) SQLRight Squirrel+ oracle SQLancer 0 12 24 36 48 60 722025303540Coverage (k) (c)SQLite code coverage 0 12 24 36 48 60 720255075100125150175200Coverage (k) (d)MySQL code coverage 0 12 24 36 48 60 72020406080100120140Coverage (k) (e)PostgreSQL code coverage 0 12 24 36 48 60 72304050607080Query Validity (%) (f)SQLite query validity 0 12 24 36 48 60 72020406080100Query Validity (%) (g)MySQL query validity 0 12 24 36 48 60 72010203040Query Validity (%) (h)PostgreSQL query validity 0 12 24 36 48 60 72106107Valid Queries per Hour (i)SQLite valid stmts / hr 0 12 24 36 48 60 72101103105107Valid Queries per Hour (j)MySQL valid stmts / hr 0 12 24 36 48 60 72101103105107109Valid Queries per Hour (k)PostgreSQL valid stmts / hr Figure 8: Comparison between different tools ( TLP). a-k show the number of unique logical bugs, code coverage, query validity and valid statements over time for each fuzzing instance. We run each instance for 72 hours, repeat for five times and report the average. 0 4 8 12 16 20 240123Unique Bug (a)SQLite logical bugs 0 4 8 12 16 20 2401234Unique Bug (b)MySQL logical bugs SQLRight SQLRight−deter SQLRight−ctx−valid SQLRight−db−par&ctx−valid Squirrel+ oracle 0 4 8 12 16 20 2425.027.530.032.535.037.540.042.5Coverage (k) (c)SQLite code coverage 0 4 8 12 16 20 24100120140160180200220Coverage (k) (d)MySQL code coverage 0 4 8 12 16 20 24020406080100120140Coverage (k) (e)PostgreSQL code coverage 0 4 8 12 16 20 24203040506070Query Validity (%) (f)SQLite query validity 0 4 8 12 16 20 24010203040Query Validity (%) (g)MySQL query validity 0 4 8 12 16 20 24051015202530Query Validity (%) (h)PostgreSQL query validity 0 4 8 12 16 20 24105106Coverage (k) (i)SQLite valid stmts / hr 0 4 8 12 16 20 24100101102103104105valid statements per hour (j)MySQL valid stmts / hr 0 4 8 12 16 20 24100101102103104105valid statements per hour (k)PostgreSQL valid stmts / hr Figure 9: Unit Tests for SQLRight components ( TLP). a-k show the number of unique logical bugs, code coverage, query validity and valid statements over time for each fuzzing instance. We run each instance for 24 hours, repeat for five times and report the average. USENIX Association 31st USENIX Security Symposium 4325 B Generating the Motivating Example 01 CREATE TABLE person ( 02 pid INTEGER PRIMARY KEY, 03 tid INTEGER REFERENCES team, 04 leader BOOLEAN, 05 ); 06 CREATE UNIQUE INDEX leader ON person(tid) WHERE leader; 07 SELECT pid FROM person WHERE leader AND tid=1; Listing 15: Original queries that lead to Listing 1. We collected this seed input from SQLite website about unique partial indexes. To prepare fuzzing, we collect the seed input from the SQLite website about unique partial indexes, shown in Listing 15. Together with other seed inputs, we launch SQLRight to test the SQLite. During the initialization, SQLRight scans all seed inputs. When reaching Listing 15, it extracts the SELECT statement at line 7 and saves it to the select queue; other statements are stored in the normal queue. After that, SQLRight starts to mutate existing queries to get new ones. When lines 1-6 are selected for mutation, SQLRight first invokes the preprocess() function of oracle Index. preprocess() determines that this query set is compatible, and it removes line 6 to avoid false alarms. Then, the mutation engine randomly modifies the remaining components, i.e.,CREATE TABLE, to generate new statements. By some chance, it deletes column tidand column leader from tableperson, and inserts an INSERT statement that adds three values into the table. Meanwhile, SQLRight invokes function append_output() of oracle Index to insert SELECT statements. append_output() picks up existing SELECT s from the selectqueue, and mutates them in a controlled manner: it will keep the SELECT keyword but can modify other components. After generating a set of valid SELECT s,SQLRight appends them to the previously generated query set, i.e., after CREATE TABLE and INSERT. At this moment, we complete the query mutation. Next, SQLRight applies the validation to update the operands in each query, to make the whole query set semantically correct. For example, it updates the table names in FROM clauses of SELECT s so that they use the table created by CREATE TABLE. After that, SQLRight sends the query set to function transform() to produce functionally equivalent forms. transform() picks some CREATE INDEX statements from the normal queue, and mutates them for new operations, like adding the UNIQUE before INDEX. It inserts the generated CREATE UNIQUE INDEX statements into the given query set randomly, and updates the operands accordingly (similar to validation). transform() returns the new query sets back to SQLRight, and SQLRight feeds the original and new query sets to SQLite for execution. After the execution, SQLRight invokes function compare() to decide whether the results are consistent or not. In this example, compare() notices that one result has one row while another has two. Therefore, it notifies SQLRight about the inconsistency, and SQLRight will raise an alarm of the potential logical bug. Regardless of the consistency, SQLRight checks whether the execution of SQLite triggers new code coverage or not. If so, it extracts SELECT s from the combined query set and saves them into the select queue, and stores other statements into the normal queue. Now, SQLRight completes one fuzzing round, and it will move on to mutate other query sets. 4326 31st USENIX Security Symposium USENIX Association",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[10, 58, 70, 71]",
+      "technique": null,
+      "sentence": "Although many efforts have been spent on testing DBMS systems [10, 58, 70, 71], most of them focus on finding crashes and assertion failures that finally make the execution terminate accidentally.",
+      "context_before": "zers. Then, we provide validity-oriented mutations to generate high-quality query statements in order to find more logical bugs. Our prototype, SQLRight, outperforms existing tools that only rely on oracles or code coverage. In total, SQLRight detects 18 logical bugs from two well-tested DBMSs, SQLite and MySQL. All bugs have been confirmed and 14 of them have been fixed. 1 Introduction Database management systems (DBMSs) are used extensively in data-intensive programs, helping billions of devices host trillions of databases [23,51,54 –57]. Any bug in DBMSs will affect a large number of users.",
+      "context_after": "They cannot detect logical bugs that make the DBMS return unexpected results, like leaking extra rows. Since logical bugs usually do not crash the program, we need an oracle to determine whether each execution produces the correct result or not. However, building an oracle is time-consuming and error-prone due to the various language dialects and features [49]. Recent progress on DBMS oracles shed light on logical bug detection [43 –45]. For example, Rigger et al. constructedseveral general oracles by transforming SQL queries into semantically equivalent forms. One DBMS may process these quer",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 2092,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[43 –45]",
+      "technique": "norec",
+      "sentence": "Recent progress on DBMS oracles shed light on logical bug detection [43 –45].",
+      "context_before": "Ss will affect a large number of users. Although many efforts have been spent on testing DBMS systems [10, 58, 70, 71], most of them focus on finding crashes and assertion failures that finally make the execution terminate accidentally. They cannot detect logical bugs that make the DBMS return unexpected results, like leaking extra rows. Since logical bugs usually do not crash the program, we need an oracle to determine whether each execution produces the correct result or not. However, building an oracle is time-consuming and error-prone due to the various language dialects and features [49].",
+      "context_after": "For example, Rigger et al. constructedseveral general oracles by transforming SQL queries into semantically equivalent forms. One DBMS may process these queries with different code paths, but the final results should be the same. Any inconsistency indicates a potential logical bug. SQLancer, the tool that implements these oracles, has successfully found many logical bugs [30, 42]. However, SQLancer relies on a rule-based generator to synthesize original queries, which may limit its capability to explore program states. Specifically, it creates expressions for WHERE and JOIN clauses based on th",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 2653,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer, the tool that implements these oracles, has successfully found many logical bugs [30, 42].",
+      "context_before": "ually do not crash the program, we need an oracle to determine whether each execution produces the correct result or not. However, building an oracle is time-consuming and error-prone due to the various language dialects and features [49]. Recent progress on DBMS oracles shed light on logical bug detection [43 –45]. For example, Rigger et al. constructedseveral general oracles by transforming SQL queries into semantically equivalent forms. One DBMS may process these queries with different code paths, but the final results should be the same. Any inconsistency indicates a potential logical bug.",
+      "context_after": "However, SQLancer relies on a rule-based generator to synthesize original queries, which may limit its capability to explore program states. Specifically, it creates expressions for WHERE and JOIN clauses based on the grammar of the specific DBMS and the schema of the underlying database. Considering the huge query space, it could invest a great deal of time and effort on similar queries, which cannot inspect diverse program code. Coverage-guided program testing, or fuzzing, has been adopted to test a wide range of programs and successfully found thousands of memory-related bugs [18, 27, 32, 4",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3014,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "However, SQLancer relies on a rule-based generator to synthesize original queries, which may limit its capability to explore program states.",
+      "context_before": "rrect result or not. However, building an oracle is time-consuming and error-prone due to the various language dialects and features [49]. Recent progress on DBMS oracles shed light on logical bug detection [43 –45]. For example, Rigger et al. constructedseveral general oracles by transforming SQL queries into semantically equivalent forms. One DBMS may process these queries with different code paths, but the final results should be the same. Any inconsistency indicates a potential logical bug. SQLancer, the tool that implements these oracles, has successfully found many logical bugs [30, 42].",
+      "context_after": "Specifically, it creates expressions for WHERE and JOIN clauses based on the grammar of the specific DBMS and the schema of the underlying database. Considering the huge query space, it could invest a great deal of time and effort on similar queries, which cannot inspect diverse program code. Coverage-guided program testing, or fuzzing, has been adopted to test a wide range of programs and successfully found thousands of memory-related bugs [18, 27, 32, 48, 70]. The core idea of fuzzing is to utilize the code coverage to guide the input generation. A fuzzing tool, or fuzzer, picks up an input",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3115,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[10, 60, 71]",
+      "technique": null,
+      "sentence": "Previous works applied coverage-guided fuzzing to test DBMSs and demonstrated the benefits of detecting memory-related bugs [10, 60, 71].",
+      "context_before": "nds of memory-related bugs [18, 27, 32, 48, 70]. The core idea of fuzzing is to utilize the code coverage to guide the input generation. A fuzzing tool, or fuzzer, picks up an input from a queue and randomly updates it to produce new test cases. Then, it runs the program with the new test case and at the same time collects the code (like basic blocks or branches) being executed. If the execution triggers new code, the fuzzer will add the new test case to the queue and use it for future mutation; otherwise, the fuzzer will drop the new test case and move on to mutate another input in the queue.",
+      "context_after": "However, no research ever tried to apply coverage-based guidance on logical bug detection. In this paper, we aim to understand the benefits of coveragebased guidance on logical bug detection, specifically for DBMS systems. Our study reveals several challenges of adopting current mutation-based fuzzers to find logical bugs. The main problem comes from generating valid SQL queries. Since we need to compare the results of DBMS executions, the generated query should pass both syntax and semantic checks and successfully produce meaningful outputs. Existing fuzzers can synthesize decent queries tha",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4274,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[10, 71]",
+      "technique": null,
+      "sentence": "Currently, most fuzzers use one parser to handle different DBMS systems [10, 71].",
+      "context_before": "executions, the generated query should pass both syntax and semantic checks and successfully produce meaningful outputs. Existing fuzzers can synthesize decent queries that trigger assertion failures and crashes, but many of them are not completely USENIX Association 31st USENIX Security Symposium 4309 valid and thus cannot be used for finding logical bugs. To address this problem, we propose validity-oriented mutation, which defines a set of policies to improve the validity of generated queries. First, we design an automatic method to convert the SQL parser of each DBMS for fuzzing purposes.",
+      "context_after": "Since many DBMSs use their own SQL dialects to support unique features [49], the unified parser may produce a lot of incompatible queries. Even worse, the incorrect queries could trigger error-handling code and are prioritized by coverage-based guidance. Therefore, we provide one query parser for each DBMS to reduce the invalid queries. Our second effort is to build a context-based instantiation algorithm that enforces accurate dependencies between SQL elements. For example, DROP TABLE X removes table Xfrom the current database. Our method will follow the semantics and remove the correspondin",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 5440,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[43, 44]",
+      "technique": "norec",
+      "sentence": "We implemented four oracles, including two proposed in previous works [43, 44] and two proposed in this paper.",
+      "context_before": "for DBMS systems. SQLRight first mutates existing queries cooperatively. It inserts a set of oracle-required statements, and applies our validity-oriented mutations to improve the validity rate. Then, it sends the query to the oracle to create functionally equivalent query counterparts. SQLRight feeds all generated queries to the DBMS, and collects the execution results and the coverage information. After that, SQLRight invokes the oracle to compare the results of different queries to identify logical bugs. At last, it inserts the coverage-improving queries into the queue for future mutations.",
+      "context_after": "We evaluate SQLRight on three popular DBMS systems: SQLite [51], PostgreSQL [39] and MySQL [34]. SQLRight successfully detects 18 logic bugs within 60 days. We have reported all of our findings to their developers: all bugs get confirmed, and 14 have been fixed. To understand the con-01 CREATE TABLE person (pid INT); 02 INSERT INTO person VALUES (1), (10), (10); 03 CREATE UNIQUE INDEX idx ON person (pid) WHERE pid=1; 04 SELECT DISTINCT pid FROM person WHERE pid=10; 05 -- output: 10\\n10 06 -- expect: 10 Listing 1: A logical bug of SQLite due to the improper optimization on DISTINCT when the ind",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 7974,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We also compare our system with the stateof-the-art tools, including SQLancer (using oracles to detect logical bugs) and Squirrel (using code coverage to detect crashes and assertion failures).",
+      "context_before": "n-01 CREATE TABLE person (pid INT); 02 INSERT INTO person VALUES (1), (10), (10); 03 CREATE UNIQUE INDEX idx ON person (pid) WHERE pid=1; 04 SELECT DISTINCT pid FROM person WHERE pid=10; 05 -- output: 10\\n10 06 -- expect: 10 Listing 1: A logical bug of SQLite due to the improper optimization on DISTINCT when the index is a unique partial index. tributions of code coverage and query validity, we conduct unit tests by disabling each of them one by one. Our evaluation shows that both the coverage and the validity help find more logical bugs, but the latter makes more contributions than the former.",
+      "context_after": "We port our oracle APIs to Squirrel to help it detect logical bugs. After testing for 72 hours, SQLRight reports 12 unique logical bugs, Squirrel detects one bug, and SQLancer does not find any bug. The result shows that combining the coverage-based guidance and oracles can help trigger more logical bugs in DBMS systems. In summary, this paper makes the following contributions: •We study the efficacy of coverage-guided fuzzing in detecting logical bugs for DBMS systems and confirm two useful factors: query validity and code coverage. •We implement SQLRight, a coverage-guided fuzzer to detect",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 8969,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "After testing for 72 hours, SQLRight reports 12 unique logical bugs, Squirrel detects one bug, and SQLancer does not find any bug.",
+      "context_before": "e to the improper optimization on DISTINCT when the index is a unique partial index. tributions of code coverage and query validity, we conduct unit tests by disabling each of them one by one. Our evaluation shows that both the coverage and the validity help find more logical bugs, but the latter makes more contributions than the former. We also compare our system with the stateof-the-art tools, including SQLancer (using oracles to detect logical bugs) and Squirrel (using code coverage to detect crashes and assertion failures). We port our oracle APIs to Squirrel to help it detect logical bugs.",
+      "context_after": "The result shows that combining the coverage-based guidance and oracles can help trigger more logical bugs in DBMS systems. In summary, this paper makes the following contributions: •We study the efficacy of coverage-guided fuzzing in detecting logical bugs for DBMS systems and confirm two useful factors: query validity and code coverage. •We implement SQLRight, a coverage-guided fuzzer to detect logical bugs for DBMSs. SQLRight provides general APIs to simplify the oracle development and embeds validity-oriented mutations to improve the query quality. •We evaluate SQLRight on the real-world",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 9231,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker",
+      "surface": "[43 –45]",
+      "technique": "norec",
+      "sentence": "Recent works construct functionally equivalent queries and check whether the DBMS produces the same result for all queries [43 –45].",
+      "context_before": "or example, in Listing 1, the oracle should indicate the expected result as 10. However, constructing a complete, error-free oracle is challenging. Senior analysts can manually analyze results to identify bugs, but cannot handle large-scale test cases. Differential analysis provides an automatic, scalable solution [16,29,49,52]. For example, RAGS sends one query to different DBMSs and compares their results to find bugs [49]. However, due to the diverse dialects and extensions, popular DBMSs only share a small portion of features, and cross DBMS validation cannot detect DBMS-specific bugs [49].",
+      "context_after": "For example, for a given query, oracle NoREC shifts all conditions from WHERE clauses to SELECT expressions, which effectively disables most optimizations applied to the original query [43]. The number of TURE rows for the modified query should be the same as the number of rows for the original one. Oracle TLPsplits a condition xin aWHERE clause into three subqueries: x IS TRUE ,x IS FALSE, and x IS NULL. It combines the results from three subqueries and checks the equivalence with the original one [44]. As oracles are built on high-level semantics instead of low-level implementations, we can",
+      "section": "2.1 An Example Logical Bug",
+      "page": 4,
+      "char_offset": 13306,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For example, for a given query, oracle NoREC shifts all conditions from WHERE clauses to SELECT expressions, which effectively disables most optimizations applied to the original query [43].",
+      "context_before": "s challenging. Senior analysts can manually analyze results to identify bugs, but cannot handle large-scale test cases. Differential analysis provides an automatic, scalable solution [16,29,49,52]. For example, RAGS sends one query to different DBMSs and compares their results to find bugs [49]. However, due to the diverse dialects and extensions, popular DBMSs only share a small portion of features, and cross DBMS validation cannot detect DBMS-specific bugs [49]. Recent works construct functionally equivalent queries and check whether the DBMS produces the same result for all queries [43 –45].",
+      "context_after": "The number of TURE rows for the modified query should be the same as the number of rows for the original one. Oracle TLPsplits a condition xin aWHERE clause into three subqueries: x IS TRUE ,x IS FALSE, and x IS NULL. It combines the results from three subqueries and checks the equivalence with the original one [44]. As oracles are built on high-level semantics instead of low-level implementations, we can use them to test multiple DBMS systems [30]. We can define a simple oracle, dubbed Index, to detect the bug in Listing 1: removing indexes from the database should not affect the query resul",
+      "section": "2.1 An Example Logical Bug",
+      "page": 4,
+      "char_offset": 13439,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "citation_marker",
+      "surface": "[44]",
+      "technique": "tlp",
+      "sentence": "It combines the results from three subqueries and checks the equivalence with the original one [44].",
+      "context_before": "ross DBMS validation cannot detect DBMS-specific bugs [49]. Recent works construct functionally equivalent queries and check whether the DBMS produces the same result for all queries [43 –45]. For example, for a given query, oracle NoREC shifts all conditions from WHERE clauses to SELECT expressions, which effectively disables most optimizations applied to the original query [43]. The number of TURE rows for the modified query should be the same as the number of rows for the original one. Oracle TLPsplits a condition xin aWHERE clause into three subqueries: x IS TRUE ,x IS FALSE, and x IS NULL.",
+      "context_after": "As oracles are built on high-level semantics instead of low-level implementations, we can use them to test multiple DBMS systems [30]. We can define a simple oracle, dubbed Index, to detect the bug in Listing 1: removing indexes from the database should not affect the query result. We can implement oracle Index Mutation EngineValidation EngineResult AnalyzerSquirrelcode coverageSeed InputsCrash BugsDatabase Management SystemFigure 1: Overview of Squirrel [71]. It utilizes the coverage-based guidance to test DBMS systems for crashes and assertion failures. to delete existing indexes (or insert",
+      "section": "2.1 An Example Logical Bug",
+      "page": 4,
+      "char_offset": 13848,
+      "cited_reference": {
+        "number": 44,
+        "text": "Manuel Rigger and Zhendong Su. Finding Bugs in Database Systems via Query Partitioning. Proceedings of the ACM on Programming Languages, 4(OOPSLA):1–30, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "citation_marker",
+      "surface": "[30]",
+      "technique": null,
+      "sentence": "As oracles are built on high-level semantics instead of low-level implementations, we can use them to test multiple DBMS systems [30].",
+      "context_before": "lent queries and check whether the DBMS produces the same result for all queries [43 –45]. For example, for a given query, oracle NoREC shifts all conditions from WHERE clauses to SELECT expressions, which effectively disables most optimizations applied to the original query [43]. The number of TURE rows for the modified query should be the same as the number of rows for the original one. Oracle TLPsplits a condition xin aWHERE clause into three subqueries: x IS TRUE ,x IS FALSE, and x IS NULL. It combines the results from three subqueries and checks the equivalence with the original one [44].",
+      "context_after": "We can define a simple oracle, dubbed Index, to detect the bug in Listing 1: removing indexes from the database should not affect the query result. We can implement oracle Index Mutation EngineValidation EngineResult AnalyzerSquirrelcode coverageSeed InputsCrash BugsDatabase Management SystemFigure 1: Overview of Squirrel [71]. It utilizes the coverage-based guidance to test DBMS systems for crashes and assertion failures. to delete existing indexes (or inserts new ones) and check whether the DBMS produces the same result as before. One functionally equivalent form of Listing 1 is given in Li",
+      "section": "2.1 An Example Logical Bug",
+      "page": 4,
+      "char_offset": 13949,
+      "cited_reference": {
+        "number": 30,
+        "text": "Manuel Rigger. SQLancer. https://github.com/sqlancer/ sqlancer. (visited in June 2021).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M14",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "We can implement oracle Index Mutation EngineValidation EngineResult AnalyzerSquirrelcode coverageSeed InputsCrash BugsDatabase Management SystemFigure 1: Overview of Squirrel [71].",
+      "context_before": "he number of TURE rows for the modified query should be the same as the number of rows for the original one. Oracle TLPsplits a condition xin aWHERE clause into three subqueries: x IS TRUE ,x IS FALSE, and x IS NULL. It combines the results from three subqueries and checks the equivalence with the original one [44]. As oracles are built on high-level semantics instead of low-level implementations, we can use them to test multiple DBMS systems [30]. We can define a simple oracle, dubbed Index, to detect the bug in Listing 1: removing indexes from the database should not affect the query result.",
+      "context_after": "It utilizes the coverage-based guidance to test DBMS systems for crashes and assertion failures. to delete existing indexes (or inserts new ones) and check whether the DBMS produces the same result as before. One functionally equivalent form of Listing 1 is given in Listing 2 by deleting the CREATE UNIQUE INDEX statement. We can use oracle Index to detect this bug without any domain knowledge. More importantly, we can extend the test to a large scale. 2.3 Coverage-guided Testing Coverage-guided testing, or fuzzing [32], has been widely used to test a large set of programs and successfully fou",
+      "section": "2.1 An Example Logical Bug",
+      "page": 4,
+      "char_offset": 14232,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[21, 26, 60, 71]",
+      "technique": null,
+      "sentence": "Recent works also ported coverage-guided fuzzing to test DBMS systems [21, 26, 60, 71].",
+      "context_before": "eats the input as the proof-of-concept (PoC) of the underlying bug. For input that triggers no crashes, the fuzzer will check whether the execution reaches new code paths, like basic blocks or branches. If so, it will add the input into a queue. Otherwise, it will drop the input and pick up the next input from the queue for another round of fuzzing. Coverageguided testing has been used to test various programs, including but not limited to operating systems [11, 22, 37, 63, 65], compilers [10, 19, 38], web browsers [17, 48, 66], document readers [12, 64], and even smart contracts [20, 35, 62].",
+      "context_after": "Figure 1 shows an overview ofSquirrel [71], a recent work that aims to detect crashes and assertion failures from DBMS systems. Squirrel takes a set of inputs, i.e., SQL queries, as the seed of fuzzing. It first translates the query into an intermediate representation (IR), which contains many structural information. Then, Squirrel takes three mutation methods to modify the query IR and create new ones, including node insertion, deletion and replacement. For each newly generated IR, Squirrel builds the data dependency graph between different operands (e.g., tables and columns) and fills the o",
+      "section": "2.3 Coverage-guided Testing",
+      "page": 4,
+      "char_offset": 15998,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "Figure 1 shows an overview ofSquirrel [71], a recent work that aims to detect crashes and assertion failures from DBMS systems.",
+      "context_before": "ers no crashes, the fuzzer will check whether the execution reaches new code paths, like basic blocks or branches. If so, it will add the input into a queue. Otherwise, it will drop the input and pick up the next input from the queue for another round of fuzzing. Coverageguided testing has been used to test various programs, including but not limited to operating systems [11, 22, 37, 63, 65], compilers [10, 19, 38], web browsers [17, 48, 66], document readers [12, 64], and even smart contracts [20, 35, 62]. Recent works also ported coverage-guided fuzzing to test DBMS systems [21, 26, 60, 71].",
+      "context_after": "Squirrel takes a set of inputs, i.e., SQL queries, as the seed of fuzzing. It first translates the query into an intermediate representation (IR), which contains many structural information. Then, Squirrel takes three mutation methods to modify the query IR and create new ones, including node insertion, deletion and replacement. For each newly generated IR, Squirrel builds the data dependency graph between different operands (e.g., tables and columns) and fills the operands with randomly generated strings. After that, it translates the new IR back to queries and feeds it to DBMS. At last, Squ",
+      "section": "2.3 Coverage-guided Testing",
+      "page": 4,
+      "char_offset": 16086,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "Recent fuzzers utilize typebased mutation and semantic-guided instantiation in order to produce valid queries [71].",
+      "context_before": "oracle APIs to update oracle-specific SQL statements to prepare for result checking. For query validation, it again calls oracle APIs to translate the query into semantically equivalent variants. After the execution, it relies on the oracle to decide whether the queries trigger logical bugs. 3.1 Validity-oriented Query Generation SQLRight demands high-quality SQL queries to stress various aspects of different DBMSs. Any queries that cause syntax or semantic errors would not be useful for finding logical bugs. Unfortunately, generating semantically correct queries has proven to be NP-hard [29].",
+      "context_after": "However, their validity rate, about 30%, is still insufficient to effectively test DBMS systems. Even worse, most of the generated queries are not ready for detecting logical bugs. Therefore, we proposed several practical techniques to improve query validity. 3.1.1 Cooperative Mutation SQLRight takes two separate mutation strategies in parallel to generate different components of a query set. As shown in Figure 3, it maintains two queues: the select queue contains only SELECT statements and is used to generate proper SELECT queries that produce outputs; the normal queue hosts other statements",
+      "section": "3.1 Validity-oriented Query Generation",
+      "page": 5,
+      "char_offset": 21267,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "To support the NoREC oracle, we fix SELECT_statement ,FROM and WHERE of the statement and leave others for the mutation engine to change.",
+      "context_before": "CT statements to the select queue, and keeps other statements in the normal queue. For each fuzzing round, SQLRight collects a set of statements from the normal queue and relies on the normal mutation engine to generate new queries. Then, it 4312 31st USENIX Security Symposium USENIX Association abc SELECT COUNT(*) FROM v0 WHERE v1 = 0;opt_WHERE_statementWHEREwhere_clauseopt_FROM_statementFROMfrom_clauseSELECT_statementSELECT v1 = 0 FROM v0;from_clausewhere_clauseopt_FROM_statementFROMSELECTselect_clauseSELECTSELECT_statementoriginal:mutation:Restricted Figure 4: Cooperative mutation on SELECT.",
+      "context_after": "invokes the cooperative mutation of the oracle to create and append several SELECT statements. After the combination, we leverage the instantiation to build a concrete query set. During the cooperative mutation, the oracle preserves the query elements that are useful for the correctness test. For example, oracle NoREC requires the SELECT statement to have both FROM clause and WHERE clause. We provide interfaces for the oracle to notify the mutation engine not to delete these necessary nodes or add new nodes if the original statement does not have one. Figure 4 illustrates how to control mutat",
+      "section": "3.1.1 Cooperative Mutation",
+      "page": 6,
+      "char_offset": 22755,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For example, oracle NoREC requires the SELECT statement to have both FROM clause and WHERE clause.",
+      "context_before": "= 0 FROM v0;from_clausewhere_clauseopt_FROM_statementFROMSELECTselect_clauseSELECTSELECT_statementoriginal:mutation:Restricted Figure 4: Cooperative mutation on SELECT. To support the NoREC oracle, we fix SELECT_statement ,FROM and WHERE of the statement and leave others for the mutation engine to change. invokes the cooperative mutation of the oracle to create and append several SELECT statements. After the combination, we leverage the instantiation to build a concrete query set. During the cooperative mutation, the oracle preserves the query elements that are useful for the correctness test.",
+      "context_after": "We provide interfaces for the oracle to notify the mutation engine not to delete these necessary nodes or add new nodes if the original statement does not have one. Figure 4 illustrates how to control mutations on the statement SELECT COUNT(*) FROM v0 WHERE v1=0. Given the IR of the query, we add attributes to three nodes, specifically, SELECT_statment ,FROM, and WHERE, to mark them as immutable. As a result, the crucial components for oracle NoREC are preserved. The mutation engine still has full flexibility to update the from_clause node and the where_clause node to generate new SELECT stat",
+      "section": "3.1.1 Cooperative Mutation",
+      "page": 6,
+      "char_offset": 23186,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "As a result, the crucial components for oracle NoREC are preserved.",
+      "context_before": "operative mutation, the oracle preserves the query elements that are useful for the correctness test. For example, oracle NoREC requires the SELECT statement to have both FROM clause and WHERE clause. We provide interfaces for the oracle to notify the mutation engine not to delete these necessary nodes or add new nodes if the original statement does not have one. Figure 4 illustrates how to control mutations on the statement SELECT COUNT(*) FROM v0 WHERE v1=0. Given the IR of the query, we add attributes to three nodes, specifically, SELECT_statment ,FROM, and WHERE, to mark them as immutable.",
+      "context_after": "The mutation engine still has full flexibility to update the from_clause node and the where_clause node to generate new SELECT statements. The IR-based mutation engine supports various patterns and different conditions in the from_clause andwhere_clause, and thus provides rich capability comparable to the unconstrained mutator [71]. 3.1.2 Dedicated Parsing We customize our mutation engine for each DBMS to improve syntax correctness. Most popular DBMS systems have their own customized SQL dialects, which only share limited features. Therefore, one unified SQL grammar for common functionalities",
+      "section": "3.1.1 Cooperative Mutation",
+      "page": 6,
+      "char_offset": 23685,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "The IR-based mutation engine supports various patterns and different conditions in the from_clause andwhere_clause, and thus provides rich capability comparable to the unconstrained mutator [71].",
+      "context_before": "vide interfaces for the oracle to notify the mutation engine not to delete these necessary nodes or add new nodes if the original statement does not have one. Figure 4 illustrates how to control mutations on the statement SELECT COUNT(*) FROM v0 WHERE v1=0. Given the IR of the query, we add attributes to three nodes, specifically, SELECT_statment ,FROM, and WHERE, to mark them as immutable. As a result, the crucial components for oracle NoREC are preserved. The mutation engine still has full flexibility to update the from_clause node and the where_clause node to generate new SELECT statements.",
+      "context_after": "3.1.2 Dedicated Parsing We customize our mutation engine for each DBMS to improve syntax correctness. Most popular DBMS systems have their own customized SQL dialects, which only share limited features. Therefore, one unified SQL grammar for common functionalities only covers a small portion of DBMS code and cannot find bugs in DBMS-unique features [49]. In contrast, a catch-all grammar that aims to support all features could lead to many invalid queries. Listing 3 shows an example that a string with the single quote symbol has different meanings in different DBMSs. SQLite can run this query",
+      "section": "3.1.1 Cooperative Mutation",
+      "page": 6,
+      "char_offset": 23892,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M22",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "3 Context-based IR Instantiation The previous work Squirrel [71] identifies data dependency among all SQL statements to help instantiate query operands, like table names and column names.",
+      "context_before": "r front-ends to SQLRight ’s IR. Thanks to the well-documented Bison format,01 CREATE TABLE v0 (v1 TEXT); 02 INSERT INTO v0 VALUES ( 'text '); 03 SELECT v1 FROM 'v0'; Listing 3: Queries allowed in SQLite but rejected by PostgreSQL. SQLite takes flexible rules to identify strings and can treat the ‘v0’ as an identifier, while PostgreSQL strictly takes ‘v0’ as a string. our tool can easily port different parsers for SQLRight. With the original DBMS parser front-ends, SQLRight can support the full dialect syntaxes for each DBMS and guarantees high syntax correctness to improve query validity. 3.1.",
+      "context_after": "However, the dependency graph constructed by Squirrel tightly couples multiple SQL statements into one static graph. When dealing with complicated queries, it cannot update the data dependency to reflect the dynamic SQL context. Specifically, it keeps the static data dependency unchanged across the whole query sequence and cannot adjust itself to fit in the ever-changing relationship between different SQL statements. To address this limitation, we design a context-based IR instantiation algorithm. It dynamically updates the data dependency based on the SQL context and fills in accurate concre",
+      "section": "3.1.3 Context-based IR Instantiation",
+      "page": 6,
+      "char_offset": 25839,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M23",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For example, because oracle NoREC shifts conditions from WHERE to SELECT, it only appends SELECT statements that have a FROM clause and a WHERE clause (both clauses are optional for SELECT ).",
+      "context_before": "nspect the results of DBMS executions to identify bugs, SQLRight provides the append_output() API for each oracle to insert proper output-generation statements. The SELECT statement queries the underlying database and returns the results, so the default behavior of this API is to append several SELECT statements to the end of the given query set. During the query validation, these statements will be filled with proper table names and column names so that they can produce meaningful results. Oracles can have their special output statements or apply other policies based on their functionalities.",
+      "context_after": "Transformation. The core task of an oracle is to transform one query into different equivalent variants. transform() accepts the query set as input and returns one or more equivalent query sets. For example, oracle NoREC returns only one variant where the WHERE conditions in the original query are moved to the SELECT expressions. Oracle Index may insert or delete various CREATE INDEX statements into the given query set and thus can result in multiple variants. SQLRight executes all variants returned by transform() and compares their results. 4314 31st USENIX Security Symposium USENIX Associat",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 7,
+      "char_offset": 33474,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M24",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For example, oracle NoREC returns only one variant where the WHERE conditions in the original query are moved to the SELECT expressions.",
+      "context_before": "atements will be filled with proper table names and column names so that they can produce meaningful results. Oracles can have their special output statements or apply other policies based on their functionalities. For example, because oracle NoREC shifts conditions from WHERE to SELECT, it only appends SELECT statements that have a FROM clause and a WHERE clause (both clauses are optional for SELECT ). Transformation. The core task of an oracle is to transform one query into different equivalent variants. transform() accepts the query set as input and returns one or more equivalent query sets.",
+      "context_after": "Oracle Index may insert or delete various CREATE INDEX statements into the given query set and thus can result in multiple variants. SQLRight executes all variants returned by transform() and compares their results. 4314 31st USENIX Security Symposium USENIX Association API Basic NoREC TLP Index Rowid preprocess() check applicability basic basic remove UNIQUE INDEX basic attach_output() append random SELECT SELECT COUNT(*) FROM x WHERE x; basic basic basic transform() - expr: WHERE→SELECT /0→T∪F∪NULL insert CREATE INDEX add WITHOUT ROWID to table compare() #rows = #rows #rows = #(TRUE rows) ba",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 7,
+      "char_offset": 33860,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "4314 31st USENIX Security Symposium USENIX Association API Basic NoREC TLP Index Rowid preprocess() check applicability basic basic remove UNIQUE INDEX basic attach_output() append random SELECT SELECT COUNT(*) FROM x WHERE x; basic basic basic transform() - expr: WHERE→SELECT /0→T∪F∪NULL insert CREATE INDEX add WITHOUT ROWID to table compare() #rows = #rows #rows = #(TRUE rows) basic basic basic Table 1: Oracles and their APIs.",
+      "context_before": "HERE clause (both clauses are optional for SELECT ). Transformation. The core task of an oracle is to transform one query into different equivalent variants. transform() accepts the query set as input and returns one or more equivalent query sets. For example, oracle NoREC returns only one variant where the WHERE conditions in the original query are moved to the SELECT expressions. Oracle Index may insert or delete various CREATE INDEX statements into the given query set and thus can result in multiple variants. SQLRight executes all variants returned by transform() and compares their results.",
+      "context_after": "Our system currently supports four oracles: NoREC ,TLP,Index andRowid .Basic is the base interface that implements default behaviors. ∪combines the results of different queries; -means no functionality; basic means the default behavior in Basic. Result Comparison. SQLRight allows the oracle to define their own comparison methods to identify unexpected results. A simple comparison algorithm could require all results to be exactly the same. However, this method could be overly strict and may introduce false alarms. For example, in the results of the Index oracle, the order of rows could be diff",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 7,
+      "char_offset": 34213,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Our system currently supports four oracles: NoREC ,TLP,Index andRowid .",
+      "context_before": "E INDEX statements into the given query set and thus can result in multiple variants. SQLRight executes all variants returned by transform() and compares their results. 4314 31st USENIX Security Symposium USENIX Association API Basic NoREC TLP Index Rowid preprocess() check applicability basic basic remove UNIQUE INDEX basic attach_output() append random SELECT SELECT COUNT(*) FROM x WHERE x; basic basic basic transform() - expr: WHERE→SELECT /0→T∪F∪NULL insert CREATE INDEX add WITHOUT ROWID to table compare() #rows = #rows #rows = #(TRUE rows) basic basic basic Table 1: Oracles and their APIs.",
+      "context_after": "Basic is the base interface that implements default behaviors. ∪combines the results of different queries; -means no functionality; basic means the default behavior in Basic. Result Comparison. SQLRight allows the oracle to define their own comparison methods to identify unexpected results. A simple comparison algorithm could require all results to be exactly the same. However, this method could be overly strict and may introduce false alarms. For example, in the results of the Index oracle, the order of rows could be different due to extra indexes. Therefore, we provide a loose comparison fun",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 8,
+      "char_offset": 34645,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "We take special actions to handle oracle NoREC, as the new query form always generates one row for each record ( TRUE or FALSE based on the condition).",
+      "context_before": ". Result Comparison. SQLRight allows the oracle to define their own comparison methods to identify unexpected results. A simple comparison algorithm could require all results to be exactly the same. However, this method could be overly strict and may introduce false alarms. For example, in the results of the Index oracle, the order of rows could be different due to extra indexes. Therefore, we provide a loose comparison function that merely checks the number of rows in the results. However, if aggregate functions (like MINandSUM) are used in SELECT, the comparison should expect the same output.",
+      "context_after": "In this case, the compare() API should compare the number of rows in the original result and the number of TRUE rows in the transformed one. We present the detailed steps that SQLRight takes to find the bug of Listing 1 using oracle Index in Appendix B. 4 Implementation We implement SQLRight as a prototype of the first coverageguided DBMS fuzzer for finding logical bugs. Our implementation is based on Squirrel [71] and SQLancer [30]. Specifically, we adopt the query-mutation module of Squirrel and implement our general oracle APIs and validity-oriented mutations. Currently, SQLRight supports",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 8,
+      "char_offset": 35490,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Our implementation is based on Squirrel [71] and SQLancer [30].",
+      "context_before": "MINandSUM) are used in SELECT, the comparison should expect the same output. We take special actions to handle oracle NoREC, as the new query form always generates one row for each record ( TRUE or FALSE based on the condition). In this case, the compare() API should compare the number of rows in the original result and the number of TRUE rows in the transformed one. We present the detailed steps that SQLRight takes to find the bug of Listing 1 using oracle Index in Appendix B. 4 Implementation We implement SQLRight as a prototype of the first coverageguided DBMS fuzzer for finding logical bugs.",
+      "context_after": "Specifically, we adopt the query-mutation module of Squirrel and implement our general oracle APIs and validity-oriented mutations. Currently, SQLRight supports four oracles, including NoREC and TLPported from SQLancer [43, 44], and Index and Rowid we propose in this paper. We port the parsers of SQLite, MySQL, and PostgreSQL to SQLRight to test these DBMSs effectively. Next, we present several implementation details that could help the future development of other DBMS fuzzers. Memory-efficient Mutation. We redesign the way of storing SQL queries to reduce memory use. Squirrel uses a library to",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 8,
+      "char_offset": 36015,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M29",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Currently, SQLRight supports four oracles, including NoREC and TLPported from SQLancer [43, 44], and Index and Rowid we propose in this paper.",
+      "context_before": "FALSE based on the condition). In this case, the compare() API should compare the number of rows in the original result and the number of TRUE rows in the transformed one. We present the detailed steps that SQLRight takes to find the bug of Listing 1 using oracle Index in Appendix B. 4 Implementation We implement SQLRight as a prototype of the first coverageguided DBMS fuzzer for finding logical bugs. Our implementation is based on Squirrel [71] and SQLancer [30]. Specifically, we adopt the query-mutation module of Squirrel and implement our general oracle APIs and validity-oriented mutations.",
+      "context_after": "We port the parsers of SQLite, MySQL, and PostgreSQL to SQLRight to test these DBMSs effectively. Next, we present several implementation details that could help the future development of other DBMS fuzzers. Memory-efficient Mutation. We redesign the way of storing SQL queries to reduce memory use. Squirrel uses a library to map each node type, like select-clause, to a set of IR nodes. At the time of insertion or replacement, it visits the library to retrieve a set of type-matched candidates. Although keeping IR nodes in memory accelerates the mutation, each IR node occupies non-trivial memory",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 8,
+      "char_offset": 36211,
+      "found_by_all": [
+        "name",
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "1) USENIX Association 31st USENIX Security Symposium 4315 ID DBMS Description Oracle D Status Fix Logical Bugs 1SQLite UNIQUE PARTIAL INDEX ,DISTINCT INDEX 5 fixed c2f940b 2SQLite JOIN and LIKELY/UNLIKELY NoREC 3 fixed 2363a14 3SQLite IN-early-out optimization TLP 2 fixed eb40248 4SQLite EXISTS (SELECT.",
+      "context_before": "ation. After removing one node, if the remaining query is still valid and can trigger the bug, we will delete more nodes from the current statements; otherwise, we will add the node back and move on to delete the next node. The minimization algorithm keeps the deletion process until that removing any node will render the remaining part invalid or miss the bug. In that case, we will report the current query as the minimum version. 5 Evaluation We evaluate our tool SQLRight on real-world popular DBMS systems to answer the following questions: Q1. Can SQLRight detect real-world logical bugs? (§5.",
+      "context_after": "..) to IN NoREC 2 fixed 16252d7 5SQLite Misuse aggregate in ORDER BY NoREC 2 fixed 0d11d77 6SQLite Aggregate in ORDER BY NoREC 1 confirmed 7SQLite WITHOUT ROWID with DESC NoREC 7 fixed f65c929 8SQLite unordered WITHOUT ROWID table ROWID 2 fixed c21bc5a 9SQLite mixed table name and CTE NoREC 2 fixed 0f0959c 10SQLite inconsistent constant propagaion NoREC 1 fixed 9be208a 11SQLite IS NOT NULL optimization NoREC 2 fixed 8cc2393 12SQLite large value loss of precision NoREC 4 fixed f9c6426 13SQLite constraint check in ALTER TABLE NoREC 2 fixed e379499 14SQLite equivalence transfer optimization NoREC 2",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 8,
+      "char_offset": 39957,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M31",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": ") to IN NoREC 2 fixed 16252d7 5SQLite Misuse aggregate in ORDER BY NoREC 2 fixed 0d11d77 6SQLite Aggregate in ORDER BY NoREC 1 confirmed 7SQLite WITHOUT ROWID with DESC NoREC 7 fixed f65c929 8SQLite unordered WITHOUT ROWID table ROWID 2 fixed c21bc5a 9SQLite mixed table name and CTE NoREC 2 fixed 0f0959c 10SQLite inconsistent constant propagaion NoREC 1 fixed 9be208a 11SQLite IS NOT NULL optimization NoREC 2 fixed 8cc2393 12SQLite large value loss of precision NoREC 4 fixed f9c6426 13SQLite constraint check in ALTER TABLE NoREC 2 fixed e379499 14SQLite equivalence transfer optimization NoREC 2 fixed 8b24c17 15MySQL ANY ALL optimization error NoREC 5 confirmed 16MySQL UNIQUE INDEX with NULL NoREC 1 confirmed 17MySQL GTID_SUBSET NoREC 4 fixed 8.",
+      "context_before": "e will render the remaining part invalid or miss the bug. In that case, we will report the current query as the minimum version. 5 Evaluation We evaluate our tool SQLRight on real-world popular DBMS systems to answer the following questions: Q1. Can SQLRight detect real-world logical bugs? (§5.1) USENIX Association 31st USENIX Security Symposium 4315 ID DBMS Description Oracle D Status Fix Logical Bugs 1SQLite UNIQUE PARTIAL INDEX ,DISTINCT INDEX 5 fixed c2f940b 2SQLite JOIN and LIKELY/UNLIKELY NoREC 3 fixed 2363a14 3SQLite IN-early-out optimization TLP 2 fixed eb40248 4SQLite EXISTS (SELECT...",
+      "context_after": "0.30 18MySQL large value in GROUP BYTLP 3 confirmed Crashes 1SQLite distinct aggregation-4 fixed 0e47898 2SQLite ROWID from views-2 fixed 0f0959c 3SQLite nested CTE-5 fixed 94225d6 4SQLite foreign_key_check-2 fixed 68db1ff 5SQLite UPDATE from VIRTUAL TABLE-3 fixed 2547cfe Assertion Failure 1SQLite pLeft==pRight-3 fixed 240f749 2SQLite iColnCol-4 fixed b986600 3SQLite memIsValid(&aMem[pOp->p3]) - 2 fixed c9f0b9c 4SQLite target>0&&target<=pParse->nMem-2 fixed 7072404 Table 2: Detected Bugs .SQLRight detected 27 bugs, including 18 logical bugs, 5 crashes and 4 assertion failures. We have reported",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 9,
+      "char_offset": 40262,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "citation_marker",
+      "surface": "[21,43 –45,71]",
+      "technique": "norec",
+      "sentence": "To answer Q1, we use SQLRight to test three popular DBMS systems, SQLite ,MySQL andPostgreSQL, which are commonly used in previous works to evaluate bugfinding tools for DBMSs [21,43 –45,71].",
+      "context_before": "-3 fixed 240f749 2SQLite iColnCol-4 fixed b986600 3SQLite memIsValid(&aMem[pOp->p3]) - 2 fixed c9f0b9c 4SQLite target>0&&target<=pParse->nMem-2 fixed 7072404 Table 2: Detected Bugs .SQLRight detected 27 bugs, including 18 logical bugs, 5 crashes and 4 assertion failures. We have reported these bugs to their developers and got 27 confirmed and 23 fixed. D (depth) means the number of mutations to trigger the bug. Q2. Can SQLRight find more bugs than existing tools? (§5.2) Q3. How does code coverage guide the testing? (§5.3) Q4. How does query validity help detect bugs? (§5.4) Experimental Setup.",
+      "context_after": "For Q2, we compare SQLRight with SQLancer and Squirrel, the state-of-the-art bug-finding tools for DBMSs. Since Squirrel cannot detect logical bugs, we port our oracles to Squirrel, denoted as Squirrel +oracle. To answer Q3, we compare coverage feedback with three methods to handle new queries: dropping all, saving all and randomly saving some. To answer Q4, we disable each query-validity component of SQLRight to find the impact on code coverage, query validity and bug detection. We conduct SQLite experiments on a Ubuntu 20.04 system with two 28-cores Intel(R) Xeon(R) Gold 6258R CPUs and 791",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 9,
+      "char_offset": 41925,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M33",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "For Q2, we compare SQLRight with SQLancer and Squirrel, the state-of-the-art bug-finding tools for DBMSs.",
+      "context_before": "etected 27 bugs, including 18 logical bugs, 5 crashes and 4 assertion failures. We have reported these bugs to their developers and got 27 confirmed and 23 fixed. D (depth) means the number of mutations to trigger the bug. Q2. Can SQLRight find more bugs than existing tools? (§5.2) Q3. How does code coverage guide the testing? (§5.3) Q4. How does query validity help detect bugs? (§5.4) Experimental Setup. To answer Q1, we use SQLRight to test three popular DBMS systems, SQLite ,MySQL andPostgreSQL, which are commonly used in previous works to evaluate bugfinding tools for DBMSs [21,43 –45,71].",
+      "context_after": "Since Squirrel cannot detect logical bugs, we port our oracles to Squirrel, denoted as Squirrel +oracle. To answer Q3, we compare coverage feedback with three methods to handle new queries: dropping all, saving all and randomly saving some. To answer Q4, we disable each query-validity component of SQLRight to find the impact on code coverage, query validity and bug detection. We conduct SQLite experiments on a Ubuntu 20.04 system with two 28-cores Intel(R) Xeon(R) Gold 6258R CPUs and 791 GB memory. We perform the experiments of PostgreSQL and MySQL on three computers with Ubuntu 20.04 system,",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 9,
+      "char_offset": 42117,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M34",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Since SQLancer requires the particular SQLite version 3.",
+      "context_before": "acles to Squirrel, denoted as Squirrel +oracle. To answer Q3, we compare coverage feedback with three methods to handle new queries: dropping all, saving all and randomly saving some. To answer Q4, we disable each query-validity component of SQLRight to find the impact on code coverage, query validity and bug detection. We conduct SQLite experiments on a Ubuntu 20.04 system with two 28-cores Intel(R) Xeon(R) Gold 6258R CPUs and 791 GB memory. We perform the experiments of PostgreSQL and MySQL on three computers with Ubuntu 20.04 system, 8-cores Intel(R) Core(TM) i7-10700 CPU, and 64 GB memory.",
+      "context_after": "34.0, we adopt this version for evaluation. For other DBMSs, we choose the latest version, specifically PostgreSQL 14.0 andMySQL 8.0.27. We compile SQLite andPostgreSQL using AFL [70] but enlarge the coverage map size to 256K to mitigate the collision issue [15]. Since MySQL is multi-threaded, to avoid data race on updating the coverage map, we instru-ment it with block coverage and disable hit counts. We collect the seed corpus from the official unit tests of each DBMS, specifically, SQLite TCL Test Scripts ,PostgreSQL official test infrastructure, and MySQL unit test samples. We use the same",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 9,
+      "char_offset": 42881,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M35",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer is a generation-based tool and does not require any seed input.",
+      "context_before": "ion. For other DBMSs, we choose the latest version, specifically PostgreSQL 14.0 andMySQL 8.0.27. We compile SQLite andPostgreSQL using AFL [70] but enlarge the coverage map size to 256K to mitigate the collision issue [15]. Since MySQL is multi-threaded, to avoid data race on updating the coverage map, we instru-ment it with block coverage and disable hit counts. We collect the seed corpus from the official unit tests of each DBMS, specifically, SQLite TCL Test Scripts ,PostgreSQL official test infrastructure, and MySQL unit test samples. We use the same seed corpus for SQLRight andSquirrel .",
+      "context_after": "5.1 DBMS Logical Bugs Due to the limited availability of resources, we conduct bugfinding experiments for each DBMS with different durations. Specifically, we spend 60 days testing SQLite, 14 days testingPostgreSQL and 7 days testing MySQL. In total, SQLRight detects 27 bugs, including 18 logical bugs, 5 crashes, and 4 assertion failures. Except for 3 logical bugs from MySQL, all other bugs are from SQLite; we do not find any bugs from PostgreSQL. Despite the different testing durations, we can find the similar pattern of bug numbers from previous works [43 –45, 71], where most bugs are from",
+      "section": "3.1.4 Non-determinism Mitigation",
+      "page": 9,
+      "char_offset": 43577,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M36",
+      "found_by": "citation_marker",
+      "surface": "[43 –45, 71]",
+      "technique": "norec",
+      "sentence": "Despite the different testing durations, we can find the similar pattern of bug numbers from previous works [43 –45, 71], where most bugs are from SQLite, and very few are from PostgreSQL.",
+      "context_before": "QL unit test samples. We use the same seed corpus for SQLRight andSquirrel .SQLancer is a generation-based tool and does not require any seed input. 5.1 DBMS Logical Bugs Due to the limited availability of resources, we conduct bugfinding experiments for each DBMS with different durations. Specifically, we spend 60 days testing SQLite, 14 days testingPostgreSQL and 7 days testing MySQL. In total, SQLRight detects 27 bugs, including 18 logical bugs, 5 crashes, and 4 assertion failures. Except for 3 logical bugs from MySQL, all other bugs are from SQLite; we do not find any bugs from PostgreSQL.",
+      "context_after": "We have reported all findings to their developers. All bugs have been confirmed, and 23 of them have been fixed. SQLRight detects more logical bugs than crashes and assertion failures. Since we design a set of solutions to improve the query validity, SQLRight generates fewer abnormal statements that render the DBMS crashing. Table 2 provides more bug details. Column “Description” shows that logical bugs come from diverse optimizations, and many bugs are due to an interplay of multiple optimizations. For example, the first bug is caused by an improper DISTINCT optimization when the table has a",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 9,
+      "char_offset": 44101,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M37",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Most bugs are detected by NoREC, and other oracles detect one or two bugs.",
+      "context_before": "we design a set of solutions to improve the query validity, SQLRight generates fewer abnormal statements that render the DBMS crashing. Table 2 provides more bug details. Column “Description” shows that logical bugs come from diverse optimizations, and many bugs are due to an interplay of multiple optimizations. For example, the first bug is caused by an improper DISTINCT optimization when the table has a unique partial index: DISTINCT and UNIQUE PARTIAL INDEX work well alone, but the bug appears when a query satisfies both conditions. Column “Oracle” presents the oracle that detects each bug.",
+      "context_after": "In fact, several bugs can be detected by multiple oracles through different queries, but we only record the first detection and treat others as duplications. We present the details of two logical bugs to help demonstrate the necessity of feedback and validity for bug finding. Appendix A presents the details of three more logical bugs. Necessity of Coverage-based Feedback. Listing 6 demonstrates a SQLite logical bug due to incorrect index lookup. The first statement creates a WITHOUT ROWID table with two columns. v1is the PRIMARY KEY of the table and is in DESC order. The second statement inse",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 9,
+      "char_offset": 45081,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M38",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "0 12 24 36 48 60 720123456Unique Bug (a)SQLite logical bugs 0 12 24 36 48 60 7201234Unique Bug (b)MySQL logcial bugs 0 12 24 36 48 60 722025303540Coverage (k) SQLRight Squirrel+ oracle SQLancer 0 12 24 36 48 60 722025303540Coverage (k) (c)SQLite code coverage 0 12 24 36 48 60 72100120140160180200Coverage (k) (d)MySQL code coverage 0 12 24 36 48 60 72020406080100120Coverage (k) (e)PostgreSQL code coverage 0 12 24 36 48 60 7230405060708090Query Validity (%) (f)SQLite query validity 0 12 24 36 48 60 720102030405060Query Validity (%) (g)MySQL query validity 0 12 24 36 48 60 72051015202530Query Validity (%) (h)PostgreSQL query validity 0 12 24 36 48 60 72105106107Valid Queries per Hour (i)SQLite valid stmts / hr 0 12 24 36 48 60 72100101102103104105Valid Queries per Hour (j)MySQL valid stmts / hr 0 12 24 36 48 60 72101103105107Valid Queries per Hour (k)PostgreSQL valid stmts / hr Figure 5: Comparison between different tools ( NoREC ).",
+      "context_before": "WHERE v1 = v2 AND v1 = 'x'; 10 -- output: Error: ambiguous column name: v1 Listing 7: A logical bug of SQLite related to column aliases. SELECT statements should return nothing Oracle FeedbackMutation Depth Max 0 1∼3 4∼7 7+ Depth NoRECSQLRight 24738 9761 3125 2443 25 SQLRight drop 24738 10478 - - 1 SQLRight rand 24738 7753 2561 - 7 SQLRight save 24738 10030 356 - 4 TLPSQLRight 24738 9662 3594 1580 21 SQLRight drop 24738 10294 - - 1 SQLRight rand 24738 7217 881 - 7 SQLRight save 24738 8916 - - 3 Table 3: Code coverage triggered by queries with different depths. SQLRight maintains higher depth.",
+      "context_after": "a-k show the number of unique bugs, code coverage, query validity and validity over time. SQLancer does not implement NoREC forMySQL, so we skip the evaluation. We run each fuzzing instance for 72 hours for and repeat five times. From the seed input, SQLRight must accumulate seven consecutive mutations to generate this bug-triggering query. Each mutation triggers new code coverage and is saved to the queue. Without the guidance of code coverage, SQLRight can hardly maintain such a deep mutation chain, nor detect this bug. Necessity of SQL Validity. With validity-improvement techniques, SQLRig",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 47686,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M39",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer does not implement NoREC forMySQL, so we skip the evaluation.",
+      "context_before": "8090Query Validity (%) (f)SQLite query validity 0 12 24 36 48 60 720102030405060Query Validity (%) (g)MySQL query validity 0 12 24 36 48 60 72051015202530Query Validity (%) (h)PostgreSQL query validity 0 12 24 36 48 60 72105106107Valid Queries per Hour (i)SQLite valid stmts / hr 0 12 24 36 48 60 72100101102103104105Valid Queries per Hour (j)MySQL valid stmts / hr 0 12 24 36 48 60 72101103105107Valid Queries per Hour (k)PostgreSQL valid stmts / hr Figure 5: Comparison between different tools ( NoREC ). a-k show the number of unique bugs, code coverage, query validity and validity over time.",
+      "context_after": "We run each fuzzing instance for 72 hours for and repeat five times. From the seed input, SQLRight must accumulate seven consecutive mutations to generate this bug-triggering query. Each mutation triggers new code coverage and is saved to the queue. Without the guidance of code coverage, SQLRight can hardly maintain such a deep mutation chain, nor detect this bug. Necessity of SQL Validity. With validity-improvement techniques, SQLRight can detect bugs that Squirrel +oracle and SQLancer will miss. Listing 7 shows one logical bug that can only be detected with the per-DBMS parser and the conte",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 48727,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M40",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "With validity-improvement techniques, SQLRight can detect bugs that Squirrel +oracle and SQLancer will miss.",
+      "context_before": "Comparison between different tools ( NoREC ). a-k show the number of unique bugs, code coverage, query validity and validity over time. SQLancer does not implement NoREC forMySQL, so we skip the evaluation. We run each fuzzing instance for 72 hours for and repeat five times. From the seed input, SQLRight must accumulate seven consecutive mutations to generate this bug-triggering query. Each mutation triggers new code coverage and is saved to the queue. Without the guidance of code coverage, SQLRight can hardly maintain such a deep mutation chain, nor detect this bug. Necessity of SQL Validity.",
+      "context_after": "Listing 7 shows one logical bug that can only be detected with the per-DBMS parser and the contextbased instantiation. The first statement creates table v0with two columns: v1and v2. The second line creates an index attached to two v2. The third statement inserts one row into the table. Two SELECT statements search in v0 to find rows where v1=v2 andv1=’x’. These conditions are not satisfied, asv1andv2have different values. However, if we refer table v0using aliases, SQLite will incorrectly optimize the query to ignore the first condition and return the row. SQLancer cannot detect this bug sin",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 49192,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M41",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer cannot detect this bug since it does not use aliases when generating SELECT statements.",
+      "context_before": "rrel +oracle and SQLancer will miss. Listing 7 shows one logical bug that can only be detected with the per-DBMS parser and the contextbased instantiation. The first statement creates table v0with two columns: v1and v2. The second line creates an index attached to two v2. The third statement inserts one row into the table. Two SELECT statements search in v0 to find rows where v1=v2 andv1=’x’. These conditions are not satisfied, asv1andv2have different values. However, if we refer table v0using aliases, SQLite will incorrectly optimize the query to ignore the first condition and return the row.",
+      "context_after": "Although Squirrel +oracle allows alias during parsing, it does not use it for instantiation. If the table has aliases, Squirrel +oracle will resolve the condition tov1=v2, resulting in an error of “ambiguous column name”. SQLRight adopts context-based instantiation, which avoids the collision problem and successfully triggers this bug.5.2 Comparison with Existing Tools SQLancer contains three oracles, NoREC ,TLP, and PQS. Both NoREC and TLPjust modify query statements to detect bugs, and SQLRight can easily support them using our general APIs. However, PQSrelies on database content to construct",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 49865,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M42",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "2 Comparison with Existing Tools SQLancer contains three oracles, NoREC ,TLP, and PQS.",
+      "context_before": "ndv2have different values. However, if we refer table v0using aliases, SQLite will incorrectly optimize the query to ignore the first condition and return the row. SQLancer cannot detect this bug since it does not use aliases when generating SELECT statements. Although Squirrel +oracle allows alias during parsing, it does not use it for instantiation. If the table has aliases, Squirrel +oracle will resolve the condition tov1=v2, resulting in an error of “ambiguous column name”. SQLRight adopts context-based instantiation, which avoids the collision problem and successfully triggers this bug.5.",
+      "context_after": "Both NoREC and TLPjust modify query statements to detect bugs, and SQLRight can easily support them using our general APIs. However, PQSrelies on database content to construct statements, and we consider supporting it in the future. Therefore, we use NoREC and TLPto compare SQLRight ,SQLancer, and Squirrel +oracle. We launch five instances in each setting and run them for 72 hours. Since SQLancer does not support NoREC forMySQL, we skip this setting. Figure 5 ( NoREC ) and Figure 8 (TLP) in Appendix show the evaluation results. Unique Logical Bugs. Based on Figure 5a b and Figure 8a b,SQLRight r",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 50302,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp",
+        "pqs"
+      ]
+    },
+    {
+      "id": "M43",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Both NoREC and TLPjust modify query statements to detect bugs, and SQLRight can easily support them using our general APIs.",
+      "context_before": "orrectly optimize the query to ignore the first condition and return the row. SQLancer cannot detect this bug since it does not use aliases when generating SELECT statements. Although Squirrel +oracle allows alias during parsing, it does not use it for instantiation. If the table has aliases, Squirrel +oracle will resolve the condition tov1=v2, resulting in an error of “ambiguous column name”. SQLRight adopts context-based instantiation, which avoids the collision problem and successfully triggers this bug.5.2 Comparison with Existing Tools SQLancer contains three oracles, NoREC ,TLP, and PQS.",
+      "context_after": "However, PQSrelies on database content to construct statements, and we consider supporting it in the future. Therefore, we use NoREC and TLPto compare SQLRight ,SQLancer, and Squirrel +oracle. We launch five instances in each setting and run them for 72 hours. Since SQLancer does not support NoREC forMySQL, we skip this setting. Figure 5 ( NoREC ) and Figure 8 (TLP) in Appendix show the evaluation results. Unique Logical Bugs. Based on Figure 5a b and Figure 8a b,SQLRight reports the most 12 bugs across all settings, including 6 SQLite bugs and 3 MySQL bugs using NoREC, and 2SQLite bugs and 1",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 50388,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M44",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Therefore, we use NoREC and TLPto compare SQLRight ,SQLancer, and Squirrel +oracle.",
+      "context_before": "t does not use it for instantiation. If the table has aliases, Squirrel +oracle will resolve the condition tov1=v2, resulting in an error of “ambiguous column name”. SQLRight adopts context-based instantiation, which avoids the collision problem and successfully triggers this bug.5.2 Comparison with Existing Tools SQLancer contains three oracles, NoREC ,TLP, and PQS. Both NoREC and TLPjust modify query statements to detect bugs, and SQLRight can easily support them using our general APIs. However, PQSrelies on database content to construct statements, and we consider supporting it in the future.",
+      "context_after": "We launch five instances in each setting and run them for 72 hours. Since SQLancer does not support NoREC forMySQL, we skip this setting. Figure 5 ( NoREC ) and Figure 8 (TLP) in Appendix show the evaluation results. Unique Logical Bugs. Based on Figure 5a b and Figure 8a b,SQLRight reports the most 12 bugs across all settings, including 6 SQLite bugs and 3 MySQL bugs using NoREC, and 2SQLite bugs and 1 MySQL bug using TLP.Squirrel +oracle merely found 1 bug, which is from SQLite using NoREC; SQLancer did not find any logical bug. The empty outcome from SQLancer could be due to its extensive",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 50619,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M45",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Since SQLancer does not support NoREC forMySQL, we skip this setting.",
+      "context_before": "column name”. SQLRight adopts context-based instantiation, which avoids the collision problem and successfully triggers this bug.5.2 Comparison with Existing Tools SQLancer contains three oracles, NoREC ,TLP, and PQS. Both NoREC and TLPjust modify query statements to detect bugs, and SQLRight can easily support them using our general APIs. However, PQSrelies on database content to construct statements, and we consider supporting it in the future. Therefore, we use NoREC and TLPto compare SQLRight ,SQLancer, and Squirrel +oracle. We launch five instances in each setting and run them for 72 hours.",
+      "context_after": "Figure 5 ( NoREC ) and Figure 8 (TLP) in Appendix show the evaluation results. Unique Logical Bugs. Based on Figure 5a b and Figure 8a b,SQLRight reports the most 12 bugs across all settings, including 6 SQLite bugs and 3 MySQL bugs using NoREC, and 2SQLite bugs and 1 MySQL bug using TLP.Squirrel +oracle merely found 1 bug, which is from SQLite using NoREC; SQLancer did not find any logical bug. The empty outcome from SQLancer could be due to its extensive use in testing these DBMS systems [43 –45]. The generation-based method has reached its limit on finding more bugs in its supported mutati",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 50770,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M46",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Figure 5 ( NoREC ) and Figure 8 (TLP) in Appendix show the evaluation results.",
+      "context_before": "ds the collision problem and successfully triggers this bug.5.2 Comparison with Existing Tools SQLancer contains three oracles, NoREC ,TLP, and PQS. Both NoREC and TLPjust modify query statements to detect bugs, and SQLRight can easily support them using our general APIs. However, PQSrelies on database content to construct statements, and we consider supporting it in the future. Therefore, we use NoREC and TLPto compare SQLRight ,SQLancer, and Squirrel +oracle. We launch five instances in each setting and run them for 72 hours. Since SQLancer does not support NoREC forMySQL, we skip this setting.",
+      "context_after": "Unique Logical Bugs. Based on Figure 5a b and Figure 8a b,SQLRight reports the most 12 bugs across all settings, including 6 SQLite bugs and 3 MySQL bugs using NoREC, and 2SQLite bugs and 1 MySQL bug using TLP.Squirrel +oracle merely found 1 bug, which is from SQLite using NoREC; SQLancer did not find any logical bug. The empty outcome from SQLancer could be due to its extensive use in testing these DBMS systems [43 –45]. The generation-based method has reached its limit on finding more bugs in its supported mutations. This demonstrates the necessity of bringing more diverse mechanisms to det",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 50840,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M47",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Based on Figure 5a b and Figure 8a b,SQLRight reports the most 12 bugs across all settings, including 6 SQLite bugs and 3 MySQL bugs using NoREC, and 2SQLite bugs and 1 MySQL bug using TLP.",
+      "context_before": "cer contains three oracles, NoREC ,TLP, and PQS. Both NoREC and TLPjust modify query statements to detect bugs, and SQLRight can easily support them using our general APIs. However, PQSrelies on database content to construct statements, and we consider supporting it in the future. Therefore, we use NoREC and TLPto compare SQLRight ,SQLancer, and Squirrel +oracle. We launch five instances in each setting and run them for 72 hours. Since SQLancer does not support NoREC forMySQL, we skip this setting. Figure 5 ( NoREC ) and Figure 8 (TLP) in Appendix show the evaluation results. Unique Logical Bugs.",
+      "context_after": "Squirrel +oracle merely found 1 bug, which is from SQLite using NoREC; SQLancer did not find any logical bug. The empty outcome from SQLancer could be due to its extensive use in testing these DBMS systems [43 –45]. The generation-based method has reached its limit on finding more bugs in its supported mutations. This demonstrates the necessity of bringing more diverse mechanisms to detect logical bugs. SQLRight combines code coverage, oracles, and query validity, and can detect significantly more logical bugs than SQLite andMySQL. USENIX Association 31st USENIX Security Symposium 4317 0 4 8 1",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 50940,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M48",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Squirrel +oracle merely found 1 bug, which is from SQLite using NoREC; SQLancer did not find any logical bug.",
+      "context_before": "on database content to construct statements, and we consider supporting it in the future. Therefore, we use NoREC and TLPto compare SQLRight ,SQLancer, and Squirrel +oracle. We launch five instances in each setting and run them for 72 hours. Since SQLancer does not support NoREC forMySQL, we skip this setting. Figure 5 ( NoREC ) and Figure 8 (TLP) in Appendix show the evaluation results. Unique Logical Bugs. Based on Figure 5a b and Figure 8a b,SQLRight reports the most 12 bugs across all settings, including 6 SQLite bugs and 3 MySQL bugs using NoREC, and 2SQLite bugs and 1 MySQL bug using TLP.",
+      "context_after": "The empty outcome from SQLancer could be due to its extensive use in testing these DBMS systems [43 –45]. The generation-based method has reached its limit on finding more bugs in its supported mutations. This demonstrates the necessity of bringing more diverse mechanisms to detect logical bugs. SQLRight combines code coverage, oracles, and query validity, and can detect significantly more logical bugs than SQLite andMySQL. USENIX Association 31st USENIX Security Symposium 4317 0 4 8 12 16 20 240123456Unique Bug (a)bugs of SQLite (NoREC ) 0 4 8 12 16 20 240123Unique Bug (b)bugs of SQLite (TL",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 51130,
+      "found_by_all": [
+        "name",
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M49",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "The empty outcome from SQLancer could be due to its extensive use in testing these DBMS systems [43 –45].",
+      "context_before": "oREC and TLPto compare SQLRight ,SQLancer, and Squirrel +oracle. We launch five instances in each setting and run them for 72 hours. Since SQLancer does not support NoREC forMySQL, we skip this setting. Figure 5 ( NoREC ) and Figure 8 (TLP) in Appendix show the evaluation results. Unique Logical Bugs. Based on Figure 5a b and Figure 8a b,SQLRight reports the most 12 bugs across all settings, including 6 SQLite bugs and 3 MySQL bugs using NoREC, and 2SQLite bugs and 1 MySQL bug using TLP.Squirrel +oracle merely found 1 bug, which is from SQLite using NoREC; SQLancer did not find any logical bug.",
+      "context_after": "The generation-based method has reached its limit on finding more bugs in its supported mutations. This demonstrates the necessity of bringing more diverse mechanisms to detect logical bugs. SQLRight combines code coverage, oracles, and query validity, and can detect significantly more logical bugs than SQLite andMySQL. USENIX Association 31st USENIX Security Symposium 4317 0 4 8 12 16 20 240123456Unique Bug (a)bugs of SQLite (NoREC ) 0 4 8 12 16 20 240123Unique Bug (b)bugs of SQLite (TLP) 0 4 8 12 16 20 242025303540Coverage (k) (c)coverage of SQLite (NoREC ) 0 4 8 12 16 20 242025303540Cove",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 51239,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M50",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "USENIX Association 31st USENIX Security Symposium 4317 0 4 8 12 16 20 240123456Unique Bug (a)bugs of SQLite (NoREC ) 0 4 8 12 16 20 240123Unique Bug (b)bugs of SQLite (TLP) 0 4 8 12 16 20 242025303540Coverage (k) (c)coverage of SQLite (NoREC ) 0 4 8 12 16 20 242025303540Coverage (k) (d)coverage of SQLite (TLP) SQLRight SQLRightdrop SQLRightrand SQLRightsave Figure 6: Contribution of code-coverage to DBMS fuzzing.",
+      "context_before": "L bugs using NoREC, and 2SQLite bugs and 1 MySQL bug using TLP.Squirrel +oracle merely found 1 bug, which is from SQLite using NoREC; SQLancer did not find any logical bug. The empty outcome from SQLancer could be due to its extensive use in testing these DBMS systems [43 –45]. The generation-based method has reached its limit on finding more bugs in its supported mutations. This demonstrates the necessity of bringing more diverse mechanisms to detect logical bugs. SQLRight combines code coverage, oracles, and query validity, and can detect significantly more logical bugs than SQLite andMySQL.",
+      "context_after": "We launch five instances for each tool and run each instance for 24 hours. a and b show the number of unique SQLite bugs found by NoREC and TLP; c and d show the average number of SQLite edges using NoREC and TLP. All tools fail to find any bug in PostgreSQL, indicating its good code quality observed by previous works [43–45, 71]. Code Coverage. According to Figure 5c d e and Figure 8c d e,SQLRight triggers noticeably higher code coverage than others on three DBMS systems. Considering that SQLancer is a generation-based tester and has a high validity rate, the extra coverage of SQLRight over SQ",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 10,
+      "char_offset": 51667,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M51",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "a and b show the number of unique SQLite bugs found by NoREC and TLP; c and d show the average number of SQLite edges using NoREC and TLP.",
+      "context_before": "overage, oracles, and query validity, and can detect significantly more logical bugs than SQLite andMySQL. USENIX Association 31st USENIX Security Symposium 4317 0 4 8 12 16 20 240123456Unique Bug (a)bugs of SQLite (NoREC ) 0 4 8 12 16 20 240123Unique Bug (b)bugs of SQLite (TLP) 0 4 8 12 16 20 242025303540Coverage (k) (c)coverage of SQLite (NoREC ) 0 4 8 12 16 20 242025303540Coverage (k) (d)coverage of SQLite (TLP) SQLRight SQLRightdrop SQLRightrand SQLRightsave Figure 6: Contribution of code-coverage to DBMS fuzzing. We launch five instances for each tool and run each instance for 24 hours.",
+      "context_after": "All tools fail to find any bug in PostgreSQL, indicating its good code quality observed by previous works [43–45, 71]. Code Coverage. According to Figure 5c d e and Figure 8c d e,SQLRight triggers noticeably higher code coverage than others on three DBMS systems. Considering that SQLancer is a generation-based tester and has a high validity rate, the extra coverage of SQLRight over SQLancer mainly comes from the coverage-based guidance and full-featured per-DBMS parsers. Although Squirrel +oracle also utilizes code coverage to guide testing, the accurate parser and context-based instantiation",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 52161,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M52",
+      "found_by": "citation_marker",
+      "surface": "[43–45, 71]",
+      "technique": "norec",
+      "sentence": "All tools fail to find any bug in PostgreSQL, indicating its good code quality observed by previous works [43–45, 71].",
+      "context_before": "Security Symposium 4317 0 4 8 12 16 20 240123456Unique Bug (a)bugs of SQLite (NoREC ) 0 4 8 12 16 20 240123Unique Bug (b)bugs of SQLite (TLP) 0 4 8 12 16 20 242025303540Coverage (k) (c)coverage of SQLite (NoREC ) 0 4 8 12 16 20 242025303540Coverage (k) (d)coverage of SQLite (TLP) SQLRight SQLRightdrop SQLRightrand SQLRightsave Figure 6: Contribution of code-coverage to DBMS fuzzing. We launch five instances for each tool and run each instance for 24 hours. a and b show the number of unique SQLite bugs found by NoREC and TLP; c and d show the average number of SQLite edges using NoREC and TLP.",
+      "context_after": "Code Coverage. According to Figure 5c d e and Figure 8c d e,SQLRight triggers noticeably higher code coverage than others on three DBMS systems. Considering that SQLancer is a generation-based tester and has a high validity rate, the extra coverage of SQLRight over SQLancer mainly comes from the coverage-based guidance and full-featured per-DBMS parsers. Although Squirrel +oracle also utilizes code coverage to guide testing, the accurate parser and context-based instantiation help SQLRight outperform this state-of-the-art coverage-based DBMS tester. For example, given the same seed corpus, SQ",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 52298,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M53",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Considering that SQLancer is a generation-based tester and has a high validity rate, the extra coverage of SQLRight over SQLancer mainly comes from the coverage-based guidance and full-featured per-DBMS parsers.",
+      "context_before": "age of SQLite (TLP) SQLRight SQLRightdrop SQLRightrand SQLRightsave Figure 6: Contribution of code-coverage to DBMS fuzzing. We launch five instances for each tool and run each instance for 24 hours. a and b show the number of unique SQLite bugs found by NoREC and TLP; c and d show the average number of SQLite edges using NoREC and TLP. All tools fail to find any bug in PostgreSQL, indicating its good code quality observed by previous works [43–45, 71]. Code Coverage. According to Figure 5c d e and Figure 8c d e,SQLRight triggers noticeably higher code coverage than others on three DBMS systems.",
+      "context_after": "Although Squirrel +oracle also utilizes code coverage to guide testing, the accurate parser and context-based instantiation help SQLRight outperform this state-of-the-art coverage-based DBMS tester. For example, given the same seed corpus, SQLRight can correctly parse more queries and achieves higher code coverage even at the beginning. With the higher coverage, SQLRight is able to find more bugs. Query Validity. From Figure 5f g h and Figure 8f g h we can find that SQLancer achieves the highest query validity, where over 80% of all queries are valid for SQLite, 99% for MySQL, and over 28% fo",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 52562,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M54",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "From Figure 5f g h and Figure 8f g h we can find that SQLancer achieves the highest query validity, where over 80% of all queries are valid for SQLite, 99% for MySQL, and over 28% for PostgreSQL.",
+      "context_before": "a generation-based tester and has a high validity rate, the extra coverage of SQLRight over SQLancer mainly comes from the coverage-based guidance and full-featured per-DBMS parsers. Although Squirrel +oracle also utilizes code coverage to guide testing, the accurate parser and context-based instantiation help SQLRight outperform this state-of-the-art coverage-based DBMS tester. For example, given the same seed corpus, SQLRight can correctly parse more queries and achieves higher code coverage even at the beginning. With the higher coverage, SQLRight is able to find more bugs. Query Validity.",
+      "context_after": "Our tool SQLRight achieves around 30%, 25%, 10% validity for three tested DBMSs. This result is reasonable since SQLancer follows well-defined rules to generate SQL statements, leading to more valid queries. Nevertheless, SQLRight achieves significantly higher validity than the Squirrel +oracle, which can hardly produce any useful queries for oracles. This result confirms the necessity to improve the validity of detecting logical bugs. Another observation is that SQLancer can keep a consistent high validity rate for 72 hours. Mutation-based fuzzers, regardless of their configurations, get a l",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 53191,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M55",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "This result is reasonable since SQLancer follows well-defined rules to generate SQL statements, leading to more valid queries.",
+      "context_before": "nd context-based instantiation help SQLRight outperform this state-of-the-art coverage-based DBMS tester. For example, given the same seed corpus, SQLRight can correctly parse more queries and achieves higher code coverage even at the beginning. With the higher coverage, SQLRight is able to find more bugs. Query Validity. From Figure 5f g h and Figure 8f g h we can find that SQLancer achieves the highest query validity, where over 80% of all queries are valid for SQLite, 99% for MySQL, and over 28% for PostgreSQL. Our tool SQLRight achieves around 30%, 25%, 10% validity for three tested DBMSs.",
+      "context_after": "Nevertheless, SQLRight achieves significantly higher validity than the Squirrel +oracle, which can hardly produce any useful queries for oracles. This result confirms the necessity to improve the validity of detecting logical bugs. Another observation is that SQLancer can keep a consistent high validity rate for 72 hours. Mutation-based fuzzers, regardless of their configurations, get a lower validity rate as time goes on. The reason is that during fuzzing, some invalid queries trigger new code coverage ( e.g., error-handling code) and are added to the queue. Mutating invalid queries will lik",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 53468,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M56",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Another observation is that SQLancer can keep a consistent high validity rate for 72 hours.",
+      "context_before": "h we can find that SQLancer achieves the highest query validity, where over 80% of all queries are valid for SQLite, 99% for MySQL, and over 28% for PostgreSQL. Our tool SQLRight achieves around 30%, 25%, 10% validity for three tested DBMSs. This result is reasonable since SQLancer follows well-defined rules to generate SQL statements, leading to more valid queries. Nevertheless, SQLRight achieves significantly higher validity than the Squirrel +oracle, which can hardly produce any useful queries for oracles. This result confirms the necessity to improve the validity of detecting logical bugs.",
+      "context_after": "Mutation-based fuzzers, regardless of their configurations, get a lower validity rate as time goes on. The reason is that during fuzzing, some invalid queries trigger new code coverage ( e.g., error-handling code) and are added to the queue. Mutating invalid queries will likely produce more invalid ones, which turns the overall validity rate down. We also measure the speed of generating valid queries and get the same pattern, shown in Figure 5i j k and Figure 8i j k. SQLancer can generate more valid queries than others, but its efficiency varies over time. Overall, SQLRight can find more logi",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 53827,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M57",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer can generate more valid queries than others, but its efficiency varies over time.",
+      "context_before": "validity of detecting logical bugs. Another observation is that SQLancer can keep a consistent high validity rate for 72 hours. Mutation-based fuzzers, regardless of their configurations, get a lower validity rate as time goes on. The reason is that during fuzzing, some invalid queries trigger new code coverage ( e.g., error-handling code) and are added to the queue. Mutating invalid queries will likely produce more invalid ones, which turns the overall validity rate down. We also measure the speed of generating valid queries and get the same pattern, shown in Figure 5i j k and Figure 8i j k.",
+      "context_after": "Overall, SQLRight can find more logical bugs than SQLancer andSquirrel +oracle. It also outperforms existing tools in triggering more program code. Although SQLancer produces high-quality queries, its lack of query diversity makes it less effective in finding new logical bugs.5.3 Contribution of Coverage Feedback We compare SQLRight with three different feedback methods: SQLRight dropdrops all generated queries and only mutates seed inputs; SQLRight savesaves all generated queries (no matter whether they trigger new coverage or not) into the queue and takes turns to mutate each query; SQLRigh",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 54391,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M58",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Overall, SQLRight can find more logical bugs than SQLancer andSquirrel +oracle.",
+      "context_before": "tent high validity rate for 72 hours. Mutation-based fuzzers, regardless of their configurations, get a lower validity rate as time goes on. The reason is that during fuzzing, some invalid queries trigger new code coverage ( e.g., error-handling code) and are added to the queue. Mutating invalid queries will likely produce more invalid ones, which turns the overall validity rate down. We also measure the speed of generating valid queries and get the same pattern, shown in Figure 5i j k and Figure 8i j k. SQLancer can generate more valid queries than others, but its efficiency varies over time.",
+      "context_after": "It also outperforms existing tools in triggering more program code. Although SQLancer produces high-quality queries, its lack of query diversity makes it less effective in finding new logical bugs.5.3 Contribution of Coverage Feedback We compare SQLRight with three different feedback methods: SQLRight dropdrops all generated queries and only mutates seed inputs; SQLRight savesaves all generated queries (no matter whether they trigger new coverage or not) into the queue and takes turns to mutate each query; SQLRight randrandomly saves 10% of all generated queries to the queue. We perform the u",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 54482,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M59",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Although SQLancer produces high-quality queries, its lack of query diversity makes it less effective in finding new logical bugs.",
+      "context_before": "son is that during fuzzing, some invalid queries trigger new code coverage ( e.g., error-handling code) and are added to the queue. Mutating invalid queries will likely produce more invalid ones, which turns the overall validity rate down. We also measure the speed of generating valid queries and get the same pattern, shown in Figure 5i j k and Figure 8i j k. SQLancer can generate more valid queries than others, but its efficiency varies over time. Overall, SQLRight can find more logical bugs than SQLancer andSquirrel +oracle. It also outperforms existing tools in triggering more program code.",
+      "context_after": "5.3 Contribution of Coverage Feedback We compare SQLRight with three different feedback methods: SQLRight dropdrops all generated queries and only mutates seed inputs; SQLRight savesaves all generated queries (no matter whether they trigger new coverage or not) into the queue and takes turns to mutate each query; SQLRight randrandomly saves 10% of all generated queries to the queue. We perform the unit tests using SQLite with NoREC and TLP. For each setting, we ran five instances in parallel for 24 hours. Figure 6 shows the average results of code coverage and logical bugs. We also measure the",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 54630,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M60",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "We perform the unit tests using SQLite with NoREC and TLP.",
+      "context_before": "Squirrel +oracle. It also outperforms existing tools in triggering more program code. Although SQLancer produces high-quality queries, its lack of query diversity makes it less effective in finding new logical bugs.5.3 Contribution of Coverage Feedback We compare SQLRight with three different feedback methods: SQLRight dropdrops all generated queries and only mutates seed inputs; SQLRight savesaves all generated queries (no matter whether they trigger new coverage or not) into the queue and takes turns to mutate each query; SQLRight randrandomly saves 10% of all generated queries to the queue.",
+      "context_after": "For each setting, we ran five instances in parallel for 24 hours. Figure 6 shows the average results of code coverage and logical bugs. We also measure the contribution of queries in different depths. Specifically, if a query is generated from seed inputs through Nmutations, we define its depth asN. Table 3 shows the breakdown of code coverage regarding query depth. Unique Logical Bugs. Figure 6a and b show the number of detected bugs. For both NoREC and TLP,SQLRight reports the most bugs, including 4 bugs using NoREC and 2 bugs using TLP. Using NoREC ,SQLRight dropand SQLRight savereport 2 bu",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 55145,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M61",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "For both NoREC and TLP,SQLRight reports the most bugs, including 4 bugs using NoREC and 2 bugs using TLP.",
+      "context_before": "kes turns to mutate each query; SQLRight randrandomly saves 10% of all generated queries to the queue. We perform the unit tests using SQLite with NoREC and TLP. For each setting, we ran five instances in parallel for 24 hours. Figure 6 shows the average results of code coverage and logical bugs. We also measure the contribution of queries in different depths. Specifically, if a query is generated from seed inputs through Nmutations, we define its depth asN. Table 3 shows the breakdown of code coverage regarding query depth. Unique Logical Bugs. Figure 6a and b show the number of detected bugs.",
+      "context_after": "Using NoREC ,SQLRight dropand SQLRight savereport 2 bugs, and SQLRight randonly detects 1. Without code coverage, SQLRight did not find any logical bug using TLP. Code Coverage. According to Figure 6c and d, SQLRight achieves the highest code coverage: 40.1K branches with NoREC and 39.2K branches with TLP.SQLRight dropdetects 35.2K branches using NoREC and 35.0K branches using TLP, which are 13.7% and 11.9% fewer than SQLRight. SQLRight randand SQLRight savetrigger even fewer branches than SQLRight drop. The result shows the benefit of coveragebased guidance in generating more diverse queries.",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 55643,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M62",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Using NoREC ,SQLRight dropand SQLRight savereport 2 bugs, and SQLRight randonly detects 1.",
+      "context_before": "perform the unit tests using SQLite with NoREC and TLP. For each setting, we ran five instances in parallel for 24 hours. Figure 6 shows the average results of code coverage and logical bugs. We also measure the contribution of queries in different depths. Specifically, if a query is generated from seed inputs through Nmutations, we define its depth asN. Table 3 shows the breakdown of code coverage regarding query depth. Unique Logical Bugs. Figure 6a and b show the number of detected bugs. For both NoREC and TLP,SQLRight reports the most bugs, including 4 bugs using NoREC and 2 bugs using TLP.",
+      "context_after": "Without code coverage, SQLRight did not find any logical bug using TLP. Code Coverage. According to Figure 6c and d, SQLRight achieves the highest code coverage: 40.1K branches with NoREC and 39.2K branches with TLP.SQLRight dropdetects 35.2K branches using NoREC and 35.0K branches using TLP, which are 13.7% and 11.9% fewer than SQLRight. SQLRight randand SQLRight savetrigger even fewer branches than SQLRight drop. The result shows the benefit of coveragebased guidance in generating more diverse queries. Mutation Depth. Table 3 shows that SQLRight maintains deeper mutation chains, where SQLRig",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 55748,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M63",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Without code coverage, SQLRight did not find any logical bug using TLP.",
+      "context_before": "nces in parallel for 24 hours. Figure 6 shows the average results of code coverage and logical bugs. We also measure the contribution of queries in different depths. Specifically, if a query is generated from seed inputs through Nmutations, we define its depth asN. Table 3 shows the breakdown of code coverage regarding query depth. Unique Logical Bugs. Figure 6a and b show the number of detected bugs. For both NoREC and TLP,SQLRight reports the most bugs, including 4 bugs using NoREC and 2 bugs using TLP. Using NoREC ,SQLRight dropand SQLRight savereport 2 bugs, and SQLRight randonly detects 1.",
+      "context_after": "Code Coverage. According to Figure 6c and d, SQLRight achieves the highest code coverage: 40.1K branches with NoREC and 39.2K branches with TLP.SQLRight dropdetects 35.2K branches using NoREC and 35.0K branches using TLP, which are 13.7% and 11.9% fewer than SQLRight. SQLRight randand SQLRight savetrigger even fewer branches than SQLRight drop. The result shows the benefit of coveragebased guidance in generating more diverse queries. Mutation Depth. Table 3 shows that SQLRight maintains deeper mutation chains, where SQLRight can accumulate 21 mutations (column “Max Depth”) while others at most",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 55839,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M64",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "1K branches with NoREC and 39.",
+      "context_before": "Specifically, if a query is generated from seed inputs through Nmutations, we define its depth asN. Table 3 shows the breakdown of code coverage regarding query depth. Unique Logical Bugs. Figure 6a and b show the number of detected bugs. For both NoREC and TLP,SQLRight reports the most bugs, including 4 bugs using NoREC and 2 bugs using TLP. Using NoREC ,SQLRight dropand SQLRight savereport 2 bugs, and SQLRight randonly detects 1. Without code coverage, SQLRight did not find any logical bug using TLP. Code Coverage. According to Figure 6c and d, SQLRight achieves the highest code coverage: 40.",
+      "context_after": "2K branches with TLP.SQLRight dropdetects 35.2K branches using NoREC and 35.0K branches using TLP, which are 13.7% and 11.9% fewer than SQLRight. SQLRight randand SQLRight savetrigger even fewer branches than SQLRight drop. The result shows the benefit of coveragebased guidance in generating more diverse queries. Mutation Depth. Table 3 shows that SQLRight maintains deeper mutation chains, where SQLRight can accumulate 21 mutations (column “Max Depth”) while others at most accumulate 7 mutations. High-depth queries help SQLRight trigger more code coverage. In NoREC, besides the seed-triggered c",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 56005,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M65",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "2K branches with TLP.",
+      "context_before": "nerated from seed inputs through Nmutations, we define its depth asN. Table 3 shows the breakdown of code coverage regarding query depth. Unique Logical Bugs. Figure 6a and b show the number of detected bugs. For both NoREC and TLP,SQLRight reports the most bugs, including 4 bugs using NoREC and 2 bugs using TLP. Using NoREC ,SQLRight dropand SQLRight savereport 2 bugs, and SQLRight randonly detects 1. Without code coverage, SQLRight did not find any logical bug using TLP. Code Coverage. According to Figure 6c and d, SQLRight achieves the highest code coverage: 40.1K branches with NoREC and 39.",
+      "context_after": "SQLRight dropdetects 35.2K branches using NoREC and 35.0K branches using TLP, which are 13.7% and 11.9% fewer than SQLRight. SQLRight randand SQLRight savetrigger even fewer branches than SQLRight drop. The result shows the benefit of coveragebased guidance in generating more diverse queries. Mutation Depth. Table 3 shows that SQLRight maintains deeper mutation chains, where SQLRight can accumulate 21 mutations (column “Max Depth”) while others at most accumulate 7 mutations. High-depth queries help SQLRight trigger more code coverage. In NoREC, besides the seed-triggered code (depth 0), 14.9%",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 56035,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M66",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "2K branches using NoREC and 35.",
+      "context_before": "we define its depth asN. Table 3 shows the breakdown of code coverage regarding query depth. Unique Logical Bugs. Figure 6a and b show the number of detected bugs. For both NoREC and TLP,SQLRight reports the most bugs, including 4 bugs using NoREC and 2 bugs using TLP. Using NoREC ,SQLRight dropand SQLRight savereport 2 bugs, and SQLRight randonly detects 1. Without code coverage, SQLRight did not find any logical bug using TLP. Code Coverage. According to Figure 6c and d, SQLRight achieves the highest code coverage: 40.1K branches with NoREC and 39.2K branches with TLP.SQLRight dropdetects 35.",
+      "context_after": "0K branches using TLP, which are 13.7% and 11.9% fewer than SQLRight. SQLRight randand SQLRight savetrigger even fewer branches than SQLRight drop. The result shows the benefit of coveragebased guidance in generating more diverse queries. Mutation Depth. Table 3 shows that SQLRight maintains deeper mutation chains, where SQLRight can accumulate 21 mutations (column “Max Depth”) while others at most accumulate 7 mutations. High-depth queries help SQLRight trigger more code coverage. In NoREC, besides the seed-triggered code (depth 0), 14.9% of new coverage is triggered by queries with 4-7 depth,",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 56080,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M67",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "0K branches using TLP, which are 13.",
+      "context_before": "3 shows the breakdown of code coverage regarding query depth. Unique Logical Bugs. Figure 6a and b show the number of detected bugs. For both NoREC and TLP,SQLRight reports the most bugs, including 4 bugs using NoREC and 2 bugs using TLP. Using NoREC ,SQLRight dropand SQLRight savereport 2 bugs, and SQLRight randonly detects 1. Without code coverage, SQLRight did not find any logical bug using TLP. Code Coverage. According to Figure 6c and d, SQLRight achieves the highest code coverage: 40.1K branches with NoREC and 39.2K branches with TLP.SQLRight dropdetects 35.2K branches using NoREC and 35.",
+      "context_after": "7% and 11.9% fewer than SQLRight. SQLRight randand SQLRight savetrigger even fewer branches than SQLRight drop. The result shows the benefit of coveragebased guidance in generating more diverse queries. Mutation Depth. Table 3 shows that SQLRight maintains deeper mutation chains, where SQLRight can accumulate 21 mutations (column “Max Depth”) while others at most accumulate 7 mutations. High-depth queries help SQLRight trigger more code coverage. In NoREC, besides the seed-triggered code (depth 0), 14.9% of new coverage is triggered by queries with 4-7 depth, and 17.0% is triggered by queries w",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 56111,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M68",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "In NoREC, besides the seed-triggered code (depth 0), 14.",
+      "context_before": "rage: 40.1K branches with NoREC and 39.2K branches with TLP.SQLRight dropdetects 35.2K branches using NoREC and 35.0K branches using TLP, which are 13.7% and 11.9% fewer than SQLRight. SQLRight randand SQLRight savetrigger even fewer branches than SQLRight drop. The result shows the benefit of coveragebased guidance in generating more diverse queries. Mutation Depth. Table 3 shows that SQLRight maintains deeper mutation chains, where SQLRight can accumulate 21 mutations (column “Max Depth”) while others at most accumulate 7 mutations. High-depth queries help SQLRight trigger more code coverage.",
+      "context_after": "9% of new coverage is triggered by queries with 4-7 depth, and 17.0% is triggered by queries with 8 or more mutations; in TLP, the contributions of high-depth queries are 18.4% and 11.4%, respectively. SQLRight droponly mutates seed inputs to produce 1-depth queries. Although it gets more coverage on depth 1, SQLRight finally wins thanks to high-depth queries. SQLRight randand SQLRight savecan accumulate some mutations. However, without the coverage-based guidance, the accumulation is slow and falls behind SQLRight. Accumulated mutations also help detect more bugs. Column 4318 31st USENIX Secur",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 56596,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M69",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "0% is triggered by queries with 8 or more mutations; in TLP, the contributions of high-depth queries are 18.",
+      "context_before": "hes using TLP, which are 13.7% and 11.9% fewer than SQLRight. SQLRight randand SQLRight savetrigger even fewer branches than SQLRight drop. The result shows the benefit of coveragebased guidance in generating more diverse queries. Mutation Depth. Table 3 shows that SQLRight maintains deeper mutation chains, where SQLRight can accumulate 21 mutations (column “Max Depth”) while others at most accumulate 7 mutations. High-depth queries help SQLRight trigger more code coverage. In NoREC, besides the seed-triggered code (depth 0), 14.9% of new coverage is triggered by queries with 4-7 depth, and 17.",
+      "context_after": "4% and 11.4%, respectively. SQLRight droponly mutates seed inputs to produce 1-depth queries. Although it gets more coverage on depth 1, SQLRight finally wins thanks to high-depth queries. SQLRight randand SQLRight savecan accumulate some mutations. However, without the coverage-based guidance, the accumulation is slow and falls behind SQLRight. Accumulated mutations also help detect more bugs. Column 4318 31st USENIX Security Symposium USENIX Association 0 4 8 12 16 20 240123456Unique Bug (a)SQLite logical bugs 0 4 8 12 16 20 2401234Unique Bug (b)MySQL logical bugs SQLRight SQLRight−deter SQL",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 11,
+      "char_offset": 56719,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M70",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "5Coverage (k) (c)SQLite code coverage 0 4 8 12 16 20 24100120140160180200220Coverage (k) (d)MySQL code coverage 0 4 8 12 16 20 24020406080100120Coverage (k) (e)PostgreSQL code coverage 0 4 8 12 16 20 24203040506070Query Validity (%) (f)SQLite query validity 0 4 8 12 16 20 24010203040Query Validity (%) (g)MySQL query validity 0 4 8 12 16 20 24051015202530Query Validity (%) (h)PostgreSQL query validity 0 4 8 12 16 20 24105106valid statements per hour (i)SQLite valid stmts / hr 0 4 8 12 16 20 24100101102103104105valid statements per hour (j)MySQL valid stmts / hr 0 4 8 12 16 20 24100101102103104105valid statements per hour (k)PostgreSQL valid stmts / hr Figure 7: Unit Tests for SQLRight validity components ( NoREC ).",
+      "context_before": "ets more coverage on depth 1, SQLRight finally wins thanks to high-depth queries. SQLRight randand SQLRight savecan accumulate some mutations. However, without the coverage-based guidance, the accumulation is slow and falls behind SQLRight. Accumulated mutations also help detect more bugs. Column 4318 31st USENIX Security Symposium USENIX Association 0 4 8 12 16 20 240123456Unique Bug (a)SQLite logical bugs 0 4 8 12 16 20 2401234Unique Bug (b)MySQL logical bugs SQLRight SQLRight−deter SQLRight−ctx−valid SQLRight−db−par&ctx−valid Squirrel+ oracle 0 4 8 12 16 20 2425.027.530.032.535.037.540.042.",
+      "context_after": "a and b show the number of unique logical bugs. c-k show code coverage, query validity and valid statements over time for each instance. We run each instance for 24 hours, repeat 5 times and report the average results.01 CREATE TABLE v0 (v1 TEXT); 02 INSERT INTO v0 VALUES ( '0'); 03 CREATE VIEW v2 (v3) AS SELECT v1 FROM v0 04 UNION ALL SELECT v1=10 FROM v0; 05 SELECT * FROM v2 NATURAL JOIN v2; 06 SELECT COUNT(*) FROM v2 NATURAL JOIN v2; 07 -- output of 1st SELECT: 0\\n0\\n0\\n0 08 -- output of 2nd SELECT: 2 Listing 8: A false positive case related to view affinity (SQLite ). The affinity of v3is",
+      "section": "5.1 DBMS Logical Bugs",
+      "page": 12,
+      "char_offset": 57534,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M71",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Figure 7 ( NoREC ) and Figure 9 ( TLP) in Appendix show the results.",
+      "context_before": "1, including cooperative mutation, DBMS-specific parser, context-based instantiation, and non-determinism mitigation. SQLRight -deter disables the non-determinism mitigation techniques. SQLRight -ctx-valid disables the contextbased instantiation and uses the dependency graph-based validation from Squirrel. Based on SQLRight -ctx-valid, SQLRight -db-par&ctx-valid further reuses the SQL parser of Squirrel .Squirrel +oracle does not adopt any validityimprovement techniques. We conduct unit tests on three DBMSs with two oracles. We evaluate each setup with five instances and run them for 24 hours.",
+      "context_after": "Unique Logical Bugs. SQLRight triggers the most bugs,shown in Figure 7a b and Figure 9a b, including 4 bugs in SQLite using NoREC, 2 bugs in SQLite using TLP, 2 bugs in MySQL using NoREC, and 1 bug in MySQL using TLP. SQLRight -db-par&ctx-valid andSquirrel +oracle merely find 1 bug using NoREC. Other settings find no bugs. Code Coverage. According to Figure 7c d e and Figure 9c d e,SQLRight dominates the code coverage number across all unit tests. The results show that the validity techniques help generate more diverse queries and explore more code. Among all validity-improvement techniques, t",
+      "section": "5.4 Contribution of Validity",
+      "page": 12,
+      "char_offset": 60958,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M72",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "SQLRight triggers the most bugs,shown in Figure 7a b and Figure 9a b, including 4 bugs in SQLite using NoREC, 2 bugs in SQLite using TLP, 2 bugs in MySQL using NoREC, and 1 bug in MySQL using TLP.",
+      "context_before": "non-determinism mitigation. SQLRight -deter disables the non-determinism mitigation techniques. SQLRight -ctx-valid disables the contextbased instantiation and uses the dependency graph-based validation from Squirrel. Based on SQLRight -ctx-valid, SQLRight -db-par&ctx-valid further reuses the SQL parser of Squirrel .Squirrel +oracle does not adopt any validityimprovement techniques. We conduct unit tests on three DBMSs with two oracles. We evaluate each setup with five instances and run them for 24 hours. Figure 7 ( NoREC ) and Figure 9 ( TLP) in Appendix show the results. Unique Logical Bugs.",
+      "context_after": "SQLRight -db-par&ctx-valid andSquirrel +oracle merely find 1 bug using NoREC. Other settings find no bugs. Code Coverage. According to Figure 7c d e and Figure 9c d e,SQLRight dominates the code coverage number across all unit tests. The results show that the validity techniques help generate more diverse queries and explore more code. Among all validity-improvement techniques, the per-DBMS parser contributes the most to code coverage. Because the parser understands different SQL dialects, it can explore custom features specific to each DBMS. These features are not reachable through Squirrel,",
+      "section": "5.4 Contribution of Validity",
+      "page": 12,
+      "char_offset": 61048,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M73",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "SQLRight -db-par&ctx-valid andSquirrel +oracle merely find 1 bug using NoREC.",
+      "context_before": "dation from Squirrel. Based on SQLRight -ctx-valid, SQLRight -db-par&ctx-valid further reuses the SQL parser of Squirrel .Squirrel +oracle does not adopt any validityimprovement techniques. We conduct unit tests on three DBMSs with two oracles. We evaluate each setup with five instances and run them for 24 hours. Figure 7 ( NoREC ) and Figure 9 ( TLP) in Appendix show the results. Unique Logical Bugs. SQLRight triggers the most bugs,shown in Figure 7a b and Figure 9a b, including 4 bugs in SQLite using NoREC, 2 bugs in SQLite using TLP, 2 bugs in MySQL using NoREC, and 1 bug in MySQL using TLP.",
+      "context_after": "Other settings find no bugs. Code Coverage. According to Figure 7c d e and Figure 9c d e,SQLRight dominates the code coverage number across all unit tests. The results show that the validity techniques help generate more diverse queries and explore more code. Among all validity-improvement techniques, the per-DBMS parser contributes the most to code coverage. Because the parser understands different SQL dialects, it can explore custom features specific to each DBMS. These features are not reachable through Squirrel, leading to the significant coverage gap between SQLRight -ctx-valid and SQLRig",
+      "section": "5.4 Contribution of Validity",
+      "page": 12,
+      "char_offset": 61244,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M74",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "First, Squirrel parsers do not support custom functions in SELECT, including function COUNT, which is necessary to generate NoREC -compatible queries.",
+      "context_before": "elp generate more diverse queries and explore more code. Among all validity-improvement techniques, the per-DBMS parser contributes the most to code coverage. Because the parser understands different SQL dialects, it can explore custom features specific to each DBMS. These features are not reachable through Squirrel, leading to the significant coverage gap between SQLRight -ctx-valid and SQLRight -db-par&ctx-valid. Query Validity. The per-DBMS parser again plays the most important role in query validity. When testing PostgreSQL andMySQL, parsers from Squirrel cannot generate any useful queries.",
+      "context_after": "Second, these parsers do not support UNION ALL, which is the key component to implement compatible queries for TLP. Among three DBMSs, SQLRight has the lowest validity for PostgreSQL. Since USENIX Association 31st USENIX Security Symposium 4319 DBMS Oracle FPs Main Reason SQLite NoREC SQLRight: 19 VIEW affinity, subquery ordering SQLRight -deter :42 randomblob andjulianday() func SQLite TLP SQLRight: 8 subquery ordering SQLRight -deter: 53 randomblob func MySQL NoREC SQLRight: 0SQLRight -deter :18 rand func, auto insert TRIGGERs MySQL TLP SQLRight: 10 GROUP BY NULL SQLRight -deter: 64 rand fu",
+      "section": "5.4 Contribution of Validity",
+      "page": 12,
+      "char_offset": 62126,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M75",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Second, these parsers do not support UNION ALL, which is the key component to implement compatible queries for TLP.",
+      "context_before": "verage. Because the parser understands different SQL dialects, it can explore custom features specific to each DBMS. These features are not reachable through Squirrel, leading to the significant coverage gap between SQLRight -ctx-valid and SQLRight -db-par&ctx-valid. Query Validity. The per-DBMS parser again plays the most important role in query validity. When testing PostgreSQL andMySQL, parsers from Squirrel cannot generate any useful queries. First, Squirrel parsers do not support custom functions in SELECT, including function COUNT, which is necessary to generate NoREC -compatible queries.",
+      "context_after": "Among three DBMSs, SQLRight has the lowest validity for PostgreSQL. Since USENIX Association 31st USENIX Security Symposium 4319 DBMS Oracle FPs Main Reason SQLite NoREC SQLRight: 19 VIEW affinity, subquery ordering SQLRight -deter :42 randomblob andjulianday() func SQLite TLP SQLRight: 8 subquery ordering SQLRight -deter: 53 randomblob func MySQL NoREC SQLRight: 0SQLRight -deter :18 rand func, auto insert TRIGGERs MySQL TLP SQLRight: 10 GROUP BY NULL SQLRight -deter: 64 rand func, auto insert TRIGGERs Table 4: False Positives by Non-deterministic Behaviors. PostgreSQL enforces the most rigor",
+      "section": "5.4 Contribution of Validity",
+      "page": 12,
+      "char_offset": 62277,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M76",
+      "found_by": "technique",
+      "surface": "NoREC",
+      "technique": "norec",
+      "sentence": "Since USENIX Association 31st USENIX Security Symposium 4319 DBMS Oracle FPs Main Reason SQLite NoREC SQLRight: 19 VIEW affinity, subquery ordering SQLRight -deter :42 randomblob andjulianday() func SQLite TLP SQLRight: 8 subquery ordering SQLRight -deter: 53 randomblob func MySQL NoREC SQLRight: 0SQLRight -deter :18 rand func, auto insert TRIGGERs MySQL TLP SQLRight: 10 GROUP BY NULL SQLRight -deter: 64 rand func, auto insert TRIGGERs Table 4: False Positives by Non-deterministic Behaviors.",
+      "context_before": "ignificant coverage gap between SQLRight -ctx-valid and SQLRight -db-par&ctx-valid. Query Validity. The per-DBMS parser again plays the most important role in query validity. When testing PostgreSQL andMySQL, parsers from Squirrel cannot generate any useful queries. First, Squirrel parsers do not support custom functions in SELECT, including function COUNT, which is necessary to generate NoREC -compatible queries. Second, these parsers do not support UNION ALL, which is the key component to implement compatible queries for TLP. Among three DBMSs, SQLRight has the lowest validity for PostgreSQL.",
+      "context_after": "PostgreSQL enforces the most rigorous syntax and semantic rules, it is hard to generate valid SQL queries [43 –45, 71]. False Positives due to Non-determinism. Table 4 shows the false positive numbers after bisecting, and Figure 7 and Figure 9 present the code coverage and validity of SQLRight -deter. SQLRight -deter shares similar code coverage and validity with SQLRight, but it produces 95 false positives in SQLite and 82 false positives in MySQL. Most false positives are due to random functions, like randomblob ,julianday, and so on. By disabling all non-deterministic behaviors, SQLRight s",
+      "section": "5.4 Contribution of Validity",
+      "page": 12,
+      "char_offset": 62461,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M77",
+      "found_by": "citation_marker",
+      "surface": "[43 –45, 71]",
+      "technique": "norec",
+      "sentence": "PostgreSQL enforces the most rigorous syntax and semantic rules, it is hard to generate valid SQL queries [43 –45, 71].",
+      "context_before": "plement compatible queries for TLP. Among three DBMSs, SQLRight has the lowest validity for PostgreSQL. Since USENIX Association 31st USENIX Security Symposium 4319 DBMS Oracle FPs Main Reason SQLite NoREC SQLRight: 19 VIEW affinity, subquery ordering SQLRight -deter :42 randomblob andjulianday() func SQLite TLP SQLRight: 8 subquery ordering SQLRight -deter: 53 randomblob func MySQL NoREC SQLRight: 0SQLRight -deter :18 rand func, auto insert TRIGGERs MySQL TLP SQLRight: 10 GROUP BY NULL SQLRight -deter: 64 rand func, auto insert TRIGGERs Table 4: False Positives by Non-deterministic Behaviors.",
+      "context_after": "False Positives due to Non-determinism. Table 4 shows the false positive numbers after bisecting, and Figure 7 and Figure 9 present the code coverage and validity of SQLRight -deter. SQLRight -deter shares similar code coverage and validity with SQLRight, but it produces 95 false positives in SQLite and 82 false positives in MySQL. Most false positives are due to random functions, like randomblob ,julianday, and so on. By disabling all non-deterministic behaviors, SQLRight still introduces a small number of false positives. They are mainly due to the special semantics in each DBMS, which we w",
+      "section": "5.4 Contribution of Validity",
+      "page": 13,
+      "char_offset": 62958,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M78",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "We find that the affinity issue is commonly reported by third-party tools as “bugs”, including SQLancer [30].",
+      "context_before": "fined affinity. Specifically, SQLite assigns each column an affinity (similar to data type) and uses different comparison algorithms for each affinity. In this example, the affinity of v2.v3 is indeterminate due to different affinities from two sources: SELECT v1 FROM v0 returns a column with affinity TEXT, while SELECT v1=10 FROM v0 returns a column with affinity NONE. Therefore, in the last two SELECT s,SQLite is free to choose any affinity to compare v3. In line 5, SQLite uses affinity TEXT and returns four rows, while in line 6, SQLite chooses affinity BLOB and only finds two matched rows.",
+      "context_after": "Although SQLite documents have clearly explained this problem [50], we suspect that DBMS administrators may still miss it and create ambiguous queries.FP2: Ordering in SQLite Subqueries. Listing 9 shows one false positive due to the undefined ordering in subqueries. This query first creates table v0with columns c1andc2and inserts rows (NULL,10) and (NULL,NULL). Then, it creates view v3to access particular rows in v0. If a row in v0satisfies condition c1 IS NULL, the minimum c1will be assigned to c4, andc2will be assigned to c5. Both SELECT statements count c5, but they return different result",
+      "section": "5.4 Contribution of Validity",
+      "page": 13,
+      "char_offset": 65224,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M79",
+      "found_by": "citation_marker",
+      "surface": "[43 –45, 49, 61, 68]",
+      "technique": "norec",
+      "sentence": "Differential testing [29] is commonly used to detect logical bugs from DBMS systems [43 –45, 49, 61, 68].",
+      "context_before": "unctions such as WHERE ,ROWID and IN.TCL Test is the built-in test tool for SQLite. Its code is available in the SQLite GitHub repository, and it contains 1,272 test files and more than 46,000 unique test cases. These TCL Test scripts not only contain previously discovered bugs but also cover many SQLite -specific functionalities such as ROWID. We plan to introduce more high-quality and diverse seed inputs to SQLRight in our future work. 7 Related Work In this section, we discuss the recent testing approaches that are related to SQLRight, focusing on DBMS testing. Detecting Logic Bugs in DBMSs.",
+      "context_after": "One direction is to run one query with different DBMS systems [16, 49, 52] and check the result consistency. Another direction is to run one query with different settings of one DBMS, like versions or optimization levels [67, 68]. SQLancer proposes the third direction, which constructs functionally equivalent queries to test one DBMS [43 –45]. This method requires a deep understanding of SQL and DBMS [30]. SQLRight adopts the third method to generate equivalent queries. It also uses random mutation to diversify the query and relies on coverage-based guidance to help explore the program states",
+      "section": "5.4 Contribution of Validity",
+      "page": 14,
+      "char_offset": 69936,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE), 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M80",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer proposes the third direction, which constructs functionally equivalent queries to test one DBMS [43 –45].",
+      "context_before": "s ROWID. We plan to introduce more high-quality and diverse seed inputs to SQLRight in our future work. 7 Related Work In this section, we discuss the recent testing approaches that are related to SQLRight, focusing on DBMS testing. Detecting Logic Bugs in DBMSs. Differential testing [29] is commonly used to detect logical bugs from DBMS systems [43 –45, 49, 61, 68]. One direction is to run one query with different DBMS systems [16, 49, 52] and check the result consistency. Another direction is to run one query with different settings of one DBMS, like versions or optimization levels [67, 68].",
+      "context_after": "This method requires a deep understanding of SQL and DBMS [30]. SQLRight adopts the third method to generate equivalent queries. It also uses random mutation to diversify the query and relies on coverage-based guidance to help explore the program states. Our evaluation demonstrates the benefits of code coverage for finding logical bugs. Detecting Performance Bugs in DBMS. Performance is an important metric of DBMS systems, and many test efforts aim to pinpoint the performance bottleneck [25]. BmPad [40] estimates execution times for a set of queries and checks whether real execution times are",
+      "section": "5.4 Contribution of Validity",
+      "page": 14,
+      "char_offset": 70273,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M81",
+      "found_by": "citation_marker",
+      "surface": "[30]",
+      "technique": null,
+      "sentence": "This method requires a deep understanding of SQL and DBMS [30].",
+      "context_before": "ork In this section, we discuss the recent testing approaches that are related to SQLRight, focusing on DBMS testing. Detecting Logic Bugs in DBMSs. Differential testing [29] is commonly used to detect logical bugs from DBMS systems [43 –45, 49, 61, 68]. One direction is to run one query with different DBMS systems [16, 49, 52] and check the result consistency. Another direction is to run one query with different settings of one DBMS, like versions or optimization levels [67, 68]. SQLancer proposes the third direction, which constructs functionally equivalent queries to test one DBMS [43 –45].",
+      "context_after": "SQLRight adopts the third method to generate equivalent queries. It also uses random mutation to diversify the query and relies on coverage-based guidance to help explore the program states. Our evaluation demonstrates the benefits of code coverage for finding logical bugs. Detecting Performance Bugs in DBMS. Performance is an important metric of DBMS systems, and many test efforts aim to pinpoint the performance bottleneck [25]. BmPad [40] estimates execution times for a set of queries and checks whether real execution times are expected. AMOEBA [26] constructs function-equivalent queries an",
+      "section": "5.4 Contribution of Validity",
+      "page": 14,
+      "char_offset": 70388,
+      "cited_reference": {
+        "number": 30,
+        "text": "Manuel Rigger. SQLancer. https://github.com/sqlancer/ sqlancer. (visited in June 2021).",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M82",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "SQLsmith [58] is highly customized and effective for PostgreSQL [71].",
+      "context_before": "ance bugs. SQLRight focuses on logical bugs,which usually have a strict policy on result consistency. Most reported bugs have been fixed immediately after our reports. Detecting Crash Bugs in DBMS. Most testing efforts have been spent on detecting program crashes through two methods, either generation-based method or mutation-based. The generation-based method takes well-defined rules to synthesize valid DBMS queries [33, 58, 59]. Since generating a completely valid DBMS query is NP-complete [29], most generators focus on syntax validity and take practical ways to improve semantic correctness.",
+      "context_after": "It generates constrained queries based on the understanding of the underlying database schema. Several works reduce the query generation problem to the satisfying constraints [1, 31] and use SAT solvers to provide valid queries [2]. Differently, Chandra et al. [7] propose to generate a proper database to boost the DBMS testing. SQLRight is different from these mutators in two ways. First, it does not require any predefined database, but creates one for each query. Second, it utilizes code coverage to highlight promising queries for testing. Several fuzzers have shown their capability to test",
+      "section": "5.4 Contribution of Validity",
+      "page": 14,
+      "char_offset": 72162,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M83",
+      "found_by": "citation_marker",
+      "surface": "[71]",
+      "technique": null,
+      "sentence": "The most recent work Squirrel translates SQL strings to an IR-based representation, and relies on the IR type to increase the validity of query generation [71].",
+      "context_before": "st, it does not require any predefined database, but creates one for each query. Second, it utilizes code coverage to highlight promising queries for testing. Several fuzzers have shown their capability to test DBMS systems [3, 4, 8, 9, 24, 27, 36, 53, 69, 70]. Grimoire utilizes grammar-like combinations to synthesize highly structured inputs [6]. Bati et al. [5] proposed an engine that inserts or removes grammar components from existing SQL statements. However, due to the strict grammar and semantic requirements, most tools cannot explore deep logic of DBMS, like query planning and execution.",
+      "context_after": "SQLRight adopts the type-guided mutation ofSquirrel, and takes one step further to cooperate with DBMS oracles to produce high-quality, oracle-compatible SQL statements. More importantly, SQLRight focuses on detecting logical bugs other than crashes and assertion failures. 8 Conclusion We design the first tool, SQLRight, that combines code coverage, validity-oriented mutations and oracles to find logical bugs for DBMS systems. SQLRight contains a set of general APIs so that users can easily use existing fuzzers to test DBMSs and develop new oracles. We also improve the query validity to boost",
+      "section": "5.4 Contribution of Validity",
+      "page": 14,
+      "char_offset": 73221,
+      "cited_reference": {
+        "number": 71,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Proceedings of the 27th ACM Conference on Computer and Communications Security (CCS), Orlando, USA, November 2020. USENIX Associ",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": {
+    "url": "https://github.com/PSU-Security-Universe/sqlright-artifact",
+    "markers": [
+      "sqlancer_source_content_match"
+    ],
+    "evidence": [
+      {
+        "source_url": "https://github.com/PSU-Security-Universe/sqlright-artifact",
+        "source_type": "github_repository",
+        "excerpt": "Artifact Evaluation code for USENIX 2022 paper: Detecting Logical Bugs of DBMS with Coverage-based Guidance\n# sqlright-artifact: The code, analysis scripts and results for USENIX 2022 Artifact Evaluation\n\n<a href=\"Paper/paper_no_names.pdf\"><img src=\"Paper/paper_no_names.png\" align=\"right\" width=\"250\"></a>",
+        "note": "Repository names this paper.",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z",
+        "excerpt_is_verbatim": true
+      },
+      {
+        "source_url": "https://github.com/PSU-Security-Universe/sqlright-artifact/blob/main/MySQL/docker/sqlancer/sqlancer/src/sqlancer/Randomly.java",
+        "source_type": "github_code",
+        "excerpt": "private static StringGenerationStrategy stringGenerationStrategy = StringGenerationStrategy.SOPHISTICATED;",
+        "excerpt_is_verbatim": true,
+        "note": "MySQL/docker/sqlancer/sqlancer/src/sqlancer/Randomly.java is SQLancer's Randomly.java (8 of 8 identifiers match: StringGenerationStrategy, SOPHISTICATED, cachedLongs, cachedStrings, cachedDoubles).",
+        "content_sha256": "sha256:cb0a0e87e414094ea53bc63a6ad1aefbb64e7919425edd6a94de532b9cd1daa6",
+        "retrieved_at": "2026-09-10T15:15:32Z",
+        "first_seen": "2026-09-10T15:15:32Z",
+        "last_verified": "2026-09-10T15:15:32Z"
+      }
+    ]
+  },
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M8",
+        "M33",
+        "M44",
+        "M58"
+      ],
+      "describes_as_state_of_the_art": [
+        "M33",
+        "M55"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T01:51:46Z",
+    "is_model_written": true,
+    "summary": "SQLRight adds coverage-based guidance to logic-bug detection in DBMSs, arguing that oracle-driven tools generate queries from fixed rules and so explore a limited set of program states, while coverage-guided fuzzers find crashes but have no oracle for wrong results. It combines a coverage-guided mutation engine with validity-preserving mutation that keeps the clauses an oracle needs, and adds non-determinism mitigation. Over 72 hours it reported 12 unique logic bugs where the tools it compares against found one and none respectively.",
+    "narrative": "SQLRight is built on SQLancer and Squirrel, and two of its four oracles -- NoREC and TLP -- are ported from SQLancer directly. It then measures itself against SQLancer as the state-of-the-art oracle-based tool, reporting more bugs but acknowledging SQLancer's much higher query validity.",
+    "roles": {
+      "M3": "background",
+      "M4": "motivation",
+      "M8": "baseline",
+      "M9": "result_comparison",
+      "M11": "definition",
+      "M18": "reuse_component",
+      "M19": "definition",
+      "M20": "reuse_component",
+      "M23": "definition",
+      "M28": "reuse_implementation",
+      "M29": "reuse_implementation",
+      "M33": "state_of_the_art",
+      "M34": "baseline",
+      "M35": "background",
+      "M38": "result_comparison",
+      "M39": "baseline",
+      "M40": "result_comparison",
+      "M41": "result_comparison",
+      "M42": "definition",
+      "M44": "baseline",
+      "M45": "baseline",
+      "M48": "result_comparison",
+      "M49": "result_comparison",
+      "M53": "result_comparison",
+      "M54": "result_comparison",
+      "M55": "background",
+      "M56": "result_comparison",
+      "M57": "result_comparison",
+      "M58": "result_comparison",
+      "M59": "result_comparison",
+      "M78": "incidental",
+      "M80": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "yes",
+        "mention_ids": [
+          "M28",
+          "M29"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M28",
+            "sentence": "Our implementation is based on Squirrel [71] and SQLancer [30].",
+            "section": "3.1.4 Non-determinism Mitigation",
+            "page": 8
+          },
+          {
+            "mention_id": "M29",
+            "sentence": "Currently, SQLRight supports four oracles, including NoREC and TLPported from SQLancer [43, 44], and Index and Rowid we propose in this paper.",
+            "section": "3.1.4 Non-determinism Mitigation",
+            "page": 8
+          }
+        ],
+        "reasoning": "M28 states the implementation is based on Squirrel and SQLancer, and M29 that two of its four oracles, NoREC and TLP, are ported from SQLancer. The artifact carries SQLancer source content, which agrees.",
+        "reuse_kind": "implementation"
+      },
+      "extends_technique": {
+        "value": "yes",
+        "mention_ids": [
+          "M18",
+          "M20",
+          "M29"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M18",
+            "sentence": "To support the NoREC oracle, we fix SELECT_statement ,FROM and WHERE of the statement and leave others for the mutation engine to change.",
+            "section": "3.1.1 Cooperative Mutation",
+            "page": 6
+          },
+          {
+            "mention_id": "M20",
+            "sentence": "As a result, the crucial components for oracle NoREC are preserved.",
+            "section": "3.1.1 Cooperative Mutation",
+            "page": 6
+          },
+          {
+            "mention_id": "M29",
+            "sentence": "Currently, SQLRight supports four oracles, including NoREC and TLPported from SQLancer [43, 44], and Index and Rowid we propose in this paper.",
+            "section": "3.1.4 Non-determinism Mitigation",
+            "page": 8
+          }
+        ],
+        "reasoning": "Porting NoREC and TLP is reuse; what makes it extension is that the mutation engine is adapted to preserve the clauses those oracles require (M18, M20), so the oracles run under coverage guidance rather than rule-based generation. The paper proposes two further oracles of its own alongside them.",
+        "techniques": [
+          "norec",
+          "tlp"
+        ]
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M8",
+          "M9",
+          "M44",
+          "M48",
+          "M58"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M8",
+            "sentence": "We also compare our system with the stateof-the-art tools, including SQLancer (using oracles to detect logical bugs) and Squirrel (using code coverage to detect crashes and assertion failures).",
+            "section": "1 Introduction",
+            "page": 3
+          },
+          {
+            "mention_id": "M9",
+            "sentence": "After testing for 72 hours, SQLRight reports 12 unique logical bugs, Squirrel detects one bug, and SQLancer does not find any bug.",
+            "section": "1 Introduction",
+            "page": 3
+          },
+          {
+            "mention_id": "M44",
+            "sentence": "Therefore, we use NoREC and TLPto compare SQLRight ,SQLancer, and Squirrel +oracle.",
+            "section": "5.1 DBMS Logical Bugs",
+            "page": 10
+          },
+          {
+            "mention_id": "M48",
+            "sentence": "Squirrel +oracle merely found 1 bug, which is from SQLite using NoREC; SQLancer did not find any logical bug.",
+            "section": "5.1 DBMS Logical Bugs",
+            "page": 10
+          },
+          {
+            "mention_id": "M58",
+            "sentence": "Overall, SQLRight can find more logical bugs than SQLancer andSquirrel +oracle.",
+            "section": "5.1 DBMS Logical Bugs",
+            "page": 11
+          }
+        ],
+        "reasoning": "SQLancer is one of two tools SQLRight is measured against over 72 hours, using NoREC and TLP as the shared oracles, with bug counts and coverage reported for each.",
+        "techniques": [
+          "norec",
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M8",
+          "M33"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M8",
+            "sentence": "We also compare our system with the stateof-the-art tools, including SQLancer (using oracles to detect logical bugs) and Squirrel (using code coverage to detect crashes and assertion failures).",
+            "section": "1 Introduction",
+            "page": 3
+          },
+          {
+            "mention_id": "M33",
+            "sentence": "For Q2, we compare SQLRight with SQLancer and Squirrel, the state-of-the-art bug-finding tools for DBMSs.",
+            "section": "3.1.4 Non-determinism Mitigation",
+            "page": 9
+          }
+        ],
+        "reasoning": "Both call SQLancer one of the state-of-the-art bug-finding tools for DBMSs."
+      }
+    },
+    "disagreements": [],
+    "unresolved": [
+      "Whether the ported oracles were modified beyond what the mutation engine required is not stated in the mentions."
+    ]
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.json b/_data/papers/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.json
new file mode 100644
index 0000000..c9a5e8b
--- /dev/null
+++ b/_data/papers/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.json
@@ -0,0 +1,583 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-08T14:33:34Z",
+  "paper": {
+    "id": "paper:s2:84c637ad3297e0d4e45a4f5245d0472a82a59268",
+    "title": "Demystifying and Checking Silent Semantic Violations in Large Distributed Systems",
+    "authors": [
+      "Chang Lou",
+      "Yuzhuo Jing",
+      "Peng Huang"
+    ],
+    "year": 2022,
+    "venue": "USENIX Symposium on Operating Systems Design and Implementation",
+    "doi": null,
+    "arxiv_id": null,
+    "s2_paper_id": "84c637ad3297e0d4e45a4f5245d0472a82a59268",
+    "url": "https://www.semanticscholar.org/paper/84c637ad3297e0d4e45a4f5245d0472a82a59268",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://www.semanticscholar.org/paper/84c637ad3297e0d4e45a4f5245d0472a82a59268",
+      "retrieved_at": "2026-09-08T14:33:34Z",
+      "chars": 93367,
+      "content_sha256": "sha256:412f4f908f92635fffe7a506a1632740c287fa34555732f98b71273b0977abbc"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 18,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1616
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 8882
+      },
+      {
+        "number": "2.1",
+        "title": "Definition",
+        "start": 8895
+      },
+      {
+        "number": "2.2",
+        "title": "An Example",
+        "start": 11200
+      },
+      {
+        "number": "3",
+        "title": "Study Methodology",
+        "start": 12776
+      },
+      {
+        "number": "4",
+        "title": "Are Silent Semantic Failures Rare?",
+        "start": 16435
+      },
+      {
+        "number": "5",
+        "title": "What Kind of Semantics Is Violated?",
+        "start": 21202
+      },
+      {
+        "number": "5.1",
+        "title": "Sources of Violated Semantics",
+        "start": 21240
+      },
+      {
+        "number": "5.2",
+        "title": "Categorizations of Violated Semantics",
+        "start": 24419
+      },
+      {
+        "number": "6",
+        "title": "Why Do Silent Semantic Failures Occur?",
+        "start": 31365
+      },
+      {
+        "number": "7",
+        "title": "How Are Semantic Failures Manifested?",
+        "start": 33598
+      },
+      {
+        "number": "8.1",
+        "title": "Testing",
+        "start": 37371
+      },
+      {
+        "number": "8.2",
+        "title": "Assertions",
+        "start": 39649
+      },
+      {
+        "number": "8.3",
+        "title": "Observability",
+        "start": 41357
+      },
+      {
+        "number": "9.1",
+        "title": "Design Overview and Workflow",
+        "start": 44174
+      },
+      {
+        "number": "9.2",
+        "title": "Instrumentation and Trace Generation",
+        "start": 48253
+      },
+      {
+        "number": "9.3",
+        "title": "Template-Driven Inference",
+        "start": 51534
+      },
+      {
+        "number": "9.4",
+        "title": "Rule Validation",
+        "start": 58950
+      },
+      {
+        "number": "9.5",
+        "title": "Runtime Checking",
+        "start": 61336
+      },
+      {
+        "number": "9.6",
+        "title": "Optimizations",
+        "start": 62381
+      },
+      {
+        "number": "9.7",
+        "title": "Implementation",
+        "start": 64114
+      },
+      {
+        "number": "9.8",
+        "title": "Limitations",
+        "start": 64468
+      },
+      {
+        "number": "10.1",
+        "title": "Generation Overview",
+        "start": 65244
+      },
+      {
+        "number": "10.2",
+        "title": "Checking Newer Violations",
+        "start": 66612
+      },
+      {
+        "number": "10.3",
+        "title": "Performance",
+        "start": 71889
+      },
+      {
+        "number": "10.4",
+        "title": "Runtime Overhead",
+        "start": 73178
+      },
+      {
+        "number": "10.5",
+        "title": "Rule Activation and False Positive",
+        "start": 74571
+      },
+      {
+        "number": "T",
+        "title": "Ramani, N. Govindraju, X. Li, Q. Lin, G. L. Shafriri, and",
+        "start": 88435
+      },
+      {
+        "number": "M",
+        "title": "Chintalapati. Predictive and adaptive failure mitigation to",
+        "start": 88496
+      },
+      {
+        "number": "M",
+        "title": "F. Kaashoek, and Z. Zhang. D3S: Debugging deployed",
+        "start": 89104
+      },
+      {
+        "number": "J",
+        "title": "Cutler, P. Enriquez, A. Fox, E. Kiciman, M. Merzbacher,",
+        "start": 91723
+      },
+      {
+        "number": "D",
+        "title": "Oppenheimer, N. Sastry, W. Tetzlaff, J. Traupman, and",
+        "start": 91782
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Apache ZooKeeper releases. https://zookeeper.apache.org/ releases.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Google cloud infrastructure incident #20013. https://status. cloud.google.com/incident/zall/20013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Google cloud storage incident #17005. https://status.cloud. google.com/incident/storage/17005.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "HBASE-11536: Puts of region location to meta may be out of order which causes inconsistent of region location. https: //issues.apache.org/jira/browse/HBASE-11536.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "HBase-17125: Inconsistent result when use filter to read data. https://issues.apache.org/jira/browse/HBASE-17125.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "HDFS-12217: HDFS snapshots doesn’t capture all open files when one of the open files is deleted. https://issues.apache. org/jira/browse/HDFS-12217.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "HDFS-14359: Inherited acl permissions masked when parent directory does not exist (mkdir -p). https://issues.apache. org/jira/browse/HDFS-14359.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "HDFS-9083: Replication violates block placement policy. https://issues.apache.org/jira/browse/HDFS-12070.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "KAFKA-2960: DelayedProduce may cause message loss during repeated leader change. https://issues.apache.org/ jira/browse/KAFKA-2960.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Lmax disruptor. https://lmax-exchange.github.io/ disruptor/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "MongoDB-12355: add \"invariant\" for invariant checking in server code. https://jira.mongodb.org/browse/ SERVER-12355.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "MongoDB-50971: Invariant failure, wt_notfound: item not found. https://jira.mongodb.org/browse/SERVER-50971.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "ZooKeeper-1208: Ephemeral node not removed after the client session is long gone. https://issues.apache.org/jira/ browse/ZOOKEEPER-1208.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "ZooKeeper-2774: Ephemeral znode will not be removed when sesstion timeout, if the system time of zookeeper node changes unexpectedly. https://issues.apache.org/jira/ browse/ZOOKEEPER-2774.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "ZooKeeper-3144: Potential ephemeral nodes inconsistent due to global session inconsistent with fuzzy snapshot. https: //issues.apache.org/jira/browse/ZOOKEEPER-3144.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "M. Alfatafta, B. Alkhatib, A. Alquraan, and S. Al-Kiswany. Toward a generic fault tolerance technique for partial network partitioning. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI ’20, pages 351–368. USENIX Association, Nov. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "A. Alquraan, H. Takruri, M. Alfatafta, and S. Al-Kiswany. An analysis of network-partitioning failures in cloud systems. In Proceedings of the 13th USENIX Conference on Operating Systems Design and Implementation, OSDI ’18, page 51–68, Carlsbad, CA, USA, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "J. Armstrong. Making reliable distributed systems in the presence of software errors. PhD thesis, The Royal Institute of Technology, Stockholm, Sweden, 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "R. H. Arpaci-Dusseau and A. C. Arpaci-Dusseau. Fail-stutter fault tolerance. In Proceedings of the Eighth Workshop on Hot Topics in Operating Systems, HotOS ’01, pages 33–. IEEE Computer Society, 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "M. Attariyan, M. Chow, and J. Flinn. X-ray: Automating root-cause diagnosis of performance anomalies in production software. In Proceedings of the 10th USENIX Conference on Operating Systems Design and Implementation, OSDI ’12, pages 307–320, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "M. Attariyan and J. Flinn. Automating configuration troubleshooting with dynamic information flow analysis. In Proceedings of the 9th USENIX Conference on Operating Systems Design and Implementation, OSDI ’10, pages 1–11, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "T. D. Chandra and S. Toueg. Unreliable failure detectors for reliable distributed systems. J. ACM, 43(2):225–267, Mar. 1996.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "K. M. Chandy and L. Lamport. Distributed snapshots: Determining global states of distributed systems. ACM Trans. Comput. Syst., 3(1):63–75, Feb. 1985.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "F. Chen and G. Ro¸ su. Mop: An efficient and generic runtime verification framework. In Proceedings of the 22nd Annual ACMSIGPLAN Conference on Object-Oriented Programming Systems and Applications, OOPSLA ’07, page 569–588, Montreal, Quebec, Canada, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "T. Do, M. Hao, T. Leesatapornwongsa, T. Patana-anake, and H. S. Gunawi. Limplock: Understanding the impact of limpware on scale-out cloud systems. In Proceedings of the 4th Annual Symposium on Cloud Computing, SOCC ’13, pages 14:1–14:14, Santa Clara, California, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "M. D. Ernst, J. H. Perkins, P. J. Guo, S. McCamant, C. Pacheco, M. S. Tschantz, and C. Xiao. The daikon system for dynamic detection of likely invariants. Sci. Comput. Program., 69(1–3):35–45, Dec. 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "A. S. Foundation. HDFS high availability using the quorum journal manager. https://hadoop.apache. org/docs/stable/hadoop-project-dist/hadoop-hdfs/ HDFSHighAvailabilityWithQJM.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "D. Fryer, K. Sun, R. Mahmood, T. Cheng, S. Benjamin, A. Goel, and A. D. Brown. Recon: Verifying file system consistency at runtime. In Proceedings of the 10th USENIX Conference on File and Storage Technologies, FAST ’12, pages 7–7, San Jose, CA, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "J. A. Goguen. Semantics of computation. In Proceedings of the Proceedings of the First International Symposium on Category Theory Applied to Computation and Control, page 151–163. Springer-Verlag, 1974.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "S. Grant, H. Cech, and I. Beschastnikh. Inferring and asserting distributed system invariants. In Proceedings of the 40th International Conference on Software Engineering, ICSE ’18, page 1149–1159, Gothenburg, Sweden, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "H. S. Gunawi, M. Hao, R. O. Suminto, A. Laksono, A. D. Satria, J. Adityatama, and K. J. Eliazar. Why does the cloud stop computing?: Lessons from hundreds of service outages. In Proceedings of the 7th ACM Symposium on Cloud Computing, SOCC ’16, pages 1–16, Santa Clara, CA, USA, Oct. 2016. USENIX Association 16th USENIX Symposium on Operating Systems Design and Implementation 105",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "H. S. Gunawi, R. O. Suminto, R. Sears, C. Golliher, S. Sundararaman, X. Lin, T. Emami, W. Sheng, N. Bidokhti, C. McCaffrey, G. Grider, P. M. Fields, K. Harms, R. B. Ross, A. Jacobson, R. Ricci, K. Webb, P. Alvaro, H. B. Runesha, M. Hao, and H. Li. Fail-slow at scale: Evidence of hardware performance faults in large production systems. In Proceedings of the 16th USENIX Conference on File and Storag",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "S. Hangal and M. S. Lam. Tracking down software bugs using automatic anomaly detection. In Proceedings of the 24th International Conference on Software Engineering, ICSE ’02, page 291–301, Orlando, Florida, 2002.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "K. Havelund and G. Ro¸ su. Runtime verification. Computer Aided Verification (CAV ’01) satellite workshop (ENTCS), 55, 2001.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Y. Hu, G. Huang, and P. Huang. Automated reasoning and detection of specious configuration in large systems with symbolic execution. In Proceedings of the 14th USENIX Symposium on Networked Systems Design and Implementation, OSDI ’20. USENIX, November 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "P. Huang, C. Guo, J. R. Lorch, L. Zhou, and Y. Dang. Capturing and enhancing in situ system observability for failure detection. In13th USENIX Symposium on Operating Systems Design and Implementation, OSDI ’18, pages 1–16, Carlsbad, CA, October 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "P. Huang, C. Guo, L. Zhou, J. R. Lorch, Y. Dang, M. Chintalapati, and R. Yao. Gray failure: The Achilles’ heel of cloud-scale systems. In Proceedings of the 16th Workshop on Hot Topics in Operating Systems, HotOS ’17. ACM, May 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "C. Killian, J. W. Anderson, R. Jhala, and A. Vahdat. Life, death, and the critical transition: Finding liveness bugs in systems code. In Proceedings of the 4th USENIX Conference on Networked Systems Design and Implementation, NSDI ’07, page 18, Cambridge, MA, 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "S. Kim, M. Xu, S. Kashyap, J. Yoon, W. Xu, and T. Kim. Finding semantic bugs in file systems with an extensible fuzzing framework. In Proceedings of the 27th ACM Symposium on Operating Systems Principles, SOSP ’19, page 147–161, Huntsville, Ontario, Canada, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "J. B. Leners, H. Wu, W.-L. Hung, M. K. Aguilera, and M. Walfish. Detecting failures in distributed systems with the Falcon spy network. In Proceedings of the 23rd ACM Symposium on Operating Systems Principles, SOSP ’11, pages 279–294, Cascais, Portugal, Oct. 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "S. Levy, R. Yao, Y. Wu, Y. Dang, P. Huang, Z. Mu, P. Zhao, T. Ramani, N. Govindraju, X. Li, Q. Lin, G. L. Shafriri, and M. Chintalapati. Predictive and adaptive failure mitigation to avert production cloud VM interruptions. In Proceedings of the 14th USENIX Symposium on Networked Systems Design and Implementation, OSDI ’20. USENIX, November 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Z. Li, L. Tan, X. Wang, S. Lu, Y. Zhou, and C. Zhai. Have things changed now? an empirical study of bug characteristics in modern open source software. In Proceedings of the 1st Workshop on Architectural and System Support for Improving Software Dependability, ASID ’06, page 25–33, San Jose, California, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "X. Liu, Z. Guo, X. Wang, F. Chen, X. Lian, J. Tang, M. Wu, M. F. Kaashoek, and Z. Zhang. D3S: Debugging deployed distributed systems. In Proceedings of the 5th USENIX Symposium on Networked Systems Design and Implementation, NSDI ’08, page 423–437. USENIX Association, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "X. Liu, W. Lin, A. Pan, and Z. Zhang. WiDS checker: Combating bugs in distributed systems. In Proceedings of the 4th USENIX Symposium on Networked Systems Design and Implementation, NSDI ’07. USENIX Association, Apr. 2007.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "C. Lou, P. Huang, and S. Smith. Comprehensive and efficient runtime checking in system software through watchdogs. In Proceedings of the 17th Workshop on Hot Topics in Operating Systems, HotOS ’19, Bertinoro, Italy, May 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "C. Lou, P. Huang, and S. Smith. Understanding, detecting and localizing partial failures in large system software. In 17th USENIX Symposium on Networked Systems Design and Implementation, NSDI ’20, pages 559–574. USENIX Association, Feb. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "C. Lou, Y. Jing, and P. Huang. A promise is not a promise— demystifying and checking silent semantic violations in large distributed systems. Technical report, Johns Hopkins University, July 2022. https://orderlab.io/paper/oathkeeper-tr. pdf, Accessed July 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "H. Lu, K. Veeraraghavan, P. Ajoux, J. Hunt, Y. J. Song, W. Tobagus, S. Kumar, and W. Lloyd. Existential consistency: Measuring and understanding consistency at facebook. In Proceedings of the 25th Symposium on Operating Systems Principles, SOSP ’15, page 295–310, Monterey, California, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "H. Ma, A. Goel, J.-B. Jeannin, M. Kapritsos, B. Kasikci, and K. A. Sakallah. I4: Incremental inference of inductive invariants for verification of distributed protocols. In Proceedings of the 27th ACM Symposium on Operating Systems Principles, SOSP ’19, page 370–384, Huntsville, Ontario, Canada, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "C. Min, S. Kashyap, B. Lee, C. Song, and T. Kim. Crosschecking semantic correctness: The case of finding file system bugs. In Proceedings of the 25th Symposium on Operating Systems Principles, SOSP ’15, page 361–377, Monterey, California, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "D. Oppenheimer, A. Ganapathi, and D. A. Patterson. Why do Internet services fail, and what can be done about it? In Proceedings of the 4th Conference on USENIX Symposium on Internet Technologies and Systems, USITS ’03, Seattle, WA, Mar. 2003.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "B. Panda, D. Srinivasan, H. Ke, K. Gupta, V. Khot, and H. S. Gunawi. IASO: A fail-slow detection and mitigation framework for distributed storage services. In 2019 USENIX Annual Technical Conference (USENIX ATC 19), pages 47–62. USENIX Association, July 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "D. Patterson, A. Brown, P. Broadwell, G. Candea, M. Chen, J. Cutler, P. Enriquez, A. Fox, E. Kiciman, M. Merzbacher, D. Oppenheimer, N. Sastry, W. Tetzlaff, J. Traupman, and N. Treuhaft. Recovery oriented computing (roc): Motivation, definition, techniques, and case studies. Technical Report UCB/CSD-02-1175, EECS Department, University of California, Berkeley, Mar 2002. 106 16th USENIX Symposium on",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "M. Rigger and Z. Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI ’20, pages 667– 682. USENIX Association, Nov. 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 55,
+      "text": "L. Tan, C. Liu, Z. Li, X. Wang, Y. Zhou, and C. Zhai. Bug characteristics in open source software. Empirical Softw. Engg., 19(6):1665–1705, Dec. 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "T. Xu, X. Jin, P. Huang, Y. Zhou, S. Lu, L. Jin, and S. Pasupathy. Early detection of configuration errors to reduce failure damage. In Proceedings of the The 12th USENIX Symposium on Operating Systems Design and Implementation, OSDI ’16, November 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "N. Yaseen, B. Arzani, R. Beckett, S. Ciraci, and V. Liu. Aragog: Scalable runtime verification of shardable networked systems. In14th USENIX Symposium on Operating Systems Design and Implementation, OSDI ’20, pages 701–718. USENIX Association, Nov. 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Z. Yin, D. Yuan, Y. Zhou, S. Pasupathy, and L. Bairavasundaram. How do fixes become bugs? In Proceedings of the 19th ACMSIGSOFT Symposium and the 13th European Conference on Foundations of Software Engineering, ESEC/FSE ’11, page 26–36, Szeged, Hungary, 2011. USENIX Association 16th USENIX Symposium on Operating Systems Design and Implementation 107",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 54,
+      "text": "M. Rigger and Z. Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI ’20, pages 667– 682. USENIX Association, Nov. 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[54]",
+      "technique": "pqs",
+      "sentence": "Several solutions are proposed to detect semantic bugs in file systems and DBMS, including cross-checking multiple file system implementations [50], fuzzing [39], and testing using pivoted query [54].",
+      "context_before": "traces from all test cases of a target system. Adding profile runs or a dynamic ban mechanism can further remove the false rules. 11 Related Work Semantic Bugs. Several studies [42,55,58] analyze the prevalence of semantic bugs in open-source server software. Our study analyzes semantic failures in distributed systems. Beyond the difference that we investigate distributed systems, the study of bugs is in general a complementary effort to the study of failures. The former focuses on analyzing the static code patterns, while the latter focuses on the dynamic manifestations and system misbehavior.",
+      "context_after": "Both cross-checking and fuzzing focus on finding bugs offline. Oathkeeper focuses on a complementary direction of inferring semantic rules for runtime checking. We hope our study can motivate future work to extend these solutions to detect semantic bugs in distributed systems. We observe some open challenges to cross-check distributed systems: distributed systems usually provide a wide variety of semantics that are less rigorously specified compared to file systems, which have well-defined semantics ( e.g., POSIX standard) and many implementations. Each distributed system has its unique semantics",
+      "section": "10.5 Rule Activation and False Positive",
+      "page": 15,
+      "char_offset": 76035,
+      "cited_reference": {
+        "number": 54,
+        "text": "M. Rigger and Z. Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI ’20, pages 667– 682. USENIX Association, Nov. 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "This paper studies silent semantic violations in large distributed systems -- failures where the system keeps running but breaks a property it promised. The authors analyse such violations across widely used systems, derive rules that capture the properties, and check them at runtime, finding violations that crash-oriented testing does not.",
+    "narrative": "One citation, in a survey of solutions for detecting semantic bugs in file systems and DBMSs, where pivoted query synthesis is named alongside cross-checking file system implementations and fuzzing.",
+    "roles": {
+      "M1": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json b/_data/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json
new file mode 100644
index 0000000..9c15061
--- /dev/null
+++ b/_data/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.json
@@ -0,0 +1,1368 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:33:14Z",
+  "paper": {
+    "id": "paper:s2:c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+    "title": "Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis",
+    "authors": [
+      "Zongyin Hao",
+      "Quanfeng Huang",
+      "Chengpeng Wang",
+      "Jianfeng Wang",
+      "Yushan Zhang",
+      "Rongxin Wu",
+      "Charles Zhang"
+    ],
+    "year": 2023,
+    "venue": "USENIX Annual Technical Conference",
+    "doi": null,
+    "arxiv_id": null,
+    "s2_paper_id": "c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+    "url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://www.semanticscholar.org/paper/c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b",
+      "retrieved_at": "2026-09-09T01:33:14Z",
+      "chars": 71229,
+      "content_sha256": "sha256:45417770f7ef7fc606a877626653a302b096cb19077eccc7dda52bc8394d2c8f"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 15,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2860
+      },
+      {
+        "number": "2",
+        "title": "Background",
+        "start": 10194
+      },
+      {
+        "number": "2.1",
+        "title": "Database Management Systems",
+        "start": 10477
+      },
+      {
+        "number": "2.2",
+        "title": "Logical Bugs in DBMSs",
+        "start": 12402
+      },
+      {
+        "number": "3",
+        "title": "Approximate Query Synthesis",
+        "start": 16470
+      },
+      {
+        "number": "3.1",
+        "title": "Approach Overview",
+        "start": 18208
+      },
+      {
+        "number": "3.2",
+        "title": "SQL Query Approximation",
+        "start": 19632
+      },
+      {
+        "number": "3.3",
+        "title": "Approximate Mutators",
+        "start": 23326
+      },
+      {
+        "number": "3.4",
+        "title": "Mutation-based Query Synthesis",
+        "start": 27914
+      },
+      {
+        "number": "3.5",
+        "title": "Summary",
+        "start": 33945
+      },
+      {
+        "number": "4",
+        "title": "Implementation",
+        "start": 35246
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 39111
+      },
+      {
+        "number": "5.1",
+        "title": "Experiment Setup",
+        "start": 39540
+      },
+      {
+        "number": "5.2",
+        "title": "Effectiveness of PINOLO",
+        "start": 43047
+      },
+      {
+        "number": "5.3",
+        "title": "Comparisons on Detecting Logical Bugs",
+        "start": 44903
+      },
+      {
+        "number": "5.4",
+        "title": "Impacts of The Seed Query Generation",
+        "start": 48240
+      },
+      {
+        "number": "5.5",
+        "title": "Discussion",
+        "start": 50105
+      },
+      {
+        "number": "6",
+        "title": "Threats to Validity",
+        "start": 55153
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 56001
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 59810
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "Kinan Dak Albab, Jonathan DiLorenzo, Stefan Heule, Ali Kheradmand, Steffen Smolka, Konstantin Weitz, Muhammad Timarzi, Jiaqi Gao, and Minlan Yu. Switchv: Automated sdn switch validation with p4 models. In Proceedings of the ACMSIGCOMM 2022 Conference, page 365–379, New York, NY, USA, 2022. ACM.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "Jialun Cao, Meiziniu Li, Yeting Li, Ming Wen, ShingChi Cheung, and Haiming Chen. Semmt: A semanticbased testing approach for machine translation systems. ACM Trans. Softw. Eng. Methodol., 31(2):1–36, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Donald D Chamberlin and Raymond F Boyce. Sequel: A structured english query language. In Proceedings of the 1974 ACMSIGFIDET workshop on Data description, access and control, pages 249–264, 1974.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Tsong Yueh Chen, S. C. Cheung, and Siu-Ming Yiu. Metamorphic testing: A new approach for generating next test cases. CoRR, abs/2002.12543, 2020. 356 2023 USENIX Annual Technical Conference USENIX Association",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Tsong Yueh Chen, Fei-Ching Kuo, Huai Liu, Pak-Lok Poon, Dave Towey, T. H. Tse, and Zhi Quan Zhou. Metamorphic testing: A review of challenges and opportunities. ACM Comput. Surv., 51(1):4:1–4:27, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Yuting Chen, Ting Su, and Zhendong Su. Deep differential testing of JVM implementations. In Proceedings of the 41st International Conference on Software Engineering, pages 1257–1268. IEEE / ACM, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Edgar F Codd. A relational model of data for large shared data banks. Communications of the ACM, 13(6):377–387, 1970.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Joeri de Ruiter and Erik Poll. Protocol state fuzzing of TLS implementations. In Proceedings of 24th USENIX Security Symposium, pages 193–206. USENIX Association, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "MariaDB Foundation. MariaDB Database. https: //mariadb.org/, 2022. [Online; accessed Dec-2022].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. Griffin: Grammar-free dbms fuzzing. InThe 37th IEEE/ACM International Conference on Automated Software Engineering, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Patrice Godefroid, Adam Kiezun, and Michael Y. Levin. Grammar-based whitebox fuzzing. In Proceedings of the ACMSIGPLAN 2008 Conference on Programming Language Design and Implementation, pages 206–215. ACM, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "William E Howden. Theoretical and empirical studies of program testing. IEEE Transactions on Software Engineering, SE-4(4):293–298, 1978.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Free Software Foundation Inc. BNU Bison. https: //www.gnu.org/software/bison/, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Google Inc. american fuzzy lopa security-oriented fuzzer. https://github.com/google/AFL, 2022. [Online; accessed Dec-2022].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Google Inc. Go 1.20 Release Notes. https://tip. golang.org/doc/go1.20, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Vu Le, Mehrdad Afshari, and Zhendong Su. Compiler validation via equivalence modulo inputs. In Proceedings of the 35th ACMSIGPLAN Conference on Programming Language Design and Implementation, pages 216–226. ACM, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "MaraiaDB. Bug #30249 of MaraiaDB. https://jira. mariadb.org/browse/MDEV-30249, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "William M. McKeeman. Differential testing for software. Digit. Tech. J., 10(1):100–107, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Michaël Mera. Mining constraints for grammar fuzzing. InProceedings of the 28th ACMSIGSOFT International Symposium on Software Testing and Analysis, pages 415– 418. ACM, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "MySQL. Bug #108937 of MySQL. https://bugs. mysql.com/bug.php?id=108937, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "MySQL. Bug #109407 of MySQL. https://bugs. mysql.com/bug.php?id=109407, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "MySQL. MySQL Database. https://www.mysql. com/, 2022. [Online; accessed Dec-2022].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "MySQL. The MySQL Test Framework. https://dev.mysql.com/doc/dev/mysql-server/ latest/PAGE_MYSQL_TEST_RUN.html, 2022. [Online; accessed Dec-2022].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "OceanBase. A deployer and package manager for OceanBase. https://github.com/oceanbase/obdeploy, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "OceanBase. Issue #1100 of OceanBase. https:// github.com/oceanbase/oceanbase/issues/1100, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "OceanBase. OceanBase Database. https://www. oceanbase.com/, 2022. [Online; accessed Dec-2022].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "PingCap. go randgen. https://github.com/ pingcap/go-randgen, 2022. [Online; accessed Dec2022].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "PingCap. Parser-A MySQL Compatible SQL Parser. https://github.com/pingcap/tidb/tree/ master/parser, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "PingCAP. TiDB Database. https://github.com/ pingcap/tidb, 2022. [Online; accessed Dec-2022].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "GNU Project. gcov——a Test Coverage Program. https://gcc.gnu.org/onlinedocs/gcc/ Gcov.html, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "qaqcatz. The Complete List of Instantiated Mutators in PINOLO. https://github.com/qaqcatz/ impomysql_binary/blob/main/mutators.md, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Mian Qin, Qing Zheng, Jason Lee, Bradley Settlemyer, Fei Wen, Narasimha Reddy, and Paul Gratz. Kvrangedb: Range queries for a hash-based key-value device. ACM Trans. Storage, jan 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "Manuel Rigger. sqlancer. https://github.com/ sqlancer/sqlancer, 2022. [Online; accessed Jan2023].",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 34,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 35,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA):211:1–211:30, 2020. USENIX Association 2023 USENIX Annual Technical Conference 357",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, pages 667–682. USENIX Association, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. Intramorphic testing: A new approach to the test oracle problem. In Proceedings of the 2022 ACMSIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software, pages 128–136. ACM, 2022.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 38,
+      "text": "Andreas Seltenreich. SQLsmith: A random SQL query generator. https://github.com/anse1/sqlsmith, 2022. [Online; accessed Dec-2022].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "Donald R Slutz. Massive stochastic testing of sql. In VLDB, volume 98, pages 618–622. Citeseer, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Thodoris Sotiropoulos, Stefanos Chaliasos, Vaggelis Atlidakis, Dimitris Mitropoulos, and Diomidis Spinellis. Data-oriented differential testing of object-relational mapping systems. In 43rd IEEE/ACM International Conference on Software Engineering, pages 1535–1547. IEEE, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "SQLite. How SQLite is Tested. https://www.sqlite. org/testing.html, 2022. [Online; accessed Dec2022].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "TiDB. Issue #40015 of TiDB. https://github. com/pingcap/tidb/issues/40015, 2022. [Online; accessed Jan-2023].",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Jianfeng Wang, Tamás Lévai, Zhuojin Li, Marcos A. M. Vieira, Ramesh Govindan, and Barath Raghavan. Quadrant: A cloud-deployable NF virtualization platform. In Proceedings of the 13th Symposium on Cloud Computing, page 493–509, New York, NY, USA, 2022. ACM.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Jianfeng Wang, Tamás Lévai, Zhuojin Li, Marcos Augusto M. Vieira, Ramesh Govindan, and Barath Raghavan. Galleon: Reshaping the square peg of NFV. CoRR, abs/2101.06466, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. Superion: grammar-aware greybox fuzzing. In Proceedings of the 41st International Conference on Software Engineering, pages 724–735. IEEE / ACM, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. Industry practice of coverage-guided enterprise-level DBMS fuzzing. In43rd IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice, pages 328–337. IEEE, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Shuai Wang and Zhendong Su. Metamorphic object insertion for testing object detection systems. In 35th IEEE/ACM International Conference on Automated Software Engineering, pages 1053–1065. IEEE, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Dominik Winterer, Chengyu Zhang, and Zhendong Su. Validating SMT solvers via semantic fusion. In Proceedings of the 41st ACMSIGPLAN International Conference on Programming Language Design and Implementation, pages 718–730. ACM, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Zhiyong Wu, Jie Liang, Mingzhe Wang, Chijin Zhou, and Yu Jiang. Unicorn: detect runtime errors in timeseries databases with hybrid input synthesis. In 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, pages 251–262. ACM, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Dongwei Xiao, Zhibo Liu, Yuanyuan Yuan, Qi Pang, and Shuai Wang. Metamorphic testing of deep learning compilers. Proc. ACM Meas. Anal. Comput. Syst., 6(1):15:1–15:28, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Haoran Xu, Shuhui Fan, Yongjun Wang, Zhijian Huang, Hongzuo Xu, and Peidai Xie. Tree2tree structural language modeling for compiler fuzzing. In Algorithms and Architectures for Parallel Processing-20th International Conference, pages 563–578. Springer, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Yibiao Yang, Yuming Zhou, Hao Sun, Zhendong Su, Zhiqiang Zuo, Lei Xu, and Baowen Xu. Hunting for bugs in code coverage tools via randomized differential testing. In Proceedings of the 41st International Conference on Software Engineering, pages 488–498. IEEE / ACM, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Peisen Yao, Heqing Huang, Wensheng Tang, Qingkai Shi, Rongxin Wu, and Charles Zhang. Fuzzing SMT solvers via two-dimensional input space exploration. InProceedings of 30th ACMSIGSOFT International Symposium on Software Testing and Analysis, pages 322–335. ACM, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Peisen Yao, Heqing Huang, Wensheng Tang, Qingkai Shi, Rongxin Wu, and Charles Zhang. Skeletal approximation enumeration for SMT solver testing. In Proceedings of 29th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1141–1153. ACM, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Yuanyuan Yuan, Shuai Wang, Mingyue Jiang, and Tsong Yueh Chen. Perception matters: Detecting perception failures of VQA models using metamorphic testing. In Vision and Pattern Recognition, pages 16908–16917. Computer Vision Foundation / IEEE, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Andreas Zeller. Why Programs Fail-A Guide to Systematic Debugging, 2nd Edition. Academic Press, 2009.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: testing database management systems with language validity and coverage feedback. In 2020 ACMSIGSAC Conference on Computer and Communications Security, pages 955–970. ACM, 2020. 358 2023 USENIX Annual Technical Conference USENIX Association",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 33,
+      "text": "Manuel Rigger. sqlancer. https://github.com/ sqlancer/sqlancer, 2022. [Online; accessed Jan2023].",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 34,
+      "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 35,
+      "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA):211:1–211:30, 2020. USENIX Association 2023 USENIX Annual Technical Conference 357",
+      "technique": "tlp",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing tlp",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 36,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, pages 667–682. USENIX Association, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 37,
+      "text": "Manuel Rigger and Zhendong Su. Intramorphic testing: A new approach to the test oracle problem. In Proceedings of the 2022 ACMSIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software, pages 128–136. ACM, 2022.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[ 10,34–36]",
+      "technique": "norec",
+      "sentence": "Therefore, DBMS testing has attracted considerable attention in the industry [ 14,23,38,41] and academia [ 10,34–36].",
+      "context_before": "lt set returned by our synthesized query does not follow the expected approximation relation. We implemented our idea as a DBMS testing system and evaluated it on four widely-used DBMSs: MySQL, MariaDB, TiDB, and OceanBase. By the time of writing, PINOLO has found 41 unique logical bugs in these DBMSs, 39 of which have been confirmed by developers. 1 Introduction Database Management Systems (DBMSs) are widely used as a key component in modern enterprise software. Their correctness and reliability are critical for many enterprise applications, such as online banking, e-shopping, e-payment, etc.",
+      "context_after": "For example, fuzzing, a widely-used testing technique, has been extensively applied to DBMSs [ 14,38], showing its effectiveness in detecting crash bugs. However, as another typical kind ∗Corresponding author: Rongxin Wu (wurongxin@xmu.edu.cn)of bug, logical bugs would cause DBMSs to return an incorrect result set for a given query but can easily go unnoticed by developers since they would not behave with apparent manifestations like system crash. The predominant approach to detecting logical bugs consists of various automatic testing techniques. However, designing effective automatic testing",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 3109,
+      "cited_reference": {
+        "number": 34,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 34,36,39]",
+      "technique": "norec",
+      "sentence": "However, as pointed out by the existing studies [ 34,36,39], differential testing cannot be applied when a generated SQL query cannot comply with the grammar of all selected DBMSs or contains operations that have different semantics between different DBMSs.",
+      "context_before": "tal technical challenges is to characterize a correct result concerning a given query for comparison, which is a classical problem in testing, i.e., test oracle problem [12]. To tackle this challenge, researchers have proposed various ways to obtain the test oracle. The first category is based on differential testing [ 39]. It provides the same generated SQL query to multiple DBMSs for execution and resorts to the querying results to construct the test oracle. More concretely, the inconsistency among the result sets returned by different DBMSs indicates the presence of a potential logical bug.",
+      "context_after": "Although all the DBMSs support the common core syntax of SQL, each of them provides various extensions and forms its own dialect [ 39], which dramatically limits the generality of differential testing. The second category is the oracle-guided synthesis approach [ 36], which does not rely on multiple DBMSs and thus mitigates the limitation of differential testing. It first specifies a randomly-selected row in a database table, namely, pivot row, as the test oracle and then synthesizes the query whose result set should contain this pivot row. The failure of fetching the row with the synthesized",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4474,
+      "cited_reference": {
+        "number": 34,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[ 36]",
+      "technique": "pqs",
+      "sentence": "The second category is the oracle-guided synthesis approach [ 36], which does not rely on multiple DBMSs and thus mitigates the limitation of differential testing.",
+      "context_before": "cle. More concretely, the inconsistency among the result sets returned by different DBMSs indicates the presence of a potential logical bug. However, as pointed out by the existing studies [ 34,36,39], differential testing cannot be applied when a generated SQL query cannot comply with the grammar of all selected DBMSs or contains operations that have different semantics between different DBMSs. Although all the DBMSs support the common core syntax of SQL, each of them provides various extensions and forms its own dialect [ 39], which dramatically limits the generality of differential testing.",
+      "context_after": "It first specifies a randomly-selected row in a database table, namely, pivot row, as the test oracle and then synthesizes the query whose result set should contain this pivot row. The failure of fetching the row with the synthesized query evidences a potential bug underlying the tested DBMS. However, since such an approach considers only one row each time and the synthesis merely focuses on the where clause generation, it would miss logical bugs in various scenarios [ 34,35]. For example, those rows that are duplicated to the pivot row are wrongly fetched or omitted, or the values processed",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 4934,
+      "cited_reference": {
+        "number": 36,
+        "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, pages 667–682. USENIX Association, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[ 34,35]",
+      "technique": "norec",
+      "sentence": "However, since such an approach considers only one row each time and the synthesis merely focuses on the where clause generation, it would miss logical bugs in various scenarios [ 34,35].",
+      "context_before": "L, each of them provides various extensions and forms its own dialect [ 39], which dramatically limits the generality of differential testing. The second category is the oracle-guided synthesis approach [ 36], which does not rely on multiple DBMSs and thus mitigates the limitation of differential testing. It first specifies a randomly-selected row in a database table, namely, pivot row, as the test oracle and then synthesizes the query whose result set should contain this pivot row. The failure of fetching the row with the synthesized query evidences a potential bug underlying the tested DBMS.",
+      "context_after": "For example, those rows that are duplicated to the pivot row are wrongly fetched or omitted, or the values processed by performing operators on the original row data are mistakenly computed and returned. Moreover, as pointed out by some recent studies [34,35], the synthesis requires domain knowledge of the database dialect’s supported operators and functions, and thus the implementation effort is high. USENIX Association 2023 USENIX Annual Technical Conference 345 The third category is the metamorphic testing based approach [ 34,35]. It first transforms a given query qinto another query q′suc",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5392,
+      "cited_reference": {
+        "number": 34,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[34,35]",
+      "technique": "norec",
+      "sentence": "Moreover, as pointed out by some recent studies [34,35], the synthesis requires domain knowledge of the database dialect’s supported operators and functions, and thus the implementation effort is high.",
+      "context_before": "the test oracle and then synthesizes the query whose result set should contain this pivot row. The failure of fetching the row with the synthesized query evidences a potential bug underlying the tested DBMS. However, since such an approach considers only one row each time and the synthesis merely focuses on the where clause generation, it would miss logical bugs in various scenarios [ 34,35]. For example, those rows that are duplicated to the pivot row are wrongly fetched or omitted, or the values processed by performing operators on the original row data are mistakenly computed and returned.",
+      "context_after": "USENIX Association 2023 USENIX Annual Technical Conference 345 The third category is the metamorphic testing based approach [ 34,35]. It first transforms a given query qinto another query q′such that their querying results satisfy a specific relation, which is referred to as a metamorphic relation. The violation of the metamorphic relation upon the querying results indicates the wrong result of evaluating qorq′. For example, TLP [35] decomposes a query qinto three partitioning sub-queries, each of which computes the result sets for a boolean predicate to be evaluated as TRUE ,FALSE, and NULL,",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5784,
+      "cited_reference": {
+        "number": 34,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[ 34,35]",
+      "technique": "norec",
+      "sentence": "USENIX Association 2023 USENIX Annual Technical Conference 345 The third category is the metamorphic testing based approach [ 34,35].",
+      "context_before": "DBMS. However, since such an approach considers only one row each time and the synthesis merely focuses on the where clause generation, it would miss logical bugs in various scenarios [ 34,35]. For example, those rows that are duplicated to the pivot row are wrongly fetched or omitted, or the values processed by performing operators on the original row data are mistakenly computed and returned. Moreover, as pointed out by some recent studies [34,35], the synthesis requires domain knowledge of the database dialect’s supported operators and functions, and thus the implementation effort is high.",
+      "context_after": "It first transforms a given query qinto another query q′such that their querying results satisfy a specific relation, which is referred to as a metamorphic relation. The violation of the metamorphic relation upon the querying results indicates the wrong result of evaluating qorq′. For example, TLP [35] decomposes a query qinto three partitioning sub-queries, each of which computes the result sets for a boolean predicate to be evaluated as TRUE ,FALSE, and NULL, respectively, and then constructs an equivalent query q′ by performing the union operation on these three sub-queries. NOREC [34] tra",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 5986,
+      "cited_reference": {
+        "number": 34,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M7",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For example, TLP [35] decomposes a query qinto three partitioning sub-queries, each of which computes the result sets for a boolean predicate to be evaluated as TRUE ,FALSE, and NULL, respectively, and then constructs an equivalent query q′ by performing the union operation on these three sub-queries.",
+      "context_before": "ted out by some recent studies [34,35], the synthesis requires domain knowledge of the database dialect’s supported operators and functions, and thus the implementation effort is high. USENIX Association 2023 USENIX Annual Technical Conference 345 The third category is the metamorphic testing based approach [ 34,35]. It first transforms a given query qinto another query q′such that their querying results satisfy a specific relation, which is referred to as a metamorphic relation. The violation of the metamorphic relation upon the querying results indicates the wrong result of evaluating qorq′.",
+      "context_after": "NOREC [34] transforms an optimized version of a query into a non-optimized one by the customized rule, e.g., changing “SELECT *FROM tWHERE p” into “ SELECT (pIS TRUE ) FROM t.” Compared with the aforementioned two categories of approaches, metamorphic testing based approaches are much more lightweight to implement and have been proven to be more effective in detecting logical bugs [ 34,35]. However, existing studies instantiate the metamorphic relations as equivalent relations, which are still insufficient to detect many deeply-hidden bugs. This is because it is highly possible that, owing to",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 6402,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M8",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "NOREC [34] transforms an optimized version of a query into a non-optimized one by the customized rule, e.",
+      "context_before": "roach [ 34,35]. It first transforms a given query qinto another query q′such that their querying results satisfy a specific relation, which is referred to as a metamorphic relation. The violation of the metamorphic relation upon the querying results indicates the wrong result of evaluating qorq′. For example, TLP [35] decomposes a query qinto three partitioning sub-queries, each of which computes the result sets for a boolean predicate to be evaluated as TRUE ,FALSE, and NULL, respectively, and then constructs an equivalent query q′ by performing the union operation on these three sub-queries.",
+      "context_after": "g., changing “SELECT *FROM tWHERE p” into “ SELECT (pIS TRUE ) FROM t.” Compared with the aforementioned two categories of approaches, metamorphic testing based approaches are much more lightweight to implement and have been proven to be more effective in detecting logical bugs [ 34,35]. However, existing studies instantiate the metamorphic relations as equivalent relations, which are still insufficient to detect many deeply-hidden bugs. This is because it is highly possible that, owing to the limited search space of mutations, the pair of equivalent queries would still share common buggy oper",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 6705,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker",
+      "surface": "[ 34,35]",
+      "technique": "norec",
+      "sentence": "” Compared with the aforementioned two categories of approaches, metamorphic testing based approaches are much more lightweight to implement and have been proven to be more effective in detecting logical bugs [ 34,35].",
+      "context_before": "tion. The violation of the metamorphic relation upon the querying results indicates the wrong result of evaluating qorq′. For example, TLP [35] decomposes a query qinto three partitioning sub-queries, each of which computes the result sets for a boolean predicate to be evaluated as TRUE ,FALSE, and NULL, respectively, and then constructs an equivalent query q′ by performing the union operation on these three sub-queries. NOREC [34] transforms an optimized version of a query into a non-optimized one by the customized rule, e.g., changing “SELECT *FROM tWHERE p” into “ SELECT (pIS TRUE ) FROM t.",
+      "context_after": "However, existing studies instantiate the metamorphic relations as equivalent relations, which are still insufficient to detect many deeply-hidden bugs. This is because it is highly possible that, owing to the limited search space of mutations, the pair of equivalent queries would still share common buggy operators and functions and eventually return the same results. In such a case, the oracle from the equivalent query cannot provide any hint to detect logical bugs. In this work, we present a new metamorphic testing based approach, named PINOLO, to detect logical bugs. Our idea to instantiat",
+      "section": "1 Introduction",
+      "page": 3,
+      "char_offset": 6881,
+      "cited_reference": {
+        "number": 34,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M10",
+      "found_by": "citation_marker",
+      "surface": "[ 35,37]",
+      "technique": "tlp",
+      "sentence": "Notably, the metamorphic testing based approach has been recognized to be state-of-the-art in DBMS testing for logical bug detection [ 35,37].",
+      "context_before": "ations. In contrast, people are often unaware of logical bugs in DBMSs. Typically, the developers of database-backed applications may realize the abnormal data retrieved from the database, while they are still uncertain whether their application is wrongly implemented or a logical bug of the DBMS is triggered. Therefore, detecting a logical bug in the DBMS has been typically recognized as a problem with both high impact and significant technical challenges.2.3 Metamorphic Testing Recent years have witnessed tremendous efforts in resolving the test oracle for logical bug detection in the DBMSs.",
+      "context_after": "Generally, the metamorphic testing based techniques attempt to construct multiple SQL queries of which the querying results have a specific relation, namely a metamorphic relation. If the querying results violate the metamorphic relation, we can have the confidence that at least one of the queries triggers a logical bug in the tested DBMS. For example, NOREC [34] transforms a query into a form in which the DBMS does not apply optimizations, which yields the test oracle that the two queries should make the tested DBMS return the same result. Besides, TLP [35] gets the equivalent query result b",
+      "section": "2.2 Logical Bugs in DBMSs",
+      "page": 4,
+      "char_offset": 14445,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA):211:1–211:30, 2020. USENIX Association 2023 USENIX Annual Technical Conference 357",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M11",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "For example, NOREC [34] transforms a query into a form in which the DBMS does not apply optimizations, which yields the test oracle that the two queries should make the tested DBMS return the same result.",
+      "context_before": "Recent years have witnessed tremendous efforts in resolving the test oracle for logical bug detection in the DBMSs. Notably, the metamorphic testing based approach has been recognized to be state-of-the-art in DBMS testing for logical bug detection [ 35,37]. Generally, the metamorphic testing based techniques attempt to construct multiple SQL queries of which the querying results have a specific relation, namely a metamorphic relation. If the querying results violate the metamorphic relation, we can have the confidence that at least one of the queries triggers a logical bug in the tested DBMS.",
+      "context_after": "Besides, TLP [35] gets the equivalent query result by splitting the input query into several sub-queries and merging the results of sub-queries into one. When adapting the metamorphic testing, they take randomlygenerated queries as the seed queries and then transform them into other queries to ensure the metamorphic relations, which automates the testing process for logical bug detection. Unfortunately, the existing effort has not resolved the test oracle problem perfectly. To the best of our knowledge, previous studies only leverage the equivalent transformations, which are supported by the",
+      "section": "2.2 Logical Bugs in DBMSs",
+      "page": 4,
+      "char_offset": 14930,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "Besides, TLP [35] gets the equivalent query result by splitting the input query into several sub-queries and merging the results of sub-queries into one.",
+      "context_before": "t in DBMS testing for logical bug detection [ 35,37]. Generally, the metamorphic testing based techniques attempt to construct multiple SQL queries of which the querying results have a specific relation, namely a metamorphic relation. If the querying results violate the metamorphic relation, we can have the confidence that at least one of the queries triggers a logical bug in the tested DBMS. For example, NOREC [34] transforms a query into a form in which the DBMS does not apply optimizations, which yields the test oracle that the two queries should make the tested DBMS return the same result.",
+      "context_after": "When adapting the metamorphic testing, they take randomlygenerated queries as the seed queries and then transform them into other queries to ensure the metamorphic relations, which automates the testing process for logical bug detection. Unfortunately, the existing effort has not resolved the test oracle problem perfectly. To the best of our knowledge, previous studies only leverage the equivalent transformations, which are supported by the tested DBMS or conducted by their approaches, leaving the functions and operators in the query unchanged. This greatly limits their approaches to explorin",
+      "section": "2.2 Logical Bugs in DBMSs",
+      "page": 4,
+      "char_offset": 15135,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "For example, the logical bug shown in Figure 1 can not be revealed by NOREC [34] and TLP [35], as the transformations preserve all the operators and the functions, still triggering the buggy evaluation process.",
+      "context_before": "seed queries and then transform them into other queries to ensure the metamorphic relations, which automates the testing process for logical bug detection. Unfortunately, the existing effort has not resolved the test oracle problem perfectly. To the best of our knowledge, previous studies only leverage the equivalent transformations, which are supported by the tested DBMS or conducted by their approaches, leaving the functions and operators in the query unchanged. This greatly limits their approaches to exploring the code of the tested DBMSs and thus reduces the chance of finding logical bugs.",
+      "context_after": "To improve the capability of the metamorphic testing in finding logical bugs in the DBMSs, we propose a new DBMS testing approach in this work, which relaxes the equivalence relation with a less restrictive metamorphic relation, finally supporting finding more insightful logical bugs. 3 Approximate Query Synthesis We propose the approximate query synthesis technique PINOLO1for detecting logical bugs in the DBMSs. Basically, our insight comes from the intuition that the mutation of specific grammatical constructs can induce the approximation relation between the original query and the mutated",
+      "section": "2.2 Logical Bugs in DBMSs",
+      "page": 4,
+      "char_offset": 15972,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "citation_marker",
+      "surface": "[ 34–36]",
+      "technique": "norec",
+      "sentence": "Compared with the existing techniques [ 34–36],PINOLO considers more SQL features, such as set operators, arithmetic expressions, sub-queries, etc.",
+      "context_before": "the approximation relation between each synthesized query and the seed query based on the principle of structural induction. 3.5 Summary PINOLO automates the DBMS testing via the approximate query synthesis, which discovers underlying logical bugs in the DBMSs. The syntax in Figure 3 ensures the syntactical validity of the seed queries, and furthermore, guarantees that the synthesized queries have valid SQL syntax. Meanwhile, our approximate mutators enable us to obtain the approximation relation between the seed query and the synthesized ones, which perfectly resolves the test oracle problem.",
+      "context_after": "The expressive syntax permits us to test the DBMS more thoroughly and discover more logical bugs reported in previous studies, which will be evidenced by our experiments. Besides, the approximate operators can aggressively mutate the seed query, which may remove the buggy operators and functions, revealing the logical bugs more thoroughly than existing techniques. Lastly, it is worth noting that PINOLO provides a general framework for discovering the logical bugs in the DBMS. We can further extend the syntax of SQL queries and instantiate more approximate mutators, which can promote the capab",
+      "section": "3.5 Summary",
+      "page": 8,
+      "char_offset": 34420,
+      "cited_reference": {
+        "number": 34,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[ 36]",
+      "technique": "pqs",
+      "sentence": "Following the best practice summarized in the prior study [ 36], we restrict the number of table records to be no more than 30.",
+      "context_before": "extend the syntax of SQL queries and instantiate more approximate mutators, which can promote the capability of discovering the logical bugs triggered by sophisticated SQL queries. 4 Implementation We implemented our approach PINOLO as a DBMS testing system, which was written in GOwith 8,055 lines of code. The source code of our tool is hosted in the github repository2. Next, we present more details of our implementation decisions that are important for the outcome of our experiments. Database population. We randomly generate the database tables by leveraging an existing tool, GO-RANDGEN [27].",
+      "context_after": "Besides, we randomly generate the tables with the same attributes, which makes the join operator yield a non-trivial result. Particularly, we avoid null values in any table, as the test oracle requires non-null values as a prerequisite, which is stated in Theorem 3.1. Seed query generation and parsing. To generate seed queries as the input of our synthesis algorithm, we utilize GORANDGEN [27] to automatically produce seed SQL queries. Specifically, we use the general-purpose parser generator BISON [13] to write a context-free grammar file describing the SQL syntax with a series of production",
+      "section": "4 Implementation",
+      "page": 8,
+      "char_offset": 35665,
+      "cited_reference": {
+        "number": 36,
+        "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, pages 667–682. USENIX Association, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M16",
+      "found_by": "citation_marker",
+      "surface": "[ 34–36]",
+      "technique": "norec",
+      "sentence": "They are also commonly used in the evaluation of previous studies [ 34–36].",
+      "context_before": "inux kernel version: 5.4.0-135-generic. To ensure fair comparisons, we allocated four threads for each DBMS testing in our following experiments. Tested DBMS. Our focus was on testing four widely-used and large-scale open-source DBMSs: MySQL, MariaDB, TiDB, and OceanBase. There are two main reasons for the subject selection. First, these selected DBMSs are representative DBMSs from phenomenal open-source and/or commercial products: MySQL and MariaDB are the two most well-known open-source DBMSs; OceanBase is a mature commercial database product from Ant Group; TiDB is developed by PingCap Inc.",
+      "context_after": "Moreover, we admit and discuss potential limitations introduced by our selection of DBMS systems in § 6. Second, we chose a DBMS whose SQL syntax is compatible with MySQL as an evaluation subject to reduce the implementation effort. This is because, although our approach can be generalized to other DBMSs, the implementation of the seed query generation and parsing (See § 4) requires a grammar file that describes the SQL syntax of a tested DBMS. To obtain timely feedback from developers, we tested the latest release versions of the selected DBMSs: MySQL 8.0.31, MariaDB 10.11.1, TiDB 6.4.0, and",
+      "section": "5.1 Experiment Setup",
+      "page": 9,
+      "char_offset": 40340,
+      "cited_reference": {
+        "number": 34,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M17",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "We compared PINOLO with the three state-ofthe-art logical bug detection techniques, namely PQS [ 36], NOREC [34], and TLP [35], respectively, which correspond to three kinds of test oracles.",
+      "context_before": "cuss potential limitations introduced by our selection of DBMS systems in § 6. Second, we chose a DBMS whose SQL syntax is compatible with MySQL as an evaluation subject to reduce the implementation effort. This is because, although our approach can be generalized to other DBMSs, the implementation of the seed query generation and parsing (See § 4) requires a grammar file that describes the SQL syntax of a tested DBMS. To obtain timely feedback from developers, we tested the latest release versions of the selected DBMSs: MySQL 8.0.31, MariaDB 10.11.1, TiDB 6.4.0, and OceanBase 4.0.0. Baseline.",
+      "context_after": "Similar to our approach, these baselines also require knowledge about the SQL syntax of different DBMSs for seed query generation and parsing. Unfortunately, their implementations cannot support all the selected DBMSs. We also sought help from their authors, but they still could not fix the problems before the paper submission. Therefore, we skipped the evaluation of the baselines on the unsupported DBMSs. Moreover, we tried to use the same random seed as the baselines. However, we found that they can neither export their random seeds nor import the random seeds provided by users. Therefore,",
+      "section": "5.1 Experiment Setup",
+      "page": 9,
+      "char_offset": 41043,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M18",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "352 2023 USENIX Annual Technical Conference USENIX Association 024681012141618202224 Testing Time(h)0246810121416Unique Bug (a) MySQLPINOLO TLP 024681012141618202224 Testing Time(h)0246810121416Unique Bug (b) MariaDBPINOLO NoREC 024681012141618202224 Testing Time(h)024681012Unique Bug (c) TiDBPINOLO TLP 024681012141618202224 Testing Time(h)0123Unique Bug (d) OceanBasePINOLOFigure 5: Comparison between PINOLO and the baselines.",
+      "context_before": "nnot support all the selected DBMSs. We also sought help from their authors, but they still could not fix the problems before the paper submission. Therefore, we skipped the evaluation of the baselines on the unsupported DBMSs. Moreover, we tried to use the same random seed as the baselines. However, we found that they can neither export their random seeds nor import the random seeds provided by users. Therefore, we generated the random seeds by ourselves for PINOLO. To understand the impacts of the random seed query generation, we investigated how PINOLO is robust to such randomness in § 5.4.",
+      "context_after": "(a)-(d) show the number of unique logical bugs over time (24 hours) on each DBMS. We ignore the results of the baselines on the unsupported DBMSs and the baselines finding no bugs. Table 2: The demographics of the DBMSs under the test DBMS Version GitHub Stars LOC First Release MySQL 8.0.31 8.6K 4,766,086 1995 MariaDB 10.11.1 4.6K 3,727,410 2009 TiDB 6.4.0 33.1K 985,518 2017 OceanBase 4.0.0 5.1K 2,722,881 2021 Table 3: Applicability of existing logical detection techniques and PINOLO for the selected DBMSs DBMSPQS NOREC TLP PINOLO MySQL ✓ × ✓ ✓ MariaDB × ✓ × ✓ TiDB × × ✓ ✓ OceanBase ✓ ✓ ✓ ✓ 5",
+      "section": "5.1 Experiment Setup",
+      "page": 9,
+      "char_offset": 42017,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp",
+        "norec"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "1K 2,722,881 2021 Table 3: Applicability of existing logical detection techniques and PINOLO for the selected DBMSs DBMSPQS NOREC TLP PINOLO MySQL ✓ × ✓ ✓ MariaDB × ✓ × ✓ TiDB × × ✓ ✓ OceanBase ✓ ✓ ✓ ✓ 5.",
+      "context_before": "C 024681012141618202224 Testing Time(h)024681012Unique Bug (c) TiDBPINOLO TLP 024681012141618202224 Testing Time(h)0123Unique Bug (d) OceanBasePINOLOFigure 5: Comparison between PINOLO and the baselines. (a)-(d) show the number of unique logical bugs over time (24 hours) on each DBMS. We ignore the results of the baselines on the unsupported DBMSs and the baselines finding no bugs. Table 2: The demographics of the DBMSs under the test DBMS Version GitHub Stars LOC First Release MySQL 8.0.31 8.6K 4,766,086 1995 MariaDB 10.11.1 4.6K 3,727,410 2009 TiDB 6.4.0 33.1K 985,518 2017 OceanBase 4.0.0 5.",
+      "context_after": "2 Effectiveness of PINOLO We used PINOLO to test the latest version of MySQL, MariaDB, TiDB, and OceanBase for 24 hours. Table 4 summarizes the results of PINOLO. The column Allshows the number of problematic query pairs that induce unexpected results, which indicate the existence of logical bugs. As we can see, PINOLO discovered a large number of problematic query pairs, ranging from 4,675 to 46,772 for the tested DBMSs. However, we found that most of these pairs can be attributed to the same bug. To relieve the developers from the heavy burden of checking the duplicate bugs, we leverage the",
+      "section": "5.1 Experiment Setup",
+      "page": 10,
+      "char_offset": 42845,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": ", PQS, NOREC, and TLP.",
+      "context_before": "bugs have been confirmed, while the rest are still waiting for the investigation. We sampled several bug reports of MariaDB submitted by others and found that developers typically take a much longer time to handle the bugs. To keep track of the status of our reported bugs, we release the bug list in a public GitHub repository3. Answer to RQ1 :PINOLO discovers 41 unique bugs on MySQL, MariaDB, TiDB, and OceanBase, 39 of which have been confirmed by DBMS developers. 5.3 Comparisons on Detecting Logical Bugs Logical bug detection. We compared PINOLO with the three state-of-the-art baselines, i.e.",
+      "context_after": "We ran all methods with a time budget of 24 hours. The comparison results are shown in Figure 5. Note that the baselines do not support all the DBMSs, and thus we only concentrated on the comparison between PINOLO and the runnable baselines with respect to each DBMS. For MySQL, PINOLO detected 14 logical bugs, while TLP only discovered 2 bugs. For MariaDB, PINOLO detected 14 bugs, while NOREC discovered 10 bugs. For TiDB, PINOLO detected 11 bugs, while TLP only discovered 2 bugs. For OceanBase, PINOLO can detect 2 bugs, while the baselines cannot find any. We also manually verified the overla",
+      "section": "5.3 Comparisons on Detecting Logical Bugs",
+      "page": 10,
+      "char_offset": 45034,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs",
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M21",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For MySQL, PINOLO detected 14 logical bugs, while TLP only discovered 2 bugs.",
+      "context_before": "ug list in a public GitHub repository3. Answer to RQ1 :PINOLO discovers 41 unique bugs on MySQL, MariaDB, TiDB, and OceanBase, 39 of which have been confirmed by DBMS developers. 5.3 Comparisons on Detecting Logical Bugs Logical bug detection. We compared PINOLO with the three state-of-the-art baselines, i.e., PQS, NOREC, and TLP. We ran all methods with a time budget of 24 hours. The comparison results are shown in Figure 5. Note that the baselines do not support all the DBMSs, and thus we only concentrated on the comparison between PINOLO and the runnable baselines with respect to each DBMS.",
+      "context_after": "For MariaDB, PINOLO detected 14 bugs, while NOREC discovered 10 bugs. For TiDB, PINOLO detected 11 bugs, while TLP only discovered 2 bugs. For OceanBase, PINOLO can detect 2 bugs, while the baselines cannot find any. We also manually verified the overlap in the bugs detected by PINOLO and the baselines. For MariaDB, 4 out of 10 bugs detected by NOREC can also be found by PINOLO. For TiDB, 1 out of 2 bugs detected by TLP can also be found by PINOLO. There are no bugs detected by PQS. Figure 5 shows the logical bug detection progress over time for PINOLO and the baselines. We found that PINOLO",
+      "section": "5.3 Comparisons on Detecting Logical Bugs",
+      "page": 10,
+      "char_offset": 45324,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M22",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "For MariaDB, PINOLO detected 14 bugs, while NOREC discovered 10 bugs.",
+      "context_before": "que bugs on MySQL, MariaDB, TiDB, and OceanBase, 39 of which have been confirmed by DBMS developers. 5.3 Comparisons on Detecting Logical Bugs Logical bug detection. We compared PINOLO with the three state-of-the-art baselines, i.e., PQS, NOREC, and TLP. We ran all methods with a time budget of 24 hours. The comparison results are shown in Figure 5. Note that the baselines do not support all the DBMSs, and thus we only concentrated on the comparison between PINOLO and the runnable baselines with respect to each DBMS. For MySQL, PINOLO detected 14 logical bugs, while TLP only discovered 2 bugs.",
+      "context_after": "For TiDB, PINOLO detected 11 bugs, while TLP only discovered 2 bugs. For OceanBase, PINOLO can detect 2 bugs, while the baselines cannot find any. We also manually verified the overlap in the bugs detected by PINOLO and the baselines. For MariaDB, 4 out of 10 bugs detected by NOREC can also be found by PINOLO. For TiDB, 1 out of 2 bugs detected by TLP can also be found by PINOLO. There are no bugs detected by PQS. Figure 5 shows the logical bug detection progress over time for PINOLO and the baselines. We found that PINOLO is more efficient in finding logical bugs compared to all of the basel",
+      "section": "5.3 Comparisons on Detecting Logical Bugs",
+      "page": 10,
+      "char_offset": 45402,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M23",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For TiDB, PINOLO detected 11 bugs, while TLP only discovered 2 bugs.",
+      "context_before": "confirmed by DBMS developers. 5.3 Comparisons on Detecting Logical Bugs Logical bug detection. We compared PINOLO with the three state-of-the-art baselines, i.e., PQS, NOREC, and TLP. We ran all methods with a time budget of 24 hours. The comparison results are shown in Figure 5. Note that the baselines do not support all the DBMSs, and thus we only concentrated on the comparison between PINOLO and the runnable baselines with respect to each DBMS. For MySQL, PINOLO detected 14 logical bugs, while TLP only discovered 2 bugs. For MariaDB, PINOLO detected 14 bugs, while NOREC discovered 10 bugs.",
+      "context_after": "For OceanBase, PINOLO can detect 2 bugs, while the baselines cannot find any. We also manually verified the overlap in the bugs detected by PINOLO and the baselines. For MariaDB, 4 out of 10 bugs detected by NOREC can also be found by PINOLO. For TiDB, 1 out of 2 bugs detected by TLP can also be found by PINOLO. There are no bugs detected by PQS. Figure 5 shows the logical bug detection progress over time for PINOLO and the baselines. We found that PINOLO is more efficient in finding logical bugs compared to all of the baselines. Within one hour, PINOLO was able to detect 57.1% (8/14) of bugs",
+      "section": "5.3 Comparisons on Detecting Logical Bugs",
+      "page": 10,
+      "char_offset": 45472,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M24",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "For MariaDB, 4 out of 10 bugs detected by NOREC can also be found by PINOLO.",
+      "context_before": "The comparison results are shown in Figure 5. Note that the baselines do not support all the DBMSs, and thus we only concentrated on the comparison between PINOLO and the runnable baselines with respect to each DBMS. For MySQL, PINOLO detected 14 logical bugs, while TLP only discovered 2 bugs. For MariaDB, PINOLO detected 14 bugs, while NOREC discovered 10 bugs. For TiDB, PINOLO detected 11 bugs, while TLP only discovered 2 bugs. For OceanBase, PINOLO can detect 2 bugs, while the baselines cannot find any. We also manually verified the overlap in the bugs detected by PINOLO and the baselines.",
+      "context_after": "For TiDB, 1 out of 2 bugs detected by TLP can also be found by PINOLO. There are no bugs detected by PQS. Figure 5 shows the logical bug detection progress over time for PINOLO and the baselines. We found that PINOLO is more efficient in finding logical bugs compared to all of the baselines. Within one hour, PINOLO was able to detect 57.1% (8/14) of bugs on MySQL, 85.7% (12/14) on MariaDB, 63.6% (7/11) on TiDB and 50% (1/2) on OceanBase. Code coverage. To understand why PINOLO can find more 3https://github.com/qaqcatz/impomysql_bugreports USENIX Association 2023 USENIX Annual Technical Confer",
+      "section": "5.3 Comparisons on Detecting Logical Bugs",
+      "page": 10,
+      "char_offset": 45707,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M25",
+      "found_by": "technique",
+      "surface": "TLP",
+      "technique": "tlp",
+      "sentence": "For TiDB, 1 out of 2 bugs detected by TLP can also be found by PINOLO.",
+      "context_before": "support all the DBMSs, and thus we only concentrated on the comparison between PINOLO and the runnable baselines with respect to each DBMS. For MySQL, PINOLO detected 14 logical bugs, while TLP only discovered 2 bugs. For MariaDB, PINOLO detected 14 bugs, while NOREC discovered 10 bugs. For TiDB, PINOLO detected 11 bugs, while TLP only discovered 2 bugs. For OceanBase, PINOLO can detect 2 bugs, while the baselines cannot find any. We also manually verified the overlap in the bugs detected by PINOLO and the baselines. For MariaDB, 4 out of 10 bugs detected by NOREC can also be found by PINOLO.",
+      "context_after": "There are no bugs detected by PQS. Figure 5 shows the logical bug detection progress over time for PINOLO and the baselines. We found that PINOLO is more efficient in finding logical bugs compared to all of the baselines. Within one hour, PINOLO was able to detect 57.1% (8/14) of bugs on MySQL, 85.7% (12/14) on MariaDB, 63.6% (7/11) on TiDB and 50% (1/2) on OceanBase. Code coverage. To understand why PINOLO can find more 3https://github.com/qaqcatz/impomysql_bugreports USENIX Association 2023 USENIX Annual Technical Conference 353 024681012141618202224 Testing Time(h)100110120130140150Coverag",
+      "section": "5.3 Comparisons on Detecting Logical Bugs",
+      "page": 10,
+      "char_offset": 45784,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M26",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "There are no bugs detected by PQS.",
+      "context_before": "between PINOLO and the runnable baselines with respect to each DBMS. For MySQL, PINOLO detected 14 logical bugs, while TLP only discovered 2 bugs. For MariaDB, PINOLO detected 14 bugs, while NOREC discovered 10 bugs. For TiDB, PINOLO detected 11 bugs, while TLP only discovered 2 bugs. For OceanBase, PINOLO can detect 2 bugs, while the baselines cannot find any. We also manually verified the overlap in the bugs detected by PINOLO and the baselines. For MariaDB, 4 out of 10 bugs detected by NOREC can also be found by PINOLO. For TiDB, 1 out of 2 bugs detected by TLP can also be found by PINOLO.",
+      "context_after": "Figure 5 shows the logical bug detection progress over time for PINOLO and the baselines. We found that PINOLO is more efficient in finding logical bugs compared to all of the baselines. Within one hour, PINOLO was able to detect 57.1% (8/14) of bugs on MySQL, 85.7% (12/14) on MariaDB, 63.6% (7/11) on TiDB and 50% (1/2) on OceanBase. Code coverage. To understand why PINOLO can find more 3https://github.com/qaqcatz/impomysql_bugreports USENIX Association 2023 USENIX Annual Technical Conference 353 024681012141618202224 Testing Time(h)100110120130140150Coverage(K) (a) MySQLPINOLO PQSTLP 0246810",
+      "section": "5.3 Comparisons on Detecting Logical Bugs",
+      "page": 10,
+      "char_offset": 45855,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M27",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "For MySQL, the improvement of PINOLO over PQS and TLP is 2.",
+      "context_before": "0] to compute the line coverage achieved by PINOLO and all others. Note that we are unable to provide code coverage results for TiDB and OceanBase. As TiDB is developed in GO, we cannot find any feasible tool to support the program instrumentation or the code coverage profile for system test4. For OceanBase, the instrumented binary has to be deployed by the specific tool OBDEPLOY [24], which breaks the functionality of code coverage profiling. Figure 6 shows the code coverage over time for MySQL and MariaDB. The results show that PINOLO achieves a higher line coverage than the three baselines.",
+      "context_after": "2% (3,008 lines) and 4.0% (5,316 lines), respectively. For MariaDB, the improvement of PINOLO over NOREC is 12.4% (2,835 lines). Code coverage is well recognized as an approximation of testing capability, because a bug cannot be detected if its buggy code is not executed. However, larger code coverage does not mean more bugs. To better understand the impact of larger code coverage achieved by PINOLO, we further investigated whether there were some bugs whose buggy code is uniquely covered by PINOLO. TiDB#40015 [ 42] is a typical example. The root cause of this bug is the improper exception han",
+      "section": "5.3 Comparisons on Detecting Logical Bugs",
+      "page": 11,
+      "char_offset": 47309,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M28",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "For MariaDB, the improvement of PINOLO over NOREC is 12.",
+      "context_before": "ge results for TiDB and OceanBase. As TiDB is developed in GO, we cannot find any feasible tool to support the program instrumentation or the code coverage profile for system test4. For OceanBase, the instrumented binary has to be deployed by the specific tool OBDEPLOY [24], which breaks the functionality of code coverage profiling. Figure 6 shows the code coverage over time for MySQL and MariaDB. The results show that PINOLO achieves a higher line coverage than the three baselines. For MySQL, the improvement of PINOLO over PQS and TLP is 2.2% (3,008 lines) and 4.0% (5,316 lines), respectively.",
+      "context_after": "4% (2,835 lines). Code coverage is well recognized as an approximation of testing capability, because a bug cannot be detected if its buggy code is not executed. However, larger code coverage does not mean more bugs. To better understand the impact of larger code coverage achieved by PINOLO, we further investigated whether there were some bugs whose buggy code is uniquely covered by PINOLO. TiDB#40015 [ 42] is a typical example. The root cause of this bug is the improper exception handling in the function vecGetDateFromString, which has been covered by PINOLO but missed by other baselines duri",
+      "section": "5.3 Comparisons on Detecting Logical Bugs",
+      "page": 11,
+      "char_offset": 47422,
+      "found_by_all": [
+        "technique"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M29",
+      "found_by": "citation_marker",
+      "surface": "[35]",
+      "technique": "tlp",
+      "sentence": "This result indicates that the logical bugs are typically difficult and slow to be found, which is also consistent with the findings of the prior study [35].",
+      "context_before": "MSs necessarily. For example, we received an appreciation from the MySQL developers in one of the bug reports with the S3 level: “ Thank you for your contribution. It is our standpoint that all bugs should be fixed, whether major or minor. ” Table 6 also shows the bug impact duration, which is computed as the interval between the time of the bug-inducing version and the bug reporting time. Surprisingly, we found that 10 (25.6%) and 21 (53.8%) of bugs have lasted for 510 and 1-5 years, respectively. Specifically, the two earliest bugs of MySQL [ 20] and MariaDB [ 17] can be traced back to 2014.",
+      "context_after": "False Positive and False Negative. According to Theorem 3.1, PINOLO will not produce false positives in theory. However, we observe two bugs that have not been confirmed by developers for more than five months, and thus suspect that they are false positives. We manually inspected the two cases. In one bug report, a query returns “0”, while the approximated query returns “-0”. In the other one, a query returns “0”, while the approximated query returns “-0.001”. Although PINOLO considers the above inconsistency results as bugs, developers may have a higher tolerance for such inconsistencies. Th",
+      "section": "5.5 Discussion",
+      "page": 12,
+      "char_offset": 52206,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA):211:1–211:30, 2020. USENIX Association 2023 USENIX Annual Technical Conference 357",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M30",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "However, we believe that it is non-trivial to detect new bugs in these selected DBMSs, as they have been thoroughly tested by SQLancer [ 33].",
+      "context_before": "ision of bug deduplication. 6 Threats to Validity The threat to internal validity is primarily associated with the implementation of our approach. To mitigate this concern, we have employed several DBMSs to cross-check whether the mutants generated by PINOLO accurately represented overapproximations or under-approximations of the seed query. The threat to external validity lies in the representative of the evaluation subjects. Our proposed approach was evaluated on a restricted set of DBMSs, as explained in the evaluation section. As a result, the conclusion drawn in this paper may be limited.",
+      "context_after": "In the future, we will extend the implementation of PINOLO to support the evaluation of additional open-source and commercial DBMSs. 7 Related Work PINOLO is an unique DBMS testing system for finding logical bugs in DBMSs, but draws inspiration from several areas in the literature, including DBMS testing, metamorphic testing, differential testing, and grammar fuzzing. USENIX Association 2023 USENIX Annual Technical Conference 355 DBMS testing. Recent efforts on DBMS testing focus on various aspects of DBMSs. Most of them target discovering logical bugs in the relational DBMS [ 34–36]. They us",
+      "section": "6 Threats to Validity",
+      "page": 12,
+      "char_offset": 55725,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M31",
+      "found_by": "citation_marker",
+      "surface": "[ 34–36]",
+      "technique": "norec",
+      "sentence": "Most of them target discovering logical bugs in the relational DBMS [ 34–36].",
+      "context_before": "w bugs in these selected DBMSs, as they have been thoroughly tested by SQLancer [ 33]. In the future, we will extend the implementation of PINOLO to support the evaluation of additional open-source and commercial DBMSs. 7 Related Work PINOLO is an unique DBMS testing system for finding logical bugs in DBMSs, but draws inspiration from several areas in the literature, including DBMS testing, metamorphic testing, differential testing, and grammar fuzzing. USENIX Association 2023 USENIX Annual Technical Conference 355 DBMS testing. Recent efforts on DBMS testing focus on various aspects of DBMSs.",
+      "context_after": "They use specific metamorphic relations or multiple implementations as the oracles and generate syntactically valid SQL queries for metamorphic testing [ 5] or differential testing [ 18]. Some also attempt to improve the coverage of the DBMSs and leverage fuzzing techniques to enumerate the queries in various forms [ 10,46,49,57].PINOLO uses a new design, termed approximate query synthesis, and does not use other DBMS implementations as the oracle. Metamorphic testing. Metamorphic testing [ 5] has become more and more popular over the past decade. It has been used in testing many software sys",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 56381,
+      "cited_reference": {
+        "number": 34,
+        "text": "Manuel Rigger and Zhendong Su. Detecting optimization bugs in database engines via non-optimizing reference engine construction. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1140– 1152. ACM, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing norec",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M32",
+      "found_by": "citation_marker",
+      "surface": "[ 35,36]",
+      "technique": "tlp",
+      "sentence": "It has been used in testing many software systems, such as compilers [ 16,50], SMT solvers [ 48,54], DBMSs [ 35,36], and AI systems [ 2,47,55].",
+      "context_before": "logical bugs in the relational DBMS [ 34–36]. They use specific metamorphic relations or multiple implementations as the oracles and generate syntactically valid SQL queries for metamorphic testing [ 5] or differential testing [ 18]. Some also attempt to improve the coverage of the DBMSs and leverage fuzzing techniques to enumerate the queries in various forms [ 10,46,49,57].PINOLO uses a new design, termed approximate query synthesis, and does not use other DBMS implementations as the oracle. Metamorphic testing. Metamorphic testing [ 5] has become more and more popular over the past decade.",
+      "context_after": "Metamorphic testing uses one type of metamorphic relations to compare outputs produced by a seed input and a mutated one. As long as the two outputs violate the specific metamorphic relation, then at least one of the two inputs yields a wrong result [ 4].PINOLO utilizes an instantiation of a metamorphic relation, i.e., the approximation relation, as an effective testing oracle for discovering logical bugs in the DBMSs. Similar to the skeleton approximation enumeration in the SMT solver testing [ 54],PINOLO performs the over-/under-approximation of the seed queries. However, PINOLO has to deal",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 57013,
+      "cited_reference": {
+        "number": 35,
+        "text": "Manuel Rigger and Zhendong Su. Finding bugs in database systems via query partitioning. Proc. ACM Program. Lang., 4(OOPSLA):211:1–211:30, 2020. USENIX Association 2023 USENIX Annual Technical Conference 357",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing tlp",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "tlp"
+      ]
+    },
+    {
+      "id": "M33",
+      "found_by": "technique",
+      "surface": "NOREC",
+      "technique": "norec",
+      "sentence": "Therefore, it is able to detect insightful logical bugs that existing approaches, such as NOREC [34] and TLP [35], fail to discover.",
+      "context_before": "ong as the two outputs violate the specific metamorphic relation, then at least one of the two inputs yields a wrong result [ 4].PINOLO utilizes an instantiation of a metamorphic relation, i.e., the approximation relation, as an effective testing oracle for discovering logical bugs in the DBMSs. Similar to the skeleton approximation enumeration in the SMT solver testing [ 54],PINOLO performs the over-/under-approximation of the seed queries. However, PINOLO has to deal with more sophisticated syntax and supports more flexible mutations so that more buggy operators and functions can be removed.",
+      "context_after": "Differential testing. Apart from metamorphic testing, differential testing provides another testing paradigm for resolving the oracle problem in software testing [ 6,40,52,53]. Utilizing another software system with the same functionality as an oracle implementation, differential testing techniques compare the system’s outputs under testing and the oracle implementation and reveal potential functional bugs with the divergent outputs. Although the techniques can generate the inputs of the systems flexibly, they can only be applied to software systems that have other implementations supporting",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 57884,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec",
+        "tlp"
+      ]
+    },
+    {
+      "id": "M34",
+      "found_by": "citation_marker",
+      "surface": "[ 36,57]",
+      "technique": "pqs",
+      "sentence": "It has been widely used in testing many real-world software systems, such as browsers [ 45], compilers [ 51], and DBMSs [ 36,57].",
+      "context_before": "eal potential functional bugs with the divergent outputs. Although the techniques can generate the inputs of the systems flexibly, they can only be applied to software systems that have other implementations supporting the same functionality, such as JVM [ 6], ORM frameworks [ 40], and SMT solvers [ 53]. We believe PINOLO and other metamorphic testing approaches are orthogonal to differential testing techniques, which can be applied and strengthened to each other in testing DBMSs. Grammar fuzzing. Grammar fuzzing is used to generate inputs that satisfy a specific language syntax [ 8,11,19,19].",
+      "context_after": "The generated inputs can always pass the syntax checking of software systems, which avoids the unnecessary enumeration of the inputs in an ill form, improving the effectiveness of generated inputs. Instead of relying ona specific grammar, PINOLO mutates an existing seed query to synthesize an approximate query. Our synthesis process rewrites specific grammatical constructs in the seed query, which ensures the syntactical validity of the synthesized one. We do think it is also promising to utilize existing grammar fuzzing techniques to enumerate initial seed queries automatically [ 36,57], whi",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 58998,
+      "cited_reference": {
+        "number": 36,
+        "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, pages 667–682. USENIX Association, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M35",
+      "found_by": "citation_marker",
+      "surface": "[ 36,57]",
+      "technique": "pqs",
+      "sentence": "We do think it is also promising to utilize existing grammar fuzzing techniques to enumerate initial seed queries automatically [ 36,57], which can provide more opportunities for improving the coverage in the DBMS testing.",
+      "context_before": "8,11,19,19]. It has been widely used in testing many real-world software systems, such as browsers [ 45], compilers [ 51], and DBMSs [ 36,57]. The generated inputs can always pass the syntax checking of software systems, which avoids the unnecessary enumeration of the inputs in an ill form, improving the effectiveness of generated inputs. Instead of relying ona specific grammar, PINOLO mutates an existing seed query to synthesize an approximate query. Our synthesis process rewrites specific grammatical constructs in the seed query, which ensures the syntactical validity of the synthesized one.",
+      "context_after": "8 Conclusion This paper presents PINOLO, an automatic query synthesizer for discovering logical bugs in DBMSs. Given a seed query, PINOLO mutates specific SQL constructs and generates a query that over-approximates or under-approximates the seed query. We posit that the approximation relation provides effective guidance for discovering logical bugs underlying DBMSs. Our experimental results demonstrate the effectiveness of PINOLO. Benefiting from our approximate query synthesis, PINOLO discovers 41 logical bugs in four mature DBMSs. At the time of the submission, 39 bugs have been confirmed b",
+      "section": "7 Related Work",
+      "page": 13,
+      "char_offset": 59586,
+      "cited_reference": {
+        "number": 36,
+        "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, pages 667–682. USENIX Association, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M17"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:13:36Z",
+    "is_model_written": true,
+    "summary": "Pinolo detects logical bugs in DBMSs by approximate query synthesis: rather than constructing an exactly equivalent query, it mutates a query into one whose result set must be a superset or subset of the original, and reports a violation of that containment as a bug. The authors survey three existing categories -- differential testing, oracle-guided synthesis, and metamorphic testing -- and argue each is limited, the last because its transformations preserve all operators and so miss certain bugs. Pinolo covers more SQL features than those techniques, including set operators, arithmetic expressions and sub-queries.",
+    "narrative": "Pinolo positions itself directly against SQLancer's oracles: PQS is the oracle-guided synthesis it says considers only one row at a time, NoREC and TLP are the metamorphic approaches it says preserve too much of the query, and it states that metamorphic testing of that kind is recognised as the state of the art in DBMS logical bug detection.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "definition",
+      "M4": "motivation",
+      "M5": "motivation",
+      "M6": "background",
+      "M7": "definition",
+      "M8": "definition",
+      "M9": "background",
+      "M10": "state_of_the_art",
+      "M11": "definition",
+      "M12": "definition",
+      "M13": "motivation",
+      "M14": "motivation"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Nothing shown says Pinolo is built on SQLancer's codebase."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "Approximate query synthesis relaxes equivalence to containment, which the paper presents as an alternative to the metamorphic approaches rather than a generalisation of one."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M13",
+          "M14"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M13",
+            "sentence": "For example, the logical bug shown in Figure 1 can not be revealed by NOREC [34] and TLP [35], as the transformations preserve all the operators and the functions, still triggering the buggy evaluation process.",
+            "section": "2.2 Logical Bugs in DBMSs",
+            "page": 4
+          },
+          {
+            "mention_id": "M14",
+            "sentence": "Compared with the existing techniques [ 34–36],PINOLO considers more SQL features, such as set operators, arithmetic expressions, sub-queries, etc.",
+            "section": "3.5 Summary",
+            "page": 8
+          }
+        ],
+        "reasoning": "M13 shows a bug NoREC and TLP cannot reveal and M14 states Pinolo covers more SQL features than those techniques, so they are the approaches it measures itself against.",
+        "techniques": [
+          "pqs",
+          "norec",
+          "tlp"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "yes",
+        "mention_ids": [
+          "M10"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M10",
+            "sentence": "Notably, the metamorphic testing based approach has been recognized to be state-of-the-art in DBMS testing for logical bug detection [ 35,37].",
+            "section": "2.2 Logical Bugs in DBMSs",
+            "page": 4
+          }
+        ],
+        "reasoning": "M10 states the metamorphic testing based approach, citing TLP, has been recognised as state-of-the-art in DBMS testing for logical bug detection."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.json b/_data/papers/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.json
new file mode 100644
index 0000000..8663d6b
--- /dev/null
+++ b/_data/papers/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.json
@@ -0,0 +1,563 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-07T17:07:44Z",
+  "paper": {
+    "id": "paper:s2:d88db0b52f31ed5444602a67e9a275faf57e15bf",
+    "title": "TCP-Fuzz: Detecting Memory and Semantic Bugs in TCP Stacks with Fuzzing",
+    "authors": [
+      "Yong-Hao Zou",
+      "Jia-Ju Bai",
+      "Jielong Zhou",
+      "Jianfeng Tan",
+      "Chenggang Qin",
+      "Shih-Min Hu"
+    ],
+    "year": 2021,
+    "venue": "USENIX Annual Technical Conference",
+    "doi": null,
+    "arxiv_id": null,
+    "s2_paper_id": "d88db0b52f31ed5444602a67e9a275faf57e15bf",
+    "url": "https://www.semanticscholar.org/paper/d88db0b52f31ed5444602a67e9a275faf57e15bf",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "usenix",
+      "url": "https://www.usenix.org/system/files/atc21-zou.pdf",
+      "retrieved_at": "2026-09-07T17:07:44Z",
+      "chars": 75885,
+      "content_sha256": "sha256:636b4ac7ee9022a1fefa63c0148dd8f7af1d3eec1e8bd4532d2f97b3b66fc43d"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 16,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 2261
+      },
+      {
+        "number": "2",
+        "title": "Background and Motivation",
+        "start": 9796
+      },
+      {
+        "number": "2.1",
+        "title": "TCP Stack",
+        "start": 9996
+      },
+      {
+        "number": "2.2",
+        "title": "Study of TCP Stack Commits",
+        "start": 14908
+      },
+      {
+        "number": "2.3",
+        "title": "Limitations of Existing Protocol Fuzzing",
+        "start": 17767
+      },
+      {
+        "number": "3",
+        "title": "Key Techniques",
+        "start": 20503
+      },
+      {
+        "number": "3.1",
+        "title": "Dependency-Based Strategy",
+        "start": 20922
+      },
+      {
+        "number": "3.2",
+        "title": "Transition-Guided Fuzzing Approach",
+        "start": 25681
+      },
+      {
+        "number": "3.3",
+        "title": "Differential Checker",
+        "start": 30064
+      },
+      {
+        "number": "4",
+        "title": "Framework",
+        "start": 33080
+      },
+      {
+        "number": "5",
+        "title": "Evaluation",
+        "start": 36735
+      },
+      {
+        "number": "5.1",
+        "title": "Experimental Setup",
+        "start": 36748
+      },
+      {
+        "number": "5.2",
+        "title": "Runtime Testing",
+        "start": 38746
+      },
+      {
+        "number": "5.3",
+        "title": "Influences of the Found Bugs",
+        "start": 45031
+      },
+      {
+        "number": "5.4",
+        "title": "Comparison to Existing Fuzzing Tools",
+        "start": 47831
+      },
+      {
+        "number": "6",
+        "title": "Discussion",
+        "start": 53399
+      },
+      {
+        "number": "7",
+        "title": "Related Work",
+        "start": 55737
+      },
+      {
+        "number": "7.1",
+        "title": "Network Protocol Fuzzing",
+        "start": 55752
+      },
+      {
+        "number": "7.2",
+        "title": "TCP Stack Checking",
+        "start": 57598
+      },
+      {
+        "number": "7.3",
+        "title": "Differential Testing",
+        "start": 60381
+      },
+      {
+        "number": "8",
+        "title": "Conclusion",
+        "start": 61239
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "American Fuzzy Lop. http://lcamtuf.coredump.cx/afl/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "ASan: address sanitizer. https://github.com/google/sanitizers/wiki/AddressSanitizer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Steve Bishop, Matthew Fairbairn, Hannes Mehnert, Michael Norrish, Tom Ridge, Peter Sewell, Michael Smith, and Keith Wansbrough. Engineering with logic: rigorous test oracle specification and validation for TCP/IP and the Sockets API. Journal of the ACM, 66(1):1:1–1:77, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Steve Bishop, Matthew Fairbairn, Michael Norrish, Peter Sewell, Michael Smith, and Keith Wansbrough. Rigorous specification and conformance testing techniques for network protocols, as applied to TCP, UDP, and sockets. In Proceedings of the ACMSIGCOMM 2005 Conference, pages 265–276, 2005.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Steve Bishop, Matthew Fairbairn, Michael Norrish, Peter Sewell, Michael Smith, and Keith Wansbrough. Engineering with logic: HOL specification and symbolicevaluation testing for TCP implementations. In Proceedings of the 33rd International Symposium on Principles of Programming Languages (POPL), pages 55–66, 2006.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Boofuzz: network protocol fuzzing for humans. https://github.com/jtpereyda/boofuzz.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Yue Cao, Zhiyun Qian, Zhongjie Wang, Tuan Dao, Srikanth V. Krishnamurthy, and Lisa M. Marvel. Offpath TCP exploits: global rate limit considered dangerous. In Proceedings of the 25th USENIX Security Symposium, pages 209–225, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Neal Cardwell, Yuchung Cheng, Lawrence Brakmo, Matt Mathis, Barath Raghavan, Nandita Dukkipati, Hsiao-Keng Jerry Chu, Andreas Terzis, and Tom Herbert. packetdrill: scriptable network stack testing, from sockets to packets. In Proceedings of the 2013 USENIX Annual Technical Conference (ATC), pages 213–218, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Qi Alfred Chen, Zhiyun Qian, Yunhan Jack Jia, Yuru Shao, and Zhuoqing Morley Mao. Static detection of packet injection vulnerabilities: a case for identifying attacker-controlled implicit information leaks. In Proceedings of the 22nd International Conference on Computer and Communications Security (CCS), pages 388– 400, 2015. 172 2021 USENIX Annual Technical Conference USENIX Association",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Weiteng Chen and Zhiyun Qian. Off-path TCP exploit: how wireless routers can jeopardize your secrets. In Proceedings of the 27th USENIX Security Symposium, pages 1581–1598, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Yuting Chen, Ting Su, and Zhendong Su. Deep differential testing of JVM implementations. In Proceedings of the 41st International Conference on Software Engineering (ICSE), pages 1257–1268, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Yuting Chen, Ting Su, Chengnian Sun, Zhendong Su, and Jianjun Zhao. Coverage-directed differential testing of JVM implementations. In Proceedings of the 37th International Conference on Programming Language Design and Implementation (PLDI), pages 85–99, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Clang: a LLVM-based compiler for C/C++ program. https://clang.llvm.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "FreeBSD commit bc35229fad1f: add PAWS check for ACK segments in syncache code. https://gitlab.com/ FreeBSD/freebsd-src/commit/bc35229fad1f.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "CVE-2019-11478. https://nvd.nist.gov/vuln/detail/CVE2019-11478.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Aled Edwards and Steve Muir. Experiences implementing a high performance TCP in user-space. ACMSIGCOMM Computer Communication Review, 25:196–205, 1995.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Paul Fiterau-Brostean, Bengt Jonsson, Robert Merget, Joeri de Ruiter, Konstantinos Sagonas, and Juraj Somorovsky. Analysis of DTLS implementations using protocol state fuzzing. In Proceedings of the 29th USENIX Security Symposium, pages 2523–2540, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Pedro Fonseca, Cheng Li, and Rodrigo Rodrigues. Finding complex concurrency bugs in large multi-threaded applications. In Proceedings of the 6th European Conference on Computer Systems (EuroSys), pages 215–228, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "F-Stack: high performance network framework based on DPDK. http://www.f-stack.org.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Fuzzotron: a network fuzzer supporting TCP, UDP an multithreading. https://github.com/denandz/fuzzotron.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "Hugo Gascon, Christian Wressnegger, Fabian Yamaguchi, Daniel Arp, and Konrad Rieck. Pulsar: stateful black-box fuzzing of proprietary network protocols. In Proceedings of the 11th International Conference on Security and Privacy in Communication Systems, Security and Privacy in Communication Networks, pages 330–347, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Serge Gorbunov and Arnold Rosenbloom. Autofuzz: automated network protocol fuzzing framework. International Journal of Computer Science and Network Security (IJCSNS), 10(8):239, 2010.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Jianmin Guo, Yu Jiang, Yue Zhao, Quan Chen, and Jiaguang Sun. Dlfuzz: differential fuzzing testing of deep learning systems. In Proceedings of the 2018 International Symposium on Foundations of Software Engineering (FSE), pages 739–743, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Endadul Hoque, Omar Chowdhury, Sze Yiu Chau, Cristina Nita-Rotaru, and Ninghui Li. Analyzing operational behavior of stateful protocol implementations for detecting semantic bugs. In Proceedings of the 47th International Conference on Dependable Systems and Networks (DSN), pages 627–638, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Dae R Jeong, Kyungtae Kim, Basavesh Shivakumar, Byoungyoung Lee, and Insik Shin. Razzer: finding kernel race bugs through fuzzing. In Proceedings of the 40th IEEE Symposium on Security and Privacy, pages 754–768, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "EunYoung Jeong, Shinae Wood, Muhammad Jamshed, Haewon Jeong, Sunghwan Ihm, Dongsu Han, and KyoungSoo Park. mTCP: a highly scalable user-level TCP stack for multicore systems. In Proceedings of the 11th USENIX Symposium on Networked Systems Design and Implementation (NSDI), pages 489–502, 2014.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 27,
+      "text": "Samuel Jero, Md. Endadul Hoque, David R. Choffnes, Alan Mislove, and Cristina Nita-Rotaru. Automated attack discovery in TCP congestion control using a modelguided approach. In Proceedings of the 25th Annual Network and Distributed System Security Symposium (NDSS), 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Samuel Jero, Hyojeong Lee, and Cristina Nita-Rotaru. Leveraging state information for automated attack discovery in transport protocol implementations. In Proceedings of the 45th International Conference on Dependable Systems and Networks (DSN), pages 1–12, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "Seulbae Kim, Meng Xu, Sanidhya Kashyap, Jungyeon Yoon, Wen Xu, and Taesoo Kim. Finding semantic bugs in file systems with an extensible fuzzing framework. InProceedings of the 27th International Symposium on Operating Systems Principles (SOSP), pages 147–161, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Nupur Kothari, Ratul Mahajan, Todd Millstein, Ramesh Govindan, and Madanlal Musuvathi. Finding protocol manipulation attacks. In Proceedings of the ACMSIGCOMM 2011 Conference, pages 26–37, 2011. USENIX Association 2021 USENIX Annual Technical Conference 173",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Guangpu Li, Shan Lu, Madanlal Musuvathi, Suman Nath, and Rohan Padhye. Efficient scalable threadsafety-violation detection: finding thousands of concurrency bugs during testing. In Proceedings of the 27th International Symposium on Operating Systems Principles (SOSP), pages 162–180, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "libFuzzer: a library for coverage-guided fuzz testing. https://llvm.org/docs/LibFuzzer.html.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "LLVM compiler infrastructure. https://llvm.org/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "Lars Lockefeer, David M. Williams, and Wan J. Fokkink. Formal specification and verification of TCP extended with the window scale option. Science of Computer Programming (SCP), 118:3–23, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "LTSmin: model checking and minimization of labelled transition systems. https://ltsmin.utwente.nl/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Matthew Luckie, Robert Beverly, Tiange Wu, Mark Allman, and kc claffy. Resilience of deployed TCP to blind attacks. In Proceedings of the 2015 Internet Measurement Conference (IMC), pages 13–26, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Zhengxiong Luo, Feilong Zuo, Yu Jiang, Jian Gao, Xun Jiao, and Jiaguang Sun. Polar: function code aware fuzz testing of ICS protocol. ACM Transactions on Embedded Computing Systems, 18(5s):93:1–93:22, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Zhengxiong Luo, Feilong Zuo, Yuheng Shen, Xun Jiao, Wanli Chang, and Yu Jiang. ICS protocol fuzzing: coverage guided packet crack and generation. In Proceedings of the 57th International Design Automation Conference (DAC), pages 1–6, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "MSan: memory sanitizer. https://github.com/google/sanitizers/wiki/MemorySanitizer.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Sandra L. Murphy and A. Udaya Shankar. Service specification and protocol construction for the transport layer. In Proceedings of the ACMSIGCOMM 1988 Conference, pages 88–97, 1988.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Madanlal Musuvathi and Dawson R. Engler. Model checking large network protocol implementations. In Proceedings of 1st USENIX Symposium on Networked Systems Design and Implementation (NSDI), pages 155– 168, 2004.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Yannic Noller, Corina S. P ˘as˘areanu, Marcel Böhme, Youcheng Sun, Hoang Lam Nguyen, and Lars Grunske. HyDiff: hybrid differential software analysis. In Proceedings of the 42nd International Conference on Software Engineering (ICSE), pages 1273–1285, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 43,
+      "text": "Van-Thuan Pham, Marcel Böhme, and Abhik Roychoudhury. AFLNet: a greybox fuzzer for network protocols.InProceedings of the 13th International Conference on Software Testing, Validation and Verification (ICST), pages 460–465, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 44,
+      "text": "Lei Qian and Brian E. Carpenter. A flow-based performance analysis of TCP and TCP applications. In Proceedings of the 18th International Conference on Networks (ICON), pages 41–45, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Zhiyun Qian, Z. Morley Mao, and Yinglian Xie. Collaborative TCP sequence number inference attack: how to crack sequence number under a second. In Proceedings of the 19th International Conference on Computer and Communications Security (CCS), pages 593– 604, 2012.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "QUIC: a multiplexed stream transport over UDP. https://www.chromium.org/quic.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Gaganjeet Singh Reen and Christian Rossow. DPIFuzz: a differential fuzzing framework to detect DPI elusion strategies for QUIC. In Proceedings of the 36th Annual Computer Security Applications Conference (ACSAC), pages 332–344, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "RFC 5961: improving TCP’s robustness to blind inwindow attacks. https://tools.ietf.org/html/rfc5961.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "RFC 7323: TCP extensions for high performance. https://tools.ietf.org/html/rfc7323.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "RFC 793: TCP (Transmission Control Protocol). https://tools.ietf.org/html/rfc793.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In Proceedings of the 14th International Symposium on Operating Systems Design and Implementation (OSDI), pages 667–682, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 52,
+      "text": "Joeri de Ruiter and Erik Poll. Protocol state fuzzing of TLS implementations. In Proceedings of the 24th USENIX Security Symposium, pages 193–206, 2015.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Mark Anthony Shawn Smith. Formal verification of TCP and T/TCP. PhD thesis, Massachusetts Institute of Technology, 1997.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Juraj Somorovsky. Systematic fuzzing and testing of TLS libraries. In Proceedings of the 23rd International Conference on Computer and Communications Security (CCS), pages 1492–1504, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Dokyung Song, Felicitas Hetzelt, Jonghwan Kim, Brent Byunghoon Kang, Jean-Pierre Seifert, and Michael Franz. Agamotto: accelerating kernel driver fuzzing with lightweight virtual machine checkpoints. InProceedings of the 29th USENIX Security Symposium, pages 2541–2557, 2020. 174 2021 USENIX Annual Technical Conference USENIX Association",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Wei Sun, Lisong Xu, and Sebastian Elbaum. Scalably testing congestion control algorithms of real-world TCP implementations. In Proceedings of the 2018 International Conference on Communications (ICC), pages 1–7, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 57,
+      "text": "Syzkaller: an unsupervised coverage-guided kernel fuzzer. https://github.com/google/syzkaller.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 58,
+      "text": "Possible TCP options. https://www.iana.org/assignments/tcp-parameters/tcp-parameters.xhtml.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 59,
+      "text": "TLDK: Transport Layer Development Kit in network. https://github.com/FDio/tldk/.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 60,
+      "text": "Andreas Walz and Axel Sikora. Exploiting dissent: towards fuzzing-based differential black-box testing of TLS implementations. IEEE Transactions Dependable Secure Computing (TDSC), 17(2):278–291, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 61,
+      "text": "Zhongjie Wang, Shitong Zhu, Yue Cao, Zhiyun Qian, Chengyu Song, Srikanth V. Krishnamurthy, Kevin S. Chan, and Tracy D. Braun. SymTCP: eluding stateful deep packet inspection with automated discrepancy discovery. In Proceedings of the 27th Annual Network and Distributed System Security Symposium (NDSS), 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 62,
+      "text": "Paul Watson. Slipping in the window: TCP reset attacks. Technical Whitepaper, 2004.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 63,
+      "text": "Shinae Woo, Eunyoung Jeong, Shinjo Park, Jongmin Lee, Sunghwan Ihm, and KyoungSoo Park. Comparison of caching strategies in modern cellular backhaul networks. In Proceeding of the 11th International Conference on Mobile Systems, Applications and Services (MobiSys), pages 319–332, 2013.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 64,
+      "text": "Wen Xu, Hyungon Moon, Sanidhya Kashyap, Po-Ning Tseng, and Taesoo Kim. Fuzzing file systems via twodimensional input space exploration. In Proceedings of the 40th IEEE Symposium on Security and Privacy, pages 818–834, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 65,
+      "text": "Yibiao Yang, Yuming Zhou, Hao Sun, Zhendong Su, Zhiqiang Zuo, Lei Xu, and Baowen Xu. Hunting for bugs in code coverage tools via randomized differential testing. In Proceedings of the 41st International Conference on Software Engineering (ICSE), pages 488–499, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 66,
+      "text": "Yanyan Zhuang, Eleni Gessiou, Steven Portzer, Fraida Fund, Monzur Muhammad, Ivan Beschastnikh, and Justin Cappos. Netcheck: network diagnoses from blackbox traces. In Proceedings of the 11th USENIX Symposium on Networked Systems Design and Implementation (NSDI), pages 115–128, 2014. USENIX Association 2021 USENIX Annual Technical Conference 175",
+      "is_sqlancer_publication": false
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 51,
+      "text": "Manuel Rigger and Zhendong Su. Testing database engines via pivoted query synthesis. In Proceedings of the 14th International Symposium on Operating Systems Design and Implementation (OSDI), pages 667–682, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Similar to SQLancer [51] and libFuzzer [32], for inconsistencies that we identify as semantic bugs, we manuallyStackTesting coverage Found bugs Branch Transition Memory / Semantic Confirmed / Fixed TLDK 1.",
+      "context_before": "and 48 semantic bugs. We reported these bugs to related developers, and 40 of them have been confirmed. We are still waiting for responses for the remaining bugs (for example, the mTCP code in github has not been updated for a long time, and thus we have not received any response to our reported bugs in mTCP). Besides, 23 of the confirmed bugs have been fixed. Output inconsistencies. TCP-Fuzz reports 15.1K inconsistencies between the five tested TCP stacks, and we analyze their root causes to identify semantic bugs, through our manual review of RFC documents and observation of TCP stack execution.",
+      "context_after": "3K 329.4K 2 / 26 28 / 19 F-Stack 7.5K 46.8K 1 / 6 6 / 1 mTCP 1.2K 47.9K 5 / 9 0 / 0 FreeBSD - - 0 / 6 5 / 2 Linux - - 0 / 1 1 / 1 Total 10.0K 424.1K 8 / 48 40 / 23 Table 5: Results of fuzzing TCP stacks. Branch transitions Branches (a) TLDK (b) F-Stack (c)mTCPTime (h) Time (h) Time (h) 0K1K2K3K4K5K 0K100K200K300K400K 0 12 24 36 48 0K5K10K15K20K 0K10K20K30K40K50K 0 12 24 36 48 0K1K2K3K4K 0K10K20K30K40K50K 0 12 24 36 48 Figure 11: Covered branches and branch transitions. fix them using the developers’ patches or by ourselves, to reduce related inconsistencies. We iteratively repeat this process u",
+      "section": "5.2 Runtime Testing",
+      "page": 9,
+      "char_offset": 40210,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {},
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T02:16:45Z",
+    "is_model_written": true,
+    "summary": "TCP-Fuzz detects memory and semantic bugs in TCP stacks. It combines dependency-based fuzzing to generate packet and system-call sequences with a differential checker that compares the behaviour of several stacks on the same input, so semantic bugs surface without a specification.",
+    "narrative": "SQLancer is cited once for its practice rather than its technique: when an inconsistency is identified as a semantic bug, the authors follow SQLancer and libFuzzer in fixing it manually so related inconsistencies stop recurring during the campaign.",
+    "roles": {
+      "M1": "incidental"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is cited, not used; nothing in the mentions describes reusing its code."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No technique is extended; the citation is background."
+      },
+      "compares_with": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "No empirical comparison against SQLancer is reported in the mentions."
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "The citation does not characterise SQLancer as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_data/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json b/_data/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json
new file mode 100644
index 0000000..1457a69
--- /dev/null
+++ b/_data/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.json
@@ -0,0 +1,1004 @@
+{
+  "schema_version": "1.0.0",
+  "generated_at": "2026-09-09T01:32:44Z",
+  "paper": {
+    "id": "paper:s2:d933042eb3c2a8f2e208515490d6b6a400e00fec",
+    "title": "Towards Generic Database Management System Fuzzing",
+    "authors": [
+      "Yupeng Yang",
+      "Yongheng Chen",
+      "Rui Zhong",
+      "Jizhou Chen",
+      "Wenke Lee"
+    ],
+    "year": 2024,
+    "venue": "USENIX Security Symposium",
+    "doi": null,
+    "arxiv_id": null,
+    "s2_paper_id": "d933042eb3c2a8f2e208515490d6b6a400e00fec",
+    "url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+    "also_indexed_as": []
+  },
+  "sources": [
+    {
+      "kind": "fulltext",
+      "route": "supplied_pdf",
+      "url": "https://www.semanticscholar.org/paper/d933042eb3c2a8f2e208515490d6b6a400e00fec",
+      "retrieved_at": "2026-09-09T01:32:44Z",
+      "chars": 93200,
+      "content_sha256": "sha256:388e6fee4d93e389eabb23096f26b6eebd3b10ddb51d99f2c1c8375f0c9dd6a2"
+    }
+  ],
+  "document": {
+    "has_fulltext": true,
+    "page_count": 19,
+    "has_outline": true,
+    "sections": [
+      {
+        "number": "1",
+        "title": "Introduction",
+        "start": 1871
+      },
+      {
+        "number": "2.1",
+        "title": "Diverse DBMS Interfaces",
+        "start": 8174
+      },
+      {
+        "number": "2.2",
+        "title": "Existing Challenges and Limitations",
+        "start": 9947
+      },
+      {
+        "number": "2.3",
+        "title": "Our Insights and Solutions",
+        "start": 19256
+      },
+      {
+        "number": "5.1",
+        "title": "Annotation System",
+        "start": 27399
+      },
+      {
+        "number": "5.2",
+        "title": "Semantic Analysis and Validation",
+        "start": 39263
+      },
+      {
+        "number": "8.1",
+        "title": "Environment Setup",
+        "start": 46716
+      },
+      {
+        "number": "8.2",
+        "title": "Generalizability and Portability",
+        "start": 49392
+      },
+      {
+        "number": "8.3",
+        "title": "Real-world Bug Hunting",
+        "start": 51219
+      },
+      {
+        "number": "8.4",
+        "title": "Contributions of the Solutions",
+        "start": 54864
+      },
+      {
+        "number": "8.5",
+        "title": "Comparison with Existing Tools",
+        "start": 63968
+      },
+      {
+        "number": "9.1",
+        "title": "Complex Semantics and Completeness",
+        "start": 68022
+      },
+      {
+        "number": "9.2",
+        "title": "Common Database Interface",
+        "start": 69319
+      },
+      {
+        "number": "10.1",
+        "title": "General Fuzzing Strategies",
+        "start": 70288
+      },
+      {
+        "number": "10.2",
+        "title": "Language Processors Fuzzing",
+        "start": 71663
+      },
+      {
+        "number": "10.3",
+        "title": "DBMSs Fuzzing",
+        "start": 72602
+      }
+    ]
+  },
+  "references": [
+    {
+      "number": 1,
+      "text": "ANTLR. Grammars written for ANTLR v4. https://github.com/ antlr/grammars-v4, 2020. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 2,
+      "text": "ArangoDB Inc. ArangoDB. https://www.arangodb.com/, 2015. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 3,
+      "text": "Cornelius Aschermann, Tommaso Frassetto, Thorsten Holz, Patrick Jauernig, Ahmad-Reza Sadeghi, and Daniel Teuchert. NAUTILUS: Fishing for Deep Bugs with Grammars. In 26th Annual Network and Distributed System Security Symposium, NDSS 2019, San Diego, California, USA, February 24-27, 2019. The Internet Society, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 4,
+      "text": "Cornelius Aschermann, Sergej Schumilo, Tim Blazytko, Robert Gawlik, and Thorsten Holz. REDQUEEN: Fuzzing with Input-to-State Correspondence. In 26th Annual Network and Distributed System Security Symposium, NDSS 2019, San Diego, California, USA, February 24-27, 2019. The Internet Society, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 5,
+      "text": "Bitnine Global Inc. AgensGraph. https://bitnine.net/, 2017. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 6,
+      "text": "Tim Blazytko, Cornelius Aschermann, Moritz Schlögel, Ali Abbasi, Sergej Schumilo, Simon Wörner, and Thorsten Holz. GRIMOIRE: Synthesizing Structure while Fuzzing. In Nadia Heninger and Patrick Traynor, editors, 28th USENIX Security Symposium, USENIX Security 2019, Santa Clara, CA, USA, August 14-16, 2019, pages 1985–2002. USENIX Association, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 7,
+      "text": "Marcel Böhme, Van-Thuan Pham, and Abhik Roychoudhury. Coveragebased Greybox Fuzzing as Markov Chain. In Edgar R. Weippl, Stefan Katzenbeisser, Christopher Kruegel, Andrew C. Myers, and Shai Halevi, editors, Proceedings of the 2016 ACMSIGSAC Conference on Computer and Communications Security, Vienna, Austria, October 24-28, 2016, pages 1032–1043. ACM, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 8,
+      "text": "Surajit Chaudhuri. An Overview of Query Optimization in Relational Systems. In Alberto O. Mendelzon and Jan Paredaens, editors, Proceedings of the Seventeenth ACMSIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems, June 1-3, 1998, Seattle, Washington, USA, pages 34–43. ACM Press, 1998.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 9,
+      "text": "Yongheng Chen, Rui Zhong, Hong Hu, Hangfan Zhang, Yupeng Yang, Dinghao Wu, and Wenke Lee. One Engine to Fuzz ’em All: Generic Language Processor Testing with Semantic Validation. In 42nd IEEE Symposium on Security and Privacy, SP 2021, San Francisco, CA, USA, 24-27 May 2021, pages 642–658. IEEE, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 10,
+      "text": "Nicolas Coppik, Oliver Schwahn, and Neeraj Suri. MemFuzz: Using Memory Accesses to Guide Fuzzing. In 12th IEEE Conference on Software Testing, Validation and Verification, ICST 2019, Xi’an, China, April 22-27, 2019, pages 48–58. IEEE, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 11,
+      "text": "Microsoft Corporation. Microsoft SQL Server. https://docs. microsoft.com/en-us/sql/, 1989. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 12,
+      "text": "Oracle Corporation. Oracle Database. https://www.oracle.com/ database/technologies/, 1979. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 13,
+      "text": "Ali Davoudian, Liu Chen, and Mengchi Liu. A Survey on NoSQL Stores. ACM Comput. Surv., 51(2):40:1–40:43, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 14,
+      "text": "Andrea Fioraldi, Dominik Christian Maier, Heiko Eißfeldt, and Marc Heuse. AFL++: Combining Incremental Steps of Fuzzing Research. In Yuval Yarom and Sarah Zennou, editors, 14th USENIX Workshop on Offensive Technologies, WOOT 2020, August 11, 2020. USENIX Association, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 15,
+      "text": "Jingzhou Fu, Jie Liang, Zhiyong Wu, Mingzhe Wang, and Yu Jiang. Griffin: Grammar-Free DBMS Fuzzing. In 37th IEEE/ACM International Conference on Automated Software Engineering, ASE 2022, Rochester, MI, USA, October 10-14, 2022, pages 49:1–49:12. ACM, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 16,
+      "text": "Felix Gessert, Wolfram Wingerath, Steffen Friedrich, and Norbert Ritter. NoSQL database systems: a survey and decision guidance. Comput. Sci. Res. Dev., 32(3-4):353–365, 2017.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 17,
+      "text": "Patrice Godefroid, Adam Kiezun, and Michael Y. Levin. Grammar-based Whitebox Fuzzing. In Rajiv Gupta and Saman P. Amarasinghe, editors, Proceedings of the ACMSIGPLAN 2008 Conference on Programming Language Design and Implementation, Tucson, AZ, USA, June 7-13, 2008, pages 206–215. ACM, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 18,
+      "text": "Google. Honggfuzz. https://google.github.io/honggfuzz/, 2016. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 19,
+      "text": "Rahul Gopinath, Björn Mathis, Matthias Höschele, Alexander Kampmann, and Andreas Zeller. Sample-Free Learning of Input Grammars for Comprehensive Software Fuzzing. CoRR, abs/1810.08289, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 20,
+      "text": "Samuel Groß, Simon Koch, Lukas Bernhard, Thorsten Holz, and Martin Johns. FUZZILLI: Fuzzing for JavaScript JIT Compiler Vulnerabilities. In30th Annual Network and Distributed System Security Symposium, NDSS 2023, San Diego, California, USA, February 27 - March 3, 2023. The Internet Society, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 21,
+      "text": "HyungSeok Han, DongHyeon Oh, and Sang Kil Cha. CodeAlchemist: Semantics-Aware Code Generation to Find Vulnerabilities in JavaScript Engines. In 26th Annual Network and Distributed System Security Symposium, NDSS 2019, San Diego, California, USA, February 24-27, 2019. The Internet Society, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 22,
+      "text": "Renáta Hodován, Ákos Kiss, and Tibor Gyimóthy. Grammarinator: a grammar-based open source fuzzer. In Wishnu Prasetya, Tanja E. J. V os, and Sinem Getir, editors, Proceedings of the 9th ACMSIGSOFT International Workshop on Automating TEST Case Design, Selection, and Evaluation, A-TEST@SIGSOFT FSE 2018, Lake Buena Vista, FL, USA, November 05, 2018, pages 45–48. ACM, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 23,
+      "text": "Matthias Höschele and Andreas Zeller. Mining Input Grammars from Dynamic Taints. In David Lo, Sven Apel, and Sarfraz Khurshid, editors, Proceedings of the 31st IEEE/ACM International Conference on Automated Software Engineering, ASE 2016, Singapore, September 3-7, 2016, pages 720–725. ACM, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 24,
+      "text": "Bo Jiang, Ye Liu, and W. K. Chan. ContractFuzzer: fuzzing smart contracts for vulnerability detection. In Marianne Huchard, Christian Kästner, and Gordon Fraser, editors, Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering, ASE 2018, Montpellier, France, September 3-7, 2018, pages 259–269. ACM, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 25,
+      "text": "Zu-Ming Jiang, Jia-Ju Bai, and Zhendong Su. DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation. In Joseph A. Calandrino and Carmela Troncoso, editors, 32nd USENIX Security Symposium, USENIX Security 2023, Anaheim, CA, USA, August 9-11, 2023, pages 4949–4965. USENIX Association, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 26,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. Testing Graph Database Engines via Query Partitioning. In René Just and Gordon Fraser, editors, Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, Seattle, WA, USA, July 17-21, 2023, pages 140–149. ACM, 2023.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 27,
+      "text": "Jie Liang, Yaoguang Chen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, Yu Jiang, Xiangdong Huang, Ting Chen, Jiashui Wang, and Jiajia Li. Sequence-Oriented DBMS Fuzzing. In 39th IEEE International Conference on Data Engineering, ICDE 2023, Anaheim, CA, USA, April 3-7, 2023, pages 668–681. IEEE, 2023.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 28,
+      "text": "Yu Liang, Song Liu, and Hong Hu. Detecting Logical Bugs of DBMS with Coverage-based Guidance. In Kevin R. B. Butler and Kurt Thomas, editors, 31st USENIX Security Symposium, USENIX Security 2022, Boston, MA, USA, August 10-12, 2022, pages 4309–4326. USENIX Association, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 29,
+      "text": "libFuzzer. libFuzzera library for coverage-guided fuzz testing. https: //llvm.org/docs/LibFuzzer.html, 2015. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 30,
+      "text": "Chenyang Lyu, Shouling Ji, Chao Zhang, Yuwei Li, Wei-Han Lee, Yu Song, and Raheem Beyah. MOPT: Optimized Mutation Scheduling for Fuzzers. In Nadia Heninger and Patrick Traynor, editors, 28th USENIX Security Symposium, USENIX Security 2019, Santa Clara, CA, USA, August 14-16, 2019, pages 1949–1966. USENIX Association, USENIX Association 33rd USENIX Security Symposium 915 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 31,
+      "text": "Chenyang Lyu, Shouling Ji, Xuhong Zhang, Hong Liang, Binbin Zhao, Kangjie Lu, and Raheem Beyah. EMS: History-Driven Mutation for Coverage-based Fuzzing. In 29th Annual Network and Distributed System Security Symposium, NDSS 2022, San Diego, California, USA, April 24-28, 2022. The Internet Society, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 32,
+      "text": "Alessandro Mantovani, Andrea Fioraldi, and Davide Balzarotti. Fuzzing with Data Dependency Information. In 7th IEEE European Symposium on Security and Privacy, EuroS&P 2022, Genoa, Italy, June 6-10, 2022, pages 286–302. IEEE, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 33,
+      "text": "MongoDB Inc. MongoDB. https://www.mongodb.com/, 2007. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 34,
+      "text": "MozillaSecurity. funfuzz. https://github.com/ MozillaSecurity/funfuzz, 2020. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 35,
+      "text": "Neo4j and openCypher. Cypher Query Language. https:// opencypher.org/, 2015. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 36,
+      "text": "Oracle Corporation. MySQL. https://dev.mysql.com/, 1995. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 37,
+      "text": "Soyeon Park, Wen Xu, Insu Yun, Daehee Jang, and Taesoo Kim. Fuzzing JavaScript Engines with Aspect-preserving Mutation. In CA, USA, May 18-21, 2020, pages 1629–1642. IEEE, 2020.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 38,
+      "text": "Van-Thuan Pham, Marcel Böhme, and Abhik Roychoudhury. Modelbased Whitebox Fuzzing for Program Binaries. In David Lo, Sven Apel, and Sarfraz Khurshid, editors, Proceedings of the 31st IEEE/ACM International Conference on Automated Software Engineering, ASE 2016, Singapore, September 3-7, 2016, pages 543–553. ACM, 2016.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 39,
+      "text": "PostgreSQL Global Development Group. PostgreSQL. https://www. postgresql.org/, 1986. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 40,
+      "text": "Redis Ltd. Redis. https://redis.io/, 2009. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 41,
+      "text": "Redis Ltd. RedisGraph. https://oss.redislabs.com/ redisgraph/, 2009. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Prem Devanbu, Myra B. Cohen, and Thomas Zimmermann, editors, ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, pages 1140–1152. ACM, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020, pages 667–682. USENIX Association, 2020.",
+      "is_sqlancer_publication": true
+    },
+    {
+      "number": 44,
+      "text": "Andreas Seltenreich, Bo Tang, and Sjoerd Mullender. SQLsmith. https://github.com/anse1/sqlsmith, 2016. Accessed: 202309-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 45,
+      "text": "Snap Inc. KeyDB. https://docs.keydb.dev/, 2019. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 46,
+      "text": "Spandan Veggalam, Sanjay Rawat, István Haller, and Herbert Bos. IFuzzer: An Evolutionary Interpreter Fuzzer Using Genetic Programming. In Ioannis G. Askoxylakis, Sotiris Ioannidis, Sokratis K. Katsikas, and Catherine Meadows, editors, Computer Security-ESORICS 2016 - 21st European Symposium on Research in Computer Security, Heraklion, Greece, September 26-30, 2016, Proceedings, Part I, volume 9878",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 47,
+      "text": "Joachim Viide, Aki Helin, Marko Laakso, Pekka Pietikäinen, Mika Seppänen, Kimmo Halunen, Rauli Puuperä, and Juha Röning. Experiences with Model Inference Assisted Fuzzing. In Dan Boneh, Tal Garfinkel, and Dug Song, editors, 2nd USENIX Workshop on Offensive Technologies, WOOT’08, San Jose, CA, USA, July 28, 2008, Proceedings. USENIX Association, 2008.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 48,
+      "text": "Junjie Wang, Bihuan Chen, Lei Wei, and Yang Liu. Superion: Grammaraware Greybox Fuzzing. In Joanne M. Atlee, Tevfik Bultan, and Jon Whittle, editors, Proceedings of the 41st International Conference on Software Engineering, ICSE 2019, Montreal, QC, Canada, May 25-31, 2019, pages 724–735. IEEE / ACM, 2019.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 49,
+      "text": "Mingzhe Wang, Zhiyong Wu, Xinyi Xu, Jie Liang, Chijin Zhou, Huafeng Zhang, and Yu Jiang. Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing. In 43rd IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice, ICSE (SEIP) 2021, Madrid, Spain, May 25-28, 2021, pages 328–337. IEEE, 2021.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 50,
+      "text": "Peng Xu, Yanhao Wang, Hong Hu, and Purui Su. COOPER: Testing the Binding Code of Scripting Languages with Cooperative Mutation. In29th Annual Network and Distributed System Security Symposium, NDSS 2022, San Diego, California, USA, April 24-28, 2022. The Internet Society, 2022.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 51,
+      "text": "Xuejun Yang, Yang Chen, Eric Eide, and John Regehr. Finding and Understanding Bugs in C Compilers. In Mary W. Hall and David A. Padua, editors, Proceedings of the 32nd ACMSIGPLAN Conference on Programming Language Design and Implementation, PLDI 2011, San Jose, CA, USA, June 4-8, 2011, pages 283–294. ACM, 2011.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 52,
+      "text": "Guixin Ye, Zhanyong Tang, Shin Hwei Tan, Songfang Huang, Dingyi Fang, Xiaoyang Sun, Lizhong Bian, Haibo Wang, and Zheng Wang. Automated Conformance Testing for JavaScript Engines via Deep Compiler Fuzzing. In Stephen N. Freund and Eran Yahav, editors, PLDI ’21: 42nd ACMSIGPLAN International Conference on Programming Language Design and Implementation, Virtual Event, Canada, June 20-25, 2021, pages",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 53,
+      "text": "Insu Yun, Sangho Lee, Meng Xu, Yeongjin Jang, and Taesoo Kim. QSYM: A Practical Concolic Execution Engine Tailored for Hybrid Fuzzing. In William Enck and Adrienne Porter Felt, editors, 27th USENIX Security Symposium, USENIX Security 2018, Baltimore, MD, USA, August 15-17, 2018, pages 745–761. USENIX Association, 2018.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 54,
+      "text": "Michal Zalewski. American Fuzzy Lop (2.52b). http://lcamtuf. coredump.cx/afl, 2019. Accessed: 2023-09-14.",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 55,
+      "text": "Yingying Zheng, Wensheng Dou, Yicheng Wang, Zheng Qin, Lei Tang, Yu Gao, Dong Wang, Wei Wang, and Jun Wei. Finding bugs in Gremlinbased graph database systems via Randomized differential testing. In Sukyoung Ryu and Yannis Smaragdakis, editors, ISSTA ’22: 31st ACMSIGSOFT International Symposium on Software Testing and Analysis, Virtual Event, South Korea, July 18 - 22, 2022, pages 302–313. ACM, 20",
+      "is_sqlancer_publication": false
+    },
+    {
+      "number": 56,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event, USA, November 9-13, 2020, pages 955–970. ACM, 2020. 916 33rd USENIX ",
+      "is_sqlancer_publication": true
+    }
+  ],
+  "sqlancer_references": [
+    {
+      "number": 26,
+      "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. Testing Graph Database Engines via Query Partitioning. In René Just and Gordon Fraser, editors, Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, Seattle, WA, USA, July 17-21, 2023, pages 140–149. ACM, 2023.",
+      "technique": null,
+      "matched_as": "project_authored",
+      "why": [
+        "written by an author of the SQLancer project, but not itself a SQLancer paper"
+      ]
+    },
+    {
+      "number": 42,
+      "text": "Manuel Rigger and Zhendong Su. Detecting Optimization Bugs in Database Engines via Non-optimizing Reference Engine Construction. In Prem Devanbu, Myra B. Cohen, and Thomas Zimmermann, editors, ESEC/FSE ’20: 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, Virtual Event, USA, November 8-13, 2020, pages 1140–1152. ACM, 2020.",
+      "technique": "norec",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing norec",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 43,
+      "text": "Manuel Rigger and Zhendong Su. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020, pages 667–682. USENIX Association, 2020.",
+      "technique": "pqs",
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "cites the paper introducing pqs",
+        "names SQLancer or one of its techniques"
+      ]
+    },
+    {
+      "number": 56,
+      "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communications Security, Virtual Event, USA, November 9-13, 2020, pages 955–970. ACM, 2020. 916 33rd USENIX Security Symposium USENIX Association A Appendix 1// BuzzBee’s IR is a tree structure formed by IRNodes. 2// An IRNode has the following properties: 3IRNode(id, type, text, annotation, children, parent, ...); 4 5// BuzzBee lifts the example shown in Fig.4a into the following 6// IRNodes, forming the IR program shown in Fig.4d: 7node0(id=0, type=testcase, text=NULL, annotation={}, 8 children={&node1}, parent=NULL); 9node1(id=1, type=cmds, text=NULL, annotation={}, 10 children={&node2,&node12,&node22,&node32}, parent=&node0); 11node2(id=2, type=cmd, text=NULL, annotation={}, 12 children={&node3}, parent=&node1); 13node3(id=3, type=hset, text=NULL, annotation={}, 14 children={&node4, &node5, &node7, &node9}, parent=&node2); 15... 16node33(id=33, type=hincrby, text=NULL, annotation={}, 17 children={&node34, &node35, &node37, &node39}, 18 parent=&node32); 19node34(id=34, type=terminal, text=\"HINCRBY\", annotation={}, 20 children={}, parent=&node33); 21node35(id=35, type=key, text=NULL, 22 annotation={\"default\":{operation:Use, 23 args:{type:\"HSET key\"}}}, 24 children={&node36}, parent=&node33); 25node36(id=36, type=terminal, text=\"k1\", annotation={}, 26 children={}, parent=&node35); 27node37(id=37, type=field, text=NULL, 28 annotation={\"default\":{operation:Define, 29 args:{type:hset_field_type_resolver}}}, 30 children={&node38}, parent=&node33); 31node38(id=38, type=terminal, text=\"k1_field1\", annotation={}, 32 children={}, parent=&node37); 33node39(id=39, type=increment, text=NULL, annotation={}, 34 children={&node40}, parent=&node33); 35node40(id=40, type=terminal, text=\"1\", annotation={}, 36 children={}, parent=&node39); Fig. 8: BUZZBEE’s IR structure and the IR program of the example in Fig.4a. The IR program is a tree structure, which is also illustrated in Fig.4f. When lifting, BUZZBEEparses the test case, traverses over the AST of the test case, collects the corresponding annotations, and then creates the IRNodes forming the IR program. The complete source of the test case is stored in the textproperty of the IRNodes of the type terminal. When compiling, BUZZBEE traverses over the IR program, collects the terminal IRNodes, and concatenates their texts into a test case.1cql: 2 navigator* property; 3 4navigator: 5 ’.parent’ 6 | ’.child’ arg | ’.lsib’ arg | ’.rsib’ arg 7 | ...; 8 9arg: 10 ’(’ num ’)’; 11 12property: 13 ’@text’ | ’@id’ | ’@sym_type’ | ...; Fig.9: Context Query Language (CQL) USENIX Association 33rd USENIX Security Symposium 917 Table 5: Real world targets we test B UZZBEEon.Popularity is the number of Stars of the DBMS’s GitHub repository. X* in Data Model means the DBMS supports other models but we only test the interface for its primary model X. DBMSLOC Popularity Data Model redis 281K 60.1K Key-value* MongoDB 8.7M 23.8K Document* ArangoDB 6.8M 13K Document* PostgreSQL 1.6M 12.5K Relational* MySQL 5.5M 9.6K Relational* KeyDB 348K 7.6K Key-value RedisGraph 1.7M 1.9K Graph AgensGraph 1.4M 1.3K Graph Table 6: DBMSs supported by fuzzers. Fuzzer v.s. / DBMS redis RedisGraph ArangoDB PostgreSQL BUZZBEE ✔ ✔ ✔ ✔ AFL++ ✔ ✔ ✔ ✔ POLYGLOT ✔ ✔ ✔ ✔ REDQUEEN ✔ ✔ ✗ ✗ Grammarinator ✔ ✔ ✔ ✔ SQUIRREL ✗ ✗ ✗ ✔ SQLANCER ✗ ✗ ✔ ✔Table 7: Real-world bugs found by BUZZBEE.BUZZBEEfound 40 bugs in latest DBMSs during evaluation. * means BUZZBEE found the bug in the latest DBMS, but it was already known by the vendors when we reported it. †means the bug is known and not in the latest DBMS. DBMSID Version Status CVE redis1 cb1fff3 Fixed 2 395d801 Fixed* CVE-2022-35977 3 1ec82e6 Fixed CVE-2023-22458 4 7.0.7 Fixed CVE-2023-22458 5 7.0.8 Fixed CVE-2023-25515 6 7.0.8 Fixed CVE-2023-25515 7 7.0.8 Fixed CVE-2023-25515 8 7.0.8 Fixed CVE-2023-28425 9 af0a4fe2 Confirmed 10 7.0.10 Fixed* CVE-2023-28856 11 7.0.8 Fixed* 12 af0a4fe2 Fixed KeyDB13 6.3.2 Fixed 14 6.3.2 Fixed 15 6.3.2 Fixed 16 6.3.2 Fixed 17 6.3.2 Fixed 18 6.3.2 Fixed RedisGraph19 v2.10.8 Fixed 20 c425ad8 Fixed CVE-2023-25310 21 v2.10.9 Confirmed 22 v2.10.9 Fixed 23 v2.10.10 Confirmed 24 v2.10.10 Confirmed 25 v2.10.9 Confirmed 26 v2.10.9 Reported 27 v2.10.9 Reported 28 v2.10.9 Confirmed 29 v2.10.9 Confirmed 30 v2.10.9 Fixed 31 v2.10.9 Fixed* AgensGraph32 v2.13.1 Confirmed 33 v2.13.1 Confirmed MongoDB 34 87d48e1 Fixed ArangoDB35 v3.11.0 Fixed 36 v3.11.0 Fixed PostgreSQL 37 v11.12 Fixed† MySQL38 v8.1.0 Confirmed* 39 v8.1.0 Confirmed* 40 v8.1.0 Confirmed* 41 v8.1.0 Confirmed 918 33rd USENIX Security Symposium USENIX Association",
+      "technique": null,
+      "matched_as": "sqlancer_publication",
+      "why": [
+        "names SQLancer or one of its techniques"
+      ]
+    }
+  ],
+  "mentions": [
+    {
+      "id": "M1",
+      "found_by": "citation_marker",
+      "surface": "[27,28,43,44,49,56]",
+      "technique": "pqs",
+      "sentence": "Fuzzing frameworks and research related to relational DBMSs [27,28,43,44,49,56] have been developed and advanced extensively over the years, contributing to more secure and trustworthy systems in the relational DBMS venue.",
+      "context_before": "ocument DBMS, and graph DBMS have gained widespread adoption more recently due to their flexibility and performance advantages in handling large-scale, unstructured data. Considering the prevalence and criticality of these systems, it is paramount to strengthen the security and robustness of the diverse DBMSs.Fuzzing, an automated software testing method that injects random data as inputs to software, has proven useful in uncovering faults in DBMSs. However, there exists a disparity in the extent of fuzzing efforts directed towards nonrelational DBMSs compared to their relational counterparts.",
+      "context_after": "In contrast, non-relational DBMSs have not experienced the same level of scrutiny. State-of-the-art generic fuzzing frameworks such as [ 4,7,9,14,30,54] all show non-promising fuzzing performances when applied to non-relational DBMSs because they cannot generate test cases that trigger DBMS behaviors effectively. There is a need for an effective solution capable of testing both relational and non-relational DBMSs. However, multiple challenges exist when designing a fuzzer that extends to non-relational DBMSs. First, it is hard to generalize. The interfaces of non-relational DBMSs are diverse,",
+      "section": "1 Introduction",
+      "page": 2,
+      "char_offset": 2981,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020, pages 667–682. USENIX Association, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M2",
+      "found_by": "citation_marker",
+      "surface": "[ 27,28,43,44,49,56]",
+      "technique": "pqs",
+      "sentence": "Current research has made significant advancements in relational DBMS testing [ 27,28,43,44,49,56].",
+      "context_before": "g one command is unlikely to abort the remaining commands, command sequences with semantic errors or inadequate semantic dependency may still trigger fewer effective behaviors. 2.2 Existing Challenges and Limitations In this section, we discuss three limitations of current approaches in fuzzing the diverse DBMS interfaces. Non-generic. As discussed in §2.1, the test case quality is important for DBMS testing. However, the variety of DBMS interfaces complicates the creation of a fuzzing framework that can generate high-quality test cases and remain easily adaptable to different DBMS interfaces.",
+      "context_after": "SQLsmith [44] generates ran902 33rd USENIX Security Symposium USENIX Association 1// Partial grammar rules: 2createtbl_stmt: 3 ’CREATE TABLE’ 4 tbl_name ’(’ 5 ... 6 ’)’; 7 8// The test case: 9> CREATE TABLE t1( 10 c1 date 11); 12 13 14// Bind \"TABLE define\" 15// to `tbl_name `. 16 17// When we traverse to the 18// node `t1`, we know a 19// TABLE t1 is defined. (a)PostgreSQL examples1// Partial grammar rules: 2match_clause: 3 MATCH pattern_part where_part; 4 5pattern_part: 6 node_pattern; 7 8where_part: 9 WHERE node_pattern; 10 11node_pattern: 12 ’(’ identifier ’)’ | ...; 13 14// The test case:",
+      "section": "2.2 Existing Challenges and Limitations",
+      "page": 3,
+      "char_offset": 10370,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020, pages 667–682. USENIX Association, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M3",
+      "found_by": "citation_marker",
+      "surface": "[56]",
+      "technique": null,
+      "sentence": "SQUIRREL [56] and later works [ 27,28] advance by introducing an intermediate representation (IR) that effectively incorporates SQL syntax and semantics, allowing it to adapt to multiple SQL databases.",
+      "context_before": "o\" // stores an ASCII string 2> HSET k2 k2_field1 \"123\" // stores a numeric string 3 4> HINCRBY k1 k1_field1 1 // increase the stored value 5(error) value not an integer 6 7// > DEL k2 8> HINCRBY k2 k2_field1 1 9(integer) 124 (c)redis commands Fig.1: Static Constraint Examples. Example test cases illustrating the problem of static constraints. While using static constraint sufficiently models the semantics in Fig.1a, it does not correctly model the scope constraints in Fig.1b and the type constraints in Fig.1c. dom SQL queries according to predefined schema information about certain SQL DBMSs.",
+      "context_after": "However, these frameworks still require non-trivial adoption efforts to support a new relational DBMS, as mentioned in [ 49]. Moreover, they either have not supported nonrelational DBMS, or inherently cannot support them due to the model design. Considering the wide use of non-relational DBMSs, we lack a generic solution that enables efficient testing for the diverse DBMS interfaces. Static Constraints. Mutation-based fuzzing has proven successful in fuzzing modern software. However, for DBMS fuzzing, mutations can easily break the semantic correctness of the test case. Recent studies [ 27,28",
+      "section": "2.2 Existing Challenges and Limitations",
+      "page": 4,
+      "char_offset": 11830,
+      "cited_reference": {
+        "number": 56,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communic",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M4",
+      "found_by": "citation_marker",
+      "surface": "[ 27,28,56]",
+      "technique": null,
+      "sentence": "Recent studies [ 27,28,56] highlight the significance of maintaining the semantic correctness of the test cases in mutation-based DBMS fuzzing.",
+      "context_before": "multiple SQL databases. However, these frameworks still require non-trivial adoption efforts to support a new relational DBMS, as mentioned in [ 49]. Moreover, they either have not supported nonrelational DBMS, or inherently cannot support them due to the model design. Considering the wide use of non-relational DBMSs, we lack a generic solution that enables efficient testing for the diverse DBMS interfaces. Static Constraints. Mutation-based fuzzing has proven successful in fuzzing modern software. However, for DBMS fuzzing, mutations can easily break the semantic correctness of the test case.",
+      "context_after": "In this work, we use constraints to represent the semantic rules the test case should obey to avoid a DBMS execution error. This includes scope constraints andtype constraints. Scope constraints restrict where a variable is available for use. This is often enforced through data operations such as define, use, and invalidate. Type constraints further restrict the legitimate type of the variable in the data operations. To model the constraints, the approach used by existing fuzzing frameworks [ 9,27,28,56] is HMSET k1 k1_field1 1HMSET k1 k1_field1 1APPEND x 413413HMSET k1 k1_field1 1XGROUP CREA",
+      "section": "2.2 Existing Challenges and Limitations",
+      "page": 4,
+      "char_offset": 12609,
+      "cited_reference": {
+        "number": 56,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communic",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M5",
+      "found_by": "citation_marker",
+      "surface": "[ 9,27,28,56]",
+      "technique": null,
+      "sentence": "To model the constraints, the approach used by existing fuzzing frameworks [ 9,27,28,56] is HMSET k1 k1_field1 1HMSET k1 k1_field1 1APPEND x 413413HMSET k1 k1_field1 1XGROUP CREATs g 0HMSET k1 k1_field1 1HRANDFIELD key1 1.",
+      "context_before": "antic correctness of the test case. Recent studies [ 27,28,56] highlight the significance of maintaining the semantic correctness of the test cases in mutation-based DBMS fuzzing. In this work, we use constraints to represent the semantic rules the test case should obey to avoid a DBMS execution error. This includes scope constraints andtype constraints. Scope constraints restrict where a variable is available for use. This is often enforced through data operations such as define, use, and invalidate. Type constraints further restrict the legitimate type of the variable in the data operations.",
+      "context_after": ".....(>2.6k possibilities) ......Level nMutations(showingonly1outofmanypossibilities) HMSET k1 k1_field1 1HRANDFIELD k1-9223372036854770000withvalues crash!Level1 Mutationsmutatemultiplelevels ......Fig.2: Random Mutation Running Examples. This figure shows a demo fuzzer we made for testing redis. When performing random mutations, the level 1 mutations contain many possibilities that cannot form a data dependency with the initial command, which consequently affects the fuzzing performance at later mutation levels. first to parse the test case using the target’s grammar specification, and then",
+      "section": "2.2 Existing Challenges and Limitations",
+      "page": 4,
+      "char_offset": 13174,
+      "cited_reference": {
+        "number": 56,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communic",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M6",
+      "found_by": "citation_marker",
+      "surface": "[ 27,28,56]",
+      "technique": null,
+      "sentence": "Existing mutation-based DBMS fuzzers [ 27,28,56] perform mutations randomly and rely on code coverage to explore more program behaviors.",
+      "context_before": "te on k1_field1, USENIX Association 33rd USENIX Security Symposium 903 a field pre-defined but storing a value of an ASCII string instead of a numeric string. Adopting the existing approach, we can only let the AST node of k1yield type \"HSET key\", and let the node of k1_field1 yield type \"HSET field\", losing the information that the value implicitly says k1_field1 is an ASCII string instead of a numeric string. Here, using a static type constraint without looking at the values ( i.e., literals\"Hello\" and \"123\" ) loses vital information to model the semantics correctly. Loose Data Dependencies.",
+      "context_after": "However, we observe that random mutations can waste huge efforts in fuzzing non-relational DBMSs, because they tend to generate loose data dependencies that trigger thin and less effective behaviors. The issue becomes significant when the database interface contains many operations that are notdependencyaffiliative, which are common cases in non-relational DBMSs such as redis. We define two operations as dependencyaffiliative when they can form a data dependency. We observe that test cases triggering deep logic require dependencyaffiliative operations often. For instance, a test case containi",
+      "section": "2.2 Existing Challenges and Limitations",
+      "page": 5,
+      "char_offset": 16501,
+      "cited_reference": {
+        "number": 56,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communic",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M7",
+      "found_by": "citation_marker",
+      "surface": "[ 27,28,56]",
+      "technique": null,
+      "sentence": "Inspired by existing works [ 27,28,56], we design an IR to incorporate both the syntactic structures and the abstract semantics of the inputs.",
+      "context_before": "ciency. Semantics Abstraction. To support fuzzing the diverse DBMS interfaces, we propose to abstract the semantics to a point where most semantic differences are neutralized. To achieve this goal, we first describe common DBMS operations at a highly abstract level using only three basic data operations: Define ,Use, and Invalidate, which in turn defines a symbol, uses or updates a symbol, and deletes a symbol. We then use constraints to constrain the highly abstract semantics, such as specifying when to Define orUse, the type of the symbol that is Defined, or the type that a Usecan take, etc.",
+      "context_after": "Afterward, we design an Annotation System for users to specify the semantic constraints. Following this way, we neutralize the semantic differences across diverse DBMS interfaces and stay generic. Context-sensitive Constraint Resolution. To avoid the problem of static constraints, we enable dynamic contextbased constraints with Context-sensitive Constraint Resolution. Specifically, we craft two additional features for the Annotation System, through which users can specify the constraints based on the context. To achieve simplicity, we design a lightweight domain-specific language (DSL) for th",
+      "section": "2.3 Our Insights and Solutions",
+      "page": 5,
+      "char_offset": 20460,
+      "cited_reference": {
+        "number": 56,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communic",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M8",
+      "found_by": "citation_marker",
+      "surface": "[ 27,56]",
+      "technique": null,
+      "sentence": "4 Generalization Mutation can easily break the semantic correctness of a test case, as mentioned in many related works [ 27,56].",
+      "context_before": "tion file to constrain the abstract semantics of the target DBMS interface. Next, BUZZBEEconducts semantic analysis and performs principled mutations using data dependencies as the guidance. Lastly, BUZZBEEvalidates the semantics by fixing the errors introduced by mutation and generates new test cases for testing the DBMS. In the next sections, we discuss in detail how the components of BUZZBEEare designed and how they collaborate together to support Generalizability §4,Context-sensitive Constraints §5, and Principled Mutation §6, which address the three aforementioned challenges respectively.",
+      "context_after": "BUZZBEE uses the abstract semantic model to check for the semantic correctness of the mutated test case. Similar to existing works, when BUZZBEEdetects semantic errors, it tries to fix them before sending them to the fuzzing runtime. In this section, we discuss in detail how BUZZBEEstands out by modeling the semantics of the various DBMS interfaces at a highly abstract level to neutralize their differences. BUZZBEEabstracts the DBMS interface semantics with three basic data operations: Define ,Use, and Invalidate. Each operation is associated with a data type and name. Define. Define represent",
+      "section": "2.3 Our Insights and Solutions",
+      "page": 6,
+      "char_offset": 23522,
+      "cited_reference": {
+        "number": 56,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communic",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M9",
+      "found_by": "citation_marker",
+      "surface": "[ 27,28,56]",
+      "technique": null,
+      "sentence": "Inspired by existing works [ 27,28,56],BUZZBEE’s IR is a tree structure with a one-to-one mapping to the abstract syntax tree of the original test case.",
+      "context_before": "relations. For instance, in Fig.1c, if we uncomment the DELcommand at line 7, it will delete k2. Since k2_field1 has a subordination relation with k2,k2_field1 should also be deleted. Similarly, in Fig.1a, if we delete table t1, its column c1should also be deleted. After an Invalidate operation, anyUseof already Invalidated data will become a semantic error. Consequently, if we uncomment the DELcommand at line 7 in Fig.1c, line 8 will yield two semantic errors, since k2andk2_field1 are both invalidated. We design an IR to carry both the syntactic and semantic information specified by the user.",
+      "context_after": "Besides, the IR captures the abstract semantics specified by the user. This allows us to conveniently lift the original test case into the IR, perform semantic analysis, mutations, and validations on the IR in a unified fashion, and compile the IR back to a new test case for fuzzing. We detail the structure of the IR in Fig.8. 5 Semantics Constraining After BUZZBEEmodels the DBMS interfaces at a high level, it uses constraints to further express richer semantics. In this section, we introduce how BUZZBEEenforces flexible yet lightweight context-sensitive semantic constraints for different DBM",
+      "section": "2.3 Our Insights and Solutions",
+      "page": 6,
+      "char_offset": 26121,
+      "cited_reference": {
+        "number": 56,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communic",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M10",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "SQUIRREL SQLANCER redis 2 Sem ✔ ✔ ✗ ✗ ✗ ✗ ✗ ✗ ⊖ ⊖ redis 3 Data ✔ ✔ ✔ ✗ ✗ ✗ ✗ ✗ ⊖ ⊖ redis 5 Data ✔ ✗ ✗ ✗ ✗ ✗ ✗ ✗ ⊖ ⊖ redis 6 Data ✔ ✔ ✗ ✗ ✗ ✗ ✗ ✗ ⊖ ⊖ redis 7 Data ✔ ✗ ✗ ✗ ✗ ✗ ✗ ✗ ⊖ ⊖ redis 8 Syn ✗ ✗ ✗ ✔ ✗ ✔ ✗ ✗ ⊖ ⊖ redis 12 Sem ✔ ✔ ✔ ✗ ✗ ✗ ✗ ✗ ⊖ ⊖ RedisGraph 19 Data ✔ ✔ ✗ ✗ ✗ ✗ ✗ ✗ ⊖ ⊖ RedisGraph 20 Sem ✔ ✔ ✗ ✗ ✗ ✗ ✗ ✗ ⊖ ⊖ RedisGraph 21 Data ✔ ✗ ✗ ✗ ✗ ✗ ✗ ✗ ⊖ ⊖ ArangoDB 35 Data ✔ ✗ ✗ ✗ ✗ ✗ ⊖✗⊖ ⊖ ArangoDB 36 Data ✔ ✔ ✗ ✗ ✗ ✗ ⊖✗⊖ ⊖ PostgreSQL 37 Data ✔ ✔ ✗ ✗ ✗ ✗ ⊖✗ ✗ ✗ Count 12 8 2 1 0 1 0 0 0 0 triggerable by syntax correct but semantic incorrect test cases, we mark it as Syn.",
+      "context_before": "s), we mark the bug as Sem. For the bug that is onlyTable 4: Bugs found in 24 hours by each fuzzer on four DBMSs. We run each fuzzer five times for 24h and report the bugs it finds during this period. Property Data means the bug needs correct data dependencies to trigger. Sem means the bug does not need data dependencies to trigger, but the triggering test case is semantically correct. Synmeans the bug is not semantically correct but is syntactically correct. ⊖means the fuzzer does not support the DBMS. DBMSID Property BUZZBEE BUZZBEE!g BUZZBEE!gc BUZZBEE!gcs AFL++ POLYGLOT REDQUEEN Grammarin.",
+      "context_after": "As we can see in Table 4, BUZZBEE finds more Data bugs than BUZZBEE!g. For bugs 2 and 20 that are marked as Sem,BUZZBEE!gsuccessfully finds them, but BUZZBEE!gccannot, showing that Context-sensitive Constraint Resolution helps in enforcing the correct semantics of the test cases. Moreover, BUZZBEE!gcfinds two more bugs than BUZZBEE!gcs, which are bugs that require semantic correctness to trigger. This demonstrates the effectiveness of the Annotation System as a whole in enforcing semantic correctness. Finally, BUZZBEE!gcsfinds one bug (ID 8) that only semantic incorrect test cases can trigger.",
+      "section": "8.4 Contributions of the Solutions",
+      "page": 13,
+      "char_offset": 62474,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M11",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "We compare BUZZBEEwith general-purpose fuzzers AFL++ [ 14],REDQUEEN [4], syntax-aware fuzzers POLYGLOT [9], Grammarinator [ 22], and SQL-specialized SQUIR REL [56] and SQLANCER [43].",
+      "context_before": "whole in enforcing semantic correctness. Finally, BUZZBEE!gcsfinds one bug (ID 8) that only semantic incorrect test cases can trigger. Other versions cannot find this bug because they will enforce semantic correctness and filter out such test cases. In conclusion, BUZZBEE’s success comes from all the solutions we have proposed. The contribution of each solution varies when applied to different DBMS targets that have distinct characteristics. 8.5 Comparison with Existing Tools In this section, we compare BUZZBEEwith state-of-the-art works regarding code coverage and bug detection capabilities.",
+      "context_after": "AFL++ and REDQUEEN are widely used coverage-guided fuzzers that perform syntax912 33rd USENIX Security Symposium USENIX Association 0 4 8 12 16 2001224364860EdgeCov (1E+02) BuzzBee AFL++ RedQueen PolyGlot Grammarinator 0 4 8 12 16 2001224364860EdgeCov (1E+02) (a)redis 0 4 8 12 16 2001632486480EdgeCov (1E+02) (b)RedisGraph 0 4 8 12 16 2004812EdgeCov (1E+03) BuzzBee AFL++ PolyGlot Squirrel SQLancer Grammarinator 0 4 8 12 16 2006121824EdgeCov (1E+03) (c)ArangoDB 0 4 8 12 16 2004812EdgeCov (1E+03) (d)PostgreSQL Fig.7: Edge coverage comparison with existing fuzzers in 24h. unaware mutations. RED",
+      "section": "8.5 Comparison with Existing Tools",
+      "page": 13,
+      "char_offset": 64121,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M12",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "AFL++ and REDQUEEN are widely used coverage-guided fuzzers that perform syntax912 33rd USENIX Security Symposium USENIX Association 0 4 8 12 16 2001224364860EdgeCov (1E+02) BuzzBee AFL++ RedQueen PolyGlot Grammarinator 0 4 8 12 16 2001224364860EdgeCov (1E+02) (a)redis 0 4 8 12 16 2001632486480EdgeCov (1E+02) (b)RedisGraph 0 4 8 12 16 2004812EdgeCov (1E+03) BuzzBee AFL++ PolyGlot Squirrel SQLancer Grammarinator 0 4 8 12 16 2006121824EdgeCov (1E+03) (c)ArangoDB 0 4 8 12 16 2004812EdgeCov (1E+03) (d)PostgreSQL Fig.",
+      "context_before": "y will enforce semantic correctness and filter out such test cases. In conclusion, BUZZBEE’s success comes from all the solutions we have proposed. The contribution of each solution varies when applied to different DBMS targets that have distinct characteristics. 8.5 Comparison with Existing Tools In this section, we compare BUZZBEEwith state-of-the-art works regarding code coverage and bug detection capabilities. We compare BUZZBEEwith general-purpose fuzzers AFL++ [ 14],REDQUEEN [4], syntax-aware fuzzers POLYGLOT [9], Grammarinator [ 22], and SQL-specialized SQUIR REL [56] and SQLANCER [43].",
+      "context_after": "7: Edge coverage comparison with existing fuzzers in 24h. unaware mutations. REDQUEEN does not support fuzzing C/S programs, and we skip its evaluation on ArangoDB and PostgreSQL .POLYGLOTis a coverage-guided fuzzer tailored for fuzzing compilers such as Clang. It models the semantics of language processors effectively, but cannot model the semantics of DBMS interfaces due to its insensitivity to the context. It is a great tool to compare against because it acts as a coverage-guided syntax-aware mutator for targets outside its support range. We also compare against Grammarinator, a fuzzer curr",
+      "section": "8.5 Comparison with Existing Tools",
+      "page": 13,
+      "char_offset": 64304,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M13",
+      "found_by": "technique",
+      "surface": "PQS",
+      "technique": "pqs",
+      "sentence": "For relational DBMSs, we compare BUZZBEEwith SQUIRREL [56] and SQL ANCER (PQS [ 43]), two DBMS fuzzers specialized in SQL DBMS fuzzing.",
+      "context_before": "port fuzzing C/S programs, and we skip its evaluation on ArangoDB and PostgreSQL .POLYGLOTis a coverage-guided fuzzer tailored for fuzzing compilers such as Clang. It models the semantics of language processors effectively, but cannot model the semantics of DBMS interfaces due to its insensitivity to the context. It is a great tool to compare against because it acts as a coverage-guided syntax-aware mutator for targets outside its support range. We also compare against Grammarinator, a fuzzer currently used by RedisGraph, that generates random programs from a grammar without coverage guidance.",
+      "context_after": "We also evaluate SQLANCER on ArangoDB because SQLANCER recently adds support for it. Table 6 shows the DBMSs each fuzzer supports. In this evaluation, we feed the same input to all fuzzers and the same grammar specification to POLYGLOT and Grammarinator. Coverage. As shown in Fig.7, BUZZBEEachieves the best edge coverage in 24 hours in all targets except PostgreSQL. BUZZBEEachieves 69.2%, 19.8%, and 76.9% more coverage in redis ,RedisGraph, and ArangoDB than the secondbest fuzzer POLYGLOT. This performance gain demonstrates BUZZBEE’s contribution to the non-relational DBMS fuzzing venue. On P",
+      "section": "8.5 Comparison with Existing Tools",
+      "page": 14,
+      "char_offset": 65522,
+      "found_by_all": [
+        "technique",
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M14",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "We also evaluate SQLANCER on ArangoDB because SQLANCER recently adds support for it.",
+      "context_before": "ing compilers such as Clang. It models the semantics of language processors effectively, but cannot model the semantics of DBMS interfaces due to its insensitivity to the context. It is a great tool to compare against because it acts as a coverage-guided syntax-aware mutator for targets outside its support range. We also compare against Grammarinator, a fuzzer currently used by RedisGraph, that generates random programs from a grammar without coverage guidance. For relational DBMSs, we compare BUZZBEEwith SQUIRREL [56] and SQL ANCER (PQS [ 43]), two DBMS fuzzers specialized in SQL DBMS fuzzing.",
+      "context_after": "Table 6 shows the DBMSs each fuzzer supports. In this evaluation, we feed the same input to all fuzzers and the same grammar specification to POLYGLOT and Grammarinator. Coverage. As shown in Fig.7, BUZZBEEachieves the best edge coverage in 24 hours in all targets except PostgreSQL. BUZZBEEachieves 69.2%, 19.8%, and 76.9% more coverage in redis ,RedisGraph, and ArangoDB than the secondbest fuzzer POLYGLOT. This performance gain demonstrates BUZZBEE’s contribution to the non-relational DBMS fuzzing venue. On PostgreSQL ,BUZZBEEachieves 92.7% code coverage of the state-of-the-art SQL fuzzer SQU",
+      "section": "8.5 Comparison with Existing Tools",
+      "page": 14,
+      "char_offset": 65657,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M15",
+      "found_by": "citation_marker",
+      "surface": "[ 56]",
+      "technique": null,
+      "sentence": "Referring to SQUIRREL ’s paper [ 56], it achieves a semantic correctness rate of 11.",
+      "context_before": "eves the best edge coverage in 24 hours in all targets except PostgreSQL. BUZZBEEachieves 69.2%, 19.8%, and 76.9% more coverage in redis ,RedisGraph, and ArangoDB than the secondbest fuzzer POLYGLOT. This performance gain demonstrates BUZZBEE’s contribution to the non-relational DBMS fuzzing venue. On PostgreSQL ,BUZZBEEachieves 92.7% code coverage of the state-of-the-art SQL fuzzer SQUIRREL, showing comparable abilities considering BUZZBEEgeneralizes to many DBMS categories. In Table 3, we notice BUZZBEE achieves a correctness rate of 9.8% on PostgreSQL. This is also comparable with SQUIRREL.",
+      "context_after": "7% onPostgreSQL, which is only 1.9% higher than BUZZBEE. Bug Finding. As shown on the right side of Table 4, out of the six existing fuzzers we evaluate, only POLYGLOT finds one bug in redis, which falls into the category of Syn, meaning this bug only needs syntax awareness to discover. AFL++and REDQUEEN cannot find any bugs during this period. None of BUZZBEE,SQUIRREL, and SQLANCER can find bugs in the latest version of PostgreSQL within 24 hours. We roll back PostgreSQL to an older version, and BUZZBEEfinds one legacy bug missed by SQUIRREL and SQL ANCER. Notice that BUZZBEEdoes not perform as",
+      "section": "8.5 Comparison with Existing Tools",
+      "page": 14,
+      "char_offset": 66553,
+      "cited_reference": {
+        "number": 56,
+        "text": "Rui Zhong, Yongheng Chen, Hong Hu, Hangfan Zhang, Wenke Lee, and Dinghao Wu. SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback. In Jay Ligatti, Xinming Ou, Jonathan Katz, and Giovanni Vigna, editors, CCS ’20: 2020 ACMSIGSAC Conference on Computer and Communic",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M16",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "None of BUZZBEE,SQUIRREL, and SQLANCER can find bugs in the latest version of PostgreSQL within 24 hours.",
+      "context_before": "dering BUZZBEEgeneralizes to many DBMS categories. In Table 3, we notice BUZZBEE achieves a correctness rate of 9.8% on PostgreSQL. This is also comparable with SQUIRREL. Referring to SQUIRREL ’s paper [ 56], it achieves a semantic correctness rate of 11.7% onPostgreSQL, which is only 1.9% higher than BUZZBEE. Bug Finding. As shown on the right side of Table 4, out of the six existing fuzzers we evaluate, only POLYGLOT finds one bug in redis, which falls into the category of Syn, meaning this bug only needs syntax awareness to discover. AFL++and REDQUEEN cannot find any bugs during this period.",
+      "context_after": "We roll back PostgreSQL to an older version, and BUZZBEEfinds one legacy bug missed by SQUIRREL and SQL ANCER. Notice that BUZZBEEdoes not perform as well as SQUIRREL and SQL ANCER in terms of code coverage. To understand the reason, we manually analyze this bug and find that it involves a syntax structure SQUIRREL does not model, potentially leading to its missing the bug. SQLANCER focuses on finding logic errors using specific syntax structures and does not discover the bug as well. The result demonstrates the capability of BUZZBEEin discovering real-world bugs that existing tools cannot disco",
+      "section": "8.5 Comparison with Existing Tools",
+      "page": 14,
+      "char_offset": 66983,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M17",
+      "found_by": "name",
+      "surface": "SQLANCER",
+      "technique": null,
+      "sentence": "SQLANCER focuses on finding logic errors using specific syntax structures and does not discover the bug as well.",
+      "context_before": "yn, meaning this bug only needs syntax awareness to discover. AFL++and REDQUEEN cannot find any bugs during this period. None of BUZZBEE,SQUIRREL, and SQLANCER can find bugs in the latest version of PostgreSQL within 24 hours. We roll back PostgreSQL to an older version, and BUZZBEEfinds one legacy bug missed by SQUIRREL and SQL ANCER. Notice that BUZZBEEdoes not perform as well as SQUIRREL and SQL ANCER in terms of code coverage. To understand the reason, we manually analyze this bug and find that it involves a syntax structure SQUIRREL does not model, potentially leading to its missing the bug.",
+      "context_after": "The result demonstrates the capability of BUZZBEEin discovering real-world bugs that existing tools cannot discover, proving its fuzzing effectiveness and contribution to real-world bug detection. In conclusion, BUZZBEEoutperforms existing generic solutions in non-relational DBMS fuzzing, while achieving comparable results with tools specialized in relational DBMS fuzzing, in terms of code coverage and bug-finding capabilities. 9 Discussion 9.1 Complex Semantics and Completeness Some DBMS semantics can be very complex. For instance, the foreign key constraint in SQL DBMSs requires us to track",
+      "section": "8.5 Comparison with Existing Tools",
+      "page": 14,
+      "char_offset": 67464,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M18",
+      "found_by": "citation_marker",
+      "surface": "[ 15,25–27,43,44,49,56]",
+      "technique": "pqs",
+      "sentence": "3 DBMSs Fuzzing Fuzzing DBMSs has been an active research area in recent years [ 15,25–27,43,44,49,56].",
+      "context_before": "utation. DIE [ 37] and Fuzzilli [ 20] propose strategies to maintain the semantics stressing the just-in-time (JIT) engines. For compilers, CSmith [ 51] specializes in modeling C semantics to produce entirely correct test cases for C compilers. Tools such as Nautilus [ 3] and Superion [ 48] propose genericgrammar-based solutions to fuzz more language processors. POLYGLOT [9] further advances using a semantic validation strategy that performs well when no context-sensitive constraint is required. BUZZBEEfocuses on the challenges in fuzzing the diverse DBMS interfaces effectively in general. 10.",
+      "context_after": "Tools like SQLancer [ 42], SQLsmith [ 44], and Squirrel [ 56] have emerged to test relational DBMSs. SQLsmith generates random SQL queries. Squirrel employs semantic validation and coverage feedback to test SQL DBMSs. SQLancer uses differential testing techniques to report inconsistencies in query results. They have been proven successful in finding vulnerabilities in widely used relational DBMSs. Researchers also propose some solutions targeting non-relational DBMSs [ 26,55]. Existing techniques present unique solutions targeting specific DBMS categories, but fail to exhibit decent fuzzing p",
+      "section": "10.3 DBMSs Fuzzing",
+      "page": 15,
+      "char_offset": 72605,
+      "cited_reference": {
+        "number": 43,
+        "text": "Manuel Rigger and Zhendong Su. Testing Database Engines via Pivoted Query Synthesis. In 14th USENIX Symposium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020, pages 667–682. USENIX Association, 2020.",
+        "matched_as": "sqlancer_publication",
+        "why": [
+          "cites the paper introducing pqs",
+          "names SQLancer or one of its techniques"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker"
+      ],
+      "techniques": [
+        "pqs"
+      ]
+    },
+    {
+      "id": "M19",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "Tools like SQLancer [ 42], SQLsmith [ 44], and Squirrel [ 56] have emerged to test relational DBMSs.",
+      "context_before": "n-time (JIT) engines. For compilers, CSmith [ 51] specializes in modeling C semantics to produce entirely correct test cases for C compilers. Tools such as Nautilus [ 3] and Superion [ 48] propose genericgrammar-based solutions to fuzz more language processors. POLYGLOT [9] further advances using a semantic validation strategy that performs well when no context-sensitive constraint is required. BUZZBEEfocuses on the challenges in fuzzing the diverse DBMS interfaces effectively in general. 10.3 DBMSs Fuzzing Fuzzing DBMSs has been an active research area in recent years [ 15,25–27,43,44,49,56].",
+      "context_after": "SQLsmith generates random SQL queries. Squirrel employs semantic validation and coverage feedback to test SQL DBMSs. SQLancer uses differential testing techniques to report inconsistencies in query results. They have been proven successful in finding vulnerabilities in widely used relational DBMSs. Researchers also propose some solutions targeting non-relational DBMSs [ 26,55]. Existing techniques present unique solutions targeting specific DBMS categories, but fail to exhibit decent fuzzing performance for the diverse DBMS interfaces in general. As a result, BUZZBEE is designed to address th",
+      "section": "10.3 DBMSs Fuzzing",
+      "page": 15,
+      "char_offset": 72708,
+      "found_by_all": [
+        "name",
+        "citation_marker"
+      ],
+      "techniques": [
+        "norec"
+      ]
+    },
+    {
+      "id": "M20",
+      "found_by": "name",
+      "surface": "SQLancer",
+      "technique": null,
+      "sentence": "SQLancer uses differential testing techniques to report inconsistencies in query results.",
+      "context_before": "solutions to fuzz more language processors. POLYGLOT [9] further advances using a semantic validation strategy that performs well when no context-sensitive constraint is required. BUZZBEEfocuses on the challenges in fuzzing the diverse DBMS interfaces effectively in general. 10.3 DBMSs Fuzzing Fuzzing DBMSs has been an active research area in recent years [ 15,25–27,43,44,49,56]. Tools like SQLancer [ 42], SQLsmith [ 44], and Squirrel [ 56] have emerged to test relational DBMSs. SQLsmith generates random SQL queries. Squirrel employs semantic validation and coverage feedback to test SQL DBMSs.",
+      "context_after": "They have been proven successful in finding vulnerabilities in widely used relational DBMSs. Researchers also propose some solutions targeting non-relational DBMSs [ 26,55]. Existing techniques present unique solutions targeting specific DBMS categories, but fail to exhibit decent fuzzing performance for the diverse DBMS interfaces in general. As a result, BUZZBEE is designed to address this with a focus on generalizability and effectiveness. Some optimizations proposed in SQL fuzzers seem effective and generalizable to other DBMSs. DynSQL [ 25] combines the dynamic feedback from SQL DBMSs wi",
+      "section": "10.3 DBMSs Fuzzing",
+      "page": 15,
+      "char_offset": 72926,
+      "found_by_all": [
+        "name"
+      ],
+      "techniques": []
+    },
+    {
+      "id": "M21",
+      "found_by": "citation_marker_project_authored",
+      "surface": "[ 26,55]",
+      "technique": null,
+      "sentence": "Researchers also propose some solutions targeting non-relational DBMSs [ 26,55].",
+      "context_before": "ZBEEfocuses on the challenges in fuzzing the diverse DBMS interfaces effectively in general. 10.3 DBMSs Fuzzing Fuzzing DBMSs has been an active research area in recent years [ 15,25–27,43,44,49,56]. Tools like SQLancer [ 42], SQLsmith [ 44], and Squirrel [ 56] have emerged to test relational DBMSs. SQLsmith generates random SQL queries. Squirrel employs semantic validation and coverage feedback to test SQL DBMSs. SQLancer uses differential testing techniques to report inconsistencies in query results. They have been proven successful in finding vulnerabilities in widely used relational DBMSs.",
+      "context_after": "Existing techniques present unique solutions targeting specific DBMS categories, but fail to exhibit decent fuzzing performance for the diverse DBMS interfaces in general. As a result, BUZZBEE is designed to address this with a focus on generalizability and effectiveness. Some optimizations proposed in SQL fuzzers seem effective and generalizable to other DBMSs. DynSQL [ 25] combines the dynamic feedback from SQL DBMSs with fuzzing. Ratel [ 49] tackles the challenges of fuzzing DBMSs in real-world settings. Griffin [ 15] mutates the input without relying on grammar by tracking dependencies am",
+      "section": "10.3 DBMSs Fuzzing",
+      "page": 15,
+      "char_offset": 73109,
+      "cited_reference": {
+        "number": 26,
+        "text": "Matteo Kamm, Manuel Rigger, Chengyu Zhang, and Zhendong Su. Testing Graph Database Engines via Query Partitioning. In René Just and Gordon Fraser, editors, Proceedings of the 32nd ACMSIGSOFT International Symposium on Software Testing and Analysis, ISSTA 2023, Seattle, WA, USA, July 17-21, 2023, pag",
+        "matched_as": "project_authored",
+        "why": [
+          "written by an author of the SQLancer project, but not itself a SQLancer paper"
+        ]
+      },
+      "found_by_all": [
+        "citation_marker_project_authored"
+      ],
+      "techniques": []
+    }
+  ],
+  "artifact": null,
+  "checks": {
+    "checks_version": "paper-checks-v1",
+    "suggests": {
+      "compares_with": [
+        "M13",
+        "M14"
+      ]
+    },
+    "suppressed": []
+  },
+  "analysis": {
+    "analysis_version": "paper-analysis-v1",
+    "prompt_version": "paper-analysis-v1",
+    "produced_by": "claude_code_session",
+    "model": "claude-opus-5",
+    "generated_at": "2026-09-08T16:37:26Z",
+    "is_model_written": true,
+    "summary": "BuzzBee fuzzes database systems beyond the relational model -- key-value, graph and document stores -- where SQL-specific fuzzers do not apply. It generalises the intermediate-representation approach that SQL fuzzers use for tracking syntax and semantics into a form covering several data models, so that semantic correctness survives mutation. Evaluated on Redis, RedisGraph, ArangoDB, MongoDB and relational systems, it found bugs the specialised tools miss.",
+    "narrative": "SQLancer is one of BuzzBee's baselines for the relational case, run with its PQS oracle, and also on ArangoDB because SQLancer had recently added support for it. The comparison is reported per bug: on several Redis and RedisGraph defects SQLancer's column is empty, and the paper attributes that to scope -- it looks for logic errors through specific syntax structures rather than the crashes and semantic faults those systems exhibit. The paper is even-handed about the relational case, noting that none of BuzzBee, SQUIRREL or SQLancer found a bug in current PostgreSQL within 24 hours. SQLancer's design also informs BuzzBee indirectly through the IR line of work it belongs to.",
+    "roles": {
+      "M1": "background",
+      "M2": "background",
+      "M3": "background",
+      "M4": "background",
+      "M5": "background",
+      "M6": "background",
+      "M7": "background",
+      "M8": "background",
+      "M9": "background",
+      "M10": "result_comparison",
+      "M11": "baseline",
+      "M12": "incidental",
+      "M13": "baseline",
+      "M14": "result_comparison",
+      "M15": "background",
+      "M16": "result_comparison",
+      "M17": "result_comparison",
+      "M18": "background",
+      "M19": "definition",
+      "M20": "definition",
+      "M21": "background"
+    },
+    "relationships": {
+      "uses_infrastructure": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is executed as a baseline; BuzzBee's own implementation is an independent intermediate representation, not built on SQLancer."
+      },
+      "extends_technique": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "BuzzBee generalises the IR approach it credits to SQUIRREL and its successors, not a SQLancer oracle."
+      },
+      "compares_with": {
+        "value": "yes",
+        "mention_ids": [
+          "M11",
+          "M13",
+          "M14",
+          "M16"
+        ],
+        "quotes": [
+          {
+            "mention_id": "M11",
+            "sentence": "We compare BUZZBEEwith general-purpose fuzzers AFL++ [ 14],REDQUEEN [4], syntax-aware fuzzers POLYGLOT [9], Grammarinator [ 22], and SQL-specialized SQUIR REL [56] and SQLANCER [43].",
+            "section": "8.5 Comparison with Existing Tools",
+            "page": 13
+          },
+          {
+            "mention_id": "M13",
+            "sentence": "For relational DBMSs, we compare BUZZBEEwith SQUIRREL [56] and SQL ANCER (PQS [ 43]), two DBMS fuzzers specialized in SQL DBMS fuzzing.",
+            "section": "8.5 Comparison with Existing Tools",
+            "page": 14
+          },
+          {
+            "mention_id": "M14",
+            "sentence": "We also evaluate SQLANCER on ArangoDB because SQLANCER recently adds support for it.",
+            "section": "8.5 Comparison with Existing Tools",
+            "page": 14
+          },
+          {
+            "mention_id": "M16",
+            "sentence": "None of BUZZBEE,SQUIRREL, and SQLANCER can find bugs in the latest version of PostgreSQL within 24 hours.",
+            "section": "8.5 Comparison with Existing Tools",
+            "page": 14
+          }
+        ],
+        "reasoning": "M11 lists SQLancer among the fuzzers BuzzBee is compared with, M13 states the relational comparison is against SQUIRREL and SQLancer using PQS, M14 that SQLancer was additionally run on ArangoDB, and M16 reports a result covering all three tools.",
+        "techniques": [
+          "pqs"
+        ]
+      },
+      "describes_as_state_of_the_art": {
+        "value": "no",
+        "mention_ids": [],
+        "quotes": [],
+        "reasoning": "SQLancer is called SQL-specialized and a well-known tool, but the paper does not describe it as the state of the art."
+      }
+    },
+    "disagreements": [],
+    "unresolved": []
+  },
+  "provenance": {
+    "extractor_version": "paper-extract-v1",
+    "inventory_version": "paper-mentions-v1",
+    "checks_version": "paper-checks-v1",
+    "policy_version": "impact-policy-v1"
+  }
+}
diff --git a/_includes/head/custom.html b/_includes/head/custom.html
new file mode 100644
index 0000000..69ef51e
--- /dev/null
+++ b/_includes/head/custom.html
@@ -0,0 +1,8 @@
+<!-- start custom head snippets -->
+
+<!-- insert favicons. use https://realfavicongenerator.net/ -->
+
+<!-- Styles for the /impact/ section and the derived homepage statistics. -->
+<link rel="stylesheet" href="{{ '/assets/css/impact.css' | relative_url }}">
+
+<!-- end custom head snippets -->
diff --git a/_includes/impact/artifact-evidence.html b/_includes/impact/artifact-evidence.html
new file mode 100644
index 0000000..5ae0310
--- /dev/null
+++ b/_includes/impact/artifact-evidence.html
@@ -0,0 +1,68 @@
+{%- comment -%}
+Explains, in words, why a paper counts as reusing SQLancer infrastructure.
+
+The raw evidence does not read well on its own: the link evidence quotes the
+repository describing the paper, and the marker evidence quotes a line of Java.
+Neither tells a reader what the connection to SQLancer is. This turns the marker
+list into plain English and puts the two halves of the claim side by side --
+what was found in the artifact, and why the artifact belongs to this paper.
+
+Parameter: paper (a paper record).
+{%- endcomment -%}
+{%- assign artifact = paper.artifacts | first -%}
+{%- if artifact %}
+<div class="impact-evidence">
+  {%- assign marker_phrases = "" | split: "" -%}
+  {%- for marker in artifact.sqlancer_markers -%}
+    {%- case marker -%}
+      {%- when "fork_of_sqlancer_repository" -%}
+        {%- assign phrase = "is a fork of the SQLancer repository" -%}
+      {%- when "sqlancer_source_content_match" -%}
+        {%- assign phrase = "contains SQLancer's own source files" -%}
+      {%- when "renamed_sqlancer_package" -%}
+        {%- assign phrase = "carries them under a renamed package" -%}
+      {%- when "retained_sqlancer_source_files" -%}
+        {%- assign phrase = "retains SQLancer source files" -%}
+      {%- when "sqlancer_package_structure" -%}
+        {%- assign phrase = "keeps SQLancer's package layout" -%}
+      {%- when "sqlancer_provider_directory" -%}
+        {%- assign phrase = "keeps SQLancer's database-system providers" -%}
+      {%- when "sqlancer_build_file_reference" -%}
+        {%- assign phrase = "builds under SQLancer's own build coordinates" -%}
+      {%- when "randomly_java_present" -%}
+        {%- assign phrase = "includes SQLancer's Randomly.java" -%}
+      {%- when "sqlancer_copyright_or_license_notice" -%}
+        {%- assign phrase = "carries a SQLancer copyright notice" -%}
+      {%- when "readme_states_reuse" -%}
+        {%- assign phrase = "says so in its README" -%}
+      {%- else -%}
+        {%- assign phrase = "" -%}
+    {%- endcase -%}
+    {%- if phrase != "" -%}
+      {%- assign marker_phrases = marker_phrases | push: phrase -%}
+    {%- endif -%}
+  {%- endfor -%}
+
+  Its artifact,
+  <a href="{{ artifact.url }}">{{ artifact.url | remove: "https://github.com/" }}</a>,
+  {% if marker_phrases.size > 0 %}{{ marker_phrases | join: ", and " }}.{% else %}shows SQLancer markers.{% endif %}
+
+  {%- comment -%}
+    The evidence behind both halves: why the repository is SQLancer's code, and
+    why it is this paper's repository.
+  {%- endcomment -%}
+  {%- comment -%}
+    The note, not the line it was read from. Marker evidence quotes source
+    code, and the same declaration out of SQLancer's Randomly.java is what
+    thirteen of these papers copied -- so the quote says nothing the note has
+    not already said, in a form nobody reads. The link goes to the file for
+    anyone who wants it.
+  {%- endcomment -%}
+  {%- for item in artifact.marker_evidence limit: 3 %}
+  <div class="impact-meta">
+    {{ item.note }}
+    <a href="{{ item.source_url }}">source</a>
+  </div>
+  {%- endfor %}
+</div>
+{%- endif %}
diff --git a/_includes/impact/bar-list.html b/_includes/impact/bar-list.html
new file mode 100644
index 0000000..df0e7af
--- /dev/null
+++ b/_includes/impact/bar-list.html
@@ -0,0 +1,26 @@
+{%- comment -%}
+A labelled count series as bars, in HTML rather than SVG.
+
+The page charts are generated SVGs because they are shared, few, and reviewed in
+a diff. These are per-system and there are hundreds of them, so they are drawn
+from the data at render time instead: same visual language, no generated
+artifacts to keep in step. One hue, because the job is magnitude.
+
+Each row carries its own number, so the bar is redundant encoding rather than
+the only way to read the value.
+
+Parameter: series (labelled_counts array).
+{%- endcomment -%}
+{%- assign counts = include.series | map: "count" | sort | reverse -%}
+{%- assign top = counts | first -%}
+<dl class="impact-bars">
+  {%- for row in include.series %}
+  <div class="impact-bars__row">
+    <dt class="impact-bars__label">{{ row.label }}</dt>
+    <dd class="impact-bars__value">
+      <span class="impact-bars__bar" style="width: {% if top > 0 %}{{ row.count | times: 100.0 | divided_by: top }}{% else %}0{% endif %}%"></span>
+      <span class="impact-bars__count">{{ row.count }}</span>
+    </dd>
+  </div>
+  {%- endfor %}
+</dl>
diff --git a/_includes/impact/counts-table.html b/_includes/impact/counts-table.html
new file mode 100644
index 0000000..c4fbec6
--- /dev/null
+++ b/_includes/impact/counts-table.html
@@ -0,0 +1,34 @@
+{%- comment -%}
+Table view of a labelled count series -- the accessible counterpart to a chart,
+and the relief the palette requires for its lower-contrast light-mode slots.
+
+Parameters: series (labelled_counts array), heading (column title),
+            label (optional caption for the value column),
+            link_base (optional; when given each row links to
+                       link_base<key>/, so the table is a way in to the
+                       detail pages and not only a way to read the chart).
+{%- endcomment -%}
+<div class="impact-table-wrap">
+  <table class="impact-table">
+    <thead>
+      <tr>
+        <th scope="col">{{ include.heading }}</th>
+        <th scope="col" class="impact-num">{{ include.label | default: "Count" }}</th>
+      </tr>
+    </thead>
+    <tbody>
+      {%- for row in include.series %}
+      <tr>
+        <th scope="row">
+          {%- if include.link_base and row.key -%}
+          <a href="{{ include.link_base }}{{ row.key }}/">{{ row.label }}</a>
+          {%- else -%}
+          {{ row.label }}
+          {%- endif -%}
+        </th>
+        <td class="impact-num">{{ row.count }}</td>
+      </tr>
+      {%- endfor %}
+    </tbody>
+  </table>
+</div>
diff --git a/_includes/impact/dbms-detail.html b/_includes/impact/dbms-detail.html
new file mode 100644
index 0000000..097dc8e
--- /dev/null
+++ b/_includes/impact/dbms-detail.html
@@ -0,0 +1,214 @@
+{%- comment -%}
+Everything the dataset records about one database system.
+
+The impact page can only show a bar and a row per system. This answers what
+those raise but cannot fit: which bugs, found by whom, under which rule, and --
+where the project uses SQLancer -- the evidence for saying so.
+
+Counts come from stats.json so they cannot drift from the charts. The bug list
+and the adoption evidence are read from the records themselves, because they are
+the evidence rather than a summary of it.
+
+Parameter: id (a database system id).
+{%- endcomment -%}
+{%- assign impact = site.data.impact -%}
+{%- assign row = impact.stats.dbms.rows | where: "id", include.id | first -%}
+{%- assign entry = impact.dbms.dbms | where: "id", include.id | first -%}
+{%- assign bugs = impact.bugs.bugs | where: "dbms", include.id -%}
+{%- assign adoption = impact.adoption.adoption | where: "dbms", include.id -%}
+{%- assign used = adoption | where_exp: "r", "r.relationship != 'planned_adoption'" -%}
+{%- comment -%}
+  A proposal is superseded by evidence that the project actually runs
+  SQLancer, so a system with both shows only the adoption. Seven do -- Turso's
+  page claimed it both uses SQLancer and has only proposed it -- and the
+  statistics have always counted them this way; the page had not.
+{%- endcomment -%}
+{%- if used.size > 0 -%}
+  {%- assign proposed = "" | split: "" -%}
+{%- else -%}
+  {%- assign proposed = adoption | where: "relationship", "planned_adoption" -%}
+{%- endif -%}
+
+<div class="impact" markdown="0">
+
+<p class="impact-breadcrumb">
+  <a href="{{ '/impact/#database-systems' | relative_url }}">&larr; All database systems</a>
+</p>
+
+<p>
+  {%- if entry.url %}<a href="{{ entry.url }}">{{ row.name }}</a>{% else %}{{ row.name }}{% endif -%}
+  {%- if row.supported %} is supported by SQLancer — the main repository ships a
+  testing implementation for it{% else %} has no testing implementation in the
+  main SQLancer repository{% endif -%}.
+  {%- if row.bugs > 0 %}
+  SQLancer is credited with <strong>{{ row.bugs }}</strong> bugs in it,
+  {%- if row.first_reported %} reported between {{ row.first_reported }} and
+  {{ row.last_reported }}{% endif -%}.
+  {%- else %}
+  No bugs in it are attributed to SQLancer.
+  {%- endif %}
+  {%- if row.repository %}
+  Its source is at <a href="{{ row.repository }}">{{ row.repository | remove: "https://github.com/" }}</a>.
+  {%- endif %}
+</p>
+
+{%- if row.bugs_rejected > 0 %}
+<p class="impact-meta">
+  A further {{ row.bugs_rejected }}
+  {%- if row.bugs_rejected == 1 %} report was{% else %} reports were{% endif %}
+  filed and rejected by the developers as invalid or duplicate. Those are kept
+  in the dataset for transparency and are excluded from every count here.
+</p>
+{%- endif %}
+
+{%- if used.size > 0 %}
+<h2 id="adoption">How the project uses SQLancer</h2>
+{%- elsif proposed.size > 0 %}
+{%- comment -%}
+  A proposal is not use, and a heading saying it is would be the exact
+  overstatement the policy exists to prevent.
+{%- endcomment -%}
+<h2 id="adoption">SQLancer and this project</h2>
+{%- endif %}
+
+{%- if used.size > 0 or proposed.size > 0 %}
+{%- for record in used %}
+<div class="impact-adoption">
+  <h3 class="impact-adoption__heading">
+    {{ record.relationship | replace: "_", " " | capitalize }}
+  </h3>
+  <p>{{ record.summary }}</p>
+  {%- for item in record.evidence %}
+  {%- comment -%}
+    Show the repository path rather than the full permalink: these URLs carry a
+    40-character commit SHA and are unreadable inline.
+  {%- endcomment -%}
+  {%- assign path = item.source_url | split: "/blob/" | last | split: "/" -%}
+  {%- assign label = path | shift | join: "/" -%}
+  <div class="impact-evidence">
+    {%- if item.excerpt %}<q>{{ item.excerpt | strip_html | truncate: 400 }}</q><br>{% endif -%}
+    {%- if item.note %}<span class="impact-meta">{{ item.note }}</span><br>{% endif -%}
+    <a href="{{ item.source_url }}">{{ label | default: item.source_url }}</a>
+    {%- if item.last_verified %}
+    <span class="impact-meta">last verified {{ item.last_verified | date: "%Y-%m-%d" }}</span>
+    {%- endif %}
+  </div>
+  {%- endfor %}
+</div>
+{%- endfor %}
+
+{%- if proposed.size > 0 %}
+<h3 id="proposed">Adoption proposed, not yet shown</h3>
+<p>
+  The project's developers have proposed adopting SQLancer without anything yet
+  showing the project running it. An intention is not use, so this counts
+  towards no figure on the impact page.
+</p>
+{%- for record in proposed %}
+<div class="impact-adoption">
+  <p>{{ record.summary }}</p>
+  {%- for item in record.evidence %}
+  <div class="impact-evidence">
+    {%- if item.excerpt %}<q>{{ item.excerpt | strip_html | truncate: 400 }}</q><br>{% endif -%}
+    {%- if item.note %}<span class="impact-meta">{{ item.note }}</span><br>{% endif -%}
+    <a href="{{ item.source_url }}">{{ item.source_url | remove: "https://github.com/" }}</a>
+  </div>
+  {%- endfor %}
+</div>
+{%- endfor %}
+{%- endif %}
+{%- endif %}
+
+{%- if row.bugs > 0 %}
+<h2 id="breakdown">How the bugs break down</h2>
+
+<div class="impact-columns">
+  <div>
+    <h3>By year reported</h3>
+    {% include impact/bar-list.html series=row.bugs_by_year %}
+  </div>
+  <div>
+    <h3>Who found them</h3>
+    {% include impact/bar-list.html series=row.bugs_by_affiliation %}
+  </div>
+  <div>
+    <h3>By status</h3>
+    {% include impact/bar-list.html series=row.bugs_by_status %}
+  </div>
+  <div>
+    <h3>By symptom</h3>
+    {% include impact/bar-list.html series=row.bugs_by_symptom %}
+  </div>
+  <div>
+    <h3>By technique</h3>
+    {% include impact/bar-list.html series=row.bugs_by_technique %}
+  </div>
+  <div>
+    <h3>Why each bug counts</h3>
+    {% include impact/bar-list.html series=row.bugs_by_attribution_rule %}
+  </div>
+</div>
+
+{%- if row.top_reporters.size > 0 %}
+<h3 id="reporters">Who reported them ({{ row.top_reporters.size }})</h3>
+{% include impact/bar-list.html series=row.top_reporters %}
+{%- endif %}
+
+<h2 id="bugs">Every bug on record ({{ bugs.size }})</h2>
+
+<p>
+  One row per report, newest first, each linking to the report itself. Rejected
+  reports are marked and are not part of any count above.
+</p>
+
+<div class="impact-table-wrap">
+  <table class="impact-table impact-table--bugs">
+    <thead>
+      <tr>
+        <th scope="col">Reported</th>
+        <th scope="col">Report</th>
+        <th scope="col">Status</th>
+        <th scope="col">Technique</th>
+        <th scope="col">Reporter</th>
+      </tr>
+    </thead>
+    <tbody>
+      {%- assign ordered = bugs | sort: "reported_date" | reverse %}
+      {%- for bug in ordered %}
+      <tr{% unless bug.status_is_true_positive %} class="impact-row--rejected"{% endunless %}>
+        <td>{{ bug.reported_date | default: "&mdash;" }}</td>
+        <th scope="row">
+          {%- comment -%}
+            Some reports have no public page to link: a bug the project records
+            in its own source but tracks in a private Jira, or one the lab sent
+            by email. The record it came from is then what a reader can check.
+          {%- endcomment -%}
+          {%- assign bug_link = bug.primary_url | default: bug.links.record -%}
+          {%- if bug_link -%}
+          <a href="{{ bug_link }}">{{ bug.title | default: bug.id | truncate: 110 }}</a>
+          {%- else -%}
+          {{ bug.title | default: bug.id | truncate: 110 }}
+          {%- endif -%}
+          {%- if bug.primary_url == nil and bug.links.tracker %}
+          <span class="impact-meta">tracked privately at
+          {{ bug.links.tracker | split: "/browse/" | last }}</span>
+          {%- endif -%}
+        </th>
+        <td>{{ bug.status | replace: "_", " " }}</td>
+        <td>{% if bug.technique %}{{ bug.technique | upcase }}{% else %}&mdash;{% endif %}</td>
+        <td>{{ bug.reporter | default: "&mdash;" }}</td>
+      </tr>
+      {%- endfor %}
+    </tbody>
+  </table>
+</div>
+{%- endif %}
+
+<p class="impact-meta">
+  Every figure on this page is computed from
+  <a href="https://github.com/sqlancer/sqlancer.github.io/tree/main/_data/impact"><code>_data/impact/</code></a>.
+  The <a href="{{ '/impact/#methodology' | relative_url }}">attribution policy</a>
+  explains what has to be true for a bug to be counted.
+</p>
+
+</div>
diff --git a/_includes/impact/feature-row.html b/_includes/impact/feature-row.html
new file mode 100644
index 0000000..4f7bba6
--- /dev/null
+++ b/_includes/impact/feature-row.html
@@ -0,0 +1,80 @@
+{%- comment -%}
+The homepage feature cards.
+
+This mirrors the theme's own `feature_row` markup and CSS classes, so the cards
+look exactly as before. It exists as a separate include only because Jekyll does
+not evaluate Liquid inside front matter, and the card text now has to quote
+numbers derived from the impact data rather than a hand-written claim.
+{%- endcomment -%}
+{%- assign head = site.data.impact.stats.headline -%}
+
+<div class="feature__wrapper">
+
+  <div class="feature__item">
+    <div class="archive__item">
+      <div class="archive__item-teaser">
+        <img src="{{ '/assets/images/sqlancer_logo_pos.svg' | relative_url }}"
+             alt="The SQLancer logo">
+      </div>
+      <div class="archive__item-body">
+        <h2 class="archive__item-title">100% free</h2>
+        <div class="archive__item-excerpt">
+          <p>Free to use however you want under the MIT License. Clone it, fork
+          it, customize it&hellip; whatever!</p>
+        </div>
+        <p><a href="https://github.com/sqlancer/sqlancer/blob/main/LICENSE"
+              class="btn btn--primary">Learn more</a></p>
+      </div>
+    </div>
+  </div>
+
+  <div class="feature__item">
+    <div class="archive__item">
+      <div class="archive__item-teaser">
+        <img src="{{ '/assets/images/supported_systems.png' | relative_url }}"
+             alt="Logos of the supported database systems">
+      </div>
+      <div class="archive__item-body">
+        <h2 class="archive__item-title">
+          {{ head.dbms_supported }} database systems
+        </h2>
+        <div class="archive__item-excerpt">
+          {%- comment -%}
+            Not "of them": the two numbers count different sets. The first is
+            the systems SQLancer ships a provider for; the second is every
+            system a bug has been attributed to, which is larger and only
+            partly overlaps -- 25 of them have no current provider, reached
+            instead through vendor forks and tools built on SQLancer.
+          {%- endcomment -%}
+          <p>SQLancer generates test cases for {{ head.dbms_supported }} database
+          systems. Bugs have been found in {{ head.dbms_with_bugs }} systems in
+          all, counting those reached by forks and by tools built on it.</p>
+        </div>
+        <p><a href="{{ '/supported-databases/' | relative_url }}"
+              class="btn btn--primary">Learn more</a></p>
+      </div>
+    </div>
+  </div>
+
+  <div class="feature__item">
+    <div class="archive__item">
+      <div class="archive__item-teaser">
+        <img src="{{ '/assets/images/bugs.png' | relative_url }}"
+             alt="An illustration of a software bug">
+      </div>
+      <div class="archive__item-body">
+        <h2 class="archive__item-title">
+          {{ head.bugs_total_rounded }} bugs found
+        </h2>
+        <div class="archive__item-excerpt">
+          <p>Every bug is recorded with the evidence linking it to SQLancer, as
+          are the {{ head.papers_citing_sqlancer }} papers that cite the
+          project.</p>
+        </div>
+        <p><a href="{{ '/impact/' | relative_url }}"
+              class="btn btn--primary">See the impact</a></p>
+      </div>
+    </div>
+  </div>
+
+</div>
diff --git a/_includes/impact/headline.html b/_includes/impact/headline.html
new file mode 100644
index 0000000..b776840
--- /dev/null
+++ b/_includes/impact/headline.html
@@ -0,0 +1,28 @@
+{%- comment -%}
+Headline impact numbers for the homepage hero.
+
+Every value is read from _data/impact/stats.json, which is regenerated from the
+underlying records by `python -m tools.impact.run stats`. Nothing here is a
+literal: editing a number means editing a record, not this file.
+{%- endcomment -%}
+{%- assign head = site.data.impact.stats.headline -%}
+{%- if head -%}
+<ul class="impact impact-headline">
+  <li class="impact-headline__item">
+    <span class="impact-headline__value">{{ head.bugs_total_rounded }}</span>
+    <span class="impact-headline__label">bugs found</span>
+  </li>
+  <li class="impact-headline__item">
+    <span class="impact-headline__value">{{ head.dbms_supported }}</span>
+    <span class="impact-headline__label">DBMSs supported</span>
+  </li>
+  <li class="impact-headline__item">
+    <span class="impact-headline__value">{{ head.dbms_projects_using_sqlancer }}</span>
+    <span class="impact-headline__label">DBMS projects using SQLancer</span>
+  </li>
+  <li class="impact-headline__item">
+    <span class="impact-headline__value">{{ head.papers_building_on_sqlancer }}</span>
+    <span class="impact-headline__label">papers building on SQLancer</span>
+  </li>
+</ul>
+{%- endif -%}
diff --git a/_includes/impact/paper-detail.html b/_includes/impact/paper-detail.html
new file mode 100644
index 0000000..225b2d0
--- /dev/null
+++ b/_includes/impact/paper-detail.html
@@ -0,0 +1,212 @@
+{%- comment -%}
+Everything one paper's record holds about its relationship to SQLancer.
+
+The impact page can only carry a sentence per paper. This carries what that
+sentence was derived from: each classification with the reasoning behind it and
+the paper's own sentences it rests on, every place the paper refers to SQLancer,
+the bibliography entries that resolved to a SQLancer publication, and the
+artifact if one was inspected.
+
+The division of labour matters. The summary and the reasoning are written by a
+model and are labelled as such. The quoted sentences are not: they are stored
+verbatim from the extracted text, and the model's part was to say which of them
+settle a question, never to supply them.
+
+Parameter: key (a record key, matching a file in _data/papers).
+{%- endcomment -%}
+{%- assign record = site.data.papers[include.key] -%}
+{%- assign paper = record.paper -%}
+{%- assign analysis = record.analysis -%}
+
+<div class="impact" markdown="0">
+
+<p class="impact-breadcrumb">
+  <a href="{{ '/impact/#papers' | relative_url }}">&larr; Research building on SQLancer</a>
+</p>
+
+<p class="impact-meta">
+  {%- if paper.authors and paper.authors.size > 0 %}{{ paper.authors | join: ", " }}.{% endif %}
+  {%- if paper.year %} {{ paper.year }}.{% endif %}
+  {%- if paper.venue %} {{ paper.venue }}.{% endif %}
+</p>
+
+<p>
+  {%- if paper.url %}<a href="{{ paper.url }}">Read the paper</a>{% endif -%}
+  {%- if paper.doi %} &middot; <a href="https://doi.org/{{ paper.doi }}">doi:{{ paper.doi }}</a>{% endif -%}
+  {%- if paper.arxiv_id %} &middot; <a href="https://arxiv.org/abs/{{ paper.arxiv_id }}">arXiv:{{ paper.arxiv_id }}</a>{% endif -%}
+</p>
+
+{%- if analysis %}
+<h2 id="about">What this paper does with SQLancer</h2>
+
+<div class="impact-summary">
+  {{ analysis.narrative }}
+  <span class="impact-about">{{ analysis.summary }}</span>
+  <span class="impact-meta">Written by {{ analysis.model }} from the
+  {{ record.mentions.size }} place{% if record.mentions.size != 1 %}s{% endif %}
+  this paper refers to SQLancer. The quotations below are the paper's own
+  words, stored verbatim when the text was extracted.</span>
+</div>
+
+<h2 id="relationships">How it was classified</h2>
+
+{%- assign order = "uses_infrastructure|extends_technique|compares_with|describes_as_state_of_the_art" | split: "|" -%}
+{%- for key in order -%}
+{%- assign entry = analysis.relationships[key] %}
+<div class="impact-adoption">
+  <h3 class="impact-adoption__heading">
+    {{ key | replace: "_", " " }} &mdash; {{ entry.value | replace: "_", " " }}
+    {%- if entry.reuse_kind %} ({{ entry.reuse_kind }}){% endif %}
+  </h3>
+  <p>{{ entry.reasoning }}</p>
+  {%- if entry.techniques and entry.techniques.size > 0 %}
+  <p>
+    {%- for technique in entry.techniques -%}
+      {%- assign definition = site.data.impact.techniques.sqlancer_techniques | where: "id", technique | first -%}
+      <span class="impact-tag">{{ definition.display_name | default: technique }}</span>
+    {%- endfor -%}
+  </p>
+  {%- endif %}
+  {%- for quote in entry.quotes %}
+  <div class="impact-evidence">
+    {%- if quote.text_is_letter_spaced %}
+    <span class="impact-meta">The paper sets this passage with letter-spacing,
+      so extraction returns it one character at a time and there is no readable
+      form to quote.</span>
+    {%- else %}
+    <q>{{ quote.sentence | strip_html }}</q>
+    {%- endif %}
+    <span class="impact-meta">{{ quote.mention_id }}{% if quote.section %} &middot; {{ quote.section }}{% endif %}{% if quote.page %} &middot; page {{ quote.page }}{% endif %}</span>
+  </div>
+  {%- endfor %}
+</div>
+{%- endfor %}
+
+{%- if analysis.disagreements and analysis.disagreements.size > 0 %}
+<h3 id="disagreements">Where this differs from the pattern checks</h3>
+<p>
+  The regular expressions that scan for these relationships are advisory. Where
+  the reading above contradicts one, the reason is recorded.
+</p>
+{%- for item in analysis.disagreements %}
+<div class="impact-evidence">
+  <strong>{{ item.check }}</strong><br>{{ item.why }}
+</div>
+{%- endfor %}
+{%- endif %}
+
+{%- if analysis.unresolved and analysis.unresolved.size > 0 %}
+<h3 id="unresolved">What could not be determined</h3>
+<ul>
+  {%- for item in analysis.unresolved %}
+  <li>{{ item }}</li>
+  {%- endfor %}
+</ul>
+{%- endif %}
+{%- endif %}
+
+{%- if record.artifact and record.artifact.url %}
+<h2 id="artifact">Its artifact</h2>
+<p>
+  <a href="{{ record.artifact.url }}">{{ record.artifact.url | remove: "https://github.com/" }}</a>
+  {%- if record.artifact.markers and record.artifact.markers.size > 0 %}
+  carries {{ record.artifact.markers | join: ", " | replace: "_", " " }}.
+  {%- endif %}
+</p>
+{%- for item in record.artifact.evidence limit: 3 %}
+<div class="impact-evidence">
+  {%- if item.note %}{{ item.note }}<br>{% endif -%}
+  {%- if item.excerpt %}<q>{{ item.excerpt | strip_html | truncate: 300 }}</q><br>{% endif -%}
+  {%- if item.source_url %}<a href="{{ item.source_url }}">source</a>{% endif -%}
+</div>
+{%- endfor %}
+{%- endif %}
+
+{%- if record.sqlancer_references and record.sqlancer_references.size > 0 %}
+<h2 id="references">SQLancer publications it cites ({{ record.sqlancer_references.size }})</h2>
+<p>
+  Bibliography entries that resolved to a SQLancer publication, or to a paper
+  by one of the project's authors. A sentence citing one of these numbers is a
+  reference to SQLancer even when it never writes the name.
+</p>
+<div class="impact-table-wrap">
+  <table class="impact-table">
+    <thead>
+      <tr><th scope="col">#</th><th scope="col">Entry</th><th scope="col">Matched as</th></tr>
+    </thead>
+    <tbody>
+      {%- for reference in record.sqlancer_references %}
+      <tr>
+        <th scope="row">{{ reference.number }}</th>
+        <td>{{ reference.text | truncate: 220 }}</td>
+        <td>
+          {{ reference.matched_as | replace: "_", " " }}
+          {%- if reference.technique %} &middot; {{ reference.technique | upcase }}{% endif %}
+        </td>
+      </tr>
+      {%- endfor %}
+    </tbody>
+  </table>
+</div>
+{%- endif %}
+
+{%- if record.mentions and record.mentions.size > 0 %}
+<h2 id="mentions">Every place it refers to SQLancer ({{ record.mentions.size }})</h2>
+<p>
+  {{ record.mentions.size }} sentence{% if record.mentions.size != 1 %}s{% endif %},
+  each stored verbatim from the extracted text with where it was found and how.
+  “Citation marker” means the sentence names no tool at all and was reached
+  through a reference number that resolved to a SQLancer publication.
+</p>
+<div class="impact-table-wrap">
+  <table class="impact-table impact-table--bugs">
+    <thead>
+      <tr>
+        <th scope="col">Id</th>
+        <th scope="col">Sentence</th>
+        <th scope="col">Found by</th>
+        <th scope="col">Where</th>
+      </tr>
+    </thead>
+    <tbody>
+      {%- for mention in record.mentions %}
+      <tr>
+        <th scope="row">{{ mention.id }}</th>
+        {%- comment -%}
+          A sentence the PDF set with letter-spacing arrives as single
+          characters, and the word breaks are not in the file to recover. It is
+          reported rather than shown: the mention is real and its location is
+          useful, the text is not readable by anyone.
+        {%- endcomment -%}
+        {%- if mention.text_is_letter_spaced %}
+        <td class="impact-meta">The paper sets this passage with letter-spacing,
+          which extraction cannot undo &mdash; the sentence is there, but no
+          readable form of it is.</td>
+        {%- else %}
+        <td>{{ mention.sentence | strip_html | truncate: 320 }}</td>
+        {%- endif %}
+        <td>
+          {{ mention.found_by | replace: "_", " " }}
+          {%- if analysis.roles[mention.id] %}<br><span class="impact-meta">{{ analysis.roles[mention.id] | replace: "_", " " }}</span>{% endif %}
+        </td>
+        <td class="impact-meta">
+          {%- if mention.section %}{{ mention.section }}{% endif %}
+          {%- if mention.page %}{% if mention.section %}<br>{% endif %}page {{ mention.page }}{% endif %}
+        </td>
+      </tr>
+      {%- endfor %}
+    </tbody>
+  </table>
+</div>
+{%- endif %}
+
+<p class="impact-meta">
+  This page is rendered from
+  <a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers/{{ include.key }}.json"><code>_data/papers/{{ include.key }}.json</code></a>,
+  extracted from
+  {%- for source in record.sources %} {{ source.route | replace: "_", " " }}{% unless forloop.last %},{% endunless %}{% endfor %}.
+  {%- if record.document.page_count %} {{ record.document.page_count }} pages,
+  {{ record.references.size }} references parsed.{% endif %}
+</p>
+
+</div>
diff --git a/_includes/impact/paper-list.html b/_includes/impact/paper-list.html
new file mode 100644
index 0000000..a2e5c98
--- /dev/null
+++ b/_includes/impact/paper-list.html
@@ -0,0 +1,113 @@
+{%- comment -%}
+Renders one relationship's papers with the evidence behind that relationship.
+
+Parameters: papers (array of paper records), relationship (the key to show, or
+several separated by "|").
+Only the named relationship's evidence is shown, so a paper listed under
+"compares against SQLancer" quotes the passage about the comparison rather than
+whichever excerpt happens to come first on the record.
+
+Several keys may be given when one heading covers more than one relationship.
+The first the paper actually holds is the one quoted, and the rest are named in
+the "Also" line, so a merged heading never hides which of them applies.
+{%- endcomment -%}
+<ul>
+  {%- assign shown_keys = include.relationship | split: "|" -%}
+  {%- for paper in include.papers %}
+  {%- assign relationship = shown_keys | first -%}
+  {%- for candidate in shown_keys -%}
+    {%- if paper.relationships[candidate].value == "yes" -%}
+      {%- assign relationship = candidate -%}
+      {%- break -%}
+    {%- endif -%}
+  {%- endfor -%}
+  {%- assign entry = paper.relationships[relationship] %}
+  {%- assign record_key = paper.id | replace: ":", "_" | replace: ".", "_" | replace: "/", "_" -%}
+  {%- assign record = site.data.papers[record_key] -%}
+  <li>
+    {%- comment -%}
+      The title leads to this site's page for the paper where there is one,
+      because that page carries the evidence behind the sentence below it. The
+      paper itself is linked from there, and directly here when no record
+      exists to link to.
+    {%- endcomment -%}
+    <strong>
+      {%- if record -%}
+      <a href="{{ '/impact/papers/' | append: record_key | append: '/' | relative_url }}">{{ paper.title }}</a>
+      {%- elsif paper.url -%}
+      <a href="{{ paper.url }}">{{ paper.title }}</a>
+      {%- else -%}
+      {{ paper.title }}
+      {%- endif -%}
+    </strong>
+    {%- if paper.year %} ({{ paper.year }}){% endif %}
+    {%- if paper.venue %} &middot; <span class="impact-meta">{{ paper.venue }}</span>{% endif %}
+    {%- comment -%}
+      Only what the paper did with SQLancer. Everything listed here relates to
+      SQLancer somehow, so "a fuzzer for graph databases" says much less than
+      "its framework is derived from SQLancer" -- and a list of two hundred
+      entries is read by scanning, where a second sentence about the paper
+      itself is what a reader has to skip past. That sentence, and the evidence
+      under it, are on the paper's own page.
+    {%- endcomment -%}
+    {%- assign note = site.data.impact.paper_notes[record_key] -%}
+    {%- if record.analysis %}
+    <div class="impact-summary">
+      {{ record.analysis.narrative }}
+      <span class="impact-meta">Written by {{ record.analysis.model }} from the
+      {{ record.mentions.size }} place{% if record.mentions.size != 1 %}s{% endif %}
+      this paper refers to SQLancer.
+      <a href="{{ '/impact/papers/' | append: record_key | append: '/' | relative_url }}">See the evidence</a>.</span>
+    </div>
+    {%- elsif note.summary %}
+    {%- comment -%}
+      A paper with no record of its own: the older note is all there is, and
+      its relationship line is the same thing the narrative would be.
+    {%- endcomment -%}
+    <div class="impact-summary">
+      {{ note.summary.relationship_to_sqlancer | default: note.summary.text }}
+      <span class="impact-meta">Summary written by {{ note.summary.model }} from
+      {{ note.summary.written_from }}; the quotations below are the paper's own words.</span>
+    </div>
+    {%- endif %}
+    {%- if entry.techniques and entry.techniques.size > 0 %}
+    <br>
+    {%- for technique in entry.techniques -%}
+      {%- assign definition = site.data.impact.techniques.sqlancer_techniques | where: "id", technique | first -%}
+      <span class="impact-tag">{{ definition.display_name | default: technique }}</span>
+    {%- endfor -%}
+    {%- endif %}
+    {%- comment -%}
+      Infrastructure reuse is explained rather than quoted: the underlying
+      evidence is a repository description and a line of Java, and neither says
+      what the connection to SQLancer is.
+    {%- endcomment -%}
+    {%- if relationship == "uses_infrastructure" and paper.artifacts -%}
+      {% include impact/artifact-evidence.html paper=paper %}
+      {%- for item in entry.evidence limit: 1 -%}
+        {%- if item.excerpt and item.source_type == "paper" %}
+    <div class="impact-evidence"><q>{{ item.excerpt | strip_html | truncate: 400 }}</q>
+      &mdash; <a href="{{ item.source_url }}">source</a></div>
+        {%- endif -%}
+      {%- endfor -%}
+    {%- else -%}
+      {%- for item in entry.evidence limit: 2 -%}
+        {%- if item.excerpt %}
+    <div class="impact-evidence"><q>{{ item.excerpt | strip_html | truncate: 400 }}</q>
+      &mdash; <a href="{{ item.source_url }}">source</a></div>
+        {%- endif -%}
+      {%- endfor -%}
+    {%- endif %}
+    {%- assign other = "" | split: "" -%}
+    {%- assign all_keys = "uses_infrastructure|extends_technique|compares_with" | split: "|" -%}
+    {%- for key in all_keys -%}
+      {%- if key != relationship and paper.relationships[key].value == "yes" -%}
+        {%- assign other = other | push: key -%}
+      {%- endif -%}
+    {%- endfor -%}
+    {%- if other.size > 0 %}
+    <div class="impact-meta">Also: {{ other | join: ", " | replace: "_", " " }}.</div>
+    {%- endif %}
+  </li>
+  {%- endfor %}
+</ul>
diff --git a/_includes/impact/plots/bugs-by-dbms.svg b/_includes/impact/plots/bugs-by-dbms.svg
new file mode 100644
index 0000000..b2d1d3a
--- /dev/null
+++ b/_includes/impact/plots/bugs-by-dbms.svg
@@ -0,0 +1,227 @@
+<svg class="impact-chart" viewBox="0 0 626 1216" width="100%" height="auto" style="max-width:626px" role="img" preserveAspectRatio="xMinYMin meet" aria-label="Bugs found by SQLancer, by database system and who found them" xmlns="http://www.w3.org/2000/svg">
+  <title>Bugs found by SQLancer, by database system and who found them</title>
+  <desc>Bugs per database system for 40 categories, largest 302, each split by who found them.</desc>
+  <a href="/impact/dbms/duckdb/" aria-label="DuckDB: 302 bugs">
+  <text class="impact-chart__label" x="142" y="23.0" text-anchor="end" dominant-baseline="central">DuckDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="12" width="376.8" height="22" rx="4" fill="var(--impact-series-1)"><title>DuckDB &#8212; Found by the SQLancer project: 286</title></rect>
+  <rect class="impact-chart__bar" x="528.8" y="12" width="17.9" height="22" rx="4" fill="var(--impact-series-2)"><title>DuckDB &#8212; Found by someone outside the project: 15</title></rect>
+  <rect class="impact-chart__bar" x="548.7" y="12" width="2.0" height="22" rx="4" fill="var(--impact-series-3)"><title>DuckDB &#8212; Reporter not recorded: 1</title></rect>
+  <text class="impact-chart__value" x="558.0" y="23.0" dominant-baseline="central">302</text>
+  </a>
+  <a href="/impact/dbms/sqlite/" aria-label="SQLite: 222 bugs">
+  <text class="impact-chart__label" x="142" y="53.0" text-anchor="end" dominant-baseline="central">SQLite</text>
+  <rect class="impact-chart__bar" x="150.0" y="42" width="292.0" height="22" rx="4" fill="var(--impact-series-1)"><title>SQLite &#8212; Found by the SQLancer project: 222</title></rect>
+  <text class="impact-chart__value" x="452.0" y="53.0" dominant-baseline="central">222</text>
+  </a>
+  <a href="/impact/dbms/cockroachdb/" aria-label="CockroachDB: 202 bugs">
+  <text class="impact-chart__label" x="142" y="83.0" text-anchor="end" dominant-baseline="central">CockroachDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="72" width="166.2" height="22" rx="4" fill="var(--impact-series-1)"><title>CockroachDB &#8212; Found by the SQLancer project: 127</title></rect>
+  <rect class="impact-chart__bar" x="318.2" y="72" width="72.2" height="22" rx="4" fill="var(--impact-series-2)"><title>CockroachDB &#8212; Found by someone outside the project: 56</title></rect>
+  <rect class="impact-chart__bar" x="392.4" y="72" width="23.2" height="22" rx="4" fill="var(--impact-series-3)"><title>CockroachDB &#8212; Reporter not recorded: 19</title></rect>
+  <text class="impact-chart__value" x="425.5" y="83.0" dominant-baseline="central">202</text>
+  </a>
+  <a href="/impact/dbms/starrocks/" aria-label="StarRocks: 183 bugs">
+  <text class="impact-chart__label" x="142" y="113.0" text-anchor="end" dominant-baseline="central">StarRocks</text>
+  <rect class="impact-chart__bar" x="150.0" y="102" width="240.4" height="22" rx="4" fill="var(--impact-series-2)"><title>StarRocks &#8212; Found by someone outside the project: 183</title></rect>
+  <text class="impact-chart__value" x="400.4" y="113.0" dominant-baseline="central">183</text>
+  </a>
+  <a href="/impact/dbms/tidb/" aria-label="TiDB: 172 bugs">
+  <text class="impact-chart__label" x="142" y="143.0" text-anchor="end" dominant-baseline="central">TiDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="132" width="202.0" height="22" rx="4" fill="var(--impact-series-1)"><title>TiDB &#8212; Found by the SQLancer project: 154</title></rect>
+  <rect class="impact-chart__bar" x="354.0" y="132" width="7.3" height="22" rx="4" fill="var(--impact-series-2)"><title>TiDB &#8212; Found by someone outside the project: 7</title></rect>
+  <rect class="impact-chart__bar" x="363.2" y="132" width="12.6" height="22" rx="4" fill="var(--impact-series-3)"><title>TiDB &#8212; Reporter not recorded: 11</title></rect>
+  <text class="impact-chart__value" x="385.8" y="143.0" dominant-baseline="central">172</text>
+  </a>
+  <a href="/impact/dbms/dolt/" aria-label="Dolt: 167 bugs">
+  <text class="impact-chart__label" x="142" y="173.0" text-anchor="end" dominant-baseline="central">Dolt</text>
+  <rect class="impact-chart__bar" x="150.0" y="162" width="219.2" height="22" rx="4" fill="var(--impact-series-1)"><title>Dolt &#8212; Found by the SQLancer project: 167</title></rect>
+  <text class="impact-chart__value" x="379.2" y="173.0" dominant-baseline="central">167</text>
+  </a>
+  <a href="/impact/dbms/monetdb/" aria-label="MonetDB: 139 bugs">
+  <text class="impact-chart__label" x="142" y="203.0" text-anchor="end" dominant-baseline="central">MonetDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="192" width="126.5" height="22" rx="4" fill="var(--impact-series-1)"><title>MonetDB &#8212; Found by the SQLancer project: 97</title></rect>
+  <rect class="impact-chart__bar" x="278.5" y="192" width="53.6" height="22" rx="4" fill="var(--impact-series-2)"><title>MonetDB &#8212; Found by someone outside the project: 42</title></rect>
+  <text class="impact-chart__value" x="342.1" y="203.0" dominant-baseline="central">139</text>
+  </a>
+  <a href="/impact/dbms/cratedb/" aria-label="CrateDB: 81 bugs">
+  <text class="impact-chart__label" x="142" y="233.0" text-anchor="end" dominant-baseline="central">CrateDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="222" width="105.3" height="22" rx="4" fill="var(--impact-series-1)"><title>CrateDB &#8212; Found by the SQLancer project: 81</title></rect>
+  <text class="impact-chart__value" x="265.3" y="233.0" dominant-baseline="central">81</text>
+  </a>
+  <a href="/impact/dbms/questdb/" aria-label="QuestDB: 55 bugs">
+  <text class="impact-chart__label" x="142" y="263.0" text-anchor="end" dominant-baseline="central">QuestDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="252" width="68.2" height="22" rx="4" fill="var(--impact-series-1)"><title>QuestDB &#8212; Found by the SQLancer project: 53</title></rect>
+  <rect class="impact-chart__bar" x="220.2" y="252" width="2.0" height="22" rx="4" fill="var(--impact-series-2)"><title>QuestDB &#8212; Found by someone outside the project: 2</title></rect>
+  <text class="impact-chart__value" x="230.8" y="263.0" dominant-baseline="central">55</text>
+  </a>
+  <a href="/impact/dbms/umbra/" aria-label="Umbra: 55 bugs">
+  <text class="impact-chart__label" x="142" y="293.0" text-anchor="end" dominant-baseline="central">Umbra</text>
+  <rect class="impact-chart__bar" x="150.0" y="282" width="70.8" height="22" rx="4" fill="var(--impact-series-1)"><title>Umbra &#8212; Found by the SQLancer project: 55</title></rect>
+  <text class="impact-chart__value" x="230.8" y="293.0" dominant-baseline="central">55</text>
+  </a>
+  <a href="/impact/dbms/datafusion/" aria-label="Apache DataFusion: 54 bugs">
+  <text class="impact-chart__label" x="142" y="323.0" text-anchor="end" dominant-baseline="central">Apache DataFusion</text>
+  <rect class="impact-chart__bar" x="150.0" y="312" width="69.5" height="22" rx="4" fill="var(--impact-series-2)"><title>Apache DataFusion &#8212; Found by someone outside the project: 54</title></rect>
+  <text class="impact-chart__value" x="229.5" y="323.0" dominant-baseline="central">54</text>
+  </a>
+  <a href="/impact/dbms/yugabytedb/" aria-label="YugabyteDB: 54 bugs">
+  <text class="impact-chart__label" x="142" y="353.0" text-anchor="end" dominant-baseline="central">YugabyteDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="342" width="15.2" height="22" rx="4" fill="var(--impact-series-1)"><title>YugabyteDB &#8212; Found by the SQLancer project: 13</title></rect>
+  <rect class="impact-chart__bar" x="167.2" y="342" width="49.7" height="22" rx="4" fill="var(--impact-series-2)"><title>YugabyteDB &#8212; Found by someone outside the project: 39</title></rect>
+  <rect class="impact-chart__bar" x="218.9" y="342" width="2.0" height="22" rx="4" fill="var(--impact-series-3)"><title>YugabyteDB &#8212; Reporter not recorded: 2</title></rect>
+  <text class="impact-chart__value" x="229.5" y="353.0" dominant-baseline="central">54</text>
+  </a>
+  <a href="/impact/dbms/clickhouse/" aria-label="ClickHouse: 48 bugs">
+  <text class="impact-chart__label" x="142" y="383.0" text-anchor="end" dominant-baseline="central">ClickHouse</text>
+  <rect class="impact-chart__bar" x="150.0" y="372" width="7.3" height="22" rx="4" fill="var(--impact-series-1)"><title>ClickHouse &#8212; Found by the SQLancer project: 7</title></rect>
+  <rect class="impact-chart__bar" x="159.3" y="372" width="52.3" height="22" rx="4" fill="var(--impact-series-2)"><title>ClickHouse &#8212; Found by someone outside the project: 41</title></rect>
+  <text class="impact-chart__value" x="221.6" y="383.0" dominant-baseline="central">48</text>
+  </a>
+  <a href="/impact/dbms/mysql/" aria-label="MySQL: 43 bugs">
+  <text class="impact-chart__label" x="142" y="413.0" text-anchor="end" dominant-baseline="central">MySQL</text>
+  <rect class="impact-chart__bar" x="150.0" y="402" width="40.4" height="22" rx="4" fill="var(--impact-series-1)"><title>MySQL &#8212; Found by the SQLancer project: 32</title></rect>
+  <rect class="impact-chart__bar" x="192.4" y="402" width="12.6" height="22" rx="4" fill="var(--impact-series-3)"><title>MySQL &#8212; Reporter not recorded: 11</title></rect>
+  <text class="impact-chart__value" x="215.0" y="413.0" dominant-baseline="central">43</text>
+  </a>
+  <a href="/impact/dbms/h2/" aria-label="H2: 20 bugs">
+  <text class="impact-chart__label" x="142" y="443.0" text-anchor="end" dominant-baseline="central">H2</text>
+  <rect class="impact-chart__bar" x="150.0" y="432" width="24.5" height="22" rx="4" fill="var(--impact-series-1)"><title>H2 &#8212; Found by the SQLancer project: 20</title></rect>
+  <text class="impact-chart__value" x="184.5" y="443.0" dominant-baseline="central">20</text>
+  </a>
+  <a href="/impact/dbms/doris/" aria-label="Apache Doris: 19 bugs">
+  <text class="impact-chart__label" x="142" y="473.0" text-anchor="end" dominant-baseline="central">Apache Doris</text>
+  <rect class="impact-chart__bar" x="150.0" y="462" width="2.0" height="22" rx="4" fill="var(--impact-series-1)"><title>Apache Doris &#8212; Found by the SQLancer project: 3</title></rect>
+  <rect class="impact-chart__bar" x="154.0" y="462" width="9.9" height="22" rx="4" fill="var(--impact-series-2)"><title>Apache Doris &#8212; Found by someone outside the project: 9</title></rect>
+  <rect class="impact-chart__bar" x="165.9" y="462" width="7.3" height="22" rx="4" fill="var(--impact-series-3)"><title>Apache Doris &#8212; Reporter not recorded: 7</title></rect>
+  <text class="impact-chart__value" x="183.2" y="473.0" dominant-baseline="central">19</text>
+  </a>
+  <a href="/impact/dbms/databend/" aria-label="Databend: 18 bugs">
+  <text class="impact-chart__label" x="142" y="503.0" text-anchor="end" dominant-baseline="central">Databend</text>
+  <rect class="impact-chart__bar" x="150.0" y="492" width="7.3" height="22" rx="4" fill="var(--impact-series-1)"><title>Databend &#8212; Found by the SQLancer project: 7</title></rect>
+  <rect class="impact-chart__bar" x="159.3" y="492" width="12.6" height="22" rx="4" fill="var(--impact-series-2)"><title>Databend &#8212; Found by someone outside the project: 11</title></rect>
+  <text class="impact-chart__value" x="181.8" y="503.0" dominant-baseline="central">18</text>
+  </a>
+  <a href="/impact/dbms/postgresql/" aria-label="PostgreSQL: 18 bugs">
+  <text class="impact-chart__label" x="142" y="533.0" text-anchor="end" dominant-baseline="central">PostgreSQL</text>
+  <rect class="impact-chart__bar" x="150.0" y="522" width="21.8" height="22" rx="4" fill="var(--impact-series-1)"><title>PostgreSQL &#8212; Found by the SQLancer project: 18</title></rect>
+  <text class="impact-chart__value" x="181.8" y="533.0" dominant-baseline="central">18</text>
+  </a>
+  <a href="/impact/dbms/turso/" aria-label="Turso: 18 bugs">
+  <text class="impact-chart__label" x="142" y="563.0" text-anchor="end" dominant-baseline="central">Turso</text>
+  <rect class="impact-chart__bar" x="150.0" y="552" width="21.8" height="22" rx="4" fill="var(--impact-series-2)"><title>Turso &#8212; Found by someone outside the project: 18</title></rect>
+  <text class="impact-chart__value" x="181.8" y="563.0" dominant-baseline="central">18</text>
+  </a>
+  <a href="/impact/dbms/virtuoso/" aria-label="Virtuoso: 17 bugs">
+  <text class="impact-chart__label" x="142" y="593.0" text-anchor="end" dominant-baseline="central">Virtuoso</text>
+  <rect class="impact-chart__bar" x="150.0" y="582" width="12.6" height="22" rx="4" fill="var(--impact-series-1)"><title>Virtuoso &#8212; Found by the SQLancer project: 11</title></rect>
+  <rect class="impact-chart__bar" x="164.6" y="582" width="5.9" height="22" rx="4" fill="var(--impact-series-2)"><title>Virtuoso &#8212; Found by someone outside the project: 6</title></rect>
+  <text class="impact-chart__value" x="180.5" y="593.0" dominant-baseline="central">17</text>
+  </a>
+  <a href="/impact/dbms/citus/" aria-label="Citus: 16 bugs">
+  <text class="impact-chart__label" x="142" y="623.0" text-anchor="end" dominant-baseline="central">Citus</text>
+  <rect class="impact-chart__bar" x="150.0" y="612" width="2.0" height="22" rx="4" fill="var(--impact-series-1)"><title>Citus &#8212; Found by the SQLancer project: 1</title></rect>
+  <rect class="impact-chart__bar" x="151.3" y="612" width="7.3" height="22" rx="4" fill="var(--impact-series-2)"><title>Citus &#8212; Found by someone outside the project: 7</title></rect>
+  <rect class="impact-chart__bar" x="160.6" y="612" width="8.6" height="22" rx="4" fill="var(--impact-series-3)"><title>Citus &#8212; Reporter not recorded: 8</title></rect>
+  <text class="impact-chart__value" x="179.2" y="623.0" dominant-baseline="central">16</text>
+  </a>
+  <a href="/impact/dbms/stonedb/" aria-label="StoneDB: 16 bugs">
+  <text class="impact-chart__label" x="142" y="653.0" text-anchor="end" dominant-baseline="central">StoneDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="642" width="13.9" height="22" rx="4" fill="var(--impact-series-1)"><title>StoneDB &#8212; Found by the SQLancer project: 12</title></rect>
+  <rect class="impact-chart__bar" x="165.9" y="642" width="3.3" height="22" rx="4" fill="var(--impact-series-3)"><title>StoneDB &#8212; Reporter not recorded: 4</title></rect>
+  <text class="impact-chart__value" x="179.2" y="653.0" dominant-baseline="central">16</text>
+  </a>
+  <a href="/impact/dbms/wadjet/" aria-label="Wadjet: 16 bugs">
+  <text class="impact-chart__label" x="142" y="683.0" text-anchor="end" dominant-baseline="central">Wadjet</text>
+  <rect class="impact-chart__bar" x="150.0" y="672" width="19.2" height="22" rx="4" fill="var(--impact-series-2)"><title>Wadjet &#8212; Found by someone outside the project: 16</title></rect>
+  <text class="impact-chart__value" x="179.2" y="683.0" dominant-baseline="central">16</text>
+  </a>
+  <a href="/impact/dbms/mariadb/" aria-label="MariaDB: 15 bugs">
+  <text class="impact-chart__label" x="142" y="713.0" text-anchor="end" dominant-baseline="central">MariaDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="702" width="5.9" height="22" rx="4" fill="var(--impact-series-1)"><title>MariaDB &#8212; Found by the SQLancer project: 6</title></rect>
+  <rect class="impact-chart__bar" x="157.9" y="702" width="2.0" height="22" rx="4" fill="var(--impact-series-2)"><title>MariaDB &#8212; Found by someone outside the project: 1</title></rect>
+  <rect class="impact-chart__bar" x="159.3" y="702" width="8.6" height="22" rx="4" fill="var(--impact-series-3)"><title>MariaDB &#8212; Reporter not recorded: 8</title></rect>
+  <text class="impact-chart__value" x="177.9" y="713.0" dominant-baseline="central">15</text>
+  </a>
+  <a href="/impact/dbms/oxla/" aria-label="Oxla: 12 bugs">
+  <text class="impact-chart__label" x="142" y="743.0" text-anchor="end" dominant-baseline="central">Oxla</text>
+  <rect class="impact-chart__bar" x="150.0" y="732" width="13.9" height="22" rx="4" fill="var(--impact-series-3)"><title>Oxla &#8212; Reporter not recorded: 12</title></rect>
+  <text class="impact-chart__value" x="173.9" y="743.0" dominant-baseline="central">12</text>
+  </a>
+  <a href="/impact/dbms/tdengine/" aria-label="TDengine: 12 bugs">
+  <text class="impact-chart__label" x="142" y="773.0" text-anchor="end" dominant-baseline="central">TDengine</text>
+  <rect class="impact-chart__bar" x="150.0" y="762" width="13.9" height="22" rx="4" fill="var(--impact-series-1)"><title>TDengine &#8212; Found by the SQLancer project: 12</title></rect>
+  <text class="impact-chart__value" x="173.9" y="773.0" dominant-baseline="central">12</text>
+  </a>
+  <a href="/impact/dbms/firebird/" aria-label="Firebird: 10 bugs">
+  <text class="impact-chart__label" x="142" y="803.0" text-anchor="end" dominant-baseline="central">Firebird</text>
+  <rect class="impact-chart__bar" x="150.0" y="792" width="11.2" height="22" rx="4" fill="var(--impact-series-1)"><title>Firebird &#8212; Found by the SQLancer project: 10</title></rect>
+  <text class="impact-chart__value" x="171.2" y="803.0" dominant-baseline="central">10</text>
+  </a>
+  <a href="/impact/dbms/matrixone/" aria-label="MatrixOne: 9 bugs">
+  <text class="impact-chart__label" x="142" y="833.0" text-anchor="end" dominant-baseline="central">MatrixOne</text>
+  <rect class="impact-chart__bar" x="150.0" y="822" width="9.9" height="22" rx="4" fill="var(--impact-series-2)"><title>MatrixOne &#8212; Found by someone outside the project: 9</title></rect>
+  <text class="impact-chart__value" x="169.9" y="833.0" dominant-baseline="central">9</text>
+  </a>
+  <a href="/impact/dbms/risingwave/" aria-label="RisingWave: 8 bugs">
+  <text class="impact-chart__label" x="142" y="863.0" text-anchor="end" dominant-baseline="central">RisingWave</text>
+  <rect class="impact-chart__bar" x="150.0" y="852" width="8.6" height="22" rx="4" fill="var(--impact-series-1)"><title>RisingWave &#8212; Found by the SQLancer project: 8</title></rect>
+  <text class="impact-chart__value" x="168.6" y="863.0" dominant-baseline="central">8</text>
+  </a>
+  <a href="/impact/dbms/bharatdbms/" aria-label="BharatDBMS: 7 bugs">
+  <text class="impact-chart__label" x="142" y="893.0" text-anchor="end" dominant-baseline="central">BharatDBMS</text>
+  <rect class="impact-chart__bar" x="150.0" y="882" width="7.3" height="22" rx="4" fill="var(--impact-series-2)"><title>BharatDBMS &#8212; Found by someone outside the project: 7</title></rect>
+  <text class="impact-chart__value" x="167.3" y="893.0" dominant-baseline="central">7</text>
+  </a>
+  <a href="/impact/dbms/cnosdb/" aria-label="CnosDB: 7 bugs">
+  <text class="impact-chart__label" x="142" y="923.0" text-anchor="end" dominant-baseline="central">CnosDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="912" width="2.0" height="22" rx="4" fill="var(--impact-series-1)"><title>CnosDB &#8212; Found by the SQLancer project: 1</title></rect>
+  <rect class="impact-chart__bar" x="151.3" y="912" width="3.3" height="22" rx="4" fill="var(--impact-series-2)"><title>CnosDB &#8212; Found by someone outside the project: 4</title></rect>
+  <rect class="impact-chart__bar" x="156.6" y="912" width="2.0" height="22" rx="4" fill="var(--impact-series-3)"><title>CnosDB &#8212; Reporter not recorded: 2</title></rect>
+  <text class="impact-chart__value" x="167.3" y="923.0" dominant-baseline="central">7</text>
+  </a>
+  <a href="/impact/dbms/cloudberry/" aria-label="Apache Cloudberry: 6 bugs">
+  <text class="impact-chart__label" x="142" y="953.0" text-anchor="end" dominant-baseline="central">Apache Cloudberry</text>
+  <rect class="impact-chart__bar" x="150.0" y="942" width="5.9" height="22" rx="4" fill="var(--impact-series-2)"><title>Apache Cloudberry &#8212; Found by someone outside the project: 6</title></rect>
+  <text class="impact-chart__value" x="165.9" y="953.0" dominant-baseline="central">6</text>
+  </a>
+  <a href="/impact/dbms/oceanbase/" aria-label="OceanBase: 4 bugs">
+  <text class="impact-chart__label" x="142" y="983.0" text-anchor="end" dominant-baseline="central">OceanBase</text>
+  <rect class="impact-chart__bar" x="150.0" y="972" width="3.3" height="22" rx="4" fill="var(--impact-series-1)"><title>OceanBase &#8212; Found by the SQLancer project: 4</title></rect>
+  <text class="impact-chart__value" x="163.3" y="983.0" dominant-baseline="central">4</text>
+  </a>
+  <a href="/impact/dbms/presto/" aria-label="Presto: 4 bugs">
+  <text class="impact-chart__label" x="142" y="1013.0" text-anchor="end" dominant-baseline="central">Presto</text>
+  <rect class="impact-chart__bar" x="150.0" y="1002" width="3.3" height="22" rx="4" fill="var(--impact-series-3)"><title>Presto &#8212; Reporter not recorded: 4</title></rect>
+  <text class="impact-chart__value" x="163.3" y="1013.0" dominant-baseline="central">4</text>
+  </a>
+  <a href="/impact/dbms/seekdb/" aria-label="SeekDB: 3 bugs">
+  <text class="impact-chart__label" x="142" y="1043.0" text-anchor="end" dominant-baseline="central">SeekDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="1032" width="2.0" height="22" rx="4" fill="var(--impact-series-2)"><title>SeekDB &#8212; Found by someone outside the project: 3</title></rect>
+  <text class="impact-chart__value" x="162.0" y="1043.0" dominant-baseline="central">3</text>
+  </a>
+  <a href="/impact/dbms/cubrid/" aria-label="CUBRID: 1 bugs">
+  <text class="impact-chart__label" x="142" y="1073.0" text-anchor="end" dominant-baseline="central">CUBRID</text>
+  <rect class="impact-chart__bar" x="150.0" y="1062" width="2.0" height="22" rx="4" fill="var(--impact-series-1)"><title>CUBRID &#8212; Found by the SQLancer project: 1</title></rect>
+  <text class="impact-chart__value" x="159.3" y="1073.0" dominant-baseline="central">1</text>
+  </a>
+  <a href="/impact/dbms/falkordb/" aria-label="FalkorDB: 1 bugs">
+  <text class="impact-chart__label" x="142" y="1103.0" text-anchor="end" dominant-baseline="central">FalkorDB</text>
+  <rect class="impact-chart__bar" x="150.0" y="1092" width="2.0" height="22" rx="4" fill="var(--impact-series-2)"><title>FalkorDB &#8212; Found by someone outside the project: 1</title></rect>
+  <text class="impact-chart__value" x="159.3" y="1103.0" dominant-baseline="central">1</text>
+  </a>
+  <a href="/impact/dbms/hazelcast/" aria-label="Hazelcast: 1 bugs">
+  <text class="impact-chart__label" x="142" y="1133.0" text-anchor="end" dominant-baseline="central">Hazelcast</text>
+  <rect class="impact-chart__bar" x="150.0" y="1122" width="2.0" height="22" rx="4" fill="var(--impact-series-2)"><title>Hazelcast &#8212; Found by someone outside the project: 1</title></rect>
+  <text class="impact-chart__value" x="159.3" y="1133.0" dominant-baseline="central">1</text>
+  </a>
+  <a href="/impact/dbms/kyzo/" aria-label="Kyzo: 1 bugs">
+  <text class="impact-chart__label" x="142" y="1163.0" text-anchor="end" dominant-baseline="central">Kyzo</text>
+  <rect class="impact-chart__bar" x="150.0" y="1152" width="2.0" height="22" rx="4" fill="var(--impact-series-2)"><title>Kyzo &#8212; Found by someone outside the project: 1</title></rect>
+  <text class="impact-chart__value" x="159.3" y="1163.0" dominant-baseline="central">1</text>
+  </a>
+  <a href="/impact/dbms/sparq/" aria-label="sparq: 1 bugs">
+  <text class="impact-chart__label" x="142" y="1193.0" text-anchor="end" dominant-baseline="central">sparq</text>
+  <rect class="impact-chart__bar" x="150.0" y="1182" width="2.0" height="22" rx="4" fill="var(--impact-series-2)"><title>sparq &#8212; Found by someone outside the project: 1</title></rect>
+  <text class="impact-chart__value" x="159.3" y="1193.0" dominant-baseline="central">1</text>
+  </a>
+</svg>
diff --git a/_includes/impact/plots/bugs-by-reporter.svg b/_includes/impact/plots/bugs-by-reporter.svg
new file mode 100644
index 0000000..f569864
--- /dev/null
+++ b/_includes/impact/plots/bugs-by-reporter.svg
@@ -0,0 +1,9 @@
+<svg class="impact-chart" viewBox="0 0 660 76" width="100%" height="auto" style="max-width:660px" role="img" preserveAspectRatio="xMinYMin meet" aria-label="Who found the bugs SQLancer is credited with" xmlns="http://www.w3.org/2000/svg">
+  <title>Who found the bugs SQLancer is credited with</title>
+  <desc>Bug reports by reporter affiliation: 2038 in 3 states.</desc>
+  <rect class="impact-chart__bar" x="0.0" y="16" width="454.0" height="34" rx="4" fill="var(--impact-series-1)"><title>Found by the SQLancer project: 1408</title></rect>
+  <text class="impact-chart__value" x="227.0" y="66" text-anchor="middle">Found by the SQLancer project 1408</text>
+  <rect class="impact-chart__bar" x="456.0" y="16" width="173.2" height="34" rx="4" fill="var(--impact-series-2)"><title>Found by someone outside the project: 541</title></rect>
+  <text class="impact-chart__value" x="542.6" y="66" text-anchor="middle">Found by someone outside the project 541</text>
+  <rect class="impact-chart__bar" x="631.2" y="16" width="26.8" height="34" rx="4" fill="var(--impact-series-3)"><title>Reporter not recorded: 89</title></rect>
+</svg>
diff --git a/_includes/impact/plots/bugs-by-status.svg b/_includes/impact/plots/bugs-by-status.svg
new file mode 100644
index 0000000..adf5aab
--- /dev/null
+++ b/_includes/impact/plots/bugs-by-status.svg
@@ -0,0 +1,9 @@
+<svg class="impact-chart" viewBox="0 0 660 76" width="100%" height="auto" style="max-width:660px" role="img" preserveAspectRatio="xMinYMin meet" aria-label="Status of the bugs SQLancer found" xmlns="http://www.w3.org/2000/svg">
+  <title>Status of the bugs SQLancer found</title>
+  <desc>Bug reports by status: 2007 in 4 states.</desc>
+  <rect class="impact-chart__bar" x="0.0" y="16" width="532.1" height="34" rx="4" fill="var(--impact-series-1)"><title>Fixed: 1624</title></rect>
+  <text class="impact-chart__value" x="266.0" y="66" text-anchor="middle">Fixed 1624</text>
+  <rect class="impact-chart__bar" x="534.1" y="16" width="2.3" height="34" rx="4" fill="var(--impact-series-2)"><title>Fixed in documentation: 13</title></rect>
+  <rect class="impact-chart__bar" x="538.3" y="16" width="31.5" height="34" rx="4" fill="var(--impact-series-3)"><title>Confirmed: 102</title></rect>
+  <rect class="impact-chart__bar" x="571.9" y="16" width="86.1" height="34" rx="4" fill="var(--impact-series-4)"><title>Open: 268</title></rect>
+</svg>
diff --git a/_includes/impact/plots/bugs-by-technique.svg b/_includes/impact/plots/bugs-by-technique.svg
new file mode 100644
index 0000000..94041e8
--- /dev/null
+++ b/_includes/impact/plots/bugs-by-technique.svg
@@ -0,0 +1,25 @@
+<svg class="impact-chart" viewBox="0 0 646 226" width="100%" height="auto" style="max-width:646px" role="img" preserveAspectRatio="xMinYMin meet" aria-label="Bugs found by SQLancer, by technique" xmlns="http://www.w3.org/2000/svg">
+  <title>Bugs found by SQLancer, by technique</title>
+  <desc>Bugs per technique for 7 categories, largest 1690.</desc>
+  <text class="impact-chart__label" x="142" y="23.0" text-anchor="end" dominant-baseline="central">Technique not recorded</text>
+  <rect class="impact-chart__bar" x="150" y="12" width="420" height="22" rx="4" fill="var(--impact-sequential)"><title>Technique not recorded: 1690</title></rect>
+  <text class="impact-chart__value" x="578" y="23.0" dominant-baseline="central">1690</text>
+  <text class="impact-chart__label" x="142" y="53.0" text-anchor="end" dominant-baseline="central">Ternary Logic Partiti...</text>
+  <rect class="impact-chart__bar" x="150" y="42" width="49" height="22" rx="4" fill="var(--impact-sequential)"><title>Ternary Logic Partitioning (TLP): 199</title></rect>
+  <text class="impact-chart__value" x="207" y="53.0" dominant-baseline="central">199</text>
+  <text class="impact-chart__label" x="142" y="83.0" text-anchor="end" dominant-baseline="central">Non-optimizing Refere...</text>
+  <rect class="impact-chart__bar" x="150" y="72" width="19" height="22" rx="4" fill="var(--impact-sequential)"><title>Non-optimizing Reference Engine Construction (NoREC): 77</title></rect>
+  <text class="impact-chart__value" x="177" y="83.0" dominant-baseline="central">77</text>
+  <text class="impact-chart__label" x="142" y="113.0" text-anchor="end" dominant-baseline="central">Pivoted Query Synthes...</text>
+  <rect class="impact-chart__bar" x="150" y="102" width="17" height="22" rx="4" fill="var(--impact-sequential)"><title>Pivoted Query Synthesis (PQS): 67</title></rect>
+  <text class="impact-chart__value" x="175" y="113.0" dominant-baseline="central">67</text>
+  <text class="impact-chart__label" x="142" y="143.0" text-anchor="end" dominant-baseline="central">Differential Query Pl...</text>
+  <rect class="impact-chart__bar" x="150" y="132" width="2" height="22" rx="4" fill="var(--impact-sequential)"><title>Differential Query Plans (DQP): 3</title></rect>
+  <text class="impact-chart__value" x="160" y="143.0" dominant-baseline="central">3</text>
+  <text class="impact-chart__label" x="142" y="173.0" text-anchor="end" dominant-baseline="central">Constant-Optimization...</text>
+  <rect class="impact-chart__bar" x="150" y="162" width="2" height="22" rx="4" fill="var(--impact-sequential)"><title>Constant-Optimization-Driven Testing (CODDTest): 1</title></rect>
+  <text class="impact-chart__value" x="160" y="173.0" dominant-baseline="central">1</text>
+  <text class="impact-chart__label" x="142" y="203.0" text-anchor="end" dominant-baseline="central">Query Plan Guidance (...</text>
+  <rect class="impact-chart__bar" x="150" y="192" width="2" height="22" rx="4" fill="var(--impact-sequential)"><title>Query Plan Guidance (QPG): 1</title></rect>
+  <text class="impact-chart__value" x="160" y="203.0" dominant-baseline="central">1</text>
+</svg>
diff --git a/_includes/impact/plots/bugs-by-year.svg b/_includes/impact/plots/bugs-by-year.svg
new file mode 100644
index 0000000..350e55f
--- /dev/null
+++ b/_includes/impact/plots/bugs-by-year.svg
@@ -0,0 +1,55 @@
+<svg class="impact-chart" viewBox="0 0 584 264" width="100%" height="auto" style="max-width:584px" role="img" preserveAspectRatio="xMinYMin meet" aria-label="Bugs found by SQLancer, by year reported" xmlns="http://www.w3.org/2000/svg">
+  <title>Bugs found by SQLancer, by year reported</title>
+  <desc>Bugs reported per year across 8 periods, peak 411.</desc>
+  <line class="impact-chart__grid" x1="44" y1="222.0" x2="572" y2="222.0" />
+  <text class="impact-chart__tick" x="36" y="222.0" text-anchor="end" dominant-baseline="central">0</text>
+  <line class="impact-chart__grid" x1="44" y1="169.5" x2="572" y2="169.5" />
+  <text class="impact-chart__tick" x="36" y="169.5" text-anchor="end" dominant-baseline="central">150</text>
+  <line class="impact-chart__grid" x1="44" y1="117.0" x2="572" y2="117.0" />
+  <text class="impact-chart__tick" x="36" y="117.0" text-anchor="end" dominant-baseline="central">300</text>
+  <line class="impact-chart__grid" x1="44" y1="64.5" x2="572" y2="64.5" />
+  <text class="impact-chart__tick" x="36" y="64.5" text-anchor="end" dominant-baseline="central">450</text>
+  <line class="impact-chart__grid" x1="44" y1="12.0" x2="572" y2="12.0" />
+  <text class="impact-chart__tick" x="36" y="12.0" text-anchor="end" dominant-baseline="central">600</text>
+  <line class="impact-chart__axis" x1="44" y1="222" x2="572" y2="222" />
+  <a href="/impact/bugs/2019/" aria-label="2019: 221">
+  <rect class="impact-chart__bar" x="49.0" y="145" width="56" height="77" rx="4" fill="var(--impact-sequential)"><title>2019: 221</title></rect>
+  <text class="impact-chart__value" x="77.0" y="139" text-anchor="middle">221</text>
+  <text class="impact-chart__label" x="77.0" y="240" text-anchor="middle">2019</text>
+  </a>
+  <a href="/impact/bugs/2020/" aria-label="2020: 302">
+  <rect class="impact-chart__bar" x="115.0" y="116" width="56" height="106" rx="4" fill="var(--impact-sequential)"><title>2020: 302</title></rect>
+  <text class="impact-chart__value" x="143.0" y="110" text-anchor="middle">302</text>
+  <text class="impact-chart__label" x="143.0" y="240" text-anchor="middle">2020</text>
+  </a>
+  <a href="/impact/bugs/2021/" aria-label="2021: 22">
+  <rect class="impact-chart__bar" x="181.0" y="214" width="56" height="8" rx="4" fill="var(--impact-sequential)"><title>2021: 22</title></rect>
+  <text class="impact-chart__value" x="209.0" y="208" text-anchor="middle">22</text>
+  <text class="impact-chart__label" x="209.0" y="240" text-anchor="middle">2021</text>
+  </a>
+  <a href="/impact/bugs/2022/" aria-label="2022: 224">
+  <rect class="impact-chart__bar" x="247.0" y="144" width="56" height="78" rx="4" fill="var(--impact-sequential)"><title>2022: 224</title></rect>
+  <text class="impact-chart__value" x="275.0" y="138" text-anchor="middle">224</text>
+  <text class="impact-chart__label" x="275.0" y="240" text-anchor="middle">2022</text>
+  </a>
+  <a href="/impact/bugs/2023/" aria-label="2023: 271">
+  <rect class="impact-chart__bar" x="313.0" y="127" width="56" height="95" rx="4" fill="var(--impact-sequential)"><title>2023: 271</title></rect>
+  <text class="impact-chart__value" x="341.0" y="121" text-anchor="middle">271</text>
+  <text class="impact-chart__label" x="341.0" y="240" text-anchor="middle">2023</text>
+  </a>
+  <a href="/impact/bugs/2024/" aria-label="2024: 193">
+  <rect class="impact-chart__bar" x="379.0" y="154" width="56" height="68" rx="4" fill="var(--impact-sequential)"><title>2024: 193</title></rect>
+  <text class="impact-chart__value" x="407.0" y="148" text-anchor="middle">193</text>
+  <text class="impact-chart__label" x="407.0" y="240" text-anchor="middle">2024</text>
+  </a>
+  <a href="/impact/bugs/2025/" aria-label="2025: 264">
+  <rect class="impact-chart__bar" x="445.0" y="130" width="56" height="92" rx="4" fill="var(--impact-sequential)"><title>2025: 264</title></rect>
+  <text class="impact-chart__value" x="473.0" y="124" text-anchor="middle">264</text>
+  <text class="impact-chart__label" x="473.0" y="240" text-anchor="middle">2025</text>
+  </a>
+  <a href="/impact/bugs/2026/" aria-label="2026: 411">
+  <rect class="impact-chart__bar" x="511.0" y="78" width="56" height="144" rx="4" fill="var(--impact-sequential)"><title>2026: 411</title></rect>
+  <text class="impact-chart__value" x="539.0" y="72" text-anchor="middle">411</text>
+  <text class="impact-chart__label" x="539.0" y="240" text-anchor="middle">2026</text>
+  </a>
+</svg>
diff --git a/_includes/impact/plots/papers-by-year.svg b/_includes/impact/plots/papers-by-year.svg
new file mode 100644
index 0000000..789fab8
--- /dev/null
+++ b/_includes/impact/plots/papers-by-year.svg
@@ -0,0 +1,36 @@
+<svg class="impact-chart" viewBox="0 0 518 264" width="100%" height="auto" style="max-width:518px" role="img" preserveAspectRatio="xMinYMin meet" aria-label="Papers citing SQLancer, by publication year" xmlns="http://www.w3.org/2000/svg">
+  <title>Papers citing SQLancer, by publication year</title>
+  <desc>Papers citing SQLancer per year across 7 periods, peak 65.</desc>
+  <line class="impact-chart__grid" x1="44" y1="222.0" x2="506" y2="222.0" />
+  <text class="impact-chart__tick" x="36" y="222.0" text-anchor="end" dominant-baseline="central">0</text>
+  <line class="impact-chart__grid" x1="44" y1="169.5" x2="506" y2="169.5" />
+  <text class="impact-chart__tick" x="36" y="169.5" text-anchor="end" dominant-baseline="central">20</text>
+  <line class="impact-chart__grid" x1="44" y1="117.0" x2="506" y2="117.0" />
+  <text class="impact-chart__tick" x="36" y="117.0" text-anchor="end" dominant-baseline="central">40</text>
+  <line class="impact-chart__grid" x1="44" y1="64.5" x2="506" y2="64.5" />
+  <text class="impact-chart__tick" x="36" y="64.5" text-anchor="end" dominant-baseline="central">60</text>
+  <line class="impact-chart__grid" x1="44" y1="12.0" x2="506" y2="12.0" />
+  <text class="impact-chart__tick" x="36" y="12.0" text-anchor="end" dominant-baseline="central">80</text>
+  <line class="impact-chart__axis" x1="44" y1="222" x2="506" y2="222" />
+  <rect class="impact-chart__bar" x="49.0" y="214" width="56" height="8" rx="4" fill="var(--impact-sequential)"><title>2020: 3</title></rect>
+  <text class="impact-chart__value" x="77.0" y="208" text-anchor="middle">3</text>
+  <text class="impact-chart__label" x="77.0" y="240" text-anchor="middle">2020</text>
+  <rect class="impact-chart__bar" x="115.0" y="177" width="56" height="45" rx="4" fill="var(--impact-sequential)"><title>2021: 17</title></rect>
+  <text class="impact-chart__value" x="143.0" y="171" text-anchor="middle">17</text>
+  <text class="impact-chart__label" x="143.0" y="240" text-anchor="middle">2021</text>
+  <rect class="impact-chart__bar" x="181.0" y="185" width="56" height="37" rx="4" fill="var(--impact-sequential)"><title>2022: 14</title></rect>
+  <text class="impact-chart__value" x="209.0" y="179" text-anchor="middle">14</text>
+  <text class="impact-chart__label" x="209.0" y="240" text-anchor="middle">2022</text>
+  <rect class="impact-chart__bar" x="247.0" y="141" width="56" height="81" rx="4" fill="var(--impact-sequential)"><title>2023: 31</title></rect>
+  <text class="impact-chart__value" x="275.0" y="135" text-anchor="middle">31</text>
+  <text class="impact-chart__label" x="275.0" y="240" text-anchor="middle">2023</text>
+  <rect class="impact-chart__bar" x="313.0" y="135" width="56" height="87" rx="4" fill="var(--impact-sequential)"><title>2024: 33</title></rect>
+  <text class="impact-chart__value" x="341.0" y="129" text-anchor="middle">33</text>
+  <text class="impact-chart__label" x="341.0" y="240" text-anchor="middle">2024</text>
+  <rect class="impact-chart__bar" x="379.0" y="51" width="56" height="171" rx="4" fill="var(--impact-sequential)"><title>2025: 65</title></rect>
+  <text class="impact-chart__value" x="407.0" y="45" text-anchor="middle">65</text>
+  <text class="impact-chart__label" x="407.0" y="240" text-anchor="middle">2025</text>
+  <rect class="impact-chart__bar" x="445.0" y="101" width="56" height="121" rx="4" fill="var(--impact-sequential)"><title>2026: 46</title></rect>
+  <text class="impact-chart__value" x="473.0" y="95" text-anchor="middle">46</text>
+  <text class="impact-chart__label" x="473.0" y="240" text-anchor="middle">2026</text>
+</svg>
diff --git a/_includes/impact/plots/papers-relationships-by-year.svg b/_includes/impact/plots/papers-relationships-by-year.svg
new file mode 100644
index 0000000..63c8bd1
--- /dev/null
+++ b/_includes/impact/plots/papers-relationships-by-year.svg
@@ -0,0 +1,54 @@
+<svg class="impact-chart" viewBox="0 0 490 264" width="100%" height="auto" style="max-width:490px" role="img" preserveAspectRatio="xMinYMin meet" aria-label="How papers relate to SQLancer, by publication year" xmlns="http://www.w3.org/2000/svg">
+  <title>How papers relate to SQLancer, by publication year</title>
+  <desc>Papers per relationship per year: 3 series across 7 years.</desc>
+  <line class="impact-chart__grid" x1="44" y1="222.0" x2="478" y2="222.0" />
+  <text class="impact-chart__tick" x="36" y="222.0" text-anchor="end" dominant-baseline="central">0</text>
+  <line class="impact-chart__grid" x1="44" y1="169.5" x2="478" y2="169.5" />
+  <text class="impact-chart__tick" x="36" y="169.5" text-anchor="end" dominant-baseline="central">5</text>
+  <line class="impact-chart__grid" x1="44" y1="117.0" x2="478" y2="117.0" />
+  <text class="impact-chart__tick" x="36" y="117.0" text-anchor="end" dominant-baseline="central">10</text>
+  <line class="impact-chart__grid" x1="44" y1="64.5" x2="478" y2="64.5" />
+  <text class="impact-chart__tick" x="36" y="64.5" text-anchor="end" dominant-baseline="central">15</text>
+  <line class="impact-chart__grid" x1="44" y1="12.0" x2="478" y2="12.0" />
+  <text class="impact-chart__tick" x="36" y="12.0" text-anchor="end" dominant-baseline="central">20</text>
+  <line class="impact-chart__axis" x1="44" y1="222" x2="478" y2="222" />
+  <text class="impact-chart__label" x="75.0" y="240" text-anchor="middle">2020</text>
+  <rect class="impact-chart__bar" x="130.0" y="201" width="14" height="21" rx="4" fill="var(--impact-series-2)"><title>Compares against SQLancer 2021: 2</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="137.0" y="197" text-anchor="middle">2</text>
+  <text class="impact-chart__label" x="137.0" y="240" text-anchor="middle">2021</text>
+  <rect class="impact-chart__bar" x="176.0" y="170" width="14" height="52" rx="4" fill="var(--impact-series-1)"><title>Uses or extends SQLancer 2022: 5</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="183.0" y="166" text-anchor="middle">5</text>
+  <rect class="impact-chart__bar" x="192.0" y="180" width="14" height="42" rx="4" fill="var(--impact-series-2)"><title>Compares against SQLancer 2022: 4</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="199.0" y="176" text-anchor="middle">4</text>
+  <rect class="impact-chart__bar" x="208.0" y="190" width="14" height="32" rx="4" fill="var(--impact-series-3)"><title>Describes SQLancer as state of the art 2022: 3</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="215.0" y="186" text-anchor="middle">3</text>
+  <text class="impact-chart__label" x="199.0" y="240" text-anchor="middle">2022</text>
+  <rect class="impact-chart__bar" x="238.0" y="170" width="14" height="52" rx="4" fill="var(--impact-series-1)"><title>Uses or extends SQLancer 2023: 5</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="245.0" y="166" text-anchor="middle">5</text>
+  <rect class="impact-chart__bar" x="254.0" y="170" width="14" height="52" rx="4" fill="var(--impact-series-2)"><title>Compares against SQLancer 2023: 5</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="261.0" y="166" text-anchor="middle">5</text>
+  <rect class="impact-chart__bar" x="270.0" y="190" width="14" height="32" rx="4" fill="var(--impact-series-3)"><title>Describes SQLancer as state of the art 2023: 3</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="277.0" y="186" text-anchor="middle">3</text>
+  <text class="impact-chart__label" x="261.0" y="240" text-anchor="middle">2023</text>
+  <rect class="impact-chart__bar" x="300.0" y="170" width="14" height="52" rx="4" fill="var(--impact-series-1)"><title>Uses or extends SQLancer 2024: 5</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="307.0" y="166" text-anchor="middle">5</text>
+  <rect class="impact-chart__bar" x="316.0" y="106" width="14" height="116" rx="4" fill="var(--impact-series-2)"><title>Compares against SQLancer 2024: 11</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="323.0" y="102" text-anchor="middle">11</text>
+  <rect class="impact-chart__bar" x="332.0" y="170" width="14" height="52" rx="4" fill="var(--impact-series-3)"><title>Describes SQLancer as state of the art 2024: 5</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="339.0" y="166" text-anchor="middle">5</text>
+  <text class="impact-chart__label" x="323.0" y="240" text-anchor="middle">2024</text>
+  <rect class="impact-chart__bar" x="362.0" y="117" width="14" height="105" rx="4" fill="var(--impact-series-1)"><title>Uses or extends SQLancer 2025: 10</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="369.0" y="113" text-anchor="middle">10</text>
+  <rect class="impact-chart__bar" x="378.0" y="22" width="14" height="200" rx="4" fill="var(--impact-series-2)"><title>Compares against SQLancer 2025: 19</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="385.0" y="18" text-anchor="middle">19</text>
+  <rect class="impact-chart__bar" x="394.0" y="86" width="14" height="136" rx="4" fill="var(--impact-series-3)"><title>Describes SQLancer as state of the art 2025: 13</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="401.0" y="82" text-anchor="middle">13</text>
+  <text class="impact-chart__label" x="385.0" y="240" text-anchor="middle">2025</text>
+  <rect class="impact-chart__bar" x="424.0" y="170" width="14" height="52" rx="4" fill="var(--impact-series-1)"><title>Uses or extends SQLancer 2026: 5</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="431.0" y="166" text-anchor="middle">5</text>
+  <rect class="impact-chart__bar" x="440.0" y="75" width="14" height="147" rx="4" fill="var(--impact-series-2)"><title>Compares against SQLancer 2026: 14</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="447.0" y="71" text-anchor="middle">14</text>
+  <rect class="impact-chart__bar" x="456.0" y="117" width="14" height="105" rx="4" fill="var(--impact-series-3)"><title>Describes SQLancer as state of the art 2026: 10</title></rect>
+  <text class="impact-chart__value impact-chart__value--tiny" x="463.0" y="113" text-anchor="middle">10</text>
+  <text class="impact-chart__label" x="447.0" y="240" text-anchor="middle">2026</text>
+</svg>
diff --git a/_includes/impact/stat-tiles.html b/_includes/impact/stat-tiles.html
new file mode 100644
index 0000000..3c30bc8
--- /dev/null
+++ b/_includes/impact/stat-tiles.html
@@ -0,0 +1,30 @@
+{%- comment -%}
+Key-figure row at the top of the impact page. Values come from stats.json.
+{%- endcomment -%}
+{%- assign head = site.data.impact.stats.headline -%}
+<ul class="impact-stats">
+  <li class="impact-stat">
+    <span class="impact-stat__value">{{ head.bugs_total }}</span>
+    <span class="impact-stat__label">bugs attributed to SQLancer and accepted as genuine</span>
+  </li>
+  <li class="impact-stat">
+    <span class="impact-stat__value">{{ head.dbms_with_bugs }}</span>
+    <span class="impact-stat__label">database systems with SQLancer-attributed bugs</span>
+  </li>
+  <li class="impact-stat">
+    <span class="impact-stat__value">{{ head.dbms_projects_using_sqlancer }}</span>
+    <span class="impact-stat__label">DBMS projects with evidence of using SQLancer</span>
+  </li>
+  <li class="impact-stat">
+    <span class="impact-stat__value">{{ head.papers_citing_sqlancer }}</span>
+    <span class="impact-stat__label">papers citing a SQLancer publication</span>
+  </li>
+  <li class="impact-stat">
+    <span class="impact-stat__value">{{ head.papers_reusing_or_extending_sqlancer }}</span>
+    <span class="impact-stat__label">papers that reuse or extend SQLancer</span>
+  </li>
+  <li class="impact-stat">
+    <span class="impact-stat__value">{{ head.papers_comparing_against_sqlancer }}</span>
+    <span class="impact-stat__label">papers that compare against SQLancer</span>
+  </li>
+</ul>
diff --git a/_includes/impact/talk-detail.html b/_includes/impact/talk-detail.html
new file mode 100644
index 0000000..3bb5cba
--- /dev/null
+++ b/_includes/impact/talk-detail.html
@@ -0,0 +1,114 @@
+{%- comment -%}
+One talk: where SQLancer comes up in it, and how far each of those can be
+trusted.
+
+A talk is the one source in this dataset whose words are not directly readable.
+What a page can honestly show is therefore layered: the speaker's own slides can
+be quoted; a frame captured from the recording can be shown, since some talks
+name SQLancer only on a slide and no transcript reaches that; automatic captions
+can only be quoted as captions, with a link to the moment so a reader can listen
+for themselves; and where the only source is someone who watched it, the page
+says that instead of quoting anything.
+
+Parameter: id (a talk record id).
+{%- endcomment -%}
+{%- assign talk = site.data.impact.talks.talks | where: "id", include.id | first -%}
+{%- if talk %}
+
+<p class="impact-meta">
+  {%- if talk.speakers.size > 0 %}{{ talk.speakers | join: ", " }}{% endif -%}
+  {%- if talk.event %} &middot; {{ talk.event }}{% endif -%}
+  {%- if talk.year %} &middot; {{ talk.year }}{% endif -%}
+  {%- if talk.publisher %} &middot; {{ talk.publisher }}{% endif %}
+</p>
+
+<p><a href="{{ talk.url }}">Watch the talk</a></p>
+
+{%- if talk.relationship %}
+<h2>What it does with SQLancer</h2>
+<p>{{ talk.relationship.summary }}</p>
+<p>
+  {%- for role in talk.relationship.roles -%}
+  <span class="impact-tag">{{ role | replace: "_", " " }}</span>
+  {%- endfor -%}
+</p>
+{%- endif %}
+
+<h2>Where SQLancer comes up ({{ talk.mentions.size }})</h2>
+
+<ul>
+  {%- for mention in talk.mentions %}
+  <li>
+    {%- if mention.timestamp %}
+    <a href="{{ mention.url }}">{{ mention.timestamp }}</a>
+    {%- else %}
+    <a href="{{ mention.url }}">{{ mention.source }}</a>
+    {%- endif %}
+    {%- for id in mention.technique_ids -%}
+    {%- assign technique = site.data.impact.techniques.techniques | where: "id", id | first -%}
+    <span class="impact-tag">{{ technique.short_name | default: id | upcase }}</span>
+    {%- endfor %}
+    {%- if mention.excerpt %}
+    <br><q>{{ mention.excerpt }}</q>
+    {%- endif %}
+    {%- if mention.image %}
+    <figure class="impact-frame">
+      <a href="{{ mention.url }}"><img src="{{ mention.image }}"
+        alt="{{ mention.note | strip_html | escape }}" loading="lazy"></a>
+      <figcaption class="impact-meta">
+        The talk at {{ mention.timestamp }}. {{ mention.note }}
+      </figcaption>
+    </figure>
+    {%- endif %}
+    <div class="impact-meta">
+      {%- case mention.source -%}
+      {%- when "slides" -%}
+        From the talk's own slides.
+      {%- when "frame" -%}
+        A frame from the recording. Nothing is transcribed from it: what the
+        slide says is shown, not quoted.
+      {%- when "captions" -%}
+        From the talk's automatic captions &mdash; a machine transcription of
+        speech, not the speaker's words as written
+        {%- if mention.heard_as %}: it renders the name as
+        &ldquo;{{ mention.heard_as }}&rdquo;{% endif -%}. Follow the link to
+        hear what was said.
+      {%- when "watched" -%}
+        {{ mention.note }}
+      {%- endcase -%}
+    </div>
+    {%- comment -%}
+      Why a moment has no picture. Without this the absence reads as an
+      oversight, when it is the opposite: somebody went and looked.
+    {%- endcomment -%}
+    {%- if mention.frame_checked %}
+    <div class="impact-meta">
+      Nothing to show from this moment: {{ mention.frame_checked | downcase }}
+    </div>
+    {%- endif %}
+  </li>
+  {%- endfor %}
+</ul>
+
+<h2>What was read</h2>
+
+<ul>
+  {%- for source in talk.sources %}
+  <li>
+    <strong>{{ source.kind }}</strong> &mdash; {{ source.status | replace: "_", " " }}
+    {%- if source.segments %}, {{ source.segments }} caption segments{% endif -%}
+    {%- if source.url and source.kind == "slides" %} (<a href="{{ source.url }}">deck</a>){% endif -%}
+    {%- if source.note %}<div class="impact-meta">{{ source.note }}</div>{% endif -%}
+  </li>
+  {%- endfor %}
+</ul>
+
+<p class="impact-meta">
+  This record is
+  <a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/talks.json"><code>talks.json</code></a>,
+  under the id <code>{{ talk.id }}</code>.
+</p>
+
+{%- else %}
+<p>No talk with this id.</p>
+{%- endif %}
diff --git a/_includes/impact/talk-frames.html b/_includes/impact/talk-frames.html
new file mode 100644
index 0000000..1cfb6fe
--- /dev/null
+++ b/_includes/impact/talk-frames.html
@@ -0,0 +1,33 @@
+{%- comment -%}
+The slides a talk put SQLancer on, shown with the talk itself.
+
+Some talks name SQLancer only on a slide, where no transcript reaches it. The
+frame is then the whole of the evidence, so it belongs beside the talk's title
+and summary rather than in a gallery of its own -- a reader should see what was
+on the screen while reading what the talk did with it.
+
+Parameter: talk (a talk record).
+{%- endcomment -%}
+{%- assign slug = include.talk.video_id | default: include.talk.id | split: ":" | last -%}
+{%- comment -%}
+  Any mention may carry a picture: a frame captured from a recording, or the
+  slide itself where the speaker published their deck.
+{%- endcomment -%}
+{%- for frame in include.talk.mentions -%}
+{%- if frame.image %}
+<figure class="impact-slide">
+  <a href="{{ '/impact/talks/' | relative_url }}{{ slug }}/">
+    <img src="{{ frame.image | relative_url }}"
+         alt="{{ frame.note | strip_html | escape }}" loading="lazy">
+  </a>
+  <figcaption class="impact-meta">
+    {{ frame.note }}
+    {%- if frame.timestamp %}
+    <a href="{{ frame.url }}">Watch at {{ frame.timestamp }}</a>
+    {%- else %}
+    <a href="{{ frame.url }}">See the slide</a>
+    {%- endif %}
+  </figcaption>
+</figure>
+{%- endif -%}
+{%- endfor -%}
diff --git a/_includes/impact/year-detail.html b/_includes/impact/year-detail.html
new file mode 100644
index 0000000..07e2e9b
--- /dev/null
+++ b/_includes/impact/year-detail.html
@@ -0,0 +1,130 @@
+{%- comment -%}
+Everything the dataset records about the bugs reported in one year.
+
+The chart of bugs over time shows eight bars and no way to ask what is in one.
+This answers that: which systems the year's bugs were found in, how they broke
+down, and every report with a link to it.
+
+Counts come from stats.json, so they cannot drift from the chart. The bug list
+is read from the records, because it is the evidence rather than a summary of
+it.
+
+Parameter: year (an integer).
+{%- endcomment -%}
+{%- assign impact = site.data.impact -%}
+{%- assign row = impact.stats.bugs.years | where: "year", include.year | first -%}
+{%- assign bugs = impact.bugs.bugs | where: "reported_year", include.year -%}
+
+<div class="impact" markdown="0">
+
+<p class="impact-breadcrumb">
+  <a href="{{ '/impact/#bugs-over-time' | relative_url }}">&larr; Bugs over time</a>
+</p>
+
+<p>
+  SQLancer is credited with <strong>{{ row.bugs }}</strong> bugs reported in
+  {{ row.year }}, across {{ row.systems }} database
+  system{% if row.systems != 1 %}s{% endif %}
+  {%- if row.first_reported %}, between {{ row.first_reported }} and
+  {{ row.last_reported }}{% endif %}.
+  {%- if row.bugs_rejected > 0 %}
+  A further {{ row.bugs_rejected }}
+  {%- if row.bugs_rejected == 1 %} report was{% else %} reports were{% endif %}
+  filed that year and rejected by the developers as invalid or duplicate;
+  those are kept in the dataset and counted nowhere.
+  {%- endif %}
+</p>
+
+<h2 id="systems">Which systems ({{ row.systems }})</h2>
+
+{% include impact/bar-list.html series=row.by_dbms %}
+
+<p class="impact-meta">
+  Each system's own page carries its whole history rather than this year's
+  share: {% for entry in row.by_dbms limit: 6 -%}
+  <a href="{{ '/impact/dbms/' | append: entry.key | append: '/' | relative_url }}">{{ entry.label }}</a>{% unless forloop.last %}, {% endunless %}
+  {%- endfor %}{% if row.by_dbms.size > 6 %}, and {{ row.by_dbms.size | minus: 6 }} more{% endif %}.
+</p>
+
+<h2 id="breakdown">How they break down</h2>
+
+<div class="impact-columns">
+  <div>
+    <h3>Who found them</h3>
+    {% include impact/bar-list.html series=row.by_affiliation %}
+  </div>
+  <div>
+    <h3>By status</h3>
+    {% include impact/bar-list.html series=row.by_status %}
+  </div>
+  <div>
+    <h3>By symptom</h3>
+    {% include impact/bar-list.html series=row.by_symptom %}
+  </div>
+  <div>
+    <h3>By technique</h3>
+    {% include impact/bar-list.html series=row.by_technique %}
+  </div>
+  <div>
+    <h3>Why each bug counts</h3>
+    {% include impact/bar-list.html series=row.by_attribution_rule %}
+  </div>
+  {%- if row.top_reporters.size > 0 %}
+  <div>
+    <h3>Who reported them</h3>
+    {% include impact/bar-list.html series=row.top_reporters %}
+  </div>
+  {%- endif %}
+</div>
+
+<h2 id="bugs">Every bug reported in {{ row.year }} ({{ bugs.size }})</h2>
+
+<p>
+  One row per report, newest first, each linking to the report itself. Rejected
+  reports are marked and are not part of any count above.
+</p>
+
+<div class="impact-table-wrap">
+  <table class="impact-table impact-table--bugs">
+    <thead>
+      <tr>
+        <th scope="col">Reported</th>
+        <th scope="col">Report</th>
+        <th scope="col">System</th>
+        <th scope="col">Status</th>
+        <th scope="col">Reporter</th>
+      </tr>
+    </thead>
+    <tbody>
+      {%- assign ordered = bugs | sort: "reported_date" | reverse %}
+      {%- for bug in ordered %}
+      {%- assign entry = impact.dbms.dbms | where: "id", bug.dbms | first %}
+      <tr{% unless bug.status_is_true_positive %} class="impact-row--rejected"{% endunless %}>
+        <td>{{ bug.reported_date | default: "&mdash;" }}</td>
+        <th scope="row">
+          {%- assign bug_link = bug.primary_url | default: bug.links.record -%}
+          {%- if bug_link -%}
+          <a href="{{ bug_link }}">{{ bug.title | default: bug.id | truncate: 100 }}</a>
+          {%- else -%}
+          {{ bug.title | default: bug.id | truncate: 100 }}
+          {%- endif -%}
+        </th>
+        <td>
+          <a href="{{ '/impact/dbms/' | append: bug.dbms | append: '/' | relative_url }}">{{ entry.name | default: bug.dbms }}</a>
+        </td>
+        <td>{{ bug.status | replace: "_", " " }}</td>
+        <td>{{ bug.reporter | default: "&mdash;" }}</td>
+      </tr>
+      {%- endfor %}
+    </tbody>
+  </table>
+</div>
+
+<p class="impact-meta">
+  Every figure on this page is computed from
+  <a href="https://github.com/sqlancer/sqlancer.github.io/tree/main/_data/impact"><code>_data/impact/</code></a>.
+  The <a href="{{ '/impact/#methodology' | relative_url }}">attribution policy</a>
+  explains what has to be true for a bug to be counted.
+</p>
+
+</div>
diff --git a/_pages/found-bugs.md b/_pages/found-bugs.md
index 4377f5b..e7e943c 100644
--- a/_pages/found-bugs.md
+++ b/_pages/found-bugs.md
@@ -1,6 +1,22 @@
 ---
 permalink: /bugs/
-title: "Bugs found in SQLancer"
+title: "Bugs found by SQLancer"
 ---
 
-SQLancer has found hundreds of bugs in mature, widely-used database systems. An incomplete list is available as a [JSON file in a SQLancer repository](https://github.com/sqlancer/bugs). Additional lists of bug reports can be found on the websites of [Jinsheng Ba](http://jinshengba.me/bombs/), [Manuel Rigger](https://manuelrigger.at/dbms-bugs/), and the [NUS TEST lab](https://nus-test.github.io/bugs/).
+{%- assign head = site.data.impact.stats.headline -%}
+
+SQLancer has found {{ head.bugs_total }} bugs that we can attribute to it with
+evidence, across {{ head.dbms_with_bugs }} database systems. Each one is recorded
+with the primary source linking it to SQLancer or to one of its test oracles.
+
+[**Browse the bugs on the impact page**]({{ '/impact/#bugs' | relative_url }}){: .btn .btn--primary}
+
+The underlying records live in
+[`_data/impact/bugs.json`](https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/bugs.json).
+Reduced test cases for the historic reports are kept in the
+[SQLancer bug repository](https://github.com/sqlancer/bugs).
+
+Further lists of database system bugs, not all of them found with SQLancer, are
+published by [Jinsheng Ba](http://jinshengba.me/bombs/),
+[Manuel Rigger](https://manuelrigger.at/dbms-bugs/), and the
+[NUS TEST lab](https://nus-test.github.io/bugs/).
diff --git a/_pages/impact.html b/_pages/impact.html
new file mode 100644
index 0000000..a66255d
--- /dev/null
+++ b/_pages/impact.html
@@ -0,0 +1,636 @@
+---
+permalink: /impact/
+title: "SQLancer's impact"
+excerpt: "Bugs found, database systems reached, research building on SQLancer -- every number traced back to a primary source."
+author_profile: false
+toc: true
+toc_label: "On this page"
+toc_sticky: true
+---
+
+{%- assign impact = site.data.impact -%}
+{%- assign stats = impact.stats -%}
+{%- assign head = stats.headline -%}
+{%- assign policy = impact.policy -%}
+
+<div class="impact" markdown="0">
+
+<p>
+  SQLancer finds bugs in database systems, and the techniques it introduced have
+  been picked up well beyond the project itself. This page collects what that
+  adds up to, and — more importantly — where each number comes from.
+</p>
+
+<p>
+  Every figure below is computed from structured records kept in this website's
+  repository under
+  <a href="https://github.com/sqlancer/sqlancer.github.io/tree/main/_data/impact"><code>_data/impact/</code></a>.
+  Each record carries the primary source that justifies it, so any statistic can
+  be followed back to a bug report, a paper, a repository, or a published
+  resource. Nothing on this page is typed in by hand.
+</p>
+
+{% include impact/stat-tiles.html %}
+
+<p class="impact-meta">
+  Policy version <code>{{ policy.policy_version }}</code>, in force since
+  {{ policy.effective_date }}.
+  {%- if stats.bugs.total_including_rejected > stats.bugs.total %}
+  {{ stats.bugs.total_including_rejected }} bug reports are on record;
+  {{ stats.bugs.total }} of them were accepted by the developers as genuine bugs
+  and are what the counts above use.
+  {%- endif %}
+</p>
+
+<h2 id="bugs">Bugs found by SQLancer ({{ stats.bugs.total }})</h2>
+
+<p>
+  A bug counts here when the evidence connects it to SQLancer or to a testing
+  technique SQLancer introduced. Bug reports very often name the test oracle
+  rather than the tool, so a report crediting NoREC, TLP, PQS or QPG counts even
+  when it never says “SQLancer”. Reports the developers rejected as invalid or
+  duplicate stay in the dataset for transparency but are excluded from the
+  totals. The full policy is <a href="#methodology">below</a>.
+</p>
+
+{%- comment -%}
+  The undercount, stated where the number is rather than in the methodology
+  nobody scrolls to. The figure is derived, not written: it is the projects
+  whose own evidence shows them running SQLancer and whose bugs never reach
+  this dataset, which is the part of the gap the data can measure about itself.
+{%- endcomment -%}
+<p>
+  This is a floor, not a total, and probably a distant one. A bug arrives here
+  only when something public ties it to SQLancer, and most of the testing
+  leaves no such trace: reports against closed-source systems are usually
+  private, a report is written about the bug rather than about what found it,
+  and a project can run SQLancer for years while quietly fixing what it turns
+  up. The dataset can measure part of its own blind spot &mdash;
+  <strong>{{ head.dbms_using_sqlancer_without_counted_bugs }}</strong> of the
+  {{ head.dbms_projects_using_sqlancer }} projects whose own evidence shows
+  them running SQLancer have no bug counted here at all:
+  {%- assign silent = stats.dbms.using_without_bugs -%}
+  {%- for row in silent %}
+  <a href="{{ '/impact/dbms/' | relative_url }}{{ row.id }}/">{{ row.name }}</a>
+  {%- unless forloop.last %}{% if forloop.rindex == 2 %} and {% else %}, {% endif %}{% endunless -%}
+  {%- endfor %}.
+</p>
+
+<h3 id="bugs-by-dbms">Bugs by database system ({{ stats.dbms.with_bugs }})</h3>
+
+<div class="impact-figure">
+  <p class="impact-figure__caption">
+    Bugs attributed to SQLancer in each database system, each bar split by who
+    found them. Every bar links to that system's own page, which lists the bugs
+    behind the number.
+  </p>
+  {%- comment -%}
+    The segments are drawn from by_dbms_and_affiliation, whose segments come in
+    the order of by_reporter_affiliation, so walking that series in the same
+    order gives each swatch the colour its segment has.
+  {%- endcomment -%}
+  <ul class="impact-legend">
+    {%- assign affiliation_slot = 0 -%}
+    {%- for row in stats.bugs.by_reporter_affiliation -%}
+    {%- if row.count > 0 -%}
+    {%- assign affiliation_slot = affiliation_slot | plus: 1 -%}
+    <li class="impact-legend__item">
+      <span class="impact-legend__swatch" style="background: var(--impact-series-{{ affiliation_slot }})"></span>
+      {{ row.label }} ({{ row.count }})
+    </li>
+    {%- endif -%}
+    {%- endfor -%}
+  </ul>
+  <div class="impact-chart__scroll">
+    {% include impact/plots/bugs-by-dbms.svg %}
+  </div>
+</div>
+
+<details class="impact-details">
+  <summary>Show these figures as a table</summary>
+  {% include impact/counts-table.html series=stats.bugs.by_dbms heading="Database system" label="Bugs" link_base="/impact/dbms/" %}
+</details>
+
+<h3 id="bugs-over-time">Bugs over time</h3>
+
+<div class="impact-figure">
+  <p class="impact-figure__caption">
+    Bugs by the year they were reported. Each column links to that year's own
+    page, which lists every report behind the number.
+  </p>
+  <div class="impact-chart__scroll">
+    {% include impact/plots/bugs-by-year.svg %}
+  </div>
+</div>
+
+<details class="impact-details">
+  <summary>Show these figures as a table</summary>
+  {% include impact/counts-table.html series=stats.bugs.by_year heading="Year reported" label="Bugs" %}
+</details>
+
+{%- comment -%}
+  The narrower breakdowns live on their own page, so this one keeps to the
+  figures a reader comes for. See _pages/impact/bug-statistics.html.
+{%- endcomment -%}
+
+<h3 id="bugs-by-status">What happened to them</h3>
+
+<div class="impact-figure">
+  <p class="impact-figure__caption">
+    Resolution of the {{ stats.bugs.total }} accepted bug reports.
+  </p>
+  <ul class="impact-legend">
+    {%- assign status_slot = 0 -%}
+    {%- for row in stats.bugs.by_status -%}
+    {%- if row.count > 0 and status_slot < 4 -%}
+    {%- assign status_slot = status_slot | plus: 1 -%}
+    <li class="impact-legend__item">
+      <span class="impact-legend__swatch" style="background: var(--impact-series-{{ status_slot }})"></span>
+      {{ row.label }} ({{ row.count }})
+    </li>
+    {%- endif -%}
+    {%- endfor -%}
+  </ul>
+  <div class="impact-chart__scroll">
+    {% include impact/plots/bugs-by-status.svg %}
+  </div>
+</div>
+
+{%- comment -%}
+  How much of the total rests on each clause of the policy -- above all on the
+  weakest, which admits a report because of who filed it rather than because of
+  anything the report says. A reader who wants to discount that share should be
+  able to see its size without going to the data.
+{%- endcomment -%}
+{%- assign rule_rows = stats.bugs.by_attribution_rule -%}
+{%- if rule_rows.size > 0 %}
+{% include impact/counts-table.html series=rule_rows heading="How each bug was attributed" label="Bugs" %}
+{%- endif %}
+
+<p class="impact-meta">
+  Reproducers for the historic reports are kept in the
+  <a href="https://github.com/sqlancer/bugs">SQLancer bug repository</a>; this
+  page links each record to its upstream report rather than duplicating the test
+  cases.
+</p>
+
+<p>
+  <a href="{{ '/impact/bug-statistics/' | relative_url }}">More bug statistics</a>
+  &mdash; which technique found them, which tool, and how each bug manifested.
+</p>
+
+<h2 id="database-systems">Database systems</h2>
+
+<p>
+  Three things are tracked apart from each other and never conflated: the
+  {{ stats.dbms.supported }} systems SQLancer ships a testing implementation
+  for, the {{ stats.dbms.with_bugs }} it has attributed bugs in, and the
+  {{ stats.dbms.with_adoption }} projects whose own developers use it. Only the
+  last is below; the other two, the system-by-system table and the projects
+  that have proposed adopting SQLancer, are on
+  <a href="{{ '/impact/database-systems/' | relative_url }}">the database
+  systems page</a>.
+</p>
+
+{%- assign all_adoption = impact.adoption.adoption -%}
+{%- assign adoption_records = all_adoption | where_exp: "record", "record.relationship != 'planned_adoption'" -%}
+{%- comment -%}
+  Projects that have proposed adopting SQLancer and nothing more. A project
+  with an open proposal alongside evidence that it already runs SQLancer is
+  counted as using it, which is what stats.json has always done.
+{%- endcomment -%}
+{%- assign using_dbms = adoption_records | map: "dbms" | uniq -%}
+{%- assign planned_records = "" | split: "" -%}
+{%- for record in all_adoption -%}
+  {%- if record.relationship == "planned_adoption" -%}
+    {%- unless using_dbms contains record.dbms -%}
+      {%- assign planned_records = planned_records | push: record -%}
+    {%- endunless -%}
+  {%- endif -%}
+{%- endfor -%}
+{%- assign adoption_dbms_count = adoption_records | map: "dbms" | uniq | size -%}
+{%- if adoption_records.size > 0 %}
+<h3 id="adoption-evidence">Adoption evidence ({{ adoption_dbms_count }})</h3>
+
+<p>
+  One entry per project. A project can show up in more than one way &mdash; a
+  fork of its own, a job in its CI, a script in its repository &mdash; and each
+  project's page carries every piece of evidence in full: the quoted source,
+  what it shows, and when it was last checked.
+</p>
+
+<div class="impact-table-wrap">
+  <table class="impact-table">
+    <thead>
+      <tr>
+        <th scope="col">Project</th>
+        <th scope="col">How it uses SQLancer</th>
+        <th scope="col" class="impact-num">Evidence</th>
+      </tr>
+    </thead>
+    <tbody>
+      {%- assign adoption_dbms = adoption_records | map: "dbms" | uniq -%}
+      {%- for dbms_id in adoption_dbms %}
+      {%- assign dbms_entry = impact.dbms.dbms | where: "id", dbms_id | first -%}
+      {%- assign project_records = adoption_records | where: "dbms", dbms_id -%}
+      {%- assign evidence_count = 0 -%}
+      {%- for record in project_records -%}
+        {%- assign evidence_count = evidence_count | plus: record.evidence.size -%}
+      {%- endfor %}
+      <tr>
+        <th scope="row">
+          <a href="{{ '/impact/dbms/' | append: dbms_id | append: '/#adoption' | relative_url }}">{{ dbms_entry.name | default: dbms_id }}</a>
+        </th>
+        <td>
+          {%- for record in project_records -%}
+            <span class="impact-tag">{{ record.relationship | replace: "_", " " }}</span>
+          {%- endfor -%}
+        </td>
+        <td class="impact-num">{{ evidence_count }}</td>
+      </tr>
+      {%- endfor %}
+    </tbody>
+  </table>
+</div>
+{%- endif %}
+
+{%- if planned_records.size > 0 %}
+<p>
+  {{ planned_records.size }} further project{% if planned_records.size != 1 %}s have{% else %} has{% endif %}
+  <em>proposed</em> adopting SQLancer without anything yet showing them running
+  it. An intention is not use, so those count towards no figure here; they are
+  listed under
+  <a href="{{ '/impact/database-systems/#planned-adoption' | relative_url }}">planned adoption</a>.
+</p>
+{%- endif %}
+
+<h2 id="papers">Research building on SQLancer ({{ stats.papers.external_total }})</h2>
+
+<p>
+  Papers are found by walking the citation graphs of the foundational SQLancer
+  publications, so the list of citing papers is a fact about the graph rather
+  than a judgement. Each paper is then placed in five relationships: that it
+  cites SQLancer, and then whether it reuses the codebase, extends a technique,
+  compares against SQLancer, or calls it the state of the art. The four
+  judgements overlap on purpose &mdash; a paper can do several at once &mdash;
+  so the totals de-duplicate and such a paper is counted once. Reuse, extension
+  and comparison are what &ldquo;builds on&rdquo; means; calling SQLancer the
+  state of the art is recognition, and is counted on its own.
+</p>
+
+<p>
+  {{ head.papers_citing_sqlancer }} papers cite a SQLancer publication,
+  {{ head.papers_building_on_sqlancer }} of them do more than cite it, and
+  {{ head.papers_calling_sqlancer_state_of_the_art }} describe SQLancer or one
+  of its techniques as the state of the art in their own words.
+</p>
+
+{% include impact/counts-table.html series=stats.papers.by_relationship heading="Relationship" label="Papers" %}
+
+<h3 id="papers-over-time">Papers over time</h3>
+
+<div class="impact-figure">
+  <p class="impact-figure__caption">
+    All papers citing a SQLancer publication, by publication year.
+  </p>
+  <div class="impact-chart__scroll">
+    {% include impact/plots/papers-by-year.svg %}
+  </div>
+</div>
+
+<div class="impact-figure">
+  <p class="impact-figure__caption">
+    Papers that go further than citing, by publication year. The categories
+    overlap, so the bars are not additive. The last series is recognition
+    rather than reuse and is not part of the “builds on” total.
+  </p>
+  {%- comment -%}
+    The swatch order follows the series order in plots.py: reuse or extension,
+    comparison, recognition.
+  {%- endcomment -%}
+  <ul class="impact-legend">
+    <li class="impact-legend__item">
+      <span class="impact-legend__swatch" style="background: var(--impact-series-1)"></span>
+      Uses or extends SQLancer
+    </li>
+    <li class="impact-legend__item">
+      <span class="impact-legend__swatch" style="background: var(--impact-series-2)"></span>
+      Compares against SQLancer
+    </li>
+    <li class="impact-legend__item">
+      <span class="impact-legend__swatch" style="background: var(--impact-series-3)"></span>
+      Describes SQLancer as state of the art
+    </li>
+  </ul>
+  <div class="impact-chart__scroll">
+    {% include impact/plots/papers-relationships-by-year.svg %}
+  </div>
+</div>
+
+<details class="impact-details">
+  <summary>Show these figures as a table</summary>
+  <div class="impact-table-wrap">
+    <table class="impact-table">
+      <thead>
+        <tr>
+          <th scope="col">Year</th>
+          <th scope="col" class="impact-num">Citing</th>
+          <th scope="col" class="impact-num">Uses or extends</th>
+          <th scope="col" class="impact-num">Compares with</th>
+          <th scope="col" class="impact-num">Builds on (distinct)</th>
+          <th scope="col" class="impact-num">Calls state of the art</th>
+        </tr>
+      </thead>
+      <tbody>
+        {%- for entry in stats.papers.by_relationship_year.references %}
+        {%- assign index = forloop.index0 %}
+        <tr>
+          <th scope="row">{{ entry.label }}</th>
+          <td class="impact-num">{{ entry.count }}</td>
+          <td class="impact-num">{{ stats.papers.by_relationship_year.reusing_or_extending[index].count }}</td>
+          <td class="impact-num">{{ stats.papers.by_relationship_year.compares_with[index].count }}</td>
+          <td class="impact-num">{{ stats.papers.by_relationship_year.building_on[index].count }}</td>
+          <td class="impact-num">{{ stats.papers.by_relationship_year.describes_as_state_of_the_art[index].count }}</td>
+        </tr>
+        {%- endfor %}
+      </tbody>
+    </table>
+  </div>
+</details>
+
+{%- comment -%}
+  One list per relationship rather than one combined list: the categories mean
+  different things, and a paper that reuses the codebase is a different kind of
+  evidence from one that merely uses it as a baseline. The categories overlap,
+  so a paper can appear in more than one list; the deduplicated total is stated
+  above.
+{%- endcomment -%}
+{%- assign external_papers = impact.papers.papers | where_exp: "p", "p.is_sqlancer_publication != true" -%}
+
+{%- assign infra_papers = "" | split: "" -%}
+{%- assign extends_papers = "" | split: "" -%}
+{%- assign compares_papers = "" | split: "" -%}
+{%- for paper in external_papers -%}
+  {%- if paper.relationships.uses_infrastructure.value == "yes" -%}
+    {%- assign infra_papers = infra_papers | push: paper -%}
+  {%- endif -%}
+  {%- if paper.relationships.extends_technique.value == "yes" -%}
+    {%- assign extends_papers = extends_papers | push: paper -%}
+  {%- endif -%}
+  {%- if paper.relationships.compares_with.value == "yes" -%}
+    {%- assign compares_papers = compares_papers | push: paper -%}
+  {%- endif -%}
+{%- endfor -%}
+
+{%- if infra_papers.size > 0 or extends_papers.size > 0 or compares_papers.size > 0 %}
+<h3 id="papers-list">Papers that build on SQLancer ({{ head.papers_building_on_sqlancer }})</h3>
+
+<p>
+  Grouped by what each paper actually does with SQLancer. Every entry shows the
+  exact passage the classification rests on, copied verbatim from the source. A
+  paper that holds more than one relationship appears in more than one list, so
+  these lists are not additive &mdash; the
+  {{ head.papers_building_on_sqlancer }} in the summary above counts each paper
+  once.
+</p>
+
+{%- comment -%}
+  Reusing the codebase and developing the technique are recorded separately --
+  a paper can do either without the other, and only two do both -- but they
+  answer one question for a reader: what did this paper build on SQLancer? So
+  they are shown together, with each entry quoting the relationship it holds
+  and naming the other in its "Also" line where it holds both.
+{%- endcomment -%}
+{%- assign builds_papers = infra_papers -%}
+{%- for paper in extends_papers -%}
+  {%- unless paper.relationships.uses_infrastructure.value == "yes" -%}
+    {%- assign builds_papers = builds_papers | push: paper -%}
+  {%- endunless -%}
+{%- endfor -%}
+
+{%- if builds_papers.size > 0 %}
+<h4 id="papers-uses-infrastructure">Uses or extends SQLancer ({{ builds_papers.size }})</h4>
+<p class="impact-meta">
+  The paper's implementation reuses or derives from the SQLancer codebase
+  ({{ infra_papers.size }} papers), or the authors extend, generalise or adapt
+  a technique introduced through SQLancer &mdash; taking TLP to a new data
+  model, say, or building a new method on the PQS idea
+  ({{ extends_papers.size }} papers). The two are recorded separately in the
+  data, and each paper's own page says which of them applies.
+</p>
+{% include impact/paper-list.html papers=builds_papers relationship="uses_infrastructure|extends_technique" %}
+{%- endif %}
+
+{%- if compares_papers.size > 0 %}
+<h4 id="papers-compares-with">Compares against SQLancer ({{ compares_papers.size }})</h4>
+<p class="impact-meta">
+  The paper evaluates its own approach against SQLancer or one of its test
+  oracles as a baseline.
+</p>
+{% include impact/paper-list.html papers=compares_papers relationship="compares_with" %}
+{%- endif %}
+{%- endif %}
+
+{%- assign sota_papers = "" | split: "" -%}
+{%- for paper in external_papers -%}
+  {%- if paper.relationships.describes_as_state_of_the_art.value == "yes" -%}
+    {%- assign sota_papers = sota_papers | push: paper -%}
+  {%- endif -%}
+{%- endfor -%}
+
+{%- comment -%}
+  Recognition rather than reuse. Most of these papers say it while naming
+  their comparison set, which the section above already shows, so the page
+  quotes the ones whose claim stands on its own -- recorded, with a reason
+  for each, in _data/impact/recognition_highlights.json -- and the full list
+  lives on its own page.
+{%- endcomment -%}
+{%- assign highlights = impact.recognition_highlights.highlights -%}
+{%- if sota_papers.size > 0 %}
+<h3 id="papers-state-of-the-art">Papers calling SQLancer state of the art ({{ sota_papers.size }})</h3>
+
+<p>
+  These papers describe SQLancer or one of its techniques as the state of the
+  art. That is a different claim from building on it, so it is counted
+  separately and is not part of the
+  {{ head.papers_reusing_or_extending_sqlancer }} above. Quoted here are the
+  {{ highlights.size }} that make the claim on their own account rather than
+  while listing the tools they measure against; the
+  <a href="{{ '/impact/recognition/' | relative_url }}">other
+  {{ sota_papers.size | minus: highlights.size }}</a> say it in the same
+  sentence that names their comparison set.
+</p>
+
+<ul>
+  {%- for highlight in highlights -%}
+  {%- assign paper = external_papers | where: "id", highlight.paper_id | first -%}
+  {%- assign record_key = highlight.paper_id | replace: ":", "_" | replace: ".", "_" | replace: "/", "_" -%}
+  {%- assign record = site.data.papers[record_key] -%}
+  {%- assign quoted = record.analysis.relationships.describes_as_state_of_the_art.quotes | where: "mention_id", highlight.mention_id | first -%}
+  {%- if quoted %}
+  <li>
+    <div class="impact-evidence"><q>{{ quoted.sentence | strip_html }}</q></div>
+    <span class="impact-meta">
+      <a href="{{ '/impact/papers/' | append: record_key | append: '/' | relative_url }}">{{ paper.title }}</a>
+      {%- if paper.year %} ({{ paper.year }}){% endif %}
+      {%- if paper.venue %} &middot; {{ paper.venue }}{% endif %}
+      {%- if quoted.section %} &middot; {{ quoted.section }}{% endif %}
+    </span>
+  </li>
+  {%- endif -%}
+  {%- endfor %}
+</ul>
+
+<p>
+  <a href="{{ '/impact/recognition/' | relative_url }}">All {{ sota_papers.size }} papers calling SQLancer state of the art</a>
+</p>
+{%- endif %}
+
+<details class="impact-details">
+  <summary>All {{ stats.papers.external_total }} papers citing SQLancer</summary>
+  <div class="impact-table-wrap">
+    <table class="impact-table">
+      <thead>
+        <tr>
+          <th scope="col">Paper</th>
+          <th scope="col" class="impact-num">Year</th>
+          <th scope="col">Relationships</th>
+        </tr>
+      </thead>
+      <tbody>
+        {%- for paper in impact.papers.papers %}
+        {%- unless paper.is_sqlancer_publication %}
+        <tr>
+          <th scope="row">
+            {%- if paper.url %}<a href="{{ paper.url }}">{{ paper.title }}</a>{% else %}{{ paper.title }}{% endif -%}
+          </th>
+          <td class="impact-num">{{ paper.year }}</td>
+          <td>
+            <span class="impact-tag">cites</span>
+            {%- if paper.relationships.uses_infrastructure.value == "yes" or paper.relationships.extends_technique.value == "yes" %}<span class="impact-tag">uses or extends</span>{% endif -%}
+            {%- if paper.relationships.compares_with.value == "yes" %}<span class="impact-tag">compares</span>{% endif -%}
+            {%- if paper.relationships.describes_as_state_of_the_art.value == "yes" %}<span class="impact-tag">state of the art</span>{% endif -%}
+          </td>
+        </tr>
+        {%- endunless %}
+        {%- endfor %}
+      </tbody>
+    </table>
+  </div>
+</details>
+
+{%- assign resources = impact.resources.resources -%}
+{%- if resources.size > 0 %}
+<h2 id="resources">Resources ({{ resources.size }})</h2>
+
+<p>
+  Material other people have made about SQLancer or one of its techniques.
+  Pages that merely mention it in passing are left out, and so is the project's
+  own material &mdash; this site already links its documentation, papers and
+  blog, and repeating them here would say nothing about reach.
+</p>
+
+{%- assign resource_types = resources | map: "type" | uniq | sort -%}
+{%- for type in resource_types %}
+{%- assign of_type = resources | where: "type", type -%}
+{%- comment -%}
+  Every kind in the vocabulary pluralises by adding an s -- talks, tools, blog
+  posts, documentation notes -- so the heading counts things rather than naming
+  a category in the singular above a list of several.
+
+  "Talks" alone would read as the project's own recorded talks, which are on
+  the site already and deliberately excluded here. Every one of these was given
+  by somebody else, and the heading has to say so where it is read on its own,
+  in the table of contents.
+{%- endcomment -%}
+{%- assign heading = type | replace: "_", " " | append: "s" | capitalize -%}
+{%- if type == "talk" %}{% assign heading = "External talks" %}{% endif -%}
+<h3 id="resources-{{ type }}">{{ heading }} ({{ of_type.size }})</h3>
+<ul>
+  {%- for resource in resources -%}
+  {%- if resource.type == type %}
+  <li>
+    {%- comment -%}
+      Who published it comes first. A post's title says what it is about, not
+      whose it is, and whose it is carries most of the weight here: "Fuzz
+      testing QuestDB" is a fact about a tool until QuestDB's own name is in
+      front of it.
+    {%- endcomment -%}
+    <a href="{{ resource.url }}">
+      {%- if resource.publisher and resource.type == "blog_post" -%}
+      {{ resource.publisher }}: {% endif -%}
+      {{ resource.title }}</a>
+    {%- if resource.year %} <span class="impact-meta">({{ resource.year }})</span>{% endif %}
+    {%- if resource.official %} <span class="impact-tag">official</span>{% endif %}
+    {%- comment -%}
+      A talk has a record of its own saying where in it SQLancer comes up, which
+      is the part a reader actually wants and which no line of prose here can
+      carry.
+    {%- endcomment -%}
+    {%- assign talk = site.data.impact.talks.talks | where: "url", resource.url | first -%}
+    {%- if talk %}
+    {%- assign slug = talk.video_id | default: talk.id | split: ":" | last %}
+    &middot; <a href="/impact/talks/{{ slug }}/">{{ talk.mentions.size }} mention{% if talk.mentions.size != 1 %}s{% endif %}</a>
+    {%- endif %}
+    <br>{{ resource.description }}
+    {%- if talk %}{% include impact/talk-frames.html talk=talk %}{% endif %}
+  </li>
+  {%- endif -%}
+  {%- endfor %}
+</ul>
+{%- endfor %}
+{%- endif %}
+
+<h2 id="methodology">Methodology and attribution policy</h2>
+
+<p>
+  These are the rules the dataset is built to. They are stored as data
+  (<a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/policy.json"><code>policy.json</code></a>)
+  and rendered here, so the policy the collectors enforce and the policy
+  described on this page cannot drift apart.
+</p>
+
+{%- for section in policy.sections %}
+<h3 id="policy-{{ section.id }}">{{ section.title }}</h3>
+{%- for paragraph in section.body %}
+<p>{{ paragraph }}</p>
+{%- endfor %}
+{%- if section.rules %}
+<div>
+  {%- for rule in section.rules %}
+  <p class="impact-policy__rule">
+    <span class="impact-policy__kind">{{ rule.kind | default: "note" }}</span>
+    {{ rule.text }}
+  </p>
+  {%- endfor %}
+</div>
+{%- endif %}
+{%- endfor %}
+
+<h2 id="data">The data itself</h2>
+
+<p>
+  The records are plain JSON, each file validated against a JSON Schema on every
+  change:
+</p>
+
+<ul>
+  <li><a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/bugs.json"><code>bugs.json</code></a> — bug reports attributed to SQLancer, with the evidence for each attribution.</li>
+  <li><a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/papers.json"><code>papers.json</code></a> — papers citing SQLancer and their relationships to it.</li>
+  <li><a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/adoption.json"><code>adoption.json</code></a> — evidence that a database system's developers use SQLancer.</li>
+  <li><a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/resources.json"><code>resources.json</code></a> — talks, tutorials, documentation and other material.</li>
+  <li><a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/talks.json"><code>talks.json</code></a> — recorded talks, with every point where SQLancer comes up, what was read to find it, and how far those words can be trusted.</li>
+  <li><a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/papers"><code>_data/papers/</code></a> — one file per paper that cites SQLancer, holding every sentence in it that mentions the project and the classification each of those sentences supports.</li>
+  <li><a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/dbms.json"><code>dbms.json</code></a> — the database system registry.</li>
+  <li><a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/techniques.json"><code>techniques.json</code></a> — the technique and tool taxonomy.</li>
+  <li><a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/people.json"><code>people.json</code></a> — the people who run SQLancer campaigns, linked to their GitHub profiles. It decides whose reports are worth reading, and attributes nothing by itself.</li>
+  <li><a href="https://github.com/sqlancer/sqlancer.github.io/blob/main/_data/impact/stats.json"><code>stats.json</code></a> — the derived figures this page renders, regenerated from the records above.</li>
+</ul>
+
+<p class="impact-note">
+  Found something wrong, or a bug, paper or tool that is missing?
+  <a href="https://github.com/sqlancer/sqlancer.github.io/issues/new">Open an issue</a>
+  — the dataset is meant to be corrected in public.
+</p>
+
+</div>
diff --git a/_pages/impact/bug-statistics.html b/_pages/impact/bug-statistics.html
new file mode 100644
index 0000000..a367546
--- /dev/null
+++ b/_pages/impact/bug-statistics.html
@@ -0,0 +1,68 @@
+---
+permalink: /impact/bug-statistics/
+title: "More bug statistics"
+excerpt: "Further breakdowns of the bugs SQLancer found: which technique fired, which tool, and how each bug manifested."
+author_profile: false
+toc: true
+toc_label: "On this page"
+toc_sticky: true
+---
+
+{%- comment -%}
+Breakdowns that answer a narrower question than "how many bugs, and where".
+
+The impact page carries the figures a reader comes for -- the total, the
+systems, who found them, what happened to them. Everything that slices those
+same bugs a different way lives here, so the main page stays readable and this
+one can grow. Adding a breakdown means moving its section across unchanged;
+nothing here is shared with the page it came from.
+
+Every figure comes from _data/impact/stats.json, exactly as on the impact page.
+{%- endcomment -%}
+{%- assign impact = site.data.impact -%}
+{%- assign stats = impact.stats -%}
+
+<div class="impact" markdown="0">
+
+<p class="impact-breadcrumb">
+  <a href="{{ '/impact/#bugs' | relative_url }}">&larr; Bugs found by SQLancer</a>
+</p>
+
+<p>
+  Further breakdowns of the same {{ stats.bugs.total }} accepted bug reports the
+  <a href="{{ '/impact/#bugs' | relative_url }}">impact page</a> counts. Each
+  system's own page lists the individual reports behind its share.
+</p>
+
+<h2 id="bugs-by-technique">Which technique found them</h2>
+
+<p>
+  SQLancer is an umbrella: alongside the main tool, components such as
+  SQLancer++ and ShQveL count towards these figures. Each record keeps both the
+  tool that found the bug and, where the report says so, the specific technique —
+  they are never collapsed into one label. “Technique not recorded” means the
+  report identifies the campaign but not which oracle fired.
+</p>
+
+<div class="impact-figure">
+  <div class="impact-chart__scroll">
+    {% include impact/plots/bugs-by-technique.svg %}
+  </div>
+</div>
+
+<details class="impact-details">
+  <summary>Show these figures as a table</summary>
+  {% include impact/counts-table.html series=stats.bugs.by_technique heading="Technique" label="Bugs" %}
+  {% include impact/counts-table.html series=stats.bugs.by_finder heading="Tool" label="Bugs" %}
+  {% include impact/counts-table.html series=stats.bugs.by_symptom heading="How the bug manifested" label="Bugs" %}
+</details>
+
+<p class="impact-meta">
+  Every figure on this page is computed from
+  <a href="https://github.com/sqlancer/sqlancer.github.io/tree/main/_data/impact"><code>_data/impact/</code></a>.
+  The <a href="{{ '/impact/#methodology' | relative_url }}">attribution policy</a>
+  explains what has to be true for a bug to be counted, and how a technique is
+  credited.
+</p>
+
+</div>
diff --git a/_pages/impact/bugs/2019.html b/_pages/impact/bugs/2019.html
new file mode 100644
index 0000000..b27431f
--- /dev/null
+++ b/_pages/impact/bugs/2019.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/bugs/2019/
+title: "Bugs reported in 2019"
+excerpt: "221 bugs attributed to SQLancer and reported in 2019, across 5 database systems, each with its evidence."
+bug_year: 2019
+author_profile: false
+sitemap: true
+---
+
+{% include impact/year-detail.html year=page.bug_year %}
diff --git a/_pages/impact/bugs/2020.html b/_pages/impact/bugs/2020.html
new file mode 100644
index 0000000..2d7fe30
--- /dev/null
+++ b/_pages/impact/bugs/2020.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/bugs/2020/
+title: "Bugs reported in 2020"
+excerpt: "302 bugs attributed to SQLancer and reported in 2020, across 10 database systems, each with its evidence."
+bug_year: 2020
+author_profile: false
+sitemap: true
+---
+
+{% include impact/year-detail.html year=page.bug_year %}
diff --git a/_pages/impact/bugs/2021.html b/_pages/impact/bugs/2021.html
new file mode 100644
index 0000000..9cbc35a
--- /dev/null
+++ b/_pages/impact/bugs/2021.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/bugs/2021/
+title: "Bugs reported in 2021"
+excerpt: "22 bugs attributed to SQLancer and reported in 2021, across 6 database systems, each with its evidence."
+bug_year: 2021
+author_profile: false
+sitemap: true
+---
+
+{% include impact/year-detail.html year=page.bug_year %}
diff --git a/_pages/impact/bugs/2022.html b/_pages/impact/bugs/2022.html
new file mode 100644
index 0000000..d0ffda6
--- /dev/null
+++ b/_pages/impact/bugs/2022.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/bugs/2022/
+title: "Bugs reported in 2022"
+excerpt: "224 bugs attributed to SQLancer and reported in 2022, across 10 database systems, each with its evidence."
+bug_year: 2022
+author_profile: false
+sitemap: true
+---
+
+{% include impact/year-detail.html year=page.bug_year %}
diff --git a/_pages/impact/bugs/2023.html b/_pages/impact/bugs/2023.html
new file mode 100644
index 0000000..52a61ef
--- /dev/null
+++ b/_pages/impact/bugs/2023.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/bugs/2023/
+title: "Bugs reported in 2023"
+excerpt: "271 bugs attributed to SQLancer and reported in 2023, across 21 database systems, each with its evidence."
+bug_year: 2023
+author_profile: false
+sitemap: true
+---
+
+{% include impact/year-detail.html year=page.bug_year %}
diff --git a/_pages/impact/bugs/2024.html b/_pages/impact/bugs/2024.html
new file mode 100644
index 0000000..dd72488
--- /dev/null
+++ b/_pages/impact/bugs/2024.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/bugs/2024/
+title: "Bugs reported in 2024"
+excerpt: "193 bugs attributed to SQLancer and reported in 2024, across 18 database systems, each with its evidence."
+bug_year: 2024
+author_profile: false
+sitemap: true
+---
+
+{% include impact/year-detail.html year=page.bug_year %}
diff --git a/_pages/impact/bugs/2025.html b/_pages/impact/bugs/2025.html
new file mode 100644
index 0000000..18ab783
--- /dev/null
+++ b/_pages/impact/bugs/2025.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/bugs/2025/
+title: "Bugs reported in 2025"
+excerpt: "264 bugs attributed to SQLancer and reported in 2025, across 16 database systems, each with its evidence."
+bug_year: 2025
+author_profile: false
+sitemap: true
+---
+
+{% include impact/year-detail.html year=page.bug_year %}
diff --git a/_pages/impact/bugs/2026.html b/_pages/impact/bugs/2026.html
new file mode 100644
index 0000000..22f58b7
--- /dev/null
+++ b/_pages/impact/bugs/2026.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/bugs/2026/
+title: "Bugs reported in 2026"
+excerpt: "411 bugs attributed to SQLancer and reported in 2026, across 23 database systems, each with its evidence."
+bug_year: 2026
+author_profile: false
+sitemap: true
+---
+
+{% include impact/year-detail.html year=page.bug_year %}
diff --git a/_pages/impact/database-systems.html b/_pages/impact/database-systems.html
new file mode 100644
index 0000000..7cf3057
--- /dev/null
+++ b/_pages/impact/database-systems.html
@@ -0,0 +1,160 @@
+---
+permalink: /impact/database-systems/
+title: "Database systems and SQLancer"
+excerpt: "Which systems SQLancer can test, which it has found bugs in, which projects use it, and which have proposed to."
+author_profile: false
+toc: true
+toc_label: "On this page"
+toc_sticky: true
+---
+
+{%- comment -%}
+The system-by-system detail, kept off the impact page.
+
+The impact page answers "which projects use SQLancer", because that is the
+question a reader comes with. The rest -- what supported means as against bugs
+found, the row for every system, and the projects that have only proposed
+adopting it -- is reference material: worth having, and worth not scrolling
+through first. Every figure comes from _data/impact/stats.json, exactly as on
+the page it came from.
+{%- endcomment -%}
+{%- assign impact = site.data.impact -%}
+{%- assign stats = impact.stats -%}
+
+<div class="impact" markdown="0">
+
+<p class="impact-breadcrumb">
+  <a href="{{ '/impact/#database-systems' | relative_url }}">&larr; Database systems</a>
+</p>
+
+<h2 id="what-is-tracked">What is tracked</h2>
+
+<p>
+  Three different things are tracked separately here, and it is worth being
+  precise about the difference:
+</p>
+
+<ul>
+  <li><strong>Supported</strong> — the main SQLancer repository ships a testing
+      implementation for the system. {{ stats.dbms.supported }} systems.</li>
+  <li><strong>Bugs found</strong> — SQLancer has attributed bugs in the system.
+      {{ stats.dbms.with_bugs }} systems.</li>
+  <li><strong>Uses SQLancer</strong> — the system's <em>own</em> developers use
+      or integrate SQLancer. {{ stats.dbms.with_adoption }} projects.</li>
+  <li><strong>Planned adoption</strong> — a project has proposed adopting
+      SQLancer but has not yet been shown using it. {{ stats.dbms.planning_adoption }}
+      further projects, listed separately below and counted in nothing above.</li>
+</ul>
+
+<p>
+  Supporting a database system says nothing about whether that project uses
+  SQLancer, so adoption is only recorded when the evidence comes from the
+  project or its developers — a CI job, a test script, project documentation, or
+  a developer saying so themselves.
+</p>
+
+{%- assign adoption_rows = stats.dbms.adoption_by_relationship -%}
+{%- if adoption_rows.size > 0 %}
+{% include impact/counts-table.html series=adoption_rows heading="How the project uses SQLancer" label="Projects" %}
+{%- endif %}
+
+<h2 id="every-system">Every system</h2>
+
+<p>
+  One row per system that SQLancer supports, has found a bug in, or that a
+  project has proposed testing with it. Each name links to the system's own
+  page, where the bugs and the adoption evidence are listed in full.
+</p>
+
+<div class="impact-table-wrap">
+  <table class="impact-table">
+    <thead>
+      <tr>
+        <th scope="col">Database system</th>
+        <th scope="col" class="impact-num">Bugs</th>
+        <th scope="col">Supported by SQLancer</th>
+        <th scope="col">Uses SQLancer</th>
+      </tr>
+    </thead>
+    <tbody>
+      {%- for row in stats.dbms.rows %}
+      {%- if row.bugs > 0 or row.supported or row.adoption_relationships.size > 0 or row.planned_adoption %}
+      <tr>
+        <th scope="row">
+          <a href="{{ '/impact/dbms/' | append: row.id | append: '/' | relative_url }}">{{ row.name }}</a>
+        </th>
+        <td class="impact-num">{% if row.bugs > 0 %}{{ row.bugs }}{% else %}&mdash;{% endif %}</td>
+        <td>{% if row.supported %}Yes{% else %}&mdash;{% endif %}</td>
+        <td>
+          {%- if row.adoption_relationships.size > 0 -%}
+            {%- for relationship in row.adoption_relationships -%}
+              <span class="impact-tag">{{ relationship | replace: "_", " " }}</span>
+            {%- endfor -%}
+          {%- elsif row.planned_adoption -%}
+            <span class="impact-meta">proposed only</span>
+          {%- else -%}&mdash;{%- endif -%}
+        </td>
+      </tr>
+      {%- endif %}
+      {%- endfor %}
+    </tbody>
+  </table>
+</div>
+
+{%- comment -%}
+  Only projects that have proposed adopting SQLancer and nothing more. Seven
+  others have an open proposal alongside evidence that they already run it,
+  and for those the proposal is superseded -- listing them here would say the
+  opposite of what their own page says.
+{%- endcomment -%}
+{%- assign all_adoption = impact.adoption.adoption -%}
+{%- assign using_dbms = all_adoption | where_exp: "r", "r.relationship != 'planned_adoption'" | map: "dbms" | uniq -%}
+{%- assign planned_records = "" | split: "" -%}
+{%- for record in all_adoption -%}
+  {%- if record.relationship == "planned_adoption" -%}
+    {%- unless using_dbms contains record.dbms -%}
+      {%- assign planned_records = planned_records | push: record -%}
+    {%- endunless -%}
+  {%- endif -%}
+{%- endfor -%}
+<h2 id="planned-adoption">Planned adoption ({{ planned_records.size }})</h2>
+<p>
+  Projects whose developers have <em>proposed</em> adopting SQLancer &mdash; an
+  open issue, a roadmap entry, a testing plan &mdash; without anything yet
+  showing the project running it. An intention is not use, so none of these
+  counts towards the {{ stats.dbms.with_adoption }} projects that do use it, or
+  towards any figure on the front page. They are kept because a proposal from a
+  project's own developers is worth recording, and because it is the thing to
+  re-check later: some of these will become adoption, and others will not.
+</p>
+
+<div class="impact-table-wrap">
+  <table class="impact-table">
+    <thead>
+      <tr>
+        <th scope="col">Project</th>
+        <th scope="col">Proposal</th>
+      </tr>
+    </thead>
+    <tbody>
+      {%- for record in planned_records %}
+      {%- assign dbms_entry = impact.dbms.dbms | where: "id", record.dbms | first %}
+      <tr>
+        <th scope="row">
+          <a href="{{ '/impact/dbms/' | append: record.dbms | append: '/#proposed' | relative_url }}">{{ dbms_entry.name | default: record.dbms }}</a>
+        </th>
+        <td>{{ record.summary }}</td>
+      </tr>
+      {%- endfor %}
+    </tbody>
+  </table>
+</div>
+
+<p class="impact-meta">
+  Every figure on this page is computed from
+  <a href="https://github.com/sqlancer/sqlancer.github.io/tree/main/_data/impact"><code>_data/impact/</code></a>.
+  The <a href="{{ '/impact/#methodology' | relative_url }}">attribution policy</a>
+  explains what has to be true for a system to be counted as using SQLancer.
+</p>
+
+</div>
diff --git a/_pages/impact/dbms/bharatdbms.html b/_pages/impact/dbms/bharatdbms.html
new file mode 100644
index 0000000..feef2cf
--- /dev/null
+++ b/_pages/impact/dbms/bharatdbms.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/bharatdbms/
+title: "BharatDBMS and SQLancer"
+excerpt: "7 bugs SQLancer found in BharatDBMS, each with its evidence; the project's proposal to adopt SQLancer."
+dbms: bharatdbms
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/citus.html b/_pages/impact/dbms/citus.html
new file mode 100644
index 0000000..5bfaaa0
--- /dev/null
+++ b/_pages/impact/dbms/citus.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/citus/
+title: "Citus and SQLancer"
+excerpt: "16 bugs SQLancer found in Citus, each with its evidence; how the project uses SQLancer."
+dbms: citus
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/clickhouse.html b/_pages/impact/dbms/clickhouse.html
new file mode 100644
index 0000000..a0b102b
--- /dev/null
+++ b/_pages/impact/dbms/clickhouse.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/clickhouse/
+title: "ClickHouse and SQLancer"
+excerpt: "48 bugs SQLancer found in ClickHouse, each with its evidence; how the project uses SQLancer."
+dbms: clickhouse
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/cloudberry.html b/_pages/impact/dbms/cloudberry.html
new file mode 100644
index 0000000..552cdc8
--- /dev/null
+++ b/_pages/impact/dbms/cloudberry.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/cloudberry/
+title: "Apache Cloudberry and SQLancer"
+excerpt: "6 bugs SQLancer found in Apache Cloudberry, each with its evidence; the project's proposal to adopt SQLancer."
+dbms: cloudberry
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/cnosdb.html b/_pages/impact/dbms/cnosdb.html
new file mode 100644
index 0000000..70705d3
--- /dev/null
+++ b/_pages/impact/dbms/cnosdb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/cnosdb/
+title: "CnosDB and SQLancer"
+excerpt: "7 bugs SQLancer found in CnosDB, each with its evidence; how the project uses SQLancer."
+dbms: cnosdb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/cockroachdb.html b/_pages/impact/dbms/cockroachdb.html
new file mode 100644
index 0000000..aa80336
--- /dev/null
+++ b/_pages/impact/dbms/cockroachdb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/cockroachdb/
+title: "CockroachDB and SQLancer"
+excerpt: "202 bugs SQLancer found in CockroachDB, each with its evidence; the project's proposal to adopt SQLancer."
+dbms: cockroachdb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/cratedb.html b/_pages/impact/dbms/cratedb.html
new file mode 100644
index 0000000..ae9f504
--- /dev/null
+++ b/_pages/impact/dbms/cratedb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/cratedb/
+title: "CrateDB and SQLancer"
+excerpt: "81 bugs SQLancer found in CrateDB, each with its evidence."
+dbms: cratedb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/cubrid.html b/_pages/impact/dbms/cubrid.html
new file mode 100644
index 0000000..cafcbef
--- /dev/null
+++ b/_pages/impact/dbms/cubrid.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/cubrid/
+title: "CUBRID and SQLancer"
+excerpt: "1 bugs SQLancer found in CUBRID, each with its evidence."
+dbms: cubrid
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/databend.html b/_pages/impact/dbms/databend.html
new file mode 100644
index 0000000..0b41e52
--- /dev/null
+++ b/_pages/impact/dbms/databend.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/databend/
+title: "Databend and SQLancer"
+excerpt: "18 bugs SQLancer found in Databend, each with its evidence; how the project uses SQLancer."
+dbms: databend
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/datafusion.html b/_pages/impact/dbms/datafusion.html
new file mode 100644
index 0000000..4c4e15e
--- /dev/null
+++ b/_pages/impact/dbms/datafusion.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/datafusion/
+title: "Apache DataFusion and SQLancer"
+excerpt: "54 bugs SQLancer found in Apache DataFusion, each with its evidence; how the project uses SQLancer."
+dbms: datafusion
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/dolt.html b/_pages/impact/dbms/dolt.html
new file mode 100644
index 0000000..30a8a3d
--- /dev/null
+++ b/_pages/impact/dbms/dolt.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/dolt/
+title: "Dolt and SQLancer"
+excerpt: "167 bugs SQLancer found in Dolt, each with its evidence."
+dbms: dolt
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/doris.html b/_pages/impact/dbms/doris.html
new file mode 100644
index 0000000..a80e82d
--- /dev/null
+++ b/_pages/impact/dbms/doris.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/doris/
+title: "Apache Doris and SQLancer"
+excerpt: "19 bugs SQLancer found in Apache Doris, each with its evidence; the project's proposal to adopt SQLancer."
+dbms: doris
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/duckdb.html b/_pages/impact/dbms/duckdb.html
new file mode 100644
index 0000000..0552681
--- /dev/null
+++ b/_pages/impact/dbms/duckdb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/duckdb/
+title: "DuckDB and SQLancer"
+excerpt: "302 bugs SQLancer found in DuckDB, each with its evidence; how the project uses SQLancer."
+dbms: duckdb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/falkordb.html b/_pages/impact/dbms/falkordb.html
new file mode 100644
index 0000000..8e54bca
--- /dev/null
+++ b/_pages/impact/dbms/falkordb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/falkordb/
+title: "FalkorDB and SQLancer"
+excerpt: "1 bugs SQLancer found in FalkorDB, each with its evidence."
+dbms: falkordb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/feldera.html b/_pages/impact/dbms/feldera.html
new file mode 100644
index 0000000..1685666
--- /dev/null
+++ b/_pages/impact/dbms/feldera.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/feldera/
+title: "Feldera and SQLancer"
+excerpt: "How the project uses SQLancer."
+dbms: feldera
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/firebird.html b/_pages/impact/dbms/firebird.html
new file mode 100644
index 0000000..ba03f0e
--- /dev/null
+++ b/_pages/impact/dbms/firebird.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/firebird/
+title: "Firebird and SQLancer"
+excerpt: "10 bugs SQLancer found in Firebird, each with its evidence."
+dbms: firebird
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/greptimedb.html b/_pages/impact/dbms/greptimedb.html
new file mode 100644
index 0000000..b6968cf
--- /dev/null
+++ b/_pages/impact/dbms/greptimedb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/greptimedb/
+title: "GreptimeDB and SQLancer"
+excerpt: "The project's proposal to adopt SQLancer."
+dbms: greptimedb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/h2.html b/_pages/impact/dbms/h2.html
new file mode 100644
index 0000000..fc7e366
--- /dev/null
+++ b/_pages/impact/dbms/h2.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/h2/
+title: "H2 and SQLancer"
+excerpt: "20 bugs SQLancer found in H2, each with its evidence."
+dbms: h2
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/hazelcast.html b/_pages/impact/dbms/hazelcast.html
new file mode 100644
index 0000000..32d4941
--- /dev/null
+++ b/_pages/impact/dbms/hazelcast.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/hazelcast/
+title: "Hazelcast and SQLancer"
+excerpt: "1 bugs SQLancer found in Hazelcast, each with its evidence; how the project uses SQLancer."
+dbms: hazelcast
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/hive.html b/_pages/impact/dbms/hive.html
new file mode 100644
index 0000000..00a402e
--- /dev/null
+++ b/_pages/impact/dbms/hive.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/hive/
+title: "Apache Hive and SQLancer"
+excerpt: "What SQLancer records about Apache Hive."
+dbms: hive
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/hsqldb.html b/_pages/impact/dbms/hsqldb.html
new file mode 100644
index 0000000..01b8cc5
--- /dev/null
+++ b/_pages/impact/dbms/hsqldb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/hsqldb/
+title: "HSQLDB and SQLancer"
+excerpt: "What SQLancer records about HSQLDB."
+dbms: hsqldb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/kyzo.html b/_pages/impact/dbms/kyzo.html
new file mode 100644
index 0000000..97e30ec
--- /dev/null
+++ b/_pages/impact/dbms/kyzo.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/kyzo/
+title: "Kyzo and SQLancer"
+excerpt: "1 bugs SQLancer found in Kyzo, each with its evidence."
+dbms: kyzo
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/mariadb.html b/_pages/impact/dbms/mariadb.html
new file mode 100644
index 0000000..fe24f67
--- /dev/null
+++ b/_pages/impact/dbms/mariadb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/mariadb/
+title: "MariaDB and SQLancer"
+excerpt: "15 bugs SQLancer found in MariaDB, each with its evidence."
+dbms: mariadb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/materialize.html b/_pages/impact/dbms/materialize.html
new file mode 100644
index 0000000..a5a66f1
--- /dev/null
+++ b/_pages/impact/dbms/materialize.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/materialize/
+title: "Materialize and SQLancer"
+excerpt: "How the project uses SQLancer."
+dbms: materialize
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/matrixone.html b/_pages/impact/dbms/matrixone.html
new file mode 100644
index 0000000..6d18429
--- /dev/null
+++ b/_pages/impact/dbms/matrixone.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/matrixone/
+title: "MatrixOne and SQLancer"
+excerpt: "9 bugs SQLancer found in MatrixOne, each with its evidence."
+dbms: matrixone
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/monetdb.html b/_pages/impact/dbms/monetdb.html
new file mode 100644
index 0000000..68268bd
--- /dev/null
+++ b/_pages/impact/dbms/monetdb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/monetdb/
+title: "MonetDB and SQLancer"
+excerpt: "139 bugs SQLancer found in MonetDB, each with its evidence; how the project uses SQLancer."
+dbms: monetdb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/mysql.html b/_pages/impact/dbms/mysql.html
new file mode 100644
index 0000000..6408dba
--- /dev/null
+++ b/_pages/impact/dbms/mysql.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/mysql/
+title: "MySQL and SQLancer"
+excerpt: "43 bugs SQLancer found in MySQL, each with its evidence."
+dbms: mysql
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/noisepage.html b/_pages/impact/dbms/noisepage.html
new file mode 100644
index 0000000..3adff8d
--- /dev/null
+++ b/_pages/impact/dbms/noisepage.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/noisepage/
+title: "NoisePage and SQLancer"
+excerpt: "How the project uses SQLancer."
+dbms: noisepage
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/oceanbase.html b/_pages/impact/dbms/oceanbase.html
new file mode 100644
index 0000000..448f70b
--- /dev/null
+++ b/_pages/impact/dbms/oceanbase.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/oceanbase/
+title: "OceanBase and SQLancer"
+excerpt: "4 bugs SQLancer found in OceanBase, each with its evidence; how the project uses SQLancer."
+dbms: oceanbase
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/oxla.html b/_pages/impact/dbms/oxla.html
new file mode 100644
index 0000000..cacb9d1
--- /dev/null
+++ b/_pages/impact/dbms/oxla.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/oxla/
+title: "Oxla and SQLancer"
+excerpt: "12 bugs SQLancer found in Oxla, each with its evidence; how the project uses SQLancer."
+dbms: oxla
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/postgresql.html b/_pages/impact/dbms/postgresql.html
new file mode 100644
index 0000000..a8ca852
--- /dev/null
+++ b/_pages/impact/dbms/postgresql.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/postgresql/
+title: "PostgreSQL and SQLancer"
+excerpt: "18 bugs SQLancer found in PostgreSQL, each with its evidence."
+dbms: postgresql
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/presto.html b/_pages/impact/dbms/presto.html
new file mode 100644
index 0000000..735fb18
--- /dev/null
+++ b/_pages/impact/dbms/presto.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/presto/
+title: "Presto and SQLancer"
+excerpt: "4 bugs SQLancer found in Presto, each with its evidence; the project's proposal to adopt SQLancer."
+dbms: presto
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/questdb.html b/_pages/impact/dbms/questdb.html
new file mode 100644
index 0000000..8a8d401
--- /dev/null
+++ b/_pages/impact/dbms/questdb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/questdb/
+title: "QuestDB and SQLancer"
+excerpt: "55 bugs SQLancer found in QuestDB, each with its evidence."
+dbms: questdb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/risingwave.html b/_pages/impact/dbms/risingwave.html
new file mode 100644
index 0000000..d10da86
--- /dev/null
+++ b/_pages/impact/dbms/risingwave.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/risingwave/
+title: "RisingWave and SQLancer"
+excerpt: "8 bugs SQLancer found in RisingWave, each with its evidence; the project's proposal to adopt SQLancer."
+dbms: risingwave
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/seekdb.html b/_pages/impact/dbms/seekdb.html
new file mode 100644
index 0000000..f8814cd
--- /dev/null
+++ b/_pages/impact/dbms/seekdb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/seekdb/
+title: "SeekDB and SQLancer"
+excerpt: "3 bugs SQLancer found in SeekDB, each with its evidence; how the project uses SQLancer."
+dbms: seekdb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/serenedb.html b/_pages/impact/dbms/serenedb.html
new file mode 100644
index 0000000..694dc9a
--- /dev/null
+++ b/_pages/impact/dbms/serenedb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/serenedb/
+title: "SereneDB and SQLancer"
+excerpt: "How the project uses SQLancer."
+dbms: serenedb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/spark.html b/_pages/impact/dbms/spark.html
new file mode 100644
index 0000000..cbd816a
--- /dev/null
+++ b/_pages/impact/dbms/spark.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/spark/
+title: "Apache Spark and SQLancer"
+excerpt: "What SQLancer records about Apache Spark."
+dbms: spark
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/sparq.html b/_pages/impact/dbms/sparq.html
new file mode 100644
index 0000000..22b5201
--- /dev/null
+++ b/_pages/impact/dbms/sparq.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/sparq/
+title: "sparq and SQLancer"
+excerpt: "1 bugs SQLancer found in sparq, each with its evidence; how the project uses SQLancer."
+dbms: sparq
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/spiceai.html b/_pages/impact/dbms/spiceai.html
new file mode 100644
index 0000000..0ede947
--- /dev/null
+++ b/_pages/impact/dbms/spiceai.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/spiceai/
+title: "Spice.ai OSS and SQLancer"
+excerpt: "The project's proposal to adopt SQLancer."
+dbms: spiceai
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/sqlite.html b/_pages/impact/dbms/sqlite.html
new file mode 100644
index 0000000..9cb17b6
--- /dev/null
+++ b/_pages/impact/dbms/sqlite.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/sqlite/
+title: "SQLite and SQLancer"
+excerpt: "222 bugs SQLancer found in SQLite, each with its evidence."
+dbms: sqlite
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/starrocks.html b/_pages/impact/dbms/starrocks.html
new file mode 100644
index 0000000..9db94cd
--- /dev/null
+++ b/_pages/impact/dbms/starrocks.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/starrocks/
+title: "StarRocks and SQLancer"
+excerpt: "183 bugs SQLancer found in StarRocks, each with its evidence; how the project uses SQLancer."
+dbms: starrocks
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/stonedb.html b/_pages/impact/dbms/stonedb.html
new file mode 100644
index 0000000..eb86425
--- /dev/null
+++ b/_pages/impact/dbms/stonedb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/stonedb/
+title: "StoneDB and SQLancer"
+excerpt: "16 bugs SQLancer found in StoneDB, each with its evidence."
+dbms: stonedb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/tarantool.html b/_pages/impact/dbms/tarantool.html
new file mode 100644
index 0000000..3615fc9
--- /dev/null
+++ b/_pages/impact/dbms/tarantool.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/tarantool/
+title: "Tarantool and SQLancer"
+excerpt: "The project's proposal to adopt SQLancer."
+dbms: tarantool
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/tdengine.html b/_pages/impact/dbms/tdengine.html
new file mode 100644
index 0000000..4d45b9b
--- /dev/null
+++ b/_pages/impact/dbms/tdengine.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/tdengine/
+title: "TDengine and SQLancer"
+excerpt: "12 bugs SQLancer found in TDengine, each with its evidence."
+dbms: tdengine
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/tidb.html b/_pages/impact/dbms/tidb.html
new file mode 100644
index 0000000..466b274
--- /dev/null
+++ b/_pages/impact/dbms/tidb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/tidb/
+title: "TiDB and SQLancer"
+excerpt: "172 bugs SQLancer found in TiDB, each with its evidence."
+dbms: tidb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/tikv.html b/_pages/impact/dbms/tikv.html
new file mode 100644
index 0000000..d605b5a
--- /dev/null
+++ b/_pages/impact/dbms/tikv.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/tikv/
+title: "TiKV and SQLancer"
+excerpt: "1 bugs SQLancer found in TiKV, each with its evidence."
+dbms: tikv
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/turso.html b/_pages/impact/dbms/turso.html
new file mode 100644
index 0000000..1adf5f1
--- /dev/null
+++ b/_pages/impact/dbms/turso.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/turso/
+title: "Turso and SQLancer"
+excerpt: "18 bugs SQLancer found in Turso, each with its evidence; how the project uses SQLancer."
+dbms: turso
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/umbra.html b/_pages/impact/dbms/umbra.html
new file mode 100644
index 0000000..3ad5f87
--- /dev/null
+++ b/_pages/impact/dbms/umbra.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/umbra/
+title: "Umbra and SQLancer"
+excerpt: "55 bugs SQLancer found in Umbra, each with its evidence."
+dbms: umbra
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/virtuoso.html b/_pages/impact/dbms/virtuoso.html
new file mode 100644
index 0000000..334ad59
--- /dev/null
+++ b/_pages/impact/dbms/virtuoso.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/virtuoso/
+title: "Virtuoso and SQLancer"
+excerpt: "17 bugs SQLancer found in Virtuoso, each with its evidence."
+dbms: virtuoso
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/wadjet.html b/_pages/impact/dbms/wadjet.html
new file mode 100644
index 0000000..baab920
--- /dev/null
+++ b/_pages/impact/dbms/wadjet.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/wadjet/
+title: "Wadjet and SQLancer"
+excerpt: "16 bugs SQLancer found in Wadjet, each with its evidence; how the project uses SQLancer."
+dbms: wadjet
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/xugu.html b/_pages/impact/dbms/xugu.html
new file mode 100644
index 0000000..5fce0e1
--- /dev/null
+++ b/_pages/impact/dbms/xugu.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/xugu/
+title: "XuGu and SQLancer"
+excerpt: "How the project uses SQLancer."
+dbms: xugu
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/ydb.html b/_pages/impact/dbms/ydb.html
new file mode 100644
index 0000000..db35211
--- /dev/null
+++ b/_pages/impact/dbms/ydb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/ydb/
+title: "YDB and SQLancer"
+excerpt: "How the project uses SQLancer."
+dbms: ydb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/dbms/yugabytedb.html b/_pages/impact/dbms/yugabytedb.html
new file mode 100644
index 0000000..675b7ed
--- /dev/null
+++ b/_pages/impact/dbms/yugabytedb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/dbms/yugabytedb/
+title: "YugabyteDB and SQLancer"
+excerpt: "54 bugs SQLancer found in YugabyteDB, each with its evidence; how the project uses SQLancer."
+dbms: yugabytedb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/dbms-detail.html id=page.dbms %}
diff --git a/_pages/impact/papers/paper_arxiv_2105_10016.html b/_pages/impact/papers/paper_arxiv_2105_10016.html
new file mode 100644
index 0000000..044d5b6
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2105_10016.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2105_10016/
+title: "Testing DBMS Performance with Mutations"
+excerpt: "AMOEBA takes TLP as the closest related effort and argues it does not transfer: its equivalent queries are tailored to logic bugs and are comparatively simple, so they cannot serve as a metamorphic relation for performance."
+paper_key: paper_arxiv_2105_10016
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2206_08530.html b/_pages/impact/papers/paper_arxiv_2206_08530.html
new file mode 100644
index 0000000..e920a37
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2206_08530.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2206_08530/
+title: "GDsmith: Detecting Bugs in Graph Database Engines"
+excerpt: "GDsmith says plainly that its framework is derived from SQLancer, carrying that tool's approach from relational engines to Cypher, and its artifact holds SQLancer's source under a renamed package."
+paper_key: paper_arxiv_2206_08530
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2304_10044.html b/_pages/impact/papers/paper_arxiv_2304_10044.html
new file mode 100644
index 0000000..39ee3cb
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2304_10044.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2304_10044/
+title: "Finding Bug-Inducing Program Environments"
+excerpt: "Two citations, listing databases among the domains with dedicated fuzzing work."
+paper_key: paper_arxiv_2304_10044
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2310_06433.html b/_pages/impact/papers/paper_arxiv_2310_06433.html
new file mode 100644
index 0000000..2275a75
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2310_06433.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2310_06433/
+title: "Retromorphic Testing: A New Approach to the Test Oracle Problem"
+excerpt: "PQS is the paper's worked example of its own concept: the authors argue that although PQS was described as a technique for a specific system, it is an instance of retromorphic testing, and record that an author of PQS agreed with that categorisation."
+paper_key: paper_arxiv_2310_06433
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2311_06728.html b/_pages/impact/papers/paper_arxiv_2311_06728.html
new file mode 100644
index 0000000..8b4dc0b
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2311_06728.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2311_06728/
+title: "A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Experimental Comparison"
+excerpt: "SQLancer's oracles are a fixed point of the survey's taxonomy."
+paper_key: paper_arxiv_2311_06728
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2312_04941.html b/_pages/impact/papers/paper_arxiv_2312_04941.html
new file mode 100644
index 0000000..f28355d
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2312_04941.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2312_04941/
+title: "Detecting DBMS Bugs with Context-Sensitive Instantiation and Multi-Plan Execution"
+excerpt: "SQLancer is Kangaroo's baseline and the source of the limitation it targets."
+paper_key: paper_arxiv_2312_04941
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2402_00292.html b/_pages/impact/papers/paper_arxiv_2402_00292.html
new file mode 100644
index 0000000..1e6b1da
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2402_00292.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2402_00292/
+title: "Effective Bug Detection in Graph Database Engines: An LLM-based Approach"
+excerpt: "TLP reaches this paper twice over."
+paper_key: paper_arxiv_2402_00292
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2406_09469.html b/_pages/impact/papers/paper_arxiv_2406_09469.html
new file mode 100644
index 0000000..726dda8
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2406_09469.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2406_09469/
+title: "Conformance Testing of Relational DBMS Against SQL Specifications"
+excerpt: "SQLancer is SEMCONT's generator, extended."
+paper_key: paper_arxiv_2406_09469
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2407_04294.html b/_pages/impact/papers/paper_arxiv_2407_04294.html
new file mode 100644
index 0000000..3230cbf
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2407_04294.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2407_04294/
+title: "SQLaser: Detecting DBMS Logic Bugs with Clause-Guided Fuzzing"
+excerpt: "SQLancer's oracles are what SQLaser runs and what bounds it."
+paper_key: paper_arxiv_2407_04294
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2503_03893.html b/_pages/impact/papers/paper_arxiv_2503_03893.html
new file mode 100644
index 0000000..9bfb662
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2503_03893.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2503_03893/
+title: "Parser Knows Best: Testing DBMS with Coverage-Guided Grammar-Rule Traversal"
+excerpt: "SQLancer, including its QPG configuration, is the generation-based tool ParserFuzz argues against: because SQLancer restricts itself to queries matching its oracles' patterns, the paper says it lacks the diversity needed to reach memory corruption bugs, which is the gap grammar-r."
+paper_key: paper_arxiv_2503_03893
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2503_17322.html b/_pages/impact/papers/paper_arxiv_2503_17322.html
new file mode 100644
index 0000000..bdb192f
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2503_17322.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2503_17322/
+title: "QITE: Assembly-Level, Cross-Platform Testing of Quantum Computing Platforms"
+excerpt: "One citation in related work, noting that grammar-based generators similar to QITE's approach have been applied successfully to compilers and similar software, SQLancer among the examples."
+paper_key: paper_arxiv_2503_17322
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2506_02617.html b/_pages/impact/papers/paper_arxiv_2506_02617.html
new file mode 100644
index 0000000..840bc6c
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2506_02617.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2506_02617/
+title: "Toward Understanding Bugs in Vector Database Management Systems"
+excerpt: "One citation, introducing NoREC as a method for detecting optimization bugs in query engines, in a review of how relational DBMS reliability has been studied."
+paper_key: paper_arxiv_2506_02617
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2509_10819.html b/_pages/impact/papers/paper_arxiv_2509_10819.html
new file mode 100644
index 0000000..2945554
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2509_10819.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2509_10819/
+title: "Arguzz: Testing zkVMs for Soundness and Completeness Bugs"
+excerpt: "No SQLancer mention is present."
+paper_key: paper_arxiv_2509_10819
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2510_06663.html b/_pages/impact/papers/paper_arxiv_2510_06663.html
new file mode 100644
index 0000000..91a9c54
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2510_06663.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2510_06663/
+title: "Automated Discovery of Test Oracles for Database Management Systems Using LLMs"
+excerpt: "Argus uses SQLancer++'s query generator to produce seed queries while deliberately not using its predefined oracles, since discovering oracles is the paper's own subject."
+paper_key: paper_arxiv_2510_06663
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2511_17377.html b/_pages/impact/papers/paper_arxiv_2511_17377.html
new file mode 100644
index 0000000..639c350
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2511_17377.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2511_17377/
+title: "Anomaly Pattern-guided Transaction Bug Testing in Relational Databases"
+excerpt: "APTrans's database and SQL generation is SQLancer's, extended."
+paper_key: paper_arxiv_2511_17377
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2601_15074.html b/_pages/impact/papers/paper_arxiv_2601_15074.html
new file mode 100644
index 0000000..0ba31c6
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2601_15074.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2601_15074/
+title: "SmartOracle - An Agentic Approach to Mitigate Noise in Differential Oracles"
+excerpt: "One citation, listing database systems among the typical applications of differential fuzzing."
+paper_key: paper_arxiv_2601_15074
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2602_19490.html b/_pages/impact/papers/paper_arxiv_2602_19490.html
new file mode 100644
index 0000000..ddf0bd2
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2602_19490.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2602_19490/
+title: "FuzzySQL: Uncovering Hidden Vulnerabilities in DBMS Special Features with LLM-Driven Fuzzing"
+excerpt: "SQLancer is one of three state-of-the-art open-source baselines, alongside Squirrel and EET."
+paper_key: paper_arxiv_2602_19490
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2603_00311.html b/_pages/impact/papers/paper_arxiv_2603_00311.html
new file mode 100644
index 0000000..238b231
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2603_00311.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2603_00311/
+title: "Towards the Systematic Testing of Regular Expression Engines"
+excerpt: "One citation, noting that Rigger and Su used metamorphic oracles to detect logic bugs in DBMSs -- the precedent for using them where no reference implementation can be trusted."
+paper_key: paper_arxiv_2603_00311
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2603_19434.html b/_pages/impact/papers/paper_arxiv_2603_19434.html
new file mode 100644
index 0000000..cf7ed37
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2603_19434.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2603_19434/
+title: "Computer-Orchestrated Design of Algorithms: From Join Specification to Implementation"
+excerpt: "Two citations, treating SQLancer's line of work as the conventional database testing frameworks CODA is measured against in argument -- the natural choice for verifying a physical translation, which the authors then explain does not fit their need."
+paper_key: paper_arxiv_2603_19434
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2603_21530.html b/_pages/impact/papers/paper_arxiv_2603_21530.html
new file mode 100644
index 0000000..381404d
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2603_21530.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2603_21530/
+title: "LLM-Based Test Case Generation in DBMS through Monte Carlo Tree Search"
+excerpt: "SQLancer is cited as the traditional approach whose handcrafted dialect-specific generators and sophisticated oracles work well for a specific system but need substantial manual effort to move to another -- the cost MIST aims to remove."
+paper_key: paper_arxiv_2603_21530
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2604_01442.html b/_pages/impact/papers/paper_arxiv_2604_01442.html
new file mode 100644
index 0000000..006cfb4
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2604_01442.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2604_01442/
+title: "Fuzzing with Agents? Generators Are All You Need"
+excerpt: "One citation, naming SQLancer as the database-engine counterpart to CSmith for C compilers: a generator producing inputs that are more than syntactically valid."
+paper_key: paper_arxiv_2604_01442
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2604_03024.html b/_pages/impact/papers/paper_arxiv_2604_03024.html
new file mode 100644
index 0000000..76d3a2e
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2604_03024.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2604_03024/
+title: "BugForge: Constructing and Utilizing DBMS Bug Repository to Enhance DBMS Testing"
+excerpt: "SQLancer is one of two state-of-the-art tools BugForge is compared against, run in its FUZZ mode, with branch coverage and bug counts reported per DBMS -- 441,160 branches to SQLancer's 201,299, and 18 bugs to its 11 over the three jointly supported systems."
+paper_key: paper_arxiv_2604_03024
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2605_20473.html b/_pages/impact/papers/paper_arxiv_2605_20473.html
new file mode 100644
index 0000000..a641c9b
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2605_20473.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2605_20473/
+title: "Code Generation by Differential Test Time Scaling"
+excerpt: "Two citations, listing database systems among the domains where differential testing has proven effective where formal specifications are unavailable."
+paper_key: paper_arxiv_2605_20473
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2605_22992.html b/_pages/impact/papers/paper_arxiv_2605_22992.html
new file mode 100644
index 0000000..4395fdc
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2605_22992.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2605_22992/
+title: "Finding Performance Issues in Database Systems by Exploiting Dormant Code Paths"
+excerpt: "SQLancer is used two ways."
+paper_key: paper_arxiv_2605_22992
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2606_11132.html b/_pages/impact/papers/paper_arxiv_2606_11132.html
new file mode 100644
index 0000000..13567fe
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2606_11132.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2606_11132/
+title: "Operationalizing Property-Based Testing for Data-Intensive Scalable Computing Systems"
+excerpt: "SQLancer is the demonstration the paper builds on -- credited with showing the effectiveness of query oracles including TLP -- and is explicitly excluded from the evaluation: the authors say they do not compare against it because using it for Spark would require an additional ada."
+paper_key: paper_arxiv_2606_11132
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2606_14164.html b/_pages/impact/papers/paper_arxiv_2606_14164.html
new file mode 100644
index 0000000..8e0640a
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2606_14164.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2606_14164/
+title: "Investigating Metamorphic Fuzz Oracle Enhancement via Large Language Models"
+excerpt: "One author-year citation, among the metamorphic testing work the paper builds from."
+paper_key: paper_arxiv_2606_14164
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2607_03741.html b/_pages/impact/papers/paper_arxiv_2607_03741.html
new file mode 100644
index 0000000..a4e6c66
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2607_03741.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2607_03741/
+title: "Graph-Aware Fuzzing for Graph Database Management Systems"
+excerpt: "SQLancer is cited as the relational tool that synthesises queries and validates them with advanced oracles, and the graph-database descendants of its techniques -- GraphGenie's transformations and GDBMeter's adaptation of ternary query partitioning -- are the related work GRAF se."
+paper_key: paper_arxiv_2607_03741
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2607_09072.html b/_pages/impact/papers/paper_arxiv_2607_09072.html
new file mode 100644
index 0000000..d6090c5
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2607_09072.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2607_09072/
+title: "Agentic Proof and Property-Based Testing via Property-Templates in Data-Intensive Computing"
+excerpt: "One citation, describing SQLancer as detecting logic and optimization bugs in database engines through constructed oracles such as query partitioning -- the testing-side precedent for the paper's own property-based track."
+paper_key: paper_arxiv_2607_09072
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2607_13276.html b/_pages/impact/papers/paper_arxiv_2607_13276.html
new file mode 100644
index 0000000..c599ced
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2607_13276.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2607_13276/
+title: "Aurora DSQL: Scalable, Multi-Region OLTP"
+excerpt: "SQLancer is used in production testing rather than cited as related work: the paper says its fuzz-testing approach builds on SQLancer's, generating millions of SQL statements and running them against both DSQL and a reference."
+paper_key: paper_arxiv_2607_13276
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2608_15709.html b/_pages/impact/papers/paper_arxiv_2608_15709.html
new file mode 100644
index 0000000..3820e46
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2608_15709.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2608_15709/
+title: "Logos: Certified Order-Sensitive SQL Rewrites with Mechanized Semantics and LLM Guidance"
+excerpt: "One citation in threats to validity, noting that tools such as NoREC expose optimizer bugs by constructing differential or metamorphic oracles -- the empirical counterpart to what Logos proves."
+paper_key: paper_arxiv_2608_15709
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2608_23402.html b/_pages/impact/papers/paper_arxiv_2608_23402.html
new file mode 100644
index 0000000..413b5e3
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2608_23402.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2608_23402/
+title: "EXPLAIN Yourself! Finding Query Planner Stalls Across DBMSes"
+excerpt: "SQLancer's work is the related-work ground this paper stands on, and all seven of its oracle papers are cited: CERT, QPG and DQP as the performance-bug line, and NoREC, TLP, PQS and CODDTest as the correctness line whose query-pair approach it describes as similar in kind to perf."
+paper_key: paper_arxiv_2608_23402
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2608_25573.html b/_pages/impact/papers/paper_arxiv_2608_25573.html
new file mode 100644
index 0000000..1e5cb62
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2608_25573.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2608_25573/
+title: "DBcover: A White-box SQL Test Generation Framework for Coverage Improvement"
+excerpt: "ShQveL, which enhances SQLancer++ with LLM-synthesised SQL fragments, is one of DBcover's baselines and is reported to achieve substantially lower coverage because it lacks the seed corpus and context DBcover uses."
+paper_key: paper_arxiv_2608_25573
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2608_30385.html b/_pages/impact/papers/paper_arxiv_2608_30385.html
new file mode 100644
index 0000000..31f67d7
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2608_30385.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2608_30385/
+title: "Detecting DBMS Bugs by Constructing Equivalent Representations of Intermediate Query Results"
+excerpt: "TLP is one of ERIQ's four baselines, named among state-of-the-art DBMS logic-bug detection approaches alongside EDC, Radar and EET, and included partly because EDC's own evaluation used it."
+paper_key: paper_arxiv_2608_30385
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_arxiv_2609_00381.html b/_pages/impact/papers/paper_arxiv_2609_00381.html
new file mode 100644
index 0000000..d90aa16
--- /dev/null
+++ b/_pages/impact/papers/paper_arxiv_2609_00381.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_arxiv_2609_00381/
+title: "Bounded, Indeterminate, or a Bug: A Condition-Aware Oracle for Differential Testing of SQL Aggregates"
+excerpt: "SQLancer's oracles are the paper's point of departure: it argues each of the three avoids the case it addresses -- PQS restricting aggregates to a single pivot row, NoREC rewriting predicates, TLP comparing an engine against itself so a consistent rounding error cancels."
+paper_key: paper_arxiv_2609_00381
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1007_978_3_030_71058_3_5.html b/_pages/impact/papers/paper_doi_10_1007_978_3_030_71058_3_5.html
new file mode 100644
index 0000000..d23dc9d
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1007_978_3_030_71058_3_5.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1007_978_3_030_71058_3_5/
+title: "Artemis: An Automatic Test Suite Generator for Large Scale OLAP Database"
+excerpt: "The single mention is the motivation the paper shares with SQLancer's work: that generating many queries in parallel yields statements which may be syntactically correct yet never run through the deep logic behind the code."
+paper_key: paper_doi_10_1007_978_3_030_71058_3_5
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1007_978_3_030_88494_9_12.html b/_pages/impact/papers/paper_doi_10_1007_978_3_030_88494_9_12.html
new file mode 100644
index 0000000..93f98a4
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1007_978_3_030_88494_9_12.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1007_978_3_030_88494_9_12/
+title: "Differential Monitoring"
+excerpt: "SQL databases are named as one of the domains where this style of testing has been applied fruitfully, cited alongside JavaScript debuggers and C compilers -- the SQLancer papers are the citation behind the SQL databases half of that sentence."
+paper_key: paper_doi_10_1007_978_3_030_88494_9_12
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1007_978_3_031_51479_1_17.html b/_pages/impact/papers/paper_doi_10_1007_978_3_031_51479_1_17.html
new file mode 100644
index 0000000..bdce725
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1007_978_3_031_51479_1_17.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1007_978_3_031_51479_1_17/
+title: "Syntax-Aware Mutation for Testing the Solidity Compiler"
+excerpt: "Data-oriented systems are cited once, in the introduction's list of domains fuzzing has been applied to, alongside system libraries, web and cloud applications, and compilers."
+paper_key: paper_doi_10_1007_978_3_031_51479_1_17
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1007_978_3_031_94706_3_7.html b/_pages/impact/papers/paper_doi_10_1007_978_3_031_94706_3_7.html
new file mode 100644
index 0000000..dc2621f
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1007_978_3_031_94706_3_7.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1007_978_3_031_94706_3_7/
+title: "Fuzzing Graph Database Applications with Graph Transformations"
+excerpt: "SQLancer's line of work is cited once, in related work, among the recent efforts to test database management systems and graph databases."
+paper_key: paper_doi_10_1007_978_3_031_94706_3_7
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1007_978_981_95_3182_0_3.html b/_pages/impact/papers/paper_doi_10_1007_978_981_95_3182_0_3.html
new file mode 100644
index 0000000..eafbea9
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1007_978_981_95_3182_0_3.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1007_978_981_95_3182_0_3/
+title: "Ghosts in DBMS: Revealing the Security Impacts of Silent Fixes"
+excerpt: "Query partitioning is named in related work as one of the techniques prior efforts have used to find these issues, listed with validity-guided fuzzing, graph-based query transformations and syntax abstraction frameworks."
+paper_key: paper_doi_10_1007_978_981_95_3182_0_3
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1007_978_981_95_4721_0_7.html b/_pages/impact/papers/paper_doi_10_1007_978_981_95_4721_0_7.html
new file mode 100644
index 0000000..91a1d44
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1007_978_981_95_4721_0_7.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1007_978_981_95_4721_0_7/
+title: "CypherFuzzer: A Tool for Testing Access Control in Graph Databases"
+excerpt: "SQLancer's techniques appear as part of the body of engine-focused fuzzing the paper positions itself against."
+paper_key: paper_doi_10_1007_978_981_95_4721_0_7
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1007_978_981_96_4506_0_18.html b/_pages/impact/papers/paper_doi_10_1007_978_981_96_4506_0_18.html
new file mode 100644
index 0000000..494c173
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1007_978_981_96_4506_0_18.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1007_978_981_96_4506_0_18/
+title: "Review of Fuzz Testing Techniques for Database Management Systems"
+excerpt: "SQLancer is one of the survey's principal subjects, given its own subsection as a well-known tool for detecting logical vulnerabilities and covered oracle by oracle with the bug counts each produced: over 60 logic bugs from PQS across MySQL, PostgreSQL and SQLite; 51 from NoREC a."
+paper_key: paper_doi_10_1007_978_981_96_4506_0_18
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1007_978_981_96_6465_8_28.html b/_pages/impact/papers/paper_doi_10_1007_978_981_96_6465_8_28.html
new file mode 100644
index 0000000..54c45ac
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1007_978_981_96_6465_8_28.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1007_978_981_96_6465_8_28/
+title: "Fuzz Testing for Database Management System Configuration Errors"
+excerpt: "SQLancer is described as background: a fuzz testing tool integrating TLP, PQS and NoREC to detect logic errors, with each of the three oracles explained in turn."
+paper_key: paper_doi_10_1007_978_981_96_6465_8_28
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1007_s10664_025_10662_w.html b/_pages/impact/papers/paper_doi_10_1007_s10664_025_10662_w.html
new file mode 100644
index 0000000..b74752f
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1007_s10664_025_10662_w.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1007_s10664_025_10662_w/
+title: "Detecting data manipulation errors in android applications using scene-guided exploration"
+excerpt: "The single mention places SQLancer's work as the database-side precedent: the paper notes that existing studies have detected CRUD errors in database management systems, but that those approaches are not tailored for Android apps, where the errors surface through the application'."
+paper_key: paper_doi_10_1007_s10664_025_10662_w
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1016_j_cose_2025_104564.html b/_pages/impact/papers/paper_doi_10_1016_j_cose_2025_104564.html
new file mode 100644
index 0000000..863f842
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1016_j_cose_2025_104564.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1016_j_cose_2025_104564/
+title: "Detecting DBMS bugs with context-sensitive instantiation and multi-plan execution"
+excerpt: "Kangaroo argues that SQLancer's oracles constrain the SQL they can test -- NoREC needing a WHERE clause, all three putting limits on queries -- and positions multi-plan execution as free of that constraint."
+paper_key: paper_doi_10_1016_j_cose_2025_104564
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1016_j_displa_2024_102854.html b/_pages/impact/papers/paper_doi_10_1016_j_displa_2024_102854.html
new file mode 100644
index 0000000..67d1e97
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1016_j_displa_2024_102854.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1016_j_displa_2024_102854/
+title: "Using query semantic and feature transfer fusion to enhance cardinality estimating of property graph queries"
+excerpt: "The connection is a single background citation in the opening sentence on cardinality estimation, reachable only through the citation marker -- SQLancer is never named."
+paper_key: paper_doi_10_1016_j_displa_2024_102854
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_ase56229_2023_00106.html b/_pages/impact/papers/paper_doi_10_1109_ase56229_2023_00106.html
new file mode 100644
index 0000000..028472d
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_ase56229_2023_00106.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_ase56229_2023_00106/
+title: "Perfce: Performance Debugging on Databases with Chaos Engineering-Enhanced Causality Analysis"
+excerpt: "SQLancer is cited once, as one of the works establishing the standard differential testing procedure that PerfCE's setup extends by requiring consistency between an experimental and a reference group."
+paper_key: paper_doi_10_1109_ase56229_2023_00106
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_ase63991_2025_00095.html b/_pages/impact/papers/paper_doi_10_1109_ase63991_2025_00095.html
new file mode 100644
index 0000000..073f34d
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_ase63991_2025_00095.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_ase63991_2025_00095/
+title: "ZendDiff: Differential Testing of PHP Interpreter"
+excerpt: "Almost every citation here is to FlowFusion, a PHP fuzzer co-authored by one of SQLancer's authors, which ZendDiff adopts as its program generator and measures against -- not to SQLancer itself."
+paper_key: paper_doi_10_1109_ase63991_2025_00095
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_ase63991_2025_00151.html b/_pages/impact/papers/paper_doi_10_1109_ase63991_2025_00151.html
new file mode 100644
index 0000000..442be06
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_ase63991_2025_00151.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_ase63991_2025_00151/
+title: "ARG: Testing Query Rewriters via Abstract Rule Guided Fuzzing"
+excerpt: "ARG integrates SQLancer to construct dynamic database schemas and populate them, and then measures itself against it, reporting 1017% more rewrite rules triggered and 15 more bugs in 24 hours."
+paper_key: paper_doi_10_1109_ase63991_2025_00151
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_ase63991_2025_00322.html b/_pages/impact/papers/paper_doi_10_1109_ase63991_2025_00322.html
new file mode 100644
index 0000000..d166e63
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_ase63991_2025_00322.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_ase63991_2025_00322/
+title: "LLM-based Dynamic Differential Testing for Database Connectors with Reinforcement Learning-Guided Prompt Selection"
+excerpt: "PQS is cited among the DBMS fuzzing techniques whose success does not carry over to connectors."
+paper_key: paper_doi_10_1109_ase63991_2025_00322
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_compsac57700_2023_00273.html b/_pages/impact/papers/paper_doi_10_1109_compsac57700_2023_00273.html
new file mode 100644
index 0000000..115d92e
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_compsac57700_2023_00273.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_compsac57700_2023_00273/
+title: "Detecting Hidden Failures of DBMS: A Comprehensive Metamorphic Relation Output Patterns Approach"
+excerpt: "No SQLancer mention could be extracted."
+paper_key: paper_doi_10_1109_compsac57700_2023_00273
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_compsac61105_2024_00141.html b/_pages/impact/papers/paper_doi_10_1109_compsac61105_2024_00141.html
new file mode 100644
index 0000000..3e3cbe9
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_compsac61105_2024_00141.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_compsac61105_2024_00141/
+title: "SQLPass: A Semantic Effective Fuzzing Method for DBMS"
+excerpt: "SQLancer is one of four baselines, compared on semantic correctness, line coverage and bug replication speed."
+paper_key: paper_doi_10_1109_compsac61105_2024_00141
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_dsc55868_2022_00057.html b/_pages/impact/papers/paper_doi_10_1109_dsc55868_2022_00057.html
new file mode 100644
index 0000000..15524fd
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_dsc55868_2022_00057.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_dsc55868_2022_00057/
+title: "Fuzzing DBMS via NNLM"
+excerpt: "Three citations describing SQLancer's oracles -- PQS, NoREC and query partitioning -- as the established ways of detecting correctness errors that do not crash the system."
+paper_key: paper_doi_10_1109_dsc55868_2022_00057
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_girst67753_2025_11382165.html b/_pages/impact/papers/paper_doi_10_1109_girst67753_2025_11382165.html
new file mode 100644
index 0000000..6e367cb
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_girst67753_2025_11382165.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_girst67753_2025_11382165/
+title: "Testing Spatial Database Engines via Spatial Equivalent Transformation"
+excerpt: "SQLancer's oracles are cited collectively as the mature body of logic-bug detection for relational systems, against which the paucity of spatial work is measured."
+paper_key: paper_doi_10_1109_girst67753_2025_11382165
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_iaecst64597_2024_11117732.html b/_pages/impact/papers/paper_doi_10_1109_iaecst64597_2024_11117732.html
new file mode 100644
index 0000000..5aa319e
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_iaecst64597_2024_11117732.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_iaecst64597_2024_11117732/
+title: "Detecting Logical Bugs in DBMS via Isomerism Fuzz System"
+excerpt: "The system is built inside SQLancer: the paper states it was implemented within SQLancer, and that its query generation draws on SQLancer's design principles with modifications."
+paper_key: paper_doi_10_1109_iaecst64597_2024_11117732
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_iaecst68792_2025_11415166.html b/_pages/impact/papers/paper_doi_10_1109_iaecst68792_2025_11415166.html
new file mode 100644
index 0000000..adc157e
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_iaecst68792_2025_11415166.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_iaecst68792_2025_11415166/
+title: "Semantic Hint-Based Fuzzing for Time-Series Databases"
+excerpt: "SQLancer is the pioneer the paper builds from and the limitation it works around: it is credited with pioneering systematic bug detection through oracles such as NoREC and TLP, and then said to assume relational semantics, static schemas and Boolean logic, which is what makes it ."
+paper_key: paper_doi_10_1109_iaecst68792_2025_11415166
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icccs65393_2025_11069832.html b/_pages/impact/papers/paper_doi_10_1109_icccs65393_2025_11069832.html
new file mode 100644
index 0000000..cf0005f
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icccs65393_2025_11069832.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icccs65393_2025_11069832/
+title: "OptionFuzz: The Fuzzer with Coverage-Guided Dynamic Configuration Scheduling"
+excerpt: "Database systems appear in the opening survey of targets fuzzing has been extended to, alongside operating systems, virtual machine managers, network protocols and autonomous vehicles."
+paper_key: paper_doi_10_1109_icccs65393_2025_11069832
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_iccste65902_2025_11138310.html b/_pages/impact/papers/paper_doi_10_1109_iccste65902_2025_11138310.html
new file mode 100644
index 0000000..2782d26
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_iccste65902_2025_11138310.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_iccste65902_2025_11138310/
+title: "Research on Coverage Statistics in Fuzz Testing of Closed-Source DBMS"
+excerpt: "SQLancer is cited once, in the introduction's survey of DBMS testing: it and Troc are described as employing pre-established oracles to detect logic and transaction isolation errors."
+paper_key: paper_doi_10_1109_iccste65902_2025_11138310
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icde55515_2023_00057.html b/_pages/impact/papers/paper_doi_10_1109_icde55515_2023_00057.html
new file mode 100644
index 0000000..ee2a7d9
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icde55515_2023_00057.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icde55515_2023_00057/
+title: "Sequence-Oriented DBMS Fuzzing"
+excerpt: "SQLancer is one of the three fuzzers Lego is measured against, described as a state-of-the-art academic fuzzer that generates from custom rules mainly for SELECT statements."
+paper_key: paper_doi_10_1109_icde55515_2023_00057
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icde60146_2024_00011.html b/_pages/impact/papers/paper_doi_10_1109_icde60146_2024_00011.html
new file mode 100644
index 0000000..15e83cf
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icde60146_2024_00011.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icde60146_2024_00011/
+title: "TRAP: Tailored Robustness Assessment for Index Advisors via Adversarial Perturbation"
+excerpt: "One citation, describing TLP as a way of deriving multiple queries by partitioning the results of an original query, in a discussion of SQL generation."
+paper_key: paper_doi_10_1109_icde60146_2024_00011
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icde60146_2024_00441.html b/_pages/impact/papers/paper_doi_10_1109_icde60146_2024_00441.html
new file mode 100644
index 0000000..7f66e02
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icde60146_2024_00441.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icde60146_2024_00441/
+title: "Applications and Challenges for Large Language Models: From Data Management Perspective"
+excerpt: "The citation supports two sentences about what DBMS testing requires: that comprehensive bug detection needs a large volume of SQL queries fed to the system, and that finding logic bugs specifically requires generating semantically equivalent queries that should return the same r."
+paper_key: paper_doi_10_1109_icde60146_2024_00441
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icde65706_2026_00180.html b/_pages/impact/papers/paper_doi_10_1109_icde65706_2026_00180.html
new file mode 100644
index 0000000..6245be9
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icde65706_2026_00180.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icde65706_2026_00180/
+title: "VIREO: Human-in-the-Loop DBMS Fuzzing with Visualization and LLM Support"
+excerpt: "SQLancer is one of four baselines, named among the state-of-the-art DBMS fuzzers VIREO measures against, and run with its default PQS oracle since the comparison is about crashes rather than logic bugs."
+paper_key: paper_doi_10_1109_icde65706_2026_00180
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icde65706_2026_00224.html b/_pages/impact/papers/paper_doi_10_1109_icde65706_2026_00224.html
new file mode 100644
index 0000000..1da68e8
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icde65706_2026_00224.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icde65706_2026_00224/
+title: "A Set-Theoretic Approach to Detecting Logic Bugs in DBMS Inner Join Optimizations"
+excerpt: "The test case generation is built on SQLancer, which the authors extend so that all generated columns are NOT NULL, and it relies on SQLancer's randomised population of predicates and clauses."
+paper_key: paper_doi_10_1109_icde65706_2026_00224
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icde65706_2026_00240.html b/_pages/impact/papers/paper_doi_10_1109_icde65706_2026_00240.html
new file mode 100644
index 0000000..c5a5207
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icde65706_2026_00240.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icde65706_2026_00240/
+title: "LLMSQLMUTATOR: LLM-Powered Test Case Generation for Database Using Bug Reports"
+excerpt: "SQLancer is both a baseline and a component."
+paper_key: paper_doi_10_1109_icde65706_2026_00240
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icicnis64247_2024_10823213.html b/_pages/impact/papers/paper_doi_10_1109_icicnis64247_2024_10823213.html
new file mode 100644
index 0000000..73698e0
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icicnis64247_2024_10823213.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icicnis64247_2024_10823213/
+title: "Dynamic Adjustment Paradigm based on Genetic Algorithm in Database Index Optimization"
+excerpt: "NoREC is described in the review of database optimization work: the paper summarises how it constructs a non-optimizing reference engine and detects differences between optimized and non-optimized evaluation, and notes that experiments show it effectively detects optimization err."
+paper_key: paper_doi_10_1109_icicnis64247_2024_10823213
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icpc66645_2025_00021.html b/_pages/impact/papers/paper_doi_10_1109_icpc66645_2025_00021.html
new file mode 100644
index 0000000..8279fb6
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icpc66645_2025_00021.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icpc66645_2025_00021/
+title: "Sembug: Detecting Logic Bugs in Dbms Through Generating Semantic-Aware Non-Optimizing Query"
+excerpt: "SemBug is a direct descendant of NoREC: it says it transforms the optimized query into a less optimized one 'inspired by NoREC', and its evaluation treats NoREC, TLP and Pinolo as the state-of-the-art techniques to beat."
+paper_key: paper_doi_10_1109_icpc66645_2025_00021
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse43902_2021_00137.html b/_pages/impact/papers/paper_doi_10_1109_icse43902_2021_00137.html
new file mode 100644
index 0000000..aa12333
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse43902_2021_00137.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse43902_2021_00137/
+title: "Data-Oriented Differential Testing of Object-Relational Mapping Systems"
+excerpt: "All three SQLancer oracles are described in detail in the discussion -- PQS forming its oracle by fetching a specific record, NoREC's semantics-preserving transformation, TLP deriving partial-result queries -- as the DBMS-level precedent for Cynthia's own differential testing one."
+paper_key: paper_doi_10_1109_icse43902_2021_00137
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00024.html b/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00024.html
new file mode 100644
index 0000000..296e844
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00024.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse48619_2023_00024/
+title: "A Comprehensive Study of Real-World Bugs in Machine Learning Model Optimization"
+excerpt: "Databases are named once, in the opening sentence, as another domain where optimisation over complex objects is error-prone -- alongside program compilers."
+paper_key: paper_doi_10_1109_icse48619_2023_00024
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00101.html b/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00101.html
new file mode 100644
index 0000000..1dfa694
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00101.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse48619_2023_00101/
+title: "Detecting Isolation Bugs via Transaction Oracle Construction"
+excerpt: "SQLancer supplies Troc's generation and defines the gap Troc fills."
+paper_key: paper_doi_10_1109_icse48619_2023_00101
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00173.html b/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00173.html
new file mode 100644
index 0000000..32ebcb2
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00173.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse48619_2023_00173/
+title: "Generating Test Databases for Database-Backed Applications"
+excerpt: "SQLancer is described in related work as including a series of techniques -- query partitioning and pivot query synthesis are named -- that generate SQL queries paired with known properties of their results."
+paper_key: paper_doi_10_1109_icse48619_2023_00173
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00175.html b/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00175.html
new file mode 100644
index 0000000..5dc0e53
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse48619_2023_00175.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse48619_2023_00175/
+title: "Testing Database Systems via Differential Query Execution"
+excerpt: "DQE is built on SQLancer."
+paper_key: paper_doi_10_1109_icse48619_2023_00175
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00003.html b/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00003.html
new file mode 100644
index 0000000..2acd47f
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00003.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse55347_2025_00003/
+title: "Coni: Detecting Database Connector Bugs via State-Aware Test Case Generation"
+excerpt: "SQLancer is both the motivating example and a baseline the authors had to build."
+paper_key: paper_doi_10_1109_icse55347_2025_00003
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00013.html b/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00013.html
new file mode 100644
index 0000000..4dce7cb
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00013.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse55347_2025_00013/
+title: "Janus: Detecting Rendering Bugs in Web Browsers via Visual Delta Consistency"
+excerpt: "SQLancer is the database analogue Janus cites for its own oracle design: the paper notes that in database management systems, tools like SQLancer validate query results using predefined equivalence rules, placing it beside formalised layout guidance for web development as prior e."
+paper_key: paper_doi_10_1109_icse55347_2025_00013
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00045.html b/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00045.html
new file mode 100644
index 0000000..f8db077
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00045.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse55347_2025_00045/
+title: "PUPPY: Finding Performance Degradation Bugs in DBMSs via Limited-Optimization Plan Construction"
+excerpt: "SQLancer is one of three widely used tools PUPPY is measured against over 48 hours, finding 6 bugs to PUPPY's 35."
+paper_key: paper_doi_10_1109_icse55347_2025_00045
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00183.html b/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00183.html
new file mode 100644
index 0000000..c6fd6c5
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00183.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse55347_2025_00183/
+title: "ROSA: Finding Backdoors with Fuzzing"
+excerpt: "Two citations, both pointing to SQLancer's oracles as the known successful use of metamorphic oracles in fuzzing to find intricate logic bugs in mature software -- the precedent for ROSA's own oracle."
+paper_key: paper_doi_10_1109_icse55347_2025_00183
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00257.html b/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00257.html
new file mode 100644
index 0000000..9c79bec
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse55347_2025_00257.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse55347_2025_00257/
+title: "Thanos: DBMS Bug Detection via Storage Engine Rotation Based Differential Testing"
+excerpt: "SQLancer is one of Thanos's three baselines and is named a state-of-the-art DBMS testing tool throughout."
+paper_key: paper_doi_10_1109_icse55347_2025_00257
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse_companion58688_2023_00041.html b/_pages/impact/papers/paper_doi_10_1109_icse_companion58688_2023_00041.html
new file mode 100644
index 0000000..d25e151
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse_companion58688_2023_00041.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse_companion58688_2023_00041/
+title: "Randomized Differential Testing of RDF Stores"
+excerpt: "SQLancer's techniques appear as the relational precedent the paper places itself after."
+paper_key: paper_doi_10_1109_icse_companion58688_2023_00041
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icse_seip52600_2021_00042.html b/_pages/impact/papers/paper_doi_10_1109_icse_seip52600_2021_00042.html
new file mode 100644
index 0000000..d1a9337
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icse_seip52600_2021_00042.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icse_seip52600_2021_00042/
+title: "Industry Practice of Coverage-Guided Enterprise-Level DBMS Fuzzing"
+excerpt: "SQLancer is one of the three fuzzers the practice was built around and measured against, described as a black-box logic-bug fuzzer that combines three oracle strategies and credited with finding over 400 bugs in two years."
+paper_key: paper_doi_10_1109_icse_seip52600_2021_00042
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icsip61881_2024_10671554.html b/_pages/impact/papers/paper_doi_10_1109_icsip61881_2024_10671554.html
new file mode 100644
index 0000000..3b56a33
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icsip61881_2024_10671554.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icsip61881_2024_10671554/
+title: "A Review of Fuzz Testing for Configuration-Sensitive Software"
+excerpt: "Database systems appear in the review's opening enumeration of domains fuzzing has been applied to, with network protocols, hypervisors and autonomous vehicles."
+paper_key: paper_doi_10_1109_icsip61881_2024_10671554
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icsme64153_2025_00030.html b/_pages/impact/papers/paper_doi_10_1109_icsme64153_2025_00030.html
new file mode 100644
index 0000000..f1f754e
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icsme64153_2025_00030.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icsme64153_2025_00030/
+title: "TSGuard: Detecting Logic Bugs in Time Series Management Systems Via Time Series Algebra"
+excerpt: "SQLancer is TSGuard's baseline, and the paper is explicit about the form that takes: it adapted SQLancer, an open-source relational database testing tool, as a baseline for comparison."
+paper_key: paper_doi_10_1109_icsme64153_2025_00030
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_icst60714_2024_00012.html b/_pages/impact/papers/paper_doi_10_1109_icst60714_2024_00012.html
new file mode 100644
index 0000000..bd34410
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_icst60714_2024_00012.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_icst60714_2024_00012/
+title: "Differential Optimization Testing of Gremlin-Based Graph Database Systems"
+excerpt: "SQLancer's techniques frame the paper twice over."
+paper_key: paper_doi_10_1109_icst60714_2024_00012
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_ictai62512_2024_00085.html b/_pages/impact/papers/paper_doi_10_1109_ictai62512_2024_00085.html
new file mode 100644
index 0000000..f1f65cf
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_ictai62512_2024_00085.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_ictai62512_2024_00085/
+title: "NNTailor: A Neural Network-Driven Fuzzer for DataBase Management Systems"
+excerpt: "SQLancer is one of NNTailor's two baselines, compared on code coverage and on the syntactic and semantic correctness of generated queries."
+paper_key: paper_doi_10_1109_ictai62512_2024_00085
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_iemcon67450_2025_11381190.html b/_pages/impact/papers/paper_doi_10_1109_iemcon67450_2025_11381190.html
new file mode 100644
index 0000000..1a8e9fe
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_iemcon67450_2025_11381190.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_iemcon67450_2025_11381190/
+title: "A Comparative Survey of Mutation Testing Across Large Language Models, REST APIs, and Database Engines"
+excerpt: "SQLancer appears as one of the representative database engine tools surveyed, and query-plan-guided mutation -- SQLancer's QPG -- is named as a contribution of that domain."
+paper_key: paper_doi_10_1109_iemcon67450_2025_11381190
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_issrew55968_2022_00056.html b/_pages/impact/papers/paper_doi_10_1109_issrew55968_2022_00056.html
new file mode 100644
index 0000000..72e674b
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_issrew55968_2022_00056.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_issrew55968_2022_00056/
+title: "A Disjoint-Partitioning Approach to Enhancing Metamorphic Testing of DBMS"
+excerpt: "The paper is built entirely on SQLancer's work: it devotes a section to query partitioning and ternary logic partitioning, identifies a gap between the two, and introduces disjoint partitioning to close it."
+paper_key: paper_doi_10_1109_issrew55968_2022_00056
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_missf68264_2026_11521893.html b/_pages/impact/papers/paper_doi_10_1109_missf68264_2026_11521893.html
new file mode 100644
index 0000000..84ffa80
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_missf68264_2026_11521893.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_missf68264_2026_11521893/
+title: "An Automated Risk Scoring Framework for SQL Execution Plan Analysis and Performance Regression Detection in Oracle Database Systems"
+excerpt: "The connection is a single background citation in related work, where existing systems are criticised for showing raw metrics without merging them into a risk model."
+paper_key: paper_doi_10_1109_missf68264_2026_11521893
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_punecon67554_2025_11378586.html b/_pages/impact/papers/paper_doi_10_1109_punecon67554_2025_11378586.html
new file mode 100644
index 0000000..504452b
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_punecon67554_2025_11378586.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_punecon67554_2025_11378586/
+title: "AskDB: AI-Assisted Natural Language Interface for Database Querying and Visualization"
+excerpt: "One citation in the literature review, describing Query Plan Guidance as an automated testing technique focused on the diversity of generated queries."
+paper_key: paper_doi_10_1109_punecon67554_2025_11378586
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_saner60148_2024_00096.html b/_pages/impact/papers/paper_doi_10_1109_saner60148_2024_00096.html
new file mode 100644
index 0000000..7521dac
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_saner60148_2024_00096.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_saner60148_2024_00096/
+title: "Testing Constraint Checking Implementations via Principled Metamorphic Transformations"
+excerpt: "No relationship is visible in the paper's own text."
+paper_key: paper_doi_10_1109_saner60148_2024_00096
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_sp54263_2024_00109.html b/_pages/impact/papers/paper_doi_10_1109_sp54263_2024_00109.html
new file mode 100644
index 0000000..656a255
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_sp54263_2024_00109.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_sp54263_2024_00109/
+title: "Chronos: Finding Timeout Bugs in Practical Distributed Systems by Deep-Priority Fuzzing with Transient Delay"
+excerpt: "SQLancer serves as a workload generator, not an oracle: it produces the SQL workload used to exercise MySQL-Cluster while Chronos injects delays around it."
+paper_key: paper_doi_10_1109_sp54263_2024_00109
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_srds69199_2025_00038.html b/_pages/impact/papers/paper_doi_10_1109_srds69199_2025_00038.html
new file mode 100644
index 0000000..fbb03e8
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_srds69199_2025_00038.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_srds69199_2025_00038/
+title: "Diverse Database Replication Based on Snapshot Isolation – Performance Implications of Improved Dependability"
+excerpt: "The single citation supplies terminology: the paper describes faults where a row is wrongly omitted from a result, or where a DELETE, INSERT or UPDATE wrongly changes the database state, and notes these are referred to by some as logic bugs."
+paper_key: paper_doi_10_1109_srds69199_2025_00038
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_tdsc_2024_3521591.html b/_pages/impact/papers/paper_doi_10_1109_tdsc_2024_3521591.html
new file mode 100644
index 0000000..ec22f29
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_tdsc_2024_3521591.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_tdsc_2024_3521591/
+title: "Improving Multitasking DBMS Fuzzing With More Accurate Coverage and Testcase Trimming"
+excerpt: "SQLancer's oracles appear in the comparison of existing DBMS testing approaches, where NoREC, TLP and PQS are grouped with DQE and Pinolo as metamorphic techniques for finding semantic bugs."
+paper_key: paper_doi_10_1109_tdsc_2024_3521591
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_tkde_2026_3656491.html b/_pages/impact/papers/paper_doi_10_1109_tkde_2026_3656491.html
new file mode 100644
index 0000000..ecd61b7
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_tkde_2026_3656491.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_tkde_2026_3656491/
+title: "Effective Bug Detection in Graph Database Engines: An LLM-Based Approach"
+excerpt: "All three of SQLancer's original oracles are described in the survey of prior approaches -- PQS selecting target data and constructing a query to retrieve it, NoREC comparing optimized and non-optimized forms, TLP partitioning on ternary logic -- as the relational precedent for g."
+paper_key: paper_doi_10_1109_tkde_2026_3656491
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_tse_2025_3574328.html b/_pages/impact/papers/paper_doi_10_1109_tse_2025_3574328.html
new file mode 100644
index 0000000..c9e76e6
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_tse_2025_3574328.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_tse_2025_3574328/
+title: "SQLaw: Detecting Bugs in GPU Database Management Systems via Rule-Based Differential Execution"
+excerpt: "SQLaw imitates SQLancer's statement-level, syntax-based reduction to simplify its own bug reports, and takes NoREC and TLP as two of its three baselines, describing them as representative state-of-the-art approaches that have each found over a hundred bugs in non-GPU DBMSs."
+paper_key: paper_doi_10_1109_tse_2025_3574328
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1109_tse_2025_3625300.html b/_pages/impact/papers/paper_doi_10_1109_tse_2025_3625300.html
new file mode 100644
index 0000000..4f86996
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1109_tse_2025_3625300.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1109_tse_2025_3625300/
+title: "A Comprehensive Study of Bugs in Relational DBMS"
+excerpt: "SQLancer is both the study's reference generator and the base of its proof-of-concept tool."
+paper_key: paper_doi_10_1109_tse_2025_3625300
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1117_12_3006402.html b/_pages/impact/papers/paper_doi_10_1117_12_3006402.html
new file mode 100644
index 0000000..5ae7070
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1117_12_3006402.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1117_12_3006402/
+title: "An empirical study on configuration-related branch statement in database management system"
+excerpt: "The single mention is a citation in the opening motivation about the consequences of DBMS vulnerabilities, reachable only through the citation marker."
+paper_key: paper_doi_10_1117_12_3006402
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3372297_3417260.html b/_pages/impact/papers/paper_doi_10_1145_3372297_3417260.html
new file mode 100644
index 0000000..4f59917
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3372297_3417260.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3372297_3417260/
+title: "SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback"
+excerpt: "SQLancer is cited twice as related work on logic-bug detection, with PQS described as constructing queries to fetch a randomly selected row."
+paper_key: paper_doi_10_1145_3372297_3417260
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3395032_3395322.html b/_pages/impact/papers/paper_doi_10_1145_3395032_3395322.html
new file mode 100644
index 0000000..f57f14b
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3395032_3395322.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3395032_3395322/
+title: "Testing query execution engines with mutations"
+excerpt: "PQS is discussed in related work as the closest neighbour: the paper notes it also uses equivalence mutation, arranging for the mutated query to contain a specific row from the original output."
+paper_key: paper_doi_10_1145_3395032_3395322
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3428261.html b/_pages/impact/papers/paper_doi_10_1145_3428261.html
new file mode 100644
index 0000000..2a94416
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3428261.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3428261/
+title: "On the unusual effectiveness of type-aware operator mutations for testing SMT solvers"
+excerpt: "SQLancer appears once, in a list of domains where testing has been applied, cited alongside work on SMT solvers and OS kernels."
+paper_key: paper_doi_10_1145_3428261
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3448016_3457559.html b/_pages/impact/papers/paper_doi_10_1145_3448016_3457559.html
new file mode 100644
index 0000000..f219d9e
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3448016_3457559.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3448016_3457559/
+title: "FoundationDB: A Distributed Unbundled Transactional Key Value Store"
+excerpt: "SQLancer's work is cited once, in related work, as part of the body of approaches for testing the correctness of database subsystems in the absence of faults -- the query engine specifically."
+paper_key: paper_doi_10_1145_3448016_3457559
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3468264_3468540.html b/_pages/impact/papers/paper_doi_10_1145_3468264_3468540.html
new file mode 100644
index 0000000..5e490e1
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3468264_3468540.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3468264_3468540/
+title: "Skeletal approximation enumeration for SMT solver testing"
+excerpt: "Databases appear in a single sentence listing the application domains where the underlying technique has been used, alongside bioinformatics, compilers, debuggers and machine learning systems."
+paper_key: paper_doi_10_1145_3468264_3468540
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3468264_3468573.html b/_pages/impact/papers/paper_doi_10_1145_3468264_3468573.html
new file mode 100644
index 0000000..f8614fd
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3468264_3468573.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3468264_3468573/
+title: "Metamorphic testing of Datalog engines"
+excerpt: "SQLancer is named directly as the relational precedent: the paper notes that Rigger and Su proposed a series of testing techniques implemented in a tool called SQLancer."
+paper_key: paper_doi_10_1145_3468264_3468573
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3471485_3471491.html b/_pages/impact/papers/paper_doi_10_1145_3471485_3471491.html
new file mode 100644
index 0000000..717e4d4
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3471485_3471491.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3471485_3471491/
+title: "Technical Perspective DIAMetrics"
+excerpt: "The only mention is the title of the query partitioning paper appearing in the accompanying reference list rather than in prose."
+paper_key: paper_doi_10_1145_3471485_3471491
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3485529.html b/_pages/impact/papers/paper_doi_10_1145_3485529.html
new file mode 100644
index 0000000..722b3d9
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3485529.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3485529/
+title: "Generative type-aware mutation for testing SMT solvers"
+excerpt: "PQS appears in the evaluation setup's survey of generative fuzzing approaches, cited in the ACM author-year style so SQLancer is not named directly."
+paper_key: paper_doi_10_1145_3485529
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3510003_3510093.html b/_pages/impact/papers/paper_doi_10_1145_3510003_3510093.html
new file mode 100644
index 0000000..64780d1
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3510003_3510093.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3510003_3510093/
+title: "Automatic Detection of Performance Bugs in Database Systems using Equivalent Queries"
+excerpt: "TLP is one of AMOEBA's three baselines and the paper calls it the state-of-the-art tool for finding logic bugs in DBMSs using metamorphic testing."
+paper_key: paper_doi_10_1145_3510003_3510093
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3510457_3513034.html b/_pages/impact/papers/paper_doi_10_1145_3510457_3513034.html
new file mode 100644
index 0000000..96ae6ae
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3510457_3513034.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3510457_3513034/
+title: "An Empirical Study on Quality Issues of eBay's Big Data SQL Analytics Platform"
+excerpt: "One citation, naming all three of SQLancer's original oracles: the paper notes that to detect DBMS bugs, Rigger and Su devised a series of novel approaches including PQS, NoREC and TLP."
+paper_key: paper_doi_10_1145_3510457_3513034
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3531348_3532176.html b/_pages/impact/papers/paper_doi_10_1145_3531348_3532176.html
new file mode 100644
index 0000000..82e4f1f
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3531348_3532176.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3531348_3532176/
+title: "DeepBench: Benchmarking JSON Document Stores"
+excerpt: "A single citation in the conclusion, where SQLancer's line of work is listed as related database testing research."
+paper_key: paper_doi_10_1145_3531348_3532176
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3533767_3534364.html b/_pages/impact/papers/paper_doi_10_1145_3533767_3534364.html
new file mode 100644
index 0000000..f97fa51
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3533767_3534364.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3533767_3534364/
+title: "Unicorn: detect runtime errors in time-series databases with hybrid input synthesis"
+excerpt: "SQLancer is Unicorn's principal baseline."
+paper_key: paper_doi_10_1145_3533767_3534364
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3533767_3534409.html b/_pages/impact/papers/paper_doi_10_1145_3533767_3534409.html
new file mode 100644
index 0000000..5e37ef6
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3533767_3534409.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3533767_3534409/
+title: "Finding bugs in Gremlin-based graph database systems via Randomized differential testing"
+excerpt: "SQLancer's three oracles are the relational precedent Grand carries into graph databases, each described in its related work."
+paper_key: paper_doi_10_1145_3533767_3534409
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3551349_3556924.html b/_pages/impact/papers/paper_doi_10_1145_3551349_3556924.html
new file mode 100644
index 0000000..3360efb
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3551349_3556924.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3551349_3556924/
+title: "Differentially Testing Database Transactions for Fun and Profit"
+excerpt: "DT2 generates its databases and transactions mainly on the basis of SQLancer, revised for the transaction setting, and its artifact carries SQLancer's source under a renamed package."
+paper_key: paper_doi_10_1145_3551349_3556924
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3551349_3560431.html b/_pages/impact/papers/paper_doi_10_1145_3551349_3560431.html
new file mode 100644
index 0000000..f5b68d6
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3551349_3560431.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3551349_3560431/
+title: "Griffin : Grammar-Free DBMS Fuzzing"
+excerpt: "SQLancer is Griffin's baseline and its motivating example of the cost it avoids: the paper counts the lines of code SQLancer needs to support each DBMS -- over 8,000 for PostgreSQL -- as the adaptation burden a grammar-free approach removes, and reports finding 27 more bugs than ."
+paper_key: paper_doi_10_1145_3551349_3560431
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3552326_3587448.html b/_pages/impact/papers/paper_doi_10_1145_3552326_3587448.html
new file mode 100644
index 0000000..9a36ee6
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3552326_3587448.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3552326_3587448/
+title: "Fail through the Cracks: Cross-System Interaction Failures in Modern Cloud Systems"
+excerpt: "One citation, noting that data-plane discrepancies of the kind studied also exist in traditional systems following POSIX or SQL standards, where SQLancer's work is among the references given."
+paper_key: paper_doi_10_1145_3552326_3587448
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3582016_3582053.html b/_pages/impact/papers/paper_doi_10_1145_3582016_3582053.html
new file mode 100644
index 0000000..1cbcaf1
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3582016_3582053.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3582016_3582053/
+title: "Finding Unstable Code via Compiler-Driven Differential Testing"
+excerpt: "SQLancer appears once, as a bibliography entry for non-optimizing reference engine construction in a discussion of differential approaches."
+paper_key: paper_doi_10_1145_3582016_3582053
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3588909.html b/_pages/impact/papers/paper_doi_10_1145_3588909.html
new file mode 100644
index 0000000..52ad1e4
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3588909.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3588909/
+title: "Detecting Logic Bugs of Join Optimizations in DBMS"
+excerpt: "SQLancer is the work TQS is built against and the paper says so directly, calling it the current state-of-the-art tool for finding logic bugs in DBMSs and the most closely related work to its own."
+paper_key: paper_doi_10_1145_3588909
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3597503_3608133.html b/_pages/impact/papers/paper_doi_10_1145_3597503_3608133.html
new file mode 100644
index 0000000..f193aeb
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3597503_3608133.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3597503_3608133/
+title: "EDEFuzz: A Web API Fuzzer for Excessive Data Exposures"
+excerpt: "One citation, in a list of systems where metamorphic relations have been used successfully to find bugs."
+paper_key: paper_doi_10_1145_3597503_3608133
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3597503_3639112.html b/_pages/impact/papers/paper_doi_10_1145_3597503_3639112.html
new file mode 100644
index 0000000..f10b939
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3597503_3639112.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3597503_3639112/
+title: "Mozi: Discovering DBMS Bugs via Configuration-Based Equivalent Transformation"
+excerpt: "SQLancer's three oracles are the point of contrast and the baseline: Mozi argues that TLP applies only to queries using particular clauses and that NoREC's equivalent-query construction is limited by how well SQL semantics can be modelled, then runs PQS, NoREC and TLP against its."
+paper_key: paper_doi_10_1145_3597503_3639112
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3597503_3639200.html b/_pages/impact/papers/paper_doi_10_1145_3597503_3639200.html
new file mode 100644
index 0000000..6b8903a
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3597503_3639200.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3597503_3639200/
+title: "Testing Graph Database Systems via Equivalent Query Rewriting"
+excerpt: "The paper works from Ternary Logic Partitioning throughout, but at one remove: its comparison target is GDBMeter, which carries TLP to graph databases, and the paper is precise that GDBMeter is based on Query Partitioning and reuses a test oracle designed for relational systems."
+paper_key: paper_doi_10_1145_3597503_3639200
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3597503_3639207.html b/_pages/impact/papers/paper_doi_10_1145_3597503_3639207.html
new file mode 100644
index 0000000..2bc5e8d
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3597503_3639207.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3597503_3639207/
+title: "Understanding Transaction Bugs in Database Systems"
+excerpt: "SQLancer is cited as the work on logic bugs in single SELECT statements, which the authors distinguish from the transaction bugs they collect: the boundary that defines their dataset."
+paper_key: paper_doi_10_1145_3597503_3639207
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3597503_3639210.html b/_pages/impact/papers/paper_doi_10_1145_3597503_3639210.html
new file mode 100644
index 0000000..9297555
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3597503_3639210.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3597503_3639210/
+title: "Sedar: Obtaining High-Quality Seeds for DBMS Fuzzing via Cross-DBMS SQL Transfer"
+excerpt: "Three citations in related work describing SQLancer's oracles -- NoREC, PQS and query plan guidance -- as the generation-based approaches aimed at specific bug types, in contrast to Sedar's coverage-guided seed transfer."
+paper_key: paper_doi_10_1145_3597503_3639210
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3597503_3639212.html b/_pages/impact/papers/paper_doi_10_1145_3597503_3639212.html
new file mode 100644
index 0000000..1203428
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3597503_3639212.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3597503_3639212/
+title: "Combining Structured Static Code Information and Dynamic Symbolic Traces for Software Vulnerability Prediction"
+excerpt: "One citation, noting that the CVEs used for evaluation were also used by prior work, SQLancer's among the references."
+paper_key: paper_doi_10_1145_3597503_3639212
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3597926_3598046.html b/_pages/impact/papers/paper_doi_10_1145_3597926_3598046.html
new file mode 100644
index 0000000..b7a79cb
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3597926_3598046.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3597926_3598046/
+title: "GDsmith: Detecting Bugs in Cypher Graph Database Engines"
+excerpt: "GDsmith says plainly that its framework is derived from SQLancer, carrying that tool's approach from relational engines to Cypher, and its artifact holds SQLancer's source under a renamed package."
+paper_key: paper_doi_10_1145_3597926_3598046
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3597926_3598052.html b/_pages/impact/papers/paper_doi_10_1145_3597926_3598052.html
new file mode 100644
index 0000000..75b6afa
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3597926_3598052.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3597926_3598052/
+title: "Dependency-Aware Metamorphic Testing of Datalog Engines"
+excerpt: "The single mention places the work in context: metamorphic testing has been used successfully across a variety of software, including other query-based systems, which is where SQLancer's techniques are cited."
+paper_key: paper_doi_10_1145_3597926_3598052
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3597926_3598068.html b/_pages/impact/papers/paper_doi_10_1145_3597926_3598068.html
new file mode 100644
index 0000000..44b30ee
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3597926_3598068.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3597926_3598068/
+title: "Exploring Missed Optimizations in WebAssembly Optimizers"
+excerpt: "The citations that shape the paper are to work by a SQLancer author on C compiler optimisation testing, which DITWO says inspired its differential setting."
+paper_key: paper_doi_10_1145_3597926_3598068
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3597926_3598130.html b/_pages/impact/papers/paper_doi_10_1145_3597926_3598130.html
new file mode 100644
index 0000000..a0daa2b
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3597926_3598130.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3597926_3598130/
+title: "GrayC: Greybox Fuzzing of Compilers and Analysers for C"
+excerpt: "Two citations placing SQL among the languages, alongside OpenCL, OpenGL and Verilog, where randomised compiler-style testing has been applied."
+paper_key: paper_doi_10_1145_3597926_3598130
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3605157_3605177.html b/_pages/impact/papers/paper_doi_10_1145_3605157_3605177.html
new file mode 100644
index 0000000..9b94e9b
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3605157_3605177.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3605157_3605177/
+title: "Beyond the Coverage Plateau: A Comprehensive Study of Fuzz Blockers (Registered Report)"
+excerpt: "Two citations, placing DBMS fuzzing among the challenging targets researchers have extended fuzzing to."
+paper_key: paper_doi_10_1145_3605157_3605177
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3611643_3613893.html b/_pages/impact/papers/paper_doi_10_1145_3611643_3613893.html
new file mode 100644
index 0000000..b8a73a5
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3611643_3613893.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3611643_3613893/
+title: "Adapting Performance Analytic Techniques in a Real-World Database-Centric System: An Industrial Experience Report"
+excerpt: "A single citation naming SQLancer's three oracles -- query partitioning, pivoted query synthesis and non-optimizing reference engine construction -- as analysis techniques Rigger and Su apply to databases."
+paper_key: paper_doi_10_1145_3611643_3613893
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3611643_3616286.html b/_pages/impact/papers/paper_doi_10_1145_3611643_3616286.html
new file mode 100644
index 0000000..eb84e03
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3611643_3616286.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3611643_3616286/
+title: "Property-Based Fuzzing for Finding Data Manipulation Errors in Android Apps"
+excerpt: "One citation in related work, noting that SQLancer's line of work uses metamorphic testing to find CRUD errors in database management systems -- the same class of error, in engines rather than apps."
+paper_key: paper_doi_10_1145_3611643_3616286
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3622819.html b/_pages/impact/papers/paper_doi_10_1145_3622819.html
new file mode 100644
index 0000000..7ea4653
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3622819.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3622819/
+title: "Towards Better Semantics Exploration for Browser Fuzzing"
+excerpt: "One author-year citation, in a list of domains where grammar-based fuzzing has been applied: compilers, databases, and deep learning frameworks."
+paper_key: paper_doi_10_1145_3622819
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3643779.html b/_pages/impact/papers/paper_doi_10_1145_3643779.html
new file mode 100644
index 0000000..2eeca2a
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3643779.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3643779/
+title: "DTD: Comprehensive and Scalable Testing for Debuggers"
+excerpt: "One author-year citation, among the differential-testing work the paper places itself alongside."
+paper_key: paper_doi_10_1145_3643779
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3643781.html b/_pages/impact/papers/paper_doi_10_1145_3643781.html
new file mode 100644
index 0000000..a46a5c5
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3643781.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3643781/
+title: "Metamorphic Testing of Secure Multi-party Computation (MPC) Compilers"
+excerpt: "One citation, in a list of the software systems metamorphic testing has been applied to: databases, AI models and cyber-physical systems."
+paper_key: paper_doi_10_1145_3643781
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3650212_3680311.html b/_pages/impact/papers/paper_doi_10_1145_3650212_3680311.html
new file mode 100644
index 0000000..a937c26
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3650212_3680311.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3650212_3680311/
+title: "Testing Graph Database Systems with Graph-State Persistence Oracle"
+excerpt: "TLP reaches this paper as the technique GDBMeter carried into graph databases -- the paper describes GDBMeter as the first metamorphic approach to apply TLP from relational DBMSs to graph DBMSs, finding 40 previously unknown bugs, and notes elsewhere that TLP was first proposed f."
+paper_key: paper_doi_10_1145_3650212_3680311
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3650212_3680317.html b/_pages/impact/papers/paper_doi_10_1145_3650212_3680317.html
new file mode 100644
index 0000000..40d4391
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3650212_3680317.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3650212_3680317/
+title: "SQLess: Dialect-Agnostic SQL Query Simplification"
+excerpt: "SQLancer is the example of the problem SQLess addresses: it is named among the tools that integrate query simplification into their workflow only simplistically, deleting whole clauses, and its equivalent-query construction is what produces the long queries that then need reducin."
+paper_key: paper_doi_10_1145_3650212_3680317
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3650212_3680318.html b/_pages/impact/papers/paper_doi_10_1145_3650212_3680318.html
new file mode 100644
index 0000000..ec32d31
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3650212_3680318.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3650212_3680318/
+title: "DBStorm: Generating Various Effective Workloads for Testing Isolation Levels"
+excerpt: "SQLancer supplies DBStorm's SQL."
+paper_key: paper_doi_10_1145_3650212_3680318
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3650212_3680392.html b/_pages/impact/papers/paper_doi_10_1145_3650212_3680392.html
new file mode 100644
index 0000000..d37d52e
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3650212_3680392.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3650212_3680392/
+title: "Testing Gremlin-Based Graph Database Systems via Query Disassembling"
+excerpt: "Query partitioning is the technique QuDi positions itself against, reached through GDBMeter, which the paper describes as adopting TLP to derive a graph query into three disjoint sub-queries and so focusing on predicate-related bugs."
+paper_key: paper_doi_10_1145_3650212_3680392
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3689031_3696064.html b/_pages/impact/papers/paper_doi_10_1145_3689031_3696064.html
new file mode 100644
index 0000000..fe32f0a
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3689031_3696064.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3689031_3696064/
+title: "Understanding and Detecting SQL Function Bugs: Using Simple Boundary Arguments to Trigger Hundreds of DBMS Bugs"
+excerpt: "SQLancer is one of three baselines, described as a state-of-the-art DBMS testing tool widely used in industry, and run in PQS mode with default settings."
+paper_key: paper_doi_10_1145_3689031_3696064
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3689491_3691821.html b/_pages/impact/papers/paper_doi_10_1145_3689491_3691821.html
new file mode 100644
index 0000000..dc9c6a4
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3689491_3691821.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3689491_3691821/
+title: "Step-wise Execution of Data-Centric Systems"
+excerpt: "NoREC and TLP are named as state-of-the-art approaches for detecting logic bugs in DBMSs, the line of work the step-wise reference construction is set against."
+paper_key: paper_doi_10_1145_3689491_3691821
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3689757.html b/_pages/impact/papers/paper_doi_10_1145_3689757.html
new file mode 100644
index 0000000..14953c2
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3689757.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3689757/
+title: "PolyJuice: Detecting Mis-compilation Bugs in Tensor Compilers with Equality Saturation Based Rewriting"
+excerpt: "One citation in related work describing PQS as generating queries designed to retrieve a specific data entity for verification -- the database counterpart to PolyJuice's equivalent-graph oracle."
+paper_key: paper_doi_10_1145_3689757
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3690631.html b/_pages/impact/papers/paper_doi_10_1145_3690631.html
new file mode 100644
index 0000000..e16c5c1
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3690631.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3690631/
+title: "T-Rec: Fine-Grained Language-Agnostic Program Reduction Guided by Lexical Syntax"
+excerpt: "A single citation, among the compiler-testing and programming-language work that commonly employs program reducers."
+paper_key: paper_doi_10_1145_3690631
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3704870.html b/_pages/impact/papers/paper_doi_10_1145_3704870.html
new file mode 100644
index 0000000..6cdf286
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3704870.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3704870/
+title: "The Decision Problem for Regular First Order Theories"
+excerpt: "No SQLancer mention could be extracted, and no bibliography parsed."
+paper_key: paper_doi_10_1145_3704870
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3708533.html b/_pages/impact/papers/paper_doi_10_1145_3708533.html
new file mode 100644
index 0000000..22555e1
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3708533.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3708533/
+title: "Software Security Analysis in 2030 and Beyond: A Research Roadmap"
+excerpt: "Two citations, both to a paper by a SQLancer author rather than to SQLancer itself, placing DBMSs among the areas where the technique has been applied successfully and noting metamorphic transformations among the approaches discussed."
+paper_key: paper_doi_10_1145_3708533
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3720504.html b/_pages/impact/papers/paper_doi_10_1145_3720504.html
new file mode 100644
index 0000000..d99835e
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3720504.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3720504/
+title: "Checking Observational Correctness of Database Systems"
+excerpt: "SQLancer supplies the evaluation's yardstick: Troubadour is shown to verify that a trace exhibits none of the classes of bug SQLancer detects for the SQL fragment it supports, and was run on the queries and incorrect responses SQLancer had observed, reporting errors for all of th."
+paper_key: paper_doi_10_1145_3720504
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3728908.html b/_pages/impact/papers/paper_doi_10_1145_3728908.html
new file mode 100644
index 0000000..ebb8218
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3728908.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3728908/
+title: "QTRAN: Extending Metamorphic-Oracle Based Logical Bug Detection Techniques for Multiple-DBMS Dialect Support"
+excerpt: "SQLancer's oracles are precisely what QTRAN extends: NoREC and TLP are two of the four state-of-the-art techniques it takes and carries to eight DBMSs, and the paper tabulates which systems each currently supports to show the gap it closes."
+paper_key: paper_doi_10_1145_3728908
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3728953.html b/_pages/impact/papers/paper_doi_10_1145_3728953.html
new file mode 100644
index 0000000..d4f3351
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3728953.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3728953/
+title: "Detecting Isolation Anomalies in Relational DBMSs"
+excerpt: "SQLancer appears only in the future-work discussion, as background on what else DBMS testing covers: the paper notes that SQLancer and its approaches -- PQS, TLP and NoREC -- detect logic bugs in SELECT statements, that QPG uses query plans to guide testing, and that QuDi, GQT an."
+paper_key: paper_doi_10_1145_3728953
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3728965.html b/_pages/impact/papers/paper_doi_10_1145_3728965.html
new file mode 100644
index 0000000..40913ee
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3728965.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3728965/
+title: "DepState: Detecting Synchronization Failure Bugs in Distributed Database Management Systems"
+excerpt: "SQLancer is one of five baselines and is called a state-of-the-art DBMS testing technique."
+paper_key: paper_doi_10_1145_3728965
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3728973.html b/_pages/impact/papers/paper_doi_10_1145_3728973.html
new file mode 100644
index 0000000..974cc07
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3728973.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3728973/
+title: "Hulk: Exploring Data-Sensitive Performance Anomalies in DBMSs via Data-Driven Analysis"
+excerpt: "CERT is the closest prior work and the baseline, and the paper notes it is implemented in SQLancer, referring to it throughout as SQLancer-CERT -- one of only two open-source tools available for the comparison."
+paper_key: paper_doi_10_1145_3728973
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3729175.html b/_pages/impact/papers/paper_doi_10_1145_3729175.html
new file mode 100644
index 0000000..3450692
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3729175.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3729175/
+title: "Systems Correctness Practices at Amazon Web Services"
+excerpt: "The extracted page carries a bibliography entry for pivoted query synthesis."
+paper_key: paper_doi_10_1145_3729175
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3729319.html b/_pages/impact/papers/paper_doi_10_1145_3729319.html
new file mode 100644
index 0000000..ce4413a
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3729319.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3729319/
+title: "Graphiti: Bridging Graph and Relational Database Queries"
+excerpt: "One citation in related work, listing metamorphic testing for DBMS bug detection alongside differential testing, with work by the SQLancer authors among the references."
+paper_key: paper_doi_10_1145_3729319
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3731569_3764841.html b/_pages/impact/papers/paper_doi_10_1145_3731569_3764841.html
new file mode 100644
index 0000000..7f35b34
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3731569_3764841.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3731569_3764841/
+title: "Fawkes: Finding Data Durability Bugs in DBMSs via Recovered Data State Verification"
+excerpt: "SQLancer is cited as the established means of detecting SQL correctness bugs through metamorphic testing -- the class of testing Fawkes distinguishes itself from, since durability bugs survive any query-level oracle."
+paper_key: paper_doi_10_1145_3731569_3764841
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3744916_3773102.html b/_pages/impact/papers/paper_doi_10_1145_3744916_3773102.html
new file mode 100644
index 0000000..f52a211
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3744916_3773102.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3744916_3773102/
+title: "Locus: Agentic Predicate Synthesis for Directed Fuzzing"
+excerpt: "One citation, in a discussion of feedback signals being too sparse or indirect to measure progress."
+paper_key: paper_doi_10_1145_3744916_3773102
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3749186.html b/_pages/impact/papers/paper_doi_10_1145_3749186.html
new file mode 100644
index 0000000..c4b7682
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3749186.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3749186/
+title: "Testing Graph Databases with Synthesized Queries"
+excerpt: "TLP reaches this paper through GDBMeter, which the authors describe as adopting the metamorphic oracles from TLP, originally designed for SQL, and partitioning queries using three-valued logic."
+paper_key: paper_doi_10_1145_3749186
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3757347_3759132.html b/_pages/impact/papers/paper_doi_10_1145_3757347_3759132.html
new file mode 100644
index 0000000..0a1d822
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3757347_3759132.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3757347_3759132/
+title: "PANGOLIN: a Comprehensive Testing Framework for Configuration-Rich Key-Value Stores"
+excerpt: "One citation, noting that black-box testing and fuzzing have been used to find bugs in file systems and DBMSs, SQLancer's work among the references."
+paper_key: paper_doi_10_1145_3757347_3759132
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3758316_3763249.html b/_pages/impact/papers/paper_doi_10_1145_3758316_3763249.html
new file mode 100644
index 0000000..fb92877
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3758316_3763249.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3758316_3763249/
+title: "LLM-Assisted Dialect-Agnostic SQL Query Parsing"
+excerpt: "One citation, listing DBMS testing among the tasks that rely on query analysis, with SQLancer's work among the references."
+paper_key: paper_doi_10_1145_3758316_3763249
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3764583.html b/_pages/impact/papers/paper_doi_10_1145_3764583.html
new file mode 100644
index 0000000..c6edcc7
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3764583.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3764583/
+title: "Unveiling Logic Bugs in SPJG Query Optimizations within DBMS"
+excerpt: "SQLancer is the paper's reference point and its baseline: four of SQLancer's methods -- PQS, TLP, NoREC and DQP -- are the four baselines TQS is measured against, and the paper calls SQLancer the state-of-the-art approach for detecting logic bugs while arguing it is not designed ."
+paper_key: paper_doi_10_1145_3764583
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3769779.html b/_pages/impact/papers/paper_doi_10_1145_3769779.html
new file mode 100644
index 0000000..1d38088
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3769779.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3769779/
+title: "Detecting Logic Bugs in DBMSs via Equivalent Data Construction"
+excerpt: "TLP is one of EDC's three baselines, named as a state-of-the-art open-source tool for finding logic bugs alongside EET and Radar."
+paper_key: paper_doi_10_1145_3769779
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3769828.html b/_pages/impact/papers/paper_doi_10_1145_3769828.html
new file mode 100644
index 0000000..c5d4eba
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3769828.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3769828/
+title: "SRS: Detecting Logic Bugs of Join Implementation in DBMSs via Set Relation Synthesis"
+excerpt: "SRS is implemented on SQLancer, generating its database state with SQLancer's syntax-rule-based random generation and keeping SQLancer's query generation strategy for features outside the join."
+paper_key: paper_doi_10_1145_3769828
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3769832.html b/_pages/impact/papers/paper_doi_10_1145_3769832.html
new file mode 100644
index 0000000..2ecfb5f
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3769832.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3769832/
+title: "Test Data Generation for Complex SQL Queries"
+excerpt: "A single citation in the conclusion, placing SQLancer's work among related database testing research."
+paper_key: paper_doi_10_1145_3769832
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3779212_3790244.html b/_pages/impact/papers/paper_doi_10_1145_3779212_3790244.html
new file mode 100644
index 0000000..c7a8763
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3779212_3790244.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3779212_3790244/
+title: "Understanding Query Optimization Bugs in Graph Database Systems"
+excerpt: "The comparison here is against the graph-database descendants of SQLancer's techniques rather than SQLancer itself: GDsmith, GAMERA, GRev, GraspDB, GraphGenie and Gslicer are the seven state-of-the-art techniques the tool is measured against, several of them built on metamorphic ."
+paper_key: paper_doi_10_1145_3779212_3790244
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3785021_3787993.html b/_pages/impact/papers/paper_doi_10_1145_3785021_3787993.html
new file mode 100644
index 0000000..8da253c
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3785021_3787993.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3785021_3787993/
+title: "Reproducibility Report for ACM SIGMOD 2025 Paper: 'Constant Optimization Driven Database System Testing'"
+excerpt: "This paper is a reproduction of SQLancer work rather than a citation of it."
+paper_key: paper_doi_10_1145_3785021_3787993
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3786673.html b/_pages/impact/papers/paper_doi_10_1145_3786673.html
new file mode 100644
index 0000000..095a940
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3786673.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3786673/
+title: "One DBMS, Two Modes, and a Bunch of Bugs: Catching Logic Bugs in Distributed DBMSs via Differential Testing"
+excerpt: "QPG is DistSQL's closest comparison and its point of departure."
+paper_key: paper_doi_10_1145_3786673
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3786699.html b/_pages/impact/papers/paper_doi_10_1145_3786699.html
new file mode 100644
index 0000000..dcc6673
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3786699.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3786699/
+title: "SQLBarber: A System Leveraging Large Language Models to Generate Customized and Realistic SQL Workloads"
+excerpt: "SQLancer is cited as prior work on generating diverse queries for testing and debugging, with its query plan guidance described as feedback-guided generation aiming at diverse execution plans."
+paper_key: paper_doi_10_1145_3786699
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3798226.html b/_pages/impact/papers/paper_doi_10_1145_3798226.html
new file mode 100644
index 0000000..ca4bad4
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3798226.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3798226/
+title: "Metamorphic Testing for Infrastructure-as-Code Engines"
+excerpt: "One citation in the formalism section, noting that Rigger and Su applied metamorphic testing to database management systems and uncovered numerous bugs in popular systems -- the precedent EMIaC carries into a different class of engine."
+paper_key: paper_doi_10_1145_3798226
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3798232.html b/_pages/impact/papers/paper_doi_10_1145_3798232.html
new file mode 100644
index 0000000..6130ecd
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3798232.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3798232/
+title: "Beacon: Detecting Broken Access Control Vulnerabilities in DBMSs via System Catalog Consistency Validation"
+excerpt: "Cited in the design discussion as the logic-bug testing line of work, with NoREC and TLP described by what each compares -- the correctness dimension Beacon sets beside its own access-control one."
+paper_key: paper_doi_10_1145_3798232
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3798245.html b/_pages/impact/papers/paper_doi_10_1145_3798245.html
new file mode 100644
index 0000000..0d19b13
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3798245.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3798245/
+title: "Understanding and Finding JIT Compiler Performance Bugs"
+excerpt: "The citations here are to papers by SQLancer's authors rather than to SQLancer itself, listing database engines among the specialised systems that domain-specific tools target."
+paper_key: paper_doi_10_1145_3798245
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3799227.html b/_pages/impact/papers/paper_doi_10_1145_3799227.html
new file mode 100644
index 0000000..442c3e7
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3799227.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3799227/
+title: "A Comprehensive Survey on Database Management System Fuzzing: Techniques, Taxonomy and Evaluation"
+excerpt: "SQLancer's oracles are among the principal subjects of the survey: PQS, NoREC, TLP and QPG each get their own pipeline description, and the experimental section runs them against one another, reporting for instance that QPG detects bugs more efficiently than TLP within 240 minute."
+paper_key: paper_doi_10_1145_3799227
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3802034.html b/_pages/impact/papers/paper_doi_10_1145_3802034.html
new file mode 100644
index 0000000..d036fe4
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3802034.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3802034/
+title: "DBugScribe: Automatic Database Bug Reproduction from Community Reports"
+excerpt: "SQLancer appears in two roles, neither of them reuse."
+paper_key: paper_doi_10_1145_3802034
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3802053.html b/_pages/impact/papers/paper_doi_10_1145_3802053.html
new file mode 100644
index 0000000..440d946
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3802053.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3802053/
+title: "EPSC: Testing Database Management Systems via Equivalent Prepared Statement Construction"
+excerpt: "EPSC reuses SQLancer's statement generators to produce the statements it then rewrites into prepared form, and takes four oracles as baselines, two of which -- NoREC and TLP -- are SQLancer's, alongside CODDTest, which also came out of the project."
+paper_key: paper_doi_10_1145_3802053
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3802061.html b/_pages/impact/papers/paper_doi_10_1145_3802061.html
new file mode 100644
index 0000000..de6d40c
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3802061.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3802061/
+title: "Finding Missed Optimizations in DBMSs through Unbalanced Short-circuit Query Construction"
+excerpt: "CERT is SCor's most direct comparison, one of five performance-bug detection tools it is measured against and described as tackling performance bugs through cardinality estimation analysis."
+paper_key: paper_doi_10_1145_3802061
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3803437_3806090.html b/_pages/impact/papers/paper_doi_10_1145_3803437_3806090.html
new file mode 100644
index 0000000..83a5030
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3803437_3806090.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3803437_3806090/
+title: "Why Property-Based Testing is Necessary for Data Intensive Scalable Computing"
+excerpt: "SQLancer is the paper's central example of the approach working: it is described as integrating multiple property oracles and having uncovered many previously unknown DBMS logic bugs, with query partitioning, semantics-preserving rewriting and pivoted query synthesis named as the."
+paper_key: paper_doi_10_1145_3803437_3806090
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3808109.html b/_pages/impact/papers/paper_doi_10_1145_3808109.html
new file mode 100644
index 0000000..f46d120
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3808109.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3808109/
+title: "Eidolon: Perform Noise-Aware Fuzzing on FHE Libraries via Equivalence Expression Transformation"
+excerpt: "One citation, in a survey of where metamorphic testing has been combined with fuzzing."
+paper_key: paper_doi_10_1145_3808109
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3810991_3811632.html b/_pages/impact/papers/paper_doi_10_1145_3810991_3811632.html
new file mode 100644
index 0000000..da5182f
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3810991_3811632.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3810991_3811632/
+title: "DIRT: Database-Integrated Random Testing"
+excerpt: "DIRT takes SQLancer as both its point of departure and its baseline: three of its five oracles are reimplementations of SQLancer's, expressed as generation actions, and it explicitly declines to modify SQLancer itself, arguing that integrating a new DBMS into it is too laborious ."
+paper_key: paper_doi_10_1145_3810991_3811632
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3810991_3811634.html b/_pages/impact/papers/paper_doi_10_1145_3810991_3811634.html
new file mode 100644
index 0000000..53b00d4
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3810991_3811634.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3810991_3811634/
+title: "Cloud Analytics Benchmarking: Where We Are, What Is Missing, and Where We Should Go"
+excerpt: "One citation, listing database testing among the related work."
+paper_key: paper_doi_10_1145_3810991_3811634
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3810991_3811637.html b/_pages/impact/papers/paper_doi_10_1145_3810991_3811637.html
new file mode 100644
index 0000000..23e15cc
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3810991_3811637.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3810991_3811637/
+title: "Boosting DBMS Test Coverage via LLM-Driven SQL Generation"
+excerpt: "SQLancer is the reference point for what coverage-oriented generation is being contrasted with: it is described as generating syntactically valid queries paired with logic-testing oracles such as PQS, NoREC and TLP."
+paper_key: paper_doi_10_1145_3810991_3811637
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1145_3828685.html b/_pages/impact/papers/paper_doi_10_1145_3828685.html
new file mode 100644
index 0000000..4bfe38f
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1145_3828685.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1145_3828685/
+title: "Programmable Property-Based Testing"
+excerpt: "One citation, in a discussion of domain-specific metrics used to produce interesting inputs, alongside SlowFuzz and PerfFuzz."
+paper_key: paper_doi_10_1145_3828685
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_1177_0926227x251370258.html b/_pages/impact/papers/paper_doi_10_1177_0926227x251370258.html
new file mode 100644
index 0000000..a20b964
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_1177_0926227x251370258.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_1177_0926227x251370258/
+title: "SQLaser: Detecting database management system (DBMS) logic bugs with clause-guided fuzzing"
+excerpt: "SQLaser adopts SQLancer's oracles rather than competing with them: NoREC and TLP are described as the oracles it works with, PQS is noted as not yet deployed in it, and its argument is that SQLancer's rule-based generation constrains the code paths reached."
+paper_key: paper_doi_10_1177_0926227x251370258
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_14711_thesis_991012980220103412.html b/_pages/impact/papers/paper_doi_10_14711_thesis_991012980220103412.html
new file mode 100644
index 0000000..88a3110
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_14711_thesis_991012980220103412.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_14711_thesis_991012980220103412/
+title: "Duplicate-sensitivity Guided Transformation Synthesis for DBMS Correctness Bug Detection"
+excerpt: "PQS, NoREC and TLP are the hand-written transformations the paper sets out to synthesise automatically, and it argues each presumes an explicitly equivalent query pair."
+paper_key: paper_doi_10_14711_thesis_991012980220103412
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_14722_ndss_2025_230530.html b/_pages/impact/papers/paper_doi_10_14722_ndss_2025_230530.html
new file mode 100644
index 0000000..7276850
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_14722_ndss_2025_230530.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_14722_ndss_2025_230530/
+title: "MTZK: Testing and Exploring Bugs in Zero-Knowledge (ZK) Compilers"
+excerpt: "Database testing is cited as one of the domains whose testing-based quality assurance MTZK's approach resembles -- finding errors rather than proving their absence -- alongside CPUs and operating systems."
+paper_key: paper_doi_10_14722_ndss_2025_230530
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_14722_ndss_2026_240198.html b/_pages/impact/papers/paper_doi_10_14722_ndss_2026_240198.html
new file mode 100644
index 0000000..89921ee
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_14722_ndss_2026_240198.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_14722_ndss_2026_240198/
+title: "Efficiently Detecting DBMS Bugs through Bottom-up Syntax-based SQL Generation"
+excerpt: "SQLancer is the paper's reference point for the template-based approach it is arguing against, and it is unusually direct about SQLancer's standing: the most advanced tool in the category, and the most popular platform for implementing the latest SQL testing techniques."
+paper_key: paper_doi_10_14722_ndss_2026_240198
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_14778_3712221_3712247.html b/_pages/impact/papers/paper_doi_10_14778_3712221_3712247.html
new file mode 100644
index 0000000..9e8079b
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_14778_3712221_3712247.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_14778_3712221_3712247/
+title: "Semantic Conformance Testing of Relational DBMS"
+excerpt: "SEMCONT is built on SQLancer, whose syntax-guided generation it enhances with coverage guidance and uses as the seed pool for its own query generation."
+paper_key: paper_doi_10_14778_3712221_3712247
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_14778_3749646_3749661.html b/_pages/impact/papers/paper_doi_10_14778_3749646_3749661.html
new file mode 100644
index 0000000..b5d4091
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_14778_3749646_3749661.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_14778_3749646_3749661/
+title: "PBench: Workload Synthesizer with Real Statistics for Cloud Analytics Benchmarking"
+excerpt: "One citation, listing database benchmarking among the uses a synthetic workload serves, with SQLancer's work among the references."
+paper_key: paper_doi_10_14778_3749646_3749661
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_14778_3797919_3797928.html b/_pages/impact/papers/paper_doi_10_14778_3797919_3797928.html
new file mode 100644
index 0000000..12209c0
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_14778_3797919_3797928.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_14778_3797919_3797928/
+title: "Dinkel: State-Aware and Granular Framework for Validating Graph Databases"
+excerpt: "SQLancer's techniques appear as the relational background against which graph testing is measured."
+paper_key: paper_doi_10_14778_3797919_3797928
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_14778_3828612_3828639.html b/_pages/impact/papers/paper_doi_10_14778_3828612_3828639.html
new file mode 100644
index 0000000..d76552f
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_14778_3828612_3828639.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_14778_3828612_3828639/
+title: "ReSequel: Robust LLM-assisted Query Rewriting and Optimization using Templatization and Sampling"
+excerpt: "One citation, noting that prior work on SQL testing compares query results directly, in a discussion of how rewrites are verified."
+paper_key: paper_doi_10_14778_3828612_3828639
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_22399_ijcesen_5462.html b/_pages/impact/papers/paper_doi_10_22399_ijcesen_5462.html
new file mode 100644
index 0000000..5562732
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_22399_ijcesen_5462.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_22399_ijcesen_5462/
+title: "Benchmarking Autonomy: A Taxonomy of Human-in-the-Loop Checkpoints for AI-Generated Transformation Code"
+excerpt: "All three SQLancer oracles are described by name as the database-testing precedent: PQS constructing queries expected to retrieve a selected row, NoREC comparing an optimizable query with a non-optimizable equivalent, TLP checking relations among partitioned results."
+paper_key: paper_doi_10_22399_ijcesen_5462
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_3390_app13042519.html b/_pages/impact/papers/paper_doi_10_3390_app13042519.html
new file mode 100644
index 0000000..2b68ef9
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_3390_app13042519.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_3390_app13042519/
+title: "Squill: Testing DBMS with Correctness Feedback and Accurate Instantiation"
+excerpt: "SQLancer is named as a leading example of generation-based DBMS fuzzing and described as constructing functionally equivalent SQL statements through several patterns and feeding them to the same DBMS."
+paper_key: paper_doi_10_3390_app13042519
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_3390_electronics14193910.html b/_pages/impact/papers/paper_doi_10_3390_electronics14193910.html
new file mode 100644
index 0000000..565bb2b
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_3390_electronics14193910.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_3390_electronics14193910/
+title: "Nabil: A Text-to-SQL Model Based on Brain-Inspired Computing Techniques and Large Language Modeling"
+excerpt: "QPG is described once in related work: the paper notes that query plan guidance was proposed to test database systems fully automatically and was applied to SQLite, TiDB and CockroachDB."
+paper_key: paper_doi_10_3390_electronics14193910
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_53799_zb6b1375.html b/_pages/impact/papers/paper_doi_10_53799_zb6b1375.html
new file mode 100644
index 0000000..695ec9f
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_53799_zb6b1375.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_53799_zb6b1375/
+title: "Optimizing Database System Performance: Design and Query Optimization Strategies"
+excerpt: "The connection is a single citation, to CERT."
+paper_key: paper_doi_10_53799_zb6b1375
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_doi_10_7717_peerj_cs_1592.html b/_pages/impact/papers/paper_doi_10_7717_peerj_cs_1592.html
new file mode 100644
index 0000000..ee130ee
--- /dev/null
+++ b/_pages/impact/papers/paper_doi_10_7717_peerj_cs_1592.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_doi_10_7717_peerj_cs_1592/
+title: "DAFuzz: data-aware fuzzing of in-memory data stores"
+excerpt: "SQLancer appears in the related-work survey in the ACM author-year style, so none of the three mentions names it directly -- they are reachable only through the citation."
+paper_key: paper_doi_10_7717_peerj_cs_1592
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.html b/_pages/impact/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.html
new file mode 100644
index 0000000..664e284
--- /dev/null
+++ b/_pages/impact/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_s2_295c629691d5654def2d9d85f72c756c67c68583/
+title: "WingFuzz: Implementing Continuous Fuzzing for DBMSs"
+excerpt: "SQLancer is one of three state-of-the-art fuzzers WingFuzz is measured against, with branch counts and bug counts reported per DBMS -- 211,620 more branches and 25 more unique bugs."
+paper_key: paper_s2_295c629691d5654def2d9d85f72c756c67c68583
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.html b/_pages/impact/papers/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.html
new file mode 100644
index 0000000..84f7a4a
--- /dev/null
+++ b/_pages/impact/papers/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb/
+title: "DynSQL: Stateful Fuzzing for Database Management Systems with Complex and Valid SQL Query Generation"
+excerpt: "SQLancer appears as related work rather than as a baseline, and the paper is explicit about why: it states that DynSQL and SQLancer are designed for different research problems and that no comparison experiment against SQLancer was run."
+paper_key: paper_s2_3a7373bd891702ae93d7e058241a7e8be25e89eb
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.html b/_pages/impact/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.html
new file mode 100644
index 0000000..f9ebb1d
--- /dev/null
+++ b/_pages/impact/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42/
+title: "Detecting Logical Bugs of DBMS with Coverage-based Guidance"
+excerpt: "SQLRight is built on SQLancer and Squirrel, and two of its four oracles -- NoREC and TLP -- are ported from SQLancer directly."
+paper_key: paper_s2_77faa12e3aa52f3cb41009e9e8f65c79c9ef9f42
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.html b/_pages/impact/papers/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.html
new file mode 100644
index 0000000..1f49b9d
--- /dev/null
+++ b/_pages/impact/papers/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268/
+title: "Demystifying and Checking Silent Semantic Violations in Large Distributed Systems"
+excerpt: "One citation, in a survey of solutions for detecting semantic bugs in file systems and DBMSs, where pivoted query synthesis is named alongside cross-checking file system implementations and fuzzing."
+paper_key: paper_s2_84c637ad3297e0d4e45a4f5245d0472a82a59268
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.html b/_pages/impact/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.html
new file mode 100644
index 0000000..94d618f
--- /dev/null
+++ b/_pages/impact/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b/
+title: "Pinolo: Detecting Logical Bugs in Database Management Systems with Approximate Query Synthesis"
+excerpt: "Pinolo positions itself directly against SQLancer's oracles: PQS is the oracle-guided synthesis it says considers only one row at a time, NoREC and TLP are the metamorphic approaches it says preserve too much of the query, and it states that metamorphic testing of that kind is re."
+paper_key: paper_s2_c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.html b/_pages/impact/papers/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.html
new file mode 100644
index 0000000..e84c954
--- /dev/null
+++ b/_pages/impact/papers/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf/
+title: "TCP-Fuzz: Detecting Memory and Semantic Bugs in TCP Stacks with Fuzzing"
+excerpt: "SQLancer is cited once for its practice rather than its technique: when an inconsistency is identified as a semantic bug, the authors follow SQLancer and libFuzzer in fixing it manually so related inconsistencies stop recurring during the campaign."
+paper_key: paper_s2_d88db0b52f31ed5444602a67e9a275faf57e15bf
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.html b/_pages/impact/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.html
new file mode 100644
index 0000000..2452874
--- /dev/null
+++ b/_pages/impact/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/papers/paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec/
+title: "Towards Generic Database Management System Fuzzing"
+excerpt: "SQLancer is one of BuzzBee's baselines for the relational case, run with its PQS oracle, and also on ArangoDB because SQLancer had recently added support for it."
+paper_key: paper_s2_d933042eb3c2a8f2e208515490d6b6a400e00fec
+author_profile: false
+sitemap: true
+---
+
+{% include impact/paper-detail.html key=page.paper_key %}
diff --git a/_pages/impact/recognition.html b/_pages/impact/recognition.html
new file mode 100644
index 0000000..4d51b36
--- /dev/null
+++ b/_pages/impact/recognition.html
@@ -0,0 +1,58 @@
+---
+permalink: /impact/recognition/
+title: "Papers calling SQLancer state of the art"
+excerpt: "Every paper that describes SQLancer or one of its techniques as the state of the art, with the sentence it says it in."
+author_profile: false
+toc: true
+toc_label: "On this page"
+toc_sticky: true
+---
+
+{%- comment -%}
+The full list. The impact page quotes the papers whose recognition stands on
+its own; most of the rest say it while naming their comparison set, which the
+comparison section there already shows. Neither is more true than the other,
+so the whole list lives here rather than being cut.
+
+Which papers are quoted on the impact page is recorded in
+_data/impact/recognition_highlights.json, with the reason for each.
+{%- endcomment -%}
+{%- assign impact = site.data.impact -%}
+{%- assign stats = impact.stats -%}
+{%- assign external_papers = impact.papers.papers | where_exp: "p", "p.is_sqlancer_publication != true" -%}
+{%- assign sota_papers = "" | split: "" -%}
+{%- for paper in external_papers -%}
+  {%- if paper.relationships.describes_as_state_of_the_art.value == "yes" -%}
+    {%- assign sota_papers = sota_papers | push: paper -%}
+  {%- endif -%}
+{%- endfor -%}
+
+<div class="impact" markdown="0">
+
+<p class="impact-breadcrumb">
+  <a href="{{ '/impact/#papers' | relative_url }}">&larr; Research building on SQLancer</a>
+</p>
+
+<p>
+  Recognition rather than reuse: these papers describe SQLancer or one of its
+  techniques as the state of the art, the most effective, or the approach that
+  established the field. That is a different claim from building on it, so it
+  counts towards no figure on the
+  <a href="{{ '/impact/' | relative_url }}">impact page</a> &mdash; not the
+  {{ stats.headline.papers_reusing_or_extending_sqlancer }} that reuse or
+  extend SQLancer, nor the
+  {{ stats.headline.papers_comparing_against_sqlancer }} that compare against
+  it. Each entry quotes the sentence itself.
+</p>
+
+<h2 id="all">Every paper ({{ sota_papers.size }})</h2>
+
+{% include impact/paper-list.html papers=sota_papers relationship="describes_as_state_of_the_art" %}
+
+<p class="impact-meta">
+  Every quotation is stored verbatim from the paper's extracted text. The
+  <a href="{{ '/impact/#methodology' | relative_url }}">attribution policy</a>
+  explains what has to be true for this classification to be recorded.
+</p>
+
+</div>
diff --git a/_pages/impact/talks/645a47ac1426.html b/_pages/impact/talks/645a47ac1426.html
new file mode 100644
index 0000000..bc19f3b
--- /dev/null
+++ b/_pages/impact/talks/645a47ac1426.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/talks/645a47ac1426/
+title: "Fuzzing: Practical approaches in ClickHouse"
+excerpt: "Alexey Milovidov at C++ Siberia 2021: the 1 point where SQLancer comes up, each linked to the moment it was said."
+talk_id: "talk:645a47ac1426"
+author_profile: false
+sitemap: true
+---
+
+{% include impact/talk-detail.html id=page.talk_id %}
diff --git a/_pages/impact/talks/6YGqFRTe2D0.html b/_pages/impact/talks/6YGqFRTe2D0.html
new file mode 100644
index 0000000..3a335d3
--- /dev/null
+++ b/_pages/impact/talks/6YGqFRTe2D0.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/talks/6YGqFRTe2D0/
+title: "[FUZZING'23] \"Three Colours of Fuzzing: Reflections and Open Challenges\" Keynote by Cristian Cadar"
+excerpt: "Cristian Cadar at FUZZING'23: the 4 points where SQLancer comes up, each linked to the moment it was said."
+talk_id: "talk:youtube:6YGqFRTe2D0"
+author_profile: false
+sitemap: true
+---
+
+{% include impact/talk-detail.html id=page.talk_id %}
diff --git a/_pages/impact/talks/9d70cce25a55.html b/_pages/impact/talks/9d70cce25a55.html
new file mode 100644
index 0000000..ccd007c
--- /dev/null
+++ b/_pages/impact/talks/9d70cce25a55.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/talks/9d70cce25a55/
+title: "ClickHouse Release 22.3 Webinar"
+excerpt: "Alexey Milovidov at ClickHouse 22.3 release webinar: the 1 point where SQLancer comes up, each linked to the moment it was said."
+talk_id: "talk:9d70cce25a55"
+author_profile: false
+sitemap: true
+---
+
+{% include impact/talk-detail.html id=page.talk_id %}
diff --git a/_pages/impact/talks/BgC79Zt2fPs.html b/_pages/impact/talks/BgC79Zt2fPs.html
new file mode 100644
index 0000000..b7fd280
--- /dev/null
+++ b/_pages/impact/talks/BgC79Zt2fPs.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/talks/BgC79Zt2fPs/
+title: "Keynote 1: DuckDB Testing - Present and Future"
+excerpt: "Mark Raasveldt at DBTest 2022: the 6 points where SQLancer comes up, each linked to the moment it was said."
+talk_id: "talk:youtube:BgC79Zt2fPs"
+author_profile: false
+sitemap: true
+---
+
+{% include impact/talk-detail.html id=page.talk_id %}
diff --git a/_pages/impact/talks/CW4Ntdtp7lg.html b/_pages/impact/talks/CW4Ntdtp7lg.html
new file mode 100644
index 0000000..b74cd35
--- /dev/null
+++ b/_pages/impact/talks/CW4Ntdtp7lg.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/talks/CW4Ntdtp7lg/
+title: "Fuzzing databases is difficult"
+excerpt: "Pedro Ferreira at FOSDEM 2025: the 4 points where SQLancer comes up, each linked to the moment it was said."
+talk_id: "talk:youtube:CW4Ntdtp7lg"
+author_profile: false
+sitemap: true
+---
+
+{% include impact/talk-detail.html id=page.talk_id %}
diff --git a/_pages/impact/talks/L90MBb6NLBE.html b/_pages/impact/talks/L90MBb6NLBE.html
new file mode 100644
index 0000000..ba6a205
--- /dev/null
+++ b/_pages/impact/talks/L90MBb6NLBE.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/talks/L90MBb6NLBE/
+title: "FUZZING'25 Keynote: \"Constraining Fuzzing without Paying Too Much\" by Miryung Kim"
+excerpt: "Miryung Kim at FUZZING'25: the 1 point where SQLancer comes up, each linked to the moment it was said."
+talk_id: "talk:youtube:L90MBb6NLBE"
+author_profile: false
+sitemap: true
+---
+
+{% include impact/talk-detail.html id=page.talk_id %}
diff --git a/_pages/impact/talks/QRwxHGpWaUA.html b/_pages/impact/talks/QRwxHGpWaUA.html
new file mode 100644
index 0000000..14a737d
--- /dev/null
+++ b/_pages/impact/talks/QRwxHGpWaUA.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/talks/QRwxHGpWaUA/
+title: "The Art of Database Testing by Alperen Keles | DC Systems 011"
+excerpt: "Alperen Keles at DC Systems 011: the 3 points where SQLancer comes up, each linked to the moment it was said."
+talk_id: "talk:youtube:QRwxHGpWaUA"
+author_profile: false
+sitemap: true
+---
+
+{% include impact/talk-detail.html id=page.talk_id %}
diff --git a/_pages/impact/talks/V_qzqY1bb7I.html b/_pages/impact/talks/V_qzqY1bb7I.html
new file mode 100644
index 0000000..569cfbf
--- /dev/null
+++ b/_pages/impact/talks/V_qzqY1bb7I.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/talks/V_qzqY1bb7I/
+title: "Reliability Lessons From SQLite - Richard Hipp | SSW 2026"
+excerpt: "Richard Hipp at SSW 2026: the 2 points where SQLancer comes up, each linked to the moment it was said."
+talk_id: "talk:youtube:V_qzqY1bb7I"
+author_profile: false
+sitemap: true
+---
+
+{% include impact/talk-detail.html id=page.talk_id %}
diff --git a/_pages/impact/talks/wHo-VtzTHx0.html b/_pages/impact/talks/wHo-VtzTHx0.html
new file mode 100644
index 0000000..fb618dd
--- /dev/null
+++ b/_pages/impact/talks/wHo-VtzTHx0.html
@@ -0,0 +1,10 @@
+---
+permalink: /impact/talks/wHo-VtzTHx0/
+title: "CockroachDB's Query Optimizer (Rebecca Taft, Cockroach Labs)"
+excerpt: "Rebecca Taft at CMU Quarantine Tech Talks 2020: the 1 point where SQLancer comes up, each linked to the moment it was said."
+talk_id: "talk:youtube:wHo-VtzTHx0"
+author_profile: false
+sitemap: true
+---
+
+{% include impact/talk-detail.html id=page.talk_id %}
diff --git a/_pages/splash-page.md b/_pages/splash-page.md
index befb035..fddbe7b 100644
--- a/_pages/splash-page.md
+++ b/_pages/splash-page.md
@@ -1,38 +1,17 @@
----
-layout: splash
-permalink: /
-hidden: true
-header:
-  overlay_color: "#5e616c"
-  overlay_image: /assets/images/mm-home-page-feature.jpg
-  actions:
-    - label: "<i class='fas fa-download'></i> GitHub Repository"
-      url: "/docs/quick-start-guide/"
-excerpt: >
-  SQLancer automatically finds bugs in database systems.<br />
-  <small><a href="https://github.com/sqlancer/sqlancer/releases/tag/v2.0.0">Latest release v2.0.0</a></small>
-feature_row:
-  - image_path: /assets/images/sqlancer_logo_pos.svg
-    alt: "100% free"
-    title: "100% free"
-    excerpt: "Free to use however you want under the MIT License. Clone it, fork it, customize it... whatever!"
-    url: "/docs/license/"
-    btn_class: "btn--primary"
-    btn_label: "Learn more"  
-  - image_path: /assets/images/supported_systems.png
-    alt: "Supported database systems"
-    title: "Over twenty database systems"
-    excerpt: "SQLancer can automatically generate test cases for many of the most popular database systems."
-    url: "/supported-databases"
-    btn_class: "btn--primary"
-    btn_label: "Learn more"
-  - image_path: /assets/images/bugs.png
-    alt: "fully responsive"
-    title: "Hundreds of bugs"
-    excerpt: "SQLancer has founds hundreds of bugs."
-    url: "/bugs/"
-    btn_class: "btn--primary"
-    btn_label: "Learn more"    
----
-
-{% include feature_row %}
+---
+layout: splash
+permalink: /
+hidden: true
+header:
+  overlay_color: "#5e616c"
+  actions:
+    - label: "<i class='fab fa-github'></i> GitHub Repository"
+      url: "https://github.com/sqlancer/sqlancer"
+excerpt: >
+  SQLancer automatically finds bugs in database systems.<br />
+  <small><a href="https://github.com/sqlancer/sqlancer/releases/tag/v2.0.0">Latest release v2.0.0</a></small>
+---
+
+{% include impact/headline.html %}
+
+{% include impact/feature-row.html %}
diff --git a/_pages/supported-databases.md b/_pages/supported-databases.md
index 5868f8f..a33c818 100644
--- a/_pages/supported-databases.md
+++ b/_pages/supported-databases.md
@@ -3,26 +3,22 @@ permalink: /supported-databases/
 title: "Database systems supported by SQLancer"
 ---
 
-SQLancer supports many state-of-the-art database systems. The following implementations are part of the [main SQLancer repository](https://github.com/sqlancer/sqlancer/tree/master/src/sqlancer):
+{%- comment -%}
+The list below is derived from _data/impact/dbms.json, which the impact pipeline
+regenerates from the provider directories in the SQLancer repository. Adding a
+DBMS to SQLancer is enough for it to appear here.
+{%- endcomment -%}
+{%- assign supported = site.data.impact.dbms.dbms | where: "supported_by_sqlancer", true -%}
 
-* ArangoDB
-* Citus
-* ClickHouse
-* CnosDB
-* CockroachDB
-* Cosmos
-* Databend
-* Doris
-* DuckDB
-* H2
-* HSQLDB
-* MariaDB
-* Materialize
-* MongoDB
-* MySQL
-* OceanBase
-* PostgreSQL
-* QuestDB
-* SQLite3
-* TiDB
-* YugabyteDB
\ No newline at end of file
+SQLancer supports many state-of-the-art database systems. The following
+{{ supported.size }} implementations are part of the
+[main SQLancer repository](https://github.com/sqlancer/sqlancer/tree/master/src/sqlancer):
+
+{% for entry in supported -%}
+* {% if entry.url %}[{{ entry.name }}]({{ entry.url }}){% else %}{{ entry.name }}{% endif %}
+{% endfor %}
+
+Supporting a database system is not the same as that project using SQLancer, and
+not the same as SQLancer having found bugs in it. The
+[impact page]({{ '/impact/#database-systems' | relative_url }}) tracks all three
+separately.
diff --git a/assets/css/impact.css b/assets/css/impact.css
new file mode 100644
index 0000000..bd02068
--- /dev/null
+++ b/assets/css/impact.css
@@ -0,0 +1,487 @@
+/* Styling for the /impact/ section and the homepage statistics strip.
+ *
+ * Kept in a standalone stylesheet rather than in the theme's Sass pipeline so
+ * that the remote Minimal Mistakes theme stays untouched and upgradable. Only
+ * `.impact-*` classes are defined here; nothing overrides a theme selector.
+ *
+ * The categorical slots are the validated default data-visualisation palette.
+ * Both modes were checked with the palette validator against the surfaces this
+ * page actually renders on (#ffffff light, #1a1a19 dark): lightness band,
+ * chroma floor, colourblind separation, normal-vision separation and contrast
+ * all pass. Two light-mode slots sit below 3:1 against white, so every chart
+ * ships direct value labels and a table beside it, which is the required
+ * relief.
+ */
+
+.impact {
+  --impact-surface: #ffffff;
+  --impact-surface-raised: #f8f8f6;
+  --impact-ink: #0b0b0b;
+  --impact-ink-secondary: #52514e;
+  --impact-ink-muted: #6d6b66;
+  --impact-grid: #e1e0d9;
+  --impact-axis: #c3c2b7;
+  --impact-border: rgba(11, 11, 11, 0.12);
+
+  --impact-sequential: #2a78d6;
+  --impact-series-1: #2a78d6;
+  --impact-series-2: #eb6834;
+  --impact-series-3: #1baf7a;
+  --impact-series-4: #eda100;
+}
+
+@media (prefers-color-scheme: dark) {
+  :root:not([data-theme="light"]) .impact {
+    --impact-surface: #1a1a19;
+    --impact-surface-raised: #222220;
+    --impact-ink: #ffffff;
+    --impact-ink-secondary: #c3c2b7;
+    --impact-ink-muted: #898781;
+    --impact-grid: #2c2c2a;
+    --impact-axis: #383835;
+    --impact-border: rgba(255, 255, 255, 0.14);
+
+    --impact-sequential: #3987e5;
+    --impact-series-1: #3987e5;
+    --impact-series-2: #d95926;
+    --impact-series-3: #199e70;
+    --impact-series-4: #c98500;
+  }
+}
+
+:root[data-theme="dark"] .impact {
+  --impact-surface: #1a1a19;
+  --impact-surface-raised: #222220;
+  --impact-ink: #ffffff;
+  --impact-ink-secondary: #c3c2b7;
+  --impact-ink-muted: #898781;
+  --impact-grid: #2c2c2a;
+  --impact-axis: #383835;
+  --impact-border: rgba(255, 255, 255, 0.14);
+
+  --impact-sequential: #3987e5;
+  --impact-series-1: #3987e5;
+  --impact-series-2: #d95926;
+  --impact-series-3: #199e70;
+  --impact-series-4: #c98500;
+}
+
+/* ---------------------------------------------------------------- headline */
+
+.impact-headline {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 1em 2.5em;
+  align-items: baseline;
+  justify-content: center;
+  margin: 1.5em 0 1em;
+  padding: 0;
+  list-style: none;
+}
+
+.impact-headline__item {
+  display: flex;
+  flex-direction: column;
+  align-items: center;
+  min-width: 8em;
+}
+
+.impact-headline__value {
+  font-size: 1.9em;
+  font-weight: 700;
+  line-height: 1.1;
+  color: var(--impact-ink);
+}
+
+/* Deliberately not uppercased: "DBMSs supported" becomes "DBMSS SUPPORTED",
+ * which reads as a typo. */
+.impact-headline__label {
+  font-size: 0.85em;
+  letter-spacing: 0.01em;
+  color: var(--impact-ink-secondary);
+}
+
+/* On the splash hero the strip sits on the dark overlay. */
+.page__hero--overlay .impact-headline__value,
+.page__hero--overlay .impact-headline__label {
+  color: #fff;
+}
+
+/* ------------------------------------------------------------- stat tiles */
+
+.impact-stats {
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(11em, 1fr));
+  gap: 1em;
+  margin: 1.5em 0 2em;
+  padding: 0;
+  list-style: none;
+}
+
+.impact-stat {
+  padding: 1em 1.1em;
+  border: 1px solid var(--impact-border);
+  border-radius: 6px;
+  background: var(--impact-surface-raised);
+}
+
+.impact-stat__value {
+  display: block;
+  font-size: 1.8em;
+  font-weight: 700;
+  line-height: 1.15;
+  color: var(--impact-ink);
+}
+
+.impact-stat__label {
+  display: block;
+  margin-top: 0.2em;
+  font-size: 0.82em;
+  color: var(--impact-ink-secondary);
+}
+
+/* ----------------------------------------------------------------- charts */
+
+.impact-figure {
+  margin: 1.5em 0 2em;
+}
+
+.impact-figure__caption {
+  margin: 0 0 0.6em;
+  font-size: 0.85em;
+  color: var(--impact-ink-secondary);
+}
+
+.impact-chart {
+  display: block;
+  overflow: visible;
+  font-family: inherit;
+}
+
+.impact-chart__scroll {
+  overflow-x: auto;
+  padding-bottom: 0.25em;
+}
+
+.impact-chart__label,
+.impact-chart__tick {
+  font-size: 11px;
+  fill: var(--impact-ink-muted);
+}
+
+.impact-chart__value {
+  font-size: 11px;
+  font-weight: 600;
+  font-variant-numeric: tabular-nums;
+  fill: var(--impact-ink-secondary);
+}
+
+.impact-chart__value--tiny {
+  font-size: 9px;
+  font-weight: 500;
+}
+
+.impact-chart__grid {
+  stroke: var(--impact-grid);
+  stroke-width: 1;
+}
+
+.impact-chart__axis {
+  stroke: var(--impact-axis);
+  stroke-width: 1;
+}
+
+/* ----------------------------------------------------------------- legend */
+
+.impact-legend {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 0.35em 1.2em;
+  margin: 0 0 0.6em;
+  padding: 0;
+  list-style: none;
+  font-size: 0.85em;
+  color: var(--impact-ink-secondary);
+}
+
+.impact-legend__item {
+  display: flex;
+  align-items: center;
+  gap: 0.4em;
+}
+
+.impact-legend__swatch {
+  width: 0.75em;
+  height: 0.75em;
+  border-radius: 2px;
+  flex: none;
+}
+
+/* ------------------------------------------------------------ data tables */
+
+.impact-table-wrap {
+  overflow-x: auto;
+}
+
+.impact-table {
+  width: 100%;
+  font-size: 0.82em;
+  border-collapse: collapse;
+}
+
+.impact-table th,
+.impact-table td {
+  padding: 0.35em 0.6em;
+  border-bottom: 1px solid var(--impact-border);
+  text-align: left;
+  vertical-align: top;
+}
+
+.impact-table td.impact-num,
+.impact-table th.impact-num {
+  text-align: right;
+  font-variant-numeric: tabular-nums;
+  white-space: nowrap;
+}
+
+.impact-details {
+  margin: 0.5em 0 2em;
+}
+
+.impact-details > summary {
+  cursor: pointer;
+  font-size: 0.88em;
+  color: var(--impact-ink-secondary);
+}
+
+/* ------------------------------------------------------------------ misc */
+
+.impact-tag {
+  display: inline-block;
+  padding: 0.05em 0.5em;
+  margin-right: 0.25em;
+  border: 1px solid var(--impact-border);
+  border-radius: 999px;
+  font-size: 0.85em;
+  white-space: nowrap;
+  color: var(--impact-ink-secondary);
+}
+
+.impact-evidence {
+  margin: 0.3em 0 0;
+  padding-left: 0.8em;
+  border-left: 3px solid var(--impact-grid);
+  font-size: 0.9em;
+  color: var(--impact-ink-secondary);
+}
+
+/* Evidence links are permalinks with 40-character commit SHAs in them. Without
+ * this they are a single unbreakable word and push the whole page sideways. */
+.impact-evidence a,
+.impact-table a,
+.impact a[href*="://"] {
+  overflow-wrap: anywhere;
+  word-break: break-word;
+}
+
+/* A model-written summary. Deliberately styled unlike .impact-evidence: that
+ * one quotes a source, this one does not, and a reader must be able to tell
+ * which is which at a glance rather than by reading the caption. */
+.impact-summary {
+  margin: 0.4em 0 0.2em;
+  font-size: 0.95em;
+  color: var(--impact-ink-secondary);
+}
+
+/* What the paper is about, secondary to what it did with SQLancer. */
+.impact-summary .impact-about {
+  display: block;
+  margin-top: 0.35em;
+  font-size: 0.92em;
+  color: var(--impact-ink-muted, var(--impact-ink-secondary));
+}
+
+.impact-summary .impact-meta {
+  display: block;
+  margin-top: 0.2em;
+  font-style: italic;
+}
+
+.impact-note {
+  padding: 0.8em 1em;
+  border-left: 3px solid var(--impact-series-1);
+  background: var(--impact-surface-raised);
+  font-size: 0.9em;
+}
+
+.impact-policy__rule {
+  margin: 0.35em 0;
+}
+
+.impact-policy__kind {
+  display: inline-block;
+  min-width: 5.5em;
+  font-size: 0.78em;
+  font-weight: 700;
+  letter-spacing: 0.04em;
+  text-transform: uppercase;
+  color: var(--impact-ink-muted);
+}
+
+.impact-meta {
+  font-size: 0.82em;
+  color: var(--impact-ink-muted);
+}
+
+/* ------------------------------------------------ per-system detail pages */
+
+.impact-breadcrumb {
+  font-size: 0.9em;
+  margin: 0 0 1em;
+}
+
+/* Bars drawn in HTML rather than SVG. The shared page charts are generated
+ * files reviewed in a diff; these are per-system and there are hundreds, so
+ * they are rendered from the data instead. One hue, because the job is
+ * magnitude, and every row carries its number as well as its bar. */
+.impact-bars {
+  margin: 0 0 1.2em;
+}
+
+.impact-bars__row {
+  display: grid;
+  grid-template-columns: minmax(6em, 12em) 1fr;
+  align-items: center;
+  gap: 0 0.7em;
+  margin: 0 0 0.3em;
+}
+
+.impact-bars__label {
+  font-size: 0.85em;
+  color: var(--impact-ink-secondary);
+  overflow-wrap: anywhere;
+}
+
+.impact-bars__value {
+  display: flex;
+  align-items: center;
+  gap: 0.5em;
+  margin: 0;
+  min-width: 0;
+}
+
+.impact-bars__bar {
+  display: block;
+  height: 14px;
+  min-width: 2px;
+  border-radius: 4px;
+  background: var(--impact-sequential);
+}
+
+.impact-bars__count {
+  font-size: 0.8em;
+  font-variant-numeric: tabular-nums;
+  font-weight: 600;
+  color: var(--impact-ink-secondary);
+  white-space: nowrap;
+}
+
+.impact-columns {
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(min(100%, 20em), 1fr));
+  gap: 0 2em;
+}
+
+.impact-columns h3 {
+  margin-top: 0.6em;
+}
+
+.impact-adoption {
+  margin: 0 0 1.4em;
+}
+
+.impact-adoption__heading {
+  margin: 0 0 0.3em;
+  font-size: 1em;
+  text-transform: capitalize;
+}
+
+/* A long bug list is the point of a system's page, but it should not be the
+ * whole viewport. It scrolls in place once it is taller than a screenful. */
+.impact-table--bugs {
+  font-size: 0.9em;
+}
+
+.impact-table--bugs tbody {
+  font-variant-numeric: tabular-nums;
+}
+
+.impact-row--rejected {
+  opacity: 0.62;
+}
+
+.impact-row--rejected th::after {
+  content: " (rejected)";
+  color: var(--impact-ink-muted);
+  font-weight: 400;
+}
+
+/* A chart bar that links somewhere. The hover cue has to be visible against
+ * both the bar fill and the page surface, so it is an outline rather than a
+ * colour change, which would collide with the categorical palette. */
+.impact-chart a {
+  cursor: pointer;
+}
+
+.impact-chart a:hover .impact-chart__bar,
+.impact-chart a:focus-visible .impact-chart__bar {
+  stroke: var(--impact-ink);
+  stroke-width: 1.5;
+}
+
+.impact-chart a:hover .impact-chart__label,
+.impact-chart a:focus-visible .impact-chart__label {
+  fill: var(--impact-ink);
+  text-decoration: underline;
+}
+
+/* A frame captured from a talk. It sits between the mention's timestamp and
+ * the note explaining it, so it is bounded rather than full-bleed: a slide is
+ * evidence here, not an illustration, and at page width it would read as one. */
+.impact-frame {
+  margin: 0.5em 0 0.75em;
+  max-width: 34em;
+}
+
+.impact-frame img {
+  display: block;
+  width: 100%;
+  height: auto;
+  border: 1px solid var(--impact-rule, #d8d8d8);
+  border-radius: 3px;
+}
+
+.impact-frame figcaption {
+  margin-top: 0.35em;
+}
+
+/* A slide from someone's talk, shown with the talk it came from. The image
+ * carries it, so the caption stays quiet underneath; and a slide is never
+ * narrowed past the point where its own text can be read -- which is the only
+ * reason to show a slide rather than describe it. */
+.impact-slide {
+  margin: 0.75em 0 1.25em;
+  max-width: 32em;
+}
+
+.impact-slide img {
+  display: block;
+  width: 100%;
+  height: auto;
+  border: 1px solid var(--impact-rule, #d8d8d8);
+  border-radius: 3px;
+}
+
+.impact-slide figcaption {
+  margin-top: 0.4em;
+}
diff --git a/assets/images/impact/talks/6YGqFRTe2D0-1509.jpg b/assets/images/impact/talks/6YGqFRTe2D0-1509.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..9962491a865adbef0a51f2c49f79392a62509795
GIT binary patch
literal 73144
zcmc$_bx>Sivo5@c8DJo|1h>K6g4+N?aDqc{PeO3l;4rv539d<y5Zv7f1P>5`TL^B!
zK7Q|c&#5|9x9(T>R-L+Qs@6Yi&DuTPvwQW^-Ot1P!!H0&>AB)_00aU6(BlVqSOjDN
zGzbJ00zpGXMMXnHL&w0uz<@$A2(Ym+afk_?J|!j~A|fHDqa-1vAtNH9Vy2>j(K9kK
zKBZ*kU}0dVV_;<X=Md22S23U%_!t=Y45UP)4FAXDp%cJ|qHu#d!5|obf)4`YgC2ST
z%10qlP{5Cp{?|Z9Lxn)WAe6^d6FdMwLjfRQFghCaQD_J{2n7rQP|@(w3FvqUr8Uf9
zM9#!ezM$`kDKgqZ^qTp#Pjyq>Yh1tZe@)Wra+|auVGwi)UQO0Hl4W{E%4lWHEPxR9
z2q`GC3_E^2pZrlzV9-Cx`mgcN15nW(*Ph`&Di91pMTPz&-@mqiQ1HQY8U(x$LTMu6
znn`obAXws&b3R`e|0?RUV|s>%9{|?lPkG#h4@d!<DhN0N0r`J9VgB?PBl#MpXhF1z
z8YRl(SoM;mx&Li!Tb+EzU!J57PVbal;vaxb?hCBAbi;S9gWEQiT=DX0$lZ@mGg4iO
z6r0EkM_1hv0hB=KDJ2jP;$k5S>eJ7CUn!uL*^>tqV0P@5*ZGCZ*Nhf2jk+CUHs_=g
z?07Jp9_R4x<_7FqnA*NMzn$^3))>=vYYipVk_i1<A_2(_iUa;L!g(S>rq&Q?QGsdM
z=MTU$+C0r1iLQgTxH?fs4#qE9F#0@Rk(Pb`S98epGPOX{(yK2dn>;Tte<R#)xI*$L
zi~43fG0P84E*^lr7_a;`7n;rKy{lvl<wNGVQaIW1dX5|A@(pe{pPgWAKE^C9SQMwc
z6TKpkuBt-qbgYb){>J&@H40qZPCi!8&S^nMWeDjL7#Q()oANdA#2^vN)SI)&F{Oa<
z0npxf17)H%CsrJm2=yJ8Xmzk%O1h4KLpXFdvERz^oJL$}<z~hw9UQ6Ezxn7=HEwN_
zYFm)LEzAQY`JH)7CQA}|^^FAy8s90oi0H|2zY0`&8Rz)2EeIJ;c<lcG6uIB4c#nnL
zCaiEDrPCL{e7=bzkx*pd|GSAejxGlSA&ozGFme|Ht6L5vb$i2btn=atIbB25b(d7N
z6;ti%W!vt?Y03)w#AfM+VKt$jgI{K-chqEX1#Y#PMatuzYIC}t=X*39k-|LE&h`l%
zJdO#Uf26FK$I;!=F{-^;368czx@=XPD;Lj1Ve6n^Qm&cgi5-Ya`XkT_HTO*{dF#V@
zKhPMSWSLL+oZr04&d#pP+R8{fvnY8Jndaw0b*I~loAN4mm{@d|q4S+$2#C<_ePCuU
z+zo}FY}@6)uuz##1RVR$EMVdN^9~pKKV_f}F7=E<R9W;HvQ<Mi7-*QT3XYff^W=ry
zdp5cx&X2X(72PyuDlPITA1x?VVCIMHPl^gX_L}q~h$iV$gjnT$fR(=`=-@TSZ<j?y
zV$?F95@+EX^N~OsqQddp55lUXp`?m?--S2>dXKZBYN=1bjiC`;&eYQa1<lee2^aC)
zv=0D=)o;|@AF`B=2y$zpsz9mBG}}D$a{@{%nofGLYh+}>OS9ZeJ-$3p4OwC*4%%8c
z7igof>cn!mJ3hjg$ut}%rh~YqApTI4<^gCB|J6k!E_(89h)W%TI5+xFSMwrFSCs%`
zMm;h?ETbpy)`mo6B-1RTI1O<tl4yY}9|Y&CFGs0gAd!%u>Tz<b9DjII6+ZVTgPB%*
zeoPNGF9e&|j$|hd-Rwa7-_8wR&jTuCh^GPv-4dZX_FZy2+;c?^(}QF55VgIzun_^-
z$r%qF%ITWN{tWeNxnXkLx$I;28+u<NkeZ&(n5Wg{VvdcuF`t4%chgBrE7^uV4(TB*
zGO0s18#OAxLA3sg7j}*G_x9ai6v(n7I{ZnMtgMSe#oX>j;3QjS?70B<z!O@t)i<2e
zF2U57dOaF4KmL@>OzKT2todj2GM#SIjeKN;YlD6T;9ra%)PT=h<zg$kO*?bY_ttvg
zi?RqE?iKV8nvxms=w1kWO?_pVACA#9_t7E7pvBdz`S5zWxiVZw#rLpK?LZgO64vIR
zKo1n^A@CiQ>;urt#fJL;_y`?G!9NuERaCrYvepsNqLa1Zjevs=$l@)NzQ3&GoL|xt
z)fX!XNuun!03II&v;6Hs?&F%0ru&JEbXiG{3z`u>G)f1iBmCTwLRFxX{V8jRQc$bJ
zj8}!a3u6I<kB5){KRU7GSZj6nMmgRW<<9e=>DULJC#O?lk3^K-&*ol`!);k;rOCEV
zO*cG#-D5ZgV_)+M#SOHy5XW|?(-H;Ug)s17j|-_gFZ826{}F%9dmMl}4y&Qu&rmVv
ziL0AxU7{&=P1rG#y4^OtRNH%7>Onc9#jll+w-mm|n(|Kl{ZQQt)3wO^*J=`kyRsfu
zQ8heY<@nt}SBdYH{5gM5{GIoqrF#ITv5FQggsg`M1nGQ&t#45LgQYtvXhPRn+cP|J
zGb8J`hZ|#dg}y%kv{WaS?!P1rMZyi(8Z2t%>xMzh+zBe~GwYQ85Tp6>RoaR^IPT{_
z(&dKVId(L~?b=S4&VO^is@`9R4ofCp>%LI0vmCZ{l2hNIl_tdNn(xD>4yKlUwlWo5
z^LUs_A=9fky{PCSPiE}`2qHK40dQMSN8wuu9Ini5tT57`w(9JZTzB-fprA=Msnu+W
z7JSM0*36?Zb06&|O{=lTZ?V&qTbqLeHN&^hTT)_=L)Ocn8IXdiS;6)$ybK*3&D{Yx
z-lv(aCgBHZ^7`*%eR6hk5G^!xIIHySlDx}u6n}n#9{{M2V?ik1{R9Sjem_=2M9hr2
zjCRz5PK*=ZPEd~pO|}@}TEAs40^_B}Id>Q;bUjh9kGoZR-uK7SXZPERDY**!3X^1^
z?xg*wtChP~rW@NCn;Ki(Z!%VcjdFDQV;=y*ya(XMb=-*GFJXoGD4lr?<`V~l(ZXOT
zFk0yV6?7%$9;VNj=B!i;1(|0Ooj}iBK9jg}md+XRrwJB(2J)gzOwa@X1<~Ucs|l~&
zgvRT=mKL3fF1eID9Mr($xeFA#VeANPDx^FF${K8HZ~`J=YEW>EqD2bON28YCksyQG
z>#HVisSu|3XjFRX>XVYcU!Q8{&#Mk*JOHm9Trd{ve`mApejh3bW*{ufA1g@6pGL?J
zGE+B_yE;FKeOZN{M0fv|`?f2wSCnQiyyt!h2snUuJ0}&w2Hz!~pJhy~>owPuuv@<;
za${-07p^d)F;S;pwyF#Br*ZOox)3&|tY=+oB}=k~^Q)gYw5t$f<cIdBj=PoV0QKA(
zqU3^7h^A=xPoxIV{sgU@@*SmfbZdC`4X3wtVzXP*jH(bjX`~=+JYrZg-V7&qFLB&E
zwC0y7q0`7Mug-qPqQn(*OppfOE9`WCh%c0)`T`Gxu@W+8C8cm<*fO$I`Ms3J>TJwF
zXO?{x8u;$GUDZzBg1)_rhtgtIO{&K(0!_mB{l`*n&Ovn5(+(ndpoMQI`R=rPBO&v5
zZmF8iBYbkLD394V%)W}CWnRGxvy1(PuQntke>qmI2fkzF*JekTvIfUfS8V)p#FMo^
zoj3#VXK!K#?zoU1SNH4=jQIAZZgNlYaJnjU7np6hDdzd5*R7Q*a0uDTa|)|@RVYBX
z)n;14FM9-p&3=y8QOu?3C|Byd6mp*)ds5|<Cuv4srS@I&ujpg^h!oesy+3nYCrWNI
zf)!N2<h1xA_j3Zx2nqe`sSOjPCg;v0`b_te14xUo94xqnSNlC1*)Bfv^wvyw$?q0a
z(Y^mTwM=7DSY-ypd#8=RixcMME#*3f-41#65pqpVdD_h9<X3~;Pv_p9L@V~h1aSXF
zywVm?{8>~HPGT3-k_@v35D;H~N4HKXv!O?~9p*p{_4i!`Jpg~?gOYv);Mvty;?ue9
zJ8&8#>JDcc>uT`qm|q}W4i{nL5HtN(Pd^@1xyoUCrm!7rHaTo|1&$%Xe|GX{MGxmj
z3{H~o(E}%)ene~;Ky(#zqpTzP)z!>J>T2+55*XP<|B9}AOJp^i#r?D-0}bS^za3{}
z{kCV}>(>jRHE4A2VS7g=SZ)2hcGBwD$>{SE8#BA}<{YcgoQc16yDTMm+~wnwoOIMt
z%3@3FI5r6TiKl+B=wzFG^<jrG?i>Ej|8W2T9oR*;9WF#y?72jhrt`~$_k*aLh18N`
z_g;C;aVu-f-w$;xi!h!Ax@3_`E7gf*yHN_G4L3Nzd9QI~iBu?l(}E}bBD7+4zrFor
zP*&l`UndbhtR40o&YBSRqK5^i<qV)+{yT2eSXS|IM$j>yQG4ojgkr7u8|i(Epmt4)
zyQ!T_F3M6l6Mj4z?bx|1=0%U+6N=Uvt9_ZqnHHJEK?!(TEF?zFV5ptw`R&1PcfksZ
zMoCWP7v>>`q8f?METj1{ilsUZU#lJlo}x?7*`H6S{MEl<GcDI0un;l9!64@K8O2ne
zyQ+1&$C&>h-*fQ$8~gpViqu|Ky=p@FCut&KEt@0<<yB#@0C8w1+SZO3jqijT`N6Ek
zqKMTiRIxwA`lO+B!*Hdqn|opt8GvkdDT^<<B`4)qwHob@A|=AzFO(@NG+>meV~`Sn
z7lDtH$3JB$(7bCf_W(c`Xbnvmm;mh^yTGs=%~+q1yc3+Msh`f2VT7#lZJRi2wwqGR
zynnnUHk+sIRJI4^iTr1=dmb@a=+1Rwjf(hZeB6W}I`9@5l2pmI3+(fR-@#GK`dU;F
znt2Xd79a#zUW5tbw1vRD?GkB*UXLIyxLXl&Y@@0teX7q!8IF47fISKEA}S3V+ZNq#
z%F}~oT5hMfJhu$jy`#?AqjanL8M{N?#H?B%La1Yzt@PNQok0nCo=<bgbaw&zgf9Rv
zEqykeW=&Vh2^|<5o*`_ui?ZR0LY;7Z4H3FJ&y?k*Ltj?Nul&IRS^ulIplq`#%|>6M
zc?P=DI`zx`Kmi->DX$=&tE<^^Kbg@HiZscY9w4F3t#v?EFieI2Fkf>;-}c_~INBIO
zC$9M#m5(a(evTKAHGcqvq?~db4Oxcjsrhgan3TnAp~5NoJAgz{{MqX|U(5a|b&sXU
zT_O5$j1y#hV1S67QJwY-vC79}PJTkXZ7hV^CF{QfD@hV*W|p$xp;SZ)OTq?^9To~X
z9^P4k*0)0UdR~vMxp1<=Pga$xqlLtnY;-5+wG$9@O)n%K#W?twLV)$<D?c@zF||(Z
z*IhJN#B+%)=xbmm6=lM*ZWe>d$*=6Xs?h~0qLHD#89iI()CK&!c=U5dJTB|899Q8_
zSwcK!`mH#{dAE6xq#{iXpM6)l80Y{VC=>|1#zH_j)ljS(!{w}B^>9pzKKD?<{HWI=
z+pAK7fBW;UUUavEwdiJN$l2ZUWx~&`j|Lm6D-5fi>?M+TFzJ=hOu+wVdq9^67>_6O
z1C8GM^?2T!l(6Rx(fONHq~SPu7V!yz=LzR}s9rUZ{NyUHqA(a0j8h4k-;P#5&ba;k
zgF(!LL9(lQPc&C+4N*xNeckO}m+SgmPn@gwJ*GQwnt}OCvZ@|Q5|Pd?M(xYK493Z{
zeq2*3k&$wEgrsgT`f2||ZbcXZIuPYY@HW?ge{jz}$MTvV4|v|!=$HYnrUc;fhSVKa
zRTSY%jy?(e-hDPm!M_C`bq(~pPIjJ2*=4VxDis2l_cS!PuAWvmeRdU?`W$&zYs^1K
z!R=WqYVDeWZ@W$SCIzo6a-+(rzvd4n{uCgr%#;WgyI&APVB%slDvX3<t*{6_Su|WI
zUze;ldS<OU?FQPQz^AD<&HE>o`By;mKR4;Rp1Mpi*dyP#JOF0HFbD0qtDd(8huoA1
zn)~Oumz(!pi3{3z*e?ikv#mYM?y*lQXle8V$eM?w`GTh94>Byu%*irJ6x7N+dAmNp
zwfZXaeR>-Pr!@!7efoVovvT$53Uc^Ag>BM8E_}cC<@fm;hZ(D$t3LqSSkui;0oi&u
zuIDWGPGf;tdZzbARd*VZv4q6JN7owHyPA=)?eA_fOGZcuu}N2ADh*f|I}A<R`Pr5=
zj{d&j=1NuWpspTHG@E<*14jjvClYdyQH7jyo<5^YpSAe$J290fLM!_?OHIWRPc~yE
z=8A-_^0Zi|o@x!Bugq$Y6s|gIj(0J3ShyQ;|MK-POiE|&7(JJynP{v-WVw?M<TWW@
zwd{T?z3qW9OWEY5oau}&fll^6j&=U`P>MbC2dqs5YI4sje}$$#q@Q=K$jk&hMP2y|
ze*las@rrT!F#2*+V^R;54fy`_WkB{)?j|#%N-Tn!Vuh0F<Oi9$6sV=?Y1mZ>z2sin
zl+Js^!e3OroE`cF`QC#2F9T%%BG1zE!JO&Xi-}YbzOwC+XgBJl$>C@FEWKFz*;}Qa
zZ-LPvjd0(>QN*gwc0MxOfC|0n*mdS~WU7ybjne~N1%dV!Ws27)6YA=69n<N)H65kJ
zQK%i)7+!L`uLhY9{WBoV86g~6)2vYyF(HE(Hh;RoNu+{qE_Q{Z!A(|%@EhL03K^|b
zV`a%~f-^NhrcjQ91KqdwZ*xv-({r;Zuf3GhAB`c8|DVR7JuH6W`<5vqOhPWX1c$>u
z3=qZ4<R$d790up(io9LL#T!XDs(bRWxK${Xwd}`)nuJ0DEv+TRCIGrIQ&^1b`-5w=
z;hE!TonYAEd2~(ncHm%vO3UjFEhb3o(zCK}<qf0h@z5RFCu($@FV;KzDWVP>y+ta`
zEIR|NSmOOGKAov+^4?VxW!1w*E(g!!O{+V2z^-*%;{d}JcheHW1C)3A-YbACI$<-c
z&w=-CjP005_dz~Q5Q374>WpsQKbHdo#;65ddt$U~c((ZEW%YOca-+O5Ga91a+h22L
zWMLoI*?nUm3$7d<%k$AwHEM0;Ae|KTQi8)$IJfF=UwGmx*;ivGR2G6RQT_0H^1RH>
z?#vFiAHOBLQn0k5_l&K}*V3<wEBRlqHs#dO_uy?f$=unKwwv$r++qOtTE}8=nzO_g
zM`p^CbPl4&>?}Pe;yPhes`-nRcf2~fx54tAv|@I~=U}&$l-gxyuZ#;Z&5HrD2BkmC
zmbBxcd0L|Czvx$F^RV4l>LFnSQoX{>VRws5K7XQm$eLfTtfLmLvIc2`n+7e;c+hhO
z*2Uktr?uMEnLP3P3~oTPi*3Y!1k=JO*HlkIg5*mWG7RX!$02;RYaYtFhXfrigOv3O
zu?S7vEG4svOz9PMn$HvRdCKNjKWax>(A-E>4>sK`zXf|*Oe>(mf+`&}p{Jxz+N&9%
zxtMGiFl!YwJXS<EKtUbU{__ii=UJdh`7^tNE8k>Y%MmVZnXw@S6E*Wm6|2ipb^P-n
zZbY1LM1S)w%0)%W^sM0c=WH??D=WZU`?eF6sjcO3Q}k2gciZ7TE1h1h_+%rm;8>7|
zp9@hq9`CqA-=i(e6q8rolkK9SKjmvF@<ooj#swAw^T_xFB?oZtMbpO{HcuVgohr6#
zxiQyYI8Jo!<-=w>7V%{&xZj^|_61<4ezM6c**gC&Ajjjmsr^Ih0hp@3F;k48`TEr+
zdX`JmWyT_C>Loo0>GC%)LK<U*_UE(BhE}%x-koEmkC3dA-3pZ5YQi4%{4ckH5vTE=
zN8U<uXEZb}3v6+|k!gps<Ksq)p!aC-?#Q|z&w1Q&(;(3Vu>pNH+L>=#U#E_naP{ZE
z^;9Y{Ln)8}ka(V3_HNhK9Jz}9BU6ltbRkwzmVF!SA0=gi?MJ2b0kNDX?FCe{s=;O;
z03X0VIQTHwBVIp1y_qBI(PU@z{$^6bWsK5=0fhk!O5xrnk~L=jsn}mIXLab@>Ppl!
zY^s}pb*m48$A<a@A^QK=dqXhz{L1nP=u*@>HHjeok7+7^+LC^snp1>PmA%4_o}t0h
zE;5Z>28Qv(jO8SQn75h8b6)&r%=Raf;J0IrzNA#{^3N3pT+zdqft0(*B6AQ3Cm4(Z
zzuv1|O>jL8c>x*<M1EW+N!7wNV>y(OVpz)2#{>OJN3VYbTiauP9qZyBYSdu0!+d_p
zlve_olKD>BDQqR6<76JY3KG*S9u7x)eciGJp(=`cv0@MrW<{We_8P?|*G;_Il&v(S
z;&Aq{a+ru*pE5IceAEKUsBsS5&)P5pGQiY=7TU$cp|;cXlvl^eIN+NBL*xE*ABbi(
zC^b@X;Hucvt=(wZ<1LTjQ={ExwtmntR5LP2W`7Wl-r6^L{5|e4tm-QCc^tmYKrh>8
zHJ7WLO+0`F*FM|4Dh-k<P6&`<!YHV__F;YzyTBZcucE$iUCW4a^YqGBo5hr2jB!y&
z!~|RrCQ&MT8{9mHYy>YFu>ZKZw0mE%oSze<u#$qF5tV!z>esFw#DiG>gWGR3bK#%H
z=Ne)46g$Z!QPWC80b8FkS9`sBMd;nSkg#%JYx@JhJ^1!>O>|wB-*1-SQlie3jm-z(
zX=!C!c90kXxqL@QI1?88oGidOZnnb8Xh{eq7|uvV*~=0)u6d1Ased&<PAOSSQFo<R
zZ|6dF%NH{;8c>GzA1<>>pWvCo_LFrwK9x`N#pC?;F^lgTTO6W~SZ-ycm*BG15v3!$
z{5cBWvX53MuJ0DnaSamPCiU_0KnkT)u5q`)s1=Nad~0`1B#G-paf1Z|gE}Y-Uu*z2
zMtp#EY!8JA<I4?_Blg3`VB+Z1bj;d$VVBH6Xrv$`O8kd`K{F#IGafq#L_)@0+;6sx
zhH5><u(+CAv=k-XLKK+4j#<}<z|-w*kAE|7Mm#r`@bbsz)%KHA*DnN>0L+>CmqQ*o
zM#?&qv1OtYd{#(!GZjC3fp@{=Bx5863x$mcQ=+AFLm1}<5yj$l6!LM;Hrq()BKG~n
zm3tblf;xRN;09uG=lc8hT|OiFH}%g}tNX<u!GkHjiX2~}Vr=9+2;9!VbhFD+XfT_u
zcyvjpo7Cmx(05gqiMi1&$`wWbC8%!7e>0nPosPv`-|~EZtZwW1Zn^PC-Ki9;NnYGL
z^MPg1iUPqiC2Hb+Ouw4CL5bIYg1ciNvmAB+W}(ZDd~T`B8TFS>jzz(8mbZELG8-L3
zF9=h(TKP^eMC#VOe781Naxl`|o^0iv{%U8Wi6Zd&fG+HMa(~I~?`usXxK6Cg7RJh+
z(K2Y~7>H{$6dt;t3>Uwl?wbgWb82l|i2ts&PEuT5?6F8K_6lnS4}=)U3_Qczu|Pth
zQ1}LG3@V5R37~7>pJcgT3e?{-kIiCJd&v$5fHZRx{Ly8<%bP0PhWAH=rEOMy_Ss&?
z<HKxnFk|7T-G3$ZA)?{MOPwyh3dyHO%c;Mdt`EqTD~H?E%<0sw)J&oe{beoa=c@Dx
zmy{~4d(j&0<AnQq7<0(4k9bvvG5`Xe@ax$2)f1(<ORA~QqxL+1Tb1@sHX~?*RXPiF
zpbT4tkrDZDLC{=dXUdu;&ri-t>&CEfrF@bKrLsL{v*r0W&r50TI_*7o+0TqToP9Q<
zhRn!Bwd<9xI1aEf5mC^?6hxjFD?+yMEQtLZ&}qK|cc1W2LSHQ<;iqV)wMvaYR^D8=
zT}xtA=}tfufwW1h;+mRdr%_XHWPHZ(inPs`|6*@wSN=)IZ@rHt=SdAw&#Cep+~h7#
z8K3UU9euLHa&cchcH+$X3u3l6MjSFxW?DE=u0-f&8kw!@v7<ao@56=PQ@tNqWHQ^$
z)CXt=c(UPeT+YV|sRt0m(3@}OXZSXg+ui6-(s0~b3h?LSdm^k-+VIZ4&A*Ft7e2!$
z4}b#dK5F4jJB?Kx5!W0HQtJi*n1y#rM18kdaO)wJuf`OPVq!K4n$-(jV9bC>6jkq&
zB^l4b06N|mXIZ^5#!zf<;-kN<f5WtqN#($ZI5}DF6tWni)6v?u%sT;2@u>r;3&G3;
z)bhJ{&pW&BE^rey)%Dr5vlm5J<)nP~B8RDzz<d16ImbgKRB7~44FjLKJ}rNxPm8RX
zY<1=h74gpk@K;_FxGDS?h$gWg?s`foG-TT0G*P)ks}8ezGYX>&7m1L0_daAni1!cQ
zdV`lloY79D?cxj-!nEGS^hbmC`=Pu)^wqO1e^4*np1u3t86hUm(nyEHnlVI)h05az
zn#=G}*d|=(Z%vx5X^FvQ_*hwwkEW|!qQwy4rh?*xQv?Eb=-G^LhuL{5DK;-JpI~wr
zn2Pg;f-TIq0F)H*UIZMVeW3yS5hp@sXb_%>L*X4;*Q`!pcD_xQti67&<e;$t`3dK5
zK8@M08?t9P{w=u8C?Q0v|7o<BRf9LgOfx9rm0)WyB$Kc_J3jh#NG=AVReQKJJnwby
zyrT@Y3}OBM3Ou4QKNf_Sp4|NksMJFe@@LM|8R}+@<3N~WG=KZM@#TCl)CS*+cOX?N
zD!ykcb!OPw;W4^3#p*r)1Ifo2OGe*|r0%4p#xW>{mS1(&9%^>n)|Wq4nABgK27BK&
zmfbN*odyf#EvH6kJ^$VE+3>z>)NES^b^jRh00=Xs_Pc6-_8Cfafo2rY=rh_@(($Sk
zl<0P#dsP|x_VD8#%4zp{7lWocHFtFgdG>SM`2z4PnLq#W7O^<>d?)htDc5LCgk@9g
z_GE+DG)vs2sGNwtvVGxt;rVkrR=e^LqaNNs@&idVXCSS{GyXl58C6`Tja~>U-rBvS
zWWhi10rOpP8W%2MdbIw|-{`6n*YO;haWd<BzD+;G3!UzN=PIGmaiP2?4rGx^=CcTC
zDzx)ISh2#XatRv8)Y}nfEID;XvM*Z_)!RUgK^@yD)W&f1cU?IpqKajVgb5`Cal<f*
ziNB#ogC&-!(sYQ;71mI&`3V-(?)<g3KkV3pV{*fGjzk9%1)9Lr+Ch<PVtU=KDxaRr
zIqLG<8>H(pQ_0gsjcDZoZEWF46Q;;E{uWg;#AJ;MEK6voMQ5RquEtgU7GyO)g1NSg
zuPvLRO@Pi5QF7i$>Ee<Y8|8#mAWbD9Yd*4gVb0YwZ1VuD;YQek<kNrU1|rd`8S0J(
zcX^JIzXzy`$Sz+?Fc(B(<!kd|GIP)H1lmbT2My%#BZ(<lh8iQ3lkoR+asFug^);)G
zyGp=<{sjAAL*rlc>mBEm!g}xH$|OpzvNkELp1_VLHI9iaf>NnngWQ)y(Y9G}CYwLv
zTNY%dGS7%0Q7mplHYLz@f^8HCR0stxOw=b1I}|7_`<ACW*86Ecb7>`5voI}85(ijN
zsNfE@YY(L4pG16Ri0u4AnkhsG>Em&vSC=*aeDg(DG203VOm<5~aZJI4;}~Dg75#=O
zmM{_E6%)pd(vr{rje&aZQkLlwVzwF!$c0(|WpB|ACG=7}LApZGQ$9K;o;xVIhd84a
z$2MBL$?Va4D)3mzvSY#rw8}kD=b_O3YBjlvsJ97P+F|P9JB&Pb=)ZGm7YGV}5i@jg
zzD>ZSY(ATNjc1BGxmw#gdMO~HtDN$K_mh!|r;K|N3|whpr$BvlnD4pI=9Yd=x@8T|
zg?<s!v$y{QuP2=fRkO-wGQa$Gh&DB6lAZIxVXOsC&>+nK_dznjxgHDaYbpkL<*ACz
ztYrb8ToM4)6C?yGMlsBG^g&|lcU6i`Rzt;?>qa?3(n|F7bU8d+jr#?yF+Oo$nl}G{
z9?GPJaNV^|LigBc*G%T&(~}H2UR+QA>p*uA$wCCNdQ@4|EA?FMEf#3{1`X@!5vN}O
z^2*IG;H))h?CLV;)7|3M8IEz@Jkh&sEw(J0)-pT<#=WOv&u9lbmtOZJ4i~>&+BY_*
zJhDE5*H`uiE)^8u<m7y~7_N}N+M9i8QAtp#Xaw3IXnM!h%e?lsO|&!Iev<HbCA^=g
z7yecGoZy1`W$3!9g@CwO(U7B^BU{a-K&i!Bwn7Y*1$Dtr99nQe+?~`H7b0~PH62JO
zVahWtm+(%B5X^$C4Wb!3ZQ4uL5Hl&W&WJ0?0kCVP!Qwg@)rFiD)?i44<EUi`Mr1P)
z4g=RH`N0PMS&qtMW&FJHv99x}!)Dd|j9IXO@r;6YItK~MLL3gu1j$}8QHb<rZtK7Q
zAbRDk@h&K78<;O~*cA4K!_j{SQQa{?x<bDj;0qAjb{}W=jb<X2B7tn132%3ds(y4k
z;n?nO8?TT=k9x_Ydd{b8HY^#MLI;=tEqE~?G^Hl%m4NV=yK45A1+JMIJe|(c;U_NP
z^^8=1RXGkxJd@0Iz0TQtZPR>^c{;5UseBp!2f*7#JIQ&x2LLC?t^spRijaoZN^j6m
zM76?(DoQm;p^r6y>YkuVqi@M=Mc%6VOM{<Zr0r^zOvh>rlanR#q)Y>tNUps6ofBdi
z+)%OFXl_RYnPN20z}Mj3CgWwvk^KRqmu&P8V-B|<`s#ocG9VBegc}73g#5s#&fWvf
zLNt7zpVIhu{z$jxX|(!P&RN-70+BVpz*Rhs-kQFpU1NIJ!zUJXR9F3ki7O^(T%Hw)
z+pNKYtG?ox_F_wsKcj>f(-O}^18+X?hS@OOei=i~jX*EO!^8ztcj5rS?sZ^pu5IWI
zobwcRpmCVgW7kh?U=LwI!(+3-2YdMfpo8nFC*2_puFid7^AZg;Hg#`M(bKGHsJi^<
zBY4iR1Fzi<%7^5B<ZVh}yJ=~^zgAP{c?XtYcZ+P-9;EC<r6LlID8k{6QVJN%g+3Yd
zPdhI|y2QWmI~6z!4zTXQnrwGfSsKRPv^X2bd~tT$Q6qdKYkqIHA=I5eDC5O?g6E-m
z5*{Vld6H5|>$GY27~qgPt^R#8ttEanvw&#<+l9|D8}9&8b13=VG7;F4F^+q4D5+L^
zSULlve~9OVhQ<5uG2X2j+t!H-^=DL~U=|x8NzogjDd?s>qF(8N*~q%obIgt_gR|VC
zgEe?APqRue4CBxrkeBVSYDfrE36udIqj~_{3`%LSzV54{hs6C}YL`FtZC~pnZ!&0-
zgsqq*Bj7hq`$AY-?@#q(R1@54Tc_ed4$_yPVel|Qi}tGPA`*O*Q>Ot*eY{hxM37=E
z8F?^`b6f=?0=~Sm5^44PJmC~U&`BqfBReHm@dv<Wpek-}#@sT|@fB}(_q+eAP@X$F
zVJ9=z(7eg}UT~goKI5buuQu@pHxiU57~*rbboS4cCW}|#a+uN$s=h{p;74o{d+AZt
z4liyJ+pHZg%IXCr@&<Kc2>d~!myLfyr~90@hK}z{AvbO9=<VdHS7H@MTcef{q=E*_
zwiPlbf0dR))=Vl46Yk~ofq36ZpM*Svj}W`U5<aO<urT15-S~E{NkUif^7?Wty~kgr
zucyVNyhIm>tgtAkLG;Y<1)*nPf%j)kzhG80H?$JGLg3=+Mjt<2+bLYTvQQ1oc}7m%
zp5*m1S`QN?T2M|ly(TUL8u5cp3@i8*uZjgxxAX!64pEZ)rhw06@L01yrV63N6umCr
zfd(oc;_)Wg2&;}^^jMr1L?+Ms+ZEnBLB3%!@P(->NQ=nA&4GFJ>Q_rR<eF1U$EJ9L
z4nGC-(@104b+Hhuhki2X$2evek&0FI%9N0ovlbtWdwtQi<@$z(CsAmc@Z~sh+Jen#
zx>a`@<X{=)MVY{QEl3GebMZ}6?wJBcNhTD2`#lAaf>Iio8yp-?oQ$Cz%xGvf8*9)h
z!5lPO=HuGoX=-d+Zr*bcb%UVE5BQ(MT-e_Y_H9V<B9x~^_ZLJ^5Cp1HSdj@u`+tJq
z;G_3e7BkOUVS<Qq<S7P-2Uw}hL-=_pbYF&dYeFMknC{?W=o|@+q0TSBgKE%{>M+;%
zg|c^=*%vlD^qc$xasndWQ>;z$)>{A-ZkdX%c})Fijjy=(kBXwO5La4vLK?4@Qz)fc
zPu0f{mZZfFZ_<{FWArMbg9Qcfla+f}ei!p6;Dc1qtM|XAP+1Es39PsMkxnAhe#56U
zy8^bsAVVlIumA~&Jkg&GS1M7rq~n~&y-xd|t}7actzpcHb>>=56%2Tjkedk)oIC<!
znc{4pt@F+xoP}m<MLrgF9*`0MTA9Za{%tU~^s-^?-HX_-P6i|7nln5v5q(mz+RZ!}
zc@hYTWyc?Qr6kS`;y75GPrqTOP=94ZNg=1{L%sHv5EyB1$w^X$!QiTVM4{YpR#`ZW
z)jk@h^gu7z&Cl^H=kcc8cM@K!)oGMSJsLh`RaO6JEhB)FlMtM5z(p%%XkzT%VK!RH
zn8rIUnUpI&O?$XCF)$v?rHjyF|DYrLis^5phv@uY#RHF>CKXL%7+c0A>o6MV1`WDE
z6k32tW=Ux+o`gT64*DHL<4>o=^lMfVK%qYhnwMJVD0hx(TaQCg7qOkx0~-ib>8kP}
za|r2VGvE^0$-(7`nixc^YFcgt8G>Mm4Q^&;7MFd4IaT<2oOkaYHQkn<kSyiZ0J5%0
z(R$fWBr54SuNwvu>Hltxma2~PRzgKK-cLz5!Kb7bK#1LZ_+p@%tQ`eWZDwFq9S1Z4
zfMRYl*IJH`a+oL#oV`y92uiVZuBSSPKLB(+SBqK(k)sD=&Fz?G<irq6t=E#qwuWBr
z1LBHfPx<2tltC?`uX!{ZV{ylqHUZWcXFLD}E!4C9yBUo2zd|>Ff<_}3VzVF+(}br`
zk=@i8h(^=U3YCHk+aCQ9c$G32RWY`?oa{Q6GMFLViQ)bSiPpMl<lG{jjlQnxZh^`6
z`*J`#l1Q`yvdNPofq*kJsZF91vkl1cv%RQrX2FO)!A<$A3S;_rsK^Ll$Cq`k$GP{l
zdP=O)@ho}Q4rDSGK>0ZJ1%+AWe=%+m+C2kIH1;{L*`HRd-;?$^ZHuX>4CdVRWEAG_
zGkxK;%j#~TD9}+|mPr75E-rH|qAbeT82-amno(y5Jt6Md`zd;idhk}m>6qgUs$p?u
zRE7G2@&yu5Fy$V`NOhVK5`0;&?3^NO%t$qbA|Ctljru%gpjsf+GqF9<oJoiRCT5;Y
zQ13IQ2|L-%_hwyrWU^o@K>P0+M)<drvwal0TDpi`lCbQQA6U-*an0zh`~qi@1E0Q_
z5E;~#!^e$^ta74Iu`@N9oOq)^TcS8WhZr7F#H068a|sXpfgR}M;NQ6(LEu#y?<^Z$
zW}Z>Om20~A#0|4&HBRdpK}LZkCE^)i?3pI|_nge1fn8l)Tvo7mFroZ&1P_IbSTdVk
zHmoL1v6(@Ad0Z19eYddis|3@lHSAbmWGwGzn-B`xn#b$i4x*X-!7uxEvSc!?apWJ3
z(&FwkXuR5)SzCMnZzVK|c&VXA#@_xqR{a*e?Qc;kRZ!Z^Zfg{ZpV;m^>6N9N>93dP
zxZ{GAPH;xX;erthO&17cMuvh+ii((Eh#`$e5UIJidoQ1?`ga{apIL`N?G)SRrMSTa
z$v=+~vT)EyW%fxTh_^*0V&^Wc&|sl^q<%Qv%Ssb;jFU+O5@PX20*AkIDjs145nm*2
z(krQ1UMRqHjSw5}om9ocQwgJm{)RucvYrR^XC*ncaQMbI7>bR4{y?^dDfG#ye4M2c
zg?t+qcNw5wZ#wUb&QZ&SwRiNMMEm)Q+2%2Pgd~V0*Qz(81eVYh0ugFvWP@k(&*U^-
zG7Tn5|8Xe+%x~N!#GY1g>5MNEa+p@+BT>w*(Oa=8jhDm1MfCqH0FGuY+S5)6XiY=}
zff)juv)2FXa<HtaQj6T4mu*NW4x`jKOL}QuP$_<Fwn4lp<qC4#+@GmzB%o&_J7*kG
zE<qopSPqPE4$Izzg8n>qDQK@1IuMCfJ8@<$k_n6g`I=4%DZ44s(kPVRcULFgI3eS&
zo*~9I?GHG50yqx<sR4>(vh9<tsXNvOfXo~Rb3uZ&9c4Po>$TT_h6M(u<gR>*gjj~q
zygRR|u;+-&X4YZOd{Oj^NIG_ct(O5mt}IaLI@~<yf8*N4CALmcu1(0A@b(<!N{xt_
zGLB2x)zth7t0=K($p+*KzvTQ<S5avKAb_Cy&zA79HuRjCnpMVj$%Wd|_zA5ws4P<u
zu(Upk_VYHxsO|QWU5BE^+kok})JgYJ!1=)KE1#E~H3RZSYcw1`<%wL>_af`5hpI`Z
zKHTHo6Mm>D%XmXHA`(ma2Z~9FK=_=eO-C1mc3GxI)hpz=ARQMtB9l|{2tye5_6nZ4
zN0~XY9$CRr?4f~b#iF4_98($+s@6<vK|gIX7%L;)pl)d<c;SsTaekE@L*Lj7sFOyz
zWgdGMe?9i#{*!npO}}><;XY1hvw-=G6a3qZuNO^b)C=Se>|RzYY;CYEebVmiKg`2F
zeNQZ6jD0d7Q?l3ew}@K=xzMG}tGq`G{(+EEJl`{94f^bZ_@=Ae$zM<=Eg~l=y|p0Q
z@q#t0$0d#=wfH<8msB^YyS+_r=1BiNjztQj69uL|4^0t;VAwEN<Psg=cABrzetC6@
z)Qj$jxNN0%uz%Zre$lU-wXIr1zEqN0&-LEBV)**{_l))j{Lnme6qk5=++5?%<N`VU
z5B9yEM#59CDaehyi$C^_I!Uk2z+iJGyFD`Wqmq-eB_ssvZ6}TWrpeX<?&dl&0ic$s
z9-7p>U!uGU%=q(mN)#8J*=y3J>(uVC4dorfd&hbCUyK=2WZEWU4uW+4hD-LsmwS9L
zl17D#a~ONKNpWaob+0>X6xSTy=WN0Nn0UXI?6epFxX*WHz|8l!hHX^w8@%;X8%t!G
zq)|JSbm|U?^61+f8**YQg^NEdzT6D&VSD|&mRNnTQ6h2r*qBA;loE=HhY|?jzB^*l
zPe>AX86;fY5^`wMQM$&_+|c-Bn^dCOgmJT%(NjHVyyMSgO1$QvvM6R9zGc4?SC?&B
z*>^5l!uOn4G?PE2`w{l{py)=T>N&rS5_!FW*6*AJazvbWV~TiG1mZ~ZQZdN{9_C8d
zQ0<o~qu;hXdl3@b7+Cx!Y$OqFXfEaNmisuv#@5|(VB8(8QVm_nlHM5a$G=aTYAIj;
z<~9o9V^#mvuaVV{>sA2@g?DXMzxVLdUZbkQccEm*9oS8mP$&wB^~oR3B;<u7Z~zGv
zm?*Y4(3xe10Cy-7S$ms4-5asA7)~d&H|i`AA@JGRGG0(J&gk33LpiJr^cj{XbIpn8
zS{39>`c6Am>>gcG!G*5>>x+Z9Dp4Tg1co5ELvlidihyzODm%k`dU~eb-A#G(GH!Gs
z342b&fGGpOMdrtd?Y8HE#cFVvtq|i{J^V~(7MqLwtyX{ETapn2xkllvp@<S-{w{%$
zcuw^TCRcfrmP&yu#RMQo&pSV^s+0b`p}E)tXI^CMC=DeN{sSMhLknI)lMP?X#jiEU
zUR*LID|3#j<2z80e?nPBl@dtnnryQ3b`T<wG<0Thr)0STd(2K!LrSZE)>AxQfiYPF
z$e}xW0-0h_dd^d&f=QAmY|!@|QnTtnK)>TuQKL2Gi=h`80kE3q({f*J#)Nbo?Vs`*
z(lX5=;UF-n{0$nUih#Omy>Nr)Q-Al~`-X3gnL$FF3!j*!oduh4WZm&t4y#H^WJ!jE
zOMTTpM`oKlsCx_ySbUVnj3m@|qs@9RZ=O$`{}~ETfx~^FSZWgf)ZC}f^|cl$h}fwt
z$B5rGN3d0p?WWR3hK7NtO^gErl}(?V-aY_QobK0|Ydvw2ysZdBVRhqL7DuBs`eCMd
zv>xf7kUD}I#e@y+Q?cRntK9`{KA6Su;EQT6CR{H-ypH4Of8RIqf3u20@>lN@)M=sO
z0~rouEkZWiA3fzx*2EPl=Ocs5tN#`3a}xKM#CO@=b^Kf7cxMxT#hfD%E$!{_bg+n<
zRlgRP<7&cQoa%VZ7)c<^qIAI^>;@+01>rWhDr=NK4L(`nY$P<OviuX6BGKap`a@&*
zb-_BhsnUqVb~4H~$Q%9>FD_)NY*ezXhTcQWoLBp^)JdZE&bJ2uVD30c^nCy(AAnzw
zif`7U!gn*xsrCloNWHH4G4KPhfqt_+cU!zc(?0b8px<BXL{eeB+w-ZjH;~$7y7kSh
zy2rgc{pA(mLw~*)uo|$!uECrizbsm3uRqrNr01Qh<1P2i_P3Lh{mblsEl*OBOH8}7
z{rDE=>LQi1_y9b)zoxsRaUp3CH_KO?C}N~D$`HY*YzJ+KG2#}=@24f-G)VPQt5_5;
z9YZ!M+%jE#U`$c}wgC#~`g4gl{@#JCrbe~XnI?&zGMOSl4A(vxKZo8mm8x~84Y#2%
zB)+fL!7*b&pQTN&Y=4;&3t!t!T_g7+cI`fEbhmrUG#0m@S3+>hn+SGwWfEH$mk30$
zX-pg8@QgRhs5;R+8K6<jw*He`MT=RH$rlKAo)1J<pTBJ%8jM9i^^h-Zti8MYDApPU
z?z9FRP#O=$6G1iyP5$c%#Z<3lmv9y1DD7t*yPt4TX`2gqb@JO}X0h68;>GE(cK*O;
ziElLr;p3ktFTvM6gyPqGPOW`c3eO|CFs+hZs*B)uvOpl2DV}ORxASaKm|^^n!n5#(
zRso;Q3f-`$yUhgmqr;#JM$q|m@;S8=>)ufa*x<}@x}SaQH5X{2i6CAoy6x2Su)Q{}
z+_FK$=4H-&&4A+xX2a8chUOU|po_$c`vr6;L*mY3-ZQ5;F}oc!L3MK}knYXNCdiig
zN|inL>UEi`zc~sLm+LNpJ|`<PmzYOIjG?!NUfevURipNkvpK+<giW-Y;#mlNjlS6Q
zcJ;F^Uy1&LPaKZ^y1X2J#%#O>8$7bEQZs@lGSq{vI_VvAq_+7p)m>KdTQLSfS`D6z
zexA7Ae*8Tc{gRCZM5F7I$Iqm6zURYDJ#tjbZN^PHTw&Uy)VtBY$q@G4c1f2zYqNnf
zub%J5M#Uq$R+`oWIm|bx+eD%ym>KHuXt89nNn8UopWmW=sqm8<bpIfXC@&c8n#;jr
zqPE911JXvpH1$>M%y<XJbB)p=4?yUZ9y%AYffa&1Ea{V!17AC6r+2HD&Y(|qIyvQP
zdADntxq`BC%Ak%%gImP|67lhNEGcp6feg+4<)6^XSNnA-5Z&NW>>Y?%9wdN3<WV<M
z?YU5>OXN9UMn3acdSM=IV;-b0%t>FT<sCl{QfMjJLiyc?z2g%BKm2ZUW&a|UJ1O|<
zEp1I-Z|pSKKmt*YX+bwKsCf!~Wc{6E@b~9A&*3<Ka9St(y82z1IJK~Ys_X4<EfML=
z@7f6XfN;F3osZ{968Gdq+z92W0HbjWV@8{7KPsO7Gm8T)sN*IGl*9oK_d<qUaP}+N
zEywY{2{1m@tlLX!%(Es?SFN6hMo6DuTf2&ON)kr?silac79J`4lS#iu8ze1Hyq}T>
z{$eMH0$O5*SwN(tOZwAsI|5|8w%vy+T$pU@qB5V>DAz<JJYobC$(_p#YGO-n!(%x^
zyH`Z5z0UkKBGg}MgkW!*@Gs?_HANVl<DL&)2%tRqwg0IUQqj=!(`<IX$!_&@n#wNB
zb_0UwrKOZ_YC(S?S&J<l-&V60i#<yfxU!0X^-CN-CY@*8&d-5BQ6A)BHDtpM{z=m+
zJ&i1NlYhzCL&r`j<LeiZdih|pJ%IY#xw82G1m8CFK5gaW(+6Po^#c&>-7_>QRot-G
z*KwXD_3MMv33|=H$XlZ0t3+eN?n)YG__<P~D|ft7q*Rll?|3`@TD;@ISw!wUWnKq2
z|0{A;qIn)a$H{!@5P%eq-pD_DedqE$z+?L=q;u=+zp;fCPHER4v)axrjBiag6sf+`
z50_KG3hqh&LK7e;kO;yjky{z}&IW?Q2Zg>*I~`=H=%^BYhJ*INC!^E<K?YQ7;3BkL
z)|nAvdxNKDK6?gGkcD;E+B(o`B%Or0lUTXqB#<FuGERxyB(P`Ed|<l5sX0Qcvt*Wu
zjS3`BivUHC5O+*|gdEph0T0J9@6|y_ivjue;SaEonjzDd&z(!1i81Q<T({><3I7mQ
zBX!0bBTR!nBa*i&^(aa=st_7;zsDW=1$#yP5;!&R=<}8!N*%cT7nIM7O@6f1rVYUk
zbc5y2N114M<n!XC)Ba6-&3(f0eCNG2Izpwu5~m>#m>*>#&i;OklD!>8JZ+pC<>sBD
zFci-#FGDV*Pqy>lxU#7pt<q-uNJl4^dKcU*KKk`fwA6ShJ$b->+xc10>wkotKKRaI
zz;VzWn8kuO%;Nl8Gtf@{2uf8$pF<OynmH1d>Zs^ys)m@>5sMfi7OY4K$)~nIyRiyi
z?&g|M_1G)<!bXH8N61530{Tr$w?ymT`=5Wh@(n+kx6&EwSD9p77u`GBf=N7pAR=f+
z_He{svMb2$UzFZRWA_h}{A1+iABDBKs2WHV88Aqd7_cve%&0@Y>*v831H{JJyu4!@
z2f0+r1QAMbnB10VDbG$3i%-lGh*Cd#R(bYZ<$;mUTFek7CUk)T!WYO00+Yyy;e7e_
zZyF){;C9N}t&gjrPI*gt^HeP9p$_UQzP>+Z|3)Z0CH-&u`bXoDdcC!OZ@af)6*ddQ
zY!hN=(V6NvI#{&yUv_-mPN%=Z5o)ClN@Eq{YAB2f;p%WPB;a_bE@q*7<@=s@O^{;|
zR^_l->xm|!l=XKTM+-yV2OpPpC2Q;JJSt{-`bb#_(YW@g)px$|EuI}wbbYe<EK`Zu
zl$f}}t({vh^d4fZAgpSn9q^c8bl)KKK!NF0$I;GhE1NL%)uy0jUhmHm2TGi#pGM~&
zoOH%Z{{O#>d`l93-7UiW*{FD`d|zBwGaYf^RlyFbUV#i-TeD3ehH1)wRh~}#`LwiD
z4f~^%5J~Y#CbQ`pV-$@QfSTiP;S&Jimy5=)pDXgmjmjWzWW0Yd-pAx_g+3t+Z~Ukk
z>svpDk2;6z-Il0(1Wx!XC5&!5dWmqKTg>Z!g|v+QAizl8=i!Phs|~I|1N{a})B#}5
zH0`w)MnEC+bv`wU>CVs<1ZVIcJ7BQ*lS>(W&0TJ~r#<0LikK`sA5pwp%w59;XHWhY
zE0pd<(Khjk38|Jg1qDoz-Rw)K#*cE8zyu-yo(G<?tU27}q(uA1B^=vRlHL$IrMR1_
zK}f~tLhQt{g9F-ZUlz4PGvQ=oulzywKKp4^HYr?=8G~2`8eu*WUh1lg<a5ZjDJu!a
zAW4ZVQD;@44g^i0KmyJ}hCZzMF}7D%sSk<rSvzEfQDMHL3oJ0Z*m-Wo8{YT|eDy}5
zWv_Ut+$_t+X|;vU<;w_V&KlDMbbnC1D5%$t7l+o|qw&Wc3jCZ}lZ~ze8;(bi{TmO1
zaP6_{%<GNP%F_`q-vFmJXXzGS@}%IQCY3fQ{qB_Xg-6iqzQ**P6N{H7hS7Rt1;pKK
z^QA`}IM&3hqConzufnwht23kci|cnH@2D|App_FdsI*j2$Hz5SSJ(ZK2jH5|PjYou
zF3HSD!O>m2jeNnf&ghxkM1%;S{N)tzxco}c63bsQu=C@)Re=2EEE!nF3*ETR#TK^H
zHG#W!<2Uf>N%*V2rLM7dlyIzQRH;-BwQAa<7FL|c7Y!q%GIH9D>RhPCZbY(QxMBot
z>^D1?MuU~7%FjB-Uu68r7K{WDPCHUvl8(nM#KNDuFGrm>7j$cEn)DQjE|RHAC;2ga
z&{|Ym>w{;$eW$@}lxZ8T8~eCa(o#?Y%X&_b4Y%|FQH5ejtPr}yw1~D29LE%d5Pr@V
zrv11`fG=nKc}p{;8$iW|I=Jz#n<#WV;g3-MY#_qL<^(r+257Vd5(31U>1FQNX9qkU
zjiRI^_+Rv{X|Se8cmB*>^~aMxV^1~6@nE8hl7I_JE%4W$V;DQV;Fl*~&2D`uZ^erD
z>ZlKP2*=XcIvSc5n9J*|KKj`c<r%A(vWaU;G1Y43u34ejaiB%5PC)tMU=K!dnp#0E
zurPUqD&96F=<s>Y?;}$g3<*n|CPox5C|3r8YUcbEUCC`$J8(Z5utkMBy*GH1Nsktz
zg#t&>z%?GEt_@p|n47GtlOg_C-(LAuvk&c&j`dzmM@77uh8MKke#G!eHJ&WXP^Wc}
z%lZE!?k%I@+Lm_V#tFgQT^o0IcX!ti++7+8?(Pr>PSD`)P9V6uyF(!4cJ@B!J^Q@l
z8{fG9?w=lW(5u&~s#$X`X4O+qz3>eizua<UdjZ)Ym>$I87Ey`dbU*DpfrHVA?BS8|
z{1Pbym29B9t7H<VqMHbm9Spjt>p<#JhRVvh?fG@p_X{)kAV?W5Iyr$p(F=#q)70ES
zFhZ6oRtEA;s_4rC_&=|?jkLg#M!%w(9urRs-7fpIYhXW2&xW>ovBFxkj&7UyFe)#I
zSDg;7H2+~L=8<DZ8@Wa0U}4)PwI+L+a-9ZAZN(vhHKZ&dpgg?HEsI-r&wy$d=|E={
zr(wj}MFJg0$c~0@a)H!s*`NPSXk;h4syR}C)KHV<6k(C3*}-Z8T0k1sK^^cEPU~+h
zlotfqE2G3ooYXgY`)YTYPoBBVqDn^Ye!$v!;R>gIl&615Z*!RW0F`NO&xY7eJVsx=
z0Y#N)wN?(C-dz$;@gTby_avsh(VxQNs~M8nmJX(kIwB3w)RN$w5_^%Jof3C&PGPLB
zZLEDmTirdc{7R5{d&t<TDO`r4${K}_h7qrL0v(=q4ad~9Gk%60PgOdQY_`mqr2Ar`
zaXDoIG2<PshxF`z#Ex<oqE2lYceQmXGx{c#x*|p;0v*BBg6ZwvRRGDCiOeo&9&j@j
z^%DBHaQa1*O0yDL<aLK?37!@o-MW7c&c9uOF%KiFILIkCwt>6m?6mra?iMDDI+wKh
z7#xe^NxBFHWbbRuPB_YKjA`CSb47d2Bqm0QnAIn-QF{Og(>DXkbk4J@-cfjtB*8*+
zsEuFVFFD*+d$H9co4-f5cSb=0J~>5j*S6Nzq9Wi6kq;s=!<B9`79%Uao6{+Nl+8ht
zQVMys%5{oYZMK<GC%nqE`=WE&ha8WQbtj?=t)<wEz0~+|y)yw)K9d4MKpzZ=3hMD5
ze})6D<#ICu>nd;+nP9V9a`g}y7dXS1p+r5Msur&nE`mzJk{#h>PNRMF=&RA7mQLB?
zW1Gwl1v?gC7D3guwpM(8!WTq#oz2(Q_hkL)7vIR2U(!ez7_<q5Z7Sl~1(QnTAMmtq
znsNJBxf$rInDmKO7MOlqi(-xtMJV-|p1z`U?g%O8UNeVJ9N{^w+~@aJ%sss4RTiTm
z0|Cm?TSNRM{DSSShdM7L2V?YdV{O|Pk?5wHdssafajZ>Q0+uw-h(G|ryd`V~680TA
z1F_K~Hu|wskEK~MmqL4;RQ3)G3=072&ytNMv_8N7h#YAO!bFB)0(MD4c!2^65f}nW
zSX9)*O>VBXu%C(n*c<h|_UEi(CfM`5@<TSd)F{SSV`;%6L^wk?Cy)^Pon!m4@;S2t
z!;G~T1QxiTZn&WJW`2%BdAz}+oCDnWc;ZJF&x1x6sWDe8oEFAJQ8!0<egpIe?Gn3Z
z`o6P^oS`<`T}#ZeDZ*=jAe@-zkh7=bb%on#Gb@T|;Uf}y0L1<@_<Gx|ctTOl98yKK
z?=?wiUDz?Dhgc<hrA>@}F6!VJgb3_nlf^Dx!|qR_&M{Pwpvfc$`crD40=T2s6!=k+
zT@w`3=}V58lq)_XIBgb$?MV+!1gz|Wx;eEM`8<9Ddd)}>qs%P1)G;zRh_i@ZlQhc}
zjy2g8#Uhn7z}t;rkGHrfB#trLj+AG@0rbGd2`#<zr->yDmuYO(<uO$>4ESGSaqScK
za0K;PT?hm`olbpIV}pmi3(Y`c2n=DC<V{p?LYOw&Uyur-n)5@(0bXOaf!-mFO>uoW
z?a%7fEp?}iz!Qm(0v@=kc71KOT@skEQp91Yx(j+UH6()1DAoB3%Gq_Tp#C}ExJRhY
zLyX%vlJ=5@t_f$q!p4^`E1hONq2yK1U{GNg?O6};=KaKmZJsld5o_4{aC$8X15LEW
zm}_Da+e;L-PQL6C(EHV@URC;Np+KW~lA`g01+fZ~y7_8F4t?bQ57%%laDnyu@5HT1
zWQ*4Q*7ktE0m%g*fve%ZYu>jSZu=KfzpF#RaorD|PfQ-?u_0LtpK8ShD}+#i2_VM|
zFiPXaQZ^p5Vd6eEPEk6;tYzH?SfPc*ERz+@PbV<P8g6;;Ag|}uncV$_^W{E+k4nA6
zD^_%?L<ld24}zAUHa;MUbC6Z;jo_PW?13M?ds3UjwsL5bSx+3oIT=xUM*mWP4vs){
zTywM&Qzicw(KhVpF4>~6*MiT`J<fDvhR+0|pDzf~fT3Osdh|G#XRHlNo#fJ1lM<nk
zi*_5o0jweA&$8ziMEjnR8K>=W2pSxp`wY~|xW29-vAqN2wu*`_$3P#iiV0<>)qEkM
zmi6t|T;8>Ab*hPr%;mBR&@|wNcpOoE<?}v)QOP{;4*8%nZzY~){;cVKnrQu5msBFv
z${#kASGvso_F<wJT(8pR2aZr|-wF4<BP76=TBPqI+J2>M9LCkwExxWXXxh`*t?nbh
zcqjiEjpuG@RINI0M)=F8_1@F9v_?G+Rf;+VKMF!<E|ouq*Yj%3=<I3ncTUviq%T$Y
z0ih5MfewlUbqp;gI0jmLy=r`O{CKsP*4ibUIKyiLtdO!7rD#$YAqAMls~{H$toz8P
zc{^Oh6S!RG@pVa&QUM!}PIDzf-TVP}Tp`sRdsO{vhVsOWH13h6&0Zc{R={GL<Q)XM
zR$h=ufy9Ggs9zgwM@K)ic<Qxwba8!G|M>~)3igGDT`xDIfIG1Za%+s~&jO*TO^a)W
zTZSED*c%l^Esj<sx<yMHSP2b9H8j=(NWd>u7Vl?x1uK}qG$&x)1D0u!(FSEmvF2A_
ztqj_vJ>DjO>FY+fwLC0DkOhYy(<IS4JLaW}s>jor+G9*YE-NI>Y-?xU652R^02$m&
zs<y<a4HOA}10n7!%6SJ-c19QVTWx$!o0-*)5r5BMb@rwf?-iYjv&K3X*;J5x5pF5k
zk@Mb?U#FEVN&khNP>jUjVASM!T8SpWuGJeE0X{>d5Nb)fG*Gv*m*DV`C)!W#5Us70
zQ`5&DNAt!Yl849z9%I7C{l!F!CviC2CJ%L|m5G1b@7j0FK~)A5G6;dZx&0=rGnqHz
zT5tY`0Ka`i^;*2k@H?P0_(g+Oi-}kQ3Q$FQft7v*Zd*waiW-HGRXOEx$CjF;`izy3
zW^JO|`xXkqt)mg=mTUr85bN@McPz0k5-cM*^xD2^CZnf1qN)21qFr7@+@p|zS*d2<
zG!z|vS>dHI#>{^azQxQT*VJ@TK|W+1mRvgHQ(z3lV<T`Zx?iv?iLE(>(Hc@Ecl}}s
z?-I^~=|Lpo>#<h8YJR$SP0!$-<+QNN>GI17UvriwW+~!<4$gc`nK9yFdTPxPab)=y
zzAFOf^<k{tg4folb}AW0#JkY~(f}tnB6!ca_lmIg16duKi|FjqnjfcOKIW7vtRpeN
z5*x!yfWbuY_eghZRqZ?Z=)h}@+xeR_(J^whR&_juP#E~g@RMMWgDDT-H{f_h_Ax_#
zqCF@m%H@^)5DWf`S{44DMll_z<nA&0b0U+_MqEj1G3VjT_3hfHdP|3j2yJQ$142Hp
zJlebIN3@_(e1cHNV_bbM2xy~b#a@6f2)BUm4G(Qc8ZqRUJws66>N)8vEG-~@?huu=
z@l`k`1E@;j(-iS#qDn?WMy-kITZBt^3;ywJv$8CTS@E0FFB-NX`m@jeTG0zTf^n3n
z5MU{PcEjV#S_x=PJVj=hWz2t)w8alMQCs4*L%>6i^q(6uo<bvv$naGpWar*hMw_m$
zFxiR9+z@JuHCieQkGzgfkxu!fTqiC%=l_}Um7(Zc^-ef?UPq0~dgbC#Zv7ZF75P47
z+(6L*3FTwZoJYz+=JA>B+Bb^!do2gcWc1+C2sc*GHNjebFZ%()$_PSwPpkQ_A6-f+
zss?1rBK0%P;Ne(K%V5KO?%yeQd<UN*ZS|y^TLU<gSp9x+Ih?L=1Le4(;##9f1^R33
zW`>>4mYJXq+&S%xSzFz0_9awV5ag;NGTQexy6Qwk1TFbCG38d`i9bd;`5ct*!hJ<C
zw>oYKE+jR9f{>6<a6$t+3IHCXxJlDwGya>`isi>E<c|wZ&e;1{-aY+a>Fz;%9>_`g
zsu|agg5x(QgaKoDb+Me0KNv&83O<h+4M@B{Ts1R2WoX7si~2?3eCndz%IU&kOIcK_
z;73lVz(<>1i&F1*Wfb+JlyQeW{Qg3(@ZogO&K+!sf$5h)v<u=TJSO-LkG!2!*?zIq
zdu_!5#lA(&NMN|9Il$X{%=pKAiC8tePNh6`q=uiX+nkLXZwlLv^PaHq1;~@3?g|+7
zBWqDeFxEHC!ZV;^a-!Z$MVEE6!|)Ns?&YV9>WcAJRo2{Mt6wW)R)xiC2@MM>A{cHl
zLTYwXwHU$UY7mG*v67uLPU_-9dPRVsxJs&mlp%XSwzzyVm3$x|vmIs{&oX%qyup`$
zB!Sho&w4`O)bWK4Y4PjQKQHedit!}>m}2H9I&?DEq6ZqIJ`NUt(M08OM~BM@*9~rP
zO>&-P9-O5nbBtF}6?~85V4n;<Rq+YMEtKi!^BW6W^{o);HHKjKaIHSV7_h@@YrX%`
zzKis<27Ija0l<a|jZC^$lbP}?kYbN%QaQan^F2m!Dd$oppb_B$0?Hx)e#w~8Rms6Q
zQsYE{l#Dm{goz`<4xMTypMPk0PKd*$tCDPG3#isi@_g#IF#u)MS5*TY-dX6yqaxN#
zlSaV(s)K+)4YJ!sc6JRH>Q^n-#2mGE%cc=26bY~h<pVj^zF3-5e{Xo-#h|;5sVm4u
zDYu8NiXm)=P;TN1JDi^3UT{8JRei9yoQ`=}+sxSF@2#z>?GikN?Uk2`C2%KlmMQMb
zGSQ8f%QmvEJF|Y3^>gNg3SCp5MFJUa$+zShAR+^B*y2Bjv>hI`%cW=AD4W+8`vTZI
zm!pyyzC{P!15&VgEu*?+2pCKTKi>Lil=IRC!@zhC^sXyck)55huaTE`6>U!v1t^ry
zHHx-iIikKEFssf;Y?wh5q}u$Xmv~D`OQ)`J!KzUC5@{G^zKtisVprg>`XExa!yAv+
zk|KYZX>WasKc~YZ5|?fb$Dc_1Qtt8qW?U$_T#yw&-+{WIFnzbYQW(W$zXqIKRy`6`
z3AqOeG(i$kzdw4E{3tD-LiN#TKXI|ktzOcQm9$KBA%@@r>QwM2*Yt|{Q*0mh1Yw=7
zekm}Ka~2(UfXKQBYEP(^>!~t(FZEf8{gat3o4KVQFlPY;;~GBrW7u#?Zj9+&>MIn*
zj$VgxZNCdcR&{T#yjq*i2n!wJRw)uD7l}#_IV9Ox2J;tjX<ZylXM8ILLL-K3$;Ic!
z?9VoB6kX9?ozY+2HcOi<d3Y_=KOM-Kap4+v2__g)G6kJJ7F=D5WFG<H?A?!{#NSWy
z3=U*x58Y#H%+G&o8YlfA-w5?SIOq-@1ov6o%;A@K)v9r1tA}UOCihVNnIlGdi3`II
z!65j;@2SiSM7aJx983D2EJKbS#cd^4Ss9CcOT-SzT-r$6%#^WMQD>sNwuI{W<Pkpf
z`_+&;raFD~>uTyf-q2odu+XL<Qz9qN1o?!8npmao2AG@nX}9vDn&|u4%;D=NV)jZH
z6had4atR$F_m}xv_D8-dH+)o#p!if3v9hl;O|PB<DnNq*;2!Lgh@igVCHK<WFkkV@
z-I#>+FD++kDD-MXh*yY`0)S@ewfASj?PR|JMJ2+n_5)Pvo#<7I08yJXs|#Yf6c)#T
z0>fWfK_h9cc|7eYOQi|(3}-`Ez2tnPIm71#OfQ#X7IvvF**GKs4GrIIT>X-|RvC8=
zfdX<O_f7|GJ*R9aQ|sJH1G!wB<H}vMUV;>4^-I4Vq4v}~e)DI}283)x!<|Wkc_vSU
zVE#j0c2qet04zvCM7pJ?_`o|K%ga>@kF&`<N*D4yHS$q*Go8d2RuD9T|7}FresZ*c
z?D)||t%r^cLz*`EH3H*07}p+&K!wKb1bGV|*M1`IjZ8lO8-TK=y6Z>W{r+wM`C$D<
zU<^WJi`<OFmp67*biZ4c`GDuXy=pg9Nu|o80&8s(_cI0>ec&>x;PvZ9f#8<kj1{5S
z08f!(_`!p_r=sI(MbQUF0xxP&aMg6W=o+7GO}!iPhk-mwtIYR^*h3Xb`9%8L3JpUW
zta;1_#!up<>5fg7Eg_QDFR`z@4Yo%69t^!pyiQqWbVU=QfCCc7v10bM{9xTCo$t>$
zt_kx-9nZZXS#v!<d7JA=9(}_{zxHhY<YctlCEwv|{+eiSZL9G{M=5Oq%qiJqZkV7d
zit%2*MES}J#xH)sRH_n9%|uMVbqlO8;dVKd$_OmFH&XiXC{Q={5v$z&gIRV`30Xbz
z7&)o92Uin-oY-Y`XhKk7H!NKf=?50Emk_^NdiS3!3)BkV;wofnwsY$n|1h6|EL8C=
zT2vAj9A=MO{n`de_jB7!fMLv7O^%V#wKnyTZdLpQW=h`&(``V;g_0;La=>;0@y005
z{M9VO=#y;P9yg~_;yAI8=&shGj9gPzN&&dsyVL)e>`EA~Q~cei3u0u{dCiS9GGSG4
zd87uNNTC{z_?*BU@e*1$yyngfkN9mKv#Fx4x&|_qDAwvz#gg?9>jT9-fFHK8)qkR;
z!Q$oUWRu@yJxO^j7~qyBfh|*d?Tu2HT@cpe-Ze|sYZtIm1Tt4QH8}nXYW6V>(`5x_
zv*f362(=^0$-nZGs~@FIfl|_7o5@2l*r26{08c?Am7C2vL~BmA83I#YZu_~DF7Rz0
z8QRny*-w_QY`28J0UfHuz59%rm#Fox2$lGQa}CAr=|&8NRt>ts5JhfLf*R-%j{sEU
zPH2%2#`xuzYK|Zacei-P5MsZqixp00wzV7(B&o5`N`?baWySej6WIXQ6MysGS5>pl
zVLNqNrb<@={k;=NLGBuBkr8un$vNu<g}PZLwX$StT_;&@`^nx&R%{wA{7ptN8WKuM
z+0|$_IwKq6a~$?tNoFRqw&sV6nd=n3&LHyiTG$XXzT0#mVgj7Nv^snv1fk)nbk^@S
z=fU>9WkJ%-84(Z1r}Vf=;u%*Jk8FmUD310(dnO#+wu$5wkP~q>?VBk`T<Y0G`zd*V
zAZIlse#zqNjMPIosbX}m1c#&qf<GbDxDQF6NRsQgliyb+EQ9rxS*F|}#l%w6TNlX#
zRcKI@H6pS2Twt(Bu~Tg|Qg3t(fnOI1O=st6^Y{<g0i`;&;~a!a?ss+vq><y*>TpgZ
zO56PZO^Wk>osRsI=CXnOn^Xt3vDgLseyVcq_6a$O4d{X^=b#Q7E=JUbI6GM9)4>*$
zi^T>Rf^D%U6YOAmLQYODWJXT0U}**}s^t7P?FbqK#o1=aIom{Hm#VR^eY{a&c>nvc
z+Y}gy<}P*2-p`kmxIv<QEfT}nBiN%=O!oEk^J-TR_emgAS>bgu5x{%qnLo^#)pcl)
zG{S(s)?dL(U5-h+F=yE9pO)EG!sK%{b>t@q7+OsTh0C@l=_*bH+yyihw)@lJ4t4uV
zE?l!GEVKS84p^{)2+$l@%YPvZsr1RW?*qhBr!673`}Kk1`Tcz4_`2+2pV%*3{q{>p
z|1J@Q{o5o^s%3wYX|aD&xCq#ueO$V+w*s-b^Iu_3IX*!O)$l;XWW0R@iHi|^Fe2%F
z@H&vOD_nTkr0jtH1f9E72DL>k*z<|jn){Rz7n~R(0Va%5x?t{P#;5{VJZXQ9r6WE$
z^oONX(`WHJJ!?Cpcx=yt0IKR4e)q?W*AS2Zlj*QNasZU-pQI50|F2XdSQCjqlNKeE
z#ETJQ3}bJy49OH2uPCIRBl|UF9bBzd-9v*->?}94B1?12X#<(*0V}ErPwo)l&}C+L
zxlWvY3TA$5`lZwS)*$_gF-45xg;%MtaIl}a!mHRSYMwG)C623928cLXz1)9-?*Fgk
zwFGpEP%KHcp7O=oZvd({&Zzj&%+LKVZ|C2y;esark+e-v{?Ym1HcVK-Med8aoM|3%
zPv5ga>V_<W#;n4;aDG_s@KcC)-0*6w(_nl@YCtIvaVkU}59uX1Q1J0`Q7_1K>;}_W
zz-H>X56BejYu?KDF#;kE2I+`<)^!(URAs{1XCkN`p;@Z8)c}o30(t=AhR7{Li9Mjg
z<@-q^IGBew+jBXOP4i1H{XwK-092V}bUn^9*5Oyb6HQ#;#W|@r4}3?Y34qHu<O=sM
zspif(;L#(L9WY#~^n1jyyiScbfSqSWGy(-~Ec68Q290=YcTO11S>C0<@1`+lF9!|k
zV~Rg8a=!c$c>CVuoOSI!3^l}t&_!>7&C9u(g6=YivwevWQj`)2vI`bHUi^~gvbd1(
zh|6Of&)r(Dniad2II5{{b|j}1rf!HRw1goc1KZgoKJ@y+sQ5=llHql~jrm5kg2Xuu
z?g`gy@vhI^O;!yEaJUCVYW>VR`=9iw$AOv~B`~t%HxcJ-uI@xB{6}z3)Up!Wj%bF5
zrxwR;W+5UaO<Jw2GJTGy#b@S@wXaF6TT><)ULB8^MdRmeb{_|=J7gZj%xQyrz*HsR
z{}sfICfy$tz!<W2FhA}Nn2dN}DzFKrKk8N7vwtW>{+mTeT|=W($5wf0$0wl&lf`^l
zpM>xqb=)?)I&F5?K;-WMYh^%_CDa|*eU|WAp9y4F#WW|`DZ%ujoOPp+@Z-iNn_nWB
zT2S$-WFmsomjyS+3vd)>5_8R3a=$2D5I&LcaQbLiS7fX5JX!E$w6oyUqBPlP?Rma)
z-sQK@b|B=ml#4oF#@(R{&H)L&-y;`Zg|vw54_~adf08*BbRMJ+y}2jE{j3$;LvBHW
zk%9&-UNe8%O6JTtBpqq8#IAYaqG7GcgY+xg)hk{d3KJ2A6~qTS(A((^jVw0SWuMVG
zuu!U)$8GM0%`~<w&Vd{&$M@^=PmfoA&x|+}3MD;hC0cgvXHBzHh(z}JUQPUXX0><U
zpccDAAx&Cdygw?uKrq^eq*;0<6h>`D1sh`q6Uo_oqFB6hp@hn!O0_hW;dI9DXzGjm
zDkrh%el_&iDzaOv?2lkd<hzkf)z^&U6_3NlMBF*Tz)X*3mKwrooS$pa+E^di@d73t
zd{<9}5xaoLX^dpQz9Xq)4=;8!f|IF6GeLMj>YeevRc>qc)l$^xB00hvW@g&NT)3>!
z0NI5Q$}B(?fDUGma9eT@;!z|x4Q2hQy)Y$bPCg3N!h_QiBu7#3O8G8mXIYq#*5Kfw
zb^aIc22;7M?gE%5Xqhcr@&L(qU3^JsxoT?VHd5f-nC>2q<BxAcguD(7CE<|orxij9
z)JxUkem-mj!|cA>*--ot&zQNNf_0Y>@s0f(QQ_7&6AY9!?+8TK=fy86ws!jZi%qpw
zr4h?xlDZT};u$4Kpd<efp6cDOlhQk@H<dv8dU4G8k8RH#1~}Uw?i!SI{k_xG0JWoW
z$d<A#s*n?d1`21Blk)b3ea(@NWnTz)B)$PiAqkLt03!Kj&LL4b+9QrUh9TwA0&b~|
z$*xO$?5;<zv601GWcw3PwP57K9`9tBuYccI6agILdk3+#CA_wrqAk>RbOIwW6BZlj
zbBy>71w?1Uq}6BfA^EWv;?fVgdZlD|I7Y_`Vwe&e>nG+3JJ^!$cbW~IM$0{`94eFG
zn8iCxUb#9EL~>#?)#(^y*@4}qrZFDV{?b>a<_WtC2Lmk}1uazUNMKoVKYIEbDoJxy
zf#4*1`H;TF+^fo%ovWCa1GZH?LNq5Bd2$Q^ol^jSB<p+@IUtO>d84zGy+(YJ!icSJ
z^qEOC7SKISl=ws!E~0k&z0C7IiIXwt>oKkl5e4}N`(y)k)Duxw5%<Me*Vwx?ZdKzi
z#8N%3h^fpWX3^U&(l9V-5G*L6ru!!%XAn@yigD$uU)*P*=UD+tB1A!o)t|eC8tR=7
z5sF4y?oyj{KW1(GsCydJ)2y%1K_{0WJwgu1yZQ}qG-|lSnPPK)(j4&d&6YGU*a_Y)
z^q?fx#H7XaRv;(I!6&Zeg+s^KvcgN&SyVa9#CCLe?oTVAsy;*sy9#Zi#TA%5pg3iC
z^=lZd;BmTMU|@<kswhFmvR1N8a-et-M-_BG++P6GFcEB>WSoAhVz>?<+t{RL__otF
z<D@Ee@_5Zb3QPNl<nb^j{N5?&d(;woI<|mxw>53mHvXPo<roaFlF-f2g@3%FV*xa>
zlEf~0hubLr)w}Hb&0t-~haxFkG8&eI_jBVeAa`!&hHl;HHv^7-X4b<ehrpD*%BpLw
zhZu$*8l%;j;ZnTdnix@to6RrCI%5aEsqOvNursMpFZ-fqR97+GL)QJDxlNDI8QZj8
zq6y8~`e#NNj+|e;>#cMeOQ9EB>h9Nm5(jl}Jxq1cE$*$cZOzWfXX@!x>w`NaAl!?)
z@W^5ie%{X5zVG<?Vq)0xd0>^0v07*$E2Rm?8!D;MdcH3Y58fUSkt059f^pHzKTe=;
zewT@iTkmN1Y8>ThR1-CmfhnCL(MOU4AcAXp-7_9W<keTAcDi>Ssm_k=FDx31>H$p1
zW%}Rw_?D2c5{(n>q_yPZ@R-Js1=^fvX7kf0IyFt4$*H2UWliIXL9e0%zJ5$_?&lj9
z^=mqa3JcntJjVyWF9}Zo$WyM?o~wZlUj{#^>6y3OV5Q?k8}W$@qr~cf)N^wB$6yzo
z0n?6CupX+J*YC#jn7eynqNwZ^zH)EvAqyO^#()X>R}C$37}Q>(2rWHBI7ZH_t;?3s
zW{b25F?9P%QKwyfVEeB~bVIj)8Kz;}fkv{8jCeX>b*^ZW8+#yBMGzs?@(m{dkOa3u
zBS$&zy{vvkaU4tywtgo+bG~b9G{#h7Lnqpa{b$>X@gC?X2EWp7X>ppO!FWI7dW6{}
zQU~!W7?n1b2=55w@C7&;@U|gb`n|<7X6CTW8DZvrx0#l5zio#Eu2}rDTz-KYieo@B
zP4c{D2UTak3)CSuj=cT~_1H{h5nO-y4(vlF)CtIaaTzKx*{F9Wlf~tj>x7q&4o|Ew
ze?dhA+FN`@@Qd;gx*Ga(eWoxbBWv3&DIiwU1(Z~g1MN|e0Z<@Y;;f+y-U_pN8><m6
zWs<nL|HRB%$v}99jX9Ayz(o!agof&Fg6<62o2oyvZ|onF{ua$Z4PkNnYRX}X6fbhQ
z`R)4@i|~XCn?GiS!;dyU>=QvUn5`-yvTg^H>Fk!UZzPm!Q^yTVSS$rovHAKdIyEN2
zT8ua>e&ogKtuh(@D#)K-1xa0$Xo#Oi6L=g4Rvvf_-$dg~<yoK#F3S2cm5!v#fKCvo
z(}9(%)WpP1velHmT+jH+^a1L4^i*-`L}u?o<?EP5r`<N|VF{u-MEt`(VcFAL5Um?6
zp=Dd7ZIa2Z9TK@8k*qRXZqmBi`ka=(x8!U5sj_{x_?T2(=5)>$@8Aj%64*l$I=@r9
z&n;j#bZtYJyN_XukgBkEgnTE$+ZtL-obwY2@GO!`?wniAoz<|@zkyS~L$qwIZ6QxZ
z#N?#9jd4a-QV`xoPO&*Vrq`l$N?Ndk{h@!c6=PM){XE%YaCaACK{|)?GYJc_2}#2$
zq(f><H>+>VvTB2EoX%>}%=i9^c9lon^wO??+_o2f#N9Mcchy++&W7?`D^dIOdLELL
zsK~4eClarmPt3G%fr4&H!l-O7?(u!xU-NlB+_v8-oO2*k0Shd~$khR!ND+5ITm!pr
za*K2m0SVYo*@#_ZNjbm-!%@BbP4u5MU7=`bNZCKEctma<33jaX`&a1~n`7t`BzI|0
z6sPJC{Ot-Ig|?r3P2_Yvq&889>@L?)PsohcE|A4MxoI6mZ^lXyTnSe>Z1fGn39Bp0
zeU_?q%GO89K7;zFOKBg7joLFdzvRBF=FxQ7YjRwM@qu%q<5EG87UUc$ur2qyr|5Nc
z@0)r*!z`Jo$KN#8^)6Y|uYic}n%Cz^a#0a*TK!t!qDMpc6cVv7fHRCb1#EbE(N@rw
zKZI=p21oFbO~9~Vwmtg9H<#uexC)LqHf)gZdciQxwjn$==JR*EOpD;~!kr|PfXM^8
zmm!@Z-|s!8{sthnSY}s+*}7V<Zox<y;|F{|LMG7>Rhf=^AWrV4?EEHvH0B#xMVK^v
zAffqGSVmgP6XfCG@bb`;VdlIV>CL!mU>gox);H2NtHCSrtprX!Z8^&n8kqy&gCso2
zg&9&>WH#y;oF(a8aauKP#v`Yl&P_BWphUk7tUw%ZmN`}-hG}OA44ofka1ArqCx3II
z1#Xx5#`KqV1%5F8V4CKQah-HBRl~Ji>A}3#B)Q+3?35*qo@#c~bsRE5WlU<Ba?s)5
zIjFn;;P$OttJ!HU7Nr=v4aJ?67(AUPEW8kO9Y#|*SG!QPqk8O;JCl+uKzUp_byvdo
zbK*5QU#>phpQE5PAd~1B3A9^kNq`}JcWdLArO3q?-j67Kj=$)H)tUqE6W&;AsSp;$
zezsN?VIt5MN<x2L;!+_BL@eDC5OY@?WUxh4OhVvpv1>cwVGTvjNcI6B_peS#@P?6P
zn-B`w4@mHYxLoW<5>#)0jsc8}5~K*Ql~(saLM(P0{TTH8aJxw4W%yxy^g5-oW&hWP
zi>vLEi4vFuVjy8?X5_p(#lWJ|+>G43;p<oRgqY|Ww-Pc0az}@Ijtu8?YWq7oc``VY
zgJ<m`wU9vs0yK;xOw?Oq<L#i4XQ}OB)OD*xUROB81aM^=t65?)OGJesN{9Ueah^I%
z<W)5J@R<!$c*HW)q6n2>_J?p7{Wb^`uVUr{_cK9X8JqS9k8gNs&SAS^*|x9Alzi|)
zP6B1$XjE7q1ivp91g=9R7#QwgR1>N$>VOX<{q6Oli4k?7;6_8zR!H7=>A6z^&bRo!
zF;0O*rHuXU^w1ajf#36Tlw7M!S#FWRJ&B-w;Jz$=KXN5TY$}~n`|jY7{<NpkXmi!W
zxCjiHgBSVvd5<_37LH;^beW*?4S4N*fIduCTTt-DEV(qih6l|PSpJ#LyKx!-_XO*<
z3p2vfMc=X=8z%qJRn}Fbs7c9?zF84z4jQn1I2{NwQM%@-usam0aM}$fA>6i$^+gqJ
z+@60N8`yZtk?!Wf<num~UgqW1*1ff7x7mzx+U*f3dpgO_l;praK)WZtqVNle(zT5v
z7wUJa`WTyek|5iEztM&X0{onjtr0}YNyWgx{e+SJhkW7=Z^!^HY^;1ZrZvIxnNF%^
z+W=u}uZyCVh#*mQ|1o-h|A)*PMO%IS4jcx4hu?M|;<>tFn+>6Yh8z{QSa^cSz%Ryb
zEymw~su}Qc<2F3c#coG)KDC-LpZwQtxa_y_AF{1jJQ-*-<ax)QOMD&)`8g<Ne@96m
z$vGj=1cJ^aL}dRF5J>d<XRyGFaO1D|#lJqVHpo@^N3Zl(w1`Oth0H46d4=sWvql#z
zJwwMhd8|1MSeIXK4b^Gl)+%!C)a4TlpabrSbGgfTB;aKgm?Jv4Cz;P{xiq=m9#3EO
zCh}^wa_isBpbm(nwCfe9$C0;7QEwTxk4Q*Irh#%v?d1eGCeR9ng8kqIA0jz%s99-^
z436~aUmL8Wn;hUNHL>LDbGm|2dD_O(X}6_F$AuB5@LRFJs%mTyI7my;f?%Tx+%S%O
zB0Nv2_!X=Ud3WHD*@x^gq#Cm&u{|U;0IFn~f+>LR+zktzUqmf7y(<o~aahLiug727
zKWWH>oRP(jzoN^KiF|@g0TctK0Uq$ENl(gXEuOiX>n3&;4i)uS5MyT>HCvhWZ&nbu
z$VI;a1&V$R;eK+h&JmVVv!C0$i&b5DDtSA&hUkVZAtS<CqlmbKKx`fns9ks=ho{bw
z0+&EbrP0=txQVYBUnj3N-o`|aFX8XeSlmv_^S2h>yMAI$&s*n!$8NS2vfrkKJCIoq
z^@m%g%m~80K7<(Y3X+!lPN3PwjJNXTOv~UWbd*~5=Hj8X1fN`E3EH3%jo!?{2c&Xa
zYa2tf)znNT8o{6up^NC}0%5CW8m}M}t0gU0<w5epCw7}v3Vstr;#!~PP<)6bT%-rM
zZTkBq56h_451AVt+vPv%3R#tW`d>yepk8$m1pTK0$flRPP<-IMcGsPNM<%MUTq0ay
zyFR1v52Jg~fFB0RC~b>JKcC%j*lA+6IKRyImGD6q95cc(IAsiDhdXo&ajqR~;|TLi
zDVlio4o9=OzMqXZeVL#>J}A`PY58d%lrqKXpFCG1gi)KLeK+q~2fGI<w=TieayXMU
z^)i$Aq=WT}dWBiE<SOJJ6FAf``2YZKX!!nMKi;+h5{u^Mucws;o4mkYZ371j{FDO0
z@TTDab;lSl#wnXiWUYb-!bWacLD!u!=pT|7WG6zSlmqa~4A^@)KE@oF)6Y3~#b$8V
zQ2B#Umj+!~w|NU&i%4F_wwbh9!W^w-vRNNspmH>5!+PY8XFHO}VWGIx>EvQG`+w9J
z&2!3_NM47QSLFTLIpYUIyA$*1`*W&qBYy+NW2&Rli4^(n@$iaWHt06qJwRF_LR5o<
znMINfM1q_r<fC8d%pS1^mZsN#WLjdHT(1VnN)<S7#25>r=$Z{Zf$<nP+gNWr$Vq|z
zZrPvKswu`K34w;z(7u?07Igx1@T1*GcI#K~7+_W+ni}vEtj!@>lgU5N@r64r!Ra>k
zN$HL&_#P0h6qXlRF{)WFOH~V~^8z5~LVK5>^xp8m6!LJPuW;>y4f5_cP^69MNNWvc
z3Kc=2<xB+`Z<hymBm+`*S0us9$WY14dKOUyZT8u*{?g<H&h*YIN)+HsRJyYD=mM!o
z6$>$y$+MK2ol6mMlVd_V4&ADdeySG8{*KkX0pM7x97aeuuJ?<?;p$Vwq<pi(oGuHn
z=UonrTDISMhzTTL@|j5SUIx778lw>5vx=yu?GFcF`Udnonf&Jm5N>zqmmlGT<v_jN
z?YJFBm<me=c}v3x%mHQ*)?~m^9Efth{Gh<E5CM`BCbY1W)-}@C(a`-Yah?mwVK%lE
zS#O&%z`C|dvO)?5phWX$4%eUx?vo@$f@!ZVfLEE5jY_zENk4;5i`uQ32@!zoU^aj8
z38>QX@|1$DU-QiG>rWOPQESIsT&lKz6gX&fn5T%qLkfx3ot`J-N+5hkqBYuiLPBDU
zRHq+On3Xh9SQ~@_xKBTSuR1sAMEH`{Pq4rjkhRYiyO|*)nRGg#!s4&)amtVgiP7)n
zn@TspZ$!G+$nkU?uj64+Q>{I49U~|yi3ot`%3nUFCBziQ(YxBWxH?E_!ivvn1=6@R
zhgQPy4wP`eicrk)AAJC7C{&bKOcBua$dR2uvErtYDTk`3<;B)ES!g1cQPW_&3=?H|
z2>SL!Dc|`x<-_Tk#ePM9|DY_k%pC@SIk@pab_%AOc(Jld8;(WJtNr*PdZ$W1QOr+&
zryUDM4Diw?25@5uuSzXaA?qlDovL%7?wGp5zHhQ<a?oI)&-=RVJGc<1?jQ8ZUdh?l
zk~)Bl)mhcSDE&RoF3~AkYs}@#IwUS!f|9V#I<)6`vP`H*Ec+|+A=!9~uhl>VUgP^-
z-J@KQKNK=3Zo>Sv->b%9THAIWz(i`~Q2I*ANo}mD0A7jx88>5I@QRf*ppSUvn{s*Y
zhm<NgDJ53Mr&+W}{4SgMfK<qIF7WOht?4_9QS%X4qwS<eSk}1O24<iH1W`)RX#x1e
z`=Jn;KhvbO69r>90v8=?fjVFsZ%JD~zYoggRGO}<=T)$w2x@!5gHWY$yunqAA+=b!
z506XwQ3Z|VBPzrRDGbWj0}ASlAc$SBa9kdh!3*Zdy2AEiwlUh?)n>?~Sio*4ZkNv`
ze-#gw%lDv#rAiujojs3>-j~L>KqO9Vj1&a`#U6mfF1HZccyJB!l6CuBl1j~*=u}!^
zUZnyE%?$K8n2x#F#eV0WMHYqrnFE>AlUC4q2MFG?z}Ss@;?6IrsAPV_(3A8$O(niD
z>a%;4Zaq~KE4@xKlM2RDq!At@AIhdS@UKqUN&@qiEP8dAZYakrBr|^lK78NKfNn!w
z0>Fdpd{14>BT(O0BU#$~BBp0?Dg1s|axqdLiUYA7Bx;W9|F#}V^*y$QXu&mfQ9~UK
zhNwAUxp%ArB(g`dFddV$18tcbIA+r2X{;J!u8Ra!vPTAgABcENk}h=!ab&WoGB<Li
z4Dju@Luw=nKB_v?QXMM@d*uk-?Gn2jp}HXz%Ta|gTXWui(DO%2?#9mT3t1}RL7{$|
z4TP>=-^M{{S&!QvDT$b8snV@D!~!5A;3|*;j9E^IEc>a!zz~F;MZ<cEp+Sx4lrS`8
z^%M_v0O$;wTBCT6789xfw!J3~qxV97*xi*(5hV*r4iWT5>iX2N#kYUAm@^!XSiN@5
z&fBVS-1N1plmd1NzGO<113*r7fuA%WvMXl5veQS%=GyP^ji*8DldKs2bhWVc6>n${
zVcVmEAXJ7gbjMx?=LjSIRVVc^H-@=`s_Jh5TMxKPiY3JUuZse~FP*3qK5O{f%|uM>
zSnv@V?d7tC(*J&W>=E&QuRtW;X^+k1(rZNnXIbwE3Y=7+cU*4Hpt}siWM;~+jDY6f
z>!*k|XSA@`JN>3d3N=y2z^$-ap5?;S`}i43L+*HH3@L|xrecqC>-&u+_DKS!BfL?J
z>UASf1s`YzFMzclVl2<B%Z}yhZI<|~w=`9kP({4uj3<}Eegp32rkn}FdD-y%kzGN4
z=xOfuQk_;PC?nF0uOOs7vbwi-OcL<D)x!CQ8B>b7x!5J;1X{uK*<93Ek}VP7$kOwf
z`|L8s;}QzN*C#s%;;Jx#`Dm>TxS`aZOz8pUIc3AhPS^>)b6yfu3b^x@)Y$3W#-)2A
ztXZx5j~oG$yYYxUb#ef5br1Brg`%J}-lkr?O~E$V5N=P$Pg{$%L<Q4rsE{dsFVC_<
zq|yS)eN1dsH8AyG6*0Jr8`w+D3PGxL&`@P?`3|#VF~r;Xgv>YhG7*|Wm68=S;LRO~
zDPN2J3@3U@>4l3n#2&Jyd0oUXPM8R&{ARU|eE<Z}V4y-}Ti%>=wd{xM9p-S=7Zj7z
z`0b6pL#W94G@}G57$B0q6f}IgMiSE+C$I=;d;Wy4mP^jxCTe<ZKq?9WxO$D9=ANxC
z9@`A(vf8P}@HWU2W>?h)PO1~BK&4xPcPU{yngMw9=I~d!UKhK{XI|BKeeMyX%pV@#
z)AuzCNc<5M?RpHs428>zJ=x8E2<{<kybcRnBxTQS$0WHdDku};$nAjp<Na3t0r-4$
z?N|o}AnqO_>&eQ36|In0QeS<IjLJfk@g&;uf_DGW>nW*i9(M;2NvolSfbT;@`GgxH
zK%j)k9q9}p{_6+0nU0k?$eBiPTac0HvkpZPYlF-QFC4&(i1Go@bbsFliuBK0o^cO#
z{t#I?yStpl5kqUcjy66VlZHXu-%p%Hb8z{@{UCcnD!v>5m$lm=qb{`l)BXF@uW~&d
zEkW%*Ajr@qr{w;K<R{|uV6ENpkg?@JP94H#zPgErJ13I%icDDZC9P1Qa37m_1fje3
z(QQKepynu!SAsLg1DTz+rS=*f$cJq5rK8*GexIg|$}YXwqPX8JTNe-gRZf;9=aah8
zL-+(Y_YSmU#MUCmtMNpAxP=J4eWHU@2{4;C6hyenoc>YIGwhljeg|9LJ(Q#5TlFZS
zWD3RpAggsPAe^hCzwF_XdMRiCncV3&fTp*_KCv>T#qM~%NosxBfYc%CZQ^xk|8=8i
z$)A$skT?9l)2_q=5{O|W@$~5y2jx$tNV61{2J!vP*QwlO&}+ri{bqcN-UpHq2I_{z
z2J4!HSY|1M*BH>L3KAe^MKSo7C`@QB!%0{rlKBqZb>CjQiV>pL0R%|$%<(ak7x5{J
z_sN?nVGw@7QOMnPMP_(vN1(i(K7yqb0JLvh!N=bKYH;=6ng*A3Tb!z@#+W)L0uaHx
z2*NK^T#Ot*F%9>_9!z9b?><)6pg`z_G`1}QgK?s|6p=0_FgPROdB}&_*6HZhxuR9d
z7<`~|paV}p7st<&!jCC!MkEj_=Eyy9V^s!>;`)dhh0%Ys6f|a}%1rA68!85;)&i>+
zIR}+mhDCuOM|VwkgJ$GJ6GqVq@f-?ee3I*~A^QrfW_$msGVn0WBub_UX38=OVuJbv
z0K+t12Z2U(_b#zy(w=_mj3mm#y3KeeE(rw990P<H0P{Bajgio#%$AQ{&{Y`-I(Psf
zFJi7#T;?jNC8YzdMb^W`$hwv5;bvdqsI8A@H3p)A6Qyju6Dkn<yyNxmC-|iChCT`&
z;;|N9WhGppGGEvZ986`^XrVO<AT1`S&}Kl0P&IaAfebe$ia#X3k%CETc%x{8S|$3N
ziQw;FZ{dt}W11RTI}}EBbb^m_SK?!g;ompL0bse6RcPJLp{mJADgRLzKGOA1kOs2j
zUs?VCryu{TCcgLbzdV*GBJ3r67wUv28%5>S-1LMb2F1*~_8w)33+ez>@IPn!f7ZK_
zKmVf?_KLUjKPZv^HELOlBVO50%apv8J3VNeq)(THo@IxuqCSu)MZ<U(G9l&frTab4
zWcDXzzK_e^A*-s^xq78czro-RI)lt<-;{73QedI%W6}SNiUq`B=i9K(Wa=nFDDoWF
zC-7`n)s#C0u3Rq<qe+gEejQyS5-5H0MVH!j;llj$N0g>7-98Y6FV8#uKPkI?z`vhu
z-&;yDlctjATkOtY->m<V@8i4uf#f@E_YCFB%nmMBF2Vz)&RQH|oOib;t7VP7Y{1>q
zRRg$bMl{ObPvBR&G#W^4S&Q{!02i?SePjPb!n1%bixDBY;J<|cUlI@2A2H7NxiuN>
z*+%}c^m|DDZ-9VKphWtmbcD%@r_Tdr?$Dn7*5Pl!cWjaws+XC64F3qSNj;GK=P|^j
zf&cnh$N<X!i1C-AT3|)5T>mZH@!|j11)Nu2L9f@QztD|41^;vHzfGmI>++8ah*405
z^B9B+<sXX%SnT<*&SdptM;3c`;BE)Q@4%KgBj?@dd>u_dsW0db_+!n_Mgy^687Ja=
z!PYYW%UZX+VCtIx+=RbNF_Ol)mB5SlH{f4)`R6bUsnoN^Om<0X&vyF**b)f80e{T=
zzXbi;;&JeC-i!HcW+YUi0k1@aw*dzh1LCV{#+}4uAs^Ro>||bIM^^vw^nVP&+VFS!
zqW_P!HE)s38~GoLa(@F(*{MZuG;l75BTOFSz|JGI>%xD<9)Fmvd=%{ed>E7yD6kSJ
zf$``26*u~&L{EamzKBB*pE8>T*?AfBX1)+|;J@tqM~-Q*Hfs*j?~3x=4<|BSc>hd&
zn7qJdYewCR&HKmQe@XJkMgO7Xe|Z?e9}oL$>i^?m0{<BP@#4igDhF-~<M+aEUbTPe
zVj;~>Qj^JslJu{c0bfCJ!A%iTlo#M&F%Y1a^y~QgFAM+eYkwSUqcf7;xs?!&@Jf1f
zA`@;@#kgDO6><0ZVD=|z{$JAzz6m+_G2ojxu1)Af_=1%wh&askzWpDQ{pH90-?=Hi
zL-D+MbG${+Z033KcHfTH+=&Xkdi|N+(K_<K1j)U-y2UnHpToIS&WHQA-0`Tr-!7iH
zD04#{q`I&JU$6d}nfU+F3;cbf;NODcKY9Phh5j)NLn^y?Mh7niZgSnN`0h+rzyo+X
zr~VS?+aH1c{S3vSJ<qL!Kg+Ef%&>a>*O2-%xauR=|LM=p0k(Q1c+n!qc|T_+ll<8_
z{GR^dwf}O}H3zAU|IGM*Mt^+pbSVk!9?kz)>Hir1S@eGYdoA3R9d_q}T?6>{lH>1m
zUP>aDURrNbS>>@;bah!6ZOGbR$LY6%&j=?6Gk=2O`lazmr*d@ZMisP`Vcpd^Ccl8Y
zk$?FMJVrS@*P9J1=aL8#R`77JcfH9rL?ER&oOO-x)Y-tD%wA+<gH}K&11yv04})uk
z!hXUw@h!Pu5PhKQN+<aTb4m}b58O87q-vyzL)0j@uGmMY&}7ZNa1{g#q{~}WKHEp>
zn(B@)d$p@u#12CZJPTwkc8dWmmWoUSQUD|n6ydM3(drvtRpt_25P&hILcB#1`kt6s
zI=pk}hpWj&D^&!6Q(<)+vRoNBrw`qZV2~@}d*=JB?7*skUlxYHmh*iB0Z<{5IcCu-
zLK~6bW9J>-Z0}fA?UMc|eiaK>Ao3G$5Akor9QfV(&w%v*4|?st$u!U=O@0H;15DpG
zHeWN6vC=Km?(no&b!HkQbqyIRH8It*b2&wV&1?-h*y;6u1LSW}9v;AaTLYPrD_m}c
zK&EvnrL{Mp6t9RR``fiY+^UdgW(9?y_RGw%S;y`t44j>M8>elBb|5q^?Rwv2nq&+7
z*&a^{ya;ijDe5@~;WfRGT&w)f)zy0ik!x?{Cy2w4mkr{^PWQRlooFeDoU5But)u1&
zPn*i(5wRed!Yu-37C2`Nl-1UN?=G-9B~%|jnz$r#B=E|?UVNTVg-j5Xknok8I!ML@
zcPi4@j8VAqG|%+cl8<E_$!Y=G8KEMoGng=g8$qdVDuO_v8)65IvkeW7CNtV}VN;RV
zbaBQh1^n-Zw+C|rM=z1QyiFw0^$vQB1i<a%!07->Xu*&o=z4wAIi5#67Au~0GtQG{
zhvVJ;La2#)NhT07IXN{N3G01YB@9kR<myi?Hf1}i!-vLuOS=3nn`xALNq0FV$7FLZ
zG+pj@SQ;ASkvUxI2_hoBCy*HNExrV=&x6%Qk4Pn>Z=5S%rwM4*ifxjt_Ms#4Lf^0Y
zUhpS<fUA=?;KvZOELR01*>LO23ZmZVP-TXeWndo^ZcIy}yiOW9C%(gK3r{Ya@T^9M
zOTwfuMXc8nB3RC`92Fxy+xf&NC<@>|($vw-jlmH(NcI%W^BszP9{4pO?pP23+b-9Q
zOOI=bR}B-}Y87>$dlHw1k#2r<oNmQDLa`>8yn6r~cv_cNR5n;n^@p)pkP?NwNgmSi
z7FoHU>*P?y-FdaTy^@%?^hK{*E6Cb+a8Cvf(`J-Mf)GuRcOW*^m7gW$>riq}ABc@n
zl^@5TNabGQ(wVhXSwF;Spw+O3uaQT^vpxtJ9U<>)v{3U7M%RCW<>;$KGi!NTux{Sd
zngrLUQwzz4XLjsK;hR4zrM-k+mdsb5tVW93Qp-k0!2$U8ky3n70{S^F*wDE?`<tsU
zGAueMJCnMoB^(kDvOX4GNWDsa9j>ZY8@QfOD_o>eZAIt*lz6E0wFQorNb~`2o?&Qz
z3CjZ`!kkdJlD6vW`47l*b^i~<Bpt3O1s(tt(nKkIt66sLd~(~Afw9@d5cO5ON}P&<
z_rUNtNa*7RYv~raswH!)p-D2$kfEtrh<%KL&q^vJilLEtX7>hCyYBTKU*DLT32H>^
z2r~NKMGX_1LIF@zrXTS1C2zdg0~n3WCPxO7-4kl8qYTj%11sDSh(g3ZI$UCX#Hh}X
zRpq4;pbu0DNwjV(!d!!MQ`!PKy>Pd-B$ZcZYRc@Zvcry<#r25K!HVRFY)>&qIr&M@
z<9Buiz*DD&#QXP6VWBRARyQ*qj?1}r?`3CZwCBI5(BU;H0>D*&p#@N{><lf(CFCFT
zR+$ym*(>d(?PoN~fo&2hThw%jP7vV52e^AT$JMwsH#K;x;-GlTxRajHcG=*lM)_@s
z{waOv6q-@g@|5*tE*UdfWdImWYl=n}+9|=gMvdQ3#6CsUHt77SS9g!hvlb_o>uDtl
zgQH(oWpYBZXaaGBA?Kvd6@6-3$vHJ}GBg^SCDwowyZ!}gj7C%}2jI_zQRZkPy4f3>
zjq2X%1=-5`cs$-YQbV-;pfa|};IYtEgQc`mMZjg<JDtnMccZR{th(^xGXX28DI3om
zZ_ay&BsbPOww3JrO%o=7u2%vsLkViK?zmNOBQrC~3cR@$g4Zj}hllxl3A!_qMl1EW
zDEbPGp8m6?sV$GC$)IVbgl6>Hrvj(F{rgRO87O4?zyTUhJlhGObQ=2fs3r$}n%9C!
zasrV|jD1v;^p=TaN}S%SrYFHcG)IP-Djarv3Bt%kMF1){UCIrabVpXJjk)4^z~vnU
z;|m4_{YoV#Vm1L6Qwp<7VXG3^17Pa5+d4CaY@Jk`QiTm<34jI(hWbN_#IUBHJ&lId
zspAgU4XTrWN7l1MHYY-bK-QFSgM<{MHb?8Te#*_X(q-I15Vn&i$zCaRN7RUsi2jU5
zC&AnHuohLTsyyk>7`S5An$oqi(phBI=JczAaq?bM)hs3K8CqJceSl%Br3CqstO~m(
zl`dJf*7cUU5L`_QWl7vQ-2A2B>JYPJiund-;r{slaQ2pMaYgO6X5kX7DBRuM-3byL
z3Wq@9?hxF96Wk@ZySuvvf_osi1ouF?-rc9Kz4y1iPW^;hYtHq|F~=CUP;K!So_nwz
zAQLePODLZH5;@)_Pegi%VZt=Z`8_p*$JEK*7s;B!1+cHw#n1lPT}<Vt87|OJQ=FVT
zz>+PUsW!8Rk@&+QlP-`d(;cW=%4NCo*~=HOQ~f8<$w(Uk|8l(<?(H)vBsqJ3bv){w
zUN-K!3~2LByU5tno>bJ2!5%>?NKg?8Nd85}Yti2l38l90hk{v#_>hgOOQUP00wd&A
zVvUMG5d|oUjIV#WX<>!{B3@Zo+)1ZEyKf+dD-w{Y83y>N3%FYQdR&N5FITd$jvv6W
zz)Uly{GtQguXqDC_a5&2cnfwfq9%3lI{<WA1~q%I=n|WI8hNWQbZA!LPpu^TMw-jG
z_~A+#!bRm4xN2RP)n>?-^@=*t(Kfx0yF5$Ws|`6LN~MC+2#nGYDj@)-de*c<RW&r^
z^APt}EaD1}@j1W<Q-S1KUTeZlNfTCmuv}Z8P$&K`cNAKu4dMW6L_)lGT+`S|!z|w(
zx{IaqnF`Bhg?WtHK4W%}b)oaKR{<f<uTq<3kBm&Z9=)8HD84^8E!TdmLcBQ*@|7Z9
zmS>3+g1wb8u=kuj{xn~wXJL})e0Nj(>0sL^UW5F7*LTt$DgU7Cez`x@=`iaQy4UjZ
z?2J`1Z~H};*$k_y>TAZ{t=09ea&=450Owk+9ihy5FZebH1=$oIl;$3fDN#jbD@_U5
zy@cdq6U|2w242hJ!>{Lh{JB39-qvlZRl@bT=+ZzwMr#a`0N$;R<8$t&(MLK<#4u{7
z|Cr>L?sqrkXy-19M8*m24PNwtZ!i+&GcsSmk^W$lyrI@2^TSZDn#O)fv(zZ)tYTW|
zF78t4OlhX2uRkXl#V9p~nWkHtZKF%o=SWgQ!07<xhHvRt2IKXGAet!!fi?pmLpc0#
zThdTW=&2_U^k;Uqv!ch$HQ%x&oTCL{QWD4s*!B;-$aEyRipMb<S|-F((8W`YUA154
z9zlbYQo)u+8y&+Ly?7++utP=F+`RTqp;RTC4-eeKJ~>%?r#*1R3ED!4U$xHC-`)M~
zWrBAk*UfDl?aym!u{;d7x)2pp`DW$!ak%5YKXOW++RgclkA(^t^BW8;ucfBqe(B|0
zX;v6mcub%`jsg_z17LpY5HOk;6Pxi`3#dnT9Gp&eyJmK1!DBew1U=j5!>|i5WM-6I
zpY40l&2!~#Y%O)UTGpIrvi$1#sqC7u`vCZ3Pj!F<2t+Vnx|$gekOtgj&v~p`ux#=V
zG(g(g?WE?j?uvy|bX3jNyphFqy)Iw9I`np@u1DHQO;d&DbpkLhl&BBQB7?>2>G@T%
zDtfp@Ab8OawK>wWD3PXkgdbq+ueAg<B?=uO3>C2muyGNKmb&TRGFG^?PtNezz%t=#
z(JUG`7?g`p>EFo}M<qic!nqDjEoaK0kG5zqK2zE@JNp9bFP-Dk#FzBMRnfwJo7%)`
z!A!E{Yv}HPjT{x73Fo`~61(lJX&2z}Ce@?x=|t4N>vOciBoqH!_6TFYLY>h=9(;p&
z!B0vue$FL!E(@kGpny-1Q@M(uT8D9Zkhf+tGSr4Dx?Vn?Y>q~iBw+>%LAk7BmsW85
zu#3SSk<{@M_noy$Ca907Gh8T$%7_5YEFYr~03fCFc3y~k{Jz1etSK7PSy>uw3D!*I
z87JnP{?EE4ti%V0yM8QyOdB!)XyozP1%rI8`5RIKM8ws}$f~q=0)eh4sQW=ZZ?_)r
z076x$pMta6yj@ut!B+WV(=b1i43MGCDG#6r5y(M^Ft$QIir}mOxrJu_0Ti-IL3SH>
z;K--T+&0^UxNg}+So}bS*%n9U`D%@6V}3jtiX0uLAey<rk#*Tco@#;2;he@XD#zSV
zA$&^RfsjEuS<bnr6l_l*xq3?2J8BWaUL%^C($mvwVhMzbiV@dkjyM5n9F1~gnl9z0
z!S}O~xa~d25#c+>4+aDMHb$b{F@e;IWyVz4ltGd~m<^wu4Q@(rEvr5No>JKM{nf7y
zKjjzKOIOMn&Q?Rl-58JS&e7OLd`A8%(L|KVd_rZYnt8K&S+~KtY{)piImcaV8alTm
zYTwC>_bUzJMFGixGG4#(nerR|rZ140y}tl)H#HF7Hi%6VUYyTl(5uD3W9p+LSZqD*
z?g5pLi?3df{VnZ-Oksl~Eaw36<>QV~sc{xoBg8&dpHZK82eW!c@8GXu7ClKjpTCZg
za&l@dmGtOTR?!A^S91Rd{=g@7>2n3+`jmZluzh$Q)L=WQ|H;q|f4bdXzY3fRuyKyg
zFEC_N_07<NIb4lbhq+JQjnfFq*#oFO#)<_W5xR_ktQjkh4+1!fBk)jyzd4=9Qe^a9
z`?Am+!MdK!iJC=`m^p=RgVi`q@LJ;+DODnw$B|Qh7oNL|%6R49p^!8ev`=t!i5w<o
zZf{kk{5_nYLY#I_0?L2rQ$~*6pkv_emLDcEWYggr!@V&Wqh3UH4GKnv+m`}Z!>tLu
z2-H1AEv%0P6dym_)V4Sw+-KuezVA(?ggzn9B#YR8+`;iePVMdJRu_QbJ?mXha0n^+
zp|@cP-V_{LsaEI3urpYi*<+KLq@2$D(5nWme9();V^i+8XD1L`x^Dc$h4o=tm<dr$
zg@UY~tyFzVw(z#*{`G-<(G^9wa^$-oswd0rnZXrmgRJVUS4TaYL1>UYGu?+M?*TEB
zL+qEc-eT5_UtbPIp~ti)qFRM-!RJa1n+Y;tA(V8>BaXP4KJ_#b6S08b0Xh!%$z|r-
zHSAIugY+WByJcPh^q)RIS`;?N=_tqknwjjhG|WfjGBG&>kPTAqXzH$b(D<OCn~&EH
zyQ7b|hJ=XUdv(AY9IzA6j-Z`(NFYgF6Y(*-W677y%k+zOBLDFh+0~Z+jrm;BiY~F;
zFUcEn^#JjbC`Tp$A7Njbt(4v`+mN(y)@VLj_%sEe=<rOFnXR<<@cMe5wRib2RP0?o
zSf_Le4jG%W4y;I?{gnNN+;T*cRooZ^<*-n}hF?O9hje}7r}C5VWUX^a5mE~a0UsT}
zpUqA+A|tL&&5|O&t((PHUW4!DDqIX$+g)7!oDgYi91L2BGJnh>D4n7R-y}$vPa)6)
z7bXS#O2I1qvaSgSh$?AJKDtE(6gfTsr&(#O2fcUTVCQ3y?Wk5d27(TsLaBvOGt=Q|
zK{Rz;f^>81k-&@t8+y3&OQNK6*hrzUmU4q(kQh}oa)ke5E*LUKWYBg`;yBBc(!9M&
zyuj!0$!doyB!fHY5jv!_M~odDA;H~KQsaBQz&~Rc&(41iW_;T*4qcZ`ZDIM6J^VMr
zDS`TQ+I*tHRKIovJ44!1YxtI&V#h`e(j~$ef$K(MQHTzd9rW-Rk>SaL003_IQw85T
zVqNEjpxX#hS$SSsPle;TUXAq#Jq0n2N{7VmKY+QMn~8A?vr{y8iew)<e>6RYvd?8}
zF=$wY9I_gLq*PF4Al~bTOqUxI3~C1tma@$v0aqW%g2iqu`C6ZQ_~1t?YI3#3HIhs0
z42AGcgw)*CIPc(Q5#F}9%onESwT$VV$RSoV=gQ}-1o{bMb$&~`s6rr$3M~M6z1fle
z6^5zb;)hwrb1?fd=FH{J8kIha?GE~+!!5rSF2R;9PUTbeS*}~61xgdno)G}a!O`O?
z^y`<mI8vj8ecT;#D-+WII@f~X3Y2ftBw?!nZ{+x#65aAr{!(|wTB`=0A5`9gCu!Bf
zpO;%y@0oXgXHQ_reGAmT7hkLkjB}z9yz`5HV)V}Z2gpeT3atTwu%H)%_iCt+qF3~e
za)b%%hWS6b)c?z5_`fbm{x2mRr#JR#@r?~=trfPV$RSzVd-@^y#l=6sL=@*=xv3n}
zjAJp#620oCkZiU&7zj%$i(cD#R^#Zm!>hD34F<NH{K{5G?;b)enPvX)*>&@X$`-b~
zGI2PXD&Lb6jWk?m)!8)>f9AV^?&d!1oOd>@^)Xlhgk%cq3B*)Svv*&kOf-0wPu2Jo
ze3zh;bI6dsI2b58-LK@+GEbZ`rBF(5z%hHR__(GVpwt#?)pE?R1$!GXogY{&bpP-J
z4pfKMqts+aG(0^J=|*f$<+t%DJuc%LJE2Rst1CcsL*W&a@kDDV0|K3uaYPvTCFKTL
z65>fG?1S-Zm^6sYXcDT7%`(cM{STCvK8S_6&!v|1@;5T;4|a6}wo<*GhmGqmnfR5c
zwM07IIP+>ln;r*!=ACK_++(&xOB_sL_KP`RuQBL3BFZy-B3~_AbrPtfp|$66D6`pk
zN4g3<Zjne5JI(GOgrPWE{@Dh8{6G2#I)jykjzD$gCknOgVL4+X^BFz6BA*%v@MeL}
za%H$G=is8TMAl5)c5eD7L-vC?WdwwRtamMZ@<|7rVoozI=Gs<O4-eBAz0!UaW<J!Q
z%=5);r1&Y@#yym&gqX92Q<9%v6Vk^K*R$n*Oq}kjuTy=lSNdGBh2M#wTL`_LZR#1m
zcUiLejq+?8#@dE(U8;qTqYQ46t(hzauTAdDO5eOYaoit}bKK^+8&cPfrCC~fBYyhD
z%_+-Q<|l>`+91=p+>><pXCT)bMlNonl`=xYB83qUFi#jc#jQB4itnAsQb5w?m?x+h
z9kxEy1EVO~DjYeHRzvmW?VR5RJon3iTkZ#0bMHh}?Y$&;U_O8r7md0n@~qAqq6$;<
zL5P@kv(NJ|iWX#R^}+32b@S-$x38sf*4BqKWFh&9;6?dpQmRVLXv2XXjP66?Nbf1b
z59Fdr6`|=&B;_r(NU)g}N<x6MRZ+P~8J)DsHmABxyi#5#v(;2jp7*U+PTg6DEkA?g
zN=H+WodnFW+FfeQw0;rVAZyj!*aa6~THJKgXcPfPV@Uq_C=98;5#a08ZaN|JFEv;9
z>xaSoAYNL#0xjj&euqb_lXX3n1MEuV=!sNgb#aOuz-GZwovMBAOGV{E-OI;wB?=L{
zaBFjKiw~_|W~DV%HR?u~_U`+0Wn({=M=P9rf2v=N-A8<he2)BM{r6R;4r<FbdzKX6
z5nF7lmFWlD6JZSkC=hk?g~)~$&DivBl$6F`vmj|<Wb~io*EaQdapMkm`vqJrlza1u
z0POhyZzUDgA)=h5tMmhJE3l*1U!b?)FAbruH*nq<B^CHJMp%>lt|7w|ucOKm9QX%V
zxyr0n=NPWs0rL}4_rbW)F6>hUC(O7jpdRd&0xfow3NaIJm7=p$+0&B_y}fuPHUueu
zX_&t)Ct}A>20EI~>$?Z3JsXE6;;1FriCV;$(_!au7imL1wiyicuECO&05w?&3`w$$
zyH_jCD%(SejE;KiO&s*(5#5rbXFgH0iZtIZg>%LKS!{hir?u-Zsq~Z3#kK-nY=RN(
zXHx+aO>Tok#QHDV<sD76+9_ygTP@BM9e^z*SkudfGq!y}ElL$>m}rw#f}(n?T$Sw>
zlgR4pN!Fz8ZrPOoOv884Zu@;8*WBn**1<^;#e0L%T9z$crj6aWt}&{LSmM>VvHIy*
z@+cist$zUC<u8Si)rE&brH6vEE~}BT6ZyON@~NIsrq@UGRGDc+hKPFU;uGR%DE3k`
zhU=-Q_@P5OZ~tuNJi2Paa9Yc8*jbC^Nm%}2_o9fIxzmFMefFB^2$E!%{rFWRVn0)V
zn@j_22-fFA5@Y>D&1ox(3&Q!zaXw0V55lZI*~7Cw!cxqoLsxTuSstNdq$!U?{C59s
zVUD(fcHSU+P><JJi55s1<W70;Q{Qf)3&dRvZPEjh5V$v>?l&G0>Rfz->=x)@a<)p9
z!u8;Yw2B^>`~?*$OcgC{nB!a+GceW8gw892GJGSE4lmc(C)yRvpoY)P5zYevoPH|x
zJ95d?0J=jq=Nz*mJk})ldn<-!$LK-|O;TyQTBs7Fo=p2_lVlUe_S5+$5oz_d5%<UY
z53GIj%v2B#{6GO`+v4cM6yF24-cpR>9Y|wj!!oBIrX`gTpj?{|OD(LS7K2qLf1nnQ
zo-#rj9J)%$U0GT7Ovj4)$7blXdR33@S|@f#Cr4%JjMk4g5qMc}N3O6GceR~If^<gH
zbd-iigR{BXt*hk>h1bDhnqW$#;XN$Pt_XYnSZ8YxYqY7>B<N!RoYOiym|uv|F7Ki1
z=NB?@sK1WHQBNmIc*`TKKwo5bWj#86>JU1{y=lmokpN<;<+y4b#BV_o-zhs{sh~gV
zLn<hk+omiqMN9K{?7E;{&$mKapF+ub9F$As$mdTlj%r}d!mJ5Wt@jR8XAtNBlQC-g
zrt>d7skb*oSh+zkIsq9bjqeVg?eDM*RrKjj)Hm~t80LEs(gIWK<&ll4m{3lBZl3Bb
z)QL87N=CUK>)fZ8U%~i<psvLWwkkt0*wEosA)Lsg6dscGvvp5XK{l|f=IXw(wXQZH
z0L)>14hB<W8Zl7+iBPb&g3OL^s>ihcNrsurL{M~wxm!j1*?F@d^kZiph%;IE<Ey}L
zG^rmr_UsCu&Fiaw4Yn?EL`JBXp4GX->O)oJdowL}z;syO5UR0M9y?nf%Fgu#pwPuL
z2nT(fiV0uBN#*S=xzCNMQAdYRpNcvpMtP8^QeWq8d}F)H__`u9wR844a*G>dYzW<*
zob^TdwU)^2vAqk;VUihGW21>^TUbUlP7s{WsYDuXHF{HC3sb+=fH3UsM`@Vzdbj+@
zN7CguL8<riI6uKls0=nmY(DMOyysQD3pE!L&Gux)eRn2oWblsxVdf9WG5F`lmk|a}
z^vUahc!7y3W$nEM3O8f*b#yHsF;GC)qavv|A#ID_(v2~*4ugQ|<4-k=&%^WvzCK5G
zF`f@-%ep-a%8-l!sM<lzC!g8;_J+O+nF(LYqp2Q;$9t-54Yyp)*=eFgBLc3}b|07;
zBbuEG1wiyaw?W=C*?N4v3bPQ^w#Fn^&ePjGm(FpS*mj!jE8Z*cbc;EAC4jl7?XuV@
zla*N}ZZL{;97r7TCYb7xQ%Q8!3|l3d7`VR=#u#0GJ=JBO<fN{{UkjDHMn<~DXV}YF
zwEITB#2xtZ%vN4?dtVHtg50k1)t=nb7#Q4;k5S1&g;`x?)kn2ks_a7_!{j5V#iJ-C
zR2-~B0tR-TXp{w+mo@Pswk4x<%zxODisOD=V(Vpv#agGRr~zb%+>17iFkjpd#T<z-
z<}N;kZ00+;S)rqR6&d^Kh>+x9m*8bvWP3{zb`1j#LY9uSS41QQoZ8czbRNTfO;?bS
z>FzJU_)m*<gtFrB5>~cUs>dKx%%`^23s$y@^acRZaIHII;!5+t*#%01|B#5qAN;>@
zPe7@19$Kc4Sb{2Wy#Y?!J4Vx2o7y22RD3K0;LcK2lJ3dCCG`7Oy3Nd)q<L?C8?8@6
zL>|{(Uk<9OgwtxLH8bV@Ajrzo0uHc=Z#`<WvQkB;YpVSwem-X6CK3=}tTl~|3MoW-
zKKk8X<Oz-eNi{!njJhl`8yg8Lot3laJjwIQ_gZ0{HINOk;ChWY>92BmrA#|u7*!^*
zg$`XVq(gn$ygZpRIK$J|ItN|CJ%4)})Bgcz7HLPuNEJr182n80{z|FH#5@gBiEIws
zOa((Pn#8zb=wzDVX3YTj(i?R)r?Wtb60Rw_43d70sA&Kq@EoR3Fa@lo5~EM6UbyQI
z79Hegt7fgWC&p1+v!kb~8w$E3lQ@SJYh1l(AZs?N7#^XT$fQP*lLDAvk^1^fwk+HB
z1Y#Xuf1H_9tJAN(UY{h9P?OI^r94>l+V<Qdk*c01(Jx*}q%gTVk2I@Re@bL`!xBE#
zFP{+S)vp4nB3rolS%gzd%_cRMA=Z4DRK@Z%J?n{ljvKYfMy}9g@=C4-r}~{Fa0se6
zlxlVN!%YRavEbeBo$bO7WVCO0lM{6R0~}tunYlI@=}$kDdD0T5AO8b<3v*qPc1ZZ9
z{`@r-<|>2{Iqn7j$3<F;vrw3^S#87^%$Pilc?S6=ddi0;q>LaWZ^){$m9{-W##hh#
zDmTV^)XEYHs1kplmb1om!Os{sD(I)!vaJ1sK}(zaK{;*-3-mr+B5*!b_u0Nq-`O+i
z2Zg`uC*e|Rvq^RXZg)nSeF@$sySiPr!2^H*9dj`f_HD4tZUcL5k7!egf$ct!>2#G_
z8H60HB{TvX8Igb-SFD*#<DBe?JK2k8hWo`VR-{JDxME_q@T=~tV65r(bC0xS!Oh~>
zinDD=^bCUCX5#HfE2c;;*S~{N@I{U1aP3I>Q&T$1+oh;2M)WAz`J~jrlj7DYjT<@)
z`}&Em(gHAZh2Us<n16t(o{88F-sbG)hP$V44y^Y}XD+%w9JMGnGo{brpwXEWkq|En
zf?K7=unXef)gS9XS4%j`la59~$4KWHNP>HjE;Os{z(RYXrN6#ocH}9Vg0U1*=W>$u
z==W)@b=W>p_>PUAipw+@5`#wg$0@gEH4T-(I^<nIN32CLQNvczzu%2Iv}^JD=QPBP
zN)eVL{sB~O5mX0HeZruI!fYfs39$Sy3F|C8f^g8?S^hCuFiVA@cJ1*DcD-d$(xXgF
ziV8WSgqPU19#<Wu%Jpx{68<IJCFm5d94qTF`gNG%ytm=ovV!di8HpvFYsCmq*b5LU
z*SU&wN`OXoY5Gi?)P*Gn_;tehisrri(FzeHZa!2IYk_2m9T^Wfh6|XWFD@XV1FNe=
zmzu1pNW0+-oMYt%ffNo)sXf;NER3_`SC7yVCb%Z)(Oq&vO4BM#JS!h(;ZK=pF<o$?
z8s@zUfDZ}=EW5dUde-Z|{;)7RWG84)pWctqdW*aJOX-if%{DR}+Eqdz4(6i*VVrF!
zCu+s*mD+>5tDB`LBAy3x<qeahLGV5}Lh&|^b>?A4zP1k~`;JRgg_-`SYIA|+mvPBL
zjbP}PjWP1qt=qA%YU1}vDc$hC8lSvVoxU;tB$BxGfR@w13gdqOl5P3q6G6-?ja%-2
z06{7$k47pgc>24x|1J+wF@1w32LDHDFahcWM7h$;l~dr6MM!t957tzsawFIi4FU+q
zzGiH|v1J#hN@y{T6EjZKW5@mdDl9Rvp8pV~LF=IV77~rbiF^Q~q(qhv$1^ws!Wn~Q
zqGfBPFb-x9P2YQ95Ys>Dn!;V?l;mocP;u{vp}^3@RKs+CMUMy=>doDsyxMCL^@)na
zHOntb%?780=^}T`YTRH-98gBXiuwDV!h75{1xmkzT(E_l2MloR@)#z72kH}8K&8-Q
zx^wI^t3>BF=sbMbg1`_0i!pU&fFFU=@o6{1W7jng+L7th+g%#Ro2uSaV|80m4OS?k
zCHh`n3g#+z;f%-$rwDd;bRfWh+xSr$l7&|@O#wh!jq6gLNS)q>)Mn5r`R6B2w#IaF
zgsOiUaCA)o$w5TDWzaXu^rHWN6GNcASpU;DNuBlM<#R@0TN~8OOH*$3(zU0J8F#Fd
z#8l(>OQb^+L&NEPv28i<=ijlnj=pD3S{!3t?%&F(RXIKf3*y024asT8FiC7P(*<8;
zq$L~P*KVH!&viK00_h9l0-qO9<N+QHn~%~9`O{phQ&xo1N>Orb@3e(Tj#+@x#QCm_
zPlsY_ab#p1M8wgi4Xsxfn}maBe<hwufLLT2n1|`5&7f$iZ?#dKa|PpxXoaDZQh<!1
zW0*yAul{C?-?)JZEQr^2)th1<%p(w#zTOXNihNANB^Ww)3=cnfZD9`U@<RW4Bp*V%
z{}@Nz6XZc?J7hCNI&5>w17Ibhfx4Eypc226P7QD1jIP~idsL|W9;_4`A(*2QZ|QAv
z97cs(l&6E?U&~MJLPNY5^$nEC#8^NMDCCP@!SotsL@A<xhASc!Nw)iwQfKJ~AO2)*
zt*u;FEDF^uFg9o84Af+Z48t{#>v7QL+NGf~6q>t6&2ui05|nUQyNh)a`f&fz#96q0
z`3I1%B0c;gs;MtjxQpJhWD>)#=Wx{`3qfKZDm43If&F5gFZ68b6!lkdtlrNt;#s8s
zTcbp?ZH&%W&R)vF5R)Siu`l0id~uSY^A9qQ(6+&4Su=1Or6ecZiAF|Iy#Rr@FofHH
zY}dkwb;vG#+4J#7W_n^_n%z^-s`xL!NuQH<^DWQxrSbaJMuYqaUt@CK@8xQ@DVD4=
zG&UOv)<_u;RX$xEY{Ugm!X0Vcm?@_oSerW0R6taPQy+bGXDej9(Epa|^P4l`usbPl
z?N7<iq35`I^O_T`;pfH-9{7;#-K+R%VGa-UAdJ7I50Zl~#FOiu$T6J>CH&k7c8gcD
z5z5v`bzkBjky|SfN0vh+EQg{Izn?M$SaCnu?zM9eN=S4btvHOIMK|APdV%Vd{5Non
zTw@~QyoY=p)i9m?LrgZ4t<PZ7Y7of@E-pi0VY^xcm(*munlN+bNu=7{Fb_;}-E`r^
zLl|ZH856}EP3OFkV-JehE;O}fTT>5XMiSo~NNq)&9Q4!%5W($Bk}L2UY|-Ex;&xvT
zKHGtoM}o!;xxs|`;u^)aOvy)7`p2c!zro9vH!7edb3w)%Yy%^(s2+Cy(cvQ*fdGXk
zocqaxGQ?SjL>j+wG`p)+@>_Pk`~HYaixw8TE6P`)^i}SF8c_9YYvjGQO;t$vR0P2a
zF5CE@Y+@_*2%T!-FG*Cg4k?&aJ~U*#Zmh#Tv4K>5@1M-R^nl}38ZDU9r)Z~hq5N<)
z&auZJ9t8FY);kdvo4Cs987{w^@uoStHMt;$W4L@231MG4-Uzr#+cftkl4M5vq80@3
zbn-8TN|#%X8|Z9tclp^@es}C4E@vYoHFa+aVjAt-MPSXA-G}>GH+b`}V6`BCPnvNb
zu6E!cKXVd;e2Z@nQT)pWR@v|>7gI&|6MlZ#a4GGvCJunV;5J<(COqYt*WbV^&R+he
z>L0+pQ6WG_hMW`hd^zi@Fcj|bq0_Tx>!N6kHEv~(tn&yb(&TJh4IS`m?iHEE>hICl
zdU7(YD8Ac<aV!vRFx^LM+Y)SlfsdmmG|D8Ah(4+5nj!(ezM~h`GA5o7f62lf3*fQ^
zYr{ybA;3slF9J|y`a^^%a71};Ci=)m>>6?`dIg3?pJ=eTDCqfnDICsa{C#38O9u3V
z96w@Q-pJk4a1bAkv3n=*MVM&1uSPVvF5YtX2rrb4MchS=l;&t}f=i4%d3CoF!f8zn
z5@w=Jf#hfd)5JT*L)2Ne+TH4VY-+F6GIdqgM+miMhUK?Zny!(Vn)vmRujchK<4G0@
zP~trWjG{((P(Eo8E!pp{U{9TRM?FKMpY6~yKyj_hZsKgG{XlJK9Y@Ynql}Z%OsTI<
zWz@X%nEl*PLoaiknxi5e299ET7)P|Zmcpb7MSMC(giYMUlhHSb-v@>DC3Ee!LDuN(
z5;QX=IaC56EU!7Cc!{KqUIMc?j{OGsM7<gfTc9Qo^g5YQdPru5@z`*A#ziPbZj@GF
zxZOV$!ux08OO_3>tl-4G#CFMpa@P{4mXgEDHL&ber%bexdl=lfHwu#J82P+^FTW}X
zJZt%y4<%puDE>B_P$Atq%2=`$V;wq!kmZ&mX_5{Ew7qG(nI>)cY4rBdH&6D@Vm#lV
z6xMOv;zT9aV-=h+5#X_uygS$p`*yYvwCe_7w>0h_sME-k^imEz&isYJIVglU4N^b)
zW;eE6J)(U=8UvXy`AHwdvXKh`*K*jLs!hpNDIOI2fXMvapZc%o$jld9OPWHwU1^Be
z-q+~d;0LmC>fUt&p`{JzCTQ=-C7^)ms7tf_dtlDpev6xtp~p)d&xy61fS9Px2(e)d
z+mt@NrG~h6PmvDwbg~TSm<r(>k?b|4(&~ZXm%JHRWyMA>Fr`L+<VU2)2^S0u2iDe|
zsD#)WQ_rs%?uZ}DHcq~}KPM!?XNW2tScXIQU!@JupHD3=zu3ha|5m!5cl1<RB`CAB
z4BuC8m|7j-uY*=NUB|teDf1b;HRWi;yG_%De(3Gj7_QQ&Yu8wf!nt^1YBIrGN6qUd
z`a^w@YM(C93S~n#{;ZE~v~CJtr0Wram{&!@T_Pi8ehuiqq_cMlZ%M(2`*QG>yH!uW
zOLsHlgms)eU5qi%XMEe)6rsi2{#F7^0&jc*ryA`HXJ<?js{HO__4bi3%P7n{xDC8G
zAZ4^Sc>;^!>oR~ZJPa|TcW+9Mz-wc5g5QpM?ru_MMRl6SwT2oF^h>*9K7;@Ww#K8R
zZS4-w0aNen5^2mI*K@nt920DfW#W#1xutxU-BAf%rI$^k{cM-KWbUZc-d#ym>?py1
zK9s{lhjqVrRL0uqH2wC#v)Qdjq_@YD=BH{A)LbhYE-)XtDGXVa`RY@jos#bia|-nc
z&~Nz)<uJ#l!^E{Jgc-1WnLXbqn8#)z#>VJ^X<=Kl!eHjA(oUy%nJio@ghV^prCu%K
z40}yFgn#81Be?1e?;>q2k(C=IP9mtwV`vjj9ri(v8e2P9w<8%5Zt(t&*OpTyaQMQ|
zO}on`8W_br!9`~)?r&MKb5q^Wfqf_tiJ}M#fTe(GxJ{~sCdPLK&bp>J1@&7fkW)F<
zVyWj!7<?j+>9&;ZY;8wVW5;Ri=cx)4WjrY~!b7YNGI|`Xr2fP$mT}9;0{Fv4N*RLA
z(cn(uII8&Gl~3D^z7)=v;gbtQ#I`$yGh3<<&&?&?oB+F0jWR!BsY^#N91FQ&x}e{s
z_fqT;5<Ti}1s{LP(A2JnCjoy5VhDLbubQMTil5?^zmiJPsd2IJPS{}9@4%P-O?rzZ
z=!>7OzvASikkqp4W<cE5hVCj7SK~DZyUVBYtSJy=XQB>sa7?CI)jiL*681{p_S@56
zP<?Gt$I_5%94@?tNlj1GU5kuUT~fw78S6-iM~H2z{-n-+Y_WHgfh9H@n&*zSz-gIW
z<H3$3<n0f`0v^FUhSudI=3J!(*<^rMI+`n=do*if(i~-~P$OQT!L_r<=RfvpSF>5U
zQ)RQ8;$8KgK`r2|tZhbU^L_>rBn8?f)U6KW^nTkp{hlkIOgO~ux+(L}sw*lc@%{`F
z;BA@`WifUpGnP7gFfq<zz*JU#?Y118&f(e}*+`Au<to2;o;R?BEmns;-n}_SG)(yA
zz^tV-QY)F<S^)$;x6lu2Ng#Ddc+_K>w-khd`6tyUkl7IS&|o{L2wtgdoW@1vj6NzG
zPRnf5D?5}80tpUqT&Tk`d_Ne<kyjK+;^X6$;4U3<#+ifIel!4YcRuT-?6J|y>Gv{5
z%q{WOsoY%iZWeELcb`y^3{sIL-Dm#dU8`K0&FKmlxPXKU>)5C6tBXFOM|u+21+DP6
zm9(mFg|@lRq}T^<GXI$QaUU0utGv=ZE0`_IpPBvn^!xWF3wH8i(-ATX3qC!NYoF=I
zV`k>Fw8uo%tu?6g1gd+N6u*hs!R|%#6hgF!rVd}4`}b&=qOBv~9}iY06<ar{s~S4|
zZ6~+?K;I|KIK_^!@a#O+4Bk19Ly*Uh57#Z_mE{_aRkrJ6GGDS7@{F0nNG~MGN>}g6
z^-4rWfDTzUjXM{hICU$mcIel}P`t?Iv##2>VXyfT{-7ISAeIdL3+WwOkPniLctVIw
zv?OViB7pjo0`$Q@a%JCB>&JFxfKLR~3H-PTETJU2{lHNjV>ZmnlwgxZ)=4RylTdSM
zj3REuh>Lw$w1{MPXPNnFZ_}V7RU-uQR6@@c9k_hLiLbyrb;8=j^iHgft}rjuCuk(d
zsjc;uoq>=G@mabwv|J%10<mIjb%R@O%t+V^wZq+FX4))MVm>q=ixj@>AK>a$!l!AW
z#y_b_XA#?+ohONcx;F@B<sz5o7UyC5#4bSCSF4KTfpNtAIsnIc1>3bSLgQhC7Kia$
z1G8(6*1<EN5c^^Tz<TH#Vg>*(Lirh+SR#>8Na?004VmY4;U0uw4{7PnK;sgD6g?r!
zd|WMeHTHn5AbMlY!Ae6jDn6OyF1ckZN7m`CBQk~zcnO&;Urc59lM3S&(5)(|FW)~v
z@t?&qx-P4~5_{TmCQ|H1s=E3=e9JN0>DBt^Ui_OrMTjs!9D&UZrp6;^6|cM`JoN%7
zD#438^bF6|MfrD0U?)sQJDWeJK_qNpd#AN`lX;ecDd-qqg#k!5p>_kdrFY?Fe3D5_
ztG(nyZX$Ed`=seZbMA?PCoL6~o8oG9Dq3ieBC_(vF_+jM;irom)%9?8?n?-$C$7Gm
zgZY^J#(QLI;$JGF!&QeVa)hD)1JmSI-b%H*U8qRA3EhESNqQXedvl*_QBjSALM)B}
zrWXxae_4k=YvYYEj4S*DH1k(RpQXqJqNcqKOHbtkoIanX{s0S7`_M;hXXA}2Efj7g
z=zU{F|CT6^#Qm|7O(Q-drI3PrfLe^?YKMHPe&5x;AzCMGr6Q(fvlGZss17`!6nYJU
zIiZa}rn*2NhWfE(HKL&WcsLwsDLCo}Lr|%SM)Su@N>)km^(QS%c;X17n`1b?uA9vt
zc=FMf{58UXC6%R?V4-cYc}qjpjym`3RGS7%*2;}S@H8O*=3LPQW-YYNY1C7sytt=T
zj6Kr%wDW+(St%rrB`1^aTZ+IfK^R7%n|I`dG|_O?wS|MWxyPdR(%ScSp;0*G9BF+_
z6C@jhz=6qhuU5Wtqt7czGpe}1KRR~vr1P~j%-M@gbh`n2_;Xpyc9s;z_|yc<q}@!v
zT~oFIvliN2&!;mf<~99<m{K|gN=;7zqHV%1MANQ~kEA0vwIpQGccXAkO$wi}YPD6j
zL0`g1<L(=_2j$V9WTh5>p9rjPma5E2KP+xi{Z4ePq~6JZt6nSI&Gk_3^is9Is6C%o
znxHrw32^yOD|Z3_0FYbOueFTY?!JCf{SbIDPf&`JBaS}yl$d#_2jxW7+vcYCw<x$_
zhK(lz&`cTJ`PFYP7+caj-9>2idimU1RcR)`kRe4^KQ2UpbXoqxTRVfB<XPcx9W#RU
zMcmU4#4|#(@`_p5{7mT9-d6i&DQJvmroxf7>N?KmbJ&Taf@%XHftSI7LprkmsMDkF
z?-4V0A}T0{Az<X^l&PQ-qVao{dAqQEt;AxsBU=3t%7MvIBEA%MQxs;3U4NvXa4tMd
zVJ=3rFTjOE_n~F)1|)X+L`DX1y?z>8_Rw~)H!>$gZ;TytSn#{d<Qdc7VF1{>i@C}F
z#jWPDYkF7T-_`mx%GHwp`alOm8s$}U6JL5zerR)%{g;0E_~et|(C%pFZtB$U3OH0R
zS$t%=hCA9L(n#N-vjdi2!e`WWBD99_d=+)Z-7}mlHrSORlKIqaq>*YEe~U4pg&=@g
zkp}#p%fP-WTgco4{rH$d){pW$AVQwYftXdnI@WOWo84ldhUmLscOs4ZkNZ0HH~Gh>
z!yDN=<#8xdIz2t_c>)NkNd#h1ikI~LH(L6C*;z#Y16Ym>S{)swf-@r~h(wS5ue$C3
zJ5Ngw)Z}mxZuY+UpO_Mxho-!uX1Nr~m@FItT!uf~GK5+BAHd#?`~jM3+aFGu&iMxr
zyY9j%hXE?)COt%_PvvayiZ?<HE-DW`zb%eVFja9jNN5W5IMgaoixDsvi$O@xE1%n3
zW<5?vb*>*-#cKQ%(J?srh<B%=Z%KScKULcgFZMd047Fp&1YQ{G)L*^57@ta!Vpp$P
z&x9sc{oL9PzWhb&n3;@{85*ZSKSs*;$Xu9O;3QsXdVEi->^OG?psCXf5WsgH9>Sc|
z8BNA{4Z}A~wlZrGe#;(fWvEx6szqc#8TO=}q!I-ne&lt?TQ?Ko-rX3hRdcePmusb`
z3PC09kWta{+5~vQJSc^weTda@DZ{QUEVmyf;5DAvRfAazDk$#X^7nh4T9{H-`nwRo
zeJ7`3!QJk&6^v?Co85MJGF=%)sU$YQai5jGh8`gJ@z2H0Fyc3Lu2BvGq#OVr;4>6u
zK+$cQdxW*AXOZ@WnWYBbG1YP5H_K!u5lv=?A6GHuoR7Work4<xi0qpj>S+UW>jdy3
z?W0yk{)Z=M=WA4a4@~G>7mD?PO;lu7%YxMiFwK(&75Y4c8Y_5LA$O<TdR)0xTK3X=
zN^~5qFpq!Iw^rlx6sSTMf662LET<^YM8Kh30_>lx_>%RD@>3}dVdYm$)aOV=H+1oA
z>!HUq7Q&KIV$n<T8~4aqrRs+9gPt1F4Yw%Rt4^1s8an9&LKF!~)}Diqw&x~SRD^mR
zfp+f~q!rthJW9v{vOgs%KhQK<p1a2(GQ1W1^gsv&3F3wMEE-0<7vr&uKX+~Z9{S*=
z>xFU4pHU%OWJeOoTset2ZxHkv&MD*TD@u3!j=#^m*lY(LTIS{ZmabsZ^}LGXGK~ns
z87s8bwo<xRtDR+aSS9mOZ#B<QK(<h-D(t5u#~%+21P1@2pJ2^dmCMNEy;u#$gM96n
zM~v|@eu}6LMYR>l&B}%%n-LZ2eYx>u{FX0!Q7g?RF679QQJx+uYJaE?gxVS7mwBh$
zs`SI7c<PY$>uj?EwMvxu0zO-DOx{^TNe{n)S1TbKx7=8%)jRY$`X?`adm9uU-)?XU
z#ZOcJkRL)Jez<9#<ghcZcO#G{WfKq#yj+Szp~#j5Kc|RSqjpy~Ut}Bk^Gsyv00qV&
zt_lJcW|M2^@OPr$x*y49^erKAitOuDGvg$KyI{f49RA*Z;*(V&`Jw2n<8Obsh5RTg
zJEG=ASa$IyP^YSs81Xw6IdBSYBJc7TyB(|C1ZAN-j(j>?kN`m^VD%m&3B8T?kJs=<
zbIRz(K3MlMe1`8sFFZC=Dcf`f6x_8(-Q(0+w#HQZ#UtVxV`~J0gO*1jVXD71$0Cy|
z;EJ4X)Jlhu^tPT$WxJxfF4QDKN$vb7^sf7RVnZBJN$F&8*gk;?z*MYxlhTh!8W|VS
z+#=w)Rv@_9E(;rQQwali4b1VM{*PQiLIH8*J{l*<EMyQCoMuXOgHA&|{y!s(gIV=L
z!`A|&zOyZ0$g$KH%7nQ$tw_auJS8cZli{P|fcivr87iIP+*fuXImr$hgDM*W5{6W}
z2<jN8KnYH`3z)@MrmtKk1uE(k5gY{Y@VsYz3P@M!q~L`=G|f+lsj;(abx(+&2tN)R
z4DoX$$erJ0;SdsUW=9b8VVsf>q}45eTSA8UwUx2eMg$|~lt^LeQ=6w{uI@VLldduw
zKaAa<tYl?%2x|(tYsHr#P|#&O3Shil_#hsFXJY6K^KdSjoUq(g2{n@Qpf`4jpQ1<6
z+pA*Lbf8t$!zbzfyKe4q$gar&^ULpR@dGYIB9A!rvrRJCO&;cKIEU^o*j$d%1gKg=
zK5Q*LlXn%-5Zk<t-4~qzC-ddP3K~@)7&%kOcZ>tzaj|HEC$EJQtTiGPJ{$x!@jeUB
z@>AdulB3icDtJKG{_Rw-GIvy;E!u+GD9q{$l2ZcBMvU_r@lbB%OG(uDwfkS%oXUn2
zb&Yc*OAx02*VqikLNS|_&p#JexLGbzlNF;M@i5b=J!_J2!CG69YJa#g!NZ^Mb-(r6
zm}zGdX<-S6qm-LIMWCcN%#y~;?31k(j5HEcbPPKe2uA<hJxTil^L$(ga8zx!`a!;e
zbH&3r2asp2CRz#agh4h7iuB7umnGQqviZ*IO;y?ciXR(r?%2?ZuR{ERJ=$J}PG<BF
zCrH|7XM(Y0?}hxv9C}}*QNJ&VkRl6@pawLAQIX0(GJ>M1xpGum<S=$iDx)%x&DZ^i
zArWpQ1LJC#_Ho=K-h;a27+;*%B<&w2+|kN!Zol}29SMFFxibrdZza)M;s1{R_OTIc
zIBeadd+kL%s%~YPWTKLx)d}Y5W#q4@qyfFDV@G^$)aO2*n8nFejvpxQk&ZC8zc=Lw
z`8MeSVC)x0cIw?X8m=VCCU%QmGtboU{v@?dmC2+n-H(7qoi6|*w~hVk`>Bmw>BH5v
z;;+YCT~YhB#sEt-Etq!z459Z}oNeL!=&OA-Si4Pr3#mTn)!(YSNJx{D7)Ya~732(n
zhG+mv6hy~lZoSUatz89BW)KL0v_rB3DWA{+Rf4=(hCaVhDn1Uil+iAG_z=VBKTcoS
zaJ3be&r}nx_7xT6P)35{D9Z|vi*-TOaP;6}T6i_KobKYIANfQjbE=lEuK8!4MVMIX
zS5P^?gdzU$j&O~nS|xqu)A`DLo`zY&_+V{RoCdD>*HH{e3I;LOJ>VAQk#RK-r6E?*
zvo5O{Y{z%<6iBq|Mk)<Rmm-yuAd@`d6oJ(H{aW3o8Cj3*OYsCO4W48uy5_5`MZ+D;
ztefI!8o9`VN|zo>8M}+|9$1XZ-Lt!G8>8M_=_-^UfAit$GpMaR26Bk84=^K@(?^+p
zff;}tz&R|h=OOCpD)h0d&EEi35Jd5nH;<4v7!H}xa)iB?Dd2IEyAHnp+V6~a)~VV?
zztiEiY@n5~-&;Z+X4ZM`5#&MGZ8Fxyt29&dwx;~tlRhRmkhCZ10-;;1{#lLxXRES|
z{VSCNRu^#PvwdrRr;0jNIYU+jG%s9g!l`_N>`wS$PZx}jb?Y}q(*x0oCF*8%<56*e
z_r_-XDw8kv&|P9&IXgwV%YApfkPG{HzmlW&t&DvrpL$ZHl)S138Swl`6`Jtuk~zbB
zgT@~163rXTs}gPHzN!2YE(9ymW4;P+UT{s|>}6DDkbIb>1tC8Bdz@S(NcZ9!2i~|Y
zIKT7;ANcNT6H6JuRuP2p_aYs`2AFIgDzkiSi|RI-{B8gjW>=9c)xo2oJCUeqtz^Q_
z?u@CBCl$HkEB`IvtzsuPDwP+c734lEo#IPS3Ip>h<<aT0mSj>?#=x&6_h0r?xdjL%
zueG(EOXV<=t$u0|jVHW(A$3jMSo!Nm$@?fH7ksMjaU;{%n>1@tjh-xa78sg!Am>~V
z9R+mwyH&b|L|@h>w1D+NC))BSup?k@8p>%TI6#W@ihtQBJ34k8oNQ6s&ahJ&1xSwc
zIZQ$e#8JR2tZUR2756)HqF!aH5lGIYj0pmuQU%>A!fiuPVni$V95JRdJos2TYgK+J
z>t0PUxg)ECLV5tad~w>Dna|DVxxzT7uOy+a_eEF?o$g!h#HZ1dDcMkKq@^%@hp+m_
zqL-|j_0Ba(x?c_QsOKK{Up*+#J9D=X4HEVVOR-9g<l95OeJ`vh|3w_+K_5;<bPq)7
zvmH%*$>AL39M)nLZVPXRPvQJ335z6dbicyc<x5?lr0*24WA))(rSR^N-!%g3lv9ew
z5#G2*=Q34rEVDzpKtiOqEn~_BthA4S>Piqod<X{!@A0CF^$6yAr?@742?G@(V@JVZ
z8d-<XuWVGPkKBAhOOI$K;V*GEzewl0U`c29bih`V7zN&dkR+m?zf^AYr1g!myE3O!
zVuN|dQE3WrC>}V`5TTFJufuaK$(R}X7@=W2umoiwkg!HUWiq6uMA?;Awn%F-Nd}Vw
z^Zt3fi7>xrZ&WjJ#{xMllxS?My3VBHtGawKdafjUW`r9mW2uwYY~uX`yVjap^-OB(
z9&H-O5<IS{@c=!OFy(_7<ye8V5pRXX2Rw<hT`}G^XiHMTb<AF1cF(M!mKY!D(B=xo
zC~@id!($%b3PEULegbg<8mw40DQ5Mg?o}Oyeqs~THo4BSewz_($ho6pRh%m{p2Tqs
z@3m5Tq-@xN|E?WJR$WAERN?a~R97%j;d1|JPp}o2$3(4!*)@TWPV8?)z0&n|i$dRU
z`2qag@?T8`HTvg`#Rw#fqS-HJUbiK-n81oE$Lx?CB*k{|BTVC8Ae_C9KMHy4iw(p2
zhQFW6fuZ^(d2sI70_l+a6bbKPzqWO44oyy?^?C27vhwfht}(^YO&&>l)Sg0Ca5w>T
zCqm%^nTCFx-}#Wpk7YG({{gal#wiyJv%}hZP0hpX?ERT_(6MSETXSsL(m66^YDi`5
zs&7J{E&}15N0+L5+1mEVsdsB@5*PW08Op5(O~qc$y%v6IbZe>Kg0*wg^qDh%PG2~B
z8(5_pm`$K4`ZWLKic_la<Iea8$T031nzhOmw;oI1+iG+8aqnmGpE?5waY7xcbFA@R
z8O1c&9FH^Vx}fto^nWH&h>V2sc?pUghNR+WZrSA&?!cR+Fgto6(6e0tf<h#yG)l7|
z>ibYlkn7&{&e)@+hBQaM)_xy)O3F9E!j5zUJR|!|26zwCxMY=1X=Q%PK2lADBmvry
z?SC)IrFaqn2#Jam0=CPLcO5ob4E*-0!tyc^o@3EL92w;>L*vpC+@2c%XexayY)M^{
z*Mv1ee`F^F?h+>#bhvtdc7$VEnAL03NBAUH$$#hzvIayfUmk(B-^|?NE++OLUE13n
z`H8=c%*i<NQ=NE&wnq1UkAMHe4>$2ZS~h#sm5dA_d~Z1wuORy`GfAV6rT9DZ;y=Lu
zdu+B^xFP>lasaLJq4_`250V!R?HZRLPRQ2K$I}|?1f1D88j47)-!(yACj#MVlxek%
zHbsgSsO4A_Ot1i1WS4m*Sn5TjzIvg!GFGuPTb2k6LIKByL_Mf_ES&<7K}y&?ItS)x
z8wgC|EAYg-x-coE()<IMD3@xyAD`t%x&v=95;%`WZU(Ib%(O<ZNu_c+u``PK8JmUY
z?r;8Z9~}}mY0_I~v8FJNqdvh4M3*Re6-D_)ixeI8CexI<q%bAPXKniNM4B&9@GAj=
zsCyz*Ldc{y>3TfX{k0|#mF=Gm3hw32<sBjS^UC#)(IlgDm@4(0y+q@YrMuiT4e*Mt
zn64SZUsc;M@V@NouxVZ6>mI60AH!t~3Ipd<|053kpJpwT|Iws%!~Q=@w&1eo|F<#Z
zzkMRqKG1y82HMbn89?apnRk6jiYS~CXq}vt&65MegznE^5VYEQ)mJggr)Rw0IR3v#
zjKz`>-tdu;`A_;w24{u<YB8&<+Dh-}wl6(O>YzBbGL5UWxbQX#$jnm0-@htypQ_s;
z>ya2DI!)1^<NY5qk)B93Hwid;y%@CNT=~}J0GbJfa6866XDsr9lJkNawMoO<9mxU&
z{5L_fEuXn%@-2*7l?p}j>2>pA#d77k4|6)$y>oS<Z6d*3-wDC{0~(c1_Dr9(a6s1A
z(LrDkT@ZYa4>B@5M1-SqnBCZz-`DbeU=KkOoL9G(A%O&;5W2;}v{G;qjFgHR9)S(A
zSE9}b+;5&aUNkjEWeZ5U1|7@NqB>*to7EUDboqn2RB;g;(R8Tp*PGWcq-ECjLMqRz
zwqVpb$J&FCk<$Yvpy{OK=s<j8ap56+Qo$5k>8fcYf$)wL;Hl|t8gCxAM4?A+t5UwB
zI;c`3h<Ks=@j(^rvjn4r=e|X$c?z?{=>4u74;@Ua*O-2b2TFUU(JCoKcSIBFN3(g=
ztYs!CM1>AgqB&$p@ZLwPxE<jRtwPYqnoWvAV^)}j^(>TNUDPE{MZ3F89l-OBlIj+k
z^1;tz95~{@?-U0}7NAxm+-SI3^EgEGzxmu^C2n}<cghu=#Sy62!&ia(kO{WEW)%yW
zQ^r>(9OaokH>4v#ymy`IVEFwHy>{=E{pAWy0fw!nw!B7)2!|kmo85*m<L!|bTN9Rp
zVOGsC{*&dPERU5fjHd~ZEO@23No%$CFtzGcd$`6(E9i{ELy;mX;#OuW+(BDuad`O0
z=3s*|SB3snc*DEQ%R$S`fz_-&{!B*4RkAAu(dyl*(;D~cM?!;O^)O!DT6Y-DYI?)X
zG0s>~Pcb*1n6HM(!;<F96nBS8zfAs0`Gp>aw6`HCk3cJUmO@I7{_=O|O(j)&S-LA}
zD%lPk3D5=vj1)1`s~Qw9^$Sw8s3DsQAcodPw0VCk+68-9tn;UNE<YoLs$sWi%$Nx9
z?PQ_?U}i4TRhoXjc*mCMx0iX+%gk_&DD><q`<J{tBY2F=peGl?kW?AAW03&dVK`;r
z`RfGl>QHS-%z~x%2G=GLeX<Xhjm9FY!jOdloL|N2B#uE?DlFS;`i6XC^;YT1Ou3nb
zaY>$;f^llAK{)F`@+7M(F>0M7^DoiLMml;N;?$jwUk|{*8{j>r{dR2+&-NF&-@6x9
z#OdZ-hGKop8r7e3omCpvj7zgD0gV3-YiAYIW&?it;8I+QTW~E>+^tXuPH^|)?$A&u
z1TXIHZpGalN^vjl?$&Sf{bhD{W-oSUHW%+rGMP!{l}FC`ohj8r<fA$Iz{L>4dlJ^w
z{pQ7a7M*a=9WoXiw*V({lPDiT&1WAJR8#^<i=U=5a+yN0xxQ1Q(Sr4AefB?R@oG#S
zMU4=2Iz4nVnc~keb`P8b-NK{KZqie~V%l(eCCA$SRyA0zWT`CgL*|l@k^9jpL}W(G
zcAkk_U}?Z4q(lJIhxtPg%-`Y}rK1<ick&WpCb_$Bsehwfyc2#7^rV64@m;+ZRX+Q>
zWH8(gT<x66oX^_T%rAWp@Kke%Cm3hBpQoGAB*4ML3BiO@pp-me$?I7$i?c#o-z<`8
z>#I*3@F?DgMWuVf0(h0i->vn6uWVn|Tz@W737|_wh*8stgq@q>zHM?QBV{++s;B)v
z<cG}}OVuH<xu~(axC!i{-Of#C+<jHSa}{4dyQ)**t5ds^G=U^=sfl?zeIf2>rj$2T
z^uG;cU{x6J=8qnSK>UgW2C2LRlq#9s=Ru^%`q#X!1i|w1mSUhVK_G*6_m>}Lx_0j)
zJK3ch3XhO?@yq^{li~wV0u;r}!{!eNA~f}q)Fk#<!b{^?HD&gWH4^<p7!0-1_*4cB
zEoMN%S<*p@k5PAwsMV<~@uN(s_kvi+AGRsz+vfGyhlJJINnbP-{$gp-o6Nt>ukXuP
zeL`qb7#-%8PZcIh1OQ}2KRNCjd^Am8*IpB7YOe7yN=uM9eUCVh*<VL-=o|aR359()
z;r__xAAt5Iwzp_@eM$f8<dnJU-}#+Sr|E@HR+BLrF^Qniy#vlzQ(QFf(d0nC%h}`T
z6>naa7O=FERo2dd=R<5dAPcE-E#fbVa=sJ*;B-n7Q(|s8Kt9Qqz7-rw;J(krqAU0T
z7|AAz@n`QE8weT2L{VBN%4!eMV)7}>B^HNRD|Hqq=q+kF%_C;;g2q`~D0|?#e+I-0
z%}=a&U-HvNBuPh$jR-`Vry4ctwh?Z#eMzsFmBp@Z2`dZpy~p?ws><oV$~d0rWV4-J
z-cIS7S{fhbi|md>1tW`_?5FgX%REUO<9sun@ri<~yT#U-By`d%oaJ&LAbKUKaxpdA
z?s~q@s>=UXvCn1zhilNk=4ao2GdCf;)rb(fbn`JDTp<$BBWSyf`+(t$i6ZtKTm5DU
zoleK;lWup^030e<D!?OLJ0hK^GAnar1Su``mAQ*kpUWJKi&(6OxkR(3HNDf#;<gXI
zqu|#~@7by-E_Y>Fm$+e?cy`pU2BL99e52@zfa|Wc?h&)GrozC=h=B13gZ{JL98&{p
z&ii^7J8)K-i||zsSH;3a9!c|~370{45gi9g;_KANkUSx#7K47Wk(B1jmR8BObF9<5
z+!lYXBz`wkJM(KtgNbrW0US+ga_Llb227~$u}9Xb#qX<_J$U*~w=}6Buj>nq&(jHp
z-UXh?*ra_r+e_QGYz9KytXY;=^`UT@pyH?KPOp=Xnp;K!P3Wr_y8P(cjg;<SBtb1;
zW}g@}Oj>Au#9qc0e-t#M#)~Q%B@G7&{FS5LskiME;9cTL&?ZJ)0Jix(xUU$uMiJMV
zk$<=GoB9M+lS14A8h9eMF}Uz_5$K6D?^Up#+^S8uLiVqp5XP)DRiVW!Os))4clg9T
z04k=aC&@mn@NIrw`ta2>o}l0>G@kUf3wK#AtX~wUenMjiQ`p-6$M=2d&(FaJNFv%w
zCu?x(nmcPNIFxEs79A{i`AGPcM{TkQuFx#?L_mO0>{p9(<_Hr60tpFU2;>x5D=KNG
zmj{<cb{h5Q>`OW8OkJ~Cr%;J*!EL6S0Q2-*9|M2DbfsU&ohAv6#ixdr`i4jICc+%S
zD?a3N*sMnAXtr={-=~3`xq>jO&|}W`jPDCwS_CpavESV_wqOYi^z)dcK}0x0W3qSm
zjZIA@XB}6tEoS_XEPc?&3J7#hvSQxuPx)5afke!jWXUpHZc50i{BSP^+e*oJOpb|V
zB!7AoRW!&-mV&a#rgA>DmIG-1<Y&JIFOFHwc(3=yXlhGpFbD}S<bmj^9K#B)`uD3q
z>MgG8aN_ewn#?kr48xU0x~|~DF|!D8aU(21DSEZN+QHPKJhJ>vW{eUzMI>;`1aoE|
ziUx6!Q}%@6S9&>_Eq|=}@SdmHlG+9b0X19=+r+5`c5zB=+|oRTJc7-Dl4`Kw<hwbL
z8=Qog8VP=Sg_5w+$Zt=vhXdLJ1A<b$jlmQB_T&9zYeK`t2Dh#J^5C`gCbHX<Ir3)b
zoQU|t(v)W@CfI!qq3qj$Tia>>0j`atd{lN;94P-g%*;*PIUxbJ$(9M$27zX;Sbko8
z$?NNI5y5DYD&y!>^Av_XQ~5bFa#9ga08z7LTdckhQ7Jx&77L|yf|`!=tJBK_gU{^%
z!2V*B!<OKcYtjb-SB&MK<`xCW<>)=|f6&%89-4m}APzFi^qIgaJQ=xMlyP<~gO>d-
zY1_~UYcT!6J013v*3sp4g9Slf=E$FVY)DGejE*Kan_EO{tksPp<QBY(L<D@wV@R$=
zwmnic)qQ4AR3{_Xwe?kNf7pZmU63zoVV+7&inLf44S*?CTOJ)I5e)B_5X*Xbq5EW?
z;j#6Q@5Kxb4*jTL7R!}VuA|fF`<29Fy#>G0Qpl#qi3St7k@wKsU?J^BF2<EDbd3ZM
zY(d$!|5-F8<rsp(FS)4ek$5UD#vy-0gF<IJnPeH4%BM4SXn-am7*=}HU8s@5HmH>|
zow6PtVP;Gm_D;uCc}_H7urCS`jHgvu`_BDU`Hi>k#fnP|O{<>pSJ|kmZTkEicUX<h
z&1dA7AQ@>(?0ESIj(9pkLe=me!m0c4kbnX){Wkm4e5)y31B8~sJY~2?46mr>^qPlT
z{x5$@cm3H)x~)O+D)trRzr$!5tF+(REEhr$`@0;Tqh1CDj5WeH1Unt?XWk+U;o2&?
zjR*Of>u`={&8xGD%Cw_ZnWHL-j}i;UfF25}aJLk(K5Nnh3dqb51)5qM%5PL`HO36*
zr&FPOB1Y3`YSV7FS9+f3tZ-Qw2Or`r5}XYX1!EjM_Hc22DSC#8HMMr!t*Dn<0a>!=
z^X0Qc4tay=<`I>z$_Qbj#c=}6?gYOiXgA`<)M0R`%-xhLQL2x0Y7ZHeO0qXtT_6=b
z-wH3l1OagU+7<7m=2a*3!*l*uDp{+f>4ycxdFBu?7)olrw6Uf5!3RILW|`zGfRdN6
z-P@$jryKKJn<r|}-*7{0GD9;<u%(t&f_aJ_9^YVCd<pN%Ft1V`zNXYX{#JYOd5>&E
zh&F&yjUW*Lf}MM4X^t_GJkDM(BSuJ<rrGtLiA@ccyHE9D!0XpXu{Z)S7f<cR_i{fW
zHQ$-O(9c9l^Qt*I;pPA7;be5zTT>t8ezS*p5dO14ws0T6tJC=Jy|Usz09YokQBK#g
zkHPR09W;Dp01Cy)`Mr8)vH3hnb*UoYC-@zP7}w?huVm+*{|%Jz{}9gfKf6v;Xb_q9
z*IDxavJg-~o06W4<dc~eamB!<Iufq+wyp2a^xeW#aC%|#$y&w1uWwuvUsp5!<tk`9
z38=2jGJ5>r8gXJK7XyWN(EIjCwZPxPQNW_+kHnmK-DfAYd6k+1-Is2!2Brk+zxRvd
zla93nnN-;ceez(F2dCCMTdK1BQRC~8o0fd#p<E2pxiLZ!8FyM{x}I6(HTo7&X?mF`
z7?$vZKR;GA-tsFdvz?3Inn(M|2jnwJwejzS2vv)Qf+0eo!P5&?+~kz)Gy^k($>2W1
zBY67X3;-@@_D?ljtp41!h}Lqe_SvF+u{g9S4Lb!MTE=Cv%6DN})hWm_MeJF5^0|pq
ztrCLj_+E<+llLCHI>uxfd_${A!8nDwlLhWF{?q^}=RU(*%J`k+%dHp{Q)Aoj(cgKi
zT#)D}YXu1Yr2aL5goZFqs$kfbmvfN%n?I<WCvZDC7%)%mN#HC%JaPd`w}83I8-nwz
z9sHuEX@c9m_vVjU!ghTVVIrf;rBAb0zE2+rrQy_>&>-KWE+lAkPp*{|ji_d3FEL==
z#6`cr_K*H3@2brykg&|YS%Ua;0dQvw)~{lf0g#fA$Md*qC3yY=P!Bp9y6~~kqOst5
z{GY~UQLuwXlD2G?Ux)$jM;U~izULyo{-2E^yG||{Fb5FUh<3s0shVrpB>SfvBtFhA
zLnnhtMrAzkbbqsq3`CW?<OcT|4Z0P9zT+=DK~beT7L+AxW_FZ7#Gu<9UhT{``Lvd|
zV06tv(AYLQKeBARt}fynuIhH<n=4`P{j3vmavCv(JF98_gh}hw^hx<u#-K&l$O8h4
z6G1@rIO=-2pzSb^as#PQO|1vyNSA2NyBlaOGQywe5@RTmPwiqjV0X}~8JSYuC{RU>
zTQD?^d#9L6B?}SVruL!c>g|XjkEUr5h}#B0Zf-FFpP7{@!7SCB<}3$&Lv2505CxST
zRhfw2grK%I79#9loZ_fxt6Q7$8g)58dz>u5f7`<fno#>U9wkAAH^$28Y<0+bh9|<o
z5~h}}Uzu}rN<UT`oNY9d(yCYb!zPm%CQXcpu)g#ab%KEr99h?tJM21vJQPc#xySup
zs-;w05x+DRiqO?LfU+Q8&^HvCt6sW{EMtu>#M-YDIrA7xke7=?VS^6$luMd5=g%kA
zb=JApxZWjg@+)5}LZ=*=)!_k;AQ!IgE>2|rea3XA3TE^3T9k&jA-Zd((wG@E%oz@D
zZmz9PIXmkv0*2p-o->g7y2TXLzpbOergYXnM<MLKy*$#YbB+N|PuX?ku_QDRqa+oo
zCh<3^?}R#X_d6khj06@E<@4J16MumrnVltn5bu7?R)gGm2zq0Fg)W=9JB+I8@(cP3
z!7HA+q8^1oaTtIII`a@NNs~iuR-86`l<F|aRx78Tw)&s4!D|Jq0S4i4-1Wj&Kq=MN
z8F+~My(0T2c9|pRtQl%CZ=)3AhtOYuf-w!a#>F}5(ub@7dlxX+=w)LH_v`JjXN6py
z!j>0Gsh;^qt%W0DABPdSj+0h#APSYAEXOU}!D2`97>w8;2-1K&C3>~ynJPyfgO4R;
zab&f&SF>nlplTmD`d7<p!goNIrvX`Hgzv%a>fs6=Vb+{b*z5V!TFjs!{thhk4?u~z
zctsjzNU<(|tLP+gSfM<uujrUjRU+uE061)(3jVz#!q<sD5g0KK-$2ID=Mm$?1(I<`
z6l}isNZ>Pva!*#Qa@Oj)|B_sBd{W23#u<RcDc7C}^N@dP5Xjd`{}pzrOfN_v@1D=d
zbjyV@B4rmQ`cVT{*yG+lRjZA`>X<`Xb50<ve@!sLij~D8sZNcPe;z2i$MDvkF7QcI
zWf0HM>L~)wOwF^wDq2g`|Ghk&&@TY?0_6*dUv=NA`es=I#m)=LZWS$NIXx~H+M)@7
z0%VHSRsX(WnDv^@Ue|1?$$RuJy_hRaP;hmzS+##&Bjy?6LE%(+IB1|vR>!whGi2o%
zpgWI`UnbF>PDwzW1q?H(w){a+&H?D3n6J(GftW8ESmsW}HtF&ql(>;JH2{;{gx=EJ
zHl0O?5OaqqTyzg{J>YB{%QAMx{1FqG`HP&7?PRMw?9>N=DlFy6u6#W(q3LD<e&1ML
z#GR7Tk|m-&+}s$}75AdNIFk95O_<)K5>b|01Ap4Z-!~rIEK_v<=6C<edq&ti^%&<b
zy}dVbRClK^*y)=LSjS_oiU?zZ+(4gvZM{8-3<P^&4~|YXn>2CeclvsA5-q^|psiS@
z(4X)O8%4(l-!V2Vu$+#aI9v`c&;f_@<8TUH9TOUW(6@%hKtgc)FmAq=rTrPOi~ieq
zi}3)s?>og|AI+VgJ0{3)N)Q?|`SIs{V9Qh0@8lSi-b^P-bI?;C$A&(yd<hB`Ir&;B
z-W{`IC!DnTtOm)zDep7ma6<8Lp!X2#L|)cRlgpooP|nlhiwl6goKd|FR(w{R^XUVB
zxgpnBJ-YTg2AL#U`u=_af{zGeCOB~SA;VO>JVtkJ2X7y=gk{Uj9%Tglwwk_1s|n|M
z%A(J_AoVCPlegnMc+woKG0OeMfwpa3E~EiV_j_+Vt04V%np}6VH3Q~&3~iX!R?=}}
z&DFeGUw(TEhbG~B=h6qk3DF_{-4-|UvOVFHF%GJaZI1&9)DVILz%MuVV>sFhko9Kx
z?0weq+WM!~A;QnGV*Q-Y=(oDBXEqTvB3u-e8Ll^2_u@<wi#Q<uOpV^U?SBA~ww6^3
z0LnIM*B{~(<5+I}L#++z!~NE#ureVpW_gDePLq+-M!IrX{W7#IdFHN_<9uCL)Xa*}
zbMGszI$W5UD>O{{D?q706dxC?Q7Q9gshqhnLuymJ|JaD-l^{IKrC?><D3r}Uc@#24
zOA#|rbW*?~maXw2ZRb&pDq3TPRe~s)IVfYE6F*hUP}+lQGVjpvj&pJ)5S1pjiAS$v
zoW&}9iiI57N+U5uxliR)8e6$s|L4RGBhNlD?UU3(uS_qs8X=UD<RjdeVWK+)-CbFS
z2I16cphu3=v&@XvAo?)AS8o?p$4?77jN&Kv<p?M=>Agu_SH2%*`rFpc$(|US_m{vT
zH#!ypUxi4|YBf!cXh-yF2Ik;QkPwzb=x*sr=qY<0Kk@o{Qc)i(>kdMS9iTcb*dSj6
zfZSM}C2EzVg8mojg^`cMQ^_%b5WO7Ju<h*!akq9VI^1N9ed_+tw2g(EWueoTgd!34
z{PGK8yMb!%b-WpCtLj{{U`eI6{4_1i7p@3XthS5U-E=`G?CHqes(CpKmw40Ck2s%N
z+w<^w4dl{f5ksl`<zUX~-&#Xg$U5WgVGaw1mS^jcgA=8Y!<AnIyt{CmDvx>v8LTMw
zH*DFh?~pLc&-7cdBylgsYjK&d>t#hIWVO1-adB6#VDdZ7+x?A2YqZI5=19z9h8f@I
zg9_qf>OcK2^zY8YB@o>yIzkP5B&1EKX^!z&M>!Eh{;D|tO-o4DX*Ykjib75MV#4%N
z=H`V@+y|tkN5oTVW6qC5PydpLDurSOaR+2BH(Pn-dp~{5XACL8NXmK5b<t@liVu5=
z$O@vW{#AmPv|Yj%fwJftRyZ@b%sC)19M7Gnvc*iM9yhyilqyS*ve6@fQhPGKmwZ|h
zhIw)EYlL`BcSgT3wm)s&rzgMMxXlCyhvZJTUm+sX>UUgyjfkRID&-PWjd0S0$s$*j
z*yJH^bwgk$u8ao$GOnGUM-n6Xq;~y>A<ih*;qP9$Q<|gNQ#o;aLH?G}V{x92k~Chq
zoQFcM@hX2-tsMl+V+wEgg*H(&UiyI4$V(89dsL+iHmtT`L=K`IK_?@v(ME(qYIIj3
zBi}_sk7NcWy1{fzTrABWpfViMe?8j(1<QT2HY?F?`KIvV@bE`qHJ$mLZVJ&Ax^@-n
zpQ;rB?^6ug0y>0z9Cz`~y7aVAB1P-$ZDLyF{IR?;a(yA$`cUfonSirXK2fom6i{<$
zv}@%3GCYyzT$2s|`}}q!Kz_cI@E!%^c_KQHxUZYpnEo-vk&11UfvSlUm5|#fOnF=M
zlaU~_IJMCL=zf1Dvy+zAHD|MuUxK#%R^BrE;k%_q48{1ahG0l}IS?p;gL{z+m&#^M
z);7Vl_kn|x?pnL~OPbkIOA|8^FvA}b$9oRv>wf>axErn7)Z?q8TltaS9~$x`2Q&q#
zL<K~Sj-TA+tKTqLXN1vD+nG(3%~4<1f_o6h;DzxLSM~wv8k1^Vl_$k*ux@DH*R|gH
zclk;za%okks%D*iY-nxx(_J0~3V(L6dXe`)rp$I<%Y0cPFX20a_3{ZUde+87DtDFY
zJtm*eyag7ijk4t;J@Irb76@3w5~P^1nPs~jJPv+85`Z;rjd@L%aP`%!bAYQ^tR53i
zRIE;GKDL=|utwqJ*=cf;ns6OXF3%S==z$1KcDIH%Os4v6nj~@OPh3yRahS2Y3`(u<
zWDJF=z&S<Vy-z~hK%kdnuReD8K0lYtAZMcRNiD=PW$c`I=PtD~Swu=umsLHJO2|-6
z=p4i3*qmx{cFa8OG8&P~dTBN@`RN432DeV3tq9jKO{q)W{iv(Io4MnLU_$R^NQ?`o
z&^$HK%7|<yaSg>S!Jtk!BMgLb`5r9oKnSLQwMV(<hR>5Hcdle!ErH0NSoQLM#|e`$
zSo&P{T}l7|#i|mXs<?1CH}MOK-xhWse-rN{R7WC$px1QN@McBT>M{MbCzZb^1ilz1
z>vA{IzL5H9(GmReGpeEK<$CR1vXRouwq<ZKIPrP>k|%Hw;zv-a{l0nVCT=ugV0AmO
zbb(DSdd43N{j>KF;iwAu6;?U7#sBW8{#29OKZ()FP`?d)rKZ+97B9g_sm3hf2L*VL
z_?h<;kZG;)di~hcX>D?3*YZ5-AwEj@6;!SMaGR`0s<m{+itcH!i7^PzuafE;u0RS$
zWk1zr_jx_beQ}xU?BmX&`thKU5aT@yG5RNmm}aBiF}20#LiFLV)DvZ7@o}d#iuHma
z+wLnu%@vtrVjsbGoq#)zH;aQr^TRz&>@o_DHYQP(KGj1gTu=K+)g(qNQq{;6!4){s
zd86gKOimrA8yY^&j3>>KJ6W2^^lYTj%HTf$oJMyg`Lo$VLvx*K#2Ye?WFx#*lnP!V
zV1#@a2jqnL{yyD+&!mcxn8W8Nb*HMr7I2}f-NrqV3KALQg#Pt}SiaD1`Pp@Ziv)e&
z>%_mb)w0cQRs9h0!;|S0MPrPTKyzo}8u`q4xxDB~gdfX*eVa{n59iPn+a^|bc7scn
z$8mGo{^V4n1iVFI+F~T=*T33v!))bMp+j==2PugY7YNAsp`c0uxW($Gdn)5ic$E%)
zguftfR=(pk6x9sF(b$p)niV3f@sVQJD`Ii7?A3nQRj_N7YTrVi9-t139DxDUJ*r~b
zZf{6R5-<B2=sqQ?`Ts3Cqw(ic(IrRExA+A}sgZ=6t8~+x)aVp#^ZWRdLZ~V~$q8ak
zm}%U1TxzS|z+;hA@3j>kmRfkI><}C1B*^ov5O)_LQ5kM7S5*2mQ9+rawzsm9yKvob
zV}&HO*u=d#D!=U{{{=)1SKTZ2XdEG0Vsrm|lj?K<eoT?&yx1C}t<R^TwhLT_DfC)m
z)9+&|r>VZ(YH&6gZPX8yWv%IyGk>&7Ezg7S31LgzkPQ1D*Ty)Hi{w#~@R}@d$y%JA
zvPEBYjFGJiYGCyNO>i*pGrx^;-3x;8SM>2R^*TTr;xs=^?jg^RF`5>F(H$#?s``%P
z8Ds?p3tZ=25pIIu+7BE*MIn6Luf?`puk*^y_m}?w)=Nia<_kFmmB^yv$3vHBAEw4E
zR?>T)OUuTpp{W4BjrJu++XL@*|F3i<eRZUkQ_;N`&y#rF+_qv$>=4H73}38mp$q!N
z%YFJul@#=ftiyU_I#3@a>E{dcG15a?UN>c`#r+6Q&xYpt1@I6)um7Nttoo+&EzZ57
z+Qh7q=1Fn4Zg+e`gcYN0c$CSaeM^{<>o|r}_+oy<NEh<OSeZeh;`wPV+#HIbMMByh
z_|Zz}czc478x4SiFVE+J<@lXBAfHF>eXlfmt={B<q?hNN<|`FmZ=hMR(uFLw4MMX|
zYx*_@Po|HrA`;POdJq`B9C=U3n(4F{?jB>D-hDAX>u6S~d!s@FGz(gdjBP#VYbSRp
zpY(PgrvRVK;n}_^@-V%YXa1q71uVSz-ln`ZQ#xe3c;-+xa*x;Lml8Rm!}!g|rPQi$
zI46sSRu0Wb%h_Y)V+RZ)P83=KEQgVc0Or%q;2#K*i`6evFEO~q{{ZN}#&eE29kUt3
z$>*iBsD6O$TGDaO4uov?nKSgN%oz1moGyaJvh)c>Jc`6H2kmQ@o{itB7|P3iPz7x>
zGyD<#gNdCD5ywpm$ui>H%4U?-lZAzed<<+D4kX<`k6<vxM>olRQ{%MZKce67U@5eO
z>YB4Jc&qCr_B>t1AiMq<IyWblt|>0@{M^q1ivL!fW!`RgBS&55K5rzOky^9|2ls?f
zS>v0372s{Wg<muOWtizH6v;hVBf7h^Cu4iS<iv@B)(RZ81&2vo!V#S6|G=<?vsSW`
zDNL&PO}}gO;(VP>n&F3-Gkm&<=&ck45c|`z%k5yY+ri0c`o2qu9Ycd=^F7Pg|I*$3
z7BGKaD*s#P^Xle9JDvLvAm$%HjohYv$GQ1yyo}75nT!mo%8|o!z0;dj>?32QXlCED
zeqbeRiua!kRudo)o?!u3Vv~AQ>m4l{q5UjRm=P&N5)Tq2k!yMg&%h;pmFydcohHX;
zZ``G6grJSiLnH63Vxd!(B+a}k-FY463`0`|DTmv-^jo2n74z8f$`?$N4SDYI6a(l~
z{{YZp9YYodB>2~)D;FqnmWuC?-|0nQEy8IQsv!g^{9xU>AzuAO7T_bwyI3kwBePjP
zxcq)@80_Ua5nmyir!J8n5qTZ+oofl22=D}&Y)asw)!S1yM}L}!2o`@f6#xz{3WajU
zA+P~84iV+44T)gCh?Db*2ce%nPG(HF9TB8uOi9Or!uco>uRt~-1Yv=z;etC8bM&_D
zRI&^E-im@ETecB|9yF$r5ALP7!It2JLat*|GBH5Pp+Mja7w*zY#|7knRptEOM&H5n
z!SI*Gg0U}dU$4!S?7MkQ!x#-XeuF*iDH$dS;<+UH%&`|JVoh!`*u;-Q&BWjS{ha&`
zRAA4@ZV058b)w#)RDhH%!j#%Ok;4a7`}OyJc)@?jm$@RN=4*)EZCFW6S(F-Q@?RPb
zX(pKLIK@w#(=}{$#rt2WMHZokW@l>xC$TtA9Db%={Q(qu`9?5j9i_qbG&)YF#fdPe
z2`jBBP8BFKvRVQ(M4X}@2p(6=R#I6sbZ0fE7?kJ-HqIe(xlm=FUXTV4epSuo>2qVm
z1@jkK)Hn<eEOnf`2I790_$IDV4xd{j&-y*_Ux7jY03=NiIma#PmTQ6^7yd0SCi3u<
z)nnUL1s17=q6mmk@z0b5H8hNCSi)%Y3Ke^qRwN9iw6{fMBud9y`$4|3Qutfu+5vYl
zbL^}yR?sfKvOWdA@|~Jb+E1T}t4|8Wj571GR{LXl1>^4#w2pcOqh@3k+!#-n+$pns
zmF5g`{~)Rw(n}QN5lkWfdW#2+UJbQc6S%BNnzy1KlbSmJfwEZ)fE);l@KFp=i7rVB
zbQI31te^J2BN&HP6cy4e*l47VW{n*7uE`lc40{!q{`P1*yZQ;_o*I~|Sg4wV=h#Dx
z67^Be1wj432`6n(?7utER5W0kWGH<#JLHEaSK6kNb`2U1G^a&*lqDqx2BhzylCIKt
zj~G#Gk5i@bV*h0;JCs0IU2>22dRN+=;4hX}wYw|ifa-zpmj{KbJH(RPJU@bAW%BHH
z$6{7<UVX~Vm0(yfLjB)Gm<0YFOnM+GqzKGIXoWg+qO9O?j*#X+MZg;G>kQ3)sdLf}
zuF$zqi|9(u_zVbzOaJ!VO?))QBo<;7k}GGAb6t$Dia3CCc6_EreaS*8D>BFU0-Bx_
zTD@2<_j)W0+2GT8g|;}B^|#9BfnXUd(_rUFXE;L(HMm3gyq@+~Yf8hN7*?JR`U*C#
zYKFZ@2opF%C(9#ACa57v;;hPNV#wNHyMe#6W8(5)`IA^4Dx0=_ZE7VR-~pns<e$%@
zyT`l%-Pj>IU3K8CuS@o11ZEb)CVbLF`ulKjnEnlAM9`M~%lDFwMt?~m4*qQ<$QB%N
z&g0m9msU>eYO94+>7@w0veu{o=%W!$6FtX6?JXuK`aN#oIDdIlQUC0~3f+1i$1DGz
zdPl+V7tutKo|O~)v4Kwwjxa~BZ(On`13e{mJS0A{85q12stV}8M?LrQKpTD!P9pe1
zuRlo6y~LK?Mi3oG*qa0_;R;KX=n-a2h9Hc&(i221<54SJ)2ULs{o73(iv<)5Ufx!H
z=}zefA6ED*X(Vo&FprsjkOpAgT|cWJS&XHYZ8MI(to7CmNjhB^E?6$#i>uc~y7`&R
zR8a3P6{5)9N4x-vmj2>n#s*iUqfg=IU&(6HkJr}gr%DW3MRcqO2;P%sJP4+4bfv5G
zEq33Q=<DlpZCD*^S<Mm8dX7}x#48Fqm&K0%0ZJU2TvNnW{se9j$spl?dF913(|URz
zL&CzLU>B+=qT)IwaQW4~jdo!WAs4z^MrmT7+EOCIxNBSApEc3EZOa%=6}Sk`&x;F1
z6AVBWsB3C(w<OhR-1-r$=kxe!e8U}6IdC~MLgTqs0S~{0qvOZG0wBBGI9N`EhmBkT
z7@`pj$Kuq=qB_nCcSC$y6PGAx#*pGVr<7La#Vb#EMk<rL2g`5U5by7U_jm7AD1wDZ
zH75?Xv<SY-&Qboywdmw~8@^Y%(^&{xO<>2`n$w+vAX}U?Nk~w{4{kyIlGjmJjA(*z
zYJ>?oo&TQtfz;f{Erp0T4W<OdCy^}Z?&xomruIlUu5E1(iGocGOvu5>dM$S0`oFE;
zCrd_S@yH*=p<|$*5>dAx1ULhpyUzDmdu22_JF99Q%FS9kD-$SMWfQ<zf~sY}a3onQ
z$3Z6K0SouVh$p0D<k}&Dua*w#w1w-Xdmk}cP$aof9rG#U-@LRou8Y?7^SIQ!<9_f{
zAFgoPl{Cq4u(rhYPS~8(HkEYcmV%Tn&(#Q0fB;cQkC=Yk_^hmUc+-RB1@1q9vQf?9
zt$+`BC%Rif2^;l(3KD|x?;RBzCvve`(N&-OfE4sD2SoLH#nN7u{TW(w)YNVQL}(F5
z#p2l(!KEON^>V>!T<<Se;Uq#27)edEV7c6?8h4aAY4wxA$c035(GK@Z2}!LI$2f$x
z2y#IMRqRBoSy=+G5|8YD%~bkw*748DM-p*7yF-_@il%VVL18ac>WL6e0?jL7Y-5QF
z4y2P@+{MZu)+3^)oA1Q>`7iVK)ApCai-|qP?kKyk5I8UT;X2kx^Nf@&J(gV5zz3$2
zM-QIK2;i3p;Q)tsYMPVheVyA}rF9jR-~MJ<v7fA=Gtq(NEh$n`(UCtGA`kc}>JuGy
zStaDB79sp7;gb0SR0_Qn%KZ9sAfvO_TR*&ALiv7li@eej3rL*?(bNv`?~$^&@s+W4
zJ?_B9&(*YB&$UxzaL7hS_EyJcmS-WBlZ8dCZ8(TD38qF^J8jS8H^i6D3j&YYf8a{G
zYXL5{GBIoV0-i=9G5Z}^1wx0GlPL|f3Q5W%2NL|>;r4cO`bbZpOc{B{#j`_A7Kt22
zke%n@VD={y6kPa3xg~hv7e;pPZPo~zjP)wsFFrb0M{Nq<<P7ncM4^@d3~{RmQA2lq
z{49?Ina<Wk0ho+4E}OqM%66K>rb{B?-s?j=Ysz~$XJ<_U0k|efxCaUTUkss)(uRZf
zk-(;rc%r-c8cYp?URlg7I?W!yFb*7Ms;>u0;m&a7z!ACo^{3aOU#j17py&jjt8Uzo
z8&6s<VcMi3d2%2NF`qho&;-pfTxZlypB4v8i(;CVc8ku0Ru^e{5Z6E-TIL`EK`~`M
z+=7L7P~S1gzr_&DDdOb&Jo>)hL|cxCl0@-E5^e(q^Rp3Blw(!CKLdLn6_0e4-&rj}
zm3#IO%9)bfL8V^Q<cPiZUfEfFs6pSszd@*E{DM6ld1sqnFqTbjQ#Xk?JyT~r^aVKu
zUGK~V1Si_)IXbU!kAbveb(K;D;Xe1)0lpEjOT4~%`y(V8g!u3lSf-o=(izgM>xWIZ
z^L&kuehxn~8~R40-h+0p<nu66Lwoi((LAp2gTS0B1B9rE7n2jDuh!(t;-XY&4W5nY
zrM?d-4CST!0x{4$Nbs2G-G<TQ&ow(vbaK(WDCy;nFFI;)6xS(IE`QN@U2{Y<0(qR~
zXGz=*s&DuR(vax!Lb#`@rd&%#Tp@Jop#uj?_i6nCGU47sVD6%qZ4>yvqJ<(&B+_YD
zW1~cs(4!(=DKr)}!Qd=&caB?8M07Z=>ZO(uo(dr?o$?f&1vbs)<_gwJ$2c&jy{0PW
z++W|N+XV7PLOq~E(ZNptvv!#_8Kb6_QZL}JR961Kby-NBOJk4p&d`b%NG=*oq9FaV
zh3|{LS|+WMmcWXqd?BxW7bhX2=0f5gXMkbYm!j{BW}i1_)QtWtw45s_F$E8j6YqJJ
zn>Jv>07QU8IVnolhH;G>5ZKSYY5eTEZ6?{NlF=~gQ0>Gb@p8^R2>Yd2HCG%G_1M^8
zr(|$EdYqe}RYqc|Qgxk(V8#>M>gJgu(79k1hN+Mt_nB$8smardHaep|yqKa5YD4dA
za;)57^XQ`2_Fxbp^c@6HHcLcwhc{gq!gK#tPQuL*H^;a@qYVGAd+fKAE|PwGxiXQ-
zbhZ%pG^Bcy1&mnk{e^yVvjyAG(Xk+W*4*E5D9xN5W8|~J)?{ds;H#Ovf^wiKMa;v8
z^`mG8kHrS-^3KoM1LK^H+Oavh`nkuY6MEd5Ww~dv3iR%z$0mU1TngdmiG;=Q#X9FJ
zbJxOk+H#j$cnB00zX3rckA+`uK2U8UNgBa=3Q-y>Z>RA8&ZJ;6NI0-*FIRJnZwSK_
zjUB;h;$F=V%J6Rm^=B9}Jcxqbe=rJ-6{m@N@56^}bK_5V`$<PwJmQ0Vh|mlKnyD$4
z*T|16!vRRjF1W{}Y2<-<HSg;u(x>T|*B3N9Nf!u<RV9J0Z<ZJUnhDeUab<Kbuk9*v
z%2fwB4So7PCRwy?oINf?e%O*KY)1^&eSK;WX+}TMU5un><rP_anU8ZYR4(>%C2Fdo
zoE>a5)r0RGXwD!3xQqY_Gs{TVs&wz(DU%FUeo5xo+`l!bbFM0Ik1++QF`*{z!{PqA
zy`b309$hfA`gj^$dJIe9+~J7x2?2uyQ9dmPuRjGaeyb0BIMaBLK-{==$95|GE@0O`
z)<w(HGFDGW)uy~K>PEdpPVo`)ou64HA8Nm{^a%~3>GJ-CrK_1}xcaHbZsnuW=&bx`
zzA81bkDS#);r#eci7!L-`>w2Zfs!&IJ$!{2RY^`Y@NXrqQ@2cXwHFy{Q!Fq#GEIbQ
ztSVkiPvG)!Wk5q_*UJ2M%jbFp-$mq#Dx0DobdfT@kM2;)Vnk6=twvY(TCB*kxdBzo
z*KEOCBm_iAF@t42h=T~v4}ja`Yz@Pu3gd1%kIwMjnAI#b`D$G5@8=S#u)C9!8<JwC
z)`kVq$4Q8SZ*XyZMe5$J>OSUmO7*aj8{T5@qY`e~Qd12W1zfP>uRV}6pr446F@Q@!
zX4*OCh1lW}wQSy-&|V5UH`J*-%A0UIql*Y&pm@H=J!zFMz}f>Fxi>Fpbb`0>Kb3qo
z`YF_fX2{Q-(}JF;^yuV<n_-#M;wiE{^{b*33Z>TkfmvEN{R`FDX5=03P;zy3i7*Eh
zeY9YJ#rpf$Nh=Sdpn=tg$kkH7ML<Ey-gAf5D*Cl!E4L*+HH3JV35p6qg*X`=KwD&c
zWkxFFi{O4mSi|x%mw}*|@NQgzi6r`OG1&Uj>gd%n?+Kn3G&eQ-jM{JjODY0iWxk8!
zyA`MQilMhwSm@(Yx{+|V;a#KxU<t9e%n=^~90JL#ZO!QqLc{IPSBrD$yijo^Po<dv
zrm#HCC3Ku2yr40L4=~yn728f#fFUm@jqboH$5L{vAv*1k;9w@O7FXvC0^}vTXEAYN
zZ5PYM;2!{gRr$Ww#xl!&5;VL?jAj-u{I68#1np_M3S$6s#v8kcuwzX1lCl_!|F=L_
zR}?ol<lwX5>WI&e5iDtFo0&rZ>;Y?X<}Lvn%`GyoV~#ifAgl9Cdx@TjbfK_&H_sO<
zvx({JbLy3ag!k$&#nccSANxs{09hee@aqs0!rV^oFCXNhl7z@rKV@;O>2gMnlzLAk
z_hb;kXU#e4dqE6%DyWL2d&s*Y4a%<-1!0}nz9Z~>ULwK+0>L(o62G*C9aXEUAAcC&
zYHp2)6zsyoEcT22=h$kh+UhzTj|a(<YDy;H@q%!5m|smj!9Q1Z={9Y*SC(K>Lwjd6
zPx*6NW4zQj!#aLt{dXn0F-E}&A$ZswmG)k~D2({Ldch)ES!7*fRaD5MNX?u;c}Bc?
zPdWR@tIGE-K60q4GL3EKw}oLt$RIC1zj|gqDGr5_r#l!#k+f=sIph3k&lZeoh^GF*
z3?(lhh<i9b!py6z<7aC2klbFj9H$CDcaar)?1@pubKJqz<Zd)=C{>kj&9Ph#jQcC{
zT=RJGJQr|3KDsYHN++YdKqn&+@;7<2g5e*)0!(fG&8L5P;zLpM@o4v5y7-@L=X-b{
zwH-+u28-V@fcb6=`4`!AsZx)Z31-wByS*fWT^hBkquN4a?PjewAO{zrQEov|KxwS$
z-j*X}PIXF|fpJ9mXzeS>C5LeuK9DaNJ}&VpBnQU<J*Ttk;eI+b==VAyslexcFJ*OI
zsw-PuH3a-!D(2dPl(9Ax>xNpMZ#e@<vb`Za4t6HO7u7j*Pf#k>$P|Sv@Mb4i7NFop
z;YJLC0pQ?i*AM~XfQ%1p_Ed{BdfUCr%el%@G}};ogFeT+ePP^Vo_uP5x-@lAA;**A
zs@pjbUV-XU>>@kkgX34INaSwbt<9V3@2{Q*kHJ0rxBmfX`9B;`=fnP|f6A0QhP2?v
zuFWp!g8hqqg41OfIIc(JKb%&eS*mAq7`2!GUuy68xqpdpe7RuyeoiONi3v2*#e<(_
zE1jQ+-KpY&o!Jq`5fgNk2LZK7P{be~(w?QU38Pl`lL^quk_&lWsgSHLG9KuTzJbJo
zb}7E;(3%UfOd4vQgANy6KX_vE4Cq<iS{3`Z9X1J#mVZT&VBF<<k_(h3p`;vU@RDyI
z)S_X|^4F6u$pwbNA9if?+=KQH;+e<6T)6nXJRp_#Sk1WB`bp~+gN7d2tGIMA=ypj!
z92m0ne)5yQXqg1Jv$Mx~3cafGB_Ylsypbxi#}YtTQAAFwi+}m~PxP24ErK&mMf%Vk
zABm&PeQVx`)tRPFZn}xu5`nbQrH{uMJsyN0SN#@%iTIQwAxdX4X{2(%4}`{`J1X{n
zdPqa#7J8{rU^PbBns@H`L&+T|VL|~pwh{Ytg3@Te`ugUr&)_8ctW{#1)x$y!5Pi0j
zpg{Qf!`Ie!MW?B_%ERJH1+zgjs=oM^S=#)AvVWaT_$$z+B&wW%ocW7VLeps0#@gyK
zbp-TVY!X6P_b(ZQ8+S@c-q=1~0!iN+t7U;X0j}?#;Qt^NGl!We2e`T3oqep}!J@Zp
z3^y`=JVVfHhMQVS=7`;?%!vy@C*|087z*qX4DtU5kTU<e^oYfbfJkr!D$kffm%Q0u
zcnjh$che9b`_Q*yJZhX^enGPpz&mT}+Mv6~lP(1hqTm!~f#U1!ovexJR;Ik`IqZHX
zQUPpmzUuDxm#P+&g};GoN+rsZGoi4e+0Y*howwc0z%8I92;U)XF(jXc<2)9XUf&0t
zl>KPm?qj-K4mAb+imb3GXX3N}BO`9(b{#Hfc$as^ZhsGrsHtihI2q`~TIgT_R;O0@
zC2SZso8)QP&m4chjK~Xwg;ipEORs(~qc|9oa>Dpl3#T*&?~K(WbV6GZ!}1~8v`Lq+
zj(I^qh?_ThMR*@Y<^Vo)a&ZIm)MHfx`)^wDn%0xee9o8tvC!7pwO>eJwk&E*K=C3F
zwr2zyi$VdP`1_ADt?AoG(j&@6TU4=1%c>YC28!-c@6yoyv&*+Dr)L^!RG~kLT`dd<
zb|qx^s2Q&3t}{Li*Aq@{6rXoq55VCweCsg>pkx;+=~i*^&whTzRU$X^o&wGiX)Rk-
zRF^N*U|e|+;^4cX+5mWWDK&@}zb7(nP1k8tw+}fjs*N2pnIK{0O-@eMo>EhjKGAo8
z5aGB&002Iq=(jKF*QKoQZ0r~yJ$^!0(tVU|&+F~CNl}ecb;9BX_#cODa}4{6^jIch
zc|_8CgK2*pHh_Bwg2#lJ$R+qpj+U$fV{xoyzxsJcjI*2{sWUZ)SK*+O!lx{A$FQaT
z)UC&&<d@a{v?mcJ0GCYRB#X_J&bJ&RD79PV@WC!kfYTn&(3!dMT2UF)_~LX5<4Z?;
z2_<6KDzH1bt!H?D@2-s~*TYlH2)NR~>M@IzoEtOqPIGZ>6H2~Mp2|us*6qJ*oe;tV
z5Fq49*O;URGFnZ!*n32_S*BCWM=qRYWU20h6x@&}H}&>5b<N7gN}aU9o#1hl6WYI=
z(F+J@9tXMC$!X)W;W;};mw#u-+BX(InL`^USriSf#uvIsiwx;kSeQ%Gzi)(~oFxVQ
zxzzEtu@DTD^)pC=Y{KuEu-MwwQl&_l++#F!PjN*wFljH|RYk>Tqm31OgZ%gh&_;8@
zjnZ1B+;AY>ZLbBdDn1YaLA#DM-9H0HrG0@$c&jvwtI$n;oH*{;_ZKRgNUsGB8FkKo
zX-yL`bHdbm3znr)4w_x2#n;fEvEJvfHcqmN&Pe0=EkGkFBX4?ci4Kp10El(Djeirf
zj<LYPSSh~lyxxPuRYK~4RSu#m6&>wV!&vEFL?G6wl6jUBIGIh0E|F!MF2TsBS7C5D
z%sJpjyJhC}q*iZH&$?2w@2Hi}mPjLkk3D%zb}V`j>D;h$^+GBvQb1kr_^NTh(->in
zQ$!K2n0ttPlRUOMHlBRkT3u&*h8R7T4=qJzmec<gJIYIJ)iMu*Drcl_EVTKOt)Ove
zkX>S3({kA9>&rES9SQ$*SsAi?75tJ!nbr4RrEFvNlvL1)8krgufNJB5Es^f!&ur&$
zqu5}y+T2zF(ok5Y2M`IuSfzpKI7{P0Lwq~}``ofsBKQJF*(@9Ep;Q#U?GHXkgR}Ec
zw20WGWqDPK!$Pbe9-_IZdIf8b_n*Bh*Nd51ln!N3CQ#;3r0<nFr+q@zD;t&cXN9WT
z$q5vc2}tuisjH3SI?a0aggsd~0BJsfrDT!~0AGwpzdq#sx}R<u@f+&AO{!r3$#@7_
zxyy4OJWW0H>DaiLEZwUzrSr9u@aT8Bz*ZK?i4DtY(x(Zgb%X)^VruHL9wD>=-&nPc
zH><Z|;O`x#k`aT-+8R}`Bx10^H}lYt{xh`9HAn3lW0I`yF6Q;`w!_3W<1}9BH?4-(
zgkd%b&M>q|-UT)jijgT!+yQor!qW!%y7n@IoVuNw+W7M>A>xRmTdd@geLhXqXo2~O
zNODPfb-LG=`@lGXZL8~2qsy+)O$=##i3n|OTQjDmvoJYXshFM@%AHNWi`G2jzW1fZ
z^P4LcUS29EAalA8U<7<KQfzl2<j$<UXF55q_=Sy^Y)0j*2*oa9pnvOvlKxjG_4X~0
zQSRK{<=VBK{*wC=a+?Z=w^8s~_>{ELSEpDV5hlXr)J#EgyV}%ZOs7?j=jhKd)^v7X
zt$T$NPt{}l38JM4Lq=Wk1W~?{@=+hkx-pNB&-uBY-1sR!E0gCLdPm<iu2;QfSWIMC
z=41-$C-xS&666A4;VSERjA^4>Ts*^9hmIP91wP8N)FnRZkHtLN;Ut8Bg1kMPYbsqz
z)*jE;WBk&_@p`#aoLv^osTRVIV=zes+=ctm;?G1gt{cLgNHso|{PdPuhW<3>l*ZAi
z`hC|KXq_Z3&gCpqtOiQ<YssvEWQM`uxh5-H!J>nP4I*DF9;l$(Yr6CyZ6xUg++4qj
zc-}7!ELH*L2PoH{@GBiNNE{bcyC$xq=-<rsf9PKs6p#P&jziO>x=IbxHM7S^mFrDL
ze32a&nAxcDKK$j65|@aMFu7z_vOYM0qpfehiV#ndGMgCfCDZ?bby`3HqQ9i8zMGOr
zC73hd9gL)asQK3l6LJlGp}(g;+Gi~Zf&DF!F$4>`AB&NjJd^GTin#Q2>6*EpIZt~`
z`4^UAo+rz($m<K=$+0Z!=&b4GTafDs36MBS!U5pJ%pgv=h)n}tnaPclW8TZqA+BWa
z?2oflHA3B3%QW)k<S<?Y3{2l_F4FJv$@`O7G%KN&F5R4(nwlw=Tk1YOYD{MV4{obM
zJ($1aO8JzD^)mYXea+_16Z^iHK0T7P(lV1V#sTF%0Ovu*YmeXl(NM)MoOjqjxL$xq
ze_Pexa$rinX~19E^jy?(i2fjTw73Izw!|#&bN@WC82GGO#mBd7m7<hzJ8$bp7$%u4
zlbsm_znHTlc_^wAQnj?>bt`u!l+)RnKy9Dm^dlK;dhFja5I$`Hb7kW(K{G0zhTabg
z7m${Wx4<IdBMaMIW`3&{)N+E}8R0aOja5|mCqEzU*=}$f6W5}f3H|cRCB^jG^a%|A
zc<}G#lZBR_mNCKUp6bRT5^?@UpQ5`y7M|sVBmFrDK|=k7fftKpkw*6iH~aT+%`SIq
zM+#hJ*#rS7sw=TDz=oIog?OI?H@$u1Mce>$L~Z4=HbT?_rW|GH7{N2iWm|5Tf47Az
zacT+=aYOa90ttLo7d&dPL;~A?ujQjr$pf0eRhxyChJChNa|GRxF#cToZk7|~ew!9O
zqyu5`d#4JYlVebZq7)Qph`cYh`Q0IVSGk_lZ`D4q_ei38Vqj}z#}EM0KyE%u35aqO
zdIEaYoIaAp*)d~1s=&4hq@+Ym_tSW4N0cqjTTZxOnCyQKml9$W^ou@$x=4j4&riAF
zdwSs+jm#LekRZBP0^<bFsP3e}gi%0QglS%^I{(mN>TIJ>m}ES4I2J5I5paWhNAiO*
zVI*y0!-*N%uzaYJc1uw>5T%d3(5Zb#I}Hm$rIf=E*Sj;Gixr0jq59nATy@`-zMQ4R
zO?-`)hh|O4LxWJ?I1+*Woc4^I1Y&M-r7#n`$ky}_Y)_@EGUOI^gm)J0N#57{rkLTr
zvdmCM0?ZMQ2$Y7n*`usuLhq@Vhe9h%D--PFiK}Yk2tjqAPTeV!LkjE*zah#tN{F#!
z5b3MA)9vo7`3~9F62_emYb}xSg%y?Ry9VKj_^~~u)Mkfm8zDl67;x1=ol(QKWo+Lw
z+FrG$I@`Xm#ftZ2se$^G=xTqgBBU1M;$lIBPzgUoIsDDciSqxCueF@rx_ao?GLKoM
zFawlw#8(Pgxff56;z14})bu!u4%n6JR@vl-F<2KEzQ5E#BHVEiLsy<KINC!SnPFJ1
z2W2gji7;l^m(x{t(G(`zx=WNF=qQI&A^y`aW=M+YD-4@<9#n?!wC^YW+Iq2j081_S
z|M3<*c7fbUa>7fq_yDDxG0*ZpfW7IjjN&ha8MlsA{+m%B%h^VM%OI|u4$BMz-ojr8
zsNM8{)G&?2|KWgWHQ<p7i<X5Ymyd>XzhF*Ci~ld}pGxdMfViB00J2|mjwy#fP`z$l
zaDF<76Uf)sJ>M}Efwo$=qb{icWyeLA6skYP1Qw;n2)s~8hS9idPtFN-Y-Y_BRQW1`
z6laW1H#Rku__bX$8#LFd1r=BLQPB{3*?mz!Jq~yHXd%9~mLC!-J@l$P+D784Anznu
z%Hc#LIZ<69&#{RcaB~BD#6=vI`Zpdsz;;JuBy{pVHJXkncnS&20y?mPlAc^0SyQAG
z>xc_^MhJ0#yS<%&rmbD%HlwahoF+|V_mu^IK+Ln9MgwOvS!T+2tQvLJPq7$YigYeB
zs4YJ_zEix&6~9?lr66%GIL0{_cwKan_qfOT-^u=_4l{)(_LyB|^AV%@oTl}<>NyaT
z0SDQ|3<sqJPZVS2iqoBC+eQvdHB{{{WSMP}xtJ3lH)udx^+EhNwRN5}+gY@9%M>r;
z<tBKEMc85mKr7;tWc14%Bv(noA3`Y~;uasfRPh1WBS0PxN3vB@-Giwlov|m6FRF|^
z!K)Nm-+I@++K6_2{yOSy8f&G+e$KJd{tlj6)8~hGzKO&+J(e0a7iIm@wOKz%2JD^X
zbk<m)Vmb2>Y;b~+gUivK&?355)+9|mBgxQz6Y|k`iv(B498;QnOAKeP{*&QqzL3j=
zDr_s%1q`|m|0&iMk4{cG@WbRPtiK!F{)<K%Sd%2qNwC&KA(!jaT{*h9gh^?yMPnPW
zrv`yG>DCW@mT4lzK|nEdM~fkUygI@c@m^iFIlHehB-G4x(dr%;bMa^S)|OYmieGx?
zIfeoWo-O%RmXlB#)-YvJacLj&bT95=GEGpz{)KP_M$QBOj~z<W+oCD}yv$|ks@rWp
zu0B=fYJYPzRO<cTtNAhqb~(S|%AcM5>HM1H_9+&VujWXIK3X5D-jdEQD_i_2zh;4&
zT%)ed6*t`*b4%YmuDgBzm$J;=<f<83Cl4htD5Uho)-07grgi0RdW^2Oc*>&AjmKtK
z)g~mBu*wv#&{+J9p_u=srR#;G0&<6>BpVkU%qX1IUTWHNRBXN9!aEFed{SK=E(Q*_
z=QX}Ge!8=(FxWa@m~&m{s=zkIqGefsW8Mgg&Uo~tp<761H>(_%LaP6#ww+bJJj_z3
z@;*GiTXVVXChG%@-OIl@6|c)n6c8}ny6Dq42HxzT^kW;ddfa-Zl=3>VaemG;mhj(~
z_gnD&&M<rC51Tbs2(E3?on|rDEn{Cm?-}L$cNnkjoxFT+VAa|sWtp{h8*i_gs~H%b
z*iu@2_?-OR6}_uObM8diXt@ciEG)B=bf`If`_0US$$rQCazz)UT%Bcd<@L(k4@;N?
z80r@KFkY@ZnX~($o^ExLx9X&to40%~XH8lwJ6F(CiLEQ>Q@`Kjpx5lqLJ!~mt&@&e
zwT=IxVg0e<H9OC~`z-NHwy(h@$#PcKjLvz6*=5;Q=Z=&t69wKgko4@pyW&05wl*Ex
z7JGZEc!h(xmd!Dy35#r7N}pa|b0l$E{)`X5#SR4>kB+$&TX8;m>qfKL3+{*ri@ahl
z<Tv0fKK?_z(`fBF<!gG^14Skqot;`zY;LrRFHz+C>H9@TrxiSwJrkO=w>w>9M?#=-
z>CH7Md`|t(8m6CejX9^eD&N@9xc=y(i?_c%m^U|2N;&J*>V(I+R%RT@D^><Z?Y=sB
zNuDO7m*OeT6?RF=H*D43^8_4U^D(+l;i=2I+Jvu%C$C$4F1U$vSt654fVop!gB5#i
z%kGN>pXCo;*|7Qet7|`=-u{uMIo+@|FK&Uf!FR!TMjD$QUS4zJ)Yi;HmmlZr@NW;}
zn>Xc!Iiq~d%VmMD3VB~Ixzw{Ib?>V!u3Imeo%wdxYsIQ*7Ax)7JmfPBjthI!uD`?b
zy6}u3|3ik`LXJ4E^ip7uwOM-gIp3%7^B<2~Uw!Ds#ML#kPHM)eYduq}p0Jz0IzjHH
zg8e~_+`_UQSG$=kfhWjjE(sK57M@@NyaqQwKEqe0zIDev^Q_+0N%lownK}bQA0HL}
z(BSOE%FNfiWXb0(XMB!Lt^{Td+1+MqB~lCJ7AeUk+g!=3-1#i_()YlX&mI><zFQo8
z_Sl2XciooH?md{ktJmLLy6JqW)Vw@%)_Jc?ZSLf8+g{14+;K;&dc~E`g*$Y})3y_0
za`!Z3QwodFE&9T~Z27qdY<iKs>t7$OTKJ#g&3aI4q*MA@-;cGSf4x9wT6@fo{?8!!
zpCPIrxc6A*JZSIny#vs5sgU;y!_KP0CMNj!Pq<F*q`UU3RE6H=gn1@4Z@sG2BD5q3
zcyi4m`Pn-^f7Yx2C8zr{>OVt~$3{z|A`7EX({SHs^|~w1PnxmC%j)NJ`@j5Rf7Twq
z(?4_be})Sm^H1HWJ*jkLd0WxF=kr%ppYvVu`dQ9@hA%eJpSH^v*`=QU&)|G?zjonw
zm2Sa@bGGk%{;GV=?7-`1fc5!{itSJH?4Hz|o&T@ByDseUJ5MoFImy%2^XisYp4=*1
za~dY@j-+~Gl8y1S?>2w^K2OfF`Z*innqq`&UMU~8%r8EF<-MiZ;`KA(5+{Dn?vP*j
z=>F6lzbCo&x`SQ3@?KG{d%ZERB?FRJ|6RS=cRsK=uo&*-JuoMqKeH3zWVq8jr2|)d
zI2V2A^H*!2*PlNFPOV_^dhd!W73sR~Du2y=u6CvLvl%#yqx*Nt8y7Nvv55zTl*g`K
zj}Pa<p$5HvmZwnH-u(|4C@7vUm70tkh<6;|k+Gxkmz@4jpqVPMZi|t<50`*g296e>
zWk}wS2fG3iUulonUsUV|Ma#reseYcPd(Ybk?|E{oZy#6!8ed2FFKmRw`zup9X+(sb
zg-blR4D^^zIM`#ifOdTXCJt~=+k@-^N*o8K4Umx$>!0=k?b?cLR~<s425e;fX_?xS
z%2`N>L*@@$0url9`=?p_p5%5V3z0f3!O;ztnE2UTz~1R-y$&!LqQpfpLP8l7<gH!6
zgPGr<M=D&x{R}X~xkdkMJbq`A+2TM%RGddh^v?u_xS00OWMEX}B1gr1gakM$Afey^
z3La4284VtIxDC_b@rNXAa3%ufE?!vfT9{n1&1YW9y-#nZoLF?X=`4qvx5h%??La=a
z<&qj07}Wn5{g&#y-Y;wMvP+}e#c0*aRrx9|-mB0KmKfz?nGj?d6>}L413G36u1EKb
zb~*~A{k(XJ)o`Bh92=7^)r(w=PjbJ`uXq@0+uF6H;$@kqjPw$2&GTo9{uZ;}zWipU
z(WUECAMDXKaXc}5Mkv$oLw1`}cc(<Da4O6c_F-`W?mo?NPyS<IjeT$f+knTYn359v
G|C<0??DZZ1

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/6YGqFRTe2D0-295.jpg b/assets/images/impact/talks/6YGqFRTe2D0-295.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..3df407e262ceb43c2bfa70d9a4dd7ccbf1fad8f7
GIT binary patch
literal 59480
zcmc$_1yCH(x-L3ckOT-42s#kl1Hs*4u)$qJfZ*=#t{EH#clY2H++72~9fHGPm%Y!~
z=kEKeZq<8L@7~i>wPsZ>`MX#5Z(o1wW%1<?;Ju6_ND=@C2LQmm{s1p405QNDL_{P+
z#5YJtNN?V}K}N+uMMXhD#YRU*!}x%WhxY*+7Z;z9oEZNjkN_8#goXt8iGqrX3XhnM
zk(QEyoRW(2ACthn4n;*l#X?2JqWp;ak@ElZ@zMprLV^E?(1if^2>_1;hkyn5@(V!n
zdXez(2>-a=e;vqgkPs12;NehToAuuV;1J>A5a8hv5#PT>dIN_J2af<iM8bN5O^(cp
zBck*P*XV;i8wwspynoG<D89ObBP1tbx{gz|_Qx0HuTC@FD(oCwiE5MqRAN23#?%|)
zK_3I_C7fM;ZT>xfouBYk@T>HHiT=kw96TZd(wo;dF09uTM1Vs=M0o>;^dDW|;IR-8
z$yu@Om5^%KM7nWszr6XRJY~f0AD{C<RORbm3J0;5Wx%`F312&60fYc2a^k|`;)ws9
z1zg+!@1MJ<DymM4KF;~0N+de*%&c6*)NRXQ=)A+?J*`i53NNgXHaw=hf>L~rOEQVy
zERcRNH)j2apt+}wuCJNefV#W_SwlZUOYFP#W>AR*nsQU+Qz-ag^k~9WV5g{0g!wC5
zGEOdVh&#@we;VS`BG*5yp{6Fka`{VwnQ#sDyj|sryU*kWKzJ1S<SZ9$fMchf9m3s6
zhH&|f;rf<|@Cm=n+YgH3MA}E@TgveQAed@9H9FQPbL$PCQ%LJoRd|*wx}lb+Bo8w*
zZ|Saogck<*K!(T}$dLTNqotlqnQ)WqhY2QbpA_g0r-&W#vee%^&ndXn{5<@Fs9|W?
zIEXkF6Tk;p1a`*+Tk3&;{;VH2xkS15$_^Ro=V3&mTu0M!<i`7Pd@fl4&H07S+1ZDV
zr=9YO*}1i*+7CwK0|wD_f+{AM4gDvbO=wn&<)E{>!M6x6fb1bu*mv1eOe2~DVC)LL
zzG{A6PcVcXlE#BJ&xx5U6KU~mGWv1OW4MTTDV0N|c6K=EPmW|uR`vT8X(Hu@eHtXb
zE4C>7Lvo2K;evehXN+CP;DTB6?=l@doHT?m#Y}QK>3g`6nAH<RafPj<|K?W-F`Vbe
z^Ez4Q`jnyXRXEe&mq87IN~L(o7r>q0>xsas$SL9_Z7^;<v~?OSrk-zPF9uR>7W=o)
zV9%Pf)nfexU~ggIJFAF|Kfa@3Kw$n0x>0cS_eku1`!cpb0~pXZTfP5Xqj-jl&PXr}
zFy<lWQfNx#{6S)qfL9AilQ|I=uokq(l`egnoy&)vz|lFARzIi4SiGQPA^A|uV3<`u
zE`jH*TJHz0QZKYp;WV04<(&G)XkDL(3GISD&Fy=?&*~{D4^c#(0vq3)dcGrrHU6kd
zNsdrWONZ<8Sy@52Q#y0HwU?(75-j<{85?OXR3dogzYlqabfUF^B}XWHbMu;31ieiX
zIEg{)!(J#P&tXds)FDzr;k`e5wcbT*msNXc%T^*LX4R?}f3oS`Ic920-+!w}qBE)5
zeKA?KN|C@?Tj0I6Rf2fd!d~;tIJT!PGF;((m0w^GTEA7IRA%9jiby@9pdsXMPWJ+c
z+M8Osp7}9<-0&MpPlMie97?Z=`8iYsk%?S5w%*3uxqIg5bBd*C2U{I;ad3UCxz6>n
zPwXR&K;#gD;px{=>lDlI#gjI1g_V^5?pNn%AYtub*PZ<_6Sx_^=sn(e&*A&bGlKae
z?@@eI-YkTtmVJ#*?fA2UE3I*;Z>6qLfS{2Ajqh0V=Zjv`dFSoMl@aT9Dl*DAKHK_G
zbCW~Edr}}>oJ}v}KpT(b25UAOJJGUn;8!>q!$r=Y3YVL@Z?U0sucl;-E+VXqq_RV1
zH|^|O{J?hsFvg56XWVAEc0yIilaTti6ho5hMwPs5z?}~yJpNwAPFYmKYDtJ()%MYR
z)R*4p-QrRK^jRAD(NtXf)2MrRYKxw%0XpBE2sBXG%8B3k2KEBr55>RN{_EsJEyP~c
zGcEOKhV@u^XZNTpY4`#-F1j(vZDnAW`F;aO_XOt~RHgmwcH_tQjwI_kpRsEAEc|%}
zIo06mRX}3w@GVBxC9wfn=SCY2tZSrS!i>1+Tf+-r@zvH?hyWoLN5*U)o1z<)%SO{;
z)Q(Q$60H~vn5wReh=7T`oVebVKqKw>TPv>@K;W9n;|~w0_3-Ti$o(zjszgCvfegs@
zTSWsfr2F~>pl&{<t37tdI(B-D15_VU&rZX`F<U8J782v5DIl}mHT^B;U3!!1dU$63
z7V+lJMkK~D(eAMknJk2;co7X;?GGcA44pa_!1N96w-UP7TagI0e70c6U5)^&1$zW?
z%p^6(;?~kbKW>DD;4n!Y5=Zih=ToW=Er|*XH830kM@wRAQNZT$uAFB~!|U4^YNkbN
z3}Ssr5LIkZc+k@Q)v0C2fa67q+QbBdO7rT+u*^{YurQP`zyGcx5tGTG?2%@mIYuPv
z9g)-Ya{BG0dTjn7whJ85ao%~-z4iqkN?p%|HFu$MA{7yH+<PH$a0%Pp!DcC47Q5Vv
zWx32_InN;qQaPq?ro;=NEu(n5$vxtY=cgoz*1U?YVKhaQlSORjMa-ztUBMjP91e|W
z*K$B2d}0M9I#6Nxq~^{9Il<_s{K`COJJcpbNiIpR>N)`uiDu{2nzKYFD}3yws&e^-
zroQz{;0F!~eQ*5kk{G(haUE++j9^MyprZHysX&a|6L0a-y_HOiMwH!fQpJ4HXt~Z!
zs4^)Yp$wW#J*MMTgz%CyA0Oe>7o%e!YX4ph`|?Cog~Vdbb#g}C$q8IgQa7C@0G}(b
zTH(sMq(7QNsijb=GW<ek|L>zVP%I}gzsM(2h%VkS?5ov{4l{%AQ!njS*k8|PwWGAk
zV2+MI(h@<<({ZTi(V1t}Fbe@155yx!&?*>-6HjH>LB4**2$B5CdI7;(!hG!t6^Zw(
z0H`OO$>I1ceO-SgU3^|d>ezC?_>8ai@MZnfq<^3Q;aSa~j>7d~v6GalJd||1YPWnz
zrJQ#`TiCj~y+G&GHJfWi$~$U{)M*S;9V+y^Bs80L*bSvIk(HG5vHO7;^Yg*Dkw>Vs
z0eCl#&y9!Tii`6^<ue+KoN<x=L~UiM@4x*5P{!GPScW-QFg`wA#Xf^bi<B1bxzF=)
z?Ywv~ro<^9aTANN?l5*3G8~R4aEBw@G7{5<oD-=w!Aja7GI)jC^CuqL^F0>hj5-tg
znz$01CXkNNcXE(`KXTq4#S4IKv!i5W`E6~LHEhW7w}AtIl4Kv_DK<X?N25HAO=fN-
zGy)PCIA(m0jGJRFb8Cx-9cv<wQ<Q#EUpDj2&u47Ze^WE~VBq|i`yQd9ugF4a!9gyF
z`<B}*=5x|V)0IKljzgIwY~_mY@s+*Ce@E+}W}|eLv_8{{;7JfVH2yEq!P%Y3#h)QT
zL`e&#2G%CgcFA73`lQJ>0m|%%^X@TE?Ud5DsScHFih(8=@!4nE3b7MKMFu4t*7rK(
z9yLbi<{&cX_qPf!0BLl2to)JPrpg=`3l>g!_oq+rpT6EBlU4YYh7yY<F?MuXG^k7d
zuHLwkk-ALqgsb&L3H!NL<ItezZFG}*cy)`1)7jnJciYePC|Gp=?Rm6na)Z$Xw>+7t
zOqL<R%Qwxr`UrRGuXT0~%;H+<9v!avA?qBt)`5Bd*0+YKhDZzC!P>cTEi^)NV|*56
zd*=<l7<x*Le44n6Ju^Ul4DMWvU3~$BJzxLz9n^UWTeLP2yz74fgwDPI7_*jQ|7)-*
z%dOC+P`elK0`CRzRmky~;$flt1wd|lLt)eadDtR)2C6*P`kMax=#<WXpZecKMu*HR
znw%e!y`2LI`tfHiWsI#<9tP-m7Zg8YSId=}{X9e5`nj{#+HcoR7R0$s%UMKf_e}pj
z(^csX^G?Q=xLc!Sn;%bi>$`D<7HmeefjS}NsGe|9avE8?>9>qvQS~w3sJIMsO7176
zub;^NvE=_33T3b9+SreKD7;VTHJv?&h^;bG>A3lKz1%8Clao(e@mvS(*Th@x+K+69
z7y9)KjS#1#wDlP}M9TPsc8^27>NZ!@)V}$_m7cBDT9`*tLoIFr$x-toykchUbwY@c
zIFlc{ALcag{NVkNN;H+c5k2R@@^mz?t~qle>;=FA780*F_fp=FLf;xZ&LCy?h4Gq>
z=t{w7$x(B^)_uR45WzbZkWCOz7)z1%Gz!Li@_PZ?pgx6bH22=++S1wDXUGiCs^(h8
zY(sYqaMynQ+O8S)Mhto)Uz4+1wf*{|i(B?2AciFr81OR-ImYe#^%nP!7r_3e!(@HS
zsV6f$;y6_SBRsak=H+ZX!+@XNZ0u<x@SYor3D{U~_Y%6WXS6Kec&aRlEp2^{s+TMK
z{LA>R9~tfNcfZm8w7qeGtWkD?RAo_A5kdEErdTt-JVPb90G-MS5^ikV1NFCgJ1?3*
ztudHX=9K$tWzoQ@nD6g$jeEGa=eJl3HebS@6{r$hVrXFsCdSYM8fDWG(OAnMdpwoS
zu{K4aU#PhQfUPFt<<B2yi;lm^3W}B$4v63|h?agdM0brG%-Z~9LN6`{rrfT-o1+x?
zmJIYZN}G!lb{Hv|LS47*0UN^6REI^z>ITMXbg7EPkdNRQ7&K^?>8<{o;-cma+&zVy
zFz}@7xO`#B%iDLgNjK}Poqr|?Rc=?6oUn$g7&u`2-C*{`^+0p9BgXIjp{&!gz*ABy
z%s0PE4Ol0v*>RQkOr-Nz>T9a)Vil~z%JDx>2>w@-n(3&#Z9DA?AY}dpK>w<w|5o>}
z!_hgD5!z_z94>lQ_<z;D_L>^kEgkB*61OfSWygsUStFDTB>8?o0iQs85UxD|a+{+L
z8R~=)C!OXP08zymw72$i&>>N5OlO?GFMyyte^;h>f{Y32dTbT_TD95Vlt9&@C#tL!
zmaTnuWKVHt@sU5UQ-e=mcw*we!>3&*R2CkP!3t<@Mkvmc&#y?@-Ai8p-rNg+PhJ4V
zJVof>vnrxzlFs9Azg8Q6{lLNw|Mck%X0oF$KGHy4+{ABKrdgK6H<lIUo6V-2`~V;1
zwm+@gf?r~H>hB`&*`IFhYcJIlgBgPW3bVq~o|MpS59xu6R%T4nuaV?b-RV`+cuol(
z8Wm6JDd3?#wX+Kwdq)VekQ3#iHZ#><<R$}cxSYQV{V^SIKop!%%-uz1_b5l2fzv04
zJaA?0!bu6CJ@4YQpAZ(zA>H%h$GHpdNrbxlXR+&56$MH)H^2N$tZL}fugYiX&RgsF
z*VwV~_yrKLn&ZZId7?$C&OMGCy=`9<$IB#;poP)sx=LYy0eyX1h$}<<ms{gh5bm}C
zGI>v<!ZoId(<+d*Xo3s2lvA9qy3WsaihhSxD%~~0!3NP~nI+f!qGEWFV_}M)Un-lk
z#%u1u7rtZf)bXjDzNQ>m(&i>5tjp^4cUre&cF7&v0`X^pa@i6~)jE&!7fo|kn3}8d
zv2JYbxHkXK8bB(R(HxrorMS+%154kXc;<xUUDXnO_A$gGZ*-4L-*Y9d1MmN|d`I5&
z4T@rC<@IY2?8YuCnvIqE)=kFxXWrKG?8fCESIUxMkDnyX&G+zd=3adVdz{-M1*%Hf
zLiTfWW@k8EBbwRym8@ifkGWa8ql!rr-x#iZVXXTW^TanoaupNC9Ph9+s~<I7scciq
znn#*%pw%P_q3%gv(}=9;04^s2v09^<{={P8YC1IM;+NOrFMvEw5g@@ZB!DVcaezk^
z(=QE^6FYC&j1+7$gJsRDzLTrwR_dn;hr>=F)o_<LZDjFd;Y%201B&WJRJI^UdTqCb
z7$D!aKIN3v`1NTDp*kLmZI((U!I*Z(Jq9i+n0|ragDowWQU1?3Utyvt<W5JJ!vez+
z_^bo@DdG00hDqT{HpLFg^03nD`^gqox3CMwe>&OljpumMO2ztgq5aWz^dGG~FMv;$
zu$ucHeJ=n{H-l|?Uv<;3EL1rmk3*X;fc7?)3+m1?U@E`CZBy{G_`<cR>Hl+V-QK@V
zofg)X6l6a3`C}BAa`V)<MwO09VHr)$W8W(N7^sZ6nDH#}{P)~otHRgN>|I)+HKGVN
zy|1_1`hTeJ9@4+m_iKW(oVZMWSl9<&a<?<A?Q#h(b(587R{d$F>^W`4+<IH#1WcP<
zbNzE>_`SGN$NANEFsojb(L@<9_FTRb<X{$(P&uF%b_FD7u7M}-_K{oFiTQT4-oJP6
zDA<?3@?j=U+QbqKZ90L72w51IayP;7m~z)+yWR{=aAdS^H*(3eNCGi&&F6w9(VyCz
zI;IW7uxu$Gp+EPiwqq{1zGM`JNX!#fZb4W=TBd^#qoDyHax&=CguD0Vqjx2Pe@V#K
zorelFkc>gwZ42xs*C(N#>!=DWOs#IW+H3$+I!ao%X?J<A>%4%A4WUF&09^rucf%!q
zxi<yuUZ~YvJ7SQJMBW_oeN7*&vo@e!)rME~(H3|nN-q9i60`6UF(xtUR{811v0xRq
zg=!*2CG$x*T8%+JDceTtE^!;9&-^X1<RfOR&N|+hMvR3O0h_`ykiwpBg;@InOg-VT
zUkdMn=WFV%^t`yWE2lFxt&uDaq?e|p96UWRc9Hqzb=Bpg`q?TaYxUo0o3n6|Sj(4F
zi5ZvB6M@e5y{SWScU0YdI97uZ1w+#6pCphI;9HjMa`c04mKCy1zZ%{rGw1buwjABo
zsortoAE9ETq|>qgl_L;N(wNhn@gEsxc+ie+OX&M5k%#^2dV`LBo+=Y^aqhzwWre|$
zP8H6&)WZyqJoQTI;dJMlqx{c3nHtJ-)126-2tGxV!h^r$`Nx$qs6XAP_ox@rX*i7d
zBUsRh%EWa1Mj?u4x8sR;3`LC{Rh12|@~Kl#!$7AcH%9}+LZjCF!i(p%1DCkYtyfH3
z%Nj#7D?eOK24iFA3Ig_z{>o#J4DVms2=+S?ubj+#+cZKt;t~T^BCL+#eQ0?f$V_wp
zW%`V)E{4^botc>%*STy|T#hVJNf*T|WTa_4*XB8v-&p)JLZrrzyjbRMX!`*ns`if=
z6pD6TlUDfs{Xl}3#(e46w4W$WWxNtbPdzOER<`iAt!D-4mwzoZoA<~1d8jc071rle
zeH}5(q0bpUJwVP&{TIO8Y*0YZgEX3k(z?wr%lS7puVO3|M-+$Yhrh3w4i#UN2@+{&
z)E_$6Gi>M7tem9VVPU~xz5m}7wE8Zcw+0bkZh5iy8biP`?HFrTzWoL!ETbus_rT~n
z?ji$+7DlT4WFkNbuQPJBKnM1$Mz^ZQjD>lwWlApB_<+39wwab=39cnPCszUKS{4t>
z-_|0Ue%4!o`li!Hu#7kI^d}XSKjj_@b7?<CUH~A-3!rNVIo619>(1k6qn^sx0$4^p
zf4|r&3p|(32d?)nB3AJ{v4xJ06yDdWd`|+wTeHQP!K-j;lkDN`Th|71EsMZgu>MW&
z3Mz8=x)n!@F*9<rRN5QMzo}zhpU5P`i;6Itnq#r4BUl@~!{6@TYu(rrhRM`8F9oC0
zdjV`P;a<yhj_{wk9?4V<e6%ja&P7UAnQxLBrM!`Tb|2{XI#2mJsbC#asjlwNM8;X~
zou?Q3hpn^UP3>K5{Z7K+L<y>T6SF~cBhKOWn^l>HB|ZIw=T?$0mYkj(j`*^RN7t9Q
z1nu%&C$7_Y!H>VrzvgJT2cCbBIx=M1T31eraK#OA{VM{e{p`D!EM*zq8uDNi{l&b#
zI!c93eOHO;ppFN-OYW*BKTd#Bo>O9WToju8t}2<`BsrnTldV`SkDPf}TzptD@wn)A
z9pQP%^LjCfKyTRB`>)V(a;ILng%ZR2Ff96n|4dL6A0wYOHrEHVpS#8h#<mU4F!7#w
z7jE9zjVSDk;{+7lE=0coFxGFrK<!kYVdtG2E6@FORFuEOOQru42#QbMAxs8Guu0I_
zt0g?C>7cAVR|2J)+r4<bea413-D>pGt$6f%rQMU26&vT{@+&C62VfRcY`{}rUatyt
zViDfmmbP6ll;ojB93uX_ZHbBfj+x+k@HfTtuJ2-k>ayHMBX#$jqej(y>1#W`e`UzL
zrhx5tkD&O2yH<f`Y>Do>Q2+DtTDHX)rl=D4_M|m2Po^K|(v(5Z**M6SJt+!e6Fm6j
zJ0DkVv>Sz0Vd*wh`ddhi!y0ufwK1(8B(DC1b31LF8FgEa67}0B)Hg_@(ILg#Av`Om
z1rt^uc%7tC;7sVIsC2Gy=;Dnt)Vl0OF)aCh^e;CJ>*g}5w<TM9OE+%F5Ia-I*r7+q
z|3#jc(|SheyO$I!6k-i3vS?uPRzo5Y(wIcp$K5ZgMyR<%y@-vyYh;7+Ch}8pGHfxX
zC%%VdvYjL1Bl{e}q?*^u(J)b)vNqcSGtWnr9r!cLB_bH|w4a4NOJt9)xdryP7O|?|
z&0V<enTllSyHZ3CFZ@y}F<EUS!b(HeF+AJXl$lf}L6uRvauM+1O_-A-<qT1HR<7Vo
zYmK{ku7^oYd*%qx9zG&er&NNspfsZ=thRB6huc6Sq<p5|fs+a_g|KDUb*Zn2=2}<o
ze3{K*o(X&Af6NVO*9V%^4@=vv_)$5D6{jJ}@RIFtf<Hd|a)-LypXaov)tLnEIk@fU
zFgryBGBs{$DPJ#r+jqT-@6=+xU?KQ>s!9KDon&AYmhLlxB!88(@&XXX-M@MPlpXL~
ztRh63edf4(7DJMFR8vmqGq5dq2f0BaTmYSV3)+$E8x1?0ZtxeLei5>6P@C8~QuPiU
z?kN9oTwp}Vi(sOIu+=^~uM^EDmSJB-$}@cb$sWH<cWBYu{R8#b3hKVJ?e9lJ|KHxI
z3nd8x1|R<}bC+?awK<aZ3(iFIUc0Ouh(+4!p^<q>Nk7rG4!ePgcjvS!25UYz$bar=
zY~OyOp7oYdWtV*Jg$M4?jWmzi=@JpdPng-Oonz{P%sA5;nOlIgFxwkuPq8NHvyQ#V
zA(lSpW+p+xIR*6#=-}tev4Xxu*;72xiKbH~0Z4TwywA{tZ(kb=O?Xat!vHOF65rQ6
zkciaW?a^d@oaD2>!{|72?B>r7rgC+>NcQh_9U<C;(v-(pJ)l*Lmgx%P0Q(6h2RUk4
zmRl$n%hxi&>%NvCDx)|9ZA=p;a`9l2OjH|=a<{V#drmvutS?YWn}%MOilgEBbX1#W
zdpu=S7?CDIlkr>*Wy9TVL~~{<R39y&RwGr-6AZEHSoTArCk-_OczU4r_4Awe-kCJn
z7;--mdnc<*7HNa%57uBJl>6R+qKIVaOWMz@-R>eWMyYi}wgy$F>Rd9V9lE|b&Jw<T
zaCDU4(PJ`)7+zz@^$F@1z~fHV3&7b%lkTo4aB||=PHpRBf344fo2IPZ^v1O&cr(~L
z+s4fPj?spG=K)eu-xn2`bEH7#>TX0%E_E^qk4UhMAu?D)oRwt9%cEm$u!E`Sig=i*
zeiRClhzUG|6fM(;&6K)rdF^qG9~lgD#I^M5x`uMk{o>qw1rFNr5`&$+X8pzoM#_8%
zxYjlo<z7Kthh5Irdwx635_5ByRWLe()JcZs^M?U@HMT_BP5OnZB>Mqd&Y*ow<^CkK
zJCV=i85Bzx&Ka8-o~%Q4nuknwhf6*!$+8&BwFQVJea(lJeq4DgW~Lf;_fhQ%2>qYA
z({GtIMlZb5@tD<HBC2#{i0#d#eW#FmxN25mm|FZhk9s|qv*+~}x(>mGtkxmsIp2k{
z8(%-l2X<t_E#72|-yMYu3?vQ^Z!KPisua8cN*=VWnXy$>;~!LNy*c1N?4+!zg`O8t
z6+z{bY7R!Wi2F5~<K>Uuk#($V<NY~RnT1n$*dH6$8ve5C$TZ^RFs{6}))q>xH7;ai
zLt9a3%>Z^Po29A@MOi5Pl|<*(8)|%x_xbwg6T4#V1tb3ynnuee3axTDWdaJ}(w#hi
zGQ2D64nLUrXQ^ot5$O>Jp$xdqinl7i0jLlc)8#cQhi<?o;E1p>F1*XT(^x(;;5P^1
z7R$@ecXq~jSSWe?f|a_>X>d`5GoUyYGI<MZHJU|kTvw9ySt(Pv@}b1J84^4pY^JwW
zZq;l3kR@V0pCPTRp1>{w#46G1xl0Z>UuZwp_&A{E)VM|UU4&kl7{EmNWA!~p9z3w$
z;oVuR-D*;Y>fQ@LK(kA;BQ$G>sGg;O9^puyGAlbjK``b=PR*JrH?>gO7TcZfoIOOJ
zz7IcRkc9#&vAzA___}dRKDJLo^Qkfvl;BHpUi5y`vU(REPA2!b;g~l&A-C`aFbNYs
z03ODyf7{x6X4ufxpv@@6kC`s8CkKgD1j0UQY$3usWj^@{`xEDvCu5$Uv9^!YF1@R?
zjGJ_rld8O4(2m@O<YX=5HB{x@oXT}r4C(cqTRF}$EtVc$y4^q;b!nAE%PHw7WBp0O
z?nVPi$^LNES`2^bW3^sYIjB3MsyuL_rKJ->gl1z}+io#xd69AnX|*)VSnHTHrB_3E
z_K8JX5~_Mo-XP+n^qV-wxErul>I~6U#Vk3GT3aNmt}s@7NZ1{g+JYA*p=oSdF$LZ0
z2HyL=T^(GS&T1OKdq}&YS5wY=+%+4(=?<wqep()AN~7mN+&>&r=Ec<5ke*$Bfcp17
zja{ohpTaul1gom@j<2q+d@!JsWTfPJr*s;>mUO{PBj5BV=+ZRIg&coZHllx3eZP<^
zkQ)d6JuuKV<4F!_-CP}j=<O1uttM6Mjc_PA6_(N43^Ya*c5cS0RQ|F`w2@B%;=LxS
zNe}Dj{r0oAPFKijwi0B~q*(zj0@Cw)170fK>)XH#X8WkGwZW73)!L`JF95lsbD56c
zHi)kMZeyPMK4bcIzRZ1)E!u4Xe{XXU?b(jP*~};E?)f-R%sI$7PvoF|6~XzlrC9tP
zQPX|zwbKlqba4{Cw{=C8<z%o`@sD*=7ksL2w>5*f>1~bbYlD%g$tr8%hiE?#N%vVU
zh-j^7s{8IwcTK>@h~AIAn<HxV9`E<<nrHc7*9ld<@SS^d(W~Hacal}-qEa2x7FjS`
zZDRIAI1!@sL+aN{e&w_Go0^ZEl%pJjz#0qK?~w;Qrj`{t@_hcPOKBPlT7{;4%8j%h
z<7PbOeg-C`GKf`C!7ee8On>*(lOlB$#U?kYPOPL5PefT<>2kp2%Fw!KXl{bHvd^TU
zxa<~Gn*Y;0nPy{RL%P`9erq~-!D_3AH(BEd^r+_Sig<fKNNsdMT-mSlq1Ww<*nkF>
zcpM|zemGy76WMWz;H%21s_<hrZ{H<54ld#RoC`%Ai~Gv}-P2k+^`<sV2NNrIPaW3M
zu;RH^*!#?}apShdZ@tDBt4<#L5UqA|m9)@)9l#gtDscM~%OPXjJ_g_Q!${V4n9Z(z
zT8ah4VbGPVdebM`WN~rY!~axTK!`xA3A*jpfw>)}qD4H;<CX#CXc6rfJ5s0fEVZo=
z{7X}81zq7?i|tpCJ~u;?0(!<`@JO9Dt>F<H=0lsLa)-kY4^NKTCmQ&5diJ-NzaH5S
zGby(;X4DQVd+4it!==-Eh?Bg9*-?y65V;bK{;Z14*zt<yE|1i3;4EkPl)uyhLoj3~
zIabb6yQ0}O?a&BeO}sF}?qVIjtWq&Ujl<A4-RHC-)3m*!;)lWz{iM&X*umvV&0w^x
zw&gN4cf3!PphAsPc2cm*+@(xN*pt{tD9*rV1tmv<&=Pq!*io5kZA_L)K`|W&7%Yr)
z(^cjsL>|}Bci~ja^HvM#tD}Z}`C@%Xllq8_If=|u4dBRuWydR_*WGxSd#+Mjae~^C
zbXRm&3qCJ&4aK2_{@SV$j*K+(W^K>N{v)F)QY(`&U4t54Igj;rqY%kXW`)@awZ{56
zCPG7!EAuV>0P|(lTL!o6v)^cFM+hEhi->WYKA$N@EAh!k-EZTUd?6R~A3o0=HY+r|
zjW9uRR-3J?tJ1@LX31Hiw)k4R1g89b{*2Ex=-{@T>^59GPn`2@-r*^~1Z7Iw^Xa{6
zJ1Qr~04t;3(3(vkU0tFKmU9>1A}UBmv=bhLg9SMcMIYT{y^9JZ&HEa|DuBDJ4Bkhb
zuTQ|}nmpxNLz231HfOZ2l$>V6_?8eO8z*DKU0t9B2O`~2JeFy95inthH*>sU<>~*p
zO&}A)FgQ5CVU#?f^NyC%iNZTTatQTmy$Ade<4ee9{c1sA|NMZu{q#j@WemhQ4vzQE
z19Lj9(oUcKj~xr9n#gL{2qSJ*ft4lL<C)6{9yn(5MW4kWzQ)aK+~Ct3Ba&AE$2Th?
zW_5Q9`-T#F4R7!F?W8MM;HcRxsP=hD#qh<_j7!17Jj-v`fJ4_40nuq(XWYq^r`Nan
zGooKYaZPcqK)n$SOX|#IJZP1B90)~GAPgpRax=<|+(%3ixo$=tvZV=AUsQ(7oMd0;
zl?YJ6D_rMFwv&mMJH89x#|*DM_2_Y!aKKf&{qXJK2!TvAeeK4_?^I~&f!daB=Mfna
z0{-#pzrzN1G)uzUqlU5cOK0P<N;@^N4m)zy^4cFcSZwZm{W=qU8TWAeN-s}C=g%#h
z?PwXwT_^bfudJP@B~rGH-gu}q_Y+Hd<hmwo2(;e%0{BwoE#P}(X-5ua^0$q(?_}Qj
zzH|BMly_ITRm2?`mz_{1FqrRj54P7fSQ#Qau&NT~j-;C|5dWr6sxJ4Wk9+uN(Tdyx
z<w{yaTlx`O)3g6kvR{WP$rV9fqc)1UVW35tlSmN`q`21V2Paq;2(>37I3xY$R`~Xi
z1HCjLQb-G=BG<4)Q+-w)ef5+r{Oiu=`cz7VXQX_!`kH?Yw7^>avu<*Zvs28ffactz
zMK!NVFPv-)Lnx@E|F6^MS*0oUXeW*DzNHX9uA7_mzyBLd1M59@)%IRTRriI$k+<N2
z_0cPmp87oz$XHb7XM%igu$-$x6Qae--Qi__H{eCizDtZke>jC8>^$1Q@Mg?(O%qUT
z<$pj=&MO%D_xqSv?5Ew25ho150s3tV?hW&p&{=(rEz<p0!iYhRVAl7HQ7^TzoFl<c
z1M(l~QjYc$9~n^xXF6r)C88al?M`gVp@cK?C+*3^n&b>?Nm5l{n->7qMKUv0;_!}r
z-(x?!u{AQ97}<~d@3>{1K7UTFGn`c#^#%X_tnRq9mWAPv=o!M*agun$As?XDMkulM
zu6sSSQ>!-bwLHUF<y6@-*b8`f;v&p-&fo~dIfaJb$r<14iVblvYZ`%E_aU^;><y|F
z@-hi_NhOm6h|X`6*A?wCA@8gF4oGWvQ~);)4R4J=%yjk1vXL4ivj7u?-_I>)#sYuR
zO0)xrrF>q45U=gTUuD|HZ<Tgh&wN67xhfCtJiJVq?AcG8Xj4*j+&KA8=WV{6B9T=f
zVulIu3jIw&PT*qovMM?}Elwb&X^~zy9E~h;%n7u9zjH`EH<lrjVJQ%GlgF*@Nx&hd
zD^oM1as`4E;TI6NY=85#fJn{3$kWovm+rz+=@VKZ&kuim7tefJvHSiV<(8o`!h)w#
z{!hvk8Zh3sjbG<(#bA-3m#6I?h)3stARgwwQKGm1lssK;`SvzFMeSvL5b9{Ny>od1
z)I@+ltpCJGl-jy?@Vw;v3C8)NT&d2qB~)uya_SnWq&&%9;rxrc0sM<K{P?>uuX|}d
z8M)7wjau3PNyF*u%3S9HAswFBsrMS0cIC*1zK<drHIHxO_Gk)HwjNvK@1$@?!t#b3
zoh#G~)Dy;>pE&#kwNcm4O%X?WI%(5MMYg>1$b&a+xZ--)Rw~qL?8!NKhbHOCoc1ZR
z8bNnHQm6J_F2;iN8_+tvCIj;kTBh0t8WVEQsfBUpfId#vYp*AnI&<rYeWRpF0c=RI
zpO7@_-Y2`Jdthjd?#5NaFCorl*Mi%(&U3~qiZF3Yy<WrHtA*^(e7fZ~ROmT#-}Vcd
zh?sLst^oO%Y6u;Fyq?G(Z$4E>RtiCK=QRA7I$BX=h=GU2nZVghjU8fHe+MCIxAhqN
zf=FWJfd$3<1#c4u^1h1iZv|i;A+25lnEkdppLMA_v&!u&U9T-NbZ`Q#%b6_aaeq5S
zlies$n6-8kH3JR(ml5I$>7JaH4b={gZlAhOt{!1c6$-pOUM>o?6-u4TznzZpHQgD6
z7pm->c?0{Z!?&M_w(lg}nXDu+FH_zTEty3+9|sl4+{naclTjHk=iHfA8YFrf_wrW(
z!?qxHc=S7m>SF>yZj}W|VVZ}M;Q(iHvi<#IuRpGMg`PNZ7-{Y@oML3j-sva?jmTB%
zv9@#Z_#ZBJ#o{h7K<39j)KyqrrM)&wuthFnrT`HcR5%gsJFfwEqV~W)=<cq5TWuCC
zH)Y=#xcg1gz$Np)%k$0k&m41=HL9H}FuD#c?`HZlR9LznODyKG((!7NDi`u=3$EEh
zCRKxCUI0jxq%Qzkl8g=Fb<ByoC#zk89Xv*ws%6)F-USEt`T!5`6kGgA+qx^!ScTmh
zO|R`mM&=uwuV=L|EipP?Uc|szr~Jp(OxG0OA<-f%vjOL0BXo)k_xh)^6JFJQMf(Zc
zRgjQW<MpDAn%D1`jgt$j0oE-+am%PnD_L*O(pTfLoZvt02mMtX6TebI(>PG#&vX&X
z_~qB5KGZ&;SGW9d<$Dpn;e)MbnFhGgg%4{L=79+s+pMCSyl}-dppdX_6$s5R9l>b%
zaE$A^q#kOMfSH}?y6q#r4OOYyVw6pxaZHtLN*^>{B+oHtssLGK-bgDe7XU+NP7Z0f
z-uViFlkRB(mwOhww0C~4@VxV3T~MxlWr%tx=PX{s*Sr8~1r;g#J88?SD4Xs6f=Nty
zIO)@a3%0TMO)#*ICQ<mF1Yj;i>22v30ggzKn=JDisj?dPycyQzj+G&#CchXdvIC&o
z#Wz<cpL~E5hiZQl`4uRj96db9BzR*j%+;|D59#zW!Un%GF;@5%DXLRW?Xw0Q$}xi-
z3C6{D-X=k-4Ms^Jr5)wuf@fVAad1!4B+x(nbwa}_KX*Mt^|sy`nafJo!t_`IFa{40
zv?rf<EFoa5PVtRO^|Eb>ablYRouv6gyh8W{_Vp<mvHc(GyC$dqzw<@Z_zS8=<x==w
z0c{Uo!f?w(UmBXS!0pgi|5o@<|0XnQyhrl)=yq!940g#AaenGcr&k_Ls?U1D&zU5c
zy%J1^ri4l_Fo;+j+sis+MMWcS^(dEcm#g|&^~!~p&nGoSTieInqcfbo0}AyCSJeBf
zZ*n_huDkK6OHn)X<9tts1_-P?_-OL|$kgmF>F4$kqwU3s2JgK!7N(t=Ne~(Ocb6;a
zWWRS;*d^)g;>XF6Elu~ScA4*@efQ~&t2R~#+8~RKp;(>yL?Fc2I)@7gZa^MgM)+o{
z3=n;cC%4r6p6KdI!dbhhW&Zp_CNl?UvI2kxL46e1F^IF*CP)wWsU@hY>OYbk(wL}U
zBT+HR_6XfbsYzC9oqBgaZ4}m7Ai7OVxBA{=VIi)ObUs~1VFidA;DdOZ#E;2s)aLEc
zpB}m_EZHr6m9Zwu+nb*%*4x4>q|P?FVku8VJqhs@&?Lh<T(q9JN~qx~yWG{GyJc9h
zL);Ijzg2(oebWh*s%Ybr4{EL9FrLsCdRL)O&v4FixHqQ?!nFc;BWs%>w4GMJ04^DT
zGjau<nMHlgHF-?D<v%)9Z!9Q=eoQ!;5?gD(0uEO*6eG_f6<X6DM7tM;%v0M^(5zy$
zh%TtvYlR8ZS$9%1OEp4LR?oJ@6@I+?e~0?N`wxexad-6U5W9IE!~f+FM_pto$Ju^5
zPc~Fopx2HVmkg{5hKc&n3q7kW>l=S#*!@<^xW({L8C+qhXq}`(q9C|&C5qf9^#U07
zltq5O37+vi8(kE=7$?hVjPpp(UAneL-ppGm@db)3f-|JTI@P<z%|wYun5dZ8e$4)U
zGg129he4>(U|`Y(`}Zf)6>`48utX5nd-gBf@4)1hzrFf73q_JLWe06WK=!ouf)M~1
zR^6JLZ?uus?&5fRc<!!x%UikMDs(KIft|Vd75=Nwsj0O7AqMZZY`pgt0{R)wti64m
zXI?Spg`(z_1E2D9$e&dMuJDSmAw$-r^J{Dz{kn$f{Lg?%6hfFlNUPkoaF6PrNBzLX
zz0&hz6^t$Hd}SSH#UGjs=K&}{DNnwIaULMb7JuvV!wUc*Y8cOVtaQ%lr_v3)WY+_=
zjXHn&$l4z-e5N6WIh&Pw9hvEo2s)V|ZI#tm%iieU>KJTa9)kMut>!!9#iT1GkXrn4
zcC0AmNcdJ@VBDks?;1~7Z)<Q)K>vI{KZ--1^CY38PdjKnv9s~3+q_wF{qNR{wPeyh
zMf12rV3D&8;-lKaJS50Z@QZ8g{oC-1y6=Yf7+hqf;kTr+m`He2&+(+~LbmxB3Ibhs
zRXFYo)x@6ok<J?h&FZy8%Aq5MwgzVTt9|Ef%vg9+QGLq!=`kfobViVD{sxZ{51@2b
zY6fU0g-Wr1T|qmwEYbemv>P|CRfQp>0C`g(rW3J$^&%E!VB6=UX}J1W?{-eLdIu6*
zFeyw&@yA8@41s*$rz^>(=}^}O>s_6-i0zY7P}3<^yR89J*31|1mn|QnK?h@+<jp(F
z1<6h+pIp2K%b9Pl>r}?m4CLE<lsLS;Tza%&Jrdg0sfQ`53vlKAT_2b>_L3iAarA+U
z+$JFFvDUfu=tZTInsNpaB4!1MzuU3-5>%9=uHX16kmti^1|n%AI;VP$d}p|h_?(rQ
z>MU#f464;J-QO7!__c}HO8Q$-lx@=r4^$qDBohaz(aqaXqS)~$^bYw5V_J10Sm`k(
zBO{FR5#l4&_5i&AV{YMPl%#E^w0VcL@Ge!~$;`~7f~T%e;e<*MV!v!*;0#t}gy=*V
zRj$%lcpru-&plF9zTOb0gYIieeK1=Px5aJgY!06`Dj?sG<L51kE%~Dz6Eu>to8cI{
zd9>q@TFODYbZ`T-pPIhuACQ9SkUZGfes$zNe&)$YqQCZ>%_5QS==J}ud37wcf}jz5
z&X4>-Z0wI+%BnJuT%j=0iH(?I5*o)<k(vbY-#PuvA9RcC@Z@sVI3vNJQZf_EBq7*E
z+HMGNzEDhwCAv??9HjdK7Ru~Zo6u~xs8dbY7oCL&)^lBfXueVU0E19@snqzV(g<Rt
zqvoZMRS3d4M0MqVoN=5My#U7H2qkBwUfBX@7gflTB%{;K3Y5n2hO?aB=ac#fCV)qU
zFo6z|KiauUk7g6Zd+N=i(@zZStV=4&k=F{-M}pIMxN5byE_thRm5zgtLQ&iCQKJ`f
zdT+}BA;4-#F~Yv|Q?U^GAyN#b+BT4ZCu+}45EwlEhjn7v+IG1nP85oI>Nf+DQO(NV
zQzAQ?lnYJuO5^vA?Ej)c9$EBb__s}&jlpudQLeAJSN?ld2)fJS^?ORq_~>;qGuB6y
zB>jipm~Z7qb`=Jd*Pq&38rT}He@+RBpdBEQ(H`5jRaXu1dm;z*u52%j+OYLY=BdMz
zI#HB;M-r!5$j3R_7qnxlVP6~FD%(^`b~v$*krmS9Z0WXLnTNqbtt~sYjW@HHp~m|2
z#;k*GmukSDbaMmM!Bns#l~OT~_HdYT2I?XJ1$$ObQY`6@J^1(68|_T9?;7&0()L|%
z854xHJ3y3J%@M{U#CY?(^B|~3m_sH9XC!7}7oEB6y0Frnt&~9kQY{q&`F4iMBN2~2
zL;BW28V??A%XOp#6MO?8A3n}ke2Ro(Q||YWyI9c18GVQpWRreHw!bCSWmvhsIom_}
z&gb*>$r@5FH9~^pGs7>uUiaf0A>yWK^<|(F5*@|O^7X-kHw~{l6>$;1r$A7%{<VtU
zf~I$YGv|h|M$TPye(bD&N;s+8$jBnIQBXIlt3Vik!ZqibM<laf_J+c+sEg;LhieT}
zjKd>>=3@+Kw4HsrpSO{@Lzm>kDhLjwb%#|0vtv7F$Os)=m5LA?AsCm49%#8hC>A>d
z$Y2^)c;R%c2o#~8(fy6G_TaR>h*DwPCc3p`0!Q<%UC#^6t=r@+BNfu1x6XZ6aXD#y
zSEB~OCwA`(%0Wdi5r=kJZc(%d8uVXaQfz&0^JR?~GFt0jZZO})pKr{tmPSPkypn`~
z?bqiK6LH?h==nR6#17>#$*b1d<V4tpB1?qgcR=zVK~oEV(rnx@BfYP05xfFFmJesx
zLzHhrn@`F4l%+}fgWeL$s8bB3csy$0C+_e(^MM}9fBx0e@m9+~QHN+N*Q>sn>8w9A
z#TvhBx!c>Lcf6Ykxq_W_WLbL7H(jn<dvuNe>N<zI_VXkhQ^f8bd`s4(Y)d<J#%rqP
z4egzLh74TVQrmZsjkT!P-IX@SD-D#XR>MS-vICUr&o-s3k9<+^_90O$KkPZ&Fp0PM
z>zHEtI+Joowwb))Z0M>B5NVppY)@vZm3J23Ma~iwrR156Mxxazvj5eXCsYW4M0@zx
zv<r{?>iWIsWlW_0)S7zi_|EtkEh(wV4_zSJyY9Io3?^9HH0G7jr%#{~Z_gOQM>%sZ
z1BomV+yvNf=NRrs(<1D9zsBUWiKFZBccFwsdHAW@DJ}RKxgULcl(R8&6#uEo4rcc+
zfDfCcC}S7zUID#qnBqe>y^e_<Ifyjhb_YQS2f;YYUM3~YJ@g)fsoZ%}!qTETZYH0y
zIH=6y7~5CISjdG1h?aN)H4b{GWmi8vmLF4qsZ^y@5woYCi_}kz6Yprl^()2b_|zsw
z01tDY>Q<QfW{Hk8AMQ{p|J)aPuB(UUPSh5Pub;S0H#|>V9k+5I;}I^?l$V1bfE_;c
zIQHx33jo<Md0}X7SPA2e)&~QHbywr>AfZNtZ-IZY7yP$Bc7UTSX+qoeuen^XFW(-5
z$o8Y3Lf(i{SrjGwv_3xQn%`b5s`WSNEY0na86GH=8WY>*8xAGBNPp}czSn)j#}y#S
zfR{a3Hd(OTR%Kh#m*Xu=N$vqU_I`(|OeI;w>3a(r`ui}aWz<+x@$EqKEuVL7Zm5U<
zF`8Nx*gw(d?Vd$ORoKOvm_>pGMJ)HlQ+2NQ;RcdCOJ|=j)=JYbVg+`5pL-*i!ICti
z#jYO97%;B|8>-GT4=&NWN#Hm^yg6KhHB!ssY<Cc34s%5GU3kf_OuTZpLaYj8wVgY<
zJThGuDzmk|N|CBuV&H@r)<tdA{d64qKF3@qvL_1BSP{;D51Lw6xMkFeB|nEU=_fRk
zdXhwWtcd*Le8{2KoLoX!$Op#ezh#s&5mX8qT$Sf_*H&kw&$Xp$%tWKF%vB~Kyatg)
zkH6umBCd<dgP3th4D-B2Em6B_K?AxK=abWHRs?X|2#<+(8Q<UHjhI2Z6ekqxAZF#v
zWi|oTDtEox(s%v`=V@NATLyTGOn+{sM145yL+ogao~zDh@C86JqrD8ZH<kIN#%dp5
z0MSAz!&@cO@LO|7rFp~7W$29_-oFw7PO7-bqv#`T|I4;unT7cIX>L;qF!zeQ-`ebE
zfbu`L4~@`v*bnHAxYhB=0<}ieGsVao(-)NMFHq+PN&Ql_Z)I)J?$}f`KrvTR*Q;MW
zBd=T^9ehotSB?6Ug9;1v#3~>z+6vvL-ZUC707n~r=Qi{%9|NhT%kf^f6%t4M9#-~B
zwnj$wskhepB?2;EfJmIl1`W<v7OZ{jqqZ06_sryCK6QzdSV!#2%cD!y4VLrYeloji
zOi#xy(YgnRG8V&blV#H_YQu$4MPI)j%=;{PSnm#Yf5CBc7yWULV^mON+|+Llw`TR3
zGl%8k3k{XZJA?Rv=RLVtdnTTfV@l`+Xz#=`Wq#%|r5ux31%h3jrNl!xoJ&=OH;+WG
zEFc06u7<9p>O!gHtSo;fwi20(6NKWpoQ$Tl(5!ovO&c>{ANEM~Ywa6BV7v`D)!~LV
zbP}~D_;Ju|PY{?QlvXsPgS3`w+xe{xt#_f-09PV|Ywjl+^KmLO9+QZ+1wbn%`>;=m
zn=^NHe0$Ahl6k?8?#)Y)`t9@jsyE@Zc$%{9h_It#5>R+8<--~N3n1k+oM-;WPLXrp
z`)HwygQwJ-`^Y{aPj2uwy05wrt6R~T8S0~d)#1>+e8mgkzq+OT`{*=aN}I}UUC^`e
z!nN@weuU4BUDhDsbGcXHE5&`?t@E)z>)z>k)cZE_1u$y<0!RiQk717L(>*Np_>%pD
z82djZ3aVK8p0yb~8-pKPUbmw>ys{m3*<S!7MK`9QyDYd*#~VU%98a2$eVbYB88jDf
zUU{>04T=}04uM+G&zIoB<=z|R5KCI_tw<G0DmrO_A%RYPGq+eQp9=qs@gNq)dPx27
zD*^uBlO-Gusfa(UE-DrwnE1*)BB;(0(K&{s1LEkT`Ke_lIe3=|6m2q;VPUVS)IU*h
zY=OsYx}`z4zy!4ty(5;Ny+-J40S#qW-#p+}ZxV2YLmuO|kd)42)MEJ89BChhi?1Wp
z9avUWj_VKA8z+--on%CgIod5OYcKUBFBE_*{hyZz+3hvc8!A_tB%G!<+~Jrs;DkRo
z6@MUc0rGH4W;++XO=@zCx$6wW5EvvJJb(NxlAu#hEX&c}QGaU}O>$-!9qn;wQfY4@
z)WtvE^f^-RC{=dYu(XC*5R!vb8?&N_wnTV$rczCq#j%D+@^!@vJgD_9nPw|huEA>!
zK2aM(W1w_?)-U_yD?orsMj1RkDTKqc{^N;=dmY=*Lbv#uYZ){6^935-`u@?e$OC0)
zRquP=(MyHd^G@pX_Hm}9hLH=Nw-N<3qy-J;HUhK1vwZd7=)j;QvY1r&9($9rA>FZ2
zDA$n%u}iSz3Galmo<`#zds-!u8&MbJC2A6JjH-oSX4Q*ee6tgt^uDQ_5RdeE3uB3<
zV!>Q?_2GovxcnGsc4jo(ii7AMUFP2%v_~0csTXp}OgoHg?ho9BT6)1h-($)ukRj%o
zztUt930N%1#M@;z*X)c&3%r6GFMiUe!0r*18N?cO+`YX`5DIEV1HyR$GCI0yyIybO
zRNSPLk_;$e6H}#P+ffltoeGtMcdvN#jlzHh2SOE8s^^<cEGv1N^}r8%P2oW?dciv|
z*gzq4dSO66VQiZ|@n<amMn>=>!`XS8wWTxK_Zb#6FcU3$)rv5U9v$5xNq%d|rCzp#
zAhKb>sjc}(*%a;MFC#VD6=`cdffgpUm{Qs6jyVUN8D-0j1GvrGQ8eV6OIDS#alhSR
zZ>wVTg1&v!p%arw29OJ&A(G~Dv=2Q>W{dF~M442#EexnRIIp@N5-Y*x*Z1i~Ns;`g
zH7aat5cW!pE~${uFz}C69R?bP>A3gAOuSJ@JeBMDXm-KopAN}H`8z+jk33u46fY}V
z|Dos7$w_>F5$x(O!!8()=$~_$IGRbEMZ7ci?IQKvDG->j1)*oPNFv~x<|QIuK0*^6
z`SxS30^s7hPxH-Kk=}>7SGC*i7CkDzJTQ^4Qp~`u=0s>(Rr;-DM{HHmh%#kP$dK}1
z4T7dy`MK86ZTamJb(`kf5%k}7#X*)9<n(JDD%I!3@6|{xn%r-{?9OqoIUjrSJG59d
zeGQDJznDCHAg3pqqS5QWLY*%Zig6_G=QkPr!u_aVKr?FF63I34Gjv6`{+v!&Mm08n
zhRhPQ{z|T|;CoxVu<^uA;#DI?d=~z9ag9fZ#r|;h2kG_g@m8kM5xC^4VqrEh9~<Mz
z8vwAaUmCP`UhH!3zWrS0@_S(ILsIFlpD-0QDmY{v#ka3&G<2j?OrpQdvg^ywqLin7
z(b4dW*7&TngTS4JKy<pJqJRNp?mZsRh*{YRYKm+RZ{O55j?r?t;ouJFGgA#9wj-~N
zIk-k!mZsQi9)X_QAPA)SvKKkbvS^#XTXrRhb7`7agbraK(%K5gVOFg$a`Th&MQ_q?
zvU2E@Nah(eYDwkdyZ~^{DPcblDv9+dczW9$duzPFEHM+Cwf}>-w+w6R{kBEXA6BF|
zMFIs1K?^NTa1F(RI}|7m#a&xUfkKcVEnWx?!ChM1p;)kDL5jOePyT!FbI;xPx#yfO
zcc1%dJx|tJSu4+*_nq?{bBr;)j+XrI9n<BBb5q1myyBRuf<<cEpDuWM(qFloY7Q~*
znudbxxKoGc+jVxQ^-<iFYc8@Y?@6*SlJGB<KNZ<tJPV&?vcHsF_8dK_H)?T5p1cOy
zWS2ymt2&G9Gkqu2&#|?nI9HM035ZVJod4n?;vN(mwlD@t_a!i{P$FUR*?G5zeg4d7
z=avYstoVQ}$fhjgM1)~<^KetYt)82X5-t+?BzESPGdqZN2~3Fz0yWO<8VeYf8oqIH
zekWe*?J(oQgNDjf<uE3&Z~-H8Z4IG4$i12ViaT4OoRAOFs+z)4CMFde+jVex3xMXA
zLe!zoR)}Ta=6eQT4khT;!;n+6-LukS_tAWK$124nN$cg&VnA}a%$^br(ESmB%BJ)?
zVs-<Dk4@eUYG}Fh%n~<M*Q`C$GSFf!vNqv>D4!JK8;tn_SgfBJ*V;)QJ6zFC{<eR)
zqN&Q-;-gpLGcTw$Ewk4T75$xNdc5u->N_3t+M~?Nu-w}W?a7g={L5fCv_ycRuzvpq
zKl6++Y2C@GYONw{nlBB*tr*xKoa4=QRHJg8b(xht{nKNN=JQsHlK@OHtOZI$RjwcG
z+-~**e^+i&7Ej?bC6vIZ@L){uBI+Ps(RH2QrgXlKFRF?<EcKRNTN}uNx8uBjs9Rg=
z1hoswh#5UGtu)?dV_AB;hf1-fUbt<N(?(Xp0v&O(zYFb3kz1{|(U5j8H9g=g25hb;
zJ8mv=SKa+)RV6eY6vxQI%oX85kA0dzbG%e1z^>A0>@DZ3-e>(8uxf{vmQ<e<AfIpn
zXD%<OCBnCEZ_4w<6&-^`|GjVeNouXXvgoDlKD6F35%I3qZdTh0-=58@;KCa-Z%*Uq
zt`K%bvj$Dpj}r}ooEncW;(|W4VvWU<)XQaWCkf=Uhay@cgvm6bney8<S|c2hgo7LJ
zFa7^<*iS=ilBy0yv`c0SMCTwAb8;c^VY~Y63j03!TFgZH^XrX4DIYf&On8$APPq}%
zP#xOMImJr`TTX|nRUD4o2vY55%Uf_KV!yqd_dOmXE=wt2e2El&&C0a(XVfQ7A^LK%
z;umhk*w@{wZ}#;IIZra(H-Q@uB*7csI2dSV1A>!QJ<X&tcCp*=NlAO$K+$UebJXRt
ze9sKy!A7}WkB~BLt6pISrMmDxULkJ{`Hs=P2_|PwgFo4tU27<eLiG+WcW}w(k)gjN
zvU3)LYuu-AiW%PzPL8bBwn-!`w`TUBW_p~gdlwAz{Pp%pQg|E53OMtNgS=}i0AGgQ
z`gy<D_baRb&r7DVXU1#H>w}}}Ot|D`lL~KToV~9!eq1|hhAEJAcvK1{R`=bHL*id_
zJ={kI3Vn6?a4Y=HR!zs+Lgj(V&4P}9-DA|_k#@73w$@-zD9a(zu(EVdK8u|h#)<VB
zQDgKR&lo6j)sMCIDAj-Rrd~=rc3|2J-U>BwnbP6&23d9}s}f==kg0+^v`@ywZad^o
zK~bfmQ*Ikhz%=S#D(=$yGVJI@`O$)o)sa~q0OMs?=3NruMR_@!bxi3W9}0I7ru?$~
zPF8~#%<?1owx(j_-aGNfyW=)xV!7`0N}ZUzr~yNV8Xb$>O1PT44KYb!OjM|j1oK-Q
zqqIIF&9s{MTJ^serH{6ytxjro`G^($1kxh?`o&q@lT0IB$-uR01449(a0>{s7cqpp
zTnKl%YuGzG-=ZHCnu8xc^QbYO|FbQ!+s$3F{ZKiO*Y%F1K2E3Dzn+J;74yIzr@+Z%
z#RYTOjB()AkNcEwBOcC7kWGfZ7qOyyLe+Z>Wc0Emvn<RoD$YpzKa7Ko=+)n-y|)>P
zv&x_&65DO6WHei27zM10DB`MiS2KHG1;TdR5kJS-wQJ(Q99#>6LRZU*>`%scU6wAK
zT(J+@ij3swow<$K`GA`mJ&r<HPfpsACz20eXlUs&g50n8j9LwNqjiY7MOF1DZdCS&
zj9800UGLIz1UFw%UYj%>N*i6(%oiobRu(4c(q07uXP#c?aqdch8cPDEW%}IXL7Fk{
z`#wZW+2888zqbZnu&|m)J4kRGyyE!J+7ybX!@3mx2PHBISi#aCUzl6oG_!s&Lk<E^
z(P)t~GVP}%4D)R>c1e-y^(CNIkaA{vfe@46+Sz`nWdJ8gbIjyrzTah_FA}jkB1vU#
zJ3cWJu$}CmqjBbti?b#tF`7F0X1KNJ-6y9|J!XP8dQ>Gg=S<t1Y__v%-D(t{-=?Dp
z>*{(6V>>qM!<0%JA)Ne=Mi0QTUaClFE@Cq6XRP%D<s<A?pq%6KyFO=jMQur}tK$6w
zW~uYOy-!AvCyN>FN47tVn%R~FZ@&?~#y}Q8Pw3z;a%Hp+Ue<dp>-R#YvYRGv!vv!e
zwRB3K+ql_KA3i1?y&NkmPI@OK@NRHnE4L~N69iwO|7u$ppwZ42L;!Ah(dEL+DhrvS
zdG6e}amHK0hRr%AlY1IHJ<w|F2>9}j?{bOp5pu4WQs&fJ{Zzu${eP<V`9E>uWzgbX
zweK<Oz5A<V+|ij|y@zERPaj275Z$~+?Y~VEWAX6<o%yRVsDeT~v3-Y9L@)Z?5xguz
zyE|Gdd1&G5mqD)i<wX^(J3AvuxVQWOUEUGS;};3<fMA@FC-hq*B%?WZ9LZJrV}@v2
z&y*BYLR)ptT^!)N>x)e3JVV{;tTk1HPLuAu$)@x1k#Ls0Q3&}hT?A;oczWmcn7J#B
zq(Xbigo26do$jw6#2Cn(eNcZfX0f0@9UjEiEMU02kK&;k-eR4>6KKPU4=4Y)ywbj4
zOaaU!JD^h>fz#j+$aqa8@gv=*K2CU*2<sam-=|iX%E231P$w7GV}cJkfvikhLz8-E
znMC%7Zg(&IO<vJ8=N(+;5PA#i$I<;*y-13}Dy$t?n$wcV)L={GC6Drc``njd{r`la
z!2k3amN!ki>sQKi(374quyjV2JCw$b!GRkx`KgYKKzcjz^oZNylkOnRcPm8)*dJ%c
z$J7np4UTG##EtptSmC4vQxZMV<@~oua17(^kcyV3&rgh7wm7G@K7QUrPzS=z!dm08
z2fmfsSku7Z&IKN7ko$?_1CTnS&{XI(u(CfMr5YLW=Ti)l6m@2eT@q&CCNxQ5Fi=?_
z)Wy?J%9|{R!+P!WuE7b=MXbv~-{&zl=N-v{G_#$C@vE^oFdFaf0`X?VnUa!=C%&d;
zSH|~l6IjHzjM05rW?K1uF)vfA+NOO)w8lmFz9Wee@?Caf<%hl3)jZ<MYVRi}P`}5g
z8tmO?#Yb(r1-Tp*!DiF=MvTmC=z@YEXxdR}>a8LJKclx(ZG1VWjxn^>Y=*3k&4*D|
zAuG!bi4vpeB2S*5?JQ(-$vXwWp8;5Ib(r428NJT1=J5jS`F*yXsM!mhA|Etvp_S2W
zT5-GRi3m_0P*%AisH|X^=|HUcO`8mYO8U^&7oscYACDVNX4}yaMFAV>q?FuiDLI2V
z>7^Wpdy2sLM)V&LFiBz7|Mb`$FiB23af3f`Lm<@+FgOlWUQH)L3%LrRI|OTvO0B%;
z^Q<}OyEUsro0)!HORKFvw-fmM|DUzb=g$Ptal~RzNBf=`=<2cRD?!|bwVxCAwKR$#
zwlc%v^o>{c=E-%Q5lzytnIk0r;k9ZO=uEt&Eq_3lQm+*QEr-W^)!<`#xfjP;{9&hg
z+KqSrhF)UYC-|iHFGkhZtq#Dbo^eb`PSMnpc69;DuKcGI$g2M?Vfzn5BLMu@zv2zW
z`fTc?E*z(TI7uUntR9XQ4ax$sZIW|wV#PSvEJWPBb}C#>rnxo5lOW}Zc)vIvHAfy-
zuBB?9x;G}DlGyRPD90ig`h>ni_)M+bf?X`AgkcrpGUvMzyHX#_#)`I;#nGl%qJJ?S
zpB>%p^n8R9;Hwa*Ji{QykkZ+2r1You2=O{Y+>b-i+zoW~umd`l?fn<y>j_#vBeZfg
z70+fS`EuiaXWQZ?2k;l88?9Lk24qR1X^gMXOVzbg=TQ-Z;u!7!=1ZP$KM@NR_eT~j
z2ZjE{AohJYpItLo-w^BYG$C@>S08aI_M;batoU9;mbHV^R9OI(3kk_i${G;E90iW}
zfM#Aqs4B*Ciu32h+Drerr_4i_#9r7)x_%TIKel%P%0Yf7zzXa!4@kO9j#F%<dK6~=
zyWsG@-W$cbne-vi`Y(pvy_MiAx<RGm2Ib?>J_-tp8|TcT=AaN!olLji&*pBD#pAXp
zKl<}aCHUZmKR&l}$#Aj7W%ilgRZjcK1Rlr5BAvx^d-yocd%XNxQOE7%yj#RzE#5Im
zLDPP_O>(<#GOgk~Zr820zxHS~G1G}U<(GS-xvY9&Y>Z2dd@QlIH>}`8?NnaH?Hg|x
zu#W+(G^mQ9#`kK#pTUh5-v2AwX`#&QQvD;a>NXY{*q*t=1;)Q6R5~w6{_}<~#)D6E
zDjps_F^_5T9(!d<e)=|uqt8PvSdr7xnXULjlwyuJ$H=vIl+Kzclp+SzOT17O%i?S9
z@|c2~0)u(s6#F7~UETGz8fbr1$C%R>*JWU|Ebhp7E~`cF011$CpwT?m6IF{_vxriU
zSUd}T^MT#jM}V~I*^#Y1psZ{sAei4`D+%-$gV)$&RN4L7gbLSNpq?m<*WTqqVnUaq
z2H!x_BUspX=k<*N4+7T9;sv@qNCS$-C@Sjh-Vv9u2`43x6R=uR)Yk4`<0YbkVkzTW
zo9YCyA{hQzRD8o3A!mkZ;-G+?+fGptFhaT<I|o3&!yNqu`W-^7GuHp;4$kL?w`67;
zBRjoGjpgQyRlHygUW8utAI@8%8C#v~JdkMzgk#8l%kpXP^iOa{XMG-VOB7MC83>G5
zWTHL4JU)cut8p}Y+pnyt(IJtc(>-2!Xw9A=q@zu}&wE~zgzpSgG-I{EY}Z6<ps~+e
zo-=dwR&c69U|2HqY?oNUf<kA!nHd8;Rt_ev?M}T{yzKT|q%2i&!d|jF>}GdYv+W`;
zp8s(?H3xUNuJ1sM6Z;durrw}<l@Gts#L&T;oW<J^egrv$3_TUtne8!NW88q(O*d3j
zz<ldTusEC;Ck)aw)Og1g8(Lx88_H&JH~7l6#BkjJbg23Wi2aw>N8*u(=JCD3sUq5}
zEYW^9quibQ-!3ObW6W$87pUcpoV9Dj1{$OGfAMRE#_+pA@PEE{|KIP8Rfg<Q!cR$L
z$|p1Xt$#!83t)ydAHLr(Rpg=}!!xama{z_V4deal>yyjF8KEPfz3*;$Nj(e>G6N|p
z%iJ>hP9x<VTnOaI$zfsS3Fu#f-}vOkRgjv41Nt*q9y2=-TL$x(yc>nM>+TB0LSD*Q
zk`|0#?>!zL=lrODNJ!K;*UfqU9PPwP$p7sIwJd%zt10pZ+J>+0C{w_F?w!T<=XS59
z{$#rJ8gJk}C%Jmym|>=eJ8`ufWBZSTEN=;@fgEGMARD!+(C2>??-9!C*}>X<JW#I&
zut0EFv$e!{)M^xk%xT|4@+Orv{$jYjGz_61yZ7DB^s(Uf7-zRmn)RrZXHjD^cF51A
zIK(*s{sNPXwYqIsdR*tH`dbf-7711chZm`qe=8G{-Bph=mD5d<^UAuylkeDE>|VN)
z9`;{JbT<mL56Px^p32$q3{0e+((+n`-oS<f=>ffw^pnT+B`in#<Dh70$EHU_H*FWL
z(zU?il#w{(n2aVzm1pu%r)?`SZHXHu!q<3wL-pxHG~BIn5z`T{im8t8uVwe1J;$?u
z?mp``S2@*NUuP;~AlHYkXZ?X@Zv0n!?xioQcz>!2{qXqmD^T<F(EU*aRo<sxnn7xy
zX3O4l!O>c*>3&;FZ=`^XXt56Jx#Cq=C{AVJD`B+`EfY`KV#mFCC6)Kb0<iwVTxY9N
zmz+VXkRH?$Cr2J?5RhoWH{6-E;aj8M0@@?N<<A4L24GKJaAfZ@GMUnqW@c3Ed@?U=
zDow9_`jlek<^LlK`M03V8p`z-W3kQmx-6%BI*H9~)8ft-%@m^Yy!e(3W;gmAt>5(@
zxvTz-+e7}xM<KN9<5t`DX6NN6ifiOwjM_}{8P`bbz_s(QM~tD%OcdADoi*tcJFk|}
zvA9i=<Cr(pmhy;8v-&B^|5X!Ixf{*Q2lk5giT81HEwc~%2f(shQ1*saY1DE^_ybdS
z2YQA1K+dwdtk$T*GD5$>ke$Yinj_qw(+G|kbYLqcZ-6$|@v2VTn?LmT?7o{jGw1m}
z*V^CwLg2FcaAObIs?O<gMjWPZe5*i(A<M}|Mp{5X!mfS9zWPm*)|m2aV53;4o-!<p
zrJ(+>qMtxo5vb7+_~&b6-_0;)Vk;C}?Z4~S1`pT&^maxn#wU~Q<HxdK8O{%rV40U_
zri%#WDNDD>CfC?}r+1Oq*RDwwP)L7MY8Sb%QUZZgcE;C7+s%Lc?t;FjJn`OZHl-Gz
z!}_#9DEA@D3hyMGg<Zm+FuC0ocb82k?$oU9k>bNfWsy*S&X7&njEOYFIiDH+(o~#F
z8)%ueW4+@uTc{(kHNr&xsQ&S-=RB96FQ_~f9H;3~&;Eys9lVPJiy&-~ohBe;Cz9Th
z6sCXK$CS-4lG*<5gFn~7`W{~UETzgWo+XA*dsaZ4bC!DYretu3K=cOf8)0AZF+<L(
zUcz<_)}oX4(PH*z%<<dJjEz`3<dXH<Od@JAQyXr&Bmb0dP{rs882Egx%g!}6v3wMs
zsvy3NPCzk1`aI~3#Pv^_O^HX$6wGQZxQ2|4jSLK?_Veew7#|CWGT%_+rK4)wq&j&B
zLQ;ix`?e0)*i~$>0;VxREy@O~&-JQ8T>9-i=Q8C@?UMEH*6I!6j=j1pMd>Q@z9*Ly
z-*Y}aCLAo|#F4-1zB!AxJxQx&rO7V)R4|m<78euDQ!!BmwZb9SUxS2U5Cd`?s)CR)
z6>2<}nwSG8`qf??r)KmikhJWwX>v1XTrxRTMtRH6<ZNJo#3|!5(G=^G!v%AB?#R}J
z!UxOO-7eMBjF852rRfCx75|8nqOT4c?m?PVsrdu6)yc(EPi)<0fryYHV_Yk44x1#Z
z?=pnfGO2HsuI>-|i;e1Pzl?)E!JT&XYvXi&zhERJU`CV`hvk3IO~TCZZ5!_w7`NGx
znmI|%3(oZGhu5VY@UEPJ9YDqqN@2MM3G61Cxx}Nan-I-U*GP`U9w9$7W5}?N6ARaE
zo;LQ9{<zgKUZ)2~CzA&_#UO6ojRaDj@+GiYeH%u7YeMpANWUWcLD|yxmg9oiVqKeN
zbBO4NL`GiN1|8RkS*_|Cq=m9jjV-dm#1NA)9zDAu4yI@Ef(Wv_;p9GliR)PJH0O>P
z{R^|Y;v!cUYm?c&#wrDZ=j;@$n<K&U@o!V{5;onIbH5}NLy9RMjCap<a*J`2e6V~}
znRlHwgHZjP*4GaUUZ*#qV=sQ2gT7I<{B~F0E2&A61vjYNkR&&eGk~`OGa)EL9#=b{
z#u?R-v~f#PVZWK4q#VSI{+5|D3}HqgqzMx}6%%1W4f{`<pr(^U)?<`q?2C%ev-T4R
zedj`6hPy;lE}j;d9krtq_K=RWCd3Mll~lnm92yz5QUe16!?^|D%B-xTx=P!&pJ{`B
zi%}jKl(n?JG#ULw@KmslKRgmm!Gwb$i~4n#hk^JMBj6&!Q(h7WH+HC!1b6fxI^i?Q
zPqKWyOTNlZG{HtDBTaXf?vLTl&GOMbTlp){at<x@l|eb@QtZNk%w^TLtnYAZUOa=c
z6b&?G`0uOjnwh{xza2Ui{K-vRTvdp_R!*UaK4~Pn8*J-;z9AMPwaYc7+-ZB9b<y<w
zJBG@>cyu^l4>ap6l-4?yN#RUPDZ`05)2!67JfvQKQ2KMjRr_s#Ti~;$jozCyY>&d=
z{Vbja+$C&+dY=Z3#)e%ZEp~1&MC0dd{a(iee>JD6^>B(mKhP<TNnlVQ&8+0&xNK(P
zBjI+9DH~4$0V8W8!ny>Nwk1e;U)JqDVqVqDty|+14CDSIS|B_x3kfFB7^J`Xk?>s6
z55TWm((5;=6)!*R(d>vYcHx0IwDQ$04CutlXvul0U|^}{xZHeT8&9rYa;XGn`U2dj
zGi+kQq^RM!xPQtW>TqEQYb$xPRj~U@$>DF-qDn$#%{cSH!~BI6#wqkyfV$ImvIQv)
znVbYHh#1Yc49Qn1o$=k<oE{89a`mW#8L$OnHu<1KSsR>z<D8&Tzk{WRro+YmZfNsZ
zHlEu|s}P^7LHT>bhe`PtRiy@&+v^$6<DeEFyWB}7N|8829`Wj7L%$GTynmn$W-*H>
zE+Qr&apg&Mao*5$@g;%)Hou^$I3;m-<jL5Z*k&RkJHQt`X5Qa&TGZxpgsmcQ;t~bN
z0E3LCDM_^P4|Agh@QpQ}h|&Yip4US0?sS59Vh3sm3!lS!DkOo8HZlLvuq>zNG{~Cv
zehXVj^U0%JtI@NyC<doIS$+{XK+y@QKnF*c9|d%WQGVF~4z&a!H+IE30a&LlJob`o
zf)A%WD|ueo2BU&eCiS_}wHrICTGE0$ZD8ZH`Tep_pTMU$win&dz%0(nA+nHkyejQq
ztTwwY9HX?<FPq!qA@SA}*7(NMCs*8<Hi1Rz=q*YlZuXB46-NQ#A<-Y#+ftU*)wfrE
zZXQ?G95s~Qyhw^Ht71_FLL!n^rU6>YUGG-^*IP$+ogvWxu}1<}pG^jzZjdm8inhiN
zcQ#*uJvoxwT$%0tQaGM68kV>Jny;ff^{FL)h@-hx{f%B~jNO=ZJ9agSPxQ9{xE7gF
zO3gskdliAz|GmYx0-71y{1+pZU%NV#C3YaPK$Ot2MSK<#@N=?3SMQciS_MOPU*mJq
zib?K=Lv9N|fz8t~=f$!y!OGf%w&Cp4kYpd~q}ReA((dBD<F7yW&D4!N^{BkgTYuLU
z<dbkY7P5xs<*s0Eq`u<pF$l=3DI7X)B$t5WI$;o($*?Qi_#(&U-1vLjk{79U68h@j
zB!TX@ICmH^WfgF8zWd0q2KNVp*MobEgmoO}g5qf8clVIPflC`6GyTSwrUM()%R?<x
zwl>c9-pxDaF0fkEp<;aJmy(d*!Uh3dWt<|7lwFgs3R!s>&hz4@lNy7wSAx>b(ris+
z<z}#!H0}&l>iOH2E^({~-Q<oGwC4Fw7(Qr=ls#i;bI*Fue9$<!++JE&)x14Pg)IVu
z?UHe9+@WvFgzIcLc%)C;jTf`Nsm#?I$ZU`0yJU}fjYDGE&p6;-CRZqX+rVJRt(KCi
z8-X<)NwUfqnP+)<txk7d;fQS7t4wUCq1IIz<Y>NAmHznk^nZwB@OHcuWK>p$IxmkH
zAGx;ogxI3#ZwwxPqdI#{AKn#AM(e*tN9+H9bEqm!(4%c79T$2B{3(Xc3-kUxFD$~T
zH_0<xz7Bp7*l%$a_3;rGv0sRD`{pXX^N5t-p2%WerN|!={Q0x&zX0)SG$KSEx4Fz3
z_;u|VW<T?8(rFuLTC+H$qH0xtFdFG&;1<<qZr5AFscJk;4vYI2X2jyEyf#ucytU*D
zZPZVr+o+?j>yqoDFG`f_o1|y@{GJ`{z>3gRO+ZHoJfDV;ijkuDj2;j=D3{(f_nI++
zRlrG8Hd*ZQfgEP7s*tROX2P9njbULU{aaI8D+lqLrwf7pTV;Zc+3v)z7r?3Y#pvzV
zbW$)5zH+#qMS|AAHqdUxR`gr?o}al?19G;`plW><UL1svOF(~39P}JPusxqG)+;J(
zf(z}pKGxqAv*!DK-o#UHcz{_L5s(D8cPo?eO|~`hJ~B}JhN!oS!h<k6SgC5=L`cvL
zW04R04X1L`4CKHy%F;R|7dLddn=RRezbY&L;(EvvH$V)zp+rLkHzYq~Y}U@j7}{(f
z+xJb|wnqGPkmchzm%}(Zzgleg%y_|m@Hw31*Scvx2f8e)#u{$N@E3z-SYKGCB}%@2
zp}cT-5Lq}KO40mn{DdjA|72-6daX+IK3h7`wDDbo-r|nEo_ZoY%7nQ@kzRafc>`HZ
zPsZxXirP=3-LB^grg7K3t~V8QrizEREw)N4PZRsD?1BT$#kc(`X?jjLzdy9VTvj}}
zqw`ZKkv5{cu%A~+N|RapMb^mu<Pn*O>+zTy4QZke2s?V?)_VVv(Y7JiyPu_Sj$e+q
zYRm}|S5mb{o48@^ZVB!mR}Y>y$R|em&HTKx12z{?=1NcRXs_H(dVWKo9E_Tc_b#|w
z9T5gqr;E7G7sT8J**kOl-g{WBV~GY(zgF=nAh8}iD!6cx{t8Pg6)dxJ9p_XRDJgT?
z^cRD5?O^MhlJOWgq8P<wEyb4^7&2kpDTB{Q0Ai+cF%!oQ$V!2buTw@A0J)wq$3?v0
zr|1zq{AW0|T9iRwsyi&QRX^W_SAGL5BOXqZLt*CuNPcjoK6Nq})%V;ZaBy>h^9sZM
z$QlP@B`;b}<kZ+S@&9AL>Cy>1S$J=LfXBr*d{b|sQ}RB(At^}(pViJzx$1yjB^+r%
zdc5CZ0reEwH*{lefjOT%;ciE>vTT}Hith<wiJccxgIPb1A4AjK!Y%#(!2`tP^{l*|
zn*YZQepHca%;|#fOk9xlz?<cB@Zyb+VEpq^N9z(~kr`VF@5?zkOb(krK8LEW-r<mU
z5N!|!u>w(CO>aD#hG(5w+D?zy*jo^irrJ8<dwdvh%*y4gJKxRW>+y8y(Z%iaUERBG
zJlk)g<xcU!isb<i$oi6}h2o5N3P<O!9b}(|0M^&dN0H<$CXo><fhT#3mlJai^-W|1
zBzek?peW0C+^K6~`uy4JVxjb1v!i!@Y3{<MV(vn)6`jxSe7KxsN|UloElCP8VFo{f
zVomxiGL(t+t~~r(g=?Sa4X_|QzAx>e&9B<A{$fz>JCjEFf89!VbsKaz2~b&e=D<}Q
zFE{UMOtocnwBK!24A|i%X18Vbod6IB_TZbqj(N7!((}_gyT{Dv7_5a3tc^#XA+&?*
zIU~}@3*bFu6@WVD;AT^0=#lpL_2mJRv;k3kS8d|$?-!MUU8OlHI|SN??OFR!^+5=;
z{ynr;S$MU%LsC3=xL*HLOe@BpTNUYdFbWoWDT?jJDj1^mLc!zB;^m%)P}HQi>V>dW
zOyhYBWMYE;-h;#DJC0|<s=@D5imM%@?HBQ;(&JJ7@?bo*R&YRYfSmWTezht0H>1@|
z4(P^99!Pbcr{Olkdvp2dkXA-B*CO-md{xlMw{=_&f<8g7AjDuZ#Ymp$ejMSOm#c>C
zYHo;f%G)?^0^?IRiY>7*OF#uQP$=A7Vh*8QK2&`$Tc_)@mnY80RsdDsGfy)4OgYYQ
zQl7?m@{Sd1^Gx2K5Ir%<L_d}TSpry3qTWhG%c7R0wlY~NDKhe-%2L{6y2L&WSQGcW
zE{Npe;cdhRvOWDS1#$mO7Sm=ibYyKdxLBf{*`SlAGd5o^6IP&8IhiCLFo5YqpfMKZ
z5vU$2Y$-@PD;YhQ?NVp$cCy=aX*OrX7X`xyUZ~t!?=W>E*Ek&d-?j?RnNUA-*)mlH
zFQWZ_w4`!B%zY*^_<X>PxABj_*H6jGBOBGWJR4ufzNF3U)yQ7L#3>F6Qk<C{l-F${
zhx{6JfE%v78KuLH`SB$Q@4AM=C`+c9S>>nq3myrqyWO4f82Gq}KK5A?wOkR457;Bj
zGj^~&fmm3ceQl|o@l9j6%d{EuK`tDSF>tDt5RbKUN$*qm<2(*!^7mk&e4c0&!yfVQ
zHI&68gCS-#h-T@=K(U~zprwcmE-4;02~fDiey1jY6}?F!kTERL|EkSD=1Nd9l;xu5
zSMZ91a0UdusPH#|>nb7w!4+df!_Ai4M)#>)4P1}b*jK{Rw7l?5d)(|s1`IZGxV_Uv
zz+fu#+opQFlT7QUa+7jAdIjBl1&m>JdCYP5fge~duK!|Wh@qG?e^NzM)c-d7Mx(ev
zxd=Xt`V5?%kn`Q!t5BU~Ru1}$5%0lv#wRjzy#@NxI#*aeT@Cg4VWm-btT`n&O(>V7
z^Cws#UNWk$nOfqb$Fg=>g+{zZ!q$oFE2!Gzp7-QbMP6<r2E8_DTG<V@Z;I@UgToLP
zffuxO;}C;`wy?Sjwut2X6C5mNg|QP2r;*y`Y}t&TQ(YAd9<$VOug%nUtIqlz-<M;M
z$Aw{l?)FE*R==*aNE(>VI!p93S{Xe(;y7piR_grcadC%CLRE=bRt!LT%ALu3+m~gB
znAsfPPA_P+Em*SB;QkX>e7FEhVFZLz@W!q235AU(u9pt@oYGcq5;c#P^WCXxbK%+)
zv*rhChHP)Xzz+xbg~oeGi9m;Hzfn6*c+BbZGdf-&%?kQeaJhWf2^-{myrcwW|94fo
zl+NdW-tT{Uc)fBNovLbj%7P>v0H725x;Oo^fiX9lo~&q#R<!mP?P*${1+e&nt{f#O
zHCO7og?@*gDC{!_1I6*yRxId{<1ATc&egM>ij19?3TGCicel*}Ry!fgIk&*=KPdsv
zIV$|0|L2nNe_eitA21A!ZhBZkFe3)Zp-wcad<|2jcs{dvlL6Ax8qmq>?Cy#5Ldl`z
zHwr4~yp4H9+RpzUi*$DYon=k@+OGlXBG=!xihO~<<;f(eAGL^R0;Y_VL(<q2&1a`R
zB6NNJIh=M9#)~OAC6-Nvt|ct8Dn^_Hvp*%4TUCih9@z&xi;nMOzicd`B@+~pLvEl>
zm!6|w!U^$13C5>O5fYar)<N!@d<Gk1LPP##l8z}sWbdCsjbqBau?UbM8;QG0xqKno
znWdXg#yL=#Pid|{F?FQ}oFV01TgYV~lW_sQc&J>ZN5P+iZO2#5y#hhp&@!=%jq~2y
zixoe?Pjt*#6&BLp5kDsbZ^1etw5JFF%72{6^1!|Sy3crMHseyiZq5pxIGddSQo-In
z`9*(_v%y5QvE;_vEe61Bgw`ga{C>syG$_EHPg!x{nlOF5C~I^J$E(8zz{;8q#fFwP
zf~>s6Xr8$jX7vlBu@@#j{yah=q*c2cvKq(DGbSp%IomgBLsY{v=Ix)dL%UPbxw6pZ
z%Xwb%BnkVZ?}-)nzMN8!Rnyp*_k2b}H{L^`p&lb~N(}Elb5)oXAj;s5pdYscXb&64
zv61=X&A#Kd$s@epUu*M1TPK=8{~)3xaoQ=z9r>8|Ssa$cq~Ahuo<zmA73MtiqlpnQ
zVakj*5lE<pq}VY%Y0x`jEOBKTRGE8_4x3LE5&Mhr;&hoP-EUJgQ<3RSOhwVsyX%nr
zl;1$`EOkO4oa_8(>0apNQHX|Q;t|!!8%WwGPf$O<w7p)98tgzAm8Dpp<+hshsa<)I
z4M6t3e1&0fgWtBv_X*10`(xeY`)F<d?wNY;;>(@=*LZ!n>XLUrw8@@CPmnZm+`wEn
zD#KmVRuHSCp+5L|YWTh`IC}Om6v*e1t!5X0ESdH^p!nU4q{qRxT>~xo@Q%s|<zm8v
zW@`WL-nPzx8#OB6$s_RHx5+tpTJTK0546Y-BpokI66K!1)66B(qiq+?5xvPP=EB|g
zB{&wnvObXJB0Bp~%Q)CS?q76rT%UVZst%ovoxUk$tTylViP1KJgYg-~4~4VPMH=yE
zvzusS;MKNd_i1qNocn#nZqrED&Q?}&4Ue=Q5S%!jw*)PKsII(Evb`cg9Z9?od3}Zd
zW5chvO_oo@<oPKqtxUK^bpmO99?h8YJvf*>g9?@HzdlAB(S7~6O#KiuCV8k=o}>#i
zGx>HHAyHIhd_n&#{0*wK%kYHeRUq1G<s$Z|YEYkVcvmlmjbhe>sU0k9@Lppmka0XG
zIqrz5jB~V<F7xtLvO&SRNBTZI|G<bDt1<BMmW^T;pDBNO<cQ>iKhbY+Gqa^WTTDK6
zYxdXM4B})}82b-6+^k*qp9T=g*kxegkL##3j};DqG07;+_(|6ap&3en48z9h$lNs5
zi)Y4-H9oIF%(k7Jd={~T*LFz=%$y>JEgg#rU0%4OobD-J$cHLY_sb!Jk8XY|OjBRu
zK{;a^Hy|F0)QZWbv&i^@hNM+`89>~eF9_Oa(H6P7q?p6FU4?EQ5Z$}s;b6<9#WZP(
z@6A1#l7A|GXFE*Vqhk?1`(w_rBG8Wf8);J)CZ_p>+-Gr|xTE7AP*?xk3k#xk$&W9+
zQ><CMg{X8^4CFA4r>3bY9;a@$Nt5n>k2aBVl*KxI)NVK7urcT@w~<zAoCiHr5omfl
zwdI3W5&ZM3-c8(%qI2EruCYQfrdl4q_BdKvLOwHs#b@vKI^abKek()DBh>z5;+-^u
zB1WBMZ!-oB-xF3inGmS#ckCY3Uia^ZC!X46h;>4ng-nh86zRQB^%b2R`;^^`#Wlto
zFj-sbSL!lMxFF>E*(qegBJFg@`>Xi2%BwW<ZEweuH_(}d+YX_OOS3=4aUp^bJX>Fb
z;G8~bn;a*aIG?z%kL?qnA0O%>ojN*3e{B4J+B1s4<Wyj1tr`2?;$IO?<MdKBQ#t<h
zDhCXkV)>q%<$N$r3SIjqr2XnsQ+)E$3OxPG^qY(La}i`sAgTH3Kc|E-=x~<<L7OyD
zssgUGyPf}H>{LK&|3I$?8dGh)4xg$a_wrO6HY3@&E*Q~GOFo|cjw%|23h9`d*unp_
z`fA6%c=k5BV@00oDaEv8?V6U#JTvLnSL1Cf_dTI1>OnaV&pPoR{N4NA<9S14-&R-J
zoD>h!d~J1^7RMfNm^lY#760=X1A;^0Rd7*s%PFqqz0k9#4=hY)|8`)z_a#^dgt!|3
z3g_f`UJ6oXwp*OD9(L$%rr6_RVNv8w%3uc%`9@ZtviZ|xtJ>DN&ZQPzdD^<wOV~Ll
zpD}i&sxl|(Ook-lV97w*x6pDV)>}Ccnl$vUqutT*GQT`Snnk03#&}5&;ipAHJ-Jlz
z{wBGQ)S0}+qG}`xi(^QatOKZEeL}0|ZC;A{%H3cp6VSSh3d5)=D5)V!9fAl`buK!G
z(E|KCZ$3XJK>z)aTE--X;;v6Nlbq_*`hL&{+6Gvr0E@IzJ52V#Ff<Xu2Na#z1OOc0
zUz$+F+p#2pPlaY|60IUrTbSJIg_Tuq6@=3;4+_ef(5;M$qsa1e>yl4#G<CLoiFF9F
zihJcw<x?j1zO?PRRzxQFL=^aNR7r(h9dgX!e-6JjL&G#8r-+Bv{?M4_+fmi$N{@Pf
zF&h1jP7I`tN?w20OaGi=`u~Ou?TOvk{r8e#p;N@MUGwnzOh-f0o#SLGNlLKHl^p5M
z&IlMN7_VnEnV!YuAZWFX>hS$_qcd-%)Iwc7aioRT#s9cwaG~ADv1aj!@TYGsEch-2
z1YpQNVHg4U2%~RbM9f=N98#0WK20Q`Yv%SCc)1kGkXp{?&*lDui4@-LkzyKWJJV{E
z!6W**AzQ|nQfeZQE?*um45ea{e((B<I`Yyu-RenX2sy^mkV^h(g1pB2{pWz(wpgwW
z!{bdhHlV)FAtTxtkUVcuA$%F9u(yyG%h113W*<yRg_z7tqW4WOSTz+`%Ar|RTPF(T
z2j|xx>A7?U^NGTMmcq5|xMT!Uodk}ZL$78flatqJk@i)eu1^JpvV&9j3%hbm0<{R~
zm69Zg0S&mj?wz(VdAL^@_PkbgDa9Nm$`i!5$)}0@M4L-ju(!?XGh0myoYu=!e{4tb
z$hwE@|6=S?YnjeF+ZS-7k@j1Hk51oCCASGs&CBZXKR2Fr4(+~*GcGB?)QcI?v0(wT
z6R}aTCC@!wY$MoW6Zp2BtyG(~ZH?6-+9wf>Si)K);oXxlAW40U3DZ32NL%=){mt3X
zeRuM0A22h-0(w%2CTVSVRHv)Hm#TNun-EW4a<aJs^?f8LWH;$=&Rlj1ds}O?+T~&3
zNTk71s5_}WEuNH}q`Hrf*BM%)B<cpdAw9E*^9(BWJ3aoYRt^b?^D$N1k&jRP@Y5De
zDLndB;e^6+fv4mrc}(dZR9ks7@))W)5%P*+uyY~ve&Ja-FC@6qNVq|THd@b5uT91K
zxgj)eJSo5ewwehLP!-on8t16($0?NUG(>y7(rYGY45a^3cu&~FYfr*rPj6{QYe&^U
z#|bA)^Sz8T=*Rv)zKK@Xz5IxGr?8XyfC$i@CVL$fK(;3M-5$S)6-;ty5~ab$OQbrd
z#cKKCG(kF{T5@um=cI!BCIsZz-M~wkg3>p<!sY^jWsP<j@noDG%pM1u?FJ=J9{Dr2
zu3x)fHeEiwT5x)HM|5>iYi35mzeG4?b6y5#;V?eJa*L)E8NDe2o<6V2bRQ9tlo#C<
zQ#NZETJp;~NLrB7{bzuwlQ_tiyE%u;|5*cf%4q;6s<hiZv%5~+z}R!8VZjs}%D}C{
z%;WoOex;~~<7?cwhciD-X{^ymJ%croMMD~ITk0e{D4e4i7Q#JrlcmYR8smu_8MM;N
z4>a?eorRWBqW_~QR1~vP36re>0dMe91-|YVrHfKIHlSMu85r|(wHi18k0arEhg9*o
zRCNgyJCnY9SoU(0yX1A8Ktc?-%>L7}92<shnEzCp)Ky~jfu(kz{hSvod4s`4T4~);
z1N<(}3~2FIF2MUZmGHjv+o{D1oA-ts?`PhpHx!!&FTUY*Yl=#oHc=tOAjj%RJfGON
z1hajNP@MqKb?YBEm2j6?rA;P(H#XG5+gwn3v?Cf`S!+|><2UHDtLZJ<gu`gQy1>Qu
zQ%G)&Fc<?yM1bp}n)II81hBR1%a?~69{E&0^#T*WOZQxCQgVIRaV7AOZ3Iz!^&RyW
zn&q!9;=BXRtPvJcIcCa8r(mbu97DUkXS<gz{pKCPQY{HD@Sc7-*%ol94DF?Od$2#W
z(sxGv7ej%hWqr`45k{&5#`qW(5{*5DlCnI$USaZlEwVwvnIXczHU0Q1UF4R(X)ji4
z%C#)m<>@>#;Cm>ZA7ey%Yg;~E{P1+Q>F@+130(@&)J9IGnLiS$&Yw%e1(OlJ9!4SD
zoBYo`+A`8(FRJ`@8;D*cUcn-w)Hq#il^0ozclG11=I|LmMpORU?2tiy9pe%6I2Jy+
zPedyusQ(C}T(yGLOL{m;z$`sGY!rDbZ8K7a<BzDyGmPJWHl5Vh5y7<wccJE&Ad&X-
zAoOrrMw5DIxNmUWhzRHPEuk)U+<oZkk={`Z?x%%B*%Lcj*uh-479`ykbVEVOL(P$6
z{GZoVdQg?NFIC4Es}mgDgr^*VU=uk=t|}LYP4*8NSg|eVCi&@~#kWI-{(8T8c{o{m
zvVeHzF@UFHGy!EkPmO$x)b{Ar6fJq>_n8WnERuh<_^q&tu_4t_gB}1Qbq_VGvAPSd
z<dvep*++8vEtgqylXN_H^(EV7m4&|;Qj?1@cbVP=C6&424YMs$;eQy-R(+7a-IjR0
zi<Yp9i-IrCreB}Aag-F1)$PX>L<oZv_KH&drIax@W25_x^`70lp_Ym)Uu9J7d%0O8
z8z{dXjCqc0B28d_O7o&<EtDx+y65C;%%MT~R0={r*P=4JC=8P66_JhJlKAfGO1;Gy
zrXcQZmX|DJYAr|A)R9LKnJ<ICmx+2|&VgGKSCGd8qi9L>DYf6BJcHAuWOF<?TWUCD
zp&fn~ODNw?zucacW1)M9Gc2T&^1UrNy=T#K*`=VPWzBj?%_f0TXYac#{T-Ww(_nLh
zpKp@Kj%&`W-{_rQ3D*1$#KeVA)o{hL9-#h)#4Y)EmuI}upc6dV2Z4j|#&lq3wCdMe
za+?ZuCKzK=>u(+xFB)vhU?@AJ5x6gH9d;-sZc5{<coHQ8nN~3b&VfEb&t?Urg2ru~
zQ8gFH_-ds~6}`K}Oz&jdQuyRt;2vOWsIMi)yTWIA)#o1q404UI>j4BNnz_ZgWt!E%
z<9?eN3bkP@MwGwq^nQ3vDV$ru%mz!U@Wr_=Wc_?&&7ZV&E_zTY+uuy(442SItJY(*
z6*6^@nAES4FJ?}_KU%=)jYI39emOphFH-MH61Z&XQ$~h)c_|~_+flV0^2LD5Hyj0F
z9Wgx>({bW8SXzOX+HCd#S!3<j+bNzQ#Oq@MIqrrc*Hv{fP6erH7YN%$Lgm&&Z?6m7
zvJ>x$pm)lt8`|C-Dnst)jt>S|>+^gDPH?w}7gV!rE0YG3vv)qz1Q*nX-rGhlTeExO
zS7a$z1<uaSK_Wyo-I}kvp~6vSD-DsHxcu^V@+d@&Oo-<2My~}w$3K~_Tz#POfR&bE
zLTVS`LR}S|Xsui8DsTt3wb|y^{cIi5p|+5f7ih(y|5l9~{r(h$#;nc&9OhJ_P>?vU
zjJlLb{Vv1cG;XvsGs5(SAMiHY{FYnn6C%F$BF3557U(r6Eo%#k`BaaP=QXt$5Zmuh
zJhD9fDk;T@Y_&=~P6al_3UjjoTrRV@cHWoCDxk%C=wi9@X%t6$Ul}S+LpA)z^X^i4
zs7+b36%rpC&Qljb2(29d;RvBW6X9q_Nt$~yT9947>u1p*R#b7P%6WLILV}?Q!Mx?@
z7(frpicq;N=X9%n?B$%}%vfhxO?_Bni_*-blsMy9fz~!mOq#&$>-6SeuZ^pv`<G3}
zcBq4M!>K8^j0>9Ysg#-^Oq!S8k?ktkjEX#?rxkgats$wt^5k}|sWk(+l$9tKljg&T
z-8$_lB4c@U?c1%k=3S)9++~4)QoD&WMTDPd#Bxi^rMUOm%8<Tfc=*~pkJiwcaSD%h
zHPGzCMVA`yzI}0ifHu;qN@Lu0OLqJ%@A^GPvMo`G`8Es9Njee8*1)l3xv-eB$nh2}
zSa^_H44+p*=!P`c%%n0{y1BxCXbJ7DOnT5PfD<A`!-}NIsEEr@i`(KI(9(E;&C_MT
znd0W#1ZiP$LQnx54-MdWVUZe%FGvJG?Ks_DyUN&XW{48p0Qzl-o>}<nbbU<Wr~To~
zRF$*XbWFd?lwTDts_rxo7M;;oUvx$aXRG^Qml0f<>Ysl+no(A6oPX{ru~k&Vq&Q-2
z-tAEf<(X6m?{@QW)gWi78v|W{(#ncvb;$_TgtN&Sg0gNBcB-_*`|Xkon+3MlraPWy
z0Y_73NI!CziUG&Qqg@ZznX|r&Vc@*;3zoKa`V>PG7xBpA0=P>GIlk64)4Iv2DT-Y)
z?RW(I%*wmgaIaxMJ{dVX<x;|e7r;2{+kVe%zAh6aJ-pPK=0WoO_nQ@3|I9mtgeRSG
z%~Ex&RjytU)nPz_cmV7lm{K_TFbc}DeWT3N2pzSlrrqPI>@q`DPP><VVd0)~W$ESC
zC<w2SL557Qi}1c29u-85vLK_0^T3#D6Y~?-e}anGa?ea7xcy)CUr@#8S?OoBQP^&^
zWLHe~T~0Qh`~C^`VW%M7v|vF%*U(6^q~B_V?XNFB89YrqK*FlR;E9?_Sc3PtS0M3G
z8^IcwKo#;CgE%RKB06@)MX+eDX1IdGvVw?E1&MHHT^S+(TEw0L_0IL#X9<lg%n@X(
z()yE9iC02Xd1p7SN=x$8<(HZRv~8hG$^?VvsIH2Tm&y$QE_C0d8H!wJpO%}$IUsHs
z?`?X!{!oiyO?T<+c@$QBCnT7LoBl;{)vv0=Y2!Uc%b<%rmQcXM=+9TMD&IDDlh{)=
z+9G17nd#wywr4VY@pZncbA-(Crx{m_gxPq#{z8Qb{!k?CU>Tq4)snYQ7iA=OAm<K<
zQR>^Z%mI!{V4{D?zHh?u_3Znla_h;Njm64L!Raz?Wf{4JgL-ypHCs?!UZ;`QDS)rB
zugxWOlXh0lc?Nz__^odK83+oVn)|TQbWKY=m#EMG^{iZ45Ov*Qa7A(9&+9~DpaCi(
zFV%cK=a~oAmRbjEwA4uq9Rh}#WHsZH$7-Bcne149snJ1g8_Dvme6UsAXD-N|UT+lm
z?n~VM$QmTTNy1<pd==mR$)zT0uVrzR$AwQgx1c!aCsU1J$xw3$UK(<}?dKAoc6o>J
zXlW`Z(gZ{zi0KZ`1M5%9wK9GH`i=9O<n{Y$5Yd*4(Ch8<FZ=KWFI{}1Q#O(_c1Vr?
z$wkOk{0H;wr34W<P5+Hkss{r<T^|a~i9D5Dmix`>I7|{#;#dPlS`imVCKQ~X7vw;E
zMcZIOi4X*#W1la6FH($NFSedroa~0h7S0Q@{_}^sq&-37^UD`=36FsT3<%=U-IKOY
z*0>*>^-GPKLp~ICSY3o>%P|lU*cnf#vSmSyUdn(&(*&H~1^K7;&#UV2{}U0Wt-8NU
zif(SZ#+9uTJhVi^wg=`k4z755{6O_e-+!%N_mO;HQl~dNCSb(38_ilq_ntG*76JfR
zFCwq*88`HX-OJyK@!nD;X$^A|$gm~-6Mg}D6m%3*%QBPe3cnL#$#z`-z3I<hp$OIq
zOA3)z(2%y>wc|62{`z|-0Q$L3cXet??{$yM=_U`4g2rBdTLpnTh+72`qU8Pn>Ae5+
z7X$36K*C>&HpNL;Ik(<T`HQh3@<2CCCK^&=@hBhq3;G|Y>|V>P-NHj}i2q_Jc^Jq!
ziZfM$(sZh*=XbQ`ejQ46DO`|=P8q>CfxIkSOdkmb|E1uORzr7T)AL6>sUC7hmCp<|
zDlW)wJcvIY8@}fVZb`22LTmi$6%^)_R+?R%u7)-LVmy&PvcBRS4Y@&Ey{tt&>?gfd
z-}qmh<@@Mzm%7y%RQDI-xcl)^+G7cMl$!HM8@fFB;`zGp!?34ChjHk&6}ovM;Q#gv
zo8+gx?1`Ssl;-*mG@kxzj#+vojupmvmTiPy5V&^|d5HMY<)C1@PwC6w0S7FF#wv33
za$e{@fEai`2!@}&5;XiAP4GWxQ0#%<-Y?wG77EvA@FCToqy#0YSu)iNR^}#mhF=dq
zkwKhQ(rIS9mx*#iJN5WVNE+WZWA>TKMNFV~IzO@WxcnTuea?)y+mK>`+-bd;CXM>3
z6NTC{RVD-&9A<YxgHna6?pWN)ANjJ0JtN!4Y7XC;SI)dMQ8YeyFstJv#{(-HXg!la
zo6~y&?@|_;KlfXEGE_dQ)9%C3AO>F<Z=e;l*B*zBB>C2QjSW9s%dKDNnStQ*iw`%*
z&d@(a=fu_f4}cukhetu4O(uzf^&4*)`_sm1ulKJ{>MB?1EB6q*NfmcHfmUqJP6e1t
z`(r<qa&_+mkU$2)<e;Bs{#{eyN4<N$AT3z8Ay5lVjf=&=#QxYI?j#F}^_@++kv(sL
zqbK30@lW0AM&N0kdd!&#64=vr=fN>%WjZ$^oxxYhI$~whhbH5!3DvfBEJ_uyTy|!*
z2+Ax{`Li6_;|k^XZH@G48+32V+6rE~Ecywjku^+W>Bc8uV1&ucehNd=0N@5E-G|2O
z$eR23OUMyp@KwTHdBX?mC0S)~gbW57)54+mvgOe=Z-z2^JYuL;w`oj&sKT^Mkp6NO
zURL6u774PKgL)){yZ{SeHO8!-pD=W@405-v>-oe5Nb3_|HY5xhXh0DMo!_hgC&YNy
zP!5zbhx4m}8j=l0KNirN5+}JB2ne4M8@bJYg4J#vWl%j<a*A^J^|n%M$-@xC3&?G(
zud7-4_~kBrZRDMp8(<^e*t@YRV!#C!K6%s@mOLG-RKE!MN%kDs_SC4ysD~By&Sc%E
zg5J#%LVpPZ-%ihY7L^O@Efs~9@rj1enGg?qX2xZZ07troj7)HuS#rB{Vds0D?-9@;
z!?(DeS?|$~9lb)}y~|Q*y>pWykMCXPrMIC}_Wv@rwz^0+#fxPu87T8$QXTSW%VCiv
zQY5X&Yxq_&`x?$pT!2+0;Y}C(`DxSfLu=#ClOxgbtrJS#?|Zb1vbjW`K`K`aV7XcH
zz(4o7*Ac(><Jj%|`$C1H^W8<#`{vbT*?YK4m{8FwkjdmF>{ImUl)g8XOe~|jhd%f3
z2@YkhpZsn{@&axX_7Owb(x9>iugG96)92nUX3*AOZ}_m?DqCGWM!)P5J?+3rgeph1
z$j+*8-V-cI)H16leR|KvE7L}S0h2Cvc{=lT6FKgr?D%?0HX3S*IjR8Hw}OD=JA-Gn
z@owpM+^d<LSILH{lgPTM-ift^m;hLOMM9Dn!Afz`v`i^$|BQC>Ci1c6#(>kd&|Fhx
z3iC-xuFlz<{@n=0GM!7zlDD6RT5`ATOT)Do_1l96xM4-!xqf6sj&_xDo}dc4{VCwR
z__9>@LZAMB<L#}3;`-Ws&klqj2@os<Z8SIpg1aR^6Wl#G1RBz~HH0L%L*pJ?8h2?3
z!QC4txI?hU`Axrb&VBEhnVPzFtIo`SP<vN3^scq`TI=~fpHIJW|8pUew#K8_BGA%=
z7$dT2<#Ar8_0NQPWD;!$&+zmHg)&F2!^=hROVG^hI`N*Gp?ZOH)y)8iPTTQ%3g0(&
zlP{U%G90Z+Udzl`mvNYvozC$*=qRYHmG#cuFtqz4l`7&Gb>ARDCzuR0llzrl&Q<5m
zf-K6%zl~+?qLU<!JqhRKz7awAJL<oR?UyMQIE0S}2`0zfuHmuW6Q8_f7*+n0REzS)
zd%e}%X2TKN<Y?Nbi+|@NHA#Y;R#3;5hMg#wu_X_AB}E$?wpLid>9*eDbaFD|-s0z%
zUC2B(PF`Fj_XlbReATIrTNl!C|69SGOgt$2A$#9;LBfoC`R=16Ne}t;Na4YncY{qb
z&}pJBQr6H~>BFpUVf=c$HWgFWs&jelIu<A-n)i~!Wxryf>W;c_sB&J-D(QRFvq4RT
zIf;**r`az=z774U5fbsxW6!`>u$!p3<Dzd!H8Q811TafKxrlMiWoil}F&78SvLkrN
zZBuN~aS0l0A&pQd*R!NEZCzY-H9cvK{oR$ucgcPVANB?lt>7#toLi3T7e5}bDNq-I
zk0q$S0MoO(m#jBDq#G_}+t~)Kc6muPLxRMyafiNO7{>QgiL2EuNlQPg10mt%!w$yw
z)cU|z;~_xnI^SLIs#Wx~au_S-r#iZ|O{cKr+~ZBaVo_kMCpcW&S1NV#E|iNezd+)>
zDY8Wid31Q|>~qw7l>KTxAknhcJIeqzR2^1zR1OuIx26v&CtB{#XZwU9jeWSXbxWsp
z(-@;OSaTh@_2OuW)`*xE93ij>UT)ZZV$Z72h{{t~<K_3``jAXr@I$0LDo{^6<EG}X
zj5|{Hc2FnhS(wCFdE8GdDWMH&J^EqW%1smKY;8*6DG!qJ30Hx#x~5vCF7m$1;;kgG
zI%+dT9Hp0PQ=9ZQs%1p?h(5m|DQWdXD7u*{mtBlSR`C=gVRMxd$n-AK+Td=kp;+oN
z+deoQ(wITH%OU#<Z*PL`r9yyGUhJvF;R;!4YL-K4dwhX6#S4G6dGkj57Fb*ky5&xZ
zE;EQQGLGS?W917*daXDfu>ltm#|BT}l~MANeZym=nbi-$1%TqQhiCinkiSroR?@n+
z#PeDEr6RTCJ9|f#Ji30>SQ;RIEDmGG?@42ZXWZG8eG{7$)OW03O#ADG1WApVnYrBE
z>?OaSj;*C+(|>v87Z}qw?aAi*s)alqF%mA1t+|fz%}r7C82*@FFOPFhv)1w<n}Iq)
zC~lMd244fTy!Zs4?FG|Im>>)Q+l}`;eFxO~tp`cw*k0W-FqMnzwJe{=GyMy05PgM|
zRrdifHc{!KeEsr^UW#sPn6iur58Yd2y(!|%vreXT5RA;zc_tyyDY<TF7$9$Ou{$B<
zl%E%T>K>j^6T&cZ)|TacXyhyq<21zCco^c#TB4}PO@og=bo@iQa!1&gVpOBg_1Mt8
z=!<ey!-66f_Z|F#HgBI_43&twmy$tX$%VYg%ri}rp?EsSs%8jF$2yLe5U8#Eg?_ag
z@A$nb66<P=u63FrFPnEdGI9?k(1X_2aXP+qU#$jGm=&!1+IIv9By4*}%-@mvpH#*x
zf2=80aSzG{<Ep6+ew`Ovoo5mbe82I?>ygat5P!z49J`q+I?P@F-oU=lusrBd3Pc4U
zV+G(!;fCe?DTQT~$e$UvE}15-w!Q8Yra}?_F)-BBtEd-D%*y%Z&WsZn(cF7_ReODS
z-v4E#x9(hLnwp-T`{MFQGUOEdPYNO5y0fjH$pCNDV8vpyP5h#DQt)&=vDg!C2!`mo
z{~K5G<i725<MekK)zQT*bK8}^pEQ-jo<Jv`7+#3_B(7@NSH4^)jC_6ike*Ca*tzI2
zDGLO5?m;+|XaYf@4{W}nL<t#`t0cEOoI_U4*uo>cO4|1PNrs>W*?v46Ss$%wH8st!
zl^^5l2)cgW?j|VAhWoTaBebV5Vr)j)8LJ_0w$NBM5cE|KQYqKWEsyugtf%3fa3qM+
zW$D+L1!u*w@v5BZ8EKVl!Z^vex~BV7rObh(wV$4)K+oR#iOb>d8$)kWYyF|7q#-yW
z-aX$ThYmwpwVr1JLwM7aso$kAL5_FS^;I=N`1Tf=o~0VRB4Il52N+SnV#TtqTF+$l
zH5PC|awrW7NvgjIgw2mXJipUc_AB;wB=bTWDxLrygO+XsGUX>quLRDrwoSjE>vvqD
zjRLvYMoS^y@Va*cE?9P+MJv1eI1<bzbO!;F8yB0SKj4bvs2TVzh4Ro1=CSheS`2N)
zf%akPknUm?iMik_{zhy~?X2BBMJ$OHnWgiGkXIvmJjx4x209mY0S6PT!wXNiS!+Kr
z=S|igWzyZ+-4~Is)FT`o*ytYjK~r+)?r{1`Np&-@A}D+kUclU;HB~z^Q=?ihoOY&O
z9;Lf}5v!a&xdgTcJ<oLjd8|C9c_b36bw>H9ss6=NtkTiZ!}5JoqcdLgT_J~KQ)jQE
z=F7u@(RRmVQnXZ7s{)lqFs}!#M|ko$RTFVne8J^o=!DX9b;gD}51Llo)RJL7;XjS+
z5F0If)EMGC!XE16Ml>C(G3V}9X~us$GQD>;LtdDgvc(W$dZ4SVxvJZ_3K7|=O|s6|
z1kZe~1nNAkt<O5z!_(WKf5L`c1ieeBJ-bL;ZK(|SE^<E!O%|Gvs(XkKumD46eD5j7
zwqyR|%EVzaBJ82)F#Z4aia{BPwbx=)uCk8%k9^0h<sY+Y2R)`z)E`}Ch=|c0)&JKR
z7smhJzcMQi?s5Dv;1BFLNBftItc8Zj7kj-9cp#f4Dh-rOEXdYk;AxsWk)Mvjn#6D>
z>7qt-z%tF>uP-OV*i-}WiD|Y=_O=X&&OHACz?9g;PpI>94F_1JTI3^cbf&Gm@ejZS
zN^%vBSU%C6sytQ}X?(+DYRa?sDp?t6J`Ab<0C?o9HHaiIyKCDV?w4cMSS}++>^|~P
z{sR!QyjFU!eCN7^QrlvE@ZBOzi&Co=6;xRjpv<=5G&$EOK{50!T=hK$_nm?s?A4N=
zqlFEEQsH_BAh)yCmv@6AYF>9lBx8h=yZ@`$-hXRC0_XqnX)yE&|C&~Rn|E>#{K}!Q
z%Mh!#lU+4aw$-_@%lHjpx^SUa&rCjsv_j?IDX#atIvT|D5q)YxIwpUW*%X@I-2^#r
zQXvl%Yhi1>n6V~31y^QdNhHT&6$KvHh>jOWcxOq1nwqO?6PeCuMnNm?DTehme=i>K
zY;4NZVrY2zwO;NX?A8;FmDUoGU=7-OA#YlNU1|pu=_#rLk4}Am(N$ZUotDGwL5<t1
zL7WG5sTUuUsRpKh+ZJ0nO^_eVdI`65A1ottPEh-{Gw(=;yG=EREX7SmTrMVDvZ>oP
zo>PHf#L>CuPvmK!wb)6&t*V4{lsP{1aIm2_^ljhb6^8x#7MNQ5y?K(Fla$(GXuEeH
z!s9>E4jW=&S<BNu-Dhh5N)(!nzF2HuM+=TdwO(v5LJL<Gq&2nkIm$h*iDTV`ukRE%
zt$t=ytCXfYW>$}R9o);Sjw3t%0mxZi7x>o|DM$SSxTOO2a9kFMdYgqpx2TKSprK(q
zA^+-H`41Hp+@+EUB1vYRHh(|k!-I%#SX9@9X%aN%e@9!22(SE38&7_A+>i&9u>|`t
z>^)<G-XX3RuO`gyBRXbcmalqO5#i{A5|hXBEQ(zdy6%_`&s8e_TaJGKah|u;|K;B3
z|Lx1QY<=R@T{liTugkgJ)01_Bgp)KECP7;JzWKP+@5WmeiHM2+Q>XRs4r%%SuYUT7
zE&jM~c8~K9faaal+YDs>X?mt1a(v}MiiYp{9x_r(9-Z|i1g2H-$H%Y8xGkY_ECUf7
z(k9jO@o@dSGE-6hK|^1!d-@=N0Gc{6cNKCSmPzBb61u(Tu(KKcxlLZxD1P*`^4vq}
zNA~PHp|=_T9`$4H^HQ+9%M%!}?BDk<JmmT>sQL>Ce8EizIQny1B~-wC!2nH#+N{Jy
z{R0)_!n?#nzMO_-$2~>H^%sNF78y;YeI)iNAp5gdT0?r?GfV#Tkt06#u0Ma;=LGAn
z{`%}{IDaW7(f-D4MEgwJ4_!b>tZ_24nZg)aHZf7&Ge!AJ+$pZ~Wj0@?y-BM%$|1c-
zPOpHq1fUSue{oG4N4;qtMWaP*nHsC@=n4{ZnGQLMY2of9wdWar(Cn=16ExXo*lAU&
z^fk>f%gASF=0*Jp+E_dH$Yus-o7@Qa&sj<}OK%<Mt-y8yv|#XF`&V)X4GCsk<+^bN
zZWx<2#5m1<9#ot8>^vSYY&L*-^MKtQl62o3b6UZXOS@b?1IsR*UPT)qAU+q;kFIlm
zihJ9(?_xJalqrnii3_dPP$$t2O@Fyf>>SLhq|kF1MaHaL>=vgrUqkZet;EEp{YREf
z>`p>y>`!m9elmPcRU#s8j13KK3mzsR`F1p0Tw~&fx+V;hVtXrT(lYZ(xgwMB>vh6e
z^6G#B-Lc&8k?`6$Dd*FCA0|6R(4)-K_pLrc=Y^s_ac`ssuph4t{{eiezU?v<bPTdy
zu?tO`eUqdo68?)Xxn`bSAC=^z*{W8MUwj{^L6u(Uo=4ud>;3&KW~<!w*~l$5B-<uQ
zq9=Ba3U{B5cB5J2CK3Nu`Y`B4ka&T%QLcT2`_gep$`>xY`==i7vZQH21Hgd;ny~J_
z9<RIBT~k?1o8ua$l2ZHPU-H8U{=5`I!+XNrQ6kRC&5cy+_q3xt^ytvKoJg$4&vG3l
zvht!&LU~o{{;~TvFF&cBQ)F$<nev#VQ?wO`yYh%qfIUobo*P^!uJ%f6w*~HqWps#i
zrs|Qp@<e?yF*yM%f$>Sdy^OoF0L6<%FJGOftSD`j`Hgig>BVio^)gB}-o(kj<wSA|
z)LlaPnX-Q>BD?QDc|^F9^flWMCKfJKrM}^Lldxa*slH=;S*=vHul!*3nO-c-eW8f8
z+ZHd&ycS-6f&G^T^i)^CS%j)^BfXMnY*t(I%j<6wi#i{b>meX}?Pvo=!XrXCytl}H
zT;w~;(7ja4Jz2nZ`Dxx1Lw~=&E*yDl@FFAX4cu2AC1Z%u;)x<T5IyM^e;B?_vPeMP
z=wT1%4z0ImaI<kh90ir;ub{Y;6DRu$mMhqH7VbMDTZHV^B1n;qCB5Tk`={Lzl9vrJ
z`gV*^Oa!ce=893y%+cgNr<q%>r%i}HC{IQ_FJVPjL~<Y9m`4xA(^q|DDyo)ffVyLR
zQRDno8P5?}#Pn-8H>SKaYpe<VZdd;6Nn**8)x8GoXy=33`L^!f4(&QrnzrmSx>ia`
zQC9;K@q?RF`$Q5Q0Quy_7?j>s`4{DwOC+n7a+9w9+A38$5~q`nf0oDiOsmMF?Z6(v
zi?^76JG17tUQ_X}w)8r`J~F($%6^~E*3h^&o#Q5@3m)&E9Mn}QeRZ%{0sqnRJFuI>
zw1qBCbZ4yED=i?Qh@&>@kFKz0c2$>2(I<PSi<VW;15`2&^Dl+j>6oZ%*K1lxuhSgn
zT+XcR3*FpXhHNl^$?pzcTY0I2S(n&MK)Dq`c3@q7(ymW@2cDCK^<&EyW7ZuUrIYgU
z*VJn{1D)Cj515!|sR`&}-P~jtXWdDDi(9X7jox4xDMN-lyeum9xjoe(oAh0T7a{ut
z5oc(WO7lmxyZ4nxi5TYp{Yw<wAKOtOO(U}dn8l3EkCZk}x@|E&4MIVi@Ffjp{AYuq
zRLI>h%$xq?zKx($uZ^^~?mrg>Nhz_UUaHoME^cndjkz+K-J3tMTb<tJ{~?{7up&EY
z<|^wLT4OlY>HQ*J%ab_zuo=AQ>%_irl3NurI+^qwTN$%}-o)yl>6&7Ec5Hxpna1;*
zNFM&{iByEaud4e1D*M~EvV(?H9QPLlQCv^!5}9D-4_r-(0HS<~IBJscl^zioe9*LX
zh}C?iZhz6Jlz@p#<;P`A{@0w$Th;>8tb~-Ux_>Wl(o)K~KG;e4z<tdOt&3!hd>g0K
zTnH^bT=?5qv;2<*575FdY(n`^Ee`Wg{s!~gKY$1PnDxZl$xLpce*n9V0i>0)u)QsR
z`YV`A)+vh7C<+*t<>jA@ejn%Dgbdb;8}!A2!6B7PnrVfTcXZ|WiJbSuf?__|Y7OZ2
z4xz2~hZS`H$C2|tfY8*>mQK^47b7mo&95|j^*G{H${KWLH_G%+P=v9xUkSSGp=)~g
z%$0k%@?rKVurHNnUfTZva2#%Bp?6)C*UDy-#9_pZ-zM9kYg+g8=-aMN@H*DG0u(_^
z@Bh^Om%DEt%>E{I<TDdPBYxSNtPE<KC$=6Jm}5?`*l5>KVAl|mQ;P!Y)qXK}TRtX@
z{ZYG~G9E~sujVC;ce;<2Iti>qJqMOEF}^wZ*Q8GIACo#8GkQjcDKQX0{tRb6x*%MC
zH5+AE3Yk~Ubc|!d^(ir}{Q7a{=)ULpQt))+d_#t}MVDi+6%E++8$$aMy{t~7I!rFg
znVuG)xvmmSF>64j$buXG7ao`)>j!6CWlxfH4S$!FEeyicL7K6ZIs9KGr9?#J@o0H^
zoi^G^`b7o?a(mA%h!nL^id>`@PzLzby_yQ4!3T#khnm$Wh&cb5E@;=r|56Z9>{fS0
zLn8r=S?ribqnyLxUcLf)!>Z@^*le%&ZwSprL^(>S1*x3#NtxwQcsMkd)hqTJZ8#6K
zPez1u$N1}AjLI6)F-kBxG~<0hK8WP?+e4oLBbKB-b9VM^%?Cq)cxfI3cdi^@vSujB
ztQd;|j|Mp}lEb8YXI;W%x+%yX3qvL_bbU$hJmC9=SL<w4Ov(6rGRYqo?t3)M(vAlg
z!#_xFIBbKTRPcL`$OIR8nU+sNlnA(dFWt>e<i9Fn`kY)pJ@k`~f=DuWQVfeA=|tRw
zvviE#CdoaeT$L`ex<vEAViRqp-ZX|p{-w5#R=M;W1Ga758VDZH#ujjv<>!_8+?(EA
zqD8>2Y4e<!)B@zrn{Z^2M(pz##2;Do%W+>K#D(sC21(b=vBajeE8_idbO6HxmNt~V
znDU8yl$+fy)6+n>&*5-LlAlWA&J98&z3WW=Q{|tejVR%ktjFn!Od7H$i$xn922f{-
zNMFHY{8x**N@;s_TM$HkT>)u<85c3mgHi{^n<20CRF`a^tyd-MPLy?4e@TKVa=FA#
z7e`++5QJ8b)JaOz^WLAFITehY8>=y)G+sms*sv|5O|yu$DfyS!lebI{^rwgZ1N_6U
zY%?JUiPYR516yUY16GP;Wi*M+wUY#2MW3SP$CvQR5R67yN@M3ww~XWr9Ho|(iaD;V
zE(#=|CWef=w{||a!$qNf{yULg0+M{kG89-y$V~F&gd+#s+HjBQ)_I2;g05fw%h|x2
zBqB&zsM?E`PqhkhXx;tVUeZ%*G%GGzW<udEs6p@Ptkxmgji#5n;2H161H_M(DR?m(
zS{7n19iuL@vcviBwfRK1Eu?wd%$EY(s)DTA($kK5z&6qYO?sYU*WE;oS@NNrXL7-f
zD}7Rz4vMO%p-g%cmH?WUzJ7n4yU{EBaaV~*2f<N%PbyU<6q$VsnmAz3j;o0<d>V-9
zk3J}{TO744-eR>VUj&ym1Ha5zos^t_>6MXxo>H`<|Ba>ZaqC_cY)chR=lpiHIgObT
zfh^MK-d+}I9E#T|I<AtE(L6WPZRu=XzUUZ3#mk)7r~{wb2Vb9$Lk`NtpmC2xfXvNz
z^t>o7uVQrb`E{*<PzACSES`3Voq=;@qW6=eLx+a8>)G`Gxy6#9!;c35U;h_eR#@<$
zE4M_jEyLw~`v<K^uIG7`B|7b0m{5Bv#xhHj*5ToQeNlZYV;j89AN{$Pd*@r@x!>fm
zV>^?wlo%dW{O<5B^{%AeFNAH|H(xG>=Ni``s<wWdl(ZN5sPBnkPpQAtAbLVVld{>Q
zDxNfxGHh90@@4PYGE4w@EgHHs@4%>XD|@ax=aQ??1XK>=Lhr*1X4)N&0;@@YVVhpY
zx5m1RTrvH@(Mg~6<=U|~eUzp*iUz-kP=8NZ@NTnCA91dz6^w1vs5Hr^iIN_apB_YW
zFgbnasS6e^?5jjG{C@}HNZ%=?EqNw>x&aHAbZ(Vc9Tb@0xs>Iav)|D80Doh$k$mg9
zO{6w26HKuldXAb`j((b~2fxS%YUQYg405iPsPt0R+6R-H`%ke!I^*{*5wP$3ZiMET
zSu9w-ms~A(I!ABM6O)hHf|u~PIE?nxX+soz-Wm2W{MME`s7~suNgI7;%55Mkr_5;1
zK7aUA;t5**M>4nqZh)OF$r^!t!UvwvIry#)LNq$GSae?PE?6vk3XPu=VC5Q$9_F}x
zavju5<NZ8*JtiQ-4Ua!Fd}24-n<aDoi0hsh@W#kL8GBH_bivpg*SUb53j0DnMyU}?
z+`vAjS8S#fO!9UgoK3iFC%Ch!27k^(kW8%FSA3Bz^+gWC?q~IYBnaCFfnxRw72(^i
zpz@**G!k?tG6@$QS>XBrzB(+1THFOelF1J-dt*kr5j1dqh5HA6t>OyL8P*A@gi`IO
z$X0W+?jjTmW=F!%NFH1$7S-;f8LO~Y)Yi^B(EPHELqaCu*x7_tbo|CMoH|(yGW4s|
zy(M*CO`VRg>MHI5416zjeyN9?nbm0Z7xaCl%w^?R+o7KI759Os!G#vrjtGtTJSxvt
z&x-E5D7cE%8#1g}Y#>sp(OAv7VmYEpNo=%6{9@f#bIM7$dzTlY63!YL+Q^pOYP>Tr
zqkrGsgdR03S6v~$cm2Dsy-Z+-)JY%X4sv4lhv=NH4D@#KE)Uzo_8QMhmDEvgVl>|c
z?*yf8-d!nF4$K%Ryl}g8=8BsGVr1`++5I^wp`!3hH7ilvn*&PqB=H8kS-2B0k<SF?
zAy$h}stV4VE51u8&BU0ze1SRt{IDslshjK;C={x5+lC_<AV0b}agLdj8Sue7pl__M
zNYpizcy2=W;u|T>+(Y@OL|oefM$)2O(R4`-+3mT1`n=IxX+_<_b89r%OwB4PYo#X9
zelMzaRmv5KM6-3CKbNszT9J{FQ!t^FdU&UQJsR9wlN@v5`wxJ>dC{a-UAsXery>jP
zWS5>)fAJcua7`6n_nPmhBK<t|E)Ti6AJ+z4lO#*r@T3{FFd9On2eJAL!bo8kEV~cH
z&l)?X;-%0{<Y=O_l=p>H;m&2D)EU$r=~Xa!ZZC^ytN0~)G@i;jMzrvHudk#coMD0l
z<yJK#ph3bU|532ZJ^gI;@#HiB)%Xvf`{Us-dXd%f@Gp7Ce=f2-f11eu&#1<;A>c()
z07ihRgG#W}9_QoJ$2MewSnf-u)-&|eLp<x4<MgkJjZE)g0^a?xefZXQfRmr>{gXk5
z=|9hnLeJ**#aqwN_ORHuBMZWj7*3M~=Ja6l46g#G3)=3lfRjBw@p1o8yy3t8is~;h
zKWe4CVDGGW&IxoZ>4R41C%QENXaXf`B}T*@!nTvytTOB4d$x)T6f5I<^*Qyyc$XjU
zz9pP7UJj3Y{`L|R|FQ<(JV&kE?z^xKbU>AxW?)gw-NL?1fs4`(9r<`3o*CzoFOULV
zR_{JqA8&lY9ocX((~<)za%fo-&_Ri16RP~RqP*4E>j#BTfUYU)pyv1F;N+=r;9cVa
zj^=l{vb{r2fF3Hl#)RYA`B~%#|EQ&4OJ;}Iw;ZPZ%Qe(Bg`Jb?2^Fxw+8pd&Taz_m
zr0X^3dPePKIgM@U>=VnJ-0lMQead6Ez)D3T2Jx6NC-W~}3-gY?-bLOz6=D{5u&#7(
zd}bfdMeOQBf<T(S{?sejxunk5jTQNgo3e{{4ONNu#wFhS_&=H!hmdmRS2ApDq7Nde
z+ba~Wkb$|ucl}eI3GHe9`k%+a`vudSi47Iq_E6E;;cDJBM?5cdLz2^@CEc-Kg9`<f
zF2<3>QQWxbWey1(fZg3$bZ)A@^XyE4f721E5@H9zt7uq=jxTty?#z5hXAH2KTk{*s
zs7%l-f*w|w(YPcW8&x|4N^#IWJksbOqKom0y@~llg1Z4smW}dPw!E-`O-hf!nx^h`
z$Ete&;6e;o#cw0Lg&634VG-r&PvNa#<ju^Wi@?0K&U}f_Ur5sUhjt7Z4a_Qb7WKCX
z!U0!Ft=@eXgi1CX92w<$IBpSd_~N4c>zclpW+=9$Ar)oHD@b$qqlUqv;wFH96h8th
zL8*53m-}dvXk&E&K}CmFyOMG!<g+nHw8qBujCc#oh^wwBak@q{?AX^&DT`uV7$pKl
zXeUZKFXk^}6KkVeqo1==QeG~^^geaa8!(PqCHq1kTjrP%?Z5>pKuW;&z)kz{RV2Id
z(lp(xcZ;ba+aInU#S$4F-_($Mau{!<f$Zwq?>K$G9;tF55iyEObMPCf)0Wv{E)ggB
z7O$a!4msM^MYu*MqUL3k4@rl~E(!c&uuwNCggR~8KZbepa;e>IwO)r@^ELGMUriqm
zEjU1E?g>Wp2^taM$}!A`w+TM`NKTfsGYdLB(o;_EU(g4|D$xV-%kPEYDv#JB-aT1b
zL2&3&gC#J&Vy~_BHI{}0^{z7<t5M47n@1Ff?f40zASC))?0KI<hm6RmLI8<1VeY-j
zV&LGIZ_Z^R6B9AKIP=8!gwM~c7kuiFxa<Hsi<tQ=SXegE>LvW!o>BTGF<zY*!B2M#
z(T%0-!5e4tj*<Bx-nd$8=F+*rlx{Z1hIE?eeo*>j;g^ZA(VeJxp$HM2O=IXCFk^p{
z{qta&R>cGvuw0tTdbYQ&%5zn2T1}_TUXG^gGlbK|NFcgWrR(6;{`_uLA~`8ye>QX>
z`WGd`C&%lwnP;adhqSt92G!}RmO=@&j_DpCl6lO0fNUs8@eo7i@RN;r<6upjQMA@w
zl>5TKmWw+?CmTv*XRanPbxeBiCbXcS%G-Z^73Uo|w#{w}*86kdQ=g#3h1Edw<wwh^
zTk$0p^oN7xtn&R>h^5FZ?y#WVsxrGFja>Z_SIbbSo~CXP;d=~bU$V6}LBUbogL&g)
z5Xy|Q&3Z@=epkU&;MZ!;MDSbA%;t~QyL)89`t=TO^HoBdXUyWSx`(B9Y7m&A9bK(u
zg|yojx@mi0_j}N6n=GMgINKS6{{YY@n59YV6+m*;sbnn~K}p2<o@TtX#-!1tXYOi8
z`IwEkk}_9BUbu6d%SR~Q+@?haT=kDEXBF&XuTxV&g9=Md{=4uw?dvYHX2WeEw#<!_
zckkOQX2O~e?=EBcfAD-dk*iXlxcg-OK`+~tTP?fW{DzB~Sa{L>K)7yW@*e;a4v$@j
z$&QWNIRdZ)qPNAuy(4%(j}Z;;z<T_E^j!uJJLONz=`-Q58nT1QhRL7AYF&o4w6A|#
zRdOGRk1_m~uqe|$rgmRnMH+%7*seSXrb!mds*$~57c;_BC$-;@d)Xe({q@s9uug)(
zXkz0ep8K?nL>G*3%;u?y+F-^8UpcqL-Bq>IN-vM-`P@AVGD}$%kw0rihtEupi>Eo}
z+Ws+b>7@VWqu$fL+}Z0#2O*y=4rjx@+lEPP%`-zYxA?eel6J@Y{4lh<J#uulZ*$*@
zFNHHnvSWs#G2Dy)XYVf*nuMbHgbEPd>ucNh`a+`5(O|&(WPamB!lFGQ3ED?vRKlj$
zq<GAqmm^Ry&L}A}CQ{~?%+<TQea7Y)qgB38KDp65-@j<|-dgp?A^-O|x1tUg&jsta
zVUbn7(bGqv4~xd2y9}sEd3;tr5W<~|X&xICH<<-bIGvbobOFAa7;(XF#^PM@#f+;N
zd|h|UKJI8B>p3->$ShZkA+0`v0of5(ju?>&Z^;;=>)WiZN91Qji4_Q%ecFYI))4CU
zuUxiY1%{zP!w_oaw~-N%t={6_nK-#%u!bUu^Jz`OFPSIO_sqC=;zRhJL&h=DXLw%w
zO@3)drDd@1gAH|hU=RqVHld~nKx+an6U?yxTz3H1l0$x%PL!4r%>M(J7Bx20937jK
z7~8+Pvq`(R4|1g$+$HDsH7@C+ZiD`+OKhN$plS60(xCs0jkZO$)H7w_0(hMzaD(hD
z_iJEX+eQmZ8Pf2mM5%H!8j`%>C$~>ws8eHrknw$e6ta(%%OzO!O-&l<_5-~!qjTQi
znO`L`Kj+!t5`c|bz{h{`X6yGZqKy5Ek-JVbG*MB9cE@j5fE!Kh)c#RIJi)>bJ1(SM
zD|pO(qNrDjw{m-#e*SZ$5vCuv?zM17sy!>Ghi$$zv+LiAU~@dvtb`ZPutRW<E<|IB
zDvGGIZ*(z7_#e@QD1mclp<W&_!$CH@1Jw*iSP300mQ)OwTpq`l#lF!TH!$V>%2eTv
z7J;!^b0yd@hq+!*dO>jy8+sR3@Z;|>BVmRJ@?1QGD?x_^vR@t7e8O7O?tv+d@6dp-
z^$n!eIwOyTWO4MKi)Z=NA7n%g!ir}sU*XxaA#twU5~lZCWF#mt%cNDhZxm7?ek^a>
zp>5HoN$*}q!1yRQMXUD(@ot}UsD4#{S|A<eF?VMb*@4CnHes0(>^8K4KP<`-gtR#F
z=W~=0Fn_}5r2P{pyzhm`UiLP&aggX2!8a}lC~aF&T{To+y$xhj1a-__9@44D#caPu
zQ_;PSjML+^N3jpAOdJ*qIZBdM+}zP3KQ@}FJoxR&8XtmX+c<)^pQi8Z@VwqNFxfaD
zdcLQy$qs9e`z^XzQFM+bY@x9om72j4^yS*GSL*GtbKK_fMyf}>Ezb(K??*Tm761Sk
z7n?h2NAqx31l9+s10-iXdYvWG+YMXxbxbwKJ@-dgzvW~yIcDM~)Gj>Zf$expI`<#_
ztq@7I#h5YHA2ezy4XZlHbb2+uK%6&;J;9XS$pjg46)#JHl+!o{gTv>k6@(mct+&~j
zRSdid{6ju5D>N#Kv0kujmGHWpP;T6qnihg7#Cxf@bP{z{mP}&odusA(78-CgJ(oNV
zZSez0APX4*B;PF$is2ChLP>H#$3<o}iDd3AHG&mC(MQK?nB?0iApFmY;z*g4VbzK-
zGxDKowWUg{-YxpKE#gOyMXm8_n9GVl*qH5}P01OhAc*R7=%)#2lHB*4Z=hHO=Ymz}
zX4H^Z54Uaa$?1J~mWuMxoobTKBqH<s++%*lsWXxJH3HH$w&LAZA3igq8SmmTw3|Oz
zUnhujR0WuyAD_m{G-GyA@Kfu&A%@qu7-ZD5CsU(HCL5N#74>m1u%$hT8wdOfhE7~Q
zjtXH9E)*qJW~(wJ;YM0;vMt;x>U^sqgw(TM0o>5lfkj}#6;<Wu@TsItP*GG{`UUSG
zS{+VJ;J$~0%oV#_2^5{qH9c(<uZ$n#Ac`5)7%>}?aH%j%g1p#JUs!(1=6;0^{cJGh
zC(>LYbBw^Qv`R5&Y1JZ+(#M#u%5)Zba)&MS^f<o-&L&u*q7xmTR97rVj@{V4vwTrO
z?NG1r+aYecw4+>4xt&$rbkHYcLeFWCrhB_}J<U6EN`$k*R|*n*)aYliRF8(bzasic
zN>q^(n>$T11HXS05DyU*&-F@NHrf*||L*-pqTorRI?fZ5)}lSyaR$E=_A{;`@sNZb
z7h@<MHRyM?UTL|HiAh${%Qe|q#Ys2S&|4!G#MgM8M`iml3jNKQZw`yPWCqWATfRGR
zx<~gRKltfISK}27mL!Ak{Bj*kre+1>3%@JmH9G%{J3yQ>RlHO3`*`}T5%JRwx;!&;
z%hGss=eyEDWk%!E8C-g1#@=*=rL>ju4PmeIc~7`{DtgbJ`!Wx-7{143Z>C^7r6r({
zX<G*2*3c3gVjJ)OR4ey7W1`PD2-Oc(+N(%do+O_H%&m3~4Vj8{^PP}DxQ+JL;cRJQ
zU!Y7Su@hHfP{v6~$`gJU1miF3&s(V`83_cNHBsCCRlnB=X6t?wuo@|t)Q?Me?thZ^
zi%D@Yecvgb_&q=ZJiYe0+PbWv+Dxn3JmlriLfr}6z2mGQ0FCma)VqPE6}>)TvDD2i
zs^L=PY=#a$zxD!qP(vX6Apah!)@yzGNZC1E)}PWG@GjAesXTlh`()vY`U?{)Mm%6o
zRPT4y_;?nlmn)4kW<*a%zc<6-IwuvWOMo%PvWwQlc;dxDQBgxZM)r_+<`85q+7BhV
zrM=K?ySH)rc!0|)=j1CL<oAn1;=vZ7H6+qVHf-(%tx&dEUX0Z5UmUdjD{~dQ0Mbk{
zG8d!7V_%p-EF_g82sYfm7>K`ZPVvqtK=BEhRy!<tJM(Mi`1G&3d{uq4HB7BA-xv<y
zX_b3ZHZQKLNBt@dd0Kg=n)@87N$?9#Y5Y(o5WZf1MLz}!Y9098HnV?4Tb`U*Tw1uO
zq&M!npy%e!bOn5C?fh-KxlzpU#=$STfSR4pxhg|wdC$y5H-nX_6_K;H$KwBei-}t(
z@P4`XP1vB-S@?lgcx_z3vb%id?Jm08;Ylw0g`Y%N&lx7HZ`&IS6yD;(-u=$4Fg}~b
z`aR`J*s)eS$=^U=SwHnXnk#mtuj&%o1GQYaU%d@?v%h<sZ9lraYdPsqK~49&aI0-{
z27C4cA!i`rwg58wGETm22_E4af$s(NtXTu15_g%|u}S=1R8b7$YAX~o=0RhXkwh*2
zIkX1{s~;MtW-ibm*ddD3N7jzPtSc^em*C0e)k6MQ3RH-K)2hM_;xFSP&+tvFal*Ix
z(QQ_fzH`G${X0ba+ervIS{?hcJ&~v8tArWN{-*#bzxXkX{u{DUg=m)~9)@!>w}+Lu
z%OoPJx{5~6s&*f@C=aUnH_9qrJ1?i&l)9Up%ff2(^BS=ym8|%oqpOM+`92P9F~Tf?
z>L_N*M8i?O5WWU(sbH1A_69RSYz1&k{}Nm#458d=Op>jSQ#UW9lG)&;{Wd?}mfEXS
z!9**+<SM{Ffqkx>r^Y|V@SG`dHx7AAN<DF9eqT-~N0w3&sB^JgWk>+@?v#>K+Q$Xh
zJubSSv2CZS&m~2K>n=B=l`l117rzi>mmTDx$F}NYZLepKn@jV7_X5R=%-3cp&ikqG
z21=ERz{x~b(zt2tPq}%;WE|Al{nTYCr}qS6Ok-rXC)Cb7(8l7B<%*N$$$VyZMnpf&
z1K?&P`rWWMN!G{O;4O+BUT*fS;N7*AUS>|S(pYGHt?i8REG)MpW$$`uMv^F|XyEf1
z-2>8Dw<i*$eco;k#tORRu(Y;{-e2mTJpCZq!CaEroUutEC$r<IS>m(P)T2lJlsGO&
z^;;+vi|n&^{lKgG4W5k}smCJ*#QTxuKD6?n|7fEpyXB@({=taG0Ka6qF8S*%8-OHt
zS)rpuEc5%&DLX}l%=ghbqxD%LHSQ|ex^E^5vMs@_sEdmfP!?mSmZ}J=&*|&K*7?Gg
z44~4LPs_JfPlYLoj;yE9xHBo9%INh(oea5T5w%%9<;KunsOP?b`VUyuAmHKU(I;}U
zc01HX@+!$0tx{HxRr+S)Q3%_z7MauB&r*k$K1H@KbpfPTbO(+0rG7Mh6s5tuSFXap
z4ey%q2&QVs+gz3@Aq}e0X?i6xb4jB-eT0-hx2(7M<~?cLT3RVs9*~04l#~EwiFUv=
zPE6Ru2T%IZ8=V#IKCaEg^D&EsT=U*|2OZ~pdRFcSQjB=*fXS4y8)GT1jSO;JWenjH
znqe2)(oKk+q0g(r@BurR$DUEx1x<8*X1||I*znGw%L@-Pi9SA$N!X$<q3issuasEb
zv6_lN8YdsM;HJuLK4FKbWsmR9>2_7EPv_r%tLg9(*(aY?n!D+&yh(^bV-TW5-uOnY
z?XJ57cov4$zlYW2zim`c^yhBOQ34YEO^O!VARr0L3)U5T5~-Xr9`bn+980^ucklA&
z7Q{Z!Vs+yB9e>@}<msxa9&f)0gR_nf3sGuY(Kv<1trAEY;4HW2gSt$+cm1tDD&ik;
z<RYo#k|kp9b~<|oGO@vViq8a41fZVAVG`cR!iVO>-(Zp(cs7$A<_SXFI@KD^4zeNs
zCmNWGxGRs7%)Y;k2g5~pKX*wS>kM%sjxEx;@b<dW;9rxbC+6u+OspOlJeeiIoxB-a
zEqIMK56_#B5f_ykE2!*>8vcqzI}~H}Q@cSF$1g8VHHON8h2b+H*P}M-0EWx;BCZAi
z?a+4<=5}<`7ooF0wZGvb$_t4mr=VR=jV@kis16Z7hrNeQ1r&j@l<4ULfzPN2IUmJG
z-?;-RBly1%x-|z*Zm>fRg+F<3=Lrck2F+P5xzGId%$M9ZV&Sna*MbYKc||g#**$<v
ze7}z`XwB&QqS}URdF*8(6<B^Z9SqV0HmWci-8j~$txf;5^=d*Rvv6(s=2a@J2TQFg
zJ&s}XZa?w^uxbKtU9^X<m2n5Wev1c4rEF4o7_@LV+8lLm{m9CS|BA@&rpsEm9ui_&
zOBq#@qxpbr8G3zCfATg7M>T1%A#m@y!wN^ILLoq^v)Z58A5BUSp&qD=-qelvNl@*W
zgq&o-I;FAKb^Sk>cCb84x9o1}O;eMztA((s^*`qZNP+bCycT2BcdnYJ1dvxF3j^ir
z;UUtP(j|;0Fj-C9)ttMJX!r&w1gEq1M=y8Re$7S&6C$VJsHf!fG2ptmb7T-1Pd3$6
zqfo#;?p)JOxEq#aV0b2rcde^Lhv8#1E}6YUB4powNxjjIwo8VrG*EOGOs>oNe#N`<
zLDAMKP{kWqoE_`iaTtO`ztTaoMIwgHd>Tpw3l(LsMqy7cith>cKl_1o_ofYD4O7QO
z@0p?w4zpI?-Zy3o@qM;p-5x4-U2?DUm1ZTUo|9+|<S%acE<2Lc<8pBPlobDn(ZA(F
z5>acSw^U}ZLtQ%9H8mb0fb!2+=(dL6u}YU~!#vd$ag#4y?dv!^k}MXpO<3pg0Pgmp
z6Uh8&`Yk56;;#xW^7HMsi;EG8G6u{_OvF+k`xg-ofM<!-XR3v&qPD?;9PjEcRbLG#
z;X(|L1wF5AgH5jprT33If_J(XK#eVZM{{&zAD65{H1qlXvR&st4tL<e1N<8HR@EnN
zd_D$=r%`Jta>&MekX3GHSS)gAwwE5~Gm#M&FcnALW=)wbfuCOiyz4n;q(f8M(5{H3
z&%KI`1>a=9PJMNItoyr&Vw+KBMiee+rtjh*yRR_+#@oQfbIdQ`Aj1~luRl{SRUhnE
zA-crlPvZ$%?|uEYOA0+zH<#-cve^veSk*McW18U64@JafEIBV1WTPKxeyj^osdhO&
zsjYGseryPeVxQ2sBUU@WwJWrA;m#C)^(|V+%i|wFxny?d$9OCIKi|lDx%b}79TYV`
z^`GEV&LXLo5=zLPpZRtCCp7Haf@-ZlnjyaY$n%J%XQ4xcshKwS=oA0ed$R}lKY&vb
zF4YU=;y5Du|8Y(~XZaY9&N3W~cvtm3FlTpWBh$`o-579%{1v6HI+1r+F|))(37(sQ
z{{Z~(1FG~N%t_-<_~k+QabvvaCwOZHJ5_pS9Zr{S*qlFubR;=2SNG!f=Dt@3cgGbO
zi%0Y{+eS8<F`N}W8QS2{@FTOSZV1=bnyScuXumr!m-9KnFxkv9S-aN?AK$zY&v6P?
zs%uqZb4+#&0w~BzKyg+s;a4duBs&#EqvvwiG0jdDD)#J7@Y{WBCxD=j=Uk@e>{SWj
z{Fc`-csqd9_8n<TK}2&Xk6y(o?U1g=?)a*69|&#MseQNNCy;f{5)l`!Wh=}FXg!KM
zekRfFP<DtktNi`t4FSjIM{?d_Ti(}2K$CKk_4dw<_qfj&LMB+I!OsjD>C^@;dl%1a
zlTdZ})LlLv(*FR~C0JL`7@&kfSMg-pp&LT_fM<Ig&4vrFd)KGQDjiwOK~#8DG-HeV
zat(2!t^M!Bvr}CHh&QOV3>;;~)`Yxo)kxWVl{j?UMMG6v>KA=8gq_R>+T45mI>O)?
z{Z?aCRYSrDLdM)JV=Xn`;MuVgH;&HgVj4}EwhmMYr>kMqXWUX*J1*Xl(L_Al^+V>{
z$`vs1G}9FpBJTk4*uj>!!A|=@fqpG@39M?hMTJ;6*R4s_E!Sz7`U-^L>@T?2D3lyZ
ziUls%WjdZVV2h0TH4m4sZi~!9?wzGh^m1l`UM*=r0^S=*g-u4e#bg|2{JT2;-=Co^
zy?H;w`-?A{5vqhAS7~-$oK&KIwh`vg=N$x>VZ(K>$|O85PQO6hMnhKdkZUl^RCBee
z6%8!YGwj_zgniEbMMpq5KUk873zxtVAW#{4lRckGZDUKhNWj<XSX`NJ*M#AoSG|f}
zcDGd?SBTMrlUB`9lxDjDdH4=%d5H5>i<Cu@&CMYxa7({MJGq&LteRm9^-1mD)xvg`
ze7yQH;i~gWvS?U=gQ%#KNLZy!UL`rXOO4VzU{2Qk*2ZPARplGS?Xt!coUlCCY?Q%5
z$k+<#d;nY2a&+U2V#$AaQWAkOkt?-d&#D9F1D38)pKHg2rfg~z!Std8MW8$1RCr>;
zRG9MNYAHdTEDac~u~Q#OK{>WCk8W_W68n-Y?xHZhKvjBP64wQuj!RYZ!mYa8$RHJc
z<H9fIDc<z2&QehQ12AEHeCN@Bt~*jUYZRaemj}2{X7&9&R8uOQY}c+vB@)i1Tw2L~
z2-=p`L#JWUACmRCbn3J2%;%2%CI&7^>yBq`8NPTpUrOYoZ^hE_;jhXbz`2w2x45z~
z8wU-KzZY~LwiS`HsS}ikyBWLvGi5prDuWNw@q#;+9NO~W=EgkEw4+2I<mY!3PaWz9
z(%wv-!|FCxkp!cyKuBjAT-%{l;NTi3^qI%(Hp%^r16a$lwGZ|X`(=$CWx`5VbWnP6
zbSFRX=6!a;P?SjbiLWYsCl7};i{&W^v&bJYt#faYhfGJ>W1P?w{QUo01#>=i)_Ex4
zt^+NaTR=Z)BEIPp`IfN%nE}WDcOf(VMZl!NJ??{d<r_~~x`gCSsL76aLvrlo^pqR$
zKwpVoY_`FlnE&)7#?@45)6x8GIayO>{=pzH7wHV0&HIZE7_$V3n{QCFM0g!|Whk3F
z6=NAT<lZtT3tabYO3JMfRVDT^Fl#FQmbwu}cOJ#t!nd6!+Gtx&tF4nNxL&laiLq<S
zTxTccXTN02sQkUek-Elv?N~vUr$_m9naBKn!K~`nJy5)7#qY*>#`~Df4&~=gS+0F;
z-%!tECpBT|dauCre?GKm6BdEdg%*il2^rGiXiC1`li^Lrts4Fl>LBEER8r55k?gX>
zw5-e_?mV)~Wvk=O7_E(qC(m52_cK&fFh$Ec1>}<*XFjm^`o3{hQzMjgR$Qs4%<6hp
zsb%(kmVnWI(<K!iJ^pSKtN+|?*0#&csq-tibNoC%>CNUp091l=OF{05X(zDf-f|D+
zSJ=pX=Hk`&2HwYhTE5~*_rPqn+sB+YK5iz%0=K`uN4U=LaDL*WmJ=8-aN5`U4xZU$
z6(4IvlLwVFz=|4&aDwFv|1Z4a4rgAc{{UY3FV^!maK4$bWrLS_=<XAvD<w3zPf;uF
zQ2)6Nsf1OWb??$tM^^(imxQb%8ydv=ytN^t`O}9Yq^obljZ0R5h%#<K%+6bK7l9d@
z(Ehr)_jODd6+!f$sQ};eN`^L^x&E|exxm*`8x8J_9*0H@L9#GhIxbNbduDcE2_{>w
zFacqx^*e8IWga10A<F|Hyk*TUWIfH-0P@oqb555i<zZ@%E76hDSC2KcT{9+So}_V?
zZa>^E^JxxYm-wigv9@;u``gJJ4`yL5>|U*Kwta79&YZWm*k9B@;c?SqMaB4?ZSo~I
z*ct7=DSB!Z$gV2XxFGzcIyzsw@LIHM3OK({#uEEK(?3O-R=-b%SB1|$x7Qjj82@8$
z<ychgbFgz7SWD8C1*Y7GU|K?_>DuP>lps&cxO1Cm*S&cJY4FL4mKci}Ev@`6CD9a~
zZ_1%~EcqC@HU8=n-RRlXzFxRCKc85wA5;X`Mv62D47WB}ZB#pfwO4fNni^wCb3o$&
z6xwg$Gk^HW$ika|E{BtgnPoy6B~>Pw$V5XdbWjYm@K%PGn6Mdw>SuC%4|cBKkf)s`
zU3*dt*i84$Cca1KLT9?VK%I0xSIEkW99ss6PQZacT0l|JlB`~3Qty^gjJFWsz)i4G
z<VJc;BLsZeY?uGcnBIgPM{Isi&%z!49b=hTvU1x|Cn9kgA`<+!p#mz~00y({S4vE4
z3S6{&0|hE-P`>dQzKNg+ZI4x3x{2^=(^_wk<1m4xU78%#WfrrUh|$Pn!aF;<k*<s?
z*=4>-sS2~OmwlR@SK1#s)rNOTa>chQp~&#&fV`#KS^|lQc6Yb@Oy^5FuzdHa<DFl6
z%m&|p-P^Jm%Em*oAhxQyAb`I=q9s)gLdm|b8e-u&Cw)nRMp;cLLb~b7s`lxeNY54b
z&`+g&PmFO7Si6I{q{6xK{!sPJ%r>ijzi1cUlrEyZPi;U#9{-U<kg6uJCYkyH2d&VO
z?Z=|J1~#r?CNX#s3TnT675`K4*<IXu;4w>L%i?E{c3lwb9$hAxKaLT+0oQOc_Spc7
z;Y$hmo7t@8zR8P|1GKCh#u5T?jCPy7U7e0iOv1PYHRNQQpwBEvyZ@SjnV%?wv^#o~
zL|Oj&?I3BvIiOjz{rWHvbNVCsyRG_b8}bu{eZv;gJArhIDkVryQ`L{aY3WPG4c;a~
zX3e4PiNJn~nmOuPc(Oh`-kSAx5PRd*bRru&vfKF!*C}IJ*5^L;+60}*XKog!o;tF#
z8u;`d`Tx)nyCEtv(Sp}r>!TYXLgnn~H8lZyAugoyk-;AAFZSnbAFdETU;Y&Hqci%c
z?cJB~<(>dV8!c0&){*4LH_4IDv52LlW{CQbFkKfjlXhdq(x@ipOA@t_y0p3+go}81
zrk;<~{Hm@R9%$I2$ixKHC$~#+GVkvrS)Ug_Eq4qjo3WhV-{lxzX{?wc$~eQEGZr*r
zFUPN+?OUISS0040fzNW3>{*)eD32A9UgS6#r0~KmUEsKkcYYI5dn^w2<GaL0^?jdo
zL)|Bg_F^ZP)1VN(GY)Yh4u$7x&tYRW+;y!u1u-De*g+gT2*@x#R#2MeF7<cXS>_8`
z5+aXYAb)w5Rnne033_pExZb%CuhGuL<jKy04ySmL>%$N%TX=pkT((%ggh?AWEr)1O
zBE_4@#Fo*^W5N9RM1U{t9SO&8?;ulfx4nxa)CeEhw6FwD_L0ke4^oYeZUFgzd3VY|
z>U~5OiRD-C@Y*r!i>SkR4Mv8NBXQ2p%zTeyv8$yT@^sMTyrC?z29~gF_SNr`Z&;tr
zZ+DJX^DVcANxa$0SWK(_)ySJ_>JYAP<}1AMeH$XgnW`)d=Oe5Js7Y;~bR~9aHYLU{
zO#Nx8e(Cy-Dog2mhN(p+L(f6jlWek}P@vO~pQp|Omd(!;h(25a$38_vDt%u+%D{Yu
zuoX<>>(~tSl&}@lne?;dMv|cwb$4Ig-A)h)pX}#H*mCc?v2pfs_Nf?mifPv^zEBsD
zp=)vat&-Py;02x|W@hdNx7^if&r6!f0fr%hlickY_(_5EfoYY<zsAbXoLBK0*PTR*
ze}@$~wyqJtUicYjM1%?ZfYQSg2?uu&Iw1X3LR(uJ82s!rvk=kdkB;a;vrTbdfu59J
zKF*eHmAVVIpM-8Dc<M}US8By`H^W@cB9qFna_gqbH(hHr1i5MR4)w9GH4JXC(2-_G
zj)w4%ZfA9-{uKVRh4Ag6b(xUIrVP7*WH*Tp#}=c#%T?Xp2Z54|{}Sv(E5kwO+=S=J
zz0Kunvxt&QSe?X)b7h`tu31ENhl+CfFV{rDXfAivOh(1kiz)JyMf?=|@EWa;$EhLe
z6My8<&?6t<(f$Xa#_MyK7e%b^S)()%*!rQPF9EO8GW|nuKa2d+!_>jZBypj-pdZo8
zi?ApgXZY?;P2K9=Af;a5OY~M~J$afiS5@dAz<EYM(>u!59hE3&lj@v+^MQQp68fk7
z)S!+)$s%6)mx+Sfwfl$|ITQ|Rc-2s<bm|{KkyOj;hfZQmkheq-zOw=?_CbaBLnjfu
zC_y~$?Dg7%H$Nl)4`vXwh#Ku*{_1>b?!6gQHcvoby~LF7pKOO3?$AaLeKe8tMz*MP
zBp=GHeyey*?e*ApeiujvP++KEayIiMJorDVJI|n|zIRav<tI%*qlg%30s<l+C>;Wb
z^j<<q1O!115F=6rA%F<ddyy^@TIeNoBE2acLJ1&6N+<#*)PN`d|D8MMp1EJ|ojK>r
zp1t>)S^LYH+3$Ma^*&EurIpi;1+1*f@8_e0Q3ZVsH}zY9#xJdB&^Y<;IFpfSS;4=6
z?EMhBAO`XZD?buX)K;eaAyOD?*3aK4(va@PQbXa*1n1A@w=~a_l6Z9mi(s_p84}6#
z?sVbf)7-ix7f-5`FyiCc#0uy+=J;{gE~t1*4v343*)(*fWZkF8T@~Ceq%!Fqr6<Lf
z>XF9bN}3xUBGv_=aw3>IlQ@?ouutOGQ!`d;C&OCDleD$!2gpcNKYi^_FJh69EkCl<
zChVkQ;b?5>J(HbH^G6=)4it6lu(PcrQ|{;Jgs&dIk`*^<aE7w2IjS{NAD^^#uV>5q
zi1iq|<003Y-|wrSjr6KZ7eAKSCGWQK`iJjCe4Hivmd~zF4#$<R##5cc<V1^ZE4UI&
zcMXGT15QXk`QEkWJLI&UA30(A1WNxYzDLjsC;At#L;ZpO1vsN9PX@maHlJHQeJ<0t
z+l<J#dv`~>yQH<?npKjl5#zO7kBxt22J$6DaSh_&v$Zwlc<}?iL;kgeX3E;W`g!M@
zeuz-=0wE#W|IeT7KK9eAr}1vT=l$J|@B9VW%Q~3l!ccDw{7#1ksaEJM%^mF_+`HqU
zm58%Xi*Y8|j&=XXO#Ukwr-#`YR9W;?LWdF)M(;0RboyV>cosF_uJ+en(OY~F>GAJ?
zn{L()7e48Kk$fibA;McGYyWb(clHlL6n9Ax&lI*j9iI203M~{9aHgz1H^hE~V_2G3
zoEN!13>z3Ng+VnN>`)K)xp{ZPb0-^F|8sci-@_)<;U7NPvB#{z^_@D_|2rS@<OzL5
z;{Ii25GBVo#m!3Ia;7IzF<`qcenuxM)|~;|5`(OQlutRwFTEf8E}?Bp@Y-W2GH+AW
z5i3pGWRcH2DR5$7JR%w#-@e2rw#3=Hv_9urnRl<tUDgDwQpl-~hS))(AP+{$jJK1+
zHB}92$IA7c7cZ8*cba>l7_r=BIqxEePH~nqp3xPpcXcU>H6!Y((7wv!CkGkkwYiqe
z&WkQJqk2gj4e$uYzQrohM>0E@O;$OG=B_NG*S>kd%e|?~2AsO)GVB_0mZMQGu1p)9
z`lt{N55Y3Tpp-c>xR`zCl3ya}9}f=?_f6P>ALaC0;1QCmw7uci4LRVf31QKEpCg46
z#N;}wicF#*1Xd{n(lfZlO+UE}Ds|N>OFAd)y)WiAsSJN7eY=yFUPNb@=o`{E>SOU|
zd)`0?4KZ)6Dc(xwns%NO;`F24D!|}jzst2tqrcfpuu$zI>>3F%s0?eLb(iwp#K8Ki
zDQWes*Axbtecq<_V+W2u7`GF>CWk7BjjYq-H2Cj2t)>Z!vPlaMm)|zrI$GP7ocd}t
zK`LD-j={#j!Nn)mO+N1(C};AM=3_B<_(Vk+HnHZP8xy4n{u551^xc*cG0AQ04z$PA
ze*i9|Hdv7)$^gz0iVE=bRoToIcMjRVtQjNQ3h{Qb67hy2S0oRubzht21oM-ywO2_5
z^U-W;mahljd&FP%WW)@lMrt+%_V8Uv!|!vj+v;#)b7Dmx2z8hdUMEqrBk#S#EHdI5
zGWkdIUCA~$O`dNp%=R%|u13!@u<0_vl*W-X<x3agu|f)1B1)UdyI&(j9Sq|KjO071
z#%BLP4@xvDWU+zoX<!$qK|x$5u{aN`RD+eK!x~*MiIvx~7}Qqujoc%|q)<Y8|9s~?
zd|+OrY~7Qb8=h8@RP-&@osFc4ils<eRrl;3ir2+fhv%G4l1Ci-*G&uY`H32x<0}~k
zF?{VSd%C8Hrim)AXUP|gHVd~Hb!ztv#&FixN>zSUG;2OsBDyMD^x40a^6T6})TJ9u
zDF+;6qL!zc*|_+YZCk9$O=BEw{k}J$)tRmQge{*bmaZ50Zuxp<#-cT_@(I_SBxj{k
z5u%SqFU1Gb<ivti44FEY311I=)+ZoJ&ct7N$>f$zs(Is-$K&upIOJ#{exxgD#}zvG
zeEvCcK%{JT0}6RLqkz|<5m52G1but^D{G%(86>3NqIK*xbjsC(u&Htnss-Y#D<37K
zh%9eR6p`i0mM<#)0S_R8ziN)V2bLP#s@Jo=j+T<B!33kVX!3!v!1C^kYQyWuH353Q
zh`kIO)^P9Y_+dE0ac8Bt+J3Duw{F&>?n1h;V59q`g!M;K4>G!Lo8W<KUUJMKnvXQh
z2lVM|j4FcZ=&C}-aDUWWxD65rpQjy`;R4AVY$73;LgB<qW1}NnANioP$VElV_N8qr
zNcv485}bD<aWhgtR|s76uY7vYce;BPJY3*+iclOb_3c*#o@>d-$P6Iz&lBc6pvF)k
zP<)ZRpSp0ToW?dB+D*F`JjOd4S)ZS<b3b&_#6J!?7~QnBb&xP)0*p-z0Biw{3?v|d
z4%3&h+(HKfEqsMkrt_HxE;zXHXwVImu#00~B)dqqJ{C90Yj!d`|B?c&NRl;+F#_m>
zfzuvl?BBLw@1B3WEXQ=2PqsN)J7is0ZJ`RoBoH6$Qz#@X&-Y7En###Umh;SmbHz%o
zV1ylcCs=Et(`RyI0$6IZw^t#i@K^U=0I>bDwySa=bSnN1*rNNm$--MfZxzm){QAKx
zv)@z!vs8X@Ji5LDmoI%?+4-mg2j_FN>NWpqb)yP9-QaQauw$rds4A2dTeb*NQqFe_
ze-Hg%jzC7*>v0vDy}g_+OZt0vCFUhiVKaz9qwSgXmTg6}rK!!x;ppysJ00S}BfzbU
z=|TPQ?xV5nXNZ;0FU5ZIRDtr)=aD%DvrsL=aziaa|AY!iMSZ`ge(@C1#nR-W<K(oC
zxrqYp7W4Wm&n%lZ`;zImV@Lf4QyyG1b1-_7$!wA~pMzig>riV-I`wev+HD+j!f7JT
zzEpl(%yFOF;<OHdFnE;7l}F|(jRJo8I^%e}4OZxul-h6Gk;?w6#$i&t>uGzi0Oy41
z){wM_wnp-JKbCzJs!Z)iWWML*!+2awdDe!EB}6(~fqwmj$#Ta36G`FX;(hOEw!jXf
zZnKvltA)lXb)TWuQEEqt%J69Klrq<X>`g<7Cj63(&N`!}xt!Bz)Aa+cs?VljcJD&t
zt0G_CV=BQf*W8#DakZw=VdC`TIW)Rh2~L+!kP6?LwR5qA*QLF2FRcS!PhyRutIUF-
ztqKxPJ*L%)RS>#50Ano@EhV)7vDpygjL%xw(eflWY{$<qufEnpSoNNX#!ycM5ZDnd
za++<2SQ6uXzbBU*JEca1)L=IGuhagvD@XbTY>+d*ltaRk#%!uZEBY_R&%zmWG^gn7
zL?{mfmt{TgL`l!u&lSU68Je-clyI|J5g6yU6X@j1rxdX)kk4V9&6=l`hEr+5RTWEb
z^r};ciKSFX+qC)62pL>GbiP_QaO7sA>zACk280{Pyjz+Tn`U7dJRC7^j857+b=Y)f
zj%<QsJS#~HRYS*hnU42hs5WB|caa2pz53;BG0M-D?HSFvp4)J6zxhmP<Lym_qb+A>
z#1|$zWNqYfhVq^50?;y9E-HAe+V16->%Fd`TxkC>UqE*7+Aq0ge-zuQ$k9O5Y`$sK
zV=e^jakx<;CXb@g2BKD&PY$G$(rk_WGrgSLgHM^@_mFDC?s26Cj}|M#?D{u)OfNSk
ze(CNI-hi4O<wU&-+&xzBv8D7ScFgs@T@6vDG3AZSABGSz>#n&@BnPESUCGv`Zs4wM
z?j>i0a`m{`^6%+PuSOC70@8>z*c;Wuf3lXJu?6@iH-`P*8;P&Y5EI)@oT-k^KMT&m
zraZN~XQ@qlJxGI0I5>SNFAP!ZeaH4$x3q4X*A5^oM1}2h2W%<$cm}HeaZ6f%Q*0oM
zohs{oe#;EaDL}8E=q30QaE-0+VSSnJ(2*mT1R@fX8nM}HU2AN!>|w<$taQ(EyEX{`
zWEspWpqfBFFqXe<qH13o)8Em+sZu9a_-eF#vPAC0tt2AWpq~RFL_K2%00K7+P&<be
z^QumGSp&JrYOJjWgJbQT{kY*gv=+WTrxuWTv23)hPoe>A@p;hJKBpkNpy$nT_{160
z4Vdd16H4OL!M`M(O1q(+Otd~ZoHGO0P2hi@S#zyf3}Ijkv(g`R7RghWA91<zGQHvV
z5a#otJ&Bd}^{aXvA^$tQkIO5#Zbb$J^LXu8DZ6FQdCSyA#<5n`f3QFYmva}@+n^`T
z(YJd1^6IL`9wEMQbZ$$2@;36$5`Y1rmz{S*6(B6>aI^73@jXUJYspMd>`+ZsaI7c2
zK@H34ImaS_OZar7S0O6zS0@!)ldA11Pa3VQ<S~0+YBvi8tw`UO+LB-8b#|@(3+OOb
z5l}Al)ml!uk2>07^SjzE43wcbWS3aHm=&?_&~G05fJW5EQt`i^lp+`1*i{$22xu$+
zKXxhdY1*4AB<*biYuq-22TtJinV87H)r5EBZV!H#Y%V%6V}NMBmS+m+ip-Cj8yBne
zTD;4zbQ!$Z{{F%?G#^tx80uP2N3e9$HB{@@2?GSy{2nKJ#kW_EPwsPD>kvI^Zp1FS
zunEvj@N8Bni)(kOF-3N<VYs6U^cs#<n;4LoT@p2T8%o-A1-<Jxv|3$Xq*Q|-$kC?X
z=<2PK*vF%|+)anPZE#B~6~<0j9cxYxIKG%J9G>|4IdbWIZ(Vefl@6!o{(J9Yp+TEG
zcab0=i-|3bIWk`|YIj87!T`Nj&RM1k(aox7tifok5)z8J0<?0i$h)FMr+sUj-ls>n
z-3_Vqj1JT{*87z?q`9el(nCZeV?*c4b>*8M7SwzQ<evu5d)Stq-k_1&vKu%vZ}-o=
z|8-#s7LM0A+*lYJ)#6~~*t;AtwKfr0U`=suJBcyXoS5tnq^n9F8Ns_}^(@^rsAKzP
zQX@Lq{`_V|X6&X1OsD&l>nF1HJ4^Lq1pz`3{J8%#tL)LJf6L6|)*O#?9$Lgy8yYql
z6vl_*=t}4*k9BM)>|gb^;l4iz>6~mzsrn;RWMQaXQar=U+ZuJ96n&~d*uHjjSv`$;
zd7*2>Gr{4yg|)kIP2Boptzq)9p^WS*J1-pLC2{jqngbyHXCk&1OIM)tY3fYtW~)_d
zuj>%k%KrBUk?0f4iPl8g_!NS!FNu?6==Z+B!u9yX|6a)?j3Kq_t+Um8lh*C3jv~_x
z&;IiSTn;d=&en-PD>FKiyRQBumDr4AQl)knf9kP#`;GTktrfj#&dJ9s3ct4lLs#&r
zy+=I~t`Q}+jsz(xPG0D`qEeC{f84Bwroa48#Sh}$vbKER*EZ*!nZsUpgFxolRf+PA
zi$bQDdjjo1l`Wvmyk5!j20l;dp-B_u0+W3sh^8W&<fAg#+SflIig&zRH3Ndz1oP4Z
zs9I<hfD$bW=J{WM)NN`VfTQ0Y@nu0sDXHW}&Pfz`RvO^PD(I(L5#16buX4d^{Eelu
z*y(Rf^!`Z3kZ>|V{lj6Iz*12BOhWZ|JHQB!xVZ4v%lV(M=XaXZE&E01-TXXb+b<R`
z{CMVao*@(1SsiDiU*dS<ce$NsJoP<AWO8WZ+osy}BRq)u@uavgeMKZ6CKhMKP~ziK
zt<^Yu5k_Qw6#&_#Ta)qj5ED<J{8^DrBUs$TdUw=XbuU$27z^XAQyQRMcMoDT-Kfqn
zd?)<}b=$51%XP+{rVf%(^Yhx2j6uDEnrG!|jv0(fC%JIMw0n0=%E`YIzcVDR;Fso{
z)2@FPQx%e|X)_#1>Ko=9Y;osNo8A#Ij5zWNe%e)SQu}^+>eoI~UB;U_M@EOzJN=_R
zAHYii+c|Vvq|aFjK})rJPrGqKE6u96p9_+u=g>hlIE>NzgwdbIQ?Clqg1>4G6P60i
zti=n9F^omSpaZ67b;Fk$ef*@#gC{WEyER2h8fAQ(HeVOkoNKKZ5JQXn%-j=!crNw0
zAxCeV)S~R*<9G8Q_QzSO5xqK3y=kW&r$BnHlAc;%KZ^J_-Z~Y)d_Ol^RK#>1xO(gD
zNS{%D+SlHTYR+~QG5y3-3NGB3?2{h>I``PS9{Er`INc){^(eA0Buy8-Fr*EjcIYv5
zy>*#!TDaY*m>wovIbjTP=O%puJ>WYfe7M@BxEoqM?IXK7(FuE{gdo)<l|YrKbR4Cy
z_*&4!TleWEoep!S)_`ThM-_Rp2?U|5Qy>1M`n+e2GU+z$KKo(v!$<T~>bySC4)N82
zL~a-_^*a7O=d(3#*GQg0B*UmET?DVN7FPhx6#eZuYGeDqyW9VD5uqFl+g@u;LO$Ue
zMjmV=Q&}LTDc(OiFI;L@LWO*0d;NOsbmm=;4<~%}N)k4w4t((oz`y1|{Sj8Ph8TnV
zbp8(e;~zFZ+9K=2S#3Hzl>f(v$4!fu8yt&&2Rw?&)jmtuO?B04iL!!+{c*B*5m?w-
zy)EniygrRATxZ`7o52NN9w?74L>u^iQ8}cIy`Z_F@4l#y8iCpA+)~zEr!usU2j29@
zWo2TMl|@wZ=^o!lqAn6yH}syfwH(S{N55iDa@JPHNV0y>sgMtxiOJ0XDa#Emxd{12
zsKGT25=KiCV?uKhQ=9IJq`^P65Ey!R?K-H(W6ppKqzsm2&V01!8o<Tt@mP7@-Q!+!
zrwSLz!ErMmNR@xzCJ#r8v9Q{@%ettVpTh7kz`s*ZJ*nWX7UuPay;;<wB;H>lu<cO=
zl%izNB7glTJ$D@!&SAgi3)U#OKBz3$&}~DGuIi?8F6*0e-;qf)$3y7|mWOr~@un3<
zETfuRgLc+V*3M_vmvB&-z0q1(8|<jbf<fKbs4pkSKwfE1lBo7V4-`1kQdXF*K3{dC
zy+m?dg;p|d58!O?X41T@ZBp_e00d&PBnfKb*UB?~*88=O@Uzj<lHV&BvdHfuv=oR|
zGtn&$dr%136(Yf4(cKSA9f*KF)+s^a9iroij^A_)!!LBx>{OPYsp4lzPnz@^rN>x7
zGglG%u19B|KBd@E7fa#zlS^NXuiT1yqT(b)ef+<ucK`2*H2;O1mvrur{u*X0TBL16
zzmsj-qgSDS`G6ziT$q9gy^Sbn^hum=3`t}~1c8BqZm$`0GZH7v*Hcv)TO1v^PlB@6
zs^VV^O_H#gCtS-4$`8s7=Kca`W0Mvs68sSjG`Ij0YVA*?<LD!2WQe+Qpf)$9ntj5Y
zEah1i{rvrT@4=wWuT`IG+d-bWxeOM6Jg<dUC*N|*u{2unBNYapHIB!#etNU^7cf-C
zw=1e#DQCKr0oA&DMyA;n;2t75oD`3r<F&88+A%gwc_+Hzc}2(9gYhpwE2g7tEbv^_
zDKqEH1kDq5BU97v;5Ys+;FMpZ**XI8B;AX>x`iu-IQV?|seBsD8S)XWrs=0#DcWXg
zaRAlI#GhTAguHfaQ!A~Q&Ez3V4-TQij1|V)>wjdp&Cz>V<6OR3AisGDq@CaIw9r<m
zcGXhY%1xNc-1Nz^-80}l&z#0)GJCF5!DM41=J(Lm?9TAQW6#<68@2uen8O=ottZq>
z@wG1}Z&p>w(!p7o%<GBk*=q?*t(<N3(Sa{!=uW=0`~^hF8V2v87LU2dkSEn<8~YF8
z+dtT_3kOe7FkfvzR7^|{d5x!65ooduS2#MP+wC(uQlm$m^Tl%)FOXkWF6s#yL{Y*{
zwKd814;-MrZfZ&>_>-YuTZ%KlCF8s;7iOf}z%hFbm6CLc>MEdAS%6vEIZ?mwr{a6-
z&Md(!ZLz4sWW$%^+c;t`nS*}2h1TN?UAc%6sqS$hXOd=k{8<tQC~dA>e5Rqpo|!UR
z;80Gn_&tNR;agjDug4S~x^{L+0Qco7HICyJ30xbQ!B2E6|I6C@|Gc#S$4;MWqn?HW
zHWtAlFts^9q!rV;Ms0Z)Y5rVUI4zZe6{V#6kG-QKO4k&zplfo-c_-%&<dt6qHjkcl
u$F}BSig>mKpPjQ-`DA!dix1V1nOlfFT7Kj(%@;YI|1>N7ce+FUo&FD?$~Xo9

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/BgC79Zt2fPs-2144.jpg b/assets/images/impact/talks/BgC79Zt2fPs-2144.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..dbe74acf0a1c3cda4945c11e3f83a8c210468794
GIT binary patch
literal 32980
zcmdSA1z22Ln=V**2*KSQf(Lg9Aq0ou9^5G;IE52Dc!1ypceldbA-EI{!QCCEa&r3b
z``@0YZ};5pnYmRwYqQt-<o#r?TDG6(o>u`FG9RQr0AOGM02t^O@Vo#J2Oz@3Bf!HW
zA|N0jA|fIoqah=|c!7+Ciuw`_7Yh#$7YhdmpNNtKpMabY2ZxlNl$?s1mX;Qegn{`r
z4HG2|EzNHsFi=+H7s!~%$e1()I0Q8R>CbaJ0P_VbAzV8g3>5$t69x_w=D8a{@*5_?
zZ@B;Xz`(&Hz#<{OfVOI40AS$YVBz4A5D?%IkdRQJ)X+Oj1S~{KY$SFiF&smxSk>_t
zF%HTcKm2g1J1X(sadJ7*7&#ME<4=74mer}w!z=DjyLLjy&G$h?&BWBnB`&85DwznX
z5)S5{^+F{hAVS+1F`;b$7<ecUShzpPU@&3fu;3}#vBi|AI1CXSI>wd#D%YMNfLBlg
zSWGxffGFV3y}MyC*4IOE`W;0Iv}D;8ReRwN+!a-=&v{g(LWY*g%W;<<L+GHT?0*v(
zv*6&DxAhQTw6hR@8>|i}5_B0&52*Hvi`?Mc|IS;T!?(h;nD7e|-i3D3cLndUp=E9h
z{oW4@d1#q%kx}$Tj0alkOn4XlcaRw_L*>G*SOJ7zhtPE}<3LT8MVEo<B>Prz0fl3@
z`|sSMAPn*%Q;SAF11`~@tc%NJ8;FkgNTP@XKHht2h&qUV<y{R=(mrRN1~P}O)=#TX
z^!0pR$j((v+SPa?-|e8D$K6DnQ`}ux(y}#FhB;N4GbcEGSL&LI_;U7Ia0Ue~u<%Vw
z8B&N4e*UtyRpIM>B3qiszC(?+o^M-5pD2$i-`du?ujjT|WnHo`kq_i1ytUEp6uX!=
zK9R%tXuZxK>?_<=eN)urFGj{!b;ZHQd9In9I;F~vi0=(!YN4W4XR8HjC>KmB;>O~M
z#J6(%S!%l(Fk)+I^_t!W9aCVfrH<9I&3xSbE2+!7a>hZaB7#XSzkE$1x06Frt7iU7
zuB)VB_nxWl5wVd3x_%V-K8N}}T;@;8_1LO|jW3fd^9>t3ZR=;7JzieiZ{Wi5)M3*j
z@Y0F7(Kd0#bp`f8UY^?XMVGQG*W)!--SuF)9Ti{Vu_glPZ5b+$`?+f|I|fX$IQLf~
z`h^qpUVw0P8PStS2UK1;ijN;o(kOEYM;W+VCrfl7C3a)9<jV{ENL)!Gg2+Yo4g>2{
z-qG4jcXU+6?pxv;yWM2&ybkdV33(CTcgf^4GI>X0s>;XMJWy5&q>N0BHiE}4nIj{?
z0KqFueA9!$Wc3Ycw!PWISr8Sit7JE@Is=ZX5|Aq)Q`Q8iAm(f32bcPwbq35+ks%J)
z`-XshRPW}VQjT=hSMnEdlOV%6bedZ7tclnK^-{L6(rZqnV)(C0ZevH**0|lv;%t+N
z@O<1qid7@|;K5u{I+KRCRR~LTRCCtv-Cq3KiHRaI@w);WFRc1dLohsO-oJgXp5*D<
zV`TlAj`MX+e(dLVV6nJtsSnCL0)&lWlI50M*#$_p73O#Gs>t)}FOj!rOXw`EnFz<W
zo3YC}F)89!ruFdOjtXsgk}m1+nxx|t10OKVl<ZT@htt<aK1xLQUc@=fD9a&7x{-bA
z^R=;g?;<4jDcs<p)4(-y^(kq<=uF32Na6cV6-zzW)vk4e*0Mjs18YDL)i$ZtKJsp$
zH<~|7lUhpQF!&&d(LI2)FgKG6m0pQ<Xn5wK)Ap+gdQrcVtFy{PH@7SU-LS<W{k4Ul
z6&u@RS+>DlT76PVr-%|+tLm5>AQl+_wl3|e@PD!T?()TpzXJAOUw}o{?QMN1qyaC3
zp8?!6=!W`?m+)<O+oJoe%icPulEJt`OCx5akHN3bY>h-lgUk&`bXSUU_Cf-WsgWS7
zV7TyR-Hp3=xJl<OlIV-=%8aH`fxL|RG=gWq42nL#g6PF5$nbOBZoPrp#)Kf?j>)qk
z=Zf>%SzG8UX4CymOGOKFlG)5^R?he^$aI=;3QksPqsO7d#PIx0>zeP<iqX{@GaUqC
z^8)*b$Oj7h3YD{!Jl(XME#pG<Hfk352k_2f?+t&{6kfi)@egBCZdoc;{OXZ-{OwA$
znq$y5On{$b6U-+S$13d2ocwRnBoVB@xT360!qo^{POOGyP`W^p!^uWbz}!43&8Rss
zU>WX`sY?MR$BzA|lImB(b@6u9rj~j*w>rPAUeY0TB>i4F7kYAhl}Z4YDa7e2WhF4N
zsKdsBzJDXP7!70SMf@<`XiKnZuph-!SXaKYLywAOYHx+WhuC{fy;M=|1B>%-?wD`)
zG-8-3EBVXtgmi34*Y}&kYVp}pk8T=y`9W#ZZ#hzUspSxsP>7Yty%wGUjr)=9L#(?I
zp3Wb+%P+Ijl1YT&S$Q?y?GZu3B%`#ENxJTcna(6b67*{BhjuS}T5WyKMPEMy&WY^q
z_iM^n&K;it$;dtsoR#aDp~@!ER&~DSNL%=ibes0%<X8d|)SQN?8ScD34YyCVk4AmA
zM+rxz)73?cDRgwn1_O+lsLV-~dq~nWnB8nz)s>ppKoUY{1VVaJIq5fVV%_C0n9vG0
z=AQwyO?ByE$vpfwMR8elh<j8VAi%l}hUABOJXu$#1(@Vv1xBV~D#yI&H=IgXd@g8k
zz9`*JUGB%ix6c3uzSHkP0*j`gU3L{|I42d434ku((sx48w%l9)+8Z`FOH(_!xdm0o
zZ6^n)6l+$~sY4wKj}4mh7d*lby|rBW-l%PxXPX)*YssEZ1(_Y8ztY@b3PyXPfLWK1
z)^kQ|D)Hi7P}#25lWl+T-AMxFH4;ab_*^gS=nDB;e5Q#5dS@ROr}CfbTf6Kj{*8sz
z<0EQPD@zla_Krl$2#Zn;Ev0kKtaHUV`=d{SKiz&+-aHltf2+n0DJqT#a)rA|p8yG>
zM1@vM0yONG9u*ge1+CO=`Olo<ou*crOq}AXB<0!hZR989LmvWv`lC@#qfXACB`Pa5
zwiRf4+}+*2S8ceH1IbM~RN5n?`9=%V%(2P0xz2vPLYH?@`}D4rz?Dp#YS1$r{K~ax
zb=#u0<OPvfeFziwIOL`95}0<c@1vue-?u{1dwNn96T=!MG*p+^BY5;qL6ELne9Ucb
zX-@wkFk4AN4M7!c5fg?P3#np0?}{SY|G0!yas}jCv;eZUIh9vFC`BHZ{@A<2iQhj?
zod$su7Y}J^QS)0N4k*zhaN+#C?P`j+*{!G!@o3~x=Hclp+8PlM!-tu%KcLWMKDxob
zEu8Z?7-&n~_esfkn3Q~)%6|qt^!q;pib+2%;w;1=RMeqZj;@cqQLri^A{mrI?RRbn
zpD&|{wm~wabz9CkvaNFh0aGKqH+qg<$|Tm23qW}4WEi9yE@j_3>cRz+lN}Q!CGUBq
zoU_$AYW3A6632B^CBFnKe4nR(T^$|;)7yS+jY5BJ-1;WkY5v}u@lGZ+QYH(Vgk5SD
z98(Di9|^5k1spe`P1o<!^&7Cq$D^SoXT=icP-x-{@WcYwy5mRg+S=+7CzkKwaDz)s
zh3TxoW9pUT+*==zx|xv=R<~VT%K~cJ&J>ZKe&#&`zD)Od8}!_zO!!=7KLgg^`KRCQ
zu_>+Xe_Oz|7<qqFWMTZKt_bh@F0zH=Hr=eiTOrgw%=Tj4-rc<n$@+wO+x;b+lnkWQ
zVw~8Zi8}gs>)L2305Kvz5*r(d1Q{5gRHC2H0J~?vl2!h>(lcNR{p#H3h~k*ukGf&r
z^BFMx^C>os{TYxYy58aicn0(y*OaGwjA$r|pA$}MCxh4X{a<%wFqfkmBXK`S3jxF~
zh*Js!%Py1Gj3&D=7d1wg*mH=nx2#=(LKYb7Ooj3~mJ1zUc}yZIPU`BTD$0i9_G|~b
zm87+_8c#}sc$RwNZ9uK<GqKVafQDy)*pH$;a<ywaHIuxAS2ou_`a&hX`tFinY4!%}
zIyYl@JOdKFZZlREeE78*n26Q5>69+Uz?D40>_Pqr$p|;;({-F)+vc~y!#213TK8xv
zSdfBM_8N;gIg=)}^2dC)a@w=(#LPa!dRx}2s`P}qS6!RXNLiE2SWXsoN8(!|fsL<R
zyu&>_$7~0`O2|3mjCKx5<D90{ZkC3A30UXW4?}q{JK59Fb({}ANWCCdv@zN`8=n?{
zVXjq1C+mXn&1nK8Ej1SGT#t*iOLr@HkG=-VlaXcR#j(9Y_*!7jR1poJ8%1jOC*do$
z`?ffZ$7#Lg65&*-wB?Y*uP-b#`pVrB)ix6y82&<dQE;}nWba7O)s#13tV_AIDg!sT
zAl5TSC0MLH;e#(=-BzmQaPJ$*ex8^1kKHygdz`5glJt%8^gJOKRyM7(=HXS6RlXqo
zOM2ha`l`{fV>bcJrlR_-PXf)Q>a1EqId7`?c2ev*T%xp=wkx!|noor@#v2<WBPK4{
zfU|1P0F{ls3@sdkyV4+h4DF8fBcewkiA0;hORVt16;{D0K^e`#sCh>TErrR)?p0Lz
zdvuEn9Kw}EDxY7*HmB6r9M6CrkJA-i9dGiT2iudp8bTXkC;7-+|DVjx2W$%+BbKC#
zpc%W(XMoG$qN|k4CX1~PwycnSKRM2{L*?P){!a|~&eM!Hg3ED5YTrM7w@EIjx>Tzz
zNv`rRE_<E4;LGhotT*aF0<aLvA4c2s54U%U`aD~wAWH(d{USfL{|UdbtNT<4KX1Tu
z(MDT9%RImD=w;Gd+Dd_M?`O#Yx_a$i>eS^9%4_w<Qa@rf*K@LqF_UO0xJ!)J@M%t9
zyLwz+F6~x!V<o5c)C~hjjbsXR7d#|xGPg1q<`R0U&k0w!-(Q<m(Pa-#43aSb5Qpu_
z1D7k;7GaN=w%HQ4Ce<$$3;5-$QL-SP+ayGS61um~S(m7vBCG8HYs;+>Z{NR%hYmge
z%L}-<R<%8p($;#UtVaE<)*|8XO_3}VyR>$n2dB0X`|C-qsKY}a<dk2tmrtKO8*oc!
z5_Ni!nTG4fE*L+cgbgUGSP8xRA&c6-T24DSW-P278q6_^Sz+b=2>Vkf=`>+(ZP;5U
zImj&b^*mZ!=VLLYV8IAseo{p8i=JdV>pKmR44?kpKoL=*+dRhr9g?6?2jLD=`azAc
ziQeHJCQC*ddR>ZuO#Q(A)2i8ks6_HI;pU@f0Qm6MdOhX)>3$c#q*5rR;@dl6xrA<=
zkhNXyBW({CadA}67uHnt2jI1b_?PYuF_Qx-Lti8_{Tw$|cC^L52iPmu6(QA*`wanH
zeEA&iK|X5vB6n^N90z<GaT#~v23x_>t9Sy99NH3znT7sKe#VRxPWiG#y4%D6A(^gl
zmy6q*B7z59Wu9~mUs)7C9_lXt%Nt#Spbx(&Q-d2h9tB>oI7$hCLe}}NOVATbI~w`T
zl9Pqs3jY|H?weejE2mgdOYQHnB9hgHn|s?d->yB>M7(GvVB9x-nl`IYUsia2;rfcy
z#9Ja3+!0|2ruT(=2Ka-Y_Pz|YFcxK_+2FDV2gP*rc{4rLYu<?zNZ_&5Y{je{Q$|j0
z#Tc#&3y0Iv?PM-GOb(CB%L}ugE&3U3>*{ORm3C|IkNsLyFr(Z$>E3o=%fiEVJs2jt
z?@AR}%i{h<_hHZot5H;|zOqS5gx2gLeLY#xLA(BAR(3GCBmyZXzr9NDTsClEV#XH$
zz5o=pM8rfILrD5yXSq9910XLKq>mVKu5m~zp8?D*tls)p85ie%v%_4Sx$CLvhxi8<
zpN7Ubv+t>d%<S;VFiWyV2j`2|OwJGow^Hh6L^6>&>QLr)g?qTt*8>uKn#g5DPr%QB
za>ZwWus(X?Qz|dw0mj10J^v+D<=8E_VUR$Rv>4j&AXClO<BmK11y*cGa)|y9Um5gw
zOW0h0ye0CnDRLpQa|QovA@{cm_aq-z2AV{xX-9|~w&~&DZ2qP5d?b;(nM-P_baJBt
z^y2a=BW{}&dkqt4aW`oYA=RVDME{lVej$shAL_oy%!A))Nbc8N@meBvmu$r?sz@Kb
z=BwAerb9x|3Fx%bOCwV>o92~0^KBjwr$DG9lIW;Qr|_4!w*j+W^j(y*hX$BA`QRO-
zgE&<%VhcFTZ{4)_Q(Bc=!Xa0rzDr(y!d%X8S_#)o#)Y)T``fn~Jj7Pj>O_$vy7wQ`
zzGi35-Za+=Qcqv)!tT)HzGMI1C;&J#N_eq~b`K#U_8Bm`#_L(QFQyM2G<z*H@`>z|
z7f$MI^|$+)<q-<OaGmy*O7lOV74tLeT{u|gu%jE(5u_E3ch=y(VxL2?XiabDgY4@~
zSMoevJ!af{e3|hcEx=Wsi9bF2wbnOH^%&r{+JgAUZ1}~v<!0~8UDn|}QlPNyRyWb@
zZq}RbH6<DkEf8eHz70#GVgeLqDcpm^k$-EwE6+8O)TF=*&Rdn$;DmUO=^Hh&Ad1A{
zN}!@5aB<aCaB;jgG%%*NvxlqHg);rlg%sQJ#Pl?~SBvu{2}J7ZQ-Zojd4--VNz1e&
zB&?oNwKE{Qi+XdQZkHp091Rz4X%3Fqsqo!<Az?il$X0uz_I8U$3w=Xg+~P;a!`-on
zz_!okOy9eO!ARlKDzcnEoSIqo7Xl)hfzHVbq}-4uK^PUkP9x)-AH~K=%5bT>^a#26
zIs;Skp`^`hokNp<Cn$PtT{^kx8bzl;p-`R(?6xE_D8p#O)xm*gF+L!3=V`Ee!fa>9
zzBfHlpo{DsqUs>^cPvH3y=p<lwc22WL#SQQXG?6bZ7DIA{6_Z~Fbf^0rnh*H?$(XZ
zvG`mpzl<ZhCC)wPIga0GXz6ZT4lkD<5_QT;^qL45wDV3p6v-BdS;)Y%9wBfGxIUB2
zR;%@fWdos2Z)M+_@5(8n6RYwCfCQ>Oiz<~mld>K$<Sz9f3L^RS085<Q7I{Qc$R+XD
zg3|m-k-F~`b51q(pbtWVa1T6H3fccV*R($odB^Z4IpC@OkEtN*IcrnmN;^!yOh!*n
ztnF&?QpuS+VcPWCsJ5sU%y%@Q9dP&X`&Mq@yW1QXf%s<Ox?%1b#<^$tbo3k(UH``0
zZEg9I{o)zG#^SPaS`B$wL&SD7e~92fFGDU{#+W)np5(=r<j(`k3O7eW>T~GPjTlL@
z(4o|8L*!Y8f~zL`NlQobblA9Lt8-%4%bZKo7v8I$mnIwq<wOyf-$LH+aGxjYxM{_H
zQTbAO-8P@}Qv+Jn>4t+WjKik5D{&(Cq9O#O?#esK9PsJZ>hp!tPFR-m2&?lKR-zI~
zSQLd!Hy2Fn6YY-|5qCu+-d4UuhSc&x3POGTlulm(bH3;WZdd-mZZ^sK!;L7>rtPMu
zEB9M7?+k1A-Rf`W&BBnZXtJ@EI&u$XTd4(!?=fWw9qA?~X!T}g8+N$c*NO_3bz8{O
z`E<AYly5?*O!+aJbYp88u)E)U?n!kMRO_?&WFtXHqAdtYXj~s{0rTDFW3P=}^T?w6
z_sYwPiKdhkAPP%BzQ|KYwD7Dpr&HT&Y?xw+3x4Mc+98n$Hh9E+)uL3C(}RTa*;0=8
zeg@R=WN2*YbIkQr<k<B+GATg!elKFn%J^W9Pkw<+;lt!~P?e-NK3b=`2Qw%?sS)n2
zB^u2YD=)5AklRH=z^aCB*EXsG9^tw@$AHl)sGXvmsrj(FhrQV?xMkjla_DDQ8H@WO
zbf(dE(Y?8ET2ziDw=K6iK{U3EN}7&Bv?cvv^=jyar(eEKPigNJ{^5H0&?owcUJ8<p
z+^$o55s`TJ2Fl*4jn_q%d5+8Ngv{!qO{{dC+rB~~)8+6B?MH6>e6>1J%p`F<MOrlW
zF<*_l#3qXt`BKoO9%kl(q&<q-?<L^PWVuN76%nHIQpk45I^0Y{{C9F1&?p=#Mc|9+
zB}j08;@bD`EDjKz!{@-tD#Y2_Nc69PSk8I5*4x+`F)>}aF)?8<G5$X+93n=}W+lL`
zXp$U{18Q)^MdTlifSkI71<UO#VFx?KCFc4?M5m5+VRqOGs84&)rM(9=t!IFm`NO7y
z{<mSz82*T*_!1gTK4s3}$J`YqmDRralI!O4jpAp3=;T)7!?z9coBA@T8=n9u3-c{w
z?;DJz{TTeiXMngl`MvC}sy4Xv+~+>gD4BU=BL;YYd{)9sx)Q1<edhHw1`>Hx28MkG
z?0j7Z-D`}^$kO7orRUtf?Uj~)B|2T(_3e{*b2h89POx3}xvJI~Fw{DAqp1o$#_%a*
zlsx&41rzUf@^R!ewYTlAvu#>r>*wu|aPX~3PPa-c;!!eKbK44Ls)Pu1*yA##y3E<I
zJLP8GHPTAzd^#c<tNhS9Hpzd)pFg;b0()oRbSGH~Db2gJ?USLcTP-FTfIF~+h`;5n
zz(3ZD!FBBTdiUk*nk2Q_GtO4TTh~FcTm0=q_s<0S#M=Ocg~rFs@YAQ4wHiaC9E~Z7
zDygNOM{b^RFTTI>wFmfh`I7fI=cwPDUpZd1hlK6_cdhqASN;F%7m@$}EVQ|GZ%<)m
z9z0Bk%_4L<V@~^KypB({E0tvCl=|e<O>7Snw@jQRs=If9gL|sR22{g6jaBz%+ZA@5
z^`jou`NFohp@6F||Kh<!-#Z!;iSZ960%`N@Ug4aP<pE4lx#fAZV`}dQvS~CEz?qR`
zBrj3E{5?^WD5>HhSfXy=&TM$}bJ+HrcdGpo5MS27^9+Ea??5t^7<$vjsmD61R+Yfg
z%U~M4Ygz1fQj;MYQt>FWCkHY${?v!W-MQAjh`H>q1nnW*6*4acffv1D3H6^<N8q_X
ze>Y-(w~K6#FN9<cPgaUg{0snnihl-Nm-;-uQ}n+`;Vl;yUeS7S+Ti(M)~I5!TSV=|
z7Jbe85s>kgS8NQ^9e6rb+BErdzO8=GEcohM?@Mo8);N`(Cg3~-zwqCk?7p)>KKi6%
z=*(t&NtNE{)uP08IyTb(eF!DH`VfS2m3jI4UvjKDf>-rw<kK@C!>~t<f3&^xy4%3s
z0LJeb5LI3Ag$(?^)9)F(lKtuiG7En&=GTLcU-=NFbA=E_d;tzL&=GWetZx36hLKtQ
zGeF)Xv%S}f$<ivj7^i)1bS4;f=FZF7Z^MKzEp(PNe9GqQ_hB^6R&WFx&Uxof2DPMh
zN`c2pKKZ-r;`Nn8kqA2vg?wYqBq3TCYVvT$%>?a+VckwmZGthkhFI?*gsbS)e{|lC
z3Wd%;aVI_F@m{_%wFTL00CpUv#Ai@xe06O=Qak|c_OgTqd?nYTK?te7sXDcl$+f~G
z&N}QMKWQFff<3SyLR)?$rJamE4E5&-s@ZaRZm^frF@IelW5$noI|q=Yd!@CV?F39q
zGk)7%x!O*nVa={9^$yM8<kXZT3h}ESiRhO0&iz8nqTJ`l`q%Z!uD0!NKQA6rtdU4f
z6gX8?9qF9XD=aK{sB2cHA2srCV`&~5fm9&b0(9tp&F5EFDqCUFOKM6rE_jX(>pM4!
zu_64?{s>DsK?19;DPG|}oU)4vY)gUHn+^^`@?2P-i!A9VTJg0vLRQ5{#O^@}I;MsN
zC&ZGd;hBEX6CKhhM7`l~d*N}*(ll}F()w>MvgvtddM<nOiSHAYcvJv(Fofu7rPK#c
zF9d?tZqu`aom3B(GHN-R8uOwPezM?p8w8kJR_jV31(m@*Ch87(I?DtUE-CKDt@X_s
zRq-tr|0wI5;h1>4EsJayj9~+cLZftA`6fq;iY~o2AdW4x{GrHvO5RQjqN_QkXb)FC
z0h&$Oe<BKtD@$V^>>}3E(9@hNB3Ygd*^Cks$<PhIhFJ{QWkWctAEi4eDtOGOC$Hb%
z$jd5?%uV90zs5ylC+=M)57BRHFE4s`yT2{?44~#YzBg<;Ne{rCCEe6&dD3UJF6LR5
zwHrWFPEV`3Jgoi&Q$NFpuDZPPLyJUXI6|HWHZ0EafO&a4r)OLYLCq6p<cn}67az~C
z^Wkp&xapG(pOgHe+wGv@dZVIjCC+z|G+-mqYaJZ9K9z=`aJ^9j$5bzxSVc$x>4ZYr
zS|j*N24$qbl|ZsIyqc;bhCZf{wiszC)s={UbL-`8WM;=vl1tvq0+?kf$$veIk-P(w
zw3G%3OEMxb*z^cI``&8BZ$;Q5IkYrt;AK*lu_Rzin%xdzyY&-q-VS8JhgL6OH(lIG
zRbz&4%G5AJ+5teOb%}0Bb9Krp2Dw5XR^W9!GOXprL!qwaoX`}P|0JvxdSF2VOXA7O
z?D?{7kN1;(8vD}C0isoygQAwqUj8imhMYW6V2$<>P2Hkk=lY{YMnQ}G0PrDIn09#W
zOw%KVi4pQ-ojxMgkw0x`9N6dl)8m5FrbpTA$+Qt9riWsln@lhqBcxXjt3_%gB+>aQ
z!e_+GMS!5HR*z#5=LR_y-W@qrtAwnRbwtP$u}pX$3@q)5TxuX9Iij<5Ms<2;kuDJt
z<wpij>@V1PE)2{GD$WrsV=%V<zG(6jmD-oZHZNn$R~>2dv!rO28|sHUd2z}4Q4h2q
z&|?j7N>3`!fEayF&vf7-kG1WQ^k=tJFUR?4W6ix}7b$IiFW&}Ygo{z}2UN*h#U}~R
z!!u={TTbXY3PkQ15EPJm{Pe{p?et4Q&SPLy-z>o*L$M`=4O`&VrSt^i50EbL33M*H
z$>w2_*STG1W@SF4Qcqh$c?7n34=yvkPu@y#MFO;b0WA=tEU(Vg=*HRr^w~TsB=5}&
zykr)?)$`q^UwwE6Af3~J`W_RWEI6~bv5{4Rx5j{9Bh<}g62;Is_~BBw5Ceu;t>puu
zJ!LwAD@*zJMEk>GvrirjuT4vHBY#{>=&RC>z*j87Ie=Uc&VLSk!t-6{C9{}hpGOyf
zD6!)^QoowcAxDYpYwZGiarm~{-p)S*2J~v5#(f^RFT!q{wymB46HT_geOCd`fVIp+
zx0Z-n@`zl=t6t2}Hx;oKRNza=yW;JP6ktZ(RTIyax$W4*`uW5R9``pnOhWCMMNx52
z?MDh?AFH8>j}I-Eh}wKK=33_twc|BGh|*@^DCox;oXf;af@4p)K(C(0;vdUSYPdvN
z!j|sLEcPv3sWw}|NW3+=E)++PMoSBJz>h$DJC<BBkE}We63VrVooY+7e2?W=LBYhL
zy<oKpQijfY>LkzD5y5bVg4N1|A6<2`tAJDb>NDMxt&5er7Xc3UYTe||fC!!A9j!(Y
zU|k(K-ONvM@elJGf$N8?sIjuqATP7}`4^l{K3s>nZxkEr@O<oyt;#6MlXA!TAcu&D
zc6M%UK1Mrwf_r6bJ~r;IettK}FZ7vA&y~5(7Rd4WVmfEE$Z`z}92by8#bP32RR+X3
zBg+J2HFamvr6@4GP&Pf7OBFbOE$4naVmVKWlcyxNZ{~}dm2Xoem0#gDrq@wFES-Gp
zYG7B5WQlD=jOKF-3QDox=nk+0w2*b)hA*_dU*=k{Ll@I2vf@<T-K5N-=kgc_w=@R3
zt+t3to_y0pO9D0u5G0Y<_H}U%N!0dVA@TP$$I5KX39%qzZ<Dn$d&-3ge6`-oiO2-d
zcmi41)*SmT#m4T^W=nV#Ii~Kt<4Cm{cu^(79cqMh+a*yy^mbqQ!tiXgGb3Ve*`_l;
zvSNDzN$P$kSNOqSCbAQ!puVroK{;s&A>+{v&I1iOC^_qZj%;juikEKr8GnTgylO<n
zeie!LKwra!j)q!w;bk~{iBhbiYDm8`V-%Fdupor%xon+99pFoZ1dG}PYKnGAFS+y%
z&c#^EFYS}o!c)$f+&Ag)#{kgwt}bdm$)Z+xJSt9ko%5ie)PF`oC#Gc?9&jLuH1{R4
zfjMa^-_gz9Vto~%C?}vVpBkjef$m6I94TgA`;f=oey1_@-BX?nolrKSZ?aEbT9()-
z?Dc}djGt{cet4u^+xI=;wnqJ$y1J$ErKE;3`k=4T{i%XF_OpNWv|h8C!)hYI*g6^n
zO{@OWI#;zhQNLP2x9ojg8orAuvDtE9nZ;m6t)-qjaq<4P+@CbpCr4uY9$x3Tir{Nw
zm!QoBE!5)F#QX17p)Bz;4ZtC7F$!Y)%43a3+5H(&qobEz<>8&PNAf)%$4<p1o&nkL
zx>h`NzZ@!%g0W3^coSu=W-b;r^}4>DE16Urqvs{0+v**htde%k03v@%AL~{-U!s^D
zl2w|}LsEE|8taE+2pE@SnJy$c>Mh9XF^I;bfWfBsA5AQt%yW9oEh<6hMq50o&gJ8{
zy1!c4*Bo7H=CD{L4!wI0J({NPSYwUZlU_F4%)i**L*>JiA<UkhE{H|#NgT~!7i=m>
zDYx3lTOuZ6sSXXMT9(aCl@^NqzG8NRu|(N@mmPvK@3~x`Qo&;_{oov7DQfeY6P%IU
z#yW831`J3W1UArYw!N{#QP?oj*^9Bhj=5tBgC1{4)w9nDFa_Go?3N~^swyG~nYJoF
z1LnMvHwxV5bgIL4Ky8?kViTUDtS{Z>r4g~*lCUL`iRR=3+ayMx6bcv$@04hD9k{Ef
zP3SoKkxC9Je`KbwJ$&>?k9HoypSI+{A+NTonE&eh^@1BoX%d7FXE^7E@s?^$$C-Ky
z4!iOiUPm7{q<pa{a25aK7<8^$#tIh^+R`uwXV^XOco-tVHga?c-T$BbaGel^orh08
z9NWzHzP@)WbMe03(b&UBr|HajT0Oapu!iX4Hh)jXmnT;8?00BC>qJZo_FP$b7MDT#
z4AsR&{WwRZ;Lu+w0BLI}@EQoy2ct>mR<vXu?T51?jBHd~HaHb%O8Pea<v1JnFdpSd
zT}=DQ>7j+b!lA3WAkauJI^tA4nNMDv{3eCYuYg?s$UotDsUa<@zL@j2C{hBK+F8iY
zl`qa<CJG<h0Cq#^VfaX34?C&%BGF#Wpq&l($M%fDp*_qOodX)^zGLHKlz2t1;1&NK
zds$i#qqaNAycHcji?xkpPD%<28%oeV+sCuHA|k!0aY%pVNN9|*qQ0^g!B+s0TTr+r
znn1MEQ<AeJ^0ZgOLo0ziX6P1Cm}NV<-&IRQ1BN8VyI*be@Dc&bXdqwf&so1xV9an8
ze;ug4)H{RZNs_+puTJiV{v_*Law9Hu?sq%YGpPjcu&M1ckZmb2x*Hy@YLmeslV^va
z6~^7B#8Ib}^w^-r#V;LZxg-_sE^6UIx6iBf92Sg#AbYK+@0a)3ZRaY7B?+w6wzjX?
zJiY8}67VVW+85DyD2N^KwI3?iEGkKfJBDP5;cM4rih^q#Ktqds8E-Afvra_cR7eKx
z`^cTUUDXN2%j-~nuMWsDHZzvEzmZzIqPP~^*y4qmeB9TmbL6|rWUyFJhI~&Y9R($$
z7+JI2L>ztwcDQ>)CGVwj%hN2x<=sB;P?fG#&d5p8xGZuH15P#!Y%9Sw!er*#9LWN6
zT3k*SuL_OHSh=fUwShZA-eWDZBNEKqq&Oo2>PU@*48Oi+95f}5JKSN3s*)WGxA8nJ
zKopjeUVX9Y{;b1EXE8m34ybtJ61yLc(6{Sty&}Xq)Ra#g4eV_k9eiVDtpmcx%vXFY
zpLSd>oEnMQ75(7yP{pTg^8SN8YOUsnjteY)FFX10teoCzVpJ}<tVl{Y%dQW81Vl~?
zRWoZQWaN3I7G2I7^r<;r%Bmc%RK63dP!d@Yo$d2y*Lah4ErUNG@}j3PcR)9)UY_5%
zTig`4#UGf`;TL!*^nTP-R@SH^c>6R&%$$inX8(xih!RA849qAcjcAu-+dI+`J^DsE
z^1m}}{F~R2Ebb^h%e)@#SMIW#==|KBG@ws*QxBiNaTy&;p<I>g<XCA&esT)fr0*ML
zWYA_MrhFZC?XJShkXfyW@gc?kIBR#uf0my6J;7jMA@!DO6@DX~_FX!W6yE8E>6!q2
z!}Zax4vi_h2i}9CoxK_{@`N}0;o|ZK(ru;QOj%hl`s~{~uC5;%_rEV+G$~hqUUU3v
z&z5cI6umbg>jszGF?@<gaaEp=i8Q#Rol=x$2jOfCFi~B5ql(3oqde7+47rZdm^KYY
z4CaF@f(BO0J)$HV@8Zrb1Ab)BgOhF25NvRMW_i9J+>(H?{KnHB6l>F(T18RS)qUfD
z!`?s9SM>t2l|Cje(YgAFu{-*S<XDx#l@+ThiZnItwEK2rVtxa(1-3k0UG@3c8dybf
zYi_)G7kIX*qE~Y!C`b!-x!-o0QghqF$B=LEhr#2)x2!~btXH2e;Z<o2ofb}M<!F+r
zCk+p4eZ86mXt}WG_|P!GYnG-R3Whtm(nNXGH+ie8u~lxm?Hbgb=>qF-6R03a0ZE+f
zTvBx8;?MEhFp>l`DO243EOgfCE7n6!7TvKa<7FQHiWvqwD<VzMyEr)ZR=`&35#Jet
zbFOqK#KQevG0e@vawC>E9FOnAnAn9so*srhTFAlIWzND!*`kp(CI$zv)-McMA&JT<
z_DLy)xOEi8t>&Y-a8J+bZH~DFpl;AEMD6@GOlt?mab4Jd70z6{-5=q?|K(jEbHFwq
zqetQ_%W-IdHxo;hOAw<WSC+Is&UDafLW*WXiuTOJm^7G)eJUb5W6-n5M)wepswx{-
z`~@zzL!a9ZfgB!(lS6W~#f!o!bcSg;=TG{1O^M`UiY#h+A9&n;1W~$5t#<<8n#)Dc
zcKru|t+8vTNy#t6Ue4)>G|<~7)}1wkM@dnXC4{`5sr{LqRtgUuZvY6%hutFOXrYx}
z_y^F7i%%@3OT<sZe!k6$3H*Yhsp+6^M1`K@inzCbtx`a{@J)xX$tbxtk1=c@{QG<J
zQ&c-47!tH-#n1ao^t;aVGIe|%DpHKawA0*rQb6z|-h%Im1ppvS7Jtkf!lyB4JNEMs
zwKOenksxeX_x#uU%03k`|Esh1h6_Cp59p(!B`_)S(grQ<``kbj^=2yT(MS8otzvJ%
z$ezg<7<+x_c)x)+Xt7=8@IlyCS(7P)rX)8?%h*YK`%-V0Suk3A!`Ml4&tiFEwhoUg
z0P?~@S9WX2%>3E~(w^25T0E#ap#eUQOy;=e&J&qH3T!7YI@&cM7k+3zeC%s`(;`~h
zeK=ea+oQ_JSO`W+l6euiF1BWbK5o*uJ1IY@)bu7zX3jxUC8euIlAAap;=MsE3ka@b
z)Dj;)pesaga45#P+~{~X&lsig3Q=qP7rQ5I?-0KUw>;u%v_c{o4F*W;a5bRYPT_~M
zPJxarrGMV;q54dIY^-?PfQHm+|GG6wI-yJC*o#vkA2r!*V$?LLTZ7UCyO;`TH7&v2
zP5TCu5|C~uaLvJZhgScH@u~Ha8bzsgrK&Q;DjnRMTy#ARcVFGn(OOA?**=GDI?f>Q
zZc$W{bn-w%+e09VS$pxT>9Egt))|va`F7Sd10QsFlAq>wt>qUBFl3_xPQ6Y1DpBT%
zS*3Z=XBh;mO9~bgzAU|~3w1s2%dtvFRzAr@N%aoFS4-*w+JoPJxa3XlN901g;Jum}
zd-_ABY2a|*Fj-G`SKZdUDsl8AZ2xD)QY00-E-f~MnG0Vaq=i?aQK`W`C6T-Bdg|ir
z$fjq0zz>onPe>8t(D_vS7Iqc!Dkbw2Tw2@Qh&?S-cD6w9GNP}py1TV3GWMaLM6X~G
zAVt~Jlh>`)O-#4B)bepZUJY+)LC~19nq6pXK<8Vr<9cjYJ0DEfqxxc;^&DGnNrG5=
zt4ST5D`((57qJgv%+S+t3CZk-Noq*@3}8F`+;;hLMbd~0)g+>-cV~HQs{yY8vpoQU
zKg=g*dv|NSl$7jgHeI?zEXu?yf==s5gA^HUgPIC6A%;f*`|{2Y!*eOL%dQpdERNd9
zi?>p|K&%hD#v;rj9&S?L=90;5Bh88mt^lAo&>DyH`n>ZP0H2S{U*5-|G;Def7yK~v
z{5o~^$P1YBmtx^*@zYn*Vh;GVT&3cyQ7UL*N?5h`=0*;>P{=-kbqA^E(;UcZpU=F~
zbV*-%)uy$>tD)McUtQ#I&&#Q(h2XwD)=0H$WUQ5~g@=bYi?UnHQ1uG!G^e0s?zTwA
zO`qlc-28eZ9SZoE9fJtTovFjK0yx||y=&vPxW!XUs>x*TL-!+@JZHDE-a!wDVK1Cf
zeT_gqLbto9Mqo8pamf{F!d?C~#6*A+5AF1U^$yq18^ztT=@OkYsU6suHc-P^&M~0)
z0Xa4a67E}p*poX7fb`_oV6-N`C+FTeUsBlb)gcY+0zS<2FN_V-mG&!KOK=5LwMX!-
z$|tiq8Yv=N1ahI~=7%AIk8EsSA#5E{pg4%os=g=D*^0~8TDq-1JwM!Y)Y1wq;%3c(
zJxU6!6OatWLtLiX#}uDg&au_iCD3b|Sd2$=I1SO)(tYLRlsCiSF+zf50srXf?W<%$
z&g0o&<Z35>9kd1^e5{QDea0pnH7pg;{0=!1C^bh~)QPw`eYu^P#yU7q;X1Rh<QBH5
z>z<s*o^B+}%EC+`oZZ{occCvakKx%-$s7F913`jyXi*TKDWgwC>iXlXG0~@qS>kBa
z5Eh!a-frD+=EyNfQ(p<GPSzgXQO8JqMT41NWM*L-ha{}58wUEiKzInLFT#a$4AQHo
zxSu{Pc9--A$v@{-s`)>G65IuDdFsmUPd=4zkt4({rqPUSyt&}Uy{snSw|~2&=$vcF
zL*|RDM~P&PLoRCT{ngC1?Hx}LPcpR_xI%SY_LUlWlS{>d2!~IV8!Ymvh4#dyV=3o4
zbq8Af46l7c2%3ztjvT4Aqf~8+XxPet6Kmap(9t6VHV$ZJ-ebB;hl?S?zk+R6LD9R-
zTS2@?HslL_(COM@HvWKrheW*H?U?}?Ei?;<ERF~q`jGA4xKVUIWEt<XgNh5d)XwYj
z%968QThUpv+QimLi-(BtQj{e|j2zQnwoyi9SWJr7BZ!2Y`$vbG!bCV~nzn?@=(5-$
zX#5a<osqj>TVTq#pp`oKbE|mEut~g`0C^qwS^AWwEaCV2CL#Y=DE}`Q!2SChrp!qG
zR1})7R8+836#rfUCHZ&T|Icon(J~Iqoe<*A?k=@52Xe8+KSto|I+o~(7c4XYZcH4l
zicN7&0B#(QIbMDm+66URtd`TJi<%8Im_u6Fx*!;y#^#`^mr+MK)~%hbrr8-llY$X-
zg(^#{xvg$168#d<4p3`~ov5&iN<e~!K$$NRht6p%hZ}hft2gT&Z&{)wJ&^CGY7--q
zoR>v$R1`9C$Katef3k>~<V~7sermumxk-`XY^(xrb?GfjsF;WgdGoQ!1NLawb=pCO
z|I9PMO8-!-J}vlIa9$Mtyy9^iM88_w21&gC2;4TjAH7oS5IPWbUmmz&vAfEBnj&_e
z6FF~*J5dxwUlxH@+A41%F8}9b_Wh;E>n}xM*>C?U5oy`A${R)J8~7LsDBy2Ef9TqB
zSwUmKZG*o7k^cqgk8boo_H_KQ=P6zYG^Y5^+}Tkc4#!Z-|Cw@=6>8!CTHt>pnW)=6
z)YpGp;QcR7{mTW9g+8_aLAO52c<27lq_dX{cfbAmPeS#l^7sBtOnsCy3jb~NbN&SS
zpV)Q&6PW&yKL6`b|C>mIwHwXqYx^1hk!OEp#mnF%b5LubVsaETEnYMIvuQ7b|Egx#
z+JL5q%*&^LPRAcx8^^9yDkFW%$0nKO|D4l|@&(2b0S(t=D4>7iFJ&9#Hlyj-Ix!dF
zGSWj{F_pg`nZI>wFR@)^F8p2ha!$G~`yjrEqQSbWA;3QOWB${X*cIc>f9N1U07_$?
z6BlY_weCJ{KkpM&hNE$|zge(Ive}e`!(-l!cX8#B2fl`p3FfJ%cLV9Z-%-nkd?;xB
zE8(^sERr#LP6?O_Vr03^epzc*mW*$_iWzJTwli&D4V)-JWwPQ(iKqm}%}5WR*7_QQ
zf^(p+5e-z#?|DFWt@dafH79geXFYOmaHn}bJY<``QavZNcR)@0LhO=)xuYxGhHI<0
z^pndPQ=vA{0oUb(V#$m#?@eu^#NvjbcgIzU$in8#8};tho75Ghu8NXVn9gopHW^`|
zUF{DeEZ{>2u<)jAT5irYN+~r$?g8CeS;alxFI_Dek|MN<?IMFP3zuAep!}=(7bwL*
zw=4%NtF`u%RU!3xg@R2EH=9c}M24xk2QOx2fy!FSY9y&%;ZkcGiGzKtmQ-RC7)3>R
zOr@3*|46GBecZicgSJ;sIagPKC0?QnW9v+(?BmVKCWOR{gd+}YuPm(HZtN`%j>NXj
z_5z*U0dsF`67e`}x|8cw{pfjvva^-~&Q&w_4&~x;;;~j2@QCWP^CqX=Bb50oxJMnB
z>tUTv){8ZqQZ<fWpU0{x)uqMTl(1Lf^JwsAb>!?#J_AbgHmxIeaa8Xs%&e>|$>mAn
z2InaKhB*ku;6-OxYC|6=^m`Bm;z}=ZH3!4;4$n-+6<IjnmrXL6M!(@|r~#bX+pzML
zgH)3HKd$ePs)(U{j*sQRl$*CB{$%&q<O0AA6qUQ0W)Uo$msQ>g&tfs&%d!9Z;pB?j
zH$&}5R|T6`Z11G0Wj8l7E2>*enRU2sA6g%*;LBII>e~bDa&Q<wxlsj+tO#~n)GM%r
zRo0=WJ;UaXwf92XfAsXQ2*tn3jg+QiVFk^RO-MBG^hJdYw3!>3hQ_oqJgS2ON7B|N
zxO;!`t<1n*7bmQ~rgj4vvQYh~eTzhlOteI?(74E)htWRLTHM1cKq9A1VEHROhwy+n
z>s=pWio-#J8vP0`j*`sa;ESwc>>d;XT2{lE5rbHSo?ttKL}GISs$@!-aW1>i4XyQA
zzJ@4CEcIy}J7RScr|34_$cW9u=wRqdt0lY(2F}zg6y8msJbQ8q*6a(*G!?$scO#VL
zER^*mAXdSd$z!V`n3&|EWH@+djLC^z$6EdgF1)?9d1FV@)Nq>($}r7^S6^gd^4&el
z92}8p7Bh(BdfqP>6Mr;k3lkf~47QApWL}*`+eUH8ecLI~Qv8`~i;={D`)i~n4VF-)
zUrEJOr^M1{iTF)A?<|)5url`BqQTyGz1)3Lo>8PZH)zwYtYba$DAXV;Y1_8tm%;n?
zKI;gxU?i}vD9pPDUA}7KKxUZIh#$ZBlYNlU4jYbT2zc#YrR1NzSvU0Z^jza470$N&
z3fD2S{=&X$ctDFz76BK?@-Y+p{k+@+7dEyOc-2k+P)4slB0Hu^0_8RW3)MCnJUR<}
z9|w^S#;rA5?wd7C2~Orm#nzq);(0e|c7+njAU^UmwiwU1$HsiO%Rf2%eq4gsG!}pu
z%wI;N<LHC7nLr#wq?hT4RuF5@kBlWZ*4J`f0eyE+put{)Z2?Eo|MhTfPUqRpuXEO!
zg>Veg1#36PXFx^ex_`1?1&qAhRV!Zy4?^Dh<Ad9&44EBnbFu!Hye5V-LTc^TCd}SJ
zd=2>Avt)SnTjdO{re3CAA_77tEQM_G@d9}QMw`*dmi49fYo@Nr5Y;62tQ3_sw}LQt
zS&|Q}n)blbIWOClca1+58XlY~Wfif%#31=z3pJ#f;A1IT(zxVO%@CL6Q-7dJ2&XUv
zl?-}luA@22<y`T_bY!|fX4KUPQ?nV1ikp|iJOt(nMhL!QoE!?I71v$6qNAa3$1bQ6
zYVQ>1bITV52Jr|GE{?F!M*;HW>mw2=AlH(V-|-qE4JWR-w8|VoD88pMEFpyhzfNz%
z(r}kO_$)Wmvr<10lk&ly`mVAS+yt!ksJEMNh|{BzI-92g<x#BRz8VJ6W5HO|Tf3dK
z;esS;D=%;zRo_}23og-=c!{U>ftTtnSBux{9uWzTy=Xoof&GP*r9NqQ%#C~ZM*b?0
zr3wjy!NiKCqWbv%;BQ9KGGderb{`wel-s?^SX>z3K5Xp-eh`;UMfjQdX`fZ1gLH3M
z^L?~Py2bmkX1s(Ld$AMB*9No$CMUK)rv267scCr<)x+3v0ghE^w#mnI5<WfcdM{NO
zVrI=50lfDER?;b7UB7iZom@7AKZdiDK{~snX{p~FV92a)*GsQU8nzsF9jYya^zm?F
zJ49H&0*2IZv{&Y>x?I8aV00SSD}2f;A7~6qkQ~xcf<rN7nIv%m3-P7kno2iUu}jm8
z`l?QF5a`o^0U)>>N7ivW%_}ebqF{V+MfJV@%C&ej$8;0ab=@^1kmuW%i-*nZ$qbQ%
zK~lgU5FgY~>b~Z9t~X%4Qx!ah$??he^o~}t?vXrp3=p>{nWhw%sfj6Y)mX1#=y(@r
z3R?*+h#Qrq9`8>5EpU;qAU*p^kiR+@nKb_t$(xNFx1z%+@Vzs6Rj>fh3&<I(_}dE7
zmH{#U^1|VDgVN&M$fQX>YJ$G~YN<8H<tqDz);BWQujGSX?C!4qk{}i*T0!(%g)N&2
zs^M~h&uJcY9v#?jGgC!s4NCr|CW}Hd{*xB+)?sbW7rCl#cx>b9f^E=X)s(r$gC;J9
zVsp52^_4B~UWylS8HF>oLy($T?OR-3avhz3u&r60EOC?O)lK~^vnk?ee0i8abyQFo
zKYVXO7*+>oN0aR0x-{U%g_d%M11#XlBj6eMF)jwqxh<jdT4xLitB_=gqF?!=wym=+
zo!c|O$!QnwL9+eyKnuP?hQ%~&Xwr-_LEG?+jfw!0qcO8U%7Ot%GWSy!l@9cLh8Nr1
z%6f4T&NE;|_KNqWXfN|WbUqPYmsO{&kww0N-dh}FXSXO_HYaqE;CxvyY)>l7XwvSx
z@lMcIPGKz1+CZeZE1uNEgm>}Mp-6;keZuFSZ_!+PZ@RsVe4_OtyTHuF1R-bXg-SU&
zK|$#PR&L!6AK<jdS8<K!LduHzRK8RNb#LE2W-WCnfW094L+iyVPyB7eqWz3J%(X*D
z6Hod$vusVWx1MTy6jihIO#`={D!S>1^qZ#|1$pDLET8Xd+7f%9yvvFOH^tqu45nrJ
zu5o)J3}uQgac4KhQ@?Be`nT{8x7s@Gx2iuPRreE@e%(cdHNUS)(B8bZ^kECLPF%``
z*0extJT|Z0MY7kkbz6_=r=aTpX1gTE`CHK6HA8<PI%fMrQw<e8a7m6!1m%Xx`z`3V
zXz%Rx{J#Z3_5Tk<zgx<cm~+$rSWxgcyGN+$U*<nb{@>cB@0z$L!gz@_sP(rA5ufjy
z+mxU_Nl+frZ9Rm7`5%q*_;u|r$`%GCs`;Df=Wj5Zr%F&~CH{+-{`TAN3+_5?xcp~O
zb4Fqh_rTGaw?3|G2OsUXn(03QP)$E(g9g9_*&u<1KlWeAWDYU3_T&EYsnwrLONA^v
z)B`zxIpcTTLo4H12QyaBZ+oH9PKF};?WT8s;r{JJ6)3XbzTEj8f<L(b<~tfNfmY>Q
zTZ+>B_R=lWkO$7fSOW9yZv?T9G=JIH^d2u6s!f#UFQ`mS3{_gPEXq&<rBjtM#ow9n
zmwu?@{sO}={0Gw^hPL&eia+VY_^%cJ6GfbE`Z4{lvT=kwu<)|t9}A$q{jK*8+J6A4
zwEV;R4;@%NEllBgi7F_lDeX^1o66=lp(4G{09>_q@BcF|RN`2+;jC=oYANMCsQ7$W
zGX@d6WM&QSX))@_4Sv@Xlvu)WV}qLzR5Kmsxvx;YP{k?h8DiZF(n#F(>ognX#=0f%
zJWVH&rtrt3oiF)SBleB(>+o@M7_SN06~>-S92c;fbwzb<U@ttJ3x_mb#Z-<%o`yL-
zGJ(n3z<G$RTh~qR^TV`)IiYNHmYEeeT6q&ri$#2yoa+OrZY8I-)BMtxlN0svtt#ay
zM=N^ho0vtU;#)<!Bixw11`e00Gbc^$@R36|dPv$~Zuw~JQUh?hnzHf-k{0*s`+>al
z={bFfO1~I)3rx~mrS|DK-rt&G!H~5u)10g$1m%wNIDU=?d&b);G*)hr$;tU|R_M0(
z33uec-Gm+<i0zvc1hEJWXpY#v0xnrMT*2(B;Zqj($kHj4#f_Sz2wtmlc=KZ8(0&Q#
zSZ^nz+NjhqJ-aVT=CeroRveU$E=Z#0h+jJ~IPf+^s5Ym0mp-t?gfy(EEIo{xTkf=r
zfC9gNb=+~jyYyEAUzSsqa7jeAO6rCIlE*2i4WGvvzji_Q_G>n!%<V0{we*B+DBc0|
zpwpT2yj+h2ttw(Qt+3~q31Q5SnBaJLp!Kb3)(rVvZB*&!fbt$8mAy@=0j@OG2|MEj
zyAfOP+w`n3Z#{W4al-KQMsX!IxmoUx&SX|i8rZjV&j3q2)x6>3*K3(IwVUV460we7
zU06*J7W@0_k)$Tc(H54FG|w^4AKJK64+U;rFKd=(>z$`1`irbN&#M?xw4#G_zu=sO
z5)A!c?R|AnT+OyGAxIM3-Q9x?4k18r8DMaN1s&WiBtUQ-2yRKR8Qh)V4DQb0!6gBL
z2hN>*=iKw&JyrLsd+NPY_r0q7Pw(B`d#&zTYo>emlHXRwVkxEl1lX_}w+(5Al??;u
zTImdv#zcrTA<)R#7u2_*cSS9Lq(>|*F-^;zsG%h+eaCJ=+hZ_yKL?~IMIXDaN?)lz
zZFXcH?7FjXF!9u>vzp$&u_FWy)E$?WbJ})b9>u7k+<WiY_iHzvhpY-0?Qua9)_Z&}
z%NNH0h3>`Vq3!hw>sN|AchYoB2G&IF<5B$qN?4%^Wy^$f2rTZI#rENQkSTdZpjy12
zA-Y&#{LWM)?|gXpezT+((eh8(If(lcbA;OGb4;0>K*$EH28HZ_GvvDgL2)E~5Q%o4
zNb?)S>}cj!gF?p?r?WSM_rsRR6c2M}NUu<Y!=|r4zE{ofnZVsY`c>{1_W(bRf^K3J
zgaa!AR!YR=%jb)FB1+Vk_Y57V>)!-yOa#qUl;wYHdbxl&9f*|+7=Q!D$jI>Qn^_0M
zExEY4mRUI=$0eE7MXV6X(Wx7mImT}I&u16^cT!r=S>Hvb9l4x>K>vQD_2+iqf9U<&
zdSU$%4=!?(*U7noI*>-Q5B2$gR^vdkA&mPY_>H<;8|&v)d;(-8U+cRpJF?J$`g~$Z
z*7~k4*N!?&m)>dR@uP-1g*^w-sv~79RJYXMS3ukM&Nb}``{bmK)xKpxL0IB+_r$=K
zd9Bgp0%}EOguesL-jhGQgg88TlKGz%`C7=K(Z0m}b!sZ?GsFkKpN^CL-PVqrX$bJu
z=T#!D?@w5W9iDlip{dIX{~s2=+5SY+n5lV&3{@rg{R&l5;`s2OS>4g6`=`kO|FDcp
zMMhMkKaoMM)^y)Uyk6q<iQGiopd-_T!qCa6QcA?|3GhQCUZS#0B7Y#fx(Fnun0p#0
z0UDxkXm0)WZnaA0@4AFeJ-hml36ymO`sw#oL5OvmSFa!#oXEH6#qknpTYbs?ce)R<
zA{_2WfHdTD|BKn*8VLwcx7COKR^k2q9vJLiw2ee?``C<xklW@(V!Ab}raTGx&GDaY
zB$Jn?s-6RYLK<Bhnx1gf-rWTK-r0b^_qadqK>yT#04X%;KZIuM&H4PDpUA!7uaL5n
zUSgUhy~keU5sBwE7P4c~=6$6nIpTGv4VCM)C&zjj+&KWo^x4_=k*EUHow}C!J7Uek
z>z#Tcy;I)!>P2$4Vv_PAE0_T%tH%wbZuQE=s<M;n(k$BY(19H)Q1hxw%1;1J?X+`V
zv9IJ@{<=I2POwf^dEc6Vmn5i7d7VX-+vB<ug;(mfuiF7TwXVlK*WfNjYhR)PW7><W
zH=S;DX?OT&jtY{9`YqTwur60k>C1t3pkNC06^5C;;jZ@O<<85JAN5t=4L`Bc_lvW3
zFNzgjB4H$KR?90_Ym=ST-Vi8q*D6z)m1;mu@NlQWCb&|i-p-VmdNols1N#M(nN(CU
zCKBJl#*v#v97Kc7QNMzYYiCAEduFRgJz^|977G%+P@KSgJgN8j3W_laWQw|{szl)8
zi914H9M=0r%D8r{#W_%qoGK4UmM5C^zz-z}bAcq;2UoZ&VkA7R^B^?>^=YhkYji`#
zsf%F(MPpL9(V_^A)usTcv`QW*<i$OPjBd+AMb@Qp+-zW~ddjj6`9P6JVZ4#e%()@`
z^g-6a6d(9<(I42%-w*p)<VVFsG2#tHP|ht~wU;|TtB?>@Ov4lMLf?^aKqG|&^;oTg
zEM>fFhoxEF_8Z6)k6N4!<3=aAt!9gV&cY9>KvVj!5I@yQVX3adJs5E!&Z&sH<zuXg
zVMu;>38!+&7i;;>^evC~9-YTE^#zEgmZJ|%xBgYPkzQng-p2)tb8NxY>rM-eTWgw=
zgMuy`Y?)LlZ4qtL;w3}}4z(6@W9CReqsO4}rCKrSzBdv@oH79W{cOhP{T#6&r9juW
zt1>w6>8wv`b^SO^;)uYEzCz!>rVVzAJ_euu&@B~V<)cfCpaJSF9I#FoHUKD_goqaS
ztiYE=SF<q4MoR#zjryoWDfS2|MV3C*@c4K6V!3Qh6xD#o?`qmL&YuIqle70RZ%yQ8
zX5Wg3SN3dx%lZJXax45xe6vrinR?ypyXTlq=&Hg5^{1zG0Z&=5AtRpe=|~&v9-@;P
zX-ko*M$;lpwBFil{5<G-FGY|0Cm&I%!*45^>FkW&oUW=8d`!wo;$M9WIS43VN-1wg
zcwS!eS>W14`$yX6XH(dI7^jjnrCT)?;dyoI?;k#f4Z_8a>ba9&i}<GZmNy~O($fDm
zX$z2u;{qJWnDStt-D1;kyF;~*XACNLG9jIm_ApG}wIdN{;pa!J`rc0*HNwCv{`*(7
z0E%7=pRd}xf@*Hf{l@^Q(01Jf30a`YQ2?7qu{{sZl;;8JH?x{`U-td`>2vj<i1^@S
z*~Z$_`G`&`b%L<qG1ob(rU2zrzWUB127i0bqWDI1bsoQvFLtZ@_RU6Z{ptOst#{(e
z?Y{y@Uwy5}qEiqy1)@GNF7J(bFg^6{!=CEXwQvzGV~F>7YZGAr+56G4P#0M&!g_l#
zQx@HyC6Q2Co*Di0Cb^f|WMaA>=?<@2_yMv*f>g~A%=d5wwuNk*TsZa5i5!Wnm-F)<
zXRB8L310k}@ASXq8(qwc($l55PVt1O=WR0IQT+mG!vufH=UKgrF41NFXCPH%k()T*
z?^UtndpB(Qf#Z)8bH$OY^L7m4V@k@kAIy#^>|Di2#{aDU{S%QU%w}s%hdyWSwk9Ld
z-Dvi;iWhKc)xB7Q9E~bg1kWurPxW+7*YnQ8y;Y@y#h6t#Nfuo`p_tdsjbX$oW<kBR
zuj?nL;IU)b$UqmBJ#2s1D%8^Pgd5WTV*SwgD)3QzrMuossaeTp6$ip4_Sk9#hFPW@
zXj3jWy}_k#X>?rgp7({GapG|2Rn_xkt0Hv~wsYP6k(QR5THyV=(>UnT4RshZcVJPt
z(pE-VJZzfK1ifPO5mN+yEpdAD=gNE-BS3(XNn>wGxwc}uXR@=D^hD(#RU59LsQ{8v
z6uu%9tFAArzlH=Pz0e143nW>u=L>zThI3>1kSTdpp+2ISZ`L8m`cbi)WLfmJ^r%gk
zguDgHVMw);D#HIoao=g6k;v*-TK3?vTDV>`qp7C4F%Eul+^C-`?hz92R5xtTBSUjx
zp(HyZEvY!M`@!UG(OQ@J6~^^_G`+Oa%8$$lI6qHfohS~EUHiEGNtF?t_;@^i)AwG0
zWPzhp!Ah*$%7MQ-mUe319}zUh`P}#DgRmD#2I;gSGeqIBRd8RAD&HaDi%pls^;*1K
zQP`+_{4|x}2=g_Tso$^g^YCX+4j77)M25L6&CBeg8eDm2_ep7~btH8JwZOUl4#u|6
zwP$y$#r*xnIn$=dYJd`T5_08-LJNqx)5e{ufPmtQg<nS(Uej=HH76a81Eh7U#>#b+
z^igp86-%H0XnPvPkua~)%RRqi=KhnpL?xyJQj*poD_V`$rnH%4Dyw>p9=H)0h#&RQ
zaY2IGZks;0x+M*>fy*ugFYWg^S^ArVXJ!dW`?Tijst8{I*5z6y^u$}u^=EB3SnY%B
zjO<OBopQSB&R!m_#MQ64b(+Mu0StXbOr3%v+4t>sN&>%!RXW0x<P4!XdhORsD+jYV
za$lh<FDZmK2x?7x>ZxQBZadIlH`&uPGT%mg>YvnwXAbLY%vGo0#wFHKCG`v4!kdz|
zZc$*)f!sIIb_(3(;l68@SMM&JNquRvyx$_D;yFz31ZhA<^hrVT&dF(Xn_gi@WGSOw
z_*>PUByLtC9-!{qU>uzo6Qj8ZdpoD$?861)V;z3SQ`pGG%Zp{pI@|Eq;tfIyTukw~
ztCjSE%4GH5d$YV<pQQlEBybs1qkjHGRSS$w#e6zQgukieQ+n`2s?E3T(EU{){h$sU
z`ZZgZKmKPUD7s61=9w&yge6MvYYK@qx1na#tu{AMdZbI-84|XzMN7hrduf1c#PdvL
zSCx6Pru;{^*!LL}djHNxTh@@`9{rxCwnHjn;xUgr{x@TihYj7Q5c74X!^izY<G0^N
zf!LnM1&;26f`-gvzPQ*y5HKF?DkpSRAo@1iR#sz!sMroWF$0q)Y$<!$)xN$%tMs}O
zx|0=^kk8ZXHf`<dCG_#uZwx2UqQ7el9Qs~56wH}IzN<GdxT0^YcFcIHz>b3n?@4g4
zU-p9uehjiWDU_sidgM0k7C`eNNaZ1)O-`B15?UI<LH!6Pppykm6l9Vk-ro{brY9c3
zJ8}&2dMnHkR_5th5$&!@;p&vHw7J!rb;YIhV^%jys_ABrVL(@1M!?92&<aDjzBTfP
z)Aw8M20lR@)0BY;U0}3@3X`X826SNlb?=q8Hy6!rOHRG{7f4yLw2iI2sYEvAF#p)Y
z09#!7;$~%Q2Y2nA6v_K_2?F0G(;FUVz3`}NpiXqA0jH`Bq;EIl9jHYXEyj0IZF!h0
z0a`A6eakup1Sx28LsT}`>8wh6y-|Eicr-0zy2fX&3!(3U3K~Q9qghf`xf4&!-RyDD
z)*_!>yVz^2!}{z#)4m2M_kEw|g$~~X!Nw6$FLRs8vX)wVot;$!Va1S{wi_$n=BV3Y
z)wfJlp_|XzM{IMV2CVGSaTH!uKqe0(%@op2Z!b9_GcrR01@_w7c`uwacFDcFDB{kQ
zQ36a<mv%%z2niI)7c_08J}gNQWLV58HElzu)NZ|_ngI##=|1bt)7J76UFZoD?R&=z
z?ubmg)mJId@HS-+6eLQf4n1Be4wacb7I__=N6%JKnooSe?ZullaD7aj8FnvsU+R0s
zo~@=kNEv0y*qt*piH>6o<D@TwB`M`P7!L<3d|F7oB3z}j)0;e*-sRV=)e#WUR~+%1
zzn?2T>ud8exQ|YThA(c;we?C-@-g7vq#WnnljbT@j;g)UPBPkJ4R6q(Eg$UI1FdKr
zWDzhijOg^JGUaC+d(fK9@70%g|8m(v<X4d8t|hU;_6ht--hLY_g>k51Q{VDynb2Lo
z;pW3`TzJ}2T7M;Ye)|FPVu~T&Nvo|%ME{Z|2#(GVO|cDiqtT(I4JM_MY03!A4PkpV
zAD!XZmS&tfhSGZQ;$9C>8tCVv*Y~Y6Gu57)a@Q3SXV0*8fyR75KaO!hhJRsKO|I4l
z@{<3VGBYU)Bz7AS)8qxG-z}UTiJ&`~rN9bCCV&lb6-6*!a4}ysxJ>EMSE-zm#b51$
z2o2W46a}#6>%~I{7#az-&=MchR-L;8*#%x)TR@6R=8!+2epV3v#I1_YtKu_pS@dLc
zU_kC1pdq{Ya_Vv4=pBj5!GhaN((H_c8W2Z=-sDq747Sf}u{x(xzYIu=93H7jM8eUJ
zQ)!w57(%G5;>;!fb387Xh9Tj~$s-FY)cSQmAhK%0dBj>?dzfmQA=7b%w)TQbMCr77
zy+l2iU|4=mYqkv}Hu{T{f5A6&s|y1}%CekKIW!0}WeWX_J98{uzQnsOMbGIO5DTIX
z9VJn<EO?^{7PHO#@k0~iAyP*un(|m37bPa<l$47G*FYUo7ORS7Z?`bXG<rhoo0W%!
zM!{?_bkv_I2aJEqqGTaCFKr9}yjLGhVLx~G%QnabbkV+K!-qdEPcpz?n<4rsn=Y?s
zsMZc(1PtnoOt`quttrtId}>#)ptZ?$=-?stA{4csxrBQQr5>Q<J+^jHXv&8LPmQVq
zQO>0eTe7;wc&B8pJ`I2A({)y%v|HrPUN!#a=^yGV6k#uN5Ap@GdK(qR%aU;<*4QDI
z4>otP=xn>t&p3rr7%Q0QqYcV-Mbyft)Z;2Ixm127ZwyXst@EEdU}gVU*GONI2x@pD
zz+U}xkxhah6rWzJL%wKHCBiM$c=v4J)4wAc{C-)Jt(5?|LIvyfpK|>DoU2>%cauR1
z%74k%@<-?i>gCo_WF&jn@*frg&`R`2n%Z~i?-D_#3n8Jcpdyv9`cr_2x}>;z&*z{t
zZ{rJ6&ZA4HeD8>Q51w*^J-c<dit-`-_N#$90W;9CJ+lMp){@Tc>yOjAQ#a9Td|D>E
z3ZoYWs`Z4c)S_kG?E7XrYps@S%R(2}F&T!5p4AT6SvmG7uGLPRN9z?E{W{Y#a)OlW
z*A{2Ttx*@ec(mu8Zu*fQqrT1b8O2#MgMKD!Ydnp%Azyt|T}{eGVpwO#)k|xm0yD#Y
zg=>*Kmn0y|?be9zD3R(!nt|Az<`~he&#KmvE^&zX>f`Z=_#|?`&q(2pY5|XS){1rd
zQ!3=+cxrVw2L-pL=X@ZqU9D#lH@6g)P}DsCUUu(w3iexU@*z<eN4hcb%TYAMB_b(c
zkhxSRBwt8a$jEFE200rLcAMGqu(~XZ<{Zxx_-f-pf-QOXIz+Wa5U8=mpDq4Wh*YY3
zjHzX-j#)wn*;h2}|Ft~Z)CUJ}Bu&;xdy%GHs?{+%2%4EQFa#1O4y~v`3gc<kV@QHw
z3)f2};^OuWTnM|&>-DT`u~@6D!jW0$?p2zHlsy6?3YdaAG4W_oTyRx&2%ydzBiLLz
z22Zt?pI9^Qha8uh<@2dGY(EH%Z}Mhsf%3g7i>tL;0zEmEuQ<`=b>?Jwn`upn8@@I=
zxJpX!HhM3>#dg#QpHf@1bRjgK8+P7G(HwF{O=D^DJ{(fN=4pbWOL~zAzCr=ZpE*pR
zR6TyL?;I}Ds*`V(E%DTjMOUkcG2>82(7k`k{Bz;Zc#;zcypxU5Uipy}pU(p9YXTct
zUmu~=R~6PJ)E=;oR`Elom(<mrR&QA9DG-kpz5T+-G>xuU$L=<ZKdJyUQyq#*Jlo8g
z37}Md)EOEy-G8vz;b`TqyGZGf{>{^mab6eYjQ0hKaci|TP@!=5RbOBC`mHk9u7GG*
zi%uPoT=L>^1ir6fendzUl~MW|wsMkY(yxW(j;B%mQI0E2T8&PvWwlDF{01x4vAy$O
zTawsne5MsucFB@qv0t-peJ@L6<2kXDVFfAE0xE{s)zWPHJj|qdb@)p9PzWsHc~p!g
z$K_Y)H14BpOCVS7ejS;Ah@yyUp7WU~PduW~3%}6wq5>y-rbb^w4bqnMR-d+$*HMDI
z23jtWk(HsSrdUM!Ha`xQ%TXYVeiR3hJ~PYiZ`G0erAf=hq!}Nd*-}wmGb46k<W+x`
z#^0h4o_Yy*037SHC&TjI$SW3cEZc;e@E!T(fusQCMuq7ust2>z<8+FPU7Ucg0wq4W
zs^~jzu~B@7f@#-sn(qyDb<@GQ>NcD#5#G~%Eo=%ykk6;-*+yqr;pOJZt2DiI=<U+W
z`5_wMf|5+_LGpc;kIiFv^?yN0tQlgA4&7c``H*0<$%xGpys3nk2WWp^#S&4XY_0oI
z)84;fcPj7qN-Q-j5fX%6e#ADtufGQ&Ks!6JEHW6$U4h%M1g1=Q#B6-S%IjQSJw6dR
z{fYjv)NZqsiaeD=HT@04rd;VfsVj>Ogq1$t6dsup&V{8OCGeF^lRGKAeoB+I5r4AN
ztP^*;Lh&iBN=+v0!Yi>Q{UEMtfN~;$S0V12iaVOi20WFOv6jOPwX-fNtAM7$&Yxva
z?!0=3&4Yv5b1cGktRHV>cAYu@_R9f~KwS(N>P98{u7otD&UJojSXFBUsItQQtcF3&
zq}4gR2=cKuV>>RMk1PqUzq;Yf)YGhEgZS{xrYzUKaF&VL(fXl$sdGZ<RH_4%F3%AW
z<ESO4cI)jCN^N9(c$%QxG~E-HuvB3h8Lni%C8&E)>=wYXV|pdk$kuSBzZb?I;XKju
z#V>nNCWxBZ+$(c)!_vulCe(SV3rEc<vnn&Sx-xOZ>;0t7!nFZ+CZpEeo}C@@d!AlI
zDKG)-E}-`m@4FRtk_76!an*yvF#&PSz{Y-!D}X``QeR0JwB41)RP##|O5SjD;NLjT
z9iGefs`HMPw;rJxwN7%zpz5!pKRu{*!xmU%ui7aUONt`&mZZT9T{)8=y25TdLdedm
zNz@<=BF|xgOPTxZvkuX&Rd`mFk{q?s3O!`gvECaMGO145@_x{o2qTJT*jHp`^@NKt
zQs*h_c)u!|Yul{o^CrY_-%<_~vD|J3kByfME&VWyq4@ljf!TpQ^^5w}LH)=u_9^X}
zctI)}rr>})KUx!K^*hao&)OC6=;nP}LUA#U+}HCcI9+J?F00R4<vGBrFP0oyN@|%3
zlJeD0FiR-8QS_A}wIZYJ30|^%W{*sT5HlJPBCDzapiW(K(UN_}WSL&&-(d$O$(Xl)
zHFJFW2dBiJSAoUHS-*OawqKae<>-UOgDglE7E9#0jm*KDPbEqv9SD7GpkZLlh~sA=
zX5knqw2iin-a<e9k?14oUZ{qahL-W_pzB|6OneHvpKO2lD=BGPRSCOnB{v&8&m>0n
zr|eIS@|5l&?jbAq_CrsHusmgp+5hvd|47xH`+Iyb3r%hw2u52~c2$Es<yp6T_aZ)c
z_~|g#o8P1Sy$k3`d6t=7RDL2cwbhOsxHy$~Md?30@h8N8)`d@r`ca;E@&4CuTN3oQ
zQK{p%QR$CFga2v~k|N{yhtXZ9OZ#tfF9{uaX9x-Hv6jsH3yRompPJSD{c6od$>kQ<
zo5xN(rO};0U67M4ntgv%D{w+A#2|ReWH`a7H@K%a%tgH#On&f8>}({BB);Ta_s6rh
z0%WM0bIn(qlqLa~ROn#CsRi(<C;818E_Xg$RSv3nJ&b^3US5;p(u&BYML<O<7?8Ni
zGow4=Y|)XX6j8@E?R*^@ENu#RAekGb&gd?cD^UN&5TVr16YPl0M+wHQQ~~j`QE7!C
zwPd{FMGd*^4#+*WJ1H%#e8ee#Q(2Pckm~Z%o0De&`8^69K^kuj_C2nGe9yir9FSy@
z2+)O2`@|`7YfpAxrt&!xlQ7mn4%O$K$ASy;^8LJ*5toFMy<t}RGb3uqBXOT<QXVn@
zZ-{sdJp9Hh^9LSS)4v}V0`wc2vJD*qDeLe;Fu$koXUdhG1!6@}^R3s^IC1yArd6Rl
zI`UANepD2|A!kew)zt|JXv)GmecX)xE#UAeEAV<+>r$nAhk}lR;GA+rBSGKhlW1a<
zaJyddIi8Jpyx|uo`Vz<F>Pe4BnW9}zaMGMNYZDC-g`T6snjoQ>8d;%p5W#kb|01(J
zxUQ}B{if2!JFIfkX`|c3e!X2oAy+LyIus{%3osS83$An{kzMDH@3IFRBCh}qDo8MC
zITn*-1uCp343mU(L5IfY&r$=tWP=Ac?N|y+Gj5b)`M;812!Fs@S^MQvDWTfvA(WLR
zl37^1eO|&49xrAxHTP<B8p!^*7Sx2hiZyNPsyPhWEZKlL5^k0FZRrR{giOtg)ETW`
ze(E|tDc*Gd7&vrx`!tJqS|%O-B*w}QTql`eG>5j_Zd(<Q8tSSb0Ua5AD~7666bonK
z<()F}U5KZz6w#Tw2D{dYmPB8Pa~g22^E!I-?NWs<_b4|m)wJr{zf*@BTuG-YAA~;6
z(ZpoO?|jid$b2d-V~OG6*%Mh>T=-VXRDPO&DqN@L8-D?6fWn>Pqds7ScMs4Eq)fvD
z&Qemf&Qq0lV#q9C4jO-1*a|4SOA^g@ocR%si8YMT$lpH~K>dWr^?`FL*;+h-%h-6d
zAZv6@w?>EV`Eox+x4K><)}`;K!P%2c2S+JZEkzw6uW&#QwG$=cH7|Npl`dyv8L>wn
z_D3-?sfh-^4{n&5He~kbmr_?p9hUI%T6f_r?=&2cX+lPwOV~u3wuEXZdfDZ{);on1
zx%wL23+~)d7R?vCIs%y%IDE*cHfHiJY1l>NR?rNEIL`ELEeU!5*5PJSPPu22(%jyo
z`6-uoR3vGGJFw^J^ZNQyqp|oClvBG>y$iyhkmz8f21}Bvuyf&?M+`pM@ucoQA806>
z*7|QdSUmE&WF%t4pB$*O$P>YJ&{~AU+vA&B0~Ny(E!l6=VxjRBh72Wk@nutTSo8tK
zkZwTxeEdA=jB}+!;{$3+U8S&q(op1wjZ=9O#@H;(uWKL7B51ecA}p=V%e+Au104>O
zFbA}v<SU~$ab?_-r9Cg`SMLw{ogq3I44Iq*W+Ew0EOI6G%^@#n&0zUjC@W(dy>7Ax
zP!VTAwv;R?726~wlN|~_-fc3GQz4m}&4UkFhj_GJ3)%<C+JjjdSCKSBZ{nSc>N=_j
zB}j7l-l)bQ#zvm)EmO>wu+wjQ)T3Yuz*Xy(gR7*r^<ayZUmR|5ENDF#(7lj}`382A
z%Ui{bQ=$47WX~+hlJJDZio%r_iwMzG4Z9Fue8u4nLxK|PR+n-<Qqtlja@>xAp5J35
zP|r|iF!4n%OHJNszA18&;?dB_laoXH`O^)6KhIwiL%3@;NAF6FI;dH{Q@Zqg-GR$!
zpXm)SGhD%Z9WIoh_rQ?fhYoN-7;|RHie-(VP@YoraueCl4&mNiK8!om=O1`8=HtjE
zr{*y{EwlROB$MW`4W@M5O_ZsTm?ix=mn${)tJS8+Q(N!mN<hU)yFt}x^z9M)^`(gq
zG=D*iTUW>#Zp##2k(#T*z_ss%`C|9LrL^=^==zclKV&ljy0GstYfutCNN%m`IiazD
zu<{FZ^6-+VE<h~!lk^7!^o0-3Y*ER9c;pRkz`dDx%;Ap@(tN!+wU2C74kBb2P4;U%
z7YVoUk0YPkAD>PxZ+Lm+EluqR;FUx-y;!V1-`ckNGV9$@-H9*=y^wX0orz_6Ev-_V
znQlg8!5jQTROy`lLP+S;YqYb`eNds5U(lR&{3(L{6P`fY^N3eQj_W&@Arh&0!qavh
zgaBK;N6cE&0`%AQLgIu<ykul(qA|W)El%-6Z$BG1E(%M&sv5W&M5839K6jUM`87I(
zvofPb9hl5f7O65fuz72Xt4i%SKX_%NX~EfgHTF!*5bID#>2gL?c$9vL?z?neX7!k(
ze{7+;Ve!y22gfi;&sj*iSM#X#b=BG|GRj;}hHWa^|F$ee-j9R93u$`k7jpO}<O&H*
zu<HAc#QpQSDbhaPKY!l8Zxx?1t8Ue<y-o$id)a#;wFRJ}i_FZUW0Aj6-VBSw;dS)y
z8)SA{=dA%5_(vbn%8>!}rgt*9f$i1HDO6#ZJX;o9oAqVOtIl}@PodLy0N-Lv^_xse
zV;6;!ke&x%!?o87&vOhp>ZGS?yaXdmzbr+(MH_o@w>?r;7Xn3|>kjgJ2EI3@R``~o
zOz<Alo|z>0of;ib1>vP((m|&-R5bN!;jtM-I=*VoJ7Xoc#z*f+jA(+*cx<x=Y^-vf
zorv>VizX&Gh9l`BPrWPHy3l{M%XP!$&~*a0)&vX9gX_ixs^Hfb0`qoGedC2XX+b@T
zwWrJRU&6oW&7~U;6*<>TGnL^jKnJVA^p#mmaU+1t68n<ZTP)HI;pzpc!EU{zsCeHh
z%xYHd$rKw6&Ft-|@OTU00bfNhQH_JT6|E<XCoX8?^SO*;Vc|x88>LA}Wa$XYfXvYs
zVk^PDW)YuFEE`Xx>#Wuj+b=*`eO#lGnt?zM%u?9gg^Lxq+^l*8GB#~$tDbUS>_%rX
z?L^a&3GbO!rq9CPJBv-Y^mKj28jVBLq`0(VfiU`xzwWdi%9gvPo8=*jc<V@l8#QAC
z9YoWBj6S+mrpLv39L2fw_IpI}z|^F)a>rENlLIu3Z-l4bIB(M!j8ACFA%ojF5V&>z
z=fID=pCFaRgdtPMMiB(PzP{W%`YO#gl*-FUXiCL`%v6I6W`*&Dq=||C96cNW62s;r
z#gOwS3r^@kDdlad&h+~et94f*)MI-1)yWApX_6PwlO=%e^VvbBHRd40k0+iZ9x}8=
zxb#>t_n*zjl}9G@6{Ix)4f&}%3g}|-1We3x%%1*{%$HOKhNi%u2gT%<psh0@&y>b=
zrM<m!stnewIPprTJtmL^!<Md%gJnr5E50d`0UB31)oG&ah790YQU;*-gUVJ^I(yIr
z=p^@T0;{8KrrJj4Urp@Rj~t_+K3h0SzGq12Y^2{1eiW_baAS2humU9<8>a1+;*7)S
zO{GSNm6RpsRw)CyTNozb6ijc-A9<Khbgg+(l*En36m`{#-o{J}^h=A>k&zuVB%*98
zGh_LbdA7^;$SIS^h1nxRu_SBlU)$Iu^qUNuJaY5c*YdWf(ulzr+8l)DP$$lub=Ads
z7bUF`1WiC^oFozFvSxSwd`oT(cBUK!)V0K^7jiD0ArrF9w^~m6Go#JQz(K(o6`}qu
zZlRMvt8_lPGE8ur6qliRR^w(-w2yehnTHF0kzbOuVO{3((u&B|J9LfYVmGa6@l4KL
z@|%zDB08cEn=2hZ%Wru)KE0bRJ@|FgZp=kdoOUgC`SY~CLzfEo^A^}DZejCEbVioK
z@l;yi`rFsED&si}1fGQ<PkHm^ih)~lNK#2g*mrYlIichNh})_txAFqo?8FQ<5)f40
znn>vEFDPX6$!qte4>9}=KirRhDZ|AJbYBg=3FMtDsE>E^PKHOJ<x1k~l#cab=;cxG
zjM+fhR5Qnc4QLCw511hh&=nzkP)ka!xSt_)l2}!obqkN1MgOmt%Y4z5WtI8RkP>o3
zl?QY1CND2DzHf-+v;Ju(l_l|FO^jE7R7DAMX|p-20{WHyhx$lUQ8+bxo7<=So*}6i
z@iA%otKaG^KXrDoz?V^mIgGWJ6*Azi-7jAp!={EQ(g_wQRyB*e%=2msvuHHcdDNUU
zpPKRsy;^7HMH6kEYi7MV2IbXmoeL%RfpoB7BR*hF=#TzH^fpWqdpO>&l*4a#yigx6
zbkC&P^lLOXk9u{R2zY=4GIK~$&nTIRMprF0S8tdq0aFM;@{Qxxf585L-b;z$^bBbM
zWaycn^`W|jSNXbPO;|i1@0%hg-^rl8QrZ<R5CQ%q)QpG%j^>ibY^5|^49b+yhLIqh
zs}}ju;)BaTt+^#-qtQpZ9hxB(&Bp&}N+8F-pRQnv4~%BWm@v>k6B8yctAYoA{udOp
zYjlwI)L!-#e@i{p(xvV<SSqX7NwK&w{PY7EJ2a|@@xkQILJ{!JM|wfll!Vst;pA;q
z_ToL_&I0|7u3zn4^JTY(?Ul#z2NDQyXZ&P$`RQoh%6J35-aX+%m|~VpE1e@7;GH*;
z+VDS88$VvpRCLvC{@+=6#(!)4ti7vw0$+YQmba47fDgJSUir-z*?{DWY|vBq%{1>f
z)8AA22c~~MrT@=Kb>9aP_B3&tEVd)1FIF3X6#cUXnl5gMkitJ?L-dEV=}RpT#1GL4
z02icuG#pov`U<Jxl>9?wVt)wx*#PwX50(9)uP0oL`G2VFH+h~TeHoJUw`eu~Q1!ot
z6u$hoQ4<=7viM~-254@(pW0rX-N!v}mP`TgH_tm;=IJa~{;6oaM__JC0|eY#)Sw)K
zf`ShDZ-#LHm*Xk$rm3<d{)>x0w~PHx@4sBh|0Cl65%K>c3z>WN8psEocCWn0>hvQC
zbyN<ufBgYHR+7diOg!4Y7Z-8?80~Q9$t5u~jSI84`j6kw_L`1vin`J4i68LFKfqj+
zZkDQu>6+~V24)n>)cf)dUD1AYSe)qnG__bUB;QF&#);2Jj2SSlj6e;#A~IKxphd^D
zR_HHn>1nCX?huRU?-?O;E)NWfU-ym2Odv7L(OVUe#=s_(s(>bB;Rh9+Qw5lqZuMcB
zDfF)&JRFq_P<xV#KXvHI?!hu<aAm3s;zOr9>uL?G1DRFr#R{sJ$897Tey3ZcZ6pk>
z`LcEDW#+4{!ATJcCy%K+5D$y{dD?o76=U+F16`%IjH)7d%t?ccjy7>hBP~%Ft1;56
z>AfS8?i<mCj=v7z^R&}}=(0+`GQS6Nos(Bx@}Vz+&v@Et-?VL?tEjxVD0t?qJ`Vq&
z8ohDLT`Pe3A7b?XTdHykk)grPCQ<=DCn`0j9cGCI=bAn|kT!sYGi?>d*3BjJtXO7v
zn|ggMp%`MIXQhmQI02lyicS627?sC7Koorqh&g^9G?An-NKq$9AnmIl$<h*$65mT&
z<$i1_N=IduKnSyXgQ!dM(!=D?=VE>{^ZmJ4TQm8*aO0z|wT>s_iVn9V_<=VWP{g7#
z1-B@54tGb}j!fci+F?Ucf|g;nC@!FRvuxk)yzVb3wTR>9#FU@dALUKaQs;IF=OHdf
z$%|X&UlZi%1h%T~B50C|D(<5`rc9_Us;W!$w~Ak~t^z>D@d(t2A)0oGN8mz9>!;)s
zNdEhyt)*wTApF2XP1TeL9Dwkh4jVlb5}%&FRuJBy|EOqvdG?qQ;pC+-(+XMCf1|Nl
z8W{e$w!TJtz^HSrd)V6^wdW_+KAP<LsK*sL_>kIr>?NPz%OL?O>;lwzp|n$sZ~Kj>
iug|~uE+M>S*q4^XqxGe|Nu%<)c_{;VTeRu_viNT~OKNuj

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/BgC79Zt2fPs-2170.jpg b/assets/images/impact/talks/BgC79Zt2fPs-2170.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..a9dcbd3d9594b74094b7f419014c240dc56253d3
GIT binary patch
literal 54045
zcmdSAbyQo=yDuDEO0nW!oCGZnEf(AoBoGL+Xem%A4n>N4(c;nq#U%uX;_g}~?gfgp
z!QJKZJNG^7zJJ`i?z-naXWjMgWY06nJbUJo+0QeX{mf?`<{nl7q)=suG5`Yu0Kj;B
z0Uj0s3IJSeY#eNCTpS!6TwGi{0%8IJe0&1(r%wrqDaom*D9I@(sOebfscD#KDJU2?
z7?@buKp+qmJ=b$iAP)-=1pHSe7>}0{;1iG$5Rd_BC}@EH$HzklfD9j#7OMjbgB5^D
zhJi(f@z4Wccm#=wiS;kM|9;@%;^SaoV__0J7QZD0U|?ZmU}9t8VB+Hu5@M5JU;?nP
zamaA-$XNvOpUG*nQkcaHIR{e0+1zS!<f&d(Rf~NHqJH7})&(e>KqJ}-N)qXs2<Fro
zSI`Uj&XJqft!tUcuBePKcMt7ZJ3M-9gYL2MM*#mX1ORMIEF28n$1+Z`f5Cf%fQf<g
z|KP*G#3I9fZ2TFHwp=HxnUHgQ)%e2#fC%eRlMIUtAOjfOB+pBL{hGUh{knWd<Xd#$
zICZx>B0g++_#aKAVNo~dNBp1O^e_0G|Lgp>YV97Ye)m7VqTld={hDWbcF{%~CjKbu
zRC?|XKADX^ioN(F8Sk|(AO0mQ9<?st{nsfof$lICdujyC_hh{PbqW)^NOK(6#YWt$
z5Orrfibb~Heiz*We;&p8z-3nJW3T*6R6lC@{8uVJ+4%S|<$rYCe@e&P7achB_2u38
ztt~zPW~lG>Upw=4B>TT|;8@&#{3OHH^RjcxSh{Vj0@$#7H{>n-JwCx@T`^x>nrM#y
z7s36T>$VuRQ+Y3ngIe~I`>#f9$j`uY{ULw-KTUFfXG+bh+uoeI8&p7KgM%cSUu;ey
zR8$Nye=ELO|Gn3Mwi5k@JkCh%@}YzEm~s(E*_gQ5o)%l0noR_GxB5ol2L+c}c!kyZ
zyMFYax#K@}HJplh0CaJ2GCBIRDk*Jqa|TJA{1m)uRxGwoIfnFq@^0GOE}e#^aV9}i
zoJzg%a$9z5V2&BGRpi$}%D$bW-c8eyZEe-Nkzyz|T(^+SN(t}tBboFTwguP572dvC
z9Xi6GL?9%NA|p1if14>8^~cbrM9COO_xogCgON@D$`O;9`{(UxOi@G7D4{>ZTmb4N
z0ohw*o7<p<7MU1tiPZ79AyVu{zYQ%#QB&JM=L+ljRzs##@Ny#*t+vK|;(P=yJ~q?@
z<(#lp>&i#utnC%5=BpY|1yoZc2;oa-Iqhe>AENd>UN?lej5|t9aOIE}+N@@Oh3H8r
zYh~W=0;%4;c{lJ5*Gg;S+r5KKcHOkX+maT#?WqYG^dK&?9M}qQzT5R{Wd6-NyGmgR
z_q_pOn}2ZYts}c%f6&;^?m&eT5af3yNF*O;uDtM32{dzlCoT4<Db}8JRb}K@z2J83
z`?mT<-!*tTlw#7gG_yK0E;9kLmqH#aCR8)>d4G(cq~++vw1Rx_El57F3$MQka5LbN
z64f^6N!JrK)2_cWne3_>T2!b?9i7B0-G3DuZf5rG5%5&=5>DZyv9sWHP(mhT2%}$L
z9$J-EPb6u#-D2LxmhgQ6oW_l##aiH!E!&7W<VB^Ev1%pe%`j%6oP{#3!<j@CQ!Q9d
zeSw!JGQ;Y8zc{?^q#6@4#Mutw39S!5mlzTc6dc1G`dLX#j30~Fz}K7Y8CvO}{{YyY
z?wWU=ImkclS8>(^BmA}zfuX)3L^qZps8O-an72~qhUPg@#;De*UkFx~$V{zJH()+?
zVDJ<~r*y_Mbbc>s2UW4QjLJ-7nSP$1!NS&)wK}I!x_#?5Fe9xw8}D6Ju1#ZFAo2_V
z43fp*(v$p1D#sA+{t5T-PQCjtNSvj+viWD}9{`tK4}jL8Z7~u`PXd>x+Yf+~!PxRt
zEaa_`rqOm~*4|~S+K4{^(P-dYG203Mpc`qZe;?bumeUpY-KOtm*ut8!isnb*<CIw)
zO___TKX-@zge65xH-FnV@hFIVu+GER8N!whw!0F$vXW}mT#BMMU`)4_yHTx5`W8wA
zakOL{2CCu!?*@YykSYu+>)z>Hl3&Yi@T%!1>716C#773KacP+tAv!f*-&~IRawf-0
zl@$mJK8(3i!;#Q`E#g#qeia+xwX*OMV*jL6mmA)>pP6}io2Bv=`loVos$3+3qJ=R&
z`aWh{;&TbjTYJhYx0m2=qU4J?_0u>kLMTX5>>F|?8Tj910v-cj2z8J$Oxv4F`vmN3
zy=&}iJY?szP7hn0QPt-a(gz6Tfia{x#P`B%tE_**T<Qn)r&v-bhVm0bt+49Nc2#Ba
zW}P=Jw5CEvwAgSJkrf@5*kMk&trJEXt-8Zn34C9YlHJ9*`W?fE)1&nEuSQK|G*1v0
zHaQw&>J5_^$z36%u|F6d0L^@Kky0MS?sZ(Eg4m@#iOTS7KWs6Z>QG(EnsKC=)<8bf
z&oy+4DB1IdnP1yO)K5Gc?*~Bl0!0siz7J}lw@VKI^Wz5qjm`sL9jtF#yCQn)VOwd)
zX_a!aqXSnU`X=&~0U`_O|8YdT$S<^5ch9qM-YSt>Om6Ug3qJt`NcfsaqXh=WsW|O<
zxs?ZOzYwy1Pw?XUEhG>-Yc=s)I*eP3?T%{i`>E@A4%`5xs;Ek^%_`&rIO;Gv={BSs
zc33yW**_9PR+L6S(_{l;a)DZh3}J5=IHE4>$9x|E5UK~jBB_d@U4=0S<;gWUsd1u;
zp@p@@#CbHl@9Dp|OCbfaM3yxFob0!cRGg-L2UQ2E<^>Uq1!DWSFmMbj&)OCgZT!e>
zD-cPUbfzpxf&I-<MiBq{(?E0aT+%1dX-o6v42plYT)h%M4@@Q><>d>FAPFKRl=8jr
z5He1bH|5{UK%AY`H>sI)n!>;S-73z94MGCvRiWOL9&B~aVM;UHXz_J_$;>sE`kyu_
zp6c%DV1$viFp|_OCCEpLFt)Kv<&Du=z00Sn!tLaAiVJRNlb4-xINC}vG7T>MDse)S
z;NC28w~Ku9Y0;O4ri9OkrBLgNii#v9HOKt={2woUC;{551Xz0<81o)`2k|0|A`{#D
zFY$Bc_Ngg!mUNlr!{mdK$D`knO33v$@N0KZiFmsY@;`TZdOXTk%W*$cKm3}MLXaq=
zB)_Bz>o9nwSEbRLUu1f!o<Uz5nTrZS;n<M{J}1Y?ULzdE2MX+cz+*e`TH&$1uFK@I
z(O3vLh3m3_aV_~2BPc3J$;m0B*EhX*SIoAgX<xLa+|+5=rFcbtpPU}gO%^Pa+E`_b
zlkL!>APJ`>@ofq4nX>+RUaRh(>2bETQ%~N1`yO^r6fJpMHS`Y-U%&a!?-V-UVon3K
zY*5dpBE^VRwHf-HlF3!YusRYLUM9F}+N~{HFFE`JDtS{-Y+!cXGhQI14b@Z`p1@4V
zrjWe31>f(VO<h}k^?i5dZSdZ{UX9-MCt}FUTro3V#f`DM4dc=+tmee5<u%JXVve;U
zAK9>kViD{nDmI9xqns(ocosN^aWqU&Vp5XUZd!+!Lh^b3T}x5I2q-{_BslLaHLRyz
z>;=Pri9FHBts;e+lq5g3A%eV)$8e6u^o|+Q59=K;NWCt$xaTjthw#y#;N2A+$sDCU
z0Fpw(`pyPgVyVT(a&4YzmrJD4OcUDV!irR(@~INZ(V1~1j>PL#zG`jLLm6_7www#+
zUK(lrJ|z?hrxS%B`P}^AXn-7pY|I8dSy)woKRVFxR_*~1nR>KU)GvDvhTXFfY($ew
z<B>2mdfZxy_1~@DBOIAiFQ>0kW9i2bFoiAQ&o(wik(u~;ghth+AbdvDO@cR|n}tH+
zC%ob-Hw_oKlL?bi!a0<ATxw)*A-VpNN6C$=L_@-#kTOIygT&6Q)K6(-R$61io&v6?
zXYa17KnG18*9%47Lm<GcN<5N75mBSCp&KjiK^QN%E>Gme5x0{C6Xye9H-`XkPY?XX
z@V3P<_3p-`P0x-pq9iThZ<Dp|mxP!vU2<5PSklt`=?+)E=_BBE&9qvk%3^!gaxo$Z
z4!LU|w5ci!%{cGI&rR+1#5d8dR!>i7Vy4r>s9HfPLOvlTGh;P=jv~&FB&To%S#K&L
zX|I%7D2`6HOiAGS>5`mO2&BZ?m#A3|^INPxhKa2l@Ju)!+NAZC`xitaJ3uLA{!CbV
zD_8(bDYXOJboUE2ohIvl4wSlbW5%=*5RXroA-Zf6hd=;9SKwt3{{HnZ9Y@_4%LXUK
z`Wlp=t@#hhA0|h#LzCi?&Hz46T5^}}m(2((zQwwm%s<H2#lX^{+Sl6_nP`eNEw^h6
zR{-9J1mzER<XsZ$d<Hq|emIpnIvoz_ls;#Ii)GF}!^!U~0R<RKY%;~*utsX0GvWh0
zti5T&%>{|(0z~y|=>!@wdRIvQ>X6DvMZ_qsfHdE;$xD~=yiaIPVWaYr`x?XT)YlZU
zx5b|*N#TL}<;$jy&S5W&J;|dITN-vf7T$*F`VFOU54k=7_BjX`Ah)sMH&33fr1IVJ
zmtM#m3EeyZZY`S}(?9z$ZTmJm7WPovf?Xjo1z1lw^_+y)=OGX#;s*eA>+wz9&dfgZ
z?0Z|snKXm&-*J3gsCD_tTkbI@O@#u_z~#SJ%H_j-bO>F2o2G>+jn|@rJCV$BvJ578
zwA|VZ0jg&GUwG5#Bp(1|yS0OHJ)!JvX5?PU(OtD~ytAk+8?Kthi{FDXGZ!nTpm9((
zgurJW=?-K`<;}06EVF_dM;|53%@}S2i7MIzCF}xD`88VLITqIP2%@<#nmK7|*I+&X
zYla#@m>B!rsu_QH$i7(c{cP01w04<(Nd!{l`vgv;EJhav-kSc1w-4Y$R6jp2y0C0h
zTCT}f0L80`IMq4LF9h*V4SA_-&d8n<GEv8u5{-0>jQr;(-NE}uYH?G_SCu4=<*7Hn
z^im_~qe;%r9{~5u{1**U=xMY0bAvOB`t#fG405+;9)4Eh{?-tkt-VtQhF=jlDmC-t
z6O<GQ*Qb>m(b26vHjyc8z+rW3SY(nD<7=T2_mB)J>&$aI>))h-bJ7>HxoOXVq)9QW
zx5j+vmQF<pt{fwj%qOLwq^#kkHf@(S$orU0r@R>acUrovqM?*|OP8%;m(~4y)_a_O
zYlG`1ZJCk|<45barLG>=AqXC!QbpJg+Q;d~10dG9PFG2^Fb&%rB@f&?bc-~ONzvV+
zbo(LjmkcAdB-lFBaL*T~zfkN=oK8pk-ypoTv4xFayyIYjt)sHo*;0#dDw`)#O5FUf
zjbf9UDb?jXxd5?M=xsCI2KD#c@E8CS^)t%z-D5#T#bHysPm0|_0o?LQ1pW0j_4B(y
znYy#x+M(srZJ9y%aOdM&{C&w=lMuPrZ{x|wz}$pk2Jrl3qRZ%;kJ20eY+4}$I3O%6
zJes(=G2JvDioJXD3MGBpUM5r4mQ9226O%vHiE6VWo`X2SgJJ;x`&7m~ZElTU)rA82
zojNqHAdH0fDgH)np(LEWPs>F5KEF2@kZ+1qW%iE!e(ufnBHgbq(ldbzU0BHerAvUS
zCHjo<?&f@yNdfeG`pbtZ#hKkmFXMKNssr~A@8+yO+Y`hKeF;IuZfQEXB#<<Ek0$#(
zSuZffJhMB>&HuEbg2Nyr!CdrNpfBT30T%!x@f>gmj3CcZHIR?XmQToulRLt>3wRt$
z{BT0_DxS9(HP`R)ZyyJqffdxb^|%HvU9GxzM;Y(v!TG&*tqZ@a3nh4Vi__XDV9e?$
z$dGfWPr$R)3_Lfq4CXS&>MJGx6Z7Bx_tg4#*Z#NlE3nl2ml~hl1uK~5cZxO4!nU^_
z0H5(W3CHyO3D{`+Jy=h(0v!?$pR<l#RXeLZ0G?#s>^=al9KG7;CxElO3%^U-&k@!6
zGw{wcp$7mS7ckk)&Ehh%XKlF&)=>|+q@p9-W9@cXK)z@~v4OUh$%4}IrnUacZ4aRF
z^6x<(V?E9q-n;W2|K$DWDOqVQd7&9;vi2={lo&2|!S6lchx1rXWvJ(*W!0;lQg@9>
zjmSTR(;NjJvU<gw7POiG)d4Z|A#;2dy;)ClL3`xEi!UO-m}*gf{9C4afGGU0{xCvK
zO?}QGW#<eMfQ!i+m<EG7_ot8AB5E{npe$Th`lmPbZ2dfMU1t0I%z^}8d++HKI;!LZ
zjrCmYyJ0VCTFDW5-ODTGx@Ow?GcMx+AUwS*!Pi>5Z2wZS7jJ5Me)9~Of2J1;@q~<$
z%YrAe=r1Kn9)C803}#o`XSz<tb@tzBSU2ua{{n<Qs_kf%I5Yb6hWK)L-DceFZt<^-
zqu~$JU34$v1<1%R<6E0!QuHNpe@&HZ$*SR_jdX8hWZW9dxcxR3j@x;^<8CWgk(tIu
z1f?WbnF#t24U%QtXpV)p;WQx*pR>{V^ZMv$2)$D-qMJ&L`)X5Y0NWxwG>VXO%&#5q
zR+7`e0X--98%d)@8a{al_$WR5?-v@E439DFQ#U&5^N1izzZvVGQVN_3ExnV0?t>R>
zwBCE(QQz<2r|CZu2A|^CyjLPMQZt3+EpDPmg`9*+EpCl#9-%ICUs92s?{v~p-MPNB
zZOyT6c>Wfe21zC?*o)0#NOC@iZk?&<YbsBFQ(QT>Cpe)sougWe0Tt3Dgu)C4x4{vN
z;F?9^v5AU6hero_DqU9m`g=2wO7bbkgyAZDDqKA>5l4l4KNm9SCFxI$et|MvW_bEd
z_q2YBW}lR`P@BFZ!JDH5|41f`hT+@fzXLM8ox9BTzu|*D0KhW$hIQNY_Q=1GS2Mf1
zDQ5!``a*I>B}|yL<fXn}-5<@I_tD@&_1)K>!r~FCWw|r9KWJVBUwX3r-R{?4W0S_}
zozdRRiexj;6V-xhI-v#!WWMaUggxHwN9XD`HLqN#EwOVS7HvygQKDW8`i8-UeVO(B
zuBiE-&FjB=1HJH{{M#q4S%GHV(oX;8(_utcotIRP@5ds%RYSbhS59LYOo4Vx|A{IU
z%e<j}=zlf&e>eBW?gdF_r5_Q^e^5&69l_!wWh$r-z5(uB8wYPp^eM;_=YJP&7LNZq
zFUwp5jdW`9C8H@G0DKmlP<1o$OXTV;o(IX-7{^!PZg+W7Z(H|dn3A=MX>Y9`XYO}v
zaGA64$IQ*MBdMfs&bq&l_XV!)I@s0_S`2GKJei5BZ~}(G{+gp&Yl;ior-kJyErm_B
z=2M?P!p$H;w?nFc!0?)kj09!87J~KHOR-Kx-z0NUbRYFuGAT$UDS7IQ3D>LjU>rs}
zF%Ah{D*6`p3Z47>OO8%4pzC@s&JgD*==W1D<6Jf|wmL$HmN+B!$8%q9X<%G=8R8-h
zfb!SI%>*%NyWk^O0b<Np$Bg&P86L#?)KTL%hQo>bet*OZ1t-6&oNVS6&+BN#Mf~*s
z$WsE^t=-#n-%sO8Dn`1%<R1VzS&UhCpKG04r5oRfqMG<}>zGs_<Xlsv<&7YS7Dt2Z
zX&;3*50dORgI!M&0Kea^F;O*(Ps2rBbBza5H(y#e$oFEPryA*+V}_T{zm`RGv(^_?
zo>^B2jj+(xjpEURMMxh^Q>safe1~97YPhh=(0>{7-}ai~FZeiSwI|r?Lo4*-wrr*>
zraU@!jqs;TZuu4Qo*h>U3#G`i7=%#IH@2G?C>)Y~>qoqkZJMfX!`T#g77!1(iNj~x
zml^qee0i33HY3^P_#{20xtrm-S^2)8szqNz-{h;6etEV4`LlVpQ?E@{s3=aZs(cE$
zSBhZFRma%E`4ecOh>c)Cu9s@RzGKv$oyg2~lqIFH2+AmeKrx;uOBET1>w~d()goYM
z?UO+(XsB;J&26(44?QCdpPdkps5%m3vi3{K=T$}(3k6b;T?YWXV*lQJ->FC8fWTTm
zXkofMVCGd1@eAxW)OrlzpfrjLK`Y;NN1#fJ26Hf9(6ZGzZK!zehqdkQyt5L`{?D3X
z(%9j?Bf{-?_on0jJku?Vi}glW;LQuH|8<7;i@=(3-moa~k1C@IxLvIlcUz7dE^rbg
zhod5f2VR;Qt;;tYR`+y_dG}sIR%OVzCF5Il{rz_AZP5`&Qj}|`_##gLA$A}%KEvCp
zcmqqqOXSj4YGjfw-M`{8gd`cPo0o$CE6@NZAH(i<MaYILv03ifX-Jh3CqzcbJd_*S
zE^neBEwD#9NnD)u+!2<r1Ahxw-gMs9f19Toy}ovtW#h<^GN={eZ@5pR?3||f>@fsm
zqVUOm^2ad4`mv{-l1Sy#rgm~Jp0NqGDowwY+FPZKnV(<Bfz9?l5s>}Z0@>}oqv3%%
zzm`P?%8iA&oY<Yvk9)DVPnoV@wRXqx3btvqZx?qWbHyJ3on%WsGS3F6{L~fu3ij+p
zzjd(`9Um!Z7E|81yr!rM&{rpUoZCv5wizHnXRx!FQm+J@)ye#j02csle^lB(BoeCy
z{)Kd#V~NlAPrg4s7QI-6mT8ScJioapR#%A^X(F=1$NXzn?@%qx3p>S4+>Xf;x?E;(
zbyj&_SxEp~(w*tXs9|)Ow7U@ZorjmOFqW{u|Lw%=Yv9)}wEw?IGUSHuE$oc6KSmDN
z*HYW2<&poOfa8y2lB{iO<8iaK0S=Fj%=l{xk4(8<Uzr=dOxJapL~*H$Qq+*@6CPz)
zxHAXg^KEpjw^?q{i22RimuHUij_B)h1H!k+PR+zeRmEM{`VJ&zw%clM-?neeS=fuL
zXu~Mx&}XESv7eNUvA#A25i@=_+}BwEHAXH3Ym!EU_|ic1T;#@rTMhmHZkrxIhElMl
zKAzOy7w)X|@!YFc4x|z07%RZ^NVPa3gs>Ac+yKtE2g^N6S;N)N4(a9n_H|L;jt|3&
z|GrOBRneDsDKKx`&kw*Af`z-5^qp+|qfpQDZq6?4%q=&AfE!&r@Fe1KK?M;X8~Y>H
zo&12ug?as{h9PE|jxnZH3`tR>W~k5nK?W0%4?|41EA=o#(>9APW?c$*7DNTaU`>eT
zSfTBX@_dqRQG3|i9#=ag@&3L^RsKyG1g<@$OOZIC;+hnHEyHK9B{>t^=hZeYrzQu|
z7<cW;A<t1FfUXJgWaE%HFM>jb#u!kZ{NFtOv8l6vsa3Bv2POJGK0{GnSSYxH1R_7i
z0suDz9&OLItQ%!urOiggxpgxOGv*cYkkbTAuaIDHM+Z1I5GN#KjDBrv6u$YS)~2E2
zRrXz&0!!3RxpC`<z2EtP@hkZj0m^EpHJ(vR`U6e(5q?u>`<GnXIvOr#gt5Q(le>eX
z$&R*}Y>C3%_D;O3okyF&*LM=Rh0!4(e3)+H^Gg=2r3|K;D$tyXgy(Q}g>kHYQ(>~3
zUK-&hs!s6IB|9^wB7m7!{CxZ770U^Kn+fwvNfOfK5?)45<Opg9)<e-B+st>?^B8kC
z+azqX+s-SlUde{Q_s&#6ax&pL#x=1{vTUpOrw;&u7Rze5j*^Cde(jJ}Ql!M%=@#E|
z0>PE{M|UwlgpcI=c9*i6Bzx+qt_JcfJvs|^)$IXj;`9tjlPX?+hd{%D_gQd^om#7%
zXckP9MA2GxJHhgXh2TPFL`st$1ltO8og3PAljdwPvL6H&u`8XC{J=+#ZKzGTJbig>
zbfREC=|HGkCkZrzcV>WOl|hLz2NCkHIYzBX9R(E_16V}6n+Srh=$sGYR6PWc0J^Bx
z9{?im;Is?I<{AT=GuxghG8Re|L0+p)kRoP~w>i>$&!g?t#^X|?wn@KelUfj&l#>^J
zk}~Wyn>h2OS%0S8<@Kk#{*yh@$sB51F2x|%XkI5u>|*Sjz_^^l&i*LmvHP%gy{BsR
zGHO$38p120qgp{(E?SJEvIiEwdu8!1r!voex=jBPJ*^(~`MH}sB@eQXD$WVkCUm}^
z-Yqk)oOi3EaS!m|$Ho1)@0pm`DD$Ri+}Sc^{%PT~;3u`vo*+v-C&9$P3~~VWI&Jry
z_3e1`LL*&ltmlhCB@JUI7y@KZcBLN&b-`c`o={fCmu3k$93|Xr^J{gkVE<`5bsp~E
zN-4ssO=PV<r6w^n5d!=Jpv1m;7Tt6{HY9adpv7U=ZgZ|`_dZ^_mQ|pjE0i8kO{JSf
zD$%?eAJ+}~`m;nlve)r9(vvFZ8S5u63k`-BD#%8}oFcepIzFp1mBO(8(J*exO7{%T
zTP+H{Kl;e`gpjXR(cUoyI6N{;snR1t+8H`Mw+9L>=z8zzbNt=u6!MA%XNx2jW^$^G
zi_WevSU4LKo@(~ysQIQiH=ogm(nZcv(gEaeWFy(Emdb)It{*cI2~p+1jMK%l5B8Im
z^ChnK3j^tl)5<}W;e)3rasWp@xs#LpCjx8_M&hEHtN4Y|jMOT0%AR}XfevA>XdynE
zDY7vfG7?tOBoB=NFk&^(J{J=G#vid$pZ=+|_zE+Ukf*cwmAX~|vzHhQa;2;MVS0Z*
zn%gI@U8&E@-sFcU=FYf!L);t1AffWwkjpq9;Z)>WxkdKYKN`OFylrZ(r^4X{MG4w0
zXt+!A2FBcN_wB@~pX~#nif8P~#&q-NZwI1yDXU>BZcHM*ACso}$S|1@O=M7ew|emH
zVK=W;nuN8!NC}kfGe`{{gDFc3Y%u@q^TV)_$4j|PsbWfpRZeKV)L2yuu0?=XlyA=L
z<88U$GO(|p`P;iCzO)eaJKnLVvvWJKI`(fs-i4TOsHh0q<c-@fL_S{o23uD*^kYPR
z5s%iLM8F-*yMM%V5<L^&2m8jxG;EJSoZ7uuZBUW5{y6IRgndGI7I6>`CIK4IjKfm<
z2t5R$@OofI66_mOZbpdW-M~?<VC<o{0>t4T1TTM5t%>lwN#)BhEqh%-rOJttZYx!6
zFP2XN3)AY!VteeO&!mz^Z`?4??Pav;f6hq|C5`L3mKUM~S4#4c&*E1BOz(VNS?GwB
zM0jSJK-Y)M&Y?8O{h6ep$Z&E<beUjsfdaS-)^Yjf!ta-<xX{5+nx)3}84f)qnLgiF
zt^)algt!p;)x?#p=fHDkleM9j>3A<}(v(Cfpmo2Pg<!yN?PQKiSl5TZOT4NJUq6$r
z?ba73az)#E{)^3)y)4?32~cLpArD|9F~~U%K+8b{>etweQ4Nbc1MPlT%x}nf)=v;h
z9zHe-^NI5#W@|jIcyScsY;v(|GHGfCpNN=t)FRi}INzsmKHL|8*|o-gZN9^u>>&NM
z_rYrlF)fnV)$%4O!JBMXC6I}<%dyJ%nNPLhenfaz+Jg1AqQyowFUza3Hlf+YAAGIv
z%A04oAO$&rZsbYPnMwJHfk(S%*Cf}U%O3z#I{v^^)f0x-b(79#opgImpUQ&_B$Adb
zkckCO3O}W-`;=jBK0Y)bkEq3Gk9+rXIb9ca?32lOosA0O%E)=SRsRmCYC1tWIA~lJ
z`yHR_w>C{?PpWxk*G{~rNOBVuG*WEuP9`AGhkLciFjXFsj7Ftg`KQ>Rb{Zyn?Xzdm
zmIAq`&6uW8l96?>&Mp<Zi$@2_0QF@oSE@BLN^V4ny!aNLkIKx&x}}8K1xCj&Vanm*
zVY~CccigmH*8f6xFJ%lS*+qHDAk}jS&mvnzcp66qlbPO;4~t!VYs5tN`S?qBT7OEh
zY1n27XD<M$cEymqk-PUW+>|zNe{ZBx?m%^I>$S80E`3IEyjZ(vD!j~?nUjfZv6<j&
zrX*yrWxZ94$85Zoj_@(2Wh^u6ips5GyM{otcAcC6w-FvOPqq*)nKh>RuDLObF5d^y
zdIiAnqagV{?|gSXxx;P(rJpZH^_Alrx6_8o)JpYw#RD|i@kxEqnqE#`?bJOV#d<!3
zvAg)*x1~p%E#S=W9edd{7)gPo`$4iCVgss!{U5^A=3jnGwBGWQ_`d%F!?Jz8TvbWl
zT+i&AkZRCZvZZq{$5L$l9PXQ@Uv~pu4*<HkAAihB1Y{BDe!6F~z~MeGt*j4=$Bdiv
zjJ%!;)yqX;MlH$_7PUHFKcR2}b)nMy{F1dP4B8tShvJr<ZTL)~_}Su=Nbf6Q9K`~l
z4^;Se{zF^icY_zhl%nTjh75?N0>=Cg#NAjuq(1n}K_s6TIUUnIp3XRdcnb$L<ZgWA
zG{3x!Nh&O)?Svr<24mJ4JjKNa?j#pj57PLPmUI7T?@Xj84PBF#7RYn?ibeK!Fnx(_
zUk?5A{v@jKHDC2lgT2{*CRntgs2t8n5UG<FVKdUV0dOPvhw{XsVpwZ5T^_|%f=-|Y
zQ^IBAxi}bPgX9x4=NTDS#jFPRgUwoO^%wVkjVn2Qb(4=|Ll9cxf953A!jKJ2$Vohx
z&?{Dt5j`<VSE>*>HLXP`f2o;M5ON{ABF<kwCK9$!f8KH0Tt4(o65edbT__5vn<&rk
zqW)=trtNcbz5sVw$S1**^NXdK8P_H}+izGG+av!x0B}C_fA+9AW>0mPA6K;2F``n$
zg=ys#Sjt3U$%IK0j~=mw)%)o;|9!FS{rPMg2w#vN*Kuv&bX8+bDL7>V5_c;6_UB+d
zvJ$sLq_<>v49r;D8{$I*G)RW&H;Bs<gp{E!@`0Uwl2p$U#nN9t8vc5wI~X@NGxSBX
zC4U)FxO_<OtMJ=?vHrub#np5AyMYS_zHx288U=_byyR^W8YQ0;lMGVKz?8hrTMkJW
z6<ujeNsPJty;x(!)r;uc#5UDXB14mpG<3!KUCNF^4tAx~vL}x9r5BOyWgefY=0!tD
zObj@sp~$r(ZmiN0W{-W61uUX)D-x%6@u_3csP}EZ-gEJuzCfI1cTi@j<fwr>Ioh7>
zZolw@ZQUaxz0V^os}}9JlC|Z|hM@WB`4@v6jkf``dkNia^F1y6Bh;Tq4?L#aZ=~GI
zaz`+W6ge)%wc`i_20O8fmVzkoT_*9be+*l<xU??DhQ<_98k!0FdgYI}tTL~1FtjuE
z&kyhcXRhAveEgGsA1L~?mFmV#JVhb|WKtcjZnY)eC;ee>mx0e;FuinfYq;RGXqsW~
z4Bhm^!Z^8ZQVu>mXydDraR=a|Yb<>VE#nv+%Cjc<vfZ+_b8tq|pp04^`>hwLFq0Dn
zJRcdP69-k@mwGa&h~#B~h+|tYMG@9>I947NrW@A46hJ<%sL$-!s-%dS$}Dn{PcFX+
z{a=c@IkD!ZN1YR_#ksMnZg4}xddL{@-B(0S=e`}>7Q8$j^!|l?(A|_3d&d~-|Gg54
zZnG8D2s&X%xp8RNp{y?EjNJ(kq+|l<3g}$L;Vo?abXj{0!e@9pEPkB%<1RA(q%4h1
zAEx!}<dmc7CoBS@>Fp!y=Iin0K+ln9D&0-!4XSBTFUsrlEBHj^S9uIE=ER>m2fw>n
zX=lX_-l7&o3>mI#$y5+dxsipsJ@~Tl#KZuInCmiLQiY)k-@bvb1%+z;yv(6nPbdN6
zB=Ce>Ra&}3Ao*2`-<)Ur$7K(LJGJmV3IwE%v#=cfI+#(e6Fmsv5C>!=zgaaqDgK~&
zReJms_w!JSjToqjlyjfrBLTO*#UJbf2sZ<`QjU!!B(Th7=^rM}l~bUBWsu=>6BKHs
z6Owx9PSB^aet9E~v0g#PD|l7Bfhd#X(x3$?u3~)gZtclfsd8-jB<jHTNLgA$j@{*X
zm3_7i!~zdV=n6^6-mR|%2z7rLzP+KQu%lY!KXX7GfxAcyWaOQKGppnNMZDYHCOok<
z#23<acIc&m>=z_6cArIq{UzLrMog(gON!-O=e1}CuM<}o$3TZ9vF;Sad|P`u$-fgM
zr6zTjSYSp|ln{bsuET8_AeZ<BUVR&5#h1nzK2Vc^{m?ARk$mZAtupf$=AdNkkiZED
zhx91t3C8>22ysW<jVIWUe9D@v(obX)Unzp(Y62Jk$PdU^k2N1$eJd->*^Ch`M|~sJ
zQ4mXkq&Oup!)UoVR`J~b3H2;%>LD!nZT}UYksGAIbT`>{q$06nREdg&V(TvMCk^hT
z1eF5XGFFR$?pk<&OR~(h3*a9<kqB;hy)7+P`!JlsZ|(P5Cbzob%2(T8M<v=2##(GO
z#blGOiXVoM2{9cv#jo|GoP>N{Sw8unMOgYO1lkPGd2bvgm8-zO;o=Cgsfj5ybW$-^
zJ{JcKbAS-BfJ&RUMTtO^RO=CgDfxD?NGcR*)o8@=Y?K+;PfCD>VlZ1V49eIh&3|DO
zznU$KH!uz=<6;+cu;gJ|L<z?pstxoTY}gYo4oW1yYjovNL;$$O<1e$5!4p+iOpZ>f
z7iP7yC2tyE*-w9;U{~eldYdAezrn>V%iQ99QX^$MD;i0Gq}{5Be=mE)%$ENEkO)4m
zw()!Wu`a;&`na?iCl@Qya_IXra>yZu4K(1Q-wB-At-JbT{lhQo3r(XY_^35SvW@|I
ziXsl~66fWoo6WdEl<b{C<?!}*vj@=jdb7;?sc7Qc`W)_TVO5N{q<k*0IN_0Mr=}{)
zMET1DfedpwVEv7KkxP}1vib=ixnz<N^a^bn<^n~WH$TET5?%3rVH~Vm{wd0`a>TMg
zx1S)%f~QDKj0J*iVcv+>J6S$_0;*fEnzbpFI74wGKZBJzh1o!aG}<%<RdKMOSGMAL
zAM0G0QoL->3PTLn64r@q^8s%O$aN{=127m@*BkiPN4yj>p7#IM56aKMyWci0d4GK8
zq=GemwAHGk2KGT{Ps6~_5$j}6kcGP#LkHF>lVeIll*MvGJA9@Ll}-HwicW^V`%);3
z?$9bvApfw^UAfz3R^QD4#t{)@W))kPCyx_&WV|!7I4^7#t(!PqJLHoV0I|J+C@ueV
zet{f0MY8P6e1N~~S{uu6?mN0HFXdnMsJfMRx*;m!T&E<((=)0J;FRvx&9`*c4e$XZ
zrdPhP{APdC9paIH;Q(+QH5JJ^$y_(1R+)C~-!bnmnljb&zk&7Zu8-=h=TA>MJB*d>
z()H?YCy?oLnr2sEErV-TiHV7Y1GaUx(-{&+{NmYXy<<))mn=2xO|vtUGz0X=L);mH
zw0r~Zk7<Mc8qRX>7-WvWn>_fA&>8ec$B25V8q18dMMI{1gxBZJJ;qnpHJ#Q_eumXL
z?1bMJ4_<73<wOmD*u=Kwt5)y;8y?tJziHggWu7mYRdczNm{OrrAVP+ly5I-EuM60T
zn1Sy*^k&Hyw@5?Kad}g}|2Kj16rCU~3n)c$zE6HkwtN8IeRC+U)z>{gY=a+#{oZ8x
z%8_Vbbg`^EcIJ0-zln@nlWw&&@TA!a=fy8X^CUt~%S=~q!Q`kb|3cGALKd$L*Tp-b
zs|sx4^)e%L%$b$jb^{V4uIq%60xaN=*<?fCnZOW;P9+Uqgzk7JHm4Bec_aap92Ww-
z1kiu@VZ`IGIjoMgEo1e|+QoWD-UZh?={fNxF#3F-#Ce~{l9KxCXNQov&kU*Wa-d)`
zFf=!HDwYPV4NoW_a=!|cNq<7}g;95=JcGeh$GjQ7MCKVUB0!V8!XgK^<Y?&9*$4Oc
zPIwFh7ajnK!%rC$ZDK6dU|Qcxdb|>N&99OeH0y^sSmw~g%c7`ZLs3z!5|eh#+{%2c
zDLA<#*&+i0aJ^fa*k|KT$NZ6gmz$Gcr|_;>k2|y26_Mx~4jAO(23&M9RGrgPQMJ+D
z4~a7UPIElOsx{&%-uq4|I<1CZR2cCATCprI2fXWOaj$+aZ7I44X2@NvvK7?UmjA*W
zeTpm~puu8)o;f;>+>x3!bxR0~tI#v};ZmOB<%L9cDHfFr#RsuLj%27Pd0}zlY6%+l
z`qVbpGd1Of+1Ut$ViL;od{B&ryH7Y|^`k}8^3Z(4qST<ZxktP9FTM|6F2D(uaRw=0
zI3Pekc<EV&cQbKL-iLNF)(XSv_oXk>fr;D(UTMYT@nm{JrIa9r4oh&K57XxdK(Oaq
zUeu2Pt>ry$+)0XG1xivy6GcT>)z286lK`4NlDFO||2#d;l*oG71h<qm(2=LYxZ5an
zA^9uI2mA9q)M;x5AKPzC``&3MRg-^#Z)v^6vBqyyiLxzFU<L^Fx--?3Ee~05*nV|!
z=uxgUQ{sLv-8RQtNcP%P75P?}r+`5*_E~V?{3mV~`kJDl75d)fNXuEs_Ih_!D(%hG
z2~|I(6qTWmNHO&NdSQeh_9Zqpk@;>Py_w_MGxya<V##?>B=>^#RQ{9>Pnp075FVEk
zCnQixw#XcR-sD!^y``d8tx{~L*@{qsF-=Yy>vOBeE3Ia@k$RI5Wyi<5Gv2O9$Hnmd
zO3r^GvnLAEj2g^VCAXOJI1{|zDLEUScD84dhyhnI=+$T{CF|zs7JucKA^`rW*t~r`
z>Nm^m+@5;lSJt2U+2#QN=R>3e*#GD-Dr-t>1$p1*pJDpwLR_11(?@URV0_KRy%zI!
z@(8{JXmY`mVC2R*JVC{8!%{)KKZ@iZ7?TUHzL;&Z$@<4B9aB|IN|*{_T&nW4)IF9~
z`zrRYbpJ)afD(WDx8F*A$$Z@JiEL!XImWSLf5z=e>328t@}Us}Em6*e<#Sun{TeXX
zFA0(O;NWNrj!g{kCzy-c9(oDkU-hAFyk!~{zEB>=E#XX`KSUsS4_wDOWO9Fo7c8bS
zf}?j}HSc;8)9#z{X%g&1bshQ2ipyBB8j61<ctRZK;<B~c#e8CV7h&Q*41Zkp4Wm6S
zNGVJW7VGw|&5pctNeg|aco{xThtIM51S+#h{o*7u)@?)cWrJ;<r4D;tt|~>aR&|IO
zIY<SFwJS|Bks6y{x`whR-+g=b0C=n7kgTc?f5!cmfDn(NLkrve+>FEpv@LDD=_fhE
zDF`a2I5OoSmkdbGhXh}-F#BiTtlzqZ+4;|#Fx}S7dwy8;{zJ38pZYD+hO30VND0JT
zlnlpK$*2GZ#$p4cyDr-Q*4;e-#41ihr<HF0`P-ING%a1PpErK#9Bs8{rZ?ZHX4&qN
zJcjt)s`-O=A%te18a0=gm5O2d(lLpiqY)TC=kCfRNM(X3n&9HO*h?GJhMWy3>fQJl
zvrTbo?{K4OJMilLslD5dtk|6&036E~nxYzYXbo>}ga)O-L_$z5h1@pR!1g}AIBh2r
zgZo`qOZ~7O7vDdK-_tmKiMT5k`CtK}tc)_>eiKRJKPN=uCr+3RW14skqVa$3_f1p`
zek;CMP+eD*CnI_fRSCdg!|0?B>&=KATrw+L*ms(L?^t&|`TVc?rl1!@PTomY4-suO
zKbBPW<$dnwS3{&p8NLAmeU7{x?7hPPA`ESI!1)!sJ3l+qyqm{SY5QQ`MGHgEY#F`-
znVpOh@MTlG9*YK-iMez(%WEvWC&W<|&K{zNdu~JXIck<ob0r+YDrZruf;X7(TzowN
zVgfuO$=i5+ieGl8G^E)DovIQYGGQe%Y**jd3nUAy(tPu`pBemaRSpfGcE#>v!H;Il
z>~9miwp=><g}w6r>RosHu0HC?H+>{3>ML}}7vx&Z6$3oIP{olU**Y-aET?T>dP&Ep
zv_5q0^)xe?MOZ=0nE@_~#uMxF@`IDS6<i|imaE?}x~sIM=0RMe^Hgk6f!b4Lihbm<
zvD%u?Vx!8Asp<`jJ=ODS^k=p7)nG0#^(o=&pGjZ!WTCA`sqO7HOIO1bs<SVwcfF=l
z5F9(8!dLJusHSsBv10T^ADmb+)p?VV7TJ8b%pQznPvg%^AW_@S=Oly@aBDL&eR<qn
z^IMGo>yb>XUAyvo$Z;d#Oltft$-cc8!L&;KX1c6q{w!UiE<bw=qC!OON?%lrPaq~k
z6gIx0_S9rN&(tmnmjz*}z$>MUaLp@&bPRTOm|^H>NJ&FxP(EjLZaIp6Yo%T7Uxyiu
za|*&fiJC#l$x-lv)0@&VW7Be%t91vFlpo#So7f-!1~zq1Ca2kP&X^08khdsikFGEu
zcRry?>)Z8i*4+NW+_4(l?XPeh!rBl%#=L#P1f48$3^wqMW!S7V{mBJ&VQ|rTc&1QW
zRxkQ`o%Tj%UFO^M>rtt#<@-g`2SDKG1F_jKcT~A-VuB}MU4r)l8G3viPT#L_e7ccE
z+>wEixsd?zlPc@+uG~m7fvl%B$t)Cy3QWNfoIrj8^uXAVoF)c#v^jI-o@e_vGN)4{
zJV3Yh^snmJBLnj$&^1dS>-TKFpR3FrEo{+s?cQp0N1|Y{oXUU;d1%e+x`U2T1L8nf
z6Q`Fr#P<C)uNqz2N$tFgh<Cg`SGf{28cL_n3ajfd9XciP6v%Oa#~z;`&%)UXXUYV+
z0>anx0tr+Z0`N7ldQGHKP^l(=3Vym3T3mO6u^Je3f@3H7pYQ2bYa1ZZmgqB<Ph!&(
zE<#BG&KF{NVGWT-MaCC{?qmw0#QnMjJ>ix->>zHV&>&pN=RUr3&YLf|6`gu(bvvbV
zw{0lJ8QEc4kL&GQA@^V0+PsuF|7da<c=?si09rm?zqN4vvpZ-aqd!sWoGRc{Hq(|G
zqf~cSKJ#OTHM2>vKr0KN$_@JwjR{v3qs6#m3-;u`8Sol?0DL#*b)&}T$xEwF0VVaT
zz~~aR>5XHxZ!IqRd`BT>2{1CLKMtu%yL!vgbYEHOSd&E)6%#|oSd+2|+gMn@VLb`Q
zdtLWt4aAzSJ|*nDD-KGzQVhm_mcWLL0-h#da4{za$GVWrkv#yIJ&mSn+t^E6wPt4S
z=jycyYo`YJ-XfVX^H0qI^cm*uVNr$K%?#_d$HPr?I~*y+RJ6sQLLOi-DM}V9d<cWE
zt&((rLsgz$g|@%WnJRaSx(lO*VsbR}yd{8Kxxk@QxVRXj0+Q|@{!lyF7s;((@0rvo
zE?G9`$o~<bQU3-(qh+NDRKcw$iRS`&p3lo;Ok`TdgKwWx3)K?c@B!CvpIwO6ZGAVn
zU3Kdy&sXt{E{2?w=76nEPluQCh7W|Dlee4in2bZF5T+azFx_yyh6h02Hmhc*f;5}h
zLC`^)X-l)QtL^Qg-<me|g&PZ`C`g>BEg|)C!{C;10O&pzvHwa*$e>YYM6c6p;W?sO
z!66T+D-v$}ws;+?;@VHc$c<)Ba9%dy|Lal&J`$37XM!$y7w!|!jaIdJc6jv;_WP0S
zWMZQLdnzHp5-B#Bu|@!}I`#i(hI?f^$v?A**l}Kn*NSTE*8-uD5cSMBkg<}?_%=ME
zzXQ+fkr}!h<#5lFbxZxxtsFYL^P-jT?rr^-P~)yh=tp%FS_S?ISCbfvFg8UHDyCy{
z{U#IrWiWQ-EH$HvNw}Zjgs1OQY~D^p^pDBQRzdEntQrQ#Vx)70_~BU^w+1|S(6bPA
zPnEl(&D{I^7j_PmYk(PR?()Vv-HOU->*u%ayaaLfrj&{|24D%Of%zA-9E144UR*{1
zuI!*Vu=1z1|GMwv6DIWo;P+%}QAX33_1|gMy4lpcGmMY)P*!=<e&7+~jpOFCK`DEL
zeA7aBA|-+n9wdJlm$-gh48+c`AXQc<bClvAX=$|4csK7-qoW7sWdAgc{Y{_<d3($&
z%jxz8Ael9}B$3gTq`TL`ruItZCyMZGc10INm-){=2{-UBS;-XoZ|C;*VSk&E+%Ubq
z&5{V%eegrO)~kb=xm=Al)axMm0w{ShvqcbG73RD|N(|2YWUN2Jdp=j@a#m%V^rAQK
zC#rFBV``8(D7;H$MA<j-s&f3wxS&RVM=D;$@VVe?lr0fCuM0M_?Dv7J9&h6Q-rISx
zn$*j-1jNy&S(~}%!m7o&1@Wq2zj+K(cJDVyrB)<kLv6(hRexIo&5C&v=1mc&CfU2W
zSsDsR-#jjv^n)FV0~u!qj2X8Z%P5N)1=b(Hi2-tRXoNJ*d$z_=V&}x5uHhc~e&@OR
z$R7%>Q|MhoBQ08@pZ$|N1<d5cE-H4U^5M+lEtg{Y5tk(nvECIMF^Z_XmkVE{(=ub?
zf+AhtU?oWUkp*B#0~59iX%=%I05Obz?t3Z@nf-m9Vna_bhJ=d&0}(SJB@7+9f7q_V
zmiDg4C{6bm-@ITr&v9iRb1oE=PlBp<oc2f-6#;z0R#xxC+M^s7{5c(NL<e=1gRTUP
zIX}S*0=JWWy0I8ydn!>^<Ej$iz*n@?7ORg#Udo>x$2%sCdkE8N7`12Q`acI;wlA9P
zLv<mUDJ{B#XYenR<3IDynHjHGhFqPw302>kNW{~@osR&xWU#b~UCoPoc>0w^>j5Uc
z(|OUYN$dgtPKHQ&5sP#0uyGBY#mrPWCG*tt92)fq?Loi|NpO_t54I7%67$?@g9@)w
zT@h`3Gx;W6d6iF)D30Ec#I1iU3uZn>nWf~)e&u=pg9v}jZIFLB@QS#=Pp<P!kG9lA
zEp`f62;Kg&;{~hM*NV$(<-$92Z}4e7XbJPrZ*-Yi#BqDd-IVB<4Fs9)E0ZT>Gr3@B
zGv4o;Z7^U)ef-Hs7X;<Z&ip1Qo6oixsLI?Q01i~$J2abI<GWQtJ^(1??w^EzEU%gU
z6JvzT)iMZqhCig~l3&5}i%FubXp3LBy0^&i-K@4<0L`1mK~_B^8}Ob9?LKCg9k**l
z+gbnX`Lg51lTcsLuAsWFf=XIczOVJCC<=JR_7bP}U%mH9>Q^s&X@&%`x2a{TZ`|_*
z(m_95CqDorQ(x+2Kj{s(?;3G<=Xk@jBP(?M4e|hp8uK}G^=ohjo07WlQ#<IZT|E0?
zhpID9o(7{D<)qW9VRmu&*cqhRvSRx;X|be^-d3#Ijx9y6<-gho(EVdfODIo4tdn5@
zAlTl8aL|eov9_k`EP7N3Nd9U@>E(v&6&0=t%}!($5k#LRNUkD*Z$93f4=4B(+UtH7
zq1KdWG+eFSrHa@{P{hVAF9yHT)<59I!@5Ijf|u*nf8Xo{XSgpfl<6u@djwAwkqxP5
zZ^`szyz#@@Ka(o%TBP(~H4uLiYn7eb(vfq9M8@?+#$<M6U}ffS5;Cq6@kkxr3oe=8
z!?PX$`Kbw`I9uf$D;~bHuNFn!46Ju}DaMhJ#e-y{UkPt~LO3yfV6mN-YUU?ktzfeG
zT5I`Fgg-;vIQE!21%6(YA~o0A-qSI(${TFLX1R5idO&b^#(f05on!3ljXh{M>2hNJ
zHGWz`t~Tts4G<@Lj80YTGVJ4Te*T?F{<D8#mOo13j?3tA=s}seY2bnSZk#dmdx3L!
z(|>+}mOPg8UbSuO7_wvzQB<M^q0Su0WMJTg920-@6QP!dBj|+Aon{{G#WN_$r2}o_
zJXuHPo7!}Ttk-%`O1C{}={B1}K#&W~z`^<|MS^#bl+!-@*p{zC(RDA1Lwb#1?*XtZ
zeL;8lbSv2HzU`R3BmM4<0X@#Bu2n)qIEW>NeVPxziz4X#-!sdCNxIrTW&TmpR?sJZ
zS6CdOHB8q+6^e;3^96KML*!L^Xv*Q-s$tAe@0jE?&kYnJR7(OYvp()KB@7e;SeT3L
zGvq~GN;exU*m&5Fz0vFct{F_nfo7F(jSrs8dKS}N2_OgTZ_fNLH(HzJHNJOOYa4c$
z+d)6o=$Z><jfX!^lGpVvW$P|kETJX&_#fC$)D<|Ap`tYGbxR?n$G2jt#FZISg};3|
z8%?UL4EO#>>Vk3Zya}T``LXagRHw2V>ew3ia#=;Posp|Vlj@REc>6ceUhp%24K@6H
zc|6%CQ7heQy<_B@2%H>@Y21|N(UV~YoCEOQ|9{y(?#3s-$zU7J^`5&98nGH#t)xO0
ziKCVbw7C4}&hJ@aN3~oytm5@ux3KV2wcvB1qB^<D*tf)Tj6DK#HKmTXpDT_EP)`c`
zwM8guM!0?=b96&WgPa4;F?U^0a#8qy!V@CkR0+yL_ycBtM}83zUAr&-s5^|Hwr?1p
zWe2(P>Q1uDhWd~|#AM9(1%z{tUy7iY9>ZrZx(y|4P+Ar-nVHX6uZ3gsc5#&@Gxz7;
zwoKZ+F!-mJy7+YZCu=UhHkcc&g>B(YPsOnH@4etp`7Ae$fITiQd_u33<c-iP-<mRO
zqhRxmbk+ld+nQnJh-poPh_diZ9Ko(%FP`Opw+So6q9?!cgu}V5##XL1>HvN%@?P=3
zvG<lyZAD?bV6Y-ZiWXX&P#|b&X^{rE1P{RpPAOX4y|`QOqQTv*MT?hG911ON!HPS*
z`M&#O?p<qUt+}(lJM(A$=49=YoU_l_`+Xk+>FT#9<!pF*s2{sysG9%rO{=NYbGzVG
zhGv6Udbi2A(p@b<*$Rf+5G#)ETOWbL|4!qi01OM0mDztW_()!DVBee6{+c`Awm3r&
zw_*hf>A9tKU-(a=Wlpr)Z@aqylgxX+OvMkTuqRD}bcuQz-TPu!sn2J<%Mr!nnl+Pl
zL^Z*wi#o(EgJ9S<u3cR`5%d5?Chfz?5mip5hFhA-ML`w$Bo9@}giUZvEdJkz#*BYG
z4VQzc@yedO5H^$2pDheb2})9%$03t<`jI@Isl&X#H^)Ow-9&92>m2!1Rf8fa+ag%i
z?}1eQKmMklSm@S*wLo<2KwQL+>WIz)i$OH^N=|{GPdq+qgH*ou_;c~|>OP)h*qCsM
zR2U`89&;|%1+ccE;VReyE=50)q`@jZl^aAycJM#*N^@0oD$MvGP37zIndcm3uQ7$u
zH**n!A5MXO=^p<B|8%b9>&4*U8;M09+><`Jy%(gV1%ix4Rt(`AMzf<HMP0u*tEahP
zySAosvLF}~A{{2heC5s~7W04eG3BBU|NiSE?J%BrpEC?pPC3W$nw-v(^M_~zBTn6+
zT7cD;sTKp8Tg$l98l6D{`FeGD0u06JSp)%3$}5;1hmY|;anKq22e=uvY}7{VfOZRF
zqTn~tK_3{{+1Xn0zcG<+4*P3)ZkzmcdmMEEvYdb$mM6b<O=8LcvV%p`FF02JVudVy
z|NgS$M+#SIgN;CCLIxiNbK)8nR0jQ5>tVI;F4Ad@H}m5ayK=FQoOBn_boDy^ydr*h
zeUd_?-;qebnM2my`>wyR{Ra!rdiyz@aMOC=VBNgJv}QHmq!NX{B{TVQDfx76Bl%&U
z%(mw3+ophLh1Xp4CG|%as&9Vo9?dOGM|rM962!gCc#)82O2D|WavzATG!SR9N$aIH
zdLOmQn#9-Wp}Sj6PCB0=^MVHdu+oFcL7}1X^oxw?hCll(H(aGYPrA+QA!=kcB>T>X
zgQ?84L;U?kzQc2@HG{cx4~75(bcJ~C(F3vh!QWfN*(td5asa*5q@?2C3zqHfa1(UW
zpcuEr1LDLbxcp^nb!N(cTDWub;jC_AB0jC8kM_!ZVcjYf_r*+?vwd8s!iA-f#u-e4
z$lDUh##HigA}o4Dd)L^$qNcXLVyV2W+Jm*N`X$-QhxlTDJu%OVtN5p>@eE%uqV9@C
z<KwOeW-p39;ueyN`hzXjasSGn9|cF{RCWtSdHgitxRFk*FwNg0U?ZjpFPsh+4DZYB
zvE1aYeMQu|OVHz@dS)z92o9G{y)*=n?Rf!ce=!trHZc1oU?=%5d%Q@4%10+VzPKHw
zhL;+`m&AUl2s3?*&%{_p)FmKJ{3h8eYrbg}&;NISS_e4Ot%$2~4~SwS4rO7JcrTN5
zkppkk_56jn`IEayGu=;L7Xc(VmrkWzQ!pbuBpB);bNXt)dGyTkFt`&qsw6}_oCx|m
zC?#^&1Nd}IM%4rw$&BadfWLC1Ejwg+lDwbP3YWZx+|FCNCpHeYqqV<Afh=JK>#7;w
zK<XOue)aDVE_$2oerd@cZ=UgLN(rS(Xt=#sM~(hw!xA343t+N3?_ruhEV)B4-8;No
zE2HQVqK|?}F$RaDsh>#v=yP#8)q3Azbkr<lI|qs`YH!#n<FZZxBD{N3k|Z(m@DlOu
z*1nSluY9-|g8Gx&q(f#|4NL>~{4o%_Fa(JKb4rA9V^-%{AI(8D$G@aGrBd+~#1+G(
zn~2kIw$(M5u%(@HgZR6)`7-&@e`@r8f<3E$6BxHb6Ch093QN7Z8xNYb$0{L+kI*E!
zkwOL`ps?UHKVA2@34R+m9E}qlpMm})opIqH-lc2Tv;%iS&``DZz!V{DdF~>N_$Q+?
z&Tw-3&JV7(Bn#NKZbB3wT<$|;4FzP$Cs;iu-WQ_}aA=q?cC#(2UZr41m7$VEDC!d9
zl9x~MahdD{BEDq(16Z6(44h&bu$0*w#>JuBbi25#Jvk}^!Z|1_iHola0e4HPPMAG;
zMDqRa!xohh(`KfFLE{qN>)EoZvV!s(SfnKdbxuO%23Gvj(+=m&H))-FqA|^7<*2So
z7HYzDc7aVl@T5XOe1ub1(m%ktjK#xSiQ%H_(}M83RnI1o!xvI){{Z5&iu5I8XYbrx
z@@qBpx!;gPeFZx%mc5g9mIMg{t?&UjD{3Zaj51m9+&$}nx+1|<sknr}xtFCfL(U01
zz=w;WBPKE1mq{V5bvdBvB$l7+!;`7sAhRHWF&RG50Sa|TjO3h4{2M0|etwrrRpu=<
z-`BBs=1wiL86lAgq3Ki1xNd>oT7gt9lSS_0`@lHF;<?hJ-$feBZi+R?G@{xlq9%-T
zt9uSDFB+!M4*_?%U5N1KR4yS}4D8-!5{`lGUzQ`BX<iaZsSnBv7C@cW-pRz>eS;|R
z{R4ax)ZpSSLfB9fB{>rj*SaS7)Ap@r^TN|=1hYM)5JVisshszUse;HLo+ykcvw`$}
zaLoS40<MmCoIsaU4YZc6=CdV(wK%RN1Nc5%Cm=eb)ML!Q0~+t5;>=plmxdmb{n}1`
zNN~U6{1U?;G#peSQJ_4)>^gOr_S<GA6-;2+pKhLz?Hos4yOx)nc+^ZOu#ivxo+JKz
zr_PQv@kdB9O@tIZ&1}yqK-=2Av#&$gxsI|zuTHNNB6kUy-H(E6#z_|k8BHqR<+i~0
zE}%r;21%aI617Dr>a5`5F){~#@mz0c|KM(;;-ggmYsXyK^=w*$odsO)jRm?J2Ih|S
z_#$lb>#wVS?;K1hkO=1+1`Br+N~e~>^8kMj3;KY6o;euCyXAgIB%5g`dd5{R6NYi?
z2c4sn!0I&dY2D|PE547vx5VDgU9DJm?Q{JDJaVUqTlns7PbS~8zsMk@MJ?ycY53;L
z`@A6-m3hyAsLN6|)+Ig@{}}tPIpU{EnAVFO%_uRNL1<HUm7{wlpY3|U@AKQw4(BJ=
zNpoClb&$Gfdmc^AU$D+xgt6$jGr=(+XAn$!<hRKvJ8y@ajrv2>0l|n8RuVzgM6@u)
zzh@6iwMgVp`=qOz!`fP?Zm7icy|mO}_Ie2;7-7t>7Uv_J$}e~;HyNu_!9;0nTgioH
zZJ1`RV(XV=$MyU6@KI*qi(JccE1B5@f6DnAXdCypGweEAHIvYlbqOw*;=u|;n15Y&
z!$^PChnM%0)q2KZbF>YZwt^D^QkLlhNx~GFKHxL*07o3$@)v|>?OXI}mug-q2o#kz
zNlL6f;fcHM76Et+9W#+0vYo`X&n^0wAfmO~v|xq~wn+vknJXfhT{5KK7k`aIi6_o!
zcy?c08ua<yGI`6~z1!fxHcSi1Sc8a5XN|d}y;fd=c~4w9V7e;$YBOM%B~02!INtvh
zi>9raK}=_M@1o<*8l6KV6C$-J48(@?&>diO4&dz8;}Yu$FR|!rqtmXP*%7~}uoJiq
zm*V#bA{IG35u*M41XS`x*i(|X-*panJG0c6n#AIY+$+0_ZjjUpp!i^C7{PcTxCi_>
zb#HWfw$rfbc%<qT1l*erCz{|Sl<fUM<dlrLv&WZCFp(0a-)}0|J5pSU_DIl47WWKt
z5Mx<EY7YDx5FrTalaMEVa{JulYQv_h|I;1eyh`?`*8aTR7T!!7xmg9NXWAyxy+kK@
zxs<C{pP#yH{Py1LrMhkU2he`1r1i0&Nq<XUE7)RJ6+$b_CQfF)C1tHZxPPxe7c3wl
zp+yQ@y7hY55b&kw!LauH^ywErohQF|X}+p9XO27N7gWX7;XIJZigxOfJ`$}395b>E
z8&Y>ExvkIt1K7@UyWNN};ECcPK8{ku&0b}wD>88sj(0`4I$*E^s*bm}><yTH$c+ob
zZ27{b_E40y4$~}bakq#I{M`~u)?WdIb0M$**$S&a*VDOIT<<s>wMXee_X^R{TJet9
z)YmJQgS8&jTwBrj+?vyaGSqUNyK?^kp$6r(!1zu<f;-5#-39Yi;MyOdjnCWK6FjLE
z{6vQ64+SVGIk;9nR~gV^G+z0E)n(1sSVy|iVUGW$hQo|E$*68#O|W7GMU&g$%&vK9
z1<o-SmpCI2-|feAFV2f!W131+hX(c|f1oxq%GK%l>c^^*z?BP*+Hr@YrokK&u8Xf#
z`tF0(bLd4f2pm3R#g&Bv#Fk`Up%4^sbTo8!eb8PR`|}UrGi~Q+|3+yVog+(@sV<t5
zM}<X1!~mtE!|y+Ii!+^QIaFwnYuy!Ue^^!4Bq2Vx)Ri7)=Ncc8hX|t~bmBaMEE8=8
zX_fC(Y{H6nUX)i=XvneSiG%fqj=^^nf`a}(A0al^?a@iRB|mscsDFAl&PIE`D2eo&
z)K32@>2$&C^$#EvHWKj4hduPkQgbzTr@oyF7qhlok~&!siz3b$Q0y(wi7!W#Mf+ZA
z=0RBSt&}E(O=-|BN-9!vn8PlR)S2{U@CNU|Z;0FU5tYzH0aS0NST<63Xi)0t#`7Wm
z`CX?DeA>%lxvWSZNy6hc9M17K6c+kK0;Y!Uv)+7a2D5wJjJtbY$9yj*T{FP4GvXaS
zJTsN#v^aRf&rT2Cl*Fv4+}8g9{F-;=*>$d^2=?akf|tmCq7B8Xgd~Mafym36msA(v
zGVi#esyX|XA!m-Lgf(HRcQ8fVB=D_?rxiHYSmTt1!D#AH#TT3N8^6iYkg3c9Wmsu8
z&;muCJkPieu4f1sF6t!uY;L;no-lRW*ctAk>{Eea=N|B%;mK{g_4<)l^e(`rou0Kl
zbt^;)<im-_A}7hjw1avu5h%W{UaF{g(?L?@JvKmKH;3KIO(XHCm4a16+3$-K?J>XQ
zmW*xXQmgiad(GTl(>2nUJX3Tc5}@pCYJ^jFms^qnE!(U7qM-vTu*%7&7*dOeC1OeZ
zP+Gnor&~8L`3E@pV*U@%o)}Kyl~Hp;tyXrtd1O-mQ$9t>RT1vQUBt|Z!sV{*IrLxI
z_ygJ$li;y^e`*`2Z_S#}r>Vq+=g3btpb){l!E_(*Q?75#)81dcAV<VnR8kc6nuQGj
zNW8xvzS>+nB1C|M#ym1@ct>7BWFvu}3R{Zv_d*h{-EY&O&($)7xh|q&tspbi2C==#
ztog1M5rxbvfzohQGBd|VexvipgXv;M+jZPpU-c@^V<T`T<PfV|I6!t;oSC>QW(!Zt
zJ|m+R`wF#5%ZHj{E&p%apfuo|BB&h<6Tl`LQX9R#O7g=r<<wZ6`7X=$otzMVdh|Y&
z$Iq0V%XXst3*cDSnt{1l>sQf)E&-=_=Ifsqp3R`6T%|vdI9F&e3-5QP_r6Rj&<tAg
z=U84n)GgoC<wI)P3Z+85wFbCm)C>HJh;Yw|QKa)1<I&Mx&-!KRyjo^HI0zJY8|TGY
z%GRY+<|#J@g9l(@?@|>B0i-?QMw?DWjRt53p{W8Qs4nvLWaWi4g&^b}=MzSC7BKJU
z{aY-$rlsRGH%z>xw4{bnm=JPXuaxF-V0^PzY$P<HZXG*-bbMiFl`rcK^odz#TRC1=
zpZAIm;)TpPc$3fbz-{#lO7lUmU^q06CeX(s`4mV)kyAC^Ts0`yY3JUdrC<Gq6Ul3t
zqZRT!0Sk<bTgO5iuO$(X_V>R_lNo8x_93AL?fe6T8B$Cs);7sD=0+S3F+Wr%_8R0q
z|2Wri?-ty{o@-Vig_z7KVb{#7*it`)DQFzc)8~>N7Ih0&#I%`vu1784!X-0r#5?=)
z!6y>JDM<l;1XB2yhSjV$4RjbZ7RvdAW+saDtYmw*iXmuVsmV5>nodiE+gR>*!#aDn
zUjy+ahRx_jC(er0&vGwi^Bg5vfZ2=-!gTm_ZXfxkN0z-5+&2k(47p3Ss0r02LK0W6
z<x|K`bD9}6Q$3$MbuBd+5R0h<AFG##<}<wxbmM{y;qrnQWG(Qpd!g&q{w0gedQIFH
z9mbXW8hO>CqJ@cX9w|bYpExaJv{86H9uCSK{NLS3CtVFDZLN2Sxf&Au3)I>4y$45E
zKjo4EHvH7m&Jh0qNe6_#&f0$mxRiXO($qqlg03Uv{b7V4ZPDbTn9i#0JD19Tfd0a3
zLRzTUvf%vHHmqft|0ld($1(JANfQc3y(lGE-6h6#8#kA1ydi#OD{j~7#^$~-=HrDA
zJ2X(viCq`JdsR_UB30_ja<a|;)LAF<xk<QjmX%g<D*NOw4Gx7;KOsf9Ai;j&%<BD<
zLmvg;3FAz)VWrj=Iu9+dpWyDamLc`jv;cLy+BbUgYk}nEPK_qpJKHz^0Fs|Pnu}?-
zBL!;mMvzsU3`6@;I*tl;q(VBG$ls;wS+JiqI_C=5?`s6ky<MZ&UWn&?nwC5k#yNHU
z<wG-A>81FK_c?oW^=~j(U->1x1b+(tvx03O4SMi+?d8=p`RmF>Plbmo9?d1ET*QXw
z%qxpzioCcBG*v1&bunq1-+lUFj_;Bos#r^wr#7w#6_it5OA2#Y!~w@i_ANSXrw_S*
z9kN~|tud}Nvc&D7hDL>ikX9$Bpil{f47HBCIobzs^F*%~m6GxLnT<U}$A4aftRXn7
zIVtx&7uS)j{I7hiK>+1wS!-B$BxCvDy=CZWve#_9Y!}StB8%O9xNZi>`<m+W*VnnE
zg~85_s+c{s)54Bjjzzh3v;Je!YZO9=i;tUpYWhb(X;ftQ515!mP-3=)<!+lyyceb3
zW~g&qC`8Tcm2`SJr$vvxq@;8>gslt+q(^NqS(SFU6gM?Y8|=Ru9XP*0%1IcyK<0Mk
zyRg6bmJ`w)1sBsC37h0F4>3Lk`QCp{-G~F>y|XS476`DI{09K0$(^dHF$CZ*Z3h-B
zuBE2Ba&a)1Oe^9t4XY|om5z16>L*gpGQEjBBd%Nr68aqeS3C%M2~RakjW0$>v}m5o
z*^j|n@E`)k@s$JF9Bo(lqCnp={1&UT?*enVUpn_)#bV{$k?*Tp^$xeu7R7~9>LuBf
z(?kk&Xm$zG0{W&OODf(=O<&#btAHhzbFfI3kc4EAB!r9ZU=|5{LKg3IxJ@knnApQO
z-Cw4(A<KeZ#aDg=<!zK7rB<Sx!{E6)z8LdRy|}p@S#h<kQ(f9w_Tt_<ZLR`K$|&M!
zGA5q#lru@pVtrxM)3<iO0Z{Tsc+jD3_Y%yy%8oa(R;J|+h%m(hG3Wt`Aj}*LgVDpO
z@1MBc!n)vI#kF=$`rJ>gf`6GMwY}T|au;X>*TgB_E^B5sk~EJF(R6D6EbaO*H!06j
zlc(mU97yC$xt6Eyt;jHdMe6&WcCPsppT*het9VJ}G4f);z$JdS9E83}ig;2z(i;VP
zB(I>s2DlPSF=P<x8a*&(m>efMbfem`{I1G|&`~#fb=#(>>W{-`)+f;a8z;F<NqTy>
zG4xn#wsTW5Sq%b{o_eYZVeo|D5?_S^6#Dx2eY=h?j>=CShFuozr|@Pc$|2lIxhw3M
zas_hipufVvxaD>u#np7FcPypL`|p~DE|rQ3>f@qT;aW=XCe8|+_INfqmh14({U2=H
zlnxgw=B@YNWR7P4L{RcXVD4@{0}+G5{%F?U@62(2>nfS;+cnON$%L8rCQwRtkiz(D
zYp@K?jV`-n$+u8<?~ZU@XX6*TMKhK3PA+vzdj$eohPG5FDQK`j6R?b}jc&jv=3>~3
zX5N(aT6gA1H2(RtK~3{Wg_}m(7;_ZnT%4RSc;+jL5-#a*>>yl=r=tS>44{$(-*)4p
z3z-5#cO@Y|3`koL!5QpJk%uahX?G-2#C>vy7OIB*7KvB=<IbyJ9F?e*=nCsi>RLEE
z9wPf&GM`h#-4#&lPR5f({1Wr()fjqbJBs5|yIAcb-i81~l`jxisF!F{5E9Vsjw`CG
z<9etIk5)O0ueZ4zl#>Yd56s!x!KM{>&M#%Dib<2#2o(u0hkq#U<9npk#0U)L7_MjS
zb~y>Lr_Lc<P_+<ZAt8am>GY(~UQy)VAt{N%U(QBF{_S27KaU!<qeh|;#FFrtK}PeC
z^C5rpgZcQcKRk(h8-`t}byUIgJ0Hq6%F0Wd$Enl@1j}gDB^8+{48<z`j?LMfOsEw9
zwk~coGRv{abk)j36=;B>yu%})Ns8mm&L?c7Wh2kr-k%oDE!el{N2|C732JJ|uqJ4*
zs!D38`6n+i)ow3@h*X`Zy|TWbb%Xxlsp>G?F<0RQ_fRm&%c4v{n1?|Z;6X%p**D?k
z(_*~i+jsC`p(x5{eE)pZuvHC435<?JU>fB=#N7AyQ^k(bW3?w9HGc&yc&BDc$UjFR
zgQTn(f>XT*HEvZ2FWvu;tY!Vl(DeTG=4PkjjnCO<H}S*~7D?dQrxgfCrx=DZPt0Bt
zXSi&-SV=!q<;)(gS26!#mn(!jIE5mw)&iFvTQjBdiKD!f*R$_0kJ4WAHkg&pc-g|5
zl+@}M^P->??@9z+<FiT@%OGeNJm=e|#n!$(qqeD}Hx+{WqRTYPye;jGwA5O{Iwfj}
zp<KWsHVNzk!13XUP;D=HZn!(zpD5X&WPy8gk(#+i3sPk~#>pn|LfV;f>LlKw`r$~%
z^c25+Zx-b9rl)_A7jYQJf|bcRORkQJ!Q(v3ri{D#;ggma8}`IF%}8wifZ9iXy}@9;
zVb(4Qso)^(txSMn%5XizySXz8Gu$CuM!)%VJ+pJWk(*O!Ee~s72ckOyWd#CbtT18+
zFk#Qou@~QWGJ5Vq-^gCWG{08RuotEw_C2RjJR}+k4@r`i?Ac`lnghs#$*dCh`)J&E
zG3&6P0b|$4k_ikEn76(J4dQzXq#FJdmq&;*@2>bmm2iOJ)8{6V-oat-OOD+OFa8(5
z0VQRq7;1h%O+Q;3G^vR=W+gn~dmoJmhxw-J#~D!7n;+m--75I5ik!|g5@gelic2rV
z<KpP@@4oV2>ppt(mCds9q{Mo5cOQ9TWF{w!AToPlarPZD;@|Uo!{|9N3!~)EqeZ6F
zat80HU~|vnmY<aqoMzIcH{KKRe{jH^Y-EKM3m<)Zde7U%M=Q1WglN$EHa4P}d3hYE
z0^Syc%uleO(&7Fie7<J1=5NwQe8c-`F#4<HSpK)x2az}4qU}!#BF*V`@uWbwWPtpG
zn3xVVGTZD=b%!l{GJj$%3+UDbe(HW3A&^$wMZ>)gLSl$xFvQ1&u`A3fU@P7;^o69+
zn775={l3liydh(!!bE?|(7Y4CQNa*_$3e@+)^LS*vp9ZNSox=93Ur?J=&QhIH?3i=
zOkZB^9o#?&J2OD=5rlnXT|oG0w(zB0%SY`Tt)IKGej`gE{d$tQ{jbg0m;)54N>M?}
z-QnO@#CE^DaueDpjY?;lss%b1tKZU&y*}5Hk`{iFAL^Qv5;UJUT_Aw5-F>Q-_m_7L
zUZgOKu;E48q)Wcqn;xbJhx$uQC|<3O(!MA6Nd4`<SLp<9Uhr<2HrO-I%vTD|XW&*8
zq#$Ja;OrZr908-jxZ_IUFG*CsB1SZ59Es~)jP`>c)fhBd&*^>j)RHAM<mh}A=$J<a
z!8O&zeR}R&@cVA`;Ej#`J*-7%sw&z0w7iuZxZNx{M5n>T2{5rpGEz59mlC&jjlRDb
ztTJ@Vg11mFa4T1!O4MNO1gZpzhmxFt^|iC?pPxt5Jl&@>NaNk|+aI^WQ-3A{OJg~8
z&q*bPfp?6G&9rXzr5M46EH67q8M7IAA}tg2uIqU7Pt@u#l@B<sIV$-4W!x(bI>{U%
zkV$p&(Spno8H)k&OBc1O4KiM&4v9^Ldd)AlHZ(Pu8u5h!mAtsXK?kq|5|byW8XfjL
z1)eIfF1~r()V^Je)^_ubD1?v3LHluXkYp(pK;k3;0JB9}B5^P2{JGe<TBeGJdDw&=
z#&dMZEcL%H@utBw=BH#|M80{xp$q+(*fes`0O`=HaywbvLz_sbXwrw`QD9T$rh5Pq
zKV49)9K6lE{2@^%S84y{WBt<Mo3nHcJpk+>pE;YF8};cg8h69P3p8(v<*1$2Xn6h9
zY?C2#Na*NST{>4)KbV_2C8-H#xd*2WP@3e3?)+#slliC5{EyvKrgvuRizYagKf5In
z&glqE<3rFU@T(bndY~X~PJ-VLv3K6#@m@;Ew|gYxNe8o|T2Z|XqFh|>q5aAk_}yGa
zKn@HG{K-3Ah#^N|iQM{Qp&Z2+H<-*hwRx*Xrdo#NecqGYQ$q!_1(Byg?$5_^&a0L$
zALD6B<%D(ORa@qZ11v&)%(;Yn=otB5d=eOddh<5qcv<N$*qouin)OwWWONoZQjMv@
zYHM>=D3JhxM-+r2hqs1pR!eT;UX9M|@!FO1JIV4aaCn&piz&BsQUMm%exB68<6#=p
zc;Byi2Qy-innr)QRoQlGX!e%xNyk~Rm4KPuLCn00>I}y*;zj=eSt3jx4a3zKw|?Iy
zgssNei8lNkR}+tz3!(v)l)$5Sxpo4s?Il{gnlfv=wyT~Y7wR+X`p%l_OjDKy40Gjz
zDCqd71O;7ih#yL??jh{cqR+G322%B*BQ~__nS`2?t<&$Mn4Y@4f1;CSFZ~Ztbx0qn
z<<+RIE@kXU0ULhj$MsnpHOM8tPQ>KtpgsNZ(!p<iZw9ZRzIZ{B{9H&wA$|2*IY;TO
zJ9FDiZ6A|IWE)vjn?r+R)353d)hhQ0RvTJ&{RgD3s<-BH-lhmXEsLz^9Gl(q`1DfV
zo7k*)2X`CU8P)<H<g8#KDHc9%AfQJaH#tz?<a2HB+fCy=nz{kkX;EY_LsB)h=HxSI
zrg15B73v5_S#=V@4l|ondRz3H;J?d4d3$XX(=-8UYDHX%AXRX!v%p2nf3<6EPvHMb
zeehp>_Wysp|9T7J{}-<5!rjA`$$MWK`i3{U7*0kO|C;q$NzCx;h*-j7*fanG^Iwej
zrmjK^35&{u^EkfsGaHGo-Ov9I^o{??NGuX7D62pH|LW*}<$>ose;dYT=jsDc$(9H=
z1h_mkm5wVf){-Eh)DlMT&xY~iamW@Jd*k>vg(S^n{<T<$PG(e8yEmwZg#I+C$TbBU
z<;L_RIexT))XT(keOtV|AhWL`$~!WTyBZ1nhT&dFUcHO+T-?@q_%!_w5Ci-Nz~T|B
zy_Rnp!v?CJHx@w^b?(f3EJUj;$GuP5nhng`m&@G|f}1Q-uaMO8Sl6{wYSNRWj%3jA
zQ%kw04}@gbum1rY9domx-d>&wN`f$ps7<U$g>Muzq!P-tGuH!cl}^7gJ|@+&thQ0`
z3>G~#ePwR8zQyjYEMw$jIlt<pDue3j&}qp?Man`z8S|a^obD+J@YKt7Jp<IR$MxFM
z>zHGgEV@a*+Q<R}Tu(%ay(l^R2T)@|74d=*3~Z>})pt%|)LACKd^=B7KIuHv&l<b!
zUB_ZSE22=R>5N%Po<vpPv(+)=j%RBbZY>ExpWDkxvzHYixn!MF+$yGN|MF|vw;MHe
zx;Fi}{t?`$Rrx{;AuGqEpjx>K>UAc9Y2^$uXD1-~zN?Z6AA2&BpA+n3czde^rmWW=
zF^*c*zM9sl)<|CaFVFoT`t{Wq+?-G%lSpa8t7@^lGLr-xHs2;UYEmHwlx7P(1BP<b
z9xa`5+-)OPLrLFk+mFZ7Am!woG+GTNeFel#@UNpW(D<8&BZP0-R!&?vV^xK6(fJ1L
zh@Tyv-@Of{_b*U;a$>EaZRcZVi++)+$ZaP56+(357-~g#_t`^j)<N9*O~w7uhpsu;
z-Yhz|7}WxUOp0xwOg6|;u(g6A;`8njFP4QK?dDv!6^dM5Yu}`a<80)J{Nk?3RtpHx
zm{qjY!rdcKKU9<=!h4KCXfReAwZ|JwJnM0l)M8c)c2C=tkX1~uRxLi|QZ+Z%6o7Z&
zQ80mUUuLVVbIqTt!x``P?ovK(Xh9hyzOY=D6IDuPQi*fZYGy3(rD<?L;w)$|A0DZ$
zjC5qtAoh#Kb4tDxxs>^1cNblyY4--A-L7|%6#3{W#o(HBcpcOEr^DXvJp(3o>-Jbh
zROfzkhaUuR^GX%F`+{AZWq#}NOCDJma2L#O1h}<a7jGN9v(2A1vK8@oKU>qmU=Ky<
zX1XculzfZ(bn2F*>WnG*OcT5hmun7c`3icx^4bPIg=Xp{b;f@uT|f6vKA;gbXl+pQ
z$dB4EdqpJoBay!tiyeWn)53PScp7<wSX4uuO1%FEkjIds1`!nbWpS^o$a&ujA=z2{
zZhV+8|FShKxbRhJv|oR2vaQomkz*KHq}QU9cC5o!4wH=^V~(E*Q{L-{``Uc3QTvCp
z+APsbWQ-@sOo)DkLW1=?%|T~^#{<uxTFgE*wGO|?aTuP1VhXxq8vEwk@v*_nea!~#
zM?+XOllCiOXHMP9Fo_H)r-Yj`vWx3JAlk5h=1tSGxiwchIko?`f@JnclOU<NxS$AO
zohZZ3*~5U5`gdFmn<h7EO}Bt_%Xi96%Zrg<jWMF$pLNF=8=r->?KIxtWc|(G(b9EB
z0k-I=f%XpHvg9gGn75kw@va86lO6D-kVi=D#iHrdj(__<fO}!HCgmh@dWdq`AK;mM
zfp1iI`jNrTZ78qVzInJD1?G&jeTmEs_U<0O#p&mhEUm(*HeM<i<g|bIpy8-hpG~@!
zLwkzmV(TkCE4}~PNqRYKq(%|tP3L)GjmqXME`A;hlVoE^MY_h&U&S|tX^!$Z%C=eR
zo5vD@VAGBi_>mm*!Y3`?-_k6i$9NWFu@X5@>B+!I{QDD_!fC6_chZ?=w4qGt?$h?1
z=&yFL7u&#BZa1~P(&|KCZ-O8aHb%=azNP4vKj4<73!><x>NKdubZC!&v~}=+I<}*9
z1TBa7z93&;y<T&cv%?#u&+G$<3fMMY*pp=4>)rREds@TKy!p{{$ht*Jcj@mog|C8G
z;rJT3J|@IplvX-E&qawG+P{D{sVHdbu-6x{)dSVXOPCo4nf;?!b}wTc4<4-ew8<)x
z5}!WrE=l_E4?v-g3as?|D#BDj+*-7CC|uR?VoI%5Uq`#?##%9r5Y`ZbmpJKv`D>xg
zR(`6YGx)Z_ptM4a5<3;%xZPLAg<Epo_pL_Ikjr$<$*gP+ZJS0CP(fce!^0EGh6=?7
z9Hal<GK7)wr`qP?Vm6kKizfZTY`qUV%Zne0FR&2dQaC3rZHw>-{+RJ^GUolzVZcY#
zltKZl5_+OY6O144^w;|tqruN=b!|0;=7p`WD1QBX>oPTE#UTc$L<Y_&Ka+1Z!ltO!
z;i74z2qr`2Mn0fs!K;|S8ULMK!X>ixd5q@BVoIuh`>#wTi`QG_wcV2XlSn`cM&w>Z
zx<T(@nt5|1up?&K|7P*clzl~|UN7@wPS|)4i=>->hLctHO~!7c80$)=Zi+)7<+Bk8
zW6}dD(<w<W+!w56)xL+E87-*PPXK#irRG?eN`?Z-$T6g-G}~>zoC>Ftz5HeFv^gCu
z?OcTCp#Btn1gr}~RIhwd!>b8-``ET#g>^$ub^$gGk?7-J`ih*&5uO|~Y=1V^JD(Xb
z4o#+m2XU;x10`?&Nc{1zIZzu>ZIH9iaD*GZPmOiXTZM!`SW~eseqaZ%BG_X0vpF=%
z$`RX^ZJNs<qVA|nE^KQUAt5+HI`a5<&{yDxnD($$r@q{uIcHRL%Nf#7?p@G_ZaVN_
z>lGLropKhNxyevJ*7(bI&l0zQ0EC^I_ed4L6TrUq5n0uauu7w@(pY+`^}eJ016B{z
z0)~yG&*$h)$o7Z&)d(zkA-?&3lIz?f)B@un)XQQD|4z<)nnt-lM0F&H&yc;RL)4v!
z=a<g1M%~vf@;1Tb*_DM}g_PzcVb}c~l5qK>KBsB8lUG%kK{*9y5$p$%!Wt736y{}m
z|8Pwo`kS>GbvZ1Uy_ixFw5m|#$d?W1*W{3x=svo@IW)3kA&e!Iroecc0UiN}MgF(M
zg8d(ZdXrR0&0K~^$HooHxjdmTdsPUM_jSx)BC9)E2#1baU{XVN?3**6c|YTnXT#W^
z6R2(A9Ij>rty_I|!_{@uV(fWK2)S#=8Gc*5;5z2v97f>&9Lz8&>PG!7GySHQkQV44
z!<lfMvX;(ePPE7}qMjgGoSwaYzaH+z;hg(@YuRejwcNYVyf7CH+S`mN5~md;-%ml6
z0vzF5E8ELw3<J4$lo+JC1K+xSZ#LFBMnkVE#T{+pvJ2}?pH%x6D3(h553tP|_5364
zUdrp;-LTA+2y-UfV!S}{mBUntmrHsO%a*CJrtV0y;>gEpQ)3HK0P-twKDh;<=cKC;
z>K8rMsJ5woj`9e!`D3U~ups^*Zu`*<>AKnjVI4->@QFBJYwYCH_eX<O*7{VhmXx(X
znfl9eicmpp2pu4#NJN8!@kjp_k%Ur~aJ6_rb!366TI9DKIxnmmwCF<_MM0gO{lz!_
zV<KHjCY>$G;d|q;WtF*_OVQ^=lj90{h>z-8ys{Iad^ict$p=f=y~L*?)JE|w)=Ne9
zpQe+3Yn&k^-6`1Glal<WGbnGu>rN^xN=xTH-u@m0RWH~UC2+}fVd#g;2l~iMd?DlW
zMb!y=MDwLXajOQ#$u+N30Y-?=yuQyrIc1esXTaQSz`0sjP@*QQE}fIYA_0(q^*z5U
z)0j4NGm<Gll_7XlR3KqoKOIX`OE(C0#lKOI?cBF=XKR+1Z=+6Pf0aj_LbMwg5Mwql
z#qLK!JP-8b32J84YuooOYPB|sWpZH?3ic|ndr1FlKgNPijw3&kh8Wj_Eyd0GCiL@=
ziMW(8R#6L_)!A%%zYJ68#-ZD~H;MS3^2gb|Z!@k*eGJskzHMo70Rffnio?h3dtVA|
z-}3rIBf%qd%=abw?X7IOJ!kE(gi0SPRN{rnZHIkU;MHWGZqx{`{PNyFHHy!ntX=IJ
zUXnz{5CGHJ0GQ@4m$Xe;KfT`Xk>6zKbenSZ=U12acDmT^S!YViRDGVMQQCHhq_J)$
z*I-2l*eU(y93N3f(sQ%5fqyuk?EXA_M~MLrvRg#Qq!-!Y73V;8XcDqwDJ7V4dAY@%
zcEcl9y5oCilUIFl+*GA{;jUS^K-34|u57*%L*O8+-Q_~;;MzL*$D77RPA*WB1^J3X
z;j8qcF9t<+jVZ4^>}?h`)jmSxc>L2OgE;!Rb_ei3lHu2izHhI%7w@$@dw-<zw2;I!
z{z`WuSXAoDB=M#;`pbg#%%a-UdzE&)X%*{o6`OU*kr)IBI>dNIgnxp!)A+aq%eb-p
zIdWvV<ma56a>#6t&NFgWcBo_ZV0#dAg1k#^^pG<aEgQ>EK12Kg9fROQYJI;pEeg)6
z_u{3Kg~wbFoa=P;R?vCP!R_Wr2mEJca|K=<C@k}sL?q5@x&U9T8yh>4k=b>-e?0W8
za#qRVt3%l0hy0TKf)F!dW=Uzv8?qz6Sz6w<sA}?4CiaW!IK48wejVDtgT0<N-tB5(
z3b634G?tUJtdMcM_N-Wo(wn9s71<_N$%_HXXRyjm*<A9!m!m^pE1+|4ZC-2aYkP}j
zbAmC|4LklCl63d}@uY1w*I@hS0WoB*c3w$k#Fj`3HEW>+)p_!#i?Amg=~b^B=7n=%
zi$_CRa_HX_{{S|{oYG*RvUCzuAzYp!`Fiufvt!Jm#B6=kuf{ZNm0@yDnJbt%d|Cyn
zE)Yzeo01slvSOqHp>mNgMM-Oi1|!*JSR}84!F}-+7#}D%H)#3$Df=}!`b2U=YmR#~
z9RE~cAm?%OZ!hDVt=|?XINWu{*$}P$NaI<YpK#Bm2-_Kie**IwQ~vl5P(${Y(M;EE
zRJ_JPu6fwLs6``hLq?5O+T~dISi<`7yd+Ktr}iH}_eY2QeI5GdAHZNIUT1nhw7rvu
znP2Q5;7q&ZLBHgVO025mP7{;gGP3ue+$Li=>GpN_i9+qdhw6x8kU&&=?>CEe;k&;~
zmd}3Yh>@N9(V$JO&&0EKdL@F->f&O2U+(u&f5{A3S=_#*XZrmr_Dk1E>|3hC`rmx7
zo2Y3<%3YLTCW^3iJSm7SKF0bm%=htFIm1)kDc&?W+h)<ndSxZ*=~J@gXxt**@y+DR
zFKQ>z@0<6<K0hqD7ZleN<owJBQs(gzF*f=S-rlEsdBtpH1@O)4RsQN1GE9L-!)}VA
ziy_-an1ya58*A}^ec^gfr)=uomk;isZb}^`crJqZfsUFg3Kk$;Afb?OvrP6uG`9aO
z_lug3s47)&4or^YtJ@;kGYmLWhknD*nFLb0FHYK4qsl2~D~sXf(k2*UqNnJi{+NmI
zpW5WtscDgI+iZZVXaN=JVEKZI$OQ0lLIAT-R@B=Zs=O()f<JD;MPDYVKdEQRa8*En
zs13#uy4|Coc@M9Q_jQXxixzJ>RH&%n0@!RNp<qsV2nWt7`_$nQ$v|5BE_a37W9M=p
zSP72Mz+;LeV759lCH~7eX!6GkXS(iKD|HNFubiBPveiq<t%Kr+i6V2!!qgf{PE)TP
zw@H=qCYrp6XGbszk@!+0AdVkTUk;wZUVZwDd;8^UBl^vAYNWOjqBt(<S7X`Z;$VRk
z9!`^OMqs?l@4bC`^CEN$mQ&r{n+`WKb#;#5C&-Xo^4z8{mgCMDx0BUTWc$I5T=@*J
zH?srvB$W9wu88>#Yy5UnA^HcwdrW4d3dI!ARuM+?DIrfDgFA>`3=Z7*FY8C`J+#@k
zX(`Y~zJg0r<pn|r$NPZyM;>;~8Vglt7q@=zT&Gip$oq)R31iv1vHoI6ZvQ(1O#hz@
ziU;I;l1BiJ;4hM~Xj-&{H){U^=~0P7^Mg2nzJfT14gX(w?OO=*kEI^eSwE{vtuX9>
zFLLC3)lh6Dz_UIZ?isoY&nsFd6PR)RIn9=0jcqBdjg{dAh{5F1cFxjqq&m{H4EP>N
z)f(6w-X$vOID{38fUKuL!>P=9@SNj#^2D|6Nk>6ly6ZkIJb6F5!sL>{e984g{|qLY
zNB-PZ#c>wh?GYNFFHedB3vWEz<6mLsjiEuk;Q<mhsXe7p<~UB?n@}0I8anYAKjh<i
zOEvpk^=0|_^vtwonI`;4{&WMH$3r3kYZ=y9#R6fGyu|gy^aaIf8uQtq>QDt|NH?*F
z)i#N&5EnX3cT+7)Jum-me2Ry7EcKDU#>~_xYeAf68mli18LY5_&TeVfcDiKx?XFMc
z39{i#ewIjoy`uw<g!~{7#|!QGY#ZAg^?6FJeqSg-22ZXx<JB;|$PNT}L3b4QqaS!I
zjg=(JOc^wtC)jt8PT#CccgOl+6GVXVLl6c7>6w~hjtOPuWeEv}6x4_BY5f|&@~c`x
zG(|p`I$cAJ1Wz_`D3^*;zJTCXiuld`8WWEXV%>V}=vN2Hr1woe;`MuPWU#cP>LY)K
z@jshGo3dq(rIS#(*vSdHey=Df2sr>A?ALvoB$6-a%oE5MLviBU!V6!Ws31hQ1YUR|
zIgcfEK};plWZG88g5RBZ&FKh%RwtNFDdQF8cbf5$=U*+Oj^%8FGl@h%6y+agk>~8z
zio+l_S`9Y`pN;!b{jGK--dwZlf)-vXOZYJ-E@P3Y5!>%D?S|cP&lcEpajH)!P>5(y
z8K#_4x!niI?p((aud@HOY)>dOGu@bP96pES9N$d3U&#TbC%>j()twbY=1V~&5;3B{
z^-G|hS6of%gjSVdh3PC<U)oG=Ksa#{+fQv}B<L7@ak#wzVhsrbmP(0ZL9xOMB?!59
zSxTdFpy}kVyN;d8C8@fGJaZ>*YhPVue%M`yHz-D|D~_bp18@%5&vu%hg9koJM;%CM
z@jp~Q3*j|(<PpON(=vqv1$jB~`$_|ck&hKhk*LDUpo7cW8>?*w*;f>7$j5{6(4an0
zG1c&H`G?-qxTO4C%q5kGH)8FK*I+R2m=?U*=~zTny#7#Aq@{iwqMTIs;NaZX{C)2(
z9HC|Vvso0)HK(SfVpo<*r*W0gI8Z<({!`{L&O7Imv>^k&wa+;!Irmyla;(<h-<O$_
z<IGLFEHdI8dOe0X71ltJ5RB1iN;YRdvo8KYkLkSCF}tk5`ZSVYJRBA{dh-~frmoI4
z;F9wlGr#}y_ofOZE)%NruT{6FE<df((<FKAHE38tpc)|9_d7Sl?JQX`Dgtd{$bBdH
zLK0Rpw2gPmZmln!l!uu^q*A7%U_bf(_eG_zXP(;nzWVfu3P}Jxy{1fCh6V?6F$W70
zT$+ylpSe{2uY5+Vek%6kd1}K~26IFk#JUI-%mqxsjfc5afsG?Rg#=1=Uydfdt2yzz
zOcEkmtsMf{+9L@Awa!8^exf$quog>AV&cieEjK!errKcTM0-or>T{Cu%32yDZ2F1W
zXBEuBk5t7htoo16#3gE-zDCj@yc2p<X)F?Lb^aq~q0Ht|7~uOzOrWme6|o~G{-OvI
ze*t}fU~&t_9*wEUsSycBJv0tdJfWfH7z=z9ux32>dHkH;=s{5D1nOd6Vn3H`TL`VW
z$jis|dQ|^z8x_!Q%IGxy&bTGEK6MZ=_|^wVSs({$&Z{X9OhFQFy^g!WBr{xXsc?92
z_>GL2peoCob{XT#oGXeybHe736iKh9eQ<*o=`=WmowI$kg7qlrbw>4P>nT!-s|Dd|
zfjIp1BG?A;JlEk_aSpe5mRUym5mcI*!@HS9-qw1@`$e3k>muE1rm4?55o`9E5l!QJ
z;>fn8QvMVdd2;M(0{yL#Kw&~a@H@4LZV}RdfMMS$_t~rG#$B3<=@%qnnFQ+@giJlW
z8s`vZM!G9sqVS#xk)tTjj_B8PQw~82=9Njp7^a4PQ5KeD6+)z&fjMS<NxVz&S2^0Z
z;~2R`@V)V3-sC36gYa=I3-CTeRYH(f->y@mls|JrM<vVh>u{sCEJsq2U??jc)vI7C
zB4o@;PJV9P`Od(#{mDPT(6ftECF41}kU<$et>p73$uaD*o03!;l>7OAU4Exw=w0W}
z5UJcqkiBiOGWs(N5HYB#sByv8$~MT-_cX^KE%nMEv7(?p^RXMQd`?c@cb#+aiK?iA
zDn4W|;9GB1pW#z}(D8l!ZmXS6i2j;#%0-9Uv#DonvE16)&IG`UibY{Y36d`S@}q&J
zP5-rXgpC6YT(i6&TeULfNeV9rr<)P^#A$w=PyJ;3J_^0OKl-5MAlw)b7rT6g$xJv{
zH%%g=5eYk{*&cWN;Nvi?Wiq|HDjgW7LBWC4IL9tx{5`yjyFu)JXr$SyWK+Iwq_<Ii
z`0iKJuGyExe2nv;&QKjiF)lmeWf=!YeS08tTz{xTLy*HS-XBv9j{WHJ%3?=uPPFY$
z(@cKKQ(`=v6{+E)#Fe`?{eqU5yFPSQWR9o()6PuqM(>aT<cb_KoA5qTzUDi{9(_~)
z;OH2`i<!d1G$@yjc6w;1SC|DZWddbjV~hups@PSV+IPk(aZJ|G+E%GIpZg9=2Rk9d
zq&I$f?lnjQ6S*O<e_gz^IV0R$kQ%3@fbh(C{gR4;3M1Q1jK5@K+7|9YqHDe&`t@nD
z7zUabNao5svVy4!Ie|q=^1eRa;kDpiz~9jwy*G64x=$qNe@$arq^2#b)mt>A#5-)$
z<AbFIEV-Aewp)IKN_nmGjR$MJ?YDjx$_iQ>ieVY1QeZhfA_#*_>je*6c`k2ezfs$*
zZFI}yQ28Eg+_NTv)&Z1i4LwBy5#iN#-Le1DcH!?MdNxS?p_`VvT{HA`-)pKVRF5c9
zP;<8S5g{9XOA!Bh!k2p6XUM44W^|U^s9{bp-i*SW<_tL(Zb*?|D&YXu@S)#wtizo5
z?{~R|)oHsHrB%3+HZ<baKchkbg~#IGZdb$lqCfC1z3n-tPo&oJvq#vMbCIwscXwxa
z4@s*7hVY<tw~V<mtNGuxg)bc-OMK2c2+^YA;*xS%>tLQ$a0&t4fV3ol*yn52>;hlk
z20Ak*NZPvWAOX%ro>LgHiPr7D0{Cy2uNsl8lPdn?o%7OhM)Y&O)EQ)OQY!&L@HYod
z<OwhdyW!6fHNJA+kDZVXJ-mB{?cEm5qoOx6Si2gE_#6zAs01;_Z>BR_^>#ij87f`6
zP5VL@QO|<AYBE1tRlQ!3hmtSX!P*s;59Sx#cqJnDmzJff_g+684P9HzESh6&g(yK$
zsHCTdP6-5zXq>g%cSf(a`Vz=`kpZjHy|JUus0T+o`)d%dd85!7d;|J!FUn<WxEa`5
z0eCobD+*?U`&pN%@$;fW;~#S!JGBp9zus!s&tSg>>MOtk9}+Hm&=+L(TYvun`mQdo
zeyANqj%E+uryTsJOB~I!L%<^6^Af5Bz81Y14a8*Yiwxr^nP(%?@xLVL#5Wf%YtA`L
z9(c+hAgb?93h$$SNsMA}PEZNUQ2<)C?@mwaE-BqCH|>z2eA7={&iLZ_=NAof+Krk9
zJE`R8?~>e!vccd<d^2BP@O~zuyyerEZw6=4ep~G=MxB~;xYXR^5KcV^vpbq7o(^E~
z)b+&gKEC8?pzt<f?=Je`ttkI3Ye2g@4WstYw??WPqR-`2>@7<qOG+r<YD(%Mbelau
zs%?%gsJ#3mLZ<+Jr_PSv)J#O7)Nej2j|iJYrpfZJ6d-u^MqmP&od@m{DIcFZYTvmn
z(qojJ0kwU7-v=lrZ#LCfd73K3)2dDb4I0K>F##*2$Guu-<eywe)yw<?>@rwF!LN8>
zMHw$S*lpN^75~x=`aan4|1tM^ZrbQX`c=!eG&R`ZAD|$!z1%lcHdJO7g^DN64fn-n
zj0zw1kFT>a+rzvTo{fFlz4!M1ro$Zi{gEV0l1Rb-kBC}Fhl9_Ll1irmr0r6rQv1SD
zO3L(JGT}=}S*JTRaKnT`N-ZFz{qXmuu~Kmh+)lr_A5m3JpruK_Rx(|pip6|x!AaK0
zS(|!Qj8XqV@{9gRlbqz1b6F@C$bhh}sftK{>mJ`V9Nt~q#!n6FhorJ2UNn>GQ8Rww
zMC%OarqMBqxG`}XAa7mP-KOoRVYA~919l!Q*rj1z2r91do|PzKJUR^&X1#%PNqzra
zLj3AgEdC3J85(oXoVl~<(X#V=1q~hOq^MKkdhV8cfKY%2na}_iRZW>30~1OiUUyvW
zboK5UK5zeQnNtUDRPCC&W7#rB1T0UHNAe5>QzkD8<G0M;{&`QJlMbQ~kc#_sex2^^
zoR*q3TU7`@ddr*dq4Y(La)~wR)mjq(Y5^K%F7*t`c_2ON{nF#!c<sBuuOos)XjYF{
z?_e}I1$n5#De3q~Ea+{Ah{AU#Owpu?k#;f1Z>k@UNAEZ39P!-6LgYgsg50d71d78j
zbMh+PdkS4^&s*uN&)kO{G`pQs2HrsONYJKqQHF&TAruo14m9s;@g>8)cxT@9$5Slb
ze0s(wYS*l+A};k~NjhH+d^j*N0gDC1AR~PB#HiuZIUncZ&R$td#f)6>F7a1KF0F`?
zun$nbs3Zk-<_)Xv_@)KBRDb+;kss@9j~&wVSB<G$XI`JmmV41qk*Nc*)n6{HcT@D0
zjl-(M-<vAx2i|0TG*NxW=2Ig&#mC!a$f+NGhb<GRh_PZ$E7<Nj&I-RTU%a3AI{vxl
zyaVyTpL`AL^ZWMhW!E1MF?_Crw>ENarCNUPM)gb1kaPq9tjWYsVh-@95jDWMcP)-y
zBR~L;ik9iR@oT<ms9|4=o#DcFiRtG;Hmgm2x*wQ7RYs&VT1{t4O3(-thy#c{*iT1t
zbPwNYRg$z1*Nqu|p!uXZMHsB;EJ?uv(fF&u^x=nJPBEyikMaHcnl6?Pg~8kJDv~jK
z_R~<;A#vfy;~#@aZB9wX+7W6`m+WZ_4)Ff0dc+FIQ3DIqRPZFrYzJ4n@#~I~Xw|E4
zRr)e~>t}UzkmJ5H+-&fvPoXNbwC7}tKO1r*rTCBHyhx$^pbo#6NX4=i2s<$peKZBr
zjQa;*%28`u@GY@^rG2bDb_>Oh0*A>>_0LMsn+XQzU$}8%is?4bxm)=?lNRGMcG<l4
z(GY?R;7YL~HLFv6W2qkoUE_M1Im!g`AE~5x!9CXGhb`^1PQfB`N~rVaUU-<AK7T8?
z`fpQ}Yf97Uk}4q=`8V(X#@$;-MH&C`o<m8ffCvgmcSwk!APyxpL(VY6&><)#-5t`x
zkOPQF*UZq}DJ`I+l*G^}NS8amyZ7$d`_J9IXYbjwXa6P+4?NFvzI;CK_v`hRHL&o?
z2<Mm@?zzWmR;MVtbZWiz%I{Sw#Xa!A6}0ANWMLHxgFUj_pMIqUw;ad~&7w*68<NE&
zC*0bRg<NS=7`J>|*fZGc1q;U7uNS89cl95K3?|lk{OJO>NMExCrZ>=3hg;kGlJN8i
zeS>Lj37e{RKh<$Vk#L34$afyBkX@PdA^riF=>7p%Ch^ovPSoqIe*i1pI2;ULBdMcN
zA^vc374dNA%r}21vrZL3J63+EYLTG2ElZ;yFKajFl6q=a8%PVse{vDzsP!h&uKl|2
z=t$a;Kk$7s;!w%-2y7ne$K5-}D=;y?`)Fva^o3g?|GwaUe!sG!%_gyX=5_}!QRBJH
z@9EOLHoc2G(YAN%F{o^qJcODLSteQ{AoU>W;=$rKYU89qE1sQ8iS57kzwC1A-?$>V
z-@v7|Vg*$Yg9U*q?8Z@hGw0suRl%1W`!9SwDz(34=$^wLWM-|aSA%no*!}_32>ADb
zqrbY{*Dp`M%ri1ip}<JS9k*J3RYZ_!VYckoRi2VuuNU(rc%?If+BNeZzMs*jCBe9W
zXVc?sy~hHVbSJg@18BG(SOz}3%1E%;@TZyUn80J@)E%^E!l-t4vy4VKHn+A8djkuW
z<2)X;Coz_5dHmwZg!?$pQcS==9W>dA+cPsQKaRdl<St=s#Vk{3zk_pg3P2rQq;K;Q
zE|PE&=0~n-Cs!t;s&r#kRl>sM5wJ*xThHtz``*;B3xlfrRZ@I8&(!QO`tpz>Jw0|s
z%GGz0d%G^(ed!H(mdj0Q{-ua5VV@JX!A#0{3$kLcQ%N2#X@Lpb8EaB<`CRFS`Vk5{
z{65px2OFWmh{@v;rsapC6<Oc<IA_%~^`*C&O?jU0*4e+WFyhQhaDO;aKmmS*f1ecq
z<sR;9)qiX(a+^M36=HOg)~tNdY17z?@#awIBDTA0kmwT^yz8V)^`07kFCCUewAqPG
z7<oT>qC3uu{;6c}kjdZpQmi&%a&kpgis8@h5$TXjY5%1qezaLMmeR=uN(on-=Ezp!
zae{`RSd;NjBtT_$g+hj9rF27A`cZ3Jo<VAXvYP$6SHAcl@ij&YT;W89U*pfWMkK`|
z&+eR<miEE9l|I8<G#f6XkWw`Hp`}uPMpUm{WbQp@_`jxy{r8BW|1Nv>-_m7aDhUnk
zCToK3Lha^C(4~U3QP!N7g+5JS-wdWx;b8oUoDyFwIrMG6!Kb5E8*9_vc(e<=cD8PW
z8rftwMQ!mY9-?$<UL|JdBp`3P7mfl6w_6JI?+-57@d;OCE2nRBhnD0s1!c=q^lDN?
zo;Px*DSDU+Ui;eB>yE?6rWP=Q=As}&V_BnqtzIzznN*+C^Gj)#)b8|=RE&{?o({sb
z*62~e=>1-K5p90*EqWp%fS2!?XNvh9wN-pw&sZPkUVm<#T$ZO8UAc9u&p#q-bW;X;
z{jP*l>5EE1_N6q}k2D0`t7w?K6mRh=>fwr}F*?oWL`ZkBz^=yNrWnj;9S@;K`P+6v
z5UvOFspyZNYPSR4WCh03DV)8Q37L3*d$%Z?^N+<+{NDEtT&BiStX8r>Zo|9|;*EP4
z0a<xYJcnk&5-l0K-su_ZW6tMfrg}%^MKg}n#$}UdvB=!d@=m#HqA(C=!=T(n^I50l
z74w<@Vy(ujNrTwvjW-W2C5DcngTyGlPA#y`N=a5~b!}7l#<wObws{+2X`#<ADWVm9
zApFo$0H;(ZIopGbim2<%O`F;s1aoh0a;MOW4jl&!#8)UULwrSN)!r`WO3HTF+d7aU
zZ~e=LPF&R#O<Rf0)k7>#g!y{xA(yTq%yM!H>L8s;N_g*~#4BQm5{-fmqg)5xdPsC;
zPB?ZUJ4xKH(QPr~K>IAC5D5+~U`<hQ#SdZJ4TRg_MIxFnHgT0@U$z2L#Go{B73Ukk
zzqrhUfbV=vc3n-QbOTPt(P}a6Cds+5DRnMYFo+F6Bf~3PRN<{T<}hvJ@M3VNCt{lx
zE&*(_EabN8Bal0et6=g++P$Z3lUxx(e6q3nWSXa!S;W$6tZ)M=AnKs==s4>D$lASp
zyfekTzSHzGx7c_;^4SEx>q7?O{8LTo0Kne&uDmwqfcE~e*@CuQG`LpXs_2;K*|K7h
z)ZXHsh&Fr$_=^OxZOC++5q$nw>t$gX%8k!#>FROCd&c{$pqj(e<~!~0)VFej)eMNo
z+_3@>$iYQwUE|zxv!U(Y2~e4$0YUr~|IPvNOUm?-oa%=;%Kvye?`ZZ&n!@&>GLO-(
zOjC$4xYrlAONI4V+4DJRo%8g`hd=45d&Wee{ws!HawWs_`rX{;Q1by9lM`WZQ{jnj
z)6TnJ-28B3;$Vt+%I;!nc{Yzxi-@uM3lm(Ew2*sKF0yVJ=t9WV0N0v^#!#y$REByI
zb$I@fqBMH{V1bp`=J=90T|ymR{{&hj!4EfD&{1OVpit;wmL#epZ9E*x%Ny9o?(p!}
z8K>LX!V?rQp~M!;nOT<^=ieoKMyhSbcax{oxS8DCXwh3pF8|g6t~0L8)vU)gz4~cy
z5P|&H&K}LfuG9;Bsp5t2l+<7k=^;{*;6$hAyOb<f4oe!+9y7}`4nODbchP=JPQ!p7
zME(O%eUM^KvpL<<U1+vLHQmgJ;pKqrG?A%4N=QiVu{taPRhXR}T6<2nn*-XSb4ZN3
zY?Ks#=Ll?V?fvrTz=P-SI<V;AaltkXYEFe3{wX%8(;R%3TKTztstUp6qcg#$(k~hR
zD1X8GQ>sQm5=uSR$HVS#za@rdQD*>XsD3>^ZM>hRA}vd3)9kVGVkg6Wee>}>JdQj$
z?-{OWt?w%SiYd0sAd7E$4DsRF6vkFH=GhHUKE@4Y)zCTW8StJ(Ts+|iq!UD@9Zfb)
z^9rd2Rwn&Ry8FE=Vc@kC+9l)J-5Z)77@&}G@ZF`WGF6>#2=I@*^TL;bx-iF5I=abv
zYz9bI>Ge1}ne5wW8uF{|%$W#-1^fIOZF$G+qUiL*%xARP-!hYXgG2$Wq)bNF4V#|+
zy&83xl!kdrtcEQ<`5px&q&s(BQr#C@yJN1l8o)Evs7XQZ!G}rV)2;2YSwHUI22g8K
zs+gV5YNlHMkeIa(=%Y>ese?kx<X%qY7SPdXH4gZ$n=}1%68<XR&b&>(^AGSL{j<O&
z3aTEE+ll{oKH|GCaFPA6AR=i`{IoQfPOz{j1IGCyt=Fdu|K9g9OSOB?s3D?NyJ@h-
z)B0Y0ASZw4_#6JQScgCct_S>g7IN|IFsrhfDUx*CZEyYpG6cBkQDV}^T_;obv^O?K
zS-m!OFzx6k;qRfa^p7z0iiE5V=O9X7e4dk)bneG>)us>Ky5OWZe425SZkljs?XJ4O
zDHNI<oNy@BSDm6C7U=b(gel4b{>40(4pyL`NS%xNR=JeV=SkXlHScV8^ttbmV4@xB
z{@^MkG1AVKrg+Rh4h$7jsdJI(dtOa;g*=yruIzk;kD4GGV6e{Poy+Jr2scpPgq~iC
zs5z*f<;{ww-0#OKZCePSPaDHy4Y*5m(Ordvc;YC900$gHC!590+TVhGzn8fOld}4O
z(`Z}UKLqF)3_pxCNI;&%v`o)@@j2YtU7~5S{4hfhv-R1X4mETax(XrSQ-OByGLHa=
z+I!Z*KlX{kGx*<5ylA!+;D8ecLNxf;a%{N+IGbRJ<qzhmuU^#?9*$%geG0?s7tlFp
z7OxjWi(pnOnN@PK*82y$&u+^H0-}KWDR)b6Gih8mqb#&p%^=D~IekVD%%MB&#N{4E
zUSb|gtNPFhPbv!lyCjV#UpoDNW$XCAvmyT*f%5;^Ph}Oybg}++r}s+h#KO1^qM<{q
zno^is!w({P!_E&@J?#a_+%l(GN=uFhUw9j}Vzo@+O3ZY<(SZtul2OmFoM=H!-ymgQ
zRzjwfjupx`Gg3_pwQwC9pMqSRCD=lA0>(#zR${}-b`^E%vrw9T5bd#T^Vy(a9?4S2
zcidJ@ZK03pL_k2f{9=}JL{TU2;>=?0{h}<p7k2hC14a6LyGQB|5w8W9o%{Ax{uH%u
z8o#t-y|l7h-2M3vz$u858b|ff+LD*#hrIgI<Cdi%PuR&X*bAN@jZ{57<GMqmsgJ7_
zVQ?KD{4V0ol*P@wt>`mQ8~T<uX<j*(R8^*Ljm^JJxV!s$*|2K+3Hz*#4ofE{_l+@=
zMcy=+ru+*FfxZHs<svC_ZZ7_=>M$^Fd5zL1r))N10J`JWg?8$6$d>1L3Q*AoX_7j}
ztY*BMZCF_z#FWl9{Iy1xNWhMVUbN<I!7M1)2>zOp+_L5Di)QPYoUrI;&(#mdX%*E^
z%0tZYP(qGGh{*8x5uW~0cKYWay5?pUo9=PR2wa1uKJnXv_ttMUwOi+d#mGj7Qayg1
z4lkmlQueZUq@nK8_9%)I?nwMHk(WTx>*0c42flLX(#`o9H%V*M{B7<8nYirTaPVq)
z3PPforCu<gl1O0C0)93@3F;&;q4%cfdbv>s!L0bRnFt+Mz@}la!W81sEDf$OzTzNx
zNpfJ-)wPO#+w@K%+=W>8+^#Sgq!}#-pibt(_eQ&_Qz^U$Dt90JfG({4QcvNxT?v1p
zk{3+dC9!VO?zeP$Wpb8)ks^vd2N%{Ad8Mzp8~e-V7&5<l3ll}N(f@Iwa?Yse_wc&t
zwO(7e+_3z8@Pt^`-$XR5aC(~>Qw)t^ctq&r6S+MY*KT%)1v6LUB9i8h*_9AVG}eOF
zf;J0tX#?+S=PhZ9f+HDv<y1Mpr7(z&GL${h;=J&h-Q%CDQ`+-EO_?gepL1PSFvVnE
zu*)&wxcIE__0;Pw7YAJ7(Th9~Mi=zdy@DiQl>T^o6;{Acv6sOVM@Vm}tG_bMF?-RB
zbT8jou;NT_%ubPj&*jEZznRJgsB#v7h=Y7dE2<!Wtt1ST7c?4_inU?9FRuLgbh(M@
zp^^Ip+Q7-IA$s$~l4ZxI^GOx^bzc^;Y|>y0AK(fkFlxel&^EoUGz|sq^L^3vlsjGv
zhD`_a8~vWfu5a1{bx)c(T3>xiw&nq$5;H=i&)ALG9{jS*vKpT$XUba}S+j5c{UWL!
zXi#QOqF@VEtSpXMq5*+AxR~OdL+I6w>W5%VAiMf`!M#a%e<}LXGWRP81lkf9*goC8
ze0|lQ>LB*x{_OxW7b*}zOOKW}cm<=&>_FZ`G>7@k{rD}2{xHYo%Z{EdgHzw!um9N*
ztju)TYI4D_ciK8CvvVa}-%<5*Pg<<*)#PvU1PR-Sv8|W_J!P3L^3mCD4)?jxdgFi@
zQ%SjWOd`V+wJ2e7P+nX$_pXV=_N=Mc5ert{U-{e}tTd)5G&KtoWJ%N}q~i!D+wK5f
z#sxZ(zB|GVe0emWQMoghv6a44rnNNy%O*^YDdrPljsnRa?B6AC&6F^#%LM<uTe<MK
z=mW(>?cEQB#Bk-9A!xb9TqBnmj*aWVifO3~uD|6;yMBr9@S*gJTEArZ`@M>81B6Ee
zQisHJyljbNQY7K3&--NXEAesgTC2lM$*6>)yW5+>U(dECMr!**-91L*K!+^&Z5v)Q
z$)(5BIPG%hC-Vw;Rk7G6PakSSOA#G(p`h$SIy6^JE|FRPzBSKq+QAcTA(p$SCIpc?
zL@0nOFPCwttUp|`q-F&?@Tsb=!tM7^caeAfi7g9RvfdFFTkMX2m~VbrUP(Sj3coJW
zxv~G@(TFAKLVgXbI#tw3&Q)R==o(Z&%on^_JS1AWV$t*F_}loa{6%bLodI?&y&@Mr
z6jaSv1&}1_1nFI+X?$2RuJ@XK!TK!4|A|41mdc0hsD}$W1faV_=SpETlrqrJTU#HA
z4_6y6j#7F`cdnu)>>rT`u%)}L`-D8whW`rj{g}DLWl^W8XIY)I=<N{FJ#cDFXy^{B
z^|bc4-cHPq#Ief0eI1w`aK>sU=F(1o1jm}T&6hnXtoC5O)1t!<&eL_Gtu;F~!%0C%
zcxBjo=YD53bAAVgOuzVp7NM&YFnE$W?wK3&Sv9Um_60zD^s0~{Hgo1f!*i53ugwpk
zSG{587>y0@=5LYs*dY<?PqAZ`8GN4p6!38Ocp3@Z(17*{52y@OE$oXY+$f&*d{cKJ
zjzRNn_d_~>O*&a0MEiT4l8uH8n=U+C_jM-&zPCbjKum2>{o!uO*8?NI)bb2}Mu1;M
z;vmCJKa0nVau<+|js@WC=o~VEOdKC%JK>#VmpnTVDPu}BUorT|D<S{DZ@O`QpUG6s
zr!ZvA9-X@5SXCtUhJx+~l#KbPoCukhMa{AwXSN!^xHjx+H0!apduktS6De%w5>nu1
zY1IUVLH$(ODh7JJ{sD%XPZxaGz91giBKm8q#T9J7L<j9a_yx$c+AWbCW{G~vEX`)s
ze_v8g&Xp%W(HQryE6aJNYlu>>9T@XXB>gV88(^kP`8<VzV{|0lvA8JOAuMk~&J^5B
zn-Q}V_jb!@w8{9`n6hfitR7_Q4*jX6!N>voN`W{YREbn2#&t`Ydk?+#s&h6K|Gg_}
zf@yr)6uzfJmZ41+87-C*A4V-l`s=#2ZQAl^dEm>Pe}EF>ty8#)(8!R^@eoFiXo{GI
zoTyHo*wQOTG;L=2jK$&g_z+*?PKr?II8SEDFx<Ns0nUjzRpjT_R0hL*@7-j3^=HKP
zU!@e?DeakEfFXd}%W8}QWr7!>{uP$YCfi%ks+?c<a1G`%YnhTCe6%DFZ;-~APl=Z8
z8gCeivGp3{&(K6omkL$1fYA?lyB+>0v0|bSk#uZ(A1SG$CEf~wbtEOw4oGT~E-&HJ
z?{_}}WpLGFHtDUS=4B+X(@S^F{5ZOVrG(R@;L8?;iiL~%??S)V4Bxho+T|4J4RVwi
z3yCxAr}e403le?XSkdxez|45X_GI+59&YgTWgl_dD_K5-@x<DeBuylyE8)fsEDxb{
z$@qok=m(j;G~Y!YXUz?gmC0ITRQWvj2|CxP74n8p3PwR3!0H4XbKcc^hcwnnIy?6o
z*Tl{g<s>$}de@so>_SZEUc&qB<8nsJ`_;u~hP-UQC5B!y=^F99FAkSArp?a<3FLP?
zvpRzu{@K^6p7RbJyPzECoe-PaOETIMuc^G-WNS;23eiPZQRs{>kaM$nGADL<t_W*=
zOmUBLoS2%i??<BD+L9Ba$oauqx^k?o*CPk1QNOi6e5v__JRtckzIBL4cS9UcRG}Ns
zitcW09js(toI0N*r8k-wr!QMoo8Y&o+c+U+uDb|~Id@<zIj|`ND#u$fAoKnCK&H^R
z<9t|Wi;fF=e%M%d)V+~@FRB6VuT>Za;)Z%a2>7KgT;~(^Qnf2=f15QD8B5pR`RR6S
zxt*9v<YFWS18Ax{7m*XO1yPE1i_eKLUswE%-S8|fnW#g@Dvc#_0)fyD^~7vy+ORlx
zu6WlKW!I$@;F~1xPuBDMjO&dRy3@MwsO1>A`YV3z$4P9XmhpE84v=N|`#U4%kM2f4
zQ)kzc3AefXmvwEH+|SEWFqLreNr)h4o!@Ib>X;aqg6HbW{Q59)wfFaJ0|YXu8b-!1
z81wc^c@trdCV&M`ICS)2s2I6B=6#g?=f38Nf2_nyLECA?3eg0(B>#h?xO*o>lIKj4
zzKBm>FXnd2r&`m?qLJj_htZQ`a)_8q=K`WO0_M?gmKDPY<g`QJ>}KQ5D`L0PJe!x2
ze1nq|F$;IWpo^x;thB3F{{Y0U4$F3nqs%VECv+}DV(7OiMJ*GP<Rdf%fgpii0;zrZ
zk-O+!d=OV17RZ99Hw8brH)$2=*V!+<_~@?>b!*CYGjT2TJV!pPXbk=;`4z6OgSs*<
zoT+@>H33ueTO!2w{s>qtY^;!P3&TGd7TK`_O$?w1Slx`1qqgeRhLxL(P0DG}vF^0u
z&tbiv3HSvRD8ilh;x{u55mku8T8RUz7KG9EeqvcJC!7}&{op8{YLvtjTj!E>YFC5D
zV8R2r3Lh+{ve^7;boTt|Us%^6v_hq}i8BC4U;fQ3U|pd65SZ1%O2uhn)ohpHt^36f
zRgAQq9p^3DQbuMcOo!+swr|E{_=)YhF+~g-u@AVNYUExRx<slyo*0M01q%6>^;yMe
z%a(AGz~-1o3~i$<*CT(wE-IzO!6pmD&kXuQ|4tULtA&fHg0;!6!E9%zMD?z-qHWu4
zW;H-3hm&UD^q7K<kRoeMeu{xxhG!zTUjxJb9q0Y-r{0gvu#@73d#7Q~gfWXsLW-bd
znvnb=PPPX#eJqCF5$0<c-`&)6`$K_Xr$UkZDAj&F6IT?|E%mVnQa?52b>qGSg6t>r
z??uPV%<61&v4<==Y%B|Jq*%VOTInAdbL3Sy3@J3}CJ+X*u!X8Egu~Y|$*f-&eQ&m(
z$}^Zgn0^xL-zAg4_19`kI|(Hu08H4WXq%^8J99L2eBmU6CkRKMnd(KyE$h(1@l`;K
zs;roc)}Zf|SxiW`;DvAecrT-nM=!sU(IKr8F@-Ko<D-&$SwZ_+t1AI_)=kt__ZkJ8
z4bbDIwBJz^_#o!9Tv$GB-h=?~3b|;9^+sW4UkV9KHyiF3$ElSS<QtS?l24*hbWF#`
zOSo>S(?5O}hE<M{)n9fW|H4gJXxP%paRUR|i)PG$lGl2&pJu}Kt9wKQjAHBeu1<QR
z{{bqE%G`d=Ap|(ksipjS2>?M>riW*(_~4V=Y|Y+B(~l<Y^Fq#AAb~X{@d5(dyMTw6
zqVd-~M_>O&yMK8%(nEKzdQ$i1RgaOafr1ZhS;vD#Hfa?q`?LFhkss!mf6{K2lqyCi
zh;kJG3YECbX#5j_^StX_xdvsdW81t5&|x9Py9N_3usn4T?;iKSt$}Q%kkxg@;!uB=
z*8e!?>r^CuVUxXPxaxT+-BtRMlksh_K*S`{vZzW^j7*!k?Pr8UHH8SZvr>KJ3Ce&E
zV_8z1X=;K&+iZ!I1Mcz8u-zlLxHjJU&sM;KW%^cX_TcJiVE^Fl>-F*P0>^k912<5B
zH(f$A_&4|a=MyqN2xRk;9a@Bi{e~S^v*vi7E74e7vc3U&DI{gKoL=kyIrYet!_P#M
zO0~>dzZJ;yJZ381<D#J=$-a7Uy%g`Kr<~vve)8HnMaF67QCLtunvaW*Whri%B`NZ0
zx@MNk?*kXdPqPo0#S(mB6L-Oozt~bRB2Y&X5$->98W$v`oH*+~iyz)AeJmPNe^26r
zjyogR?x)`iDv^?wUlDSZALpI4F_&;ViPA1hcY3uwm^4;a;&smh;1%Hi>&?`nh;H))
z`P_M6e3`BRc&tD%8Fd*b5Xs-quMG$h&DL5!orm-@MOdswop%Rd3!D3GP(YxYnLFU&
z1f?7UI6RQLpZ!pNZ71;O2DT7)t7cp^`Z`>346bvMHgMG`yC1~RF;sV`g+GBn%4?fV
zI{k}(o%&293)h)J@orgi;zUpfgABk7(DkLO9mg<eN?Co4J(+E2j=LywYc`TF&aE1)
zon0eq5KfT*a1m$oeH`<tigD+<0uJMb)_%Cyo9vZ1Os2@r(i9pDTCea<Fn|nvLc*bC
z8B54-ZJq%HO_L6P_h%{RgeyHuSpPI!QmlM7+1@+s;G$OcJvEp6C30Q6_ItxB^Q=ax
z&aSqWFeBIRv#%%rV4EeDfh~;BWB({q;v`3ap_5^MsTq-48hukechK^s8B~=!TPXx2
zc6NI+L8*cW>Xqv!Y7^CD+U{oc+r$q6Uu5Ad1Gza5E=q5^Rj%N?Hx3)aOi(FLqx|Y8
z=sJH11KU;z6?QTdE!nGb5cvFT&EYxI=e$^>`Qa9yO3(>Dlz0woa(Ny2g4Ji0X0-cJ
z1OJ$Mm0bcm0q3Q1GQsu!%OfkNyrfu%N9Th&hH9<}aI#)n!XKH5fvX9Zk6+l~13!!>
zWurSjD<O%Drih}V@^LvWU0`+0a`%-;s$s#SIe5)s-GVR{VFbq5D8AD!&Rh)wMjedF
z1J2FEQ%7Pgq1)+u>7ra^C5`UgjK0O8+4)NWcB=lu`UP-VX=moAcox-4G_i<m%V(#+
z4C#QcYx_+?z4u4sUo@U#TUqW4&>|zvwXX#_4?Fh(0IPOS%LT7chu5(s4$AvBfoAam
z$?}ss=I=~Dak5DMc`EVSN?B6v?P&Y>e<ZHWOF2lEiS|#qzARcl`#5c0GMh~^vX<f(
zZ4t+pm#=hD0ZcI)^oRa?kM$VeXW2|EAtrA}iZo<PfBxLISY(hizvHd*4`6g5`VVlS
z7%2F>e=D&-v8L#dHeU86)1fI9@_EZ{J=el;_vxN$*0`uoK-~9Q3n{7VK&y|5^cJ2;
zK9T%e^@6Z(9BcP`#MHw)5X!V!>)MFlXEM(FZWHtR_wPl6GOKXZVfefr8#GyERa6%7
z`?p`q<kY^HLptT?3#?B@5|lc`{Ig1r9t*=ETL?Lx@L3x-RQr3U7V;q@wNayH*pVGp
z^HwI3M!vrvSQ$?VVCWc-xyR39vAI$2RKx_C$S?HbDnu$KEC|ppbhmhS-u6WtqGZp6
z7NovJZ8dr#qDQ%X_D*Og2@5lG_;$#(iK+8fEvdEIJEw$OsQiANH*6tp)(h8kPByVh
zU0xJ}oM0bCJ%IrhUa2#4?kzT6^@<<>;etAVWwDbRSN=d}6yE$71?1!L4nRL3z)c_|
zSx1gc1{`h@-e=ccW|oiX^pTo8oBR=<ft!<bb9lA#AbVfhq*r*o{<2%!>R6qbCEU{B
zFe9_PFV67}HEAl4)NKCva&ksW_@MNL+crU>3B?rzhAKv^vwkA>3Y7ol{UnM#2V42`
zAmgVyIwiO0cC-kc3)_Z$Rvnas^hGNNOuPQ_Yc>|Tg6j_Ve({?vP;wT)6dVsO>GN_0
z0jn4DL-4M~D+Afo!F1+c;JELsz>DTdnfoHeA!*s#e{{(Y;>-YeNA^>J>yjAnd|W7Y
z{o8{xS?WgLmN*uIz9**Cu~{gIh24c=*<BbnVbYK7lZpb?p3KEhOMXnq*Yvg?LtisT
zOXuy%x54($<jS!E@_NT{gC)U<%6I#Q98+zjV_*APQM)jY3@aIaW(NswN^(7zpnwO-
z(`)no^=5UigPN4q>JFak0J)98A=Z;iHGF1q4GoM7Yi9}K?h`+)jnbR_({1iW7GOV?
z^khZ6eSR7w);%8SO%+(@g7TzyJP|(wN7|rX)BPlv{6#{jN`w{woiPvtx#90#Qe0$W
zONu#CI%|HG<tk<NEBt}wU0g=W-Z~F3X>^g{c51udwVDdN5K37|jKUUa64I#RNl}^S
z40!oDfZ&JU21I-Q>1?eyNNn+aISl5^Mu_Hw%4whDu;YERdLFN_vxUDLd>wh_Y08UR
zbq7df5yuwWJG!#a(0_om>-~&3Id~7Y`wV^|WnWhHd&`zWFclnDK%TK7(76Q#bq;_!
z*-l+mn6^bXo0PJQ!}#w+f96Xrx`bx#AhvVWV(Mf?>Svk9nm^!T2YiN=`cac(WgHDD
zRW7Qho%p>`Cof5ujBydFfzy53URt-Ca@?$Cg2GM68y4iwEwHbm4fPrGMG-Au*f&1B
zb`aM(eSqhK{Jzq|Cz!p_^VW4ED)Zd40(;+tk!gyMdupWYdcX=w{1e7zCF!eav@~!*
zWm&WL$T-D6;KV*|Oi9rhwZ@j1NT$uthBi%m%3<PCJq`E9FW5A;eM$Ey^f(F{d(Lxo
zEwvnQjgY>$<FA*^VV+<{{%x%HkU2iHJ;^mDM9fV_B*1&F1rXbN(nkhyVL{nco9<6O
zH`#OMdgBXr=1(F*gEf7_sL2^l>4g*EYkw2#7`gOLP4qF?%6r^SN(3j|a(^he&hS2H
zba-M<nceI^K*|DM7*t&?@-_#(V}OCHVw|D~wJGEfaUYm&%HT&U@tL%fLHI`V$NoT5
zeMbx9o=2gF_6U|4_|=POTewc)+boD2{vc|?4CduJ2bME!;pE5meELfy+N*Ie{L$~{
zGzx7R!OQxIMdmxZ10m6kv`bE3AR2MB<8LZf#MIYw0c=Cd-f+C++kCAk?tfrA8#X`r
zHAIJZ{PI7KyI`NL76SR2=<h0iT+8mj*xQTkD%J>x@aD4ZoK$ikVpc8w;1+wrOp}gf
zq}GMj)u{HGzL*@?2k{MPF;-d7Yi|=1W0}qHcC&y!)j9j?p0%TCy4x&o<pJ{$OBErY
zH8eRsFEr$F(#PdwEa%edzRnUFZxe`fo_fJh+nR#nc&@_;P7+fImtzYd;GBQ(E5Z<c
z^tZ)f(k;P7mUGCnMn{!fTs;0#0V=~=!Fq2gAt^@mi~E=UU-+Ka+B5nv|EF+{=Qor!
zBo@a*u`RtvsJU1P+XrTfZ}&XV;uTtK6zYI`Xgy%mK48@0;?$*<`>D8i#n`xVT6Qow
zQgVDLG&qR#7r<@Acm?Yygx9TCVy%q1{`B(}U}AyF49}3b1uiE1omm;`{+Rdx5luVV
zbaZn}h-^ytp-aQb^?iTU6x6|RFdkRAb)8;Fzk8Jr+YwWU-uz}KBcP?wA?xJ(B`|gG
zXhV%8)p1QaHlURnOsAxZVt36Bl!rddu&56xYQ{O-DaG9%TsD3;%f+~0V}P=TF-Y;V
zGnn`uTP~evdC<x7*sxaCotRlJ&^*62cXxnRA?91ApxT{**y;BTW8c!%Vu&{6k74l-
z=L5L8iux8tUQ2v?6J=d?+_mpuT4Nig1^GC3d}YDHE3%X7^?rQt7=}&3*u7f`LP6%u
z7>tbnjwvha#8XAGosonwo^{Q=Ya2C0cy2I<NV|y&xeKI7(2{j&>8hZYV@^Hdgf4GB
znc_~m%RUIU&lQ%Yz$P&$eD>@GO=3LCb1&|(R>{#xf3xKSgL)POrmQMQt}K&G)hOG`
zokos#4}l2Scb=*pa>#p|TRF?$1H4cBmM>m9!#B|%FMz#D><U|d8(!ER_Cj?F3ae4M
zH>i5^O+8F-ACUDE$4lt^E|6<(-=@}a^3{n3*T4y1_RjYx{yiR=nvDj|?wJjVcU*MG
z)lj%X6naA;FHs_<z=o3;FMJVUtc&9*hW`fNJhw6~WFY5A&!K=svRd{hlCzrQ!B(o6
z<vr+PtIkj3#uN(dv4m5vc_LE(9Sm;^RxyPz{9yrRRIYD$R?XBrESVVZ(RXVu7RZ$F
z8t0}M__9J70682uCAm6$j2c3){I>Yz`0}}5vs_tI;L;^~uxWb;2-<wg+ug^of7y&c
z&3KlcYGB{Z<(EVogqFHN#@q-9wXC6K)xpS_zD61DaG(><xKgiQjzj58X&VX%<KP<A
z4naYAv?+Ks{Uc>$n{wX_ibx26&z1kw#OGrS-r{?YCyBS-pN3F*$=mj|HIpj?2qmg<
ze&_M1hfg{GaK*&9opqiw0l)jnHOJAE$Gwm~o>u+#md|dS45>GUB2h_5nt;kNk2=ro
znQI(b%da$5ko7lfB>cIhd`abP7%S#F3{EECD=LL*b(|2~d~1x0@~eg)25e{rJ~1yZ
z3==DWg?xsDq1f^t{!K{8`nd0Gl}gatJU@pxiJAV)kuF<wX{yDNLkcuj>1Mq{TP}U<
zidn$G3wXD1>4n^azxYA&tX`n2%m}2Hz*_yL?q2xfkVFkF_@kKIM>Zt}trhuOYJxG9
z4nGs^{Y#B6@$R#qNKP|`K{9m7EU4bfrJAgFRQ~{dANv@~xl^re--pb!us@oC|K%`n
z4@<@y56BVVxcTn!q0`_Z{R?5i*iVV*I)P1ekzjswwCY5D!Wc|9FNRS5%2;La$LQ<n
zz-nDQ3!g;=pQR_$JC6<t%u5Gf6&NVk+7+mRpz{3+comuJ;(XGIAG719*9}Fn^^K>M
zm6(<x+MhSFW0zhhw3I$gap5xe*t4YY$?{TZ#GKuy3w@@c^)IZOFUqfLS<lTcU3pVs
zM^oc!{?at1?ONb;M^)%$s~!}UaYFU`*2-l~Sds~qH;jD}l}#f|cO&;{qEDXm&#D)0
zrJGL3Z5-weY7yo_@!XH48Hlqy?tOk%EE_<zypzhgJvKQj#ZflQ{d!w*>es8kkAv^B
zsY>(t;^ua}xqFR7<a0I=_g}2RKJC%@9jU!UQx(#p3i$bZwe~lpZZ~LtU#CsHaZmTx
zaWcNo)c8%}S#f`8*5*kzGUp<wtF_qKA;)3O&|p+}Y@xiSUYt(HpXpPgw|3d)$*aiG
z4^~{j_dU0gk+(oR4wCStl+$9$(96e&fZ39=mtTH>6m>Adkz2vLJ`WRpBKIb=BEqHH
zpQNVJPXDFxeth{261|BH__XDEPLnV`J$W#Wi}Rpsm%6lW*!CF4&<5VO%i3D!{0Gnx
z36@RlzzeXK_LA&N*3KJnS_QDGNyu34B2VDSU{W6TIF?N6`nQEEb2mcyqi@}ZdIJGI
zEZ;ja#jyghr@=?bTxSr9tzO{oXBX0AslJQbFF-5X1yvnBv?`7!a?q?spw+(l=NE<=
zLeW2!Y;$8&)RYMMSLFVXf#q1|kNt{n()VP0;HCrovn=H{Re}2M*wWPL%fMOM#Ieg(
zJX_6%e*}1#LO;CfO@4B^c8n~GmZNHW&WX9Gy`3!%{KX%bY}hYMz4Wu&uDRIs{tu&2
z&O2FS$Ct0#7nnA%7nwyVehs|IFI8XYRAT1ehVTIi)VA^FhYjGHzNc*M630-L??2Qa
z0$b>9J~S+<gExmQ8U5ynE!DBz+qXD-imTN`sw`VCKZalDJC~7IjtO(C&Sb{<y|t0o
zmEHKsrqHZ;VdH0hJGA_z!MU8^CmDXWQ>f-0`5Q7PKNB?(7Z9_M$q9MAa!Dx9XUpJI
zE^PiIRv+b@vYpq`17kbQ$xp<e&qcq3iDN_M@^idcUn1pC&t>CyWRG7Q-fB|);dXVW
zy08D#&(!(Hhfl^#631h2ge;@(CW0PiFd&wbmggz}fa466F3#{hLkgwc1zGEFNx%Pd
z`84fkDpK3G@2Po{=GMx_2jIuGe|v~ejs?O|P-?;{E^st~%qr2v@_ftghercw2|OU|
z0mvC?Bak1<MW@7|syf<Td>+{9%uDS2j#S9xz{96$px7$p)3%N6TV|7t5EydnHl8iZ
z$*2!|ArhW>#_TXz9Wap}D-SLVp<~y<&8R`6h;aA_tF`%$MeFsPBlog5+9)0cjJ{}2
z2!4i_j|Xs3ZmsE4B!z5KC!ho{6$FPs_Qeas7>?Ca-pkVaF7^;|^Mt#KKy&iAPIf8g
z%A3)hNXh9Nx-FdXO~I8aMHKHzE34L_MV(l{y=Q>~lkO6v{D|?1?~w(?bwf<@D$KSt
zZ1FPlgn=q^f3y6u8y!x4{khd$wd56Yw5dxjZ4HiS0_N}G_9*{0drRiG?4n-UyIR~N
zkf;}0sBuf_g(<Ec^YO7{Yyi&vOi_V}Sj*E(zO6b;Bo;DdCs0tha~X!`1_0mW-(z*r
z^wV?qn+UXEl)bPGyp1+0TFbklnF<^jzHW)zNIqA0(fw>O6}$Q4E)<6-KfW!(jz2D1
zo_KcXQ6@ev`*S_%!xV{m87;X%K;HP~&b9&S+{O94a;0nJe<^d6*wXx)PgY8+h5lU5
zEN`QU2wifO09bFQ0Gt=Bm`Db&!~vUe!1;#n|C6lEu4ik%Nvn@;{F|ya+#0koA>7|W
z$3p!&6<KcNu9=f!&u4pjF3+oe<)x0n!fgsa%R@)m0q%T3q(OMR48WuCKc3Y8M1Dv&
zAQZN2iF7o?QG8rpXY>IvT_P)O1QpH_`&D~3jYiw+*dp;A4QM>{;RId6Si)j;GQ)pj
z$rb-MU8Vo0%Not8SO^R6bZMrK=?YIV$Wmu=KW4)+oQ}|Y&WwAdCg=<re0aiP-#gEp
zuG&f@f<}f}NOc@ue;xQwLVD|Tp>8x^pQm-vHdct30*nw*NDava?8a+F`8Nx#d@`^3
zxllTGIpLnb<7K(Uzm0dQcJ9-A>i?HKb!NHMOsS7QvhczOeQ8UfeXr`6&MhQ6;TNHn
z3%xzV&dJqHyLHu=iM`H^(tXRfdjI(bptH{X*DRg?HJv>s<(WiEXD7ONKhv%v5iR~Z
z2I}@;0>x6Ybb1lMvMaoca{#Yfdu`*ZZv|CD1|m8f>@-#n(tP6S=Hp#)aaLR(<Npuz
zDQ-k*?VI(J_t<jG56pPQgb$&)q}<011=kps^1dubQuYeGi1(TEWKo)VlPk$+Y)`K0
zT78F=M4_1u?_iik9s_F}ZnsOxiq*!b@e!{7iA?4v^<-=XUjbYojZY)&7$3Iv^HYo0
zWBBT5w+eIxm)1f|1!E<Yemg3}5^{@7nn0f&eMIhP#r1?GWJdVT{>-<1Y#i%`!D=)S
zJqs4x6zn8o=<NHnbr^7gM=!cbzjnKutznR_8Vn^#K7LyCjUWT}R7m}Q{_p>P=PSC{
z)>*|KY4SC9Ti$$zcr*$;6R+%Ji=}3}Wh(xNVr3#&+B1R-)&&?$!#!(zB*f!En-gQh
zA;pLoB(5r$xzqV^TA=U~e7vIeAHcvpJzA<DlYY6lpg3uqE!&J=pqH#yjPf;b;c94_
zddiun9IoMp4bNS3{``D|Z2-ykFf{vNOy+7*TwrBr@qc{{P`396Zc`3j!x7if3&GPD
zR5$W7fzg<MfCR&PhD)uZNju_t?j|oQem;03o;YP}Hvtqaz@qt7x$fcmzTIl7;r1Qo
zO6RVPd<S2o+X@!w${_2~%pX3?jF;o)5+o+M=Sk0gcAZP%-0jvDr=?$tKB>e_PrMHJ
zjqi`RXNettYlE#Es>(H{FUgqpY+W9@ZS0OB{#LZ~4W`y!d`ykIM=M+PQ2vBNGtB44
zBn`2$-76MA<vJq0Hj3A7*T1b%QWG+w=?X0@;8yLLqH!T0?YK6wdF&{4!f;j-cuw5R
z`^6^V@jWA&TC*XzdT3nK$@80ziagiTq2>=Qg4Rui;WJlv)*K5kjf0e2(K%`$)fhUf
zBFR(PFYzr6lly<SV9$1I9?kkHy3uxx3yD>8RUT)2CWUnNX;mBwe;o*1`Dk}5QuhxK
z8X@#&@f3~@q{9zGgXj%*ir_+Yikkq8<U|EkxEwd`Mbi4?<ck*z+~x4eQq<#y6oigZ
zOKwU!Co(upq-lWSP-Y=`|48@XA3#YrFh1#;W+{-nqBrwp6F8YmwHeceLavhK^_^w4
zModKw8?I#~@q1k*mBK_Prh|`b-9F~iQG@-+!HKzvVxsbg9+bxG*uwhCADd0UoX20Q
zTG{=@EPK=wDJg<Dc>&l)072qb%jzq?zQe>Z(Oh%le}KOv4d0byzi8}(S7b1+yh$Vx
z$mi61x9{ow7aE^@ZKxtUm@Ca3@~ViUNzcmV2QLvXd1^%j%&^6r_7ClON6a}kTTv@{
zO|s=+Cdyz@>20|d<F94A@^bh!{d|H+EC0BpiVH_EY1F0P#;Ked%Q2g_Y5P1oJ#oYB
zbq7Bk_$mVyW~tY`_vzWjSpZu0WGysqEgb9iMI}4@O5P!)lbP1Esxa6)9WqgJ`LP%`
z_e(0L@0#=mxox^jetRO_)7zFg>4<KC<Kadr+@k`OnaNx<TJv|w?r55>QEL4vJG~XE
zF+lIV`23y5$bppS9ihsHjC)J*0ABh52LC;g&o7%rIDR<)1MtjPn`0ZfFo|uU%YwVk
z@#Mrb|K;~dC69I?<vX%O)xI_^wKEuLN)L&{V4~Qe1IX3mQ-R*+p&3>T!mXY)`?}@N
zng9!@6w<@J?>H3wxdJlWHn;V*h=`=vGM2-QB<sI4hEyXTT37!PifsDeIiN1Ct*TX+
zxh%y-4HD1>So*G>y2SsXWZy91uRpTa_$vNINFvE9TfmqK&9;r=+V~Sf_p68<G$N{}
zQ*L;WM)@*Zcc{id;;JB>jtUc=9Mk~Se4d<v++etf=$Z32{K7jenKIgHCtl+YY8)n0
zbnl0TKx_Eo6;v;RHD3n^$9c`Z?r?T*GbqiRdp}n~JAl8FM`%zF>2DMGx3sS(LN0I3
zUJ5$gtWEV)7@z}?>4h8gmmjE>bdv{Bob?XhSax5q*kT_$kGnC>K$8a`&e3q<bHd_a
zHvg{+_2z3=w|OIVU&Riz<Rpf=9vs*5$10c=<QK4Ev83mak8_>!V=cphR`ag6(^}Wj
z#vg~<utM)YJIo7h@i*+Ctdms}7M?u~(%P)}c<O1+ebjrLDfPo)j$Yj^GhH;@YjqU7
z4k4G5JRKOgB5+GLaVnYe(*5YWV1w2eTuV}UqkJ!*9wL)=5yiR2tawGZY1C5-`d$<(
zJ_oM&8mi=0N-le7(+MKI+2@g6k2qXkPxp~dWihnh3SqR)H*(ud3nWBw<Kx<I&g6On
zzE>=CHhCi(wRzv;-8+!EsVo!ew|#$h;ux5tFRN5$seY{rKV5&NsXcC2c6{8f<W|qE
zT9Rwf)XGl?B4+tkpeK2@t5S`LrV<O7)?f-;$2S7wPmDPGQuki%w9Y%e^!KS!L%o_5
ze}pO58T*2^1*&#zGEb1scQu!K$F9jJ)jWBhhy6LZOO^nIO{zTN-&)@L0Wdh6(n7cd
zviah8b_E+i7WHMjp}8v<>tU!)439~|)E()7rwz{s+{K?AmJa4T%U2WnX>Ql&PTU(q
zHO_nBP1w!wdd5Vj?7+{{3wF1d_ThleX7<v+PcHBl>0U3Z$6B*@Z1^J2RP5Vf4ZB!Y
zZTolZu7&K2>&(X9eMr%OzR}O}s@{g5NW4G=m-vw|VaqFj>lPEGV&3c%i=sstdUHCv
zmgPsd?$obSlNz*-rH!8sW+I`wX+;XJBEre#cU2-BYk-SnjtCiLT+{vN1G=YhcN+CD
zp3mHTUAj<Fa$fu-;k7f(v%ZWe8<m@{TVVFS<?H7btn16)GhwQ)`K=F@CA5w(<}<0u
z<)FVLtSpgy?UtC|wlQBRPMU7S%6l>_IF~kk3IjJ#H5PbY<i-UbD;gc0Hxj6BM<F^J
zW7qiyF`g9H5OIj+VwIAHoIN_iM@7<uBkD;#{!$IfuF+jrAAESdfG>r|cI&LDaWh-j
z@#R+S>A8|=bFpMDvsqG;59m_fk#1#uB&>H)oi=^P-?B$a`L$2ei0s!fzXlzblFPMO
zjXTvwAsJJ_Fkujn1+QFRkgPD(VhT-I>X-W1Cl2+t*Y%dU1<{*j(T26;w^YUPq`Kt9
zmO;)TV&wF0fzRo_tTaABC}}iJrAs<k{(2O4JxQ@0QPcj?d~*EHp-AhXpZj;|R+2X6
z!JFJMB{l7Lh4B-N?sN$ugwejI2XPg^>Cdnsmg#TB_ycm`dVDQWJbYVc^=4OS?zs7#
zW9nCwZ!GhDq1~UF@1vuPA;e+V4N|f{Ss%|FHBziBGy9cTluQW8rk1teAQuPf))pFe
z-seP1fVie+lXJ*5!KfcjfHNtdk-K%nHY-2_WG*?|XIc;xm%uFmP6p%0l%&mf=l*99
zIsdc#uVR<lEZep%cehvX`VAe`5;^JQ(|X3@D`2mJBnMr0PZ{Y*_NR+xt?R&*`d;B*
z`LJ0Nf+t)DNfUCT)r%?jIBuljtG=EBrVeLw=h#V%;^G^n6R>Fko`tl^4=`toWA$8E
z|LW>HBeDI;|IW!3&oZ-iqRoBzh%(_Y%8VfYe#tAQ@b9luB5l!oFsG@3?MW44o6p{(
z&@MU1Xq%=f;zOO5&*Z_Q!MDLc`+LWSs0?1TGn$ikIq|fl`|&Z7Ykh>d&L_RV9=U2q
z4u4nX{haX;B~1;#u@txDsPi)N=4*MJSD;h#5ry~TWUTc@<5)gNNC-+FnZ?j4xeFLz
zyLTqVJyOzjZ*8A_9&<sl*gx-W3Zh$2=jxV_05>TH5oUz=QbT=ICI9Ri8@jCs+pO&<
zj4ZfT{Sb(w$gLe0c>rQ(qwCL^jKLdvMz<1=oYRyHS?P@GDDM>g&`=V8sV8*&BZ?v_
za_BLr*Eut3Oxbz=ra_>uKYch>S7YuJ@$<(UI$*=tQ+|K`exRk-=~j2rOhlL)1v<Al
zn0`A^g3YR%iy9{)Ja6T*vhbp6)ZdtK`+b=)SmSA!%_g6MiVLY<ej_C9&2?(e%f0$k
zx8bT_H;#pO&(f2J_|+o{q19smsoenfV{eP_Q-jjkq=bg`w#-i!!R&Oje>2mzgP7Si
zy7BdJ(n^3Qu~cK4g*zUezkxK(i!+{Gsoe39P({J<^&_=3u2l~T2=FR6IR+;TnUln%
z<GN3beBT=Ib7(N~Zz1@(+}Y*jO5%QN?+1(-1%2lTAfA?bqFp1<*?1SRJ4gnhYu?g*
z%)0`?Ut`RJ2wvmT3KM(xl)=n$g({NqfiWi^-NdB0A8q5Jz4j_{8zyX*!=BLSZm%kh
zR|O%trDi^sOYUwcZn>p<acPzw0WWXK*WBzMImky}*tPA64sTDqkviU5BEKe!>*fPs
zIoA&N_D!^^sqT7X#A$mQl+)WmEEREo!j|0e$!Dh(q|=Jw{-Rrz1LHeCT0@DL6%nK5
z=bYH$drLv`0=vG{xpz}PhfA8;`1Kr-WcnN6-c9e;z^#u3c62yg8b;p{gcS#=boh_D
zw*MjN!iGI<3B-=7aVwD+2ymyDz84=Xd0J{5+zV^#UfOtXe($dc&93x5w%A}#Fz4aS
zHb-LiD8A>SZqAz@<T@{fFs~f{kn$ufrIKxYS_h=TqwLGiWpzAa(vxPf0r1AV^L|gm
z17`6YpHve5{Q5$26t4m-7v>v0K373>HnA#Wz2c*3FsW4O8Z906G_#csLL8YIL!B*s
zT<js6bLv7Q+UNE5uLS4=<FS+7(1Tufr8Q{wz}OwaDlK=5`8?snZyDl}qmV|Z27=lD
za{p1&_U~mueQ*er{9x==)WAN7d+=VD(9vP<beZrN=0pADm@vj*ZGvm8UpF%`lwdx1
zkzW}4-kicaXxVV%>7vAEdsK3bTk;*)#4tyHqG$nI7#qPnGHIVBWcmGHe#8e83<gWM
zl@eD3-=g6b!^jeX4k<Xlg<75LAMvND_>kT=s3{XG9)pWa-2MoM%?l)ECW$BV7AR7g
z-2q%@=%5cbWSrp=n)R3oyR*Bm)?x85SaBfH<Vto@$nzDI@6VTxY|<<inVfi*(JpH;
z8$K;##-S7A!cBQ4QFIUvRd)V9H)OXIYy2M%JJLs{rEO*h&DT4*Hwng|&2Pzr>I=pX
zD7Sh6gH-LEL*KhNaRI_@+bz}2doSrA0t<fnaM5T5w5a9wMN%+w(Z*e%jC0k2cTU)|
zjVEQn(6$L%W~XEuqG-hxBqqm3<D?J_SXx_E6YayWaIKm^=7my!>t)Vq*w&O!6jlG^
z;};WTyZ*%mP-Wmsl<Tv4|ChAq%VW0gviVy5EzTxA%p`<S05OKL$`7F=>xYtBtgM}y
zvvNcv3j0A2AX9yZ+y$O!#U3EzrFpDzApZy@l~$bN<3^m|`(dlGy<@|$&X-T-1BxbK
zg_;#D{kiyUdKaD$TU{gi=f6#M{#Vwz{@0Yl{{Qz~oHY(m9uz{EbwlzWvWt&-V9qD*
zC)9U9yu}1_z-!i!MimdRCE~;-?m<5XVzA}JgE<1>{Fk?3Z{AN&tz?BkO2s{HYP$jt
z<Ox<a=-=j~=jFV6(M;%T)zQ|<D!S)Mww{<c?C_a7rlhF6{%xRdBc55&{<iDA)xBv~
z%IOcsw^hUT5t9sB2-BUVvc860-?M=JHTyxO@oYW&W%#qpim(&j31#O5RGh~{B1v0H
z(>dF+I}0Sp5QB;Wwyw~A7}+7~r@8yLX)hKFql-@I9_NV{(og^7;nKT}yl75&C@2|{
zg?b5YNO;P4b1%xEiB#3ZHN%rxpNK6cbB1cf#At3Aefdt$+8>(FMx3i;S@I4*UBZ%s
zzdH2~D~w3!0>60*-sicA`^2mU(Pr-TWoiFX-z_}XJ6GVbp<*oI4iu*-Lh-47#1hkP
zw1Lc>|Ho;{|F=GBPViyMH^g%Y^&$Pnzc+3>xm@Ex2@eFhC`wv?b}oMMGCjFO18c$s
z4NV9ue##0~Dp1q1q`c51pzn}j{+Oj7E<1X1G^EcVakmAW`9O<YOZ!sqk35*PDCWVh
zD;)4D6=$FNE`Qq?9=@e7@F)sBWd`+^qCs&>5y*-;l~ev^5n=xv4*ZR&v&)Z6xqE*s
zMQa%d0+I-+?esxa`bCio5q_3u{$Q}--pk#iQU{zs>ER%UK!{dv0SziE^^)t{cj<N9
z{}#6ZNdIOS$!)Xj7W<VockD}~s)l6DQfcHJcvf&uJ>Rk|*Ywc+A=mCoA6P2#s3i7D
z9XJESzlbj-S%yE@>Mu4*Z%rDV%2^dCk!4Y@r?Dgyj*>!#aom*rK`+L2>=Iuhy2q^C
z0^(dM0dPo4y)^XqAOV#R908APJcecv6qT2s*~Z-#sdh~?aVjVyT2`chq>xmx(en^a
kMlp`g=fYO%>vgv1n%YW=WiE6SiK8eg9Uj>ZN@gGb*-qmKmH+?%

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/BgC79Zt2fPs-2455.jpg b/assets/images/impact/talks/BgC79Zt2fPs-2455.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..6328aa9e020d55ed5ce485d01fb312035358164f
GIT binary patch
literal 52448
zcmc$_cUV*3*C!g9R4D-zkWNAgf`FiageEPa1VTbD7El2}K#Fvw6Oa;muc0LLDyS&E
zDM*(l(j)`{L3+J>-}g6jr`$hg?z}Uz&-1K(p1t-y`*Y4&=j_$?#rKP!0A{!rObb9p
z1^|#<9)OD(z<t0KN=hn9$}3b<R9CKCp{8Y|rKO>v1zx>M$H)%6c8wj##&(@cfcrWJ
zA150dj|dN+pb!`gzQ!$fQxqgF00M*lxd_?iP+A&V7Ft>s5C<Cv=zqFgd<L-4kaJRe
zrXUjpkh73cu#jDJ0(dS#l9N;XgZDob>MJx<lw=g-w3qA$%m6Y93JP*c3W_UN7-;CJ
zsmRCy6qHmfSEzvk(lo3_YJv~h++^M)u;(Myg=*@qdAbkZzUxuTap#e7+w~Dy`Cza~
zQa$J;wxC_^exmT>j)gy>np!BkLJ_pRSLo>F3|yB}UrzsD%n!JPLrHbzk|)OU54e8{
z;0ndR!y_YSp%4I4O0%jNQMt9T3CXBGd{c8V4Pc<S?8-vH0#E_;b%~ZPXMHMfO^@(-
z{=Z22HqYCL$4_`xYX8HC`1@n(!nHK-l!5;>;!(M^Ch|Xcp;R^!>jmJJl}gL?Q%STN
z^bB%tIIFmO9CiVqU)>R5q2JYdVs%@XkJp^91$#J>zth<wT*WG4vz18MAw*bAJN{G;
z0!1k0%js~*ywg^45GiN{p38;u>OYUDR$(1B$H&ma-?UK|)~<tvD}!Tep#GiAf!|is
zGVp0D+KF3y4f_f4=cx1Hsdlh(?Vk$UMBSU|J#2}`ye&Ok4n&2RIH_Ig?~L<nfydzn
z*Y<=|rmnH<UjX(ZipxH9Xfwx%D_v1k=*)L)8quS3OSS$uo>4}{t&L4_E$4foUHF0I
zP|`G*c1_4e-EYrOIbx=++>GGD?u|mW8FA^Y1CCy^?X7L?H{Z|r8#Z$A;!aEC5(J(I
zf+c`Ogub1}pKh2c4(P)nM2LNHHno5ofiHE;Ke}c;+KP76%QA`W2mJF6xL-bJPDc5g
zwQ{z0bW%4MYO-Nf+m$}rWb5{5dPA^6Q?A*YQn)}~DG4s5!t1*nGQ*S~9W5`&2js#Q
zfb<!MNa;ce0?#be`%zdK8f9C<`z;$et8;rR-L=@I{1=KYiq9y^GTw;_Bhj%!q6pid
zw$RWoUl^tsK7Uy|Nvv9#Ilf_!x&wLNS6QJi5StbMkdpPwPu^?}E6#miZsl628E58A
zhQZ8=13G8EQrWqJdsprI?)Ha};{`y%@$NdyRiV3wb+cYKCH3Z_uBKdh%Hfi;FYdf2
zQGys<!whU#kFeepEzZP5EgN@=M1gh(XrWm*%opK%Pb0^FG^V1yeJeZxEpzkx?y2%6
zMTiou10Kveaotk%Egv#Kuk?nM#Khm?6txJ5BPxHndGpKXuc@8yXQ_%ut=**4HNi@W
zNffbGKy=K>QwD|Sr1y`jgl4mEXqP?8vnGe6LMZ9tb&T8Wp`UsF>?;l-g|ydn8}qCc
z@rlmG*03<hq8;u2P~Fu5dEu!=UBe%xg6}N75{)L1dbF{xu9<|rc6HTw1y%Fnp;)|u
zSQozl>^Lqtzgv61ru4Y*T?sil5RV?EJ7T`Qkpd*h&wF`CZoKu34VQo#`;~scYux8i
zYNK1m#sjY~!XTa*L(DA&GYBv`aL6&k_gRFPJ!pZJr|)crwmI9@rRiE<^R(yn0l^iP
zup?tNDRLbU1sKv}(8;W=&XfNpJ&iN{e*@t*E=@_AujHGfKff+6p7*NioiJkM>&cRh
za*GwCc1h=vQgheC`JWEijcIZ1s<(SK>*MM_g^hJ=399@FiKI2!L?78Y^h_TXJ|4-v
z5+wF)d^C2aQ)P*IxuyR>P+ivrAe7tYb(UA)@Z(1ARx>Ss&QqL|rw8(;05Bguq$ld>
zryE{S8auQ5gwZYEs@Ct_md?QLxRd#32!b-?$QozrD25cWT0;)~4w<~=H#8FcYS_H9
zRZOz96d6-WgZP>n)TDt*HZjJ*HtAX-IvVw#SC(0^l)+7fCad{g5b%z4o;@CTz2PEW
zYewZidk#%z{T3gBuya;l?aLAj-+SC4C6&=8APwq0KhqLbfW*3BT*;S&S0Hz==5Men
z-S0#8LcqJzX3w55w(_&q)v%YyVx8MEw>w3e2Ynw8Pe1qNrQN)9<k%i6)1HnBc76F8
z`jM9EPvcIbC<Kt~09Q+Zh<9Hx8r^;G4Bp57dd3cKpnXUi^~|tsh}&b2vtx67=vL;=
z-l=z*jVJp3-<-I2stJ9r`L_!vMb}R`-6(0~$p%xx3JO5Q1~r4h$xI21-1vN1_Hj1M
z$+zAj?gOIh?-`EBWE+W(9U`;07MD&C{VL}R`WJwtUlH_&&F7<kgXE2aR_>X|O~35&
znzN`ss}1+5;`NjRN%eIn3SkYCviz-U>6>65QZ^npWK^MuBuo+PS#f4)vLV0-^7t1G
z48$^aPb;p`kVoi|aCm&8OED;o8Fn@jhzM&z(cZ8mJSZ%AOJ`kN^e2i;DKaLfW0}MM
z_e4Rn<6L94O%*E!L+VA9717WvYhZXH)V@*cOrJTI?-6JHf>tAiDjO#u*m0$#HQM6C
zB0wqGHdCngz^}ug@qIdp#M-g(IikZ-W0xQ#rVcC&OlK=+Ht=KCuCx~F&>}R97;oB?
zR+j05oBm>)OmF7aamC<yphJezn`Z6@SKhyBoY5b%eQ_`S0gKivO*)CCFQxJO>SWqM
zL8@1DYoCEP&&#JjRaJZ(o{^*Sgo_lW;1g8d7kwrcfuVFQ2?<$`>k1t({nkg5J?85-
zl3PRI<}0H0(-8cUTDI!hN>NT>jnRR;4v{`f&TDn;Ee}pyj_p7Z-s1aQD{@Ul1C#Pd
zCM~fui}e^FG8HP+JWY7z)q;JCo(XwcSyh#!sr#a+sp#unD|Ud<cOdnbENc77<8?P#
z#%W^9`?vm5q<djqYzOQ#%=v9<@P!yRN=E*r+3WZz5=6qf3_SO?t<}71BJTpwXR_ub
zNRAPRPj-V4RZsHvHVel-Mqn(aEn`zg*xAsAXw?ce4qzL<1taszU|I$RzfaAITLvzT
zIV00h*#>yaH!KDoD)J^>H$hNS{2eVVUAO9?O|&J6XK5<;t8c~$o+t;-DX5U$^LSRI
z=Ma}_MbXI;&qf~c=*s;cPOeOtx3sxVRwcWGXZ}_f07nhRj=wmSgIcLomu{l%S_4nC
z+A}ueY)xO)QaLp3GDe<J$B0MEH5sTQPf-I>bXg`@|DKw=<05yAoh%iDq@)Gb$U5jH
z$*c-3ERtjM*c4BW{f=$K%XVKx1>E-iSRtD-gL}*N4c-5$J6)+DhMhO0l5F=E?@(9x
zyPA$hLbawX%-F3NrT#cv4A~$kE1qE9CeH)lZGI@{aNJyZjo#i~J-E2EDBOYV3K_Bx
zCc--hbL-w(-%u-R=$@Z0--`;y?pc*O89su4=%nhF#>yy)S%0(&Df-l<EAlts?l#PI
zHBx0GyZ!<&=yWb~jgM>H?Jh6RPOiEGEpd`ZpTRLq&PZgS2rQtRt<9;V3T4tu_|^)Z
zLA#`UTbGb|Yl;`$lfxXDh=FNwdRpwjc(>OyLh(Xn)RYf@FK}+ZvAh5rUI5PFwh@@1
zi%ZQOpdxYs5D4AZqUDWX;%j<&4$OR{(#O|bu?##vZXMZP^1+u8{k1k4^ikW!ZEO#9
zQ_!LL6klzT2vrU{E-W1?aT|P}N~BciTSV-Pt)A?ys0!<%jl%wV(~*q&(UJlo6Ba>N
zXwBCL{?X|!nVAL8cabJ0j_)A?Z^7^dld#g$U?farnOmj$4|<b#o%D4An`7bUtPQ#G
zDnVZ6Y2*c<OZ`fqV^1A@M8sR4pvh_9BlA*mR<s;=+~ZrBw6L;tXz&%_72lLbr-2Ra
z$ms63ISqViYtDj|@(eI4pj%9`dymjT@~9+kL2FA*&3?SGrN7_T@%t_M6rCptqDZ5j
zxSK#ldys$5CvDr(h7t^qrSa1;O--5d+7V49$9@xJT0DxW<mTSP>c;h}ee*XqR(yre
zt|X3LYd%j=ECi%jltKM`JwJa2#IF4H_*FUS(ZDsTy561{rpc32fREi`prR?&PW~Pb
zfU+@_@`t2$|H<|(XmW5G#8zZLEV~+4A*&_H59yIQa36r*&%BVe(edf^yY6#@P2V9M
z^D<L&ghSn!E_6h!J#Td-;WI!n3wt|v>ZZ`gx;Z>g(|2bTeFRyHqv8?%ag)X3VJhTH
zKN*~&R4WMR%j7P3m&EhpN(YJb(Z?&cat-}F1yK}Esm%j)JCI4(79ZQ^_`D~T6F1f0
z|27jh`}MB1>1lD(cxA2VNFZAWU8I=mfvJJ5av`J3dlxX{>Y>}1_cSsT73h=#nRX4T
zHUKJ!{D$`QhTr(aXnyZjv2X5g<OP7M$zc0sNgkK-3}NllORsv<&NLEIsy6H?1@u@<
zOlY)<Fn1U*p5Z%J&*hcUc;<jBjTKYqlCyed5#(Dp#u~qt!|&N8Oq?+H>d+YP^1@eC
zXpAS0^15jx1p}59x$Vj{JjethdLnuBukz+iLv?M^8AQdZ(lm15-489Qp+kYOUtPCf
z<`dWKb8==%Az&RPAvEMMJgyo7A*zUvZOzIyI{VZcN}4Se6%ts|!2)7%=66~!DB$zK
zOLmurZ|xbz#-$!puPKJIQ4j;=$51mZ8&+png!YeH^Jp!cO{MTlU42orA^^95YQVU^
zc1a$i+pCp@55^{&IX~xQzd)j4{neO2F|jD!zcM*OzkjLj)A3dbnf-6kt2onN@4qU0
zcd0PWWL^Lc{6Ebr5unjsJRKa9cmG}h)>qE2cdQPI^L1?mVw<Oe7EY%H^Uf6>e8N`7
zrOVY;zLl0N&UT~K2Dxu)PyId}9Lmp7`UdkqmS<~V_=M3RkMu)vI5691XlT8egUd6%
z-{@f`jxRk;-~Dc6Tr9>MKzpjiH|yb%LW<m^Pj^%`Lt_1?k<BhtU@_6cDC3)cRST4-
zxuVvNP75XBHIRZFv(rI39%bz+0&&<n!Rm+9&Zmb#yfZ-!3dzS~nf7Qmd_fF|7iYd1
z*pokw(bOM#l#l}uQhB^n*GqU75xj!+B;aX@+I<tUze-A5#!ZZ^0T{b4m0;`JlX)jr
z2PUI><EC%wtiAU>&6nWx6euVtp=fs|-oXJ=JsH>CaoP1`tH;`4(Vt|Sws+pIkdC-o
zIve(TQ!;!|KsZ~T0nJ`V)AUVsPGeiPo{S~&MMEUAC}H%MVr*-bmIL8=bS!KY7l+k|
zf+6^s8Ux;Y`oHs>dx!*-8mOzAWHFTSP(Pj1PmR_#TiVzTM43#~Px~C}H$pn;KyVx-
zn8zKVwsoGWKJEr_b}~g@(}e0dSRhr2s@!!KfE%~0HJi8ZKi9uIA}XPwRSAc2@Gh;+
z{PqKrFjuuOJ`tnHflGDk>5eZI0qWzuKIYxA>t!Qh9+{e1EZk(LJ3>0+d_3v(_{o^;
zliJYUbUk&n_O^(;0*re0<0`xFwZ$s+-yZVy9`WCd+O!Q~nT@;C3JqY3p|1gbM}V_b
zF8jAS31Su~I#e=4Fq5lo$-fV27q<<!uC$*2o4ezq`uFcusWP(?kz>w{ok5;AH=qTN
z8h>oKO|kN@@1q86=|tj%(I9M==OI9()|F=z^O4@oH2>kXpb+7)ps$ZF00yNO0O!An
z3zu}%&N{|n{z$=7u*UGrYJN(cj=b4@Xb$vPp0sQO*_G(B2@5qpXYez+0LTrUDF*HT
z*z>!azdYJ_CQrC^n(SOAmsA~~@#%dM7T)aMx9Q6o;dOa~vuL8(GpBle%nH8Z7>|$a
z^J&s3b-*i)$4OA9Xfwac{ZP3ycR<*+j<d1QEHp9-aHdJVePr{f@NhJXyC`JS=HOqL
z>9&e-HD>G2P4pcK3!914MGM}lCS{^_oCegqc}r15HU6B#jwnyud&jLiX|hj-rpI4?
zvVlATqAB%0m4u0ih$q<#$q!zUVzhaH(8Hu9GyEwqOOYzli7zVkZzI}r&D%9u_?){m
z8V$cZxd7-JL^cO}P#s_*plKSWi9}$l11?ye{F7y{`paWpf+5ja1z+ivnZ5XWY{Goy
zwvi6lLJ5aubj6al6nPIleekb4HyzF|XIZC}!#tA?NXcC@MgOWQH=~GS*xu)HOBuJg
zj&ORBcY3DkcmcR;0CA{+EbT0>O)dyQBkm7>mnVE{x_sk2tXKr0WZfD!oEc%MZ&_Ax
zrL`bKhyl(H)b8RG;efw}!n}F^^?Dv77OET_iuHO&+$vxSv?+M|qO-gzs^0o*$M?|O
zvM9Fd5#;9_miu(%EF8N4FaMl>jdQ|(P}-QN`8eR)Y>v}>PFhhc!<*7QF>Ogo7!{d~
zvEM4cVOh9av2r&PoC2X+9toha(lupaLPW&6<%EcEDpyWj3dcLbx9T{#oK7Ot@8w;~
zu6&1gqL1Jiuw**iT3j3to7+~*#~-OP&f<m7+3(!oDTctg>c!ak#o$<1B+{omM>}8W
zliqGe{iS@mKz47+PFupJcG1`yJmHCJ5`WmmnA6#X>0;628i`n&*!j8^$SBOOAioZu
zlX4uw<4;qTq%8`{iCPKuh8o8(#;*Sh5T+*_TFQ5RU1Xy&_pHfz<c4#R2hfhor3?1L
zst~iHBcA;DTZIOJ@8-hP;lDp`cIb<SV_i3Wt_EoNWCni?_11D+#7*~;qNmlH>(Yl$
zrXwek-`-?Hz6>)lP&?7SEA}tIaufHi67($#8>cFECyqsHji?mKk+hqEPk;Hg`fWw+
z25c+GD4U<TDZlMJmO6U6*Li%GmDsV8JIP9%MHDin1X&>p-Tvdd1T?V{Eu(gCGRz|K
zT|~ZB-*!_DdzY*J`D^F#wZ#B>ic1jxaqP8Re{}2@^0?nAJhzW&K}%G}XYyx{I@)rZ
zMwupz1EkLErM91L>~`!W&FiU{>nWL^{L^;#A12eaz<=8G|6z7?{eKO}z3jv~wV1O|
z#y1eXZ>OqadrLk4c<5$pHa}nF2g3;qMa*A(1GL>TH^ERcc}-toDV+|;;Qx|T7(%>a
z%a>mH!YtDF0#Gu~^}<gtAn3NX(X^uwwiu`8$=nwT*5x%@n~n2+SoThnDeW;<%vD*A
zK*@5Pd)KsJ`bc7z5L~>a<7gj35>O}OBzs0jH?-leK^a@L#ME+?UDv(F5>iHpEGspH
zM5&6gklXb~a_~P=K2lKN;5|(!cpUL$UZvsC)oGnw2h+EpVbmW}Ce0222Jl=VpvLt@
zGxEtbV?$)@M1J#x{BVR)*f48MT8IWeD$RCS#05PPN@04=mZOTj#}c#adwjLn7Sd+x
zlMllI`P+bTY0Z@I1OPWP88?&7&n!=-!tQOsggZ?QeAQ=38)-NZ7QW+QG>HaMn+`;N
zvRc`YJtx2ni>peJg<6!xtB^>)r>Cjg_&cOjeKUQBmP+66{`nxW*P-HB*phhAvdF<>
zIw87cnoXV;{cmApI#jzeRu<B?Br~7XrRn~4^z+Kf%c!|d=2mGT-=04v1np&%xXy)%
zy4KM9LDIlZ^g#KO(=xusam2_llsU4JK3+U^UF(YU1%T}%+_YOY*Hpr{ASb=aV}5$&
z`M9UYJ7BL}+#Y(8aogqX-Iv_Ev#<Oipp!t}O@4MV=`z*ets6n5lRUD!FfnnkOp)lx
zgJ6|6+ebN2-_zt+n}|27E~{EEbIp%amAB31(63TF)5pkYb$EdH{QRJg3b8SNHEO8J
zDf%9BtjAyRTmBdY58W>BOcK+M(%}xLUvcE}b4E8}@G16^8X)bP4}u;;K#V1M(0ksV
zul}ayt~{H<f1l#6N9tuX+gT(E8gaG1?1|qP`|01)L^!T1Hr_)!*ly-UmIIiCa#QQ3
zjCEa5^y+HiX1&5<GUSmw2%Dd&9ae5?7B!{}c=h;tLT~&MScSG%UI^qd$g{1GCGU4T
zJw4Zb`+vyQCv}4?J%p^9GwuK1$^Wd<n?o=p`Dvx*TRTrTa+xI_lSqg>-c8zxRxrLq
z#@Hc4MNlaTQ`U{rE=BCUo8Go{j!d_^>N$v+VM(dPb1*9=3V!x*nmAl@G8ZjLu{fgE
zQ_aau+I{u@<wUB}*e}22=X>HRl^@>hGq!Dq&G37wJYJq!9y)HHCdJhsSS||V>@ebr
zLcHA$w~qn7@a+9@nbIS8VVFoY%W_=5m8xWvO8t4FyMb?Mi%8&5|9otjjP?42-34Hk
zhI)$J%KKHZP5%4w*>y;rvAIlfQYb({bT1H634hNU_f@WpzDIb&>C$IbqB~gjRsLmJ
z`R<sTsIIY*AfgsP8ED#wEBw|Z+Ql<~w63sA*oM5N*Ev=5wfkZpbBaLNGB4}4CdJj?
zYW)kX8#)8iBf_i&@Irq#{1Tv#Kz*Ga+ll)ueH*Ml@!#XsJu7-C=FhuhKOGJ-;mQe(
zsN&+_Vn_3McTku;C{X(*u?ol)+gZA{)q5SI_S$s#Hpq9iHrGlw!+lPq%^D|ET-*He
z>;+EykGp@@D`=yHDBbw@3XB7Ctg}<qL`bcFvg|PTQV(1|DPT;tJzBMp!1v56Rrrb7
zVJ?kH8w{+n)A%8*LVReXFdq<SoB0P360H+2=G*u)@Al{#v+|I4g5H<M86R%Uu+f~G
zSzQS_<5H>E*{}gR?`NbvjW#tEGkW?_P?<j<&NTk!&Ld%IF_TgAXwz9iix*70vR#hX
zllyTDR4_dzzO=*ltvAoPCH~R|?T459V5yeYM$}amK6CE96jEF(NZzB}*tz4J^*by7
z0`SMZK2uLt{DVRBBXip^NH&O|-1(HSkN<P@yrOu)$iLExWM-8}gJpQ_0eCZ-citp%
zI~QKMPH|fSbo(~NZK?lbg!N78>MGw~WT4vFi{Uh%qn^{T-1s0%pMTZ6Pj~~}F7;xO
zWxk0Y|LNq64{oT__q$_hP~}(A8Q;XB_4GD{<F`oc)N`Gxk6t_N5j0Hr-gCyyhEF$6
zj?%ssd(^L?-Kf~*92A&?u({|Uif&n9cz?34%s2=uj!h}V7WiSUoaL|0q~Jd~=>}I`
z&**b}V$@4{O2Zq4c)<5o11!W5FfG~-kSFp<z5}RGceK0|U*ncd<h|AT<^Bmi3H%4G
zVrpe7E--K$rjm~=Z)M8P8QN5lLE~8RDjel$DL|=`F#{Pb?MbboRcA!fnPwwi6DQLJ
zf~nLak9tRHMh)1}f|D$U9Ed=Xnt{fQZw)!S?9FOJPghOl%{x|2kZM{vQm(W-l&U+X
z(}<@VYR_<l5{(ABol8qUQ_vlyLhZslpw{*7W9`OnEvkZfkd?!bR>lJ|)ORv5v(T8=
z!=h<15r3_f6HdDiJ5aeRP?}#=UO%C2sb(x8r>BhtMOo?G64<Zm)bOTYFEnB6Kn8NG
zYisBP1b;%Eo+t%blhG_IJn$aaM&FT<9`6UIJ^kVt=m;!bLXPCetG~JYp7(`qkD^qx
z_deK7&8?S?CaOIgW5qi?iqw#!S3vOz28^CE39psJ7J~*DqOfwWiz-C;gI8ITp70#?
zUjS~45YHR`CZ9h+44=Iiy&6&O^PgLRmxdJ4^jM|h_+j<u%jnvd{~P4z1%P>`-y<1V
zKZkvI^aIEDF<GQS_UZhw_EfK5ZdIAr794H;OAK|mfM8LX0zpICEGs778(HIByg8HQ
z(5P_OwE#rZcV4|%h%s0rTr*k1Q7H+>jSbM9wIsT#C-G%p6{WmhuO$TYu$VGuUhUuy
z<DgpKWVc%-{LLUxydO6i$0AG-RzqT|jB?J1uGv<u>!5+f*Enw0H$=WK1L8J_rw#QM
zF~jVHFM1Jpw-~mQ)wvn2h3rgm=PXnKUEy^_k2k*nuITl+%pa2BefIJC@Mb(t9!Xgk
zsclns;<mZIZOyLF^7`f<ozI9`-<3bY?<*ZNqRk-&s#N`A4C%?)-VL|Fa8(L`_D0-`
z-b3`=q1W9eK(fBiq(IFHy9%W5&cMQeSsA1I?^xz^Gx4e?Fom!kBaT7oh;fO1Gv-R*
zt*A|GWv&a4y?f~%BrsvDObo!5n05LiTF}MO2HO7~+SPjK@CdE+P0(IUH5f5_-UN?D
zy)I~0JRlwIzVxJG#feiFTV6r3E9O^)Yv#M<#W1ifse~+y%`p0Cg}Lzz^0;6Z#ZV}=
zYvn(4$5jM*u)(`!h0>7)a70TK$bVLtYZ3ul>FX>CAzBSppOi@y<DugIX&L5BT(F7q
zBgeD*JG*Yl`XE+quG8@{?JcoafU`JbQ^ay#+6svgS(NjYHTYqUYA_i!e_izcZ8w9;
zYWje7Ll+N#fL%^);Z)G%U-ytkFxh{rW{m6!?+_&fyi<k(#**sg{)Cpf>O|4%820|g
zDh6{{bN28pTyL=b+aGZOFuircl>THM+B|mG!V+DD8bO<)dj-SU1XWE6zcZ9D+<Lx#
z{M6Q=;<2e)VU4w5WzLjHvPr1G{eBajwq8zB;>_@Oj(;?eC|b&#4g!Vz4+M*!IdCkD
zbSzbb#VVEgvRnYz_s&^nj4uF6&n_eNm+s5zcG%O3(G2V;;rSn%=gFlNd8I2W2Cr33
zXkkQEXd9E=fA_hOGDbaRUFKm7ir5pxeWkutZjhL}m_qlfqRW^|bwl0Qnlt*@M2gNW
zFU3!Uyiy3uKpSkOGbslHIG{Rqgu)z0OnjriZq~4Qu$}r(C1KUw)wNY=H}79flpj3>
zd6k^Mx`aitzW_Ykl>M>u?5edDhF-F)361U>W@hJ@s3XQ#%LH~`zw_k5lP6c6mXBrR
z3>8TLf81&q{`O5B3CdB`ff)h3DEB~DZrt&;7AybF^=tZlN>+21TcJiVyDo<z*#mOz
zPO{(mM>cE!UFdgY^Q15Txo?idEz6)cccbhz$dhR-`t8ig#Ef(VK+7H?b78}2|IGra
ze0Ar@Wz-9w*)UzkmLyyHS2+DaeCR4Ck@kQ=;mhG2@r@?rv_s`SrU?BFEs}McpZS`M
zmhWr7;G=*3W<${)?+wj}!tbos#fYUp5n>|Q4{217u#3l4Y$s@4^zH?qpVq#u4GU2~
z#SL_Y)(>eHH8f^Vnk>1FraC&kRYZ3^0s9*fTG<kV6Za!lN)M>@9K@4br~E#s-8XkL
z<uWURN2Tr1uD#ZFr?xe@cJ1Gu+rJSI2<XRHzBbcD2ujGfw&JcH?8xv$RM}lT?_o@!
zpyzQzP93^2#l>XZNE*c0pSquX9>HZaYHxjilVdqY^U_#Iuq%T@`;@ny>LPk5Bdl?v
z?ha*#CAZ4@bZ(_!<Fk~(MuXP04QL?uf$wxMYM-5A@m8o13L9Dkgmrv(`Y~i*_5rP}
zi{P|ahSCzLzoa&_ny5L678E5eMk#@X=2Rd2d~e?^FPrl=&rpZG!L^iqAE~(SW;aq<
zfYzzQ8o>N4S4V}Vj;+=$qQHleNG6j6Yh>si@;&td!l9}nM?R6fPWaX8JkMPE6K;p6
zy@7&8S)T6bs%Lo6lALxXgv~V+EDu>5-%uH=Z=j+<%kDZF9;&rp0N!-0=1&-L2y0Tl
zcDI?xM9c~o%h80X^V7<(gvRxZ_YFUPK9*dQUWCjy1h}aU2JCDOJNxgBX(cfG#Fs4U
zMW>td@#lQ~JGOT7#W4)C9hUIh?OUrePiwy8XEj=$BRguj3jp4c>)|hWl>c~EX!+B)
zHLq;Bg2!(~Enqvt2grAA2GW~1euWhB8f%*x#NAsP6QtfsidPZy;{pv9rsFh}vp&wQ
z;03TXuo7%Cl{AB(*<%{9y>6?Qcn@pmkD`Q-Y&Z=`PsMvfsQKAcBIAaPV6$L0l05$e
zj?h!qA4Sk1L-p#ll7ow2IT3$+?+GqzL?KPdO=mDTp{nmzQMx5>UEzcndZKg`S<k?G
zJxp=br!iuHG0ZFKp%K_hSVtvL=F;M~{L(`A-9^@Qe6(CrEdE+DldBECqxkia{(1qo
z(js?S7>WAWYGL8PNZ-2n`m+l8(mYePFy7+Z<C6V|kvpN~Uj{q1{w0l`U#;IwBh(^Q
zlA5M<9r3qU1d;FwYIh)+{LvC)_n(fg;_z{xi~82~48h6SGwE9o<u|P|Ru;Hx>EP|o
zRmjoKiJARbgrP-)X622Fl-8Wq@aKPWsCHmI>dG3vyj2f=R?)N|3>~s*ZNZxv1sKKS
zHaQ8|H8C?)RVsgkQSSx7FIBNxF(q%>njT+?oqOT?WZu?VO2_Iif3hg(f!_&$qOp!a
zeJs-tC#Hll+x>72BKkXVG#~j5Co*6zt>dm0pn4i7C~;4^X+r$1+~7>nMp8rl0Pw!-
zGXbMgzt~ES@!UUOM?cNhTmV?b`#wGBU}_BUtUn*O^#)e%I0}~%S|Zn;tr-nEjDa%c
zsR3jO*Z^&BN#YAOE?ZNQna7HK?fS#2GDA)IDs^=`^WRw^Nr@W?Ata*m_1%@$PcIwJ
zNcbEve5p8^T_z=7&+1ov8EjXDhI4m_F#W2&C^+MwY4TZty14O6>7&p7nwOZ!13oMC
zOq=N+&)b;Y+39{ikzc7VahV<hP>@6~3DD?}5Qz1TC<mnn4sv7dPP^v?O*18?9P2A@
zATg(Z7<Eo$)_4WC`pV`%$~iS#I8~((>=hwFaw8%cB$I*$X4b77+My%<_XfBhhn`v)
z<<+Lo&ug)<#Z>A-Xhvjg8jK=Nosp*Jxm2$6UB??MjXA5&@t8_X8M)chD0BFTAG59C
zi{IksQRmb&ue4&%)S4O^2`jCw(Zo&qV>@*FyYZC@TBFZ~JPPg`o4T35&f7l`bALWO
zI;lOT8*RRa0|)Qoi-s!|(_zK&0{)3(p8UH$0_W5>E)9rZp44X&e7}{<XEs=TW_f(%
zupO?Ld5e6d0tb)<0FE$9irI5NI23X5;ake4K9eatPSc&vG;^q8P`uOg3=_QWmWMSj
z2muNpL96Ky5x|~lU3|SudC7z?e6Gx)>{l?|q0+9V7?lH$FSo~#cI>&d12}KkR~YLV
z1rFBiYzRhxh1tm6eGeEPe9zqw$6Ww2BmR_y=RWt>HuqLR>y|vbISEl%3Wj}Ypi_g<
z@U#5MS|yaBUX6|Hjsx2(<a~wYl7OISYHGgn#VwjoGZ}=VmXecQg^*6hif@_bPh{&A
zEe*y<7Q2c1z*0fOHzFSfI5y@fW>4Qd?5R8K9le|R@oKHaU7uKZf8^RogXB$6#ifJ)
zV@s<6e%rf5wiwr9<|w=O!=a)9iLBI_yv|-mCkC=x<>7w0kooY}Si>Ku4I}cz7Z_W`
z;OzTR_;~IpoeRJnriYu_dFwJ;Px5BYYHJWa&1=!@?R!?Z012gfH#|H&4%v|PG;$CX
z#JCc6>-0{2)9Lq*7l4JXGZ(}x^!Mlur-ud2bK{C6Nfu!X3mO1wB$!O36~IY?oycrQ
zJlcO!^{N4^RWatV7J}$8ECcCDM<$v|zFg5tiUCv1UuM8eHdb}A51U>+Bovk9r!-<_
zB;>eJ%vibwIIxh*l~KXn^x<av!-m7}qsi>On5AMDlt;e_!}eelhPuq)N)Y)k9(41B
zwL(+vI^La7Q6}AU^GU*e8JrU>$IzEO>PGr5=P#3}<G(t^J@;$(yx?hcSHX77;z<CO
zq@NnJly;9*RGxkrb^r0;Pg(iKT}O)+yWvR<hhbx2y!y3wW2(#d-1|;-Y1IkcyzwQY
zhSrZ8X2BIXK7#u1zi6k0<}v+{P>oN*y536+{`K{)k3qrc#6Z6Qina=eDb>K)<w*#b
ziOh4ps#v)HOt!MF*4p!0QKP1Z71|btU|3FDA|rzv=|Ex2)HdIJp5DK&|GXd=6U(Ni
zkKlTQ0WpASbUp)o0UVBdDRs3(@vEBqXC`9L7CsNkl(G2<h#eSR>KFH4J1V^}A4VA?
z3fGrQsJHq|LT|<qLlcf|qPyov6{gt=cJyL}nzG4Z6t&BcpR_0B1}a5XGfg+pg~a{A
zQKytP7kq4Ouoxs&mKM2c6-fn$bibS|vB{R|m{;2hOr4du<=$$MtO#oOTt*PJ2OyYf
zA$x7q)Rf!-ar6h`wo?bJp183+)Fhc*j-wk`0jaS_2^MgGQ3JBXc>?Bo><bldbr$Sy
z0KX(b$}1Xr!?>da$X?O$UI5%g{sj3&slk{qBuDu*v0?qAAG5h&=_2i5!QQ2q2i5&o
zW2#0?C7PNLMb6`1>vt6ruR<^FZ7fKs*4-~FXA8aPv1He9efPn+0;7IWL2=$|H5lTm
zk{Bp*QB_?l@P1*;2SkvAgyz25sXK>i>y7=}aF?947OMvGN_Ahyt&+I45|)xUN<yg^
z)LnObn_^))+R0gVI#=Jy$v*!$v_n=7azzK_i*vA0T}z0aqUNzm_?dgE*?Iv`4EhuQ
zoA;J@9&YOBXR-qjZED;DgpqWN%zJ)fJKi?lSnYWT8bA0>l4LglV<0tL5$OtaI_f*V
zb~uMYzj*5<hr2Fw?;A1e$&K@->KMg021TeOY+<LkP}Iiwl4Mvx1|J>QCeA?+5*-t0
zu6&B7QNi``(thf!P&-}DUBtGp`E0X3sdI_(^OML}0y~m^79A&*@tTUM>&OqfDo{V2
z-S|G5Y3!isx$nS!bGbMJ3%IPjKSKeWoFXN*HgVS*hHN?5zVCm$OtASFwSB7HCyC#t
zWsk+%W9%w1!O}UA8d>njQxjcUNL=@i?34<Y@ij{vRKbugMPVrc#{0oHh`#zTVyv;k
zdSS8kz@DIKO=Yo`%2v9TI}c*kc2%Qf_GA7%J=$>|H9l={6?(Nc<C_><Ij0hnP5!Li
zFV^g^_TH-U$JrMxnf8bVW$s)1T)H^x?rhI*C>)vwu4j+bqSJ!5Jl%fS_a{HF$ss%G
zgN@{#(TL%QwFuUsNS>A#Wt)`vjd3~ap3U5=xJDPRB7JN}@h!|E-7kFV0E-&OSwnnf
zIuWo5du8y^*2g3+4ld}a{WP}5lMqG0`@A2Hq#J4_S7b`RrQm<p!lUBKgaxVogU@j%
z)x=j}hJjL^YuB_%S+AktQ1=d4vkj!a@lC7#qoSs9oAt8f_PjR1#~AQQVZtB^`bX;C
z(_3%z{H+|UANm*9Sve=CC}MDt`6=eCAh6v}Fq}iIZw)K8y|d;xaYEnBez*TUCHf7k
z-XnZ@@onUofTKvf;qP+)HXfUe8;K2;*?0#+!;hr$YLk8zy7daVS0%mBYxQoF-Xtvp
z!0yA$L0q$6s>AdQn<=3SYPY*HCY&yYQ%B|D$kh~tO4_Z9v${)#q+!3bkn&Y2J`5fT
z2d>&NOK&R*&OM_Kn>oxYc!gw(^Jn1!>`yS2tuXa=!qU<*7#36h`eky!h>YUX$@kUH
z;lItBEP`7bzv^2fco-IQ)J#H)k{?5*5PB*B6out2Dw;E;5-7GmG{kRc!N=Sy=31x*
z@n#Tqk%_A30`O!H@i*^WBD4X2U@%l97oI!PFg9LmGM~CK3Znx)R$FG>#;Wlks&Bn1
zKj?uId2DrytNQUcY(dianYpTzR$SZ`_s-B~EdW%<ZQ?NDocSK|_85xd0<Z-p-A(DZ
zU-rR}n28>w95WiyWk~^h#e>}mQAfP;D#F*;54%TLCbdA-jvO1cRT%McI(3wanTUeW
zZ_+QWjTa>q(|V-o=q%dW07XYl8~0>^`wO2pVZ8wFz7E2Ruf^ddPdsk|k?oI)^H9}=
zuRGydba3rP>EDY4o2U<)wxNDJ1%pME3&2SwHwi0^cEu>|)SS;hcZc*|cLaz;&oILY
zkA|lUEEP?A{=R-AsPvR`E3QNG@d#JzP-QvW<P5uESwSf^rlw2BkciCl6Xji4>dxJ!
zH_xjd=S`Z`Z)y3~1~bTD*JqH&n6<y!Q~zwBvlYhd*Nk|u)_8nX+*o)HJ*MxHQ#7C?
z2NXl9-DReOqsO6uBR*vNdbymYWX!1N<nedN)8YbEi=|?11va|Zhgw>frsNd#1x{t+
z>+xz}wTTt!g{j85bAm@%GRg_-$*bFU$b|Cb>;e8Y`{44Y>#X*`nEC?nA?WaLb*gOZ
z`jqFBrdHxH#F4oE+g`l@W|+WLU7Z^m2K3E!<(nY7Wam0c$fn-=D(9>rhbB*_{YMrn
z!R87=#S~FqdWnnoJydWa+Y2YJM){vv(ujQg9>mrNyheNx7UJmWJa&Mg$H6yQB+bra
zWlKSzOBrM`)i#a!%8-z3-oVdWOmmql^fFl^$zu#cs6dGZ)5Kt5lJfcdc38v%iY`)v
z>0!yNZC<`!l4Zc-)TBBArK`w~LB*Bt_-b=L4jHCMo(p<(WvWkkChecJ6K?@4MTKHP
zoJixW9qwxCc0VAgr`skRkoaWmHRI~q(q*>CAlo&+0Q`);OJQMf4hAdDErp|qak$(E
zvt8)gK-JMH`N3e8*dToMt{T7f#449EEgix*VUB-S=M3lho9v;&#>ATWU2(lJ$9o>0
zvL3R3B{|tGfWbMmbjVm*GI{7<($dUObbt`dwY0d-ysC-hYaB%oT_ER&s)>O7dpZ48
zSlCd0w$(bL8IF8~g*gQfDmN8iypM#v_8eT46tH8H=<JQnR42Z-PnoPgZgVThrb{W{
zQo@wc^cM{(OKL3pS99oRzHIS*>f!8EH9C&5$QO`f8x;xt#yW!yU}U5LT{n7vm^FM&
z@vXM5O~#?)b2NEUyCD|etRVVGV4#$n7W5OKO^9j4=!R)Wwq$NpY0uZM#BS+}=Fqih
zYI9xxEV>5(k=HLXr3cO^6ghK6cWNzIxg@!vD<@4MYY8&s*yKuk##c~U0JpDwXp#T@
zHAVfV{K>nrXd^Y7W})JIe#rv+d`%FDBJTNe{FVFV!-q|yrG#S516`6X4-c%<Gg*p)
zllL9ZrIB*#58s$5{V?J1*0`u*cs99Y1v%+LL$|Vxd#J`&FO$WrxbI6pANf&?@O;Ms
zIy`JT0P<85e8h|rGUWe@;7=~0^dom8b%2SapH)TSl_@BFgd!`*oqt6f#!1PiPG8}Y
ztRJ4JYO^}Fn>b;ffXH~qFK;p2>%pUKwKS`2GCyiox@%u>^|p!A&6?37qc3(c?AjY+
zduub)m%oJd9R3SH{RH8B@d7ZQ%SiR!t={F$l%Gvm<BW^cT9JTgF;0grn-&!&q?CL-
zrQ$=pl_TNd{HaeLKlSQkg`uF&W%yu8zmI?0BC1Cqi`#>`*!qUa1>keb*ZoTe-vsK{
z+sILqxqA(YKC)~z8u)$=Jl-Vfby<;xdX&+@R$R5zuMMw_wYl#N`j(HahQI3NWn{oZ
zLoaucM78mUgak9ywR~{>-Y+kqEN~z*uL*om^pEMn09I+G&fDi>Ijs^Jdcx7rJn-&C
z&S_iYu*txc?Ar7FdW~{-B$5GhM1bfy3_XJFy~Gio)>Gw-NhD2c^*bDHn`Xn)O<L3X
zQ}R%VAxXAr!Ax(=#{gZ^KVv9ioa+t|#ZYRCI4t6;SzcoBihHuh`+dv0<*X@X?Rv5J
zVZn-?2;7LAtz3d%%$(oD!tZA=Z&ep(_qub*0X%4(MPU-PX_Gvh8f~9*@RDu(K$CK$
zUC`o7XelV*o7>o%@irSV#m&a)Adiys8dipEhPMB_{bMN(6HUT9Zdmxryz0J+%Hv($
zH3-t}UEHSjUcb8Tx>*;n++IT-s3vs0vtwUFvy+nqddigWT9JCd%8J=n8JIjiDs~TR
zK(=j{$-lB^EHy~&-E){4zZAZ*AQPPAwz*;Hi3?J;9tM@2jZsS2N^cxnyRkW1gsb(u
zPxAP(Hfmwxz+4{%HZT@e%njPTZg8US)b1J651kdCT-7ZofWPViVt`q@MC(UzF&21P
zJNwEqqpi%obQgGYou6V0cJPr5dI8Y=^rPPQK_1&ToVsW#Fbb*#%+FnLI})!Hto>1E
z|MEUj?1ccjPQdIWfgy0R>gO+l&5!AdA853yS{8_(byyS-v>Q12xfY(OuQZ#)0=)hQ
zCsZlLJ7R6jO;20F5+4^`M$FYaZoPMO?J`}veJGB(w&Sya0NbKGR+`iBFm8x<Xu_eO
zC?Z=EC6)M{1#6`lINN{Nmcdme2wL{R@SOMXS$XrqkAoc{jwLakM$}~URn}Xak*4zD
zFm0b=tV+12DC-GSu>m?!A%krO#O$)f^D52URm3U;_pioiDY8zA{MsLXU!^aruAu{L
zJ&Qxs^m}X&9yy7%6okoPX=CuHXQU#Ar#r3PB~3Fxqg8&gQA{M$jm;Y$&8m$kG3&D-
zF)a;GYAVX&%X_QKy;rcC#nC5P&J6j@PP%A~9CaleVLrY6{&vc*FvGA4R5ZL+RfK{s
z;?cIfVsLrx?;XAdMDG1ENpqHhO1r$AHqC*BrsgC+^@CR_3b#N$Zo;=kUX8IK7B^eG
z*tX3uoOWuap!93(J!g_i^&^$|TguN)`+)9rM!`Wsbq*(!Bi3|J^y>3pQ4Fl~D3=Tm
z5Zl}Q5o2eAf)AwNs+>pAoWZ$IQcfAKmaU><o3F4_F$FFi@-^7?;&J*KyP`>*dFt`i
zY&Cx=-dCJ^Vd%5L9;m)3SG{y?cAysf_mvfiKIslg^TTgnAkP-kJc0b8fdVJW4fp@+
z=q`in_%+?4AGdB^m)~e0<i%!4wG@3W1R@goNgp0~^KJ=!4B1F~yc&K3+CPOq;Idc`
zuor|#ilM}F*!gvGWn9WYo;=zwOxG7{huvBR^7W~xtRI9Eaecdg@CXT&_oPbxpB~WJ
zQ)qf9*&@Y#lCD=>wM%0{zIAu)P#1fW8@b855f~l7lOG3%^%&A9S4I16Bf!7DO1#&3
zwgHnYEK|dXu9WmUO!&GhxufG;LRKRZB*NLUb88(5!cn+5^EteWY4cdw>g}rZ*W8cz
z1O8Ckrc18gHA|r;<^BwM-ofB6<8ZkM`M$LdNLhgyIFNd>p^$GeIrA(WK3QMdf!o$i
zO#yECrKIN`wI&l5jxnU-NPBmUqHR-_P3?BKwNysu^G%m7$;6Tw0wdOH9Hw68Kw2(N
zmNCxP*fViW=n<haRohC<xEju3zL2U<(wA<&uGl{SZf8X}5NNQD4y;_uRPuoM+gO$B
z+*WIz6ZzwWvFQk1n-}adgTfgQj991pn<+jTN+AvoN)O@%Kr53$T77cyo)_hq2W%!y
zY%Mr}hNf{0Ctb+@!1FUezQiMc@_63Xs<O$6ic}*#R!PH*5z~0~hF(Bai&`ztZ5C@w
zl)S+p%{aP9?eY6si=TDXb1@>(L5Sr(gV%Ik5-3omWbSOMamtCx_U1`diNq)j4TXZO
zpR?5@N`R5fi31To#E8_bGzyzLs|+2CRm#HN+UAxEH4bNTa*d>hB3+~V4A}kxC^i`s
zO76F+5)!|ck~wK~=<s2L)ZahwqmY0yhOqJLlp6o)iF~bk7zw~?H!HU-U!#x;R^pp>
zxfAfAxy?O~ke_UdE|VVXEFCjm(H^5B8y@`qg@21JQ_XeOT1C^ldNd>YV`18{#~CRS
z9b8m0uKHT_pzpx-JY!q@U52=B2To*^Hm4=_R&jADJ{&c`rP{8Z5EogJ$Mc?H<yZ6r
z?}wP2>-!-_Jf?RR9a}6g>InikO{DyQk!<>k7<DOtb-vlOZ_S%t-xlJJHl7uzv%(cz
zck=xT7fX4xNNKM0y88-aU{6~4=Oq=OtemMt=BJTiV+7i;hKVCTroyqa7YZO*rG1gQ
z=j{1o-5HI?Pdyw#>jvY4O3m>NdJXQ2tlwfhJ~O3y)NE-->Yj1kx`yxfeHF+C@wG>e
zAVUt4!@NcKWc-wVbx1{5S>9t;?m};jCp(}DqGOP~@<-!j>IU%L<(k8%%iEbwa_5V`
zvnKDc--#(BSli`mrQlN@6u}dE#1<8UrT2e6_&|Ct)#jv}HcR*RxYnjYKlZ1%W93+K
zGEy*IOEfqsnKGFyblc#{?NJVB_TR22qmG<%8eErI%1G3NR_T%^QY_vStW<N>C#9H_
z5Zl@X_4`oKA=U0|`pGh}<qlndfOd&UNky{T(O=79sSNkT?@Zq*f98COSD-*UR~C-^
zxW4{-?^nxuU^zaxvH&FJNgtEWa#DXFf<fMnO&{pXy3JrY<($j)>7gfk@Rn1`j630-
zB-{tjFWsA5Aj^^D1AwbR_#P$g+Hh`WG^vtWz6$u(+Bi6y+T&TJafxrihE!r^Cek6X
zp#77QsbBO`9SU+2ThDVE@DXTD&$0d5<ohcVI?ZBuLmjaM`Zh8%-lf@K9t0Bk!~3r}
z@)msKw+Vkyq&-^JPIJB-TZKVNF%`R!{w`g8&EZ<^we|Iy1-*CruK=*D$LVEM<DX1q
z?z@XSm(KYaMIDJTIdmO;7JO%I5rv@$K3?F@Qs_!^b(ZG)Qtx(?Ew}hBggT!OK($yr
z>v-}he0{TU)FKsbBVI}(ogO>3h&Izo$FpVaB2ihqN0bjDQYFGU-ny1gWQMhJ-7vS3
zZ5W(-f_`rJtpntUTntwrLt!8ID`BGgdR)@Gt(R=gH;oTVhOcut@3|Y@*J1_31&pGo
z59a7zf8M-z{Nc?LwhAIHr^!)HU+yMwNU3oR8Cd{Brnr448%t%LzG-os+rAfa1eO??
z3T)Dq&`7GG;2zan)$QZf>5~^w^;=lZl<G<QU}EO@oE&pf_+r%h%Ok~Tw<35!ik>Im
z_1}rVjpO%i*gfRd^~|bL2_KtH68!OcyJK`zkZ{mWH`2Gr-wd$Vu6FP;mok~^ft}w;
z>v#H|s;m2!*>!H-cYN19?Lp}4<fV2w>;f`z%8!$<Qd%?{q#gFOM_)%)eZL1BjrpcD
z8CbbA@~9_^-5%~n8TaC>nOtCo{6z`hRwkN8QR_UbO_r5+KHl)nAN4wpWjXpI`qfym
z?eJX@UznJIj3UFqdPMq*f)1_^XcE*GC2=g%^R%VUvGfhQ+#Z|9di>XIX2E_f0Jpo^
zR>H>os>Ta@<@vQ1eHHs@s4b6#qHl$f|I3?aYu8?R8eITZzAL3Ym+JmXoXdV(>NPmY
zn?8~{^StS=5s#Uagz=~*SMaM-u|uELPoxmi?hLxc;{=Uctg~z^F&$9cmaw+SYbSww
zkb#U*JXmPN5P0SB!%%Rk5qeSA;RTycwUx(4dkzj#h*_5R!R8C7?(@+FCuN-gNz>w;
z)&{RfJz9({*RE=_lg62gv=ab=9(fO`%XnM?y}k0iJ~zKJ2td;-W&>W9OJ>X<B^T^R
z-9cS0wfae6US#$lir3UXpksxxZ`Qr;47=z5s2IW%n#M;*$IwqW?ZaRig@fjYC@m)S
zdkRO@?d<cG#jV2`afo%EJCb*APzmg8rr^=Q<XDZrVt|~ygn{ok1&^=2+hipN;#w09
znRhU)bCToR9FH&sF)aE5zpK?mgx+=EQ_Aij3h-kAlm|WU>3F+{1F^v~^m4Jk^Y?~v
z?q4|+SRk6oe})F+Rg@Y}M^L64*Gh_uN>U1ZME^t<s97ZWl;lyAc~3ly@h#}^%4|C;
zp5m!&_9Qa1Ba5jxwIFN;Mlt*AQ-gk%2sz>DHDBMzqVQ)Ms9q~~5KhPNtpX))*R**k
zRefW0vPfh_Z)IuIcn3@ok*pM7Cf#?^c{a`xGBY()Rj&N~@vjqSm(u(Rk9!qO!}x-2
z%^ni*n@pLyHgi$QejB2_=%AbCz~A@9_1^Gw*#jt95N0!J+gj=VPw^vn?$@VGi#rn4
zw{yKXrW!N&Og9^UX|`A#&l1pA8CC9B+EYRLeG6%!>PVQ00bTM&#KEFPpGWg%;2g?u
z#>sJ^q!gB-!;c*0IF20ZRQ*h(CGDuC{lEEjOi`07?G3zbpbv5%2KKytkCQ!Rh6Mla
zmwxDMCWN)%Y!XDzBf{MYUlJS-kWpun*K>e`5agS#H9R%<$J1}XoGOI*b?z2ynj=UJ
z5;#WWBputTAHRKN%yKo0r4RtTVtd<YLUPVNma^*PR-+&Dp{xuTEiDk_QA&~8HBmdT
zM-D#gJq9UO*^JT)x_)<V_{feNAAq>4QYrAv6(0q%X-bR6vt2R%?P~r1@b=z8O}%Zu
zZxr}ZMCnqb69hsL5fqT#Qs^P{CISMX2uPFOy8%K+2sIEw73l~{k&Yl;ib{u26zSES
zXYVub`<}h$IkWeiGv}YonsrasT3Ku5p6kA@@Aq?QYFY9Yz8iXyMH)A7)@oEV>Fq0<
zF9jC{<We7d>12Pad$lF~!KmCrw)gEIM30`{vp|P4=cwGVqu}h1Airt~L}q()*~1T)
z2w*G%TVeOhnspH!o|Wj`nkYIM6`!@Hk4Uz+Bc^8Qy+AKDr%ix=A!rgpwdv2io|Y|J
z)^-w&a)MAK=69YL@xJTI&P`GO?5Cr-6M<=8M2$X69fD6r3~^qMCYjnMe~DC~N~_8A
zP4ydK$|%MBkVl?=3V*-JftqjhZ3`VuF4?H2@%(i!sA%#ovz~O7w6V^Qgp>yp`rjN8
zGD5;aEJmIo<O5BEJxBOC!iq#p(YK0}<vaB67Z~z?`c&NidOy!mic-S1nM2|@V^J5X
z8qP0lB>~IKto;ixOPip<mTop&@*pqDz0KVQB8!q6W)G#ru?UqMNnldULfO7|kVCQ9
z<5nC9HzhLGTdg6s5W1&)<jI+fF`Uq77-dVMea%;w-&ht-RpCTrt}yz0_ufL4U!Vu2
zTJg8A%phdTYwkN%p5i4zk@BL0(fHhyCzSO5QMMI_a3-WBj8u*r9;>pP$|LQ+={)%%
z3R2Ff>sh%rf1Ci4l+v#p_i2d(=6E^$k>AF<x^(zNH&pxu_<&c%rY|aFq8{zMaH#x0
z)iD3hf~WsjQvdG{c({e)7vbcu*kU&QjQJsI+sd18EKR?a8w}%7E`Z}sWR5sTbB(m3
zH#7=A>lqFvhV)8%?;NJv4{CKO$5c1>i0g1o$(^1A89|s?y~{GfUq-24^6@KQ45MH{
zrIKn9yyyJ#L!C#rk)q?Jr3kV@a(;oubLY2f_`~P80!3!VHI_~e`RkVmBR64ScoyAk
z@Vh|SQwG{$JMXQ~fT!hUa)S(kQl-^4bYVcOK=Y_?Wu@X>E(IlJUk;weKi_T7=Wtzx
zdV~Ha9*h1?sPS~+C%G!p@#xp%h^)a>0?7v4H2k+c`A0nqq6^5esy3~Z^bQ{jInEn7
zN0FK@qF{2sduiykxxL$4zXg&_O~%qBiw%H_7}&#L>6}LH|9FHF4v*jyHIzWQO{-<h
zUw|N`vXRHU*pv3_lK8HJ7vqRmx@+me1CYTqMw+FV4s@q$8fo8c3C6Hcj9h@Mb`4@`
z0>~q$q*#xN=SrxZuucqxlq+z{dIAKmAD~Y@HZ)Wc;Xn6~YLL=c+DE_hm;4jfcv?%L
zFJR*sg%+BWm*wZ!gIp2sO@@VbV#ZheP|KdYjMEZw7Ncg(ke+HqhxRHN9}a9TS*{)=
zFP<CUT`p1qnyse5wqt+$T~#YL&7YB$d-YRrME`T^hbk0cOu}d*&`7fR=0Hw3$Hsc1
zzq%MD><Rqy>cprfB=x-&qTw4+__SZ;w9@Y-q|o=L-TSgrSp#S!k?wWYuC2XZ3VFBA
zuNV8rGno77i6^#2Wh`mi?UB0N;Q<r(Mr4e<j=VKJ;(it{#Qsabv*Y0OuqRTQ60`+H
z?{hKO{hQ0!$ZlDx3|%dDh2|*u(3>Kal%{Ev<*>y^p`qsX`*DSL7X^?dltO>DY)QGS
zc=mp{_#5y&l7IG}+SAdlx(@UY<~&maGRfC{G9?r`K@!&Wp1cTt51c3!D;2b#Z&G|W
zJIkY@I_+*xq2W~hWT9PT_8RPF(h$)8YTRhT_>({hF1m=xj*{2JOy}tl2d9)DQDZvn
zZ+w9-zUYzNh~)TsIqLgXdGW=Pw!3luOX01qz(5EWe^|-^*JD<DOCpGYF>S^|NC;62
zQ#;@%Q}VY<>_B+(noIzZu(4Xkbl@tnd}hVxxA+&pQvdOU{V(804BB6I1?)Jlc{?ab
zOEN9(*x!^t$DXuEGS&bM-C;3f?N09yY3x8XAo^R3r}zc_@e|Cjz3_C0T#%$lPPb2(
zb@lP@b$ds5M00B;#B%EYw2R_KDXq*#-#hkNWc|R#=Po8O@3pwkbmZRUVDX%lR^<jD
zKaNixa#_>6+wT9*=khOE%}9!Zwo-(W)35})sNXVGxZlyL{MD=#>-16eic`>svQpgS
z48nW`Z&n<Mm#WDL9oS6P)k+$?Fk0R+smDI9#2k>xqbriqO{Eb|Mo^MMR#xs4Y7v*y
z#)>MYKa0Aajgx(bV0aktk%nhBeE{u||Al@HWV2%NTDk{qzKO-D(AWzJNt3VTim!fe
zu#C`-d^<&0blo@!#$)N2g2KQR7w?hI_lx7;->Q!<&<rPU_*M8<j`sy<*nGftJ8#>|
z)@O2#TRToXnkNU7lgb6PY(RZI1~W1{5&)LX>l5o|_vcw0k1&(?{o6tODh#6Rw-5v&
zfuhts4R^-WWaTsQLF1}+;u&(^NLaZYUm%%Iv<+2w*N#2;#KM{bS_}$({4<RV_0O2`
z1V?=`;DFGQ0k4D#+J8w}pKmRObvHkTbdRG|-nC`sY5L%j52PHG{fN>1jV0Wu`iu_;
za|d)&N7il2Kfz7gzPH}>Wd*G2Ya-$p$*XS+Eqw5v>7cJnn{E+n3o>x8{~_3!L-7Wd
z78kqSBX|kXKbGkYIBye(UYxC22of-V*@3G*dBMxN#z%#bpkcxVH+3l=@hR-^pKC_(
zhbWfd!;8bZK9t`)&BxGy_x5W#V=fvSu7Aax0$&S$8r8%%lw!W#-FKh#MVh-xq<4a{
zEscWcQ_#i7<cK=IsCf=$bB^10D@wH0@DwUUzZh*%003J;Gv1}{P(FEHJy7flYyGiD
zXTAxq?|qR>e=~j7i?sDi(GTs`@hL%UD<K|IX2u|p&SnfkrvsGrRCGBb+1c+h_xk5O
z#P9eIPQR_rZ<$zdOe;!y%w$J)XN&uY9Ji_A-+s5@yXftvDc)3l>#Z&1Wb(&S6+@8%
z&K@ivnP!w)%!SY@?B(UN6!%X~hk>ysy-G<W883nc9tH3?*g2E^=<d}RUWN+N^e<^F
zwK#K1L$^Hq5w8SN47IUw&>FwjS12|0tgseIgqCzsG7_lb)o=WuG$Fla4R`~ssy|N7
z2p8TRXJZwUIFl?-7Cfui8hgey&6OU#n`boDDywjTFB@+hT9db5&bwt%tCB9r4Gb+D
zW%s1q%|t$P(3>`IDU!Ucgq<{t&OqTxL<~a9`8#)DdXyx)A+k?>4<EPVC36Dw#;2wt
z(}QHgi_^juA}eo(P}erU|7s=Gn_Fvz6Uo$Ukmv7RN!29H9&VQ2^awjmCNwDU`ZS`K
z9f4DaR*uFuOL)~aXb!vlu4TF@Ir7s*E#<u>4H56b@^byRP$rX%=_QHb6nnRdEKW{b
zs+gtbg{6XSB5n0mx^Aw*5VFszgoL=#qgE{hr%W895dV}5tv)*-7nk?jkPiOEak$on
z-d-kjbcC0(;b93kf$@H6jB)g~38Zw3tEx2FKs5+_6|NT<juip}ZtQ7RmwDF12-hjH
zja|q`eM3z94hs3NyD+?B%%TphYz9c1*6JB#3;&~ONJP>TbOu17nGPeggDA{*u+Qvz
zJX0O9^;O*y<tBZeVpe=V$UT^AtjkGdf>&j%paf?}>q%iR$iolg21r1U{Z5w7V@zF-
z+b8b@`Kph<*CopP=C@x&SwV#09-ZI<Dp<#+G?}9Fgqz`k{>$oZ?S^x=<+kAbhK++f
zExYUAWjE|i6ltk=gmj;K>h=p{pFB}BFl2)^ct^{0cD^y(ze!U_*pZylGg67$Dn&VF
zNM!Tr!4V2-USiDH-%*X?W3l&3TM^nlBzmo%$x^pNZ-%HQ3U5h~M_zR{fzTU<9$%ll
z7H=*k&?ke%l{SG8x$tNinjY~X*@`F*N+mo#JU_cD@Y~JTTp7B?{eB}WY$gPTg0(kL
z3s1wr5}Y>``5^a)Ebt<i?Vay+M|cr4S4vC5{>w&eh}YC4ub8z1uf(;K_pdW4Z1u@I
zt{m6HSB9j)?^LKluV~mF>pE1K&mgO2=B}5_UsqLDpn;!q!XtYa5?4{hAJq~84+-7(
zD_e^`<HoV3HIKKGb-$m(A~pMVmvuNtmiKl}{dB&Y|C}~AsrW299&X1p-pN}EPu7v)
zJ4W4+AkK;kDEFp4CXTX~qHM*cgCfyI7_M?Ax^e}JL0v>zSjr@4mW`)_a>uXU`K`fN
zr#)FbN(QUKPw_@L4GX#f1uduc5errkQwKHoJ{F2RmK9zLE^F$0Sq(K5!RaS|Ay@q@
zL<S}=+!{L+5u@p3nyfPfr7($u3Q0h*5UOzn+mCHEn<K;HPovleC!VwhgcksY*i=%0
z&yM*)fR&9zS^K)7g361jaw$oFtmb5h64uxp0j6?<XANe4mD!|ET{y83RZ!d@`nt}!
zD$365FjnBK{70N0UggTUOguev`3taS3pbB$H2u1by%{DU8lt<t{Gc`0j{V^3!CUX<
zuk(S{>K>Kl(HQhC)(q1%Js54u%?@7OXWe{6+@1pO<JOf|>W#O}p&M1ol!%o6*vTRf
z6cVDl<tZ)mFAe;E3cCO6*52=vVB(pxeZ_w2J>+<<4_6o$4}rXn{He%z{xYv|5oePC
z3Rfi^<ic0~C<)HYK#|_Kt$h3djWA>#@Q<}$Sv=S*L)qc&aaWSoL8`||b>QLa7$b?d
zt38EmrEmnF?st6-^5sjX!pp<-60$QPZh$PA|Ju$JS-0Nz-DAQxX#)uP#DvB}c8I{l
zk6kgB23K_^j%JOt@ai>_?#-FvJ;NU_q-_@o-w;B>g5ml)DUn%|<n%%Nd)wlAnm6O8
z>uQuV2Sg1Tk5}AMte&IaohS`}6hftCkCZtYt4(f>7ez51%E<VRcu=UVS>{MQ$_I<@
zUN;q#PCOi&s+QD3h5BqNxJiYN8kk|8D^Oh7dxdx-CL1RKh;raDNx{&;hj#CIt5mTn
znaVZfWNQO`e|=9frSUuO&A`Eiu>EHk?9O3hhT#_;a}$}@?RvRKBvW(7&&vwIowU1~
zr=qe)V3Xsm*oLzVR1&O6zdukd@n=WzxF;_h0pP^^dd-e6`@{hVCR85K@XB^LExJ_3
zt3o*n-7EJ`KOJ&l)}x7tJVfrwx2r27f4a~Mt!2LCl_%N9X=<tgDts_CyM4~S#RD4!
zmoU(7uIzF(IF_W4qxS50ZoueuIasVbj#*!iFVy{$bnnv+(y4Wlb!b?`0@v8lt(vq;
zn7qNxwXnCSsFb1t4?G1W@a~9gFg;sir2V_sv|de^T0aRTY*eoX%xdM9o?;`Ia{tk9
zyxa=UE_1oQpq{`G2@WDZgtx2~pEm6K#8{JW)lOslqbNU9eyuScQ~UMxb0%*xTM``2
zknUW%|77EkSu$1(nKcG3dTn=|*iJz37BFd=t)vU}1e?7(rQMys?L<tfRs&lH(IxXm
z1NtQFQ)4s(nNucfh&mVs*iGtDwpYl*_eW8KBm1Z;B^wj*AN48f`}=W*0Y>Y6+7_*0
zNOg8tfy(=}+@$YxyF(euCp&8){}>wM-;lgg_a_*;#n<zO4xy8!FKt+_l0D{2p|8?6
zbWM+_1;h89i<&>PweT?M-B(H_$dJSg5|w^U6JSlSE|Zj;YCLb$4GN`0fETrVc6PMH
z=?`KYZuURo@MfaI(B~Y#*f2-_s#0#L99w{mZ}6aInZ&%)J>FOnI2Zd1SP4uT&e5}T
zARuY3dD+YAkCCFP?_~YIdPF_?I&J!uv|nKh4pB@xH+eo9d%L@AVO}BjqlRA2dsR89
z+Ood@by*{fDrd|uB7|;n+1S_^?%MHv%Z#YHO3%m=28AkqlU56j%ukj>$$M-rSq_dh
z%zIQ;%X&7YFm8j&(k)5HD)vNzia!q1?Q9S|YEQl0C)=P!Qiga+kg;QZClf-y4E-Uh
zCt~;p(Qb5^p8w)*d&BR}f{MrM^HRt2qORi`U=FQ@5WRXl1FuSG;!VIZ#onEu1?anU
z-`jQj<d2iJRj|dl3d|I-U|~<W)@kP-PQJ#sI7j4se!hCI<0vbvP8e`6Z=EkE>WzQG
z<?xXuHe_nKB|icFoZrsmX`*+90~Rb7a0|j|AZ-qQwJG!$WeRkWj)krANguCn0m)R%
zJ-2e~9MPCWRabN3^~YFO?a>&B6>Z4;yWskc;><rA<x-rLGwy=%{sCj)Ecvtsh;832
z?=?B-wEexI)0@w=+xNVpDUk&?B+6PH!c>XEBQnTk(!8=GVj)lf**aB#FKT@4CTRLA
z-ZMAkwZNK)lmh>eWLp$YlD{s_Vin)esh5WO?yAqBxUV{J@ny<H>TX!!*C&!X7541z
zJiL^04GgrOWc=cmIwlNjo&*nwen#TLKf?rAJC|d@EUCw2LW(Ktoa=7<ADs{Dj&2WT
zl;ZOkz?7)`BngB&O^?#CP0NITZ~|=B(Zy%AKS!3ufRtDLj~-+<b?=PX*^+<no1sfh
z)T4339^CymmKjmhbZ57goH#VKD<wyo^x*0uk|$JNlVJAd_2e2`RJOn6j|pZ`o2Dlf
z>N&4@S~O|V*^uM{yqchhK=SSoO7SjjE{6T&f!p>*`P`MeZIRh!C)kiD@hYAaB$c0i
z;|%c&^T*XvOL6-iRN2&3f)rFoO6qx?n1V)G^ZsMiZT~@IR&p<=bX@klsP3y@1Cs&B
z3}>g8!i|=pr9v0TvM1S0o&Bo*3vjw^Q(WQo0{w_^FK@PwN-o~}chU+~e?J*)D-^I7
zaA@;s-x=q4Wb|V*VWiO>>fDT|_QEB4JxsI<pi3t?AX}X(r0sJlQGde4FRsD2GM_}3
z7<`BJ!;Wk3*;4Ol@pM}0Ux0I@kxCR@l3co|KEuH+_$va**sfw+g<ED_cZn}2)rI!>
zcnS=}tb-*hl&8O&EW@t^cuMP$!76Sg?uVo>+IK_Z_7AOG7GS}@F1k)veh<~Sam6J}
z)tD{dC0~@4el&Gsd?8()2?jp@-T@@0dO><JMgg2XUU$@fCzMtBEVfsDFj&~Cs5XYi
zZ!@sf;h!a|*y`=k7z$o+TCzA@vV^p5TkoBQgoDvZ)PYZbDWuKS6a0g~qR)*jP2zaY
z{p3EWye1|npT;~4i8`JQIZOC>MO34i({QYfn>`GOvzwZas1S78Wt3c(u56%7XPu-<
z;=TbysmrkY?*lR!e|7wHIj;EnN8j;6O=F+>D<0Hp{&J#J(t&EVk~&=W2I@56wvBZe
zPEKdUYXK*>D{la%->oQ*gDa(FrF=?DT;s8N;m<C(+F^9Hhf16@qi=)l9^Dj<5yT4<
zst&CT<s}78WI&XxgUv}gS0-*Jm}=xGm27Znk)ft!vN@sww~E{6fckP2edAzNZ)z-e
z^8wwf3K~br;s^1wkrieo5Mx@y6D4vImR-!A<N}AL)1(4@mHyneZMo%#G5tv_cLd`A
zWHln+{m-M9wO)7Mc8*?VLNDPaw%<Du#V@_x_RPyC8_Zq8O_$?fqYz_3t|Z!jf_2lE
zB9zRI7CE-vkGz4*6?s8{DSbz=^rdnDL_z8Vu!3X^cq8Wf=-kprxAlX$Qaw6jP*Jv8
z0NDsH^y)!WZCj?>H`~|q?VV;j6XVlj{{C^i`Y0tOQah5deJW@#g$NKd5^*`z*68X{
ze81_N;Lk1hA{9i6G%)Hn3`{p5(4?GvD@ufIUK#27eX_O2YwxCsQ>o-Jo^@+AN9+a+
z!Y@Q>BLvwDNT*-D(AY=bp8THR-J~?F^M=Pw3B94$`iw#E2JlF46=O}Mv>A{I8Gki<
zOhjEs#kjkA^^W*v5*)>tB*VRcuk@khr29N}!W{7$CP?2Q!(EQ)AIgFA_-5C0%nzdI
zYZJFxO3ycW$s>nW;y&4vZ6l9MD4FZlaP=OPI(xLJo5Prjx`mc*O{&doS-Aau+%Vl|
zSEhmOr7q##k=GV9c@B-^1uv(h&Zd6jW^al&GlZ)*dR1u4GBbsc-6tj0p!gHEoNu>L
zdrZy0*zs+ItLb4UKW2)yxsNVhGcpqY`aMbwcR~O34(fO3*(4=3Jq3iG5kq>x?sX;d
zuvgP+Y8SJF)K+bl#m@wPuy8=U(;t917}5eU{Ct_8Q9`6EH{{8UxsZ7_g$@54Y?@bp
zH!o?7oayF3>`w$qzQPSAOP2(MA}Bxgb1S^)`r&f`vuqGC&m$9I%QC7wJAv5>u3RYD
zu|LRw_79MH2M4@snbtOL&bfc&P>jihnhH<QHsiBCQ&-Rzdb+dhq(xpheAKvG!)Wxd
zyvg}x<qP{yZ<DUl;oM75*ajsqn6!o2-3P2Z;p$3=j?ppSVpe^b;qQ-I=JW@?Ayd}S
z{Qwo(PMzWZ>KneL9{A&SVT(E57FUeV5Gu?TD2YXEaVMU)%($PsUYCyk1&kA^QgH^<
zj*ZT3n7rSCaYA^Yy0*GYBhDTci_kAe!fS&%Xba>e)`CiAgP0xWw~=B#w}QqymLT!*
zt?|uj@gOsm<}yWZK53RdmlF?#^sh{<zq;ho9`~oq2XXWwH*J&#v4X}uDF;IETKO*=
z#ECFab2Dud#^$Q)efMp9UAE*D<DafNyN4(q$S0Y8k}`LHalGq`SnT`{Cgy<KH`5$#
z8;7JM5Jg3?Gix|Lfibiqd0GvVPlEAC^`-7Y17lk1xWV+?nxw}t1|+%lXM*kBwVqB0
zbGjC9-tKI?_Kc*jtY*&clR47g;W%J9m9CgFsF$L1fQp<?pbf=<+?Q&4Co(=Wj>JoQ
zL<%|P*#)*l_6L8`JXx&ByKwMrO>i$kb$ryc;G>=V%pNZoWg#sjJxW_0XZ9*FH-?+9
z=XesocvNQKAw8PxWPk8#tB)8RIX|}Up`JTGGbm|L&-y0CFN-K7YLo9%_5t^ed!kUM
zGW~?jYle6CSKNRN<Wetea!<D9azE;DncK^2{PIh{F9aCYm8R<IsXW(K8(4m!moNl5
z1jzss4*~Z3N4p#c-CzFbHa~9jYG}1O#s;}~_$ZtHqsDL{ki3re=lotNQNFVAM#|t#
zb3>`t>-Eii;h78iv*1KUZHHwCbpppNj|A;)N|z@sHdkGEL4NpH^_n6DH@(EoFzGKA
zeA+eD0Y{aG-kBDQuE*Qn^zVQ}S975BekO+95gQ)P?vE}@QNMc7_8M>I5Tp5054j)6
zAFOp|W(Zys@AwU^$U@3_$NsS27sQttG3=zQ`Ec1bm3`iu1&~iJe%u6~&uUFgdY!Pz
zY~q)-CXeXwLbk^Cg{q`n^~`S!749|NQnJ81KW;K-Lqd($sbsUM9afpXLagK=cXo}N
zGak^&3Lu<D=uEAdzJMhAqlQ=`Zmp#ODlgK0UR7t;1@nlUt8{WEEXX)lF*_p?V!q(x
zoNqz(^=|QjV9^kQxQ1-)^7Koqqg{)y58S=-<hy3?rd1PT20Mmbpj`o1%8gk$2q7W3
z?)QN<PE-@Vxs5N!CM9>p7~V`hAf*kq`R&`?^EhGBAGA_ZU!O|2mG_OP@D3+hoCnA=
zAY?F>O@O@n;oDTKu-B8wxALg<2o6tY-VDd7@BC08+FencDdJ#=lG*L?Q+4$Qgix~y
zu9~UosV4|j0kYiyWi+1!g5(YQ_*nWJ?w-Qt<)7lWUK&>^zl<leQbmr#I}tLU?}GV-
z!#1J)zvgQy2DF(b++5@XOt*;tV9lt@yU(`dH7B{w^f+UPRWKH0)v;!9%zg;etrlJs
z9^D#=hdGk7^a&y$B(t7MM)Foe&~7H<geqs!O|29t5`H%-F$!1&Q0B}E_U%&aRv4NL
zS^xd?FJKQCeEx`QDXQ`xjgZG&A%C1^kzQ^@Wj&`*gKLiS?K(|YDGCe8_%dibzefb#
zAdzJ98h$#TP=ssPh(Et|yE}9JOprIDNK&wD7$rSf&C1yms?fJn<?v!Ygej+Vt(*WX
zEUqk4^Uz)sdX~;hFAHRd_^hf>`0It4{eHSyeM-cO{=b09@)+SB%*@uU7o#%^t~vhm
zZWWCnGG$Q_$~Q_NwStnA1=`{tUk9EFyW?72HPS2aDe&PoFJ(>r0V8tu)MN63eFdf2
zD=?ZI7INP6(zP~!&KF9g=@p-9-a7b?721DRvj4}{*mmBNyOrZJ_vAit;i90>E5iV;
z!INPPlV?GtQ&6zEIW+K3Dv%yTrctEQHL(Xcmd}5MN13a`9r>c!WMTEXlFF~G7^TA>
zEs*37u36KbvzXLuPBH!t16y9ShLfvl?&!|3vsyw0C8{%U{0wE0vE;u5WVTOZj9&2D
zscDJa4JdNWoO5oHMx+IbS9=H2(HJC!q$cmZ{l$@YtBj#&?E&3qL^Iu;yn}^i7-n?D
z!Y+U5kmmVYes;kDsP$n!_=v+cDuhvMt*Ol|xO~#df)Ri2Xs1YDe9?FPr0M5T+TFwP
zf@-_72C6;m(Txk`wXWnyAn?X4`3`nf!LmcWs%s9;m-*~0=2lv7e)7<M#*A~i#X6$y
z=!oaodO*io;`@+Kj$z|Gl@<da^x<dZKmZI+Kt@FNql0dEfsjIYdx``Zh0brg{$`E1
z%ZGz^pRP>=N!7303(kulXY~Yc(Zo0KaKJ*(<kRP7bnRoV$Bp=};Fg>#zV{8?H_2UE
zJ6O(j9*cDN8_&l4$jSBA^V)x%?EB#3Xne11Q&(VuC!Uw&7x}N-YvQ(4EO+ADx{`L8
z`_Q`*@54`)6x0bT<!A~)<M&ll-_t>Z@o|@y{;;F`4(mH+mk!IDyQS$qZ65l*4|6w)
zrxMCQD5$Z$J;Aw}l?6JqGy$nAvi52Bpe>zV<IS^|ju`CYNtA1xwdBA5cI-F85Zu+e
z99-&>NSEK5|H)=~uiN!-KBDTG)>quTBL%M9_+96z{%QUgr&k}_^EVF*0BMN;`G`DZ
z<j`)`qjz-wOC{NT&&b?<O;CGq5p0n)Ajl+Ms)P<$NtO;ioUGMXiuZT44QJ_N1hc8Y
z!!`k%Bu6jUE>L-Oi|bl5_tZNY5oPhl`dY-INCAkvc)gEWuP65T8;Uqe3K%^A2LA^@
zGOO5VqhCEI`mtpi-&g}SDaGlpmNRK2;?8sLRFA4FLtRTz8Sy!6W`4bN)PfW$<OL)J
z!8_eC9`DT`P^gF*oy$&Yct%ckW#81rQIXT^-Lt<fMKh)h)@I_OPbyP`Y6a?g^iV%r
zy*3^oUYQT({kK&_#=JHe{WJz-C?(4W^rugjRZtIVjNj66s@U8j`~(VqmUBu0qWVZ_
z&#1cxb26wvep_1w@f3XQ`s`3j*uSoTW`=rLcrMB95LJ`@$hv;5dQHNg(xD8NL0%?y
zdTH(d>SaA9Mc*<;XI`{UxDbuW6Egh^c$0XY`RoO?{GH-4e{*9-JdxwK+ay(7WNe_5
zUd}fr%zE`8a9G&qcH`GpQ#fd-)KyYAnX$^lKmZ-a!i|2ElzK4!)B)?o@A~Dvu9;0i
z#(HgQk_bkgT&=_h<_Z0@g|;lra?*(%9|Yz-u?HRj$39;VZJ*I#Kdw(@UPmxrJn$Ko
zV!q-kYcLe?Y8n#lSfd6%EGi|f1YfbE@*~U+qcL^n{0;Li8FMh!Fs=HriNc5dG{LG|
z%ptx_FW8Wx%%xS+Q)XL|S9mJ2fpvU3V=SFeNTop{!$E*wA&TNFqkgIMSaMn+-wkdb
zKmhsSuyk5uf;W@a5u%>4{)mbEktU-MY)$41LTO-|aLN{=?$mC=cX{=^abehccEnKD
zPVj<~mBwpW`mQrsna<bJ_)*v5CIjHAswS1KPLMmf?|#U{V@m_@q4AI*)0ek0bW<wX
z0EJ@tq}gMwZuK%RvCMJx7v0Yq=b5;>X0&zEgjC0H)XB$&(+S>VJ<gj#pkdG&9fomW
zEg!MU&KdE3?LBcOt+w{r02BNV7ao|sm~gLvkc{F=VHrw0lzrBp51i{!n3g%)_tUex
z4UcDIAC7BkMX7~9wUPdG$@wkIIkilr|6Jm&Hb7E{0;{N#I8ow5vGqIbPvDsi%8zKT
z)Jm!12dVU`gzZtB&}?OSY`RVwwR&h1eXmdeF!tvgL5M&Lk-B}==UlHox8pW-4j(eG
zG8`}%f`ns*Cl3k1GmbCVr7Ndroyoj>&eaTKdsMu-Qn;0NFeYCVbhC!L2k~f3Yo6DX
zp1K_U9bpuLjNf)FVk}1cDSu;c86hO)@q<U14^9yKqlW>(F}^Q2C5hH7#!aF$EE+@6
zOUDSIfPslGbBMu2w;|O}3{aCXB7UqsieHs@kdGQRuuS{M-;u@ec82E?sr~s_2<cw{
zxM?%+jX{I#BU5uB(aZoc_OwYPRhgC`9gDjs9Gz<;z{2;=XWFKwVq!WWlGb3_Y!g{n
z#!kE)47|21B$QO-8~|cYH@`nflxStI{CLRn@moNn&8Nm$(YxNFKJ%8h=}Q1269z>R
z1y4zOsM~`}=C6b*rn4WM`+xIH;iE(s<$+8%!z}kt+b_hcw-#H+*>tRBKbP(rZM{o+
zOyw2#@Jhttuze(`F?ySPVnJ}Lf@hU!ZIs?l2W%ToF|_)gCUz;F#eiDBkdB8(NCkcq
z4j@{#734d<y47nm?@ltNLCji&B!8ER7ZX&J1_B#xdJHjhZQ*XPrg2aUNjy?ZNTS4+
z<yLRdXdBqEN!^;iSC&f6s9(AoD_#P7H&JLL0apICbLP@*=risgd?jbd?95TOB&j&N
ziZ4}*2Y*y4fCSbIAOEUh<|-&TAH&=%8_>;Rm}lS54Xh)6tE(|OV@<2oY|ar>K7-QN
zW-()mNJ%DAH3&IqOgzQb*0XpDTjDeXq5SVpF1Yf4K?eO7YL^sSh4vU>p2-rkMD*vU
zF>)dDpDR53z&$-@^ihQXdMoPB$V4~n8KB9!Q*bbUNB+BLC3Yc!Qp@I{rI50r$YSd?
zzFqmX_-qDqX<2+2*@124dLsQ%W`ehC^qRhFi&87H)jJK!b`h@l^&rrXAb%^e+q`G+
z(O8EHBjRujVXs;-Z$MkX!CYBp>Uq=iimw!br%)P<+vM`-g>pS{95+Lnzk%Bu8L$~n
z9(xThJzmk+Oz*8XoKY9?J58Nf(z~#%(L)a2qE@i5;kS_{<9ol16xFGc*Ao0m*{>nP
zfM!>H-KJ|TJKb=pcH|M4?L2`%%2ay`qfMu{1q#ISj4hx7#V7bxC~3O9-PYRIZ;UV0
zjfH7ejq-ihpKV?O6K^aI3JzpG*bWV$HwHY)tjrF4%G`@q=kf&+92{f=@~wVF@ju6Y
zkyVhb(=|SfJ=oi;ZtY@lumQot0^X99;`gn|TBGv}rVJr`@scRKexNMukz~IT6~7v3
zs036#n$ynj)$FfTZ(eVyYk=F4&@i1~NNaDfb!E^M`|tR_fcI^#+KqD>nD5Ey)5D|<
zo=Nk`7eNuVzL~Q-qc)*0918wZ2J`>-7VpUBHyFoYkyKi)&;xhinZZg)#1S(=A$aum
z%i~+s>8;w%AM}U7#*|&fCVTSAC*+fgUuX0dL~ADIlUyZ<1@!w$3fhPuVC-d5*+9;x
zLH=~%iXU)c79UIBI@wG&ys=1f&V^MlRX;t1<g(<1{7`m;>vN1@MHcLy|0}e(?Aj2U
zhLpj)wxQWq77`)VIBwH8rLwl>c`>Q_#sg8ac>2QM{FeQJ=E7g@_B71lSB^3%jo?ZH
zBo$>-f`Q-~a7Xi(@^wV>MPHGn&rd8@m|8(<Y7*F0xck0YksU5Gp3g|${rWcVSN+dU
zKwGg9M@ybE9PSf^G9|?L)v>(jcU-9<i2F5XmJfJ|y+MjnP@V?-3Af7GtJi>$5Ln@R
zFsz5Sj<fcnZdpndm3y~p*C82EJd21xGmI%vp<hKz4OaTyn2Z@|oX5K{dB7)>fkzC2
zywWSW{DS9zojG#n2C?OtQp|u-F}VoTm|H0NL4(!CDyQA<{Vu1g_hC{n73W_d9bjd;
zqVFVkaQ~H<i$Nn-cZt7aH|_{QnXyg|@;7`|`gT3Y@REjy*hddFS~yX>OW^yzEf(bD
z_w`#_d%rcqo!TtFH%_?~hYQi?K&fHbn!3DLL2m_O*{)kWyGw!5`?z3+?S4C3c&XWz
zYr}I2{2HN!R*&g3_&7c&;AhMeacRxZ?R#6#=3%f3?cTZ{dKI}4iSc3FBwU||Pl$V^
z;FWWjoI?n*RdK`dV?s*w!7J~oKF!l$?dkAUn-5Uu;-!bluUnEe(nTD~r3GqgjQf61
zl?sc_Ik81ARaCr8_icqBfmmE_ztE!rC2Cx5vc@aa7wpFz&JnNKA)U(dcza7%pBxsU
zjBJPs{NsZRgV__Yqb1_<lFdMM?ue|Ka;3V%+^Mt@zAg-#>dREa#1V?WMb;mkXsf=c
zV@$x#g@;z!^OhFU{oxW*vp>KJ)8Tq=n2^T`05aeBh$h!oFZoQ~Tk$WzwHxVR)k!n0
z#iW;emG@HjlsH0fL~D#$Y`WD_B?Q?^*#Un-Ny}4YE|huxSOHzn5BFmCMP1(QE*=@S
z`fusfPpQR6NGb;-;+L7enFvD3%USszh|VhC8wqNiKN)gyblfvcJ4`oZ!BCXKZL?mN
zbI~hzwfeV25~>t(evNE*m`#`u#Z2ThUw*O5eU*EOdFqF4ve>&QYdYSRf^CUODZ6*d
zGDXm*gTUND@6aLSwrzfjUnQKf%W(oN_mBBNs5Mb#eT(KVh@`fje4USv8HOi5YtxtJ
zsHwC4LrgLQfq?@d|6LZ5T~3rmq<{0M6cX5{xqE2T)JjOY&9?V70ArT3O)UX`T~~n~
zR<q|(JK>I^J_oER-uere)6@;Q?%N--5r{umHFErn_^0HK2AB9Nl9?pmDc3I4{0&?+
zK0E{9Zii8dj5x+5ac#<;j<);-NYX6xd0%+PhcKAG$EAnVR3t&STD0ZmIb=^?8T0ci
zGz)~MVSRT<sPjrcm{$k|SLbw~{ayACinBo)K{6zX9VR?Vw94c-o~2*0RVgKNyMk{q
z=;~%w`%0V|^DVFU>Aglb9`-*vAoU&~a2jo?<~?vws2ArV8pK5;0~%~v36{4lUEW)N
zeMWTEFdA3etMT1bL(2%@`dKcmwxRlybM+0SsE&52Bgsh*u#}0ZULF)e7MNr}%u@Do
z3V*Cm-@ZQ2JUx{D<^Le_EjW7jHd!OF`<W~O2@zh4DE#vwF_K}25xkNp-j9>Y$SJR2
zbVuG(YXI)k7S&1A1&QDBJfUIUZIgO;z74s5`(8jC^JjxUWf=1?ZKby;r3`+~yns*)
z=8NatTLs+2IqwgrsiS10MZvBjDLP2GTz}^&Rayb19tX{A*QmMaZ8C`sjLEVJ_<M>C
zpF;Kn<_`+NK?}RC)>Sog-+!jh?~#emMb5^%!;4Yup_G6`0F6U(TUb%$`GwY`>B|&N
zQMTBuQXy(2QA~n6`{4Y$j7T%E<exCNu*qEzVcI@_&G3*?h166=*LDoEBZ1%WZ#3m$
zuoRQ@dMNl)tkkZTxOmp5X|p7J3kV%`ZiXAH+@VgJxUUwmm4@DrL3*(TYJ7hnOB7)Y
z85+B(>tmbdze=RA0Xq50^wPppCRk<o&xkJ2gl?mPcg}Ap4NSFPmY1I+w_VWbeY_q+
zC|x*}O4|nM&0ExEIZI$&Ij;jlWybNV{uyp6>siWy;JP7pI=rb-){4?#T5)I!0qXbd
zgE!suY1~|E7V3GnI0PNknU!)WX8-+8n`wn&4nE|Ai0I8VHT)wpVWAPhzkmr!gYFK6
zV~@ps!qIKa4{atx`?>HO%*!3m5xq#B5E}vV-+uE%9%7ih!^<6szknI8fQL(eMw{8F
zrf?1I-o5KDLi=8ps4JsDvR=5zuq01qg}k64>YSpQk`{x~(s(=-j_JfcTB)9^7L~N#
zv7w5L7;QE((C^NDojRB`k*|d&=^aTb%yelwLF}$^c&i<&!jR9^21jp_B4J5Q-05j(
zatrdxaoOj(3>R^@m*=I8l}>GPKeW|VNk1qL$Y}(1&<&`0^Q%^~SY8?t@Hz7D3R;|o
z7HSV#`nV?N36NE}hBdk5051;qc~!kMb0I7SQTC8Ky~<=jh;waeBTh7<cy616!zZh%
z8ui%{!B^g-{V;_hmtUP8n4ifrgF7PpF}B{_3CM4Xoq#XcHQo1`7}zL!$ipsP`7Gtf
zu8ki`4HjB2Bn%}GC6?ANRSPBYk{ZbkTm6J7lZ3#Vbh0U*2EuhbQc>46f&Wjj>Hj`k
zwtE@rhqpx-Xk^nCh7+5Jup3qd$~$eJuT~I!(F0QP+ZTlw4O}@P<wMlnbTx04E4)g2
zO{)g0qR5F@=R~f0UmA5wOH*S+yng|_K2VOzZ~u}LnMSPrCLb|tbJ=`I#$OnU5IE9G
zsIQF;kJSF8nb|~7fwW!?3^PWOo3_Sn4422E?Cw9@16K^SO@DF{uFGitY|lG$jkL;7
zUYo866xSO70kIvE;I%Rem1Qcx`GvGbn~G)DH%(WQ>hLJivu!;n<#tcLL(`<08)nv!
zG^7KUNT#_*58XX(ss6U)=Il+ttXDw<#L?AMWxcx9ujR{5AM^T8jT+$aRLlZdxRqF%
zTKiwnr>b|?mFOnVzW#p87uxrXv(!V}2p%OMpSpc4H(2s}q<BIQ-C=gVL(N!i#eJ>B
zG2PcF3)W=>{divV#*qa!#R896_HQ{#dqvGRaZkSLs6O`bHopoFF*D&;&WU*Dzj~w?
z`s(|SQ?)v?OeVLgiYLglOKWXTeSo5(w_Gd=s3sm6nUs3oou(@BbFQ-Sf}dA#chVhh
z&?}y`qyz*1<a)vq^gi`}(nyi(%)C`EesUjje=@_%5Mh9N_$7{sF--n69T=_=l+gVC
zwob;lk2aBSGxDtIv$853|3h$UIwygczc|lqeDTa(Ao0hpG5B(qdEZlJchfr~y4Lvu
zE^3AfHP$cEm$j7;Vlg2lNlOl(?-KexdisFxtYh)=p$z#8S=Yx&*N)h05&8BFbrbJ@
zve>NFR0QfNRuZ-bC&^j%d+gW`m*uZ8T_X-f6URB4+KpViM3`46si{CXz4$-624ACV
z+{GG>wP`|H=Ee*yp)WCx0A1d@Y4msG_kVB+ya^y{YV)l*+<j3oV@zuK3vCEJ60EKf
z{1e*C2<FxSTcjr0d|S&wGg3iFIU~7z87wYEcZ0^a`9J$W3nYYdewN`aJQ#jOqPkbb
zSyYbaGooTXKhDueMEy{=D^S_Wlw{G87+D<|PzL|<h8yI2$_B`Fk<wDgkoEcgqPg(E
zIaW;?%6hWcGy#ip;r%!WRE9@g05q^owC%477Hd5$m89ke0BMw1Y@fu9fH{*FtJ*$}
zu%hq1M*IQ#3euNlDEJNKR9`d{yUOZVYF`#K>8fPOZGI>p#}0e^E)@C#i`z={I_c$l
zs=xa^YpN+B==%vWzOinBF~=K%^?(?=8{`TSu^h#J0pY#V!CvP?DUZZTzGCTli(LMD
zR6t-|m%!ZOL(k0aj#cZmFsD?#`>&T0fouA`lprpVa@%X+kEgaXW^VZoj)H_5qjOjS
zazk6on~CC$egtm@Q+NPv;mqzUSE<{g9PD0*$$eE##^*ql$k50FYSOTImaQ1(nNzE|
z`KYhPWLs-$h-zdsTR9z_%$1%$O>#FyjZ4Fu`)&L7-W~HZ>v{uVNyMLLNfOdyk!0RS
z%bM#uj|&Ut<tJv$9>GVM8vW;fD4>z)q`Z2Jb>w^#%dqjn?eGmvjTy%avg><S)dOT4
zT~zw&>_pHbjE96><xyW#hs4sNO;zK($)4y0#0`^KTCeRg>}3qyfCnBr@U1||goJoh
zfvD>pDa?Z>DxKEYEw89jqP?Ih1Ir|JAEV6s><!?jASySGudLvLQ*G<PzR}24H~GGk
zUw&&^fUALP<L(|2pJlvQeY6iMDtZIU#ZpprX~hdHnsT!P(3(T@_AE{F<&NW?Iez-I
ziog<@L-F_z#vx9^qgO`{_BZ`CdspWhFB}`sZ|&ux_U%55O~8#nhIeQM29*btp&Jt^
zeY&mdqQ)gV=W_#6=qn9j(StNR1U89L2T^grmG?+iEF|x~ZF7=7%=4~hOAyg3qS&$n
zM)H$TcD+y5kY!(sd{E=V_d-n8vrNz24be!O#Uy9w<=v@=3rqPenEYxpm$~D2W^l)k
z#&*qiU3r+$wp}>dX#k#13)7MCPi91ruO0`d_(A7R@dlu(6IMRxXmab?@9U4_Yi-&^
z;oYufjb>Xs#YLq`C>CBF`pBebEdJ9Wts_@tAvgSfEIkfge7s?%2i%u39REa9CMf+E
zfL+CpWz&`j+6g{uy?!&s_Uh#p$VqaoMN)i9b7L@tH*>(RH<&jlf<m>+?OoHXOQtJH
zAm*McVbNM7oGm?exLg_*{wgu8oY?X>rf1&(CnPoHlR4l1)PSjL+rWZ9osn>hcGaEO
z|MA9YT9JIXXF!Xy<&Hwc0_?!^!yJ_x)hA0!5q!LQsJM)6QOCH%RsG!o$Dk&>p=<OQ
zokd9+TWcJXNTEP8nYFLN>5a%IA2sqJFF9u{g)?j##)<@u^$LHNfCX1|XBlZ`e_eAw
zQPnnUY~9x+TaK=4-PXdxAVOtY@zx4aCQxYzY3^0TVn5r2veQUs{k8LCMt?>5c+Epr
ze&IJJeL&o^k}FhkS>Ti1>KycICl^ky_dU%pLYW1Y0uDnkr2a-zvT_8w@U}g;q}&eK
zBSt&bG6CH)$C=|F7aT5*<ah2lVVw@)hX|wEw>LXs<#JO;WPzifG1dV!zD*P7!SmRi
z$wp^&>C%D{a3-~~x&;H;0}b8Nq`2A1lG!VURhFh|5M((1d~5pDc2d&(HU7xw2SKK|
zapvmfTOFhkEx*tpg!GVr0dW@O%3SBatY5tU4mYhgAO4`>;cJPh*YT$*k%=CooCBQS
zM|GR;2^Teq`4NuDD)L;MD6}o*qJ>~Vlw5v61xt~lD}3%Q4b#Fe&`r~wrshW~bp<N5
z?{3nk^Yf^Ak*50`a*1zTe5tA14B{sv`~{dEdeMQYm>-ZmAXB>Wl^e5Hqq@}Xvgv}J
z`zKdzOq<s}1L|OgHA~I4e@#m4n)1n4RJYj0zqhA?X)n@nK21-HAE}FAFT-&kcHc3=
zd^A6j&p;nI<r4&*`3`I{<8*-=9W;8!G-1zb=MO*Uphu8p=l*QcJ$j$}E5d^${4+=Y
zy@L4PN{jyw23#I1xKPiYH4;Dn9k`!H8DIZT#&rK1w)k%4SAWiVUiDIq$={6WA)mFn
z$}*O0TyWx`$JXdNZ3nfgCxT3hu9i5X96)_TV!i+FkpK2v{~ru=mj=CmbSL|Gn0VtH
z|7!~DcCIXHbI-8C*{gD99%FC)0F;vr>caE$6@d#jp$ea*HAWu91}$Q>&!Z2NOR5ak
zylT_V>>ZEwP7!qqc8M3%#OA?%{Z(U&u5UW#w|pfhtm}`Em+w5vAi_zLQKV(SJ0B1d
z;OBqLGVoLc6J$A7t5z#=fIU{`7Li327;8fq$@tf^bZKqebhdY$j&31%jm591@F^&#
zhFS~JC$&c;wPU^yi;8Stgy7Sw4(*He@oDgTZ=`cSF^>)n$jdx)XezF?|G$vvb^i*K
zzq!;tvd?0z`*tJjEe~<eITiUka?Zr3FNphlU4MVSaxL=p%+&toCa&y?mNf)PjrYD!
zub?`OB_bbcA3Zi<tsAEP`Q~c!X}9C#B%SM`=%BMFF1$=~k|K}S3duq2Q0CEBe7b+H
zHK*IvYdsxbX9R|tT8>BsyGj80p>N1p^tL7|-cxSnS-U)Fn0x!u>u7z0*S@N{F#dH$
z8flbBVrtmfQ&keV7mdHH8gcvF?F{^17;KqM%hkBFS&TE%aABkD?J*V~vN?!9;+FWM
z8N$7y%8XliU+E~*)ip(x5_8GYxGPwXzp$Ivw(XjKFf`6WRwkRPx6oGI7n%}Kuve{C
z>Xoj|%6>}}+PfsK5IeTVeG+I3KMv`!4_@D%%X_F_wuQ5erYvRX)5GF^o5BB~;5ZY{
zBU)(RihLT$7$SLeljixS;cwLcuV1`}nLQ4%dHRLkvE%2~TeuOCheP^?B0}k&Lbk~;
zl~&SFFtL^m=JtEFIFzxyWj0VFu9A@6=+!AwTNZyGmYP<P^lnp`q)gXs^p~&PQ>Vq*
zmQG(>6~1AiI8m)-Ydk$J`6(Ij0@ciSc<2*UJAdo`l+&-_X&VuY9mZ8%tA|w}m1Uq$
z3BUsIv2lO#f~9XcTnec@Ge5$oSA57Mf%WA3W@Xn&MN|btxe^Alo}(~8b$a?(V&x4>
zth{*v<UICWU25tqC4H90RLaRPIPgP`)TmO(!6qdY>sgZV(=855$LV7#&9w}?Y>rtP
zS)c&ujn$^!v7S7Ii0std<L>Gz#T;*@cBVtC=k|Zn8U$|XxO<{$S}v=Z_n*G1nhzqw
z3LREiT2@n;9V*3^mzlj*i%pLmAP0JJ3@!Sh7w4+TZA;#7y$wx5Sb%7GQNU6tZ-P7&
zO8$!-Xy-l9=K$M(czgMF#TXruDnj3vY3z#VZuCs+fwkVrqwLewxZ6%rbTrkv^91bf
z9eJt^JbGlxNA<_4=poS~*c_g++jGP*Kp^D6zSRW6w5`i4j_ijtoxF5#f`V0pLh#cQ
z0a6E6R{bwJN&s@{J?}dFexpD7hZ-YNpy5mGIEG-H;Z8pPZN=!5%T<%dp-Kf^)e(PP
zbq3abQf~}F3cl-)wFnvHye>EwZ<>F=HfrOm(T9CG7heH)awRt35><D)Zvq5SZbE^y
zSOGq19@9<B;@3P2Z!D{J<{oJdysC3LD?V_0f_Iz*DBjf{wq>DYf$;IGipAi3l2o=_
z14><*-uPCE`3}ZstrcPAkaF+WshR8o0VzMe0x#rCKYT=vjQ)VVdDnnM<~?$3nm9`x
zfWho>a`X+)!YLJ501kkg{<V{Ps0E~J@`SC)*6zq8!-%e&6FR5$xxo`YR4AHhrkVS2
zw0QC56vvRoj`y7w!@auik>*0{<qb-@6H{WQe1z(cbvj@bLcngFnYG*F^P*q+xR=J}
zs|TN%gqjUW`qS<0xT$l2oF#|8?E!xQB&Q#<cmD!<)w+iqNZ%UNjc;#P7d0;m+Bfn#
zihAZRP6!Gq3nrEK@eAJRBzWnBF5Et-&vPh0I$G}2v}gc@htDD+i^L&RpY#BBUV9K$
zfr?e+EpP3i&OCajWGwz(kQ{-M^^{JWXzd%g<g9RLlvu3sIi3h8o0+Rlhx6!<Xlf!D
zIOO8=$p8U&`gTM_?!Bz^MlHB)oL5ZyW<RL+e)qbuE%N7CzE|2Rri2p&CCwG|lJ=Lq
zz-|XV$=BOrt8y~ORvyEfoi;bu{;cBR%%3nTDFPswFaH2IhSp3TZ>Tu6|9)>4k|A4t
zRKAH2P&43vq|QPKZ|?tK&&T0$z**z^YAnED`5ycSf8)@r2INzVaL~6xr8rGJ5JX?4
zjtND|M9ull=IF)Od)f5?Q7@uLd_NGb_}z`uE#g0WO4@QWc^j)pFjD;sqNgmU`w_nV
z>kivn?V`74lgL}Qzf>Ar-L`_g8{kn5kLy$Sf<rowkB0uC)3~<q1<dc>6}<s9%Soux
z%{2BXO{~~lCga1$EDRXI+QW4?ZW-?=6nr%^Tf&Q8)Rf;IU(~@DA(4!+r9!Miqhi_H
zi5u0NKYISDn}TmIiZ-6V*guOon7xIoQC|KaWY5E6*(;qI77itO0Im5Tw-v-+^wKQG
z0J#hH!(mHO^m2#{E3shdMB8#;1&2hJ;jFx?=fdsUhVM1Ew9;1xNXM}}3N|W(Dz&mV
zH6A7Tau56U)XAfDH{E|l{;M%odmGVYdD^wkO6+|yaK&&*t*CG)R=IF-bljx9<@|ZB
zx`Pm<oIWj*WlOb+C1Qub%Hh?gz&m>eZG_wQOGA#@@vV5C@?T9tCDDbCvWICY#+bMT
zjt@c=#;qbcm+7nKT5zE-*Y`iqgOBcfT?w8*V4a-n^t5@&)FslM@;iiwDms$?5ikAb
z^T1QeCq<`Kwu-m4h|gG8TizvKAM`Y0R2r7#t}whbpuqAZXTJisHwT6r4VYT1Lhw1T
zavggM+a)+-&>(T<>YzOIs^WO})EjrmBvy`F@NX525=YnZi}4)R-PkV(e>Jwj+5T{G
zHc-(isHtq4Y3lq5!dh&q{`lMFECDBni$s1JFnYSd!g&?MqW;wB_2O*@oUmj^)}%3m
zzGoa1;~(EbQ5N;z+I#PyrsDA3Gbl(AL7D>62`v<<O7AV9hXe#EA|M?B2~~<n5u}DH
z(jkxp2)!4TUPPpXUZe>G0YQ4v&2M(@?(gp0nJssBW@q>QlatKkoIl<(bI$kszR&Y{
zKt!kjSnz~}-Ox*+GqkeSGbTn<>t!?O9{?Cpk*=l4bQ?((k>iu7`H*E#o^EFSe)aF-
z3iml9pPlFJcrESs={(zZd2rGK@J~K9KAnOA@&l3uMaP|Bra5GH*)IO*xFOlFwdHS~
zY7G^H7$Kwd4cuRF9{s?GP9e(r^{h2JRs`SR*SkPS0Y+pXw4tw2hiXDx$&9q?-hbmJ
z+CH^6Udp^?N)5HE7n2Y8dW=nwN)u!%XW`fKA4-~0P)ht|xmkB+JCAiu!_Z^Pz6qIo
z0rkr{3)M#RSl9G}`ATKOr5LfYr~HVGNcPV8{PKpA&9L$iPID7I118l%!LqbI*UjtN
z2AXz;N&7|;0p&UNmo*<_>K?(|SlLZBuO%mrC5;V4eAJKtbKB0GADm5%oKbv3cj^yJ
zOs&z3+^Dz%V+?;X8;@L9^WX^yk6{Sg^{R@k%aB1)+gQ=AhN&qcR{E&=i@XCY5<}c2
z<wpVxgZ+-}uw%+Rd{r~$ja4oEjMHt*7ZxdrLV1qQG;7J9;WT7VA=N*wX^S6g^e~0=
zUFCv-E~Tk)SS-Oooix%xht(wKSrFisPWk!RoY|(yN}IKHacr28Xjg(WCoT(q*-M-J
zhurqis5t$K!>x^sKfZZ;Xz0RHIFhLxOOXdF9!YS~UC&9$VN4E@52_AF#Iv@(eI-~K
zG#3aGJ2`j3=ntDq0o9%rT77c98^ka@U7mSmPY}j`w0N@GXuT&bfm=WcG2(?2K+3Ms
zx?mQugw6}xba~SvlV9K(233m8NF9xjh&79%(E|Fj$c}chSM^Q0vi2M4CA~4*DB*<i
z$^E_C&HFMPLFbP5uoOs_5t`UZ*1xCT^*3L!#s6E-#+SdeGFsm{f;x+i{i`aYk$HLg
zc$k<8Gl!iCTWr7D1yM2$_Y6%h`~43f3VjHZ#MKT6alK&x4z#4x(y+bCd7E>z9kI$u
z$(U*-@-acjC||i~?p<;Jjt~RMh29<3hI=b#9~M~OJC^Rqw9YeIjeX{~@;$1)0f2S|
zRXXubB17I@thMH`AMeU=S5&nbUoqf7uS6yrypZbA6ld`7NzZ9$orZshVMQ(7Y^*&_
zgV1QlW++7gj6eZSxjm?Dq^di*{>p4MprG2>1D30x<@(VW=A~wp1{YlYz!;=6v+<<=
z?_Fx1Pxphc2Wl0+UQGD6d{<LKNuGPb#JIpIA-Z68@0i+yhz7l)Bi5-Rl#nESkq`K=
zh_I6<Rbtgj=xX7W5H*X$n4}_+4mR4MhTW^DSGwO-zNemLWx;)yMm>LSI=dl$3a`UX
z=3HUjZUq?_?^m(xBoL@<{90?G(QJ&v@0PZ?MA2uJBp9MXaWzXf+up1e<MJX2*q=Hb
zOJYs?{Fc%}Rn7dzl8w3B(y2?8N3Qi_rFo4)KOsz1#mvBxoMhDj-p&vkrC(`rQ=NSO
z0LtIhOD4ox8`leyTIgHWRN}^!$GdZ#{sE%Z{{iwrD({~zylmWZU3}G&Pngi1*3&Xz
zQg=>Y|N2};D(98(6FCYalcW>@C62tYe*o$z;`Lx*0Zq`UxffB&T)2}EQ}65vHMn0o
zbgG29m#qYIduM`^-X|pb$xB4z(%!?Evn?6XTh&Z=2gMV7OkfWwQb{8#skZ8KIDVWE
zZVrnUPY*6slDSX5knS~+)JI4N)bgvX7AR}RC#BG`9yj*x8b%eBVmdvw#=uxJ?!-J<
z!o8aK!_}NC73D>%KP#)F&+0sSC?@W+UZ!J<ur90U%CbYV;^frG3rB&fx$e;(vBXZo
zF0H^%y`zk<kyXpqyxW$crNEwwz5tGbs|!V&-}8kH_i85L@Oz?qd>O(H^>z5ZSfCNU
z_=6(t4U)^UPli=8GW%yq=&+BVPjRMwoTT{r`lb+5wqtS47CRJODfkBDFW(9IC+~BO
zJE0q%_iC0Z@tNw|Pm8uhC17AZdJ~CgInBU7{OKCl7>zeuw{Gc8xQ-|_cRuiJ9Im7(
zH}GJG41gmXu`Ks;m~=!SO7BiDgCYoi)(_k(m#9WW%W)G=p2Iz~<OzM`l<Mi-eAYK$
zUQ&N(G^Z7W>gUK{HOu}y1dOXsiIxE=;x-_3jB=E2%gFD3$I<fBieNnyuJyT#=!C1U
z5OkE~dPCkg<tMkm%%SYYGT@@Q&gd6@&CI#Sby2B#@|0a2ef7sBu~5B4xJ5^@)5d8o
znez11=N>lSMdX17%D1uDthlNPt|8-rUAx1jd9dLmKpL)m`kaE7d~EFD(t~HLx;vuW
zmAm+pWz<EFjzwHx#En44LojdF?%Tu;W=N1^cU$YU>t2v4)Xb{>cqzM-Y0Mg%Rx&K2
z1WAUy9Y{&soV5;C44U0#LL(A8Fm+F4=znEWZ2wGwze8%Kqd>Y)p9os0+7;Ulk=f#P
ze6J+&w5*CL{XUf?1#fn?%cNPe4@CcU5|q}9*8M)m*^~0pqNSmY$w`eHa2>NqEIcye
zb&v1P{`bGyELD1}D`jzV_eY%7p{Sr_g=U6cn}fxA_B*$7CK|L8H+&LIM`207lDlXC
zY9}CW>yQXTg^hIpSNz-L-xGZ(9QMxVx`ipEi>pTjGk$l3UIZPgRq|Iv2Cl%;TA$4)
ze{McIZMsf10f`$-c=j&Qs#c!q*=FLavm>9rZ+o3BEgs6Hj~{sSZK8jrBo#_RD~4e}
zQr_hLK4^hY@EXgp|NikvNkyBvQMrjv2{sxeI3)Bum&bx{m0^Q4uh6QMBUjwWiE~L&
zA5UOMO<rVtalqD(YKSnxjv0nN=zI{8GY0sp2OXap+dY_h^b*k|fjF(S#kkd96x-0h
z6wZ)=c;^(dkO*=VmlbTCFFiPAfLk_M%d@V9A^*6g!x@(ue+G;<cfCXM<+U3z9CDom
z?H13mhHTYY$@tYz%b+!xAaPIdvaoQ{D5gpudU{&1T&4FDzeLPhU(y}joprQoY$lY`
zfqA~cnj?Kk1#@WD$m{M*9l7oNY+nN9-Z#Pg1EdCpBCEeg;k0-dy*U|c4|r1qWCuTQ
z&tpQ<6^vRgJrIR0yNxIeUb;5)_lSF{Amj}vb<H0rEOmEsle})hldQ?v|CLW!qlRtJ
z1Z$patWa5AkOP%PX%g=mSqSA?xVF^9Li48Aqc!L+vZ-b7Cf6Ygb7NE>!+e><5Hg2h
zgd?{@?N^uo0aHiQE=#fdqaC07E|d3%?37IU_XU&b<@l82KtQJ%;N`Km(Z#!kbm!%b
zS-kz9q>1aW@eucX_{LgZ;3nI_bYd7==lyrV3G-&NewR-cw(+buGh2`r=?drrJ$(^*
z6IVfj((StW#{2HBN1F1D9$zct!Y3$EJNE_gNpZ@Y!O3!L-Wl{dQ64{9-7d94VvHxT
zxJ@?*FcB3N8o=m%z=;6HJ1x-V;erCzR$>f1u4zzh3%)kR?ThPL%n^k$X*LjEswcdw
z{ibOg`ECP0|2Gr2|0_T3y4x+bWXN!()4nBfxHzm7<M$!S^peZ4KPO%?VUZwt$hoU@
zUSr1$SK9h#q<Q~3k$RMJvVOIsUPlq*sVK9ZdppqJZ)<IxiA{2#4%|9XLiiY&`04(L
zX3B2Ni$!a6lV?NYqJ~#hd65>GF;*SZ1%@`KydB`kwO+}3&Bp~KMi9~5n_Obvaz!wn
zUbA4uT9vWrbeNX9QCh7bSeFWHXk_J(q{=q)m51o6zI6O<J}V;{GTvrFAd4Ui3Ea7-
zO{qIb(9uU4D%hRxCB@8p{vc3=7<qP>yUbq+K@1!PY{dnNh}}Cunh)en8kr(e6xy`6
zA1OpJ1>&>gjxnAM##rWZ!iHSQHwu^w@OXfIaQhegdjt0<zG6?YAe#NK7F$vLGL^Y9
zJ=Ot`Clv1xP7h^Ns{6IxXUDQ@I56Sy`o6sF23gm!mNj!E)C?FX!3!Fo9ApoCb~22a
z3b4>>-8O|6A@HY=ak#ML*aQZmdJc>Nu1(1C;zy~w<mNh(t9PGc%U!S_TxRx{zM_pM
zuXs{((n>-H1c(8PYnUL>mn(gV{{ZFu_*N(A!Vzk<;1G>`kL5FOt)|iySjJKA#7NXp
z*n<N+N}F7<9wz0-1{3C17{>eh!jinPsv96KjOLHyn_NyaEx8{e5p8V(j*Do)4Qx54
zaZ&Kq5&q~J7Ou+&GHM<eAW$eJ`9G;UvoFlX$tS!lf0^AYUQ){HIta_Mm+kW&B;M7@
zP%-xxbe&`FLw+gn11P~Q{aZvChXK1P(T=?eOeXnavYq-SM~>a)XGH3r_~i>a<>`Qc
z+FctI`qR^@`=G(HGHbcSLW#(56q(ZsU~|QP*SbWcmC$Ws|H-5j7x$VTsS9OP5CXBQ
z$Do5t?cy8qC_L!RnZ{i&wp^%cd5|f$Ep^Jc3TWqJ#J3E}C)(hpuDer$4#Iho*lf%I
ztbvMBc3l-fG7a!$@z4At4bJB4scNlH%L$^3W@cGA^$?ja!&IFpjDBQg@U;4EL{$T%
zr%b*P9mEkCL0j`GI{mi`s-h{Nyy|iG@Pas6R5~t7yswtlFaU^t9WA+T-9B^8+h`v5
zu`xfcg-J%7HIVxFmS!3gnJ@ExS`3$Xk!R=F+^J2kh7%5lZTg}mn|muoaPKaZA?k+2
z+h3<aw{<3**?B!DU<fJL#CIZwuMTjpqM{rpA39MjMKSHy_&|SeIvs9<**$$ix6yB{
zPtFs+88;pzNYngFJGi-T;s<hs%!jYlb!e^?hGx<rleB~Om+_$Nf79}_L>J*pel=Pv
zaO4y=nfAIQG8hU4Y*48&8=Hlm`1lyT&Lp*MXWDMPtR+;+N4L&u@w9~S6zVV84hjR=
zf}B)b0@=QDS-j?O?wNXub;f178Wm{<QXb+rD?#gKMWzcEG~p!S>q--IdbfVft`@Y%
z7Uq4>^2PhEQ*r~JB^Hwjl&T%k?0uEkF$$0z_Zy0R(YoZe_qnfLizQZqn~Imji8P&V
zf%;9}xqoHh%)IvK$SdTmZx!w_V^4ZqayZ>*biDV?<4+pM*+{15CHkYjhK@;_N`Kb(
zOmOJe)P6SgCuVhvYQP9&a1Pt+{LYy(_=G}OQ1rRAt<uSgenIQYN8D`2Xg_m+AxtS&
zNnuUQ^97yRud~;+dmDSmU7xlEXB4XR!lCGcppfx8JavE}KlaaK6$ep;V&~d<vHOq<
z*<BKKnTbdSerVd2)thURdHkSAH~UMQ+tLmG6Av8!Zg|4HO66L5Vk3uck_Bw{<gL@>
zW%<}l{A$ddlV@3LV8NDS!<l68%zuD@bs))@&^c{SX4nflnTIj&`NX!%y-fpJK+RKP
zb=z_Fhb@A4b==pg%4a+qyInkH=q49PCm%mC+32@5NOY7NfQ$!;_@5^|U(c{hn0iRY
z8k)GmcAB7P?SoRUbt!ry9{BcqIq=QNv#TY|r=n?oZXrso%SP-ne9|s)%=XOs&Nl&o
zbCK@8TI7Q$;1NQQUQN+8PLp#E5>0Ux_1V{k#&@v_I&LF+5x30<OZUd>7yzAZWGe&(
zGBpyjzDXWZW@#x@t&BqU(k)u`fF~cnkbo*mb$OoMBF}m|rb+9icHU}}@!PI+*IX96
z-!wP;l}<QwoS*Inkat#-UiX|@G42K9ZueZSOpB$DF#FsG<6gZfq$Tujt7-qu7=3$Y
zHr;-;EjFIlcgF%}3>(WC*Eio389m%TxZp6Vqw<{DHO!m)D4nsx_DU2%bsK}tRfB#j
z5SI!E+{6KxS;WL9L+{i#{5&=FDd=3NOo8b-N_R;^vL+36!Lr1s+{XF?@js=JrW3ss
zeCpDq4wpadrXKUU|9Vht_bB*CSV+Bn0K~26SY{(JC~&5>WnGzN&!$Wv>QYJd@yQe}
zRZdWg)s&kfr{%8lf*h8;cwGA8hBY~RJ^!F2m4nX1C80p_ufqGITx?e{O*%d=nCCTq
z6!nq~=NeYD8hC=pU;|NfmG(+;hdOKhhYIlbT_?HYb3Z?tJ$d<9&o`~^yavjpjV2l<
zOg2=J6hQw`b7?6GTi<rE?cDwjy^x4j?C>k}j&ZXwFb-kQ7DpqF%l)#mooQBm2C5#P
z?l%dU#GmTk^>JB^R2yA~r?Z7*sXRuCbXAo<n$~wy66MHhoi{Fb7gBc_qx*sN<fW9z
zpXKtslxJ&}2&E9^Z%Pa5Ymxpint|11k=P#2kowdA<Ip1ue)7lD_wVeJ)E5&+Ly3}A
zk}f6YLaz7YeJnddHHeNL=l={n`adz5{D0MHf#EM+e3p`W#`2+>`s>kuA+Y8===jT*
z(vTe?s^nhBvuTr+Dx>~0tM6SzCqbJ8m@+-j!cGjwYw}C-KbmY3MB~*gzz<X4`sQEx
zG#!LM+i58TrT3I%AF{L`9Je*JsS!O*Ta{_2OjB@H1dj`2c=BC%N>a^O@v`;LZ0TW+
z5xnSaT_}vS=dQNeudCM{M2oCt*!iGsuh6hx!qAwp#a1eRRW-M6S-c?Y*&9;e>lx$5
zO(MLX6{o~qQNL)G*<pyg%Wq=^H2H;Oyn!@o@(Qu~aUkrE;LArm9@a44!bVyVoSjsO
zQ9Y!^H;n6|LWyzBMY`Mgp$FaVk9YV>n>_x$DTr9X;)HHPH_lRY-y$j5>%!sbz5pk3
z5q`6PdGteK)?9tY3ie@>0zbN59qQ&%q)HV9e3KIie3Kd<|7K5o_wr*qQ;NHlCD9f1
zO$ddR{wXC5PVOHv6jmeW<ysG==9l#PF=F(I6=ywG{yqDFZg&83!SpzFgqxX@R`QlA
zz~?6LSozzhG8xY+&mzHxbKZ|G`69J>)d^iB<taj4DC1Qy*d_=_lK56>k=L{GvXKAW
zb*3`Yjm<Ke`(w#rf0m<8R-(E@f8p*g<`<^254=uaTG*c633h218|Pm$%!8lGhcjls
zrhP{p0Wee*AYK1KUNUJ}7W_zd!WLohZN9vou%QoE7wo!EY3|M;BqT6%%(6vfnh1Vl
zTsBy4aapXc-nqkK&E~xbT3G`oml_)5WXjp#$uz2?rMK5SmP(fH_+z$YtPx5w`I-i7
z?-L`_0&Dv>@@^~1(ZhRAZ?Z=-8^#})zVv?;^cbR$5MwmxZs4e-W_6mK1%0<|r^LXl
z3ntTyV*8O3YL*n_AEQo$@pp8LJw4_q#L^YbN&+D2)3ZA7bfFTd5hSDtS-W)igO~6f
zkK@@E(YaLfh04btm0;U(MMzR!Iwa*va%5W0%?P5l!z#Apa_ni5yB~#g&t0RyNusqn
z6RJ#+tR5<<z>r=-o#QXi$gEF&&~4nk%KP%y4qk>eZ!9c2nbLI^T}B$|gcFZMl7FJ7
zQIP9ab1m8N;B=i7LLpP&)kLK|$yYscDkGaSt{<1t58O~)!Um1F6<Le845n|@Z_*i2
zd^((p<eGE42NS_{7S>?<(iROqM%@gf@>oGq!DNXjPMyBx16*RP?IF7frPRYsTB+fG
z0K{SDDBJa=AK}2h``N-x?4~?h4>wiW)z0Yfj=@hvze=Xq;uEK0MXM#1yOy0RRrd@0
zW|#>)`P=tR7+tLJ41dSzhKrzi1^Fp*BvWpuI<HTl#~NY{R+&_+xK0CbqVCU|`+SmI
zj>5!2#JUFcpnD9}-sU^z`qEj82L6w+Qd6$LpON2kN=f>Kp)swg$r3w<G+fautMXOa
z3Bg8a5Pt)f*St%}4%(y9+Fp?N`G)s0$5$!7z95FMqs|rA!J}Sm#x}b<-7rtO3B4)D
z-GbBzMx$3-Tk^DiO}wufI}eN;OXKc%xMu@m@$~kIrNrLvS%o)o9HzYZuDgG+^g(oG
zSH)kAEbXF0r$6y|wj69RA5Q;rc0_X)#-APW7nqn{+g^@l<UeL_`QfwsXZX9PVSC|x
z$OEs*^a07z58M(&74|rKg@V&%FC?5VDmU-`+~eCl_jU)p+O}Qe-fdyh)Z$_#wCHK1
zzVL;ZRlFX&!kw>7_uMPO%jX6OTcjCSmFT))<DUz??m@tyz-LZ?WG`BEnkNp$idy&w
z%Pp55ni2B09#dG*6F#J@rmm_LTCEF()LI7yktF^=Jd8;AwRoL71#3(Q0eRkq@Mb&#
zk-9+(P5K4bb)Jg{uR9>9rMPQ<gwY4@d(Q{Ch2$H&8}n^ZcfDiv?l$A^O#n)u&_r(k
zJINpKgw?fG1?_D@h~v|cyrP8~wh|s)*Bj7wZ&18Ild-(tFm<2Poo&v$;`m>9&z4DP
z?d;a6VgpnC$TZ5$?e6GcvO^=B@<SbJi{pQQcX!q>&DQm~n)UUJ!)xo~%!ScX3tUXc
zmNoImO4qf_FP<v8&OE%E9esfP_I4Vx;RDTNQi8714e@EeV#F}BY<YRnXw&!{!r$bH
zoy9FxIySWx>Uh*GW|CDv)l5VspU0<fxNaN>0ps7q>^IEkw_M~~3Bw6NZt0&is4TI0
zJeg4<-Y7Eg*}I`p!*jwDhj~Rgn?EmUjtSUS(RTWIXO}N2;$JpO`Z>IqM9Qy=w<HeP
zDT9?d4@jz{#X7YU7CBz^0tX_%PU@PISmLToNa^0&l)P&iwe4t@7d!J&n2oVk{y%7!
z?KMosBBV94m**X1-~@>9s)|*)>+eD9o20wV$MYb&SU(*RMdBJMyjKV~TGxT<%97ew
zId{VdA*U1Bp!S_0%egkPy$`T*_;Y*y8#TYn1<#6A_<WVxst4G^m*TEhQ5^=>uBfI5
zJ}c^)^g^%Cqf&D&ysyP90zRzV!@1sJXioPXKHQ98-be@-W#Shi28%15Mpe#EJ&nOW
zWDLqU#GVPEO+|FY`x1r!07kpWYwVB}2OoPm_I3<9@F%TzWCG!2(nQLcye5UcE#G+@
z%Mdao+<)ZPQtz(W`^HfNM!5fc#WPbL+va#KrlA6baNW<R0_$DY17VUP+EJp`NXmy)
zsTJ+oX4{CJ=Dqveju8DNSNm{*VfHDWJ_tEZf%?S%nz!TW+qS1}-X#+f<60G!sok5R
zJktFpkb;T|zK|QW;;QF#%hbT#hUE^4R`a29rHqViCN1e(g|66GL9ni|kve10dW`X+
z(cPxqn=Q#Mpa<+^)<k}JcC3C*f0kD;n{U8ojfU5^s)~OA^HMq=#!W)|@<R@zteoq=
zxx^&(EdMtD23+crrILzE8&BRMkMnt#%X1qDKI%0DtQ0!MqB~2&e8AHAYS*9@Iqw+9
z^0J8(<7`!o6e=QT8#o0X-E2}fa%XmATH-@Zs0>Rlc=j!CT<hw>6t8esS2<3J@+Z5~
z)KW4z0xSI?*vE!(*s%~z!B56q5Qlf(1~A}SKBwaiRW4rwMZeSSc}0!Ao{OcG=Z&z?
zAWR-{D(-Se@fr8)0mubOo)CFUf6K3K8Oc#98UD@vZ)3}76yBc&22x9BJf@7Z3>>HJ
z$tm#d(|JxY5csC}n9;IL&$8;bwteE79=ooZ+{z$VN`A_KUDN+v3&gwkE(4s)7k2d%
z!1it9Ch$QjJ4|#9=trj#eLaRN+53BZFuCtcM}hWtm+Ux7w^Um%T$YikBt0}-wF{aY
z^n?0XX;2JIWOv#nDnI%M=&4k>TCZA7+kI4Zd)@->UXrIr+UxT<3VrRq2-X!i$r}x8
zOk8W5y6w@p=v`81&2EB*5}-wlR8(HF*En#KDJU)?`vZguCc#6Lh^ea*Go7iv#m-r?
zG`-L8cx)e$SOhz;R)dzLl}#XHxn95Rzm2MbTeX|Jf((>GBf9-Ec+em$H$9owdS5`v
z-JmpLONZz1yr-?ZR)%%;z!TqUb8!A1{#2C?mK*>{_DQZlkUGTwSf{7pjxvYu?(OT9
z;!FriGqpuuTM?x-rr#qAG_{#h`->}f3{-A?_?=je5Iu}LH;g+H1I;O(>)k4rWjT_g
z>Gq#9AS}cMxVJ7_%x+iMxxm#$$NnCA(Kb`)Dt8!??7WQ$T`<T?GVuTrYh<Wr<PlBF
zEe`TR0r_Qo<ocmTP$P_;<NOo*viS}vHeJZwk8nXs-Orq@deM#V!cOuRsxBMCSRUlo
zca-XJd}y$IkwOpF(qWRE3J*XRk^ti`eSHlW>YLafwgpu^g;@#lmvoerf&#`1fq_t?
z9wVdK6Fc@ZADH*P&1SE~8*eydNWJqwHTQMR;h@fz4`*z4tEF|?<a2b{=|G&|o73c>
zc>6<!GbOQgck8{Xr^Q~62<K$)jp^dDej4q$(nzF+2PFfzB4|jqY!aIsAM5xEqYD?V
zEj;f*z?!AeQ#mZ}j?bVZ{rWj;0wj@PE#AXMD;la=vgR^VeD*3>Qwh$3n2f2ju{#mj
zvg7oTP(C4DW&kNoqRP@@9(sXD@0sv}V1+G3k4=V|)2WD3`@RxVKz@Cw4p2?M8O@6V
z(q~0F+RRe+J+c1>cq^MMJ%vdFzJf2uDi>%=T=AU|98flrM@y=*O-@zUU%#o%<f5BB
zwDR^fyqclYVx7%+KG?P~C7bG(js3CnMZd%~ujQv^YdmYZaUhUw8C(d`j_05k;5zB=
zQ{kmrj!)Xvqy1CDxupBI;b-jF$An^O!`u!6v&h^VUq(X1!(>SjEbvAi`93gN>B|7K
z|DB_G`Rl1M?-|kK$$C?sTKQByRTCoEB3hZF<t#7r)g#Fe57g1PXp_S{LM$g52sH=b
zsDvPD<m{^J$ps-D^3_FF6&zfR8PlSOHq$Xen-;Y;m>X>*$H1y>q!nHS;H<5+8mU)^
z%pLt+FQ2s4%T|l=Hn?F3{czVSAYbEqQbTW((w$Xa_*3A-X@W(WkSLzw&=ur@)U4GG
zqWfLIQQRq!DZaZD)wJ7}QZV6%$34M7AwQpVm8R%`EI4KbnyNqB4NWadBmSIsmYty#
zt$*wMl^}%e$*O#MD6$8mv-ZHl7!D7yc2>fxUI7?B0nRr8-KAGTZXqmK%-$mdComHu
zLIP_m`brJ>9c8CQMae98v(c-IZ~Vks%XeA+(i-^>z>(iMXlwbdoZrOo{$QXk7ehLA
z4By_pVb43(tr>TM+cG$drI<`hH;wCo1LVGkx`SU-(icxNyZ!+V=Z(SGdepeE!Kx#r
zK%9-bkOO-OBo%*qOX_<>(2?AT*xRnqPW<Dbz=pl*#3CfMfDk4F`#j0FpaPJy_GwGx
z5o@N}R!VKFErbyAxo+HioHg#F<2N#pWgybQ9GIAEI(c)C!u#S;YJlJ2+{eGkZNpEU
z#jjxqGs{&Ipq66GIphm3d+2r`OUqwQcgr0u%cJWWmwE#So!h=&_*~SzZY0+%L_MPq
zUU!<7VB}zb+!{SosRYKk*w-f8AJTc}O8i#pr(q{Kvh)Al(eb!rKx=3iebpLw`Pw;8
zzDN5mrdS~M_RUhYev@St0j2<}zsvVu+M+7|%7}%|R3@TkFFmYM&E7|?etL&St3&M7
z8BUGEgVumZqtM*y)sLSd6KvJ6<u6%vo<P6%E7pSPhc+St#%sZ9P?FKtPRnC4-&O;{
z7DKdfFEfL+mhyx{934_k3l`SnQKUy~s-G{}!@M8;d9wSr8$L$J?tzayDv0nx6Vhe%
zLm@~K&`r+!bzk*f(@z*rs43>T*Yre+29ELpNR&eKrrZp;-KQ;Iv>92iN}5dR^lB+F
zM%_%S6qr>F*O}=O`~BWj>(QvGhJFdlN+sGrmtEJwOgn5f?XOOK^<MPYqD*_^j*<-3
zF{uws2!40&42_{QNwIm!sfugaIcv5HWoUYAh@bpYDuZAZ#L_Z}GQEOgHN!H94do3n
z^O=1b-(V{c&OE3Ymx*mQkh5~(5#&(`6dZ*bwI(W&cdB#M?rCrD=QUL{I)6+T%PvR&
zeLMy0+vt<1-e`zS$7rT}555GfRql>8dC*lB6nt`Zks-Fjoe3b284N;|Q9zeY9}Ea-
zJMp>t%ha%!f`4Vzcxgr~gDxy<!)jbZ(YySFsTU|nI}uKdf-d;&jTWmlHP3Zp@)(Du
zPft$sYXu@V%V;(MzJxG=7Z0lFEX~#3_~v+8+@6N!PS*Q+m&y{l(qaqEVNn1{`AyQG
zoS1vo(j2G15=fgM!U^tvanBfAc8@y|@}_xFCc^XRy-Wqqe}I@+Vo_Wwuh#U`FlZM#
z1Cc@X1SQ>@Xx@ElVIW5b4TYF&W_RAn!tT3?Y4(@;ZGv6;^j1zt@st^_wMklS*2<@?
znI2QWH@!1<jOaayNInsSlxm`2E)~^kZOV7HBiZfwY1&bH-+Mm^2@J<UQr=v3#ynU}
zu=;&2n2+zEF~rC_LfJ6v`$zIJ-;9N6=hoHo5`Sa3SFXmJ^Z)=O?JVWzwf&S#YJ7n4
zf`PC%TAtIQmx=A1OuN+50umC@MMBkI^;`U4{8G+Zqit%wcm7p<W#6gMt3=CeU%Vh*
zzdk>f&HJ=kA|}(5d{%|JW_S0OYlx0_^iH78!!ms#Wf+xlS7tiG@**XN+Tm2GESlM5
zbqP4yJ}DvSC|pnqzNCIn%kQ5ZYvR6CR#K1@9*5_-8%f(5S>iyOdDogY@JhkPU&<Cr
zuzY~;Or&sGiR_iX2ZJzz<!*B9A<(Gk+H6bLcM~E~T>U(Zl8i`kwe!2(WLHYt><Dae
z2d}1^v+u8DEUzhvW!F#eM(-1dp(;V;{->!n?y$;9=Tr;NsWjvtR$g7EV?F&ZEPT9V
zT$ec^xKWHEkf6s=kg?M^mn%&tk}oDyhGZ!K@4ew6S<i1GpH^BL_GlM0hq}INc(h{k
ztZh<@E5}s3VR#dw0J6A8V$94gF}<)>Ows1O9zDI9AhoQ^=xr_Nkm7dPX{>L1-`4X(
zpvEdyV?^eQ2KgO_cpVv*L@}NH0Tc}&DDR0%w6%YzqQaNfL$!d0scz5CFPuo!5b`-D
z-tTt5v9=+j)b}5k&!rpAAk&ZIEQAm*YGPsa-_a(fI;PTF`t4rvQ2_ds{Wm$jTiwH5
zL6D<c8+^P(9v^`A)%w@tzm%H^tUoWQGE#sUpP|U0a{4maE1dHTdup=>-<^|;9q*#%
z(k-;^U2auCa<zI((4jZt^vfQqkzi4NWLSnOE{4UyINXulO7cd(4pkMycXQ&Snf6&9
zQ@K|5HbT2W=AxoCLLay`-@xLyI5pxqdsM=hW7DkapHI>vJ%9oiR?%T<dJEQ9_m_q3
zOY@Yh^-R1{Kycq<mlUgYfijx)aHUBfqvDT`pT4NPEGu?#$0KfJtf;186B`6KHY$K*
zH{>0uJ`bGV{MGG*dg(`HP}E>uh9t3%7^dGKswGIlkz~N6&>p&-vF*lDD-T>X9M4J<
zhor<{zF=VYmP^kJ10n+AkI9SQOsuH*PTZ)lhd)hqfF7cw-2a;FEDx>4U$#fju@^P$
zKFWe)GfXWkV6TVfdIe%XJoyX=22zJBJ6u-xtK7=<7dxN$2WT_D^5JCYA*wxhX*I4{
zY4>vr0(7Uc!%RlJ#<Ra=B$Dak=*dP%mn_Tzgu}TmgMX|nJT~e^?b~`<i&dgJo?<X7
z2I4R&ZbyecpP)C16+B7z$MkOfHip3qwNv~qeBw`p^xv-&Nz*p_;--j(hlaO{1j6SM
z1oHSv14E6dH<fC7$4|=~cRzk|B$BC>Beu0+SjhK8t5>JVg(+z{v{}5$EpJr5JoVFh
z_*0_X`Oe1-#iXkI!M@;gEAKzuckV|HpfF1fFcWU|40#A@Ffm+PmvVF!dfhJ97j>3^
zg(4}{nuUOuY-p!L`a>1RM~3tD_hP@t<yW<t9mpV}6xAqyMbbLN=YLQMXI!lOy&d%Z
zX`ym;dB;<%3+f_Ajq2V##c}Ti`_d=oie4n=Wni|jZNjIHgs*KZ*J$a?KMpFNcb2&P
zUTn5Anpp;Wd`k?UFZtjmBV_XP9{nw%sZDFi?~j2|alKZi(c~zT<`hBG6!+u{rG-)R
zsFiq)loSzQ_5bk?d2NE{$VV4ZUvuFRK7>9zHQSdzMF<+vobGUFygk-`v(Ga|sUj%t
zI_)5g80gFJM>b63EN8s+6!zVv4N-JS2KULxTE|37_vHyfzv)^iAwzP@RQK>7y~VO4
z=G{@a=%>_wwZLeIs>OQ!2D8b)SEo*;EbH0v#(u=F`Eng^Y~A>@Y0<LD(KuBJiVxD~
zBa1%rH~X(d7Ea))2X&5PmY<AVTrVuQ-5wtDgIu^F`p$fIH$Q+#5z3$2{sT4kU+$Fr
zxAukGhRfS+Q5Qe>Fwsl)4BdLY(6a0Z)9)!s?X<Q|&&n#wu+S978muNM#yjdpiGzbf
zu=vS;--epquKMx?yF;Vi{+;Om<^k}Z9ajHs6HS-QqUhi8rzRF;om;|go(x=iNA<dp
zbXHnNL5wQ(j~`BA{Y@Veo-OPEI$X}Dn$^ETdqDxiH$IOt(R6sJ4FK;a{pEjj|B36O
z#}#Hm+BVk?8zEczh8`uUUQZ*}^=?SF-?Lv#>-R{5d2aPMV(`<sx0M)zG8Z=R#5Ost
z3_vf>#D?^qsH2SikY%5g@aaI<=pR5%_zWeNF@l&7tpAQeOgW%U`xvQfoD_9Nx^4c`
zcK@Vtl=6X_xhD#Z6z-R^Q^&_GeGnr3J310M`Vw1b1_^sS@-k2NBEJmE2&L7(QpyRe
z|CnS{c}&yT_@zwHgQ^yvjTkZ10o8qj9h%Y0U$LBC_U<;WIn<Zpr>4xdC(Zd?2PPOl
znW$vRMmg9Q^qaE;I1o1xO-}#ulzeae<;Secdqcv;KnaSn)e&Oy8Nk~T%$iTfT5N7x
zUO8`3L{nH)U{2{NL(iyL_t(f6AS*jK$nLdfU5djr{<^m_IWB7Ql^=Q|b%6}XXIqfj
zIMcSYXL0f~nw`C{!u)|ga+t?;6GMS?{KKuK@s(X$V!oQ1?@s+PMus1ospZD5@+fO^
zi_BzVpvWHmc_T7)k`2NB<z`{ngV@f;ZiWr44C*ea=;Ir_^pjD8eIPGYS=l~A3IQNV
zvV11dv&6nU6^trDIPPd$?n3HeJs(O-D&o_>{Jar(qt=kb3~(G=**hm1;=C}_(LXO7
z#g^WX5{Wm432~LLP#T|gG68>szi_>sjS-vpG}mP4TJHGPSNge##G)BbX;o=zn$;k{
z!BI6nB}bT-Yh{#{nQ%I^>(=~KbROUETMdvEDDziGcGXtZ0b)MDMpG777n5o-5NofW
zf)=780rS?7FGi^#n&jsj<l+9qk&<xT7E;E^nq(%$68c*q;Q(aF&G$c-40ccr>}txB
z6Sw8<WvEoq=6P8)MBVkt&S@C=iv!w9-^8@Kxz$N|A0<66G#b=6KRy4|ki<gyv(L>M
zU`|l$;){|4N=zl<7Q0Yc{Rm@nRa!doPdde_kM}^bHz9-mUczE4`>V$Epz&Gl=5#PM
zAp_}%(5&G?7box0s?!$!TDzmL`WEJvO|dPVxIF0li3SSimf|ceO_Bu*b2;SeC-wfm
zdKC6$)BGP`$vg~?{H1(no9n%+6IxSOziwgU&pmRuHv}+vF8*^j<L~1B1;K#4@~HnX
z=)%~XLVZv(QXrF_X<lfMl%9rd>Y&c2SU*(Mm*=}I#&_(Hce$ebj00S@msAr8Z+V)2
zD}7<)ei?8g?LNyxQ+{c)Z#o~qAD}tKE6Oc<3MxwDON0>ttaH5?a$;xFG7l&5Fl98}
zt7JV<(_fbK6hKpQ?Avkiclpxv8qaxc_t2#+Sfv8{n1+2|z8oduk6zXH8HWkdyf*D#
z5m@0W1E8K957RQN-h8?hJkt5m*H`gTu;c4asF<`*I*$}V8%Cp$lI)En@mE@0de^u<
zSLs&{%Uo0oG!sNWH@lIVd?OZc1pGm=T;(o{JLqUF8f4=86x4{z3x56+0O!ApK99dr
z1Fc61-Ut@t*js5=F!5agRm$Of$(FK_`|f4x+jK&!vCl_khYTseY)R_kYW@1+_=Tp;
zS4W%^3k`o4)WG%`uzo%R#&ek5ZhowW3UB<$aEs^C@tsfOW};x{f{1;C0Cb{k+=fJw
zsI%~GX7WSXZ-3e@GfOJp7cb`4l(bne))xh36>~saBdBV=N&4K9>M!5vsazcio86^*
zH8IBXz(A`@2HKO=MXGx^CF{+8i-Py&ShPX^Ls<qfiL@yD-@^%wM*i5xmT~8%=57&l
zqUq0bys|1pp5^Iq+1I7m8*6odOg`(YBEyOfgdN407#|3<!Rv*296mtq<!yh3uyDTN
zi;VF)U_c)$7gyGMym&8<ydBhR<dv=d4!ryOgjrS=xW2|^xA$m2%=KYV+c{lj{ra4(
z3Hk-|6wbtGLYDNHG(Kg4jiY62wKAf8ZzTVvop#*VBfWH->jwuJ1QR#C9=}&0jeR44
zU4gt|vQ;NgW~p;L2zgK}YStps;GJ#&e#&8fKRf+y#Q?aEDO^eKPUn?-$;FPPvFY|l
zwj+f9Gt6El{%f5<FJ%|6rDUi`<XXW=U(o8x%7jKgs=mGfG3Q!4f_`C9+N=kcFOi9s
z3>e5sjm*jM<v1Jo;xqrhygUEbx9I=l-=Y+u9!sbN>xV=>$fNx#bSX{>k|aZSpp&8P
zQ$+5wajAd1v|+3p6fnp}SE%tlGTc@o>vmM*!<PvDM0{|JZE#bFsM6}EZ$-A*QRnDC
z{Bm7j-lB<R_J#4UTnXWG!T7g2gq>S;_17Ygvr)7nLs91`AW5VT(tCMv+IO)oppg<D
z<TqH#GiB-h3gikdwq@!aZkClZ^r9rC%PTkzl$;r+AlKNs^JRKI1>F2JiArShq-S-}
zDO!I;HIGk~A)FEf%t4Z?-duC-;A@_@ZS_}rS?F;SYFTA3<M&6PpMH?^hm%TxgP>Cj
zsbUN<i(3iqvKV$3ZZ?1Cj}EWy_MB^qsj1@522VA`*{cXc%FmR2M$p6Od8E)UY;+NH
ztm^c%)i+5l?6TBtcpKZGuIJK!(c?l|TFXXX9SwK}IGoqhBXWKagA96($qxzZ9s9VS
z#^AE_lC6428Q2TZaLFN)AXmJgcKkX$<?*GCf3isTgA;SDy?cohK}3FzZx>O0m{)80
zG(1=3Bo@z1q@`{o0;$*6IQC9gh~!P*dkeRfiwZAT_i@=Tg}*}nl9B${046=(u#%$C
zv-VgjYd?z~mv4-o|2gaNotsHU*_Av)Emeh5F=>fCrtPDS>I3R7rp1wt$<};-;_%7X
zvU1tPqaMbMs0&0JrKL#13@J&ENRP+}r<|I5H1Ufvo`&H?28W<6Qkj#6!M8IvR0jqW
zp=#`E?A2CRGsI}^5ryu~$+Qb-xq5%eF1=x_>keog#GBOojrPYaBf|p6{Zk!R1S0wr
z)lj-f`;BQXG!aB8dRWEr+N7jJ@P?`!B#lv$_Bq?-o2{=BRb~Y1UyH)jbK30n2GI{9
zNM#TWuN*10gVr*{<%7vM4RQU=3S<_Dnaf9UwRkO=Z#^c8!pXl4G_%aJNUn3z(eW45
z7cKoa$Fl6B*f)C~mDTo_IE~jo(hyP6Gku+_A(X@htG+R?t{u9i6l~wN7=W02I+ZYf
zf%{VK(Fs@jRy?dLS*RYM??`$DT;sYtepN2sXQ1+YahVnmKQWLiv!AkbEqY}FDG`a~
z7AU147m~fnOX}djv5zS6<zLyeS^e&o&|i|+tLF5jpk9L?l0l<pOnNj51rKsj0}jcl
zKmYHqHUGz7{@ZKt|52C!AJ*lHfGsF0Bj#N>|A+vl4~oQ5Lg*EcMyllHM_~+IaqrBj
z`b>BrMl;&XuD&NV)r4#pg2<`7YxDM0^Tv;AeMb-!%sWG$`gZ0`%GiN#-;eDP`{&qB
z3*lZk84#%>@`efwNtVP#IyIRPZWrt+WxRDi8E-P!!jkoRPc~3CnM`Q$o3%LV>c^3u
zC|kmcz<q{6j#IuH*4;X;mkC8pdQ&dtX2msFR$Y)loN;R_QLleXQWf-`g`7pmPI0En
z*D`TbE1*di<Gf?1|GT231VhC_PL6*4tKFe#%+|BjAFT5-yK`Sh4}~#n)54y39H$|L
zRZc3{DAjHP7`~S^nRiZPYVM&9dtX%IACJAz{+PHzOF6=%)%OZioZ^F(n(Y7VC?q>L
zOnF?pcAeR7z9@Y>l~KDUGxY$zid)q2&N{TVR&-^9p+d)4gg~dOt9R6h?Qbpq`uZt?
zP|ErBU9|vAYiVAt?nMe~LEJmRLsb@|9uirQa=c{7Z~M-tsP+*@Evpx6q!bureig8%
z2$~)eS(5an2Qio3nECQFv?#bk6V7-aw1shit&8OZ(w)2LSe0SUy!V3Sua$r&34x)L
z^$2sw$4dVAvQ`ax!n!42PLW!#?r2YZxVpm4)V*$=+HWMKe~x<dRb6*p;M%P{r!Ou9
zg`cBwu<z4F-X9|DF42b<DLpibUe-;PZXJO{fO1hsklMWz4K>C3hORYs@J|X#^Nf^~
z-$-0X^RXQt9^b2>Cqv9cCI*se<v0-y64xXLzcR6+T24i6^XssBM0HUSQ(9U&w+7db
zV|F$}_j`+CaV$cJ>$%#Z1D6U|T6%*k)?k@&Bh<&o`#!^gqbw<#(yfS6vpSpX@o6{w
z(4t$xxHQw`WW7ur3T{<<BYg-}5+$h9L+wYpUOiwIuzV-94DS~#+If3X>jLy3cRb6a
z-%py!zt!1{j%_f@*ETbUE@ikdl-~7nV{Wljx}nuT&zidFc7iBE20jD=P4LNr4Y;wS
zQKV6+0~~sk=l571W(@aj{hUIaXu0$*Or?Uu4xcv|prOKUTz6l=S;~`6z8i5-l9mDT
z5_M<wim)F3<*>tLEjaE6+OfDD@?oQBJrb??0cn(-*g~Zgv=Y6y*_!Q&_ivsn&1kiJ
z|J)w_9J7VWkfj<@H#SR48TbQUTV@Z*ZF_k4A>6#}9&W*sT}oyQQERo5aQosM5}}hp
zY|E-CA$53X<ygAsTn+4+x7w@)T}Ha{h7q`Mvq7qLXyxGsS1;AMkGjLxXKUPweH6pg
z<xa7zO>j>w!ewZR*=3GkeJ^t{oM~pfUo(YMMOq+#KXw%iS{FykKl{|?*FF`T?weh-
z1~+?G={M)duVC$-HWqs?0tIuVV#EJNk{bh*6sm{V6)>W$*!lAzyku$VAJp3vHKTsk
zAx7~KMkMDV@=^=!Y~VqKg7H{zU#~;r57md5##GrHI69L2lODKLHP}RqrTMiEhwuzx
zT^;{m!IHY>k{OWZauVBnx?9@8@(bBzUi;E!5Xv!g=m!T}<-PrHKE|WgtYwP#fa_Gq
z_!Yr{Egg-qXzPtUj$YhxCn|chOlZB;ny*RE{Eij|zJsbnL*ema4#KB-X5OdK!8w_A
z6Tzs;f=nH~e*nQrqNd8LU?eC5ZYn0P#SQrsWHJldpTI|(*Y#vyPb@rGR64w^F9T;x
zZ%tvW6GIl7;JBWvGLGrOFAGpWBF<C*J-iwwa>}2Pw^_0#^XP1|yja}S$QM>u_H|Ht
z3qy(#N&_Z?2flnEbPFTIP>n9anwf6xgos2r-UK}g5VzhF^&pxKSOLe_Do<5V=NgJQ
z`$kTCN&lWSfrBy{tEWN)SQv!pjJ~Ev+=W5GLV)B$VUGrsXI(KauaU|bZUiz;0hmz;
zeKQ2xdH?sd>i(<d;{U-**?q=U&x@I}-cE32l)}%aHH<@Q(p6Ccad9D(?3}rq29`3E
zsK*cr&<I<n?=7_{z8_`0zk#6-Uwj4mTVuu^V<D`T*N@et&Caxx#8;uu-}%K00XbCG
zQ)Bjn<~_L4qV<0B8F@`3VD&u4ns?2Clzp-VWn{eUpJl~{zREVX!O4)zWUaG0WbEwn
zH+UjQNx3PH0P+nlMm(-V;2v<p)P`YE7ig(T{IQ7~=%y?rmK~PEK-hEdD)XHSp$%>k
z+whrqL}+c{xr+na3uLPYBx=Z-P;ODOa8x;TFqJv9eYat9D!E#F8MD2ORTpnA{=!Cb
z$}CWLOf&~Qsq;uom^ei(RR;Mig1t;k7<4#cOd)$WYo+#ksM~<m{fcgJFh1}9NMnfp
zpVJt2v*5pi?TV!r@TPOp8=Z*|G@d$6GvsxHzv|!_Gz>0#$AA9LgV09fr(*I!5k<&F
z=I9hMVwRgd8?->}CdX`dy%g2rgr{`E&rzKhl@llDf`TfGcKWZ$RF0^;O!HrJbnass
zr*$S>2I?2G{T4hV9)3+s(~2w%92M6wt~#tsCH$)Bs^On?-u69|CLC=v&!q_DWQPfp
zNQUK5T`v3s41EiI=~(95kO_V|(%DP2bWJd^FupTMbXA%xUex&se(3sDVjAvpqXQ9y
zMUSWZ%d=u9M8^1q=;JC#`aytoUrtZ)zl4HHTrsu;JJtj)#z^tK?)d=osT58TpkN0Z
zk}$8*dcIj#5bvj&=Mi=VxiS^tgfbu|S~t3gEhxvgNerxE{-?UzJMr;XTE_-z0ieXE
ziJ8sjMtUL$MkFhucrb+m%8<-(03^$g**IxWe!+ZzNHv)fqAH%4mgoBkhm=fUw6xW!
zkjf;H4h}TUgM?#ompJs8EB)3SCILMnDq*ddPd#j6oZ1-5#I{!RIZ*zqgxO*tc1zd;
zF90zc5(F`r(ko|ZuBfT$tlm`8-60ZH#t>37MRj_`g!Pi>wjKNUiJ|1w2r3;mx*Vh~
zS9|FT=2;CVOxc8wOKSo-u(1WGj;HoL?9BmxJqBFq6l^S$q&rt7?g?Y+qh3qI+NJ5N
zi=<IZ^a_Dj0f1*TkwoC-H*g(wptu<Y@C_?RQ#iM`sP_QsgJLQr(zO3HZZ-Jm*ZxO7
e4<h-|^Ib(T8U@NF<8dCMWK*hLFSYwO`#%98V@~-1

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/CW4Ntdtp7lg-110.jpg b/assets/images/impact/talks/CW4Ntdtp7lg-110.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..00256c745d11d6c75b6d556041e1388e9b6ef6b4
GIT binary patch
literal 56320
zcmc$_2UJsCw>BCCq=P_^8juo!(3?_~4hcOZ^dizhdIwQN={<A=q$Uu0?+7AQIsp`<
zcMy=Opy=iGJ>R+S_m6YWz5jp5{ckeH9%E(gx#lzHeCFPB?=|!L<L_m_Eto1)6@Z5a
z0N`DJ0l((~N`RXL1cU?xHwg&|Z{EB~L_$eILQG6TLqS1GNk?=0HXRKRNYBL1LeId)
z2n2$-L2MkHU@-VL3okzp7auzpnCq`a@UBHkh)JkPNT|6OfDBy!)1Tj60BYhJ4ESC6
zcpQKm)Oh&Rc)#BRK!1TG`U~!VeuxNf;NQd}ASSt<HoXPF!^g)Xz$YOiBECs_10N6X
z1^}OckoqP&jR+B~f}T~9#|t2bD6u|@uEw)g3_hWFFTdNXj#J#n`!W6G?PLkWJFrry
z^@{2%1}<frmj)>XNR_+Xl1Ho8i!)uTy<YI2y76uh;1k}wo|C1%cHgyP;%mQMtN-g9
zyc^W`1T^eG5j|Q3D?)hw#N*pZqI4WLJ@j8Fu6*t0tPvAGQUXha{+<JnUkhBHO$~Sm
zh+Vdt*bol#@Aw#4c7=LABD9{qyd!F3Jb0R5>2n$2dFK21SoqPNP{*N&9_6z@Az$X9
zJLgA1)j#Rh<j#+Bf2{R>6L)_|yV$8m)flr7ejFKGZ}H_Jq5oKLyM5<pOOLLfI)N9;
z59Wk?S=$TPT<F`Vx}S+HxiKC#(SQEGl-r6qkYtNGbmMtB-+RzAo1(B^@XhKwUtiGv
zn%p;*e=f>A#C|^Uf1=41YhHfF-SM=5%8kB_V&m2ihmMcWNUvdD<LUoz6#hHVR{y_^
z`2PRYh}?fS{%^VeFhYkXVqWE-C*Y3{*g}8WcP@%8xeEQkDf`DG|H4W3H_qnKwUF<*
zr+<6$eu~l`+lBtve!eCXkG~o;^|#TBf3!v{zrA*5?6>*Y14*i=!~aHI`{lJ=55C;j
zp=yj+h*%!`FT`g5t2rUwm-9*oJxl-1TmXXgI>Kx(pZ;y#brc?Ox3{CIT&S+2pd0j8
z6e#{_-SKgT+%*)r08Q38wrja-uKeYdf0q*uatQt(<c|K9yPwj%pMT8=nSXkv|Jp0#
zKdS%o%3tOk{>wZy{3E%wzo-!W(d{}G*Oq+syLQ*LXrCVo)%$M=*Dk(Z>@RjwU9<BV
z$6xGZ{KL)%e-``0>IYvQ68#6O?f+u6|6i<TyH1BcT7QE6P6v;i{{LHbf7tx5WcW9A
ztpCO%qW|FE-|9}k{<uz3*Ic%1X8#W;Zxm2nCl<w2_aCY}{7<J4tp@dAUjD^C@2`Ji
zK=q$7c=0{=>(26Dsk0#acj*3un12$eUD{vnI*sq^5MAeV#PV;zzkK;uR#N8v&uJU`
z+m)a{dYJ#l^Y!^ZBM|XNPx#+>ezN`3lPiB4)%pj|!+-b4-$uFqfkV~(>z{G^|3MiD
z3coFy3RY_1qF-KOJbbGjSVg{Lx~TnplIM!l{K(07_fuW2i;$uA^h9i8B0?L$$X#>t
zUOgH*(68=?k|6^85SC#B!H}>7VLD}<0!d8w=L0zpxmfeEVA+nhFDKJl<T%r9PB(69
zEyt3-vX(nO<XR{6jcv~GiQ!!k2tDH(hz)@b&WDKA;8F<zMTrXYX^T0Z(@P(NRqRVo
z*P1V4tvr=z&hzrh82hyQ8>UKbynwt?i`Jf6CaM|9#+C(s<_0?|4BuJdUg`J1wQxj+
zG!%2~GYVGUkXL8G5AiKvp3>};WXJFQ4Y=I=wz$UUHOP|30VW)5C-w7PbEBVXWu46a
zLRm_s!l=t=h_@nxANBemtU$)rN?nX6Uy4|}mwF}$WDDkG=<*wX!E{5-(9+FkAFU5;
z$y_jd^fF#>i6S;CM*NUHs0ZXah2QG=Uc*`al{J2WUN;DO@STl6lE+xbI^!0QnDfrr
z11L?PEC?HWkUS?HqfsH|XiW9wv>40@(Zz2ylfH|5UaNCa{1f$)nR`@djV<iQP}Mo~
zH$eO?%JQYnMek4RukB2oeI4{?ymt0c+vLMt!^8i34^)$^Wp91l-n8po0^i=h=1S$&
zbmy}pwQcyR<J$wJdsV+a7b;DnfWoRu^AEMwThN7Muuz1mDjTupQAlDDKgDZRve&tk
zX8vhWwG$mqTs*waM$AKUY9ZceUU&f~oF<HApOJkrh(#@Ibkf(H2SaRRl2%+!enW=}
z-~+|&?Rmo_#Zrz<_nuBb#|cT`MYcAF)SKD4vy<k0Cc3ENMA`)}O3w}k=Dhf(Qk^%q
ze(A|An*$wrU)cadp)j3Kp!iZ>DD^?o_xrpSQi5V)*q#Q+v#Hjd<Z%Jw)U-#Vqs}!f
zn67<wG5)B&URA1CdQZ;kR-GJp{wKC~&zHw{K=HivEljDg%hJ>UqCg$FU`)tYnO8T2
zeIslg&3GV+-7r-@WK+9Yzf!TeXg2cK;KR-IV~wv*-blWUeMW!C)DtZ=c&>2){w}SP
zZQ1vb?4%($u2pCkK0HP{&Y1ZkGvj~nfvrf+@Y(LtL;u24-e+#lvd;AK%<YmpOUYn<
zuIkLsMbqbXMCcZc&k)0F^&BU~&cxbB%bU5!8vE4WTfsH35*&HS%16t>nTg30>8wi{
z+iBWua>~8xO^ajbzTTo+0R4W_qQLd+9s?x)eNB|t!|Pv?4~&dv_oog_D+-I{w1UYT
zPZCopt9KFuI*G)JYIz-aB}j&zC;tY}`|q4lC-CSr{|3bP;JlkhNiS*WA4^U;ftQ$v
zpoCKMA!>%JTYR-x31DWp+OZ{At7*_S-gfu~-KKzPyA0Ly6+aAEw!qA(7vYwn`?8Ko
zpDW||4U~(=KPM<Lu0#CwTa7#emcCqE<|HtyKA`8$!8#kWmWTF*NtRv3cv`W7BPn!%
zlO4wvSPm;n1S3gq4LqH8{Q_&FazA^Ju=^!<fpxug)JfVfR$}b2_WqEsb;HPJZAQdC
z59AApP^v)y&}*`=eIhM4)gEL{;`QJ+AfV{H<dD9`HYJ-x?|aLh<BmSIEWiKOM-8+n
zF!e(hpTk#*7sDf=1u9won~}@UYLp(hdA)l^gjCHfji(_<JSVr`b#0B@_r?%<ee7Sr
z#+4AW804i3tAu0G#>rFp9N>dTG-A~(Uk}=%rX+Zc3}LC%JrM3=sXd;kCBYm6XdEuJ
zM_d9B1irq&uyK!8t<fJs-hSf#Ir$6vj3Rl9{wU$yk<$;R`iHpU%-p+2>vL{tt!;+m
zkd?<Zh}lWB5_OtrHem<}vyQi{?UpoW_FP^1OzPYd&+&pZXsQ(L6bW6YJdvya6RPk6
z&tA5#$}WD#nkGM{pNnJ<{02<=T%88H8kYoEm3tRA81gNIW5eV4VC1#HWuC9hWM{!|
z%@&;k7$j1nG2cGU(<9qC-Ur{|qPkyaPqf86E75Rx&vAKDf~2tQ_|w|<3Z|I4LW79Y
z)t7R1ZToC)_yQ7z&WN^~>I??m&9b7%?B5ka%pG#YEXg62pr(+gAE_CQ7QYZAy}9My
zmXTfC={a`@>fC;%1ya?L(3IlIwr2oYncefVERJ~Z{i$sCXqFbDB}NX-cCWdt{9Ih8
z;)TnK4)2{T;>$Pjiw#%?7RVDdAN{iSr1_S|yK}%%OO}Im)T6;ZhC(Atmg%^ww;~1W
zzvrfv=O7H}H-71G+Ua(&tH^Xi`dwKS3dBJt8HTp={JJXq6YuSk-~+c_aY_z8ppDqO
zlKMc*-(G2vger(H@PX?PrF(M%20Oxhzbx+Wy6dN#HR1v-v%UpC#59<Rr}|zJK*7f7
z9nKuGg%cRbE*BeZzfSB#n&_134*EQE<KF&yV{e5w6Ye%ye0uw|ab>kPU46{wRK@`!
zMgr>hw-h|T-QJZl_M6Gx?AVFPEzoRYSd-8<qo*9C+@-v)@To@LKYMq6w=|%<!;BXq
zq&qbZFBR!USjkL7cu8eSN4vygw_>JPxnF{SXVbl#96`Z!A9LT6+UK6=r`&fnz)dk>
zGs;RJ$L!cBFsHu<?8UN2Q|oaB(e6mVbQ*(@AzzV!KQ?Jnfu?{jNnI9Lk;lc#Sw+=g
z!H&TI?X%YGg0wO_60NcW8ZXg{PWI*TwO1ONk+vkE%8CY=<8aEa^L37Q;@4iJ*7wCX
z7Nq<0_dr$`K|xr~hc6r7n7-5#W59Mqp{=t%NY+r@*>w)Ezw*!PRMKZlVj2T$a`BCv
zPGui4$Ze_dNs*Zkk35vvp?BE-T9B`+n)=Rg^3e@#uQ~S}!$`*+G!|Wf8^N?{Wh!D&
z4z8A#Em59Xp*;j}C_ylPS(f}O)$C=PZyLFtvnh0+dvN!Iq@9x#4t3@D%VNtQOFSg~
zk#A^6FKj<lC}A#AOTL&`8|hEtOnD3F=x(MLDbM#ikwRCKXLd?w*9x*q3rgl9JAceA
zh&|?2;2TM6?e;IL`^+w4Ffs5rP1vtGe}To4^o3dxsjt05&uJBJfQRYB`wevGCjsFr
z`gRqltUM^?T*Dweng}oU>w#S0*zq*2o=*8n1sW3{oeJEH5&_tN?MDZ`FXZTm`}awX
zsODCBQNOJ{Imn5dm5+&kHGMlqx6O=S;<WMjR=z%$zXIv0c2EnY@>TvYg^<UL+3QB5
zeUD-vr0)DA=_pOe4H7-M5b|b?5;=t1y?OrWnGYLseOMqPg^;4&q_rclf%f9?k<XO>
z45W&2lQ*X{@|SsOw%5lOLflh7Yi6XebFXufb*5+aN}Xc0ZXfKx{o^pDYXm^lb+Kec
zRC;+b-RCoTj6_ne6-5UQ(k3M()yrA%mX~R~tSI2HJ7PoVX>E=@crNn!uM5n5s$XoA
z{$Vq-rZ&wuQ&;g1=yJ5hdUe&LVjEfX(planJ0nSh^>?we)_QwB*X#@PHG^r3(h^9z
zaSMc28-}Rp^1Sy2)IY{rrWeh9x0V+jpOJ}jjQXMLlWUmo4a}N1<An8_MI%c_rG4&{
zJS(T|X^W7>Wxsx-eR6lKz|b$+_VariBypr>P_I(B@1>Bf$*}k=!)C<^{V&*@?#i3|
z2&@1+wN=7}S1vUB%9YK_IeGH9=3bA$D&)kXI3z_?SC=-fO?y%30UO=qMy1GF<%!(X
zm--3e2+b=ZdaQ(54ZAgR@?wYdD&s8nyxlF6wMv4eeX?$8<88MQ2TJl1*rkhX_jsR%
z>8(#SobC-ebX~H3totI{cw@G#_r&zmL{LRl=!i>-n_>$J(V~AE^6^GcQCsVe>Ar78
z{HP_DbV$$w0q+iYTfdEi`t%3|B(fmuumGbm+(?POS)r{T$Ew5mWRjQ<g37oPZcT(G
zdwahsdyadgCfHVmm&j_&!qnuLW-Gs5>t0@Ei<-cbsUEz%XxS7-u@@aPPqm$nDCDH1
zOUq6yX71T%-4=^at3qZSIYqbW?FN!THt^3N+TT=l2uhV-<SzTCi%YRR-6-ZaWQ16?
zm~0p&4CO<iacUuO9r80)N}eM*i2!M=^U^Qh<GWFD+wQ|-o`qt<Vq*CJ%ZG=Ez`Wow
z99swVCBa+1w}{LC9JT3<XuffhyLV46JUBBuy1+pDw#c$K6xKYBY^RiA^Rkv<+H-*!
zHiY<r6AFxV0w&VD4aviKZ?vlnD-hVQgl-tS_h+<kRZh+(W4=`DGp@jpdPsO^eIG)U
zCM8tVBNV#tY%acrZE295@WfIk$qb%GtbTVtdVus2EHu3^rx+sUtSVNtrI9C&_a9ss
z-Qo6r`QDycX>_xgCSb73@ZBM8U8|Xw6*d~3RTyZ=Um7Z*UiYfVrVSy2l~#Y3+*+6l
zykf~-I|^!s4{|x;KVH){f_2bQ<IE9f#3c%xym0N?VII&gLYUJ7+c#~2=SMs|+`OCw
ztc0JIupzF+8d-wO#TK4w?q~JWx>L8fEJ!<mFHlbbay_y>#7l9egQJwgw}Oa0sPA<|
zy&kD7rgN`Qk-Gzn09K7~zBgZ3KU!q?T8m1J<z-i@j%~AmYU_OY`XgjMx!oK=eF?|N
zg{;516%0&7Ao-pW>i6P`1MRz{+qv5h;ig?OLyEd@3T}#SG5{Vu!JEXc-d8GS*!05{
z@K}cmM^3dSegF1UD6P_b=Y|@=Pr(JPcZJ1i(nbZku?MN@wW1!`<dhW&A!dmo`G{17
zJB$kk3x_fBf=Md+9gA6|;n>0A96FJUp~T2pw)!vq-cPAoj`bB=Y>%<84$2tfWq*D=
z_@=SvWp*64y`P(sg-*O1p06Qw1wVgYYZ<cs=V|P1Uc*u~;+O4nuD5UCKUaT=d^gp<
zUl)%O@ZldmxkoNpCw$iN<~Lxs`D##TNdMKZG7r7M;jh?i?}>3u?4xJ7UC+O7%&LrK
znOMMo1N123Z8g@(3XeOp3D!}5O&$v?>Ezxsh0TsK8=&23;pp0=plvQ2)&i7tH#6B+
z?>Uw>>1ard%kcd6j)cm}j)j|<l(|l2VuksG;^-qCAS`A8mrtYL3uf3GW8LcuI2_TE
zD?7caMSarop_i8azH3ca4~w}&(WXP=`-H_(?aEF(bpzNS9itdsmMTdo%B9AkTeY4_
z{vsDcC50Jns}ia7hWBU-)En#sIvJrVHcnp)H}I8z`FOMI*uk5zpxq-Gq*pU9q4G9G
z*)nh%Nj&Sj$t@W08tt&%ur6^f+7JbgZe9<Q<6+kH#0#V72)@20%<{}?eE1D`?!8&n
z`uRz{@Lhx+0(V-&AS!IliEroiGTn~KSU{}w3+`0<8<hFg#G}-{ltHHS9)GYAFL$$r
z%<z&C!<lv_lfIU*R>61?TI&A5BEnIM7^Yvt`>A2-b}}+STorORW3GMiv15~;%OF;u
zgHFneEBeeX)z+EsTVvZOS8{K#E6Zh(dW%o-hK5O6RnX08mN(fN8yok_S)}PTb0h-m
zKDZaeX$T9_ar=YqA&RJqS2npnjHH$@;#;Bzm*@y>M2~`{5=hixg-0RIkrVQ%jtHbv
zVIq4vieb_7uX?I|Sa7evh5u0V`Beq(O&8y(hb;FDGu(owrB{beVyKcpPYG5+3oK?Q
zRO4FqO&6>`sa<g0qMNc1oHA%Nne!rbd=*8*b&zM1%VxXUJ~5jz_Zwg?1cy(IKuHT<
z++akIzBTN3_|$Mdxm~M@0R`|b08KBF4Yu;G9BHD&=}O@tYMs?Kx$lzSbe~`L;>>tj
z)P6;n)hODc<9!s}^>50k;Uy-nP)P>txNKUaqU+!$knMB|7`Lbv2Z9);-Zw@z=Itje
z6^1z_C(?HyZChyNDU4={T!*F^MV`Rc=G#RQWV+M0+ykmt`jLp<ol3hY5&E}kk7uJ=
zeRk^%J+T`LYUxnf#D0tVNi19^vlS)JT^1xXME~}%Ea6E1Y3)_%P_TyITO&ckpoI*_
zOJ!Wy!1gqoDhCF36IoBm?VlX-Q^z}#jvn@@P)a$Krh>bk4#k$gg48k@KGQT!+g{pZ
zPOb7*nzAT2^mY=ousE%NfBTu_4AWI3wu62;XlgsWJ3j4x+@!p1N@%4iaX+x<#?+l|
zQj5!kwPsnjQQ4eLlK`ty`xYGyIdia<XkHpa3Jt=ORD0ny@lUha*V&uD+z)+WKWxET
zoqRfboXQ_a@0x$D*w1+%a<ThUOwb&Oc>lZ~iMH?Lcs}Qd<v+DcoD!d7x*DcXkk-^i
zBBS#PgbAQA!A8FUj<7>X%iRv&w5!Uq)HzShhMK$i%D({+Tl$`=4m*c4*O=yy+vN@Q
zsXpV8Na+nIOo8+Hqxt$9vTnrd81d~EW6xIP%)Ap{vgr+Wn3V}!7hd4$weqz3`!lrn
z1qS>u_c3q690^wFw*94^6Ki9WG9eEa4!Le`^&3zZkKsIPL@8A6aFXdNG!&(*FJFAv
z^$)$smtecMD{)nuSXfksSk1C^5h=xkC~(%>JCC#!j|B@2YjGMsVKPB8XAi)v%2N6Y
zNk~U~sbpAu{VXyjH~UMv(|FodL})+w{Yvt$e|KHXtQk6=J>Mz8a$!;tK)utXl41o2
zF~JQP)qcpCUhAxxx6cgIP$`yL%Yv3(y|$vO=*Bf41VPTD!E+7v{C!E7x>*HXc61CT
z$${s};z0Z6LR(b~D*H>LXXluEUS)&947RdZtk_-P4i~{1k&uK1Tb6e~$;Y2l$hCAn
zjMKe*vZ`gW{Q<lX$mWIZW&d<#zIb~U9hvs!<I*x?ok_qJ)!=<d(X1JEcN~>6Cb~f)
zD-VzAUtQf2XgO2L7bnlM2m))Fl_3>dy){!XXW93taaVuUWMf6K-YQ$3^;*Sg&u3o4
z@V`+%4YK+bGZ1V5(KAJ#fixYqrz93U34(7G-5^^1nHfP7ATcU+VT?2;$3dzZ4N@wN
z5-X<j9T`ZWGLld3bedEVp>qVh9oJ5@T8762u1e7zKc*@bClHu73<#Q{sjRho@aKj#
zN~2%q<o&Lk44N-8Rhx130{1>fBSIC3f}aW>ll}&nI6r?|FJaJeUqFB%WzeE-B5GH4
zheTV=z+GEM|J|AW{qjpp){K_5L|Wds=>(H)Q7RN(!~>B6nW?XRyjbZktwO3k&8VQ(
z?3a9VygqC)!BaK_Vq@bMPt|UpnNIFCXeONiPL^DVtA)C2f1<Cm_XAapW%td_Qd&DX
z9!g+&5;k{AyN4b{f}%i+wSgbv&z?2R!B?om>!6<@)?74Eu^QTH_%CaET%Vj6YC6yH
znadSwVrLv7np3na&>g*4*NZo)ec?IEPWJbniUlmtwob6sAbKlkhULsJTnsJrD}4lz
ziUKJu=0N@7yiAz?Q284LIrAbvR@yOb7ru<9H5Db%WUD-5O%fuIbTb70fU<kV<DQU7
z1vdq$z|Lek9xU>`;;qM@7w{SS$^BT_hM)O5Tv-+Fl{a>9l=;}IZmH=$4wK0+;&xK?
zzai_|T#7@M@6}no_KY`(+&M%YIh!*FBnYPDZ-lV~h8aqJAp}mm3G$?i>u#ldFt*Xt
zBc3kNVqJ7vp;bA?T9P{rDIyh{th6bHX_NXj<9+tO`{FG2p4Pp%`(^wEKir~4$ejsg
zHbuJ0V2wPKg}rCH3C0jkv|aUfg=h>vk_1l2@eUBn^C%@>=6+l8W&*{S=5%x`BqOWJ
zAIl*yB~#_oxc2(tnTf@gfS{<VFs(-pl&CpT=|sh%vUAqyVKAf&Tjr#1l9d>ByBMqi
zJ5FlrDZ{7a^o+<&TQuEaDg)UT)YckfxiU1!qe$?e_{>i!Y2b!9(*qGRUM>-Yf^aT4
zo>$jE;aOt>)Lgp!6CcKVZt7<W)=wq3{H=4R@yUZ8C9O)%W@Eaylf9rPo(f@Wq&!9p
zx^qW@KAySgPX&a(#>!lqG~}2#AE+7YP2N#dCxH`xlpk1?6b-f!Pb+QEpK`8XamKPU
zF2?bqL_2E}%Yj=pt3Ri=4$o$~a>T6ae@r`^5WH|AbcO{30r$Y1`1T$bt-*K8Se>lt
zXf>4=-9*Ia+Zy*Y-}jxe+1a1Da{ezD{eK=99{!`&f6MpJQJems%D+x*cX+;uzW``+
zv<s?StJ9}~#U`k`+w?>^^Xb9kq24w2_IlmVJdeN3ER|IGULfpS)UyJ1;3!1xE80$!
zWI$>65j~T<``g#4K5Du~B|Dsma7nvVD2%kR0vh|2l-d9M_R_};tm6`89A4pGi<N=v
zH&VhL=ZY}ki3V6c$=eAsV2m#rn~h(elFiEK;lo9pR4=KSBU2&tRe%nd9+j5_CR6b-
z3M17PR#R)gXI<vAZldixlOgQ5pcs-#O42lHOGC1z!;n((rd<9I<95rucFM7RnpSQU
zF)eWtb+z#tD_(9Tf9t!`m+eeNBQ9ahq>y$S))kP|QhWo^<8J8wZZL_7;dTida}iQx
zkj^@~NfVrM?lZL!w1ct@^yG`WjPRHXk5}+2mFJyjah3@@PY99vu#pu4mCEHshav%_
ziM~9Pj6~M;Hklotw}2`Y6n8_eiW6@6PEQ6X<`rmK>4n&&$s=fV2wu$3955V+xaX!>
z{W>Bf`YHg8R!2?=kY=f1RSsMjxyyP(W*dJjh-iINHugl7dRDlPhD5D(-wA|(V4!}_
zz*p~ynUNni&KCGjR3yZsLX~Z0dSyOAS7p8nWz1jdXJtj%|ANYt9unP-S`yu(Mz}T%
z@xE%H3*qNKo&UtFI{pANSr1kLLkWu3ZQKP|9HQ{3i+R<TX@o)A5bTM+H~O_oao#}>
zNv1M>i1b3WsQgO0Xgt^Zkfb%a%BajbP{ma!=7jG8{PQyX&l}ZWp)<d_^)FI?qJ9Ny
zoWy+ZUkR1~#a?Pp|9L|A_Z=Kc_U+N9&gRr^@t;(T36bQc9O1QczhZsC#srjW5wRC`
z-3byhQoK>|r|*Y7mCdcjI^4?lkLR8C9u*d3JXSAR+rCu3qT}OA`s}`w*#~;SFH#qs
z5qWUeE+BUJ#~nMnn|6=>KROWYzClp6T)zBP@K8Q|!~UNmr-0JreYLLPu`<cwVg3)W
zOv8?x#l^iA7V7?PPPyWl3+Kx7d83iwck#(5Ctx-b=ZX%*X}+<E<XQYI{#j!Y7QIO^
zar?Wp$t5h%ztk^;S2GHlc(gavKeN{EtXu+HgBygA7el2;L3wA*oHhwMW(^CQXC>`9
z-l%CDFG&X7X&a^xrl^19|EOX0#}qBHNrdq%zAe52sOJ=vqlGks7>6^GrZNpk;`_;{
ziq+YJ5>TkEfKUd=;Vv$*AV6_v4{a75MO%bQE#h&}U74-pWbhJrSXPy0uIKS~d}qL}
zvIL&Qkd^A70uhVD;kC-$O3MBUn%wJzmLKB0=CEb@a*dT`AMW2xFG4~FxqTppIjy3;
zl#;u1XMNI#7Y?5%Q0;omQhaeoqudwgmYAlTqzE-9m0c#nyN$UpnPOgIECwZ^>NU*)
z&!b$eNm>KxH13y{2R?n_(>6X{61SYW2}7tkuxNC-UcUP#x?KrO+Y4wOSDOF_9pmd$
z%5JC6kpN!e@^feLcb%MlBzCbYUM3qQC`D{o*`326Vd;>S_&he=Zd7K2hpl&#Ho79)
z$9&IJyP$D#ZAZWmW`cW>Jgx~;dxf943F3dFW&UlGf>YE%A}+q>4j5V)Mhf=@P4jub
zoD;UhsP~{@_wJ5f=G@me7#?|7%ooM|&Uoj`gD1;D_C3hJ{KDqRd2~v2R$=7$Ep--;
zcGdS=1ms6|R=N)CpFxn&1`CIu8aUigjKfLtT<?sf5B~y(A9Rk5=2j_PYZB_mN{C|a
zG%L8+1VwsyaQSAqYCZjA)AG+w+9ieg)zaU2_VmNQ@+_X|zs$4$9ebu!(-xTD0IxHm
zL*-{LgodjAITXa5=yT5{JQ?#3Hl7Q@+t1{;%xr3F%o+ctQJko<k+KaXO>`q3N8Csw
zG6rlBSs$lB%;)y>41IE8yS=Q5AEcv)>fZ0~{lq`F9a<t>T+(;NYM;{SK}Y~=t@$t4
z=6+fkYL|h|O$e*dS@FdP6fmIM2rM=52}iaQJ-c}N^a1v9)$qtEx~+4&!PV4>0H<PX
zPOzSl@LtyX0Yy&!5WnG?OPcXkL0MKQbfHgP8y217mtdf=85G13iLSr~bjaz1uZ(?j
zy-Fs{(FiEldw(>chKr<auVPBQz)#nj!lCRdt5V(+!>eVwLe9_912S`+SHJGx<7p6A
zyL!Ic$oox>Df<0)hz(onG?3>7a}ow$Y7M31C}%q}DnKD_RS?wyeE%G*tL+HcaSA0J
zkyF-p#YBf10Ob#sAIh^jJ4`tvhZpmeA0<A5tCx_jwM|?a|CFg3s~kS99<8r5LN#!|
z;B<Ie`*?nF3_GbB5Y>UE7A%`#LKjr1*X^VXYEsflm@6nexRt0DNwdltHk%?n6J;kJ
zZ5sAK7M%neRC|niHc6#-M8EauhCJXAB|x%wAntyQ`3JcA%SoihdNs{MT>lRBit&Su
zuOFhbc-ttepb3<6B{|!<D<kJWC>~Vhz5_?`RaF#b&w{?_K^bE_C0MMv$+}nq&70@5
zV@#ae45Bb$$ajvr5Rz@jJitJ;9N6U}jZBa)5qafYXn1YoNkF0bjf&$1-r0>k60!vy
zd6??;d$hg<+*Z8C3UKN5F!qSPZoEJ0Q(U_Y<JUQ%nGPjwY@5ovtX4oRs+o(FL@uKG
zyQRNkYByJF%di&it~f>}kNl|TlOF+}oYB|TUa36;S#2SbPnnM!mKM$)W?BvkZ4JNu
zvrT=<+GI<pvuSv&adGt5HX&QsInGydM+Q+RHvIAb%{$irMcIeQz&BU_dNm7q`;XDz
z1=<6mMiRM%py-D--`*IKNx9x<($FGQ34XebDMrVB;4hPe(MY1cKHpkXsIa~Dxs02a
zQ4U0o!S74l*=i`%HLvGNG<F<5DzF(br4gGH1g@`eR*`md#^3DUY?GZ(YY%fR_RKC;
znlSo8)UX5+LCo#9Y6(bTPPCHhGR?jemrwP*gko1Vd|)CYx{m2gc`?j>&>w2yY81uY
zG3M0r$5DE%bv48f9|hYETP$LQd&T^kUaF{qj_x}~kq{(@NG9XXSN(V{kgO%dxceXd
zDxe0Ayoh_qMBjn8>-b&41){zljhacy7Up%E7&;kv?`gWA#O*poC!W#?sJy?y^ahXR
zd-ap_m5(x&VACthCnE0&M|XTGl36{X9_KI~5QILbr>E9nqM^A-a}&S}pv)(DuCiRE
zhtwexthxu?ayXFxu~cSW=?&4io8P88mLMU)x3k9G))hhOV-Yp|^Q|s&l|V;To7R|d
zDa`y`HJ-u&kKa-)fjXOwj1MB^Eag}q(G47jp`tjG*r6smw-i=E*5Xrxn#RWy_vif4
zX`Rx^%?k0Bd6nGVbC<c6qtA4n!_(XvCD)Ph(1*siq>wiTKfUZP<bEWlX7wrJg=txw
zyokoh0RF=vP?azLgz5a$t~HlQKpPT;!t3tq`Dj_xq8C($I9?4k;#T3mH}EvKLbQZC
zFA1Ly!d<czBxj{?(Q-12oSvM}B9}lC7nGpD2KxDyWf^xiz|g(!9{YZYzJp$$)@O|w
z2ij-h73H%guNK$0*U&Jej@AwRZAG9u*I7vprY9c52g5~^4zUBB=-?#eSndGU_^z8h
z+h<?qyXV4rx<%hN0IOVY)W)7i1T1EI0N)ESnR&(YWGpcM2DCaCOSsG{o7=RS7~VcD
zA<iYy&)8Cz0-U|<CuI@6(nCwoQI^$An(v=*s5^1OqT$trOpyfnM$7rxw|+k7cT6JW
zV|83>=o2!SGg;z&Yj~%`WaHNRoKBl$jLeYE3-Fyc^!g*%{@ZimmbZE|)r^@jze+a8
z(mIjsB*YdZaynUA#f}+_Twrdi!R4b~F%iEwDUQ2?bbepx{!aDxFVAAW_g)cI{yHrE
zIrIzq{K4P-MV%Kv7*9K}jG0@`|7u0ab$h%%Q-w2tX-w0}b%;y6P^H5e_L3MuyH5}b
ztE@u7wFZuD9a`vRb~?lmNum<OR@);dTDzyeGy-4FhO%bqQJy~!8#6`SI9!Jie=TVA
zvrn~{jc@qCs93{`%vvg=M|K1I6iGe&<io_Wltzrh+uj>PYt}7`JNd9}7pEu*4L(5L
zfVwIXElN7OF5%|QCvk16{j;9+v&QS8GGY8FA4*=(>D^TbZBB3C1NGo~l5%T_XjqTM
zV#9AfmS2_sT8Dq1l_JsqR@X?<9BXt4!TOWco+q1G5n7+?8;}wLGe9K)Eq-JY!2n$c
z@FlN|^bv035sf^HcHhllKWDAr=oN|9Ee!#%4FL&zik14nz`$4f!9?FsGW>fAdz7}O
zm^}ynw7p?0M!bnmJyP*5tcZ|+(Cv<k4k&Z=oykaI;OEhPOqmZ!1yD6RfDW>|v4hrp
zUZw;3d<mQHa<SsS-q{{M{|g&)4F0{KDq~P#XU~(7>}s(AM{$X{GMEMy(aPlf*ajg2
zDaaS8EOtCirqP~G%Efe-JfX{ayxry!SUS9|$aR+_86sR1MkOU93F?Rim;Wd?Fj;{5
z7uL;NYc{0TMp`MR&;Sd&Y`##XVM%y)?a<i-gXGb1)>7|K>YSYnruJeTj%{gg!sjL4
z0pG{uK%5;TjSPM&CzAA25?Of;Z@u_{FO?nmtX50rULLHpJ(?+Hl7|rmUeX<SCz-NV
z^)l@`rZ=CgMR`eU-S@1hHWT`)m}<H|*-N1u_6S(1c>s{iUan3Pv$2+92;{;;x%WId
zb`0=&jwwOq7C_Xw;y((0M-`JSVmR+$xSb+@$XbXb%`v!Sd|9ih?HX2R(PkO_h7mLg
zxOZk|rKz~49zFMzpDz$hgD1Jt_2t@&2t8~7HBc$olZX9|56Yb=c&TOHnSYiv#-}eA
zW-?@(u^Fxqrkam4@)G+}XX14D%|r&-L6TNkYtv0i8mgF97$WLR<kT4*WE-qS2v>yg
z+9E`YctF!!%dy{;$jCq8N0DEHU|@WmKv7QqiRcC4f7z?WR(_?~=lv}IYWvMN`-RBm
z6mIO`KSmMx?u?mTRu@w<eCdtO{N9(*AN4hmmipr(swuxJk+%@G+H!Q{E`2>oGgr5N
z?(fN!lx#GA+;^W=;!k<!XctGiC{l9pDh#hHDJTB>uPx1oPuW$_#>E)j90e6J9@0I|
zwuRuacw87q7@;E^QGiekCmsw=n8EBvOj!S5zc*>(6W9cw{}Kh}DNQ~bV$mm6HFzq?
z@HO$uO+$-<P8@+dy31oFA1^sBOp_=KIFcmUJ#y7jEon0|5L-`i2m|$S403lc9Jc%h
z$oXva;T}z;1*T3``fI7!Dw}d?8ncTOB`BAXavgOYv@^xbCfl(JMQ;2CpxlP2?;e$l
zUxgfeJ_L%_(!QYC0C6OE;Ez7Ivp*-%nA<tc1S2crEA3MzBaK#}98V3=iMp^rGr7p=
zcqlhH@J~<ZsY#FCwyVm7l4;^OO2Oc%+*OW)QoGrnZl*e=mE(@ibGyz04bj_)nRf75
z=)D~{I9zG#&Ib33FG8^si(?FjT0a6lZO0*pwImAF<BvR*^-E<2F>!aH>iY}Y>Y%BJ
zxRKE^kILvYCdz22a3JLJqoeJ7&#<0M(K%1iHx^LnC)3*%I16JviQvhZtexRw7OdlA
zo!!~^WnKfxfc%~~!{<er_l7jgD`BLqM!2k@+#Gn&3P&V|B9+(Mv@8yeWAwujTCt(K
zV~+Zt_f6iRI>(S&9Jj5vcef$qG&M}!fMptBauX?()1D<_t;5jfjnXdmhjOn5YC#HS
z@O(zzx9Iz%TCC*B5H@-@haK&j%4>uzXobwv49<i3qfQco;HmCkwtL>m+3lAW&!!-g
zFfQ*t)3Wa)obRw6M+D3zd_d+LX-*{DhKB<2i0<qYe$zD){1z=e^dYzwTZxL|!f7$7
zuKP1`a>Zu5bPamGemtB4+~wxVdkEcDe}em2IyF#M3dAL=rViETOmG#I!bs(&@$1S%
zWD;DTOsR;7%$@m|WcDi~HQiFkns#7JSW}NQ(k#-j_swW(TAEjpvWVn5Ng^2#@nl9Z
zZP319lHT|&-Iu@Wyy|NDg6)U)Y7Zm~%lOA<+;G_Haei;-p`<V{*(eD&Hz~g3>dJ;o
z?U`FcW|{c@DoTn^529IJ1fb(kCpdf}HAX*w`>BNfMU%!lcCf4;apt2{^+|3AE*{sI
z)kFs?DUX--(JUH$E1p<cA4>^nUil!!s~TR14oTsVl$OBwxfjT5U#jo2D>qxBy{`<~
zI}z|+q>7RXzj`!MuojHS=+?7Tqalsg_GtH|JR|~&+e+{IF^{Ix1Q<7tW`-F65TQ>8
zqeSY8n7v|cT(gPZAk#tmPPFl2NLmZW%-nOBZ#D}0_(wq#xgC=BGv#5DethMH|M|+C
zh>kNuCDG!4Ji{*j-3@xlYy!yt`*ti_bTNJ1?%TGNt3IVB!9V<6f_(TL?`4MrGiB$r
zPs+EicA@bpIv^g4L+yfcm%d5Be9hun)W-Rnag*r9xChBPbWcc8Fs<W!YSJ!F8F{(I
zezN^p$h^v9K9^~)Nksp)7U>Z14IY&rU$M_-g-;V?D0(RMQeV5afsN50GG&{Cu#5f;
z;8L4wSe@~#ULF31jctk!=6(o~^7oMlUQK#Tx)?FQJHey7(|G7c9gpEiTX6hkdpO<r
zJ$4{MUkOO(^*&|vjxY<M^#sA0J)^(NnEvO=b5oV9QV&S{XcRS!mVp${Y0=>5hpDA2
zYLoCcKt;`%5M6+;WkRH*rGMjYluaOHLAUSgF27jY7oSGvDnZIOHM&T1zWP(>vGMj4
z#6HPccM(q9Z6LD6N@cb%lY(fN38J95`6hS6E=9+qqA)*>S_>ihE~O}70`p?5$U;PG
zRKQ%T#APmGsPt}&Y00z_L7hI>C{8&!sjJL!{!`6Vz~=O?3jy*5abJd6B}DYVw9b=g
zj#Ap5=lG*K?nhSHCqxLyY@T(^(v!<T0iCDBIqF4C{iGEsMWH#XHv?6U`gH>YhNknq
zQ_o+keTJN7JsAw;LiR|Ie`p>p?jy+*19OuO4k>C2v+&`vzf3b#MmMJI=vXCx(A5(k
zXjILNJP?YMk!%x}-JG8JVIwXwrr6?`e;gQi;draHnX~VP9xo$+;$=;^)?x8mHP5su
zkET!i_nGEmdf%a>hvux<Z-ofE`)6Ade4USwow{Z0CjD&8cJkE309|WBF*h%<o1<qi
zgS!Pa(@*NDkmP*%5*e7gVTy4$Jv~-!C6&%}T4jFuyeN4RY#z`vu>W~uI<3t`4oItu
zHBr}+6L?zhU~Vq!-?t0qyo1+M6aElgL1$==bZby>)pF6Cc8t`6vMbuMo9Y%*<UP5Q
z&TuBzI`(v2dVJVBwxLndIxF=y6Ebg{HQ77gGm#+V>)uK_{(idCp4?>ao{~Nn9d0se
zHIsW3);(%r=)ZYh+&=H>J4o$RT-2&EG2IJQjC_R&3VOP>Z{uNgs{tfe`PO(xj)*RI
zT16$Ws8O)BMi@`G7EVH9`Em4sQZkjLwLvzEqc#HmY0ud^o47F4$~ko2Emyi#DzQpJ
zl)rG4h{52Sv3uY}S><KumYP?24sTgOD3d=|7>H5vF@wUV85y(Vvym(PXAQXd-J+&G
zXI;*6ET&M|ht3fMSHFqO?VPzSW$KW7gk9aPFxZRceT2d~9EfJ@6e#L9LW$?#xsrPU
zO`;_N_brU|cO~tn&aRuQ-X>xz2UW2ShKvb8?x8o2h{`FO$m<)it=sdpu4z$8EGwi@
zj!AePADF^{VMSlyI@VcE59}-(<T5_-5q2^=l+fu-!Qpp-%GEhb;@lzgP4(470ajk|
zV?zgzW(C93ntQ+v>fyyQ6LhE8ZuqltB3J1}EfqAsSyzYW;-RHuVIguit5G9NfJT~O
zcUoI}n}jvWK-hhG`mMg1OBz{4bfyO>b8S8nhGc-;1=Deo?#?}RD{_8)>YYaFz$csG
zc&1+Vv>M~99zES&fShe=cqL!ZKIw%I-Q68pxNqBt+iCSb0MhEgBU@!*wBzazK6383
zd^FE}=Hyh?pqg2bm9<1e5g*+YNNUs9ah7ka{dK-sI=M^MQr}A-?)2I$t+iMeKB-+J
z0OR4M*4fZtq2tvlZgLCz;Knp}^1R0MNfYm*+}ybvj>W=7nkfKzh2$u8qDQg}_R<!6
z(Bcarfvejx$am{;3hQ4#m}`Tni@KAmk5;NdFQo#Wic9OOo1Tu)sTd^2lNFOo%<>#~
zi;n0SPiB{>))&13Q0^*fzq{8Ay0W|OH%G>2EX6EN8E^b(d-=S0VPW)}opkoEo{<GT
zOa<b6&S!qtx@atE&{Q>Qqi}R-n%7jn$yGXrd`hEC`4f5?IMX^WCt=+3%B0fp5<H^m
z=nhSAI5Aw%9mUTyK1EX8E}ij<xvJ8cL)VY%PfsfP?FtNvv5>~$LJ@J9Bz<^-3?HR(
zO3F#9VxelD;Vbu3a7vzNEk*kK_3xVbT0z1mO2{6pJr2E^>5R4BzN=gV*8M;!%z=0o
zzsy3UPkyL3U;nEDJNfLGqvt6u^**73q0W;~6KU8$M0U=XB7&q)5*__ar`JBYzX2ci
z?YTZoip?5Xhdb##Dl$`z5j|j-We2E>)RnK>_Bg*0Ka(5UdeY2+&V6B!DQ&{T!#hPT
z%zvaS&%kxI6|>#$6$D>&F_h7aY?vHh;G~{bP*aT5XD3};`gDFQc`h7znU*`{F*~Or
zUohxYnAfIn%2k1l24qE%#PYnV`m)6!xt3E!G(9&-*83p821Mo<P6`$PdQRwPpqmKz
zF#*oZ2E+_93%09G9oGJD7fJ=sd`MJWSZ;L_RC5>dXetqNf~;B0wsgJx5Kb~$TD4dc
z2I~rd!H9->q+L1wT!jp>$<4buripo<sU@{Qyul#1ZWeAKaC<Ir5SzDi7fyd0pqP*!
zzeTfb^Wj&Sb5tDAeL&zFT>%$Dl}mvnF$9=gV=7~Q(Gu~j1UD0>-r{(5RM$%&j=dF!
zL$klKl?!4C?`(8V^-=LrxEYv<6c6LX4^U2iF<oDM`k+$lE_zE4i(H*mGQFABpkb+_
znxRPX_`Sm&n0~_}+MEt0DvF(whTj08P2BX1sd?v?Q~LzGz=_m@z(;tvqNWtzinq_J
zpNh6*bEcxSbzX3{DvF-@i5<gk(hXX!fjc)cwI3P@i){Hw>}57p=zv?ZR<Y6=E?_w1
zDJnEmSm|xMXl=@of%FIBHj_4$4>{iiO+OUzPkeIbN+Lr^>t;pHjvq!P-#L|+Q?g@y
zi1F6S`VDZ|@Tsk3xUMX7Mx{%w@%XUH&O^G$_gjAW?=}Lz<@yEgLxT-+uxKs9$I~D=
zk3yK!1fDt}Nn+eOCVq6U>usf9(L8_JHujRZ+)G1Gu{pVo3$C753_9R;V6aNs)pEO?
z-?S*;=@6qyT$Bt<1p+r21qNXQ@1t2WIIR}<{G{X*pk@-8SakTams-;r^stc1xJMXL
z#^s<&%aWF+(lZHckU4ytcNxo$j4muv<}2rP5H00|%s0JT4-7(lKJD*(a6U*QcU)*r
zp$28Jcx-V$WDLJ9hPV!Vuz=>+sNk1=+-p?UTbU_wJUtmj2cZ*Dz~9cMP4$qGVG-G+
z$VGKmRh?%6y}eWXNUXvbM|W>DQqk6!mb=apeoU85OtuP_4ixwpvY<%cnR8g=HTV56
z9Xp~DSkXRPN?Kl#iKGQ``+%LKBtZ%h`GOq1`Xe90BVNOdM$kb<rKyfDeZ3&O%K1>E
zQ&m-A869JPbF*6~_-Scn?Iups6e~>8G6?GV)no55(n3<|GSI^7&la~k?VNZeLeHGg
z=x@{Bn<-^^%&)@*V<uK|7CRTJ!e5#!?)x%4@ka87rZI=QCY2I{I3P5MiFuWhUCZyp
z&G^q_dS4pNIj>C=pg%l8VRPdLH9NDG_?_^{vP0-+h`H7~jStQj<lHwpumVDlu<>V3
zg(^=|#q*t0MWM~^&j#nTzs_0Uxo;ViHS3!>;skU#cC=dqm$>udB@kiOe0t2VbB}{`
z_KI10?Cw0B&7{0tbq_gESs4K1G&uH)n3LMKqCPNDUGi_((_(kXk6F@^dU_m9Nbr(B
zB__GR%;rcJ$}8h9tCJ0DoOJ5O?BALyK=o*V>Oauj&^)>Gtig$M5^NLujb&cmizd3e
z#LeXL&@0Cz+#5$+aeCSukA~{+@EX5^(>3{l6nc2kY3Km%94C~qfsM%+9Zz=hk`-ga
zv%1RCb=6eG?TR~wNaW;ToARrayP_eaBHJZrB`$Jq;L6D?Pl2HF$}{ovXoT^T>sEX{
zLv17!SgV?7Yv#bH3(8paZKgKF3L@t`7ch2as!J|e#ygX@ZiKG)og-(vq}~OTVfC=K
zxWTdzLmlmC=)^czk1Vgcy5gD)P#B-M*eO}l%x6|e(06xFK<mgbKD@yo5X!r=8A+ZW
zgjL6T^hNSOieO4_JD3cKbu6FxkTbbo;_R@V+nKEe;@2z5=}NBVH6Ha}S($GCh?`C?
zQ~4+uo@wAH?F2D#LiFp9UQSQMy~=HG3k=Gfi_q$K56wg%#hXatG;s~}gITN@*{4yk
zW<IxD?;yKO6{0z&dAny6rNEh~V&$4CEZqnNe^E)khWtsaBe&$2o_7~@E~VyKV}atA
z7T7dL+pzmFq?+m_ckm}F`2ne1K5Nb>jgbXQx$~u`X`SF%-N`I2UR^G<5tOSy`Kzwf
z{O6}kb(R)BvEua&i)uP5>WC7aZO17BVt5DG6|>Je*#nuJ+2w6f$Cf(s&gxw~DTpXW
zJgO?>oy$I2kV|+_Zk}~EB4?FnGdUwEATl+h!BRc-3diBFx<p3f!TfCCQv`lVL-5Sf
zqk9ZouXMA%ADCie>AVLRpo?D)9(=rrE8&=)zN`_`k16h^DeyVvbdRELE$1$?CLaxo
zqZja2jW5ZYnu#ga&M!8%^-+13p=m(yx&UZD0q}Jh3=DH08CrY{)p0~-(Ji_RY28RE
zW^#wMe=fmKAnWz+7e8Z7X(*Ug$C6?ti$tgTE!CqFz{aa3-#TdgzgrS&x4ku(cSF$m
z<m!VC(9*4+!0HGTlu@RLuPg!ZPJmBmeysdVpy6JtwR}kucQP@Nvi`a;w;w~F?^eT_
zdPTvwY-PPgDgV9Kslk_xwUDfsFv(1jUTT11Vla&tN_Cc}a4W4jUyDf19c`Avvppf3
z5{al)r40$IVuZ^ti{V)6w7l-??l{|=x2otNCbW2SgbgMOyp?!=1B@9A3o7@B`%AaC
zU70EmN?oe+sXdRxlnA1hj(rGXtz{In-Kpr=&5xIgOIvKAx6`X<f=-+;bdyMUSS14M
z8H%?n16^F}sp;8WeBx9V*H^iQL=rS0)7}PbCh3w_zNxyZ+K|%nt#12FXN5Rc)CvVE
zs6=-CeVr_Bc!embHNNoE80On2%#B_{I=<zYR*c@hblBlAfEp`&)cHiv!Y1AxUYCAJ
zd8B1q*r4MPUuis{b$ivuQ2plDy#-nM-O+w$WiJ-=j}pQ34h4x?>~W2Fo@tVyiu4Hz
z{3`c|h3(-zi4DrKqUyO4K8UM&eZLUYQuYRAjwhd_vh}#4Jm4m%5tb!KJp?n?)3R;&
z?CNJ)j)QA?glI>y#{mwCxgSBe9am2%pihv8#*U;|mxoT~0)nXMMib{0tO^(yV3^xk
zn}Q<NvFKX4m$E7MLv0*4UxHNASJ2JpDCH|n)}yguxt&rL+E&$IwmkizbJ7xA8EP`*
zBoveY)lOzPy|_*gcO$9q49Y>c#$9LD^cnAm)33wk_<jRACK2-$n-F78xWa-+380u;
zhAu+8#M~l(E50O6Lp3`$tt-=StC_n1gHhFwLNCliUPuCt+<SCS?(6PqAns)q>nSMd
z81Ov`MWqU(Np5&-9()D)@6Y`}j+&c-_wQ$-^U_KY==j2C@p3~0gM+l(42kO$7}q{(
z+4%~jpR0)z%$UD{?qruvZ_L=C7%3sUqX5;(ke?HaebMk?%G_<{vdqkpFFdP$a#Byh
zy$n+pPBE0_m=9KPPL6K6ZsT!rKdyV-TF__;n*|d%*3D#H@)Q&)a)v7cy3{|l)kTEO
z`fRaQw43a?oY4XGeyk)J>g%XS?m}*9`_bQF<UQ~OykkGUoYrwIoJ?!mxE<Tn8@P>Y
zbsF-SHvGJrWCZI5oEkIVoahlWM=wvdIyD*0O@Ol&{4wo<x#5k!0qWRjJxx@eH4y<g
zNHP}_VbU^){c?JH*CfDY5A~5I3)WIH0ag}ynZ#rx;}=--@ry}IRmr~3eW-|QdH4?8
z8N%!AFd+^Sky?UL75JVBN8JvlS5PWF%h|oIiKb>sUBnU7h(yq3jHl*%JiXTJ)H`wc
zvUq~VC0ygkvq0F6sX*lh@-VufP3>_(8Uwa8d^|5S@&1dCivAIyN+T5dlqDs&&hK|&
zyBY6k3)E31wy(&EQ(5oB0QmT&2ZtjR8otf*&S6?|?Q`jqdP)k~85|hn)w~dq<k{?v
zaF=1M_`+$b?aVY~e5=X!?l`TfeaIAIT(v;aI!!D#><Ob(HzgG>MByC@9z@*p!AD5k
z=VnP`<tJSMoHuHSJC`q*iMLnzWoXgDX{3zVYxu5nZe=TW$2U#NDV2!>qU<QYps1){
z{bZQ|mib@&2brxxPV?8+F56<%)eSb((mzHFfpjrqFV;J3N6oFv__p%s3dUT;Dya#n
zd5uZ*baJ?h{M{Kq{jpQJ`cpLZoT5A%a6YG~ug#yZ!T$$q?;Q<S`}hAML<xc*2oj7o
zN}^1ZA<+gi`V^h$y+@baK@bFC^cu#@m>D$~odnSn!4Qe)-6$#AoqC^nzTfA!*7vOS
zob{Zw&iQk%wbz<!&%X9`?frSbKd-mlUJxJo>LtEE_;g=9@;8&IoAE#%R9%AY0^qE}
zw_YQ#?)xJ>#jCtisOj*ur}#v@5S(|^jd&gP;W+0vppC2t%N4n|h15)gp)XP%<niH~
z<Q4B^`bgXm-V|+re{QgR^RhKC9rNXBMFYwQvi{r9n8(C|aZIENXpj1qyApo<)56Rm
z4&R3dmB1EUA*w}^N~7rd#j{{0E--P#;kAYx!HuH1oY+tmmv<_NEp~#!z9z;c=-!;=
z%R|ls?PTR-*9kRwL1Vl`HrNwa=BLNUrW+HjUmh4;{*`-W;rQW3q2D&Sq2$yZUE+{u
z39-1}uIMT>E$_?1bECWnuJa)>V596w&bOC&p<!>~e&tpQ+BdMUd|q+$Y@9*RXIUMW
zTE--t#zMn<nKXLAy5!|V{r0|pQOLJzNiXQYbj&VCAu$}H&st|JI81ZiIUwklv-}p4
zXBC)RgVFRy@qf?dRt=Pqhrr75Q4l|Ei8_zD!fOGof7u~PA1J{4JnPvm17!N7+WW4|
z3B$_Ud9mL4LYOKwAuxF+UjFk~U^-6=w-y}Dg<&27MBdevWwnPP>6u*_J+5L6Q)hUR
zlCTi~7d405Pa`!p7DiN5-`$p|sZV?nxoP#ui)^c)Q(`QoYe434i6;XNVLCvGxJfIG
z8}0#DPpj$;y80-_iM*t^*v)(2qBcK%yL8f(mj@sk&=esHx?fIjGW9@{eVvYCjFfjW
ziDeZX64$DnFSPzc*LscDbS~-vH^_RsT_P-z@$81ioD{y#(zaG!wNS&Y&DtCHI25=f
zou+#u@5+d`^Rl9SHXR*1eUqj~Hbe+(!<cAtDwSG9wmjeAQ!zz2I*<0_dR_39WFsg+
zA%bu2R!s;BUVj09A;BJL$hV+-<6Kl9|GoU~Eh=#1vGE|lddmn1L4_G2D=RtG*zPxd
zU0oZh%@u0f5{A!yXKIk;$<HeHg3?LcNi<YhlpJ!tx6AgB)yy`}#>7F}Z>rPN(UmT!
zhE7r<ej#>QIM40RljPc33sVae_&2Yge_lz-G{VpPNsmz!oXuy8CDn*=#S1wr>FfgS
zpB!vSSpD`}%?&%#=CkLn$B1PG;u(pkD_m-u@T@tW*olL7L*-7EBbs;x?~G!`JWu}e
zam7pF<wT5B9U87@PK|q3hBlwE$eibe6}!rQtr%G@VTnIZuu1AReXa2t<$QM6;U~wx
z{T>_tY70NP(|soNZ+F#zMfts9#qZr5cl`>y9tauI9E2|S`~M46u&sHoAfhXAy!JF=
z=A&!NNeaupqH<o^rZ7TCN78m*>$=_M=~vC>x2}BQJpFNOPE3=mnlQg4ERV57&>NR5
zv6PFG!FFybS%Vz!R$wu}t2PQiLLxsPL*g*sJ^b+pPR9mgsdT`2hli=(2@;#lx2hc=
z<DKo-E9iu!9_F=-i69GwSc14&c%ZD=JUm7|BM5w>(cJXzf&vOK9i#0T&?@_w84KH>
zL&v@#CncAi1xlhWrs3yE6<18GLePsvI^Z>59Xl)P@Fg^fHHXI&36}!J%E^iWzlbs=
zv+pC7YwpKNW@UQ{LXUey8r@1FOkKIgJ+HC5G7_RYn%X8Y$+);=;H615<4+ZDv?th`
zZie~H3;lgmFDcz)YdIMeJVTBWkqe)-oVczVNUXs*s!KunZlCR$gVh{Mj5m|LG`L#j
zg?Whp?sa)tZ^tGxdIleCYECvky6oAtg}rS9Xnx3O(cFuF4~bo&1&l_#WgC!8F`jK;
zwP%rjZR_A28cuo5OT3X!4_vZ>v(TN1ehz&T{-?ZlRtP=s`s~kJt1x0#3CJAH0Gp)O
zj&=olf4ongU8wc*_0r4VJRURmWCf?gdEp@s%tz4gI%mRpj)Fda4iXMJ+iB?<pny{<
zTv>_+Z={y7vT^R+5xs1stGZ?=gU!C-31N{Aj|%NiW~XM!O4%A%H9#s@^`O$PMf36{
zA#6}KW+s5k<Busq-gnz~$AA0=#k~y1u+&i#RP=$!;;S=g9=OtC7ytrD@Z_%7y$}PD
zl@5=rcr{QE^#ENtHXWn2B&F7_snVOZ=&l}a#G5N`B*agyV1?0sfp!j))_C4)vFuTm
zJVueVEU@|ud|7x*qSZ{SpziT6kv=K@kBeqn<AqhG%7XVZU!hUYnXjtOX!_OI`uRsb
z#}%mi$!+gicHZyGO{4)}5UgOW3CmqMRC-E-4|ynM&Biq9L3*+BNv)+di=esk?bA~&
zlQHkT3=zg&nGMmkWl=?W1(lb4ea+#l;#gu1b7Brp-sOyoSt$)D1*Mr;1*<Z7DM_AO
z+8XcbTK-YiJ4Khb$wWAPHMuhJ0c0=y*hnThWW3n?Rq~DG>sR?{V|CJ+-uO$9y9Z|Y
zjb)5RKB4+Z)>v;&LwR15&tb8D<Yj2uHdOO<s9`66v?MFwIah>Kg{Q2qsWP9TauZir
z&eDNvV7Z<wpAgp6Uj|BV!m19{QjRS*sVPyXh_B_a+{54i@s0HRw6iE?>gL}QEz75R
z*B@t6UeWIIcBHms^lbQ5*nXgF$t5HM2D{?}wf<akXL@Y)c2jQ4^Lg6+6H)Pc`B}=R
zVa7r@#4uS<LWzqP)zY;I)qDR!2#%94V@yC>BqYmI(XwK81_VccVR)sksQ9q#6(J9$
zI@)`RH9W-lfHuiPC~w&4({(%qbr$bLA4UFMt_%Ul$rONwtA^yDP*L%Lf`~E3DI5?j
zC;@B3)-y6is|+26Q-%}C{Xn=9+EPbwZH0CfW3t<6td7~ENsZI+Y#p2U3dK+VZyJA3
zYlzPmf9wC8zFVpP&EY?fF6*_Z#D|GQ;iqfE^C#bsCE18csqY(DvSUW?NYd?0Ub_FY
zgtc{g3>&EO=o7X2AG)^D$88mSmvtV=x_gI2>zBBB1zxf;eX}4M({7GkWYB{GFKVbo
z#v3`up6vujyt^*al?or~!6t~OOfh3L(Le9iXGuxK?KV+VDZMy4xLyZ$;zpr@##=KR
zw{%LVnVIJiV=7T1naQLB8^9#(vV7eqH~!*TXd!d!Qdh}WV%__N0j+|DE*USu;cu)9
z&qb9s{wix-An*}u<&sSi7A9QS8l3Tx8M~9K>?&JRS<Y+v*>6;8=4?}RW@t3#)8fk9
zq1hvtcMnzYP}Xv^swD_3(LpRg<0@vdvrl?C!f=))bM+u5t-aNfF>V{SK4t#0^E^>!
z&XLO?1qeDh;?K!FQsC9m^EY8*fN{KIzoe7U57XCm8$Qh1rOwiMW8p*84F`q5hI0!G
zj+yw%QIeaf%5cw-aeQPAn8YNle&Y~)>3K&~4SD-1*{-_RhH9^WFEmk>ny(G>D2=->
zuKk*$wDSglbGg~^&i3WIifV){@-8(toh>4!wsW$4j92pq;JML;NN;>-BtD%~YM^PA
z-nBhD;QH<;0D~@hDB?<Ft1SwTV%@NLQ~3`a2`X&e(AU2NzL7~<*-TZ29@o;j6E8G@
zTMIty&Ck>oduf2FpEAB~%rC9qb2tfE`>>+5JXr{x`&?TZ{L;+9y_Z_5K=S401x38T
zK<ZY5F7zPQtdDj@Jo=bn(|y{KfJ7F!Ih(NtqbP=Ret}SocN1zxfDlTTy^%Ozf=7JF
z?I__|;<{Gzyz;)2X0C&n8jwepI(&=2E~c6OLsk1=VIlK2M8$mXL5lwn!gK^MymHLE
zadlAb9rQW&9;#1=ty()yzx^KnX7@jIp7$=LE?I$hlP7#c9|espZgH8q<VFVyvmD-6
zr;6vr)QWPp+i%4FO?qmVIsa$q-=^qj+BMS*UtEp<r$6Cy?vBmZr+cL@%>K2t+h6o@
zy{c^=*yXiVPqiYrV@o{XpDl+_iOJC3od7T^>srPB%YC&#3jdOmo;>%AzKx`2z7wA!
z15aJ%fX)YA!$-`_^fwY&86#2vgnS;T7Bm_-jnX8{9=^pW1P!&d+q%$1<n*q#>5w_C
z+NrtsY<By53P{S-ypd@l88IuU){M=pj&@1h)_O4$`^04bH{H#Awq(##LPg0(K>D6<
zPDfQA;ayel+@~jObw9FBYYIB$&;9npHb1nOXjU(OSv_<oG#Vb2j&<8yLz_eFiVZD8
zpP4-4$PoYYys~XEJS7?1jaSjVLndW~^rpc>ZeRIeexYEW!|2V^vL7?=oB}|%)6KKZ
zEVO|LjxInWi=)wn@PrJPvd8b&919(iT9u}t4whZUTgcr6I8V}OgWLt85&2C}>;_`F
zK9#%}d=Bj5bzzYzhv|;6XZ1wc#tNM>ihiXa_sD6W%~<m^<k#_xJgcJ|*_y*r!7rSb
znaIlr1%E;8q+j|HP8lTo-ZL73U~9W`9cWbEA>hlx8Z1KL5OmT`AJxd07m62(c|E?8
zOy?pgol1A*#R9Qsu~zm9#OfXb1oU#!Ne<&l0QO35q<r0%mxu+n4%bX+${4w&40;yY
z3HbTuS`ukoP3ag2Y(XXa(m~=Lv2o#@qwiY>bL^hIrsfR6qlP6VLFw_7qD1Bsx6sG!
zigcy8SORw-D7Qpl3&et#(-!iOU~<(`d(ook88WYlISeo(_0jz8q)KBt#Dr#EBJ;m4
z+TiIBdM&p7TVlhKt~OFspO?pk-4kk>MjW5PsB_&e>q*3)$!l(u09D&dWNJDwHy_oj
z?v<FtOh&O;e7tvGSeUzNt|akeFsd!(JPeDmZ_l^j(90Fn?yg*0Y~MG7`gP7a$ut%U
z$dR(V%dd#+J^S{jFf%El&2|U7tuU}3*&+Tp%)0P=n1Zkfh5@a^uJjT3;=N$Xkzd!p
zG`Ou)zx?(M4I4e>n{2+8W~RNpr!UN0EpxoO3>|M<%^KkZ%(fGQ#ZS6>V;fvHtp(?1
zvg*^<MWU8a4`=r?qq=(oU%kF-4U`139KKo=@P6>}<@{cQ)xJq)v%#QsX*rqujU_K3
zJ?U|b#%f)Nwp|P|WQp8+UQjqYI}M4cc#E>JanP2@Grp1UA~Sp^)K$XeFho->(?eBd
z$_S+_3*=9eN{|!dk=0m&Kd}>3|I+=%@7j=j9*=H#eOlxy!VnA7#WSMy?xX_Hp$h{b
zzwwv**JZGsNaWHf`|Lt`1u*t0S;+T&Q)N{3vujbD{q6PV{%o}cFWUXaV=vKoYBGda
zzj{z~8&(%8?5e?KdRtQG%f2Fu(`g0v<p{vW;c&Q3mcN)!n8K5Y{UIHdV(2uH(lw{%
z;6k3KUMfei(1r8N%3u>-i7rOjdOnD+#?;LZp{!WF5?4@GM%N7iRs@)GVxF8^-tFwm
zwlNClR#N}aVO_5BQvG{P)kycGx||fgzLL#Ds;w<E17re6tc9;)YYCOaED}1#mnv?3
z680LCf8#7s{i<rOzZsM;$Ix)z^<%?a4|By7DPj3jCxp3sDCw`p_i8bn<!@@`n-LQM
zZBr;~_>5xtMfIHwG2=pkh5C)M3PnUsBWR)F<9OU5^^trRhjkai8VfLXV*&uOFY1J7
z*r$9UQOA$N>p&5s$OuHS`L8`2Z2t3B&WmBEW{<k32cb;blg$Ez8j$L!0hUEl%R`51
z-A_)6uI!|mY?J0oqEHB4-&lLiJ$4?wD@=k=DZb&l04sG3<{)R41`yx&v%;QU${7_)
zK{(ZY%9mv^-fPjI2U@FiaW1Nd+SNxW3{XBpoPKO+_mpo~0H`w38Vj0dhYMSCHk#C3
zUl&pxp#rG>0~l`HY#=ik1=mc8Y&uzDM1`$@IG^<{HK;}J9<+>D>5C9~&&DIf<*}7t
zN7V<O@O*&=YgR*jf}A-nrdctKm%X*_vSmzZXvYpz=dJEsxDbANm1BEbw2_!1BRM-C
zF~Y_t#N&s00SL;DGvvH|3g$QSCaNw0@(TP92UaW;=O7juFzDe0!nr<6KaLMnYQA5)
z<fLu(3_cQ-&h|2fC{f@ikjHAN$2!1+dYvUKJTncth}q_^XmqORjmq==(Yac&Z<vY-
zC+b<~Dy~BrGphlCH`G~uX4`8yVvqQ{$$=t{#d}iea6Tz>bD9$<&Y5(oK<zsy9YQU1
z`J-9jz&YhEB=;cOy3(TSRq<J&7OQV24s(9FR?5>TBViQZtt4X$V^~fZJ61SadDAhc
zzTgw2v6eWX85^k8=`N>Z3riyQ_kRfa^0cje_I%S!Wab~bu)6e+yfNVxbaGNzjqF;}
zVHg#Ri|EoU+TE9~fjAnnQh3G<rgO{Z+*98nSf&wg9$H*XoEAz1r<yJIZ6J(LPE*ym
z3IVM7rl0ykPG6)3dEui~4S7VlvF+3Ip4(LW4%tm7rXD0J8O9L*^kd?E^eS6J{(dzc
zgYwkH;q0-AJT09(v$}A!X4<Q<BaM9i_Sv(a`DT*^w6Udh)w#t`G*(JmqJKADyR1)v
z2z@WCR769pB%QoU>(Chc-7CVUOft4GoL*$G`<IX(&ux~;Q8K661IN}t>OrB9;NeWs
zOg<c4ot>C{tNgKh%J|%mw218Y4L`$BrYw%3rle%P;ryNG3+3^fJMA~qFO_?015UW-
zW{n_7ED*yepJdV%$;Zi<*)hR#9O1;*v8*v-DxnwDW=qj2pHfpqcnQXf5Et!xf`q=Q
z{zLbx7Z(P-<9PdddX2Z;eq02_Y(W)YUfodL{K*52{T?s?^)7#a3~i#jO$T<C)W2ur
zUzgIBi1)K6@VykCBH(<qE3?>DoDfmdxr@{)@?JXiQhdke+(ejiwa0Q)ahLkx{)hD;
z?L?ZB)NW#FWSN=}{LL?oVl^$AT7@ivvaN>2T`$GvAQZxkG32^@S(pY;I!!S=GlEo|
z!op1q+Dk9u9Z0b5{V=Q+l2cyFDteLIBrmGo4j-PkP(f<jmwH|)m-5LcfeG43vNpfX
z5UX02FWTce(dl4QulNK1JF4LG{Z<F}+?7keW~W{K!J$HkjL)^;N_UrqwbOQIzitq3
z`0XWVR5k7yIJ+N-6%*7@r#o1fE%}U^vOsA|VY?}B?W9dCJE(d2rakU4MqqQz4~8Pr
zL@N2^EwVGN$8rwt=yT8v^{(N?K#5cN^u0n?<2!lK=5bKPna?@*C>UJb6U#+^_t|Xm
zOK)e=)!C|n=D!4?<6lAy4M&*0=YFJ<_L-LB4#yobHKW{Wo%AwZWWB1RRrlRVdv#=y
zqqd5!tp^<_<P1TP6;PDRmM|*wLt5K4$=14~2YEb6=8Ie?-Qufg?p1yN!`P}9R$e(F
z6iV9w9PXnxf6ZiVL`|L5exJ>^nQ&=D7hl}dk*#&ju8GFh9#&tdbx{-?jP2|XSMzD$
z3tco#xhH3UHR+44jmt9Gd+AquRO9I1>$D#vrI7R80H-U`A$)-)p~}zyK&z^6CA`T#
zcwmnMgfdm4k!g%Ajb%SX>E*wEQb77q!l7HKs9D0RA2!?8SwH_;V1Q3jt9Ez2*!>#+
z{jzgmju%w$a0v+%>0j>D7a<*CQbQY2r0DcoWKdLGCY2m<4y2YWkT+s|H|6X`IBH%b
z`B|4^r$YBQo$smf)DHc!S%Q6iY?@|ji8CU|F*&QL&Mw3w8}02HZ~2wtcvX*UMp6Fh
zKz~1UC~r=koGi>-J(TH{9BzFpu9mM5_Jc&t&Z{)D60;vLqR*Wb9$0ILfGm6{b^5Ea
zbIuY=zxizvcjMpNr1!7cIW_;ka;MwaV9%x7=Y2`2Im0A?(Bs_9jf`xiChaim^9PkP
zW&Jg+)^j8B*4R13hy@v1h=ajq=q95gqlFdgJKrxRYx)Csh`iG*Vmx)Vn2Zp;`>!}h
zoJhy_o94cI;@50+;XH^L3e4U5GI!4Gfc%{QLtSjA5e8V#H&I<nY;06{*O4p!^@b!M
z=T$CRbJ<gyqwixwXJ%A&x}QIme?f@ArwdEN$D3w*7s$*dR0wM8pnvV4qB`ag`X(qR
zgtpct#~dR-GE}c13`ai;U5LA9YLEZT@xfW5XYooHD=VK{M0j_l3&)ZAt6>4}p7%Y2
z4`#X|@tBhDRv|p*2~nNXm~vG1sl;39Xq=Nb`E#N5p2pM4qXix<;D(V6$IOD#P7qN~
zV+LG`m3zH*dTiwkQh@t4s%ds7ja?(643ZtHn<D5}W3JzfAh+v6?nR`ew>>oM@Wf+^
zU{@>{0ET<;+#%St(7FC3JF?SZl!9x0O<VZjk9zO)C2uskT*ZtxR9@W9Ez^N5i&9Y;
zwppm*F>j<#(dU@Kd<A{9q%41KWl6g-GExnPKeFX9jLhm$R|@QQ3^*NaY@bKYSl>Vw
zuVq)cRDg~A_q+X;bRRaRtcCT=&!H$~gYvl|yIJ?QA6~!0vPSTmCul^yF>q9jdb-j}
zNDDP{3WCV<QqmHQ<1ecROil_6=F<FI&JuI6-e!~mIfax&ie<MOSGJtTB9z`p5U_DU
za^Mr5_SV&bOWz^{MTD9Oq}XniLzHg_9}z;pI(N9X$b}Bc19(l+_(FQxx}HRqod*!k
zlbJ?$N<a5uK;!r#X{}DRp`jkzvxvb}=3kln@Y#BH*hj~i!FTH#J+l`x*J{N0yU-Rz
zUPm92ET0y%LQI(rwqb**TDbJNd;g)uqbdl~d)i*qfHP!Ci5W*5uNv>)xvJRMWwT_%
zv%l{LoMZP*QBh6y<j!IhTxE9WwEo}Q%9-~q5O+I&JUPca(s{Dk{vSsvqRC#~biY>K
z9+};vf-+H`EFADOtI7P<B_##$7qb?PWeL|0hT!TKKXl6|NQzoAp~1Iw@J-ttsQ*(8
zm%j{AL0>eCL}Ez$A}RRGIC@1XToJ!@3fF2H(q??k;_s4r;_2k$Yz}XjoF07f<NGLi
z2_h+)*^7MN&jHa=g^v=pbl|8YG(B_olTT?4zZRkwTfkeJLPCGaaG~{<$N}u=tp}Sj
z!pWEEg`HziZ+rE%j}pH#DYT*2%ve+Kwo@26O3T|EjSm-<m0CrauL&!*P*V5B*R6}v
ztfbXfZ|%GburfF94LOw%(b!9GHvHSL^ids*<e6AO8f3a|Tv-gn)=3HcLzkyxc&25(
zqTE>?(CKq6GanUMTK;}rQvlO8PQ;`!KAur9VCT-*$O>8Q^2+p=F--%S56~}&1efi}
z7WbIs{T-=MN&%O^;1zTi7A}f@+=H<QTCs{?*Ql`^UDnttNth+faK-|G5hD1pB`M=P
zwW2JTLo7C%^{o!1vhIbo)_zkjUW$hM{c1s~^xhpUHz6I0Z-K((U5!+~Sny<^YxZt3
z23?rFWnss0wf3kli)bY@A)I}o8S>^OxH~@M$zFIioJE7#C+;r~s$B2U7jK&aZ@&XX
zVF_EIzVGhgcL>CTv`3;wh0ct}foXk+A_<dnuH43KAv_HYBPwBCuj^INE!okc8#ome
z@P`(&b+4FuF~3Y#=(_M2FsoTuv`?a`)0NTvM#h)8EXeELP2uaw%Y^PszJp^MZRJ%h
z<>y$Dnp?CYE4#6p#p=EcE;pF^F23u~^uB|e^h4#5qwD|8X{+FQK>efiNPqtyI+y>p
z+5>1-d$-S}O_VdDm-kR54UrHr$ezz5CFAbcEVCw~2TS`xN6}SKC8bAW(+v*b04VH9
z($%Y3dMRFE?CaK#r-bC}C}u)rNovWOQBc8}9gh!}yPmtCtzLT1k;JtN>#aU!cS-xf
zN_(|KGo(v_6(;Fs5Fsl-GY-mdBfGlkGe?iDi|KazlI&lUi`;Q7^A24hsH=GfR(;IE
zk==7YBoGUSzFqUSipW_uy7V*`xYgrVxpNi!tP)yJbbZDTV@NUqc|L`aF_zNBtjph`
z)8mmzk?fs~w_3%+fB)Gs@@QM(;A;Rhko-=hV{X)8q1y*6R@YR>DLj*&Ciq6s3riET
z08#-sV>~ejY6vk%J3&{>2oLN})VsR1PlR^)c9)9?8!)|7wbg^F3u?~{ig@Qk$LsI&
zDN_q^iem}f#VBe0rW@?}(MsfrgN2)LDiHL3LEc8_hwUVvDpx*0*jsS_&P5*U`*DWZ
zvua<(T0i>lcgZ3c%NZpd&~oa~18L@-&r?1MHW4VePx$F1b_Pa}-d&Zh&JK{#p3Y(i
z$nQIvq@y4}-ejBu36{cU9+?Pswjro;_<=YgUYaiilS8aV7&gsNiU(g$(gc=a(G1ls
zt+!s=z4;vLb&jWQRQD{@$^a$JPeLuYVKmV^Yvszz=xNR87N3R&_FX!y;BwO?(QI(2
z;00~Fqb79A#A)G8;K}Kke_$(eaRc9mSfomt+U~leRf!PW)okQN=i7l-85gD_RqwO0
zN~w+#myY>l1Z9}puZ7Us4<F$)v>*0ZkBLwdx4d!ZeV8_2ns3%Yiu(CL<XMl~NOgo1
z4(sGpF%;66sFFNp!3x(M4y{<;U<KbH9HPDr*~o{Z<g!vPz^epb)0M^iLuVB9wjr=e
z^`Uw9>`5EYFLf`YKMblBpTp||h36-<u5yY5D4oT;{bcO(QNB><P^}JIM{0M4!)<2R
zJk1kBOpT<Yv%ob$*FCg4C<S0tTde~%{VBwAl6SUWA8OzS$YP3>EA&}<$EDu%#B|;1
zy42jwTY<4bCIgafPJ4icsxE|tYz6Ny?m}0za1CRGdH|Mp6w8jL_2VD9nrLkOl`_X)
z>)l5_yOa;&r2m%YBCErkAYH3HANi!Pt80fx5<`iC6>TPfWRRA8Oa^sp`mMsE$Luj>
z!_zf4H519<w@YbeETl{R5IY2>BYgkkhF`r#ET$X(7U5T)vWa8F=UC<d)7Fx{YPc$T
ztM}~whSYc_Uh6IBOOB9^<ko^{i%s4Fo5z+MSCjcG@O)}<3K5JK-RaO6@uXt9G2?d*
ztYDSY{0BJIMY19Bd<@pb7VaoXK8dq3>}2;XjMG?G$r>0CX?Yutf86G%TJKdqTSyGF
zjt1uHU(ZFH9GhvFpV5ud(Z4rYnt=KIL@lR4K^}7jLR3%F^mLddiD1pCso@GdwusUa
zaFt&Vs+$tJYHF`Qu0gCw6?_f`ohbcWjL&C)!TF<pEme1Yt+4m`o4nmzMLB*Cg^0jB
z;P^~>PjD*m1E)%V%B^=dyFRKqw(iwXuGXGT9?BEa+3XH?GnIMj+@s~o&H@%TLw!qU
z=h;4mM~>s(TZ%)0l`O}rFm2_kS!M@Tcde5732ools>Hq}ykdShnRO-Iwrbh*aVz+C
z-#x#HmAulRijKMO^5ZqnF;BIHAE#dqzcuo45<4A>miydu>BS<`r|_GjG?@F6#Ep#>
zkzkaNnibrcc#zKz`@11Rool8+98jpGW^|7e&*OFfT3>nh2PTbf2S36C;zi@w=y6S&
z)5I{gd3&*IhReU)pcX)$s4js)xA3w$HYPwWWYO7|NzV5IgIYfg!bnc8)e)kV-!?2j
zL?G{McV!8YE0J6i%npbETZ=mChg0}X-lB$KF*j{q+xu3Ty4pSS`qlZ)O8r883j3tI
zoyIkhvs$|;%JR6JJZODIFHM{tG?R_V%YCj|KENEY&TqyW6zgv74D>E2c&JsxTBG!9
z9h+U5{7~36MSfCzXrdhC*80hh+(xRHcDkPovb3DB5=tIn=;fPe4cbW+4*6{VUDpEj
zM&M-ee0QtOBDi43A`GLGi%XVS7WD=5H6>nV7!XpK2r4LNC%9$}KI)|rkga(2G)asG
z@%b_g^{dsb-j7@AS|7wZ6oI<nn&ep{6WeqvODrQ%*P%P~&;073fOp06i8Fe}e)4xs
z1k;~G9tDaNFox)pD^?V=1<@e{ZAhftGq0c0G%@>layYG-a)4d-+|y*;%}%Sv3SM-{
zu714{F6`~k5;h66jft4}ZKNOBP(5e!H&FpOjKSmapM)fgrp==!k=0FfYYrDQ`f@1w
zLNqJByz2N8x~uBe=M8Y$9dwVr^y?kriK!-^xL$zueooiI{vHAawRd0KL-fF_;*CFG
zn_dgP{%uvMdNHf(n%O*Z<rkUv%j63tGMf&KC`c#rb>VB=@UxUpl?SnS%`pIKrU75{
z3QH%!?ar(nGr{g!yt&vt!@XY)lm9l?kRw%qsmAItWVvTF3TSr>vU}22&6o&%lALM;
zOT|7!^6yzzs#J}Io~G7DmB#VkE;|X)bDtnoIZ3)s8@nLhvr__Iv)Dyeulx3?XHtHd
zY1g-Z==AqUmjH!_CUNPcy`c0}30ynIKum7(O@3CBGi!K@1%x3;Nf#N&r(z-^;xT>1
zI>5OiFgi-5tts00sS<qio&u?2Fr=VQ#>z@k+3lDj6fbKflEnS&Qn$@SuakZVrXA%w
zkq65EIZ(hilx)$ZTt7XZBCxNOcslXhrbdQi;A_p1DzB@Xiq_YlN|`T$hl+FNsk0}M
zInjhBM@&0!@^{#(rQzP{{n^X@tP&b;>{Z|v0p~MvlCir?8nxeYxOaaD=e(?Vlc;5p
zPVw$o@wd|Vt0FXzY$b2KN+ffsmt7PN*a2y(z^$D20_JXO{v8DJV2ky9t+)M-V|sqw
zKjrCcL9M+a`AjZTfnlzrgy#0p)h>3G^3L^LcBWqXhc0^#LkTg%<Eywxy|=Ci@mTYq
zDw>r4jz~_{q&+3_R+FTpPx}CzDao9@v_^2m+7|k8Qzh)P4TXY2ncy1;iMavagd&25
znne1<uPZz?rgB(H-ZlY64BgtmqO>5o8mp{_N4N4+iW<$ACGd?zr{;RDPgqmbQ=|u5
z1YwnFCqZ%N)cPmILF^pgg6jq-CWr{Wg+pJ-9MY@_VVDX(Seh?M0*TOT5l*$?B?xoH
zRdxr^Xg$Ausov*zE&rTY>~j4o_|K!ksk%3-G*ZD@*#@R#F~;{HmJEMuV<RsMb4~K}
zw=gdM#o5x``(iubwU(s5m53w~Uj=-6y;jmejRmYLY?S%>kHfd&cIV0Tu{U`IfZBIN
z^C9FvbW(~tet&b;Rfwk4Pj`iFg`5w!4FwSER=o>~+3dccZ+F(+8nNIJPggY(qJvq+
z(-{lXU3fgLS<)8B(bCZ9?p8jk{YFt|PjYV}Ga_Y9l5eJ@82;);d0exsgt{ys(U>vX
z<hUMPr58Afh8O|E%1^x+9}TYlb!fTF$er_X;a4X?gSBoC#24K2ew2E+M?khVrx#5Z
z&4zL5W$DsoM@Fa*=6cn>A>!V?KyBNGKb6UZu>x-d7_V2)GAF$>-V-(Vja5)F;#ePo
zIE0~#*g9(re`KT!S~7m=R~HyzzN~5_6=8E)v1COE3_ponuuC>5A{mgftNpCHDlm24
z9oMUVl%Hx~Y9T6}pPYypx!eXvG2mjY@?IZ@ntXe4QpW<ohANc~g31g2LuVrnznwJ{
zq8+WcT)5R&|Gv5^{hd`%vU9*3YFEifV1J1*SpJ-j)@SaYGPq?POMI%IEYy>SXyIu$
zVXeWZ^E9^z@no(mm9cv+dOaw9f)fJ{e_<AB3>bI|=OUo`l<6k&IXT0z$yl5|EkFkm
z{0z&+cyx+eE~$SzA$dg$eKuGMwGf&KOWL`1TLb#>#WJuvA8Dqjx$?p9kuohgXY%`<
zh(`jyZtiAA9f0or3v?6yT=T#3STwr%`xkJ<&-rP7CdhN^Lo57ATMOyAr5OjX2DeM}
z!;5;DERQGL^_K9uXH?flqS#j>rxg3O_`)a@k|v43Tq*_}M?7!UnOr^+7WGc4X9D?E
zqsVHh<Zk|jCD-DLEh{-1g;sF^ON5$-zkS7C7PxfTF{`2E)9>)wnsVgv5FUsUPP69C
z50+qza#VxOCc==|O_$XTu{}9^l@6S=1T(XsNQT7iq8`M}{2$h}ma`;0`kDO&*>Xd@
z)K~z~fpcjQZ8CAHdVUQu=-@P*OG2k~p~FqW>iUA=$bpx-RyPiStMHuGkiY!*u6B0z
z)GzKtfmK!`fEdr$;>c#8>K9!*Sy4rH<w^6#>g|GD$++qj8HP0Vj}`63dz!ux9FIWZ
zhkzFidtq4?66sKe*!v<@_Gq-ZN)~<pCXMR~1#rK5oG3a&mULKyu}wxLd|Gi8h`5;5
zYg8G2u^>}P{9Iw7Q-n?0+{OOfn9cHz?*-paFHVl{n^7Nva8;DsRJOUjZns?Rh)JmK
z`7qJ;+Ro(UI&{hSYlos#HkZNEQ4Qwk1MNnCLWyoyC(cqEW+#>uNt#-q7=|5uI^+eB
z&z~;cH3@GpT{FhuuR%1G8;vtb60f%TPo55EPoac@%^ibf+bL3#q(rJO0%D!N<%Q%F
z{1Eky>(_}Nm8G9hlqIlt`TZ%M!4wXf?kmA|lE|-fNBCCee7Om-UgWJ@rdawM3iSYE
znp#D?WIXxBR*#aoYOEna?tH}e3GP$M!4#O-x1-B3ZYSGd?Vp`aN`c4(D~5}dswXzo
zTu7YdvgFk>rv*$Yc95jvWtGeTuFlzd<M-DFS>JbJvmQGv`?__Jj#6u_bDXPlt=j$4
z_O)v|HVW!Gp}G;<OJ*W=iLP#KESu@3ay~lh5*-%I+v`+hvs_#(q<k&e)5zzjDXYZ(
zRHqy~?vRBgnx34rkAIj~&};vp%Alr}_Nj}nEAmV$%G!6?!(IlI|8RY^%fZR?$v6rD
z=Y3jWlgDDXw2>j4;lg~0Ef3Cxgc&jBHbs|*A<ck1X;L}9Mwop0?B{e^0<l6w*3RRr
zR(38G^bS$eytCoHXq~(XHGQFW<w%csN75%hq3e3PKLTxdZ>55=3AHuKl3p@zp#8Rt
z8!4e5R$X&ET|NDQ!hAhpS^-4sCcOeOLzpgL^Q`HlqP%$@9ISLop8Rej*82X_{u{&?
zoR7l|#)jEV35h~{e5DbQFWVb-u3HYSQM9Sn70e0^?h?8uQMWSYm?~1u{NOYogl^7U
z5AI!*lmXx?qyGRFQL<2_*OSF&$ho!iBc5{#7?iu;sd-7|bXih($kCSo)N53oC676p
z2(k)*H7(SywA2-5g-vGdIMnUBx7?FGC{+F2P(X?Z!j4`XJst$EwYv`uq%tz-*l=CG
zisz<Seo(xs%|TQYem&Dx$~T_Q6*vzJ$5;4`qbt?exS~85JvNz9`+8%OPsxuhKl;0#
zxA_NZbNIrhJlidJ*g^pwkGqVo8Ciwg_A92+U9pg_(qedz%KC}pQPet8@R=O!!YB{U
zSVwc$d?!x4<?l+}scW*MR*{Oq5=iuj_$#QE4rAa^hmkap?1bsdBAo|UybVTo$#^v8
zK$jl>rNeM0TDP?higHK}TSJrD@Dcq2DUwhzNfu$A%a$Tp15~MeNgCXHCN>YFe;M|E
zL9*5ISduUD)6_}}(y)}guqmm=gD#&FSK8jV7N#B`63o+c??~0A!a^4sR>30CZdr$s
zg&}%~Uobl0u-k#u5Z4HmQ98$C8OEb{TyDb0)o{MLqwo9RYSx9wXt01$X+Fqv`Ed!q
zwoK#6`sqyz1uY)N34{S#0_=81d^0OOrE+M;;`n0XWBbJ_Vz-=ynOQ(G%Gb0NE%R3c
z-`E*IBANjzA@(ZIOg`*{P+%}N@(?KIbdFQ-x3kZq+0XK?pe>S<LS}j1e+&-<ky?!B
z?1-esyqSU8@=xQ^rbNqSqz^dCcocX|_(^GN;i2!OHq^fo?YCwD+{%);Es-v>!DQfZ
z6o1k2W<gAiC9-)!JhzFjj*#`U_(KFJvK%!I(@8c7!Jw1FplGbPX)tfaB75`)f0(W0
zoPEJZ_henUMxw#B*9XSZzF*l3cLh#g-jw@Zj`D>y63rrRo;vtG9nEDC5kfNKGS|w^
z3b(NeRs>l#IIlWy8CSo72yH!?w1xZ@tRXL#x33UYXV3OWau!k0PzuGc_)UT!YiVd4
zi<WYE17gyjqj|FYu4Z7kwaIB9aI4Fnu(_O7?y&6kBDVv1XEIVfEIkdAJ)VX$cyfsI
zkn7vCpA%1ik5e#8&kM82OoD8W&(wn_vhx`bHC&vRFU$}>&#fB_y{$EPD$8mb!91yc
zUAcynmwfZuIK18PqIn27BbfvW7K6c_*qdKsNw#?72QH|3A$T%b+T0HfPsG^QK9tSU
zNg7qj&2716k?@+TeLb@E{^g>s6f*DEPl~r|^YS7Xs@V8sX>2eVy3K=)55E@o_i!J7
ze7wyuy(99t<<)?C*oqPDd%tqHzGZsOv9VD74S$hD5wE45A)h5S-~GWMtB9jxcU~;R
zsU&IxZZrMfCT>hKyA&opP+Qw^)EK13V+?<GkC_F5(4>QT=idmS_n;dimJ2SGAU8MH
zJUp7735h$5e-`e?&gG%_gu3=2>@TMA1jIY7xbKdk#@g4_1dP`&3J=WNp`f+jU40rU
zGj><Z!V3JttsNm|aCf+|tS#6wBO~J_IDF<L(Ye4pJH+|xvZxx_LV0TsrJA0*myd4v
zhmK!RWc!(=-+o1Pw`>I8R|fz(mWV4A7kGn{5Y+AvxSXrMvN^Lb{5Cezp$}pwOu9R|
zJn*Db@<Y{ZKFx;oeu7RXkxDSrn~WOGgy`Nl;|Y7fi5#3>DA)+tO6i!U`K82AUH;{s
z<Bg#)7lp`0Xuj2qPR_;@g`16?QB{>W7v*@T#*}CYbhH$Z2W7u33mE~9XfB)rOu_|#
z)@KF4&)<{tcfdmi6+c-HA9R=mw;cb~oK=k>K0(0WGNxKC!*YSCmS664XGYW_!wMbz
zVSM^bfy*Vdc6ostyw*mymUAR_>w#=vwq9KTj#?_qd@ie<yc2qK)@AS%hD@s01+LSp
z%S&WcQEWs#mJVW9OdRSLQnIHI2x|u*kms3n2_Q(8M><A2{43*C4a#G4yUv$E*jCF0
z7b1p2uJ|FI<{3^RDG~AjgtGWsx$Wfdb39|BQT(mmS=7{t8S7S&O<5>7@^Ok_;Adz3
zhpMbX_blwMm>|=WO#*OA#!Z(o=H^DuU`NCEi><Rl6D9t0*6P0d;lJOLXr64@at4`d
zdL}0^>9-Q$2Z|TLX~c4FfgCMj##bFX);o&q685USmqDE?q~MTbN7dhb_fbF>iP)BK
zx}uy4Fcm6zFcw{N>!h#mB~>@(r7gY+HA&LcG@iw#u1}7xnRC95zss<8av#bWt8rB{
z`iPKmUNL8-`L$Pz)IT~t)_>@F?Q};G6xe8C<&&x6Myi-IO0Ug~Wuxnu+wkky3BBNE
zl+*#fQ_)GWC4O$E@H&uyS50__^xx%N019lhXx+P>Kl=||YUxk?)6UF6`1{2G{!-qw
zG}Xu!Wekge9FnvQ&x@UWrA43eXiO}Cl4qam<eafyutWnX{3@%#tIWn=E9JcOl*6Ju
z&;202<@d&BGnKnZg;-$1(3=rTjOvb$`3`S#`~%PaRWc(llU#UN-j3G;_WXK!T{$=n
zB*iTFr782*`uxdR{XMIIKg7Mb1pXhoTV!M}U?;Zxu8(Nl!NFi*qtU&HCjixq+t<SL
zzabp`3zgmBx;g+lbK~{Z9|gGbEm@d0cC%|5p#^0%_*NxBl)G`Oe0J9iVi3$qY-`}N
zbI*c(A!EZ??dUZ=wQF3vF!r4RTV0d}v@||%>+mRhL_iGgfwUirKJT*KkWoE{x)fj_
z9;31Hc{SR_8h7Znp!V7Iv{|A}RCnFI<MdX@W=z~uPw(n|prroW<7ZdptJ?icE>>)V
z2JI~X1y3`1)@zWIG;Sv(Zd-Cd`F=vXM0|*jiFj0$hwb7?NcHyLV9|N4W-+A(JKm%p
zvOvIPI;!in&V)uWO2MI^arkg)51;P>s`!X^Nj_~|O_$LnU=rMbRNWdW(kiB$oUssh
z2xQ0o%y0&eVc%R?7#hQCR2(09>N)^XFd6=qSI5p?wyKTu*Ab)r$8EyGo=I`DB|<e$
z$K&>Fep~wbP`|2@HcW{lpFt6qn$_bSSp%2ou&;*^F7uj$&L8W)E!4cAOjbiYYtSKA
z%or}tAbBG<Cau;@lJ*G2K%nUi6L+HCC?HaaP!5A-d_z|-)@?=$7;(BUd=~OE5yAU=
z`l*7lx-KB>ESsuBit2*jyg-LUCS_HAd#jJK3z{n_w50gLO1c9f!k^}WPq$@&zSbGE
z)bgMm#-PeT#=A<2F<5izcE0ZceC)Cp##2vB1?DC!AqTD=>Q6G>=2&kNb@;Zv#1t?+
zL!u~erqEJZ@+mSguR=dA9E5p|2hC}w*w!j0#yW(;723uV*w1g^=m-PPK3?#>;Qz$a
z&ypSn0Jtzjb8ak#jj;`^=Y-aUz<G5=Zw*o`b<%-<%xScfNL+t3Nf%quGA@~*&d5}E
z$lMrJt9aeEvG7t`*)~=2Dl0W{kFiR59LY9hO;7hl{bLJp;Jy|%S<nL@vKp18d-$bG
zZ<Zk*o^p<BfA)m@1S2N{l3a1{af0B1Z(`wWngH=|^D7pHR}z!~<{_*ajNCh(19$wm
z8hokA(h1$BM%~fP-o@M}&;o;D^C*T?^SF>*czPZZ1C<8kqHSZ9O%4MpTCAoUHe{<*
z6Uzo?{t$+JzQGMcM}m{rt?WN{J<rpOT*7Gjdbcl<-p&-DoyHKr!Xi)0QAsfgx-Xnu
zC(Xep<>Kdm3ckL3I7j<P;c+W59#3J_8xPYm)#h-rF(G^}jh<GXr!@HgDXVF-p2g`J
zs|mPYS|0si3^$g1Q$?%H_Xa-v1-x=uv4&_SFnZDWL6H&5g}U%j>EU1F$wamlSYEz2
z_|@w9^6PYZ9R(e^8NrRULzfC)pcFbL4<(;2XaCGTZA|;y?tjU=|C^BeKL^rYhR`B5
z?nt+}X3YOJi24@{J?g*w$Rlfy4E+D0b0z%?K?(jh^Q!NAx8Rq^|0BiyOIr?g76D`G
zVy{4#8cLkcV;lO!BaD4&Q3ARKrIn0H<yfvQ`-%MFr#8;@C1Jrb*fs}5WIhPaI6)J8
zVAa;Q1$mJ4>|*;GON!6BP`0v-d*3`9s6h-L(f~!$k%3c<V67{U_p}l>#(E(^O^d;O
z4lbms_8W>-;e|5WGGng<M7#+RP5pv9@6BO;X8LBlXVl+dCG=(4UC7Lxe3oG-JUW9@
zWA(e=NwSaz%6B$z3TYBX>PAkFTf%att8Ub=3UgZU3pi*k)EdqVj+aJO@y_XeMm;E4
zhC>IcxtRcr?pYRBYa6A>fhC`4b_)~3YKTBdLxE5$6ajmDhRqPh27dZ5$mXrlFk@p9
z0m{NeuXC@s?fHy*Fvoy@Rzyu7mn}Tcw=alW+MaR8Bt$||_o;rsToEmx7Jc>s;Pj%v
zW39?>6a+0R<O<$OCr$Cv({Ud6Vqmi6hW1Tp`O=$*b={gZy3Vbc+90P`Uug^pp2OkE
zr{|sQZ^8RnI`N6CYgWtuhitX~mHYdjy%_&PXU6{%wB9w-@vh5%&mE6{@zP#mY%YHQ
zzWi6&=i|T2K7B#M4@5|I#b15?1vC{1^>P&rj413nvfb@^sL-fj&ARJG<jZ~+!X+0y
zk;hXed@w;WGr{i1JdVPcQCD~i%av=LoIbM3;0jiwVA!?Y@<l%Pr}5Du35Q7zJqrqF
ze~=EX8L{XIyD(?mN)hE<FRv_83_N6%Ze-sPp;4%GLe_3K>UW_C<(dHp{hy6f_2ej9
zYMOO16Eeb~{jvukCKsBtT_P`XPflZr^;%>If5n`T^QE^xZgtnAAhGUf0d;)w1wp}|
zMvaeiDEp;ZIh%9%l>SHG<WhU*PQFr%)!sM|mjDlW3<uT>cMn8%dN_x<D?~lV>?%)^
z-&rlqg*UcQ2NA09I-sk0(TwiW!A92pXKXqJ8_|Z{<})O<Ix3*1tV!sKit4{GWE3lH
z!@DE9G`V<i?8ti5#_Z10%ggU%hgB7EBZEkb^=g8+4g|^=af{*M7XszkU9rjS-lr*j
z+iVI^50&hoA@x}naW?@mpnf`b<Fxzn*yz~Bi?OfP;s@?J+_b-H>OcPDl1E!AB9ih4
zqGkfzX-C-bF}Pg<n;OwwDP!-6DpF(hJ;~=RDycS@!Z&Z0B+q<gC=h$KbVX4Irn3;q
z(ZR`cTj`Z+(ukG4MAI_$FniHgF_dyM6cnCRg!{k=w|K(o_<xU8`2QF8{-4$7|G5_Z
zKc_fDrLl7DvkhixT#eqrCHE#Z{yjbSJSjh$TaGW$5+f-8>c}9Y%-}d)h*GDCin=3h
z=Eq!2o0{))an`23ph(#zvZEFJmJE!4z_cAAHOa$?ec#RwN<GiYxyQscjj@>y)8iGh
z&pAgX8b<^*iKN?IU_{V+%{T8z7envn@t1ONv>n!-qWVhuK3ZslngxK&B}A|6V3{Z@
z0jr0x`a0A0oyQ35(QJ;56@t_ldPO=$*=ULwD%S(I)w7P-Cbl~4ez&StyhY145A<~E
zzcQJ(7KXb-FJ{yzV(|nP_Um2%Su^qQ0fusBEX|8)CbHc#QxaU%yFxbLz@F0d_+dro
zZzU@xZ?B;m;fuRKLQS$78@&)8K+R{=>KZooM$l$*r0Z(Q87XC_j(XKBggj=56Jx-y
z82DDgqItwOwolL2Qf^J&l_;=Rq`X<aH)ZvHX33?qpN&cYBzu_WL&Ux%qM9TPE~}&t
zzPee~XuvNwbTmg)P=hQ;(l7y2Gs($tNfTKp8%?k}P<M@2Nj-5aDyg(|ppvyQkF6G6
z4NeMtXZmt{ZO3_V30=uDA<V6JOr+(mIMr0YbUGDbN-12C4Mcx;G_lG;(zmd#Lr0HX
z%9>t%e7&b3Xv60;JX3GxWnEdF0wWyytiX2?WS3Vi4ev_c34wlkBDVRvIsNp_TX1m4
z%#aP#3dt?cQzM%ll$$8vday3?4_$uFpikd&pk~%|82X(oYG4J%&j%I4%9`^UUVYrO
z@p$e@UK5ptC@UM0nX$58wYPUDvtpm821`WUxJGcMKQ>{BeJ9nqYu_cJ=GuptEa^Bp
zo-m}<(NJaxp+r;Ip|l30#%~L0ET(UnPz&N?l588A!mQmd0iA@tyD^$9D=23Rz8%TW
zwO3(OcjCAJ3BQ6_GYM~VTvM60fp3a?BnJ_=B5^ujnYQ@4uA1nN*>hNyE*kfk*r|F5
zV6}mX^xr~fdCNWrhiDA4_)Kc}2U*yEwtu(0FL^ZP*D?EP4~J<9l2NR3qJ#IWs=<+Z
zsqe1F`yM&sH{4M_e*RTODHyaQse9FCh5=*HTAMf-!GYK@*A$gBWbDN)e183|oLlQG
zDe<*jj#f^wK{nBM3C-jYM0ouA+)%jN^4lZ_2*3N8Z=qzlxc2bN{PYydtk?5L>2W+)
zPM(S41yr<ZAU;q~xEgHlDDQo#6CN;OXQ#>A6pstSG6Fo%&aM-awtKY|y|<s@vF(nB
z`|V?PW|C6huY6D6$Rxm(df?GtTv>}>%h>=A=V_((GZ3z=WI{It!dAOelOvlgd#&Hs
z0#|mh^tCJS0U;f~qKk^0D<}mX4H?dJzyNSk&dV--n4tvh8H26<St!HDaiLCM5EF!Y
z3~>|ss|yHx2ZZI(J3m-OjKsgX5ah&H_PqO%z5II<vxj`|E6(XjRUNNku8tC5s&!;d
z6tny-sZjo!ygaXw*xpZ|a?+lzmw^`^7VW_#hfDwj)iTrZ=%wiP$B^&IpLiA)moTIx
zwv}vfyyYDyR-*ZGEb|SkDO(N<);h=PDO}uTy0Z`Jv(RtjdKXSbNbB8NuNM_g`d!0)
zkjb(GsROUToa&%^vR4AyUn#~mzYFd~PG3s8Js#is*#Oa#GqVvQpGBW^$9vAFUGejM
zn(7)G>xmqd;F%re7LM76Kcf?M`fDpsRWq#lA3FCr{v}<LpaONdA$fn@1QP%tb8QEk
zlaV3-R<BnR17sabCTI>!)~@eS5YS*aKY$2*_DFE`amMPE_RGSQU+q^#)a&)*X8C`E
zgV7N)@P_(J_9QRk<9L{%o)JR>Ucg;Fd9oaAgui(kC<Kc&Mh1kV%}IfT5;*5c@%}sE
zZvH<=)Z+e0hU&QthawK|r04g>L%cAgB418$!78)2(=}1$t@KuAgn}rMP*Kl%Tj)dm
zu1-d-gVs|qStBo_@>NA&#F+0IDDndcgsFv3>vW};%DX-kXF&8|)0Db!*id6vj*r_v
z912x6Q&Ljw>peZ&tp6|C-a4+$W^EUR0>z3$(Lk}{(BkfH!CgwR;_gsNTZ$BSOM*-A
z;_mKl#XYzd?@8Zxt?#V8&-%{ZN7f%nelzpTWS-G`?)#qW`k6d2q1I2#TCC$Gr%*4Z
zgWk)z`x0zA9jNXRS>yRt3#LeZgw;qHDRfR(SjfaE67quj3xAl@AdY{HrDJN|bnPwk
z(e;9z?GBm<vc@Z{J+F&8(#0|)l+)E)i$+W=tm~eD%xQOR9-F#Stq%su+RUViXletf
zg%zFp-M%ZzD$nc+Z9I6g>(mY38OvCLU$w?@XIh%E(0VA!Z=jm_&#q|8@I=kHG{Krw
z;_Rkn?L0&bbi-oMNe$YUW#t!=NHB%xfUYY-iQkkS57d_qdB<3_k-0|5cJ}k-2aufY
z<&k}QrmVMf=g1SciTIYZ>9hMan2+yw4BVQXqoTcR2OC42E4oN`<Icmkd9x}OkB9{^
z^C&Yhxrs^Y;_*AV+M3beDptBEc715IM{SQ3z@c2n;3W347p%9YxRFL5l##${QE)5q
zjrHH3FaDby{{QV(t!QgIoU~Xt>v@UU@f*RSEq&;3w%2TrK0+R<zbuGGe_Ige=<x^|
z`mo&I+SsH*_{>sMteZhQVcU+GMP*4}XD4+eUxB$9>EEF^J&W^nfw7+m_>?~Euh=kC
zXKG^+y-<uIWRuWfEc~3u7zvVk=X?d$pIFa0D@=KqXb{r0dog@PO{hCO$i|UbeVjH`
zVQ|d+OmV5GJd1!%n{f>}IQj=D6^A-!2b@xGXG~pkoEVnhAY)&c-7JU2`5Cv8EFM1&
zVG&J@oZqZ{)-26dMZv!Qd`dx|sX*i4s78BIt6NQ6(n5_<V9ga(GsUKO2i)p<X2AUD
zsU2l&7~7zs0|jZyygujC3I2M7Y9--(lp?DX__n&GE+vh1YG%~^DJ4@ICs_d<6Rutu
zOs?nsq?L)_>?-*(&tYES#w7KtZQa2UjYl$vhechOnK3wy3^SN)Uq_jSe^yXDt5Q4>
zF{oyJ(Oh$}w9zhwn}-L5V8uoqfbWIAPsq%79ouEKT>jh&@9Cq%(#gfRscK8f)FT$R
zJhVp2Ub@P?vVionWZat)q+LK1rHMxX%2<+`fwnGdoq3@xyGFjW!&^2I3}z%Rx--Nz
z`%cc^2>gpqw};RfN~1Pki)v^hmpk)7%tYu}Q^cJi=c%Z+=?Ct<Hzma1b9>2{1!L(Z
zkrN3E+@s*&(`Qy@V4G5H+dBQ*5EV@gb++S}9!Wt?Uv2mMn+13s%jow<tv?*F9e229
zPls0P?XN`%941_s&$(v{vuP-zX2g{Xz1lkc=xv2(o<#N=L1mG)aopCUR7=fB)4vvx
zRLxk+QZHijlOEI3Pxtftv}Zz|c{R)BaSJRQ+M)G*9)z7)Z->!Wq(fhUW{@g^s;~*G
zUL2&6WeaVTx~sv`mSw{klP7^}%r@)F;X=~#2!Nl6ab>mP_wO6dKiNQ=z9#S~ISkqb
z3tOSE5q*|MK1RV2kDwHy#2BV|c<7oc-W~a_^SZ`rOaVY+3CY&cPzTnR^O5rF^tmZs
zypHtcwyA__=fMh(wRv2$%f4#$Xup7t?vv0nqlVahiRfgVy}+N*)-ApByXxIbdlXK$
z#8PqKm)wn!Pkcv@gf81-E@EWzb{Hh*Fl{;$?N^}HD)Bx)yKoT0>$U9lx#wB0_o=tt
zn^7ObfSk(dpW3gBN~S!jq`yv0zlId#B_Jizc%LdDJ0CZ981+wFQjGpEd4(Q_<}f*7
z%&z5bq8gr>D;Yoa8n69D)V{NRiuF};I!axVS0s~^7a#hJ|2HUDUA4Y@{51{6X#k*r
zM99R<X2%0-O4(-7($1#3`r21(Vn#FFx#&D0PhUBf#eH3G;Y$#x42=sd*<&tgDL-oV
z(#@#Ywv5=XhF+T)x(Q%ClZ(;%5!;CUwsOopsjXJVd3%NpAYwUcKyGaLV_iuJkBcGS
zpamg-1|JS=-r?n%mTxN<3S-KTBh6Mbpo^4YmY_hR4KgF!yF7|lI`2x!UUIIV!wC2F
z_Rod}yN62bH_v}EnNjG@uqW}vE3ZyeE#JN_i96!P*I&!eD$h1%9teI*Ek5{?ml0pK
zqo(tI=f|{f63JTgDmyV+wordmwj~>rNWDnF9D^Ci=19{vv+4pHoSmhRbfqw%&XB{9
zrweTRvaoZsc#;0HWimnM18tLSs#B7d8j1G121Dc;#$-3dw__3Xh=vlQ?G!JpNQpTd
zby#GpYjte}fy~uiZ?z@}-9NnmOln?#;^#l;3KL9E;q`HUURbGAJmp_n9tU!FnKD#i
z6P8y_WkR5H=%c7-9HvEX?wktW-EPwf-@YGtp-%61$=uJB|5MbFYv10rpTj^`?}H#f
z$iG$LeaVm68K8<-T-gWMyHvv-e%0{`Z_BRNW*|2&_4zCMow3(ouuoxO$z`>EDhi8Y
z4l=GE5SyH0RnhI`gx$9G;DkQi=idmj%x2YqN%5#6Cgmn?tmR41M+VH)uO@R8^R+q)
z)q{sntvbS43C|`PX~Nv;E__-=w^{6U0YI5gz)opnZP|7q<87E*-nTL)VyrYwi}%?~
zIvMU(8gMO1{H`?}KgKYvN&H-+)r{E`bMExp{q1qb`x6DC5+zO3SMl^V>6Brr!xrY)
zM|^^Xm&1$ay9Ebo(1W}uj77pBRYmIp6vNy83s-ExsO2csDzp}d8NrtW#T;|#D)!aW
z#4TdFxY<T5%sIu0v{wSX(<c_e&V9tRQ)C;f@3ynqmKhX03#Jq5<oCm|i3vY+2>pCM
z5y9^}Ogoyh18Quwt2zIklFXP@8m<Yn`aJt}T85@?q)a1&@R_Qrt5mY<FGHIpk*F!G
zA>F}Z2SH$~;KyxyB88Z(ZEjdEexjFrhqG~t3<$6Ia2DqG`2&ypoh5-09Y|*pM#Yws
zlmU4||B6EXc^6oSx1qES&RCAwTT!QPOE+0NG+cc}$kU&rT=Y$^m#FO}Hg0o#&|W>w
zFip>ak(w4vgbz!nsny8FVK7eLQ7a+zlg4mqkQ%K!V-)$qW*v6fJJ_;-@8sq6;?N9x
zPHWnhJO|<I7`+x@^UK5xB${Yqbp5YG8k!6H9Er|)EVeAAAJlNQfK;+I5ihny3|+TN
zhX0%0|KEuUkJfNUGO2$!lBGPF!|cd(bFQ#Y#a6=L7XqBWQHEMO4Qw8YoUZ_>%ubTY
zV_RH6Z47c=%fz`K%T$Kr;y5#|1S&md-#c?2IhC18jQ}TUqk}p;W<BqaAqq<B$3=Tx
zO*NPX{ytdsD~+S`F?kZ42wWXfL0)9t^25Wi|Ed}=*Zz%AG5r^LJm|-tO9?lo6U-<7
zbXxkS_Dc3&l~@0y<oVZP{&R8vx4>_y8JLOyx+OSBhqYPnoW^@aRnPcERbx7K0m6!y
z%o6+2ITvP3d9^j8L25_fEUVnL?2MOR)Sl<&VgvBJ@<u3%E+{pXjg8s{jz7X4s^zR!
z3RNA$tUroOMD`%uJ##}#?6q?KpW18nf7x7cN|QgjzSF;TegC<p`FAD%Pt64<g83(E
z9*Hk0yt(W&rhj~0vmekbMMsmDUjsB?Z-vDdPjI|<$A3=9)x|_T*wWPPQ^h8xG>W~M
ztwn?Yhk*C|7A4-2fwqBzJ2eSbTuAMdbD(rT&{adE&W+i1A=f?iQi5_L(;==NxeZXv
zaMjin_S339AKt1xY@HC@E-`ii<xJ*U`!FAylX~H1RtF<h_I$4$-ZQEfDk2{|v=}aS
zPsIf@%bzBThnCxaXQ<~zRr|($D(N@UK)!@eZS*pJ(?d{MUadnC7&d9PyhIl6mmvk}
zz3TOxU>b$>!rGA5yeRbFvfBO^URdoPyf87$JL>;}6kGk9Lhje#*q0h5E>X7s9i*QB
zZ-l=8S9SB1SJ8C72vwm=^dB}xoI#G6j|Ho4H8_7l?C*)n9B?GVt1e}Q*|T)wd+0>v
z>H<Pt!?)*|X+^xx1a+A$49pcnCOH!K<uMUXX@d?OLY21?`UQ#cbzD91+N0Qwh?w{<
z-}Z<<U~jn%1}6|D+9O-`c3(4;gzaLthdu9L_h)j9w7tybKa}tK!8{!c7ky{1!?#%1
z@)Q`p$q^eWE@_o#p^JCQ0>r<xH(6OQF?dMv&+kPMGSd|a5k;I}v$IXs+!3i%=Q2!<
zjty;ip`*j8z4Q5haVo5+Zy0kO$pa@}@wo~(3FmW)(5`Sef7xY0Mxpp*Jch(GWr@$4
zi(D|Ni(0>lM$Q{w9Y4GGeas;D_&6HD%RrtqwNuv0OgR)&73Vj?W<C!KJ?3SI=?5<W
z9AHG@())%CPlVNV{r(?xfa`zv<s(kn<F~(AZcqQ{ETcQ`x19gWlk(rqjsIxeUHSC=
zu_v5zDSDj#R{gFRZYuwaa|s??Ok~-LF1eb%zT@W|13GF5YJF*GPQ-PI8i)Bg2WKZ1
zBg*iPsy^pDFz{;cKos91uC<GFx45%oKE?`~h48)iM4Z3GEb7iD%J)Gv@n>xe#t(M1
zcHjEs7xm5MgQ2Z4-i+Mq;e_J7!go!1QtmMlUCH=#u|iE=stONLJ3n!;DO7N)Ex>a|
zNO-jE2ma2gmH58b9!&#9R3E%RR;i8Cb1zM-6)hC&abX^CT`y{0M3z*wNDJ#a-7gYz
zp0-M@zjuf8e_UA;d;hOD)-Elr{<^W&k^A4-N8t9l{}ko#N9dK<J)|(um-Z6<qxQds
zJo6N9X_w}Gt!F@*B9=#SOCoPrlg@`n5f6#;otZs%DXqLeNzm3GP?$_Vi<3C`@Y53k
zIRXK}Z|FyuGiWlA2rC3}HuYK58;L{ee6%<*9z{7X+IHhUhj$rFqA}~s$AU2x)mz?u
z=3bydkV34@nlbA-pNaHV3?blGFGSwk^qo5t@F6vJDdhx_WFzOsqj5I#Ya&7A1iqF+
ztUnNO9-ssM#%Fj>^f%z)Un83MQn2oyfQP+*7BCK3oq+!n@bEv@i2uHOP+{e8OI^bv
z{IcvHGG*O=1pnq%hO;S~!kJ@-@2Vn(yPTYq2keRK@?Rt^q_%T$1*r)ENLLa98P+-c
z0SK22sN)O*haz)t6fd&mmzv`n%R@S0aF|VT>>y4B3TJP8UjFa#R^9cNlbXvL05iCQ
zo%qd4N^h_We6g=1pFu&$R6ugF+uQG~mBkoy3kf*SrbBr_P3vTSLJnYd&pxn;h;O4_
zq5TAd?^fCg4DSXQ5T{Z#Ys47?8NL_rsVW=PSg31VJ>><`wSoU3-;BdKh327w+;Fe#
zF&|NvKVI3Yj=c^XcFMySHUAI9xPK?w4bT5I-1{ly-&vP#JyNea{P{Ag_3b|b{Ay)=
zagCqTL3e}Hm?oHb$9=)lK`Iv8v!-u7n=aF164_dD^GbL67!zgLh)ox1E#Gu{(-9oo
zK57xk6`<c~U&lka%?i=nkCwI#3&0Za()}D8s9Ca|Eigl+_lq@-t3P%KWX+J1$6@{S
zt*vNPj)iCN5y0ScHpM7-M$&wC745#LS0&s+kvVs8>7KaM_fm|dIZDY5A1EjP^T9^`
zMyZ_m%NV>UHdZFE9~F&HEW<;Q0VE<9=R&mb(L^Ag<DsSD3`R?pIP0<Ky&ET5VLv8G
z!4-~vWxu0XzsMB7UL5(8vgDyp*kM|#ODEa1+r6#3-~fFA{FTM}<>r;i4%rw8rIL~?
z>q_*iFZ5u>S92kDy+~}r9YQ!6_);{##~6)&l!tak%%FI(egQuE(yE?x$D_4CdU7DY
zDu#QhTR<T3#BT(Av9T>scq1-Ph8GaZn=hJ_RTce=PAqm=4oZIm5z{nie^TZo);|Lg
zpV7|E&lhJQEZk9YXpwbYH2%o&SJbBZ+M3b=oQj^ay|TU~G&O(-oGlXun&#dEjDdVr
zmz^SNjB<I!%)rmFv=5^&5w#=QsTrQW4%MJr<3OqjpGP_v%qI*|H~|ZLcW;oi=a1V<
zK|rQBK{lNjHrJRf8c4cyW>U%*_HKH$D1E7EnLxUa7k5l7TE%Ai&Yg(mH$nj({I$}b
zCsNqgV4VI;enic8Qw~(V^Gj!R9)10KAUEWpg|nGYKrH)?ruxHHWTRI3rZN3Pn8(5H
zkSKU(kf`c=VclNtrXVxDw_^<sl;2(1pBoFmm1i)saWuP}?$P|`f<(x_lAnL+5e=6W
zdlIZFrc(UjdX9<8xjk_@rjmEL95<t+R>@k4#F{>YM83cb(4%DVHwcG9XVl*U_@b|v
zx3Vvfebu-=AwkG0ek1%Wfh-I>GN|p=mWcgQ;<$DVR5&<b+;@gS<z}4;Zm(tn*IC;Z
z6doB2e0HJzk4Zz136M#*I-tUgRy2+yGPaj(SKWa9pC|Y<eCD8IkBcLZ5s)dzAK-Xs
z5g@gF`U0X|94V3#xHa&|=j|4v9s2{5@0v4ErB5HI=-U`udDKC*8=~Wf$?+$$4Ghpf
zWNn)tc;r1`+81n$ZOFi<Sf2rR4*e<Ad=#J$N%^DdXSiH)jEXNVA~yt-p|NE6R?>=~
zB;S*Y<=+<Ft`}REYjeacASm+HH8P3i!{=MVl{RffsD*=2&SxO1)+%>7(&xaj=Ox#?
zriXM->3RMeGbw!%mapJzWBi??@1S)JKx=`5GO49qmD4xGE+{9%pc)XXAL5o3^BV!Q
zZ?rwL)b&Om`{X`MEPqMs5@amDMMQ(4l}*T1jo2|Pf%@g*<YJ<@t!qK7BL2=PaLlS4
zfg$V1*Ql%)UmlKYl!3OFQ`|XQFeJAR{c8LN><1TjKI_Ffi;RaRl;4aXPmX;1O9{#8
zL@2#_Q-q0cXtb&6V&Y_-(LK^V*=&yF*{R@%6&v{42f_E>_;a=!_pKmO;1_;jOBLi?
zL<uEhGx06s`=MR^R6{l4@2}bIym#$;aMEUdIF^%d+X7ayK1lE_`e4=zJuiQ_gwvy~
z0HLj)D73wFA6cGo`XJN3qVVzj%K--UL4;JQ!<vSw&%W)IW0~E7y6-Ao*M&bgVw}#F
z&h4mB1y0=;K$9cWaTf9JYF(Ep90<xcwM(9KC#Z_+*x+E6$(^k{(jn)KV=xB}8B^e~
zhsK$Zm}9}rZ-k@)_$O18H1{dK9xQpjEOy?)bXsbCUVjTw*)E>?3CmmG`M4uT>*E?l
zrt|5>D|*{&Vj#*S=UZG2)eHTHvZ^}>a6!f-M4JO$#KJS`{#VNO%OI-r$n@U`j#QEY
zX4&Q>vWZnM0@pPT#Gr4k)kVW#mTM~Oi$X`ukX)gYJ$!+{+?-49nUuf@yepxi2{L2R
zC{Q)d9bKSi`3Qeoh?t&Ws{hRjoH6y9_jK{;Yk}N;_Z=^}{BMM8Ir#HaymY=V4c`M@
z<KEs<0AU${U62xha;s11tT(k>J7%Hgu(uS`>8rMEv>!{fmu%AtHx;iw$^~$1^^>Rw
zqo2wm<vu|Rs;J7Z1;M{YtH)=GgV`ibrOS)p*24~yNEH23$?~E{0>aZMJXx^?oB%@e
zL)xDqrfCH?CT$BMK)+q*AIn6Y3r7Bg2X`PT;EHzVgo`MK<s)$KlcXGS(1yL)+)i3s
z$iZV&LCZzIP5S=2MPuL*-5o^PfJ$pD>bt7(9C%O${y00(=A;&A<01yNJM5#<hu0>E
zb{kxqJU;1_=gmZT9Hn*8<2*@R4&a2eq+>(4cb+H$EMVq&Yg^|brp4c4ngib%ce`4r
zXgE*uNvucU3_dP}l<?o?h{cJh<_nsT!8<Ujl}w}FTM>2JE4E+MqWoKOAr+Zk-z}u>
zbRN5n+g3Q?t^gVmHUYaBPK47&^@x&%KGg+?mOc#V{zeE15+%zGa9ZTY`gCx?TQCGH
zZ<vwlwYjp{5o~*~BsrbQnuX}7nzKA@I#H5!w^^DUV7GC??wIZZ-LKTn76Wt}LcKfe
zfZr+tR3A*Bi&yW6hO8pMd0O#N=ceV(l&2ek8nF9T$WK_2549+i?B|^>Iey?pQsCeT
z?F^S5mFkHlFi1&k`(tI7*<_&lfuQ-GhgNeTL@QD)faMA#icob6nD)(v5$xM4Y53#G
zn-*2fyIj1i%NGsNRvOyBl5Bn~)pVr9@PG^+_rbn6!;z+nnZy=yb!Q)WN88@m`A=Ma
zf#&UW`(|=HT9pFtxgoz1ruKkP6kg#YW4`c+K!rOt5JO*0)#Jw)a9J`K6^ezrIFo3?
zEdA(MeV4fyk_Ro0{+vo>cu6bxBvAUqkP9$oM52PyhcddSCGBL*r>Id$2B=>aUY6J8
ztl>qU*B_3Gk--sI@B>7m$`WwppjMH<JolFJ-w1D!9?J75p}s`4qeacdPcgAix}>R-
z{-%av4?jTRzPLW3#bk)WD$Ayq<u^hTcjc}}sIS3$ZIOB(3tleFof%u7S;Oo#$G_qE
z*10aS^XhNiTxooiyo-n34oN#ty@g;&7d(v8RF&K$l?C3~-lY>>zCk`G{<8`s76HF0
z;M{_`&$*J(;cz8;IA05&bU{_+R7FqpruwSG?r-2k;K}j=WRi265@Ly$GK0pbI9a)+
zNWT$YQ3Y=g9|$f>Ukr%p|Kc^zijo@w0dPCHTj28=H;}d&gqf%I2E=DgiNXI=Ao5~g
zFrsR-;cl$qgTho!^@a#JR-rc&?H%aSW{BG)IDZ9#!fk#o@Lh1!KK@bj3gW&<5n7`r
zx2LXA(1C%v<s}0U^p?Oqm4y_oSeyJPixm$;`uc(DFhorrAaRu8pkVG~<3xH=#gEmJ
z*K{~cV*lKVto_g)wf0@tEYi}Oald~-voXtlkWqL1G(!DTU{zQDh+}El?55S?V=XYM
zJopi~pIWE?(6*SE*LvOVo4CsHegWrw%S6CZv|srF`?fV|iyEZh_<Ny2Rv&4=(g~VJ
zXqS;va6XNQK+A#$rn1Ropyv*{RI!&GJT{N?78dP!jHxKnBG(kKPU?||?MS~d4TgpX
zJuTIztNlhG&^-Rx!RQe>W^UyCdZz9=4dQwHdOEO2>%QQPvg2B0Pj?G%4b}ZJu+b~t
zR`+HQ;(k9t-uTtQE{gF8E_D;D1uh17-pNQg2a5<p6@%y)&*txNqx!VO^Y`T;xx}Xe
zVKSf%k<tl=0Q6`lyK=YsEjKii2w=HOBOg+*i%*%UL&zf+pDV}Ut|l!y{ZJ`2b2D%z
z<r84JbTQr06H2*RYbN5<qJHE~a_#IIf+M-FC1e^XQ}rcJ-HRg3Cv5V4Krj1dZud4e
zd#L%B%bie?$(Js_5xz!mwehB0iz$BSh-CQigfA9%gqDz>`$gd`Ds7H#`%dU3!5R)g
zYe&#WVn9K%)*CECHB!({HTqQb-9zj*g6Te_%eHkg-+HuA$Buw;86ExAkpRaAV?J}b
z^1^;2WmOc1*@e~=vkh)$<}=@DU3WN^owhbKU4lIlzR8e#+b_Z~5$N>Q=`t<#7eHSQ
z{K8Xh40OB@k>i`dfS{`ky@&qhMz=b0@>e&eZwg|W6;yXLO15^HH~U(9WA)Tl$mQ~`
zzeJTqzx8oH;XUp3g)rr)Q-VdYN&L>Ww~pj71YUW(+eIAHjvh5J7V4_zJw5WD&}74>
zws!x7wlKYSPVa<*a<vteQ-#=A;#Irg<7x3mO&E=R5f0z%=7E44K#8~@LlzS{1wXvS
zpwoWn({m4JY`s+~4iL#{?GQhG)hZ#|N-8Txcm(t1#(ewrgkIEfzfUhfu;in=iE(;Y
z^;<Ox`IA;3rrDbK9dma?a25RctU@*_!3gVEf{=&yMGu=8>-+g4As+1}MEw8kG)J9O
zH=S(fS(ua**ZRRsxi$tj;~l!KR;tK20x{c!!Ep^S36x{bN905X#eIs|xCleWOE#DS
zJ2xhoYxd6e#C00Z(59xu{ai@_Q}vCsHii}`(2@@S#+^O1Udl6wZ%D8?bbL;V*$N@I
z$fJb@k@P^7@bSS|4(qnm&&w1pU%2^Ir61>%<TtD9m0UW{7};hSD@IBO<4Ix2lMlYa
zSD-^cI&bcObWbJsitc;_ew^~CN%j@m1~-xkC9<;QiwMig=LbvUCn887<DyaDdRl-J
zeKwmv67PM4Wccyedzltg7mYRi^h5XZ(xYq3X@ovAmSAt0grT{z9gBSR0*s85m5y#E
zFrs5u=N-cJ9cX6np^b5?R0_K4gy_N;GZ8mJMoA8VmSNJeMLSD9Z`VF@P|dwz&)?<l
z70(q<$YQz^X~XO`E(%SwZKl=EKauc`&dn*5PPvTlP&f_YM04g{buOxz>C;xCj*=gh
z0GL<&!i>CbEa#}3(*AmUIq_QA(}G89KCJOh?Sz^&L8wRxYN!Zcq}qofKBS7G1~AR<
zhY7TdQ+CS-qP}@Okz-!&GDGGm6hv+`U|A8FLGH4OKL4xH2wR`_!dt*09~pB-6>)(~
zi}w)KE~7u1XQ=e;&V^3kGQ&DMZK`lQkzCEVd4LRiT!@80qv&RC-_HQUXjzdAiYPAf
zOnBfMpJI+{2z97qoAlmUB%2M@joy>e#o2fBOUk%~d^Ypn2$)puM`%g3I;~M@ni02O
z5f|lslmodT2;L?0#?menc#U~@Hy?F>vh#?W=kcGSc3t6ou{tB^n^Be>xm~SYrm7m6
zfE9Js)KE<AxW08My6$VH>>13Z0(yH%-9q9$z`g+i>1_mk6HH42MqWf5W+wZ`O<+xq
zsRw}d<+n1iGKi^VPPfaLMgEKJCv}b~k)89fkLl3~9v|Tu<8Qu+hAEQ{qBw{b8_6o?
zX0i$KJG^?5)B1(Zw4qh3-s$^k(JyFh;BHJrw6*_1j!HVe!M=f&bFM#&?g}IN2V;Bt
zDfq{Ud%T~I-zLmWEa%GQ(f#h8Mp!oN5F$PCgQ7eyeQ9WwN^?$$uP(V&w1A4y_TlpG
zp5cs4ke<grgn8fk`rNB;dYN@)MfK(`eH+UZ3^6tTE7khnCt7@@jn3B8>31zlb8=$v
z4*(i-udJBd++{pKiZ7-P4*yEF|JLschNpA*2dE}$mkQfjoJ1G30%vabudc=L;BD0_
zB|uuQG{aiZD_WpXUL?#A<J~vCyR~c$|9gv%4Df&0aiwA_`i+1GFG6KIQEbEVL9{^Z
z^?@+F_Ece;dl2c{jF2Jb!0v&Y0M&AiKixv%pI-7;4dB@W2zp(C{>%YIC(nO%7A%5K
zo&N5r=<c&iMbA-%lVD(t$lA({*1hHm*12QV7_hwZ1}qw;x-cMeT6A5|OumW%vGwa{
zRhjkqQ|G_?m!XwNbo$p9zje(cpf}$@<DXvX+wi`l7qX`1!ze2PdW&F2>Kegw|NSQB
zXTkya;HudBb?rYHT8T{Hx&Nh>n!T<B&QPj@Xzw=-+Sa?#|3jIg_6>sUKd6*i0hhWb
zhxqTV0Z~<VV)|4w!15mx8~g=7G+z#F>~G(xObuOT_~-o9_;pm#*~Sg_t?Xc6?y@jM
zp;i5_Cg*<$z&{!i{S$jK!KN{aUlseMvim7+ruVTCJ_yNRuN`bmtmnq%E|Ib&fW!rj
z5c1vg+Tsylc~t>{5(qq_t&EKy%yQ~}YeagR=(XIcHFH~Zb*&rGyCf7lq(t4z2Ye5D
z`pl6HAM&+Ofxl{4l<gxxdIbe@q9LJH3)tuD>8RyVL~W)8D#D=K+VIg_uB#%nL%eX|
zUVJChD07|clu+|$N@U!?1nDQoQsmp<`1H8=r4oMPSYo+X>coo$a|_%DA5v@b@&mv?
zs&UV08;$Hl!q5C??ze{8JMh#=5AHrbA)!ZlD%-%9tGRay4~E;ux@qtc<)X#P)i$XS
zK=;J<_xIm#_rn|dk`$INe-8c{VeSY3#~PQN-2t`5{=7hZNwyI2!osSqol|j5tL9%X
zEJrRUUCve2S~j~XQhE+9hc{4V@5BVl@Zkw~uH|BJl?nDR4WC_+IoZ1Uiz{r+hTCj^
zO{mD6EDuoWMu*%#OX43CB-USoKjCM%E%J{?vUN!(T9*BW{0si-OP2(9W@}YLc2Uh4
zA9!QKbI@ywaQPQOFEkGh;fcId(c8d-KSq*A_^srf%&BC9M$44l&vE*+dZf#060eSq
z+nD0Cyut>AfWddI<m&?N<eEwEJxd24A@?R3mGN;IBA>3Wk*;1=i92{8v<psB{Cs9C
z&I=G15XZkRp+}MXM9Q8C@XLz(=7s)vT0gDDf*z&qAS-PAi&h0BW5MLzudaH${H11#
z?@qzIbw;yY@<(Qc5XNZ7`hZtYk(zYrwo@<N+rrZoBt+@JYL^X{y2`~96+9zXGw0q0
zos{voyuwci2lRb*JY;aVkOb&uRxqO|E^ob!f|BE89ZYetCv^=)nx3CoqPHumj*yL*
z*>L2AM|HSHRy;qmg2KYjfY+skP23Yzyq@&f8=5t7fc9i`t%UUn47oO~f*%56w9O&P
zOmNqzNysmakFsAG@EFV+ktUHELTEBX)@MT0?7nki1rL?Nj~&W7A1%eu-2|R@ZeP3q
zM#wu5<A!{#kFo(P6e;C~_+iH#erI!RNoY#3=X|MXOwWFc{W+NyB)_<MoIo<AGq#>{
zLxIk?kPNu2TOP7Lop`{$EY`#vyOVROTsVeEZ9zq_w7I))b5j!hxVp}8O7Ex0vrUC4
zcF6tQtDuJLxr)`k%(V4;R!Vt1Uwsu&+mA6)ruk#nal$}Wp{mp6r1-_TX3VHp7^AD(
zI5$mKZ(ipUNoUJ?<8@PJZ(a7~mlh>Z<dKAl;#EIzR4_&%hu9WYcUI5%m8CW8Yp7_^
ze~@3|Hk3F{IK!zNI7q5#r&$pzF;MSoia!-C0|WDov}F(SnOCiC)-G@EY(!Ii_g{Wo
z1o6IH^sId4(F)k6s`fiEG&-b(`!|ViWbW}WQgE7bXikB?*o@oy5v+O+9|)siF9|Gs
zH8Et?sE;!NpZa0Giz=X19`2_+H%a|=Zt@FU0gt*Sru7^1!Tx4u8VRav1|I|&3i4I<
zoCuL{j_KQ16jE?jNR11;Z^tb586Re59?DA%I?QO{!kA#*yws6u4d0Gdp*F#aPRvkw
zD?CN;CWYn)y^Bym<H<8ZPBhy3Z4NUT1|~XhfHjrmL@cjY5xp`ON-;mescV9cBDf-J
z*(=s0zS4%ak+;>uJXz6IT4hSV3}4<k`mM+)M+&Gx$T$~B0S@mFiK>&&vGggk-B~IY
zaIaFPHayBI64*;}gGfKgbz&PU2tkoOC>pg@HYQtm44lxXHXGy;UQ2XS5HCz0>b#iZ
zXup`^O={e@9MrkKY;R#zX&rJJ<!HZIwXEAEUh^nFGNoEZ|B^86{{SH!$;i*y-MZ;H
z?-^O;qp!6JSomh0Pi|$6j@Y#{h4f7^a}4;df{Tl8h>KI!Q(LVh24VgeVUKNR1y@JB
z<!+tQ<QCqhh<&iryG8;Uq7K!3BCc@`M+9QcJdkvO-s42eZv;fKy`l4QQu$L<n6ddU
z*<TOHzY%gchHNh7J_$FrYSr!wDCzkj9)9y)OzT?8O;-Ko5qxwMRb6xuEL6zep=k(|
zLQl)r>LcM_;qZiA8Zo{vQ6wUxBX8#*?e#u9X;w1S0TLE|@pqR2yix!zF7h|~LiB$!
zvc;+8Qrj4KspVl+Ml}oMzqWR@Vk5TrW!FNa{qf;T4jzP4JVvpy9em%%u^5CV?XIQi
z)SJOr=safK>F=nd>zP)Dn~0oeY^yr5zUGhQM9jUmTBBhtl4F%PkRO*VMfaUWeM7-V
zCu}y(W}!K)u*?7*dpa_oc2NkFwL$$8NSr-k$F<_*l~c4WeJ01Ph#6@EXVcRY2ArOO
zjkTGJ6(@r;Qo=XRkgC>w_<eTW8;hl~GXe{2qAfTDC-?K7;F_m~XJhSaKykkp0KLV-
zphJ<bhraAZ>I~1euU(!JO2v*RZcY7L!~Aumceahs@NNwR;?j8m(q@&(7Chg&3Q=3;
zXd>rI=e!xr9R(yiq@oPonX4|zJprhYK6eU-kX_FAd|nYd-Max4>EsjZsxlYN;qkAK
z<CwwNXIcxVrPVpoh2=b@fBE@4Zj&WJ^&-9|5=QKPuQHUO*&@<{q7|x+G`|$Wyd$Yw
z;z);Z3z42p8f}1#33IYC?BAAK`&<joDEICgFo$h%aY5xqOtrTr5Q18`Lp-^!lU0ld
zncXHbNv(uwWv4a|-6z!h>3Fm)m_3}4zM5BvCQLbvXvAy((r8mqWqKzqz7+`-DJW?a
zwAs<<MvtIpW@3&3<+`-v<Xt%V&<(CEynyzL$ygWnZoIw_VFFFyzGe;{AuD+6*cZpg
zRegGJ)|zCkpPo<C)zy&WQc|l2zp4REOOw(1nkZd5DZ4*^LHzl`@BVZAP1d;*?~T7S
z>_gc8yGvq#xx3_?l&)8iBsMHUi5bq{TkK`(x!MHtkz)vKDhq_j^}C>eTkE=MpW9Yy
zMx}-xKlQUUl{YypU-8$OpF?~Yg{yMUAq;X#lfH&aO>WCW$G+Cd#d9u%JwkXYjf6m9
zt%@?AEoF7txV%z1#8)24m^=*J2ev!fA~9qixEfz;xG~>n27q@B_jh`;^ua8InAGU<
zX*sBq3=;~+9={-_tkn_nxiZ(auNo{LQ(4GyW3SH^eVZbd@}|~1zI>mSIT;iiJ0G@!
zDBCFwO2bq<iK?0NGjl{~+-?M!*~q4pqnTc}l1x+&yy%sV#zudTinJem;xRtr->;Kt
z8@r>fP1p(F1`&F#L}R{sU5ez6^ibIa7~hbm7RQx9{XpIM(aMIgRFmdN3Ri|3_A`w;
zf*#)kZyu*>-3N!I2IE+#fvpCU1^k^nSdLzu5~<tM`^~`YzKdlCS&OiNka<>md{BG^
zzj3h-NSj_cL5DunOoEAre=t{U;1IAvzf0CxrpU(T4(1BZ2Z`g&WZ(?0C$l<?>%iRI
z_Y`}CnG?~~`d3I0ZmN=WySQ(50lpcP-UXF1qN%f96%34JYI^;~G7_@>(z~Z#^pn1e
z(}MC72I5O;wxHh#?G#VgoK;2;vVnkUO~>!lNg*=^pB-dS!280*l9NAjr{%()<I697
zGE(u}%KxUH19FiqG9TRvepMu}`T4*KVB*WAO*A2`*%S~nmZn<_s=ZJkA_`F9BvNVC
z#2=Ip>a&Z{7$~zViq%x%fqk$bRSs3BHbLduIr()(b=tPe!N2_^Ry4aBTApCk0(G4#
z7~dQqlpa1_rD~)IzIc}6<YX#NiGQr=#-O>ly;sb~z_bZa779O!ASQtj&5lU;6>$H;
z_^>NMf+95vbf<h(-$XCU6y8v4&%M;ZFhzF0R7A*hJvnPGf*dk+Q&8T~vsH!+`;H<T
z(-LZ)YKcmqsm#kldL!ZDCT|*8!pq1)6qR8m@Cl~3xjOG}zq>)k>aMPu1H1<26lkc>
z!*Lr_UW7-<?<eJ!GN905Q+G7792eHBzRGPd+D-VJ8M;Nc=He2f9WdVrKjt#DuY0kI
zUd!BueN>VHejVZ7Ki`7$hY1E+iak9p`Ts`PTK&iGrVv;3Ccy*q$55LbvC<rMO}{<I
zR)xlvCx{R@tk>U>ejCJW4Jn?!x{~pGOZ5W_)*|DRKRw6Kbevcz9sW6i8b1dA4KKnu
zww`VHAndN#p%ZEl%m!d(PSk%@@cz>ou0IfW#*YUybknxidN8awk+wTXjVu`^Bkg?<
zX%cfQ0Ri&AwV0avHc_L^7q&b~gO>%u?GH1TjxA)uL{v~aK>fg;)|j%lpI4Y@SlRu#
zhH{OR&EX`bdZpiCxK!Iv#W#G}p%e@m*nF;uQRkIZhH&#-=vY77gJ`EpYpDq4m>4hF
zfTratFRz>C9Ws_zv^`@`YkQY)LJgp|Xjl^sJ=tu&^$GmdIv=<~@Ef7KAs~R{t@+eU
zh!t;9{>4y6cTURMXiRCe;W6vYm1Fe?4Nlq*K+6t&J4~ii8X`a~)qIlONbj<ulx+Fq
zfLn6HD1R>&-<L(<l-L8I$VUEv-&MGwC5lO>iZh`Uj@=_-c|M>zY&;uYn?L-8#Q$n=
zxj@$oah9I8nBCB|zMCF5PC|8R{%it13>^oIWLWI?q$kQJ41kfdm&*D{n@6?&8xJ^P
z*M=WmKB6i<E&LH6z9)=gK<N+&b}6y#vGox65jbx&b%v|wWG6M4c7$nlefGMTy?e7Z
z+&sMwth2_7q7o7_mYuVK3a?rP_7ph_N9C*yYXoStiV6D|Xt_kAq`3IyuP=v_E<PQW
z*&hP2=R7VL9SgrHv&(TTsGl{?aLSngT`sE9*{nt<+Y@5sG^cptJ=QgRCjl5HcVX^a
zL7IpD;!TAq(Mooo&Boi;F9zP~h|dGmN2)xmd$(V@bLHBz7oaIzMK80go)SDov~kdm
z13!tW9Z)suwyiPTqTISLi~v6g`?jh-EV+ZAVeU-I+4~%XJOjxJtfOaWTFE}xgT`Ag
zDk@!TG>!Kocn~5S6sLOFv1V~oXM;MJnbxCF`cOdgNr7=jY;#GPc1>iVn>1;~V+s_{
zTE`z+`Q|82bjp;SiayO>^NW1C7vM%}1f{q~pu&}|th5}=K~JTu;OBYXC#b0*-(7nt
zqvfVpNcQPsKT-)%#cl&djF!#gs!^dTbvJDxrk#|N#@teZ!GJ++{{p?8RG<cL<1)Gz
zQUYS=(9*me{g_&Oy|Ync$0R1mF#xX-=e*X!n_@I)f+HSSz~hDDNr=<Y?IwM$QqjSl
z&1uqUJU))fo=MpW*Ye4JRx0+GCiv-vAMzHh@hh!)M5FinK5?<1v}?{${Df7FF<L+b
z{$%3ASSX7O8DU)FYr<zZc$JIP`ers7S~0mX5{qaDe*V3rObXh9UOB?=*P~#-Ei;9w
z$h#6&Fxpreq`dR-Ku)OD^I~uiT4-}X5SD{M`}td^uv7dBY%m~UW^kE;;MLH6E`YGA
zovZ$HT<kpB+MRLj4rn=aA-wK7uc;1VqqN@Tx2msxui}C(W<?#cz@}(!Dw{G#a~79?
zja1#Zo=89&fNG5t`x(c~b1B_q;!^iLhh6F9xP+{jHVJB}oQ){2v4u;K0*wseoeU!#
zL+42oVLEco5xOq;zCh%>jcvR>EGJGhmPJ8Mf3*FuFRcL?iY)mhrGql|&Q!BZk>4^t
zKtVt>1{E<N72tUBz1x+IYuL(KJLHR-*FQvwQ4_3@)ng3cM#uoWp}){?1Qx2}-rK+u
zhqU)`wPa^%+T2A3zKp-VJ#(qE8-yG614J(2a1n5Si0?gc6xv#tt>8d$q8ibQB1vER
z;|RiC$3D1qlIJB(f$?L-U7AU2)29R2L2Yf5eVfei5K({yVNCc+oDTsRUsn47KdFF^
zsR<>6%#aX0iUpAid&K~9Sgg?GmIXz`q0a{)iQ{_j#a7C*E|cXtaW-A8rl_KZepN-N
zHkm)dadc3hEt%GY+SJr5vp3xdh-(P^s9qzYIc2LPdj~qeaJfC)iV8yP9t63_VG$t7
z@)xYPgR3xdt5MRfb_Jdq8awyAQJxpaofg1gQRf>9uI*s`f|(duoOE@<Sn&1f4Ta_m
z=W8}TcY}HieKEq=TO<1IZd;Rzx}fW4)BcDkBM^yozsuAOpFj)qrt#NDcEy?c%S3q0
z!q*n|A`krJ3Yu;cb<2;V;bo~ZERwyFK^sCvB4wiH*v8&dvrP{+GQF_`gm)}!)i;E^
zH7U(Prepz7PMQ7UbfVe|#|2J=di80Ifm!qJu!rC+og?dajf^yDyQuQsG?Hp%=M`tZ
zPipG@*--L~(=QzauUK(NI=p)hlIEI}6;sio8VMs3B|uwIFx8c714DD*`ohlqL7H{j
z{e6zAr?*K4`GnvGbU>E~U<6HCrumGfKPDBwPj~GZLdrSDWtd&AWHMs7;iE`!;dOua
z{l@enIsF{&&lym+!Lm*K6+UwAqW{Opm%f5<gDfBM$#K7aYFEcQ|K%L%Bm==_pVMha
z>sw}phfGlu)oB8h(G*aakt^alGF8y<ZTTzJR7oJwbS4fi{Xl~N=|1~p0BAmbbMSj|
zTS#E=+{Xd<ZGcf8-R1A)r$79$arT;K*goHd!$#w?l<gGurkY)4pZ5`8nJ%aQm5%#i
zh>hpzE*zukRhW9IbZ-eoP|T3GXdhO3$-Upl(2xh%{>g*Wwb+@{t}ZZeGq4gqf^%;3
zEBp+8BOJ+1Up-9eiIfIj=fLcw;8rdDv)9*HcDB=r-*dChqhj$=aTF-`dc}{98!q~)
zuS08XpaaZ%$r?q(q!E|J7$p57puU)@(djFpz*AG}mS-N;t3t_VffGJ$JVyYL+O)l|
zmhMA)TeN#DA>I!hXfqK1W78hELrHHjs8y)+NB(CcM=e?NrD^$je>s$F%Ht~GwE~qm
zA=*coNlHT#tSMEw7a?5gh3#dNU?!0xFvHM@P<vKII3deb&Ct*G&3fMAa~$#>A1CLe
zp`xu?EzlU#h~T6*x`yh>HYK06;mkl$O=jJcr9zlZZZx@T$!%XN(5Hc=LyoHEk;?1#
zVL^_K^+_rpW;ZLQZ`iC^^%)-xa1nP3Li*LXblY{V6j7KP{&}8>T3%UA+*}+G#P4h^
z{&-yb(fJ{6+u^0pNcYx$^cCY8+`HG5QzWFUc5OofsC+y%HK5HH{x&WUadEENNu_*1
z8$fgL8n57$i$)XDAWh^O*_81&2@$BaadCqD(jPW^6pz@qrf9&lbm5(v-OU274kJ17
zo~(X&K09@WdTIPG!WYZX0ePF4u^Mbj>VpZc3SPx_x`v{6wRHB1en>mM&KnDn4+~Wm
zOQ4_WD_(fooFb%zs8+`L0S?%@o;2H>0d#q$Us4y@<qftu5lIv!=!(r_xOR1zZNGEK
z8Wg4f2;;a6<(It-X{Zns`Hg`3)%(GO`x>1B-F>~#9i;6#fO1hUGF1P4OZs-ryju8!
za^)<H09;)U;-Ry(+qZh7-KCb%L!%uf(&y#v84@C4U>3uMoF~<6qNJb|UQfU*c#Su2
zVqecI!1I2ruF<E)5TQ>oB9QT;ZZ{#0o*NGx8$H7ZT4Z4>?+Ai2VF>s7(hCPiw=p51
z4FXx$$PGa__|PP4K#kX`V{uDYsR}wesfcQya_Qk=pZQvxLP7+(?yCQj<{L5o0)SSz
zs-7y50PP3~d&Kr&k5Sj7`D&8NW!rOD-pAU6kwb@hidMnYnd;bRq&*e)P!dGI4s09X
z!JdUbrUQ`7fK*d)v~mU@f8=<-)n=6pFqSh`%52YS|Gck32!7Qhx#hbF_|*m%lb4^^
z1smh%G>T5x2M;t#vstMye>fsCV1yZ8!(axSQ<Fy?osrvHr`RU!^4GQ}AqMP)l2!F}
z0ApixxYL@-cTuW<{mSe6(cERGxvs$GB_em0I-pYWAB9ZdU;3C8=dU+Nm$PN{;-yT!
z@+$*b8CZ~hj$ywg)K9{KgH|-@ziRhjVJ&GKDlu2d;pFtSzC(q?W*Sn>Uux!2ekx{U
z`AXd-K4_i_@nJgftOUC!GGa(hRb1s>;r^l?e7h+AVER0%w78_M)Z;r#{2-k;135m?
zCqNdOl5<wtk|7V>{&E~WA*6Ic@f*dEmX?oBcD*8Y7w4x_EJ*C&syEyU9Y%ln#t-Rx
zf(S}3lW@G^Vs@|C;l4Uxs_wj@-hVa&Qv{(8!~01-0N})Z;St>3_(AkEZ~|x0Pr+Rc
z=kjjMlHoQDY*8?<ag6E!P%W$DC8q4kbn$b0f8Ze+<bUUE<rm(T6m%mrW6JA~J(@=8
z(f%8u890Fb$J`g;09Qw$8ps-N5NQz9%qBF&cI#x~$0@+D{cNW+v<abE#`x&#*P!eK
z<=hF$A~Zz}t(<nmnl0WIDoxGc=~(1|D=LD4qid*;cQ@{*vKl~c>sdtvHv1X2n`U_U
zn2e)5t~6icES$vXGgX@0CjlE0uYlXkukLn5idwOs-*^%`q=MBV3nMRBMN%PUr$2iL
zdW7S8X=6FfnbetRktj@h4EjbjMj_$PLe`A0J=>INmNB_M&~Pz#U?$4gj^WJ<jizd{
z5&>A>k`lb?oJ1p+L#k`O)32>tCKoE%rGaX27OnwJPBO}iettt1Cu-DEjMB=Zh%@fX
zZ<)fgT1cIt<$5Dbp5FG85YmGaXX1M5=m_61Gy(}l6<n9rS8w%xBX~ps13iDjQTyOJ
z5X7!yc);i%0g$laZv^%>IHjGuLu!Wb7V(Z{k<KGfcpX{FAeIxT;}VR0^I)VfL7Z#K
zeZ-ojpGFw^8W1ku>u-)SSA5xRkUeb^{?!8ybVDy@X&h4T`}1|}&vFBd^J->#83uHl
ztxI~K)3j9b#QTN0ecBhm-JE1!!I$2~R6K{8y;xcv?mx_JeMMuux~g1vM)FCt*e<Vl
zG|fl$@g_1Uk3D~72-lw;sm_je!{x1mhshS2Z(S#@T7teC1QADgUlOT43E(gu*~P8v
zx@!dLz6ij;u}!T0_36Qc4~yr(WM$^{f|Wk`IbP@*a0{s_%q6YQI=^DcEHe+)^=%^K
znL#uW6=rflVVb*v!c_N4W5aI*aMbyzCbXC#bIOYuYNM<vg>LYYKld|lGugz*!sauf
z8e9m~K~cYAbTEUpPu8~0rGVQ6&&Ky9SlHni$!$S?Qbw6U|Cm_;!<y@T9dGHGc<y3R
z0oe6TqB5wP?mD*nh62uxln<?aANH<sF^Nn>)9qUZaxeV~It9Kwa?~x{3u?M%@q_kj
zuMuHO{Y}vLS}pj8&H?87>Z$=TgOauYIX?Qp(xNxQX0}=a9sE|aJOXw*h5r7{h`Oq<
z-fJf<S-2e{CMO-TIvv>g8zG~4GZ{{cd(?(!%GgYm-u92!RuIo{dO%d4f@4~EnZhdt
z6nUnO0fv%a`m@cXT;86{E<HO>d*CrdQ@*22pQOPa%Ff)7=#_>%j|OTrwa##~Sw62A
z#ckTInT;10l{d1`hUzJ&<-oM$y@TU72K!r>GL+g(K<t>qzEFP-`fZOQcQPMsjRFsk
zAMRlQzc26o?v=XT3pk3)39_NP6baS(E7X2;>Gi0%IQs1DJ!I68Z+jE6&=8pvWuJ+!
zkrsDznRgXLQZltE|8b<Xg5KA7JiP=4bw24m9(}ps^hfLgX#cW0dSHbheespOf0@xD
z=bT3RTGHHrkM<QHr*XV4SY6xKXnKk`^;+(e$r$RVK%wiCr?RT$yf!eV6C5($@OXUJ
zZS;Xd_d($DmvzMHy%-mx6Q81;zs_C#oLWnJO%cCpr{`w-1uE;p<?S0~&7^eYOvmE;
zUDqd$J9dF6#P$*WUxS+ua6kQ&HQ3dI;ZrZp(xW7v8_h~Fdtub3vhQ0h<bAb!&fJJ6
zsZFn2Lwn<Z<Lm=IaGFI1<Izg2&(g}kS$&qKjFJI9vrSe8gROKZM&B0>txNZ58^*kX
zaOA?R&~be9GXd8J9dDBy>kEUqV_o%o;kI<$b|d_8F~3qsh8SJ_f?D9=`ICbwqO;~d
zg_6+~yx$1cVn<JT<f%vdLml>SeLZ^ccVc9}CNdr(ekRdXe--sMkvMLAf+k#K`bhsW
zaq-G2Q-&}=G1erm?NtD?nxtZaxcs65%-hX!ahW2$Vq;-dcz{k>p2gs^25+1Rk4w3+
zMLH*q!iDCoQiE$fw9+rMKaAns(Q$HD&sA55^@UQDeUoVtScQvga`eA0!|{{5If%W*
zS9<A**G!Q$TtfRaUmp4q_jY?#y8OF`Go%4)4=VUGy84sRZ9lw=YUIwuScVFsN9Q1i
z&bJa2n#pQ~3L9viu1AB}35OfM3`jr$G(bJimzg#AX#RP2!dk}cygDH|fc}YA&r_TT
zHifn9yPQ~g`SoRogyBY)kOF()B4u|@BZYmjhm&Ivjlu>%AN9+31YcqOj0L1T8w|jl
z=9?Wj;C&HjAwITc{)lrEO7B?W<ta$iu@Z)kz!GMmXbOEgG;aHnzD2yt>4}5zF?`0<
z6Ta|vjGSX$qzVh;44<)U!OsGD>p!6%lA9YbKqz_#AMK?kjfYl@GqXDkjZ1>JGS0m4
zjXxC_DAmNY&{RYn%XQwexn@T#u+F0^5(}rCvdLVsOaObSFcYTx3ok*oan7X!H@|cW
z-adP&M4hHRSjM8RewNXw5AkBso<5NTZ6hukC;Ng7<FbsLCCv6<dfkLE(qObmCdyZU
z;Z{Qj3y<qN-@>Yi5i*^OYuvjpV~YnELB6%W5k3z4F`*9nqS`IzTK&>+gYOu?K{s4Q
z-zdb{qh~ObLE<===cI(k=W*7VhskJ{sl0FR%f(JY25>wAxl24nN07wrHlV6M$s+JY
z8Ewrk2!ud|ks2)M3JetjGY}wiqxuv}j+`Ea(WPNg>1D^;-k6*oW6PCVI%<oMIm=O7
z<Cne$jFmtVWEb^CrAo3H!VIHXNkZOTj?dl|u1A)R$9YAoqzoJlm}7=9(p98m2s^mi
zFU)IVnUcJdzZ;1;KsJ+OWjM7!Aq_Q~HNO#R$IgZpa_;lK>tA<<)MbvDBiZYx*c|Ou
zVO2e6r6AqUh-~>k0=NoC^%u5a4qtk(&}s^FQ`})xv7vn;LRM3SV;o0?_poKDs%e<2
znY5+WT|I8o(MuJ&6~u{@7O<%b@<7L1n7y%Zn(L~$+OBUcJ-st1+v2&|t@nt8Hws9q
zP&jXmsqh{l_7cpsHBB>mX41ndIZt7gWl>9A06%3i1p(g~b<y&#Pt&X}Un{A{*^*2o
z_;)~ggE?yaLsM&YlCwuaOKztW^mOr4#GVL^I1ckOhDn*3ytJf+q@_S9Ct(C+7a_Q4
zO*?YQlG_wZKW+kJKZei2&)LfO^4E@UM9hB7A$+8=DOp8TdIx6mPvm##`B|swmKP|h
zsw1jdFq7f{Ji(c!d_z+)Zn9P=sA;X#;+~QSso_rqMw|zInadSZGdGl;^`xx&f}s@`
z8g<Je5z(ZkRVN9~cmn0$&7PS500TGv;eT7oS;V{1tIb4g(_F-U>}oz->4?wwaaYHg
zXY#qF4xY4I2vr@nmK6>POk;R>&et(|;Lum8&0Jcl=6{^F^mx6!5YXVYS4$0jB-FJD
zDu!ej)&a&+c<p%S^J%6(z{~#txL?-tXHYq5hg^D$nD=9<WA>IW{Mu=c@G}1Z?icmF
ze>S>f{0zVQh5c_RR~M%-4xDzQ_>=WWs5Bjd*INy3B_ybx6>p(Nb*KZ3a3^mUJ!Z?7
z9J#V(o8+oz_ed)=CZDgamb#*nMN~?@k!lkx0<Qo)$3HtB-jw3$uPt@NMN=~o$jjax
zFJR%Enp#4VQmvstgWz%-c80Lpt75IBq>&TEa)lX`tQ=#3FCZf>mWN6+SB@qqs$%@x
z4@~MpO2|q<B%A;`+_{m)sa6nxRHYmM9!GAsriwhP3YA`w?Cu9mvf7$NZ^!|aTz`|?
zeZ=5o><<=fyIZ4^ZX7oJ)666!5=a1XCFA!yk*iUPGDJx}>;4X_t<2OD3?ztw+J2Y(
z9I@XH8-8hX&6@uJmIh{enTBRyW@ZLw(@}4zrrNX;pQN3k#LPzNOhnHoGd;E(p{KC)
z_TXvGwEP4nNzVm&sr;^8ML{iUz(^u6eC3`2<gmnuB1~u)N=5_EpCm+V!$u5qGqoz-
zJ9?9(uTURg1o+(kX;TDBk!J}=$#|TyDp?mIxTum{G}Q0ArlPIZQs|W>Kw;L;V;hw~
zgMuEN>G&ntWua0ARtI(TzIeWHK87S_W>jWo8ww+=y@?b$-7LcAN_!sqQP^Fkaba|=
zDg<`Eees-RI^Qb2HF#RfM7lN?aSH$dOX7RpJ*BLbtyFWSF8=_*&GHNLDZNmTGd2!8
zAEb2>&-G^Ulb`z>-ypv}``t;iFhq6_vLBh^G4}reH>KhHs!L7cjGvzT3)jBXXZ?D6
z#mrT1qd)g}z00KADE|P%Mebjn7J4avzO;VbBg9>4y3rr|%0Cy|?_C%8SiQ@sHkvQ+
zv3H%V2)mUxIy`>U8?84gKk+fXv1y?H00|Fy+(nw+t~8MMoXjHbF>$0)kOf43kn3_w
zwL+Dor9~`}nH9WNobBA-Ime0RKJeSRL?0!OYO%9p#kCzhH3p4Du|yLr>*)8hz!fV$
zfO8Owq02hdMqamvW9p+xcj@CY%9JEzsDK@yw2!lw*H3Xw1u75&f)5rRfTC1X^l-oy
zfzig4U=gze4)B^Y!bE$Z;LJMaX{U}P(I}Y~5<gz9B;&EgN-ZopTvT+5PB!`Hi{&&g
zw92Np+o&jvl~+htShRwQYH}C3pJ=WFz%pF2&_b^DYe@uoItIi1r_P-xZ9nSd9tD@*
zL2ae931p2*N=kg?EnEQF<{(<sdZLR=(e(9F6C&~>i91xY^><u5UpB52$U<O+2+qXq
zC%*CBX?<P#^*v7cQ~^a0jOPd&2Y-)^$RS?p^NhKQOW7Fg+{0S;1x+@?#GvgDJF?yA
zF6#-7Y$)vjjAg+45|JoKD#l)AZk063Lh06eM36=tS=VMJ5gzX>RP^fuu)KqmVmU3f
zk|Y%#oU0&g+6K(_?BMalOBH=RNGgoa3?*wv+D^_md9N%YMJ|(4z{d4Vr}1{CH@H>0
zET7h|dteTVY2t?@tS4r6zc%yZo3}-31=w``(7{Dck<-actSU)5)4V!|U~q{Vw*3@}
zS}yh5%TDTvokuC`J9s;(5ckbB!kW4iOHWZ0$dGR&p;^f8I+j9``~@DSy~R2z(Q>Rz
zi6JPFGN=^efE#u=>B>)78o8?=vg+D_GA1DrSFd7_dV>2$b&Z#xibJU+r>*S^OsL*T
z3Q6EYttlKhG8M-u;7M}Iv%OO(?NM_P=DBT8?Gk)vvq$$!7PIYvnV!}N9`elKXtOh$
z12Z*%hGt-UW^-UKOwC|sW|#(MabRX<uro6tW@fT(Gc#DY%*|rsGdY9|%)rNkGlZ#i
zGdYwrL#3IZ9`{+CgEKghP9E8ujbWLXY;1ej%;H5ZWnoBFVaNvvBzGB|GX0XiKfV+9
zsUu9NC|Y-tc#hU*=RGNN``btU9UbQPwvX6H-ez;PS4z7S^d9!n`w08Z?`<EjkG#y=
zQ_^fx(0kiQ>?7|xHJ18RQp&WzfS{5=W_G5jyl+KRQ*KOknV+QdTf@uK=#K+4%c<;X
z({{P$w^Gh>xKp0g>@Sse8g*?%lBc{`o_NVra+Zp*(>&oo)6sR*N!|~l3oRsGl2sx8
zQT&X}IcuKOMk+cT!6`rqNB|B1F(Dw5Km&;^&au8%ozTbb*DoqpLlON_PSf;{)aA_2
z(W029Ee|@Xrm4J+m9KDT8D@7v!j@OX(#*?4XgE+k*VN3?i5Z#Akztvd#fD~U78#kW
zSbVDVkA<1SVW(OnD;drin3XkDF;-E}NjOo8c-0cXI6I_Z;Jh#Sp|)pb%C{d8A5$}>
zc&M-Li>;3>bbYg9`JT3a_Lbh@4`0mfv;P3JuJ;J~nVmRXURO>lFPQ#kZJ+(Ef4E1~
z>i%c#gZ}`^S9^qgOwGdL@|;dzGk?t;w14?3-*AUm<wd*??;6i=hGyY$c}^!UnNj6c
zzi0YJuVK%ihbFr64$pVCQ`_df-4PrMD+)>@$tOH=JFLyZ;_{qMUouJm09Bnk{{X0O
zKiS3gH~m(0@BW~?{{Uwf)Xe9H$Ix*Ze8>G(bnpJ4y#D}a7u4VNS<}Dzg7f{HUsE%1
z`1%pVW%Dmlda-HhI>;)8`tGv~iI62x#e;*K9gZNK4MmcCvZrtIQq0q)$32PSa`}(y
zH6K-FM|m%T=gSx_QXu>JafT3rh0~!3+><{`ukmJa!s7CrPG2(-SF1IEfS>CH-vfv0
zqT$|j{{YpGPmUUb@%@xvQ!|bm7eY9!zGrP0tIe8+(7xSO;f^n<&S*J5YoV1g3!QX?
z9tuJ>{tV5-V)Cbo%jS30^6?cl27iBUQF+Bd2nfIg5K=a9AiIjub-wg2)ds4&(^9Mu
zQxans&mOL}X5q0%^rMN(=3rBLuW9L)R;@#M$>aLCzNbj_TGONcV7#CFxW1-mZXX{(
z#O3oY#p=hW<B%xduFFqaOmSfT43AblKODR0zum^%fw>mX!I`*xeFqbl%!_t9Ni3G=
zXoXCbLKK9FCIBeow+?0H@_hs|HxG}Y97azu<;(gBb|hJmYFPjiw*i^noK(}km{zQn
ISNBW**(3&wE&u=k

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/CW4Ntdtp7lg-827.jpg b/assets/images/impact/talks/CW4Ntdtp7lg-827.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..c0993a592a66daca399bc716b2cb803e9dde1a7b
GIT binary patch
literal 64634
zcmc$_cT`hdw>}yKr1vHQDovW8^se+0I)NZvx(U4m1OycX1nE^;s38!lv_L2pnt=2s
z2|?+-H(wCFSiblD&bhzy8~2<)&V?~{#vtrDpZUyZ&b9VhYybH4;|qWmsG+R^ARqt$
z2rhmBKX3q5z-1yLVj`l;#KgpxFJC4hqb4IGB_*S$q9UheqGx7iqGx1exys4Ta^*TJ
zBO?bN$8|1lUS3{ic7a>`JVKm2ygWaR5M10!MoLCUMn=bTh4Bi{fBWahTL2yDB{ssh
zgallGOLPQ;bOb*-0USR;lKh1G_YcWsB0^$<OQd8MU)$0G2nY!YhzLn3$w*0vNeKxE
zE)fEVi0LkK(vyfX7%AI%GjcJBrI02)V=?i2JpvvrFjn!Yt9{2Uo-8dPDa8y*<?*#w
z4bymT`aXQl;R>(1TH#F@$D)q0`CWe4g@daXt}l%KyMKa97yd6_e5XKnK_THKqDv$s
zM3*iwTp+ygPDf-!&nU{tpiFH1nz<k;+0L6wjOp@-$}`a1?mG!pZgEKyhp-=503{*8
zMVN$ifO~-03BS#$R$G)48gC{klapz%*hdpMb}A~^;*wl71DQIUf=*r?_vCC`XIEhu
zmaHL8B*~mml-nqr35ZI7;scz$Z=3!l7(VOwJytV$RdK~&wR7LBC4lz3@4eNlrwle*
zGpc`m`~k>!UDdZ5Xv=NZ$v(4b+;^gBT|2*$z&ji~m^Tu=nCCys`EB{hu;d{(9{u8p
zY~CSN2<fu>Cfs=W`&4CXhVqWw6AkWOziS7`L;FC!fBxsheg68O%Q>N~ifnc;cWU#S
zh7C{N;j^Dc-`q6(Z9L)MjJNzTP9|E0e%)vF8GH2R59mOVUojZ)oT2v^R>Z$Qt1gh-
z#&_P!89Ke({@hpUSoSA!^u9I8e?Xq#zx_Azzl|#lw8j5I{_QvNo2?nDKO^8v4#71=
z9c2ChC`}Fjb0^QgtT*mk|BIJjAzbkC|F-^<7oQC2<2Lt2%=Mp{x~5Zbl6$X+q4(}T
zTx<TF0lJU4PfH@6TS2&9zq>!_Q~1AP;m-WS3p#~AWa%QAfg({qbNiDQZ<^n{T>Cda
z;+_tvEk9YmV7o7G<i(!|;Myl1|H_QwKQYHU&-MreJaL_z)y+RiyTBZJ@%an;zuWv9
zJAGhwFgpT&UTyi8Sk$b^g~C$!e=F?Y;`>iYYWbO@-}1`Zx2Ap){#!Bs$_?p1xGog)
z+j!s~<H@TbzvOH1HvqbSjQ<V2{SV!t!2cWV|3D8E3;zjS?_UDr4gN2Q{3&1C-?Dx2
zCk8n1ABlV7J6_~n)CT1_>2l?T1_F-TAd7w113?!W_`_$f%~tEb>P@r7^G_B2*1+~r
zb#m!1`T_-i=ae_NaqsV({&U~Yoc@GjaMSRY46FZqE)M-$CH;Gu`UG6a@dC=P#_<!%
zuat)Vlu6%3>AS4{N|(=@Ro2s=P!_i@<T%uu`m351GJh2}U&=3;q+OK8(C^Y%dNM42
z_*0I(GZ$r=_b2l|+59QTo4;5T`u;4h;Fom%__=_d@)P>KzghfMu5jH87Fh<(&I#|S
zDB>PP9Q^k4GpBsM7jpa~L%Kg9C}eKqI&bGxpI#1qenAn%$}d0c7geo)(T^&={f5l_
zCxSrni+*G)Yx^@V|G4?7uH1q1YV*H-BmP^uE*EKO|INieUVe4}`HTKix9{}3zx=Ct
z{iOWyZz}%r^0Nc{q}=5XUdcb5Kq33L#h|}&&tCBJH|{@O=mPhzMi40d%c9LcxZ5u1
z`BfF4EwM)#|EV$FJp0AOPm9;KT2=n)JNcPd$v;}*bj!bf#&NHRZV~fK1^)$<iQkQ}
z;_sIJTLk|Fl=6!L{Yl}U9`yJ4;`xihwx4qPCtzRjZwhPwbBq4`i^AZaYWgSOUj=kQ
zq0_%q`>!V2{!eWGcp0$#E!zG+Ly*GnzRs5t@=HuRzcl+ZQNO%o{b`~9JqB<89t<x!
z=x@pD{npaYocxyTPcQM%iw62ru{Juts_ef@_n#5;|D5@M=>Ctx`F~0G&k^yzOZRVi
zT>A~5q4&R{``3{5Kc)M3cl$TNDpI_Ac|K;c3Utk#uQY|dEDa?0LkRE+fC2B77!WPv
zRyk_2)~~HGk4ivhNn0O!C1M>ic@+?im52N#6sBw>JPZ7+-Tt~SD|na3BN!qm-BMvR
zeh?jlIYG977Dl<AUJ0+x9*svR-wBs@rHXR~XM~IC@Pb0GF#ff->oW|sdjS4gRmW$o
z`k(D>dVJKz=q5S1DiNBC*xNy(XO|SR)ps=RJpjH}y5WYC<c^fGj0y18dst{BB;oeJ
zXOF10MUQ^iSl|j#tfQ2aPQHwZg-VAV8PV6~UUB`a5!A88+?Abhhh}_w3}<(FGD4k-
z?E|NY#I!aZ%yxyZ`I;vBtUG%o8k~wCBhvy<JtecFib_Q@nl`9W^2dccdX3%Hw}N>f
z{Ku4V19a+o5%kIMSoi6i{`QL-`z~%Y!OPKBAb83vk`1OrWTffppwW1P(3@OAD)vx(
ztaZsVD8D+#9l}9Fx8O@w%-N3s<bs|WH?^q{^+W_UAzp?l(^uls<Ap!%@Zipu^j14R
zC%b>&D><|W3ll4j$bPiXP0K>zf!R)%sQWJFNW2pkrR@3&bkA2B_*v;U*BTytIqspn
znYB`4;yk5MV~COS`hq2Ovs0}C{U!HuHa7Harc0tpB4`B(4b=v2lTjFCy-s4#g>_U)
z4(XPD%l%DlKUg>|u1(|loz|zo%VKf~R`dQn8%bYPj8P5@)1!hpM^1pfsL)5oKLFp_
ze7;(KPA}Vjy+ZwM7x4q|SJ2Wf`i(H?T6M9+YS4K8i|FW8IywS6+W*}P!^HQhAApvt
z1`#HqCi~Bi0?K=~yHn;&uV+ZC-IKqG{sBlhD>SJ;BYO;}8Z_QC&|Z2y;a8PzLQk-S
z!>8*k25!mZxpJ>(n*J5D10)l@tJCd*$)w`lweprX?N=C;7oCuI@>t?uO>GE|T>+dp
zt&Q1M8iX3q%Xp83l*~|<O)=(_5@6t}3?A-pa|hWE+R7M2d^ys4-}03@n->}DiQ5Sh
zJwuZuWh^1h-IXf7#^?{fu89_{%h^V$c=1u}p=j^+0pCdXOTABRG(^Pm7c)ry05n|R
z?djp*P{`ooGUchM-$vP{&n{BFQtNSQdXc=`POT)C^ltuENJLIR{O2LTq~&P4&^u3>
zG&?yb`C0|oZ8dVfB<lk<#QnB}4H}W`va@UqH8kmvPS0oIn0y}^qink_#BulwZb8CF
z?sWIkz1GgHr=#a|cJ`Ng=Jdat!#CiJ?LPn_R-ueLOPMPha;nZz6{YrE_CZ8yN@{BV
z%a>mneg89~X$nP(F6(Q+vsRH}<EzFOe`3>|8;~=9QQaa^d-LHIhHHSHUQFwf$yZ|V
z6T@aG3U6XDq|_D}#O}9!z!jaxE0iYi4yX|<F5sk`kNRF<w7C9eooVyMJTl-uT67Va
zt%3%p;4{AEnw<mNOZQ>s7n81PM@<fSIEUr!1^Tg~q_hFca%QaQ@U{EPckla*-3OgU
zzz(k8@*Zo6#g#kD^!X>Ep6|3)e95DA1B4%ajCyMF17IAIbW2*@7_Sq@oRDy;78}3!
z18|1VC@IDFJ<QD@_`ncFy%ASkCy#7~+gi28n4Q$H^{$h%eqB0%l!p|`5i{my5@{_U
z2hC!jueQh`tvUmi8zv^$S7utVo|JKInypTsc_)MMHXo9DbOS5jnf%4SgihcTVn>^z
z8+304+k*sSN^=ecU!bJHC>N&}8LgL*rx9li_K{_3fcL$My&n8{eey=1xEH;jcniLX
zO@{s@cpL~s;MQl%!b@`BD-nm@=A76XFf{~Ue*c&ikKF9xkjJPqxfHR<l1NIdZFpU`
z_lvPu8D)p&TC*+!wLdl+DBl+}UZ~(F+9cxH+=zbl0>1s4L8em-^giSBQh&nx+;4S?
zpK_Qe_do9DUi1<Fg<a%DAEB-*mV6jAUhpD1VwH|zIJERIbpnQcWdf?*)iL?HM#~h3
z+Q@9kinkdsD-nM_#+sw>m%LL3)LQkH6x!Mx52Q^-@E-wWhfLRp?hEr3AMti~85MJc
zU`hJPVfAA4wW9To0gj3ay5NR|o4M@}D&aOGP5qJxLX<o4!U7Ym##<$M8Xx=DYZNEk
zG37F`F5;r`xfxkp3d7D-@g-peCKIVUpt}jO=tRTflXtjVA6|EZVIEEM*6s0VJqqi=
zs!YmJqp0MDrbN>PfyZ?@XDT|bUuH2LUk4IDUEf_Xmn5huPGlot>oR}KI^76;T~Si*
zH{oM{(?RB=fGD1~^VQK48p*p6mMT90Q2!(vN2zU<Z1Z(;Hu0G_?v4S2s?^mQ3!8FP
z9ukfZIuH43N;ijm0$v_3C(sMVa~yo?M7&p8R@&(w_cJwP88cAVJp%O|SUM3y6~FKm
zD%w7JcwS@#6>_c}OI;7dI9oxOViDtCO22;KyK!b5rDJL%(R1zS=_tnfWsy5mBL<wB
zE_Zo9hw+-K$46bJAmhxc7XJkWi51ycgA>beKLQx?inpFzgw#A4F!6O*%DATEvYNF6
z5*9!!Q~n6SOk_Np)3I$RG4F-F#1P(=!w^QMSloA2pTe~dH(*+jP1ub};s^a@p~5!H
zs&AU(7}vyLdbQ;-ooQboAM3{T1Jb!Xo+JiQltjdVh7~dK>C6$uy8iy9V+3SFn~gNn
zTHt-0Q{}@nYp2@Q^BX)QllfR?O?zPcXQ*HH@q7Y3^n#G-E2#*2|HfSh8$479vgUWN
zw9cX`Nop|~7D1Qvaf`@5E?GAuZA@W71PSX-H7poces|p@>$YW;x4O|iZ_zvRJlG!q
za#+OOtnVoz!__=~myqvO%eada(qdN{Ek|4LOkgHwx4JtX`qZXnPyoGz1@@ly67qSx
zzgV+)E#7q{zU}l%Y9)gr;!vb;()7rLSsuE+4q<5DmktXOZEHdERXg???IxH{eRkb6
znS{xDqYJ1_YB@A4DGvNs*c>N9C&Z&s(i=<7(b{`_Bww{F*4#66a@EtiA8r9Hpw`|H
z0AnDo<*}Pf?fucmcP6m-G?V+A)#GD4+Zp33Y(s`dD_dI2?)K1ySWVnzULmzim<!dB
zO%#y*jd@<NQwJuaQMfh;9r-qRe>rC^zG5Ohpjs-}7*%R+Jd9S2)*|K>{G{(sx-9Iq
zx2GXJ<UpTRsDJE1;o5(c%MwI&_bjyv8B&Yj74%kwXv=bG4~>|NnO4QE2UwP=oJxun
z)7DqZ;CmESYV}yQro{Cm)jo!lFxwm4-qm~|F$uHI9fNV@3Lm2MjjhSLeVcRkkC&(x
zBC^`wA-nUNEttD%5DiiN(ef;D_@^r#YQ@k^-FcxnOb;m1xL7~iUB4-ESgfIxm%Ae6
z86R|zY~z%rbUN7J-YB?v<W8!v;4F9zJi`$V$jKaSskiLmFk;1D(b~Jomi~0Su~J{6
z<rr<%%7{=&OBm&kpbW2T-qbzvBxs6LRzdan;riUkW{oYph3;PuqYjA%Hh>|W1KyvJ
zj|?b_Ni)og%Sust+!EW(#tnxMCw^WhE40!Id@*nOD0E)+u&_YRyE9CT7M+-OM$~UL
zZh@wg7JM<ducmvb+LWp3*S7TXEGXp6chlMmcI#&5_;zWfCcu4TV@c>JfJRa(g0XL$
zdBTXiFn>e{h>r1wYhrl5P9D&8ahwKhoN$x*h~Mm%F)v^D>p9FVSDARYYHBp6$mKl}
zySN~`DLy>I-8lSa{Sb{WEb;zKRISLBr44@H>dq~G4L!;T+1l%w-=V_>IYA-)y|1}<
z%6|Ze-7!_Yrv4L+PWLno*-YDJbC4l-@7Q$0Yh*vTcFuC({q7cqAS+)Y^2^KoEW5v&
zd)HLHddHW<e%x=g#wPhWSHWOD;jaT37gp^gF5)*&bJUwgN-8U5Hry-advykRDY$pF
zhw6AKzUQ>8xyDpZm4~QP8EXVdr^LqB1k&Ld)URfnPITxQiJuhF_o%6DhH>5C;=1%d
zyaWqPygboX-|frmhfBiykL=DzjUJ03_m@`JR|576o1*C2Bls%x3t=t}FR2p_p;;hT
zjR~IkZ@v20JP~e4bgkGM1`g>L-8BB3PU_q(C}B8o>2q2NOO89e;drZi4Wy-Tiz};k
z!J?aRtz-shAuq+m|I`$n7)$4YLNazc$xVnER_-c*j)2l@=4JRbqqdfQ2t$yq2~ILM
z-^#t*B5PK-2^UTQamX@}rydS^k(K~rZgpHl&S#WVh|e^LiZf7Jn=-LU49X63vPJU<
z5R#MBT?UU`uqg$6VY52Bv*1=-#Vc|L?faz{Y!=@YG8&v*;f|&Dm75Tk=(ZB*G?nq+
zDEC7sJ;x+SJ_ymQ2)KDM4^d?6)uB*Gz?(h#?%}KU#XZZATpRfG#sQIUYHVP&_+8^Q
zpmDs3w&HY|!#g&u4ybq4JhRfZDS>KYuY-!R9<%5}$9ZWp@!G#`T+#+mkI5H{m<UOH
z@rcV6i;9u|{O}xPGFdWfXjx`FDit;#(=MAqMt1Qa<wKulN;fhYb2GRPN6D4T7Tw$1
zm+GH(tu|;$7?i#J#Yexl%XB?Au_kdJt*z}XoSdksSer9mH=NO5ChQ>pWfAB#F=<yA
zQ?{QbPrfVN$`s6I5$$Aau6?>EIJGR)>ZB;H6EXtc5Ae4uzV?vA1ru6kxbmuM?`0lS
z<>A?tMu?TSQW8cSr@&Tv`Suj&1bFucU|TSRKr=`FVC8%Mcbn#K!aLo9wji)Q>!Eja
z8j+fAT<4)CiewL)_f7CgpUFXMTxjLh1^rc2%jpsE0E%e5@x#LkhIVCeAoVId08xgj
z<n4pR4&IAc{r2Hmg?lSmfC-+Zw4C<kdlVyd)W*+lAoJ5@xn^w!B~irMxkF{;>Iz}{
zgNQ>xa^u>n=R5A;Ss%8&yIYze^UV3pxI8^|4n6xBr}3~$qe@tDF$MZ|oZn{#VHTY>
zOFutdk|52T!F8W$!VJHU{(h3UF+GqqNZdesZOn1NJoRGz?gv0zj_u3pHze=p`*8lx
zUxU5t#~@3fF-7oud}x7NL>UBFlj^=Grk&i~FwdOc?7gD3epP2cd=e#PoE1~N0xKl{
zYjfyf@=ir}ZH=qYQsx=)Kz6vUQ4=o4^)6S;KEa1|xJS0O<jfezKYm1aMf7;cbUCQ-
z2Y?whf;zXsw@%hPJxev2c*~dKk;eD%*q~0t+_i`5+#0eyndSeDGus{-si6Y*WwpTU
znjUG3gZ6~9CQsi`$s0T_@!4AJ^M6e1>A6#k6j+LTCALD=)K#>tpr9W+IYDJi?oY*e
zJAB0C^>n?DA)8{p+7dW<6S7gl55Ll!<9XfQm-YL{&*gaxo~L{L6E_P2G3e4nT<rMe
z_l)#~%E~>WiKMCDM1BC4ufMPjtL|5O({b+k4o__B(*QZkHu)$_3fqX#i7^ZbydjmW
zp{0fEM>hhogfA>?M!L6LgDYD{>)fk|K;y{)9DB+BZreI~Rf0Pc(D5<d5v*ujlK7a|
zD~XdI0Ljzvi)rl2o*qy{bT7sDA^fJOOPyd^7SGLGY^p#a!X#}$zE58B<x^2ne;ZB+
z^vV7HNreA_?o0VY^~UwV0LH4g*U#z{Mnj^xy!(x@lG2P!{`F>(<|-B?9P)wqKFzox
z6y$Z$o+NYO7V8T%heM;0q`kaH+}#GLKIj$|q}&(~!g4dq-6hL-Ud!+_>a}=|`8bo6
zz#QoMNHJP>gE!_TOA{%C#~m!|Vt4#8QPJ*EMsb4+O<Con(3T|3UdY_9e)4!1Xk3w7
zRFfp;A20Us8s|eI^P|%Yjd?>HjBTsB2nJbfoJ{occVYkQqu*Lpch+Yn28l7NWznS*
zKG94)n`+GZO0{2_sr!aEE<)1C4kqBnU827#FetB~1;5w+M0h5Y@nk7;vx*Gw*WjMK
zq*?Tky`<I<?C0LPCO@GgK<p}u8<!vJ&ZwnY|Dyas(je3vo>0gI_qiCT6hv{2eKTt`
z2D@Svu5gl>!?7MjP#P2C6|-pzVD{I&E!`ggqF2b_8L826rq05u6XRNfD~f_)sU(|4
z8;Lp-Z(c|Z(3Oek+H8!edsHK?7yHUb!$`Pgwr(Ve&pezzLLp7oh#{8YrV`9%5%v@M
zq*^w+Y%M{FG+VmZ0V?c~<l0*Fmn!>;zx<<r09u80yTzgvZj9=#aTs(0Pf|wEOSD+#
zf~Qx>*7W7!GRHh+>vK<R%H~B{+L{OJImhFR4^jAQZy6*zze}x-<yXf+`X?0i2ER<H
zrB4&x79!x)bg>T=pbRlTDuIt=SRSCyikO$z1u|8E0e597c%D6*5P)`VWCfp8ffTGn
zb^UQfADy{aFO=qLZ02g&7=iG~O8$uMY8$_-lKRD5f4?t*OM5fvcAP~?_46EsMx!+o
zt~h8QqgxXjk=XO%WG{bDcHiyH<#_&C0grE5?Ra-n?a<8i_Zu|mUR6T!YoC&qU&iFf
z&u>RTd#YU3Ya7@ihUa6%)-?#@s)^ijHAhMEPzfWCftbQsNUWX_-DhM@V4>iSKuTnQ
zMf@Eve?*)9)v~|?BaMyhXPm+9&|RrZVzMw@7t|^36r~8iH|IFlf^iyQ?*m@hYqK}G
zjrLrWcc-GZAt5B-><FE>CLGXRUAf`*Ak6iK<JKn<ABQ&EuW`kP)zszg87)haOs;93
zP3gIV`^m)s>UUVRUSB<*O2zfc5m7x;4esmge2NQ#?WB8o#vA&a<K`m3l6{Bz?W){J
zEa~opFll4bWzjmfin6iB^&yMy@g;dDVO)xr0JlAy^alS);B~0VUBJ4fM*6&pNC0ZN
z^kJ9y>RQgAbgC3pu8;GSf_WphX+yKZjfo2X@M<w63d5zTmCVST1RqHZK7b9dkDy+?
z*5`8n7?SzjG^UPYO!NAqA(Wqf1G4Uutw$Cdp2<YxDaA5JZp25T5P4Y&)?1e#X+QTC
zSx%^2O{&h;c4aYciRVi|9q(3PWfNb;734U}m=VYu;UH}6=OJ+rE=yJ6K;(j!!8*R{
z$?BZ+;>A`)NX!>7y6UJVixnp(&0c3bfD~pq7#KuIOfaeGeuRBD8}!95S_bS#@O|C-
zZU`S-NN=s4TaLD0fk47-&Q?pDra-N%v)a5${yBWtX7)npM<?9nU&@!MmKrA*k`oQ7
zoOXM&0ijzA9tE#$*2>o{Dpg;)w=VKX<?D}3s+}kbA@{03MI$9v-kkXsdedi}TMI#9
z;VgLHU=R~aR@8`+t&SaAoYbcqls<#Uvxog_m^fvR36GiT$NFO#6d*((x<E@)Gf&!U
z$E|ztTNpdR{5cEt`i8Jsee+^=*yxJx^PQ^!fbKnxGH$qB@irbgF(0-83bn?<>lNGI
z5Ll!(CYuhy*qXOKkxeTMc*H*(H|x^f{isPTF#r9t{o%y4^Oo<ZoY^Ee@(_ig^QbHq
zkvY@nb(b`*PxVX|WBA~B;>+catXV%h+K)dkYG@N;)Gd%fH^GxL_o?wq+z^uWO}I09
z9e%x`bIEFERF6-)!cU;kP!QyBy6FP<-(E?LH94GN^)woVw%n~#FJSCbjf_mrukVWq
zrg-#W=;2mM#h4x))LHN&DE_d^=tMW5Gj-4>rJ-iBe?cQ5z<d~&X<p?dkziPK=pn_<
z$~MzWv@fp_Tz7Bd**B?I7eg#^UN^mIy1PHLctVOqVqQ;OG7|Z|OJxjm-B&<HOFqCW
z$EAf!qj;W$L?#aju^LC;#8JX?CHjTbRfV|8wcg?;>fpT|Y?FVv9$Dsqaym8pYvltX
zH&uTC_OXpmRujG(Bds`o0G0-9iJBg}zQSd&NHBGTPq8UtFj+I}l{?0xI8%5M84&`o
z*{$_%1t+Q?Ht(PGdRG}t5w{f+oEHw}em9=R<aw*w&+Q5>NgQA2-(uzxX79YHFS=N-
z>&$x3WI)po^J&{3)mOO1R3R3lfmT`#L&roa_4aNgeqZR!iu$^sv0PQbBo)kyU&aEn
zwJ5oHE_sP5$s;`i^-df1ZuS&-NCVjpzLR#nqY}U*lL47yzoL3>D6h<9G{u>|Z*gvH
zfXW+BkDD~W4CDg}smmTjIFRdm#}y}v4+b&V^B+d^qGsk%!Md7o`IG%L->*hP_bp(|
z_12iRf};iI<155KNDYT`@yd--q=#tTWu+(gi}wf&aB1bM?gE7>sae2mA2MwtHe-{=
zcKtG90YOn=0jPDhr6Mq&{#C;WL3s3nE>n0?YnYT!P12ywwbxb;UT*-&73GmRlCgKb
zmIc)j-6*|u36zV>?0t29-J>NaMxZeIRqTqyJmd%9^gs`U!w{*dow58+AAf#5H3f@L
zgr8{seg+bUe>{3{J~Hz6Gf?3zHMP~jSw3o+^pw_WhpGSpS4(i$mH?aR1nsK`u~cwj
zQCwYSbD7Ki%kL=<MA}EpT<<oQL5AkG4dhJ?OfK*8^U#yB1w1ok3yutOqx@i5c|?o$
z=8EM_hx??b6(bSd>%bWyc>^vL&<BPDXI&T3h{8?WfRT)umq56q9(KxVy;bsr&TvGy
zj5MUUC?JHb141;X<nv)8396hu*xa@rQGcbhfXT!os%(IX3va2z@jR?3Oe~Gko#;a}
zLec_8{Q7o_{+2Mvr3i)aqg|4Jg-xcWCVfXzPwR*;M7c-?o;UlvY`-;N{_*^iGrq`u
z=8MP5C-EywnA-5W<WJA|Qq=GBE4~kwPZ5hU{IX|4;{y}Y^QDGiV_wlgKkscAB4-r#
zsn>&_s<)w~Ib2{+N=z}IqE%0i_-Mo}u(8Z3ma;v{O2nfym7ifcH-A6lWr~TePDVK6
zM_pNUZqz0j$Mgs95*rZ2?i7JO9gL;qo?Z@4K1t<J(f0WaX1x41k(q2KT*R>9OR9U6
z7PU~17kq-F#H>}r=~2l|Z{Qh?wBe{VL|My>i*sspm=zmL)Fp3t&SBY3`YsSz^G1e;
zGX&{r({ZZ|v!3QOHQZ-#XL2HR<lB2q5_qs&=a6>%c4ePnp(o$N4~pM%-l4}1&!v`U
zc09u8(?Tj>{i%mrW6~pyq^vxPi^+k>PM=Wi2ND>NnpdYO2cd+e1<o9(4Y(fT_nd{1
zKDW4U`-P-_aYAjVz#urYgF4>fjRHaGy^N|S6v2Bj@E-t8(a&(e`1Ss&dl7A4yY#c-
z6NeBRf}}4pvq@h(eevR-3t_#zCKdMv71Tke4k;O#f@}fRrsm7^(Gou6CKFP=GRtAF
z^fmKH&N3+Gk%#7aFayK$N8>iNseV}R4d%!(?OLf-V>u^}_*ni2sfc^Q7~Q9f$v)aV
z2YiiOW_m>p$mrdLYV!^8^Tn)^pd>oU8-IM0rHloIRqNRD??`And5%_x@COn+yuYaO
zY?)M?mHsA|1HB^cP6ar;-H*v4JgPUe?fKIPJ0zgz0S{s9F)P=eZ@lC;U8k2$inNX9
zpA5Z3-?XB^mkEs6?#KfYyZQ25M=V@nU4A`})-p3+oE4g7EjtjY8K|rL57#dK&$sc_
zDc-;5=Vwb<_O1D1foBLzgXduB5PTzI5nmM$#PxxWPOgbo$?zTPWiH~Yr4w(SC$&sE
zEw3gfEzlR;USRM}tGg9EzV5h^6-@c;#M<1WZoU6}@APVLt`4j_Gb_MLiB!c>8?3w+
zfQ}~--_rUvKRn8~@M=bB7912g4e+~`i)a%M9-jiDN*-b%*rmwi=R3p#I`~<zmp3lH
zvi1qhkv?;D9)I7G=Jg|3bH>97m%7orqsj!0U?rji49iomf~hSXW)EKch9{)WbkcHt
z-7r2Lm7WBT!=}rASpig%@XE_;ZM|}5E0suV!{vez4R8Fas{1eYtOyn*(x{@KVX3|t
zDfjomwHq1ZcoD1%a+GmgG9B}54o;y<KcOSvvF$S`-JCMiBG<*ZLZ@L-g?kl$wL}2k
zZW+>U>!Ubq2AV1TR9O^4GuHle!6Sdu*ubq)b%;h1$laDnA=DJ^G7=9q!B(<JV6Us>
zFgoY*t4nHp^hhkE$3?w5L$Pj&QC}=lsG>xwCZ?=6fIxJ5B;5F__NEFwHQ?G%oxnS=
zw`jyI^Su>DU7e}BS5`j-ghaog^ewyF`7DQf!Yl%s?lutsM4_%matJ^XA)G411Z=3g
zccybQJL8fYKpirvv#+I4b7e47k@6jfHsR8g`sCmPkiF^0OCEfRTypveA@W~m@gD}|
z_9h5RGuZ4r+_k1BUC4onUIpPdXlo;gsqG!fJ5tvTx3s^OC<+MPAV?819D(LO$VpB%
zc+D_kZhujydH0S;_K$|<JVQl6tAOSIumIvW?iU%tXKtvb_7VHPzOARGWZ)cB6S}(E
z&I<a=x7&9bUU3~GcRSKF)T%+zYTHl!RdkKGT$7~K&3XEiM~Lc)4jdt@!VpzkRW=)z
z%jW>GA?q7bc)$^!=M#7&_XA))i%PRKFRRkymjRk1eG64+yt$p_B4@W<3LaG>nNSkE
zgK_Dnz;r{wvL1-{8#cmr=_s^9x$9#tq;GIni2U0pubxcr={FW;!goL;<Cdr=>`p9e
zUl*+W@ZGqOlG`R#N!_2SOZ3T8jg_dm>+*7PbKmjGUn@kv_V8w*2={Ev?Z(;3+5~_S
z?)dKHq3mw(3LF?G&0fhWZD2I*qLRx~GxGP=1L+3NWq?aOa_`sQUn@&FcX-S5z0d#L
zXY@$%{4uJGjYGR+sJkpaE4OJK9~>Gv3gTG%cwQ$~8FU_8YS$NlD$8*nPvyUAR8JS*
z99N9+h<jc;>jfYC2(J?v=o@tQ750}%cXM_J#nydOtVE(+%O&+dTQx&k3sCPaB82=v
zDP>AOA0ZC@qO!@@62f9`X)+)wc6lETv*RahcS##h%vZTs9wceBD0f<41id!01~2i-
zbV&63l{LO0x-N~)+{R`h?+lfXnEJQNIk9K-gfI%~P??yG+uawD)GEw@EpGKR@L>h9
zu5sLwsh(&f-F+1ttS+^Zuby@3=6%TXZHeavvWq;hFKH!(QR?JV)?N+XWj+Ow(R&td
zkMfKCJ!wsX4<c|@bPVYn;P_i(OfHxwS3V30@Z5)*mrh=sR3YDq8#leV$|zNsW_ivM
zC@|v^P?H89b(}H_YP59aFLQM*Tl-w;C9Oa9@_nwEAw9z?4Uv9bt;rp{N;%3EI%a&D
znj^&d4N1nT5IF%K2ygV6>nAp~yQc9@#=97Ery%oFw3%dMrjDxf*pytrk<K@UJ^lA4
zfjpT=3tN@Cc<p<PPNJC)qW)85kw0EMSs6N4{r1N9;*><J$wuiKwS|6dPp8_)Pyhdy
zR{O<rKSuE1YAaD(T2flBsUE>gCTJX{w5<J>J}n^F)rF#Igy%@Gfbigh_6FbMs**a~
ztqQUIha0(|G@9qR6O4hCE)^+l;Y0q)Y>L9t;Y=Q!uEx>#jg<uyHL)3J57!*2wR+#+
zAsS`xmoZxqS648O!(DD0)@yXTaULvS!A5@U=*dvyf(ho!6P6$|I6>1zl&0O_D0bOl
zm~>WsH@(Q}_N0{eqn^}emw0CIRd*_>;Ezj9(f4-WM2KuWwytX9X6E64&dGy-75lzh
zW+SoVhAhf0ZJRM_IF%f&yMwxn4PLKp3?f6^%!R9`%19_I?xf@HEt~IY2Fs@4`O3c+
zl_Ec*TCt_M+=^E%Ezgjpd5ppI3Wnxum8o$?BX<Ub#RbUROA_f*<G;`&-vy0_M8~)3
zfyU~(TGl8ghUT_tL~NTxJLOqH^>LOvu>^VB&k^sw7rx-b-qd=(jtR)IZ%@LSS<06K
zlYPR+i<4RvqqjzM<7YAjH)dJx!;V<>T<_A@JWuQjzCmKCn0coC10ZmUfQ(Nt%=%=@
zZPX9V=XCVDH$@pTM-cOD5wVBMrAumLw(jnK->O}Hju=@e&^|W>Jq{b27!Owd<l?u3
zHXgNhTM|5NNrAC&HEkZ>nk`v$CfT(xG(MHcA->Hzqb(zN9BI&?2px3?bKJu9dN*z{
zGTK34d;*jOtmIc#YFt_}7pD`;C#NRXvJ~ktmHjfHD1KfMnRvUj>osIN1Ro!8g!zmc
zncmMbhn+&Jc0Wp@jwrHvOw4RJ5Z*CCPoIxTHS92BwLVeH(m!_3MK%kpyUv!49b5XW
zm#1%0yyPWHZ5-AA0m!P{D&C=1t-fh|C(2S|;U)0IpO_#_GoVGJzpJ!UwVKZZQo?uo
zWW~>5&Xzg1PG;hzQL~Y`#zwD4R&$T^lrA+u<)A@(cCjK^9PM~Ws>S{kE*JvO<s?cB
zd;CDQycyoRMosv68@+XPC+(U~(c2AngndCrnhaJ&tM{yv9t2<PFKb}@&i`NvQifLW
zJ51H2$7`;$UzE3wBuvK+0#wo0@b7n@5IBYd-xA$Oia+d+?zRbFccJ7`_>Z-HdMQsP
zU}b3Q?_wV-U8BU<{KsMk_PGj=Ho)QO)7FNAT5DI47GrhrVNn6b<Os%6#xg!4x-d$r
zf9niQUE)Xb=Y%3HBgK`$(SmrbWqn!d%{c)fL!hz|AAlfPmr8Bx_!=V6?e17CCj4UW
z9cYp=u8igTb=*osT=!hbtG2%?GNpY^g*=R5()&YzE`|56<5(i1`bVH%=iH2IuweSg
zQ4>VKQ>=5Q9X4vXLs({Gke`#PfTs+&4sgFoLMIQ)xB_!mz~a2g1uU488IxthXt<>f
zbk8l@n8c~u=ZLnY!VnMK`7RMRUCi_|=ycCwn@(9_y7fwy(aHiEfsXsE+i%go&wvd%
zwwP)gsr+SMm5U8XCJ@Gir2#ZFI?$;UMzjQc0LNFPnVcn_#g678&mYqbB^!=|-e~Q=
zDs!6T(CCtyCpUqAl~?-{f}9HD{OEID9g7!a7q7}FD(PYRQ@1OT*_IwRw#z8AfyqHH
z6!N_<+|pOR)ayn?pC0oI<`yH*VJcUR+~2$h-%)I0O#1JfO=G~$qCOT3(vC1u{s26_
zn1at|ZsnO8;E7q@uRmeEOy_my(^;5jfzsmhGlr(5ObyQX_O&zT;)Z)rj8Owf>ofq2
zjjB;zCf($I2sfNhpe}E62zC@Tggz9Kj^l2T{-7@w1;T1*#>$?^8ayx%59-HvWKG7r
ze01pN|E1MsN2Y<{=!SkH>Q#BU2D~h0p35uh&{U;1ljb?2r?2*=*f(UpXHlV@q85E&
zQ0n3KDY<wDtMef^&8BXA9BkK`6+q@w29XhPk~fWC`;_z*G-;9=Yd9vUqdUrKmwFlJ
zPDrMr^cDcq5?^0}Dzo&zj|7A9Y%XP-^6SUIlufdv`5k6gzX97siBm(@4!gdrG#;UM
z?_0-t70VTlHv#dl70wPMXXc`!)+nZ%W!Fd>Ucn7oZ6t+xF>{Za#!YNk(c@fMRj4l~
zx;TUKb2c|(UKs<f3XbM(qU^7h6;K!x6ZG)(N@3e@Ei?%@nw}+};$WZH7_G?mbfH^A
zVQxA}W`H5K?yGM{X95caf*28Q<W*Tx<ml^tVvG$*c4l`ymngyb&5WW#Lus#i)y21`
z1O#x*%m-#-u^UH<J*KnQ_S2l?u7WrZcM5JRw2eQCdY}R@BI_n2RBZ0BKEh23Tk9{U
zO_=GtB&K3ARRBjs8RUeeyw)bEiS?y-o|WrsT!<GR>s^Bju4d_TuTBwjx?`7Vp_LO<
ziTP%++hMW@2bV?~$9=u3>H?Gww9oi^7q2QYHO&t|aWH72^<Hywi@K?}HZp&lQ3z!!
zc$v<?WZFQ>@a{U8QD1HhKBj={3I_U}&>c~+Y2ksc#Gpr?=rb-^O>dR<Pop5^RC2B<
z>W^Uj+k!|-3Y8fpi_2f%?-nPaU5q-*0)doeT5rMIgLClW;<=-etS;9Yw6>c2$8AO9
zC|X7g;@N8;UtX+__Bvtebm$3@1BVT~+K9Q3o3(#n2;v&X{&)mjd!W~tk(`xETboRF
zkW#bHojTX17~nM)YxvACcadMNBCrMb*rrTjR^n(%QzksP*pe}n64C|LKO>5;V>LX`
zzW$J^j#w|Sv7klF)C)mODqtvr6i{Sk({RQp2+rT+TR6FEb|eg|?vQ*e-a_7X+h}+y
zYa_-|PfqDljz`=QL2RYf(+^A0>ugjESAE9xT+vdI5D1P`g<g!rd^_347ne!s2Bv7Q
z4}JYoy-lPR1y6%XR+cfobSIF}KUT0W6n(zpJjQ)FHt_ZuR?d$kyk2XU1@|ay!J}bb
zSdiNeE1pynM%jTvMWY0ey--iv7WRM=ryR?dFw1_hZv9`J$UENi*JHraO6xtbD!~Vy
zI3>HG(U7s}naoE8qUs?03xl6K#m5ETZGZ0+hY+g(xZm7;yiT*Uk=JqUn>}LZ@HxbR
z;pdVE_QQ*3(z|t0)fex36@I|qFM<dQ7M!31VhnSr!t0ZjGTX(y0<CV9<@><rT~}Sv
zqY*MsGid7<Zak}r<%rZiRlRS0D7jZYk#<;T@`;}lGOPMRVYC2C+3QUhjR}u7tuUFz
zJ91_6-U8LcP2z^c<h5m%oheeG%~{aG(%rpT=B<PVfljqq)NP^rP!8^QR6I*zd**J|
z>Z5HUS+MQHryhwk!Qz7Wtd}to-y~XWr@LUwOCV^|8+}5Za`(0}uP+WyYBKrEo1#jL
z3WYHaMr(rEO=y!h#{C0UR5JaqKHT70N!&_3p}sZMTq<$YJ?36qKMd)Lz*SZxm*8+|
zObiV`GsrU+gEb-d!Le%8XBxlst9RiEu#Z&RdA{9es`QUsBQ2<cvJKWW6&kbC^BX|p
z4aQN)7R7<?OitR|LPX(p&O{lmUlGWGBG9n8pM@}R+z-#NLp~vAX{*JO8Yab^7bExA
zW;#RHbEGXK&TKJ_&b|;?B!R!vf6f5aI6hXeT9N6b$cW=DE|-v;JVHiH$MiG3ixI4*
z^cGS#wlXr%N+fyVlTzoFxy>#!*Uzqga!<LvaL#h15Z-h*c|yG0O@QB+w68lbC)UL=
z%EV$q;?&ac6BV6N+Z1Rb3Zr{DNHWc1fz7vVsF^E@Rwkq5aD`Fm_AN-ahZ*)VR*&f&
zW@!5iPfBKi14IbOa@*may`nYig;PD2!gR^V$yJjC>T-uzvS12RkKvE#g>1j-Iem}%
zyghk#v$(d|*^cSf8yJkFV;D^DGt4%S!~VhH!T5x~u$g+93R_t@qJk+xB!Q`G$CLeJ
z-*L-lzVi7nG%FbVs>1GE>zzkoN!7e92)K4>Hzz?{c6ulspD1?}q?e2L*^t)pO)Ix1
zHgb#ts7PBFfsK`WxgRJ-xgka{u6pY(9u4(aTT`n{_wg8=5bj<wNUc;EHd!<=kkUPQ
z_SwYtLw5tcV0A7bV=TK<AFQ-k2KR#B@WY~)Kdb17H=DlngX%F!Is2whQ&z64)5vCK
zJVc#<m-_bgV*SC2_zw)Txx>c;G@TdgwUVapxa*?^re*p>mCJ12WX6f!uYvj_a1HIv
zwGe@Tu~cjwP#RWfTD7IEZLVXpZ~a*(hAkeUMCs0^*0PO#4}<A_gOv2&z7KNTWip}v
zq@|^=-3{~-jSrcQZrZyxqKAAnc516HGNFVzu98Ho*Pb;{xP2`pfBT-q#f>z)w8?F-
zV*Va&_0f0$Tbyv?e57SQx+xuB27|CqRz_<P@*E_d#XP=OGBPk-C2?zV6oj|%2Ig0J
zQ-mgU3K8>af67-TV!i(Zz_R8MfHzs1eT!;gE-qCEjgz7H{d31@=pK4eJ7DfOBFjC~
zEnCLpKDluXX#+<*dqNBpPNRD*W8}p@-MRdM!Irw-H0uYznF$sYY+{3AF*G+Xun;IK
zvz(4NM15-T-SfA1N+9=}h&k$%)4}U?m!#;R_0On5rJOzV^HDL%)5)Z)i?Y3;Q=P$D
zBgX@ZkH;ohXm}JB#-aAjx>C!hVJ`v!k3Mm~!o?+@pS&z<1G~oLai>2)xPfXJ#$4V7
zapktQx^16`zGA)SgQKpf7?G4S?>h^512R?AI*=@<ZU#2`S404$pXqCDe{4*@%47a0
zb~ep<Pk&1?V%c<V5YaiFFPAh*Nzmk@fN!Bp-uwnCufa;B3AaZJl#d0FY$P7M?6X}9
z%tx|a788e0C)+am4G>>0E~#1V5#EMgl@?g%@M&ru5df0!U6$gOm09+8pC+VvFU$v>
z4C?4LLzvu*H8L3&&=rC~gd0srH<tPvb+c5XvOFy$AHBg=2CTnSEW35*rH>%nE~+nL
zMNo(vM*4P8Y9Td6Cb%LpVw|ESOZ+~Qwvo<Al=MA|Y5ilbKo^;eLvS;4tg!_Vx#ucv
z$vajAI$V8JyWk=9Sd;ekhu8%YyJzCib{30@0KCYY^NF^5JMrP<ZtD1yu`mfSVSxQX
zP58JIxN1(4K~6cK-zn?F4rM>zMOKV4;ao7x@aKKk(m-_W!AFj|=*C>h@~wbeMMQQ{
zQNCTDQ|*#*`Z>^wNC4>j`jI>-8=sj7?=)`Ij6)uIcy3BsQDy(8L$Lhi_D-WV+w>L1
zkHLY@X#J#ehfhJV#{;4oE$^!vk>Cv18E@X0sBoULnpv@=`Z@DW94B)GQt~md78Lav
zx#lrYgs=;a=N35tw$S+Nk3V*Q4O*X+EUY&K7uodnbU`SafYK(7BMD574v-IQj<RC`
za}Or`H})R1^Ou9nh9f+7=7Mp}n>%k~CHcopE;j!8QY&F6<S2L5;bs#|!Dv%$!=dY}
zzz4z|nzotbgmR2=)y*;9u2fDFrUD&8UhaIbJ}-wfL+MP(MEV)x`K09$m1inFR#3+S
zq(B}+y?gBr8%{%N-d@JDl7!|6B)w!v{yjP>YnA-u5_%X^zmPs_C_c32e!cY2^mqpW
z@-MC$dxwP5*{oaXISARTXyp(zDWA1Cj+A^=v>zYWGX_n=OvMuns|r{!U7YC@y>8`R
zcOKiv#(EOMr}@qHn1}QbRm^&~OZ0O4F*wui0&L;;eZ$HxxChTLaa%1uj-Bzc7w7sB
z!s}D@X{zW3P*y;yhu|t*I!19hyC+6}Ac`9<`}X|^B;<ODUqtd><=B>~@y{>Ikw|oF
zdW1YFct@*HdDckwz2i{6L=em)ah^JRq`JkXOE08BqI$fjQ1RM+uW^s_>R`|w&zUA|
zMZ(1?{wVMo+Vd9H%G-}d6nx*C%zPH4^C=)H&rP3oK5TU>t0Z|E>M`1~q$jlbh%lnw
z=(0Aa0Q;ct&>c~^<0u2{?U?$tJ@7g|i89E{SWEJD@f;K=I?pEy@ZCyrp^@N7Ff=bi
z3DzH@TH4Q+km~}>F;J+4BOt%wd(8olTQ^U2H7xFKQx&46YBEcxHj}ie9oxO+P~@W1
zlE9?qyOG&P&4)EDag!>a%n5Bfziavh3T@V{`6BfBT<F20&TGRO4;=AxuyE~3S3$g<
zk+rqkSi{AOfA(?8NGRURlTYx*_S0OXi>&=w!QrEZhuIAXvv-<LN=x;C0jft-2y#Mi
z(ktYKSaIB-v)^&O8SN2uow|C-l#$yTnF;$X$qZ58iPls4G_kcSvsbV*c4?SuTch3c
zEh2n-<R{l#H!FL#>eE8x-K5suDnIh|Lh>Yic-yhQ&4lBxH|;PQHm}BpY&hvqrF*22
z!c6rl?|&w^!8}ea#hNr8B+_MB*)WEU#2kRTl7}>{qF&B<mL`4@;ECmk@^i^(;MK{a
zudp4l+|I3C>%*AXT(d9e!(jH$S_Z0vNw^0#ui&u0l0ox*A}3UKU}3?BP#p*6FUh-i
zsV!&r28Y<>a~gf0=~YQgX1fCKH#leLz0ZnYJUjF>nrymwVMH=a>pr2tjb2tkenVqz
za*`YI9XBJ75Tc{$RPv^f-4bCl&28!(MNC$zY)ks?Wzh1$_D7EG#`uLt6%}aQn0ry+
zgm9KxMa7-)UQSK;4Gm%5PCUf)eTc7DFbL0HQD99Du>sABbL!LPr%Wf+f<AqHTxnFY
z25T%S7`H&p1qZ%mP&LK@of<zFXw6Tp@F%@iR8ify`C6|?ZCz2~;meXkloQG6#pH(9
zGq4GlnOf_8X|V0KddZRXv$F=_h82#2p!AwC%&>VMPM*|Qn{|WsCQ5|T0s?C<s58B1
z=BJK{Ms0+&Md9pAT4gk&TYWAii?@=1dz0n{!gZ&c-WVvg-Fy6?#jVu#nRS3l83$Eq
zsmm1@FSmWo9sOV)g`!Cd&;BVO=A2dEeWDb21zB8XsXx5E5AtB&Qfy`mPLbNVn&SzE
zY_w5jGcsS*<To!tm4T{S$DT&)v<yAkVfQCxvJh8Ui?j4-vwi3`m7S+-5h<|AnD#+W
z$1QacgGfq?-2{-`aHR+#E;5`P!5R5@B&AVxZwvdus5X0nJ@;i<w^nYlIaiOQ#!ogL
z$Cv8~rEwJ|ty?B%mzP|5=hl+IO0>97<ZLhVtO7oTg%^H35`?J6Nxf$U_Vrn+1b|wL
z8h}0GxJx&166vucQ|H&AEvVh&f~Kz=KI^Z%1&~D|w*=}L>C%oJ7k#HCy~dc^R67U!
zmS8D{p9QZpH;))$cd?{`@_pV*ZW{6=Ze7NE(7}B-Gq%~({rLRXQe%u0W#z*4>uj>h
z-mUJ&n!-&>uM{jBXR^jqSz4$pCvgMA_{13CXn&=QhBRaToqC7nL81dhy-3aoyPi*2
z;pdC}{SLi0Z=`d<BGYIX_Eck1i4*$vt>L2ohmb<Q?ws*<P^F9RdU+x3SH^Kh4+Otc
zvUmoR@s5dHjkuugg8Y*)3dz$9{rUl6%&6P8;ngtX1ztg5O{*ij5L#ZXIDcLw*G1n9
z5>qK|)grikc}}aJg&r4In$K6{eSdGPuIYh;OnG71;k8OBjHZoGCKe3E(j;1`Sm_5a
zf@%_d#AH9b3`W(1cRe-lyQQC1(pNVm-06FK?=VE_lpfIv;NUWNaMvx;N_DbyDg$Gg
z?LUq)?xJ8+elFXd$fnIoQr;|IF%{o}sBg<k&onu@8iZIcSw|m?3P8!PiAX(2@F3dR
zh81`Op%yyMKpDmBS0xm1W_&Z+$g)xhFVL;=vT;h-rMl@m8C2p$^*;bq71Fpk$;raK
zz!DTyWB;76)P}171z3}_#v?x1zptmM$HimGx@-%hX0(nyPLC35y;T;OV-Wz}aCfl(
zFeLsglW(2F>Ndte?~?+P9^1v#<aXMQAD-)^F4nD+^wnpILpE&5L{V=W+OHLRzXDo$
zn^$)fQxfojpIeB8q~75{a^((i?dlc*s7zws<8a7Q#N85QbbsE8iQui4H|GtHT)3Nb
zS0}ntp1FEoCB|KXa^#V;zS#BxZ`;<U+n{f(u40C*NIyczEzU5a%_yJ*Q*czum|vGn
z3yRBMfj=<V+zsjpsI^K{b(vtE7_qdfw%_nIro|g4=FaDl93)fEu=ihWh~u^G5Mdvy
zb1!Xdu|Zp?yQUGS<SNP;ujjB8J(1;=e{3WGU$9QZ?*vTxA#Bq;Q%xDuL|S}|hkIc%
zyv0ecWpUxNSCX6NiK+CXYuC==;T~0}?)?l+W8?>Sy~?)@jSEj6u{+#|Rmzk}>3B(<
z+-UYyI0FPiWp)<EZ!WwjEozG0G7`96^Z_dcFC$yjYT5&S?9ObP_M6QrYxv7szyopH
ztnB!T7I=@cQ3;R329ohX<twY!x9GS0HMU5d(;975I6%4bMT34TNs5M|{~z8(Wx&~>
z@-XDLBgVgrJxULDodBIEtM6j+x|Lq;cZIluu~)OOVfh2%z`*xODVRj<w>xB^Agk4g
zZk(B#DMpahl4n6?R-5&7kw$*q(_DUHuWJNiL1gjjzI<BR1K}mU3cdjY(IK&`6QCBq
zYM+IGiqI8{ZSvVt@Se=lrgqiRTk8<yXKK)}v2~^+Bf`(~QQws_igzS!nF3Sytev0a
zw^|5oZNs$#(|F*!H(anE@h%-v<;6?$;)<aYT34>Hmn@;6>FZePdY|w~QIUb9V@0i3
z@ZE&as($U!0I42$c5?4APJ>5oUQwvgh6it3;Z|K-?%Y<Z5rBO!8o#HK694+<ZL{oA
zh^MrUcnkn6m6iVNUdT1Ur#}ED#XkV$T=2%ZbEMU10aF{Vq9q_w=*nB~iUT%(0X{+j
zdHJzHR24;C#(?XVL9|Iob6{Rfj!bjm;g%CU^m_di2898Z7^2N#)busf+g`6*HD(=B
zUnr0a53O2U-Leb?u9jsXEG*W>nh?c_h>GGZ!cHx~*4@rLiIS%iK@yLIFIOSU(2hxZ
zW=bZs`lH6GPmvxogz<~7;?qMH?&I}Nm&;$eRf<Sz;FPocD%{BhdWCMp#x2oO&p7*_
z-tZg5iR<5#Z`~;r6#fDDT!D>1L)@pp1ByDujLj;$z99|#*DQ;_g!kv&`iQt)?xZ4O
zk15=7QLh(z(vj`08F=sEjjsjww`RFC<&U3^$)zLTZ>L9IE5qY&7nh+)*9EnS;=h&T
z&w0E=^D|2<5#Z%J5ZZo(Q~wWX?;X@s-}U{1G(kY6O0@t&Lb1@BQlte45Rw21ASfLY
zItYjg3JMC+l@_Fg6iB2blu!hvSAhg15EPIuSP)RW^g6lj=e?izexEaE&dfP~WG4Cj
zvG<<rJ-@xy`hM4ErQ~R;>)u`*8imb`T-!%ZiBa-l6u-%L5{BA90zm%YO*%^e;;;ZR
zGg<gpfxVfxd$*R0)YfY_>1e7FR7$=1vVw0nZ3`#smstwdtI`^_209svJV<J8=fIq$
z$KKtzQ{Unt0zG-%m6n9*OL}ef>A?V1RNHW4DTst@k(gd;N-mxHWV8M?=P)2-+c>oP
zPSf;m(3i1NYi)I9DYLh{mtXMCXB{0=e5ysoQu0+k@Nc*T6)k^_GahbIX+7Eh@jK~)
zRFwEM-GkJ2$u3{{I0UX^p0;HJ7CxY6UwTZRH=5R<Mj?VFZSCBxK#vo)D{n6=fzRr%
zymLU$@}i_wq3@5tOub*K=hn|my;XRX|KZmWD-Ef@`0caUqGorAiCW|i-SBNK0iX&u
z;p5WdKs^tpUc2Kq{Dv_VM8CVuD1<Dar<uI#o^P4Yuq`mCE~tnbCOypp(T@}$A!(*&
z5u{I5w_5Z^Fk0Km&p^z-#BrsGLE)jVAU}EfS#Cqq6*@kc?z#|>irH&#ETRG4RkV(j
zKobE4T6nHHqthd*>a@5aUrbnf!eUNvufBW42du0Phn$~t2mGt=_~dD=zZ~a+qLFy1
zt74toZGw8X#j(O&hB-eTMWm6*pCWEAzArOon<{>fFFyUlHdUDYp9_$K_6*&k<a5fr
zIsZ2H*iFkzZ3#J{?T5y_YDkrkya(+II~$IxI)y212n`9D`_o%kneVVCC3?dH07xQU
z?7wuI6ET8a*T`+N8amRc8#_^%lkHn7cr#fUD6<bj{wk9CKh1*Y&4zJ1@hQv#QdBz8
zL)~Drl>ZD`t4ts1;M0FU>*z<Xb4sX&;qdnK#?Y#WDyKRkF{AsiDR<Pr<%|<siL#|r
zqP53)0f8)hRI)kh4B;AQPMAUJtRMS4mrgypmzGukP9r{t3e_pHV9=GY?MNGr2WNb(
z0%afc@BWwTD=n+hRU<xWs9-zqF#*QD0?a>G@KhgZy>e+eU|{<{y<x>1IUF)JT}k--
z=EvG^9OgZmSOD3ongB%-j_*m<3o+v?(>%wM1}JL#k~E5~{3JuNnQWOlM)u?qfICAu
ztZVhQ`|U8rTZKWtm2Xq0;4t{yb?+kJOHMqZKDDeb_IRv}Z?SwCEeO;MP}h1~li1|q
zSfzN+JU|N4ls#b5-y>AXzCTKv>7aDSNg{ydV{~uXw@2*zBcnXck=Sd;E{RspA;->E
zXJK0!MDnV9@E8-}DX!<Goo^`P>JOPnwo@d(iIPW9jO3QL4F7+|9vD+M3BO)c{P1vC
zR!;M8>VRF161#qk@R!fgS0_cdq_WzkW$cQlAAZYK0pcl!i-tOwM;SA`hVNVs8&n2#
zx43$$KTYGbu+*=Md!tLF0K>p{F8+%b3Je<JvG{Q@?}WIeCq_R5Z}XMjG!#F)_UG>C
z<%aY*GXQb{z}Xvl+9w(`yX7P*c91BJe~zOOr=n}>JpdYc$Bq)K!VDJTM47qsXR_Fq
z&O5U`!9s8BH1Ry9=3o`g4<E$aPhVX(<7Hehx+i0~;5^3UvTdx5juyfYdMmQML!oF{
zfc*>qDEW~b-uOlZO!6YN5$uW9Z6b-~9g9vBk`erv>So>db5K7V89~0)wKb;hKGl12
zb=Oz{rU|t)YzTZgBk4WmG>6iG^M4f$N4&qCNj=*jg4zFBVtD<fI{(I|dQ-wO|L#O6
zVuQ%B1#$<1+U8!{6M3x90s)F%C-iMD|Bj>^7-SC@=hK^u<K4|pn{TJeRtq><yFxex
ztbyEBI6W!82A?=WQUC2qy3Xp>I=o?WD66U^Ii-nG0yt*F%s+DU%+Sx&`ngw?bn$3O
z={{S=1#$!7t$|J!mgC~(7)9=+oQUw;F|$m?!|scda+H9S^G&rs$nN)s)%L#MMR~l}
ziuLK$)O>^}@+0(UZwd7(&INc(?8i3O1VMNb9?iYIe0O4PyEZ09xl{2t)d^VIu-g?i
z7Q?xUtFzE<y>o&{>vwU2ETz-2_?kXLP?`}ve&i)4<FV1n!UTl<g@>}RXvK-wbHJB>
znNggh5<X3;&2)xm<u$LabvBgU0PQMN@5!OFUP0cge;_d<Aj|q)_<*0=Y=`D0x=AUz
zXH@jWXM8o0#1PwypQM%YhgMm?ZftTKhDxq#i#HW5n|<@BB!n0+bD77lN3k7qz8h=K
zqM4s_S}$QnO0Nt-Pwq!v%2hBU?H{=_HfXYW3{Bvx`;?EXC0&od0sDTG-#>wB>4E=k
zUY2|ZU}f$oqeLw^5+PXozg7^HZKT#TY_fdV=Wz`4{?9vWU;6*%I7Rplxbm-+hd;Zc
ze%$<<;|X*-VX)s}m2H;!FnR47|3PvQ@BK%#S&8CO^9&~{nD$I$;x&uEi^bQ7S@@89
zyq1r@wq5rQ!}(_P_o<ohk6?zw?^PfdW0ThI7al|DK2AyfFxofcx+i{9S-pLz8CvL!
zr^0sOjOvD8QG}Mx2Rllks)m(#&N@3uwIF4fJu|rXq}*ai-^Jc*Y#Bux@hK{HM=)SB
zh~#3AgQNEb14GMu`SL2Vc%6pj{K@_o^5AW2B0#OtvHZEmMK^8elWj0WiAHsK_0Z&(
z?XCrxszPq!9xO4YplJgK87Ob++_O~zob^TDOi(to9L+VrsBd1|RQ0Qn=UFijIBSA?
ztpBCy9_vk4iQ3MmvC%BmC~ESo+v}mnBnwe690zfPDD54WsX*M0ihUYVruvC?d9BC8
z<Av+Qcdd}gDaI>@ut3HSb1>zdYafd>HaZ`XHNDlFz?dBW==0WX9koFU>Ja-L)gj`z
zb-VulTxsjL56(*Dv;CGW=A~`a9jDf&U!jUk@fYPRmzpgjbRy=sWXegPS3zy|d^wml
zz6BY&9ZlMKkV`SNOv_8e!Hf6WC<+^re_OwEp|~lqIB{qk37!kAhXcMyhg&Ee7qJGO
zdS~U7)o2@J@LMjivI8;wfH9HdG^rd1<WW*oiAm<?d!;#(sXXIO`TYhw=@wE@?9q)f
z*cH=S;U)<z2n;HI{LB4?g=8%gu%K4?(BI0j8nXhL-ivT|jeD36k`Q}0#=#x2tvIMo
zvo|ikh4Sx&2QuwV!(KrdI#@W#TkKN<iA6_eFJt12sycRx6Abpt+@qpAp*SHd|N6Ma
zlJ<lkPJcI{eJw`gynD#L_Pb3~L2+;U6(TW)tQM7=&xnLvwA5q@zu_|6;(wcgGC5^7
z=jEmnucfwP<6u=Mzb;aN56qN%r&SqU1b*xYU0|mIv<gCR3OB&_Y*VZ?{~6(gm_Zr0
zE!AooyK6XhT?i65S3RsU&!cjRPcBmP=2FMlc;JgBIwGVwia7{tolDWRUXfBx;MbRu
zVsZ|g(3FUmyWl%<Ktij@Lv7lR`O=a<HaOWOCpX|?!`o{^<Z=S)*GYhAMiFR7jno)I
z*8wqSwi8yYeWxAkY<;#lvW|`eghT!Ux6hTf-3k_yEV3P55Jg&fj{O1~2%B}CVmWTf
z<$U|mjE4~jF9$8Sf-5T{-rHKw8B)G~Qs+@pF)`E;iNu7RkB{Fze;@=S^`%g&$#W4F
z^<vlq+vDW{?~g_K16{kN2Bb7~e;=FFVs1es)sXv?4es|c<z|EHauCOg6XctP9Q#Ln
zP6tC>8Y_ufMUej7O;sm%wgo4*3Z!c(9Ow0|r;Ku4Tvjr}G@*^*GzrJq0%y#){G(@N
z=7lQx^cA^EK3!Csp8Hg^{4g{~HKwS#k)_@Y3KG8*3ZWVFE$l*$q`mHz{S|qhxqb3R
zEoCzS1uvb733?1GXENagG!Wl<dD%pCYPt8!$7CI^Lcz~QPPBQv3W_c!F)Sw}4hhb6
zK>1gKHbs+V;WmS(-s$sak(Mo(xS^UGgI*Q*4rfFAa6AqRh^VQOznnJqX0FlCPLTEO
zp#zfAl-lGRNaJzeQ?IYv8=S1D7Q>|*T&t1|J#g?kVt3FX_hP#_n4JGI)CRfz-Cl24
zmovg8FV`{8N4=-?VH{E*finwnP@|#CR8>-PxW)|0>*ND9C4riC)0331#5i0^y9^J{
z{Omcr^U7~(vyFZ+5Ytw4G05E%yrG!e*73CCNmqYZnGIwsF*c-OhK8()Ok6vF(V(es
z;gI9rdVM|{M0JBJ%UF$_ynrqHp{(+ldcPLZ3jIi(*c8D<lNQ--XqrXtkAFE88*fXk
z-*cXJYLWBpZsxvylZ__Y6jp*=SBJ%{hDhI`r>|g#`U3)WWpCfd_>@kRwt&xE8wv(L
zp?viledDhcE$6F=o}&EjU6Wiq-d_EBD3i7D6s;RRWE0=!z{I!s?2Q$r6x4_kYXa^M
z%3t(V8FE_S^^H0%efrPyK-`|VcJ}$~X~O$XLx<lRT7%r)pp{|2ea$qDOFdgD8C%;;
z1x1YL#+iAKT3bm0bRp&Ct(}&8;mMX~9Q>_Zi|7u{74L>oY)X$s7gqoD!ZDL)`nw}N
zZr{ZlJv85Jkvg-9AW*(W^3MrDrDHg`bMsdp%*-gIk6j2d*y^=leRq^5J~nM=O%21`
zd*Z_pquMsys-iW1GL+`v;PFX^Q3uc<u%f{Vs=)w&Qe&A|UWq8<NKCyGlWgv|8DxAG
zmm3nXRE;Nw!m#N&(p-B1A#y8vmi&g{g_o{YtH;~A1QUZb1JLP}7G$cBP6OpkOI{8;
zi7ojy?w$40AzZkj7dPV-&j@InYd^zi80}5|fV+i@BIzkSm(3i6g)J>qw6ZjG4N}*Q
z&MgN@_&2+~ju*UnhN~W_|82+Q5j0-cHW2AqT-iFYA^jfU*och4)7<k$!CP%%T{0pf
z7%Q~3rm#A~+%|w^am#PUf{-FyX)o7tP(aR@w)IKwm=s9WS*8pk#(NVo8fHvQUHs}<
zqa?#TaY+HsY$kN=?3uMbYop}Gv{d4BGiZ({I{Pb_7{B$XwY_d(b9k;Qq_3+)-Ze=k
z%2&H!us6dntnu9jz?E)*8rpNl6U?&5(nkM0S@VML=<JK0F?Pn&4ITk#AY?=^@Wj2*
z%hrLVOQK83HvOUcMb%4*e=<Mgb||a)v|^r9FRFjbLz-@TncpjuIeSeWa=p#Np1JwE
zS1XSq5bf<<q=nX+{hLE&u;2aY28t3;6`q2Tsj+lwwJ^grE)F9`&3DRkDw=h6@^WW0
ziUl9KY$?%-@9zh{(aJd2y2U~nQaNt1F&{ht<L>k1(_g94(n}LaLSSta5hOP1`?-67
z5M4LHFW@+?dCywvvQ7P0;1-S~4%dPp$nY5;13u*<;P}ORD!cQ^H^rN+N6ue(D2+l4
zefUMu5^|jV-YY)bwJERW+~K%nGF&5L77-CX?o;t6GhnXcvt^6Slr;^yFjq-#n%$)D
z8Gi2%aN`RC$DAxw=;+-B{lUR26S_(XF=-N;1rlSs;&zHRqkBY~jeh-w6E}ji6sNF(
zZ~+a7!bXMwGadm#(E%*X<w3)tmEnuZ!CVIDU%dvucr*&io8KY<J82G<R6_p{gv3zG
z9XVMil#SEEl+^`fXhzgjOj7q;n(|$sZQJqBRRS(Voz25#cPZlcm&_Bc&}g&T_Aymh
z3jm2^Z{~i?R7LLM*;`UOCF~w0k*(T&1XA0#Vl<($98pps35-}GK*8We?psw)c1p99
zRgLdIq#$bGPH(4Y(P#=n=WGlf5>NuFgNxNaA?bIG@eV3|wH&C?z@n&u`yr->ehY$A
z(C^=O){1LuKC!WT$|pmYV0}fMpPjFVG<p=a>pcr11a0tLgXS*faet_kZIa4hFgq7(
zYc@7(im~1_L=*%v1!<n3f`=j>C`O|$o>zCi(KHc2mYbfMps*|kaT&-ond@hNPm~4w
zlm=_9rcu8tAd&9ahD#nGH~XT@&CQ~6MXb{3((%$}FJ&3Khtp`F=O^Z!f&l<ZyVkJ<
z7vND_CER9-mCa#3Rd0(~bw6;W(LF!A*p?E9qwGI55r;s;#MQhiK_#2H_S}Vw%JKH6
zEvM`0g7aEbZS2(vP|<WDj(B0J<%9KaO(5Tha@Jr^xp|R;?Ykehh@~>2?1Qa~5NCp9
zU5D&VEx(yrf8GDtU&pTs4M+d^Jz3GAhAb3DxcCMYR>!*Gur4gK$hrftpU5ltyoMI@
zPU*+GlU`oVlRH((RSyjuPNSzV`vWI6aF!b#f}LkP7bwkf=-z1bY-mLdf~FYtp!7i3
zTA9mhXXN-ve27Na-yD|jKke(NUtlXr+A0AAlRUr~?tBzWCY({|T%~t!W%@T^>w-(0
zJ>`A6$DN!-6Wlt`rL2>4!Bp;npWD)EebaR{Cb~D2T=mkU3JO?T4}2^}&`?3MEx#;!
z(#Y<J(3j1j;dn&#F0A^K_}S=00S$xXmPR`f(R`h`n>@B#@OhhtHnR=&tF_U;z-~-Z
zGcrWbeIxKOk%j8bQ{}&I%&PLo2&TqpD8St)KkDcRwS)NcWiZ%fw4kJrCXqv2fjhG6
zU5xFFhFZ`uYT~y}teC~GmDV<2_|`R!$xF%ErRiU<31XV68WvijZ%#|hi7RE2wo)u;
z*mP}`h7V>(a_Yov_|SrDR&#38V7d9d-k}MH1foT6o3D}*B=1xJL1r`DTTxs#<OnD<
zsOaE(zD5DW^@{+aUp4DR?qt+sYn#DwHDLm>in?+q4@+)N2g#I@Uq#W8Tg(k9MDA)W
ziU%lngWGB9P!+VW=b7H|SOL1vvOtmJiHuI@9J((5NcA97B<3W>(iY*Q^V==DWqw0T
zE}y3AA&symke#*cRBG>So8?7NPd0KqDSU8UQk#?h-DRKo!Z+s7V!~;nX>40t6KG=t
zay<SsZ7gmCojapVvoFq&4nQ3kMvnsQVd9Y%s1`^>e#n_*{#C)FHBPU8Yu{lYq`Zry
zgMvquU>4oQhMOLSsbz|(P*UYBLHV8Vv#ed{M#3I-I`F%PaZxL=U{uMXgMp4pm~;H%
zX4o(LsP7$_{f4R`)xn*sljm2<RKIXu>$gmutjx(N$#R}$NsTn$U5bw0q;Hck<35d3
z45Om-dUW+V1G)q@byCdIY-Us1-RVV;sc1GTfi4LrF|Jr)o2xOH6-PI$Gg&K3Y%`Vh
zmQQ2SM{6mg*{%`QIKQp&?j7w44$<N`yZM;D&gwQ#8>4^bN;c@JI`DY)5;Fh!X+4_o
ztx;&o#j$;CPuo1e8e|HN%%%yOj`^zJJ-HREF&)3EE&)2PwzUBWaINBelAGHp%vhs&
zflT)AZ>NcV63@4~N2`AW`RUr>c#oXz#uc@{YF?Zl`OG#0mOyHl%hw<<&gOBSRkBi4
zQpnn|OVOxlaoRBWt-)kJ6o&;wmXpnf?%XM`a2zV>#P?EWCZlVvBe9P@`mxgKvXcXp
z%1iHjIOZt~<ro{;Eo9oCttLiiG|>v)GL)3otf8!YvwVEhn?(UB)3NaE5-Gl725dV}
zAOo0J|3O@wKt`nBwV6~he8Gir-LYd_5D%q4ohw7p)PE(G539O`e5k84RYtQr_`Sfl
zuhwrAld~IjgDD$rd)s?08kATN!<kXqViwr+U|TkGs!$*}V8Z;)oy;aoMM_i9<Iu~M
z?aP2;-8Nf}@*{;&e{+b{gy?J~o~?J)5N3`A0Ji|Tm}>onQ)e<h$yFxk<#c!~&^mXv
zm#};K?z^j4H2w&55xyV(KIs&-pt$apR~vLgT_1;@4$^Vl(@yN9TGC1cLp88eS%PxF
z8KGQte2w}c4jNW4j>K)-U|&{LhL2?k3?YGzYB=S+#FjP8tPt7#I&H?u9l-`Dkf_h#
z_Qr{c;ukD`+p?Kc`mZB1TDUIv>WaUFe%hcH86w9w7)=2|^SIxa9g{&wND}i2#_>D?
z7<{E-CO>A&Gr5`g{-(X>H+Pa+QwL-50!*)RHCbOUvgdhFNCDCUe2<Y=&|$bp)ulQ@
z8-wFQoi6Yg)=23e14q^^pRtDF_V&gd7Rb-eZ>$ruW6d*WoaoJ+y=D@mr5w%d<^om@
z`JiX3W9fZJv$j<TbrMk{gc60SMR`9HHW(Zj7-Q9)i}olQt!c^ck?EM6^$+aOgUfiY
z<@Fjn4@FfOI=1J&@6&9!=~5n*S@&u~8V*#0#vV(F-ylc94H|Zp@~TK0$_YXI-8OQ5
zHksk^`y?w^;<v;?jhQJTX3?(5{gP+CES|;|&kQ3jsa@?s)a0-BEZe;eB&x!!<H{5#
zmtQEE5Jp7vF^k3nqNUjFvo*|V`gQ<O2JvY>9f$Yu4p!t=QIw2`sSP>vY5Qd0Te!zf
zmlrQ7lne+WC~)5Xt5AqJ8X~j?wemZ>nAn=z{#9TQ-4GP(oH*4i@R)Thk5up1q)|Fk
z;&vK~2fiy2{833M1R|q7crzD;CI_u<`z=P%HFRF`dQ*=p>$W{KE3m@6I1*-f#s(Jq
zIDtkTZqyly+3u8^s(zsS-8-v&=Vnce<#@zzO`BVc5U7)mYX_~!L;vY_U_>E|9|BG=
zf+j?cCqB=OzAi?UyJ_H!llKwQf)KTpxWxf_-C=8ede#o{+)Giuokd?Ow3f79qx07(
z#n@R)S@-bcNT7A}n|c*96Xk-jzdCBalS>oKoi}v}YGUGAGfjBl-AV>9E*DJsEca>c
z`#r8FkvTVR)!=zsh|{WHtb%J~oR8NnCyeL~<%l3y<Hbocx-m=nV~ai(nVB<wlsN<m
z{b5Q>YpyNx#R*{>!`ZR<RqkE#cSgpxetgX&EfzelMhP&_duHenpWzF<e3@22k1%Bl
zci%r#f+5c%TjxnL28aap)al=>&h%wBbK8kF=#!6C+pkR(&>T5x{}aEuKSRXrBtKq^
zzWM~CgkkPMf4_CQt%3WRb?&cg)?5?^iLEV*_u$}!$GXiPHMQD6`~|hUVksgJ$}iVb
zpIH4q`*{;)&<F^>+zrq4AO0EzWc${ohgl?qv(t5v6r`G@LDkEgH#|qLZe8HodM;CH
z;Zg8$kG2vTpHp3$@KS3-W_{;F$eG<!hS&C26)!ID6xPGn(|@f*tJwD?)qkL>m(J+s
zy~s+w^UcOZVf#jG2FzK#4DVjl*>im2u#lyWM;&ZJA5MWQCv}7k;CUCT?$2aY(zw$k
zFgkN;@u~R9x2%Oa#v<eFm&~HVorMhjuh~yRlnr%4vUG-0$`2nr)6KmXYtqGc-gmXI
z(f)m}GYS!Z;g3O^?Ua`6jmd+DqV|EXp=H-9@pQOKiLh7X?mh6x-k@lPK)ut3>fI#A
z4nNv_kcg3(H;#2vv2{vGamA2x@Jftf!$~EO*(3!JfDndMsbFo4y)r5SeZlQG7J?OL
z&eJaeuQ2?Wt8bYr?x)=NAl^{Cr)S@@riR$kJTW~DW1QJmN!)A6-B!s4Og?yl#-8DP
z$IHEWzU5J7)^|3b<y#*=mJL*}<`B$DIOk6hG8Xr2t8<<YoFiakUWTLeF6ZZ@pg+be
zB5`s9Wb&y(i=iKYq9*717ty0*(~Kns&J+J4b}vZDSfO&8TSUB#o{%ufkjR>jCVy97
z-x>%jeGBgfagOQa-{fjDPR_fR#h!vPV80YM&xjlDUHj~>4`Pa%H&<y<#mhqpYsEaF
zdinr{+{#X7&&N-q=eB%tV)ti<{Di|Bycfd3G-$`e0t1{u5W6$>@^`E2x>5OZ$BJ(=
zLM}Iui*eS7OYK+cFEvyY)!Y3b4rTA9X<L;5ESxf>Tpa;QK<<+xQaJf=!ImR*-o@=5
zyXsR!aE$8aIcGYuSg^$`&&c_!mBhxEwL{lMHdp%RHvaCjox6f;%s`phzi@3$|1#lU
zV&mF84|25rgxtOc7E=5-h1f2?VhH`o-BbL#7KY0%C^pDr-ucHFwC;X>v)+zLUk@8q
z>=WI^{`7EP6HVe2y+Dkj_FJ(;Lug7>bU-SRBQRk(yyvrWsHY&~mz}%TvQBf&8&gS;
z){OJcg#>k#EKPb$L{b`?Dv*mVT}=7qG&>cC8usm+U&a;?HMAe$Y=>oE>$ZQ28H&%8
z-Aqq(&p3w)_E>yvpo4UTz;iB3i0Pnjq7T$e3Mw2VF>JVmF`4{PJgBAqnkHPGF@X2b
z+d?=~IA@fGI<&(oreavv)`Pn=96G?=pl=OL^&Q#obe@>7tJ2(7SDdnEI8SEJ{5M>C
z=-=Yv|7T?4zYg#pb_WNMaa26<&bmscr1Ot~lXuQ3vUeJ5i!EJO*SF9R9<l0*k3-PN
zqz70K$1gAIHUK$>raCH0(UrCg+VBK#J!!vJg^zvmyaenlnl{`6_bs>hOI^v=+#k)c
znY(a*ab^7aG(c<9L#ZB%rn`uJ<l2|v8@o<7jIVP4o8t}U!Ix!7;Eu8U{<pxtE}2qd
zd5uC6A{6FG^q3-vfB-;IHP)L`h{6oZ<9v6%kl`Q<2(7tpBr9eWTi8<A4eC7YI&-Hv
zNIISAW0X4Yz0aWn$~@AX)IFCp5SjIdfmlJf+x9B;$OnQs`C5&$Yt@8;N;M-JXq@)G
z22AeXjFU@cfv#8RUyH>gRinG@p-9!@@CvuMdb>q{;|}Nal~$DT?D$gN`ktt0`lb)`
zSqT7L+OV}lP^Tn~20L*o%$~EPJ%uwRKapQt-#NEg-M9PztEMy(p<+D<j)@$}yLAa0
z6v~_|pjD)kid0>-LC)W|P#XG!l;jth$)HyZ%(HBKhmmWwJ-XVjQzkm*m%Mz-+zw4+
z=Xdoki0EH<wF+f*CbPHPYRqO!_E)O?ZNsMd3x~9km7@G1$|SrIvO93P!k6Dsk#IYy
zFDBC1nI4;Q$%hW~bxE+`>8j6}C3tIy`C~5YfQDoe<r)gL=vuXqx~JOJ$fMSuIUAgx
z69J16Yr!9iE_R+f=YLZPJ4snC-UK`JYxMYvo-~HSC<7paYm3}5(L++t@LpFI78Oxr
zzA`a;QP16YxF|7Wd5pQXjG^=X6hItGMsSL8oL5%vT>I4Ad_Gmn8b#GhF3%i%dfONK
z^jldv@{&fRs9o!S$wCkRB@4Y`t3v7ueR4l{#~trpU|l)&H^;KLdcq*;>i?mnTbkq1
zh^xs(SGIgBD#g(=aLP(>URty@_{&)-M}@xq7%#ikipDttT6&2<6bLdPvV<A0t22b|
z6(RCTGINzXx%D2S8|R$E{hIpAwYGB|7|=<Vx%Wg5C!{q}L#bD>lGBa*iCk64r<_XW
zxFa3zu_aHV8#>4NOwG{R0S-;phB&|OEd85#N2zY{qzM;$q3WpZie%z=<cq31FZ8cH
zvuA|L8n##P&kl2ZdPW9SE$(E`5T|OL^iYQcTD<am-wP5VNPX4<w-4%w+13TI=mHG2
zVg}WDrB@Zn5{Ny4)$ACrbmFcwesl4SBg@&3(C{O;`aR6CsJT$@Y`;mDo6#<%>uD4`
zL8oRWUgIF9yi#17={;AutqV3VD5`Ewwh7c*SSg(tkFh;W-NHrR$wf-4zRKU4h{9SJ
zcNYUH#PEo&Xl!jr-_Bd%iMF$D)4Jb&SG@25p4cm_8|8X*7Kk&^=TwZfSp)7VpNzI7
zRauBHZLU}VWjEr*no{3le`C3m+Q}P9axWQ0p0mW4(q=S5b4V0@$GM{4-W!PIw~aF&
zymhPRev0_F%a*S;`9f}AdaFaz!Jk6!nmUf65QHpT-pqr%4=PVG9+YHs%qqQ1Kb#oW
zrvTZc^g!Dyg5b?~$jrNqORw7IN4j0!qje3AC3cbtp^!PGGc@DG#UU}$3g>&nn6B^9
zt2kR%YKG<0ni^CLZ{Xqav(E2qGbIky9_D=cnpgYd*RoF*k-R5;Jk`Kr`;vv_mkzd}
zeG|7K@{I_82omXD;Nyyw;bk@0g{B`eYOJZ}*jE6!e??eWbTYLS&3CC&bbbU|7cR-p
zgM*uf?ODutT`_mH9jx~;1)0GGzG#n$N1~y%UJ7X-o+PkE!(;X1-|v>oU!V@z$3Jv=
zuhrX@YR5!TGAn2k;D^xdbDr9Fs}#zIlDar@lQQG7;J@%}B-{tIWcrM1(4m%=N2E+I
z;FR_LgYh}NW>W1Z=c$S7?@>fQXYgy=07qZ`G2P3;8&Zar?@qG+scO=U?ZcC|qJ65o
zwZ+8oZOju7j#Oodoc2}5T&!JgRRmwAL!5&Us7olJTk<7U=y_Q)6ODFW4>tu2+<L#d
z1x4B}GT3ve(YW9dtj*;;Wz9UgUW$BlhQ!D%eXz!X%F(tkr-rsU9u7Y>qUtS;Ypayx
zetLdVBXbYjcfSs?x06jK)qlRSQn7N4T|wG}VDDrr6v38-t$4u_9%`#4bXw6Usuk|8
z@eFnDbOW%amjIGXkq9pn_vU!Q&rews+4IXOhjmWe3Qxb>nRT)0?WV{pO4f_8gefTx
z;)PhljaM*M5Q~=o(C9+d*{<?o?uh;fZt#bg?2g&bX$<YW-@F)mxkwjfiCD7}1&`Ee
zxyk%WdgJ;J=xG~zXDeumD!8GU%N$Z&Lzh&DEmNc&9p;?x^cQrL_K1W$j)+nM@|}Wr
z34b}`HIn!+?eL8qqL{)A1v@+GQ1w#d&rcu8r&{&dtR1<zcwf69Q9pqWYGG?MsP<72
zvbNTFB34*MoLlS3iOc#qigrPB!T#y{hPhEud-9Hx+Tv6>^)tTO8oUJt7Mg=RCW2|)
z0WD=U>#Ah&T2HuqD?pVf5J)}CQ1k|>09f3@Z<TO*8N`KF&B&Mo$V#wSO$BQ213~Sj
z;lRqk$F4GIn!G|7xw_f0-lXWU@3+(fOf{xZkWekJhY((uvX*Te7HM=3G8}D2y#KD$
z4043OBLJq~NOW83W=$fwp$|oL4?6$s`<%md&J#kYaLb?qb`vB%Zk6K{Vqz83QCQgj
z16R8l(l@ewh=r#6IM>{Wx6sl*)0(k_S@p6&dRP=+Ti-tUp*bix0!?8f4D#J>Yt5Hv
zh!Lo2-#BiyRifvId{`)^v)xze&(-Qrw9*r{Pn26E%EN5h^L@~2A!pV&2Cn07<=1GD
zIGULkgr_F4zI{yRfH1AbQR@Z1sDo>Lt;7osj_xXQEuGuOL*?4AK(~bpw35XjWwfQ8
zfNdHHe|uilEnI*MDHE1B5LpFJmOk%`$0`oEhaT8cg*uaI)R_$wsUT1qesPF@kZbPw
zkN?e|`v3infBWTscYPb#{)d<6jKz;fS^svIvfWWV|Bgxj8ILiUBaxGH`S>M3S*5R9
z%Zn!`a^en@pv}&I8NRxAN99_N16*7_B4Kx}`oDtPzd8DViN_pDFn|5r`0x0)^mCl%
z#$WG#JSst0vs>+pA<%!o0r*I6uSe4V1f+Xq4`U>wJldew_re>s!wK<)1-T~8`^`^s
zAmuZROYv<FV~sCG-^eKpY<Gq`9m`Pi5D}~Du#3!JE(*V2@+}G2gAaP|`WsLalzjlW
zpXSqEFld5kr>_+_c;&;S(FZo*JqruW(y*Xf(-zrT9e^@_?xeuc)z>q!A*32MO-oPj
z4FLxM=a0UJKpo$1BkK71<*NM%x4fexa-n4WReKlQ+sPClf$<8~xe<cCpUk-`EbL{I
z4J1p@e%U+RcUK|e-z%KOF(G1Bo=2al*181WWP@bpR0&eCd*~~#v+@tpWV{@E+H7r|
z5Ql&b*O>A(;S+P#K*-ZDp=wVu#T}JzQ$>DJgC)Q(4v!Fy>95>lKQPw$Gct11V+Au2
zP4Ee&_l%-cv3J)K6hgMy$#=Ey+5*L0IxvNswzM%VGHaBasUw>GsjL55O^e$O9P@NY
z=a8n6Z#eqH92j~#`bkD3?XeIuGmUmw>=7xJ8J917v8svu?NN`wsZt##3xPxlAn0Li
zZ&)s!`z5;rj>v65YqNM=s`f%}F{U)&sEwY0z6__KCYPK@-7GoqWzYhOafA9cho$f9
zaqe}hrRbBAj0$w}U=*!=S>hK8b(7>@WiChvvbxRLz#|4~QkRs*1xejVCd~j+cL&Tt
zzyN<K$3DwR-dUD!<`x>&AFNKL44}G(=(|7@(7=Grz`O8fIY*Tn*Y6~g3ow}SLl>rB
zmH6eQw`Lw3AM!A7k3(BOCZu_s<Kcnt1&jjYmHb}#%by>YL62qu&cTksx-CiPQmcy2
z*jukZUFXnKy6cY~ZHThs5B@AEGi4AC%co2iiNQ!iMaoj)jJ&GvHkSMgUz?(no{I?D
z)bGf$?p8ECW+}j6`MT0X8q87C4y)Jgb<d^`JijMs<F@L7iw<dLGNZY%)`)-6D^|E3
z6&iY$&^*i?G)-Lt=-;F~5U!A-&^S|lV4sdZ*)DcLuGy!CzQ5>!bPH>A(Y653i2)(N
z^h{ys<rDEqYe#I-B!XJ_5#3V-bp7avy;wDk7jedBIVm{J(zQFx1j|9-7D}*Kq6S`8
zuN&X0_W&g>CJ@+*7vVSKwXqyc-%e7wM$$3Fx5xxGO{ah0|8ir9=FXYBv}}~>H=6nh
z=!(^$EjmH&EvrWfypO&bs34_4JLA|x_Q>31^y>7h=i^_xy$*!FzEAwDEOqgB>b9-j
zsPjH;^?4$aWY8p**OG9==G1u}L7Or}*v`*#iz8&p*7{apG2Wbj@;<66wNAH|*Xx^L
zLu$0gzELaY(6uB8FBCBAoTAW{)W9PPDp02~V4YkJ7LYwpP&_XaQZlfrYcettILd#U
z9^a5_r+e+UQX=TJHQqOH-FbidQdaiN5zjOO_G}3ld@6`97fmp&Pny&s#_DeJd+x}b
zVumlrHjTjip7l2LZMO}H=N~Axdtk+aIe7&7WLysIi^2LPUDrr84Ib*>No;Iea3Epf
zH}d7hs<UHX6?kL%N;_cA+AxQJ7Gv<{8&rFGG~t+&9zVCBh@IeV(4isE!ZN$qD^8{n
zR{4dSh6NDC3eO$6{F_xu1N&s1!R7WX_i0sXCm{duIwD1@naDAv>MWcrdwdJzGBp04
z><Y@StQ|>|swPogrmp<ttX{xD#1@T`O=Z}&*ptAifEIFfMZT6e2#*Ej{W8UhYl*Ij
z3;J-*VVr5lgqR!P6$>6Rja2Z;by*M)XmvX&=}4;L1nMp98A{ZN5HP)MKI%T#RGm^F
zFumyPo6wb6d-GJzwy(H-CpL=89`FLy32VZbnoYgu3psX4m9#HIoo$wDMr_r{1dn?d
z3WO<in#`)dMUW*yxwfX}Fc(LC${P@(pTg?}>1lJZaWZwJ#W%^};u=$$HU-4~A(#{e
z>)pK5>{csTM2khT)jTEOx=T?=%ZX!!r(|Xa<|rn4H|!3_O;}p_JD;xsHn&tr`2sjt
zx7Ls?NI@o&BAgg06k6|rCK+sZ&U(1cU9Ck50=(tE@eYVx;ugy}+AX^~d)eRiea_Ta
zvHF0{YctmM`xk_&o+N{HjtG0*Jm(1$hp1$`6a7=pO(Seq_$nb?gVru;NwYwn^i5B-
zW4?jWY63Z#Dah{1!I7Jz`Z94Q<!O1+h(AHSCmP&Govc7oUtphi5^Hbj?Tlc`-cnt*
zcq?gVXQb~86x^jnDcL51FGNPgBpbd}dX(=X<B?cV*}bC$(+a(5EN2cvxo&lftf11j
zQPpT7ZN|eE{CSfVmvZ*lXOWBSz5Qi;cPtJILr&9HH?s279n@2}l~i|a;Fs>t&H&c&
zz@2CSwj6thuop+VIE#2n(dnue)ef9EsOhY%o?35)YrF5JnH3`i1OgTB!K;?fD(Ij0
z)hn(Hh&o^KO<fMwN3d7Kt_1@OxJlmEM5f<M$>;#ND@CI3J~-GsSuIBn>;vox6!XWv
zxPD-NpnJr9fJ2ocxGgG9jDk(sx$lH%6p#+3rJDh_uXfW5N&xW7$IoK^;=iMGH%ntR
zkMUEZt^2-*e!>g*hfuxu3w^qgjheP~tm5WlIor{L<u1Vdi!Ubr=0KiGG8gGgeiKW(
zxcps$w%LT(o%c>5seI1Tv4yf<g(R!{j!sa<t0&*>w6HaDAI?B}3SZLXiTyaOz1e<Q
z<Uu}Ep=(%4)!SZ8nU9i-aeW?~pR?h={udTSs@ULlS`;L7{MomBC+75@e4GD>RQ~U-
z<Ns+K|6ipfC0f}c>o2aawmB?^FFSuYf4uO{9X0+ZCI6d4PoqrmF1sjaF8a^P@c&-<
zCrokuhyD3an35!tcs)33h3_PRMXxRtERZPj2b!$E{n@vn|4zrBdw=ldbipz<nB4SW
zwxl9{o$3Ci9}s4_I@(pi&N__ettCZyMg?66IOksD%2VGE!0qI7=l7g2?_;On7{P!k
z0%fg$;L+}){3E}?Rm$>P>gSU1ckxxsa`@*{M0LCBLJ4Aqs<Tb?+y3-9%CJwtWDb}i
z9_mk1E)jQeT+xiG!AC>4RR}`jIS)ifk_JQ#VSUs)$2;5{U-HP*`o1|lInF<~^vm(w
zUX7h+R5PE4-=5hX*Un~JM9%O6u!~Cyu*zxFI4%<8S=H{|@|CQn{Y>ED_W9%wGxtbt
z{`_8mE6Qs3n~l~mW$o_7*u#XBn;qvSHxI4yNGrIUOz@=Pai5v|;NF}0opP{6a0=h)
z4OnNLr7*0dBs1*%ontQK_q{^Ebt23_(sLMHdm&0pl%ldHS1kAl&MdXXskF2@d#=3B
zT%QvpJiD%pH4h><n%WqpZoIRip7ByVVcIQQKdUOwK6Sh&X4&<#-GippUqdd|c|Qqo
z@LNO9Bsn!n>0ribK6DZJ1&losA)2xSomRa?0JyAAQsuo=V@p1}%(&EJqu)px@;1id
zCb^z2Y%*bn$D=R`Po4;3mac4M@kB=pJ5%jw520@q2~FvPApTX8kD5T$3$e%yLnxye
z$gGzBjdZwz#mapEbMwz6YtP8ssi}=;1;lJNq7s`6t8ZyTQu8<<@05R^`NS3SNH|a?
zqu+6z&y}Hjs7N=25-l%;l*7JT)OW^4uBDWwEo^5A6cG-UC!9dPYK>-xWCJaLHajON
zwhWixjOZJ|#G1`bU~a|?D-MJR9~MwP#~m>Zi2ieI(+apjwRVn;*Vfr7j{-1M@F@UF
z(vouB-Y<x^XJwmeW^<ca&4SMKr(<Cd(SqhFP=1P7X$rE7y`37pM<%wp-%oDZ@PNv(
z-XFh${k-$Db#N|oj1W3Bb6fy+{UPiYXsQKAgN5cTp8}4BXFN|jrM2?;(CQ7!5Lv70
z(NgbFTV(!pTCK%#sMrP(5wnBwu^DDl5+~-PU#4#S_Prh&lT&`-77BUAUg$KVWxDHP
znOK$LYFy=OqLE~`LF{dryH)G4nXCJk*mGp`w)2mB?ohQ<nqNhdVlvv%&ZIp6<sb)q
z20YK9HMnGd9PB3UDdebP`;H}^cyhh<(^#B;6Id)`tKf%ek5g^NfRA(3xg`hNS5voE
z^ygZKTQLw`EXXW5t0G_V<B2fSTo~U8vfakEMc_;41Eqy`-cR}Qcu3xBWaRCe`UA#}
z3OAxUmbUs6<p}HRCGi79koZqi#i%Q)GFZde-Nr`QRb?=tED}_XuBKD9&7EMy$x&=<
zTh*=i6&gO`VmI~rMh3)2S69Gu_LstH1DjkXbsiYKC}8a8aK1~Z2ul7MLG7faA4~C2
z7z;r@F{%Q%ZJpUpTDJ<_i#n{l?*6WsF^i@!^(!{;(arMC-2rcERNhHg?j)Fsm@QIK
zu_sihS025rjUjjbcpV(zi4P-K)@M`#jROYutF$ErddTHB$S;6cr3*Dx`nN=L#P?I<
zKlZ+$l5|ptG3$(c+T0tjA415d1VELOjSc+6DYv=o+y-8&C$Wg24CrI=`mP#yX*HSw
zIuViMc-6Z8hb?elaqrF0eoaeaFnugD2lsHtMW3>3o)PN&6y$R(yqC6A5^iCONYASr
z$%(L|F??;+jKK_ZQo|nu_+*2~s<7pk?fw|<u%#4R39|e=RY->P>{jxgLQmDAX%kc6
zcI@LxF9M!s+yg#0M}mp_HB{O<6AMZo9z{Oob?Y|a3#!K;{fjX*c)V|cJU+VbIky&b
z8ieG-tg~mrc}ODfk?h0BMDf6`;}2nzf`Z3o^j8|V9ovJ9&8zhQ*uMmpk)gE_yAj#~
zW<zqSW;tP%<6~o1foqpM>&%VpNpS${h>G-#gQKSw!l7)l&+XC%(>pXGwyNQ`?M_oZ
zp(~{p&!ak~XN;rlj|By7s(Ne?%c2j$L_0T!^6y>bb?`c&z)FbkU}EHI%%$tAc{=k>
z^Ic`-?ElSikXhEKKNj18pE=%UAF`}EV$f8A{+r_(MO^>@ID0Mm=trMpH@-$-D?LOd
z1^Ni?hQr>+0qXd@y!-d{V%FTh1&M5<Y|{vi$VD*zAyi)+_4Vl;?z`ky3w%;PQ1&Z`
z4{!E=-6##l!I!Z%hP~!`iu-FN^_A%Jt+KX|ncFR%Q|<+Ci%}-x3%qik-w=?-C9WqA
zwtMcauDKIagJ4L=buT4#0}uEwZ=+;z#HHCZVsHjZ&OdbuM!y8hA(u7_xa5v}7OFE=
z@NF?V5V>_Ly;yrSnT$zpi>dgsi-tipC)99N{7hYNc)PxoYFN>4XM6ya>d{9$h@yK6
zP6oE~e^Is2wT_3@!Rn0*`R6hrLd4xjkGK}#vMX__@%6DGjfnozl5ewW5C+|mey+Np
zd^7fK#505&C@0|oDrtQS7V&=iFOBM&#_RY^X)<1ovG5)h?QqfY<w?^k7dhRVA1Wvg
zU%__-sY~{rM?V7=^=&-&BbEx+=1sK#o6U+(M{@a%Si|>L*PJ|y@6Sm0t2(XH)mJ1%
z7vPa<&2Et4&?c7ofQ8AMLC}D(;F`eJTyw}#9Gj;WOrVtZ_D;VFh##~RIV^w<ZR!f|
zFCc#>Hq@jrTom4H5;^_p$C}!rLB0Lpb{?bynnPsyHhn=S)BN&Dc04|G&52CdbJih!
zyzH*TZ|rT9^b9rgIKMpDgSO~$4aX}HxZvz--gdu<uY4O$pQS2`1j@CrLTOK5U=gRA
z3T9jMUe~#UbWTVZIo9cX0L6qpo!au0sHlE_hpH}4)?JJxR!J8X`Koveqd)AVL@cZo
zt~IG{0f@af#T}>+@pgIh6ji>Lw@@Q2RA18yB&a4B2n-Uw3ffHxo*fRHJEFVseM_vl
z7isvo{qfLq^Y0XF@Auv-C)dvSU-8q)9L#@f5m@s9^VP6h*KX}4FpF#0#X(Sf0Ydsx
zSK9SA2bi3te5!9hGHp0E;V}s39K9EyHOJz#9pWMFB(%l&H_Lvou)R`A3qnhO1xPXv
zi|=dIRN;Lu;PDqL9U&k-*4G#>>*NAz&X4|TV2CQ=&ja4|DZA=vBW~v%p$~U<{_MxZ
z&Ar3;pc}B&*S%@avKS_uAC-KRMr1P<=k7O$Jkq%WWPVadmf<Faewxka5%P~cex28u
zz0_W<FTNIP<j1{zqo!3_$l3H`qS;F8w%s@OUFu}tC7u%^OZtR%l+<kDnXk5>1-BVy
z(o4ZRe^x}TYiddBllf<P0L*onBdI3FvjA=UXR$phj51>|6qL?(md<;x=(s!S>$8mg
z;LJUgJ-HjNZWCb8w&Z5cP<2xUo#^;x*w!yK4i&`lZSIK4S~Sv`B$E4<3}QHnY#_5?
zWM0aUOEU4v+)=c7HQ8okS{=<}Kjl0z$E1&>Aofn#L}Pu#4dzOQcrPX>@}hbvOYW#3
zv$wH1RV8AWx&$Ry-`9wCb=6jP;1)dT0kM2V4`PrC9ys}@*!nuj{n`?0wP|6`xRW>&
z0W*etZFQ75lH+)1f3)rVxejT@@NwaBA^R?*b7<)8x$*TWkIs0=A07XK*!>%(_Y-Jh
zXL(l;kknxS&z`r6(WK5dV`r#K4SOtM;5r#bsqHxIn~k#u{rVtRcW&z0<KOom6j2vF
zi-L)J$PoG~6H1l1M4n4aRR~A+LHNaz0Al=#O}R#Nresa9*0L%-F#*1L|1k%Czl5{N
zJln+)0JA<F<xKGI)sl!6xYzepwvqLaxCw+zo=%*B3>oOWo`%Xe%u0PU%u>KSXrAAW
zH@9MFd(@~hNo;U(k*&eoUiEW1JIBGR{EExUMh{-b9oBf#mP+Ns)7M;3gxWyu3$6=7
z)03zL9{yEF8^d>5BADICs>+CTc9=M-iqp6_Hw2PMAn@^`9<wG3TBE-o0H=dxcQ<7c
zZNfeFxN?s~FZ)&#X9%c2QgI|2u!1>E_;x?g0%o9dZ>lTtgCTg{Cg;=M&e|KWG^3ok
zS)3rQ8bs})0w`)wT%|CoFMRk4{XSLu!t@nd`m@r+HDUO2I=hPJrX2W8x@EfFjtFPZ
z(!e2->65Z&1sp$Gk6kwS5PxRVwxd6jx$wp%3?Dy9Ja*sn2-P=BWbE5*u-n-DZh@LZ
z)NRAJkL#R*4)e;X>=9^{FPk1zH&=b7g_$ZYpHlWyF6$j@fwj7-23gp?Na>iYN%VRm
zTcfWpveLO1MJew=?!YJkQQm>xD`(VV!(zI|!*39Z7S3KT^PBzHad>|c67}f1<ti_`
zJ?*L3ikqRq$6x~epy0RHcT%>bK^~J;Q4u(Fv0ZD)#eSJbIj51ei3wE9u~Ez&!OTL@
z&0MbOQy$QuIgYb`@;%O(oH_-MxTs^2PDS{w5pP}nAkA))J=kVeQyrYAo{*}iGw3R~
zdPDs|)*j65R#L2b>jcB;%k~R41sF|I#|0Q=PK?D3VOEfqo?482HOX#UY%>&-CGYKO
znAqAF-4aT;BL{1*zAamWh1*`gSynv1azx#o+KKglU~u^Z>)VH-HT0fcD^%VXIA@jv
zNx4vH4OHgyn>Ct8v{gEVKRwjTuUKC(7MNWRlYqa-sd&0uQgRa==b%X>W1-tq!1BpA
zAXQXFBP~VVN7$tB_5wL&#zPl4mc5j-bm$wZh8rCmn~#Y$c`ES%N&PY?Yqi1Cb=?yn
zd3r{`NNC|%0(#BniJu2!Z;w80AHx{xCJ|31reJ5#dm%-ZF0T%CrA|pH_#L_OX?HKN
zuW8FMD0mA9e6J8&mH3)^v*x@<O$RG@16ftm3a2}3A2xX;+L{@5l^i{(oWV@~l(+Ht
zT=3>eltDyN?<CeohnnXfXZ~64zSK%u|FM*AccbaR$=>k|8<@uD`(>YbA1h<mzcynu
zGg`9xD##V})-c<hcE@{6+s8nF^~Y?CNi-P0#qOBoCseRY>fhtY4Gk&6UuC?kq?yBv
z4q1RMv#44dL<XXkmWG1t1u#~Y&l0Sa6F5C`q`&$#5gXGGSx|cW55E=1sl;<bG%Y@1
zk&<s*L|bi1Ad~oB7+*_>e|bw47ycS|Szm4`vH=md7UYQ>*K79pCuSUjR>b8vXdcm<
zF-~Z8`*GD%?j^AaR~|&#$+iu+0<6nwED2pMh#A_aiZAT-paHXJNrCiY*Jk(TfO$B=
z%F=QX6nDg+=N@~TeU1|wnWi^&xkhO{go;S$t4j4*xlf|C@ri3}cdYzvfJTd~;TTV(
zfN|^g$kF=RJ#k<8oW*GDR*Ja<q8KDg_NL~Qe~kL5dRIP-HTI_ATH9U%C7$+USx4x4
z1O=Ex$DSwUF(oY#Uv0fI-g0zv=?`Gs;>i_Q!KU3ETI$J`NqB>K;M#+=tQ~&GniU58
zD+)G5d7BCC-HW>3SYnb{BrzLFc^9Rb0pHvz)_(k@`V>W7&}rh&Gwv(xzE)hy#>Glo
zB;(#xy+(&hWOU7DG*`m*TfwUz;tabvP^^WuG|s@!u^^DPlk<Fy6c*8(_a+peqKpBV
z_Do3@m#RmQ1>J+CYu%J!N8@b5uk8*c4JMu50&3d}{4jSynmjIe;WbTwQo-0`QOOr2
zDhFBfk#i+6g9&Lm>6^cK+bUG4FopA93VNS8?vQ$o>P(%Fu~XdyMfvzO4@cw$IFyEP
z3vXzKg2cd}RPbx?{Me3eaVg6h*0=12sO~!vB=4MJ{HZj{@RchX;kWs_`ZI_I+Kk|;
ztxa~$I=V7{&e8U1=bke$?!Y~cfr89CL7v_&ME^%+=Ksx^^B?T!e<Y`W{j`m~`ljs%
z-_o;xmvT&cr|OyuF6wYm_8dwNpP)Bo@SC-_?;eO>$-z>Zvhtn?<*qJZ|AezTrT=kU
z7{IQJmbTqi3%Ld|U~5|<zr}@MLnNXBFg7U2lMTAj%E7ArU&Ot4R8!B}FB$=*C?b(2
zU5XJ973sZ7l@<skGzAd|y@ZYk2nf=oD!qjskP<@gg7g+@q&EpwARtXo{Fe9r?mg?g
z>#TL}S@)l;o!NV4&&)G>=9y=n=kvh>xVM_)Hww2XX=fMo<iP;Lr#dqVpM98Fdzp39
zZR&F0Mcf5CHg2jjvNCS&JR4gOM)TqtY~m4=ZOI*8E&`u-#8b*8ct4dhZ<<_Ez_J?5
z?Qz9)UnP5`8Q;h$xTdGKC2(j;#D+;!tDE>EylRRg^^5L)$-#(^6dh;;aV6fySdVr%
z+Lp2<heqs?d~XPtJ35-96CCp}o}cUfwoI_^;b~bk`;e&;BFsg_0WWG(9VPpqvFxOF
z39R`9s^c+(CAY>|4WYCp6ai;_(at8`ut94{TA2iSo4-&9xnh99j2qid7+nfWl%zZI
zUNK`-C5bweRBq=JR-r;YjIm73#XIQTfUrzvDn`I53}Iezsua8KK@YpqoTTngCa3XM
z2q%lC<~v@QQa-z0mVE%mlKODk&MKh`>nuu9km^3qG>L_LaHI(|T`@o+T&R{t-k9Yp
zutqIQdw(2)3&TypYLH7R;8gWDXyA<X7aL3TyEX=K(DP)Wf@mb>tRwANZNY3nUrgu(
zJ&=e95*nx0@$OFB&UVnFuFOa292n0_S7U@DrbWjmV<XKKt)mhWN)yh^zSLqOFAmJp
zaw%Ah;WQE&D$Rc#llgtAGfFIpROd=xTxDWL-mSYAYG_>-G-MO&_;k5^=+JXA9juYI
z2~O4eLH>}7IZskNh)>w!=9GrtxIwh2rm=;gcNN(5x;TX{yeq!RtwUMT2j2D}_nAbY
z98Q`cnYr2I{S6E-Im@xDVGJXl&(9SVQ!M7<7fLo+lVUPAlmRio(a{(eysB@2V`Rvf
zBEd~viyPiia-pN3dp}KzcL!wQsF@DO0Dg5H)Px9l#M%g~U!|<NN;SWoywe1lO!4D=
zBxC@}$cI;LDF$4biW4uUOlZoPBO;6goAGbL6A}fXw8v@A+RsXo!1Oe@TelWfbt&dK
zJ)TDw5>A#3N%3Pm_Bg}B5}j0{@=p|C#gE$Q`FY}C6(feAdtHrc>X!j}LN#S|mx}$A
zwc~bu$+<27mV#%g31JdwH51CQJv%+22pz?kyMK*O^$xZB(w#&G5t4<$4L=;}4j!mO
zb+sbnfCm)mj;ae+$%i1z>2c!E43d|Q5tFt7SpK?}FXH6z6J`y~9F=&lWVv~1(?yL+
zj(hJGQHT1l>&>Z)vq<gP&E%CLg1&jCIcB0oKAU7U#00=T#Qv0aZM3bpfgTsveGNO_
zsm-auaJFL|DMM*#6B;*|oo%j>h(u)h$>BVOWsl4}VOfN5;$lG34I6J(lR9j3x}`;n
z3s!$~wvWOj`g3h*NKT~9EgZAm6gja6KUblova+XKvgI>qVfb0>)0%PgmUkYNRPC+n
z1j`VCKr?$4RCFMn;OiXjn`fw_Z#aXrFG3V@Q}{S>*U+v=M(}d_n6=lE)B`!TSZ`12
z!mzJ0#yx~l5D1+w>}1v>ss@#xLJ`rXF2A+L!+*hwDV-Dv5ckNZk<l~P)nQS<>OKF7
zQ)r*lCH~C3@sga5YsY)(9wKiNl?q)@StFgEhw~BQc*BU^-4x6A((Shey(}L0=c@xn
zFa~PC8dSc2Y-oJ5@{n7;%c*EjgfOymNrHf|W|vOLbXi~X<SdKREN!TGOA=4E*6p#s
zhxJ2~I!)xmNe{E{+fMcBTh^07uH^95n=r+fd-uKMg6?;H5^BAPlBq28Ef{v(w41R|
zMrXip<OOaMRb4T3(c({+#hm!zAz{g*x@(LNe(GtxsMNaBZGgB1l+15S$+bp3Ol;bu
zI2tkkZm4=T!mO-MlR|wQpA(lecg-?BJ{K6L1_IoiW{eOu{i%s`IJTbKa&SVaj}gbj
ze4v0QvNhGV&#SK0M7RxiP};xF&nrV4pTT@*1IY_rT4og^dI2{+Bmt@350Rt+UwpQu
z)F>`bBCr`0PhduYydfo&_p(S?B@*Sd=F(o-n7JID(dQox!cByqnU!mlXqr4MqT$Rl
zBID_KK`P!mK<MpDl@7@yuj?#R4`h1p2iVczd}?BXc(^LKHvBl5(!XM#q$UJSRqZe}
z7-03xR2w0}_L1<_tH=p-Q%!9`a2XunN<`Q7QK%&t<NF0dH0Y%F6(L#Cusk;Ic>h2R
z{0@-rw()R$e*rRK!RNp~vp}y-N2!77V}BqbutJ4OkFrW9hjD!=@eU{!GND1yH@X&B
zR@_YiFPkd7ThCaHCknUS;PL4G(7eSsF>3Uk-eAjilSQ(f%#O`Yo#2ewN=yZO-eAk3
zZPPMt1lz`QtQf{#vu=9Sab;B}fAzkXd<sgI5@isJfR+ESe=sibdD;i9g9BEd(B#Jh
z%a?*4bY+y(J*wbgDE9FUSS$F*h2$3tEh8`rs>}C<jmBRi$!>9(g4lh*>bj$NVtMNB
z6zISSFzsBF9r^J@wL8~HDuM2G`(CEAMfRm7Wm?>+L^(P@98C?k)KuNgA1bGn5gLVJ
zFG}IojeQ1b3Yq!3JdP6$VV!-<WAi_lHtN{bNAKCwQ=-iCA#VJ&DJ*6!IW#C$=-BKh
zSEa@K_oJ>fxcS^re=kcR9D3d{*fKtr`vE`B|3zl@78>lL6MFwm^-poaaBVeHgxDir
zM>b^i!AjuCbJktE#T3l#K$}{ZG^n{lQIpQ}yk`=I9?JAxBbMDA%mngLY#<?5=ITZK
z^eV=`*KU1P%d^g4sJt>%mz;L}oT;Ck)=PZIPRyr-&uNND(!lE?&%wU11Yr(brjC1Z
zWdo&{VoL7GwJ*j~LD?(v!%CX>6BT!WWB@Q;4L~Z2R<9zPBNe~R<gip*+K4eNE+TNL
z<W3&&;AX(V27MS+K(LR4xYQdgk#U63<bIEV35vE+djk3%ej-dV8k2Mt-FyC!xIqOi
zP{Rn`bALMLQPXQwEn<r7k6;0Mod}|{qN`utP*RO3_ZILOJ>HSlk20wDw5vfTrz=cG
zCMAT8B*q&&mRhQoyvK3UkzrL>%vf4CD{;>heuEZqJjv*)Lay-<<dSe)Nd%6dEBB`=
z8E0(Nz;1nV+fwn;P;*U%7Nq%T)6nwNq{jiezhs1|O+I9XNt@_dtnxv>732qK^7mva
z+B&Q*U4^mJI;{SYBCk0sF0yf=qkAsnhdcUnNopiyiBH4o4m+FV5JQjDAPjkhFu*;^
z7x)W4r8XTsM0C}KQV&Y@ayj1|B^=^EwhOpggij-E-;v#wH02`LF0sU2+6|BN57jSv
zS0$|-vqGUAze?JjYmSQY0`!L@3bBeXIOr-F_}z8s_K(0swXo679c}+61LN;Igu{kw
z{`>@UDurX+4@@qRC%(Vj1q%`MEvjmS>A^8a?<=5m%L@7(*EAvwTykQJB%f;t2pVMY
zyVPOze7_6YJ@uoZlA^n!5ySk3RK^&~p?E>;wpIlH!V+KVnWkMPs>;jez;p6Ap`l83
zj={jI-cv55@*WmMfr`n=HFK;TD*SkpKmUHyTaPWZj`(RV+BlWDfh2S5KCg7^RUWj7
zuEI!8a1wDa+19{Kjz2!_p#OkM`tjgDf?R^=mpT4ccmHP<_#YQ59ygQXUn~o4vNYF)
z@GnMgHAkLBoqj0}O!smQYyJ5}JI78&!f)CC5ED4WRM^ROVOa>>>``HY44NTg-bD;f
zzk5eb6$Bj{&Jo`7l3e7;cgyH|ozCg~8DVw3M`plbE8$UDAh2e#sia2F?!;fPO)=85
zj2!{;W03pW+bzi6^vqC*ZVRJVoc=?_xzdTJo9ta0<f(=*Qc7{@X@%BoxyZ4zfl1O+
zh3{Luz}|oq1($ptBT^DaV9TY>e0+_|o|?dGRPL!=U8|X%ez*h-#w}J^N0-0KY~{qo
z+$J3dV19FpGqVAOz4ocayyA$ilz$DbDe<+&%-7Rdee+3GfkMppZY@8-d!?vV@v@2X
zGLbokQZG4H7~a^;mCsR{Zqv>obi><G>$?`{qnF$&;J9F4BD|95)yM;rhpE5DCg$r6
z`u*^qJabYcI7Y-dDb5>R1ZO_cIVT>~nlswZ(mNV3aiUl$3K$CNQyU`l!+!&es`TBz
zb<ev2G@YV0h^4<%E53QBhYkSG+EXhaHXFJklhV24y{`R`eghx)!l}7G9`1-07B;(%
zholBaHmCB!67}h(Uk0y)d&G%`wNKrBoiRZ>Yt{FfC|NtPhy^V@1`@LyaJDMi-gvxt
z<?EICpiQ%<?rsZ3!uQH+O=_Ty6Bzu}#!j>--I|qF;gyzm_cIRDopu+g1(JKFF=yIO
zWv!|=(;|^H5e3DN?$r_-{$cJ80{=+O8psm;hU*|RfDp9bb&KjTyRsPpC}B#$l>-5p
z`~k5jvc=>mEK#<^k%8lAS7?BOcavjFUxgSN^YNxre*V&tF-2Fdrf#YRn*JDdV}n?L
zNFq;NN4iVeZO8g4?Uyr5dCP>{#LR<AH^-icj&Ew)T5}!ppUz<Y0j)YNA5v&vqItt=
zYxMPWCP?q(@67S>O8O4DX6g;f^Dq&~9p}MittI{9IL)pi=ICKlC;f9ctg*aL&g<le
z2b0_rre>}KIr8kQPKV;e%mtncMcWwPzP>L{HJdfcNAxGxnTz}piTX5bqjfdud45d!
ztJI%1T;jw5(nWJjo|H|;?m_f*Z}HZ#;T@g2uf`@{;daT2G+D0t;w<LXRX6?DxOq5j
zUI>zRGN$?l@O3H49P%T#ZYkP9+aB^Ew7NDv?UGZ;5yugck$${JqKJWfm%2G7#1G-j
zr<V+#d-3o_j0r;Rcu0`rswM>rMD5XY%h7Ag`d(UIVXn!LxY65(K@YM?@9-xyCxpDY
zHn+LaMCd+g*PvnI=^$FT<5|p7hkfeK@ZChn7H?#t9{0d~At*OjVWuFy_e<#n@8ydJ
zc;5QQc41kLC<zOuH$N$=;>%Vf=Q7-fjqVIx^Ib>lML9*mElJw5dH7MYv;CT06>f8K
zJ5cbYo|6a6PSP_pV6N`Y_COSUAIt<qCH7unr>fJ&tso;*ujb3Z1ZMII;jj-ud&?94
zd`CJSU}^6`S;yK%wg>UE`3<<`dk-Oc?}9o~!ye`$BO!D%q?=@e*jYf}-Z!LA=S#ae
zSu=(A(N*WhdJrYNI>RQ`RmByPbc^P8e?9y0_wL666YtujgGqCE_G(ka3ZL3T*28iO
zved+_eHRPmtj{_|uSnmWmiwx8;--2cx0Vt#FYk*CGU_gbKl<FiX}+%ku=yhM#Qp0*
z+%PdiTaN)d;EzyQV{$Q*p70D}VTk_CX15W{(Tzs%uPqts#?0s*pj-5|Q$n0l`&<dE
zjBZf+B@(r1!bt_AfVC3mRG2CD&au<0@q)p@BJOv(f*SLR{HTL4A}8#cO~z%k7n&G6
zc@ZsT`RegXkl!ZH{7lG1OtDC{otMRaf);d}p79#*HXNe7q@t^zOtncv6W;yfiX1@2
zC@zg*Re*u?<1OhsMfb}Z!Yw4iojgu_lZ7%EdF)2r_C3p;Z-nMJ1&dg9&vD(pUva|c
z7(daeG{MpDaNwEa(^sMa*Vc*@a%TV8$CZAZ(C70T-~p$yB>FRDWveinf#i0ZrYHzX
z&YGcqlELiGjv07Rx);sL6<U`tozCYH_w3_L8F~aKLnX3QWim)>=2BdkQ*u*K4Q;93
z7h9zz*ymH>aV2^*#y<~EfVM?v4!Lle70e#LueQhH4^EuRbDOc5h#X1?V(KB4*t$-0
zn_Id&w-JY?@6GX7RX9y1_gvVpp_9f9`bSet^?HMK`7G-uHr^yJi~-b9j+u6CEWs3D
zpL{kL32c<7q{$9<4*f2)o?BCJbdEz$82;pjFh=xfg{~6COI{S14;-f!BA4kBtMo?}
zQlLAWHEN9p#bjmbvPM(S$mV%Kuf8#B%hKc@s>)#INgt3=+nby0F~$_C+D5`CB{iXM
zaCYC3s7Swd^pF>et5UuX!ie-`n1@i=y;E&93oy?#t(x5lR&tI}VIM(^4T=Z(dj1BO
zA<B_+{O=VZj(5B6!sHAHR8r)`6%=C92`t3pLPvDXl8~}9Mh$LZ+Nbfz5{<l+Sc<ax
z=+OpM-Zm0PPGTpLu(QnV4>n7D4u+s}1S3sN&gdEKPpz>sHky@+DuubO%fxf)AoO$W
z@JlK=7;A_@bJvc5b-j)-ysSmgyw6Z4dDOH+FZ3=S$YU{M+r}a3XW3V@%Z413r2%Zp
zcmJt9oG3F};*R8I#}g`>2M04%3jQ~Z8201{^lvOnME8tmoX*=yn=|9fqeUtB+Uo1I
z4mTZP&jlKr-c*Qfe0EF_?mOyTQMt+X^h?FHf*e$i6ZcZxRK!ZKd(a#yHcse`&l}e%
zraXqW_U$gRlb5C0Eq2#;GtgKE!}sqQlM7J#^c0=XQa&0em^#X<QOt2N40Aq07t>!p
zXc0EnvqZdJ$X&>B?oCpM-pUD1Zg`q?$8JdEw$yn8VKrr7{w(6ft|T;B({d&CDJDld
z4}297zsN`a!11oskIWfJXLSDt72f<qPI*CI^Hj4YxWfX@PtlzQ^2xwIw{bcq$KFl8
z<{p$;Uq_wrwASM$x&{lsS(IzC2gPmP;m6RMR7>F0b(5m<9cR6KUg^D;R=HVcUE-J;
zxg@t^wsWShGdezzn_a9~-&7LPp!L0bAfxaJ^`cwb8^9_c=(|ltV&~+wErYZ)j<hqW
z6<ZG(QNcw;a#FFO6C2aS8Z&#oCC!o#aQ>2Pd#DPqFa7EZ$q-)hji}?V8V%`I9#*Bm
z+y$ex+9gf%HWkfhI9BETuk-3bcXnUO9hn@5A5Agh?=>}=?`&v~mzPQS3vl;T1qW%2
z+W3jLbqiEKF(37`R}flHZX6|`>b#XPuPAOX)3B|1RB))%rgL5=-rgJMM?gu33m><f
zm?nop{BT8Hd0Mxd%Dq|inB}C}j@eZxoxKX^S{~g?UYvyO9hosWE~Q}S4UTuIq9lG)
z=^JGtCQK#AJd)EMW;8;K#}0*~Ov@<{ZlAu0H?acv@2crwn`?pv*u_*C+L**0)<g(}
zKE5}^J8#yy)sM8_s+GVcCQq)_tLxpQ$p-mwn`rYFqJbQ^eoFI3j3?5n5H4Ih_3CLu
z1}kt41~n_zL^ssXwa*X9x2Y1ys|wI6I>+o!Cmmy5i9<d8XoEKtN~3ev^w%7PHNkfM
zOPz@_qF?7(0&<_BrB$5F_ViJjUIbY3oRQ61Nz-sSTF8AKif@MVl3t$fU5C5hrP^dL
z&zWCRrq{X!-7Xgux=C%z=wZNF@aoV2Rm<`tjp_Rwmkqmn`v)Md=P0-|D)AVL;Jb!8
zqe1CdklT&9F4G<`^}O<V<%1>`ZyOlE!k@xRwdS^IHm%xrza;23MHP9#=CwArXP+xm
z@Fmex-C07U>6jKXSJckW)j_VYYIKBUc&mdETpxzJ+tzqy##Ypff2Qv^?tc9ZNZjcc
zQbiyYO9HNC-{%8tPb+uU%3Tmtv`^WIF>0ziYeb^@SVzVWQharg`OVD-M;J{LzN}YB
z{?=#vuCVqy^Qx<&^*06(v+?o$oh^4q#qB@Om=zb;w9gjS|1{HlP8B=$Ac+~j-M%4p
z9+bXl)67)7(IEYfH3VVCtqrWRD1p{tQ|xIxJhUEl=+N~eYmrAt@jll{Pc^u~+SV>}
zp7u0fk)CVqf;}&+XrV-rMbhd1u0(t;QLv8}7weqb5Lo1JhJi(Kn&5B*XK6;Ncoq~k
z>0u0@HB8A<yOa%pq7qf6TNO=mlG!odahnhVhS95HdYw?2`2s!-?&rw+Lg%X5Qa0#v
zRi^Bf=QI5_Jbx)NE*x|pDxN+fgRc-3KWI5WUPZLTaXW$X4fcsYvADU7$ByA*7&hO-
z%liu}Xp_Tvsf{(<>UfhJc$}QMSC{N4)R>3H1lY~5h_`*+Aw*6Py!WPCP4RS}Cn<$8
z*gR<;=;=*D2i38!mKnS0`b#*w)tIzAPpqFd6HW{3qP68U-Cq$-5ykbCE+r>^-PU$0
zapk%jcf!B${t4hYfivfi214`%ZOJGZ3K4l9Iyr{_0!<BC8y|z@>_!8uxbjyTt5X~X
zresz8mO=FVLQ{++7<1b*Mp{Xow?bzfab;H%8o#p607Sv!Tn@X+I9I|FS|%6E$Qp|6
zP8&boS`*1=oOK+QSP13VT!bgfk9lL=L>u+VNTuCw;6#c(VnjudC?|taj^eILosIMZ
z(HI)vUY2U6hO8>-h`PUmcK#oMHx;<DP}W<cFRW$faelJHL;_0a(fUi*V-gRiK-0jN
z%G60;lg~tD%MCx@n>7|!Xj(noJ$d=Ozb@F7pI*N?x=?mqD!Xr6!!L%_X{G7oUCc4U
zYFZ$K?}NBvkbf1~I+cLL+YkTgKg*B#4Y;|t$QEguj0+6++Nb(Le1pLH*>$>6)S51;
z9VYwNF7mL)Upq53{Iv5Q9Me&+JmHu>JfMGYO#eQVQL9Y}Lqpd+@-VUEEzw%I=|Z02
z@5~Bl{S=AMQcg#WE=%yq8Qs#<;u3^{P_?a@Mx&QGpAH%22_!x0ST4a*g~p|z*K^`5
zICyubpl9_`CUUa)`!_F_M>!N!MR97X>y^V!y}^Y`%_Uo%lES0(gczl9ud;%oMo2@x
z$f4aF<=etL;<L(wA?aj>w}s1uGJ2F27OK8=st05zAsi}vZssWRK(%$RX``JO?HZ+7
z-ykgy)u3yuRj}fYWdHj-i%C39`A+#%G{NqfI(L;+k>z3{wU401{j}Xfl`uFi<dyY7
zNiq+(LXx4oJy=<d==p<mUSxv_{{#x%&;QBG;Z!8Ec{!Z#nhOnp6wYmO#o)<vHEQ17
z<Pt3ZP{EYTi)RMlh5~at(~-|EN@=@I+~w7}n4f0H-S1J{9QGD9slmLrMgfPM9;GBK
zBkn{`Bc!Iqv<(9hwTUpMtNh*D@g>*oCqRWo#OyxVZ-L4HUEo|oXwr!ZQ%XZ4s$^8@
zOS{2r=#-e=NL}KF)0LsQk8CvLyj*G)M9zZ~ggrRC$_J910{+sZzc$+q6m3x67YG)K
zihI_vxyytxcZfFMrEh#1wg5gT-ejql&|A`?65|4BU76PRBm=&EmYhF6#lquTQf+2@
zV-j2nf6nK_MID!|@>rgPbvq52ey4G296N%aAC!kTJYmbOnUK-b1Ba|<8fl$82RV=h
z0=>pBc<ScMCQ>3xkyV3gisp_J>UCneBXzek+tE$~l=Zs=LmxSxMLy~q-yR+GuDh2I
z4dhW`qv^CcMM5<>1vYwFkVr1#Ha`4@E~G2NMA6plYI3JFoEw*lrZ6f{j~CU;$S+D?
zyzwc@<}JN%hW5)U@D#S>Q%mhi?fLtgw0bSK$PHh~lROU=VE{6##yXAIq0q%Mdb7S4
zJG2pfJ^I*w!#8DOKn1e4hq0DU?}JVxS~>5|$-l^ZeX2YGX^sq!G}JM8a_gE1_2-w{
zy)#5iU1{6Qg7tIbo<-%z>?)m^`F&Vl!wH2>cy&zFb0ES&s*yX|zVqyoO4IgrZB-<a
z{*}S`b%gnuIuJSHj?H*0m$Xa(<P8@##Bx+jpzY0KFc$j;CFM71!sm9>0VL_3nF_D?
z&#ut<+1%Q6PQ>hedr(^o<1x3JJK{pFzF8&{W6wXvw}lIUu+C%;>6Ptf3R{gdm&-p+
z>dju&!YfAaCb<Lh`2ZPX?%cQ?ZbSJz<XLggCn3~?vu;jK<_BI|BDL-AX$G-`rNzr7
z#-J_5q~qf>M6IxADH0CT5E|fT4@Q0h*)kBP=LUwQfF9MBYGyWI{SK<FXQ7dxs`>9A
z?+UT|bL}8+$~bPagxA?i&6=5C@w*Ft)eFd@ihG$ztos-98_P{2*>Yl__^{rX(KC^y
zvaN-}Jh#+Ok;<ByW@g}%I|a55yvBRm0LRCV1wfH8w@t_0J_|f`uQf%U;Br*d-1DJ{
zU%kY5jM#a&+`Gk}y{Z{~Xq%d^3o+YCN($;>^3D9vfznG%f8qC-d?WT12@2>|x!+g2
zQ~0UXz-&i^fQhdEg{)3Or=;ih06V}%u-bd^)ieF{N6M-0rnC6#`gq7>v&39`Xl-6Y
z|9dbQXW1K7ToA>w_HoNivxhXv6Z+-^VAjx?Y<vUO=u`u*Dr3jMpmp-<$DL0svYo#6
zDK{TM=&6Wn2zafvn=n56^!u8E@}xdoH&&CZo|AmM)xx#rwq(7&P_*Ou5`yJrYDACg
zno{fBRzD0VSuhZYVh*W&BqiZV02*!{Gj|llf=d$=4v$Z06d$Iue7Oqnj9QEiBOG+N
zu3U{-D5(43kAfC|5vCM;M=2Dqs>;um88BvTO-c|yuP3yi)yTA|Rn;C^bFa`h$B|^u
zo!EoV!^TTI=&n+_5P|l^-dV!0ACI<&#0O0c3iGAVt5fRbrRo4ANjbFI`AWnd71`mP
zkL>y)Nu=hQ=_N`K^<g{=I=$eDZ)??AFUhZQ{qS6ZIM%jJ{K~}J)tgvUIL-GTy$1(R
zRt}R8349Hy)t~$M$Qy;D6T%vrJcX~}LaRzdSS1HU68GrSxnA6F138VbD}GP#%w8IN
znmgnyvNcgaKs23Gu;XQ^ZP2%yH%v_;y3+C4MY;1Hg$&U=Y&Y>(FQ8J`*W|^<PNaXz
zrfy?%65z2a!&g$y$9JRD8nyTL=WV5As&%K!2I|2j60XqG>j7;pE{VYkfdejFgN5Gm
zN{$V48QPu*l$is$BhPhc<Q?wk9;I_?^9n(2-QU_mT|pY6*Ktms4H_BnvtDb<<>gd#
z<VhdI2#g%de{=J8Mp|E6{NQfCS)mi&^L^UY$Uf2Q+9FymIdeNK^+AdR9_IgCOaGg{
z{|)!b4{CO3-{f*I4M6MKXhwCu2sjtP=cDYB6ox_W>ot|_d1{PV8$m5L4M#(kqf2&A
zxJLjC#6T<)7;>EWte!)U=>s*n3B1ZB-@#~YZdWVK$n~ybEu0@!=>#e^K7M)1Iq1g0
z#YG6EB}CW$`T2LDfB3H#MNp`ArsT5!5k~f3EAdF>hIXcJ$f@tfKAHeok-MjC&tff4
zeLF<$pEh!|Qo`{uU$S>GDK=hCwP=>IDz0F!44)*S&tz*+p^mupEyn=c2AA&O$grYD
z|Ay2TKJsc;iIU2ro9o)=rz>CzMM2Cs{I8_)<xm{LD-q?;Xy<CH>7|6aTbS4ZRu2<=
zBhN(g0mZUBzH9h$+yhNl%4ho!KU5}?1Sm)Z-dg;+TIROFWfUao>aJq~XD!M{q9FGW
zZ=prSn3~|n%Z%h;&zR1tS5pKg*B<Tt1EgU=aj6IPT|sFM9I;o-O5Hk!kb9mFs!k6b
zCE~}T5TTj8TaDvrn(F~$@ym0ZUwH4XiCtSHzH?`Hpt;(7#Jr|9*)GGvcD!PyDB-Em
zmyYSwaN4?D#j>ItV=gvPPTM#Uc5A%joZ+RmTYaM`KJYDu=T&ic{MbBmaOAbFcv2pE
z7~UYI<g+127ni10k}I+s6IAH<rLk1|E#QqQ008t;6DN|Z3u5yi@N^xMdPKt;egiD4
zO*Vn;4d8mmhlY(hvPK%Ape~S;JOfY#Gn+eV)@h$nkcxU}C|hDtp?DX>`$ARc`j9}d
z_G`tWAja9LSw|^b6pq-#0XkhJs=lVY{5j?6U57%3;fyNdaf#)~wQPrE5j=mEw&6_j
zxzqcMW?@Q94Qekm=P5-;m@3cAr)#++y<PSM$hoT8>i(PNCgh5gll5B3*NYTfZZ1xf
z8oDXj>`4eha2ev|_9#7rYRAHS+FDdqweFbS8^Ft;VA`ON4^dzFZgI#IPSNeCWbX|g
zMOd!5O}zRZsX3nb*%<6q-T#8!YhA0PRdq}9bzI{VEX)Bf!Yg8HZ_Dxm=Gr!PkW}43
zz`Nh!1OmIre(upcfH&k>jEq^_-`O_bVnCwjV0IFo6ltnguPel>lGspmcW~39EwDT3
z@?+N<?0Gd%SIhM4&M~z=v^CYyOgf!@#aDf15@OuQl*m^IF#pz>3+2%A5)Hok8CFC-
z$?dU*LU8tn7?Ud-f`*zetcs?j1nbZ)3dIj5&rO7fNJ?+;7d}~(6j_9;n%v^#*q!9J
zvz!!Bf1j2-YYh{9cx0%kzEZP$H3{T8IHIn`q^#8QBQrH7MBQaKG7>SHY>aBTwS`Q^
zEN(XKbfVbPSES>W#-TkrwkO|wzrc&L^|DEcU((danDmO1Vu8u0fekoyyy+uaJne)!
z!^5VsIEiqbU9#H~wbj~Ntjxwo6FoZPcITlnRXOWZ2D6w!jYrSZX+lEtV^q@DBexI^
zbh{i-q?Z@2xOB&1Tt^x_s*cOhC<cEfz5Dq?x<lMDyYC}S`)Jxi6lLa;1IB>QJi{sD
zz8SISn0*6y^!lbL4~XehvW*bXOf=^*Hk%hFGz;4)O0X<L&GGWxDzr<+!Jr;D-tJy5
z#V9L^1u>O3%=VjfrId?E+dep2@6k{m)H}-Hqe7BhdDaqt^LDje`z$xc^}32(vsb}x
zS(^RC>#KTdesxfu>mF-b(HMTpEU|p|nGp7ua*?xG@R6@_%XG9Ktch{ML^TgJ8`oiK
z2lD2b*yYL(z6KcRaGCrKuq~{G1w8UY9ymVC&Qnya_{OIxiL3~aTy*><6~h&Ai&HpF
z-!6b=GJ1SqVY-1ML6_sut>OhLjAsO(s{Z5&GEI$0;nD>9r5`e1T5NiA9AnsWs6ous
z3+^SI{`fJ7mgYTc?NV(DqJA!`5@R&@l>MQu#33A6?00IX#_;gQipL4pM3*GKV7L@M
zN2>|`2=R>1%Rwj>OKy3&R^4*#HDYqV%bLE-S~{j*4mN(2<l(Uk%pFq&6U7Px<N+tj
zGjEaQrS*_TV-ql5Q=@Oeb;Cd(m06`NMh!oYiJqi)<S*I6TZx21Z8v$gkVB7`vv>>E
zMs}m_^AXorvQ3=Xn{8&8JauhCzbI>l3AiIPc96CTM5-)zRi_C|yf2vsUXj=-BSt)?
zqUj4tON)1p{jRd8o7XjJ<!Ln(y5(mKi)$i`_6A{tR`%}U&eNHg$uvZNKh1>Q)2i_H
z&5<$rp<STpyPaVd$2fCABDdwsgSld{1iwZo?kR*)16hm)lPz4SA&%NvC;MQpxs^d%
zSaE!|<&^Uo>w&X96L5mfYT9~KDKr?Ws~tv&lR7`p7Vjb}N;1NuZ)|&RCCzLx9TT8@
z;#Ly1ng*?$`22Tih|lEvJ$hI4z9Aj9{K`{H&HQNS16ZCa6V4xtdu&gDbV&_wSA#w~
zd#&Cc_a)GF+0n5SR?g$TfAmP=$-kuoISy83XZ?=lv4cBh=e6)&d|9ujhr}BiuyB9b
z_^7+X(pg2KheXfTEL`JB&2KzaYu(hlS+CDvqIv)A%dn0%@Ec;g9>M#YV$s&9AG@c;
z36Qaw@`squyd#sRicaVTW-=XY2q8pU`K2_bVO(d6l*#SVv$*1Wt3e7GnTk2^F=^%F
zIF8-?NiE}TGh&J5QGXLu;_gq%Dohv?ycR=8gF6Are`Qb|76dsyk}uqCa&tD+n}?0-
zS&U7jX&h<goLLu#G><)q9+|!)J~S8XQ~ku&E0S?^OY}oJCgrNGlZW-yVTQcqO^bBB
zn1Qy;n&y*dqju_oPEFI>DW0WlW)rhpgp~)wncp99g$L}~_#ryYI$*=Z&P={cELOdA
z%OjC7KJrtEM&t}M?Ycvb5)Zrcjl~!+sS?=&Y>Lp;+=Y$<I4pl~L0?}{$p_b3iZ|yk
zULqzFb@JkGV$vPs>GDfzVg2QW&EMd9DB5Tz#;cObw@Q=Av_>+bmpGqj{+EvNe`y>4
z%d7uDa7^f`{ts?C5{q{cL}M7!c}Mc$7ofNl;9~#|IIrl;oPISb3LoLhr`|L{JpaV{
z)>3o?&yIRjJ^ur`l(CyiFDxE3km-CC2ZG#0I5>bBj*cdg+q^xMHy?!5MVCr7X3We2
zZ^eC>0rj#vcGVW%{GonJUq!l(x@8Zqc%1*dZ&ky30z@z|viJ;ff8|h(AW*SPUjfH*
zy?fSQmyBjHpWt(WgR-hc7dNH)h69uzN0lp~N3N_R?S2EG_7C&2Hc8!IB!uu){S#D7
z01Py}qn7@!eMP}Y|8%8Q#uvu6adrG}LM@%c@W0I}|I)+!ZC1&C7tx;ozwEaEb2l<4
z@W$Sv9gOVqzn&BxY(y4THygHk=v&+<zNi=KhLpRK_3?hdHg9Wn?qoGd*DKu6a8b;P
z;9aR(7i=igq5zw#$^i=P*aN`vwn??fV~m?m)L{&fQ$SQ=4HWEIi(EL!1V;<U6fH6N
z_J#An8z$lu;?qMyBA9Pq_fG7~3m6@&$vh~Ceb!nP^ljFL1If@=MY2OnNx+B<=CbrO
z5opX8Npzdle&X{9Q+v<{<pw#MjyiTZD)co*O09AkzT4-tGw+a<YEUJ1>g{ygT~im=
zUAKAU-KQW)-n&O8A!)3&InT*bpZ;0R)@F83GmMe+ZcWI1#z6$LL{&Psq(Qma(6-35
z1Dp`q@XGD${JhKMu37uTu5IO2B~^mn%U?G`8qoay=LW%_9N`lCIRF`dR8aJTIAj;e
zd+f^1!Vt7^^JkVcis~=IwATc}G;Q1e;sg4lMbGak`>|NZ{{JQ2A-u)qsILDX(MH)+
z2N)8L7zwg1xu}Qv7Q5&2D^XTT)QR3UPJ5dWOqT?VOCoZ|*tq+z-nHk2##4}~ii_`L
z6y3_Mxsj=4b0_UxZy4HCry~(aW%=4T;E|ZjvERY8pc}XCb`!7mymqBfNmyquID*(?
zr@I$+t=ClH8yKlRR+(5tJqdc0E>zy6PP6ogWiW(QMEsR0d?G9fEq%(&GQRA&Tc$dv
zO8i-~pQmVxwQE}{1j`p;&KvG}C*nF`ul2w%@J99Ql_wXKCKr6I!Isq03@-)<#!3R3
zW;(rgnDifGCHZ0aYyacF_cE*(mId#d?;QPt5#sh4RsY!l{h#}U{-=`3xA!1>bF}oG
zynzqByZuBb{7#-e-S4aDPcAG;k>!0bxFPs4!<0q&x#K&gh~i3e4h}i3YVio-ddGo1
zcA$cmeAo9oh4S}j1{|qLjvUF!7>J2Ha+=~Ly183EH4zzH^Q*2Y;)AJ5f8SMRw0o+0
zXI+H#Vi&6WeX98cvf(9JwXH@O$|Ds4T!ztzNA>ISy@TF+p&uS!9i2V2#wS|_ECE~H
ztQmTqh_~^nih0eWd(4e5BS3z~CDSF0NFi?Us7XqMi7Vane39@EXz3rD@L%1`^QW6n
zFtYt8e<f(0_$ta|l~w+r78Clv8JPc@P4WNk=J|i)x!W`Ir3($bkHYkSnf{16_9{8q
zjd{P6B%8voZ77*RZGLaP(2&T1wZ13_qe(m^O75V9nI;&*+3y_+1FwxG5+d0A^b)NQ
z#>{U3C+i2ULvFp~OyQQ{Q*#MK+vrf?_2^Z+XUx&A9iMt^WVT<o5^&Za=|E=60q3HV
z<kL%n?@^MZtNc~tDOn+S*8YfRHGk2&4+s2|3^u*PYZw5oysv%@2w>?||GcH_Qk%%5
zzTkdmTXpr`!lfbsD2oTp{EZ!;ug%U~m~nYM$c~2_bZ6C59XZQD_pL9U%JlT~4_H>#
zUtqCv%AGQ=Em~8YEUpu<V^bgb2--FDg>l^Ttor{3C-vV_o=9{}su&$Vce+<4(Nit@
zVUnE#Fw9?_JszH5FTMKdPSAix@TbGxaP6}&&r}MOPfZeO(^!k@a`eChrBbJ1H;<)8
zIIMBt!7n@g`9uuW>+&yfi^VaqK%ug41rh`Pi=DK$#WJp!B>S5Qd&}-R`N!kDeah(9
zH{Z7t<|T_>&TjOU%8+d0GWbXh(L!gR(lp59C3{4cq`Ejv%{-fN7tYx*&6Wv<i&&kl
zY_Roq@;u=>in;cK#lI%Q`+_6rbnVFHdIaBE0UOJS3EQFi5L%nXiBModn9doOvUced
zFzMIOhlL{NwUiIIwJt=~s>p{qh+g1>vaKTmmPxVp(CtpiyXmSW3#E^!(+`VmnJ##{
z`ob;^#eyr7?v2GPm16PtzyYBA7_`;8i`Lhtz$c&y6SFLd+<~l<;J`nn<{)V|Ov{RY
z1B$9o5wZq^Qj>2|)XE!uC@-Bbgy`zF8eFaqF}*-ZWM{^ua{*neGke*qb9s`2*XFuH
z&|mU~FAkeC;O|ETVPz|pb{}zxx!RPrl027A+2z`cSoz<8gZ|AC3Hv>mWChAz`!}G$
zqNmXAR9(!PUFn_Oh=k7RY|P;fl<?Z!6zmA&qdgqI;*JR+`sMiRy4~z+a3+m#1dVRz
zCGiRVT9Iss)62k9YYpG)VJpLpxG#bApJgxNn<iwC7SGb3yse&TDPwDp`0MKmMt(P^
zD;zGLyuSCC5E7iKq)FR#Dce~J6v8L>`c&lDO*Dk#EbLJ6)$tUb2ZD&`XNCp>0w>+f
zySifAu`>~*zX3mPYaALZT_P*u9elg=*pC0cwqK6^<}XqE^k>{V-%dg*wDsIXFK3HX
zD{uz(8?Y1Y|KST$CW{Hjw`sAsIN!F<gycO!YC=xWy7oE0*=1!#yjY%lXX#651bGh|
z<0;bQPm*V*WR7VK%K5na&)^(yH~3omss8$h?%jms<kI6U=;>WkXPym2$!|a{>Oh6Y
z$pun%zbn)x=6PV%L>3l$%Kl@7>KIymZg0Wms=RU4gd=|}o<<_}1#?%Z`br1zi3g$R
z$rd$O+T@i~ltgk3$`dMCGH9jfT*(<JD#klQr$Eyjb4zZUrTG*MP?HBxF#+lP8n687
zBkW%t>{rt(VtNf^e?seSxnEdWaAn^AwvX#kGEPUf1x;$TNvwjHeS6ONi~<)JPw&OD
z>KM>?Af4K)&sQwC2w?_<#?)&qhxA@oRg?Pcb^{q_F^6Z8{RwU)cy1X@4wsn9eC}kB
zFsSh=WgI{lWeIOaytOdu{0^$27S*{@=q3B|0u0n~Jx)4XX9|4wjZx!0nO+NUtWRES
zNQ7|WAG`F8fRUr#H%jj0f6*HF@LLPr$qKE^n>{N#Y;^Sy{&fkxNrn}^7jwj)TRhQn
zd$FJUE{vBr%Tpy^IN1X;E?UxUT=PLtFg%mP_l``FC{=UGW<UwVzX}(i!*)K4s+`Z4
zRrj$d{~p-8pS6VUEZ$JfjMFauc|O~0K0%0c?K9IZ#*5BR?49eYoq<~`Yy$${X0m7t
zD;;f_7m4JY%|T2n+6&|^4iKzGv#zpb7(?sV*BxUA`DE*aWNI!DGA!B;y6~0(;F|0W
zzosme1FdmfT3%0vXEPf<PF5(^+rzx(B7eWcM*)PS#^vSeZJ0h%{|zYl@Eh=;B^DAu
zk|n;?(K-*f7;E|Q9i8=IN&kRL`*|#6i1iAJ-u>6l*J?hjW&Yx-SQS^pB?Df=c}+I#
z!Eb;~s9q1X^v5iNa;CzJ5~SG~s=pZTvQ!dyWElIpcsSW`-tpcz;W@kg67L2O>Nak(
zQ};I@yXXXGdTNdFZQdVkJ$5DC&mu6^3~~gz+XWb(svLMf`5~n3D%w`gEZnc5o7Dl1
z-5~H84^vV1t3qP~as#Cbo6Z7xL5to`_VZuD3E^Tc@8t>y$3%jD0~E|vC~BT44}xrd
z4RHs2uf}P&?OQKK#Z+#0?{BdVoGg3W<&Lgz+Kc@Lz@If_vf0tNjnlpO^*uJ;a6Zjq
zB`Y1&p0OZ{PB;#)7LD^$IwKPjNn);<NISnB$kj5RG`{_YsS-LGV^V{R3IKynG+X)H
z4?_r9q+NVtD@B-$S3|S#&(1i~xRI>ZwV~2D52`^)x<2Z@$i%=g?Oq<)hT;SCkKX{5
zkmDB5Yb;W!E+=s|3NZ-N;w<U)i39Cig#}e{uwLw$+9}8~K0~T_mSvUv_Ic@Vz{Eef
zyslgOHqA0t&s$q?)xGDQ=zI&^nwT%N=#_bQEiO}4b&pHyIL3tA=0yKD;Qh+4SkfK;
zLuYjuseeDqpc&5kA|Anb8`8LLSK82sd4CET;ctE=a$+5@S#jZdnD|rHzD;dusLC)6
zQ31Lf4dklJWY~hAHmDy5&e0!e`|oTC)DrrsbtN{vuswSIx;BOIa=jhX@UbPgqW_&`
zrfix`rgAEi=l-w!svXpaci+$F`)gkl^|q6uT$X5^;r+32pDoi&`$}D0#WA$VYcl9`
zqfUTfSN(jn2+z6f{rCrCfJI$=#j)M!f=_F*sdi}(Vc{CKY6VYU9AX)Em12GHGDumr
z)%_RCn6o-pev7l)oUKOF7YB#0k<6+V8TNT9O9q^W+OMqY38fD2IvQXu3Qwi(d+)H8
zep9p%;kUiutvr2Oq+L?mEHzh6b&_yoQ^5fabc6jGPC3^OrQ$4{&@jz}ClfkviZD(I
zRVaCllG+s^TtjSr8E<J;a?VS?`f(t2-$$)gA8C{;p)rwBIVn24o^&6tgdrWmOy<X|
zJiLQb#q!A3p{g%+4?;Q+>omSOW-Y%;v)MwPRTI|gvuuNa9kXYWn8-7|acaUBEZl<b
zxV<!X2o2C-AxsFZu-M;#wMU}3Hw$N)rg=Ywo(B#}vj9QkvwIdxq~RLFWoJ21l`WK}
zZ!1v-@yZjzl+fVdGaI>GaYmRCI0w|V3-IHFWnGqPaes4$m>QRm-;?Fr#D|r;7$ZO)
z)L8m&C9PTDb!68*)N%(J6Xxdl*KF-e<5f2}Co~j@lOX0v-}c|e5?099rT)_LKR$T6
z@0a|N=QvW=M&PRn-0`x9=b+$>I3#P?g{U0XvCLu#QgNtJgm+phkv%d@`U%nUuL5-y
z;(S{Vv<IbVw!@0?@IP|`ksto1f7OAsMI9{m*e>by?Q*u$e5DAsB3Ns$thm^VyLmyT
zBF@PP+0qAu*9a9mhnULP4~}`ZI##_F1c@bPIOY0&%S!(Z0NvnLa;Z=hd7QyxbG$ib
zy}FN8=1fki`VDxg(tFP+aoySX@Gxt2DeMX(shTK3&$_>-i#||KKFZ9r$9wWYB{A<#
z??JgV&l4t#PlpdIX?Q9LoKD#LD!v(PTstDAzW%d}^7GcRT%{4(z~+th^am7-`UeHi
zz0$OMB{e1QA3`0x;s{Z2VkO@?Gcb56OJ}U0E9aBvy;}kala}n9)VCN|bb}_SuLLpt
zvm+2l{dl`zw~hI>(yxWHSKhre6}zBwY}4E0*ZJ5&r&eOWgBw3^)XShI86JFVt<Apa
z`O3^gcu};linU2*m$$HCEAuh8=O!+CTqckAh2r9bRg&dzz$@qWE$Oq0nWQC9(coKU
z-B5sKUEog1<);zL)`gzqdvY8e)SxE<OYdV_15dcy20It>j=9}TGOl&K><E4EE#OSn
z(o{qA(QD9OsR)Lsk2+!HiiEXVWm_Y{X=AUOa4KVfV6w*KibdUzpBr9q{MFTvyqkIE
ze9R5geTV=)d7=E))DM<+84-t~KtTA>-_*ybn?GAQ%rd3yoIi8|{aR%!f2Q`hZ#<7B
z!71;a=d2cYM+EPZlF6=?1{j@3_;mwzw@kgWrVXtaGh`0TXQS!IaFX5A-8{O=xGalg
zqj`tpy)Kv}$k%Vy+YU3*InqlK<~!r*NDYd8Fr;DDYYI(I&d<?Qo;sNg{GofQ;M<q?
z(@Pd`q1R6n9q4ambEf}VC$li{uQc#}86wy%ly&JR!i;`DDljbT-7`f9Q-qFzx=R2u
z;nC9z7TKR_Ki}@R`Y7{5&o^dn%*T`pf-jU1C7UNb3XpZa=FHyx<nJq`;bOkkyhQ^j
z&DN1T#LzF1jdZj@cv9mi5U>#vA;Y$^_$jvaYhRwga22?iDv8g_-;v=*{fXb)S9R7+
zqR&mvZoJ!D8B%2UoI8-`%B1>*Jl>DQpr6aO4LSrTgmivsQB+5VOOnc18a(JC7=G{Q
zuhcWJfD>~I0aqPKY@B6$E=ET&FXY9q7?JnRW*QGqf!boBH%e-iNeUcqh65qZ4T1v~
z?xCWFEiQQtkXawfDu1#evFyI;m(SrEj@Im4IaX5W7}p2fg=24DCb2Hpvu@`@qn2Ox
zRlIzzt`&W`*E+Dc=e9Sb2)WD3qoYg0oY^w-+FaW939=0r9sda<rJpf(K>|=Ul&M_U
z9a3lGVGliEQV5%?q~i!Ixlu}@36ji{`cdm-^Ra7*>z^q0e_f~kU;dR}BVw_DUru6+
znF)?2mVHhfakT77PK?6RQ^Vd=FpH(##&XQClj$I*nn(l1XB|<uE$n%x-P&>#IUlY8
z8oFhZoXi6DuSs$y0Sc_HMV(Yvxp(qE8C>gz)(hyo)th}VaerF(%;0?q;3QT+`scXO
zT3vzUO2tKr)R5jTYT|)TwXwBjWK~Q3EUCGPM=#0a(W{bMTyF?Iy#tHGj}WZ9DlXb;
zuu|)JEWF7adE_uPX6l$psUIrE<QghMA6m#sp+(~+zMTnI_t%}jJ$~6r^R%Q+X5jK?
zJR-Um+3TO*z2_Ei4Prc(;nd$){o;ume~=_u##@b`$D#QJ9=9L9)e7dk$ILBQlRq>T
z;lgo4Ox+q&;K=haTCYgA@ILT(L-nxan;pqEh)HdiT;SWQ4sJ|=d;`)wu*5w;c<Z^1
zCG-lrTAun37VU;P+Bumg?-XD_@yx-c)2^D)$7Nc&<TaWjpx*!q@4%zZ14I|sAa*nj
zyTNw;B#>9>p0Q`ba>P7Hv+(7f2VPBKa(i;!T6xv!s}{|p8Va%};acNMkVvj{@y7n@
z^nRHvTC+{H!h~hQ3KM&puLAbvqX2OOi*nucz9lOpqQ&|927B9`RP*^Ytc#&0g2CSs
z)iaYA%B(K_e6#6AzgUI|v0jI8zQFm9GryA95<&STvVVE)@l*6EK!X{L4cX&g7iXc<
zN5<LC6JM*wuW3h_*dDkVNiRNmrIwZ#RTDNgE2<OsA>QYT;On&`^E@-BZ+6S~EN#To
zeG~^raQ9IVQ-^_PhRH0ykgYG<n#41Zc`O|^U;Roq*$iZgd}WJ_7Z&@va-R0&A1b&{
zNbG;POzb<a*rX!dG|8-PZ8rwSb{Zb*u`;$+eLS%kV=W8Nai8@1$;V+8-P^7=O2g@q
z=L2$bPiGsHG;7z)N6KH}xx;l#F0*x}I1yVyJxnORJmTG}O$Zoy^aShRNiWz`IZ66_
zum2Ze@2Y3se)@8F$5~YcYK03HKDNmp?v0Cx-Jd#3J*Mhs{r3sKzw0BuC3$g&>?6U!
zF|fGe`9<Z@FkX0@Em?Z%D7=(SpBZ0qdgJZ0z;Id2)aTd675_Z6+nO(Jbe6rby|@3;
z(yB=!@9(U8e`lTiJL_e}#_HFu9&fnHpBlZ2ZvQdxL^|)F_r&Dy1|NI<e4;DxYP<5%
zkWh4cWAk)!^XJz_Cx&A(g#L)dJlNZ}Ubd7%>K+^askcGV`Ga%PwODunC$49L6i4Tk
zmGI~`^%^X;YM79JDk+m7jB@FTqgX<{+5VQm7ay|)9QJIbZ}&IAU}1`tfWJxTzuRY_
z8NY6nJu^(2`;$egd=Kw<>AG8yA3$i+rNYo}z%_$=eE;S#^g65kRrgP=Ugy{ZhVb2{
zUGoYob>52QYrHfGH2p{HvN{CGCPhRZG`3El@q^bp3on`e>!u2!zCWLg3Yh<0+~3di
zG-Q6jTj4_R*9kH(`k3JD9hm35l}exDjw-lJV8!|VsV&vt9bu*GNwFUHDL^>Xiszg3
z3P+&5ud-}`Oc7Lg(q_;Qn9^a0`P52o0e<OM`PVqpmb4EHF5nh@TTv{q#D{zKPqH}H
z<bqa!r-?g2Eyd2j!pqg<8u<)Ipu^Tm>x}@NkCV_(-G&_c-n_5Jvd6h@Bz@r7-EEwR
zO_%UK;#u~{dzOIOKB*n_NbI@fZH|!H*}613?PwZoo{jLK6#D*yd45l;iB{J3>xsJ=
zu4m{r{r4<kDGeOrhZZizPw@Wxe)UNGII;w`TKwVI$Fm$-nB(|_FLz9anSulyPVVdp
z|3wgdp}0N#XR@y~ocwi?%q@Ky*O=nl1j_=;Iw6Pc1ol&xhAyu^wXB`);@5K<T+)fx
zpI6cU_^zi^wEyZipiJg#>@^jF%;yHZO4-CzVwG{;Nvt<w(fFPa&K-i7*Z;H!)RPC7
zda8BDNL_2c+d6wwIv6VJXuol0g$o5ioFuJ1xyuvr<SmJKvJ)E%!f+#7-`TAPE~xvy
z^(ZV-X_O!gCz&GgM!#|;)L3SY=oo9!-_MbXI>QOR$BLMHC&T(#e>W)8X?}OvWRprK
zZ9!mnsFfhnn}1SVTl_vWS<Phx{`t)2n&-MUd`0v9P!)d$L>juWD1DD%KGqZUr%MQI
z5b$?hQ)t$ie-dI~vmrWx?4~VbC#!eXNMdQFm!!qJBui@@X}!v=Ys6{<8h}lUp+)Qm
zyZ46$cb$GVn}E>Yvy>M`_4F;GURkjmj@mYA25h>%>)P{#`EI&Le9k=D@^obH0w_(Z
z4SXjb0e%H!yd8Wy)bZkf^z;=#aWw7LB=7<uxCMg4A`5}wF2Px3fdv*Pf#AMqfB;D#
zz!IFrE!g7j{^IWL!Ce;#LGthS-TU9FshXPVsi~f>>FMX3KF@i$vBoNGWCd!G&wz27
z#L3RZ^=sWt1o<xwmm|`@vLoOeFv_olRca93>~F@5eqs?Df_*m4#y0Z5-KL6#C@>>(
zC17*-Ri;RSEM4tok?`RkcJDHCtOMeDwL$Vuh51b_s+*TkB{wC@nbr1?=Br+Ziu@tu
zYc(rXt)Z}OWiRND1cfaz+a~*T8!B}Uuorq=`HPXJn{^*dRPOb)+Id018_i2SW1!CX
zSn{TNl-;rwYlo}~myarA(}k2DXiv6DtyU1ae-uC*05tl2L9vEd`Wz5dvw^c2c9^n1
zKt85wwaFH?%*|o6n#z$>MHTn`8{A1HfmI-KVW2@h>r6c`EGt4&Q8$oq4=n)qxlCbL
zOZkgIB14rK^uH@j8>cG0WMD_7@ZAp?@(!Q_#iXu|I2glFR;rgMiL}^ndWu1c&Z;pL
zu3K-_Gt-!nGH^PqO-8M*T1H_f?aFMbHd+xE+}~4x#U8H5gU0N<1(<b&0d{WrS-1tT
z(k*|{aowK$UY=cVP&fC-R^5PithWB0x8n`mg6bb}!)ub{TKV}5zIBBPedLhZx*V#M
zrKXMO)1O7_PN{KxBwCwq0(`VMO;j+PkEwQ;ZBG!rrspN8h<5Tl_l4B+B#F$P3aqkT
zyhcO-@!rxO<c|``nUCMJd&&mVlZeF%Rrx_=DU!IGdYsvqiJEnydfN9zH&$VZ8$sdv
zFUPnV=ngsoD8Q%!!ZIvO-iNQ%ItTj}i4qH=v(y-4#`y&AyNKMvY)n7vyCT>X5F2^7
zPX+O_u#Lz3a6P&WX6x)Bh|e4xQNUWz#$o(*%a-%cm0oj{%_ncGD}BaeBUQg#$85ME
zlGV1@93fxUv|CEO(xVnu=t-LhoADYNZA-imvD{I733ocrkHNsPORm+A-TO(upA-w7
zdb&9{9rM%{?amkK0UPR<D;(PfmRhkym31eSmUM7`!yRCZ+Fr9_4*PysZhyFDqXTae
z!0Tzo%G)L-GDvwXUY_jy=J!@Z`A6A|+UjMvzBZyby*r;`R3On>%fzTF?icM1^?Tq^
ze5(5j;}|)ZwRqb)<s;i}u7Es^Tl4B@z)W^R{kmAdOqu_yy1VT6yk`MTfx*@m=Btlr
zq84`y+T3DxKxo12Z7qDA92UV2P*A}i9n}bW;mn5og$9Cg&h|{)J=OMP8E}F6RYDtU
z+#o`fNW9p$+5&UYuX8r<#-8DAyNe{#DPSfi(2T6?I=@4FYPfH1`->4A-IwFDVFo)J
zSAaR(diwpvc)<{<fa+@9T2B!Qc)!}9Hf3UAkN%<b%{>#wLLUfWb_-i@Fkz_WF<4p5
z6*O=lP$l-<ix1%F7_)A46#g<Eds$&=W!0PQ=dnuP_^XNzHYp#^!iGiYER|tlp%lF_
zH3wMo<H};qa*Tt7jS&!jTBgIB?$rEsNmXz<d!`YpvML$sE%4H(5$E*TE@j#7zbdYD
z#Vqv}LBVJvykQ-CV0yw@YA$XO@>BVAHu){`kI=7YS`$N>8-(Nb3dRKTLP@Vf7hpWK
zI}-M)Q*Pw^HmN$*O7u8T;Gtin_?)JEBZTM!;P^&CPhI&Flv9yWc<kRV>;~!iZtU-r
zu)dn3=hSV!a27*O$#1!Kpq}<FwPr!+t#9sMj2TX5zqmmQKJlJsZ$hf5GM<UF01~QG
z$_D-1L@cvM`DI#bo|pspSay`pPXnI)&U=}!WN-N2=^s1wKY$T3QvzYon+{L)to3ps
z2ray!c3y|r8+P^4JD!SW=1Szkz?I}Rt+K#zGeQC25irAP!w?jOz~`omB1-hf_pQEb
z$~ye+q3SDsQ$n^$HybTh=ctuH8c#+BH)3ONB_Uw~H0WrBLv@crEtGj`W`(qMN4d~0
zdjK!b%&T1U0%sm1GMg2jaR(Xg;xRl8_8V8UFBuXq<~}Z~e;M^ux8w2S)F$R8_{0++
zZtRrboOjnljB#o?Je9%Td8nJ<C_l8O+A$iLdl<26t!}|gTY>cNi>b#WPH-tadoJuo
zBT-$<47CA<b+Vp#elDH9c}tMZ9olRuvQ_Zu!=8jx5$~Wx`1+{kZDNO{knKTu#iWLI
za!PQjXnqk9>r^I+;ikViD*p)iG%i3(r>g6Gry@bUD{!+7+Mm23xo4)ItI^J%UQIK@
z0u%eKDnlNNU5+n-cFlXIiFmJ<GS)J7zs)!E3N%#fU#Z;VKE{gEU}JaAFzp=HL7k_`
zm+Q#K7eHh>7jl_Pm-{CK%vD#k+&c`DfNOl8{=*EIypEVdw)4EohN7Yo2nG@bIwazJ
zK0d=_J*BfvtT3LJi1)kFD1jlP*>%PC9&(R(XLBc)kRyk6GD$-skU;givJK;_Xx-Cm
z$aLho4bE2dL3?#|;p_P<#te8#0hRO`R)mvq2#uNs6L)=Va6Tpm;V&{hd?DyiN}&^x
zVAXe~&Q{zrs=+@Q>0AF6gl@)rRFEFL<!MlcISJ}Hh^l_;P}(>g-1Z9rR};c8m07@g
z%=K_iatge@(?CeOU)$E=T2F?kOHq3JQj(f(b#t>Ms~Wrg#OEo$wpg|XW&0Ao;^~0*
zqKhJf4Eh(qG;X<l?lR`$G8{r~n=7ZxvP{^Yyvf3nS%)h9xrV*!E<{#WAI_X5-9)`o
z9d?dgc;{4gaM2z_^N24pv|VKcxWCHUf|w7u^|iZPHOTl__7-nFH=wkjI5FM%L3t~K
zPQRS{lc`$Wa_&0$zXg|;=2m$+^@%BRMoui~`Y9N6|FYZIb^@Yusm=U>HhBgGxPG~x
zez>-+F4RM(b-0jt-J@234`jo(l6i7_eBEFp9hmd>>Pv3DdJti~2xy>lfQCpcCWOv(
zS%fpZbh?`!c9ZAKc=QsP);Ogz0sInbz{8d2@@elAoGXg`5W-IYq_arluley8V_5j?
zTgl}}dZXvV=ub3%d1p=EM*mL+#jD~53iZDjwck&3i)fZQu=?+c&ZWiT8(;xHIbMFR
zD<HW!n1fV&IR3M(W^C(fZYviGiNpVQt(H6PYN?@!MEwqBR`84yr&)}_nfhREzvRSi
zOnL!oa4l4$&2RFv5D%gpH5GC=xlU)gh=_UJ9JQBIvGJ^HRUV7GKxelxXy7)Q8DKNr
zka%|E>lgHaJ)g?x8H!Sv<hVPkLslcz;<{**wCQuH*Sd9i$%ijTZ@7CX@FTv~SbqN)
zZWMD&2w}K|T@Iub2d3$3ervi=G3_SFF#4B~WTI)jl__`9>qO&61CKU^>)9=jU!nc`
z2DQgjsTt3=%Xv?uEI3Nhm061<lexz%HwwT%QE7iM`lD0nY*&5!+74g)zMkqd?<p;T
zWlUEABC7$`Zp1Neq%nS9H0VO;1Iiy;>UyF38<8>HS{Ia$j_vc4I^_ud?q1wAOR}$X
z*508@Qx)26`f9K@Ny{p*>Ln3bO_rLDXlwZsoh=8hFc!wS4^eG7&}zLLq1^**H3FGt
zO=GUV7?qD&f5fk#T+@o=(<CO-$yqJ9(`1PBh&`54uQ4+pe8C12=pk`jKK)N@b@NL*
zn)eQ=NB@kbzO+bwNpRL`qJ&1j9H)-dH#^ud^YblpIe9-zO!>^G%6^43fP3P{r3QJH
z)1XzebwVzH#DpomH=azu3*~M2I3d{cdt(;rNj7v&C#Om&)wVb}bu)eBv4%8?L~Q+3
z<9iU*=d`>c&{CZ8r{Td;AISmz0E2<l5qKD)XM`)(6@;2l_H%3x4@s2Ni*~W>)qsO?
zD8qM>GjchNAY$>ngD_hZG<!{e%Svaw*~AQwWI4*(y*!F)IG@7Wg8EL=1P1oYd*LcB
zp^)7;TMJF}Rrn@p@glhpX>EK|%$%a5U`drq{le5q!sP6QJeO0ud&UOW`yI3O-4G>n
zQyJv@4eAsef#Av|kOX{CY>XX`bHlbtnT#Z8#=^Q1$8stLt*pRlXV-jn?0zmi_nHkf
z#FjKDQP9M(o^MP8Yt5|U)_`&9Ih$KEv#tz;#l#D^Yrk+2-Qxy6ID*%X8$W+g$nV6#
zRyEgJPP-@U7Kd~xO=0DSb-99L%6c)y!nzi9@q%x@vF2C!D~vJA5}#1xQHd^d?of$}
zk^W)88CzsbHci7R*C|6~k(xM7?73cplwMibQa9nMN*X<`nss!$Toh*dN~41jxeIhK
z;)~9)=|G1k=U-d>P?QzJ_YHR9YVb+{k+7bgISZ^6IXVEiXVVM6nuaO^lbgcAiE||V
zWwsK;w7jUH5=>61gDA{H+`^9&Fh-Qh%ALc<bQ^{xPFXRroKESt)g6<MR^P6!`?l57
z=X@%oz~k^{0fPMqeW(c!jo#=sPwsw4++!<dhyqvg0l}gP(X^K4a`G=%nQc#J)$0$<
z-ocDZ(8g5kERGpMI=zC!NM!N`{5$@03n$6ldZC_ygiajY40(eEl)j4l35SKe;6VPr
zLspz8up*#D*>pa}t!~9VtNnhmlI3Lvyi4KuMd##{^@R|8j!6un_uZLp&8;uVC>>eI
z`VWvNz9G{+h_kr0doA(m*VE>~k$)hqf;VeiXIhVi2Ua~v5+nFa2G(evQ#`IMdy_~L
z&oc>Czy`H*0>c{h|HL2F{~Lc0KyvT9--l#HsElg=x1=2K6BL4Jt7fZ8jG?N8EnWaY
zunw_o0?M<FiNxA(4Nf%*ki{3_GsEva2*UL}uDuJ|_;&9W(!JdE9j(h`N6W;;L?c5C
zR-30C@$8GF(v~SWi)QkqxA2S-NuN*YjpJ0_Tj+u|lTV~(H{fS<$!CO(ks(W<Qy$y$
zGwn*@QC_LM?V-Km*991jKhuvfQ>14O8i)^{FKa<GnE@9cTOqD@Fph&0jqI`1Z2Lm*
zleaNvFLZb0-^9>*;<CqR_cQ?yrV=C|y3e%TJ>n#Id0DMLwG6??Z@#}{dPgjaQ>&`4
z86q}O;J_i1r@zE<8a*yt{Dy2o6bY%20&e@QD!q5Atqsn8@j_JzgCsfXJQKBKs-Te5
z6W>_&F!L9~#{o6i2U>wX*$1}P-EG(H%Un7=o+o}X{an8oIX)y*I(5}Xxu*hai#*<1
zB=vhrx2yO)%W%RFCpSu$J60<60gC5>y>g$nyu5gN3m85X@Ay4%#c<QTeDxi@LLR2g
zvvHm*GuW)88_oH8?h#LWhbwa1zfhHn0!fAfQ~2q$&trA#9Cdw6xN=}T&R0}pgs?FM
z3pY^Ag{RlmmAc!thVc&Or>_RZqsA~OXdgkX>B`E2+EpTua8WGpl~oG<CEVq{s!h$l
z$R?{{h&W8Ezv<p^kTHoC+~&-P6Qi7t{s8jLOn4ACCRbJuoC)OFWR~As&~R{399CN~
z)EbJJ^u$5`?4*JCjphc$PS9?11?K#^`t11{mLpOgGX#lE1*8ej`^95CV?U8Bw4!dS
z$}($#|Gn&fMQ?ta!d3sH=6S??OJG9jbuBf<41eBpKSE*uPchQnUd2=1GrC3JOg<D9
zq-b}t#D%&J+(m%=Ahv$cJZ+!t^2`dV^N@P7*;X-5HHKWzT7%67F98ixXxWO=bP`6!
z;X0P#eF6~J-!Oa>Pax#PZ7)}3Z|KinSNHuh<dwc_)cyclczt3%OPqe<>|IY1&uja~
z0-|6UWo<AbBf7RUkD*KRt+Tv|uY|I1;0@2FSJoYAv<~EzZ9B^>TUm+!!CFd^E^+vG
zpMCa3^ejM;a&!;P?w{I~{A&De^8U%N;WNeiM9Z!2id}ykpTI=Rj9&C^YMOVa-kmjA
zQvAtg!aE5aSs>A%p|s)lF9tA<86Zh6vw7Y2vVPJ8<B{bpCHzSBs|?kjopz$xBHbiT
zIX8r8=)1!;EQT=k0t#L11V)se{0Dz`%+8rCyylXcTw8zpxZ9I}w~F%=KAJA+>}7fn
zOU<7t&}?neI};$cZ)|GnQpmv~!JAi<UeA>VDeLr6_6O(sGHkMu0_NgThd(|QlJBTM
zZil>^2n&rTNz@O{6y3v6cPuzp>l)}Xl7;B(kB8Vtk?FvTVwoiwK>TArdJ{CFU?Omt
ze(N-_PiLQKnA&Eijqb_tE->%fc}6&`Wyt#5uU<7c_)py8Ag1WZKqf?M>NT~I(dLO!
zNw22Lpo51<OXSbJEH{nS7*w6E5wg{=MDV9s!IRI5G|gR4>hA{IAv>M{O;d?HKcEa%
z9J1ia*W^Df5#r%fZ@pB9FeCSjC4z6>IDu~7MM#~XYY$;2Hr;sNS5a&48}}!AjIxx7
z`NU3${RS-%)$>|0ke<=2Pdk|7t$Nj&;`!eC6yDQFHF2IJPm8(Q6pJpkICWYP^!+3w
z-WORGIxxyM?^T1<>f|%L&pIq6dD=AVW;k_hOkIL}^fCQcPcA3(OJlA}6Lz&HUAvpS
z#kFMwS(gBY!t{GmosD@^*XmQpHubC2pU(5<g4{k4IzY^p;PfmTe&X|^yLBgY`L0WR
zLAyG7Nqc7(27E9{HJ^|O>*cO>&tCe1Xe^_;<tSZEeJZ+&UHz19zT<OoW!S0!p`K)7
z+OD5(s7cDff}#M>jyC>!Sdg3}$ANaPLd?`H-$Bim%kYH?HDE-f90q2|^-MD%Njcm^
z773ODXKx6CO0PkN^K+<hm<^})V8ScT9cVLORqb6aYDg?+w47?>+O^F+97ac%y+^WJ
z7<G{}2}8#)TBmR|vf62~**-S&Sj9;h|7wY!BV{>d1wUHyBA%tg-;Bl8v9bcSF;Ez&
z^pYA1C8oI_)9O1SP44-&8=8R3ypzPp)2!3&+c7{v9J}mn)Z&;Rq8VJCH}jU8zVjY}
z@)o9a)ob=cs#;{rCu4g~18zvT#0^;VAReky*@zfrH$vP_>bNqBfvKj=5)R4f6h7fY
zQztIjAzgm_akr6)S?32RvIM!W*P6xq2+6dhcR|`|Y)5WHT!UPjFUcX$pAc8CvBVtv
zP?~A-F+<Vh^%s7Bx<xiz{40t*&-43xL*Aw1r#KWW4_p59t$JJy_EutzCiHt~?}^6K
zNAA37UZd*rNpTQEZ)^~`ALot!`1aK0pM1XKLYnYdRp|&(^QRKW49zkz$?RAAqG&1T
zxw$&=^Z_aD-~Htm6CZNopCZjh))Cq_6fECzdkef3>!O=&B%2~M?qoZv8_#bjs$|W2
zqVVc9e}GRkc8yKdXead6n2pYN1E$)-ewA3|>_`;n_mQ=>X39I}QjB<29P)38P!hYE
zP)QBaQaGj*i&(Po_lN2P8l_YjgMGw<Xa%ZUwVMq$=TFfYpEQAJ!sY-K9iY6%HB;SA
zZw5&s1v#~4U28Lv?E0awtNEf&*K^Z*>PUoA;APbowkf4ubh{3|?rrd#@*+njcV1Q@
zwIPd?`FDeI<UhsFUm$L1f*K?=xIgaW!nR$2Kq}G;T+7Zlc8YMcL~=5Ia-%yM<@hk=
z3rc>JFDSaEBeh#|t*zzqCj?mLT_(bNcF0z7f#r1w$Hw8eWw@>%zF09qEfpGF(eT{9
ztjtvH`lWS-nw-iD6)oBu_n1R3q=jEsg3V}a*g)zR*qf4CRBXxD<6xsx@J&$<`vT5R
zj3E`fW7G<bmHNT$P<&LZoxvIHCbS(MET;)vNSyLg`v&v$VA~BoBu17z8m3Nv@!?Z0
zyOkC#6P3-d>r7MCG%ja&+#<hb&t-xy>0p4TSkKFiUI<jeC%{%#+)aA}OjztE^bpsG
zB1j6N!N7qBi*gz0a#HdkWmzZ=UMl*Fu#sqOTdIh^maG{AgK$Kx2PniRT*e2ItL2`Y
zZpIbUT9D|{@MP$2LK~<jZ{PjOG)$<N;Wu#gC>#XccNeSo4Zv4fdK7LlNfX9mQ|lyO
z>x4m`f0pCt1?`-SwMwV^9C1IrR<(LivfP!N?Hlq{m)=S+#Nt3OqD^u<y+=e9xs_E&
zeMP+90LZ?3(Gs>p^1PWvThgTvj}l|FLg1g$(*+)Ww0pY8z<h28cYkMMy{}(*`6|^W
zlgIi*n3?q~m>B^Xy)L^2_rvcq!TjLO29(a@PQ^HZFVRC<;Ah?yT2%S!Zs1-lr@R@V
z&wKz%1~m|$Y<D&OWcq%2VMRfHST=wS?K)XpOvgen$MhDkl{vz7M9Q{5-0WiH9G=UP
zCx^)sm0w_f<Hg{t<B-84Bvi_pF*7$ES5NMRITyeF`+_MmdFFbFXZOg3w|Xe*M3MhN
z@419BBy^sQK~cedNwh(8FL&!N2Fn)iFKZ-rkg47B4-o8?u%R}M@#j+@dx^_#`bynt
zW_@*j+7CT0+SCGxLz!ehSm=+TwatxR>5o%2J!+M82ljd2&6lcvokXp+j@JqR0?3#3
z^a_&<TmFbE!N>0sDv3S{yz*}|@$5aQT*d=DMWv?Dcq<RjS?W){c4`x_RTP#xq?KYh
zC2B3tUZKrC(br_kP4f&u1ChO&&C%rKM_H4T>OoVZzBAQsP8dx$3ri^oI^O%k2?F_%
z%cCu^RQtSs@TNYxoN3E&%Ct>k-dTV+6N!kaMHK!6h#*#yrJLl_us@N$^p5?M-6vs!
zxs`n1KzvNuM~*xqGrEz47Cks-9&BSJ*iD-ims91q->OnN(6|dns=ZfM&*;;t%mesO
z<bD6RG!j{FGp}Qt?bDq^Ti~QBW#eHvKp%2+otOh9Y11}3y=?Gdk}7D$iY7m1P6$bM
zG)Fu31H3W*{hcS3r9MuO1}#aSr3)~SJAI4RW-hnmRSfgP&}*7>P;199`ujl7^U;=H
z#Sb<7&2NnioWz}@qN+P`{42}0J=CE5IX+**G?qPAuACZk1jH_%_P>%Ec*Bj%Yq#|q
zYi7bd_l`)-S}$PLQ%g0IO;dDV|9rvF^Z>ZujD)7ROfIFxhvfuzShe$*5pSYtI1bW9
zA;PZ<DqYHzpc^)|;QF2EMSF+DI7#3Ni&T+HpMRV`9AI}M<T0nuDqSaPWORGTMY|-?
zC{@rXrZTIjG}<1Us%nY7X!Q(N<@mYU<ovu+n%Zk|8zIju`%vA#7{cXQ|F~aTzTJpM
zRYjkh{0<I(U2<3F@tZ3vo{2J~WpYsnQ!imGF<DIwqy>Ub2|E(|a;B811+Y~7QGV08
z!|LKuh8`ng_j`#ffM5PW;P0TU#g#i2BCfa@-}p=QX`eWB$%pA%@j2Tf2g-&u=pm`0
z!Qfe}z~$JhH}lRg6LtT68nk3R)0-&d3C2>uUH#8t^j=2yMpnAY9_Nie`q|wd(HqW%
zvZEi>cN6B?X;gVBYv=fh-o$Sa-nbtRMZI25Z#lGaPQc6~Umgc4IP#Q;KI|>UGG8N4
ze1ozt(N_9sB>f!`oB27~&C|AWxfH)~v^gtVK79Rsgf6I0ECuifnyKI2zttyj!&wse
z4c(a+fpM198{??<<)?;iYH9~BZS+6|cMS3qxbFS`&;{`aEzqF<i?R9_0}=@nw$A=l
z&p4`)M^VVi7zf6ns}v3f_69RjaHL}Z`kfn)@U|UuaG{fwX{Tttu_CjtVnl0s;0_(6
z^*SZ|dhKnunp2wSgC17!HxLG)<)H5#Ovqt>S=37AzQ5etv+1|&n`+mI-GKSWc$&A?
zV=@$A?+TmqsTl<g^`8IJ@_ell@WL5NfyHTtxk>bW*wxp}IwOp}4DcDdvW9!qh?Y~Y
zZc13Tcxc<V^_)h~axKQ=CskziqsKNSP4jNalOaK!lWYk$14k?195EdV)rwIzS?<QS
zUkO@86nAR>`!8Z=Cw5VgGJ8--8E>Mn=cLvpa~ny-)xnwO?a4@fauORnQt|BT*CBSM
z?ab{3BTd`b@5W!Li7l}E7Dp7)8PmnryrSG2gNv;9oC|Om9R<RS0gXfTr`N*@o9~@k
zKV-VX)jn%HS*&(+D{ql8R<W=tMX{sFuCL|m8@U(o7_AJXk6GN2=B;mg)I<DjL)&U9
zXL`J#I*R^BY2Q`7`rK8rLd)8+P`u~}=G$O#GxDW4AM)61HQ`{UG(8%-?L;l<B2Swu
zhX=9?HDT0aSK0&Mvz!ryJ}cx<Z%Bs~vZ=@Z{0DTN`G<yn*=WlNWt{j>{fmD9N}J!!
zH3%pewV(BqZ-$y-<McC)94UA~bC_|&Jn4k-5Lj0@c(11Hs@smYU}9&_`55tpN#Jy%
zg1b*Ch16pwO43C}@wneI2Vrc+nwI-uwTc%_0|KT0=2wW?4mZ07ZjaK}mHx#zqqf*|
zolG$DU|>S)MoR;4BT_#$Nl_E0=xe$Z`bcbGG5T)}nbc~h76*Uh#}O0I3$E=l#>8;U
zU`}9z3O*u}Gh?@L$i*;+ta$JXs8ulOzNiW3kP15{L~qDDwv(rv_Up7Fb)N}{fFP8y
z_2%?0p2n<)Hp4c=svjWL85s~=TfzjoyG;86Kl45KH;3!Ov(e`F!#da%D9A!uYQ$&+
z_b7E^jF5<+1@`8%u-79iGawoPozGju7$ea9>ElBOmoL(&h}=B2w9zkMB`>G5k-P=o
zI*M6EZ9M^aa7Vs}r!YcDDYIzrf!*CN55xI%FHr&|BZ4hA?>z5SRKC4yl$ILi?n3sW
zQ1E*J1b=RN;}>tzvP22}y6q`-+Q`gX=N;(bW)6(QL8o)2G}X9|<i_kH!0*G8-6}gy
zOoDa!v2IrNu;`z`hbfjGk*A81KG7%QUW3}+3UIxStY-6O2>IpM<?OmlvU;gPBHpQ<
zTRG<*f1Qr{sFeibnjnA96r?&)u8LDRD812@#;O`#WLlKNewuH5KtEOg=;j&bQB%}n
z?e1I@%<mmkK@kvsKB;-P!1$=7t9|ZS`O|^t2;!$TEYWl`C_^4rMc{tmZ}OiW^etSi
z)_^OGe-_|ELZc2_tt#3BMSyW(ai~+1;^xMZ^daNf*UU+}vwS+*(0e?q>IH&)07?3u
zW7mC*ap0zBRwZ0UL2&6RBxjXRkzz!_f4xgV8MjRKU3q)c$C6%#R4eWW(P_lXmx4cv
z4Yutmb)_P|siWR)r3cR+5NN={%hfalLyXm&vN`0GND&bj65kv%YTOkR_=YmK6n-SM
zBjFAZ60M~)PEB6iK7}~{0q*q+@%07eLH3oguOsF;edX6-$#qcv7^(kjq2Cr>|D9oT
z%(5MSA&?yA{iv`11o|ELLVCPg^`+2M>aSIQogW2ezvbd$C@Lcx&j!0#_($fiMc$6+
zdAak(pGpe;Y%<Mi;%P4;lsr?{LKI*eFJ6Os`Udk=cn*0?!;noQ%1T>gDV3Jyz1=4`
zDaMvDlyPy&)=GdP^Nmr0kp14y)WalbpKC%sBkeZ{iAxt|ch<SNoSnkrRCW?6<5YMR
zYY#@@7i`crTfn>{c@beV{{e0L_+w3)M$OM>xAA}EcY|&NPg2C9u8*Hy8^2lX-z-f2
z59v56fp>x3%i({z=j-j+Y$JHgWy4zdS<XUY%5mwbIHAq&kDWa;TRr5zDQk@Q>1t8c
z^{*Na@hE0KGOuWCIuMhPeToqhF!IkKKI)GPQ@KI%B*qT)%R5D8`Iy2<ZOsunwTUs3
zx)I*He%C$9FWbV|Y~U-2XM`R}IDEIRS;gw$=0qc%a1?~!GH>hqJo#o*VVsrV%x#M2
zjHYcU|MI`-9~?=yH6rYFgm~I9SS;&a_B1<MeNas?&yrb6+Aw5hX2k(k<`Yf}5sMRY
zc#Q=W)T+U#UdiVcqK#0FdTK~{tvnUDv7WPjaTiAvm`8Av7P9dFp{R61Fxk@~#T7mK
z$li#M4ZJ2j&8@_q2)u{K2+mFK{SN7{S88V|l=^vkQfdjT#aufeV}ws9K4eX#2asgy
z=UE7-&0Ga;XSJU95&iK(6D6`GHh*r=XD~Rmf#FK^F;}pR&zWI={@&WBG2}LYeEfR!
zX(B2k_bH)ciFEwRdM#ip=+T{n=Zf-&5N9z1VY=Hpcl(H1h<f!ZSKu7WP4EJY!A^Q4
z6y6$YN`nZKBQiH+14Hkn%C}TQ(|lGhB|U;7r)S@Gez(Z?i(=0)BbS_BUP+(e_(UK&
z@CXTDdeeloqp|8rqNkFJ&R9v!nMMrfnfnW@3D`*xxq>!CQ?Lp>;S*@5W1e|6*2K1<
zSL+pTg`Yt;pJB2{qWBV;ku$(K+trG<TRnsTti6_cbx0H|Wk;)sTXS}ZYPcQ57!YA)
zxD4kSu4sMl3FcN*(;kAK@UP*R9FT-Tj51EQ$YP#t7tP_vPAN^Y^y31xHIcMNcyP~&
zI-ANZQf!@?DK5KzWB+1=s53<Z)sge9w15&D<`<-#QG>y2gly6c+Quh4a>#t-H|(i=
z_cK+xj{zFrx~|Q-HZEvDC)zRC#TVM9kOEE<F2GF+n6@Gr<tv<|Yj{|s2U+b)%hj|<
ze`eOX%vsh@{IcZnA0Zuk_hhhF<`pVPs2Gf=2ai?C5zbkKyf=C;`wvZl^T+RYee|KN
zg0n)eXX+v@u~Kx*`(y32aMx6r<O}x@LDDB-ja9#HU)uJaiYvTz_Pt<cAQH+?J?*7?
z$rG|B&1nn?0Rwu(hTbdy#VOVDocJFNIkl)qiY!@+KOiEL5ys@XQBB@LPV_yhIh7Dr
z797mX#ZO$;fb0TUeI{<KU$<6002W44Ka`c&^5P4pIz)=n{E_y00vwZByYA1-?e6N^
zouW|vyCP?1wh_sPcz10ou(5zPRRlHLZVs<iHCqlJ6~D%)1|~c1aIO%UcJNLkpAeOj
zaae>(mz#nT3${vF1T`}=FIr{O8OQve%P@slGKkV{SQ?y)e~-<RiVR|nAMB|CMphAv
zu?u`k`}Z(vkYhNd8f#?@07_`faY@%mVKeDjS$vLPOJqVq=Ej@eltM$J;#Q<+Hz?;O
zfnwnIBHA>0AQ}gHU#i};%;O*L%ypL`D1eN|tnApHWMjVSfiVX%XEOGZp&&Goey_ts
z`o!v2w4q(Yur026zrGaBG!?k^G&7*^sqXZXu4DflVf+*7CO>sqQGmQ$JSyG?I531X
z(lP1y5_oP^)CB$5TxX1A_GA4-W*8)elX)QRdN$b9m*yi;&csEh0Fq_zqq|_Ij@OZT
zT|1dxHEwW_E+%ViQ%DO}FO%+Q(u!XPr4*)HT*G#cgl)6faP9U`bvGAl{Ml)kc8slu
zNh>NnLp>=SCU>zGVVADc+?8KKCP-r!*kOE0^XX2(tS^Btp%hV*Ky?jBn+|!OZqkpi
z&%Y(uIY1X-_kADPas?)H!jOFU6thBLtEQnjV&TzUW4}?lJ1f`vx`dRG8S7VJCbk<E
zx~lKMRwi7$3JMX$3JR9Y%={!I)Gt)MQnWe*hdwLm5P;*B5^CL2d)n8N0R?ZIe^%Z3
z_Uk*m^^hix5m&eI#=}Xf17x|`Pmi5x?CV_)BqWb)dG)@6^%EvuQ!JmTXIE|r**qAf
zO_b~eb=i;qt`o`DA7%!B6v$$+afxU~%{TmJDrn+%c`!;s-x8N_G~0#pZj{h1(|Jc)
zXT>2{PB*zAHgJ(Vl00b?6$77J&5{WkKmJp-vfSq>)MoG(<G|$3ve)L9sb|r<;(L`}
zeG^m$<MYfn*40)`#(pemEzQ(zU8?g&uqrPk_)^t<qQqJ`z&R~AA%b^V=|t6=7_pZP
zUBUMpNQ$RU*e-LrK%kPme>1Y)&gLBWwCK)9YPdK`mIlApf1!dw*uLGLrKg6*EB0mU
zizatAIhWQb?xE#H+SS!H_rDlj8Pj`uE%#hmZvr<U2dLuJ$As}jUje@49j9Y=Z*QL&
z-{MuAdp+l;fxTgI`r=#74CvW$)aROsJHB|@vbHGa(=%z2!zK?W!>0t}u)}j@X++qP
zO9+wOy^v7Lkh-`bXwAUx9Y-8UbR^5ox|5$B<nj^yg0GLjBy)!)4jztcK)WH61H8np
zgpB0AZh*jHg9qpMRnGtPgNBX&V?smKnCa?X=V7N8)nANFKkplTX~XWPk|dk?R!@yu
zuRz?rVJU)mp;q6m|Gm}afy*?&)3~Er>~eS}v($U()x@JPIt93gHXnraAh{lwJ)>`2
zlXCM>^J7VGY56Djq&aTXiCkoPlyxTeSwotYikWyqe7#Lvy8H5f9?K@C=7%1x`5<%}
z&`}r8qV!0?VA_aV{f$^jTJb;V;<2OQc@e)u7ZHmQ{ah}V?%pE0;dwzBvwr;3@5__}
cj7OR@>ux&l<;=Ks6a~7<^zF+(!v8M*KRaqa0ssI2

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/L90MBb6NLBE-1703.jpg b/assets/images/impact/talks/L90MBb6NLBE-1703.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..b6b02b92abdc5cd64df2afaadf0cb5de686baab6
GIT binary patch
literal 45502
zcmeFZ2UJr{_ct0rKu}O=N>fmJQ|VQtLy#7lgkk{%Lhl^}1*A#~NC=%kLKR3t6$AwY
z1c8JWigctaRr(iw-lzQk-&?<O@49Q<WUX1V&+M7mv-fZBbEcegI&wM%xCB(wR0Etj
z0|1;M{sE_B0A&FA*|TJ4&ytgok&%;=Q&7@SQl2|UNlQ&lMZ-YL$jCrTPk)7(gZ0YQ
z>rC|YY&>k&Z`|bK;$mdw6XfL-;Naxq{MpDEVpYm>l$R+fFLPd{zsmU^|4!clE}uK&
zO!|)W%nbm^<ujz0&zyb$u>BPCEGbdaKYnE7XDLWY&YYtp7MolGoFOGWLrP9baqb)$
zDJd!084^;!S+dLIv>bOR=os`973otR-{=P!7!O==GrGzsk{oJil7I7@sJldEZncuv
zlNL^~FcnYbX^-bsNvbcnxOvpA@`w#H6E#G7=BJwe_>qvFIZH-PEEBza20#KJp(Hs=
z{yPh?*UPk^v-)%#cXIm~6v-aD-4LO#44wX#s3g|H`TdjtxIi2O$z{^Z00qF#eU*DE
zDrf(1CfW>w^&8?nO@-d9K<G1?5mtl!r$0o6EO@)j=u#H;p%tfqB3Sa#K*`~t2BpHE
z)Vu~|>5UsCH`xFEgkn(R<+GPB&qdLp#GPU^^lMCm>kZ6M;}0bwuN#v<W2b`eJ<?J^
zl(M4i$_(69tGZD_J3A%7lvE~5!+T6TFRHMc4Mi?8hzVI2$fDuOE&g;u`x3pqnHPIf
z(gwiiDgzoQptV_3go)efR?RHQpC<bdjnl{lVs35#7B8Fv%*!6#;Guvp*=T5$=aRjz
zdWHZP`bV0E_7Vrdb){$@j628&$y%YSJyAkZvuYTz=x<8spt(rjr)2iDJRryzV?w^F
z)4`IqbCFvhfi>s>m9c2(=uPwnA3@T1Tt=o1i4<nYAnUb(&YRtgnx4PJ@B3@K#3a=u
zW};eGPujJ?#Gr3LaxW|BG2iu+y?1Y;?%OvRcsf^<h<OXCyHub6SvEDdt7P6u@oWOK
zNN!M4o=pO<bl$!!C9-8ijTkaljw|s6(VtUD(Q&<e`OM`To(m<_5CSARL&(Fc6xUR5
zYy9k<R@7nVmq8LRaagVv%0l%KsVnwwpH|tCB&BZeHWIxKu$>OgW~&To<mq+eNf;0g
z$yTJ~JL}D^_2ECn+STWz3&zbSSrFYVV8sTMo%7@`#D9Q{UoiR`JSZcY{bVBSzHZxS
ztI9$JF-ZGyx74vk`k8y21w2P8aU&b)sZS;?og`rg76N8ZtBoy&9h&Zdx9D>iw(QQr
ztxt5jNtWA+dA*bzmbO#VOW#Z~7YP)VG>mG?a;VInjiuY7jFY~aH&5Mj1!++|f!?Ce
zylKv=tt}AaW(0`CkX)Q-ca@4t6YJoWj5(ezrg}??TwN{W*Doofjz|oJv+`yhaq`?C
zxNkh^t63I6H08w?=B-$6Y0JR*>`b=|<s$3J73R{6ztqXhnXVbQdQX@<v7w3PXM}+u
zCWb}3_rO0KlXTC(fk`_hb(Z~LuebI=%-&TalkYLuVsGO}6hWVrmFH}1>?xqJ8sq*j
z<ZbZUEKpW$o@}kg1k}xN*M3ZE-kzf5=D|XbN${|Ffn9YTxbI4UQR`c3oQa9CV<GfK
zn8Xm26l%-zf-=ab%d%Xp3g;i9f-_qtI~gg)4nIha(3Ga(ef13?Kd}G`EbiY-=ow4K
zTijv}BD7f~M0-`<pSoB4CA`!p-<BofsB3V7&8^1$eICT#u9yu?D7~du_keDJ<KAQT
ztm>$SwDUsCWirP5HW&qOi->O%4)~NGi{_3O*b;nSGpxfPyKS&c=IFxc!Uc)CMY}N&
zv+-6enF~whory6s2W_+Ra)OBx|Da1zC4!`B-ql`bO|CM5aOuuFHwC$mFz0W_2cOzn
z3kS{Y-PZE#1pqYIGwe<P4JMXTfMA<>Z^2x3N~W%N(ZHU(@?(6AGg6p9TVwwP=n5)G
zl4LtPv7==+F?VoS#;C<sF!~Ew*=RXHN+mELL$M|rM>m%0hAvLY^){_yvw=?Uys9Bb
z{$M263z1y?&mKkm{Jjz$nd82iC-*&fPO43ooq~`~@nE(_i9sQ*w{2uHIO@1I-*I{!
zNv(CoQXhHw&Mh%X$d#{<a2td0wB+!!r~#1M{3X8ZN%z96_=y63$bgB?;C>pLhT14E
zAW0%%&&Q{$T6v}HGOE0HV5!@GX@%KFhYx}&&iM54tP}k!?j@G))XawEGX*l-lgBP$
z5ABz)FGTnib}(eKr_060dcEstOm!&1zUedC8%rqqhUG6|QWuL$S;cDTs0CODC*TiO
z9=e;_HG>68Jtci6T=CMTI#r!c0$`uB@aMHI`|~^n&~m@K)ek-8??%>|7hf+!xkt=}
zbAp6wbi7kb=d4OzdDf_wSlVw)EZ_>uV8KfR7?%cZa0IRN0mMrR!*`FaO&Jg(MwOi`
z|FDa@wg%!uZ4+*X@Msb8p*Ii`yt}D6X4&grK>dCZwq1ysC+ju5Hn8V^tl*x5#WBYq
zRLMm8_@XEtvj7!**QJ)zU?Ct&!Frme;lh2X=CNyNy(g1B5Hk|Qr>Qel`B3Y%Yi~Ds
z65A{U(IoFZ`sy=d!lyf~)lWhaMwBM>r1d#WM_B=(PFh_NwL^xIZ0bej2Rfe}3%&6A
zX5~DPJd98hRZ3ZICFXhDWV#P>->orAS1h6!hizvod25oFNVwuN@2kfPY-pVOzMO(h
z?KeE2NiA$CmgYd>2QT2IinYq91zZXd2m!wHoMA~)F3@1iK+XBo%(-JmC(bn#ecBNJ
zvQX*Sl6M(h*l4Ve&)jS|Sy0aC&}u(E)5*v#7}17s<Q@;_65@>eYV4a3upl#tk8?N$
zOp*^H)|W1hI;Ku0iaAV<ZSYPp(0r4dm{cAXb^OqXu-DdnIy}?_5{qj_6!R_hFtidF
zY3n20HF$ZBlodwUO;-A4oJ?d`kX3z>7B2c?pV1$c#=3wSU2chCK1|x1CS~ie)|SAE
zI;vp*h$SP9VX?Txo|-+6=2xk%IKnBw1mZByAmC3j{}ek1BXk!Jgr%&)$p@05Y^5&K
zQ7ZO6UKQNy%RxfAdBT!jCga&9H6~RB(MDX8d-S?6ijY9+warl^^PEi_c!i#P(a6)N
zRt`;%!|`N-luH1M0Gd%fsd)WmF`G5#Wg!6%nt`npC>q~=j<2;2pAvp4@JoG;?{`{?
z<O7!Vm^t^<G5<bH4Mz4M2(|*D4VwWQM030gcYe?->DKaTBWaDz^XY!^7wky6jFpzT
zoicrDci6XzbL`<#Db}kz!R+>P+>6NVFI-|ZD?LG#Jwocve!KRgHcBWiex6=aS)R}}
zIxpek`S_iJ!Sb)<@S+?8VCj>tSpkDygSf+q-g)}i8P~am&mpz$k4-wgR_<Fk-Kum!
z7D`?T3uJ0zV2ewl(ecQBt!<pCTTwcmeN#msx%++15TABkG1}3WHaRX~$<hZv<2Dz{
zh1$1$drR9+YcjdU056Mz_$8}@+@OxLRt^nR8_TrqHJ0ue#E#dsq}qX{vA3-Ut?-(r
zcV<E$6wl{H(YKl!pHtT1%ll%D_A9oWSbv~N$pOtmn!}{EjSz_l;`;VnBEQqaqIFDT
z_1BoU;-u?iUs20AeNf_pm|?aU@BSmZJ{g>eco%iIl7rD0|1u#@wKWwRijfG%8dZTn
z0WAt6Pcun_@CVguU*K<I7|NG)1CWLGzygl27lPI(UTOZDO&0mO6Y3k}B^%Ea`fYW8
zNQSSTe9m0-sLX0~NhwQ>F}GU>VjzOMRPzrZ8%Ao0If>ZR@dvD%YCg|IR49XW#Ia4y
zD{QLv6~P^zxz!?IBO^Kid(2>BLLVrqS8}mfZF$0}D<ZU}NZ-;Krc|6+lfhsNvF4Q_
z(C1`}x4P$v4hs9pFU5FAhp3HiTjo_7GhZ1Jt^tf_K_NOj$Z$F|f^Fgf9rL&tTWd{n
z<<o+>thL>;H{m9h{NE6KijPp*rAR2$iw5wTA&qZoNIbCGuExoPMVULdX4qkI#FU!G
z?uzv4by@k5X@T=uId8%HDt*QRCC2{s^~>O7Q{#zS@HE=QBigdBn#&uhtO!EI(3*D6
zj!k`=neWrCG;k>*AJCY9ZB+zh{*X?)5)ibZ@1EJYp06FzlmvwV5k`?oq8f+-N5m&q
zQc@-HG+x$Rh}JxJq<=T0dX%4<w#4}YYTl@c0#18o3nVlDRzc8YwsQ!Ls>shT7?jZm
zKF#GNyDwEXCIHxIVVV%y-hyQ(t|_9*7S$&<b{!Bc-B4<yN<LFys@%f(*;p6Sz*lnF
z5LmgkpL&A{Vchw+gwn6XOhw3v{SgNRt5NK0mh$m@_mamrG3JAL?MKcZmPc-Z6QISp
zg0=T(Jm4Z8a2h7s+~M)OhlF7Xr|XYu!Gb{?Dh~c?Uq`{l9?u5o5+oXa^ww$OBx8!$
zdrW!D1{to1_4l(6VBvM>q`2Hy+!w7kQj81+E48hgH*LJ|=o^V-(e>PYGccd4AgOQ(
zHS&{7B8+L;`;5kPDDxceh0L7U6b?m1?0)eXvvxHR&eEK^OO+66`Oaf5n*7B?!!!$n
z`%N2CvlpO#aVU#9O3Qcu?mS5G<?XvR#K5|=P;d$mD?Zr=k?5}c7xF)vv=Ett42hE#
z;zN82wxAlaeE7_}6UDcNZvZfwBWT#lx6=8Ig{1G#g|CUn&WIYx=IJRs^;6p;y!vo9
z==J>dFa8Rb?l**5_-;&_Tq7ipbv4BegdX1MQfS%^W`F%hX3E4;_J`CiK5NowFAzwv
z+QEBr*#9HU>W|FN7N%y$KmmRx{k+$!S@RD!)MqGP{ho;@^O|jb{>n*nZT_Ki)V<^1
zd;UWNSAhz)Bh~<jt_%MG7wN85N=VSxBgH#^WZsoG`h0U+L&-Omto7Fu#~`h~H~H!V
z-SyuytW+x<Ar!w2VS)wm&ztx<@h@aloctF;{L2<pa)Lj(zuI_v{U!I^-!oDS!|^9H
z@)({PXb|l`*YfuNu9E+i8UJMm|4PaK+a3H16a2rd!2iGdzo8en)F4rwQ+ve)7N0e*
z)sqrohM-O;jb+{4{WwNm$<#Iz@;PFd87HmHt<V8AxJ(P3Q@QtlEy$$~UXkHfC-q|R
z{8SqIqIMbNwnZmw3p)k)$_8;6H#oWAu*>0ej~FgY*%;(M9bLS3B=cGFE)K0mCbri9
z)ftXC=2HFfb-P1{(3F&S<mqo4gL{+@mcTgzlUK+)*{UOz&6-7QtNgFUMU`Dt7;?qf
ztliDT1pW6)GRzbE$?}PPLmQt=S6``p7>ZvkorZK1f@C2Fz?kp-OYy%~`poDl-mt|T
za5J~;OZuHr$^J_5{JIq>aMf=1?=n~wIX%?W50;S^ZOf!~KI+W+_|?H@;lB6%ZAWdp
zb^l(*(ziu8Y()XGneE!^r7@EM!-w&&?ti!aP1IkauqA#Sf%T8?z=cc$bboK|&hM4d
zXL=)Qf~`MW{}M&^=6@iAdj1m!b&r2X59$Ar9)9YV?Xzj?U>f?sdvF_iWi}tj_Pg@3
z1b-=DS4Zi+iUC@vCo#<Zl3QrIJPRH8;;$<H|EP+$_IAAmY-$>T_LweMt2MJLc`R?s
zdRDik@+NU|^&Kv~WEvp)r+!kKe<?M6X70gP3n}MGLa*qHob-;LrV<_fDdL5HsL}UN
zYc3vg%W@OqWVJM=obyIfjX?}mGgLYkNt<gs@@G5;A`aTGmA=jV-3b5Pa#mF`j6@Kz
zB?n!(0%lr~^8Y7@i2oZR)bnOm6p5_WqkkWjzZy*J!)|<fz4}VEZRq)s>yB+TAxQe~
zpi+p4O4+}m@>*O>8Aa*z^4|sU6;HNv_e^WBw7vq5rcNG6$XrUOc5&n~r`K~05qy<D
z<DeX5b<x<e{|{t(u9YnD-|A2=^sW9)ljJ?ltah(gGk#)pBC9&Q<d*&R?>sWUn^Nvc
znKTgQX;?gx(2CbfqoV0bk$UF8Q2rCS9oFhCl|08eJ8-F$e**9h>&f;s2qLohD%zv;
zjmXMNyA-TU=X%r+2~IEc<^Re0e{-q7c}K@zyo1=$KXatNu?GBfq(4|ge!1{pd~R+8
zX)F>G0AZD6&)g~xuSu?+5?qvO@BHL1!Ht$2y1iqwCj5Jp_)9z*f735kmL~Mk$=nrv
zDFo~KTP#osB!Ztwf$9CLoqy&r>OW(G)hXaNUjhFE_qwH%oo}G!CxK^7xCd^f>GF8b
zQ%~&2?G#XDRrS-$SXQq7TX4*OvZC(&UjtP77aed@@~_Yueeg3%vi*vZL_Gc*QBsyC
zh(d|>O8?r6%oZ~Wip>sU^daJ%7%bENPOvPl(&JF&&#~ucYvW`ZRvxQ<Me>{YzXgeZ
z7+1W1(^8eoZ{GCJLh<j8_m@yy95aY%dvRl04Z%AFIMC$5TIw>}3uwP(F6>43SJA_U
zG}v8bp9BcIIR&s`{a&IMGis~}D`=G-E5Vm9;qg^B*x5f^y)J?|1#EADLSRNAfv7V)
zJAb9lT)y;k?fl<P?iT-=d+17U?j1|3pRj}sp}xMCo9~`(ultQ!I=XrCiePEm*1UK;
z9W;Q%xa-@nNg6qreJuu>J%LN{>*&JrPc9ReTp*~qlacN{9<sVqz}cto1|5W*1g!Cw
zn|n7lChk=4QPsmFcqdR^p7=x6)Ra<0<M(%)-P0wvW}?L8@0d*?6`cgPimZ7`*bGK9
zQ@_IpMr}pDhloZ>tCkFKJRDp)FCwr#SXU;vp<xm>Z=?`j_l7{_W8*unKFRL&#8NVE
z`?1l$9Pbb>-wI2<vKLK#0^KX=mMW?!p7llVrFfkhe3yZi<XETyXKJVCe%)26qm10Q
zPAKML_p&Qp-wsifX7Hu1XHyq@IAkT{P65q+lW->~=FEI|Bwj*Z;PC*Y^hrcsRbDx^
zFf}z-2gOPNxkLGQqDZ~D<<h*;0>WDodMvCquxa$)w4y(rH<>i)wJBJA_Z}kvEs}P(
znlawmZP!y#sZ;!`W$2_TwUk}vj;V~~%YaYYXFeZ+L%HTRj-p3rhi-Gzq{TFvn-_|8
zmj&HV?dx<baZ}O(WxuAqKTmMwxuVY3&FAGf@;w{@w}m15hXl-Ekm;ZbOtY^UgElBH
zv4}n{p-Dhi_9)DlVSA|OO5qN)DJ*@4B{-Ai2}1J$%}ox@S9-|}%}q}-j82xU+}a51
z4D=A!$2QQ_wTpsN^OJ9fikaLick)n=I|dMonR-)3LpHt{;v0br<JGT@%(M$r3RKkP
zicSHh0*MEfk=S7fS8>#Nw}r3i;?4+kiW8Y^Sov|*QsHsuguING5^9MJi7YBFG`cRH
zd`O|AGu;s{dgpx_RgE1i*FColgvLmXaqc7aXYQ?t>}Q&V6uH*+%+z60pOa~NhMBPv
zH#Sm7D}9rr%)E|o=3Cjf_QlGY6rEJrOm;r5G!=)EiNC=6(&49h;Q9~7)D$VsK1)uX
z-+*Im3Oj<0deV3;ReK}4wTuP)W~+QS&#t<*Uu23&L~Yo4qvj?YNTGHM%M87&^>8by
zl?n0kLr*tdA+H@OgVkss&4*tcT@^PkxF*ECqLUXA4r%WGP_*ADSJgj%7<g2r`bBAk
z>FJ&eWY|P!EiXN--`(s(AqvRX<*Ihuqk(pQ?N(lLv4bNufz7!P%B#hXM0%tny&heH
z(ue6cM4Go7$-wqZl_tnkhbjri->bSNxNfpa`tk4wnFJw9f-7S%R$k%&icGCY`*%{-
z``jlcEVnu)^)rho6j}~=JeyJwwJZLIM(xlK7!_Egkp?au)E&q|wXr|1tC}x1=!TuJ
zL>uYj2F;3>Eo#yR(NOVNK7QG~*zd3FVY8;{U;)%}76i##DhEwy%;FJQJJ$9`l~`Eq
zfjjsa2ht6qU7?Jh>5n>!O=Ci1?0bdMZowMm0{J^8%=eE{Yj~FUOR|C^68Z<qOr652
znjF2Hu89z2E9t6k@^*HJ!WUFjhD24fCMWafCVLR?oW#YqEAXJ&+?JsUmg;#;5uJG(
z%(A~MceZZ2qO!Ypvd|P?A5NP68b&~r68EJXU6@emigN5HqY8q~rW0nX2h(vr-5DBA
z=PFP@urHpS%i6Ktx707N`V;nj6avB64gb6$eG0e;YH+$yFZ$5#6>i_U`)M>Tt*CS`
zrfSfP3FTiFiA+6+df-H+`GDuF>po%#TD=6FK(7r%w--oyJjGN{_-%d#5<U_pC6mrv
zUH_b9MX_UXP+VC*yuMRzh)#-_FnECi#9orA7P}KzZ$M@Fq$#(V6qKf0RVAHQ(P%K3
z&Nq7JDL41=;^NUw7GwJR_YJo9Lj+4~pS*X&8AHxPqqcEFU#=BB&1j`hzHbQVl<KM+
z<lq<cf-*l@<>iy(Uc5K27-U3KhQSUnDjAQoceRrfcZ2dQ@HJ`UPQ%V#4rVT}6y93*
zG+l;Q!t0{R^;x?+=Tm4b92kJh(W#xTtH|W!#Ch2n33G6-S55Qh03OJDv_}Qjatvyj
zfHRI1n-HID!WsB_SIzM%->)yA@|^PPu#NiirC$3>zha^E-6pWz15boxxc*irlD(qm
zBk0)-V9VRr=aZ8IUNYc@g=*Ejky={t51ZD6iY&A8Mesa7>WbkeYG^~Qu&XL*u{5yB
zRMxQuvxTX%g^3gF!8Q*TQ?=|~X7&;no1R)~eB~q0c)XZiQ|2;3vI|G@JzPydT$C)6
zbHlsa$agjv?!XqlgC^u<G52x!W~GmEGv{;GM|n2<xGrk>)%KwyP!11m(0p1zy;6W-
zCiHrjq_!B;i=KfRKhQXP5$hhwV%(J!7~!hZZJ*H_67_|}=m{TFFry+GH-QNlG$ORZ
z>gyQmc_3Gd3~ZoyUS5*89FcqSNy@9iudU_1kZ62fb}recu^y!`f~$0OTz>OeudXu_
zkC6|{A#cY9-IrmS%Fbyz10dhfxzvqimbV6jw-nrq-MX;*Mff60;hTeZbvZjnD?R2N
z^F5!YI7#WV&Pwmkb(&~%6l6V})m4QQ7ot>iuRNU;=TdB(-F+-S+x7Gi`?{a8DU<5W
zV67vv$WO(X*W+mFnoGikqb<gaV>}8XNtJ|n!<XXy11)uPKQQ6=V95{?Gq^+J`#>Ya
z-fvMUczxLNHn&=L%!c17KtyLad`;u?(uIm$|G}c+Ae&SOYYjJcq*9~kr6$dzX$xPT
zCf5=i!F{c41MNKOSNx$q6~*M$!BT{>+{{NOE+k3OuoMc0=uH&sZC<M#luGM5AF6dX
zxMTct&xCM$4ZhRMIc6<HYK9bIBI6E{y0|8ykkS_KZ@#&I)F`>-Ya!-mySblN^1w*Q
zNmZu52UTPk7FW8OpwS-ieORm)pNC=KP0DlE+z>@ADfgx?eta2KX#L4QD4OL9&ye;l
z!_=|hIanm$b-lw1Y-8xw<;2@_4cdga=MYK`*D3S$Eu^SUD0w)GN&*)Nw{013+;H#m
z@celn6cuG!WD*%7PImMCn<gE~ZcU6YG4dJ!yQA;f@uBB!t!<lqJcnIzsRG8?&4jBM
z=JF@2gEElwFUsDa!{{|wFT*dt+xukKUe-J?h&QVAEpyH*B3q<gBKzd}*&L^f6vn?S
zQNF@lh%?PTN9lSpwqL&}=;&K{nK?*gaw)UoymaQZx^?UB_1J!i`VW1L*X_H&*bI-?
zx1?r!E-x~Wz)8O>B-iUV<~~4`*2q{XYu#^SGdQzwMQ3wdtFyDU@q0=Z6Ro&^hrZMp
zb@`rUfdP^)$Kvi|)^qy^XxGL=U*xxpIIBSyD_f^Q>(D`=ckkK|c`lBFR#wxN)ePlY
z8ZR@;Im=T{k_jPO%%^}2x)HjL1A_xy`%}Ot;(BqT?xQDx#3%b^w!29Ox57@&A8WkK
zx=A1&Z@HjyF&%sgIPgO$^zb(~89vL}t`A;r%&=HD@?kF4N<^0UxzrngbExon`Jr)R
zpRfmcn-BLhaUIaFq0<^!D}R!&if?h90@5es{48HyqmNo|{tNvl*Z-51%yo)N{Opb=
z#TX+7O3j_bFkeyb7_7)VI8{PJU-soDr8>@Cu^?KnhR=ciHq|rKO@&dRSQc%u<K)(Y
z8G?i5n|J#1eEJ_#U${{>Yqa@Dpcl!JNf&)(ix{P8`s2&^#0JMcFRJ%T*TSp2EUqUv
z{qV#3L@hHr-}4lBMGn<+?e#eYv}LO#>#on{=pG|FIxgnvJ5@p9g<MRdWh_}*AC~>c
zgV&h>rvSoupz8tmhv2owY(6vYq}kcXe$lPkfTF|0{QUsAD7ukPC##K_w>I?08@()}
ze23R-I*Whwn;-dEMrCIeE}fjLPMHO?Rm!8YCk&ERLq>lZAqt~VAIYtL(zP_(wXGdj
zpS&grXHoueS^lGfXVZP%<6kxm4_-2VU3Aj>lSfl+RC71QlWV#*>x;>HZfAllvOG(!
zLS-t%>+b!IW9UztXRsLY`0Iz;zMKho{II^6`*}FS&&x3AVc8}m`%s|hl|Vnw)qeRK
zIu{yE=|)SAN2H~6sRXi#m&LLt(vlNT0X81r2mQJF;oHx{WhbRFX975`H}G&T7y1R0
zc@os_oC283n_*_xCyJ7rmSxl8vsIch#Mk<%uV;L`97WvgD6!?SI_aICE#6iitWIwJ
zv~DR=p;EIch5YnEE>tF~S)t(}PbS{y7Mn>P#rkOHVh~$*htb!Jeb$FbqP~%cCpX+U
zi8G$9v|HJ@l#uC~V@Ot9mdE_of%~f?iBmwC!N3pQ<Ll_gFEJsX58tz>Jh^ihu{o-F
z`Q-<ue<!&mxXnH9ou9uj;gRjf%V+m2kMc|G`lsp-WA_hJphAY|du~v~2U-yk8kM-}
zE_Wv@zSvPSsD-3Mwf)|;0<5&oY?@F?h0Plh(1^~e%cyLx%vO3~L|(+`<9@>;-7x{2
z&v{i{nRmh!yBX3{EG<OnSa<AYYI!zGjCpNveZa0%6r7#?>^iSrQw21u)5LpcSjb}&
z54;HoU_w`q4E^k*BprG+b*9Dko~ZZHpQU?)PIQ&Gt_EtN-DC3!lEt%~;_$&?p~D#T
zc5!WkrVeP<UW-rnN2;;f5X~y}VF6fLG}lPh<g%-u_Xj;g1-%23z+sqiZ)?CCojY?W
zYZVrjYlh_mh%Wy^bI<1L2d$Q)uHW?tEK#tGp)0A|$2!^CN~%qW)|I#y7v|vbYtvav
zm==Q7@hRYFQ%AT*VYv+ACsH$#@V?v}I=xNAgq@qymJ75@T8S>RY;;C#zgvIlyzFpy
z&%$>j6J6lW)YuXs&xtcUfl|MBirkk{1HW=OZ*$XdhbrF_%|aRE`SI4JFe*h$5w}nY
z&}K4niQnO&s;W$n!wKyS+hc)yK3B&y6g<~iWeTh)>{lcfclM82q#6`7DsIg#Ba=Jh
zQpXX6-#sfn12e3GFAV7Pv(qCNBiXi`u5)?{A*Ma+YbPBn^PTW~XpS>1*kQE0dq5;5
zoy$CM-VwK4_-HFl-C_8_n(`Xr1CoCIog6*Q(MOnkNN}Fbf);vE!)xB~vWZR_*4$(g
z<Q*~dgb&jj;<GpZF=MqqjR02*$9ZmJIn7<X?i*V~aXh9Jlps!;f&B8m6MSn9MHx7g
zP?c_@m-gD>1^5)+QtSOrkR*LpUBPlg9vmx$xZM=1lu-m61t!tvX;M+R&G`06t|_|W
zq1yOy#!&<92s357bINwCv?=i<o{tTfZ_(NJ5+9za?zn`Dzb>_rYfiCNyYQ&D=Bn>?
z*i&yXRz461wpdee433|7u<6Lux*BwzSNg$&FB7-hxlA@QP&gOo@Zv&KWk)K{`Y}qH
zh<EM1lcLhzr7SZlc1OpXyDNt>yXN5OPfJJ5M#{RK+BJEJ21ecn!uZ@hCf71F`bDk^
zCodg3uhfl|J97UPLQB`U#(Y1=8<y^=gS<#++zeB?n83P7n~^_kQY;G;dTdyMLqoap
z_3uYA_L*FQUQQHiD0R^tv0DWew^7rD4uMjpHT9&g`Ot;=0Ikm3fBqh@Ya!Jj)-k=i
z{xS7LXU*To5Db4hRg_&+C#kb?5taX$<YHDzK@vc|D^Yth#ICeF_es<AgPzudHKirm
z_nJatw3stBnTOveS`7wS^S>-ZP|HJhs#+Jhz8%<KN@IL;n<|A|muc_{A9GSz>_D~Q
z&_{!4C&*)dx1?&QBX4Yxv>c}ti>`%m$d|Oeyg;Xn!dJI8J|7P15%xu^2CE?mWZkf&
zXT3q=ET6KbAVM2Tn%qG_*I>&{2rh?(7MEaJU<cih@~g_422O~xYc4astdz_f{F2Pt
z@(ySMWs!xr$Vkv4Gcd&v!S>aJeqH2prb9rHdwTm9M}An8@$TlxcvtN!zAx>LkJm)$
z_H2$1Zc1gy+3YSJ9~2XUwbp#72;gXe@F!)uV+?M`d3upG0=8D1tFc*It!@FyW%4S*
zP^bwE(Wig={f|QG+NniKGR&h{_V^p1ZJEtu7csduWkoT=u6!pddww8HzZ2MUICii|
zHDT{vR#aZ3S6~mDlNjr%)JervJx`gUVOjWwg`Wbh+*(|!gKK9C8!a9M`IPv_oiB8h
z=m8O*w3m%kU_K9!Ksqdel}Sf(R(d=znV~DiJqD?2;iMlFUS9XYpb7d0%As+uH4rx`
zGcVrsu-bsO?mN%j?7X<bAKu-OOkT$tu~%C%g&hlCw!~f)NxDfc*^-iY_0Ds0kjkU>
z3&e>2oMo!(C(m=f8@H{<FZ{uG`_bPn1DAA{<cX^scPdW4%ZJK^6m0G0^`8P>4-<Ed
zvfgDYe6Q%xEoh3IY7YAXy1^5@`||fwvA=fJe@{XqoBfIuPLeR9Yr+~Q9|UlBHtF7U
zsrdr9Tn>9>s~vD<{ySY#`$_U5k)Us(Te@m>?PONVOFukqqx=qnV-6h@>APk3()pa#
za7&-hG&$zyDetBx$qgc;sbU~h*DXg{lWwbQy19w=EOYTC$qDgqFeNRSP||6K9cXNJ
za#jRxe!rQKO;<e6`V=2>=?JInz?4C|xaKyOP%4lT!|K~fPEXEhAixrDGwZxDE2(&4
zpa<#I)AB&h+Cx_>rQtPPRQ!PT-K?xvFUyZ82*JX1OD}()dym;oQT7@8qq)#1o7;<g
zbLH+mWk8rQs!}^wAvm|-#JmS+bi)_Vr;g%H9i|EzE27qVK&J#vf2*q>Pu<v@c~2Wy
zYIEH^u_|OJ&(@fYNn&3Jt_Ra{&?mOOb)h@o!|UcJ`xk*dul7<m%$M<!cqxhl9fe|_
z$AgsV<8y7YEwE>WLkR@{kvqxhj>T7!<W8Kfe|p2iZ;M}v=V~kSb9(5zSa14_X8-H7
zZBlwvugR0R?C(RkqK}j&J38;h_^t(+MW3j>I_wt7p_!I~_7)G>mX^GpY+e+4`EjJ!
z^=Ron)iPs8x*~D8_2ZJl;LYg0-srg>cjTXjbd}t%y04;};YnO;<z@5eOnp7LmohHA
z?IN$23K{`j2a|u~VF}&rI#4xlUI<yDd-7EJJ@*a&%8HK?FWUV@O4x!K7Yn^MFXo)Y
z7Jsji7l98mNE+`jB1`gB$1_|)YoqJ^TT_vVRa9ND&spa>+b^V+3_FAQomUFK83*3j
z?BRzjr_9FXQWa{!r|(I#J<uUPO0V~xj1Tv@70FJx^bIq39ai@Fe&F-33i<qvPM`0G
zmDEpqG3Y?uH<ZoyOP|yY>VqdO?tnp{OVv=)_rY5yd<H|1t_E?MJbRc*!GksQ&V}3^
zs31y{6VYZHd?$86Z0&75!#HN2qAx_v@W)uz>t8wikJ+sB;K>}kIJi2`-pIi{Q7oA8
zV82EJC6Ro18)JJGWh!-jnkmhncF(VQV901ZS;;<G51bziJr9Ij%IvbF*K8lUxAMV$
zEI>@4?ws_wK9vNa!L#vFr5-q0Vh@q00NgM6UwSXMXZ5Aoj(LKNCtSqopCF{i$aLwU
zb1qwY^nuA1#l>sIVpd{|g!-8J)if@HA&Cj#Yya8%B4!*tA83=3Co*5-u)AMP++F5a
zl&&$>=br-B$G3Z%l*G8*`k|)F-#X+)Xv#4e^WLwEFa{4DzG`#XSzM1qMZS+KV-GJy
zS{KZ@a2v1L%XsoPVT-FOU*P-2fdC}0g#a&$ibr~<v1Zqg_RJa>!}trL*y$~pLDEw|
z>SEzV4rWSDQYX~eJ?0LI!;e&=59&QZ11966xU`ZLv288)T$@=P7OvJbdhak}^!gXL
z>#T$^ZcES{$v8VGWV$1C-c$JQv|Bsvv?J<_^<3PX%eqvYtP4?DlE2V{BBJNc=EdB*
zv8eAI@}eBqcotvcuv<zEulMGzD+Mtb%*Oj${gHdB^nb9>HiVr(_e#x(xML*r`Tj>y
z7KayxU~&2Kg#qF1Pvujs5Mi}9#p&lvGCpmwf75^0D+T`vBDl=D`Jb@-m4gJjeZF5e
z58j=iE{yMU7QmLTfvy@EScl&pP#YLJHz{@I6kv!QU|!Fgbr{gT=CtQfd@ZNc&6IH7
z=Iq)<{x>uywZtJuTKu%;^?uA<sl~#(N@7gYPQO&5#heYA^kRz=^P;E6uMe5vYP%9J
z6trEH$<I4%BhxyC(3Uxk1XC<VZ-Oorc8GAgaZW&-lEYb$H|y$_Mn?~`_v8fCy2hW9
z{+U~jvHZdqqdHKWG|}&as2>z{f^L1f6pkXZUe}YQV7Fp|-6bq9Y-6epjWiu>Vj{Y`
z{je5(^nxkYS4ebpbR^?Cd-of)ej3y5r!nJx$y<>SM7eFF8!t^<m^?pI97jZ;6q}R4
zdAoV4kG%$S%NYwqXOZ$6?`UQ}p!?dD&FX=>{zp!PUhNkM{}&5ue#4F8G0RWf5DiT1
zflxv0L4aj~tGzz<j#I5?K(U*wOGGO^^%U@dmS!z~$VSow1F>1kXm#G3f=cQb;!^R9
zsW~r*i#$GI`LV0t+Gm@@)*AcD=hHskeIX|VRX+tx)GfVA2%`9Hm>CZB(vvz)40~PO
zNqQ6K1t<$qX3mC<h3|lYI5dXtofo_-QWZ2|heniV?as!VC$uUryUv?g0OjW{kOU?c
zIjs`g{?iBk%!zWS5LM+VQ1>cWahs<f`U@C;<g9sr^P~SGg+80qxw_|V^!JO$w?93<
zbJAQhl(+8UqGv9!ica0?tx<O~5`lq@4H2th5=S3@IVe0>NsK6zr34YJe+XqjJd9$A
zGcoM1R`$1ON}XJzTh^5-kTx_2Q&KeMQYYz}-qn2FH*@Wt7lw?(%ATL~oZJQbF~3}k
zPa%H<f&Zp}pM?ScUn=anHvBOwy2c#kz_&I?=@$u_g@`+|^0LN_rh}D4j|@X@Cwp+?
z<jojfepOLvPyD-^oup@jw_KczmSo(0o0h+y0;ae)30f7o=(xElN>K3kLnG$*njp@&
zJXXEWQ2T8SDsuBD+c+0agr>2Z{CS-RVy{w8%CjTg;aWPxnW+(8yF4=`x22J?bB?|w
zQ+JXzFwa~tw8^Zt@xo`?@wE(`Q|<$^D7|sX?`iRPjwA^@xv$31ZGMC2ZQK?NJaXi=
z!h#+~QG>yQ$~G*IA3YzJ>V|=$!=Vf*kEq*Z=d>J{((uv8x%>Gv%{<~bA+D6T6>!tz
zC~sUW@wro@WYI9z?pqq~T~qJ)uLdG1&ZRu-B7OVqHm%MVFV^Q|&uh>MQTQ;u5f-z7
zr!ANI+4g#=Z}uATcQNq<8=P#)!sR`Jw3^=(Dl91s^MG|F7MU7AZ){2G3L2!+I7=eN
z#|%*jAY3aMG}s)wgb=arPJ#-Jn+pZ}=n)pwX_R{Ev`uvih`G1&@uaiJfr74BHPpEO
zZY`XZL%7s7>}BSN1;b$-Z~D7pHvu+}D7{S2lv^}57F&5JWT{s#SFmZnB}A|*>;_NQ
zE#hLM>Ui^l`eqcL*jo+U%Gd+xcFC*s;-Bx0+`2*)eLygQdlx;ro2ZAze$!}GN}!>U
zs}ePI_wMO$hX^4N%3(F{AxtQP(E4!ly2`f{_EG@XiF)h!cq^1eW&bBH0W8^h1lq1z
zTy3X4XJ~kU@5JUECgsgo0WHBhii;g^Ms{Ik{rMXRFpXc>FkWi5jND?x&rS<T!OqM1
ztkl-q7dV{%dfCL$(}NAT7OGODIMQ!nWFAqRu=9xH8?1<gbxT6BNb0bmF=<2BKE^q%
zNNo25o&(NFJ`v=6HuN2j9sDB9D2TNK)L?7D+HuhENY7h{IU}H@JM49_k+_qp1#KUF
z(y*MSaUbyos3|57_r6`NG7rp#sNRIZ7>hFwFN9!AxOZ6K!wUTTOOTCUDHYSCuIEZ@
zUlmOKwKf%F1hu?!9E!tMpexDUT}{V2&O57Sywss`5W9jGmS6_cFJn^@$Bp1wqXyeQ
z2SQZOvBeu+PxpXF+Hbr4Ul)64+E7t?m$jv)ysFpJ>^iqXr`n+`*`JcyAd_BDsNGV$
z3~eT^BJrI+-`g*X*J!_AaUlI_$Gb!Q&Pn;mWm~@}zWz@kC0LoC7vIJFq==gL`IVkE
z)U^E58vmUNMfEcRyNfgpZy)wXhO0}|ho}0rv;<CiyWh1p{UvLrr^of>wNPDIyFCBp
zp22qS5+<Ab5>=X%N-y~VuAT3`+;l*TM^8O8K$mm57*Vwe6{Qf8c6)mVH10YZ)Xv71
z9!SR{eoBfjAE7q@)3oSnt7_|NtendU@}i%w8G>>^q)js62>~r>+V*Bz)cWWNi7wt^
zlLzm{cx(pOoCLb}1{=EhoBkXF8J%*D=W$bY>qx#H*ss@#Y#xR6ZaL4ZD5WVaC$q3z
z#$s23(I*T=r!Mgl7Qo8l=N-3v+q+)gb?q=O_5V|vOKAM3ejG+yhXA%oAs<un<zjzy
zW}kah;FsQHelos`j{3K$RLKl+u7hw-@y73uZTYEX3cDI#6g-bR;!^+Bp4YFG?A#s?
zv)=(T^A)C2L9=mTgNqk@>=BpaI;6+RDNH;wv16~%{dq~R8^JD8!u3TAjf?&%nj!&z
z@qsI}CV!zA{(Xw$7FY%sg1j}(g7mOIgTcbWb8A3SWbCZG!1upeT=@;!kl9Dr31q*z
zi3VXrY0QmJW3)vnRF`Jz-UO$0(NDmQe)(n0%lce@QReSdkGP?>NGn{!*kC^N@n4Ts
zNLT;5Nw{Cmp2^>Tk@54b++UOOgdT{$Wl;M=vb=EmH8+;n`lUl1C4XQ7-MEE@e0u5_
zDeYTGaqSO}0h8~7{r*^~su$*b^p$NK_{D>T-Q}eTDW~1EqY`o@2fGzYH7?r%><5)-
z>pxy!zIb<YVcBxt0a&@Zgst2vQ$H^3DojF9P^tCFz6yLZ@7EiUrllO{2jhf?>@$Uc
z7nt<fa3w4$pu3h9Ar}i%*-kvBi=)>;<&d0th3SVTpKoPZ>YbA&H$X`PqtfW1F>1`<
zs`%ACKOTl}$J48J+@pFI^Y5ZEjMo|JA7GajlaC4=4KjZ8vf3TO(_Yp2cBhK~6D|Gr
z3vLR?np-278+TmRX+QD1yUp}!Vs)m4X>$eBAHuk|%e~ZCV2*ErALeW3Q=?4b`=14~
zDXAqkOMQ9<X-KuSPXT%-qfabEJ-U0?q?bX>gqs&R(bYaR^<J2IyTx;Kc}p1#x)N<t
zXJB1h;(XTW?<y!_v2T)LDWc+1sDASS%v^8ud#=v&qkh%+oDi&2sjRrm#hr9JlUSzu
zx{&YA*BY;{X8O8yUflJ<NTCcO#-|Y7OM+=4s1_9+E=$vQ#hxHTOD6vtJl$o#-p00q
z9J^US@CWVTp2N{YsMn7Tn<#aAZ@f;`<jh)yv+bv9&O5D$r+#`Li+O17xH?{(rmx>~
zg(QMR|3xMO&O7V620iqD5>>y_12Nu{_Eoy9M^ytzp{LF&Ygajf>))Busxe<r9bV#1
zbXi2fb#R{w<~ag1*%LHd0Bj3q&j%tVP{Oe2`litMXEl12;K}aSL40O5nk-V2hOXy$
zdNzB7Hb3mCn_v6-w3vS>kVzY(X@niiU6dg4?owk+@`<(T<_DkXI@J*0T!O)1^EIyd
z+H5uoQquJ=3NAqfyuBlIzNzgGc+%;Z4JtnLZy0A<yy|w1c81ag_5kGeBA)X?a^t(m
z(s{;XptDfX{i-c2*xJy7MBo|q8c9w>lx|i|DB2@*;>Jn;TmN|yR8|^3wl9MBTEWxF
zj|LCCWYsfeDFEvmEY<ua8$8B`^X*mW;qk8riUoJfdjwjj2XmzXf{zkw6BD1u4Hyp>
z@;dH;$Ku$*S03Olpc?mIYH9M_(^wyWq?n&_W1dT3sb&M&B-~To>pBWb?sYu{=w1}W
z6b#pNg(AI6u?uZ!kA$`?{lW#J2TUq!bESCPCf$SwJA+Cm3@kIk2dQf;ot*}_zJI!r
zvuLj<)q9j2yUyd;_l0wtdg-8nLVI#p(eTYoLQ!H$34D>VxA!=|L;7Q=_he43<-SDe
zTSI-}LBpN5dkK9W_S$jb^a4^>IUntd^4bbBhzTzVfSd_QS@C&o)6Uz@gUY>fK)HDa
zE=ULbsTT50^R97yZ}dTQM|PoKih^;D5wsvQDS@2a9}%B6P!77$?N*l~>RB_Uiq1{L
z68N^<-S2T`CIqI3Y8y;Ad#`!)EVGntMyD@FjG;4Ad>PDaDi0-TdX?Kud)+3Y9C9DN
zT@GN=dV33bpjMcbi*9%@LeHlUPM{A<2?MHAAE}f=wi(1**C)K32x~?<gBqpIp~EeY
zY2v0rW2e}rYyrLg4l0s})E5Jm&YwY&HI-_IC<9FbCmfLAL+jQ=CRu?h_xJ9(*YLEc
zzHn~jN%R-Bq;RU?&@-y<K!%M)T<Ws5oovb`0lC9q(;KV$F6BW(thr7(w0WuYRS4dV
zyRKJB0e&_+Tlenx<AolEFKnnj=qAr>CiglcRFHYdp((}F7%$o|^g?@mNzvW=(AQ?n
zkLs~t?X#2;YXopIIVndc*q8A@oWE!g+yuR4C={_g0=~4;<;q@`fN)%+3r&9K41H0!
z-3FUi4#J!8?=BHkTM>xcshI+@=NxbHf4s`W)V>e;?b{FcZOVj)1sJQ)WifA;3YW_-
zh)wyW#6Yg<wI~|*dUsBFz|)g$Zh7<>4re4m9M*>+(JJ^GNHhG8Z^3Sy^y}^3ih$Kc
zrcm3itQ;hI)ZPhyMXk(-<t@;jZ+RR9x_Xu0EMw0Zs5^fHyyUFk(>qwOKa_B+J=$Tn
z%nEF0_9AYatj{K8uR+B-fM&vb^=6m=OWOIHG$J_RXNAdFMG#iwaMF%xIbnzOHGv*c
z9$#2Lz8J|8BTxV~m|91%oFBQ$&ZYda<LaeT055acJW9}WsN04;uehWf5p+u-T;R2F
zZ=bAejo-+|v=CaNrU8@}mbBzzhGy84A*eZ@FCKF(6PC$MvEsO@sqrOq{0=|A&iys)
zuF%VZS_>xcF~c1L-K`cK!Fu-@V-0DlwX_rh?WmTn)8q3<-Wn5G%OPXQ+NvtWiN@f5
zvnIPshH^#5hWulu`cE!^g;%_anL|PJ(;yIn2d*a9mlJMvcGlYlIS*RTRB&iCvR{TU
zI(O?nmo_kq$WT=4Gbo{yWl?tz4w};}3E^hU2vU^^cj`O<i!39*PWtk}RHe>wkurUx
z(qrObnh-AD<`P$f=iyc-gbh<XQ%<Oq1#~POY;A3bv2uN5f+VldOBW}$90o$<@V+Rq
zBT;wTxMo^qHio>4=`+v+6(FLYCUx8MaB|%w4f(|N;rGvg#NkOc>o3L7F0quDMQM5?
zxHO!YwxCt#2=dA|46aaMx(}YL&pY55fzjTS_6NFSsGb+RWmXRE-toO>&c+@202kDj
z&-3CLYn-~uO7fK7+c<-eTsEAS43cPOE>Q8Wl|n}PnS<kEG8`Wgk&6=7<-3`#VaGQd
z?_=#s`W8r=JJ-9WAJ}Is8*ZUrj0#Sq@*7a(x0GCD+k(=BR^vurV@$*c)^HN0kH?<d
zd9^1#BLR8)002qMR@!~`0&H9`LKqBUc$Tw-1$YVoy=3X!xXtzFM?|=;bg^kKe=&Gm
zoYPRAl;{Hl+oHzoxpc$po};T<q}E9{y9NyEE7T-Qt^gl4n1AgD&RkRWi~$<-a+qnF
zcIkY3WB@P3S@8A9lpZ8)X?fP;G3x3X-Ek3m9*d=6VXi4;yc_A%1&jnG`l_mgGx{?X
zf>AP0(#Lw5HuR2(55rBFj-tK1j$w^Xg3s>mlQJzQ|2!uwMBG)MXFbXu7Up^za+IH)
zd@!Z*;V&P{pm96}FkJt{IsU@-2lC`G_tb9#Vt==p&1YVW2U)Is<;~_b@S@>RrMTMq
z@^#&stsX{dCKEj5i-adxUu&q%#@2^Q^E{mgNuB~Wiw!23ff}C!50;+{30x%i)<pLv
zp&dZOcpli(TzjuJD+h;QZliv4xcNG|Wv_zaf>dq_iiWMDyXmp~mHQ2u-y^~XQl6s`
zyz3<Xwz`wNLcI1=D7S#hDjY;#VKmD4uGsvpmp=<*%X8^oPN9blo5`L2IQ;?PFHYyA
zy*Qmvq>US2Mn%2DhYV#Fb?^A(>hV2<&B*rUeqf42VZs?c6bPr*MOc1Rh6=sNwTd)<
zF-N@3=lKDf?6Kb(Z2VQtC>6&E9hYQnY3@?6(D()QH%!=kv$FjV0}Afko!IuL<w9@`
zCB30>>PLsPmbF`aW*&s~8b{v0ud8%}6HGjMpmOz^?L5<D$leREZ-iaKT^G?)fFF-p
zHihRN(7|+^M8TJ341uUmH3gnizr4V%xnqnQ>TATZo^SPuKN~n(@8xu-w^_sg;rZBa
zg~1*N`3cRpyEE*8<9ZLsxMW*WDLHd|o?4F82-!X(rGrgeYd23T<8Z8(J1%g{gny@V
zPaAq#?i$WhV%<yS(JgoM;1GT_aSyl?zLA6%*oxXOV4>=q<w6cFUf$Aaq}ufRF#xl7
zpBDJ=#>V$UJ+R2eZt86s-@{b*vCVE*cOJoR?&z5wghxQkDd3syM@$I1il<vrDCC>a
z#?;bDt_Yt9+;nIZZrCh{H=2jj*HJU{3x#`t;a{W}*5Q=N(M7Z)=9@hZ=2J#dWm;**
z%yWoqb@doiFP<SMo<lAebUtHwFa-zyZWr>@R#mi<X(i*O;S6Hd+m^aQa7H)P)Rdcb
zw}`gxT}cR2i+Z!ja#xA6*oSl3Qi+!@(e{>O9A)366vWyFHYWPq+JAMRLUW+~MgQ`E
z^TEs4%MB?IGu1==F=r?q-9&=Z$^#~pA)sQ0?&{ZN*+Akwex81ZbUF_Ocddtsix{&Q
z!?wpAz+E6nB0^@$COFv#ui}}}l*h&GmHD<9STuAEY%ef|tu4s$>qQojkO!|D@?69|
z`@&zE*Xd<sA<K`XJf}#LBVbLRS@HcEQ@a~U#gb%@in?aq@(lw<Hl1Ds+Q^*Gz8IxU
zBh$-Idq&{iq#ET}TkpUKeZC<{n3h_~;tn;#jrCOpeSHhpqVquDJw7ER;60Wpo5^I4
z-(=wyLk^~SLj1Kr8xr;WK71o?(UD7rX7zjz10~*;vd2$1ZonPEt&2Xyo3gS$7GVe-
z6k7wgaV%ykhxqI^JUF{EL$19fDX*gQQ%!Vill)%cG3td=J2UtmBL)E|LkzM|fTpey
zfu4Fsadt6Eq}yc0p(nM?x#0$O)CGw|*w2me^WkSP7X?_f!>t|9z-_VQdcO^`Vflvs
zP}gDT&<_6AVD0>kvQpz4nTk@6A>gE(Byx?dfGvKzD?!{F$d^C)wKPI~ep)G>sAl!u
z?!Qnf=$C(=O$D98ojaJOB~e~uHd;scqHSa57-M@o=bMdkK=^>`xfZGRJES*%n(Su5
z#Li_tA&;Q)D$~9UC1&?E@ByD)hNlfzn+s=rPQv|RtD)^00q<<Dg`wz}76T4SmJB~i
zE_)MWDvi(w+rL^E-?_|tMOsCec-low!k<tma2-|ThA*(8vN7Xf`C;p~spWLNi=-s*
zTJ(yp#b-m~K@bL9sPw?i4*c=n%|;G6x43q4^*=Qu7MO1)!2jT0L9Mv9v4u95RUDuP
zlt9TrAw88G8w%I*4>tG;`DzClL0jO31xE6eU_b^CiFb4|^vpAgkZY^&Du~B-e*<=^
zkaRCPl8P~|Al$L^vF2wtHV^pura$0XaIfKyZow-5pg_rTMc^y4buD#00c94Y1QzzN
z*N`(rxU{PGu=DQ-=Lf%cFDZEA>11@%?B2N10>fEkX#vm1nb7AA(wq0n{}3fb$h)lE
zN*!FhWM;oIF_c-Y(|^rY%4`huafen!caGLXbz<4&K&RiU^c(|&C(^Iy#cGk=_jl`;
z_$~i4dH*^PPN9RkZkzaN^N*VDu=I{Zw1`MblISRJnQRwB@1nZ(Q%vzw=lF4!f(m=S
zEpNsFxKe+!s~+I_)=MUZpTJ-vVlXOx9fr+|UPC&YFCU1ssil(Me7+O2RXKl?SNB*A
zXeYbVy5oR+x;-XjDYQd{D051DL3V+U>n5_`B}1ep5$ZBTYtRm`Tl@qI4@F7p*8iaF
zy~CQ=zWz~)fS{mKRHO(3(m_EwNRyHvErpU$6ln>B-a!-;QKXZAl%SLl5~>(NFQOs@
zL3#<jOZR|)py&;r_x+yp-sgAUd+$H<WM*gftXXUA*?X_@SuWLLLV^VE|B;o$n63T6
zXopornD>{I_6tDm-$sh-XK%W%4W*P<Up~ND?-P^nx<V&j+YoR3rjt}T3?2u8Hax1?
z)7}DLlqi9!ltPsJpfIjf86&W&H8A}AlM82w%%Q?%QK;wmW*70kupOu3q~ale)ilhw
zn6?Yp&35SA$c^mxtfE8V&fM6vw(BS*{ib__LrM3TX)LB^lD5Of<#?P@I6Kf(%laTJ
z47nzt|L{qFjdHT$6iWbT#l=QRK)^tHa2H}$wdOJcCjH=$)LLHVi;t;6cHL)^e>6qI
zzTgnW@kzIzu}M8-Z(uiVD_etx58-S;2P43}h=#?{Fql0QHLm2w=BkC6R8f5bKV=Jx
zHiU$vjfkztLvOU6sE4rl)jZ?o5EpWJfi63M39|cF7+9MUY#bQBfX>)Vi5BqO9*if=
zS?}=b#&~6)-R{-ZqZn^7`^<UrynNT$XXg=KGK$8@<q_|(+H&BJ?3|nA@(Lp!S0Heq
zA!ahBX`H!Z<?%2INSs?WmWvd_#5i?&Q7v8bd1!E3sL#+QV=ih)L|d8V2Y{#2Du_B}
zf)|p05)oSp3`;F&J1`xtw(U?V7uP;A2onQRqI@-$gN8KcS1uQfeK*xUn4`_f)snm5
z^%}^Si;<7`X33HQXySqEn-TRApuM{=0+<?n*3hL^A8jniq3a#1T)y?<vpOn!Ad6yP
zvXl1c*X8{z;C?qNahc<rBpWa|hJ!20Up#)X@qUk{@eiCNu(8-Xbuclq&*8T5C4+<3
zC6j*ny||xg4Ge0U*a9M|;9W!`U{d6T)vUZl(?D;bW}`uynOis*1PNOX2Zc$zuOsOB
zgIZ>u{H8ksAAxltOJ?pgT!9CqPgOv%($w%}&I+0N>W26k^K|R{=BfqSybZ)NNq4DA
z`kwOl2i5py4f@lj?*li^rU~a+VPN)4$Dc|Hjd7R9Wd@seJf3$qrLETH9f1x_WQPTq
z-uk(ZZ}-G~wj0ewC}TohtMWCLIhzJRIq4rTRNlEO;t$)d2#MNv{kr!nMNA9_<$USy
zDZ@TG%MkBjBG!QYRcDU7e{I~IRrWU>i%V7fa*Nuh0hc}$)dMwjs@Om$zxwJZXHhIR
zW&?F_Y(|D15yM6DM>e>|7zHNOw98A%OP1u_oyM(wkxPr$i`Z-fnMCp48mU1>LH^3(
zhSRXHU-PhN?5Rec+z60hXimHk6Y3r-FWtQsMWF!LQT~5*#()_ht2dm-p3lg|g(&nP
zl34DL7=nDG*%AS2q6f}bQ~P&0dJ%CY2^u8tF+y&UUwR%XwW^Z2_pJzD)m4Kp)jw|S
z85vYi#jBXk5N{-OfqD8gf{MI)=Eh-!pQkTmj|$ClxCI%{$0x2#J~x?MSGiPuJ+Q#O
zo>)iyP+70oljyThs8)Sci)s1OiMniLpH{L2_HN-OEVX=MQEMnJ45)MgYJ8-kKEC-N
z21u{3p{X;A-FcGNzrUgHWnZ^8$v|6M9K_}I%>p}rP&_fAg^7)G)>`LV+-V8p$K>hc
zm8MgP_6Vz)d?GSBcgS3CNPqMlMQ<=-f*<?dTgV&Z-1P7FK$}cGcO;oR|KXC5?XCYG
zjfGo>ddtOYaIdP3+p?&k8msgaW5$(Hp~ozW??e(66Z-4^{UG7ufIzvT#$GGCbcez`
zRCk>H3e(Vxe~iwfI-!S}RRa9k4X7oIOWl>^z6oo`TjQk}aA(y74B2qlSTvO*m^bC{
zm|H1*9(&K$qmpE)%+15Ha<4=6F=!r)l5`-|6SvUpM_P}*78YM$*xQzs-e=|!^r3!k
zH&o-|GGokV#l&_0oQTwa)5NU)YGJiJ2mWQ%`%3;85?qq+3|_D0E+j}UUwEl1Dl<>#
z*6dZ$i{Vi>9LHZO%?AgdSMYEQBxh-utLl8b1h<4Q-{ONm6ea#);>qm5%SPiX#a%tx
zpKd&<8{vR7B{TPJXZp*B{fl@P&iy~CE$~Nb=Pm!YMv#g{uhpb14>v^Vc7N@m3DS4i
z<e{)m%pGU%xAWrvW{K|g&P7>h{xl#wzzPZ{bjvE4-kem^cM-btxFmE!aYXD}WR5pr
zg~F%kE%}>{*;CA{_@(B4GOMbf*$VjIV}v5M^)Ky|$#<ZQ#{UeU|F!YbU*0QlQ&daN
z(oe~lH#PC=8=~>^+Z2b2G;++i`4nN*N~-h|u7(fjBOQHKkAAu~+UyMaS@OSmlPG!P
zRE?IRcg>XDD7d<4G2o8(*H~7mrM;*L`mrf3+v`q!=q#@TxTB}7Q))F@jB3BiG?*5v
z{7gKI7PyRg@c&M^Ph`|>vX#|Ifzc?B|FgGYUgRHd!z0En?M8tbqiMpbwPR>OK|M&d
z*#%$$>)AnbCw%`!SoYvaFJpNJzSZr)#^&h)Pc3icI1r(*we6O_;+>L%YEKbwjRi89
zy(ECf=uxWI=7BR6N6US4v+v)-)ZeRJ3*BzHRn0J6UAo*o-DYK_<vK30w92ic@n%TU
zXxeE#i5)L~*N6=;VG)M=h3SXZSA88KKcvHjcKJF6lJ?V75qMl0^|{YHP+cK`_?LVA
z>+TH1pntR-W3Xq3Iz#VJbzBV`E_cC-4(!gz$bTB|Iq?W_$Sn*FgbsNnHR|V3%@??Y
zVH=S2up5@WRlF|z%m8r;V9qTsHuidtUAFam5w0_O!jY%9kUo6)P)`3dvXue@O%*aj
z)oP52qE&vUwmvQJGa!4!L(g5C$ncyH#H%4b2pyb8--FinpSel3W999b2sUOdnwxPN
zx7sRx7q<>T&)vB@{g+-~W@3_(yZTgWEz%_JO;+5ZbTv?WOF9gk!L0H6`m&gaGIHe1
z1*y)1{o65yuNaaJUu)VErz4tMW(WJm*3uWAudz70HTwkG(z$vBhDs{NWne_jF0QV>
z2-^Nlhoo63o*<>C)$iZK{-!HJeufZ_>{OEKL1>RD(GQol=UPOg3dFCN^hxhM|C!d;
znrMC*v6|}=q0;qU%J5n<=Wcb+;kQtg)dn^!ZMJ(=elRtXZbSG~5imy9WOe-c=rEoP
zdwqRl8grpEi}1q*EL%Q-MPs943J=G&=G#ynS(IC!4Ry!g&*8;z0&TgAz_@yq<egKz
zz*ILS`En?7p`P#5qjC>;_V;cV|L6>%W`4@gH?W-T9wvWShB&V{2hic5R$h5)G5GL3
z2~;$kk#*rCDD^?PS=uL3Kwyn3;gjiIIG%@yv?cpz^1Cs~n1<gg_W?)S3Rq_3Ib8`s
zb{xCPxW&3`Rv(a5sM1RL<F+O)`qF_MgR(W#<A!4=C%`!z2NBa!mrxLH==0)G2_nwm
zF6&*BuR&-Z?iFKUMm1h7LnFHYa=C})DqEL;X=av3Kd)|BNe*qX$d2Yj)5i!L@G4xv
zr|hG4QO`YYT#ynQAAJ>=9b@uoJ-v#SgW*<pSJy~V>*%Z08eZ}e^0QA^ba@>jFwwKI
zu&P=ZT}!W*{^}Y!Ct;!MRAgo2q|*Ji;f~jey1F)!Xu)C|Vt_xw5@Z>Fx#=i1yLQFi
zAgp=qlx&^yA>2MZ0P<POB11X(VF#P+ox9;x*UaXhStE{E0H_xGEgm#liAPBZ5sBIE
z1(&|MfxhP+{|KEc;3LXAoDPdKxeta)`8H%|%rwXrXM|P)E3fCWYHPmsKsVhy5^w<{
zD?k!C>=5x*>^Q<e!@)1ek6C589&CkWEGbJId%=?qHo6Oi5zaH_%zt=aDpY!pbxewv
zQ7!9Nn}?G7hf(guWDQG5M#+?^{y20+9MN0txhUqw7!)z_l*cu1D6fnjN?0$^f7l;l
zpfIW17S+xQc)OLtqyEl9g>ddpk^Li7)}^qFRcS0-Y)BQya5m+o^^@?$O;UcXRwss%
z5`eli%`o~k>?{s1B8X(B7hxoN<n@t)6K<E8I2r+<w~y*Gvw~J%Xs$`?q4R?vn$wM%
z4X-=~vgS65<*ly{m?Mx+qLxR4uu!(I!+VNV)gl+iEcqH^pQY44a9B+M6gkxQy~a-y
zg3@w|@(YULN#xZ;O6Xt?UCR5Hq7D8A)0@<`C;NV;aT%?rS|Po$jJxhIHqdzHMONMu
zTN>@_rayBK@Tt@Se2ulm%D1Wy+BcA2;tuwoYQ$Bg9Jw-rklp4uq6AT@Y`A74ad5y=
zhJ5%zKT+*5Mp$*M)l%bGC*yU$;gRPKZ)(OSi%8l4M=L~fw-yj%(;viEw^(rT<YY^U
zNW@6kKp43JRvlvedKm5=9l?-=8?msly$CN96O|0<V5yDRiJ3QWl&9uDsoV3s;n_Qq
zU#7xF^YYCDX8I-*ezJC6+>BBy^u8b+6L=Umw{}Kq(*Dzpc~k#zV_Doi5%Caedqip>
z`pM&Ef%;YA8*|Wr(uHe9=pRy^wT#99AqfUU-D#=se{b}jK16H%fKADS5u>Mw9{PiU
zn1p`eC#w(-AVk4ow4K3_BjS-7=KaUR0s}`Sr9<qH;$0jbUy96YNCViEo45<@L|jj6
zbo02VDP+D@W@?FBmnz9CG(~+_wlp^a;_hg!CN3<5(rc&6)KkX-$u>mv@{NrZ(QCu3
zNBJNKDfD`b?`nEI-Ux!k>besm>!==q)pCvK0rjFOs9q(yL?l`-s6`UJ#-Hsd<Afet
zlOQscukNB41G^k^y1LP2r!08c*eorC_*=2WMfLvf?GcZ=S@n4iJL~@43RUUpeKc@P
zzMt!lhCV*-D^N4ZYnH5M+kX_Sf+Qk((N!c;K6Q{muVYMVlrAOnGO{-;wcj%^x@gGK
z-oO!V2}-ro_wdTI)!=)^q}BS@S_WF3p+5b|2Qy-p6KW(ZZXf8stUMb`uhG1}!E!fB
z-iUHFr@VsuwpE#lwOY5Ttqp2>RP1hbO3b-OwSp@sQ%dfO(gI|8RuN%u+EMwVit#LX
zLEL?vJ*SwC!+2Imwh@awKlTRZLELa1aq=BO*t>bZEVk6^e^_bAN$0bCVp*VnY4Md;
zxqCKyBBlgBM5Z{Cq;f)sp=Mvaoz*G5iXYWjo<=3x2FjCSEwPwF3FkXA2jY?_PkeD9
zxqo12FpKykQ%@iwo;IU0*=%FCmuT70f6VAKcg4GOAiYI*efiw#go)|E038G;v71UK
z;7kd_qc3G;vr}eWDy^F=^+u(U)i+P0OI&~!qFWk#8hiflovi#stU&hC>q-GukNb#4
zxN3z6StO7XBO2|OY-SjuQ&f@jYIuD860)ucHQ3{jjFBVO4P8k`r<1?p*njlQwn5b8
ze$!bJzIC%t5tCY{2)dT;oZz}!OrpuGz=s_jGq(zeB*DV6s+D?VdnvT(bd17A-9G<~
zYQR$4*;x&>aWmGpC%hF(d$YnyOb}Tf_AdI~>iy}ZN+dW!%F+VG2j2soWYlV7`Um)e
z(Oz%D3;Ut!7=R0S7Uvlw%@?ILt@)As3sk}Ckmp36^r~$_)Wj_idk%9^Kn=%AOA)`H
z6Pufxnaq_CDa@=fjE9LOOD@{7I$>PLyWLfhKWmlkV?`itO)I*nd2HsSSMPZ~cy_qm
zS|;oCuX9n!+`+9W{hE1`)ofnwrp?p-eyP}uF~1QD6FJFGzditMCP<s&9U;NMSUu#W
zGH?;1yqeSv8myh~t{pIu0s;j<DqJ5oDc(T$?4|6IIy*w~x-u<U&kLI!$d?%B#t--<
z{TMHN%WFb0QM5DHY~8v^#RXdRS09DFalTA0#lvbubuP!k!s6mvE85NhYI%<ShV=tV
z#r6Zc>Hq$5^69_QMxwg^4qQp5!DRpOyZr4OJH{g~YVt`4EN=O<+R_BU50~b2HiaLl
z4?bAJI8G6D1-T~T0a*kJPkFL_U{+A0aE?@pu2)5=w2$~3&f<lS;7OV{zy7(BB2V$s
z8TFVPH@@~?E4TeLEMkx!V91j<_K|J~Zj29Al48Cf1k6s!@`yQK&<wi>RrDsspjgt9
z^`KR+G@SP|)td5Y5!rwJwH*6yqGE%VT^z2@T8GGLe>wp*+qTl+5Z%2jpQT*zeaqT1
zZG(V`Ixuw0mK#y(gS1r9O9TQ$wlkfv;8xkBm(1|zNp{liR69~BRq<&!!y^nkuWu`U
zFo4PEr?kV{UlV{scp1B&9xW8~B}-EcsA5&|Jz<cW<MKF|7G7_xu#~%l)Q>`8D)XkI
zK@X~&Q8!)&eO=s`;CEFBTTUugs)k<GWPwad8gG7$IXCJr1>>*^jJet2233UQ5}%x{
z*tnRe#TBQes`?ie<8%io2j2}>T`pe7tZa>esMfnFR}l|FI8(WN`>WlxR*Q=DKHhzs
z8V@tCD@eN@6D6{1(m0&npsM}OkQQP3>vY~qoqs=ujtxEW<=NgKQ796!<Tt2EM6h*B
z2YY25-CBz)2o%oGk10kc)1B~r4tv9odu=7S@HT@Bcd($&Rc%p5;bnPTx<Z_GW3MyA
zZS@QoX^#d|9Mkq5!PJuz#(|*UbY$p_=alHiv}NBqs{j4;u1l&)NvZQVMEa}LS-q%s
zem?OZu3x``Qo)UZ&zzO*^vM0(Y_;cPu{b@*d;;o5>0f$Ja}MR6RRM-JG~Z0n<tQ(>
zNcZ@ci+$vtrSsNw>9s4Gw6k85L>j65l{)D4tM5T7c(B|%^_#Z~a(YOMUoc4=k%)}S
zxIQfp4_4sC2m59usH>Q*nu{krb3ojxYlvU0$vtq6{m5FWc{&&Q4L-bNrl`4JF}bW1
z^M3i!i(j3&=r7^{JMER>*dY~>-*i8orq70#z5NMb7cVLKw&BF25#y=)ir>XC(BY9H
zUW=R@)~eV|VxtnBId%&lcz#p2eq0zxclDmPeaKR+^yi2<T({lQ%3G71&Ubfx%*1TD
zEN!qCuHre6RvOF77t*Ct?^_rxC=8i-=d4)qe8!i(5i?#I|HklDoJTE}%WM>TfsWH6
z<4vYxx`#B)JJnild(xyog+csj>C>mId8_TC4sFqI;y1Y>p4UD2^q^nJ?pwB&$$Da%
z8{vNHQ*XOoc^fN)%Z;7DfKM#_bFVA+vq_+r5`2HD^m>>mLGvg5h-LU!7nt8AD1lZ%
z`HIKJ;p>CV9V^OV8Do;?cX2UYO@YM|gg&s&!<y(}I=c8A{(bruqA|q2>>ExmpN!3a
z@{SfEz>em6Sg;@-bjLj4pU0<PdShN($+S6Ps{78d#(YC%KIm6d|M><qBft@^H;Fr}
z8Ld<DbRp*$%gdttGj}TK_}-t6?f2Z3uTlr(;`V?+(*a8CH?mg!y~=#qeEQW_AtanQ
zlLO){zlLiy(&LK#U5tIPK89Y(g=t~2HH15U>vI5GcH!8nqjNwSjSJ`v_QL`HV!x~>
zivZ9M9SnIO4`|$@67?=}<*q(KfY9i$g2lUJNdtgT3p+8O7@za{JRSY?3!bFEnQ&D-
zW9<4OizJyiDPm7BRRYWOslxrRcl_%*Mz`<k0z^7osZA)jKERNORkWLQLBS9gam!S(
zHVX*%d5)fIXBMLz|HY4IZs3tRzry^Wcc89Buol=Ndd{kd!R)L8mLjdz`Z-|0Yj|^L
zWOy7&I%<!bH8+-q<6z0vkQ>k{WuO_un}!pu73!umz8C@C3^sQr5D)Z;PGgY)qd7&j
zHdF%x12L18wk!)#?lNB9Rt?;}+xw5f3VX_-@5Mq^J6MipZvTrE1tGw{^yL3`2CkVz
z8Qg23mOj6`yy_*#ygj9C!=fSO5Y{1dMmP%guNyH#u!IE3dy#7vUDfne=)5O}R;z<P
zoG)djy_<XOic!1qgF;v49@YmCbR<9FMd_z>=67_85C3@yW9S;3ou?;`y4#+k4a4O%
z%6LXUie~MZ$4=i|DY?C-5t;9)@_FF8eNST&<8pM)J1<EMoB0;rpOw;P$10xkM_a(9
z(;G@WiG7Ox1%8BNlX%<$<`svUAmG*N<jsrlBPH*p<X18BFU>kEX}g*Egs)P$|J-XJ
zjPPZJadTtn^rAXM%Kf6m|5m_b8jU;tKc97))lrzUk>vbm<4+Mw=iwTtVy0pVQ5?%f
z+Y=A<{Gakv91HF5admyS1+5+)L(NJ#PbmaJhqMgLZEyvr0$&A1>1Y515mUEHUi~lP
z`QJqMKiW8!?Vh{P47BeF#zhEkNy#}qT~UrsbK3*G8i%r=@E9#EJ%j#=)&ty(9JV_(
z|NKqo{lKR@TmKL$L2QD>!57;TjA>Tv-VhfTU5(y?R5bm&&BEMH|H5i5<?<?}0Q}um
z6pm0MIW;ja4R{M|Y^15Z)&D`)vWOm8p%`IdTP-J>{{+(<2OH=teg4bxXPaq<;e)uO
zB~Guu*PW+zqjnS>-8QBQ{=i7@f=+XZE3LFQ42HKHfsU);=vIYSx<#IL>$gDnB$RyF
zutpHs#+FvIMVB8^-*-_hsUE@dP<MRbDW`?;Y&P#GDPz?Q<~xTfF<M}`d5u%R1`gMN
zVdmV=6uG!&D&cJU)XAMo7f)FQC*{PLd$`1xsHwfKuiPv4tPZaO2j>$8_hdR;z{v_O
zfbwjq5tUedpo?@Vtq#ri0&!KXDlzv~(a$1t=QH2;^sbE&xD29g#B6f<4ChNm^&r)P
zGm7$=Zk|gna^|G*fZ^Ekz}pF!O*W8!l;*zuwHx@WvSkA*FlNw^Dc1(MaDxfWe{VUZ
zk@|MRIQMO3AJzNN)1nbABgs_7t!gemU3zpbxFTQlGRu9`qhb1NwU0_KvgP7<3Lptz
zLf`+SC~Rakys*x;Su(QCIEgnCSAIe16bMP4qpF*TxR#d74rPoun0O8qAdDE15Xow!
z8(-I8&?<?sk>^h4g}cTp=NscPuhP;XwFkwj+nensYcI36HTmaEc(M{*$7XhKoXj@~
zPkvl`{^ZBh+kFYC$zeqgg9S@7t1$=f1lp`UZtrB2#G0~wxHr}$;zG;y;JJeXKXrg#
z)|{<bN<B@Ggco$*Z&lsQCjTW1?6^O)?|f!Gr*2{k{mI{Fy0q}3KQq9xDW+uGSQ+~<
zqPZ}*2eQ$cYsMe8MBoj7;zAXCo1CB-wm5I@J+Fa~m*--a$|zcBs5iuI$xsDb=Vdbe
zzsR3z<<i_Lp5{PGN7qkV?|#~@d7YNIoZZk69QVPE1_tLiBHw1hp;rFIUO$jTXouJv
z?LN!is!t!sN1RjI;;yr)8b1jPihIf{zBF^w|C&~l-3BFAO!BCnUUY^&rSa3#4Mtc4
z`iA}UpS#Ud$#Ib1bdMjuY@x+cRfUUi`rE2?XR>EK+>r=iVCd(0^xoA^I{oL&C&|w;
zOD|J9o=g&2D<Thy>u4^k#@DRxl_ec)dlDSUBEx#IdnQnMft`Bm*PtZ174==zZ@N>J
zgnrD3lK(@lW?#_GwD{&U&BR9YwM7i$w&s<}J^|V|clmFbpR=UJlE^L6GQepo&eYVK
z_1F<dsvg@D)z%ODlRW-)_DVeMDU$edjsYl)$o7PL9>q-%CrEVyKHEL2MWE~gaOIz#
z!_t|skj*0OZPb-l#n_DVjk(?|KLvIaLU@d-uFpUPJ6b;s!GP;?TO<TY^jmdiI$!Cy
zVVAscK1S@pT&L=k;$nL6eGNjOPMsPU54{yz7C<saP@fFrxnpVIe9kkc-2|LmrWuO&
z3^myb#x7ngKx-)#Z+g>qetYO_e{|W^7yZn9#x_<E0VzHkKC+%G<>-um-Q+5}sq~YS
zAG=Dh^w>R~Pgk?=0#BUPWt(h>phPlRuE!sfcTM6{b?Ty1!#Ht0D5D&`0bi9cL6-2L
z2uL;{8qWdr;|NI+Mumla?>aQ2CA?xI6Z;IQ=puW!Uml_^p+aafhYoM2+GeI(D^V*Y
z>6k<&mV9?P_Z9~^>jce3CMb+zX>q;}AtO4=UMRZZyk#lsef5?dp~RafYxsjF3~cRa
z3mk}bXHqmPO6B)wdCDlVtSI@NMw#M6a+$Oon{!M`^eblb#i)c7oIEUYCBQMN01Pij
zo05;zOakUn`DfAV*pITgT8QLC_lDIkF95Ht)euf!C!tz-lm)^b;vEq9&|#QPO6c*1
z7nkC3&wq;~>UFHdw{1pcWE@Xb4D%O<P-)wCW4=2nEtFtaP?mk!t=RGrlb-Yv=v-YY
zcLAisUHN65*%+IcNT`O<v#IbRz*ngx*L$rm_gtATmc4<DlG<xZ{U&x3hwQyix~5hU
z<kjor&q9;U=r7VaFLTMjgW(A6@3i7S8=Sjc9!cVNDFJ0#CUi*~IgPA;HDq18@qUt~
zDW5(@LBbUoZ<|ih1oVj}pf}fUeDaR$g+&Dq%LS+y64idZQ%c-HcT(??L~$bSC1D^A
zWu2{0o&Wq_q8om<5y`iRAzFVBV5hhDz=*N>NPbm@8@`>d16oBQO#2Y`jqsKTp!<a&
z*AGGWhZ#*!)q~*rv8iM;t}5KQM+p;Gt^a;GqE=>9+{PuZeVwBxi?;awolkZiIRw0Z
z8NpO|(S?nwfo9=K_yzDq_CW{HdraTLZx4Z|h(sdRtcMEo$I{<QRAVz-Njuis7sox`
zae_M}H#3(8jmVhxm48~)udsVm_VP4Uub;tpnX{_WZ20;!GTv~w*S$=|RR3OtgT%$2
z$xCDWU5ia=v~)9%XO^OZS6v)wYUZlR`kT)6jt!g(@;p$=1JsXvByonhdB%Zx_;Z@V
zlGi`<7hT=@^Z#QSE;;=_HMqv~_N!aqC12R0T->ER1!4rH1dO`mBmSXWR=c2rlvS0w
zx{KSB)2m*VyTD~ZB@iWta>U4QJ1qGGTI2cC6dEUI%zr2f9!rn?y2`NtkG1Vn>i}2q
zv1I@EhLiUJ#=DWZb^YK1)bfe|g0##Bln~+=)0MtuUuHe2;)g>qOg#KxFqvPe7Mppe
z0Bi1`JEpp;n`uoOeVIuN52D;t7<t6aCzZ_7r0z=H?Kbh_39KkS{Tj$&fld3^t;eiA
z=%QO#df~oSam5wFZ#wq8I_5Pd9hB^e!FzWRqx=fVX_>I^T=S3qd}#iU?DDl&>@Fm@
z3vqnEH21F>dMwI+FF^h)h0})?czW*aS*O!yGQ)lx2LC}HYl1|xz0aW`YYa32ej%Vo
ztA(elUZgq1cao$Pd#w(ktL$2@O43=>Od-S6xR$q7h9O>pcXGo4nLpA_l?)$U2KY5V
zNY8nsxY;bC@eFNGd$w}#v%GtK<4t!Lsoj$6bYQ9N1rD&+x_Hm261!zfyuSO}+mGs2
z&kvPJd&}^s4`961qsu-3q*Q-(h6XJ%t+w}1$|}I7F_K222jbb)b5?tqnPJ&HU%!6c
zZh@M(^#%Moy-1l`MYxagIrvU3e-d*Zk-QW^HF*87kdDrbt0Cas8Aq=RBRa+Rk)5pU
zj~*w!I#GtYq+XT4j)@a1dz(`#k)nFTew23(uKrOB?mZ|+ZjMawFtP2JOj<{K71W%<
z+&4fb2N4~lH5xiGe9B!oMQLL~0g!&?lK9ag7Rv6*fU)TX2FKGPNE@+r?dE2P^NZJ_
z-OLJ;>O<y&&GCAv`QTq7;=K4`n9kRomdK^ltHWOe8J$P8S_qPtH>9)LP2(?Z>RQ;w
zXnA;iWa5&(#CghHd+q&WQ$Zx76R~h!O+)7aDeXKt3lBIm<d*bT(#1Z?`TiUCBpzpI
zM88ht%_cyU2g6?rk5yMTdwXLYdMRYAWzTQAT)U#^n24FL0Sjjp0NPu^vOOS)&}IZL
z$naMk0g7Is0zzN;zw<N+a(B0=N@8-(SqN<u7wa2PW{Xi2T7K|CH1vmvhszMs(IiQ|
zP*Und{8>L$Gu_DnJu=_@$z;0EL9=aYZ$Bz<(S-2U6SzDSHc+*f`BS1p%uaIUtK)sL
zULrlLk~Q15u~+ov_e1c)@~xG(xP8^Xd=UaJ5EaYC4Afv^9m}1%u@WitbKG2ROfvI%
zOf2iqXlVDFS3K|RM79TLAbJ`eCgJZszyJ8I+cPdnBKy%V+qPwR-bT0UOPhhghALwe
z;N(F~n$WL4YWuujeh06}OEaCd`8E?J7&Q0Uh+rGUf5PqM8hOOtqA683Q2(agr;?CC
z6W7eF=l*-kW8+6vH1o5Vz{1%OiozpD+YJ)rTaU0=EoT#+lWNupdNvobZ&sFVIJRPD
z4D-tc`abrjoK7|y$xh03os{=7C>k6Or<appD4S66;$rN#e;M{=>=i;DLYOS-4dKd(
zKe!4WXw1+05ht_%++%nx_02<OxC6`C@D?KYxqElOZRsArQ>mXIufD>}7qL67zv-R^
zb+}du`#HRybKo4wnQjlTi|_w3RipW)2>b4f_A2M3$;XIceqn^#jBfV%_%ZjM3(~eG
z&v~~L^EWqkz?*k>lVpbNE>6qtixl_<_+||+g_s(=Z8b$Yrqg!+`ok&bG!1SYr^~Sg
z9aMOdFYUsoq4ow~6iY5LHPbttPm{YTImk#*3*_)*!@!@c=$~I~U?8#)3lC%wMu^l~
z$W9LB1tu1Ut$5TADg^~*5mmmdT(A&ibr&K9(T`b(_^9$Np1aE(7=+U9Ct070ArwjX
z)&s9jz>3#$Lds-oABT2Y=vA^#obfzeXIfFla~|pdL{bY(*~-~`J(C&lBpZgHLh;@6
zs?^=_3vit5>ju+4V#W&|IkEPBCPk#Eu!%67T<OCc@#Z{uWjx04Y1ote5eLwXr+PMU
zkB*3&bJ4zQ2_N}g?iG?;*zst#tae#Jj`5%@i>?MKRu0a}O4hA*MTXZ!gYnQ+Sl1~?
zwHWm(JaqZ1TH(gBi&~GLPj~~a>5T(ot-5?M5LZifFY{8J1QVHrtD-QzbJtt<DK~c<
z9lcvT_GNFYS^D@TPJK||1BmCq>~t(Y`t$&{jZ1<u%ln3^X=z=VJi{hsIbetqLcfAX
zwY9xJif%)aO;dG!x3xS;(&`5Mf^Xo3t1y%9YiBpOp80-I+gAn|Le8Xmd)UKs31l#v
zN3=kl>KaOT)kV3GxsQY+NWEp9Q5?SY0lL}<wZRs-906?a<kk188QAP{4#4{Ll2kd_
zg3`|_`{d{T^v<*fH0VPSrv7GjYj*c?+#wSLbX;oA#<^&8j!!%DPaWIcqqu!vzo~b$
z@^ivz7U&^xS@OKjOc`ya5fwamn`luw!YA9mfsmhgL6!uBT#M&=9K(q%Ww_>_djLf6
zt-<gx$g~&w0uuNbtOn3EJZf6NW%&f9SC`_Q%R>0x)Y7SBdb>8fhMG_~EscEf22KtH
zyb`xuRgAJ^YJ7CzLzU?3?aY=n)=K_01J(-Mmk5Y6LK1h@iy0Kh<wOk=e?7C+{(AA(
zhJwR?`)O#-8vQ9HRPKksYHA6wuD+`vhUgj%jQgCy&w}oJ_EgC#XNWK~m=Od2B3(DW
zsD=+bBaeu47K2NGf_%^m1fW$&3ZpoZUDvyRFD$#M@0U|zrm&g3b8&pmY5K;T%wQwf
zhqjbBiW}!V4?C`XSXENgS+Migc8pPI%ao@T9L06!TK|x53Dm|0Z{g{gS5^2Z6;f9E
zpa_(n7uNrcdjEMDJMzUqFfuMU!?-IE2ia7}@9(Fh!gMXne0&qWDy;1T)sYmWIFZl5
z)M5JCn8w5yuhq>vuFu)`GA!XA0)>2KFYtSCoaWDSVF|VZxeAkE$-D1N9p$a0GnBn*
zxGQpte9aWjFWHP6cN%|wKrbJ#gozwpdf5Q(wNms8I)G)>a7=&Yc4+UNIahbWtNrJG
z>n9C;QD<jEs$s-zL)W#KC*xK~&kj7<!DbM$2_gOf?U<cVA49#3ezR18FyNzCYuCU-
zvtYTOJD0K^y!ALVxLs6~pSfzinnD98RZ2vCeMEZH!m8EI?EB2k6<!XLNZ*w#2)wOq
zZ3ZRTYKTaDj#P~Vgwj?Of0h=q<tpmZ23OYPcsxO1l?1$ljh@k>!p*Hx8D))oHB#eP
zb;)c!J1IRfB|2)X)eJy=+pap(^&_S{Zgm`G4ORj@nK0=vdDd~toH0oh=e?)FoY-IU
z$eTncfKmHi1I%jBc<z`JvyvwsDumoQy|06=^T^gJj~;`~5<@TCiUH;4V!g`VzOLB^
zDQHe^J5Ae7Ql4dzlXu)hHFV{bEh;q$8gE~8Ky8U6FHF!v0@%RP8-JR-!Or(3Fe9}?
zuU|)3Rijn(&MsvjZC~CTAsx0$7hDL0)_fKfYr_^)S~1U*VRI5|c~1W)1NlTu3-X6U
zZn0Bp;b4=s1xB_WX~n%m)C(zb5t<bd2~7iYZbuox^~<F;aoxAN0+pQydK=tPCf&gE
zZ@Md7#CfRz-ZrzJt!2dOqx|$Q*^WxSVa4QuI1_F&RX&rK!R}E@Up2#UfNP5f*6PkG
zmbU^$E^QnFZ>Yw>0<+X(IhlicKC4tIXOR4^H@k|WJjK2FY^D==qb6?Yn1sjvNi14?
zUsL;lqIRHNe&T;rfBrujsji$pqcweSeWA=+BgXW85!SoGVm?Wxxe<fmG#9LLN$R*D
zHd61iD8N<5*eI7<X1lJI8?wk{V}GG^>)Q$;KzWja6G92n53x2F`<3`q6{|qOmfdd8
zT^-K>7PISXT66O6V*7qn`9=0`*LZE7fec0+0Wi}1xd=XvTxd-8eSPw>A2()@M_^g}
z0dJCLV@rS8N&O^+uZ$<ugKn2d?h0rfXo-3IV#t=Ug#N~yz-RZb4E<DCDe8W7|C4>$
zgP^ReJG+rD>S{oJVu%95jvZqnXEWRC<_k4QMSfvF0T<gwJ{7V<b1*oCRbQ%G^+?h$
zA}DbuiI!tMeso`k>{6WOQLnXw?KiU$$eZ%fmoLub)fNZ3f+Mav&W?R@e;xT^;gDo?
z9qf1X{LY5I&cR0k|Ev5g<BZWDH-XZ@!a&iUxRyEd6ou=&TBa`<UMNw@D(ap<uPDEZ
zrz>YRhL-L~F(x`APcF4i*UpLOAMSdx=c&TfYr1uZE}Uht8&wgF4lxNxA+d!#x>5yC
zNesySswfZn3gXPXTN5L_l+8zh<m7(OeOl;u7M*+XoXO)D=};*vSG-v->*f`IyvAQc
zQZ?OA6WmOF3Dxx|DPJW?5cG$VXK|q~gVQ{*nn|k5ga+iE>H}KZ*0|Mu%EIbaQo|@J
z!-vb(xPQ>XNG)E*Yd$k2wA6VgXfn<h4}8-dH&C|qY#xG%*mzd0zPGJPh>6(fSlh*n
z3;3-J2aR_wP3*=N8Fm$1`qh81Hf!MO)Yom%QYMWYm!(CFG(Ku!)BU*Hr5}#h^!#=?
zS>i$CvuBMSTSl>v+Q#0OO)fZXO7)rj#B_45vej=o5enxOkpzD|&LeSzrsolc4SQ)#
zp_Yne>9n}ddG)HV+u|?2Jk|$iNS^D@?;D$2KhAjonUh){`%RZz{L}lW5CSk%DkwDi
z?)q3<ax*>4WkbaPlr}7o(Tyw#M*@Gcy&`C``ilasL|c?1AA@+3M{W>)kcRY^{QQ4J
z^c!V$Tl}V5rAS=~$|g&vlKJzwkA_!cf78MIG;xFCn^%h;?7E!cR3zEgNvBaTA%aeh
zpATvAYSf!n6|6#eCx6owFM+KRl!BTizB6$p7J5tHM6xVu;O-ZNDiqP2G9~TQ)zqc}
zvy-w9RT)RbO`Kq%V&`sB1^`HwU4ew>7!w3)nJqdxgrDCTkn1lJ@rH>ELf?2yrh@2z
zhIBvEQi8&Gz2UYCwbBe8((_6e6sJ`$i01K`jQY~nc+R(U%;Hj2i_hfrEZBE_rxho8
z$_aPh9T-;?mQltL=t+ckq^cZ-g5j{mGu9OWAK0C~5WR17L-Guy5OBCoqR8W1V01q>
zb|(HQSL2*kz}`pIBzvtY4w$_hNf~HhWk-bL@BmghRUdkBx=0%?i8s{jhPn^~Mb=u?
zQA8l4-(Xa$V5Qg|Rm*vg;xH1F*97_wZN~A)ptv^*R6FKWx;_aOF!-Yc>%;Hm7?tOa
zv4}+WS_Hh?OZ1K8G3zJ}-gCQ4;-A5M<$XK35M_-j7%}&(t{eX%&n?$GLK-qveCN`h
z9x=pGAZg<8dSJi)Gh>bY3KDq@!lzeL%ppdzzbt+F;$f=lb^pS^jY!^_Zh!cL`Cb2a
z9-TR@rG#;O;obTmMii-E@9g7ER^y9wrhi_NBC)zg*TARS0J_>F=Wh@r%{Ja16!vBz
z=fzOERMvjky1ir$@s6ySGRB#fYvVoJ&_ZN#>k*;R=Kqw&Tgf8n6=bocL)S(*2DL*7
zxqPF5m-^XC?CFrQ%XsmotTpLK%)bMV4=gQKt>&b=xcz91;mz{7H9T4Eeb~wd+C{G9
z_3)%M0$&emu#0|W18iK!Z_$q={xpqWm9}?`WdpXaj_B6sasQ?xf!%o9X~?7xqpWw1
zAI*O(BMNLF5Z5$rQ`=i>S5#~^j+B%Xb>rvwtC2U9A^4cfqkkTzdy93-m&SNEyW^WW
z)LiVtYT!tMq_i}zQ+{>Cq=ttT_iBBr#tc;R_z&&W<F1j5eg!0pCVNNFq>{8G)pvx0
zGium%35(iG-Z43aLYxAo`kmBx8>LOA>B;NkLlwQbaxC1%OiWG#`<ScAW<F~AnYU+M
z|CFSr+_RoQEUgT0wB`a0Q|N&uGn4s0x1C{DHn)t@_=l}R`6srHweR(xOVR&BA=!k?
z{(1iniT}E7?T<<^%s)jidC5u`&*Av#4_vUr2dLNm-=UQ1EGq!(Dcd=7y6cUtL~`GM
z_z$Nxzcu%C`m#txTSE91zFO(WHlFrYFpMK-%>#$*`8Ec@jac#|$`%-Fb4|+Zn1_EW
zFVk7^&|?WE9$#|i*1`TG)MU&5h4n1<`m$vs+?<a^2|!IczRncxOI_xZmx{v@H-I~k
z?wtx5PEM+plkqGAdHi^gO#c~#&kBasBb%d(W=(?&{n9i1H#cUKevZxl+FtFuPo91O
z-=nk>v-Qa5;o5=rw*+QdOa5rr|D+8*QYMA7VLP$8wcSztA8?lA_!s`MKbcT<C;xU?
z{aqFHKh*%I{w{gC`(Gtb|2bvxS6y*1S<CJNG{85ZxqP5riz2p@x}WKD6;CQ8*CeOv
zP%U`1HFtZ`ORL_v-Hk0rBiSFHePr>F6QmOS*<|*6AQ?TGV*=_^x>pSvgo7wj+DX6`
z)koO*r=RNeaf`BT*Q=XhlU{?~H@9Q#)y92Z&WTqn=kV6{txNxsd-)gF3YyxN)9y{g
zHILepf{5lRie(kwV<Ot%*L@3?(Wds=Gyp^5%iA+@H{wFy%?A9upMPJW*(9JKZ0_}E
zML}C<%sS4ED7DEQd$9zsaYSouBi3i0)KHR1d-T@H(3SJ~t?&vvTd=N9m+RC7$Hste
zMbYWU-%T`Ge!WUof>9EqhAJ=l4v7c#8^29iNW7|bZC(SIZ+{1v14`sn_LuS4ttLLx
z5&O_sy72hSJ{s%r2E0)_9uOsbk5>v!wE~4dUA?Xxn?d?%E@uOrdxKsEKEvf3ugXbl
zkBZ^MC9R*Vvigz*P)6Z*FN!QF<l_7u<6V|A(rRat))jEoHa0Rb+WxpAZ$3DyB>2U4
z?m6!Mg`G0;yt$U>kBEHOkeSt-AZl|Dt&-FwwviQ=yQ35L{_dM)k$Y+i=&&(-j?L$d
z)U)88j8^0Am-5}29;kZA%1BpG25)JhWv6~EK)l7p&ef4c;)l!7qrU*pZ#jK4oh~bG
zvTW=$>uNAC#hm6YA>LD=MXO8JaDqsj7{&=#`zrf@-S011YqL@CSdXmmi`ojcKy5T1
zl$=}BhN`&~*325B{5m;2QOq|MxeKlyKf@WYB6ai1)!Ki@VLimeic?J+7M)5wz|{t3
zk+Z+)h8(TCM<kVbq59ob@hDkc5pl+=u3hqmV;iNGicny3zfV5Dc~1Gx$Hp$75-!aF
zoU4mE_DS#A$B%_IqCwZ_4)9F!@o!XMG7oydzohOf0yoi_zv*)DKOu*1&al`0Rutog
zoQQ#@s$0)5sNX*Go$}y=d6C}*b@R`(+zXzYnSIL7Fwifl<CChGB_lqZ_%ix8UEIst
zb8-PMu>;2)I_*w%x$b?tWzNG%%d0Z|e3P5y@K$<?wfsTF;VrhS+pYd+t($CSoq1(!
z7W|5@r$;1WCbzn3rpmUghDm$qS0$+ZwhkqZ(%ieDTkjl#dtC30oml2uQ0up<20KU&
z#zL|wgNsgaJ^h`=X7rnMbab0h@rTH|lKU%@;=y$jR8uTIH*m46t=rOviF?6W`kp&F
zs(~?JW~HpG5_&cD003(%SKhoU)4bvMNIqu@C)vu%;ybqac2-eZCLdc!8f~*kPB~x}
zmc@zC(W%ay<c4l8+Tu13Q9l;PqzdGQt`PIcCZ?Ql{!x&w#x@NhE@7oz4ie^?Oa9?g
z0~u`G*@DXDwMQdT4=b^`OdhJnpXoERjM*+)oCv+eUHF%=wrW)3w=zofB1LQxG7?r;
zbDPI3?gG)O{C<y|`q{F#tP^NjMV2yKQ)@W9d^Ar#@n(bRSs>q!F2TrtBXs<;RAr;H
zDY=a%Cv1wxo9e=J0+o^!()oqoRg&-;YUZcov0o+Be{{cuSgioH<O(&cjX;;hisB+z
z(N%@TGBchF=!D#DQBJS(BL}i>0CJx?8eQpLSMa&0SR-|1vewsXidJc-my;8~%Bbmc
z{l!OYb(M0h8{f=5Hg;J@n#1Tx=3%|7Z2WW|XltFd*cX*oO^0z~1qgY!hjgD`OFRFg
zi)^p$=e30_^{O60px>T>U#P-y)kG~3Zp91y*uPzta4>e3lU`JmAtYI~^?k<>1T&FA
zC2geU@nmGxhlMc$sc}|Qy*qo)E}C)jG@E>p_9<+(lL?kqaF1Z04J2Rhi3L&k`A&_2
zd=$9PbXO~!)zj?&>Tc6(&@twCM<rwpTA|D-Vj{GpfI@tg`dYvSdcDCGLPD1?@+GIx
zR;Y8@H^%ejf^^Gw4aCY0A>>s=qX5~EVqHlk?*)6m)%f{FcwJm=jQ60ZKF)R3#utMU
zQcaC6>Y+T;y5-Hxq}7T4c8O-1ZExA)kqUJ~RD;H3OX0=Ig<vq=g+ws8b$U5eGdNBD
zIJVYF)@taRhZG+!FgkA|yX$msZS1&&2`k$KvmbXA@R>#py~6sJhape7<t4g^4OU~M
zMX`oL(2pnNs+@X0;0upvVB#!>pwaM`<eKa+@>?Uk#Fkuxw^z4+Sac(XM@ShahOMTL
zzNW+S<W^S*Zx6GZ0jpXEi>9Ay>pRjJ(4<_w({5lz?G<dW^mF4-bJbBsZ&`7HvDruO
zy)x8Y)5`~CR4yUAV%6YbWko@mUecLMlw2WCAHIH$kgZHN7x6gV>0E)Grj$Vqp}T8J
zON8YOj)BQd^XXiB<Ho+)YV@^Hy&d$q$DYL^&WUPUImq6l3f)n;z)R7u1C>m#R+Z;g
z)eU1{p)Ip{2tlVOx~wzC(~9RQerC+(T#45#Qt2cCj1NT~pYkv@ilKg)UYhpExqE8#
z!S`nr-L}Rizs&$<@xNU(=gu*-?KtPP^cSVhwMJs)#QOV_vZ`|F^&3OQq(i>QnaK2v
zkOsDzEJ)XHPnr;Ym>ve5<t|+p7Q!CAQ>epo$N19yoMRR(ppEVI!^kz$5?KjifnRCH
zOBXY6^L#Cs05m))soJiZlxxMG-Z#|ePP}2I4D<accwa(xCtLy^2a1^fx+yf@pg<fo
zq#HA1{(4PZ;=e=apFTZbqw|&wjmtRYdEScOhF1~*I5EHUV#{<@JzRh2hGf%~A4Rgm
zuXZqzC+YDZV<3zM!UYVT=e?Mz?t9YsBCD~0j?TMpMcpME>e0L4h<x(kh?a?E_xeUr
z0@+F}`O72;61i{XJ1OO0#$)Z<jD!zVfFWRL?K>0%R9TdH-&mIo#E@vr%jnU<*)VE$
za?I?^xkSZE217bJj$(m-qp17|;{9(ZmC@w*2fW9SDu1q?I1cT#YWbCmG~WSMY+-~%
zfBox3NZh~pA+{%<A-F#9tOl00yIQuS8M(OX26Z+A&oR~w{tK$*@9s7OX(lkn#&$PL
z#@bJ-|6Sr8OZ)!O{Q^wr^w!52PMe$xeB9Wy;#tfjy~OWkLM%Xh8z?&r(5$GZrY9Hd
za47ORzQ7M$nhlH2qtyaRob@TX6C)FXs@~`@_AcWULT`$QZY6hv5R@tNK&`p&#CyeD
zVEE@j^-8IW_c<gK(1Mb!C&w5QBPtbbWF?V0qf%zd58DmHmSrCw64qaA=}mAC7q-YP
z9bPK2XRkJk<z^uN<lpc**pVFfmKdoX8kWpXu(#(iO|6;-?%Hdx3luf4j_a{n=&m{(
z^(rd2-G}TMwrN53ZmU_BE4gOAf*QDtWoNC)NYs(7f$0Ol6<-o3L&z%@#g?Je;2(Q=
zReT1yNtSZkyA%S#9?m`>yaY-oze7-k8;iGikBwAlWv*3x70ha4gS<=_D<u};{Rk1`
zCw+Crh5~6W<$GN&<1l-#VLnrrz6N+XXmAjko`qP{<O`KF{0am?(w6T%`3dNTV9PCT
zxMg9^N>g+Neo}lHN|YdoV(t%zJ_j>>7F6xDcx|W=uuyoPKn3lK%Ib<aDu;X$M0lES
zYun!^fW;x0`Z*%aYBOLbAbgT#ixHd5sp14VPZZD!ulKl6o$5YWjt`WTffh5G-@XW7
z;{EX|^-an&qePP>;5H~D_ilbm{rOkeitK9}ref;jsAU_J7r(Hy1(V}7P1jTT1+|WR
z{oK3CLEad-mDN{h>qfpYroj`cfD_AZZcmQ3fjds>0`>6jp`bG0WOH<SYI8d7^Y?T%
zp1O9{Ee-u{RdGgUW>;NAcUg3dCpImfU$8)_4r<(Gt~vinm`pEMTzt2Z*=)pWvrZQw
zeZ-!c1{)a~RDL`#B{eMpYPB@h<G9{#2@|Nf+^FW>44NWb8iqU%st%NuM3<LINlT)P
zp2|wQ1^WrPofz&d$z;fXk<riHJ8r%~KLZ}ViRcXEE^B^reoZnoynoDs&H9sSX8Zww
zMEmA_3-cN`y)|^O{q9NSSqvNjv(Q``j+lJsBFPU}<zxX@dlwgss7PUy1DRxJpM0^C
zI(nnyZ%2V|(JjJb`Css+M=yKT$0o2tm8`GLBBl_!@KgLC-_`@Msbf6@s^}MrK-TV{
zaoWG>L_38}dw+G^kyaJ(6dy8o9+<t6dv0c8hyRyiH;TP9a6W#b^l^^w3+uFov~NSo
zmnesPN-oe%a#@X3wv^I(y>h3t_)j5m=cstEsLb9XD;$ogT9Sp0@dSQ(HRp11Ax`p4
z{W}9U)^Q87zHPVoR|l&3uSjbG9{M`3yV>HPg%_U3S;{cv14Pr(WJU{KJJjh6bq+l%
z2n~3KH{gR3>19FbRaOtpn60ZZ>_1*PZ#)hotS9^#a_JkT^4#5yo66%uN@7aUge)N3
z5^;(Lg|3OlT~^(Q!KUVxA2iPr0`;x=%04vopEJ<C??jEe%&dC~)kcLtn$`H14ApCv
zn8Bk)#N1BLqx(!!agKx{gsw5*l+6jA9MMQ{ZQv&(WbLZ-&9rj+t2p>>1yWH5GpIm4
zcu35Tk!Fni>i?)V*aYRbi;_hRPh0xo!QT=`G5zx3JSl0Y-gfmoUDQM0)<_%PhP#5a
zME>K$RY@~aGh8~C+!E&6lx3=`Z%Ewk=OTA?NWqaGvrek=3OLz1%4fijHebFNL>A%-
zhi)?TkD9m$3!bTN#C~L>Uet;-=1sbL^FRHS4qNYy*oS!SQNRY_rUYdtio#CBoMmd{
zFKU$Z#mU5#*W>BrvVvwi)P7;zCwu!>_&ojFd3LmAkHIO2MU6Mu<X{~NoOKsG#&>)W
z4uH-!Y-Z+HcoFkURWYIs7X2_Ef|R5%z-&hMgx6%fY3thGPp6R}?#ozqER8=R!M^7p
zbFC(J682XRAnqUWq*R|%ajYkP*}4>azI2yh;w&M48J83DlESpblnE^qxO>y>CTzc`
zMXbuI;LEKAtbP0~`};`z2%Zh0C()dGI*f5cIXu1^udJBsA1Z#PnV)-tLVJ8Q)K+_Z
z<v7t249Te+UyUN(myLObX{Tig7t3;EMt~!Bk-1dRwMF=k1Cr<_y$IdQuaDEDd(79j
z5J$Z3)q_J<$_`-x<oRNUeqe3#nGoLAgIDJ3$GZP#0Bi~cR=$L4r5qp9YNLk#{xZ{s
zjoBQ_KTOhidI@R?j2$q;+n$kdk;Sm*Yd(snf*vY>QVy;vjBs1QdtTR2H{wD0-zUL5
zyCdwcE*Q3PEMVL;Oxin)H#-$P(9g#^NcrPBStAB0<HFL&D9p`85gV!lC_Xv=<&6i*
zcc3QauTUp_18)%YcB=B{bgVyxF9U1g$!J=|;b*6K0dk_KGgl8{*P-c`B8@C97E_dx
z`feWa)WuQ37|nT#3Bs^<N&%b`h9{aJQId%MiCkEzWR2zj)!ubSHI;RHnsgbZC|wkg
z8f0inQ2`Z5=t&4Ah=m#;H0gp$Q;{MB5;Q=N5)u*u0!ctb6cErj6oJs2fG{Ld6dA;V
zFEaDZe1E*R*7w%>zO~+4YyR2m-m~sGcb~oYIrpA(_TE2NFE2|#q=VT2tR`*%G_3zj
zlyc3^oMDeCHiusovAk)Smn>MsQ6r(=-HoE{7CZzaIo}KzJtO!bHPi2c`TwYT5q~yO
zqoC(0#BSizI|>5$PQS1gqB~;3&ZN)O;JHkFxfwLfV64#>J>*b`S7=`)+E6&)pa;Ra
zh<)R|oVv84i_l{<=&Zi?;7~Apf`UAL$12xqt{u!WcM?=t?OOdLHSb&CbYKh_09&t^
zO)=h8R!8m8_ho>t=^oBT%>>%sr7Z(V1DM_M1Ltq!yYGPWQ(ETJVSLku=$%i<OC~7n
zi$jCStT&5tt|sg50&)ow*-9}{=C7{)Y5ThMaTdc{b7+q-?IFlS0BbvcLlCi~;4qMp
z#@xdf@@({t8uQBVaaju7IkaVvilp(QZl(_xf`nfo$(mtX$>j}GI9)y1pEd&u&HKVw
zgQ-$jB-R~l{4I?cDT>hq<pfGErDvNU^Q9xR>LSP7pwSL<zemsghtROk+OU?Ukl*qF
zKxuiUa1*aCa-(hGIqs&VQ<83u#yFYrt}AabTY&wQp;9_7p4vv^*jteLMo$9J<L1vT
z@B0ce-@b8C9CLo-BJiEdB@95^J~arve6Nrb<S_J%pQM@BXO<%dHe#aX!{oNOb*dtx
zi_74zD(fchLzX?{A(6{K57#YJOxVe}s2}Y(1(E~~!j!x;Uj=*hGWTfG-3-bBDK{PT
zW<IRSHHBahud&DjOt{2l`W@9{#Sori(>>SdKX<=UlyW-J+_X-*8%#VH2{S$a_};o{
zBH(li-6IGatX61YtRX!sH0@iwWS>D<d>wtxo7y)%%*xr9Pdrf|E?!ZzEwJD8NA0SS
zw5Ka*UMv+;a()Dl_H8V@IXsezk?2Lnk98k9Cc{{&cz2vs$y$X4CHkK67K-|fgOtTY
zov^p6`#66r7uT@c3Q^pP8~6smjMX>$VvosLd^Qde6S6tk5jM4e;hg$*hBY-tkf}~k
zSYtBwmNW?Sb4Q%Q)D*-u4eD*ide2XKAX)$uzzE0|@2?EfQm7_Agez1<!5vJRebCjL
zd08Lqza*-HmxsUU>yYjz%&Q{!=+*o4H|51oYiF;MGX0Cal-o%^<Dz2ONmJd{h>1=Q
zvd#h1sPmv3z&KW9m5&$~%CvHX`Xsw-p{Hl!oG|-7H(*F3wP2rssJwqn476YCkkgBV
z+y(uo!o=SgNy8<Bz&e*HvXsXrd<&EdW?}LxR8sl)ZkXP%$pg!^lTOx`{~!<SUW_9R
zeK}Fs;gvLJeUF_uMyg&8Z7JJ9|LN13fL-P668V*8P@8|XA)Rg5imFH!`Q>4LbxL=v
zMDc{}=vuA8=D=$UPxk4K^Jj$I9dpD_#2qd9!N*;ps;)TZv&o%;x}izgrYHUnBI<H@
zL#tBJ6=KpSR##btBYyOz0&ugeuj*3TnHzN7Mt(tew$rnj1@Z2`nDkV$cXxLBW{HGa
zux>ehc_m8FIgyuNy_pk}Vmyraa<0=+vwp(biO|lAA4b)LDWS5e-?ynLxMOfq&MRmh
z%0}B;X*p-SHo&6*aSoScv?VsMoFcS16we4xPH0FUkd2*ShWu)KJ&AIOsVnW!<|ZxN
z^>gv0|DoMJ><YALIpc8iQ5MqRlwuA~U|i}(3NLSOS8PZTiTW1ToHedl=&L-`+w29m
z4UnvAkAPWI-ePusyj^jZ1&8cx?2O$MKzT1@kZR;K2owTYJMo0%O#k(WpUUwx=v80q
zAH@8A`rbc=C}(^f^}91J3`ZListJLmD9enH3JsN%oUr)Y{8pb_tm#npt|rf)|F~JU
zBYQB&S0UC1HIr(9ekai_^F{~o+qkLy5%pg+-0=jfXb)d*4TqRj9_v^aYnmsA(4j*2
z{zWz}pV@tWm+bM&8Z3}~>&hHmp-BZD1|>2gs{)miIH_0Dn_phU*Tn|jgJheY2RAoP
zg>Ap`cxcMS#Xa<gQ3w-He<mlzYGa;LJ1mp$9k*LM#W1eBIr-8QsL-vl_MxY*AIyY8
zhE6+hR<q)jo*ovEvzg*~Q@PDQ`uo4+TE#WAB7)ZZc2JL0nNv9t`#`ExvrS1+NnZ5p
zAlhEa(BJTlw$YZ}Y}6BK%EP+0GsNsiG91w}!+#Jbof?dY{knk;W*o%v9V}cL7a2We
z_aeU4;hoeLCPXSy@@}2uS*%gO<RwkS!viqM+{u`E1@M=4l74p7mD*j)X?}wbcJ+6U
zZ|$=H&vU8bP1>PO0$j(#KG?6ht)N9+AWuT^;iOLF&|&*zMHm<3R@^g%WOw1lK7oAy
zYrDp?GZ9qUFmMz#Wz(r@qIbsbfvOTtjpw}0g&*w~b42n(hq-&9t%Q9dB&A%{-xHwz
zhrT^Fj|;wbe>(!QA?GSea6%4}(n*GUka=j$&&|zS=W=v#NsGfz$8<e|kYe7Cdx3o0
zJ`y93r_RLVSSyL%z}g1{O2x$!=9LH+xS){2^HQaf#rsxE<=Qcrnk-lFl^4w$*Cd7e
zN_sNY6x=?XyOvcy`5JuZj&}nDG$oNFCb{!cc-%(%P09?<X|T&s9r+`9ZwpH{&ah>U
z=e9_PjEur3)5X*MhK<<IWLrsQjEBjcSN4$JlVL<1np;bLDN;>MPPonNc#K`ixpJg|
zp(Int@a_{3L6X4=#4KBQM7wO=6^(H?eOcS`=G0jYs0<w+P&pQ5VFPLtmNQ>S78RDA
zAqGaC$1O5l6xvKMD0IR3%dHeg2s3%s+HAa`N!%gUJ9_zQ*oa5#0C^<MX>_>oknW~4
zGDIw{F+BBwqA*wSQh2uznD+8*5?vD;P9mkiQVrN>DXpmXRCNsvVKpTl{iI{$NPO9x
z4o<wiVBiH(>N-_9Sb24xtY{+%NZi>ml)=}V+S{R75;jYz8lAT+2qH2v8<6gZ@Rpyn
z7pEVzj)ul#Y4m^#bi?>d1sxd|%xtr$c|BWPdSW}q=rq6ddRKu*!(O22YN^gH9QZ&%
zvZW>JL4?`sO#42B>6v%K_46X?HC5-}oU>G8IB{(3lw%X_#eQsWaWmyyg~qgbIL%X0
zRUk>@(%I;JqUD1%&L1$?j;lT5<g9!BNQIo@l6mi&ag&*wc1gr)T8wQLsyT}lsWSm>
z#1tg84^%P)rDtOn94`216C5+R@h26<`mM&s$0sJn$LDFZr@HE|5Y-Q-Ny||Jph`9r
z+vz+ajMqU?4r@BuScRpPo|wvB)u+f2N`^d67=1nl)oOC47LYZA;yw;3<+%0`0^W)9
z>nCM0>hD^Psgun(+j()%ZtWW-lHgT7+Lz$m);%xu6H|cj$k#m6Z^O)k4KO$|{I;bv
zmw0&+4|X`;Z0fgx9$f3<F|?E@W-J8n`O2Utw^t+J%VSm3`WTcSc_wu{&b*-+vt~f0
zMKO@95mrV%liR)+BR+a6WQQ*)GtSrAy)<~Ukx*;va@D0TRY3l7T~~YZwO^l|=tf02
zq&HNurh0&+xW!_~%3&6sQdwloKqZHXZETn~^xGW=EtU~3ann5>H|Hyb*^F5{`&z-M
zw^+^qxIFMIqQn<>lgX~IxEq#El3cvkc9xIl@-`LwSe+G{i^`SRXL@hL-Cd42G<-)Z
z3fWVb)kSVU1lIBt^eb+WD$bnskFZ+Z&Z$>LX@9h7F^*k)Nr4LybfhG1w)JwYPRDFM
z!*9c$G+H3asUPptO=GG=WvN6wbVUUTxR^B`#=BD$R6(2HlJi3aY($BTv~KQ!-#6Zc
zIV<5>Th*(Netq`aRAECF5~j--Mf$>X()PI|b*S+sWKsHuP2RT5jm0%8lGX?2k5$o1
zSBg;;ISG@5@NQ|z4ihrO%TOW9z`+SJXzG5!j%^F-J0Szp3yYoq0fP3?uk4A(1f2$E
z{tkks93<jS1SL<px%U;`HONdDsP?yTLsStF%cF^@{im9-QOBLA6cKmE*LMpk)1r^P
zLE{K)8G<f&ZZDWEE$818bro*Ci>JbaDIIBS+>dy~d<`EAd;%D*avB)E;FAil(NgRA
z9VU$fB^E-1$9s)kBbCu0gOkzcj~#Brg`}5drkO~zEX>DF7OkNT58QYNXYa9<>`l(T
zf<=T@+3Ugt-I086R2J6Rg@KVs5{YRYc$9(#89ln0(+QeTZ3)gNrY!N8%-T7(Y7$2i
z00dqN{wh+w)Ln+*7|SZG#CrItS7Gr4BtVbBi!3!0&(Y;Lw}kvd)9J%!eau(Y-1q15
zJ^~HF4a<VgM`O1`T};>151c2QUhhHjpvDfx!HxN}q8+S-pbJ(Pz0*Fq6y`sM$F>ZB
zVVZc6Yc<5?o(M*CPO6Id{;UO`y&V$D+2Z_n&yEpmvL*+x1d<GR=mlr!5HFI~Bvq}2
ze!;y+nslgFJ|Ylq@*?xL!gKZ!J*60l>f4;+2iwkgMV?y-tq?>j^fGqD(H306^V=1=
z)A4=$n-Y(CT;s%fHp+sxCK_S_pYA;u6ucoAjiVqX%Z-(SaU0z={3avlt>{`;#gMW$
zPU=+}M(luT4_T^lQB&i-)9F;RccQEFN%|7DVn%8BO9-%qu)O5YT@c!2$!N{7Cibsn
z)y00Mr8AUTiPw;JG~wRZ3VAuNX@FO^0hxn1XPdKpV$T1>v-&jDgKSkIH8f(vaO6+C
zJ2pu6F@drKiel$ENB_T6N2pkp4Ggb<Uj=(+IU@u($z#$CBqIc00B4f9Ws>Y7sGYti
zDPPc1C*OwF_d*~*DBLf@*Sz4ss8Ve-P}(%LWdseX8o+c8!p!!&!_Ovfhybd(hN(+@
zp)&k7o(8J8Unh{vnmINiK$j{zV&r9RCNHz`R6C~yF~La=li>>oNm8mrEZxKCPEHXh
z)bY&A{3nMtLzjiwv{EdC@tKzsIE)F@gt^)^fHCXx0ztm!OMF{Vz&kxx-GS8_D?m<q
z#&p(ubLzv1&Mrx?jb>POXpnL||F;jwCju7(@6_kDd3ppv)1W5oSDa|HeGoLtftxon
zF{Lh&_-NzUtvr`EFU^S;3<};S&sv4&&G4kw2Nj~r9mW7KR^(}L^KrI(_@d%<B{M3X
z%Sec?<h?>W^tT>3ktWq~pFR;l%(Hd0Tk&?~dH5^5YFypMacYb=eO_hcLACE4qx<Jv
zG|DX}e8&UP)LBb3wF1qH5;ATpnt6S|pMAeZgLG!)Fr86I@`zg~Ulq_z_!@T+On+ie
zt}=MiyTn0Y&B*dBB3K{_U~a)CU|N-OJSsPnZ;g%a#XX9Af<ZlQM+C+#AP*v<AK};L
zawzkQrq3Um939$w$(QjtROEU)Ou9P!VSyeHteY^ouq5w;CyiM9G&Y2mhGGdmmxNxb
z#l2K1UO)kl?)mb24qT-%<zh}QT;m_kyPO?8rGNXsRZ?KK=e<GK&l4u^YEO||-W}w9
zwkJg3rdhrA&ikS1wWssEuhqqz;kGTr$wF8!gAE7vi7MQO4lH}fW&x{YoBEHU(%ESz
zRey$D<w|bJ8T*hruUwO;5?lz-B?pZd-|MS={b7*4QqW1$2rfJ7O<-zmW>;<{uE{9g
z{@e|@+VY(%^QW^ns5<p4#Dl=FfuYF!$SfC1{*hZ3Q8)8>^g9zQ)@HPgP5Qg$f;2;h
z98_u7+o%&w7Pc93lkIhnlocBMeMi~ug|3xe+nYYftC8bJ8vD$?bB*X^CRJ`ndOOVo
zP=NI&FUgSx^2d@Ta9=Nb1~D@1VO4%<z0Rv9@efPsSqeJ6&%o)UTFyWraFb~6{o4k6
zw5Vp~9*9Fg+iKS<LBLgIrC)Mu3}EGtf<{;$*^!YnLlEABygy<Lh^#kCvp<p~HtX+0
zkhl6a*!!ub?^a5Dmt@4f9#(=fqO-e$I@fze0u{4Rzrl&tUG*bObzeO#nt<`T`<-i{
z$9cT3Lp|`CO%yHaz0Tx4LSPb$UQ~l5Z8SLLx8gO-e=?1i6ezkf)qbU7A;ZlrFNj1x
zObDzJA7n<dnyO>L6V>4U;_{+v&r?$}L!N(_rUx=TG1FkZID017^^ezN<h)<kuwatk
z%F)HPZv8^AiA&X&+}rrduk&hXI@m@{G-A~>gWQkNXCq4sZ)-MQe-7vjcYsQwxtm1S
zIc||JYSLQ+$cf$uQN6Dy5Y?Q)$vLZbs8lv(Az|I`?FZ+CZ$|@bhraIr^GZVbxoG&G
zcYl}vEua7Y_Hf<eW!hpk9f$9AHUxlV!?hqfSKvBaSL(xX{62yl47DU%sGHU#nHZ&A
zFU!Gt5-C2R*?z+-d@3hq?VWg~uP`$IaNsx?%<bylEz|pNpTbWz)EXaUx=vk)eAGVx
zslk>`(T4~FD61FprcE}iUeLBp6TKrUTQ(oEiqIKB^6vz%-#f-W$>@Hf`QGT1;OD-G
zx;0RX^WEh5{7-blASs4%0?vMCZ%UBISl{WwQ^)dpUF72k1)iUaeN_R~lL-sF&v}an
z#v4&M=T+M?X!1ImPN~ospDoK9)}Lvzli9?*^`1t@mLK9v^nc{Kh4(tz-DY$OkfES;
zVg2PN9nnqN(^vh?wx}oG7S)uIjB)^AFZ5PP<WO4mCa&Jkt!eqtPsGIMBUkKOmdu=g
zNe0o)mB_|p&*T~%_y5(P=^t#{-%So*fuH{^<6Lb4ptmQ_yRyffqQBHvIJz3)m*c(|
z8a2kZK??T;T`0_a|7C0j_h@1gD|{ynYJ9DB3C07fC`Oq}%eAc@8jfGeYHj#heRZH#
z!h7~^^>;2U#gF9Yqo36DDJy|Jf=v5BmTYZ!Dp>}hzTt91p!lhpG+$hJ(VWNem*oK>
z<E@jhdNZbmzs9uw%u}V{uiX+|CGI=0lgqB@b?er{*U@+5UFL!l_Wf|w>9c?`Plfz6
z^vR0La_ewQjEc|ds)D71k7EfZPsD#y=hO|(Dr@^Y*JjDCWD9>tcH8ZD$@2cnfA;T+
ze+%bt%lW@EI>gobzH|LGgDA64|6{V1{AXt-)z0?jvgHeq3$P_)2$VG)70ZhVR3XmB
Q0Ws^lk$V3+;r%}LU+!j5YybcN

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/QRwxHGpWaUA-205.jpg b/assets/images/impact/talks/QRwxHGpWaUA-205.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..5061e3980cdc13b1002329017ffadd15cb3854a3
GIT binary patch
literal 57444
zcmdSBWmsIxwkX<IummT#y9N(5fySY6cMa|mT!Tw+hv4o6w*bL45IlHrclX=L%F5dJ
zowN6M&wD@Kd)<7qde*EOW7M#!F=`f1-=BT}&?LpcVgM*8000W|1w73IgaOZBVc}q5
zpTWVwJ$v>H9uWl*5di@a<HZXk6l{!_FR?MOuyF9n2yt*p@UXCmXoyJ2DX6HZUJ}wV
z(o!;zQBqO<1Of%=iim)Sj);g(iHn6x`9J<VeF2~&K;yxDfq^0iK%+y!phG=%1BiZ-
z2?zC)-oHQ4Fi`N%;9wCEAdijF08lV6P%zM_h{%Y~;GmwtLqP*zVByf8VUWSIVhWC7
zD<r&Bi-}jY$C?<A!!aONR#K_*<#1qAjD1(bm7U{=K%vgrA-u+3E#ySm`MUP4h~Em8
zp(w7AbMET#2{nzeiGKhD4L*c`U!T8efN&2Q76$Ga<QXeE6aX4R1Plx;;2F#>I-#J^
zVH7aPuvlR+1%27!$k`0+W2&ABz0K|zTRDE313ZT`gET=0fB;{BHR^htbxW3Yw=0~+
z(j}DEwEJQpIK*{Hadm!n_I)pe@aMyw(!*bdhfxkk4jG>he?>VQ?r&jddwvwub!JU%
z-xWuCxIa9;zkj$&IXnWx&Rxu2KhdTwnzA_i1x4N71^ziROfm{tKTEO`<~PT#RH<^I
zEC&>*VOILkcPP=!S&YTOEIWRi$Vj2gf)jAVgrA}vQ1Be!rlg5Rel)_9H6^ed{viCc
z2X5KuV@k?)9;(-=-mN%DvZ_}nV4igzLXK;iXO#Ld#?XQ>>r6GJe%kVK$N}hULe(9+
z-BQD)qjHPV(~Q#HDo|79V(<RU<^%8pcWJ`FtTbsM#(E)+!Aoc6eO*O)<`L1Jn73@J
zoBLu{7|KI6u}#_%CdL@6+M^V{Mxs4}!;QrOs1&~M)X3}@hkd8SDe{Pv9oRanA=hRz
zlxmoAp&T8NVAzPm65&Zs#G1x+7Y}oTre(%q=8<6nYBf_UcB8@l{jYHhH8ot)Wo|m@
z^Uui(!Lrr<d0DHHVd>b-yN;@7j71DN52`YP&J(Y34^=G#Md6IZ=@PN`Gs(8mj|hV|
zm2wJG-s*Chc)wyzNd~7R<E8j6&Mx-&rle%2Bom$P4^t11BC)e0*VMGo4v!$$u>1XJ
zX>9RZWgZ@8Ms5@Sz!RvG9?TFof+kf&2ovoBt>2W_Y%mNg@g3%dC;du|ob1?@Y#@(A
zs*SOUE5|I*$b%w=In-zm&wN-S(6rIf?*J_JPx2NA*sY@om&R9*;8|4{CJsevF_+vp
z{}c?)5cfv)zCfGKN9K{(rk~Pr;6=lGJ3_hMOjw@CJmPPho~|&@sFRU}m)uPB(i!D2
zXJl~Qfte;I0K0$xutMA<AP$Auh9;pq9+jaAMbg<s!I@dznJiSGv56t!dkt;I2j6#O
z-HjoQF-`p~FR?!Oq7*MDN!PH!tH8V~ATl~->nwpOixdY#1oCN;K6?r!706too1H%o
z*U#Rh^67ARRwYEAXg1o!R4^r3J=m1yEn6q^j~4bUse(c!6eKq+<?ZqrrJM$;6nUkb
zwhtABqPA?qpBabmb%ohk;_Ah<U6$Lrqe~Z?=mGvJ2&Ccu$6XJ;-F$L+*Y+}U8qU??
zj7~}=SV$HU@q#eCXA$3ha85Im8giy4#tPz>W=6<qh;i#q#;&_aPtkOi_mD;IU#%Ix
zP$7KTsGx$umWS`yp@cTFkqzF!yBcjH*;*9t@<>k#(J2rSnZlF}*)iq~h^ahMb)RqY
zI@UfX`B+%gTI*yEbSX1=J8z?Uv%@4sQ9(cvNA_5e=pXzUc4Bf|bzEFU=XD20vJ#Vw
z;+Ja?QwK%<EH)mgI)5Fzy>OwNAq}%IyqoH5Y%2+&_>8dJm^T47+x{D#?2q|%rNBAe
zjOp>HE5|1QA^V|cx1+bBX{|kjm~&_;y)~=c7FGaavB?~41(ky-5t<-;{{tb_xMS*!
zu9nrshOSv_lZm7((s)3E)mC<z!NcK|RKsD8j~M&;>E#CQkX-*>kemBi--p1QJtf<T
zRQE31I>o^v3O`AfIyyRD5W?X`*P3OyT%HqA{Y+6=ZOp8bv=Zu@96!uquzw6G_w0zk
z;ut(Ux96O0%cQ2|q1mwN5;Zjq3F{y^(8tSeSku|CtY&aQu3^mb1&!q@Uup&dSEpIU
z<Tp}EVeSQN-+ok@==>XQ&O$Shs~6Tq6oS>}*TXGr&qH?+t`T5zBl?IO5~Btt>g!ue
zLo30Rql{}L8oi*<!Lyx2F8F1u$i}13O<g4(y7{G3_ac*t-%AzD!>7b;ga9Gs0{am$
zEr<Z9Z`(;*QP%;jC0XJcg|+LKopvQ;Vu0`WXq7bLsF-^f<sRnSI}WL9NxLE4!=~ut
z??5Z?bSG#%ZZPpdb42#bZp0^L982-jUZ2eHEi}V42Zd71EO)?>iK1xA@&4ZySR-ZD
z%1%iTU1QWLD%9L9(?7`dhVgnPD4bz|!3^L9ly^-x?wJFYX4_Zjxo_UG^CX8xg&Sf%
z8ZEtL58l$ro*}wo^c@tqT@tBW%irCx&WO}{nHqu3Qo_khkCs45$4yCi$h$O@{i;yn
zm~F*7zhT~@l)UAPjx&B$usZ9#*Arm+s(kO_+;R0bRmbElIgiK|Xax9*Cu9Uy&CDC^
zsAb(@-C+Ib@f*&>6W|+1gJGY{SuNb~t~Z;e4QFj_C}-WBgCLfG2tU*UhC4xKo_o3b
z_|^3I8g^*rB4(OXUx=>K*$V_THw4DB$`eGRh6QWo+pq~R*0UKs&YU_{Qc7qPMG2V7
zNB!g0*eSltBT<WdywqwVfhgCaHIUk?$tC}pnJIT4A41Pfnl+O9UB&y{SH$Af!VV7Y
z<DNSCRUMrl%3l;jrt^uF=QrTRVF<<N#anxlvOkuQ_`DMRhLxC3(q$QpX33;H;}MmC
zgil){HYuOpWfj>5zPbXiCiP|C6iuC<m1CuyuAwIn53i9-zsp=no*XuPE)z3@va<2P
zmYAk#Iwwl)JEIFSN8ei`Q8Bc}Sf9!CE$%(<hA1a1E^bb23ywwQIxfG~663+x1S(2X
zqb={Xn2Zkx=a0FXGnW^&*XJW|-;qq)<7TdrOenCYN%Kgrj@IR+%rma$WvH?2M5?j$
z7o{KSb5jiii{V**5c)IHL$4D;Ilu)9*li-`Aty7Iq&UDqd^ZQZep9eC4ykUyVu$@u
zqZK11QCEn|^J0U=K1QjMHl1q7oZ8~c*urZ(&)iLaIBV5nyo2LxYSqjv1(|a9rI3?M
z-;R`MI1km-k2!*8=c)WfsKp?Sw((NFYeVLtbR2u;!=3;sco$PDbk!{!PHr~FgJ5!^
zPiqXkI4&k4$RAQrxQ#@+n^wAmVUa`MF*AmlCiG}mWaHV%8;Jn>*B#>q_b7RxtBH(&
z5-82WkQr5MOM$G1+0f1@KcxZ3lg~&a{rJ1p2X=NCc9s@iTj!xB3{%2-E%JSx6o2NP
zln$^>><y7qGXp_dd{Vki*rIVWgME^3N~V9mi3t%h{P{t{Rih7g(k;pLrw5X1pIWLX
z0LA@@0jVBScQl|R*zK;i{ktdl+cZCwqLRXb&2GLZmC&Y8@kSIqr6IieqxjilPDbX;
zt&X+aysk#K@kTLst<+i4@qI~EkCs5e{48CvT%@e+N7Z*nrlxyc!}uv2h1b#J9v*MD
zl-7Bl0LUgfNk^Z)b31*x1(wDh_NyoteOI0sPNC6sbEXh--zs~P|L8NRE#0W=8qJx9
zuxV77j+`GTenc*|o^>Pk?x?8Ey{DaN&vuuCfmLFn?yU)A5s>*%cvx@H`~<)%Ff2O|
z^w92acmmK&k8+onc^3w;N8EHh7;6y+Jd$#3(5aXemiezx(Z$de=V{!nV`a91j|icO
zh+cOdAa{az+H`EHCi{~IMH*g&i)BQ1!qZ35h+KR;c>-iiob$)FC8^0S9RTh7GFl!a
zm&4zBw$mk8JHESUD@dAjx|Mto_oy!>FNyImN5+#3Vo?-V_Y>Z;f2}T9US};c5|+SG
zO9zdl?gxz!!G{4a5(_?UwA-wd9^VbYEG<${F-u1Z5~VGAhr~S__ap6)0)C0=vBZh`
zG10cPa1%H)e0_31G{Z;GYE!(z)s{T9i_AAFvuMJTTCn8SubMlUllcUY^{mvJNn?(Z
z(b&0Q^1Fb$lbtLROD*ahT-8{Y?3)TdTyA`Shl9GZCqJr2Ip@$k=3LvKQ$!!99)Axr
zPz(Iwlv%0{wpm$5#?BI*PTcZnRHd%?q*!VLR1POfuW0XpsT|bn7@HZNXiN8cRyNO$
z&XG34aJI>=?ZbZ0iiv(_-2J0Bzaa0k@{UxmQMCp4{fLo6$Odn&`kw6~Wc|jUguA!n
zm|nlw#lR7$JpF~98PjS%H8tURL?tQDZJ)+=DxM&4x-%awni&lP{DtH>pMPX5xh{j=
zbRd;Eo48+Gv`?&@=%vN2#pwDD2veiBuaACMmF6xa<q1%pn*3mQY_}Cs_nEIrSq7U?
zWfD8)j5FThP`dy09ARKDVjZV}9tGJFu?AaoS$s;``@(#SE;e@=;HB?If*Qnu&#R!g
zaDJGh%ypNp|7_kVMuZyMw-w$Mmz(f&69rG#4$ZaGw2;pRO?8iW`bAJ3HA`ma`q0@$
z(s|4-6EG>A;e0R6Au1~LgJEBI;#x*4{wSKOhuzA-2);$}N?BfMR7PHT@DYu;xQjq|
zc$6pK=(WS+$H~lnjkcs8YhrmdlY)hg>5)~h@x6GZHcwUaHrNIN{4m@}txh9mxwk>4
z$9o+1x)*q{BBTTyS=O}A(;{2WE`ZI=?IEof?T)?QhaJLY`Pp&P@KOp@DWi_iuPC<N
z`UE3@70}vZ#VQWg=^EE2At)}E2(!KJcpI+#pyP?3O<II@np+LNctpf5a^f>ketuUh
zBhtxQ9qs3mG<GvAUYIv$Pq;H<Q!}$+duPyrZYNB^zDMXa2$YNK9Ob;yt-gr@n%(A9
z=nq|OzF{!6T<&GNZGXt93+}bzi<SdlpJPWBf=@EOS~L#GjE+xJN)+WIMY$s`RLFif
ziO`d?CbHPx8_7Fq0H@u{-Rp<HwP*CTP304=xK8Q1O~Hs?qD*xcXcJT*M(Aq~C%Law
z_~BL0tFm-eS`oeaf*nlxk^Z^^j*MR+(4}j0w^gK-wGTP{GVS0deNpiq^J=?)4)1Cr
z`l^k4X0~=Ijw>C+hA@Q#r(@xag?j1=hH)XK?>%yDL(JdNbaxk*ai7>>MSuAO!1tLa
zdP`2R(Iew|CAYHl1kh0~FA4j2YZ84inV*%JzemWw`j#WTt-i;tR@7tSVi!3z#uQ1Y
z;YA*1OV(?TE?2K1&#I`E<^vM$$akfz=@Hp!))ZOysWdV>Iv%;}RXG+6VwDB)18_3Z
zfwbjO$0)LF>=9kOzRRuWiX4`OmH1OBg-s_X<Bb{V?zY7E-6GPmhk^XbE7T1|SeFb3
zqcRIL+(8*X(i<hC$b~|m0HJFKI{C~%`UfTBW-90$LbO`K91)@&MX%MXA=~38KvA(}
zPNuw?CKFd<A`YiwDMsac6xsg0!|gU4Jf}dE+UYYjEi9w#0zV+PF$Lka7+Ckwm!A77
zvpsuev^WAJR|-sZO-Z|0<72(8!W-3%OP?#d=$19etud~Cyqh#Jjn@YHPM{dZ1d22T
zO8|Wr=po^h!K9Rd0qJJ!4kWHc42Dd}zQ)RM3qZGba6Qtzeyc}GMVNCga9zxT3yjpr
zZQ8<ZT1aggkl5Ca@*s(@h4<GKT)kVFXg_2e&a6{CyjundZ0DInTTD@SOqCeol1xmV
znHUpQ4{x_rxj4U)f6sh4jGS}Wv-rKb3=*1$1#Q90x9D@n`Z_x}cZKID?yv8QU7|5l
z@<gu==LTbJY9oh)+B-1!1R&VUzjI~`foh!=aQxC$5`-;J<oMz+-e6k5-2xk?<&%q(
zxSdW)K=rIW${}6iPz@io+K};V^5BiLi6bQR{GQ0z`8Cm5UyUNGYH%;J)lqHC0qi!t
zT^TbI4WwVUPvZGC^_mJUo9hKs+|}@quO;!}xac2#K?K8MAZ$mfIh89CmP;xSVH38n
z>*J*aOpt$L2`nc)$H;;Cz|INdP{TvdgZEP5lWKKl#~k+&cg_1RG5u2PIf4rqJcuM(
zpIqExU*ukkw=EK`Z=`%;bg3HHVBTNvEN{mh+3afJf}J9tW~nO{aKy?H{=l9rn3D2V
z4K6*s!-#OA|Hoz$e!PD&Gt{udr(_4XFFf!n3Xmm~;3OpRz!RwkSjSaHIV4+Vj?U2O
z@{fk(DC|}Vece?~XPx`sy0BFy&e@qtQze;N)oVa_(vz%><JlPOVxVsYSw4+HmPo)L
zols{}(bL1#qr_*<5P3lu7LYQ#_ze<yrr$na4{0&(Kjm&5E^}!ySbp9gNtE`X%0UFd
zo(vL%NWyZX?GlSauUGpzGmCI1qC;vMlSJP;$rG{cW(OL<>oCU~OcA7;GuG>*Z0j*4
zn&6n47&tqd;0)DkB(u$$nhG-|nHYFNp8C$FfejqAarzJUE5zZjn~|KRx0zzTn#zrk
z-;*xc<E#XaCcitY*bU#g>DCJ(fYh?gMvOQuDx`zBKOA?t{j6kd8fhuv0rLNB?ZX49
zHF?c5<D}~NdGl4*i!D}fT&vzK;9<$(_Ra855O05j&`2lw*|<Gj^xB|?piH-=q$Yn$
z{t?#}82l52uG6AQ`m3M6-6lc5k^FBBAUNy}?;l-ISAxzOeV)JBgkW{a&deNHxo27W
zSvzd=uHr}-2wg$~A;azw*Q#TrrL-pd=T8@G$i>0r<tv@3*|DF!dEFKuNLW7aHMaoy
z{|3o32Ek$C=y~`Hz<rigm-b$BO{~<}%>sy)FwlsEo`42;cmcgmlB_y$U*a($RR3{9
zX^AOW)@gBvu)^R5&g=DH<cR|gl3t>mYfCa@NP8&e-uy-5_tidH(M^hm_eNJAOkV3b
zw6ta~;H=`%%O%OVHX)~zh&j4^sA7ST@8Gb`ceR2e`6@{q@}L~W`&;CRRp(j7ks0$*
zH}fP27XtS}Gk6X02**uqt4U|+r2i0bwftVEi1jkwB?`Qtv(3*IdE&y_qEIZea<q8V
z;<6lE%0YrB?t1>QlmqmuWNp5SUm~9NW-gPzH|AR#xB0tNXgXALsbNZ5-EBqfRa)Ys
z{<|Eu@km%a7_$A}yAZ7&8opbA&}P++qq6M{!MpSN?~*=a!^Oj4|K3YI`?}MkFK<yj
z8+09T6$M^WXokSre-k-6Fy&4Xp8MHdc2PC^dXTqgNoCP%@bHrT#7XnR8toBr+~1``
z#3+PgW%pjc+CgCbE*EltVmAbls1*H1{=1}IxMFCTKD^ga%^saYUt)%o5CnaBNA!mf
z+<T_$yLG6&3QNqZM}j8-5Uwpb{qB&ip7pEY;_q^ywCPaErF`-Fwo35iCx4rNmrZK#
z77C7Rh77K6YXqZee~m)pufAK{KY@K0jwDGc#rli!CFj4&v<-s1rIuFF*&FxDzrcN8
z-1cVq$u`=>lvWq_lpA2B@Lza-mrY_2Q#y}c6+Zh|hyDcUrh%}FdQbG^CnuZ4c1BuS
z;d*Yqj)yz}Rxy{D7nO2u9Irk&{Iu9@vhE<s33BLf4hgpK75n+wB;HwX%9=zPV|!iO
z6!@+%IrM@u?o2S_ZK9nNSD<Bt!VI1Xy<Xv6bR*XRDi%i<jmtvGcQyIP+oWoDWzieJ
z6YlaJ=9omdeb-d>T0NrGRcI~+N)`vgBgK$jggWI&7KVc{B)_>~#Y8V;Z4g9`d{)MO
z7fW7{V&t%&l?H4TZz7gN$~4i6C(Lz;SiJ2?-SoNjlB;2U5NY;YRM{W;4C(Y)cI6Hj
z>;!t?s+`5KH)y_}SKI&K03qSH=IR3a;LO^znrRZb;cG%he*-O#*wHhsSIcu3KTfq6
z_+p$7oKMb-O{<wtBW55SjH@3WDwg;y%P&P+bUu5|Xh8r)j3~5Z$X;3J-13z#dH;O$
zEp^PUP`7(h1-%rnGnaG)!fyJ*(*qn+y<z%Crq`=4&yd<b$Xo|P>S-4RL!ItBD_3j0
z$(%W`GqFFOAgATNi?=yt`lhu_zo_X(M{_T<T{5HXbL>PCqBCrhmg}4`Ji6H?KVnBo
zZ`OeaCxCv@o~@h)zY-t+sa%u!n{v@Ocyp8)zO8F34wU2lBaP*Rer$hKrnQ6HU{xh|
zHsd2|diZ1Lou6G-cd2*{AJ-M^N^q!Zl!?0)zcc$S%^_kJDn>c+@cH}#W)LxzCSS_V
zr<?Di+nJ3t&)-PadazH}+R5U&CYW8C7IbYH{=bOA&8j^o<G&r>|D7igo_PGXJgK>*
z`ooi@m<mXq64sHbH*!%c=OAu!<dcaAYpi(=XixR=)J<Zk>x2T7U7LowBICrNqTI|s
z#DPi5%!B5qn3w*;KS}FfS^C{a<l)QpZ<#dw;jY11UFN}8`@%-5Y-VMy-3J`2RQEs_
zHiPgu8Js~{)5?AxiGgjt&L}@Zx<<tw6{$SNxDGa<wO9^}guGKUhhDumvk#bmWNQ6~
z^zWYgrsr~h2>4foRheU;KQdkV!_nZ8UI>YQ)aD=VknrOD{|QCw{$tob#Q$LpMBQzz
zkN%BbYk~faYJHsd_%}8(+uG3nZ|p6i_22PpX@2?N@c!$;|F=8s|BDZKJv@2;Vx^Aw
zWAkK3t~KjdY{DU-Zc{q>2od!5imt6Yc9vPDCxCwNp$>8(rQzVi%az9!_=RR1S&7Gn
zDckRdlXrUJzUlt_YRC=wvJ6xd$Ds0aPNA@aoi{BQ{<k5X5A>j=jk~Blub~K<3?(U4
zQShMGDi5wXcw^{X5V*6Bk(<D`x+9LKFuTrN7LUyukU)W`E5L)huzmS$cK20+BwAK_
z+5QvYZVy<?=`gMo;*yt_7FjhkYxNqz>Q=L$j0!PX!XP9=g~7<MHCDGI-un3#O^*9v
z#1kNRD7Hv-_;Y;WTh@&w-93B~9ML&!%c|~_XH>~&Lqt}ms6__*w8F?#D9xC`M)^~c
z+Bi^?P=4B9^r04b-ygytQrb8`eTOH-CF*KKxi-)zPA`d-`ThtdnZBYa+O_{!b)O0<
z@93IcnrE&bg8lSfZ1cO5kWMEuk^%D_b3s>I*K_FMVH%moH$Kk%Nf*Ta`^ry%Sn+Qp
zL#ej4;#cOkdXpBYRZ0+e0tPz2w(9=FMZO|3$AWGxFP-B8IRYG}LSZFVW}_e{4xjY|
zz_a|yQ^c{9Qyi=iYE+hPq6bu<myw`38n1|o%kF;ySnyL<j7_Ion(Ndb$c>13V>!?`
ztmv(fo}*5TxtFbK)Ed9%rs&Dd-&H8Ad>3by?&@buQq7ooN*twrwAVgCC$A_JD+upO
zkptQ0u}2QNvLVrK3hQ?3d7c+$(M|J164-aFGDEZ#dyU=G>i;aZebQ#(gO_x0TscP6
zjoj3<^H|2Ggfz!QFLymG?ilf|dAg(@a_@@ZIO4k>PPWT+jG|`xZ{4>v=WD5gw8}d~
z#g9mRX;wyQ1SxJu1wj3&eh)4)z#z%nHiwetas%`p4Dc-8HETAX!SwAyxkG&#&lPXF
z_3j#9XPX<n#qLWr$}k>!nkT>#z^8y4v4B%7<CA3R!X`d_ICJ7M>EO!9X|>NpX2<On
z2o>R)3?&_HnUhjs)M=s7JgQTarp+BmZ<h$AWmtC3u$#(o-E^Z-ht^vD><VelLZThT
z4H|Qq>c-kVr_Vz}wED!hOnpNdGvDF}CMjr9%v9Dr#Qd#t%@CnNp_}F;M$86PpqOWv
zj$Waq#hba*2g?krwU3d-NDBPYZ7ncBc2y;%B5I3}w+eVrmWbT93#+L--*B)(V&G%I
z@}v(20kdgF320drqWOq~B??~*sO`ecwqzF+$wjvCBz;MW7cHgD7&QG9V`9NypaN1R
zf^zT!5`-BY@(3{k;ZP@!7{|W&Q&y96pr|kjpxhX6ISWmY3q@z;#Y7AClik?hcsS)l
zO9gr{;iHllCeXsSsk_hKY};_SaJ%i|H@g?Lqu@bLEW=Hv=HNR9$}rkfra{-ihCf)*
zv!7n5`mas4P4T_f#80+RRCzu&2-QI=ugW`2&wfUbjzEb&s4up5U*c|4>(IyMoU5rC
zHwNxE;4;zwnV0zY7taX2TN?L=&fGH#FPHwQCqRx%<pMp3tAK=6iq(XGv(~3k@QOKM
z0JL%=Rc?I;OX9Gj-ClE=G__fAAXj(p3Tx0$#thjTVMWbMvXm0T)*=`G;iEN=gG3sr
zSQd~^L7DO1-P*5>z)5j4E!r<4y^D8Q^{Cjqr2Mo)GM^sIf}$t}FM6X1EY(6l(oi&t
zprMIxS5Am4V8xSW%d&+%-MBV4(2?GW;6_sDy;5X@{=x^3?KH%vAT5=E5J4Foeh7Y3
zNaQtYGgj}RtU_MWwj=lIbt{@xG~!-b)a0d>NzOR!*s<1nlkS@vmiI*)h3}Ql(i$bJ
z8mdvXw_MsT!gk2Q%qCw(r#u@+Ao1keq5v_lk~}Ylj-xc!#KKB=?gxDJ?fnhDqUcw{
ze9hZ#n!|TZm$Ae#j=?HW9b{7yb`tO`MDO28=tK2C>q^9hiFS5DL2hZ`{lv$vBE#qs
z2rEA=oncLaLR%j<FyrxYj5E+lh3~*DC`WyqLcBxr6AxkSiX~o>|HwLF{d~yjM3TLx
zI|mNb6X1T=+on{TpXc=&FY?p?<rARo>$11^n)m#72A{hGAOBZ*o2lg$lE9;9df*%K
zNgU-<wPF5-qk<`%TEn+%EHDG(=97BX2B0T^W5uxsP}%~6$9%UhF;zD1BHgK&>~22I
zpX9PI?~ps9_@r&KW+Q@+`ik;$<z_Y|vvdHc_uEyv|8|u=%yiykB#VVRJ__6XJ#sgc
zUQmI~|3RsKs~j=GRP1t*b&ZydW7U-62$vdu-yYFhAS`I~_2|mxhG4ND9fQEcI<wOZ
z?Gu1b2Hzz<K4Mpu2VudT+pSXf1Zk}6PA`1{XJk?Wp)fBz(!tnA(S3It^woRYeL}^u
zgg2k?6zrg!HpN6KkR0njL_8`=>2>nvbEQ+L$#-jDYUmrYTv6xP#2qCSqe!62nD$Gq
zf>5nEHkCAMLQlriwIw&+z%<gVZ;m?^W$4p-l(IPdY{^g?(mvA_N=NV!*WIaiFH#dT
zI)tqoG)fYI`BYhlDi&^rnnA}H-se8|B&GqLPT150&&rDsKRP;IVW+`e#L;aNS2yz!
zw>b5Q>?ArlGFYrOjE!pWbjynstGW+JP(GkE@_5QvW5sRnt;=tv>A<dVZ;W)fTHi45
zIklarsprzGh^lxlEHWageF>zAp~wOd44fuT<5S;sY4tFwjEfSUQe+ku=S8pCcXWIq
z^?#7nrfV5jZ}ffJsM=3n5NVxIE)E4{MhAQMrO5a(aJ3z2mphKL+QPC7ODF<^@}OI-
zp;pm>MO@={crAC3g1H@u)2g{$=-a{`B~odW1yaJPu^Bi0Z7=Hd6M)<PmF~iIV6kxo
z#&D@SHOrVn-aQ6ay-*0mGu{gLWS#4V6|F42Z_>=Kh^&^o3iD+&qlGai2twpDy=e$M
zDX<~isqKgMtq6~HU-KE?uun-WU)h~ZBcE!P9sc-ra2bhCDblWsRx@jq`a9vrD<k6#
z(8`S7%;F2#<fcjHtcjk{aWHv(JQ@O$AWb~P66cphRtGWK2rSUDV>YGu_CcfVU8j>7
zH1c_ba7vRfM-kyq03}nF462z&Y7CUd?0)zsfUWq}%hLux?<n$tzRmT!M<+R~CjiD1
zpx_*|f9{iR%W#bOs5aMj=rdA0@bIZ>BRYfWk!(-Ioh^0_J8+=_O#rY#N|bvJAD=<8
zhPyk8#zEvyW2|D5uaELx(d8;u*KkNN>Uf8Io9==uBR;y+se6Sc3q{dK6Qs558qc1^
ziF(Ixdzut={>E>puUyMY72{)oNj}V*!XK57oNs))GrzqMz6)61VB%MN!}{uY#y)n@
zzEOQs9(_i!&;GsBu_SiXR3^nDD8xPkcRK3P6*NN3Iv$(Oi}Qk0Ql$nfR#|g5O4w+)
z%yd`b=uP^)&=X*YetC%$B0Kj?I${FqgmTfEV6NubXNCo}eQd#>lYJixL$9r~^;b_C
zEr|@7L{0hRi;8z9$b`_Iu@8KPiUtt9Jth3uNLup*fb8ydLADuDvrus7k9karXqF%4
zo0@B_1>Z*by%G4dP{uw=l(9v_swCNjh&u-*%!eKb_xuQ!FX=j^&&)9eucsGu%5<PD
zw(|LnkZ2XVwsH2ZXRp3t$SYg!1qU%6;^KR$!*3j!2MAAq{`$y9%6cD^X7Y+fDFxD|
zgCV4sWy*&G5;PC|cT0}4zBSuO^kKZO)6P^dLe)kG>YYD5JLt+H(d>D6A$V7-w~o57
zKG+R<eOwxN>!AlODtU!i!lt-@L+IJv0s~u=cDVH5-X%R_N55k;scEsHK~-Sh&52cq
zJV%%^%l$>3lmQ+OmhZzgBveIK^{GR>mz%wpET#yy_YY6KuJ^&V)#El@V=R;b(ig90
zu*XZhuf0L&cOn7FQAvE#Qp^sdPDJx;fjTE9fTZ95!*Y^@LhUhhsok#VelAnzl+$NB
zeeVh>>T7ZEu;Q>r(3myJA)41vz5_w7Q+cMo|Dg^+^diA-=mYbwdDfr3(am#9O^%GA
zIGPSo{rfrvAI@Y^LvNi;C<UMXy*n6Ma#jpf^Vm?ADj={x5^>+~!h#ozN+6`lf9BP!
zOaWPuI`o_8r?y^=J7V9iUzLKTDF`ExxE6^lL?q}#QZOI0>>jqn=4f~o1?tfIokz9~
zvt<!+H&C+zKBZHlPY4pSQc@uh(7|MpWWi%S2l{5|FjwnvoHmCV&~P;RHub~9`a`xR
z43Qxj`E^`v-WGUg98}^xovT@|?X_ECLDxGCAKB3x?kB+D{l%Ntg&RF~+c@N_Jjlnv
z)hFS>yAcL3Zr-k9%Fo#L-ih0qMHLTmBE=ejkx={^Fo>)+kZ{+Rwi)k9r>KHvQ!aTh
z^Y27_oHO$u#Ix?OLF+N+@TZ%Z2k2hk|3}Mv;^6KjMlEh134O61omJ&cd;u4lvL@ib
zgK74RN4cX4rot!@lR!a{wuSUNEbEX-u??#XZDRj>gNib>15mNWP91)JVygvPNwx(&
zHsF+remg2RX16NQn%nj0^-z7q=ZdzqH+6MW42TjCi+4ba=?4Pf<I9@8<AN_G!;CBl
zh+4GN#4yRq)3WAQ<Od`XVBZB8(qx}^iJ6D-<9(;Qs4opFpfBkOp+zU8i&54BQ_AMi
z9;reUpYN({%FNR1w!)&t7R~B4oOCYv*IrS@ksR66tWSdcLRajGL=Sk@QIb+BV2>$H
z$p$&?xGFvH6bzY~rpO9qV5IiB#|TGe*M-;gVUyY6d3ssDFCX8x=o&`JC4ksb0Fd&B
z7fL%yqkZ84&J*CxG)^n4iWH$+VGwWZ%LJ*Ih}y&cKH<1m^t&3&9$=9HX-8bb{=z2!
zLSgo2kI~IXrBkBB_}w0p<(a68kxZ$bhr8nuEN(BsGRFXgxe6z&xv=3(e>h`~_8}#R
z8&zoqpJIr&5sQ)1Q9)kdYxM8erOnk*q#`yc4oEkv$(+0_(M;FCW}aIsq$9%5hs`W&
zkJ<K_)EQoNw9dTlHGBe~Ec*0$=)NM+jT*hzS>dw@ZIW@w4pjPH@=`ezgMG6*^Ru41
zQxqBT-IZNK#MG^#g!SvG0WFh-fPQbRTK}bcN|XlS>}j`N$#hYNFi8zMt0~t0*o1Cl
zWF+tgtfTP#PLDUSve}0`Y<^oJKI|{$3gY;e6Bt2k@7?D+NNzzbI}b*dS^?u@54{xk
z3&2!NiH$*;+oWB~a%Jo}lj?$0Wf|B>2S`@SAt5g-T5u24zd-*4NN%>h`u^y5ngoAz
z8`K_*!|Mg4D=cUTeT$Ufswkg_t?$uAgNkXlp+Y;Oimi~%P=sx^IbQQ##eeX@ymb1c
z*GKMT>vnO=;=F7V2#M57FF-Fr8*SMS13r^a0KQ|GM>DM_0GYx~_30J*u9L+%=dtRT
zQlGRr&Ilbn_{-783Nb44j|exhQX^w0?csSk+z~DmHy6+g5{*&G+cYNJ>*Vgy^NY5C
z>5EF*v?l;)a}E@0JK;`p=?)rcNVrCP0x%rtxq!BmkD~~Ung>eEUo27efl~4)84Hh`
zr>3t?uVitib`dz35ONC}E0Sz<0)=BU+j-WnmZOU9qZ)QBEAgGv?mhc5U&*`zb4{l+
zD@<(h;hn4uro1@1j>-Svo?ELmu-K3NNp6DF>w%SXv*boLb<iW+E+U^!$r|?5X2W%4
zLaMCpH72LinwaXC;=l``;`P*ASS)LO*bSQX#e3WGOf@N8bIv~5;@t;Om087GpMtNY
z7*=}STpt@L_j5J?fg+BwIK-gA(^nXvjmPYVPRLyC#~mZwO$5?1Xy3iPrSFhZjgf8k
zg!V(wfJY?Eeg!;1QI_dyIWOXoJQy#gsX$_T47lgX4k4}nIC?xoSOv57a%a6PJW`>C
zZtfx0V^n`NaNmL-T1m&bK0z^6wCoA6TnpW<{;IMAiT2Qw+%L!=xkN^qf%r5r%R4Dt
zMTL<NeKOT1FODzvFhtfJQTb!Lx>C_FhR*;TWI}d*G`R5zUcS+@xvvx7QhspwryBH6
z^;UP8n~@7-Ga@we8@=3RfX}+tP!wd_0+S?5x&7B{+rY44+LdlQqpcI>m&(mSh#{05
z9skOwQ$nKS?y+V-6Xl?&@};m+@qt9$Oz5O9v>VbhmI|ez)(b0rr<eXNkNB@n;SbGY
z$!;nG7ld`gk`x3Yc%53@dO94e$8ycY{j&XA67HDYQ*}A^cvUkS^MMZYQnZ<0Ev7{-
zB=6B!p4msk3sWBPW_-N|`UIAO=7Jw-Y*!_Na2ZuS&dtA9<W>ry)4MPj`IVU_*$a^E
zde$5dOGwH`$G^vT26^-UaC=PJg7!ypI39)M=roj5#{Z*fcgFHDs_-sps|ADSv(wP|
z+!G+T_;KT=lAx)Q&Wq&0#lovHHv`eSnt=bTYWbUM(x_D`m6p?K)Z?o?RF_DMobz%R
z5}tQyJ}0IZ!7<r|h)_49i7|!wFm43PbqFw7s{t#M5<?U-;=~v#k!lB$cH5ol$>V%m
z8jK~7_v{fNfBD1QRQ)U^8R9HE`NM2^M2r3`(Jm-Z{Vb_&Itwr@$e%sR+6ztZ0(=tI
zaiW}(&exaiqE6fS^j?_X*ehLCz^*V09NK|AF4_^2Y9=J?ve&)eR|dhHPNcb{+X?W>
zv2Qd=zsSm-K^@HQf1p!6Ick5?63KU^pGaV0`h88g@J>YUXQB-MUlL_LwW3#skWjDa
z|2Wk1Y?)4~v7n<}Z=U~As)N<(`rQBMC4bBj(Y*gI`*Zs~O$H|C*h0Uncd4@&rRZOk
zb1><Zs0XX+W7<vemR^5;mVvEPDxj0GqyJgRC2&HU0^HXYxL-u#8hp$9K@-1J&)lX^
zLAeJ`nVmX^-;pR30oE#(hdo2>CiSfAg`r`LP>rk_R^gpG#a6V>8O^Kl*q-ctIrs~k
zGb$ojI>M5~6=+5N*20)ZEoar3@`4GMas%t;4)xjzVe`Cam8qJMYa@q}7(i0vm1ZxR
zB8EJ)4*h27YWr&1^_@5Aog+xQA$6Sf7cK=an<eq$6slOCX&}i835nCh*R4{s(&@&H
z-Lj=dW<(HR9rt0l_sbHO_sls>9?!^&;|_kFtaQ<!q<ze~ypqkypf92<%-|TnDyG%M
zdQC^E?|%WppRNm6Wkm(2WWE|qgKi$5=l&qK6<2_zJTU}~Ri}@ghQ@yMqjml|qeWpz
z%BWtP!?HxWTA2w^o(f?W5%!{5Zr4*r@194mr~*T(Kn2bVX;RuAOd6>!E=a>053k2_
zmB~MJ|HhIx-9pJxhO_>8L{naxWSP%g{1(li@YPP*=i9XxGuO`bGR*$%l}7v1vW>Y{
z{U6{jk^&ud2AO!u4sei$3+U-5>8<C-p*o|c+CM)5-kL}7i017pCtA8o6y}Td)8B1m
zDCQo;f?aJx@1ZK2bm?t1HI#R;rPySI$(yPZG86+H^wY1oHyoIzRSk2URv*i1An(v;
zMQ(uE_Odq3CkTrb;)@~$P^sA*j+c)QgsIff*U}NGuxjHZYNqF)C@zvl)Sw&RC$H<t
zPlSxC_$uJh5CPaf-;D&n;JoMxYPe~?JX69Jl*U#b6dRs}w*zk=!JZQJEQpA7Q@hvu
ztm>UF`9EZ2jxepGAX+T)S*(1mPi(uf$}bYyoC~vvCo>-oqaaN?<+sJ*)+uTX_R4OV
zGb{*$5(=3NYm#=L(vnm~0GwK*mT$K|q&Zb>&v{bbRyZbiOw+JD;HN8@mN(3(%-h7P
z4CIC?_L~<-f5Zh-aPu}H_bzwokdLIa=;SRm-(J}61>~7t6g6&pzqDXt=z37Hl+W3%
zo1*4MqUMF-CE%TkSoSPzIc|}h++M^MG%ZI*YP_6z#ZjJjKqqp@NA%tGct*p{Cq&Au
za-~i?%UC(Chnvqfu-Zp4UGqco=VNNk9}dP<>A6n;yC|c_Pr4K3`-&DiRL-@+Rii2s
z%-okFm<@wznY5@%6YkV^f}KdMG@5jQJ<1noJvGeZEY*?@Wi}gouRHH6?;%N}Ap)Nb
z)B6?InsCR~^4(SyNt7PW+c5lYn0{lCP#WS#p_U^~yG*GC4)3tMnnhUh5;`iyru>6K
zBn=90V9(xy^-5ruUYAu-&CRFsJjHKq-ZJltCfSGl`?GI{ns(!eh>G-8)Skg2-0g7O
zNSvObM%?qvh^utD>VpX!r<BC1(${FLyz}|?WVER2%|BRIRM9Hep2u|1f(O9}fE(fN
zvEx0x<GC9xe&58%<|)nA8nAK}67Y+kn0*=?%)lVwBkIQfBe_jNt-X1PrI)n~$JsP3
zI&7GrdZ!o)%EK;7<q18KBOERbrCKlQDpPMqSGryaXCQIWD(rd4%TsJAL0O@=S`;z+
z=x26rU`qHZ`;VfFP9uqa^JZvxW2&JEqu@_mLdq!jM0blLEL?Xa=bZ3|V?NO!k1plQ
zC7@ylA-dXk5qnZH(f~pL3uyexo209PJg9b0GZin7LHr|<d^AQHQP~faNW#wU+RBxQ
zdnS#?Nze&M?t}&Om^nfGzOfIoMWGh}BtFU+^Ol)Z^4HO}Alo#Wzb3eGc}{dR)yqU+
zKco3(uG|Dlncgtv&FQ6*Z?wygi^Wmx0wC9(+_?_PI>i%!*1KW7u8WDYJ{IrNZF@xs
zr`mzbPk{S&xqJh1Ht9cwtN#+8UM&9;xF&n{F!m%{C$?|z`ZRf4j-24cx2u%-984}d
z+|NYq2e;{cew^_xoW12~av9ko<RVGb1HEVo1`yPV9qs#+y$O$LUCSWwk(_)KzwH{J
zvl|ouR*@F#0}XDA6&`cCCS%^ilw_z$^Ux&ps7ggM1Dc>aA@9cn){zl|InqaxnFXp#
zppTLO4v-4o#$aa*Q&UKsNwSB!(-iek1R1Z#U0-ttO=$wsz`EGcefG`nwM~XLM|$7j
zYR-Bn2Glj412&!bB?7<NT8Mig{8A=GQJn)yOEH`U7nq|8M|~knd|B&<Dsa7t6;J!~
zSSd_WOs{R*tLzse5!&^QsE=>AsCwTUawU(5uOzWF=?9s3MT$!PU+2sFrvEal;;Q}U
zEd1}MD3G1MU+>Ygvp*w&t`st#I{?cR)6{pC=Ckl(zL@2MqjkUl!XBZv4L|>R_V0gs
zxxsHgm%X?P6yKb@J&GxP0*D;xH7a{sOY^@?d^^mwr*>`)T4?87yL(0F{qe6GzO_3N
zI`4@)o!d;)?Twli5S<;@A;A+sO4(cZk@Qr+bvbD5wMCl48OaHJ>}o(XR}(&Lc6hng
z4M`K^ROGs&Vsy;=-TV$^h(ssgzETCsuM-L`tV50!u0u{6_P<n4^R+`+2^4VQftMMe
z3KTe8CF(-nSo{vjpu?*^0i+?jck4gJde(Paw0WfroCj^74rz5w$PMXA3@}#puV=O~
zk2{k1&+3xhr`)C)6G)MbfIbte$%+E1wKlvl!}m*3YFpO8d&Xn3#|ci08lbM*1&Xd0
z?Cb~jB}xxDuUbf?ZX*OQ>3s$$wox~}g67EYs?OVsCQ05F-pBI4?`-)a5y<g|-?DkG
zmS@48ROa}=bV;OQkP}&xrfB{(T@m+73>d~T9d98+o4-$<JA19%|C7b}NFkihMnu#h
z)&1$B6ba_`5$1frv`2nh;&zi^PFWFpvWc>RF_?;h3Ld>IB8v?}-R;)52bIW*$X~6%
zChJ7FPYYguYA2)1a_aJ9IbUZreY1kF@GGv84ki2*^;z)ToMdGGwfeO>Z}Vs&Q*Rg|
z4q=_+NjW;+WI{ibwEmja`k`EVf0xuQD%Kn4iHE;|T7$lGx|=Rn&H!ze3wP_FkCfTw
z%@uQ+R#Mm~quHxjTuRF_6!*q0TClB?O6?Yln$s|lN>tc{DEi@%c)sxVuFV*fUBHJC
z^O`?)luo>scg<d_&)mmvWuM!ebk_8+QUh09UKdf5iYqq<ASBAJ1jhMon-fXNmF}}F
z(Y<Bu;)u;AEn*<Upu>Z;A>Cj1=m~0=B&K!U)Bg#>2_}pf_?`DzrSz@l#}KSrhZQPV
z)sOkp78ZK^WmvRApQoQkW0mJWkE<@D;!!~4&3v1ws8LcR<ESRdE}fj4?trHVjSks%
z%Qrm|@`W``2r8XK`LD1qcPgh-J>pfHDjZqWhS`bew<Hn_mjrqEh|dDWa;wMH`fQ6@
z=2yxS-f}2`<pcWb(4(Qj1Kgw!R-Cd~_P^!wv4~LqWV+)gGaGY`Y+~s(V)P`7_ONTD
zYX@U#xM>i0uEhcAi*L8=D+}wic5SdNH+PEPYM)l(%?b%pUilj2U;G*DqQGi!WIYOm
zpCIE8p^iA_<T6feD@Kt837nHP3(%O@BJ_%i2%5@5DhatiXahaTn)VZ1J{m)>i|X`7
zrYMlHF-oz?%O@}x)AR89QUK1SUV#2kXR-^A^~ULw^9$XSE9TTY={d&@)!f7rai;2c
z1{-%bB~p5c0U<k=(#IvWjfzznzep|m9w)}y*BZtls$l^L`m`ba&Y|4fhJ;OwJgtA|
zeEmnuM=gpb>m{YE)PNs)+VBypzN$c2)<f;ar7R2C71b-+uP<?>x52SMlAs55#JUHN
zu^o<fQ2sbvZVD>?Z-t}zpkR~TB(EhZf~hjY2rHBt(mC#zK>W^F5)ORLeNCGFsJ30&
zK1a)t)iAY4YraR%`c-KNs6?5vz@;cFdx@X)hqSN%z+Wc8$W%o7s-|L)U!WYCRVY&D
zM5Qim84Wp{gG;ijAfXtgZgsj5)cQ}4e^hZpuxChnJFmD=i`lu<%LTMqcR|knynwWh
zmM9P-t_FV+`h7pxC1-W9MBRd>Ny}#1HO59~2~Jfc^D0$Dzt{{>q4PZ*6;fx=w}nok
z|Bi!&hSR5dYDyA}uNka6IQ*(J@Q>qjayX4BL(!C@-@O-HVtb7<F+LaeLdmeOS@8L%
zcYzX#h~AHg@;QOMm0t8<5~UZW|DC*#!n=h3k$&?Z=i;%(^6g(;&EG1WaXrkz|KBNi
z8~Y2?{}Tk=Pk_JN7X-r?(EraZ#G8SBd$Y@74%t6<AB*0|oy`5Rq1GA*iofq5I*WPl
z{7Yy58pGdr&AvZjmi_dQzXtKIDC|LfI=_5d8~byGlEXLRV^|zXnyoU1jT0p+U@0mf
z6Md+pbPY9o;JZRtIMbL7mnheb7O`=k`tr;R`_BQm|8-IHpO-?HncH(djoy|sm%54%
zs=sf(L3R@MylYmLi!Z$+4cS4jVeD62pPeY(H9Y}X+<0E$*|p!kExda7sDEl^r;L)L
z!?H^i(3Nb5Vh|{DPU!!hSpf3-`_CjmG9*K=?(n>rccbTfIUSrNDbyGv1B$rk%9%yQ
zw~4;`KtvXofFt*ZAr-ZE8!Sp+<&ZGArWid2Xq;c4zL8laD5us+{0{Ht0E-@dQZCxN
zrX-(`Iz>t1AK+j*kiMi=qh2wNh&hTq1)H9bPYIpobla7qeT9FSU;WvmB=S{)&yCL@
z@!bm*PEF^X&2azE06CgzV&4)_!5;q(WG2S1Nv%ke)++Vmzpf2iAgVc@BsAIval7-y
zWWG-Vn^6n7qm=2n5^-WCka%srR03b)?g58D1F`o-su%D&;w+`R=48PdDVL2?bdO^=
zi&6OxV*kU0eo$10-i@K*U8v8ShK+B8++iJN4XQc@rNu2{h*MbXs`To`lXb7&%praU
z)2XsY%<|tIANy;VfAm2qJFiO(8<kPOH<nyYt><8Z&(K{(swUo;zl=U9>6%6GGQp3G
zsamwKywHKWk|<d;pTtN3pC!MH&-xM)h9v(kt~xiSCFYcTC8)4sOl{j1LveDsBt}LQ
zpq)7&CoWkVPbRrsGiqS0dwX^I=EFZw51s%Lu2Tm40sJ*Dn3+&r+mtLzQVuw#Wh>`n
zgrx+E<v&Y=7|HRGi76s7NQ$vJ%OmaqaUST3orohQV#G}|W;Ki|{M?EROZ;u}X+yFq
zI9C)7;26w#4ZTN&-z~NHj1zNw_aH+pJDPutJ}H0<q}2rXGFV`7f3lt-f<H6$huL95
zUEEx$Wwv(V2xc`E5~8%A!1{hfm$f>NM<ueA)7(1|`FR^<dj%D7DxoPEjLtDM?LAFJ
zgarX4!l?OwpdQ@)8HbwYn1$sFNbb!2(|4ImhVp!l!62Rq!cQ_dJk`!BS3B}?F}$Z4
zGDotR0=w#tb<;COwe%ZOUNv*sVFl(&(nZ*ExI*1Al`GH!uqaI(y~JAoz;=Gf^R)PM
zDMX?X#M5H8bfr{&n82>2&0xm-T2d=O1?%FQFd;cB@ecGF+Xfsw0X!@NK|||YPhKE%
z<J9Zj7B4y#&&h(synOoT9BzzQ_Q5np2AAnuqIPo<{4&1i;AxxX+NcOF;(p;hmT#^u
zg@Iz9n@dtFkiN1PTA03!d3oNaq!c*v-VNacy+j`kFnbVrP)%Z&x_(MrEY(=K0=s~6
zGOvWZI_~|a*YR6POP(Q4z2Sudkwj93xdM<Bb2HlOBVNC>JMVFRPW7Ks=_1rTS5d}l
zkkneNW}?fZt_8z6VKq>WN~gwVhZc9xqzKvN4yMIs`XP905r?<#p%6;BHuYdT?LMg4
zrYexmN){Dw2#9SsVg*zvqS8fwPUmVIOF*ZmtK-Uu<+M@6(LsbCL`n;48z<Gb_=svm
zc9Cr15=o+MgOw(cRZx)c$8gUY_B=iS5AHQrJ{te|0f0UHT*tof(B*a!Pff32t)b7n
zB*ls242;!+S3oP-J0f^(>LOwY$<@wZ#{9O6pTe2=rNGa1Uccbvq+*IRGcUpVVC^)S
zN{=6cR@<Y%s%$9^Ob~AZ!RGyV-@1tA_1c?Am(KArt)On!LPf#0UQP@v?3yYutzU36
zPr83IKja5Q_Ev`rr{cDA!o8(GDgW4TK(?k5T#a@+%u#xgt0emDLs*i;#I&-8nw*jn
z6SfFTcWj7ipfH-o8~_^{i0iFmnRD@CMoC$lNtL#~G!TcE@&av0itVjD6(=hbFIe&Q
zehFq=OAhf9;HNeIGz~A4f5w7xHhw-B*boA&URLCyuGHZhNs%O36CEy~0=V-cMY(QU
zN4D(!qmA8r<%Ie9rb;`fwSLeMo;-LP%XLT3@O3`=M^BnL$?GI2O*xILDs`HcYJVG}
zx3T9{$mONur6>3@f=J}D8u?L8$LU3N>{fw(c5(gAZi&0GG6{EfMUfs?AeDvql6dud
zyE4p%fn%wDuXb}(e8|zoU#c4XTUDpH-3&jQ59K?aX&FzXT4PB-&nwttEA+@?Nv?;i
zpV`CO^a}7s_IK^uRdWs$mQ)xcM>5onRIkYf=H-PL$Hx;A%jB)--DHZoABjr@lDX%6
z$t%+L*F%JLYP(smJRp7ctoYM=(Ble3VcSOif7<+CF7aQS7s?qws_pF`Sv~wME8gL2
zzV;o|zpd*$I70Q8bse$)xO@WSz4^<SfBLE47V-H%A(#{YqYM<kmEkW0B!Bbd{BQI9
zi2&j(%zrtH7Kp|D<p;Zd!iVVYA9-K@rMtiAhk*ZOB@kVsSd;msYrnulyu)vn{{cVU
zxOIyMew?fKnd&{ob<GTSuaVQKEyreRd$1@g(%GG_d5&%0U-!0z%`O_e?xz2Du=r2d
z2(xbS1Q_F;Im9OUsI=E6R{4P0>b;_HQ^#@>rtyJgfr(_S7LvZ((TS=1dg`d7C3kKN
zS^&Bz!gI#Vc*p&<maq#k=ocY$`vN2>UrqyWdp9J(*sL@|+hXHh|5{p5y&|w%5m?-1
zJ^8t|kkpP*V9wDYg#_R%+Fb?fxY9T*JHf2|od*4oe=XB#@N?^dt(?)1cCwBTHaw|x
z#sZg4QANJ4Y$(HmFRieA@YX$JWpYn?f{qv5i`<<=us$_~pgKK>dG)`Td+)HOwyj?n
z1*AwX(t8jH9qC1SNoYw(=tY{+dq)rvkuC%Rh=BAG1Qh92L8;Q4bdlZ#DN4~VpxeFA
zIs4rEo^!wZJohhFSYwX*n=31GjxlBg<NZ%NV6XLQpkShBWVni_28hi;qi^VO%fOyI
z(4`0+N#U$Q!v-NR*XcZN_!h9|GT4cXT<X*hCiLCm2t0sB@K{fn<|rs~*!sgfyIG#m
z6pFu5)YH|^O7m_<Wg7w^1d~Vi`$z(J%AzJ6X#ID=S-uN?GQ4sWzL+GRVfMo6?%Gj)
zppvz|O2YG)7#BjuXNJLD0#_of@YgMiVlLG%KCt83un*;%;;)agowm*w<B7t(Oj*_P
zv~N$nDFL&lhERm+8arGgt6ir6s6<GCJFW*TBqpqjj1^=+M>u(S6sNFCc;tf3A2lB9
zo5VL7+s4`J%!ouGr94PYo-dUQMBg@7K3exN-(!08HOkk5b+1J{v-bP$0Ce_lS!H1S
zX^)&QM|XGLoxpwU(}X>FK5!T5(6i5v6Y|1YZr_>wKiGMpb(siw1{*Bca{7pBu8zys
zgd?sN=YaDs*~+N!Na4i$Ta;LBO*2o-C*3UhP|K&}XKlmfM&5nzbd9;qb+nqdO0GI5
zd`%llJgBKnKkz)V`x1~~52xg&QX8|UZ}7*%7NYC}lcx<wc-4Nkr0Bi=k?l5$FPIx+
zf84lI_4WM5aFCX;poH%>spIEv7DQb`i(8;YIOMn-h(HWO-t{ANsmQlD89cZ#y{p3V
zZ1zE{q<3pg1*od)t@RAp;>apF9BoPhk|+{F2$tBw1(vtZ-fRWaDX-W3X1P?jgsy`P
zjtuGPUVmE^E!>)c^ckUmt0}ymCKaHy`tE|Ua{S!N-)fs1A3sZ-wpV>?67y^a9M(r(
z;hV(`NMB&gbLlg*+IHFb*8#4@;hg^{+v(h##f^8fp>Ldu>ZaGN^K%N+rTB(|)+xs&
zgH1@#73*GoC919TuXM5^<Ktnqk#r0o2#$@1jjLtcw?Eua+B<XdVDePt(WpPHq9Sun
zUCtI>9G;merpXsQO%@h|V)4VWc%ZW`t*xuPG6Fx?RnYh#=2a5XiM&LpVTh9#D)FMj
zbY`wr?o}Iikp6{+nHcXhR?g`WX}<LN<@IWTnV(*3tU3^1rcF5y{Dn)L%o=`4zvF8(
zcPX6nLxV5gMclSt`fd|n*tnzbe6OGi*Tc@)S9DaTv?)HqW<tNu=S5CNVhnTi2SY2f
zXiQA+U}9Q|8AEz8W~sdRTaf_OIM`AT#3e;;JXlqs_ZgXkA|fOCgl`$`BXHKrf39c4
zrrAlD@rC)0zD6&k-wO-8o2ylQ)n*r@|DoXszuM0k1EFi#qhtKQRPE&jVix&IYw&$o
z5`{><C2K~!aAN0;ru{XC;*H`QbbgxD+Oh&t${GUGM~dKcB$2gIU&OrsAE7$@ib(sn
zrqu!%KwY@3n8{YON)M(kt6B7r2mX>%q4<)~1i<Ukn+^-yvE$UgHM7vU4HbSJqv_9C
zFhzrz80XT=;dv#6<poh_q}@8tl7^|r+pN4&uJ}+X<%jhPnm##E)loV=nGi5dg<d*^
z6h2<kHQI*rxpd+B>GOD&W8!I$++1K8>EBS}=6F73{(_b3CqX&Gl{fuOt8joS%JdvM
zBgoDwoP2zZ*?bCrCSV4_Cby_|c*L|gCv99Ot-f=l{2>m2;#&6DSFXl3rt@j~=luOS
zO&f2hLkX`(D!O;ZR~C7YsVUy2J@}rIeCQO}sOPPqWABf$l}aT_ZH<@!b{s4A5HtLv
zy65mSXf*lyut1>ZW&vb|9E#{GBTiCfW3UJC$e74T;gI(XcsmEIXwR0FRGDn=S5Zey
zeb>f0m3#1TlR~H|v34^tE}Hl%2AT91kEC;CPRX9ni|g-#vq#`a_&~8=l<jaYV+1#n
z4&s)_!xjHHD^!AJ(DD)R;PQTB65_~(Xux<4Wu&XU;ywZ`g_MCVQ)E%3Zv#AJ+jmzO
zAN|1Us(iLpa_7v{l`Z^VlimL}Dekek$0E<R<x9T_wtV=ec`$h|*Z%wb*PZhZvjH+Y
zhs|+%nA|%z2?0zhlold*=LZ&<zoH<`XUs47lB0WHA9QssS;D0XQNtvbm}PD}OpUsh
z6o~<*rB7>?CrPS+<wM2JMMjE6I^#LAweg!N#W3nP*Q+<b_OzbPPfp?m3rWZiNTglz
zsvI38yCDh&dycp7V3Kd;hxI_CEMf_qyh>5}qzZ8coL2CGCzJvoK$48PSsPJZzF!nV
zeR<_*S`7CeXDYI(+&ZODl%pIVjlG^~?1-X^$@X8AdG@T1u5ElO-S%DFOHUG~tUNA1
zykOI?x;}iu)x0~L+q-tw0B*x%9s*i6F2AD}-)d$Upo*F+@oZrHz5#4M4?#LE9V?M>
z;8^55KEVXnJ>GqLR2L=#*|!yh`_=NQzE8W3MZl<I@qL0aQ(PfXQoc>>N>?{t^MU<p
zua!p`>E6Q`^~COuQ&kxaf7$Yk%|q5xyakjY>DYrLdB&%sk1$`5pG*91o*ZN?i;?<&
z)%%PHIm0X+lx|tkCOmoO7z_L7q`rFs_W!g{aD{r}Z;~$U(i!OHZPUHypMI&3)yu8y
zD;%~P+&^pxG7PPVt6m^az&q=1>E(>ofXD}2%cXC36R)n{;de9mZusHby|%>cA6V~e
z8hlUQ3*qq8MpM7HpPi%#gcs5189GSy6<f;A_RSqmtaw<@#02*H3?DMYK1NLQA3CH>
zIs}w3iJl%f<f9`HU!uXuN~2Axl9O05+k$|jsg_yG$6}?|i}Hs$wcD2)9!k{?(;pby
z)98xW`Lv)22#k{h=q98lCff#Gs>t<*ra00}PK{wfh8%fHqmb*!#9p|5Jl<XbvmbXz
z0P`Vp(9Y4bxMs|c{>-R=Zkt|yqFHyX{5!&+3cbpk`P@NHLcMLe{eJU&dt*?u5+1wq
zOm}Bh#nh34Y2bh&Pv7KATIeMseN1bx(zgLUIL6?9y*ul#i{tzduA+lu&iIM2xInv0
zU)e0~b>DcD61X*I87AgURbOLBf9z#$Tcgd{BuqOBy{rWcFG<H`Eu@24_o?}E4uyVU
zDkupbCp?(b^Hk+CGcvegz`7jE9_s%#kBYL-X~k#E-`s1bCMcY-rN^#**BJlydQ(~2
z44+;<l-^<2vlzla&UJ@diwB1~V@e%6p1V0i?$d<Rg2<pwLxm+e$4pMm5^W}3qSZOk
zuBh-;>Z>wAhw2i^iSGx?AkNzjQw^~CR+kw>S=tJ*2fFZy9nVC1D?lj{45Lat^d%iT
zm6a$acMReI7rU-D^w^b}ZOeof5+I9#5!9CVE~#*r;4*}29A)|(N6#C1={y@A?J29C
zg5rX6g(3;ZDk8e8qsS|DT0L8AC5fbDZSspt=QFbkbE7~GfP%MnvA7Og&axpMjrWS2
z>syg>Zq32<=+l=9VfsUQK!CP0VRslvadL(7N$885=hk-^GCg)0jy~M_>f}vhe$6q>
z{(9PiMv+bOoHn(bDJV+u%Kp24-gIN|3-_Z2l^A)S+E)e>UInPU%6DRNS$(J(Sh-XZ
z(VNdwZ|Li0U0vx2B=P8GQc@c*feN>-G#4<|Q=Z}#@Qp(ImwR0I|ER6X7ykHZdj3_Q
z>~i~Jp}ji5I>+oLx6!tEPgTwoh*SG|YUsGATTxWkFq8g=H`j4JpIR~u=xg)^2#Wbm
zHMd=v5j|U<cOIekB`n!Obk7yoappX^n)%G(1v{)5=hhFb2r#sgMCX**V7vZHqC-P^
zT8Kk}ah1uamH116%i`pHP7B|J{Ck@{F{y+)(xf66JR`gr(;-C&?;ve{W7kLow>G~`
z`qGsv-4rnY`{Zh4gOE;r(A=Z+r_Jw$3hvjzVXNq;ijA|{Z>}8=GL`zgQ%lt+3Ulh|
z4b{DVlf2IYvzk}5nY`|j&G4bTcD4BSvy3z&!jd7534=Dkb&IeUQu2Cz8Ao7rGX12z
zy@M6S8xbN8we0x3giz9vHsUqHSkb3T-_xkqFIlNZMS+VPWxu3T(Ai#vfLoF#B^gv1
z^H_ogtMs)4=g%s7H2L?nNphm3(+diC64mP8VQMA0v#9iXtpSwxdmf=S>2-Pw5ReiC
zjZl5YeZHs!4}MDy0uZcAhL!SuI#F{+P&nZToIZTivCi1JIXIe0pim=8bA{3g3d{h$
z=9XR0UWxmhSnIOb(`kA=yn+}&ZeS+nAiM6VQ+F$kKLY%nkmCAvrG!RK@T?m1%ACsy
zQBZLSw7@Tc`1^{c5od)XAPL%|t5tV+osxhMUjaZ}NIMg%7x!wUN=ZKHO7&n-DV4e|
z>jyjjV$@AXQ+NeLMJBB68=v!CGV!LPr{@?76})Dqicl|c8K*YaKAh8ZK*kLB10hlO
z35nVPfXx!%B;j;G{}j4D8;`PA%f3m!6j5m({)9b;x{aBF3hhssN<(r$N6Cuk5gvT7
zT!xvTR^7b(1wHN2RR~*R8;@!G)y0~=1nC7N<TgcU-Vz6$&eAO+49ug3meB{rf+@IW
zG!{&sDu4pPXebpVDMXy3uf3~F>djnN=bm6t!=_Md@|PkPD*db8FKH@g`HvmuaY4nQ
za5k}da{QruKBv+^MHQx#IM1d-FWicLuTTeRzXR~7T!G}aT`kZ{0iAxz_XOWSK7BE3
zbMs6Wc!a??`RB0|Z4*iT=%BvjB`{ipEqse0s#Js>o^B~4)3<w;xf$4>a;JV|T2k*;
zei7o#ebAI$b(be9TZ}!aOgSe0THX(=hm{b}L{A^n;m(Uz@hMf?^)>4(!xkoC&b^QJ
z34E{G^Py1?$TiT_<!@U}@vj=KvmZ6PB^kWS<tVI8d&MbcYb$&Ec~RH<oE()7Wa*+4
z&c{sXquO)jK;sSEqX87g@ppSA>?uGegCfQG<dg@Cq<0_1`IQvxc$j<lWmiCSp$OuN
z%vdNV;Sa34`N?#YK+1-B9>NcX*g5H5=I-^AcI93UIkkF`A7RKQqy`&^4FuuvV+~Pc
zXlE&JZo@I3ug^h$ddLNdRZ4WBckK;p#d*sS?J6VRb-$+X%hSL^_Z{io3D?M$9;mSc
zNxFUw(!SU_Pj($HPDJ@|=emFwmm`+T>Vlx;xz41Y$LhgS4$jmPQ<c;d(*kVQ>iDCy
zSj+4Q5&fQbpSIqgYCjs7e-}j0f2^saTd<)$JE&EQzMUMOeQh+G;i~y$2;qv({0p4{
zeO=YX?<Xxm;ZJfanw-lx!<OU6U|CdWJ2%s8JfB(2VlM_h-JL9@5(n#Px)y4Y)6oD!
zx`4(>@O<u!MI;E&awjNUZXspWnd`)1;s=%iL#N`xtW<FL&=0Kta@)s6{Ml+ZX4}W)
z;|%H8`L<8JYb|MS;Nqih^Jb=_6CItNTFp9=ZApfEMB%-k3NPLhiv{W3Myz)9PFhYC
z7n70`;8N#}A5E!Yrf<CNd*X(@x!Cb3x|%AGdj83iBr#1Bc&pL&EiaHqFu~LgY>mvu
zd5=GRjZ{Ex_jp$8uC@^f=WrKL%m6L2G~C!{L_Own1{IfI+YgOSq61;6<pq6S{U(19
z2=Jz!jOr45L*5>hN<y7!ab^)`CTRaeO$B{RXD?%_xu$OQ1R;CuQ=dT|7TT1=m!7n}
zai;qWX5HUY4z4~Hj^+0TZx8}q4#>8pmWL}9SBY%!UVYdQpyt(bpT2v5vsuPN85+3L
za6RRvK7u*miR|ee3aZfbZ^{#;k(HA<T3-9Ys;@eOza}oN%Jx0CCcjFrms!y<&d<lz
z0_CJy!!Ke}#ZKhNIuq5eGN%MQutI5Ep)hD9qr%^D4ODF}=_~oPJdn@p=m8;s&J47l
zwghoDrv+$)5YXyK3At^_o4(Z)f5+}NgvtRZB^h|;Vaf>>V@WvmiN~I~pHdvz8qJIF
z-+ixAw3tz`q3T)?^GU;+quWBa%0OO5%+QMw&-U^beUk5HKEvnzsLn8Ik9X;t{Gfgd
zJklFEX=!2)-x0#HJC)mfA*y3V%wavpn!6*<4cfda`%}`l^@CQ{9{b6g?C}P7^{XC#
zToi2YH?+EaJE%<TzfKFX3|T|hYHHJk&lQU`!n=%Hp{o2HPe4;LiVQsM;;*{uM%Nz`
zHQImTs;WCKXT`j;8W|ov`@D3zH+%lv4#;m4q?e4`atY4$)(kfnb*G%1J~PPJUD~OZ
zZ91goK%0u}7Di-_t*lGHYi$>F$>mD>NBrlQ%}=a*<}B|Y=VFHYVTNbsdaGlG?@~_A
z$f;)v|MT#*KZg%vhBtrM!@SDG6}$DcXR-&H6c)|E=2K#2RSs+9p)0{Zu+AUY>TtiG
z@`guTiaz(Wmt7EkX*s_q|A#KW?)_x)r6Z`}>|AGA;Xie@{;f0SwE#2j*XL1}Vz-V9
z;m^?L<1TnwFQ}i}_w!yHMjqx~JH`ziDa~!z2b$qG&+VH&QxE<<wg{uo%cY$cZ<wI-
zk7F4z2IwN{bK6)hT$)+Z_jT`R3gasx@=<2mufuP39X$^i(58Q=C%eM<xd)J-4x|C7
z5Ky2lLo_;Zyd+#bG5q9U7=E(K7yR=7<hPw(MK+xJc9{JJ`&+xK5C#X?NV+ANxE=p+
zqV*J_yLOl;*9Vl23ebKX#0c{`F!#NE9z)^XTucnt)@#n~Keuss^mhzHrT^rQfxyjk
z2+YX+0|fqV_)O|~{5smt-BgIi#1CWsg}N&~mZ8!Yuy}M13yeri7|G7gVYBcHHh;bU
z6E>vhu=xw>WdB6nFBJR*b-#xH1nn=>VWM(AyyhO}OX&j07m-rKIOP}gAIrQVB1y}n
zM2taHjf%Wc3ST;jgr{w4Jqc4ahyqRQCr&m!8b9yuFxJ?0o2?xG@!bE{0=@rrS>A%f
z_n3l{M=OnwKR4H`R^C$^4r=>iUgK$DGTzqw1IzyX9Za*`4s4J4mp`y(9ON;p(-E@q
zDY1Tbp5Dc$kIo<TwZOc(&^7F)5h6@0cbOR>)BBci8X+@4ijX0K1k+o;^BO{{A#`nN
zjt}@yUaT)c;%bm857rZF9-RQfRp#$Zeha2~1)}^muC0pl%Ih!_yS4YxjR;z5s_VJE
zm=sSs4LpCY1ni!cS&!RnhT)<1EAMY~7qybmWJ1)rgV)WRp#|%CxuGcDh@ds-be6x!
z>120VjH3)nj!2`h5D4}m%TX}zoP5I-eA+CBsHWO1MR)@_=2qqDBAy9EnY#}*yfZB6
zmd-U1jw^RleEiW;&U&15m}&!S$-y9LugPQFrFuk<jUAlJN;kOz>KN@j!z>uEERH-c
zW~h)BY-0+LD;Op_oZfk5ILg$WsoKFQGI)Jb3i>$&ui1RZzIk}#s@D-&x@nfqjHx^E
zFbAvW8n)Z40%WFf(aXHNV57LS6+DFk0@rM=0Moo<BC|ONbsOD0NEJf9PU`G?xm09`
zKk;AZM5eh5^~=!dFLDR2$(%kshPg}nt*cq{pX%H{ODg*Wf4&(3kn+Dar<v^?key$n
z51jKs;`DNg(y`God?yUY?+ZQ{ESnFS6s2>nszmMXV|~54!!%dA|F$koG%%S|=w(Za
zAeniMxuag=J-%iO-afGr>cb~E)cX~~<uY||t#2xKnv;&e1T@bA7CFH;cmdxmGRq&8
zJ$?5qMDEj>=sK5RyUxmb{*qwkUAgS2C;vw&57BRmFd%vUQX@W3fCrWpz}j=&yy4a`
z{XK0*Vn>g>S8Dl&xrLc=qNR79UPpl;8eQ#!$L3v_JPBLt)AW{1lWDe^A-z%fsH!%R
z=cv<{y?iX7Swh$rbKI4WZM2^kOdT242lagOd27ZeW7~O9x>P)&7NJTU6R6vUod$E<
zmkE9MKKf^m2F#{}z1m3YnW+z`ucx;!wudTEQj4Up!$2}d=1MiOA7meMOzsC~RKVeI
z6F#4H9<-uXNoWQ)FGz|tUW#BYPJIm#KGLA%)LjE(aDb!^YgjQ3fHim+WK(g7{AsNn
zdhLt(L>vN{f<C1c)JK^F*p*n<5Lcz;WcT7LwZdOOd_WL}_Q7dSaAJgUegNSC#6aii
zX!Qwofs{k>&}C$urH*Z!STUOdSdqX$Im;YeK738HsFZpv4VYk_I*90NMS0HCaTE3n
zba62`ZF{ZmhiBFdl+CHz)(j7p!Qnvsk6=hK1q@b!HJb%X%rtuwUK~6^Gs`$+ED#WN
zfZI0&%nRLsr?KIQMB?p<U=g&!URDI0+}daE$mZQXau!tgY|1A`j6H-$z+yA4IE&Dd
z{bq`|teMSbRdFcu7rBMiz2^2?;WlOU8AADL6{48e-ocZ%$SZCH96l>AF5M4l(_9(J
z6w;*6XTLd~tN%uy@|l$RsfX=GTEB$@uXAuoX^nwNanXLBR}TW`vvb)(A1mLfTowdh
z27dw%h$poexV59Y)}*P&q`lnXTB}PhRSAM#*3Zk8n!wq=(L!=SFQJ1N861ThZoSe~
z3G2{iUmMC-Njh5vam0H`&s?EBf;a|2qBF+o!?#>=v#ZUEO69bZlLU<Pcpf>Z@)c1P
zMe1~7(}I>(eTx>$3Kocx4p5b&ZvA%2#&`8177H~pu%Av6ezgwvjcHhCrW#x~Gs8Ws
zE~BK<Q<MsWsP<E<D1!Y@ox_~5G-mx3SQ&5b?O#J7zl`RuZCI7vU6EV4oguqD>seLR
zlm22%R4so*V*RQ2r|N?{*G`^~LUM+zKBSR2YZBm_gA%N*=R>jRB-<sn0!AGktIZ`p
zOyUt=M^cgC%MR7lSElK=Df*=Cg6?F6F5L9*F8T<zG4}^Sm>7u;js<g}y2NI(l?UYP
zhPH}v8e<};ycjZd6|y$m)nk)_q~WeH?(Asr)UDxSN`m!RV;5>5dFl#WgNyd78CgE{
z$B(W-4g&1!XgT?EJ#%mweei75B*<aSI$W!albe7ePPGs#*XOgF@1tZIhb-jONFDdu
zxPsL(u}JQf8_dElUv<5T{cv}Zn7DvH3FYObf7Mn_%;|2g7IZl8x|b^hFYQ-loV|k&
z_R?c=w#uz$svhXkgl}JS5LTSIS!7@E$tjBBeIl0@%{_$HHw+JOx$w&nnj>BZiEYaK
z@RGFqxMISdDqlF+l<7`cJ@!<RUTABfjo4znHWXu_JgoVkc5#hG4u%(B9p|@^$f0QI
z{iRZ$MUC6qz~Uk6s~or{Tt4m%oo$d$s;Zmzy<CVpEx>_D7xAyx$y!klJ)<<}0B5VK
zg$YJ)yuUsjjW$bu2j!ehF5@l}6&h6*qj!f_!P^s9*_5C1^tp*Je~M)(%zcS&jplDO
zPV%t(mh?qsMQUQP&lmHoY<YC7v!M;rv?z4;hCp+l7REVvl?oL~wJI*ow!y~IIv_(c
zRdiKm{MGAbH(bJ-+(zH@G2Ayaj`FF#uOt*30wb8bc}U6&amUt&jW{?qF*$R})xY7a
zvzp;&9$L}bZ7(J(7QYW-xPwCiZSh%2SQovSivGePIeggF_=b(DaY-8SK94c1ki!ix
zSbAqvuh8irvQ%5OBF<)-kXoOH?4}zu7TMC;j_HyR(q;gvh3$IaqqnI>y{g3PpYSYv
zac4}*7I{r5-r7tn)wi3`Z!~<opx<L*Svx*m<iJ1fxMiC@kQ2lDHsbvWh38Wv;CGG7
z!l9%Vft@$(<oNacXo@S(Dl}8$wSc3SwX&^g^~VK#NxmHB>P&o6mFB1#%oeR8ED6C6
zA=fG`Ev6iUPU-6kw#xa?uBdNl>v}h?1=Md;H9HWirA=rz10krvh{UAa^hr(?4_=Ox
z=X=kixRsKwXy7cA?VFGi+VdU1GtM-w<;r0EfhCe^$6g2#NvJHocK-><6+AALc%7qW
zp^kKkiYadcV(vSQFjd2bZj^iJRR7+qQd=W{>sJ}1^cz+$<|l=GV<);qqDy^oBvuvG
zM=DWI-bPKLc*<BWgBIIYQYvRNbkf4nahds;bA0kWhZXWJEoa}chlWb&Rm0T{8NIMG
zbWkt#jou~a3@Y=#Aj?)Of?44?I?BjCnVeg)%y-YDN1O5E_$>Hp)vOMnsruNdv7aPu
zqA0Ky`@ReWOL>)gmxM{DF{IzV4<#(gs7{4H8Fj?hm7>t&S<SZt&eS^JVw5|a%k|nS
zpv}&%Fj>ou%0j56&THVO0t(ZaY@Lud#`w|utE2kflbmRT9H-Jy+KaFyJ>_&XtLJB*
zn#;c1{T8fp#=+Gl8BsH4k_5>G$vnwe?NoV}?->OjQLzbUb1Nyq5<;+EV?ST~H`N%z
zJB%&LJol#b`;Tj9l}MR)d-g>V&#Ih$V3ou!vi|EJ(!W$?=$8GZUZZW9=QdxPJ0?Ly
zP4BF2Z?+Fd-sHeo_xrf+5ouS@;-d}kxqFpM)HF^hPS=i~yq@-RaQFH70-Bc+E9rQ~
zulJ1<vsVQYAPlLG6o~i4eF!nV@Ac+<p}`wF?`jbeQ)hGe<JDS8TX>MBYCd&hyW<^I
zKEm0{NjS*+-DH`1OTk_cPtouS+si(oPzv9Sm7T|pEGY&`4y)uAYOhw{k22#FJbc2x
zv~-^C+>0;_KZBZSyVShVb%caGQA)C}7v90)H*HWq*q36K&MPf_tOqw3-K4vXg67jr
zDk$LDkos>XpCMdE=sl93kF@e~Kv(ReUK{F56D-MmPsukeFos0MDMk9cJZRzW)3=ST
zqR1D6=>sx-GmOWV0LZiISP;F{`(lv3F!IJD3zz_qdi;G+YiqI+#GgR+F+5ZxOr4uG
zI}r7FzvTIuJ0*dMmZo1KJd%Szt&`Q7^;Q*@n8=rC0Mc_Y;4NhFK}LUrqJG2x#kCtt
zehZIDU5xfVY<nv}3#gbMEDYQ-a~^uu9B0bHr8uGY<Yi*~8Ki@nxH@OQs<jeQ3>5MV
z0SN>UFx`?!>|9jV&x|>11#)vG@4BZd6|b>sQ7wib@QtAQ{e~1rfrqbNu3inGHj_%O
zR*6lMq(<@WOLzbRcd*vt*nm?YsAn}_*fb9^p1K=J<(@E($FXtu0}a#yf<7J<04+-C
zOmyu%3)+zMu+6LtT0yu|dyMH5GI%Fk<t8I^2QJA!K)K8eO5SDiKb1BqFEDu+UIZ~H
z;e?g|L-iJy6pz}~8xNzM51)e8Ca>Dq%GgC*4p^5N$vr@x?bpv-nKJV#)_GSS$^nS(
zn9B@RZ`M7KNzFu#XmuDY%qSL?89OZSS8r%?a+;bJ8L%<o0`{mfYl9sSx6&JT%eF^O
zl?M)BV<w+y92Dv!Y!!ZB;W8+H+zPsNYj5wH5FcL7sVl9Q{Zt~&jT;-D!2h}uN6%(B
zpj^|$>Xq@=$O7x!d|oY$%n#J;Fnz>BW<Q&lg9k@Wj%91~nlCNpn$-#YEOT7zuZ+6a
zt(nN#+A7(azUtLN38;FO%z~B#osQ_|$0PMgvDHD3c&*KRHHOThzMoZ#jeEs%ekk9W
zGnsaL6wGNTOU!f~Lpk@0k_$r#!ca12oKt3Se#l;%GnjV$P08?^@|@>CQ=aoQ=QO-~
z&XeI>I?ILhr{iWrqR*@z6}q^d>k)+Eeh}1N>-`Dw==I=`t9vt(AIkExs8_1TZ}p~r
zYmA}%ilKafq4Yw3)ocUP8==u&!5QKB5F4cG5GyAEHXt!hf!6a7qj1o<Y3F0rFRb0q
zp2s6zXIB|mosk200Qj$Jv*SZ#-rzB`J*%A~X#~IUqnFl_lA+(JS3_<Z9BuV(H2-pL
z3+Ddm#eJRGg}XAlz&9W5)`S)}e|leexVpW5J0Y)uN={}YF!uAt+Y-F{xp^2$>3>H_
zeRhJObo)(d`kV4gz)t<mKg6#dU&OYicKnxm+S_7n<rL4Euf!#i7)X?z<gN+4yBB<}
z>2K0MEYaSU_)V(Q-f*rTMyiwJA2!wew&|R*<rmZcg!01X3!WDsR32{rdqwqsdgo6{
z3|!0q)}((Y{by|cYLV`5<9GfvzWjF>|DJSuwqg76g&Y1MO#gG||LM`cIR9l?!_P1r
z&n}q$!iL0eO@0FEBBI+T7-(Hw?UelKatvU0egdW~@i*dXFy`gi3%n!M!7%*;Mt@~0
zJNpOXS|rLEevvl+E$MGSZ<7D5rsKb$ttHmbj2bzvqDnsLAoJSXw<|3~!{OB*BLTwT
zCzqA#7Gjd_cbe<_IYymI4jy~zK&G9JZx6dZkNe_`pPBiD+gK4D!$>~PkZDWwuce!R
z`Lh3;;!KIF7wMq#a*H;~2Q8|F=Q%4ZnP0j{$2KuJD(M693#`*CnGY)Jm{;0Xsx#@&
zD>T2KS7^e_#pi4Gt8^FTKam0e^Pdnx%S=LxjF?@A454KbLpX^7=CDDbB9`hgG+av+
zg3~F^K6x^e+m$%b&l*fqYF-E{6Y4Cjf(hKf=>)ZJF0yb-vYzyde=bd%_Z0uS0JbNV
z)r6|lO0urR*BZI3qRmv<LtuK1jMny?yTx^14@;?gR<eq8_h3aU_muA`wBQzNW;JuA
z8?aw{xVeot+qT!bq!Tw#w1ZZnf}s5gCK>8TR$ffso8`U^Wx;34k)H8|8B*|4LSpj3
z@i|mCQ@FctBhj+N9~Z@;F;&S9Y!B^mvK<KZr7mHUMk-alhDYXH;h_)@VF`*;Ix@0{
z<!gc~W8-x$%?;k2YO3tNtRIGHHF$0Jez$c05a$y&^g25j8D&IO^fKcImU1CuHxrFg
zUVO7wU>|qFR#9??t%&~UJ5Zm~mqM(AIObj>=TPL7EH2k=8`?$rbqJ(AxU^$EfFIX*
z4Q-ir<6--Yyl)<d1iTI<dGCj)u{`itNhz6NwmZXAFNV;B{OgfGs7Aj!lh9Fv7JFa@
zp92d{HxhHukf6a!$UA`W2(rvaEZsW3sbWA0D7D0U30sh&%};50geKDx?<2(}$&-OI
zS%*4R>la2|O+4JxdIiQInPy=-a|dUnb8S-|Lg)!BY=1zb(c6-5ExQD)g7#!=ARXy8
z&-n4yMhOdv%6_%PNZyrbOrO>$x%_A26}uTW%=Q*~ShNI29q~IXnv+@2DMv=(oJJ~=
zdC|`i-LLjZi+8KE<!`u$Tc9HD%&{)yE$4X!*!8bl>fa6``mevWiCIEh_S!qXNnwh1
zs+9o^LZvY3ND+KF?3RS}yVO6hlEvk_y1p$nbxXt56miwDvAGgETL(3&o+z>q0NZag
zB$llW0}<`T8r_kS-A5h=^(2+>Pn^Y;9u%c3c1zuoVHzKFuYcEPw!0075ai*Y7U+y-
zbqCf78{Za-izpJjmWh<JTKB{k#;@6ou@2r1U#77B084%z@1b)@H4ZC6W>f<BRUS7_
z3-2@#<_q3Zr{x3z-$=HTFd4B^+(IfcKIFm?u>`}ZQ4|z<NYC!p$fAV1U9miGGcE_M
zh;>%>QjoGo&XP6_dVI0*v{L{wummX1(s_MS<y7?`&eXa_wOF#e^nsN;{^kj0YsCNC
zcJc?8{ZXH8<ST6xty)_R{hNg0T^%Owf+J`Z9$OPlIyM#T+W4jAg^U%SMZ841nMKRa
zA!!F<nu4{>q-fY$5i+l>QumD-2<)5KDRi@NkOD+uje=1>%qlAM*_xO>yIbDJLvals
z7NrFCp>D(qn7K`322t`!%9x&=6I^%Kz6}LNsaexadt8mri@8hyyRXyVDmA^dytX_N
znqS0yYsqN3RrYu&qz;qA!0Mv!Myfgz1r->?{`ioL)U4f$#|<qS;vw9UwZdq{JxuRU
zA=HWjj{stMV6XK(Ux);d4CyYff*{8CXA*ldL%acDU>G<Fnf@_vegfWUc*#~1i@foB
z3ooI^7;Rja!XyD9q<@{sS0G5$hFvZT)2nkK<@vV#8`iDoWLN-3o51b$0wq)>INUv+
z<dUT|wDy!4B-Ly&w!nx)`%}O&lHw&;{5ywP%N==7ww~YdF|7-WB)SjA*$eT#Ml~!F
zx-^A@gXKn|Ceh&k35hhg*7adZgRoCIHDALx$$-5&9G}bG{fi2gdMj(;oq$*O$B4M7
zx{XI3=Guhdzv=t5B$__L8}B-S(-866$ow=OoBy)L$b(L`tArG|`9fJ5*_F>J2NyZt
zT}^N#r^9bWp-A*%yk@K~CE!zVKV)5`%8af_%K`wX0ihZYnWMcO%N2^wd^Vu*ku&y-
zD>@`18}AYH9EPwss!RPUOnqIsmw2wI+Dc&+y3d-96F`(iz=~af9Z5^X-#2ftKHt=a
z{ZY!ue7@yv`{$PTf4cQCZ~r}}5VQ5s<U)4lI{5R$Qd5bSyh+)~S$S?c^`bk_9do|y
zc|(A7GxR;S^UO#l&V^t`mb0<KJxoJ@1f*$#Aq}e;tC@l+HRkZc^UV%wMNg76ofV<B
zF7JkdejIOLOlJa6DP9U3AbI?ZaI<JEwr7Rs_E)b_?D)8vMs*b#hsivUv^`ZRl8h>G
zpw9w0=#88OWFzs{i_tnNB8<c>e)|L)jT-AiGn{ELdK#CP9|XvMqa0+0b~BA5s77~n
z9#q3aJV3{%k9*XqD{6sY$B;bajYMK4L=D|h1Ve}aK%9nEW3Gcu1R^v)#Eryvdqei4
zD*{ONhFWVchZ`ABAg|!Hxe@2Ewx~vkh+$l$DOlHhT{niKlw+Us71O!4VJ;|8t0!)j
z6C{O${piWVZqy-*XcLd&B@tJ*;xGZZEGe_Vxs$h$-a$M%uOXjo&?<?6QF5;Di|aX<
zWqM`ifj5_==tvQiRL`@lSwF@!CN*1%Y>p_{D(MX1bJ^d?5|Spp6~0yALZ<YzIU|3$
zCLb4)XU!1NIZKKvEq=FEnBOh?5*0Tn*R6q|qg~SX$Li@6y#_Y+!sVA;KbA3lupTJa
zptgqYhB=85W<zB_ajMj6L7N%#yUCF}%cde&)bvprcmisLf<)`~#+iuttau7cA5bio
z=8=bR{dbq8Y#_9E-}3qa_V!s^L48Sc#ZvgNL)n-;n~mOHdTUTAc|0ddhG(y(U#yDq
zYN@m79ftxX9-hQ;BdlWeSvr4?h@(LhH!7&0uwsMnysW3ogTC8og-)1mK=^>5Pd8ea
zzi~7&J*POD`lxj|6M31MNfLnR<WXuy&BEh0>V^w>BP0Wk)1utrzPEY|(PO8w<erwl
znsa&_&S?J)Uf#SyRz9}Ws0mLuW_d>U<RmKT2bNv*cE`~7Tu4yVvptPt>~EgBhY`)2
zL4xaIcpr*?U;!Ry36=d#^Nco1yZS!X)m|}mOIaC&?~^YP)cB3VY$Dj~N%(kv)9Soi
zhbwQHmxc~~P@gNWBsXjjb1DXvkJmQ?)B4;URpsEISjmK2I1lJvzh<I&8JIeHq%<~5
zN0VczS#y*1NFe1^W$Dy2vcMG0iAe*4>GtyaYt^4Sn<)Q@(D#2z_={s9naux^qs||q
z3;#iM&7Y#b|3^CNVvIT0F{x~(-W|K{Zc|3SchGx>lWa#<qUb@1x>yS@e&_jwfa_Ez
zo}chM)&ga;X6kma>%5vW3cTh1maxI|6X_p>cs~h)-UBfl?_F?wh~emZ!SN45@P%j$
z$NXO$e-fUH{z-@t{hQ;3=s$G)8;*a~F$1IH--|x+IyXiJW6ZhLUgv~AjrkYR{{i8J
zj^M3-LdAa_8qFG=i_l<j_HR*f9&Fpz6qj^`Z8O0;_9lM}DJ#5)+`0R>?s*dSts2j@
zvF~d|nO&HR@wCf~<o)S-^S+__=4G*<fvTEt)zzneqL}cX2z4$9sWbi{Y{kI+!Z1R!
zb2Qw=IM8MK>;38GpB$+%99u3p{wrbr|CaCqh#9|hZ2l$s7srgIb0}?`he@dL55MSd
z4M#Bb<Zo{AmTX%Z%v?43YI8I4f)InbM}tE7|06;S$8j-a)%UuZ!AY7d?FwO!++t*b
z<~!L?QB`UE^3aSG4_%I)`=6MCZZ+BkYEyrDVCGQ&M0F?}SJgm|{*y7FNQnxcJTCOH
z>eBLmT(%HYax9Md35Yy7dyGjyE5B$$pt*A>Z*ozxVDjpQB!er7?)Qz0b>Ge|K}NUD
zPisFvMCcK|#w-FyV7d&HU5)ob777`{ao^Ku(6F{4Muk~qqMVUR6-f6n!6>v_rBiym
zZ>Th>ooQEnG>UtU6EJ1Z<n{5+@f^83?Fbeg-g7DD+tB$m9^M;zbUfuD_juw!dSA_M
zo`fMFqw|(PmDl9#x?Cg2P=(5V#(}&q5wj8=&9=@IMXq{xDN>hX!1u@cxsRhmzqV%1
zTPfw`=x=7re6J~5@%*rVa7Q%s0TSiXVb;aARmVjY_(*l2;Eh_>BHMl$fksV6e*(ql
zga@5_Jf2l>)F3d9gI__PETP-H%f7(^z}jQIB!AT@>#-p3xMP%$Rgu;*6Ux$2qf0}u
z&^8lg$tkf^jz{=ii*!6Cy4`YgZG~*+*2wN)d#!*^R-6|(RgqdJXcc?luFO6=yu*Z9
z0_zFZh8O99JX7m}X!jQmJ-+>rWijp;dF}dFma^}&$gbeaMv=C2#4PRH8)vlbo$Q>D
zBY$K*AzJ+@^3~_uF5Hn?<6a(01+Aox1k-%x-L)26ttnPk7T(q9+tN3^i^bBx)ZUmp
zVQscPkqX_Q_D-o1@0(+nVB(LHcD~p2)`7FN0U9UYuDm`zx@WSU`%N^5dl-d<6`-!?
z4AcOFZmO%pVvKIb1RZwj%Y9TXRR`Dpz)DL0xR0Ih2hPhWfv&b^$<04{^4Z|My!5LZ
zNA~}HWkn-h0@>&4UiJ}+FjZiZt^$Eg&le-frygkJs9S?k@#=Glw^|KBTHd&xYUK2o
z4*lW^4DQ)r2EYQX;4mg}0TYu!jg0feh3xSwMwjjabyys_9cE;#$}&T1w2Yw$B5JC|
z4p?jv4HXy#oGwk`2YqGCdk|wAesHa3crOZZHO5}stac{HcnFpV7WZTW+`3eqX?8Nh
z;7TTXHkQ_$Ls*=vjVqtgmlkEuW*8Eh%?%zRj91d3x`aono;}a>y&fgLaW*u*3|Xu5
zS_EdtDY7ZW1m57aTab*g<?wB3@l|@$CWE3zpWr(Aa69sDMOVLrs;^uHP!&1zbyrzR
zv2Frf<{Dpou#<dR-~wsW(?uYbtEkEJ6!kb62iruJ@t#f?BQN2^I1^8>-th}+2-*%^
zGq3=u%ggcCTk7hCrK__knD_Y;OhEdiL=Maxox&eI^q44Ne_n#3g2Gy#-UT8N@O+--
zC@hAE1RmD(OR<YL%SpYak1aG|&~Z?|>GEVrZU0~zTCWCJb8lLM`7;sE29u`uD^tGe
zDOJwqoPsscmk$l#K<)+dw>Dfn;5Z%O&V#gVPK)KC%1PuD$RV6ofJvuTuP?v1x<yJ0
z7Xl8Y&})p>nA?$is=u1%Z07S)*MgW6L0oU2pO%&*_flY5mEiSaCzs5egvIK~=<FNK
zTSJ4G?D5?)9;&D?jRY1kAVA4l%NmObNs4;p3U!RWp8P~;dbtSvEf>mlLle+StN12a
zDJ(t=dr<OB_nqgbxTH$c>sjd9PR;_MZ>@vc?4Js?vS>%@D3D3{Mzny1z%lPRk^`5@
zSVCldPn8^|88_7)x-hlyiK<Z=AYa)jKI<mS_0A3z`l@u^x>2J8XBs&7ApapUJd>lM
zG!=ydQi`hon2-=oCn1v=={)Yc(c9nWd>l-lAvemyAMQ1hfJoxY_WWWiBRgQo!C+!Y
z*P*xrqE*04vRojL)bpY-2K%j-f-9i2+MhD*t&B<wfhn0tDl#7*cw=mCi_HQaDtt@z
z`STha69a9dzUym%K^U{Mat2&c)jH8foVusy(ZcZLY}^?%5N4xARb8cCc_+#c5JP*|
zl_0_?!;us!BHXl*7Wil~ZrHD)An;*<rW_Jo9bn1mK^UngnRv-2BOYecz74t2kT5%!
z<L&L-rso}8(8}%2Bq0o4mfcNeRI&nL7TS9flbK6=uaz2TtIEkzY#J4BuGf{71^8P*
z8CNx6N+FYqIvT8u%{hJ<1-RAK%>ff@>{&?e^#nalK3J<QtD2UB&KPyVgV-H}I*C}7
zF)y{*y(;KPG%3_>oqrIF0z^`jd7G&FD7h)=Y^KvJWz7>G2+iJIF6h<dY|ohx;mEsz
z=~HQ6Jwlkr2?jDg^7;Oa_>%y~S{*;ams7uo$jh4`FO9BvQDyas;fKd*JX1jZ0Uj*Z
zzGX(W;i#&dPib{YsF5_la6y_f%mJ$_`U%0*6t+L@w8Y*`rkzJL(wHLLFRE_2ab+g4
z#N7{Xuz{5fbjf3KOkG>=>+kO{JXRbSk(F&(H)im5ZY}HOUk|T0&TEZWib{ZXv4~sq
zDuq2%lw{X2-ENl<n%%5H6DvE1FF6%qYGEcO3~?Ak6>*q8hT^mC;8^Y*#D1?Hinb+2
zOXk(hlu&{$CE1b_qXa-yS41f3soXPGE{zOIdsaEi5jQ~~$^z4g;54e|%GFiWjLyYi
z*@iBeKAkIW-;Vb(3o>8nIuJS-ISkvCi@;SWz$x!FDdvrb5*m}X9^P_&%_nP|Vqxxl
zBvUD_$tf1ZANqvaK#B`3%&!RIU7o)L;IYJZoF|jGzqn*BS?88j+^j!jyj68gQ`9RF
zT<55TxRK+R$hBw3!vnmeOaP>lsWS`m%=Na)A3>yn18qXjR8oO!CE~~zm8>0AgdJ~i
zq3M>^SRnyB6NR-FPWdBZM?J9>q1k%sD%=TB3fr_8<;&2}9O3ksB|c8&a>1JDbkwy5
z9BOg=vg;q}9)hS<LHe!B<Vl-f?Dap1SHB?WP(0+WA~yD}@<}g?lY0nbpH~#QBK~Re
zLjtDYppN%V=h9WCJH-ZW=oPsEJyockIDA(6717W;uFN5RSb-5o%!3M8J|HIJBX<lV
z-b+0g!u0DCsMj(eDty#M6dz?jLD>%+W<LyA14Hg-Y8A!Zx2*Zqbnh>H6n;12yZ9eb
zV29v9x5eS;tM6qJ@eev0J<f>oe_%PA$geN><NwEP7$&#We_**3o;=$K4xE)w=x;gG
zX~MMa-rE0x<#Bq{^ogZ3;|O#7njqVhoy_Jna_&9s**n+{^YfVWPvFlb5CYDH?<{J0
zrfBee>{LuXsNCr_YNZSXGpuD~YpsB#PK3a%7ig-X_iW)HjZP(S#2y}2q3wNX1A;}=
zRgJy*E_zkm)|{xt=olNIQjU$T;;oF2O%18xr8k3A(ud3MCj7ucWADVh8nfSO7v|F>
zq*QzE)Vze1Xe$OE`qsD`iy{ea`os)?r)3D=<6&|TuYV^Qa9C6x>agxdA)gY9s|i`e
z91wEK+z<7S87`*2mI^>|aM_wj9liGZegY?+$Fl)Myo!6#_i=z16>=kG^yMfgEbKvx
zpoy7*R7m_;5*Y3W0}*c2_qa9AS;t;ZW2L^|c(RkB3b_p;oVEZU^Ow{C`xX_byOV9P
zattaxz3F86)`W|Tg7C%>;PAbaRD3NP09vw-LXly2r>P*D#|#ea(pj>3%rRP((VP>y
zO98y{ElZ^c@E|FDAtASO!pYZ=msN*T1}{rzzbOh?CfXJ9qBJwk{CXAWMb##f1dr`#
z&IP$5Cwt#MekC`BU#qF4WS6f+5u(vPL=`Fo3t-zT_fdVFi_lQYmcRRuY4_y&*MOy`
z5FQ*{o*nUq_j)*s4|Z<aN9HXv-Jl+c(PS~D0f<(nh&?`{v&Dj2e+*5}NF0-?xs#H0
zRv9qc&%CHA6~EV+w|$)1`+fUP?bHy3y?)j!!VqqV%xBD|de92t44a}h5dtz*{8kJI
zvjj(cwUsT#*$<F8CdHgn+!K1RaD&8Fyu7eD{kr7h(=wO@;AVcSPB79r-iPSUy^(Lr
zm?XsSJIhFDPdffsNcuCs*pBt!4>mW3oPrJ&nbnOR3--N#f0p*&H}9ss6W)jzrZ-VF
zaT@n`@jBQKHC&D^_6B}1kbV_YAVPvUmt&%7MpUB1B`Bg~-}PC67%AmJ5HjtXl2?$J
znE7N*l-)YKGLIj`pu&rD%KTvc%U$EIeG!LZ=C26Nx9b@eYINQrj>V6>2j-aSeb~zf
z>l^2;HQ&gQi#2#Y{rjplrJ*eR>(=l~>fqR>883{q42(3Fa*224F{_WQ5gQV}uY?M>
zH?o^$yf#%Wa$SARh<B$7t$=3?#l5?Z3(_1PMgw|(U%lr!sb6)Cn~@<mSe=>~SHA1B
z_93c8ZQb*?iho`a2LG+H>AzL{2W1N|%J%<M_7B(m|EcWEpC12vw)q!6@ch#U&(3{N
z^4s#i`XI1=_1>R8IQ3b%@WG#9`il?Dx6@b_Qho-^3)8rc@h=g}KKmnL7gw#%{x)KT
zzvJ}(Sj5U6{f?NI{6E0!-fwuFUkRPN_WN&mk^X|$PuH^ihS&Mkd!B!G?N4}Jxc1Tc
z47hOZ#nt&-VEJ<{V0sqT9sZsS|LoeoOoq9cVdYbgzk}`A?>Y8w`TjY{eufOg_y3v~
z%zjR?-`Z~cY5T7$YV%jbGLf?2;mN)(<vTCSRuE>Kv$K`$&M4v7q>*eeCs+Xsi(uB=
zk(9>4O!sXm0|?|%f0(PMR=;ixvd)jo7qK$ZV}`+>+c@6|?B@GVcg@A4EioOH9&B59
z^iX+}nN_m+wTp+leH9<Zw6>aVdZg=W+Sa*&DdTWl-ORxs?xMi^4%uKA#XgE;f!tyo
zqI<pHj)jdy%IdA07pJg(`BQ6_z^!`Ly3p1an7;~jwqb{H7J^W?woVj6p^9BQpA6=^
zcQej{5{hur*8>2G3fPs@O`o07!s9+gC}eIHy--^s|J*;qsFOeZ;vSDKztvc2w}#q|
z!0AnI?(7?p0(hB5vQG~wmo}PR%E;Tqc*2IAcz8Kd<?&KV2mCaA3P*g_<AaA12V}5*
zV3~}U*snSAQbTz4(?hU??j1kl)(P1D?q>e->LXPu(JuYFCz!q%auS;nFtwW#&wgNy
zSdjLSHn;k;&cERFdcD$@xH+<(6JS%sBCU3F-i)jC%Szccmz>W6LHJNTTD2G`2Kp-X
z152r1Hz6ojyAKVFDES196J`bw(qfL_xaoT{U|Y$0+(qbCZiN8BwJOB$UA6(>n!M5$
zn&ach@X8$Eo?hOQ$kqDD_@!Y|TTa<xg7+gQcT=ZiF#Vg>A?5~}tT+p}y93YT3|6$Q
z;BXWZQX2Q$>`OuL9Pv=-+KO1BX4TOf9?nBYZ&g*yPtgzQQN~8y!xa)e>~-+F58U=L
z8-|=b9_RwALwNN+-ox<<+!^QWR_P%U#=@V4OXKXxzLl-w67O*>6*A39PRnP2g2jj9
z;LpA5XMD>B%u%=MuZPaFnE`UYNk16q+vF`c(|A5nsFVaJV0bx>eeXan;2SKT#2a6$
zaikXRKu4%lr+7TmCQM}YzJE0Tzpw4f{S4py+}`ypG3CpUylrBne)Lv%a$IKG5~J8M
zRM|xAX0M#9L`IpjQ9mod`BLR24#78XI_th)Gx`JwslBbEt}N2YBBP;NQtBh)gVaxr
z#jl3;YRk2vy=b{6loClrC`xmyHGq@YHZ~sf;YxY+0|rkXa(eSChHQ4Tqj;D5jSOWT
zquQ_bR--}Ig|#rPS_mLP_&tgWu*V(Zi#;w#<~9y?!1S@97OM}gn80)F$x5-Oda2}~
zG=%wq&&7#7?^Ns$@bb4%dlrtMIwe(mRx?q|BZ!>(s0vrMy;=;@M|DvwGB(`ek>X;x
z`n=7zW@C!KeZ@C73QbO2O0|J}-p_J_OP0EV;`suHP3)Q5vDf${PpQ{9&yqHkm~bQo
zP(O@n@ea;n%>7Iye`e9)F*I6w%hosLk+<kR4G!db+7LO|KspyZRON)bNsLsS^}k|v
zRVTG!R<wn5%Rr77s>MfxuWBQfJ`@j$vZw3=aH>!}q>4Q`Tc$JC?5zOeU@F9*Igx%7
zMer{264_LJx4@mqk}7m%sT{S3GqnMOw@N|>r<v%W)0;kxB-Z%+%UR%#*$17w)03D}
zmYlQVXe-wXb05K`fy2an@F8tt*{7yM*(x1R!5{$B1lBe&|1-Ert-S3p!H_SwA)-jD
zB8TS|dxo%nq!QfcwUok4h47+fa+&YCjF~DwA9|_?jS^<3;%D&VVhH3^NQ#T3+mn~O
z{qluLIb;I3T^VnBJyMUIj*~!!hX+qe=5d?OrP%<+hij$^h`!IFqu$rU>Jj>S270Cd
zeV)2W8+A5dL{Zn#Ti4eB*Cp=*)q;HE4qb=5H3Ap6G`cl(@jwL8{+@;vjPP*e(!>WC
z2xlR?E?fZ-#BP;ri%{T*RHfkS6o+KFhkDlTFhrN_;tuNqSKdYPi6TVzCLF18kH2V8
zClY-=r2^cqC4jR-@%)!{&6W=7tDlY!I6PUP5|{bJ12Dxj;vQMm!pR1M=BjlFK-iw!
z9e$XGAu8?$L;PMX3;8zysllsvKAySKxm9$z4Ua-$2KJ@ex-c8`)so~U-t0@TQhRn`
zo_4i9p_tISWsgWIori`?*eLB$&?)=X^;^l{d7S82&3LkiJ24OY>1z`E3ORgEm%`Ow
zZ7|6P=cSd+`Q&EeQ&o0n1Jxy7<e(@MYi+swrvpf;m~(4JO|-i~37TMlsTF8R0bC7N
z8-PGD8}AZ(-kAwG&Yo*eZTE%^atab4{RSep!=*i$+By63b3|<w5-YOnZe5BqS1+Fs
zomp?uff)fSy%9K~lPPF5V8W$H_(l;@3nyRYE?M59byqT-hs+Y%!5?i^2Nl4Ve}}q}
zl_NLtRb7(dvFbOM`sOy<Ds}I&`XO~y;6TJO#l!1L`8csilCwn*508}aQd8LxA34cd
z%VA-p{pR#OejM&|G&T4(+|Fou6zG!>+te}lItWZm;GYs65FVbZtKn!6AVCk5&eiP`
zVB-@Z%Xqvvl#lcL)-9Wl2On%T;|i)xwpPo&q@tlbZ+SxPx(lVS!fX||V|LyIclV&E
z>$nllwsJxVCipirl5lySi}*hCUx;JV8?~s!fhTWao7UGqWWJr+p@kYq>O?uSJPO;o
z)L@f1-j<EtN-=r&|F!qkVO4eAz8eMUbW_qL-Q8WAW^WoqltvmhjYuj=r*y-nyFp<K
z5(1J+D$=clNa<bR>+AP^-}jz#&pFR?|G4))5Ap0d$Cz`(Z?3i1Tyu=EnXNss?wj7j
zsh8&9-Kg0tI>S178=eU62udmjRWQbaEE3q9Nj6#Ia5VxLiGZ@eL~jxi3i}|;6%moZ
zrqB^=r;e%*8<qOaTwa=!C7csUYM90ah$lolsH`TvV4cs=VrZ+zj00(8H-&sgiRAX(
z;MM%6@9j4}ITkE1HC2Z1Q0kb9C6^vS@r4dDUozOT#Kji`sa$6!_Y-pxP^&P*vfr50
z58Ww6p;v#RWm+tX^65TuYuoqn2a?FmG?;zERKL9-NS0G9J~1zs6@4%N4r#F0cIGt7
zi9i#+v&XH0=LQh=+SP(~JyYLeECf`G!;N^!&U<Y#v4pYL3pdLYgyf<B;u!B6>0!!B
zEXXsH#mfiDqw>UAVHh}vQ<>eH=*xsZ&EzyJj#nul?85|hCt<EreJjh7=uwCI(XA$e
z^@_gs5Qhg!%_B@M#fh{7<<a*%2pJJ~G@)#P81orHr>A&5M#X3M>ePx1R9ft}N@<&W
zQx*7oIKo_WnV20!^o3MF4^GIORbI?6-UOSmfN5poO<}YLb`p{L7cY+n8(L4)TpNzm
zZ)&My?IV(dK#<2XeN5ua<meKilJfF)VvZC@iU5IDeb!3C**rxnFu^M;L@a)25-~v%
zWxQ90r2g#P31WXKjWjbVBdv+8m(7*)I*gdni<<T+tYH-FtbM9lUHU{S*W<#q+qSTA
zqx;HMJjas7U6xDKAKB@j`%~IDY`QgXrt7s%lx1hiLP>~+iZGXMd*UDFB!TqytkzRH
zsD$zsD+}_QHZUrPJ;y5Zl7sCC^yqtEBWbM&S^XF-n5l@|)RJEmAtogFOPk)hoHh+_
z-Alt@Nh0RcQ>o1yqlu_Rkp#yFtdid<-B6Upao5-Df;_>1=t(@0VUq|V0w6t8x#yH)
ztOM-EYZih-pIAdMnO%B;dhYn-<}6%IYGJh(Cb<qy6M2)8^)vRQD9LH}a-RTa9fb6O
zu#@Ytp!xRB2c}%DY~hVrTlT>#+EJubmE;A}vA1%pm@sskmaVQkc+oB(1tsO$9)Etj
zDe#`SSdg+Ip2<z&UXg6H;oE{jT_)3y-)Ji&n#E6O{4~0tYQ+^U`e9fl2s(u>WkwEK
zG|NM;l`jt4M`K@vM&XLQdN?Gdtv<K=x-Cts6N@R^x``JoYCIuNNBmz8Z=CPhI1in?
zI!Koc$XKI3tL17uu`eo6^kgfgrO@quu4_QHg%ItdHh8=)K($Kc+u{TkHhOK40uNu(
zWJRSM9quy?bw5iufBwvg)=9+EBP+Vi#Pg5{Ufb&D@BU6hN6;=9_Ih|Z3_2XlbZ1(8
zlubpYZ}=mUVcIh*dxEW8qjpDC*LEeIHQ16|RY@qOGcN9Pi{$a0*%Di)SiOl?)sgmg
z&y+m-LMT%~J4~uN6yjfApnEZl3+TlMaC=XJp&Lx4Y-uy4o4)S}(FFytF<C?XX>-bh
zg;b{J2ab#br2U&WIDLTKvZ_Y<1KrjEl@);vAYWIQ#1uhYm_IRfqZkK22dP7Vo<sxq
zXrfLYlQmgo6_P+snX5pF<F^YFW&<LFAh!nW3vjHoAxV!im%N!iNe1LEHz}6c+p)0N
zZjWOtoDZZ>Nf5Wc8Q|;I{D_rMolyqbMO{1+;fHFuc5J1G5f>P5Ftv1t%49tVDj26_
z_qWU!T>)j>eBp<CrfZKS$!S^EZs3^88dnb`8(7u>&el^AvnrtMrHr&WsFXix9G{rk
zB^EaRmc-^b!EzH4lo&$COefCDc3qU55oO?<(m-ysVBC}*_IB1`eQT2eqI#RGSZ-Nl
zeUY|o)1!gj2_{xLUdAbe*sIU7kOWtc4wA*uE54>@)SL;1#p{H;xRBYi;O}uz-=AG<
zm~}SrVPB8(x63a?&7<uIX(TDwE@5}2^71PNS0wwkM*5cN9e`ogQ~HBcEXFLK5^O_m
z16j7PhWkX2<DSG%bqwtNB=-*rcaKn~EsL|LB+cKPv~Ol`(3WM$Y6DR?y<Tmb9oGXV
zlEi{nRB@-0Zklj*v-@s#6S}Ov2{*Q@R%I;Mc67YYw!*+6EFMF&QE{B+2l4oxTb5kT
z=}`-X_BNMl2gMiWD#<1W)$cDQ9etb_ErSgem)fOhi56*CaYSm<7fFOxQCZO`(ni-l
z+IQO}iEOQ~DcFTgU`0ko$2Z)(UMm_xPmduBJpRlmZAXm^d8xLz-q)f}U-Y~^ZzO3P
zL4T7sim>2plSSuxc|6kVvaF6pjQ7$h#jnJ@1m4Hd=DGDR{>T=1<G=jcoe7A*YDOM&
zTaKH&lm2eL<vZE@DV(OokNUYj@#SYA46EVrfsC}^K#R}}RIGP&m?*2ov*j%fy1rJL
zA>lhCX~gSm4U(swh2-HP=<PmcPKY}-ahfp*(eF(aI{mfrTcr*6<CHsx<U1zCH<pXF
zfJ)4hbEZ|uGt@KQao<m6R??f2#})>3>nUQH6m)IoufGzWNne;n?Vd~5L`+kszYEjt
zc2j%Y3=x}O-Rz(sHK@uYv-ql9v~lFoFEnWehC*&uYeHl`O!;$_^iOH|Ocg%of5^%O
zm+eJYd|v0!FlU!of#94#gfTi+*w4iQXN#X!t>nkoroWlMg2Pk@L$~+VsUHVEwZah$
znw*06i=dx~T&vEe-D<sc!K=(vt*Awb$-@%LkR#N}%`d52VPN;cZk(z1>saNiW2?{c
zEC-vT->$6Y6JuSz{Gjb8V}*AKkfZfx-A%2sPO|EfEYQl}^O}^6XM2_@2z2)~OhjoR
zr06B)uueO%-}LbZCCLrs(!^*bIfk}G%V2zTy&fdllGRWK;>GBsZnuSwHUbMvLJDm!
zUlXN?%;d``p^UvKElh>x`ClVZdAUO71||2XFhHcJ3JS{3uzs_XB=}@$XeRd(5(@!B
z=|?ivlM1RpVN}DljpHrKzCZ!7zAp|6!p7=@3f4B81BL%wALzU3_&*g9!eOuD|JQdF
zujg*HYRe)`INYYjdh7GFbrJCx-;DxrI>yU{Mlx(KYQ>y!MsZ0WhLpm;-_U<I4&$^}
zgxGF?(y?eutYy4nedH@90=UmVR8pfWZx~wRmCpvc=6~(mXwr9lAu@Vq>9IhwDV6fw
z`5pfC{KJ1HmHI`R{l7)pe>v)J($vdQQLjejy%YfR4*@Pm{Y`oZ2q5&E)bUSJs}i<t
zSe=#tF-c!xiMHurlxv}G3C^`VQBASk7eD9mI9#CqN9KS(nJ*QUa`{8iY!RI^k1Iv5
z=J~1U73nXVT#~N;l>Qg#-`eCKN4+wi&~H-5zgYS&0$5%t`X7;A+N9!2*8i0B%72%n
z|M{rDNiQw%AIiG%%L4zoT7URY$M4?>u=^+0E-mnj^j~HD?E-4RTL3El5McD|9~^t-
zt)Hm4^46cEZJyTC2VpdgS{jSr^X+0*FI+y0Nkit9#%YF57e9IgCWTvU4%Mw;Nk7O+
z_q_LuboUQZ$Zyg_fYdB&#Rd`PGGgSM%oLSG=a#Fgy+xpTj3qWm)tM<U;Yl)MOL$x(
z*Zel|1SH~}QZi$H^1@UCFJ`Rmh(hfLNO&va(del@d4Pc*aJ;1Slpdq&hNbg`6R+tG
zfx_c^6*<u&V??3!Fw_D(nHyf{k9}i`q|=X(T`U*uo;2zYh0h{_bpZB|Uu)WNGTmla
zAaI&WIHEP`H|ZZ!UC~CA34O?5xoGuu%J8BP&>a-;)IU_?y`fm7fX4CYQ9|Z73g8g*
z#GimP`7Ltw8%d+a^F`&I@o^{Masaq!W20U2Z%C<(8_$=nNGnXFT=OnTrGH;gOiE6F
z8KU}q^E<aZ#27fP%SeN5XmdI=4;KG$VQKx^zM;3I+j@t%j-r`aml$w8ti{kVVWQ%C
z!&fPj$o{cU>jPmQwab4wut|vG;d_g-lxn-lyF))fDW?l<GF8VS$`@5__wV$lY+i;n
z?;*U&2d9n%+H_@}?O!vOb4b*w&`DyqlW5B(SufNZ4tw2>GU{f93?oYlCB<aMq!VG&
zSw|0a(heeutJh)$3a)%_^4=Igzpc8$oY<Di%M$Hl9+b_cR`NuMNTE+^fAsBs1Z|jS
z=2x&Uo|Q{|sQcYuEG_y7IuA1_&g1@=k86p9iDd(1bZn?>tPFR&z!oYS5jc=6Ghb1-
zB_@NbsLO%XUN4?hQ8cS=H5{Ma5?S6>zONE#|A5J!;_-5nq}Qj3MEYQe#R!+4D!Fmw
z$ZcRV_I#o&i|vRzO~0RP#OPgTKc`3HCIsB^_^aym4{pZg9<BrFc5lWe{3FkXRvT#O
zI}cjIU0EDSX1D_FboSZzgwtUD&oWovkR`F+m~mMm!XnKa8C^}Hp=18k^nBtxG&^!S
z1Y@V=woGq)@iNo7?lvo1pqzOR*D`XHGG?6ec_5?iB>pzgaS^MdS0QdXs@0w!Jgu%a
zt6ouvve((62uh$W=1x&m<Fi(p5rWq9xryyL!B}+R@p2KS5;LmL1W)~}yLyGG4K~V(
zct&$Y&iaLYFo5y~M>sMpPDp*lk#RQK%w;PnopuS6a^ehiHIshPt!o<-PNJR(9a08L
z`67d>F<iaDZ$?^PA^zojwseZXzZ~Pn+Wt2s4+WgCQFiZJihLh7A5r=G^iE#w!aI6S
zFNXlPPYd<qFnYP@@XzrsM}gQsKm%vo%Z2CLG1M-bc_U_O$oCbd(`_a>2w~$$kL@&L
zdGI5Z=XafM%adb@?`K3SkxyX_KoHE4S$X6I#pc3%+SU0)pYaBQmPKek9%A^<T9?->
zI5i}f1$5sk9pH!zRl+2Uby1oUQbl(KEmIrhy9@)_q1sJ7?_?KE>dtkQ-H4VqY0|#C
z$}nSWNzsSRMq%>V%Y&*}wrF(MnMJRAKTJ={FzI1uIoVhH!fz_)M6T`8rde&co3G2t
z#gg+XvqoHl8$2SXQ(s|1G*#xxElj#7o-7AE!>jfdds;f9{7A@nkkic90G7r$IVh?7
zMNzZxlVZ+xSsIxL8>Tqz8bx*`#c*yy1gX`yy)0w+v6_amiM+mpffm+k7@M+0$rvMp
zo!@6KN+K6S>j!w%z>jzS<H$DKO-XUb{PsCw$I{C5kxEkRv>e0l>Zwb5Dp5)T2%~j8
z89u%f{m3ZAlG4MY#x^}u99_e*%@n8uf|VFMI+r!fS}rt+Af){x8e!(g>LG}!ow0Ty
zE$c&HOmZn@P2U-a-NwhPB8(!0jskDRT{b<CDkgej5a#?_e5}by&pYqBFSHQO!dT!T
zqBs-7$U>+xWn|GEa2^jqAv@g5Chl0U5>apWopys&vyA68N2!LDz+_N&Dy~GhD3;d3
z4!g~wQ@@Bs5|3Ed{`!$({Gtz{c5KdWrA4=((%v#VJTw@E(X&>@JHp7}X%yZIW7{f0
zf?Iw)or$^{#?WEp?cNW<>ZyIE3LFV+$r!XtWL-1NlW?$WW7>0rNJ6m(`=exUkhYrk
z4D{S>v^S&m$=R=$p9_#~Nlq03yB8a>20H4l0@|8kl?)mFw0KGvv-1r&dI^SHIwpxw
z<5nZ6rfe4xm>~1JsRWf{v+*X&VFgJnMvV5%;fhN|<Z}vQ@GB_lHiS3xg$O1s)d}i}
z{NW9BLm_lS2~x9H>5~DiKJ^X5YME9-(TeJEy$Q_1`c^W=6TIKqboJprUd~kL`m-f_
z>!V`3l`%R<6JpL1Ek!2D9?t{aks*$R;ncXn(6akJsw1?tJVW28?!KqG`=#g8w9l9?
zi@7a#(_o<$PPG|ZoPmgW#w#GEf|Ec6#KeRK!KB%Qn7fBhric$kJIcDHxuy5WZLw^u
z%&$&FTm#?Ffq;yTrwqNgnr0=R@Rn}5=jIT_GYInEmTY%hvVFEhcj6?5)H1fpjcnXe
zzE+7wpNQ})o}p#7jC#%Z%ya)Jniguh>De2vQOf=xiLwVNsU79;nAw_gSM&S_X!9u3
zUCUu{5AT7oQ|p298t=UAri?y4mR44gDi$G7gHI?L{Z#il&d{SN+(%RRH(b5UVY*X5
zexoDQ`njIcTQf|zU!;ZSLX4{|q6o`|RWVUy6_{wF)mKgh9f9Ii#F|YVP;;lxm9Hgh
zU;;I&k^Y>7<S0r~Wcs6hqOaxA!W~|2q5A_1!|jEx!PpnAp_IV#G*Pa%V?D|my4uYI
z_sQipNS>(@it1!y^^M3|Mx>tNh;DYy61tW&u~jloLZOj}w?1@_<F6%FDy*R=u7R>-
z_8earbMy2|)~?B$fz3a0WqGJ1It}mR?Tk>5h7A>q6mxtmt<g*RQpT|8ZdWMt*hdKC
zb<%o6UbeR325AAEpe}U%2)baED?1%ah^c`3C`>JMDD>hp>b_Mtym(S9urj=8QNyoc
zNEzZPso=>1bEP>;#6zF@<Y^Xx5ES;VtoH<iL-<G3YxwSM?L4p{TamQrSd_GLsWneR
zunF8-w`oEav*wf>DBPptEmM7jtwM>AxwCC~JtDEbH>SL4fm^t3F1a&s4HY&9HTlT<
zk|D-<k==I1LEMtaH3($Uq}043XBnkx3`T*(yJ1L#l5*fx``X;FOuIvFs6Jp^b1i{f
zr=J%S%y1Eu(#mt!GX(hI|A}(`zxjy$AiCNyZVGR^dzPV`!qPLOji(XaQ2gV+GUE8W
zrWMyHnEwHqx+?A$eiae-4@qGEixF`V@QCd=9~%2Nf#4U=Pj4^cxf!;nPb^PgRBYj0
zM#Ob_A=0ozfppj(qZSVGIbkK?qEI${(paToAr+uh&-#R|R)#{g(x@iJ`mX<p)N2Bu
zxUKnH%2y(_rD-b!n2Fa<q%TaPTtS~;2~(>-Ku7*x`gT|8MAarIjedYos8%=2KLke~
z?AP60;Wp4KI!E<NHM{(qos`ivaA&w&;~R!kx!S&Mpb*l+M{RoYxs(Rprz7tE+;7u0
z=)kds7Uur^qb38(6M?{{#f90~XOx9|q$h!@RQt=^`{(iED+is$4-w5>17_6kt!Snh
zI?c{36CYZ}^m3=i_|)tPlZg%@W*Ljmz{i8l6Z|yiVh_pptHrt+l?_FENi$lPw1=fX
zZ=BnBY#McPW%v36E5#B&?5GJ~Yni<^-u^PX*UvEutJ%aN4ax?8#2Up$M{AP;KR9al
zCMt|g42I<gp)0gU*&y1;XdjWlWWlb5tZTQYMXu-ZXa=m^rk5b8f5SF8UK-`GuSow|
zRE(aNn23aU@Yz0r@P$^%<D0dZE2?NS6tF=AvFB5T_#8qa0-ncpgU|LDj|Pu!OR$9D
zzIMguL#ma9%IXQ22iHoQe3?M0H3zB;)9`!5;wgfvCQkUB(GJu-;mU^hzgh<j{||Dw
z0`GZa+}WwWe5cILGMU$X{1!t+PId(Z`fMgR7!DU1YK?f-)QaUzQeTut@yw5qDi5M$
zA^wOt&&+8?Bw&6}`>PVS3{ohM<-WLzf}NyGEM@qB(hblSdZ@VV;;|Q}{4`m|b@&K!
zch~r7EwAql@)sD>)7`H9(IM3eb}}-MYs$2syAlku()evH_SAhn>lBTFG+u5Kl5tPp
z@(gWZ0A(dGb+uT;yB^RIxMVzfwCpY482yVt*pX6dccsLSfi%Y=J@IjJR2UN)O|Z;s
zpJ^Srx+Ho*YK;xFMe5&j&eyw~wI%0nKfCiN+==&9p(8k_jI95<h?6~41@MKoF5{!s
z-;ulaZAC1ljXDA~P38*oi_9E(48wDB+~dM<SkT<|D64S4j$Y~*ajD}@e#HD-1HR?G
zLmYlLom)~zELyRk{%FR6LRJ~o6%Y<-sGrS0+2)Fm++EjhQ&P_^e$t#1!g`w_OHrbi
zflrLLB<c)nnf5;2VP3QdgSt*3iyR1|o7qL{1Og1F{M&GU+J$Fm%`#S#PmGH_nw3XC
zl$|CN#ar#SU4<W<|2CUXbM#1;au*pz_{Uux@;*9LYBQu?H&7w-m!;Z%%ZjrAB{wv}
zn7xi-NPc}_rCx0D<M)4WUGre1yoO&{JF+n{vM-KUPVNUNA?5xa4BB6hGCgAAB_kDl
zlk&b53M<uJp3e^*DJ6H;uG|QK%5Q$;C#d*t4>>3f87PeE7PV-fw3edY<#^rDP^+b2
z<Cioui@Q1vb<VQOi!9<NnnX><o|edA&y8!7Izvj<yBxP=U3t09c5%;3@(P1E)2cuL
zBQ;!30o!m4g4Cs`>zHyfFRw3v#9F9aK70Lp5k!_fk(QpwE&w?HZpfv$ceruy@X-O^
zFYNk8)av+>0KV^-JHofjJLKRGS2h(=2EH-@KJshX{0Z)8>SOnx;2svGa+&_mr@r`e
z>agG1VB;+QF*W>8*r5mhb_G%dkor<^fcsBB{$7uNa{mIY&Tr7>0iX?kyA?k3Q?x6f
zFp=i`TI~>Ewf}=v&VC#E_zy$lzKCX|qEKyrsBv|!8ApPI-P@*wZQa)XTZd2EbtBX`
zoe5J0mSk8TF&V7_<~JFSJ05rcC@cSalN-Hn@&_o-t7?u-TB!KlKNQBXH~`m?SS;kw
z?79XF9*<9`67b6KOB67XFH>JWtP0TFd)jv@AW)KNubV<Iq<o5pIXHWDZmOYomJ()V
zA~OI6w-^jmUC`Xtm>9)Bo4n9!5$sH{j0#J5$IkSHjffzWN#c>(>rK~}`w`59eY^Wy
zZ-ztuILIv?VdL_07kIiG1~yLlq`vtW<x}XQj1KRf;WcJQWVgM9!$TDYl*Hz;tMF;Z
z?^w!VDVSqdMeUi62_>j<DkMEbQA1&UFFK!*z3F#r<I~J{4K#&AwVnbMvr<bP$_2f*
z0k_=g&~!6yzNJEkWHBlUE7tn1oD#Tp7``@@$Tyw%0-oq$9qHB__Kyjpo4YeuX-CY2
z#`Y=o^^?GF9&XgBt*kW(Ms%myNoDW2GR3t>Qs0aqhW{rIt(w?fMBi*I*&+q0o_HmO
zT+}Y?tXBJnnm?wF>*}eLBA$MC3!a>6&8lU3P<vO>OXU`YQ&P=VTKc4Kl^~2*&{JTu
zOgC)Qf2_dN4~r2uICGZvLFiH?lH;VDhK}^pjRQj|Iqp~ZH%DG&8dx`|>D(#fv^+|)
z_q$+DdsUy8=^9Gy=%X_$OT>&lr;6^r&TedfrjXKlYdCMtVX?Lcp8NhrGG_xqWmR9l
z7Gp-GyOuwLlBb{A0liQ0+oSd={Me^ouD@!yH3%~kpst;6gT9@w04uo|IL4=yf>ZVL
zNQsi=R9VG|m>}rraQW>yL9U$f8Ipd1CNbl|iUC9VNQK#Vj87r?Gjs%n4+m6$n(DgH
zX9>^lhRgnx8}|=Wu+@IKskaGnz{M4F{JIt8*t_UlnpKnp!2owDaPp$<>bx9)Jf=8q
zCB=o7Rq0ETx)ipJnXs`{h0hwMB4G9E^=ux*&V~}xs|Pjj7y?(x-G}a;V7wp@PV3qy
z45Wn@D0~@vmG&GeupvXFlK23vjG|9d6TKt;^HfW`anhGtC!8~_X)8rKDD_2bNxhK|
z<en2%1i=l4IU%Yu=(HB$)_}B1+^;UBeFiv=r{djAQSzc^)olF{+oX3vqpCuQKHF{e
z{s=ocjdC<5sFUXKXY6$&dqIUcg9#YlhmY~os#=OX`2rDcB-V2YO`r8XE9r$~;39E^
ze}Du-mg&!B7Jiz1DRE&>$V|0T&wz=_bm~k%?{Hrk9QW*DJGi$X3BgQ3<Ikh%wU6FE
z>MuJb{Pc*ExVLn;+1^S3nVN454lDgYWo!^%7mM|kcl5d_4&T~O{hfbKaY#jK1~+Rm
zqOV+ZE~?6DRl8Mr1MPq!_lc6C5(2Lew1*<&G`Kc<)LRxIKYt`RkT=6xkS4NSt18Q(
zRuZL#LBvK-$dZ)gk9rN0jPcnnPUcVNQT(*EgrwU$vC<8sh9`4I>7-}1W9UTbGbGy*
z&utl>PL~6RZ-z575<yEjUpZ}V7v(N$^tN~zW0jV$SRtN6@{Mmw$kF@M)UoL5{Q!9z
zOducr6(1=c6T&U*^{d%6gR=5a*PQqUPr*1g%Us?f-l1FyRW==12xDeHdSZul$`i3V
zs82BG?GYzfTA*M3W*XXXk{WoGGtolz8y178_-AEyoQA(hE(MS8;MTQsP~22Wz}i}0
zM8NDsG1z5vKu5KA=qGj6RLx#RDN;>!#;@G3C=go7E3}|c%3^qP;4`hOUq}FrIx~Zl
zxBUfhz8bg(dJ9K8Gm{!R_IdTwlS&AM#I1OrllP5NLGU0=F|4kA@fkL!8du#7r)DiR
z?F8HCqAbIE@%)t5c?zMdOjc|R<TId~%NDJ`GH3X?%mD!6-<oYMbSp`MxsAz`ZZUCs
zYMio+1)>Bc_NE^1u4y63N4oH%>^?QtrE8h)3BvEDiiI)Fq}kaTXn~<1aH^Yrr4Je#
z`hy4iRI^B&*FVvG={~X-1d{R)^P_7=8dl*KlAVZq^IC4<#v;T7$|4Q9I<bZ9w%3B>
znr>z;A|q=~hS0*J)SmN9n%m|)=Y^CEEa^s%sj8x*4<OwRZ2xhc(8cJP9#$D_!^=XG
zw8zcqa+<jEMA?f`TrjPIdB8Z4vcp1&*JMS-gWSrdjH|RiFJ$eYc?g+=N$!e!>VS*+
z9!%wNZ10@o#B^%vWyEDY#cBDA$P#{4p78u}!nY<RY#a8w;n8`Of!6O^l542@uvT9M
zDU-lZbc{rMQ9eZJ88_*eRwdY%Nh&sm1r+Jd7ZsJsEk{{$=_OtR3n_oN*0Jm~hZOSp
z%O#ayGmv{>J<^G!)Z40{xG)ESC^!)8A3EvnXH5$zc$`$5TT7IbFyNXp^<ryB-hQgA
zK&NSJ`hHnaVYHpyI}Rwh5PY0*UFX0Dc%u#(4QYGWJx9*))7Y2SCEh?<@7DQO9a^4K
zx=>xCqWDqVRW)@4nC+EoPGYIMPE#$I^u_)Ad%mJRkUWJREykyb3W+%-eLe)BmD5y{
zOS}O&SbmLUuBzx!JN0(FD6hB?JFF4mVL<4jXVrLK9b4a5U$O(#KFR@I$D$>=K?8r+
zgWXwJP}OW?yY7?jn5Y4zf@Lgl*$=Zu_M^~IFgr+xW+{<#|5ec5`7%MpEy(6$s}|^}
zS;a)2<w0`4-6ri!6ok%|)zYnCTO}#h#mngfxo33Y*g5a`JMv9J@76n{@)|fLc{;rA
zHNmg~v{gME_Sj<ExIQGyUrmi|E`q3ZoVxphw<x#l5%B@S^AI~`_*0NV0gK|WHD4|r
zI^)M;hNNXDnh#E1ZC2iGm1&Sx>?lR7!s2JWPvpQn74e$F)Gf;vHJ5iqnm_M~)L-n8
z^15X}nGaoi=63B-m6FoJXKci(VYHzoxF7eIxRXjfdw}d|AMEQF;%i5sT9&!NfwsD;
zu+juNBf?Tm7kAZee5AO^h#}9w_rKf|dt1vM--;YB8GAgcEN=^xlhU1la(Oro<Tg?6
z_m+>dpfTInp)qEz;m<pzL;HKmdWiPkmQJGKa?05b(aj}QS_Hw5**`w3LKmS3Cv(mM
z#_qYg-^Ky=+wltid5ce;Q{hXJoDkefwn$GT(I?(sn^=qqtO1x3e?5p<(x%ktTRviA
zgTw3T0B3`TK`DFaMq$At*ZymWd|1e05>-b*PB0IT3i6-V%)`6}>*A^j{Q>NGjl4YU
z!k99Q0~T-yR^(e$KSgCnTJF7SKKGVoeX^~n&x+L11uRjRlE#hBnAd-RNGbYquS)6s
znRSSa{9o$v?F&9!&)ho3iWx#2ePA&7{$$~WOUe06s(%khPwE4qAD|k=<p4GKuKT8c
zo7ekK`%;B%XIhB8#ml1<dZ~*{TjLpd(*7l#7g`HteMhH{9)H?r`QAi{FXlrv4?JJ(
z<L$m#Zxgqg|GAwrCoUdNdil-Q1(0MQ!`l0`+5HU{t_sqb#q+$tWOA&rj>sg1&O71e
z_GH9#!(GQu>jkz_Z>uXMekZx1R%xx^t?B<rqN<-*Na$vVOknj6_-$WDQm$LvsHU75
zT@&4hTa5w7ThR)+3=Wvr;pmF5+>h*I+fbKK*q&P1_5T3rd!a|Cm}x_!?*{TO2!=jB
zxaBR|y%O7BD~*a1z_e-qh)x{YQ|zUbsCkQ>Un(W?zkQ})^c(vR5J&k#SMIgsQ~e#@
z{ED<R&$qJ`U+!3jx%VY9v#YzYV`*#AUrQAJh?|GczUl(~Zt(sGh>3YE`9$c2WNG~u
zrPA2GvVpSFjYZb8&kTh!k62Hml{@p0_v&otF)0lo;)<<rOEwt<8fs*6+G*E#S+2ov
zwx1I|+LN|CG*;R|vwHAp)JU_`sdy67kn4Itn=r(&MuwXCy7G&jo5!;>Je_=kkO0kO
zU4x02vfvN>2TxLL&eV7AvM;VVQ;`n7vw@UGo>wqUEaYPg*6ZJ*%@&?LSU<^W_owP~
zb{73mH~7}#Ij6oOMe+8LuJ&t(;CZvpY`0-uwi)*=nbdT|<;g)eBbKsNUA`x1$vT!4
zH}u%XTS$soA-Y9h@&H+lYB4Cn`{$^+Cv3Ffqxx+IskrQu>D(Et;?(!-Px^hvvJY+2
zSHJenXV~~{4La5utXJOde~_1#t=XOSoYPV+NuDw<)_pUC0Yzffs~A2tMP?ZwTr3dA
z2p$}tNFKJ2w6hF@J;he%Q{kmUy&0T4aG~cbtYu)-8l_`jWcDIU@EQ)mO;#f#L)v9_
z+k{wOrX@$HTlrY4gOavk9BxtMI(ilis)kF*rXivW9j-vb0IFN_!Cml<7u)TJ)CD?I
z_4JF26vldtn1E_@A0v%l^;V=VM|<W!H&R(Bz|G8WcR~xd&jZQ`jL9pjKLQivDBt4K
z98jkH(DB)6#zy7N5vN3octB^-5e{pXl*dDs`*Pp09xrP{qPN#j5Ba`)&&jyMxY(If
zK51YQz;sWaseFKC3|y*L!l|mwDmUSEH(m^7CvBacy)&$`Fd?IqG<5tmttg*7BawSP
zP=Yr$|Lr&0>ciP@7Wa!M-f=3LXrOrp==nP2#w!rm4cwC>0u6(e`soxbf!(Ab>-buO
zg^j%ilJyk#A{BK{+Qf23A{h||+iR+zHSz%0zBhuFE-#doU9*<ciKR0q(29&8Q|gZ_
zWMx)Ny0G;3{9kz7;?{z)a8YP**rVIB(gx%6(T;HUgJNg{8Z8jIpNkbo<xOyI1tt-`
zdu36oIo6kn8Z1{^@dM<WdES;?gh#ny?B{9KKw^taWv}~WF;2}`pU5q9fSF7pwU@f@
zjR-snzi1mKSTA%Y$pHqQksV^t<nljmXSOFa#5J!aomeeaCS>(K1vlDSHqy(p6d6xJ
z>y<_njTk6u`o3CQRHz}Nm<5Zn#+l+XJz#^d*j`8-Q6f)8l8vpU6N3nY`#Q34yM8~<
z&@r)wO-!md--3cpV0uIVw5+Mju*r`v*$I@`%Pr2yeFQ$NA1xxgC)2GR`V2M38@(?@
z*6zt3P^{D{KYkT)leNSc^2At)sQqSC4hXbFcRU>tpEjweY-V7lr9tVbS9l{278AzR
z4Toma&V8%?yaXBl;M+l*2A&s1a%Y59IorlR#3t)!rc6Jf9(W@Gn^JS63h?aCip1F5
zrJ(lY)4o>xUUL8*qkP?cY1Cy_Wp>nviZeKzQNO#?Zj8vKB+;$p<w%#R_*ZDqSAC1+
z=y7@6E(5ZSdm8t0G<Z|-L+`0FvFu+{J)Nicn%Q9QODON17hRj7@?4HIF06PDhCv%l
zqnPS&V2jJGV;R}oMn%vh&n!uaq4ZMWIr9U;VB+?3j^*W+uH3_Uq3m=`q;-K}o+qoW
zF6s}^?ZTV&nqnuQwW^L|qwoe3WB`+EUVdJA%{WX}At?`a0G5|nzM{&;%*41y8;@OV
zgv6zA0r!`V*Yiahd+5r66F%Qn9k{=;@azYuU|F&-)xj%^a)VRWZ0(&IF0r1{0D*o%
zTs`6Ob5Ic67j=&c-MC85F72$Z12&CdYDUYG=T-gYpuE5;FQS|9mXV0YC)Fh>w|=H-
z-dxD{4r1qthEjijEN*Nqe{i|eOmD#P{7VL$H-dwK!PSOrZ0y@4kyzNz;y4M1hLUbe
z=4YU;eJm(L4PD~8bzko7`}!lRqNe99Tx$Z8!mNy8tcs?!8Rd!2C?ef5ADYqY`0*rT
zR17_qi3yB%Vxjb9*PQgoL1ZY33h|&hblN~jAk~Jk38!K_I}=Oe1S@;|b0b~8j0drp
z-$Y*a@;-C&qu_a`gY00XBFG~Vlh?e-Cj`UnOQBp<6=gDE?PnGb+155*udvG$fQ`On
zyRD~?d!53KeJ?TKAY#*+S}m+Jax?X&i3XM2-RH!FXwSA)LJ~{-K6W~u$=h{k84#O&
z8c{QmP+5Hy&$yf~`d%NuThnVhBgW_0DB{LSuE5a32Gv;WpsM6OA?ZY2m6`XVN^jAz
zrX|v4mU1qLPjv^~+s774E54|@7NGD<EEn2=mahl%p?nC5J$-{T35hT^PCIx0Vzj97
zw$JJyr(q|FNNe#vIWwHM<BWbxA7WeaW!%Irck`YVSb;&xahS2tkM^ibG$7m}XI&FK
zzCL??oUfzIxLI((vXq~H;5|;gFs-9G(@7{YmzfwHMDBOM|1*Dp)?e}m6#Vz|2fVL9
zUJP)z1Quv;p!0(LHniu=(`LW21g@Oovmbu*`mjTfaIj%wizkaPWsE;Lk|zW<KR)Zl
z2d;M(v#-v9h-guQ(2v@Yk5;U0Vz17xx)@E*kz8%zp}TCqVH}=*Pz`+9seQRQnxQA>
zJBWVNs@YMGW-u1gP7cjF!`+Sll%M!S<$BI3W=S@>HRD=-{kp34Bb%%!`<Pu{u6CFb
z2eqpLN}FxO8pHUt(#Ke){g_+VF5>!8ZDHda7;l9V^ZU@v4w$P7o5r7)+Nj%6K3|cN
z9rbOPxzO>^B*?OlV|L^32s~}ooT{@lXfccWwBGvY^ZnPtM818Lncodm?|*!ON|Btf
z_rfD1Mk=p475PEVS4{>uJ^AIoAFzH)8co=xw^#bxGEnBOU+r!|-V6OZb}5YR21=o@
zR~V}+y27{Rcw?)xLt?#x`1`0+>=)sM)!NFA^<QqSsmWs+@Z&=_Qicj4Y`qXSA1(Hy
zuQ3&EC8lTOYMa&bIr)W!u*l#nvLv>~<ohzJMq3$fDX6n7gs6!d`9<n{l$E?haxmZJ
z`)C`P`3_E-q?*0Drxi0ojk~!X+g9AANDsZ%Cmwlmu*A(`idRaEiTLNuNW|*r$D+|r
zQZ_a=>C+S<TlW+2m3JsiytG^4d?}Moqeo4i*v0oY0`w``Cp;#3Zu#Uy8mfSL>(Q9K
zZZEZTJ9rW+U=*jtXQ}7}r@p%dOI%6-Jwq@28U=`LCJu<*WIHq48a7^&*ja`<Qxu(C
zNqw&f!}_V$%QzI(`TpkJK=Q*KV!f}C2fT6a+6v8xdyMWbNxQl)&*`&6vV!_33pX$F
zw=Q+*8S|}}k-E~wjJXsU(DxuB7k+6SGjUQ?V4BmbX*{yZaF6L~sl$Gg&i~0@E<6|B
z=S%vV^f!O?2R913b+bQ7|Nm6~O#@!xu4n8|+?fHm`{}EnkRbgH2@e1gKYeuxQTJ_t
zbp8srQUL!iUtOYSqy`{`U-Fxt0sOyw^>3!?PPr^&BK@n33B{bZQ3*GqsJP<`)|$B@
zE{xc|SO7fwgc$tD4oy{4W<Z&t|G$0^P|fxS$TecubS;wroV4S-g-+poc`|UVEgtou
z$x-SB*V#oO)b+6kNnr6xS)zJ<1#OVB;<vBEW)F6hLiT71Pnt=raX$V|7t<!rk6m};
z=Sz^lbo0Y7Zsu=KR`y@;GbC(s@X1x_ZxZ;gc7m5KF!e|G&|kMWydo&sso&K1a<AK^
z(e-s!1Bt~&?JC~&Th7XByf3L<nk+x6b1F96?m%nt?X8*juz5XWM;tpg+M(sy%Ye1P
zr%8wXZ&alY){dm;@61SZqCC%9P#w?EB9{)8N;w>U3NuVlzNY8fc#T1YVR&f^|1A4W
z@<>8ES%BnDY1hGgUH8&PlcPR<)=m~FjmCL5&^5j9mHK~Kuhg+m#Q4HYIgRA31ea7g
zeDdze?)0MT;fFVr*?#TQSKYh+w);eU_++2w)NfWVWUb96?De?QpJXRt*$g&)$c?=}
z&DUmM?SK5>bMvY5DxOxy!UJGz<i_#eb$gBcofo=B;yy?z(P;Vo8qe|1=^ygcEdw!k
zf9IdLx@C*Q<Hj=d5%>spOwoLUp_E^vw`qXz0k)LVNZj8Tf#ZbY|Lk<)L)qSc3$p$#
zyaUU#`PIMEe-H1#W`XK&yZ=v{XHUpKbIqms?>us~o-&ljhiA&AFVQD$1fb;n50u;v
zSX_8;h6GTu_U}+a{U3Le_-C#H>-BFC`2R%jzNd9j&0t{_hT`f>AzQc$5nDcEmgAkX
zlbUR1pStBe9ujMdIuqryXVPx#X0iIs1Q}M!*O^gHIkWGE?3z(@$wblS+bNvi%u?QQ
zet$~xY$E_|nkk%#4bsY^8xOe_8IvG_{7e{1>-T!YBZysFWj#Y$1IPzgD+)sH`ybKu
z4G-`BCG4iz=I-v#`dYiuF|lo=cfzh2m%Q}>t6Zo{yc8>0@p#E9!hq_;mFIIOM_;~z
z`qO&ul>7wIk(^hsxh&R9kl!?ZYLdm-M)W;{q?4^)`-Q~sMfnp0`4=fW=v;;R)G7~c
zr9RzQ(?j_RKl1!K>F{a9^kL0K&MrsN+8O=~m2ho=#iUwCgXT<ysbMAhRTEwDE|6Oc
z_snjiqk=NJ)}*QQYrWlkrRgd7Lt|EY?6`qEP(c2o{%~P6MThI_q`0v_^i<^wFwipF
zP*&9=?I<bSaaa7Ku%q4%Yn#H3_0|{jSuKq=!#01|cQIdQKQ|<NQXC?{HRbez_=+jL
zs^=lV)VbYA(OCL5aXuGb=lQK_r?_aod_Iu<QjX8z0FUNa6n?``TA^Q~?xo@e%{ORH
z{p@++D0$W649vIJCiU4d+gsQX+K?~5b8qX*{ck|PhXbj3WcF*IH=XMn=PYc;_HCPC
z>$$;KAUI))>jSZLbXNKY7qe=UM#Ri$?$;XlH^jLIpw`o-9<8O~qT@FtT~0H_4jMD0
z4#GbNYBO$)Ip+psJ?{RPb0rCQV=(6j2pKrz>m*q<7xA427~HQrBrgw8POGp?U5MiG
zN-?1Kd~1Zu$)yXL%JAZkNyY7p9rcSIOfDr~w%xa_psWsjyU|f-MyhBs+lZLet2kn-
zTb8!<m__Ji)1Ao|eZAq{1T<G2an&t9ylnQ&W&oPc&eGlCN8<Jk-wUVwAD>%{JN@xL
z;9Q8}eJOm>MJ&Z&-@Kjz;rWp1O!jy_9q{j@i|=#2tHvbr%SM;?UYBEcZ(14j2flSG
z((?!4G-tiD39z4L>h0hE65=UVw)}bteFj>W*#T`VdpfaMm2i=>vq~(@bc);&7j_03
zqkew!3DfEitUlddtARF^#jO(4-1b{+@6BHx@t3NBBmIaq7b*=>C;Xr9g&*9CNw}~$
zavAm`7WZwrW2Lii(@E7n@GwF0^J%EKMMq?2hv^7kHoz(TTnp%<(MW%fn4MfE5f2Wc
z&vpf39zEty!(6|;*uhJm`Sb&XNiDOIvPGR-<$dU}VEm+J4jphughuNV^L?6nF=q$p
z6Sn|CXkF3%q;HhjWs}pqa^;m%>dN{)X|*5BPG*gzH&hnrpT_`0Ht9M@DFgi55qPL)
z!|)>jk?x43`^nrhR6?7Mm!{h2+|Lwra-Rn5?26l2;P(1VcJ@uFl-5ziJN;wz{q{2+
zK!ZAEksl!CE!@?F@910j1%_55_BWf#(pTNHw|{^Py}*Jrx5XC^Gse7x8_=B%xZkiz
zX?+FkUAzng8%Q;K5>d>A#b_9iA{vzAxo7Y`z0-eKdy#V%;-1QROS_|ilDc4uX8VOL
zV2L=BeQ7|=PAZwb26P9dmWjv@aTmGcPJgH~4e~<zX?GNQIhd(boX}h<q3GQA9Y9?I
zN;T!zoE`QylRsLTblC&1GzZq(!UPq-67WwfDJa!cU+?ZXw1(lRg~zi@@e~xJN=r@w
zuN}7;%gBdmI-<WDAga_QD~fO4;lgoIe{$^^-(A1qg4j3h)Q<j{HZw?xGu}32ebgMC
zn&|g_v$Z?y{$(QiM3To-*1)(7UO~n#!UHL)@M-h~1oKvalCwSDn6RsXGN1Hl_U&UD
z;q>K{g@C$&;WOTrnRj0Jsp-o}3r_c(6AXO54%Yb*UN({fjg!9`(>b@5)*Q^{?p)Ed
zJn7UhTbs82%HZBmxhu`j0*J7>T6n_bm2J*cYxc8QnV-48BtT<G^&5KoY0q?<_uP&{
zThg2UZ)h#gr9qtt*FojtsNt!$mT+=uP-|kIinV&s-D2;-ukogs@Bj&?%yOqM6LgT0
zI9c(?C-Z&Vi7Yf7+*7J9j7>F!`<eAlMVZ63Bw0`J(w{in69M#x2ajpCsE&7nD^bXA
z7<SE%`rq?vW?eny<qfS2)pt+P;3}nJY()fU_E_$VAv_ZLHVZe2^?+-(8Q4v%-%3v{
z&Gcs9YKlIwd@MzVeCjg%mSKboRo`7ILL0b+wHLCJ%mk&HJeB+dZa9gJzYsZ3o%mvP
z@3n0o${2<x%Pxu7!6&AJMV$tO8k+H$dSVb)CLo7Yi<y-e+lnPrqB1h`{=j-!Qao&W
zu1zBE4RA7QCs}W>PF##*dw1Iioyaspi(RFTpA_wBjhCwdI||&Ii+%v^#WhU3Z8rZD
zo7Bc_yZg=XE^au@=eT7&O`ad1@N8PQ0KDeU!5M0;EvBkIiHX>cR@@j@US{s`y(%{)
zSv3=EA7ry^Rr6FUs5E$P4V&1=GSJa{?rlZX^*$M&CDvUn*fe3_72CB<##Y%OE~Iuj
zWk`@4-lE?1R2Rx<*-bhqW%MpXTz_>wpy_+P=YZxft!m%Hk$1aKaASAEFaN_I2vLBB
zA|8;uY-_}B6A!_*n65SPYv!?-#XFm>ws_#T6K?&o2EMRsH2Of?*$vM5$SI_*1*k1@
zZM$cd!@^3;-V<ylD9SOwG%it@>tewEY=Ei9v{JE0lP@>}wp@%RG03;cG^y&LyIFYJ
zp%=TwG$vn?Qle)GxbW@f4Lz9HqjC8X>SSWTV|Zz1vVFI1siyMhy|=jy`0O%N+}ir*
z(;^ph-m<BdVj<ua^rRs((rf#X4$B^cC7sIZwXB?@2JGd<ikkHCPYa6mG~+V}Br1FQ
znn{V2@9l0IV_@@(m5GbD@u`?)zOUu@<nz;4HC}M8>ya&{8F~ZV!#(rwo1c>0`-`{U
ze0-k^TvnxIhVZ|s?%xRDU7VyQgodPLP+Q7sVy(DCgIn&9C8}NG2KcpxJq6qZYlW<M
zCtGP~V!3l+d0Sl{*j_d@Rea5!=3`2<nt}m<Sd@ycZ<B7LP-ZnnV@RU1G{>DX8f9pc
z*k$<>)qHmmwzzon)D7$V*ZFxLJx{-lZsTR$AGW{!W|~Kc3Mpnm;svd{{$#}s&heC=
zKQ|o7KSg|Rcx5{5>C^ZYN<$MoMp2Pu1=z=8K@rwfqeSq>V(onq_M~1bbfQ&L*b)nC
Pm!`c7^i3W2kLmvfOX~T(

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/README.md b/assets/images/impact/talks/README.md
new file mode 100644
index 0000000..7689202
--- /dev/null
+++ b/assets/images/impact/talks/README.md
@@ -0,0 +1,59 @@
+# Frames from talks
+
+One picture per moment where a talk shows SQLancer on a slide. Committed,
+because for some talks that slide is the whole of the evidence: the name is
+never spoken, so no transcript reaches it.
+
+## Which moments want one
+
+Ask the dataset rather than deciding by eye:
+
+    python3 -m tools.impact.talks frames
+
+It lists what is already captured and what is missing, and says why each missing
+one matters. Two cases qualify, both where the words alone are weak:
+
+* **the transcriber misheard the name** — the excerpt says "SQL lenser" and the
+  slide is where it is spelled correctly;
+* **nothing quotable was said** — which is what a slide-only mention looks like
+  from the transcript's side.
+
+A talk whose transcript was read and mentions nothing is the strongest case of
+the second kind. Miryung Kim's keynote is one: 439 segments, not a word, and the
+name plainly there on a table of fuzzers.
+
+## Capturing one
+
+A browser job. The trap is quality, not aim — a player drops to a small
+rendition whenever it seeks and upgrades only a few seconds later, so a frame
+grabbed as the seek lands is a blur where a table should be.
+
+1. Open the talk at the moment: `...watch?v=<id>&t=<seconds>s`.
+2. Pin the quality before playing:
+   `document.querySelector('#movie_player').setPlaybackQualityRange('hd1080', 'hd1080')`.
+3. Play, seek to the second, wait a few seconds for the stream to settle, then
+   pause. Check `video.videoWidth` — 1920 means the good rendition is up.
+4. Hide the player's own furniture, which would otherwise be in the picture:
+   `.ytp-chrome-bottom`, `.ytp-gradient-bottom`, `.ytp-gradient-top`,
+   `.ytp-chrome-top`, `.ytp-large-play-button`, `.ytp-pause-overlay`.
+5. Screenshot the `<video>` element's rectangle. Take a plain screenshot first
+   and check the page has settled — the rectangle moves while the player
+   re-lays-out, and a stale one captures the wrong crop.
+6. **Read the slide back off the result.** If its text cannot be read, the
+   capture has failed, whatever it looks like at a glance.
+
+Then file it, which trims the letterboxing, shrinks it and names it:
+
+    python3 -c "from tools.impact import talks; \
+      print(talks.store_frame('<screenshot>', '<video id>', <seconds>))"
+
+Finally add the moment to the talk's entry in
+`tools/impact/collectors/talks.py` under `frames`, with a note saying what is
+on the slide, and re-run `collect --only talks`.
+
+## What a frame may say
+
+Nothing, in words. A frame never carries an excerpt: every other excerpt in this
+dataset is a substring of something fetched and a check can say so, while text
+read off a picture is a person's reading. The note describes what is on the
+slide; the picture is the evidence. The validator enforces this.
diff --git a/assets/images/impact/talks/V_qzqY1bb7I-2836.jpg b/assets/images/impact/talks/V_qzqY1bb7I-2836.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..0379466bf4900a434688605d57047c05203b87da
GIT binary patch
literal 31216
zcmeFZcUY6zwlIzbN3e_t5v5qDl2D{eQ%4EC7)S`6QF=m?E=^?wCA0|;kS+mINstnd
zt|(nVNC85V-a!yVvHgNGb7s!H=X~co_c`~k-xr>T_uczld#|-u+iTZvgWuk89W&B1
z)Z^N-hl^_u=a1{#FxNG%1N--L@85rbo16Q<fdfDA9OdCTc#!A#kt4iE0mo0A02~((
z5EMEmB6#x5DFFddY0)!c;*ye*Cq!gq&r5*LNk~fUR<egvl;<E19}f?o#7TjZ694w^
z+cPe{gL{SdJ=?cOjB797o_&0KzP;oU-6eDXJ`SaS`yKd!d(XkW`}Xs2KL31-YtO!Y
zd-m_;IehfMLGHajaGJ&W#Lai$_z&k4`F~2YI$0qma8SJ;5=6I14Y+Yj>7H63;6+)V
zxU!}H3AAAO)#s`ODUvGAzg)vg+|-SpXnqUD6zWY%T?E5i?pJa!2ys}oZ_h5j{`T9m
zZ|{EY1DtOPd>rET?)zci{=FQK@89?A<=dzDlhX11=LBw{Eg)}|`Ngi@xTjL~oLl!B
zo9i&AAshfcuFG7k$ZJIT9qNK=|F$2Yv?Q~0u1k5N>A19n`Mrzhy#9vkpS~Qp@=QJa
z%13pd){h07U;69JYHe$~g0$aVZj&;_*Q&tf`97Ed>*K$1`Lh=4Z0@+5FWh^4WqfRg
zLJuzg+LKQg?XttIxPHmKj5}~|-<|OEYEyRCC(S!1A#y(1w2b$GUmpqWKk{YOc2pgy
zu)Od^afDXu!*Uv;@BHj>Ontis88}+fRDs`Y`~N}pTItp1?5?`sn3GS-2;ODRfg_(e
z%(;w||D8EeEQjyRIjR1R!<;WAb-y#`lTFBX-v2z_!{Pn=^3~sXk6&>e{m%Pe{vYrE
zpJM(W%?i9`fYe%zY4|DhiM>=IDW-jmdu)FkYoIzcs}Y;6I>Y@mcYNM^(B3y#Sez(-
ztZSbQkZJT;yhzvas6Sp(_543|A%FX@C+&hO*AA?<R4vSsg>u~Nw*XheW_yjo`i#f2
zjIR6*?mzTW_!aim)o1J*m+82zhfQPNUsBfp8pXczKX>lgBnLTqZQ>~c4GTp@@jm44
zytFNBMuSS)-G5!dzXh>J|1<e>=V-&%?8eI16_MJ1CZ7E3dJbHFp0;nnYO#K3!Q56R
zoNj1%VicaOBeMin2=kQrC*}9(MHGi*6RVM4PfT&{xLCwa^>b-+UIWxa26V0qb<gJB
zPxZ!~$ZuSuD(>&^d^(za>C;wgEGu{g)Q~O7tpZmUm+`4Q^g-31qzbSEqB5c2R)A7`
z!XBq*7fTZKixBS%nM#Sh0P}}Oue=x=WK+Y@6f*?*{x!bN4m+AK`~42Z{;|lt!(*`t
zO-pvBqEh)xjICQJ(83ca!=ACRDJMR88W8FViMr`53`5hX{xuLj0KDYrl{X8M&V5Pf
z#IOxS4Ig|~<L&{G*)l*6RM!khX57@U0R<O^JC4K1Y5RUXXV*0bwX2WHi}N^9Yd6}X
z6@IFJInZO(ce<w4@<FrdmAze*Q-mTqV5Q&R0vv87jk-j<yg!L;vf}47(V<(4akS8k
z>UIh(srCF3@ED(hb4+|l9dw3u-5;EMdPPk{sO#~Tdz$=u;D}Icmw!JItUT6QtZUFV
zLv5bnBM-by&C>a?A=en}k!`u8i3CU1PA|+P^$!HX(gqVP)F<rP<WjQ~e%NDu<)Ws?
z;4E!Jks2ZaYc75H!-X4;EG!wr$B!ayopedS5N35s5nD9>yZ&5nQeT$KdWr)Ux(U<0
zi~AmFZ~st}rg{{Kp&H~7h~0v&Msm}+L<QTzYom+4A&6m&ZiuvQ2cp<Ugth7Q2Gw8#
zyAFDI&hAN{?JwZ#BQjeDsANVzgArI}e0@Od*(>K8YB_5OSs*llM80nZBpvvvRl^xm
z2WcoZvq8)py|U-Idz^#;OHX_t71oi@2>2(;#kDv2Bm5(#F8qsnq-TNK&h0-FKmW5L
zf94-9_~rZ${S@ZMivGMxix6B&id=2ecb{R#@fm-@{oNbu8<!ln9y>6MgcY-Qfx#Bv
z+IroijM<uaRe6gG)8!(Y^+;g^isx|<37IIH|HasKdA>p1Q5#FsA%_W77<lB0LjaRA
zfUqtVMtr~iUinUE&x`VWfS$JfdIc8z9nL|<-?-=-?_C<vihbdfzTrkgLDNB^hbvmm
z>9Szw<`!TTW*@=x3~dzgL>_P=Q5r{<Db6fKt&SA=gEVywM(TjX#R;AhcSojHYNPXS
z3uS%rFJV+DZUOkb@u+7hiUIs~3Cm@IJZ)t3sg3Tb!ocxTxixbT4W>mZDW_NbQjD{v
zpC(ltJW%C<gt@29lA4^hboytGrdKgO(6~wIL2*gFXLu}2U4=;XOowj1qjt<bz-co|
z)l%;G$m;bzC1Tpv``WAw<HOy-Fy*}BA*lU~<#__ZK|29qV~z`wOLy8^l{8l^>+@zB
zd{aH6we4K^#<LoF-)iD!ibjK~cOc)(k#z?Qob<VUd}(jnjKiJCsO8FH_9#S~ZvYO~
zveE#MN+RU_#YhUxCfbj;Qv*idM;E%r3=3wzd{t1Kamu=Uq-G&c*dDFi;gE;TzO8?~
z+iS#aZJs1V$;&gco<?kG$PApf|NMlY>7ZWd!(Y@WZt7SD)yQEFl#%oA`L+F=%SZag
z<+)))$D9Z@CdqGir59vo2dVEb4S=PT0tU@w?mKzsyjdKvA~p`Ht`?16Dy;1e4k~ES
zsVa|Bx^(7;=P*am^JS>n#kZ*!pRW~fIE}7&tH7Y!a{99Tisky^J_y2lk}E(^1S9Zt
zbvxiwZ6BMBoxoF>FgW|<=n=5?z=2mEvNp}rUwKO(c9xLj!o}8g>W76TMNiy*+UGH<
zE*;+n221NtS$J#InK{MS&U9iYF2p4P9enCC7@tZQ!P)N@Nb;&_JsCG=+)GNpDC>`N
z^3d5#Ga9%V`%e3`vHYhWsnuxwlCo$-P={FA2%y_5MHT)9$7we2D5W|ESA837Zby`t
z0Ut4G*l&cs1P*^ozORQZmDuO-8K)vzB|0(RIZdjS$V!?nAv=>`cE`n@G3HZG@15vs
zOXk0+A{A5Z^R-WoI%Ii^5NkN9*;rfP7S)fH;_J+*7kC3{k{Kw<>J41;@mnG@l^!{@
zT5bs+y?c<lRX@1cYYY5DkAncls<TCHmcUM7xA;?T9M~{^r{KNQZ60CKHuDqQ(wG?H
zF$k23D|cX)8cbUiruv*d6f^0ho$|SBII6;OTp;<j<rk^Z<kHI@CRE9q$}=H>l#1cX
zNY{PShK04{Y+*-=AwUR~Xb<#*>FQ*WJNi?;ap9KM!lJ)%g_4-1xJctAI}1vz2AVYO
zEH;hC;yR_ZrHt=jzHx0v#c$1hJm*0Baq~#mV)@s}#N(`$_MuX(4}FkpC2Q5{gC$9}
z=RFtjfxgt#8SkF2K6<L%NXt2=2Tl5HQnmVws3CSmP;;SjYa`Kp9g!LQUhu|ZzNV(?
z)O~v`UZ;ro&x7!tCF$^lcT2wCxYWZGSS??M+6CplB1M04BCCuo1=2d+*LH1eVlZW-
z!pVCu%OlXq9HA}y8}NgflX&2oq&KijtMSE{ZyJ1786k`vsIf2K(&#oj37bn$Xuev#
zR!s28gJfL=eWo3=mET17Z@%v81L}y12SY}UD`)R`5DcfukH2wk)n;%8Tu$}-{Ip)y
zEqNchU5${j3LlO2JWrBx!hnL|u02~0&)dlZ6?x^Cr;)R(AD4=97w(S-+cmu20Wu>=
zeQ*tVMHbDOh(_97PYmZ9QOe0^wx7O#Tj*n5W7vXp*w0sor|axsHgyfbi>S=m0HCX0
zt)a*aBJfZLEcC=3=?1}DV`KM&<csP@rs%<oH77fbY#%6YRn+kH1eQ+L9RVX$i7Zla
zR|mYir*VhTFkz^U#%F~kJGgym?|2gDuvxdguaR=HTQTVAml%d+D8=-M?vi+vopt@k
z&s8MgIv6|^UGUMs_xN^^Y<m3BN2vvk2+bz`=s<hT6Eg(TIHa=_;Z6qDzt#I(7_qpT
zJ)pYw`ouVB$y+<4VULr4L?7GEMpDNfUFYiToPV5z-IDLJZ&KtjBx+++3QOjPi)TN<
zBLkErhd?>WhLIJv^%vq9HBR?$>maq>Db!{M*J2yCfnPM_WylitjUIf%+OSZnHUMl-
z6-{`n1NkUBf9lcDDL+P%bE;eb$7vAiF&8<#flpxwTLC&k6|!A|*agR`L#*25!rcQk
z{DFn+xHNm3#^<xJ3xC&Z{h1KB)7kWjGCaOew`6)qo&J}^pD_RQ&Gqo9E0-`nrQhSz
zR8D+~_WrZ?{PiCf<5nNnn;QsEOAkH1{$6?6+nA(Q?^L{?^zhL~!z8c)*D<dB`y#aP
z@!`SYK~;8@=`RaPj4IK;N|3hn$iaS8lYm+Co9_C;HaW{!DYGlQU~<U8om`tsWv|)T
z3WITAInDU6r<{%vRf+0miF)VgL2f%Wd3*V*IWxo5P@hnpTmn*!G(CjgtW1hISDe^h
z(^&?9+!mYQ<Ga-+>B@!dO^@Q&wH9VoyG`ggJ%38kOf|F(s>(=z5m{`M1$>k>P<=5#
z<RL1F(`g8zWX#vb%vfLGlvK##5%cgGNI3;DQQ5h<9-48!jwffLUPDSL#$j(vpU;_4
zAw;GZ+bKLlNcgy4QeK!^hwS0r4(E3-zUJn~@w6_5{Ny$osbgQcr3yE0c;3D<oO^TD
zW<aCuvs@=>3dreN4d3id^Zn6}u~TSKVeOgy#RtRa_UT%ZvGh#EN@=Uj2wrRkV%K#i
zQFfKvsiWRzxI#wwOO|SFBWY|pQc?xI8Mv+i0EkdJ=TMy1@}JM_u;mh%ixkdJ4p*xS
zI9J=@S;L;QSpW`pS0boNsE34lHWLi)1R1kdRT_)>S#d<#rILJ0D`N$Np_v$QgY5E$
z3A&wg=>dyGS*}Bl#4;S74!0A=+Xjz#N*3Fo>x@qi1L~}oT5k(p*?%p_t|2*SM>F3P
z>=*g`S3`+7tw{rB@F*=Y4K}l${$*}8OK!%#k{OucdH%!;_>{Nv&xV?=?3)uX`|Qi&
z`KMhz?=j!5uC=Zvo4dtwlw!OrC{GR5e%ul57gy2Jk!1fNH9kf<`8Cu6^j<SG|3tr3
zaBvnee@Tby;ORMa_=)*VCO8RHlA+cOU8G{(N*ImMC_~K}k)_&OxgnJPme8^8{7_^_
zii}*~n0%VyUXaV1&7Lv;{Ddt{)LpIzzUo2gqKh`Cyc*Ozrtfu0%GvYA$!V=8;6i~4
zwhG*Bq=&^i!Bf>4F!1~1r4iNAL&0IH$n$+b(bJ0M<=)=@6D^(a3;XS^t&X+Sb%S;^
zZp8InnfC8ciKj{QYg{M|d)v9O#mC2YtkNeuu+JyMe86V1t|P$P!HBV0dQZ%XL=#RB
z&(JQ1a&fDS2G10X=H)dGVH=G7jO&xou92F0Wv-|mgPC6KEkU{7)n=Xk1y5N%hco`j
z#e5k3l0clLC&e^JtdusFSMJY+|MYTH*7h40bir;j<M#PnJR;)8ldAn;<`ZRuFt}82
zdz+KeHL8kgYG_zk_>rAVx1vl)`>gycw2p>O(xU$zRdA=w9j@!O4lTYy-FoR-Y9FXj
zmK5&gtq#4UJgIQ24q95Xe=|1Xp!v2_ZTRUWxm5bFgN}o@bHD9yjM0jz#W${)Ff;(p
z^R`sgKK1%Pb`o4%+-EjQ-+_X+cYJo}SGza=l6dydiu{>>>e_+pKNj1O-nkd+coeUh
zN=n%MWkCHW+<)ec_wtO>tkufe=I9nKCVSy8iM#(Afl}<@$7hWGG9O`3_2+=v;9r;j
z_aOcg-u}<o2#&YE{e7&ojZ66_-u~Ys@2}A3zj0yiWZG3X(L-2SCw%dLN&FS;pS*XT
zRJt1p{L6HN*zlj)aF$DMI}z+g?^CktW{-O(frp%zfTwshB6Z<F*DFu=wILep6w;O~
z5l{O@&Vq+*Jl(t~y~K19&iv<pg6@Bm{l6#-Zp8tstQ`j@$xa=RCsT1`UihcOsx<V~
z9d}iGDnjBRVfL~&VEDA)FCC$Pi+9HZk6d~@00Y45j!3cxY`k4nVEzWVl8t&Aos^pj
zCstOlg@Xh_$rR&<?5iH66=fpPh44w(tZWclM6iWVZ1b>ZOATgkXrDx@@hxR27``OT
zDEFU8)XmyRDsMFKVPzdMH3i76q|#PN`sT-J<*FBp5)DR@{9Q$edAy_;lm8L-{z0L=
zkCwjZL$h$kB0FP8_^utPp`n>NXm6H|{j|S$@!Mc$y=pmgo?v>4h`JL|A$7z<u3Bqp
z&7`zaZJr1*8(ch(v1idlN_4?459dI^Xp!tgz2rKZu26>1OqNgVf_s&m<USvPGJCmY
z(x-F4hW2UGVleV#f)A@LW&*o0w?wa9H6D9XP<csem^@WLw4;>GGgD3#Mi{htJIg0R
zR;@HHY6vpHy1aM)GZPhmGd|8l#h;U#TnSD=Goo=B1IGH6$|aj`%}U~V7e<w8PPRq>
zXR)^?UZ$oc-?ojC`-!+npr~Y#s;1;5j+Q13`6ne|%4j|FwfY7q)Qm&DI0Vj<$7P^?
zJ0x#dD=@Olyw{b8%G_eKFSf#zLerg^L%orHvyHl79}SHQB=ds@Qpiaa4pye&m80Ia
ziL6<*lZZv<ydMF`zSB=MFNru(;Dt|?m;G9KKKujG{g1Lv8?r_g^U{NgY*A9O*}g-=
z_o!fSFgzXt5llY(CVq8t-ItNRMNL_8Zg3>#tMD^l8w*gABr{hNQM1zf2L~s*&et<y
z*ZhY=i%J;$VD329r!Tcx_gtSkWP_i~=Zfbr^;ID&{+D?(%}z<MZ>A(IQL>}Tf4Yep
zFh;IRn|l+Izy!>ibFHcY{or6<e8Rc|tywI9)4u=39$_ztsG>b!hjo+_N?O8Gjro60
zw^d0pn7auiL)O~|RAF^yFH(6kEudF|B*)$bu#S6n9x*q>IpYhH$ljalT@dTifWF}4
z;32(=8%hLJjlciW(Jv!|3wZYh`++*zP9s&izlu6JIAx9XLM1tiC-NnL1C;8Oo|=^I
z^|4&x^MA^wLD^TsySaBV(?Is@EQH6KNby=-Uwx3XL-LQF_RMC5=+D@9z=PjKdKI*t
ztn|9CIIB&rdBMR6cFo*S^4bmGHzJ!U`W-QPG``GfQjPBO&^|WS+t|>GJ&arl(Tcp<
zZ`H4<ixy9HEd?X#2WM-Nk!Zi1w*=H(r)Rs>%I=g}C>ygQZk?Z_64I>~h{SVHeTK=l
zSJ^{qvwE_s<H3g{?l%HEcqhcte&bsVwo-ahw@`H}(1S{ZNGby$YNJA}qd1(U7LXK&
zX~yjTPOJ8S=IpQ>4c6<N$M$eiqScQiBB=aZ$ChH>uF@HH^P8=ib`7rJT2Rq)9<A~f
zKGvjAxRS7fuYq*jHwo2%?F~HSmf7@zxC{*m3oX8fk?m{&-rxk?wzFCyN=aZdk&A-A
zjlxiikxu^AqmyNxOUE7Mlkw{{!g3at*UXRhmx;`gG+?lHLZUA04f)w3eW&%E+#*N{
zPp4Ec881H)-EbvFq}zRuhJ9FF_5)%Xg1jR@_e0NkLU&uhU1#A~NKZeMIBT7+5-89Q
ztyKex=OrLWSohe_J-7CYeJb_6J}UW`9tPW{72=Xev^9$BrDg+ENs=~pr2#^E`X=s0
z@^Q->4HgKH+>yx`$4c63C9Gt)-%5s*r3|C~%Jz3g>C=MA8D_;t<hW_ucC=3THGRDy
zyW={x%<z!~(dQJTC9oczA|`+3*XSShDM32it*CsMr-2*;L$68)b_%0lY8nn3!s4(5
zfANgiepe-8p39r>6>G!mvSq6fpNi16RuFo}^;uO2puLrh;nzV%OWqm-u)}JNL=|wN
zp`^%05XRVz;;MW$$$ryIA1KdAvTBOu<9A1`bHw12(|C>O6}T&gSWQCT7Xjjf)_M%}
z0|Wwz5}W>ximySRvaqav-dlxLS3EnX0@Adwn?&Y{csv~r&zq%D;C?1l_@Vo6@WKab
zqt}D%biA=kE;S_}?ZEAWS-ILCgrq=U=bQmAa&}serOXR1!@IlH$_e2iEyWb{K==Ms
z+E(CZogzsL+YNFCPccxROJ8!22z*_yA1*%P%?djrqZ%38rS#nIsifU0aTp;9(|47a
zL3_ANg(=6h`k4>DZ*Oftp$5c&z@W#czSF$NT-7_a_J~JT9NR(;uD5u?Qvc|PgX%u^
zwnk54IY|{VKRW|~jO`7$mY!5@Wm@9Okn8vJw7tw1q&#UTW<xH5)7lqK@sjyYCURy=
z$@Qw!F>4UyL)mIIn*}*RyV35~A6^hC>#C!N5v0Vx<nSyOmIRa*p$b0O%dFFCkON$v
zFJh=?2esx(Q4D<iRFl#*qyt7~HIf2E5(6JHn`bV%@}D((#s&Gt1^mXPRPB2Qw?<Ks
z9XJ1L>U&H6kN^I~dYwR<BO5n%fDn~tsIn!ZqJn1D8e?iO72W+B9Ur80AVRF=P_xMz
zsF`TcMV#$Um;Box<xPe_n0LN2(V3V+TF|2@lZ@uXB;=5fC<w1mGxj7Y?yi=4h`Cw<
zqhdbeOEf8w;pAdxiG1uWT==3dn_sWRid3h8EJxMJPdp<^s*R2cWJz0?7rGbFKLtP{
z(XI|yv}evNZQ=B*Ww=Igy@XXA5+$B0!ke<&PVN(~<;6kHt87ctMZ03jswqoxQJt+c
zhA#mY?lzB&g4tEh;X=izktV5$LmgK>MfMgHaRN`)ggEudh=f;ad2u&l!f~9td!D5e
z<9j4A|NedZsGi!bAct}K(wSrHLsmovCzM5R2|LBiA1?-sD`-Gp1Xn%D>8M&_v8xch
zfTSfr3+T*yST$Ug=?N<z*+h+VM;uLdcYZ!J^o;*3hsM2HrKzTMrmsd}4x-#OE{PGY
z^pc~oP=8q;mxlyOm<AujvINbSJJsRlmPSnasBh_Dt~`YTBCKX_yTkFi=vyY2eR<>4
zT9wH6ugkcuL47u|hE+{{-pLEq3I(~84Y}4dClCYdQ7ny<ttB~rPyI>q7qOTbUei1E
z6o)K?+>N^7-1Ii9#8~W{$YyzMt_b1b$K$GSMdbvN*i^n)$zu+UT&IyaffaU$LW;i{
z=50fThC}(o7Q~8t*i)AxD-dg8nV(^<U*91r%Q)j)7_ps)@tm9;OZF7xWlqMoao-@P
zpY(K1FSZ^$*`^7;)1imG)%!d;-P#wUa^hjsv(>6CyPIxK#wzswD)*}-s?SUF;ISUT
z=lVdP6#rfRBSSDYIZzdclsFnY+E??dc=3YBEN#_?-Sh#|iJEb2z+Sgj5%%UOnistp
z*yV#RO_9h{6c5gu)1aD20Ys=^vyz{i?sD3&$D}$_LV>|hw571K!au9{WZ7mPUVkB>
zg7|#ni^4s5M-(NQw+2CSF0C*u?*P1S2}CX3W?(|XY#%Kwttw+FAn_ralW1|<1pm5&
zYuuvgH*WvDg#<2!2UpRUw%&sAjYX~^4;O8D3Q^J)peK13768KK$&;@_?F<@h<gy&|
zk|HOos%G$OwvOwv$_#ydiFM;^wsYdbvfb7jaX>L;Pi+9SnWUh?TeQn^Th(Yg(PCeo
zu!T#lp*dsfeWb!#xfpiq#ko_hhiZHnNH5dN{&uY1!CpW{>-&nN2d=>%Z_VV{Un$;E
zOOIm;f`YnhSuFvL*S+Xp(w<h?NaJq;XHN-H^^6!{Gvo<X(LC+`0uSEKL$PsXFx&a`
z=vdp64(29Qwzcz+Y5H6#56iL^)rE;d7gaM?zj4{XY(0gBJgF9Nx@3=l?+FuDA^MXC
zb7li`5hxEkQRrc7=Xx};+|mY7R8pl6=J!NMK-uYXX+!=K{3!e-p7lMEyRAA~$4rS~
zq?4SE)ebn~$S_(tQ|o7<OQJJo&Cs5}qMfQtVs~-`FDNk|;|oMaULG{eH4h>+=sQ#+
zqd9>s)`l;>UtQs?g=(_r8QP08N4cUs$wjaoRG(ihxMcM`4*sE7E{IJ!&|>Tq<I_bf
zM8aTHc5+d0zj#epv7Z@O&hD2bnrO6~c~Pk*0)y$v!iW*@=}FkKAuHLX<7D(lw$<Eo
zxPl1vH!1uO{k8LyEc}FnXvLL98;fFp;=?LIU(~IS8s~BOB|%lUPr1=;H;@heL!1Q%
zhv0Y_A2S(c34gM4IILUOxn9DCp9lvuFIIYbk?oT8Ck+pMb}ss;bGmcI{Y|S83|T)m
zK|l7~txvNyJV3_j2w=>Jbh(_5UdPK!OtvS9aaI9FU_zrM1i-#4mj<K~DmA;!byR<d
zR~3sb1j&ZJA#6GN>brwjH3~Dy*Q)xvB@y0nZgvPu+F6^5e&<qO?cjI0C0o;tyPV%+
zT%8?O?)k;k9DcgiEu*nRFCoGoB4-bS&Tm<f3){Gnm_dJ_0s!3cW;7OGu{e4|n-<n@
z3cg+4CLvI!G(H))oY<Y%Z_tW^b=DyLw|b)qaWS2p>d6i`3tLT4n!3%Jb>OC<YqTNg
zv97&rqLtWk^+hoItwRUo_K7bHhex#ST478iaxI8?+2sW9C*-`FwZ+8=eJ`!&ciK(y
zinlP+na==ccS%M0{=~B0bnw=D?Apy*bZ>!(QwUpmsxFBEj-yF3<<nr;qAtvFe@PX>
z_{)jR3aA5U7&H;@N)(VCPI0ekXX&+W)%5tQ@-wfWAC?ab7WTI}<u+y-bQD`+gvxYB
z={FV#pANQNn{rN3D@2o9Y9`?x>rY10mvQ_&2|r`Y<jk_0ED%N%KEmc!Y>ExgJz6LW
zf%33_4-!=_aK`kBiDMl4W;96EB^1*DM@9e)K*|K0JyF^voojD4GAKG}JE*~y7Erm;
zJnVxq#v5P7KsV?s24%o9zd8@1m{jLn5<u|Y+-Pw`b)>nfUO~;+Tw{xFBgoW`CfO*}
zU$Zita=y=@k~Do&!+xdc%_Ea_+L{j|#ff~bfRS+$q#$stMW=k+#V&y?U>;WG$eSzV
ztzju|55(I@_Gcg*Oey&eI+4vnJ0SmS0BqyPqJ}eajG2Dt(E7tU#ke<A;e>$eK(7DQ
zp%P~8_$Z`A)tHiI>(~+;l$WBJIqm?H)PYhkJuk0c(grE_QSE1jdiX_{z=AjFEna!=
zxEGDiTg{jsB2E!oPY;d|NUyeV5JBGkKZFd3Y8ev<E45a=rq>Mps-Z9|*~E|}u;t-0
zy(gVim;-Jz2@Hl!#h4vx#ic(|$xX_JbQ6EBvTO;>`%y5R68E6vG{kS&&naaL)D|se
z{DdQ>$9wc#YFqS)WWus$zJ;}Skf4lMBn(_G6pygd7!$c=Eat83xkzk#1}9<VL-}*}
znfg|DL{-xnaMpmOz2Yqj#i|5jS$lH4GW((~Lh%(sOiWDtR#FB1S^<H$P#U~QEP84y
zPdC(j4W*IVPneV>-o(1Q_P~M#r%7)Wv4KwWNO^C&2Jb#3DW$NuWXQ&v^|GDa;xFbs
zQHH-?4=>ezmU5RvC!gt@Tyd=h+@YvOFEI~L98{)kh0zfUbl?qYLW&5+D(9V&bj`Q&
zBy7P7JRIPVYs!*I*gR^!sHApZjYu072$Tn0#Nq&e>%4YywvH+@tWxMA1q}DC1W7y{
zl&-eyV}c+M+o39cdFQemT01A^b<YA2Vv&@gek4kYwk<>sdVY2t&Z4W<<N=^?usHme
zN2E;MS>_}vCqS7)A=f@^gG@TA0HmgER~TYky%Bt&&gO$+O0uVB7P*K~1#+xpR%r#k
z22s!sa)AqyFimOA+apZ4S2gd$>0!MIm|&(2kcch8rDp|;Phyf1SDYphIKX8Kbg<9+
zEK^BSJaRRNmM69u2?$#*Ct<9cqB%zQ(Zol%k~kq1T!Gm)P`7P6U@$!FhQex|>|58v
z9A%?BheQT6y7GW?;yL7jl&dzCQMnE>cu(0?N=gmm6Jjy$Bui6YV>OIKtLKl;K;PbN
z$lgp}`Kt6$B7r1_5lT6tn`XXnqD!~4(n#e_XP)$1tHO5`6%%vFQ<EVJ1j4}eLF(2{
zUK12HLIiSjy|3+vtRhyiALM*fN@R^200%v{d~h5N$DL9uDP;U=#jxx>qO|}#hY)Ln
z@ser?%Q%Sb78~p%h{DbYR6$lJlZH!d2p&k%S-e@Ez8!>Ot$7<l_-My!3GRUr-;mZ@
zVX&+AyJXr00qsQbRL?M|@y-6ODMU^4wI_8EPd9qyCEGU~Rssj=V6N0M(hj2CX+H(r
zOcp`K$1K{P7f3bf_G4Tg$;Yu<`*`~|pIA7^Rla30=wcC1=Z0Xxd=S8`su7$YIi+ac
zHr~N8vE1ARpOQ7Wb3Q%&)FtRGFI=YrU@H0kI&5BTq>{-^tzCCeZ5@4JF=m_K3bhae
zihq#mB+-KHQIGuHWnR{LYVA{@W>r}i5+aSgH~Z)ky$Ds8Arn`~MX4)wyX`pUz0lzk
zo7V>#zwNq!K~$AK=~FVIr4|OW#hu=O)YPh~o;puF*N_cMH7@_gr9&+T4$XroL6;QG
zA%|>dnz2;|BLiYs+A<8hWX79w{h%nxfSOP-u24%|4@U`?K+Ej%uK@((tw_r>pg2Oj
z>Jd2yIEB*S3zdPj=Y9%-h9rW)k<nGy=upzndDM%L6PjOR0)!0WJG!TnP2?}iTb@jR
zvrFfXa?JDTLsl85lV7F_7V8^sgy~?|z0VWU9vCpJ93^p2hH1Fc^4G@ZG@sZdB;Jb&
zozHob+YJluy0ww&&Cg3@SpnG><y8Xun0hHcL4>hr4hnxDFsj%>DZ4Q`QrZx|DyYfs
zI4_cneM(ZqKznX}<>?Nv6gKq{#+dUc({c!>R}NP7_JNRI>rnlx?W~I_1;ZBdT^JJ8
zo(h8jU|}0h@p>sw<TL0DtC*5VB=Hd?=-OaHAc+bGVTjK!lHOij-E9fC;uwrACqpEb
zs(mr{p*bMV@9+@i1uPgZRJBf14MqEI^oiSVOR=fFF>UCTGD~|LGd(_!+{9i2;fe^x
zFgaZ2GJbxb(8k3^WDdl%bQXbJsOP(#v6HI|+wne7P)v@g+~|`#5~Nb*Et`O-S+7ar
zL<w^o9|VDfVGhaRLjYEyH?BEc^H1K4LTTm;RvWYo7ex%8a`7<}75pU)f&V2(Z0g`W
z?OkHUwc72JJQlZ|^5LSwoHac{pH+dBp>SWIL5`JKs<Zn<8$?P@lWb<WSSJc5VLgdE
zU>7vwDk0@3kVtZp-dwEhEG_iy8*X`*it%N2+@hQl={Rb@tYX9qqqY;K*>P5n{A9aK
zTS`Z0x(mQJ2<^OewTxbqj89yzc1=QnEEy(#Vqq^rE-W1~x!xT>EUaxa+5y+l%i<)^
z?W392@10#=-O<S_ZpHz4y!a%JwH*hN6bTI22Y#;Kw#aeft+>#514PxRcP);IsA2V$
zYm$sH0<>e74vwwF<;-JF2!O*=J-f=A)QqcgifZLcib|R4!$Gb+2%ePa^dY19g{fbB
zB+;LI0&=#2n-NTYk<_e=+kp+1s^m2W6S|~tr^^P#Nc4Gw`cPYmuYyBwq1&gSP-vT^
zT-couW|*(lPjF*@m#nDtlu(a(mzyP(@t`UHveKRt67!!=kT?TFAnpSHU9SJd#gP4W
zw=)n+QYi#VHB>XJpyCPxO+azxTL{D!PParlp2FX#Q~Da%os6MdE`N?x_5)0I6Ec%s
zh6K;^YSGsUi6z^b0Xmm<bjUhAIWmid^h<#I0^#^^eVTg>Jt;r(G&66`>?lb`dBXZ9
z{o2UQaI5g#i|dKMJv9lNz4zB_7l$e<R*uA`YbKWiN{WMrnRN#nJnv5yJu1kEm*4VT
zGN=}UW_rvTJ@osv+D&)5UNHG(aaXH)@5-swz=?;GO)?G2KAp|HI#a^)?-RB)Z~gxC
z8;70Fd|Pt1a7?i=*`qfAfBN_55C0`!@s9#QkcE7n*?Y_i<RJ%BF9y25R){FmR}xkO
z5KhvD@F-rBwQ^WosP&Yg1QPcvbuQ#E4V*ISvSy$)p{+YQg&Ynj>jOTtZYf9YlEdK+
zyLx1y&AjmrVmttUWVd$S@A>)cia+yzqejBn5wTDi18BqwSua|Vsx)Ex4R+;8;Hl>t
zZ_hUF63(eYuDVNpD@fwPwSXeSDSRf44=W)%6y|>DdFVUGJvnAln>{NW)U&^#R_35c
z@GWF>o-ZPyVeZXOqLTwZ{l=kP(m46ey8?*0!F_sRyKk47QyjTfa^&_43D1#Z6nyhm
z&4->&AbMpF$&IBMtW~*LM%UTUi5EUi)g7`AfSx^?9kGU>S+{^q6y7X!MWT8eB2|68
z8wX5&wN&l#SF{jZcZ`7fpUMMm#9OK@Q?C?`{;nyzgm5&+Z{WX(Vpl{6j;Q`BqTTc#
z`Db@EjDNV5oHbnHRdu4j^~0(xW)<T|nDxRRmbQ^N@9mHs#vBOKY5*%;s*0m}9lt4{
zCtmZmM{rJF=hDY7yIPv=Ia$;FUq$g9>)%AQs}LOWzH8DhnSY4xkNjQ!aYXo>sgH}0
zt~~@ZC9m)g^cssQ7m`wPPjYsPLY_`x%<rcNsI|F|Rb>}NN>Ve=#%$EOt6B)otj^ZJ
zIK=#BGRhBHY#)u`e24yG9=p=8;iz`we=(as@<qQ(6Ww4;nk{qM{6f{Vo^fS`=zbL0
z0w?l|Kr}!8D*wCT*-_QJmmJ9Rpd}d#qxf)^kvsP4=mUABr{Lxdvm03-vfHLi#>;E=
zCjX(oyU(JiIWyr!;%}O<YpMocdUE~%5QSa(Zd1F0+*SDBHOeoElyC9@#?G|jz3r;o
zS%z8;duoy{x{u__()2l0zgdEdNDjAB)3J{$e>zKuxs+nM(Is3M;bm0=Amxf)7O;GA
zv(nee)GpJ9)?dYI&9r9$uWo^1o$Ii6x6sl~Ig{b?r~gfTUpV{|VuGMMe^d8eG5(#n
zsvY2}O*TWDpFEu-LJk(8=^&;^cbenDM^o(_d6Bvs5M3>jlVL7FLBjU!E-%j~>IY2Z
zDD>ECa-5eQ$9cKsN#~4Ze$@aSJHKt-n2bF0o1bG{{yP&dWViihPqgIU^#pYAJG|%b
z@ckTkp6~ur+M30%t1O-@Pl3f^d1Z=Gy1FnV8bzoN4bTc~fil^_LoiFR4sAx6p*WLI
zEr$Yzy|EDfJUkeLdXr&QCbwiNoiKTk|7|#k)|D)~0qjZIk#^rnyZrR-i<a@yG!y@U
z+P7RsG5>)*BjoRUMv8vhGZOXv5JoQN5XPP##~S3oZt=v}bRKt%J09v{0@r{@-6Tz1
z+WYf9ReFO7ef7odQ(P)sw^!ALjKPPQOy*%c?p~e0(N?qS<#$Zh0p!wxcVK`Ot5;M<
z<uJTMHa?a{WEJT!G9um9$}Ozm)(hfqDNowobW#`dT$aFQ5QW3QIn>+LA}pE7{B?Bg
z!lk^?Ln%gwT;f__8m5{-b(E6nG311QYC(`<-6*u1m8(OTArT_;)&&AHvVvDM5<feb
zMAUj<Zqvf(+3@HsVEYB@H5PC9inez|iY3gluDF=|;C|XhxLc`h5C|TP=hh5-Uzmr$
z+bDP?<-I!;oa%fZhQ{v<RxlT<v*G!NfwLsv9wA&{+Cwg@!96FM#(=9CKX`VDkOk%+
z(zd!M(;%g9t3MA&cbHF@^MqBX&~KMZc`el;j9VgG5jr4lIrG6ew{qJUXp%nua|Pfs
z3m6b0^%gu?uaFKJm494aRh8kLcf3tj3HRz(gR!HD3yf)}rVg_cb*_?G(L%f@Er6kR
z<*nGcDVQUIR1brrNa^9-wdHibU~E})IRgX%5$x_8&C^rxM3Cil>AG}ugHihG%Z-?D
zdQS19liKOoG|`RmBhArn-R4TJRI@XCv3M1zJr=tTzQ(dtbbZX7Y_TY9Yat|b(9R~L
zq|4K@;{HUBe(BrRF+`KnEsL8{Zm;9A0|FNveu8l8S{Z|E<YtN4ro$%XDY6r8sc+11
z@BH6x!Q|2^So1~?hy0&1IB}l5ZjiTSg`~);)09((PS^80_uwKJ#bEk-siB`S8$FA4
zF?Kt)oz*x|?MGXeV~8D;>9}_pG7FQWSOe+AlooMVvNK&S03MCS+n(4+GFnqs4~O_4
zj4Y=*2-<OH)#`V-68!aj$>`1*y#EFs)5Tm7ssC!U-s;93Wa)}lKX7cH%-*|?e@<SM
z1MA4?1|wB&F~$(Wm8_<Oz_N{n;x;L~MS%ZXGnyWjVcE~ncdG!4!e$HVOr5`R37Vdf
z&$9PN1WBta>|cT|Ke4`1MsJUk#lV7oTtJX!KD+AXk#-*Dpo!0k&kxDqZQ#m9a)Ii0
zNvUMiFV_b&8>5To+2Q2D1nZfQ9U|nS81VB0|F3|l7@IiXk|O+5i!MX>A?av0vP9xS
zdIPx|aB!<}0#XPppgFjwglqR0%=;X#lfT#X{E0~uW1#esaT{+1OlxFHNoojpc5m_B
z-3a~v>UT)$X{=GAa766MaoOa@xh2j*)Bue2Q(fIzyiZEv+l6owf5lhf=O>X_)6FzU
zTpD!IVsW2vQ~JBJT<rhl6D9G8L~(zHAVPDn6k<v5!;L^w`ilBC$1zSkN`#z>u6!kn
zBbHFQ`h62ex$`~Qq?$(D+O<G(Y`Lvax?V@&FIYAbi~LXxYLj;?{}2;BAl4Gb^jA|i
zC?UnQx4sLOdGypg=Jv6!BT+?6E1^i;x>r@~ew<1s!|yu1k=NuK7dfnbg=V$kPonw9
zKR*P@Hmbb=<5ME!^O$=qXZ_-2O;`<B<h4~VDb32gS&cU|nr-L?Bj3O4xX@sPA`lEV
z0Xv13o_0ZOiVW*^-~1fiy&%I)Ur{1iJNR^>zEoIF4m>k)=nGHPH?HEfQYj!!tM8rC
z&3B%LGDEU{dL2cEu1LrkF?Dxu&e5jDUP3U}bs8UQhjmEgwGENHAR`|{flF*D=q9De
z8-pq~GKiEVr+WgMp3>IN)tc!oe)v)^Gd3YJr6|Qv{3&@FIy)E8yKz-tg3qWXMNKCX
zt}2jd=j;;Jzr#|cn4c(o!)Ti^COmpn2_i{tJ2k((Mgjh8izzn5+TOouQ06VgR)rs2
z$qVgtmOh9eE}e}?qqk^#l-{hBE<}qkq5iI#eGrHx;U4!>#M04>TH`R9BZb}%y{rz3
zu%y|OyYO)}A3Z;tew8qJXQnEYjuT>gkUXcop1gXQqD~9d;7rh(KT22~%MzS77S0v$
zffc#ItnU-~gdTE!d5>EjH8iB^no>6$78gC_FsdqP7H}w7s8-WsBr%Y>=mU>FrlTG}
zsv0?$_2ougTTs2(ASdK13p5D^*zxn7CJoXL-ERYQmI8npLac{zmvSb)aV;I&t~ns^
zZgd-z1b2Lur$u8WwIGQ#Ba+shGR`)x=#z8HG!3c{iu~@}QMWH4;o^mQ<M|(~WD{VI
ziMj#&+32uK=W~(m4u@8$wGjFXZXBr>7V=4^UpMWED!GzX?K#Bu7xVkz@An#3-4cr%
z4Xzrw5siV!0GD975`CpoO|D%T)#(H4Qi6GAsGUw^I<b$I7SLj8stM~5=+hAC5PBkp
z<v3vrH&C{vum7VtKQ8DKZFq)#Hj-$k8@K^qzPX>^U!_wD>x|2h)(*;hjpe>n_9dgA
z7N}CK1q030F-w93_fu`$vK$_0d3N++y|Ahfq4-!FA5Zg%BR9Ut6-Ner(#1cp8>fj@
z)pmIGZ&snQKBc7;X5UKDQ$)AE49G+EfD-UWk$EC%wWLx$AXZ~YPMmZ4=5m6CHE~5H
zL$wC1*{Dg}x%mEMVy8m%@|@>o$dOd1B{KJ7n1jA2UB0+*-4m#NmE@E8a~?RZyzKOd
z{Am=brMc9~5Yro5NAFm)giD7BbX-{O6R?dQGS&giJdsl|Gv|jAZHm33`c|!QtX|6+
zQou}#$sU;H?b`Nu@ZCK;|5uLfccA~!f_sx%mL)6ew*#{9-2<}pzZ{TFwEp_*qA&H@
zH?EIUXVuBGQ*(__=m@3Ch}~+`sA%1#H5czKh-$RbEAgMpnz?;T+KONY31j*-*e9lt
zIEL>>h&Orprm^(6y4&M1GYqY7VadD62;SGH%MJ@0Ld@b}iNx5FU#nE#K8ef4G?%v8
z#pNe&Qq#Ij8a9SJiXxk1r;~jcK~TjaKSLoBSxpE<yzkTL=XkOAS!lKErr7A$jK=zz
z0<Z2To>`!9q~}tua8X1mT~(OBD1sXPUI;~C73L7)@(w-_9f~|yeTU<Y&J=y)a@tHR
zddyKeah1!a3gfWCPo7fZDrZc6zrVVjZg^CxtHvZ$evKc<s@tq;Orqa<%GB2p^^S-W
z2vU*Z7i!fPe;WFu{!C^M{NhyVa2+DcnD!1nwtTCmHmVT|z+p<Pz%VUmXr4zOL}@_2
z*<1Pt5G?fF#epV^rLL_pRBu>OnXMO1Mz_3nT-2T!q;({G;CQI4b1<9H><R~^nFx9d
z-mdSm45ioJV{qn;zMKHbIL{Z7qpg6gUz0&kriTLT8TPJ9IRXzQZ;j$#g9Mo;4IlQb
zT)Jo7NN6@uRGX{y9J-r`G;dqzbv(D`f80W~<Qvx+yiVwYhV~0pg>$u_Ayt}yy6Cx(
zVnR(h^Ke?}FeO#^n3z;>E=Jet?PTU;=GDw0)fLA|Mmj=fOD)x<vf>iLYS@!3vXdDm
zO{2ij6t7z!Z2ZNj5XHQ5-+K1W4+}_I9ySz9$QzfucEZCg!W1dOV29`p04F-fx%ps`
zU=y^%sfO%^PuJ}zsv$>|uMdruNuwb4&gc&as$Kuf1Yd>8;Db5sP87``6-@9i<)OU^
z6mgL4$11*u+a-2pP((ypR|XH`(bYme&^X;e`Xe)UhR3tSq;R%a=<ePs&RrXcf7pKU
zFJWo<Zc)#N&e9nfVT(00=~+gkAsJzVvbGGJ@|W2jTi|B#+qi(|hrT}fn%>`7`vq%p
zuHg_2ykl%y-n4ncJ$(W=e3SZ=*<bPyUCkSRR2c}YdmsyRMiE_>U+~f7zx-I2F+2|0
zxmH|qJ?gYjJ3e*-;-FH|R>|2ylgSq$;Ka?@XIEd_A=kAqmb*jqWj`ESQCYD{Zy1@E
zKV10}bjJf{K%lD9%m?9|V@yUdFF2<ov$aFz!{Q|WC=pla*pEtiyvO3%zvC7DBYWt-
zkB2;e-$Va>JfyOFDE&7&4$`P@bjKk3HO}TgV%(}84$foJVjqP)<HP#&tWoWCr<PuR
zvE3If@;o9}wgPN?2tq2!<QKL8>O2~P`gXprcRLaG=p4f~v)o+3XT2T9*Q@LFc6|r0
zsdYN1dWcj~X^P`4{L7LhjzPV|Wb+kVAwzEdP0}Y9p9kGiU;m1+ys$Vr^E^XG`lst%
z-hi^;nIt1y*}%h?RGk#ig}66l_MF(%CBf6uI-CgPyT$9U9?HJgij*s7R|Zx>OT4bX
z5J}W4BX;n|O^l{s=i?Ea&S)YO-&k+<lH(e-6f4uXc&+xk{gOm$lK#fk@(;WNSgUS$
zp=97wH_Wv+bjK~dylhL!{OaIr0lG8M*C1ZRU*jQx@n+NHIA!#;wNQo@EgQ~U%miig
zOB0FYq!2xHVU_S3)w;_vBLJmfkHy70Y6VLfHoyR_Y@zMmR2Qy`9Fs6XmoIPC>g#k>
zHW)n2Nxt-OsF=CK9y+3*NT)-80=b2W7bh7y2aNA4((OXW#~|ckL@AA*A4*33BY!ZX
z;`L0AhM(G;E9C?SJTk(p{A7Aw{k5^_li`qeBc-nN(*F4s;Q-fkqju`iL`ux2XJ!`Z
zJ(e7*AON;dS<0i)#=L?Ly@CPOYbU$oXZESxB;GqEMF%>Yw_KjyDG9!KBlNh=gx%6=
zt^C3@FCz>&Pp$m&3AA&@OtqFg*hIc36y4T`M8a0v+{rej>9O{iAce#JN}*V!#H%1v
z6Y(0H1mecvTeKkR<*c3|U~K`H`gyTu<9=nF<JT<D%RMD)=Z#Hox}H%5|2I=_vRU|a
z;BkZXLg|*k(O84Ug{j9RsJ@lP%FVAoqU7np$}dpvGQW(MTrRFGr8aq_C9N4{7#YG{
z^Q~Ey*SwLy0)yftaKPC6wg-x@z*ZV{8v0TI<>$uIZh)`-tNlV&ks2kdG7w^Ji0|Ow
zxnexYgh_vv(oXC*{92}lfrvB;9@S_yX8vKd4}BIFWqfR@>u_+%Bx5s;aRG8=yZJry
zYHRy;p?erjSUeSZaQX#C18@6^>SVUL#w15EOi!760*&r+8~<CD{vVk>|Lqvx?llvC
zJI43t!45gtj>U^WQV=XCYuxhsrfy*QvjL4;&y`3I@;CX7gX=t(ts*CPj(~&o2&2ia
z@qgy6>Fb;O;*!!+4!((28FINr6Jp^B4|9^<=rWaa&?b6UxbEL~`(=UhX|eK-x5Uf8
zy~@5Ar~u3tf9hyhAW@toMc1x5o^m*FHhOV5)jP2yIZL%dHo;IBW3LK>U3`1>vCsQ-
zt)O&UUT@J$kBZ`gr(9@?W;Eb5ED-fxr*qD~E2qXD?HyF2nrJ#!Z4Q}_e=z9w;kdn%
zk8w0C=<88HfK77DViRBBsAOJB7xL*(#=cAyJ9Fe%2Ek?@04BK3rx+}b2r7hV)?_Ah
zbV&Yp!!nZ^*)|ovz}MjU;fu!7OP$rL26|h6v$AG(a^B2sEmfB^h-Y2&2ogkCumIS&
z7r={ky*qEYHG?YF#nP(Hp*FKDt22ERaOrC4dPXQyr-K;<{B_g!RkDpUGE~c*Cl6R(
z+kdMx1PP}0pUMrYVIyzEFjj~K8SV&aon$K=^7D?$R+4{nZdKymPo_P*Gc3*8nsfej
zEo9KJgpwchMfQ_SvrDBA{w(>;Kds3C^KOCqyOHJmwGVavJ$^1yw)agxBsn2#RY!>(
z_K)>+CK+!SwTj!eYEThvZAXXSkb4g;^6UK3=i=qS1eMubLN(KAk?4^x)+mI?N+&6z
z2-&6!im|SI8YK5Kdt8>Ao7fP!fw58I+CX-TSus6BvzktSwrG63qffV$%#;Z0XmPAF
z8rc%}Gm`&qAIK<%lMLxB`5s1~yhTMx%0|LWU$x^c(9fdHe<uVVMYeli2zAvd=Fft1
z$=LpT7WDnv7V6(!+ww>1-?+Brw;99Vu5J0#p6>pYlt07E{~BrP`G!@EdlW^K-JH6P
zF)1)YcsdWSKk#?P^yuE?EP4}PFgry8zv4b=NJav{RpH)2{MHMP)2(MrGYT<O{ni<9
zPuEj<fMOTgq#EPkTL&M*C3XA8u{cX{uJ+8=<zYCQ&=!(-n(2LwU#2`6>R;M!MHL_j
z9cRAh9ff;4DEpSn`V234ckALQol%N}Mq!5+yaVfIC*jh^#crqm82m?2sTbn8$=kA}
zcl<_NwGLhH)qY>wsLN(y>pzD7n(`L|@Z&_6HIBxoH|OP>3YqY1tAfU1$!g0>X3=ak
z$+SgVTKv(Re5b452oNj7_r%e&$!L?I7P&I28bERnKv3G}PcxIT{w1aJQ{Y4j)a<CE
zhHj5hDuqg`(m5AgGi;kgF2+TmNo9@+(Ah6(lPhUmWdjKU!`GIiI&NdDt|Nh*X*#)8
zY#!A4Y#=5uN`9cI#~Q6B)VAShTuG?0RL&BY{HfuKYX6;Fo3ZQZ=zkeV;zM_0YgtwC
zrjoF{v94F;bVG-rM3-36bZ;>)S4u!`3!exf)2tCZY<<Vr)5#&KPv*q9L|<dl=#})*
z%P>o2&TMXYw=lx_LBRuWI}x8&mpl8-w9#E{CuMUVeIzxp9YvMvfLdavc12rh6^`kl
zi_-J>xHJoOPqnEaS;Hr+nmPkiTxEPruaai~N88Id_Y+V)K}1I@)1NB90@Xf=M&IkL
zd<}0zI)KgvyCN$@2#nbzL`~D%x{SjmsJYEG^GK`e72Uql|8WOH`|COJKN4JfLca*q
zgpv`#V)G-3(9dG-&8-(B-q<|9L+%sg3=mSv_XlKGJ;DgyYki<M>LakAeK^LNM99Q&
zHbJ0n$&h7CUN*nvQSy2Jp}r8m?V|iitMRK*{^+lxuM_rO&#q}w*dg5s_UusizT*h7
z46nO_iiUquVVp_9Z3kz5$z1%gqyJOgmAGh|PfCo-DaG5t$wNs6pD_w`l?NY5HdTm%
zKLUM2Jkv&_uufah#?{wiXEUO@XoAzk@N<Rl9SBX5#zRN;it&Uj*FnA-Wt0oKzqccN
z<NBU^mgmlL#c#QnU9*xmpbfvi7URnJ+NFBtx8gQ`LOxh7wTRN=l)vm6mb#(Bfjsx-
zQB>!a^t9(LA~^A~_%5Q;Jl_F#5q$^Tt?~~*4n;T)p430_1bxSY2XKft|Bd)7DIIm6
zrY$tp_022C?!=mZ-pWPq-ZiBb?2LPp2>YxR)(5*ZnK_Po-=xA004Edx9oUHda{%?a
z$`>KtcLEa<#c=GfW&Ff&x3RkWQ1*-Yql=%v-`}LgxxdLMP;XOZHK(r%q!U3iB^Nx7
z&T?_p%3*|`WYzFlX1Qu)EIMp>8g&w6uvuj;d3rBj(KYxzyY7E2tAYQ2?R|GpQ|bOU
zy1I4|10sr`gc_QZARU#2CPiwbOUEEBKtQUp(wp=S5+IO-BE6Fpsi8wci-7drq?h2E
z-QV52``)>8XYb7KuXo;LCi6UJ&g9HF=kxu3pKtp-B_8TDKWqJ|=;%^t(N!DReoA@l
zpTja*lP(zQyiqH>+uh3wnfGzfO$;9RL`|I_3B`H)+%*{sNI=#QhI!K>6wClN8Opad
zig$-WI7$)vSR?;ir83|_?H7Rl0PhHPF|*{G#A=h3A#s*q897l{JfV(l=+hhL)2m*M
zzm^jigC?oLbb+2?c(iV&6d$5NGoRQ>=P`?Sn3$0=^&WJ@RT#Z2ZRCU++ULEdtIIL3
zv2cB@Yhh_EzFPEfeiJIfghrU||3lH4i_~d+)Dy%$j0I3t9zpCS&~tF>$YnU#>+Aau
z5i5T<*MNWuRvE%~{YNNjBiUzduQIqkKb_&yBzs?FAn1K+s96te3l0{>MHXOgerhcG
zf-V29y>Yq4%W5LaJU^#uY>6!<s{xQX07U5`n|L*KVI*}9dD5rhQnJ3jvQr&oWN|iv
z$DUZC!8=jpz|z_v&BexsCT!{;<|OE=o<8jqsHiNtPudY4Zr>A(e3f!X9R9EcQSrUC
z8G-1bc5WSX^;lYBhVkeWjcuG1;pPB`_mLen1|nHjpT!RjiSs7W!)Y#)X=oI{K|7W4
zyTfZYGi1IkVJZ5yk>qE}d7y7Dq`?-?ycj(vjr01SX7y^sw|Y5bQobA+svNECg(wAX
zzNARjOR4oO9$QiN-oGX>zznD+eNdSf5nW}*(Cb-tx&C1=W;sE<%bR4nCa^Z#r~AsV
zg{-1O-f8*i)i?i2G5rVe>AaS(M-I^F9wT(?!-f|+Mu=FW)AXec><}eC;MJAW)bV5V
zr`}xK9nv0r%)nh!Q#ID(NOdh~K4c|GsdefQs#<3CRl?ftHznCfT@s*dhs)!w<7qA@
z{E5!LGvfZe1KIAc9ms!7Xs}n+a56yW^kPPhy%eqQIIS|c|Inm8Kgh3q05qD>5k9p>
zOfqOa3Q?b|&QnO7u?c~t;1VghQf{mVL4p|d%v!YQAYDSR_BQ^F#-m@T%e)O%U`4*$
zr*$}sf1X&?1+lC^^mW^xoe&C15DApinaH0Nt_DKnV-OE5&%|A%VQ$a}pGGWeghzik
zbKjCiI20kCl}2q7I`=dE+^?U3Ob>CtL<c+;m&#u;QG{oGHaH(x9MNRCOD8+cvf}9r
z)+UDSvv@3zmO1AeZI2!Yd#!a(37-g2ecZcI!o}a^D*8XIdOAx^IoM$uw(zbSUg|fv
zjwz2CxyEJRwbrR8C;Lgst=-%E`3H@kfp3Gx@uFMTSGDaQG$OT4nUh1AMW2rs9T3)c
z4c^|D-hce(vsSQ`O_^-ODeeuZ{yFTD@e(o#R>w8ITG`94Hdxt*xe_aodBR*oj%(6M
zcB$ngyNn%V6WW;+XR7%Oz3v#Mk7yiS+UZKM9(M>#-q|~Rl^w9Ae-r8z4@96)t7yps
z9?7j2H*?{h7i-;xfci*){fkg+M6a}!uJdM^w0Z_l9yan~vUHb12d_c!LbfJS6IRS*
z<PB%L6puq@J}xlTG2;YEl^>lqKgzx&bUW2$Ty*u<zpUmT{OL1uU4hGb#jlnXIM9|a
z8RmL{mh*=)(gzM&z#?$=5I}0`-D(L<$+#R>$w`M@8w`=r$IHc^=~ih>{Fr`uSMr)g
zti7Jtn#7%92{O;tE<>2s2#Q$iTiYx8eXz8MBMKdMSF`xMBAej(CX@i9|97@qD|*>P
zSfaekP&tENHpzruM=$8{#+?_8*{+e^k&0|$#)Sz^)Yd_wwF|_uw$~Nauq?xXs3ewR
zBQX_-FCr2$xXWgKYeY4I7c?(xpi1qEmH!(!lhW8w!N80IjL%TrbH*Q3qwACg;CKo)
z(cIRGcznk2CC$awe+I`oovW7Dz?gNIkJPK|#y;Z#9uW6j!InN|!Q!S&rA)VrQJYX<
zC+u=)GIUy1Q19}!C-ii~r7#ZZ22u4;o920HJFTLS<{MC6k0JhXA!~e{-+H`kvei;j
za57BjQF1`6%9357UQWS~WRd^)`LP%J%Ft)(4RXQM6bm*TR3VnJi&z7(anjwW`P#2-
z{A&Gj=6VGlF2nj!FN!dZ-&e_0-Ker%Q7fIg*K~{2h~|_OlSP&4SKe8*H)END^Z<4N
z09D8~zfSSX|0ad_8;E~z>oxc*(Ze5;w;<A4>kLBXg0|pS9v;0#82z4O?=MHGvEr!=
zNHvKsc&*J0NB?wuwM&VJIr*ly0A)^cmuqGjaDvV|zUV$GPDUNL?CWe-^TVc#_-5V9
zqK8^Hr3}(&pJ=Brr#M3-?lTW2cVEBMHShoP^N#nzbZpEYK2z0z@R_T-Tx}=e*UJxn
zouy%?IT!x%BF${1u#<2Y@xuKx+%z5GZ-w5)G95fFAyrXM-45{4PE}QO&o)Z-^H?OQ
zs^2TwZ@azvIhskZv8;h-j!I_bN9A5rMj)KTA7@U`Hy%vZx=UQAU!u%B+X&g$KsgJL
zE*+R=9;R2)pHj*xNZLvH_e*~qIH~aY?SOVHKj!fb7O#(g3O=yn$YJrAq}DU~CN!)S
z70Lrlh~PZD`$ES!r;VTmxfF^PhO}%eoHoHTMi+n3>{T8=2q;;OR~5+1{y{TFUW$2c
z=c`$B#UZs6o&KqB%uuo??n9+OykTi^n~7xDho7w-A0?pp+kpoEw90XRhP?0o=6$aY
zm#&c#C;KD7&h_k<o-ZruDKr<4DAxId877;ObqKqhI)O`Gl%Yw4->8&Ze7yVJlggF%
zaRTvm2(jd%Pz4|f`6*)5UEpG9NLTg7T|J$VnXTJvdzX*&^U)cVQn_s{(Vy5VB8@X5
z$f}$Dj5AMD)*DxH?++1$#tOzHFJluM!edgk)bvI5;cKp%jnZsbFr5>ITKV=lbz32{
z;V*`m4c3dU{T~B8_##=cTLnyW-u~^M&+@+*{U-~C(@Engz~5!uh2(O8NROuZ@F`R}
z<tDTcwMacB4}R|`_C#03uZeh?BXdh|am3fJHcvfk69>mj0M=Pj0Z97~Tqkl}+XwtG
z^CCW}=ii#8G-;nQ2nVr03`RhH;^~?Xq`R*4*77RT#SesY^5Pe7P+?2cwCzYmYruRT
zwNK^xi_gD2iGvWZY+PWz?PZ*+<48zvzDY%dxNpV0GqJRxAF9yu#i}!BrW=5p7ALRK
zxn4RDg<B+I;SYouk)N@JF?Ci0vRr=66RbpIi|CjV_J}nv#@Lh^C|~U8gWlV*QTl78
zC2?Dg>3~^kxTIcPjT9u>GE6UQbMY+t;mg^V(JgP7)m1Dkuc#>=B%jL78rh!8Q+6Ac
zp4ltTEF2JPG&e08nG7dBe4jaqTI)ZOVhQBYegrc0)~I!H4i)Y0iJiQ0#O+ix$Tg)s
zb82H1AcyBSHU4G4Z{~Ssfa9rWPv=cPUxUV)3maLOj)Z}a7ines{`KY275+0`r1ED}
z_R4RGnL6-lffu0y-mxUxz7!$<<h2ryr4<CcnVsc=963#N<oQ%+XB3u(LKR=#0;bR!
z>MYBsooAM8lmBWkj$ub<Rdk>GE(8c|Ucg@oM`2NdXC3Q3JwW9g>vvKD@<ND3V~JU=
z?Oh-cZ_<Q-gQi%tWU2n*+EK6OctJcP*(9T~&tPGzwfBkMQ6f<)+r=PyE<$Of?v1Xl
zt`0=weNTSBN2AZ5zVpJrOTzr?&HkPF?{&a`xR$^3k^M2D(M^(Lu-eC$$0gxhU&WPj
z9h$;<TaVn3@Q8d*$H|%HemUN#75QOnBl4?Zk~xdi^bZ<5srYd6Aj&x2#U5>{Cwd%o
z;e1?Ci%}A?eAz~%A}{N5THkD1<VD2O3!)MaTCQdRqvIaFJIm2=HSMwd2cci?+x*O!
z>g&R7_NHp#CoX{tH!g5Iq|+g6bDqyQR4vs|V-ukBKCer$@m2;JL=bpg0x|Xt<H?ee
zK{naxrQmL;=_P~A*38Xa8`3w=DXjhpBOBNt<g<*aPPp+h=fL8x!PwL~oPdy)kd}>r
zrZ`X=gyz>+0&)5&@;)SLT&rGaUctVIFVq)f(DVS-b_UQFW$-to34Ki@Eek;=3|r@(
z^;c|OD}FnD#VAr=yk?O|ElH!s){;BSiEhQ#-Mow!XUF`$EZ^tb$S3M-)ffWqf$~h*
z#i5u7?w1~T$#Qfs6Oy5X?;?oKY0?)>z7^(RTzS1@XUiRIncRGy=hJf)Jm-y~Ok-l?
z<Sv}!it`vX&*PGlJF%Chs%SQ;qVy2=>TcS*xm=A8f|v_`&<w=kGlKfQgt+HxG?mRC
z=UIgvIgobS<$SK!MzvB8V39BI=rup1U-nb<C4|RB&zxD1{8vbv1sBTwBRQ$8G*o`#
zo3=a6vj4#P7SVg(x<0{Y)(C4A7^GsjM*qlwylw3WHI<q|Thf7xqI+HFjPN|c5$awh
zs9W7++wKP(qKql_K&W-S;j(p30=KnyL|z6LpW_J9X}(z02w?)<e_=~Vf972Nb?qp$
zvc>TA>#b6_t$Ct-X*OLuh7IJ0dpe%$?&Uf7i1oJd8@Hm|7_32Us6_tKSjTKEXbuLh
zQ*Lqj>=|4oY8tcg4+UOD^`zKf6+dXUeha8hP_q_)38?<-&HkPFf073>!fH>9>ZI(I
z^U%Qxl~Zt+{bDos?!cmUuvD2qGz!31gocGrw>HYyIBJsti3*d|hQ++6yE^#!;&B17
z$XiFQRQ6iD)6}=`@zr5r*N4(orJipDI4HTi`g~m}=eqUtCZ?x~KyI_M=bmvoy(F9y
zwz_)t-G_Di!=?+~=o6KnoorZX2)7ZNZ7i}`Se_JYs2M;ktU5a<nJ{7y8q&Q!o9VqH
z_O91)iXE{KQ{~T+G`-~-fgy&)mf-p&w3*<$Ec1*Ppmgzb9eR9(L1f8iO$}yjYpKVb
zx|<d;ZWVr0CalC9UpH6Vm=cE4NBnwN=U#)C>AWd>yq@tw0etp#jiK-(+FO=u<|P&|
zFvF<V8!OLxhH2eaC(ST8W{&rhc=DGp^=DV@IDB}m7mdO_aRzgcrY$PhvGNO|#hsy(
z1QZ*cr@sCZ$7=1a9si6;oqCba;l9ivneR;fx9w`g(9w+c17_>yGH*89L=uWF^Mq-p
zx;X=kPW1(D%ik7>Ws(o6WVkq#9nxL+I;^8h)w?<JRq9oZ=t&MtJn)fWz5DHvp{99W
z<2VEs!3HhkjpgaGx|$~{%<q1#H{91d;OWOEF|nhhjn_OUWjlyY5Lz>)9h-q=!^mYL
zBv#uvyr_hm5kc4FOn#F#&##CouKmJC3(PD?GCucJ<M~q+ugw`_IBc`vAeOo$6I|*v
z9llSQPY26I+B9c&t(W+z8=ThAZj^tW{bKv!+=YzAe}&u?tJBXvXQBO>upmsjh7_DF
zAq8SFvIB(}VFk70YW&TuIX6SXvgNXFA_p;b{EcrZmtrf^)kf`iE<FU}wH}$>%j$nV
zV2p8a8Q);o)&dUPA`yWj<CUv5P7&(j?2nmJ&@QTw-;z9U|Mk*;YyKZsFhc<?Zb8e-
z_|;E29EN2@4q{>pL7|}H<%r8$r|D{%6<1Bt4^(o0&<Oc1A*(pMnYq(d9l!Fq2N;Kj
zAG;l6O}XWoz8r?^Ul*X|F$~`Sv?h-Ug5T>a?6|JvEFW`20+kGX7;JSTx`Vz)m5u3F
zNV0I~!?R2svA2cKoGpHzDgGGKBI_iK2#bIyZ>Fj1U@DA=dxgE@lGSp!%Uej8a`uuw
zmB1C_1C^YPwa1hvAptzgJmdg%?)oT(1_qXsnQJbwL9IULT$~%o;_A}vYoG6sPr@Oa
zIk~Sxwkimj&g~|+^|;DCdmW_+#Fj!HgK1I#Nw?D!u{C0JkeKbN956p@2AiG@5|aZc
z+;Ztg-a7A*EMK(Q&bo(W$|%t$wP~<*g*1WT9x8ZR|1?=ZX}745cK204t&xrlVa~kD
zs<SNN2MugDDWEWvaSdFoE~W1<yz7=;qI{0DVPI)@AJS=13e3O_mj?$seRBqcL;xft
zovS^vX=6@@?ZNb0Lnb`7@do?2S&eTt4DxNxHH`@oOgBNYP#r}to#R<&T>_T+sox%Q
zni;GK5k-YG2=S&9_Bp)mZVCaTBo<npSu0#MQxPCbNPuD7ZaDH<>QRp>!oC8T+DqbW
zcQ?Q!1K~C<2n>5i+jd-hfAJJaMiIY%8-fwkiQ9=~$A;^~7_FjRrldkZRqY*CE-CXh
zx0D~|*9(<?2@K9aQKcp*%wX8ae#{e9G*TzhT!<(lNL}rHUu}Bwt=!h<TkjjZArChH
zCTCYs$Y6=C=0G+gNj;}+obx6dl~p=;R;P(t8~pUMp=E9*dCpK*)U{Mw!fqrQbyDf^
z>}V`BNrtTU)sqX%dtiME_i2$Tct$z4JDQbi9PD84PBxDZ4IKsA^=lLL=46+aN(A@U
zidR0J9zV0;d{CZ$_NQLgza;{H`=tLQE7+qqWOpf9zrYCJRV!`~GTk$GfE4h9CZTOb
zk733~G9??%yun++e6SMjyBu?wb#u%26>}|iN9zd3Mx}CrKHb>2r5G_Aacb4kZJ3P(
z$<wbvrOdQTuD|N4+9|dpX~L41^P){ix<3sBKK~Yq=m5qriL<4Z(g`S8UFSw4Zp>ci
ze184Oo4jU=2e78Yw*+R?sKvyo1Y}v<@(WA;zCvc}Ea52Rm$G+9*pf<<bZp4&RkjkL
z_eR^=r5W*CkA69*(e2Su7tl4rVgfLBO!3q@$Ac!A{1s$6@Wodv&2hyV)fx`^n$|F5
z5oCfQUuZ~%jCW)$&gZpSwld3@l2yLtb(hq{3f4F%WH!yYtC4IwkPLDVR#+hFD9uH%
zc1P$Q@+^CEsiVww^g*}uij{AH09Ng3FY;$qi<_Oae;ovXhkK0OoZSp~@?vCH?oKsm
zJw7_h^XSsnw($p=mXr^pCBKIqMMm%%${07!=o;i7zS0s^6bi=9j?R>?+nt2X5!M`V
zS=n_@=M)ABgsAg|2`Ch5PNYMuwn4r{Y2==%YsKT8*KTT-i%HXU_Fv)w^U%Yd(1*4u
zRoPHLh;E_qVnO~^3O5>p0QVwAmbrfQ5R0xS0cMPDvi9eH%vQ=yP?Qk~h$djR%pL}7
zQ+&>=U6Ya4s_3%&I@_jQ5-G7Fy+OIk6s~qafW3s>Dm{&a&gSDK#??+a1@KLe*0YqU
zRNV2}3lba<3;~9go7VQ{Hb#z0W@R23xc#7cI$7^lt0#DpE?50%<4%5jg)#x<xQ6Kv
zaO)tHX2}iIZKS6|yg*-14W9sczJ^se<QP%ofk?**$)POZA$`A)+*API(EI1o+==Nl
z-0~nrr17wGmDDX^WCx%~H6F71W?CS8$PD_r>^*i!I)JoOv;1d$saHJ93-J$IwYt;t
zq)Rjb3%U?)u2Ecurf>U@m0XYeq<G_ob2N5%nACu4yPizQbPZ_@>6eO#Mg{Ebh(4DB
z9V}&=pV&4-Wl|6*o{a=js6xYCQvd@sX%Jl_ZxE|JX-L>|S`DInr&^C4!dNYcYo{ka
za)zF|Z-g$E6eCLQV>m!Oy4?@sv<`a|PxBbT|F^~QZ=Lv$%FW^biz?}l*%?iH5m#{K
z!otNdr0<E9B$v**j@u*B-Q!iEFh-O?Hn?f=q#3u|j(2X3eCMoo$1_;o4o=!BROH%1
z!**jphG}XJtz;^p%c1genzyJyd_{4AdlND{>YSUV1Yk4#bbuvLM*(xt>RT>ZR7e)i
zQ*2**CuRYW_L1W7CCF53yx*x*n3;P3?2naF)RL32waL-YJ>TXqab0O9mpN=GNEC4;
zlBg{qS=pns!?{>i1?18jV<JD+Cr4*qnHLB4*?Sl@F@JX>wBf@Muccy&MB+rs5gl0G
z6zffDIlDlRaHTT&oUYL4yQu-u1znjpL7R}ncnw2-dV9zi+I0Y|iJUdgCAb}B67_c(
zHg=E`D=Dp`hIg{{z`9hPKa7uQ_>}pCF=~0J3caNd3sMw`W}G)s@sbu~P#vT*+D4e0
zb!l)b_v89jC9v4EL?*udMLpE8F74ALWOA!m)hm<Wd~9$!yn5D{?3$NcZ4ZHm4aW>?
znv~QGu2$$@G1_T5$%W`1*4%SbRYky<<2qmT8=P2KSH8eKfJWWrQR5Rw!f>~DudHXl
z6rD2GKsen#C?NW4N+?lddGxdlo9SJY-kB;r3RQgFri05&Hb5@%R*2<dF?N!+saImM
zx(J-)w7DkGQN+|%h7E0$!mXXI;-V02U%b4t<znbCxE<F`pIi5($jV-x5v|zXVbUff
zN1C}5?sk)~l&<EAc`<AwETAdDLpv-iWJs$(ROk?mDSER80yC~NwM}DMQKQdv%(`y|
z(_Clx_&XlY81FMR`S)&)zrdeF$K5YFj^4E8GriWIz8jFFU#6!v{`dw%`QJkB-#+O-
z6;_<J|EV72h%J#Ln_c4U)8dEdd>2F6#y_xQU<#Lb7}mzFCe<%=DWrUSEN(#V+M=Zn
zm5^HdFoWQE2}!PCZXYb-b*+Y|9LjoFnAQeisl<VZBB(}Sid;RS_mW9W)w=x17z^`V
zu3ev*G-Oo+VqPg?_hF{TQmB7nJ<jZK>4kfxe+8_mAkXv%O+lu2aZby=#@F0M`$9jU
zVAtgFy^z<jtf8sLtb3CBP?M0RQd(!z*_n#+Dn7rkoP!~*BTLt$wy(Kyl}eaa$#7qD
z%1n5|TEa<)_G#4dy(NcQ3F$jys=lu=KWM^zBI{}n?Ck|<eWy-G?}-WFpJ71|I}g5)
z<;@>7B_kg!D0YMM8udsqlBYSgvNSbH?z#gt!xCbT=K^6hGL#z+WyIvZI}R+qh?kpZ
znIzhv{B%__F|;aL%jepr7E9B8`0{KN^BJmPeKoR7toy-B2T`+=SyPKS8i!SA*>m^k
zDY{|XC9;PbN*nlQ0>cj)l?}*TnYW2kUvZ0Y-szb=@u=B}EQdvS*(n$=?MG>2k1VOj
zF=cj{Dpi|r4?k1EGd%SZ?^-kVTm+hi=4P}ij!nzOdIr>4*GTA<Ex;+;8RLGZSV1D*
zdk<s;W2#c7Zh&ND+8wA#>gXBP-8$&v{Xz5IrxAE4`1Dc!q;&M)7j0g4iwtaVa+8@X
zYxt)&%N(f7)Y40slWvGN!@*e1(_<{%_o2tJ@*GkF;4db761oguPgv(YrlAUKT<nsc
z_ED+2sQ|n09$9&~szH+2okW>lYoH0gLUD6udwHibF>b5jp0>ycChBuXb96gtaXelf
zTJgMA2o#~8^j^}yAQ{eryD~P1)MX45RuSb;gQ8d1s?17AT#yKpH$9pzKWH*+AYD2M
zs#`EAIimT%{xh$SN@Jzj){3tcNQffw@9N$mJsin%Jhwi&HC^79i3l(R*?T-29wtJ-
zq9ozUbkN;9X2uZ#z4<y)T>`b*6&YOmYCssj*Q*J<@D|T}L0*1N2Di=YktUomkfF|~
zhApic3<E2CA4BnXyrj7YfBU<Poy}X%)FurWi&840n@TuZ&l6B<0>m}waqdGwYWz9O
z!q8Zm3#$DrWXhKgX+m9>JJmB@4(TeR6@z!|km#*wGsg$(#?~q*0UE2;QhDwbWV_?b
znInWTpR<IkS=;-y?q0@~_)i#0xV!p>UFGl|gIvkGoh#hJLA?`8<7>#V)26&Vp>e5(
zBC_d-&S~p~KE96xRynVK{!^Rl`d*#ApPBaYZPv%7zi@j~*3&13?Tc4IeirxsdFhw`
z|8oETVg6T8IsW&^yzm<`*Z(~-nSVoO-M>dBf{M)3e~-*F|L+UJ|LViMEB_ag``B(>
z{Jn20c6>vIryt$y8*RY70o2X6s9T{;53y})=)8N&MU66SemW-mPW>dPlbHIRS_6FK
zW~WI<N0w+?0jrMQOo4b=Ouv`CvevEr-gW&Tnrfy(DK3y7G?(t2bnbTwp4ID7%xZv`
z^HRPPy<?nKWXa#XX5v8}o`?BNJG_smk-G8ek(O#EXq;=sv>sWdbqr^UJuN;4xC$2b
zHn}STvu9rzV(&W=a?<6N<x9Y7g2vKIM*#!TrMV2^=<9nb!vb_3(XJ#>znFAc%i{&_
zJJ^6?s0)|Ik@iNcD|?b~uz^07e|;tdU-pB>r3B1TM^>Y}RTc2i<_<N~)+9HS<43e(
z-L6%AFFtL0ukSpn&D|o+lh2P&G))gh9LqVF9>$##DASHVkxRui*i89*e$b@a^n~q)
zQxdC+LG6uhj5ivr?hYM9x*YJ?L(B!)p(cK;!sE<JA#uGcQoFn{#!yW@K&66Gq0XpL
zUI(PD?bj6xT|f}^3x{dP-ZU$4KR;^#=dzc!Ei%2-`Tpq!Z*&2oM&#GPp<WJqPlaaF
z&s$A`RIArnT8jBDv50$bayg?oKTXC&PWa}%@yEr)KB;SvP^NpIliGfX#D|wWQJ-(P
zv;U<sB#r%z&H;?bep~^~)FO6#$qtNfPdC!-e!9v>8p5=QO!8{OiK_z6Ae(QSDyja7
zjtNFYGTUB3^kkv-#yy7BGA2wl&B8x6d=+qw8bzf}Y*FBbv~H3=XllA!0Q^nP3(ez9
zsDg|eXWsvh!Ta6n{V)AU0HfuCj(VEY4Lj~pawQ*E?}|0*Az*nFI8tmwt4G6ckGSdY
z$hO!mZ_uOFu^xH6VBrS!dZQ-nm+NfvlP7cw?AB3_PF#pqe^n@upR|-774VFPF`CV~
zu3n+$HKvnUlfKh!oP-H)#%z|*N1kb5)9If2Izn`OqOuVwdfX;ACes#Zy9i?`P=dUj
zh!&vaz;^o#6DgOOW8dg@^pIbh?yyT=TjERN{0{0D+7lSVGu!R=DA(7<Em&0uQfPOA
zZSM|*alWd@c&;K_4icy`(G2idUnzZloRblK_1uQC73HWBrompA9gx4z8(pWa;|M4L
z=RWl#$WwBGZ(Fi>&W$y`Nt!<>S0h+(&wbO+#<;@gMxic=AY8XB@Z*&yZv9nlidRsf
zDV)F>ENIRbEM@N*^USDJiQq9&3clMyNSet&bzO?ozht^Zft^MQu2q+e7i}9aD5D`z
q)hRk%=c8U4PI&Tv+qC#!y<w^Rnb#Fgtvj8@Km*J&DxdYo=>Gvy3igix

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/clickhouse-2021-cpp-siberia.jpg b/assets/images/impact/talks/clickhouse-2021-cpp-siberia.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..000c776ee464df02c94b5a039c84db6db18b9e97
GIT binary patch
literal 70313
zcmeFYbx>Tv7cMxsBxsNXCqR%1kOa-(7A)9cfkBhtmH`HL2^J)1fB=L0;4`>eaF@XX
z1b25>esA~ftG8QktM=9Yvs-&_RoCr0eY?*+=XT%I-}&yt%)?&*snQ3<4*(1d0086h
z1U$?EWB|C>*f`kOxHvdCxVX4@1jGac`1l0mL_~zdl;l)Yl;ji?)HKX=)Xy1TP*Bjb
z(=)zgVPj*XqT_tc!3tz%Wn=wMCm4^n65tb%5fG5EKBsuj`hWX)Xa$hrV`^fxVqv@l
zV3J{AkzqV^0_Yx<#KgpU)bxKp@NjXkvG6f4AEl<G01PZ_05%p5HVzIx5djVX1|}8&
z8wZz+9FO_!Gbsv<m-wbXq3^){l%F+Oq%*0i_=GB8)RnBy9Ru2&V@Cx9UHGfn;+%w!
zUTKZ%b}S#tWVLI5`8p;mrxO?ypRn>c1kIzV{`>I(7??QtSlEv;7P3bhVm~VP-x^?I
zYmifF#*RL!|F!~0>H!US`X4zN78yVSu=i0`QdSoG|1bV$Y@h<H*`NNLccu%8q{k6I
zTbXD~9B(n7kZm7zlEJH;c$FT<Q{H$SO{gBe{zu~+qq+R`5g-5q&*Z8T`y<<dQ1@Gu
z7wwRYH460M7L<}BSnBDlx;~l8TKP|fb4LIju!8nEE;LrolyD#b`Tq<}Ndk7&5A+$;
z>5gge*L=%`@v>FNElCzcl?$#w$F(Xxb-*Zw1y6W#IEe+}IVB98WFujJFvv*B$o@wK
z@CxA_<#;?ZIsB^DgHYb+Df2K+k6<MMmV%;~f3l7CSnP#uPTh#eDp<QRWY#GtKa!9Z
zPZkrzC`_yIzjaEgfMo$E#b+_ueU2aM7aEEm7S&rxYtCs_+_cE$Ezlgvz%q2m=|wO@
zJ0`_mNFVg{$45byU$cD0K-WeO)YRO(eoT@V8>)tNGx9ao(i-0zGEx08G(}AT8z(In
zyCjKivBaPOaq@}k{udLO>`cWnhc8zq`dKy*M7)rP*5q8VRWaxEg<(F7hv_be{M7)q
zigMcGf?2M%;ZS|z_cXC5H9Wc1B!jX9VMr%x;wEiEWfx7e;sT9j<&RUy+=3gx;=q%q
zvPU*D3gD@nOakl#j!lfcH)3JAw$1tnicp>Y1j=s?Qs&90@7tQ^me~#ScK<V1CK?Pk
zUkh$WB#c#rb8$Hz)!+0JNaIz<u>dVP9{}7AFL1=7!Z-9!&Uo|ELY~QPeBYjM1LLWj
zvBF+rSNvaj<bUX@|6cG+^~6ywW!TUW%BGN=@0h8`U`oKna`o!pRNV+43^@te7WzB+
z4oNrdL&=)+0ATi?u-7z!Ns!Oduf7|$hpDN9n_pN*h6v-4(HYVBVp%ED*T;HZ#CGj0
zQzz<jfhURR+sDqErVF>Hs#<<Ig);t&nM~aK?e%I+h>+$+8c{px1P`gtMYV#ILqE;P
zq<}beZ@Sek**krCtFpR8bTw>hMMgB{<%1BsCdTu+{qh#Gm_2>-q{mB7b9yMDjw|)B
z`fPByJR|uZdIHeM$kc6)*f25lAIf$Rp?$n;ZNGUvHfUyJp{&$9D@}Qe!MUA#K`%M=
zV^_{j$#OHJP^VVUPL=-FUt(7^$BS|NM_lp?aWx}RYvfj6`{YD$^A?xxT+Yv{U*v+^
zhatflns}8-`!6W}v~}Jfw7BywhkZz6DSRe520lO!%NIpEk{N+KNy;kQKimevgaPpV
z`om0LH;yYDa*{8lLp@S^nADff8{YDK99$*InCesVwhBC?=_KIAOQlP%c;bK%dwA3}
z<pUtnclWw~?w*Qqq>dv@u}%DK;}e(Ud_<X&4*El^CU}I4aX+D@n!xjN`hM89Wl7`i
zNer9ON8hiP(HBJ-w^BTg$$z4EGsy1GG#>z=Y(tpPI&i-d0d#6a%_JA_G5^GZzVG;l
z7Ua#%N=d@L7HVaZO^)^L#A{64wrvuXXQj6Ga)Y;YpN#W>^=4P}W^FRo1y3aDj20<z
z7>nR8R3fUyjy?~7_~ZOE#BtV<i_eagS0Tl=TP*F}QaF0F(oKq<B5lX_YmJwMR*Oo1
zs`QL|cf9>3cBUC?G|tjjf-5Bg>{tN(pW_T}(YZe>cAc+X&<ZbM9dv6^y2BqmreXaD
z7eHdr0+1V0N@0DmV8=mlw4kw76QeN0F5l~#b0P$2ynSb|d!@q|XTXqE=vBT$$;EK&
zxAOOsYHT$JjytBnRqMViIbik|S4sDrLwl5={zjmK3Zc!E+P45dI9_#{pZHf*<f`EA
z>Pq^Su}g!o4fOV1@&kY%1tiC~$N^i%f`k2-3aytk;bQ*i?rj>kn7L4+hQk7SZ}$0}
z(vD@a2F;1d86z7{q;BQ5R&Miw^Fg6f42Duo%gd)RS4)G~$I&=KxR}=V34|2Y9B#Xu
zL-Yw{XmZ@)>P`37pLFiX5A*u!Vns}9<<fpkyDz31`Ig5}zs^0pij!yMu8TGt?RZfi
z$J;Q|YKU^^1GBT6NddGI?b=N-x4ceVgLc+e5Z4Y~rRn@)qkfrCHjm>cYUJ7#aJaK9
z@pNZUGcOl?{u*Ow2ZFz4OlviakW(yaosF>HW?fgJLRp0pvAfO}Is%!=hU`<iB?*lL
z<#t)ITT8056k_EkF9y~au6%|9S$-|@T2+Z?Y#YFHtdrF&qe(NFpq6n<n%o#<we#2?
zR=<&vVU;}qXpCpmP&d0vr>cs^3BI!k?z^qr`>=N$mt*$>do4@m4}dI3#rc*x@a&bm
zNu5spi`<;bH#b8Jz0!Kv8E}6`R*WifT6dH`cWto$NPG+oR2nVx>sH%%Uwd-8=++l6
z%Z-15?Pax*Si5()QMFdCEs-Dg@IQ$sHExkuNzM9ZN!KC^FQ0^(O}9~kXgP40EHl#E
zcD>6*Xvx_x-8Zo2$GeMOG~^N)mvSj<`?Car;+m;3y~0MZ-N{ZSvYk-L?{)MBti{29
z1^VIM?PSLKW8qX3nw>dE#D&rt2OLp3r^nm$Q6z%-PfR-#qPDPYeKZdoF|#gZE7V7j
zCof!-2G0N-2KsJHY(uwFG>fRUu9n|nb;&=?OEzj<Ui=czx2Mwo$7=B6Tdoj|tD%E`
zJ8sh+Yv9S^^?@mx0ctYM@cA!n=B#LwyGE_B29L6<wLHN#(?HDaqoXHYo#O!(ya$1j
zaxBW%4Ci5w@W)VMWfnDblOIN#KU>kBe*O<3yQ<1vllyv0K?B?8is7@@?Ym}_J9~8f
zNrZlA0h^5m`}KwB0D%%T{JYk&B`DzUeoSUd&i390-+&Lx`w&Q|X;O%-TUsNBwoZ8#
z#Wlt>O(;|POME3kT%=mRi1DGFnF@wsHc#{T0$4CSWFmp7#1kBYPkYXgAxs~ZLEJpY
zVlNy-Vjar35Wj6;%wFS|ebuDm>ygc7jn4b&<ZfoItr|wSqP~c9CZNRv1L2nna?53q
z)D7yL(#hq{?O9g^r6#|+PVF)WwMsnGvFcK0MP|u!mWrPnokHil2@il8ywk84qOpvZ
zmxk-XiLnd5C3UMpW^BrN1-f$CKbWmJ_qx*pC<Nx+Z53};Xm6<d9suaTYuVPaTQ=V>
zR1bjj1+~-fA7g1RmF~N_9{_vx8Cy@2nP8A~p@t3qt5K;w5xYT`8%=A;qQ8s3^&GnN
z6UrLNBLV4iv25(^y1-AUyo5PJm~tCPKvk!fPUC!wmHh!wRrujDXl+O@;si4(^^*!{
zJI$uU5MqZkfbcMj(wSiurGoj18(`=A7Kds-=u}VR!E4TCtCMg*zO?ehwr+;HaNIfB
zRcn=F!}9)g-|u$hg;nG9goYM4kv6-4<-(c^yC8lO2nX0Y&nEl0I+>@&*M`;M{?v3T
z();{9&0;|_Wft6c2?nKkN}eRy$gMB<!mXFB5qXGgH#?H?TB=~5&oJH<v&nV3%kCrj
zFp9{$#9u)dbNhQ1d&|p=@lOwUub#PW@DW*vPBa@VQ7$#*psHI)R2Vu>$H={6V?eMa
zJ=f;jmyYN)S$*XA{OdC&&xEJi+(XR}<)FYTo4q*Yrzq9F7iI)@$T8BO{7T79z-|6K
zanZ79cYhtpz^gjf^l6^8&)$kE^qxVnV^q!|<k{05OLDAn9odKlPwu$QRb69?Y|m=}
z1-#YIo#U5C2~T_m0dDIprJ9M%PD-2YrQ<&$cuh?|9pBA46)>xbePQm{1M7nouk)}K
z4<*V*XQ3na^>O`Y25A{-CYOv4fI!Ekn{BnD$NFRBNN;vI**uvQLf!L1|EG$<O$~|V
zPH4722Oe6&E&JS`olZCDIpVD)a#VfBu9u%f%5L>zz(FY9(khD=WT$B;?-qQF4kRAn
zt~o`~1}h!Ygcs%iR4)g&@i1*Q5eF`bUCb!9kY5_~9N7lU`sDv{1cekB$w||A|D&4H
zoL0gRnINcwf~t=AbT!q6Zi$dZ40cacgviw^F&+S1Vy%4D?O4zzRm6!ZBC{pp_4^<#
z(ZqNTm!OiMjK%kBO^O*QbXiCz4Z(l&2$YR4oQJ{#Pg?l$(zUDH6+u=ugmd8QFeX$t
zO;@&DhSh#_J$JV8u5TtQ+qTsDOaUDo3l{b$NCSxFeeofE>)DV5nfkg?G*=-$wrQ!M
zwWYssx1y!p+!Msang(OtsqjY=-L)tz&7H<1OfM_bG#CPTc>c}G*lRbVCcbaSkjLgy
zl8P-ZeB~Wl)LWfzDnYLP>N;R+G!vpGYg4GtDdZMZ{oW2HA&-_6n)`ZXD^e_O03%`#
z>&hiKFW6K-6f?QT{Eo=<IL*<SFP$jJS2q0`QlzCC*Xa>MAR^wGO#9d1SCdoo)S}%$
zWBPveyx<xo4ohY=|0uh>ENP5aES?>?>R%Tam(mzTEWTK%*wb?uFMo~kJ!K_CY=f?O
z-rlXIZ6!Pv2KjBMIjJZP;bhH~#KhmJSp0k$R2!Zyv$RiPU&`T;k~}pP3wF~wk)@CR
z>CY*<Esk+d{=Ua=_Q=3|WnaRiX1p*Wa(0q~r(z|SkqAfpIF>wT*OoZnqC72b4as`{
zBuKxaFJ;;2?1{B>5vG!tv`(&rY}~E>!5U`&|3nD>=EYv$g=GigtabqtLH)gO!isY&
z{bfFhB&?#z!Q*5KZeFyxRlGHySDbI18G@TQUhqV^r#x_^mpaY(>92%ohB{+t5x*W!
zwewElVPSXSHYT|i7_u8Hxv7b0YD;;((M{PLTsC^OF9fSz_;tZ>I%r48QpcgE6yv%!
zz4z<(0kGozVj;1NVr@voqt<}Ojw?Z}G^#Z7ORyZiq6TZghata%DV)G%k+*g0^Nsa0
z2b50=@2LLfnQagN4-KTo6vZx`;o|xX;K0LQTOD<!ai4`G=5>Vol9i|zRNpMmFm5L>
zmMf+%_a|t3A=-5YC;VON>ba~_6|vvuY!U6cGTFxMvCjhOj{ik&qge9S<#~1g{;RfD
zE*obJN{RJ{S07{=g%BuiW}fUn+hrI>oRj=`qVmjqA#F9uQ(!$~-uo$EXF%pgdlp4T
z`N1BaG3#YSZ(XF)UPSz_-0iHqT(*3k=xTCK|E=oH_?*qVH4d8@6U|L>_u!-NduA=F
zV-OoJU3$eC{Cr!cj6oBUAtP~t$2EP9_hQ0V`d^*nv)D+-^no+Ew)LHNw4t~ar!Zx!
zeHtk-g(a}p@C=*C>ZfP@J3RgpXXW?7z2y0?Br3h+Zzk35nUda5J;hq-A?|t;K_6Z@
z8uMP$A<H^>vbtZ65Ddm(vBxUwvv>du%i}x%4nHQ{w`JUYC16SQLRTF}Gs+G98?Vt8
z0CIyhswFl1u`CL23h4!XzBI4lMsr>U_jAnF&b{?A>QcCtM%y<{a*hSg6hLi#OuXZW
zj$q%7-ih!?B+PCxl$cNHSbvfxj1S`Vm-1rirP#a?6wTA0JPHidM$c+iQ|A~`_z_|8
zx3wEgFG%~3VxRAwOa3VB;AW(}yf~SeBigWB6}-DNV8FbDH8)JVGkaM-0L+om2lWz}
zF7ok|$pXGzR6y^3yOn7Tm*s-|?42&vHoNXQ>1z_-h1sj({_>yV0vMvmYRhL207&*V
zwaG_1<oz%-<3OADR57Nfd-ztm!q?mP0vIR3eeHPh;{FW$0FX_7>Sc@h4M_E9)Ux}v
zZ|@oxHMcIcY+9Ef+SR<qA$Il_&D6BH4zVs+Y8{<b>iIQGokvr|_Z|d^zz+g)zZ_kr
zP3oA65}SKIcBJg5Ri<%xwT&{q1w8<Q&1dD?*V?Nn21;NFm*X{rE{V$uQGRK53hWE&
zOhoV)uy4GG4Jh}hz9?ANbdQo^oMIv^qn;YsJ-o(4O%icyULc?xwt6wQHoAt5Qu&Kb
zoL}}iORbk=Kv}+%8kn)(jf>_k_}i$?J2{Hns5d}6Bg4ePXl}=tbPC;lhaUhRr~L;#
z)dE}7i6Em0dglkg=FKZCo=eyDmN)cp9o-DHy&_yczI#jAtoda+_cvydCH#Zk-v%w0
z*{aWMzEG}fN1ezyA(5P2N_yxtB@beLjsQwHZg<<knEqbIv(u||UF(YF=V>e3{bw9<
zFi7bkp}FE%B7nVor-l^dzK_>JU2u2>3UcrgqA_jx<9OD48gv{%MgSlgwPgrAGh<oB
ze%8kj__!iGsy1)Xcr2gKam084VO(i#P<;~oZpV_#3gYd&<DQF5`gxJcN=hOA#q;+O
zChh}3a31vlu+vk!d5qF_vTx@cm+orJB=&V50EoB&N&l8X)MLCgr=xkl=6k*Qe`ua_
zLds&V=a_!sw`groEpB4usFR-m`c+Cb2XhUS&ESiTt|4pGqPgN*(nlX2<=!RF`8E$e
zz9;|wm`AuX5(JfNQQt1*;DsE<%;+B5fAQg$60JgRI^5SdT(h$#1nP70-lkQcfpbTk
zHnz5!je+aI?=bFVFq03%Y3P3EaqomqYb6O>)`s}@gRS@qFYO)xOqV*nJi_OiHf6P@
zXJtuN_FAP!*g*X!3=mENLMf&IL7@%XzUJ3<LXa<s&4%dx3$O2`Y(=J93d&q823}%H
zp~p1S)vZNdf(#^~qFeRS-pnc%QghOUqI?;#ts+OdX*`Ws;Aq3YM;DB1_*ll<zN|l+
zY3pS4`n^058$OIzZSG)$LPb%zAfkA9@t|x$q47Js#5KmW=jMw2s+QwUzGNs7Xs!lf
zou(jRI0*^nAFKlD@moVT?TIVN-H9ENE@{pX{`cRiF}5QmiEUZ{1r8y;$-2k=HjLCW
zI4VzU%o8v7+N=POxtx`6Wrx8IL1!m1A^dcG2Xr$X$8?Bzh_(GB#ZFEQ449RtBB?}P
zb(t_JkJ4P5!TZwus-@y}+#Qu~zHG-aw;IcGR)W^eQ81CagW+ZE!u0V#PiHTCtlo3R
z;Eb(Naq;Idl)bm^PgL6|0RbY>^u1&cfHqp^)fayZ732%GZ4|#nH?`8DU5{sVkB9n(
z@}~41eS;CE^A5-e;Dy(#O<xZh>`RVWW-cZM9C$2-7zGK+qoL&}!QX=i1hl&^s*#z`
zIo(v`I>s@O_4(p!>i@<(Ik*_ByV#MozsNr7CsP5Fos9-a2`oCH2vBew929qQdKbD-
zULx3UTP2*uxs51G;2t3k1CUj;yh*cV$ouMEVcT+&6!Tg~L$9d7zW)$`JGHIgd`OAI
zgo(i#pp~BH;b|rj-}3BNB4Z!~`a3;*%f3!u|FlSnvv%ySA_z3Ysas99h=a9)??h+(
zR%U?ZGw_J+G$ZR`1d8hEX*q*k4fWqjKL9?@>fU_xjS8DW_?ijbo9M3A!Bi5nMt_7S
zfYGB`0VyG;>bZ@X{}P<B$__J<)8a1!nY|MFlvLZYvB@?LvT*U1fj7?|=Nxj|H|O2g
zn5FhlEOAtKk-iBN4*-R)e=s<br&-I5Ox|rZeA=2Tak7{Ov+}7^?yWF--kHcoR`P~D
zjYxX2hF(ZEJ?mi5Rtj#eNW3&S{O-q;0d0k~G~%yJC+3U!cAMiaFl*P(dLxR}FF5--
zY$@F_5iDiZSm%;ab(+@}KOX?35o;G)ctCDlJQ<+c*IKG2TGhj=*Uosx67l~oGjm)?
zHgx!sDN41530;TPR`b6>+3)_SO?`R~^k4I8UC&h$JOMf-c1y>vQSudnDOSRT$=c(l
z{qgTv=65z;A#9J^=LT(6=Gd(7L^gFJ<+YwwcEBy^-E#u7MBC1{ahelzQ_$u`R`bd~
z8vRyn)XZ2T10-19Tj;zMw>M`bQK<@XVrM8RuGWlQAR3BOx)6;`pl9ohW>NsltGJNQ
z8XpCQH4pno&Nx09939QGk`tB$Hw6V`pW@<DuYSIr`?N1+BHiYbR|+%3s#PLu2YEei
zDP#T^llLri)p5#Lcu}+{f8~U+TIQN%<&yNqvTFMs$MLLNym>Z|ZUqjfngQ#|SbQ3#
zW49z!+>U2G8{N*HI6atonsLv);EQ(f{ZK&f^xtn-(&#T?8+IY_TCUv=TGUDj22|~6
ze!EO5W&(AQ6|gU99QD=a-<ErVINuq*2SADc41R*R-pR}9!PijT9XSCzfp^x_A2oF1
zbG;M^ShrXsJv)XVOfa=o&3ndn@xF|s*|Z{y>m1)Ki8@f}Q652yDG2+t@Bu(yh)(n^
zlD}rTi2)TKN!-=_mDn760E`XnnWWiWqB;&fpy-E7CZe<>-u^Abm(mR)FJHA`f-NtR
zviE<sAHOwUnbhf)6Myer&;*M4DLF<8&WBw{x6RU?EHCxpxTY*!)^_;xF>cYN<vqgi
zgj4YN9NYi0F>&rgM)T-)8WDL6#3b_QCT;ED&|pkHGkO{O-DiZH;HuyN@h|$weV{m9
z=u2kjo|o1GAaA5bTbuQ!Ehy&67sJ}HZSO#%Z@^@^q2(jFf%PLB;67i_r{71`+oQ85
zm4rRj#c~`BGGUrk=cy``;j_v%$Z;VRd4(}Ovv}fA=?{Ks_G=bm?s9$eN-13q)F}G<
zN<{vMl{PQg0msb(UJQdI!RzCL^x7o80}`|?V@{JY#d@w!UPgx+0IWH(dxj1o+P6Cq
zU+61QycsmcnzWhI(hAlWAhCfzexY17jv3#vS{&~5KZ!Jh$qP7rD~h<APqG|2_9#cU
zT&LWVXO6O5YMh^j9C!^@Sp42_ycjV4^uuRhWiAV;$+r@G$ScLQjT<jd9-SWG>GFI0
z$h}|IFFxWg792C=0dVbabUCM){iJ6k){6zIp<5NoU+R02KgAs2G<d5zu$PfSRpe>C
z`AHin?vK}D3ICSwL$>%msM!gN9(60_=nt&sR4FROtVgdY2`d>YGT{){6VfS)x0d!M
zC1#Ln&%=>5fhT+n4P%J;Ct8ynmXvNLr;3V993*V4X_4OL@9L76-nR_Uyy9^5F-7xe
zP3*f33Kpr61beV}%9xzLGIHOuGfZ9X-m`<Yu54~{Y)w?m$`Mz_`o@|tURI!%?D<_9
zNhq3@b{O0-TJvhL>~^7~7_4ayE%@|*5)Z-*i+^6p%S`5heD6BG2{cm!=wiz)F{;$n
z?kfM4z03&yE6;j#_@REn_hTLKe=5~Tq|`4~zamu%&naT5aE~v$$@12i<FISZi;7DB
z#ju<h8+nI48;T$5uQ&MKF6;z1Y5bpyzW$zfPv~FwmQw@Oc=5q)86mZxo8Yj?$>SZd
zhFODEpo++T0n6^HZQz+=+K*6rkM`XqGC{A!ajBR^z0#iko8l5F)6r#C+x#NwHl13j
z>EFB0O8Bp(4ef|Arj-;V7BZk%#;10S_hD*}p?1ba_Ng89f4$D@l(fmqi_=ktP>Sak
z!)P52e7HU-7~PKu34O++*_wrf!Bm)))fO}qSbdzPe_3ZT^B_aWiBEF^Boe(lNyua3
zgK_|X5v)G1#tv-)0$8;$#>|udG9=(p!o~!^^<&VCymqlp<rW{eM<O=$rTuSz{0pUt
zck#4qGIMsOi0>1d=;xG@kP=*BvYtZQUGOkMWR$f#M%)J3JZ=Qp7uayvnkpCQffT)B
zZu+)N-q;(VjKr05TYq8I+#-wd7>xbmF+s?fAXYXeO8`IiIn&FRm@og=!VA=)mLEJd
zvf|pr$rQbop{S%D6Ym%A4+m_rS70r#(-hdLh@$E23@H+FNPR{uaOm@v95l4)iMZeD
zI`%96N2x4Kx!Z}2wYd0MBE+N{a-NH()J*NY#z%}pd8)XPO`wNPs|XFlEQgBqjTzBo
zpqjC~iq4<?xjnYdjNSky8LrFJe}jaS#viYfv=KbG+&0R6;zKWEgzl@O%h9w;of0Be
zeA~s^DlJr1g}BLT#8J3mnwEH#@j7k^2tVf~s~Hb_?Z3`2@w^P89<&>MaGrl@;Ax^u
zirx3IIR#})g)#6?I9{)$Fq&RUCRAz6Gngol2>D%|6ipr!)y5*JR^Y!9OBI^{kfLF_
z8hmA|lVZ^&BiKikj%V8W#;^1aZtozAJ?8Vrrl{bIWUgSr_dM!N0&x4o`mW@vOIlK^
zaw&5w$S)_btmGPE3B{?&<dWJ1e`veSpOCz=%A1wPvQ&Dus{l|d2&IEj3}_?u3(Qcd
zVk0a6aq@V6PswRih}9nd(^}&8xqNAb+ty_KNAQehn%0bHC9)3~NiSvw%vLr#Zy(7#
z!OID*j1Kw@TGie;O4&)|))xd@(E%-=T^EVU9IMC0OYY$4@`$s1Gk1f|j~H)39#>s2
z?m^$IO9{zr5uCsppARfHs+#I4sbgDX^<B7QessFvH1_<4X2$e97bctg$gBszDCYxU
zcl`d;Co=oLRc1mNiQdf-{I#kUOgm6iz`A0emn9amO7;ra;jY!33uD^ZdH<4$A_fkY
zWIBDcB8?|y-eZiSp=O(9c7PUaCoe0YO&!HrcHR?SF@Pw5$1k+N1dOH`yx<CTDOln>
zONGAF^ps;Kk?&}=?=_pmU5f8wh8<Y5JS*>)RSj8bE`agh?+Pq~xq?5eW95nufaCwO
zh75ve8cAvL_#9^p7D-jYemlx849e>GGfp-2q*SHW>#zz^fr7nw!GH0i8snO5h8^5&
zfII9BXf-Gxgft=NLg{;3h6dIea-<KFR$Be%467x<*?*s<{U0aZ|NnOX7p#Q;*BR%H
z-r8~Ti}tdlBG&4n6hi7Eed3&WN@naxKPi^{>W{kxC$2Z#{Sr+A#-qO|Zq1K|yh5HM
zzd9!f1NWxKP&z3>Rz-R`@##$IZ)yNus8VI)*X}=}OdvLz+76DB9yMAiA_+}hP%3%`
zQ{&FK?CfI++v|T9=bB-@9#$>c<cuo!_45Z}P9NPT*>b1~t*jy28T80YI(@azVC|M1
z;*M7b_S#TXUUU2uMwKdqS|}bW#Pj#w#Je2?e+Wb~awn2golRt~VhuF1kLOW)e4ZHI
z)+YyEY<yKz=*Z`qE$|_G!8j^kxOSM@DCitC(fm!~yVKC9vDO%k4PW&&sD@?#`r0`d
zQ{jXIEkBECKvJEr)eFVx^vh-9fC4<pUo|~q(Yjh|Y+~NPNp9ay{SIu(^5_^Bc7tP)
zuGZsK(RKQo3rNAAG9rlOHbh-I&cUv=U5UcyfcgM*U(V;T?7!4cMe(^Y^8+8WdaC@y
zRcpcONpN-XMHoy$m-e^f;ziRhavgz$1%&9|fk(tQ2r+Auo9U7OM@vcnmI8|E&(1P9
zZtg8Ec75hBEX%bo^;U0BSAZIScCBLeNF6j18Q6y|a?l)EZ`$%}Kh1`qM^!69-dSJY
zd-sX*vaJPIHlJDN)GhF8ag>%cG^c|qPr_D!$~}vUiXa*meg8M0n~(8$bkf3;t?6tS
zon<EH$z^J8DDBDjiYmqmXEKI#W~kWQ>^%jMXQh;`pxdSR#%j6sy1&Pj$kO!{Dw!z3
zV9<;+u!T%vni^G7@6<!seQ0D%SN6ObvWrT^Iz1Sj7_Fv(eyxu$e8M^UNnK+`=K|BB
z1d)q+QxuL)wECsC2;IOW7a#LbtM~YBJu7p%k33PPRgNs+<YE|+mwCiuVy5ORhhPSr
zMTb>MW1sfoi*u3hbvj;a)*ib}9mgDvO}02Q)KAX}H$Lr2yiDSAaIW1rQJuI0YE#u?
zYgb<8ctzpq%-F{GPXG=5F|n6DdpS7^c<jIioDvG-V9>E1ajq&;E_Q%M)6SRKw+6W_
z^Y5Y{s~5$^ZP(j<{PidP7@|t7XCs>uoZ!5FQxrYGv2ho)1LRtC(kEGR($uS_)95Iq
zsZ#Z<$tJIc?|1Nh4^-m7L!#cqYh||7?1T5f6WGeK>SR&x89AqNBT;V5DxEd@+QIrf
z+1-Bm!x?*tePQr8S^2R64z{Nos*rf4)-alV!}uN7fwS#=oV<o)V!<|;&Xh}=%hwGe
zMp$o24uVa)cgcsNH{aL9Ii3e3*Z&7Fjq=!z@Ra#D-fX~4naPqR2->tMSc>r%!XwpK
z!D07{;^97>Mmcq-*`1HADl3g-Ru{6HME^|lLYbj-uE-nT)aICR8%QLFse#K#i-TT!
z1=8v~0CZHRfC8^XzQU_USOOA*zi*+{DmVco`;vZ#2p?mcdnO(?v&G6}9l6O#4Gq>e
z@rA@p=>Q#2TF}qwfv}1r-eJfAcNvup&s)upgPNsWMWw3zNzSvngwL|)JoK&=l6UAe
zXAs5%A~v$~6}JOZ{M(+FQbvb2!Vwn7d^>zrUcGJa{RhK~glH&*2thzcgKdau4I9_=
zTs28G7v8t+;|RHwQ8mXVLk_zE5SnsjQZzAE5wJ-~w)Y#^nE1yF#j|G9Frup=)C>xD
z9Yc<0R@Go-X70t<yTaNKeQlhGUod-%*h#&ZJ<9rIFHRy(0uB8?Os>*3OO+@t(JEzR
zl(`et)D-_Yr_#CL&UO{|;ij1s{umMjThU+`ddcJ>b^m2KzItL(%`EZN)^^xaP|<oV
z{pjejs6I&S=^x#Hf}Z+Tl*TZ4j*4}MbJukclX%rE$LvMe%qF|;l}O)5&|NQWo+7MT
z2m*6VuTv<1S<f*%*Tn<iDUQ-!#hT-!pmyFX|Eatwc1Fb}#%Q_bgj7E}#~6{^Cu+a3
zu5M9O`&Or+(WJi-{BiDpjnLa$Yy#7Jb;6lpRN!FNERb5prg^b+s=D5Jwzxpqz+$b_
zqt|zZR$gN5`OJ>*17Pv(%OT*D#o)mHsz!gt#!(i+5X(+W=TZS~1=I&)s(A-p95bb_
zW3L_&{Whp*))a|ZNph|{E_zO+Jx(E<z^pVnx{d!%p<+Tr`wP6l-ORq><JeU8dX}H!
z^-$VU#c$NJe>@O}UfUWfh{u6e4Ha4_o%oVdKcPFm;Ms^3{aIDjy2ZD*Ete>p2Y^-8
z-E;qGAER}zvvo82u?^FfRm;yy`t`<`mGNxKOgKao@Y7MO+Qi@TX)nah`hV1zrw^O=
zA9aOWBxr5g|Kwsf_*$|h0sA7mV*(qEc>n;9o6VMMalg2C?49eUs<3rUbIlhjC4t9k
z!Gtj}a6)j}v!<(lyQJqR^gUJOZe#>%%jWxs;Zj7w^78UGVZ0#u8ZO}M375FO|IBOG
zmzvMTkZ17*dr#}FXdMxyDM8mp??CZoUTmy4#7%MOsh}>K=Q%zZtX1`sUz%~=jeeRX
z5xIOeT~r?xSf)2+P4mj>cX8aMz+r8p_wy519Pm+mR_-Dmaa6<O9K)OK+DlHib)2Sd
zcQ5V-4Y-7*njji7QCpu_1+zkEnOoCN#m}P}PW@**_jOIW$;Zl)iecrV|8c6!f1LVZ
zgPTXr98q@1sOSPOC^?jqe@x*KI<<=>wG&^e_82XE#po{q_YWd$PiPKOz%RG=ifSR7
zo7-XB<JA<^fgR1)lDB=)PQddyJ0Y(36@#*?*trcxJbT)-268^hN<_RN7Km8wa<LTu
zS*6A$YIsB5pzcq7t9@mrS3moz*~ZjN+f~oK<2>@NY_e+K+7;GmR2YI2xKk_`lnh~<
zv>Hl^vjD705!LurVZ#T&tF_8tUXCQsl6pLb4<8$1jwY7%#@f~F$g*NRHV+!B%|4|~
zv-G)(M66qv<d*u(uW^5qsrU5f(yiALnp9B0KHE2O846gHbIzyar*uBErQ~FuqI;sq
zv>6$dicKcGeW25aH>#!!&CYipuAM8(CDffEm`$lg04>^gZS`JjeU4h}A+w9A0m{i1
zoN2HZvi8Ov4$&vhADez&lkA&UQLdh*ZTOy6eeD-Rl{oKi08vw`gOfPe5=aOSmAVmS
z7UzzDkZA?h$($A;p@r52c(PG}o)Vnt`GucLTd3037eAbJq65C)kAj+@`CndM8ny52
zpYBdgRv2~hIf%#9WVYx)II+)0rNMD~m@Ya!(df4+WKv&|8go{deYb=<ftD*`Lm?`^
z$ZAgbhV4bswWgn%K8Pr)$=h20HLM-0-;URU#|j2VTOKERnz6r&v#QHmbM7*z<K(zw
zsP|wW>$jd=JpvoBNk-bp+LXD|EWCCtKIlx_i+H7`8P)v5BhAJ;+<cLTy%Q@a1lMmC
z@5?OPt@I!F3UyG|4qHL5F(lWrz$Y20^b9V{KLdyia9t|FuX}GHlrR7@S<3lV22TA{
z?~yTjMn@;kj)&W5c;Ic8$~^e)ht5s^4!vWk03|gVRfymzZy$?>(MeVlBGd0Ze<2?;
zDhda;v5cq5xx+;^3FgrLSFg|ZAUo60!l#p<R(Eb-yBat^o?7K}o`CUJGLaEyfN|Lb
zry0!Un%Y`l`45*A7_2aZO)l|P@J;HVs!Yf%FJ+jOfWX9St$G@4CQeI25Ox|-hHcpU
z>>o~`q@))t7vd)R#g-6cbN<%oXJu;`9iG*bRZ-SxG3QtZphT5{?#;OJ9l6QpGK$72
zYC@D9s)Re@8&t_uKudrSN^}J+5XN9~Wx{ydm@;(yqc8o&sh7U8&Mdd^bm}Jf3DA_%
zi<VdzbE?eAi-{q1N7y~r4Kd(QeOuCFP1S!k7;=y>=nBtV(hB^yn&C`hdQLx>4%OR)
z-pl+GOf`GQxmmDhmZ&ee8I+T6d}X3YtX0{Mb3t!M1R$Y_$8`3(KoiL@cOnfPA{AE&
zR%kiTcV?OQ9d>SKit$u@!ZNq+tN3JidpIi*_RE6d8%2=Jf*vj$W#@gWK^YVHrIPwK
zxdQDdwtty0d4hNQy<-DVaMj!}_&i<v^MdnrEL?0DpE$^&<>uL2Ll_F(PHAkYtU%=k
zHimH!inmn?y@cGQ2eb{{Yoqt#OUFjR>V~vt@h*1EKjA`CHjeqvDo$AcC7XY-&TTSe
zXw+8EZ3r~|r}1ViKz-a&r=1@Zi#zle!uPs)!**Q|!xE*9<VNKKarC9hTPV3JDI$UW
z^0PdrLJ;rXQhx1;8YSB;(<E>+)-Fhl_3c0-Xn=1~YvuH`FQ*!3x@v1{9xUPf0y{gD
z{xH1<(fi8L2%5_CWLVXZ{dtYXY-;^mXnv`E7<`P2^gB4`h)s-uaHsk3)y#ika{oV9
z5d7$S{d(>B+E9=B$&adUd5+2ULI_qiE2f{#sXH6*pqM<F|4O@Khq^*f*tw~`zfb{L
zeLCIEz$e1S$!%s57;j!1nhyAqXsf>B?2<=YS(s2>uUSc_6w=hG238&Vv%@!v`3`yP
zab%WnKW@9Oqo!zLqi7w>VXDgS=fTVyPZ{Jt;=cXKJ*2-+)dam+$XKg0SH!{fQ&$C|
zpt(}LBBiFvB(I)bXB$fS^?0$h*{I5YQnM&;Xn}^KS8r^B95<<A){_px;Pk|7E%<)y
zjF#Amm#wIJw3N^~%x*l_T1kZ20UKH<#xJiD#@tCb*n*V_AE<nq-Q&JdR`OQ$P9#jx
ziqndX#Zb_NU~x@bG!X&M_UQRKTsBo=x&pVe6YQJlqfpee6VFM=q*qg|KmQ}mByF(Q
zVR!0=#QFi@NVlHqmXZHRo42xJRr;C|N(?_+dfY2iiS%(b44UEh@j)i<#~r*WpF%RM
z5@TxMdpK*%ObywH>&5!3KMYJ@C*n_hF-yPxlnOOOy^nd<mUlM6On8*C*iJ9qrqn7X
zf3Mi^wVg4)#$fZELBMi+M2aZL-{||uj68v&60IE*E2+oTdEzgBnh{6)yr}8h1t;NN
zuD3z(>InS~DI+p$-1tj{Z%8~<`lvwJot#?JUp3;?5fCU^cd_8}Vy`K21400!Ip0up
z-I__hrZIcD@8E@9G<*H5gRWe)0W3#w#{l1!r9)F0n_GPU-21O<mXOffMy|l5#|%QH
zXDPb*1XkqGiKv@oqJE;1Z;EeMy~RPp9?VjBewt+Ik7FvlN98oM6Xk1d{%`U9$38kM
z?boH>xA)m9(P`3D7|h&h<dPRZ(mpfE-`1~DGEj#YLL<^3u6>@RkTU@`=X-LUBuN|`
zQTf>=AE#8{8e95VKi2@}Kw^nA3C1#23J+T>9r3evErMCR;US|eY$2~p@1nx|E}Ig)
zVC@XH9`3<8+B%dfMfmsF*bMhdM2MV`Mds-I1;wRzVfQhlTS_xodtv)PFyHPhN=Fm3
z2GnFX0&fnZV^Ns4PmQCKj6M|C`pup&{I@0d&LVtL)2C|u(&MPM{t(JAZ?sr{oREFE
ztZ=|p`b%aOYo&5rryea*{PY_qN-19I?}`Y+#mbVT3u`|qQ+G4%{#hNaMC@0C&N`$0
zDUYdb;i5-q&N(lp96t7}XP;bN+x9MwGs%$(QO5H~`S9v)A+MhY>Efmy=@pz8hkTLG
z8-_MFv9^%Dq(w1pAN;N@>qNLWjV<dM8-!mVv`g_6z>;(v@}$dZQsbf9N=P*C=s8-S
zN8;Id<@ee%1P_IOlaHo3lYC>&8m(7(r4hH<!ZSDfjZgKtN1AR@oM34s{1u%Z)dzt1
zSe5N1%%%Uwx^Qj_m;USb1gm+uj?q8nraOt-8>PC1B^7Dt{f|XbUg8<J{|W!{=>vBl
z*kB*k_oOKytBXp`CJRV$|6EO1qw7L(7ZVA(ppAbG{=PQI;P5%&RPT|e!!1|%Z(`aA
zXnDuY$DJu9MZ{tuyHm(+&FZ}QaUUF;5b%iI9`khye(xb9Cl~Xxl-wQ`+!tPF8T2e2
znoRuaCa`usr(z}}Sf#CKTphSi6z%4`cuw&62^#+1D4lnbymn;>5wdQe?2aGFDmsj!
zM&OvSb&>6gvLjnpuq-~fSlKa;V2&i%91#t}0W|b*DXkXb7r3IRz1k#xUYC#dnb!KH
z%kY)Cm76wVD3QY)+`Az$Wk@;v2MVtxpQ!*01vz}AEwJcxsalvO|Gh$^i&_wg{VHjb
z6G`*0e`+pxRPf{2g=MAT_Iycc&3I-k1xlc2xjrWWNr|WzIVWSS^=BYfM&ki#*8bK`
z6M|ru@n?ZSqJ14*69tx*Tk?kWVb)!*FCg^AV~g(yQ|~mO$^s7Mb6m@yF=+_E(q?bl
z{Lzhdc$|jy=9BNe%eWh75S09(S>H5}eX)6CE_er|WnwQfIx8ZOqV)*|$8!it1_FE)
z8(TPE>27|9wCAqtlGT?`%Ami_Ve4E~8M>O=leTx%Tc;UM0y|e%_L{j*4+{6y0>;c<
z5DR%06c?Dy`(*YdLTuTC$Et&3N+Wmj<Evz!s_I2mB(4gH;3yAOc6BE^_d`peY+N*~
z4tX0q@4wp6_Us{b=%MQcoT<4oQ8r<w1O(VSdDreAe@Om#@vjM1f92UQd$Joav$tNV
z4db**G24?(r#RxSL{J36g+@NEvGZ&jR~p&Yz0!8=-Q2HUPgX3J;Z6igVZ)I@XCm|$
z7qWGo_$jEEh^AGp?#ibdO!ep#v)7~JrB9KZm$3F+3OF@I48QOndQS)N$~xAEbyM>{
zgNAAa)g$j6>`K}FEsc=5&WqRX%CeHBhKJV5%?>90hL2o@yxvJ=m3J{2M6JheYE<3V
z?$)~dzq5<1?~L6$`JN!S)<%bE>eJH_GfXdFS$}2xf?i+!TvVvQOlgJ)Xu>)XHPa1g
zcIE=B@uCj^3~l{1dBTVkg}t?o!aDCZPTtulfp{@p^~nzs+g?=a&sq<D_yu-VH2XPE
z<)$#1Ky9td=v}t!O9Iy1D9<8_>;%HG!8zwvG`m0fM*qJA$^MTCp29ZPv+!&mM5gJ}
zH|R8|{3ZBjrB473?#57qSaEXv49kQ-a8H_bvMxaLFaJ$sLT0Y}5n0)Vv_f2g?&}`P
zM#tDa_j@QhdBFN0{Limk=tBK?sFO;~T&<-giPc03AL5IF%^NS30dQdvB+5)%<uP{!
zrs5>uR{H%7Yow5g6yxNEN&hM%wK&Pm%=|+Nvw_ubQw9vDz=%#5_ko94E$T{HNxL(~
zJKL#j{;ZW}<^Wi{&G~L;Z|awb^3K-&LAc+F%wEi@g(MP=VDLMPj~n-RHy5jGRHbT(
z3K&`3d7T(xS3wO7oI1ON2>|uO)yH(RTVrtl(0_cQy(+9-_RH=3B}ZUbqvoWzwU-K2
zBPnq$Y})huG_ej}EXwfhkoVt-GoilzraCPlMuqxgYVcU&+wNdtz^d8;Cwsvxc6w}!
z%7BOH)T$SvmroSgIY(Lcn2jyrhcVD{BN3!K5i&}|-13ErAo`jqyf16I$#wO0o?%Ts
zDx{^{%^l(dPJuowg&!`m=OP?=ERx+5h7&(EJ%Qmyv$iY}#hu2T=GL&~B-rQh!}+s<
zqJmssL7h6V>Q@cwU9#f1n+>hR3>guR2`Db%o~TG+bTuIbWvTyujGBr+R{1lfaZluu
z9_+@<hBC79cxzshYXu*QrzCY1IIrr_=cKD}ChR`-q(IXLsI;svD9{=APyX9i#$H<d
zO1<N!!M6eTC;Zf&sW#b;aLy2E%E>O4km3-?;!SH>mW;4K63cq?yG0wP!OABh7Xre2
zIP``8YSd&$^pr4Tz+T#(4Q%p=voZZTp%eqB*PjcF8U+mZC=bxZ23DdR?%fH9H=;;<
z(D^>DuJAfrG{)y&^ONd#X4d`AF4uw)H3Bxb9Fhm;X&B8XME2{sU2`==I7>;h+L`qi
zflnue$JO`~7#7*TSE`GDwbm_(%-cJ9A1h+yS2Md?Jvdj};!kW@n{#`f%kC?t7-GmI
zkBqXZgQC{Yvf6a6x)8i0uTp+omg%^8Y{zd}rk(HbI#Zh0e@`zd-+S+!BdqT(3Umm?
zYR~-l;nc^e%q2C=F7@d~Y4?KG9-EoKd0n@$l>yLBl8DsnN6kW|XyB`s&%*SG6Ah~~
zq}IZI(kn>;E=%kj<}A3hRxY_xwZidLP{i-AAtar=FEFiwHIQTP!7qTzN5UV@n&?$Z
ztoI!*_YnJfheX~i+Tx7V)Oa+QH7#QHACm8>?m7E*w=6h%c_}8TT&17aPvG6+U>BQl
zjflB`7{89R9!xWs2$*)4xKfuQe>et&>4qojwrOZxnX&N?Ugfq3lwCG&dkqi}{J8&l
z|5fX6R^*XZSB(%0SmCgbu-y%|<ZUti#;MQ92wi{>aE_jQiD*i}&Wy!S`pt?%e?_EF
zc?N6N7k|Y=^6BlkPoWO~n6@gAOr<rN*2_S7)5A~T<k!91*W}2lw$J&^fAe}2uYJOd
z3Km`zpaY`fTK$`wJZ|AGK<LZ)@Kv|Lx}0+YDbLahw4KYT?sSVJJiI80PA3L!z;wt;
zoD-E_eN}0|D@@JL@g`+%())bSGs`0{cqgDEfxk=u!T-piu;dh>-F`{{X0_{$*c)eB
zJ4u+<zYa%dek-*F9oHtbCuxxj?E;jD8h8eZ>nNiQe?ps@jT?j~SC*wSXlhcJZs3I#
zSuXtk<ph{`rYo|)@-%<wRTxBYms{vIBBGY0Q43kMlb8Lb3Q7qVTJfbg`qM^ftly@5
zHI0LRr(WIs@)e`|GF2s~W>;`WM|~0Zk(P*p6Wg=VOulgO467Pk*#CQ3Z>?zApor{)
z<vvXC9<ag47=&Qp-n8-g^Q)UDdhw{ia4auPt85C)fAfQb6atPRP>nVU`Z*=tnA*Ot
zCy>hISj^U_U)tNt!;l>D2)XGetlacNJ=hE-F~k6)ix(s8GciEaZ*=z4;UgsPS;Bi<
zYOT4`B&ras$sga`3T5EV;Bd@X*iugAy563ua^UC4m9uEDffFd)+-k|pfJakcv}cOs
zrxJP2T^@F{cm1oLueF9+qwf4#LlcQ@qo1T7|NG<KmO-_I_I<xnvl!;qp^CK^QX5@Q
zEIdYF90VX({i~`;c;9nu1b%uY4LYrvxRAI9=YxU1hq<8-fIiK8z7dg9a*`g0X)mD_
z&`G!<qGponM|je-{;1aRZy8mhLIruU3wikuAr%sw0Tm1TSJ&@<CIdxFyKQ|+8(@{C
zYNUfe6;1}0Fewmvj2weTL~jSHq9I|?#7y9q5xa4W#_aD#8VWbW81m3guHs^X0MCmB
z(d0K&T9pBLvwK2qgfmc*`by5OjRK`)QvtQD4KC$eQ6mS74%_{NyNZNXvrSE)#}YCn
za~0KRSY~7iyBP;;UYQ1ruYVLzvGMq&Imy5ka`pDZt!%+PA{x!_wVn@rAHKm?qPC>*
zrJ!bRP#T>Ly^v1O2}+~Jl%!+v7QGJKO>_A3?MEQ>x{B_rT~~^+FE0`_wg2qYylGw0
z3}qM!Wijp~jsBf@)nw80wm~jn0WzR;%R--(uPuz4qinx3Cc#%!RN&g0wveU2@n}t<
zfX631gX0c#thgtHyss(RikgSiYOqbdGm_Sdp#rdawlcmwOM|WvD>O#)^j-N{4><yn
ze45tUa+K|v)^cWC{*t<ciGt!s^wrM9LD!d`I}yfzuU8Y%4oTUMI73YeM6!S(c)F3S
zHCIqx8%j3qBaLU}%0;DWBI34AbU>;j=9DW}7!Tr6&HtzcwaN1EL8<0@u_N;h-a{Rc
zUlS$<)<h=<k8U?6t*)(66N8{or|X;me!=t<R+w1mrhO0FK(x^N3aKw;h~k3GDm=S*
zAWYdVzgB-#=IdcF474?IS1v-vQi7E{dFA`Y>Eq|TfP`=__f!7<!9qFvfLx*ihK!Ci
zW>v86%&~H5&b)hZ3dD(Z0$D!?>&{g8oGaz2$%$=Ku!$kEbK#ZO4Gf&GLby7;3;vO}
zWr9TMQ|fR>5Ve4n5?a&6JPR3)#24;OM3{g5<LY`Fk#Iz{of=!9$4jVhilU#6+4QZO
zd9lyttD_MKDN{yGLbUbUjreOg$0Hqj9{?wFnh4>2jkwuw2<R%wKUdIdM8eA@qn*Ap
z?kKkWgM%;oSWB7gnjda_2lnIXlHk>AC=nQJJ3*yI$5{VzLhEPuU-pY2Ojs>_RoFB&
zchaKzNf}|yn7KRA_R`NJNa^PtjaAH?NGJ*Oq-6A4#J?9a%jo|~Df(~ZqW@>E*fdjg
zXuCB77GCQJ>6s%|g2p+gb-<)!@mTqAF*jj9#`iX%&%?cxdQ>d!#}+D+Mv<*~%4Rtk
z3qglF9_I>(@jqU{Q4q~9%C%rkL+D)J{_{JB7cdM0JxwcTOHzDrtmI-nLsQN)Df7Sy
z^|~wHH*_(*H+R6F$$Uc2oyXr7OQ6JT`dl0gwBCe`6AB8OLVR^bw6W^ur?XS^n_M<3
zq$mUKnxVwhqOZU(LK!~>%AA`Ip|Sju5nJWEQ%3_O4ty;<(;vH<!n-N$b7884Uu*ry
zrTgSVYZx1y0R*a#uc3`S=KwLp05NfyAdq+yJ#S-BW}Z@C=sc`myNn*AAct2w6r98k
zNJbs+GQJvo3skPrp6YwrJgT3X#xk*`Ur2haRE@4}T4TyNjN98O<Mbb@us1*9rF0xO
zRyM7~T`*l4r?l~%;wH}h&vVcKaP+scq(8JSu5<YP^yNLGLy<HZy%un$3ZFMxh8CAm
zw&zM6Ce>e>A?Tkq2OS{)Ws<%Wp%NNMhd}7MD#AsGwujJDb_ofSHl}_2+;RrQaMnwI
z)-;Nr6}w3c`X+^I_wQ0aZx6z34(jv_bu>FxCxqP+;i4fX;?nvn*zW1a$47>-BkMC}
z-6>148gHx&>`U3SrU@Ngmqj`zEYAq(TbSg<Vo@osv^7i(=(iDK6*cySD%y2gCtITz
z5IRo$9Ynr&?lF^0;hL)CTX8G>+JdGJn$V&m_}$hr{x@O<zB1?2oyGr!x%Y}{@{7Vg
z0YQp?5RsxFok$Z<q!*=#K<I=3hGqdI^cFe@NJn}A=`}zC2{i#iQR&iqAT;T{cR?M#
zX;(9|)~xl-T66d2=3Kma&)NIz{p{akeI#nJqTTYv`U0y%BQDQQ=Clt2r)+K`dqEx3
z`fQpsY(VVU&lg;;(;YrOJ^sQgSi2zNbqADsb~-wY)iJGyn;U8XI^n_hXCUO~nJ(^K
zSJr>+SA5#*zN~oRSsh@bIo&kSJ<p#ihQ)$Pfh$RM!d5OfL>KZ;>YwKz>rT#gwb~*s
zQzI_G03AF#9R+(SWHK^vsMY<QwDvv+P9LHpZi#3boormLDE4tm1{acU6Y5$!sP><<
z*T@Gs&wEGwrpa6vq@ZY<F<*q@aiI-~#WJM7-1v!9JicGvO|#F)J{+_a5VS`Mg4G;x
z`uI`<V0aF?vq5PI(&WiXc5}z%>~x*~u&8bB<{;<xk7Rg)X@ZH7;GOH*0-Qk!aaFe?
z(DHKUq2B0|EK^&NJolNjGk-v4e77eCfh?2MnE;DaEo;H-HC5U)?rcd{V~&iY?kOgu
znK=9fC9Y%ywvO4WtqnVF<7O;GFQN>HIj}Z(=r`^HV!(15#%WQ?&(SzEg{9eG7$;bU
z!Ad&U6xYe|c3SftGRED7S-|GC3~t3|D+m&qXjF2kId~E!lF=|OU)r8Xx=z{Dz8I}6
zuPq0_QSP!ks)8|L0+nNg=)S37wow@ZeKa*mKDXR&eD;clU-Y22*)aFq_vg0VLW?z%
z3aUl>s&1Dgms~s)gAmly$t`u4TQPoeR69c)Go1*Mj2Z`<RS2szB!|EFc5nD@hKgrs
zkgYDubr1m#$*Y^U%fJ+TvgD<ULqNC^3P=XE)G}w@*4H}M4Ug#?)JV)Q?uK!{d`mA`
z)8v2dXnE+2pa!d1+}V*}^rRV|)|fup`e?P)$dW1F6Yz;vP4ZsT@O+#{kE9?Kt<zwE
zMf7>M=Co-DfiqU!jzTi#*t!`col4w}<JEq8APEpHD*MImb(flrv-3Ek@0<rCxo{WP
z%shE2&jNW#Qw{Z)b2>jWT4D^$dNcCJfbT<Z#j`hlwG>l==B>zgL&l4SQKqak)cj%e
zB-@zGv>tZHo=g9MoxRHYUYiNaUNQd#wWO;tF=^0P!gHk+MsFvQXvlBN^WP%_^vQZ8
zT3r5#UV~RWEmy=OIgZPe2AdjawHNy5wPutKv|89;n>)K}cZ?&Zef<c!sE;D)lw=I(
zLnxHL<j?Uj1=necgwkhU|GxQml7n*B%i^V2se!IZ)cY7pT;_*5S8UPvroa}a$QJ58
zM76dcyn!)I#mps)LyS(rIwBLlQ#!j!P9}%t{;cFDs%LETUxZBV^rg=d*G?7jeK&1V
zbTMdLs($~A9P?3PqB4~RU)+9x;_dIdPaKc76{?Sg6?=CC>I=N(J&evuQMoiUFz$Td
zSfTnW;cvMuHO^7&9zBh){=bg%JG??&hZVj}Ht&UmMC%eLIRgd%G6rjo9FgAy3axKX
zE(3$QWOMRXSRR|IQ5+%Y43Kvz1T|?4<Tb8ZI#S=0ljZ8!e7=6Vmv&RkmK)cpy)Jn1
zkEC_+{M%hMy5lZc`84S_y(~JCVz-~fv!!3E1h#ji?EMuDK1;b)ij|G++ul<jaHQm|
zuW8#oQ?~%WPul<{GlJ0kjXVBYC8F-NCVmfjoI<}Em!Etr5?YcVZZ5f8v<~I<eq8$h
zDzyI33DE!3R{;p;JY;tU^v=c~JHK(RjxkI#XP+?~{~v>Iv7g{~zr^O4N<H^<AA92~
zMyT;Tm86wV)%y77DG}Tjq+T(w^kz^#$3XT?{G4ee6yYj&uMUj}Vk%G)<X?G(d^>)|
zVR%$CVkl6~&h+2YQ?ojwa%2$V9Mpj*<6;s<D(byM*2Z~!q;8rQt*iSFJZ%1;2t0U`
zUYR2x8LuN*V@7GPtyv*l9l;OQ&NM%D3wg*r>7%VJpmzJM;9Q2Qx3VGM6q7TjQH>?p
z0LZc3m-k!iQuFXni(fJt=e#5i7|xbfQik6YMFbe?PDiN?xM0S?n+^9FHBd82wioH*
zPkdSpd~jA}HOxZHoob4A5$>szhK!(m1O?w`Ov|U~_FSlihj8Z(#KX~y9!ErF-;9cN
zNSvzqM{OX`LT)eE%q?z541vg;qA86%x=#L)JgIo7M$68#C%QrlaX=!YVD%?hz`NA@
zV+opq2K?rh?t3%lgJoR{V6`AIr>1c*1whk=kWqkG+>%RLlDnHRs(oktP#73Ztq<T=
zSFek9>tE)oI~IziZC*(;FHGBheLQ{&*Di=e=Gr;wG=zz$h86E{L~hu)B&YRvUvsT(
zc|R_&P!u_iSR!ofvUa+1lqr0|v*Z1T`7aVfWszP3&L-(j4_k$!W}L9P1|#)uwD>=M
zBO|&pj7tGh(X6a&ytV-o$D^BWt6lQG#bDIX8!x=jn3-D{#g@EZvJvCP$RnkLcn;o5
zf8-{aTeQ&+<dJQsD&Y%eB6(OLOenb|IqGS?-=X78UoYj{Q+@F5B|67iCSEmC-_&jz
z171V_{wSDUd=VyD=fA$kC$Ouop8u532W}r@X)PN!Dz$s{05>~+M6fW=RDT4GXeyQn
z7jRO|2Yj^jpblA2ug#lvjh)yVNxHq;wyxz_k4uYhxL_QM0>TsX{Z15H=R<^EeLm1m
z$eitKgR~mV??_ou!>Cjyl&rw+iA#~qGQrSW%8F8j%8FR&2kUWNweW_&YMPEE-hgk$
zk0-QV8RZY54jDrW{SQb&F@N8>43fuXt)=zf;<#b|c<ZPjADB6nf$~NCA#Jxqc>AMs
zfT5#|^7k4Em;f``&yK5sR-1hjn13WHS0}B-wJK2aLVaIyM*2eLP|uVOQZ`B=;2()V
zzc|f}-i5S%uCTaLf^6B!quj1J^RZ8;A^jAWNGB(cC#5Naf(EfYZKYw&SM&24*oXyS
zA((^2&LK@DR+kn`Sl!TxjHFY7a=h!pQy8@Y(GLce+1R-X+-s+FV6PO)qVSMlYW9X&
zS&yE!n3fDvoF|&2VjoRu`rP@W2MgIln@mjiTj!XZ`NtLr$=koK%;;c3q$Ov`lrR_n
zO(ehjx`;9C^P3vAv67s=aB&N_;vB@CEyml|IJ#E-CvF0d@cnz<@@F>jD`I0P{14YZ
z#2J_Nr6M;{CxHa#P@JtiQ_pe0@$H~L_Lw>|ovyW`GD|eVFkBHUB&gD$63G@B8K}Yf
z6hN(0x;81gCmYa{S5{Vpn<~$#pOm}ee>W~DsK71H$B<8&nr$ZCBhuW5Xm^oiPZvNm
z8Z?3QEME1RdgV-FJ7%I(#wC=e6#qk*y3;h9rDTsa5A)=}BpMa{5;G${qjQ>v<uy0_
z?%%4`elZ84ZS4)6apShrKPGp47~ZmGJab$5T{XBQe`3KLhB73QDv1=unX3oNdF=~5
z)OAXlu$Fqn&v;@)a7d-x6zYousda5YBT_ojemnvi;z@KPk797@I%&fmpVQ#J02`*s
z941(KNh4_NH|$BKN5@Q@W3ajkV@jd)@I5k@3Uo>RsNx&bE2EQsu7+()8GuTIjW1=G
zR6xS~AIa)mTT7;tOZlIxf;1}&wvBfaDftYG>eSPh{RuLzTRT_jEZKfjzaLhhpn$^K
zVzdUt<a^NyQ#<;4^&{IlZB1&Qvq<&_a6|adAa;3yinpwk*}8SCK2CB|tn=8h3Tif<
zpA}Q7=sxt>s*ZD^<EGw=)_b*6cg6Lp9DBPI2WRg;#!EUBwp@S{{*mZU*vQv)75&|5
zfQJWuu0Y9+`IdIXskQEfjqAUk3^N)TlaX2|^DK3S=Y}f@t4<e)UQcxFPDfCC=T}07
z3<b^oa(4Ys_PGHX5ItArTlMK04;9kIYu}4PW}3#);T8@CE(mI<*0S#>7{6*M(NBAJ
z+gLf#cvYc!C3SvjE2pBY$4+%nxG6@{kS3x3VE;HLH8|RM&DniH^nkd1S$gVH|9wL8
zsodc_Bve1=k~SwVQ9L1K))yQ0i|_=6IsAY(dj6_D562T+m-!rhD7Ai1)YY5&!#v69
zZSr=8r?NMdUw+};zaq~*@a314H3Y4r(qcSiWa$2kbQFzM^jV3|ASk8W0&K!-L<mBx
z-3!1hVs94p;%IZz_vVhUdjncH2lwVCyMaUV*E3Ei_xQM`C?{gfX;ERet6;~46N7|=
z)a(8W%sOJLG;In4aw<T81D*afft`5yWFn&sM~mwU{oELf5$b*m+0-uleU4MGJCX$a
z4^7=as-EbI0^pL{M@?3eKem>yd|NfM-(M&ycHK<Ck(vl)t+Hvn&j^<rkiP>_rxw2I
zqYS#ZwZbn7?MD%wcULv(s=n{Kr>s4v6F9k{k98A6J+wifxz#Oz(ZiHm49yyl=|O)i
z$Mc-Rf*D;kuZ|3-H3yMM&(qz8Wc?dSs{%ky;VmaG^8M<nu8F^U#KHR0XR|Tm$2MSz
zajBD{HUr&J`XzJzg2}e7({C~i3ff*OS-lk#g59$P&TEFZ?qGoMUb?Q{P`@179Df+k
zNVfHbtw-E+rL0q|U{R|PDtsJdSeRKDy~m)R&wzCzKT*(sER!T&<9H*;TYK>?da4|4
z+{>m}Wa!Mq*!uhn<%2r{S^-KTK;z$K!HMz=et{npM)1>4zGJe5n-2C>oqSCuDkFBr
zQn>3^{$Yu!m)-=c{xF1zVT_y!%6#FL04QdWP43pms}#0q8%Eg@S>;4C!y~)xnM253
zKEPYY`272r`RPX>*#UnX$C?4i%7&(**g~o6&%>M9c4cmkqNwuD#TU@U1>HGYkHzz=
zhn+8pR)0>F5=+QZq;SS-sTZF$kSwgMueU7>6BT%fgAws(daHuihq`7tAH6F)7#c=Q
z(O=JRf$RA<6+xBbXa!3)J0WUnjXgb!eipItOHO)ORbLvr0ly~_$h)JVX(+=-k3`9H
zkqhs0a_BuZLZtO*l9NgG$<N5o6EQHj$DQ;u`--&1j3t=ZDE;x*-*)Zqx*CE7okrKA
zj7aA%0!r$oKuL3#v3<aDz*$;I=4z{5d!40N@M7r=j&Psusi6|}W}$S${bHcg-vnvq
zEDJ?CJ!>4g<OX9Pk5Hq2V;IaTB4NyssK<N(hmC=#`yTD>di1$JdU5K7wc^0m4S*#_
zXM@AWCDwl2Q86k;XmZg-4w7g6-}edsUw1>F$vR7_fBj(cBxtd76T`Um&PH*pdPH@N
zF6!zrgiH0J6_oA1_fsc!MMiS#y2;^vX4<Cp!JIu39>;|gGBhG$^OetZzQ|KHubj_T
zKJ6E28$SITd7>uL)JZ3=^_r9%ty;<&O{t12uBMm!6Gaqq6&8Nbs{E56PV<}Yu_=CL
z18QN5B}ooPI&kH01pBmPr9X;g!yB(>qZ>x1bl91yOoXr{GIu;FdOgYW;T!iltEgBv
z0kp5E_QkvR(&?Oz{_bkM+OA_>nWeB$fwc{>aq-4hW^4+a^;_(j&Fy<_e|rK>pK%bS
zBbe=ZG8&!IA?R{>`jD>SWGPc;+x$HtcR!l$5_o?c8WF2jr*j;5i_m^k!wxYya2a$%
zRe~o6JHzeM7GLdBg$fq`)tU~xyqeA#LES@5ZE{Te7jXLlH7|sZBDnA_siJijJ=eBm
zC~kG&i#j*f{jlG%{Z*$2ouKFL(?dumiD1{TAHN^<G+(Z0u!HXt<=DMD{ns_|MGu+?
z!`A7q%b^(0XcgqNN)p1SxQy{2FKNE`8$Fk>l<2vL@ZMQ$_Y<acv%E?#%$#Ta@YuNQ
zVgp8udU%nBf<Mq7mX7M{!+m)@pL$f&xFxqLBZOwu0pX#bZ=UEX6(b`F4K_k{;u87G
zpVx+{iOhw4=4RV(44vhI5T`7Q_iaY28U{kXA`}8rGJXy>-)3Xh+Uu8|O&_zHDhmp{
z3e#rX0*Uno=%=BBAtpV!$!byeq-85u%p#PdXS2IeM_+71x(9kkfGY5M#$Z-YT;Z({
zK8_U5u;p`p+3)VgPhm@E{1tSD;?vNQH1oTk>-862Ayxg@^=MxA>^%M&!gj>Ve$y&N
z7h~r(EDABI6EPsV?I;QbEPs6+{U=%FD)(<f4k@OM?};Vi0k#LL)3JfTHbteoPttyF
zks2~p1>$oeqUa+f_8KF_F9Tg`-c)Jrl1-Yrm!HI#sie0ZF)fOdMZejs!2|>N`$L{*
zDCS;9da!@_J7j}wH83{2?}*nQfoG;&fIA|yfKhj@0rCsk@Kbu8CLP>qzo-qdRyZdD
z|KnpfGS5CFHj>8PKF^07ZLgyFOG$&oaVH~b1s%GXU@8Vbhwww$%f6Y6*QHBh>*}Ra
z?<O&*M%^>vck>^JOh(nr?hYB6$J}V%e`XUnTbILM1iekD7(9C2=hp2DjwfQq;|&NP
z5TinSo-bvby8021B@a3O;1d_I!anoXcV4C2+25u)swa`5lqDt)L|V~#T!^^ql68rv
zK%%x89TxHgD?RJiX*lXE<>yUiUh$6hptOKX7UAE?cRKr(hn)9qwTa2G151xnc8do~
z)##}47+R|o8VSd8%YovG_KMGRfAY`oj`?)EDrA^vtl4~hgNp0g3@S$!l@2cjA{0iG
zi%)ly_>XczqVB!c`Oxz$dyI1k&FmLD=1BacI=egG$x|S*A15lssANF10cu_NS?OsE
z&hof_yQPSw5<9gEO9c!;7&1e_y2X*qVPyxrK^x>d$$liaS1&@s`<&~uL{h;x`G>7w
z+>-R`^wH7D`9DX=m~bV>VE>-1uM@s6WvdS(W)AMGeg%(<<6~1dY3n9f>GKPVU6E>9
zn+287lx%x)@Yna!{U(z6YaD_~<?T@g%zfBuVQN2gNMX7e-$p|+rRprTX|M0vmqS#=
zUCtt8w+qYdI7J=fSGbMv`}P|(a6xljF*6cA@9oAH+Aq{2KTHQcE%`QWC+b9pWO{<I
zp(G~zeLj0|Pg>kHv)q}xFH=Ef+T>GF7v=<eM~5e|^bxc;7vPvEr8m@=?!YfDChdTg
zymQ^yyNiC&N&%h#1ds6^3WvAB&ZwQLr_c=(C7C03#m@1esS5v9&0pwC>E&6j4D2cu
zA4gJ^4;|Q-TQrVKx<k9l1^Oa2#3V3$2rJ(4fcnN)YbH-?V!sL1umawum#e+QHfR2h
zSt3GB)#<ViCK<JgLsgVG{N={8ZN>Pz1&(I@7T-0nq-eUnGulftVyH@`!}GCo_L3X;
z;w+(pvIZFt$JBJ3qbC55mlIF^M-p)=#K0U}C{w5IG|+48-d~z6(+V<)nIJx?tCOtT
zWYSU&q|w)F-ZIosGT2C}2^`hM@s-=-sy|{!W}q>BjE+1xHMLmR0&d=qGX+xlO2&>B
zdk~YkU6y!LVbS`Rmb(`CO_zewGLLpQbIbb$ZHzR!zt6Uxzq!GP^;AC+J~8HKQiWfN
zu*1}_iIbaefD_pA&zif1uof2!*khRx_TzR=ytNO40#Fi7?4D^hcu_2kju_q>$FwQ|
z*FHppmzxI~Y-!rq`YQ;TuiuotD`E9RMW02|7{g_U5t{o&E%7Z8Yd#_Drj?>Ihke+|
ziQdeak7QY@#wXI+ojUH`2!R5AB|ZivgX!%8L0q=e_K<|*LptH$xvDPY_|`rcY^Nq9
z%qB2wClOAMb%QMppZ%QY(r;Y&(7K|~T3%2kv8$u(8~#mWtd(`^t5$HsD7d<?IKjoT
zXi#19>+*Sj%I%6S-#6emGbffb^9tB+W5M~wMczVvAUtdv6}|U5WLut+o_ALKhiCn<
z*;j7}ivLYPqtL=8r*VVTuhK=UyHkF^gVmNH9{h389HhNYB7+m5Mhp2a_(tbNL-2gh
zoynDMdY)M3H`!lovMr0q8?i~x^)@CTSSgWWgu;uLF-$cn=!39Bm(oE~UlH4e53?2B
zA|?<Mfo&wxhr?#TjoTc!0XyrfQ;M2Gv>w%eeF*6M;=Gw?Dt<Ep@(2f@Q&iW#-&TjF
z(I<ju1-!DQ%w9Notr=@2jT6ml9<Rpnq!H_lQ!_u3O|+%cc**=Y?_ndMmx4VO2$@)p
zwBT4(^Ljia=%5Qhfn)kQ&9sqXGIZlhoI=8G_LIp;YyQD}kvT-MI^MKKcBY8dt!L}h
z=z?NuNop<l4>oGPTZM9QKm85J7}qG6Ca+SK-sp@`H!VCXDa$$Bv1EM|l#G1E*su&R
z7na+c-G1#K^2Ws&V{JXHYAT$j<pfTuR8=8DoODHK_}lf>@sUjEZA=K8xP8=vY~6jU
zpzbn^iPrNoATi8MUg>w+<d11qKGN3Dfswvx|MQgP{~L4s|8MJ29~|$3isIJwiC(uS
z^Kbvj`8Zp09Lp%1C*m^c^<DB#i5Mscc2|18$SPCm`j>fWyPm{3OV|?2_rh@3GIjxP
z%5?#sbaB&$BinR2Buq5T35Z0>?`>;x|2R7={$pcxyXVu7JkdjxTx_7uvT$#<*HSD@
zjK(~0Js}1w2pm??Pvv9cAcj2eC|*|RE^D+NT56<I1SQOmXl-ZOxD*zk(qtNdiA(;8
zo>W{9<G(ld5}rNEXZ~WpW9o^|BqCfd7-@iThH$YDarTRpnv}}WS!#MC@sjahFbe7g
zhR}$Jrk(2&hW6YK4^_57Hmx&z9Kj)mCG0d>X7f0I*vRowVyp6}{K5<WzT7r9d?pQe
za&lz6Xd?6t@xuI*y*$^69@X}Kz?XibS%J+<CfC1SN2goD?6mkVG}gvg{0bedppe}z
ztAfxu{`Bgt&iDNapyK&gqp7whQd*M9n0+5SD0~F6IScf82z))c93!W4f7o#+WyWKv
zdq66$q6KoKC8;Y9t2M}qQ<KLaPGo_kGw6og7?RMtO3ESr(H!q*fBrm57=i5^9ks+V
z4TU%oFw45e5vdcR=tQskXm&EaUem@K{-NnXQ)k!6>s1Q}MvO?E)DnI1RjC8<w|~(*
zAs2f~*2K7zR(JxYC)FuW1;XH$>fqeGHm@#8NT07QxEA9_o>?dx-t@#&ZpA5#h9)lQ
zy{FpalZ>^QmNM(nYSC4-LpsE=mASE~+q0Th8bAjU!(i&~&V%<Q4n*xNs|O8Dh3%^3
z(W5^d>#$pU)JSMD*}bc_8K8nLKu*!`R#BxR^%6jLJ=<8c(+<Bhwd$;qRSWe^>>H?x
zm=UC*adYPMetR||Kij<=(2s7maCvar&0LPPP9wgWESX1@<miCxr!<|0ef-K?Pbcfo
zzPG-L^1YX9`_P6K=z;@w3wFZZzXZw~{m@bq9yL7}+vnhzg%EU)y_a~!da7pWO+KJs
zL=ueJoxyd>(34Hk8@AFX{{$OFN}ZkWNtks1z~q#P*=^{?=Edkj%<s1Q9fb@@wQHIr
zI*l~PsB`xQci8W%-RP@30~U7P+S|M1!ofyTK4;Usm6kcbwNI^NM7{jBC@a4Lh=jKd
z;ChYDfvpI59kRWc<$<P<<9*B@^{)#W4Awwny?$?z3|>>r%lF70m);K(in?Px?@-3x
z`K1;nuXfHz7;NWhbL+p#d*&l?M$`LJ-pK>7)&l4oJEVe!p>Ya=TeX^_hq(+Mu%CH$
zP4kb$uGZMhVcQa4a|PclQjc1iT%gMY$|He7Bwn|MijNNa*PonSvFU#BW^vbAr$<zC
zcU`}0pRnoN524l=5n9N;(+?DV*?$ZQ#B|p<uwBf}Cr8by>b{6127P50?3&e6ZS$j7
z;(;!k5B(G`);CAbwp2>doR<D*5M9(wSKvYWAaV}Lw=wlCzDh}@OkfbBL?sw37h(F2
z#*M)23Fo6ty}v@jmfFPZ(|IiYv-*{R+}I@fm|*vKmErl9(1L<l2gtVYC7oxNmd#qw
zi@Ez1vI%B6Hr+;5hKKGM4&<aWkmC?;eS+a^*%Z34%)T(QfV3+Ys;uG0=SZ*a`f5GL
zr`fjgHE-`Ai2sJt4iRQ?vRQxQTTJ;w>88>#`aAw*=alQIxL1|Qtf*I-Jti$ugvnuP
zxyWrruH;06<A_>wYGiXaq&3|PG8u*zGNgl(PV3F=<Qa<A+ccz!1*|!(N@d3Iqv=3R
zFeF_%?fYi-q7c4KehbMl>;6f}ESq%HZ$84&Sq;IdLXXt}?DlSS)CZVs?!+JElE2yd
z=dHyWwrf@0t{|v&(P#hwkfWLNT_rE5OAyz4C;Vvso5dl<+-ffk_|X#mK5l}$D_e#3
z=`VTz`3;{J=XVWZqw3RiT8Y-*=+d{!|HK#0FIG#{8MPRWwKhf9CI6zj%2M=%o#*TH
zznIe!Y3+-4D`j8~FWwQnc{hQXdG73`>l8z*?Yo5Zr}9k=SP_7nMH35A<wC|Vc?tS+
zSzzJ4r^Yr~E8e0%Ccv*`%mHU+??Nc=)q)!_mIG<BZrQ^eOkQ!_eZEel=mJRJTVqR~
z4Q+bRV!HK2`nS}5W2kxH-GVhnAJw><!|IU*-FL2mFZ&9*kIKR9lBkkHuVy2m%!u$v
z`e~S^reQRshg#znGQ>XGsMh8UyIRdA*uTvIoysV35eIs1JYZNZcfV~~liiOL;n0w}
ze%a_-HA(zOlGDH*7&Glr?i`KUHy=<B+xq(pMzs~JmphoRnZ$dYkoB_Q*rcj2NY=I4
zGVneOzBASkhWLe`_kp8tk0$1%3S0(DX?G4E*DRZvnZ=iq<NuK;2&CmP^k=5UFpOyE
z`-}&EkI(n_QX;t|M>0gm9NUJT+I}NgmI*y0X1jGtdm>Q!9^XfOLPJwVl;Vl|C-dAl
zXaWJW=cNdN0zY#I46iTGN)=&&IN6iZHGEglRG!+dd3~h>FZAu8kxAJnlagHLy)z_T
z^Tca@_w2CGp@ri`TZkVLTp@UtB}8ETs>yGmuCqs8z>oy@5b<ilqPi1IQR2er(Nyq&
z_=Jk!jGD~wAv5-jyv&Y7d+2?AY|GaCGNHoviWcOY_UAd4_FD%DWt_0`2&LbV;(!13
z194!uwQ<&GBdPyHan}gRB}~x-R=Yh=V{z%tw296JU58$-)%xzoC+1o1(;#Rg(LRnH
zMxm076<-S>7gS_SzN?~`DB>n4yC5lYL3=c0kFKAb!0b$@d_9RiWG-c+UW5wlA)b_x
zuI_HiU}i@wGRJIAA58SDcw^CwUgYR+`atgvL}dWAhHkWhqBPn0Q`o~UNv)*nIC=j9
zoy~tFTN?rT7K+qcNi+HUmGJMTzNuzbpUS($Ogq6@T=r3fF~iKXd3$U=?5h5C#6=yI
zKI{Ay@GjTBbY@e$D##?f0Y14RIcRJ_F#aKd($k|>4X|G%(2b2@x^=$izE|s8*MT|V
zo4BgpvB-BlQz<3evqM6$FCu>4aZ3a5%On*MKKN})*SzpDbz51&@?_TFBOlukpq2=A
z)<2VN!@&zM5gTPLx{lKj?v8>v-;#34wEkVo=bmbR&$gxust%>Mm<@G%Q(MXT_<?-6
zS@+W)t|{tt3tI9gEl)f+-87o7$-$2E6l9i!jK)0DZCT0P<l>ZJ@kvt4%1Qr%8*hpb
z?(4%&Gw#a`7=X>5aAUb&xfW=>Y&N*%r!hRmHaj3W7~m98Uoym;*<moAM~zonU{cwV
ziBTXg_O_zuoxa(MX8>XL{-c?k`Lwj%KH`-mEzqcVQnV$qS%m+sTv$$V?tGwjdwDkO
zTx!qckr_?H!j3WF5p+8irt^H6G4N&x0}Jz{<!F({WQ9EL==$)6>D!YkeJdwt!Cs*?
z!z-ZvXk9?m5pNkU@v#`=;Md;D3=ss1A4+&zp~Dt0@-{MBkA?nL)z#ZKOX}t8w5(4a
zVwX?Adba_lr)&ck(R0c!W`=r+^2cAB(iKFRuf)X@)|eK3@@_fHdV*CoOjWS^h)|zi
zQpLgg1xSeStJ2lkp5ks3XhFRhs<J+|M<~C3R8SC9Moj~XfkXl6NjVPB1{^&yz)X$8
zJ~*9t@Ulr}1lT}hrm|jt0CNBQTH0V3$&1t^*{-<LuQ3>#&GKL9svq`)j16oQH(<HT
zt`}`QCof1P2fSAE7TPWjzVZLD;NG*Ju&mN9K@qGm#Xq0exn(w)4dT*~EZ;R;pX4&}
z@h^}E?y*xbK;UpbG>{tj`X|2<6)?Fj9H4xoOy%0S$K9{`6u&Z?T8<v5&aB^EYAc;b
zS$r_Vm$n;NQ7JDuCMulHTYqXp#8PxUW%V<o{jR^T(Ix<ePAoNa-w^~kIpD_&i}Tsf
z<@xmiGD(oj8H<n2g+Eh%5zKUYvElOeCAJ6@!km$cgI+*R<#(u;CkbC}%JXOsQQUFb
z%V=vEtfj2C4rirJy^vX%b3wc)gUZO*+eE4mU#H1=z55Qpu^OqA>gxGGjnZ_MEucBY
zR`V@NB&@&-Y2$SRU!(#yg!Y&<JLfpO^>50!!y+ZhG+p!+yM7N@`=XPfe9H>#FG)Kl
zA<uHh`ryzZpFb!tUt@#^>w(8^KfeTjqr^|Xmsqt24;x8DK7#TCeQ}IF@Q^|;4L;cT
z;Oe9C`e{gYtzAJ`k7x#&Y8sJM4It#(N?*7H$3!WSJLaALG7}bKzJh_-0Txb*ifZnC
zo&z#fsW0-SRMazWRch9ARn!VJ&RrJtnFSuRl~<cTY#f(^3$7T6AKQe=ByPzu+*L^?
zW#f3Q!xG!c{L-$x$vjU-k)2H_a2-4?pfUYa`n}q&NBsV8+Ba$@VtXeSKW)HN=u?rK
z;lev%@CKBtPx9xnGQMDMw9ao`mr*d}X>20a2p|hnDz)Jd(kuZ0%nXiqSILy4&3<~=
zK5LF;t&x4bz5Q*xNp}XRPnsNYai+ZAI`^1id4Oi?OU&}S4mU`1a^YB8*5@T8-u8iO
zTtq10e3syPJ2Hb_?2y;5)#nZJDQSji1R4ja)WOvxEy3kqpGVPGwJy)i|MtG;u`XD*
zguM|G0{lvApPBY?Zw04KuF6M7j)%eYl>mx#(KM9B)I0zO=|Mr)zWXdMwO2cd_%wly
zO;i0lNZ3ReLXf}oJ^1X_^K`a2k-|BUTct1KNgZ17JjY-ojL+%SpWp{<&v@#6=Pa~7
zItZ2ja6Y;C9HY#l_F@F~A5sah3QF*i@}fI$Vc)Vjig%QJ+h<qpCrsORYT;VX`uo|d
zV(3_R8v`(knw1#ybC_eSjAaULu9^WhsmQHl>TyF=xSw=#RjP`POY6_<Fu5KgIy^Tp
zVLCJ%&gSm?_S?_Qg_ixR3}2M6xT}G5SbFKwOrh$|qy$XYX~86U{C1g5&2l-AJ{JL$
zyLFsrv7(rKEYCFPE1PPyC&h}tqM&_O<esXVW{#qQRi%C&E<*hn*^<$C{xWDSj)U8?
z`>@xHqN}xV+XXcK#E>4r5S+`5KilvtOC%xx%D@O@7VGc5N;7v;NMJ1Q7R49};WC6#
z&Io)&a(*h*ZYN)3Q<8mcu4^6SY0Mj8vSK`=)mMtFR5Q~Cbl*E4Z39n?=JmMbKPFGQ
ze=llPYKQPBzehw1Qnqb&NY5-$oRer+u%b+U12FLUf4bR!Ch|~%w}wwCmX3RHL_-6w
zpHrxJf*VPW9@hoKR7V;vkjo^|*<Eb4!%+33v`w%NZutwy$PDGMtBXYF3qk}8V@L&+
z3Dj}~O=(tJt;M6hAOY4pi;o~z{zF(?q<*mox2FDxCL6g@)Pa__nDBE0*J@j7?ZQ(n
z7gtJ-22?ff;Ttvz$f73yh30|>vhiS;@`A1Q9;R|KC*kE$y>+UlvZhk#96*?HgWpC(
zik|w4eDwzHr=+R@q?ZTqDh27wjp?di)_UA3`$rPC$MNZJf%xgb=*t1r^7yA8jCphR
zS}#%)sC9fqF6{Y`k9gI0?{7NdmemlraH-b~G|H@3T(IMpI76aubnMC{#pJnNH)@9(
zmz9G(tr4`MxkC+uQijt|4wZEGnjRWoT4{N=S|P=oBD8k_IXdP-_9%j7pQpXaJ#~6b
z@{<?8_)Zn&eui|h`WE)Z6*!gI6=+iM`AfbKv_slu96+Jd?<&8Kl7UX><@T~#(&?L*
zdEt!{YO66<pN;n3FO^S<&xuD1LzK^_sMrIS|MsMGnQ5Oo3;ULoI_WB$-r2>4#iI<I
zlt|Aq8nx{63*4Ly4L~6DCtn>2I8$ITU9CO&y$~BO`SkWUo8;tm;O^+$7jOyYi`aSH
zsz^ghSMKKaZpO8T>SB1P#y^t(QS~np(%dx@Ff|)|uC7g7_>0wmay9-<x>dO@+b)8r
zB$o=FZ}1^uCDiK6?F@q7*tN@Q*`U4|eRwc8IGJbV^A<sKM^#g#cBzbh*kG^nrHiVR
zuk9bijpbJ6>2xJQ`1D~H5M*VcZ*>S4N@iM=QM%*kMf8&HoYYIu9~}qznIJRV;vCu_
zq_11sNBt%M@{c&~(@KAjCqKxiW4xlQ3dn!OdG``f7@E8r4W1R8`c@y%aNpU@Tnv5@
z9&lrUr+3%lEaPP~<YV3{(@0mA;h6Rx5!Gg18<Bz&CSfaX!VUwF`S&wZ2Q^_bq-t0z
z+aqg#8ElKLO@S!421F&fpfFjA>P#u(awyfbX5L8N>|K+=EHVo}Gh<daNP!SOB$`n3
z4!!cc>@V$j5ph>_>JuStf3qX`0d52Y`t2pKN&FaK$;t`@eRB?;Tz_5$os-(qZ*8L1
zrz|@TA?&n;&LkY;PF?%N>UZ47f{G9)(sHbwAP}-Er445RQ6{+oNiPlW@1_reh^`TP
z=~p=ZPYu>gU>cj4-@h{v;|LpS0IBG`HuNn*SdJ%yDiZO&8u8q)&qV^W;l{1`GAVa#
z?D-AUw-kzl=*_7@sWq-hLYUTm|09uk0ok`Y)~Y<=d3!nSU^~}8YbDfTVkz=<qtNf>
z2*%0bCSGj<Gz`ZRf%@o;+&Q_QKechC3|m>s5GjVj+g(>1?QsR;g7plk1*&R7Y~+E7
z+&jRmG}XtbmSJ1zC%O1Jdtbdq155nbR|KD1s0z^Q)9v$x;!<Nvw^_9-z(@Yt$n6O;
zS7>!~J!tEzTPf52Cc=M*gj!yX(Ca0Y9GV<bGY5G7zi(*iGwMSf`}&dQ(#<lcDeRMP
z7ZGSwSr<}E_9*|S<Mp?7oZlJ9y`=7$Pf`82{x{=F+WOHk8)0gL4I}Ewrhd`qUbbiJ
zFGRJauC$Ft5R;v%HrTX*``N;Ric0#ARQ=?5yl$67q*d#nN;1-8KSon@P`Gm1EkMtI
zN!H(uh{H?NwkhsqZtsQ=nvqb%$|pfg0)31ytZe{fbbe4X{5oc0ha`!x&hHQSJCQt-
z;#O-eh+)(+Ow5ddg%_yZ*tB@c@i>d7MnVU~r=1)b57NdH4b^KcunLIM5`^_L;Fxw@
zd<ZU?0sdFDjtbHTP8eA=U-uSz^7r$pkwyhwftFqb$~CuuEAOS~+;z!w&ZdzMvVTQB
zt`5S$DbpA_5!8&_mYRpxZ`Zg;hF@F5n#0rv+eE$AoV@v<&2RI6Zm0zQe=P!QW8$Y(
zU6xKfJx^vo+8;oGG6_Sf6*T9PP|nKLG!rfD(VW0-H3&WNbuS!Z9LzEO&_ev`w&uZr
zdvTowQrO<k4QF7~r%Q&1`MgsQ=4|syrRB8FSu2<7rLn#V=g3U0Zf9}{cz7Okc>D7E
zm&T&XW4FefgbUidLjytIiJIJrti@@yzstVZ57-C4MammzI6;<4e`7V=G|9sR*!`u?
zbUQ>!CP$!lDp|a0d}NYms$c51)zaB+eufoj5;e~3$=~$(n~0&F+n0CAr<vFIk1-$W
z?y8<{RXCL}SLNp_y6#8IFV1&c=GdBLsi&XnK+w}I6U^z^zItX3s#_&)k@~mCBO_n2
z-#d>_7&1K7Qqxi*gllRK$ontc(x7I$A_tEc60pA}#%lHF&%bYxjh#G|`}1+Arb#mE
ztJ4P);uqqybuo0?V(zRRbyxUz*TtyBKN1omPI|`DFBS84_HZVb)4HtQgYG6|n^zW;
zp`hT9F2Nse3R<w@VnCixSh2ABM^(S)SjLn7(YiD!(J5{0T%V|7_~JYN9VrxzXG5AD
zGK;QE@OwnEOwW#PB#2BJ=WxAJne%O_Q~D3LLFD$e;xSwOb??L<)|Gr#M>z{By(o2R
z<A|O!XNEDTn=XL*j%6sEWSZdgmE9L_Dtwf8aA=2>DlJEG6d&9Kjch?{;jn-qYR>G5
zLMx#fbDbEog{L)j$P;Nj&<7MAZ>7~j83w)ZiH9qMuuBuYlxd}ilXcvJhJ)iuB;~tj
zgLg;jEGsAd<UWCBx(>#=|8ARwbh?^N_5pw!55-z^v}@YcKgMY)FY2u+-7hCxXU@Rg
zy;j}u$23=Nz2yRPlFY333ABuRw5i-iKXpe($#lmtnQRMhh?l4=7@3d-OalJnJf)PG
zO^+@5OkmqOu!@K6{6eMyawz1OBJkJAn(@q*v!J2c)NbAh!9~WfdCo}9>AJtO{oFjd
zXJ7hCZL>UlVrI8BbP4ZZvpN|dX;MX^p#Br0C==<+GXdFfNQ~GjL!#&xX0^}&<(F@H
zi`Sj%B!VMG;Q(G`K~@r_2#@C4fC)?4<Ta+RUM~i!nT(vrS~IOI4-M|jY*FsTXOkp2
z5-xnal4kodgbFGejK_p(qMeM)84Is^o|K1oGLdNLZzU@cVUIKMG+LpCF}=n@3@IIn
z+wb0T+D&DL;4&d;3}GSzWM~pFX2FrFUV(7oz4}#d;j`!DKaxGrQ}aVyy~`RiAk|`P
zpZOYw@{6HqibGFGt<n%C?%Xh}xh<koLNlD40iE0cy9JskHr=a@A>*)y(C}hGAy%%q
zP_9=E6!qz_#y%4^W*d)QF;BQKwHblQSgKuEpPb1doG)*YKX9=0?qj=jkG5JovoXV}
zu9~-v(88RM9Ob1o_WvT8aw?f7mkP@lNadiv{qs0vV$Z6G8n;;ZBV+xI=reC@K`-P=
zi(oTWR0x0R83(n~-y-vF>4Pne+11#ZC)H2Ic1@0;C0buxt4x>DP?#&RpB|G$OBtEj
zp>kDE89n#q7G}-j5J0^tjZN9KJ_iVM$uOd5uPhiC)V@kazm7&S%C2#yS=jHL$ehZ9
z86~l2-NrL;AyeWb09WGYsXTb^+Z(?Ta}T^$T#5wVh6T?mZ9L+)g6f)r`T;|X8w_>-
zZgQQ><27M+TG0akYgDJ2;&C0-H`M1UOiqpQEZaepw2{6{40#YBVRzIIqx@To`I}?-
zyxUF~{PREHKN8K?rydW^j8{i7IbnxrZJvT*EZ~KWg3|V?ywaFRRqpL9mHRE5T%<P?
zW~yIMY8C&RXR-{3HO$=%na;w_6#Yqp`5V`oK5`_Klsvh*(nIyoVVax6+;GYCOrrUN
z)S$b9;VIOtf5MF5Z6Qp@8}~{Cz6}J1{AJLTKZ~BuGy_xI!6<5}>F6%!1v*K<k)NQq
zDt;qzBgPGpLQug4vrlfixKzeJVeyGi%^R_%6%C3du@5C_Y7x0|U?U1j#_+Q8k5Ob&
zM*bnZ+FDPm+^;eeej<9E>(0{cx2G{BI=Z9wZ$uoN;{4YGC312uHQQWURJHVvU0SsJ
z;;|F;%)v$pRk_1*W9p{PkC5x((2}6^@i|nkmL&BrH)=zACH*5g-coqtWme^}V*`xy
z;6CS)GytUiQ!Dxj&Tx^~O*PuO;CQNy!^HxfXdL5Q6R8N(WVGdh&HOLgo@e#KzXX-C
zs$~${l1s|06oggM)aQVI_gP#Vd*=70_C6ke_|Jv0!hTgN4*w^xYgNN%%381rF8+vP
z3tlvPYgjY9*UQqnYFWb!_y|CwA;>$8A!d8RC2(f}G;+4k%=wFlN-EyF!y%MMd0!u_
zi6wfF%>*@oAX*uc9|pja!fiHxqbW!`{hOh~3Vi#(sf9P^u}}A)0c*K7AEBModneSm
zne1PJr{&OafG!`sR4#9R!0*1jdwpTcqAUMMiY*Y0lK#z2Fje~xb9PTi7*uf7tox!p
zPPE(tDu!@AyPGQr6}>KhnknWR?Sz+~Fq-#VIOxfd{U*b=`7wB1abn%Mp24~#gR=kP
zaoOL3pih6}7<_Ngvbv>d1MJW3Km8t!NWPP|X$QMr7c2dE{t54LZsjq@u|B;M!n_9g
zM{@VNY4;z=Q~amb83V3_%b!H~Po4G&zk$}-C1#*#Hx14IBW3#mt=aG_c)|6E8?s$#
z`wKR6YfVv1L{`Mg(-hj<oAt+u{f=%0H@3B2?=I`mt9%k8P2T)8^VjLjs5Y02)l!M-
zm51?anue<|9YqHRFBHN_z>{Q0DsGdvO5-v^3>A^;bO9`}(O6nq6p3@0L3>Bo2?`Rr
z283xV>lxT?a#^mBNk>`pRv9*_fN%1>jyLYcdaCZ#6BgNa2+c=As*VR&8~QtNmxA&7
zxhd>XpW=@rN$xY(vkG8!VfApK0Q}|;_Wl5+;n=^&em79FV8-eGsQ}6=KBs%+oZSp6
zd~<+IW$)F^>>^9`)9DKk0M)k%U!QUZ3`y45nD?%ZNwKEw!B6VF1MGjlXVDGMgl*>k
z+CVs12kX{Jqodu|Eu~%id`%Lk(VB+aRwXphx^ve8O)HYiGQNaC^EU;EQYTzSiA@Pn
z-ib<yl=CfnsRC8_p2s3KpxRC16fN&zf%RO-AHG|z^`kIVswC)0MMK`;5f|;gVRB4b
z#Wi{bqY5#as3F`C7KLPh$!(0`#UhZMobAd%z50Hy%xy>f(>{rQFkw_)+dhE1LuMFM
zqi%Tn${G6BG~rn5cHJ_QbFIY&6r_v2@KePv9ORr_y?yN)<YL`-<0<BK_`ARZQPgNe
zS&sa(<tzy**Fg;S8Heyk`fawB81G75(TQcRF=vm7_B;c|%+%JivSc-);*+tF5-OqC
zHsjpxkAaa%Iq~)}8DUa0MmxBOZ{<LJ67?yC(o|2z)D7B+(Q5r*FNXgS$DF^=HLn{t
z>n{9cjdWxE-4xea&sU1{!k{~4P?3Ebz)8k@nUF<xb=C1u>qGd5@kUE&k80zDBmZ!D
z`1T9v61n4z_1JZZno&QpxTYjaq!gm8F-&ky;Y=P|5ONZgj-MPcFb5{<pMIcN`@Yaj
zqmr!Kt$V=L3O3(+^rPp{fp%K-BD=LTV7B>HB8xdQ%1bWyx08_NpAs#Fxo(G=R*1E+
zneBiRm|?V+bi_>+NOfbg7HO4zwYo-q-B;Y8_{LRQYhWG;u4i2uc9h#>B^BU=QQuN>
zyu7lhp@cjca&J8hd#0J_K`B)1JtnPX@g9lCbEpQBBa?2N)e5Tf8K@=^JEPBCE~2Uo
z#`b@dw$DOI)jx)d71RDz0ovn6ger`M;<Lx&hnUB@UQSM~%l@C+&X0{y>&oNbbh~ot
znM9AN%3M2#lKd_{$N7Wo%YFqHiq+E>{BZLJwN`GYvYH^5Pent*9R|P*rL7H9eFBO6
zW1XP;L1MAr^X{7~{EwyFrJLS0_aN~v4Yblem<zeR&Mho#?*wT~%TohCXDaYk^Hye4
zWn$`2652mW2!nyE56gN`IVXGZOm#Y199xmek?$3ok^%2H7t7Nv)UT~7{*f>-oQD4L
zx@LRe+|iC9kS#4Bk*LvhqoawAdZyKyfaW?t><>G%sE^&U(kQr1in8j#J=@nZtvuDQ
z4qa2;Vc3p2E|N+`H{8BJLhD%uJOW}FTFmUXWH#gGJmIIbf;O74+-xBl%AS<!Z?e-z
zeKy0JDq7L3G{aMzj*BSVh>0K#s<^aFBqtwFG-*F`^b9dyb;^2Fy~w=fsMDM7*mox=
zr#DwrGQ6U!?2yDSSx+B;AsHUt{eP4XKh4Osa{BR%SP{A>q)02M<XN@An#@?02-_h>
z`~^u1KN_t1l9%4dlphKPyZ8!hehc7(-?-uihwPd3YtgZ}Rt;NNxrte!!U3ES@qDQ{
zv}OB+Ud5-EgIW>ZcJ<Z?%Y(mDO+Z9~w|BdPohr3~9@u4{fi5z))I3-9dalD>%yY;m
z^+6E73Lc!`ANq?ba4x9+7Pa2*qpQ!BnS)!G37J(sn|atVvnf;gTM>!)CH1NLx0V%;
zraEigW7rgJ91@Jo2JHK^%UF#VcPJ{H1U%H=$@3Q{>^&NL^lhc}nP*pa%R(kPP(;$q
zqf|`B>@dnIMkSIazDL~>&oI^#V=U;)^!DJPQzb))vsj4S@*5#yLS3F??5x5|>aQs%
z1p#M#l7qnx{C6kpr2qr}v&(64{I4h0QrST#ooZ_Kg2{7T#JY>OBHU`wB?<+l#LsIs
zo$(*0lcL9Mq`ffAY=PqAkrc!XZ8-@gR&_%!jawGlW!LK9r#DWn_Et7)v(Tn+YRGU^
zaWPGpi?t4C)G4O&LD%*sCNy(KJ#n<j%yg9gqC~x!VJq1ymAtHY(?TwadHb%kqj&iP
zAhU2Gpk~sfKvzE&OM@JPkW3%^oZ_Qv=2y1_MSROKU1c9!w<|?7&+g%fdD^=at<id|
zXg+}oKF$W2;3WfgKid8~U)9vsj^hYcja4qR!?R}Sg3=SX<L}61y5v+TveddmPqs}$
zug6hb@qbsDbYXllE>6Xj78?JB-YALmCMgKigO1au2ybDz1!|jBkkK>45$KjT=sXK@
zSHGb8G~3t9S2aprQ+pS8>RC}i1Cqeu0pl`;Ic@^mni_2EJ^Q1eOUW^jHP@rRZe`p5
zNH!lk+xnvSuX$#kY5`!a{it%kJu&lg)5DnL)PsmjWkY2vcTIlPUW(ZN*4Wc!)covj
zcy|`q?-Ne6{uD7WHutK#QdX3vxRQkTju%)(@lAnEJWVqh-aovyopB;t62gAG9h^S5
zW{dCf0%_!1h6L6Dt)VwFAwfH<5jpUQs!>)^`9&ES5|_CpPVbpU{-aNiyV84SJx2(`
zW6!me#_megCx^BpHRu3ub7Pj~&lmiUL(=+uGd^?(BhfBAi|$PmydAdFjwN6<;LLnX
zullRfwL+fH8Iv?$jO+Ou{AK_Yw@|jl7SL4PZ7p5AD|<>SH->#Dhpr|y8ay`Dc$q61
z7>M#~LMYxjtG9CHd#Md)MAPmcls9lXUahZ!n54AdZ^Wdb9C6K=(cAT5Rdu*g&1uW}
z5r8Hir(MXqI*&ZgM)Qr`Q#i{u4~B*+-*4Z=LLMBWm`ohfNgAiH%fP01crbN{!I&L7
z&Bjk14eCu99ts~OW8nbWuwAV;`Y$%v@x7HD7Hy&1eE`Vr>3@?Jdst62k*(BIu$%KR
zIls>Cl2739t{5(0c%)z`I-SLM#>EJ<^aS2e9BTMYU)eA)>~#A$#I@(L=v@~S{kWhh
zPA71Y)YBLb1k0KpaNWQ9wm7T^tRpXYnK8T?aQwZKM|fjPa8uL{m5;H}e(^&jzch}r
zCq6$^dAn%1Kf0`>?pm{Xch&CWNxZ@gb7}5IsvuqRwjF;dOK#IvfOWmFj!TE`e*>)3
zUagrCzcLZ^gm6HWez<^dLcqZ}so7rvwy#|u;+IS@ACoYLiL)C{1bmXIgF5S!s~|PV
zG42N1O%w4=77HHg2gaQlZ?jZGeaFSK7B)`*CM+8abBNUj$^`QUxj(7++a}xok3`Vn
zTZyfP`K}Y~wcOmyjcE(dj5b4}1OpjyG7^5L<Xu0TD6>E&`V<BuP%s+IY&;MfA5R)T
z;PNFMPgdfXj>jH*bKos^(-;N@V-8)-&@nW<hGYGVykF88^j^XLOwfKFcX`|EFZmx)
z_RmLDb3??}hWAJWPT_D=!LxyJ8OmRJPdA|+lrD~+b_n9~4b*(SEM~q5sXWzk3^jAo
z-la0&@Ly6;{4ReS;<FZHc}Dx$qOz_oVZsRXw5AvG7SGUnzYP)mI`QuO3zZL--u+$f
zJDSp2IkbUAV+OSbCI(hOD00#$9w34c$mOHuXl+xh_0AFiWmonwgry1)h3Y6ht`Hgv
z&vn%H7r`Y@DrwS_tn=L_r%VlFH4pket5(hd@c~gG&L8t+iehaGwj6c=8RxegZ<3cm
zX{AkmP42O&`_n*@fR!j?GMesY$#3466{r3uMU0=4y&=Vu21aH=yF#|9xLM6Zqmm0V
z6_UTAJV}^e3GiC?|AoExjA|-++%=IR0s@K@rB@+9K&4kH(t@-k)DVyo>AeXE2vP*3
zN|7qP1PDDqXo?_6htNZj-i!3^<oCa`X1>gwwdUTLwdTY9lKmwoYvru7vrqPU-{%<(
zUv*qvlTh(lX1EKi+igpw8cPoU&VJo(b*hkexSe(4`xfqBGqHSDAUfVRIw~1$Kbr1+
z?t@IGV<mC9iTtoS^sMB$%)+zMh5QzHSynH9N)!{EbIVLq5kO*16{&sc#btv&Mnosk
zy@uG#4iyHR&II_ZlmaW>02H-iY3}T|1NJ#}gZ<ZF!R$a|eI?L`Ed`LL#KpIA3e?N~
z_YT}$ti5*5cQOzXpZHUuiKC^^xVBUpze45dBgXkP`su+P`(?O`#SH$gCS~K#+Xc7E
zhKh8Pu5Xwm!BhUjC89tuU8~-NzI$25M&4gZ#{PM~Tww91bAVf3ZRIn9Y1SXYllR5?
zM}H_^LNY_zdhq4t`vtngHOVCao~mRH_K&TURBh&bG(}c!s5(oZvY6^`jF^<LvrXK@
z`7DS`ee#>Gw0cCUkz8fk<1l(vshFe*^Q|<^aWPLHgv2Hr7ozI>88|6BliQ1E{b^s;
zXn3DXARb`xHZeAfq}$wAk<W>%{~Lh(DEh(bP3@DWb#LiQ(;)UD+p$qoCw;|y-F?rT
zT;8xjkzAZjkRyfgb9aZ09=AbLD??I%P^x#*4SZT1G^wdRHJTFd_b!V4gFn%ACIjp1
z$>|d2CQD-!^p`>h#V7B!Kgc2n`n%_>c7dDZ^54Z)1$+cr``8A2-TFW025YIg);z^}
z1t^D!ic!Tq<Tqc`FrK)gs)6;2_|x;oK^L4Q_zgCn*fxGy3V}<Xfnx%ev_BBx2-ycN
zlJzp*%Y&_tqrK-$o5OhIz>nS3fwJkLg`FW(=Q?ha)n5*>OdFTQ&Ko*Tq=hON?J-wf
za$Ua+BsDo^l9CEOP91zuKb+k`3a1p7i)SR$q~zzsUy|`o&y&v0>?&KL_2EMLd_{Q$
zb~^HhR~%$ayalCJEmI9Esr=I7&u38y$lY}NnLaSC$dfpksCzQ>fw<&{*`YOqXqA5i
z3{57LLp;$4E0Ty=NNyW(S&pn)j?6@xUqTk;w%{tt__I}VlNdeBvQ^%;v@eDSbE&{I
z0|il2d}7?nlb9eCq8noD|M=CHO3E;9UWG>tAt%O&LzgqU7G!gQ-+fZxcH*6$FczL&
zteYy^-h5=)>#W{DlirJ@FqyRox!WTkJs$!2*<`vJqbZuXwpJ@fEsjNv<C12WjWmn|
z%U%&JG+qv6css@02wR`n^#LW5i@Mzc=S@UE1r`*d_AsV~<Vu(g;@h_*qpJgW%T4r5
zP1xi)+P^|sbnHZ{c#hmPp1c7?8^g-!Z`+Sxkf}yCuToabXzmjE@V+Of2y<$395oO=
zS9UNo_wRe-td!EpS-jXyWH+3eZf4LyZjc?NQRb6N+7kx=yvLud?~l`tp1_;cruT(N
zYJqErR2mF;I@5C1vqhgBGl!mhZC-NFt+c599pr<xj%2YihO-*~72+?UglV|YKy*gO
zMT_9Y@AMsWe~nKoJO~JVpvY<cmm#C>T5;WR-c(sXl^?`VVeh5zuC7NBG&+A1y+>N>
zVKiSMbgc;;$A#FN=67amfu}eWouB5C|C!puYi6$$4udUG>nJ&Ap)u&pi%=6|pAPo}
zCn0iCuMs`Hvqm56d+D_P{Isb-7JZwyN9LO|E#g;oC(<|$N^xdAa0@z7Y{8}=yhiXR
z$50cSUmxm;SJ8+fTn!w)Gd5OY@@XbM#b9d;YEVIqNvJqZ6KItY!<$?qPPa^ydqfAz
z3x9NyhcWTvL#pg-bqa~wQuUu<e;G}SO1FO5mcb7IPu!F3N|7F)3i+H8B_j>c(%#!#
zW^a!Oqn|ea5#`HmnckyZrFBL|ZGw<K1|#Imc6-UZ)<k)p{=wBE*CSrdHWI!LKFt-t
zX-4mwZ(Uv{PRL@WJ9&A2gkCv1HEu_}>$ZMJeOsc9eUqXqW6llVm6EV9j#t}NdW%E>
zJ)1Mc9h`O$vx<5j{$}`p)Jw+}<fTQ*IgLF=GoVE2m{?p!)&D1gGW^~fd(6k#R|-Fk
zT>E_<H;lo`(I!j*JfNs~M0f~zkt;M&_R{m4!E2=-u_o}^86d5v_*qoast9FhE<zc^
z{r)>4V(D&H&46>Q$#gxeWV$fdf|bd0msedW;Df((T2bapHf?dY1e?4Yav0E$?cO_C
z?!OKZyW6MzoeJM!RrgJ>v13-A2K;x+Pk2<e*ynMsWiW2mn}}hZW>)GLD{q^flrkV5
zrJoCejFmdl&+!wmvir1tAT`eXM5js3U*%}ICY0i6f>kleI%vZZ0GReWf+7yb25~#y
zIDLIDeac6)`;!<}q+8uaOO;q>rQM`P>jP1z2?0c29=a$bEXt;A{)A4`Nm^K{!cAbt
z^J%Q!M=Rs9*9{5Ad?<?$O_kHrqO*lC_^)FbHbCa5$I(I@am#TG<)?m3etq|fo@V95
zphG|qY!6Lf?L*g9dJ?Y|<Ow{|gkOll$jwD4)5-oC#1eUt030DAOR|uhw|hM<u7b#G
zs_SjcprZ_<AKJ=ZJ`~yOi&&o8P4m{&N)L~nT4#s^;go-Re6llDK1UOf`foh7WY@2r
zH&`^v>YOeAI#3f*6%v1?<=;cvsPWP>AU&;as^U{C|9Lt+$$;llvcBod79fcmvCDKD
z1~2A&lxS(d+z}~ic%{WMm_@#<#Lq<j$FQEa@#S+#ppnpXE77OZmJK7wdAMjVB0h1%
zm0*!{CWnCgMEa4;gJ;N&>aLkR@nqwY7?Ocir~%~MDkQh-0kod_g*s^vpuwXGCoCQz
zc*aQ2Uon^?j<JE%0k~cf%6(MclvpC6PM5qHJaNj)$@%$rCwn$T^Vo+!v12%gK+(>W
zo01B+uhuYVC>7L4k?#v{=4v+DnQEoWO2suyJH?)Nclj%J&-*9<RPYu}ym$srJC`?g
zqCywVDj9@@B=aa+D4Y?$rhopya>`^YHFj#FBC32c)^6nIl7k=Vin!nVwLo_0Mw3<u
ztkUvKBg(|Mk*8zJOqsw{@E-w<+MB>Pn%1w#Ph5oEG-K6&Y3W<!_#OTLWmG_dS~PU#
zWRAFcvavi4RWm!W22xHr#RYJWEo3P~V^LVzzrt+PYu=^ven10j-3*CqP6HM~@o~Ep
zNtr{<GVTc;-{pnBz56lAh_VJB!eHfkYuQBc_rVq*QIg?2QHJP<5SB+u2gfVcMC$4t
z#q_kdK+JYyBph~fKS*<F@|{;Eg@F=J_n<}rY3b<XU%1F()}_``(wvuXQYC|bn7ia-
zM&y*uL0xcxO+K&Gc=`yM_n+kMCD2VSX8+K9hnHi888&^MGcsfXK<<?nDz)YoD=uDv
zRa**vKj1g$SH+w}@elMH0*$%K-c=xI+nD&(+4F7Xtzem=nO|4uNFZXKjZVqw;mn#9
zl2LDH!?<@I1mr||d}2&>P-(kz4#aup730I`Iuv#LIY0`g1>L|mY`t%)+l(yU#kf^g
zTN%udx6ypK(xQF{ko&GTO|-(3$8LyfY3&mB*>Vh7ev^Z21}D+m&$^sN0%UZ|<ydjM
zk`5PEVYB<cdjy_mz|LM(2>BIOk@sS!)p`<g^L%EEm+n5Wkeot4cc3?UsH~;GZCrfN
zP2}>I?)-zVz^5<mb{VPBl$Gp=aOXNaL{R&ZKV9?p(mgLWFRL{PiKV-bo_4tP|4uRK
z5gaq*#!DxocXzeEEQFphT|7Cn|5?4wbUcI;923<Z#Uc}upY<p@74iYq84vlH<gpWP
zIZ0Gx^S>v!mo)Lx)J9>F8dFV=m*<y>g^S;)bJN_tOiR19H@enT{+74-Zl{#5q3mM?
zCrC+Mt#*zeAOvA~H+iYgsmEy$=R77leyl?&*`;h@YgqfHU#v3aol|-m#vxG5vDgWd
zvi<0L&%)>VbB4B+VjH-h<CbyWUFkUHI$bk|N+^2kt+OFP=(s)yquZb0f-p8YvV{M?
z)u{gOB4D5J!e6}i^nd*2tKi-R<ny9W-#8--u0=vI4&?a`0>y`@em~q4smUK$*l8BL
zJs$f<p!x>-+#ycNz1)5}ps4B3PQ6>9b17w$iZ1j+UC12{zXfA{$T&LyqqYSYu(bs`
z<tjYd)zv<$=!r<0%68qOj+zkZPyO1E&S>+<h29=PT;D+*w)C*m&vR*iUA*Z=RnXD6
zwofLTUeB&fnf7%dKNw}}y6G7PU^4jCb82TcMbm?4P%*waLL3(eHgrmc1Yq_nGbf8$
zVFLF0nwoC;0mm+pMc+Uv-|C2(Xy15AF1J*+dQbx@?dw(qrS#sFx_o*Eq|5<xndnl+
zx!xHo`LfREyUIg?>Wi~BniN>*Q~^#zdNzR3Pu}G^kS)S&^^D0$OAObEYN;kipB2;W
zYEq^oEJ@P;VJ5p)3y2tW%iDK$57XbMEYN2RJW(>!a121s7sz;Z0n>sy13JPFXOI;;
z%Sk5`dEj;r-JHvi_pkmDFhXh+<RtP}Mcq&^D}B?I{_1gD#;x;kt+MHOQc(YZa!kA_
z@RXatQ^s*0Mx-b$fj1}qm)CFfAo<;IS6W*iS=dDJqtr4hF-nM(TfXu>K~Xy7H@Pco
z7nlFbr0s1_+NoQSq0fOcQ^obVx!MqL!8gjpRRFj{EuObWGgjdD^^3o~I6I8Vrn6Z6
zM3I^I#=%nQGK4*hXQ3rNJD8n?AfDZ(DJvCTUA9Jz|4Ni1K5Y^z2DDR#Oe);*WV@%4
zK6s1*><P7@K6bQDyK%vdPx~h0saM-a_D1U|7j9<;HPgWI6%B+-qqiAl>VxDjYwGIT
zYd#WL-s@=|{x7_n301ATVtzRtLzBh&(F5e<;)UUJ8-<dp=k>#s3OmZ$xoiKW6|DX*
zt)Q*Qma?R<0LqZroWHBa0pa~J&Tuj>52vVwUs=H7o`tyrs>iXSA&+pSU+84WiY(9%
zpx_y$J0^{h1P-Ng{Z#7zbCctLW@7!H$9^qGcFaR!;DXw181Fq%6u5MZ07N*d`d#j-
zzg!{8Nj9&)+#3UfH)lxr`AMrL{-mTp0RG@z@6$}utkX`~<ms812yV=5c4L!M!SS)^
zwiCc9f>w0<My9cfVBsJpc;3X4*wa_Q9=M7gF$YKA)s|JOa4H07iL9Gy*Hv<=lSY_K
zv048zmE!Q~0g9fsM0{m%$daQdmwELKgjK8;&r$S<xEyzlW%+`&-gY<oM3?@aB|Xq`
ze66E;^hB$g9;@|Tk|wFCyQ}T$aO`wERK#Jha@yX3*$vl6+b@)vLtOWSn2_3|wNFqX
zN-Fqcee7b(c$<j9u5uG4QCvmh2O=Jf`v)2|NZS!xgnQeW+iGh_c!XTbOnLGt5;3OB
z5k9dbIiul5DVzIiMuWE)?zpX7khVm^x!xPCr6td%rf|4B_O8&%%d=lRKIe~;I+-4C
zfKQG8zDESlszd#SNSc92;^Z;|G9|Igs&1F#)PDF|^O9&=n1Uxcv1jU3uzC<d8+5V$
zRIL`bh%uCAk?`c9Y#fXV_!a+w@PU$7j?Y-uNvgDWQkRSqa2!*>&fVCy&F=G3MC$*r
zegD5s>id8A$-J3%*BY1Ij4x>dtK=fr|1(zqitGGVPHpM?U!z<8U*2t^sqFt6nQifJ
z{(j1v`JeHB9rV9W#P<K#74e^I?)&NgF1-K0ngYBe_EE+gUbUq|V%zOVHplxchiA^e
zcEt?+>mm4@M@%ZKZ$RS9Y3_|XYp-)4M_c{-VY++5U$HPVurl6WV1NICc1R?tg`Apk
z#@Zg8VgG&dMT>^EKCxIp@sx0oK|Q)EC;5GBeLKL=Dl1hG0X@#_>~Q{y1DV(*d)O)E
zMMPKeDDM^A2K^>n&ORca>_hXhG>w`f$6ZTA%S9BrKQCnx`7KT@tUf!HcZcma_-R@h
z6(K;FVj_#JW>5yA=KVy)F`HEI`f3225}$9c@Fd5KO9WBjlb^YFbIl%eaH!=hppBmQ
z@5^E4OKnlJ)s4m*a6M{6DR;Zh8d2?Dh3PGV>pwPMGG->Tgvsxf-ta|HkMfujbsRK?
z1fHIC)#AV%cY9c|r4OIuGuu&d7f)Xny-rOy!`{Z<?%Zp*{`2=tK)|l#?`+#2TO<SM
zD9NB^+oF@@VJDr?yKr-zuO@SiwNA}2CcDuQ=S`5^tTe%2;xA`@XBSWI9)1z(+o5_Z
zI(16|OTY?qmYHlBqntj3?VB>{PH*P7XSR7T@%JF5n7+V;4EQtMSL`Obi`;XCD)_*O
ziB$#1TeP`E7V`m9EF~p34=sCz>uCER&a)NhrQK&-ZOPT0q;;<={t={~_B|5s4K=E9
z$Y+qoXCzk9v&A90s2>4p<ZTdvzA`VlZ-m{pw%?V3u&;o{Bc402ZVD0+F*V?V(8Pr5
zVb9=4y4R}94Wbz2eJjy2O(FVBK@-qV2bR$afxL*S-+B?sJK_hRcxEL7n(9Kzn2k6t
zPV?|pK9gE4fHEEx>mb9H$lZG9S;K2I)~0=WN6Z<Me)ZcE%mYm3QK|bY5NZB%^lgOi
zo%M6AF6P-EfI+t}6^5i`Nw{7wl=K6p`Lr|9J1kUZ!wT7_Vy!XMQjeSel0~AgVvEAB
zcaw{6WKt(Kph((EKRk`rJ?mo2`K`frI7jdODq)UhGkb_ji5s0ENuhQQt?NNZ{M`u1
zMH;kwp;WdWZ9(*;#Dlf$=OpxGksqt>t5u31c7QEyFZi>6PoHsXm{G0(3W?);U3N0$
z#Q&``8`*;_c92qzYCMce;rr+>cbp91x*$kxWL48(_JHuUiv;|7mT6f9bM~beWDw%H
z^ENRnw?3)hcuC&eW)R`_nd&G8o?d3=A3tw978VsHldBwQ^63m3Lclk&jmr7nZ7wxw
zTW2QV1Z_Np_<9GA4nWX2H@Bh`nOI;<LOa0br-!DHFG6Lh3v5;_0Kto!b(e*xHt=%v
zrxPlPCI?qM`dhoubPyvnmD$iTe}zVNEPt44B~4?+Th1Lly>GZ(@uUH2-5??KDOZl}
z^=L$6GBwWw?;t*!*RK?`WV$9&y6h@2lK_pvJd57cC`e~cD0is|tBt~OMxxxhwv?fF
z$G+E&bwimI)H}UUC^p9S{`h7+izK-Qdu-OdO~P$N^QieQt)V?#sb_sTXKrHMK}oK%
zl~vDnuR*Q6?eqaI3z!8p)a+U+tvLivAcLO#8Q%PrTSX&3e?;=TU#Tsc{+El;TVocu
z^lprx$%0!gD8A@dB7uUzm$t0;q2+oD&fd<}9#)*_DYhP=Iw9DGUCFdT>Yt>m8EIbl
z9d>FCLj4YpGVuTh_qO4NQ6o?67#3Ow3rriCqfp#Igey<cDOY%0NxXT6u{OlVRlgPM
zErkD7l4$i@Vzwv#9Zu*wYI)dv_}tNQH^8l(j($tcKH7hhjZlMZp?^btlJ;k<;}A=3
zuh4Z*J50#ut_qAcWelATdR6YmK~h7_rX*2IsN63iph{uG;DC+G%hKTpz54`G28&uC
zSHh<SMAQh^fW==OO<C(9rgr}b)Y<^woSWiEED$nR&Abh0V?eqa#(=5RRsP6U9o8bf
zY{bY<)MsOJl+ex>wCo~+hf^}#?tK4}$4U6kleZRh<l?>QP?~4CZ0zK<;#!1?|8N9X
zoY5e5>BH9_^6mS944ENEZ}Ppy4W~6V%0U^6AkD}n9rE_*f~nsxWXyQ1k*)Lg37U%P
zu@cN&LF@!a+{XkyFDQRhX(ro<T|_F(`zM2?UadZWTT#`qmMz)F!+98G<-$K~3bvdf
zZ0rAsOGl+5zao1|rb@kT=%Dzz*6;}T$uq&nlhTgHBzuWNdJb(<Th3$xITH3dBy>5H
zx4SSdzwLbCLxOcxh#zK|zsdA@dKsPS%VY?J!%$gZsf=M;_D6*c@2M@hFI55?^}b%u
zWV1bEN1ryCzL2*%;=hP8sga&_?~bh-6=2N)sc6CWC=bHQ-UEoHca9bmQv7=co0o{_
z^UIJ2yJwd;eWtFcvH>8Odt~y*{Q7-sqBY`qRPB7#XHo&!wEo~FQ$x!xQ}Uv`oOrq}
zR}8zG*9dWJdeXefq(1g=o31mXi>;y>zC))o2`6lK?-*mj8{0;viG~7Hn;$=d^XSWY
z35a=Cir>g=XvRoS&ln_%(^7wBAGBk<mYR@*{Vw-%p@Gh@Ah>U7rE^{;YUuL&p1lCd
zHapLjWdVCjeRz{~w8x291)_`ywj-=?SO|mVI!oJ)Pa#??Y7H7)>MG7!URPSAcLa{y
zkaw^AOKohd=({?u-xln{h;~6Zj&ghtldnqJyz`&+4OabrL9%5&Ew9yme$g`5O+WzU
zy@V@u48<#fR0n&Aj)IFl)?;Glf-uI6RjU$=(U9EqV&Li0l~YMfY_3_r`<BnO7QW*V
zU+L>j9mPNa=vl!QTujQG#gA(N#jX1h)opKTKYF}OClBmOm>oXpN=^krRgP<D!uCIh
zNcyy8oU&Q>9;T&kIbHlCcwskNjL7`l3{g}6nQLE6ncG2;DMzz5J}%e2PT1wOJ}Bi<
zS!tO#9zPFoiUlCA!Z4xDTA+fPKZez<zFUYIXQ$=IZEVV^(kCijMXj9@+N+N&?2UD~
zJ-HqB;@|(+eD;m9?sfhbgIDqN`~%9B`;|{1NYcRh4`dHb#;5^1;wR#iVhqn7YLDtz
zE=0BlWWA>9u+IEdWqIx(6<~#Gb_@_K{dVV$v&_;)0gud<r?=nnNSdSbjYiTG@T`|$
zqoC44jFJ@6QS}99h5EeR1!qARV-9Z%Qy(dQfXq33M6zyg!t80SZlIX<GtSquUvGa`
zv+cuP&OQ_Jph#Lpmx~9dZ#T!Hf1Uur>~#xI{xVCePu8xBfT_--e+lS27GORXU8e90
z(`s?fGOHE6`GOLqVhJM@7uNy5(dJc1z`oh&x#C{ek6>#xFnL_R;NpV&p#Q@g`Qpe^
zxqY9mdJ_KqAHkcxyU(L0El=p2^;c7`KS&aQgkr@D>cAQ^@4(r`)d>tdFEqRI8H&qt
zej)eqh7uL9kWO8zpY<+Rr@3~1>3;4o9!B$ZPua0|r!KWj4$NP|Kk^G(%D9j~?E65G
zn6>kDQsgb`r=o^612i~k6n1F;%WgEGppaBed4;>0h<MM!*+!(e(;zC?tt409-~JEl
zLnr=4u4w<Ig_NQR&3HY_uQvgwi%Wi>yPAXNb~90G@vJuGaTgz(UrJv11{!>My(F-E
zLN0eVTE&iS?qyQ+p7E(xl6yX|Va1DNCga=2Wtj~oX)Cex_CXARM2D{b_pHCS*3<5q
zR|fp5WeG;Rr5WG<rjw~#46;7ZzCCM^4icP0_U|V77SgjX@xN4`7(q4F6I0(8PgI;~
z7``elDtPJHWL7QwiLuk_UD?H>ie=HUwdAPKXC_WOT+#NMISPyygAOYA?pDxwcvWk>
zW&LC9?i5lNM-9zA+<KSv?lgN&gYiP4an5(s3T@3G!{p6ew5spm|GiYdZPwh0|4OOl
z)&pq27Jy!lF>_}SJiBt@$*tMmXw;*JyK(TmLMI07;rGBFFQ=*M$9ZU`*>=08w<J@;
zVZy+u&gXyPzvR4&fbZP!OmelaJR#*fk~m~GDP}5ho&$$Z1r`dY>McI{MegslS;Dw;
zt+G^<5FLid=Wl?qE%G%&XT|`>`NMxh{NAs&KO5TlYWCC}x!MeIz?ll^6-XK|%Wr)d
zBcyI{rz%=&eIdo+(5WA5mJ!%CF;|g}OO_~Z0NCqLvwhCbOPFPgPn-9eV(kii0@%@~
z<~Xa^$SHQ~3>fDVE@cB|f8(Psx;S!hE#Rdb60NYk4Ku)8oF&@I5NyMkwmGj{o?pb*
zfkjbzro+EAZG~=AtUqMk>8kLT-U9LFO3ScLk%jiuTxkRK>@3%|OWb#j(iwGei1*rM
zRfbjX*PInv_B$Q0Uo7f*pY=d9V)01d7p5*D9}=pH0_H+w8yX+=jO<RRHUtJ*iDNEL
zjT_8Sp5omU`ez_=&Rrh<Zr8$nRLhA-zcK2u+4H~>kEg8*3o2aWltOED04)He;D>D5
z*wEni8rADG%q;M_?WZ8^RI5twWI%C+`>F@1-RE^7@cX-8n4LSV30RZ)9Ice?FrN94
z>_tW0IV~PdJFnCX8O={zZz5|s7s8_=%vM=ii)jm=rXZ17HTmssLRr*(?%?1R&m=0>
z)w9pu3vX+eV@136IU70Ww7^S}*w@thO_eY1FkaaJqf>pZ=q3{XpsZGWUKI}V_nI8c
zOqFVX;s6jiWo^Vfnz=h~*S{^#;95{G6ua)zJf4FY$4u%h%fm=jtUbwe`p-r(8nu1i
z4mk}On8YYM6nR$bYd}ZrE^SB;DoCpnw1Bsp_G{f@mB;ip`)4vRT3XK}N*2L76f3(Z
z+3@7Uv4KS@AoF!E-6q4UvaFEsHe}i>8tm;K4wDV2xT5^r_taua1VXGWVXL<x!!j*=
zY#Y_+nZ!<+s8dBY!1Xecak4~8I6?9Uhc-_*RBaK~18R@oB7Ma6V5%(YZQ<2h6%9m*
z=Zk^8ym^Rf{iB?B4OJ3ruWHuc_tYKUh!2g@`bbX-IhUrtt=M_jXYzs6WNQ9YtTcR0
ztR`7Dw0Ssu7_hlR2F#&~1fr99m3IkMn^<mUWMK&nNTlD^8KW1zlV{LZ+1fFBH!*ul
zk3b>$Im+G2m*t{7z!APPX)|ij#WNEL?R`&mO?=GTu9LL<TWZMqS4~u=oj1OcZ~nTM
zpM`#gjfKxi;X!_Ws~1s&2VVnoF!SnKAV5*srJ}l$@%<cs*eYRmMW>Urveu){+`1FF
z@H3IGn_Npig6s+Z2r`au+Ocm*j_wccJlQ3xM@!dB17?BHXWNK|(#dm!k7KM6tE2iT
z=Cf)&vpsKbw`<hWO-(LzoFWIEngQxg>1c0$UaI88_Ndo6M)qFIa-f_gWEj{fsbmo7
zr_OYRu%ha%Y4Mj`lQG((Hck}XJpp-52<U!CC_JE1ih##f@d((6R5UP~93?i}w!=zx
zDNDO?m3e|Kb%o5Ya=EYO<O5cHD?feK7lcfh0;RN&b9z62xK`AcimnLkWk1tVZaexW
z(6-=moi@Ple*bMgGg1XzS_gRvPVF68bomUBBbeqg2QW{1Xhhv=;hyEj&TrbI(7paJ
zx(1UVO$7tpgeB4>Pc^L&nd9K82{Ze`lm&}UX>25g(y(_jcT5QNS5EiWLJSrj^qqoe
z-fwhNxRUfDSboh|(g*bmmxkWAIh>q#Z)l>@M`T9E*3*w)jtlH&{P#n~kIb4cb592%
zE8#b~S8oj5XK*K&Und3<4wk#_PU?L5zWeuNvxVQ0WcN;-93_N1xeYaC#6T@YTQC*X
z(Fw}>h7^Sm(A%`CoI(JH92Au$#?5N=cIR;WLHTcfzRU{eWUYd#nGRZ%Sm>Gjz#bIo
zJ85q5HhnSq@b<#8wyloCkn>2NvbOsSTJ^C?`V*DI^5pUJke9L3_AY(hDD{8|MmJ!o
zQ*KVYhTiZrn^b)5dQ+FzL~LS1vZ0KZzz|P^{iG|x&VqQOyugc{3iC+Vca20W`dKv|
ziI%Cwb3#t*BU(<c`NvUI6+<JcxB!pbv&j~ToOLt^2M*2&HDD?U(-VqmLXz&(WAglY
z<b$r+_1uxn-E8ZAM@&LY#__{aRx0ghbd>dl=b38AxYT=xN2!9!FZx`3l19y%{9U^@
zCCP8BPAa6cIVyO=<ox~-R12A~aWOt%cO-lNWA*1WkW<RWy}8!4Wz4W%cg_JA62?Lm
z`en-HC{cvcR4OfO-sy?<bHf<&F`+L%e2-;o662L~ii=fz1IuUDge=#dSX<06iW<WA
z#GGKeW6|6^>l1!%568F(G+qL1Yz4(X?2|@zNS;gl>_TA{QvSR$8u&+$_V}snFJV~U
zOn*q;Oq`~MTTZ!?=n0V9j@+D~<zofWsfNk$uEWF@FABPgHK-^M%1s&XOvz-Y<7RtC
z^=s98J57G!nc#T<Dmf}5+oJdC3+iuBA~?>A+W|9y$1;ja2%ZQ>nHVBzlL3P0V%kTu
zp#@n%-1oK<&g<9d8dfRIY?<@n&WaGD@5ZE^2t`=|M&8_!MCgw<(@sLZQNt@qqV=?6
z<E7wu+@*jHd6eJ5@!{9Z^|X7PC<7b|X23Y8eYPW2%efG&t88~Q2(;fwx^TR{i5GI0
zT#J>9mr1mI;w|UE5K(;@5q)}aXyUe(rJ%tF#yI*@_adsatt7r0%$A=_MD@{KPJr!q
z>p|qYN^RSpO#A@6GxBC5LeB(f=Zob#QQlX3FU{Z3RWz$&{EC^}21&Tg*!j49kJ$i^
zArKGg$-ZtW>k0Pd6FUp<U&0$@O151uBJW_Y6&1N-90J;Y%P45bDQR30-<M~Jsg78h
zC=bl)*!I4(e!lcDuZFaxvOh6MD(nZ!xYg<GsRKW{R?VpxNmeHrN}M{qWHzR*x1bPv
zx&Oi!1>WPbQ$ffjuKJhy9%E}&6k>Vl>j}Me4ho2~()4oIdEj)nIGg^e=4^pW{h_;2
z73;WwROUmaq&SRr_r%>H(Q1m_031tywZRm!bs5!IFx<9I=<wLp&MC~B{M4n`JU;Ed
z!}H$C$rOOYGm7N!Q1N=G{^VF)1s}B+u#ne&dhdN`uT4X95q??-8>I95ZAB6Lg<)l(
zvZ{&Re>8Ody6(gQBNJ=j8;2Bfbo29n+!LEZp)cC)q2dxpEOGshAN(?z_w`yc*^`hr
zGkrlpl7_uxOh`v(SU6MvBk*<QORNukpc-+85ES8TUlAOa3hYGwZD2G}wk;`D?mAGM
zCXuT@I_-PND5o}tM~rg~x+GmM|8nhnVV9L(THub}qMx6@vX}5E!ZuqSj##HTsjwPl
zEcMY{Dyzt+5cg8M`*KjyHZ=AOtW%}6K^E1!v)$w7(3MtSSU<~WUI-eW=5Z=z|CZ9<
zDVEmmwx&Flh0e~Jfgv80V)O;eB8!<{V(bW5+#c4m>=z8KFnCvQlk^^63AJ=#l4pt^
zOf=$2L?yv(8N)}xYO7v-!1|e;sjMqfR1W>69;i9be%d|FjL|chfSRE-cXc=|-?z?s
ziuc8yaX`rTTNV2rK)U^+n)=p*E9yGYh(}!`8eCRKF9DoZ4dFGf199(*EhdZ_Arjyj
z7NST19-4snl3#$(=KZVFFT~wocjb(~n5T3jy;HKrM5Er~Nrm*N58pUW{a;ZE87?)T
zD=(h9eHnQ==hfag!<a)ALCGUxr2$Hkc4Hy*!gDvxpEa)4rD+3oVsL55S|}7Ck(%-W
z`qN+a6L*ZKzz((Dz=!qYv^f?RN=i6t)#?4{Bo6tV3M;ib<_X}cPAPfZlHEC>7c(d5
zMFGS=%XUxJBtkITaRsINGFcguK0o*z`0X)jbrpV!ni(!EY#cTaO}9@O6`e`#+UF_?
z-dyFGTI8)Hdnu>zW1*6vOj|ttlGzv6g$+Z=c!_K(E*Uf)*L#lo-Cz=Igaw*{-~2vg
zjG8x|P0&i1!x|N9IY{R#GV^{4q7Posu}<=kEwMOt88y${OARYsc^0MVA<tC?Zpg3X
zz?NP;6-c@6l>M&qN5-PO25Ms8#WgnWK<>&{99=?`%i6=7iR-g>H8YzRKIXSJfN&6g
z>r!KcE@grg07o0I`<KLgK-VTChW>W<=h*V?X3)dg4<exPP=dq#8N;SlH~WDo39pJ;
zFnrfjY;1qB>Rb=v7#6&gLlvgZEudxvOI77(EJI-alUb3CUt-pnGEtx6+9h?BUxMym
zBx_4$7y-rV`lIFk3PD7SBpY!RIxQ*VlcoaN$6l?Gj=?m|PQ3^6HmgtQ8t9O}oqU-*
z2ZIuzb-230x@?+c9&1O|V#~J|g`Z51uAjyjtU*))a!M?>Rl7jp@t4vmNs1H4ARk;1
zATaxTgfOjds)X~`6U<2j)Jog|;ou|irQRhsu&^^=BArYIz@1&(x3kTQNM?v~wBOcy
zQ>>!3NZSWz5+I0Ry?vm_byW)}J-iyg)-3M}ieohdOoprrC=&y}-@h+JtsjKy+^v72
z&j}EA7T1E^QwQmyxP}STcPpjTFw-wSlYX6aY-X-1<Epo&tP`RY-7mzX`(!Xuu?yrS
zzVGh8F04(Jf_uBfA^#ERDUGwx!Ii^jLzCadf(s6EyCtSU>{(RlW79(~wj7xrHJT%t
zgc+>~`iT4kR$1wBjHe~Rr_Of^!`t@LXZDUSSOp+)BT#niivi6I*uHq(gQcWLBP_?n
zF*`d)6678HzCQR`3;vfuXarY@@1LjZ{tVshCTv|P<p6CSvSkh++?RIqBKw*JpS!{M
z^fg~QqH&>48rqlf)Y@&o%?CuJqv_UJx_9yESVR9+b3)w<Ln7K9W*u1+Yb&XOxcF<z
zSt43h>Tr|iozkqLw2SqayC0*ulliDE^rSB|i376oh}{MpE&s+_f2Gu~3F|EarY1m)
z>f(ujcJhCxcZW|C#UpC`A8@td!`IPIgZeeI0v;uerNTZ^>&Zez4{5?eRAVn`WCk34
z9%3v_`+MvFPJ^ux*5&D4rR3@=EXE{4yzuWv3of>&nke`eJ&VTiU;hYPd9lMaa4=-f
zekX}KJxSqueI97+)wqn_!p#BYSbB3qTW(~SP46drk)1`#5X4-c5kBpL^b0|Z1YB|6
z5B1~gwU2u1OgFbOG<m3K3L=zZN5MMgx3ABUt8$kEeXr$zY7cD+WC<zjcr^hNMHBIz
zJ%ubs-Z$@0xV7J`=`59-l<azz8_e(Pm*nOazb|1=^lo7IIT=PaEqxm4NR%^8e5J5Y
zM4180YFcWd1X4<RqiRBOQ%2(nGr{pW3GH<k1FVKK4vw;zZlDjp7sOyxSG;ehBm`!(
zM}^Ur{@h!*XN9#m?X~xb$CJ|b>Z$0Qv&dzE90Z=rdRsozFS<NYD#%G=H@Eu6IW8(F
zkZ(_k2C8sz99++rxz>ql3TN1Kw6((}vmoUB?_=7CTTBdkCF2y-<s?Rp|C$D8HZY<W
zHjR9v{KKT}nVja_IQVy{FfKdEv)D8>rZ&B0ZwERvwRrtHdlE{KNXHjFn@f`Qp6dA9
z{EYkAJYN=s_n{QOSlqfbH9(`t;u$Y$zU)$+J92G-s%k0Hf<sAkO(aFXBcLnjsLO64
z`i4z=-8k*~GrK>i970mP)$7zuim%Y=s>V?nCuFUelXkenZn8DQW8S@zha(#DuYy{e
z-{ydQmBS_%B@>EUWU-UcVUPyK6R|U*!K|c@>p27q4St(KfA4Ugkn!$3;N1W8T9c#i
zxq#nuKO26bq~yiU!g>Pm%nZ!I5G+IT`l){}EbFb|2PU2usWy2kK*}-p4-dLTKDpJi
zoZWI@Z_;>eo@7<t>9!Y6b)kD*2gW(pWb?9lscAS7^?y0uLRfYUAiHY>yrKnu?itmd
zy$Mp2(LH2!&Q);v$@9|9@AlED?{_^biBONj?JrVF)pmg$bNu{hN$7zacQI`Km^$fn
z|MMJDYTcqIJ=df{bOID*hzw@t^U^dX<MN*-_I-W!`~@tWEgEV~e@S2VMWIf{p!clP
zuCZ<@Ay01QV|82B{UKdL2%OJudt;6o0<B9;bM-c%75{SljHB7Tww&p!)Ib2Wun@xx
z)A$cD+LQa^b;FAw_4(vUys#|MxNPr<fk)pCdI270VB`RAz;tm@L(!W^Fiv+Eup#Ox
zeP#VwHBN7GU3QH8sl>;?FDqTO_j@)q1>l7kP)cf^u`6%xCf3g2(`1dVyCX!t=a!AN
z(_yEwJziH|1hhboMgC$m#~f<Bj8NEDD|W8Y?(^W!kXMPdGDx@gE6Z9v0azU7MER-y
zn)5t3Sx8i~L{tNobqRdrdmehdd@dD=1&aF0_{iP^NnC2gvEN<Eo=GYiE#%HsVNd$T
zg~AerBJitrT3<aSkCqH@Tuxp0HmWhhkW7AG>q+r95AxKz_VXRK{HRGheVT-%=8G9n
zDZ-<61i-}f42XZy^i;sVhtKUQ`z0u9>xyJ9BR(Gz+zt-i0pPc~4W0iHNHv5X0gXv6
z%bv6u2|RBJ$w~n8A>o>(0f}DtIOs+N>+;QVSqu5Cgm07eu~$j!QYr;u5c0r*)v4nW
zO8t<8IEHF2phmd66tC!!ecMcowiq&}nVDC~+Z)%kvj9qKko?J`B`1%Zyfu~>QqtI>
z(V1e`Z$gCQF#yjT45W*}M`IPGI+(13it&<CYnMJ7<(>5Nfo}2AL3fbmYJK{9w~%im
zVgLp5ph|dro{o+!&Chsq+vm|hr?str1jTuIwwS}7X#(FnYZ}8uGuA%JLyVzP4i!Dm
zqFcj%_;>;J=FGh)v?tGGp}b`lX86dN`(3LHBYiA?gNd``<&@Ux53}eP8a7#Ai`u;7
zHz6QZyD(q%`jf1sO+{t>KL429!bUkS1l&p`<zaq1b?B~*AJJ|~c}mfy%9}>$wB1)^
z)3c7*VvECfEGDYwI%K3J#8j;(VpT0gabjH+O(gdKZ|mc0KF-z0Mk`(vD#Z1Y&|g{g
z9K_lfyglEO-&1sH)5}X=RRL}3>0&FkgU9}Q>ar84;Us(!pUq=+5^a1{@YX<K1R7A$
z#_k49s)*qRDi=Q-=BBn8g2?p0{*_PMplg<_KA=)4R#EL?%-dOv_}Rgt24~@m`M828
zd7uDb|2Xz=>Jw}4{<CUP8%g*AGbGYZe{7YqF}Et>4!KfX4<A{XTR&28wi9jTLvHU0
ztAwKXLn`<UpOm^Fr;X&K3qlg*e>>i;+<URLFZ$v$8FO0@sb4rIH+*+IwU{3#YKM>}
z84KR0DJbx~NVDjx;$K6VRu)w2$>_Os3ncP069=v)p_9VOY=+e8YAi0YRJJ*OKX0xV
zlLJot24~$RZ{!D<GHXdoVs>JqSXGUkpH{)Pw3_ZqnKBv|D_<Rqt;O9lz-(ku#q*gi
zbP3b_BglQxZ;<RI_4K7d>JtPxnHY3i+AC2<>*3_~ZZh&z<h4&`!{@f~5K|YZp@52<
zfSrow>vIQI#!q^BhXD31_3PGt*x+A~%&-*}DiE2Dv1-MRu^%)pm|N@~avB}W#dA5E
zTU_v{B0Sfq-VTV-P2hPYqTVw;@1Y|Qq8i$^u2(U~r{A2z?0>EK2EUc6J<h~5n)B^5
zAoQzs$E~z@KM<0tcx^XH^i)c&t4xwMMcEIU&6yPj*6EH+JS{EAtN3~LPBoJ0;`<!N
zVAXgMc(@(l%wTL;={J8O$0ITh#2^cSwNu48+qsiCiA`7N!iAi2_=2&Z<Z5oZAvlND
zZeG(0#0mp_c$w5^beN9JbVr>eOl`S^j=qcbO>dxC)Ku5*%h1vC**r|zGILw;u@}aQ
zI~$sB?|?`g$D_jIw|C=~m3imcgR6Igtye)cPzxwj%IQ(vq&?**(1ATFi7StV_)%*y
zYytiecoY&+VyKH956`rgA7we1V>FTcAj{AC%0<=V`>R)!TWnj&|JhDbz5%yQSKs;X
zYP0@Z9A+y0WMP~dO)U3U;wF0G<vt!2TKKxyL#XVlp)5(dY~BwS^b$gJt7*c7Uk(li
zU^|pT=)nXFR6X+rX@j5T*R$QXR^z6B-bEKWCcWEp<<fBF+u67vGi*0%YD!yn=;GgS
z(`0EY4ILdf;4JNpP*xy#_b5l)3%?4GG>HD(o$XltgB~8ci`|?-)*Eo2SafpJo07IW
z5gk5-_|gbH*@fGPxU0Na{cU^m@D(=5J+0^iRY_pYgvQHnXb>7EBhExgnVt-Sh>DgJ
zI%)1H>C%v0g1+&%_Mu2Va`wHuPhM@TdvU<LtV2F_Z##E_<dBnHr77{HhR6DyyZ{^Q
z$cUlogP3DTj_bz@0`B}(&RbV(1__>uvPPqvO%hdyH3g}?8fo7H*kACx3c@6YK$G#L
zBytTVa?&?(w!yTABjdA8T?6EoVaYPDZ&-*lUVb0dt<62;VZTAN%m3WHKoq{k<gNYA
zw>V=u)<;49(!?$O;risCpNr>3m_%h-WS%!N(7<vB`0!DOM|=2O#=gX95-Z2)Z}Hqv
zC>HZG<Bu1)lgWM|A@$+o2gN@pg`I@|Jf2WRHV;?mFtsr{YexkXb`i(|54MZ-P_>Dx
zkasTMo*=YRV$!#}SowR<*tRxWAPB6%G_|3@+qmEqnYLPl&zzd;>RLME91SU0xVAyg
z>J=tz_W=kPkanZQ1--WC5@GH5c4Fs<QiW3Mca#cj+vz0soqBmOZ=|$Y9C>2U)YUzM
z3Nb^(wSv_?ywoHQW3U9e;d$M09v@Z+Vs}|lIK2|L9_MO#MAf7Jb<U%uOH>f<07y{O
zP*!b!NbC7q_7&asJUA|B2$%b1O^(&WXduUeZtOUgvyQ)|1uVIFGR_srTRB}q?_cN#
z^zQBZ7|HLP5W!EIFdNEU61Nh+AXEhzFhzXEJ#RBNuf$Yzf3=MLvnnF06_!*8EF`8p
z-m{(r{}g-Os=Y-x?aS|EE0+{Tc<=HF&9c$iiwkq5a_1jTHP&eHERzy8WQAxYR=kkT
zRZ>*x1NRBrHM_>CjAc1GO4F{QtWtC!yj2U_nC(e4Tz-*d>oNou>Z>2vImE`!s6|?L
zZI76HH}N7Vl>%2qY|fL<>+ydBzYUg1#d)pv(O)r5VBV*cg<;MzJkdC<c+^XhS2Waf
z>b$5(xito=h9ilG=wYMXb7wXchhi|el4@lsMN$6pH`l!Jmbd>`nx7S%g%oyUVI0G{
zb=jQ)xopZ@!V#nsQnuqyf2&}YjX@)Z@LZw9Z^iV5eKNg|e?AwexLBn=6sIjp)Q*nL
zj>+-Y&{H4la`&Mx{psuTwtg~Zl9nG?&bl}WH7NL2`zBCVpy4y;Rj>rar^`3&@<p`0
zoA-5s;w3AmPeG$P#42!L=rdQw`1DC7N73=IH>epd9p&=LxTHS$bQv`Es7faOeZvI8
zPKqbC)l+}rX-Ol(hj1%P2{pv<*v2MLx95z22>l~9!pV~wXXxM^mDcBn1M(+m@pgcz
zJ(C<vF6@b4UN80HSsA##{eJ{MUH%c^5`~5B%l7CILaSe9UZ`m)s+ZTJlSGp?{6h)V
z9&80Ymu&x`Cj5`!R_#~NKLV4Xe+0E6-VT(-%mt#}Fa%xA*YZZ=d<2laWP5&8foVhf
zI!E<7yaYaxFV`|=JFdSHhi=PCoFY>S;&{=r`E&Pd=f#;9vZM)p(}sl>C1{x}1Np+o
zDg(%*&Gpt-M{VCY{v#0aE&FOkg-$80bkMq;QL~ZQYYRRC)p_k~bKg%K%#2{c7^kd<
zW`r?~75XiZGLrjTIFk^4$!x2`xj9;KyhR-&yKs!I=-uDr*j=X+?5l&|$(%_IlZJ_v
zy4jAk-9VsYA;x?<aJEZK4tx&zbHH*D$I<HOa+oR)8#y9k&Z}GY!Ks89GzUl)-Ee7`
z)+~*NZjyEY_B`)wi@jjh6zpya)=yuAPF%$|<DHl5j^!nSBo%Ejn6QnQ;Ng!)r^olr
z?qfpVE4}^XR`8->=a^bwe5NrmwUKFRAv;oc_A7!k<*U|wpL_eNlyZ_okMKT^j)XcW
zy%0g9ZpVE#nRY?eo;tEN{zUcJo)}C=2Z1<Y!fFFJbR_C*dF^-`*eqqf_pUXZEKLRM
zUE=KA`n(?1^C>k0G&BIDtF7ud%2fg1w?LPk!}qCAk?n8*)(DyI@a6N<q(;fal=yF+
z#38buW|M`{evaY4U26^8{Om7%09lLs9d+?uN84PQBn%R+(ISrr=*pPev3gwC5jYzP
zLawQtCu>k&1D5ubiwKgWDdgfWEI@?GF84asq2b17nZrb%t_RW9F1>8$xc+|x7x+qS
z^NaNW|EERWG|fCbXv`_i6hvRBw20R4UFgFx-!0WfvFP;kK{elslImstqV~B*a07^|
zF@)?2d5>{Bw?<-}P^+IWt;y05!~cOoXBaA8<!_5FHCh1<Ts83i^^1x{J&mt!w}-iD
z72r*R>KX5sdi3MmZ@26c;f<ybFU5o71J=!hX4Q%MKWN~$1@Hd)5@9jn7$D1y+hju6
z>mVfWP6Ub3J*IVubiy7E9hZUrAnRv0wl>Gt=$RKcPZBe<8Pb&tY%4gd{mGaT;-tRb
z`k4MTy%WoWS~kmwaaCUkd)2KEFdHZdjmu2(^CcjSXVrbk#V?q_)qJ$JY?agKW@X^u
zV2>*PqaCIhd8S6+(3Z0usH#y#6E+leRxz8~%T~-IpwITPdX~olq<<aSxAVhBa@2!a
z`Y5i{^ln>%W3G9a-;B;x4!eh?_$kn+FeN;VLl8R*L4S}7uER&S@Q)2;43PbCDW`_C
z5Dnvaj`fITAOV=V&;I3aLDQ@r!+7`0$t`qrauZv=!?dAFt?p1`L5VL5@wY+Ci+!8Y
zAo|?je2-z{jrBh=FuHLq==dPNgxgB<lQ$aNJNr0JB!j-|d2$nJL)lyL8NiPnSv4A{
z8m=`E{JO!^&V~NcA(^7oFR`KkZvv5{4I@yze8RLElc|<@K5y`ctifzkKW-eOoruLP
zTY&x%sK??paG6&<#fa`qZjQR-GG!YD$K#`<e)CMDaI1IDBetcDO$J-gy25S$0bUIz
z`NI!HIjJryBBS*swiVX?<dMNEuM{G0Ip$y!nV{i9`;^T@y=uq~_zwTb^~o`fu>qc=
zIXs=PF83VM^6`#e-ppurlQEL98Jx?7u(#gTV6%VM{*0%Ot!>o$RGMOrykoO$N=4ab
zCAXFBs|@KHyP8B)vf(vo`IwrU-l_3K#fpjTMvY!K|CkW80AM(JPDDR+uydlJo{LA!
zIcs|#%miIFkX-}JZu4<Ggfbrq1y?TUE{a8*2lMp>+M<+nT(Rn-q=)@yV$ZGuHDe6N
zC&o-?+(SYNoyb9xkCo!Wb=s2R>BSQk)_1?Yw&VZVG@fX}Py0wMxh}euFixQu$1IGN
zLynM7)~Gj_5W}7H(I+Ljbe?<93aJ+<G_?N_q!63;+?tsHs%9bFD)5bobXvJZ{*}Yh
z^Q6Rk2a$`DW*<!+(eMj@_V7<TIo~Cb>t-)N>x#J)ml~4OCc7q6El;;PI>tDgNZC28
z(Q_!<b)nGBBwBg%A9XIZv+FHJhZpV~MEReRmLCcJtZ~o|{Cq@F+%TyR9r2?`Mpqy9
znk%s9+M0bLXC$huxy$qzZW72Tpb!1&IiCErP~hw?qdCn(UD|jBdB2I^FU)aGUSVb>
z?gD#uDU6~EmGVr0qI_c*)U$_xrz-y}$x}QK!to8y)YCt|tlHZ0GchP5M3s6LFxZ`D
z70PIAql9LHvwv5z4eYpYzdQKZ7LqWNO*E}?kALz|z4*@R*P*|l`^`c<?)-SIZ>>=>
z*wnPxIYxjt7|0=aH@tc8H3Q5`Y-z#R8L6lbglSrIe$3O&QM^^=?03cvIKxd~UAx|W
zcQqUf_?<)Ait3?*;v+I3xN7XVci?^!src?@amlJv5vp(t{V)N7-X6-TSPJz7-xOpQ
zuM51NYeXc!JE^JVs|TBG^Bpm|If$bErW&8s)p+~eNbE2s@;<B5MEQilHj|AtH4<@I
zmWw*z5ATV`(b6$Kf8^9g-^_N2pe$BW$TR10Z|i0R#pK(uBV*}dlhELTFIaU2Ij_^W
zDsJ2ylOYBCSIlt;_IO{kdbI<bV;yj?;8xP)?%sh+PDXzFRS1epf3Or1#4alTL-76Z
zZ&g5unGk>+4*n`SR#A-1D)E_hV)sofbU756H(IhG{qFena+K&e>1X$L`+~r{$3KF)
zyqA1*6m4wr?rT|BMI!y;8x=nt#CHlUwiNs~scGo(8`20VodKJsgiDpwo#=dcOK5tp
z4@I3m-0Q0N9`Wfs=J+TZ=gSjQw0sHU^lleu`dj+_W^rlTDbb(fTO+~Y2I8hD!=m6t
zpRH2?6)guD8Az?_&alE~!Su%WDl3(;U%H*DMzPra-$H_bR{;fm^O{k^t7DGClkKjE
z0$CRF+%oqx{}C2w2pNk%NM0e?;(E7cMW~!}I&p-FM`?TnIF|Fbw)S#nk?lG6!8<5*
z)gCcrPb|zBWz>PHG%yF>$SsxfDI2QAj~vXMy^E(xZuMFSlKS7;d+(qo+qQ3zDuMz^
zRUi}rfgqqrkt$Un^b|@0DpI8P&;z0Z(h&$ql@@yFy(&%VEg^ISq&MkJbn`s#?mqX<
z?C$K&?6))D%$q+lnQM|G*L9rzJbxv%KXNxfc`~t|76$`c!fKgrHkC1nW})9mue8ma
z=$1ZaTcr|RD6JYnluG;4st-=r6{lkn^i|^}luEhpLv9ny{RYa#T_Xp_E?*9~sgDI_
zR*Kl^D!}{LN()TJW`JCt$Fff!lO;fm<Jt!UAq1<W3xDEKv9_yCQ)-HtruIwWuz^!5
zzo}7$pIa-uW|s7QV!juTOD9qnT>L_#MiC-4F3FudYWoef>twlAB%BM06&~UV;@EE7
z1U0-Hu=ULmLx@lbP;611r|_50f?s$mwEW}>wscs7uyTA8Zk|WU)$T!Nur;v$f>-FI
z*z2;&6r~x*io@xt8|1y1m(aL^PEERxljZN-f<%8pY`Y)-F^0{qyRoBkdA>Q=C!6Z)
zni|=?$bk>H#faR>^CEmc_<~<UC7jpyQaOGvFE1}(KEh%z)&ewI_!a4Cl7B0CvMWL~
zGLd&hG{g>@gu62%VmWivZ&%}MR%Zjk4Crqy=w}g}TURd5Kg{{Q>xa(<2}?k|Wsd|4
zN8;;>CRn>B!Z)#Ya*8NxskJbV$4{$gDt^}`eN(r#n1H!8jNw_$8+fp{66>?SK$WJR
z{^kaq&Fn!<M8t3m&)s7(oEc{e>*9lwo-NStHtNm0+1yyh)LG$m$4RHI8%}9nUcX4-
zYwn(oWUGeFIWQ#&Vz-4(>0c&tyQr>{B*v&M<D~*PX3HK3MJ6em94}#Ny_Jy&rKg*&
zpNFpbr%d#?a5VK67E7?ZPZ?l8-%DhEcdVH%pI08Z)_)UThOu^$aeN)E(KzPgeMDyI
zyNC=-AStOqe0ARmf{QRc$lVih&Jc`^)3|*f@hFY`PLG^p6@>G?(-C%@|3zd7H{*H6
z=OGIJ{hKIGxtMu2wvxJ^68hrV3q?ZUz>cSl{MOMTs*^gAzL}!~kJII-3k<+_N?5lt
zo6d9|k5*(~Z1K@of%xLRS6`}od($>ADSoN*CQjQek8eUrSJ_R$T9?=3Gk^cIh%Pu^
z80tSE?hZSB6YP#i8Bx}*1<Jd8{1bDgKf~A5lrJdsyrdW@0;!nJKz0KtczEpvwn@H3
z`l?_IJZ_}2Jc~VNXZ0$c(cDJ2yfwc1dKBLhb<XO9y42kgFST}pT6+tKieM~3x>Db*
zxN01)-fM}IsV>}JIMm_#0~zZLRryWmmdQvr6Ew-LUM{MKi+*n~Y-w8I|JI?n4o7CU
zD=484N*GNQ-tJkVx+U;EDVs1XZ!y_db}pNr`r2`a(uqEo(BuK5cl?!bHrh0Fac=O@
zIJd2ku5cqYz%MY+U9GM)&3FmF**G?}<IC_aCl1n7urXG~HCJR6eYLagPmymvx2FE5
z-}IFSpj(im)B-wWHtGf+d0*&Sq~)TH7?qu+zTX*;O|%y@PIJI)$`ZYbgn8BSVjuks
z_;3>5{_vBYXb*EZymHzbdPtOVACAb*S*in}%S8$k%7xO1+-r8+zd=Zh&%-V-sOnc_
z|Cf9z2SN~3ajNpa_J(K#B&x^z&6IzCK+q1<9O^{8+<Q-ySaFFAc`=)cmLJR27a|PE
z0v|*B1Vaq-_I)&8+)_ycj*f!D*$K(c)ztg<HCO!Kd)`}7zY<I=C92&%t{k#`**|T&
zS=fCsM*niAEK70nY4Mu5KQozr<DmqiD&T(ig<#|FO>5_{7L?y}CVs!g#nV40%`xLa
zVnd-qsBY}DdHeQs=AOvzeIEtIH}de&rM;O~&(5V!KKF8+<SvXzKW<sr$~dULJZs-)
zBpaF!WF3@#_;j?339ddqaW;4LE0d0nEAzCy5AkM-d>VO=<>|hm;a?KIZ%4iH7m62w
zqhu*>*K`hPO+|Po5+U`k8$~B;8h-dnlFFTa5F#NjZzk*{GhqZZL0~ea5fWR{?T_@G
zd%ypZd|Eij*xS7;_9fr)t|}qDp1a2dC-I?iic4LWZDfAznD^bOb3vrl10<++P1Iem
zJ;CKHt~;|p7S{I-rVpmt=b-_|>#tIcfqpT_t#V0}eF}tG_dIAwJrE9=3wvQ#W0apB
zL%Z$Y%Y8&~3djrR9&*01zm`w-D8Y%<Mb-32;K$`(ikoK*4GMeBtq#KmVox%yGoF|0
zjxPAxU%d^{G=$lgMJwJ?n&hC?6nLH2Gv{0BkNsZerNc_=aUYgq(HsKWGSnFZ5RNtZ
zI3+!KcNm|B6nyIVxpp{cQk3)STgZh#&%V4K&uNPi6<z&I*)i}b?Qc+zxN!#8is(;t
zu}|ho;F{0fnILJWQ)jd19zZ$&#Tju<X0LA(x3<J26jpIrbZr_h60w!TI;ojoh75gH
zkbTJjWFnO90+9ft6cL7-x+nz`hQawAsz#qb+&3k{5X0sT$SX`iXYEN2u+<c}43B-4
z(GpGdq%gJ=G50;Qh?!QrW4v=I3Bh4$S@3)&`2Hbl2K1A9GH1)oj(ywc^RuMsY5z_^
zYmqRT;|}lLc;q<z@$1x<h1``o1?OhS%;1JMvm2ypX8sn0hr`~pB*dIV=-`i#N#CPL
zRBCg=WWSzY-!=Q^dHc-e3sQD9u<?S{F8wv$%UzkcT=M>)bkEp&{jljEWq1aX?|vRI
zPe{MOvTRygYjR_Z;u5_o!J=s1nrvnZM}^u>m+1<>zg57*VDLs$!6%W#sLf+7Ddj>Y
zItutpklCf!CVLv5$}G1e6lM^(VH5_wP8>Ad2DS06{Q?j6*`u9AnE6MiWt#}Zb0DYt
zDWP#WOqg5OaL99emxG1-HIIi%hk9Q(&TQ=v@>5$OqTxrf)8-`nxE;PXetYH@&QvKH
za)(i}kw+nGiNj-~$>Vy-Xg}K!5w-Wx{1nHyhQ$;wJ5*Cy8qdn|<jNf@tfWI@tpQWT
zvyUBd^yUfeiP!+$za%yTL4ktpJ<uc#ixZ_2BaeVE{%;!_W<C2<?ymvx_(NN)x-^l3
z7K5WZ#RY-~<rSi?>n#kW51j`!%m~TIH>F{tFwC0Lp*|bhZ`LVUet-GHxH6%A0_?kt
z&WtZxafFrCPP|>n5e?IoZ5cC?<*L!_L@&&^NCoY!2X&eCd1Mwa!Xz{2;_Cw&nxPk0
zfYhO-&f3ipDk7E!qi_+&twpl!h@={VY&T{&^{1rV2W~NK<kpOi;UqSd_nFK%Pg%`v
zeKqTu;1%=EmQpjKkK$h?m24q(VFqTK3|o>kPHlV*(<+3HEM(i<pQ(Do5b)a5TfWw`
zQx^%gHZ#wT)!D_0vY~Ca*l!4NCq@@;p^nMC`ADmj*ye|wG%L$otj>k;l263M=G#q+
zG7$HcuD&)Ig7W~v9iU*ofZ+pOJ$OZnbjrDg=%=Eh!it=bG&DJkUZ@)!vLWm!c)WPp
zo)Bo`0Iv*ueGv1p--ch?rf<J)!EMt|dRF4N6a!ikq*<V8%FMVp_a#DtOKQyzdUajP
zYhb3u88mbT_*-UHA-8cc?*Wcn57qj7w}ugf^bZFyo@|*Kvxg3}kB?yM#@WAsRP7pI
zqg&WZkay*|qc;Q59Ko@4xfok*88C6T#EELEwNV+?v!h9}gW#8^IGSj_1IQg!c<&#m
zp|Qff>kNH<z@eQ4MBt4OwL?3a$+$SUS4wx4&jn{6MP^v#=ewc8&Ogi4Km2@>OLV|4
z*&UqT^&G-uaYj-<6ffR}pMIs-_WhF1nQyU2>moj$$<Q!TaEm^)+*Y9SJiPbMzz3Pq
zDDkaS@U(MXRIYGKP9oD3R<BgN)8BAEokL4gz;8;S!O!c{A6;YU)HyotmClTdm&3#Q
z+84U^O>F@4dwp+VUsTX}!l?QcG6$xCvp;zc7h*dX&sxZ1!`kUtQ2_<qn^Lmj*KN+O
zj7;|_)va}>+vb#ZpBN+7=R~j*?+Tnkg6J6<lzGeQ<X5JL-@HmMuUQZlwBKdvcfh^J
zCp#D|nNZWr6fRAHj{IX^27B1Vr)I1r4a$kgpD+7rI`Zq?lB~TYrPngB1_H&r5!hxd
z@s6~SaHH9drR$J+)2;qkxHzwOu0)VulYJl7vx|FI*0$%|p$j!RE=aq>^4J&+F0RxY
zdt?$Ioiw}$oun~t&_HEoWZ69A=p%~HciyE00Mv!)Ww(^;KL9Cf6B#r&X0>kyD{?lP
zhtImE>1jipk={lQ#VlbVkRp=0i?<42${4ZqqHP|Z(^Bp<9ZJR0HjZ22eU3$Hj3*2x
zO%O2?4YhVc9`8W3B=WHyN7P?ay-bA>4{D~6W)9$6R3FOicpJ}$F&Lm0=!=kXbB-)m
z@{lq|3%E3Z$Un3=TYYWcy$jvVS=RQ>K`|Ze<uOZb*?R==ohSq~-R}_#H#kfa7!Yf8
zsI8Tlm?=q$wC0u6wq^>qrQ^@ziw%fRs9tMNy!mc+z2sZv@A=i8w;wNb;3pzF<74Gt
zX7~emG;-c>e;gr^J=bz_9G|rtU8=yV3NO0ycTpCgzK(7*PRzdQxM;K<YF3A-*3(2d
zglLMjc-1En4NGwWI3FqfI+yX)WccqjrtceNcwNgAoGBKaxPep-Gb^#M2Pb8$|3XXS
zJ}}ZTrxA@&w+=vFg|u>B+NnE5-rdW`G@8?BEFAzqi%d(3C`Hf4T<Rn1#v!ID5EX(!
zh*RJD#6Q1E9k)Ou0E*Z~ftR_xji#QkDYSxqyt#03@hz4-=DgPIJ+kYUnUKf?FMD~H
zl-a(jnOn<HCKm%tsL^Kn4@!*Sv7tcYL%!|&KNG^yW_!msWiMGat#J)Ael!M}QzV&b
zX)F#AE=KShC5{c_u$n<A=EiPZO$dqRMq8Eq+Wc*c_LqeI<BsTW&Qkq|dGyG|2s@z_
z&;6#XZRGV+V53EZ$%7M)m^o);T3THXv_2YALebFYbJZ<aP&S;bFL+npTN>NqAtuuO
zq99)#@G;f+E|Pn{yju_7U5%xWODez(=1g4H91Ug{+nD)2sEK=@$%MAGwRL&X&%+h~
z412OR{7|rL&xjoX#R>a;j0h1=MgQXCjmisQDwvm@ug9LGaGLUa(fqNsxCw<S*jEY1
zgh<jQGxY&1tg;vhszmm50@=4dMF&LVpmiCd?)^wB=m6mtavV06kCO`7&Lg9y0W>iU
z?1@$zW=;Z{^52bwZrKaAh_QApywA!MS@%A1$a{aTK*U=om6=;_Obdz$d42WdKDr@>
zg$~_5FdA>SNOFc7O^eLNeFuu2vZXmyn2QP(7Ui3&V&86RG-y1(N7S44`;wYoV6ld9
z>w&vXPz!?NXwvv><Mg3NCwjU77jvrg%^sM(V~?Ap^`sGX`ZC&+h%^*~A!Sv=g-Ol|
zA$dhWs$b`^LcD7eH1gx=cNB8AmW$t34VyHDviUj@31uB1P1*UTfGi!ZbIb3ac@|b=
zYb}Wyogq{16AoqCEi+X(xJjWq)_yUUKIu0d1F}O|@wxlp=Y^uGK*#5m+YvM9Ik&AF
z5BQs%B2oyq4tE8j8qOWxWy$h!UYrBo|0UU0oK%o(SA4g|qh~;5MGcopMFP@9>KE}$
zoR~VX#26xlx2BXVDRK1*&$oxzYq$FHv!>x#(LWyN;f{}gs@W7HtnROTJW{Qdofmuk
zGGYN0`-f*Tp>p;v``BSld`TMJk9!k&Rp_!5R(&p8cZ^6F(})IrU><6Bxn^jYAn(t)
zGA&x?5Sh%*Ov_uVgCi2)Ku+@7i^9gwGa!V*iI#%rjDb~%b9KAEj{#MUNsyl}hNKH;
zu#Pz4F7FUq!3EY?>PdnHd7WyxbLFbWr@e!ZR9yf_z5L@le&Z%9;w3IMz=A@@WBoh)
zz!QKYz-!}CN(`&>iqY%$3vSbqIlx`xo}FZVs8Pcy*6E8+Vnq&WXZvWQ)pMO&w|u<P
zAn|OwX5C4->+A!~b%X8fwl05bi8X1iUyYE^`|%l`PA1)hH#Px~mH>J7`3C(dU?Z&j
zH@`=Q4{9|h_q@@%Ca1NJ<vBzJd&)#2EW{m0x^r-z68W{##&z$*Omt2RQA3i#GT1X3
z*J2Kn1Gx!w&?taZ8fx3Ve3|`zzLLHKC1M!cz)znI>Z%CAi6)rB(RdW@vQXjpM7xI-
zZDYI@c{IgIaw4fWqpn$K-rW?pboS=Qq7sGY)IuNh4}Z{7o6KJlL8AHK<(R?uZQK-d
z{>tl;g@Q7>X=I$H5Y6NyZzYw+L2%X8V@aXwg%$<zXv+5uZ?c%=09=oD(o^LoV*d1A
zU!MI|+I`ODquW(vdhKET19f(OJ*Z1?2>s)Vyw9JhQ#ty3?0>xo-DH>3e-fDhV{$-^
zEdnKlJ;JRD|5OrI$WVv81K#Hvfn(&WMwGgJrQpws-?nz`zGG`8c8%}~CiRrP!Uo#m
zNWS&A`Y)-pnuj<tyyd4m!(op<%DwcPEy1ApLqeJ%Ml=lu(+v_?lLW9ryA^c<AUY=N
z{Oh`uAHCz7ip(3V;@i=ScP5-{E6m7C%{QED#PZP;+)4AGbb$~An*u6xs&<e~R^8(P
zZtWyJVmZw3+PG4Q57HC2cf8NE$;M-*@%>WroMK%W@bz{sc6}3ku1D07U+p8jn|*<;
zkuktf$Qdl>X9RGweDRo9TBoNz*&FizxR~7BgNpenVakXz>VE|VVv7*LlKuh+l3$t2
z4GLLiOZTYQkGApR`Sy36R+A+1XO+OPlXoM-6(qzLr;7irCqzClf!(bK0iWjAEz+dO
z`j`XOw=B2%`60L`)eC&Fpe@3`lE<${K~hUP$v7Llo&NFGlTZ1t(3DKu5`8ogeF!B@
zpu$jC%E@p6-05Uu9Gl%UK?D7lClgfz_t8C67kwfRZ8p&l<~<)tVjgkLNrqBWQd6V&
zU*G1l{}jD&ocw=GG+G)#=V4S|eu<X(8ai7&AmW7pZ+wIn_j!LUG*x+mcyW7^bgMkE
z!-IhKR<tq|{7a$)9Uq(908K5UI^N6a{2=YYDJj}%0pz1TDKNgk9>=N?wootCWj>3#
zu-e}Qt-UJAD(J0znV@Iifz~^j0dY8~AxRRW0lJ`@a&rBTg*lbef!!{-6(TNR1NcNe
z^stX;88cD##r77(C)9res!Vy^9$;xKD_x+yvm9=YVZlp#`gZ*#;bQmzqWW&TH#27P
zYfUEKMd*(YWJ!0r)1Gg77OC%cw7A}uy4T3-cfTF6#Ti~caq;C%=@Wh8Ibs`V^~pTj
zG4tyB&KI`Rc~L&W%-g$fsB|5XTWfr!{H9e!!1l~RleD6J#*J=_vk)bl+|u&ld~K1(
zk#1o9#J`_4n0{F&sx!lQ7DEl*nLOsM1j|)+tSHO75KSL)+W^0R=1Q~IjZ4HQ&0MEQ
z$a`~b*kcTJ@vl1}S21vxXWQ~xUFA=PlqDickMIw0{LZ&FARkQhX^=Tm*=;#mZU4I6
zZ#fRfZ50xHdaf;(<lpydKi#SzOqbVQtM&BOFq+COdg`ksblIzD`BNdoCg@+Ow;Ywc
zSTMP@Q=XX2{zAAajbY2qL0ktdNk0JtWTT<7vLv2FJ=Uf}0r2R{j*%GC5evfVl+XQ%
zNE94y2&VqA$S<Ij!I^cmBli41|Ic&NBgr(H-Go_kDSyf!3iR`$#Zxh-por!Pd>wN7
z&B>(#U`<FhX$^eczcr-b1fed)T8q@YFa?p`(ArdzU8hfM{YMFeIn81U8`iM%2YvqL
zJ$RSOarn9r$JH4fztFx>o{%i(FQZuVZO#On%f8T!$By~}olX;J;2kvGy&vP6OdEyy
zZnDN|1@i`5S;&Q!q<>VBXR|*$oPGar<1<H^pwOJQ%)-=U8pbbgwTN{Xr~Q5VzUse|
z^X~$@_WsvN3jKdFb~~T--Q1&H%h#p=_y$a!2rlvrcj_?HbOO>Wv$5?;;)Y`^7N;l!
zwPIKA<m~h}Pk|5b$N#|aq$ZiO87%2T3z4Jey>kDFPXUagQf%`OV;vL+mhkGpTUD5y
zbKI;Yz9X#2(1K3gMH=&p>!iGh%()8rgSkQN9`C59*2v=GXF}?f4kemJDqym+2#tm;
zFpa+X@oefppOT~PYkAW4+sjWYNG9R$VnV8<sSKD$&pNKfzeght*QuWhWp@t)xmOsS
z7Nt2C)6yc5XwsjU8d*nwDlM^pltcIqM=HhzP2d0?y23n8jRD1_WhzgC`i!#5Z&RP}
zYw4>lXv|0!POmu?&kmv!E$}PvuOV8XiU-g<(kx6B6>MuQ_deHuQh3YdsS`msno4Y)
zz2RQ!>aJ3!U8$-+i0fDE9vLcK)$c@i#Uf?i<Kv`#b<OfoQZuDynJP;+w$mJZExx+@
z^p@P#4ufIK;)Gf7j0Zth)u7oL>|jUjru8e6T-p5kLg2SoR#*S(5rP>P_I3J(2V@62
z>D7~3rQuR`!i8me*!XyYn#c`J1xak|pJ#3K{j*K@fy*-Mcry`nP#(6h0Bm~}pTt3<
zMVk#|{{e!O-Sg5LYU+LWk23RWk@{KYR{tqvJTOW(Y9=B;b?FQ<PsCH1$sR7OJoe3(
zn6M6?t6a;3TyGc`ie=`tuJQ|U7*-=MvinD`KxyVW<Y;9g+yL{RY^A&>*veCeL3YcT
zBnoJ1Q0t}AjH#dr`A4$7kEyEhmY+}&qi{G3M#(6#=L1?&hJ#5*xg7(+<LjwcE~Wo2
zEi%rpkLX^y@4e&7x`Q8P*R`KlR~?oeMCd+QV@T*A1tm%$38lTOXA0-rW$%dQe71Ib
ziHUtVfvGLR;{-@T8>Nr}Sb39vYCGQhU%g4?zaw8iyyKsOuJ3A3fGm6Ut2!0>*|VCV
zPSbhk`0YvfxsNyIjfd`LSlm^(Xr^Rgo3SREf12!wjN<ETi#5lRGH>tYo$|IT$=Bb1
z`kxALn^joG+K(tGiK5d+Xt|pIsU`GpGhTKzyhStf#d>CBW*~PyaX5Ue_c!Jj^FcZ3
zGA7TwYm;-qjKNjXe*B{1qU(7VIFy#K@Q>SEyYtOmhZjAO2FOl>Ss#SFm#V=l)<Cf8
z)_R;01%a>4W&1*(MjqOKP5Rl$Yt$l1sG_W>H#P76krW`|8pCkAx;>r9=lI_TCL{L}
zGWQmv4GYu)iJ9$eJ|vv|S6>1));S96{%2u8?uOANr_zw5p<Px!Q*T)=YSiF|iVuRn
zR^UD7J5->%_CJc5!_ovcG&;!9DOhMOl-2qeG$(GbC`vZ%S;n`ex`G=-&{2BT*~oXb
zC^o6;#`ENjQak~Tf%`^$dRpg;0pann|9Hy4&OX1MbGCG{-4($Gp0EJvPo`kDpe|8~
z3M~1^uXBGJ;9xIe{P)MQuP*g8YUahl99Uh5u&5M=D8C(Woth%=-T08+KL!kL$7Sl(
zVetyW{$<IhRt%3IkjbT?+H6U41RUWx6V#nW%||+--TIuzEhxn&Uk5#Y!LDAH6Q9lh
zx=YxE#5EcTklWU}@sF0lM-V!oc==%RMjf#pM5RC1+K;6QUOr1*pr0osHJ=ak=!RIc
z46No#f3<h&D=xQSIu*9=1*u9}P|dmt>qATFGJ>YWVgJaUY8UGOTUv;&p4t3Ubrr;7
zK~BAha|3JT9=6+ar!srzQ-5d6j7{;1wqthZE|XY8`?D+U-I{#3+?cr|8H>^hXO;dx
zqESN+{7#*J#y6s@8^;<EMkkw_v;6rTENkr9Bt+SD5`Z+JZY94xVN%yf{wkx?erPN`
z4+aD#Jex!cyn8PeiCCJCxvEn8N5t!5Nq3!G9{d_`ZoGW8Ys8L`R2`xTVIk=C;X6CZ
z0BF}Z$?TbyT92nU8*N}dxfA$teyPYqY1vq!2u@A(wt&)(xxeS7RpdXKF>t`OH0iT!
z{_w+kk)c^yuJwy86fs@J`95`Me!I(JEy75^X>3Qu&57jH+LR%f@oP_`=_{UBAt1bX
zLqf;2N?1!=;<2HEmI7b@Kf+3^U8DCLeo&Vx4B9Z*0Zc`#(W-`!)J5;Xsa}`cp-gGu
z=$~3o`=wj|)e6#KMyc?=*F{9#+5Ybk2mRaZpwJJKTa#b5@LRX?dIS031=&UWPNn1H
z^Ddr)pXb<;d`5;@+CS{`ZS5qz2(MEx1_Dim!2yU6Hc*&bJ+hJam3(dYLy)muWxuM_
zk-SO<m<c1&A?GHdR=D-L@i5Jo^R89DQ|Odmo}{NWP97mD#CbtC<`qw`O2*gnGQDT0
zNbr$aRq>a31<<7)o?s}{32|4|nAH#UlhUipZM(w6Dtr}_W<Smq?c6F7bU8^%*120k
zbA(Ua-0eH_i{|`xwa7R8k1l9P_AiNqaCxar+}JWpvz(T3#31_+RPmw)i2+c&@kF&s
z%80YYSdYwq_5kGclo=(nE7=wIg$XTK`;FD*y=95_JHO2*_K$$=z+c)i09{{fE6lLf
ztBn9$gwD~>*i9O;{djf}C*^lOpa}8(zU8|TWmzYeFn}m3gt@jnjF*>`>{0P)@>U?>
zl6m;=Zv4r(U9Rd@q4ss{EltoIHK{#$1#sd!=XbV`(*GXT{%huJtG4iuTA0lk^q)p}
z?LBeh{XciccKiSP->uLeYwV;kcJUC;w4S}mBqGTFw{^w;9G^h#5)MG8*@oUN!A(Fx
z(W@zgWqpYpbY0dIk5WvRQjJ>lzT&It{dqGBW?$W@uBYBP)@1p);Q`J*^7dfD$_Qh%
z_><Eo`wx;X@G(jGb(wV-pgUT!MsNH+mbXZRIgiqF>qM!!EAK^PXQYYfu~9O!XIXI$
zp@e121FwadoTL+?<`FPQ&f5TnN_EY)S-N2|o9-;>e)FFx{g8r6@Nn5v)EFJQ@b1|s
zzu79ha{P>2{Q}(CSSrIWrDFO67(r5s_F`K|e_PsUB$R#_e+7vLF9=^TI{`$Z(S~>*
zs6DwoJpN;TuSXf7?6V*sI-I~}R=_xT-Ww+-l;1YXR*xGUowB<E7>9sbot%XP*;FD^
zPm||(r})#wo6rNFAA`i<mnj8hjtW}PDOIEK#l7GvKc<hk%sIs|l`RHEeT9nTl8^fO
z#;!#L<|=mTUnBi@4U<7{RRUgSGUB#gZaObZCibmzTI-~zE<ZnhnhTle3+*E=`6avL
zCB5V=iwG`HSTz&LhjloNLW4A6e&T;g@~?!}Gmbb#kH2m{@sT^4KW>-#)GLFIdSxsz
z<HE|N3oe#4=22ur_e`2RBT54$LbJ%8^&Z@MurGAhF6mZUyLn+x<p9Ul*DklK8$VlS
zCT;IwH=tDQ&%fB35pKb*g$O(pg`M<80Gs^s%!=(cXarwq+gnu7K$mb&3>MjmDsi40
z6BHOm_?kyfHDXg#nl0Z>a3jvl_cW0qbWAQ@r@+K_H2kNxpPcoLWIf=qkOYz*@xm@5
zht%+gn-<9t#g|sKyq_FPUHdb&>P4Foo6}87v~UreD4yn_-_F>yriMN*VMUnZs@G18
z<4x;}N0+HD@_i-xUlpUrq!d(NuE*pK+&+0nW*#2=`a$jZ3z3G-@=|l)x~`?}nLUqg
zuCiK5g`7fxcvU}G9QP{Z_)n7g?FRmGS~JD#FQ`~qYTLv_e?7j<n;H}JvUDV;vUqy4
zXZS+_qn6TPNdbt(0Mo?q*-c(c7C3QSDYM6M{ME77qI-_B#G;422yq0BX~WBDv#ZR3
z!78LwehPsGV*A3zLzb3zb`)}omp$!L5goN!hB<JPAInQX0GW^9_JFg(ki0g3yW}fO
z5w@VLP$eCsPNbi7V1+id)$KS?vPAKR{fDMQ1`~H9%*xAZw%R_TMXo!co4Z==xZg+r
z+CwO++JN^t115LRZujnq7>%;Y8og*>npvQgGBc*%#c8y~1&ry1nJ`4@IL+|YPPuJH
zh1L}b)<MQoF>P^B`G*<Xq_}M_^0pPfmBCris96Nf^k(Q5Aj^IU{}AJ**EAEcsA3RD
z4etFzibL|*Dm8YMESdFws|NO!pvdviQ=nQ%tMUubiloXdf3e_~44JJxU8CP+zJ*)X
z%a%B)>pDSj+-xfbhH~Ht?tpnPgNiJ2bd%dW#{nVgX`F^q=$QlKVSs<>Af*~Jm}#+7
z+%@l)`5bM5-<}aJwzqeV<?T{$dz;c%{5H17V3JQ@6-h>5U|zNHc5I(^y>VRp<4U2`
z)LfR!?PuAiY(Y9Ih95>B$Tj>o#>`3sRy+4jH~sk~%`GJ*1D-`Ap?Veg*YuD@5j#>8
z5qEbG>=O>_YfOn*lD>vT#^CX@b1pr)lx^s~g3h3*><SR%P&VyEp1}3pZ8>gc*HLux
zP1V*e!{qC*y)d#G1IQ{}Ex9y9h~1hEmw&V2OkrjtR&vjW2^~?_mI_B7uZMwr^eqWB
zUk@S<B4jgdCgy}&WI+#7u^_MP&M*CTD%6|)e8}JjT|LY77xFoSx<2(0!CKHvDd;88
zPFo|s0ZKZ$D<vCKg|Ti&V112(V1|jayVvWk0fFmWR%b?C*FZ<3<nv3KQ-liN<KUoh
z0POITBF`yo5am!4Q!`!$t5iCSf>1Fq#chsd@W2iD1qu&waVCZKWf?@yzG`VxT#X2o
z9awXc`}{4pyIT0U@<$`Xn}6{BjZlIZS1VLR6e?>5w}WO%(3X_F0o6Wir)`{24%E}t
zPQ<m<_eag!Y40*KfpJ+yLRt!OWY5BqZwJe$ZTa@O(`158tDy7Fc3!6`#62}23zHBG
zT*IGD$o12h+mWMpg@05A=P^@WOCoL=js{mVVk7$8sR9{oUG86c=;Ei7Mv9Mcg^J49
zABulYPyJBw;Q9mi%(qRM6*g1LD`5=lc9&xl(?{MNd0WN$uedcI@!Ys-YkqY07TYu5
zd!TJz8HFnDs&;4qru(#s(S2tJ#8yv<-`O$0q+FvEy?LVL_PvxKhd^(qrN*xiVIo+h
zD+%;nj%@5Mk~)x&{@qe-de`?u!me+UrDmF!225RNkP`1$Goz};C!kUHC*GpFvg_4E
zwzKmZvN2CYdw&<ne;V_<VL0yKmiWHBo1GmEJSIL)>yrN|$LcCO8b7^`W%b12=kA)T
z-n^E?!mexpG$Aye>NV813$orx6#gvttHd0oD~!gw)IQsLXNu5P+rlNrxR{B`6Aiv%
z(zeEjtPf%=EkB_p$?^CY`R3%`y%G0~!8-;df=MHj{7PhWm&jn7a9w>Vx6i8?--J;8
zA^dHVKkQ3k2J+)r!C&h>_L_#58#`@>V)3g+0U8|7)x8Q4Qa^pH_4qLlE;p(5o333J
zDh+l;I!iNG@6#&X81^#^Z#ziUGNgz|JYED0wIM#sSFUB;_w{R>evw^TfND-V!hzJu
z7ilc0r=gp;CkWT*9f9ZOvR}~mATo`{`s$&w$3|LiJ=4q*8^=eq=%#pcO==Qf!3?hY
zsWFeA>BHTXH`>=7274Wi{lws*znM#$VxHclaDCTRRTfNMbfQ$JR=n6U@a=JGYM(IT
zSj2dgq+q?oE?Mu=P67R;f?De&JihN>ON!ZET6^wAet{@|9x?56ZZDA6Rg{5-2P8@Z
zYNfBtI;&SDnI&o-%-I#&#U*ptEV}ri&G_P&o{W*QQ`L0h5Bsm@Vv)(B%%D8$J)Psh
zCTU}p@n!<eXX+nRDCt&u#~Z(Ot{G>TJnW<Ko2hl7u}@;h*u588vrl@~flwg@4^Q3~
zpBf~ta*bWGob$1fAt<dpb!HfER9hGs%~6dYSsa*}MuGlguzNT`coh+0{d=R?V8ye3
zCZud!r(1!NO>i^Pwoq7+w^)VbaO8%<^_@mxE3>8U{br1LRqjzO<cOQCTGc5qo2i-O
z`Nc>)l`iL1midyAaAQK0!7}vjqelKM-LI>k@(V3!T%Xi4cYMr}l6^~e=}<~+2$#>e
z2X7F(C;Ei9;8`y*wPN9G)QeQ+VXC;J;FNI2vH5nj!*<D{V)r6XPcIs2+_)Ab&rVx_
zs9XTtn50dq?9KG79hjz&HdVV487Vpz5;1HMYLO6A)hl#G+eSz4`EQUJ{Kv}R|J5fn
zCy{V%-T3-6zvSYv$hgslV=mEi*s1;+qoIb;Sc)&kMO!8GgY3x%0IFg9v2nONZA7TS
zAqbnhq%vNU@sQq7G+>G3bvo9B^H*CD_xT(6CjY*P)RSU={ztXqrWIIGgkh~cT8l$S
zXp5ja(^!k_!Ls)CBb|D}n<64D=X0XF8E%L=n5k+^?F8Rl*l)>Au9dlO@$Ll{$#PLJ
z*JxHsP;=r08W!Ae;oG5+e~L6GC9`yvdF<=`P(~yn0%=(3{_4jMr*2fYXojrb?~9X@
z{#qAa8E@AHs4;F?M`W`=d#|9!Q|j>jc9Cv~YrKAq&aQUwvJ9#?bKJjr95xCO0$O01
zd?;wi-p12-%6i_TBo}Bj#05NClz4!vS<w}#`cW_qs)OH9T3kAD<jc^Ffd$#iPwf3A
zp(di{MpxQ;`*l<Sg;_<lCPj8!^=EAvI5)R~ZjvWj0`@=>+LG6KeF(Q^-OOe8MuSnQ
zZu8NuW|K{=nLdyYSZ_Fvh?7F~6VGo$${dhS5Br0z*bK`b%-V}^Tsq?bFmN$^48Cw%
z<4xv6$%}NbfBkY&32bP^GbhI>btWgS>07+s0+_KQ-U~V6xglS-oRp@H>}v}RuLC;A
zgq3h;@wHTB-qIVcF3`Nwa;>iGgxb<;XotHt;y!pdH2X^&0~P)EcOx;E=if6g<~$mu
z6-S-u*lOXnsut}n`Z;psYuH}K-`5V-pFh<uV9irEjsi-AG$gesd3P^Ty1JnJwbfZ<
z?CPI88yCaxPdAv$^O&3og#RU>l^0wv_XNl)g}AlcyG%Xs3k{C3no8@j-}zwYSa0Fz
zbA6FROL)IYh!0Ts#*2FIab)#)<#75l^g#q3v9IwCM#s(AlYA@omiH)kN&S*|KjeFx
znUR@UgF=Ksa>k2@B?b%zQ{q3JN^@NSl>>*xz+nNeE^HB3{Q5e@QpMN4s`7;}k+wR#
z^i_%}dl$Ck{2ek<!WtkL`)K)m=(&m@jpGXC^9%ROm@$3Bc6zXqH(SlqxW^x5RHn-Y
zAjKb>u$g={dVR=N2b8~#c0gqpykE}@pOU|4Z|DXSbhBGB)u%R8GNzX0bNs+8_5nv8
z0-r0^f@P=e+(<T17CH(Z_K$th*#_t(lBP9;KAFoMZ90u8$i`QI+cKUY7YeHT-0@=&
z!dSDGQ74RPkGpPJFHcBVi;er;NVr~w`wP<Uy_@qQPqvuwr40GTOAQsh+;0ugf*1vA
z00qrjr{A<CMOi@!*wDYA+2~KQc9>pRf3cj$aPbqr&ph|G(?aGo58d<U!HzCJ(~S-m
zj_o<+c<eQjNsW%nGHzy8=@~BGbsF^6?hZ-b;?Ptrm)eR-j+dkSCFu5-1lx!%?{MqS
zeDr&wxCUJO?MrAk4dlxt2+9;khQ!{wNkiGK<?#8;FDWZ-RT}!<aVuQePL@jn`T2UI
zA>p?{sGU8+w@=<w2S))o)ANffV?10mkUb9%=sQP6uHueLtyn@G5;VQ_axgqjUQ{7=
zJgR=|>|2GE2dB(E#T#1>PCy`FggtUhVp-`e8~6;DKq$lG5~>e_0l(!($7}oi6R;K2
z9EIf#MJp8R4XP2EODF3j_v;0*M@0VBjrOJjr)&A+&$p(zCQ8q;(EfUFxL%TJ8_7Q(
zjGo5Z*NjMOsns?tr6}_;KZwu8?G@ekb?SDk3-Gk`WI6d%!;h&D0_)S0opxmwpnPIK
zr(eGf-7#9?cxzLXr#M|AbQZVd6FjaUML)^Emrl(!>}(?T{8xKoi0>OyIj&yOWy+p$
zU8{VSd{rrCOuzjTbg?yQ7KBi)@4UJ~;Ha73BnlJI+~4G?d%9`IlA~-r=<vbzl0Kc*
zaf;u>{9K5laK>JVyit(>qm52V)k%muVOKEGPQ*{Bjmyei7AtFA1Xc1OvqibH?P!Rb
zS45Rk?fM^5qVQ3(T^}RCWKp`>vFOyC$so`Inf7}OxmD(y>$fRmc1)9Boz)AL;Lx+S
zd!h!<Nfp)$nK-66yJIPNXC}-z11G<As|uAt1W^Mf_adX-1iSzssIQ#q#QHrR^#?F4
zP1t?(xN)z>aN4;(v943%K?4BxSdfIFo=Rn@57Nf@?!B(V+5>!|M^4U`e^s@4baPBj
zY!^~%!yq;0AqiDDki5`fqP615V{58E+!fb$UPrV>$ga#jp@)K*>^Oq%wapwR{!yPi
zWKVj8e6^5{*nv6C*^hf<6;<8C{A@GwDqu3#GQLG{n<+?4t<-zIbf!{lRsp}8>sRtz
zZUy8<Ku!ud@T&#XONc8Pyqq#0HAb{4=p4I$p;Mfm=$j?qPZV?u^bbDnOMlctel7c3
zO3RH-XdhCuSlfbBSx?A}hX+JLrnv_+D{(7!w)YUtT(2AV_xjO~^?6Ulk@M>>3G!<j
zhPRTSyX9_J0~dTCHSKR*_Ayl!XmxGMBJmf7`h@hMZ;;5N2Rjp7-yKVhp6b!S1B7&&
z-UKv^sMH?JXF+6s;YGfhFw*f5-Q@ahN0zN9`OR4?G(JzhUXX8{Aw0lkM%;A*dMShe
zCzILq;7d@JX6v%dUcEWusg$u{4NYu_fpmd%D;lYww+h~mo_42l+MN?jD>knylB@}_
zilzK{g^%W34s_+Y`Fti}!Xx2RDWYmxj<y%3^Mt*mHovGZ;yL|X5e+1qhrBMd=X+{;
z_sW$Qpr4~EC+CmL_s01RhrN68i4aU($M`ZG1r|QJXn0PW^4PkirpqJQB#65TGJ$*e
z1RiAXgrk26zi!lqz|DQDm<?jC5goRvg@6rYF+x4yOzsVXv53ApsOGxXENwH=x+-Vx
z*)jQ1WRBbQqdq&WU$_ka7PgI<x&l*9U6_J1Q4<Q%=-TOY-mOa5^5@=fh@tRtF4?aM
zu>-jbD`mdRTYmjBS*%n#$%SRsJ~gv4Ox^}$=9-Q9!^d3`BCrkMkC};>aw1z5#12%6
zmY3z%DnNl8CS!?N-m{(rTlTt(K19e3=0q`xE8!hXkGbv64P|zU+bcy>0)NC0Pp@+I
z|B@)1rR>i-<0IgyHB)w!fE()Sbquvaf>TrEA2MhY#%q53v6)m*-_1R;R1`U<cK(Si
zL26wWBsCAB=FQL2kSdn-3krE((@$F$^2)(Y3#3G8UjPJuI6*ULous(_Jcww7d|+nK
zss=lLzRVj~dQ_t7^P^{6WwM6mh5Ck<PXH$;C*381UhmKb?~q%CuyaaUWz^Tu95b=V
zi&wSP90N-evTn)F4Hh3Swsa4P;OrmZRwrboir)N@#{J4SYO|mL=M$teUMHMCW@VFR
z8U&2f7gtKXk@eZ;^AXrNjaU3)wc_`sGCktdxy{`1=-6BM)!QN<Yj;;K%@W(_=vdS7
z>_uONnTxdZ$Cik(?rsTZpHeEVtLfHUHGhihL8(6DMs1oeKOKg<)kwP4jLR(+W+QmD
zxnr`n6WSf?jNRFp?IW_kA`;@0CZ6q~1uC9m#ZG6P#}gorbbWH1D^vYKH&a!5Iq0%>
zfa@BD;&f7B2#A9wn2AR7<Q;hl&`I?7-I+9#X&7`&b7!2;+YDH_Mjd6oQFfB%5oHXC
zhr1b5&5|T0enx6@9|<(1zz${@JZ|f8mhj6|@a@xZ^1G6M`?lbJIp;s1z{PX!;!^sm
zDw(CgJ}0BAX{>W79-X(%S#t7$Nf;lo9CVi4071A5hXmu6MMPjsx^=Z!MX84CZhQ`f
z3iTtpWtub7)vwR3%Y~2^+nb+aFNTW7YK>-a%1*lEqtsN&TKbvoLT-0jQ(8U_pqXeR
zJ8L>VBK_sn<9yyq{%eY$58OwHcqyyL;S8jjYj4=wXBIM3X4A}%kF3WOWH0#K3KEsE
z3%8ki`lv*ljTr}hoQSWPPB7V$)lkLq*ijWJSS`8~R4uUp?WLaEgwALtR*VMb+UX6&
z$Jb`WYY`yTji!z6kCUY%(XhT|)_lLJO1$D_n{ZacQVcE^Lt3SC_FYCg)~o`y=v8dv
zWc1B`y4ZvolIPlh77hzHXUfV19FwD>Zgji+Cd#s2)&{wL_g*}huxljMxysC_b4Di~
zJM0_8u`JIz^HXXJ!!^0AQd;;}e&74-VJ3U7dou`7`>HPV!3q5SdZEx+%YAk}47;J>
zuGX$rF!_NCYwrQ4;ppg$#}<savFCX~X_k9OSh|HZhz^5iRR+rOp-E@8T~~OICZd#m
zdaPzxf4k2&`N`rdl2eFWJa!s~4;uoiAT8PRqa^KNei>ApIcF2mGC~8mNzd5t<X|h0
zDgiN(#ZJo(5)rCG6{Vj>($FCxz~EqAP(BBOGAYc8uP$$ccZ!T}*u!Dfym!}XQtU`V
zvvoP5;1y=eukT++J|Q8wk%ioRI|B6wI)6#JL?%Kw!fAm;nX-E-saE@B&v|-qKB%^J
z0e9W%AD;%_b{oE|o3h>5LGxH#yo}*%?~*7J_kwC}ROx&*JdjR?O4LG2mvDi!aB}Am
ztb;tMvV{=PP>4fIYYaK5)FJnm<in-L?~mSp;#2l`>1kaa8RTMab~OA-Ae7VPy?1K{
zl@GocWDLTvVI(w*rk1q%&C~(M*yxi_i3i_qKSTW9RZq!#-q)fE4P(jVcm1``MCqE4
zi~UGev@d*>1pP}=zaF$9m`Y&t$~?=d#ZJW4N=C@m^(9n=_`Q`<2vdvxJrI9z(;c5U
z19NN{l}0?nWOt19k`&R<Xsv6!BFz+Is}Z~R>{&6(;9L_#Eft%|-D`YX14XRIO0k@p
z8eJ<gngq^!IXO=wVdl~ge)n|12Mc$irG4CoJ?kfOn&yn&V*=>2ICn~Ndcr4Dx-FB?
zRzlwg8Ev*x|0~)zR#qyMLu9#KQ4-Pi+C!&XO4ZT1u?cxrx-X&$u_wwF@x7&ABa@KH
zbi}N3w0^(sT+%kaTs~{-ER~gq`5BkTCsN%;_2k0{V3jfXQ=oG4!K(#j)d6h<g}23%
zcAqi2A+>C-kOBkDS?2;(q&+tSYU9zXfcBn;ZU%0bsuFrIhi~5Dda}FCZX#BNTRWc|
zVtE_Ak&UhH#odK?gNy!HS0P&`JYyB~eIdWv`8G0GC$H{t)Zu4}db5&IQ*Hf-vWO|8
z-FVI-P&^X^7b7WM_U7bdu<iBJ*DvH`tom*MTXORA%ER{Z)2pzQ0TFddPq7aIaNp*d
zrm!c|j&$#U^)*$-pyJVIE7CMdDk(-yoXgqz%D^Ym4@6-y==>|=2*~0@B#-Zjpn`2_
zF}B!^v}$XJ_OeKGgi`Ixz$^d1Bv0h`+JCb;{Uu?B%sAU6u6ajgl<=cs0W~HL$x1uw
z*5bwX0K`kThP?f3L5*sE1x2t*e$S4Cci+qwlWx67nNwXckW^kTo6D86Y2!qPb|&$O
z<iSPrPA354_hTu(-fRgDKG%P>L`wa_W$79Pxy};g*SM8W%k-or5QN**pdnLub@f9&
zfpBh>vd`?EtE)(_Vh=yox0axn8fI;&;n%PypL~))SmpZ=MSs@?P}DPSgSn!I7O))O
zrKrP*n1N&k3R(#TE+RVW^$B+<F?rHj(cs*-*~NKSZTJL-ZT!?SIHD$AFZ7sDbCdbq
zUy|E*T~1@dej{~*LYXNJu!%xSLaa2nBEDm-ee_5#f0P$q7ggLY`D|Jg>6nwT8K=)M
zqdX1v+seO6xZ$*719<)2zKZPzyTmE5ov{Z=Cl+Eo+~)J7LD@Kkj50oSQNtr&7e)(V
z_T1#-pWwKm%)P$;RzpQnW>xr3x8tN!cC&hDa1+fn90rrr%gqc(sM6XK{~u#E{y)UQ
z{C}dLPZUtnI|-abGnw8`SFQB&avJzw==A-6T#o+t)p2UmVa9ia@xZA^TN~|;pJql5
zJ8Mu6|NE=|oq_+(z<+1p|2_kcH=6RF8`E%rVyh7QL@`c{bN+){zeg`m_x2Be=3f38
zJz+TiOTtlkTcPjc?f-)|pzx#HlBQ|@XAYYGe_r!L3lS%4+%a8hT2XD^_Ifq;w)VOn
rI0+Wz2_Ey2y~e3c1B2QF{|KTkT|N_f*!K?9)VAOLlg%FVcj|uuTqIOb

literal 0
HcmV?d00001

diff --git a/assets/images/impact/talks/clickhouse-2022-release-22.3.jpg b/assets/images/impact/talks/clickhouse-2022-release-22.3.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..604fbd9c2b4e91089b1f3cdff014277fd6542c43
GIT binary patch
literal 53560
zcmeFZWmFtdv?kiP2A80XO9LT5u*RM4BuL`~2~Hph?iL6hprO$Ox8M!|f(3U84#8>M
zz46DLS#xLJ&-d0{Gw;W%TI;M+eRS8Z+I9B#?bDC5k1GHYs4_$ufQAMDpgo;{$A16?
z05&Ej7A7V(78VvZHZ~3(5gr~cE*{ylXZS>vWK>j?WE2$Cw9NF>&w(@)6bzgUKo(YZ
zc6KUyZeA`n9%eRnw*Tk^?WrjqE*>c!9x2;%isx+qkB`Sc08(6ZDvUoEXe<D9QZx)w
zw8t(0!_y?u(J}sGzW?LE!N$VGz(q&Hdnz_10ia=`qhX+9JV}60h>wedh7NejlVX$M
zFbj~&X|hn5;ZiyY#szB0vr@sRh0L?7be$91$3<(zYS~=kC)`{UUWo)Ds`Wbl>L>)W
zbF6CT<V`;36t=K*KR$UHfbMDJr_uk{@t=mn!o<OSD&!}9nj9Jc|LGSDtp7eeG;~r7
zO>!(sEyP~|W->XmK+L#yCma?**y{1)Jb)16sR1blDL@8r@LK4lkPycIpZveM2a@iu
z5vwoHI0%z07@#_wr7YbE$~l=C334-GZG=^`5#$OiY%z)<%(hG+7)TP8LbILk>%?4t
zqQW+Mg`;jA8V~$Z4>WIE0;YDBuC6|WzWts;^F8o?Bzu;Eu1><kX(0oXE0~;%RUiws
z(X9D!Vw!u2waj%9@`q*Wn<1g7V~cbrc~|ilB=L`ctCN7zvoAp+MD6dbt+B2Ddt$T3
zos-7Z%|b;6g+K%U0*S}(EB{mDll@jFv$o{Y(0fBoZeu*eP=7SLA*XC%U<=0jq1ms|
zcMJ)+$rYDrZu2utyWCei1K-K3c0~}%OM@8yA-tL61P6VRDey`7GOj8s(!zpG^zh@W
zwIuO1#VS5Q<=PPT4-)t|eN1->#BYZF3=I*E6t?ahqpllPc8AdVA^=*K>=y#JnH3X%
zZ!-M!4d?!+_!AAH?w*-WpO^D{9bQ|1^n_GW8#|pJD6uNZIt-}iHvCT+{)?X99n-bn
zAxm;gDJPRGSEVZ9(Gd8WJZ5!?DJm;j+@I15eqK;Vu`|$GqHjPc|2C5yV>jv{CK2-y
zu=e}eB;n|~2^Un0O?%I^l(ed%icICeBqP~E*GCie6`9_aE)Vu1%bVV1Lx&Mo-rTLj
z!gkE&J{$LyI57$&jQ(St>zzilW9}8^edIik&%RyW8l`NC4{L0;Kv<4vbExnWDMH&G
zxI}96{bTY<{kP_hx@XucJrC)=norHdBK0)l$^3L^wE;kPBuwV?byNUp8ngl($<R=u
zCl3<aE_Qna=y$$^CEgANKU^{l6RI-4p6WvuuL=CB)?<I81L|trh`nLQ!%}?sb5q~@
zcd(5z4bJ}g(C!c;czxq?y-3qgJ8wTYGbnjt1FHLH7T{$z-!%~OkTYK98?#(6^H~X{
zt<I3sT>q(2#Dy-cFrHU)@IA$>#@WPVEI#D+Qrcqe02n5HYIyjWY)Moyp|{V8jbkM4
z?@<;4GhGG?9arOg2_C(T+1UpE6nJvs2R)(|Xz2_fZ@5d>|Fonn%<@Lt%;3Rar>mA3
zd^?%fG=s4}>UijqX6h(@#rGzer%cOVhXnSQ1V$lM=}s2y=FJ9F*V^sy@%Nj(`3;_!
zI$chl8M*NgLTohl$ka9y4HO~E&k1H?%M%qTz`tSH8%#`;TyXFhka{Gu{fb(b=lW2k
zk2;lmuHR?fWNS3nBo*c6aAYPUv^Tms+Ta`6#xvsHl)NNbzIxOB8pc&*_r`7q28Jg-
zqbQO%#qJ$64mHDj;18JfHocLrM``c0q}+))GCKT|1Tqb2H#c2b#KDRs?BpS2WvsH^
zo~j-)-AA{?Zyy1wx{m<-4Av@!?Sn@E^??kq%UWCs<gus1m4rvunnx};DVn!8_Vsqw
zdIF_#;7vinW%-NfZt2~3jfAe?EKfWD&g(x{1@#N5e5Ryf8x-ZH)ljpbR~1E;=xV>x
zz0udyd&(MjPs&A#&jv|#<^lyu3j&7W@kD)9V9yMuoO13ESOnWxIfT>l5g?L06VMaV
zMES|F!n|wj0%FVa&p_?2^VDo^&2>>>lffeM7{t)Co{NWQ^v`45<-FpCvc`W{3hk4Y
zZ1}yksujFm{MLDyd*h{EAU?vtvh_;>+cSi^N){n!G(7bq@wc6$)ki=Cq~E9ag-?{>
zV1VLF!IGc8!QJ-NxinC%bp;zM@0sy*hkl~Zn~#7ne~FQ6zE-&a;XZr)S`tdi*APX>
zj;iVyH~9EDG+kknu^jfJT7S@Wy@pR`HpGd;6Fx%<Q{7cx60N!cYD+B7#}Le$yyAq%
zo_-T=%Ik#}HuW<@8A~+qmgPNe`!UYn1ca5-rLG#6y!E|7-umsZH5cY-@&xk8%WIT9
z0(c;EKy<fbv$9#n0V^Mc>-lPf?P{5na_R6ad|Y*Py=m{Z1n*WLKWyXvdSZ`%C1dfM
zcB5C7<b><6HnEeH;R*lnio{f(kSBjv&kJ7$uZpENl<|hKadKn>d1c&~7-b)W{8#kd
z7oBrtW(>7%)H1WOvMc3?D|(Mhy@G9c78q^P`fWbOG)C)Wzm%`o&En?#bAFP2g`;tx
zv!pPWamP42KRBGwu_Pschqt4y?T>V3XUj<R+|ghpJ|vWQe^;!{YWy1Nu~Sn(<<BBf
zFJtHF3tK1q`n|vJ{ZHMDye_~zl65^x7GpE}7;DT^ebx8$PFwO@<M+Fbh2e9&VORZ~
zTQR*@iX1%9Od&1C#7bwi5A8qi)<O>l_~}y5@j?<usNI80=w|LMul4vNVD)(ugB)Z$
z?398mdtKoWXWHaWu*~^VJ?1srw>Wz${DeA6@<#w}{F*+Lzic_W1VsNjdK`H7F%o&d
z69vEbn4_MS>9Opu@(Eh^dy`~i0CW=Iyyq2e!=)MY^@w%$U)=Z-I`mp(J*wnS9Ht5v
zP%7r@4cRU9CU;4Zr#2uwMtmfD((~@S%6wuO=O0q_+Z%kH8>6kIjL$L8{t=FRritYR
z5u1Ns-h0hCju1Yb)7ySc^xb0}VYt=OiKGw31ijJWQSEN$FNP(AU~WGGTs2wRd=8gC
zyr2*JD3%|;+x+&^!}gy?fWq!lz?VmWvJBqr?KBH_6Jk&<KDp2f7p5ED`=AVe5b%q=
zlbNTw74BvS_UExY0{UR_520iPr|i+t9q72uDbC=YsQ6}DYRlqO<}V`!0S;Cyg;L83
zSyiJJOgy5V2vtp7JWO=o_=-2JF2~=#7KJADQjTsFo#I;X1P%QPB<sWeqo;}`Aw%Np
zu+~`_!?KZnR@ZYo7<RDxGFF{PaRz&}5>|NMA$Q~L>Q<53^FSw&8h>a*Hn4I{KN!y!
z!P5N87i*AgJS&mswi1@5-rIihFVL&NR-l3C?tQ~6@u=o0e%f~T%{Yz@sctUDM6?Rp
zlhrV9IZWxocaH$~8<|0s+dtx&b@%XY_j*njX84HpDNj8hP_+$2mGC92ESk;h_~v54
z=%>XqRBfXJe2j*$O<Pkd(6qQn55fh}yW-y6m2os%=%~$m7Ay1m=H`lj)o_a|!r+C?
z0`I{-a=l#Xk4hC;TX8~3zv?;d&OO@9EoEOLzk|<k+<~+`zla_qW<+@rM0$Qa5Tu=s
zfKddwr|)^+ug<G%IA_!D8EUnCpNME9m0HWJOLMOC1#5Enfbv!JilpQz3XEYCcj~c#
z^SRsFodt{IN-rf$>!$DhLf;7&yCN(@c>o+$(%#SnqDkokv~B5+63=m2%`&yG+11QY
z+QaLop3TGiPo}y_v__aGaxTK8MTV^~k1Vm^-Cs=KTe^T$bFSkEued~*d7R1_W)bVI
zkspIB+L<(Vt5v8M6pv05KpbIN#zO}U3L9c*{pt9bXW$V)VeqawHF@*MqM^lWNM~Ei
z6sqk@7L2h15T2~YV?*7R&-r&)-ddp{(~Cwoa+;K+kqzJCeJV|5PxI^*6a&|NZo{SZ
zwF9sp0Vw0)w*GJTzf|TsWH0>G?x@s0o~l`Z<HjsQu@dg;*}~La8b#0i7LRrgBCoMZ
zIj|A_4J)S5QZ7X{&6(2#?^@AH0H;Y4(O(YRB-igGxu^J8w~o(7+h<7M3Odd`D8%%!
zeWHrm(cisJ2+n)xpr7eJK0SF5kLntAwjUoap(m&TMJ5X(V(}v+q}4Nd1wa0`h2MWo
z;zuB0HqE#gA9wfO`4*o}c%?Gd-<XA~_v+V!Kd*pp1N`Nh@lbbOM9G^UXC#ttJr&MU
znTKJ90kQ!;cuf5dtheyk>5jJv7^jbb;Jkp{WUz-s{iIKN=fO$SM0vFhj4h^vjG|CC
z5uF0<0tFv?jmr~)WIBSc>H2G;CyL-iYs(kA!Lba-C4Gqv^GqLDC+{4lVODEY6IJ7U
zS-`gI>dlPdOCR9ezVL$Io8U&8@>%jhbJbuQIVN@{bYEi5nQNu)%;<p--*?dqC?#+G
zw$t_Hhg>|)rlg;BvyunJ$i%my1mK5a1ft>*fcK&iB|}Oy#?)J$8&_#LbJ3c4?%IeG
zs#DV6r1iDPTR_Svl&o2c=@Gz35Xh`>Np^G9eR>B!&h4vCiMnKbco({qx>Vs`>@lYD
zrENEaVw)TDYb7rDf@n5*O1hnr^+aIguX=!`%;&wCg8L2d!^2`~f<IrCIEbL<^7dKE
zkWHDgmJz5tjtm~15Tu<WEo*<|PTc8X%YS)B;3C}DF8NI`t8qNv4TO8CDknop22H%$
z_I<7)K8te#?`a3z3%3NUA^q}zCAXQn*E6?OsO4+Qwt1vqKI7N(mpj_12ji4G%~LM1
zpN2nq#4l4h2t?v1W50aCoq;wd_>CAme0^XGIFIgaL*m?Zc6w~p-{pp1QNcaLl-L&$
zWbyvBsyrGntubwwWD&l~c$pXN@{*vd3@tskHniH)=t828*4R>o{%gq^dSv|l@pGR8
z8*{H-n#0=Wy~Hg68k&zvFVm)Y2#mg|0`{`GXToF}dTaew`aJa5;}hv&51cqATvM=v
z?9GmTDTR%l+`J<bci%A1mmxCLZTnK9_ss9^K%ORDm+JM+1$FcR0^ipuUzpa=7yoc8
zm(^7qR9#BI$wJ#~`;W@}|E;puQK#N5u2TT&d)q&xS>hk}*Y(Y&{O<(9)+xD{i=o-+
z1hENZxYEU}W62~Bq5;P?QICKb>qme+arFr0o$fJVrpUp*$jhnxdKXfg1LbDOu2CFC
zLPegh>anXdpfUj!@PKC~vrG91m><8{Ld`q^S{&i1f2OM;XUFjFD~3)ddUqa~kYW06
zTvE*sCA1U~kpNY_$t*eWDekfVXYkR*p~axp`obte(*(%i?KabGp10hH$-S+aq<)yB
z-sh=``Vi(xdJ}O^TxNW-IcKux@?1iqgMC*%?-8)`?Zds|d3q5h9Ld_QH*BcB6B|I<
z(%pBt{%fx>oPqji=*P>^w%azC_M5M_S!8afY{0L<O~i{2GnWPMe)hAJx;|&l@{5E?
zd_20@Xb@;AtVo8;<J4?knxLb~xOMHk;gaP=<VD;CFTDH%@t$hMkBGBdoGjwxl+%4Z
z>~#+>Y-^`y(6~h4QxP(obuFI(Ablmf28<}j8*XH0l!{a{J*ZF8q;An?r^5rCddD=Q
zuN-xWI`N#AA9Qz%v!}X>W#IoI&&l!LNcUrgn*74<lft;kia(hk@UTWD>9wTo`X8LB
z%G`kQz%S2rHl?!e*)Vp4f$-Dhnq$A@x<`Op>Rm`cV7vZkH{9;!><wz)&uG{#G53u|
zCj(n>Hoz9`qERU5iWP5>NZ4NJ3stv%9-_K+f?ED}hn%#C?+3YH1{ClS(C7k0-L$PF
z?p_hx&l*`g0-n73;dk?Q#E*c5M);kd=`r-eC*)!Og}@`A{}HfR7G|fC<)^!`d^pk(
z@MP4E+@vSrvNvUeZHwu_kARIwz|yW+FRJP0Hmv=xLx4Hg+0;$jzloXq1L<i_L)z(9
zS|^z`iP}q_+)nemLM-bl7^qv35;G!@*2qUX_zSMgD|l+YqiMGeArS{XwX@TAUtQ;5
zb8Ze{2o=~Divk7HcH8QSm|JcSc?6ui{1LgI$$o399PF5K(-v{WT43tG4DTeq#c`@Q
z{TR^7<~Zu-H674oguZFd5Lq9x>e*;+Xw@9^z4{j=i$+EQT#=2d#D_H?AW-PxIYMjp
zP^hZxXZs`IjO=BCss1B?Op~zQSK>n`r<seqw)DoFs?0UK243nnqtWNji(=PKzs~>-
zvLY1`6?cG_&HkVUgAL>2I(OX2gH=oFzZ1I_z&E+)u1}?USV-}f4Afea*}oKCK3()1
z=e)0bG+fXK)GkF)0BZENXpvleu6bf%SK@wGp_SY@i~^VSlA6>#`FJMsz!a1m+nrh!
zpD<%Y(#z-C!@SFCH_K_GU`h;YN9FZ&;+Q*YOh*WGszEd6$zGAffk*)NfSt1k%5pkm
zL`SZV@J0=FCTgSP5n%A-+>pGA52!IJr$jp+7xF*n=6M86xe>P?B#h20DH09QNZ_(=
zRE7lP2Q$pki#oaZb|z*J_X(ih^_d)k=D{t0tW1Ls@Xs|Q&M%0p|8EAfz(q}bK!!Md
zzGi_)G(4H7r#i&09+O#7kxD)<+HhHY5{xLbUp;e2KD-j6E#lCOKyv&&xpX2_8ND|4
zeli}1DOLf*=?O^V;(lj>xw|CrT6M&?f_*!KaYTNRk}*X6qS^k)>BzNh(W|R%Ao~%J
zRxj=z-yr^Zef6_dwUPM}I>kKK&s4gF<MU|#J{+G}Y2lq5-v)&f*4OMEK7&0H2!Wb~
zE?vr@;wVotC6)*|nSOQ<?ef<xWM`cW&J;nw;j+rL?IXYwhyQZ$1h4zHBr_Q=BtJ38
z(=HWi+%T39OA!kLYgfo<h4VcEn*4w#;ZK#Z87ffHMLJIlROIpo@G91??0;TnUWJ>T
zo$!yf-XDt{1uWo$+#D`kKQXo)**ny=cP>W|jV&CS(v<{Y#$7)!C`0>q@r1Iri(mTV
z&tl1V&U8<U6|9ZYMfCPY9=-qygpX=RvAE$kz1d|1|Ayd0V_JmE?)3EosL?UznnI0z
zf(|T*ul)jcFTfn1UoSuO7gHvN+-l?-a{ABwdzHCZp1E-D<-m;0jBFkeMeA{#IWdyY
zTe?pRskP-ZNmnt2Fp-gcfv_R`4|^>=Ii9=+jN}m9DJni!8Gc#yMCC6T32|qvvJbHx
z@hC$|6-!5;&CVXE+Ir^V{^I8TX2-eR@AVUgATjdr+*!}lL!!&L>dTMzA3x=Dzz7Oo
zN>*=t=oicaGYAvv6e+PB&UQtr5anJSau=$*ayf&@XpH<NjMJXzbE-<m<V{{-Do-iX
zQEi{wnBK3XmRZBmle2TaeXIC;=Fmg!+@E1$L2^j4BxYkO80nbLEqE_SP65w0hTOzv
z?@WrW5Hm6mAAgH9AxJ;FdxQ9{r=(h1g!|V(al9a)?VShj6MU8o?6@x1f}i^Hrxd{I
zzEax%?QQb{0CY84GwJ5A*!Dk!kZn0P9nE2N@(bq@Gj$|#2w53}fP8qdXwaCpB%)=@
z26g^kmk~ZV(leiHvIVFQZ8H=`myfMz0bv5iZVncgyF`G$>0zTdE3-zerp!r~>#%5J
z+;OTzKW<D!MINy%_Er6>OPQ|=2V0*6p3_GLHV2p6<6)B0A~iK-S!W8~JFLl>`TkEx
z^}irT?=%z2UgXqGOHlz!`0^TQ{obEFh<luz*ESZTh5Knq7jkP2BumI<=0@YjkVVS^
z_M69ePSF>>3D+1$WLK&z0e_m6Xi+LK=7Z8{>k=xo2qAJZ1#$;MFV}^$wR0>etmU?=
z<En0#S%omhQ*k1z1TsE;;%4XcUt5uTf|(%1ZYjP{d0D6|;2A)gcu$B}u#>)jI}y*&
z&ov9zc3n{wH=0{sB?tBj<9vv4sPoOLQ|RAcPLsny-cd^f-dAT!zb>sRC|2;y6$7PJ
zU~)4D{&Qgw?;~A<2Wi<<CB_lB54US~zcZh{i%$z#j?8<wS&Cy$w0;s*<Z-^jSU!OC
ze(S{5uCKaoCIl*-mqixPd_5Q1wlVn#xGf8~e)S00_pMT(z}t>|#msgDQN_3&{r7S`
zv|0G-K#AVxVq2B&KB{=ESW_@3s!xgvCL-g5w8rVNmoHN6;Trn-*NQcu?o<fB-bH-U
zxM6k1i@uh#&o3pLyNswkwn)|sEXDK;p^6e4s93A7oh!6R<zyI`V3*jghhIU}`zS7O
zBuJhN{`)%y@7~X-PmNKRVj0*5+U~l)Nh!(CIP=2{B`OvrOv51pcR2N8KGPtO>s#x|
zzpl>;%p@*wqb1nl_p$&dTQv(jQJtNhNt|%mtXhlWUenTNZiU1}hPNZ5M?R<pWr=5|
z`?CX-T<tBlI@xbNEN`o#9aKsSnNUJG1e7fqzj4z6TTcIjmO#~VC?@bRLs%6_D@{|f
zBQejrX;~o><1oO(+2s%>`YV%2NBxAiM+G!Wf=B)<foSjU!9GKO?%opWvS@25N3d#)
z$5!I99%0>D(@s{o3bdG-=Nne&?fD3>)7^uKXJ$ckGIql6Z&gnGs%Ca7j2-@-)nKgL
zV#(PYm|A^30Cr#Yn*MBYs~;I6*4+e!%0YAnbV?H*ihi(FnC6-t2oVb?(3AcYi%l7e
z%gbk#C2o*;GbC0z6@*mgd8zcuW{QlBoAZi#Gao<Y{Q1sQP&F|kB^MXx3TI?Uk_9)%
zFeyAXk^dqRh9;Abs)N=|x*ApxZa<V3;sYyNt<6f*DHZ4;`Ov!06{gl~xdD|@KmU{3
z_%C=5{|8I*|Fy1QL2|HgfEU}{bDAEKD;ME`()owpJCQdJBBjM42JiIZ)M^JIe@PG&
zML1W67@cby40Fh4mwGQEm4L7c;)Hx}{Lu8duD{U4GCV<Mr@bPu=KX4L;V6D-kG}J{
zIX1RMgbu5jfrFP6UTTHOs7m?8mV@vwY}e*Y@g|kuD1X|>x+%2>L7%~NW0N)Q5L>h@
zLwDaLmD4EUyKwlTFb_mV`C@7eQRcAP@^aC}_+X1yPkdEBL$x^QM>UC051_RY&SvjT
zcYZA;Eip=&v2|AwNsCU}+bz%QlWy|lnb4T-?(Iz1?Yeg?4F7s>{pmm097zXOr2YtJ
zaDD2Kp^_oRy5MV2>l6NgLbw7L^L{_`-#EK#kxS~htnO-PBtMk9TqxBkQ>UTBC#A?p
zhs)~xBKr!)I^3SiO<$Q!rT6cKc&WD>s#wc@w@qN;$%g<!fKQ(9BjEVIv10!)WCRDo
zS&XWA@^K+BcpowQ&ik5%!?)pO$z6tTe^bTH5cXqY6?iX&<E-dA)O<>zN&QGT{0rNg
z{;Tl2nD^B*sea0J_~V8NoRj%@An@%Sxb5TF-C*MS>gZPSDz|yHj;3|AVW^HlI8Wb|
zZbkk(2+j`1AVCw@#@Va;ke!hZ7@BXgX)odRiaTd~fGO%c5d^{aJ<lVSLmtd}u8eS!
z0+9AS72Fp-#l86-zTB%6<nbmF7rdLh_(vEo^eOQBs+YeFUr-0c11~UD%1KZ;VItD9
zE1#_pi>Js$l9*TM%^%e1t0BUFPw&b<jHd<}B98#pZVN%wg3@4k=z~Td_aC{1%~u!v
zhP20$*KT)5VvUi9HqX~Ib;^G?{+QlMvT|66>WGDrAD^{Jm^uH4h9rrlkXk1%sMAKL
zL&XbvVM*hx$$kVdGF2QCx7ARu44z1<jnNZdWfRZ86!k?4R<_DFu>35pz&7ydX({1<
z4gT|=y%6Ou_VX#tKj{8eGVgHGh2qQT2Z;E|xG&aSBVN6w&5gq|q{7!ud=`SSi);3F
zznZ&uK8EWDgLLmkY+S`2#er#CNlAcD`2(lSEaK|kJ^6N(7!hXYrGjsC;fW&{#t`yD
z2^yuV7B-(-sO|2TACfO8F}EW@?0|4@k{CoiYZcDe#6H;5U9J6mA4r<ROgmQbiIiPo
z=SB(|ajKp~dsXdj|G;cNBf=?6-o=qNLm-7^K6dll4_vqIUxD*4gMrwiq15+Zily^(
zxz!AB-TSiGAI)Ohijh!KS_=&Zg#s%T%Veg(_M!CQuvEIvpS*%6G8UID-apBF(tU30
z%KWbk3+4~Cx~yunzPixPHZy#m0Xtot<jl-Y*3|ic_8>dUv56nCDT&rMwf@EIzIWEt
zX-iV0i+HyKFYJHSjt}6rllH&h8I4Yt2&E<!^!b#D{7yJKydl)kU-s6ThQclzo038a
z9WrjA3SRo1;I6YZA;AB+pl;+it@_YdB(G;JuV<KSH}Eo_{~`|zR=c|gvkuOzFP6uD
z`-@C@^Vx_MZWYH}05_r&>t}-ykd~ATS*8cu+HdW4gv}*-7`ZpaM9~{D=r30j*Q8b(
zlDl32XLi)puG&=EJ6yb$#7CO?e~c<b#q%owp_6eMg0#C_rFCrRm6Y4q5BEER_bqEG
zLC4MNv^wps>sgonW2~3=qijb7HYRO?Ck-+GIu(AnFv@*i@UV9dFHvA;SI8z~3Gd0$
zne+|{yng1l&%MC&E7m3BS@ec@>7}1<GQXy%GL$f?$}i5DDDNlNcvf(}1V%yLk;>=a
ztuATEG?lvg3L`^@l4O;bCO)?Wj^)}S4G4y8h-;9_&j+G|*T3io(Y`%II~{s4NP`oK
z#jv7me;lYr$PD-L>qDmd@L>=}Dc36(;~%|m+bf%G;EVCZBIuKPC-sEp=HGBZQudG;
zQ%%?}N$-bE+WWaCy9&O8X1D!Oe9loH=bz&EiV>Fy-r6{Yx8L$lZ!S(Gmt!p@QlmGL
zD&wIcYo{rDLO)#J;{oz*17&-%$`CnER<6@AzD_b;B4d5wYy267aZ_+BpOQ%#(7L{U
zTpmF_t!}`^SVx}q3NYt$eaOCGly*&9Cq6Am_;#%}crb(@qZWE{kS{NrXv^>k@(*9#
zN!-=hvO$!neHKZ>DB+JdoK1cP%5U^<{($}p4;)mho>=le?sn<QA-CWBHjzRy0PaGl
zM^P^?=iF+`+l`eFtI;qpor53R4%oiS@V~O$>He~IA@*QB4dm~R)r#rA{g9~0Wv4js
zHLI3?Egf!3EYWrBJD;TFel5eBkN7$-kdw}9TmGdCzD1Ff&z1hn*49h9lsh0oYbtn6
z(-7}2$K)+UoTy^z@>j+5!`9{|33di+lG7V?k@d@C@#05-ZSE&29#t?*RRSM^;v&J7
zammwMNgwIbg;Ogu>ab^%;6`@u3tHFH;$(sL@C<;{tmg-z62CL|N9O}>rj9OS5AIiL
zgw>`_n<#%GXmY#tVQkS#AURwx6P&5#O~ofZt7~f9;rY>i$ve|U!D7Rmz2^KL5^bm|
zuC(Y1=?z0k-1}l@Uu*>&pNjoP{BhmntQ|QX2!!eGt}@!|vpmu?RM;eDG5Fm>^mP0I
zUYxvIE`?=DiO=%V@!f8W`B1Y0XF{(t?-3E4C}FW6B;DL~o*P208qCcTCnW<=iy7VN
zo7=B{G33}UW@3?<W5lK--LE6;l*6q}CQRL2UNf1n?Z>^dJ2rPcf9h+IrmHM1Ei~vf
zxu;V=sWa8|?FvzsXTOLK9nVsonLv7)F`g26g|7R3ulM<7KzCYv2$l4vG>6z!$$!9w
z#!m?BLcieTbN2M{Gf2Hdy_=<LrvB!^*lI|3T)-@%w_qe%LSE#zQbZw&B$1^>eOLxv
z@-tY3ozgP~a9ePC#2JYb)UQRqq|iTx6~>B?k;#s+9gW<Z98yPku%r(-SvG~JErCnJ
zc`lE4lu~UW_e!Ouc@ASm+{+cY1nMukXNFFq(h2;ps)XG_d|<`wC^}l0Dr+BONT**}
zP^QBpfVgd<)Ym<~zW|T7o7^x-6^h8G*cHr4hz@LxmHaeH#iW$N7W6hj6%k^ORQ?7R
z<NPRiXNqMJz9}?$W~8-`n5fYHds%Qfom<5gXvqTsF#?vZ*5ACGjhX07TGes6YGm~D
zEuUgr6c4tvl8dxrEye7iJP2=Ray{SvvMc`LL&nyELahPY#wT-!jq1&^CQ4$6rmB&h
zzWVO<4hRs0CY`}>$~fy{DCV$T#<;@qLM$cPbF$f1SDyA}EJDcb9HsuHI{M4$cj>=l
zNsOfG<hB-tCp?996de4HBMEsih_Yxh2)l`I{A!4nX!puNfuk5-hE&{8menbl#VM&_
z-&-rHF>?^UMx4qnmIUU9RWC`(mv_CfONU7^N&4NR4h!9z0+%+ol%N(ScKkOWY+;P4
zIuefu8vA6&MCYiP=luk0^+<6H&mQx=t4Ba8*UnO}dimAXxg?`Us0TIQb&|BSR}`Ey
zR*>mpRlZa&#0UFTp(<G!x-2ZdZM<WzcSgyR-nq(s$~mI!i3I^^aVJT~jpx*CC`L16
z>XA9pV=`=(ByFUXZ^9f)qa$m%X4?2XZ9B<+Yh}~MHxeVBzO}*yJWgt&TH#L3pV1@p
zbUymI>P5pCisV=)@3*EVa6X9xY5OgIt~BTT1cmB$l#RmM;J4b2Z`x)MWK==vH|Ocm
zcPX0A&PM&0qrZI>#A6#1=(QFp4*Nmzui~B)#&PJd!5s_6jxNFqJr4Vv2z7D3uAI8I
z{=19{1r`6iDt`?UJqYpRGK|zjW=>%sukG9sEiZ8TP-}wIyp~?Hyr2ie0rX_v)7-p@
zEE06t=~!F<@zQ~jI<{x;^RAU)cx{@SSn=&5gY=d>x_4bo+1`Xzu;#KFU-ZxwPCb78
zWy`<*7p1LJjph#@m9@d%+F+fUn@*M0`nG{0BJR;AUZ>ljg%ZPoDY2KmkzEk}zT`D5
z0eTLZ(FCI;-~ubf8@^v7H@Z*soo)vj%mv(TZ&lRQF2Vjje|;IcS46IZjvINE-<u2R
zODGRZ@O*hov#SmrtNr*11Z3&N7(RLz)+$wBL|L?6AxR^|%M}}}szs){s|p6A9RAKU
zeebNC*koQy)3+k#9i~)ke2N39wdm0U-RzgKy!BO;cgyi#*`JiqO3zGtFP_P=vHacA
zNxp;91Y}P8un@k37cHfd7X$%GDVgAW!I<i}tVuoDA93JQNL{0L5U`xh=x(OXPtEgG
zG&VuqkVpamjHTjUj{w@$JFSSxzEc_(4`i=)7=w<0WvPn!`PVL6yy*8YlnQV50%Xws
zUhC0}I5Ku@d|K=`tw`=e+8H7!jFaT#KvmC4nC&z*(WLQ~5$f&B;+!8FKl>!Szb5nY
z7Z6$l;jRbu&<_Ht1AghtORbLQyaKQrGkMBTSM}dI^^4czwDnvQb7#N0<du7+Tf8>J
z9U;({n2<oDGYL^I@2TI`e=nX$koK)jzZqxC%gA5kvnbnicD|V=O~*(7^7b2#bvw1h
zELm2E5$xYSZcE>q9VqRYni(=gN~?<u>evJ6w^Iv>u;o?f{gfX8V@F!YI_e9%Y0GB-
za#ONqn>ncp&~7Y$MTNhVaGC#QauAJAHnphe#8{MI70355pM};f(<g7qz3z5a!^gOD
z{V;&!Yuf3@a_f&HX3nFSiOyty((EuJp{yZX8mc>ZIF<~V#?2M0mqs}!X&v-EFyH*&
z%^4eR1_=*sPzySYzpU%ylF!n?ZjzTcfx}51CnHCz#|>h9Z)c*~6c+2@rka8*jj;Gh
z(bn{QSzmB^=kan-UD(Psfi80sq1VTj(h$p~<<|FlP>mwdts*%o#XI*-y>!i{%%d1$
z^t2^)U`$nD6?;R`NMY2@oBkX5z<E3{mcs7zA65ra_23`(%YDZn@%;@Gz88%2Es@-9
zprl5r4VAHo$lvR7$N8A)>Gps7>&4{i7t#WRZaq~CqLP7^>h^wq_V{~y%hC3RGe-et
ztJp6#XE&D}u!PJ>BX2qDkgh|PPo>Lc_E(irmDfd02APLRJV2>`DW!}r2<vF;d0zyW
zE6L~NVR6NetKz&vf%8Z$ofn8CC!d>a{M&N3=h&wy><YpF*Vg89lezjZ%FZZ9`V)N$
z`&BpkYQy06*MwaGOD-0kZOU(Nl3<FGz1_1(%2QgmJC%=(M^xF27;q1w=I2}@k<0d7
zKYiRRmV?>-%s?}Liln7Yo@-zC(tL}*0@jI$Kl7QYrm;&>mVaJ0Nc^i-7k@`pK1>du
z(+|u!c5vi@@$F}?hmTWlk)^|W2`5eLmtn#I5vhc13Drc!LT*|63O@zQuWc{RMsEEO
z$X}j^7Nlu#VEX#0>hH^sI^-%Z5z-N4wdk1|qLscf#Rr$lSNv<f1>8SEeHy^-_Oowv
zq-y7l`T=D!n%fvA8TvX-v--C_OA8$2ST1Z)y7rWI?{s(<lR5~ixY-4XouJGns$hP-
z`V{=d?H-d&-{MVO7wd~~TUgNDR8BUgJ(nuf0_I+m105O7jjc{&FkoCW6aUi6`n=fP
z?ml^bx#@6kclSwQ6Dkhq8S|%VLIEtT&jyOktL7vK>iGyzkAOOs+QECb4*78y-r1PD
z=*~|Hp2a?7vwEOQ6up;tIr3QC(-C!-pAM3v*pR}HeFSjZVkwlYdApw(ckUmGV$3he
z+0dnoaPnY--1J+d<qiB7FfGdjv+cN;Wr4C_2&TQ-oO+GE;<0ZVRlh+!!sN4#VIPv6
zS!mQu5$I~);SJRq#|g~d+gLl5s@kH5$Gls+e(=LY74f`iNctr@UuUs1F12wROhObH
zBo94T?is|dXX@E3b5i@aperGjvZiw)@bDp~W2A7O+Zxlyo?9MKM3O0^`eLF>;Lh~B
zqht^Nms6udb|E#=SCLhMW$POZ5-M-HiU7ek7!{mKo1ANs6VwsyW9#(wK$zqw+@0ZP
z3#Kg$A<WFu+stOFU|>l#W2k7I*Mc-Kel~w_$hSUjkbYug0>T}``W%g0QG$smBfA=M
z_KS>9RmZUOW%&nAZCfZ|wSBQX_V1YHk2kG)-MZpAWrXe=SDR3>@9aBxR}?sjw%#AF
zCpP|<*x>&|sFzvU3z3AgpIx<k{epgMMi3CNIl3k3e-Er@^R@F#QrRPRvyvh|(c?HW
zLX@O15*E+m9iI=xKRGo^vx!13z=2(kqAraF)C?fa+Zt|M9%c!QKaA=n3)$ZEiL^5%
zg~bE{EVli$>>vOfNQ@OGAqgbj^o()&7SA5~1Hv5Hm1s%UWntu;wF7~v$NI_GBU2OC
znKyJ$RR|?3UTx#ao|Jy58T2<ap+fG0fjE`06I?-fXhXX-<2+fZ$Clx%*g>ai1XY1X
zXq>cZ*7EfEHxfI0E<Z=2J_{fkYORDoJ0<umu=+aJpd3jy@)hz^UOvJT($iAQ=4unr
zclNt+5TuVATWDgx$O>@b?NA6Le(`5iT`2J@R(MNdEt~L;9%3j+ppGJ&T}5Hdl@Yzb
z#q;0?y~*0Xx{-56Tuh99gkismZsJ5`ChZ0Iv}1Y{Eoif0NuggH2Uz%xyYY(=>AWBl
zCf7hv?K#8sG0`Q)D7PFQNczjCuxTehhb}$*7U~w*3kB;JIdN=a_5={7?d{;bslJqt
zr8Gl2M(Nrk;e5*M@-`8gr5N{?C4DkZW(`ll;s1&5{BO@)oQl(?{Gbw~-d8cqx1Hjh
zP<2fi!ybsZ23$V>k|hWDx=VmBdg*eIi8}iF2&fHx81bul1TY*2T*e-@S$#!4(ecR8
zTV7y0{1L$Ra;4j<(8j6KvMOn<9@+t3%V2k7QM_w+GPQ4;tNH!Jl%LDYzfKO4*BXxn
zFn$qJ%Qx}skjNhq@;pSdk%5oAHPG4o#%}lwcWp<Ss_Z8Im@$ekGaPtHyAg2l6gruf
zY`dC?0zS<nckjuRvj1)TvMJSlpu*ifyXvy)T+_ZkS;U9p5D}4^g{edRh>Lv1)NB??
zm8n9-jWgSriwwo^dp@V&bE~RdOkKTPU6ELA+1vD0G8UK^W|3Ba%hEDHZ#aPkB_pkJ
zhnKyNfYODtG?x!#6_)?hh=L+c_H`d-^X>-{;%_slJyE@&RQ{gEWc==Lb3>X{CWcEl
zEP`^c6un^KnCyn&+spmM#x)nm&+}`Q@{A>QMQ&=bS<wAL%Ki(7k?Gv6k4zaO9iE>p
zLeD7BBYg*YR4BcaOx0U16N{_cAl%&kjm`mi^qn*dJJ9>!vq&)`FTI=(^16v-GR_-&
zGL3yGwb8kuOKRfwGy8bbgfdzV^4<xSo!K`(WwT&q<VZNqnr~Zr-+>#d2@<@P2F-4=
zAa+iuim6rBJE*XaO(YR{f_^h$`l>X%>~H*^Sh1HS(AVM#>(5?g^MI3Hal_aF-XDup
zMk|tOFFO;RV^%uhg9QyJxeu@i{*&30stesHp1p5Y{@WUiV0iQ$t_#*RpPc+qLdnnJ
zJjFT%{FyN}eVlxDQ-E6N0}(Ou&NSYGjwd+d#XrvQ&0>?tskFOf`$WYA9y@tLm2#8p
z`}Y`l?;4(%FQ5Ll?Gn=is8c32a*m!Ff4>FvI^yMqU*s?Ob$oW9HRuL%uT`~6Uo>2G
zS-l`C0nevyvYVFfBI&IJPZv&-iTFN4NEz9y2Rf@(N@W!{e;{|-d4fo!_XCdK^AzD+
zo9a!vJpmm2Q5qWtjAvKU^}zn=&S{+=_%&W`ArW$vqi>wD6*))xLAvLFv9qertIaiW
zQ=48f7=s^m`i#qz#KGdAV;>p2(tKCOhRY{?c%G~@TZ>YTh&eckx3QM=5?K(Z)amyc
z(rK!g$&p~*R)rL@MI#W26q$hN8~S9zb)#?mBh`z4kPWNhNxt4r`gaE31qTVy{OpV(
z)h8zNy~a2d);hrZGA{Y+#tlW$>NuN?PL!!os;Z~k<BoM`mCkOg;JpN2uIIE8&oCr9
zbTW~^57r;bcro0#f?z&$%Qcpe&Vq*rNDy?p(2M>$KcShoPT$_2Gt3cW_l9o55bg+-
zSh)-DmfrQd$O6fA$vwE)i0e9YmD-XNc@B*Z{qONHU3mzO4`$_nrI)TQ(TTZdmy5DM
zQsu7~eejQzG@2k-{M23<`RYXD)X3V&hm+Y-Q|WJfeS#A{u2r3Q=s2n}PuxP`c~2m+
z#N;>Hk$DyU9^ay`$E%?el+&a|$6N`+E5jHFlV%(UYN-$AF^Ra5gUVu9HtP|XN_ANa
zrsc+7j`GJLy@Y-2e(**!Q$qV?lsx70XH_par=X|$@4U9o;c$yj&O8iVVR+vPjPZZS
z^si;^;Q`JQxyW=Y6z}?o3M?8Idc%*Ur8C|-P`-<IepeDxJ4eBR>vT~!b*PpQ3r3_f
z#;PO+wqYH*7%q&8F3^gU9~*^QM6Iz#PJt4SbmXmn!|)JhY+wDC24KQvF=Z{SPCbpH
zGQID;gw@0ug<_rd_A=<z3_bId`m|#cv&seY@(R7lw^cf)LX&}qZlyNT?N0sIBji8+
zjzn0HYsGMF<$_5k!SQ6_0)-!+U1NH;WIP+V4H>R!EM%$*FADECm7v(Fd4;8=A4s5<
z70u1W7EInqSyT6n7;6*duONdMk>$Jm)ka%HOHoul`uR5!da)|-lL>ORo1?XQec(Vp
zaso|T^l)k?Uq<RxzcSbEn+i`#@UeNHk&%OtyatkVDB6Us&e3n`#b`D5o#1wESVE)D
z<6b5xdWQowj@@7RA2bjCAK43c#ugrdk5J0TRh@&yd;#W%{qzn<w~)DtmbrF)1ROYw
zRAt`wWZwM_xU5|8rWtzRjJBAH8$#Nx7iG(VGKK<iG1Jj%&U~s~)1UlQP5MrfU745u
z^>#sPVfUrW!`g{ImKk>@g9-Q9chyLDFj+hW;7+p44lFZe#;w(8$_%4R<cRT}$QJD8
zwZn&wDR%@iW`zY0UJ-|_@IRO)Z9IokOTW}_qNS2i<+Ql^t+Kj#U+b89C%-!#l}yzl
z&26WDs)48d8e?PJRT}wQ&q0u0q2Bf9+T!^7iN#rs(RVFtLD%4m>U4WeXm=1oxa=8^
z;XPYrs*CZTH6wQrV&j^FBg%usvb6Dlnfp8IJ8p^busEdeQlj+1F=0r<hsjMbn*EX5
zUJA`cIoPDy7Yu(0Klkfsr#kb(>7vLU2)u%Xk|af=;V;c8SdujNO04sEG^9uU--Tf^
z1H_$m&}chRgQ*Oh+!OOl_}Glk#--s<IHE3j(C{sklJxMcU?4NzQhgR4Dy&9#NYq}*
z?sr%0P;ftOZw^Aa{XS$YfFxjF=qW;b*Ysw`6Pni4fXX0!9jm>j|Gn~pONl#|eJUzA
zmg6P#51v4XEg5OG+q%zjhl!0^)u2n-x0u!HJ7OAmUw2}Z`VsPDG47tSR@G!eJ``48
zyX{T7d*W-;NE0{UeD^we?Zf1t#mPDkeZyShD-rvnylq}Ab~%$*egPh_aR%MZ{249}
z9`NkGlN7a=QP1X7-UwP+imZoKD3;`7F}(NMx&P<IJ0mq}DIUGV+HsPy=YHbYD*`=)
zx~Z5{#eNq1A`61`Vcr1<*>K7beUqK*rAUXN6p(g=)-Wh=oFuM}ydoB>x8d+?rl_Y;
z=}=2aY$1~(MJ~3)(9}(kTlRS(SvVcpM*uD+!@-JIJ=>w=V)Y2J#z>ExNMsY_#BS)J
z+f+N=5Kf>Elc%1j)JDU67T5A2xcyuI#IS{(<;mI4Zi(0&Pj*!%2<;bcjpa<Hu<Bjo
zoRj%-hDU(>#3SIt>p#5<(mI2UR5rkNCo!*+S;(-2IDP(i);E{3^J@_Td6@<8DS>p-
zA;C!_FWRp>_z>_6j;BSKvd9csX~Z+P(P^vXIm1ctml@@W^Fr^uuUWp5wdp<88N-hN
z^=@4MiEFX``{dGdx92y%y<6O_DqCt|P8TQYu~{S9BfGkN5Y9igf0VJSA9UjTRW`E4
zd9yJC?$9{Lq)jA7&W>GdQ4TMAEJ{r3HT7&4dPa(vDicVkf~e3Dwfn$u=8*d|(|_gL
zr~a$=N#4yc&@A$;HNnhg-FR>rfuTl<W_ghmiMB6VHFlp@>pu^|Ge$la&g3#8&;w{H
z(hJ-npEr$Q@ovrp@BB4$b1OGk@aOB`tT2^GRI;a%)7D^N9|j?aB7d_&6Ubyrl`}K`
zotC66M`$BOwtFF>iEN2E&<N>X1Ja^*s3Inwt1ybcbt6-#uN&WbEgJSOvwNn)33NB0
zn);;)6tO|x@Z$Jw;;za}SXN#LvAoDgZtrzn)TpDE4HY6yQz0nk$9g&syOz)2A{C1<
zdqZ!r7@riRWM;b4f;%|qlhc%LOwQJKGb&16tAG((vRL5?Ix}AziuprgM*+!EjGoV{
zp2JPz5HRu^(A}!vuvq;pzM=?#374xuqKGXr&OZBe2wc+-or}eF&&2+0FdhWjR@K?P
z!o)qd)C(Mc+V-UIk!{ndY-z(;fFHQ-in>?diP44sJJ6|m|MQm!^GdPO7ww9Y9XyQw
z<=A_3iPYq;%zdaUw(h?b8J!ktFd?WL#1KgfhC#o;R5kwMoCoPc2-nMJJm})CsD4U-
zT6v?~$-1TM7L$4s+2QgjInzTHeS4a8sApb+JpcP_|4N&DIV|DaK{s^j9j?G2VIP5F
zII9u}ryU1*Jr^Q#OLVz)r?f7!IomenurKj1^Mv{-96Z(h5umd`6Z%V(M&+42y}w8v
zMX_u~cvjHMU}b4J);~my9qt{BNau2$L9aIk^BnIsLq#a`e;``>@#W>u5eWH7?d-OR
z6Bn(Q!3{^xYUWLjXVnPkzW1%Tph8|4P&IO19?C<YI-vUUJQRb<l&Q;rbyH2u$om6x
zXNIOtkqUMEToVpo&~##Hv~B-Qy2OLYy=0?Rismh6k6^QC9g3dAVW-W6{kz5P?&a@1
zScm9+p9Zk;X{R4iN?Htc{i#m>Se>#<d?Qoj5H#8k6#srlmVtEJC~Jr-{k(T!;4(QL
z);n3{JUIon2x_*5P<c<Ruz9hseDdIAjatL|lR?){J>cH~>V{BI{XTtra2`Sc@bO=|
zL#Q%&e=(|_H=7x1BOi7wm-;t^?Myg)$SGeMKB5#%#s{rd(e^XvQ!NZi@RRgBJN8{{
z?aU=#GD`A!)+sbwirz3Dk5*%%(FEhLEZkOwKqs+-DCC$N6h8Y|GkhM=?nMgLo-&lK
zrM1PJ?l`Eht&IIhf0b3}llcMK_Jg$`prqyg$S2X!SC8%x&OCBDLKlw|qdyMyzH&Nw
z4c>VZ>{>3)8WrMNGRb4geCGGnfspB6OH;?$i|JECBVdH-y8EYL!YaZ9{PL1h*=!b~
zq@1ky3pPsoe$I)(e&-jPCL6xg{J}krOVA7@8wN>otr|FjtAwjZzCR%*k=P^_0UcGv
z7>x*Qepzz)D?gH<b}f;wozI8F9Hhc4#FXt+C?5;U%VY|ACFk^{i9Un<O;f~*i>J}z
zZv)=uYokwhpQT@`rVDPTB(P%Ue_%CH$uw1YC4|c#;cqi64)n4ggzJJ2ZtCn6bghUQ
zI*R!OZiC6t!@1c$Zl^nRb_GBLuY9OE0R<gi^4C}&S*k8Aegv(Dz6f?Ll2=%@3_Q=5
z0I?Cb3_oY2)I%F+1hRw<ovLRVyA6~5+5R>U&vM-xjeClDnRPo#xqKL!efW3kvtWMc
z9m7c$Y4Gt4w?+*uC7Ek&7D+xg{?f$mfmChpG50n2vr`5}TK8-P`silz&Ra~|rNC$o
zosWe<e%Q6d$$T}q>y^)&P%#UgdhF0)9`uW;lv2gQX|fUmZm<T9B8ke5)0WEzE7Xx=
z_1!x>`l7OhzV{J=ikXx@<vaNJM@P`}fz9`$RceX7K4#dQ@c|I$x&1cg{!X%z$#|Ap
z*NJGAUDA9!ReWhQ%~RP4g3sU(Ut|19kTPq9A)5jjPL4(_y2?IUuaO6NY*<@b6FVQB
z=HE<ms8|_@9XAgaV8E?gtBU&KD=wU(kmNk=<FLf2lidH_rM@aHGM`<$CMu^%7J^R^
zqL>w(P<UkFpb%{7#Yhl)WtzOSIv1W_&25+uC#CGS`ALbH&CV3UGvFQMF%DsL-QqCA
zzbuHAi>c#KkqX^N5_e$wA>@xt^PhCon?oaEhAPJq`b_uF1m&D>ZGc9uK)$0q>qgBM
zo_Bj;c=Ch{-*XX~t9n#<S@}@ua5^xf-|O`6!2}B?;_tXeDi_=Pod|ygME~o$51LZl
zgQh0vvcdjxf1#)_)x`9bM4fwm=8IUpbV|z@C0Y62thn3#*2uUKs_#EFnHV7aJWdHJ
zf%l-+dL<}?+g5+g)6%xjL3OY|WHL46hlwXpx0hK*Nx5f?qhjGB<M`rrk;Q=<bJhMZ
zP7|%(!JCZ4Tnj%C5M2uQx>fgoBYf#S2cG6z-=8URhl;RAtHuo3LZ|S<kFyfd5h~^3
z6VH-diMKh$9|7a7Hg^*W4<FjH(zYG}(Cf^zx5U>!sjp}};yF{BT%wO#g%aw%4Q%cY
zf(~`T+QL~;MK%G?3bXuYm{xb+Wk*#QUMd?U7XF=JZ?sUfOSv1+Kvy{=qE+;rNcIRo
zIGb>C;Db?O6TLYm27+bm7O)`;(zqN77(%@doBBnM^KIH-z1Xb*M2ExhN3sI*w11T(
zbVtf`k#s!w#$-gLIgOt#u_h{+^HueD{J(lWJD0c5bbJ*N+3eh`>ox1^1&r=<^xr}}
zBvPCt$E&IqkitaX$s0R{@3psV;|nVK*tTN>O6tw;!0)=+=RukgtU=wvBWacCay*fJ
zamorW?uoO{jwgr%!b<ZReQ`Kz*f$p8H*fu34ivrl`fuJ_UTt(Dcilv?Ic$w=Ba$H%
zIc=w()O^xTqgYv*4UyiJ!aZeBuR^V+uJ~5hKOdZJZ|CQ1D0biNrcI=m(o6`>e_x;^
zEeV&uZ0ZageD6a(#60NyZXw8kHQM&aq=0ha#Gx`+Hkyg_<g}w#rNxM)x^VrY_0(a>
zztT^BKXi1j!ZET&Ck~GMW!}xK?cY_Tz8BN6wHYy`d>Cu{j2lOr|KsW35p+niD7hX2
z*Von$FJWb`k9U&FbelC4p5b?Jt*b&ci&&?>ocrqhR0`-K;_#h6Pc<?SFt<v%=Lm45
zK_CK9A#A{m6jM~DW5brPak%5vWRs4p&}0T>yCl;Mr3w@N-|Ib7pWaiVqvJ#IgiLA&
z*8hvQw~A^jjMjFcK+zU=hadr3iWYZE@ZertON%=ccMCy_6N)<&C{|nxMT(aK2`<GQ
z`e&c9_c{N~zB+sCF><l4*0)yHr*ppZc_pr$(Od?Va_1A@1^z=5u?B}p%bs70^>7)#
zdSY%f=EHn_q!#qcR!!NKTsMW4#ChH34E?4@IvQWF>SNVhjVf7S7(r=&R+1v-kH6Wv
zav6jLoO{z3r+0>6-|AAeHU3&n#<yLNRZtGK7})=3QN5=OtM&GV>l(p=FUy1-KdMd1
zsE!wV2dpomb2y(%;%I{vVj*OuoOh_QR+L=(=1UlOF9*n~%O(jj#sLCv!z3Jsa;sW{
zzHMJ3w=Uh^Sk+NYR3Js-i^<F6gPc%N#x9A=7xg>a4^6L@k8u;>@o~t&{Kp7uzB41R
zwXl7va;Y;o(2><-p*A|TtDDTbDN|Kux#_nCIQOi($DOP%htp!jVD~I^ueK(W+Hm4b
zLlJeGRnxXifbY=9DJO>wX737ijc{vtpp43&ztV3PzdeMLdJyE!QFA#Dz2@~bAt_tl
zoxl6mS}54ZH|WDEkaJWfRs-8B8W(AC_No~*=u?Z01t<AF$r>**K9iRJ=3U0>!MkG4
zm(pXUDtw4w%5a2>>&A0pLW6Awa)ZgY+j{&(CpeAW{B~VR7)s&rDKeq?WKWETbkg<5
z!O^^&c4MhIH?rCRpw_@VCiy4jbIZszBT@p@RGWzZ2&o5SuipJb>-Dps-uXQ_VJqNA
zcNE{^!SeNaThgl6dMpkN8?K6+ZpoZ3;QXAho+%PQB#S-5Jor2P&rZ<jS;$H4Fe>i4
zH`I4Ke5=n0D?U4=#Df29WuwJp23>6{eM?2YIItc$Iqg|`8~YDU`IWookF;Ao)ex(k
z;THDPw}#1OX5x>w=v~g^CEfz8n$!ku&o0}z9y8O;GeZ1jk{vPOk=()IGPn1+_Q(Ap
zC;el#3HK>eqq^`s;<-q(_y?x?JZ1kj;KX3u!)`2l!C}LqP$_w)@SQ)YsODFVUIy~I
zuI+XUF~lkt6=2ea)XZB~+Z<D=!Y)sKjz#cUY0w>m)a%e2SOwPW$sgv`oKZAayTzOc
zjWjCKLbMb~K?!MM!>Hs<4M%}S76JR@VFq%K6cUv*4|e{L%zcf{gJa7Kq}e&*LgckG
zbhCTb$;GZ%$YnMk3m29TXMR`Wi235upRgeFY|4#KVz$MPZZKFjBr!nG(yTK=cMA{R
zf?0lnRTvOHsxjMx-5Zd1oR532I#Cj4h|G|1$=}o})Ga8p_P;ni_v&5;$A;ivsX2rH
zq1|;ET<Y~8mHK4HJtPQUeAn*Eod_ixNfNPv;C=bpm|C0l6AxB$08frT9~4g<YR`+U
zw5sDhst3F__AZ2W?t(51TAA>G_urX^8((vE9jwm_JQsvyCX7Dk^YB&am<y|i24ZEQ
zv}h;S0nse4e5zXY>JQ}3F4J92)~4y*rujYvO+mq3=r?z78vgv%#2)_P+Sr7j|63I@
zn^EJGrp|AhY-U}A%K`Tyth)3;1bDyZ8>?R(C#JB#@)vK0v#}_ILCIdm=|a&#j4Z;4
z-~ruF_8plYmSpO?_3YSGXD`Xnprk!1TpSjD0bOzQtZg_LAWi;}i_qQYB~nzD;yV+&
zisNmWFGp+|VW$RmIf36C21QKkzwlejn1c4TAgyym-e~7#fe6#I7|VmxNul^wiI8l4
zsG4m6PSz-o3OyJwA{L68J$S+nx`zK;Ss!V%KK1~gE(s0tRdTwXj^MM^$|MIi#o&2Q
zya{>G(#_)pz?rUu)F5RZ(PV-<W?b8$)%&w#f#xNw=D-~HV(qv2pTIUD9C*&}2N<!!
zTL5(PRVfZ5e+fl+olWE1hhVlhLY2#LeopUdzPQn}9&y$H0IBQrehcN8f&E)h*|mH~
zxg6@d#)dXk5E@M9d<(zp`7KL5|AeP?^Qmto<2aMD&nT&Docd)ui^AR;Yn};cWFq{M
zg<n#Nc17#B?&Zj7Rc8*}s>*<Nw%(y(zR=*&h)}a<4YQ(!77d0?o`2_CH$M&VJh%>h
zlLAxpFd=yBbH$^ny5yb?<MGsfVc^39J!NO0(@W~uqsg}~><G;9q8z0j$K8U8sdj*?
z9vWh4bKtvCKr%-NJ4qSMaWv(Yk$h~C`+yM!bvAkJTt+i=NID~wMQ2W0>Eem|TPGhi
zCPFVS{HSV#mn>%kpPy??%pXef{6Z|PCAhA@z(9glI&C}G<n2iN70qE*_Et`*e_OA)
zD!m(74YT>#S0UtXAVaH-gN`!bSYKUlgW?|=eBM$RPuVbm4ZsnC-l&S}mH$<oO_f)h
z(rtlzwTk_Yy^bjrZ+dzrIvh!KW%z%wss2Aq)BpU{?#7AX1eTxk=5DVB@QYb2DW1O^
z(r7s&*rot-om>)awbeDeB{{TB7m$SX-NN(5buMCbEdf-efEW#+>B(}0M+9uNsI9k;
zfToZ&qp@i^poaVi$T+Ex{(=+mfbOeS=XfLRA?RWBnm3^(1rnCYzxlMNTx1+X5q$aV
z<ag)s%K+fGfGJ0&)^Km`z#x69HC-vqRiSw^Jn|95YW=PtdG)v)Y3RuNWS;O<(!i&#
z2_J|5!$aOtbhSPqb+|eV!MX6|vp1!a8{RVoXqB0&*d>Sj)as+%$Him$gaXQpGOT5N
zG?HF^|Nkh;vA*~e;<24^WXMJ|ZbgOU<i`ZQ<AnR?1x62a^8{PdW`KVsjp(+qN^%Qr
ztL1SZ{Xw%YCB9%&ejWftk80I@pyz@_&ZpCBS1^U4GrD11Nut(vqVMk2R&Lpyiae2m
zMA~Ghk(fk+c)>t1RDd%;Ja$NvK43{q$S(8V)MGLN7*a^uxG&kXWn>P@+Wyt3^Tn`1
zbje}v*_-?rIT9JM%QMZit;p#3kNKi~m7^v}8x;r^SS09Bxl~2D2-2l@KmHS$SG6J0
z!IYcfk$~VeMZ=Itb#%h`$(iUy*mkr-oHsC<qAx@(#?@IcU!DEama&xf&vpL9*Moj8
zmJHuiK&T=P<U1F}C{v)<spV$A)EBV?LEwB;mx?v9GNuEr9MSVI83we4-q|mQAKvqr
zy|vje)Zqi9FwNlpLt_e(Uh^LF>ok`WEiFqevlDrHTuD9Cz|d1IdwT!p8?`aDL;7p6
zOOqJT9(S1TNJ_8*JcF)Yr^dLFOV*sTN-30vSq@DO@}DgXJT#q*_BAa?^9>ADzxeYJ
ztmr%aO0R0Wat4F?5e?gWGkY~5(3LxP=pPzm)Mi=gb1;SU`T!O~3R%<LGK2E!xnNUU
zE^O@wPaZAgAc$x(vdTj8SzptDZ!2eA^619x7YpKoA$1aG{Uie4nl><4Kbwe9BZdi0
z!hy2?F<jj739ybz{`D@GpgLC_U7(+f|1n01nJ|$JHmY-Ndi)+lJbXyo6xmk#8Mm#g
zRu0n*rIQo9BxUBJ7iY+q$e^G}XDZUCNw-oET2JItrg;vHfIFZ`$)H^}WbxS1m^31G
zd-Z(E^s2aG#RR}ISWdTTMO&N4HchHqSq=?XZ{m9m>v%GPMm3i+YMzDTFpw2w-!lF@
zYbV&|B`Tjy9fiOvG*WmpvhJpS&nYdwYwW7t5S=*lxP8r+(3*w^PK>q0EH5+WVu)jv
zpJP<+;=o4eSiQ1D_G#&h{SJ5aA7VZl=hHGWVkDhI|Fe2_Zc=C7-XwUc_)J+l-5H%J
z@G8*ppiVSC>5T7S6rw)(Nyl_Dl_m6I#;37=(gQsn*>-8&M8^H+OADJ*l&2RHEHAvS
zEwL*YH1LM5FU-OIl<dq4>Yqj*)A#LVOLKv4L8nB=bXVS>)pi{It(YF$|J&P1a4nM&
ztAz2VYVQ;5D6!-c#bshQlb9=_S}y<En&UF~>|dM)Rc(lvx~ftqBYf{J-1DSeYGEX?
zcR}cwcQ5kLZpBk4w0lLAEK>HV*~-kE!+-`eY-twY;D2b+ykaXuSB)+|4lTBg@@=u1
za*#wCmG!a&Za<Box^RvXC)N>0ZH~SBMZ<Gb1FVr#48qKT*`8#2Nuiw>AnqOwCdRJJ
z1?%|yW+!37$E>a7qz$08!tymaL5OM2#lH5E*!}q2qz^~3+p$U9<`TZtuftsTahkCd
zmKde{l4FuWorXQc_<-$ghEkdD8!8-Exi_sJuUsecVJW4|@}*|Mg$@rFeTca4RwvFK
z?81@$ab1@B)mf>tf7R+x9bKc#V)Z|!uzUNh=64lWE8OkY%gUNv{FlF2x}C&mr9G~b
z#}SPq#-|JWrfT-3PIde?xYDdM8U<}<x?gbkR%+{8vSxHxG$F&4$_0SHtdFa`*8J35
z)Z(7+@{dJEoHUx>e<Uv}4dU2nSno2|>_4HK33G6#Q#Vbw;a!#EP`V?pdCxIN&!}}(
zXYXb>x>*;%0fx_pI4Al2jO4v|<{jd()Z*GjMQ$~BY7ohw2dKgYhk#Ifmp6|+o1Gzt
z1?lhj=6;AE<1h%A*>Ka$tb@y{<ZbTuHbUy{-@onI)gXMDpzop29JXT-u>LUDL8B3O
zFp*6{xCi1LoH*v7w-amS_jESQ9B@+?BWij2q%ltL(z^M9Qrg9I%*|AN{F$HvAOlVB
zXX*^>FA@z+a@7ZYb1|1Z7mMe)Fr9h}xu%*70R)Q129hEyd<@+Xw;Xft%(xbe=Mb}1
z;!ZUj{)L;sQB1~+1r)oK);<gDh?3_{9+3Zw{GeAxEOyNU&YJC+DW)=L+i?<+CLVdR
z4M$KdzYW~?-CUmT!G`BuJRM2=Vi1oR!r#mnYsO&ZKQzISVR(q6fux@U{JScBDsA3T
z0FA;XxGtd7&t^h=Pc93QhR#0ZO5G-mLd|(xIETy-jB7Z3{q6A0F-P3y=Ab2~tXu+{
zB9UAk8=FGfJ%|ly`eFHaeJX3fowl=pTZ33L<@c}w)GCIAB@u+9@$fA0?yj8%Xq^uj
zCz%B}ql1h!`a*BQPc8TI6M2rg+3bv)8$D|@39m*_G!tWK6>RJ49zUL;w}d3Nqc&WS
zL}id!b{Z825--N`>!j4+xkUQph0V>*@@5+D%ja|vK(NkZ17x?Z+K<&DIPVxgy?UOa
zuz{eSNF^N~PZneD(ARQeQv5hsc;xg}1bkxf?n;3AZqD8ql1<8kj|TwU&R<AGTMzx2
z;?CGU_>+l$%R?+z=@_O*3t|&D%3?(OFl&K+XJ}}I*g+aP^<hYvFcLMs#WjoI_|3R_
z_-N7oSC!|ZQiZLtbs|~?So^2SEwq1D5^pt}(7R{Z+15c5+c5rjQXg4<({5gP&Er&1
zXnw-ds=D*;@Q<_-@?QsEzU|1_80><?d7Yf!=wChVmBjxxOr&?hJ<z^T`BQP9<t~6w
zET_mOi!FgzvGEfX`r}+;Z<WzqhqP7odfzi=ZTiKLnpHVfgH5@U+^9@PQGBQrVcbhg
zHi3oKl!~Ov_vS5Fh9ua!O)5kYPtZUimvrvtrxpDzzx~~+KS>Iel7u>QFSA?(Hj<8v
z(rH+m@P2Vg&S6%QG=~@l4zJz*-*D0YzuPv3uONfe<oaqCy{&Jo+F{}(WJ;w6-yxp>
zFYZzZcLR9(BCnoM0D|}j?uWnj|IlVzSMFP%cu(riyHF+J8h@+%HAy;QStI0`QPq@V
z()mf%l~6zxG93s61gFrw2rtN$U31;7i_%40>1J1CPs1{F$%(ONSa)fb&JXIs54mgB
zeHq>+yC=zya!tMWK0k=`*5-m*%O*+!chQ-W89LY^oR>iRHhX9JJ7+G%FYSJsnYL&X
z$5Le-gtz&7Yuw2I3M?4uzbSt_u&TWZ-gZv#{_>TQlE7_w*zT7V)%&4T7JoZj4din7
zrAv6L`0xq=oy2F+JlW>w!F5ZGx?*WpCQ@QSeffEjyQ%eohc`=J71D%l@utD%LIE)M
zDqlvIVH0|`W}N+G<Zi_KAn8$jE)4nJpO&FpZN|C79YI}X*(8QF_83kkZz5q|h9aEy
z%)1~Ov6@Bm$%)a;)$F{{Ps}%$=8}mOadoM^%M^uWL~){VWOyL3Coaw9r91JuNdM@^
z+=!`<D}6K22;6;5xu5mu-1OpHcgjT=$AOt`Qer6a{?IwDUGwLg{l>XC!3wFFRE`R*
z`RlW30#>OJw?Inq&08s*F*jWS(durwnOCF!m}BM8SUl8zJ9Jk?h2eRZ8HAfc+;1iS
zG(@S^(0Ff$fN1F&4_O;oOjdqL54FZ)4errcZE|iVx1LxVS@+{*M!4B*37D59<Oudl
zL|?+u9SpuUokE0KQxRP?2vcYJ+#$Wn_J-S0wXge2TB&%ej*0TK_T?Nl<xmiZ3s-Jt
zb0;I?^1;v-t~^yizVY_w#(;<wg^`46&A)`@Rpl;uw-*<loly8OyGbJ+(ru*FwtQ7@
z4gd=(49dgP5_<yQ$+B+ZGDMCz9=eL^vGy--(S7+tn+)f`6&lFd!j)o$gki=dI5@I}
zpdBu0V8xA{NxjWCu9lUiBgOmp9g-CTm3E2bF!_lw6f=ucCcW|`#2EaM3Y?WQ=O#+P
zGqYBt{*3DNJ>>l}T@G0pdU{6ll$XQN=?IY=I+IgI*x>q}!4Tr;w7f=LzuHP9oy#7G
zzsK;lj&@<-VKvqx!n*c6Fysx?S?&?fedho99w|+Yx<XwKozbUsv~)=U9Kxg~4}-oN
z34C50OIKlc*EJuiLg)`Ho)afKNSD0v<~@4&S}PT(hu?ZQdbLQ(MO;PlBBGqJYBH)b
z?_l%l-f!R(reX5UX!Vtp<cm0d_AkPCdlA~hp>ujnGPLRMyclC>Fp%fo_k^KM{ihFl
z)VY^Vg8cZSN^M`_YuqcG3=j4bN<nzqOk@w(RTI`7u9e#pM&@RtO&2u=(B?UJ_wVKU
zk%{y%n{ur+55YW4#~b=X%SYr1f2ohuO(}#swM~b%yQeXQH&s&ENHGEAj~4!4WtV)+
zgNEyuCjWZ(spYdWx*6dubT9l}kb9b*HDhBAU^D;$67Sw`e|eCyay{J>nOBCLRV@k*
zAI%lLLX95uyDl-Mh1uQw#zH9*v<?p5qf!PV3L071m{p^QT&ebmb}BIEH5`+Z0S-%n
zhc<j6mr{guy?(knt1)B=>lDe2a8sm_gG5B2k)U0m`OZ!sUeosZJd0-9l4zn-7RqYP
zXo1KFll{0dEH(f$$!Ia-VoXN0_Pj>XS6wS!$_}?<Rp&#Fm9aF8-BcxFq#a1m$l5(q
zTYPUEf~M%xj$RqLUpKZfC>J0|Z8DF|lPUQz^$lvM@c?pR`pRc%#lhm1#M^V#ADqPp
z2c0zmH>Tvx{Ym)RTD6fEue1OKlAH#g2ur4XG+P5L&=)&y<3!^2Cgy6)k$!q*l{-BP
z)Yk;V`y8V;2OGOcBXQ-SDfE$#S&Ac^HvZ{OeYOWp5^-9PQqy?s3Y*YISwC}~KDX7@
zsO3$AFD>-6Z&<z|BS9n)aCjIl=7&AT=hSqz?bYx0R}8P&p^X5h7har&j3;gJji#!}
z9&32rI%5sG0@>BSKXY|R?&@$&SLlY4__zLq!O@<1Zae7_E9H_~b8$uq4ngZ6Bxtmc
zNr@ktE{Rh*?9oTt8EL|q`ODfhDJ^lMjL7fCRgB4x${)ZumZTPO;%0Hj<|Oltx*|i6
zUutSs%HPUMWHnX5=}paE70l+}PIg}YKJ9#soq*UH*yOZEwrYGk<f`!;veQa%D2Kl+
zLEb%#Zxpn2dIj|~^*Sp`@~CH}lf{JwWGgCHsB&wG8%cqfI)ud%d?$j8-&8ooHLFBo
z**h@<(LVe=>7zZCrs=coAu{VpW{w_SQC#+8Z2scEXIN0oSW;)bzlW-MdfYKJARiI`
z+kdi9<z=lM-tsMmjaVgu$8!jttm_{n6)nTaK!2TkwU#N|5kE$duZh~QSITQLC#M3T
zd@TO@P%<bE>+wG{)$RD!3!TyWRp^jBt20~da2-P|yD?54*8&7NqcpFwHWa<u7ypu!
zvLAIkzj4Y9q1Pcd;rP^W)yAd3lB7ddSfmL2xzF%<{^HakdW|${M0><N;4(9WO<6Qp
zOzI+{xbRvy2p(LiTO3$rA|<gaYW@dr`1khWlO<q2@{Vk)l+F<nR2-QHRch_Y%-E1%
zrT>2F?Y6j7qy8d3&CIwO$fac(xBTfRgOnPR`O570xrR{Mq)mWx?3Vty{$c#vzsOCs
zikUG?5I9ATL<<e=nN$eT(HjFgi&lxZ@z<ou9O>@1CQu$b9+g;yN@WW$tc55Ercl(v
z?SG1JUGiw0vmQ4eIbBm973tZAWFPAf8p*xOkw0LP`&k&t7^{*0<kKQDu||upvel&b
zcYWTbC{wYvBegn_v%(p>uH2U?k%G8~I&6}(=Mqywsg6|rSXykm1U%uc=uj*>(}MF|
zMeE{XW#o!GZr$$>Rzdn%t|USC2T9klzpAKJ3+_1rX9rw<C<20&ph9I{z22W`qqc<n
zOm7p`blze)a@GFYo9XexXWE4WFe47WtEim$G<sj|99|l_FVwF4y=6h9pP`<5dPpOp
zpC|e<BOlH=R2m4zd7OCD6MbkK2;6Dg|H{^Y<*H)`cx^-w|1*vkOL>%a)*ktEEg!z^
zib7GpDI0d{>`NEJFJmXUsE|zrTNv~Z8gg<odG0<)cX@8m_w;+;5u^vQ*eg)u$9-!<
z8fqol;(~C+)|x*BLcKLJ>&8ULyJcY-;7oqxB2cmjZ=*;exNinm{AxZDtxmdH;_gkH
zpR&M!*^yPn#|P6@kZ?O0Rp2kfVs%FcF2zvcyTE5MxciLHwh|1SMU+U?JQvq^W2>yG
zYAc%t&4B6HisIaikXX#WHix&D{x`?dr%n=@Y(p;k1sn$4TxC5%dunMb*~Y8!7RgoL
znG&7N842l+JUc^J^?P^ddYCn`SIJ)e1{Yx%quVfX$zT_1KUQ<nL)Yn?Q&}^&1hV4V
z9cM=yg*Mf92d4Xy>NVxa>!CHbOfI{Dc}hdh9HEE8I^;2?4GR{$1vvS!33grbe9I(F
zcb^VQLeLuTU-bKGh)3v;8VsvD>DsspAWMbG(#ghwSh8Xmd#8v1>42fxBWc>Y!6pgw
z2^}O~c6{3JCi?WSpix~if3VDa43;!;oj!5PG1ll#$51KacT?}FI8|T-Y0$S?4K|*6
zY=}Z3>{&g$K%8TGTO^;7elnnCvXQPhJrK62N@U$|S*w(7U4@=pUBRjUom<Mk%Pb9h
z+33we#aOr3#9v4Wi-CFM=h3U~f?wb<Epis+QDqCsu3FIYU3R|wBio=|<AaA_TLi@`
zYyHrx^5J3|xfytQa_D?N-b?&MtG1q?_V+x3pgMgGuIxy|sdKNn3XPu;cWfMHW~|@$
z#V-9`j*^v~tR!<Wp(lL<Nuk|sc|W~c&gCUTe)VH>_7nHxhDD0IujYbX_JV>ppYkpX
za8Kb8pPnh@(-c_N&!q{ANubysPf4mQTMyw!AhKQ~K$)zhA}!#8+&Q?zl6LfIFfaZe
zS{Y}^_Q=r%_CK`2Csdm6XLHCh#nEuH2<xX!Ujb*<E7w=9w6I^tthI_mI4UU@yUFE$
zWh7vdJ%jzDqJ&Ip*Dq>i^|>;%^z$QF1j)ieXX`ynx}bgRN3HLDzZldL*_>q=XVXPi
zD^%%qxW(EgmvZB%IZ2n378T|NK7x43AxB9DgD9bp2jqtoBY!q!3ZBugCd`01wt_g&
zeDCE?j{?RGs^hnP?=~*ko4*nLGE*4kg4%NGUxCnGJSwm6b<T#L)(WifC_S6xg>E@o
z*~JR+Y^EMt!0UAtBVcI31xcZo4EO|u0lvz#y(O_*e1esmJawtoRj;`m)c97EJL=Pm
z#PEO%T4IamCv0P@e}Erw`vuy;xcjY|j8+4zCU>(mae;w>(OBF`Q>zT1!S&(%4>?q;
zFv?mxQF;%3t`r6T1g*j%bO=wxVMN-x%fW2Jcp`t1im_Zup&SayC3{0b%;*|m)_2S3
z%$L!4Q)#4{odat3;jl#4x5-hs1P1TjK_Qxn#5E#ULpp5Z0lX{LsiVfwzIGe3)YL>X
zf!BhnJdWvH(qnrV`?rbh*7rC+>j>IbEr458)tFm~zE0jVIqRHIIJ4q&m7m_;uyR5U
zW53lNmlJAJ4GdyBcf{`c>t6p{Av{P*xz9|x3w_z>pMi%(<*s=IRX+D{TOD3sA0O3g
z*^$DEG`?46`m9=PAmh<}Hd`QOT2oo?P2|x6$)NY7X_NqI@_=!XB#L$GhNCtrP#6H-
zTQI!&<Wrnt+B)7A#qoK4yWj4eZVMcupU$j^g7<06G;{iaLob+T`RNnOM%?0u-CfPb
z1g6mnnm+5ZFxo+Mag61x(IF0=mpq)idCo3e63asdDc4IPUT*%KIT^GB@m_A3ui0@V
zM=47UfyTNNMR(!D$p%ePn}Nw#k16iLzFQ1JpT;b0jNMvX*?kA$LxmM2uusHuX*)75
zpO8}aD-ss`U8-EJ$Oh_4T)z2H1p8H)9AV|lUOdn(QIO`3`mPSDRXgJh`80RB&9i!%
zS@o30^emJvY9GMHjPq!Ha*ha?#hZH9ra?*HdvncxFIH`Ly!lQcv&_geHXiN-_|V6y
zf{Q}2&Dz_Kr_)RQc@I=y8CzwjwGfcYb6~EcANz{wEoaS?T>ZjtJ~`SoPTC2eJlApo
z=cur^IoQ<v5&h*q?Ku7?O784ypq&Y}_>?wre%j_jDsWk9s3O{1INVSBJ{%=V;XNEi
zrG+WmGjF4&?rI*hPNg@m`BBv^ODLg3<mK>N@relc>IJ&NlQs36rbttV7B;GNPLT1l
zwAu<tVB&)m)>?z8$hk-xKT*FE&W^INbrW83y0m*r9Z=fsqZD@TuMM;9<AksEw})|T
zIXQ6+Fmf9OSZ|}g&Ruu++cRl)l3|=<ea1Y^lu~_yC~;sSM8|_RHGHW7Tk8&_F#?o+
zh6>$de033VZq_#2?i^3`(UswA%l#!%Va;k9V_KOcjh7(<mD7X-;r{90*e;k7O`uc7
z_Yohy?;1H@jS8D7m18DF*clk&C(4O*l@G-}8BX6EMcaB$GHp{&>SP=%zwBE-kMGx?
zusk$ugoJV^eItaBD8+pj!!mz<T~Z@$Xw(!)+HrHt`j%tD{?sVJ^lECmN&&i%BjjJ?
z2o8q6y{OOh-hUAIT$YIxWfPp%vlwZNA4;bdd)$P<FAwC)7G}o_PaN8@eH~4=Dg2@W
zur*ZEdgF`YpdyGU&I|}8FNFNKJOzY3eD{XV0%j|2IVj%rdGwyLajt1>52&o>&1UMx
zE<_f-&d^-`DbSBV9RI5ug-+DD3nN9-*v2ut-4sieo`G$RNDRUbsBBzJ6!8p}f+rXD
zRnlBQa;ly1)wp+qsq-s#5-A=1<-@W<m8?rVT}==TB(?P7@6ZT<cq3xtD~C*hSzs=$
zT^&CnZ>pr%t@AIhQ<4uYTYC+1mhQ?i>(?a}p^6F@gJ7Y|lGq|4U*BuHn;u{OK#D)i
zqdN`p{ln>Dra5VVle*u#5dWFyGeKT;4AQ<TbjmR0@%r+y66&-eHg^zX12lXo%agOr
zR$S_3r^I~k1@nMey@o+sS7nuyLHdrPv69@9*}hJ(C@eyqMUD&=hlalhltMr8t|VBq
zhi)~&{GyqA`qyJUE7a4oNj}epmoCcSs1O5^U<Hy2q`Q4Ki~h67-CynIuxsu?rv-xs
zJS4jNMPgMc$LP<?g_46zk!`P=44SooWFMc@o%1=0^jcEvp2T|p@1{1&ge_Fl(&Hz=
z*9FHP7#l{!2UjXekVX?9w)Ku-K}ouR*}YIe*4MP+u04~r?+giSE2~9heo6w#xpoj+
z8|DmccTTY?`RO_m-Vl)ZDdK@wsFlj*EZgO+&kYAnGM~Rcsnu~%sQ@Rbqg$b8tvY&3
zV4(9jip_BTxTT?!GBw`R8AAQycF)?;5}CwX%;tIDSfQeA4D-3T!BArtUI`PB{JtgF
zVZG|&HIdT;mjK%=Bm>QWiQX(mawRbM3pQIo7<TmNWQB(m_sVTYOl?qRCLfsU6<35~
zqR1A!0$n@6Oj5y7d7o?Tp+0xzgMQw{@{qhAh*K@)i<P9mtcp&oh`NUIngsK<_iaW1
zy7}rkfh~a?B1wc<Zj8isUoN4N4Uo6%Hy@6x33bPlFu({Ag&L=7MBwK`Qd>OXA7lbS
z7lef0+m%8xAfj}2G|BiHzQy_3_wM#kT^`cQQXoC%%e0z;`GwvA(^<*vZL@^!3y*B-
zq6xQvH0@TQ*I9@WOlNq7m>3vZ4_4|d+t}UGm_PkY7m}ox<Er?10zaL*hI3>r^2^uM
z?yNgV9-%b--E)fUMi}>6ot$$^NxKbDYr*srh*M=bPG{d!(rfqhs$GuTea!UZ<{9+$
zZ%3Z^>E3S*5%iv~)o-+KUIZ%t7?7MnQ1CzgLtCF#9y#|t_b+{loJ+^m(})SYH$hRB
z&RV-31dm!9Z%se=x7E=pK~~%L-gpD@Mac+zSa6;x?)n-0*@(z&IDa3Fed0vd+Mar5
z9gt_&;MvfAKUc{?181?G>0#KMgMvwLa8(wHmZXHL`m}!pWUq`0v}^dJ52+_paO+dS
zhbg<;z|mczKsH4MXSkyn?Il~p#Y#{}*IM;*hxYfQ*P9i#?FwI7a$)2I2Q_pyv9=08
z)JYz_5FNwjpnqv|wQuNcn?wlhupcEavMsPEgAOXt%<;T9OspiGq?XKqv|a}p;urCE
zPRcTue$|EDgfz7@6k&RuW4uM3jaT{0Y_TgYB{Ilk4AX=d;ENHUjdHh1Z?aEuU-RMp
zz<m~&!Jo`Hr2m;U^$XGGc3dQ*3O-ZJY|)$wE#P1}qHTroE#>KIhDW5&>z?Ohc1Jr6
ze#Rtb!zNub?_~ktXDS)+PZZF!`r@}IUTZ6sw;}E~&@*cD^&Pj<L2o3vP6B6={51b3
z-Ej!WyrnR;kOX@mgEZ~^CDNxOMgK69adgUo;L3pWV;M_M_upnxmYIr2c=%j*gs%#=
zlT-nAL-aQ3rf4Q@e4=~4{iqYt70=JOz0+lbxJYK-cfMSUbx;3mipuoGNnh}aRM*sq
zeapnT@A10FlX_oW{F!o0T!Xq3i5|_>JV4h;El&J=uIS9;Q4ppZ$?jzcOk|p<an)q3
z_Cie6C4YXY|3A1a|97`~A)K8>SW^6P4L23>yMOy3Y3^61j`S<rc?Bmn2jbzG+Ipqv
z0`>Q6zb&BG(orApD@yW^S^E@Ya*wfWtDzh}VTmY%47?%I79$49fJ{ri>&@`!GfrD5
zjx{9WW`&>Qi>8aS8Fm1MwE}t_<iPvhay22@+2*KkdvmV5S4%(X5I-GajKfq2#9~sO
zliAmw3-Xj`k4GOld8ndE$R}hoU<KL0)s1DEFA22~Ka<mHx5UR+4uzG}ypAnmqUON6
zQ0YMK6@yU)7mPdA7m6Qwa!_O#5~S3nw!w?ooWuw(h8C4(__G2U);+0jflI-}DLucd
z33-#8`l7_??cfdN1#YS-H%!jPkH;H2Y~V*S8!nTuAELy^EWY{_k!9Ot5=576*DM2P
zow`OYY`Y(RW|b>TC_8F{fMSoBM1A7lPr7|pV%@3oSwx<c(OoApDmmB$>&KIIqn%`T
z!4WaoBab=<TlK%Q8+hq<s2J2?)R^u8U;Zp_qR=-bG*Y!`8<PASjv6Cq=&f8$IF$>u
z$6qay>S8QL+{!D}2^SCTnfO2Re^&;_y#wy4#c785k6=Qj2!nT=4NPEH$Me*B8w01D
z)!*CT+Se43-&vJ4qGA`9)^ZajfYchO7})PljyI@xjU=#*gD90Ueynj?{Ev72i!^0q
zN;>=TL+w`&ulWm+X^N?iW881oHO^%g0i6OkYF7oHe;&OGCQs;zQ){x}$Yd#zNR6$0
ztyet|83BbB%ISLTwK^3I-$p5;e~8=S>-4~n{7m}_Co{H%2Wo&tkGm(Al3GYOwT@)B
zeE#Ps(;bzOH*igVOHl-FQP5Q!J4jb;gAvb=g{83W&(GDA|Mh1utIPcutn9<qe@jWH
z|5~k%mXukqPNiAil4O9W<`q+l(pL#l`Yh$yj}uP@ZKl$h7W%@Zal=bt7jINr2)%Vf
zAC_3hWN-2)lZQTbJFj%0!Ub>MHHxUExAQ8Sa0u3elpLi^*;K$`2!IPAX5eb^$?a>S
zxIuRbX@(Y_*bzI0{qj(1pkXYt>T-0%E@N^P&e@LLS(^83rWGZu#v{!=`$bhIMH<~R
z2z(ERK_wBW5p`vj*_O-AJT7jKvg!GyhW<csGn53Zk^p4GJp825op$15XIT;drP8;u
zd3z6197@b>9ue^=#3!2eU7NkcQJ1sfYX;vdRm^r|e$41NkB-(gnXa{goGzOTAa#pl
zbS0hPE%zuNmAK(uyF{W>8nTSMQkm;c?W_P<wjCtxa9f{je|R|}j*4q_1;PvjIzu5V
z?ZKW|h>VFiEM<9HAQ~EWfimm6^3jAo>)i=>CHPlyq^(p$DH?sy8pHXUr2E7^KMK{%
z(<Ix$g~UweFbXR@iLeDaonjkLC`F12>xEz7!A5(>;}{~aZ@*k2M<ljl1;vH7m{O6w
z6PAF}--=0lba%d`*jlW$1rM$&G^)OEhRmka@`te{?8z(7p{##!m!Bv@B6X2Oe{wo2
zN9|uBH{U?ztkY%S-0y<SCA!R)Kk{V0Ji1I04|Zxa$#8~Z^%ZMZ=oc3NFm{Vt3$|7f
zFZ!JIGr4*8Ggk9|7`MDLJzp|!or#&j6=Tr+E|Jg3!C+5q!b(O_m^5IeFz7_C$c26b
z#5+s7rXZG-8r^0(N+ubQ&!|#{ojU&bW<Zn<LQkntD3bXECQC;P{LtTeI(RKSY2J?1
zax*p@nW+oZ%JAFHGiH{Qv}DEz*0@Vf6t8*{?K?$s96~p-?U4rAG&1crwyJo&*d<05
zqw`pc2~d*LN<CqKA0BNtK5)Raw^BXw`8!p{S4!w5x`j9AxtI3N(v>-}P^gVhU=`N6
zXUHf^83Y}Nq;n2$t_MDuG=EZ2iCmOKg_cf4RfQ9`7M45b4TlUm;==W+B-&zqRmJj*
ze<8!o{Z~>sE^rTO`%0?NZ<^MJ-IBnngA%Ij%=h3r3F(r!;^1btU<_{4P3?dCt+eFb
zZi-xU$ZXmB@7&+gwtyv)gv1reu?j00u%_`>p(TY8<!Fg{jDWqh><EMLddaD+tb{t6
z7S)JRlqj(s@55`*TzQfHcd1dOgjh^JJPd2s!EnTCL4pZF#5CwID2Xp|=rLpSu++jO
zN0*}%55$Pd$o1Yo?YRFhU{V70j~0rIP4su-x_lu9fW~TO9F<Ez0UF`wtzeEY?l-m6
z2!#d}JpD*GkPCZ`C`caf`brfgUP2Vm{zDV(e`*S`=_Y&<J?%Nql|GRCyPJ$E@mv+D
zyQd4;Z@o}XLYWKD89XLQUmAXx?$P0ZaKqlqx;BsQCpiG{<%TCXeVP08+y$=eKcB5;
zvTt7hwencPGLKKy3O{1UiBl@dX5E(~E-99>|20W|Ag_TbYfQBfi2+Iyj3{M2#`x}u
z+fk5LWcOPl*`?ihJIS^f3o-j5i$@7}&%nai-&?3&Iq|{Jprc`u@6_ALr2xH4ooM_0
zAXpp*C1KSapNsoB4Zf`RDaO5Rr;I;w!6reZoIbb;H+z;OC^cvGUQc#bswbBhKvd&=
zcWll?CzsBu?c6y}4B8p=O$WiLoSfE_^;V$06BXbY_V<T1xrGq+b*hnyYr?qgJ=;x{
zSbZ*qSfHua2_@a2ED%r1vxe;N_)95#!;h|e9#Yszl2p_U6+dRmrS)?%FH|A|p?W*!
z;j&0+e5n976!9aW>1nMM(Xu$|;U@)}a`$m-vh%~ljl&~zWf>{a#+_m12F<`U;#v^?
zzETtRw}h?%eu)Sp^HmSv(J#|6;{$_G?fiD#hT;m@3$2?s>3(6ipA>D}6yI6m9SFN@
z1r6fJENFs>Yl{Q#%AGAJjn*Ijp{1;B8BMZ|YDYDM(1OKJ%m&G0Toj6x;>bWKeZr6R
z`(zW2QT_*2@w83q!#WBD0op~XnKTJ~^1AmPa+DQlXg5n3zfQfcV^@lP6Kp#bJeM#*
zVHIMW`eKTcE39;S+=%no<qv8K%F676(SkM}zaO`LY*P<bj~PCdaIqvI`5fl=Vvre(
zUC3tfqhwfP4g^4POV29)S{+>>65I)bnT0)ab984*5(PX<j%c-vAeRj^{vZyM`R)%N
z8Xf{(4R?qq4A_>dt1^>$%N>ki8-q(zlI2e^gBO&)U#+A=j-=7M3`j!YonU0^%hwZO
z0#-sQ-ECo=5#^}dS>PLJeVPv?ZIFKZ`$;x-a-vb5gCeq1Ra3Z&LyZDZs)Iwhzf|^j
zuyX<8t7txTpj-xvpsJ*)zMS6Yuj@)rrLVYg6T!G_z&)|WmA0Q-KBAjE?_Pb(*Nm!h
zt^T>(ni9vEN(sU9%;QN<YeK#d>5n}Ho3i31y&dad$(M%}>ye3?dcUrT`A}k0<eX|l
z#;t&XGP7Lm+sON?p`iHvy;~=pAHaRl9^y3C9ZSlDl8_zz0n2DhIx#&00oVYy{#<<)
zEO$XfYMP{yE@Er~WArD;VlL3x=E2lw#P_S*FyE2h#0NT&JE~G+vv0rpBs+c0K0YJV
zO!RUIl9CWL^ndexb>~5z(aAVd)cko%maaBHz=lN*p>>G|((k>G-*_4gWPI>IgzSif
z%=-avhVbUb^8<y2jcg=UP1CI*uwa|it<!&K3ci1zHU|EoEh116sOZ;>(Dhp!*hZ7c
zVOF%9yv7*&)%D)58I9JGgEO~lR4m)xd?94;Kdfu^+$9M##K|Z)wW_p0wLMm3!@ug(
zcVB#y=KF5TdmYl+?%!;e$(W?1Vp5+X_KsgJC8=&kDg1><dzs*lQQ&|nITe>-uH>K=
z^vW1$tJi_^w8$z)mJb8ud69r_q?_+2bA2ecGDVd{g#_pnt#~4htGJWAcrfsaA)Oba
z4YhTwo2L&ScRJ+LkW?LX)<OmSHtsq$yQTvQu^Gv}VQsdu5HMJ?xHF8f7h!WG6_Rrr
zykRjmkf3Cd1;T6rs?;}WlYhTbh|^3ebw*L}{tQn?IZbgLe<6)u+*(qzr)_|URkpLU
zRNATMOka3~xIn(yK$O!tO2wzq?mi4AZJKmNAKB^8a}Zk%1KQDGqqtg?2DsX~;^N?L
zw7eVNC%x5R{o2KCr{xXm1UoAN(~%C>ov%Ro@)*;Y6wd6oJEhFxfKQ$hU8X+t<F+Ce
zHl?L<eu!apjUT0e5Qq3fq!;(`8I|#?b61OTQGI*#-ok#pns8(8-($8qfC{rPu>@_U
zu(vakx|+kYo9FIzZF>kkXoQ>QRKl2gHq4%!HO9-c2Kea3Eh`SvV$dvjSqetkc{tui
zv)3o_rXlG<Qhq;70vmAE(thQEaE$?pU`%)lj+XReZqT?=NVet8q=q?1zWGanIOO;`
zF$-1)5t;%RLe#~H6{I9pCr*1G&OPY2z*)D7b7R15{_N~X%WS{(l-_|+u1}9H29jQ~
z^DgW%<fX+8Tkb!!hh)~LMFz@PH^q{U8*}-{qvf_*bdhmAScy3&8x#VOE0x3D`%~-<
z5<?MKs~S2(fWyA?=3k8y=C(hJ`cF8((DZ2L@pPp&sdNLf@=Z8Ge2w4Cu0tEZXXW_y
z6Y!GThU088U17~Lf3Lcv-$)Oh%a<WTH%Ur7HgO|X)u!!|o+>Hq32-N=QhKJt8*%A?
z*)6&dZjY``Rw46*8Ec2x9{SHfjAFJ|XaL3;^Hw)5yN^M`2)g+sZ06h|NNFV{({M!X
zfcXvxJQJp@wa9d9nbYy(?2V6iCVmoLfmK6f8o*A=;2S`#3(fza%)<NW8nT@`{;g9)
zEBwy}HC;Yrq(+|BJiX{udI;y?4|pkX8CJ56ejPQ?Ki1T>Uola8YSPvz_l9&Xl(d39
z=)On>jLwX2kcXN~pHEIkiAUR=*Skw(zS-*P*{1NAkE@!C>r%;lDO@P6rk%#QoPIa1
zA2pS%M5$q)xzp)i(>dn8%F7s8Va7k!amh2%Oq9B?MS>=)e<wpLjxp+WaqR&^wDB^u
zXZKsb+oH=@s!J|!;lg2wxg}MfD+MyK?gO!dWlp>=+kbcdVcxk|pZGYb?h~^~O5Wx?
zT}Kzr>hh`3b~W~nbuCQdNG1)Tnc2P?z`E*UPq#(Z7C2w_{SVZrp1Q3>?sx$rr;o;_
z^5>II)5&cj(TL@J;z<1i0)NO)y;~?fS|U4AHOsu2ian88NtCFx%8}Jyi*@9TF&}$1
z`2CasH4c|0ACWbmXqDh=^Bl#p)i|L&U@RKccJHi|Gvbzp@Gnly@OyW#^Wc@at0(|y
zyZ~!WsM4L66hp@XR4Am?F=Di$2vqF$ah872Gpf&gKcrbk%&X9U>Sd+RXzA=&=r;@Z
zY1qeX8hu_3Pu3Ylo_zh7u><1QYAmc^W=btqo<Iv`sC)e%zS#d8=(82ZH>ltDHR@i&
z63UMqiQ{gjO<CIw27F5x2Go(Isb)k@5n-{zpyztL0heMZzli;t!}%~Ie%J)x4}x%2
zl#@A|JW}?=)OOP2>pkrXekRcHhdVbPICx4}wQ@SX{18`PP~qJREXG)j&9iBr08&Ey
zj<K_>&MW?OQW85#zJo|-ZXmh}NF3r_v=!xywK9p~23+yqXh<gwtu-;W@U&PGU$VYC
zfcrAYNMYxRzYMRm4k#ErtG9b?&|Q?&59gYcQ_+uu2bo+USs@Pa@Y$Z|ADgE^`-pbB
zNW59dl+^(eNo@TW={6qLGKfmnQUr4`$Y{~tyy<Bw*$L=~?B981$z^r>-2aj&&5bJ1
z|1d66-jRgU;hA#Yi_C`H;|xNKg#MNMZewRr_@(yHLUGIL=RG`wbMuNl%nwz@mLKgu
zZJPGIG{eef9kgq3*EdX3;creiR!^;$ks=d1#hMMsUZXW7{Ct{AK&2MBd`rqz^BRv|
ztp(2Es74RK;vsG!6vJeU$5m|JU&=>p)KjUsYBEzOi8g-1^PSdq>JS_GBWKBIpvBN}
zBE@!kTD$YE9k8$uq&%f2@U(DfQGR7QpU)W*U$S0uU#N0h|Fx;m@_s66%B+Gx9DA&+
z$la#jf{TwoC`nEuEbLLv=b>2eM82DX85Qn`RUh^c&@Lm86Z$nY3h3Uj02w!X9TsSH
z`B}aefZKPpNA=sgtXCogZdW4(OAT215exQ*dH?n2xMVu_+geUqRU!-HCisjJRI(2H
z%Hb-~Sj^^-;LU!w>?}sCJ8c_>#ZqaDCM!0Dp7*wRds<9;@*`qWB_n}{eM0n+i1XCv
zn&RUgR6h%g;z#TN8Fk_Hc!A{TMH>py0A&*9M82VjA{=y-gd<Gm0!>=&zXrvN!inDh
z$=un07o__yBN+c%uJ|6gL6xJ$FYP4|tT-et3m~`=gy|9}jK63{#(QXqvgDgl7W#*l
ziE=g){Il54fhwd*a{h<LYw_gx$0#bFB_gZS)*rIg67;bIzSJz6VbP_4_Vy0VQ%arC
z;Xd2E_lGHB!9dYD)LYMyXiD1=K)g^7-%WhzmNHE_Xn+0t<LZy|lu<WR`f(>oaK1lZ
zjgvIL%F^Q@lc77cY0t)RTM8)n*n(6Ed3kbO4XxnONm6h$vlvsEzUxuBaWE6#${Mzl
zCZw+5Ti9QfbDCW)%;ug)#w70EtYi@<m(erUvy^(y65m_c6NfS?wam`eI9p)Hm3?pf
zB2E8+S|@Rnsl-jzn_F2MH_1v85BfZ*`-&Ll(nUz@<T^HPL5??VTwJLtA(SG_&Fsj9
zt6YRjqu(W%na~x{?;&Bwxh4KfN||2r+w8B1FA+*cVZDj0#c%3sQUoO)R{#9W&reE4
zwP!APLe>#5z}L<PD;qpXuZ#4G=Yu5jV+HnoT9gjJ`3u{0vh?VT&p4??Sl1gd6n1t?
zZEohnXND5dSwQ%)3eL71*V4qO14R;Fzn|iojEDiR(3d7T725>vQsrWGW!|fi^qI7}
zR0h+;l%qc{$89uJnFs5=kJG)l2IfkP=TO_y#t_w|A0OA=j<Q7YOkz`}DqwtJcB%wi
z4UdX5VDDP0A90BMIOC~fL)?CDTOB(QkS7rkne&wYP=lGG(})?nFh}CGg$U@i;*y*^
z(i?{q<4q%7nLShrVSq{^61?P3n!RIGxx#NTK!R=}cQCHPqc{4oe1(kpP_$qVLplRi
zxWl$v^O+DhEJ3(PEoSjTOzZM~Q7-RSn%TyNU&$ETcPke}wQf~*=$mmeZ@ktSJthe0
zeBO%(FBE`yF-c&Ukhde}$2IMpsy`jz(nX%XvQSYGjOAp}2-{=&J9?W5;7xdk>??WR
zDJAI(XHv=m*QVj9T%cn*%br-?7*HcF4-E!RAJlPN%8ES=$|G=f&h(?_q&E9#Kfg*j
zTonPFCcRfpKke{gDlx&e8GeQ089D^zJ50`#2yMuG!kGNoqxs?{6&W{bQU{_CdDOwH
zR$(PUXHNzJJ}8bD*8Xv$Bmajs!DniXue6h=X%2HR{8Jl4%h-NsFA~jwoRNL~0XuON
zgd*gwRL80X`>dDyTGF7*g(U1*r_O4rS53V!xbQU069*NAK=4Ew&2M3HB?cN%y^oXQ
zD^Ghm$UyP<t)?@*0h@MgU9bHbTWN<>E~fi;w@kalPg)Ft65Fb+xzc<oqIAg=eWq=i
z+WG$4|2c~Ove{$BwZEdSMmYXpmo)Pt2LQEeYn&Yx?=-qC*l8#JH`d|*wv8zh4*Dgx
zE9+s+7XfjJa@2MSBN4pjcu<^0?2Y<<+uV>_4iv0ln1S!fI_)Xw@gWVx5=A6a{X^p$
z{f8#i_k=*H-4llJv%H^@g!d2T$F|aqk(RhD4tMB#fH38PjJyLv3MB$t-?RCu<!+dh
zV`0o7&98GXE`YGZc8DG+evIw(Q`&HcqEJeVI5C5AQfI1d81D+3>4#nsTjISoPdbQ+
z0nW@+9(pP^d)4hRlsZH{79s|(W`R&tjigmF;o@F0(x(U=v=W_OxPR_CI3`<+5A>EO
z_2ZU{*sA)$Q(kfOEZMop5v6J~nAqDrU&GDVFv~o)SrWh5o3xtni3|+nj-wgZZMcjm
zRY>i@ifD1}2&Bkx6fvPIp(RH!#TP+>nC+*r{P*zgV^_x9lN^WJ%v6rt>AFlkAyvh^
zeipiMw`{Do$`{!BNRcGY09Nn12GujnjdUIx^yxb13Vw{o_q&(=(0xNG@-b0D(xX3u
z12&GzqYWaS!-uw;x+#lE3WW2l2XYtKb5bsBu0lUVu6c*mp}n?M<`vH-+()UIlnRCA
zLZ?B~*yBT+0d>9S?Ws{iXD(i!$02ESDrp07{|L^c^wh`E*R039AKg1qegNmx)<g*m
zK7>v!rhU$!<AJtP<F_pOXaKTjI8rixhdZwA1p=FGxu!<J<f@bq<N4DozPU1R5e~Bo
zALoe7;?F<BE`Wq#H<1Fy{Kvpr|3skMZjvfz4yv3!Er=0br7YvpmH%o2MaC2=?pO_Y
zWx!@GbnmRobru;2dl{#>_foiblha0;B=GC6v6RJ}FYIxJrsa57utHgg?1dQ3pSrq<
zt~F9jy@2agk@ZAk7O>%{KI{b+Pf>t2#%903>ha$VaSL_^9%w3HiQBEunxBlyl0!B3
zMjR~R;(zOnm-{F2H$r+Qrb)L4Fq3|@H9AcJEl%4qz6q07EqY=q0au<xl3~;uC1b+=
z4*lBLyuZl}yESm-yPV|baajVvdM2#6%3Vs)z5~E295!|G!97J0>4P27gMQUb@}GXp
zbqS(zsX)LssFU_LUG%;C%d+GVpR`X_jWbrR?$s^^EGddCVq<AEQ%UJO&tQ2ExF;uk
zAEoTmjv48<{8sB|V>Y#|*jQ;H4#*a|m<74Gd2jTY=<)Fyb)B>;jKnn<r2ov7@I~ZO
zI>+tj6Bdv+Xl+_JE;_B`=~JtF@cpS2>*VHUwn@uDsFBXn*LECr+eES<2#L4u%U-O-
zc1CyKWD-S1yYsod&oS5BMPM$du5*wqE|xLsIl<$#_`gw0r?3u{uD<VOJL>Rt`Cf%J
zBshNLY>jw~>%3rZ3co2j=Wh4^wD*=laeiI9CqaS)mn66~E<qFAp#^W;gKHqTO9&9W
zaSIL|oDS}g;O-tIxV!6r`gxyuKb)#FGxg3nb?QvjeCeva@9OS-OZVP;t#w^L=Fm(J
zEgY-_f#Tw~TdK&?45A8~-A|z#uyamT_MV>PPLtp#@_Hfx_C+wJoPRehh4&lfqf{Zp
zu-9~IVJ{=LRoz~(LCG-_SU!nE5nF9sdtl{R)RXx?*EIh#YL5P&)=mGt?{HfJ+6raU
zA`O(ximXP}ndM~#hg?QEt`fA4;CA$s3zxKST&&0>3>J^MMx3PM2i23wDfV__uSAKp
zv)G;yh8Dy*65kvvmW*eAD^@>7MV_{5Rxh+ov4)T8Fh%&!4<^C;5iH=_oUh4xA>Ibn
zhBCusWWLn)i!@8;8xnt5^Gkxzi|25z2lmhXY*I6}sOvVC>aPq#H;oY<g$h^><86X^
zB;=y3OLmm_lfxWKo(+%IQKp=x&S_~xkOgQ3gVA7_yex!|%_2EL%$sTM>?!<jW6Rai
zY@Z>~Or~CNK4P?VydS?ak;nuzaV};fjFD!>?kuKEqa2v<+Xy^pK_A-|;on183%oBV
z@}XG$C0W~8$wAwHE*9D5&d8Jy$?7n4%o_TGM+OrtLa}gp<!xmezU8M#F!FiQ^dcFf
zG=Ys{?-i6ZwD5=>k|2#Nu%m0WIZAz-8c`OxlG~cirKlPDX2&Q(WMEY--lCW~F`wX@
ztrh7g{~|YA51%Xxq;8>-fk~Qb&S~Uh7e!d4`fNTi)}5Fib?DzXFqY86)KsvE52t%{
zOoSc1?@Bkjp6W59imVk`toff<(NQgkQkz~+B!-DqjdpkPb~vjo_?0xKoW1Er_44=i
z^r+y&@WI(?`#va`-I7zvIszLTVzH;Ye~Ov7+0w;RJhm7>{T_|HZz?R7%8`4Kdp4M?
zgqk0aKnfC77ZqI)_1OMjkjVJ&J(T|&_aW(m!iv#<PqwQ$plst`JgnGM)sc>{<F`2k
z{<}M$Ynx~&u4x&`Q?hp77+av_tK}6GoJ!~d)1KEoX?Y{Qio%X6Y<Xu1S%R@$LGn!u
zjJ7(`utH)1H?1>IJ(q=Pmq?+@k0NLMnufD{CU&ZiG#~xsj}GishXeC^<`S2Tdp{<-
zaE7W_b~`&4a<>&ekmJ-%A=u*l`Osl$D|Nd!TO&ohtJ#z;nHT<HH{{uqgql+H5DWm4
zk{E^U>+n})?)rgpiCrwCkF73YKnu}-PN)sP9xw~T18?cee1XP3lC+_qppBm1uDJI9
zTu!c1NP+e@J+YZkl*Fa`yUYX)<yIE#L$iWZi(rZP14s3{Y4}5DG{Q;nH27=K7JG4t
z4s-t)!(nkSb`qOB`^rs)Pb1QgjCxGLApYEuXt;?paW$KQPpyGU>+Ceoh)oWBGcMCx
z4=247F<q#4%N~tYr+txUk|ygYR_Crrq*Vvk02At6Pgf^M>!^GYy{KqjB^khfwJhP(
zcdl)rz;}q2FqPnqO@!3oqpbd7b&?(DNlCxTKK6~DJ=(@uf($;jH2Qt<D32sYUyv7c
z_<==QhZ6dBG7i7D-q;iE%b=<7992V?5H0&g5}8Qe2n+ouVx20~#5FtB_G;~9?|l^7
z$xh!tpMLwdZ4kRWdyj4oY*+WuMsAiRv`f5Vk7QCTEGw~Ou|q3FiMfPH;ID8bu2sz&
z>-4*b&unByal}?RyQ+;rs|1pZHOF^j)VqJ0vR%~K9RB=%PHwpDs2bTf{oB8c#Fm78
zTk(iZb%8X#mcYqCNBwMLV!M7Fw`Xzac+7Sxyp1W^aQ^U1sF@`fzXBalCW2-EY{z<N
za*fJ>dd1z^`vczxK2))a3<1_CmoB`wraGWK2x(6FeKM^?G;{BtgL)m|#J*FBlg|q-
zI*vWPgs`7R+0c?H-2MY4JCo8#0V#D+x~H-4Imk8Xn_7O?Bnwx5mq9oC<tWaw^U%CN
z;A9ep8xw`d9`@y34yWkLjn-8DV?pPDEs16Agc$jL!t@$!`R}aKlKPBFmW&@AFjz^}
zR(NJDTEgL0P~1eFSb3rgHgqH|dB!(&bc{SdT7rUL*jrnRJ&GI#8;pd(JXH6)VbFrS
z4JF|-@lL9OhPVs-ci}ZD3Ek3a8*S6_^ofW(5WCJ~LgGx#Il&9xX2YEDm7cQi&Nz)r
zDv(LGsA(4yy9a@?Cuw#i+f&tUVzlb(u8@L;OiHpo@innJbM|cYHZb3`cD2MYoZ5fU
zE!NV)6PB`Z8%yeYOGP6i+SHm5Soctsm0xam2^|pC?$GpqeyfCEGRU7fnyoEj=oB$M
zFvk;N(dNp`p^!`rtS2UfljP!GA?O!ZC{xtOEO>Fg#%z_DxnS5b^h9>pfIwH&lr$ah
z->P6Nft$27_=ASlQbWhqv}hMgeb>FFBVJB008re~3ie;%@2<7aGOWC6%tS=HSmy6=
zwCQserWcdtIo3bb0)Q3rM~S)dZTC%#Va`onskDfQ9KnPUFiuK=D?N+h(7w6|%jSpN
zE}{QL;mQAYTSP$Ff2Pf!rs%bwZs7Huq~M2Wo#1h&2NKH+3xh4G7L^U~W7Tjhw%ZO?
z^<fV@#FlyKxqauE>h9CGSN~ca3ZC}yJllIQ4|B8!j`3vXS0K!A6}Y?_i^H#Ruqnh1
z&T<7SM5B1hPVot|Ujf<FKg5g;^VPc`GY_#acd1LCtOH{NTJ&KhyTlvhWM`1DQjt6`
zl5kU?TEpaCAq{2~MbZQ*Bxw`lGXvvK6lhWz2c7rlxz&?fBB3Ns=Sy})a+Y8mm3~RW
zz*b#X=M~1$vvi!mjGI>%N}n{g2?h&d?b-I|u$lHHodHM~H-)c8_Qst<Ry*EsAR|2a
zXPJldtn+RI4vWqKH&+Q0<t>cyVRWd|7uI)o?#MzUwp%A}ChAu4tC&+q{qi72);v%W
zXl)w1fy^>y3fk#W)!zBWe?4B#5a=yQ0*mf}CV^Qf5-1{0?kx(U@q(|kYFySn&4%>|
zTSpSnXV>oSB`{!9Jixjv`yD!qMns(=whgT+@xd&4cRuUiY3)j|6Edq}6lx#E2q<o0
zi&sKBBs&7tGY&gVw*!+EGtG5p>N@YS!6<`m#aT{1Dy@6V%dQcwc1z7SbS1{+;~bdr
z#v)oEY2p%Og`>!=_wCQQH9s;jS+0R|)VsR7Yp@>bS6bDY+AmXW*$7n=0?hQkR5$Nv
zfK-(m^lvkR;gI+{F{^I1@~rN3k-FT(Nm8g{{6gq!{Asnmokn9>S{D%enTYFDSoPGb
zwNgABrNozlu_z$h^VH|!dWt`6*rJ+>C8-GGrM|h<7NvlOBZ1#@ZB=k#66>QvKMS&Y
z3h<&M5$o{UBi{>?Zi8(|FXJB+`h!IJ)Sp33*6I3-g9<^Q*<j2d8uOITfb)PQy|u8r
zN)|Y^YMrmYoZID-ktj`@#P{3Q9HOlx;U8q=eh!|Dl}&HnDmU+k`pPt`P9j9+y;CS|
z_XATm_NQJQTr<cS>Cfpvq*gjyqj;Bdx4j%@HF#r>sjyoT(*^{;3xRgh3hsEXN_??E
z<Jd^a?p8x@3s0ZG=SrI?&X!%7B-B)d5kVAAfJxZ7!;Vv>b4%u+oaLuaI72f((lpZs
zzpxR(jo)V#eLNXmAwc`1;@P{a)w?13HMRs+j;gMYFN{M>az^uJKY(RAvB9Hl$j1Cm
z;X}*_S(hghX^l*+4m1xWwXm;yns~l?FFV5X+eM7Vge$IdzSo8Pj7Poh<WH~=w8}?3
z&|Y1#RN^0$j$;DyfyOz$mb4GICxR;M&Ff0`1@;tC>KLd2wb_NmBpnj&YfI#zjG@;v
zhW1R!o8)3!nR9@_JYLoK6hLoUf$DxAsca`Per&x5F!lRdnqdR&#;N&*_g&3R0d(+0
znJ@;5nlH6SOXvQO?VLuwyW{HSwL)=h&hnuKlEj{EE7+b?=x}?R)3heU!ZUEeUe;k9
zW!$NjBD4b+Y+8p_DEtJ8V*zpxfRN{(5A*tZ|6zTAq&fc!jr`HdXJT`A#6zEVsl3G*
zzl2UHsQ=5xfNJp-ZQZP?1}}{P=T&@S0n-92!<M^2jJSOe2sc7?xM-Sy@w!#w!}pEz
z%m$`j`rt)J^8%#=P~#QFm$|&BjcM--1W?1Lm%)0h&>F0p7Lg;-p%wbwG&7OBei(P`
zRn%7FV+Z}vU;94i_h&S7?IrdlL>x=1YP));2%2Wo4CE5kEgQI$?z~j!?0`wt(O+Gq
zqugSH5cu2&WkNzJ4MUL1*btZqrKf4NSPT!6oy|{-0*FJo;>5k;w6yyj^_>-Pv)@I$
z&0Z<qSy)KyBjY*kNBiQ`x%J6dBN(2MAB9!n!pmt}MHDL~ahap4K+gOh+dKCr;WnFx
z2m&lYmiO?U%-pJ~qY!NW_AH;1fB==>X2XIEeDfJX<`j#YY9PvoA$$eoyZl`Ag2@>p
z!38o~0EJ{$)M~L^f~fyE@;|h`$3(>UOGoCJT@}y;x7wx1QrX2D(K+lPrAjv_ipV&;
z$>CgAAS@jj`a<roGo1dj#b<<Uc~%6Z%<Y*LMGnr4Y|!;j@?~5T&5@@tbyxWB>f+Gm
z2K}uRAN#z7^!S}f<<5l3O=+lULHKpCvA+(e$FLO5Z6NKMuF>y>9zsw3DhV1NFT}kC
zM<vCXw6w4jOXPs$kme}_O`syqob8lrh9_p?&3ic5=HvPFT0x9a5PC{m8yc^a#LKC?
z4oQNqm98s)H@%qoX0;o9obOtXm;eauYyJFJv}!ofOehuTsD4)@Ys*5b4i`)Mjs`tv
z^#xzUx9Kt*EQ*sJZcDbnuVge3(+m3)5`q_i=Gx`vW2W{=)jxKRo-yka)N;SdGq{tV
zYI8$T)4D&x73n@fB*bndQEFU{4Tg?A(IT9m1(C#1L}rg}oyeh7x___;Ry<uyi4-xa
z2QheIrS^G^@3AQS#_Aiv@VR#rHE#QR`>G+ULbV}%-I-0aw<v`%w~Sqb9ACF)mQ<*7
zW}H=tzGWxp>`3G>OvZ^Qw89GS{O+>JliF6e#Ru03sjMU^t;&}bC+}wivRI6PC#D8#
zb4@uO2g_T6!v5JGA|5Pe&fH2hf{hCmDq${`paiXH>3fBsxiZA(6@$2=4)Ws;x=|g3
zISWv%?rq+pdA4OAu^O-hqN$mP0H7W|+Q+)K&qmGTi-z>N;F8Y5>tt-tX)O~{8!O>J
zb|13oxO@=X#f@L9S&L}WfOEEQ@Rm20Rbs;G2gW8N$;D}N7z9h=%PWN-7JoJxzn?d8
z-8Yihy)0XHJFHDfWtz+130wWx3g{p{GdLiczo%HQ_7@}&s&StD@z&@5bwI+Fmm)Mq
z`h#1qbCvR`<&v-b2IH@GQBh+X)=pN-mdocTQp0t7{<3rTxI+5q<+r$)8AP-gI4&_T
zl1`+`q;hm>tn2EXmx}5WOf?8LA5|#TR4fyrgpKGnboTpzVC0ep7q9H8_m?L_D#q9J
zA{H7anwE(E)jS-_zO=ckBnBShwlBHKUg#(N6vDA&nVa(}2b;Gvo{}u|rOv!t_6FIt
z&dNl5Zfux;P>9YhPcO)gDU+I3INxxFOW%E>(u#CRWKKQiGN4dtg>^V>8opzXy&f3`
zeLaW&)riGhdUjnVaeuCC6FmEaEBWc%57D?>Qx>42WBn%a!N0I_%g23Ihk{w3I5sr5
z5bwn%-tE<uA9c;Z1r2bp<~76idsi`Nw0#jqi<`Xyw;|@fq46Y1WBgv*6VT_BhFxOW
z^A+C0u22A-y7C|4S2*ppM%T&L!c6*u=81)S5HR*iE2<IVP2`r2&`2dsJ6H7jeBA~U
zTz`bK<R>!e2p0j?C-{4)+&m0non(2wqW{f1%QxdY1smPoV|!`HR-q|v;f~TcNr*Ep
z#WJQYA=cX9%9|eD=Pl|gj<0P0+VQu$R$Fr%$d(8dGUzkubN?IV%1QsMsKUOXt+!(l
z?CWe0dsZ$(UmSVR&#-6mClW_x>a7$jw0{8_F_(iGz%6zXer~xfq{^}}ZM*5@{9DXb
zEqm@Ki`>r;nsgBmI?)-8t_#wmJCI=6k)%f}St~=wxgH5ZLATXPH<Q}an+!;7%R-s}
z7W6m|f8#k`tWIzB<jjd$vo}dek!Tb7JOo&0oGgKBelb{VD8$1f5cQ4{(}>&lP3#hO
zZ%$_GiS5&myu_|KehW`#1cUdeKhna!#de-mep);rSMq$*wc-65=V{=bmNHW2YmpJT
zDQal9NcuYF4&CNX;uWrfYBN>oH;21(E<FSR`#N9b_9bD_pP$L3rNSY0^I#}J(j7{J
zK+{qH$8XDH(xpj3RTb7x;Lykr6LU>Fs5Gn`_Bo5=>RRj`g=Xe?eCP0bDnp~nE~{N<
z$nrreA(T`n!a!9Hq)xdjqws~%auGS|>dd0rbQ&%gAyrJIv(y(r7VAD(8oBdn;^Y>5
z$1p2FxyTx;oA@49Dy@RdrLJewJAL<bC2coQO`G3%`{y;23SSI_NBME`wHhEDimI&%
zRRxW|81yQ&D+@1C`J7OEkk1t4fsKvQI23cX;=4G=e^x0R+T%jSnyu$NHQ!HBJuMf>
ze!ZnC)7+`4l9X62@gmeD(<O6N;82wxkx-u*U>&>goi+(cwT?Wg2gZ?N%V?7(dx@g^
z><v;Qx~zB&9W%}|Q%3EC_jR<qw##rCO8q-#z%g{i-n11pm(Qu1IkkFTzQDns3|sa~
zsR>}gKZ%EZ$P`#%Lehx^Jv|-XV>`a`FK8-da33_MXa%9onmXwrk`ROs+rwpiHp<*I
zOl{;n^?Rwb_BZ%vozI~yPAX82N%EcM^VIoN*T_4*_2rQbDx{$tOG84>0(6dIIxcij
zz#~HNev!`RF>1^epJjkIbA{$p_Zq^-UFSh?kOahdqb81{>F0-e0=inj&+|E9?)Us2
zJ~qHQ*IOcO)h1+lqXe3~*3A1`9DjO?a;I-wt8cY1HYm734x75)>Yf@Yv;FSZZo=A^
z>E}vYbTzoT(0Z*Q%)&y>aq)+&+AfmT-Qap+=hs7)BH1$T3PF9-QcskNNq@lSX2CZf
zudwRlfUs~B)`RWy##qrM8ocz;>5bb=_j_*xNDbb?A$s_G)npc55zHwQ0!p7o$FUvm
z>(zIC=g%`rN$G;dWh{-!d3M++BvWl@R|ILR1+i{Zx)o-wL{2*g&+%i&%yn>v0QJ|;
z0c}suWKE)pp0V#JlM0sDn#j_bKy5J@&GB!*=as=yvu$GbPxwRcWiwqQc3g~wLpFi5
zP(Yp)SYz&x5q$RcmRM_Rix+$>Gos{PN!j<2mkT?GkzpKKi~3$hW2ZRgc~$B>(;&i%
zw{2!~;Xw-;?S2c5Jdc1v`gfl}qBP&81qFP-Y0;zn6dU_U&7Sz{gel#TDt_{RP~6O?
zcZP@mK?(c;qZ!%uI&nXCb{JLhtvx5xFbgiyzSqQ@#@w5H^#6T)O0}(;;;N#Y{NT`s
zt)ecI^%=1}U3*X|eRGQ>1B$&_N0HaBWD$?)z!lE?gs{by`H~b0Q}#KerVey7+Nk&Y
zct%4O(+djzjw<Q%OQL=M8IE@U0kL#5;a5MEKUVti><8RC+8ckozQ`N9w-^3bQ&|0>
zY_NJ-7>+BS`P9z;C8eY3`hFaF?G_dx-{LS~w5|8`*69{rZya?38rY%|b=zDQ^1N-)
z<r1>8b3asPC;L41?RuzK+h?$_aoHIvzUlw$A{UpW%m=MvjvWkcwI3T=+o$G+S=7A0
z<yOh6o9^2@TqUPSX4KIXOo{_4Jl=;AZTL+38|&*%KAzMJkF`0&_9S8TA~sW>-c=6|
z?--i*!Mt+%{@@WfS2!Q0(9GcfRUJsSa+42;zDi1BVrR7C1IGB^+1&eyi4VVdGh}!o
z$p9x{>(K5dV)o|BL#cElQ;@T}*nNTa4+^QMG2PAUky0)0&nu!}UdUIg^OR7GT%wur
zq{C36j_1Cf)Q#4$n45P@1Gn1@mHO@4&2$U~sFD`3F((#*8UYec;>lI7>#Z2f_uW-@
zI16NLBNtcM!fX8v=1RuW(=S3D8(t5soEQhh1n9&SSqg@1Z`xoWg-8qxAMSGrG4JT-
zfq>!IwTngCbiD^IT<MHjBNHW#>o}jpo5RG1)pv9oOJitDWqyd5b#}}A{%Lu4Am_TI
zr4+#7Fa}mgOUu7|N7PxflX;=kTmsM<GK=vkyNSOVM9w3#*J8o%-Nc0FfOf+@X0GR1
z0lY#g%^Z_fbrEBOt~<S4u&1ZwXtDtReqWb%oex3{nwk{?Gk+s_w0K~}i43g{2?>~$
zVcAcWJLh73MC$_1HL?bx{e_0F8pZ`F)Y^l8iocNP@Q7{4pXw3%VM4yJC$v>s0%I+G
zpGO!$4Aj5B2dar&OiOTow{K{2kVIRrhgZqI>Hn#nQq0(3iYnY($U$^<Qe-U!ljPaA
zg~QD^a@=oDE+`&V_{wZ97e(aeBo0oN?uM<$B6^KmzUpLNBjHUlAaD3#42Q=?q}E_i
zdcMgG`^6<u8>~^%a4y+m%$eKN<+1sMxTlK(kB#oD?N1OTyUE__k|2WM)zT<<@Q(hs
zd37!C{AT5)@v_M5l$q`3O7nBpWiC>>I@V|WuC>i?`)W8$*pLw~s|NzH9pDQ2`gWXB
zSN-btc5cr({Z2VQQ7GCzr{r;)o;)XMMZncTuJ5sE-`^**o12P{Qq6fVqA63wLx!|)
z6@;P`TrjnFW8*ZIDhnU=Mqe6hzMWZpiJ&bqHgpIWluUz`rjvSBl-Z#?8F&OtpHVJx
ze3Opq9}ZpM<F@?4U)wVss74#x(W&@L<LKBC-)QNvG}NTV#V$@(7WyO)DmNE5Hv#Q4
zZr?XkJ#wUVuPu&STbm@AIj<S^EgW0<*73MDHI4T!MiJeGXqImLHPmpE)0$yV%1US>
zmhqACsr^uquGp}Y;Aryu)cb*mf&82_zlwqAzC=GkB<P``O^loNn<eYDHa9)%*9*D7
zTCsHC#-}N4;DP;=^_(i(9|=j%NRNL|UIh0)m~G*x|Nc;4S$GBTp1A#1psC(>y8~!g
z9rdJE`Q(#SB`u@JB#$%1*P9D5qX*HU6BH6{-X|z_oY82c4tlC)|E*XgHm}0&%lQ0Y
zOEcGk4SmjiUlHE{IQaW?wnCZI>uH_p5`{`IR6S!WtNTM4Q3@ceEIjDT0e<V~^KKwL
zRtZdW&XnL6R*aX{;+Ho&D{ZdsObGIvtSpT4+!#ANr@%SJ1Yn;K{DT7e@<gd{W<3vk
zMm8{^&n70*FYfElW{+;TRc4ABp%;bIbuXEO0p|7_d0O7Yj5yPgYWvU>u`D;Fg@@O|
zN{>fJYERU0Fkj6Er^(jmuX^s&>a>j?p*SE$TZm?EoXP;2<s#Q<Yx#1#4n7ERSMm$j
zBRXf~VWt0H*Da8mr_;(eK5HzSAZSDJ8+r1UW(RNfWn4s}OtbkbJqY$>3V4t5?7};*
z>(1%}UPk2D=W}6nRb?5BF<3dEf+YbSoD`G`j7<>x^!M1KFJ7bj)NmHxj&+MlJJPnY
zF1hS*j!_dVSC#ywHir^Ycd$?061%*uH{n8~*jF_#QX16v_AC8$iFVCT@FX!R^2g@D
znBA#0YWl!vBKX8uC2lfUKA=I<{4+#TUx$AG750Hn{BAxlbws0?D*FcYXp7yGBU@uQ
z;A?;vv3X?(YKi6bRIVmTWzsF>69LN~M~)G%Rg9WnF7ugob{ctBZAyi2n%k%AN{mo?
zrfTsx4-@=p+8igX?LX7tvMVd#>mj|ZGJ_!br56sQ12{=9J2kWXr5p&mazq1Dud_!P
z*%G&0UHqsI&eYn-7UOyZ&5>%$33foyFP&151i`m6J!99wmZD*y%6AEUVcljM1?<-H
zpslJmO~es${<28QOfpBA5#5U&b90V9VT7yGI{SIF%7@s$RSo7HW^dp5SK6N|cv8H;
z0?GLz{hj>02t<3mVin1jp6&Y&!zF&FN6|#KaV*~8+ClEaTD{fgM#2ahs9=e@n_@d5
zmNnr1KO5TwRa-yGL%6^h#I+CM3JH~85z0<O`4<vf%YW5?$ptGSVg)Q1Y3kNE_OnE~
z(z=jN@v7<I7cBmAw-*gAHH+oPvT-*meShc+Oj0x}j6=gWnQWkRyXyK1wI4QqpJTfZ
z-=G|S035j)zNX>t{pqZsY*A{1xQhBgM)!nG1am4bDo!R2ptz`i?Tdr+sG_cRDJE|H
z56VcJNsqj!hp4KHE+3zE7$w>8I5U7%_oW4Tu;%;h#4CTKiIkrxL!+aX(N4)w9p+ZN
z^Om{R<IY9)Sq|dLeGd`4GUs_~)E@X}sS=HrQ{5;N2*62-3vS$m{8Fb0x;zN<#Tk>>
z){7m#ekOnCBCyT3z;QNJ63D18WM2AOti(2yjx>)A+HR30QbBVV{effSGqNt1R_Gg-
zKdIA%5h$)v<Tt8Kv<(;{Q;VbWNV{?@mKx#GzWsT(U~r@OEL%3B1Mg9P!s*D_w~7Eq
z4_hUnqS$^E8O1-J7aMau-n5+^a#oRyDGr6z8!D)4q!Yhbt$Qm)mx!2^z#Iw1QL#@O
zaxH&;BfjMAMctgL_4*x+trSCziEO@;iv<zA`syU9z}-Vj$f)k|2Zwk9jv9SE2W4EB
zJ{1gK=nqsK9vm=wx!0a3=oT||=U*~$P8FRPgTI-^ORqzI&JHLiEYitDV8!*CbG$+W
z?nxBPJZ(}pZh9?AI$3qHx4SZRC85XbXa2tKPV=uLtR|cE7p$jP9@(P$8osiA8~u&D
zE~Ik-!R|tpEOfhRm_jm180!C0$fsWAK2V|dr1b@VCVhsAg|Hh5NXPgzmULhI#YAHe
zBwSQ<_VM8FnXp60=jS11wbN*384fbwAGUW~lZ{sUq9AAio4$xo0MO#GbxtT1XM2e<
z_Pjj%4S8Sgs1@SJmIFJp$$Uvsr>01d%Cy43IGT^?JN`4n!H?0SNzBmrSwfGv3h#-#
zPY?g2w4^|hZ^Iwd?!st|JRCPaXIy~Mm-=FXoOY~+_5(7cp=kM7?5ENm?{v@u0AAMQ
zeP@31gPFcW#xt?9$|TBULX%CzaE@X&DTl9Tt7X`Uc!NA*u;s_p57C3ZA~yj&wj{p3
zE6vT|*-QTbRp#AH7)!4x7#67{NnXsS=gSILrHw{V5gTWozyU!hpmPW;NK$bU=;BRF
zIcRS0I?As>qR1g)H`}k)ah;Z*Ww00dQ7s|6!iLr1Q%Ky{n}H(>+q1*aXnC*l)mRMn
zxA4?ONHQX_*8Sr0R%q<!(oV8vi>ra~u)6O0dWkq}^o>)cZ8MXJj-e?EfQc?9I8f7X
zUI};I*0*C|s&{78d<`Tf&USM@7x?48^dA%|*@>wNe<LQ}sln}stoW({=Oz2FwbCtw
zQ7;RRb(_O%u*jee1T?7$XUz$sLBl4>6kWkLA3OE`-fu+|RS`1O#VU8M-H#nG_NFn8
zDf-f1-WU^%5|kf+Fa7rpADGa|5TF=U6x&cW9y$9w4giAaEsfYF06DmvoV!a6jj7I$
zDfo;%r#acyn~3YCc&0BHKA+ufxb*};fO`O{;spz``b;O%(zW#h1D|y~nYWAj6*7_O
zsU0jA2vHe?sf5!?D!i$8_Fvg?#eWACJ9)Q8xY2l?6fZWzCuvh7&93&s6*&F^&Aco(
z%m2hP<;K`pbh!Ae^zAG17IG^66AfiBismFXstmRyM15*HzuFDu5)ejADfB1K(}bnh
z<<S$*GjdEqM-Z0rt3bt>eZrEOb~w}qP_!=^7l)#<GK%D$`Wbpr@p(;JZ3`FAF*q}<
z`<E624+c1Y5fX~|pn^})uwP^m#N1Gv-w7VG4OI!AHrds7@WtwK`OY}Z?(ab->zDwf
zIp{<;t;+a2uEqusr5(J#SBQ_zNgT>?12|Kw5KU=A=S#w*Lu&W<dg4SOgOm%2wTTns
z53j;{r>dZaa*2Hk4zN0^O`;hFnDic06C!o-^TWo(ytky=`K^SfM%cz}>G=X>+LyR+
zc?){B#3fcX{Um|3Utp$y_8?S{DS$us^m*O#<oqC0H05KNX$|2dIV+xhj!jrkKzp!E
zBvcybRGQYer=WT1yQItw{koV;yK2mYVTAJc4^I&%;RD3%SofqN3gH-XPD)J7&Od;y
zIgVxB6Wz{e9C2IDU~!>iU9$Wzp<jIDWN@0iHn&3RIOA=N#`JcwPJYA0jI(*RGKxa1
z{63=xnNRLXZ~$>qlD{3w6#?4NtA{RR2_!8`PC<5^E`nZ~NpS0i6!pdGFKSwi{nJbn
z(XkVYG8XY3p3ACw2F<V@v8t@k)Su|VyoDSA-@)`3p6{|a8q!44zsea@wubtofD;QA
zBIVuEF~3&hw|cES8R=I!X<#i@+4K}}*ss^wwNXs!)ro2SjbL(7Wbaf9fkDBim@B(}
zh?Zop23N0D?3A=PD>bE*wh{j<2s;s;{g{4S-agXV&oy>Uqe-@L{(Eacy9Q-%)hTx;
zQ(DMkf*e^|?6;{&RKJ+4oTf!EM3_jxHbD?>ql*Jc@E|3~!(6{aSXZSj#Y$4TToM`%
zu-TO0%)nLDvf=2k*l!7I*-j(E9t=D7x$8oUbyZCQ@>5EN*Au)v!5n3~;;ayNxSzkk
zoo`Y4;P`eMd|<VU+b|cI`O_AO&WDyHqe2Ojv*Rn`>!?xQs|Ix{rZ!ON_W{Hb95h`v
z2fMQmqoCh(Q83&RD`E(~{a!G1{&uX&P&Rz#XP$Nz+33E#k?ys(Z-J3i;>=RwJaAbA
z*}JIv+mgz@dAmZ{hp;}scw0OfK0pszg#BV-T+aug_zlI_>hEm49VDBWO~38s8?1qS
zZtZAnGSGc+Ic0a}xv`YeX-T1VsKDvCErS&lO^r#ZOJZ)HIA1JeKC2S^_Bt|b{(SzF
z{d6!FFk-}iVD<JM3g?pY$wDFc2AE_NI`8l#`hdTs2mX0JIqGb~q%BJ%R$vp)n_L8C
zF$L+;yv8j0{r8-DZ*$sfiE761Elb41=OMuerx1hdxk=pOZ$(!xbYkK(#dc9UO}x%>
zdNS93MU4m8a6FYy??1aP?C%}S4Fym+r=7eG#GvMuj$7;OAZV~-#n^lI;uBGS1%tcv
z)MeozO?G82K^(!iuuX4=Y>qPq#K6ykp-oO_iyl&m(=b3_CpI6XracLns!jl*c$*4|
zF`K{Y6JEw<Gl~2GZ6zUPWl&DDR1JQgSX=BQ=kvJi3iBl3$^IBiJN%i)j8A*bDB3?V
zB_?tX3Y&vyLX(+nuO?`+0}M(xM5ChD#oLcDS-PAdNu!jKR@i0o!l;M&pz)HV3V*=W
zJ&JpgcDX(;d!5-tl!9r%#&X3vmHuf*RF^z;;d5b`ej!wHV4nqeLSt0YeV2XWQjX6>
zg`5L09zKxZM30CyOmg=T_k)stk>tM5yG(YspEG$C$U1VqcwLHQIoSPBCy}#EI2u}S
zCGtMaLNBpXzd9AXX@+xvdd#t=fXY;6@!^kL?}if?-p&A`AcR54Ib#<<09B0M^L!`y
zedDn+d}Csx$%zW2bUF9=EhruXluR6QPl}q0<1a<~Cj3Z;W6LnA)@9Q>fF1tmQbjVz
z#l2(Lgy(+6Acu|;1jV}I2LGU_&uuXl6|{~-GMEirx^oM;SjL0Jh;G_}Z_HYvqBp32
zu3htrxji9-(RtQCYi>T-l4e)tI2(t97NA2*h^1KRpb>N4rHfe2Zt2uCEqGjk?-9={
z^iMoPwS{?**)J%;x}fq6%C^}t)zLdl+>J-Y3(@)7IenIdzhNj5MAL9ce&STV6gEg*
zCr9elR@Y5{oSCr8odl6?YXdQCrA?)E-ED}XfKQ=#F7Icq1f6Y2!syrTI}?e~k*F5m
za5K9#7dGODam%NLgWkd`CWvZ1am6hK1Xfn(5&(`}rwX|C<>C#Cc^h3_jF%vRLJ$E6
z!t9RgVm-S%mW^1yUEuR>jdAd2?Z?UJ{(`-5N#hT`Bj6UVT6(sP=m~`Ii(v@f&Vw>c
z@uk24wV1$VLveZ*sY9fRhSR2zc^i3ouWH8ifR~tuBv*8e!!rEDkT-v+eRgVEL=}Wh
zFKz6XAn;aMi0YX?aXoj)^cd(86Qra8Y){^ebN?wp_btxrSn|C}FHLPzBw5wXG<HUv
zcia4iI8fg>>=7B&G7MC0%Ir(l8&>YR4HLQMC-cz9F*-ynY*%qwEAC3k1S>p=g#W>o
z&tggTe0ltiKW6<_jNK(O<H^-w)(B>Y^BEPjFh!Z95>bef047kyTQa<s+4hZI+L*C(
zIW)5P6It9cqkRYwflPb=u~D8%tNeaG1u0imG+I|vbv3qrYtT9;$2H<i*`GHuxY+cu
zY~r(qLL-P7r)^JxYAl^V*t@xpO`h+!#ME>*lS-HR5*-4}TH0LuYX>!s(wb4Q(Tbe{
zB51^Zrj{6&C6q^LmjAYE{l1$*jyjGIIR$>*x)72@TQ+pUJKQ1Nd>g<>Eg~$ITX1J7
z42h<9=_f7pTAYTwC)uz#aO`4nz0)T65DH8F-0!R#y*zDxFdqd--$wA=97V(tqpZZ<
zYf9av5W9-J_H6XFZSs4R{r2(ikVSE4#QA$}eZd0_Nbh)XZZeyq+BphM&dT&JPrIXM
zSEmTe+s(%Bx{uK6_tPca1*RWg5TPkbgXeX=wgtby1%x?&BR-b=usywdjtj)=%@+0f
z$f&sx@ZF52J&|52I+rNl$ERUK1=8e*L&n@-r`ctUHN#oWF@Y@3WI0CR+43RWIrHGx
z=7~4{;>|w6>|Gk_@0@2+nlB-b1gdYjW^2h=HzLpx)@bz0mMa61*OkCF-*8xNNa1}D
z92L0d-a<ay%W{T2u<`pZV0g29&6Zh6sq?KF%8_CUlmG&Rp%8;l#ob8CL_T01ztj*-
z9d%xM#t}32#Sj0XDR_jvUPT&|#P*7gRx?hTMojE=U^CT0xLNXmHvwvy!3&&(Pt3T*
zDO)zvx!HAejCuM12PWX<&*b#J)|O@uWYC$X`>foACRNa!3zo91j@!uxhfARs7u#Y@
zuye5Vwqesht+(}#>TZ{<wKl+Xx-B;IT?7Fou17p$ggTHfOl)jwl6c<#pkRSGH}s8$
zH<v1B9>m^7%ntnBRcT&zXo`E_I{|%#hm(xwN=cxmK(PabR(+ipjD?4UnYBZs6+~kV
zjr53knB7V90HKU`9Q!&zmHyQpB;}p6VcAimuS9)QtRU=n%!Vg(ER!NqSLCq_=$DFd
zl^-X`2hBqwD7|qzKHuTBBVmDlk}&mty>4mTxpIql8U3owIB`knc@Uj1o4Y8TvevL)
z`UA^dGgaw*z?gn6(d=^9h(zLgKa3!euv&kKcXa=$Is7lEaj&Yq7Sp5sGu?MJ#0}I-
z+49xtxNs;Kk+H1mVMZj*3FD9I88qjyu;ZZN8uh5?wnX(AYI3y{G%RpB?;{pSZt^C&
zGGqFH@lPZmL5Ae~9PFIBt^$wBlzbUEvz=wTNt&WTYH0x=bBMz%TG>juGNrf)BuY{N
zhMylgO7tpIIWe@n-kSv)4h74Ra;X>HK~xc{CGN+c2Fg4$I3>x~W~$$i+$GzmznX%k
zVA*sLH}^veA)fS%(u{#`1nbj2n}lk!&WBPQBnTEdZA;^09I3;E7YzK1lHWK#=C{AU
z@;+E71K37DlKokhkQHlJMdLZh;s}h|u5vNQj!(QxcG;Ne&iERb%m`HW#hyOuy?_}y
zYSE}X1wkg`f_@sKgT`-l6Yt0(w%)eRvOD-AwMtxV+VG?SRKYxA<`9zUEN}4q>Ygj}
zcO{VvSIQA#$+f9e^JCex{_=VCE~*gy4y#B&3`jHLii3_b=8n<CHMPZe<n!YgR=P8>
zzP=r^kQK>H0&4OlszS=ks|!l}@nQPMr3;U)mZkv9#J<(KKG#c*-&~9CGs1ZSk|1KF
zc!^#XSd6yQ^{_+<Y^8ds*edZuX1dWK%OTY3aK^}D6PKj~1Y1Ih&sVs*?@}?bB5V*p
z7F5|d9eR|EGOKyt^wRiUHFs7d@$b7p^%wS8_8^Fi!V~twZ1$BO;5Pf_O#e5DG6oG3
zlf3S(I1@M?#o^xJz{dq?0%SXIpwfPST815_B}>B(NbV<uKvfmKNPP8iZ@)a#H9w}o
zOyB-|N7X<X7Tx%;L1TGSN}OM)q(P3udM-WpqeJ%lLI`iy&#cBR4XgOhDsr!*k;shr
zp{2<AQ0yeP2V2b@O9Og+v~k*Zu!gqbf<L2-(QB<WB$<6XRcbiFwL%N!%qQSU9h?m?
zq!Wo(yJ$e(ep&zDL-O4%J-xyy`gOOb)&y1whdtV0|MaBXTDYwgJ36doE|B5Io#~|8
zgP#qTB*W!&V~Q^ys!^xEj}3s~%w^GX>wMygs1{l{dIk`8*jdaaej9j^so;YelcR!A
zJ{WCip!j?!iSc)yw-;Hx=#61`L^eQ}(AWYXe|tYBvwp3qzDaH07xw^KG{txEdN7g=
zKWk-S+z^hoqFvOA_DO||8H*dNLJDE_NLBMy$JFjwCgnROiFZSz0S!!>;vF}p-(I`8
z6=rXamlM4om(tG%vrUGte38TyeV>NZ$^}=#9k)!&t}9m=aw02;pWXF0U7vzWWeDny
z$bUh#wH}jBrnVZGJ#${LFdbPl^~n;YxT_nE=i(&-0!JD3Z;u#N_I{Yonn#VeuKYP4
zUE?4(Yyb3~RliSG0Jsdw4Iv{9{&2-P>L%`SKK-}p`=O4DH)+}*wWZB^CxcjQ2DR^t
zFL=9yvz&fOKf$jFlo7ae?M~Fu3a@JY5_&}K`aW(|wg%^Yap7(z(_`C*664fjO`M)*
z)-VyUAJf{@q@^9H!V>(Vv-is?sM&bXgL89nJb--$u4uL)dP|Y@%qIl)%(2+1r2Lyo
z2~6le#OF}liT*k~u9&nvN-~c6+9r3c%joz$Xw$_2Wwd|{IukWWqZO&u?r(lpL?^T-
z+OGs1v+2D?r8R^FBElTp8k>U4IPT=__ja*pXZJ79_a7AnO{$x+bt1OnRuQsnX*pc6
zgYqhRYd_P2!Qp9>`RT+8wNk6auG-&~=nQl->pf#{T)&S*w#eX1M+jEc?u7s7u97vf
zB`%>U;b1>a+)?FS67jHld^eyia#8vFH&nGcZCYm<-GEwTWvP3Z{vVWXJkj?L06K)R
z_m3MJT1TP5<aIfxUzC2rhux1RXBZN8ehLb+9QXWG@5^V(O9F-^#`9kzy-Hso=tb!W
z-E$3xk&n@{DXoWSRZq=ZEN%|8o(VhKbC)?o-Z8OaEG6EnAUM%}L4}A4klfn{c6{Xw
znaFTd{p1WhEGv~u&z%6x_CNG}TL_!NYH4?O&_CpMY$a6W+=m-73yo`5`}en_4gJ0$
zV7WGS4KEw!+Sxr#jDX~83#g=Rx6R~7B%>Uh#t`Pn7%ncFqF;Z<eKRPU^O~O^v1hVF
zJFb96ap?vC4%Q4!q-Q_(+s!H@#G-+%+BK1BEDTZ6=-=Tyej^HNj7r*pv*X%Pn@h`!
zv$GW2peknmQKF+rQ)=f3^N32D>sfBi=GXY~^p=4=Q&!*wj=uR7G``{zCgLq&n_aQe
zLgo6R`(uYyg<+jv3Xp|M_+uEdhVar+t}vZcC0L`QZRFDrB2|nMGwugZ_U<^PaLcc6
z8RJ3p9Mh1Mh*S3)aZbM~Gtr#o))Wmgk7GeC-Y=hf)HNtsuSt{PY;$u5n*<$>=AU>D
zIw%wI46>Z3J%alH%50wqMyn=eKbq&3O7E+a6k=^+N}=aGFG#j!iu{8jc9d<78BMnm
z&pU!jDn-mKOhTtH_GM<O;t|_3iyzB3?#2<pxYp3P&4Q8nJ;mflGYcNGgzlWad~O~j
zK^~w$m~~QCCKr&J_kF!VXhox1*DU=L2NB=jDghgI8Hha`ka$YDQon=8k8(*+DfYnF
zS+O`|(HLUbg0aK|slxMf=inJ|{3QZTk;t(Hsb8tzQn&XZ{~S&-&jxH=9SHb$bUDKm
z$*f=C;TAS_y)%|r)SA^U?9447XJ_o~yU7}>(-QS{;z}4eE$7x}lXsg-WOCtPi?vAX
z_=J;r7$?Od+uN0-7b1M?wW-CM3H|2e^gE!4q2~#4Bv@3$K%DfHXNN;fdKo7(g^=)T
z3+S?D|ND~|nfheHyKfDHN1cO>Aub_q<eIs8EUMDztoCE6f_up6Rlqs*AC#Yx4jfNm
z^BLS)#$ZOCaxU!GR!k6zd5YUq7SEOP>(b9v=)@xvL%atj^}&@%!4<^^;{@raTX_6f
z&cddP#(OU1-{?x*J_?JJ=~W~JSMSL218HUQ;^@d&0!l<?v%gVqu}I#A3~VW|&>Lq`
zX*}bnmnDta(M#&+MnB0GA2P9{#Jw2*6YmWq+$8sF<`Xh@P5%7?E7aW_fqu#%mtC1o
zjD!G?bOZ$osXE#_bA|!=?b&=kWDQ(0&nldN^y6}7WQgO4LkQxK-cOyRgG(EoM+R9L
zK;~YTp*Msj&P1A3WP*Tl-wxYYxd#x4QWQ-4Ac{83gGAmlWpo1RA`L^y_u&?|kW_Mz
zWP#<jz%c#Ft7X4HqK&t)OanjfmI4`iwIhRzd!jR(YLifc8nr``O4w1T5B-dh@!5)x
z0rX8qaGi7>a7bZNe{nw~Tqlm;N8iApj};L?O*&$QCeW2c%dXLD%M=WydvH-f9n`4#
zdpGxu%tbnK)4!-+mMOqGYiHH+pg(_gaguG26gLzs5<+Ja0;fCagF!{D7EOt{D?)D&
zAqz?*Z;5(~WD2jxAes;X?Ab35QS&2{*-qUGIHtI()>JOjcIm9*Z0e~~WmAE!Nedy=
zi_cmnwL(m%G#aw+UMhrse;N5Vk1NEI!7vo035fxa(NvIT+1x+6>-IS>&AEy?Etk4i
z_5RRa1ti$)R!V^qcyZd2S*hJ^MI_v7_P1JLC|j<ZPO7oTRxUA$E~To~rEuG}AjD>G
z9iuJU+PU`%L8r4H8fvrz>wNr6`7wcOz(Gp)kf>plP7m-2mtM)vS2i^^oBIZT%Nof;
ziS-kUH;H$KV6AG;99<UP?XTwk`FW!-=kV!Vly$5N>9G@=365-6eFL4=7*2SP4wXwt
zM`6*|36j1OE>r$N<b<|;cP#n)w$N`s%&OP?owH_K;q-noHoH!J3raxVE?OXM#3eWM
z|Ba~he^32hfK`Ub8Mj`cl5;K@o1F+Y+PdQ985J@kisAAP%9U{7g(4U)@?R)Hi!K8W
z1zWA%@Eu4}pcKYQbF4WnEVTi7Gt}vNU`VVh07Kzt5H!~x39!zuSuoH(Fu$+nc@Ut>
zcAu(D2N;&o5?2JhY&TX^%_Q337!rx)7fg@5Cf9Q-#X9a2v}J^%3i4q?-ypt0Uz|1M
zdvj}dveFalV-G5t4Qd&a4sdEsO36S8V&bIPMQ7)YKCM*9$vF%%l0eebJGQuhQUN!k
zyD1f$hnj`aAs(n^BeTlh4swhMy4U#A@f#lx9qfHBeNPvMw(U311<$S0%EZR!Dj6Vg
z{x$g)^9u`@<1sH^_qq)Vm!|>-HYSYM&#o%W5*_98u%VOrZT?JB2bWp5emZyWb;!t;
zY?Brm$adLfEqQ{gBQ$2JzvRM$ux?S^I!{L<Tm`f8qK7M37!RctSA%EfiKrrys7Y@r
z81<3%U9q7Kod?qyhJ}jIO%jzvL$?P@$;FACz8rC&Wthw{Bt;NA;i!>44+v+^kP1@m
z|5Rd8Vi5<QY%-pNx@qBl`AFNL(lP=kG})+TZ%vVLVtr80WrIB11_ZTP^sx?Z3cvWG
zi(^x89q8gUHCuP(RNjkCBmVL+>t|C<L08hyP<&lPg(qE&gar6t$fMdUK^Q7Uaj!7x
zAx5-zh_55Q>+He73TG?emRWyUFGDhQ^z;Y$cXY06NQ=R7Rn3#W4`GVtQcuQvu2u<u
z>K5^!<;|&M=AKhg2&}7kIgCs~aP(1x-N)r!AwBpKJkDE@l?|INfpZT3pfs?Bxroj%
zSGMpjgnnj3u2{-Ri;Ax5{a@B?o%e9Mc1M`$>rniHY9{{74WREebdk0c<W1-5ipV~K
z17I_CLg{jHHIMOUXm`L%o5gC^E#|GyX2mOH9PE)rap9dwf^@Vhk7xx5iLZaV%gYbQ
zT;?qL&LgyhIpAh)IwG_qA>CLs*?!2nRe1cs_R(Z+)WMGsbJGT~i9BYoc83&_{_X|1
zy{)dUF>EuSr2y2iMsURQ0&*gg+~OaGcHD474@1+qwEJ~q-8Dw0!vqzA;-?zNiiS_D
zPrSpK2X~4Fl}(sSv5HLHb0di0(hw<1%`B+|Y>BVN<o2Q>j+7~SRd%66gV!Y%ya!2)
zeM+u!s!at4`8kY%EZG|*OAVn2$MeA-)woXp`#g34T(Sx9lKfu^h5sQP!oM^k|K$(x
zxuYPmWYsMY){X48i?GC$LltnlCJ#fUq10q}TWK<2!O=B!G*`oV-KY9vGO4IcpQgpv
zn_K*)%%WqT)iQ@Iu<m3gjtQc0Gnp%m?zAl?R>0G3aoeNRP9tVn_6EmXP2ZXrxvnDp
zuv7eXXMi@h!_#Y-BY9sAcP15NS35H|0g%-{^%npZ(*>VLcM*Fn@o5KBzH!pyS(LBh
zoJ56lNd(=Ft1hAVc{FB;lITJ?S1Rz}WV#6SW!_=X1i;?he6BF&(bzUk>*%bB(<O~L
zL{+8m?W!5u0Vkjr@!6Kndv)ppUHtNl(|YSdBc5ORWRNAneb9dkA_%`CkP#59cu|yY
z^;YZ5rl`-bz=<ibfx7Vt&Tb3U4EYDewJ?ZaMfzQUgBjDq+t-}k?3`Q-!Xk1R6qC>9
z$G#3-kY{Gxn2h(#nqS<0@GB5YVaPK{`1#nRp0LCqQu72U!O*-kTO;-f8V@W<R4j}m
zeq3=HehjfBZ>Hro$LY_d1TNiQ45aA}q+a}mg|+z|J#{-f1W@=+=@281N||huLYYt!
zAW*?IQFW1r6{9rMsz+m7iz@$&XRx5DXl#0pGvhIhNUen2!Le^zrqWxxusU~8A-N9s
zCS6vEJaBEV`bIB|f}1Hu;kBR06z9y3c$e%t{K66l`#c>Uq3YLDzpKw`vuzkxdNtzV
zc29|wT4=M`xfNwnpA6832KaIDR>&0&Z(cjBhEe``&(ZZk$)&8wNY37OwhceFHTa5O
zC=rY-W{lMbbId&m)hXB{(n-xG{H#X(GJAu-oKl$_X4j=*fUDQsB`{T%k)y9Ffs?pw
zcG>ojW`2oNHTTBQ)z@WK!xw#aSJL();RlN25u<h00&eabwiDhdrlCV;R95DcI@Amn
z2n?t=qNW%C7iVIUkUo{BYYh?Fl=k+B7lf4py6O1>1>$nTRewF5*#6H1$CCet{ObSe
z<zpaE-J|&fT7aqjZQu1v>$`&J6MX;sisku*DfPQh#0Kp=<;5$LUdf;QAIMjz9<B_U
zYJL>_yXV99>yGHak5-QSt-?3^4X)P8z<>8R4RZy)ef~APWbNg8GUK>CngRQF&uW_f
zGw;)v^v2QCoz>Fikt{*lfA@6!dldZ6S!-h*S}os2>?Hr2aanDMcI5<=dA@g>{y#_h
z*g1CwJH#F7_53FBq<!qLSN!iDynhdx{QJMpW9k1tjsK6_jiTanZMC29cf|y1z_Jl}
zx)g!_7cXQ3^l~Nw`&y=u$mS{AS`mN1Vigf@`Y4Ogg~X%3SWYwCjKPd1J^cy1{(t|R
M|JRQO{<H9Z0Gg#~00000

literal 0
HcmV?d00001

diff --git a/index.html b/index.html
deleted file mode 100644
index c5944a5..0000000
--- a/index.html
+++ /dev/null
@@ -1,7 +0,0 @@
----
-# You don't need to edit this file, it's empty on purpose.
-# Edit theme's home layout instead if you wanna make some changes
-# See: https://jekyllrb.com/docs/themes/#overriding-theme-defaults
-layout: home
-author_profile: true
----
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000..1e6f099
--- /dev/null
+++ b/requirements.txt
@@ -0,0 +1,5 @@
+# Dependencies for the impact collection pipeline (tools/impact).
+# Deliberately minimal: the collectors use urllib from the standard library,
+# so only schema validation and the optional classifier need a package.
+jsonschema>=4.0
+anthropic>=0.40  # optional at runtime; without a key the classifier is skipped
diff --git a/schemas/impact/adoption.schema.json b/schemas/impact/adoption.schema.json
new file mode 100644
index 0000000..bd46a62
--- /dev/null
+++ b/schemas/impact/adoption.schema.json
@@ -0,0 +1,66 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/adoption.schema.json",
+  "title": "DBMS project adoption of SQLancer",
+  "description": "Evidence that a database system's own developers use or integrate SQLancer. Support for a DBMS by SQLancer is NOT adoption and is tracked separately in dbms.json.",
+  "type": "object",
+  "required": ["schema_version", "policy_version", "adoption"],
+  "additionalProperties": false,
+  "properties": {
+    "schema_version": { "$ref": "common.schema.json#/definitions/schema_version" },
+    "policy_version": { "type": "string", "maxLength": 40 },
+    "adoption": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "required": ["id", "dbms", "relationship", "summary", "evidence", "provenance"],
+        "additionalProperties": false,
+        "properties": {
+          "id": { "$ref": "common.schema.json#/definitions/record_id" },
+          "dbms": {
+            "$ref": "common.schema.json#/definitions/slug",
+            "description": "Must reference an id in dbms.json."
+          },
+          "project_repository": { "$ref": "common.schema.json#/definitions/url" },
+          "relationship": {
+            "type": "string",
+            "description": "Controlled vocabulary for how the DBMS project uses SQLancer. 'planned_adoption' is the one entry that is not use: a project has proposed adopting SQLancer but has not been shown to have done so, and it is counted separately everywhere.",
+            "enum": [
+              "official_ci",
+              "official_testing",
+              "developer_use",
+              "integration_contributed_by_dbms_team",
+              "planned_adoption"
+            ]
+          },
+          "finder": {
+            "anyOf": [
+              { "$ref": "common.schema.json#/definitions/slug" },
+              { "type": "null" }
+            ],
+            "description": "Umbrella tool being adopted, when the evidence identifies a specific one."
+          },
+          "summary": {
+            "type": "string",
+            "minLength": 10,
+            "maxLength": 600,
+            "description": "Factual description of what the evidence shows. Not a substitute for the verbatim excerpt."
+          },
+          "since_year": {
+            "anyOf": [
+              { "$ref": "common.schema.json#/definitions/year" },
+              { "type": "null" }
+            ]
+          },
+          "active": {
+            "type": ["boolean", "null"],
+            "description": "Whether the integration was still present at last verification. Null when not determined."
+          },
+          "evidence": { "$ref": "common.schema.json#/definitions/evidence_list" },
+          "classifier": { "$ref": "common.schema.json#/definitions/classifier_record" },
+          "provenance": { "$ref": "common.schema.json#/definitions/provenance" }
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/bugs.schema.json b/schemas/impact/bugs.schema.json
new file mode 100644
index 0000000..342703b
--- /dev/null
+++ b/schemas/impact/bugs.schema.json
@@ -0,0 +1,204 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/bugs.schema.json",
+  "title": "Bugs attributed to SQLancer",
+  "description": "One record per bug report attributed to a tool of the SQLancer umbrella under the documented attribution policy. Every record carries the evidence that justifies the attribution.",
+  "type": "object",
+  "required": [
+    "schema_version",
+    "policy_version",
+    "bugs"
+  ],
+  "additionalProperties": false,
+  "properties": {
+    "schema_version": {
+      "$ref": "common.schema.json#/definitions/schema_version"
+    },
+    "policy_version": {
+      "type": "string",
+      "maxLength": 40
+    },
+    "bugs": {
+      "type": "array",
+      "items": {
+        "$ref": "#/definitions/bug"
+      }
+    }
+  },
+  "definitions": {
+    "bug": {
+      "type": "object",
+      "required": [
+        "id",
+        "dbms",
+        "title",
+        "status",
+        "finder",
+        "technique",
+        "symptom",
+        "attribution",
+        "provenance"
+      ],
+      "additionalProperties": false,
+      "properties": {
+        "id": {
+          "$ref": "common.schema.json#/definitions/record_id"
+        },
+        "dbms": {
+          "$ref": "common.schema.json#/definitions/slug",
+          "description": "Must reference an id in dbms.json."
+        },
+        "title": {
+          "type": "string",
+          "minLength": 1,
+          "maxLength": 500
+        },
+        "reported_date": {
+          "anyOf": [
+            {
+              "$ref": "common.schema.json#/definitions/date"
+            },
+            {
+              "type": "null"
+            }
+          ]
+        },
+        "reported_year": {
+          "anyOf": [
+            {
+              "$ref": "common.schema.json#/definitions/year"
+            },
+            {
+              "type": "null"
+            }
+          ]
+        },
+        "status": {
+          "type": "string",
+          "description": "Normalised upstream resolution state.",
+          "enum": [
+            "fixed",
+            "fixed_in_documentation",
+            "verified",
+            "open",
+            "closed_not_a_bug",
+            "closed_duplicate",
+            "unknown"
+          ]
+        },
+        "status_is_true_positive": {
+          "type": "boolean",
+          "description": "Derived: whether the report was accepted as a genuine bug. Records with false are retained for transparency but excluded from headline bug counts."
+        },
+        "finder": {
+          "$ref": "common.schema.json#/definitions/slug",
+          "description": "Umbrella tool credited with the find; must reference a finder id in techniques.json."
+        },
+        "technique": {
+          "anyOf": [
+            {
+              "$ref": "common.schema.json#/definitions/slug"
+            },
+            {
+              "type": "null"
+            }
+          ],
+          "description": "SQLancer-originated technique responsible, when known; must reference a technique id in techniques.json. Null when the specific technique is not recorded."
+        },
+        "symptom": {
+          "type": "string",
+          "enum": [
+            "logic",
+            "error",
+            "crash",
+            "hang",
+            "performance",
+            "unknown"
+          ]
+        },
+        "reporter": {
+          "type": [
+            "string",
+            "null"
+          ],
+          "maxLength": 120
+        },
+        "reporter_affiliation": {
+          "type": "string",
+          "description": "Whether the person who filed the report belongs to the SQLancer project -- its authors, the TEST lab, or its Google Summer of Code contributors -- or found the bug as an outside adopter. 'unknown' when no reporter is recorded.",
+          "enum": [
+            "project",
+            "external",
+            "unknown"
+          ]
+        },
+        "severity": {
+          "type": [
+            "string",
+            "null"
+          ],
+          "maxLength": 80
+        },
+        "cve": {
+          "type": [
+            "string",
+            "null"
+          ],
+          "maxLength": 40
+        },
+        "links": {
+          "type": "object",
+          "description": "Primary sources for the report. At least one link is required by the attribution policy.",
+          "additionalProperties": {
+            "$ref": "common.schema.json#/definitions/url"
+          },
+          "minProperties": 1
+        },
+        "primary_url": {
+          "$ref": "common.schema.json#/definitions/url",
+          "description": "Canonical location of the bug report; used for deduplication."
+        },
+        "attribution": {
+          "type": "object",
+          "required": [
+            "rule",
+            "confidence",
+            "evidence"
+          ],
+          "additionalProperties": false,
+          "properties": {
+            "rule": {
+              "type": "string",
+              "description": "Which clause of the bug attribution policy admitted this record.",
+              "enum": [
+                "explicit_tool_statement",
+                "technique_attribution",
+                "campaign_evidence",
+                "curated_primary_source",
+                "campaign_reporter"
+              ]
+            },
+            "confidence": {
+              "type": "string",
+              "enum": [
+                "high",
+                "medium",
+                "low"
+              ],
+              "description": "How directly the evidence establishes the attribution. 'low' marks a record admitted because of who reported it rather than anything the report itself says."
+            },
+            "evidence": {
+              "$ref": "common.schema.json#/definitions/evidence_list"
+            },
+            "classifier": {
+              "$ref": "common.schema.json#/definitions/classifier_record"
+            }
+          }
+        },
+        "provenance": {
+          "$ref": "common.schema.json#/definitions/provenance"
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/common.schema.json b/schemas/impact/common.schema.json
new file mode 100644
index 0000000..84db352
--- /dev/null
+++ b/schemas/impact/common.schema.json
@@ -0,0 +1,177 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/common.schema.json",
+  "title": "Shared definitions for SQLancer impact data",
+  "definitions": {
+    "schema_version": {
+      "type": "string",
+      "pattern": "^[0-9]+\\.[0-9]+\\.[0-9]+$",
+      "description": "Semantic version of the file format."
+    },
+    "record_id": {
+      "type": "string",
+      "pattern": "^[a-z0-9]+(:[A-Za-z0-9._+#/@-]+)+$",
+      "minLength": 3,
+      "maxLength": 200,
+      "description": "Stable, colon-namespaced unique identifier, e.g. 'bug:sqlite:0a1b2c3d4e5f'."
+    },
+    "slug": {
+      "type": "string",
+      "pattern": "^[a-z0-9][a-z0-9_]*$",
+      "maxLength": 64
+    },
+    "url": {
+      "type": "string",
+      "format": "uri",
+      "pattern": "^https?://[^\\s]+$",
+      "maxLength": 2000
+    },
+    "sha256": {
+      "type": "string",
+      "pattern": "^sha256:[0-9a-f]{64}$"
+    },
+    "date": {
+      "type": "string",
+      "pattern": "^[0-9]{4}-[0-9]{2}-[0-9]{2}$"
+    },
+    "timestamp": {
+      "type": "string",
+      "pattern": "^[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2}Z$"
+    },
+    "year": {
+      "type": "integer",
+      "minimum": 2019,
+      "maximum": 2100
+    },
+    "source_type": {
+      "type": "string",
+      "description": "Kind of primary source an evidence item was taken from.",
+      "enum": [
+        "github_issue",
+        "github_pull_request",
+        "github_repository",
+        "github_code",
+        "github_workflow",
+        "curated_bug_repository",
+        "mailing_list",
+        "forum_post",
+        "issue_tracker",
+        "commit",
+        "paper",
+        "paper_citation_context",
+        "artifact",
+        "documentation",
+        "website",
+        "blog_post",
+        "video",
+        "dataset",
+        "other"
+      ]
+    },
+    "evidence": {
+      "type": "object",
+      "description": "A single, auditable piece of primary evidence. Excerpts are copied verbatim from the source and are never paraphrased or generated.",
+      "required": ["source_url", "source_type", "first_seen", "last_verified"],
+      "additionalProperties": false,
+      "properties": {
+        "source_url": { "$ref": "#/definitions/url" },
+        "source_type": { "$ref": "#/definitions/source_type" },
+        "excerpt": {
+          "type": ["string", "null"],
+          "maxLength": 4000,
+          "description": "Verbatim text copied from the source. MUST be a literal substring of the fetched source content; never paraphrased, summarised or generated."
+        },
+        "excerpt_is_verbatim": {
+          "type": "boolean",
+          "description": "Always true when 'excerpt' is non-null. Recorded explicitly so that a violation of the no-fabrication rule is detectable by schema-independent tests."
+        },
+        "note": {
+          "type": ["string", "null"],
+          "maxLength": 500,
+          "description": "Curator-written description of *where* the excerpt comes from. Never a substitute for the excerpt itself."
+        },
+        "content_sha256": { "$ref": "#/definitions/sha256" },
+        "retrieved_at": { "$ref": "#/definitions/timestamp" },
+        "first_seen": { "$ref": "#/definitions/timestamp" },
+        "last_verified": { "$ref": "#/definitions/timestamp" }
+      },
+      "allOf": [
+        {
+          "if": { "properties": { "excerpt": { "type": "string" } }, "required": ["excerpt"] },
+          "then": {
+            "required": ["excerpt_is_verbatim"],
+            "properties": { "excerpt_is_verbatim": { "const": true } }
+          }
+        }
+      ]
+    },
+    "evidence_list": {
+      "type": "array",
+      "minItems": 1,
+      "items": { "$ref": "#/definitions/evidence" }
+    },
+    "classification_value": {
+      "type": "string",
+      "description": "Answer of a bounded classification question. Only 'yes' counts as a positive claim; uncertain candidates are never promoted.",
+      "enum": ["yes", "no", "uncertain", "insufficient_evidence"]
+    },
+    "determination_method": {
+      "type": "string",
+      "enum": [
+        "deterministic",
+        "citation_graph",
+        "artifact_inspection",
+        "llm_classification",
+        "manual_curation"
+      ]
+    },
+    "provenance": {
+      "type": "object",
+      "description": "Audit trail explaining how a record came to be accepted.",
+      "required": ["collector", "first_seen", "last_verified"],
+      "additionalProperties": false,
+      "properties": {
+        "collector": {
+          "type": "string",
+          "maxLength": 80,
+          "description": "Identifier of the collector that produced the record."
+        },
+        "collector_version": { "type": "string", "maxLength": 40 },
+        "policy_version": {
+          "type": "string",
+          "maxLength": 40,
+          "description": "Version of the attribution policy in force when the record was accepted."
+        },
+        "first_seen": { "$ref": "#/definitions/timestamp" },
+        "last_verified": { "$ref": "#/definitions/timestamp" },
+        "source_url": { "$ref": "#/definitions/url" },
+        "source_type": { "$ref": "#/definitions/source_type" },
+        "content_sha256": { "$ref": "#/definitions/sha256" }
+      }
+    },
+    "classifier_record": {
+      "type": "object",
+      "description": "Reproducibility record for a decision that depended on semantic classification.",
+      "required": ["method"],
+      "additionalProperties": false,
+      "properties": {
+        "method": { "$ref": "#/definitions/determination_method" },
+        "classifier_version": { "type": "string", "maxLength": 60 },
+        "policy_version": { "type": "string", "maxLength": 40 },
+        "taxonomy_version": { "type": "string", "maxLength": 40 },
+        "evidence_sha256": { "$ref": "#/definitions/sha256" },
+        "classified_at": { "$ref": "#/definitions/timestamp" },
+        "model": {
+          "type": ["string", "null"],
+          "maxLength": 80,
+          "description": "Model identifier, retained for auditability. Not shown on the public site."
+        },
+        "rationale": {
+          "type": ["string", "null"],
+          "maxLength": 1000,
+          "description": "Short model- or curator-written justification. Distinct from 'excerpt', which is always verbatim source text."
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/dbms.schema.json b/schemas/impact/dbms.schema.json
new file mode 100644
index 0000000..64e0488
--- /dev/null
+++ b/schemas/impact/dbms.schema.json
@@ -0,0 +1,114 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/dbms.schema.json",
+  "title": "Database system registry",
+  "description": "Canonical list of database systems referenced by impact records. 'Supported by SQLancer' is deliberately a separate concept from 'SQLancer found bugs in it' and from 'its developers use SQLancer'.",
+  "type": "object",
+  "required": [
+    "schema_version",
+    "dbms"
+  ],
+  "additionalProperties": false,
+  "properties": {
+    "schema_version": {
+      "$ref": "common.schema.json#/definitions/schema_version"
+    },
+    "dbms": {
+      "type": "array",
+      "minItems": 1,
+      "items": {
+        "type": "object",
+        "required": [
+          "id",
+          "name",
+          "aliases",
+          "supported_by_sqlancer"
+        ],
+        "additionalProperties": false,
+        "properties": {
+          "id": {
+            "$ref": "common.schema.json#/definitions/slug"
+          },
+          "name": {
+            "type": "string",
+            "minLength": 1,
+            "maxLength": 80
+          },
+          "aliases": {
+            "type": "array",
+            "items": {
+              "type": "string",
+              "minLength": 1,
+              "maxLength": 80
+            },
+            "description": "Spellings seen in upstream sources; used to normalise incoming records."
+          },
+          "url": {
+            "$ref": "common.schema.json#/definitions/url"
+          },
+          "repository": {
+            "$ref": "common.schema.json#/definitions/url"
+          },
+          "github_owners": {
+            "type": "array",
+            "items": {
+              "type": "string",
+              "minLength": 1,
+              "maxLength": 80
+            },
+            "description": "Additional GitHub organisations the project operates under, beyond the owner of its main repository. Used to recognise a SQLancer fork as the project's own."
+          },
+          "supported_by_sqlancer": {
+            "type": "boolean",
+            "description": "True when the main SQLancer repository ships a testing implementation for this DBMS."
+          },
+          "support": {
+            "type": [
+              "object",
+              "null"
+            ],
+            "required": [
+              "provider_path",
+              "evidence"
+            ],
+            "additionalProperties": false,
+            "properties": {
+              "provider_path": {
+                "type": "string",
+                "maxLength": 200,
+                "description": "Path of the DBMS provider package inside the SQLancer repository."
+              },
+              "umbrella_tool": {
+                "$ref": "common.schema.json#/definitions/slug",
+                "description": "Which umbrella tool provides the support (e.g. sqlancer, sqlancer_pp)."
+              },
+              "evidence": {
+                "$ref": "common.schema.json#/definitions/evidence_list"
+              }
+            }
+          },
+          "notes": {
+            "type": "string",
+            "maxLength": 500
+          },
+          "gitee_repositories": {
+            "type": "array",
+            "description": "Repositories on gitee.com whose issue tracker belongs to this project, as owner/name. Several Chinese database projects take bug reports there rather than on GitHub.",
+            "items": {
+              "type": "string",
+              "maxLength": 140
+            }
+          },
+          "github_repositories": {
+            "type": "array",
+            "description": "Further repositories the project owns whose issues are its own -- a fuzzer's bug tracker, for instance -- given as owner/name.",
+            "items": {
+              "type": "string",
+              "maxLength": 140
+            }
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/needs_review.schema.json b/schemas/impact/needs_review.schema.json
new file mode 100644
index 0000000..e7a0b4f
--- /dev/null
+++ b/schemas/impact/needs_review.schema.json
@@ -0,0 +1,48 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/needs_review.schema.json",
+  "title": "The queue of candidates no rule settled",
+  "description": "What a collector saw and could not place, kept between runs. This file is the record of what did not get in, which is what makes the dataset's completeness checkable rather than asserted: a reader can see the candidates that were considered and read the reason each one was turned down. Entries are written by the pipeline and by people, so the shape is checked like every other data file.",
+  "type": "object",
+  "required": ["schema_version", "description", "open", "dismissed"],
+  "properties": {
+    "schema_version": {"type": "string", "pattern": "^[0-9]+\\.[0-9]+\\.[0-9]+$"},
+    "description": {"type": "string", "minLength": 1},
+    "open": {
+      "type": "array",
+      "description": "Still waiting on somebody or something. An entry here is a claim that work remains, so it carries when it was first and last seen -- a candidate that stopped appearing is not silently dropped.",
+      "items": {
+        "type": "object",
+        "required": ["id", "kind", "source", "reason", "first_seen", "last_seen"],
+        "additionalProperties": false,
+        "properties": {
+          "id": {"type": "string", "pattern": "^review:[0-9a-f]+$"},
+          "kind": {"type": "string", "minLength": 1},
+          "source": {"type": "string", "minLength": 1},
+          "url": {"type": ["string", "null"]},
+          "title": {"type": ["string", "null"]},
+          "reason": {"type": "string", "minLength": 1,
+                     "description": "Why it could not be admitted, in terms a person can act on."},
+          "first_seen": {"type": "string", "minLength": 1},
+          "last_seen": {"type": "string", "minLength": 1}
+        }
+      }
+    },
+    "dismissed": {
+      "type": "array",
+      "description": "Settled, and never to be raised again. Every dismissal is attributable: a decision with no name on it cannot be argued with later.",
+      "items": {
+        "type": "object",
+        "required": ["id", "reason", "decided_by", "decided_on"],
+        "additionalProperties": false,
+        "properties": {
+          "id": {"type": "string", "pattern": "^review:[0-9a-f]+$"},
+          "url": {"type": ["string", "null"]},
+          "reason": {"type": "string", "minLength": 1},
+          "decided_by": {"type": "string", "minLength": 1},
+          "decided_on": {"type": "string", "pattern": "^[0-9]{4}-[0-9]{2}-[0-9]{2}$"}
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/paper-note.schema.json b/schemas/impact/paper-note.schema.json
new file mode 100644
index 0000000..60441d3
--- /dev/null
+++ b/schemas/impact/paper-note.schema.json
@@ -0,0 +1,216 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/paper-note.schema.json",
+  "title": "Note on a citing paper",
+  "description": "One paper's note: a model-written summary, clearly marked as such, and the individual sources its record rests on, quoted verbatim. The two are kept apart on purpose -- the summary is a convenience and may be wrong; the evidence is the record and may not.",
+  "type": "object",
+  "additionalProperties": false,
+  "required": [
+    "schema_version",
+    "paper",
+    "evidence"
+  ],
+  "properties": {
+    "schema_version": {
+      "type": "string"
+    },
+    "policy_version": {
+      "type": "string"
+    },
+    "paper": {
+      "type": "object",
+      "additionalProperties": false,
+      "required": [
+        "id",
+        "title"
+      ],
+      "properties": {
+        "id": {
+          "type": "string"
+        },
+        "title": {
+          "type": "string"
+        },
+        "authors": {
+          "type": "array",
+          "items": {
+            "type": "string"
+          }
+        },
+        "year": {
+          "type": [
+            "integer",
+            "null"
+          ]
+        },
+        "venue": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "doi": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "url": {
+          "type": [
+            "string",
+            "null"
+          ]
+        }
+      }
+    },
+    "summary": {
+      "type": [
+        "object",
+        "null"
+      ],
+      "additionalProperties": false,
+      "required": [
+        "text",
+        "is_model_generated",
+        "model",
+        "written_from"
+      ],
+      "properties": {
+        "text": {
+          "type": "string"
+        },
+        "relationship_to_sqlancer": {
+          "type": "string"
+        },
+        "is_model_generated": {
+          "const": true,
+          "description": "Always true. A summary is written by a model and is never evidence."
+        },
+        "grounded_in_source": {
+          "type": "boolean",
+          "description": "The model quoted passages that were found verbatim in the source it was given."
+        },
+        "written_from": {
+          "type": "string",
+          "description": "What the model was given: the full text, an abstract, or citation sentences."
+        },
+        "written_from_url": {
+          "type": "string"
+        },
+        "source_sha256": {
+          "type": "string"
+        },
+        "supporting_excerpts": {
+          "type": "array",
+          "items": {
+            "type": "string"
+          }
+        },
+        "model": {
+          "type": "string"
+        },
+        "classifier_version": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "generated_at": {
+          "type": "string"
+        },
+        "generated_by": {
+          "type": "string",
+          "description": "Which route produced the summary: the pipeline's batch classifier, or an interactive session."
+        }
+      }
+    },
+    "evidence": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "additionalProperties": false,
+        "required": [
+          "relationship",
+          "sources"
+        ],
+        "properties": {
+          "relationship": {
+            "type": "string"
+          },
+          "title": {
+            "type": "string"
+          },
+          "value": {
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "method": {
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "techniques": {
+            "type": "array",
+            "items": {
+              "type": "string"
+            }
+          },
+          "markers": {
+            "type": "array",
+            "items": {
+              "type": "string"
+            }
+          },
+          "sources": {
+            "type": "array",
+            "items": {
+              "type": "object",
+              "additionalProperties": false,
+              "required": [
+                "source_url",
+                "source_type"
+              ],
+              "properties": {
+                "source_url": {
+                  "type": "string",
+                  "format": "uri"
+                },
+                "source_type": {
+                  "type": [
+                    "string",
+                    "null"
+                  ]
+                },
+                "excerpt": {
+                  "type": [
+                    "string",
+                    "null"
+                  ],
+                  "description": "Copied from the source, never paraphrased."
+                },
+                "excerpt_is_verbatim": {
+                  "type": "boolean"
+                },
+                "note": {
+                  "type": [
+                    "string",
+                    "null"
+                  ]
+                },
+                "retrieved_at": {
+                  "type": [
+                    "string",
+                    "null"
+                  ]
+                }
+              }
+            }
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/papers.schema.json b/schemas/impact/papers.schema.json
new file mode 100644
index 0000000..77ce7f4
--- /dev/null
+++ b/schemas/impact/papers.schema.json
@@ -0,0 +1,293 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/papers.schema.json",
+  "title": "Research papers related to SQLancer",
+  "description": "One record per distinct scholarly work that cites a foundational SQLancer publication. Relationship categories overlap by design: a paper may simultaneously reuse SQLancer infrastructure, extend a SQLancer technique and compare against SQLancer.",
+  "type": "object",
+  "required": [
+    "schema_version",
+    "policy_version",
+    "papers"
+  ],
+  "additionalProperties": false,
+  "properties": {
+    "schema_version": {
+      "$ref": "common.schema.json#/definitions/schema_version"
+    },
+    "policy_version": {
+      "type": "string",
+      "maxLength": 40
+    },
+    "papers": {
+      "type": "array",
+      "items": {
+        "$ref": "#/definitions/paper"
+      }
+    }
+  },
+  "definitions": {
+    "relationship": {
+      "type": "object",
+      "required": [
+        "value",
+        "method"
+      ],
+      "additionalProperties": false,
+      "properties": {
+        "value": {
+          "$ref": "common.schema.json#/definitions/classification_value"
+        },
+        "method": {
+          "$ref": "common.schema.json#/definitions/determination_method"
+        },
+        "techniques": {
+          "type": "array",
+          "items": {
+            "$ref": "common.schema.json#/definitions/slug"
+          },
+          "description": "SQLancer technique ids involved; must reference techniques.json."
+        },
+        "evidence": {
+          "type": "array",
+          "items": {
+            "$ref": "common.schema.json#/definitions/evidence"
+          },
+          "description": "Required to be non-empty whenever value is 'yes'."
+        },
+        "classifier": {
+          "$ref": "common.schema.json#/definitions/classifier_record"
+        }
+      },
+      "allOf": [
+        {
+          "if": {
+            "properties": {
+              "value": {
+                "const": "yes"
+              }
+            },
+            "required": [
+              "value"
+            ]
+          },
+          "then": {
+            "required": [
+              "evidence"
+            ],
+            "properties": {
+              "evidence": {
+                "minItems": 1
+              }
+            }
+          }
+        }
+      ]
+    },
+    "artifact": {
+      "type": "object",
+      "required": [
+        "url",
+        "kind"
+      ],
+      "additionalProperties": false,
+      "properties": {
+        "url": {
+          "$ref": "common.schema.json#/definitions/url"
+        },
+        "kind": {
+          "type": "string",
+          "enum": [
+            "github",
+            "zenodo",
+            "figshare",
+            "gitlab",
+            "bitbucket",
+            "website",
+            "other"
+          ]
+        },
+        "inspected": {
+          "type": "boolean"
+        },
+        "inspected_at": {
+          "$ref": "common.schema.json#/definitions/timestamp"
+        },
+        "sqlancer_markers": {
+          "type": "array",
+          "description": "Concrete infrastructure-reuse signals observed in the artifact. Individually weak; combined they support a 'uses_infrastructure' classification.",
+          "items": {
+            "type": "string",
+            "enum": [
+              "fork_of_sqlancer_repository",
+              "sqlancer_package_structure",
+              "retained_sqlancer_source_files",
+              "sqlancer_copyright_or_license_notice",
+              "git_history_derived_from_sqlancer",
+              "readme_states_reuse",
+              "randomly_java_present",
+              "sqlancer_source_content_match",
+              "sqlancer_source_in_nested_artifact",
+              "renamed_sqlancer_package",
+              "sqlancer_build_file_reference",
+              "sqlancer_provider_directory"
+            ]
+          }
+        },
+        "marker_evidence": {
+          "type": "array",
+          "items": {
+            "$ref": "common.schema.json#/definitions/evidence"
+          }
+        }
+      }
+    },
+    "paper": {
+      "type": "object",
+      "required": [
+        "id",
+        "title",
+        "year",
+        "relationships",
+        "cites_seed_techniques",
+        "provenance"
+      ],
+      "additionalProperties": false,
+      "properties": {
+        "id": {
+          "$ref": "common.schema.json#/definitions/record_id",
+          "description": "Prefer 'paper:doi:<doi>'; fall back to 'paper:arxiv:<id>' or 'paper:s2:<paperId>'."
+        },
+        "title": {
+          "type": "string",
+          "minLength": 1,
+          "maxLength": 500
+        },
+        "authors": {
+          "type": "array",
+          "items": {
+            "type": "string",
+            "maxLength": 200
+          },
+          "maxItems": 60
+        },
+        "year": {
+          "$ref": "common.schema.json#/definitions/year"
+        },
+        "venue": {
+          "type": [
+            "string",
+            "null"
+          ],
+          "maxLength": 300
+        },
+        "doi": {
+          "type": [
+            "string",
+            "null"
+          ],
+          "pattern": "^10\\.[0-9]{4,9}/[^\\s]+$",
+          "maxLength": 200,
+          "description": "Preferred stable scholarly identifier."
+        },
+        "arxiv_id": {
+          "type": [
+            "string",
+            "null"
+          ],
+          "maxLength": 40
+        },
+        "s2_paper_id": {
+          "type": [
+            "string",
+            "null"
+          ],
+          "pattern": "^[0-9a-f]{40}$"
+        },
+        "url": {
+          "anyOf": [
+            {
+              "$ref": "common.schema.json#/definitions/url"
+            },
+            {
+              "type": "null"
+            }
+          ]
+        },
+        "open_access_pdf": {
+          "anyOf": [
+            {
+              "$ref": "common.schema.json#/definitions/url"
+            },
+            {
+              "type": "null"
+            }
+          ]
+        },
+        "cites_seed_techniques": {
+          "type": "array",
+          "minItems": 1,
+          "items": {
+            "$ref": "common.schema.json#/definitions/slug"
+          },
+          "description": "Which foundational SQLancer publications this paper cites, by the id of the technique the paper introduced, or of the tool where the seed is a tool's own paper."
+        },
+        "relationships": {
+          "type": "object",
+          "required": [
+            "references",
+            "uses_infrastructure",
+            "extends_technique",
+            "compares_with",
+            "describes_as_state_of_the_art"
+          ],
+          "additionalProperties": false,
+          "properties": {
+            "references": {
+              "$ref": "#/definitions/relationship"
+            },
+            "uses_infrastructure": {
+              "$ref": "#/definitions/relationship"
+            },
+            "extends_technique": {
+              "$ref": "#/definitions/relationship"
+            },
+            "compares_with": {
+              "$ref": "#/definitions/relationship"
+            },
+            "describes_as_state_of_the_art": {
+              "$ref": "#/definitions/relationship",
+              "description": "The paper describes SQLancer or one of its techniques as state of the art. Recognition rather than reuse, so it is reported separately from the categories that mean a paper built on SQLancer."
+            }
+          }
+        },
+        "artifacts": {
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/artifact"
+          }
+        },
+        "is_sqlancer_publication": {
+          "type": "boolean",
+          "description": "True for papers authored as part of the SQLancer project itself; excluded from 'external papers building on SQLancer' counts."
+        },
+        "provenance": {
+          "$ref": "common.schema.json#/definitions/provenance"
+        },
+        "abstract": {
+          "type": [
+            "string",
+            "null"
+          ],
+          "description": "The paper's abstract, as indexed. Source text for its note."
+        },
+        "also_indexed_as": {
+          "type": "array",
+          "items": {
+            "type": "string"
+          },
+          "description": "Ids this work was separately indexed under before being merged -- usually its own preprint. Kept so a reference to the old id can still be resolved."
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/people.schema.json b/schemas/impact/people.schema.json
new file mode 100644
index 0000000..a4d9205
--- /dev/null
+++ b/schemas/impact/people.schema.json
@@ -0,0 +1,86 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/people.schema.json",
+  "title": "SQLancer campaign reporters",
+  "description": "People who report bugs found with SQLancer. The roster decides whose issues are worth fetching -- author search is the only route to a report that never names SQLancer -- and grants no bug any attribution by itself.",
+  "type": "object",
+  "additionalProperties": false,
+  "required": [
+    "schema_version",
+    "people"
+  ],
+  "properties": {
+    "schema_version": {
+      "type": "string"
+    },
+    "policy_version": {
+      "type": "string"
+    },
+    "generated_at": {
+      "type": "string"
+    },
+    "people": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "additionalProperties": false,
+        "required": [
+          "id",
+          "name",
+          "evidence"
+        ],
+        "properties": {
+          "id": {
+            "type": "string",
+            "pattern": "^person:[a-z0-9][a-z0-9_-]*$"
+          },
+          "name": {
+            "type": "string",
+            "minLength": 1,
+            "description": "As the source records it, handle or display name."
+          },
+          "github": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "description": "GitHub login, or null for someone who reports on a non-GitHub tracker."
+          },
+          "handle_is_verified": {
+            "type": "boolean",
+            "description": "The handle was read off issues this person is credited with, not assumed from the recorded name."
+          },
+          "reports_on_lab_list": {
+            "type": "integer",
+            "minimum": 0
+          },
+          "role": {
+            "type": "string"
+          },
+          "active": {
+            "type": "boolean"
+          },
+          "entry_hash": {
+            "type": "string"
+          },
+          "first_seen": {
+            "type": "string"
+          },
+          "last_verified": {
+            "type": "string"
+          },
+          "evidence": {
+            "$ref": "common.schema.json#/definitions/evidence_list"
+          },
+          "also_recorded_as": {
+            "type": "array",
+            "items": {
+              "type": "string"
+            },
+            "description": "Other spellings the sources use for this person."
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/people_decisions.schema.json b/schemas/impact/people_decisions.schema.json
new file mode 100644
index 0000000..6f7857c
--- /dev/null
+++ b/schemas/impact/people_decisions.schema.json
@@ -0,0 +1,36 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/people_decisions.schema.json",
+  "title": "Rulings on reporter affiliation",
+  "description": "Human decisions the roster cannot make for itself. Appearing on the lab's people page is evidence of being in the lab, not of belonging to the SQLancer project, and only a person who knows the work can tell the two apart.",
+  "type": "object",
+  "additionalProperties": false,
+  "required": ["schema_version", "counts_as_external"],
+  "properties": {
+    "schema_version": {"type": "string"},
+    "description": {"type": "string"},
+    "counts_as_external": {
+      "type": "array",
+      "description": "Roster members whose reports count as found outside the project.",
+      "items": {
+        "type": "object",
+        "additionalProperties": false,
+        "required": ["github", "reason"],
+        "properties": {
+          "github": {
+            "type": "string",
+            "description": "GitHub login, matched case-insensitively against a bug's reporter."
+          },
+          "also_recorded_as": {
+            "type": "array",
+            "description": "Other names this person's reports are filed under, for trackers that record a display name.",
+            "items": {"type": "string"}
+          },
+          "reason": {"type": "string", "minLength": 20},
+          "decided_by": {"type": "string"},
+          "decided_on": {"$ref": "common.schema.json#/definitions/date"}
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/policy.schema.json b/schemas/impact/policy.schema.json
new file mode 100644
index 0000000..13f07e1
--- /dev/null
+++ b/schemas/impact/policy.schema.json
@@ -0,0 +1,46 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/policy.schema.json",
+  "title": "Impact attribution policy",
+  "description": "The documented policy that governs which records are accepted. It is machine-readable so that the classifiers, the tests and the public /impact/ page all read the same text; bumping policy_version invalidates cached classifications.",
+  "type": "object",
+  "required": ["schema_version", "policy_version", "sections"],
+  "additionalProperties": false,
+  "properties": {
+    "schema_version": { "$ref": "common.schema.json#/definitions/schema_version" },
+    "policy_version": { "type": "string", "maxLength": 40 },
+    "effective_date": { "$ref": "common.schema.json#/definitions/date" },
+    "sections": {
+      "type": "array",
+      "minItems": 1,
+      "items": {
+        "type": "object",
+        "required": ["id", "title", "body"],
+        "additionalProperties": false,
+        "properties": {
+          "id": { "$ref": "common.schema.json#/definitions/slug" },
+          "title": { "type": "string", "minLength": 1, "maxLength": 160 },
+          "body": {
+            "type": "array",
+            "minItems": 1,
+            "items": { "type": "string", "minLength": 1, "maxLength": 2000 },
+            "description": "Paragraphs of prose."
+          },
+          "rules": {
+            "type": "array",
+            "items": {
+              "type": "object",
+              "required": ["id", "text"],
+              "additionalProperties": false,
+              "properties": {
+                "id": { "type": "string", "maxLength": 80 },
+                "text": { "type": "string", "minLength": 5, "maxLength": 1200 },
+                "kind": { "type": "string", "enum": ["include", "exclude", "note"] }
+              }
+            }
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/recognition_highlights.schema.json b/schemas/impact/recognition_highlights.schema.json
new file mode 100644
index 0000000..20fadf7
--- /dev/null
+++ b/schemas/impact/recognition_highlights.schema.json
@@ -0,0 +1,34 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/recognition_highlights.schema.json",
+  "title": "Featured recognition quotes",
+  "description": "Which papers' recognition of SQLancer is quoted on the impact page. An editorial judgement about which claims stand on their own, not something derivable from the records, so it is written down with a reason.",
+  "type": "object",
+  "additionalProperties": false,
+  "required": ["schema_version", "highlights"],
+  "properties": {
+    "schema_version": {"type": "string"},
+    "description": {"type": "string"},
+    "highlights": {
+      "type": "array",
+      "minItems": 1,
+      "items": {
+        "type": "object",
+        "additionalProperties": false,
+        "required": ["paper_id", "mention_id", "why"],
+        "properties": {
+          "paper_id": {
+            "type": "string",
+            "description": "Must be a paper whose describes_as_state_of_the_art is yes."
+          },
+          "mention_id": {
+            "type": "string",
+            "pattern": "^M\\d+$",
+            "description": "Which of that relationship's quoted sentences to show. The sentence itself is never copied here: it is read from the paper's record, so a highlight cannot drift from its source."
+          },
+          "why": {"type": "string", "minLength": 20}
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/resources.schema.json b/schemas/impact/resources.schema.json
new file mode 100644
index 0000000..f84ed77
--- /dev/null
+++ b/schemas/impact/resources.schema.json
@@ -0,0 +1,121 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/resources.schema.json",
+  "title": "External resources related to SQLancer",
+  "description": "Curated, evidence-backed pointers to material about SQLancer by other people. Deliberately narrow twice over: a resource must be substantially about SQLancer rather than merely mention it, and the project's own material is excluded -- the site already links its own documentation, papers and blog, and repeating them here says nothing about reach.",
+  "type": "object",
+  "required": [
+    "schema_version",
+    "resources"
+  ],
+  "additionalProperties": false,
+  "properties": {
+    "schema_version": {
+      "$ref": "common.schema.json#/definitions/schema_version"
+    },
+    "resources": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "required": [
+          "id",
+          "title",
+          "type",
+          "url",
+          "description",
+          "evidence",
+          "provenance"
+        ],
+        "additionalProperties": false,
+        "properties": {
+          "id": {
+            "$ref": "common.schema.json#/definitions/record_id"
+          },
+          "title": {
+            "type": "string",
+            "minLength": 1,
+            "maxLength": 300
+          },
+          "type": {
+            "type": "string",
+            "enum": [
+              "talk",
+              "blog_post",
+              "documentation_note",
+              "dataset",
+              "artifact",
+              "tool",
+              "educational_material"
+            ]
+          },
+          "url": {
+            "$ref": "common.schema.json#/definitions/url"
+          },
+          "date": {
+            "anyOf": [
+              {
+                "$ref": "common.schema.json#/definitions/date"
+              },
+              {
+                "type": "null"
+              }
+            ]
+          },
+          "year": {
+            "anyOf": [
+              {
+                "$ref": "common.schema.json#/definitions/year"
+              },
+              {
+                "type": "null"
+              }
+            ]
+          },
+          "authors": {
+            "type": "array",
+            "items": {
+              "type": "string",
+              "maxLength": 200
+            },
+            "maxItems": 30
+          },
+          "publisher": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "maxLength": 200
+          },
+          "description": {
+            "type": "string",
+            "minLength": 10,
+            "maxLength": 600,
+            "description": "Short factual description of what the resource is."
+          },
+          "related_techniques": {
+            "type": "array",
+            "items": {
+              "$ref": "common.schema.json#/definitions/slug"
+            }
+          },
+          "related_dbms": {
+            "type": "array",
+            "items": {
+              "$ref": "common.schema.json#/definitions/slug"
+            }
+          },
+          "official": {
+            "type": "boolean",
+            "description": "True when maintained by the SQLancer project itself."
+          },
+          "evidence": {
+            "$ref": "common.schema.json#/definitions/evidence_list"
+          },
+          "provenance": {
+            "$ref": "common.schema.json#/definitions/provenance"
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/stats.schema.json b/schemas/impact/stats.schema.json
new file mode 100644
index 0000000..88d13a6
--- /dev/null
+++ b/schemas/impact/stats.schema.json
@@ -0,0 +1,507 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/stats.schema.json",
+  "title": "Derived impact statistics",
+  "description": "GENERATED FILE. Produced by tools/impact/stats.py from the authoritative records. Never edit by hand: `python -m tools.impact.run stats` regenerates it and CI fails if it differs from the records.",
+  "type": "object",
+  "required": [
+    "schema_version",
+    "generated_from",
+    "headline",
+    "bugs",
+    "papers",
+    "dbms",
+    "resources"
+  ],
+  "additionalProperties": false,
+  "properties": {
+    "schema_version": {
+      "$ref": "common.schema.json#/definitions/schema_version"
+    },
+    "generated_from": {
+      "type": "object",
+      "description": "Content hashes of the authoritative inputs, so staleness is detectable.",
+      "additionalProperties": {
+        "$ref": "common.schema.json#/definitions/sha256"
+      }
+    },
+    "headline": {
+      "type": "object",
+      "required": [
+        "bugs_total",
+        "bugs_total_rounded",
+        "bugs_found_externally",
+        "dbms_supported",
+        "dbms_with_bugs",
+        "dbms_projects_using_sqlancer",
+        "dbms_projects_planning_adoption",
+        "papers_building_on_sqlancer",
+        "papers_citing_sqlancer",
+        "papers_calling_sqlancer_state_of_the_art",
+        "papers_reusing_or_extending_sqlancer",
+        "papers_comparing_against_sqlancer"
+      ],
+      "additionalProperties": false,
+      "properties": {
+        "bugs_total": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "bugs_found_externally": {
+          "type": "integer",
+          "minimum": 0,
+          "description": "Bugs filed by someone outside the SQLancer project."
+        },
+        "bugs_total_rounded": {
+          "type": "string",
+          "maxLength": 20,
+          "description": "Conservative display form, rounded down to a round number (e.g. '1,000+')."
+        },
+        "dbms_supported": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "dbms_with_bugs": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "dbms_projects_using_sqlancer": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "dbms_projects_planning_adoption": {
+          "type": "integer",
+          "minimum": 0,
+          "description": "Projects that have proposed adopting SQLancer without yet being shown to use it. Never part of dbms_projects_using_sqlancer."
+        },
+        "papers_building_on_sqlancer": {
+          "type": "integer",
+          "minimum": 0,
+          "description": "Distinct external papers with at least one of uses_infrastructure / extends_technique / compares_with = yes. Deduplicated: a paper in several categories counts once."
+        },
+        "papers_citing_sqlancer": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "papers_calling_sqlancer_state_of_the_art": {
+          "type": "integer",
+          "minimum": 0,
+          "description": "Distinct external papers that describe SQLancer or one of its techniques as state of the art. Recognition rather than reuse, so it is not part of papers_building_on_sqlancer."
+        },
+        "papers_reusing_or_extending_sqlancer": {
+          "type": "integer",
+          "minimum": 0,
+          "description": "External papers whose implementation reuses the SQLancer codebase or whose authors extend one of its techniques. Counted together because a paper can do either without the other and a few do both."
+        },
+        "papers_comparing_against_sqlancer": {
+          "type": "integer",
+          "minimum": 0,
+          "description": "External papers that evaluate their own approach against SQLancer or one of its oracles as a baseline."
+        },
+        "dbms_using_sqlancer_without_counted_bugs": {
+          "type": "integer",
+          "minimum": 0,
+          "description": "Database systems with adoption evidence but no bug counted here -- the part of the undercount the dataset can measure about itself."
+        }
+      }
+    },
+    "bugs": {
+      "type": "object",
+      "required": [
+        "by_attribution_rule",
+        "by_dbms",
+        "by_dbms_and_affiliation",
+        "by_finder",
+        "by_reporter_affiliation",
+        "by_status",
+        "by_symptom",
+        "by_technique",
+        "by_year",
+        "total",
+        "total_including_rejected"
+      ],
+      "additionalProperties": false,
+      "properties": {
+        "total": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "total_including_rejected": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "status_known_count": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "by_dbms": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "by_year": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "by_status": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "by_technique": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "by_finder": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "by_symptom": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "by_reporter_affiliation": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "by_dbms_and_affiliation": {
+          "type": "array",
+          "description": "Bugs per database system, split by whether the project or an outside adopter found them.",
+          "items": {
+            "type": "object",
+            "required": [
+              "key",
+              "label",
+              "count",
+              "segments"
+            ],
+            "additionalProperties": false,
+            "properties": {
+              "key": {
+                "type": "string",
+                "maxLength": 80
+              },
+              "label": {
+                "type": "string",
+                "maxLength": 120
+              },
+              "count": {
+                "type": "integer",
+                "minimum": 0
+              },
+              "segments": {
+                "$ref": "#/definitions/labelled_counts"
+              }
+            }
+          }
+        },
+        "by_attribution_rule": {
+          "$ref": "#/definitions/labelled_counts",
+          "description": "How each bug earned its place, by policy clause. Published because the clauses differ sharply in strength."
+        },
+        "years": {
+          "type": "array",
+          "description": "Per-year breakdowns, for the page each year gets.",
+          "items": {
+            "type": "object",
+            "required": [
+              "year",
+              "bugs"
+            ],
+            "additionalProperties": false,
+            "properties": {
+              "year": {
+                "type": "integer"
+              },
+              "bugs": {
+                "type": "integer",
+                "minimum": 0
+              },
+              "bugs_rejected": {
+                "type": "integer",
+                "minimum": 0
+              },
+              "systems": {
+                "type": "integer",
+                "minimum": 0
+              },
+              "by_dbms": {
+                "$ref": "#/definitions/labelled_counts"
+              },
+              "by_status": {
+                "$ref": "#/definitions/labelled_counts"
+              },
+              "by_technique": {
+                "$ref": "#/definitions/labelled_counts"
+              },
+              "by_symptom": {
+                "$ref": "#/definitions/labelled_counts"
+              },
+              "by_attribution_rule": {
+                "$ref": "#/definitions/labelled_counts"
+              },
+              "by_affiliation": {
+                "$ref": "#/definitions/labelled_counts"
+              },
+              "top_reporters": {
+                "$ref": "#/definitions/labelled_counts"
+              },
+              "first_reported": {
+                "type": [
+                  "string",
+                  "null"
+                ]
+              },
+              "last_reported": {
+                "type": [
+                  "string",
+                  "null"
+                ]
+              }
+            }
+          }
+        }
+      }
+    },
+    "papers": {
+      "type": "object",
+      "required": [
+        "total",
+        "external_total",
+        "by_relationship",
+        "by_year",
+        "by_relationship_year",
+        "by_technique"
+      ],
+      "additionalProperties": false,
+      "properties": {
+        "total": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "external_total": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "building_on_total": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "by_relationship": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "by_year": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "by_relationship_year": {
+          "type": "object",
+          "description": "Yearly series per relationship category, used directly by the bar plots.",
+          "additionalProperties": {
+            "$ref": "#/definitions/labelled_counts"
+          }
+        },
+        "by_technique": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "years": {
+          "type": "array",
+          "items": {
+            "type": "integer"
+          }
+        }
+      }
+    },
+    "dbms": {
+      "type": "object",
+      "required": [
+        "supported",
+        "with_bugs",
+        "with_adoption",
+        "planning_adoption",
+        "adoption_by_relationship"
+      ],
+      "additionalProperties": false,
+      "properties": {
+        "supported": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "with_bugs": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "with_adoption": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "planning_adoption": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "adoption_by_relationship": {
+          "$ref": "#/definitions/labelled_counts"
+        },
+        "rows": {
+          "type": "array",
+          "description": "Per-DBMS join of support, bug counts and adoption, ready for tabular rendering. Systems with bugs also carry the breakdowns their own detail page renders.",
+          "items": {
+            "type": "object",
+            "required": [
+              "adoption_relationships",
+              "bugs",
+              "id",
+              "name",
+              "planned_adoption",
+              "supported"
+            ],
+            "additionalProperties": false,
+            "properties": {
+              "id": {
+                "$ref": "common.schema.json#/definitions/slug"
+              },
+              "name": {
+                "type": "string",
+                "maxLength": 80
+              },
+              "url": {
+                "type": [
+                  "string",
+                  "null"
+                ]
+              },
+              "supported": {
+                "type": "boolean"
+              },
+              "bugs": {
+                "type": "integer",
+                "minimum": 0
+              },
+              "adoption_relationships": {
+                "type": "array",
+                "items": {
+                  "type": "string"
+                }
+              },
+              "planned_adoption": {
+                "type": "boolean",
+                "description": "The project has proposed adopting SQLancer but has not been shown using it. Never counted as adoption."
+              },
+              "repository": {
+                "type": [
+                  "string",
+                  "null"
+                ]
+              },
+              "bugs_rejected": {
+                "type": "integer",
+                "minimum": 0,
+                "description": "Reports the developers rejected as invalid or duplicate. Kept for transparency and counted nowhere."
+              },
+              "bugs_by_year": {
+                "$ref": "#/definitions/labelled_counts",
+                "description": "Accepted bugs by the year they were reported."
+              },
+              "bugs_by_status": {
+                "$ref": "#/definitions/labelled_counts",
+                "description": "Accepted bugs by the state of the report."
+              },
+              "bugs_by_symptom": {
+                "$ref": "#/definitions/labelled_counts",
+                "description": "Accepted bugs by what went wrong."
+              },
+              "bugs_by_technique": {
+                "$ref": "#/definitions/labelled_counts",
+                "description": "Accepted bugs by the test oracle credited."
+              },
+              "bugs_by_attribution_rule": {
+                "$ref": "#/definitions/labelled_counts",
+                "description": "Accepted bugs by the policy clause that earned each its place."
+              },
+              "bugs_by_affiliation": {
+                "$ref": "#/definitions/labelled_counts",
+                "description": "Accepted bugs by whether the project or someone outside it found them."
+              },
+              "top_reporters": {
+                "$ref": "#/definitions/labelled_counts",
+                "description": "The people who filed the most of them."
+              },
+              "first_reported": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "format": "date",
+                "description": "Earliest accepted report."
+              },
+              "last_reported": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "format": "date",
+                "description": "Most recent accepted report."
+              }
+            }
+          }
+        },
+        "using_without_bugs": {
+          "type": "array",
+          "description": "Database systems with adoption evidence and no bug counted, named so the undercount can be shown rather than asserted.",
+          "items": {
+            "type": "object",
+            "required": [
+              "id",
+              "name"
+            ],
+            "additionalProperties": false,
+            "properties": {
+              "id": {
+                "type": "string"
+              },
+              "name": {
+                "type": "string"
+              }
+            }
+          }
+        }
+      }
+    },
+    "resources": {
+      "type": "object",
+      "required": [
+        "total",
+        "by_type"
+      ],
+      "additionalProperties": false,
+      "properties": {
+        "total": {
+          "type": "integer",
+          "minimum": 0
+        },
+        "by_type": {
+          "$ref": "#/definitions/labelled_counts"
+        }
+      }
+    }
+  },
+  "definitions": {
+    "labelled_counts": {
+      "type": "array",
+      "description": "Ordered count series. An array rather than an object so that Liquid can iterate it in a defined order.",
+      "items": {
+        "type": "object",
+        "required": [
+          "key",
+          "label",
+          "count"
+        ],
+        "additionalProperties": false,
+        "properties": {
+          "key": {
+            "type": "string",
+            "maxLength": 80
+          },
+          "label": {
+            "type": "string",
+            "maxLength": 120
+          },
+          "count": {
+            "type": "integer",
+            "minimum": 0
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/talks.schema.json b/schemas/impact/talks.schema.json
new file mode 100644
index 0000000..e7f4315
--- /dev/null
+++ b/schemas/impact/talks.schema.json
@@ -0,0 +1,380 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/talks.schema.json",
+  "title": "Talks that discuss SQLancer",
+  "description": "Recorded talks by other people in which SQLancer is discussed, and what each says about it. A talk is evidence of a kind no other source gives: someone standing in front of an audience explaining, recommending or reporting on the tool. What is said lives in the audio, so the record locates every mention by timestamp and quotes the transcript it was found in -- never the speaker directly, because a transcript is a machine's reading of speech and is wrong often enough to matter.",
+  "type": "object",
+  "required": [
+    "schema_version",
+    "talks"
+  ],
+  "additionalProperties": false,
+  "properties": {
+    "schema_version": {
+      "$ref": "common.schema.json#/definitions/schema_version"
+    },
+    "talks": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "required": [
+          "id",
+          "title",
+          "url",
+          "sources",
+          "mentions",
+          "provenance"
+        ],
+        "additionalProperties": false,
+        "properties": {
+          "id": {
+            "$ref": "common.schema.json#/definitions/record_id"
+          },
+          "title": {
+            "type": "string",
+            "minLength": 1,
+            "maxLength": 300
+          },
+          "url": {
+            "$ref": "common.schema.json#/definitions/url"
+          },
+          "video_id": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "maxLength": 40
+          },
+          "speakers": {
+            "type": "array",
+            "items": {
+              "type": "string",
+              "maxLength": 200
+            },
+            "maxItems": 20
+          },
+          "publisher": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "maxLength": 200
+          },
+          "event": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "maxLength": 200
+          },
+          "year": {
+            "anyOf": [
+              {
+                "$ref": "common.schema.json#/definitions/year"
+              },
+              {
+                "type": "null"
+              }
+            ]
+          },
+          "duration_seconds": {
+            "type": [
+              "integer",
+              "null"
+            ],
+            "minimum": 0
+          },
+          "related_dbms": {
+            "type": "array",
+            "items": {
+              "$ref": "common.schema.json#/definitions/slug"
+            }
+          },
+          "related_techniques": {
+            "type": "array",
+            "items": {
+              "$ref": "common.schema.json#/definitions/slug"
+            }
+          },
+          "mentions": {
+            "type": "array",
+            "description": "Every place in the talk where SQLancer, one of its techniques, or its author comes up.",
+            "items": {
+              "type": "object",
+              "required": [
+                "id",
+                "source",
+                "url",
+                "matched"
+              ],
+              "additionalProperties": false,
+              "properties": {
+                "id": {
+                  "type": "string",
+                  "pattern": "^M[0-9]+$"
+                },
+                "at_seconds": {
+                  "type": [
+                    "integer",
+                    "null"
+                  ],
+                  "minimum": 0,
+                  "description": "When it was said. Absent for a mention read off the slides, which carry no clock."
+                },
+                "timestamp": {
+                  "type": [
+                    "string",
+                    "null"
+                  ],
+                  "maxLength": 12,
+                  "description": "The moment, as a person reads it. Absent for a mention with no clock behind it."
+                },
+                "url": {
+                  "$ref": "common.schema.json#/definitions/url",
+                  "description": "Deep link to the moment, so a reader can check the claim by listening."
+                },
+                "matched": {
+                  "type": "array",
+                  "minItems": 1,
+                  "items": {
+                    "type": "string",
+                    "maxLength": 60
+                  },
+                  "description": "What was found: a name, a technique id, or 'author'."
+                },
+                "heard_as": {
+                  "type": [
+                    "string",
+                    "null"
+                  ],
+                  "maxLength": 80,
+                  "description": "How the transcriber rendered the name, when it got it wrong -- 'SQL lenser' for SQLancer."
+                },
+                "excerpt": {
+                  "type": [
+                    "string",
+                    "null"
+                  ],
+                  "minLength": 1,
+                  "maxLength": 2000,
+                  "description": "Verbatim with respect to the source named above -- for captions that is the caption track, not the speaker. Null when the only source is a person's account of watching the talk."
+                },
+                "excerpt_is_verbatim": {
+                  "type": "boolean",
+                  "description": "Verbatim with respect to the source named above, which for automatic captions is not the same as verbatim with respect to the speaker."
+                },
+                "technique_ids": {
+                  "type": "array",
+                  "items": {
+                    "$ref": "common.schema.json#/definitions/slug"
+                  }
+                },
+                "source": {
+                  "type": "string",
+                  "enum": [
+                    "slides",
+                    "frame",
+                    "captions",
+                    "watched"
+                  ],
+                  "description": "Which of the talk's sources this mention was read in. Decides how far the excerpt can be trusted."
+                },
+                "note": {
+                  "type": [
+                    "string",
+                    "null"
+                  ],
+                  "maxLength": 500,
+                  "description": "What a viewer reported, where there is nothing quotable."
+                },
+                "image": {
+                  "type": [
+                    "string",
+                    "null"
+                  ],
+                  "maxLength": 300,
+                  "description": "Site path of a frame captured from the recording at this moment. What it shows is described in the note; the picture is the evidence, and nothing is transcribed off it."
+                },
+                "frame_checked": {
+                  "type": [
+                    "string",
+                    "null"
+                  ],
+                  "maxLength": 300,
+                  "description": "Why this moment has no frame, when somebody looked and there was nothing on screen worth capturing. Keeps it off the worklist."
+                }
+              },
+              "allOf": [
+                {
+                  "description": "Every mention says something: it quotes a source, or shows one, or reports what a viewer heard.",
+                  "anyOf": [
+                    {
+                      "required": [
+                        "excerpt"
+                      ],
+                      "properties": {
+                        "excerpt": {
+                          "type": "string"
+                        }
+                      }
+                    },
+                    {
+                      "required": [
+                        "note"
+                      ],
+                      "properties": {
+                        "note": {
+                          "type": "string"
+                        }
+                      }
+                    },
+                    {
+                      "required": [
+                        "image"
+                      ],
+                      "properties": {
+                        "image": {
+                          "type": "string"
+                        }
+                      }
+                    }
+                  ]
+                }
+              ]
+            }
+          },
+          "relationship": {
+            "type": [
+              "object",
+              "null"
+            ],
+            "required": [
+              "roles",
+              "summary",
+              "mention_ids"
+            ],
+            "additionalProperties": false,
+            "description": "What the talk does with SQLancer, decided from the mentions and citing them.",
+            "properties": {
+              "roles": {
+                "type": "array",
+                "minItems": 1,
+                "items": {
+                  "type": "string",
+                  "enum": [
+                    "explains_technique",
+                    "cites_as_example",
+                    "recognition",
+                    "reports_adoption",
+                    "compares",
+                    "background"
+                  ]
+                }
+              },
+              "summary": {
+                "type": "string",
+                "minLength": 10,
+                "maxLength": 800
+              },
+              "mention_ids": {
+                "type": "array",
+                "minItems": 1,
+                "items": {
+                  "type": "string",
+                  "pattern": "^M[0-9]+$"
+                }
+              },
+              "decided_by": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "maxLength": 100
+              }
+            }
+          },
+          "provenance": {
+            "$ref": "common.schema.json#/definitions/provenance"
+          },
+          "sources": {
+            "type": "array",
+            "minItems": 1,
+            "description": "Where the words in this record came from. A talk can be read four ways, and they are not equally trustworthy: the speaker's own slides are authored text and can be quoted; a frame captured from the recording shows a slide that was never spoken, which no transcript can reach; automatic captions are a machine's reading of speech, good enough to find a moment and not good enough to put words in someone's mouth; and a person who watched it can report what was said but not quote it.",
+            "items": {
+              "type": "object",
+              "required": [
+                "kind",
+                "status"
+              ],
+              "additionalProperties": false,
+              "properties": {
+                "kind": {
+                  "type": "string",
+                  "enum": [
+                    "slides",
+                    "frame",
+                    "captions",
+                    "watched"
+                  ]
+                },
+                "status": {
+                  "type": "string",
+                  "enum": [
+                    "extracted",
+                    "unavailable",
+                    "not_attempted"
+                  ],
+                  "description": "'unavailable' means the platform published no such source, or would not serve the one it has."
+                },
+                "url": {
+                  "anyOf": [
+                    {
+                      "$ref": "common.schema.json#/definitions/url"
+                    },
+                    {
+                      "type": "null"
+                    }
+                  ]
+                },
+                "retrieved_at": {
+                  "anyOf": [
+                    {
+                      "$ref": "common.schema.json#/definitions/timestamp"
+                    },
+                    {
+                      "type": "null"
+                    }
+                  ]
+                },
+                "segments": {
+                  "type": [
+                    "integer",
+                    "null"
+                  ],
+                  "minimum": 0
+                },
+                "characters": {
+                  "type": [
+                    "integer",
+                    "null"
+                  ],
+                  "minimum": 0
+                },
+                "content_sha256": {
+                  "$ref": "common.schema.json#/definitions/sha256"
+                },
+                "note": {
+                  "type": [
+                    "string",
+                    "null"
+                  ],
+                  "maxLength": 500
+                }
+              }
+            }
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/impact/techniques.schema.json b/schemas/impact/techniques.schema.json
new file mode 100644
index 0000000..6e9e6ec
--- /dev/null
+++ b/schemas/impact/techniques.schema.json
@@ -0,0 +1,432 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/impact/techniques.schema.json",
+  "title": "SQLancer technique and finder taxonomy",
+  "description": "Single source of truth for which testing techniques originated in SQLancer, which tools belong to the SQLancer umbrella, and which similarly-scoped techniques are explicitly out of scope. Collectors and classifiers read this file instead of hardcoding names.",
+  "type": "object",
+  "required": [
+    "schema_version",
+    "taxonomy_version",
+    "finders",
+    "sqlancer_techniques",
+    "excluded_techniques"
+  ],
+  "additionalProperties": false,
+  "properties": {
+    "schema_version": {
+      "$ref": "common.schema.json#/definitions/schema_version"
+    },
+    "taxonomy_version": {
+      "type": "string",
+      "maxLength": 40,
+      "description": "Bumping this invalidates cached LLM classifications."
+    },
+    "finders": {
+      "type": "array",
+      "minItems": 1,
+      "items": {
+        "type": "object",
+        "required": [
+          "id",
+          "display_name",
+          "umbrella_member",
+          "names"
+        ],
+        "additionalProperties": false,
+        "properties": {
+          "id": {
+            "$ref": "common.schema.json#/definitions/slug"
+          },
+          "display_name": {
+            "type": "string",
+            "maxLength": 80
+          },
+          "names": {
+            "type": "array",
+            "minItems": 1,
+            "items": {
+              "type": "string",
+              "minLength": 2,
+              "maxLength": 80
+            },
+            "description": "Search aliases used for discovery."
+          },
+          "umbrella_member": {
+            "type": "boolean",
+            "description": "True when results attributed to this tool count towards overall SQLancer impact statistics."
+          },
+          "repository": {
+            "$ref": "common.schema.json#/definitions/url"
+          },
+          "paper": {
+            "type": [
+              "object",
+              "null"
+            ],
+            "required": [
+              "title"
+            ],
+            "additionalProperties": false,
+            "properties": {
+              "title": {
+                "type": "string",
+                "maxLength": 300
+              },
+              "venue": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "maxLength": 120
+              },
+              "year": {
+                "type": [
+                  "integer",
+                  "null"
+                ]
+              },
+              "url": {
+                "$ref": "common.schema.json#/definitions/url"
+              },
+              "doi": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "maxLength": 200
+              },
+              "arxiv_id": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "maxLength": 40
+              },
+              "s2_paper_id": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "pattern": "^[0-9a-f]{40}$"
+              },
+              "is_citation_seed": {
+                "type": "boolean",
+                "description": "True when citations of this paper are crawled to discover SQLancer-citing literature. A tool's own paper can be a seed as an oracle's is: SQLancer++ is cited by work that cites no oracle paper."
+              }
+            }
+          },
+          "description": {
+            "type": "string",
+            "maxLength": 600
+          }
+        }
+      }
+    },
+    "sqlancer_techniques": {
+      "type": "array",
+      "minItems": 1,
+      "items": {
+        "type": "object",
+        "required": [
+          "id",
+          "names",
+          "display_name",
+          "kind",
+          "sqlancer_originated"
+        ],
+        "additionalProperties": false,
+        "properties": {
+          "id": {
+            "$ref": "common.schema.json#/definitions/slug"
+          },
+          "names": {
+            "type": "array",
+            "minItems": 1,
+            "items": {
+              "type": "string",
+              "minLength": 2,
+              "maxLength": 120
+            },
+            "description": "Aliases used both for discovery queries and for attribution matching."
+          },
+          "display_name": {
+            "type": "string",
+            "maxLength": 120
+          },
+          "kind": {
+            "type": "string",
+            "enum": [
+              "test_oracle",
+              "test_input_generation",
+              "implicit_oracle",
+              "reduction"
+            ]
+          },
+          "sqlancer_originated": {
+            "type": "boolean",
+            "description": "True when the technique was introduced or first prototyped as part of SQLancer."
+          },
+          "introduced_year": {
+            "type": [
+              "integer",
+              "null"
+            ]
+          },
+          "origin_paper": {
+            "type": [
+              "object",
+              "null"
+            ],
+            "required": [
+              "title"
+            ],
+            "additionalProperties": false,
+            "properties": {
+              "title": {
+                "type": "string",
+                "maxLength": 300
+              },
+              "venue": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "maxLength": 120
+              },
+              "year": {
+                "type": [
+                  "integer",
+                  "null"
+                ]
+              },
+              "url": {
+                "$ref": "common.schema.json#/definitions/url"
+              },
+              "doi": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "maxLength": 200
+              },
+              "arxiv_id": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "maxLength": 40
+              },
+              "s2_paper_id": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "pattern": "^[0-9a-f]{40}$"
+              },
+              "is_citation_seed": {
+                "type": "boolean",
+                "description": "True when citations of this paper are crawled to discover SQLancer-citing literature."
+              },
+              "also_titled": {
+                "type": "array",
+                "description": "Other titles this paper is cited under. A preprint and its published version often differ, and a bibliography prints whichever the citing authors used.",
+                "items": {
+                  "type": "string",
+                  "maxLength": 300
+                }
+              }
+            }
+          },
+          "ambiguous_acronym": {
+            "type": "boolean",
+            "description": "True for short names such as TLP or PQS that collide with unrelated terms. Matches for these require corroborating database-testing context."
+          },
+          "required_context_terms": {
+            "type": "array",
+            "items": {
+              "type": "string",
+              "minLength": 2,
+              "maxLength": 60
+            },
+            "description": "At least one of these must co-occur before an ambiguous acronym match is treated as relevant."
+          },
+          "negative_context_terms": {
+            "type": "array",
+            "items": {
+              "type": "string",
+              "minLength": 2,
+              "maxLength": 60
+            },
+            "description": "Presence of these strongly suggests an unrelated use of the acronym."
+          },
+          "description": {
+            "type": "string",
+            "maxLength": 800
+          }
+        }
+      }
+    },
+    "project_authors": {
+      "type": "array",
+      "description": "Author-name fragments that identify a paper as the SQLancer project's own work. Matched as whole words against each author name, so a paper co-authored by one of these is not counted as external work building on SQLancer.",
+      "items": {
+        "type": "string",
+        "minLength": 3,
+        "maxLength": 80
+      }
+    },
+    "project_contributors": {
+      "type": "array",
+      "description": "People who found bugs as part of the SQLancer project rather than as outside adopters: its Google Summer of Code contributors and anyone else the project counts as its own. Members of the TEST lab are read from its people page at collection time and need no entry here.",
+      "items": {
+        "type": "object",
+        "required": [
+          "name"
+        ],
+        "additionalProperties": false,
+        "properties": {
+          "name": {
+            "type": "string",
+            "minLength": 2,
+            "maxLength": 120
+          },
+          "github": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "maxLength": 80
+          },
+          "role": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "maxLength": 120
+          }
+        }
+      }
+    },
+    "project_publications": {
+      "type": "array",
+      "description": "Publications authored as part of the SQLancer project that are not already named as a technique's origin paper or a tool's paper. Listed explicitly because whether a paper is the project's own is a fact its authors know, not something to infer from an author list.",
+      "items": {
+        "type": "object",
+        "required": [
+          "title",
+          "reason"
+        ],
+        "additionalProperties": false,
+        "properties": {
+          "title": {
+            "type": "string",
+            "minLength": 3,
+            "maxLength": 300
+          },
+          "doi": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "maxLength": 200
+          },
+          "arxiv_id": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "maxLength": 40
+          },
+          "s2_paper_id": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "pattern": "^[0-9a-f]{40}$"
+          },
+          "year": {
+            "type": [
+              "integer",
+              "null"
+            ]
+          },
+          "venue": {
+            "type": [
+              "string",
+              "null"
+            ],
+            "maxLength": 200
+          },
+          "reason": {
+            "type": "string",
+            "minLength": 5,
+            "maxLength": 500
+          }
+        }
+      }
+    },
+    "excluded_techniques": {
+      "type": "array",
+      "description": "Database-testing techniques that are deliberately NOT treated as SQLancer-originated, even though an implementation may live in the SQLancer repository. Bugs found with these do not count towards SQLancer bug statistics.",
+      "items": {
+        "type": "object",
+        "required": [
+          "id",
+          "names",
+          "reason"
+        ],
+        "additionalProperties": false,
+        "properties": {
+          "id": {
+            "$ref": "common.schema.json#/definitions/slug"
+          },
+          "names": {
+            "type": "array",
+            "minItems": 1,
+            "items": {
+              "type": "string",
+              "maxLength": 120
+            }
+          },
+          "display_name": {
+            "type": "string",
+            "maxLength": 120
+          },
+          "reason": {
+            "type": "string",
+            "minLength": 10,
+            "maxLength": 800
+          },
+          "evidence": {
+            "$ref": "common.schema.json#/definitions/evidence_list"
+          }
+        }
+      }
+    },
+    "bug_symptoms": {
+      "type": "array",
+      "description": "Controlled vocabulary describing how a bug manifested, orthogonal to the technique that found it.",
+      "items": {
+        "type": "object",
+        "required": [
+          "id",
+          "display_name"
+        ],
+        "additionalProperties": false,
+        "properties": {
+          "id": {
+            "$ref": "common.schema.json#/definitions/slug"
+          },
+          "display_name": {
+            "type": "string",
+            "maxLength": 80
+          },
+          "description": {
+            "type": "string",
+            "maxLength": 400
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/schemas/papers/paper-analysis.schema.json b/schemas/papers/paper-analysis.schema.json
new file mode 100644
index 0000000..999abb8
--- /dev/null
+++ b/schemas/papers/paper-analysis.schema.json
@@ -0,0 +1,485 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://sqlancer.github.io/schemas/papers/paper-analysis.schema.json",
+  "title": "One paper, read once and recorded in full",
+  "description": "Everything the pipeline learned about a paper's relationship to SQLancer: what was read, every place the paper mentions it, and the classification that follows, with the sentences each claim rests on. The model cites mention ids and never supplies a quotation -- the sentences stored here are copied from the paper, which is what makes a claim checkable rather than asserted.",
+  "type": "object",
+  "required": [
+    "schema_version",
+    "generated_at",
+    "paper",
+    "sources",
+    "document",
+    "references",
+    "sqlancer_references",
+    "mentions",
+    "artifact",
+    "checks",
+    "analysis",
+    "provenance"
+  ],
+  "properties": {
+    "schema_version": {
+      "type": "string",
+      "pattern": "^[0-9]+\\.[0-9]+\\.[0-9]+$"
+    },
+    "generated_at": {
+      "type": "string"
+    },
+    "paper": {
+      "type": "object",
+      "required": [
+        "id",
+        "title"
+      ],
+      "properties": {
+        "id": {
+          "type": "string",
+          "minLength": 1
+        },
+        "title": {
+          "type": "string",
+          "minLength": 1
+        },
+        "authors": {
+          "type": "array",
+          "items": {
+            "type": "string"
+          }
+        },
+        "year": {
+          "type": [
+            "integer",
+            "null"
+          ]
+        },
+        "venue": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "doi": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "arxiv_id": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "s2_paper_id": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "url": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "also_indexed_as": {
+          "type": "array"
+        }
+      }
+    },
+    "sources": {
+      "type": "array",
+      "description": "What was read to build this record, and where it came from.",
+      "items": {
+        "type": "object",
+        "required": [
+          "kind",
+          "url"
+        ],
+        "properties": {
+          "kind": {
+            "type": "string"
+          },
+          "route": {
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "url": {
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "retrieved_at": {
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "chars": {
+            "type": [
+              "integer",
+              "null"
+            ]
+          },
+          "content_sha256": {
+            "type": [
+              "string",
+              "null"
+            ]
+          }
+        }
+      }
+    },
+    "document": {
+      "type": "object",
+      "properties": {
+        "has_fulltext": {
+          "type": "boolean"
+        },
+        "has_outline": {
+          "type": "boolean"
+        },
+        "page_count": {
+          "type": [
+            "integer",
+            "null"
+          ]
+        },
+        "sections": {
+          "type": "array"
+        }
+      }
+    },
+    "references": {
+      "type": "array"
+    },
+    "sqlancer_references": {
+      "type": "array"
+    },
+    "mentions": {
+      "type": "array",
+      "description": "Every place the paper reaches SQLancer -- by name, by technique, or through a citation marker that resolves to one of its publications, which is the only way to find the sentences that name nothing at all.",
+      "items": {
+        "type": "object",
+        "required": [
+          "id",
+          "found_by",
+          "sentence"
+        ],
+        "properties": {
+          "id": {
+            "type": "string",
+            "pattern": "^M[0-9]+$"
+          },
+          "found_by": {
+            "type": "string",
+            "enum": [
+              "name",
+              "technique",
+              "citation_marker",
+              "citation_marker_project_authored",
+              "author_year_citation"
+            ]
+          },
+          "found_by_all": {
+            "type": "array",
+            "items": {
+              "type": "string"
+            }
+          },
+          "surface": {
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "technique": {
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "techniques": {
+            "type": "array"
+          },
+          "sentence": {
+            "type": "string",
+            "minLength": 1
+          },
+          "context_before": {
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "context_after": {
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "section": {
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "page": {
+            "type": [
+              "integer",
+              "null"
+            ]
+          },
+          "char_offset": {
+            "type": [
+              "integer",
+              "null"
+            ]
+          },
+          "cited_reference": {
+            "type": [
+              "object",
+              "integer",
+              "string",
+              "null"
+            ]
+          },
+          "text_is_letter_spaced": {
+            "type": "boolean",
+            "description": "The PDF set this passage with letter-spacing, so extraction returns it one character at a time. The mention is real; the text is not quotable, and pages say so instead of printing it."
+          }
+        }
+      }
+    },
+    "artifact": {
+      "type": [
+        "object",
+        "null"
+      ]
+    },
+    "checks": {
+      "type": "object",
+      "description": "What the regexes suggested, kept as advice the classification may disagree with.",
+      "properties": {
+        "checks_version": {
+          "type": "string"
+        },
+        "suggests": {
+          "type": [
+            "object",
+            "array",
+            "null"
+          ]
+        },
+        "suppressed": {
+          "type": [
+            "object",
+            "array",
+            "null"
+          ]
+        }
+      }
+    },
+    "analysis": {
+      "type": "object",
+      "required": [
+        "relationships"
+      ],
+      "properties": {
+        "produced_by": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "model": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "analysis_version": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "prompt_version": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "generated_at": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "is_model_written": {
+          "type": [
+            "boolean",
+            "null"
+          ]
+        },
+        "summary": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "narrative": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "roles": {
+          "type": [
+            "object",
+            "null"
+          ]
+        },
+        "disagreements": {
+          "type": [
+            "array",
+            "null"
+          ]
+        },
+        "unresolved": {
+          "type": [
+            "array",
+            "null"
+          ]
+        },
+        "relationships": {
+          "type": "object",
+          "required": [
+            "uses_infrastructure",
+            "extends_technique",
+            "compares_with",
+            "describes_as_state_of_the_art"
+          ],
+          "additionalProperties": {
+            "type": "object",
+            "required": [
+              "value",
+              "mention_ids",
+              "quotes"
+            ],
+            "properties": {
+              "value": {
+                "type": "string",
+                "enum": [
+                  "yes",
+                  "no",
+                  "uncertain",
+                  "insufficient_evidence"
+                ]
+              },
+              "mention_ids": {
+                "type": "array",
+                "description": "Mentions the claim rests on. The sentinel ARTIFACT stands where no sentence carries it and the repository does -- a paper whose text never claims reuse while its artifact is SQLancer's source under a renamed package.",
+                "items": {
+                  "type": "string",
+                  "pattern": "^(M[0-9]+|ARTIFACT)$"
+                }
+              },
+              "quotes": {
+                "type": "array",
+                "items": {
+                  "type": "object",
+                  "required": [
+                    "mention_id",
+                    "sentence"
+                  ],
+                  "properties": {
+                    "mention_id": {
+                      "type": "string",
+                      "pattern": "^(M[0-9]+|ARTIFACT)$"
+                    },
+                    "sentence": {
+                      "type": "string",
+                      "minLength": 1
+                    },
+                    "section": {
+                      "type": [
+                        "string",
+                        "null"
+                      ]
+                    },
+                    "page": {
+                      "type": [
+                        "integer",
+                        "null"
+                      ]
+                    },
+                    "text_is_letter_spaced": {
+                      "type": "boolean"
+                    }
+                  }
+                }
+              },
+              "reasoning": {
+                "type": [
+                  "string",
+                  "null"
+                ]
+              },
+              "techniques": {
+                "type": "array"
+              },
+              "reuse_kind": {
+                "type": [
+                  "string",
+                  "null"
+                ],
+                "enum": [
+                  "implementation",
+                  "generator",
+                  "workload",
+                  "unclear",
+                  null
+                ],
+                "description": "How the paper reuses SQLancer. Workload use is reuse, but the paper is not built on it, and the narrative has to say which."
+              }
+            }
+          }
+        }
+      }
+    },
+    "provenance": {
+      "type": "object",
+      "properties": {
+        "extractor_version": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "inventory_version": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "checks_version": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "policy_version": {
+          "type": [
+            "string",
+            "null"
+          ]
+        }
+      }
+    }
+  }
+}
diff --git a/tools/__init__.py b/tools/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/tools/impact/README.md b/tools/impact/README.md
new file mode 100644
index 0000000..28c4d9b
--- /dev/null
+++ b/tools/impact/README.md
@@ -0,0 +1,158 @@
+# The impact pipeline
+
+This directory collects the evidence behind everything shown on
+[`/impact/`](../../_pages/impact.html). The authoritative records live in
+[`_data/impact/`](../../_data/impact/); everything here exists to propose changes
+to them and to derive the figures and charts the website renders.
+
+## The shape of it
+
+```
+deterministic collectors   -- GitHub, scholarly APIs, the curated bug repository
+        |
+        v
+candidate records          -- deduplicated on stable ids and canonical URLs
+        |
+        v
+cached evidence            -- fetched once, revalidated rather than re-fetched
+        |
+        v
+semantic classification    -- only where judgement is genuinely required
+        |
+        v
+schema validation          -- JSON Schema plus cross-file integrity checks
+        |
+        v
+proposed data changes      -- a pull request a human reviews
+```
+
+Discovery is deterministic throughout. A language model is used only to answer
+bounded questions about material the collectors already fetched, and it can
+neither browse nor introduce a record on its own. There is no agent loop.
+
+## Running it
+
+```sh
+make validate   # schemas + cross-file checks
+make build      # regenerate stats.json and the charts, then validate
+make test       # the test suite (no network, no API key)
+make collect    # incremental collection run
+make collect-full   # periodic full reconciliation
+```
+
+Or directly:
+
+```sh
+python3 -m tools.impact.run collect --only papers --dry-run
+python3 -m tools.impact.run report --input .cache/impact/last-run.json
+```
+
+`--dry-run` never calls the classifier; candidates that would need one are
+reported as needing human judgement instead.
+
+### Environment
+
+| Variable | Effect if unset |
+| --- | --- |
+| `GITHUB_TOKEN` | GitHub searches run unauthenticated and are throttled hard |
+| `SEMANTIC_SCHOLAR_API_KEY` | Semantic Scholar is used on the shared pool, with long backoff; OpenAlex still covers discovery |
+| `ANTHROPIC_API_KEY` | Ambiguous candidates are reported for human judgement rather than classified |
+| `IMPACT_OFFLINE=1` | No network at all; everything is served from the cache |
+| `IMPACT_CACHE_DIR` | Defaults to `.cache/impact` |
+
+Every one of these is optional. The pipeline degrades rather than fails: a
+missing key costs coverage that week, never correctness.
+
+## Modules
+
+| File | Responsibility |
+| --- | --- |
+| `config.py` | Paths, versions, deterministic JSON writing |
+| `util.py` | Hashing, stable ids, URL/DOI normalisation, the verbatim-excerpt check |
+| `cache.py` | The three cache layers and the incremental discovery state |
+| `http.py` | Conditional fetching, backoff, offline mode |
+| `github.py` | The GitHub REST endpoints the collectors need |
+| `scholarly.py` | Semantic Scholar and OpenAlex clients |
+| `taxonomy.py` | Reads `techniques.json`; technique matching, including acronym disambiguation |
+| `dataset.py` | Loading, merging and idempotent writing of the records |
+| `collectors/artifacts.py` | Finds a paper's artifact and inspects it for SQLancer markers; this is what decides `uses_infrastructure` |
+| `collectors/` | One module per source: `sqlancer_bugs` (the curated bug repository), `github_bugs` (issue search), `nus_test` (the TEST lab bug list), `papers`, `artifacts`, `adoption`, `resources`, `talks`, `dbms_registry` |
+| `talks.py` | Reads a talk four ways -- slides, a frame from the recording, captions, or a viewer's account -- and keeps them apart, because an automatic caption is not a quotation and a picture is not text |
+| `repos.py` | Where a project's issues live now: rewrites an address under a former organisation, so a bug filed at `cwida/duckdb` is the same record as one found at `duckdb/duckdb` |
+| `classify/` | The bounded questions and the cached, verified classifier |
+| `stats.py` | Derives `stats.json` from the records |
+| `plots.py` | Renders the inline SVG charts |
+| `validate.py` | JSON Schema plus the checks a schema cannot express |
+| `pr.py` | Renders the pull-request body |
+| `run.py` | Command-line entry point |
+| `seed/` | Scripts that bootstrapped the dataset |
+
+## The three caches
+
+Caching is an optimisation and never a correctness dependency; a test asserts
+that a cold cache produces the same records as a warm one.
+
+1. **Raw source** (`.cache/impact/{http,github,papers,artifacts}`) — fetched
+   bodies with the validators needed to tell whether a source changed: ETag,
+   Last-Modified, the upstream `updated_at`, and a content hash. An unchanged
+   GitHub thread costs no request at all, because its `updated_at` is enough.
+2. **Derived** (`.cache/impact/derived`) — deterministic extraction keyed by a
+   stable source id plus the hash of the source content and the extractor
+   version.
+3. **Classifications** (`.cache/impact/classifications`) — every answer the
+   classifier has given, including the negative and uncertain ones. The key
+   covers the source id, the hash of the exact evidence sent, the prompt
+   version, the policy version, the taxonomy version and the model, so a change
+   to any of those forces a fresh answer and nothing else does.
+
+`.cache/impact/state.json` records the last successful scan and the last full
+reconciliation per source, which is what makes weekly runs incremental.
+
+## What is enforced, not trusted
+
+- **Excerpts are verbatim.** Every stored excerpt is checked to be a literal
+  substring of the fetched source. An answer from the classifier whose excerpts
+  fail that check loses them, and a `yes` with no surviving excerpt is
+  downgraded to `uncertain`.
+- **Uncertain is never promoted.** Only `yes` produces a published claim.
+- **Ambiguous acronyms need context to be *accepted*, not to be *considered*.**
+  A bare `TLP` or `PQS` never establishes an attribution on its own. Inside a
+  registered database system's own repository it is still enough to make the
+  issue a candidate, because the repository is the context -- the deterministic
+  layer's job is to generate candidates for the classifier, not to decide in
+  its place.
+- **Excluded techniques stay excluded.** EET and DQE did not originate in
+  SQLancer -- both were published elsewhere and contributed afterwards -- so
+  bugs attributed to them are not counted and their publications are not
+  citation seeds, even though SQLancer ships an oracle for each.
+- **Manual curation is sticky.** A relationship a person decided is never
+  overwritten by a later automated run.
+- **A list is not an attribution.** The TEST lab publishes the lab's bugs, not
+  SQLancer's. Every candidate from it is taken back to its own bug report and
+  put through the same rules as any other; being on the list counts for
+  nothing.
+- **Derived output cannot go stale.** The tests fail if `stats.json` or any
+  chart differs from what the records produce.
+
+## Marking a paper as the project's own
+
+Papers by the project are excluded from counts of external work building on
+SQLancer. Three things mark one, all configured in `seed/techniques_seed.py`:
+
+- `PROJECT_AUTHORS` — a paper co-authored by any of these is ours. Names are
+  matched as whole words against each author, so both `Manuel Rigger` and
+  `M. Rigger` hit while a longer surname does not.
+- The technique origin papers and tool papers already in the taxonomy, matched
+  on DOI, arXiv id, Semantic Scholar id or exact title.
+- `PROJECT_PUBLICATIONS` — anything else the project states is its own, with a
+  reason per entry. Use this for a project paper none of the authors above is
+  on, or to record a specific decision durably.
+
+## Adding a technique or a database system
+
+Both come from data, not from code. Edit `seed/techniques_seed.py` and run it to
+regenerate `_data/impact/techniques.json`; bump `taxonomy_version` if the change
+affects how anything should be classified, which invalidates the affected cached
+answers. Database systems are discovered from the SQLancer repository's provider
+directories, so a newly supported system appears on the next run; presentation
+metadata for it lives in `collectors/dbms_registry.py`.
diff --git a/tools/impact/__init__.py b/tools/impact/__init__.py
new file mode 100644
index 0000000..a7d9b67
--- /dev/null
+++ b/tools/impact/__init__.py
@@ -0,0 +1,8 @@
+"""Collection pipeline for the SQLancer impact dataset.
+
+The authoritative, human-reviewable records live in ``_data/impact``. Everything
+in this package exists to propose changes to those records from primary sources,
+and to derive the statistics and plots the website renders.
+"""
+
+__version__ = "1.0.0"
diff --git a/tools/impact/cache.py b/tools/impact/cache.py
new file mode 100644
index 0000000..ba6c98b
--- /dev/null
+++ b/tools/impact/cache.py
@@ -0,0 +1,251 @@
+"""The three cache layers the pipeline relies on.
+
+Caching here is strictly an optimisation. Every collector must produce the same
+authoritative records with a cold cache as with a warm one; the tests assert
+that. What the cache buys is that a weekly run does not refetch unchanged pages
+and, more importantly, does not spend tokens reclassifying unchanged evidence.
+
+Layers:
+
+1. ``RawCache``      - fetched source material plus the validators (ETag,
+                       Last-Modified, upstream ``updated_at``, content hash)
+                       needed to decide whether it changed.
+2. ``DerivedCache``  - deterministic extraction results, keyed by a stable
+                       source id plus the hash of the source content.
+3. ``ClassificationCache`` - answers from the semantic classifiers, keyed so
+                       that a change to the evidence, the policy, the taxonomy,
+                       the prompt version or the model forces a re-ask.
+
+Entries are plain JSON files under ``.cache/impact`` so they diff readably and
+can be committed where they are useful as reproducible collector state.
+"""
+
+from __future__ import annotations
+
+import json
+import os
+import time
+from pathlib import Path
+from typing import Any, Callable, Dict, Optional
+
+from . import config
+from .util import now, sha256_hex
+
+
+def _key_path(root: Path, key: str, suffix: str = ".json") -> Path:
+    """Shard entries into 256 buckets so directories stay listable."""
+    digest = sha256_hex(key)
+    return root / digest[:2] / (digest + suffix)
+
+
+class JsonStore:
+    """Content-addressed JSON store with a human-readable key recorded inside."""
+
+    def __init__(self, root: Path):
+        self.root = root
+        self.hits = 0
+        self.misses = 0
+        self.writes = 0
+
+    def get(self, key: str) -> Optional[Dict[str, Any]]:
+        path = _key_path(self.root, key)
+        if not path.exists():
+            self.misses += 1
+            return None
+        try:
+            with open(path, "r", encoding="utf-8") as handle:
+                entry = json.load(handle)
+        except (OSError, ValueError):
+            self.misses += 1
+            return None
+        if entry.get("key") != key:  # digest collision or stale layout
+            self.misses += 1
+            return None
+        self.hits += 1
+        return entry
+
+    def put(self, key: str, value: Dict[str, Any]) -> None:
+        path = _key_path(self.root, key)
+        path.parent.mkdir(parents=True, exist_ok=True)
+        entry = {"key": key, "stored_at": now(), "value": value}
+        text = json.dumps(entry, indent=2, ensure_ascii=False, sort_keys=True) + "\n"
+        if path.exists():
+            with open(path, "r", encoding="utf-8") as handle:
+                if handle.read() == text:
+                    return
+        tmp = path.with_suffix(".tmp")
+        with open(tmp, "w", encoding="utf-8") as handle:
+            handle.write(text)
+        os.replace(tmp, path)
+        self.writes += 1
+
+    def stats(self) -> Dict[str, int]:
+        return {"hits": self.hits, "misses": self.misses, "writes": self.writes}
+
+
+class RawCache(JsonStore):
+    """Layer 1: fetched source material and its change validators."""
+
+    def store(self, key: str, *, body: str, url: str, status: int = 200,
+              etag: Optional[str] = None, last_modified: Optional[str] = None,
+              upstream_updated_at: Optional[str] = None,
+              extra: Optional[Dict[str, Any]] = None) -> Dict[str, Any]:
+        value = {
+            "url": url,
+            "status": status,
+            "body": body,
+            "content_sha256": "sha256:" + sha256_hex(body),
+            "etag": etag,
+            "last_modified": last_modified,
+            "upstream_updated_at": upstream_updated_at,
+            "fetched_at": now(),
+        }
+        if extra:
+            value.update(extra)
+        self.put(key, value)
+        return value
+
+    def is_fresh(self, key: str, *, upstream_updated_at: Optional[str] = None,
+                 max_age_days: Optional[int] = None) -> bool:
+        """Whether the cached copy can be reused without contacting the source.
+
+        An upstream modification timestamp, when the API offers one, is the
+        cheapest possible check: no request at all is needed if it is unchanged.
+        """
+        entry = self.get(key)
+        if entry is None:
+            return False
+        value = entry["value"]
+        if upstream_updated_at is not None:
+            return value.get("upstream_updated_at") == upstream_updated_at
+        if max_age_days is None:
+            return True
+        fetched = value.get("fetched_at")
+        if not fetched:
+            return False
+        age = time.time() - time.mktime(time.strptime(fetched, "%Y-%m-%dT%H:%M:%SZ"))
+        return age < max_age_days * 86400
+
+
+class DerivedCache(JsonStore):
+    """Layer 2: deterministic extraction keyed by source id + content hash."""
+
+    def compute(self, source_id: str, source_hash: str, extractor_version: str,
+                producer: Callable[[], Any]) -> Any:
+        key = f"{source_id}|{source_hash}|{extractor_version}"
+        entry = self.get(key)
+        if entry is not None:
+            return entry["value"]["result"]
+        result = producer()
+        self.put(key, {"source_id": source_id, "source_hash": source_hash,
+                       "extractor_version": extractor_version, "result": result})
+        return result
+
+
+class ClassificationCache(JsonStore):
+    """Layer 3: semantic classification answers, including negatives.
+
+    Negative and uncertain answers are cached exactly like positive ones. A
+    paper that turned out to only ``reference`` SQLancer, or an issue that
+    turned out to be unrelated, must not consume tokens again next week.
+    """
+
+    def __init__(self, root: Path, *, model: str, policy_version: str,
+                 taxonomy_version: str):
+        super().__init__(root)
+        self.model = model
+        self.policy_version = policy_version
+        self.taxonomy_version = taxonomy_version
+
+    def key(self, source_id: str, evidence_hash: str, classifier_version: str) -> str:
+        return "|".join([
+            source_id,
+            evidence_hash,
+            classifier_version,
+            self.policy_version,
+            self.taxonomy_version,
+            self.model,
+        ])
+
+    def lookup(self, source_id: str, evidence_hash: str,
+               classifier_version: str) -> Optional[Dict[str, Any]]:
+        entry = self.get(self.key(source_id, evidence_hash, classifier_version))
+        return entry["value"] if entry else None
+
+    def store(self, source_id: str, evidence_hash: str, classifier_version: str,
+              result: Dict[str, Any]) -> Dict[str, Any]:
+        value = {
+            "source_id": source_id,
+            "source_hash": evidence_hash,
+            "classifier_version": classifier_version,
+            "policy_version": self.policy_version,
+            "taxonomy_version": self.taxonomy_version,
+            "model": self.model,
+            "classified_at": now(),
+            "result": result,
+        }
+        self.put(self.key(source_id, evidence_hash, classifier_version), value)
+        return value
+
+
+class Caches:
+    """Convenience bundle wiring all layers to the configured cache root."""
+
+    def __init__(self, root: Optional[Path] = None, *, model: Optional[str] = None,
+                 policy_version: Optional[str] = None,
+                 taxonomy_version: Optional[str] = None):
+        root = Path(root) if root else config.CACHE_DIR
+        self.root = root
+        self.http = RawCache(root / "http")
+        self.github = RawCache(root / "github")
+        self.papers = RawCache(root / "papers")
+        self.artifacts = RawCache(root / "artifacts")
+        self.derived = DerivedCache(root / "derived")
+        self.classifications = ClassificationCache(
+            root / "classifications",
+            model=model or config.DEFAULT_MODEL,
+            policy_version=policy_version or config.policy_version(),
+            taxonomy_version=taxonomy_version or config.taxonomy_version(),
+        )
+
+    def summary(self) -> Dict[str, Dict[str, int]]:
+        return {
+            "http": self.http.stats(),
+            "github": self.github.stats(),
+            "papers": self.papers.stats(),
+            "artifacts": self.artifacts.stats(),
+            "derived": self.derived.stats(),
+            "classifications": self.classifications.stats(),
+        }
+
+
+class State:
+    """Persistent per-source discovery state driving incremental scans."""
+
+    def __init__(self, path: Optional[Path] = None):
+        self.path = Path(path) if path else config.STATE_FILE
+        self.data: Dict[str, Any] = {}
+        if self.path.exists():
+            try:
+                with open(self.path, "r", encoding="utf-8") as handle:
+                    self.data = json.load(handle)
+            except (OSError, ValueError):
+                self.data = {}
+
+    def last_scan(self, source: str) -> Optional[str]:
+        return self.data.get(source, {}).get("last_successful_scan")
+
+    def last_reconcile(self, source: str) -> Optional[str]:
+        return self.data.get(source, {}).get("last_full_reconciliation")
+
+    def record_scan(self, source: str, *, full: bool = False) -> None:
+        entry = self.data.setdefault(source, {})
+        entry["last_successful_scan"] = now()
+        if full:
+            entry["last_full_reconciliation"] = now()
+
+    def save(self) -> None:
+        self.path.parent.mkdir(parents=True, exist_ok=True)
+        text = json.dumps(self.data, indent=2, ensure_ascii=False, sort_keys=True) + "\n"
+        with open(self.path, "w", encoding="utf-8") as handle:
+            handle.write(text)
diff --git a/tools/impact/classify/__init__.py b/tools/impact/classify/__init__.py
new file mode 100644
index 0000000..75e5325
--- /dev/null
+++ b/tools/impact/classify/__init__.py
@@ -0,0 +1,261 @@
+"""The semantic classifier: a cached, verified wrapper around one API call.
+
+Three properties matter more than anything else here, and each is enforced
+rather than trusted:
+
+1. **Unchanged evidence is never reclassified.** The cache key covers the source
+   id, the hash of the exact text sent, the prompt version, the policy version,
+   the taxonomy version and the model. Negative and uncertain answers are cached
+   just like positive ones, so a paper that turned out to only cite SQLancer
+   costs nothing on subsequent runs.
+2. **The model cannot invent evidence.** Every excerpt it returns is checked
+   against the source text it was given. An excerpt that is not literally there
+   is dropped, and a "yes" that loses all its excerpts is downgraded to
+   ``uncertain`` -- a positive claim without evidence is not published.
+3. **Absence of an API key is not a failure mode.** A cached answer is returned
+   whatever the key situation: the cache is consulted before availability is,
+   so a question answered once stays answered offline. Only a question with no
+   cached answer returns ``None``, and the caller then records the candidate as
+   needing review. The pipeline still runs, and the dataset grows by whatever
+   the cache can already answer.
+"""
+
+from __future__ import annotations
+
+import json
+import os
+from typing import Any, Dict, List, Optional
+
+from .. import config, taxonomy
+from ..cache import ClassificationCache
+from ..util import content_hash, now, truncate, verbatim_excerpt
+from . import prompts
+
+# Guard against sending an entire paper or repository to the model. Sources are
+# assembled from targeted excerpts upstream, so this is a backstop.
+MAX_SOURCE_CHARS = 60_000
+MAX_EXCERPT_CHARS = 2_000
+
+
+class ClassificationResult:
+    __slots__ = ("answer", "excerpts", "reason", "extra", "from_cache",
+                 "classifier_version", "evidence_hash", "model")
+
+    def __init__(self, answer: str, excerpts: List[str], reason: str,
+                 extra: Dict[str, Any], *, from_cache: bool,
+                 classifier_version: str, evidence_hash: str, model: str):
+        self.answer = answer
+        self.excerpts = excerpts
+        self.reason = reason
+        self.extra = extra
+        self.from_cache = from_cache
+        self.classifier_version = classifier_version
+        self.evidence_hash = evidence_hash
+        self.model = model
+
+    @property
+    def is_positive(self) -> bool:
+        return self.answer == "yes"
+
+    def classifier_record(self, *, method: str = "llm_classification") -> dict:
+        """The audit trail stored alongside an accepted claim."""
+        return {
+            "method": method,
+            "classifier_version": self.classifier_version,
+            "policy_version": config.policy_version(),
+            "taxonomy_version": config.taxonomy_version(),
+            "evidence_sha256": self.evidence_hash,
+            "classified_at": now(),
+            "model": self.model,
+            "rationale": truncate(self.reason, 1000) if self.reason else None,
+        }
+
+    def __repr__(self):
+        return (f"ClassificationResult({self.answer!r}, "
+                f"{len(self.excerpts)} excerpt(s), "
+                f"cached={self.from_cache})")
+
+
+class Classifier:
+    """Answers the bounded questions in :mod:`prompts`, with caching."""
+
+    def __init__(self, cache: ClassificationCache, *, model: Optional[str] = None,
+                 api_key: Optional[str] = None, dry_run: bool = False):
+        self.cache = cache
+        self.model = model or config.DEFAULT_MODEL
+        self.api_key = api_key or os.environ.get("ANTHROPIC_API_KEY")
+        self.dry_run = dry_run
+        self._client = None
+        self.calls_made = 0
+        self.cache_hits = 0
+        self.skipped_no_key = 0
+
+    @property
+    def available(self) -> bool:
+        """Whether a live classification is possible at all."""
+        if self.dry_run or not self.api_key:
+            return False
+        if self._client is not None:
+            return True
+        try:
+            import anthropic  # noqa: F401
+        except ImportError:
+            return False
+        return True
+
+    def _get_client(self):
+        if self._client is None:
+            import anthropic
+            self._client = anthropic.Anthropic(api_key=self.api_key)
+        return self._client
+
+    # -- public API -------------------------------------------------------
+
+    def classify(self, question: str, *, source_id: str, source_text: str,
+                 **kwargs) -> Optional[ClassificationResult]:
+        """Answer ``question`` about ``source_text``.
+
+        Returns ``None`` only when no answer could be obtained -- no API key, no
+        SDK, or an API failure. ``None`` is not a negative answer: the caller
+        must treat it as "not yet classified" and surface the candidate for
+        human review.
+        """
+        builder = prompts.BUILDERS.get(question)
+        if builder is None:
+            raise KeyError(f"unknown classifier question {question!r}")
+
+        source_text = truncate(source_text or "", MAX_SOURCE_CHARS)
+        version = prompts.VERSIONS[question]
+        evidence_hash = content_hash(f"{question}\n{source_text}")
+
+        cached = self.cache.lookup(source_id, evidence_hash, version)
+        if cached is not None:
+            self.cache_hits += 1
+            payload = cached["result"]
+            return ClassificationResult(
+                payload["answer"], payload.get("excerpts", []),
+                payload.get("reason", ""), payload.get("extra", {}),
+                from_cache=True, classifier_version=version,
+                evidence_hash=evidence_hash, model=cached.get("model", self.model))
+
+        if not self.available:
+            self.skipped_no_key += 1
+            return None
+
+        request = builder(source_text=source_text, **kwargs)
+        raw = self._ask(request)
+        if raw is None:
+            return None
+
+        payload = self._sanitise(raw, source_text)
+        self.cache.store(source_id, evidence_hash, version, payload)
+        self.calls_made += 1
+        return ClassificationResult(
+            payload["answer"], payload["excerpts"], payload["reason"],
+            payload["extra"], from_cache=False, classifier_version=version,
+            evidence_hash=evidence_hash, model=self.model)
+
+    # -- internals --------------------------------------------------------
+
+    def _ask(self, request: dict) -> Optional[dict]:
+        try:
+            response = self._get_client().messages.create(
+                model=self.model,
+                max_tokens=2000,
+                system=request["system"],
+                messages=[{"role": "user", "content": request["user"]}],
+                output_config={"format": request["output_format"]},
+            )
+        except Exception:
+            # A failed call is not a negative answer; leave the candidate
+            # unclassified so the next run retries it rather than recording
+            # a decision that was never made.
+            return None
+        if getattr(response, "stop_reason", None) == "refusal":
+            return None
+        for block in response.content:
+            if getattr(block, "type", None) == "text":
+                try:
+                    return json.loads(block.text)
+                except ValueError:
+                    return None
+        return None
+
+    def _sanitise(self, raw: dict, source_text: str) -> dict:
+        """Enforce the no-fabrication rule on the model's output.
+
+        Excerpts that are not literally present in the source are discarded. A
+        "yes" whose excerpts all fail that check is downgraded to "uncertain",
+        because a positive claim without quotable evidence is exactly what this
+        dataset must not contain.
+        """
+        answer = raw.get("answer")
+        if answer not in prompts.ANSWER_ENUM:
+            answer = "uncertain"
+
+        verified: List[str] = []
+        for excerpt in raw.get("excerpts") or []:
+            if not isinstance(excerpt, str):
+                continue
+            excerpt = excerpt.strip()
+            if not excerpt or len(excerpt) > MAX_EXCERPT_CHARS:
+                continue
+            if verbatim_excerpt(source_text, excerpt) and excerpt not in verified:
+                verified.append(excerpt)
+
+        if answer == "yes" and not verified:
+            answer = "uncertain"
+
+        extra = {key: value for key, value in raw.items()
+                 if key not in ("answer", "excerpts", "reason")}
+
+        return {
+            "answer": answer,
+            "excerpts": verified,
+            "reason": truncate(str(raw.get("reason") or ""), 1000),
+            "extra": extra,
+            "unverified_excerpt_count": len(raw.get("excerpts") or []) - len(verified),
+        }
+
+    def stats(self) -> Dict[str, int]:
+        return {
+            "api_calls": self.calls_made,
+            "cache_hits": self.cache_hits,
+            "skipped_no_key": self.skipped_no_key,
+        }
+
+
+def technique_names(tax: Optional[taxonomy.Taxonomy] = None) -> List[str]:
+    tax = tax or taxonomy.load()
+    return [entry["display_name"] for entry in tax.techniques.values()]
+
+
+def finder_names(tax: Optional[taxonomy.Taxonomy] = None) -> List[str]:
+    tax = tax or taxonomy.load()
+    return [entry["display_name"] for entry in tax.finders.values()
+            if entry["umbrella_member"]]
+
+
+def excluded_names(tax: Optional[taxonomy.Taxonomy] = None) -> List[str]:
+    tax = tax or taxonomy.load()
+    return [entry.get("display_name") or entry["id"]
+            for entry in tax.excluded.values()]
+
+
+def resolve_technique(tax: taxonomy.Taxonomy, name: Optional[str]) -> Optional[str]:
+    """Map a technique name the model returned back onto a taxonomy id.
+
+    Anything that does not resolve is dropped rather than invented: the schema
+    only accepts ids that exist in ``techniques.json``.
+    """
+    if not name:
+        return None
+    matches = tax.find_techniques(name, require_context=False)
+    return matches[0].technique_id if matches else None
+
+
+def resolve_finder(tax: taxonomy.Taxonomy, name: Optional[str]) -> Optional[str]:
+    if not name:
+        return None
+    found = tax.mentions_finder(name)
+    return found[0] if found else None
diff --git a/tools/impact/classify/prompts.py b/tools/impact/classify/prompts.py
new file mode 100644
index 0000000..41e9bfd
--- /dev/null
+++ b/tools/impact/classify/prompts.py
@@ -0,0 +1,383 @@
+"""The bounded questions the classifier is allowed to ask.
+
+Each entry pairs one narrow question with a strict output schema. The design
+constraint throughout is that the model is a reader, not a researcher: it is
+handed source text that deterministic collection already fetched, and it may
+only answer about that text. It cannot browse, cannot introduce a candidate, and
+cannot supply an excerpt that is not literally present in what it was given --
+the caller re-checks every returned excerpt against the source and discards the
+answer if it does not match.
+
+``uncertain`` and ``insufficient_evidence`` are first-class answers. The prompts
+say so explicitly, because a classifier that feels obliged to choose yes or no
+is exactly how unfounded claims get into a dataset.
+"""
+
+from __future__ import annotations
+
+from typing import Dict, List, Optional
+
+# Bumping a version here invalidates the cached answers for that question only.
+VERSIONS = {
+    "bug_attribution": "bug-attribution-v1",
+    "paper_infrastructure": "paper-infrastructure-v1",
+    "paper_extends": "paper-extends-v1",
+    "paper_compares": "paper-compares-v1",
+    "adoption": "adoption-v1",
+    "paper_summary": "paper-summary-v1",
+}
+
+SHARED_RULES = """\
+Rules that apply to every answer:
+
+- Answer only from the SOURCE TEXT provided below. You have no other knowledge \
+of this artefact and must not assume anything the text does not state.
+- Every excerpt you return must be copied character-for-character from the \
+SOURCE TEXT. Do not paraphrase, summarise, translate, correct, or join \
+non-contiguous passages. If you cannot find a passage that supports an answer, \
+you do not have the evidence for that answer.
+- Prefer "uncertain" when the text points both ways, and \
+"insufficient_evidence" when the text simply does not address the question. \
+Neither is a failure; forcing a "yes" or "no" is.
+- Answer "yes" only when the SOURCE TEXT itself supports it."""
+
+
+def _schema(properties: Dict[str, dict], required: List[str]) -> dict:
+    return {
+        "type": "json_schema",
+        "schema": {
+            "type": "object",
+            "properties": properties,
+            "required": required,
+            "additionalProperties": False,
+        },
+    }
+
+
+ANSWER_ENUM = ["yes", "no", "uncertain", "insufficient_evidence"]
+
+_EXCERPTS = {
+    "type": "array",
+    "items": {"type": "string"},
+    "description": ("Passages copied verbatim from the SOURCE TEXT that support "
+                    "the answer. Empty unless the answer is 'yes'."),
+}
+
+_REASON = {
+    "type": "string",
+    "description": ("One or two sentences explaining the answer in your own "
+                    "words. This is your reasoning, not evidence, and is stored "
+                    "separately from the excerpts."),
+}
+
+
+def bug_attribution(*, source_label: str, source_text: str,
+                    technique_names: List[str], finder_names: List[str],
+                    excluded_names: List[str]) -> dict:
+    """Does this bug report attribute the find to SQLancer or one of its techniques?"""
+    system = f"""\
+You classify bug reports for the SQLancer project's public impact dataset.
+
+{SHARED_RULES}
+
+A bug counts as found by SQLancer when the report connects it to one of these \
+tools: {', '.join(finder_names)}; or to one of these testing techniques, which \
+originated in SQLancer: {', '.join(technique_names)}.
+
+Two traps to avoid:
+
+- These technique names are also ordinary acronyms in other fields. "TLP" is \
+thread-level parallelism and a Linux power-management tool; "PQS" and "CERT" \
+have unrelated meanings too. Answer "yes" only when the text is clearly about \
+testing a database system.
+- These techniques did NOT originate in SQLancer and do not count, even if a \
+SQLancer implementation of them exists: {', '.join(excluded_names) or 'none'}. \
+If the report credits one of those, the answer is "no"."""
+
+    user = f"""\
+Question: Does this bug report attribute the discovery to SQLancer, to one of \
+its umbrella tools, or to a testing technique that originated in SQLancer?
+
+If yes, name the specific tool and technique the report credits. Leave either \
+as null if the report does not say.
+
+SOURCE ({source_label}):
+<source_text>
+{source_text}
+</source_text>"""
+
+    schema = _schema({
+        "answer": {"type": "string", "enum": ANSWER_ENUM},
+        "finder": {"type": ["string", "null"],
+                   "description": "Tool credited, or null if unnamed."},
+        "technique": {"type": ["string", "null"],
+                      "description": "Technique credited, or null if unnamed."},
+        "excerpts": _EXCERPTS,
+        "reason": _REASON,
+    }, ["answer", "finder", "technique", "excerpts", "reason"])
+
+    return {"system": system, "user": user, "output_format": schema}
+
+
+def paper_uses_infrastructure(*, source_label: str, source_text: str,
+                              markers: List[str]) -> dict:
+    """Does this paper's implementation use or derive from the SQLancer codebase?"""
+    marker_note = (
+        "Deterministic inspection of the artefact found these signals: "
+        + ", ".join(markers) + ".\n"
+        if markers else
+        "Deterministic inspection of the artefact found no reuse signals.\n")
+
+    system = f"""\
+You classify research papers for the SQLancer project's public impact dataset.
+
+{SHARED_RULES}
+
+"Uses SQLancer infrastructure" means the paper's implementation reuses or \
+derives from the SQLancer codebase: a fork of the repository, retained SQLancer \
+source files or its Java package layout, a SQLancer copyright notice, git \
+history derived from it, or a README saying so.
+
+Judge the implementation, not the framing. Citing SQLancer, comparing against \
+it, or building on one of its ideas is NOT infrastructure reuse. Equally, no \
+single file name settles it: SQLancer's Randomly.java in particular turns up in \
+projects that copied one utility and nothing else, so weigh the signals together."""
+
+    user = f"""\
+Question: Does this paper's implementation use or derive from the SQLancer \
+codebase?
+
+{marker_note}
+SOURCE ({source_label}):
+<source_text>
+{source_text}
+</source_text>"""
+
+    schema = _schema({
+        "answer": {"type": "string", "enum": ANSWER_ENUM},
+        "excerpts": _EXCERPTS,
+        "reason": _REASON,
+    }, ["answer", "excerpts", "reason"])
+
+    return {"system": system, "user": user, "output_format": schema}
+
+
+def paper_extends_technique(*, source_label: str, source_text: str,
+                            technique_names: List[str]) -> dict:
+    """Does this paper extend, generalise or adapt a SQLancer technique?"""
+    system = f"""\
+You classify research papers for the SQLancer project's public impact dataset.
+
+{SHARED_RULES}
+
+Techniques that originated in SQLancer: {', '.join(technique_names)}.
+
+"Extends a SQLancer technique" means the paper claims a technical contribution \
+that extends, generalises, adapts, or substantially builds upon one of them --- \
+generalising TLP to a new setting, extending NoREC with a new transformation, \
+adapting PQS to another data model, or a new method whose central mechanism is \
+explicitly built on one of these techniques.
+
+Merely citing a technique as related work, describing it in a background \
+section, or using it unchanged as a baseline is NOT extending it. If the text \
+only shows the technique being described or compared against, answer "no"."""
+
+    user = f"""\
+Question: Does this paper claim a technical contribution that extends, \
+generalises, or adapts a technique introduced through SQLancer?
+
+If yes, name the technique or techniques and quote the passages that show the \
+extension.
+
+SOURCE ({source_label}):
+<source_text>
+{source_text}
+</source_text>"""
+
+    schema = _schema({
+        "answer": {"type": "string", "enum": ANSWER_ENUM},
+        "techniques": {
+            "type": "array",
+            "items": {"type": "string"},
+            "description": "SQLancer techniques the paper extends.",
+        },
+        "excerpts": _EXCERPTS,
+        "reason": _REASON,
+    }, ["answer", "techniques", "excerpts", "reason"])
+
+    return {"system": system, "user": user, "output_format": schema}
+
+
+def paper_compares_with(*, source_label: str, source_text: str,
+                        technique_names: List[str]) -> dict:
+    """Does this paper empirically compare against SQLancer or one of its oracles?"""
+    system = f"""\
+You classify research papers for the SQLancer project's public impact dataset.
+
+{SHARED_RULES}
+
+SQLancer techniques: {', '.join(technique_names)}.
+
+"Compares against SQLancer" means the paper runs an empirical comparison ---
+using SQLancer, a SQLancer implementation, or one of its test oracles as a \
+baseline, and reporting results against it.
+
+A citation in related work is not a comparison. A stated intention to compare, \
+with no results, is not a comparison. The text must show the comparison being \
+made or its outcome being reported."""
+
+    user = f"""\
+Question: Does this paper empirically evaluate its approach against SQLancer or \
+a SQLancer technique?
+
+If yes, name which ones it compares against.
+
+SOURCE ({source_label}):
+<source_text>
+{source_text}
+</source_text>"""
+
+    schema = _schema({
+        "answer": {"type": "string", "enum": ANSWER_ENUM},
+        "techniques": {
+            "type": "array",
+            "items": {"type": "string"},
+            "description": "SQLancer techniques used as a baseline.",
+        },
+        "excerpts": _EXCERPTS,
+        "reason": _REASON,
+    }, ["answer", "techniques", "excerpts", "reason"])
+
+    return {"system": system, "user": user, "output_format": schema}
+
+
+def adoption(*, source_label: str, source_text: str, dbms_name: str,
+             finder_names: List[str]) -> dict:
+    """Does this show the DBMS project's own developers using SQLancer?"""
+    system = f"""\
+You classify evidence for the SQLancer project's public impact dataset.
+
+{SHARED_RULES}
+
+The question is whether the {dbms_name} project's OWN developers use or \
+integrate one of these tools: {', '.join(finder_names)}.
+
+What counts: SQLancer running in the project's continuous integration; scripts \
+or configuration the project maintains for running it; the project's testing \
+documentation describing it; a developer of the project describing their use of \
+it; a SQLancer integration contributed or maintained by the project's team.
+
+What does not count, and this is the distinction that matters most: SQLancer \
+supporting {dbms_name} is not adoption. Nor is an outside researcher testing \
+{dbms_name} with SQLancer, or a bug report filed by someone who is not part of \
+the project. The evidence must come from the project or its own developers."""
+
+    user = f"""\
+Question: Does this source show that {dbms_name}'s own developers use or \
+integrate SQLancer?
+
+If yes, classify how, using exactly one of these values: official_ci, \
+official_testing, developer_use, integration_contributed_by_dbms_team.
+
+SOURCE ({source_label}):
+<source_text>
+{source_text}
+</source_text>"""
+
+    schema = _schema({
+        "answer": {"type": "string", "enum": ANSWER_ENUM},
+        "relationship": {
+            "type": ["string", "null"],
+            "enum": ["official_ci", "official_testing", "developer_use",
+                     "integration_contributed_by_dbms_team", None],
+        },
+        "excerpts": _EXCERPTS,
+        "reason": _REASON,
+    }, ["answer", "relationship", "excerpts", "reason"])
+
+    return {"system": system, "user": user, "output_format": schema}
+
+
+def paper_summary(*, source_label: str, source_text: str,
+                  title: str, source_kind: str) -> dict:
+    """Summarise a citing paper and say how it relates to SQLancer.
+
+    The one question here whose output is prose rather than a verdict, so it is
+    the one place the no-paraphrase rule cannot apply to the answer itself. It
+    applies to the excerpts instead: the summary has to be accompanied by
+    passages that are literally in the text, and the caller drops any that are
+    not. A summary nobody can check against a quotation is marked as such
+    rather than published as fact.
+    """
+    system = f"""\
+You are summarising a research paper for a dataset that records how other work \
+relates to SQLancer, a testing tool for database management systems.
+
+You are given {source_kind} for the paper. Write a short factual summary of \
+what the paper is about and what it does, in three or four sentences, in your \
+own words. Then, separately, say what the text shows about its relationship to \
+SQLancer -- reusing its code, extending one of its techniques (NoREC, TLP, PQS, \
+QPG, CERT, DQP, CODDTest), comparing against it, calling it state of the art, \
+or only citing it.
+
+{SHARED_RULES}
+
+Two more rules specific to this task:
+
+- The summary is yours to write in your own words; the excerpts are not. Every \
+excerpt must be copied character-for-character from the SOURCE TEXT.
+- Describe only what this text supports. If you are working from an abstract or \
+from citation sentences rather than the full paper, say less rather than \
+guessing at the rest. Do not state what the paper's results were unless the \
+text says so."""
+
+    user = f"""\
+PAPER: {title}
+SOURCE: {source_label}
+
+SOURCE TEXT:
+-----
+{source_text}
+-----
+
+Summarise the paper, state its relationship to SQLancer, and quote the passages \
+you relied on."""
+
+    return {
+        "system": system,
+        "user": user,
+        "output_format": _schema(
+            {
+                "answer": {
+                    "type": "string",
+                    "enum": ANSWER_ENUM,
+                    "description": ("'yes' when the text was enough to write a "
+                                    "grounded summary, 'insufficient_evidence' "
+                                    "when it was not."),
+                },
+                "summary": {
+                    "type": "string",
+                    "description": ("Three or four sentences on what the paper "
+                                    "is about and what it does. Your own words."),
+                },
+                "relationship_to_sqlancer": {
+                    "type": "string",
+                    "description": ("One or two sentences on how this paper "
+                                    "relates to SQLancer, as far as this text "
+                                    "shows. Say so plainly if it only cites it."),
+                },
+                "excerpts": _EXCERPTS,
+                "reason": _REASON,
+            },
+            ["answer", "summary", "relationship_to_sqlancer", "excerpts", "reason"]),
+    }
+
+
+BUILDERS = {
+    "bug_attribution": bug_attribution,
+    "paper_summary": paper_summary,
+    "paper_infrastructure": paper_uses_infrastructure,
+    "paper_extends": paper_extends_technique,
+    "paper_compares": paper_compares_with,
+    "adoption": adoption,
+}
diff --git a/tools/impact/collectors/__init__.py b/tools/impact/collectors/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/tools/impact/collectors/adoption.py b/tools/impact/collectors/adoption.py
new file mode 100644
index 0000000..c2cf725
--- /dev/null
+++ b/tools/impact/collectors/adoption.py
@@ -0,0 +1,597 @@
+"""Finds evidence that a database system's own developers use SQLancer.
+
+The distinction this collector exists to protect is that SQLancer supporting a
+database system is not the same as that project using SQLancer. So the search is
+scoped to each registered project's own repository, and the strongest evidence
+is the most objective: a CI workflow or test script, inside the project's
+repository, that invokes SQLancer. That is accepted deterministically, with the
+matching line quoted.
+
+Softer evidence -- documentation, an issue or pull request describing use --
+goes to the classifier, whose job is to separate "our developers run this" from
+"someone tested us with this".
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Dict, List, Optional, Sequence, Tuple
+
+from .. import config, taxonomy
+from ..classify import Classifier, finder_names, resolve_finder
+from ..github import GitHub
+from ..util import content_hash, now, truncate, verbatim_excerpt
+
+COLLECTOR = "adoption"
+COLLECTOR_VERSION = "1.0.0"
+
+# Paths whose presence in a project's own repository means the project runs it.
+CI_PATH = re.compile(r"(^|/)\.github/workflows/|(^|/)\.circleci/|(^|/)\.gitlab-ci",
+                     re.IGNORECASE)
+TEST_PATH = re.compile(r"(^|/)(test|tests|testing|scripts|ci|tools)/",
+                       re.IGNORECASE)
+DOC_PATH = re.compile(r"\.(md|rst|txt|adoc)$", re.IGNORECASE)
+
+# A line that actually runs or configures SQLancer. Deliberately does not match
+# a bare repository link: a URL in a README is a mention, not an invocation.
+INVOCATION = re.compile(
+    r"(sqlancer[\w.-]*\.jar|java\s+-jar[^\n]{0,80}sqlancer|"
+    r"gradlew?[^\n]{0,40}sqlancer|mvn[^\n]{0,60}sqlancer|"
+    r"git\s+clone[^\n]{0,80}sqlancer|"
+    r"docker[^\n]{0,80}sqlancer|"
+    r"(?:^|\s)(?:run_|create_)?sqlancer\s*\(|"
+    r"import\s+[^\n]{0,60}sqlancer|from\s+[^\n]{0,60}sqlancer\s+import|"
+    r"uses:\s*[^\n]*sqlancer)",
+    re.IGNORECASE)
+
+# A sentence in which the project says, in prose, that it uses SQLancer. This is
+# what a contributor guide looks like, and it is better evidence than a link.
+# `(?:[^.\n]|\n(?!\n))` lets a sentence wrap across a single line break, which
+# is how prose is written in Markdown, without running into the next paragraph.
+_SENT = r"(?:[^.\n]|\n(?!\n))"
+USAGE_STATEMENT = re.compile(
+    rf"{_SENT}{{0,160}}\b(?:uses?|used|using|runs?|running|execute[sd]?"
+    rf"|test(?:s|ed|ing)?|fuzz(?:es|ed|ing)?|employ(?:s|ed)?)\b"
+    rf"{_SENT}{{0,120}}\bsqlancer\b{_SENT}{{0,200}}\.",
+    re.IGNORECASE)
+
+# A directory named after the tool inside the project's own repository: the
+# project keeps a SQLancer setup, which is structural rather than textual.
+SQLANCER_PATH = re.compile(r"(^|/)sqlancer([/.]|$)", re.IGNORECASE)
+
+RELATIONSHIPS = ("official_ci", "official_testing", "developer_use",
+                 "integration_contributed_by_dbms_team", "planned_adoption")
+
+# Issues in a project's own tracker proposing that it adopt SQLancer. Intent is
+# not use, so these are recorded under `planned_adoption` and counted apart from
+# projects actually running it -- a proposal that never lands would otherwise
+# inflate the reach figure.
+CURATED_ISSUES = (
+    {
+        "dbms": "spiceai",
+        "url": "https://github.com/spiceai/spiceai/issues/2119",
+        "relationship": "planned_adoption",
+    },
+)
+
+
+# A project proposing that it adopt SQLancer: "Introduce SQLancer", "support X
+# in sqlancer", "add SQLancer to our CI". Written by the project's own people in
+# the project's own tracker, which is what makes it adoption intent rather than
+# an outsider's suggestion.
+PROPOSAL = re.compile(
+    r"\b(?:introduce|introducing|add|adding|support|supporting|integrate|"
+    r"integrating|enable|enabling|set\s+up|setting\s+up|adopt|adopting|"
+    r"run|running|use|using|bring)\b[^.\n]{0,80}?\bsqlancer\b"
+    r"|\bsqlancer\b[^.\n]{0,60}?\b(?:integration|support|setup|adoption)\b",
+    re.IGNORECASE)
+
+# Only the project's own people can commit it to anything. CONTRIBUTOR belongs
+# here: GitHub uses it for someone with merged commits in the repository, which
+# is a developer of the project -- leaving it out was rejecting proposals from
+# the very engineers who would carry them out.
+INSIDER = {"OWNER", "MEMBER", "COLLABORATOR", "CONTRIBUTOR"}
+
+
+def proposal_records(gh: GitHub, issues: Sequence[dict], *, timestamp: str,
+                     repo_ids: Optional[Dict[str, str]] = None
+                     ) -> Tuple[List[dict], List[dict]]:
+    """Adoption proposals found in projects' own issue trackers.
+
+    Deliberately separate from adoption itself: an open proposal is an
+    intention, and counting it as use would overstate reach. Accepted only when
+    the author is an owner, member or collaborator of the project -- an
+    outsider suggesting SQLancer says nothing about what the project will do.
+    """
+    from .github_bugs import _owner_repo, _repo_index
+
+    repo_ids = repo_ids if repo_ids is not None else _repo_index()
+    registry = {entry["id"]: entry for entry in _registry()}
+    records: List[dict] = []
+    needs_review: List[dict] = []
+    seen = set()
+
+    for issue in issues:
+        url = issue.get("html_url") or ""
+        if not url or url in seen:
+            continue
+        owner_repo = _owner_repo(issue.get("repository_url") or url)
+        dbms_id = repo_ids.get((owner_repo or "").lower())
+        entry = registry.get(dbms_id) if dbms_id else None
+        if entry is None:
+            continue
+
+        text = "\n\n".join(part for part in
+                           [issue.get("title") or "", issue.get("body") or ""]
+                           if part)
+        match = PROPOSAL.search(text)
+        if not match:
+            continue
+        seen.add(url)
+
+        association = (issue.get("author_association") or "").upper()
+        author = (issue.get("user") or {}).get("login") or "a project member"
+        excerpt = truncate(match.group(0).strip(), MAX_SNIPPET)
+        if not verbatim_excerpt(text, excerpt):
+            continue
+
+        if association not in INSIDER:
+            needs_review.append({
+                "kind": "adoption", "url": url,
+                "reason": (f"proposes SQLancer adoption for {entry['name']} but "
+                           f"the author is not an owner, member or collaborator"),
+            })
+            continue
+
+        state = issue.get("state") or "open"
+        records.append(build_record(
+            entry, "planned_adoption",
+            # What this relationship means -- an intention rather than
+            # evidence of use -- is said once where these records are
+            # presented, not repeated in every summary. The summary's job is
+            # to say who proposed it and where it stands.
+            summary=truncate(
+                f"{author}, {association.lower()} of the {entry['name']} "
+                f"project, proposed adopting SQLancer; the issue is {state}.",
+                600),
+            evidence=[{
+                "source_url": url,
+                "source_type": "github_issue",
+                "excerpt": excerpt,
+                "excerpt_is_verbatim": True,
+                "note": (f"Issue in the {entry['name']} tracker proposing "
+                         f"SQLancer, opened by {author} ({association.lower()})."),
+                "content_sha256": content_hash(text),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }],
+            timestamp=timestamp))
+    return records, needs_review
+
+
+# Files in a project's own repository that show it keeping SQLancer work,
+# where the repository search does not reach them. DuckDB is the case this
+# exists for: it keeps a regression test per SQLancer finding under
+# test/issues/rigger/, twenty of them, each naming SQLancer in its header --
+# which is the project maintaining testing material, not a proposal to.
+CURATED_FILES: Tuple[Dict[str, str], ...] = (
+    {"dbms": "duckdb", "repository": "duckdb/duckdb",
+     "path": "test/issues/rigger/instr_crash.test",
+     "relationship": "official_testing",
+     "summary": ("The DuckDB project keeps a regression test for each SQLancer "
+                 "finding under test/issues/rigger/, so the bugs it found stay "
+                 "fixed.")},
+)
+
+
+def from_curated_files(gh: GitHub, *, timestamp: str,
+                       entries=None) -> Tuple[List[dict], List[dict]]:
+    """Record a project's own SQLancer material, quoting the file itself."""
+    records: List[dict] = []
+    needs_review: List[dict] = []
+    registry = {entry["id"]: entry for entry in _registry()}
+
+    for item in (entries if entries is not None else CURATED_FILES):
+        entry = registry.get(item["dbms"])
+        owner, _, repo = item["repository"].partition("/")
+        text, _ = (gh.raw_file(owner, repo, item["path"], "HEAD",
+                               max_age_days=30) if entry else (None, None))
+        if not entry or not text:
+            needs_review.append({
+                "kind": "adoption", "dbms": item["dbms"],
+                "url": f"https://github.com/{item['repository']}/blob/HEAD/{item['path']}",
+                "reason": "curated file could not be read",
+            })
+            continue
+        match = re.search(r"[^.\n]{0,200}\bSQLancer\b[^.\n]{0,200}", text,
+                          re.IGNORECASE)
+        blob = f"https://github.com/{item['repository']}/blob/HEAD/{item['path']}"
+        if not match or not verbatim_excerpt(text, match.group(0).strip()):
+            needs_review.append({
+                "kind": "adoption", "dbms": item["dbms"], "url": blob,
+                "reason": "curated file does not mention SQLancer",
+            })
+            continue
+        records.append(build_record(
+            entry, item["relationship"],
+            summary=truncate(item["summary"], 600),
+            evidence=[{
+                "source_url": blob,
+                "source_type": "github_code",
+                "excerpt": truncate(match.group(0).strip(), MAX_SNIPPET),
+                "excerpt_is_verbatim": True,
+                "note": (f"{item['path']} in the {entry['name']} repository, "
+                         f"one of the files the project keeps for this."),
+                "content_sha256": content_hash(text),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }],
+            timestamp=timestamp))
+    return records, needs_review
+
+
+def from_curated_issues(gh: GitHub, *, timestamp: str,
+                        entries=None) -> Tuple[List[dict], List[dict]]:
+    """Record proposals to adopt SQLancer, quoting the issue that made them."""
+    from ..util import github_issue_ref
+
+    records: List[dict] = []
+    needs_review: List[dict] = []
+    registry = {entry["id"]: entry for entry in _registry()}
+
+    for item in (entries if entries is not None else CURATED_ISSUES):
+        entry = registry.get(item["dbms"])
+        reference = github_issue_ref(item["url"])
+        if entry is None or reference is None:
+            needs_review.append({
+                "kind": "adoption", "url": item["url"],
+                "reason": "curated issue does not resolve to a registered system",
+            })
+            continue
+        owner, repo, _, number = reference
+        try:
+            issue, _ = gh.issue(owner, repo, number, max_age_days=30)
+        except Exception:
+            issue = None
+        if not issue:
+            needs_review.append({
+                "kind": "adoption", "url": item["url"],
+                "reason": "curated issue could not be fetched",
+            })
+            continue
+
+        text = "\n\n".join(part for part in
+                           [issue.get("title") or "", issue.get("body") or ""]
+                           if part)
+        match = re.search(r"[^.\n]{0,200}\bSQLancer\b[^.\n]{0,200}", text,
+                          re.IGNORECASE)
+        if not match or not verbatim_excerpt(text, match.group(0).strip()):
+            needs_review.append({
+                "kind": "adoption", "url": item["url"],
+                "reason": "curated issue does not mention SQLancer",
+            })
+            continue
+        excerpt = truncate(match.group(0).strip(), MAX_SNIPPET)
+        author = (issue.get("user") or {}).get("login") or "a project member"
+        state = issue.get("state") or "open"
+
+        records.append(build_record(
+            entry, item["relationship"],
+            summary=truncate(
+                f"{author} proposed that the {entry['name']} project adopt "
+                f"SQLancer for SQL fuzz testing; the issue is {state}. This is "
+                f"an intention, not evidence that the project runs SQLancer.",
+                600),
+            evidence=[{
+                "source_url": item["url"],
+                "source_type": "github_issue",
+                "excerpt": excerpt,
+                "excerpt_is_verbatim": True,
+                "note": (f"Issue in the {entry['name']} tracker proposing "
+                         f"SQLancer, opened by {author}."),
+                "content_sha256": content_hash(text),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }],
+            timestamp=timestamp))
+    return records, needs_review
+
+MAX_SNIPPET = 700
+
+
+def _registry() -> List[dict]:
+    return config.load_json(config.DATA_FILES["dbms"])["dbms"]
+
+
+def _owner_repo(repository: Optional[str]) -> Optional[Tuple[str, str]]:
+    if not repository:
+        return None
+    parts = repository.rstrip("/").split("/")
+    if len(parts) < 2:
+        return None
+    return parts[-2], parts[-1]
+
+
+def _relationship_for(path: str) -> Optional[str]:
+    if CI_PATH.search(path):
+        return "official_ci"
+    if TEST_PATH.search(path):
+        return "official_testing"
+    if DOC_PATH.search(path):
+        return "official_testing"
+    return None
+
+
+def _best_excerpt(text: str, path: str) -> Optional[Tuple[str, str]]:
+    """Pick the strongest verbatim line, and say what kind of evidence it is.
+
+    Preference order matters. A command that runs SQLancer is checked first,
+    because a build script's own words ("RUN java -jar ...") would otherwise be
+    caught by the prose pattern and quoted as if they were a sentence. Prose
+    comes next, and a structural path match last.
+    """
+    invocation = INVOCATION.search(text)
+    if invocation:
+        return _line_excerpt(text, invocation), "invocation"
+    statement = USAGE_STATEMENT.search(text)
+    if statement:
+        return truncate(statement.group(0).strip(), MAX_SNIPPET), "statement"
+    if SQLANCER_PATH.search(path):
+        mention = re.search(r"sqlancer", text, re.IGNORECASE)
+        if mention:
+            return _line_excerpt(text, mention), "path"
+    return None
+
+
+def _line_excerpt(text: str, match: re.Match) -> str:
+    start = text.rfind("\n", 0, match.start())
+    start = 0 if start < 0 else start + 1
+    end = text.find("\n", match.end())
+    end = len(text) if end < 0 else end
+    return truncate(text[start:end].strip(), MAX_SNIPPET)
+
+
+def build_record(entry: dict, relationship: str, *, summary: str,
+                 evidence: List[dict], timestamp: str,
+                 finder: Optional[str] = "sqlancer",
+                 classifier_record: Optional[dict] = None) -> dict:
+    record = {
+        "id": f"adoption:{entry['id']}:{relationship}",
+        "dbms": entry["id"],
+        "relationship": relationship,
+        "finder": finder,
+        "summary": truncate(summary, 600),
+        "active": True,
+        "evidence": evidence,
+        "provenance": {
+            "collector": COLLECTOR,
+            "collector_version": COLLECTOR_VERSION,
+            "policy_version": config.policy_version(),
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+            "source_url": evidence[0]["source_url"],
+            "source_type": evidence[0]["source_type"],
+        },
+    }
+    if entry.get("repository"):
+        record["project_repository"] = entry["repository"]
+    if classifier_record:
+        record["classifier"] = classifier_record
+    return record
+
+
+def inspect_project(gh: GitHub, entry: dict, *, timestamp: str,
+                    max_files: int = 6, tax: Optional[taxonomy.Taxonomy] = None
+                    ) -> Tuple[List[dict], List[dict]]:
+    """Look for SQLancer invocations inside one project's own repository.
+
+    Returns ``(records, candidates)``; candidates are files that mention
+    SQLancer without clearly invoking it, which is a question for the
+    classifier rather than a fact.
+    """
+    tax = tax or taxonomy.load()
+    parsed = _owner_repo(entry.get("repository"))
+    if not parsed:
+        return [], []
+    owner, repo = parsed
+
+    records: List[dict] = []
+    candidates: List[dict] = []
+    by_relationship: Dict[str, dict] = {}
+
+    query = f"repo:{owner}/{repo} sqlancer"
+    # Inspect the most telling files first: a directory named after the tool,
+    # then CI, then tests, then prose. This spends the per-project file budget
+    # where the evidence is strongest, and keeps a speculative sentence in a
+    # design document from outranking an actual test harness.
+    def priority(item: dict) -> int:
+        path = item.get("path") or ""
+        if SQLANCER_PATH.search(path):
+            return 0
+        if CI_PATH.search(path):
+            return 1
+        if TEST_PATH.search(path):
+            return 2
+        return 3
+
+    items = sorted(gh.search_code(query, max_pages=1), key=priority)
+    inspected = 0
+    for item in items:
+        if inspected >= max_files:
+            break
+        path = item.get("path") or ""
+        relationship = _relationship_for(path)
+        if relationship is None:
+            continue
+        inspected += 1
+        text, _ = gh.raw_file(owner, repo, path,
+                              item.get("ref") or "HEAD", max_age_days=30)
+        if not text:
+            continue
+        blob_url = (item.get("html_url")
+                    or f"https://github.com/{owner}/{repo}/blob/HEAD/{path}")
+        best = _best_excerpt(text, path)
+        if best is None:
+            candidates.append({
+                "dbms": entry["id"], "url": blob_url, "path": path,
+                "text": truncate(text, 20000),
+                "relationship_hint": relationship,
+            })
+            continue
+
+        excerpt, kind = best
+        if not verbatim_excerpt(text, excerpt):
+            continue
+        evidence = [{
+            "source_url": blob_url,
+            "source_type": ("github_workflow" if relationship == "official_ci"
+                            else "github_code"),
+            "excerpt": excerpt,
+            "excerpt_is_verbatim": True,
+            "note": (f"{path} in the {entry['name']} repository "
+                     + ("states that the project uses SQLancer."
+                        if kind == "statement" else
+                        "invokes SQLancer." if kind == "invocation" else
+                        "is part of a SQLancer setup the project maintains.")),
+            "content_sha256": content_hash(text),
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        }]
+        summary = (
+            f"The {entry['name']} project runs SQLancer from its own repository "
+            f"({path})."
+            if relationship == "official_ci" else
+            f"The {entry['name']} project maintains SQLancer testing setup in "
+            f"its own repository ({path}).")
+
+        # Credit the specific umbrella tool the file names, so a project that
+        # runs SQLancer++ is not recorded as running plain SQLancer.
+        named = tax.mentions_finder(excerpt) or tax.mentions_finder(text)
+        finder = named[0] if named else "sqlancer"
+
+        existing = by_relationship.get(relationship)
+        if existing is None:
+            record = build_record(entry, relationship, summary=summary,
+                                  evidence=evidence, timestamp=timestamp,
+                                  finder=finder)
+            by_relationship[relationship] = record
+            records.append(record)
+        else:
+            existing["evidence"].extend(evidence)
+
+    return records, candidates
+
+
+def collect(gh: GitHub, classifier: Optional[Classifier] = None, *,
+            timestamp: Optional[str] = None, entries: Optional[List[dict]] = None
+            ) -> Tuple[List[dict], List[dict], List[dict]]:
+    """Return ``(records, rejected, needs_review)`` across all known systems."""
+    timestamp = timestamp or now()
+    tax = taxonomy.load()
+    entries = entries if entries is not None else _registry()
+
+    records: List[dict] = []
+    rejected: List[dict] = []
+    needs_review: List[dict] = []
+
+    for entry in entries:
+        if not entry.get("repository"):
+            continue
+        try:
+            found, candidates = inspect_project(gh, entry, timestamp=timestamp,
+                                                tax=tax)
+        except RuntimeError:
+            # A failed search for one project must not end the run.
+            needs_review.append({
+                "kind": "adoption", "dbms": entry["id"],
+                "reason": "repository search failed",
+            })
+            continue
+        records.extend(found)
+
+        for candidate in candidates:
+            if any(record["dbms"] == candidate["dbms"] for record in found):
+                continue  # already established for this project
+            # Ask before giving up: the answer may be cached, and a
+            # cached answer needs no key. Checking availability first
+            # made a warm cache useless offline.
+            result = None if classifier is None else classifier.classify(
+                "adoption",
+                source_id=f"adoption:{candidate['url']}",
+                source_text=candidate["text"],
+                source_label=f"{entry['name']} repository file {candidate['path']}",
+                dbms_name=entry["name"],
+                finder_names=finder_names(tax))
+            if result is None:
+                needs_review.append({
+                    "kind": "adoption", "dbms": candidate["dbms"],
+                    "url": candidate["url"],
+                    "reason": "classification did not complete",
+                })
+                continue
+            if not result.is_positive:
+                rejected.append({
+                    "kind": "adoption", "dbms": candidate["dbms"],
+                    "url": candidate["url"], "answer": result.answer,
+                    "reason": result.reason or "not developer use",
+                })
+                continue
+            relationship = result.extra.get("relationship")
+            if relationship not in RELATIONSHIPS:
+                relationship = candidate["relationship_hint"]
+            evidence = [{
+                "source_url": candidate["url"],
+                "source_type": "github_code",
+                "excerpt": result.excerpts[0],
+                "excerpt_is_verbatim": True,
+                "note": f"{candidate['path']} in the {entry['name']} repository.",
+                "content_sha256": content_hash(candidate["text"]),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }]
+            records.append(build_record(
+                entry, relationship,
+                summary=truncate(result.reason or
+                                 f"The {entry['name']} project uses SQLancer.", 600),
+                evidence=evidence, timestamp=timestamp,
+                finder=resolve_finder(tax, result.extra.get("finder")) or "sqlancer",
+                classifier_record=result.classifier_record()))
+
+    curated, curated_review = from_curated_issues(gh, timestamp=timestamp)
+    records.extend(curated)
+    needs_review.extend(curated_review)
+
+    files, file_review = from_curated_files(gh, timestamp=timestamp)
+    records.extend(files)
+    needs_review.extend(file_review)
+
+    # Proposals live in the same per-repository issue searches the bug
+    # collector runs, so they cost nothing extra beyond reading them.
+    try:
+        from . import github_bugs
+        tax_ = taxonomy.load()
+        repo_ids = github_bugs._repo_index()
+        issues: List[dict] = []
+        for query in github_bugs.search_queries(tax_, sorted(repo_ids)):
+            issues.extend(gh.search_issues(query, max_pages=2))
+        proposals, proposal_review = proposal_records(
+            gh, issues, timestamp=timestamp, repo_ids=repo_ids)
+        records.extend(proposals)
+        needs_review.extend(proposal_review)
+    except Exception:
+        pass
+
+    # One record per (dbms, relationship); merge evidence rather than duplicate.
+    merged: Dict[str, dict] = {}
+    for record in records:
+        existing = merged.get(record["id"])
+        if existing is None:
+            merged[record["id"]] = record
+        else:
+            existing["evidence"].extend(record["evidence"])
+    return list(merged.values()), rejected, needs_review
diff --git a/tools/impact/collectors/artifact_links.py b/tools/impact/collectors/artifact_links.py
new file mode 100644
index 0000000..8e8dd91
--- /dev/null
+++ b/tools/impact/collectors/artifact_links.py
@@ -0,0 +1,196 @@
+"""Extracts artifact links from the papers themselves.
+
+The other routes to a paper's artifact are guesses: code search finds
+repositories that look derived, and repository search finds ones whose name
+resembles the tool. This one reads what the paper actually says. Authors put
+the link in the abstract, in a footnote on the first page, or in an artifact
+availability section, and following it is both the most reliable route and the
+one a person would use.
+
+Three sources are tried in order of cost: the arXiv abstract page, which is
+small HTML; the open-access PDF, whose first and last pages hold the link
+almost every time; and the DOI landing page. Anything found is a URL the paper
+published, so it needs no separate argument that the repository belongs to it.
+"""
+
+from __future__ import annotations
+
+import io
+import re
+from typing import Dict, Iterable, List, Optional, Sequence, Tuple
+
+from ..http import Fetcher
+from ..util import normalize_url, now, truncate
+
+COLLECTOR = "artifact_links"
+COLLECTOR_VERSION = "1.0.0"
+
+# Hosts that hold research artifacts. Anything else a paper links is not one.
+ARTIFACT_URL = re.compile(
+    r"https?://(?:www\.)?"
+    r"(github\.com/[\w.-]+/[\w.-]+"
+    r"|gitlab\.com/[\w.-]+/[\w.-]+"
+    r"|bitbucket\.org/[\w.-]+/[\w.-]+"
+    r"|zenodo\.org/(?:record|records|doi)/[\w./-]+"
+    r"|figshare\.com/[\w./-]+"
+    r"|doi\.org/10\.5281/zenodo\.[\w.]+)",
+    re.IGNORECASE)
+
+# GitHub paths that are not a project: the site's own pages and user profiles.
+NOT_A_REPOSITORY = re.compile(
+    r"github\.com/(about|features|pricing|topics|collections|sponsors|"
+    r"readme|site|security|apps|marketplace|orgs|settings|login|join|"
+    r"blog|explore)(/|$)", re.IGNORECASE)
+
+# Text around a link that marks it as the paper's own artifact rather than a
+# tool it merely cites. Used to rank, not to exclude.
+ARTIFACT_PHRASE = re.compile(
+    r"(artifact|available at|open.?sourc\w*|our\s+(?:tool|implementation|"
+    r"prototype|code)|source\s+code|replication|reproduc\w*|"
+    r"we\s+(?:release|publish|provide))",
+    re.IGNORECASE)
+
+PDF_MAX_BYTES = 12_000_000
+PDF_PAGES_FRONT = 2
+PDF_PAGES_BACK = 2
+EXCERPT_LIMIT = 500
+
+
+def _clean(url: str) -> Optional[str]:
+    """Normalise a link found in running text.
+
+    PDF extraction leaves trailing punctuation and citation markers glued to
+    URLs, and a repository URL often arrives with a branch or file path on the
+    end; both are trimmed back to the repository itself.
+    """
+    url = url.rstrip(").,;:'\"]}>")
+    canonical = normalize_url(url)
+    if not canonical or NOT_A_REPOSITORY.search(canonical):
+        return None
+    match = re.match(r"^(https://github\.com/[\w.-]+/[\w.-]+)", canonical)
+    if match:
+        canonical = match.group(1)
+        if canonical.endswith(".git"):
+            canonical = canonical[:-4]
+    return canonical
+
+
+def urls_in(text: str) -> List[Tuple[str, str]]:
+    """``(url, surrounding text)`` for every artifact-hosting link in ``text``."""
+    found: List[Tuple[str, str]] = []
+    seen = set()
+    for match in ARTIFACT_URL.finditer(text or ""):
+        canonical = _clean(match.group(0))
+        if not canonical or canonical in seen:
+            continue
+        seen.add(canonical)
+        start = max(0, match.start() - 240)
+        end = min(len(text), match.end() + 120)
+        found.append((canonical, re.sub(r"\s+", " ", text[start:end]).strip()))
+    return found
+
+
+def _rank(entries: Sequence[Tuple[str, str]]) -> List[Tuple[str, str]]:
+    """Links whose surrounding text calls them an artifact come first."""
+    return sorted(entries, key=lambda item: 0 if ARTIFACT_PHRASE.search(item[1])
+                  else 1)
+
+
+def from_arxiv(fetcher: Fetcher, arxiv_id: str) -> List[Tuple[str, str]]:
+    """Links on an arXiv abstract page."""
+    try:
+        entry, _ = fetcher.fetch(f"https://arxiv.org/abs/{arxiv_id}",
+                                 max_age_days=90)
+    except Exception:
+        return []
+    if entry.get("status") != 200:
+        return []
+    text = re.sub(r"<[^>]+>", " ", entry.get("body") or "")
+    return urls_in(text)
+
+
+def from_pdf(fetcher: Fetcher, url: str) -> List[Tuple[str, str]]:
+    """Links on the first and last pages of an open-access PDF.
+
+    Only the ends of the paper are read: the artifact link lives in a first-page
+    footnote or an availability section at the back, and parsing every page of
+    every paper would cost far more for almost nothing.
+    """
+    try:
+        entry, _ = fetcher.fetch(url, max_age_days=180)
+    except Exception:
+        return []
+    body = entry.get("body") or ""
+    if entry.get("status") != 200 or not body:
+        return []
+    raw = body.encode("utf-8", errors="replace")
+    if not raw.lstrip().startswith(b"%PDF") or len(raw) > PDF_MAX_BYTES:
+        # Not a PDF: publishers often answer with an HTML landing page.
+        return urls_in(re.sub(r"<[^>]+>", " ", body))
+    try:
+        import pypdf
+        reader = pypdf.PdfReader(io.BytesIO(raw))
+        pages = reader.pages
+        wanted = list(range(min(PDF_PAGES_FRONT, len(pages))))
+        wanted += [i for i in range(max(0, len(pages) - PDF_PAGES_BACK),
+                                    len(pages)) if i not in wanted]
+        text = "\n".join(pages[i].extract_text() or "" for i in wanted)
+    except Exception:
+        return []
+    # PDF extraction frequently breaks a URL across a line.
+    text = re.sub(r"-\s*\n\s*", "", text)
+    text = re.sub(r"\n", " ", text)
+    return urls_in(text)
+
+
+def for_paper(fetcher: Fetcher, paper: dict) -> List[dict]:
+    """Artifact links this paper publishes, best first.
+
+    Each entry carries the sentence the link was found in, so the claim that a
+    repository belongs to a paper is backed by the paper's own words.
+    """
+    entries: List[Tuple[str, str, str]] = []
+
+    if paper.get("arxiv_id"):
+        for url, context in from_arxiv(fetcher, paper["arxiv_id"]):
+            entries.append((url, context, f"https://arxiv.org/abs/{paper['arxiv_id']}"))
+
+    pdf = paper.get("open_access_pdf")
+    if pdf:
+        for url, context in from_pdf(fetcher, pdf):
+            entries.append((url, context, pdf))
+
+    seen = set()
+    ranked = _rank([(url, context) for url, context, _ in entries])
+    order = {url: index for index, (url, _) in enumerate(ranked)}
+    entries.sort(key=lambda item: order.get(item[0], 99))
+
+    out: List[dict] = []
+    for url, context, source in entries:
+        if url in seen:
+            continue
+        seen.add(url)
+        out.append({
+            "url": url,
+            "context": truncate(context, EXCERPT_LIMIT),
+            "source_url": source,
+            "states_artifact": bool(ARTIFACT_PHRASE.search(context)),
+        })
+    return out
+
+
+def link_evidence(entry: dict, timestamp: Optional[str] = None) -> dict:
+    """Evidence that the paper itself points at this repository."""
+    timestamp = timestamp or now()
+    return {
+        "source_url": entry["source_url"],
+        "source_type": "paper",
+        "excerpt": entry["context"],
+        "excerpt_is_verbatim": True,
+        "note": (f"The paper links {entry['url']} as its artifact."
+                 if entry.get("states_artifact") else
+                 f"The paper links {entry['url']}."),
+        "retrieved_at": timestamp,
+        "first_seen": timestamp,
+        "last_verified": timestamp,
+    }
diff --git a/tools/impact/collectors/artifacts.py b/tools/impact/collectors/artifacts.py
new file mode 100644
index 0000000..1b622f4
--- /dev/null
+++ b/tools/impact/collectors/artifacts.py
@@ -0,0 +1,557 @@
+"""Finds and inspects research artifacts for signs of SQLancer reuse.
+
+``uses_infrastructure`` is meant to be the objective category, so it is decided
+from the artifact rather than from how a paper describes itself. This module
+locates a paper's artifact repository and looks for concrete markers: a fork of
+the SQLancer repository, the ``sqlancer`` Java package layout, retained source
+files, a copyright notice, a README that says so.
+
+No single marker is treated as decisive. ``Randomly.java`` shows up in plenty of
+repositories that merely vendored a snippet, so the markers are returned as a
+set and the decision -- taken elsewhere -- combines them.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Dict, List, Optional, Sequence, Tuple
+
+from ..github import GitHub
+from ..util import content_hash, now, truncate
+
+COLLECTOR = "artifacts"
+COLLECTOR_VERSION = "1.0.0"
+EXTRACTOR_VERSION = "artifact-markers-v1"
+
+SQLANCER_REPO_FULL_NAME = "sqlancer/sqlancer"
+
+# Identifiers that together fingerprint SQLancer's own Randomly.java. This is
+# what catches a derivative that renamed the package -- a rename defeats every
+# path-based check while leaving the source itself untouched, and renaming is
+# common enough in research artifacts that path matching alone misses real
+# reuse. Requiring several of these together keeps an ordinary utility class
+# named Randomly from matching.
+RANDOMLY_FINGERPRINT = (
+    "StringGenerationStrategy",
+    "SOPHISTICATED",
+    "cachedLongs",
+    "cachedStrings",
+    "cachedDoubles",
+    "getPositiveIntegerNotNull",
+    "ALPHANUMERIC_SPECIALCHAR",
+    "getNotCachedInteger",
+)
+RANDOMLY_FINGERPRINT_MINIMUM = 5
+
+# Path segments that mark a nested copy of somebody else's project: a vendored
+# baseline, a bundled comparison artifact, a third-party tree. SQLancer's source
+# found under one of these belongs to that nested project, not necessarily to
+# the paper whose repository it happens to sit in -- so it is evidence worth
+# looking at rather than evidence that settles the question.
+NESTED_ARTIFACT_SEGMENT = re.compile(
+    r"(^|/)([\w.-]*(?:artifact|baseline|comparison|competitor|related[_-]?work"
+    r"|third[_-]?party|vendor|external|reference[_-]?impl)[\w.-]*)/",
+    re.IGNORECASE)
+
+# Files whose presence is characteristic of the SQLancer codebase.
+CHARACTERISTIC_FILES = (
+    "src/sqlancer/Randomly.java",
+    "src/sqlancer/Main.java",
+    "src/sqlancer/DBMSExecutor.java",
+    "src/sqlancer/common/oracle/TestOracle.java",
+    "src/sqlancer/common/oracle/NoRECOracle.java",
+    "src/sqlancer/common/oracle/TLPWhereOracle.java",
+)
+
+README_REUSE = re.compile(
+    r"(built?\s+(?:on\s+top\s+of|upon)\s+sqlancer|based\s+on\s+sqlancer|"
+    r"fork(?:ed)?\s+(?:of|from)\s+sqlancer|extend(?:s|ed)?\s+sqlancer|"
+    r"we\s+(?:use|used|modified|extended)\s+sqlancer|"
+    r"implemented?\s+(?:on\s+top\s+of|in|within)\s+sqlancer)",
+    re.IGNORECASE)
+
+COPYRIGHT_NOTICE = re.compile(
+    r"copyright[^\n]{0,80}sqlancer|sqlancer[^\n]{0,40}mit\s+license", re.IGNORECASE)
+
+ARTIFACT_HOSTS = {
+    "github.com": "github",
+    "gitlab.com": "gitlab",
+    "bitbucket.org": "bitbucket",
+    "zenodo.org": "zenodo",
+    "doi.org": "other",
+    "figshare.com": "figshare",
+}
+
+# Text that a marker was found in, kept short enough to be readable evidence.
+EXCERPT_LIMIT = 600
+
+
+# Artifact repositories a maintainer has linked to a paper by hand.
+#
+# Forward search cannot always get there: a repository is usually named after
+# the tool, the tool is often named only in the body of the paper rather than
+# its title or indexed abstract, and the README frequently says "this paper"
+# without a citation. Where the link is knowledge rather than a derivation, it
+# is written down. The SQLancer markers are still verified by inspecting the
+# repository; only the paper-to-repository link comes from here.
+KNOWN_ARTIFACTS: Dict[str, str] = {
+    # Empty on purpose. The first entry tried here was wrong -- a repository
+    # guessed at from a title search -- while the automatic path linked the same
+    # artifact correctly from its repository description. Add an entry only when
+    # the link is known, not inferred.
+}
+
+
+def find_sqlancer_derived_repositories(gh: GitHub, *, limit: int = 100
+                                       ) -> List[str]:
+    """Repositories on GitHub that contain SQLancer's own source files.
+
+    Code search rather than repository search, because the giveaway is a file
+    path inside the repository, not anything in its name or description. This
+    finds artifacts whose authors never mention SQLancer anywhere a reader
+    would look.
+
+    Incomplete by construction -- GitHub excludes forks from code search and
+    indexes only some repositories -- so it supplements the other routes rather
+    than replacing them.
+    """
+    found: List[str] = []
+    queries = [
+        # Content first: this is the query that finds derivatives which renamed
+        # the package, which path-scoped queries cannot see at all.
+        "filename:Randomly.java StringGenerationStrategy",
+        "filename:Randomly.java ALPHANUMERIC_SPECIALCHAR",
+        "path:src/sqlancer filename:Randomly.java",
+        "path:src/sqlancer filename:DBMSExecutor.java",
+        "path:src/sqlancer/common/oracle filename:TestOracle.java",
+    ]
+    for query in queries:
+        try:
+            for item in gh.search_code(query, max_pages=1):
+                full_name = (item.get("repository") or {}).get("full_name")
+                if full_name and full_name not in found:
+                    found.append(full_name)
+                if len(found) >= limit:
+                    return found
+        except RuntimeError:
+            continue
+    return found
+
+
+def classify_artifact_url(url: str) -> Optional[str]:
+    lowered = url.lower()
+    for host, kind in ARTIFACT_HOSTS.items():
+        if host in lowered:
+            if kind == "other" and "zenodo" in lowered:
+                return "zenodo"
+            return kind
+    return None
+
+
+# Words that look like a tool name but are just prose. A paper's tool name is
+# the thing its artifact repository is usually called, so this list is what
+# stops "SQL" or "DBMS" being treated as one.
+_NOT_TOOL_NAMES = {
+    "SQL", "DBMS", "DBMSS", "DBMSES", "NOSQL", "ACID", "API", "CPU", "GPU",
+    "LLM", "LLMS", "AI", "ML", "IR", "AST", "CI", "CD", "OLAP", "OLTP", "JSON",
+    "XML", "HTML", "UDF", "UDFS", "JIT", "IO", "OS", "RDBMS", "RDBMSS", "TPC",
+    "ANSI", "ISO", "IEEE", "ACM", "USENIX", "ICSE", "FSE", "ASE", "OSDI",
+    "SIGMOD", "VLDB", "OOPSLA", "PLDI", "ISSTA", "WHERE", "SELECT", "GROUP",
+    "ORDER", "JOIN", "NULL", "AND", "OR", "NOT", "THE", "FOR", "VIA", "WITH",
+    "TEST", "TESTING", "BUGS", "BUG", "NEW", "TOOL", "PAPER", "ARTIFACT",
+}
+
+# "we implement ... in a tool called X" and its common variants.
+TOOL_NAME_PHRASE = re.compile(
+    r"\b(?:tool|prototype|system|framework|implementation)\s+(?:called|named)\s+"
+    r"([A-Z][A-Za-z0-9_+-]{2,24})"
+    r"|\bwe\s+(?:call|name)\s+(?:it|our\s+\w+)\s+([A-Z][A-Za-z0-9_+-]{2,24})"
+    r"|\b(?:we\s+)?(?:present|introduce|propose|implement(?:ed)?|built?)\s+"
+    r"([A-Z][A-Za-z0-9_+-]{2,24})\s*,?\s*(?:a|an|the)\b")
+
+
+def candidate_tool_names(title: str, abstract: Optional[str] = None) -> List[str]:
+    """Tool names a paper appears to give its own implementation.
+
+    Artifact repositories are named after the tool far more often than after
+    the paper, so this is what makes a repository findable at all. Names are
+    drawn from the paper's own text only.
+    """
+    names: List[str] = []
+
+    def add(name: Optional[str]) -> None:
+        if not name:
+            return
+        cleaned = name.strip().strip(".,:;")
+        if len(cleaned) < 3 or cleaned.upper() in _NOT_TOOL_NAMES:
+            return
+        if cleaned not in names:
+            names.append(cleaned)
+
+    for match in TOOL_NAME_PHRASE.finditer(abstract or ""):
+        for group in match.groups():
+            add(group)
+    # A leading "NAME: rest of the title" is the other common convention.
+    leading = re.match(r"^([A-Z][A-Za-z0-9_+-]{2,24})\s*[:\u2013-]\s+", title or "")
+    if leading:
+        add(leading.group(1))
+    # All-caps acronyms in the title are usually the tool.
+    for token in re.findall(r"\b([A-Z][A-Z0-9]{2,15})\b", title or ""):
+        add(token)
+    return names[:4]
+
+
+def find_repository_candidates(gh: GitHub, title: str, *,
+                               tool_names: Optional[Sequence[str]] = None,
+                               limit: int = 5) -> List[dict]:
+    """Search GitHub for repositories that might hold a paper's artifact.
+
+    Two queries, because artifacts are named inconsistently: the tool name as a
+    repository name, and the paper title in the description or README. Results
+    are only candidates -- linking one to the paper still needs its own
+    evidence.
+    """
+    results: List[dict] = []
+    seen = set()
+
+    def collect(query: str) -> None:
+        try:
+            for item in gh.search("repositories", query, max_pages=1):
+                full_name = item.get("full_name")
+                if full_name and full_name not in seen:
+                    seen.add(full_name)
+                    results.append(item)
+                if len(results) >= limit:
+                    return
+        except RuntimeError:
+            return
+
+    for name in (tool_names or [])[:2]:
+        if len(results) >= limit:
+            break
+        collect(f"{name} in:name")
+
+    words = [w for w in re.findall(r"[A-Za-z0-9+]{4,}", title or "")][:6]
+    if len(words) >= 2 and len(results) < limit:
+        collect(" ".join(words) + " in:name,description,readme")
+    return results[:limit]
+
+
+# Terms that show a repository is about database testing at all. Without one of
+# these, a name collision is far more likely than a real artifact.
+DBMS_CONTEXT = re.compile(
+    r"\b(sql|dbms|database|databases|query|queries|oracle|fuzz\w*|"
+    r"logic bug|test(?:ing)?|sqlite|mysql|postgres\w*|duckdb|mariadb|"
+    r"cockroach\w*|tidb|clickhouse)\b", re.IGNORECASE)
+
+
+def repository_text(gh: GitHub, owner: str, repo: str) -> Optional[dict]:
+    """A repository's description and README, fetched once.
+
+    Linking artifacts to papers compares every candidate repository against
+    every paper, so the text each comparison needs is read once and then held
+    in memory rather than re-read per paper.
+    """
+    metadata = gh.repo(owner, repo)
+    if not metadata:
+        return None
+    default_branch = metadata.get("default_branch") or "HEAD"
+    readme = ""
+    for name in ("README.md", "readme.md", "README.rst", "README"):
+        text, _ = gh.raw_file(owner, repo, name, default_branch, max_age_days=30)
+        if text:
+            readme = text
+            break
+    return {
+        "full_name": f"{owner}/{repo}",
+        "owner": owner,
+        "repo": repo,
+        "description": metadata.get("description") or "",
+        "readme": readme,
+        "default_branch": default_branch,
+    }
+
+
+def link_evidence(gh: GitHub, owner: str, repo: str, paper: dict, *,
+                  tool_names: Sequence[str], timestamp: str
+                  ) -> Optional[dict]:
+    """Evidence that this repository is that paper's artifact, or None."""
+    entry = repository_text(gh, owner, repo)
+    if entry is None:
+        return None
+    return link_evidence_from_text(entry, paper, tool_names=tool_names,
+                                   timestamp=timestamp)
+
+
+def link_evidence_from_text(entry: dict, paper: dict, *,
+                            tool_names: Sequence[str], timestamp: str
+                            ) -> Optional[dict]:
+    """Evidence that this repository is that paper's artifact, or None.
+
+    A repository is only linked when something ties it to the paper in
+    particular: its text names the paper, its DOI or its arXiv id, or it is
+    named after the tool the paper says it built and is recognisably about
+    database testing. Marker evidence says a repository derives from SQLancer;
+    it says nothing about whose artifact it is, so the two are kept separate.
+    """
+    owner, repo = entry["owner"], entry["repo"]
+    description, readme = entry["description"], entry["readme"]
+    haystack = f"{description}\n{readme}"
+    url = f"https://github.com/{owner}/{repo}"
+
+    title = (paper.get("title") or "").strip()
+    normalised_title = re.sub(r"[^a-z0-9]+", " ", title.lower()).strip()
+    normalised_hay = re.sub(r"[^a-z0-9]+", " ", haystack.lower())
+
+    def evidence(excerpt: Optional[str], note: str) -> dict:
+        item = {
+            "source_url": url,
+            "source_type": "github_repository",
+            "excerpt": excerpt,
+            "note": note,
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        }
+        if excerpt is not None:
+            item["excerpt_is_verbatim"] = True
+        return item
+
+    if normalised_title and normalised_title in normalised_hay:
+        # Locate the title in the raw text, allowing for the whitespace and
+        # punctuation that normalisation removed. Without this the quote can
+        # come from somewhere else in the README entirely, which is worse than
+        # no quote at all: it looks like evidence and is not.
+        flexible = r"\s*".join(re.escape(word) for word in title.split())
+        match = re.search(flexible, haystack, re.IGNORECASE)
+        excerpt = _sentence_around(haystack, match.start()) if match else None
+        return evidence(excerpt, "Repository names this paper.")
+
+    for identifier in filter(None, [paper.get("doi"), paper.get("arxiv_id")]):
+        if identifier.lower() in haystack.lower():
+            index = haystack.lower().find(identifier.lower())
+            return evidence(_sentence_around(haystack, index),
+                            f"Repository cites the paper's {identifier}.")
+
+    for name in tool_names:
+        if repo.lower() != name.lower():
+            continue
+        if not DBMS_CONTEXT.search(haystack):
+            continue
+        return evidence(
+            truncate(description or readme.strip().splitlines()[0], EXCERPT_LIMIT)
+            if (description or readme.strip()) else None,
+            (f"Repository is named after {name}, the tool this paper says it "
+             f"built, and is about database testing."))
+    return None
+
+
+def inspect_repository(gh: GitHub, owner: str, repo: str, *,
+                       timestamp: Optional[str] = None
+                       ) -> Tuple[List[str], List[dict]]:
+    """Return ``(markers, evidence)`` for a candidate artifact repository."""
+    timestamp = timestamp or now()
+    url = f"https://github.com/{owner}/{repo}"
+    markers: List[str] = []
+    evidence: List[dict] = []
+
+    metadata = gh.repo(owner, repo)
+    if not metadata:
+        return markers, evidence
+
+    parent = (metadata.get("parent") or {}).get("full_name", "").lower()
+    source = (metadata.get("source") or {}).get("full_name", "").lower()
+    if SQLANCER_REPO_FULL_NAME in (parent, source):
+        markers.append("fork_of_sqlancer_repository")
+        evidence.append({
+            "source_url": url,
+            "source_type": "github_repository",
+            "excerpt": None,
+            "note": (f"GitHub reports {owner}/{repo} as a fork of "
+                     f"{SQLANCER_REPO_FULL_NAME}."),
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        })
+
+    default_branch = metadata.get("default_branch") or "HEAD"
+    tree = gh.tree(owner, repo, default_branch)
+    paths = {entry["path"] for entry in tree if entry.get("type") == "blob"}
+    directories = {entry["path"] for entry in tree if entry.get("type") == "tree"}
+
+    if any(path.startswith("src/sqlancer/") for path in paths):
+        markers.append("sqlancer_package_structure")
+    if any(directory.startswith("src/sqlancer/") for directory in directories):
+        markers.append("sqlancer_provider_directory")
+
+    present = [path for path in CHARACTERISTIC_FILES if path in paths]
+    if present:
+        markers.append("retained_sqlancer_source_files")
+        evidence.append({
+            "source_url": f"{url}/blob/{default_branch}/{present[0]}",
+            "source_type": "github_code",
+            "excerpt": None,
+            "note": ("Repository retains SQLancer source files: "
+                     + ", ".join(present[:5])),
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        })
+    if "src/sqlancer/Randomly.java" in paths:
+        markers.append("randomly_java_present")
+
+    # Any Randomly.java, wherever it sits and whatever package it declares.
+    for path in sorted(p for p in paths
+                       if p == "Randomly.java" or p.endswith("/Randomly.java")):
+        if path == "src/sqlancer/Randomly.java":
+            continue
+        text, _ = gh.raw_file(owner, repo, path, default_branch)
+        if not text:
+            continue
+        present = [token for token in RANDOMLY_FINGERPRINT if token in text]
+        if len(present) < RANDOMLY_FINGERPRINT_MINIMUM:
+            continue
+        nested = NESTED_ARTIFACT_SEGMENT.search(path)
+        markers.append("sqlancer_source_in_nested_artifact" if nested
+                       else "sqlancer_source_content_match")
+        package = re.search(r"^\s*package\s+([\w.]+)\s*;", text, re.MULTILINE)
+        if package and not package.group(1).startswith("sqlancer"):
+            markers.append("renamed_sqlancer_package")
+        snippet = _first_match_line(text, "stringgenerationstrategy")
+        evidence.append({
+            "source_url": f"{url}/blob/{default_branch}/{path}",
+            "source_type": "github_code",
+            "excerpt": snippet,
+            "excerpt_is_verbatim": True,
+            "note": ((f"{path} is SQLancer's Randomly.java, but it sits under "
+                      f"{nested.group(2)}, which looks like a bundled copy of "
+                      f"another project"
+                      ) if nested else
+                     f"{path} is SQLancer's Randomly.java"
+                     + (f", with the package renamed to "
+                        f"{package.group(1)}" if package and
+                        not package.group(1).startswith("sqlancer") else "")
+                     + f" ({len(present)} of {len(RANDOMLY_FINGERPRINT)} "
+                       f"identifiers match: {', '.join(present[:5])})."),
+            "content_sha256": content_hash(text),
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        })
+        break
+
+    for build_file in ("pom.xml", "build.gradle", "build.gradle.kts"):
+        if build_file not in paths:
+            continue
+        text, _ = gh.raw_file(owner, repo, build_file, default_branch)
+        if text and "sqlancer" in text.lower():
+            markers.append("sqlancer_build_file_reference")
+            snippet = _first_match_line(text, "sqlancer")
+            if snippet:
+                evidence.append({
+                    "source_url": f"{url}/blob/{default_branch}/{build_file}",
+                    "source_type": "github_code",
+                    "excerpt": snippet,
+                    "excerpt_is_verbatim": True,
+                    "note": f"{build_file} identifies the project as SQLancer.",
+                    "content_sha256": content_hash(text),
+                    "retrieved_at": timestamp,
+                    "first_seen": timestamp,
+                    "last_verified": timestamp,
+                })
+            break
+
+    for readme in ("README.md", "readme.md", "README.rst", "README"):
+        if readme not in paths:
+            continue
+        text, _ = gh.raw_file(owner, repo, readme, default_branch)
+        if not text:
+            break
+        match = README_REUSE.search(text)
+        if match:
+            markers.append("readme_states_reuse")
+            evidence.append({
+                "source_url": f"{url}/blob/{default_branch}/{readme}",
+                "source_type": "documentation",
+                "excerpt": _sentence_around(text, match.start()),
+                "excerpt_is_verbatim": True,
+                "note": "README states that the artifact reuses SQLancer.",
+                "content_sha256": content_hash(text),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            })
+        break
+
+    for licence in ("LICENSE", "LICENSE.md", "LICENSE.txt", "NOTICE"):
+        if licence not in paths:
+            continue
+        text, _ = gh.raw_file(owner, repo, licence, default_branch)
+        if text and COPYRIGHT_NOTICE.search(text):
+            markers.append("sqlancer_copyright_or_license_notice")
+            match = COPYRIGHT_NOTICE.search(text)
+            evidence.append({
+                "source_url": f"{url}/blob/{default_branch}/{licence}",
+                "source_type": "github_code",
+                "excerpt": _sentence_around(text, match.start()),
+                "excerpt_is_verbatim": True,
+                "note": "Licence file carries a SQLancer copyright notice.",
+                "content_sha256": content_hash(text),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            })
+        break
+
+    return sorted(set(markers)), evidence
+
+
+def _first_match_line(text: str, needle: str) -> Optional[str]:
+    for line in text.splitlines():
+        if needle in line.lower():
+            stripped = line.strip()
+            if stripped:
+                return truncate(stripped, EXCERPT_LIMIT)
+    return None
+
+
+def _sentence_around(text: str, index: int) -> str:
+    """Verbatim slice of ``text`` around ``index``, snapped to line breaks."""
+    start = text.rfind("\n", 0, max(0, index - 200))
+    start = 0 if start < 0 else start + 1
+    end = text.find("\n", index + 200)
+    end = len(text) if end < 0 else end
+    return truncate(text[start:end].strip(), EXCERPT_LIMIT)
+
+
+# Markers that on their own are strong enough to settle the question, and the
+# weaker ones that need corroboration.
+DECISIVE_MARKERS = {
+    "fork_of_sqlancer_repository",
+    "sqlancer_copyright_or_license_notice",
+    "readme_states_reuse",
+    "git_history_derived_from_sqlancer",
+    # SQLancer's own source, identified by its content rather than its path.
+    "sqlancer_source_content_match",
+}
+SUPPORTING_MARKERS = {
+    "sqlancer_source_in_nested_artifact",
+    "renamed_sqlancer_package",
+    "sqlancer_package_structure",
+    "retained_sqlancer_source_files",
+    "sqlancer_build_file_reference",
+    "sqlancer_provider_directory",
+    "randomly_java_present",
+}
+
+
+def markers_are_conclusive(markers: Sequence[str]) -> bool:
+    """Whether the marker set settles infrastructure reuse without a model.
+
+    One decisive marker suffices; otherwise at least two supporting markers are
+    required, so a lone ``Randomly.java`` never carries the decision.
+    """
+    marker_set = set(markers)
+    if marker_set & DECISIVE_MARKERS:
+        return True
+    return len(marker_set & SUPPORTING_MARKERS) >= 2
diff --git a/tools/impact/collectors/dbms_registry.py b/tools/impact/collectors/dbms_registry.py
new file mode 100644
index 0000000..1557d8a
--- /dev/null
+++ b/tools/impact/collectors/dbms_registry.py
@@ -0,0 +1,329 @@
+"""Builds the DBMS registry from the SQLancer repository.
+
+Which systems SQLancer supports is answered by the repository itself: each
+supported system has a provider package under ``src/sqlancer``. Reading the git
+tree makes the answer deterministic and self-updating, and the tree entry is
+recorded as the evidence.
+
+Display names and homepages cannot be derived from a directory name, so they
+come from a curated table below. A provider with no table entry still lands in
+the registry with a title-cased name, so a newly added DBMS is never silently
+dropped -- it just gets a follow-up nudge in the pull request.
+"""
+
+from __future__ import annotations
+
+from typing import Dict, List, Optional
+
+from .. import config
+from ..github import GitHub
+from ..util import content_hash, now, slugify
+
+COLLECTOR = "dbms_registry"
+COLLECTOR_VERSION = "1.0.0"
+
+SQLANCER_OWNER = "sqlancer"
+SQLANCER_REPO = "sqlancer"
+PROVIDER_ROOT = "src/sqlancer"
+
+# Directories under src/sqlancer that are shared infrastructure, not a DBMS.
+NON_PROVIDER_DIRS = {"common", "transformations"}
+
+# Curated presentation metadata, keyed by provider directory name.
+KNOWN: Dict[str, Dict[str, Optional[str]]] = {
+    "citus": {"name": "Citus", "url": "https://www.citusdata.com/",
+              "repository": "https://github.com/citusdata/citus"},
+    "clickhouse": {"name": "ClickHouse", "url": "https://clickhouse.com/",
+                   "repository": "https://github.com/ClickHouse/ClickHouse"},
+    "cockroachdb": {"name": "CockroachDB", "url": "https://www.cockroachlabs.com/",
+                    "repository": "https://github.com/cockroachdb/cockroach"},
+    "databend": {"name": "Databend", "url": "https://www.databend.com/",
+                 "repository": "https://github.com/datafuselabs/databend",
+                 "github_owners": ["datafuse-extras", "databendlabs",
+                                   "datafuselabs"]},
+    "datafusion": {"name": "Apache DataFusion",
+                   "url": "https://datafusion.apache.org/",
+                   "repository": "https://github.com/apache/datafusion"},
+    "doris": {"name": "Apache Doris", "url": "https://doris.apache.org/",
+              "repository": "https://github.com/apache/doris"},
+    # DuckDB's CI robot files what its fuzzers find in a repository of its
+    # own, so that a failure does not break the build -- the keynote at DBTest
+    # '22 describes it. Those reports are DuckDB bugs and belong to DuckDB.
+    "duckdb": {"name": "DuckDB", "url": "https://duckdb.org/",
+               "repository": "https://github.com/duckdb/duckdb",
+               # DuckDB was cwida/duckdb until 2021, and those links still
+               # redirect -- so the same bug arrives under two addresses.
+               "github_owners": ["cwida"],
+               "github_repositories": ["duckdb/duckdb-fuzzer",
+                                       "duckdblabs/duckdb-fuzzer-ci"]},
+    "h2": {"name": "H2", "url": "https://h2database.com/",
+           "repository": "https://github.com/h2database/h2database"},
+    "hive": {"name": "Apache Hive", "url": "https://hive.apache.org/",
+             "repository": "https://github.com/apache/hive"},
+    "hsqldb": {"name": "HSQLDB", "url": "https://hsqldb.org/", "repository": None},
+    "mariadb": {"name": "MariaDB", "url": "https://mariadb.org/",
+                "repository": "https://github.com/MariaDB/server"},
+    "materialize": {"name": "Materialize", "url": "https://materialize.com/",
+                    "repository": "https://github.com/MaterializeInc/materialize"},
+    "mysql": {"name": "MySQL", "url": "https://www.mysql.com/",
+              "repository": "https://github.com/mysql/mysql-server"},
+    "oceanbase": {"name": "OceanBase", "url": "https://www.oceanbase.com/",
+                  "repository": "https://github.com/oceanbase/oceanbase"},
+    "postgres": {"name": "PostgreSQL", "url": "https://www.postgresql.org/",
+                 "repository": "https://github.com/postgres/postgres",
+                 "id": "postgresql"},
+    "presto": {"name": "Presto", "url": "https://prestodb.io/",
+               "repository": "https://github.com/prestodb/presto"},
+    "questdb": {"name": "QuestDB", "url": "https://questdb.io/",
+                "repository": "https://github.com/questdb/questdb"},
+    "spark": {"name": "Apache Spark", "url": "https://spark.apache.org/",
+              "repository": "https://github.com/apache/spark"},
+    "sqlite3": {"name": "SQLite", "url": "https://sqlite.org/",
+                "repository": "https://github.com/sqlite/sqlite", "id": "sqlite"},
+    "tidb": {"name": "TiDB", "url": "https://www.pingcap.com/tidb/",
+             "repository": "https://github.com/pingcap/tidb"},
+    "yugabyte": {"name": "YugabyteDB", "url": "https://www.yugabyte.com/",
+                 "repository": "https://github.com/yugabyte/yugabyte-db",
+                 "id": "yugabytedb"},
+}
+
+# Systems that appear in bug or adoption records without a provider package in
+# the main repository -- historic providers, and systems reached through
+# SQLancer++ rather than a bespoke implementation.
+ADDITIONAL: List[Dict[str, Optional[str]]] = [
+    {"id": "tdengine", "name": "TDengine", "url": "https://tdengine.com/",
+     "repository": "https://github.com/taosdata/TDengine",
+     "aliases": ["TDEngine", "tdengine", "TDengine"]},
+    {"id": "monetdb", "name": "MonetDB", "url": "https://www.monetdb.org/",
+     "repository": "https://github.com/MonetDB/MonetDB", "aliases": ["MonetDB"]},
+    {"id": "umbra", "name": "Umbra", "url": "https://umbra-db.com/",
+     "repository": None, "aliases": ["Umbra"]},
+    {"id": "dolt", "name": "Dolt", "url": "https://www.dolthub.com/",
+     "repository": "https://github.com/dolthub/dolt", "aliases": ["Dolt"]},
+    {"id": "cratedb", "name": "CrateDB", "url": "https://cratedb.com/",
+     "repository": "https://github.com/crate/crate",
+     "aliases": ["cratedb", "Crate", "CrateDB"]},
+    {"id": "risingwave", "name": "RisingWave", "url": "https://risingwave.com/",
+     "repository": "https://github.com/risingwavelabs/risingwave",
+     "aliases": ["risingwave", "RisingWave"]},
+    {"id": "neo4j", "name": "Neo4j", "url": "https://neo4j.com/",
+     "repository": "https://github.com/neo4j/neo4j", "aliases": ["neo4j", "Neo4j"]},
+    {"id": "firebird", "name": "Firebird", "url": "https://firebirdsql.org/",
+     "repository": "https://github.com/FirebirdSQL/firebird",
+     "aliases": ["Firebird"]},
+    {"id": "virtuoso", "name": "Virtuoso", "url": "https://virtuoso.openlinksw.com/",
+     "repository": "https://github.com/openlink/virtuoso-opensource",
+     "aliases": ["Virtuoso"]},
+    {"id": "redisgraph", "name": "RedisGraph", "url": "https://redis.io/",
+     "repository": "https://github.com/RedisGraph/RedisGraph",
+     "aliases": ["redisgraph", "RedisGraph"]},
+    {"id": "agensgraph", "name": "AgensGraph", "url": "https://bitnine.net/",
+     "repository": "https://github.com/bitnine-oss/agensgraph",
+     "aliases": ["agensgraph", "AgensGraph"]},
+    {"id": "percona", "name": "Percona Server", "url": "https://www.percona.com/",
+     "repository": "https://github.com/percona/percona-server",
+     "aliases": ["Percona"]},
+    {"id": "cubrid", "name": "CUBRID", "url": "https://www.cubrid.org/",
+     "repository": "https://github.com/CUBRID/cubrid", "aliases": ["CUBRID"]},
+    {"id": "arangodb", "name": "ArangoDB", "url": "https://arangodb.com/",
+     "repository": "https://github.com/arangodb/arangodb", "aliases": ["ArangoDB"]},
+    {"id": "mongodb", "name": "MongoDB", "url": "https://www.mongodb.com/",
+     "repository": "https://github.com/mongodb/mongo", "aliases": ["MongoDB"]},
+    {"id": "cnosdb", "name": "CnosDB", "url": "https://www.cnosdb.com/",
+     "repository": "https://github.com/cnosdb/cnosdb", "aliases": ["CnosDB"]},
+    # Systems whose own team maintains a SQLancer fork, reached through the
+    # fork list rather than through a provider package.
+    # Systems reached only through their issue trackers: SQLancer has no
+    # provider for them, but their bugs name it.
+    {"id": "wadjet", "name": "Wadjet", "url": "https://github.com/derekmwright/wadjet",
+     "repository": "https://github.com/derekmwright/wadjet", "aliases": ["Wadjet"]},
+    {"id": "elasticsearch", "name": "Elasticsearch", "url": "https://www.elastic.co/",
+     "repository": "https://github.com/elastic/elasticsearch",
+     "aliases": ["Elasticsearch"]},
+    {"id": "opensearch", "name": "OpenSearch SQL", "url": "https://opensearch.org/",
+     "repository": "https://github.com/opensearch-project/sql",
+     "aliases": ["OpenSearch", "OpenSearch SQL"]},
+    {"id": "tikv", "name": "TiKV", "url": "https://tikv.org/",
+     "repository": "https://github.com/tikv/tikv", "aliases": ["TiKV"]},
+    {"id": "sparq", "name": "sparq", "url": "https://github.com/sparq-org/sparq",
+     "repository": "https://github.com/sparq-org/sparq", "aliases": ["sparq"]},
+    {"id": "serenedb", "name": "SereneDB", "url": "https://github.com/serenedb/serenedb",
+     "repository": "https://github.com/serenedb/serenedb", "aliases": ["SereneDB"]},
+    {"id": "kyzo", "name": "Kyzo", "url": "https://github.com/kyzobuild/kyzo",
+     "repository": "https://github.com/kyzobuild/kyzo", "aliases": ["Kyzo"]},
+    {"id": "greptimedb", "name": "GreptimeDB", "url": "https://greptime.com/",
+     "repository": "https://github.com/GreptimeTeam/greptimedb",
+     "aliases": ["GreptimeDB"]},
+    {"id": "defradb", "name": "DefraDB", "url": "https://docs.source.network/",
+     "repository": "https://github.com/sourcenetwork/defradb.rs",
+     "github_owners": ["sourcenetwork"], "aliases": ["DefraDB"]},
+    {"id": "bharatdbms", "name": "BharatDBMS",
+     "url": "https://github.com/BharatDBPG/BharatDBMS-PG",
+     "repository": "https://github.com/BharatDBPG/BharatDBMS-PG",
+     "aliases": ["BharatDBMS"]},
+    {"id": "stonedb", "name": "StoneDB", "url": "https://stonedb.io/",
+     "repository": "https://github.com/stoneatom/stonedb", "aliases": ["StoneDB"]},
+    {"id": "turso", "name": "Turso", "url": "https://turso.tech/",
+     "repository": "https://github.com/tursodatabase/turso",
+     "github_owners": ["tursodatabase"], "aliases": ["Turso", "limbo"]},
+    {"id": "seekdb", "name": "SeekDB", "url": "https://www.oceanbase.com/",
+     "repository": "https://github.com/oceanbase/seekdb", "aliases": ["SeekDB"]},
+    {"id": "falkordb", "name": "FalkorDB", "url": "https://www.falkordb.com/",
+     "repository": "https://github.com/FalkorDB/FalkorDB", "aliases": ["FalkorDB"]},
+    {"id": "readyset", "name": "ReadySet", "url": "https://readyset.io/",
+     "repository": "https://github.com/readysettech/readyset",
+     "aliases": ["ReadySet", "readyset"]},
+    {"id": "noisepage", "name": "NoisePage", "url": "https://noise.page/",
+     "repository": "https://github.com/cmu-db/noisepage", "aliases": ["NoisePage"]},
+    {"id": "tarantool", "name": "Tarantool", "url": "https://www.tarantool.io/",
+     "repository": "https://github.com/tarantool/tarantool",
+     "aliases": ["Tarantool"]},
+    {"id": "spiceai", "name": "Spice.ai OSS", "url": "https://spice.ai/",
+     "repository": "https://github.com/spiceai/spiceai", "aliases": ["Spice.ai", "spiceai"]},
+    {"id": "cloudberry", "name": "Apache Cloudberry",
+     "url": "https://cloudberry.apache.org/",
+     "repository": "https://github.com/apache/cloudberry",
+     "aliases": ["Cloudberry", "Apache Cloudberry", "cloudberrydb"]},
+    {"id": "greenplum", "name": "Greenplum", "url": "https://greenplum.org/",
+     "repository": "https://github.com/greenplum-db/gpdb",
+     "aliases": ["Greenplum", "gpdb"]},
+    {"id": "starrocks", "name": "StarRocks", "url": "https://www.starrocks.io/",
+     "repository": "https://github.com/StarRocks/starrocks",
+     "aliases": ["StarRocks"]},
+    {"id": "opengauss", "name": "openGauss", "url": "https://opengauss.org/",
+     "repository": "https://github.com/opengauss-mirror/openGauss-server",
+     "aliases": ["openGauss", "opengauss"],
+     # openGauss develops on Gitee; GitHub carries a mirror. The bug search
+     # needs the Gitee repository, and this is the only place that survives a
+     # regeneration of the registry.
+     "gitee_repositories": ["opengauss/openGauss-server"]},
+    {"id": "ydb", "name": "YDB", "url": "https://ydb.tech/",
+     "repository": "https://github.com/ydb-platform/ydb", "aliases": ["YDB"]},
+    {"id": "hazelcast", "name": "Hazelcast", "url": "https://hazelcast.com/",
+     "repository": "https://github.com/hazelcast/hazelcast",
+     "aliases": ["Hazelcast"]},
+    {"id": "feldera", "name": "Feldera", "url": "https://feldera.com/",
+     "repository": "https://github.com/feldera/feldera", "aliases": ["Feldera"]},
+    {"id": "xugu", "name": "XuGu", "url": "https://www.xugudb.com/",
+     "repository": "https://github.com/Xugu-Open-Source/xugu",
+     "aliases": ["XuGu", "Xugu"]},
+]
+
+
+def _aliases(directory: str, name: str, record_id: str) -> List[str]:
+    seen, out = set(), []
+    for candidate in (name, directory, record_id, name.replace(" ", "")):
+        lowered = candidate.lower()
+        if lowered not in seen:
+            seen.add(lowered)
+            out.append(candidate)
+    return out
+
+
+def collect(gh: GitHub, timestamp: Optional[str] = None) -> List[dict]:
+    """Return DBMS registry records derived from the SQLancer repository."""
+    timestamp = timestamp or now()
+    tree = gh.tree(SQLANCER_OWNER, SQLANCER_REPO, "main")
+    tree_url = (f"https://github.com/{SQLANCER_OWNER}/{SQLANCER_REPO}/tree/"
+                f"main/{PROVIDER_ROOT}")
+
+    providers = sorted({
+        entry["path"].split("/")[2]
+        for entry in tree
+        if entry.get("type") == "tree"
+        and entry["path"].startswith(PROVIDER_ROOT + "/")
+        and entry["path"].count("/") == 2
+        and entry["path"].split("/")[2] not in NON_PROVIDER_DIRS
+    })
+
+    records: List[dict] = []
+    for directory in providers:
+        meta = KNOWN.get(directory, {})
+        name = meta.get("name") or directory.replace("_", " ").title()
+        record_id = meta.get("id") or slugify(directory)
+        provider_path = f"{PROVIDER_ROOT}/{directory}"
+        records.append({
+            "id": record_id,
+            "name": name,
+            "aliases": _aliases(directory, name, record_id),
+            "url": meta.get("url"),
+            "repository": meta.get("repository"),
+            "github_owners": meta.get("github_owners") or [],
+            "github_repositories": meta.get("github_repositories") or [],
+            "gitee_repositories": meta.get("gitee_repositories") or [],
+            "supported_by_sqlancer": True,
+            "support": {
+                "provider_path": provider_path,
+                "umbrella_tool": "sqlancer",
+                "evidence": [{
+                    "source_url": (f"https://github.com/{SQLANCER_OWNER}/"
+                                   f"{SQLANCER_REPO}/tree/main/{provider_path}"),
+                    "source_type": "github_repository",
+                    "excerpt": None,
+                    "note": (f"The SQLancer repository contains a testing "
+                             f"implementation at {provider_path}."),
+                    "content_sha256": content_hash(provider_path),
+                    "retrieved_at": timestamp,
+                    "first_seen": timestamp,
+                    "last_verified": timestamp,
+                }],
+            },
+        })
+
+    known_ids = {record["id"] for record in records}
+    for extra in ADDITIONAL:
+        if extra["id"] in known_ids:
+            continue
+        records.append({
+            "id": extra["id"],
+            "name": extra["name"],
+            "aliases": extra.get("aliases") or [extra["name"]],
+            "url": extra.get("url"),
+            "repository": extra.get("repository"),
+            "github_owners": extra.get("github_owners") or [],
+            "github_repositories": extra.get("github_repositories") or [],
+            "gitee_repositories": extra.get("gitee_repositories") or [],
+            "supported_by_sqlancer": False,
+            "support": None,
+            "notes": ("Referenced by impact records without a provider package "
+                      "in the current main SQLancer repository."),
+        })
+
+    records.sort(key=lambda r: r["id"])
+    return records
+
+
+def drop_nulls(record: dict) -> dict:
+    """Remove optional keys whose value is None, which the schemas disallow."""
+    cleaned = {}
+    for key, value in record.items():
+        if value is None and key in ("url", "repository", "notes"):
+            continue
+        if key in ("github_owners", "github_repositories",
+                   "gitee_repositories") and not value:
+            continue
+        cleaned[key] = value
+    return cleaned
+
+
+def build_payload(records: List[dict]) -> dict:
+    return {
+        "schema_version": "1.0.0",
+        "dbms": [drop_nulls(record) for record in records],
+    }
+
+
+def main() -> int:
+    from ..cache import Caches
+    caches = Caches()
+    gh = GitHub(caches.github)
+    records = collect(gh)
+    changed = config.write_json(config.DATA_FILES["dbms"], build_payload(records))
+    supported = sum(1 for r in records if r["supported_by_sqlancer"])
+    print(f"dbms.json: {'updated' if changed else 'unchanged'} "
+          f"({len(records)} systems, {supported} supported by SQLancer)")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/collectors/forks.py b/tools/impact/collectors/forks.py
new file mode 100644
index 0000000..86d9f9b
--- /dev/null
+++ b/tools/impact/collectors/forks.py
@@ -0,0 +1,185 @@
+"""Reads the fork list of the SQLancer repository.
+
+Forks are the one place GitHub's code search cannot reach -- it excludes them
+entirely -- so a database system that maintains its own SQLancer fork is
+invisible to every other collector here. Several do, which makes this the
+highest-yield adoption source of the lot.
+
+A fork under a project's own organisation is a deliberate act by that project's
+team, so it is admitted as adoption evidence. Whether the project went on to
+modify SQLancer is checked rather than assumed: the fork is compared against
+upstream, and the number of commits it is ahead by is recorded as part of the
+evidence.
+
+Forks also feed the artifact matching used for papers, since a renamed fork is
+often a research artifact.
+"""
+
+from __future__ import annotations
+
+from typing import Dict, List, Optional, Tuple
+
+from .. import config
+from ..github import GitHub
+from ..util import now, truncate
+
+COLLECTOR = "forks"
+COLLECTOR_VERSION = "1.0.0"
+
+UPSTREAM_OWNER, UPSTREAM_REPO = "sqlancer", "sqlancer"
+FORKS_URL = f"https://github.com/{UPSTREAM_OWNER}/{UPSTREAM_REPO}/forks"
+
+MAX_PAGES = 8
+
+
+def list_forks(gh: GitHub, *, max_age_days: int = 3) -> List[dict]:
+    """Every fork of the SQLancer repository."""
+    forks: List[dict] = []
+    for page in range(1, MAX_PAGES + 1):
+        data, _ = gh.get(f"/repos/{UPSTREAM_OWNER}/{UPSTREAM_REPO}/forks",
+                         params={"per_page": 100, "page": page, "sort": "oldest"},
+                         max_age_days=max_age_days)
+        if not data:
+            break
+        forks.extend(data)
+        if len(data) < 100:
+            break
+    return forks
+
+
+def _owner_index() -> Dict[str, dict]:
+    """Map every GitHub organisation a registered DBMS operates under to it."""
+    registry = config.load_json(config.DATA_FILES["dbms"])["dbms"]
+    index: Dict[str, dict] = {}
+    for entry in registry:
+        owners = list(entry.get("github_owners") or [])
+        repository = entry.get("repository")
+        if repository:
+            parts = repository.rstrip("/").split("/")
+            if len(parts) >= 2:
+                owners.append(parts[-2])
+        for owner in owners:
+            index[owner.lower()] = entry
+    return index
+
+
+def commits_ahead(gh: GitHub, owner: str, repo: str,
+                  branch: Optional[str]) -> Optional[int]:
+    """How many commits a fork is ahead of upstream, or None if unknown.
+
+    A fork nobody touched is a bookmark; one with commits on top is a project
+    doing something with SQLancer. The distinction belongs in the evidence.
+    """
+    if not branch:
+        return None
+    data, _ = gh.get(
+        f"/repos/{UPSTREAM_OWNER}/{UPSTREAM_REPO}/compare/"
+        f"{UPSTREAM_OWNER}:main...{owner}:{branch}",
+        max_age_days=14)
+    if not isinstance(data, dict):
+        return None
+    ahead = data.get("ahead_by")
+    return ahead if isinstance(ahead, int) else None
+
+
+def adoption_records(gh: GitHub, forks: Optional[List[dict]] = None, *,
+                     timestamp: Optional[str] = None
+                     ) -> Tuple[List[dict], List[dict]]:
+    """Return ``(records, needs_review)`` for forks under a DBMS project's org.
+
+    Forks under an organisation this registry does not know are reported rather
+    than guessed at: several are database systems SQLancer has no provider for,
+    and adding one is a registry decision, not something to infer from a fork.
+    """
+    timestamp = timestamp or now()
+    forks = list_forks(gh) if forks is None else forks
+    owners = _owner_index()
+
+    records: List[dict] = []
+    needs_review: List[dict] = []
+
+    for fork in forks:
+        login = ((fork.get("owner") or {}).get("login") or "")
+        entry = owners.get(login.lower())
+        full_name = fork.get("full_name") or ""
+        url = fork.get("html_url") or f"https://github.com/{full_name}"
+
+        if entry is None:
+            if (fork.get("owner") or {}).get("type") == "Organization":
+                needs_review.append({
+                    "kind": "adoption",
+                    "url": url,
+                    "reason": (f"organisation {login!r} maintains a SQLancer "
+                               f"fork but is not a registered database system"),
+                })
+            continue
+
+        owner, repo = full_name.split("/", 1)
+        ahead = commits_ahead(gh, owner, repo, fork.get("default_branch"))
+        pushed = (fork.get("pushed_at") or "")[:10]
+
+        detail = f"The {entry['name']} project maintains its own SQLancer fork"
+        if ahead:
+            detail += f", {ahead} commit(s) ahead of upstream"
+        if pushed:
+            detail += f", last updated {pushed}"
+        detail += "."
+
+        records.append({
+            "id": f"adoption:{entry['id']}:official_testing",
+            "dbms": entry["id"],
+            "relationship": "official_testing",
+            "finder": "sqlancer",
+            "summary": truncate(detail, 600),
+            "since_year": None,
+            "active": not fork.get("archived", False),
+            "evidence": [{
+                "source_url": url,
+                "source_type": "github_repository",
+                "excerpt": None,
+                "note": truncate(
+                    f"{full_name} is a fork of {UPSTREAM_OWNER}/"
+                    f"{UPSTREAM_REPO} under the {entry['name']} project's own "
+                    f"GitHub organisation"
+                    + (f", {ahead} commit(s) ahead of upstream" if ahead else "")
+                    + ".", 500),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }],
+            "provenance": {
+                "collector": COLLECTOR,
+                "collector_version": COLLECTOR_VERSION,
+                "policy_version": config.policy_version(),
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+                "source_url": FORKS_URL,
+                "source_type": "github_repository",
+            },
+        })
+        if entry.get("repository"):
+            records[-1]["project_repository"] = entry["repository"]
+
+    return records, needs_review
+
+
+def artifact_candidates(forks: Optional[List[dict]] = None,
+                        gh: Optional[GitHub] = None) -> List[str]:
+    """Forks worth inspecting as research artifacts.
+
+    A fork still called ``sqlancer`` is almost always a working copy; one that
+    was renamed usually became something with a name of its own. Renaming is the
+    cheap signal that separates the two, and it keeps this from returning all
+    several hundred forks.
+    """
+    if forks is None:
+        forks = list_forks(gh) if gh is not None else []
+    candidates = []
+    for fork in forks:
+        name = (fork.get("name") or "").lower()
+        full_name = fork.get("full_name") or ""
+        if not full_name:
+            continue
+        if name not in ("sqlancer", "sqlancer-sqlancer"):
+            candidates.append(full_name)
+    return candidates
diff --git a/tools/impact/collectors/fulltext.py b/tools/impact/collectors/fulltext.py
new file mode 100644
index 0000000..7cd0867
--- /dev/null
+++ b/tools/impact/collectors/fulltext.py
@@ -0,0 +1,375 @@
+"""Reads a paper's full text and finds the sentences that decide its category.
+
+Until now every judgement about a paper rested on citation contexts -- the one
+or two sentences a scholarly index happens to have extracted around a citation.
+That is a thin basis, and for most of this corpus there was nothing else: the
+venues are largely ACM, and dl.acm.org refuses automated requests.
+
+An arXiv preprint sidesteps that, and the full text is a far better source than
+a handful of contexts. What a paper did with SQLancer is usually stated plainly
+somewhere in it -- "we implemented our approach on top of SQLancer", "we compare
+against TLP and NoREC" -- and this module finds those sentences and keeps them
+verbatim.
+
+The patterns here are deliberately narrow. They match statements a paper makes
+about its own work, not descriptions of related work, and anything less direct
+is left for the classifier rather than guessed at.
+"""
+
+from __future__ import annotations
+
+import io
+import pathlib
+import re
+from typing import Dict, List, Optional, Sequence, Tuple
+
+from .. import taxonomy
+from ..http import Fetcher
+from ..util import content_hash, now, truncate
+
+COLLECTOR = "fulltext"
+COLLECTOR_VERSION = "1.0.0"
+EXTRACTOR_VERSION = "fulltext-signals-v1"
+
+MAX_PDF_BYTES = 25_000_000
+MAX_TEXT_CHARS = 400_000
+EXCERPT_LIMIT = 700
+
+# A sentence, roughly. Papers are full of abbreviations and citation markers, so
+# this splits on terminators followed by a capital rather than on any full stop.
+SENTENCE_SPLIT = re.compile(r"(?<=[.!?])\s+(?=[A-Z(])")
+
+_TOOL = r"(?:SQLancer\+\+|SQLancer|ShQveL)"
+_TECH = r"(?:TLP|NoREC|PQS|QPG|CERT|DQP|CODDTest|Ternary Logic Partitioning" \
+        r"|Non-optimizing Reference Engine Construction|Pivoted Query Synthesis" \
+        r"|Query Plan Guidance|query partitioning)"
+
+# "We built our tool on SQLancer." Statements about the authors' own artifact.
+# `we` or `our` is required: without it "the WHERE Extended case of TLP ... from
+# SQLancer" -- a figure caption -- reads as a claim of reuse. And the
+# prepositions are limited to ones that mean "built on": "from" and "with" match
+# far too much ("a query from SQLancer's generator").
+USES_INFRASTRUCTURE = re.compile(
+    rf"\b(?:we|our)\b[^.]{{0,120}}?"
+    rf"\b(?:implemented|implements|implement|built|build|developed|develops|"
+    rf"prototyped|prototype|based|extended|extends|modified|integrated)\b"
+    rf"[^.]{{0,80}}?\b(?:on\s+top\s+of|upon|on|in|into|using)\s+"
+    rf"(?:the\s+)?{_TOOL}\b",
+    re.IGNORECASE)
+
+# The same claim written the other way round.
+USES_INFRASTRUCTURE_ALT = re.compile(
+    rf"\b(?:is|are|was|were)\s+(?:implemented|built|developed|based)\b"
+    rf"[^.]{{0,60}}?\b(?:on\s+top\s+of|upon|on|in)\s+(?:the\s+)?{_TOOL}\b",
+    re.IGNORECASE)
+
+# "We compare X against TLP and NoREC." An empirical comparison the paper ran.
+COMPARES_WITH = re.compile(
+    rf"\b(?:we|our)\b[^.]{{0,140}}?"
+    rf"\b(?:compare[ds]?|comparison|evaluat\w+|benchmark\w*|"
+    rf"baseline[s]?|against)\b[^.]{{0,140}}?\b(?:{_TOOL}|{_TECH})\b",
+    re.IGNORECASE)
+
+COMPARES_WITH_ALT = re.compile(
+    rf"\b(?:{_TOOL}|{_TECH})\b[^.]{{0,80}}?\bas\s+(?:a\s+|our\s+|the\s+)?"
+    rf"baseline[s]?\b",
+    re.IGNORECASE)
+
+# "We extend TLP to graph databases." A claimed contribution over a technique.
+EXTENDS_TECHNIQUE = re.compile(
+    rf"\b(?:we|our)\b[^.]{{0,120}}?"
+    rf"\b(?:extend(?:ed|s)?|generali[sz]\w+|adapt(?:ed|s)?|build[s]?\s+(?:up)?on|"
+    rf"inspired\s+by|derive[ds]?\s+from|improve[sd]?\s+(?:up)?on)\b"
+    rf"[^.]{{0,80}}?\b(?:{_TECH}|{_TOOL})\b",
+    re.IGNORECASE)
+
+# Wording that marks a sentence as about someone else's work, not the authors'.
+RELATED_WORK = re.compile(
+    r"\b(?:previous|prior|existing|earlier|related)\s+(?:work|approaches|"
+    r"studies|techniques|tools)\b|\bhas\s+been\s+(?:proposed|shown)\b",
+    re.IGNORECASE)
+
+# "We adapted SQLancer as a baseline for comparison" matches the extension
+# pattern on "adapted ... SQLancer" and means the opposite: the tool was taken
+# up in order to be measured against, not built upon. The rest of the sentence
+# is what settles it, so a sentence that says baseline is a comparison only.
+AS_A_BASELINE = re.compile(
+    r"\b(?:as\s+(?:a\s+|our\s+|the\s+)?baselines?|for\s+comparison|"
+    r"to\s+compare\s+(?:against|with)|as\s+(?:a\s+|our\s+)?"
+    r"(?:comparison|reference)\s+(?:point|tool|system)?)\b",
+    re.IGNORECASE)
+
+SIGNALS = {
+    "uses_infrastructure": (USES_INFRASTRUCTURE, USES_INFRASTRUCTURE_ALT),
+    "compares_with": (COMPARES_WITH, COMPARES_WITH_ALT),
+    "extends_technique": (EXTENDS_TECHNIQUE,),
+}
+
+# A relationship that a sentence cannot claim once it says "as a baseline".
+NOT_WHEN_BASELINE = ("extends_technique", "uses_infrastructure")
+
+
+# PVLDB puts every paper on vldb.org, free, at a URL that can be derived from
+# the volume, the first page and the first author's surname -- all of which
+# OpenAlex records. It is the one major venue in this corpus whose full text is
+# reachable without an arXiv preprint.
+PVLDB_DOI_PREFIX = "10.14778/"
+PVLDB_FIRST_VOLUME_YEAR = 2007
+
+
+def pvldb_pdf_url(doi: Optional[str], work: Optional[dict]) -> Optional[str]:
+    """The vldb.org PDF for a PVLDB paper, or None."""
+    if not doi or not doi.lower().startswith(PVLDB_DOI_PREFIX) or not work:
+        return None
+    biblio = work.get("biblio") or {}
+    first_page = (biblio.get("first_page") or "").strip()
+    volume = (biblio.get("volume") or "").strip()
+    if not volume:
+        year = work.get("publication_year")
+        volume = str(year - PVLDB_FIRST_VOLUME_YEAR) if year else ""
+    authorships = work.get("authorships") or []
+    surname = ""
+    if authorships:
+        name = ((authorships[0].get("author") or {}).get("display_name") or "")
+        parts = [part for part in name.split() if part]
+        surname = parts[-1].lower() if parts else ""
+    if not (first_page and volume and surname):
+        return None
+    return (f"https://www.vldb.org/pvldb/vol{volume}/"
+            f"p{first_page}-{surname}.pdf")
+
+
+# USENIX publishes every paper openly, and serves the PDF from a path built
+# from the venue, the year and the first author. Worth deriving: seven papers in
+# this corpus are USENIX, and because USENIX assigns no DOI they are exactly the
+# records that fall back to a Semantic Scholar id and end up on the hand-download
+# list -- SQLRight, Pinolo, DynSQL, WingFuzz among them.
+USENIX_SLUGS = {
+    r"security": ("usenixsecurity", "sec"),
+    r"annual technical|atc": ("atc",),
+    r"operating systems design|osdi": ("osdi",),
+    r"networked systems design|nsdi": ("nsdi",),
+    r"file and storage|fast": ("fast",),
+}
+
+
+def usenix_pdf_urls(paper: dict) -> List[str]:
+    """Candidate USENIX URLs for a paper, best first.
+
+    Several, because the naming is not quite consistent: USENIX Security was
+    "sec22" and became "usenixsecurity24", and a paper's file is sometimes named
+    for its first author alone and sometimes with a word of the title after it.
+    Each candidate is a cheap conditional request, and a wrong guess simply
+    404s.
+    """
+    venue = (paper.get("venue") or "").lower()
+    if "usenix" not in venue and "osdi" not in venue:
+        return []
+    year = paper.get("year")
+    authors = paper.get("authors") or []
+    if not year or not authors:
+        return []
+
+    slugs: List[str] = []
+    for pattern, names in USENIX_SLUGS.items():
+        if re.search(pattern, venue):
+            slugs.extend(names)
+    if not slugs:
+        return []
+
+    surname = re.sub(r"[^a-z]", "", (authors[0].split()[-1] if authors[0] else "").lower())
+    if not surname:
+        return []
+    # A paper published at the conference of year N appears under N, but a
+    # winter deadline can put it under N-1.
+    short_years = [str(year)[-2:], str(year - 1)[-2:]]
+    first_word = ""
+    for word in re.sub(r"[^A-Za-z ]", " ", paper.get("title") or "").split():
+        if len(word) > 3 and word.lower() not in ("with", "from", "that", "this"):
+            first_word = word.lower()
+            break
+
+    # Where two papers by different people would collide, USENIX disambiguates
+    # with the author's given name rather than a word of the title.
+    given = re.sub(r"[^a-z-]", "", (authors[0].split()[0] if authors[0] else "").lower())
+
+    urls: List[str] = []
+    for slug in slugs:
+        for short in short_years:
+            stem = f"{slug}{short}-{surname}"
+            urls.append(f"https://www.usenix.org/system/files/{stem}.pdf")
+            for suffix in (given, first_word):
+                if suffix and suffix != surname:
+                    urls.append(
+                        f"https://www.usenix.org/system/files/{stem}-{suffix}.pdf")
+    return urls
+
+
+def local_pdf(doi: Optional[str], arxiv_id: Optional[str],
+              s2_paper_id: Optional[str] = None) -> Optional[str]:
+    """A PDF supplied by hand, for papers no free route reaches.
+
+    Two thirds of this corpus is published by ACM, IEEE or Springer, all of
+    which refuse automated fetching, so citation contexts are otherwise the
+    only evidence for them. Dropping a PDF into the papers directory -- named
+    for its DOI with slashes replaced, or its arXiv id -- lets the same
+    extraction run over it, and every resulting claim still quotes the paper
+    verbatim.
+    """
+    from .. import config
+
+    directory = config.REPO_ROOT / ".cache" / "impact" / "papers-pdf"
+    if not directory.is_dir():
+        return None
+    names = []
+    if doi:
+        names.append(doi.replace("/", "_"))
+    if arxiv_id:
+        names.append(arxiv_id)
+    if s2_paper_id:
+        names.append(f"s2_{s2_paper_id}")
+    for name in names:
+        for candidate in (directory / f"{name}.pdf", directory / f"{name}.PDF"):
+            if candidate.is_file():
+                return candidate.as_uri()
+    return None
+
+
+def wanted_filename(doi: Optional[str], arxiv_id: Optional[str],
+                    s2_paper_id: Optional[str] = None) -> Optional[str]:
+    """The name a hand-supplied PDF must have for the pipeline to find it.
+
+    The Semantic Scholar id is the last resort, and it is needed: a dozen papers
+    here have neither a DOI nor an arXiv id, and two of them are the ones whose
+    citation sentences hint hardest at a relationship. Without a name they could
+    never be supplied at all, which is a poor reason to leave a paper unread.
+    """
+    if doi:
+        return f"{doi.replace('/', '_')}.pdf"
+    if arxiv_id:
+        return f"{arxiv_id}.pdf"
+    if s2_paper_id:
+        return f"s2_{s2_paper_id}.pdf"
+    return None
+
+
+def reachable_without_help(paper: dict) -> bool:
+    """Whether the pipeline can already read this paper's full text.
+
+    arXiv preprints, PVLDB and USENIX are all free; everything else in this
+    corpus is behind a publisher that refuses automated requests. USENIX
+    belongs here even though it assigns no DOI, which is what made its papers
+    look unreachable and put them on the hand-download list.
+    """
+    if paper.get("arxiv_id"):
+        return True
+    doi = (paper.get("doi") or "").lower()
+    if doi.startswith(PVLDB_DOI_PREFIX):
+        return True
+    return bool(usenix_pdf_urls(paper))
+
+
+def pdf_text(fetcher: Fetcher, url: str) -> Optional[str]:
+    """Extract the text of a PDF, or None if it cannot be read."""
+    if url.startswith("file://"):
+        try:
+            from urllib.request import url2pathname
+            from urllib.parse import urlparse
+            raw = pathlib.Path(url2pathname(urlparse(url).path)).read_bytes()
+        except Exception:
+            return None
+    else:
+        try:
+            raw, _ = fetcher.fetch_binary(url, max_age_days=365)
+        except Exception:
+            return None
+    if not raw or not raw.lstrip().startswith(b"%PDF") or len(raw) > MAX_PDF_BYTES:
+        return None
+    try:
+        import pypdf
+        reader = pypdf.PdfReader(io.BytesIO(raw))
+        text = "\n".join(page.extract_text() or "" for page in reader.pages)
+    except Exception:
+        return None
+    text = _strip_page_furniture(text)
+    # PDF extraction hyphenates across line breaks and scatters newlines.
+    text = re.sub(r"-\s*\n\s*", "", text)
+    text = re.sub(r"\s*\n\s*", " ", text)
+    text = re.sub(r"[ \t]{2,}", " ", text)
+    return truncate(text.strip(), MAX_TEXT_CHARS) or None
+
+
+# Stamps a publisher prints on every page. Extraction drops them into the
+# middle of whatever sentence spans the page break, so a quotation taken from
+# an IEEE or ACM PDF arrives with a licence notice inside it. They are removed
+# before anything is quoted -- an excerpt has to be the author's sentence, and
+# page furniture is not part of it.
+PAGE_FURNITURE = re.compile(
+    r"Authorized\s+licensed\s+use\s+limited\s+to.{0,200}?Restrictions\s+apply\s*\.?"
+    r"|Authorized\s+licensed\s+use\s+limited\s+to[^.]{0,200}\."
+    r"|Proc\.\s+ACM\s+\w+\.?\s*(?:Data|Manag)[^.]{0,80}?\."
+    r"|Permission\s+to\s+make\s+digital\s+or\s+hard\s+copies.{0,600}?fee\s*\."
+    r"|\d{4}\s+IEEE(?:\s+International)?[^.\n]{0,60}?(?:Conference|Symposium)[^.\n]{0,60}",
+    re.IGNORECASE | re.DOTALL)
+
+
+def _strip_page_furniture(text: str) -> str:
+    return PAGE_FURNITURE.sub(" ", text or "")
+
+
+def sentences(text: str) -> List[str]:
+    return [s.strip() for s in SENTENCE_SPLIT.split(text or "") if s.strip()]
+
+
+def find_signals(text: str, tax: Optional[taxonomy.Taxonomy] = None
+                 ) -> Dict[str, List[str]]:
+    """Sentences in ``text`` that state each relationship, verbatim.
+
+    A sentence framed as related work is skipped: "previous work implemented
+    this on top of SQLancer" says nothing about what these authors did.
+    """
+    tax = tax or taxonomy.load()
+    found: Dict[str, List[str]] = {key: [] for key in SIGNALS}
+    if not text:
+        return found
+
+    for sentence in sentences(text):
+        if len(sentence) > 900:
+            continue
+        if RELATED_WORK.search(sentence):
+            continue
+        # Excluded techniques never establish a SQLancer relationship.
+        if tax.find_excluded_techniques(sentence) and not tax.mentions_finder(sentence):
+            continue
+        baseline = bool(AS_A_BASELINE.search(sentence))
+        for key, patterns in SIGNALS.items():
+            if baseline and key in NOT_WHEN_BASELINE:
+                continue
+            if any(pattern.search(sentence) for pattern in patterns):
+                clean = truncate(re.sub(r"\s+", " ", sentence), EXCERPT_LIMIT)
+                if clean not in found[key]:
+                    found[key].append(clean)
+    return found
+
+
+def techniques_in(sentence: str, tax: Optional[taxonomy.Taxonomy] = None
+                  ) -> List[str]:
+    tax = tax or taxonomy.load()
+    return sorted({match.technique_id
+                   for match in tax.find_techniques(sentence)})
+
+
+def evidence_from(sentence: str, source_url: str, note: str,
+                  timestamp: Optional[str] = None) -> dict:
+    timestamp = timestamp or now()
+    return {
+        "source_url": source_url,
+        "source_type": "paper",
+        "excerpt": sentence,
+        "excerpt_is_verbatim": True,
+        "note": note,
+        "content_sha256": content_hash(sentence),
+        "retrieved_at": timestamp,
+        "first_seen": timestamp,
+        "last_verified": timestamp,
+    }
diff --git a/tools/impact/collectors/gitee_bugs.py b/tools/impact/collectors/gitee_bugs.py
new file mode 100644
index 0000000..40f3d24
--- /dev/null
+++ b/tools/impact/collectors/gitee_bugs.py
@@ -0,0 +1,275 @@
+"""Imports SQLancer bug reports from Gitee.
+
+Several database projects take their bug reports on gitee.com rather than on
+GitHub, so the GitHub search that reaches every other tracker cannot see them
+at all. openGauss is the case that showed why this matters: its tracker holds
+a report titled "TLP 等价验证出现结果内容不一致" over a schema beginning
+``CREATE UNLOGGED TABLE t0(c0 boolean PRIMARY KEY UNIQUE)``, which is a
+SQLancer oracle finding a bug and was invisible to every route the pipeline
+had.
+
+Attribution is the shared rule, unchanged. What the reports are written in is
+not: the sentence patterns are English, so a Chinese report clears the rules
+only through the two signals that are language-independent -- the tool's own
+name, and SQLancer's generated-schema fingerprint in the reproducer.
+
+Searching is by tool and oracle name only. The Chinese phrases for the
+symptoms these bugs produce -- 等价验证, 结果不一致, 模糊测试 -- return
+several hundred issues each from MindSpore, OpenHarmony and the kernel, and
+not one of them earns an attribution.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+import urllib.parse
+from typing import Dict, List, Optional, Sequence, Tuple
+
+from .. import config, taxonomy
+from ..http import Fetcher
+from ..util import (content_hash, normalize_url, now, parse_date, stable_digest,
+                    truncate, verbatim_excerpt, year_of)
+from .github_bugs import deterministic_attribution
+
+COLLECTOR = "gitee_bugs"
+COLLECTOR_VERSION = "1.0.0"
+
+API = "https://gitee.com/api/v5"
+SEARCH = f"{API}/search/issues"
+PAGE_SIZE = 100
+MAX_PAGES = 3
+
+SINGLE_TOKEN = re.compile(r"^[A-Za-z][\w+.-]*$")
+
+# Searches that came back non-200 or raised, so an empty result set can be
+# told apart from a throttled one.
+FAILED_SEARCHES: List[Tuple[str, Optional[str], object]] = []
+
+# Gitee blocks by address, not by endpoint: a burst of requests took the whole
+# API to 403 for hours, individual issue fetches included. Once several
+# searches in a row have failed there is nothing to be gained by continuing,
+# and something to lose.
+MAX_CONSECUTIVE_FAILURES = 3
+
+# Gitee's workflow states. "closed" means the work was completed; "rejected"
+# is the project saying this was not a defect.
+STATE_TO_STATUS = {
+    "closed": ("fixed", True),
+    "open": ("open", True),
+    "progressing": ("open", True),
+    "rejected": ("closed_not_a_bug", False),
+}
+
+
+def repository_index() -> Dict[str, str]:
+    """Map each registered project's Gitee repositories to its registry id."""
+    index: Dict[str, str] = {}
+    for entry in config.load_json(config.DATA_FILES["dbms"])["dbms"]:
+        for name in entry.get("gitee_repositories") or []:
+            index[name.strip().lower()] = entry["id"]
+    return index
+
+
+def search_terms(tax: taxonomy.Taxonomy) -> List[str]:
+    """Single-token tool and oracle names worth searching for."""
+    seen, ordered = set(), []
+    for name in tax.search_terms():
+        if SINGLE_TOKEN.match(name or "") and name.lower() not in seen:
+            seen.add(name.lower())
+            ordered.append(name)
+    return ordered
+
+
+def search(fetcher: Fetcher, term: str, *, repository: Optional[str] = None,
+           max_pages: int = MAX_PAGES) -> List[dict]:
+    """Every issue matching ``term``, in one repository or across Gitee.
+
+    Scoped to a repository by default, for the same reason the GitHub search
+    is: an unscoped query returns a page at a time out of far more matches
+    than it will page through, so a real report sits behind hundreds of
+    unrelated ones and is never reached. Searching "TLP" across Gitee returned
+    three hundred issues from throwaway repositories and not the openGauss one
+    this collector exists for.
+    """
+    issues: List[dict] = []
+    scope = ""
+    if repository and "/" in repository:
+        owner, _ = repository.split("/", 1)
+        # Gitee wants the full owner/name in ``repo``, not the bare name.
+        scope = (f"&owner={urllib.parse.quote(owner)}"
+                 f"&repo={urllib.parse.quote(repository)}")
+    for page in range(1, max_pages + 1):
+        url = (f"{SEARCH}?q={urllib.parse.quote(term)}"
+               f"&page={page}&per_page={PAGE_SIZE}{scope}")
+        try:
+            entry, _ = fetcher.fetch(url, max_age_days=7)
+        except Exception as error:
+            FAILED_SEARCHES.append((term, repository, str(error)))
+            return issues
+        if entry.get("status") != 200:
+            # Gitee throttles hard, and a throttled run returning nothing is
+            # indistinguishable from a tracker with nothing in it. Recorded so
+            # the caller can tell the two apart.
+            FAILED_SEARCHES.append((term, repository, entry.get("status")))
+            return issues
+        try:
+            batch = json.loads(entry["body"])
+        except ValueError:
+            return issues
+        if not isinstance(batch, list) or not batch:
+            return issues
+        issues.extend(batch)
+        if len(batch) < PAGE_SIZE:
+            return issues
+    return issues
+
+
+def _blocked() -> bool:
+    """Whether the last few searches all failed, meaning we are blocked."""
+    recent = FAILED_SEARCHES[-MAX_CONSECUTIVE_FAILURES:]
+    return (len(recent) == MAX_CONSECUTIVE_FAILURES
+            and all(status == 403 for _, _, status in recent))
+
+
+def as_issue(item: dict) -> dict:
+    """The shape ``deterministic_attribution`` reads, from a Gitee issue."""
+    return {
+        "title": item.get("title") or "",
+        "body": item.get("body") or "",
+        "labels": [{"name": (label or {}).get("name", "")}
+                   for label in (item.get("labels") or [])],
+        "html_url": item.get("html_url") or "",
+    }
+
+
+def source_text(item: dict) -> str:
+    return "\n".join(part for part in (item.get("title"), item.get("body"))
+                     if part)
+
+
+def status_of(item: dict) -> Tuple[str, bool]:
+    state = (item.get("state") or "").strip().lower()
+    return STATE_TO_STATUS.get(state, ("unknown", False))
+
+
+def build_record(item: dict, dbms_id: str, *, rule: str,
+                 technique: Optional[str], finder: str, excerpt: str,
+                 text: str, timestamp: str, confidence: str = "high") -> dict:
+    """Assemble a bug record from one Gitee issue."""
+    url = item.get("html_url") or ""
+    status, is_true_positive = status_of(item)
+    created = (item.get("created_at") or "")[:10] or None
+    repository = (item.get("repository") or {}).get("full_name") or "Gitee"
+
+    return {
+        "id": f"bug:{dbms_id}:{stable_digest(normalize_url(url))}",
+        "dbms": dbms_id,
+        "title": truncate(item.get("title") or "(untitled report)", 500),
+        "reported_date": parse_date(created),
+        "reported_year": year_of(parse_date(created)),
+        "status": status,
+        "status_is_true_positive": is_true_positive,
+        "finder": finder,
+        "technique": technique,
+        "symptom": "unknown",
+        "reporter": (item.get("user") or {}).get("login") or None,
+        "links": {"report": url},
+        "primary_url": normalize_url(url),
+        "attribution": {
+            "rule": rule,
+            "confidence": confidence,
+            "evidence": [{
+                "source_url": url,
+                "source_type": "issue_tracker",
+                "excerpt": excerpt,
+                "excerpt_is_verbatim": True,
+                "note": (f"Issue {item.get('number')} in {repository} on "
+                         f"Gitee, the project's own tracker."),
+                "content_sha256": content_hash(text),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }],
+        },
+        "provenance": {
+            "collector": COLLECTOR,
+            "collector_version": COLLECTOR_VERSION,
+            "policy_version": config.policy_version(),
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+            "source_url": url,
+            "source_type": "issue_tracker",
+            "content_sha256": content_hash(text),
+        },
+    }
+
+
+def collect(fetcher: Optional[Fetcher] = None,
+            tax: Optional[taxonomy.Taxonomy] = None, *,
+            timestamp: Optional[str] = None,
+            terms: Optional[Sequence[str]] = None
+            ) -> Tuple[List[dict], List[dict], List[dict]]:
+    """Return ``(records, rejected, needs_review)`` from Gitee."""
+    timestamp = timestamp or now()
+    tax = tax or taxonomy.load()
+    if fetcher is None:
+        from ..cache import Caches
+        # Deliberately slow. Gitee's limit is not documented and it blocks by
+        # address when crossed.
+        fetcher = Fetcher(Caches().http, min_interval=3.0, max_retries=1,
+                          timeout=30.0)
+    terms = list(terms) if terms is not None else search_terms(tax)
+    index = repository_index()
+
+    records: List[dict] = []
+    rejected: List[dict] = []
+    needs_review: List[dict] = []
+    seen = set()
+    del FAILED_SEARCHES[:]
+
+    # Only registered repositories are searched. Gitee hosts a great deal of
+    # throwaway content that matches a three-letter acronym, and an unscoped
+    # sweep put a thousand such issues into the review queue without reaching
+    # a single real report. Finding database projects that host here is a
+    # registry question, not something to infer from an acronym match.
+    for repository, dbms_id in sorted(index.items()):
+        for term in terms:
+            if _blocked():
+                break
+            for item in search(fetcher, term, repository=repository):
+                url = item.get("html_url") or ""
+                if not url or url in seen:
+                    continue
+                seen.add(url)
+                text = source_text(item)
+                if not text:
+                    continue
+                found = deterministic_attribution(text, tax, as_issue(item))
+                if found is None:
+                    rejected.append({
+                        "kind": "bug", "url": url,
+                        "reason": "no SQLancer tool or technique attribution",
+                        "title": item.get("title"),
+                    })
+                    continue
+                rule, technique, finder, excerpt = found
+                if not verbatim_excerpt(text, excerpt):
+                    needs_review.append({
+                        "kind": "bug", "url": url,
+                        "reason": ("excerpt did not match the report text "
+                                   "verbatim"),
+                    })
+                    continue
+                records.append(build_record(
+                    item, dbms_id, rule=rule, technique=technique,
+                    finder=finder, excerpt=excerpt, text=text,
+                    timestamp=timestamp))
+    if FAILED_SEARCHES:
+        needs_review.append({
+            "kind": "bug",
+            "url": SEARCH,
+            "reason": (f"{len(FAILED_SEARCHES)} Gitee searches did not "
+                       f"complete, so this run's result is a lower bound"),
+        })
+    return records, rejected, needs_review
diff --git a/tools/impact/collectors/github_bugs.py b/tools/impact/collectors/github_bugs.py
new file mode 100644
index 0000000..a62c6b6
--- /dev/null
+++ b/tools/impact/collectors/github_bugs.py
@@ -0,0 +1,853 @@
+"""Discovers SQLancer-attributed bug reports on GitHub.
+
+Searching is deterministic and incremental: the taxonomy supplies the terms,
+GitHub's ``created:``/``updated:`` qualifiers restrict a weekly run to threads
+that actually changed, and the raw cache keys issue bodies on the thread's
+``updated_at`` so an unchanged thread costs no request at all.
+
+Attribution then happens in two tiers. A report that states in so many words
+that SQLancer or one of its oracles found the bug is accepted deterministically,
+with the sentence stored verbatim. Everything else -- a passing mention, a bare
+acronym, a report about SQLancer rather than from it -- goes to the classifier,
+and is dropped if the classifier is unavailable or unconvinced. Nothing is
+admitted on a keyword match alone.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Dict, List, Optional, Sequence, Tuple
+
+from .. import config, taxonomy
+from ..classify import (Classifier, excluded_names, finder_names,
+                        resolve_finder, resolve_technique, technique_names)
+from ..github import GitHub
+from ..util import (content_hash, normalize_url, now, truncate,
+                    verbatim_excerpt, year_of)
+
+COLLECTOR = "github_bugs"
+COLLECTOR_VERSION = "1.0.0"
+EXTRACTOR_VERSION = "github-bug-signals-v1"
+
+# Phrasings that state a discovery rather than merely name a tool. Accepting on
+# these alone is safe in a way that "the issue contains the word SQLancer" is
+# not: "SQLancer does not support window functions" matches the latter.
+DISCOVERY_STATEMENT = re.compile(
+    r"("
+    # Filler is allowed between the preposition and the tool: real reports say
+    # "Found via automated fuzzing with SQLancer", not "found by SQLancer".
+    # Requiring them adjacent was losing reports that state the discovery
+    # perfectly clearly.
+    r"(?:found|discovered|detected|caught|reported|identified|reproduced)\s+"
+    r"(?:this\s+|the\s+)?(?:bug\s+|issue\s+|problem\s+)?"
+    r"(?:by|with|using|via|through|in|while|during)\s+[^.\n]{0,60}?"
+    r"(?<![\w+])(?P<tool_a>sqlancer\+\+|sqlancer|shqvel|norec|tlp|pqs|qpg|dqe|dqp|cert|coddtest)(?![\w])"
+    r"|"
+    r"(?P<tool_b>sqlancer\+\+|sqlancer|shqvel|norec|tlp|pqs|qpg|dqe|dqp|cert|coddtest)"
+    # `reports` only in its verb form: "SQLancer reports a mismatch" is a
+    # finding, "the sqlancer report is not useful" is a feature request.
+    r"\s+(?:found|discovered|detected|caught|reported|identified|flagged|"
+    r"reports|complains|got|hits|hit|triggers|triggered|produced|produces|"
+    r"raised|raises|crashed|crashes)"
+    r"|"
+    r"(?:while|when)\s+(?:testing|fuzzing|running)\s+[^.\n]{0,60}?"
+    r"(?:with|using)\s+(?:the\s+)?"
+    r"(?P<tool_c>sqlancer\+\+|sqlancer|shqvel|norec|tlp|pqs|qpg)"
+    r"|"
+    r"(?:the\s+)?(?P<tool_d>norec|tlp|pqs|qpg|dqe|dqp|cert|coddtest)\s+"
+    r"(?:test\s+)?oracle"
+    r")",
+    re.IGNORECASE)
+
+# Projects that test with SQLancer routinely mark the resulting reports rather
+# than describing the discovery in prose: a component tag in the title, or a
+# mention that the failure showed up in a SQLancer run. That is the policy's
+# campaign-evidence clause -- the report plainly came out of a SQLancer campaign
+# -- and without it a project's whole SQLancer backlog is invisible.
+CAMPAIGN_STATEMENT = re.compile(
+    r"\[\s*(?:sqlancer\+\+|sqlancer|shqvel)\s*\]"
+    r"|\b(?:in|during|from|via|by)\s+(?:a\s+|an\s+|the\s+|our\s+)?"
+    r"(?:sqlancer\+\+|sqlancer)\s+"
+    r"(?:test|tests|testing|run|runs|campaign|workload|workloads|job|jobs|"
+    r"fuzzing|session)\b"
+    r"|\b(?:sqlancer\+\+|sqlancer)\s+(?:test|tests|run|runs|job|jobs)\s+"
+    r"(?:fail\w*|crash\w*|error\w*)",
+    re.IGNORECASE)
+
+# A campaign tag says where the report came from, not that it is a defect: the
+# same tag appears on issues about the testing setup itself. These say it is.
+DEFECT_LABEL = re.compile(r"\b(?:kind/bug|type/bug|bug|defect|crash)\b",
+                          re.IGNORECASE)
+DEFECT_TITLE = re.compile(
+    r"\b(?:crash\w*|segv|sigsegv|core\s?dump\w*|assert\w*|panic\w*|"
+    r"fatal|abort\w*|wrong\s+result\w*|incorrect\s+result\w*|"
+    r"unexpected\s+result\w*|inconsistent\w*|error|exception|"
+    r"internal\s+error|corrupt\w*|hang\w*|deadlock\w*|leak\w*|oom)\b",
+    re.IGNORECASE)
+
+
+def looks_like_a_defect(issue: dict) -> bool:
+    """Whether an issue reports a defect rather than test-harness work."""
+    labels = issue.get("labels") or []
+    names = [label.get("name", "") if isinstance(label, dict) else str(label)
+             for label in labels]
+    if any(DEFECT_LABEL.search(name) for name in names):
+        return True
+    return bool(DEFECT_TITLE.search(issue.get("title") or ""))
+
+
+# SQLancer names the tables it generates t0, t1, ... and their columns c0, c1,
+# ..., and that convention survives into the reduced test case people paste into
+# a bug report. It is a real fingerprint: 398 of the 499 bugs in SQLancer's own
+# curated repository carry it.
+#
+# On its own it says a SQLancer-family generator produced the reproducer, not
+# who ran it, so it is only ever used together with something that establishes
+# the campaign -- a reporter known to work on SQLancer, or a curated list saying
+# where the report came from.
+REPRODUCER_SIGNATURE = re.compile(
+    r"CREATE\s+TABLE\s+(?:IF\s+NOT\s+EXISTS\s+)?[\"`\[]?t\d{1,2}[\"`\]]?\s*\(",
+    re.IGNORECASE)
+
+# The other half of the convention: SQLancer names generated columns c0, c1.
+# The table names alone are too weak -- other generators also number their
+# tables t0, t1, but pair them with their own column names (c_pk, c_int, c_0),
+# and without this those reports were being read as SQLancer's. On the
+# project's own 499 curated bugs, requiring this as well costs exactly one.
+REPRODUCER_COLUMN = re.compile(r"(?<![\w_])c\d{1,2}(?![\w])", re.IGNORECASE)
+
+
+def reproducer_excerpt(text: str) -> Optional[str]:
+    """The generated schema from a report, quoted verbatim."""
+    text = text or ""
+    match = REPRODUCER_SIGNATURE.search(text)
+    if not match or not REPRODUCER_COLUMN.search(text):
+        return None
+    start = text.rfind("\n", 0, match.start())
+    start = 0 if start < 0 else start + 1
+    end = text.find("\n", match.end())
+    for _ in range(3):  # a few statements give the convention room to show
+        if end < 0:
+            break
+        nxt = text.find("\n", end + 1)
+        if nxt < 0:
+            break
+        end = nxt
+    end = len(text) if end < 0 else end
+    return truncate(text[start:end].strip(), 700)
+
+
+# Corroborating context required before an ambiguous acronym is trusted at all.
+# The same test in the languages projects actually report in. openGauss files
+# in Chinese, and an English-only guard threw away a report whose title names
+# the oracle outright.
+DB_CONTEXT = re.compile(
+    r"\b(sql|query|queries|database|dbms|table|select|oracle|optimiz|optimis|"
+    r"result set|logic bug)\b|数据库|查询|等价验证|优化器|聚合|结果集",
+    re.IGNORECASE)
+
+# Phrasing that marks the thread as a request about SQLancer rather than a bug
+# SQLancer found. Without a classifier these have to be excluded outright.
+FEATURE_REQUEST = re.compile(
+    r"(would\s+be\s+(?:nice|good|great|helpful)|please\s+add\s+support|"
+    r"feature\s+request|is\s+not\s+(?:really\s+)?useful|"
+    r"improve\s+the\s+report|can\s+(?:we|you)\s+add|"
+    r"(?:sqlancer|shqvel)\s+(?:does\s+not|doesn't)\s+support|"
+    r"add\s+support\s+for\s+[^.\n]{0,40}(?:to|in)\s+sqlancer)",
+    re.IGNORECASE)
+
+MAX_BODY_CHARS = 24_000
+# Longest excerpt stored as attribution evidence.
+LIMIT = 700
+MAX_SEARCH_PAGES = 10
+
+STATE_TO_STATUS = {
+    "open": ("open", True),
+    "closed": ("fixed", True),
+}
+
+
+def _repo_index() -> Dict[str, str]:
+    """Map ``owner/repo`` of each registered DBMS to its registry id.
+
+    Every organisation the project is known to operate under is mapped, not
+    just the owner of its main repository. Databend is the case that showed why:
+    it moved from ``datafuselabs`` to ``databendlabs``, and its issues under the
+    new organisation were being dropped as belonging to an unregistered system.
+    """
+    registry = config.load_json(config.DATA_FILES["dbms"])["dbms"]
+    index: Dict[str, str] = {}
+    for entry in registry:
+        repository = entry.get("repository")
+        if not repository:
+            continue
+        parts = repository.rstrip("/").split("/")
+        if len(parts) < 2:
+            continue
+        owner, name = parts[-2], parts[-1]
+        index[f"{owner}/{name}".lower()] = entry["id"]
+        for other in entry.get("github_owners") or []:
+            index[f"{other}/{name}".lower()] = entry["id"]
+        # A project may file bugs in a repository of its own beside the main
+        # one; DuckDB's fuzzer robot is the case this exists for.
+        for extra in entry.get("github_repositories") or []:
+            index[extra.lower()] = entry["id"]
+    return index
+
+
+def search_queries(tax: taxonomy.Taxonomy, repos: List[str]) -> List[str]:
+    """Build the issue searches to run.
+
+    Searching per repository rather than globally is the whole point here.
+    GitHub returns at most 1000 results for a search and pages far fewer than
+    that in practice, while "SQLancer" alone appears in well over a thousand
+    issues -- so a global query silently sees a fraction of them and the rest
+    are never even considered. Scoped to one repository at a time, each search
+    returns a complete set: yugabyte-db has fifty, comfortably under any cap.
+
+    A global query is still issued as a catch-all, since that is how a bug in a
+    system the registry does not know yet gets noticed at all.
+    """
+    unambiguous = [name for name in tax.search_terms()
+                   if len(name) > 6 or name.lower() in ("norec", "shqvel")]
+    ambiguous = [technique["names"][0] for technique in tax.techniques.values()
+                 if technique.get("ambiguous_acronym")]
+
+    queries: List[str] = []
+    for repo in repos:
+        # One query for the unambiguous names, which need no context, and one
+        # for the acronyms, which the attribution rules then check in context.
+        names = " OR ".join(f'"{term}"' for term in unambiguous[:6])
+        queries.append(f"repo:{repo} ({names}) in:body,title type:issue")
+        if ambiguous:
+            acronyms = " OR ".join(f'"{term}"' for term in ambiguous)
+            queries.append(f"repo:{repo} ({acronyms}) in:body,title type:issue")
+
+    for term in unambiguous[:4]:
+        queries.append(f'"{term}" in:body,title type:issue')
+
+    # Labels are the strongest signal there is: a project that labels an issue
+    # `sqlancer` is itself saying where the report came from. Searched globally
+    # because the whole label fits well inside one search.
+    for label in ("sqlancer", "SQLancer", "sqlancer++", "norec", "tlp", "pqs"):
+        queries.append(f'label:"{label}" type:issue')
+    return queries
+
+
+def sqlancer_labels(issue: dict, tax: taxonomy.Taxonomy) -> List[str]:
+    """Labels on an issue that name SQLancer or one of its techniques."""
+    found = []
+    for label in issue.get("labels") or []:
+        name = label.get("name", "") if isinstance(label, dict) else str(label)
+        if not name:
+            continue
+        if tax.mentions_finder(name) or tax.find_techniques(
+                name, require_context=False):
+            found.append(name)
+    return found
+
+
+def _issue_source_text(issue: dict, comments: Optional[List[dict]] = None) -> str:
+    """Assemble the text a decision may be based on, newest context last."""
+    parts = [issue.get("title") or "", issue.get("body") or ""]
+    for comment in (comments or [])[:10]:
+        parts.append(comment.get("body") or "")
+    return truncate("\n\n".join(part for part in parts if part), MAX_BODY_CHARS)
+
+
+def _unambiguous(matches, tax: taxonomy.Taxonomy):
+    """The first matched technique whose name is not an ambiguous acronym."""
+    for match in matches:
+        entry = tax.techniques.get(match.technique_id) or {}
+        if not entry.get("ambiguous_acronym"):
+            return match
+    return None
+
+
+def _span_excerpt(text: str, start: int, end: int) -> str:
+    """Verbatim text around a span, for a match that is not a regex match."""
+    class _Span:
+        def __init__(self, a, b):
+            self._a, self._b = a, b
+
+        def start(self):
+            return self._a
+
+        def end(self):
+            return self._b
+
+    return _statement_excerpt(text, _Span(start, end))
+
+
+def _statement_excerpt(text: str, match: re.Match) -> str:
+    """Verbatim text around a discovery statement.
+
+    Anchored on the line the match sits in rather than a fixed character
+    window, so a mention buried in a stack trace quotes that line instead of
+    several hundred characters of unrelated trace.
+    """
+    start = text.rfind("\n", 0, match.start())
+    start = 0 if start < 0 else start + 1
+    end = text.find("\n", match.end())
+    if end < 0:
+        end = len(text)
+    elif end - start < 120:
+        # Include one continuation line when the first is short, so a wrapped
+        # sentence is not cut in half.
+        following = text.find("\n", end + 1)
+        end = len(text) if following < 0 else following
+    # Keep the matched phrase inside the excerpt: on a very long line, trimming
+    # from the start is what preserves the sentence the attribution rests on.
+    if match.end() - start > LIMIT:
+        start = max(start, match.start() - 200)
+    return truncate(text[start:end].strip(), LIMIT)
+
+
+def deterministic_attribution(text: str, tax: taxonomy.Taxonomy,
+                              issue: Optional[dict] = None, *,
+                              reporter_runs_campaigns: bool = False
+                              ) -> Optional[Tuple[str, Optional[str], str, str]]:
+    """Try to settle attribution without a model.
+
+    Returns ``(rule, technique_id, finder_id, excerpt)`` when the text states a
+    discovery outright or the report is plainly the output of a SQLancer
+    campaign, otherwise ``None``.
+
+    ``reporter_runs_campaigns`` says the report was reached by searching a known
+    campaign reporter's issues rather than by searching for the tool. That
+    changes what counts as evidence: such a report need not name SQLancer, and
+    usually does not, so the reproducer's own shape is allowed to establish it.
+    """
+    if tax.find_excluded_techniques(text):
+        # The report credits a technique that did not originate in SQLancer.
+        return None
+    match = DISCOVERY_STATEMENT.search(text)
+    if not match:
+        campaign = _campaign_attribution(text, tax, issue)
+        if campaign is not None:
+            return campaign
+        if reporter_runs_campaigns:
+            return (_reproducer_attribution(text, issue)
+                    or _campaign_reporter_attribution(text, issue))
+        return None
+    tool = next((match.group(name) for name in
+                 ("tool_a", "tool_b", "tool_c", "tool_d")
+                 if match.group(name)), None)
+    if not tool:
+        return None
+
+    excerpt = _statement_excerpt(text, match)
+    if FEATURE_REQUEST.search(excerpt):
+        # Reads like a request about SQLancer rather than a bug it found. Not
+        # accepted here, but the candidate still goes on to the classifier.
+        return None
+    finder = resolve_finder(tax, tool) or "sqlancer"
+    technique = resolve_technique(tax, tool)
+
+    if technique is not None:
+        entry = tax.techniques[technique]
+        if entry.get("ambiguous_acronym") and not DB_CONTEXT.search(excerpt):
+            return None
+        return "technique_attribution", technique, finder, excerpt
+    if tax.mentions_finder(tool):
+        # Named the tool but not the oracle; look for an oracle elsewhere.
+        matches = tax.find_techniques(text)
+        technique = matches[0].technique_id if matches else None
+        return "explicit_tool_statement", technique, finder, excerpt
+    return None
+
+
+def _reproducer_attribution(text: str, issue: Optional[dict]
+                            ) -> Optional[Tuple[str, Optional[str], str, str]]:
+    """Attribution from the reproducer's own schema, for a campaign reporter.
+
+    SQLancer generates the schema it tests against, so its reproducers create
+    tables named t0, t1 with columns c0, c1. That shape survives into the bug
+    report even when the reporter never says what produced it -- which is the
+    normal case, because a bug report is about the bug. On the project's own
+    curated list of 499 bugs this signature is present in 398.
+
+    It is only trusted for someone already known to run SQLancer campaigns.
+    On its own the convention is suggestive, not conclusive; paired with a
+    reporter whose reports are SQLancer output, it is the evidence.
+    """
+    if not looks_like_a_defect(issue or {}):
+        return None
+    excerpt = reproducer_excerpt(text)
+    if not excerpt:
+        return None
+    return "campaign_evidence", None, "sqlancer", excerpt
+
+
+def _confidence_for(rule: str) -> str:
+    """How far the evidence goes, kept in one place so records stay comparable."""
+    if rule == "campaign_reporter":
+        return "low"      # rests on who reported it, not on what they wrote
+    if rule == "campaign_evidence":
+        return "medium"   # the report carries a SQLancer-shaped reproducer
+    return "high"         # the report names the tool or the oracle
+
+
+def _campaign_reporter_attribution(text: str, issue: Optional[dict]
+                                   ) -> Optional[Tuple[str, Optional[str], str, str]]:
+    """Attribution from who filed the report, when the report says nothing.
+
+    The weakest rule here, and deliberately the last one tried. It exists
+    because the people who run SQLancer campaigns file database bugs for that
+    reason, and their reports usually give no other sign: the tool is not
+    mentioned, and the reproducer has been minimised by hand before filing,
+    which removes the generated schema along with everything else incidental.
+
+    Only a defect report counts, and only in a database system's own tracker --
+    both checked by the caller. What is quoted is the report's own description
+    of the defect, so the record still carries the reporter's words rather than
+    an assertion of ours; the roster that admitted it is attached separately by
+    ``build_record``.
+    """
+    issue = issue or {}
+    if not _is_a_bug_report(issue, text):
+        return None
+    excerpt = _defect_excerpt(text)
+    if not excerpt:
+        return None
+    return "campaign_reporter", None, "sqlancer", excerpt
+
+
+# Things a member files that are not bug reports.
+NOT_A_BUG_REPORT = re.compile(
+    r"\b(?:feature\s+request|support\s+for|add\s+support|please\s+add|"
+    r"proposal|rfc|tracking\s+issue|meta\s+issue|umbrella|roadmap|"
+    r"documentation|docs?\s+typo|typo|question|how\s+(?:do|to|can)\b|"
+    r"release\s+notes?|discussion)\b", re.IGNORECASE)
+
+
+def _is_a_bug_report(issue: dict, text: str) -> bool:
+    """Whether a roster member's issue is a bug report rather than something else.
+
+    ``looks_like_a_defect`` asks the opposite question -- does this say "crash"
+    or "wrong result" -- and that is the wrong test here. A carefully written
+    report describes the misbehaviour instead of naming it ("Dolt ignores LIMIT
+    inside a correlated EXISTS subquery"), and carries whatever triage label the
+    project happens to use ("reproduced", "customer issue"), so demanding the
+    vocabulary of a crash report threw away most of these.
+
+    So this asks whether the issue is plainly something else instead: a pull
+    request, a feature request, a question, a tracking issue. A defect label or
+    a defect-shaped title still settles it outright.
+    """
+    if "pull_request" in issue:
+        return False   # presence, not truthiness: the field can be empty
+    if looks_like_a_defect(issue):
+        return True
+    title = issue.get("title") or ""
+    if NOT_A_BUG_REPORT.search(title) or FEATURE_REQUEST.search(text):
+        return False
+    # A bug report describes something; a one-line ask usually does not.
+    return len((issue.get("body") or "").strip()) >= 120
+
+
+# Boilerplate a report template puts above the actual description.
+TEMPLATE_HEADING = re.compile(
+    r"^\s*(#{1,4}\s*)?(what happened|what happens|describe the bug|bug "
+    r"description|summary|steps? to reproduce|how to reproduce|environment|"
+    r"expected behaviou?r|actual behaviou?r|to reproduce)\s*:?\s*$",
+    re.IGNORECASE)
+
+
+def _defect_excerpt(text: str) -> Optional[str]:
+    """The report's own description of the defect, quoted verbatim.
+
+    Skips the headings a report template supplies, since "## What happened" is
+    the template's words rather than the reporter's and says nothing about the
+    bug.
+    """
+    for line in (text or "").splitlines():
+        stripped = line.strip()
+        if len(stripped) < 25 or TEMPLATE_HEADING.match(stripped):
+            continue
+        if stripped.startswith(("```", "|", ">", "<!--")):
+            continue
+        return truncate(stripped, LIMIT)
+    return None
+
+
+# Where the tool is named, for picking a sentence to quote when nothing more
+# specific matched.
+TOOL_MENTION = re.compile(r"(?<![\w+])(sqlancer\+\+|sqlancer|shqvel)(?![\w])",
+                          re.IGNORECASE)
+
+
+def _campaign_attribution(text: str, tax: taxonomy.Taxonomy,
+                          issue: Optional[dict]
+                          ) -> Optional[Tuple[str, Optional[str], str, str]]:
+    """Attribution from a defect report, in a DBMS tracker, that names SQLancer.
+
+    Pattern-matching how people phrase a discovery was the wrong shape. Reports
+    say "Found via automated fuzzing with SQLancer", "[YSQL][SQLancer]",
+    "SQLancer got:", "Test case generated from SQLancer" -- and every pattern
+    written to catch them missed more than it caught.
+
+    So the rule is the plain one: a defect report, in a database system's own
+    tracker, that names SQLancer. There is little else such a report would be.
+    The phrasing patterns are still consulted, but only to choose which sentence
+    to quote, since one that states the discovery is better evidence than the
+    first mention anywhere in the text.
+
+    Two things still disqualify: language marking the issue as a request about
+    SQLancer rather than a bug it found, and crediting a technique that did not
+    originate in SQLancer.
+    """
+    if issue is None:
+        return None
+
+    # A project's own label settles it, whatever the issue text says and
+    # whether or not the title reads like a defect: the project applied it.
+    labels = sqlancer_labels(issue, tax)
+    if labels:
+        match = (DISCOVERY_STATEMENT.search(text)
+                 or CAMPAIGN_STATEMENT.search(text)
+                 or TOOL_MENTION.search(text))
+        excerpt = (_statement_excerpt(text, match) if match
+                   else f"Labelled {', '.join(labels)}")
+        named = tax.mentions_finder(" ".join(labels)) or tax.mentions_finder(text)
+        matched = (tax.find_techniques(" ".join(labels), require_context=False)
+                   or tax.find_techniques(text))
+        return ("campaign_evidence",
+                matched[0].technique_id if matched else None,
+                named[0] if named else "sqlancer", excerpt)
+
+    if not looks_like_a_defect(issue):
+        return None
+
+    finders = tax.mentions_finder(text)
+    technique_matches = tax.find_techniques(text)
+    if not finders and not technique_matches:
+        return None
+    if tax.find_excluded_techniques(text) and not finders:
+        return None
+
+    match = (DISCOVERY_STATEMENT.search(text)
+             or CAMPAIGN_STATEMENT.search(text)
+             or TOOL_MENTION.search(text))
+    if match is not None:
+        excerpt = _statement_excerpt(text, match)
+    elif _unambiguous(technique_matches, tax) or (
+            technique_matches and reproducer_excerpt(text)):
+        # The report names an oracle but never the tool, so none of the
+        # sentence patterns has anything to anchor on -- TOOL_MENTION knows
+        # only the tool's own names. An unambiguous oracle name is its own
+        # anchor: "the NoREC-transformed query using SUM ... returns 0" is a
+        # report of a SQLancer oracle finding a bug, whether or not the word
+        # SQLancer appears anywhere in it.
+        #
+        # Only unambiguous ones. An acronym like CERT or TLP needs database
+        # context to be a technique at all, and in a DBMS tracker that context
+        # is always present -- every issue is about a database. Requiring the
+        # tool to be named is what keeps TLS certificates in a Galera cluster
+        # report from being read as cardinality estimation testing.
+        # An ambiguous acronym is allowed to anchor only when the report also
+        # carries SQLancer's generated-schema fingerprint. That is what tells
+        # "TLP equivalence verification" over a t0(c0 ...) schema apart from a
+        # TLS certificate in a cluster report, and it works in any language:
+        # the reproducer is SQL either way. openGauss's tracker is in Chinese,
+        # so no English sentence pattern can reach it.
+        first = _unambiguous(technique_matches, tax) or technique_matches[0]
+        excerpt = _span_excerpt(text, first.start, first.end)
+    else:
+        return None
+    if FEATURE_REQUEST.search(excerpt):
+        return None
+
+    finder = finders[0] if finders else "sqlancer"
+    technique = technique_matches[0].technique_id if technique_matches else None
+    return "campaign_evidence", technique, finder, excerpt
+
+
+# Where the roster of campaign reporters is published, cited as the evidence
+# for a record admitted on the strength of who filed it.
+ROSTER_URL = "https://nus-test.github.io/people/"
+
+# Systems a lab member's name does not vouch for. The lab tests graph database
+# systems as well, with tools developed independently of SQLancer, and SQLancer
+# itself has no provider for any of them -- so a defect report filed against one
+# by someone on the roster is not the output of a SQLancer campaign, which is
+# the whole of what the reporter rule infers. Ruled by Manuel Rigger,
+# 2026-09-12, after 38 records reached the dataset this way.
+OUTSIDE_SQLANCER_CAMPAIGNS = {"neo4j", "redisgraph"}
+
+
+def _evidence_note(rule: str) -> str:
+    if rule == "campaign_reporter":
+        return ("The reporter runs SQLancer campaigns; this is their "
+                "description of the defect.")
+    if rule == "campaign_evidence":
+        return "The project marks this report as coming from a SQLancer run."
+    return "The report states which tool or oracle found the bug."
+
+
+def _roster_evidence(issue: dict, timestamp: str) -> dict:
+    """Why this reporter's database bugs count, stated openly."""
+    reporter = (issue.get("user") or {}).get("login") or "the reporter"
+    return {
+        "source_url": ROSTER_URL,
+        "source_type": "website",
+        "note": (f"{reporter} is listed as a member of the TEST lab, which "
+                 f"runs SQLancer campaigns against database systems. This "
+                 f"record rests on that membership, not on anything the "
+                 f"report says."),
+        "retrieved_at": timestamp,
+        "first_seen": timestamp,
+        "last_verified": timestamp,
+    }
+
+
+def build_record(issue: dict, dbms_id: str, *, rule: str, technique: Optional[str],
+                 finder: str, excerpt: str, source_text: str, timestamp: str,
+                 classifier_record: Optional[dict] = None,
+                 confidence: str = "high") -> dict:
+    """Assemble a bug record from a GitHub issue."""
+    url = issue["html_url"]
+    status, is_true_positive = STATE_TO_STATUS.get(
+        issue.get("state", ""), ("unknown", False))
+    # A closed issue is only a fixed bug if something closed it as completed.
+    if issue.get("state") == "closed" and issue.get("state_reason") == "not_planned":
+        status, is_true_positive = "closed_not_a_bug", False
+
+    created = (issue.get("created_at") or "")[:10] or None
+    from ..util import stable_digest
+    from ..repos import canonical_issue_url
+    # The address the project is at today, so a bug filed under a former
+    # organisation is the same record as the one found under the current name.
+    url = canonical_issue_url(url) or url
+    record_id = f"bug:{dbms_id}:{stable_digest(url)}"
+
+    evidence = [{
+        "source_url": url,
+        "source_type": ("github_pull_request" if issue.get("pull_request")
+                        else "github_issue"),
+        "excerpt": excerpt,
+        "excerpt_is_verbatim": True,
+        "note": _evidence_note(rule),
+        "content_sha256": content_hash(source_text),
+        "retrieved_at": timestamp,
+        "first_seen": timestamp,
+        "last_verified": timestamp,
+    }]
+
+    if rule == "campaign_reporter":
+        evidence.append(_roster_evidence(issue, timestamp))
+
+    attribution = {
+        "rule": rule,
+        "confidence": confidence,
+        "evidence": evidence,
+    }
+    if classifier_record:
+        attribution["classifier"] = classifier_record
+
+    return {
+        "id": record_id,
+        "dbms": dbms_id,
+        "title": truncate(issue.get("title") or "(untitled report)", 500),
+        "reported_date": created,
+        "reported_year": year_of(created),
+        "status": status,
+        "status_is_true_positive": is_true_positive,
+        "finder": finder,
+        "technique": technique,
+        "symptom": "unknown",
+        "reporter": (issue.get("user") or {}).get("login"),
+        "links": {"report": url},
+        "primary_url": normalize_url(url),
+        "attribution": attribution,
+        "provenance": {
+            "collector": COLLECTOR,
+            "collector_version": COLLECTOR_VERSION,
+            "policy_version": config.policy_version(),
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+            "source_url": url,
+            "source_type": "github_issue",
+            "content_sha256": content_hash(source_text),
+        },
+    }
+
+
+def member_queries(gh: GitHub, tax: taxonomy.Taxonomy) -> List[str]:
+    """Author-scoped searches for the people who run SQLancer campaigns.
+
+    Term search is capped at 1000 results and "SQLancer" comfortably exceeds
+    that, so reports get lost. Searching by author does not hit the same wall.
+
+    No search term is attached. Asking for a person's issues *that mention
+    SQLancer* cannot reach a report whose text never names it, and most do not:
+    a bug report is written about the bug, not about what found it. What comes
+    back is filtered by the same rules as everything else -- the query only
+    decides what gets looked at.
+    """
+    from . import people
+
+    try:
+        return people.search_queries(people.handles(gh))
+    except Exception:
+        return []
+
+
+def collect(gh: GitHub, classifier: Optional[Classifier] = None, *,
+            since: Optional[str] = None, timestamp: Optional[str] = None,
+            full: bool = False, max_candidates: int = 25_000,
+            include_members: bool = True,
+            only_repos: Optional[Sequence[str]] = None
+            ) -> Tuple[List[dict], List[dict], List[dict]]:
+    """Return ``(records, rejected, needs_review)``.
+
+    ``rejected`` holds candidates the evidence did not support; ``needs_review``
+    holds candidates that would need a classifier run that could not happen.
+    Both are reported in the pull request so a human can see what was left out.
+
+    ``only_repos`` restricts the repository searches to the ``owner/name`` pairs
+    given. A newly registered system needs its own repository searched and
+    nothing else -- the global catch-all query cannot reach it, since "SQLancer"
+    is far past GitHub's result cap -- and re-running every repository to pick
+    up one is a great deal of work for a known answer.
+    """
+    timestamp = timestamp or now()
+    tax = taxonomy.load()
+    repo_ids = _repo_index()
+    repos = sorted(_repo_index())
+    if only_repos is not None:
+        wanted = {name.lower() for name in only_repos}
+        repos = [name for name in repos if name.lower() in wanted]
+        unknown = wanted - {name.lower() for name in repo_ids}
+        if unknown:
+            raise ValueError(
+                "not a registered database system's repository: "
+                + ", ".join(sorted(unknown)))
+
+    records: List[dict] = []
+    rejected: List[dict] = []
+    needs_review: List[dict] = []
+    seen_urls = set()
+
+    # Author searches run first. The candidate cap is a bound on total work,
+    # and whichever queries run last are the ones it silences -- so the queries
+    # that reach reports nothing else can find must not be the ones at the back.
+    queries = [(query, True) for query in member_queries(gh, tax)] \
+        if include_members else []
+    queries += [(query, False) for query in search_queries(tax, repos)]
+
+    for query, by_author in queries:
+        for item in gh.search_issues(query, since=None if full else since,
+                                     max_pages=MAX_SEARCH_PAGES,
+                                     sort="updated"):
+            url = item.get("html_url") or ""
+            if not url or url in seen_urls:
+                continue
+            seen_urls.add(url)
+            if len(seen_urls) > max_candidates:
+                break
+
+            owner_repo = _owner_repo(item.get("repository_url") or url)
+            dbms_id = repo_ids.get(owner_repo.lower()) if owner_repo else None
+            if dbms_id is None:
+                if by_author:
+                    # An author search returns everything the person ever
+                    # filed, most of it nothing to do with database systems.
+                    # Reporting each one as needing review would bury the
+                    # candidates that do.
+                    continue
+                # A bug in a system not in the registry cannot be filed against
+                # a DBMS id; surface it rather than guessing one.
+                needs_review.append({
+                    "kind": "bug",
+                    "url": url,
+                    "reason": "repository is not a registered database system",
+                    "title": item.get("title"),
+                })
+                continue
+
+            source_text = _issue_source_text(item)
+            deterministic = deterministic_attribution(
+                source_text, tax, item,
+                reporter_runs_campaigns=(
+                    by_author and dbms_id not in OUTSIDE_SQLANCER_CAMPAIGNS))
+            if deterministic is not None:
+                rule, technique, finder, excerpt = deterministic
+                if not verbatim_excerpt(source_text, excerpt):
+                    continue
+                records.append(build_record(
+                    item, dbms_id, rule=rule, technique=technique,
+                    finder=finder, excerpt=excerpt, source_text=source_text,
+                    timestamp=timestamp,
+                    confidence=_confidence_for(rule)))
+                continue
+
+            # Inside a registered database system's own repository, a bare
+            # acronym needs no further corroboration to be worth a question:
+            # the repository is the database-testing context. Requiring it here
+            # as well is what turns the deterministic layer into a gatekeeper,
+            # and it belongs upstream of the classifier, not in place of it.
+            mentions = (tax.mentions_finder(source_text)
+                        or tax.find_techniques(source_text, require_context=False)
+                        or sqlancer_labels(item, tax))
+            if not mentions:
+                if not by_author:
+                    rejected.append({
+                        "kind": "bug", "url": url,
+                        "reason": "no SQLancer tool or technique mentioned at all",
+                    })
+                continue
+            if tax.find_excluded_techniques(source_text) and not tax.mentions_finder(
+                    source_text):
+                rejected.append({
+                    "kind": "bug", "url": url,
+                    "reason": "credits a technique that did not originate in SQLancer",
+                })
+                continue
+
+            # Ask even when no key is present: the answer may already be in
+            # the cache, and a cached answer needs no API call. Checking
+            # availability here instead meant a warm cache was unusable
+            # offline, which is the opposite of what the cache is for.
+            result = None if classifier is None else classifier.classify(
+                "bug_attribution",
+                source_id=f"github:{owner_repo}#{item.get('number')}",
+                source_text=source_text,
+                source_label=f"GitHub issue {owner_repo}#{item.get('number')}",
+                technique_names=technique_names(tax),
+                finder_names=finder_names(tax),
+                excluded_names=excluded_names(tax))
+
+            if result is None:
+                needs_review.append({
+                    "kind": "bug", "url": url, "title": item.get("title"),
+                    "reason": "needs semantic classification; classifier unavailable",
+                })
+                continue
+            if not result.is_positive:
+                rejected.append({
+                    "kind": "bug", "url": url, "answer": result.answer,
+                    "reason": result.reason or "classifier did not confirm attribution",
+                })
+                continue
+
+            technique = resolve_technique(tax, result.extra.get("technique"))
+            finder = resolve_finder(tax, result.extra.get("finder")) or "sqlancer"
+            records.append(build_record(
+                item, dbms_id,
+                rule="technique_attribution" if technique else "explicit_tool_statement",
+                technique=technique, finder=finder,
+                excerpt=result.excerpts[0], source_text=source_text,
+                timestamp=timestamp,
+                classifier_record=result.classifier_record(),
+                confidence="medium"))
+
+    return records, rejected, needs_review
+
+
+def _owner_repo(url: str) -> Optional[str]:
+    match = re.search(r"repos/([\w.-]+/[\w.-]+)", url)
+    if match:
+        return match.group(1)
+    match = re.search(r"github\.com/([\w.-]+/[\w.-]+)/", url)
+    return match.group(1) if match else None
diff --git a/tools/impact/collectors/lab_members.py b/tools/impact/collectors/lab_members.py
new file mode 100644
index 0000000..b549eb2
--- /dev/null
+++ b/tools/impact/collectors/lab_members.py
@@ -0,0 +1,103 @@
+"""Discovers bug reports written by the people who run SQLancer campaigns.
+
+Searching by term has a ceiling: GitHub caps a search at 1000 results, and
+"SQLancer" appears in far more issues than that. Searching by author does not
+hit the same wall, and it reaches reports the term queries never returned.
+
+The people who file these reports are the TEST lab, whose site lists its members
+and whose bug list records the handles that filed each entry. Both are read
+here, and every candidate then goes through exactly the same attribution rules
+as any other GitHub discovery -- being a lab member's issue is a reason to look
+at it, never a reason to count it.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+from typing import List, Optional, Sequence, Tuple
+
+from ..github import GitHub
+from ..http import Fetcher
+from ..util import dedupe_preserving_order
+
+COLLECTOR = "lab_members"
+COLLECTOR_VERSION = "1.0.0"
+
+PEOPLE_URL = "https://nus-test.github.io/people/"
+BUGS_OWNER, BUGS_REPO, BUGS_REF = "nus-test", "nus-test.github.io", "main"
+BUGS_PATH = "data/bugs.json"
+
+GITHUB_HANDLE = re.compile(
+    r"github\.com/([A-Za-z0-9][\w-]{1,38})(?![\w/-])", re.IGNORECASE)
+
+# Accounts that appear on the page but are not people filing bug reports.
+NOT_A_MEMBER = {
+    "sqlancer", "nus-test", "github", "orgs", "features", "about", "pricing",
+    "sponsors", "readme", "site", "security", "apps", "marketplace", "explore",
+    "settings", "login", "join", "topics", "collections", "blog",
+}
+
+
+def from_people_page(fetcher: Fetcher) -> List[str]:
+    """GitHub handles linked from the lab's people page."""
+    try:
+        entry, _ = fetcher.fetch(PEOPLE_URL, max_age_days=14)
+    except Exception:
+        return []
+    if entry.get("status") != 200:
+        return []
+    handles = [match.group(1) for match in GITHUB_HANDLE.finditer(entry["body"])]
+    return [h for h in dedupe_preserving_order(handles)
+            if h.lower() not in NOT_A_MEMBER]
+
+
+def from_bug_list(gh: GitHub) -> List[str]:
+    """Handles that filed entries on the lab's bug list.
+
+    Complements the people page: it names people who have left, and the page
+    names people who have not filed anything yet.
+    """
+    raw, _ = gh.raw_file(BUGS_OWNER, BUGS_REPO, BUGS_PATH, BUGS_REF,
+                         max_age_days=7)
+    if not raw:
+        return []
+    try:
+        entries = json.loads(raw)
+    except ValueError:
+        return []
+    handles = []
+    for entry in entries:
+        reporter = (entry.get("reported_by") or "").strip()
+        # The field holds a GitHub handle for some entries and a display name
+        # for others; only a handle is usable as a search qualifier.
+        if reporter and " " not in reporter and reporter.lower() not in NOT_A_MEMBER:
+            handles.append(reporter)
+    return dedupe_preserving_order(handles)
+
+
+def handles(gh: GitHub, fetcher: Optional[Fetcher] = None) -> List[str]:
+    """Every handle worth searching, from both sources."""
+    if fetcher is None:
+        from ..cache import Caches
+        fetcher = Fetcher(Caches().http, min_interval=0.5, max_retries=2,
+                          timeout=20.0)
+    # The taxonomy's own contributor roster belongs here too: a Google Summer
+    # of Code contributor files bugs from SQLancer campaigns just as a lab
+    # member does, and their reports are found the same way.
+    from .. import taxonomy
+
+    roster = [entry.get("github") for entry
+              in taxonomy.load().project_contributors if entry.get("github")]
+    found = from_people_page(fetcher) + from_bug_list(gh) + roster
+    return dedupe_preserving_order(found)
+
+
+def search_queries(members: Sequence[str], terms: Sequence[str]) -> List[str]:
+    """One issue search per member, covering every discovery term at once.
+
+    A single OR query per person keeps this to one search each rather than one
+    per person per term, which matters against a 30-searches-per-minute limit.
+    """
+    joined = " OR ".join(sorted({term for term in terms if term}))
+    return [f"author:{handle} ({joined}) type:issue" for handle in members]
diff --git a/tools/impact/collectors/mariadb_jira.py b/tools/impact/collectors/mariadb_jira.py
new file mode 100644
index 0000000..62470d0
--- /dev/null
+++ b/tools/impact/collectors/mariadb_jira.py
@@ -0,0 +1,247 @@
+"""Imports SQLancer bug reports from MariaDB's public Jira.
+
+MariaDB does not take bug reports on GitHub, so the repository search that
+reaches every other system finds nothing for it: its 14 records all arrived
+through curated lists. Its Jira is open, needs no credentials, and answers a
+JQL query over the REST API, which makes it the one route to reports that name
+SQLancer in MariaDB's own tracker.
+
+Attribution is the shared one. Nothing here decides whether a report counts --
+``deterministic_attribution`` does, on the same evidence it requires of a GitHub
+issue -- so a Jira report earns its place exactly as an issue does. The only
+translation is the issue dict the rules read: a Jira issue type of Bug is passed
+where a GitHub defect label would be, because it is the same signal from the
+same kind of source, applied by the project itself.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+import urllib.parse
+from typing import Dict, List, Optional, Sequence, Tuple
+
+from .. import config, taxonomy
+from ..http import Fetcher
+from ..util import (content_hash, normalize_url, now, parse_date, stable_digest,
+                    truncate, verbatim_excerpt, year_of)
+from .github_bugs import deterministic_attribution
+
+COLLECTOR = "mariadb_jira"
+COLLECTOR_VERSION = "1.0.0"
+
+DBMS_ID = "mariadb"
+API = "https://jira.mariadb.org/rest/api/2/search"
+BROWSE = "https://jira.mariadb.org/browse/"
+FIELDS = "summary,description,status,resolution,created,reporter,issuetype"
+PAGE_SIZE = 100
+MAX_PAGES = 5
+
+# Jira's ``~`` operator tokenises its argument and matches any of the words, so
+# a multi-word technique name returns hundreds of unrelated issues -- searching
+# for the NoREC paper's title matched 195. Only single tokens are searched, and
+# a report describing a technique in words almost always names it too.
+SINGLE_TOKEN = re.compile(r"^[A-Za-z][\w+.-]*$")
+
+RESOLUTION_STATUS = {
+    "fixed": ("fixed", True),
+    "done": ("fixed", True),
+    "duplicate": ("closed_duplicate", False),
+    "not a bug": ("closed_not_a_bug", False),
+    "won't fix": ("closed_not_a_bug", False),
+    "wont fix": ("closed_not_a_bug", False),
+    "cannot reproduce": ("closed_not_a_bug", False),
+    "incomplete": ("closed_not_a_bug", False),
+}
+
+# An unresolved issue is open whatever workflow state it sits in.
+OPEN_STATUSES = {"open", "confirmed", "in progress", "in review", "in testing",
+                 "stalled", "needs feedback"}
+
+
+def search_terms(tax: taxonomy.Taxonomy) -> List[str]:
+    """Single-token names worth issuing a JQL query for."""
+    terms = []
+    for name in tax.search_terms():
+        if SINGLE_TOKEN.match(name or ""):
+            terms.append(name)
+    seen, ordered = set(), []
+    for term in terms:
+        if term.lower() not in seen:
+            seen.add(term.lower())
+            ordered.append(term)
+    return ordered
+
+
+def _escape(term: str) -> str:
+    return term.replace("\\", "\\\\").replace('"', '\\"')
+
+
+def search(fetcher: Fetcher, term: str, *, max_pages: int = MAX_PAGES
+           ) -> List[dict]:
+    """Every issue whose text matches ``term``."""
+    issues: List[dict] = []
+    for page in range(max_pages):
+        query = urllib.parse.quote(f'text ~ "{_escape(term)}" ORDER BY created ASC')
+        url = (f"{API}?jql={query}&startAt={page * PAGE_SIZE}"
+               f"&maxResults={PAGE_SIZE}&fields={FIELDS}")
+        try:
+            entry, _ = fetcher.fetch(url, max_age_days=7)
+        except Exception:
+            return issues
+        if entry.get("status") != 200:
+            return issues
+        try:
+            payload = json.loads(entry["body"])
+        except ValueError:
+            return issues
+        batch = payload.get("issues") or []
+        issues.extend(batch)
+        if len(issues) >= int(payload.get("total") or 0) or not batch:
+            return issues
+    return issues
+
+
+def status_of(fields: dict) -> Tuple[str, bool]:
+    """The report's state, from its resolution first and its workflow second."""
+    resolution = ((fields.get("resolution") or {}).get("name") or "").strip().lower()
+    if resolution in RESOLUTION_STATUS:
+        return RESOLUTION_STATUS[resolution]
+    if resolution:
+        # An unrecognised resolution closed the issue, but not visibly as a
+        # fix; saying "unknown" is more honest than guessing either way.
+        return "unknown", False
+    status = ((fields.get("status") or {}).get("name") or "").strip().lower()
+    if status in OPEN_STATUSES:
+        return "open", True
+    return "unknown", False
+
+
+def as_issue(item: dict) -> dict:
+    """The shape ``deterministic_attribution`` reads, from a Jira issue.
+
+    The issue type is passed where a GitHub label would be. Both are a
+    statement by the project's own triage that this is a defect, which is what
+    the rule is asking; nothing is invented that the tracker does not say.
+    """
+    fields = item.get("fields") or {}
+    kind = (fields.get("issuetype") or {}).get("name") or ""
+    return {
+        "title": fields.get("summary") or "",
+        "body": fields.get("description") or "",
+        "labels": [{"name": kind}] if kind else [],
+        "html_url": f"{BROWSE}{item.get('key', '')}",
+    }
+
+
+def source_text(item: dict) -> str:
+    fields = item.get("fields") or {}
+    return "\n".join(part for part in (fields.get("summary"),
+                                       fields.get("description")) if part)
+
+
+def build_record(item: dict, *, rule: str, technique: Optional[str],
+                 finder: str, excerpt: str, text: str, timestamp: str,
+                 confidence: str = "high") -> dict:
+    """Assemble a bug record from one Jira issue."""
+    fields = item.get("fields") or {}
+    key = item.get("key") or ""
+    url = f"{BROWSE}{key}"
+    status, is_true_positive = status_of(fields)
+    created = (fields.get("created") or "")[:10] or None
+    reporter = fields.get("reporter") or {}
+
+    return {
+        "id": f"bug:{DBMS_ID}:{stable_digest(normalize_url(url))}",
+        "dbms": DBMS_ID,
+        "title": truncate(fields.get("summary") or key or "(untitled report)", 500),
+        "reported_date": parse_date(created),
+        "reported_year": year_of(parse_date(created)),
+        "status": status,
+        "status_is_true_positive": is_true_positive,
+        "finder": finder,
+        "technique": technique,
+        "symptom": "unknown",
+        "reporter": (reporter.get("displayName") or reporter.get("name")
+                     or None),
+        "links": {"report": url},
+        "primary_url": normalize_url(url),
+        "attribution": {
+            "rule": rule,
+            "confidence": confidence,
+            "evidence": [{
+                "source_url": url,
+                "source_type": "issue_tracker",
+                "excerpt": excerpt,
+                "excerpt_is_verbatim": True,
+                "note": (f"{key} in MariaDB's own Jira, filed as a "
+                         f"{(fields.get('issuetype') or {}).get('name') or 'report'}."),
+                "content_sha256": content_hash(text),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }],
+        },
+        "provenance": {
+            "collector": COLLECTOR,
+            "collector_version": COLLECTOR_VERSION,
+            "policy_version": config.policy_version(),
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+            "source_url": url,
+            "source_type": "issue_tracker",
+            "content_sha256": content_hash(text),
+        },
+    }
+
+
+def collect(fetcher: Optional[Fetcher] = None,
+            tax: Optional[taxonomy.Taxonomy] = None, *,
+            timestamp: Optional[str] = None,
+            terms: Optional[Sequence[str]] = None
+            ) -> Tuple[List[dict], List[dict], List[dict]]:
+    """Return ``(records, rejected, needs_review)`` from MariaDB's Jira."""
+    timestamp = timestamp or now()
+    tax = tax or taxonomy.load()
+    if fetcher is None:
+        from ..cache import Caches
+        fetcher = Fetcher(Caches().http, min_interval=1.0, max_retries=2,
+                          timeout=30.0)
+    terms = list(terms) if terms is not None else search_terms(tax)
+
+    records: List[dict] = []
+    rejected: List[dict] = []
+    needs_review: List[dict] = []
+    seen = set()
+
+    for term in terms:
+        for item in search(fetcher, term):
+            key = item.get("key")
+            if not key or key in seen:
+                continue
+            seen.add(key)
+            text = source_text(item)
+            if not text:
+                continue
+            issue = as_issue(item)
+            found = deterministic_attribution(text, tax, issue)
+            if found is None:
+                rejected.append({
+                    "kind": "bug",
+                    "url": issue["html_url"],
+                    "reason": "no SQLancer tool or technique attribution",
+                    "title": issue["title"],
+                })
+                continue
+            rule, technique, finder, excerpt = found
+            if not verbatim_excerpt(text, excerpt):
+                needs_review.append({
+                    "kind": "bug",
+                    "url": issue["html_url"],
+                    "reason": "excerpt did not match the report text verbatim",
+                })
+                continue
+            records.append(build_record(
+                item, rule=rule, technique=technique, finder=finder,
+                excerpt=excerpt, text=text, timestamp=timestamp))
+    return records, rejected, needs_review
diff --git a/tools/impact/collectors/nus_test.py b/tools/impact/collectors/nus_test.py
new file mode 100644
index 0000000..0039668
--- /dev/null
+++ b/tools/impact/collectors/nus_test.py
@@ -0,0 +1,412 @@
+"""Imports database-system bugs from the NUS TEST lab's bug list.
+
+The lab publishes a structured list at https://nus-test.github.io/bugs/, backed
+by a JSON file in its site repository. It is a rich source of historical
+SQLancer findings -- but it is a list of the lab's bugs, not SQLancer's, and it
+includes plenty found with tools that have nothing to do with SQLancer. So none
+of it is imported on the strength of being there.
+
+Each candidate is taken back to its own bug report and put through exactly the
+same attribution rules as any other GitHub discovery: a statement that SQLancer
+or one of its oracles found it, or nothing. Reports whose page cannot be read,
+or that the rules cannot settle, are surfaced for a human rather than guessed
+at.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+from typing import Dict, List, Optional, Sequence, Tuple
+
+from .. import config, taxonomy
+from ..classify import (Classifier, excluded_names, finder_names,
+                        resolve_finder, resolve_technique, technique_names)
+from ..github import GitHub
+from ..http import Fetcher
+from ..repos import canonical_issue_url
+from ..util import (content_hash, github_issue_ref, normalize_url, now,
+                    parse_date, slugify, stable_digest, truncate,
+                    verbatim_excerpt, year_of)
+from .github_bugs import deterministic_attribution, reproducer_excerpt
+from .sqlancer_bugs import build_alias_map
+
+COLLECTOR = "nus_test"
+COLLECTOR_VERSION = "1.0.0"
+
+OWNER, REPO, REF = "nus-test", "nus-test.github.io", "main"
+DATA_PATH = "data/bugs.json"
+DATA_URL = f"https://github.com/{OWNER}/{REPO}/blob/{REF}/{DATA_PATH}"
+PAGE_URL = "https://nus-test.github.io/bugs/"
+
+# The lab tracks several domains; only database systems are in scope here.
+DBMS_DOMAIN = "dbms"
+
+# Upstream resolution values, mapped onto the schema's status vocabulary.
+RESOLUTION_STATUS = {
+    "fixed": ("fixed", True),
+    "confirmed": ("verified", True),
+    "verified": ("verified", True),
+    "open": ("open", True),
+    "duplicate": ("closed_duplicate", False),
+    "wontfix": ("closed_not_a_bug", False),
+    "not a bug": ("closed_not_a_bug", False),
+    "invalid": ("closed_not_a_bug", False),
+}
+
+MAX_PAGE_CHARS = 24_000
+TAG_RE = re.compile(r"<[^>]+>")
+SCRIPT_RE = re.compile(r"<(script|style)[^>]*>.*?</\1>", re.S | re.I)
+
+
+def _status_of(entry: dict) -> Tuple[str, bool]:
+    resolution = (entry.get("resolution") or "").strip().lower()
+    if resolution in RESOLUTION_STATUS:
+        return RESOLUTION_STATUS[resolution]
+    state = (entry.get("state") or "").strip().lower()
+    if state == "open":
+        return "open", True
+    return "unknown", False
+
+
+def _strip_html(html: str) -> str:
+    """Plain text of a fetched page, for the attribution rules to read.
+
+    Tags are removed rather than parsed: the rules only need the prose, and the
+    excerpt they store is checked against this same text, so the stored quote
+    always matches what was actually inspected.
+    """
+    text = SCRIPT_RE.sub(" ", html)
+    text = TAG_RE.sub(" ", text)
+    text = text.replace("&amp;", "&").replace("&lt;", "<").replace("&gt;", ">")
+    text = text.replace("&quot;", '"').replace("&#39;", "'").replace("&nbsp;", " ")
+    text = re.sub(r"[ \t]+", " ", text)
+    text = re.sub(r"\n\s*\n\s*\n+", "\n\n", text)
+    return truncate(text.strip(), MAX_PAGE_CHARS)
+
+
+def load_index(gh: GitHub) -> List[dict]:
+    """The lab's bug list, restricted to database systems."""
+    raw, _ = gh.raw_file(OWNER, REPO, DATA_PATH, REF, max_age_days=3)
+    if not raw:
+        return []
+    try:
+        entries = json.loads(raw)
+    except ValueError:
+        return []
+    return [entry for entry in entries
+            if (entry.get("domain") or "").lower() == DBMS_DOMAIN]
+
+
+def emailed(entries: Sequence[dict]) -> List[dict]:
+    """Entries the list records without a link.
+
+    Not every project takes bug reports in a public tracker. Umbra's are sent
+    by email, so the lab's own list is the whole public record of them -- 55
+    bugs that a route requiring a URL cannot see at all.
+    """
+    return [entry for entry in entries if not entry.get("url")]
+
+
+def _fetch_report(gh: GitHub, fetcher: Fetcher, url: str
+                  ) -> Tuple[Optional[str], Optional[str]]:
+    """Return ``(text, source_type)`` for a bug report, or ``(None, None)``.
+
+    GitHub issues go through the API so the thread's ``updated_at`` can drive
+    the cache; everything else is a plain conditional fetch.
+    """
+    reference = github_issue_ref(url)
+    if reference:
+        owner, repo, kind, number = reference
+        try:
+            issue, _ = gh.issue(owner, repo, number, max_age_days=30)
+        except RuntimeError:
+            return None, None
+        if not issue:
+            return None, None
+        body = "\n\n".join(part for part in
+                           [issue.get("title") or "", issue.get("body") or ""]
+                           if part)
+        return truncate(body, MAX_PAGE_CHARS), (
+            "github_pull_request" if kind == "pull_request" else "github_issue")
+
+    try:
+        entry, _ = fetcher.fetch(url, max_age_days=90)
+    except Exception:
+        # Dead links and unreachable trackers are common in a list this old;
+        # one of them must not end the run.
+        return None, None
+    if entry.get("status") != 200 or not entry.get("body"):
+        return None, None
+    source_type = "forum_post" if "forum" in url.lower() else "issue_tracker"
+    return _strip_html(entry["body"]), source_type
+
+
+def collect(gh: GitHub, classifier: Optional[Classifier] = None, *,
+            fetcher: Optional[Fetcher] = None, timestamp: Optional[str] = None,
+            max_reports: int = 400, entries: Optional[List[dict]] = None
+            ) -> Tuple[List[dict], List[dict], List[dict]]:
+    """Return ``(records, rejected, needs_review)``.
+
+    ``max_reports`` bounds how many reports one run will read. The raw cache
+    makes an already-inspected report free, so successive runs work through the
+    backlog without any single run being expensive.
+    """
+    timestamp = timestamp or now()
+    tax = taxonomy.load()
+    alias_map = build_alias_map()
+    if fetcher is None:
+        from ..cache import Caches
+        # Short timeout and few retries: this walks a long list of external
+        # trackers, some of which no longer resolve.
+        fetcher = Fetcher(Caches().http, min_interval=0.5, max_retries=2,
+                          timeout=15.0, initial_backoff=1.0, max_backoff=8.0)
+
+    index = entries if entries is not None else load_index(gh)
+    records: List[dict] = []
+    rejected: List[dict] = []
+    needs_review: List[dict] = []
+    inspected = 0
+
+    for entry in index:
+        url = entry.get("url")
+        canonical = normalize_url(url) if url else None
+        if not canonical:
+            # Handled by ``emailed_records``: there is no report to fetch, so
+            # the rules that read one cannot apply.
+            continue
+
+        dbms_id = alias_map.get((entry.get("system") or "").strip().lower())
+        if dbms_id is None:
+            needs_review.append({
+                "kind": "bug", "url": url,
+                "reason": (f"database system {entry.get('system')!r} is not in "
+                           f"the registry"),
+            })
+            continue
+
+        if inspected >= max_reports:
+            needs_review.append({
+                "kind": "bug", "url": url,
+                "reason": "not inspected yet; run budget reached",
+            })
+            continue
+
+        text, source_type = _fetch_report(gh, fetcher, url)
+        inspected += 1
+        if not text:
+            needs_review.append({
+                "kind": "bug", "url": url,
+                "reason": "bug report could not be fetched",
+            })
+            continue
+
+        attribution = deterministic_attribution(text, tax)
+        classifier_record = None
+        confidence = "high"
+
+        if attribution is None:
+            # The lab's own list says this report came from one of its members,
+            # and its reproducer carries SQLancer's generated-schema convention.
+            # Neither is enough alone -- the list records who reported a bug but
+            # not which tool found it, and the convention says only that a
+            # SQLancer-family generator produced the SQL. Together they place
+            # the report in a SQLancer campaign, which is what the policy's
+            # campaign clause asks for. Many reports never mention the tool at
+            # all, so without this the whole of a system like Dolt is invisible.
+            signature = reproducer_excerpt(text)
+            if (signature and entry.get("reported_by")
+                    and not tax.find_excluded_techniques(text)):
+                attribution = ("campaign_evidence", None, "sqlancer", signature)
+                confidence = "medium"
+
+        if attribution is None:
+            if not (tax.mentions_finder(text) or tax.find_techniques(text)):
+                rejected.append({
+                    "kind": "bug", "url": url,
+                    "reason": ("TEST lab bug with no SQLancer tool or technique "
+                               "named in the report"),
+                })
+                continue
+            # Ask before giving up: the answer may be cached, and a
+            # cached answer needs no key. Checking availability first
+            # made a warm cache useless offline.
+            result = None if classifier is None else classifier.classify(
+                "bug_attribution", source_id=f"nus-test:{canonical}",
+                source_text=text, source_label=f"bug report {url}",
+                technique_names=technique_names(tax),
+                finder_names=finder_names(tax),
+                excluded_names=excluded_names(tax))
+            if result is None:
+                needs_review.append({
+                    "kind": "bug", "url": url, "title": entry.get("title"),
+                    "reason": ("mentions SQLancer but does not state it found "
+                               "the bug; needs semantic classification"),
+                })
+                continue
+            if not result.is_positive:
+                rejected.append({
+                    "kind": "bug", "url": url, "answer": result.answer,
+                    "reason": result.reason or "classifier did not confirm",
+                })
+                continue
+            technique = resolve_technique(tax, result.extra.get("technique"))
+            finder = resolve_finder(tax, result.extra.get("finder")) or "sqlancer"
+            attribution = (
+                "technique_attribution" if technique else "explicit_tool_statement",
+                technique, finder, result.excerpts[0])
+            classifier_record = result.classifier_record()
+            confidence = "medium"
+
+        rule, technique, finder, excerpt = attribution
+        if not verbatim_excerpt(text, excerpt):
+            continue
+
+        reported_date = parse_date(entry.get("created_at"))
+        status, is_true_positive = _status_of(entry)
+
+        evidence = [{
+            "source_url": url,
+            "source_type": source_type or "issue_tracker",
+            "excerpt": excerpt,
+            "excerpt_is_verbatim": True,
+            "note": ("The report states which tool or oracle found the bug."
+                     if rule != "campaign_evidence" else
+                     f"Reported by {entry.get('reported_by')} of the TEST lab, "
+                     f"with a reproducer using SQLancer's generated-schema "
+                     f"convention (tables t0, t1, ...; columns c0, c1, ...)."),
+            "content_sha256": content_hash(text),
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        }, {
+            "source_url": PAGE_URL,
+            "source_type": "website",
+            "excerpt": None,
+            "note": (f"Listed on the NUS TEST lab bug page, reported by "
+                     f"{entry.get('reported_by') or 'a lab member'}."),
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        }]
+
+        record = {
+            "id": f"bug:{dbms_id}:{stable_digest(canonical)}",
+            "dbms": dbms_id,
+            "title": truncate(entry.get("title") or "(untitled report)", 500),
+            "reported_date": reported_date,
+            "reported_year": year_of(reported_date),
+            "status": status,
+            "status_is_true_positive": is_true_positive,
+            "finder": finder,
+            "technique": technique,
+            "symptom": "unknown",
+            "reporter": entry.get("reported_by"),
+            "links": {"report": url},
+            "primary_url": canonical,
+            "attribution": {
+                "rule": rule,
+                "confidence": confidence,
+                "evidence": evidence,
+            },
+            "provenance": {
+                "collector": COLLECTOR,
+                "collector_version": COLLECTOR_VERSION,
+                "policy_version": config.policy_version(),
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+                "source_url": DATA_URL,
+                "source_type": "website",
+                "content_sha256": content_hash(text),
+            },
+        }
+        if classifier_record:
+            record["attribution"]["classifier"] = classifier_record
+        records.append(record)
+
+    return records, rejected, needs_review
+
+
+def emailed_records(entries: Sequence[dict], alias_map: Dict[str, str], *,
+                    timestamp: Optional[str] = None
+                    ) -> Tuple[List[dict], List[dict]]:
+    """Records for bugs the lab reported privately, from the list alone.
+
+    These are admitted under ``curated_primary_source``: the lab's bug list is
+    the record its own members keep of what they found, which is the same
+    standing as the SQLancer project's bug repository. Confidence is medium
+    rather than high because the report itself is not public, so unlike a
+    tracker entry it cannot be re-read to check what it says.
+
+    No ``primary_url`` is set. There is no public report to point at, and
+    pointing every one of them at the list page would make 55 bugs look like
+    one record duplicated.
+    """
+    timestamp = timestamp or now()
+    records: List[dict] = []
+    needs_review: List[dict] = []
+
+    for entry in emailed(entries):
+        system = (entry.get("system") or "").strip()
+        dbms_id = alias_map.get(system.lower())
+        if dbms_id is None:
+            needs_review.append({
+                "kind": "bug", "url": PAGE_URL,
+                "reason": (f"database system {system!r} is not in the "
+                           f"registry, and this report has no link"),
+                "title": entry.get("title"),
+            })
+            continue
+
+        title = (entry.get("title") or "").strip()
+        reported_date = parse_date(entry.get("created_at"))
+        status, is_true_positive = _status_of(entry)
+        reporter = entry.get("reported_by")
+        # The list is the source, so its own row is the evidence. The title is
+        # quoted verbatim because that is what the page shows.
+        fingerprint = "|".join([dbms_id, title, entry.get("created_at") or ""])
+
+        records.append({
+            "id": f"bug:{dbms_id}:{stable_digest(fingerprint)}",
+            "dbms": dbms_id,
+            "title": truncate(title or "(untitled report)", 500),
+            "reported_date": reported_date,
+            "reported_year": year_of(reported_date),
+            "status": status,
+            "status_is_true_positive": is_true_positive,
+            "finder": "sqlancer",
+            "technique": None,
+            "symptom": "unknown",
+            "reporter": reporter,
+            "links": {"record": PAGE_URL},
+            "attribution": {
+                "rule": "curated_primary_source",
+                "confidence": "medium",
+                "evidence": [{
+                    "source_url": PAGE_URL,
+                    "source_type": "website",
+                    "excerpt": truncate(title, 500) or None,
+                    "excerpt_is_verbatim": bool(title),
+                    "note": (f"Listed on the NUS TEST lab bug page as a "
+                             f"{system} bug reported by "
+                             f"{reporter or 'a lab member'}. The report was "
+                             f"sent to the developers directly, so the list is "
+                             f"its only public record."),
+                    "content_sha256": content_hash(fingerprint),
+                    "retrieved_at": timestamp,
+                    "first_seen": timestamp,
+                    "last_verified": timestamp,
+                }],
+            },
+            "provenance": {
+                "collector": COLLECTOR,
+                "collector_version": COLLECTOR_VERSION,
+                "policy_version": config.policy_version(),
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+                "source_url": PAGE_URL,
+                "source_type": "website",
+                "content_sha256": content_hash(fingerprint),
+            },
+        })
+    return records, needs_review
diff --git a/tools/impact/collectors/papers.py b/tools/impact/collectors/papers.py
new file mode 100644
index 0000000..b10a89d
--- /dev/null
+++ b/tools/impact/collectors/papers.py
@@ -0,0 +1,635 @@
+"""Discovers papers that cite SQLancer and prepares them for classification.
+
+Discovery is entirely deterministic: every paper reached here was found by
+walking the citation graph of the foundational SQLancer publications, so the
+``references`` relationship is a fact about the graph rather than a judgement.
+
+Two graphs are walked, because neither is complete. Semantic Scholar indexes
+preprints well and -- decisively -- returns ``contexts``, the sentences in the
+citing paper that surround the citation, which are verbatim source text and so
+usable directly as evidence. OpenAlex has different coverage and stays available
+when Semantic Scholar's shared pool is rate limiting. OpenCitations is built
+from Crossref's own reference data and holds edges neither of the others
+returned, at the cost of giving back nothing but a pair of DOIs -- and of being
+blind to a seed with no DOI, which PQS, the most cited of them, is. Results are
+merged on the most stable identifier available.
+
+The three stronger relationships are judgements, and this module does not make
+them. It assembles the evidence a classifier needs and flags which candidates
+are worth asking about. A paper with no signal at all is recorded as
+``insufficient_evidence`` and never reaches the model, which keeps the weekly
+token cost proportional to new work rather than to corpus size.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Dict, Iterable, List, Optional, Sequence, Tuple
+
+from .. import config, taxonomy
+from ..scholarly import (OpenAlex, OpenCitations, SemanticScholar,
+                         _normalise_title, _openalex_id, resolve_seed)
+from ..util import (content_hash, dedupe_preserving_order, normalize_doi,
+                    normalize_url, truncate)
+
+COLLECTOR = "papers"
+COLLECTOR_VERSION = "1.0.0"
+EXTRACTOR_VERSION = "paper-signals-v1"
+
+# Lower bound on a usable citation context. Shorter fragments are index noise.
+MIN_CONTEXT_CHARS = 40
+MAX_CONTEXT_CHARS = 1200
+MAX_EVIDENCE_PER_RELATIONSHIP = 6
+
+# Deterministic pre-filters. These decide only whether a candidate is worth a
+# semantic question; they never decide the answer.
+COMPARISON_CUES = re.compile(
+    r"\b(baselines?|compared?|compares|comparison|comparing|evaluated?\s+"
+    r"against|state-of-the-art|state\s+of\s+the\s+art|outperform\w*|"
+    r"versus|vs\.?|against\s+(?:sqlancer|tlp|norec|pqs|qpg)|"
+    r"than\s+(?:sqlancer|tlp|norec|pqs|qpg))\b",
+    re.IGNORECASE)
+
+EXTENSION_CUES = re.compile(
+    r"\b(extend\w*|generali[sz]\w*|adapt\w*|builds?\s+(?:up)?on|"
+    r"based\s+on|inspired\s+by|improves?\s+(?:up)?on|augment\w*|"
+    r"we\s+follow|derived\s+from|variant\s+of)\b",
+    re.IGNORECASE)
+
+INFRASTRUCTURE_CUES = re.compile(
+    r"\b(implemented?\s+(?:it\s+)?(?:on\s+top\s+of|in|within|upon)|"
+    r"built?\s+(?:on\s+top\s+of|upon)|on\s+top\s+of\s+sqlancer|"
+    r"prototype\s+(?:in|on)|our\s+implementation|fork\w*|"
+    r"extend\w*\s+sqlancer|integrat\w*\s+into\s+sqlancer)\b",
+    re.IGNORECASE)
+
+ARXIV_DOI = re.compile(r"^10\.48550/arxiv\.(.+)$", re.IGNORECASE)
+
+# "state of the art", however it is hyphenated. Deliberately narrow: this is a
+# claim about standing, and looser wording ("widely used", "popular", "the most
+# effective") is a different claim that would need its own judgement.
+STATE_OF_THE_ART = re.compile(r"state[-\s]of[-\s]the[-\s]art", re.IGNORECASE)
+
+
+def state_of_the_art_contexts(contexts: Iterable[Tuple[str, str]],
+                              tax: Optional["taxonomy.Taxonomy"] = None
+                              ) -> List[Tuple[str, str]]:
+    """Contexts that call SQLancer or one of its techniques state of the art.
+
+    A citation context is a single sentence, so a sentence containing both the
+    phrase and a SQLancer name is almost always saying the one about the other
+    -- which is why this needs no model. Techniques the taxonomy excludes never
+    match, so a sentence naming only EET or DQE is not recognition of SQLancer.
+    """
+    tax = tax or taxonomy.load()
+    found: List[Tuple[str, str]] = []
+    for technique_id, context in contexts:
+        if not STATE_OF_THE_ART.search(context):
+            continue
+        matches = tax.find_techniques(context)
+        mentions = tax.mentions_finder(context)
+        if not matches and not mentions:
+            continue
+        attributed = matches[0].technique_id if matches else technique_id
+        if (attributed, context) not in found:
+            found.append((attributed, context))
+    return found
+
+# Titles the scholarly indexes sometimes produce from a PDF's cover page rather
+# than from the paper. A record like this cannot be presented or plotted, so it
+# is held back for a human rather than published with a nonsense title.
+BOILERPLATE_TITLE = re.compile(
+    r"^(this paper is included|proceedings of the|open access to the|"
+    r"the following paper|table of contents|front matter|"
+    r"conference (?:proceedings|program))",
+    re.IGNORECASE)
+
+
+def usable_metadata(candidate: "Candidate") -> Optional[str]:
+    """Why a candidate is not publishable yet, or None when it is fine."""
+    title = (candidate.title or "").strip()
+    if len(title) < 8:
+        return "no usable title from the scholarly index"
+    if BOILERPLATE_TITLE.match(title):
+        return "index returned proceedings boilerplate instead of a title"
+    if not candidate.s2.get("year"):
+        return "no publication year from the scholarly index"
+    return None
+
+
+def _clean_context(text: str) -> Optional[str]:
+    """Normalise a citation context, or reject it as too short to be evidence."""
+    if not text:
+        return None
+    collapsed = re.sub(r"\s+", " ", text).strip()
+    if len(collapsed) < MIN_CONTEXT_CHARS:
+        return None
+    return truncate(collapsed, MAX_CONTEXT_CHARS)
+
+
+def paper_identity(external_ids: Dict[str, str], fallback_id: str
+                   ) -> Tuple[str, Optional[str], Optional[str]]:
+    """Return ``(record_id, doi, arxiv_id)`` preferring the most stable id.
+
+    A DataCite arXiv DOI is folded back into a plain arXiv id, so the same
+    preprint reached through OpenAlex (which reports the DOI) and through
+    Semantic Scholar (which reports the arXiv id) lands on one record.
+    """
+    doi = normalize_doi(external_ids.get("DOI"))
+    arxiv = external_ids.get("ArXiv")
+    if doi:
+        match = ARXIV_DOI.match(doi)
+        if match:
+            arxiv = arxiv or match.group(1)
+            doi = None
+    if doi:
+        return f"paper:doi:{doi}", doi, arxiv
+    if arxiv:
+        return f"paper:arxiv:{arxiv}", None, arxiv
+    return fallback_id, None, None
+
+
+def _best_url(doi: Optional[str], arxiv: Optional[str],
+              s2_paper_id: Optional[str], openalex_id: Optional[str]) -> str:
+    if doi:
+        return f"https://doi.org/{doi}"
+    if arxiv:
+        return f"https://arxiv.org/abs/{arxiv}"
+    if s2_paper_id:
+        return f"https://www.semanticscholar.org/paper/{s2_paper_id}"
+    return f"https://openalex.org/{openalex_id}"
+
+
+def normalise_openalex(work: dict) -> dict:
+    """Project an OpenAlex work onto the Semantic Scholar citingPaper shape."""
+    ids = work.get("ids") or {}
+    external = {}
+    if work.get("doi"):
+        external["DOI"] = work["doi"]
+    location = work.get("primary_location") or {}
+    source = (location.get("source") or {}) if isinstance(location, dict) else {}
+    best_oa = work.get("best_oa_location") or {}
+    return {
+        "paperId": None,
+        "openalexId": _openalex_id(work.get("id", "")),
+        "externalIds": external,
+        "title": work.get("display_name") or work.get("title") or "",
+        "year": work.get("publication_year"),
+        "venue": source.get("display_name"),
+        "authors": [{"name": (a.get("author") or {}).get("display_name")}
+                    for a in (work.get("authorships") or [])],
+        "openAccessPdf": {"url": best_oa.get("pdf_url")} if best_oa else {},
+        "abstract": None,
+        "_openalex_ids": ids,
+    }
+
+
+class Candidate:
+    """A citing paper accumulated across seeds and across both citation graphs."""
+
+    def __init__(self, citing: dict):
+        self.external_ids = dict(citing.get("externalIds") or {})
+        self.s2_paper_id = citing.get("paperId")
+        self.openalex_id = citing.get("openalexId")
+        self.record_id, self.doi, self.arxiv = paper_identity(
+            self.external_ids,
+            f"paper:s2:{self.s2_paper_id}" if self.s2_paper_id
+            else f"paper:openalex:{self.openalex_id}")
+        self.s2 = dict(citing)
+        self.seeds: List[str] = []
+        # (technique_id, context) pairs; the technique identifies the cited seed.
+        self.contexts: List[Tuple[str, str]] = []
+        self.intents: List[str] = []
+        self.influential = False
+        self.sources: List[str] = []
+
+    @property
+    def title(self) -> str:
+        return self.s2.get("title") or ""
+
+    def merge_metadata(self, other: dict) -> None:
+        """Fill gaps from a second source without overwriting what we have."""
+        for key in ("title", "year", "venue", "abstract"):
+            if not self.s2.get(key) and other.get(key):
+                self.s2[key] = other[key]
+        if not self.s2.get("authors") and other.get("authors"):
+            self.s2["authors"] = other["authors"]
+        if not (self.s2.get("openAccessPdf") or {}).get("url"):
+            if (other.get("openAccessPdf") or {}).get("url"):
+                self.s2["openAccessPdf"] = other["openAccessPdf"]
+        self.s2_paper_id = self.s2_paper_id or other.get("paperId")
+        self.openalex_id = self.openalex_id or other.get("openalexId")
+        for key, value in (other.get("externalIds") or {}).items():
+            self.external_ids.setdefault(key, value)
+
+    def add_edge(self, technique_id: str, edge: dict, source: str) -> None:
+        if technique_id not in self.seeds:
+            self.seeds.append(technique_id)
+        if source not in self.sources:
+            self.sources.append(source)
+        for raw in edge.get("contexts") or []:
+            context = _clean_context(raw)
+            if context and (technique_id, context) not in self.contexts:
+                self.contexts.append((technique_id, context))
+        self.intents.extend(edge.get("intents") or [])
+        self.influential = self.influential or bool(edge.get("isInfluential"))
+
+    # -- deterministic signals -------------------------------------------
+
+    def matching_contexts(self, pattern: re.Pattern) -> List[Tuple[str, str]]:
+        return [(tech, ctx) for tech, ctx in self.contexts if pattern.search(ctx)]
+
+    def signals(self) -> Dict[str, List[Tuple[str, str]]]:
+        """Contexts suggesting each stronger relationship, for the classifier."""
+        return {
+            "compares_with": self.matching_contexts(COMPARISON_CUES),
+            "extends_technique": self.matching_contexts(EXTENSION_CUES),
+            "uses_infrastructure": self.matching_contexts(INFRASTRUCTURE_CUES),
+        }
+
+    def has_any_signal(self) -> bool:
+        return any(self.signals().values())
+
+    def evidence_hash(self) -> str:
+        """Stable hash of everything a classifier would be shown.
+
+        Keying the classification cache on this is what guarantees unchanged
+        evidence is never sent to the model twice.
+        """
+        parts = [self.record_id, self.title, self.s2.get("abstract") or ""]
+        parts.extend(f"{tech}:{ctx}" for tech, ctx in sorted(self.contexts))
+        return content_hash("\n".join(parts))
+
+
+def _merge_into(candidates: Dict[str, Candidate], citing: dict,
+                technique_id: str, edge: dict, source: str,
+                by_title: Dict[Tuple[str, Optional[int]], str]) -> None:
+    """Add one citation edge, merging by identifier and then by title+year."""
+    candidate = Candidate(citing)
+    key = candidate.record_id
+    existing = candidates.get(key)
+    if existing is None:
+        # Same work reached through the other graph without a shared identifier.
+        title_key = (_normalise_title(candidate.title), citing.get("year"))
+        if title_key[0] and title_key in by_title:
+            existing = candidates[by_title[title_key]]
+    if existing is None:
+        candidates[key] = candidate
+        title_key = (_normalise_title(candidate.title), citing.get("year"))
+        if title_key[0]:
+            by_title.setdefault(title_key, key)
+        existing = candidate
+    else:
+        existing.merge_metadata(citing)
+    existing.add_edge(technique_id, edge, source)
+
+
+def gather(*, s2: Optional[SemanticScholar] = None,
+           oa: Optional[OpenAlex] = None,
+           oc: Optional["OpenCitations"] = None,
+           seeds: Optional[Sequence[dict]] = None,
+           progress: bool = False) -> Tuple[Dict[str, Candidate], List[dict]]:
+    """Walk both citation graphs. Returns ``(candidates, unresolved_seeds)``.
+
+    A failure of one source is logged and skipped rather than fatal: losing
+    Semantic Scholar for a week must not empty the dataset, and the periodic
+    reconciliation picks up whatever a degraded run missed.
+    """
+    tax = taxonomy.load()
+    seeds = list(seeds if seeds is not None else tax.citation_seeds())
+    candidates: Dict[str, Candidate] = {}
+    by_title: Dict[Tuple[str, Optional[int]], str] = {}
+    unresolved: List[dict] = []
+
+    for seed in seeds:
+        # The seed's own id, which is a technique for an oracle paper and the
+        # tool for SQLancer++'s. What a citing paper records is which
+        # foundational publication it cites, and both are that.
+        technique_id = seed.get("seed_id") or seed["technique_id"]
+        found_any = False
+
+        if s2 is not None:
+            try:
+                resolved = resolve_seed(s2, seed)
+            except Exception as error:  # network/rate-limit; try the other graph
+                resolved = None
+                if progress:
+                    print(f"  seed {technique_id}: semantic scholar failed "
+                          f"({type(error).__name__})")
+            if resolved:
+                found_any = True
+                count = 0
+                try:
+                    for edge in s2.citations(resolved["paperId"]):
+                        citing = edge.get("citingPaper") or {}
+                        if not citing.get("paperId"):
+                            continue
+                        _merge_into(candidates, citing, technique_id, edge,
+                                    "semantic_scholar", by_title)
+                        count += 1
+                except Exception as error:
+                    if progress:
+                        print(f"  seed {technique_id}: citation paging stopped "
+                              f"({type(error).__name__})")
+                if progress:
+                    print(f"  seed {technique_id}: {count} citations from "
+                          f"Semantic Scholar")
+
+        if oa is not None:
+            try:
+                work_ids = oa.versions_of(seed)
+            except Exception:
+                work_ids = []
+            count = 0
+            for work_id in work_ids:
+                found_any = found_any or True
+                try:
+                    for work in oa.citing_works(work_id):
+                        citing = normalise_openalex(work)
+                        if not citing["title"]:
+                            continue
+                        _merge_into(candidates, citing, technique_id,
+                                    {"contexts": [], "intents": []},
+                                    "openalex", by_title)
+                        count += 1
+                except Exception:
+                    continue
+            if progress and work_ids:
+                print(f"  seed {technique_id}: {count} citations from OpenAlex "
+                      f"({len(work_ids)} version(s))")
+
+        if oc is not None and seed.get("doi"):
+            # A third graph, from Crossref's own reference data. It returns
+            # identifiers and nothing else, so a DOI already reached through
+            # either of the other two is skipped rather than resolved again:
+            # the metadata is better there, and this call costs a request.
+            try:
+                citing_dois = oc.citing_dois(seed["doi"])
+            except Exception as error:
+                citing_dois = []
+                if progress:
+                    print(f"  seed {technique_id}: OpenCitations failed "
+                          f"({type(error).__name__})")
+            seen_dois = {normalize_doi(c.doi) for c in candidates.values()
+                         if c.doi}
+            count = 0
+            for doi in citing_dois:
+                if doi in seen_dois:
+                    continue
+                found_any = True
+                try:
+                    citing = oc.work_by_doi(doi)
+                except Exception:
+                    continue
+                if not citing or not citing.get("title"):
+                    continue
+                # No contexts: the edge carries the fact of the citation and
+                # nothing else, which is exactly what it is evidence of.
+                _merge_into(candidates, citing, technique_id,
+                            {"contexts": [], "intents": []},
+                            "opencitations", by_title)
+                seen_dois.add(doi)
+                count += 1
+            if progress and citing_dois:
+                print(f"  seed {technique_id}: {len(citing_dois)} citations "
+                      f"from OpenCitations, {count} new")
+
+        if not found_any:
+            unresolved.append(seed)
+        if progress:
+            print(f"  -> {len(candidates)} distinct papers so far")
+
+    return candidates, unresolved
+
+
+def _relationship(value: str, method: str, *, evidence: Optional[List[dict]] = None,
+                  techniques: Optional[List[str]] = None,
+                  classifier: Optional[dict] = None) -> dict:
+    relationship = {"value": value, "method": method}
+    if techniques:
+        relationship["techniques"] = dedupe_preserving_order(techniques)
+    if evidence:
+        relationship["evidence"] = evidence
+    if classifier:
+        relationship["classifier"] = classifier
+    return relationship
+
+
+def context_evidence(url: str, pairs: Iterable[Tuple[str, str]], note: str,
+                     timestamp: str) -> List[dict]:
+    """Turn citation contexts into evidence items.
+
+    The excerpt is the context exactly as the index returned it; nothing here
+    rewrites or summarises it.
+    """
+    tax = taxonomy.load()
+    evidence = []
+    for technique_id, context in list(pairs)[:MAX_EVIDENCE_PER_RELATIONSHIP]:
+        evidence.append({
+            "source_url": url,
+            "source_type": "paper_citation_context",
+            "excerpt": context,
+            "excerpt_is_verbatim": True,
+            "note": note.format(technique=tax.display_name(technique_id)),
+            "content_sha256": content_hash(context),
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        })
+    return evidence
+
+
+def select_reference_contexts(candidate: Candidate
+                              ) -> List[Tuple[str, str]]:
+    """Choose which citation contexts to publish as evidence.
+
+    An index often returns the same sentence once per cited seed, and the first
+    few contexts are frequently the least informative ones. So duplicates are
+    dropped, sentences that say something about the relationship are preferred,
+    and at least one context per cited seed is kept so every seed the paper
+    cites is represented.
+    """
+    seen_text = set()
+    scored: List[Tuple[int, int, int, str, str]] = []
+    for index, (technique_id, context) in enumerate(candidate.contexts):
+        key = context.strip().lower()
+        if key in seen_text:
+            continue
+        seen_text.add(key)
+        informative = any(cue.search(context) for cue in
+                          (INFRASTRUCTURE_CUES, EXTENSION_CUES, COMPARISON_CUES))
+        scored.append((0 if informative else 1, -len(context), index,
+                       technique_id, context))
+    scored.sort()
+
+    chosen: List[Tuple[str, str]] = []
+    covered = set()
+    for _, _, _, technique_id, context in scored:
+        if len(chosen) >= MAX_EVIDENCE_PER_RELATIONSHIP:
+            break
+        chosen.append((technique_id, context))
+        covered.add(technique_id)
+    # Make sure no cited seed is left without a context of its own.
+    for _, _, _, technique_id, context in scored:
+        if technique_id in covered:
+            continue
+        if chosen:
+            chosen.pop()
+        chosen.append((technique_id, context))
+        covered.add(technique_id)
+    return chosen
+
+
+def build_record(candidate: Candidate, *, timestamp: str,
+                 classifications: Optional[Dict[str, dict]] = None,
+                 artifacts: Optional[List[dict]] = None,
+                 is_sqlancer_publication: bool = False) -> dict:
+    """Assemble a paper record.
+
+    ``references`` is always ``yes`` -- the paper is here because the citation
+    graph says so. The other three default to ``insufficient_evidence`` and are
+    only raised by a supplied classification, so an unclassified paper is never
+    silently counted as building on SQLancer.
+    """
+    tax = taxonomy.load()
+    s2 = candidate.s2
+    url = _best_url(candidate.doi, candidate.arxiv, candidate.s2_paper_id,
+                    candidate.openalex_id)
+    classifications = classifications or {}
+
+    reference_pairs = select_reference_contexts(candidate)
+    if reference_pairs:
+        reference_evidence = context_evidence(
+            url, reference_pairs,
+            "Sentence in this paper citing {technique}, as indexed by "
+            "Semantic Scholar.", timestamp)
+    else:
+        # No indexed full text: the citation edge itself is the evidence.
+        seed_names = ", ".join(tax.display_name(t) for t in candidate.seeds)
+        index_url = (f"https://www.semanticscholar.org/paper/{candidate.s2_paper_id}"
+                     if candidate.s2_paper_id
+                     else f"https://openalex.org/{candidate.openalex_id}")
+        index_name = ("Semantic Scholar" if candidate.s2_paper_id else "OpenAlex")
+        reference_evidence = [{
+            "source_url": index_url,
+            "source_type": "paper",
+            "excerpt": None,
+            "note": (f"{index_name} records this paper as citing the "
+                     f"publication that introduced {seed_names}."),
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        }]
+
+    relationships = {
+        "references": _relationship(
+            "yes", "citation_graph", evidence=reference_evidence,
+            techniques=candidate.seeds),
+    }
+    for key in ("uses_infrastructure", "extends_technique", "compares_with"):
+        relationships[key] = classifications.get(key) or _relationship(
+            "insufficient_evidence", "deterministic")
+
+    # Recognition is read straight off the citation contexts: the sentence that
+    # calls SQLancer state of the art is itself the evidence.
+    sota = classifications.get("describes_as_state_of_the_art")
+    if sota is None:
+        pairs = state_of_the_art_contexts(candidate.contexts, tax)
+        if pairs:
+            sota = _relationship(
+                "yes", "deterministic",
+                techniques=[technique for technique, _ in pairs],
+                evidence=context_evidence(
+                    url, pairs,
+                    "Sentence in this paper describing {technique} as state of "
+                    "the art.", timestamp))
+        else:
+            sota = _relationship("no", "deterministic")
+    relationships["describes_as_state_of_the_art"] = sota
+
+    authors = [a.get("name") for a in (s2.get("authors") or []) if a.get("name")]
+    record = {
+        "id": candidate.record_id,
+        "title": s2.get("title") or "(untitled)",
+        "authors": authors[:60],
+        "year": s2.get("year"),
+        "venue": s2.get("venue") or None,
+        "doi": candidate.doi,
+        "arxiv_id": candidate.arxiv,
+        "s2_paper_id": candidate.s2_paper_id,
+        "url": url,
+        "open_access_pdf": normalize_url((s2.get("openAccessPdf") or {}).get("url")),
+        # Kept for the paper's note. For a paywalled paper it is the only
+        # description of the work we can obtain at all, and a summary written
+        # from an abstract is worth far more than one written from two
+        # citation sentences.
+        "abstract": truncate(s2.get("abstract") or "", 6000) or None,
+        "cites_seed_techniques": candidate.seeds,
+        "relationships": relationships,
+        "provenance": {
+            "collector": COLLECTOR,
+            "collector_version": COLLECTOR_VERSION,
+            "policy_version": config.policy_version(),
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+            "source_url": (f"https://www.semanticscholar.org/paper/"
+                           f"{candidate.s2_paper_id}" if candidate.s2_paper_id
+                           else f"https://openalex.org/{candidate.openalex_id}"),
+            "source_type": "paper",
+            "content_sha256": candidate.evidence_hash(),
+        },
+    }
+    if artifacts:
+        record["artifacts"] = artifacts
+    if is_sqlancer_publication:
+        record["is_sqlancer_publication"] = True
+    return record
+
+
+def _norm_title(title: str) -> str:
+    return "".join(ch for ch in (title or "").lower() if ch.isalnum())
+
+
+def is_project_publication(candidate: Candidate,
+                           tax: Optional[taxonomy.Taxonomy] = None) -> bool:
+    """Whether this candidate *is* one of the taxonomy's own publications.
+
+    The foundational papers cite each other, so they turn up in their own
+    citation graphs. Matching them on identifier or title is exact, which is
+    better than inferring authorship.
+    """
+    tax = tax or taxonomy.load()
+    papers = [technique.get("origin_paper")
+              for technique in tax.techniques.values()]
+    papers += [finder.get("paper") for finder in tax.finders.values()]
+    papers += tax.project_publications
+    for paper in papers:
+        if not paper:
+            continue
+        if candidate.doi and normalize_doi(paper.get("doi")) == candidate.doi:
+            return True
+        if candidate.arxiv and paper.get("arxiv_id") == candidate.arxiv:
+            return True
+        if (candidate.s2_paper_id
+                and paper.get("s2_paper_id") == candidate.s2_paper_id):
+            return True
+        if _norm_title(paper.get("title", "")) == _norm_title(candidate.title):
+            return True
+    return False
+
+
+def looks_like_sqlancer_publication(candidate: Candidate) -> bool:
+    """Whether a citing paper is itself part of the SQLancer project.
+
+    Either it is a publication the taxonomy names -- a technique's origin paper,
+    a tool's paper, or one listed explicitly -- or one of the project's authors
+    is on it. Co-authorship alone settles it: the project's own papers are not
+    external work building on SQLancer, whatever their subject, and a rule that
+    also demanded a SQLancer name in the title missed papers whose titles are
+    about something else.
+    """
+    tax = taxonomy.load()
+    if is_project_publication(candidate, tax):
+        return True
+    return any(tax.is_project_author(author.get("name"))
+               for author in (candidate.s2.get("authors") or []))
diff --git a/tools/impact/collectors/people.py b/tools/impact/collectors/people.py
new file mode 100644
index 0000000..1824772
--- /dev/null
+++ b/tools/impact/collectors/people.py
@@ -0,0 +1,354 @@
+"""The roster of people who run SQLancer campaigns, and their bug reports.
+
+Author search is the one route into reports that never say "SQLancer". A term
+search cannot find them, and neither can a label; but a report filed by someone
+who spends their time testing database systems with SQLancer, carrying a
+SQLancer-shaped reproducer, in a database system's own tracker, is a report the
+dataset should at least look at.
+
+That route needs a handle, and this is where the handles come from. The TEST
+lab's bug list is the largest record of who filed what, and it is inconsistent:
+some entries name a GitHub handle, some a display name, and one -- ``ShuxinLi``
+-- names something that looks like a handle and is not. Roughly four hundred
+entries were unreachable for that reason alone.
+
+So the roster is resolved rather than guessed. For a reporter with no usable
+handle, the entries themselves say which issues they filed; fetching a sample
+and reading each issue's author gives the handle, and the answer is only
+accepted when every sample agrees. A person who reports on the SQLite forum or
+a Jira has no GitHub profile to find, and stays in the roster without one.
+
+The roster grants no bug anything. It decides whose issues are worth fetching;
+whether an issue is a SQLancer bug is settled afterwards by the same attribution
+rules every other candidate goes through.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+from collections import Counter, defaultdict
+from typing import Dict, List, Optional, Sequence, Set, Tuple
+
+from .. import config, taxonomy
+from ..github import GitHub
+from ..http import Fetcher
+from ..util import content_hash, now, slugify
+from . import lab_members
+
+COLLECTOR = "people"
+COLLECTOR_VERSION = "1.0.0"
+
+ISSUE_URL = re.compile(
+    r"^https://github\.com/([\w.-]+)/([\w.-]+)/issues/(\d+)", re.IGNORECASE)
+
+# How many of a person's issues to read before deciding their handle, and how
+# far the samples may disagree. A single sample is not evidence: an entry can
+# record who found a bug while someone else filed it.
+RESOLVE_SAMPLES = 8
+RESOLVE_MIN_SAMPLES = 3
+
+# Bot and service accounts that file issues on people's behalf.
+NOT_A_PERSON = re.compile(r"(\[bot\]|-bot$|^bot-|^dependabot|^github-actions)",
+                          re.IGNORECASE)
+
+
+def _lab_entries(gh: GitHub) -> List[dict]:
+    raw, _ = gh.raw_file(lab_members.BUGS_OWNER, lab_members.BUGS_REPO,
+                         lab_members.BUGS_PATH, lab_members.BUGS_REF,
+                         max_age_days=7)
+    if not raw:
+        return []
+    try:
+        entries = json.loads(raw)
+    except ValueError:
+        return []
+    return entries if isinstance(entries, list) else []
+
+
+def _looks_like_a_handle(name: str) -> bool:
+    """Whether a ``reported_by`` value could be a GitHub login.
+
+    A login has no spaces. That is a weak test and it is meant to be: a value
+    that passes it is still resolved against real issues before being trusted.
+    """
+    return bool(name) and " " not in name and len(name) <= 39
+
+
+def resolve_handle(gh: GitHub, urls: Sequence[str]
+                   ) -> Tuple[Optional[str], List[str]]:
+    """The GitHub login that filed these issues, and the ones that showed it.
+
+    Returns ``(None, [])`` unless enough issues could be read and every one of
+    them names the same author -- a reporter whose samples disagree is left for
+    a person to sort out rather than resolved to whichever login won.
+    """
+    logins: Counter = Counter()
+    evidence: Dict[str, str] = {}
+    for url in urls:
+        if len(logins) and sum(logins.values()) >= RESOLVE_SAMPLES:
+            break
+        match = ISSUE_URL.match(url or "")
+        if not match:
+            continue
+        try:
+            issue, _ = gh.issue(match.group(1), match.group(2),
+                                int(match.group(3)), max_age_days=365)
+        except Exception:
+            continue
+        login = ((issue or {}).get("user") or {}).get("login")
+        if not login or NOT_A_PERSON.search(login):
+            continue
+        logins[login] += 1
+        evidence.setdefault(login, url)
+
+    total = sum(logins.values())
+    if total < RESOLVE_MIN_SAMPLES or len(logins) != 1:
+        return None, []
+    login = next(iter(logins))
+    return login, [evidence[login]]
+
+
+def _evidence(source_url: str, note: str, excerpt: Optional[str],
+              timestamp: str) -> dict:
+    entry = {
+        "source_url": source_url,
+        "source_type": "github_issue" if "/issues/" in source_url else "website",
+        "note": note,
+        "retrieved_at": timestamp,
+        "first_seen": timestamp,
+        "last_verified": timestamp,
+    }
+    if excerpt:
+        entry["excerpt"] = excerpt
+        entry["excerpt_is_verbatim"] = True
+        entry["content_sha256"] = content_hash(excerpt)
+    return entry
+
+
+def discover(gh: GitHub, timestamp: Optional[str] = None,
+             fetcher: Optional[Fetcher] = None) -> List[dict]:
+    """Build roster records for everyone the lab's bug list credits.
+
+    Each record links the person to their GitHub profile and says how that link
+    was established, so a wrong handle is visible rather than buried.
+    """
+    timestamp = timestamp or now()
+    tax = taxonomy.load()
+
+    entries = _lab_entries(gh)
+    urls_by_name: Dict[str, List[str]] = defaultdict(list)
+    counts: Counter = Counter()
+    for entry in entries:
+        name = (entry.get("reported_by") or "").strip()
+        if not name:
+            continue
+        counts[name] += 1
+        url = entry.get("url") or ""
+        if ISSUE_URL.match(url):
+            urls_by_name[name].append(url)
+
+    # The taxonomy's contributor roster maps names to handles directly, which
+    # settles anyone whose entries are too few to resolve from issues.
+    known: Dict[str, str] = {}
+    for entry in tax.project_contributors:
+        if entry.get("name") and entry.get("github"):
+            known[entry["name"].strip().lower()] = entry["github"]
+
+    records: List[dict] = []
+    for name, count in sorted(counts.items(), key=lambda kv: (-kv[1], kv[0])):
+        handle, samples = resolve_handle(gh, urls_by_name.get(name, []))
+        if not handle:
+            handle = known.get(name.strip().lower())
+        evidence = [
+            _evidence(
+                f"https://github.com/{lab_members.BUGS_OWNER}/"
+                f"{lab_members.BUGS_REPO}/blob/{lab_members.BUGS_REF}/"
+                f"{lab_members.BUGS_PATH}",
+                f"The TEST lab's bug list credits {count} report(s) to "
+                f"“{name}”.",
+                None, timestamp),
+        ]
+        if handle:
+            for url in samples:
+                evidence.append(_evidence(
+                    url,
+                    f"This issue, recorded against “{name}”, was "
+                    f"filed on GitHub by {handle}.",
+                    None, timestamp))
+        elif _looks_like_a_handle(name):
+            # The list already records something handle-shaped. Unverified is
+            # not the same as wrong, and searching it costs one query.
+            handle = name
+
+        records.append({
+            "id": f"person:{slugify(handle or name)}",
+            "name": name,
+            "github": handle,
+            "handle_is_verified": bool(samples),
+            "reports_on_lab_list": count,
+            "role": ("Project contributor"
+                     if tax.is_project_contributor(handle or name, [])
+                     else "Bug reporter on the TEST lab's list"),
+            "active": True,
+            "evidence": evidence,
+        })
+    # The lab's bug list only names people who have already filed something on
+    # it. The people page names the rest, and since a report is now admitted on
+    # the strength of roster membership, someone missing from the roster has
+    # their bugs missed entirely rather than merely uncounted.
+    seen = {(record.get("github") or "").lower() for record in records}
+    for handle in lab_members.from_people_page(fetcher or _default_fetcher()):
+        if handle.lower() in seen:
+            continue
+        seen.add(handle.lower())
+        records.append({
+            "id": f"person:{slugify(handle)}",
+            "name": handle,
+            "github": handle,
+            "handle_is_verified": True,
+            "reports_on_lab_list": 0,
+            "role": ("Project contributor"
+                     if tax.is_project_contributor(handle, [])
+                     else "TEST lab member"),
+            "active": True,
+            "evidence": [_evidence(
+                lab_members.PEOPLE_URL,
+                f"{handle} is linked as a member on the TEST lab's people page.",
+                None, timestamp)],
+        })
+
+    return _merge_by_handle(records)
+
+
+def _default_fetcher() -> Fetcher:
+    from ..cache import Caches
+
+    return Fetcher(Caches().http, min_interval=0.5, max_retries=2, timeout=20.0)
+
+
+def _merge_by_handle(records: Sequence[dict]) -> List[dict]:
+    """Fold together entries that resolved to the same person.
+
+    The lab's list spells the same reporter more than one way -- a handle in one
+    entry and a display name in another -- and both resolve to one GitHub
+    account. Left apart they would be two roster entries with the same id, and
+    the same person searched twice.
+    """
+    merged: Dict[str, dict] = {}
+    out: List[dict] = []
+    for record in records:
+        key = record["id"]
+        first = merged.get(key)
+        if first is None:
+            merged[key] = record
+            out.append(record)
+            continue
+        first["reports_on_lab_list"] += record["reports_on_lab_list"]
+        first["handle_is_verified"] = (first["handle_is_verified"]
+                                       or record["handle_is_verified"])
+        aliases = set(first.get("also_recorded_as") or [])
+        aliases.add(record["name"])
+        aliases.discard(first["name"])
+        first["also_recorded_as"] = sorted(aliases)
+        seen = {item["source_url"] for item in first["evidence"]}
+        first["evidence"] += [item for item in record["evidence"]
+                              if item["source_url"] not in seen]
+    return out
+
+
+def roster() -> List[dict]:
+    """The committed roster."""
+    path = config.DATA_FILES.get("people")
+    if not path or not path.exists():
+        return []
+    return config.load_json(path).get("people", [])
+
+
+def handles(gh: Optional[GitHub] = None,
+            fetcher: Optional[Fetcher] = None) -> List[str]:
+    """Every handle worth searching: the roster, plus the derived sources.
+
+    The roster is the durable part; ``lab_members`` still contributes the people
+    page and the taxonomy's own contributor list, which name people who have not
+    filed anything on the lab's list yet.
+    """
+    from ..util import dedupe_preserving_order
+
+    found = [entry["github"] for entry in roster()
+             if entry.get("github") and entry.get("active", True)]
+    if gh is not None:
+        try:
+            found += lab_members.handles(gh, fetcher)
+        except Exception:
+            pass
+    return dedupe_preserving_order(found)
+
+
+def identities(gh: Optional[GitHub] = None,
+               fetcher: Optional[Fetcher] = None) -> List[str]:
+    """Every name a roster member's reports might be filed under.
+
+    Handles alone are not enough to say who found a bug. A GitHub issue records
+    a handle, but the SQLite forum and a Jira record a person's display name,
+    and the lab's own list uses both -- so matching on handles marked 220 of
+    SQLite's bugs as ours and the remaining 22, filed on the forum by the same
+    people, as somebody else's.
+    """
+    from ..util import dedupe_preserving_order
+
+    found: List[str] = []
+    for entry in roster():
+        if not entry.get("active", True):
+            continue
+        found.append(entry.get("github") or "")
+        found.append(entry.get("name") or "")
+        found.extend(entry.get("also_recorded_as") or [])
+    found += handles(gh, fetcher)
+    return dedupe_preserving_order([name for name in found if name])
+
+
+DECISIONS_FILE = "people_decisions.json"
+
+
+def decisions() -> dict:
+    """Human rulings the roster cannot make for itself.
+
+    Being on the lab's people page is evidence of being in the lab, which is
+    not the same question as belonging to the SQLancer project: a lab lists
+    everyone working in it, including people whose work is something else
+    entirely. Only someone who knows the work can tell the two apart, so the
+    answer is recorded rather than inferred.
+    """
+    from .. import config
+
+    path = config.DATA_DIR / DECISIONS_FILE
+    if not path.exists():
+        return {"counts_as_external": []}
+    try:
+        return config.load_json(path)
+    except ValueError:
+        return {"counts_as_external": []}
+
+
+def counted_as_external(ruling: Optional[dict] = None) -> Set[str]:
+    """Every name whose reports a ruling puts outside the project."""
+    ruling = decisions() if ruling is None else ruling
+    names: Set[str] = set()
+    for entry in ruling.get("counts_as_external") or []:
+        for name in [entry.get("github")] + list(entry.get("also_recorded_as") or []):
+            if name:
+                names.add(name.strip().lower())
+    return names
+
+
+def search_queries(people: Sequence[str]) -> List[str]:
+    """One unqualified issue search per person.
+
+    Deliberately without a search term. ``lab_members.search_queries`` asks for
+    a person's issues *that mention SQLancer*, which cannot reach a report whose
+    text never names it -- and those are most of them, because a bug report is
+    written about the bug. Every issue returned here still has to earn its
+    attribution afterwards.
+    """
+    return [f"author:{handle} type:issue" for handle in people if handle]
diff --git a/tools/impact/collectors/provider_bugs.py b/tools/impact/collectors/provider_bugs.py
new file mode 100644
index 0000000..039e9df
--- /dev/null
+++ b/tools/impact/collectors/provider_bugs.py
@@ -0,0 +1,350 @@
+"""Imports the bugs SQLancer's own source code records, one file per provider.
+
+Every provider package carries a ``<Name>Bugs.java`` whose fields exist to
+switch off behaviour that trips over a real defect in that system. Each field is
+introduced by a comment linking the report. That makes the file a bug list the
+project maintains as part of its working code -- it is edited when a bug is
+found and when one is fixed -- so it is admitted under the same
+``curated_primary_source`` clause as the project's own bug repository.
+
+Forks are read as well as upstream. A database vendor that maintains its own
+SQLancer fork keeps its provider's bug file there, and for a system with no
+upstream provider that file is the only public record of what SQLancer found in
+it: Oxla's twelve bugs exist nowhere else this pipeline can reach.
+
+Where the linked tracker can be read, the report's real status is fetched, so a
+fixed bug is not published as though its state were unknown. Where it cannot --
+a private Jira, a tracker that refuses robots -- the status stays unknown and
+the record says so rather than guessing.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Dict, Iterable, List, Optional, Sequence, Tuple
+
+from .. import config
+from ..github import GitHub
+from ..util import (content_hash, normalize_url, now, stable_digest, truncate,
+                    year_of)
+
+COLLECTOR = "provider_bugs"
+COLLECTOR_VERSION = "1.0.0"
+
+UPSTREAM = ("sqlancer", "sqlancer", "main")
+PROVIDER_ROOT = "src/sqlancer"
+
+# A field whose only purpose is to record a bug, and the comment above it.
+FIELD = re.compile(r"^\s*public\s+static\s+(?:final\s+)?boolean\s+(\w+)\s*=")
+COMMENT = re.compile(r"^\s*(?://+!?|\*|/\*+)\s?(.*?)\s*$")
+COMMENT_END = re.compile(r"\*/\s*$")
+URL = re.compile(r"https?://[^\s\"'<>)\]]+")
+
+# Trackers whose state can be read without credentials.
+GITHUB_ISSUE = re.compile(
+    r"github\.com/([\w.-]+)/([\w.-]+)/(?:issues|pull)/(\d+)", re.IGNORECASE)
+MARIADB_JIRA = re.compile(r"jira\.mariadb\.org/browse/([A-Z]+-\d+)")
+
+GITHUB_STATE = {
+    ("closed", "completed"): ("fixed", True),
+    ("closed", None): ("fixed", True),
+    ("closed", "not_planned"): ("closed_not_a_bug", False),
+    ("open", None): ("open", True),
+}
+
+# Jira resolutions that mean the report was not a defect. Everything else that
+# is resolved is treated as fixed, and an unresolved issue as open.
+JIRA_NOT_A_BUG = {"not a bug", "won't fix", "cannot reproduce", "duplicate",
+                  "incomplete", "invalid"}
+
+
+def provider_index() -> Dict[str, str]:
+    """Map a provider directory to the registry id of the system it tests.
+
+    The registry already records this for anything SQLancer supports upstream,
+    as ``support.provider_path``; three of them differ from the id (postgres,
+    sqlite3, yugabyte). A downstream-only provider has no such entry, so its
+    directory name is matched against ids and aliases instead.
+    """
+    index: Dict[str, str] = {}
+    for entry in config.load_json(config.DATA_FILES["dbms"])["dbms"]:
+        path = (entry.get("support") or {}).get("provider_path")
+        if path:
+            index[path.rstrip("/").split("/")[-1].lower()] = entry["id"]
+        index.setdefault(entry["id"].lower(), entry["id"])
+        for alias in entry.get("aliases") or []:
+            index.setdefault(alias.lower(), entry["id"])
+    return index
+
+
+def parse_entries(text: str) -> List[dict]:
+    """Every recorded bug in one ``<Name>Bugs.java``.
+
+    An entry is a field declaration plus the comment block directly above it,
+    which is where the link and the description live. Comments not followed by
+    a field -- the licence header, a note about the class -- belong to no entry
+    and are dropped when the next field or a blank line arrives.
+    """
+    entries: List[dict] = []
+    buffer: List[str] = []
+    for line in (text or "").splitlines():
+        comment = COMMENT.match(line)
+        if comment is not None and not FIELD.match(line):
+            body = comment.group(1)
+            if not COMMENT_END.search(line) or body:
+                buffer.append(body)
+            continue
+        field = FIELD.match(line)
+        if field is not None:
+            note = " ".join(part for part in buffer if part).strip()
+            urls = [normalize_url(u.rstrip(".,;")) for u in URL.findall(note)]
+            entries.append({
+                "constant": field.group(1),
+                "urls": [u for u in urls if u],
+                "note": note,
+                # The comment as written, for use as a verbatim excerpt.
+                "excerpt": "\n".join(buffer).strip(),
+            })
+        buffer = []
+    return [entry for entry in entries if entry["urls"]]
+
+
+def bugs_files(gh: GitHub, owner: str, repo: str, ref: str) -> List[Tuple[str, str]]:
+    """``(provider directory, path)`` for every bug file in a checkout."""
+    result = gh.tree(owner, repo, ref)
+    entries = result[0] if isinstance(result, tuple) else result
+    found: List[Tuple[str, str]] = []
+    for item in entries or []:
+        path = item["path"] if isinstance(item, dict) else str(item)
+        if not path.startswith(f"{PROVIDER_ROOT}/") or not path.endswith("Bugs.java"):
+            continue
+        parts = path.split("/")
+        if len(parts) == 4:
+            found.append((parts[2], path))
+    return found
+
+
+def github_issue(gh: GitHub, url: str) -> Optional[dict]:
+    """The linked GitHub issue, or None when the URL is not one."""
+    match = GITHUB_ISSUE.search(url)
+    if not match:
+        return None
+    owner, repo, number = match.group(1), match.group(2), int(match.group(3))
+    try:
+        result = gh.issue(owner, repo, number)
+    except Exception:
+        return None
+    issue = result[0] if isinstance(result, tuple) else result
+    return issue or None
+
+
+def _github_status(gh: GitHub, url: str) -> Optional[Tuple[str, bool]]:
+    issue = github_issue(gh, url)
+    if issue is None:
+        return None
+    state = issue.get("state")
+    reason = issue.get("state_reason")
+    return GITHUB_STATE.get((state, reason)) or GITHUB_STATE.get((state, None))
+
+
+def _jira_status(fetcher, url: str) -> Optional[Tuple[str, bool]]:
+    match = MARIADB_JIRA.search(url)
+    if not match or fetcher is None:
+        return None
+    api = (f"https://jira.mariadb.org/rest/api/2/issue/{match.group(1)}"
+           "?fields=resolution,status")
+    try:
+        entry, _ = fetcher.fetch(api, max_age_days=14)
+    except Exception:
+        return None
+    if entry.get("status") != 200:
+        return None
+    import json
+
+    try:
+        fields = json.loads(entry["body"]).get("fields") or {}
+    except ValueError:
+        return None
+    resolution = ((fields.get("resolution") or {}).get("name") or "").lower()
+    if not resolution:
+        return "open", True
+    if resolution in JIRA_NOT_A_BUG:
+        return "closed_not_a_bug", False
+    return "fixed", True
+
+
+# A tracker that answers 404 or 401 to everyone is not a link a reader can
+# follow. Oxla's Jira is private, so twelve records pointed at a page that does
+# not exist for anybody outside the company. A 403 is different -- bugs.mysql.com
+# refuses this client but serves a browser -- so it is left alone.
+UNREADABLE = (401, 404)
+
+
+def is_publicly_readable(url: str, fetcher=None) -> bool:
+    """Whether a reader following this link would reach the report.
+
+    Only a definite "not for you" counts as unreadable. Anything else, an
+    error included, leaves the link in place: failing to reach a tracker from
+    here is not evidence that nobody can.
+    """
+    if fetcher is None:
+        return True
+    try:
+        entry, _ = fetcher.fetch(url, max_age_days=30)
+    except Exception:
+        return True
+    return entry.get("status") not in UNREADABLE
+
+
+def resolve_status(gh: GitHub, url: str, fetcher=None) -> Tuple[str, bool]:
+    """The report's real state where the tracker can be read, else unknown.
+
+    An unknown state still counts as a bug: the field exists in SQLancer's
+    source because the defect is real and had to be worked around. What is
+    unknown is whether it has since been fixed, not whether it was a bug.
+    """
+    return (_github_status(gh, url) or _jira_status(fetcher, url)
+            or ("unknown", True))
+
+
+def build_record(entry: dict, dbms_id: str, *, url: str, source_url: str,
+                 repository: str, timestamp: str,
+                 status: Tuple[str, bool], readable: bool = True,
+                 title: Optional[str] = None) -> dict:
+    """Assemble a bug record from one field in a provider's bug file.
+
+    ``readable`` says whether a reader can follow the linked tracker. When
+    they cannot, the record carries no ``primary_url``: the public evidence is
+    the bug file itself, and publishing a link into a private Jira as though it
+    were the report gives a reader nothing to check.
+    """
+    state, is_true_positive = status
+    excerpt = entry["excerpt"] or entry["note"]
+    links = {"record": source_url}
+    links["report" if readable else "tracker"] = url
+    record = {
+        "id": f"bug:{dbms_id}:{stable_digest(normalize_url(url))}",
+        "dbms": dbms_id,
+        "title": truncate(title or _title_for(entry), 500),
+        "reported_date": None,
+        "reported_year": None,
+        "status": state,
+        "status_is_true_positive": is_true_positive,
+        "finder": "sqlancer",
+        "technique": None,
+        "symptom": "unknown",
+        "reporter": None,
+        "links": links,
+        "attribution": {
+            "rule": "curated_primary_source",
+            "confidence": "high",
+            "evidence": [{
+                "source_url": source_url,
+                "source_type": "github_repository",
+                "excerpt": truncate(excerpt, 500),
+                "excerpt_is_verbatim": True,
+                "note": (f"{repository} records this bug in its {dbms_id} "
+                         f"provider, as the field {entry['constant']} that "
+                         f"works around it."
+                         + ("" if readable else
+                            f" The report itself is in the project's own "
+                            f"tracker at {url}, which is not public.")),
+                "content_sha256": content_hash(excerpt),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }],
+        },
+        "provenance": {
+            "collector": COLLECTOR,
+            "collector_version": COLLECTOR_VERSION,
+            "policy_version": config.policy_version(),
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+            "source_url": source_url,
+            "source_type": "github_repository",
+            "content_sha256": content_hash(excerpt),
+        },
+    }
+    if readable:
+        record["primary_url"] = normalize_url(url)
+    return record
+
+
+NUMBER_IN_NAME = re.compile(r"(\d{3,})")
+
+
+def bug_url(entry: dict) -> str:
+    """The URL this field is about, which is not always the first listed.
+
+    A comment can cite more than one issue -- CockroachDB's file has an entry
+    naming a closed bug and the underlying one it depends on -- and the field
+    name says which is the subject. Taking the first URL recorded the wrong
+    issue and titled it with a note to maintainers.
+    """
+    urls = entry["urls"]
+    wanted = NUMBER_IN_NAME.findall(entry.get("constant") or "")
+    for number in wanted:
+        for url in urls:
+            if re.search(rf"(?<!\d){re.escape(number)}(?!\d)", url):
+                return url
+    return urls[0]
+
+
+def _title_for(entry: dict) -> str:
+    """The description if the comment carries one, else the link itself.
+
+    Upstream writes a bare URL above each field; the Oxla provider writes a
+    sentence saying what goes wrong. Where there is a sentence it is the better
+    title, and where there is not, inventing one would be writing evidence.
+    """
+    text = URL.sub("", entry["note"]).strip(" .:-")
+    text = re.sub(r"^(?:See|see)\s*:?\s*", "", text).strip()
+    return text or entry["urls"][0]
+
+
+def collect(gh: GitHub, checkouts: Optional[Sequence[Tuple[str, str, str]]] = None,
+            *, timestamp: Optional[str] = None, fetcher=None
+            ) -> Tuple[List[dict], List[dict], List[dict]]:
+    """Return ``(records, rejected, needs_review)`` from every bug file."""
+    timestamp = timestamp or now()
+    index = provider_index()
+    checkouts = list(checkouts) if checkouts is not None else [UPSTREAM]
+
+    records: List[dict] = []
+    rejected: List[dict] = []
+    needs_review: List[dict] = []
+    seen: set = set()
+
+    for owner, repo, ref in checkouts:
+        repository = f"{owner}/{repo}"
+        for directory, path in bugs_files(gh, owner, repo, ref):
+            dbms_id = index.get(directory.lower())
+            source_url = f"https://github.com/{repository}/blob/{ref}/{path}"
+            if dbms_id is None:
+                needs_review.append({
+                    "kind": "bug",
+                    "url": source_url,
+                    "reason": (f"{repository} records bugs for a provider "
+                               f"{directory!r} that is not a registered "
+                               f"database system"),
+                })
+                continue
+            result = gh.raw_file(owner, repo, path, ref)
+            raw = result[0] if isinstance(result, tuple) else result
+            for entry in parse_entries(raw or ""):
+                url = bug_url(entry)
+                key = normalize_url(url)
+                if key in seen:
+                    continue
+                seen.add(key)
+                records.append(build_record(
+                    entry, dbms_id, url=url, source_url=source_url,
+                    repository=repository, timestamp=timestamp,
+                    status=resolve_status(gh, url, fetcher),
+                    readable=is_publicly_readable(url, fetcher),
+                    # The report's own title beats the comment above the
+                    # field, which is sometimes a note to maintainers rather
+                    # than a description of the bug.
+                    title=(github_issue(gh, url) or {}).get("title")))
+    return records, rejected, needs_review
diff --git a/tools/impact/collectors/resources.py b/tools/impact/collectors/resources.py
new file mode 100644
index 0000000..9ca0bcd
--- /dev/null
+++ b/tools/impact/collectors/resources.py
@@ -0,0 +1,693 @@
+"""Collects resources related to SQLancer.
+
+The failure mode this collector is designed against is turning into a link
+dump. A resource has to be substantially about SQLancer or one of its
+techniques, which in practice means it comes from one of two places: the
+project's own material (its repository, its documentation, the talks recorded
+alongside its papers), or a repository that exists specifically to integrate
+SQLancer with something else.
+
+The project's own README is the primary source for links to other people's
+work. What the project publishes about itself -- its documentation, its blog,
+its own papers -- is deliberately excluded: the site already links all of it,
+and repeating it here would pad the list without saying anything about reach.
+Each entry keeps the verbatim line it came from.
+"""
+
+from __future__ import annotations
+
+import re
+from html import unescape
+from typing import Dict, List, Optional, Sequence, Tuple
+
+from .. import config, taxonomy
+from ..github import GitHub
+from ..util import (content_hash, normalize_url, now, stable_digest, truncate,
+                    verbatim_excerpt)
+
+COLLECTOR = "resources"
+COLLECTOR_VERSION = "1.1.0"
+
+# Several publisher and video sites answer the default client with a 403.
+BROWSER_USER_AGENT = (
+    "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) "
+    "Chrome/125.0 Safari/537.36")
+
+OWNER, REPO, REF = "sqlancer", "sqlancer", "main"
+README_URL = f"https://github.com/{OWNER}/{REPO}/blob/{REF}/README.md"
+
+MARKDOWN_LINK = re.compile(r"\[([^\]\n]{2,120})\]\((https?://[^)\s]+)\)")
+
+# How a URL maps onto the resource vocabulary. Order matters: first match wins.
+#
+# There is no catch-all. Documentation, tutorials and paper references were
+# removed from the vocabulary because in practice they were the project's own
+# material -- its README, its blog, its own publications -- which the site
+# already links and which says nothing about anyone else picking SQLancer up.
+# A link that fits none of the remaining kinds is now rejected rather than
+# filed under a generic label.
+TYPE_BY_HOST: Tuple[Tuple[re.Pattern, str], ...] = (
+    (re.compile(r"youtube\.com|youtu\.be|vimeo\.com"), "talk"),
+    (re.compile(r"zenodo\.org|figshare\.com"), "dataset"),
+    (re.compile(r"github\.com/"), "tool"),
+    (re.compile(r"blog|medium\.com|substack\.com|dev\.to"), "blog_post"),
+)
+
+# Link text that signals what a resource is, overriding the host heuristic.
+TYPE_BY_LABEL: Tuple[Tuple[re.Pattern, str], ...] = (
+    (re.compile(r"\bvideo\b|\btalk\b|\bpresentation\b|\bkeynote\b", re.I), "talk"),
+    (re.compile(r"\bartifact\b", re.I), "artifact"),
+    (re.compile(r"\bblog\b|\bpost\b", re.I), "blog_post"),
+    (re.compile(r"\bdataset\b", re.I), "dataset"),
+)
+
+# The project's own properties. Material published here is SQLancer talking
+# about itself, which belongs on the rest of the site rather than in a list of
+# what other people have made.
+OWN_PROPERTY = re.compile(
+    r"^https://(github\.com/sqlancer(/|$)|sqlancer\.github\.io"
+    r"|search\.maven\.org/artifact/com\.sqlancer"
+    r"|hub\.docker\.com/r/mrigger/sqlancer)", re.IGNORECASE)
+
+# The project names its own things after itself -- "SQLancer Talks", "SQLancer
+# Tutorial Playlist" -- while other people's projects carry their own names.
+# That is enough to tell the two apart on hosts the pattern above cannot cover,
+# such as the project's YouTube playlists.
+OWN_NAME = re.compile(r"^SQLancer\b", re.IGNORECASE)
+
+# Links that are infrastructure rather than resources about SQLancer.
+IGNORE_URL = re.compile(
+    r"(shields\.io|travis-ci|codecov|badge|/actions/workflows/|"
+    r"opensource\.org/licenses|creativecommons\.org|"
+    r"github\.com/sqlancer/sqlancer/(issues|pull|blob|commit|releases|tree)/|"
+    r"github\.com/[^/]+/[^/]+/(issues|pull|commit)/)",
+    re.IGNORECASE)
+
+# README sections that collect resources rather than instructions. A link
+# outside them has to earn its place some other way -- this is what keeps the
+# build-tool prerequisite and the DBMS manuals quoted in the FAQ out of the
+# list, without needing a denylist of hosts.
+RESOURCE_SECTION = re.compile(
+    r"\b(links?|papers?|approach\w*|resources?|talks?|related)\b",
+    re.IGNORECASE)
+
+# A recording linked from the project's own README is about the project, wherever
+# in the README it sits.
+MEDIA_HOST = re.compile(r"youtube\.com|youtu\.be|vimeo\.com", re.IGNORECASE)
+
+HEADING = re.compile(r"^(#{1,6})\s+(.*)$", re.MULTILINE)
+
+# Link texts that carry no information; the surrounding line names the resource.
+UNINFORMATIVE_LABEL = re.compile(
+    r"^(here|this|link|these|it|paper|video|docs?|more|read more|"
+    r"click here|see here|blog|website|github|gitlab)[.:]?$", re.IGNORECASE)
+
+
+# Sources that no automated route here reaches: vendor engineering blogs,
+# package registries, conference recordings. Each is fetched and quoted like
+# any other source -- being on this list decides that a URL is worth looking at,
+# never what it says.
+CURATED: Tuple[Dict[str, object], ...] = (
+    {
+        "url": ("https://techcommunity.microsoft.com/blog/adforpostgresql/"
+                "mining-for-logic-bugs-in-the-citus-extension-to-postgres-"
+                "with-sqlancer/1634393"),
+        "type": "blog_post",
+        "title": ("Mining for logic bugs in the Citus extension to Postgres "
+                  "with SQLancer"),
+        "year": 2020,
+        "publisher": "Microsoft Tech Community",
+        "related_dbms": ["citus", "postgresql"],
+        "related_techniques": ["tlp"],
+    },
+    {
+        "url": "https://www.yugabyte.com/blog/yugabytedb-database-testing/",
+        "type": "blog_post", "title": "How we test YugabyteDB",
+        "publisher": "Yugabyte", "related_dbms": ["yugabytedb"],
+        "related_techniques": [],
+    },
+    {
+        "url": "https://materialize.com/blog/qa-process-overview/",
+        "type": "blog_post", "title": "An overview of the Materialize QA process",
+        "publisher": "Materialize", "related_dbms": ["materialize"],
+        "related_techniques": [],
+    },
+    {
+        "url": "https://www.monetdb.org/blogs/faster-robuster-features/",
+        "type": "blog_post",
+        "title": "Faster, more robust, and with more features",
+        "publisher": "MonetDB", "related_dbms": ["monetdb"],
+        "related_techniques": [],
+    },
+    {
+        # Later than the other two Citus posts, and the reason it is here: it
+        # says the use carried on after the fork stopped being updated.
+        "url": ("https://www.citusdata.com/blog/2021/04/10/"
+                "talk-at-cmu-how-citus-distributes-postgresql-via-extension-"
+                "apis/"),
+        "type": "blog_post",
+        "title": ("Talk at CMU: How Citus distributes PostgreSQL via extension "
+                  "APIs"),
+        "year": 2021,
+        "publisher": "Citus Data",
+        "related_dbms": ["citus", "postgresql"],
+        "related_techniques": [],
+    },
+    {
+        "url": ("https://www.citusdata.com/blog/2020/09/04/"
+                "mining-for-logic-bugs-in-citus-with-sqlancer/"),
+        "type": "blog_post",
+        "title": "Mining for logic bugs in Citus with SQLancer",
+        "year": 2020, "publisher": "Citus Data",
+        "related_dbms": ["citus", "postgresql"], "related_techniques": ["tlp"],
+    },
+    {
+        # The project's own testing page, which names the author rather than
+        # the tool -- and is the reason SQLite appears in half the talks here.
+        "url": "https://www.sqlite.org/testing.html",
+        "type": "documentation_note",
+        "title": "How SQLite Is Tested",
+        "publisher": "SQLite",
+        "related_dbms": ["sqlite"],
+        "related_techniques": [],
+    },
+    {
+        "url": "https://duckdb.org/why_duckdb",
+        "type": "documentation_note", "title": "Why DuckDB",
+        "publisher": "DuckDB", "related_dbms": ["duckdb"],
+        "related_techniques": [],
+    },
+    {
+        "url": "https://datafusion.apache.org/contributor-guide/testing.html",
+        "type": "documentation_note",
+        "title": "Apache DataFusion contributor guide: testing",
+        "publisher": "Apache DataFusion", "related_dbms": ["datafusion"],
+        "related_techniques": [],
+    },
+    # datafusion-sqlancer is deliberately absent. DataFusion's use of SQLancer
+    # is already an adoption record backed by the project's own testing guide,
+    # and it has its own page carrying 54 bugs; listing the repository here as
+    # well says nothing the adoption record does not, and puts one project in
+    # two places under two headings.
+    {
+        "url": "https://github.com/PingCAP-QE/go-sqlancer",
+        # Distinguished from chaos-mesh/go-sqlancer, which the SQLancer README
+        # links and which carries the same name.
+        "type": "tool", "title": "go-sqlancer (PingCAP-QE)",
+        "publisher": "PingCAP", "related_dbms": ["tidb"],
+        "related_techniques": ["pqs", "norec", "tlp"],
+    },
+    {
+        "url": "https://questdb.com/blog/fuzz-testing-questdb/",
+        "type": "blog_post",
+        "title": "Fuzz testing QuestDB",
+        "year": 2022,
+        "publisher": "QuestDB",
+        "related_dbms": ["questdb"],
+        "related_techniques": [],
+    },
+    {
+        # The repository is the primary source, as it is for every other tool
+        # here; the Open Exchange listing only republishes it.
+        "url": "https://github.com/caretdev/sqlancer-iris",
+        "type": "tool",
+        "title": "sqlancer-iris",
+        "publisher": "CaretDev",
+        "related_dbms": [],
+        "related_techniques": [],
+    },
+    {
+        # The write-up behind that integration, by its author: what it took to
+        # bring the NoREC oracle to IRIS, and what the attempt turned up.
+        "url": ("https://community.intersystems.com/post/when-sqlancer-meets-"
+                "iris-what-happens-when-we-push-database-its-limits"),
+        "type": "blog_post",
+        "title": ("When SQLancer Meets IRIS: What Happens When We Push a "
+                  "Database to Its Limits"),
+        "year": 2025,
+        "publisher": "InterSystems Developer Community",
+        "related_dbms": [],
+        "related_techniques": ["norec"],
+    },
+)
+
+# Talks are not listed here. They live in _data/impact/talks.json, which
+# records where in a talk SQLancer comes up and how far its words can be
+# trusted; ``from_talks`` turns each of those records into a resource so the
+# two never drift apart.
+
+TAG_RE = re.compile(r"<[^>]+>")
+SCRIPT_RE = re.compile(r"<(script|style)[^>]*>.*?</\1>", re.S | re.I)
+# The name is not the only way a page is about SQLancer. SQLite's testing page
+# is the strongest endorsement the project has anywhere and never uses the word:
+# it names the author, and calls them "Dr. Rigger's fuzzers". What a curated
+# source is about is settled by the list of them; this pattern only has to find
+# the sentence worth quoting once that decision is made.
+SENTENCE_WITH_SQLANCER = re.compile(
+    r"[^.!?]{0,240}\b(?:SQLancer|Manuel Rigger|Rigger's)\b[^.!?]{0,240}[.!?]")
+
+# Share of lower-case words above which a match reads as a sentence rather than
+# as a row of navigation labels.
+PROSE_RATIO = 0.55
+
+
+GITHUB_REPOSITORY = re.compile(
+    r"^https://github\.com/([\w.-]+)/([\w.-]+)/?$", re.IGNORECASE)
+README_NAMES = ("README.md", "README.rst", "README.txt", "readme.md")
+
+# Markdown furniture that would otherwise end up inside a quoted sentence.
+MARKDOWN_BADGE = re.compile(r"!\[[^\]]*\]\([^)]*\)")
+MARKDOWN_NOISE = re.compile(r"[#*_`>|]+|\[([^\]]*)\]\([^)]*\)")
+EMPTY_BRACKETS = re.compile(r"\[\s*\]\s*\(\s*\)|\[\s*\]|\(\s*\)")
+
+# A bare URL in a README ends sentences everywhere its path has a dot, so a
+# quote taken from the line around it starts mid-address.
+BARE_URL = re.compile(r"<?https?://\S+>?")
+
+
+def readme_text(fetcher, url: str) -> Optional[str]:
+    """A repository's README as plain text, or ``None``.
+
+    A repository page is mostly chrome -- the file list, the language bar, the
+    sidebar -- and quoting it produces a sentence that starts in the middle of
+    a navigation menu. The README is what the project says about itself.
+    """
+    match = GITHUB_REPOSITORY.match(url)
+    if not match:
+        return None
+    owner, repo = match.group(1), match.group(2)
+    for name in README_NAMES:
+        raw = f"https://raw.githubusercontent.com/{owner}/{repo}/HEAD/{name}"
+        try:
+            entry, _ = fetcher.fetch(raw, max_age_days=90)
+        except Exception:
+            continue
+        body = entry.get("body") or ""
+        if entry.get("status") == 200 and body:
+            # Badges first: they are images inside links, and taking the link
+            # apart from the outside leaves their brackets behind.
+            text = MARKDOWN_BADGE.sub(" ", body)
+            text = MARKDOWN_NOISE.sub(lambda m: m.group(1) or " ", text)
+            text = BARE_URL.sub(" ", text)
+            text = EMPTY_BRACKETS.sub(" ", text)
+            return re.sub(r"\s+", " ", text).strip()
+    return None
+
+
+def best_sentence(text: str) -> Optional[str]:
+    """The most prose-like sentence mentioning SQLancer, or ``None``.
+
+    Taking the first match quotes page furniture: a navigation bar carries no
+    sentence-ending punctuation, so a title further down the page gets the menu
+    above it swallowed into the match. Running text is mostly lower-case words,
+    while furniture is a run of capitalised labels, which separates the two
+    well enough to pick a sentence a reader would recognise as the page
+    speaking.
+    """
+    best, best_score = None, -1.0
+    for match in SENTENCE_WITH_SQLANCER.finditer(text):
+        candidate = match.group(0).strip()
+        words = [w for w in re.findall(r"[A-Za-z][\w'-]*", candidate) if len(w) > 1]
+        if len(words) < 6:
+            continue
+        score = sum(1 for w in words if w[0].islower()) / len(words)
+        # The first sentence that reads as prose is the one a person would
+        # quote: pages introduce what they are about before elaborating on it.
+        if score >= PROSE_RATIO:
+            return candidate
+        if score > best_score:
+            best, best_score = candidate, score
+    return best
+
+
+def page_text(html: str) -> str:
+    """Readable text of a fetched page, for quoting."""
+    text = SCRIPT_RE.sub(" ", html)
+    text = TAG_RE.sub(" ", text)
+    text = unescape(text).replace("\xa0", " ")
+    text = re.sub(r"\s+", " ", text)
+    # Stripping a link out of running text leaves a space before the full stop
+    # it was followed by. That space is an artefact of reading the page, not
+    # something the page wrote, and it would end up inside a quotation.
+    return re.sub(r" +([.,;:!?])", r"\1", text).strip()
+
+
+def from_talks(*, timestamp: Optional[str] = None) -> List[dict]:
+    """Resource records for the talks file, quoting its strongest mention.
+
+    A talk's evidence is ranked by how directly it reaches the speaker: their
+    own slides first, then captions -- which get said out loud in the note to
+    be a machine's transcription -- then a viewer's account, which carries no
+    quote at all.
+    """
+    from .. import talks as talks_store
+
+    timestamp = timestamp or now()
+    rank = {"slides": 0, "captions": 1, "watched": 2}
+    records: List[dict] = []
+    for talk in talks_store.load().get("talks", []):
+        mentions = sorted(talk.get("mentions") or [],
+                          key=lambda m: rank.get(m.get("source"), 3))
+        if not mentions:
+            continue
+        best = mentions[0]
+        source = best.get("source")
+        if source == "slides":
+            note = "Sentence from the talk's own slides."
+        elif source == "captions":
+            heard = best.get("heard_as")
+            note = ("From the talk's automatic captions, a machine "
+                    "transcription of speech"
+                    + (f" -- it renders the name as \"{heard}\"" if heard else "")
+                    + f". The link points at {best.get('timestamp')}, where a "
+                      "reader can listen to what was actually said.")
+        else:
+            note = best.get("note") or "Reported by a maintainer who watched it."
+
+        summary = (talk.get("relationship") or {}).get("summary") or talk["title"]
+        records.append({
+            "id": f"resource:talk:{stable_digest(talk['url'])}",
+            "title": truncate(talk["title"], 300),
+            "type": "talk",
+            "url": talk["url"],
+            "description": truncate(summary, 590),
+            "official": False,
+            "evidence": [{
+                "source_url": best.get("url") or talk["url"],
+                "source_type": "video",
+                "excerpt": best.get("excerpt"),
+                "excerpt_is_verbatim": bool(best.get("excerpt")),
+                "note": truncate(note, 500),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }],
+            "provenance": {
+                "collector": COLLECTOR,
+                "collector_version": COLLECTOR_VERSION,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+                "source_url": talk["url"],
+                "source_type": "video",
+            },
+            **({"year": talk["year"]} if talk.get("year") else {}),
+            **({"publisher": talk["publisher"]} if talk.get("publisher") else {}),
+            **({"authors": talk["speakers"]} if talk.get("speakers") else {}),
+            **({"related_techniques": talk["related_techniques"]}
+               if talk.get("related_techniques") else {}),
+            **({"related_dbms": talk["related_dbms"]}
+               if talk.get("related_dbms") else {}),
+        })
+    return records
+
+
+def from_curated(fetcher, *, timestamp: Optional[str] = None,
+                 entries: Optional[Sequence[Dict[str, object]]] = None
+                 ) -> Tuple[List[dict], List[dict]]:
+    """Fetch each curated source and quote what it actually says.
+
+    A source that cannot be fetched, or that turns out not to mention SQLancer,
+    is reported rather than published: the list says where to look, not what is
+    there.
+    """
+    timestamp = timestamp or now()
+    tax = taxonomy.load()
+    records: List[dict] = []
+    needs_review: List[dict] = []
+
+    for item in (entries if entries is not None else CURATED):
+        url = normalize_url(str(item["url"]))
+        if not url:
+            continue
+        try:
+            entry, _ = fetcher.fetch(str(item["url"]), max_age_days=90)
+        except Exception:
+            entry = {}
+        body = entry.get("body") or ""
+        if entry.get("status") != 200 or not body:
+            needs_review.append({
+                "kind": "resource", "url": url,
+                "reason": "curated source could not be fetched",
+            })
+            continue
+
+        text = readme_text(fetcher, url) or page_text(body)
+        sentence = best_sentence(text)
+        if sentence:
+            excerpt = truncate(sentence, 900)
+            note = "Sentence on the page describing SQLancer."
+            if not verbatim_excerpt(text, excerpt):
+                continue
+        else:
+            # A source with nothing quotable is reported, never published with
+            # a note standing in for a quote. The one kind of source where that
+            # is unavoidable -- a recording, whose words are not in its page --
+            # is handled in talks.json, which says so explicitly.
+            needs_review.append({
+                "kind": "resource", "url": url,
+                "reason": "curated source does not mention SQLancer in its text",
+            })
+            continue
+
+        kind = str(item["type"])
+        record = {
+            "id": f"resource:{kind}:{stable_digest(url)}",
+            "title": truncate(str(item["title"]), 300),
+            "type": kind,
+            "url": url,
+            "description": truncate(excerpt, 590),
+            "official": False,
+            "evidence": [{
+                "source_url": url,
+                "source_type": ("blog_post" if kind == "blog_post"
+                                else "video" if kind == "talk" else "website"),
+                "excerpt": excerpt,
+                "note": note,
+                "content_sha256": content_hash(text),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }],
+            "provenance": {
+                "collector": COLLECTOR,
+                "collector_version": COLLECTOR_VERSION,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+                "source_url": url,
+                "source_type": "website",
+            },
+        }
+        if item.get("year"):
+            record["year"] = item["year"]
+        if item.get("publisher"):
+            record["publisher"] = str(item["publisher"])
+        techniques = list(item.get("related_techniques") or [])
+        techniques += [m.technique_id for m in tax.find_techniques(text)]
+        techniques = [t for t in dict.fromkeys(techniques) if t in tax.techniques]
+        if techniques:
+            record["related_techniques"] = sorted(techniques)
+        if item.get("related_dbms"):
+            record["related_dbms"] = sorted(item["related_dbms"])
+        if excerpt is not None:
+            record["evidence"][0]["excerpt_is_verbatim"] = True
+        records.append(record)
+    return records, needs_review
+
+
+def classify_type(label: str, url: str) -> Optional[str]:
+    """The resource kind, or None when the link is not one we keep."""
+    for pattern, kind in TYPE_BY_LABEL:
+        if pattern.search(label):
+            return kind
+    for pattern, kind in TYPE_BY_HOST:
+        if pattern.search(url):
+            return kind
+    return None
+
+
+def _line_of(text: str, index: int) -> str:
+    start = text.rfind("\n", 0, index)
+    start = 0 if start < 0 else start + 1
+    end = text.find("\n", index)
+    end = len(text) if end < 0 else end
+    return truncate(text[start:end].strip(), 900)
+
+
+def _sections(text: str) -> List[Tuple[int, str]]:
+    """Offsets of each heading, so a link can be attributed to its section.
+
+    Inline links are stripped from the heading text first: a heading like
+    ``# Getting Started [[Video Guide]](...)`` must not read as a section about
+    videos and pull the whole installation section in with it.
+    """
+    return [(match.start(),
+             MARKDOWN_LINK.sub("", match.group(2)).replace("[", " ")
+             .replace("]", " ").strip())
+            for match in HEADING.finditer(text)]
+
+
+def _section_at(sections: List[Tuple[int, str]], index: int) -> str:
+    current = ""
+    for offset, title in sections:
+        if offset > index:
+            break
+        current = title
+    return current
+
+
+def _title_for(label: str, line: str, url: str) -> str:
+    """Use the link text unless it says nothing, then fall back sensibly.
+
+    The line only serves as a title when it holds exactly one link; otherwise it
+    describes several resources at once and would title all of them the same.
+    """
+    label = label.strip()
+    if not UNINFORMATIVE_LABEL.match(label):
+        return truncate(label, 300)
+    if len(MARKDOWN_LINK.findall(line)) == 1:
+        stripped = MARKDOWN_LINK.sub(r"\1", line).lstrip("*-# ").strip()
+        stripped = re.sub(r"\s+", " ", stripped).strip("*").rstrip(".")
+        if stripped and len(stripped) > len(label):
+            return truncate(stripped, 300)
+    if "sqlancer" in url.lower():
+        return truncate(f"SQLancer {label}", 300)
+    return truncate(label, 300)
+
+
+def _describe(label: str, line: str, kind: str) -> str:
+    """A short factual description built from the source line, not invented.
+
+    Falls back to a plain statement of what the resource is rather than
+    guessing at content the source does not describe.
+    """
+    stripped = MARKDOWN_LINK.sub(r"\1", line).lstrip("*-# ").strip()
+    stripped = re.sub(r"\s+", " ", stripped)
+    if len(stripped) > len(label) + 15:
+        return truncate(stripped, 590)
+    readable = kind.replace("_", " ")
+    return f"{label}. Linked from the SQLancer README as a {readable}."
+
+
+def from_readme(gh: GitHub, *, timestamp: Optional[str] = None
+                ) -> Tuple[List[dict], List[dict]]:
+    """Parse the SQLancer README's links into resource records."""
+    timestamp = timestamp or now()
+    tax = taxonomy.load()
+    text, _ = gh.raw_file(OWNER, REPO, "README.md", REF, max_age_days=7)
+    if not text:
+        return [], []
+
+    readme_hash = content_hash(text)
+    sections = _sections(text)
+    records: List[dict] = []
+    rejected: List[dict] = []
+    seen: Dict[str, str] = {}
+
+    for match in MARKDOWN_LINK.finditer(text):
+        label, url = match.group(1).strip(), match.group(2).strip()
+        canonical = normalize_url(url)
+        if not canonical or IGNORE_URL.search(url):
+            continue
+        if canonical in seen:
+            continue
+        line = _line_of(text, match.start())
+        if not verbatim_excerpt(text, line):
+            continue
+
+        # Substantially about SQLancer, or nothing. Sitting in a section that
+        # lists resources is not enough on its own: the README's links section
+        # also names peer tools -- Jepsen, SQLsmith, Squirrel -- which are other
+        # people's testing tools that have nothing to do with SQLancer, and
+        # listing them here would pad the count with work that is not evidence
+        # of reach. What qualifies is the link pointing at something SQLancer,
+        # or the README's own line about it naming SQLancer or one of its
+        # techniques, or it being a recording, which the README only links when
+        # the talk is about the project.
+        section = _section_at(sections, match.start())
+        about_sqlancer = (
+            "sqlancer" in canonical.lower()
+            or bool(tax.mentions_finder(line) or tax.find_techniques(line))
+            or bool(MEDIA_HOST.search(canonical)))
+        if not (about_sqlancer and RESOURCE_SECTION.search(section)):
+            rejected.append({
+                "kind": "resource", "url": canonical, "title": label,
+                "reason": (f"linked from the README section {section!r} but the "
+                           f"line about it does not connect it to SQLancer"
+                           if RESOURCE_SECTION.search(section) else
+                           f"linked from the README section {section!r}, which "
+                           f"does not list resources"),
+            })
+            continue
+
+        if OWN_PROPERTY.match(canonical) or OWN_NAME.match(label.strip()):
+            rejected.append({
+                "kind": "resource", "url": canonical, "title": label,
+                "reason": "the project's own material, not external adoption",
+            })
+            continue
+
+        kind = classify_type(label, url)
+        if kind is None:
+            rejected.append({
+                "kind": "resource", "url": canonical, "title": label,
+                "reason": "does not fit any kept resource kind",
+            })
+            continue
+
+        techniques = [m.technique_id for m in tax.find_techniques(line)]
+        record = {
+            "id": f"resource:{kind}:{stable_digest(canonical)}",
+            "title": _title_for(label, line, canonical),
+            "type": kind,
+            "url": canonical,
+            "description": _describe(label, line, kind),
+            "official": "github.com/sqlancer/" in canonical,
+            "evidence": [{
+                "source_url": README_URL,
+                "source_type": "documentation",
+                "excerpt": line,
+                "excerpt_is_verbatim": True,
+                "note": "Line in the SQLancer README that links this resource.",
+                "content_sha256": readme_hash,
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }],
+            "provenance": {
+                "collector": COLLECTOR,
+                "collector_version": COLLECTOR_VERSION,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+                "source_url": README_URL,
+                "source_type": "documentation",
+                "content_sha256": readme_hash,
+            },
+        }
+        if techniques:
+            record["related_techniques"] = sorted(set(techniques))
+        records.append(record)
+        seen[canonical] = record["id"]
+
+    return records, rejected
+
+
+def collect(gh: GitHub, *, timestamp: Optional[str] = None,
+            fetcher=None) -> Tuple[List[dict], List[dict], List[dict]]:
+    """Return ``(records, rejected, needs_review)`` for all resource sources."""
+    timestamp = timestamp or now()
+    records, rejected = from_readme(gh, timestamp=timestamp)
+    by_url = {record["url"] for record in records}
+
+    needs_review: List[dict] = []
+    if fetcher is None:
+        from ..cache import Caches
+        from ..http import Fetcher
+        fetcher = Fetcher(Caches().http, min_interval=0.5, max_retries=2,
+                          timeout=20.0,
+                          headers={"User-Agent": BROWSER_USER_AGENT})
+    curated, needs_review = from_curated(fetcher, timestamp=timestamp)
+    for record in list(curated) + from_talks(timestamp=timestamp):
+        if record["url"] not in by_url:
+            records.append(record)
+            by_url.add(record["url"])
+    return records, rejected, needs_review
diff --git a/tools/impact/collectors/sqlancer_bugs.py b/tools/impact/collectors/sqlancer_bugs.py
new file mode 100644
index 0000000..51eaf85
--- /dev/null
+++ b/tools/impact/collectors/sqlancer_bugs.py
@@ -0,0 +1,288 @@
+"""Imports SQLancer's own curated bug repository, github.com/sqlancer/bugs.
+
+That repository is the project's primary record of what SQLancer found, so its
+entries are admitted under the ``curated_primary_source`` clause of the
+attribution policy. Nothing is inferred: the status, the oracle and the links
+come from the file, and the technique is resolved through the taxonomy rather
+than by pattern-matching in this module.
+
+Every imported record carries two or three pieces of evidence -- the upstream
+bug report itself, the repository's own statement of what it collects, and,
+where the entry names a SQLancer oracle, the verbatim entry text showing it.
+"""
+
+from __future__ import annotations
+
+import json
+from typing import Dict, List, Optional, Tuple
+
+from .. import config, taxonomy
+from ..github import GitHub
+from ..repos import canonical_issue_url
+from ..util import (content_hash, normalize_url, now, parse_date, slugify,
+                    stable_digest, truncate, verbatim_excerpt, year_of)
+
+COLLECTOR = "sqlancer_bugs"
+COLLECTOR_VERSION = "1.0.0"
+
+OWNER, REPO, REF = "sqlancer", "bugs", "master"
+BUGS_JSON_URL = f"https://github.com/{OWNER}/{REPO}/blob/{REF}/bugs.json"
+README_URL = f"https://github.com/{OWNER}/{REPO}/blob/{REF}/README.md"
+
+# The statement that makes this repository usable as attribution evidence.
+README_CLAIM = "the repository stores a list of bugs found by SQLancer"
+
+STATUS_MAP = {
+    "fixed": ("fixed", True),
+    "verified": ("verified", True),
+    "open": ("open", True),
+    "fixed (in documentation)": ("fixed_in_documentation", True),
+    "closed (not a bug)": ("closed_not_a_bug", False),
+    "closed (duplicate)": ("closed_duplicate", False),
+}
+
+# The upstream 'oracle' field mixes techniques with symptoms. Values that are
+# not a technique describe how the bug manifested.
+SYMPTOM_ONLY = {"error": "error", "crash": "crash", "hang": "hang"}
+
+# Preference order when a record links several sources; the first present wins.
+PRIMARY_LINK_ORDER = ("bugreport", "bugreports", "bugtracker", "email",
+                      "email 2", "commit", "fix", "fix1", "fixed",
+                      "documentation", "docs fix")
+
+
+def _source_type(url: str) -> str:
+    lowered = url.lower()
+    if "github.com" in lowered and "/pull/" in lowered:
+        return "github_pull_request"
+    if "github.com" in lowered and "/issues/" in lowered:
+        return "github_issue"
+    if "github.com" in lowered or "gitlab" in lowered:
+        return "commit"
+    if "mailinglist" in lowered or "mailman" in lowered or "pipermail" in lowered:
+        return "mailing_list"
+    if "forum" in lowered:
+        return "forum_post"
+    if "sqlite.org" in lowered or "bugs." in lowered or "jira" in lowered:
+        return "issue_tracker"
+    return "other"
+
+
+def record_spans(text: str) -> List[Tuple[int, int]]:
+    """Byte spans of each top-level array element in ``text``.
+
+    Used so that an evidence excerpt can be lifted verbatim out of the upstream
+    file rather than re-serialised from the parsed structure.
+    """
+    decoder = json.JSONDecoder()
+    spans: List[Tuple[int, int]] = []
+    index = text.index("[") + 1
+    length = len(text)
+    while index < length:
+        while index < length and text[index] in " \t\r\n,":
+            index += 1
+        if index >= length or text[index] == "]":
+            break
+        _, end = decoder.raw_decode(text, index)
+        spans.append((index, end))
+        index = end
+    return spans
+
+
+def _entry_excerpt(text: str, span: Tuple[int, int]) -> Optional[str]:
+    """Verbatim slice of an entry from ``"date"`` through its ``"oracle"`` line.
+
+    That window identifies the record and shows the oracle without dragging in
+    the test case, which can run to hundreds of statements.
+    """
+    start, end = span
+    chunk = text[start:end]
+    date_at = chunk.find('"date"')
+    oracle_at = chunk.find('"oracle"')
+    if date_at < 0 or oracle_at < 0:
+        return None
+    line_end = chunk.find("\n", oracle_at)
+    if line_end < 0:
+        line_end = len(chunk)
+    excerpt = chunk[date_at:line_end].rstrip().rstrip(",")
+    return truncate(excerpt, 3900)
+
+
+def _dbms_id(raw: str, alias_map: Dict[str, str]) -> str:
+    return alias_map.get(raw.strip().lower(), slugify(raw))
+
+
+def build_alias_map() -> Dict[str, str]:
+    """Map every registered DBMS alias to its canonical registry id."""
+    registry = config.load_json(config.DATA_FILES["dbms"])["dbms"]
+    alias_map: Dict[str, str] = {}
+    for entry in registry:
+        for alias in [entry["id"], entry["name"], *entry.get("aliases", [])]:
+            alias_map.setdefault(alias.strip().lower(), entry["id"])
+    return alias_map
+
+
+def _primary_url(links: Dict[str, str]) -> Optional[str]:
+    for key in PRIMARY_LINK_ORDER:
+        if links.get(key):
+            return links[key]
+    for value in links.values():
+        if value:
+            return value
+    return None
+
+
+def collect(gh: GitHub, *, timestamp: Optional[str] = None) -> List[dict]:
+    """Return bug records imported from the curated repository."""
+    timestamp = timestamp or now()
+    tax = taxonomy.load()
+    alias_map = build_alias_map()
+
+    raw_text, entry = gh.raw_file(OWNER, REPO, "bugs.json", REF, max_age_days=1)
+    if raw_text is None:
+        raise RuntimeError("could not fetch sqlancer/bugs bugs.json")
+    readme_text, _ = gh.raw_file(OWNER, REPO, "README.md", REF, max_age_days=30)
+
+    entries = json.loads(raw_text)
+    spans = record_spans(raw_text)
+    if len(spans) != len(entries):  # formatting changed upstream
+        spans = [None] * len(entries)
+
+    readme_hash = content_hash(readme_text) if readme_text else None
+
+    # Only quote the repository's self-description if it really says this.
+    readme_excerpt = README_CLAIM if (
+        readme_text and verbatim_excerpt(readme_text, README_CLAIM)) else None
+
+    records: List[dict] = []
+    for item, span in zip(entries, spans):
+        # Hash this entry rather than the whole upstream file: an unrelated bug
+        # being added upstream must not rewrite the hash of all 499 records.
+        entry_text = raw_text[span[0]:span[1]] if span else json.dumps(item)
+        entry_hash = content_hash(entry_text)
+        dbms_id = _dbms_id(item["dbms"], alias_map)
+        links = {k: v for k, v in (item.get("links") or {}).items() if v}
+        primary = _primary_url(links)
+        reported_date = parse_date(item.get("date"))
+        status, is_true_positive = STATUS_MAP.get(
+            item.get("status", ""), ("unknown", False))
+
+        oracle = item.get("oracle") or ""
+        technique = tax.technique_from_oracle_label(oracle)
+        symptom = SYMPTOM_ONLY.get(oracle.strip().lower(),
+                                   "logic" if technique else "unknown")
+
+        # The title is part of the identity because a single upstream ticket
+        # occasionally covers two separately curated reports; the URL alone
+        # would collapse them into one record.
+        identity = "|".join([
+            canonical_issue_url(primary) or dbms_id,
+            item.get("date") or "",
+            item.get("title") or "",
+        ])
+        record_id = f"bug:{dbms_id}:{stable_digest(identity)}"
+
+        evidence: List[dict] = []
+        if primary:
+            evidence.append({
+                "source_url": primary,
+                "source_type": _source_type(primary),
+                "excerpt": None,
+                "note": "Upstream bug report or fix for this issue.",
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            })
+        evidence.append({
+            "source_url": README_URL,
+            "source_type": "curated_bug_repository",
+            "excerpt": readme_excerpt,
+            "excerpt_is_verbatim": True,
+            "note": ("SQLancer's own curated bug repository, which is the "
+                     "primary record of what the tool found."),
+            "content_sha256": readme_hash,
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        })
+
+        # The entry itself is quoted verbatim, so every record can be audited
+        # against the upstream file without re-deriving it. When it names a
+        # SQLancer oracle, that quote is also what upgrades the attribution
+        # from "came from the curated list" to "the technique is on record".
+        rule = "curated_primary_source"
+        excerpt = _entry_excerpt(raw_text, span) if span else None
+        if excerpt and verbatim_excerpt(raw_text, excerpt):
+            note = "Entry for this report in the curated bugs.json."
+            if technique:
+                note = (f"Entry in bugs.json recording that the "
+                        f"{tax.display_name(technique)} oracle found this bug.")
+                rule = "technique_attribution"
+            evidence.append({
+                "source_url": BUGS_JSON_URL,
+                "source_type": "curated_bug_repository",
+                "excerpt": excerpt,
+                "excerpt_is_verbatim": True,
+                "note": note,
+                "content_sha256": entry_hash,
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            })
+
+        if not links:
+            # A handful of early reports went through private security channels
+            # and have no public URL. The curated entry is then the primary
+            # source, so the record stays auditable rather than being dropped.
+            links = {"curated_entry": BUGS_JSON_URL}
+
+        record = {
+            "id": record_id,
+            "dbms": dbms_id,
+            "title": item.get("title") or "(untitled report)",
+            "reported_date": reported_date,
+            "reported_year": year_of(reported_date),
+            "status": status,
+            "status_is_true_positive": is_true_positive,
+            "finder": "sqlancer",
+            "technique": technique,
+            "symptom": symptom,
+            "reporter": item.get("reporter"),
+            "severity": item.get("severity"),
+            "cve": item.get("cve"),
+            "links": links,
+            "attribution": {
+                "rule": rule,
+                "confidence": "high",
+                "evidence": evidence,
+            },
+            "provenance": {
+                "collector": COLLECTOR,
+                "collector_version": COLLECTOR_VERSION,
+                "policy_version": config.policy_version(),
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+                "source_url": BUGS_JSON_URL,
+                "source_type": "curated_bug_repository",
+                "content_sha256": entry_hash,
+            },
+        }
+        if primary:
+            record["primary_url"] = normalize_url(primary)
+        records.append(_drop_nulls(record))
+    return records
+
+
+def _drop_nulls(value):
+    """Strip ``None`` values the schemas do not allow, keeping nullable ones."""
+    if isinstance(value, dict):
+        out = {}
+        for key, item in value.items():
+            if item is None and key in ("content_sha256", "excerpt_is_verbatim",
+                                        "note", "severity", "cve"):
+                continue
+            out[key] = _drop_nulls(item)
+        return out
+    if isinstance(value, list):
+        return [_drop_nulls(item) for item in value]
+    return value
diff --git a/tools/impact/collectors/talks.py b/tools/impact/collectors/talks.py
new file mode 100644
index 0000000..555b288
--- /dev/null
+++ b/tools/impact/collectors/talks.py
@@ -0,0 +1,457 @@
+"""The catalogue of talks that discuss SQLancer, and how each one is read.
+
+Unlike every other collector here, this one does not search. Nothing indexes
+what is said out loud in a conference room, so a talk arrives because somebody
+saw it. What the collector does is turn that pointer into evidence: it fetches
+the speaker's slides where a deck exists, folds in a caption transcript where
+one was captured from the browser, and otherwise records plainly that the only
+source is a person who watched it.
+
+The routes are deliberately not merged into one notion of "the text of the
+talk" -- see ``tools/impact/talks.py`` for why an automatic caption is not a
+quotation.
+
+Frames are the fourth route and the one that reaches what the others cannot:
+some talks name SQLancer only on a slide, and a slide is never spoken. A frame
+carries no excerpt, ever. Every other excerpt in this dataset is a substring of
+something that was fetched, and a check can say so; text read off a picture is
+not, so the picture is shown and a note says what is in it.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+from typing import Dict, List, Optional, Sequence, Tuple
+
+from .. import taxonomy
+from ..talks import (TRANSCRIPT_CACHE, deep_link, findings_from_segments,
+                     findings_from_slides, frame_for, talk_id, timestamp_of,
+                     video_id)
+from ..util import content_hash, now
+
+COLLECTOR = "talks"
+COLLECTOR_VERSION = "1.0.0"
+
+TAG_RE = re.compile(r"<[^>]+>")
+SCRIPT_RE = re.compile(r"<(script|style)[^>]*>.*?</\1>", re.S | re.I)
+
+
+# Every entry names its speaker and where it was given, because that is what
+# makes a talk worth recording: who said it, and to whom.
+TALKS: Tuple[Dict[str, object], ...] = (
+    {
+        "url": "https://www.youtube.com/watch?v=QRwxHGpWaUA",
+        "title": "The Art of Database Testing by Alperen Keles | DC Systems 011",
+        "speakers": ["Alperen Keles"],
+        "publisher": "Antithesis",
+        "event": "DC Systems 011",
+        "year": 2026,
+        "duration_seconds": 2431,
+        "related_techniques": ["pqs", "tlp"],
+        "no_frame": [
+            {"at_seconds": 2087,
+             "why": ("The closing slide with the speaker's contact details is "
+                     "on screen; this mention is in the questions afterwards.")},
+        ],
+        "frames": [
+            {"at_seconds": 205, "matched": ["sqlancer"],
+             "note": ("The slide behind the speaker is SQLancer's logo above "
+                      "its eight oracles, named in full: PQS, NoREC, TLP, DQE, "
+                      "QPG, CERT, DQP and CODDTest."),
+             "technique_ids": ["pqs", "norec", "tlp", "qpg", "cert", "dqp",
+                               "coddtest"]},
+        ],
+        "roles": ["explains_technique", "background"],
+        "summary": ("A survey of database testing methodology that walks an "
+                    "audience through SQLancer's oracles, naming pivoted query "
+                    "synthesis outright and putting ternary logic partitioning "
+                    "on a slide; the tool itself comes up again in questions."),
+    },
+    {
+        "url": "https://www.youtube.com/watch?v=6YGqFRTe2D0",
+        "title": ("[FUZZING'23] \"Three Colours of Fuzzing: Reflections and "
+                  "Open Challenges\" Keynote by Cristian Cadar"),
+        "speakers": ["Cristian Cadar"],
+        "publisher": "International Fuzzing Workshop",
+        "event": "FUZZING'23",
+        "year": 2023,
+        "duration_seconds": 3638,
+        "frames": [
+            {"at_seconds": 295, "matched": ["sqlancer"],
+             "note": ("A slide of fuzzers that made a difference \u2014 KLEE, "
+                      "SAGE, AFL, OSS-Fuzz, Csmith and EMI \u2014 ending with "
+                      "SQLancer as "
+                      "the DBMS fuzzer, with the bug count it found in popular "
+                      "database systems.")},
+            {"at_seconds": 1509, "matched": ["author"],
+             "note": ("The slide displays SQLite's own paragraph about the "
+                      "project's author, attributed on the slide to the SQLite "
+                      "webpage. Its wording is in the dataset already, quoted "
+                      "from sqlite.org/testing.html.")},
+        ],
+        "roles": ["cites_as_example", "recognition"],
+        "summary": ("A fuzzing keynote that reaches for SQLancer as its "
+                    "example of a fuzzer that found hundreds of bugs in mature "
+                    "database systems, and later reads out SQLite's own line "
+                    "about the project's author, agreeing with it."),
+    },
+    {
+        "url": "https://www.youtube.com/watch?v=L90MBb6NLBE&t=1703s",
+        "title": ("FUZZING'25 Keynote: \"Constraining Fuzzing without Paying "
+                  "Too Much\" by Miryung Kim"),
+        "speakers": ["Miryung Kim"],
+        "publisher": "International Fuzzing Workshop",
+        "event": "FUZZING'25",
+        "year": 2025,
+        "duration_seconds": 2786,
+        "frames": [
+            {"at_seconds": 1703, "matched": ["sqlancer"],
+             "note": ("The keynote's table of what fuzzer customisations cost "
+                      "to build lists SQLancer among them, with its "
+                      "contributor, commit and paper counts.")},
+        ],
+        "roles": ["recognition"],
+        "summary": ("A fuzzing keynote that puts SQLancer in its table of what "
+                    "building a custom fuzzer costs, at 28:23."),
+        "watched": ("A maintainer watched the keynote and noted that SQLancer "
+                    "is highlighted at 28:23. The page carries no transcript."),
+    },
+    {
+        "url": "https://www.youtube.com/watch?v=V_qzqY1bb7I",
+        "title": "Reliability Lessons From SQLite - Richard Hipp | SSW 2026",
+        "speakers": ["Richard Hipp"],
+        "publisher": "Software Should Work",
+        "event": "SSW 2026",
+        "year": 2026,
+        "duration_seconds": 3261,
+        "related_dbms": ["sqlite"],
+        "frames": [
+            {"at_seconds": 2836, "matched": ["author"],
+             "note": ("The slide behind him is SQLite's project timeline: query "
+                      "fuzzing arrives around 2020, after 3.0.0, 100% MC/DC "
+                      "coverage and profile-guided fuzzing.")},
+        ],
+        "roles": ["recognition", "explains_technique"],
+        "summary": ("SQLite's creator on twenty-six years of testing it, "
+                    "crediting Rigger with the idea of fuzzing for "
+                    "inconsistencies in SQL rather than for crashes, "
+                    "explaining the sub-query equivalence that finds them, and "
+                    "saying SQLite's own fuzzer had to be extended to do the "
+                    "same."),
+    },
+    {
+        "url": "https://www.youtube.com/watch?v=CW4Ntdtp7lg",
+        "title": "Fuzzing databases is difficult",
+        "speakers": ["Pedro Ferreira"],
+        "publisher": "ClickHouse",
+        "event": "FOSDEM 2025",
+        "year": 2025,
+        "related_dbms": ["clickhouse"],
+        "duration_seconds": 1611,
+        "frames": [
+            {"at_seconds": 110, "matched": ["sqlancer"],
+             "note": ("The slide \"Testing with Fuzzers\" names the fuzzers "
+                      "ClickHouse runs, SQLancer first among them.")},
+            {"at_seconds": 827, "matched": ["sqlancer"],
+             "note": ("The slide \"Finding wrong results\" credits SQLancer by "
+                      "name with pioneering the comparison of equivalent "
+                      "queries against an oracle.")},
+        ],
+        "roles": ["reports_adoption", "recognition"],
+        "summary": ("A ClickHouse engineer on how the system is fuzzed. "
+                    "SQLancer appears on the early list of the fuzzers "
+                    "ClickHouse runs, again where the talk explains detecting "
+                    "wrong results by comparing equivalent queries, and once "
+                    "more in the closing recommendations."),
+        "watched": ("Reported by a maintainer who watched the talk: SQLancer "
+                    "is listed among ClickHouse's fuzzers, credited with "
+                    "pioneering the comparison of equivalent queries against "
+                    "an oracle, and recommended by name for query "
+                    "correctness. No transcript has been captured."),
+    },
+    {
+        "url": "https://www.youtube.com/watch?v=BgC79Zt2fPs",
+        "title": "Keynote 1: DuckDB Testing - Present and Future",
+        "speakers": ["Mark Raasveldt"],
+        "publisher": "DBTest Workshop",
+        "event": "DBTest 2022",
+        "year": 2022,
+        "duration_seconds": 3849,
+        "related_dbms": ["duckdb"],
+        "frames": [
+            {"at_seconds": 2144, "matched": ["author"],
+             "note": ("The slide behind the story: \u201cand then Dr. Rigger "
+                      "came along!\u201d, over a DuckDB bug report.")},
+            {"at_seconds": 2170, "matched": ["sqlancer", "author"],
+             "note": ("The next slide puts a number on it: the bugs SQLancer "
+                      "found in DuckDB, which the test suites DuckDB had "
+                      "borrowed from other systems did not.")},
+            {"at_seconds": 2455, "matched": ["sqlancer"],
+             "note": ("The robot that runs fuzzers in DuckDB's CI, and what it "
+                      "was running at the time: SQLancer and SQLsmith.")},
+        ],
+        "no_frame": [
+            {"at_seconds": 2268,
+             "why": ("The slide makes the general case for running many kinds "
+                     "of fuzzer; SQLancer is named only in what is said.")},
+        ],
+        "roles": ["recognition", "reports_adoption", "compares"],
+        "summary": ("DuckDB's testing, from one of its authors: the keynote "
+                    "credits SQLancer with around eighty bugs that the SQLite "
+                    "and Postgres test suites DuckDB had borrowed all missed, "
+                    "argues for running it beside SQLsmith because each finds "
+                    "what the other does not, and says the robot in DuckDB's "
+                    "CI runs both."),
+    },
+    {
+        "url": "https://www.youtube.com/watch?v=wHo-VtzTHx0",
+        "title": "CockroachDB's Query Optimizer (Rebecca Taft, Cockroach Labs)",
+        "speakers": ["Rebecca Taft"],
+        "publisher": "CMU Database Group",
+        "event": "CMU Quarantine Tech Talks 2020",
+        "year": 2020,
+        "duration_seconds": 3852,
+        "related_dbms": ["cockroachdb"],
+        "no_frame": [
+            {"at_seconds": 3507,
+             "why": ("A slide from the talk proper is still on screen; this "
+                     "mention is in the questions afterwards.")},
+        ],
+        "roles": ["recognition"],
+        "summary": ("CockroachDB's optimizer, from an engineer who builds it; "
+                    "asked in the questions about random testing, she separates "
+                    "SQLsmith's crashes from the logical bugs Rigger was after, "
+                    "credits him with a batch of GitHub issues against "
+                    "CockroachDB, and says she is trying to get SQLancer "
+                    "running in their own system."),
+    },
+    {
+        "url": "https://presentations.clickhouse.com/2021-cpp-siberia/index.html",
+        "title": "Fuzzing: Practical approaches in ClickHouse",
+        "speakers": ["Alexey Milovidov"],
+        "publisher": "ClickHouse",
+        "event": "C++ Siberia 2021",
+        "year": 2021,
+        "related_dbms": ["clickhouse"],
+        "slides": "https://presentations.clickhouse.com/2021-cpp-siberia/index.html",
+        "slide_image": "clickhouse-2021-cpp-siberia",
+        "roles": ["reports_adoption"],
+        "summary": ("A tour of everything ClickHouse fuzzes with, giving "
+                    "SQLancer a section of its own: who wrote it, who brought "
+                    "it into ClickHouse, and what it does."),
+    },
+    {
+        "url": "https://presentations.clickhouse.com/2022-release-22.3/index.html",
+        "title": "ClickHouse Release 22.3 Webinar",
+        "speakers": ["Alexey Milovidov"],
+        "publisher": "ClickHouse",
+        "event": "ClickHouse 22.3 release webinar",
+        "year": 2022,
+        "related_dbms": ["clickhouse"],
+        "slides": "https://presentations.clickhouse.com/2022-release-22.3/index.html",
+        "slide_image": "clickhouse-2022-release-22.3",
+        "roles": ["reports_adoption"],
+        "summary": ("The release webinar's section on continuous integration "
+                    "lists SQLancer among the fuzzing methods ClickHouse runs, "
+                    "beside libFuzzer, the AST query fuzzer and Jepsen."),
+    },
+)
+
+
+def deck_text(html: str) -> str:
+    """Readable text of a slide deck, for quoting."""
+    text = SCRIPT_RE.sub(" ", html)
+    text = TAG_RE.sub(" ", text)
+    from html import unescape
+    return re.sub(r"\s+", " ", unescape(text)).strip()
+
+
+def captured_transcript(url: str) -> Optional[dict]:
+    """A caption transcript captured from the browser, if one is on disk."""
+    found = video_id(url)
+    if not found:
+        return None
+    path = TRANSCRIPT_CACHE / f"{found}.json"
+    if not path.exists():
+        return None
+    try:
+        with path.open() as handle:
+            return json.load(handle)
+    except ValueError:
+        return None
+
+
+def collect(fetcher, *, timestamp: Optional[str] = None,
+            entries: Optional[Sequence[Dict[str, object]]] = None
+            ) -> Tuple[List[dict], List[dict]]:
+    """``(talks, needs_review)`` for the catalogue above."""
+    timestamp = timestamp or now()
+    tax = taxonomy.load()
+    records: List[dict] = []
+    needs_review: List[dict] = []
+
+    for item in (entries if entries is not None else TALKS):
+        url = str(item["url"])
+        mentions: List[dict] = []
+        sources: List[dict] = []
+
+        slides = item.get("slides")
+        if slides:
+            try:
+                entry, _ = fetcher.fetch(str(slides), max_age_days=90)
+            except Exception:
+                entry = {}
+            body = entry.get("body") or ""
+            if entry.get("status") == 200 and body:
+                text = deck_text(body)
+                found = findings_from_slides(text, url=str(slides), tax=tax)
+                image = frame_for(str(item.get("slide_image") or "")) or None
+                if image and found:
+                    # The deck's own words are quotable, so this picture sits
+                    # beside them rather than standing in for them.
+                    found[0]["image"] = image
+                mentions.extend(found)
+                sources.append({
+                    "kind": "slides", "status": "extracted", "url": str(slides),
+                    "retrieved_at": timestamp, "characters": len(text),
+                    "content_sha256": content_hash(text),
+                })
+                if not found:
+                    needs_review.append({
+                        "kind": "talk", "url": url,
+                        "reason": "slide deck does not mention SQLancer",
+                    })
+            else:
+                sources.append({"kind": "slides", "status": "unavailable",
+                                "url": str(slides), "retrieved_at": timestamp})
+                needs_review.append({"kind": "talk", "url": url,
+                                     "reason": "slide deck could not be fetched"})
+
+        captured = captured_transcript(url)
+        if captured:
+            found = findings_from_segments(captured.get("segments", []),
+                                           url=url, tax=tax)
+            for index, mention in enumerate(found):
+                mention["id"] = f"M{len(mentions) + index + 1}"
+            mentions.extend(found)
+            sources.append({
+                "kind": "captions", "status": "extracted", "url": url,
+                "retrieved_at": captured.get("retrieved_at") or timestamp,
+                "segments": captured.get("total_segments"),
+                "characters": captured.get("total_characters"),
+                # A transcript that was read and mentions nothing is a finding
+                # rather than a blank: it says the mention is on a slide, which
+                # is what the frames are for.
+                "note": (captured.get("note") if found else
+                         ("The whole transcript was read and SQLancer is not "
+                          "spoken anywhere in it.")),
+            })
+        elif video_id(url):
+            sources.append({"kind": "captions",
+                            "status": str(item.get("captions_status")
+                                          or "not_attempted"),
+                            "url": url,
+                            "note": item.get("captions_note")})
+
+        for frame in (item.get("frames") or ()):
+            at = int(frame["at_seconds"])
+            image = frame_for(video_id(url), at)
+            if not image:
+                needs_review.append({
+                    "kind": "talk", "url": url,
+                    "reason": (f"a frame is recorded at {timestamp_of(at)} but "
+                               "no image was captured for it"),
+                })
+                continue
+            sources.append({"kind": "frame", "status": "extracted",
+                            "url": deep_link(url, at), "retrieved_at": timestamp,
+                            "note": f"Captured at {timestamp_of(at)}."})
+            mentions.append({
+                "id": f"M{len(mentions) + 1}",
+                "source": "frame",
+                "at_seconds": at,
+                "timestamp": timestamp_of(at),
+                "url": deep_link(url, at),
+                "matched": list(frame.get("matched") or ["sqlancer"]),
+                "heard_as": None,
+                # Never an excerpt: text read off a picture cannot be checked
+                # against a fetched source the way every other excerpt here is.
+                "excerpt": None,
+                "excerpt_is_verbatim": False,
+                "note": str(frame["note"]),
+                "image": image,
+                "technique_ids": list(frame.get("technique_ids") or []),
+            })
+
+        watched = item.get("watched")
+        if watched and not mentions:
+            sources.append({"kind": "watched", "status": "extracted",
+                            "url": url, "note": str(watched)})
+            mentions.append({
+                "id": f"M{len(mentions) + 1}",
+                "source": "watched",
+                "at_seconds": None,
+                "timestamp": None,
+                "url": url,
+                "matched": ["sqlancer"],
+                "heard_as": None,
+                "excerpt": None,
+                "excerpt_is_verbatim": False,
+                "note": str(watched),
+                "image": None,
+                "technique_ids": [],
+            })
+
+        if not mentions:
+            needs_review.append({
+                "kind": "talk", "url": url,
+                "reason": "no source could be read for this talk",
+            })
+            continue
+
+        for checked in (item.get("no_frame") or ()):
+            for mention in mentions:
+                if mention.get("at_seconds") == int(checked["at_seconds"]):
+                    mention["frame_checked"] = str(checked["why"])
+
+        mentions.sort(key=lambda m: (m.get("at_seconds") is None,
+                                     m.get("at_seconds") or 0))
+        for index, mention in enumerate(mentions, start=1):
+            mention["id"] = f"M{index}"
+
+        techniques = list(item.get("related_techniques") or [])
+        for mention in mentions:
+            techniques.extend(mention.get("technique_ids") or [])
+        techniques = sorted({t for t in techniques if t in tax.techniques})
+
+        records.append({
+            "id": talk_id(url),
+            "title": str(item["title"]),
+            "url": url,
+            "video_id": video_id(url),
+            "speakers": list(item.get("speakers") or []),
+            "publisher": item.get("publisher"),
+            "event": item.get("event"),
+            "year": item.get("year"),
+            "duration_seconds": item.get("duration_seconds"),
+            "related_dbms": sorted(item.get("related_dbms") or []),
+            "related_techniques": techniques,
+            "sources": sources,
+            "mentions": mentions,
+            "relationship": {
+                "roles": list(item.get("roles") or ["background"]),
+                "summary": str(item["summary"]),
+                "mention_ids": [m["id"] for m in mentions],
+                "decided_by": "curator",
+            },
+            "provenance": {
+                "collector": COLLECTOR,
+                "collector_version": COLLECTOR_VERSION,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+                "source_url": url,
+                "source_type": "video",
+            },
+        })
+    return records, needs_review
diff --git a/tools/impact/config.py b/tools/impact/config.py
new file mode 100644
index 0000000..4206b6d
--- /dev/null
+++ b/tools/impact/config.py
@@ -0,0 +1,98 @@
+"""Paths, versions and shared constants for the impact pipeline."""
+
+from __future__ import annotations
+
+import json
+import os
+from pathlib import Path
+
+REPO_ROOT = Path(__file__).resolve().parents[2]
+
+DATA_DIR = REPO_ROOT / "_data" / "impact"
+SCHEMA_DIR = REPO_ROOT / "schemas" / "impact"
+CACHE_DIR = Path(os.environ.get("IMPACT_CACHE_DIR", REPO_ROOT / ".cache" / "impact"))
+PLOT_DIR = REPO_ROOT / "_includes" / "impact" / "plots"
+
+# Authoritative data files, keyed by the stem used everywhere in the pipeline.
+DATA_FILES = {
+    "techniques": DATA_DIR / "techniques.json",
+    "dbms": DATA_DIR / "dbms.json",
+    "bugs": DATA_DIR / "bugs.json",
+    "papers": DATA_DIR / "papers.json",
+    "adoption": DATA_DIR / "adoption.json",
+    "resources": DATA_DIR / "resources.json",
+    "talks": DATA_DIR / "talks.json",
+    "people": DATA_DIR / "people.json",
+    "policy": DATA_DIR / "policy.json",
+    "stats": DATA_DIR / "stats.json",
+}
+
+# Files the collectors may propose changes to. ``stats`` is generated, and
+# ``policy`` changes are a deliberate human act.
+COLLECTABLE = ("bugs", "papers", "adoption", "resources", "dbms", "people")
+
+SCHEMA_FILES = {name: SCHEMA_DIR / f"{name}.schema.json" for name in DATA_FILES}
+
+# Cache layers. Kept apart from the authoritative data on purpose: everything
+# here is an optimisation and the pipeline must work with an empty cache.
+CACHE_LAYERS = {
+    "http": CACHE_DIR / "http",             # raw HTTP bodies + validators
+    "github": CACHE_DIR / "github",         # issue/PR/repo metadata
+    "papers": CACHE_DIR / "papers",         # scholarly API responses
+    "artifacts": CACHE_DIR / "artifacts",   # research artifact metadata
+    "derived": CACHE_DIR / "derived",       # deterministic extraction results
+    "classifications": CACHE_DIR / "classifications",  # LLM answers
+}
+
+STATE_FILE = CACHE_DIR / "state.json"
+
+# Bumping any of these invalidates the affected cached classifications.
+CLASSIFIER_VERSIONS = {
+    "bug_attribution": "bug-attribution-v1",
+    "paper_infrastructure": "paper-infrastructure-v1",
+    "paper_extends": "paper-extends-v1",
+    "paper_compares": "paper-compares-v1",
+    "adoption": "adoption-v1",
+    "paper_summary": "paper-summary-v1",
+}
+
+DEFAULT_MODEL = os.environ.get("IMPACT_MODEL", "claude-opus-5")
+
+USER_AGENT = (
+    "sqlancer-impact-collector/1.0 "
+    "(+https://github.com/sqlancer/sqlancer.github.io)"
+)
+CONTACT_EMAIL = os.environ.get("IMPACT_CONTACT_EMAIL", "sqlancer@googlegroups.com")
+
+# How long a cached source is trusted before a reconciliation run re-checks it.
+RECONCILE_MAX_AGE_DAYS = int(os.environ.get("IMPACT_RECONCILE_DAYS", "90"))
+
+
+def load_json(path: Path):
+    with open(path, "r", encoding="utf-8") as handle:
+        return json.load(handle)
+
+
+def write_json(path: Path, payload) -> bool:
+    """Write ``payload`` deterministically. Returns True when bytes changed.
+
+    Stable formatting matters: the weekly run must be idempotent, so an
+    unchanged dataset has to serialise to byte-identical output.
+    """
+    path.parent.mkdir(parents=True, exist_ok=True)
+    text = json.dumps(payload, indent=2, ensure_ascii=False, sort_keys=False) + "\n"
+    if path.exists():
+        with open(path, "r", encoding="utf-8") as handle:
+            if handle.read() == text:
+                return False
+    with open(path, "w", encoding="utf-8") as handle:
+        handle.write(text)
+    return True
+
+
+def policy_version() -> str:
+    return load_json(DATA_FILES["policy"])["policy_version"]
+
+
+def taxonomy_version() -> str:
+    return load_json(DATA_FILES["techniques"])["taxonomy_version"]
diff --git a/tools/impact/dataset.py b/tools/impact/dataset.py
new file mode 100644
index 0000000..d00d0b4
--- /dev/null
+++ b/tools/impact/dataset.py
@@ -0,0 +1,257 @@
+"""Loading, merging and writing the authoritative impact records.
+
+The merge rules here are what make the weekly run idempotent. A collector
+produces candidate records; ``Dataset.merge`` decides, per record, whether it is
+new, whether it genuinely changed, or whether it is the same record seen again.
+Fields that would otherwise churn on every run -- ``last_verified`` above all --
+are only refreshed when something else about the record actually changed, so a
+run with no new evidence rewrites no bytes and opens no pull request.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from typing import Any, Dict, Iterable, List, Optional, Tuple
+
+from . import config
+from .util import normalize_url
+
+
+# Record collections keyed by the list field inside each data file.
+LIST_FIELD = {
+    "bugs": "bugs",
+    "papers": "papers",
+    "adoption": "adoption",
+    "resources": "resources",
+    "dbms": "dbms",
+    "people": "people",
+}
+
+# Sort keys giving each file a stable, human-meaningful order.
+SORT_KEY = {
+    "bugs": lambda r: (r.get("dbms") or "", r.get("reported_date") or "",
+                       r.get("title") or "", r["id"]),
+    "papers": lambda r: (-(r.get("year") or 0), (r.get("title") or "").lower(), r["id"]),
+    "adoption": lambda r: (r.get("dbms") or "", r.get("relationship") or "", r["id"]),
+    "resources": lambda r: (-(r.get("year") or 0), (r.get("title") or "").lower(), r["id"]),
+    "people": lambda r: (-(r.get("reports_on_lab_list") or 0),
+                         (r.get("name") or "").lower(), r["id"]),
+    "dbms": lambda r: r["id"],
+}
+
+# Fields that describe when a record was last looked at rather than what it
+# says. Differences here alone never count as a change.
+_VOLATILE = {"last_verified", "retrieved_at", "stored_at", "inspected_at",
+             "classified_at", "fetched_at"}
+
+
+def _strip_volatile(value: Any) -> Any:
+    """Deep copy with bookkeeping timestamps removed, for change detection."""
+    if isinstance(value, dict):
+        return {k: _strip_volatile(v) for k, v in sorted(value.items())
+                if k not in _VOLATILE}
+    if isinstance(value, list):
+        return [_strip_volatile(v) for v in value]
+    return value
+
+
+def substantively_equal(left: dict, right: dict) -> bool:
+    return _strip_volatile(left) == _strip_volatile(right)
+
+
+def _touch_last_verified(record: Any, timestamp: str) -> None:
+    """Refresh every ``last_verified`` inside a record that did change."""
+    if isinstance(record, dict):
+        for key, value in record.items():
+            if key == "last_verified" and isinstance(value, str):
+                record[key] = timestamp
+            else:
+                _touch_last_verified(value, timestamp)
+    elif isinstance(record, list):
+        for item in record:
+            _touch_last_verified(item, timestamp)
+
+
+def _preserve_first_seen(new: Any, old: Any) -> None:
+    """Carry ``first_seen`` over from the stored record.
+
+    When a record is updated, the date it entered the dataset must not move.
+    """
+    if isinstance(new, dict) and isinstance(old, dict):
+        if "first_seen" in new and isinstance(old.get("first_seen"), str):
+            new["first_seen"] = old["first_seen"]
+        for key, value in new.items():
+            if key in old:
+                _preserve_first_seen(value, old[key])
+    elif isinstance(new, list) and isinstance(old, list):
+        # Evidence lists are matched by source_url so re-ordering is harmless.
+        by_url = {}
+        for item in old:
+            if isinstance(item, dict) and item.get("source_url"):
+                by_url[normalize_url(item["source_url"])] = item
+        for item in new:
+            if isinstance(item, dict) and item.get("source_url"):
+                counterpart = by_url.get(normalize_url(item["source_url"]))
+                if counterpart is not None:
+                    _preserve_first_seen(item, counterpart)
+
+
+@dataclass
+class MergeReport:
+    """What a merge changed, used to build the pull-request description."""
+
+    added: List[dict] = field(default_factory=list)
+    updated: List[Tuple[dict, dict]] = field(default_factory=list)
+    unchanged: int = 0
+    rejected: List[dict] = field(default_factory=list)
+    needs_review: List[dict] = field(default_factory=list)
+    removed: List[dict] = field(default_factory=list)
+
+    @property
+    def changed(self) -> bool:
+        return bool(self.added or self.updated or self.removed)
+
+    def extend(self, other: "MergeReport") -> None:
+        self.added.extend(other.added)
+        self.updated.extend(other.updated)
+        self.unchanged += other.unchanged
+        self.rejected.extend(other.rejected)
+        self.needs_review.extend(other.needs_review)
+        self.removed.extend(other.removed)
+
+
+class Dataset:
+    """In-memory view of every authoritative file, with change tracking."""
+
+    def __init__(self, files: Optional[Dict[str, Any]] = None):
+        self.files = dict(files or config.DATA_FILES)
+        self.data: Dict[str, dict] = {}
+        for name, path in self.files.items():
+            if path.exists():
+                self.data[name] = config.load_json(path)
+        self.reports: Dict[str, MergeReport] = {}
+
+    # -- accessors --------------------------------------------------------
+
+    def records(self, name: str) -> List[dict]:
+        payload = self.data.get(name)
+        if payload is None:
+            return []
+        return payload.get(LIST_FIELD[name], [])
+
+    def index(self, name: str) -> Dict[str, dict]:
+        return {record["id"]: record for record in self.records(name)}
+
+    def url_index(self, name: str, url_field: str = "primary_url") -> Dict[str, dict]:
+        """Secondary index on a canonical URL, used to catch id-scheme drift.
+
+        URLs shared by more than one record are omitted: an upstream ticket can
+        cover two separately curated reports, and matching on it would silently
+        merge them.
+        """
+        seen: Dict[str, List[dict]] = {}
+        for record in self.records(name):
+            url = normalize_url(record.get(url_field))
+            if url:
+                seen.setdefault(url, []).append(record)
+        return {url: records[0] for url, records in seen.items() if len(records) == 1}
+
+    def ensure(self, name: str, header: Dict[str, Any]) -> dict:
+        payload = self.data.get(name)
+        if payload is None:
+            payload = dict(header)
+            payload[LIST_FIELD[name]] = []
+            self.data[name] = payload
+        return payload
+
+    # -- merging ----------------------------------------------------------
+
+    def merge(self, name: str, candidates: Iterable[dict], *, timestamp: str,
+              rejected: Optional[Iterable[dict]] = None,
+              needs_review: Optional[Iterable[dict]] = None,
+              authoritative: bool = False) -> MergeReport:
+        """Fold ``candidates`` into the stored records for ``name``.
+
+        Deduplication is by stable id first and by canonical URL second, so a
+        record keeps its identity even if the id scheme is ever revised.
+
+        With ``authoritative``, ``candidates`` is the complete set the source
+        now yields, and stored records missing from it are removed. Only pass it
+        for a collector that re-derives everything, such as a full paper
+        reconciliation -- an incremental run has not looked at the records it
+        would be deleting. Removals are reported, never silent, because a record
+        leaving the dataset is exactly as reviewable as one arriving.
+        """
+        report = MergeReport(rejected=list(rejected or []),
+                             needs_review=list(needs_review or []))
+        candidates_list = list(candidates)
+        seen_ids = set()
+        payload = self.data.get(name)
+        if payload is None:
+            raise KeyError(f"no data file loaded for {name!r}")
+        records = payload.setdefault(LIST_FIELD[name], [])
+        by_id = {record["id"]: record for record in records}
+        by_url = self.url_index(name)
+
+        for candidate in candidates_list:
+            existing = by_id.get(candidate["id"])
+            if existing is None:
+                url = normalize_url(candidate.get("primary_url"))
+                if url:
+                    # pop, so two candidates sharing a URL cannot both claim
+                    # the same stored record.
+                    existing = by_url.pop(url, None)
+            if existing is None:
+                report.added.append(candidate)
+                records.append(candidate)
+                by_id[candidate["id"]] = candidate
+                seen_ids.add(candidate["id"])
+                continue
+
+            seen_ids.add(existing["id"])
+            merged = dict(candidate)
+            merged["id"] = existing["id"]  # never renumber a published record
+            _preserve_first_seen(merged, existing)
+            if substantively_equal(merged, existing):
+                report.unchanged += 1
+                continue
+            _touch_last_verified(merged, timestamp)
+            before = dict(existing)
+            existing.clear()
+            existing.update(merged)
+            report.updated.append((before, existing))
+
+        # An empty candidate list under `authoritative` almost always means the
+        # collector failed rather than that everything genuinely went away, so
+        # it never deletes the dataset.
+        if authoritative and candidates_list:
+            keep = {record["id"] for record in candidates_list}
+            survivors = []
+            for record in records:
+                if record["id"] in keep or record["id"] in seen_ids:
+                    survivors.append(record)
+                else:
+                    report.removed.append(record)
+            if report.removed:
+                records[:] = survivors
+
+        records.sort(key=SORT_KEY[name])
+        previous = self.reports.get(name)
+        if previous is not None:
+            previous.extend(report)
+            self.reports[name] = previous
+        else:
+            self.reports[name] = report
+        return report
+
+    # -- persistence ------------------------------------------------------
+
+    def save(self, names: Optional[Iterable[str]] = None) -> Dict[str, bool]:
+        """Write files back; returns which ones changed on disk."""
+        changed: Dict[str, bool] = {}
+        for name in (names or self.data):
+            payload = self.data.get(name)
+            if payload is None:
+                continue
+            changed[name] = config.write_json(self.files[name], payload)
+        return changed
diff --git a/tools/impact/dedupe.py b/tools/impact/dedupe.py
new file mode 100644
index 0000000..f91da24
--- /dev/null
+++ b/tools/impact/dedupe.py
@@ -0,0 +1,325 @@
+"""Finds paper records that are the same work under two identities.
+
+A paper reaches this dataset through two citation graphs, and both index the
+preprint and the published version as separate works. Nothing downstream
+noticed: the survey on DBMS fuzzing was counted twice, AMOEBA three times, and
+the download list asked for papers already in hand under their other name.
+
+The test for sameness is authorship first. Two records with the same full set of
+author surnames are by the same people; the question is then whether they are
+the same paper or two papers by one group, and title similarity alone cannot
+tell those apart -- "Dependency-Aware Metamorphic Testing of Datalog Engines"
+and "Metamorphic testing of Datalog engines" score 0.82 and are genuinely
+different papers, exactly like the two SQLaser records that are not.
+
+What separates them is the preprint relationship. A preprint and a published
+paper by the same authors with similar titles are one work; two published papers
+by the same authors are two works, however alike their titles. So a merge is
+proposed automatically only across that boundary, or when the titles are
+practically identical. Everything else is reported for a person to decide.
+"""
+
+from __future__ import annotations
+
+import difflib
+import re
+from collections import defaultdict
+from typing import Dict, List, Optional, Sequence, Tuple
+
+# A preprint and its published version, when there are no abstracts to compare.
+PREPRINT_TITLE_SIMILARITY = 0.60
+
+# Near-identical titles settle it whatever else disagrees.
+IDENTICAL_TITLE_SIMILARITY = 0.90
+
+# Two abstracts this alike are the same paper. This is the strongest signal
+# available and it is why the property-based-testing pair is not merged: same
+# authors, same year, titles 0.75 alike, abstracts 0.09 -- a position paper and
+# a full paper, not one work twice.
+ABSTRACT_SIMILARITY = 0.60
+
+# Same authors, plausibly the same work, but nothing above settled it. Reported
+# for a person rather than merged. The conformance-testing pair lands here: the
+# published version rewrote its abstract and added two systems, so it scores
+# 0.04 against its own preprint.
+REVIEW_TITLE_SIMILARITY = 0.60
+
+PREPRINT_VENUE = re.compile(r"arxiv|preprint|corr\b", re.IGNORECASE)
+
+
+def surname(name: str) -> str:
+    parts = [part for part in re.sub(r"[^A-Za-z .-]", " ", name or "").split()
+             if part]
+    return parts[-1].lower() if parts else ""
+
+
+def author_key(paper: dict) -> frozenset:
+    return frozenset(s for s in (surname(a) for a in (paper.get("authors") or []))
+                     if s)
+
+
+# A tool's name as papers write it in a title: capitalised, distinctive, and
+# usually before the colon. Common words are excluded; so are acronyms a whole
+# subfield shares.
+TOOL_NAME = re.compile(r"\b([A-Z][A-Za-z]*[A-Z][A-Za-z0-9]*|[A-Z][a-z]{2,}[A-Z]\w*)\b")
+NOT_A_TOOL = {
+    "DBMS", "DBMSs", "SQL", "RDBMS", "RDBMSs", "GDBMS", "GDBMSs", "TSMS",
+    "DBMSes", "LLM", "LLMs", "API", "APIs", "CI", "IO", "OS", "AI", "ML",
+    "ACID", "OLTP", "OLAP", "JDBC", "ORM", "PHP", "JIT", "SMT", "TCP", "FHE",
+    "GPU", "CPU", "DISC", "NoSQL", "NewSQL", "PBT", "MPC", "ZK",
+}
+
+
+def tool_names(title: str) -> List[str]:
+    """Distinctive tool names a title contains, such as "DBStorm"."""
+    head = title.split(":")[0] if ":" in title else title
+    found = []
+    for match in TOOL_NAME.finditer(head):
+        name = match.group(1)
+        if name not in NOT_A_TOOL and len(name) >= 4:
+            found.append(name)
+    return found
+
+
+def normalise_title(title: str) -> str:
+    return " ".join(re.sub(r"[^a-z0-9 ]+", " ", (title or "").lower()).split())
+
+
+def title_similarity(left: str, right: str) -> float:
+    return difflib.SequenceMatcher(None, normalise_title(left),
+                                   normalise_title(right)).ratio()
+
+
+def abstract_similarity(left: dict, right: dict) -> Optional[float]:
+    """How alike two records' abstracts are, or None if one is missing."""
+    first = (left.get("abstract") or "").strip().lower()
+    second = (right.get("abstract") or "").strip().lower()
+    if len(first) < 200 or len(second) < 200:
+        return None
+    return difflib.SequenceMatcher(None, first[:1500], second[:1500]).ratio()
+
+
+def is_preprint(paper: dict) -> bool:
+    """Whether this record is a preprint rather than a publication."""
+    if paper.get("arxiv_id") and not paper.get("doi"):
+        return True
+    return bool(PREPRINT_VENUE.search(paper.get("venue") or ""))
+
+
+def _rank(paper: dict) -> tuple:
+    """How good a canonical record this is. Higher wins.
+
+    A published record beats a preprint, a DOI beats no DOI, and more evidence
+    beats less -- the surviving record should be the one a reader would cite.
+    """
+    evidence = sum(len(entry.get("evidence") or [])
+                   for entry in (paper.get("relationships") or {}).values())
+    return (0 if is_preprint(paper) else 1,
+            1 if paper.get("doi") else 0,
+            paper.get("year") or 0,
+            evidence)
+
+
+def load_decisions() -> Tuple[List[dict], List[frozenset]]:
+    """Merges and non-merges a person has decided, with their reasons."""
+    from . import config
+
+    path = config.DATA_DIR / "paper_decisions.json"
+    if not path.exists():
+        return [], []
+    data = config.load_json(path)
+    same = list(data.get("same_work") or [])
+    different = [frozenset(entry["ids"]) for entry in data.get("different_work") or []]
+    return same, different
+
+
+def find_duplicates(papers: Sequence[dict]
+                    ) -> Tuple[List[List[dict]], List[Tuple[dict, dict, float]]]:
+    """``(groups_to_merge, pairs_for_review)``.
+
+    A group is every record of one work, best canonical record first.
+    """
+    considered = [p for p in papers if not p.get("is_sqlancer_publication")]
+    by_authors: Dict[frozenset, List[dict]] = defaultdict(list)
+    for paper in considered:
+        names = author_key(paper)
+        if len(names) >= 2:      # one shared surname is a coincidence
+            by_authors[names].append(paper)
+
+    parent: Dict[str, str] = {p["id"]: p["id"] for p in considered}
+
+    def find(node: str) -> str:
+        while parent[node] != node:
+            parent[node] = parent[parent[node]]
+            node = parent[node]
+        return node
+
+    def union(left: str, right: str) -> None:
+        a, b = find(left), find(right)
+        if a != b:
+            parent[b] = a
+
+    decided_same, decided_different = load_decisions()
+    settled = {frozenset(entry["ids"]) for entry in decided_same} | set(decided_different)
+
+    review: List[Tuple[dict, dict, float]] = []
+    for members in by_authors.values():
+        for index, left in enumerate(members):
+            for right in members[index + 1:]:
+                if frozenset((left["id"], right["id"])) in settled:
+                    continue      # a person has already ruled on this pair
+                score = title_similarity(left["title"], right["title"])
+                abstracts = abstract_similarity(left, right)
+                across_preprint = is_preprint(left) != is_preprint(right)
+
+                if score >= IDENTICAL_TITLE_SIMILARITY:
+                    union(left["id"], right["id"])
+                elif abstracts is not None:
+                    # With both abstracts in hand, they decide.
+                    if abstracts >= ABSTRACT_SIMILARITY:
+                        union(left["id"], right["id"])
+                    elif score >= REVIEW_TITLE_SIMILARITY:
+                        review.append((left, right, score))
+                elif across_preprint and score >= PREPRINT_TITLE_SIMILARITY:
+                    union(left["id"], right["id"])
+                elif score >= REVIEW_TITLE_SIMILARITY:
+                    review.append((left, right, score))
+
+    # Merges a person decided, which no automatic rule was going to reach.
+    known = {paper["id"] for paper in considered}
+    for entry in decided_same:
+        ids = [pid for pid in entry["ids"] if pid in known]
+        for other in ids[1:]:
+            union(ids[0], other)
+
+    # An index stub can carry no authors at all, and then nothing above can
+    # compare it with anything. What such a record still has is the name of the
+    # tool it is about, which a paper puts in its title and almost never shares
+    # with an unrelated paper. Two records naming the same tool are raised for a
+    # person -- never merged automatically, because a group really can publish
+    # twice about one tool.
+    authorless = [p for p in considered if not author_key(p)]
+    if authorless:
+        by_tool: Dict[str, List[dict]] = defaultdict(list)
+        for paper in considered:
+            for name in tool_names(paper["title"]):
+                by_tool[name].append(paper)
+        for name, members in by_tool.items():
+            if len(members) < 2:
+                continue
+            for index, left in enumerate(members):
+                for right in members[index + 1:]:
+                    if not author_key(left) or not author_key(right):
+                        pair = frozenset((left["id"], right["id"]))
+                        if pair in settled or find(left["id"]) == find(right["id"]):
+                            continue
+                        review.append((left, right,
+                                       title_similarity(left["title"],
+                                                        right["title"])))
+
+    grouped: Dict[str, List[dict]] = defaultdict(list)
+    for paper in considered:
+        grouped[find(paper["id"])].append(paper)
+
+    chosen = {entry["canonical"] for entry in decided_same if entry.get("canonical")}
+    groups = [sorted(members,
+                     key=lambda paper: (paper["id"] in chosen, _rank(paper)),
+                     reverse=True)
+              for members in grouped.values() if len(members) > 1]
+    groups.sort(key=lambda members: members[0]["title"].lower())
+    # A pair already merged needs no review.
+    merged_ids = {p["id"] for group in groups for p in group}
+    review = [(a, b, s) for a, b, s in review
+              if not (a["id"] in merged_ids and find(a["id"]) == find(b["id"]))]
+    return groups, review
+
+
+def merge(group: Sequence[dict]) -> dict:
+    """One record for the work, keeping everything either record knew.
+
+    The published version survives, but the preprint's identifiers and evidence
+    are not thrown away: an arXiv id is how the pipeline reads a paper's full
+    text, so losing it would cost the merged record its own source.
+    """
+    canonical = dict(group[0])
+    others = group[1:]
+
+    for other in others:
+        for field in ("doi", "arxiv_id", "s2_paper_id", "abstract",
+                      "open_access_pdf", "url", "venue", "year"):
+            if not canonical.get(field) and other.get(field):
+                canonical[field] = other[field]
+        canonical["cites_seed_techniques"] = sorted(
+            set(canonical.get("cites_seed_techniques") or [])
+            | set(other.get("cites_seed_techniques") or []))
+
+        for key, entry in (other.get("relationships") or {}).items():
+            mine = canonical.setdefault("relationships", {}).setdefault(
+                key, {"value": "insufficient_evidence", "method": "deterministic",
+                      "evidence": []})
+            seen = {(item.get("source_url"), item.get("excerpt"))
+                    for item in (mine.get("evidence") or [])}
+            for item in (entry.get("evidence") or []):
+                if (item.get("source_url"), item.get("excerpt")) not in seen:
+                    mine.setdefault("evidence", []).append(item)
+                    seen.add((item.get("source_url"), item.get("excerpt")))
+            # A positive finding on either record is a finding about the work.
+            if entry.get("value") == "yes" and mine.get("value") != "yes":
+                mine["value"] = "yes"
+                mine["method"] = entry.get("method", "deterministic")
+            for technique in entry.get("techniques") or []:
+                mine.setdefault("techniques", [])
+                if technique not in mine["techniques"]:
+                    mine["techniques"].append(technique)
+
+        if other.get("artifacts") and not canonical.get("artifacts"):
+            canonical["artifacts"] = other["artifacts"]
+
+    canonical["also_indexed_as"] = sorted(
+        {other["id"] for other in others}
+        | set(canonical.get("also_indexed_as") or []))
+    return canonical
+
+
+def main(*, apply: bool = False) -> int:
+    """Report duplicate records, and merge them with ``--apply``."""
+    import json
+
+    from . import config
+
+    path = config.DATA_FILES["papers"]
+    payload = config.load_json(path)
+    groups, review = find_duplicates(payload["papers"])
+
+    for group in groups:
+        print(f"same work: {group[0]['title'][:66]}")
+        print(f"    keep  {group[0]['id']}")
+        for other in group[1:]:
+            print(f"    merge {other['id']}")
+    for left, right, score in review:
+        print(f"undecided ({score:.2f}): {left['id']} / {right['id']}")
+        print(f"    {left['title'][:70]}")
+        print(f"    {right['title'][:70]}")
+        print("    decide it in _data/impact/paper_decisions.json")
+
+    if not apply:
+        print(f"\n{len(groups)} group(s) to merge, {len(review)} undecided "
+              f"(nothing written; pass --apply)")
+        return 1 if review else 0
+
+    if review:
+        print("\nrefusing to merge while pairs are undecided")
+        return 1
+
+    replaced, merged_records = {}, {}
+    for group in groups:
+        record = merge(group)
+        merged_records[record["id"]] = record
+        for other in group[1:]:
+            replaced[other["id"]] = record["id"]
+    payload["papers"] = [merged_records.get(p["id"], p) for p in payload["papers"]
+                         if p["id"] not in replaced]
+    path.write_text(json.dumps(payload, indent=2, ensure_ascii=False) + "\n",
+                    encoding="utf-8")
+    print(f"\nmerged {len(replaced)} record(s); {len(payload['papers'])} remain")
+    return 0
diff --git a/tools/impact/github.py b/tools/impact/github.py
new file mode 100644
index 0000000..abb2134
--- /dev/null
+++ b/tools/impact/github.py
@@ -0,0 +1,181 @@
+"""Thin GitHub REST client on top of the caching fetcher.
+
+Only the handful of endpoints the collectors need. Two things matter here
+beyond plain requests: issue and PR metadata carry an ``updated_at`` that lets
+the raw cache skip a request entirely for unchanged threads, and the search
+endpoints support ``created:``/``updated:`` qualifiers that make weekly scans
+incremental instead of full rescans.
+"""
+
+from __future__ import annotations
+
+import os
+import time
+from typing import Any, Dict, Iterator, List, Optional, Tuple
+
+from .cache import RawCache
+from .http import Fetcher, with_params
+
+API = "https://api.github.com"
+
+# GitHub's search API is capped at 1000 results per query and rate limited far
+# more tightly than the core API, so searches are kept narrow and paged.
+SEARCH_PAGE_SIZE = 100
+SEARCH_MAX_PAGES = 10
+
+
+class BadCredentials(RuntimeError):
+    """The token GitHub was given is not a valid one."""
+
+
+def _token() -> Optional[str]:
+    for name in ("GITHUB_TOKEN", "GH_TOKEN", "IMPACT_GITHUB_TOKEN"):
+        value = os.environ.get(name)
+        if value:
+            return value
+    return None
+
+
+class GitHub:
+    def __init__(self, cache: RawCache, *, token: Optional[str] = None):
+        self.token = token or _token()
+        headers = {
+            "Accept": "application/vnd.github+json",
+            "X-GitHub-Api-Version": "2022-11-28",
+        }
+        if self.token:
+            headers["Authorization"] = f"Bearer {self.token}"
+        # GitHub rate limits search far more tightly than the rest of the API:
+        # 30 searches per minute authenticated, against 5000 core requests per
+        # hour. A single interval for both would make reading a few hundred
+        # issue bodies take an hour for no reason, so the two get their own
+        # budgets.
+        self.fetcher = Fetcher(cache, min_interval=0.8 if self.token else 3.0,
+                               headers=headers)
+        self.search_fetcher = Fetcher(
+            cache, min_interval=2.5 if self.token else 6.0, headers=headers)
+        self.authenticated = bool(self.token)
+        # Searches that came back non-200. A run that silently found nothing
+        # and a run that genuinely found nothing look identical otherwise.
+        self.failed_searches: List[Tuple[str, str, Optional[int]]] = []
+
+    # -- primitive --------------------------------------------------------
+
+    def get(self, path: str, *, params: Optional[Dict[str, Any]] = None,
+            cache_key: Optional[str] = None, max_age_days: Optional[int] = None,
+            upstream_updated_at: Optional[str] = None, force: bool = False):
+        url = API + path if path.startswith("/") else path
+        if params:
+            url = with_params(url, params)
+        data, from_cache = self.fetcher.fetch_json(
+            url, cache_key=cache_key or url, max_age_days=max_age_days,
+            upstream_updated_at=upstream_updated_at, force=force)
+        return data, from_cache
+
+    # -- repository content ----------------------------------------------
+
+    def tree(self, owner: str, repo: str, ref: str = "HEAD",
+             *, max_age_days: int = 7) -> List[dict]:
+        data, _ = self.get(f"/repos/{owner}/{repo}/git/trees/{ref}",
+                           params={"recursive": "1"}, max_age_days=max_age_days)
+        return (data or {}).get("tree", [])
+
+    def raw_file(self, owner: str, repo: str, path: str, ref: str = "HEAD",
+                 *, max_age_days: int = 30):
+        """Fetch a file's text from raw.githubusercontent.com.
+
+        The raw host is used rather than the contents API because it returns the
+        bytes directly, which keeps the stored evidence excerpt and its hash
+        aligned with what a reviewer sees when following the URL.
+        """
+        url = f"https://raw.githubusercontent.com/{owner}/{repo}/{ref}/{path}"
+        entry, _ = self.fetcher.fetch(url, max_age_days=max_age_days)
+        if entry.get("status") != 200:
+            return None, entry
+        return entry["body"], entry
+
+    def repo(self, owner: str, name: str, *, max_age_days: int = 7):
+        data, _ = self.get(f"/repos/{owner}/{name}", max_age_days=max_age_days)
+        return data
+
+    def issue(self, owner: str, repo: str, number: int, *,
+              upstream_updated_at: Optional[str] = None,
+              max_age_days: Optional[int] = None):
+        data, from_cache = self.get(
+            f"/repos/{owner}/{repo}/issues/{number}",
+            upstream_updated_at=upstream_updated_at, max_age_days=max_age_days)
+        return data, from_cache
+
+    def issue_comments(self, owner: str, repo: str, number: int, *,
+                       upstream_updated_at: Optional[str] = None,
+                       max_age_days: Optional[int] = None) -> List[dict]:
+        data, _ = self.get(f"/repos/{owner}/{repo}/issues/{number}/comments",
+                           params={"per_page": 100},
+                           upstream_updated_at=upstream_updated_at,
+                           max_age_days=max_age_days)
+        return data or []
+
+    # -- search -----------------------------------------------------------
+
+    def search(self, kind: str, query: str, *, sort: Optional[str] = None,
+               order: str = "desc", max_pages: int = SEARCH_MAX_PAGES,
+               max_age_days: Optional[int] = 7,
+               accept: Optional[str] = None) -> Iterator[dict]:
+        """Page through ``/search/<kind>``.
+
+        ``kind`` is one of ``issues``, ``repositories`` or ``code``. Code search
+        needs a token and a preview Accept header for text matches, which are
+        what make a hit usable as evidence.
+        """
+        for page in range(1, max_pages + 1):
+            params = {"q": query, "per_page": SEARCH_PAGE_SIZE, "page": page,
+                      "order": order}
+            if sort:
+                params["sort"] = sort
+            url = with_params(f"{API}/search/{kind}", params)
+            extra = {"Accept": accept} if accept else None
+            try:
+                entry, _ = self.search_fetcher.fetch(
+                    url, max_age_days=max_age_days, extra_headers=extra)
+            except RuntimeError:
+                # A failing search must not abort the whole run; the periodic
+                # reconciliation will pick up anything missed.
+                return
+            if entry.get("status") == 401:
+                # Not transient, and not something reconciliation fixes: every
+                # search in the run is broken the same way. Returning quietly
+                # here made an expired token look like a world with no bugs --
+                # the run reported "0 new" and nothing said why.
+                raise BadCredentials(
+                    "GitHub rejected the token (401). Set a valid GITHUB_TOKEN "
+                    "or unset it to search unauthenticated.")
+            if entry.get("status") != 200:
+                self.failed_searches.append((kind, query, entry.get("status")))
+                return
+            import json
+            try:
+                payload = json.loads(entry["body"])
+            except ValueError:
+                return
+            items = payload.get("items", [])
+            for item in items:
+                yield item
+            if len(items) < SEARCH_PAGE_SIZE:
+                return
+            if page * SEARCH_PAGE_SIZE >= min(payload.get("total_count", 0), 1000):
+                return
+
+    def search_issues(self, query: str, *, since: Optional[str] = None, **kwargs):
+        """Issue search, optionally restricted to threads updated since a date."""
+        if since:
+            query = f"{query} updated:>={since[:10]}"
+        return self.search("issues", query, **kwargs)
+
+    def search_code(self, query: str, **kwargs):
+        return self.search(
+            "code", query,
+            accept="application/vnd.github.text-match+json", **kwargs)
+
+    def rate_limit(self) -> dict:
+        data, _ = self.get("/rate_limit", max_age_days=None, force=True)
+        return data or {}
diff --git a/tools/impact/http.py b/tools/impact/http.py
new file mode 100644
index 0000000..18cda30
--- /dev/null
+++ b/tools/impact/http.py
@@ -0,0 +1,253 @@
+"""HTTP access with conditional requests, backoff and a hard offline mode.
+
+Everything the pipeline fetches goes through here so that three properties hold
+uniformly: unchanged sources are revalidated rather than re-downloaded, rate
+limits are respected with backoff, and the whole pipeline can be run with
+``IMPACT_OFFLINE=1`` against nothing but the cache, which is what the tests do.
+"""
+
+from __future__ import annotations
+
+import json
+import os
+import time
+import urllib.error
+import urllib.parse
+import urllib.request
+from typing import Any, Dict, Optional, Tuple
+
+from . import config
+from .cache import RawCache
+from .util import now
+
+OFFLINE = os.environ.get("IMPACT_OFFLINE", "") not in ("", "0", "false")
+
+
+class OfflineError(RuntimeError):
+    """Raised when a fetch is required but the pipeline is running offline."""
+
+
+class Fetcher:
+    """Cache-aware HTTP client.
+
+    ``cache`` is the raw-source layer for this kind of source; passing separate
+    caches for GitHub, papers and generic HTTP keeps the cache directories
+    meaningful when a human inspects them.
+    """
+
+    def __init__(self, cache: RawCache, *, min_interval: float = 0.0,
+                 headers: Optional[Dict[str, str]] = None,
+                 max_retries: int = 5, initial_backoff: float = 2.0,
+                 max_backoff: float = 60.0, timeout: float = 30.0):
+        self.cache = cache
+        self.min_interval = min_interval
+        self.max_retries = max_retries
+        self.initial_backoff = initial_backoff
+        self.max_backoff = max_backoff
+        # Bounded on purpose: several trackers linked from historic bug lists
+        # no longer resolve, and a generous socket timeout multiplied by the
+        # retry count is what turns a handful of dead links into a stalled run.
+        self.timeout = timeout
+        self.headers = {"User-Agent": config.USER_AGENT, "Accept": "*/*"}
+        if headers:
+            self.headers.update(headers)
+        self._last_request = 0.0
+        self.requests_made = 0
+
+    def _throttle(self) -> None:
+        if self.min_interval <= 0:
+            return
+        elapsed = time.monotonic() - self._last_request
+        if elapsed < self.min_interval:
+            time.sleep(self.min_interval - elapsed)
+        self._last_request = time.monotonic()
+
+    def fetch(self, url: str, *, cache_key: Optional[str] = None,
+              upstream_updated_at: Optional[str] = None,
+              max_age_days: Optional[int] = None,
+              force: bool = False,
+              extra_headers: Optional[Dict[str, str]] = None
+              ) -> Tuple[Dict[str, Any], bool]:
+        """Return ``(entry, from_cache)`` for ``url``.
+
+        When the caller knows an upstream modification timestamp, an unchanged
+        timestamp short-circuits the request entirely. Otherwise a cached ETag
+        or Last-Modified is replayed as a conditional request, so a ``304`` also
+        avoids re-downloading the body.
+        """
+        key = cache_key or url
+        cached = self.cache.get(key)
+
+        if not force and cached is not None:
+            if self.cache.is_fresh(key, upstream_updated_at=upstream_updated_at,
+                                   max_age_days=max_age_days):
+                return cached["value"], True
+        if OFFLINE:
+            if cached is not None:
+                return cached["value"], True
+            raise OfflineError(f"offline and no cached copy of {url}")
+
+        headers = dict(self.headers)
+        if extra_headers:
+            headers.update(extra_headers)
+        if cached is not None and not force:
+            if cached["value"].get("etag"):
+                headers["If-None-Match"] = cached["value"]["etag"]
+            if cached["value"].get("last_modified"):
+                headers["If-Modified-Since"] = cached["value"]["last_modified"]
+
+        body, status, response_headers = self._request(url, headers)
+        if status == 304 and cached is not None:
+            value = dict(cached["value"])
+            value["fetched_at"] = now()
+            if upstream_updated_at is not None:
+                value["upstream_updated_at"] = upstream_updated_at
+            self.cache.put(key, value)
+            return value, True
+
+        value = self.cache.store(
+            key,
+            body=body,
+            url=url,
+            status=status,
+            etag=response_headers.get("ETag"),
+            last_modified=response_headers.get("Last-Modified"),
+            upstream_updated_at=upstream_updated_at,
+        )
+        return value, False
+
+    def fetch_binary(self, url: str, *, cache_key: Optional[str] = None,
+                     max_age_days: Optional[int] = None
+                     ) -> Tuple[Optional[bytes], bool]:
+        """Fetch a binary resource, such as a PDF.
+
+        Kept separate from `fetch` because that decodes the response as text,
+        which silently destroys anything that is not text -- a PDF fetched
+        through it comes back as replacement characters. Bodies are cached
+        base64-encoded so the cache stays plain JSON.
+        """
+        import base64
+
+        key = cache_key or f"binary:{url}"
+        cached = self.cache.get(key)
+        if cached is not None and self.cache.is_fresh(key, max_age_days=max_age_days):
+            payload = cached["value"].get("body_base64")
+            if payload is None:
+                return None, True
+            return base64.b64decode(payload), True
+        if OFFLINE:
+            if cached is not None:
+                payload = cached["value"].get("body_base64")
+                return (base64.b64decode(payload) if payload else None), True
+            raise OfflineError(f"offline and no cached copy of {url}")
+
+        delay = self.initial_backoff
+        last_error: Optional[Exception] = None
+        for attempt in range(self.max_retries):
+            self._throttle()
+            request = urllib.request.Request(url, headers=dict(self.headers))
+            try:
+                self.requests_made += 1
+                with urllib.request.urlopen(request, timeout=self.timeout) as response:
+                    raw = response.read()
+                self.cache.put(key, {
+                    "url": url,
+                    "status": 200,
+                    "body_base64": base64.b64encode(raw).decode("ascii"),
+                    "content_sha256": "sha256:" + __import__("hashlib")
+                                      .sha256(raw).hexdigest(),
+                    "fetched_at": now(),
+                })
+                return raw, False
+            except urllib.error.HTTPError as error:
+                if error.code in (404, 403):
+                    self.cache.put(key, {"url": url, "status": error.code,
+                                         "body_base64": None,
+                                         "fetched_at": now()})
+                    return None, False
+                last_error = error
+            except (urllib.error.URLError, TimeoutError, OSError) as error:
+                last_error = error
+            if attempt < self.max_retries - 1:
+                time.sleep(delay)
+                delay = min(delay * 2, self.max_backoff)
+        raise RuntimeError(f"failed to fetch {url}: {last_error}")
+
+    def fetch_json(self, url: str, **kwargs) -> Tuple[Any, bool]:
+        value, from_cache = self.fetch(url, **kwargs)
+        if value.get("status") not in (200, 304):
+            return None, from_cache
+        try:
+            return json.loads(value["body"]), from_cache
+        except ValueError:
+            return None, from_cache
+
+    def _request(self, url: str, headers: Dict[str, str]):
+        delay = self.initial_backoff
+        last_error: Optional[Exception] = None
+        for attempt in range(self.max_retries):
+            self._throttle()
+            request = urllib.request.Request(url, headers=headers)
+            try:
+                self.requests_made += 1
+                with urllib.request.urlopen(request,
+                                            timeout=self.timeout) as response:
+                    charset = response.headers.get_content_charset() or "utf-8"
+                    body = response.read().decode(charset, errors="replace")
+                    return body, response.status, dict(response.headers)
+            except urllib.error.HTTPError as error:
+                headers = dict(error.headers or {})
+                if error.code == 304:
+                    return "", 304, headers
+                if error.code == 404:
+                    return "", 404, headers
+                if error.code == 401:
+                    # Rejected credentials. Returned rather than raised so the
+                    # caller sees the status and can say what is wrong: raising
+                    # here reached a `except RuntimeError: return` that made an
+                    # expired token indistinguishable from an empty result.
+                    return "", 401, headers
+                if error.code in (429, 500, 502, 503, 504) or (
+                        error.code == 403 and _is_rate_limited(headers)):
+                    # Rate limiting and transient upstream failures: back off.
+                    retry_after = headers.get("Retry-After")
+                    wait = (float(retry_after)
+                            if retry_after and retry_after.isdigit() else delay)
+                    last_error = error
+                    if attempt < self.max_retries - 1:
+                        time.sleep(min(wait, self.max_backoff))
+                        delay = min(delay * 2, self.max_backoff)
+                        continue
+                if error.code == 403:
+                    # A plain 403 is a site that will not serve us, not a
+                    # request to slow down. Retrying it just burns the run's
+                    # budget, so it is reported like any other dead source.
+                    return "", 403, headers
+                last_error = error
+                break
+            except (urllib.error.URLError, TimeoutError, OSError) as error:
+                last_error = error
+                if attempt < self.max_retries - 1:
+                    time.sleep(delay)
+                    delay = min(delay * 2, self.max_backoff)
+                    continue
+                break
+        raise RuntimeError(f"failed to fetch {url}: {last_error}")
+
+
+def _is_rate_limited(headers: Dict[str, str]) -> bool:
+    """Whether a 403 is a throttle rather than a refusal.
+
+    GitHub answers a rate limit with 403 plus a retry hint or an exhausted
+    remaining count; a site that simply blocks the client sends neither.
+    """
+    if headers.get("Retry-After"):
+        return True
+    remaining = headers.get("X-RateLimit-Remaining") or headers.get(
+        "x-ratelimit-remaining")
+    return remaining is not None and remaining.strip() == "0"
+
+
+def with_params(url: str, params: Dict[str, Any]) -> str:
+    clean = {k: v for k, v in params.items() if v is not None}
+    return url + ("&" if "?" in url else "?") + urllib.parse.urlencode(clean)
diff --git a/tools/impact/intake.py b/tools/impact/intake.py
new file mode 100644
index 0000000..424938e
--- /dev/null
+++ b/tools/impact/intake.py
@@ -0,0 +1,511 @@
+"""Files downloaded papers into the drop folder, under the names it expects.
+
+A publisher hands you `3769828.pdf`, or a title with the spaces turned into
+underscores, and the pipeline wants `10.1145_3769828.pdf`. Renaming by hand is
+dull and easy to get wrong, and getting it wrong is quiet: the file simply is
+not found, and the paper stays unread.
+
+Order is not used to decide what a file is, even when the downloads were made
+in order. Order is a claim about the session; the first page of the PDF is a
+fact about the file. Each PDF is read and matched against the titles of the
+papers still wanted, and it is filed only when one of them clearly wins.
+"""
+
+from __future__ import annotations
+
+import difflib
+import pathlib
+import re
+import shutil
+from typing import Dict, List, Optional, Sequence, Set, Tuple
+
+from . import worklist
+
+# How much of a paper's title must appear on the PDF's first page before the
+# file is filed as that paper, and how far ahead of the runner-up it must be.
+MIN_SCORE = 0.75
+MIN_MARGIN = 0.15
+
+MANIFEST_SCHEMA_VERSION = "1.0.0"
+
+# Enough of the front matter to carry the title, without the bibliography.
+FIRST_PAGE_CHARS = 4000
+
+
+def _digest(path: pathlib.Path) -> str:
+    import hashlib
+
+    digest = hashlib.sha256()
+    with open(path, "rb") as handle:
+        for chunk in iter(lambda: handle.read(1 << 20), b""):
+            digest.update(chunk)
+    return f"sha256:{digest.hexdigest()}"
+
+
+# Publishers name a download after the paper's identifier: ACM uses the DOI
+# suffix ("3729175.pdf", "3597503.3639200.pdf"), arXiv the arXiv id.
+FILENAME_IDENTIFIER = re.compile(r"^(\d{4}\.\d{4,5}|\d{6,8}(?:\.\d{6,8})?)$")
+
+
+def doi_on_page(page_text: str, rows: Sequence[dict]) -> Optional[dict]:
+    """The paper whose DOI is printed on this page.
+
+    Exact rather than heuristic, and it works where titles fail: a journal
+    masthead can push the title down the page past anything a layout rule will
+    find, but the DOI is printed there too. Extraction scatters spaces through
+    it -- "10.22399/ijcesen. 5462" -- so the comparison is made with all
+    whitespace removed from both sides.
+    """
+    flat = re.sub(r"\s+", "", page_text or "").lower()
+    if not flat:
+        return None
+    best = None
+    for row in rows:
+        doi = row.get("doi")
+        if not doi:
+            continue
+        needle = re.sub(r"\s+", "", doi).lower()
+        if len(needle) >= 12 and needle in flat:
+            # The longest match wins: one DOI can be a prefix of another.
+            if best is None or len(needle) > len(re.sub(r"\s+", "", best["doi"])):
+                best = row
+    return best
+
+
+def identifier_from_filename(path: pathlib.Path,
+                             rows: Sequence[dict]) -> Optional[dict]:
+    """The paper a download's own filename names, if it names one.
+
+    The last resort, and sometimes the only one. A magazine extract can carry
+    neither a declared title nor its title on any page -- the AWS article
+    begins mid-sentence -- and then the filename the publisher chose is all
+    there is to go on. Matching is exact against the identifier already in the
+    expected filename, so this cannot guess.
+    """
+    stem = path.stem.strip()
+    if not FILENAME_IDENTIFIER.match(stem):
+        return None
+    for row in rows:
+        name = row["filename"]
+        if name.endswith(f"_{stem}.pdf") or name == f"{stem}.pdf":
+            return row
+    return None
+
+
+def metadata_title(path: pathlib.Path) -> str:
+    """The title the PDF declares about itself, if it declares one.
+
+    Free, and right more often than anything recovered from the page: a
+    conference template usually fills it in. It is only a candidate, scored
+    like the rest, because plenty of PDFs carry a LaTeX job name here instead.
+    """
+    try:
+        import pypdf
+
+        title = (pypdf.PdfReader(str(path)).metadata or {}).get("/Title")
+        title = str(title or "").strip()
+        return title if len(title) > 8 and title.lower() != "none" else ""
+    except Exception:
+        return ""
+
+
+def first_page(path: pathlib.Path) -> str:
+    try:
+        import pypdf
+
+        reader = pypdf.PdfReader(str(path))
+        if not reader.pages:
+            return ""
+        return (reader.pages[0].extract_text() or "")[:FIRST_PAGE_CHARS]
+    except Exception:
+        return ""
+
+
+# Where a reference list can start. A DOI printed after this point belongs to
+# a paper being cited, not to the paper in hand.
+REFERENCES_HEADING = re.compile(
+    r"^\s*(?:\d+\.?\s*)?(?:references|bibliography|referensi)\s*$",
+    re.IGNORECASE | re.MULTILINE)
+
+
+def front_pages(path: pathlib.Path, pages: int = 2) -> str:
+    """Text from the opening pages, cut off before any reference list.
+
+    A journal masthead does not always sit on page one: AJSE prints the DOI on
+    page two, which put one paper out of reach of an exact DOI match. Reading
+    a little further finds it. The cut at the reference list is what keeps that
+    safe -- a cited paper's DOI must never be mistaken for this paper's own.
+    """
+    try:
+        import pypdf
+
+        reader = pypdf.PdfReader(str(path))
+        text = "\n".join((page.extract_text() or "")
+                          for page in reader.pages[:pages])
+    except Exception:
+        return ""
+    heading = REFERENCES_HEADING.search(text)
+    if heading:
+        text = text[:heading.start()]
+    return text[:FIRST_PAGE_CHARS * pages]
+
+
+def _words(text: str) -> List[str]:
+    return [w for w in re.sub(r"[^a-z0-9 ]+", " ", (text or "").lower()).split()
+            if len(w) > 3]
+
+
+# Lines a publisher puts above the title on the first page.
+FRONT_MATTER = re.compile(
+    r"^\s*(?:\d{4}\s|20\d\d\s+(?:ieee|acm)|permission to make|this work is|"
+    r"acm isbn|doi:|https?://|arxiv:|proceedings of|\d+(?:st|nd|rd|th)\s|"
+    r"ieee|acm\b|©|copyright|open access to the|international journal|"
+    r"journal of|vol\.|issn|research article|received\s*:|revised\s*:)",
+    re.IGNORECASE)
+
+# USENIX prints a cover sheet whose last line runs straight into the title:
+# "sponsored by USENIX.Detecting Logical Bugs of DBMS with". The title starts
+# after the boilerplate, on the same line.
+USENIX_COVER = re.compile(
+    r"(?:is\s+)?sponsored\s+by\s*(?:USENIX)?\s*\.?\s*", re.IGNORECASE)
+
+# Where the title stops and the author block starts.
+AUTHOR_LINE = re.compile(
+    r"@|\b(?:University|Universität|Institute|Laborator|College|School\s+of|"
+    r"Academy|Inc\.|Ltd|GmbH|Corporation|Group|Research|Tech\.|"
+    r"Abstract|ABSTRACT|A BSTRACT)\b|^\s*\d+\s*$")
+
+
+def title_candidates(page_text: str) -> List[str]:
+    """Plausible renderings of the paper's title from its first page.
+
+    Several, not one, because the first page is laid out differently by every
+    publisher and a single rule gets some of them wrong. ACM wraps a long title
+    over two lines, the second of which can be a single short word -- "DBMS" --
+    that a minimum-length filter throws away, taking the title with it. USENIX
+    prints a cover sheet and runs the title onto the end of its last line.
+
+    The caller scores every candidate and keeps the best, so a rule that
+    misfires costs nothing as long as another one works.
+    """
+    lines = [line.strip() for line in (page_text or "").splitlines()]
+    lines = [line for line in lines if line]
+
+    # A cover sheet is looked for across the whole top of the page, because its
+    # own lines do not all look like front matter and the title is glued to the
+    # end of one of them.
+    start = 0
+    for index, line in enumerate(lines[:10]):
+        match = USENIX_COVER.search(line)
+        if match and match.end() < len(line):
+            lines = [line[match.end():].strip()] + lines[index + 1:]
+            break
+    else:
+        # No cover sheet. Front matter sits above the title, so the first line
+        # that is not front matter ends it -- scanning further would let an
+        # author line ("1st Panta Kittisatra, ...") move the start past the
+        # title itself.
+        for index, line in enumerate(lines[:10]):
+            if FRONT_MATTER.match(line):
+                start = index + 1
+                continue
+            break
+
+    collected: List[str] = []
+    for line in lines[start:start + 6]:
+        if AUTHOR_LINE.search(line):
+            break
+        collected.append(line)
+        if sum(len(part) for part in collected) > 130:
+            break
+
+    candidates = []
+    if collected:
+        candidates.append(" ".join(collected))
+        candidates.append(collected[0])
+        if len(collected) > 1:
+            candidates.append(" ".join(collected[:2]))
+    return [c for c in dict.fromkeys(candidates) if len(c) > 8]
+
+
+def pdf_title(page_text: str) -> str:
+    """The single best guess at the title, for reporting."""
+    candidates = title_candidates(page_text)
+    return candidates[0] if candidates else ""
+
+
+def score(title: str, page_text: str,
+          extra_titles: Sequence[str] = ()) -> float:
+    """How strongly this page is that paper, from 0 to 1.
+
+    Two views, and the weaker one governs. Containment asks how much of the
+    candidate title appears anywhere on the page, which is robust to a PDF
+    whose title extraction came out badly. Similarity compares the candidate
+    against the page's own title, which is what stops a generic title matching
+    everything. Taking the smaller of the two means a file must look right by
+    both readings.
+    """
+    words = _words(title)
+    if not words:
+        return 0.0
+    flat = " " + re.sub(r"\s+", " ",
+                        re.sub(r"[^a-z0-9 ]+", " ", (page_text or "").lower())) + " "
+    containment = sum(1 for word in words if word in flat) / len(words)
+
+    candidates = title_candidates(page_text) + [t for t in extra_titles if t]
+    if not candidates:
+        return 0.0
+    # A declared title needs no containment check: the page it belongs to may
+    # not even carry the title, as with a magazine extract.
+    if any(difflib.SequenceMatcher(None, " ".join(_words(title)),
+                                   " ".join(_words(extra))).ratio() >= 0.90
+           for extra in extra_titles if extra):
+        return 1.0
+    target = " ".join(_words(title))
+    similarity = max(
+        difflib.SequenceMatcher(None, target, " ".join(_words(c))).ratio()
+        for c in candidates)
+    return min(containment, similarity)
+
+
+def identify(page_text: str, rows: List[dict], *,
+             wanted: Optional[set] = None, extra_titles: Sequence[str] = ()
+             ) -> Tuple[Optional[dict], float, float]:
+    """The paper this page is, its score, and the runner-up's.
+
+    ``rows`` must be every paper worth considering, not only the ones still
+    wanted. A PDF whose own paper has already been filed would otherwise be
+    matched against the remaining candidates and filed as the closest of them,
+    which is how a correct download becomes a wrong quotation. The caller checks
+    afterwards whether the winner is still wanted.
+
+    The runner-up matters too: two papers by the same group on the same
+    technique share most of their title words.
+    """
+    if not page_text.strip() and not any(extra_titles):
+        return None, 0.0, 0.0
+    ranked = sorted(((score(row["title"], page_text, extra_titles), row)
+                     for row in rows),
+                    key=lambda pair: pair[0], reverse=True)
+    if not ranked:
+        return None, 0.0, 0.0
+    best_score, best = ranked[0]
+    runner_up = ranked[1][0] if len(ranked) > 1 else 0.0
+    if best_score < MIN_SCORE:
+        return None, best_score, runner_up
+
+    if best_score - runner_up < MIN_MARGIN:
+        # A near-tie is usually a paper and its own preprint, which share a
+        # title. If exactly one of the tied candidates is still wanted, that is
+        # the one the file is for: the others are already filed, or the
+        # pipeline can read them without help. Only a tie between two papers we
+        # genuinely need is left for a person.
+        if wanted is None:
+            return None, best_score, runner_up
+        tied = [row for value, row in ranked
+                if best_score - value < MIN_MARGIN]
+        needed = [row for row in tied if row["filename"] in wanted]
+        if not needed:
+            # None of the tied candidates needs a file. Usually this is a paper
+            # already filed under one of its two identities, so returning the
+            # best match lets the caller report it as filed rather than as a
+            # decision waiting on a person.
+            return best, best_score, runner_up
+        if len(needed) != 1:
+            return None, best_score, runner_up
+        return needed[0], best_score, runner_up
+
+    return best, best_score, runner_up
+
+
+def is_ambiguous(best: float, runner_up: float) -> bool:
+    """Whether a file was refused for being between two papers, not unknown.
+
+    The usual cause is a paper and its own preprint, whose titles differ by a
+    word. Worth saying out loud: the file is almost certainly wanted, and only a
+    person can say which of the two it is.
+    """
+    return best >= MIN_SCORE and best - runner_up < MIN_MARGIN
+
+
+def run(source: pathlib.Path, *, dry_run: bool = False,
+        limit: Optional[int] = None,
+        destination: Optional[pathlib.Path] = None,
+        candidates: Optional[List[dict]] = None,
+        wanted: Optional[Set[str]] = None) -> Dict[str, int]:
+    """File every PDF in ``source`` that is a paper the worklist still wants.
+
+    The dataset supplies the candidates and the wanted set by default; passing
+    them lets a test exercise the matching without a real papers file.
+    """
+    from . import config
+    from .collectors import fulltext
+
+    destination = destination or worklist.OUTPUT.parent
+    destination.mkdir(parents=True, exist_ok=True)
+
+    if candidates is None:
+        # Every paper is a candidate, so a PDF is recognised as itself even
+        # when that paper is already filed; only the wanted set is written to.
+        candidates = []
+        for paper in config.load_json(config.DATA_FILES["papers"])["papers"]:
+            if paper.get("is_sqlancer_publication"):
+                continue
+            filename = fulltext.wanted_filename(paper.get("doi"),
+                                                paper.get("arxiv_id"),
+                                                paper.get("s2_paper_id"))
+            if filename:
+                candidates.append({"title": paper["title"],
+                                   "filename": filename,
+                                   "doi": paper.get("doi")})
+    if wanted is None:
+        wanted = {row["filename"] for row in worklist.wanted()}
+
+    pdfs = sorted((p for p in source.glob("*.pdf")),
+                  key=lambda p: p.stat().st_mtime, reverse=True)
+    if limit:
+        pdfs = pdfs[:limit]
+
+    counts = {"filed": 0, "already_there": 0, "not_wanted": 0, "ambiguous": 0,
+              "unmatched": 0}
+    filed: List[dict] = []
+    unresolved: List[dict] = []
+    for pdf in pdfs:
+        page_text = first_page(pdf)
+        declared = metadata_title(pdf)
+        match, best, runner_up = identify(page_text, candidates, wanted=wanted,
+                                          extra_titles=(declared,))
+        if match is None and not is_ambiguous(best, runner_up):
+            named = (doi_on_page(page_text, candidates)
+                     or doi_on_page(front_pages(pdf), candidates)
+                     or identifier_from_filename(pdf, candidates))
+            if named is not None:
+                match, best, runner_up = named, 1.0, 0.0
+        if match is None:
+            if is_ambiguous(best, runner_up):
+                counts["ambiguous"] += 1
+                near = sorted(((score(row["title"], page_text, (declared,)), row)
+                               for row in candidates), reverse=True,
+                              key=lambda pair: pair[0])[:2]
+                print(f"  {pdf.name[:46]:46s} ?? between "
+                      + " and ".join(f"{row['title'][:34]!r}" for _, row in near))
+                unresolved.append({
+                    "source": str(pdf),
+                    "reason": "two papers match almost equally",
+                    "candidates": [{"title": row["title"],
+                                    "filename": row["filename"],
+                                    "match_score": round(value, 3)}
+                                   for value, row in near],
+                })
+            else:
+                counts["unmatched"] += 1
+                # Naming it matters: a PDF that is silently dropped looks to
+                # the person who downloaded it exactly like one that was
+                # filed, and they download it again.
+                closest = max(((score(row["title"], page_text, (declared,)), row)
+                               for row in candidates), default=(0.0, None),
+                              key=lambda pair: pair[0])
+                title = closest[1]["title"][:40] if closest[1] else "nothing"
+                print(f"  {pdf.name[:46]:46s} -- no match "
+                      f"(closest {title!r} at {closest[0]:.0%})")
+                unresolved.append({
+                    "source": str(pdf),
+                    "reason": "no paper in the dataset matches this PDF",
+                    "candidates": ([{"title": closest[1]["title"],
+                                     "filename": closest[1]["filename"],
+                                     "match_score": round(closest[0], 3)}]
+                                   if closest[1] else []),
+                })
+            continue
+        target = destination / match["filename"]
+        if target.exists():
+            counts["already_there"] += 1
+            print(f"  already filed  {match['filename']}")
+            continue
+        if match["filename"] not in wanted:
+            # Recognised, but the pipeline can already read this paper from
+            # arXiv or PVLDB, so a copy here would add nothing.
+            counts["not_wanted"] += 1
+            print(f"  {pdf.name[:46]:46s} -- {match['title'][:40]} "
+                  f"is already reachable without a PDF")
+            continue
+        if not dry_run:
+            shutil.copy2(pdf, target)
+        counts["filed"] += 1
+        filed.append({
+            "source": str(pdf),
+            "filed_as": match["filename"],
+            "title": match["title"],
+            "match_score": round(best, 3),
+            "runner_up_score": round(runner_up, 3),
+            "bytes": pdf.stat().st_size,
+            # So a manifest can be checked against the folder later: these
+            # files are not in git, and this is the only record that a given
+            # PDF is the paper it was filed as.
+            "sha256": _digest(pdf),
+        })
+        print(f"  {pdf.name[:46]:46s} -> {match['filename']}  "
+              f"({best:.0%}, next {runner_up:.0%})")
+        # The paper stays a candidate. One file per paper is already enforced
+        # by the destination check above, and leaving the row in place means a
+        # second copy of the same download is reported as already filed rather
+        # than as an unrecognised PDF -- which would read as a reason to
+        # download it again.
+    counts["_filed"] = filed
+    counts["_unresolved"] = unresolved
+    return counts
+
+
+def main(argv: Optional[List[str]] = None) -> int:
+    import argparse
+
+    parser = argparse.ArgumentParser(
+        description="File downloaded paper PDFs into the drop folder.")
+    parser.add_argument("--source", type=pathlib.Path,
+                        default=pathlib.Path.home() / "Downloads",
+                        help="where the downloads are (default: ~/Downloads)")
+    parser.add_argument("--limit", type=int, default=40,
+                        help="only look at this many of the newest PDFs")
+    parser.add_argument("--dry-run", action="store_true",
+                        help="say what would be filed, copy nothing")
+    parser.add_argument("--json", type=pathlib.Path,
+                        help="write a manifest of what was filed to this path")
+    args = parser.parse_args(argv)
+
+    if not args.source.is_dir():
+        print(f"no such directory: {args.source}")
+        return 1
+    counts = run(args.source, dry_run=args.dry_run, limit=args.limit)
+    print(f"\n{counts['filed']} filed, {counts['already_there']} already there, "
+          f"{counts['not_wanted']} reachable without a PDF, "
+          f"{counts['ambiguous']} too close to call, "
+          f"{counts['unmatched']} not a paper in the dataset"
+          + (" (dry run, nothing copied)" if args.dry_run else ""))
+
+    if args.json:
+        import json as json_module
+        from .util import content_hash, now
+
+        manifest = {
+            "schema_version": MANIFEST_SCHEMA_VERSION,
+            "filed_at": now(),
+            "source_directory": str(args.source),
+            "destination_directory": str(worklist.OUTPUT.parent),
+            "dry_run": bool(args.dry_run),
+            "counts": {key: value for key, value in counts.items()
+                       if not key.startswith("_")},
+            "filed": counts["_filed"],
+            "unresolved": counts["_unresolved"],
+        }
+        args.json.parent.mkdir(parents=True, exist_ok=True)
+        args.json.write_text(
+            json_module.dumps(manifest, indent=2, ensure_ascii=False) + "\n",
+            encoding="utf-8")
+        print(f"manifest written to {args.json}")
+    if counts["filed"]:
+        print("Now run: python3 -m tools.impact.run collect --only papers --full")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/notes.py b/tools/impact/notes.py
new file mode 100644
index 0000000..3c9fb89
--- /dev/null
+++ b/tools/impact/notes.py
@@ -0,0 +1,369 @@
+"""One file per citing paper: what it is about, and every source behind it.
+
+``papers.json`` answers the questions the statistics need -- does this paper
+reuse the codebase, extend a technique, compare against SQLancer -- and nothing
+more. That is the right shape for a dataset and the wrong shape for a reader,
+who wants to know what the paper actually is before deciding whether the
+classification looks right.
+
+So each paper also gets a note of its own, holding two clearly separated
+things. A summary, written by a language model from the best text available for
+that paper, marked as model-generated and stamped with the model and the exact
+source it was given. And underneath it, every piece of evidence the record
+rests on, quoted verbatim with its URL -- the citation sentences, the full-text
+passages, the artifact markers.
+
+The separation is the point. The summary is a convenience and is allowed to be
+wrong; the evidence is the record and is not. A summary the model could not
+ground in a quotation from the source is stored as ungrounded rather than
+presented as a finding, and a paper with no API key available gets its note with
+the evidence and no summary at all.
+"""
+
+from __future__ import annotations
+
+import json
+
+import pathlib
+from typing import Dict, List, Optional, Tuple
+
+from . import config, taxonomy
+from .cache import Caches
+from .classify import Classifier
+from .util import content_hash, now, slugify, truncate
+
+SCHEMA_VERSION = "1.0.0"
+NOTES_DIR = config.DATA_DIR / "paper_notes"
+
+# How much of a paper to hand the model. Enough for a long paper's introduction,
+# related work and evaluation, which is where a relationship is stated.
+MAX_SOURCE_CHARS = 60_000
+
+# Shorter than this, an abstract is a stub rather than a description.
+MIN_ABSTRACT_CHARS = 120
+
+RELATIONSHIP_TITLES = {
+    "references": "Cites SQLancer",
+    "uses_infrastructure": "Uses the SQLancer codebase",
+    "extends_technique": "Extends a SQLancer technique",
+    "compares_with": "Compares against SQLancer",
+    "describes_as_state_of_the_art": "Calls SQLancer state of the art",
+}
+
+
+def note_path(paper: dict) -> pathlib.Path:
+    """Where a paper's note lives.
+
+    Named from the paper id rather than its title: a title changes when a
+    preprint becomes a publication, and the file should not move when it does.
+    """
+    return NOTES_DIR / f"{slugify(paper['id'])}.json"
+
+
+def _best_source(paper: dict, fetcher=None) -> Tuple[Optional[str], str, str, str]:
+    """The best text available for this paper.
+
+    Returns ``(text, kind, label, source_url)``. Preference runs full text,
+    then abstract, then the citation sentences -- and the kind travels with the
+    summary, because a summary written from two citation sentences is a much
+    weaker thing than one written from the paper, and a reader should be told
+    which they are looking at.
+    """
+    from .collectors import fulltext
+
+    # A supplied PDF is read straight off the disk -- pdf_text handles a
+    # file:// URL without touching the network -- so it must not be skipped
+    # merely because the caller had no fetcher to give. Getting that wrong is
+    # quiet: the summary is written from the abstract while the paper itself
+    # sits unread in the folder.
+    local = fulltext.local_pdf(paper.get("doi"), paper.get("arxiv_id"),
+                               paper.get("s2_paper_id"))
+    if local:
+        text = fulltext.pdf_text(fetcher, local)
+        if text:
+            return (text, "the full text", "supplied PDF",
+                    paper.get("url") or local)
+
+    if paper.get("arxiv_id") and fetcher is not None:
+        url = f"https://arxiv.org/pdf/{paper['arxiv_id']}"
+        text = fulltext.pdf_text(fetcher, url)
+        if text:
+            return text, "the full text", "arXiv PDF", url
+
+    abstract = (paper.get("abstract") or "").strip()
+    # Some indexed "abstracts" are a single stub line -- a subtitle, or the
+    # first sentence of a chapter. Below this length the citation sentences say
+    # more about the paper than the abstract does.
+    if len(abstract) >= MIN_ABSTRACT_CHARS:
+        return (abstract, "only the abstract", "abstract",
+                paper.get("url") or "")
+
+    contexts = []
+    for key, entry in paper.get("relationships", {}).items():
+        for item in entry.get("evidence", []):
+            excerpt = item.get("excerpt")
+            if excerpt and excerpt not in contexts:
+                contexts.append(excerpt)
+    if contexts:
+        return ("\n\n".join(contexts),
+                "only the sentences in which this paper cites SQLancer",
+                "citation contexts", paper.get("url") or "")
+    return None, "", "", ""
+
+
+def _evidence_sections(paper: dict) -> List[dict]:
+    """Every source behind this paper's record, grouped by what it establishes."""
+    sections = []
+    for key, entry in paper.get("relationships", {}).items():
+        evidence = entry.get("evidence") or []
+        if not evidence:
+            continue
+        sections.append({
+            "relationship": key,
+            "title": RELATIONSHIP_TITLES.get(key, key.replace("_", " ")),
+            "value": entry.get("value"),
+            "method": entry.get("method"),
+            "techniques": entry.get("techniques", []),
+            "sources": [
+                {
+                    "source_url": item.get("source_url"),
+                    "source_type": item.get("source_type"),
+                    "excerpt": item.get("excerpt"),
+                    "excerpt_is_verbatim": item.get("excerpt_is_verbatim", False),
+                    "note": item.get("note"),
+                    "retrieved_at": item.get("retrieved_at"),
+                }
+                for item in evidence
+            ],
+        })
+
+    for artifact in paper.get("artifacts", []) or []:
+        sources = []
+        for item in (artifact.get("marker_evidence") or []):
+            sources.append({
+                "source_url": item.get("source_url"),
+                "source_type": item.get("source_type"),
+                "excerpt": item.get("excerpt"),
+                "excerpt_is_verbatim": item.get("excerpt_is_verbatim", False),
+                "note": item.get("note"),
+                "retrieved_at": item.get("retrieved_at"),
+            })
+        sections.append({
+            "relationship": "artifact",
+            "title": "Artifact",
+            "value": artifact.get("url"),
+            "method": "artifact_inspection",
+            "techniques": [],
+            "markers": artifact.get("sqlancer_markers", []),
+            "sources": sources,
+        })
+    return sections
+
+
+def build_note(paper: dict, classifier: Optional[Classifier] = None,
+               fetcher=None, timestamp: Optional[str] = None) -> dict:
+    """The note for one paper: a summary if one can be had, and the evidence."""
+    timestamp = timestamp or now()
+    text, kind, label, source_url = _best_source(paper, fetcher)
+
+    summary: Optional[dict] = None
+    if text and classifier is not None:
+        result = classifier.classify(
+            "paper_summary",
+            source_id=f"paper-summary:{paper['id']}",
+            source_text=truncate(text, MAX_SOURCE_CHARS),
+            source_label=label,
+            title=paper["title"],
+            source_kind=kind)
+        if result is not None:
+            summary = {
+                "text": result.extra.get("summary") or "",
+                "relationship_to_sqlancer":
+                    result.extra.get("relationship_to_sqlancer") or "",
+                # Said plainly, because everything else in this dataset is not.
+                "is_model_generated": True,
+                "grounded_in_source": bool(result.excerpts),
+                "written_from": label,
+                "written_from_url": source_url,
+                "source_sha256": content_hash(text),
+                "supporting_excerpts": result.excerpts,
+                "model": result.model,
+                "classifier_version": result.classifier_version,
+                "generated_at": timestamp,
+            }
+
+    return {
+        "schema_version": SCHEMA_VERSION,
+        "policy_version": config.policy_version(),
+        "paper": {
+            "id": paper["id"],
+            "title": paper["title"],
+            "authors": paper.get("authors", []),
+            "year": paper.get("year"),
+            "venue": paper.get("venue"),
+            "doi": paper.get("doi"),
+            "url": paper.get("url"),
+        },
+        "summary": summary,
+        "evidence": _evidence_sections(paper),
+    }
+
+
+class UngroundedSummary(ValueError):
+    """A supplied excerpt was not found in the text it claims to quote."""
+
+
+def apply_summary(paper: dict, *, text: str, relationship_to_sqlancer: str,
+                  excerpts: List[str], written_from: str,
+                  written_from_url: str = "", source_text: str,
+                  model: str, generated_by: str = "claude_code_session",
+                  timestamp: Optional[str] = None) -> dict:
+    """Attach a summary written outside the batch classifier.
+
+    The pipeline's own summaries come from ``build_note`` calling the API. This
+    is the same thing written interactively, and it is held to the same rule:
+    every excerpt has to appear, character for character, in the source text the
+    summary claims to be written from. An excerpt that does not is refused
+    rather than stored, because an unverifiable quotation is worse than none.
+
+    ``generated_by`` records which route produced it, so a summary written this
+    way is never mistaken for one the pipeline can reproduce on its own.
+    """
+    from .util import verbatim_excerpt
+
+    timestamp = timestamp or now()
+    verified: List[str] = []
+    for excerpt in excerpts:
+        excerpt = (excerpt or "").strip()
+        if not excerpt:
+            continue
+        if not verbatim_excerpt(source_text, excerpt):
+            raise UngroundedSummary(
+                f"{paper['id']}: excerpt is not in the source text: {excerpt[:80]!r}")
+        if excerpt not in verified:
+            verified.append(excerpt)
+
+    path = note_path(paper)
+    if path.exists():
+        note = json.loads(path.read_text(encoding="utf-8"))
+    else:
+        note = build_note(paper, classifier=None, fetcher=None,
+                          timestamp=timestamp)
+
+    note["summary"] = {
+        "text": text.strip(),
+        "relationship_to_sqlancer": relationship_to_sqlancer.strip(),
+        "is_model_generated": True,
+        "grounded_in_source": bool(verified),
+        "written_from": written_from,
+        "written_from_url": written_from_url,
+        "source_sha256": content_hash(source_text),
+        "supporting_excerpts": verified,
+        "model": model,
+        "generated_by": generated_by,
+        "classifier_version": None,
+        "generated_at": timestamp,
+    }
+    NOTES_DIR.mkdir(parents=True, exist_ok=True)
+    path.write_text(json.dumps(note, indent=2, ensure_ascii=False) + "\n",
+                    encoding="utf-8")
+    return note
+
+
+def summary_source(paper: dict) -> Tuple[Optional[str], str, str]:
+    """The text available to summarise this paper from, without fetching.
+
+    Returns ``(text, label, url)``. Deliberately offline: the abstract if the
+    record has one, otherwise the sentences in which the paper cites SQLancer.
+    """
+    text, _, label, url = _best_source(paper, fetcher=None)
+    return text, label, url
+
+
+def _unchanged(path, note: dict) -> bool:
+    """Whether writing this note would change anything but the timestamp."""
+    if not path.exists():
+        return False
+    try:
+        existing = json.loads(path.read_text(encoding="utf-8"))
+    except ValueError:
+        return False
+    return _comparable(existing) == _comparable(note)
+
+
+def _comparable(note: dict) -> str:
+    copy = json.loads(json.dumps(note))
+    if copy.get("summary"):
+        copy["summary"].pop("generated_at", None)
+    return json.dumps(copy, sort_keys=True)
+
+
+def write_all(*, limit: Optional[int] = None, dry_run: bool = False,
+              papers: Optional[List[dict]] = None) -> Dict[str, int]:
+    """Write a note for every citing paper. Returns a count of what happened."""
+    from .http import Fetcher
+
+    if papers is None:
+        papers = config.load_json(config.DATA_FILES["papers"])["papers"]
+    caches = Caches()
+    classifier = Classifier(caches.classifications, dry_run=dry_run)
+    fetcher = Fetcher(caches.artifacts, min_interval=0.5, max_retries=2,
+                      timeout=40.0)
+    NOTES_DIR.mkdir(parents=True, exist_ok=True)
+    timestamp = now()
+
+    counts = {"written": 0, "unchanged": 0, "summarised": 0, "no_summary": 0,
+              "removed": 0}
+    kept = set()
+    written = 0
+    for paper in papers:
+        if paper.get("is_sqlancer_publication"):
+            continue
+        path = note_path(paper)
+        kept.add(path.name)
+        if limit is not None and written >= limit and path.exists():
+            continue
+        note = build_note(paper, classifier=classifier, fetcher=fetcher,
+                          timestamp=timestamp)
+        # Never drop a summary that already exists because this run could not
+        # produce one -- no API key, an API failure, or a source that has since
+        # become unreachable. Regenerating the evidence must not cost the note
+        # its summary.
+        if not note.get("summary") and path.exists():
+            try:
+                existing = json.loads(path.read_text(encoding="utf-8"))
+            except ValueError:
+                existing = {}
+            if existing.get("summary"):
+                note["summary"] = existing["summary"]
+        if note.get("summary"):
+            counts["summarised"] += 1
+        else:
+            counts["no_summary"] += 1
+        if _unchanged(path, note):
+            counts["unchanged"] += 1
+            continue
+        path.write_text(json.dumps(note, indent=2, ensure_ascii=False) + "\n",
+                        encoding="utf-8")
+        counts["written"] += 1
+        written += 1
+
+    # A paper that leaves the dataset leaves no orphan note behind.
+    for stale in NOTES_DIR.glob("*.json"):
+        if stale.name not in kept:
+            stale.unlink()
+            counts["removed"] += 1
+    return counts
+
+
+def main() -> int:
+    counts = write_all()
+    print(f"paper notes: {counts['written']} written, "
+          f"{counts['unchanged']} unchanged, {counts['removed']} removed; "
+          f"{counts['summarised']} with a summary, "
+          f"{counts['no_summary']} without")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/pages.py b/tools/impact/pages.py
new file mode 100644
index 0000000..e1bb0d3
--- /dev/null
+++ b/tools/impact/pages.py
@@ -0,0 +1,256 @@
+"""Generates one page per database system under ``/impact/dbms/``.
+
+The impact page shows a bar per system and a row per project. Both raise the
+same question -- *which* bugs, and what is the evidence that this project uses
+SQLancer -- and neither has room to answer it. A page per system does, and it
+gives the bar and the project name somewhere to link to.
+
+The generated files are deliberately thin: front matter and one include. All
+rendering lives in ``_includes/impact/dbms-detail.html`` and all numbers come
+from ``stats.json``, so a change to how a system is presented is one edit rather
+than sixty, and the generated files stay reviewable in a diff.
+"""
+
+from __future__ import annotations
+
+import pathlib
+from typing import Dict, List, Optional
+
+from . import config
+
+PAGE_DIR = config.REPO_ROOT / "_pages" / "impact" / "dbms"
+PERMALINK_BASE = "/impact/dbms/"
+
+YEAR_PAGE_DIR = config.REPO_ROOT / "_pages" / "impact" / "bugs"
+YEAR_PERMALINK_BASE = "/impact/bugs/"
+
+TALK_PAGE_DIR = config.REPO_ROOT / "_pages" / "impact" / "talks"
+TALK_PERMALINK_BASE = "/impact/talks/"
+
+YEAR_TEMPLATE = """---
+permalink: {permalink}
+title: "Bugs reported in {year}"
+excerpt: "{excerpt}"
+bug_year: {year}
+author_profile: false
+sitemap: true
+---
+
+{{% include impact/year-detail.html year=page.bug_year %}}
+"""
+
+TALK_TEMPLATE = """---
+permalink: {permalink}
+title: "{title}"
+excerpt: "{excerpt}"
+talk_id: "{talk_id}"
+author_profile: false
+sitemap: true
+---
+
+{{% include impact/talk-detail.html id=page.talk_id %}}
+"""
+
+TEMPLATE = """---
+permalink: {permalink}
+title: "{title}"
+excerpt: "{excerpt}"
+dbms: {dbms_id}
+author_profile: false
+sitemap: true
+---
+
+{{% include impact/dbms-detail.html id=page.dbms %}}
+"""
+
+
+def _escape(text: str) -> str:
+    """Quote a value for a double-quoted YAML scalar."""
+    return str(text).replace("\\", "\\\\").replace('"', '\\"')
+
+
+def wanted(stats: dict) -> List[dict]:
+    """Systems that get a page: anything with a bug, adoption, or support.
+
+    A system with nothing recorded against it would be an empty page, and an
+    empty page is worse than no link.
+    """
+    return [row for row in stats["dbms"]["rows"]
+            if row.get("bugs") or row.get("supported")
+            or row.get("adoption_relationships")
+            or row.get("planned_adoption")]
+
+
+def excerpt_for(row: dict) -> str:
+    """One sentence saying what this system's page holds."""
+    parts = []
+    if row.get("bugs"):
+        parts.append(f"{row['bugs']} bugs SQLancer found in "
+                     f"{row['name']}, each with its evidence")
+    if row.get("adoption_relationships"):
+        parts.append("how the project uses SQLancer")
+    elif row.get("planned_adoption"):
+        parts.append("the project's proposal to adopt SQLancer")
+    if not parts:
+        parts.append(f"what SQLancer records about {row['name']}")
+    # Not ``capitalize``: it lowercases everything after the first letter,
+    # which turns DuckDB into duckdb.
+    sentence = "; ".join(parts)
+    return sentence[:1].upper() + sentence[1:] + "."
+
+
+def render(row: dict) -> str:
+    return TEMPLATE.format(
+        permalink=f"{PERMALINK_BASE}{row['id']}/",
+        title=_escape(f"{row['name']} and SQLancer"),
+        excerpt=_escape(excerpt_for(row)),
+        dbms_id=row["id"])
+
+
+def write_all(stats: Optional[dict] = None,
+              directory: Optional[pathlib.Path] = None) -> Dict[str, str]:
+    """Write a page per system; returns what happened to each, by id.
+
+    Pages for systems that are no longer in the dataset are removed, so a
+    system dropped from the records does not leave a page behind claiming
+    numbers nothing produces any more.
+    """
+    if stats is None:
+        stats = config.load_json(config.DATA_FILES["stats"])
+    directory = directory or PAGE_DIR
+    directory.mkdir(parents=True, exist_ok=True)
+
+    outcome: Dict[str, str] = {}
+    keep = set()
+    for row in wanted(stats):
+        path = directory / f"{row['id']}.html"
+        keep.add(path.name)
+        markup = render(row)
+        existing = path.read_text(encoding="utf-8") if path.exists() else None
+        if existing == markup:
+            outcome[row["id"]] = "unchanged"
+            continue
+        path.write_text(markup, encoding="utf-8")
+        outcome[row["id"]] = "written" if existing is None else "updated"
+    for path in directory.glob("*.html"):
+        if path.name not in keep:
+            path.unlink()
+            outcome[path.stem] = "removed"
+    return outcome
+
+
+def year_excerpt(row: dict) -> str:
+    """One sentence saying what a year's page holds."""
+    systems = row.get("systems") or 0
+    return (f"{row['bugs']} bugs attributed to SQLancer and reported in "
+            f"{row['year']}, across {systems} database "
+            f"system{'' if systems == 1 else 's'}, each with its evidence.")
+
+
+def render_year(row: dict) -> str:
+    return YEAR_TEMPLATE.format(
+        permalink=f"{YEAR_PERMALINK_BASE}{row['year']}/",
+        year=row["year"],
+        excerpt=_escape(year_excerpt(row)))
+
+
+def write_years(stats: Optional[dict] = None,
+                directory: Optional[pathlib.Path] = None) -> Dict[str, str]:
+    """Write a page per year that has bugs; returns what happened to each."""
+    if stats is None:
+        stats = config.load_json(config.DATA_FILES["stats"])
+    directory = directory or YEAR_PAGE_DIR
+    directory.mkdir(parents=True, exist_ok=True)
+
+    outcome: Dict[str, str] = {}
+    keep = set()
+    for row in stats["bugs"].get("years") or []:
+        if not row.get("bugs"):
+            continue
+        path = directory / f"{row['year']}.html"
+        keep.add(path.name)
+        markup = render_year(row)
+        existing = path.read_text(encoding="utf-8") if path.exists() else None
+        if existing == markup:
+            outcome[str(row["year"])] = "unchanged"
+            continue
+        path.write_text(markup, encoding="utf-8")
+        outcome[str(row["year"])] = "written" if existing is None else "updated"
+    for path in directory.glob("*.html"):
+        if path.name not in keep:
+            path.unlink()
+            outcome[path.stem] = "removed"
+    return outcome
+
+
+def talk_slug(talk: dict) -> str:
+    """The page's name: the video id where there is one, else the record's."""
+    return talk.get("video_id") or talk["id"].split(":")[-1]
+
+
+def talk_excerpt(talk: dict) -> str:
+    """One sentence saying what this talk's page holds."""
+    count = len(talk.get("mentions") or [])
+    speakers = ", ".join(talk.get("speakers") or []) or "a speaker"
+    where = talk.get("event") or talk.get("publisher") or "a talk"
+    return (f"{speakers} at {where}: the {count} "
+            f"point{'' if count == 1 else 's'} where SQLancer comes up, each "
+            "linked to the moment it was said.")
+
+
+def render_talk(talk: dict) -> str:
+    return TALK_TEMPLATE.format(
+        permalink=f"{TALK_PERMALINK_BASE}{talk_slug(talk)}/",
+        title=_escape(talk["title"]),
+        excerpt=_escape(talk_excerpt(talk)),
+        talk_id=_escape(talk["id"]))
+
+
+def write_talks(talks: Optional[dict] = None,
+                directory: Optional[pathlib.Path] = None) -> Dict[str, str]:
+    """Write a page per talk; returns what happened to each, by slug."""
+    if talks is None:
+        from . import talks as talks_store
+        talks = talks_store.load()
+    directory = directory or TALK_PAGE_DIR
+    directory.mkdir(parents=True, exist_ok=True)
+
+    outcome: Dict[str, str] = {}
+    keep = set()
+    for talk in talks.get("talks") or []:
+        slug = talk_slug(talk)
+        path = directory / f"{slug}.html"
+        keep.add(path.name)
+        markup = render_talk(talk)
+        existing = path.read_text(encoding="utf-8") if path.exists() else None
+        if existing == markup:
+            outcome[slug] = "unchanged"
+            continue
+        path.write_text(markup, encoding="utf-8")
+        outcome[slug] = "written" if existing is None else "updated"
+    for path in directory.glob("*.html"):
+        if path.name not in keep:
+            path.unlink()
+            outcome[path.stem] = "removed"
+    return outcome
+
+
+def _summarise(outcome: Dict[str, str]) -> str:
+    counts: Dict[str, int] = {}
+    for state in outcome.values():
+        counts[state] = counts.get(state, 0) + 1
+    return ", ".join(f"{count} {state}" for state, count in sorted(counts.items()))
+
+
+def main() -> int:
+    outcome = write_all()
+    print(f"pages: {len(outcome)} database systems ({_summarise(outcome)})")
+    years = write_years()
+    print(f"pages: {len(years)} years ({_summarise(years)})")
+    talks = write_talks()
+    print(f"pages: {len(talks)} talks ({_summarise(talks)})")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/plots.py b/tools/impact/plots.py
new file mode 100644
index 0000000..126a011
--- /dev/null
+++ b/tools/impact/plots.py
@@ -0,0 +1,423 @@
+"""Generates the inline SVG charts shown on the impact page.
+
+Written by hand rather than through a plotting library for three reasons: the
+output is a small, readable, diffable SVG that a reviewer can check against the
+records; it needs no build-time dependency on GitHub Pages; and being inline
+rather than an ``<img>`` it inherits the page's CSS, so the charts follow the
+site's light and dark palettes instead of baking one in.
+
+Colours come from the validated palette in the impact stylesheet and are
+referenced as CSS custom properties. Every chart also has a table beside it on
+the page, which is the required relief for the palette's two light-surface
+colours that sit below 3:1 contrast.
+"""
+
+from __future__ import annotations
+
+import html
+from typing import Dict, List, Optional, Sequence
+
+from . import config, pages
+from .util import truncate
+
+# Chart geometry. One place, so every chart on the page shares proportions.
+BAR_HEIGHT = 22
+BAR_GAP = 8
+LABEL_WIDTH = 150
+VALUE_WIDTH = 52
+CHART_PAD = 12
+COLUMN_MIN_WIDTH = 26
+COLUMN_GAP = 10
+GROUP_GAP = 16
+PLOT_HEIGHT = 210
+
+# Categorical slots, matching the CSS custom properties in impact.css.
+SERIES_VARS = [f"--impact-series-{index}" for index in range(1, 9)]
+
+
+def _escape(text: str) -> str:
+    return html.escape(str(text), quote=True)
+
+
+def _nice_max(value: int) -> int:
+    """Round an axis maximum up to a readable step."""
+    if value <= 0:
+        return 1
+    for step in (1, 2, 5, 10, 20, 25, 50, 100, 200, 250, 500, 1000, 2000):
+        if value <= step * 4:
+            return -(-value // step) * step
+    return -(-value // 5000) * 5000
+
+
+def _svg_open(width: int, height: int, title: str, description: str) -> List[str]:
+    return [
+        # max-width pins the chart to its natural size so a chart with few
+        # categories is not stretched across the full column.
+        f'<svg class="impact-chart" viewBox="0 0 {width} {height}" '
+        f'width="100%" height="auto" style="max-width:{width}px" role="img" '
+        f'preserveAspectRatio="xMinYMin meet" '
+        f'aria-label="{_escape(title)}" xmlns="http://www.w3.org/2000/svg">',
+        f"  <title>{_escape(title)}</title>",
+        f"  <desc>{_escape(description)}</desc>",
+    ]
+
+
+def horizontal_bars(series: Sequence[dict], *, title: str, value_label: str,
+                    max_rows: int = 40, color_var: str = "--impact-sequential"
+                    ) -> str:
+    """Ranked magnitude chart. One hue, because the job is size, not identity."""
+    rows = list(series)[:max_rows]
+    if not rows:
+        return _empty(title)
+    maximum = max(row["count"] for row in rows) or 1
+    plot_width = 420
+    width = LABEL_WIDTH + plot_width + VALUE_WIDTH + CHART_PAD * 2
+    height = CHART_PAD * 2 + len(rows) * (BAR_HEIGHT + BAR_GAP) - BAR_GAP
+
+    parts = _svg_open(width, height, title,
+                      f"{value_label} for {len(rows)} categories, "
+                      f"largest {maximum}.")
+    y = CHART_PAD
+    for row in rows:
+        bar_width = max(2, round(row["count"] / maximum * plot_width))
+        label = truncate(str(row["label"]), 24)
+        parts.append(
+            f'  <text class="impact-chart__label" x="{LABEL_WIDTH - 8}" '
+            f'y="{y + BAR_HEIGHT / 2}" text-anchor="end" '
+            f'dominant-baseline="central">{_escape(label)}</text>')
+        parts.append(
+            f'  <rect class="impact-chart__bar" x="{LABEL_WIDTH}" y="{y}" '
+            f'width="{bar_width}" height="{BAR_HEIGHT}" rx="4" '
+            f'fill="var({color_var})"><title>'
+            f'{_escape(row["label"])}: {row["count"]}</title></rect>')
+        parts.append(
+            f'  <text class="impact-chart__value" x="{LABEL_WIDTH + bar_width + 8}" '
+            f'y="{y + BAR_HEIGHT / 2}" dominant-baseline="central">'
+            f'{row["count"]}</text>')
+        y += BAR_HEIGHT + BAR_GAP
+    parts.append("</svg>")
+    return "\n".join(parts) + "\n"
+
+
+def columns(series: Sequence[dict], *, title: str, value_label: str,
+            color_var: str = "--impact-sequential",
+            link_base: Optional[str] = None) -> str:
+    """Single-series column chart for a value over time.
+
+    With ``link_base`` each column links to ``link_base<key>/``. A bar labelled
+    2026 invites the question of what is in it, and the page it links to is
+    where that is answered.
+    """
+    rows = list(series)
+    if not rows:
+        return _empty(title)
+    maximum = _nice_max(max(row["count"] for row in rows))
+    left = 44
+    column_width = max(COLUMN_MIN_WIDTH, min(56, 640 // max(len(rows), 1)))
+    width = left + len(rows) * (column_width + COLUMN_GAP) + CHART_PAD
+    height = PLOT_HEIGHT + 54
+
+    parts = _svg_open(width, height, title,
+                      f"{value_label} across {len(rows)} periods, "
+                      f"peak {max(row['count'] for row in rows)}.")
+    parts.extend(_y_axis(left, width, maximum))
+    x = left + COLUMN_GAP / 2
+    for row in rows:
+        bar_height = max(2, round(row["count"] / maximum * PLOT_HEIGHT))
+        y = CHART_PAD + PLOT_HEIGHT - bar_height
+        href = (f"{link_base}{row['key']}/"
+                if link_base and row.get("key") else None)
+        if href:
+            # One link around the column, its value and its label, so the whole
+            # column is one target rather than three adjacent ones.
+            parts.append(f'  <a href="{_escape(href)}" '
+                         f'aria-label="{_escape(row["label"])}: '
+                         f'{row["count"]}">')
+        parts.append(
+            f'  <rect class="impact-chart__bar" x="{x}" y="{y}" '
+            f'width="{column_width}" height="{bar_height}" rx="4" '
+            f'fill="var({color_var})"><title>'
+            f'{_escape(row["label"])}: {row["count"]}</title></rect>')
+        parts.append(
+            f'  <text class="impact-chart__value" x="{x + column_width / 2}" '
+            f'y="{y - 6}" text-anchor="middle">{row["count"]}</text>')
+        parts.append(
+            f'  <text class="impact-chart__label" x="{x + column_width / 2}" '
+            f'y="{CHART_PAD + PLOT_HEIGHT + 18}" text-anchor="middle">'
+            f'{_escape(row["label"])}</text>')
+        if href:
+            parts.append("  </a>")
+        x += column_width + COLUMN_GAP
+    parts.append("</svg>")
+    return "\n".join(parts) + "\n"
+
+
+def grouped_columns(categories: Sequence[str], groups: Sequence[dict], *,
+                    title: str, value_label: str) -> str:
+    """Grouped columns for a small number of series that are the subject.
+
+    ``groups`` is a list of ``{"label": ..., "values": [...]}``; the value list
+    is parallel to ``categories``. Capped at the categorical slots the palette
+    validates for adjacent forms.
+    """
+    if not categories or not groups:
+        return _empty(title)
+    groups = list(groups)[:4]
+    maximum = _nice_max(max((max(g["values"]) if g["values"] else 0)
+                            for g in groups))
+    left = 44
+    bar_width = 14
+    group_width = len(groups) * bar_width + (len(groups) - 1) * 2
+    step = group_width + GROUP_GAP
+    width = left + len(categories) * step + CHART_PAD
+    height = PLOT_HEIGHT + 54
+
+    parts = _svg_open(width, height, title,
+                      f"{value_label}: {len(groups)} series across "
+                      f"{len(categories)} years.")
+    parts.extend(_y_axis(left, width, maximum))
+    x = left + GROUP_GAP / 2
+    for index, category in enumerate(categories):
+        bar_x = x
+        for slot, group in enumerate(groups):
+            value = group["values"][index] if index < len(group["values"]) else 0
+            bar_height = max(0, round(value / maximum * PLOT_HEIGHT))
+            y = CHART_PAD + PLOT_HEIGHT - bar_height
+            if bar_height > 0:
+                parts.append(
+                    f'  <rect class="impact-chart__bar" x="{bar_x}" y="{y}" '
+                    f'width="{bar_width}" height="{bar_height}" rx="4" '
+                    f'fill="var({SERIES_VARS[slot]})"><title>'
+                    f'{_escape(group["label"])} {_escape(category)}: {value}'
+                    f'</title></rect>')
+                parts.append(
+                    f'  <text class="impact-chart__value impact-chart__value--tiny" '
+                    f'x="{bar_x + bar_width / 2}" y="{y - 4}" '
+                    f'text-anchor="middle">{value}</text>')
+            # 2px surface gap between adjacent fills.
+            bar_x += bar_width + 2
+        parts.append(
+            f'  <text class="impact-chart__label" x="{x + group_width / 2}" '
+            f'y="{CHART_PAD + PLOT_HEIGHT + 18}" text-anchor="middle">'
+            f'{_escape(category)}</text>')
+        x += step
+    parts.append("</svg>")
+    return "\n".join(parts) + "\n"
+
+
+def stacked_horizontal_bars(rows: Sequence[dict], *, title: str,
+                            value_label: str, max_rows: int = 40,
+                            link_base: Optional[str] = None) -> str:
+    """Ranked bars split into segments, for magnitude plus composition.
+
+    Used where the total and its make-up matter equally: how many bugs a system
+    has, and how many of those someone outside the project found.
+
+    With ``link_base`` each row becomes a link to ``link_base<key>/`` -- the
+    obvious thing to do with a bar labelled DuckDB is to click it, and the
+    detail page is what it should reach. The href is written root-relative
+    rather than through Liquid's ``relative_url``, which the SVG cannot reach;
+    that is correct as long as the site has no ``baseurl``, and it has none.
+    """
+    rows = [row for row in rows if row["count"] > 0][:max_rows]
+    if not rows:
+        return _empty(title)
+    maximum = max(row["count"] for row in rows) or 1
+    plot_width = 400
+    width = LABEL_WIDTH + plot_width + VALUE_WIDTH + CHART_PAD * 2
+    height = CHART_PAD * 2 + len(rows) * (BAR_HEIGHT + BAR_GAP) - BAR_GAP
+
+    parts = _svg_open(width, height, title,
+                      f"{value_label} for {len(rows)} categories, "
+                      f"largest {maximum}, each split by who found them.")
+    y = CHART_PAD
+    for row in rows:
+        x = float(LABEL_WIDTH)
+        label = truncate(str(row["label"]), 24)
+        href = (f"{link_base}{row['key']}/"
+                if link_base and row.get("key") else None)
+        if href:
+            # One link around the label and every segment of its bar, so the
+            # whole row is one target rather than several adjacent ones.
+            parts.append(f'  <a href="{_escape(href)}" '
+                         f'aria-label="{_escape(row["label"])}: '
+                         f'{row["count"]} bugs">')
+        parts.append(
+            f'  <text class="impact-chart__label" x="{LABEL_WIDTH - 8}" '
+            f'y="{y + BAR_HEIGHT / 2}" text-anchor="end" '
+            f'dominant-baseline="central">{_escape(label)}</text>')
+        for slot, segment in enumerate(row["segments"]):
+            if segment["count"] <= 0:
+                continue
+            span = segment["count"] / maximum * plot_width
+            # 2px surface gap between adjacent fills, per the mark spec.
+            drawn = max(2.0, span - 2)
+            parts.append(
+                f'  <rect class="impact-chart__bar" x="{x:.1f}" y="{y}" '
+                f'width="{drawn:.1f}" height="{BAR_HEIGHT}" rx="4" '
+                f'fill="var({SERIES_VARS[slot]})"><title>'
+                f'{_escape(row["label"])} &#8212; {_escape(segment["label"])}: '
+                f'{segment["count"]}</title></rect>')
+            x += span
+        parts.append(
+            f'  <text class="impact-chart__value" x="{x + 8:.1f}" '
+            f'y="{y + BAR_HEIGHT / 2}" dominant-baseline="central">'
+            f'{row["count"]}</text>')
+        if href:
+            parts.append("  </a>")
+        y += BAR_HEIGHT + BAR_GAP
+    parts.append("</svg>")
+    return "\n".join(parts) + "\n"
+
+
+def stacked_bar(series: Sequence[dict], *, title: str, value_label: str) -> str:
+    """Single horizontal stacked bar for a part-to-whole breakdown."""
+    rows = [row for row in series if row["count"] > 0][:4]
+    if not rows:
+        return _empty(title)
+    total = sum(row["count"] for row in rows)
+    width = 660
+    bar_top = 16
+    bar_height = 34
+    height = bar_top + bar_height + 26
+
+    parts = _svg_open(width, height, title,
+                      f"{value_label}: {total} in {len(rows)} states.")
+    x = 0.0
+    for slot, row in enumerate(rows):
+        segment = row["count"] / total * width
+        # 2px surface gap between segments, per the mark spec.
+        drawn = max(2.0, segment - 2)
+        parts.append(
+            f'  <rect class="impact-chart__bar" x="{x:.1f}" y="{bar_top}" '
+            f'width="{drawn:.1f}" height="{bar_height}" rx="4" '
+            f'fill="var({SERIES_VARS[slot]})"><title>'
+            f'{_escape(row["label"])}: {row["count"]}</title></rect>')
+        if segment > 90:
+            parts.append(
+                f'  <text class="impact-chart__value" x="{x + drawn / 2:.1f}" '
+                f'y="{bar_top + bar_height + 16}" text-anchor="middle">'
+                f'{_escape(row["label"])} {row["count"]}</text>')
+        x += segment
+    parts.append("</svg>")
+    return "\n".join(parts) + "\n"
+
+
+def _y_axis(left: int, width: int, maximum: int) -> List[str]:
+    """Recessive gridlines plus a baseline, at four evenly spaced ticks."""
+    parts = []
+    for step in range(5):
+        value = round(maximum * step / 4)
+        y = CHART_PAD + PLOT_HEIGHT - (PLOT_HEIGHT * step / 4)
+        parts.append(
+            f'  <line class="impact-chart__grid" x1="{left}" y1="{y:.1f}" '
+            f'x2="{width - CHART_PAD}" y2="{y:.1f}" />')
+        parts.append(
+            f'  <text class="impact-chart__tick" x="{left - 8}" y="{y:.1f}" '
+            f'text-anchor="end" dominant-baseline="central">{value}</text>')
+    parts.append(
+        f'  <line class="impact-chart__axis" x1="{left}" '
+        f'y1="{CHART_PAD + PLOT_HEIGHT}" x2="{width - CHART_PAD}" '
+        f'y2="{CHART_PAD + PLOT_HEIGHT}" />')
+    return parts
+
+
+def _empty(title: str) -> str:
+    return (f'<svg class="impact-chart impact-chart--empty" viewBox="0 0 400 60" '
+            f'width="100%" height="60" role="img" '
+            f'aria-label="{_escape(title)}: no data yet" '
+            f'xmlns="http://www.w3.org/2000/svg">\n'
+            f'  <title>{_escape(title)}</title>\n'
+            f'  <text class="impact-chart__label" x="200" y="30" '
+            f'text-anchor="middle">No records yet.</text>\n'
+            f'</svg>\n')
+
+
+PLOTS = {}
+
+
+def build_all(stats: dict) -> Dict[str, str]:
+    """Render every chart from the derived statistics."""
+    from .stats import RELATIONSHIP_LABELS
+
+    charts: Dict[str, str] = {}
+
+    charts["bugs-by-dbms"] = stacked_horizontal_bars(
+        stats["bugs"]["by_dbms_and_affiliation"],
+        title="Bugs found by SQLancer, by database system and who found them",
+        value_label="Bugs per database system",
+        link_base=pages.PERMALINK_BASE)
+
+    charts["bugs-by-reporter"] = stacked_bar(
+        stats["bugs"]["by_reporter_affiliation"],
+        title="Who found the bugs SQLancer is credited with",
+        value_label="Bug reports by reporter affiliation")
+
+    charts["bugs-by-year"] = columns(
+        stats["bugs"]["by_year"],
+        title="Bugs found by SQLancer, by year reported",
+        value_label="Bugs reported per year",
+        link_base=pages.YEAR_PERMALINK_BASE)
+
+    charts["bugs-by-technique"] = horizontal_bars(
+        stats["bugs"]["by_technique"],
+        title="Bugs found by SQLancer, by technique",
+        value_label="Bugs per technique")
+
+    charts["bugs-by-status"] = stacked_bar(
+        stats["bugs"]["by_status"],
+        title="Status of the bugs SQLancer found",
+        value_label="Bug reports by status")
+
+    charts["papers-by-year"] = columns(
+        stats["papers"]["by_year"],
+        title="Papers citing SQLancer, by publication year",
+        value_label="Papers citing SQLancer per year")
+
+    years = [str(year) for year in stats["papers"]["years"]]
+    yearly = stats["papers"]["by_relationship_year"]
+    # Three adjacent series, within the palette's validated limit. Reuse and
+    # extension share one, because they answer one question and a few papers
+    # do both; recognition keeps its own, being the signal that SQLancer is the
+    # reference point in its field, which the other two do not capture.
+    charts["papers-relationships-by-year"] = grouped_columns(
+        years,
+        [{"label": RELATIONSHIP_LABELS[key],
+          "values": [entry["count"] for entry in yearly.get(key, [])]}
+         for key in ("reusing_or_extending", "compares_with",
+                     "describes_as_state_of_the_art")],
+        title="How papers relate to SQLancer, by publication year",
+        value_label="Papers per relationship per year")
+
+    return charts
+
+
+def write_all(stats: Optional[dict] = None) -> Dict[str, bool]:
+    """Write every chart to ``_includes/impact/plots``; returns what changed."""
+    if stats is None:
+        stats = config.load_json(config.DATA_FILES["stats"])
+    charts = build_all(stats)
+    config.PLOT_DIR.mkdir(parents=True, exist_ok=True)
+    changed: Dict[str, bool] = {}
+    for name, markup in charts.items():
+        path = config.PLOT_DIR / f"{name}.svg"
+        existing = path.read_text(encoding="utf-8") if path.exists() else None
+        if existing == markup:
+            changed[name] = False
+            continue
+        path.write_text(markup, encoding="utf-8")
+        changed[name] = True
+    return changed
+
+
+def main() -> int:
+    changed = write_all()
+    updated = [name for name, did in changed.items() if did]
+    print(f"plots: {len(changed)} charts, "
+          f"{len(updated) or 'no'} updated"
+          + (f" ({', '.join(sorted(updated))})" if updated else ""))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/pr.py b/tools/impact/pr.py
new file mode 100644
index 0000000..8c8f942
--- /dev/null
+++ b/tools/impact/pr.py
@@ -0,0 +1,229 @@
+"""Renders the pull-request body for a collection run.
+
+The body is written for a reviewer whose job is to decide whether each new claim
+is justified. So it leads with counts, then lists every newly accepted record
+with a direct link to the evidence behind it, and finally says what was rejected
+and what could not be decided. A change that cannot be followed to a source is a
+change that should not be merged.
+"""
+
+from __future__ import annotations
+
+from typing import Dict, List, Optional
+
+from . import config
+
+MAX_LISTED = 30
+
+
+def _load(name: str) -> Dict[str, dict]:
+    path = config.DATA_FILES[name]
+    if not path.exists():
+        return {}
+    payload = config.load_json(path)
+    from .dataset import LIST_FIELD
+    return {record["id"]: record for record in payload.get(LIST_FIELD[name], [])}
+
+
+def _evidence_links(record: dict, limit: int = 2) -> str:
+    """First few evidence URLs, so the reviewer can follow the claim."""
+    urls: List[str] = []
+
+    def walk(node):
+        if isinstance(node, dict):
+            if node.get("source_url") and node.get("source_type"):
+                if node["source_url"] not in urls:
+                    urls.append(node["source_url"])
+            for value in node.values():
+                walk(value)
+        elif isinstance(node, list):
+            for item in node:
+                walk(item)
+
+    walk(record.get("attribution") or record.get("relationships")
+         or record.get("evidence") or record)
+    return " ".join(f"[evidence]({url})" for url in urls[:limit])
+
+
+def _bug_line(record: dict) -> str:
+    technique = record.get("technique") or "technique not recorded"
+    return (f"- **{record['dbms']}** — {record['title']} "
+            f"(`{record['finder']}` / `{technique}`, {record['status']}) "
+            f"{_evidence_links(record)}")
+
+
+def _paper_line(record: dict, relationship: str) -> str:
+    entry = record["relationships"][relationship]
+    techniques = ", ".join(entry.get("techniques", [])) or "-"
+    excerpt = ""
+    for item in entry.get("evidence", []):
+        if item.get("excerpt"):
+            excerpt = f"\n  > {item['excerpt'][:300]}"
+            break
+    link = record.get("url") or ""
+    return (f"- [{record['title']}]({link}) ({record.get('year')}) — "
+            f"techniques: {techniques}{excerpt}")
+
+
+def _adoption_line(record: dict) -> str:
+    return (f"- **{record['dbms']}** — `{record['relationship']}`: "
+            f"{record['summary']} {_evidence_links(record)}")
+
+
+def _resource_line(record: dict) -> str:
+    return f"- [{record['title']}]({record['url']}) — {record['type']}"
+
+
+def _section(title: str, lines: List[str], *, empty: Optional[str] = None) -> str:
+    if not lines:
+        return f"### {title}\n\n{empty or '_None._'}\n"
+    shown = lines[:MAX_LISTED]
+    body = "\n".join(shown)
+    if len(lines) > MAX_LISTED:
+        body += f"\n- _...and {len(lines) - MAX_LISTED} more._"
+    return f"### {title}\n\n{body}\n"
+
+
+def render(report: dict) -> str:
+    """Build the markdown body from a run report plus the data on disk."""
+    sources = report.get("sources", {})
+    bugs = _load("bugs")
+    papers = _load("papers")
+    adoption = _load("adoption")
+    resources = _load("resources")
+
+    def added(name: str) -> List[str]:
+        return sources.get(name, {}).get("added", []) or []
+
+    def updated(name: str) -> List[str]:
+        return sources.get(name, {}).get("updated", []) or []
+
+    new_bugs = [bugs[i] for i in added("bugs") if i in bugs]
+    status_changes = [bugs[i] for i in updated("bugs") if i in bugs]
+    new_papers = [papers[i] for i in added("papers") if i in papers]
+    new_adoption = [adoption[i] for i in added("adoption") if i in adoption]
+    new_resources = [resources[i] for i in added("resources") if i in resources]
+
+    def relationship_added(relationship: str) -> List[dict]:
+        """Papers whose relationship is newly positive, new or updated."""
+        out = []
+        for record_id in added("papers") + updated("papers"):
+            record = papers.get(record_id)
+            if not record:
+                continue
+            entry = record["relationships"].get(relationship, {})
+            if entry.get("value") == "yes":
+                out.append(record)
+        return out
+
+    infrastructure = relationship_added("uses_infrastructure")
+    extends = relationship_added("extends_technique")
+    compares = relationship_added("compares_with")
+
+    rejected: List[dict] = []
+    needs_review: List[dict] = []
+    removed: List[dict] = []
+    for entry in sources.values():
+        rejected.extend(entry.get("rejected") or [])
+        needs_review.extend(entry.get("needs_review") or [])
+        removed.extend(entry.get("removed") or [])
+
+    lines = [
+        "## Impact data update",
+        "",
+        "Automated collection run. Every newly accepted claim below links to the "
+        "primary source it rests on; please spot-check the evidence before "
+        "merging.",
+        "",
+        "| Change | Count |",
+        "| --- | ---: |",
+        f"| New bugs | {len(new_bugs)} |",
+        f"| Bug records with changed status or metadata | {len(status_changes)} |",
+        f"| New citing papers | {len(new_papers)} |",
+        f"| New `uses_infrastructure` relationships | {len(infrastructure)} |",
+        f"| New `extends_technique` relationships | {len(extends)} |",
+        f"| New `compares_with` relationships | {len(compares)} |",
+        f"| New DBMS adoption evidence | {len(new_adoption)} |",
+        f"| New resources | {len(new_resources)} |",
+        f"| Records retired | {len(removed)} |",
+        f"| Candidates rejected for insufficient evidence | {len(rejected)} |",
+        f"| Candidates needing human judgement | {len(needs_review)} |",
+        "",
+    ]
+
+    lines.append(_section("New bugs", [_bug_line(r) for r in new_bugs]))
+    lines.append(_section("Changed bug records",
+                          [_bug_line(r) for r in status_changes]))
+    lines.append(_section(
+        "Papers now recorded as using SQLancer infrastructure",
+        [_paper_line(r, "uses_infrastructure") for r in infrastructure]))
+    lines.append(_section(
+        "Papers now recorded as extending a SQLancer technique",
+        [_paper_line(r, "extends_technique") for r in extends]))
+    lines.append(_section(
+        "Papers now recorded as comparing against SQLancer",
+        [_paper_line(r, "compares_with") for r in compares]))
+    lines.append(_section("New DBMS adoption evidence",
+                          [_adoption_line(r) for r in new_adoption]))
+    lines.append(_section("New resources",
+                          [_resource_line(r) for r in new_resources]))
+    lines.append(_section(
+        "New citing papers (relationship not yet established)",
+        [f"- [{r['title']}]({r.get('url') or ''}) ({r.get('year')})"
+         for r in new_papers]))
+
+    lines.append(_section(
+        "Records retired",
+        [f"- [{entry.get('title') or entry.get('id')}]({entry.get('url') or ''})"
+         for entry in removed],
+        empty="_Nothing was retired this run._"))
+    lines.append(_section(
+        "Rejected for insufficient evidence",
+        [f"- `{entry.get('kind', '?')}` {entry.get('url') or entry.get('id') or ''}"
+         f" — {entry.get('reason', '')}" for entry in rejected],
+        empty="_Nothing was rejected this run._"))
+    lines.append(_section(
+        "Needs human judgement",
+        [f"- `{entry.get('kind', '?')}` {entry.get('url') or entry.get('id') or ''}"
+         f" — {entry.get('reason', '')}" for entry in needs_review],
+        empty="_Nothing is waiting on a human this run._"))
+
+    classifier = report.get("classifier") or {}
+    cache = report.get("cache") or {}
+    lines.extend([
+        "---",
+        "",
+        "<details><summary>Run details</summary>",
+        "",
+        f"- Started {report.get('started_at')}, finished {report.get('finished_at')}",
+        f"- Classifier: {classifier.get('api_calls', 0)} call(s), "
+        f"{classifier.get('cache_hits', 0)} served from cache, "
+        f"{classifier.get('skipped_no_key', 0)} skipped",
+        f"- Cache: {cache}",
+        "",
+    ])
+    errors = report.get("errors") or []
+    if errors:
+        lines.append("Warnings during the run:")
+        lines.append("")
+        lines.extend(f"- {error}" for error in errors)
+        lines.append("")
+    lines.append("</details>")
+    return "\n".join(lines) + "\n"
+
+
+def has_meaningful_changes(report: dict) -> bool:
+    """Whether this run produced authoritative-data changes worth a pull request.
+
+    Cache and state files are deliberately not counted: a run that only warmed
+    the cache has nothing for a human to review.
+    """
+    for name, changed in (report.get("files_changed") or {}).items():
+        if not changed:
+            continue
+        if name.startswith("plot:") or name == "stats":
+            # Derived outputs only matter alongside a record change; on their
+            # own they mean the generator changed, which is a code review.
+            continue
+        return True
+    return False
diff --git a/tools/impact/reconcile_papers.py b/tools/impact/reconcile_papers.py
new file mode 100644
index 0000000..e3b9f03
--- /dev/null
+++ b/tools/impact/reconcile_papers.py
@@ -0,0 +1,162 @@
+"""Makes the impact dataset's paper relationships follow the paper analyses.
+
+Two things classified the same papers. The impact pipeline did it from citation
+contexts and artifact inspection, a few sentences at a time; the papers
+subproject does it from the whole extracted text, with every mention of
+SQLancer in front of it and the reasoning recorded. The second is better, and
+where they disagree the second is right -- it is why the subproject was built.
+
+Until now they simply coexisted, so the impact page counted 19 papers reusing
+SQLancer infrastructure while the analyses recorded 22. This copies each
+analysed relationship across, keeping the analysis's value, its techniques and
+its quoted sentences as evidence, and marks the method as the analysis rather
+than as whatever determined the old value.
+
+A paper with no analysis is left exactly as it was: an analysis is the only
+thing that supersedes the earlier reading, and its absence says nothing.
+"""
+
+from __future__ import annotations
+
+import json
+import pathlib
+from typing import Dict, List, Optional, Tuple
+
+from . import config
+from .util import content_hash, now, truncate
+
+RECORD_DIR = config.REPO_ROOT / "_data" / "papers"
+METHOD = "llm_classification"
+RELATIONSHIPS = ("uses_infrastructure", "extends_technique", "compares_with",
+                 "describes_as_state_of_the_art")
+
+# The relationship the citation graph settles. An analysis can contest it, but
+# it is established upstream and is not this step's to overwrite.
+CITATION_GRAPH = "references"
+
+
+def analyses() -> Dict[str, dict]:
+    """Every stored analysis, keyed by the paper id it belongs to."""
+    found: Dict[str, dict] = {}
+    for path in sorted(RECORD_DIR.glob("*.json")):
+        try:
+            record = json.loads(path.read_text(encoding="utf-8"))
+        except ValueError:
+            continue
+        analysis = record.get("analysis")
+        paper_id = (record.get("paper") or {}).get("id")
+        if analysis and paper_id:
+            found[paper_id] = {"analysis": analysis, "key": path.stem,
+                               "record": record}
+    return found
+
+
+def _evidence(quotes: List[dict], key: str, timestamp: str) -> List[dict]:
+    """The analysis's quoted sentences, as impact evidence.
+
+    Each is a sentence stored verbatim when the paper's text was extracted, so
+    the source is the record it lives in rather than a URL the sentence can be
+    read at -- the paper itself is often behind a paywall, and pointing there
+    would suggest the quote can be checked where it cannot.
+    """
+    url = ("https://github.com/sqlancer/sqlancer.github.io/blob/main/"
+           f"_data/papers/{key}.json")
+    evidence = []
+    for quote in quotes or []:
+        sentence = (quote.get("sentence") or "").strip()
+        if not sentence:
+            continue
+        where = ", ".join(part for part in [
+            quote.get("section"),
+            f"page {quote['page']}" if quote.get("page") else None] if part)
+        evidence.append({
+            "source_url": url,
+            "source_type": "paper",
+            "excerpt": truncate(sentence, 500),
+            "excerpt_is_verbatim": True,
+            "note": (f"{quote.get('mention_id', 'A sentence')} in the paper's "
+                     f"extracted text" + (f", {where}" if where else "") + "."),
+            "content_sha256": content_hash(sentence),
+            "retrieved_at": timestamp,
+            "first_seen": timestamp,
+            "last_verified": timestamp,
+        })
+    return evidence
+
+
+def apply(papers: List[dict], found: Optional[Dict[str, dict]] = None,
+          timestamp: Optional[str] = None) -> Tuple[int, List[dict]]:
+    """Copy each analysis onto its paper; returns ``(changed, differences)``."""
+    timestamp = timestamp or now()
+    found = analyses() if found is None else found
+    changed = 0
+    differences: List[dict] = []
+
+    for paper in papers:
+        entry = found.get(paper.get("id"))
+        if entry is None:
+            continue
+        analysis = entry["analysis"]
+        touched = False
+        for name in RELATIONSHIPS:
+            decided = (analysis.get("relationships") or {}).get(name)
+            if not decided:
+                continue
+            before = paper["relationships"].get(name) or {}
+            evidence = _evidence(decided.get("quotes"), entry["key"], timestamp)
+            if decided.get("value") == "yes" and not evidence:
+                # The schema requires evidence for a positive claim, and an
+                # analysis that cites no sentence cannot supply one. Leaving
+                # the old value is wrong too, so the disagreement is reported.
+                differences.append({
+                    "paper": paper["id"], "relationship": name,
+                    "was": before.get("value"), "now": decided["value"],
+                    "reason": "analysis says yes but cites no sentence",
+                    "applied": False,
+                })
+                continue
+            after = {"value": decided["value"], "method": METHOD}
+            if decided.get("techniques"):
+                after["techniques"] = decided["techniques"]
+            if evidence:
+                after["evidence"] = evidence
+            if before.get("value") != after["value"]:
+                differences.append({
+                    "paper": paper["id"], "relationship": name,
+                    "was": before.get("value"), "now": after["value"],
+                    "reason": decided.get("reasoning"), "applied": True,
+                })
+            if before != after:
+                paper["relationships"][name] = after
+                touched = True
+        if touched:
+            changed += 1
+    return changed, differences
+
+
+def main() -> int:
+    from .dataset import Dataset
+
+    dataset = Dataset()
+    papers = dataset.records("papers")
+    changed, differences = apply(papers)
+    # Written before anything is reported. The report runs to hundreds of
+    # lines, and piping it to `head` closed the pipe and killed the process
+    # mid-print, losing the work it had already done.
+    dataset.save(["papers"])
+
+    flipped = [d for d in differences if d["applied"]]
+    blocked = [d for d in differences if not d["applied"]]
+    print(f"{changed} paper(s) updated, {len(flipped)} classification(s) changed"
+          + (f", {len(blocked)} could not be applied" if blocked else ""))
+    for row in flipped:
+        print(f"  {row['relationship']}: {row['was']} -> {row['now']}  "
+              f"{row['paper']}")
+    for row in blocked:
+        print(f"  SKIPPED {row['relationship']} on {row['paper']}: "
+              f"{row['reason']}")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/repos.py b/tools/impact/repos.py
new file mode 100644
index 0000000..63581c2
--- /dev/null
+++ b/tools/impact/repos.py
@@ -0,0 +1,76 @@
+"""Where a project's issues live now, when they were filed somewhere else.
+
+Projects move organisation. DuckDB was ``cwida/duckdb`` before it was
+``duckdb/duckdb``, and GitHub redirects the old links, so both forms reach the
+same issue and both look canonical to a URL normaliser. That is enough to store
+one bug twice: the project's curated list keeps the address it was filed under,
+an issue search finds the address it lives at now, and nothing connects them --
+76 DuckDB bugs were counted twice this way.
+
+The registry already records the organisations a project has used, because the
+bug search needs them. The same knowledge settles identity: an issue URL is
+rewritten to the owner the project is at today, before anything derives a record
+id from it.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Dict, Optional, Tuple
+
+from . import config
+from .util import normalize_url
+
+ISSUE_URL = re.compile(
+    r"^(https://github\.com/)([\w.-]+)/([\w.-]+)(/(?:issues|pull)/\d+.*)$",
+    re.IGNORECASE)
+
+_ALIASES: Optional[Dict[Tuple[str, str], str]] = None
+
+
+def _aliases() -> Dict[Tuple[str, str], str]:
+    """``(former owner, repo) -> current owner`` for every registered system."""
+    global _ALIASES
+    if _ALIASES is not None:
+        return _ALIASES
+    mapping: Dict[Tuple[str, str], str] = {}
+    try:
+        registry = config.load_json(config.DATA_FILES["dbms"])["dbms"]
+    except Exception:
+        registry = []
+    for entry in registry:
+        repository = entry.get("repository") or ""
+        parts = repository.rstrip("/").split("/")
+        if len(parts) < 2:
+            continue
+        owner, name = parts[-2], parts[-1]
+        for former in entry.get("github_owners") or []:
+            if former.lower() != owner.lower():
+                mapping[(former.lower(), name.lower())] = owner
+    _ALIASES = mapping
+    return mapping
+
+
+def forget_aliases() -> None:
+    """Drop the cached map, for tests that change the registry."""
+    global _ALIASES
+    _ALIASES = None
+
+
+def canonical_issue_url(url: Optional[str]) -> Optional[str]:
+    """``url`` with a former organisation rewritten to the current one.
+
+    Anything that is not a GitHub issue or pull request under a known former
+    owner comes back normalised but otherwise untouched.
+    """
+    canonical = normalize_url(url)
+    if not canonical:
+        return canonical
+    match = ISSUE_URL.match(canonical)
+    if not match:
+        return canonical
+    prefix, owner, name, rest = match.groups()
+    current = _aliases().get((owner.lower(), name.lower()))
+    if not current or current.lower() == owner.lower():
+        return canonical
+    return f"{prefix}{current}/{name}{rest}"
diff --git a/tools/impact/review.py b/tools/impact/review.py
new file mode 100644
index 0000000..779def9
--- /dev/null
+++ b/tools/impact/review.py
@@ -0,0 +1,199 @@
+"""The queue of things a collector saw and could not place.
+
+Every collector reports candidates it cannot admit on its own: an issue in a
+repository the registry does not know, a fork under an unrecognised
+organisation, a provider file for a system with no entry. Until now that list
+went into the run report and nowhere else, so each run rediscovered the same
+items and no one ever saw them. Three real gaps were found by hand that this
+list had already flagged and lost -- MatrixOne's nine bugs, Oxla's twelve, and
+a TimescaleDB provider in CnosDB's fork.
+
+So it is kept. Entries accumulate with the date they were first and last seen,
+and leave in one of three ways: the record they point at is admitted, in which
+case the item is dropped automatically; a later run classifies the candidate and
+turns it down, which settles it just as firmly and moves it to ``dismissed``;
+or a person rules that it should never be raised again, with a reason. The
+second was missing for a long time, and it mattered -- a negative answer had
+nowhere to go, so 347 decided candidates sat in the queue looking exactly like
+work nobody had done. An item that simply stops appearing is *not* dropped: a
+run that did not reach it looks exactly like a run that resolved it.
+"""
+
+from __future__ import annotations
+
+import json
+from typing import Dict, Iterable, List, Optional, Sequence, Set
+
+from . import config
+from .util import normalize_url, now, stable_digest, truncate
+
+FILE = "needs_review.json"
+SCHEMA_VERSION = "1.0.0"
+
+DESCRIPTION = (
+    "Candidates a collector could not place, kept between runs so they can be "
+    "worked through. Each entry names what was seen and why it could not be "
+    "admitted. An entry leaves when the record it points at enters the dataset, "
+    "or when a person dismisses it with a reason."
+)
+
+
+def path():
+    return config.DATA_DIR / FILE
+
+
+def load() -> dict:
+    """The queue as stored, or an empty one."""
+    if not path().exists():
+        return {"schema_version": SCHEMA_VERSION, "description": DESCRIPTION,
+                "open": [], "dismissed": []}
+    try:
+        return config.load_json(path())
+    except ValueError:
+        return {"schema_version": SCHEMA_VERSION, "description": DESCRIPTION,
+                "open": [], "dismissed": []}
+
+
+def entry_id(item: dict) -> str:
+    """A stable id for one item, so the same finding is one row over time."""
+    url = normalize_url(item.get("url") or "") or (item.get("url") or "")
+    return f"review:{stable_digest(url + '|' + (item.get('reason') or ''))}"
+
+
+def _clean(item: dict, source: str, timestamp: str) -> dict:
+    return {
+        "id": entry_id(item),
+        "kind": item.get("kind") or "unknown",
+        "source": source,
+        "url": item.get("url") or None,
+        "title": truncate(item.get("title") or "", 300) or None,
+        "reason": truncate(item.get("reason") or "", 500),
+        "first_seen": timestamp,
+        "last_seen": timestamp,
+    }
+
+
+def merge(queue: dict, items: Sequence[dict], *, source: str,
+          timestamp: Optional[str] = None,
+          admitted: Optional[Set[str]] = None) -> Dict[str, int]:
+    """Fold this run's findings into the queue; returns what changed.
+
+    ``admitted`` is the set of canonical URLs now in the dataset. An item
+    pointing at one of them has been resolved by being collected, which is the
+    only automatic way out.
+    """
+    timestamp = timestamp or now()
+    admitted = admitted or set()
+    dismissed = {row["id"] for row in queue.get("dismissed") or []}
+    by_id = {row["id"]: row for row in queue.get("open") or []}
+
+    counts = {"added": 0, "seen_again": 0, "resolved": 0}
+    for item in items:
+        row = _clean(item, source, timestamp)
+        if row["id"] in dismissed:
+            continue
+        existing = by_id.get(row["id"])
+        if existing is None:
+            by_id[row["id"]] = row
+            counts["added"] += 1
+        else:
+            existing["last_seen"] = timestamp
+            existing["reason"] = row["reason"]
+            existing["title"] = row["title"] or existing.get("title")
+            counts["seen_again"] += 1
+
+    for key, row in list(by_id.items()):
+        canonical = normalize_url(row.get("url") or "")
+        if canonical and canonical in admitted:
+            del by_id[key]
+            counts["resolved"] += 1
+
+    queue["schema_version"] = SCHEMA_VERSION
+    queue["description"] = DESCRIPTION
+    queue["open"] = sorted(by_id.values(),
+                           key=lambda row: (row["source"], row["id"]))
+    queue.setdefault("dismissed", [])
+    return counts
+
+
+def save(queue: dict) -> bool:
+    return config.write_json(path(), queue)
+
+
+def summary(queue: dict) -> Dict[str, int]:
+    """How many open items each collector is waiting on."""
+    counts: Dict[str, int] = {}
+    for row in queue.get("open") or []:
+        counts[row["source"]] = counts.get(row["source"], 0) + 1
+    return counts
+
+
+def repository_of(url: Optional[str]) -> Optional[str]:
+    """``owner/name`` for a GitHub URL, which is what a person decides about."""
+    import re
+
+    match = re.match(r"https?://github\.com/([^/]+/[^/]+)", url or "")
+    return match.group(1) if match else None
+
+
+def dismiss(queue: dict, ids: Iterable[str], reason: str, *,
+            decided_by: Optional[str] = None,
+            timestamp: Optional[str] = None) -> int:
+    """Move items out of the queue for good, with the reason recorded."""
+    timestamp = timestamp or now()
+    wanted = set(ids)
+    kept, moved = [], []
+    for row in queue.get("open") or []:
+        if row["id"] in wanted:
+            moved.append({"id": row["id"], "url": row.get("url"),
+                          "reason": reason, "decided_by": decided_by,
+                          "decided_on": timestamp[:10]})
+        else:
+            kept.append(row)
+    queue["open"] = kept
+    queue.setdefault("dismissed", []).extend(moved)
+    queue["dismissed"].sort(key=lambda row: row["id"])
+    return len(moved)
+
+
+def main() -> int:
+    """Print the queue grouped so the decisions in it are visible.
+
+    Grouped by reason and then by repository: 287 items across 52
+    repositories is a list nobody reads, while 52 repositories with counts is
+    a morning's work.
+    """
+    queue = load()
+    rows = queue.get("open") or []
+    print(f"{len(rows)} open, {len(queue.get('dismissed') or [])} dismissed")
+
+    by_reason: Dict[str, List[dict]] = {}
+    for row in rows:
+        # Reasons name a specific repository or provider; group on the shape.
+        import re
+
+        key = re.sub(r"'[^']*'", "'...'", row["reason"])
+        by_reason.setdefault(key, []).append(row)
+
+    for reason in sorted(by_reason, key=lambda k: -len(by_reason[k])):
+        group = by_reason[reason]
+        print(f"\n{len(group)}  {reason}")
+        repos: Dict[str, int] = {}
+        loose: List[dict] = []
+        for row in group:
+            name = repository_of(row.get("url"))
+            if name:
+                repos[name] = repos.get(name, 0) + 1
+            else:
+                loose.append(row)
+        for name, count in sorted(repos.items(), key=lambda kv: -kv[1])[:30]:
+            print(f"     {count:>4}  {name}")
+        if len(repos) > 30:
+            print(f"     ... and {len(repos) - 30} more repositories")
+        for row in loose[:10]:
+            print(f"          {row.get('url') or row.get('title') or row['id']}")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/run.py b/tools/impact/run.py
new file mode 100644
index 0000000..bb8ddeb
--- /dev/null
+++ b/tools/impact/run.py
@@ -0,0 +1,1081 @@
+"""Command-line entry point for the impact pipeline.
+
+Subcommands mirror the stages of the weekly run, and each is runnable on its own
+so a failure in one source can be re-run without repeating the rest:
+
+    python -m tools.impact.run collect      # everything, incrementally
+    python -m tools.impact.run collect --full        # periodic reconciliation
+    python -m tools.impact.run collect --only papers
+    python -m tools.impact.run stats        # regenerate derived figures
+    python -m tools.impact.run plots        # regenerate the charts
+    python -m tools.impact.run validate     # schemas + cross-file checks
+    python -m tools.impact.run build        # stats + plots + validate
+    python -m tools.impact.run report       # write the pull-request body
+
+``collect`` never writes a partially validated dataset: it merges, regenerates
+the derived outputs, validates, and only then reports what changed. If nothing
+changed, it says so and exits 0, which is what stops the weekly job from opening
+an empty pull request.
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+from pathlib import Path
+from typing import Dict, List, Optional, Tuple
+
+from . import config, plots, stats as stats_module, taxonomy, validate
+from .cache import Caches, State
+from .classify import Classifier
+from .dataset import Dataset, MergeReport
+from .github import GitHub
+from .util import now
+
+SOURCES = ("bugs", "papers", "adoption", "resources", "talks", "dbms",
+           "people")
+
+
+class RunReport:
+    """What one run did, in the shape the pull-request body needs."""
+
+    def __init__(self):
+        self.per_source: Dict[str, MergeReport] = {}
+        self.errors: List[str] = []
+        self.cache_summary: Dict[str, Dict[str, int]] = {}
+        self.classifier_stats: Dict[str, int] = {}
+        self.files_changed: Dict[str, bool] = {}
+        # Open items in the review queue, by collector, after this run.
+        self.review_queue: Dict[str, int] = {}
+        self.started_at = now()
+
+    @property
+    def data_changed(self) -> bool:
+        return any(changed for name, changed in self.files_changed.items())
+
+    def to_dict(self) -> dict:
+        return {
+            "started_at": self.started_at,
+            "finished_at": now(),
+            "files_changed": self.files_changed,
+            "errors": self.errors,
+            "cache": self.cache_summary,
+            "classifier": self.classifier_stats,
+            "review_queue": self.review_queue,
+            "sources": {
+                name: {
+                    "added": [r.get("id") for r in report.added],
+                    "updated": [after.get("id") for _, after in report.updated],
+                    "unchanged": report.unchanged,
+                    "removed": [
+                        {"id": r.get("id"), "title": r.get("title"),
+                         "url": r.get("url") or r.get("primary_url")}
+                        for r in report.removed
+                    ],
+                    "rejected": report.rejected,
+                    "needs_review": report.needs_review,
+                }
+                for name, report in self.per_source.items()
+            },
+        }
+
+
+def _collect_dbms(dataset: Dataset, gh: GitHub, report: RunReport,
+                  timestamp: str) -> None:
+    from .collectors import dbms_registry
+    records = dbms_registry.collect(gh, timestamp)
+    payload = dbms_registry.build_payload(records)
+    dataset.ensure("dbms", {"schema_version": payload["schema_version"]})
+    report.per_source["dbms"] = dataset.merge("dbms", payload["dbms"],
+                                              timestamp=timestamp)
+
+
+def _collect_bugs(dataset: Dataset, gh: GitHub, classifier: Classifier,
+                  report: RunReport, timestamp: str, state: State,
+                  full: bool) -> None:
+    from .collectors import (gitee_bugs, github_bugs, mariadb_jira, nus_test,
+                             provider_bugs, sqlancer_bugs)
+
+    dataset.ensure("bugs", {"schema_version": "1.0.0",
+                            "policy_version": config.policy_version()})
+
+    # Sources are merged weakest first, because a later candidate overwrites an
+    # earlier one describing the same bug. The project's curated repository is
+    # the best record there is -- it names the oracle that found each bug --
+    # so it is merged last and wins over a lab-list entry or an issue search
+    # that reaches the same report with less to say about it.
+    since = None if full else state.last_scan("github")
+    try:
+        found, rejected, needs_review = github_bugs.collect(
+            gh, classifier, since=since, timestamp=timestamp, full=full)
+    except Exception as error:  # a search failure must not lose the curated import
+        report.errors.append(f"github bug search failed: {error}")
+        found, rejected, needs_review = [], [], []
+    dataset.merge("bugs", found, timestamp=timestamp,
+                  rejected=rejected, needs_review=needs_review)
+
+    # The TEST lab's bug list is a backlog rather than a feed, so each run reads
+    # a bounded slice of it; the raw cache makes an already-inspected report
+    # free, and successive runs work through the rest.
+    try:
+        lab_found, lab_rejected, lab_review = nus_test.collect(
+            gh, classifier, timestamp=timestamp,
+            max_reports=1200 if full else 250)
+    except Exception as error:
+        report.errors.append(f"NUS TEST lab import failed: {error}")
+        lab_found, lab_rejected, lab_review = [], [], []
+    dataset.merge("bugs", lab_found, timestamp=timestamp,
+                  rejected=lab_rejected, needs_review=lab_review)
+
+    # Gitee hosts the trackers of several Chinese database projects, which the
+    # GitHub search cannot see at all.
+    try:
+        gitee_found, gitee_rejected, gitee_review = gitee_bugs.collect(
+            timestamp=timestamp)
+    except Exception as error:
+        report.errors.append(f"Gitee import failed: {error}")
+        gitee_found, gitee_rejected, gitee_review = [], [], []
+    dataset.merge("bugs", gitee_found, timestamp=timestamp,
+                  rejected=gitee_rejected, needs_review=gitee_review)
+
+    # MariaDB takes no bug reports on GitHub, so the repository search reaches
+    # none of its tracker; its Jira is open and answers a JQL query.
+    try:
+        jira_found, jira_rejected, jira_review = mariadb_jira.collect(
+            timestamp=timestamp)
+    except Exception as error:
+        report.errors.append(f"MariaDB Jira import failed: {error}")
+        jira_found, jira_rejected, jira_review = [], [], []
+    dataset.merge("bugs", jira_found, timestamp=timestamp,
+                  rejected=jira_rejected, needs_review=jira_review)
+
+    # Bugs recorded in SQLancer's own source, one file per provider. Read from
+    # forks as well as upstream: a vendor's own fork is the only public record
+    # of what SQLancer found in a system with no upstream provider.
+    try:
+        provider_found, _, provider_review = provider_bugs.collect(
+            gh, _bug_file_checkouts(gh), timestamp=timestamp)
+    except Exception as error:
+        report.errors.append(f"provider bug import failed: {error}")
+        provider_found, provider_review = [], []
+    dataset.merge("bugs", provider_found, timestamp=timestamp,
+                  needs_review=provider_review)
+
+    dataset.merge("bugs", sqlancer_bugs.collect(gh, timestamp=timestamp),
+                  timestamp=timestamp)
+
+    _stamp_affiliation(gh, dataset.records("bugs"))
+
+    report.per_source["bugs"] = dataset.reports["bugs"]
+    state.record_scan("github", full=full)
+    state.record_scan("nus_test", full=full)
+
+
+def _collect_people(dataset: Dataset, gh: GitHub, report: RunReport,
+                    timestamp: str) -> None:
+    """Refresh the roster of people whose reports are worth reading.
+
+    Kept as a data file rather than derived each run because the handles are
+    resolved from issues, which costs requests, and because a resolution that
+    goes wrong should be reviewable in a diff rather than silently different
+    next week.
+    """
+    from .collectors import people
+
+    dataset.ensure("people", {"schema_version": "1.0.0",
+                              "policy_version": config.policy_version()})
+    try:
+        found = people.discover(gh, timestamp=timestamp)
+    except Exception as error:
+        report.errors.append(f"people roster refresh failed: {error}")
+        return
+    dataset.merge("people", found, timestamp=timestamp)
+    report.per_source["people"] = dataset.reports["people"]
+
+
+def _bug_file_checkouts(gh: GitHub) -> List[Tuple[str, str, str]]:
+    """Where to read provider bug files from: upstream, and vendors' forks.
+
+    The forks are the real ones, listed from GitHub and kept when their owner
+    is a registered database system -- guessing a name from the registry got
+    Oxla's fork right and invented five that do not exist.
+    """
+    from .collectors import forks, provider_bugs
+
+    checkouts = [provider_bugs.UPSTREAM]
+    try:
+        owners = forks._owner_index()
+        for fork in forks.list_forks(gh):
+            login = ((fork.get("owner") or {}).get("login") or "").lower()
+            full_name = fork.get("full_name") or ""
+            if login in owners and "/" in full_name:
+                owner, repo = full_name.split("/", 1)
+                checkouts.append((owner, repo,
+                                  fork.get("default_branch") or "main"))
+    except Exception:
+        pass
+    return checkouts
+
+
+def _stamp_affiliation(gh: GitHub, records: List[dict]) -> None:
+    """Mark each bug as found by the project or by an outside adopter.
+
+    Done in one place after merging rather than in each collector, because it
+    depends on a roster none of them owns: people.json, plus the taxonomy's
+    contributor list and the TEST lab's current members.
+
+    Who found a bug matters for what the number means. The project finding bugs
+    with its own tool shows the tool works; someone outside the project finding
+    them shows it travelled.
+    """
+    from .collectors import people
+
+    tax = taxonomy.load()
+    try:
+        # Identities, not just handles: the same person is a GitHub login in one
+        # tracker and a display name in another.
+        handles = people.identities(gh)
+    except Exception:
+        handles = []
+    # A ruling wins over the roster. The roster can see that someone is listed
+    # on the lab's people page; it cannot see whether their work is this
+    # project's, and a lab lists everyone working in it.
+    external = people.counted_as_external()
+    for record in records:
+        reporter = record.get("reporter")
+        if not reporter:
+            record["reporter_affiliation"] = "unknown"
+        elif reporter.strip().lower() in external:
+            record["reporter_affiliation"] = "external"
+        elif tax.is_project_contributor(reporter, handles):
+            record["reporter_affiliation"] = "project"
+        else:
+            record["reporter_affiliation"] = "external"
+
+
+def _collect_papers(dataset: Dataset, caches: Caches, classifier: Classifier,
+                    report: RunReport, timestamp: str, state: State,
+                    full: bool, gh: Optional[GitHub] = None,
+                    artifact_budget: Optional[int] = None) -> None:
+    from .collectors import papers as papers_collector
+    from .scholarly import OpenAlex, OpenCitations, SemanticScholar
+
+    dataset.ensure("papers", {"schema_version": "1.0.0",
+                              "policy_version": config.policy_version()})
+    from .http import Fetcher
+    from .scholarly import ArXiv
+
+    s2 = SemanticScholar(caches.papers)
+    oa = OpenAlex(caches.papers)
+    # Reading preprints is only worth its cost on a reconciliation; the results
+    # are cached, so a weekly run reuses them without refetching.
+    arxiv = ArXiv(caches.papers) if full else None
+    # A third citation graph. Every candidate it finds has to be resolved
+    # through Crossref one DOI at a time, so it is walked on a reconciliation
+    # rather than on every run.
+    oc = OpenCitations(caches.papers) if full else None
+    pdf_fetcher = Fetcher(caches.artifacts, min_interval=0.5, max_retries=2,
+                          timeout=40.0)
+    try:
+        candidates, unresolved = papers_collector.gather(
+            s2=s2, oa=oa, oc=oc)
+    except Exception as error:
+        report.errors.append(f"paper discovery failed: {error}")
+        return
+    for seed in unresolved:
+        report.errors.append(
+            f"seed publication for "
+            f"{seed.get('seed_id') or seed.get('technique_id')} "
+            f"could not be resolved")
+
+    existing = dataset.index("papers")
+    records: List[dict] = []
+    needs_review: List[dict] = []
+    rejected: List[dict] = []
+    artifacts_done = 0
+
+    # Repositories that contain SQLancer's own source files, found once per run
+    # by code search. This is the route that reaches artifacts whose authors
+    # never mention SQLancer anywhere a reader would look.
+    derived_repositories: List[dict] = []
+    if gh is not None:
+        from .collectors import artifacts as artifact_collector
+        from .collectors import forks as forks_collector
+        repository_names: List[str] = []
+        try:
+            repository_names.extend(
+                artifact_collector.find_sqlancer_derived_repositories(gh))
+        except Exception as error:
+            report.errors.append(f"artifact code search failed: {error}")
+        # Code search excludes forks entirely, so a forked artifact is invisible
+        # to it. Renamed forks are the ones that became a project of their own.
+        try:
+            for full_name in forks_collector.artifact_candidates(gh=gh):
+                if full_name not in repository_names:
+                    repository_names.append(full_name)
+        except Exception as error:
+            report.errors.append(f"fork artifact scan failed: {error}")
+        try:
+            for full_name in repository_names:
+                if "/" not in full_name:
+                    continue
+                owner, repo = full_name.split("/", 1)
+                entry = artifact_collector.repository_text(gh, owner, repo)
+                if entry:
+                    derived_repositories.append(entry)
+        except Exception as error:
+            report.errors.append(f"artifact text fetch failed: {error}")
+        if derived_repositories:
+            print(f"  {len(derived_repositories)} SQLancer-derived "
+                  f"repositories to match against")
+
+    for candidate in candidates.values():
+        # A record the index could not describe cannot be published: it would
+        # appear on the page with a cover-page title or fall out of every plot.
+        unusable = papers_collector.usable_metadata(candidate)
+        if unusable:
+            needs_review.append({
+                "kind": "paper", "id": candidate.record_id,
+                "title": candidate.title, "reason": unusable,
+            })
+            continue
+        prior = existing.get(candidate.record_id)
+        classifications = _classify_paper(
+            candidate, classifier, prior, needs_review, rejected, timestamp)
+
+        is_ours = papers_collector.looks_like_sqlancer_publication(candidate)
+
+        # The paper's own words come first: they settle more than anything a
+        # citation index or a repository can, and they are what a person would
+        # read. Only relationships the text does not establish fall through to
+        # the artifact pass and the classifier.
+        if arxiv is not None and not is_ours:
+            found, resolved_arxiv = _fulltext_pass(
+                pdf_fetcher, arxiv, candidate, prior, timestamp, openalex=oa)
+            for key, relationship in found.items():
+                if classifications.get(key, {}).get("value") != "yes":
+                    classifications[key] = relationship
+            if resolved_arxiv and not candidate.arxiv:
+                candidate.arxiv = resolved_arxiv
+                candidate.external_ids.setdefault("ArXiv", resolved_arxiv)
+
+        paper_artifacts = (prior or {}).get("artifacts") or []
+        if gh is not None and not is_ours and (
+                artifact_budget is None or artifacts_done < artifact_budget):
+            found, paper_artifacts = _artifact_pass(
+                gh, candidate, prior, needs_review, timestamp,
+                derived=derived_repositories)
+            artifacts_done += 1
+            if found is not None:
+                classifications["uses_infrastructure"] = found
+
+        records.append(papers_collector.build_record(
+            candidate, timestamp=timestamp, classifications=classifications,
+            artifacts=paper_artifacts, is_sqlancer_publication=is_ours))
+
+    # A paper run always walks every seed, so on a full reconciliation the
+    # candidate set is the complete one and records that no longer cite any
+    # SQLancer publication can be retired. Incremental runs never remove.
+    report.per_source["papers"] = dataset.merge(
+        "papers", records, timestamp=timestamp,
+        rejected=rejected, needs_review=needs_review, authoritative=full)
+    state.record_scan("papers", full=full)
+
+
+def _artifact_pass(gh: GitHub, candidate, prior: Optional[dict],
+                   needs_review: List[dict], timestamp: str,
+                   derived: Optional[List[dict]] = None
+                   ) -> Tuple[Optional[dict], List[dict]]:
+    """Look for the paper's artifact and inspect it for SQLancer reuse.
+
+    This is where ``uses_infrastructure`` is actually decided. A paper that
+    reuses the codebase very often says so nowhere -- not in its text, not in
+    its artifact's README -- and the only honest evidence is the artifact
+    itself: SQLancer's package layout, its source files, its build coordinates.
+
+    Returns ``(relationship_or_None, artifact_records)``.
+    """
+    from .collectors import artifacts as artifact_collector
+
+    # A decision already made from an artifact is kept: re-searching GitHub for
+    # every paper every week would be the most expensive thing this pipeline
+    # does, for almost no new information.
+    #
+    # The exception is a decision that rests on a curated paper-to-repository
+    # link. Those are the ones most likely to be wrong, and reusing them
+    # unconditionally would freeze a mistake in place: the entry can be
+    # corrected in KNOWN_ARTIFACTS and the stale claim would still stand.
+    if prior:
+        previous = prior.get("relationships", {}).get("uses_infrastructure", {})
+        if previous.get("method") in ("artifact_inspection", "manual_curation"):
+            evidence = previous.get("evidence") or [{}]
+            curated = "Recorded by a maintainer" in (evidence[0].get("note") or "")
+            still_listed = (artifact_collector.KNOWN_ARTIFACTS.get(
+                candidate.record_id) is not None)
+            if not curated or still_listed:
+                return previous, prior.get("artifacts", [])
+
+    paper = {"title": candidate.title, "doi": candidate.doi,
+             "arxiv_id": candidate.arxiv}
+    tool_names = artifact_collector.candidate_tool_names(
+        candidate.title, candidate.s2.get("abstract"))
+
+    # The known SQLancer derivatives are matched first, from text already in
+    # memory. Only a paper none of them claims is worth spending searches on.
+    matched: List[Tuple[dict, dict]] = []
+    for entry in (derived or []):
+        link = artifact_collector.link_evidence_from_text(
+            entry, paper, tool_names=tool_names, timestamp=timestamp)
+        if link is not None:
+            matched.append((entry, link))
+
+    repositories: List[dict] = [
+        {"full_name": entry["full_name"], "_link": link}
+        for entry, link in matched
+    ]
+    known = artifact_collector.KNOWN_ARTIFACTS.get(candidate.record_id)
+    if known and not any(r["full_name"] == known for r in repositories):
+        repositories.insert(0, {"full_name": known, "_curated": True})
+    if not repositories:
+        try:
+            repositories = artifact_collector.find_repository_candidates(
+                gh, candidate.title, tool_names=tool_names, limit=4)
+        except Exception:
+            return None, []
+
+    artifact_records: List[dict] = []
+    for item in repositories:
+        full_name = item.get("full_name") or ""
+        if "/" not in full_name:
+            continue
+        owner, repo = full_name.split("/", 1)
+        if item.get("_link"):
+            link = item["_link"]
+        elif item.get("_curated"):
+            link = {
+                "source_url": f"https://github.com/{owner}/{repo}",
+                "source_type": "github_repository",
+                "excerpt": None,
+                "note": ("Recorded by a maintainer as this paper's artifact "
+                         "repository."),
+                "retrieved_at": timestamp,
+                "first_seen": timestamp,
+                "last_verified": timestamp,
+            }
+        else:
+            try:
+                link = artifact_collector.link_evidence(
+                    gh, owner, repo, paper, tool_names=tool_names,
+                    timestamp=timestamp)
+            except Exception:
+                continue
+        if link is None:
+            continue
+
+        try:
+            markers, marker_evidence = artifact_collector.inspect_repository(
+                gh, owner, repo, timestamp=timestamp)
+        except Exception:
+            continue
+        if not markers:
+            continue
+
+        artifact_records.append({
+            "url": f"https://github.com/{owner}/{repo}",
+            "kind": "github",
+            "inspected": True,
+            "inspected_at": timestamp,
+            "sqlancer_markers": markers,
+            "marker_evidence": [link] + marker_evidence,
+        })
+
+        if artifact_collector.markers_are_conclusive(markers):
+            return {
+                "value": "yes",
+                "method": "artifact_inspection",
+                "evidence": [link] + marker_evidence,
+                "classifier": {
+                    "method": "artifact_inspection",
+                    "classifier_version": artifact_collector.EXTRACTOR_VERSION,
+                    "policy_version": config.policy_version(),
+                    "taxonomy_version": config.taxonomy_version(),
+                    "classified_at": timestamp,
+                    "model": None,
+                    "rationale": ("Artifact carries SQLancer markers: "
+                                  + ", ".join(markers)),
+                },
+            }, artifact_records
+
+        # Markers present but not decisive on their own: a human should look.
+        needs_review.append({
+            "kind": "paper", "id": candidate.record_id,
+            "relationship": "uses_infrastructure", "title": candidate.title,
+            "url": f"https://github.com/{owner}/{repo}",
+            "reason": ("artifact shows weak SQLancer markers ("
+                       + ", ".join(markers) + "); needs a human decision"),
+        })
+    return None, artifact_records
+
+
+def _fulltext_pass(fetcher, arxiv, candidate, prior: Optional[dict],
+                   timestamp: str, openalex=None
+                   ) -> Tuple[Dict[str, dict], Optional[str]]:
+    """Judge a paper from its own full text, via its arXiv preprint.
+
+    Most of this corpus is published by ACM, which refuses automated requests,
+    so for the majority of papers a citation index's two-sentence contexts were
+    the only evidence available. A preprint gives the paper itself, and what it
+    did with SQLancer is usually stated plainly in it.
+
+    Returns the relationships the text establishes and the arXiv id used.
+    """
+    from .collectors import fulltext as fulltext_collector
+
+    arxiv_id = candidate.arxiv or (prior or {}).get("arxiv_id")
+    if not arxiv_id:
+        preprint = arxiv.find_preprint(
+            candidate.title,
+            [a.get("name") for a in (candidate.s2.get("authors") or [])
+             if a.get("name")])
+        if preprint is not None:
+            arxiv_id = preprint["arxiv_id"]
+
+    # Preferred order: a PDF supplied by hand, then the arXiv preprint, then
+    # the publisher if it is one that serves papers freely. ACM, IEEE and
+    # Springer refuse automated requests, which is why the hand-supplied route
+    # exists at all.
+    url = None
+    text = None
+    supplied = fulltext_collector.local_pdf(candidate.doi, arxiv_id,
+                                            candidate.s2_paper_id)
+    if supplied:
+        text = fulltext_collector.pdf_text(fetcher, supplied)
+        url = candidate.s2.get("url") or f"https://doi.org/{candidate.doi}"
+    if not text and arxiv_id:
+        url = f"https://arxiv.org/abs/{arxiv_id}"
+        text = fulltext_collector.pdf_text(
+            fetcher, f"https://arxiv.org/pdf/{arxiv_id}")
+    if not text:
+        # USENIX serves everything openly, from a path built out of the venue,
+        # the year and the first author.
+        for candidate_url in fulltext_collector.usenix_pdf_urls({
+                "venue": candidate.s2.get("venue"),
+                "year": candidate.s2.get("year"),
+                "title": candidate.title,
+                "authors": [a.get("name") for a in (candidate.s2.get("authors") or [])
+                            if a.get("name")]}):
+            text = fulltext_collector.pdf_text(fetcher, candidate_url)
+            if text:
+                url = candidate_url
+                break
+    if not text and candidate.doi and openalex is not None:
+        try:
+            work = openalex.work_by_doi(candidate.doi)
+        except Exception:
+            work = None
+        pvldb = fulltext_collector.pvldb_pdf_url(candidate.doi, work)
+        if pvldb:
+            text = fulltext_collector.pdf_text(fetcher, pvldb)
+            url = pvldb
+    if not text:
+        return {}, arxiv_id
+
+    tax = taxonomy.load()
+    signals = fulltext_collector.find_signals(text, tax)
+    notes = {
+        "uses_infrastructure": ("The paper states that its implementation is "
+                                "built on SQLancer."),
+        "compares_with": ("The paper states that it evaluates against SQLancer "
+                          "or one of its techniques."),
+        "extends_technique": ("The paper states that it extends or adapts a "
+                              "SQLancer technique."),
+    }
+
+    out: Dict[str, dict] = {}
+    for key, sentences in signals.items():
+        if not sentences:
+            continue
+        techniques: List[str] = []
+        for sentence in sentences:
+            techniques.extend(fulltext_collector.techniques_in(sentence, tax))
+        relationship = {
+            "value": "yes",
+            "method": "deterministic",
+            "evidence": [
+                fulltext_collector.evidence_from(sentence, url, notes[key],
+                                                 timestamp)
+                for sentence in sentences[:4]
+            ],
+        }
+        deduped = list(dict.fromkeys(techniques))
+        if deduped:
+            relationship["techniques"] = deduped
+        elif key != "uses_infrastructure":
+            # The schema requires a named technique for these two; without one
+            # the claim is about SQLancer as a whole, so the cited seeds stand in.
+            relationship["techniques"] = list(candidate.seeds)
+        out[key] = relationship
+    return out, arxiv_id
+
+
+def _classify_paper(candidate, classifier: Classifier, prior: Optional[dict],
+                    needs_review: List[dict], rejected: List[dict],
+                    timestamp: str) -> Dict[str, dict]:
+    """Decide the three judgement relationships for one paper.
+
+    Carries a previous decision forward when the evidence has not changed, so a
+    paper classified last month is neither re-asked nor silently downgraded.
+    """
+    from .classify import (excluded_names, finder_names, resolve_technique,
+                           technique_names)
+    from .collectors.papers import context_evidence
+
+    tax = taxonomy.load()
+    signals = candidate.signals()
+    out: Dict[str, dict] = {}
+
+    prior_hash = (prior or {}).get("provenance", {}).get("content_sha256")
+    evidence_unchanged = prior_hash == candidate.evidence_hash()
+
+    questions = {
+        "uses_infrastructure": "paper_infrastructure",
+        "extends_technique": "paper_extends",
+        "compares_with": "paper_compares",
+    }
+
+    for relationship, question in questions.items():
+        previous = (prior or {}).get("relationships", {}).get(relationship)
+        # A decision a person made is never overwritten by a later run, even
+        # when the index returns different sentences for the same citation.
+        if previous and previous.get("method") == "manual_curation":
+            out[relationship] = previous
+            continue
+        hits = signals[relationship]
+        if not hits:
+            if evidence_unchanged and prior:
+                out[relationship] = prior["relationships"][relationship]
+            continue
+
+        source_text = "\n\n".join(context for _, context in hits)
+        kwargs = {"source_label": f"citation contexts for {candidate.title}",
+                  "source_text": source_text}
+        if question == "paper_infrastructure":
+            kwargs["markers"] = []
+        elif question == "paper_extends":
+            kwargs["technique_names"] = technique_names(tax)
+        else:
+            kwargs["technique_names"] = technique_names(tax)
+
+        result = None
+        if classifier is not None:
+            result = classifier.classify(question,
+                                         source_id=candidate.record_id, **kwargs)
+        if result is None:
+            if evidence_unchanged and prior:
+                out[relationship] = prior["relationships"][relationship]
+            else:
+                needs_review.append({
+                    "kind": "paper", "id": candidate.record_id,
+                    "relationship": relationship, "title": candidate.title,
+                    "url": candidate.s2.get("url"),
+                    "reason": "needs semantic classification; classifier unavailable",
+                })
+            continue
+
+        if not result.is_positive:
+            rejected.append({
+                "kind": "paper", "id": candidate.record_id,
+                "relationship": relationship, "answer": result.answer,
+                "title": candidate.title,
+                "reason": result.reason or "classifier did not confirm",
+            })
+            out[relationship] = {"value": result.answer,
+                                 "method": "llm_classification",
+                                 "classifier": result.classifier_record()}
+            continue
+
+        techniques = []
+        for name in result.extra.get("techniques") or []:
+            resolved = resolve_technique(tax, name)
+            if resolved:
+                techniques.append(resolved)
+        if not techniques and relationship != "uses_infrastructure":
+            techniques = list(candidate.seeds)
+
+        pairs = [(techniques[0] if techniques else candidate.seeds[0], excerpt)
+                 for excerpt in result.excerpts]
+        out[relationship] = {
+            "value": "yes",
+            "method": "llm_classification",
+            "techniques": techniques,
+            "evidence": context_evidence(
+                candidate.s2.get("url") or f"https://doi.org/{candidate.doi}",
+                pairs, "Passage supporting this relationship to {technique}.",
+                timestamp),
+            "classifier": result.classifier_record(),
+        }
+    return out
+
+
+def _collect_adoption(dataset: Dataset, gh: GitHub, classifier: Classifier,
+                      report: RunReport, timestamp: str, state: State,
+                      full: bool) -> None:
+    from .collectors import adoption as adoption_collector
+    from .collectors import forks as forks_collector
+
+    dataset.ensure("adoption", {"schema_version": "1.0.0",
+                                "policy_version": config.policy_version()})
+    try:
+        records, rejected, needs_review = adoption_collector.collect(
+            gh, classifier, timestamp=timestamp)
+    except Exception as error:
+        report.errors.append(f"adoption discovery failed: {error}")
+        records, rejected, needs_review = [], [], []
+
+    # Forks are the one source code search cannot see, and a database system
+    # that keeps its own fork of SQLancer is doing so deliberately.
+    try:
+        fork_records, fork_review = forks_collector.adoption_records(
+            gh, timestamp=timestamp)
+    except Exception as error:
+        report.errors.append(f"fork scan failed: {error}")
+        fork_records, fork_review = [], []
+    needs_review.extend(fork_review)
+
+    # One record per (system, relationship): a project can show up both through
+    # its fork and through a file in its own repository, and both belong on the
+    # same record as separate evidence.
+    by_id = {record["id"]: record for record in records}
+    for record in fork_records:
+        existing = by_id.get(record["id"])
+        if existing is None:
+            by_id[record["id"]] = record
+            records.append(record)
+        else:
+            existing["evidence"].extend(record["evidence"])
+            existing["summary"] = record["summary"]
+    report.per_source["adoption"] = dataset.merge(
+        "adoption", records, timestamp=timestamp,
+        rejected=rejected, needs_review=needs_review)
+    state.record_scan("adoption", full=full)
+
+
+def _collect_resources(dataset: Dataset, gh: GitHub, report: RunReport,
+                       timestamp: str, state: State, full: bool) -> None:
+    from .collectors import resources as resources_collector
+
+    dataset.ensure("resources", {"schema_version": "1.0.0"})
+    try:
+        records, rejected, needs_review = resources_collector.collect(
+            gh, timestamp=timestamp)
+    except Exception as error:
+        report.errors.append(f"resource discovery failed: {error}")
+        return
+    # The resource collector re-derives its whole list every run from the
+    # README and the technique papers, so it is always authoritative: a
+    # resource that is no longer linked, or whose technique is no longer
+    # SQLancer's, is retired rather than left behind.
+    report.per_source["resources"] = dataset.merge(
+        "resources", records, timestamp=timestamp, rejected=rejected,
+        needs_review=needs_review, authoritative=True)
+    state.record_scan("resources", full=full)
+
+
+def _collect_talks(dataset: Dataset, report: RunReport, timestamp: str,
+                   state: State, full: bool) -> None:
+    """Refresh the talks file.
+
+    This one does not go through ``Dataset.merge``. A talk's caption mentions
+    come from a transcript somebody captured by hand, and a run made without
+    that capture in front of it knows less than the file already does -- so the
+    merge that preserves them lives in ``talks.merge`` and is used here instead.
+    """
+    from . import talks as talks_store
+    from .collectors import talks as talks_collector
+    from .http import Fetcher
+
+    fetcher = Fetcher(Caches().http, min_interval=1.0, max_retries=2,
+                      timeout=25.0)
+    try:
+        records, needs_review = talks_collector.collect(
+            fetcher, timestamp=timestamp)
+    except Exception as error:
+        report.errors.append(f"talk collection failed: {error}")
+        return
+    added, updated = [], []
+    for record in records:
+        if talks_store.merge(record, timestamp=timestamp) == "added":
+            added.append(record)
+        else:
+            updated.append((record, record))
+    # The dataset snapshot was taken before those writes, and it saves every
+    # file it holds at the end of the run. Re-read the merged file into it, or
+    # the stale copy in memory would be written straight back over the new one.
+    dataset.data["talks"] = config.load_json(talks_store.TALKS_FILE)
+    outcome = MergeReport(added=added, updated=updated,
+                          needs_review=list(needs_review))
+    report.per_source["talks"] = outcome
+    # Filing it here as well puts anything needing review through the same
+    # pass as every other source, rather than a second path of its own.
+    dataset.reports["talks"] = outcome
+    state.record_scan("talks", full=full)
+
+
+def collect(only: Optional[List[str]] = None, *, full: bool = False,
+            dry_run: bool = False) -> RunReport:
+    timestamp = now()
+    report = RunReport()
+    caches = Caches()
+    state = State()
+    gh = GitHub(caches.github)
+    classifier = Classifier(caches.classifications, dry_run=dry_run)
+    dataset = Dataset()
+    wanted = set(only or SOURCES)
+
+    if "dbms" in wanted:
+        _collect_dbms(dataset, gh, report, timestamp)
+    # The roster is refreshed before the bug search, which reads it: a reporter
+    # who appears on the lab's list this week should be searched this week.
+    if "people" in wanted or "bugs" in wanted:
+        _collect_people(dataset, gh, report, timestamp)
+    if "bugs" in wanted:
+        _collect_bugs(dataset, gh, classifier, report, timestamp, state, full)
+    if "papers" in wanted:
+        # Artifact hunting is the most request-hungry thing here, so a weekly
+        # run works through a slice of the backlog and a reconciliation takes
+        # the lot. Decisions already made from an artifact are never redone.
+        _collect_papers(dataset, caches, classifier, report, timestamp, state,
+                        full, gh=gh,
+                        artifact_budget=None if full else 40)
+    if "adoption" in wanted:
+        _collect_adoption(dataset, gh, classifier, report, timestamp, state, full)
+    if "resources" in wanted:
+        _collect_resources(dataset, gh, report, timestamp, state, full)
+    if "talks" in wanted:
+        _collect_talks(dataset, report, timestamp, state, full)
+
+    report.files_changed.update(dataset.save(wanted))
+    # Before the derived files, so a run that fails later still leaves the
+    # queue it built. An item admitted by this very run drops out here.
+    queue_counts = _persist_review(dataset, report, timestamp)
+    if any(queue_counts.values()):
+        print(f"needs review: +{queue_counts['added']} new, "
+              f"{queue_counts['seen_again']} seen again, "
+              f"{queue_counts['resolved']} resolved")
+    stats, stats_changed = stats_module.regenerate()
+    report.files_changed["stats"] = stats_changed
+    for name, changed in plots.write_all(stats).items():
+        report.files_changed[f"plot:{name}"] = changed
+
+    report.cache_summary = caches.summary()
+    report.classifier_stats = classifier.stats()
+    state.save()
+    return report
+
+
+def _print_report(report: RunReport) -> None:
+    for name, merge in sorted(report.per_source.items()):
+        print(f"{name}: +{len(merge.added)} new, ~{len(merge.updated)} updated, "
+              f"-{len(merge.removed)} removed, "
+              f"={merge.unchanged} unchanged, "
+              f"{len(merge.rejected)} rejected, "
+              f"{len(merge.needs_review)} need review")
+    changed = [name for name, did in report.files_changed.items() if did]
+    print(f"files changed: {', '.join(sorted(changed)) if changed else 'none'}")
+    print(f"classifier: {report.classifier_stats}")
+    for error in report.errors:
+        print(f"warning: {error}", file=sys.stderr)
+
+
+def _persist_review(dataset: "Dataset", report: RunReport,
+                    timestamp: str) -> Dict[str, int]:
+    """Keep this run's unplaceable candidates, so the list survives the run.
+
+    Without this the queue lived only in the run report. Three real gaps were
+    found by hand that it had already flagged and thrown away.
+    """
+    from . import review
+    from .util import normalize_url
+
+    queue = review.load()
+    counts = {"added": 0, "seen_again": 0, "resolved": 0}
+    from .dataset import LIST_FIELD
+
+    admitted = set()
+    # Only the files that hold records: techniques and policy are settings,
+    # and asking for their record list raises.
+    for name in dataset.data:
+        if name not in LIST_FIELD:
+            continue
+        for record in dataset.records(name):
+            # An adoption record keeps its address in its evidence rather than
+            # in a url field, so looking only at the record's own url left
+            # those queue items open long after the record had landed.
+            urls = [record.get("primary_url"), record.get("url")]
+            urls += [item.get("source_url")
+                     for item in record.get("evidence") or []]
+            urls += [item.get("source_url") for item in
+                     (record.get("attribution") or {}).get("evidence") or []]
+            for url in urls:
+                canonical = normalize_url(url or "")
+                if canonical:
+                    admitted.add(canonical)
+
+    for name, merged in dataset.reports.items():
+        found = review.merge(queue, merged.needs_review, source=name,
+                             timestamp=timestamp, admitted=admitted)
+        for key in counts:
+            counts[key] += found[key]
+
+    # A candidate that was classified and turned down is settled, not pending.
+    # Without this it stays open for ever: the queue only dropped items whose
+    # record was admitted, so every negative answer looked exactly like work
+    # nobody had done yet, and real candidates hid among hundreds of them.
+    turned_down = {}
+    for merged in dataset.reports.values():
+        for item in merged.rejected or []:
+            canonical = normalize_url(item.get("url") or "")
+            if canonical:
+                turned_down[canonical] = item.get("reason") or "rejected"
+    settled = [row["id"] for row in queue.get("open") or []
+               if normalize_url(row.get("url") or "") in turned_down]
+    if settled:
+        counts["resolved"] += review.dismiss(
+            queue, settled, "classified and turned down on a later run",
+            decided_by="pipeline", timestamp=timestamp)
+    review.save(queue)
+    report.review_queue = review.summary(queue)
+    return counts
+
+
+def build() -> int:
+    from . import pages as pages_module
+
+    stats, stats_changed = stats_module.regenerate()
+    changed = plots.write_all(stats)
+    # After the plots: the chart links to these pages, so a system that has just
+    # appeared must have a page by the time the link to it is written.
+    page_outcome = pages_module.write_all(stats)
+    page_outcome.update({f"year:{k}": v
+                         for k, v in pages_module.write_years(stats).items()})
+    page_outcome.update({f"talk:{k}": v
+                         for k, v in pages_module.write_talks().items()})
+    problems = validate.validate_all()
+    for problem in problems:
+        print(problem, file=sys.stderr)
+    updated = [name for name, did in changed.items() if did]
+    moved = [name for name, state in page_outcome.items()
+             if state != "unchanged"]
+    print(f"stats.json {'updated' if stats_changed else 'unchanged'}; "
+          f"{len(updated)} plot(s) updated; "
+          f"{len(page_outcome)} database system page(s), "
+          f"{len(moved) or 'none'} changed")
+    if problems:
+        print(f"{len(problems)} validation problem(s)", file=sys.stderr)
+        return 1
+    print("impact data: all files valid")
+    return 0
+
+
+def main(argv: Optional[List[str]] = None) -> int:
+    parser = argparse.ArgumentParser(prog="tools.impact.run")
+    sub = parser.add_subparsers(dest="command", required=True)
+
+    collect_parser = sub.add_parser("collect", help="run the collectors")
+    collect_parser.add_argument("--only", nargs="+", choices=SOURCES,
+                                help="restrict to these sources")
+    collect_parser.add_argument("--full", action="store_true",
+                                help="periodic full reconciliation")
+    collect_parser.add_argument("--dry-run", action="store_true",
+                                help="never call the classifier")
+    collect_parser.add_argument("--report", type=Path,
+                                help="write the run report as JSON")
+
+    sub.add_parser("stats", help="regenerate stats.json")
+    sub.add_parser("plots", help="regenerate the charts")
+    sub.add_parser("pages", help="regenerate the per-database-system pages")
+    sub.add_parser("review",
+                   help="show the queue of candidates needing a decision")
+    sub.add_parser("validate", help="validate the dataset")
+    sub.add_parser("build", help="stats + plots + validate")
+    sub.add_parser("notes",
+                   help="write one note per citing paper (summary + evidence)")
+    intake_parser = sub.add_parser(
+        "intake", help="file downloaded paper PDFs into the drop folder")
+    intake_parser.add_argument("--source", type=Path)
+    intake_parser.add_argument("--limit", type=int, default=40)
+    intake_parser.add_argument("--dry-run", action="store_true")
+    intake_parser.add_argument("--json", type=Path)
+    dedupe_parser = sub.add_parser(
+        "dedupe", help="merge paper records that are the same work")
+    dedupe_parser.add_argument("--apply", action="store_true")
+    sub.add_parser("worklist",
+                   help="list the papers whose PDF has to be supplied by hand")
+
+    report_parser = sub.add_parser("report", help="render a pull-request body")
+    report_parser.add_argument("--input", type=Path, required=True)
+    report_parser.add_argument("--output", type=Path)
+
+    args = parser.parse_args(argv)
+
+    if args.command == "collect":
+        report = collect(args.only, full=args.full, dry_run=args.dry_run)
+        _print_report(report)
+        problems = validate.validate_all()
+        for problem in problems:
+            print(problem, file=sys.stderr)
+        if args.report:
+            args.report.parent.mkdir(parents=True, exist_ok=True)
+            args.report.write_text(
+                json.dumps(report.to_dict(), indent=2, ensure_ascii=False) + "\n",
+                encoding="utf-8")
+        return 1 if problems else 0
+
+    if args.command == "stats":
+        return stats_module.main()
+    if args.command == "plots":
+        return plots.main()
+    if args.command == "pages":
+        from . import pages as pages_module
+        return pages_module.main()
+    if args.command == "review":
+        from . import review as review_module
+        return review_module.main()
+    if args.command == "validate":
+        return validate.main()
+    if args.command == "build":
+        return build()
+    if args.command == "notes":
+        from . import notes as notes_module
+        return notes_module.main()
+    if args.command == "intake":
+        from . import intake
+        argv = []
+        if args.source:
+            argv += ["--source", str(args.source)]
+        argv += ["--limit", str(args.limit)]
+        if args.dry_run:
+            argv.append("--dry-run")
+        if args.json:
+            argv += ["--json", str(args.json)]
+        return intake.main(argv)
+    if args.command == "dedupe":
+        from . import dedupe as dedupe_module
+        return dedupe_module.main(apply=args.apply)
+    if args.command == "worklist":
+        from . import worklist
+        return worklist.main()
+    if args.command == "report":
+        from .pr import render
+        body = render(json.loads(args.input.read_text(encoding="utf-8")))
+        if args.output:
+            args.output.write_text(body, encoding="utf-8")
+        else:
+            print(body)
+        return 0
+    return 2
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/scholarly.py b/tools/impact/scholarly.py
new file mode 100644
index 0000000..34dfe4c
--- /dev/null
+++ b/tools/impact/scholarly.py
@@ -0,0 +1,367 @@
+"""Clients for the scholarly APIs used to discover SQLancer-citing papers.
+
+Semantic Scholar is the primary source because it merges preprint and
+conference versions of the same work into one record and, crucially, returns
+``contexts``: the sentences in the citing paper that surround the citation.
+Those sentences are verbatim source text, which makes them usable directly as
+evidence excerpts and means most classification questions can be asked about
+real quoted material rather than about a guess.
+
+OpenAlex is used as a fallback for metadata (DOI, venue, open-access links)
+when Semantic Scholar has gaps.
+"""
+
+from __future__ import annotations
+
+import json
+import os
+import re
+from typing import Any, Dict, Iterator, List, Optional, Sequence
+
+from . import config
+from .cache import RawCache
+from .http import Fetcher, with_params
+
+S2_API = "https://api.semanticscholar.org/graph/v1"
+OPENALEX_API = "https://api.openalex.org"
+OPENCITATIONS_API = "https://opencitations.net/index/api/v1"
+CROSSREF_API = "https://api.crossref.org"
+
+CITATION_FIELDS = ",".join([
+    "paperId", "title", "year", "venue", "externalIds", "authors",
+    "openAccessPdf", "abstract", "publicationTypes",
+])
+
+PAPER_FIELDS = ",".join([
+    "paperId", "title", "year", "venue", "externalIds", "citationCount",
+    "authors", "openAccessPdf", "abstract",
+])
+
+
+class SemanticScholar:
+    """Paged access to the Semantic Scholar graph API.
+
+    The unauthenticated pool is shared and returns 429 often, so requests are
+    spaced out and the fetcher's backoff does the rest. Setting
+    ``SEMANTIC_SCHOLAR_API_KEY`` lifts both limits considerably.
+    """
+
+    def __init__(self, cache: RawCache, *, api_key: Optional[str] = None):
+        self.api_key = api_key or os.environ.get("SEMANTIC_SCHOLAR_API_KEY")
+        headers = {"Accept": "application/json"}
+        if self.api_key:
+            headers["x-api-key"] = self.api_key
+        # The shared unauthenticated pool returns 429 readily; long backoff is
+        # cheaper than losing a weekly run to it.
+        self.fetcher = Fetcher(
+            cache, min_interval=1.0 if self.api_key else 4.0, headers=headers,
+            max_retries=8, initial_backoff=5.0, max_backoff=120.0)
+
+    def paper(self, identifier: str, *, max_age_days: int = 30) -> Optional[dict]:
+        url = with_params(f"{S2_API}/paper/{identifier}", {"fields": PAPER_FIELDS})
+        data, _ = self.fetcher.fetch_json(url, max_age_days=max_age_days)
+        return data if isinstance(data, dict) and data.get("paperId") else None
+
+    def search_one(self, title: str, *, max_age_days: int = 90) -> Optional[dict]:
+        url = with_params(f"{S2_API}/paper/search",
+                          {"query": title, "fields": PAPER_FIELDS, "limit": 5})
+        data, _ = self.fetcher.fetch_json(url, max_age_days=max_age_days)
+        if not isinstance(data, dict):
+            return None
+        wanted = _normalise_title(title)
+        for candidate in data.get("data", []):
+            if _normalise_title(candidate.get("title", "")) == wanted:
+                return candidate
+        return None
+
+    def citations(self, paper_id: str, *, page_size: int = 500,
+                  max_age_days: int = 7) -> Iterator[dict]:
+        """Yield every citation edge pointing at ``paper_id``.
+
+        Each item carries the citing paper's metadata plus ``contexts`` and
+        ``intents`` for that specific citation.
+        """
+        offset = 0
+        while True:
+            url = with_params(f"{S2_API}/paper/{paper_id}/citations", {
+                "fields": CITATION_FIELDS + ",contexts,intents,isInfluential",
+                "limit": page_size,
+                "offset": offset,
+            })
+            data, _ = self.fetcher.fetch_json(
+                url, cache_key=f"s2:citations:{paper_id}:{offset}:{page_size}",
+                max_age_days=max_age_days)
+            if not isinstance(data, dict):
+                return
+            batch = data.get("data", [])
+            for item in batch:
+                yield item
+            if "next" not in data or not batch:
+                return
+            offset = data["next"]
+            if offset >= 10000:  # API hard limit
+                return
+
+
+class OpenAlex:
+    """Secondary metadata source; polite pool keyed by a contact address."""
+
+    def __init__(self, cache: RawCache, *, mailto: Optional[str] = None):
+        self.mailto = mailto or config.CONTACT_EMAIL
+        self.fetcher = Fetcher(cache, min_interval=0.2,
+                               headers={"Accept": "application/json"})
+
+    def work_by_doi(self, doi: str, *, max_age_days: int = 90) -> Optional[dict]:
+        url = with_params(f"{OPENALEX_API}/works/doi:{doi}", {"mailto": self.mailto})
+        data, _ = self.fetcher.fetch_json(url, max_age_days=max_age_days)
+        return data if isinstance(data, dict) and data.get("id") else None
+
+    def search(self, query: str, *, per_page: int = 5,
+               max_age_days: int = 90) -> List[dict]:
+        url = with_params(f"{OPENALEX_API}/works", {
+            "search": query, "per-page": per_page, "mailto": self.mailto})
+        data, _ = self.fetcher.fetch_json(url, max_age_days=max_age_days)
+        if not isinstance(data, dict):
+            return []
+        return data.get("results", [])
+
+    def versions_of(self, seed: dict, *, max_age_days: int = 90) -> List[str]:
+        """OpenAlex work ids for every version of a seed publication.
+
+        OpenAlex keeps the preprint and the conference version as separate
+        works, each with its own citation list, so a seed has to be resolved to
+        all of them or its citations are undercounted.
+        """
+        found: List[str] = []
+        if seed.get("doi"):
+            work = self.work_by_doi(seed["doi"], max_age_days=max_age_days)
+            if work:
+                found.append(_openalex_id(work["id"]))
+        if seed.get("arxiv_id"):
+            work = self.work_by_doi(f"10.48550/arxiv.{seed['arxiv_id']}",
+                                    max_age_days=max_age_days)
+            if work:
+                found.append(_openalex_id(work["id"]))
+        wanted = _normalise_title(seed.get("title", ""))
+        for candidate in self.search(seed.get("title", ""), per_page=8,
+                                     max_age_days=max_age_days):
+            if _normalise_title(candidate.get("display_name", "")) == wanted:
+                found.append(_openalex_id(candidate["id"]))
+        return [wid for i, wid in enumerate(found) if wid not in found[:i]]
+
+    def citing_works(self, work_id: str, *, max_age_days: int = 7
+                     ) -> Iterator[dict]:
+        """Yield every work citing ``work_id``, paged by cursor."""
+        cursor = "*"
+        while cursor:
+            url = with_params(f"{OPENALEX_API}/works", {
+                "filter": f"cites:{work_id}",
+                "per-page": 200,
+                "cursor": cursor,
+                "mailto": self.mailto,
+                "select": ("id,doi,title,display_name,publication_year,"
+                           "authorships,primary_location,best_oa_location,"
+                           "type,ids"),
+            })
+            data, _ = self.fetcher.fetch_json(
+                url, cache_key=f"openalex:cites:{work_id}:{cursor}",
+                max_age_days=max_age_days)
+            if not isinstance(data, dict):
+                return
+            for item in data.get("results", []):
+                yield item
+            cursor = (data.get("meta") or {}).get("next_cursor")
+            if not data.get("results"):
+                return
+
+
+def _openalex_id(url_or_id: str) -> str:
+    return (url_or_id or "").rsplit("/", 1)[-1]
+
+
+class OpenCitations:
+    """A third citation graph, built from Crossref's own reference data.
+
+    Different coverage again: it holds edges neither Semantic Scholar nor
+    OpenAlex returned, including an ICSE-SEIP study of eBay's SQL analytics
+    platform that cites both NoREC and TLP. What it gives back is only a pair
+    of DOIs, so a candidate found here has to be resolved to metadata before
+    it is any use, and it carries no citation context -- there is nothing in
+    the edge but the fact of the citation.
+
+    A seed with no DOI is invisible to it. That is a real limit rather than a
+    detail: PQS is a USENIX paper with no DOI and the most cited seed there
+    is.
+    """
+
+    def __init__(self, cache: RawCache, *, mailto: Optional[str] = None):
+        self.mailto = mailto or config.CONTACT_EMAIL
+        self.fetcher = Fetcher(cache, min_interval=1.0,
+                               headers={"Accept": "application/json"})
+
+    def citing_dois(self, doi: str, *, max_age_days: int = 7) -> List[str]:
+        """Every DOI recorded as citing ``doi``."""
+        from .util import normalize_doi
+
+        data, _ = self.fetcher.fetch_json(
+            f"{OPENCITATIONS_API}/citations/{doi}",
+            cache_key=f"opencitations:citations:{doi}",
+            max_age_days=max_age_days)
+        if not isinstance(data, list):
+            return []
+        found = []
+        for row in data:
+            # The field occasionally holds two DOIs run together; a value with
+            # whitespace in it is not one DOI and is not guessed at.
+            raw = (row.get("citing") or "").strip()
+            if not raw or any(c.isspace() for c in raw):
+                continue
+            citing = normalize_doi(raw)
+            if citing and citing not in found:
+                found.append(citing)
+        return found
+
+    def work_by_doi(self, doi: str, *, max_age_days: int = 90) -> Optional[dict]:
+        """Crossref metadata for a DOI, in the shape a candidate expects.
+
+        OpenCitations returns identifiers only, so the metadata comes from
+        Crossref, which is where its reference data came from in the first
+        place.
+        """
+        url = with_params(f"{CROSSREF_API}/works/{doi}", {"mailto": self.mailto})
+        data, _ = self.fetcher.fetch_json(url, max_age_days=max_age_days)
+        message = (data or {}).get("message") if isinstance(data, dict) else None
+        if not isinstance(message, dict):
+            return None
+        issued = ((message.get("issued") or {}).get("date-parts") or [[None]])[0]
+        authors = []
+        for author in message.get("author") or []:
+            name = " ".join(part for part in (author.get("given"),
+                                              author.get("family")) if part)
+            if name:
+                authors.append({"name": name})
+        return {
+            "paperId": None,
+            "externalIds": {"DOI": doi},
+            "title": (message.get("title") or [None])[0],
+            "year": issued[0] if issued else None,
+            "venue": (message.get("container-title") or [None])[0],
+            "authors": authors,
+            "abstract": None,
+            "url": f"https://doi.org/{doi}",
+        }
+
+
+class ArXiv:
+    """Finds the preprint of a paper whose published version is paywalled.
+
+    Most of the venues in this corpus are ACM, and dl.acm.org answers automated
+    requests with a 403 -- which leaves the majority of papers with no readable
+    full text at all. A great many of them have an arXiv preprint, and that
+    preprint is a perfectly good basis for judging what a paper does.
+
+    Matching is strict on purpose. A title search returns near neighbours, so a
+    candidate is accepted only when its title matches exactly once normalised
+    and it shares an author surname with the record being resolved.
+    """
+
+    API = "http://export.arxiv.org/api/query"
+
+    def __init__(self, cache: RawCache):
+        # arXiv asks for no more than one request every three seconds.
+        self.fetcher = Fetcher(cache, min_interval=3.0, max_retries=3,
+                               timeout=30.0, headers={"Accept": "application/atom+xml"})
+
+    def _entries(self, query: str, limit: int) -> List[dict]:
+        url = with_params(self.API, {"search_query": query,
+                                     "max_results": limit,
+                                     "sortBy": "relevance"})
+        try:
+            entry, _ = self.fetcher.fetch(url, max_age_days=180)
+        except Exception:
+            return []
+        if entry.get("status") != 200:
+            return []
+        return _parse_atom(entry.get("body") or "")
+
+    def find_preprint(self, title: str, authors: Sequence[str] = ()
+                      ) -> Optional[dict]:
+        """The arXiv preprint of ``title``, or None if there is no clear match."""
+        if not title or len(title) < 12:
+            return None
+        wanted = _normalise_title(title)
+        surnames = {name.split()[-1].lower()
+                    for name in authors if name and name.split()}
+        escaped = re.sub(r'["\\]', " ", title)
+        for query in (f'ti:"{escaped}"', f'all:"{escaped}"'):
+            for candidate in self._entries(query, 5):
+                if _normalise_title(candidate["title"]) != wanted:
+                    continue
+                if surnames:
+                    found = {name.split()[-1].lower()
+                             for name in candidate["authors"] if name.split()}
+                    if not (surnames & found):
+                        continue
+                return candidate
+        return None
+
+
+def _parse_atom(xml: str) -> List[dict]:
+    """Entries of an arXiv Atom response.
+
+    Parsed with the standard library rather than a dependency; the response is
+    small and its shape is fixed.
+    """
+    import xml.etree.ElementTree as ET
+
+    namespace = {"a": "http://www.w3.org/2005/Atom"}
+    try:
+        root = ET.fromstring(xml)
+    except ET.ParseError:
+        return []
+    entries = []
+    for node in root.findall("a:entry", namespace):
+        identifier = (node.findtext("a:id", "", namespace) or "").rsplit("/", 1)[-1]
+        if not identifier:
+            continue
+        title = re.sub(r"\s+", " ",
+                       node.findtext("a:title", "", namespace) or "").strip()
+        authors = [re.sub(r"\s+", " ", (a.findtext("a:name", "", namespace) or "")).strip()
+                   for a in node.findall("a:author", namespace)]
+        pdf = None
+        for link in node.findall("a:link", namespace):
+            if link.get("title") == "pdf" or link.get("type") == "application/pdf":
+                pdf = link.get("href")
+        entries.append({
+            "arxiv_id": identifier,
+            "versionless_id": re.sub(r"v\d+$", "", identifier),
+            "title": title,
+            "authors": [a for a in authors if a],
+            "pdf_url": pdf or f"https://arxiv.org/pdf/{identifier}",
+            "abstract": re.sub(r"\s+", " ",
+                               node.findtext("a:summary", "", namespace) or "").strip(),
+        })
+    return entries
+
+
+def _normalise_title(title: str) -> str:
+    return "".join(ch for ch in (title or "").lower() if ch.isalnum())
+
+
+def resolve_seed(s2: SemanticScholar, seed: dict) -> Optional[dict]:
+    """Resolve a foundational publication to a Semantic Scholar record.
+
+    Tries the recorded paper id, then the DOI, then an exact title match, so a
+    seed without a DOI (several SQLancer papers predate one) still resolves.
+    """
+    for identifier in filter(None, [
+        seed.get("s2_paper_id"),
+        f"DOI:{seed['doi']}" if seed.get("doi") else None,
+        f"ARXIV:{seed['arxiv_id']}" if seed.get("arxiv_id") else None,
+    ]):
+        paper = s2.paper(identifier)
+        if paper:
+            return paper
+    if seed.get("title"):
+        return s2.search_one(seed["title"])
+    return None
diff --git a/tools/impact/seed/__init__.py b/tools/impact/seed/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/tools/impact/seed/paper_classifications.py b/tools/impact/seed/paper_classifications.py
new file mode 100644
index 0000000..0b3f35d
--- /dev/null
+++ b/tools/impact/seed/paper_classifications.py
@@ -0,0 +1,274 @@
+"""Manually verified paper classifications for the seed dataset.
+
+The automated classifier answers these questions in the weekly run. This file is
+what the dataset was bootstrapped with: a small set of relationships a person
+read the citation contexts and confirmed, so the whole pipeline -- schemas,
+merging, statistics, plots, the page -- could be exercised against real records
+before broad automatic discovery was switched on.
+
+Each entry names the paper by its stable record id and quotes the passage the
+classification rests on. The quotes are not typed out here as a convenience:
+:func:`apply` refuses to write an entry whose excerpt is not present verbatim in
+the citation contexts already stored for that paper, so this file cannot
+introduce a claim that the collected evidence does not support.
+
+Entries are recorded with ``method: manual_curation`` and stay put; the
+classifier only fills in relationships that are not already decided.
+"""
+
+from __future__ import annotations
+
+from typing import Dict, List, Optional
+
+from .. import config, taxonomy
+from ..util import content_hash, now, verbatim_excerpt
+
+CURATOR_VERSION = "manual-curation-v1"
+
+# paper id -> relationship -> {techniques, excerpts}
+CLASSIFICATIONS: Dict[str, Dict[str, dict]] = {
+    # GDBMeter: takes Query Partitioning / TLP, proposed for relational DBMSs,
+    # and applies it to graph database engines.
+    "paper:doi:10.1145/3597926.3598044": {
+        "extends_technique": {
+            "techniques": ["tlp"],
+            "excerpts": [
+                "The key insight of this paper is that the high-level idea of "
+                "Query Partitioning, and specifically TLP, is applicable and "
+                "effective in finding logic bugs in GDBMSs and addresses the "
+                "aforementioned challenges.",
+                "This paper has demonstrated that Ternary Logic Partitioning "
+                "(TLP), a testing approach that was previously proposed for "
+                "testing RDBMSs, can also be applied to testing GDBMSs.",
+            ],
+        },
+    },
+    # Extends query partitioning into compound graph-aware metamorphic relations.
+    "paper:doi:10.14778/3636218.3636236": {
+        "extends_technique": {
+            "techniques": ["tlp"],
+            "excerpts": [
+                "We next propose the pattern partitioning technique to generate "
+                "compound MRs by extending the query partitioning technique in "
+                "the literature [34, 42].",
+                "It ports the idea of ternary query partitioning [42] used in "
+                "relational databases to GDBs.",
+            ],
+        },
+    },
+    # XPress: adapts the PQS pivot-row idea to a different data model, XML.
+    "paper:doi:10.1145/3597503.3639208": {
+        "extends_technique": {
+            "techniques": ["pqs"],
+            "excerpts": [
+                "The targeted node in XPress was inspired by the pivot row in "
+                "Pivoted Query Synthesis (PQS) [36], which was originally "
+                "proposed to test relational DBMSs.",
+            ],
+        },
+    },
+    # Applies the PQS idea to validating isolation-level implementations.
+    "paper:doi:10.1145/3627703.3650080": {
+        "extends_technique": {
+            "techniques": ["pqs"],
+            "excerpts": [
+                "This approach is inspired by pivoted query synthesis [41], "
+                "which uses a data-base system specific SQL interpreter to "
+                "execute a predicate condition to determine if a given row "
+                "would match the predicate.",
+            ],
+        },
+    },
+    "paper:doi:10.1145/3797871": {
+        "extends_technique": {
+            "techniques": ["pqs"],
+            "excerpts": [
+                "This approach is inspired by pivoted query synthesis [47], "
+                "which uses a database system specific SQL interpreter to "
+                "execute a predicate condition to determine if a given row "
+                "would match the predicate.",
+            ],
+        },
+    },
+    # Modified the SQLancer tool itself to reach database connectors.
+    "paper:doi:10.1109/ase63991.2025.00322": {
+        "uses_infrastructure": {
+            "techniques": [],
+            "excerpts": [
+                "We adapted SQLancer [2], a state-of-the-art JDBC-based "
+                "database testing tool, to support multiple database connectors "
+                "and evaluated both approaches over 100 rounds.",
+            ],
+        },
+    },
+    # SRS is built on the SQLancer codebase and evaluated against its oracles.
+    "paper:doi:10.1145/3769828": {
+        "uses_infrastructure": {
+            "techniques": [],
+            "excerpts": [
+                "We implemented SRS on top of SQLancer 1 , a DBMS testing "
+                "framework designed for the random generation of database "
+                "states and SQL queries, which also supports multiple test "
+                "oracles [23, 24].",
+            ],
+        },
+        "compares_with": {
+            "techniques": ["pqs", "tlp", "norec", "dqp"],
+            "excerpts": [
+                "We selected six state-of-the-art approaches for comparison: "
+                "PQS [25], TLP [24], NoREC [23], Pinolo [11], EET [13], and "
+                "DQP [3].",
+            ],
+        },
+    },
+    "paper:arxiv:2608.30385": {
+        "compares_with": {
+            "techniques": ["tlp"],
+            "excerpts": [
+                "We selected 4 state-of-the-art DBMS logic bug detection "
+                "approaches as baselines: EDC [4], Radar [34], EET [11], and "
+                "TLP [30].",
+            ],
+        },
+    },
+    "paper:arxiv:2406.09469": {
+        "compares_with": {
+            "techniques": ["tlp", "norec"],
+            "excerpts": [
+                "We compared S EM C ON T with TLP [41] and NoREC [40], which "
+                "are state-of-the-art metamorphic testing methods for testing "
+                "RDBMS.",
+            ],
+        },
+    },
+    # BuzzBee, evaluated against SQLancer's PQS oracle.
+    "paper:s2:d933042eb3c2a8f2e208515490d6b6a400e00fec": {
+        "compares_with": {
+            "techniques": ["pqs"],
+            "excerpts": [
+                "For relational DBMSs, we compare B UZZ B EE with S QUIRREL "
+                "[56] and SQL ANCER (PQS [43]), two DBMS fuzzers specialized in "
+                "SQL DBMS fuzzing.",
+            ],
+        },
+    },
+    "paper:s2:c5c2bcb84acde8c7fe8964aba2d6d27f220ae80b": {
+        "compares_with": {
+            "techniques": ["pqs", "norec", "tlp"],
+            "excerpts": [
+                "We compared P INOLO with the three state-of-the-art logical "
+                "bug detection techniques, namely PQS [36], N O REC [34], and "
+                "TLP [35], respectively, which correspond to three kinds of "
+                "test oracles.",
+            ],
+        },
+    },
+    "paper:s2:74820ccf6c039c7745f8e1c30857dee61b243464": {
+        "compares_with": {
+            "techniques": ["tlp"],
+            "excerpts": [
+                "To this end, we seek to compare against the TLP technique in "
+                "SQLancer, which is the closest related effort [40–42].",
+            ],
+        },
+    },
+}
+
+
+def _stored_contexts(record: dict) -> List[str]:
+    """Every citation context already stored on a paper record.
+
+    A curated excerpt has to come from this pool; that is what stops this file
+    from being a place where new claims can be written by hand.
+    """
+    contexts: List[str] = []
+    for relationship in record.get("relationships", {}).values():
+        for item in relationship.get("evidence", []) or []:
+            if item.get("excerpt"):
+                contexts.append(item["excerpt"])
+    return contexts
+
+
+def apply(*, timestamp: Optional[str] = None, strict: bool = False) -> dict:
+    """Write the curated classifications into ``papers.json``.
+
+    Returns a summary of what was applied and what was skipped. With
+    ``strict``, an excerpt that cannot be verified against the stored contexts
+    raises instead of being skipped.
+    """
+    timestamp = timestamp or now()
+    tax = taxonomy.load()
+    payload = config.load_json(config.DATA_FILES["papers"])
+    by_id = {record["id"]: record for record in payload["papers"]}
+
+    applied: List[str] = []
+    skipped: List[str] = []
+
+    for paper_id, relationships in CLASSIFICATIONS.items():
+        record = by_id.get(paper_id)
+        if record is None:
+            skipped.append(f"{paper_id}: not in papers.json")
+            continue
+        pool = "\n\n".join(_stored_contexts(record))
+        for relationship, entry in relationships.items():
+            verified = [excerpt for excerpt in entry["excerpts"]
+                        if verbatim_excerpt(pool, excerpt)]
+            missing = [e for e in entry["excerpts"] if e not in verified]
+            if missing:
+                message = (f"{paper_id}/{relationship}: "
+                           f"{len(missing)} excerpt(s) not found in the "
+                           f"collected citation contexts")
+                if strict:
+                    raise ValueError(message)
+                skipped.append(message)
+            if not verified:
+                continue
+            for technique in entry["techniques"]:
+                if technique not in tax.techniques:
+                    raise ValueError(f"{paper_id}: unknown technique {technique}")
+
+            record["relationships"][relationship] = {
+                "value": "yes",
+                "method": "manual_curation",
+                "techniques": entry["techniques"],
+                "evidence": [{
+                    "source_url": record["url"],
+                    "source_type": "paper_citation_context",
+                    "excerpt": excerpt,
+                    "excerpt_is_verbatim": True,
+                    "note": ("Passage in the paper, as indexed by Semantic "
+                             "Scholar, supporting this relationship."),
+                    "content_sha256": content_hash(excerpt),
+                    "retrieved_at": timestamp,
+                    "first_seen": timestamp,
+                    "last_verified": timestamp,
+                } for excerpt in verified],
+                "classifier": {
+                    "method": "manual_curation",
+                    "classifier_version": CURATOR_VERSION,
+                    "policy_version": config.policy_version(),
+                    "taxonomy_version": config.taxonomy_version(),
+                    "classified_at": timestamp,
+                    "model": None,
+                    "rationale": ("Reviewed by a maintainer against the "
+                                  "citation contexts collected for this paper."),
+                },
+            }
+            applied.append(f"{paper_id}/{relationship}")
+
+    config.write_json(config.DATA_FILES["papers"], payload)
+    return {"applied": applied, "skipped": skipped}
+
+
+def main() -> int:
+    result = apply()
+    print(f"applied {len(result['applied'])} curated classification(s)")
+    for entry in result["applied"]:
+        print(f"  + {entry}")
+    for entry in result["skipped"]:
+        print(f"  ! {entry}")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/seed/techniques_seed.py b/tools/impact/seed/techniques_seed.py
new file mode 100644
index 0000000..4b86ae8
--- /dev/null
+++ b/tools/impact/seed/techniques_seed.py
@@ -0,0 +1,465 @@
+"""Builds ``_data/impact/techniques.json``.
+
+This file is the taxonomy the whole pipeline reads: which tools belong to the
+SQLancer umbrella, which techniques originated in SQLancer, which acronyms are
+ambiguous enough to need corroborating context, and which techniques are
+deliberately out of scope.
+
+It is generated by a script rather than hand-written JSON so that the aliases,
+context terms and evidence stay in one reviewable place, but the generated file
+is the artefact the collectors consume.
+"""
+
+from __future__ import annotations
+
+from .. import config
+from ..util import now
+
+TAXONOMY_VERSION = "taxonomy-v2"
+
+_SEEN = "2026-09-06T00:00:00Z"
+
+# Terms that establish a match is about database testing rather than an
+# unrelated use of the same acronym (TLP is also Thread Level Parallelism,
+# Transport Layer Protocol, a Linux power tool; PQS is also a chemistry suite).
+DB_CONTEXT = [
+    "sqlancer", "dbms", "database", "sql", "query", "oracle", "logic bug",
+    "metamorphic", "fuzz", "test oracle", "optimizer", "optimiser",
+]
+
+TECHNIQUES = [
+    {
+        "id": "pqs",
+        "names": ["PQS", "Pivoted Query Synthesis", "pivoted query synthesis"],
+        "display_name": "Pivoted Query Synthesis (PQS)",
+        "kind": "test_oracle",
+        "sqlancer_originated": True,
+        "introduced_year": 2020,
+        "origin_paper": {
+            "title": "Testing Database Engines via Pivoted Query Synthesis",
+            "venue": "OSDI 2020",
+            "year": 2020,
+            "url": "https://www.usenix.org/system/files/osdi20-rigger.pdf",
+            "doi": None,
+            "arxiv_id": "2001.04174",
+            "s2_paper_id": None,
+            "is_citation_seed": True,
+        },
+        "ambiguous_acronym": True,
+        "required_context_terms": DB_CONTEXT,
+        "negative_context_terms": ["photoelectron", "quantum", "parallel quantum"],
+        "description": (
+            "Selects a random row, the pivot row, and generates a query that is "
+            "guaranteed to fetch it. If the row is missing from the result set, a "
+            "bug has been found."
+        ),
+    },
+    {
+        "id": "norec",
+        "names": [
+            "NoREC",
+            "Non-optimizing Reference Engine Construction",
+            "Non-Optimizing Reference Engine Construction",
+            "non-optimizing reference engine construction",
+        ],
+        "display_name": "Non-optimizing Reference Engine Construction (NoREC)",
+        "kind": "test_oracle",
+        "sqlancer_originated": True,
+        "introduced_year": 2020,
+        "origin_paper": {
+            "title": (
+                "Detecting optimization bugs in database engines via "
+                "non-optimizing reference engine construction"
+            ),
+            "venue": "ESEC/FSE 2020",
+            "year": 2020,
+            "url": "https://arxiv.org/abs/2007.08292",
+            "doi": "10.1145/3368089.3409710",
+            "arxiv_id": "2007.08292",
+            "s2_paper_id": "55a4e4a42cd86fac55491d089d07d04f09dcf957",
+            "is_citation_seed": True,
+        },
+        "ambiguous_acronym": False,
+        "required_context_terms": [],
+        "negative_context_terms": [],
+        "description": (
+            "Translates a query that the DBMS is likely to optimise into one for "
+            "which hardly any optimisation applies, then compares the two result "
+            "sets. A mismatch indicates an optimisation bug."
+        ),
+    },
+    {
+        "id": "tlp",
+        "names": [
+            "TLP",
+            "Ternary Logic Partitioning",
+            "ternary logic partitioning",
+            "query partitioning",
+        ],
+        "display_name": "Ternary Logic Partitioning (TLP)",
+        "kind": "test_oracle",
+        "sqlancer_originated": True,
+        "introduced_year": 2020,
+        "origin_paper": {
+            "title": "Finding bugs in database systems via query partitioning",
+            "venue": "OOPSLA 2020",
+            "year": 2020,
+            "url": "https://dl.acm.org/doi/pdf/10.1145/3428279",
+            "doi": "10.1145/3428279",
+            "arxiv_id": None,
+            "s2_paper_id": "d69216947188267d5537ee0f0501b2d960ecc457",
+            "is_citation_seed": True,
+        },
+        "ambiguous_acronym": True,
+        "required_context_terms": DB_CONTEXT,
+        "negative_context_terms": [
+            "thread level parallelism", "thread-level parallelism",
+            "transport layer", "tlp-stat", "laptop", "power management",
+        ],
+        "description": (
+            "Partitions a query into three partitioning queries whose combined "
+            "results must match the original query's result set. Unlike NoREC and "
+            "PQS it also reaches advanced features such as aggregate functions."
+        ),
+    },
+    {
+        "id": "qpg",
+        "names": ["QPG", "Query Plan Guidance", "query plan guidance"],
+        "display_name": "Query Plan Guidance (QPG)",
+        "kind": "test_input_generation",
+        "sqlancer_originated": True,
+        "introduced_year": 2023,
+        "origin_paper": {
+            "title": "Testing Database Systems via Query Plan Guidance",
+            "venue": "ICSE 2023",
+            "year": 2023,
+            "url": "https://arxiv.org/pdf/2312.17510",
+            "doi": None,
+            "arxiv_id": "2312.17510",
+            "s2_paper_id": None,
+            "is_citation_seed": True,
+        },
+        "ambiguous_acronym": True,
+        "required_context_terms": DB_CONTEXT,
+        "negative_context_terms": [],
+        "description": (
+            "Feedback-guided test input generation that mutates the database state "
+            "when no new query plans have been observed, on the insight that query "
+            "plans capture whether interesting behaviour is being exercised."
+        ),
+    },
+    {
+        "id": "cert",
+        "names": [
+            "CERT",
+            "Cardinality Estimation Restriction Testing",
+            "cardinality estimation restriction testing",
+        ],
+        "display_name": "Cardinality Estimation Restriction Testing (CERT)",
+        "kind": "test_oracle",
+        "sqlancer_originated": True,
+        "introduced_year": 2024,
+        "origin_paper": {
+            "title": (
+                "Cardinality Estimation Restriction Testing: Finding Performance "
+                "Issues in Database Systems"
+            ),
+            "venue": "ICSE 2024",
+            "year": 2024,
+            "url": "https://arxiv.org/pdf/2306.00355",
+            "doi": None,
+            "arxiv_id": "2306.00355",
+            "s2_paper_id": None,
+            "is_citation_seed": True,
+        },
+        "ambiguous_acronym": True,
+        "required_context_terms": DB_CONTEXT + ["cardinality", "performance issue"],
+        # "cert" is overwhelmingly TLS in a database tracker; without these,
+        # every SSL certificate bug in PostgreSQL and CockroachDB reads as a
+        # CERT finding.
+        "negative_context_terms": [
+            "certificate", "cert.org", "us-cert", "x.509", "x509", "ssl",
+            "tls", "https", "ca cert", "root cert", "crl", "keystore",
+            "truststore", "openssl", "self-signed", "cert-manager",
+            "auth_method", "sslmode",
+        ],
+        "description": (
+            "Derives a more restrictive query whose estimated cardinality must not "
+            "exceed the original's. A violation points at a performance issue. The "
+            "only SQLancer oracle aimed at performance rather than correctness."
+        ),
+    },
+    {
+        "id": "dqp",
+        "names": ["DQP", "Differential Query Plan", "differential query plans"],
+        "display_name": "Differential Query Plans (DQP)",
+        "kind": "test_oracle",
+        "sqlancer_originated": True,
+        "introduced_year": 2024,
+        "origin_paper": {
+            "title": "Keep It Simple: Testing Databases via Differential Query Plans",
+            "venue": "SIGMOD 2024",
+            "year": 2024,
+            "url": "https://dl.acm.org/doi/pdf/10.1145/3654991",
+            "doi": "10.1145/3654991",
+            "arxiv_id": None,
+            "s2_paper_id": "63b32cbfc5e6ab0043715a9adcf6ad1944c9ba49",
+            "is_citation_seed": True,
+        },
+        "ambiguous_acronym": True,
+        "required_context_terms": DB_CONTEXT,
+        "negative_context_terms": [],
+        "description": (
+            "Forces the DBMS to execute the same query using different query plans "
+            "and compares the results, which must agree."
+        ),
+    },
+    {
+        "id": "coddtest",
+        "names": [
+            "CODDTest",
+            "Constant Optimization Driven Database System Testing",
+            "constant optimization driven database system testing",
+        ],
+        "display_name": "Constant-Optimization-Driven Testing (CODDTest)",
+        "kind": "test_oracle",
+        "sqlancer_originated": True,
+        "introduced_year": 2025,
+        "origin_paper": {
+            "title": (
+                "Constant Optimization Driven Database System Testing"
+            ),
+            "venue": "SIGMOD 2025",
+            "year": 2025,
+            "url": "https://arxiv.org/abs/2501.11046",
+            "doi": None,
+            "arxiv_id": "2501.11046",
+            "s2_paper_id": None,
+            "is_citation_seed": True,
+        },
+        "ambiguous_acronym": False,
+        "required_context_terms": [],
+        "negative_context_terms": [],
+        "description": (
+            "Borrows constant folding from compiler testing: subexpressions are "
+            "replaced by their computed constants and the result must not change."
+        ),
+    },
+]
+
+FINDERS = [
+    {
+        "id": "sqlancer",
+        "display_name": "SQLancer",
+        "names": ["SQLancer", "sqlancer"],
+        "umbrella_member": True,
+        "repository": "https://github.com/sqlancer/sqlancer",
+        "paper": None,
+        "description": (
+            "The main SQLancer tool: hand-written SQL generators for a large set of "
+            "database systems, combined with the test oracles listed above."
+        ),
+    },
+    {
+        "id": "sqlancer_pp",
+        "display_name": "SQLancer++",
+        "names": ["SQLancer++", "SQLancerPlusPlus", "SQLancer PP"],
+        "umbrella_member": True,
+        "repository": "https://github.com/suyZhong/SQLancerPlusPlus",
+        "paper": {
+            "title": "Scaling Automated Database System Testing",
+            "venue": "ASPLOS 2026",
+            "year": 2026,
+            "url": "https://doi.org/10.1145/3779212.3790215",
+            "doi": "10.1145/3779212.3790215",
+        },
+        "description": (
+            "Scales automated testing to database systems for which no bespoke "
+            "SQLancer implementation exists, by adaptively discovering which SQL "
+            "features the system under test supports."
+        ),
+    },
+    {
+        "id": "shqvel",
+        "display_name": "ShQveL",
+        "names": ["ShQveL", "Shqvel"],
+        "umbrella_member": True,
+        "repository": "https://github.com/suyZhong/SQLancerPlusPlus",
+        "paper": {
+            "title": "Automated Database Testing via LLM-Synthesized SQL Features",
+            "venue": None,
+            "year": 2025,
+            "url": "https://arxiv.org/abs/2505.02012",
+            "doi": "10.48550/arxiv.2505.02012",
+        },
+        "description": (
+            "Augments SQLancer-style testing with SQL feature fragments synthesised "
+            "by a large language model, reaching features the hand-written "
+            "generators do not cover."
+        ),
+    },
+]
+
+EXCLUDED = [
+    {
+        "id": "dqe",
+        "names": ["DQE", "Differential Query Execution",
+                  "differential query execution"],
+        "display_name": "Differential Query Execution (DQE)",
+        "reason": (
+            "DQE was developed independently of SQLancer by Song et al. and "
+            "published at ICSE 2023; its SQLancer implementation arrived later "
+            "as a contributed pull request. Implementing a technique is not "
+            "originating it, so DQE-attributed bugs are not counted as SQLancer "
+            "findings and papers citing the DQE publication are not treated as "
+            "citing SQLancer."
+        ),
+        "evidence": [
+            {
+                "source_url": "https://github.com/sqlancer/sqlancer/blob/main/README.md",
+                "source_type": "documentation",
+                "excerpt": (
+                    "| Differential Query Execution (DQE)                    "
+                    "          | ICSE 2023     | "
+                    "[Paper](https://ieeexplore.ieee.org/document/10172736) "
+                    "[Code](https://github.com/sqlancer/sqlancer/pull/1251)"
+                ),
+                "excerpt_is_verbatim": True,
+                "note": (
+                    "SQLancer's own README credits DQE to an external ICSE 2023 "
+                    "publication and links its implementation as a pull request "
+                    "contributed to SQLancer."
+                ),
+                "content_sha256": (
+                    "sha256:e5f5180fefc22551e27d9fb83cb9337cad8625677ee6185f"
+                    "e8c312715e154042"
+                ),
+                "retrieved_at": _SEEN,
+                "first_seen": _SEEN,
+                "last_verified": _SEEN,
+            }
+        ],
+    },
+    {
+        "id": "eet",
+        "names": ["EET", "Equivalent Expression Transformation"],
+        "display_name": "Equivalent Expression Transformation (EET)",
+        "reason": (
+            "EET was introduced by Jiang and Su at OSDI 2024, independently of "
+            "SQLancer. SQLancer later gained an EET oracle, but implementing a "
+            "technique is not originating it, so bugs found with EET are not "
+            "counted as SQLancer findings."
+        ),
+        "evidence": [
+            {
+                "source_url": (
+                    "https://github.com/sqlancer/sqlancer/blob/main/src/sqlancer/"
+                    "common/oracle/EETOracle.java"
+                ),
+                "source_type": "github_code",
+                "excerpt": (
+                    "EET (Equivalent Expression Transformation) oracle, based on "
+                    "\"Detecting Logic Bugs in Database Engines via Equivalent\n * "
+                    "Expression Transformation\" (Jiang &amp; Su, OSDI'24)."
+                ),
+                "excerpt_is_verbatim": True,
+                "note": (
+                    "Class-level Javadoc of SQLancer's own EET oracle, crediting the "
+                    "technique to an external publication."
+                ),
+                "content_sha256": (
+                    "sha256:e61be0c83848f4b428ece8faa2aa9b0b006559d13c708eb"
+                    "b0285c4fe8254648d"
+                ),
+                "retrieved_at": _SEEN,
+                "first_seen": _SEEN,
+                "last_verified": _SEEN,
+            }
+        ],
+    },
+]
+
+# A paper co-authored by one of these is the project's own work, not external
+# work building on SQLancer. Matched as a whole word against each author name.
+PROJECT_AUTHORS = ["Rigger"]
+
+# People who found bugs as part of the project rather than as outside adopters.
+# The TEST lab's members are read from its people page during collection, so
+# only contributors who are not on it need listing here.
+PROJECT_CONTRIBUTORS = [
+    {"name": "Manuel Rigger", "github": "mrigger", "role": "Project author"},
+    {"name": "Zhenglin Li", "github": "ZhengLin-Li",
+     "role": "Google Summer of Code 2023 contributor"},
+    {"name": "Zhenglin Li", "github": "zhenglin-charlie-li",
+     "role": "Google Summer of Code 2023 contributor"},
+    {"name": "Yutan Yang", "github": "ColinYoungTaro",
+     "role": "Google Summer of Code 2023 contributor"},
+]
+
+# Papers written as part of the SQLancer project that are not already named
+# above as a technique's origin paper or a tool's paper. Both entries below are
+# also covered by PROJECT_AUTHORS; they are listed anyway so the decision
+# survives a change to that list.
+PROJECT_PUBLICATIONS = [
+    {
+        "title": ("Validating Database System Isolation Level Implementations "
+                  "with Version Certificate Recovery"),
+        "doi": "10.1145/3627703.3650080",
+        "arxiv_id": None,
+        "s2_paper_id": None,
+        "year": 2024,
+        "venue": "EuroSys 2024",
+        "reason": ("Project publication: work by the SQLancer authors applying "
+                   "the PQS idea to isolation-level validation."),
+    },
+    {
+        "title": "Bringing Order to Practical Validation of Database Isolation Levels",
+        "doi": "10.1145/3797871",
+        "arxiv_id": None,
+        "s2_paper_id": None,
+        "year": 2026,
+        "venue": "ACM Transactions on Computer Systems",
+        "reason": ("Project publication: the journal version of the EuroSys "
+                   "2024 isolation-level paper by the same authors."),
+    },
+]
+
+BUG_SYMPTOMS = [
+    {"id": "logic", "display_name": "Logic bug",
+     "description": "The DBMS returned an incorrect result set."},
+    {"id": "error", "display_name": "Unexpected error",
+     "description": "The DBMS raised an internal or otherwise unexpected error."},
+    {"id": "crash", "display_name": "Crash",
+     "description": "The DBMS process terminated abnormally."},
+    {"id": "hang", "display_name": "Hang",
+     "description": "The DBMS stopped making progress."},
+    {"id": "performance", "display_name": "Performance issue",
+     "description": "The DBMS produced a correct result far less efficiently than it could."},
+    {"id": "unknown", "display_name": "Unclassified",
+     "description": "The report does not record how the bug manifested."},
+]
+
+
+def build() -> dict:
+    return {
+        "schema_version": "1.0.0",
+        "taxonomy_version": TAXONOMY_VERSION,
+        "finders": FINDERS,
+        "sqlancer_techniques": TECHNIQUES,
+        "project_authors": PROJECT_AUTHORS,
+        "project_contributors": PROJECT_CONTRIBUTORS,
+        "project_publications": PROJECT_PUBLICATIONS,
+        "excluded_techniques": EXCLUDED,
+        "bug_symptoms": BUG_SYMPTOMS,
+    }
+
+
+def main() -> int:
+    changed = config.write_json(config.DATA_FILES["techniques"], build())
+    print(f"techniques.json: {'updated' if changed else 'unchanged'} "
+          f"({len(TECHNIQUES)} techniques, {len(FINDERS)} finders) at {now()}")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/stats.py b/tools/impact/stats.py
new file mode 100644
index 0000000..d1a7926
--- /dev/null
+++ b/tools/impact/stats.py
@@ -0,0 +1,509 @@
+"""Derives ``_data/impact/stats.json`` from the authoritative records.
+
+The website renders statistics from this file rather than counting in Liquid, so
+that the plots, the impact page and the homepage all show the same numbers, and
+so that the numbers themselves are reviewable in a diff. Nothing here is
+hand-entered: ``regenerate`` reads only the records, and CI fails if the checked
+in file differs from what the records produce.
+"""
+
+from __future__ import annotations
+
+from collections import Counter, defaultdict
+from typing import Dict, List, Optional, Tuple
+
+from . import config, taxonomy
+from .util import content_hash
+
+SCHEMA_VERSION = "1.0.0"
+
+# Relationships that mean a paper builds on SQLancer rather than merely citing
+# it. Calling SQLancer state of the art is recognition, not building on it, so
+# it is reported alongside rather than folded into this total.
+BUILDING_ON = ("uses_infrastructure", "extends_technique", "compares_with")
+
+REPORTED_RELATIONSHIPS = ("references", "reusing_or_extending",
+                          "compares_with", "describes_as_state_of_the_art")
+
+RELATIONSHIP_LABELS = {
+    "references": "Cites SQLancer",
+    "uses_infrastructure": "Uses SQLancer infrastructure",
+    "extends_technique": "Extends a SQLancer technique",
+    "reusing_or_extending": "Uses or extends SQLancer",
+    "compares_with": "Compares against SQLancer",
+    "describes_as_state_of_the_art": "Describes SQLancer as state of the art",
+}
+
+# Reusing the codebase and developing the technique are recorded apart, because
+# a paper can do either without the other. They are reported together: they
+# answer one question, and a few papers do both, so two series would show a
+# reader a total that counts those twice.
+REUSING_OR_EXTENDING = ("uses_infrastructure", "extends_technique")
+
+ADOPTION_LABELS = {
+    "official_ci": "SQLancer in the project's CI",
+    "official_testing": "Project-maintained SQLancer testing",
+    "developer_use": "Developer-reported use",
+    "integration_contributed_by_dbms_team": "Integration contributed by the DBMS team",
+    "planned_adoption": "Adoption proposed but not yet shown",
+}
+
+# Intent is not use. A project that has proposed adopting SQLancer is worth
+# recording and worth showing, but counting it as a user would overstate reach.
+PLANNED = "planned_adoption"
+
+# How each bug earned its place. Published because the clauses differ sharply
+# in strength: one is the report naming the tool, another is only who filed it.
+ATTRIBUTION_LABELS = {
+    "explicit_tool_statement": "The report names SQLancer",
+    "technique_attribution": "The report names a SQLancer oracle",
+    "curated_primary_source": "Listed in the project's own bug repository",
+    "campaign_evidence": "The reproducer carries SQLancer's generated schema",
+    "campaign_reporter": "Filed by someone who runs SQLancer campaigns",
+}
+
+AFFILIATION_LABELS = {
+    "project": "Found by the SQLancer project",
+    "external": "Found by someone outside the project",
+    "unknown": "Reporter not recorded",
+}
+
+STATUS_LABELS = {
+    "fixed": "Fixed",
+    "fixed_in_documentation": "Fixed in documentation",
+    "verified": "Confirmed",
+    "open": "Open",
+    "closed_not_a_bug": "Closed as not a bug",
+    "closed_duplicate": "Closed as duplicate",
+    "unknown": "Unknown",
+}
+
+SYMPTOM_LABELS = {
+    "logic": "Logic bug",
+    "error": "Unexpected error",
+    "crash": "Crash",
+    "hang": "Hang",
+    "performance": "Performance issue",
+    "unknown": "Unclassified",
+}
+
+RESOURCE_LABELS = {
+    "talk": "Talks",
+    "blog_post": "Blog posts",
+    "documentation_note": "Project documentation naming SQLancer",
+    "dataset": "Datasets",
+    "artifact": "Artifacts",
+    "tool": "Tools",
+    "educational_material": "Educational material",
+}
+
+
+def _series(counter: Counter, labels: Optional[Dict[str, str]] = None,
+            *, order: Optional[List[str]] = None,
+            by_count: bool = False) -> List[dict]:
+    """Turn a counter into the ordered ``labelled_counts`` shape."""
+    if order is not None:
+        keys = [key for key in order if key in counter]
+        keys += sorted(key for key in counter if key not in order)
+    elif by_count:
+        keys = [key for key, _ in sorted(counter.items(),
+                                         key=lambda kv: (-kv[1], str(kv[0])))]
+    else:
+        keys = sorted(counter, key=str)
+    return [{"key": str(key),
+             "label": (labels or {}).get(key, str(key)),
+             "count": counter[key]}
+            for key in keys]
+
+
+def round_down_headline(total: int) -> str:
+    """Conservative display form of a bug total: 1483 -> '1,000+', 456 -> '400+'.
+
+    Rounds down to one significant figure. Rounding down rather than to nearest
+    keeps the public claim defensible -- the dataset is a lower bound on what
+    SQLancer has found, never an overstatement -- and one significant figure
+    keeps the headline stable as the number creeps up week to week.
+    """
+    if total < 10:
+        return str(total)
+    magnitude = 10 ** (len(str(total)) - 1)
+    return f"{(total // magnitude) * magnitude:,}+"
+
+
+def _counted_bugs(bugs: List[dict]) -> List[dict]:
+    """Reports accepted as genuine bugs; rejected reports are kept but not counted."""
+    return [bug for bug in bugs if bug.get("status_is_true_positive")]
+
+
+def compute(data: Dict[str, dict]) -> dict:
+    tax = taxonomy.Taxonomy(data["techniques"])
+    registry = {entry["id"]: entry for entry in data["dbms"]["dbms"]}
+    all_bugs = data["bugs"]["bugs"]
+    bugs = _counted_bugs(all_bugs)
+    papers = data["papers"]["papers"]
+    adoption = data["adoption"]["adoption"]
+    resources = data["resources"]["resources"]
+
+    external_papers = [p for p in papers if not p.get("is_sqlancer_publication")]
+
+    # -- bugs -------------------------------------------------------------
+    by_dbms = Counter(bug["dbms"] for bug in bugs)
+    by_year = Counter(bug["reported_year"] for bug in bugs
+                      if bug.get("reported_year"))
+    by_status = Counter(bug["status"] for bug in bugs)
+    by_technique = Counter(bug["technique"] or "unattributed" for bug in bugs)
+    by_finder = Counter(bug["finder"] for bug in bugs)
+    by_symptom = Counter(bug["symptom"] for bug in bugs)
+    by_rule = Counter(bug["attribution"]["rule"] for bug in bugs)
+    by_affiliation = Counter(bug.get("reporter_affiliation", "unknown")
+                             for bug in bugs)
+    # Per system, split by who found it: the same total means something
+    # different depending on whether the project or an adopter found them.
+    dbms_affiliation: Dict[str, Counter] = defaultdict(Counter)
+    for bug in bugs:
+        dbms_affiliation[bug["dbms"]][
+            bug.get("reporter_affiliation", "unknown")] += 1
+
+    technique_labels = {tid: tax.display_name(tid) for tid in tax.techniques}
+    technique_labels["unattributed"] = "Technique not recorded"
+    finder_labels = {fid: tax.finder_display_name(fid) for fid in tax.finders}
+    dbms_labels = {key: registry.get(key, {}).get("name", key) for key in by_dbms}
+
+    # -- papers -----------------------------------------------------------
+    def holds(paper: dict, relationship: str) -> bool:
+        return paper["relationships"].get(relationship, {}).get("value") == "yes"
+
+    building_on = [p for p in external_papers
+                   if any(holds(p, rel) for rel in BUILDING_ON)]
+    # Reusing the codebase and developing the technique are recorded
+    # separately, because a paper can do either without the other. They are
+    # counted together because they answer one question -- what did this paper
+    # build on SQLancer -- and only a handful of papers do both, so two
+    # figures would double-count them and neither would be the number a reader
+    # wants. Comparing is a different claim and stays its own figure.
+    reusing_or_extending = [
+        p for p in external_papers
+        if holds(p, "uses_infrastructure") or holds(p, "extends_technique")]
+    comparing = [p for p in external_papers if holds(p, "compares_with")]
+
+    def reported(paper: dict, relationship: str) -> bool:
+        """Whether a paper counts under a reported relationship.
+
+        "reusing_or_extending" is not a field on a record; it is the union of
+        the two that are, deduplicated, so a paper doing both is one paper.
+        """
+        if relationship == "reusing_or_extending":
+            return any(holds(paper, key) for key in REUSING_OR_EXTENDING)
+        return holds(paper, relationship)
+
+    paper_by_relationship = Counter()
+    for relationship in REPORTED_RELATIONSHIPS:
+        paper_by_relationship[relationship] = sum(
+            1 for p in external_papers if reported(p, relationship))
+
+    paper_years = sorted({p["year"] for p in external_papers if p.get("year")})
+    papers_by_year = Counter(p["year"] for p in external_papers if p.get("year"))
+
+    # Both halves are kept alongside the combined series: the page shows the
+    # union, and a reader who wants to know how the union splits can still
+    # find out from the data without recomputing it.
+    yearly_keys = tuple(REPORTED_RELATIONSHIPS) + REUSING_OR_EXTENDING
+    by_relationship_year: Dict[str, List[dict]] = {}
+    for relationship in yearly_keys:
+        yearly = Counter(
+            p["year"] for p in external_papers
+            if p.get("year") and reported(p, relationship))
+        by_relationship_year[relationship] = [
+            {"key": str(year), "label": str(year), "count": yearly.get(year, 0)}
+            for year in paper_years
+        ]
+
+    # An extra series for the deduplicated "builds on SQLancer" total.
+    building_yearly = Counter(p["year"] for p in building_on if p.get("year"))
+    by_relationship_year["building_on"] = [
+        {"key": str(year), "label": str(year), "count": building_yearly.get(year, 0)}
+        for year in paper_years
+    ]
+
+    technique_mentions = Counter()
+    for paper in external_papers:
+        seen = set()
+        for relationship in BUILDING_ON:
+            entry = paper["relationships"].get(relationship, {})
+            if entry.get("value") == "yes":
+                seen.update(entry.get("techniques", []))
+        technique_mentions.update(seen)
+
+    # -- DBMS -------------------------------------------------------------
+    adoption_by_dbms: Dict[str, List[str]] = defaultdict(list)
+    planned_by_dbms: Dict[str, List[str]] = defaultdict(list)
+    for entry in adoption:
+        target = (planned_by_dbms if entry["relationship"] == PLANNED
+                  else adoption_by_dbms)
+        target[entry["dbms"]].append(entry["relationship"])
+
+    supported_ids = {entry["id"] for entry in registry.values()
+                     if entry.get("supported_by_sqlancer")}
+
+    # Per-system breakdowns, for the detail page each system gets. The page is
+    # rendered from these rather than counted in Liquid for the same reason
+    # every other figure is: one place produces the number, and a diff shows
+    # when it moves.
+    detail: Dict[str, dict] = defaultdict(
+        lambda: {"years": Counter(), "techniques": Counter(),
+                 "statuses": Counter(), "symptoms": Counter(),
+                 "rules": Counter(), "reporters": Counter(),
+                 "affiliations": Counter(), "dates": [],
+                 "rejected": 0})
+    for bug in all_bugs:
+        row = detail[bug["dbms"]]
+        if not bug.get("status_is_true_positive"):
+            row["rejected"] += 1
+            continue
+        row["years"][str(bug.get("reported_year") or "unknown")] += 1
+        row["techniques"][bug.get("technique") or "unattributed"] += 1
+        row["statuses"][bug.get("status") or "unknown"] += 1
+        row["symptoms"][bug.get("symptom") or "unknown"] += 1
+        row["rules"][(bug.get("attribution") or {}).get("rule") or "unknown"] += 1
+        if bug.get("reporter"):
+            row["reporters"][bug["reporter"]] += 1
+        row["affiliations"][
+            bug.get("reporter_affiliation", "unknown")] += 1
+        if bug.get("reported_date"):
+            row["dates"].append(bug["reported_date"])
+
+    # Per-year breakdowns, for the page each year gets. Built the same way as
+    # the per-system ones and for the same reason: a bar on a chart raises the
+    # question of what is inside it, and the chart has no room to answer.
+    per_year: Dict[str, dict] = defaultdict(
+        lambda: {"systems": Counter(), "statuses": Counter(),
+                 "techniques": Counter(), "symptoms": Counter(),
+                 "rules": Counter(), "reporters": Counter(),
+                 "affiliations": Counter(), "dates": [], "rejected": 0})
+    for bug in all_bugs:
+        year = bug.get("reported_year")
+        if not year:
+            continue
+        row = per_year[str(year)]
+        if not bug.get("status_is_true_positive"):
+            row["rejected"] += 1
+            continue
+        row["systems"][bug["dbms"]] += 1
+        row["statuses"][bug.get("status") or "unknown"] += 1
+        row["techniques"][bug.get("technique") or "unattributed"] += 1
+        row["symptoms"][bug.get("symptom") or "unknown"] += 1
+        row["rules"][(bug.get("attribution") or {}).get("rule") or "unknown"] += 1
+        if bug.get("reporter"):
+            row["reporters"][bug["reporter"]] += 1
+        row["affiliations"][bug.get("reporter_affiliation", "unknown")] += 1
+        if bug.get("reported_date"):
+            row["dates"].append(bug["reported_date"])
+
+    year_rows = []
+    for year in sorted(per_year, key=lambda k: int(k)):
+        found = per_year[year]
+        dates = sorted(found["dates"])
+        year_rows.append({
+            "year": int(year),
+            "bugs": sum(found["statuses"].values()),
+            "bugs_rejected": found["rejected"],
+            "systems": len(found["systems"]),
+            "by_dbms": _series(found["systems"], dbms_labels, by_count=True),
+            "by_status": _series(found["statuses"], STATUS_LABELS,
+                                 order=list(STATUS_LABELS)),
+            "by_technique": _series(found["techniques"], technique_labels,
+                                    by_count=True),
+            "by_symptom": _series(found["symptoms"], SYMPTOM_LABELS,
+                                  order=list(SYMPTOM_LABELS)),
+            "by_attribution_rule": _series(found["rules"], ATTRIBUTION_LABELS,
+                                           by_count=True),
+            "by_affiliation": _series(found["affiliations"], AFFILIATION_LABELS,
+                                      order=list(AFFILIATION_LABELS)),
+            "top_reporters": _series(found["reporters"], by_count=True)[:8],
+            "first_reported": dates[0] if dates else None,
+            "last_reported": dates[-1] if dates else None,
+        })
+
+    rows = []
+    for dbms_id in sorted(set(registry) | set(by_dbms) | set(adoption_by_dbms)
+                          | set(planned_by_dbms),
+                          key=lambda k: (-by_dbms.get(k, 0),
+                                         registry.get(k, {}).get("name", k).lower())):
+        entry = registry.get(dbms_id, {})
+        rows.append({
+            "id": dbms_id,
+            "name": entry.get("name", dbms_id),
+            "url": entry.get("url"),
+            "supported": dbms_id in supported_ids,
+            "bugs": by_dbms.get(dbms_id, 0),
+            "adoption_relationships": sorted(
+                set(adoption_by_dbms.get(dbms_id, []))),
+            # A proposal is kept out of the use column entirely: an open issue
+            # asking for SQLancer is not the project using it, and listing the
+            # two together in one cell reads as though it were.
+            "planned_adoption": (dbms_id in planned_by_dbms
+                                 and dbms_id not in adoption_by_dbms),
+        })
+        found = detail.get(dbms_id)
+        if found:
+            dates = sorted(found["dates"])
+            rows[-1].update({
+                "repository": entry.get("repository"),
+                "bugs_rejected": found["rejected"],
+                "bugs_by_year": _series(found["years"],
+                                        {"unknown": "Year not recorded"}),
+                "bugs_by_technique": _series(
+                    found["techniques"], technique_labels, by_count=True),
+                "bugs_by_status": _series(found["statuses"], STATUS_LABELS,
+                                          order=list(STATUS_LABELS)),
+                "bugs_by_symptom": _series(found["symptoms"], SYMPTOM_LABELS,
+                                           order=list(SYMPTOM_LABELS)),
+                "bugs_by_attribution_rule": _series(
+                    found["rules"], ATTRIBUTION_LABELS, by_count=True),
+                "bugs_by_affiliation": _series(
+                    found["affiliations"], AFFILIATION_LABELS,
+                    order=list(AFFILIATION_LABELS)),
+                "top_reporters": _series(found["reporters"],
+                                         by_count=True)[:8],
+                "first_reported": dates[0] if dates else None,
+                "last_reported": dates[-1] if dates else None,
+            })
+
+    total_bugs = len(bugs)
+    stats = {
+        "schema_version": SCHEMA_VERSION,
+        "generated_from": {
+            name: content_hash(_canonical(data[name]))
+            for name in ("bugs", "papers", "adoption", "resources", "dbms",
+                         "techniques")
+        },
+        "headline": {
+            "bugs_total": total_bugs,
+            "bugs_total_rounded": round_down_headline(total_bugs),
+            "bugs_found_externally": by_affiliation.get("external", 0),
+            "dbms_supported": len(supported_ids),
+            "dbms_with_bugs": len(by_dbms),
+            "dbms_projects_using_sqlancer": len(adoption_by_dbms),
+            # The measurable part of the undercount: projects whose own
+            # evidence shows them running SQLancer, and whose bugs never reach
+            # this dataset because no public report ties one to it.
+            "dbms_using_sqlancer_without_counted_bugs": len(
+                [dbms_id for dbms_id in adoption_by_dbms
+                 if not by_dbms.get(dbms_id)]),
+            "dbms_projects_planning_adoption": len(
+                set(planned_by_dbms) - set(adoption_by_dbms)),
+            "papers_building_on_sqlancer": len(building_on),
+            "papers_reusing_or_extending_sqlancer": len(reusing_or_extending),
+            "papers_comparing_against_sqlancer": len(comparing),
+            "papers_citing_sqlancer": len(external_papers),
+            "papers_calling_sqlancer_state_of_the_art": sum(
+                1 for p in external_papers
+                if holds(p, "describes_as_state_of_the_art")),
+        },
+        "bugs": {
+            "total": total_bugs,
+            "total_including_rejected": len(all_bugs),
+            "status_known_count": sum(1 for bug in bugs
+                                      if bug["status"] != "unknown"),
+            "by_dbms": _series(by_dbms, dbms_labels, by_count=True),
+            "by_year": _series(by_year),
+            "years": year_rows,
+            "by_status": _series(by_status, STATUS_LABELS,
+                                 order=list(STATUS_LABELS)),
+            "by_technique": _series(by_technique, technique_labels, by_count=True),
+            "by_finder": _series(by_finder, finder_labels, by_count=True),
+            "by_symptom": _series(by_symptom, SYMPTOM_LABELS,
+                                  order=list(SYMPTOM_LABELS)),
+            "by_attribution_rule": _series(
+                by_rule, ATTRIBUTION_LABELS, order=list(ATTRIBUTION_LABELS)),
+            "by_reporter_affiliation": _series(
+                by_affiliation, AFFILIATION_LABELS,
+                order=list(AFFILIATION_LABELS)),
+            "by_dbms_and_affiliation": [
+                {
+                    "key": key,
+                    "label": dbms_labels.get(key, key),
+                    "count": sum(dbms_affiliation[key].values()),
+                    "segments": [
+                        {"key": affiliation,
+                         "label": AFFILIATION_LABELS[affiliation],
+                         "count": dbms_affiliation[key].get(affiliation, 0)}
+                        for affiliation in AFFILIATION_LABELS
+                    ],
+                }
+                for key in [row["key"] for row in _series(by_dbms, by_count=True)]
+            ],
+        },
+        "papers": {
+            "total": len(papers),
+            "external_total": len(external_papers),
+            "building_on_total": len(building_on),
+            "by_relationship": _series(
+                paper_by_relationship, RELATIONSHIP_LABELS,
+                order=list(REPORTED_RELATIONSHIPS)),
+            "by_year": _series(papers_by_year),
+            "by_relationship_year": by_relationship_year,
+            "by_technique": _series(
+                technique_mentions,
+                {tid: tax.display_name(tid) for tid in tax.techniques},
+                by_count=True),
+            "years": paper_years,
+        },
+        "dbms": {
+            "supported": len(supported_ids),
+            "with_bugs": len(by_dbms),
+            "with_adoption": len(adoption_by_dbms),
+            "planning_adoption": len(set(planned_by_dbms) - set(adoption_by_dbms)),
+            "adoption_by_relationship": _series(
+                Counter(entry["relationship"] for entry in adoption
+                        if entry["relationship"] != PLANNED),
+                ADOPTION_LABELS, order=list(ADOPTION_LABELS)),
+            # The systems behind the headline count, so the page can name them
+            # rather than assert a number and leave the reader to trust it.
+            "using_without_bugs": [
+                {"id": row["id"], "name": row["name"]} for row in rows
+                if row.get("adoption_relationships") and not row.get("bugs")],
+            "rows": rows,
+        },
+        "resources": {
+            "total": len(resources),
+            "by_type": _series(Counter(r["type"] for r in resources),
+                               RESOURCE_LABELS, by_count=True),
+        },
+    }
+    return stats
+
+
+def _canonical(payload: dict) -> str:
+    import json
+    return json.dumps(payload, sort_keys=True, ensure_ascii=False,
+                      separators=(",", ":"))
+
+
+def load_data() -> Dict[str, dict]:
+    return {name: config.load_json(config.DATA_FILES[name])
+            for name in ("techniques", "dbms", "bugs", "papers", "adoption",
+                         "resources")}
+
+
+def regenerate() -> Tuple[dict, bool]:
+    """Recompute stats and write them; returns ``(stats, changed_on_disk)``."""
+    stats = compute(load_data())
+    changed = config.write_json(config.DATA_FILES["stats"], stats)
+    return stats, changed
+
+
+def main() -> int:
+    stats, changed = regenerate()
+    head = stats["headline"]
+    print(f"stats.json: {'updated' if changed else 'unchanged'}")
+    print(f"  {head['bugs_total']} bugs ({head['bugs_total_rounded']}) | "
+          f"{head['dbms_supported']} DBMSs supported | "
+          f"{head['dbms_projects_using_sqlancer']} DBMS projects using SQLancer | "
+          f"{head['papers_building_on_sqlancer']} papers building on SQLancer | "
+          f"{head['papers_citing_sqlancer']} papers citing SQLancer")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/talks.py b/tools/impact/talks.py
new file mode 100644
index 0000000..1c0ca7c
--- /dev/null
+++ b/tools/impact/talks.py
@@ -0,0 +1,679 @@
+"""Talks in which someone stands up and discusses SQLancer.
+
+A recorded talk is evidence no written source gives: a researcher naming
+SQLancer as the example of a fuzzer that works, a database developer walking an
+audience through how they run it, a keynote quoting SQLite's page about the
+project's author. None of that is in the page's HTML -- it is in the audio --
+so a talk cannot be collected the way a blog post can.
+
+A talk can be read three ways, and they are not equally trustworthy:
+
+* **Slides.** The speaker's own words, in text, fetchable like any other page.
+  Where a deck exists this is the best source there is, and its sentences are
+  quotable exactly as a blog post's are.
+* **Automatic captions.** A machine's guess at speech. In the talks collected so
+  far they render "SQLancer" as "SQL answer" and as "SQL lenser" -- which is
+  also why ``ASR_VARIANTS`` exists, and why matching here is deliberately more
+  permissive than anywhere else in the pipeline. An excerpt taken from captions
+  is verbatim with respect to the caption track and *not* with respect to the
+  speaker, so it carries the misheard form and a deep link to the second it was
+  said. The link is the real evidence: a reader can listen.
+* **A person who watched it.** No quote, but a note about what was said and
+  where. Weakest, and honest about being so.
+
+Captions are not fetched here. YouTube serves its caption tracks only to its own
+player, so getting one means opening the talk in a browser, opening the
+transcript panel, and reading the segments out of the page -- a manual step,
+whose result ``findings_from_segments`` turns into mentions. Slides are fetched:
+``findings_from_slides`` does it. Either way ``merge`` files the result, and the
+talks file is curated data kept across runs: a run that cannot see a source
+today must never delete what a person captured yesterday.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+import sys
+from typing import Dict, Iterable, List, Optional, Sequence, Tuple
+
+from . import config, taxonomy
+from .util import now, stable_digest
+
+MODULE_VERSION = "1.0.0"
+
+TALKS_FILE = config.DATA_DIR / "talks.json"
+
+# Where a hand-captured transcript is kept. Uncommitted, like every other cache:
+# what belongs in the dataset is the handful of sentences that mention
+# SQLancer, not a full transcription of somebody else's talk.
+TRANSCRIPT_CACHE = config.CACHE_DIR / "talk-transcripts"
+
+# Frames captured from a talk at the moment SQLancer comes up. Committed,
+# because for some talks the mention is on a slide and never spoken: no
+# transcript can reach it, and the frame is the only evidence there is.
+#
+# Capturing one is a browser job, and the trap is quality rather than aim: a
+# player drops to a small rendition whenever it seeks and only upgrades a few
+# seconds later, so a frame grabbed the moment the seek lands is a blur where a
+# slide's table should be. Pin the quality first
+# (``setPlaybackQualityRange('hd1080', 'hd1080')`` on ``#movie_player``), let it
+# play for a second or two past the mark, then pause, then capture -- and read
+# the slide back off the result before filing it.
+FRAME_DIR = config.REPO_ROOT / "assets" / "images" / "impact" / "talks"
+FRAME_URL_BASE = "/assets/images/impact/talks/"
+FRAME_MAX_WIDTH = 960
+FRAME_QUALITY = 82
+
+YOUTUBE_ID = re.compile(
+    r"(?:youtube\.com/watch\?(?:.*&)?v=|youtu\.be/)([\w-]{11})", re.IGNORECASE)
+
+# How transcribers hear "SQLancer". Every one of these was observed in a real
+# caption track; they are not guesses.
+ASR_VARIANTS = re.compile(
+    r"\bsql\s*[- ]?\s*(?:lancer|lenser|lanza|answer|ancer|dancer|launcher)\b"
+    r"|\bsequel\s*[- ]?\s*(?:lancer|lenser|answer)\b"
+    r"|\bsqlancer\b",
+    re.IGNORECASE)
+
+# The project's author, as a talk is as likely to name the person as the tool.
+# The surname alone is enough: transcribers spell the first name as "Manual" as
+# often as "Manuel", and a talk about database testing that says "Rigger" means
+# this one. The word boundary keeps "trigger" out, which is the only word that
+# would otherwise collide.
+AUTHOR = re.compile(r"\bri(?:gg|g|eg)er(?:'s)?\b|\bmanuel\s+rigor\b",
+                    re.IGNORECASE)
+
+# Spelled-out technique names. Acronyms are left to the taxonomy, which knows
+# which of them are ambiguous.
+TECHNIQUE_PHRASES: Tuple[Tuple[str, re.Pattern], ...] = (
+    ("pqs", re.compile(r"\bpivoted query synthesis\b", re.I)),
+    ("norec", re.compile(r"\bnon-?optimi[sz]ing reference engine\b", re.I)),
+    ("tlp", re.compile(r"\bternary logic partitioning\b", re.I)),
+    ("qpg", re.compile(r"\bquery plan guidance\b", re.I)),
+    ("cert", re.compile(r"\bcardinality estimation restriction\b", re.I)),
+    ("dqp", re.compile(r"\bdifferential query plans?\b", re.I)),
+    ("coddtest", re.compile(r"\bconstant[- ]optimi[sz]ation[- ]driven\b", re.I)),
+)
+
+SENTENCE = re.compile(r"[^.!?]+[.!?]+|\S[^.!?]*$")
+
+# Seconds of silence that end a sentence regardless of punctuation. Captions
+# arrive without full stops, so without this every segment in a file runs
+# together -- and a file holding windows from two ends of a talk would read as
+# one continuous remark, which is exactly the sort of thing this module exists
+# to prevent.
+GAP_SECONDS = 30
+
+# How much of a run to keep around a mention. Unpunctuated speech has no
+# sentence to stop at, so the excerpt is a window instead: enough to see what
+# was being said, not so much that it becomes a transcript.
+EXCERPT_RADIUS = 340
+
+# Below this, a matched sentence is extended with the one after it. Speech
+# introduces a name in one breath and says what it is for in the next -- "then
+# Rigger came up with this idea." on its own is a name-drop, and the sentence
+# following it is the evidence.
+SHORT_SENTENCE = 220
+
+
+def video_id(url: str) -> Optional[str]:
+    match = YOUTUBE_ID.search(url or "")
+    return match.group(1) if match else None
+
+
+def talk_id(url: str) -> str:
+    found = video_id(url)
+    return f"talk:youtube:{found}" if found else f"talk:{stable_digest(url)}"
+
+
+def timestamp_of(seconds: int) -> str:
+    """``25:09``, or ``1:00:36`` once a talk passes an hour."""
+    hours, rest = divmod(int(seconds), 3600)
+    minutes, secs = divmod(rest, 60)
+    if hours:
+        return f"{hours}:{minutes:02d}:{secs:02d}"
+    return f"{minutes}:{secs:02d}"
+
+
+def deep_link(url: str, seconds: int) -> str:
+    """The talk's URL pointing at the moment a mention was made."""
+    base = url.split("&t=")[0].split("#")[0]
+    joiner = "&" if "?" in base else "?"
+    return f"{base}{joiner}t={int(seconds)}s"
+
+
+def _sentences(segments: Sequence[dict]) -> List[dict]:
+    """Caption segments rejoined into sentences that keep their timestamps.
+
+    Captions break on line length rather than on grammar, so a mention often
+    straddles two segments and neither one reads as a sentence on its own.
+    """
+    runs: List[List[dict]] = []
+    for segment in segments:
+        body = re.sub(r"\s+", " ", str(segment.get("text") or "")).strip()
+        if not body:
+            continue
+        start = int(segment.get("start_ms") or 0) // 1000
+        if runs and start - runs[-1][-1]["end"] <= GAP_SECONDS:
+            runs[-1].append({"text": body, "start": start, "end": start})
+        else:
+            runs.append([{"text": body, "start": start, "end": start}])
+
+    out = []
+    for run in runs:
+        text, owner = [], []
+        for piece in run:
+            if text:
+                text.append(" ")
+                owner.append(piece["start"])
+            text.append(piece["text"])
+            owner.extend([piece["start"]] * len(piece["text"]))
+        joined = "".join(text)
+        for match in SENTENCE.finditer(joined):
+            body = match.group(0).strip()
+            if body:
+                out.append({"text": body, "offsets": owner,
+                            "start": match.start(),
+                            "at_seconds": owner[match.start()]})
+    return out
+
+
+def _matches(sentence: str, tax
+             ) -> Tuple[List[str], Optional[str], List[str], int]:
+    """``(matched, heard_as, technique_ids, position)`` for one sentence."""
+    matched: List[str] = []
+    techniques: List[str] = []
+    heard_as = None
+    position = -1
+
+    name = ASR_VARIANTS.search(sentence)
+    if name:
+        position = name.start()
+        matched.append("sqlancer")
+        surface = name.group(0)
+        if surface.replace(" ", "").lower() not in ("sqlancer", "sqlancer"):
+            heard_as = surface
+    author = AUTHOR.search(sentence)
+    if author:
+        matched.append("author")
+        position = author.start() if position < 0 else position
+    for technique_id, pattern in TECHNIQUE_PHRASES:
+        found = pattern.search(sentence)
+        if found:
+            techniques.append(technique_id)
+            position = found.start() if position < 0 else position
+    # Acronyms only where the taxonomy says they are unambiguous, and only in a
+    # sentence that is already about databases -- "TLP" and "CERT" are common
+    # words elsewhere.
+    for found in tax.find_techniques(sentence):
+        if getattr(found, "ambiguous_acronym", False):
+            continue
+        if found.technique_id not in techniques:
+            techniques.append(found.technique_id)
+    matched.extend(t for t in techniques if t not in matched)
+    return matched, heard_as, techniques, max(position, 0)
+
+
+def _window(text: str, position: int) -> Tuple[str, int]:
+    """``EXCERPT_RADIUS`` characters either side of a mention, on word bounds."""
+    if len(text) <= 2 * EXCERPT_RADIUS:
+        return text, position
+    start = max(0, position - EXCERPT_RADIUS)
+    end = min(len(text), position + EXCERPT_RADIUS)
+    if start:
+        space = text.find(" ", start)
+        start = space + 1 if 0 <= space < position else start
+    if end < len(text):
+        space = text.rfind(" ", position, end)
+        end = space if space > position else end
+    return text[start:end].strip(), position - start
+
+
+def findings_from_segments(segments: Sequence[dict], *, url: str,
+                           tax=None) -> List[dict]:
+    """Mentions found in a captured transcript, in the order they were said."""
+    tax = tax or taxonomy.load()
+    mentions: List[dict] = []
+    sentences = _sentences(segments)
+    for index, sentence in enumerate(sentences):
+        matched, heard_as, techniques, position = _matches(sentence["text"], tax)
+        if not matched:
+            continue
+        text = sentence["text"]
+        following = sentences[index + 1] if index + 1 < len(sentences) else None
+        # Same run means the two were spoken together, with no gap of silence
+        # between them; ``_sentences`` gives every sentence in a run the same
+        # offsets list.
+        if (len(text) < SHORT_SENTENCE and following is not None
+                and following["offsets"] is sentence["offsets"]):
+            text = f"{text} {following['text']}"
+        excerpt, position = _window(text, position)
+        # The timestamp follows the excerpt: a long unpunctuated run can span
+        # minutes, and a link to its first word would not land on the mention.
+        offsets = sentence["offsets"]
+        index = min(sentence["start"] + position, len(offsets) - 1)
+        at = int(offsets[index]) if offsets else int(sentence["at_seconds"])
+        mentions.append({
+            "id": f"M{len(mentions) + 1}",
+            "source": "captions",
+            "at_seconds": at,
+            "timestamp": timestamp_of(at),
+            "url": deep_link(url, at),
+            "matched": matched,
+            "heard_as": heard_as,
+            "excerpt": excerpt,
+            "excerpt_is_verbatim": True,
+            "technique_ids": techniques,
+        })
+    return mentions
+
+
+# Slides punctuate loosely -- a heading, a dash, then bullets with no full
+# stop -- so the excerpt runs to the next sentence end or to the end of the
+# deck's text, and an em dash is content rather than a boundary.
+SLIDE_SENTENCE = re.compile(r"[^.!?]{0,240}\bSQLancer\b[^.!?]{0,300}(?:[.!?]|$)")
+
+
+def findings_from_slides(text: str, *, url: str, tax=None,
+                         start_at: int = 0) -> List[dict]:
+    """Mentions read off a slide deck's own text.
+
+    Slides are written, not spoken, so the name is spelled correctly and the
+    excerpt is quotable as it stands. They carry no clock, so a mention points
+    at the deck rather than at a moment.
+    """
+    tax = tax or taxonomy.load()
+    mentions: List[dict] = []
+    seen = set()
+    for match in SLIDE_SENTENCE.finditer(text or ""):
+        excerpt = re.sub(r"\s+", " ", match.group(0)).strip()
+        if not excerpt or excerpt in seen:
+            continue
+        seen.add(excerpt)
+        matched, _, techniques, _position = _matches(excerpt, tax)
+        if "sqlancer" not in matched:
+            continue
+        mentions.append({
+            "id": f"M{start_at + len(mentions) + 1}",
+            "source": "slides",
+            "at_seconds": None,
+            "timestamp": None,
+            "url": url,
+            "matched": matched,
+            "heard_as": None,
+            "excerpt": excerpt[:2000],
+            "excerpt_is_verbatim": True,
+            "technique_ids": techniques,
+        })
+    return mentions
+
+
+def frame_name(key: str, seconds: Optional[int] = None) -> str:
+    """``<video id>-<second>.jpg`` for a recording, ``<key>.jpg`` for a deck."""
+    return f"{key}-{int(seconds)}.jpg" if seconds is not None else f"{key}.jpg"
+
+
+def store_frame(source: str, key: str, seconds: Optional[int] = None,
+                trim_bottom: int = 0) -> str:
+    """File a captured frame, trimmed and shrunk; returns its site path.
+
+    A browser screenshot arrives as a full-width PNG with the page's own
+    letterboxing around the video. Both are cut back here so what lands in the
+    repository is the frame itself, at a size that suits a page rather than a
+    display.
+    """
+    from PIL import Image, ImageChops
+
+    FRAME_DIR.mkdir(parents=True, exist_ok=True)
+    image = Image.open(source).convert("RGB")
+    # A capture of the player often catches the page's own title line just
+    # below it; ``trim_bottom`` cuts that off before anything else.
+    if trim_bottom:
+        image = image.crop((0, 0, image.width, image.height - trim_bottom))
+    # Trim the black letterboxing: a difference against a black image leaves
+    # only the picture, and its bounding box is where the frame really starts.
+    background = Image.new("RGB", image.size, (0, 0, 0))
+    box = ImageChops.difference(image, background).getbbox()
+    if box:
+        image = image.crop(box)
+    if image.width > FRAME_MAX_WIDTH:
+        height = round(image.height * FRAME_MAX_WIDTH / image.width)
+        image = image.resize((FRAME_MAX_WIDTH, height), Image.LANCZOS)
+    name = frame_name(key, seconds)
+    image.save(FRAME_DIR / name, "JPEG", quality=FRAME_QUALITY, optimize=True)
+    return f"{FRAME_URL_BASE}{name}"
+
+
+def frame_for(key: Optional[str], seconds: Optional[int] = None
+              ) -> Optional[str]:
+    """The site path of a stored frame for this moment, if one was captured."""
+    if not key:
+        return None
+    name = frame_name(key, seconds)
+    return f"{FRAME_URL_BASE}{name}" if (FRAME_DIR / name).exists() else None
+
+
+# ---------------------------------------------------------------------------
+# Getting a transcript in by hand
+#
+# YouTube serves its caption tracks only to its own player, and for some talks
+# it will not fill the transcript panel at all. Both dead ends are worked
+# around the same way: somebody gets the text out however they can -- the
+# panel's own copy button, yt-dlp, a downloaded .vtt -- and hands it here. The
+# parsers below take whichever shape it arrives in, because the shape is not
+# something the person supplying it should have to care about.
+# ---------------------------------------------------------------------------
+
+VTT_TIME = re.compile(
+    r"(\d{1,2}):(\d{2}):(\d{2})[.,](\d{3})\s*-->\s*"
+    r"(?:\d{1,2}):(?:\d{2}):(?:\d{2})[.,]\d{3}")
+VTT_SHORT_TIME = re.compile(
+    r"(\d{1,2}):(\d{2})[.,](\d{3})\s*-->\s*(?:\d{1,2}):(?:\d{2})[.,]\d{3}")
+
+# Auto-captions carry per-word timing inside the cue, and a cue number and
+# position line around it. None of that is speech.
+CUE_MARKUP = re.compile(r"<[^>]*>")
+CUE_SETTINGS = re.compile(r"\s(align|position|line|size):\S+")
+
+# A transcript copied out of the panel is "12:34" followed by the line, either
+# on one line or two.
+PANEL_LINE = re.compile(r"^\s*((?:\d{1,2}:)?\d{1,2}:\d{2})\s*(.*)$")
+
+
+def _seconds(*parts: str) -> int:
+    values = [int(p) for p in parts]
+    while len(values) < 3:
+        values.insert(0, 0)
+    return values[0] * 3600 + values[1] * 60 + values[2]
+
+
+def segments_from_cues(text: str) -> List[dict]:
+    """Segments from a WebVTT or SubRip file.
+
+    Auto-generated captions repeat themselves: each cue restates the tail of
+    the one before it so the words can roll up the screen. Emitting that as
+    written would trip the matcher into reporting the same sentence several
+    times, so a line already carried by the previous cue is dropped.
+    """
+    segments: List[dict] = []
+    previous: List[str] = []
+    for block in re.split(r"\n\s*\n", text.replace("\r\n", "\n")):
+        match = VTT_TIME.search(block) or VTT_SHORT_TIME.search(block)
+        if not match:
+            continue
+        groups = [g for g in match.groups() if g is not None]
+        start = (_seconds(*groups[:3]) if len(groups) >= 4
+                 else _seconds(*groups[:2]))
+        lines = []
+        for line in block.split("\n"):
+            if VTT_TIME.search(line) or VTT_SHORT_TIME.search(line):
+                continue
+            line = CUE_SETTINGS.sub(" ", CUE_MARKUP.sub("", line))
+            # Collapse spacing before comparing: stripping the per-word timing
+            # tags leaves gaps where they were, and a line that differs from
+            # the one before only by those gaps is the same line.
+            line = re.sub(r"\s+", " ", line).strip()
+            if line and not line.isdigit() and line.upper() != "WEBVTT":
+                lines.append(line)
+        fresh = [line for line in lines if line not in previous]
+        previous = lines
+        body = re.sub(r"\s+", " ", " ".join(fresh)).strip()
+        if body:
+            segments.append({"start_ms": start * 1000, "text": body})
+    return segments
+
+
+def segments_from_panel(text: str) -> List[dict]:
+    """Segments from a transcript copied out of YouTube's own panel."""
+    segments: List[dict] = []
+    pending: Optional[int] = None
+    for raw in text.replace("\r\n", "\n").split("\n"):
+        match = PANEL_LINE.match(raw)
+        if match:
+            start = _seconds(*match.group(1).split(":"))
+            body = match.group(2).strip()
+            if body:
+                segments.append({"start_ms": start * 1000, "text": body})
+                pending = None
+            else:
+                pending = start
+        elif raw.strip() and pending is not None:
+            segments.append({"start_ms": pending * 1000,
+                             "text": re.sub(r"\s+", " ", raw).strip()})
+            pending = None
+        elif raw.strip() and segments:
+            segments[-1]["text"] += " " + re.sub(r"\s+", " ", raw).strip()
+    return segments
+
+
+def parse_transcript(text: str) -> List[dict]:
+    """Segments from whichever of the accepted shapes ``text`` is in."""
+    stripped = text.lstrip()
+    if stripped.startswith("{"):
+        return json.loads(text).get("segments", [])
+    if VTT_TIME.search(text) or VTT_SHORT_TIME.search(text):
+        return segments_from_cues(text)
+    return segments_from_panel(text)
+
+
+def windows_around(segments: Sequence[dict], mentions: Sequence[dict],
+                   radius: int = 2) -> List[dict]:
+    """The segments near a mention, and only those.
+
+    What belongs in the repository is the handful of sentences that mention
+    SQLancer, not a transcription of somebody else's whole talk. A couple of
+    segments either side is enough to see that the matcher read it right, and
+    to catch the thought the speaker was in the middle of -- the name often
+    lands a sentence before the point being made.
+    """
+    wanted = {int(m["at_seconds"]) for m in mentions}
+    keep = set()
+    for index, segment in enumerate(segments):
+        if int(segment.get("start_ms") or 0) // 1000 in wanted:
+            keep.update(range(max(0, index - radius),
+                              min(len(segments), index + radius + 1)))
+    return [segments[i] for i in sorted(keep)]
+
+
+# How near a frame has to be to count as covering a moment. Slides change on
+# their own schedule, and a capture a few seconds either side of a sentence is
+# still the slide that was up while it was said.
+FRAME_NEARBY_SECONDS = 20
+
+
+def frames_wanted(talk: dict) -> List[dict]:
+    """Moments in one talk that want a frame captured, and why.
+
+    Not every mention needs a picture. Two do, and both are cases where the
+    words on their own are weak:
+
+    * the transcriber misheard the name, so the excerpt says "SQL lenser" and
+      the slide is where it is spelled correctly;
+    * nothing quotable was said at all, which is what a slide-only mention
+      looks like from the transcript's side.
+
+    A moment somebody looked at and found nothing on screen for carries
+    ``frame_checked`` and is not asked for again.
+
+    A talk whose transcript was read and mentions nothing is the strongest case
+    of the second kind: the mention exists, and only a frame can show it.
+    """
+    have = [m.get("at_seconds") for m in talk.get("mentions") or []
+            if m.get("source") == "frame" and m.get("at_seconds") is not None]
+    wanted: List[dict] = []
+    for mention in talk.get("mentions") or []:
+        at = mention.get("at_seconds")
+        if mention.get("source") == "frame" or at is None:
+            continue
+        if any(abs(at - other) <= FRAME_NEARBY_SECONDS for other in have):
+            continue
+        if mention.get("frame_checked"):
+            continue
+        if mention.get("heard_as"):
+            why = (f"the transcript misheard the name as "
+                   f"{mention['heard_as']!r} here")
+        elif not mention.get("excerpt"):
+            why = "nothing quotable was said here"
+        else:
+            continue
+        wanted.append({
+            "at_seconds": at,
+            "timestamp": mention.get("timestamp") or timestamp_of(at),
+            "url": mention.get("url"),
+            "why": why,
+            "mention_id": mention.get("id"),
+        })
+
+    read_and_silent = any(
+        source.get("kind") == "captions" and source.get("status") == "extracted"
+        and "not spoken" in (source.get("note") or "")
+        for source in talk.get("sources") or [])
+    if read_and_silent and not have:
+        wanted.append({
+            "at_seconds": None, "timestamp": None, "url": talk["url"],
+            "why": ("the transcript was read and says nothing: the moment has "
+                    "to come from watching it"),
+            "mention_id": None,
+        })
+    return wanted
+
+
+def load() -> dict:
+    if TALKS_FILE.exists():
+        return config.load_json(TALKS_FILE)
+    return {"schema_version": "1.0.0", "talks": []}
+
+
+def save(payload: dict) -> bool:
+    payload["talks"] = sorted(
+        payload.get("talks", []),
+        key=lambda t: (-(t.get("year") or 0), (t.get("title") or "").lower(),
+                       t["id"]))
+    return config.write_json(TALKS_FILE, payload)
+
+
+def merge(entry: dict, *, timestamp: Optional[str] = None) -> str:
+    """File one talk, keeping anything a person recorded about it before.
+
+    Returns ``"added"`` or ``"updated"``.
+    """
+    timestamp = timestamp or now()
+    payload = load()
+    talks = payload.setdefault("talks", [])
+    existing = next((t for t in talks if t["id"] == entry["id"]), None)
+    entry.setdefault("provenance", {})
+    entry["provenance"].update({
+        "collector": "talks",
+        "collector_version": MODULE_VERSION,
+        "source_url": entry["url"],
+        "source_type": "video",
+        "last_verified": timestamp,
+    })
+    if existing is None:
+        entry["provenance"]["first_seen"] = timestamp
+        talks.append(entry)
+        outcome = "added"
+    else:
+        entry["provenance"]["first_seen"] = (
+            existing.get("provenance", {}).get("first_seen") or timestamp)
+        # A run with no transcript in hand must not erase the mentions of one
+        # that was captured earlier.
+        if not entry.get("mentions") and existing.get("mentions"):
+            entry["mentions"] = existing["mentions"]
+            entry["sources"] = existing.get("sources", entry["sources"])
+        if entry.get("relationship") is None and existing.get("relationship"):
+            entry["relationship"] = existing["relationship"]
+        talks[talks.index(existing)] = entry
+        outcome = "updated"
+    save(payload)
+    return outcome
+
+
+def main(argv: Optional[Sequence[str]] = None) -> int:
+    """``list`` what is on file, or ``import`` a transcript somebody got out.
+
+        python3 -m tools.impact.talks list
+        python3 -m tools.impact.talks frames
+        python3 -m tools.impact.talks import <file> <talk url>
+
+    The file may be a .vtt or .srt, a transcript copied out of YouTube's panel,
+    or the JSON shape this module writes. Only the windows around the mentions
+    are kept; run ``collect --only talks`` afterwards to fold them in.
+    """
+    argv = list(argv if argv is not None else sys.argv[1:])
+    command = argv[0] if argv else "list"
+
+    if command == "list":
+        for talk in load()["talks"]:
+            status = "/".join(f"{s['kind']}:{s['status']}"
+                              for s in talk["sources"])
+            roles = ", ".join((talk.get("relationship") or {}).get("roles", []))
+            print(f"{talk['id']}  [{status}]  {talk['title'][:70]}")
+            print(f"    {len(talk['mentions'])} mentions"
+                  f"{'  roles: ' + roles if roles else ''}")
+        return 0
+
+    if command == "frames":
+        for talk in load()["talks"]:
+            have = [m for m in talk["mentions"] if m["source"] == "frame"]
+            wanted = frames_wanted(talk)
+            if not have and not wanted:
+                continue
+            print(f"{talk['title'][:70]}")
+            for mention in have:
+                print(f"    have  {mention['timestamp']:>8}  "
+                      f"{mention['image'].split('/')[-1]}")
+            for entry in wanted:
+                print(f"    want  {entry['timestamp'] or '       ?':>8}  "
+                      f"{entry['why']}")
+                print(f"                    {entry['url']}")
+        print("\nCapturing one: assets/images/impact/talks/README.md")
+        return 0
+
+    if command == "import" and len(argv) > 2:
+        path, url = argv[1], argv[2]
+        video = video_id(url)
+        if not video:
+            print(f"not a YouTube url: {url}", file=sys.stderr)
+            return 1
+        with open(path, encoding="utf-8", errors="replace") as handle:
+            segments = parse_transcript(handle.read())
+        if not segments:
+            print(f"no captions found in {path}", file=sys.stderr)
+            return 1
+        mentions = findings_from_segments(segments, url=url)
+        kept = windows_around(segments, mentions)
+        TRANSCRIPT_CACHE.mkdir(parents=True, exist_ok=True)
+        target = TRANSCRIPT_CACHE / f"{video}.json"
+        with target.open("w", encoding="utf-8") as handle:
+            json.dump({
+                "video_id": video,
+                "url": deep_link(url, 0).split("?t=")[0].split("&t=")[0],
+                "source": "supplied_transcript",
+                "kind": "auto_captions",
+                "retrieved_at": now(),
+                "total_segments": len(segments),
+                "total_characters": sum(len(s.get("text") or "")
+                                        for s in segments),
+                "note": ("Windows around the moments SQLancer is mentioned, cut "
+                         "from a transcript supplied by hand. Not the whole "
+                         "talk."),
+                "segments": kept,
+            }, handle, indent=2, sort_keys=True)
+            handle.write("\n")
+        print(f"{len(segments)} segments read, {len(mentions)} mention(s):")
+        for mention in mentions:
+            heard = f" (heard as {mention['heard_as']!r})" if mention["heard_as"] else ""
+            print(f"  {mention['timestamp']:>8}  {', '.join(mention['matched'])}{heard}")
+            print(f"            {mention['excerpt'][:100]}")
+        print(f"wrote {target}")
+        print("next: python3 -m tools.impact.run collect --only talks")
+        return 0
+
+    print(main.__doc__)
+    return 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/taxonomy.py b/tools/impact/taxonomy.py
new file mode 100644
index 0000000..83edcfa
--- /dev/null
+++ b/tools/impact/taxonomy.py
@@ -0,0 +1,295 @@
+"""Reading and applying the technique taxonomy.
+
+Collectors never hardcode technique names. They ask this module, which loads
+``_data/impact/techniques.json`` and knows how to match a name in free text --
+including the part that matters most in practice: refusing to treat a bare
+``TLP`` or ``PQS`` as a SQLancer reference unless the surrounding text is
+actually about database testing.
+"""
+
+from __future__ import annotations
+
+import re
+from functools import lru_cache
+from typing import Dict, Iterable, List, Optional, Sequence
+
+from . import config
+
+
+class TechniqueMatch:
+    __slots__ = ("technique_id", "matched_name", "start", "end", "context",
+                 "needs_context", "context_terms_found")
+
+    def __init__(self, technique_id, matched_name, start, end, context,
+                 needs_context, context_terms_found):
+        self.technique_id = technique_id
+        self.matched_name = matched_name
+        self.start = start
+        self.end = end
+        self.context = context
+        self.needs_context = needs_context
+        self.context_terms_found = context_terms_found
+
+    def __repr__(self):
+        return (f"TechniqueMatch({self.technique_id!r}, {self.matched_name!r}, "
+                f"context_terms={self.context_terms_found!r})")
+
+
+class Taxonomy:
+    """Loaded view over ``techniques.json`` with matching helpers."""
+
+    # Window of characters around a hit inspected for corroborating context.
+    CONTEXT_WINDOW = 400
+
+    def __init__(self, data: dict):
+        self.data = data
+        self.version = data["taxonomy_version"]
+        self.techniques = {t["id"]: t for t in data["sqlancer_techniques"]}
+        self.excluded = {t["id"]: t for t in data.get("excluded_techniques", [])}
+        # Papers the project states are its own, beyond the technique and tool
+        # papers already listed above.
+        self.project_publications = list(data.get("project_publications", []))
+        self.project_authors = list(data.get("project_authors", []))
+        self.project_contributors = list(data.get("project_contributors", []))
+        self._project_author_pattern = self._compile_author_names(
+            self.project_authors)
+        self.finders = {f["id"]: f for f in data["finders"]}
+        self.symptoms = {s["id"] for s in data.get("bug_symptoms", [])}
+        self._patterns = self._compile(self.techniques)
+        self._excluded_patterns = self._compile(self.excluded)
+        # One combined pattern for the finders, because their names are nested:
+        # matching "SQLancer" and "SQLancer++" independently would report both
+        # for a text that only mentions the latter.
+        self._finder_pattern, self._finder_by_name = self._compile_owned(self.finders)
+
+    # -- construction -----------------------------------------------------
+
+    @staticmethod
+    def _compile_names(names: Sequence[str]):
+        # Longest first so "Ternary Logic Partitioning" wins over "TLP" and
+        # "SQLancer++" wins over "SQLancer".
+        ordered = sorted(set(names), key=len, reverse=True)
+        alternatives = "|".join(re.escape(n) for n in ordered)
+        # \b does not fire next to '+', so SQLancer++ needs an explicit boundary.
+        return re.compile(rf"(?<![\w+])({alternatives})(?![\w])", re.IGNORECASE)
+
+    def _compile(self, entries: Dict[str, dict]):
+        return {
+            key: self._compile_names(entry["names"])
+            for key, entry in entries.items()
+        }
+
+    @staticmethod
+    def _compile_author_names(names: Sequence[str]):
+        """Pattern for matching a surname inside an author name.
+
+        Stricter at the boundary than term matching: a hyphen ends a name here,
+        so a double-barrelled surname is a different person, whereas in prose
+        "TLP-based" is still a reference to TLP.
+        """
+        if not names:
+            return None
+        ordered = sorted(set(names), key=len, reverse=True)
+        alternatives = "|".join(re.escape(name) for name in ordered)
+        return re.compile(rf"(?<![\w'-])({alternatives})(?![\w'-])",
+                          re.IGNORECASE)
+
+    def _compile_owned(self, entries: Dict[str, dict]):
+        """Combined pattern over all entries plus a name -> owning id map."""
+        owner = {}
+        for key, entry in entries.items():
+            for name in entry["names"]:
+                owner[name.lower()] = key
+        return self._compile_names(list(owner)), owner
+
+    # -- accessors --------------------------------------------------------
+
+    def technique_ids(self) -> List[str]:
+        return list(self.techniques)
+
+    def umbrella_finder_ids(self) -> List[str]:
+        return [f["id"] for f in self.finders.values() if f["umbrella_member"]]
+
+    def is_known_technique(self, technique_id: Optional[str]) -> bool:
+        return technique_id is None or technique_id in self.techniques
+
+    def display_name(self, technique_id: Optional[str]) -> str:
+        if technique_id is None:
+            return "Not recorded"
+        entry = self.techniques.get(technique_id)
+        return entry["display_name"] if entry else technique_id
+
+    def finder_display_name(self, finder_id: str) -> str:
+        entry = self.finders.get(finder_id)
+        return entry["display_name"] if entry else finder_id
+
+    def seed_ids(self) -> List[str]:
+        """Every id a citation seed can be recorded under.
+
+        A seed is usually a technique, but a tool's own paper is a seed too --
+        SQLancer++'s is -- and it is recorded under the tool's id. The two id
+        spaces do not overlap.
+        """
+        return [seed["seed_id"] for seed in self.citation_seeds()
+                if seed.get("seed_id")]
+
+    def citation_seeds(self) -> List[dict]:
+        """Foundational publications whose citations are crawled for papers.
+
+        A tool's own paper can be a seed as an oracle's is. SQLancer++ is
+        cited by work that cites no oracle paper, and walking only the oracle
+        papers cannot reach it.
+        """
+        seeds = []
+        for technique in self.techniques.values():
+            paper = technique.get("origin_paper")
+            if paper and paper.get("is_citation_seed"):
+                seeds.append({"technique_id": technique["id"],
+                              "seed_id": technique["id"], **paper})
+        for finder in self.finders.values():
+            paper = finder.get("paper")
+            if paper and paper.get("is_citation_seed"):
+                # No technique: citing SQLancer++'s paper says a work cites
+                # the tool, not that it cites an oracle. seed_id identifies
+                # the seed either way, and the id spaces do not overlap.
+                seeds.append({"technique_id": None, "finder_id": finder["id"],
+                              "seed_id": finder["id"], **paper})
+        return seeds
+
+    def search_terms(self) -> List[str]:
+        """Every name worth issuing a discovery query for."""
+        terms: List[str] = []
+        for finder in self.finders.values():
+            terms.extend(finder["names"])
+        for technique in self.techniques.values():
+            terms.extend(technique["names"])
+        seen, ordered = set(), []
+        for term in terms:
+            lowered = term.lower()
+            if lowered not in seen:
+                seen.add(lowered)
+                ordered.append(term)
+        return ordered
+
+    # -- matching ---------------------------------------------------------
+
+    def mentions_finder(self, text: str) -> List[str]:
+        """Umbrella tool names explicitly present in ``text``.
+
+        Because the combined pattern prefers the longest alternative, a text
+        that says only "SQLancer++" reports ``sqlancer_pp`` and not ``sqlancer``.
+        """
+        if not text:
+            return []
+        found: List[str] = []
+        for hit in self._finder_pattern.finditer(text):
+            finder_id = self._finder_by_name.get(hit.group(1).lower())
+            if finder_id and finder_id not in found:
+                found.append(finder_id)
+        return found
+
+    def find_techniques(self, text: str, *, require_context: bool = True
+                        ) -> List[TechniqueMatch]:
+        """Locate SQLancer technique names in ``text``.
+
+        Ambiguous acronyms only produce a match when a corroborating
+        database-testing term appears nearby and no negative term does. This is
+        what keeps ``TLP`` in a Linux power-management thread out of the data.
+        """
+        if not text:
+            return []
+        matches: List[TechniqueMatch] = []
+        lowered = text.lower()
+        for technique_id, pattern in self._patterns.items():
+            technique = self.techniques[technique_id]
+            ambiguous = technique.get("ambiguous_acronym", False)
+            required = [t.lower() for t in technique.get("required_context_terms", [])]
+            negative = [t.lower() for t in technique.get("negative_context_terms", [])]
+            for hit in pattern.finditer(text):
+                start = max(0, hit.start() - self.CONTEXT_WINDOW)
+                end = min(len(text), hit.end() + self.CONTEXT_WINDOW)
+                window = lowered[start:end]
+                # A full spelled-out name is unambiguous on its own.
+                spelled_out = len(hit.group(1)) > 6
+                if any(term in window for term in negative):
+                    continue
+                found_terms = [term for term in required if term in window]
+                if (ambiguous and require_context and not spelled_out
+                        and not found_terms):
+                    continue
+                matches.append(TechniqueMatch(
+                    technique_id=technique_id,
+                    matched_name=hit.group(1),
+                    start=hit.start(),
+                    end=hit.end(),
+                    context=text[start:end],
+                    needs_context=ambiguous and not spelled_out,
+                    context_terms_found=found_terms,
+                ))
+        matches.sort(key=lambda m: m.start)
+        return matches
+
+    def is_project_author(self, name: Optional[str]) -> bool:
+        """Whether an author name belongs to the SQLancer project.
+
+        Matched as a whole word so that initialled forms ("M. Rigger") and the
+        full name both hit, without a substring matching some longer surname.
+        """
+        if not name or self._project_author_pattern is None:
+            return False
+        return bool(self._project_author_pattern.search(name))
+
+    def is_project_contributor(self, who: Optional[str],
+                               extra_handles: Optional[Sequence[str]] = None
+                               ) -> bool:
+        """Whether a bug reporter belongs to the project rather than adopting it.
+
+        Matched on the display name or the GitHub handle, since the collectors
+        record whichever the source gave. ``extra_handles`` carries the TEST
+        lab's current members, which are read from its site rather than listed
+        here so the roster does not go stale.
+        """
+        if not who:
+            return False
+        needle = who.strip().lower()
+        if not needle:
+            return False
+        if self.is_project_author(who):
+            return True
+        for entry in self.project_contributors:
+            if needle == (entry.get("name") or "").strip().lower():
+                return True
+            if needle == (entry.get("github") or "").strip().lower():
+                return True
+        return any(needle == (handle or "").strip().lower()
+                   for handle in (extra_handles or []))
+
+    def find_excluded_techniques(self, text: str) -> List[str]:
+        """Out-of-scope techniques named in ``text`` (currently: EET)."""
+        if not text:
+            return []
+        return [key for key, pattern in self._excluded_patterns.items()
+                if pattern.search(text)]
+
+    def technique_from_oracle_label(self, label: Optional[str]) -> Optional[str]:
+        """Map an upstream oracle label such as ``TLP (WHERE)`` to a technique id.
+
+        Labels like ``error``, ``crash`` and ``hang`` describe how the bug
+        manifested rather than which technique found it, so they map to no
+        technique and are handled as symptoms instead.
+        """
+        if not label:
+            return None
+        for match in self.find_techniques(label, require_context=False):
+            return match.technique_id
+        return None
+
+
+@lru_cache(maxsize=1)
+def load() -> Taxonomy:
+    return Taxonomy(config.load_json(config.DATA_FILES["techniques"]))
+
+
+def reload() -> Taxonomy:
+    load.cache_clear()
+    return load()
diff --git a/tools/impact/tests/__init__.py b/tools/impact/tests/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/tools/impact/tests/test_data.py b/tools/impact/tests/test_data.py
new file mode 100644
index 0000000..e6552bc
--- /dev/null
+++ b/tools/impact/tests/test_data.py
@@ -0,0 +1,446 @@
+"""Checks on the published dataset itself.
+
+These run against the real files in ``_data/impact`` and are what stands between
+a bad automated proposal and the website. They need no network and no API key.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+import unittest
+from collections import Counter
+
+from tools.impact import config, stats as stats_module, taxonomy, validate
+from tools.impact.dataset import LIST_FIELD
+from tools.impact.util import normalize_doi, normalize_url
+
+
+def load(name: str) -> dict:
+    return config.load_json(config.DATA_FILES[name])
+
+
+def records(name: str):
+    return load(name).get(LIST_FIELD[name], [])
+
+
+class SchemaTest(unittest.TestCase):
+    def test_every_file_validates(self):
+        problems = validate.validate_all()
+        self.assertEqual([], [str(p) for p in problems])
+
+    def test_schemas_are_valid_json_schema(self):
+        import jsonschema
+        for path in config.SCHEMA_DIR.glob("*.schema.json"):
+            with open(path, encoding="utf-8") as handle:
+                schema = json.load(handle)
+            jsonschema.Draft7Validator.check_schema(schema)
+
+
+class IdentityTest(unittest.TestCase):
+    def test_ids_are_unique_within_each_file(self):
+        for name in LIST_FIELD:
+            counts = Counter(record["id"] for record in records(name))
+            duplicates = [key for key, count in counts.items() if count > 1]
+            self.assertEqual([], duplicates, f"duplicate ids in {name}.json")
+
+    def test_every_record_has_a_stable_id(self):
+        pattern = re.compile(r"^[a-z0-9]+(:[A-Za-z0-9._+#/@-]+)+$")
+        for name in LIST_FIELD:
+            for record in records(name):
+                if name == "dbms":
+                    continue  # registry ids are bare slugs by design
+                self.assertRegex(record["id"], pattern,
+                                 f"{name}.json record id is not namespaced")
+
+    def test_no_duplicate_bug_urls_with_the_same_title(self):
+        seen = {}
+        for record in records("bugs"):
+            url = normalize_url(record.get("primary_url"))
+            if not url:
+                continue
+            key = (url, record["title"])
+            self.assertNotIn(key, seen,
+                             f"duplicate bug {url} / {record['title']}")
+            seen[key] = record["id"]
+
+    def test_no_two_records_are_the_same_work(self):
+        """A preprint and its published paper are one work, not two.
+
+        Both citation graphs index them separately, and nothing downstream
+        noticed: the DBMS fuzzing survey was counted twice and AMOEBA three
+        times before this was checked.
+        """
+        from tools.impact import dedupe
+
+        groups, review = dedupe.find_duplicates(records("papers"))
+        self.assertEqual(
+            [], [[p["id"] for p in group] for group in groups],
+            "run `python3 -m tools.impact.run dedupe` to merge these")
+        self.assertEqual(
+            [], [(a["id"], b["id"]) for a, b, _ in review],
+            "decide these in _data/impact/paper_decisions.json")
+
+    def test_an_authorless_stub_sharing_a_tool_name_is_raised(self):
+        """Nothing else can compare a record that carries no authors.
+
+        The 2022 DBStorm record had none, so authorship could not link it to the
+        2024 paper it was an earlier version of, and the download list kept
+        asking for a paper already in hand.
+        """
+        from tools.impact import dedupe
+
+        papers = [
+            {"id": "paper:doi:10.1145/x", "title": "DBStorm: Generating Various "
+             "Effective Workloads for Testing Isolation Levels",
+             "authors": ["Keqiang Li", "Siyang Weng"], "relationships": {}},
+            {"id": "paper:s2:abc", "title": "DBStorm: A Cost-effective Approach "
+             "for Generating Valid Workload", "authors": [], "relationships": {}},
+        ]
+        _, review = dedupe.find_duplicates(papers)
+        self.assertEqual(1, len(review), "the shared tool name should be raised")
+
+    def test_a_merged_record_keeps_the_identities_it_replaced(self):
+        """An old id must stay resolvable after a merge."""
+        seen = set()
+        for record in records("papers"):
+            for alias in record.get("also_indexed_as") or []:
+                self.assertNotIn(alias, seen, f"{alias} claimed twice")
+                seen.add(alias)
+            self.assertNotIn(record["id"], record.get("also_indexed_as") or [])
+
+    def test_no_duplicate_dois(self):
+        seen = {}
+        for record in records("papers"):
+            doi = normalize_doi(record.get("doi"))
+            if not doi:
+                continue
+            self.assertNotIn(doi, seen, f"duplicate DOI {doi}")
+            seen[doi] = record["id"]
+
+    def test_no_duplicate_resource_urls(self):
+        seen = set()
+        for record in records("resources"):
+            url = normalize_url(record["url"])
+            self.assertNotIn(url, seen, f"duplicate resource {url}")
+            seen.add(url)
+
+
+class EvidenceTest(unittest.TestCase):
+    URL = re.compile(r"^https?://[^\s<>\"]+$")
+
+    def _evidence(self, record):
+        found = []
+
+        def walk(node):
+            if isinstance(node, dict):
+                if "source_url" in node and "source_type" in node:
+                    found.append(node)
+                for value in node.values():
+                    walk(value)
+            elif isinstance(node, list):
+                for item in node:
+                    walk(item)
+
+        walk(record)
+        return found
+
+    def test_every_bug_has_evidence_and_a_link(self):
+        for record in records("bugs"):
+            self.assertTrue(record.get("links"),
+                            f"{record['id']} has no links")
+            self.assertTrue(record["attribution"]["evidence"],
+                            f"{record['id']} has no attribution evidence")
+
+    def test_every_adoption_record_has_evidence(self):
+        for record in records("adoption"):
+            self.assertTrue(record.get("evidence"),
+                            f"{record['id']} has no evidence")
+
+    def test_every_positive_paper_relationship_has_evidence(self):
+        for record in records("papers"):
+            for key, entry in record["relationships"].items():
+                if entry.get("value") == "yes":
+                    self.assertTrue(
+                        entry.get("evidence"),
+                        f"{record['id']} claims {key} without evidence")
+
+    def test_evidence_urls_are_well_formed(self):
+        for name in LIST_FIELD:
+            for record in records(name):
+                for item in self._evidence(record):
+                    self.assertRegex(item["source_url"], self.URL,
+                                     f"{name}.json {record['id']}")
+
+    def test_excerpts_are_flagged_verbatim(self):
+        """An excerpt is source text; the flag records that we checked."""
+        for name in LIST_FIELD:
+            for record in records(name):
+                for item in self._evidence(record):
+                    if item.get("excerpt") is not None:
+                        self.assertTrue(
+                            item.get("excerpt_is_verbatim"),
+                            f"{name}.json {record['id']} excerpt not marked verbatim")
+                        self.assertTrue(item["excerpt"].strip())
+
+
+class VocabularyTest(unittest.TestCase):
+    def setUp(self):
+        self.tax = taxonomy.load()
+        self.registry = {entry["id"] for entry in records("dbms")}
+
+    def test_bug_techniques_are_known(self):
+        for record in records("bugs"):
+            technique = record.get("technique")
+            self.assertTrue(self.tax.is_known_technique(technique),
+                            f"{record['id']} references unknown {technique!r}")
+            self.assertNotIn(technique, self.tax.excluded,
+                             f"{record['id']} uses an excluded technique")
+
+    def test_bug_finders_are_umbrella_members(self):
+        for record in records("bugs"):
+            finder = record["finder"]
+            self.assertIn(finder, self.tax.finders)
+            self.assertTrue(self.tax.finders[finder]["umbrella_member"])
+
+    def test_all_dbms_references_resolve(self):
+        for name in ("bugs", "adoption"):
+            for record in records(name):
+                self.assertIn(record["dbms"], self.registry,
+                              f"{name}.json {record['id']}")
+
+    def test_paper_relationship_techniques_are_known(self):
+        """The three judged relationships name techniques.
+
+        "references" is different: it records which seed publication the paper
+        cites, and a seed can be a tool's own paper rather than an oracle's.
+        SQLancer++'s is, and papers citing it cite no technique at all.
+        """
+        seeds = set(self.tax.seed_ids())
+        for record in records("papers"):
+            for key, entry in record["relationships"].items():
+                allowed = (set(self.tax.techniques) | seeds
+                           if key == "references" else set(self.tax.techniques))
+                for technique in entry.get("techniques", []):
+                    self.assertIn(technique, allowed, f"{record['id']}/{key}")
+
+    def test_adoption_relationships_use_the_controlled_vocabulary(self):
+        allowed = {"official_ci", "official_testing", "developer_use",
+                   "integration_contributed_by_dbms_team", "planned_adoption"}
+        for record in records("adoption"):
+            self.assertIn(record["relationship"], allowed)
+
+    def test_a_proposal_is_never_counted_as_use(self):
+        """An open proposal is an intention; counting it would overstate reach."""
+        stats = load("stats")
+        using = {record["dbms"] for record in records("adoption")
+                 if record["relationship"] != "planned_adoption"}
+        self.assertEqual(len(using),
+                         stats["headline"]["dbms_projects_using_sqlancer"])
+        planned = {record["dbms"] for record in records("adoption")
+                   if record["relationship"] == "planned_adoption"}
+        self.assertEqual(len(planned - using),
+                         stats["headline"]["dbms_projects_planning_adoption"])
+
+    def test_a_proposal_never_reaches_the_use_column(self):
+        """The systems table must not present an open issue as adoption."""
+        planned = {record["dbms"] for record in records("adoption")
+                   if record["relationship"] == "planned_adoption"}
+        using = {record["dbms"] for record in records("adoption")
+                 if record["relationship"] != "planned_adoption"}
+        for row in load("stats")["dbms"]["rows"]:
+            self.assertNotIn("planned_adoption", row["adoption_relationships"],
+                             f"{row['id']} lists a proposal as use")
+            self.assertEqual(row["id"] in planned and row["id"] not in using,
+                             row["planned_adoption"], row["id"])
+
+    def test_a_campaign_reporter_is_never_counted_as_external(self):
+        """The same person is a handle in one tracker and a name in another.
+
+        Matching only handles marked every SQLite forum report as somebody
+        else's work, though the same people filed them.
+
+        A ruling in ``people_decisions.json`` is the one thing that overrides
+        this: appearing on the lab's people page says someone is in the lab,
+        not that their work is this project's, and only a person can tell the
+        two apart.
+        """
+        from tools.impact.collectors import people
+
+        known = {name.strip().lower() for name in people.identities()}
+        ruled_external = people.counted_as_external()
+        for record in records("bugs"):
+            reporter = (record.get("reporter") or "").strip().lower()
+            if not reporter or reporter not in known:
+                continue
+            expected = "external" if reporter in ruled_external else "project"
+            self.assertEqual(expected, record["reporter_affiliation"],
+                             f"{record['id']}: {record['reporter']} is on the "
+                             f"roster and {'is' if reporter in ruled_external else 'is not'} "
+                             f"ruled external")
+
+    def test_excluded_techniques_carry_a_reason_and_evidence(self):
+        for entry in self.tax.excluded.values():
+            self.assertTrue(entry["reason"])
+            self.assertTrue(entry.get("evidence"))
+
+
+class PolicyTest(unittest.TestCase):
+    def test_policy_version_matches_the_data_files(self):
+        policy_version = load("policy")["policy_version"]
+        for name in ("bugs", "papers", "adoption"):
+            self.assertEqual(policy_version, load(name)["policy_version"],
+                             f"{name}.json is on a different policy version")
+
+    def test_adoption_is_not_inferred_from_support(self):
+        """Supporting a DBMS must never be recorded as that project using it."""
+        supported = {entry["id"] for entry in records("dbms")
+                     if entry.get("supported_by_sqlancer")}
+        for record in records("adoption"):
+            evidence_urls = " ".join(item["source_url"]
+                                     for item in record["evidence"])
+            self.assertNotIn("github.com/sqlancer/sqlancer", evidence_urls,
+                             f"{record['id']} rests on SQLancer's own repository")
+            if record["dbms"] in supported:
+                self.assertTrue(record["evidence"])
+
+    def test_every_collected_paper_at_least_references_sqlancer(self):
+        for record in records("papers"):
+            self.assertEqual("yes",
+                             record["relationships"]["references"]["value"])
+
+
+class ReviewQueueTest(unittest.TestCase):
+    """The queue is the record of what did not get in, so it is checked too."""
+
+    def queue(self, **changes) -> dict:
+        base = {"schema_version": "1.0.0", "description": "x",
+                "open": [{"id": "review:aaa", "kind": "bug", "source": "bugs",
+                          "url": None, "title": None, "reason": "unplaced",
+                          "first_seen": "2026-01-01T00:00:00Z",
+                          "last_seen": "2026-01-02T00:00:00Z"}],
+                "dismissed": [{"id": "review:bbb", "url": None,
+                               "reason": "not a database system",
+                               "decided_by": "Manuel Rigger",
+                               "decided_on": "2026-01-01"}]}
+        base.update(changes)
+        return base
+
+    def test_a_well_formed_queue_passes(self):
+        self.assertEqual([], validate.check_review_queue(self.queue()))
+
+    def test_an_item_cannot_be_dismissed_and_still_open(self):
+        queue = self.queue()
+        queue["open"][0]["id"] = queue["dismissed"][0]["id"]
+        problems = validate.check_review_queue(queue)
+        self.assertTrue(any("also in open" in str(p) for p in problems), problems)
+
+    def test_a_dismissal_must_say_who_made_it(self):
+        queue = self.queue()
+        queue["dismissed"][0].pop("decided_by")
+        problems = validate.check_review_queue(queue)
+        self.assertTrue(any("decided_by" in str(p) for p in problems), problems)
+
+class StatsTest(unittest.TestCase):
+    def setUp(self):
+        self.stats = load("stats")
+        self.recomputed = stats_module.compute(stats_module.load_data())
+
+    def test_stats_file_matches_the_records(self):
+        """The published figures must be exactly what the records produce."""
+        self.assertEqual(self.recomputed, self.stats,
+                         "stats.json is stale; run `python -m tools.impact.run stats`")
+
+    def test_bug_total_counts_only_accepted_reports(self):
+        accepted = [r for r in records("bugs") if r["status_is_true_positive"]]
+        self.assertEqual(len(accepted), self.stats["headline"]["bugs_total"])
+        self.assertEqual(len(records("bugs")),
+                         self.stats["bugs"]["total_including_rejected"])
+
+    def test_by_dbms_sums_to_the_total(self):
+        total = sum(row["count"] for row in self.stats["bugs"]["by_dbms"])
+        self.assertEqual(self.stats["bugs"]["total"], total)
+
+    def test_headline_rounding_never_overstates(self):
+        headline = self.stats["headline"]
+        digits = headline["bugs_total_rounded"].rstrip("+").replace(",", "")
+        self.assertLessEqual(int(digits), headline["bugs_total"])
+
+    def test_papers_building_on_is_deduplicated(self):
+        """A paper in several categories must be counted once."""
+        building = 0
+        per_category = 0
+        for record in records("papers"):
+            if record.get("is_sqlancer_publication"):
+                continue
+            hits = [key for key in ("uses_infrastructure", "extends_technique",
+                                    "compares_with")
+                    if record["relationships"][key]["value"] == "yes"]
+            per_category += len(hits)
+            if hits:
+                building += 1
+        self.assertEqual(building,
+                         self.stats["headline"]["papers_building_on_sqlancer"])
+        self.assertLessEqual(building, max(per_category, building))
+
+    def test_supported_and_bug_counts_are_separate_concepts(self):
+        supported = {entry["id"] for entry in records("dbms")
+                     if entry.get("supported_by_sqlancer")}
+        with_bugs = {record["dbms"] for record in records("bugs")
+                     if record["status_is_true_positive"]}
+        self.assertEqual(len(supported), self.stats["dbms"]["supported"])
+        self.assertEqual(len(with_bugs), self.stats["dbms"]["with_bugs"])
+        adopting = {record["dbms"] for record in records("adoption")
+                    if record["relationship"] != "planned_adoption"}
+        self.assertEqual(len(adopting), self.stats["dbms"]["with_adoption"])
+
+
+if __name__ == "__main__":
+    unittest.main()
+
+
+class UndercountTest(unittest.TestCase):
+    """The bug count is a floor, and the page has to be able to say so.
+
+    The figure behind that claim is derived rather than written: projects whose
+    own evidence shows them running SQLancer, and whose bugs never reach this
+    dataset because no public report ties one to it. Materialize is the clearest
+    case -- it maintains its own SQLancer fork and contributes no bug at all.
+    """
+
+    def test_the_blind_spot_is_measured_not_asserted(self):
+        from tools.impact import config
+        stats = config.load_json(config.DATA_FILES["stats"])
+        rows = {row["id"]: row for row in stats["dbms"]["rows"]}
+        silent = [row for row in rows.values()
+                  if row.get("adoption_relationships") and not row.get("bugs")]
+        self.assertEqual(
+            len(silent),
+            stats["headline"]["dbms_using_sqlancer_without_counted_bugs"])
+
+    def test_the_named_systems_are_the_measured_ones(self):
+        """The page names them, so the names have to come from the same rows."""
+        from tools.impact import config
+        stats = config.load_json(config.DATA_FILES["stats"])
+        named = stats["dbms"]["using_without_bugs"]
+        expected = [{"id": row["id"], "name": row["name"]}
+                    for row in stats["dbms"]["rows"]
+                    if row.get("adoption_relationships") and not row.get("bugs")]
+        self.assertEqual(expected, named)
+        self.assertEqual(
+            len(named),
+            stats["headline"]["dbms_using_sqlancer_without_counted_bugs"])
+
+    def test_each_named_system_has_a_page_to_link_to(self):
+        from tools.impact import config, pages
+        stats = config.load_json(config.DATA_FILES["stats"])
+        for row in stats["dbms"]["using_without_bugs"]:
+            self.assertTrue((pages.PAGE_DIR / f"{row['id']}.html").exists(),
+                            row["id"])
+
+    def test_a_project_running_it_can_still_have_no_bugs(self):
+        from tools.impact import config
+        stats = config.load_json(config.DATA_FILES["stats"])
+        row = next(r for r in stats["dbms"]["rows"] if r["id"] == "materialize")
+        self.assertTrue(row["adoption_relationships"])
+        self.assertFalse(row.get("bugs"))
diff --git a/tools/impact/tests/test_pipeline.py b/tools/impact/tests/test_pipeline.py
new file mode 100644
index 0000000..aa4487f
--- /dev/null
+++ b/tools/impact/tests/test_pipeline.py
@@ -0,0 +1,3037 @@
+"""Behavioural tests for the pipeline itself.
+
+These use temporary directories and stub sources rather than the network, so
+they run in CI in under a second. They cover the properties the weekly run
+depends on: repeated collection is idempotent, an empty cache changes results
+not at all, unchanged evidence never reaches the model twice, and the classifier
+cannot introduce an excerpt that is not in the source.
+"""
+
+from __future__ import annotations
+
+import json
+import shutil
+import tempfile
+import unittest
+from pathlib import Path
+
+from tools.impact import config, plots, stats as stats_module, taxonomy
+from tools.impact.cache import Caches, ClassificationCache, DerivedCache, RawCache, State
+from tools.impact.classify import Classifier
+from tools.impact.collectors import sqlancer_bugs
+from tools.impact.collectors.adoption import _best_excerpt
+from tools.impact.collectors import nus_test
+from tools.impact.collectors.github_bugs import deterministic_attribution
+from tools.impact.collectors.papers import (Candidate, build_record,
+                                            is_project_publication,
+                                            looks_like_sqlancer_publication,
+                                            normalise_openalex, paper_identity)
+from tools.impact.dataset import Dataset, substantively_equal
+from tools.impact.util import verbatim_excerpt
+
+
+class FakeRawFile:
+    """Minimal stand-in for the GitHub client used by the curated importer."""
+
+    def __init__(self, files):
+        self.files = files
+        self.reads = 0
+
+    def raw_file(self, owner, repo, path, ref="HEAD", **kwargs):
+        self.reads += 1
+        return self.files.get(path), {"status": 200}
+
+
+CURATED = [
+    {
+        "date": "28/5/2019", "dbms": "SQLite", "oracle": "NoREC",
+        "reporter": "Manuel Rigger", "status": "fixed",
+        "title": "Wrong result for a simple filter",
+        "links": {"bugreport": "https://sqlite.org/forum/forumpost/aaaa"},
+        "test": ["CREATE TABLE t0(c0);"],
+    },
+    {
+        "date": "01/6/2020", "dbms": "DuckDB", "oracle": "crash",
+        "reporter": "Manuel Rigger", "status": "closed (not a bug)",
+        "title": "Not actually a bug",
+        "links": {"bugreport": "https://github.com/duckdb/duckdb/issues/1"},
+        "test": ["SELECT 1;"],
+    },
+]
+
+README = ("This project aims to provide a basis to study bugs in DBMS. To this "
+          "end, the repository stores a list of bugs found by SQLancer with "
+          "additional metadata.\n")
+
+
+def fake_github():
+    return FakeRawFile({
+        "bugs.json": json.dumps(CURATED, indent=4) + "\n",
+        "README.md": README,
+    })
+
+
+class CuratedImportTest(unittest.TestCase):
+    def setUp(self):
+        self.records = sqlancer_bugs.collect(fake_github(),
+                                             timestamp="2026-01-01T00:00:00Z")
+
+    def test_imports_every_entry(self):
+        self.assertEqual(2, len(self.records))
+
+    def test_technique_comes_from_the_taxonomy_not_a_hardcoded_map(self):
+        by_title = {r["title"]: r for r in self.records}
+        self.assertEqual("norec",
+                         by_title["Wrong result for a simple filter"]["technique"])
+        self.assertIsNone(by_title["Not actually a bug"]["technique"])
+
+    def test_rejected_reports_are_kept_but_not_counted(self):
+        by_title = {r["title"]: r for r in self.records}
+        self.assertFalse(by_title["Not actually a bug"]["status_is_true_positive"])
+        self.assertEqual("closed_not_a_bug", by_title["Not actually a bug"]["status"])
+
+    def test_evidence_excerpts_are_verbatim_in_the_source(self):
+        source = json.dumps(CURATED, indent=4) + "\n"
+        for record in self.records:
+            for item in record["attribution"]["evidence"]:
+                excerpt = item.get("excerpt")
+                if excerpt is None:
+                    continue
+                self.assertTrue(
+                    verbatim_excerpt(source, excerpt)
+                    or verbatim_excerpt(README, excerpt),
+                    f"excerpt not found verbatim: {excerpt!r}")
+
+    def test_per_entry_hashes_are_independent(self):
+        """Adding an unrelated upstream entry must not touch existing records."""
+        extended = CURATED + [{
+            "date": "02/6/2020", "dbms": "DuckDB", "oracle": "TLP (WHERE)",
+            "reporter": "Manuel Rigger", "status": "fixed", "title": "Another",
+            "links": {"bugreport": "https://github.com/duckdb/duckdb/issues/2"},
+            "test": ["SELECT 2;"],
+        }]
+        gh = FakeRawFile({"bugs.json": json.dumps(extended, indent=4) + "\n",
+                          "README.md": README})
+        after = sqlancer_bugs.collect(gh, timestamp="2026-01-01T00:00:00Z")
+        before_by_id = {r["id"]: r for r in self.records}
+        for record in after:
+            if record["id"] in before_by_id:
+                self.assertEqual(before_by_id[record["id"]]["provenance"]
+                                 ["content_sha256"],
+                                 record["provenance"]["content_sha256"])
+
+
+class IdempotencyTest(unittest.TestCase):
+    """Collecting twice with no new evidence must change nothing on disk."""
+
+    def setUp(self):
+        self.tmp = Path(tempfile.mkdtemp())
+        self.files = {"bugs": self.tmp / "bugs.json"}
+
+    def tearDown(self):
+        shutil.rmtree(self.tmp, ignore_errors=True)
+
+    def _dataset(self):
+        dataset = Dataset(self.files)
+        dataset.ensure("bugs", {"schema_version": "1.0.0",
+                                "policy_version": "test"})
+        return dataset
+
+    def test_second_run_writes_no_bytes(self):
+        first = sqlancer_bugs.collect(fake_github(),
+                                      timestamp="2026-01-01T00:00:00Z")
+        dataset = self._dataset()
+        dataset.merge("bugs", first, timestamp="2026-01-01T00:00:00Z")
+        self.assertTrue(dataset.save(["bugs"])["bugs"])
+        before = self.files["bugs"].read_text()
+
+        # A later run: same evidence, different wall-clock timestamp.
+        second = sqlancer_bugs.collect(fake_github(),
+                                       timestamp="2026-02-02T00:00:00Z")
+        dataset = self._dataset()
+        report = dataset.merge("bugs", second, timestamp="2026-02-02T00:00:00Z")
+        changed = dataset.save(["bugs"])
+
+        self.assertEqual([], report.added)
+        self.assertEqual([], report.updated)
+        self.assertEqual(2, report.unchanged)
+        self.assertFalse(changed["bugs"])
+        self.assertEqual(before, self.files["bugs"].read_text())
+
+    def test_first_seen_survives_an_update(self):
+        first = sqlancer_bugs.collect(fake_github(),
+                                      timestamp="2026-01-01T00:00:00Z")
+        dataset = self._dataset()
+        dataset.merge("bugs", first, timestamp="2026-01-01T00:00:00Z")
+        dataset.save(["bugs"])
+
+        changed = json.loads(json.dumps(first))
+        changed[0]["status"] = "verified"
+        dataset = self._dataset()
+        dataset.merge("bugs", changed, timestamp="2026-03-03T00:00:00Z")
+        stored = {r["id"]: r for r in dataset.records("bugs")}
+        updated = stored[changed[0]["id"]]
+        self.assertEqual("verified", updated["status"])
+        self.assertEqual("2026-01-01T00:00:00Z",
+                         updated["provenance"]["first_seen"])
+        self.assertEqual("2026-03-03T00:00:00Z",
+                         updated["provenance"]["last_verified"])
+
+    def test_timestamps_alone_are_not_a_change(self):
+        left = {"id": "bug:x:1", "last_verified": "2026-01-01T00:00:00Z", "a": 1}
+        right = {"id": "bug:x:1", "last_verified": "2026-09-09T00:00:00Z", "a": 1}
+        self.assertTrue(substantively_equal(left, right))
+        right["a"] = 2
+        self.assertFalse(substantively_equal(left, right))
+
+
+class CacheTest(unittest.TestCase):
+    def setUp(self):
+        self.tmp = Path(tempfile.mkdtemp())
+
+    def tearDown(self):
+        shutil.rmtree(self.tmp, ignore_errors=True)
+
+    def test_cold_cache_produces_the_same_records(self):
+        """Caching is an optimisation; correctness must not depend on it."""
+        warm = sqlancer_bugs.collect(fake_github(),
+                                     timestamp="2026-01-01T00:00:00Z")
+        cold = sqlancer_bugs.collect(fake_github(),
+                                     timestamp="2026-01-01T00:00:00Z")
+        self.assertEqual(warm, cold)
+
+    def test_derived_cache_only_recomputes_on_a_content_change(self):
+        cache = DerivedCache(self.tmp / "derived")
+        calls = []
+
+        def produce():
+            calls.append(1)
+            return {"value": len(calls)}
+
+        first = cache.compute("src", "sha256:aaa", "v1", produce)
+        second = cache.compute("src", "sha256:aaa", "v1", produce)
+        self.assertEqual(first, second)
+        self.assertEqual(1, len(calls))
+
+        cache.compute("src", "sha256:bbb", "v1", produce)
+        self.assertEqual(2, len(calls))
+        cache.compute("src", "sha256:aaa", "v2", produce)
+        self.assertEqual(3, len(calls))
+
+    def test_raw_cache_reuses_an_unchanged_upstream_timestamp(self):
+        cache = RawCache(self.tmp / "raw")
+        cache.store("issue:1", body="hello", url="https://example.org/1",
+                    upstream_updated_at="2026-01-01T00:00:00Z")
+        self.assertTrue(cache.is_fresh(
+            "issue:1", upstream_updated_at="2026-01-01T00:00:00Z"))
+        self.assertFalse(cache.is_fresh(
+            "issue:1", upstream_updated_at="2026-02-01T00:00:00Z"))
+
+    def test_state_tracks_incremental_and_full_scans(self):
+        state = State(self.tmp / "state.json")
+        self.assertIsNone(state.last_scan("github"))
+        state.record_scan("github")
+        state.save()
+        self.assertIsNotNone(State(self.tmp / "state.json").last_scan("github"))
+        self.assertIsNone(State(self.tmp / "state.json").last_reconcile("github"))
+        state.record_scan("github", full=True)
+        state.save()
+        self.assertIsNotNone(State(self.tmp / "state.json").last_reconcile("github"))
+
+
+class StubClassifierClient:
+    """Records every request so the tests can count real model calls."""
+
+    def __init__(self, payload):
+        self.payload = payload
+        self.requests = []
+
+    class _Block:
+        def __init__(self, text):
+            self.type = "text"
+            self.text = text
+
+    class _Response:
+        def __init__(self, blocks):
+            self.content = blocks
+            self.stop_reason = "end_turn"
+
+    class _Messages:
+        def __init__(self, outer):
+            self.outer = outer
+
+        def create(self, **kwargs):
+            self.outer.requests.append(kwargs)
+            return StubClassifierClient._Response(
+                [StubClassifierClient._Block(json.dumps(self.outer.payload))])
+
+    @property
+    def messages(self):
+        return StubClassifierClient._Messages(self)
+
+
+class ClassifierTest(unittest.TestCase):
+    def setUp(self):
+        self.tmp = Path(tempfile.mkdtemp())
+        self.cache = ClassificationCache(
+            self.tmp / "classifications", model="test-model",
+            policy_version="policy-v1", taxonomy_version="taxonomy-v1")
+
+    def tearDown(self):
+        shutil.rmtree(self.tmp, ignore_errors=True)
+
+    def _classifier(self, payload):
+        """A classifier wired to a stub client rather than the real API.
+
+        Injecting the client is enough: `available` reports True once a client
+        is present and a key is configured, so nothing about the class needs to
+        be monkeypatched.
+        """
+        classifier = Classifier(self.cache, model="test-model", api_key="test")
+        classifier._client = StubClassifierClient(payload)
+        return classifier
+
+    def _classify(self, classifier, text="found by SQLancer while testing"):
+        return classifier.classify(
+            "bug_attribution", source_id="github:x/y#1", source_text=text,
+            source_label="test", technique_names=["NoREC"],
+            finder_names=["SQLancer"], excluded_names=["EET"])
+
+    def test_unchanged_evidence_is_not_sent_to_the_model_twice(self):
+        payload = {"answer": "yes", "finder": "SQLancer", "technique": None,
+                   "excerpts": ["found by SQLancer"], "reason": "states it"}
+        classifier = self._classifier(payload)
+        self.assertTrue(classifier.available)
+        first = self._classify(classifier)
+        second = self._classify(classifier)
+        self.assertIsNotNone(first)
+        self.assertFalse(first.from_cache)
+        self.assertTrue(second.from_cache)
+        self.assertEqual(1, len(classifier._client.requests))
+        self.assertEqual(1, classifier.calls_made)
+        self.assertEqual(1, classifier.cache_hits)
+
+    def test_negative_answers_are_cached_too(self):
+        payload = {"answer": "no", "finder": None, "technique": None,
+                   "excerpts": [], "reason": "unrelated"}
+        classifier = self._classifier(payload)
+        self._classify(classifier)
+        second = self._classify(classifier)
+        self.assertTrue(second.from_cache)
+        self.assertEqual("no", second.answer)
+        self.assertEqual(1, len(classifier._client.requests))
+
+    def test_a_changed_prompt_version_forces_a_fresh_answer(self):
+        key_a = self.cache.key("src", "sha256:aaa", "bug-attribution-v1")
+        key_b = self.cache.key("src", "sha256:aaa", "bug-attribution-v2")
+        self.assertNotEqual(key_a, key_b)
+
+    def test_a_changed_policy_version_forces_a_fresh_answer(self):
+        other = ClassificationCache(self.tmp / "c2", model="test-model",
+                                    policy_version="policy-v2",
+                                    taxonomy_version="taxonomy-v1")
+        self.assertNotEqual(self.cache.key("s", "h", "v"),
+                            other.key("s", "h", "v"))
+
+    def test_an_excerpt_not_in_the_source_is_discarded(self):
+        classifier = Classifier(self.cache, model="test-model", api_key="test")
+        cleaned = classifier._sanitise(
+            {"answer": "yes", "excerpts": ["a sentence that was never there"],
+             "reason": "made up"},
+            "the real source text says something else entirely")
+        self.assertEqual([], cleaned["excerpts"])
+        self.assertEqual("uncertain", cleaned["answer"],
+                         "a positive answer with no usable evidence must not stand")
+
+    def test_a_verbatim_excerpt_is_kept(self):
+        classifier = Classifier(self.cache, model="test-model", api_key="test")
+        cleaned = classifier._sanitise(
+            {"answer": "yes", "excerpts": ["found by SQLancer"], "reason": "ok"},
+            "This bug was found by SQLancer during a testing campaign.")
+        self.assertEqual(["found by SQLancer"], cleaned["excerpts"])
+        self.assertEqual("yes", cleaned["answer"])
+
+    def test_an_unknown_answer_becomes_uncertain(self):
+        classifier = Classifier(self.cache, model="test-model", api_key="test")
+        cleaned = classifier._sanitise(
+            {"answer": "probably", "excerpts": [], "reason": ""}, "text")
+        self.assertEqual("uncertain", cleaned["answer"])
+
+    def test_no_api_key_yields_no_classification_rather_than_a_guess(self):
+        classifier = Classifier(self.cache, model="test-model", api_key=None)
+        self.assertFalse(classifier.available)
+        self.assertIsNone(self._classify(classifier))
+        self.assertEqual(1, classifier.stats()["skipped_no_key"])
+
+
+class AttributionRuleTest(unittest.TestCase):
+    def setUp(self):
+        self.tax = taxonomy.load()
+
+    def test_a_discovery_statement_is_accepted(self):
+        result = deterministic_attribution(
+            "This wrong result was found by SQLancer while testing DuckDB.",
+            self.tax)
+        self.assertIsNotNone(result)
+        self.assertEqual("sqlancer", result[2])
+
+    def test_an_oracle_attribution_is_accepted_without_the_word_sqlancer(self):
+        result = deterministic_attribution(
+            "The NoREC oracle reports a mismatch between the two queries.",
+            self.tax)
+        self.assertIsNotNone(result)
+        self.assertEqual("norec", result[1])
+
+    def test_a_feature_request_mentioning_sqlancer_is_not_a_bug(self):
+        self.assertIsNone(deterministic_attribution(
+            "SQLancer does not support window functions yet, please add them.",
+            self.tax))
+
+    def test_an_excluded_technique_is_never_attributed(self):
+        self.assertIsNone(deterministic_attribution(
+            "This bug was found using EET on the query.", self.tax))
+
+    def test_an_ambiguous_acronym_needs_database_context(self):
+        self.assertIsNone(deterministic_attribution(
+            "Please raise the TLP power management timeout on my laptop.",
+            self.tax))
+        self.assertIsNotNone(deterministic_attribution(
+            "Detected by TLP: the SELECT query returns a wrong result set.",
+            self.tax))
+
+
+class TaxonomyTest(unittest.TestCase):
+    def setUp(self):
+        self.tax = taxonomy.load()
+
+    def test_umbrella_tools_are_distinguished(self):
+        self.assertEqual(["sqlancer_pp"],
+                         self.tax.mentions_finder("we ran SQLancer++"))
+        self.assertEqual(["sqlancer"], self.tax.mentions_finder("we ran SQLancer"))
+        self.assertEqual(["shqvel"], self.tax.mentions_finder("ShQveL synthesised"))
+
+    def test_oracle_labels_map_to_technique_ids(self):
+        self.assertEqual("tlp",
+                         self.tax.technique_from_oracle_label("TLP (aggregate)"))
+        self.assertIsNone(self.tax.technique_from_oracle_label("crash"))
+
+    def test_every_seed_publication_is_reachable(self):
+        seeds = self.tax.citation_seeds()
+        self.assertTrue(seeds)
+        for seed in seeds:
+            self.assertTrue(seed.get("doi") or seed.get("arxiv_id")
+                            or seed.get("s2_paper_id") or seed.get("title"))
+
+
+class PaperIdentityTest(unittest.TestCase):
+    def test_an_arxiv_doi_collapses_onto_the_arxiv_id(self):
+        """The same preprint reached through either index must be one record."""
+        from_openalex, _, _ = paper_identity(
+            {"DOI": "https://doi.org/10.48550/arxiv.2505.02012"}, "fallback")
+        from_s2, _, _ = paper_identity({"ArXiv": "2505.02012"}, "fallback")
+        self.assertEqual(from_openalex, from_s2)
+
+    def test_a_real_doi_wins_over_an_arxiv_id(self):
+        record_id, doi, arxiv = paper_identity(
+            {"DOI": "10.1145/3428279", "ArXiv": "1234.5678"}, "fallback")
+        self.assertEqual("paper:doi:10.1145/3428279", record_id)
+        self.assertEqual("1234.5678", arxiv)
+
+    def test_openalex_works_are_projected_onto_a_common_shape(self):
+        work = normalise_openalex({
+            "id": "https://openalex.org/W123", "doi": "https://doi.org/10.1/x",
+            "display_name": "A paper", "publication_year": 2024,
+            "authorships": [{"author": {"display_name": "A. Person"}}],
+            "primary_location": {"source": {"display_name": "A venue"}},
+        })
+        self.assertEqual("A paper", work["title"])
+        self.assertEqual("W123", work["openalexId"])
+        self.assertEqual("A venue", work["venue"])
+
+    def test_an_unclassified_paper_is_never_counted_as_building_on_sqlancer(self):
+        candidate = Candidate({"paperId": "a" * 40, "title": "T", "year": 2024,
+                               "externalIds": {"DOI": "10.1/y"}})
+        candidate.add_edge("tlp", {"contexts": []}, "semantic_scholar")
+        record = build_record(candidate, timestamp="2026-01-01T00:00:00Z")
+        self.assertEqual("yes", record["relationships"]["references"]["value"])
+        for key in ("uses_infrastructure", "extends_technique", "compares_with"):
+            self.assertEqual("insufficient_evidence",
+                             record["relationships"][key]["value"])
+
+
+class ProjectPublicationTest(unittest.TestCase):
+    """Which papers are the project's own is stated, not inferred."""
+
+    def setUp(self):
+        self.tax = taxonomy.load()
+
+    def _candidate(self, **external):
+        return Candidate({"paperId": None, "openalexId": "W1",
+                          "externalIds": external, "title": "A paper",
+                          "year": 2024})
+
+    def test_a_listed_publication_is_recognised_by_doi(self):
+        for entry in self.tax.project_publications:
+            if not entry.get("doi"):
+                continue
+            self.assertTrue(
+                is_project_publication(self._candidate(DOI=entry["doi"]), self.tax),
+                f"{entry['title']} is listed but was not recognised")
+
+    def test_technique_and_tool_papers_are_still_recognised(self):
+        seeds = self.tax.citation_seeds()
+        checked = 0
+        for seed in seeds:
+            if not seed.get("doi"):
+                continue
+            checked += 1
+            self.assertTrue(
+                is_project_publication(self._candidate(DOI=seed["doi"]), self.tax))
+        self.assertGreater(checked, 0)
+
+    def test_an_unrelated_paper_is_not_a_project_publication(self):
+        self.assertFalse(
+            is_project_publication(self._candidate(DOI="10.1145/9999999"),
+                                   self.tax))
+
+    def test_a_paper_by_a_project_author_is_ours(self):
+        """Co-authorship settles it, whatever the paper is about."""
+        candidate = Candidate({
+            "paperId": None, "openalexId": "W1", "externalIds": {},
+            "title": "Something entirely unrelated to databases", "year": 2025,
+            "authors": [{"name": "Manuel Rigger"}, {"name": "A. Person"}],
+        })
+        self.assertTrue(looks_like_sqlancer_publication(candidate))
+
+    def test_an_initialled_project_author_still_matches(self):
+        candidate = Candidate({
+            "paperId": None, "openalexId": "W2", "externalIds": {},
+            "title": "A paper", "year": 2025,
+            "authors": [{"name": "M. Rigger"}],
+        })
+        self.assertTrue(looks_like_sqlancer_publication(candidate))
+
+    def test_a_longer_surname_is_not_a_false_match(self):
+        for name in ("Riggers", "Triggerman", "Rigger-Smith Industries"):
+            self.assertFalse(self.tax.is_project_author(name), name)
+
+    def test_a_paper_without_a_project_author_stays_external(self):
+        candidate = Candidate({
+            "paperId": None, "openalexId": "W3",
+            "externalIds": {"DOI": "10.1145/9999999"},
+            "title": "An independent paper", "year": 2025,
+            "authors": [{"name": "Someone Else"}],
+        })
+        self.assertFalse(looks_like_sqlancer_publication(candidate))
+
+    def test_project_authors_are_configured_not_hardcoded(self):
+        self.assertTrue(self.tax.project_authors,
+                        "project_authors is empty; the rule would never fire")
+
+    def test_no_published_paper_by_a_project_author_counts_as_external(self):
+        for record in config.load_json(config.DATA_FILES["papers"])["papers"]:
+            if record.get("is_sqlancer_publication"):
+                continue
+            for author in record.get("authors", []):
+                self.assertFalse(
+                    self.tax.is_project_author(author),
+                    f"{record['title']!r} is counted as external work but "
+                    f"{author} is a project author")
+
+    def test_every_listed_publication_carries_a_reason(self):
+        for entry in self.tax.project_publications:
+            self.assertTrue(entry.get("reason", "").strip(),
+                            f"{entry['title']} is listed without a reason")
+
+    def test_listed_publications_are_excluded_from_external_counts(self):
+        records = {p["id"]: p for p in
+                   config.load_json(config.DATA_FILES["papers"])["papers"]}
+        for entry in self.tax.project_publications:
+            if not entry.get("doi"):
+                continue
+            record = records.get(f"paper:doi:{entry['doi'].lower()}")
+            if record is None:
+                continue
+            self.assertTrue(
+                record.get("is_sqlancer_publication"),
+                f"{entry['title']} is listed as ours but still counts as external")
+
+
+class AdoptionEvidenceTest(unittest.TestCase):
+    def test_a_prose_statement_is_preferred_over_a_bare_link(self):
+        text = ("DataFusion uses the [SQLancer] for fuzz testing.\n\n"
+                "[sqlancer]: https://github.com/sqlancer/sqlancer\n")
+        excerpt, kind = _best_excerpt(text, "docs/testing.md")
+        self.assertEqual("statement", kind)
+        self.assertIn("uses the [SQLancer]", excerpt)
+
+    def test_a_bare_repository_link_is_not_treated_as_an_invocation(self):
+        text = "See [SQLancer]: https://github.com/sqlancer/sqlancer for details\n"
+        self.assertIsNone(_best_excerpt(text, "docs/other.md"))
+
+    def test_a_command_line_counts_as_an_invocation(self):
+        text = "RUN java -jar sqlancer-2.0.0.jar --num-threads 4 duckdb\n"
+        excerpt, kind = _best_excerpt(text, "test/Dockerfile")
+        self.assertEqual("invocation", kind)
+
+
+class ArtifactTest(unittest.TestCase):
+    """uses_infrastructure is decided from the artifact, not from the paper."""
+
+    class FakeGitHub:
+        """Stands in for GitHub with one repository's files in memory."""
+
+        def __init__(self, *, description="", files=None, parent=None,
+                     tree=None):
+            self.metadata = {"default_branch": "main", "description": description}
+            if parent:
+                self.metadata["parent"] = {"full_name": parent}
+            self.files = files or {}
+            self._tree = tree or list(self.files)
+
+        def repo(self, owner, name, **kwargs):
+            return self.metadata
+
+        def tree(self, owner, repo, ref="HEAD", **kwargs):
+            return [{"path": path, "type": "blob"} for path in self._tree]
+
+        def raw_file(self, owner, repo, path, ref="HEAD", **kwargs):
+            return self.files.get(path), {"status": 200}
+
+    def test_a_tool_name_is_taken_from_the_paper_text(self):
+        from tools.impact.collectors.artifacts import candidate_tool_names
+        names = candidate_tool_names(
+            "Detecting Logic Bugs in DBMSs via Equivalent Data Construction",
+            "We implement our approach in a tool called Radar and evaluate it.")
+        self.assertIn("Radar", names)
+
+    def test_a_leading_tool_name_in_the_title_is_found(self):
+        from tools.impact.collectors.artifacts import candidate_tool_names
+        self.assertIn("Pinolo", candidate_tool_names(
+            "Pinolo: Detecting Logical Bugs in Database Management Systems", ""))
+
+    def test_generic_acronyms_are_not_tool_names(self):
+        from tools.impact.collectors.artifacts import candidate_tool_names
+        names = candidate_tool_names("Testing DBMS via SQL and JSON", "")
+        self.assertEqual([], names)
+
+    def test_markers_alone_are_not_conclusive(self):
+        from tools.impact.collectors.artifacts import markers_are_conclusive
+        self.assertFalse(markers_are_conclusive(["randomly_java_present"]),
+                         "one vendored file must not settle infrastructure reuse")
+        self.assertTrue(markers_are_conclusive(
+            ["randomly_java_present", "sqlancer_package_structure"]))
+        self.assertTrue(markers_are_conclusive(["fork_of_sqlancer_repository"]))
+
+    def test_a_fork_of_sqlancer_is_detected(self):
+        from tools.impact.collectors.artifacts import inspect_repository
+        gh = self.FakeGitHub(parent="sqlancer/sqlancer")
+        markers, evidence = inspect_repository(gh, "someone", "their-fork",
+                                               timestamp="2026-01-01T00:00:00Z")
+        self.assertIn("fork_of_sqlancer_repository", markers)
+        self.assertTrue(evidence)
+
+    def test_retained_source_files_are_detected(self):
+        from tools.impact.collectors.artifacts import inspect_repository
+        gh = self.FakeGitHub(files={
+            "src/sqlancer/Randomly.java": "package sqlancer;",
+            "src/sqlancer/Main.java": "package sqlancer;",
+            "pom.xml": "<groupId>com.sqlancer</groupId>",
+        })
+        markers, _ = inspect_repository(gh, "someone", "artifact",
+                                        timestamp="2026-01-01T00:00:00Z")
+        self.assertIn("retained_sqlancer_source_files", markers)
+        self.assertIn("sqlancer_package_structure", markers)
+        self.assertIn("sqlancer_build_file_reference", markers)
+
+    def test_a_repository_naming_the_paper_is_linked(self):
+        from tools.impact.collectors.artifacts import link_evidence
+        paper = {"title": "Detecting Logic Bugs via Equivalent Data Construction",
+                 "doi": "10.1145/1234567", "arxiv_id": None}
+        gh = self.FakeGitHub(files={
+            "README.md": ("Artifact for Detecting Logic Bugs via Equivalent "
+                          "Data Construction.\n")})
+        link = link_evidence(gh, "o", "r", paper, tool_names=[],
+                             timestamp="2026-01-01T00:00:00Z")
+        self.assertIsNotNone(link)
+        self.assertIn("names this paper", link["note"])
+
+    def test_a_repository_citing_the_doi_is_linked(self):
+        from tools.impact.collectors.artifacts import link_evidence
+        paper = {"title": "Some paper", "doi": "10.1145/1234567", "arxiv_id": None}
+        gh = self.FakeGitHub(files={"README.md": "See https://doi.org/10.1145/1234567"})
+        link = link_evidence(gh, "o", "r", paper, tool_names=[],
+                             timestamp="2026-01-01T00:00:00Z")
+        self.assertIsNotNone(link)
+
+    def test_a_name_collision_without_database_context_is_not_linked(self):
+        from tools.impact.collectors.artifacts import link_evidence
+        paper = {"title": "Some paper", "doi": None, "arxiv_id": None}
+        gh = self.FakeGitHub(description="A radar chart library for Android",
+                             files={"README.md": "Draw radar charts."})
+        self.assertIsNone(link_evidence(gh, "o", "radar", paper,
+                                        tool_names=["radar"],
+                                        timestamp="2026-01-01T00:00:00Z"))
+
+    def test_a_tool_named_repository_about_databases_is_linked(self):
+        from tools.impact.collectors.artifacts import link_evidence
+        paper = {"title": "Some paper", "doi": None, "arxiv_id": None}
+        gh = self.FakeGitHub(description="Logic bug detection for SQL DBMSs",
+                             files={"README.md": "Tests SQLite and MySQL."})
+        link = link_evidence(gh, "o", "radar", paper, tool_names=["Radar"],
+                             timestamp="2026-01-01T00:00:00Z")
+        self.assertIsNotNone(link)
+        self.assertIn("named after", link["note"])
+
+    def test_an_unrelated_repository_is_not_linked(self):
+        from tools.impact.collectors.artifacts import link_evidence
+        paper = {"title": "Some paper", "doi": None, "arxiv_id": None}
+        gh = self.FakeGitHub(description="Unrelated project",
+                             files={"README.md": "Nothing to do with it."})
+        self.assertIsNone(link_evidence(gh, "o", "other", paper,
+                                        tool_names=["Radar"],
+                                        timestamp="2026-01-01T00:00:00Z"))
+
+
+class FullTextTest(unittest.TestCase):
+    """Judging a paper from its own words, not from a citation index."""
+
+    def setUp(self):
+        from tools.impact.collectors import fulltext
+        self.find = fulltext.find_signals
+
+    def test_a_claim_of_reuse_is_found(self):
+        found = self.find("We implemented our approach on top of SQLancer.")
+        self.assertEqual(1, len(found["uses_infrastructure"]))
+
+    def test_related_work_is_not_the_authors_claim(self):
+        """"Previous work built on SQLancer" says nothing about this paper."""
+        for sentence in (
+                "Previous work implemented their tool on top of SQLancer.",
+                "Existing approaches are built on SQLancer.",
+                "Prior work extended TLP to graph databases."):
+            found = self.find(sentence)
+            self.assertEqual([], found["uses_infrastructure"], sentence)
+            self.assertEqual([], found["extends_technique"], sentence)
+
+    def test_a_figure_caption_is_not_a_claim_of_reuse(self):
+        """The pattern needs `we` or `our`; an adjective is not a verb."""
+        found = self.find(
+            "Fig. 2b presents the GA for WHERE Extended case of Ternary Logic "
+            "Partitioning (TLP) [21] oracle from SQLancer.")
+        self.assertEqual([], found["uses_infrastructure"])
+
+    def test_a_comparison_is_found(self):
+        found = self.find(
+            "We compare the performance of our tool against SQLancer on five "
+            "database systems.")
+        self.assertEqual(1, len(found["compares_with"]))
+
+    def test_an_extension_claim_is_found(self):
+        found = self.find("We extend TLP to graph database systems.")
+        self.assertEqual(1, len(found["extends_technique"]))
+
+    def test_an_excluded_technique_alone_establishes_nothing(self):
+        found = self.find("We compare our approach against EET on five DBMSs.")
+        self.assertEqual([], found["compares_with"])
+
+    def test_sentences_are_kept_verbatim(self):
+        sentence = "We implemented our prototype on top of SQLancer."
+        found = self.find(sentence)
+        self.assertEqual(sentence, found["uses_infrastructure"][0])
+
+    def test_a_pdf_that_is_not_a_pdf_is_rejected(self):
+        from tools.impact.collectors import fulltext
+
+        class NotAPdf:
+            def fetch_binary(self, url, **kwargs):
+                return b"<html>not a pdf</html>", False
+
+        self.assertIsNone(fulltext.pdf_text(NotAPdf(), "https://example.org/x"))
+
+
+class PreprintTest(unittest.TestCase):
+    """Matching a paywalled paper to its preprint has to be strict."""
+
+    ATOM = """<?xml version="1.0"?>
+    <feed xmlns="http://www.w3.org/2005/Atom">
+      <entry>
+        <id>http://arxiv.org/abs/2604.16373v1</id>
+        <title>DIRT: Database-Integrated Random Testing</title>
+        <summary>A paper.</summary>
+        <author><name>Alperen Keles</name></author>
+        <author><name>Ethan Chou</name></author>
+        <link title="pdf" href="http://arxiv.org/pdf/2604.16373v1"/>
+      </entry>
+    </feed>"""
+
+    class FakeCache:
+        def __init__(self, body):
+            self.body = body
+
+        def get(self, key):
+            return None
+
+        def is_fresh(self, key, **kwargs):
+            return False
+
+        def put(self, key, value):
+            pass
+
+        def store(self, key, **kwargs):
+            return {}
+
+    def _arxiv(self):
+        from tools.impact.scholarly import ArXiv
+        arxiv = ArXiv(self.FakeCache(self.ATOM))
+        arxiv._entries = lambda query, limit: __import__(
+            "tools.impact.scholarly", fromlist=["_parse_atom"]
+        )._parse_atom(self.ATOM)
+        return arxiv
+
+    def test_an_exact_title_with_a_shared_author_matches(self):
+        found = self._arxiv().find_preprint(
+            "DIRT: Database-Integrated Random Testing", ["Alperen Keles"])
+        self.assertIsNotNone(found)
+        self.assertEqual("2604.16373v1", found["arxiv_id"])
+
+    def test_a_different_paper_by_the_same_author_does_not_match(self):
+        self.assertIsNone(self._arxiv().find_preprint(
+            "Some Entirely Different Paper Title", ["Alperen Keles"]))
+
+    def test_the_same_title_by_different_authors_does_not_match(self):
+        """Titles repeat across communities; an author must corroborate."""
+        self.assertIsNone(self._arxiv().find_preprint(
+            "DIRT: Database-Integrated Random Testing", ["Someone Unrelated"]))
+
+    def test_a_title_too_short_to_identify_anything_is_refused(self):
+        self.assertIsNone(self._arxiv().find_preprint("DIRT", ["Keles"]))
+
+    def test_atom_parsing_reads_what_it_needs(self):
+        from tools.impact.scholarly import _parse_atom
+        entries = _parse_atom(self.ATOM)
+        self.assertEqual(1, len(entries))
+        self.assertEqual(["Alperen Keles", "Ethan Chou"], entries[0]["authors"])
+        self.assertIn("pdf", entries[0]["pdf_url"])
+
+
+class LabMemberTest(unittest.TestCase):
+    """Author-scoped search, because term search is capped at 1000 results."""
+
+    def test_handles_are_extracted_from_the_people_page(self):
+        from tools.impact.collectors import lab_members
+
+        class FakeFetcher:
+            def fetch(self, url, **kwargs):
+                return {"status": 200, "body":
+                        '<a href="https://github.com/bajinsheng">x</a>'
+                        '<a href="https://github.com/suyZhong">y</a>'
+                        '<a href="https://github.com/sqlancer/sqlancer">z</a>'}, False
+
+        found = lab_members.from_people_page(FakeFetcher())
+        self.assertIn("bajinsheng", found)
+        self.assertIn("suyZhong", found)
+        self.assertNotIn("sqlancer", found,
+                         "the project's own account is not a lab member")
+
+    def test_one_query_per_member_covers_every_term(self):
+        from tools.impact.collectors import lab_members
+        queries = lab_members.search_queries(["alice", "bob"], ["SQLancer", "TLP"])
+        self.assertEqual(2, len(queries))
+        self.assertIn("author:alice", queries[0])
+        self.assertIn("SQLancer OR TLP", queries[0])
+
+    def test_display_names_are_not_usable_as_handles(self):
+        """The bug list mixes handles and full names; only handles can search."""
+        from tools.impact.collectors import lab_members
+
+        class FakeGitHub:
+            def raw_file(self, *args, **kwargs):
+                import json
+                return json.dumps([
+                    {"reported_by": "bajinsheng"},
+                    {"reported_by": "Suyang Zhong"},
+                ]), {}
+
+        found = lab_members.from_bug_list(FakeGitHub())
+        self.assertEqual(["bajinsheng"], found)
+
+
+class ReproducerSignatureTest(unittest.TestCase):
+    """The reproducer's own shape, which is what reaches reports that say nothing."""
+
+    def _excerpt(self, text):
+        from tools.impact.collectors.github_bugs import reproducer_excerpt
+        return reproducer_excerpt(text)
+
+    def test_a_generated_schema_is_recognised(self):
+        self.assertTrue(self._excerpt(
+            "CREATE TABLE t0(c0 BOOLEAN, c1 INT); INSERT INTO t0 VALUES (0, 1);"))
+
+    def test_another_generator_s_columns_are_not_sqlancer_s(self):
+        """Other tools number their tables too; the column names differ."""
+        self.assertIsNone(self._excerpt(
+            "CREATE TABLE t2 (c_pk INTEGER, c_int INTEGER); SELECT c_pk FROM t2;"))
+        self.assertIsNone(self._excerpt(
+            "create table t0 (c_0 int); create table t1 (c_1 int);"))
+
+    def test_a_handwritten_reproducer_is_not_matched(self):
+        self.assertIsNone(self._excerpt(
+            "CREATE TABLE users (id INT, name TEXT); SELECT * FROM users;"))
+
+    def test_the_excerpt_is_copied_from_the_report(self):
+        text = "Some prose.\nCREATE TABLE t0(c0 INT);\nSELECT * FROM t0;\nMore."
+        excerpt = self._excerpt(text)
+        self.assertIn(excerpt, text)
+
+    def test_the_reproducer_alone_never_attributes_a_stranger_s_report(self):
+        """It is evidence only for someone already known to run campaigns."""
+        from tools.impact import taxonomy
+        from tools.impact.collectors.github_bugs import deterministic_attribution
+        tax = taxonomy.load()
+        issue = {"title": "Wrong result", "state": "open",
+                 "body": "CREATE TABLE t0(c0 INT); SELECT * FROM t0;", "labels": []}
+        text = f"{issue['title']}\n{issue['body']}"
+        self.assertIsNone(deterministic_attribution(text, tax, issue))
+        found = deterministic_attribution(text, tax, issue,
+                                          reporter_runs_campaigns=True)
+        self.assertIsNotNone(found)
+        self.assertEqual("campaign_evidence", found[0])
+
+
+class SearchOrderTest(unittest.TestCase):
+    """Which queries run first decides which ones the candidate cap silences."""
+
+    def test_author_searches_run_before_term_searches(self):
+        from tools.impact import taxonomy
+        from tools.impact.collectors import github_bugs
+
+        class FakeGitHub:
+            def search_issues(self, query, **kwargs):
+                seen.append(query)
+                return []
+
+        seen = []
+        github_bugs.collect(FakeGitHub(), classifier=None, full=True)
+        authored = [i for i, q in enumerate(seen) if q.startswith("author:")]
+        others = [i for i, q in enumerate(seen) if not q.startswith("author:")]
+        if authored and others:
+            self.assertLess(max(authored), min(others),
+                            "author searches must not be behind the term searches")
+
+
+class CampaignReporterTest(unittest.TestCase):
+    """Admitting a report on the strength of who filed it.
+
+    The weakest rule in the policy, so what it must never do matters as much as
+    what it does.
+    """
+
+    def setUp(self):
+        from tools.impact import taxonomy
+        from tools.impact.collectors import github_bugs
+        self.g = github_bugs
+        self.tax = taxonomy.load()
+
+    def _issue(self, body, title="Wrong result", **kwargs):
+        issue = {"title": title, "body": body, "state": "open", "labels": [],
+                 "html_url": "https://github.com/duckdb/duckdb/issues/1",
+                 "user": {"login": "Yibo-Dong"}, "created_at": "2026-01-01T00:00:00Z"}
+        issue.update(kwargs)
+        return issue
+
+    def _attribute(self, issue, member=True):
+        text = self.g._issue_source_text(issue)
+        return self.g.deterministic_attribution(
+            text, self.tax, issue, reporter_runs_campaigns=member)
+
+    def test_a_member_s_defect_report_is_admitted_without_a_tool_mention(self):
+        issue = self._issue(
+            "## What happened\n\nDuckDB returns the list in reverse order for "
+            "a window aggregate.\n\n## Environment\n\nv1.2.0")
+        found = self._attribute(issue)
+        self.assertIsNotNone(found)
+        self.assertEqual("campaign_reporter", found[0])
+        self.assertEqual("low", self.g._confidence_for(found[0]))
+
+    def test_the_same_report_from_a_stranger_is_not_admitted(self):
+        issue = self._issue(
+            "## What happened\n\nDuckDB returns the list in reverse order.")
+        self.assertIsNone(self._attribute(issue, member=False))
+
+    def test_the_quoted_excerpt_is_the_reporter_s_words_not_the_template_s(self):
+        issue = self._issue(
+            "## What happened\n\nDolt panics when evaluating LOCATE with a "
+            "negative start position.\n\n## Environment")
+        found = self._attribute(issue)
+        self.assertNotIn("What happened", found[3])
+        self.assertIn("Dolt panics", found[3])
+        self.assertIn(found[3], self.g._issue_source_text(issue))
+
+    def test_a_carefully_written_report_is_not_rejected_for_its_vocabulary(self):
+        """The reports that matter describe the misbehaviour instead of naming it."""
+        issue = self._issue(
+            "Dolt evaluates the frame bound before the ORDER BY is applied, so "
+            "the window sees rows in the wrong order and returns a value from "
+            "the wrong partition. This reproduces on a fresh database.",
+            title="Dolt ignores `LIMIT` inside a correlated `EXISTS` subquery.",
+            labels=[{"name": "customer issue"}])
+        self.assertFalse(self.g.looks_like_a_defect(issue),
+                         "precondition: the old gate rejects this")
+        self.assertEqual("campaign_reporter", self._attribute(issue)[0])
+
+    def test_a_pull_request_is_never_a_bug_report(self):
+        issue = self._issue("A description long enough to pass the length test, "
+                            "with plenty of words in it to be sure.",
+                            pull_request={})
+        self.assertIsNone(self._attribute(issue))
+
+    def test_a_one_line_issue_is_not_admitted(self):
+        issue = self._issue("hm?", title="Something odd")
+        self.assertIsNone(self._attribute(issue))
+
+    def test_a_feature_request_from_a_member_is_still_not_a_bug(self):
+        issue = self._issue(
+            "It would be nice if this were configurable in some way here.",
+            title="Support configuring the window frame")
+        self.assertIsNone(self._attribute(issue))
+
+    def test_a_report_crediting_an_excluded_technique_is_still_rejected(self):
+        """Membership does not override the exclusions."""
+        excluded = next(iter(self.tax.excluded.values()))
+        name = excluded.get("display_name") or excluded.get("name") or ""
+        if not name:
+            self.skipTest("no excluded technique with a name")
+        issue = self._issue(f"We found this with {name}, a wrong result here.")
+        self.assertIsNone(self._attribute(issue))
+
+    def test_the_record_says_it_rests_on_the_roster(self):
+        issue = self._issue(
+            "## What happened\n\nDuckDB returns the list in reverse order for "
+            "a window aggregate.")
+        rule, technique, finder, excerpt = self._attribute(issue)
+        record = self.g.build_record(
+            issue, "duckdb", rule=rule, technique=technique, finder=finder,
+            excerpt=excerpt, source_text=self.g._issue_source_text(issue),
+            timestamp="2026-01-01T00:00:00Z",
+            confidence=self.g._confidence_for(rule))
+        notes = " ".join(item["note"] for item in record["attribution"]["evidence"])
+        self.assertIn("membership", notes)
+        self.assertIn(self.g.ROSTER_URL,
+                      [item["source_url"] for item in record["attribution"]["evidence"]])
+
+    def test_a_reproducer_still_wins_over_the_weaker_rule(self):
+        """A SQLancer-shaped reproducer is better evidence; keep it."""
+        issue = self._issue("CREATE TABLE t0(c0 INT); SELECT * FROM t0;")
+        self.assertEqual("campaign_evidence", self._attribute(issue)[0])
+
+
+class PaperNoteTest(unittest.TestCase):
+    """A summary is a convenience; the evidence under it is the record."""
+
+    def _paper(self):
+        return {
+            "id": "paper:doi:10.1145/1",
+            "title": "A Paper",
+            "authors": ["A. Author"],
+            "year": 2025,
+            "venue": "A Venue",
+            "doi": "10.1145/1",
+            "url": "https://doi.org/10.1145/1",
+            "abstract": "We built a tool on top of SQLancer. It found bugs.",
+            "relationships": {
+                "references": {
+                    "value": "yes", "method": "citation_graph", "techniques": [],
+                    "evidence": [{
+                        "source_url": "https://doi.org/10.1145/1",
+                        "source_type": "paper_citation_context",
+                        "excerpt": "We use SQLancer [12].",
+                        "excerpt_is_verbatim": True,
+                        "note": "Citing sentence.",
+                        "retrieved_at": "2026-01-01T00:00:00Z",
+                    }],
+                },
+            },
+        }
+
+    def setUp(self):
+        import tempfile, pathlib as _pathlib
+        from tools.impact import notes
+        self.notes = notes
+        self._real_dir = notes.NOTES_DIR
+        self._tmp = tempfile.TemporaryDirectory()
+        notes.NOTES_DIR = _pathlib.Path(self._tmp.name)
+
+    def tearDown(self):
+        self.notes.NOTES_DIR = self._real_dir
+        self._tmp.cleanup()
+
+    def test_evidence_is_written_without_any_model(self):
+        note = self.notes.build_note(self._paper(), classifier=None, fetcher=None)
+        self.assertIsNone(note["summary"])
+        self.assertEqual(1, len(note["evidence"]))
+        self.assertEqual("We use SQLancer [12].",
+                         note["evidence"][0]["sources"][0]["excerpt"])
+
+    def test_a_summary_must_quote_the_source_it_claims(self):
+        paper = self._paper()
+        with self.assertRaises(self.notes.UngroundedSummary):
+            self.notes.apply_summary(
+                paper, text="A summary.", relationship_to_sqlancer="Uses it.",
+                excerpts=["a sentence that is not in the abstract"],
+                written_from="abstract", source_text=paper["abstract"],
+                model="claude-opus-5")
+
+    def test_a_grounded_summary_is_stored_and_marked_generated(self):
+        paper = self._paper()
+        note = self.notes.apply_summary(
+            paper, text="A summary.", relationship_to_sqlancer="Uses it.",
+            excerpts=["We built a tool on top of SQLancer."],
+            written_from="abstract", source_text=paper["abstract"],
+            model="claude-opus-5")
+        self.assertTrue(note["summary"]["is_model_generated"])
+        self.assertTrue(note["summary"]["grounded_in_source"])
+        self.assertEqual("abstract", note["summary"]["written_from"])
+        # And the evidence is still there, untouched by the summary.
+        self.assertEqual("We use SQLancer [12].",
+                         note["evidence"][0]["sources"][0]["excerpt"])
+
+    def test_regenerating_evidence_never_drops_an_existing_summary(self):
+        """A run with no API key must not erase what an earlier run wrote."""
+        import json
+        paper = self._paper()
+        self.notes.apply_summary(
+            paper, text="A summary.", relationship_to_sqlancer="Uses it.",
+            excerpts=["We built a tool on top of SQLancer."],
+            written_from="abstract", source_text=paper["abstract"],
+            model="claude-opus-5")
+        path = self.notes.note_path(paper)
+
+        # Now rebuild the note the way a keyless run would, and write it.
+        note = self.notes.build_note(paper, classifier=None, fetcher=None)
+        self.assertIsNone(note["summary"])
+        counts = self._write_all_over([paper])
+        kept = json.loads(path.read_text(encoding="utf-8"))
+        self.assertIsNotNone(kept["summary"], "the summary was erased")
+        self.assertEqual("A summary.", kept["summary"]["text"])
+
+    def _write_all_over(self, papers):
+        """Run write_all against a fixed paper list."""
+        return self.notes.write_all(papers=papers)
+
+    def test_a_supplied_pdf_is_read_even_without_a_fetcher(self):
+        """It is on disk; needing a fetcher for it would silently lose it."""
+        from unittest import mock
+        from tools.impact.collectors import fulltext
+
+        paper = self._paper()
+        with mock.patch.object(fulltext, "local_pdf",
+                               return_value="file:///tmp/x.pdf"), \
+             mock.patch.object(fulltext, "pdf_text",
+                               return_value="Full text of the paper."):
+            text, label, _ = self.notes.summary_source(paper)
+        self.assertEqual("supplied PDF", label)
+        self.assertEqual("Full text of the paper.", text)
+
+    def test_the_note_is_named_for_the_paper_id_not_its_title(self):
+        paper = self._paper()
+        first = self.notes.note_path(paper)
+        paper["title"] = "A Paper, Renamed On Publication"
+        self.assertEqual(first, self.notes.note_path(paper))
+
+
+class PageFurnitureTest(unittest.TestCase):
+    """A quotation must be the author's sentence, not the publisher's stamp."""
+
+    def test_an_ieee_licence_stamp_is_removed(self):
+        from tools.impact.collectors import fulltext
+        text = ("Rigger et al. proposed NoREC, which restructures SQL "
+                "statements to inhibit query 125 Authorized licensed use "
+                "limited to the terms of the applicable license agreement "
+                "with IEEE. Restrictions apply. The next sentence.")
+        cleaned = fulltext._strip_page_furniture(text)
+        self.assertNotIn("Authorized licensed", cleaned)
+        self.assertIn("proposed NoREC", cleaned)
+        self.assertIn("The next sentence.", cleaned)
+
+    def test_an_acm_permission_block_is_removed(self):
+        from tools.impact.collectors import fulltext
+        text = ("We implemented SRS on top of SQLancer. Permission to make "
+                "digital or hard copies of part or all of this work for "
+                "personal use is granted without fee. Our evaluation follows.")
+        cleaned = fulltext._strip_page_furniture(text)
+        self.assertNotIn("Permission to make", cleaned)
+        self.assertIn("We implemented SRS on top of SQLancer.", cleaned)
+
+    def test_ordinary_text_is_untouched(self):
+        from tools.impact.collectors import fulltext
+        text = "We compare against TLP and NoREC on five widely used systems."
+        self.assertEqual(text, fulltext._strip_page_furniture(text))
+
+
+class BaselineSentenceTest(unittest.TestCase):
+    """Taking a tool up as a baseline is the opposite of building on it."""
+
+    def signals(self, sentence):
+        from tools.impact.collectors import fulltext
+        return {k: len(v) for k, v in fulltext.find_signals(sentence).items()}
+
+    def test_adapting_a_tool_as_a_baseline_is_only_a_comparison(self):
+        found = self.signals(
+            "To evaluate the effectiveness of TSGuard in detecting logic bugs "
+            "in TSMSs, we adapted the open-source relational database testing "
+            "tool SQLancer as a baseline for comparison.")
+        self.assertEqual(1, found["compares_with"])
+        self.assertEqual(0, found["extends_technique"],
+                         "adapting a tool to measure against is not extending it")
+        self.assertEqual(0, found["uses_infrastructure"])
+
+    def test_a_real_extension_still_counts(self):
+        found = self.signals(
+            "We extended TLP to support graph queries in our implementation.")
+        self.assertEqual(1, found["extends_technique"])
+
+    def test_real_reuse_still_counts(self):
+        found = self.signals("We implemented our prototype on top of SQLancer.")
+        self.assertEqual(1, found["uses_infrastructure"])
+
+
+class IntakeTest(unittest.TestCase):
+    """Filing a downloaded PDF as the paper it actually is."""
+
+    GDSMITH = ("GDsmith: Detecting Bugs in Cypher Graph Database Engines\n"
+               "Ziyue Hua, Wei Lin, Luyao Ren\n"
+               "Peking University\n\nABSTRACT\nGraph database engines ...")
+
+    def setUp(self):
+        from tools.impact import intake
+        self.intake = intake
+
+    def test_a_generic_title_does_not_match_everything(self):
+        """The failure this guards against: short titles made of common words.
+
+        Counting how many of a candidate's words appear on the page scored
+        "Test Data Generation for Complex SQL Queries" at 100% against an
+        unrelated graph database paper.
+        """
+        right = self.intake.score(
+            "GDsmith: Detecting Bugs in Cypher Graph Database Engines", self.GDSMITH)
+        wrong = self.intake.score(
+            "Test Data Generation for Complex SQL Queries", self.GDSMITH)
+        self.assertGreater(right, 0.8)
+        self.assertLess(wrong, 0.5)
+
+    def test_a_paper_and_its_preprint_are_refused_rather_than_guessed(self):
+        candidates = [
+            {"title": "GDsmith: Detecting Bugs in Cypher Graph Database Engines",
+             "filename": "a.pdf"},
+            {"title": "GDsmith: Detecting Bugs in Graph Database Engines",
+             "filename": "b.pdf"},
+        ]
+        match, best, runner_up = self.intake.identify(self.GDSMITH, candidates)
+        self.assertIsNone(match)
+        self.assertTrue(self.intake.is_ambiguous(best, runner_up))
+
+    def test_a_usenix_cover_sheet_does_not_hide_the_title(self):
+        """USENIX runs the title onto the end of its cover boilerplate."""
+        page = ("This paper is included in the Proceedings of the\n"
+                "31st USENIX Security Symposium.\n"
+                "August 10-12, 2022 - Boston, MA, USA\n"
+                "Open access to the Proceedings of the\n"
+                "31st USENIX Security Symposium is\n"
+                "sponsored by USENIX.Detecting Logical Bugs of DBMS with\n"
+                "Coverage-based Guidance\n"
+                "Yu Liang, Pennsylvania State University")
+        self.assertIn("Detecting Logical Bugs of DBMS with Coverage-based Guidance",
+                      self.intake.title_candidates(page))
+
+    def test_a_title_wrapping_onto_a_short_line_keeps_that_line(self):
+        """"DBMS" on its own line is the end of the title, not noise."""
+        page = ("Unveiling Logic Bugs in SPJG Query Optimizations within\n"
+                "DBMS\n"
+                "XIU TANG ,Zhejiang University, Hangzhou, China")
+        self.assertIn("Unveiling Logic Bugs in SPJG Query Optimizations within DBMS",
+                      self.intake.title_candidates(page))
+
+    def test_a_tie_resolves_to_the_one_still_wanted(self):
+        """A paper and its preprint share a title; only one needs the file."""
+        candidates = [
+            {"title": "A Comprehensive Survey on DBMS Fuzzing", "filename": "doi.pdf"},
+            {"title": "A Comprehensive Survey on DBMS Fuzzing", "filename": "arxiv.pdf"},
+        ]
+        page = "A Comprehensive Survey on DBMS Fuzzing\nAuthor, University"
+        match, _, _ = self.intake.identify(page, candidates)
+        self.assertIsNone(match, "with no wanted set, a tie stays a tie")
+        match, _, _ = self.intake.identify(page, candidates, wanted={"doi.pdf"})
+        self.assertEqual("doi.pdf", match["filename"])
+        match, _, _ = self.intake.identify(page, candidates,
+                                           wanted={"doi.pdf", "arxiv.pdf"})
+        self.assertIsNone(match, "if both are wanted, only a person can choose")
+
+    def test_a_tie_between_two_papers_that_need_nothing_is_not_a_decision(self):
+        """Usually a paper already filed under one of its two identities."""
+        candidates = [
+            {"title": "A Comprehensive Survey on DBMS Fuzzing", "filename": "doi.pdf"},
+            {"title": "A Comprehensive Survey on DBMS Fuzzing", "filename": "arxiv.pdf"},
+        ]
+        page = "A Comprehensive Survey on DBMS Fuzzing\nAuthor, University"
+        match, best, runner_up = self.intake.identify(page, candidates, wanted=set())
+        self.assertIsNotNone(match, "nothing is wanted, so nothing to decide")
+
+    def test_front_matter_above_the_title_does_not_swallow_it(self):
+        """An author line looked like front matter and moved the start past it."""
+        page = ("Semantic Hint-Based Fuzzing for Time-Series\n"
+                "Databases\n"
+                "1st Panta Kittisatra, Some University\n")
+        best = self.intake.title_candidates(page)[0]
+        self.assertTrue(best.startswith("Semantic Hint-Based Fuzzing"), best)
+
+    def test_a_declared_title_identifies_a_pdf_whose_page_lacks_one(self):
+        """A magazine extract can begin mid-sentence, with no title anywhere."""
+        page = "Our experience at AWS with TLA+ revealed two advantages."
+        self.assertGreater(
+            self.intake.score("Systems Correctness Practices at AWS", page,
+                              ("Systems Correctness Practices at AWS",)), 0.9)
+
+    def test_a_publisher_filename_identifies_the_paper_exactly(self):
+        import pathlib
+        rows = [{"title": "Systems Correctness Practices at Amazon Web Services",
+                 "filename": "10.1145_3729175.pdf"}]
+        found = self.intake.identifier_from_filename(
+            pathlib.Path("/tmp/3729175.pdf"), rows)
+        self.assertEqual("10.1145_3729175.pdf", found["filename"])
+        self.assertIsNone(self.intake.identifier_from_filename(
+            pathlib.Path("/tmp/notes.pdf"), rows))
+        self.assertIsNone(self.intake.identifier_from_filename(
+            pathlib.Path("/tmp/9999999.pdf"), rows))
+
+    def test_a_doi_printed_on_the_page_identifies_the_paper(self):
+        """A journal masthead can push the title past any layout rule."""
+        rows = [{"title": "Benchmarking Autonomy", "filename": "x.pdf",
+                 "doi": "10.22399/ijcesen.5462"},
+                {"title": "Something Else", "filename": "y.pdf",
+                 "doi": "10.1145/3769828"}]
+        page = ("Copyright IJCESEN\nInternational Journal\n"
+                "Article Info:\nDOI:  10.22399/ijcesen. 5462\n")
+        self.assertEqual("x.pdf", self.intake.doi_on_page(page, rows)["filename"])
+        self.assertIsNone(self.intake.doi_on_page("no doi here", rows))
+
+    def test_an_unrelated_pdf_is_not_ambiguous_merely_unmatched(self):
+        candidates = [{"title": "Testing Database Systems via Differential "
+                                "Query Execution", "filename": "a.pdf"}]
+        page = "Konditionen fur die Buchung\nHotelreservierung 2026"
+        match, best, runner_up = self.intake.identify(page, candidates)
+        self.assertIsNone(match)
+        self.assertFalse(self.intake.is_ambiguous(best, runner_up))
+
+    def test_a_clear_match_is_returned(self):
+        candidates = [
+            {"title": "GDsmith: Detecting Bugs in Cypher Graph Database Engines",
+             "filename": "a.pdf"},
+            {"title": "Testing DBMSs via Equivalent Expression Transformation",
+             "filename": "b.pdf"},
+        ]
+        match, best, _ = self.intake.identify(self.GDSMITH, candidates)
+        self.assertIsNotNone(match)
+        self.assertEqual("a.pdf", match["filename"])
+        self.assertGreater(best, 0.8)
+
+
+class WorklistTest(unittest.TestCase):
+    """The download list must ask only for papers nothing else can reach."""
+
+    def _paper(self, **kwargs):
+        record = {"title": "T", "url": "https://example.org/p", "year": 2025,
+                  "doi": None, "arxiv_id": None, "s2_paper_id": None,
+                  "relationships": {key: {"value": "uncertain"} for key in
+                                    ("uses_infrastructure", "extends_technique",
+                                     "compares_with")}}
+        record.update(kwargs)
+        return record
+
+    def test_a_preprint_is_never_requested(self):
+        from tools.impact import worklist
+        rows = worklist.wanted([self._paper(arxiv_id="2501.00001",
+                                            doi="10.1145/1")])
+        self.assertEqual([], rows)
+
+    def test_a_pvldb_paper_is_never_requested(self):
+        from tools.impact import worklist
+        rows = worklist.wanted([self._paper(doi="10.14778/3695624.3695625")])
+        self.assertEqual([], rows)
+
+    def test_sqlancer_s_own_papers_are_never_requested(self):
+        from tools.impact import worklist
+        rows = worklist.wanted([self._paper(doi="10.1145/1",
+                                            is_sqlancer_publication=True)])
+        self.assertEqual([], rows)
+
+    def test_the_filename_is_the_one_the_pipeline_looks_for(self):
+        from tools.impact import worklist
+        from tools.impact.collectors import fulltext
+        # A deliberately unreal DOI: a test must not change its meaning when
+        # somebody downloads the paper it named.
+        rows = worklist.wanted([self._paper(doi="10.9999/not-a-real-paper")])
+        self.assertEqual("10.9999_not-a-real-paper.pdf", rows[0]["filename"])
+        self.assertEqual(rows[0]["filename"],
+                         fulltext.wanted_filename("10.9999/not-a-real-paper", None))
+
+    def test_a_paper_whose_contexts_hint_at_more_outranks_a_bare_citation(self):
+        from tools.impact import worklist
+        plain = self._paper(doi="10.1145/1", title="Plain", year=2026)
+        hinting = self._paper(doi="10.1145/2", title="Hinting", year=2020)
+        hinting["relationships"]["references"] = {
+            "value": "yes",
+            "evidence": [{"excerpt": "We implemented our prototype on top of "
+                                     "SQLancer for this evaluation."}],
+        }
+        rows = worklist.wanted([plain, hinting])
+        self.assertEqual("Hinting", rows[0]["title"],
+                         "a hint should outrank a merely newer paper")
+
+    def test_a_paper_with_no_doi_can_still_be_supplied(self):
+        """A dozen papers have neither DOI nor arXiv id; they are still wanted."""
+        from tools.impact import worklist
+        from tools.impact.collectors import fulltext
+        paper = self._paper(doi=None, arxiv_id=None,
+                            s2_paper_id="0000000000000000000000000000000000000000")
+        rows = worklist.wanted([paper])
+        self.assertEqual(1, len(rows))
+        self.assertEqual(
+            "s2_0000000000000000000000000000000000000000.pdf", rows[0]["filename"])
+        self.assertEqual(rows[0]["filename"],
+                         fulltext.wanted_filename(None, None, paper["s2_paper_id"]))
+
+    def test_every_download_goes_through_the_proxy(self):
+        """The proxy is unconditional; the path behind it is per-publisher.
+
+        A paywalled paper needs the proxy and an open-access one is unharmed by
+        it, so the link never asks the reader to judge which kind it is.
+        """
+        from tools.impact import worklist
+        ieee = worklist.download_url("10.1109/icse55347.2025.00003", None, "IEEE")
+        self.assertEqual(worklist.PROXY + "https://doi.org/"
+                         "10.1109/icse55347.2025.00003", ieee)
+        acm = worklist.download_url("10.1145/3769828", None, "ACM")
+        self.assertEqual(worklist.PROXY + "https://dl.acm.org/doi/pdf/"
+                         "10.1145/3769828", acm)
+        springer = worklist.download_url("10.1007/978-3-031-1", None, "Springer")
+        self.assertEqual(worklist.PROXY + "https://link.springer.com/content/"
+                         "pdf/10.1007/978-3-031-1.pdf", springer)
+        other = worklist.download_url(None, "https://example.org/p", "other")
+        self.assertEqual(worklist.PROXY + "https://example.org/p", other)
+        self.assertEqual("", worklist.download_url(None, None, "other"))
+
+    def test_papers_with_a_known_relationship_come_first(self):
+        from tools.impact import worklist
+        settled = self._paper(doi="10.1145/2", title="Settled")
+        settled["relationships"]["uses_infrastructure"]["value"] = "yes"
+        rows = worklist.wanted([self._paper(doi="10.1145/1", title="Unknown"),
+                                settled])
+        self.assertEqual("Settled", rows[0]["title"])
+
+
+class PeopleRosterTest(unittest.TestCase):
+    """Resolving a reporter to a GitHub profile, which the searches depend on."""
+
+    def _gh(self, authors):
+        class FakeGitHub:
+            def issue(self, owner, repo, number, **kwargs):
+                login = authors.get(f"{owner}/{repo}#{number}")
+                return ({"user": {"login": login}} if login else None), False
+        return FakeGitHub()
+
+    def _urls(self, count, repo="a/b"):
+        return [f"https://github.com/{repo}/issues/{i}" for i in range(1, count + 1)]
+
+    def test_a_handle_is_read_off_the_person_s_own_issues(self):
+        from tools.impact.collectors import people
+        urls = self._urls(4)
+        authors = {u.replace("https://github.com/", "").replace("/issues/", "#"):
+                   "suyZhong" for u in urls}
+        handle, evidence = people.resolve_handle(self._gh(authors), urls)
+        self.assertEqual("suyZhong", handle)
+        self.assertEqual(1, len(evidence))
+        self.assertIn(evidence[0], urls)
+
+    def test_disagreeing_samples_resolve_to_nothing(self):
+        """Whoever is credited need not be whoever filed it; do not pick a winner."""
+        from tools.impact.collectors import people
+        urls = self._urls(4)
+        keys = [u.replace("https://github.com/", "").replace("/issues/", "#")
+                for u in urls]
+        authors = dict(zip(keys, ["alice", "alice", "alice", "bob"]))
+        handle, evidence = people.resolve_handle(self._gh(authors), urls)
+        self.assertIsNone(handle)
+        self.assertEqual([], evidence)
+
+    def test_too_few_samples_resolve_to_nothing(self):
+        from tools.impact.collectors import people
+        urls = self._urls(2)
+        keys = [u.replace("https://github.com/", "").replace("/issues/", "#")
+                for u in urls]
+        handle, _ = people.resolve_handle(self._gh(dict.fromkeys(keys, "alice")), urls)
+        self.assertIsNone(handle)
+
+    def test_a_bot_is_never_resolved_to(self):
+        from tools.impact.collectors import people
+        urls = self._urls(4)
+        keys = [u.replace("https://github.com/", "").replace("/issues/", "#")
+                for u in urls]
+        handle, _ = people.resolve_handle(
+            self._gh(dict.fromkeys(keys, "dependabot[bot]")), urls)
+        self.assertIsNone(handle)
+
+    def test_author_searches_carry_no_search_term(self):
+        """The whole point: reach reports whose text never names SQLancer."""
+        from tools.impact.collectors import people
+        queries = people.search_queries(["suyZhong", "DerZc"])
+        self.assertEqual(["author:suyZhong type:issue", "author:DerZc type:issue"],
+                         queries)
+        for query in queries:
+            self.assertNotIn("SQLancer", query)
+
+    def test_one_person_recorded_twice_becomes_one_entry(self):
+        from tools.impact.collectors import people
+        merged = people._merge_by_handle([
+            {"id": "person:x", "name": "xhandle", "github": "x",
+             "handle_is_verified": True, "reports_on_lab_list": 3,
+             "evidence": [{"source_url": "https://example.org/a"}]},
+            {"id": "person:x", "name": "X Name", "github": "x",
+             "handle_is_verified": False, "reports_on_lab_list": 2,
+             "evidence": [{"source_url": "https://example.org/b"}]},
+        ])
+        self.assertEqual(1, len(merged))
+        self.assertEqual(5, merged[0]["reports_on_lab_list"])
+        self.assertEqual(["X Name"], merged[0]["also_recorded_as"])
+        self.assertEqual(2, len(merged[0]["evidence"]))
+
+
+class StateOfTheArtTest(unittest.TestCase):
+    """Recognition is read off the citing sentence, which is the evidence."""
+
+    def setUp(self):
+        from tools.impact.collectors.papers import state_of_the_art_contexts
+        self.detect = state_of_the_art_contexts
+        self.tax = taxonomy.load()
+
+    def _detect(self, context, technique="tlp"):
+        return self.detect([(technique, context)], self.tax)
+
+    def test_a_technique_named_among_state_of_the_art_approaches_counts(self):
+        found = self._detect(
+            "We further compared ERIQ with four state-of-the-art DBMS logic bug "
+            "detection approaches: EDC [4], Radar [34], EET [11], and TLP [30].")
+        self.assertEqual(1, len(found))
+        self.assertEqual("tlp", found[0][0])
+
+    def test_the_tool_being_called_state_of_the_art_counts(self):
+        found = self._detect(
+            "SQLancer is the state-of-the-art tool for discovering logic bugs "
+            "in DBMS using metamorphic testing [40-42].")
+        self.assertEqual(1, len(found))
+
+    def test_alternative_hyphenation_is_matched(self):
+        self.assertEqual(1, len(self._detect(
+            "NoREC is a state of the art oracle for optimisation bugs.")))
+
+    def test_a_sentence_without_the_phrase_does_not_count(self):
+        self.assertEqual([], self._detect(
+            "We compared our approach with TLP and NoREC on five systems."))
+
+    def test_a_sentence_without_a_sqlancer_name_does_not_count(self):
+        self.assertEqual([], self._detect(
+            "We compare against three state-of-the-art fuzzers.", "tlp"))
+
+    def test_recognition_of_an_excluded_technique_is_not_recognition_of_sqlancer(self):
+        """A sentence naming only EET or DQE says nothing about SQLancer."""
+        self.assertEqual([], self._detect(
+            "We compare with two state-of-the-art approaches: EET [11] and "
+            "EDC [4].", "tlp"))
+
+    def test_an_ambiguous_acronym_still_needs_database_context(self):
+        self.assertEqual([], self._detect(
+            "We use the state-of-the-art TLP power management daemon.", "tlp"))
+
+    def test_published_records_carry_the_sentence_as_evidence(self):
+        for record in config.load_json(config.DATA_FILES["papers"])["papers"]:
+            entry = record["relationships"].get("describes_as_state_of_the_art")
+            if not entry or entry.get("value") != "yes":
+                continue
+            self.assertTrue(entry.get("evidence"))
+            for item in entry["evidence"]:
+                self.assertTrue(item.get("excerpt"),
+                                f"{record['id']} has recognition without a quote")
+
+    def test_recognition_is_not_counted_as_building_on(self):
+        from tools.impact.stats import BUILDING_ON
+        self.assertNotIn("describes_as_state_of_the_art", BUILDING_ON)
+
+
+class AuthoritativeMergeTest(unittest.TestCase):
+    """A record can leave the dataset, but only on a run that re-derived it."""
+
+    def setUp(self):
+        self.tmp = Path(tempfile.mkdtemp())
+        self.files = {"papers": self.tmp / "papers.json"}
+
+    def tearDown(self):
+        shutil.rmtree(self.tmp, ignore_errors=True)
+
+    def _dataset(self):
+        dataset = Dataset(self.files)
+        dataset.ensure("papers", {"schema_version": "1.0.0",
+                                  "policy_version": "test"})
+        return dataset
+
+    def _record(self, number):
+        return {"id": f"paper:doi:10.1/{number}", "title": f"Paper {number}",
+                "year": 2024, "cites_seed_techniques": ["tlp"],
+                "relationships": {}, "provenance": {}}
+
+    def test_an_incremental_run_never_removes(self):
+        dataset = self._dataset()
+        dataset.merge("papers", [self._record(1), self._record(2)],
+                      timestamp="2026-01-01T00:00:00Z")
+        dataset.save(["papers"])
+
+        dataset = self._dataset()
+        report = dataset.merge("papers", [self._record(1)],
+                               timestamp="2026-02-01T00:00:00Z")
+        self.assertEqual([], report.removed)
+        self.assertEqual(2, len(dataset.records("papers")))
+
+    def test_an_authoritative_run_retires_what_it_no_longer_finds(self):
+        dataset = self._dataset()
+        dataset.merge("papers", [self._record(1), self._record(2)],
+                      timestamp="2026-01-01T00:00:00Z")
+        dataset.save(["papers"])
+
+        dataset = self._dataset()
+        report = dataset.merge("papers", [self._record(1)],
+                               timestamp="2026-02-01T00:00:00Z",
+                               authoritative=True)
+        self.assertEqual(1, len(report.removed))
+        self.assertEqual("paper:doi:10.1/2", report.removed[0]["id"])
+        self.assertEqual(["paper:doi:10.1/1"],
+                         [r["id"] for r in dataset.records("papers")])
+
+    def test_an_authoritative_run_that_finds_everything_removes_nothing(self):
+        dataset = self._dataset()
+        candidates = [self._record(1), self._record(2)]
+        dataset.merge("papers", candidates, timestamp="2026-01-01T00:00:00Z")
+        dataset.save(["papers"])
+
+        dataset = self._dataset()
+        report = dataset.merge("papers", candidates,
+                               timestamp="2026-02-01T00:00:00Z",
+                               authoritative=True)
+        self.assertEqual([], report.removed)
+        self.assertFalse(dataset.save(["papers"])["papers"],
+                         "an unchanged authoritative run must rewrite nothing")
+
+
+class BackoffTest(unittest.TestCase):
+    """A 403 is only a throttle when the response says so."""
+
+    def test_a_github_style_throttle_is_retried(self):
+        from tools.impact.http import _is_rate_limited
+        self.assertTrue(_is_rate_limited({"Retry-After": "60"}))
+        self.assertTrue(_is_rate_limited({"X-RateLimit-Remaining": "0"}))
+
+    def test_a_plain_refusal_is_not_retried(self):
+        from tools.impact.http import _is_rate_limited
+        self.assertFalse(_is_rate_limited({}))
+        self.assertFalse(_is_rate_limited({"X-RateLimit-Remaining": "4999"}))
+        self.assertFalse(_is_rate_limited({"Server": "Apache"}))
+
+
+class LabImportTest(unittest.TestCase):
+    """The TEST lab list is a source of candidates, never of attributions."""
+
+    ENTRIES = [
+        {"url": "https://github.com/duckdb/duckdb/issues/1",
+         "title": "Wrong result with a filter", "created_at": "01/02/2023",
+         "state": "closed", "resolution": "fixed", "domain": "dbms",
+         "system": "DuckDB", "reported_by": "someone"},
+        {"url": "https://github.com/duckdb/duckdb/issues/2",
+         "title": "Crash in the parser", "created_at": "02/02/2023",
+         "state": "closed", "resolution": "fixed", "domain": "dbms",
+         "system": "DuckDB", "reported_by": "someone"},
+        {"url": "https://github.com/php/php-src/issues/3",
+         "title": "Not a database bug", "created_at": "03/02/2023",
+         "state": "open", "resolution": "open", "domain": "compiler&interpreter",
+         "system": "php-src", "reported_by": "someone"},
+    ]
+
+    BODIES = {
+        1: "Wrong result with a filter\n\nFound by SQLancer using the NoREC "
+           "oracle while testing DuckDB.",
+        2: "Crash in the parser\n\nReduced from a fuzzing run with AFL.",
+    }
+
+    class FakeGitHub:
+        def __init__(self, bodies):
+            self.bodies = bodies
+            self.fetched = []
+
+        def issue(self, owner, repo, number, **kwargs):
+            self.fetched.append(number)
+            body = self.bodies.get(number)
+            if body is None:
+                return None, False
+            return {"title": "", "body": body}, False
+
+    def _collect(self, entries=None, gh=None):
+        gh = gh or self.FakeGitHub(self.BODIES)
+        return nus_test.collect(gh, None, fetcher=None, entries=entries or self.ENTRIES,
+                                timestamp="2026-01-01T00:00:00Z")
+
+    def test_only_reports_naming_sqlancer_are_imported(self):
+        records, rejected, _ = self._collect()
+        self.assertEqual(1, len(records))
+        self.assertEqual("norec", records[0]["technique"])
+        self.assertEqual("duckdb", records[0]["dbms"])
+        self.assertTrue(any("no SQLancer tool" in entry["reason"]
+                            for entry in rejected))
+
+    def test_a_lab_bug_found_with_another_tool_is_not_attributed(self):
+        records, rejected, _ = self._collect(entries=[self.ENTRIES[1]])
+        self.assertEqual([], records)
+        self.assertEqual(1, len(rejected))
+
+    def test_non_database_domains_are_skipped_without_a_fetch(self):
+        gh = self.FakeGitHub(self.BODIES)
+        index = [entry for entry in self.ENTRIES
+                 if entry["domain"] == nus_test.DBMS_DOMAIN]
+        self._collect(entries=index, gh=gh)
+        self.assertNotIn(3, gh.fetched)
+
+    def test_the_lab_page_is_recorded_as_corroborating_evidence(self):
+        records, _, _ = self._collect()
+        sources = [item["source_url"] for item in
+                   records[0]["attribution"]["evidence"]]
+        self.assertIn(nus_test.PAGE_URL, sources)
+        self.assertIn("https://github.com/duckdb/duckdb/issues/1", sources)
+
+    def test_the_run_budget_defers_rather_than_drops(self):
+        records, _, needs_review = nus_test.collect(
+            self.FakeGitHub(self.BODIES), None, fetcher=None,
+            entries=self.ENTRIES[:2], max_reports=1,
+            timestamp="2026-01-01T00:00:00Z")
+        deferred = [entry for entry in needs_review
+                    if "run budget" in entry["reason"]]
+        self.assertEqual(1, len(deferred))
+
+    def test_upstream_resolutions_map_onto_the_status_vocabulary(self):
+        allowed = {"fixed", "fixed_in_documentation", "verified", "open",
+                   "closed_not_a_bug", "closed_duplicate", "unknown"}
+        for status, _ in nus_test.RESOLUTION_STATUS.values():
+            self.assertIn(status, allowed)
+        self.assertEqual(("unknown", False),
+                         nus_test._status_of({"resolution": "???", "state": ""}))
+
+
+class PlotTest(unittest.TestCase):
+    def setUp(self):
+        self.stats = config.load_json(config.DATA_FILES["stats"])
+        self.charts = plots.build_all(self.stats)
+
+    def test_every_chart_is_well_formed_svg(self):
+        for name, markup in self.charts.items():
+            self.assertTrue(markup.startswith("<svg"), name)
+            self.assertTrue(markup.rstrip().endswith("</svg>"), name)
+            import xml.etree.ElementTree as ET
+            ET.fromstring(markup)
+
+    def test_bar_counts_match_the_records(self):
+        """Every value drawn must appear in the statistics it came from."""
+        expected = {str(row["count"]) for row in self.stats["bugs"]["by_dbms"]}
+        markup = self.charts["bugs-by-dbms"]
+        for count in expected:
+            self.assertIn(f">{count}</text>", markup)
+
+    def test_the_paper_relationship_chart_matches_the_paper_records(self):
+        """Every bar's tooltip carries the count the statistics hold.
+
+        The series drawn are the reported ones: reuse and extension share a
+        bar, because a paper can do both and two bars would count it twice.
+        The halves are still in the data and are checked to sum to at least
+        the combined figure, which they must if the union is deduplicated.
+        """
+        yearly = self.stats["papers"]["by_relationship_year"]
+        markup = self.charts["papers-relationships-by-year"]
+        drawn = ("reusing_or_extending", "compares_with",
+                 "describes_as_state_of_the_art")
+        for key in drawn:
+            for entry in yearly.get(key, []):
+                if entry["count"] > 0:
+                    self.assertIn(f": {entry['count']}</title>", markup)
+        for index, entry in enumerate(yearly["reusing_or_extending"]):
+            halves = (yearly["uses_infrastructure"][index]["count"]
+                      + yearly["extends_technique"][index]["count"])
+            self.assertGreaterEqual(halves, entry["count"], entry["label"])
+            self.assertLessEqual(entry["count"], halves, entry["label"])
+
+    def test_committed_plots_are_up_to_date(self):
+        for name, markup in self.charts.items():
+            path = config.PLOT_DIR / f"{name}.svg"
+            self.assertTrue(path.exists(), f"{name}.svg is missing")
+            self.assertEqual(markup, path.read_text(encoding="utf-8"),
+                             f"{name}.svg is stale; run "
+                             f"`python -m tools.impact.run plots`")
+
+
+class RoundingTest(unittest.TestCase):
+    def test_headline_rounds_down(self):
+        cases = {0: "0", 7: "7", 12: "10+", 99: "90+", 456: "400+",
+                 1483: "1,000+", 2000: "2,000+"}
+        for total, expected in cases.items():
+            self.assertEqual(expected, stats_module.round_down_headline(total))
+
+
+if __name__ == "__main__":
+    unittest.main()
+
+
+class IntakeReportingTest(unittest.TestCase):
+    """Every downloaded PDF is accounted for by name.
+
+    A file the intake cannot place used to be counted and then dropped from the
+    report. To the person who downloaded it that is indistinguishable from a
+    file that was filed, so they download it again -- which is exactly what
+    happened before this was fixed.
+    """
+
+    def _pdf(self, tmp, name, text):
+        """A one-page PDF whose page text is ``text``."""
+        import pypdf
+        from pypdf.generic import DecodedStreamObject, NameObject
+        writer = pypdf.PdfWriter()
+        writer.add_blank_page(width=612, height=792)
+        stream = DecodedStreamObject()
+        escaped = text.replace("\\", r"\\").replace("(", r"\(").replace(")", r"\)")
+        stream.set_data(f"BT /F1 12 Tf 72 700 Td ({escaped}) Tj ET".encode())
+        page = writer.pages[0]
+        page[NameObject("/Contents")] = writer._add_object(stream)
+        path = tmp / name
+        with open(path, "wb") as handle:
+            writer.write(handle)
+        return path
+
+    def test_a_pdf_matching_nothing_is_named_in_the_report(self):
+        import io, contextlib, tempfile, pathlib
+        from tools.impact import intake
+        with tempfile.TemporaryDirectory() as raw:
+            tmp = pathlib.Path(raw)
+            self._pdf(tmp, "mystery.pdf", "An unrelated paper about nothing")
+            buf = io.StringIO()
+            with contextlib.redirect_stdout(buf):
+                counts = intake.run(tmp, dry_run=True, limit=5,
+                                    destination=tmp / "filed",
+                                    candidates=[{"title": "Some Other Paper",
+                                                 "filename": "other.pdf",
+                                                 "doi": "10.1/other"}],
+                                    wanted={"other.pdf"})
+            self.assertEqual(1, counts["unmatched"])
+            self.assertIn("mystery.pdf", buf.getvalue())
+            self.assertEqual(1, len(counts["_unresolved"]))
+
+
+class DbmsPagesTest(unittest.TestCase):
+    """One page per database system, generated from the statistics."""
+
+    def _stats(self, rows):
+        return {"dbms": {"rows": rows}}
+
+    def test_a_system_with_nothing_recorded_gets_no_page(self):
+        """An empty page is worse than no link to one."""
+        from tools.impact import pages
+        rows = [{"id": "known", "name": "Known", "bugs": 3, "supported": False,
+                 "adoption_relationships": [], "planned_adoption": False},
+                {"id": "silent", "name": "Silent", "bugs": 0, "supported": False,
+                 "adoption_relationships": [], "planned_adoption": False}]
+        wanted = [row["id"] for row in pages.wanted(self._stats(rows))]
+        self.assertEqual(["known"], wanted)
+
+    def test_a_supported_system_without_bugs_still_gets_one(self):
+        from tools.impact import pages
+        rows = [{"id": "fresh", "name": "Fresh", "bugs": 0, "supported": True,
+                 "adoption_relationships": [], "planned_adoption": False}]
+        self.assertEqual(["fresh"],
+                         [row["id"] for row in pages.wanted(self._stats(rows))])
+
+    def test_the_excerpt_keeps_a_systems_own_capitalisation(self):
+        """``capitalize`` would turn DuckDB into duckdb."""
+        from tools.impact import pages
+        text = pages.excerpt_for({"id": "duckdb", "name": "DuckDB", "bugs": 12,
+                                  "adoption_relationships": [],
+                                  "planned_adoption": False})
+        self.assertIn("DuckDB", text)
+        self.assertTrue(text.startswith("12 bugs"), text)
+
+    def test_a_system_that_leaves_the_dataset_loses_its_page(self):
+        import tempfile, pathlib
+        from tools.impact import pages
+        with tempfile.TemporaryDirectory() as raw:
+            tmp = pathlib.Path(raw)
+            (tmp / "gone.html").write_text("stale", encoding="utf-8")
+            rows = [{"id": "here", "name": "Here", "bugs": 1, "supported": False,
+                     "adoption_relationships": [], "planned_adoption": False}]
+            outcome = pages.write_all(self._stats(rows), tmp)
+            self.assertEqual("removed", outcome["gone"])
+            self.assertFalse((tmp / "gone.html").exists())
+            self.assertTrue((tmp / "here.html").exists())
+
+    def test_the_chart_links_each_bar_to_its_system(self):
+        from tools.impact import plots
+        svg = plots.stacked_horizontal_bars(
+            [{"key": "duckdb", "label": "DuckDB", "count": 4,
+              "segments": [{"key": "project", "label": "Project", "count": 4}]}],
+            title="t", value_label="v", link_base="/impact/dbms/")
+        self.assertIn('<a href="/impact/dbms/duckdb/"', svg)
+        self.assertEqual(1, svg.count("</a>"))
+
+    def test_a_chart_without_a_link_base_has_no_links(self):
+        from tools.impact import plots
+        svg = plots.stacked_horizontal_bars(
+            [{"key": "duckdb", "label": "DuckDB", "count": 4,
+              "segments": [{"key": "project", "label": "Project", "count": 4}]}],
+            title="t", value_label="v")
+        self.assertNotIn("<a ", svg)
+
+
+class AffiliationRulingTest(unittest.TestCase):
+    """A person can be on the lab's roster and still count as external.
+
+    The roster can see that someone is listed on the lab's people page. It
+    cannot see whether their work is this project's, and a lab lists everyone
+    working in it -- so that judgement is recorded rather than inferred.
+    """
+
+    def test_without_a_ruling_the_roster_decides(self):
+        """An empty decisions file leaves every record exactly as it was."""
+        from tools.impact import validate
+        from tools.impact.collectors import people
+        original = people.decisions
+        people.decisions = lambda: {"counts_as_external": []}
+        try:
+            problems = validate.check_affiliation_rulings(
+                [{"id": "bug:x:1", "reporter": "someone",
+                  "reporter_affiliation": "project"}],
+                [{"github": "someone", "name": "Someone"}])
+        finally:
+            people.decisions = original
+        self.assertEqual([], problems)
+
+    def test_the_ruling_in_force_puts_its_person_outside_the_project(self):
+        from tools.impact.collectors import people
+        self.assertIn("joyemang33", people.counted_as_external())
+
+    def test_a_stale_record_is_reported(self):
+        from tools.impact import validate
+        from tools.impact.collectors import people
+        ruling = {"counts_as_external": [
+            {"github": "someone", "reason": "x" * 25}]}
+        original = people.decisions
+        people.decisions = lambda: ruling
+        try:
+            problems = validate.check_affiliation_rulings(
+                [{"id": "bug:x:1", "reporter": "someone",
+                  "reporter_affiliation": "project"}],
+                [{"github": "someone", "name": "Someone"}])
+        finally:
+            people.decisions = original
+        self.assertEqual(1, len(problems))
+        self.assertIn("re-run collect", str(problems[0]))
+
+    def test_a_misspelled_handle_is_reported(self):
+        """A ruling that matches nobody changes nothing and says nothing."""
+        from tools.impact import validate
+        from tools.impact.collectors import people
+        ruling = {"counts_as_external": [
+            {"github": "nobody-here", "reason": "x" * 25}]}
+        original = people.decisions
+        people.decisions = lambda: ruling
+        try:
+            problems = validate.check_affiliation_rulings(
+                [], [{"github": "someone", "name": "Someone"}])
+        finally:
+            people.decisions = original
+        self.assertEqual(1, len(problems))
+        self.assertIn("check the spelling", str(problems[0]))
+
+
+class GitHubCredentialTest(unittest.TestCase):
+    """A rejected token fails loudly.
+
+    A 401 used to surface as an empty result set: `_request` raised
+    RuntimeError, `search` caught it and returned, and the run reported no new
+    bugs. An expired token and a world with no bugs looked exactly alike.
+    """
+
+    class _Fetcher:
+        def __init__(self, status):
+            self.status = status
+            self.calls = 0
+
+        def fetch(self, url, **kwargs):
+            self.calls += 1
+            return {"status": self.status, "body": "{}"}, False
+
+    def _client(self, status):
+        from tools.impact.github import GitHub
+        gh = GitHub.__new__(GitHub)
+        gh.search_fetcher = self._Fetcher(status)
+        gh.failed_searches = []
+        return gh
+
+    def test_a_rejected_token_raises(self):
+        from tools.impact.github import BadCredentials
+        gh = self._client(401)
+        with self.assertRaises(BadCredentials):
+            list(gh.search("issues", "anything"))
+
+    def test_another_failure_is_recorded_rather_than_raised(self):
+        """A single dead search must not abort a run, but must leave a trace."""
+        gh = self._client(422)
+        self.assertEqual([], list(gh.search("issues", "anything")))
+        self.assertEqual([("issues", "anything", 422)], gh.failed_searches)
+
+    def test_a_repo_scope_rejects_an_unregistered_repository(self):
+        from tools.impact.collectors import github_bugs
+        with self.assertRaises(ValueError):
+            github_bugs.collect(None, only_repos=["nobody/nothing"])
+
+
+class EmailedBugTest(unittest.TestCase):
+    """Bugs the lab reported privately, which have no link to follow."""
+
+    ENTRIES = [
+        {"system": "Umbra", "domain": "dbms", "url": None,
+         "title": "Unexpected Results when Comparing Boolean Values",
+         "created_at": "11/27/2023", "resolution": "fixed",
+         "reported_by": "Suyang Zhong"},
+        {"system": "Nowhere", "domain": "dbms", "url": None,
+         "title": "Something", "created_at": "01/02/2024",
+         "resolution": "fixed", "reported_by": "Someone"},
+    ]
+
+    def test_a_report_without_a_link_still_becomes_a_record(self):
+        from tools.impact.collectors import nus_test
+        records, review = nus_test.emailed_records(
+            self.ENTRIES, {"umbra": "umbra"}, timestamp="2026-01-01T00:00:00Z")
+        self.assertEqual(1, len(records))
+        record = records[0]
+        self.assertEqual("umbra", record["dbms"])
+        self.assertEqual("curated_primary_source", record["attribution"]["rule"])
+        self.assertEqual("fixed", record["status"])
+        self.assertTrue(record["status_is_true_positive"])
+        # No report to point at, so no primary_url; pointing every one of them
+        # at the list page would make many bugs look like one duplicated.
+        self.assertIsNone(record.get("primary_url"))
+        self.assertEqual(nus_test.PAGE_URL, record["links"]["record"])
+
+    def test_an_unregistered_system_is_reported_not_guessed(self):
+        from tools.impact.collectors import nus_test
+        _, review = nus_test.emailed_records(
+            self.ENTRIES, {"umbra": "umbra"}, timestamp="2026-01-01T00:00:00Z")
+        self.assertEqual(1, len(review))
+        self.assertIn("Nowhere", review[0]["reason"])
+
+
+class SlashDateTest(unittest.TestCase):
+    """Slash dates carry no order, and sources are not consistent about it.
+
+    The TEST lab's list has 735 rows that can only be day-first and 147 that
+    can only be month-first. The month-first ones matched no format at all and
+    came back as no date, which is how 55 Umbra bugs arrived undated.
+    """
+
+    def test_a_day_above_twelve_settles_the_order(self):
+        from tools.impact.util import parse_date
+        self.assertEqual("2023-11-27", parse_date("11/27/2023"))
+        self.assertEqual("2025-06-30", parse_date("06/30/2025"))
+
+    def test_a_month_above_twelve_still_reads_day_first(self):
+        from tools.impact.util import parse_date
+        self.assertEqual("2023-11-27", parse_date("27/11/2023"))
+
+    def test_an_ambiguous_date_keeps_the_previous_reading(self):
+        """Nothing in the value can settle it, so the behaviour is unchanged."""
+        from tools.impact.util import parse_date
+        self.assertEqual("2024-08-02", parse_date("2/8/2024"))
+
+    def test_an_impossible_date_is_no_date(self):
+        from tools.impact.util import parse_date
+        self.assertIsNone(parse_date("13/13/2020"))
+
+
+class TechniqueOnlyReportTest(unittest.TestCase):
+    """A report naming an oracle but never the tool.
+
+    Every sentence pattern is anchored on the tool's own names, so a report
+    saying only "the NoREC-transformed query returns 0" had techniques matched
+    and then no sentence to quote, and was thrown away. MariaDB's Jira is full
+    of these: the reporters describe the transformation, not the tool.
+    """
+
+    def _issue(self, title, body):
+        return {"title": title, "body": body, "labels": [{"name": "Bug"}]}
+
+    def test_an_oracle_name_is_its_own_anchor(self):
+        from tools.impact import taxonomy
+        from tools.impact.collectors import github_bugs
+        text = ("NoREC logical bug: the original query returns COUNT(*) = 4, "
+                "while the NoREC-transformed query returns 0.")
+        found = github_bugs.deterministic_attribution(
+            text, taxonomy.load(), self._issue("NoREC logical bug", text))
+        self.assertIsNotNone(found)
+        rule, technique, _, excerpt = found
+        self.assertEqual("norec", technique)
+        self.assertIn(excerpt, text)
+
+    def test_an_ambiguous_acronym_still_needs_the_tool_named(self):
+        """CERT in a cluster report is a TLS certificate, not an oracle.
+
+        Ambiguous acronyms need database context to count as a technique, and
+        in a DBMS tracker that context is always present -- every issue is
+        about a database. Twenty-three Galera issues from 2013 to 2017, years
+        before SQLancer existed, were admitted as cardinality estimation
+        testing before this was required.
+        """
+        from tools.impact import taxonomy
+        from tools.impact.collectors import github_bugs
+        text = ("Galera: WSREP fails to accept state transfer because the "
+                "cert index and the database query log disagree.")
+        found = github_bugs.deterministic_attribution(
+            text, taxonomy.load(), self._issue("Galera: WSREP fails", text))
+        self.assertIsNone(found)
+
+
+class ReviewQueueTest(unittest.TestCase):
+    """Candidates a collector could not place, kept between runs.
+
+    The list used to live only in the run report. Three real gaps were found
+    by hand that it had already flagged and thrown away: MatrixOne's bugs,
+    Oxla's, and a TimescaleDB provider in a fork.
+    """
+
+    ITEM = {"kind": "bug", "url": "https://github.com/a/b/issues/1",
+            "reason": "repository is not a registered database system",
+            "title": "Something"}
+
+    def _queue(self):
+        return {"schema_version": "1.0.0", "open": [], "dismissed": []}
+
+    def test_an_item_is_kept_and_recognised_next_run(self):
+        from tools.impact import review
+        queue = self._queue()
+        first = review.merge(queue, [self.ITEM], source="github_bugs",
+                             timestamp="2026-01-01T00:00:00Z")
+        second = review.merge(queue, [self.ITEM], source="github_bugs",
+                              timestamp="2026-02-01T00:00:00Z")
+        self.assertEqual(1, first["added"])
+        self.assertEqual(1, second["seen_again"])
+        self.assertEqual(1, len(queue["open"]))
+        row = queue["open"][0]
+        self.assertEqual("2026-01-01T00:00:00Z", row["first_seen"])
+        self.assertEqual("2026-02-01T00:00:00Z", row["last_seen"])
+
+    def test_an_item_leaves_when_its_record_is_admitted(self):
+        from tools.impact import review
+        queue = self._queue()
+        review.merge(queue, [self.ITEM], source="github_bugs")
+        counts = review.merge(queue, [], source="github_bugs",
+                              admitted={"https://github.com/a/b/issues/1"})
+        self.assertEqual(1, counts["resolved"])
+        self.assertEqual([], queue["open"])
+
+    def test_absence_alone_does_not_resolve_an_item(self):
+        """A run that did not reach an item looks like one that resolved it."""
+        from tools.impact import review
+        queue = self._queue()
+        review.merge(queue, [self.ITEM], source="github_bugs")
+        review.merge(queue, [], source="github_bugs")
+        self.assertEqual(1, len(queue["open"]))
+
+    def test_a_dismissed_item_is_never_raised_again(self):
+        from tools.impact import review
+        queue = self._queue()
+        review.merge(queue, [self.ITEM], source="github_bugs")
+        queue["dismissed"] = [{"id": queue["open"][0]["id"],
+                               "reason": "not a database system"}]
+        queue["open"] = []
+        counts = review.merge(queue, [self.ITEM], source="github_bugs")
+        self.assertEqual(0, counts["added"])
+        self.assertEqual([], queue["open"])
+
+
+class ReproducerBackedAcronymTest(unittest.TestCase):
+    """An ambiguous acronym may anchor when the reproducer backs it.
+
+    "TLP equivalence verification" over a generated t0(c0 ...) schema is a
+    SQLancer oracle; "cert index" in a Galera cluster report is a TLS
+    certificate. The reproducer is what separates them, and it works in any
+    language -- openGauss's tracker is in Chinese, where no English sentence
+    pattern can reach.
+    """
+
+    def _issue(self, title, body):
+        return {"title": title, "body": body, "labels": [{"name": "Bug"}]}
+
+    def test_a_generated_schema_lets_tlp_anchor(self):
+        from tools.impact import taxonomy
+        from tools.impact.collectors import github_bugs
+        text = ("TLP equivalence verification reports inconsistent results\n"
+                "Test type: SQL functionality. Database version: 7.0.0\n"
+                "CREATE UNLOGGED TABLE t0(c0 boolean PRIMARY KEY UNIQUE);\n"
+                "CREATE TABLE IF NOT EXISTS t1(c0 TEXT PRIMARY KEY, c1 money);\n"
+                "SELECT SUM(t1.c1) FROM t1 HAVING true;\n")
+        found = github_bugs.deterministic_attribution(
+            text, taxonomy.load(), self._issue("TLP equivalence", text))
+        self.assertIsNotNone(found)
+        self.assertEqual("tlp", found[1])
+
+    def test_without_a_reproducer_the_acronym_is_still_refused(self):
+        from tools.impact import taxonomy
+        from tools.impact.collectors import github_bugs
+        text = ("Galera: WSREP fails to accept state transfer because the "
+                "cert index and the database query log disagree.")
+        self.assertIsNone(github_bugs.deterministic_attribution(
+            text, taxonomy.load(), self._issue("Galera: WSREP fails", text)))
+
+
+class GiteeCollectorTest(unittest.TestCase):
+    """Gitee hosts trackers the GitHub search cannot see.
+
+    openGauss's is the case this exists for: a report titled "TLP 等价验证出现
+    结果内容不一致" over a generated t0(c0 ...) schema, invisible to every
+    other route.
+    """
+
+    def test_only_registered_repositories_are_searched(self):
+        """An unscoped sweep put a thousand throwaway repositories into the
+        review queue and reached no real report."""
+        from tools.impact.collectors import gitee_bugs
+        index = gitee_bugs.repository_index()
+        self.assertIn("opengauss/opengauss-server", index)
+        self.assertEqual("opengauss", index["opengauss/opengauss-server"])
+
+    def test_the_scope_carries_the_full_repository_name(self):
+        from tools.impact.collectors import gitee_bugs
+
+        seen = []
+
+        class _Fetcher:
+            def fetch(self, url, **kwargs):
+                seen.append(url)
+                return {"status": 200, "body": "[]"}, False
+
+        gitee_bugs.search(_Fetcher(), "TLP",
+                          repository="opengauss/openGauss-server")
+        self.assertIn("owner=opengauss", seen[0])
+        # The full owner/name, which is the form the API accepts; a bare
+        # repository name returns nothing.
+        self.assertIn("repo=opengauss/openGauss-server", seen[0])
+
+    def test_a_throttled_search_is_not_an_empty_one(self):
+        from tools.impact.collectors import gitee_bugs
+
+        class _Fetcher:
+            def fetch(self, url, **kwargs):
+                return {"status": 403, "body": ""}, False
+
+        del gitee_bugs.FAILED_SEARCHES[:]
+        self.assertEqual([], gitee_bugs.search(_Fetcher(), "TLP",
+                                               repository="a/b"))
+        self.assertEqual(1, len(gitee_bugs.FAILED_SEARCHES))
+
+    def test_a_rejected_issue_is_not_a_true_positive(self):
+        from tools.impact.collectors import gitee_bugs
+        self.assertEqual(("closed_not_a_bug", False),
+                         gitee_bugs.status_of({"state": "rejected"}))
+        self.assertEqual(("fixed", True),
+                         gitee_bugs.status_of({"state": "closed"}))
+        self.assertEqual(("open", True),
+                         gitee_bugs.status_of({"state": "progressing"}))
+
+    def test_a_run_stops_once_it_is_blocked(self):
+        """Gitee blocks by address, so continuing costs and gains nothing."""
+        from tools.impact.collectors import gitee_bugs
+        del gitee_bugs.FAILED_SEARCHES[:]
+        self.assertFalse(gitee_bugs._blocked())
+        for _ in range(gitee_bugs.MAX_CONSECUTIVE_FAILURES):
+            gitee_bugs.FAILED_SEARCHES.append(("TLP", "a/b", 403))
+        self.assertTrue(gitee_bugs._blocked())
+        del gitee_bugs.FAILED_SEARCHES[:]
+
+
+class UnreadableTrackerTest(unittest.TestCase):
+    """A link a reader cannot follow is not published as the report.
+
+    Oxla tracks its bugs in a private Jira, so twelve records pointed at
+    oxla.atlassian.net pages that return 404 to everybody. The public evidence
+    is the bug file in the fork, and that is what the record links.
+    """
+
+    ENTRY = {"constant": "bugOxla8408", "urls": ["https://x.atlassian.net/browse/OXLA-8408"],
+             "note": "See: https://x.atlassian.net/browse/OXLA-8408 A query fails.",
+             "excerpt": "/// A query fails."}
+    FILE = "https://github.com/o/r/blob/main/src/sqlancer/oxla/OxlaBugs.java"
+
+    def _record(self, readable):
+        from tools.impact.collectors import provider_bugs
+        return provider_bugs.build_record(
+            self.ENTRY, "oxla", url=self.ENTRY["urls"][0], source_url=self.FILE,
+            repository="o/r", timestamp="2026-01-01T00:00:00Z",
+            status=("unknown", True), readable=readable)
+
+    def test_an_unreadable_tracker_is_not_the_primary_url(self):
+        record = self._record(False)
+        self.assertIsNone(record.get("primary_url"))
+        self.assertEqual(self.FILE, record["links"]["record"])
+        self.assertEqual(self.ENTRY["urls"][0], record["links"]["tracker"])
+        self.assertIn("not public", record["attribution"]["evidence"][0]["note"])
+
+    def test_a_readable_tracker_is_linked_as_the_report(self):
+        record = self._record(True)
+        self.assertEqual(self.ENTRY["urls"][0], record["primary_url"])
+        self.assertEqual(self.ENTRY["urls"][0], record["links"]["report"])
+        self.assertNotIn("tracker", record["links"])
+
+    def test_only_a_definite_refusal_counts_as_unreadable(self):
+        """A 403 means this client is refused, not that nobody can read it."""
+        from tools.impact.collectors import provider_bugs
+
+        class _Fetcher:
+            def __init__(self, status):
+                self.status = status
+
+            def fetch(self, url, **kwargs):
+                return {"status": self.status}, False
+
+        self.assertFalse(provider_bugs.is_publicly_readable("u", _Fetcher(404)))
+        self.assertTrue(provider_bugs.is_publicly_readable("u", _Fetcher(403)))
+        self.assertTrue(provider_bugs.is_publicly_readable("u", _Fetcher(200)))
+        self.assertTrue(provider_bugs.is_publicly_readable("u", None))
+
+
+class ProviderBugUrlTest(unittest.TestCase):
+    """The field name says which of several cited issues the entry is about.
+
+    CockroachDB's file has an entry citing a closed bug and the underlying one
+    it waits on. Taking the first URL recorded issue 84078 for a field named
+    bug84154, and titled it with the maintainers' note above the field.
+    """
+
+    def test_the_field_name_picks_the_issue(self):
+        from tools.impact.collectors import provider_bugs
+        entry = {
+            "constant": "bug84154",
+            "urls": ["https://github.com/cockroachdb/cockroach/issues/84078",
+                     "https://github.com/cockroachdb/cockroach/issues/84154"],
+            "note": "The following bug is closed, but leave it enabled.",
+            "excerpt": "// The following bug is closed",
+        }
+        self.assertEqual("https://github.com/cockroachdb/cockroach/issues/84154",
+                         provider_bugs.bug_url(entry))
+
+    def test_a_single_url_is_used_whatever_the_name(self):
+        from tools.impact.collectors import provider_bugs
+        entry = {"constant": "bugOxla8408", "note": "x", "excerpt": "y",
+                 "urls": ["https://x.atlassian.net/browse/OXLA-8408"]}
+        self.assertEqual("https://x.atlassian.net/browse/OXLA-8408",
+                         provider_bugs.bug_url(entry))
+
+    def test_the_reports_own_title_wins(self):
+        """The comment above a field is sometimes a note, not a bug title."""
+        from tools.impact.collectors import provider_bugs
+        entry = {"constant": "bug84154", "note": "The following bug is closed",
+                 "excerpt": "// The following bug is closed",
+                 "urls": ["https://github.com/cockroachdb/cockroach/issues/84154"]}
+        record = provider_bugs.build_record(
+            entry, "cockroachdb", url=entry["urls"][0], source_url="file",
+            repository="sqlancer/sqlancer", timestamp="2026-01-01T00:00:00Z",
+            status=("fixed", True), title="sql: interval overflow panics")
+        self.assertEqual("sql: interval overflow panics", record["title"])
+
+
+class RecognitionHighlightTest(unittest.TestCase):
+    """The featured quotes are ones the records actually hold.
+
+    The highlights file names a paper and a mention id rather than copying the
+    sentence, so a quote on the page cannot say something the paper did not.
+    What it can do is go stale, and then the page quotes nothing.
+    """
+
+    def test_every_highlight_resolves(self):
+        from tools.impact import config, validate
+        papers = config.load_json(config.DATA_FILES["papers"])["papers"]
+        self.assertEqual([], [str(p) for p in
+                              validate.check_recognition_highlights(papers)])
+
+    def test_a_highlight_on_a_paper_that_no_longer_qualifies_is_caught(self):
+        from tools.impact import validate
+        papers = [{"id": "paper:doi:10.1109/icde65706.2026.00240",
+                   "relationships": {"describes_as_state_of_the_art":
+                                     {"value": "no"}}}]
+        problems = validate.check_recognition_highlights(papers)
+        self.assertTrue(any("no longer describes" in str(p) for p in problems))
+
+    def test_a_highlight_naming_an_unknown_paper_is_caught(self):
+        from tools.impact import validate
+        problems = validate.check_recognition_highlights([])
+        self.assertTrue(problems)
+        self.assertTrue(all("no paper with this id" in str(p)
+                            for p in problems))
+
+    def test_every_highlighted_paper_still_says_it(self):
+        """The nine are a subset of the papers the data classifies as such."""
+        import json
+        from tools.impact import config
+        highlights = config.load_json(
+            config.DATA_DIR / "recognition_highlights.json")["highlights"]
+        papers = {p["id"]: p for p in
+                  config.load_json(config.DATA_FILES["papers"])["papers"]}
+        for entry in highlights:
+            paper = papers[entry["paper_id"]]
+            self.assertEqual(
+                "yes",
+                paper["relationships"]["describes_as_state_of_the_art"]["value"],
+                entry["paper_id"])
+
+
+class TalkTranscriptTest(unittest.TestCase):
+    """A talk's words are read, not fetched, and the reading has to be honest.
+
+    Three things here are the whole point of the module, and each has already
+    gone wrong once: a mention has to land on the second it was said, distant
+    parts of a transcript must not run together into one remark, and a name the
+    transcriber got wrong has to be recorded as misheard rather than quietly
+    presented as what the speaker said.
+    """
+
+    def test_a_misheard_name_is_matched_and_recorded_as_misheard(self):
+        from tools.impact import talks
+        found = talks.findings_from_segments(
+            [{"start_ms": 295000,
+              "text": "as another example SQL lenser which is a fuzzer for "
+                      "database Management Systems"}],
+            url="https://www.youtube.com/watch?v=6YGqFRTe2D0")
+        self.assertEqual(1, len(found))
+        self.assertIn("sqlancer", found[0]["matched"])
+        self.assertEqual("SQL lenser", found[0]["heard_as"])
+        self.assertEqual("captions", found[0]["source"])
+
+    def test_distant_segments_do_not_become_one_remark(self):
+        """Captions carry no full stops, so only the clock can end a sentence."""
+        from tools.impact import talks
+        found = talks.findings_from_segments(
+            [{"start_ms": 295000, "text": "one example is SQL lenser"},
+             {"start_ms": 1509000, "text": "a quote about Manuel Rigger here"}],
+            url="https://www.youtube.com/watch?v=x")
+        self.assertEqual(2, len(found))
+        self.assertEqual([295, 1509], [m["at_seconds"] for m in found])
+        self.assertEqual(["4:55", "25:09"], [m["timestamp"] for m in found])
+
+    def test_the_link_points_at_the_moment_not_at_the_start(self):
+        from tools.impact import talks
+        long_lead = "and so " * 120
+        found = talks.findings_from_segments(
+            [{"start_ms": 0, "text": long_lead},
+             {"start_ms": 600000, "text": "this is where SQLancer comes up"}],
+            url="https://www.youtube.com/watch?v=x")
+        self.assertEqual(1, len(found))
+        self.assertEqual(600, found[0]["at_seconds"])
+        self.assertIn("t=600s", found[0]["url"])
+
+    def test_slides_are_quoted_as_written(self):
+        from tools.impact import talks
+        found = talks.findings_from_slides(
+            "Tests new code for edge cases. SQLancer — logical SQL fuzzer "
+            "Developed by Manuel Rigger at ETH Zurich. Added to ClickHouse.",
+            url="https://presentations.clickhouse.com/2021-cpp-siberia/index.html")
+        self.assertEqual(1, len(found))
+        self.assertEqual("slides", found[0]["source"])
+        self.assertIsNone(found[0]["heard_as"])
+        self.assertTrue(found[0]["excerpt"].startswith("SQLancer — logical"))
+
+    def test_every_stored_excerpt_is_in_its_captured_transcript(self):
+        """The excerpts on the site are substrings of what was captured."""
+        import json
+        from tools.impact import config, talks
+        for talk in talks.load()["talks"]:
+            for mention in talk["mentions"]:
+                if mention["source"] != "captions" or not mention["excerpt"]:
+                    continue
+                path = (talks.TRANSCRIPT_CACHE
+                        / f"{talk['video_id']}.json")
+                self.assertTrue(path.exists(), f"no capture for {talk['id']}")
+                with path.open() as handle:
+                    captured = json.load(handle)
+                haystack = " ".join(s["text"] for s in captured["segments"])
+                self.assertIn(mention["excerpt"], haystack)
+
+    def test_the_records_validate(self):
+        from tools.impact import talks, validate
+        self.assertEqual([], [str(p) for p in
+                              validate.check_talks(talks.load()["talks"])])
+
+    def test_a_caption_mention_without_its_moment_is_caught(self):
+        from tools.impact import validate
+        problems = validate.check_talks([{
+            "id": "talk:youtube:x",
+            "sources": [{"kind": "captions", "status": "extracted"}],
+            "mentions": [{"id": "M1", "source": "captions", "at_seconds": 295,
+                          "url": "https://www.youtube.com/watch?v=x",
+                          "matched": ["sqlancer"], "excerpt": "…",
+                          "excerpt_is_verbatim": True}],
+        }])
+        self.assertTrue(any("does not point at the moment" in str(p)
+                            for p in problems))
+
+    def test_a_talk_nobody_read_cannot_be_quoted(self):
+        from tools.impact import validate
+        problems = validate.check_talks([{
+            "id": "talk:youtube:x",
+            "sources": [{"kind": "watched", "status": "extracted"}],
+            "mentions": [{"id": "M1", "source": "watched",
+                          "url": "https://www.youtube.com/watch?v=x",
+                          "matched": ["sqlancer"],
+                          "excerpt": "words nobody transcribed",
+                          "excerpt_is_verbatim": True}],
+        }])
+        self.assertTrue(any("cannot carry a quotation" in str(p)
+                            for p in problems))
+
+    def test_every_talk_reaches_the_resource_list_and_a_page(self):
+        from tools.impact import config, pages, talks
+        urls = {r["url"] for r in
+                config.load_json(config.DATA_FILES["resources"])["resources"]
+                if r["type"] == "talk"}
+        for talk in talks.load()["talks"]:
+            self.assertIn(talk["url"], urls)
+            self.assertTrue(
+                (pages.TALK_PAGE_DIR / f"{pages.talk_slug(talk)}.html").exists(),
+                f"no page for {talk['id']}")
+
+
+class PeerToolTest(unittest.TestCase):
+    """The README's links section lists peer tools, not only derived ones.
+
+    Jepsen, SQLsmith and Squirrel are other people's database testing tools.
+    They belong in SQLancer's README -- they are what a reader should look at
+    next -- but a resource list that counts them is padding itself with work
+    that says nothing about SQLancer's reach.
+    """
+
+    def test_a_peer_tool_is_not_a_resource(self):
+        from tools.impact import config
+        urls = {r["url"] for r in
+                config.load_json(config.DATA_FILES["resources"])["resources"]}
+        for peer in ("https://github.com/jepsen-io",
+                     "https://github.com/anse1/sqlsmith",
+                     "https://github.com/s3team/Squirrel"):
+            self.assertNotIn(peer, urls)
+
+    def test_a_tool_the_readme_connects_to_sqlancer_is_kept(self):
+        """SQLRight earns its place: the line says it supports NoREC and TLP."""
+        from tools.impact import config
+        records = {r["url"]: r for r in
+                   config.load_json(config.DATA_FILES["resources"])["resources"]}
+        kept = records["https://github.com/PSU-Security-Universe/sqlright"]
+        self.assertIn("NoREC", kept["evidence"][0]["excerpt"])
+
+
+class TalkFrameTest(unittest.TestCase):
+    """Some talks name SQLancer only on a slide, which no transcript reaches.
+
+    A frame is the evidence there, and it is held to a rule the rest of the
+    dataset cannot enforce on it: nothing is transcribed off a picture. Every
+    other excerpt here is a substring of something fetched, and a check can say
+    so; text read off an image is a person's reading, so the image is shown and
+    a note says what is in it.
+    """
+
+    def test_no_frame_carries_a_quotation(self):
+        from tools.impact import talks
+        for talk in talks.load()["talks"]:
+            for mention in talk["mentions"]:
+                if mention["source"] == "frame":
+                    self.assertIsNone(mention["excerpt"], mention["id"])
+                    self.assertTrue(mention.get("note"))
+
+    def test_every_frame_image_is_in_the_repository(self):
+        from tools.impact import config, talks
+        seen = 0
+        for talk in talks.load()["talks"]:
+            for mention in talk["mentions"]:
+                if mention.get("image"):
+                    seen += 1
+                    self.assertTrue(
+                        (config.REPO_ROOT / mention["image"].lstrip("/")).exists(),
+                        mention["image"])
+        self.assertGreater(seen, 0)
+
+    def test_a_frame_without_its_image_is_caught(self):
+        from tools.impact import validate
+        problems = validate.check_talks([{
+            "id": "talk:youtube:x",
+            "sources": [{"kind": "frame", "status": "extracted"}],
+            "mentions": [{"id": "M1", "source": "frame", "at_seconds": 10,
+                          "url": "https://www.youtube.com/watch?v=x&t=10s",
+                          "matched": ["sqlancer"], "note": "a slide",
+                          "image": "/assets/images/impact/talks/nope-10.jpg"}],
+        }])
+        self.assertTrue(any("not in the repository" in str(p) for p in problems))
+
+    def test_mentions_run_in_the_order_the_talk_does(self):
+        from tools.impact import talks
+        for talk in talks.load()["talks"]:
+            timed = [m["at_seconds"] for m in talk["mentions"]
+                     if m.get("at_seconds") is not None]
+            self.assertEqual(sorted(timed), timed, talk["id"])
+
+    def test_a_viewers_account_yields_to_a_better_source(self):
+        """Once a caption or a frame reaches a talk, the account is not evidence."""
+        from tools.impact import talks
+        for talk in talks.load()["talks"]:
+            kinds = {m["source"] for m in talk["mentions"]}
+            if kinds & {"captions", "frame", "slides"}:
+                self.assertNotIn("watched", kinds, talk["id"])
+
+
+class SuppliedTranscriptTest(unittest.TestCase):
+    """A transcript somebody got out by hand has to be accepted as it comes.
+
+    YouTube will not serve some talks' captions to anything but its own player,
+    so the way in is a person: the panel's copy button, a downloaded .vtt,
+    yt-dlp. Which of those it was is not something they should have to care
+    about, and the rolling repetition auto-captions carry is not something they
+    should have to clean up.
+    """
+
+    VTT = ("WEBVTT\n\n"
+           "00:04:55.000 --> 00:04:58.120 align:start position:0%\n"
+           "as another example <00:04:56.000><c> SQL</c><c> lenser</c>\n\n"
+           "00:04:58.120 --> 00:05:01.000\n"
+           "as another example SQL lenser\n"
+           "which is a fuzzer for database systems\n")
+
+    def test_a_vtt_is_read_and_its_repetition_dropped(self):
+        from tools.impact import talks
+        segments = talks.parse_transcript(self.VTT)
+        self.assertEqual(["as another example SQL lenser",
+                          "which is a fuzzer for database systems"],
+                         [s["text"] for s in segments])
+        self.assertEqual([295000, 298000], [s["start_ms"] for s in segments])
+
+    def test_an_srt_is_read_too(self):
+        from tools.impact import talks
+        srt = ("1\n00:04:55,000 --> 00:04:58,120\n"
+               "as another example SQL lenser\n")
+        segments = talks.parse_transcript(srt)
+        self.assertEqual(1, len(segments))
+        self.assertEqual(295000, segments[0]["start_ms"])
+
+    def test_a_transcript_copied_out_of_the_panel_is_read(self):
+        from tools.impact import talks
+        for pasted in ("3:25\nSo the first one is pivoted query synthesis.\n",
+                       "3:25 So the first one is pivoted query synthesis.\n"):
+            segments = talks.parse_transcript(pasted)
+            self.assertEqual([205000], [s["start_ms"] for s in segments])
+
+    def test_an_hour_long_talk_keeps_its_hours(self):
+        from tools.impact import talks
+        segments = talks.parse_transcript("1:00:36\nsorry about that\n")
+        self.assertEqual(3636000, segments[0]["start_ms"])
+
+    def test_only_the_windows_around_a_mention_are_kept(self):
+        """A whole talk is not stored; the sentences that mention it are."""
+        from tools.impact import talks
+        segments = [{"start_ms": i * 5000, "text": f"line {i}"}
+                    for i in range(40)]
+        segments[20]["text"] = "and this is where SQLancer comes up"
+        mentions = talks.findings_from_segments(
+            segments, url="https://www.youtube.com/watch?v=x")
+        kept = talks.windows_around(segments, mentions)
+        self.assertEqual(5, len(kept))
+        self.assertEqual(40 - 5, len(segments) - len(kept))
+        self.assertIn("SQLancer", kept[2]["text"])
+
+
+class MisspelledAuthorTest(unittest.TestCase):
+    """Transcribers get the first name wrong as often as they get it right.
+
+    "Manual Rigger" is what SQLite's creator's talk came back as, and matching
+    on the full name missed the only mention in it. The surname carries enough
+    on its own here -- "trigger" is the one word that would collide, and a word
+    boundary keeps it out.
+    """
+
+    def test_the_surname_alone_is_a_mention(self):
+        from tools.impact import talks
+        for spelling in ("Manual Rigger came up with this idea",
+                         "Manuel Rigger came up with this idea",
+                         "Rigger's fuzzers find wrong answers here"):
+            found = talks.findings_from_segments(
+                [{"start_ms": 0, "text": spelling}],
+                url="https://www.youtube.com/watch?v=x")
+            self.assertEqual(["author"], found[0]["matched"], spelling)
+
+    def test_trigger_is_not_the_author(self):
+        from tools.impact import talks
+        self.assertEqual([], talks.findings_from_segments(
+            [{"start_ms": 0, "text": "inputs that trigger many known errors"}],
+            url="https://www.youtube.com/watch?v=x"))
+
+    def test_a_short_mention_carries_the_thought_it_introduces(self):
+        """A name in one breath, what it is for in the next."""
+        from tools.impact import talks
+        found = talks.findings_from_segments(
+            [{"start_ms": 0, "text": "Then Rigger came up with this idea."},
+             {"start_ms": 5000,
+              "text": "He tested for inconsistencies in SQL instead."}],
+            url="https://www.youtube.com/watch?v=x")
+        self.assertIn("inconsistencies in SQL", found[0]["excerpt"])
+
+    def test_a_transcript_that_says_nothing_says_so(self):
+        """Miryung Kim's keynote never speaks the name; the slide carries it."""
+        from tools.impact import talks
+        talk = next(t for t in talks.load()["talks"]
+                    if t["video_id"] == "L90MBb6NLBE")
+        captions = next(s for s in talk["sources"] if s["kind"] == "captions")
+        self.assertEqual("extracted", captions["status"])
+        self.assertIn("not spoken anywhere", captions["note"])
+        self.assertEqual(["frame"], [m["source"] for m in talk["mentions"]])
+
+
+class FrameWorklistTest(unittest.TestCase):
+    """Which moments want a picture is a rule, not a judgement call.
+
+    Choosing them by eye missed two: both were moments where the transcript had
+    misheard the name, which is exactly when the slide is worth having. The
+    worklist asks the data instead.
+    """
+
+    def test_a_misheard_name_wants_a_frame(self):
+        from tools.impact import talks
+        wanted = talks.frames_wanted({
+            "url": "https://www.youtube.com/watch?v=x",
+            "sources": [{"kind": "captions", "status": "extracted"}],
+            "mentions": [{"id": "M1", "source": "captions", "at_seconds": 100,
+                          "timestamp": "1:40", "heard_as": "SQL lenser",
+                          "excerpt": "one example is SQL lenser",
+                          "url": "https://www.youtube.com/watch?v=x&t=100s"}],
+        })
+        self.assertEqual(1, len(wanted))
+        self.assertIn("misheard", wanted[0]["why"])
+
+    def test_a_moment_already_captured_is_not_asked_for(self):
+        from tools.impact import talks
+        wanted = talks.frames_wanted({
+            "url": "https://www.youtube.com/watch?v=x",
+            "sources": [{"kind": "captions", "status": "extracted"}],
+            "mentions": [
+                {"id": "M1", "source": "captions", "at_seconds": 100,
+                 "timestamp": "1:40", "heard_as": "SQL lenser",
+                 "excerpt": "…", "url": "https://www.youtube.com/watch?v=x&t=100s"},
+                {"id": "M2", "source": "frame", "at_seconds": 104,
+                 "timestamp": "1:44", "image": "/assets/x.jpg",
+                 "url": "https://www.youtube.com/watch?v=x&t=104s"}],
+        })
+        self.assertEqual([], wanted)
+
+    def test_a_moment_somebody_checked_is_not_asked_for_again(self):
+        """The closing contact slide was up; there was nothing to capture."""
+        from tools.impact import talks
+        wanted = talks.frames_wanted({
+            "url": "https://www.youtube.com/watch?v=x",
+            "sources": [{"kind": "captions", "status": "extracted"}],
+            "mentions": [{"id": "M1", "source": "captions", "at_seconds": 100,
+                          "timestamp": "1:40", "heard_as": "SQL answer",
+                          "excerpt": "…", "frame_checked": "nothing on screen",
+                          "url": "https://www.youtube.com/watch?v=x&t=100s"}],
+        })
+        self.assertEqual([], wanted)
+
+    def test_a_silent_transcript_wants_a_moment_from_a_person(self):
+        from tools.impact import talks
+        wanted = talks.frames_wanted({
+            "url": "https://www.youtube.com/watch?v=x",
+            "sources": [{"kind": "captions", "status": "extracted",
+                         "note": "The whole transcript was read and SQLancer "
+                                 "is not spoken anywhere in it."}],
+            "mentions": [],
+        })
+        self.assertEqual(1, len(wanted))
+        self.assertIsNone(wanted[0]["at_seconds"])
+
+    def test_the_dataset_has_no_outstanding_frames(self):
+        from tools.impact import talks
+        outstanding = {t["title"]: talks.frames_wanted(t)
+                       for t in talks.load()["talks"]
+                       if talks.frames_wanted(t)}
+        self.assertEqual({}, outstanding)
+
+    def test_a_talk_either_shows_something_or_says_why_not(self):
+        """Words alone are allowed, but only once somebody has looked.
+
+        Most talks put SQLancer on a slide. A few mention it in the questions,
+        with whatever slide happened to be left up, and there is nothing to
+        capture -- which is a finding, recorded on the mention, not an absence.
+        """
+        from tools.impact import talks
+        for talk in talks.load()["talks"]:
+            shows = any(m.get("image") for m in talk["mentions"])
+            looked = any(m.get("frame_checked") for m in talk["mentions"])
+            self.assertTrue(shows or looked, talk["id"])
+
+    def test_a_deck_slide_keeps_its_words_as_well_as_its_picture(self):
+        """A frame may not be quoted; a published slide may -- it is text."""
+        from tools.impact import talks
+        for talk in talks.load()["talks"]:
+            for mention in talk["mentions"]:
+                if mention["source"] == "slides" and mention.get("image"):
+                    self.assertTrue(mention["excerpt"])
+                    self.assertTrue(mention["excerpt_is_verbatim"])
+
+
+class GraphSystemTest(unittest.TestCase):
+    """A lab member's name does not vouch for a graph database system.
+
+    The reporter rule infers that a roster member's database bug came from a
+    SQLancer campaign. For graph systems that inference is simply false: the
+    lab tests them with tools developed independently of SQLancer, and SQLancer
+    has no provider for any of them. 38 records reached the dataset this way
+    before the rule learned the difference.
+    """
+
+    def test_no_graph_bug_rests_on_the_reporter(self):
+        from tools.impact import config
+        from tools.impact.collectors.github_bugs import OUTSIDE_SQLANCER_CAMPAIGNS
+        bugs = config.load_json(config.DATA_FILES["bugs"])["bugs"]
+        offenders = [b["id"] for b in bugs
+                     if b.get("dbms") in OUTSIDE_SQLANCER_CAMPAIGNS
+                     and (b.get("attribution") or {}).get("rule")
+                     == "campaign_reporter"]
+        self.assertEqual([], offenders)
+
+    def test_the_rule_is_off_for_those_systems(self):
+        """The gate is at the call site, so assert on what it lets through."""
+        from tools.impact.collectors.github_bugs import (
+            OUTSIDE_SQLANCER_CAMPAIGNS, deterministic_attribution)
+        from tools.impact import taxonomy
+        issue = {"title": "Wrong result for a MATCH query",
+                 "body": "MATCH (n) RETURN n returns the wrong rows.",
+                 "labels": [{"name": "bug"}]}
+        text = f"{issue['title']}\n{issue['body']}"
+        tax = taxonomy.load()
+        self.assertIn("neo4j", OUTSIDE_SQLANCER_CAMPAIGNS)
+        # With the reporter's name allowed to vouch, this is admitted...
+        self.assertIsNotNone(deterministic_attribution(
+            text, tax, issue, reporter_runs_campaigns=True))
+        # ...and without it, nothing here says SQLancer at all.
+        self.assertIsNone(deterministic_attribution(
+            text, tax, issue, reporter_runs_campaigns=False))
+
+    def test_the_policy_says_so_on_the_page(self):
+        from tools.impact import config
+        policy = config.load_json(config.DATA_FILES["policy"])
+        rules = [r for section in policy["sections"]
+                 for r in section.get("rules", [])]
+        graph = next(r for r in rules if r["id"] == "bug-graph-systems")
+        self.assertEqual("exclude", graph["kind"])
+
+
+class UnreadableQuoteTest(unittest.TestCase):
+    """A quotation nobody can read is worse than no quotation.
+
+    Some PDFs set a passage with letter-spacing, and extraction returns it one
+    character at a time: "w i t h i n S Q L a n c e r". The gaps between words
+    are the same width as the gaps inside them, in the file as well as on the
+    page, so the word breaks are not there to put back -- rejoining gives
+    "systemwithinSQLancer" and guessing them from a dictionary would be writing
+    text rather than quoting it. Six mentions were affected, and every one had
+    other evidence for the same claim.
+    """
+
+    def test_nothing_published_quotes_an_unreadable_passage(self):
+        from tools.impact import config, validate
+        papers = config.load_json(config.DATA_FILES["papers"])["papers"]
+        self.assertEqual([], [str(p) for p in
+                              validate.check_readable_excerpts(papers)])
+
+    def test_the_check_catches_one(self):
+        from tools.impact import validate
+        problems = validate.check_readable_excerpts([{
+            "id": "paper:doi:x",
+            "relationships": {"uses_infrastructure": {"value": "yes", "evidence": [
+                {"excerpt": "We implemented this s y s t e m w i t h i n "
+                            "S Q L a n c e r and tested it."}]}},
+        }])
+        self.assertTrue(any("word breaks" in str(p) for p in problems))
+
+    def test_the_mention_is_kept_and_marked(self):
+        """The mention is real; only its text is unusable."""
+        import json
+        import pathlib
+        marked = 0
+        for path in pathlib.Path("_data/papers").glob("*.json"):
+            record = json.loads(path.read_text(encoding="utf-8"))
+            for mention in record.get("mentions", []):
+                if mention.get("text_is_letter_spaced"):
+                    marked += 1
+                    self.assertTrue(mention.get("sentence"))
+        self.assertEqual(6, marked)
+
+
+class MovedRepositoryTest(unittest.TestCase):
+    """A project that changed organisation must not get two records per bug.
+
+    DuckDB was cwida/duckdb before it was duckdb/duckdb, and GitHub redirects
+    the old links, so both forms reach the same issue and both look canonical.
+    The curated list kept the address a bug was filed under while an issue
+    search found the address it lives at now: 77 DuckDB bugs were counted
+    twice.
+    """
+
+    def test_a_former_organisation_resolves_to_the_current_one(self):
+        from tools.impact.repos import canonical_issue_url
+        self.assertEqual("https://github.com/duckdb/duckdb/issues/490",
+                         canonical_issue_url("https://github.com/cwida/duckdb/issues/490"))
+
+    def test_an_unrelated_url_is_left_alone(self):
+        from tools.impact.repos import canonical_issue_url
+        for url in ("https://github.com/cockroachdb/cockroach/issues/1",
+                    "https://github.com/duckdb/duckdb",
+                    "https://bugs.mysql.com/bug.php?id=1"):
+            self.assertEqual(url, canonical_issue_url(url))
+
+    def test_no_bug_is_recorded_at_two_addresses(self):
+        from tools.impact import config
+        from tools.impact.repos import canonical_issue_url
+        seen = {}
+        for bug in config.load_json(config.DATA_FILES["bugs"])["bugs"]:
+            url = canonical_issue_url(bug.get("primary_url") or "")
+            if not url:
+                continue
+            key = (bug["dbms"], url)
+            self.assertNotIn(key, seen,
+                             f"{bug['id']} and {seen.get(key)} are the same report")
+            seen[key] = bug["id"]
+
+
+class RepeatedTitleTest(unittest.TestCase):
+    """Reports sharing a title are not thereby the same bug.
+
+    A fuzzer files what it finds under whatever title its harness writes:
+    CockroachDB's nightly roachtest files "roachtest: tlp failed" every time,
+    and StarRocks' campaigns file "[sqlancer] query of result set mismatch".
+    Collapsing those would undercount -- the CockroachDB set spans 21 distinct
+    days over four years, and the same-day sets are consecutive issue numbers
+    from one sitting, each accepted and fixed separately.
+
+    What does settle it is the project's own triage, and that is already
+    honoured: a report the maintainers closed as a duplicate never counts.
+    """
+
+    def test_a_report_the_project_called_a_duplicate_is_not_counted(self):
+        from tools.impact import config
+        bugs = config.load_json(config.DATA_FILES["bugs"])["bugs"]
+        declared = [b for b in bugs if b["status"] == "closed_duplicate"]
+        self.assertTrue(declared, "expected some declared duplicates on file")
+        self.assertEqual([], [b["id"] for b in declared
+                              if b.get("status_is_true_positive")])
+
+    def test_repeated_titles_each_carry_their_own_address(self):
+        """Each is its own issue, which is what makes it its own record."""
+        import collections
+        from tools.impact import config
+        bugs = config.load_json(config.DATA_FILES["bugs"])["bugs"]
+        addressed = collections.defaultdict(list)
+        for bug in bugs:
+            title = (bug.get("title") or "").strip().lower()
+            if title and bug.get("primary_url"):
+                addressed[(bug["dbms"], title)].append(bug["primary_url"])
+        for (dbms, title), urls in addressed.items():
+            self.assertEqual(len(urls), len(set(urls)),
+                             f"{dbms} '{title[:40]}' repeats an address")
+
+    def test_a_report_without_an_address_is_not_a_second_copy(self):
+        """Umbra's bugs were emailed, so they have no URL -- but a record with
+        no address must not shadow a curated one for the same report."""
+        import collections
+        from tools.impact import config
+        bugs = config.load_json(config.DATA_FILES["bugs"])["bugs"]
+        clusters = collections.defaultdict(list)
+        for bug in bugs:
+            title = (bug.get("title") or "").strip().lower()
+            if title and bug.get("reported_date"):
+                clusters[(bug["dbms"], title, bug["reported_date"])].append(bug)
+        for key, rows in clusters.items():
+            if len(rows) < 2:
+                continue
+            with_url = [b for b in rows if b.get("primary_url")]
+            self.assertFalse(
+                with_url and len(with_url) != len(rows),
+                f"{key[0]} '{key[1][:38]}' on {key[2]} has a copy with no address")
+
+
+class PaperRecordSchemaTest(unittest.TestCase):
+    """The analysed papers are checked like everything else now.
+
+    These 190 files hold the sentences behind every claim the impact page makes
+    about research, and they were the one part of the dataset with no schema at
+    all -- nothing would have noticed a renamed field or a lost quotation.
+    """
+
+    def test_every_record_matches_the_schema(self):
+        from tools.impact import validate
+        self.assertEqual([], [str(p) for p in validate.check_paper_records()])
+
+    def test_the_schema_covers_what_is_on_disk(self):
+        import json
+        import pathlib
+        from tools.impact import config
+        records = sorted(pathlib.Path("_data/papers").glob("*.json"))
+        self.assertGreater(len(records), 100)
+        schema = config.load_json(
+            config.REPO_ROOT / "schemas" / "papers" / "paper-analysis.schema.json")
+        # The sentinel that stands where a repository, not a sentence, carries
+        # the claim must be allowed -- one paper depends on it.
+        pattern = (schema["properties"]["analysis"]["properties"]["relationships"]
+                   ["additionalProperties"]["properties"]["mention_ids"]
+                   ["items"]["pattern"])
+        self.assertIn("ARTIFACT", pattern)
+        with records[0].open() as handle:
+            self.assertIn("mentions", json.load(handle))
+
+
+class NonEnglishContextTest(unittest.TestCase):
+    """An ambiguous acronym has to be corroborated in the report's own language.
+
+    "TLP" means thread-level parallelism and a laptop power tool, so the
+    taxonomy only accepts it beside a database-testing term. Those terms were
+    all English, and openGauss reports in Chinese on Gitee -- so a title reading
+    "TLP 等价验证" (TLP equivalence verification) corroborated nothing and was
+    thrown away.
+    """
+
+    def test_a_chinese_report_can_corroborate_an_acronym(self):
+        from tools.impact import taxonomy
+        found = taxonomy.load().find_techniques(
+            "执行包含 SUM(REAL) 聚合与恒真 HAVING 条件的查询时，"
+            "TLP 等价验证出现结果内容不一致")
+        self.assertEqual(["tlp"], [m.technique_id for m in found])
+
+    def test_the_power_management_sense_is_still_refused(self):
+        from tools.impact import taxonomy
+        self.assertEqual([], taxonomy.load().find_techniques(
+            "tlp-stat shows the laptop battery is in power management mode"))
+
+    def test_a_chinese_certificate_thread_is_not_the_cert_oracle(self):
+        from tools.impact import taxonomy
+        self.assertEqual([], taxonomy.load().find_techniques(
+            "配置数据库连接时 CERT 证书校验失败"))
diff --git a/tools/impact/util.py b/tools/impact/util.py
new file mode 100644
index 0000000..dd10a83
--- /dev/null
+++ b/tools/impact/util.py
@@ -0,0 +1,195 @@
+"""Small shared helpers: hashing, stable ids, timestamps, URL normalisation."""
+
+from __future__ import annotations
+
+import hashlib
+import re
+import unicodedata
+from datetime import datetime, timezone
+from typing import Iterable, Optional
+from urllib.parse import urlsplit, urlunsplit
+
+
+def now() -> str:
+    """Current UTC time in the timestamp format the schemas require."""
+    return datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ")
+
+
+def sha256_hex(text: str) -> str:
+    return hashlib.sha256(text.encode("utf-8")).hexdigest()
+
+
+def content_hash(text: str) -> str:
+    """Hash in the ``sha256:<hex>`` form used throughout the schemas."""
+    return "sha256:" + sha256_hex(text)
+
+
+def stable_digest(*parts: object, length: int = 12) -> str:
+    """Short digest over a tuple of values, used to build stable record ids."""
+    joined = " ".join("" if p is None else str(p) for p in parts)
+    return sha256_hex(joined)[:length]
+
+
+def slugify(value: str) -> str:
+    """Lowercase identifier slug matching the ``slug`` schema definition."""
+    normalised = unicodedata.normalize("NFKD", value)
+    ascii_only = normalised.encode("ascii", "ignore").decode("ascii")
+    slug = re.sub(r"[^a-z0-9]+", "_", ascii_only.lower()).strip("_")
+    slug = re.sub(r"_+", "_", slug)
+    if not slug:
+        slug = "unknown"
+    if not slug[0].isalnum():
+        slug = "x" + slug
+    return slug[:64]
+
+
+_TRACKING_PARAMS = re.compile(r"^(utm_|ref$|ref_|fbclid$|gclid$)")
+
+
+def normalize_url(url: Optional[str]) -> Optional[str]:
+    """Canonicalise a URL so the same source deduplicates to one record.
+
+    Lowercases the host, drops a default port, strips tracking parameters and
+    fragments, and removes a trailing slash. Deliberately conservative: it never
+    touches the path's case, because many paths are case sensitive.
+    """
+    if not url:
+        return None
+    url = url.strip()
+    if not url:
+        return None
+    if url.startswith("//"):
+        url = "https:" + url
+    parts = urlsplit(url)
+    if parts.scheme not in ("http", "https"):
+        return None
+    host = parts.netloc.lower()
+    if parts.scheme == "http" and host.endswith(":80"):
+        host = host[:-3]
+    if parts.scheme == "https" and host.endswith(":443"):
+        host = host[:-4]
+    if host.startswith("www."):
+        host = host[4:]
+    query = "&".join(
+        piece for piece in parts.query.split("&")
+        if piece and not _TRACKING_PARAMS.match(piece)
+    )
+    path = parts.path
+    if len(path) > 1 and path.endswith("/"):
+        path = path.rstrip("/")
+    return urlunsplit(("https", host, path, query, ""))
+
+
+_GITHUB_ISSUE = re.compile(
+    r"^https?://(?:www\.)?github\.com/([\w.-]+)/([\w.-]+)/(issues|pull)/(\d+)",
+    re.IGNORECASE,
+)
+
+
+def github_issue_ref(url: Optional[str]):
+    """Return ``(owner, repo, kind, number)`` for a GitHub issue or PR URL."""
+    if not url:
+        return None
+    match = _GITHUB_ISSUE.match(url)
+    if not match:
+        return None
+    owner, repo, kind, number = match.groups()
+    return owner, repo, "pull_request" if kind == "pull" else "issue", int(number)
+
+
+def normalize_doi(doi: Optional[str]) -> Optional[str]:
+    """Strip resolver prefixes and lowercase a DOI for use as a stable key."""
+    if not doi:
+        return None
+    doi = doi.strip()
+    for prefix in ("https://doi.org/", "http://doi.org/",
+                   "https://dx.doi.org/", "http://dx.doi.org/", "doi:"):
+        if doi.lower().startswith(prefix):
+            doi = doi[len(prefix):]
+            break
+    doi = doi.strip().lower()
+    return doi if doi.startswith("10.") else None
+
+
+_DATE_FORMATS = ("%d/%m/%Y", "%Y-%m-%d", "%d.%m.%Y", "%Y/%m/%d", "%d/%m/%y")
+
+
+# A slash date whose order has to be worked out from the numbers themselves.
+_SLASH_DATE = re.compile(r"^(\d{1,2})/(\d{1,2})/(\d{4})$")
+
+
+def parse_date(value: Optional[str]) -> Optional[str]:
+    """Parse the date spellings seen in upstream sources into ISO YYYY-MM-DD.
+
+    Slash dates carry no indication of their order, and sources are not even
+    internally consistent: the TEST lab's bug list has 735 rows that can only
+    be day-first and 147 that can only be month-first. Where one component
+    exceeds twelve the order is settled by that, which is what recovers the
+    month-first rows -- they used to fail every format and come back as no date
+    at all. Where both components could be a month nothing in the value can
+    settle it, and the day-first reading is kept, as before.
+    """
+    if not value:
+        return None
+    value = value.strip()
+    if not value or value.lower() in ("unknown", "n/a", "none"):
+        return None
+    if "T" in value:
+        value = value.split("T", 1)[0]
+    slash = _SLASH_DATE.match(value)
+    if slash and int(slash.group(1)) <= 12 < int(slash.group(2)):
+        month, day, year = slash.groups()
+        try:
+            return datetime(int(year), int(month), int(day)).strftime("%Y-%m-%d")
+        except ValueError:
+            return None
+    for fmt in _DATE_FORMATS:
+        try:
+            return datetime.strptime(value, fmt).strftime("%Y-%m-%d")
+        except ValueError:
+            continue
+    return None
+
+
+def year_of(iso_date: Optional[str]) -> Optional[int]:
+    if not iso_date:
+        return None
+    try:
+        return int(iso_date[:4])
+    except (TypeError, ValueError):
+        return None
+
+
+def _collapse(text: str) -> str:
+    return re.sub(r"\s+", " ", text).strip()
+
+
+def verbatim_excerpt(source_text: str, excerpt: Optional[str]) -> bool:
+    """Check that an excerpt really is a literal substring of the source.
+
+    The no-fabrication rule is enforced here rather than trusted: anything that
+    is not found verbatim in the fetched text is rejected before it is stored.
+    Only whitespace runs are normalised, since extraction reflows line breaks.
+    """
+    if excerpt is None:
+        return True
+    if not source_text:
+        return False
+    return _collapse(excerpt) in _collapse(source_text)
+
+
+def truncate(text: str, limit: int) -> str:
+    text = text.strip()
+    if len(text) <= limit:
+        return text
+    return text[: limit - 3].rstrip() + "..."
+
+
+def dedupe_preserving_order(items: Iterable[str]):
+    seen = set()
+    out = []
+    for item in items:
+        if item not in seen:
+            seen.add(item)
+            out.append(item)
+    return out
diff --git a/tools/impact/validate.py b/tools/impact/validate.py
new file mode 100644
index 0000000..c5cc55b
--- /dev/null
+++ b/tools/impact/validate.py
@@ -0,0 +1,634 @@
+"""Schema and cross-file validation for the impact dataset.
+
+JSON Schema catches shape errors. The checks after it catch the things a schema
+cannot express: an id used twice, a bug pointing at a technique that is not in
+the taxonomy, an accepted classification with no evidence behind it, an evidence
+excerpt whose verbatim flag was not set. Both halves run in CI, and a failure
+here blocks the pull request rather than publishing a bad record.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+from collections import Counter, defaultdict
+from pathlib import Path
+from typing import Dict, Iterable, List, Optional, Tuple
+
+import jsonschema
+
+from . import config, taxonomy
+from .util import normalize_doi, normalize_url
+
+URL_RE = re.compile(r"^https?://[^\s<>\"]+$")
+
+
+class Problem:
+    __slots__ = ("file", "path", "message")
+
+    def __init__(self, file: str, path: str, message: str):
+        self.file = file
+        self.path = path
+        self.message = message
+
+    def __str__(self):
+        location = f"{self.file}:{self.path}" if self.path else self.file
+        return f"{location}: {self.message}"
+
+
+def _resolver(schema: dict) -> jsonschema.RefResolver:
+    """Resolve the relative ``common.schema.json#...`` references."""
+    store = {}
+    for path in config.SCHEMA_DIR.glob("*.schema.json"):
+        with open(path, "r", encoding="utf-8") as handle:
+            loaded = json.load(handle)
+        store[path.name] = loaded
+        if "$id" in loaded:
+            store[loaded["$id"]] = loaded
+    base = config.SCHEMA_DIR.as_uri() + "/"
+    return jsonschema.RefResolver(base_uri=base, referrer=schema, store=store)
+
+
+def validate_schema(name: str, payload: dict) -> List[Problem]:
+    schema_path = config.SCHEMA_FILES.get(name)
+    if schema_path is None or not schema_path.exists():
+        return [Problem(name, "", f"no schema file at {schema_path}")]
+    schema = config.load_json(schema_path)
+    validator = jsonschema.Draft7Validator(schema, resolver=_resolver(schema))
+    problems = []
+    for error in sorted(validator.iter_errors(payload), key=lambda e: list(e.path)):
+        path = "/".join(str(part) for part in error.absolute_path)
+        problems.append(Problem(f"{name}.json", path, error.message))
+    return problems
+
+
+def _iter_evidence(record: dict) -> Iterable[Tuple[str, dict]]:
+    """Yield ``(path, evidence)`` for every evidence object inside a record."""
+    def walk(node, path):
+        if isinstance(node, dict):
+            if "source_url" in node and "source_type" in node:
+                yield path, node
+            for key, value in node.items():
+                yield from walk(value, f"{path}/{key}" if path else key)
+        elif isinstance(node, list):
+            for index, value in enumerate(node):
+                yield from walk(value, f"{path}/{index}")
+    yield from walk(record, "")
+
+
+def check_evidence(name: str, records: List[dict]) -> List[Problem]:
+    problems = []
+    for record in records:
+        rid = record.get("id", "?")
+        for path, evidence in _iter_evidence(record):
+            url = evidence.get("source_url")
+            if not url or not URL_RE.match(url):
+                problems.append(Problem(f"{name}.json", f"{rid}/{path}",
+                                        f"malformed evidence URL {url!r}"))
+            excerpt = evidence.get("excerpt")
+            if excerpt is not None and not evidence.get("excerpt_is_verbatim"):
+                problems.append(Problem(
+                    f"{name}.json", f"{rid}/{path}",
+                    "evidence carries an excerpt but excerpt_is_verbatim is not set; "
+                    "excerpts must be copied verbatim from the source"))
+            if excerpt is not None and not excerpt.strip():
+                problems.append(Problem(f"{name}.json", f"{rid}/{path}",
+                                        "evidence excerpt is empty"))
+    return problems
+
+
+def check_unique_ids(name: str, records: List[dict]) -> List[Problem]:
+    counts = Counter(record.get("id") for record in records)
+    return [Problem(f"{name}.json", str(rid), f"duplicate record id ({count} times)")
+            for rid, count in counts.items() if count > 1]
+
+
+def check_affiliation_rulings(records: List[dict],
+                              roster: List[dict]) -> List[Problem]:
+    """The rulings in ``people_decisions.json`` are in force and still apply.
+
+    Two ways this file goes quietly wrong, and both make it a no-op rather than
+    an error: the records can be stale, so a ruling was made but never applied;
+    and a handle can be misspelled, so the ruling matches nobody. Neither shows
+    up anywhere -- the numbers simply stay as they were.
+    """
+    from .collectors import people as people_collector
+
+    problems: List[Problem] = []
+    ruling = people_collector.decisions()
+    entries = ruling.get("counts_as_external") or []
+    if not entries:
+        return problems
+
+    named = people_collector.counted_as_external(ruling)
+    known = {(entry.get("github") or "").strip().lower() for entry in roster}
+    known |= {(entry.get("name") or "").strip().lower() for entry in roster}
+    known.discard("")
+    for entry in entries:
+        handle = (entry.get("github") or "").strip().lower()
+        if handle and handle not in known:
+            problems.append(Problem(
+                "people_decisions.json", handle,
+                "no one on the roster goes by this name, so the ruling has no "
+                "effect; check the spelling"))
+
+    for record in records:
+        reporter = (record.get("reporter") or "").strip().lower()
+        if reporter in named and record.get("reporter_affiliation") != "external":
+            problems.append(Problem(
+                "bugs.json", record.get("id", "?"),
+                f"{record.get('reporter')} is ruled external in "
+                f"people_decisions.json but this record says "
+                f"{record.get('reporter_affiliation')!r}; re-run collect"))
+    return problems
+
+
+def check_paper_records() -> List[Problem]:
+    """Validate every analysed paper against its schema.
+
+    These 190 files carry the sentences behind every claim the impact page
+    makes about research, and they were the one part of the dataset nothing
+    checked -- there was no schema for them at all.
+    """
+    import json as _json
+
+    schema_path = config.REPO_ROOT / "schemas" / "papers" / "paper-analysis.schema.json"
+    directory = config.REPO_ROOT / "_data" / "papers"
+    if not schema_path.exists() or not directory.exists():
+        return []
+    schema = config.load_json(schema_path)
+    validator = jsonschema.Draft7Validator(schema)
+    problems: List[Problem] = []
+    for path in sorted(directory.glob("*.json")):
+        try:
+            record = _json.loads(path.read_text(encoding="utf-8"))
+        except ValueError as error:
+            problems.append(Problem("papers/", path.name, f"invalid JSON: {error}"))
+            continue
+        for error in sorted(validator.iter_errors(record),
+                            key=lambda e: list(e.path))[:3]:
+            where = "/".join(str(part) for part in error.path)
+            problems.append(Problem("papers/", f"{path.name}:{where}",
+                                    error.message[:200]))
+    return problems
+
+
+def check_readable_excerpts(papers: List[dict]) -> List[Problem]:
+    """No published quotation may be one nobody can read.
+
+    Some PDFs set a passage with letter-spacing, and extraction returns it one
+    character at a time. The mention is real, but the text is not a quotation
+    any reader can use, and the word breaks are not in the file to put back --
+    so it must not reach a page. Every case so far had other evidence for the
+    same claim, which is what made dropping it safe.
+    """
+    from tools.papers.extract import is_letter_spaced
+
+    problems: List[Problem] = []
+    for paper in papers:
+        for key, relationship in (paper.get("relationships") or {}).items():
+            for item in (relationship or {}).get("evidence") or []:
+                if is_letter_spaced(item.get("excerpt") or ""):
+                    problems.append(Problem(
+                        "papers.json", f"{paper.get('id')}/{key}",
+                        "quotes a passage whose word breaks extraction lost"))
+    return problems
+
+
+def check_talks(talks: List[dict]) -> List[Problem]:
+    """A talk record says where each mention came from, and stays checkable.
+
+    Two things can go quietly wrong here that a schema will not catch. A
+    relationship can cite a mention that has since been renumbered, leaving a
+    claim with nothing behind it. And a caption mention can lose its deep link
+    -- which matters more here than anywhere else in the dataset, because the
+    caption text is a machine's transcription and the link to the moment is the
+    only way a reader can check what was really said.
+    """
+    problems: List[Problem] = []
+    for talk in talks:
+        tid = talk.get("id") or "?"
+        ids = {m.get("id") for m in talk.get("mentions") or []}
+        if len(ids) != len(talk.get("mentions") or []):
+            problems.append(Problem("talks.json", tid,
+                                    "two mentions share an id"))
+        kinds = {s.get("kind") for s in talk.get("sources") or []}
+        for mention in talk.get("mentions") or []:
+            where = f"{tid}/{mention.get('id')}"
+            source = mention.get("source")
+            if source not in kinds:
+                problems.append(Problem(
+                    "talks.json", where,
+                    f"read from {source!r}, which is not one of this talk's "
+                    "sources"))
+            if source == "captions":
+                if mention.get("at_seconds") is None:
+                    problems.append(Problem("talks.json", where,
+                                            "a caption mention with no timestamp"))
+                elif f"t={mention['at_seconds']}s" not in (mention.get("url") or ""):
+                    problems.append(Problem(
+                        "talks.json", where,
+                        "the link does not point at the moment it was said"))
+            # Any mention may carry a picture -- a frame from a recording, or
+            # the slide itself from a published deck -- and a page that links a
+            # missing one is worse than one that shows nothing.
+            if mention.get("image") and not (
+                    config.REPO_ROOT / mention["image"].lstrip("/")).exists():
+                problems.append(Problem(
+                    "talks.json", where,
+                    f"the image {mention['image']} is not in the repository"))
+            if source == "frame":
+                if not mention.get("image"):
+                    problems.append(Problem("talks.json", where,
+                                            "a frame mention with no image"))
+                if mention.get("excerpt"):
+                    problems.append(Problem(
+                        "talks.json", where,
+                        "text read off a picture cannot be quoted as an excerpt"))
+                length = talk.get("duration_seconds")
+                if length and (mention.get("at_seconds") or 0) > length:
+                    problems.append(Problem(
+                        "talks.json", where,
+                        "a frame captured after the talk ends"))
+            if source == "watched" and mention.get("excerpt"):
+                problems.append(Problem(
+                    "talks.json", where,
+                    "a mention nobody read cannot carry a quotation"))
+            if mention.get("excerpt") and not mention.get("excerpt_is_verbatim"):
+                problems.append(Problem("talks.json", where,
+                                        "an excerpt not marked verbatim"))
+        cited = set((talk.get("relationship") or {}).get("mention_ids") or [])
+        missing = sorted(cited - ids)
+        if missing:
+            problems.append(Problem(
+                "talks.json", tid,
+                f"the relationship cites {', '.join(missing)}, which this "
+                "talk does not have"))
+    return problems
+
+
+def check_recognition_highlights(papers: List[dict]) -> List[Problem]:
+    """Every featured quote is one the paper's own record actually holds.
+
+    The highlights file names a paper and a mention id rather than copying the
+    sentence, so a quote cannot be edited into something the paper did not
+    say. What it can do is name a mention that no longer exists, or a paper
+    whose classification has since changed, and then the page would quote
+    nothing or quote it under a claim the data no longer makes.
+    """
+    import json as _json
+
+    path = config.DATA_DIR / "recognition_highlights.json"
+    if not path.exists():
+        return []
+    try:
+        highlights = config.load_json(path).get("highlights") or []
+    except ValueError as error:
+        return [Problem("recognition_highlights.json", "",
+                        f"invalid JSON: {error}")]
+
+    by_id = {paper.get("id"): paper for paper in papers}
+    problems: List[Problem] = []
+    for entry in highlights:
+        pid = entry.get("paper_id")
+        paper = by_id.get(pid)
+        if paper is None:
+            problems.append(Problem("recognition_highlights.json", pid or "?",
+                                    "no paper with this id"))
+            continue
+        relationship = (paper.get("relationships") or {}).get(
+            "describes_as_state_of_the_art") or {}
+        if relationship.get("value") != "yes":
+            problems.append(Problem(
+                "recognition_highlights.json", pid,
+                "this paper no longer describes SQLancer as state of the art"))
+            continue
+        # The quote lives on the paper record in the papers subproject; the
+        # impact record carries the same sentences as evidence.
+        record = config.REPO_ROOT / "_data" / "papers" / (
+            pid.replace(":", "_").replace(".", "_").replace("/", "_") + ".json")
+        if not record.exists():
+            problems.append(Problem("recognition_highlights.json", pid,
+                                    "no analysed record for this paper"))
+            continue
+        analysis = _json.loads(record.read_text(encoding="utf-8")).get(
+            "analysis") or {}
+        quotes = ((analysis.get("relationships") or {}).get(
+            "describes_as_state_of_the_art") or {}).get("quotes") or []
+        if not any(q.get("mention_id") == entry.get("mention_id")
+                   for q in quotes):
+            problems.append(Problem(
+                "recognition_highlights.json", f"{pid}/{entry.get('mention_id')}",
+                "the record has no quoted sentence with this mention id"))
+    return problems
+
+
+def check_bugs(records: List[dict], registry: Dict[str, dict],
+               tax: taxonomy.Taxonomy) -> List[Problem]:
+    problems: List[Problem] = []
+    by_url: Dict[str, List[str]] = defaultdict(list)
+    for record in records:
+        rid = record.get("id", "?")
+        if record.get("dbms") not in registry:
+            problems.append(Problem("bugs.json", rid,
+                                    f"unknown dbms {record.get('dbms')!r}"))
+        technique = record.get("technique")
+        if not tax.is_known_technique(technique):
+            problems.append(Problem("bugs.json", rid,
+                                    f"unknown SQLancer technique {technique!r}"))
+        if technique in tax.excluded:
+            problems.append(Problem(
+                "bugs.json", rid,
+                f"technique {technique!r} is excluded by the attribution policy"))
+        finder = record.get("finder")
+        if finder not in tax.finders:
+            problems.append(Problem("bugs.json", rid, f"unknown finder {finder!r}"))
+        elif not tax.finders[finder]["umbrella_member"]:
+            problems.append(Problem(
+                "bugs.json", rid,
+                f"finder {finder!r} is not a SQLancer umbrella member"))
+        if not record.get("links"):
+            problems.append(Problem("bugs.json", rid, "bug record has no links"))
+        if not record.get("attribution", {}).get("evidence"):
+            problems.append(Problem("bugs.json", rid, "bug record has no evidence"))
+        # A technique_attribution record must actually name the technique.
+        if record.get("attribution", {}).get("rule") == "technique_attribution" \
+                and technique is None:
+            problems.append(Problem(
+                "bugs.json", rid,
+                "attribution rule is technique_attribution but no technique is set"))
+        url = normalize_url(record.get("primary_url"))
+        if url:
+            by_url[url].append(rid)
+
+    for url, ids in by_url.items():
+        if len(ids) > 1:
+            # Two curated reports can share one upstream ticket; flag only when
+            # the titles match too, which would mean a genuine duplicate.
+            titles = {r["title"] for r in records if r["id"] in ids}
+            if len(titles) < len(ids):
+                problems.append(Problem(
+                    "bugs.json", ", ".join(ids),
+                    f"duplicate bug URL with identical title: {url}"))
+    return problems
+
+
+def check_papers(records: List[dict], tax: taxonomy.Taxonomy) -> List[Problem]:
+    problems: List[Problem] = []
+    seen_doi: Dict[str, str] = {}
+    for record in records:
+        rid = record.get("id", "?")
+        doi = normalize_doi(record.get("doi"))
+        if doi:
+            if doi in seen_doi:
+                problems.append(Problem("papers.json", rid,
+                                        f"duplicate DOI {doi} (also {seen_doi[doi]})"))
+            else:
+                seen_doi[doi] = rid
+        known_seeds = set(tax.seed_ids())
+        for seed in record.get("cites_seed_techniques", []):
+            if seed not in tax.techniques and seed not in known_seeds:
+                problems.append(Problem("papers.json", rid,
+                                        f"unknown seed technique {seed!r}"))
+        relationships = record.get("relationships", {})
+        if relationships.get("references", {}).get("value") != "yes":
+            problems.append(Problem(
+                "papers.json", rid,
+                "every collected paper must have references = yes"))
+        for key, relationship in relationships.items():
+            for technique in relationship.get("techniques", []):
+                # "references" carries which seed the paper cites, and a seed
+                # can be a tool rather than a technique. The three judged
+                # relationships name techniques only.
+                allowed = (tax.techniques if key != "references"
+                           else set(tax.techniques) | known_seeds)
+                if technique not in allowed:
+                    problems.append(Problem(
+                        "papers.json", f"{rid}/{key}",
+                        f"unknown SQLancer technique {technique!r}"))
+            if relationship.get("value") == "yes" and not relationship.get("evidence"):
+                problems.append(Problem(
+                    "papers.json", f"{rid}/{key}",
+                    "positive relationship with no evidence"))
+            # Extending is always extending something in particular, so an
+            # extension that names no technique is incomplete. Comparing is
+            # not: a paper can run SQLancer as a whole against its own tool
+            # without naming an oracle, and demanding one would mean inventing
+            # it. Twenty papers say exactly that.
+            if relationship.get("value") == "yes" \
+                    and key == "extends_technique" \
+                    and not relationship.get("techniques"):
+                problems.append(Problem(
+                    "papers.json", f"{rid}/{key}",
+                    "positive relationship does not name the technique involved"))
+    return problems
+
+
+def check_adoption(records: List[dict], registry: Dict[str, dict],
+                   tax: taxonomy.Taxonomy) -> List[Problem]:
+    problems: List[Problem] = []
+    seen: Dict[Tuple[str, str], str] = {}
+    for record in records:
+        rid = record.get("id", "?")
+        dbms = record.get("dbms")
+        if dbms not in registry:
+            problems.append(Problem("adoption.json", rid, f"unknown dbms {dbms!r}"))
+        key = (dbms, record.get("relationship"))
+        if key in seen:
+            problems.append(Problem(
+                "adoption.json", rid,
+                f"duplicate adoption relationship {key} (also {seen[key]})"))
+        else:
+            seen[key] = rid
+        finder = record.get("finder")
+        if finder is not None and finder not in tax.finders:
+            problems.append(Problem("adoption.json", rid,
+                                    f"unknown finder {finder!r}"))
+        if not record.get("evidence"):
+            problems.append(Problem("adoption.json", rid,
+                                    "adoption record has no evidence"))
+    return problems
+
+
+def check_resources(records: List[dict], registry: Dict[str, dict],
+                    tax: taxonomy.Taxonomy) -> List[Problem]:
+    problems: List[Problem] = []
+    seen: Dict[str, str] = {}
+    for record in records:
+        rid = record.get("id", "?")
+        url = normalize_url(record.get("url"))
+        if not url:
+            problems.append(Problem("resources.json", rid,
+                                    f"malformed URL {record.get('url')!r}"))
+        elif url in seen:
+            problems.append(Problem("resources.json", rid,
+                                    f"duplicate resource URL {url} (also {seen[url]})"))
+        else:
+            seen[url] = rid
+        for technique in record.get("related_techniques", []):
+            if technique not in tax.techniques:
+                problems.append(Problem("resources.json", rid,
+                                        f"unknown technique {technique!r}"))
+        for dbms in record.get("related_dbms", []):
+            if dbms not in registry:
+                problems.append(Problem("resources.json", rid,
+                                        f"unknown dbms {dbms!r}"))
+    return problems
+
+
+def validate_all(data_dir: Optional[Path] = None) -> List[Problem]:
+    """Validate every authoritative file plus the cross-file relationships."""
+    files = config.DATA_FILES
+    if data_dir is not None:
+        files = {name: Path(data_dir) / path.name
+                 for name, path in config.DATA_FILES.items()}
+
+    problems: List[Problem] = []
+    payloads: Dict[str, dict] = {}
+    for name, path in files.items():
+        if not path.exists():
+            if name == "stats":
+                continue
+            problems.append(Problem(path.name, "", "file is missing"))
+            continue
+        try:
+            payloads[name] = config.load_json(path)
+        except ValueError as error:
+            problems.append(Problem(path.name, "", f"invalid JSON: {error}"))
+
+    for name, payload in payloads.items():
+        problems.extend(validate_schema(name, payload))
+
+    if "techniques" not in payloads:
+        return problems
+
+    tax = taxonomy.Taxonomy(payloads["techniques"])
+    registry = {entry["id"]: entry for entry in payloads.get("dbms", {}).get("dbms", [])}
+
+    from .dataset import LIST_FIELD
+    for name, field in LIST_FIELD.items():
+        records = payloads.get(name, {}).get(field, [])
+        problems.extend(check_unique_ids(name, records))
+        problems.extend(check_evidence(name, records))
+
+    problems.extend(check_bugs(payloads.get("bugs", {}).get("bugs", []), registry, tax))
+    problems.extend(check_recognition_highlights(
+        payloads.get("papers", {}).get("papers", [])))
+    problems.extend(check_talks(payloads.get("talks", {}).get("talks", [])))
+    problems.extend(check_readable_excerpts(
+        payloads.get("papers", {}).get("papers", [])))
+    problems.extend(check_paper_records())
+    problems.extend(check_affiliation_rulings(
+        payloads.get("bugs", {}).get("bugs", []),
+        payloads.get("people", {}).get("people", [])))
+    problems.extend(check_papers(payloads.get("papers", {}).get("papers", []), tax))
+    problems.extend(check_adoption(
+        payloads.get("adoption", {}).get("adoption", []), registry, tax))
+    problems.extend(check_resources(
+        payloads.get("resources", {}).get("resources", []), registry, tax))
+    problems.extend(check_paper_notes(payloads.get("papers", {}).get("papers", [])))
+    problems.extend(check_review_queue())
+    return problems
+
+
+def check_review_queue(payload: Optional[dict] = None) -> List[Problem]:
+    """Validate the queue of candidates no rule settled.
+
+    Two things the schema cannot see. An id in both lists means an item was
+    dismissed and is still being raised, which is the failure the queue exists
+    to prevent -- a person rules on something and the next run asks again. And
+    a duplicated id means two different candidates share one decision, so
+    dismissing one silently dismisses the other.
+    """
+    schema_path = config.SCHEMA_DIR / "needs_review.schema.json"
+    if not schema_path.exists():
+        return [Problem("needs_review", "", f"no schema file at {schema_path}")]
+    if payload is None:
+        path = config.DATA_DIR / "needs_review.json"
+        if not path.exists():
+            return []
+        try:
+            payload = config.load_json(path)
+        except ValueError as error:
+            return [Problem("needs_review", "", f"invalid JSON: {error}")]
+
+    schema = config.load_json(schema_path)
+    validator = jsonschema.Draft7Validator(schema, resolver=_resolver(schema))
+    problems = [
+        Problem("needs_review", "/".join(str(part) for part in error.absolute_path),
+                error.message)
+        for error in sorted(validator.iter_errors(payload), key=lambda e: list(e.path))
+    ]
+
+    seen: Dict[str, str] = {}
+    for name in ("open", "dismissed"):
+        for entry in payload.get(name) or []:
+            item_id = entry.get("id")
+            if not item_id:
+                continue
+            if item_id in seen:
+                where = seen[item_id]
+                problems.append(Problem(
+                    "needs_review", item_id,
+                    f"also in {where}"
+                    if where != name else f"listed twice in {name}"))
+            else:
+                seen[item_id] = name
+    return problems
+
+def check_paper_notes(papers: List[dict]) -> List[Problem]:
+    """Validate each paper's note file, and that it belongs to a real paper.
+
+    An orphan note is a real risk here: the notes are one file per paper and
+    nothing else points at them, so a paper leaving the dataset would otherwise
+    leave a note behind that no longer describes anything.
+    """
+    from . import notes
+
+    if not notes.NOTES_DIR.is_dir():
+        return []
+    schema_path = config.SCHEMA_DIR / "paper-note.schema.json"
+    if not schema_path.exists():
+        return [Problem("paper_notes", "", f"no schema file at {schema_path}")]
+    schema = config.load_json(schema_path)
+    validator = jsonschema.Draft7Validator(schema, resolver=_resolver(schema))
+
+    expected = {notes.note_path(paper).name for paper in papers
+                if not paper.get("is_sqlancer_publication")}
+    ids = {paper["id"] for paper in papers}
+
+    problems: List[Problem] = []
+    for path in sorted(notes.NOTES_DIR.glob("*.json")):
+        label = f"paper_notes/{path.name}"
+        if path.name not in expected:
+            problems.append(Problem(label, "", "note for a paper not in papers.json"))
+            continue
+        try:
+            payload = config.load_json(path)
+        except ValueError as error:
+            problems.append(Problem(label, "", f"invalid JSON: {error}"))
+            continue
+        for error in sorted(validator.iter_errors(payload), key=lambda e: list(e.path)):
+            problems.append(Problem(
+                label, "/".join(str(part) for part in error.absolute_path),
+                error.message))
+        paper_id = (payload.get("paper") or {}).get("id")
+        if paper_id not in ids:
+            problems.append(Problem(label, "paper/id",
+                                    f"unknown paper {paper_id!r}"))
+    return problems
+
+
+def main() -> int:
+    problems = validate_all()
+    for problem in problems:
+        print(problem)
+    if problems:
+        print(f"\n{len(problems)} validation problem(s)")
+        return 1
+    print("impact data: all files valid")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/impact/worklist.py b/tools/impact/worklist.py
new file mode 100644
index 0000000..68d1f88
--- /dev/null
+++ b/tools/impact/worklist.py
@@ -0,0 +1,196 @@
+"""Generates the list of papers whose full text has to be fetched by hand.
+
+Two thirds of this corpus is published by ACM, IEEE or Springer, none of which
+serve PDFs to an automated client, so for those papers the only evidence the
+pipeline can gather on its own is a citation index's two-sentence contexts.
+
+This writes a worklist: what to download, where it lives, and the exact filename
+to save it as so the pipeline picks it up. Papers whose citation contexts already
+hint at a relationship come first, since those are the ones where reading the
+paper is most likely to change what the dataset says. Papers already supplied,
+or reachable through arXiv or PVLDB, are left out.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import List, Optional
+
+from . import config
+from .collectors import fulltext
+
+OUTPUT = config.CACHE_DIR / "papers-pdf" / "WANTED.md"
+
+# A relationship the collectors could not settle, on a paper worth reading.
+UNDECIDED = ("uncertain", "insufficient_evidence")
+
+JUDGEMENTS = ("uses_infrastructure", "extends_technique", "compares_with")
+
+
+# Wording in a citation sentence that suggests a relationship the contexts were
+# too short to settle. A paper whose citing sentences say "implemented on top
+# of" is one where the full text is likely to change what the dataset records;
+# a paper that only ever appears in a list of references is not.
+RELATIONSHIP_HINT = re.compile(
+    r"\b(?:implement\w*|built?\s+on|based\s+on|top\s+of|extend\w*|adapt\w*|"
+    r"baselines?|compare[ds]?|evaluat\w*|derive[ds]?|integrat\w*|"
+    r"state[- ]of[- ]the[- ]art|outperform\w*)\b", re.IGNORECASE)
+
+
+def _hint_count(paper: dict) -> int:
+    """How many of this paper's citation sentences hint at a relationship."""
+    hits = 0
+    for entry in paper.get("relationships", {}).values():
+        for item in (entry.get("evidence") or []):
+            excerpt = item.get("excerpt")
+            if excerpt and RELATIONSHIP_HINT.search(excerpt):
+                hits += 1
+    return hits
+
+
+# A proxy that carries an institutional subscription. IEEE will not serve a
+# paper otherwise, so its links go through this; the others resolve without it,
+# and prefixing them anyway would only add a login step.
+PROXY = "https://libproxy1.nus.edu.sg/login?url="
+
+
+def download_url(doi: Optional[str], url: Optional[str], publisher: str) -> str:
+    """Where to get the PDF, as directly as each publisher allows.
+
+    ACM and Springer both expose a PDF at a path derived from the DOI, so those
+    links land on the file itself. IEEE's PDF path needs an article number that
+    the DOI does not contain, so its link resolves the DOI to the article page
+    and the download is one click from there.
+
+    Every link goes through the library proxy. A paywalled paper needs it, and
+    an open-access one is unharmed by it -- the proxy passes the request on
+    either way, so there is no reason to make the reader judge which is which.
+    """
+    if not doi:
+        return f"{PROXY}{url}" if url else ""
+    if publisher == "ACM":
+        direct = f"https://dl.acm.org/doi/pdf/{doi}"
+    elif publisher == "Springer":
+        direct = f"https://link.springer.com/content/pdf/{doi}.pdf"
+    else:
+        direct = f"https://doi.org/{doi}"
+    return f"{PROXY}{direct}"
+
+
+def _publisher(doi: Optional[str]) -> str:
+    doi = (doi or "").lower()
+    if doi.startswith("10.1145/"):
+        return "ACM"
+    if doi.startswith("10.1109/"):
+        return "IEEE"
+    if doi.startswith("10.1007/"):
+        return "Springer"
+    if doi.startswith("10.14778/"):
+        return "PVLDB"
+    return "other"
+
+
+def wanted(papers: Optional[List[dict]] = None) -> List[dict]:
+    """Papers needing a hand-supplied PDF, most useful first."""
+    if papers is None:
+        papers = config.load_json(config.DATA_FILES["papers"])["papers"]
+
+    rows = []
+    for paper in papers:
+        if paper.get("is_sqlancer_publication"):
+            continue
+        if fulltext.reachable_without_help(paper):
+            continue
+        filename = fulltext.wanted_filename(paper.get("doi"),
+                                            paper.get("arxiv_id"),
+                                            paper.get("s2_paper_id"))
+        if not filename:
+            continue
+        if fulltext.local_pdf(paper.get("doi"), paper.get("arxiv_id"),
+                              paper.get("s2_paper_id")):
+            continue  # already supplied
+
+        undecided = [key for key in JUDGEMENTS
+                     if paper["relationships"][key]["value"] in UNDECIDED]
+        settled = [key for key in JUDGEMENTS
+                   if paper["relationships"][key]["value"] == "yes"]
+        rows.append({
+            "hints": _hint_count(paper),
+            "title": paper["title"],
+            "url": paper.get("url") or "",
+            "filename": filename,
+            "year": paper.get("year"),
+            "venue": paper.get("venue") or "",
+            "publisher": _publisher(paper.get("doi")),
+            "download_url": download_url(paper.get("doi"), paper.get("url"),
+                                         _publisher(paper.get("doi"))),
+            "undecided": undecided,
+            "settled": settled,
+        })
+
+    # Ordered by how much reading the paper is likely to change. First those
+    # whose contexts already tie them to SQLancer, where a wrong call costs
+    # most; then those whose citing sentences hint at a relationship the two
+    # indexed sentences could not settle; then the rest, newest first, since a
+    # recent paper is the more likely to still be uncatalogued elsewhere.
+    rows.sort(key=lambda r: (-len(r["settled"]), -r["hints"], -(r["year"] or 0),
+                             r["title"].lower()))
+    return rows
+
+
+def render(rows: List[dict]) -> str:
+    lines = [
+        "# Papers to download",
+        "",
+        f"{len(rows)} papers whose publisher will not serve a PDF to an "
+        "automated client. Downloading one is entirely optional: a missing "
+        "paper costs coverage for that paper, never correctness.",
+        "",
+        "Save each file into this directory under the **Save as** name — that "
+        "is how the pipeline finds it — then re-run:",
+        "",
+        "```sh",
+        "python3 -m tools.impact.run collect --only papers --full",
+        "```",
+        "",
+        "Claims drawn from a supplied PDF quote it verbatim, exactly as they "
+        "would from a preprint, so they stay checkable by anyone holding the "
+        "same paper.",
+        "",
+        "Papers on arXiv, in PVLDB, or at a USENIX venue are not listed: the "
+        "pipeline reads those itself.",
+        "",
+        "IEEE links go through the NUS library proxy, which is what makes them "
+        "resolve; ACM and Springer links point straight at the PDF.",
+        "",
+        "| # | Paper | Download | Known so far | Save as |",
+        "| ---: | --- | --- | --- | --- |",
+    ]
+    for index, row in enumerate(rows, start=1):
+        title = row["title"].replace("|", "\\|")
+        link = f"[{title}]({row['url']})" if row["url"] else title
+        year = f" ({row['year']})" if row["year"] else ""
+        # What the citation contexts already established, so it is obvious
+        # which papers the full text would actually change.
+        known = ", ".join(k.replace("_", " ") for k in row["settled"])
+        if not known and row["hints"]:
+            known = f"cites only ({row['hints']} sentences hint at more)"
+        download = (f"[{row['publisher']} PDF]({row['download_url']})"
+                    if row["download_url"] else row["publisher"])
+        lines.append(
+            f"| {index} | {link}{year} | {download} | "
+            f"{known or 'cites only'} | `{row['filename']}` |")
+    lines.append("")
+    return "\n".join(lines)
+
+
+def main() -> int:
+    rows = wanted()
+    OUTPUT.parent.mkdir(parents=True, exist_ok=True)
+    OUTPUT.write_text(render(rows), encoding="utf-8")
+    print(f"{OUTPUT}: {len(rows)} paper(s) to download")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/papers/__init__.py b/tools/papers/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/tools/papers/analysis.py b/tools/papers/analysis.py
new file mode 100644
index 0000000..5a2f328
--- /dev/null
+++ b/tools/papers/analysis.py
@@ -0,0 +1,223 @@
+"""Attaches an analysis to a paper's record, and refuses a bad one.
+
+The analysis is the one part of a record written by a model rather than read off
+a page, so it is the part that needs checking. Three things are enforced here,
+and each rejects rather than repairs:
+
+1. **Every cited mention must exist.** A claim resting on "M12" in a paper with
+   nine mentions is not a claim.
+2. **A "yes" must cite something.** A positive relationship with no mention ids
+   is an assertion, and this dataset does not publish assertions.
+3. **Quotations are attached, never supplied.** The model names mention ids; the
+   sentences come from the inventory, which was read off the paper. A fabricated
+   quotation is not something the format can express.
+
+What the model does supply in its own words -- the summary, the narrative, the
+reasoning -- is stored as such and labelled, so a reader can tell at a glance
+which parts of a record are evidence and which are description.
+"""
+
+from __future__ import annotations
+
+from typing import Dict, List, Optional, Sequence
+
+from ..impact.util import now
+from . import prompt, store
+
+ANALYSIS_VERSION = "paper-analysis-v1"
+
+# Cited like a mention, but standing for what the paper's repository shows
+# rather than what its text says.
+ARTIFACT_ID = "ARTIFACT"
+
+
+class Rejected(ValueError):
+    """The answer does not meet the conditions for being recorded."""
+
+
+# A survey can mention SQLancer eighty times, and reading all of them decides no
+# more than reading the informative ones. What matters is which are dropped: a
+# sentence a check fired on, or one that names the tool in the paper's own
+# prose, can settle a relationship, while the fortieth citation marker in a
+# related-work list cannot.
+MAX_MENTIONS = 32
+
+
+def _priority(mention: dict, flagged: set) -> tuple:
+    found = set(mention.get("found_by_all") or [])
+    section = (mention.get("section") or "").lower()
+    return (
+        0 if mention["id"] in flagged else 1,
+        0 if "name" in found else (1 if "technique" in found else 2),
+        # Where a paper says what it built is where reuse is stated.
+        0 if any(word in section for word in
+                 ("implement", "evaluat", "experiment", "approach", "design"))
+        else 1,
+        mention.get("char_offset", 0),
+    )
+
+
+def task(record: dict, *, compact: bool = True,
+         max_mentions: int = MAX_MENTIONS) -> dict:
+    """What a reader needs in order to analyse this paper.
+
+    ``compact`` drops the surrounding paragraphs, which are kept in the record
+    for later re-reading but are rarely what decides a judgement.
+    """
+    flagged = {mention_id
+               for ids in ((record.get("checks") or {}).get("suggests") or {}).values()
+               for mention_id in ids}
+    chosen = sorted(record["mentions"], key=lambda m: _priority(m, flagged))
+    chosen = sorted(chosen[:max_mentions],
+                    key=lambda m: m.get("char_offset", 0))
+    omitted = len(record["mentions"]) - len(chosen)
+
+    mentions = []
+    for mention in chosen:
+        entry = {
+            "id": mention["id"],
+            "found_by": mention.get("found_by_all") or [],
+            "section": mention.get("section"),
+            "page": mention.get("page"),
+            "sentence": mention["sentence"],
+        }
+        if mention.get("techniques"):
+            entry["techniques"] = mention["techniques"]
+        if mention.get("cited_reference"):
+            entry["cites"] = {
+                "number": mention["cited_reference"]["number"],
+                "matched_as": mention["cited_reference"].get("matched_as"),
+                "text": mention["cited_reference"]["text"][:160],
+            }
+        if not compact:
+            entry["context_before"] = mention.get("context_before")
+            entry["context_after"] = mention.get("context_after")
+        mentions.append(entry)
+
+    paper = record["paper"]
+    return {
+        "paper_id": paper["id"],
+        "title": paper.get("title"),
+        "year": paper.get("year"),
+        "venue": paper.get("venue"),
+        "has_fulltext": record["document"]["has_fulltext"],
+        "abstract_available": bool(record.get("abstract")),
+        "artifact": record.get("artifact"),
+        "sqlancer_references": [
+            {"number": r["number"], "matched_as": r.get("matched_as"),
+             "text": r["text"][:160]}
+            for r in record.get("sqlancer_references") or []
+        ],
+        "mentions_omitted": omitted,
+        "checks": (record.get("checks") or {}).get("suggests") or {},
+        "suppressed_checks": (record.get("checks") or {}).get("suppressed") or [],
+        "mentions": mentions,
+    }
+
+
+def pending(records: Optional[Sequence[dict]] = None) -> List[dict]:
+    """Records with no analysis yet, most informative first.
+
+    A paper with more mentions has more for a reader to weigh, and is the more
+    likely to be saying something the dataset does not know.
+    """
+    records = records if records is not None else store.all_records()
+    waiting = [r for r in records if not r.get("analysis")]
+    waiting.sort(key=lambda r: (-len(r["mentions"]),
+                                -(r["paper"].get("year") or 0)))
+    return waiting
+
+
+def apply(paper_id: str, answer: dict, *, model: str,
+          produced_by: str = "claude_code_session") -> dict:
+    """Record an analysis against a paper, after checking it.
+
+    Raises :class:`Rejected` rather than storing something that cannot be
+    checked: a silently dropped claim is worse than a loud refusal, because
+    nobody looks at what they were not told about.
+    """
+    record = store.load(paper_id)
+    if record is None:
+        raise Rejected(f"no record for {paper_id}")
+
+    known = {mention["id"]: mention for mention in record["mentions"]}
+
+    # The artifact is evidence too, and often the only evidence there is: a
+    # paper whose repository is a SQLancer fork need never say so in its text.
+    # It is citable under a reserved id, and only when the record actually
+    # carries inspection findings, so the id cannot be invented.
+    artifact = record.get("artifact") or {}
+    if artifact.get("markers"):
+        known[ARTIFACT_ID] = {
+            "id": ARTIFACT_ID,
+            "sentence": ("Artifact " + str(artifact.get("url") or "") +
+                         " carries: " + ", ".join(artifact["markers"])),
+            "section": "artifact inspection",
+            "page": None,
+        }
+
+    for key in ("summary", "narrative", "relationships"):
+        if not answer.get(key):
+            raise Rejected(f"{paper_id}: the answer has no {key}")
+
+    roles = answer.get("roles") or {}
+    for mention_id in roles:
+        if mention_id == ARTIFACT_ID:
+            continue
+        if mention_id not in known:
+            raise Rejected(f"{paper_id}: role given for unknown mention "
+                           f"{mention_id}")
+        if roles[mention_id] not in prompt.ROLES:
+            raise Rejected(f"{paper_id}: {mention_id} has unknown role "
+                           f"{roles[mention_id]!r}")
+
+    relationships: Dict[str, dict] = {}
+    for key in prompt.RELATIONSHIPS:
+        entry = (answer["relationships"] or {}).get(key)
+        if entry is None:
+            raise Rejected(f"{paper_id}: no answer for {key}")
+        value = entry.get("value")
+        if value not in prompt.ANSWERS:
+            raise Rejected(f"{paper_id}: {key} has unknown value {value!r}")
+        ids = list(entry.get("mention_ids") or [])
+        unknown = [i for i in ids if i not in known]
+        if unknown:
+            raise Rejected(f"{paper_id}: {key} cites mentions that do not "
+                           f"exist: {', '.join(unknown)}")
+        if value == "yes" and not ids:
+            raise Rejected(f"{paper_id}: {key} is 'yes' with nothing cited")
+
+        stored = {
+            "value": value,
+            "mention_ids": ids,
+            # The quotations come from the inventory, not from the answer.
+            "quotes": [{"mention_id": i, "sentence": known[i]["sentence"],
+                        "section": known[i].get("section"),
+                        "page": known[i].get("page")} for i in ids],
+            "reasoning": entry.get("reasoning") or "",
+        }
+        if entry.get("techniques"):
+            stored["techniques"] = entry["techniques"]
+        if key == "uses_infrastructure" and entry.get("reuse_kind"):
+            if entry["reuse_kind"] not in prompt.REUSE_KINDS:
+                raise Rejected(f"{paper_id}: unknown reuse_kind "
+                               f"{entry['reuse_kind']!r}")
+            stored["reuse_kind"] = entry["reuse_kind"]
+        relationships[key] = stored
+
+    record["analysis"] = {
+        "analysis_version": ANALYSIS_VERSION,
+        "prompt_version": prompt.PROMPT_VERSION,
+        "produced_by": produced_by,
+        "model": model,
+        "generated_at": now(),
+        "is_model_written": True,
+        "summary": answer["summary"].strip(),
+        "narrative": answer["narrative"].strip(),
+        "roles": roles,
+        "relationships": relationships,
+        "disagreements": list(answer.get("disagreements") or []),
+        "unresolved": list(answer.get("unresolved") or []),
+    }
+    store.save(record)
+    return record
diff --git a/tools/papers/checks.py b/tools/papers/checks.py
new file mode 100644
index 0000000..33307c3
--- /dev/null
+++ b/tools/papers/checks.py
@@ -0,0 +1,119 @@
+"""Regular expressions run over the mention inventory, as a second opinion.
+
+These do not decide anything. They exist because a model reading a hundred
+papers will occasionally miss a sentence that plainly says "we implemented our
+tool on top of SQLancer", and because a disagreement between a pattern and a
+judgement is worth surfacing either way: sometimes the pattern is wrong,
+sometimes the reading is, and which it is can only be settled by looking.
+
+The patterns are the ones the old pipeline used to decide with, kept here in
+the role they are actually good at. Their weaknesses are on record: "we adapted
+SQLancer as a baseline for comparison" matched the extension pattern until the
+sentence was read to the end, which is exactly the sort of thing an advisory
+check should raise rather than settle.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Dict, List, Sequence
+
+CHECKS_VERSION = "paper-checks-v1"
+
+_TOOL = r"(?:SQLancer\+\+|SQLancer|ShQveL)"
+_TECH = (r"(?:TLP|NoREC|PQS|QPG|CERT|DQP|CODDTest|Ternary Logic Partitioning"
+         r"|Non-optimizing Reference Engine Construction|Pivoted Query Synthesis"
+         r"|Query Plan Guidance|query partitioning)")
+
+PATTERNS = {
+    "uses_infrastructure": (
+        # A claim about the authors' own artefact. "we" or "our" is required:
+        # without it a figure caption reads as a claim of reuse.
+        re.compile(rf"\b(?:we|our)\b[^.]{{0,120}}?"
+                   rf"\b(?:implemented|implements|implement|built|build|"
+                   rf"developed|develops|prototyped|prototype|based|extended|"
+                   rf"extends|modified|integrated|adapted|adopt(?:ed)?)\b"
+                   rf"[^.]{{0,80}}?\b(?:on\s+top\s+of|upon|on|in|into|using|"
+                   rf"from)\s+(?:the\s+)?{_TOOL}\b", re.IGNORECASE),
+        re.compile(rf"\b(?:is|are|was|were)\s+(?:implemented|built|developed|"
+                   rf"based|derived)\b[^.]{{0,60}}?\b(?:on\s+top\s+of|upon|on|"
+                   rf"in|from)\s+(?:the\s+)?{_TOOL}\b", re.IGNORECASE),
+        re.compile(rf"\bframework\s+is\s+derived\s+from\s+{_TOOL}\b", re.IGNORECASE),
+    ),
+    "compares_with": (
+        re.compile(rf"\b(?:we|our)\b[^.]{{0,140}}?\b(?:compare[ds]?|comparison|"
+                   rf"evaluat\w+|benchmark\w*|baseline[s]?|against|outperform\w*)"
+                   rf"\b[^.]{{0,140}}?\b(?:{_TOOL}|{_TECH})\b", re.IGNORECASE),
+        re.compile(rf"\b(?:{_TOOL}|{_TECH})\b[^.]{{0,80}}?\bas\s+"
+                   rf"(?:a\s+|our\s+|the\s+)?baseline[s]?\b", re.IGNORECASE),
+        re.compile(rf"\b(?:than|versus|vs\.?)\s+(?:{_TOOL}|{_TECH})\b",
+                   re.IGNORECASE),
+    ),
+    "extends_technique": (
+        re.compile(rf"\b(?:we|our)\b[^.]{{0,120}}?\b(?:extend(?:ed|s)?|"
+                   rf"generali[sz]\w+|adapt(?:ed|s)?|build[s]?\s+(?:up)?on|"
+                   rf"inspired\s+by|derive[ds]?\s+from|improve[sd]?\s+(?:up)?on)"
+                   rf"\b[^.]{{0,80}}?\b(?:{_TECH}|{_TOOL})\b", re.IGNORECASE),
+    ),
+    "describes_as_state_of_the_art": (
+        re.compile(rf"\b(?:{_TOOL}|{_TECH})\b[^.]{{0,90}}?\b(?:state[- ]of[- ]"
+                   rf"the[- ]art|leading|most\s+effective|best[- ]known)\b",
+                   re.IGNORECASE),
+        re.compile(rf"\b(?:state[- ]of[- ]the[- ]art|leading|most\s+effective)\b"
+                   rf"[^.]{{0,90}}?\b(?:{_TOOL}|{_TECH})\b", re.IGNORECASE),
+    ),
+}
+
+# Wording that turns a sentence into a statement about somebody else's work.
+RELATED_WORK = re.compile(
+    r"\b(?:previous|prior|existing|earlier|related)\s+(?:work|approaches|"
+    r"studies|techniques|tools)\b|\bhas\s+been\s+(?:proposed|shown)\b",
+    re.IGNORECASE)
+
+# "as a baseline" turns adopting a tool into measuring against it, which is the
+# opposite of building on it. This is the sentence that made the old pipeline
+# record TSGuard as extending a technique.
+AS_A_BASELINE = re.compile(
+    r"\bas\s+(?:a\s+|our\s+|the\s+)?baselines?\b|\bfor\s+comparison\b",
+    re.IGNORECASE)
+
+NEVER_WITH_BASELINE = ("extends_technique", "uses_infrastructure")
+
+
+def run(mentions: Sequence[dict]) -> Dict[str, object]:
+    """Which mentions each pattern fires on, and what qualifies the result."""
+    fired: Dict[str, List[str]] = {key: [] for key in PATTERNS}
+    notes: List[dict] = []
+
+    for mention in mentions:
+        sentence = mention.get("sentence") or ""
+        baseline = bool(AS_A_BASELINE.search(sentence))
+        related = bool(RELATED_WORK.search(sentence))
+        for key, patterns in PATTERNS.items():
+            if not any(pattern.search(sentence) for pattern in patterns):
+                continue
+            if baseline and key in NEVER_WITH_BASELINE:
+                notes.append({
+                    "mention_id": mention["id"],
+                    "pattern": key,
+                    "suppressed_because": (
+                        "the sentence says the tool was taken up as a baseline, "
+                        "which is comparison rather than reuse or extension"),
+                })
+                continue
+            if related and key != "describes_as_state_of_the_art":
+                notes.append({
+                    "mention_id": mention["id"],
+                    "pattern": key,
+                    "suppressed_because": (
+                        "the sentence is framed as related work, so it "
+                        "describes somebody else's approach"),
+                })
+                continue
+            fired[key].append(mention["id"])
+
+    return {
+        "checks_version": CHECKS_VERSION,
+        "suggests": {key: ids for key, ids in fired.items() if ids},
+        "suppressed": notes,
+    }
diff --git a/tools/papers/cli.py b/tools/papers/cli.py
new file mode 100644
index 0000000..b808179
--- /dev/null
+++ b/tools/papers/cli.py
@@ -0,0 +1,98 @@
+"""The subproject's command line.
+
+    python3 -m tools.papers.cli collect     build a record for every paper
+    python3 -m tools.papers.cli show ID     print one record
+    python3 -m tools.papers.cli report      what has been extracted and analysed
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+from typing import List, Optional
+
+from . import collect, store
+
+
+def _report() -> int:
+    records = store.all_records()
+    if not records:
+        print("no records yet; run `collect`")
+        return 1
+    fulltext = [r for r in records if r["document"]["has_fulltext"]]
+    analysed = [r for r in records if r.get("analysis")]
+    mentions = sum(len(r["mentions"]) for r in records)
+    by_marker = sum(1 for r in records
+                    for m in r["mentions"]
+                    if m.get("found_by_all") == ["citation_marker"])
+    print(f"records            {len(records)}")
+    print(f"  with full text   {len(fulltext)}")
+    print(f"  analysed         {len(analysed)}")
+    print(f"mentions           {mentions}")
+    print(f"  only via a cited reference, never by name   {by_marker}")
+    without = [r for r in records if not r["mentions"]]
+    if without:
+        print(f"{len(without)} record(s) with no mention at all")
+    return 0
+
+
+def main(argv: Optional[List[str]] = None) -> int:
+    parser = argparse.ArgumentParser(prog="tools.papers.cli")
+    sub = parser.add_subparsers(dest="command", required=True)
+
+    collect_parser = sub.add_parser("collect", help="build the per-paper records")
+    collect_parser.add_argument("--only", nargs="*", help="paper ids")
+    collect_parser.add_argument("--limit", type=int)
+    collect_parser.add_argument("--fulltext-only", action="store_true",
+                                help="skip papers with no full text")
+
+    show_parser = sub.add_parser("show", help="print one record")
+    show_parser.add_argument("paper_id")
+    show_parser.add_argument("--mentions", action="store_true")
+
+    sub.add_parser("report", help="summarise what has been extracted")
+    sub.add_parser("pages", help="regenerate the per-paper pages")
+
+    args = parser.parse_args(argv)
+
+    if args.command == "collect":
+        counts = collect.run(only=args.only, limit=args.limit,
+                             with_fulltext_only=args.fulltext_only)
+        print(f"{counts['written']} written, {counts['unchanged']} unchanged; "
+              f"{counts['fulltext']} from full text, "
+              f"{counts['metadata_only']} from metadata; "
+              f"{counts['mentions']} mentions")
+        # A record without its page is a link the impact page cannot follow,
+        # and a page without its record claims figures nothing produces.
+        from . import pages as pages_module
+
+        outcome = pages_module.write_all()
+        moved = [key for key, state in outcome.items() if state != "unchanged"]
+        print(f"{len(outcome)} paper page(s), {len(moved) or 'none'} changed")
+        return 0
+
+    if args.command == "show":
+        record = store.load(args.paper_id)
+        if record is None:
+            print(f"no record for {args.paper_id}")
+            return 1
+        if args.mentions:
+            for mention in record["mentions"]:
+                print(f"{mention['id']} [{mention.get('section')}, "
+                      f"p{mention.get('page')}] {mention['found_by_all']}")
+                print(f"    {mention['sentence'][:200]}")
+            return 0
+        print(json.dumps(record, indent=2, ensure_ascii=False)[:4000])
+        return 0
+
+    if args.command == "pages":
+        from . import pages as pages_module
+        return pages_module.main()
+    if args.command == "report":
+        return _report()
+    return 2
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/papers/collect.py b/tools/papers/collect.py
new file mode 100644
index 0000000..54f672a
--- /dev/null
+++ b/tools/papers/collect.py
@@ -0,0 +1,135 @@
+"""Builds a record for every paper, from the best source available for each.
+
+A paper with a supplied PDF is read in full. A paper without one still gets a
+record, built from its abstract and the sentences a citation index recorded --
+fewer sources, the same shape -- so nothing downstream has to ask whether a
+paper is one kind or the other.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Dict, List, Optional, Tuple
+
+from ..impact import config, taxonomy
+from ..impact.cache import Caches
+from ..impact.collectors import fulltext
+from ..impact.http import Fetcher
+from . import checks, extract, mentions, store
+
+
+def _document_from_metadata(paper: dict) -> dict:
+    """A stand-in document for a paper whose full text we do not have.
+
+    The abstract and the citation sentences, joined into something the same
+    inventory can run over. Offsets and pages are meaningless here and are
+    recorded as such rather than invented.
+    """
+    parts: List[str] = []
+    abstract = (paper.get("abstract") or "").strip()
+    if abstract:
+        parts.append(f"Abstract\n\n{abstract}")
+    seen = set()
+    for entry in (paper.get("relationships") or {}).values():
+        for item in (entry.get("evidence") or []):
+            excerpt = (item.get("excerpt") or "").strip()
+            if excerpt and excerpt not in seen:
+                seen.add(excerpt)
+                parts.append(excerpt)
+    text = "\n\n".join(parts)
+    return {"text": text, "pages": [], "page_count": 0, "sections": [],
+            "references": [], "references_start": None, "has_outline": False}
+
+
+def source_for(paper: dict, fetcher: Fetcher) -> Tuple[dict, str, Optional[str]]:
+    """The best document for this paper: ``(document, route, url)``."""
+    local = fulltext.local_pdf(paper.get("doi"), paper.get("arxiv_id"),
+                               paper.get("s2_paper_id"))
+    if local:
+        raw = _read(fetcher, local)
+        if raw:
+            document = extract.assemble(raw)
+            if document.get("text"):
+                return document, "supplied_pdf", paper.get("url")
+
+    for url in fulltext.usenix_pdf_urls(paper):
+        raw = _read(fetcher, url)
+        if raw:
+            document = extract.assemble(raw)
+            if document.get("text"):
+                return document, "usenix", url
+
+    if paper.get("arxiv_id"):
+        url = f"https://arxiv.org/pdf/{paper['arxiv_id']}"
+        raw = _read(fetcher, url)
+        if raw:
+            document = extract.assemble(raw)
+            if document.get("text"):
+                return document, "arxiv", url
+
+    return _document_from_metadata(paper), "metadata", paper.get("url")
+
+
+def _read(fetcher: Fetcher, url: str) -> Optional[bytes]:
+    try:
+        if url.startswith("file://"):
+            import pathlib
+            from urllib.parse import urlparse
+            from urllib.request import url2pathname
+
+            return pathlib.Path(url2pathname(urlparse(url).path)).read_bytes()
+        raw, _ = fetcher.fetch_binary(url, max_age_days=365)
+        return raw if raw and raw.lstrip().startswith(b"%PDF") else None
+    except Exception:
+        return None
+
+
+def artifact_for(paper: dict) -> Optional[dict]:
+    """What the pipeline already learned from this paper's repository."""
+    artifacts = paper.get("artifacts") or []
+    if not artifacts:
+        return None
+    first = artifacts[0]
+    return {
+        "url": first.get("url"),
+        "markers": first.get("sqlancer_markers") or [],
+        "evidence": first.get("marker_evidence") or [],
+    }
+
+
+def run(*, only: Optional[List[str]] = None, limit: Optional[int] = None,
+        with_fulltext_only: bool = False) -> Dict[str, int]:
+    papers = [p for p in config.load_json(config.DATA_FILES["papers"])["papers"]
+              if not p.get("is_sqlancer_publication")]
+    if only:
+        wanted = set(only)
+        papers = [p for p in papers if p["id"] in wanted]
+
+    caches = Caches()
+    fetcher = Fetcher(caches.artifacts, min_interval=0.5, max_retries=2,
+                      timeout=45.0)
+    tax = taxonomy.load()
+
+    counts = {"written": 0, "unchanged": 0, "fulltext": 0, "metadata_only": 0,
+              "mentions": 0}
+    done = 0
+    for paper in papers:
+        document, route, url = source_for(paper, fetcher)
+        if with_fulltext_only and route == "metadata":
+            continue
+        inventory = mentions.build(document, tax)
+        findings = checks.run(inventory["mentions"])
+        existing = store.load(paper["id"]) or {}
+        record = store.build(paper, document, inventory, findings, route=route,
+                             source_url=url, artifact=artifact_for(paper),
+                             analysis=existing.get("analysis"))
+        if store.save(record):
+            counts["written"] += 1
+        else:
+            counts["unchanged"] += 1
+        counts["fulltext" if route != "metadata" else "metadata_only"] += 1
+        counts["mentions"] += len(inventory["mentions"])
+        done += 1
+        if limit and done >= limit:
+            break
+    return counts
diff --git a/tools/papers/extract.py b/tools/papers/extract.py
new file mode 100644
index 0000000..234b32e
--- /dev/null
+++ b/tools/papers/extract.py
@@ -0,0 +1,456 @@
+"""Turns a paper into a structured document that can be judged without reopening it.
+
+This is the expensive step -- parsing a PDF, finding where its sections begin,
+splitting its bibliography -- and the whole point of the subproject is that it
+happens once per paper and everything it learns is kept.
+
+Two families of PDF cover almost this entire corpus, and they need different
+handling:
+
+* ACM papers from ``acmart`` carry a real outline, numbered headings
+  ("3.1 Overview") and a bibliography whose entries start ``[n]``.
+* IEEE papers carry no outline at all. Their headings are roman-numbered and
+  set in small caps, which text extraction renders with the first letter
+  detached -- "III. A PPROACH" -- so a heading has to be repaired before it can
+  be recognised.
+
+Nothing here decides anything about SQLancer. It produces the document; the
+mention inventory and the analysis are built on top of it.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Dict, List, Optional, Tuple
+
+EXTRACTOR_VERSION = "paper-extract-v1"
+
+MAX_PDF_BYTES = 40_000_000
+
+# A heading, in the two conventions this corpus uses. Both are anchored to a
+# whole line: a numbered clause in running text should not be mistaken for one.
+NUMBERED_HEADING = re.compile(
+    r"^\s*(\d{1,2}(?:\.\d{1,2}){0,2})\.?\s+([A-Z][^\n]{2,70})\s*$")
+ROMAN_HEADING = re.compile(
+    r"^\s*((?:[IVXL]{1,6}|[A-Z])\.)\s+([A-Z][^\n]{2,70})\s*$")
+
+# Small caps come out of extraction with the first letter split off the rest:
+# "I NTRODUCTION", "A PPROACH", "R EFERENCES". Rejoining them is what makes an
+# IEEE paper's structure legible at all.
+SMALL_CAPS = re.compile(r"\b([A-Z])\s([A-Z]{2,})\b")
+
+# Where the bibliography starts, in either convention.
+# A running page number is often glued to the heading with no space between
+# them -- "894REFERENCES" -- which left one paper's whole bibliography
+# unparsed and so gave it no citation-marker mentions at all.
+REFERENCES_HEADING = re.compile(
+    r"^\s*\d{0,4}\s*(?:[IVXL]+\.\s*)?"
+    r"(?:R\s?EFERENCES?|References?|REFERENCES?|BIBLIOGRAPHY|"
+    r"Bibliography)\s*$", re.MULTILINE)
+
+# Where entry one of a bibliography starts, for papers whose heading cannot be
+# matched at all -- one runs the heading into the last line of the conclusion
+# ("...environments. REFERENCE"), and another has no heading in the text layer.
+# A bibliography entry is recognised by what follows the marker rather than by
+# a heading: an author initial, a quoted title, a volume or a page range.
+BIBLIOGRAPHY_START = re.compile(
+    # At the start of a line: a bibliography lays its entries out that way,
+    # while a citation in prose sits inside a sentence.
+    r"(?:^|\n)[ \t]*\[1\]\s{0,3}(?=[^\n]{0,200}?"
+    # Followed by something only a reference carries: an author initial, a
+    # publication year, a volume or page range, or a quoted title. The year is
+    # what reaches a bibliography spelling first names in full -- "Djallel
+    # Bouneffouf. 2016." has no initial to match on.
+    r"(?:[A-Z]\.\s|\b(?:19|20)\d{2}\b|\bvol\.|\bpp\.|[\u201c\u201d\"]))")
+
+# A bibliography entry marker. Numeric styles only: an author-year bibliography
+# has no marker to key citations against, and is reported as unresolvable
+# rather than guessed at.
+REFERENCE_MARKER = re.compile(r"\[(\d{1,3})\]")
+
+# Springer numbers its bibliography "1." rather than "[1]", so the bracketed
+# marker finds nothing at all in an LNCS paper -- which is how Artemis,
+# Differential Monitoring and the Solidity compiler paper each arrived with
+# zero references and, in consequence, no citation-marker mentions. Anchored to
+# the line start and requiring a following space or URL, because "vol. 5," and
+# "pp. 12." appear inside entries constantly.
+DOTTED_MARKER = re.compile(r"(?:^|\n)\s*(\d{1,3})\.(?=\s|https?://)")
+
+# Sections whose content is not the paper's own argument.
+BACK_MATTER = re.compile(
+    r"^(references|bibliography|acknowledg|appendix|artifact appendix)",
+    re.IGNORECASE)
+
+
+def repair_small_caps(text: str) -> str:
+    """Rejoin a small-caps word whose first letter extraction split off."""
+    previous = None
+    while previous != text:
+        previous = text
+        text = SMALL_CAPS.sub(r"\1\2", text)
+    return text
+
+
+# A word broken across a line, whose hyphen the typesetter inserted: the second
+# half starts lowercase, so the hyphen is not the author's and goes away with
+# the break -- "re - vealed" is "revealed".
+BROKEN_WORD = re.compile(r"([A-Za-z])\s*-\s+([a-z])")
+
+# A hyphen the author wrote, with extraction's spaces around it: the second half
+# starts with a capital or a digit, so the hyphen stays -- "Time -Series" is
+# "Time-Series", "STATE -AWARE" is "STATE-AWARE".
+SPACED_HYPHEN = re.compile(r"([A-Za-z])\s+-\s*([A-Z0-9])")
+
+# Some PDFs set a passage with letter-spacing, and extraction puts a space
+# between every glyph: "w i t h i n S Q L a n c e r". Unlike the joins below
+# this cannot be repaired -- the gaps between words are the same width as the
+# gaps inside them, in the file as well as on the page, so the word breaks are
+# not there to recover. Rejoining would give "systemwithinSQLancer"; guessing
+# the breaks from a dictionary would be writing text rather than quoting it.
+# So it is detected and refused instead: a quotation nobody can read is worse
+# than no quotation.
+LETTER_SPACED = re.compile(r"(?:\b[A-Za-z] ){6,}[A-Za-z]\b")
+
+
+def is_letter_spaced(text: str) -> bool:
+    """Whether a passage lost its word breaks to letter-spacing."""
+    return bool(LETTER_SPACED.search(text or ""))
+
+
+# A word running straight into an all-caps one, where extraction lost the space
+# between them: "all columns areNOT NULL", "PQS andTLP". The capital run is what
+# makes splitting safe -- a product name carries its capital at the front
+# ("PostgreSQL", "DuckDB", "CockroachDB"), so a token that *starts* lower-case
+# and then turns to capitals is a join that was never written.
+GLUED_KEYWORD = re.compile(r"\b([a-z]{2,})([A-Z]{2,})")
+
+# Small caps set a word as one capital followed by the rest in smaller
+# capitals, and extraction puts a space between the two runs: "SQL ANCER",
+# "N OREC", "C ONI". Both runs must be capitals and the join must not be two
+# ordinary words ("SQL AND" is two words, and stays two).
+SPLIT_SMALL_CAPS = re.compile(r"\b([A-Z]{1,4})\s([A-Z]{2,})\b")
+
+# Words that follow a capitalised run and are words in their own right, so a
+# join would be wrong.
+NOT_A_SPLIT = {
+    "AND", "OR", "NOT", "THE", "FOR", "WITH", "FROM", "INTO", "USING", "WHERE",
+    "SELECT", "INSERT", "UPDATE", "DELETE", "TABLE", "INDEX", "NULL", "TRUE",
+    "FALSE", "JOIN", "GROUP", "ORDER", "BY", "ON", "IN", "IS", "AS", "ALL",
+    "API", "SQL", "DBMS", "PDF", "URL", "CPU", "GPU", "IEEE", "ACM", "USENIX",
+    "ABSTRACT", "INTRODUCTION", "CONCLUSION", "REFERENCES", "RELATED", "WORK",
+    "EVALUATION", "BACKGROUND", "APPROACH", "IMPLEMENTATION", "DISCUSSION",
+}
+
+
+def _join_small_caps(match: "re.Match") -> str:
+    head, tail = match.group(1), match.group(2)
+    if tail in NOT_A_SPLIT or head in ("A", "I"):
+        return match.group(0)
+    return head + tail
+
+
+def tidy(text: str) -> str:
+    """Repair the artefacts a PDF's text layer leaves behind.
+
+    Extraction is not transcription: a PDF stores glyphs and positions, and
+    turning those back into words introduces damage that is systematic and
+    therefore repairable. A small-caps heading arrives as "R EFERENCES", a tool
+    name as "SQL ANCER", a hyphenated word as "Time -Series", and a word broken
+    across a line as "re - vealed".
+
+    Repairing them here means every consumer sees the same clean text, and a
+    quotation reads as the author wrote it rather than as the typesetter left
+    it. What is never done is changing words: this only removes spacing the
+    author did not put there.
+    """
+    # A PDF's text layer can decode to unpaired surrogates and other characters
+    # that cannot be encoded back to UTF-8, which breaks hashing and JSON. They
+    # carry no meaning, so they go.
+    text = (text or "").encode("utf-8", "replace").decode("utf-8", "replace")
+    text = text.replace("\ufffd", "")
+    # Control characters survive extraction from some PDFs -- a form feed at a
+    # page break, a vertical tab in a table. They carry no meaning, they are
+    # invisible in a diff, and they make the JSON unreadable to anything
+    # stricter than Python: Jekyll refuses a data file containing them.
+    text = re.sub(r"[\x00-\x08\x0b\x0c\x0e-\x1f\x7f]", " ", text)
+    # Noncharacters and C1 controls come out of a broken text layer -- U+FFFF
+    # in place of a ligature, "o\ufffdcial" for "official". Python and Ruby's
+    # JSON accept them; YAML does not, and Jekyll reads data files as YAML, so
+    # one of these makes the whole site fail to build.
+    text = re.sub(r"[\u0080-\u009f\ufdd0-\ufdef\ufffe\uffff]", "", text)
+    text = re.sub(r"-\s*\n\s*", "", text)          # hyphenated line break
+    text = BROKEN_WORD.sub(r"\1\2", text)
+    text = SPACED_HYPHEN.sub(r"\1-\2", text)
+    text = GLUED_KEYWORD.sub(r"\1 \2", text)
+    text = SPLIT_SMALL_CAPS.sub(_join_small_caps, text)
+    text = repair_small_caps(text)
+    # A name ending in "+" runs into the next word: "SQLANCER+by adapting".
+    text = re.sub(r"(\w\+)([a-z])", r"\1 \2", text)
+    # Extraction leaves a space before punctuation: "RANDOOP , and".
+    text = re.sub(r"\s+([,;:.!?])(\s|$)", r"\1\2", text)
+    # Collapse runs of spaces and tabs, but never newlines: paragraph breaks
+    # are what sentence splitting relies on.
+    text = re.sub(r"[ \t]{2,}", " ", text)
+    return re.sub(r"[ \t]+\n", "\n", text)
+
+
+def page_texts(raw: bytes) -> List[str]:
+    """The text of each page, in order. Empty list if the PDF cannot be read."""
+    if not raw or not raw.lstrip().startswith(b"%PDF") or len(raw) > MAX_PDF_BYTES:
+        return []
+    try:
+        import io
+
+        import pypdf
+
+        reader = pypdf.PdfReader(io.BytesIO(raw))
+        return [(page.extract_text() or "") for page in reader.pages]
+    except Exception:
+        return []
+
+
+def outline_titles(raw: bytes) -> List[str]:
+    """Section titles from the PDF's own outline, if it has one."""
+    try:
+        import io
+
+        import pypdf
+
+        reader = pypdf.PdfReader(io.BytesIO(raw))
+        found: List[str] = []
+
+        def walk(entries):
+            for entry in entries:
+                if isinstance(entry, list):
+                    walk(entry)
+                    continue
+                title = getattr(entry, "title", None)
+                if title:
+                    found.append(str(title).strip())
+
+        walk(reader.outline or [])
+        return found
+    except Exception:
+        return []
+
+
+def _clean_page(text: str) -> str:
+    """One page of extracted text, cleaned of furniture and artefacts."""
+    from ..impact.collectors.fulltext import _strip_page_furniture
+
+    return tidy(_strip_page_furniture(text or ""))
+
+
+def assemble(raw: bytes) -> Dict[str, object]:
+    """The whole document: text, page offsets, sections, references.
+
+    Page offsets are kept because a mention's page number is the cheapest way
+    for a person to find it again in the PDF, and it cannot be recovered later
+    from the concatenated text.
+    """
+    pages = [_clean_page(page) for page in page_texts(raw)]
+    if not pages:
+        return {"text": "", "pages": [], "sections": [], "references": [],
+                "page_count": 0, "has_outline": False}
+
+    text_parts: List[str] = []
+    page_spans: List[Dict[str, int]] = []
+    cursor = 0
+    for index, page in enumerate(pages, start=1):
+        page_spans.append({"page": index, "start": cursor,
+                           "end": cursor + len(page)})
+        text_parts.append(page)
+        cursor += len(page) + 1
+    text = "\n".join(text_parts)
+
+    outline = outline_titles(raw)
+    sections = find_sections(text, outline)
+    references_start, references = parse_references(text)
+    return {
+        "text": text,
+        "pages": page_spans,
+        "page_count": len(pages),
+        "sections": sections,
+        "references": references,
+        "references_start": references_start,
+        "has_outline": bool(outline),
+    }
+
+
+def find_sections(text: str, outline: Optional[List[str]] = None
+                  ) -> List[Dict[str, object]]:
+    """Where each section starts, by offset.
+
+    The PDF outline gives the titles but not reliable offsets, so it is used to
+    confirm a heading rather than to locate one: a line that matches a heading
+    pattern *and* appears in the outline is certainly a heading, and one that
+    matches only the pattern is accepted anyway when there is no outline.
+    """
+    known = {re.sub(r"\s+", " ", title).strip().lower()
+             for title in (outline or [])}
+
+    # Section numbers ascend. That is what separates a heading from the many
+    # other numbered lines in a paper -- a pseudocode step ("5 Avail
+    # s<-selectRandomAvailableFunc"), a figure caption, a table row. Matching
+    # the pattern alone let all of those through.
+    candidates: List[Dict[str, object]] = []
+    offset = 0
+    for line in text.splitlines(keepends=True):
+        stripped = line.strip()
+        for style, pattern in (("roman", ROMAN_HEADING),
+                               ("numbered", NUMBERED_HEADING)):
+            match = pattern.match(stripped)
+            if not match:
+                continue
+            number, title = match.group(1), match.group(2).strip()
+            if CODE_LINE.search(stripped):
+                break      # an algorithm step, not a heading
+            full = re.sub(r"\s+", " ", f"{number} {title}").strip().lower()
+            bare = title.lower()
+            in_outline = bool(known) and (
+                full in known or bare in known
+                or any(entry.endswith(bare) for entry in known))
+            candidates.append({"style": style, "number": number.rstrip("."),
+                               "title": title, "start": offset,
+                               "in_outline": in_outline})
+            break
+        offset += len(line)
+
+    found = _ascending_run(candidates, bool(known))
+
+    for index, section in enumerate(found):
+        section["end"] = (found[index + 1]["start"] if index + 1 < len(found)
+                          else len(text))
+        section["is_back_matter"] = bool(BACK_MATTER.match(str(section["title"])))
+    return found
+
+
+# Characters that mark a line as code or mathematics rather than a heading.
+CODE_LINE = re.compile(r"[←→⇐⇒∈∀∃∧∨≤≥≠∪∩⊆{}();]|:=|<-|\bfor\b|\bwhile\b")
+
+
+def _top_level(number: str) -> Optional[int]:
+    """The section's top-level number, as an integer, if it has one."""
+    head = number.split(".")[0]
+    if head.isdigit():
+        return int(head)
+    romans = {"I": 1, "II": 2, "III": 3, "IV": 4, "V": 5, "VI": 6, "VII": 7,
+              "VIII": 8, "IX": 9, "X": 10, "XI": 11, "XII": 12, "XIII": 13}
+    return romans.get(head.upper())
+
+
+def _ascending_run(candidates: List[Dict[str, object]], has_outline: bool
+                   ) -> List[Dict[str, object]]:
+    """The longest run of candidates whose top-level numbers ascend from one.
+
+    A paper's sections are numbered 1, 2, 3 in order. Anything else that
+    matched -- a pseudocode step, a caption -- does not fit that sequence, and
+    dropping whatever breaks it is what tells them apart. Subsections are kept
+    with whichever top-level section they follow.
+    """
+    kept: List[Dict[str, object]] = []
+    expected = 1
+    style: Optional[str] = None
+    for entry in candidates:
+        top = _top_level(str(entry["number"]))
+        is_sub = "." in str(entry["number"]) or entry["style"] == "roman" and \
+            not str(entry["number"]).upper().strip(".").strip("IVXL") == ""
+        if "." in str(entry["number"]) or (
+                entry["style"] == "roman" and _top_level(str(entry["number"])) is None):
+            # A subsection: keep it if we are inside a section already.
+            if kept:
+                kept.append(entry)
+            continue
+        if has_outline and not entry["in_outline"]:
+            continue
+        if top == expected and style in (None, entry["style"]):
+            # A paper numbers its sections one way throughout. Without this, an
+            # arabic "2" from a figure caption was accepted straight after the
+            # roman "I. Introduction" it followed.
+            style = entry["style"]
+            kept.append(entry)
+            expected += 1
+    return [{k: v for k, v in entry.items() if k not in ("style", "in_outline")}
+            for entry in kept]
+
+
+def _split_entries(tail: str, marker: "re.Pattern") -> List[Dict[str, object]]:
+    """Bibliography entries, split on whichever numbering style is in use."""
+    parts = marker.split(tail)
+    entries: List[Dict[str, object]] = []
+    for index in range(1, len(parts) - 1, 2):
+        try:
+            number = int(parts[index])
+        except ValueError:
+            continue
+        body = re.sub(r"\s+", " ", parts[index + 1]).strip()
+        if body:
+            entries.append({"number": number, "text": body})
+    return entries
+
+
+def parse_references(text: str) -> Tuple[Optional[int], List[Dict[str, object]]]:
+    """The bibliography, split into numbered entries.
+
+    Returns the offset the bibliography starts at, so that everything before it
+    can be treated as the paper's own prose. A citation marker inside the
+    bibliography is another paper's reference number, not this one's.
+    """
+    match = None
+    for candidate in REFERENCES_HEADING.finditer(text):
+        match = candidate  # the last one: a paper may cite the word earlier
+    if match is None:
+        # No heading anywhere. Fall back to the shape of the entries: the last
+        # "[1]" that reads like a bibliography entry begins the bibliography.
+        fallback = None
+        for candidate in BIBLIOGRAPHY_START.finditer(text):
+            fallback = candidate
+        if fallback is None:
+            return None, []
+        start = fallback.start()
+        return start, _renumber(_split_entries(text[start:], REFERENCE_MARKER))
+
+    tail = text[match.end():]
+    entries = _split_entries(tail, REFERENCE_MARKER)
+    if len(entries) < 2:
+        # Few or no bracketed entries: try the dotted style, and keep whichever
+        # reading finds more. Taking the fallback unconditionally lost a
+        # bracketed bibliography that had parsed, just not well.
+        dotted = _split_entries(tail, DOTTED_MARKER)
+        if len(dotted) > len(entries):
+            entries = dotted
+    return match.start(), _renumber(entries)
+
+
+def _renumber(entries: List[Dict[str, object]]) -> List[Dict[str, object]]:
+    """Keep only entries whose numbering runs consecutively from one.
+
+    A bibliography numbers its entries consecutively, so the next entry is
+    always the next number. Anything else is a marker inside the entry being
+    read -- an entry whose own title cites "[3]" does not begin entry three.
+    """
+    cleaned: List[Dict[str, object]] = []
+    expected = 1
+    for entry in entries:
+        if entry["number"] == expected:
+            cleaned.append(entry)
+            expected += 1
+        elif cleaned:
+            cleaned[-1]["text"] += f" [{entry['number']}] {entry['text']}"
+    return cleaned
+
+
+def section_at(sections: List[Dict[str, object]], offset: int) -> Optional[str]:
+    """The title of the section containing ``offset``."""
+    for section in sections:
+        if section["start"] <= offset < section["end"]:
+            number = section.get("number")
+            return f"{number} {section['title']}" if number else str(section["title"])
+    return None
+
+
+def page_at(pages: List[Dict[str, int]], offset: int) -> Optional[int]:
+    for span in pages:
+        if span["start"] <= offset <= span["end"]:
+            return span["page"]
+    return None
diff --git a/tools/papers/mentions.py b/tools/papers/mentions.py
new file mode 100644
index 0000000..dcbd0cc
--- /dev/null
+++ b/tools/papers/mentions.py
@@ -0,0 +1,325 @@
+"""Every place a paper refers to SQLancer, however it refers to it.
+
+Three ways a paper can point at SQLancer, and a search for the name finds only
+the first:
+
+1. **By name** -- "SQLancer", "SQLancer++", "ShQveL".
+2. **By technique** -- PQS, NoREC, TLP, QPG, CERT, DQP, CODDTest, or their
+   expansions. Several are ambiguous as bare acronyms, which the taxonomy
+   already knows about.
+3. **By citation marker alone** -- "Existing works [9]-[11] also face the same
+   problem", where reference 9 is a SQLancer paper. Measured on one IEEE paper,
+   5 of the 14 sentences citing a SQLancer publication never name it. Nothing
+   that matches on names can see them.
+
+Each mention keeps enough context to be re-judged later without reopening the
+PDF: the sentence, the paragraph around it, the section, the page, and the
+offset. That is the point of the subproject -- the reading happens once.
+
+Nothing here decides what the paper's relationship to SQLancer is. It finds the
+places worth reading and hands them to the analysis.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Dict, List, Optional, Sequence
+
+from ..impact import taxonomy
+from . import extract
+
+INVENTORY_VERSION = "paper-mentions-v1"
+
+# How much of the surrounding text to keep with each mention. Generous on
+# purpose: a later question about the mention should be answerable from the
+# stored context rather than from the PDF.
+CONTEXT_BEFORE = 600
+CONTEXT_AFTER = 600
+
+# A citation marker, and the ranges publishers write them in. "[9]-[11]" and
+# "[9]–[11]" both mean three references, and the middle one is invisible unless
+# the range is expanded.
+# Publishers space these differently -- "[11]", "[ 1,2]", and groups that wrap
+# across a line ("[ 9,10,11,12,13,\n14,15,16]") -- so whitespace is allowed
+# anywhere inside the bracket. Requiring a digit straight after "[" silently
+# found nothing at all in papers that space them out.
+CITATION_GROUP = re.compile(
+    r"\[\s*(\d{1,3}(?:\s*[,–—-]\s*\d{1,3})*)\s*\]")
+CITATION_RANGE = re.compile(r"(\d{1,3})\s*[–—-]\s*(\d{1,3})")
+
+# Authors and titles that mark a bibliography entry as a SQLancer publication.
+SQLANCER_REFERENCE = re.compile(
+    r"sqlancer|pivoted\s+query\s+synthesis|non-?optimizing\s+reference\s+engine|"
+    r"ternary\s+logic\s+partitioning|query\s+plan\s+guidance|"
+    r"finding\s+bugs\s+in\s+database\s+systems\s+via\s+query\s+partitioning|"
+    r"constant[- ]optimization[- ]driven|cardinality\s+estimation\s+restriction|"
+    r"differential\s+query\s+plans",
+    re.IGNORECASE)
+
+RIGGER = re.compile(r"\bM(?:anuel)?\.?\s+Rigger\b|\bRigger\b", re.IGNORECASE)
+
+# An author-year citation, which is what ACM's current format uses instead of
+# numbers: "databases [Rigger and Su 2020; Wang et al. 2021b]". A paper in that
+# style has no numbered markers at all, so the whole citation-marker route finds
+# nothing in it and the tool's name may never appear in the prose either.
+AUTHOR_YEAR = re.compile(
+    r"\bRigger\b[^\]\).;]{0,40}?\b(19|20)\d{2}[a-z]?\b", re.IGNORECASE)
+
+
+def expand_markers(group: str) -> List[int]:
+    """The reference numbers a citation group names, ranges included."""
+    numbers: List[int] = []
+    for part in re.split(r"\s*,\s*", group):
+        part = part.strip()
+        if not part:
+            continue
+        span = CITATION_RANGE.fullmatch(part)
+        if span:
+            first, last = int(span.group(1)), int(span.group(2))
+            if 0 < last - first < 40:
+                numbers.extend(range(first, last + 1))
+                continue
+        if part.isdigit():
+            numbers.append(int(part))
+    return numbers
+
+
+# A range written as two brackets rather than one: "[9]-[11]", which is how
+# IEEE sets it. Everything between the endpoints is cited and none of it is
+# visible unless the pair is read together.
+BRACKET_RANGE = re.compile(
+    r"\[\s*(\d{1,3})\s*\]\s*[–—-]\s*\[\s*(\d{1,3})\s*\]")
+
+
+def _citation_spans(text: str) -> List[dict]:
+    """Every citation in the text, with the reference numbers it names."""
+    spans: List[dict] = []
+    consumed: List[tuple] = []
+    for match in BRACKET_RANGE.finditer(text):
+        first, last = int(match.group(1)), int(match.group(2))
+        if 0 < last - first < 40:
+            spans.append({"start": match.start(), "text": match.group(0),
+                          "numbers": list(range(first, last + 1))})
+            consumed.append((match.start(), match.end()))
+    for match in CITATION_GROUP.finditer(text):
+        if any(start <= match.start() < end for start, end in consumed):
+            continue
+        spans.append({"start": match.start(), "text": match.group(0),
+                      "numbers": expand_markers(match.group(1))})
+    return spans
+
+
+def sqlancer_references(references: Sequence[dict],
+                        tax: Optional[taxonomy.Taxonomy] = None
+                        ) -> Dict[int, dict]:
+    """Which of this paper's bibliography entries are SQLancer publications.
+
+    Matched on the tool and technique names, on the title of each origin paper
+    the taxonomy knows, and on authorship -- a Rigger paper in a database
+    testing bibliography is one of these often enough to be worth catching, and
+    the entry is recorded with what matched so a wrong call is visible.
+    """
+    tax = tax or taxonomy.load()
+    titles = []
+    identifiers = []
+    for seed in tax.citation_seeds():
+        label = seed.get("seed_id") or seed.get("technique_id")
+        for title in [seed.get("title")] + list(seed.get("also_titled") or []):
+            if title:
+                titles.append((label, seed.get("technique_id"), title.lower()))
+        # A DOI or arXiv id is what a bibliography prints for a paper whose
+        # title it has rewritten. CERT is cited as "CERT: Finding performance
+        # issues ... through the lens of cardinality estimation", which shares
+        # no opening words with the preprint title the taxonomy carries, and
+        # was recognised only as a paper by one of the project's authors.
+        for key in ("doi", "arxiv_id"):
+            value = seed.get(key)
+            if value:
+                identifiers.append((label, seed.get("technique_id"), key,
+                                    re.sub(r"\s+", "", str(value).lower())))
+
+    found: Dict[int, dict] = {}
+    for entry in references:
+        text = str(entry.get("text") or "")
+        lowered = text.lower()
+        # Extraction glues words together in some bibliographies --
+        # "ManuelRiggerandZhendongSu.2020. FindingBugsinDatabaseSystemsvia
+        # QueryPartitioning" is one real entry -- so every match is also tried
+        # with all whitespace removed from both sides. Without this the TLP
+        # paper went unrecognised in its own citing paper's bibliography.
+        squashed = re.sub(r"\s+", "", lowered)
+        reasons: List[str] = []
+        technique = None
+        for seed_label, technique_id, title in titles:
+            head = " ".join(title.split()[:6])
+            if head and (head in lowered
+                         or re.sub(r"\s+", "", head) in squashed):
+                reasons.append(f"cites the paper introducing {seed_label}")
+                technique = technique or technique_id
+        for seed_label, technique_id, key, value in identifiers:
+            if value in squashed:
+                kind = "DOI" if key == "doi" else "arXiv id"
+                reasons.append(f"prints the {kind} of the paper introducing "
+                               f"{seed_label}")
+                technique = technique or technique_id
+        if SQLANCER_REFERENCE.search(text) or SQLANCER_REFERENCE.search(squashed):
+            reasons.append("names SQLancer or one of its techniques")
+        matched_as = "sqlancer_publication" if reasons else None
+        if not reasons and (RIGGER.search(text) or "rigger" in squashed):
+            # A paper by the project's authors, which is not the same thing as a
+            # SQLancer paper: FlowFusion is a PHP fuzzer Rigger co-authored, and
+            # calling its reference a SQLancer publication would be wrong. It is
+            # kept because a paper citing it is worth reading, and labelled so
+            # the analysis can weigh it accordingly.
+            reasons.append("written by an author of the SQLancer project, but "
+                           "not itself a SQLancer paper")
+            matched_as = "project_authored"
+        if reasons:
+            found[int(entry["number"])] = {
+                "number": int(entry["number"]),
+                "text": text,
+                "technique": technique,
+                "matched_as": matched_as,
+                "why": reasons,
+            }
+    return found
+
+
+# A sentence ends at a terminator or a blank line, not at every line break:
+# extraction puts a newline at the end of every typeset line, so stopping there
+# cut "For example, SQLancer [11] relies on" off mid-clause.
+SENTENCE_BREAK = re.compile(r"[.!?]|\n\s*\n")
+
+
+def _sentence_bounds(text: str, index: int) -> tuple:
+    start = index
+    while start > 0:
+        window = text[max(0, start - 2):start]
+        if text[start - 1] in ".!?" or "\n\n" in window:
+            break
+        start -= 1
+    end = index
+    while end < len(text):
+        if text[end] in ".!?" or text.startswith("\n\n", end):
+            break
+        end += 1
+    return start, min(end + 1, len(text))
+
+
+def spaced_pattern(name: str) -> str:
+    """A pattern matching ``name`` however extraction spaced it out.
+
+    Small caps come out of a PDF with spaces inside the word -- "SQL ANCER" for
+    SQLANCER, "N OREC" for NOREC -- so a plain name search misses exactly the
+    places a paper is most likely to name the tool, its own prose. The text is
+    not rewritten to fix this: what is stored stays as the PDF renders it, and
+    the pattern is what bends.
+    """
+    return r"\s?".join(re.escape(char) for char in name if not char.isspace())
+
+
+def _mention(document: dict, index: int, surface: str, found_by: str,
+             technique: Optional[str], number: int,
+             reference: Optional[dict] = None) -> dict:
+    text = document["text"]
+    start, end = _sentence_bounds(text, index)
+    sentence = re.sub(r"\s+", " ", text[start:end]).strip()
+    before = re.sub(r"\s+", " ", text[max(0, start - CONTEXT_BEFORE):start]).strip()
+    after = re.sub(r"\s+", " ", text[end:end + CONTEXT_AFTER]).strip()
+    entry = {
+        "id": f"M{number}",
+        "found_by": found_by,
+        "surface": surface,
+        "technique": technique,
+        "sentence": sentence,
+        "context_before": before,
+        "context_after": after,
+        "section": extract.section_at(document.get("sections") or [], start),
+        "page": extract.page_at(document.get("pages") or [], start),
+        "char_offset": start,
+    }
+    if reference is not None:
+        entry["cited_reference"] = {
+            "number": reference["number"],
+            "text": reference["text"][:300],
+            "matched_as": reference.get("matched_as"),
+            "why": reference["why"],
+        }
+    return entry
+
+
+def build(document: dict, tax: Optional[taxonomy.Taxonomy] = None) -> dict:
+    """The inventory: every mention, and the references behind the markers."""
+    tax = tax or taxonomy.load()
+    text = document.get("text") or ""
+    body_end = document.get("references_start") or len(text)
+
+    references = sqlancer_references(document.get("references") or [], tax)
+    seen: Dict[int, dict] = {}
+    counter = 0
+
+    def add(index: int, surface: str, found_by: str,
+            technique: Optional[str], reference: Optional[dict] = None) -> None:
+        nonlocal counter
+        start, _ = _sentence_bounds(text, index)
+        if start in seen:
+            # One sentence, one mention: record the strongest way it was found,
+            # but keep every technique it names.
+            existing = seen[start]
+            if technique and technique not in (existing.get("techniques") or []):
+                existing.setdefault("techniques", []).append(technique)
+            if found_by not in existing["found_by_all"]:
+                existing["found_by_all"].append(found_by)
+            return
+        counter += 1
+        entry = _mention(document, index, surface, found_by, technique,
+                         counter, reference)
+        entry["found_by_all"] = [found_by]
+        entry["techniques"] = [technique] if technique else []
+        seen[start] = entry
+
+    # 1. By name, tolerating the spacing small caps leave behind.
+    names = "|".join(spaced_pattern(name)
+                     for name in ("SQLancer++", "SQLancer", "ShQveL"))
+    for match in re.finditer(rf"(?<![\w+])({names})(?![\w])",
+                             text[:body_end], re.IGNORECASE):
+        add(match.start(), match.group(0), "name", None)
+
+    # 2. By technique name.
+    for match in tax.find_techniques(text[:body_end], require_context=False):
+        add(match.start, match.matched_name, "technique", match.technique_id)
+
+    # 3. By author-year citation, for bibliographies with no numbers to resolve.
+    for match in AUTHOR_YEAR.finditer(text[:body_end]):
+        add(match.start(), match.group(0), "author_year_citation", None)
+
+    # 4. By citation marker resolving to a SQLancer publication.
+    if references:
+        for match in _citation_spans(text[:body_end]):
+            cited = [references[number] for number in match["numbers"]
+                     if number in references]
+            if not cited:
+                continue
+            # One group can cite both a SQLancer paper and a paper merely
+            # written by one of its authors -- "[2-5,8,13,14,17]" cites both.
+            # The stronger match decides how the sentence is labelled, so a
+            # sentence citing PQS is not filed under project_authored because
+            # a neighbouring number happened to resolve first.
+            reference = next(
+                (item for item in cited
+                 if item.get("matched_as") == "sqlancer_publication"), cited[0])
+            found_by = ("citation_marker"
+                        if reference.get("matched_as") == "sqlancer_publication"
+                        else "citation_marker_project_authored")
+            add(match["start"], match["text"], found_by,
+                reference.get("technique"), reference)
+
+    mentions = sorted(seen.values(), key=lambda entry: entry["char_offset"])
+    for position, entry in enumerate(mentions, start=1):
+        entry["id"] = f"M{position}"
+    return {
+        "inventory_version": INVENTORY_VERSION,
+        "mentions": mentions,
+        "sqlancer_references": sorted(references.values(),
+                                      key=lambda r: r["number"]),
+    }
diff --git a/tools/papers/pages.py b/tools/papers/pages.py
new file mode 100644
index 0000000..4cb54e4
--- /dev/null
+++ b/tools/papers/pages.py
@@ -0,0 +1,126 @@
+"""Generates one page per paper under ``/impact/papers/``.
+
+The impact page lists a paper with a sentence on what it did with SQLancer, and
+that is all it should carry: the lists run to two hundred entries, and a reader
+scanning them wants the connection, not the paper. Everything else the record
+holds -- the classifications with their reasoning, the sentences each rests on,
+the resolved references, the artifact -- has nowhere to go on that page and is
+the reason the record exists.
+
+So each paper gets a page. As with the database systems, the generated files
+are front matter and one include: the rendering lives in
+``_includes/impact/paper-detail.html`` and every figure comes from the record,
+so presentation is one edit rather than a hundred and eighty.
+"""
+
+from __future__ import annotations
+
+import json
+import pathlib
+from typing import Dict, List, Optional
+
+from tools.impact import config
+
+PAGE_DIR = config.REPO_ROOT / "_pages" / "impact" / "papers"
+RECORD_DIR = config.REPO_ROOT / "_data" / "papers"
+PERMALINK_BASE = "/impact/papers/"
+
+TEMPLATE = """---
+permalink: {permalink}
+title: "{title}"
+excerpt: "{excerpt}"
+paper_key: {key}
+author_profile: false
+sitemap: true
+---
+
+{{% include impact/paper-detail.html key=page.paper_key %}}
+"""
+
+
+def _escape(text: str) -> str:
+    """Quote a value for a double-quoted YAML scalar."""
+    return (str(text).replace("\\", "\\\\").replace('"', '\\"')
+            .replace("\n", " ").strip())
+
+
+def records() -> List[dict]:
+    """Every stored paper record, with the key its page is named for."""
+    found = []
+    for path in sorted(RECORD_DIR.glob("*.json")):
+        try:
+            record = json.loads(path.read_text(encoding="utf-8"))
+        except ValueError:
+            continue
+        record["_key"] = path.stem
+        found.append(record)
+    return found
+
+
+def excerpt_for(record: dict) -> str:
+    """One sentence for the page's own metadata.
+
+    The relationship narrative if the paper has been analysed, because that is
+    what distinguishes this page from the paper's own abstract; the title
+    otherwise.
+    """
+    analysis = record.get("analysis") or {}
+    narrative = (analysis.get("narrative") or "").strip()
+    if narrative:
+        sentence = narrative.split(". ")[0].rstrip(".")
+        return sentence[:280] + "."
+    title = (record.get("paper") or {}).get("title") or "A paper citing SQLancer"
+    return f"What {title} records about SQLancer."
+
+
+def render(record: dict) -> str:
+    paper = record.get("paper") or {}
+    title = paper.get("title") or record["_key"]
+    return TEMPLATE.format(
+        permalink=f"{PERMALINK_BASE}{record['_key']}/",
+        title=_escape(title),
+        excerpt=_escape(excerpt_for(record)),
+        key=record["_key"])
+
+
+def write_all(directory: Optional[pathlib.Path] = None,
+              found: Optional[List[dict]] = None) -> Dict[str, str]:
+    """Write a page per paper record; returns what happened to each."""
+    directory = directory or PAGE_DIR
+    directory.mkdir(parents=True, exist_ok=True)
+    found = records() if found is None else found
+
+    outcome: Dict[str, str] = {}
+    keep = set()
+    for record in found:
+        path = directory / f"{record['_key']}.html"
+        keep.add(path.name)
+        markup = render(record)
+        existing = path.read_text(encoding="utf-8") if path.exists() else None
+        if existing == markup:
+            outcome[record["_key"]] = "unchanged"
+            continue
+        path.write_text(markup, encoding="utf-8")
+        outcome[record["_key"]] = "written" if existing is None else "updated"
+    # A record that leaves the dataset must not leave a page behind claiming
+    # figures nothing produces any more.
+    for path in directory.glob("*.html"):
+        if path.name not in keep:
+            path.unlink()
+            outcome[path.stem] = "removed"
+    return outcome
+
+
+def main() -> int:
+    outcome = write_all()
+    counts: Dict[str, int] = {}
+    for state in outcome.values():
+        counts[state] = counts.get(state, 0) + 1
+    summary = ", ".join(f"{count} {state}"
+                        for state, count in sorted(counts.items()))
+    print(f"paper pages: {len(outcome)} papers ({summary})")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/tools/papers/prompt.py b/tools/papers/prompt.py
new file mode 100644
index 0000000..0068bf7
--- /dev/null
+++ b/tools/papers/prompt.py
@@ -0,0 +1,255 @@
+"""The analysis task, written once and used by both runners.
+
+The design constraint that shapes everything here: **the model never supplies a
+quotation.** It is given a numbered inventory of mentions, each already carrying
+a sentence taken verbatim from the paper, and it answers by citing those
+numbers. The writer substitutes the stored text. A fabricated quotation is not
+detected and discarded, as the old classifier had to do -- it is not expressible.
+
+The criteria below are not general good sense. Each one is written the way it is
+because the previous approach got that exact case wrong:
+
+* "we adapted SQLancer as a baseline for comparison" was recorded as *extending*
+  a technique, because a pattern matched the first half of the sentence.
+* Chronos uses SQLancer only to generate a workload for MySQL-Cluster, and the
+  dataset could say it reused the infrastructure but not in what sense.
+* A paper citing "[9]-[11]" says nothing a name search can see, so the fact that
+  reference 9 is the PQS paper has to be given to the reader explicitly.
+"""
+
+from __future__ import annotations
+
+import json
+from typing import Dict, List, Optional, Sequence
+
+PROMPT_VERSION = "paper-analysis-v1"
+
+RELATIONSHIPS = ("uses_infrastructure", "extends_technique", "compares_with",
+                 "describes_as_state_of_the_art")
+
+ROLES = (
+    "background", "motivation", "definition", "reuse_implementation",
+    "reuse_component", "extension", "baseline", "result_comparison",
+    "state_of_the_art", "incidental",
+)
+
+ANSWERS = ("yes", "no", "uncertain", "insufficient_evidence")
+
+REUSE_KINDS = ("implementation", "generator", "workload", "unclear")
+
+SYSTEM = """\
+You are recording how one research paper relates to SQLancer, a testing tool for
+database management systems, and to the techniques introduced through it: PQS
+(Pivoted Query Synthesis), NoREC (Non-optimizing Reference Engine Construction),
+TLP (Ternary Logic Partitioning), QPG (Query Plan Guidance), CERT, DQP
+(Differential Query Plans) and CODDTest.
+
+You are given the paper's metadata and a numbered inventory of every place the
+paper refers to SQLancer -- by name, by technique, or through a citation marker
+resolving to one of its publications. Each mention carries the sentence it
+appears in, the paragraph around it, its section and its page.
+
+You do not have the paper. You have these mentions, and they are the evidence.
+
+## Steps, in this order
+
+1. Read the metadata and the entire inventory before deciding anything. A paper
+   that reuses SQLancer usually says so once, in one sentence, somewhere in the
+   implementation or evaluation section.
+2. Give every mention a role from this list: background, motivation, definition,
+   reuse_implementation, reuse_component, extension, baseline,
+   result_comparison, state_of_the_art, incidental.
+3. Decide each relationship using the criteria below, citing the mention ids
+   that settle it. A mention cited for a relationship must have a role
+   consistent with it.
+4. Write two pieces of prose in your own words: a summary of what the paper is
+   and does, and a narrative of how SQLancer figures in it.
+5. Compare your decisions with the advisory regex findings you were given.
+   Explain any disagreement, in either direction.
+6. Say what you could not determine, and why.
+
+## Criteria
+
+**uses_infrastructure** -- yes only if the paper's *own implementation* reuses
+the SQLancer codebase. "We implemented X on top of SQLancer", "our framework is
+derived from SQLancer", "our database generation is adopted from SQLancer" all
+qualify. Also set reuse_kind:
+  - implementation: the paper's tool is built on SQLancer
+  - generator: it uses SQLancer's query or database generation inside its own tool
+  - workload: it runs SQLancer only to produce input for something else, and is
+    not built on it in any other sense
+  - unclear: reuse is stated but its extent is not
+Workload use is still reuse, but the narrative must say which kind it is.
+
+**extends_technique** -- yes only if extending, generalising or adapting a
+SQLancer technique is part of what the paper claims as its contribution.
+Carrying TLP to graph queries is extension. Adapting the tool in order to *run
+it as a baseline* is not: that is compares_with, and nothing else. If a sentence
+contains both "adapted" and "as a baseline", it is a comparison.
+
+**compares_with** -- yes if the paper empirically runs SQLancer or one of its
+techniques and reports the comparison, including reporting that it found more
+bugs or more coverage. Appearing in a list of related work is not a comparison.
+
+**describes_as_state_of_the_art** -- yes if the text calls SQLancer or one of
+its techniques the state of the art, leading, or the most effective approach.
+An author's own tool being better is not this; being described as the bar to
+beat is.
+
+## Rules
+
+- Answer each relationship with yes, no, uncertain, or insufficient_evidence.
+  Prefer uncertain to a guess. "insufficient_evidence" means the mentions do not
+  address the question at all.
+- Every "yes" must cite at least one mention id. A yes with no ids is rejected.
+- Cite ids; never write a quotation. The text of a mention is already recorded
+  and will be attached to whatever you cite.
+- Judge only what the mentions show. If a paper's artifact was inspected you are
+  told so separately; do not infer reuse from a paper's subject matter, its
+  authors, or its venue.
+"""
+
+
+def _mention_block(mention: dict) -> str:
+    parts = [f"{mention['id']}  [{mention.get('section') or 'unknown section'}"
+             f", page {mention.get('page') or '?'}]"]
+    if mention.get("cited_reference"):
+        reference = mention["cited_reference"]
+        parts.append(f"    cites [{reference['number']}] = {reference['text'][:150]}")
+    if mention.get("techniques"):
+        parts.append(f"    techniques named: {', '.join(mention['techniques'])}")
+    parts.append(f"    SENTENCE: {mention['sentence']}")
+    if mention.get("context_before"):
+        parts.append(f"    before: ...{mention['context_before'][-260:]}")
+    if mention.get("context_after"):
+        parts.append(f"    after: {mention['context_after'][:260]}...")
+    return "\n".join(parts)
+
+
+def build(paper: dict, inventory: dict, findings: dict,
+          artifact: Optional[dict] = None) -> Dict[str, str]:
+    """The system and user messages for one paper."""
+    metadata = {
+        "title": paper.get("title"),
+        "authors": (paper.get("authors") or [])[:10],
+        "year": paper.get("year"),
+        "venue": paper.get("venue"),
+        "doi": paper.get("doi"),
+        "abstract": (paper.get("abstract") or "")[:2000] or None,
+    }
+
+    lines = [f"PAPER: {json.dumps(metadata, ensure_ascii=False, indent=1)}", ""]
+
+    references = inventory.get("sqlancer_references") or []
+    if references:
+        lines.append("REFERENCES IN THIS PAPER'S BIBLIOGRAPHY THAT MATTER HERE")
+        for reference in references:
+            kind = ("a SQLancer publication"
+                    if reference.get("matched_as") == "sqlancer_publication"
+                    else "by a SQLancer author, but NOT a SQLancer paper")
+            lines.append(f"  [{reference['number']}] ({kind}) "
+                         f"{reference['text'][:180]}")
+        lines.append("")
+        lines.append("A mention found only through a reference of the second "
+                     "kind is not evidence about SQLancer. Say so if that is "
+                     "all there is.")
+        lines.append("")
+
+    if artifact and artifact.get("markers"):
+        lines.append("ARTIFACT INSPECTION (evidence from the paper's repository, "
+                     "not from its text). Cite it as ARTIFACT, like a mention.")
+        lines.append(f"  ARTIFACT  repository: {artifact.get('url')}")
+        lines.append(f"            markers found: {', '.join(artifact['markers'])}")
+        lines.append("  A paper whose repository is a SQLancer fork need never "
+                     "say so in its text, so this can be the only evidence of "
+                     "reuse. It says nothing about the other relationships.")
+        lines.append("")
+
+    mentions = inventory.get("mentions") or []
+    lines.append(f"MENTIONS ({len(mentions)})")
+    lines.append("")
+    for mention in mentions:
+        lines.append(_mention_block(mention))
+        lines.append("")
+
+    suggests = (findings or {}).get("suggests") or {}
+    suppressed = (findings or {}).get("suppressed") or []
+    lines.append("ADVISORY REGEX FINDINGS (a second opinion, not a verdict)")
+    if suggests:
+        for key, ids in suggests.items():
+            lines.append(f"  a pattern for {key} matched: {', '.join(ids)}")
+    else:
+        lines.append("  no pattern matched any mention")
+    for note in suppressed:
+        lines.append(f"  {note['mention_id']}: a {note['pattern']} pattern "
+                     f"matched but was set aside because "
+                     f"{note['suppressed_because']}")
+    lines.append("")
+    lines.append("Now carry out the six steps and answer with the JSON object "
+                 "described in the output schema.")
+    return {"system": SYSTEM, "user": "\n".join(lines)}
+
+
+def output_schema() -> dict:
+    """The shape of the answer, for a structured-output request."""
+    return {
+        "type": "json_schema",
+        "schema": {
+            "type": "object",
+            "additionalProperties": False,
+            "required": ["summary", "narrative", "roles", "relationships",
+                         "disagreements", "unresolved"],
+            "properties": {
+                "summary": {
+                    "type": "string",
+                    "description": "Three to five sentences: what the paper is "
+                                   "and what it does. Your own words.",
+                },
+                "narrative": {
+                    "type": "string",
+                    "description": "One to three sentences on how SQLancer "
+                                   "figures in this paper. Your own words.",
+                },
+                "roles": {
+                    "type": "object",
+                    "description": "Mention id to role, for every mention.",
+                    "additionalProperties": {"type": "string", "enum": list(ROLES)},
+                },
+                "relationships": {
+                    "type": "object",
+                    "additionalProperties": False,
+                    "required": list(RELATIONSHIPS),
+                    "properties": {
+                        key: {
+                            "type": "object",
+                            "additionalProperties": False,
+                            "required": ["value", "mention_ids", "reasoning"],
+                            "properties": {
+                                "value": {"type": "string", "enum": list(ANSWERS)},
+                                "mention_ids": {"type": "array",
+                                                "items": {"type": "string"}},
+                                "reasoning": {"type": "string"},
+                                **({"reuse_kind": {"type": "string",
+                                                   "enum": list(REUSE_KINDS)}}
+                                   if key == "uses_infrastructure" else {}),
+                                "techniques": {"type": "array",
+                                               "items": {"type": "string"}},
+                            },
+                        }
+                        for key in RELATIONSHIPS
+                    },
+                },
+                "disagreements": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "Where you differ from the advisory findings, "
+                                   "and why.",
+                },
+                "unresolved": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "What the mentions could not settle.",
+                },
+            },
+        },
+    }
diff --git a/tools/papers/store.py b/tools/papers/store.py
new file mode 100644
index 0000000..d553322
--- /dev/null
+++ b/tools/papers/store.py
@@ -0,0 +1,144 @@
+"""Reads and writes the per-paper files in ``_data/papers``.
+
+One file per paper, holding everything learned about it: what was read, the
+document's shape, every mention with its context, the artifact, the advisory
+checks, and -- once it has been done -- the analysis.
+
+Extraction and analysis carry separate version stamps and are written by
+separate steps. Re-running the prompt must never redo the PDF work, and
+improving the extractor must never silently discard an analysis: that
+separation is the reason the expensive half only happens once.
+"""
+
+from __future__ import annotations
+
+import json
+import pathlib
+from typing import Dict, List, Optional
+
+from ..impact import config
+from ..impact.util import content_hash, now, slugify
+
+SCHEMA_VERSION = "1.0.0"
+DIRECTORY = config.DATA_DIR.parent / "papers"
+
+
+def path_for(paper_id: str) -> pathlib.Path:
+    """Where a paper's file lives. Named for the id, which outlives the title."""
+    return DIRECTORY / f"{slugify(paper_id)}.json"
+
+
+def load(paper_id: str) -> Optional[dict]:
+    path = path_for(paper_id)
+    if not path.exists():
+        return None
+    try:
+        with open(path, encoding="utf-8") as handle:
+            return json.load(handle)
+    except (OSError, ValueError):
+        return None
+
+
+def save(record: dict) -> bool:
+    """Write a record. Returns whether anything actually changed.
+
+    Timestamps are excluded from the comparison, so a re-run that learns nothing
+    new leaves the file -- and the diff a reviewer reads -- untouched.
+    """
+    path = path_for(record["paper"]["id"])
+    path.parent.mkdir(parents=True, exist_ok=True)
+    body = json.dumps(record, indent=2, ensure_ascii=False, sort_keys=False)
+    if path.exists():
+        try:
+            existing = json.loads(path.read_text(encoding="utf-8"))
+        except ValueError:
+            existing = None
+        if existing is not None and _comparable(existing) == _comparable(record):
+            return False
+    path.write_text(body + "\n", encoding="utf-8")
+    return True
+
+
+def _comparable(record: dict) -> str:
+    copy = json.loads(json.dumps(record))
+    copy.pop("generated_at", None)
+    for source in copy.get("sources") or []:
+        source.pop("retrieved_at", None)
+    if copy.get("analysis"):
+        copy["analysis"].pop("generated_at", None)
+    return json.dumps(copy, sort_keys=True)
+
+
+def build(paper: dict, document: dict, inventory: dict, findings: dict, *,
+          route: str, source_url: Optional[str] = None,
+          artifact: Optional[dict] = None,
+          analysis: Optional[dict] = None) -> dict:
+    """Assemble the record. Everything learned, in one self-describing file."""
+    from . import extract, mentions as mentions_module
+
+    text = document.get("text") or ""
+    return {
+        "schema_version": SCHEMA_VERSION,
+        "generated_at": now(),
+        "paper": {
+            "id": paper["id"],
+            "title": paper.get("title"),
+            "authors": paper.get("authors") or [],
+            "year": paper.get("year"),
+            "venue": paper.get("venue"),
+            "doi": paper.get("doi"),
+            "arxiv_id": paper.get("arxiv_id"),
+            "s2_paper_id": paper.get("s2_paper_id"),
+            "url": paper.get("url"),
+            "also_indexed_as": paper.get("also_indexed_as") or [],
+        },
+        "sources": [{
+            "kind": "fulltext" if route != "metadata" else "metadata",
+            "route": route,
+            "url": source_url,
+            "retrieved_at": now(),
+            "chars": len(text),
+            "content_sha256": content_hash(text) if text else None,
+        }],
+        "document": {
+            "has_fulltext": route != "metadata",
+            "page_count": document.get("page_count", 0),
+            "has_outline": bool(document.get("has_outline")),
+            "sections": [
+                {"number": s.get("number"), "title": s.get("title"),
+                 "start": s.get("start")}
+                for s in (document.get("sections") or [])
+            ],
+        },
+        "references": [
+            {"number": r["number"], "text": r["text"][:400],
+             "is_sqlancer_publication": r["number"] in {
+                 s["number"] for s in inventory.get("sqlancer_references") or []},
+             }
+            for r in (document.get("references") or [])
+        ],
+        "sqlancer_references": inventory.get("sqlancer_references") or [],
+        "mentions": inventory.get("mentions") or [],
+        "artifact": artifact,
+        "checks": findings,
+        "analysis": analysis,
+        "provenance": {
+            "extractor_version": extract.EXTRACTOR_VERSION,
+            "inventory_version": mentions_module.INVENTORY_VERSION,
+            "checks_version": findings.get("checks_version"),
+            "policy_version": config.policy_version(),
+        },
+    }
+
+
+def all_records() -> List[dict]:
+    if not DIRECTORY.is_dir():
+        return []
+    out = []
+    for path in sorted(DIRECTORY.glob("*.json")):
+        try:
+            with open(path, encoding="utf-8") as handle:
+                out.append(json.load(handle))
+        except (OSError, ValueError):
+            continue
+    return out
diff --git a/tools/papers/tests/__init__.py b/tools/papers/tests/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/tools/papers/tests/test_analysis.py b/tools/papers/tests/test_analysis.py
new file mode 100644
index 0000000..332e1d9
--- /dev/null
+++ b/tools/papers/tests/test_analysis.py
@@ -0,0 +1,142 @@
+"""The analysis is the one part written by a model, so it is the part checked."""
+
+import pathlib
+import tempfile
+import unittest
+
+from tools.papers import analysis, store
+
+
+class ApplyTest(unittest.TestCase):
+    def setUp(self):
+        self._dir = tempfile.TemporaryDirectory()
+        self._real = store.DIRECTORY
+        store.DIRECTORY = pathlib.Path(self._dir.name)
+        store.save({
+            "schema_version": "1.0.0",
+            "paper": {"id": "paper:doi:10.1145/x", "title": "A Paper"},
+            "document": {"has_fulltext": True, "page_count": 1,
+                         "has_outline": False, "sections": []},
+            "mentions": [
+                {"id": "M1", "sentence": "We implemented our tool on top of "
+                 "SQLancer.", "section": "4 Implementation", "page": 4,
+                 "found_by_all": ["name"], "char_offset": 10},
+                {"id": "M2", "sentence": "We compare against NoREC.",
+                 "section": "5 Evaluation", "page": 6,
+                 "found_by_all": ["technique"], "char_offset": 90},
+            ],
+            "references": [], "sqlancer_references": [], "checks": {},
+            "artifact": None, "analysis": None, "sources": [], "provenance": {},
+        })
+
+    def tearDown(self):
+        store.DIRECTORY = self._real
+        self._dir.cleanup()
+
+    def _answer(self, **overrides):
+        answer = {
+            "summary": "A paper about testing.",
+            "narrative": "It builds on SQLancer.",
+            "roles": {"M1": "reuse_implementation", "M2": "baseline"},
+            "relationships": {
+                "uses_infrastructure": {"value": "yes", "mention_ids": ["M1"],
+                                        "reasoning": "It says so."},
+                "extends_technique": {"value": "no", "mention_ids": [],
+                                      "reasoning": ""},
+                "compares_with": {"value": "yes", "mention_ids": ["M2"],
+                                  "reasoning": "It says so."},
+                "describes_as_state_of_the_art": {"value": "insufficient_evidence",
+                                                  "mention_ids": [],
+                                                  "reasoning": ""},
+            },
+            "disagreements": [], "unresolved": [],
+        }
+        answer.update(overrides)
+        return answer
+
+    def test_a_good_answer_is_stored_with_quotes_from_the_inventory(self):
+        record = analysis.apply("paper:doi:10.1145/x", self._answer(),
+                                model="claude-opus-5")
+        uses = record["analysis"]["relationships"]["uses_infrastructure"]
+        self.assertEqual(["M1"], uses["mention_ids"])
+        self.assertEqual("We implemented our tool on top of SQLancer.",
+                         uses["quotes"][0]["sentence"])
+        self.assertTrue(record["analysis"]["is_model_written"])
+
+    def test_a_claim_citing_a_mention_that_does_not_exist_is_rejected(self):
+        answer = self._answer()
+        answer["relationships"]["uses_infrastructure"]["mention_ids"] = ["M9"]
+        with self.assertRaises(analysis.Rejected) as caught:
+            analysis.apply("paper:doi:10.1145/x", answer, model="m")
+        self.assertIn("M9", str(caught.exception))
+
+    def test_a_yes_with_nothing_cited_is_rejected(self):
+        answer = self._answer()
+        answer["relationships"]["compares_with"]["mention_ids"] = []
+        with self.assertRaises(analysis.Rejected):
+            analysis.apply("paper:doi:10.1145/x", answer, model="m")
+
+    def test_an_unknown_role_is_rejected(self):
+        answer = self._answer(roles={"M1": "invented_role"})
+        with self.assertRaises(analysis.Rejected):
+            analysis.apply("paper:doi:10.1145/x", answer, model="m")
+
+    def test_a_role_for_a_mention_that_does_not_exist_is_rejected(self):
+        answer = self._answer(roles={"M7": "baseline"})
+        with self.assertRaises(analysis.Rejected):
+            analysis.apply("paper:doi:10.1145/x", answer, model="m")
+
+    def test_a_missing_relationship_is_rejected(self):
+        answer = self._answer()
+        del answer["relationships"]["extends_technique"]
+        with self.assertRaises(analysis.Rejected):
+            analysis.apply("paper:doi:10.1145/x", answer, model="m")
+
+    def test_an_unknown_reuse_kind_is_rejected(self):
+        answer = self._answer()
+        answer["relationships"]["uses_infrastructure"]["reuse_kind"] = "somehow"
+        with self.assertRaises(analysis.Rejected):
+            analysis.apply("paper:doi:10.1145/x", answer, model="m")
+
+
+class ArtifactCitationTest(ApplyTest):
+    """A repository can be the only evidence a paper reuses SQLancer."""
+
+    def test_the_artifact_is_citable_when_it_was_inspected(self):
+        record = store.load("paper:doi:10.1145/x")
+        record["artifact"] = {"url": "https://github.com/x/y",
+                              "markers": ["sqlancer_source_content_match"]}
+        store.save(record)
+        answer = self._answer()
+        answer["relationships"]["uses_infrastructure"]["mention_ids"] = ["ARTIFACT"]
+        stored = analysis.apply("paper:doi:10.1145/x", answer, model="m")
+        uses = stored["analysis"]["relationships"]["uses_infrastructure"]
+        self.assertEqual(["ARTIFACT"], uses["mention_ids"])
+        self.assertIn("sqlancer_source_content_match", uses["quotes"][0]["sentence"])
+
+    def test_the_artifact_cannot_be_cited_when_none_was_inspected(self):
+        answer = self._answer()
+        answer["relationships"]["uses_infrastructure"]["mention_ids"] = ["ARTIFACT"]
+        with self.assertRaises(analysis.Rejected):
+            analysis.apply("paper:doi:10.1145/x", answer, model="m")
+
+
+class TaskTest(unittest.TestCase):
+    def test_the_mentions_a_check_fired_on_are_never_dropped(self):
+        """A survey can mention SQLancer eighty times; which are kept matters."""
+        mentions = [{"id": f"M{i}", "sentence": f"sentence {i}",
+                     "found_by_all": ["citation_marker"], "char_offset": i,
+                     "section": "7 Related Work"} for i in range(1, 60)]
+        record = {"paper": {"id": "p", "title": "t"},
+                  "document": {"has_fulltext": True},
+                  "mentions": mentions,
+                  "checks": {"suggests": {"compares_with": ["M55"]}},
+                  "sqlancer_references": []}
+        task = analysis.task(record, max_mentions=10)
+        kept = {m["id"] for m in task["mentions"]}
+        self.assertIn("M55", kept)
+        self.assertEqual(49, task["mentions_omitted"])
+
+
+if __name__ == "__main__":
+    unittest.main()
diff --git a/tools/papers/tests/test_extract.py b/tools/papers/tests/test_extract.py
new file mode 100644
index 0000000..fd77c3b
--- /dev/null
+++ b/tools/papers/tests/test_extract.py
@@ -0,0 +1,244 @@
+"""Extraction is not transcription, and the damage it does is repairable."""
+
+import unittest
+
+from tools.papers import extract
+
+
+class TidyTest(unittest.TestCase):
+    def test_small_caps_are_rejoined(self):
+        self.assertEqual("REFERENCES", extract.tidy("R EFERENCES"))
+        self.assertEqual("SQLANCER found", extract.tidy("SQL ANCER found"))
+        self.assertEqual("NOREC compares", extract.tidy("N OREC compares"))
+
+    def test_a_lost_space_before_a_keyword_is_restored(self):
+        """The case that started this: "all columns areNOT NULL"."""
+        self.assertEqual("all columns are NOT NULL",
+                         extract.tidy("all columns areNOT NULL"))
+        self.assertEqual("PQS and TLP", extract.tidy("PQS andTLP"))
+        self.assertEqual("compared to SELECT", extract.tidy("compared toSELECT"))
+
+    def test_a_product_name_is_not_split(self):
+        """A name carries its capital at the front; a lost space does not."""
+        for text in ("PostgreSQL and DuckDB", "CockroachDB, MariaDB, MonetDB",
+                     "TiDB and SQLite", "a SQLancer run", "the NoREC oracle"):
+            self.assertEqual(text, extract.tidy(text))
+
+    def test_two_real_words_are_not_joined(self):
+        """"SQL AND" is two words; joining it would invent one."""
+        for text in ("SELECT AND NOT NULL", "SQL AND the rest",
+                     "RELATED WORK", "USENIX ATC"):
+            self.assertEqual(text, extract.tidy(text))
+
+    def test_an_author_s_hyphen_survives_but_loses_its_spaces(self):
+        self.assertEqual("Time-Series Databases",
+                         extract.tidy("Time -Series Databases"))
+        self.assertEqual("STATE-AWARE TESTCASE",
+                         extract.tidy("STATE -AWARE TESTCASE"))
+
+    def test_a_typesetter_s_hyphen_goes_away_with_the_break(self):
+        """The second half starts lowercase, so the hyphen is not the author's."""
+        self.assertEqual("our experience revealed two",
+                         extract.tidy("our experience re - vealed two"))
+        self.assertEqual("nonoptimizing", extract.tidy("non-\noptimizing"))
+
+    def test_spacing_around_punctuation_is_repaired(self):
+        self.assertEqual("RANDOOP, and EVOSUITE",
+                         extract.tidy("RANDOOP , and EVOSUITE"))
+        self.assertEqual("SQLANCER+ by adapting",
+                         extract.tidy("SQLANCER+by adapting"))
+
+    def test_control_characters_are_removed(self):
+        """A form feed at a page break makes the JSON unreadable to Jekyll."""
+        cleaned = extract.tidy("page one\x0cpage two\x0bhere")
+        self.assertNotIn("\x0c", cleaned)
+        self.assertNotIn("\x0b", cleaned)
+        self.assertIn("page one", cleaned)
+        self.assertIn("page two", cleaned)
+
+    def test_characters_yaml_rejects_are_removed(self):
+        """Jekyll reads data files as YAML, which is stricter than JSON.
+
+        A broken text layer leaves U+FFFF where a ligature was -- "o<U+FFFF>cial"
+        for "official" -- and one of those makes the whole site fail to build.
+        """
+        cleaned = extract.tidy("o\uffffcial documentation\u0085here")
+        for code in (0xFFFF, 0x0085):
+            self.assertNotIn(chr(code), cleaned)
+        self.assertIn("cial documentation", cleaned)
+
+    def test_newlines_survive_because_sentences_need_them(self):
+        self.assertIn("\n", extract.tidy("one line\n\nanother line"))
+
+
+class SectionTest(unittest.TestCase):
+    def test_section_numbers_must_ascend_in_one_style(self):
+        """A pseudocode step and a caption both look like headings."""
+        text = ("I. Introduction\n"
+                "Some prose here about the paper and what it does.\n"
+                "5 Avail s selectRandomAvailableFunc (Fs,S);\n"
+                "2 JackOrigin Table t0\n"
+                "II. Background\n"
+                "More prose follows in this section of the paper.\n")
+        titles = [s["title"] for s in extract.find_sections(text)]
+        self.assertEqual(["Introduction", "Background"], titles)
+
+    def test_a_subsection_is_kept_with_its_section(self):
+        text = ("1 Introduction\n"
+                "Prose about the introduction of this paper.\n"
+                "2 Approach\n"
+                "Prose about the approach taken here.\n"
+                "2.1 Overview\n"
+                "Prose about the overview of the approach.\n")
+        numbers = [s["number"] for s in extract.find_sections(text)]
+        self.assertEqual(["1", "2", "2.1"], numbers)
+
+
+class ReferenceTest(unittest.TestCase):
+    def test_numbered_entries_are_split_out(self):
+        text = ("Body of the paper.\n"
+                "References\n"
+                "[1] A. Author, Some Paper Title, 2020.\n"
+                "[2] B. Writer, Another Paper, 2021.\n")
+        start, entries = extract.parse_references(text)
+        self.assertIsNotNone(start)
+        self.assertEqual([1, 2], [e["number"] for e in entries])
+        self.assertIn("Some Paper Title", entries[0]["text"])
+
+    def test_a_bracketed_number_inside_an_entry_does_not_start_a_new_one(self):
+        text = ("References\n"
+                "[1] A. Author, A Paper About [3] Notation, 2020.\n"
+                "[2] B. Writer, Another, 2021.\n")
+        _, entries = extract.parse_references(text)
+        self.assertEqual([1, 2], [e["number"] for e in entries])
+
+
+if __name__ == "__main__":
+    unittest.main()
+
+
+class DottedReferenceTest(unittest.TestCase):
+    """Springer numbers its bibliography "1." rather than "[1]".
+
+    The bracketed marker finds nothing in an LNCS paper, which is how three
+    papers arrived with zero references and so no citation-marker mentions at
+    all.
+    """
+
+    SPRINGER = """References
+1. https://www.wolfram.com/mathematica/
+2. Abdul Khalek, S., Khurshid, S.: Automated SQL query generation. ASE 2010
+3. Rigger, M., Su, Z.: Finding bugs in database systems via query partitioning
+"""
+
+    BRACKETED = """REFERENCES
+[1] M. Someone, "A paper," vol. 22, no. 2, pp. 20-23, 2003.
+[2] M. Rigger and Z. Su, "Finding bugs via query partitioning," 2020.
+"""
+
+    def test_a_dotted_bibliography_is_parsed(self):
+        from tools.papers import extract
+        _, refs = extract.parse_references(self.SPRINGER)
+        self.assertEqual(3, len(refs))
+        self.assertIn("query partitioning", refs[2]["text"])
+
+    def test_a_bracketed_bibliography_still_wins(self):
+        """The dotted fallback must not displace a reading that worked."""
+        from tools.papers import extract
+        _, refs = extract.parse_references(self.BRACKETED)
+        self.assertEqual(2, len(refs))
+        self.assertIn("Rigger", refs[1]["text"])
+
+    def test_page_and_volume_numbers_do_not_start_entries(self):
+        from tools.papers import extract
+        _, refs = extract.parse_references(
+            "References\n1. Someone, A.: A title. In: Venue, vol. 5, pp. 12. 2011\n")
+        self.assertEqual(1, len(refs))
+
+    def test_a_page_number_glued_to_the_heading(self):
+        """"894REFERENCES" is what an IEEE page break leaves behind."""
+        from tools.papers import extract
+        _, refs = extract.parse_references(
+            "some prose\n894REFERENCES \n[1] A. Someone, \"A paper,\" 2003.\n"
+            "[2] M. Rigger and Z. Su, \"Query partitioning,\" 2020.\n")
+        self.assertEqual(2, len(refs))
+        self.assertIn("Rigger", refs[1]["text"])
+
+    def test_a_bibliography_with_no_matchable_heading(self):
+        """One paper runs the heading into the conclusion: "... environments.
+        REFERENCE". The entries' own shape is what locates it."""
+        from tools.papers import extract
+        _, refs = extract.parse_references(
+            "cloud-native environments. REFERENCE  \n"
+            '[1]  C. Y. Jeong, H. C. Shin, "Sensor-data augmentation," 2020.\n'
+            '[2] M. Rigger and Z. Su. "Finding Bugs in Database Systems via '
+            'Query Partitioning". OOPSLA (2020).\n')
+        self.assertEqual(2, len(refs))
+        self.assertIn("Rigger", refs[1]["text"])
+
+    def test_prose_citing_one_is_not_a_bibliography(self):
+        from tools.papers import extract
+        start, refs = extract.parse_references(
+            "Relational databases are essential [1]. Oracle dominates [2].\n")
+        self.assertIsNone(start)
+        self.assertEqual([], refs)
+
+    def test_a_bibliography_spelling_first_names_in_full(self):
+        """"Djallel Bouneffouf. 2016." carries no initial to match on, so the
+        year is what identifies the entry as a reference."""
+        from tools.papers import extract
+        _, refs = extract.parse_references(
+            "will be developed as a more in-depth study.REFERENCES\n"
+            "[1] Djallel Bouneffouf. 2016. Finite-time analysis. In ICML.\n"
+            "[2] Manuel Rigger and Zhendong Su. 2020. Testing database engines "
+            "via pivoted query synthesis. In OSDI. 667-682.\n")
+        self.assertEqual(2, len(refs))
+        self.assertIn("pivoted query synthesis", refs[1]["text"])
+
+    def test_a_mid_sentence_citation_is_not_a_bibliography(self):
+        """A bibliography lays entries out at the start of a line."""
+        from tools.papers import extract
+        start, refs = extract.parse_references(
+            "As shown by earlier work [1] in 2016, this holds generally.\n")
+        self.assertIsNone(start)
+        self.assertEqual([], refs)
+
+
+class PaperPageTest(unittest.TestCase):
+    """One page per record, generated from the records themselves."""
+
+    def _record(self, key, narrative=None):
+        record = {"_key": key,
+                  "paper": {"title": "Testing Databases via Something"},
+                  "analysis": {"narrative": narrative} if narrative else None}
+        return record
+
+    def test_a_page_is_named_for_its_record(self):
+        from tools.papers import pages
+        markup = pages.render(self._record("paper_doi_10_1145_1"))
+        self.assertIn("permalink: /impact/papers/paper_doi_10_1145_1/", markup)
+        self.assertIn("paper_key: paper_doi_10_1145_1", markup)
+
+    def test_a_quote_in_a_title_does_not_break_the_front_matter(self):
+        from tools.papers import pages
+        record = self._record("k")
+        record["paper"]["title"] = 'Testing "Databases" via Something'
+        markup = pages.render(record)
+        self.assertIn(r'title: "Testing \"Databases\" via Something"', markup)
+
+    def test_the_excerpt_is_the_sqlancer_relationship(self):
+        """Not the paper's own abstract: that is what distinguishes the page."""
+        from tools.papers import pages
+        record = self._record("k", "DQE is built on SQLancer. More follows.")
+        self.assertEqual("DQE is built on SQLancer.",
+                         pages.excerpt_for(record))
+
+    def test_a_record_that_leaves_loses_its_page(self):
+        import tempfile, pathlib
+        from tools.papers import pages
+        with tempfile.TemporaryDirectory() as raw:
+            tmp = pathlib.Path(raw)
+            (tmp / "gone.html").write_text("stale", encoding="utf-8")
+            outcome = pages.write_all(tmp, [self._record("here")])
+            self.assertEqual("removed", outcome["gone"])
+            self.assertTrue((tmp / "here.html").exists())
diff --git a/tools/papers/tests/test_mentions.py b/tools/papers/tests/test_mentions.py
new file mode 100644
index 0000000..34c783f
--- /dev/null
+++ b/tools/papers/tests/test_mentions.py
@@ -0,0 +1,182 @@
+"""Finding the places a paper refers to SQLancer, however it refers to it."""
+
+import unittest
+
+from tools.papers import mentions
+
+
+class CitationMarkerTest(unittest.TestCase):
+    def test_ranges_expand(self):
+        """"[9]-[11]" names three references; the middle one is invisible."""
+        self.assertEqual([9, 10, 11], mentions.expand_markers("9-11"))
+        self.assertEqual([9, 10, 11], mentions.expand_markers("9–11"))
+        self.assertEqual([3, 7], mentions.expand_markers("3, 7"))
+
+    def test_an_implausible_range_is_not_expanded(self):
+        self.assertEqual([], mentions.expand_markers("1-500"))
+
+    def test_markers_are_found_however_they_are_spaced(self):
+        """"[ 1,2]" is what some publishers produce, and it found nothing."""
+        for text in ("cited [11] here", "cited [ 11] here", "cited [ 1,2] here",
+                     "cited [ 9,10,\n11] here"):
+            self.assertTrue(mentions.CITATION_GROUP.search(text), text)
+
+
+class ReferenceMatchingTest(unittest.TestCase):
+    def test_a_sqlancer_paper_is_recognised(self):
+        found = mentions.sqlancer_references([
+            {"number": 11, "text": "M. Rigger and Z. Su, Testing Database "
+             "Engines via Pivoted Query Synthesis, OSDI 2020."}])
+        self.assertEqual("sqlancer_publication", found[11]["matched_as"])
+        self.assertEqual("pqs", found[11]["technique"])
+
+    def test_another_paper_by_the_same_author_is_not_a_sqlancer_paper(self):
+        """FlowFusion is a PHP fuzzer Rigger co-authored, and is not SQLancer."""
+        found = mentions.sqlancer_references([
+            {"number": 7, "text": "Y. Jiang, C. Zhang, M. Rigger, and Z. Liang, "
+             "Fuzzing the PHP interpreter via dataflow fusion, USENIX 2025."}])
+        self.assertEqual("project_authored", found[7]["matched_as"])
+
+    def test_a_reference_whose_words_are_glued_together_is_still_matched(self):
+        """Extraction runs words together in some bibliographies.
+
+        "ManuelRiggerandZhendongSu.2020. FindingBugsinDatabaseSystemsvia
+        QueryPartitioning" is a real entry, and it left the TLP paper
+        unrecognised in the bibliography of a paper that cites it.
+        """
+        found = mentions.sqlancer_references([
+            {"number": 43, "text": "ManuelRiggerandZhendongSu.2020. "
+             "FindingBugsinDatabaseSystemsvia QueryPartitioning. PACMPL 4."}])
+        self.assertEqual("sqlancer_publication", found[43]["matched_as"])
+        self.assertEqual("tlp", found[43]["technique"])
+
+    def test_an_unrelated_reference_is_ignored(self):
+        found = mentions.sqlancer_references([
+            {"number": 3, "text": "A. Author, Something About Compilers, 2019."}])
+        self.assertEqual({}, found)
+
+
+class InventoryTest(unittest.TestCase):
+    def _document(self, text, references=None):
+        return {"text": text, "pages": [{"page": 1, "start": 0,
+                                         "end": len(text)}],
+                "page_count": 1, "sections": [], "references": references or [],
+                "references_start": None, "has_outline": False}
+
+    def test_a_name_split_by_small_caps_is_still_found(self):
+        doc = self._document("For example, SQL ANCER relies on JDBC to run queries.")
+        found = mentions.build(doc)["mentions"]
+        self.assertEqual(1, len(found))
+        self.assertEqual(["name"], found[0]["found_by_all"])
+
+    def test_a_sentence_citing_a_sqlancer_reference_is_found_without_the_name(self):
+        """The point of the whole route: no name appears in this sentence."""
+        doc = self._document(
+            "Existing works [9]-[11] also face the same problem here.",
+            [{"number": 10, "text": "M. Rigger and Z. Su, Testing Database "
+              "Engines via Pivoted Query Synthesis, OSDI 2020."}])
+        found = mentions.build(doc)["mentions"]
+        self.assertEqual(1, len(found))
+        self.assertEqual(["citation_marker"], found[0]["found_by_all"])
+        self.assertEqual(10, found[0]["cited_reference"]["number"])
+
+    def test_an_author_year_citation_is_found(self):
+        """ACM's current format has no numbers to resolve."""
+        doc = self._document("for databases [Rigger and Su 2020; Wang 2021].")
+        found = mentions.build(doc)["mentions"]
+        self.assertEqual(["author_year_citation"], found[0]["found_by_all"])
+
+    def test_a_sentence_is_recorded_once_however_many_ways_it_was_found(self):
+        doc = self._document(
+            "SQLancer [11] is a popular tool for testing databases.",
+            [{"number": 11, "text": "M. Rigger, Testing Database Engines via "
+              "Pivoted Query Synthesis, OSDI 2020."}])
+        found = mentions.build(doc)["mentions"]
+        self.assertEqual(1, len(found))
+        self.assertEqual({"name", "citation_marker"},
+                         set(found[0]["found_by_all"]))
+
+    def test_the_bibliography_itself_yields_no_mentions(self):
+        text = ("Body cites nothing.\nReferences\n"
+                "[1] M. Rigger, Testing Database Engines via Pivoted Query "
+                "Synthesis, OSDI 2020.\n")
+        doc = self._document(text)
+        doc["references_start"] = text.index("References")
+        doc["references"] = [{"number": 1, "text": "M. Rigger, Testing Database "
+                              "Engines via Pivoted Query Synthesis."}]
+        self.assertEqual([], mentions.build(doc)["mentions"])
+
+
+if __name__ == "__main__":
+    unittest.main()
+
+
+class PublishedIdentifierTest(unittest.TestCase):
+    """A reference is recognised by the identifiers a bibliography prints.
+
+    The taxonomy carries each origin paper under its preprint title, and a
+    published version is often retitled: CERT's preprint is "Cardinality
+    Estimation Restriction Testing: ..." and the ICSE version is "CERT:
+    Finding Performance Issues ... Through the Lens of Cardinality Estimation".
+    Matching on opening words alone recognised the entry only as a paper by one
+    of the project's authors, which understates it.
+    """
+
+    def test_a_retitled_paper_is_matched_by_its_doi(self):
+        from tools.papers import mentions
+        entry = {"number": 13, "text":
+                 'J. Ba and M. Rigger, "CERT: Finding performance issues in '
+                 'database systems th rough the lens of cardinality '
+                 'estimation," in Proc. ACM SIGMOD Int. Conf. Manage. Data, '
+                 '2023, pp. 1600 -1612, doi: 10.1145/3597503.3639076.'}
+        found = mentions.sqlancer_references([entry])[13]
+        self.assertEqual("sqlancer_publication", found["matched_as"])
+        self.assertEqual("cert", found["technique"])
+
+    def test_an_authors_other_project_is_not_a_sqlancer_paper(self):
+        from tools.papers import mentions
+        entry = {"number": 9, "text":
+                 'Y. Liu and M. Rigger, "FlowFusion: fuzzing the PHP '
+                 'interpreter via dataflow fusion," 2025.'}
+        found = mentions.sqlancer_references([entry])[9]
+        self.assertEqual("project_authored", found["matched_as"])
+        self.assertIsNone(found["technique"])
+
+
+class MixedCitationGroupTest(unittest.TestCase):
+    """A group citing both kinds is labelled by the stronger one.
+
+    "[2-5,8,13,14,17]" cites a paper by one of the project's authors and the
+    PQS paper. Taking whichever resolved first filed a sentence citing PQS
+    under project_authored.
+    """
+
+    REFS = {
+        5: {"number": 5, "matched_as": "project_authored", "technique": None,
+            "text": "A paper by a project author on something else."},
+        13: {"number": 13, "matched_as": "sqlancer_publication",
+             "technique": "pqs",
+             "text": "Manuel Rigger and Zhendong Su. Testing database engines "
+                     "via pivoted query synthesis."},
+    }
+
+    def test_the_sqlancer_publication_decides(self):
+        from tools.papers import mentions
+        document = {
+            "text": "Their effectiveness remains constrained for connectors "
+                    "[2-5,8,13,14,17]. More text follows here.\nREFERENCES\n",
+            "references": [
+                {"number": 5, "text": "Jingzhou Fu, Jie Liang, and Manuel "
+                                      "Rigger. 2025. SQL function bugs."},
+                {"number": 13, "text": "Manuel Rigger and Zhendong Su. 2020. "
+                                       "Testing database engines via pivoted "
+                                       "query synthesis. In OSDI. 667-682."},
+            ],
+        }
+        document["references_start"] = document["text"].index("REFERENCES")
+        found = mentions.build(document)
+        marker = [m for m in found["mentions"]
+                  if m["found_by"].startswith("citation_marker")]
+        self.assertEqual(1, len(marker))
+        self.assertEqual("citation_marker", marker[0]["found_by"])
+        self.assertEqual("pqs", marker[0].get("technique"))

From d5cb802f66d9f74d4d38bf33121c85de27f86f38 Mon Sep 17 00:00:00 2001
From: Manuel Rigger <manuel.rigger@inf.ethz.ch>
Date: Mon, 14 Sep 2026 00:03:38 +0800
Subject: [PATCH 2/2] Make CI run the pipeline's actual dependencies, on a Ruby
 the site supports

Two failures on the first run of workflows that had never run anywhere, and
both were real rather than incidental to CI.

pypdf was missing. It is not an optional extra: intake reads a supplied PDF's
title and first page with it, fulltext extracts paper text with it, and the
artifact scanner finds links with it. Every one of those wraps the import in a
try/except and returns "nothing found" when it fails, which is right at runtime
and dangerous in a dependency list -- the weekly job installs from
requirements.txt, so it would have run green while quietly extracting no paper
text at all. The test that builds a PDF fixture was the only caller that said
so out loud. It is now listed, and that test skips rather than fails without
it, because a checkout without pypdf is a supported state.

The hand-written `pip install jsonschema` is what let the two drift, so CI now
installs from requirements.txt like the other workflow already did.

Ruby 3.2 cannot build this site. The github-pages gem pins liquid 4.0.3, which
calls Object#tainted?, removed in Ruby 3.2 -- so the build died in a 2023 blog
post that has nothing to do with any of this. Its jekyll 3.9.0 also predates
the Psych 4 that ships with Ruby 3.1. Ruby 3.0 is the newest that satisfies
both, and is effectively what GitHub Pages builds with; the runner moves to
22.04 because setup-ruby ships no 3.0 for 24.04. Nothing about the site
changes -- the pin was simply newer than the gems it had to run.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EvUdZCeqsafccQ8rt7jt3N
---
 .github/workflows/ci.yml            | 14 +++++++++++---
 .github/workflows/impact.yml        | 12 ++++++++++--
 requirements.txt                    |  8 +++++---
 tools/impact/tests/test_pipeline.py | 14 +++++++++++---
 4 files changed, 37 insertions(+), 11 deletions(-)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 519057a..9c94ad6 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -20,7 +20,9 @@ jobs:
       - uses: actions/setup-python@v5
         with:
           python-version: "3.11"
-      - run: pip install jsonschema
+      # From requirements.txt rather than a hand-kept list: the two drifted
+      # once already, and CI is where that surfaces as a mystery.
+      - run: pip install -r requirements.txt
       - name: Validate the impact dataset
         run: python -m tools.impact.run validate
       - name: Check statistics and charts match the records
@@ -29,12 +31,18 @@ jobs:
         run: python -m unittest discover -s tools/impact/tests -t . -v
 
   site:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-22.04
     steps:
       - uses: actions/checkout@v4
       - uses: ruby/setup-ruby@v1
         with:
-          ruby-version: "3.2"
+        # Pinned low on purpose. The site is built with the github-pages
+        # gem, which pins liquid 4.0.3 -- and liquid 4.0.3 calls
+        # Object#tainted?, removed in Ruby 3.2. Its jekyll 3.9.0 also predates
+        # the Psych 4 shipped with Ruby 3.1. Ruby 3.0 is the newest that
+        # satisfies both, so it is what GitHub Pages itself effectively builds
+        # with; raising it here breaks the build without touching the site.
+          ruby-version: "3.0"
           bundler-cache: true
       - name: Build the site
         run: bundle exec jekyll build --trace
diff --git a/.github/workflows/impact.yml b/.github/workflows/impact.yml
index 909aa89..12340b8 100644
--- a/.github/workflows/impact.yml
+++ b/.github/workflows/impact.yml
@@ -38,7 +38,9 @@ concurrency:
 
 jobs:
   collect:
-    runs-on: ubuntu-latest
+    # 22.04 because this job also builds the site, and setup-ruby ships no
+    # Ruby 3.0 for 24.04 -- see the pin below for why 3.0 is required.
+    runs-on: ubuntu-22.04
     steps:
       - uses: actions/checkout@v4
 
@@ -123,7 +125,13 @@ jobs:
         if: steps.changes.outputs.dirty == 'true'
         uses: ruby/setup-ruby@v1
         with:
-          ruby-version: "3.2"
+        # Pinned low on purpose. The site is built with the github-pages
+        # gem, which pins liquid 4.0.3 -- and liquid 4.0.3 calls
+        # Object#tainted?, removed in Ruby 3.2. Its jekyll 3.9.0 also predates
+        # the Psych 4 shipped with Ruby 3.1. Ruby 3.0 is the newest that
+        # satisfies both, so it is what GitHub Pages itself effectively builds
+        # with; raising it here breaks the build without touching the site.
+          ruby-version: "3.0"
           bundler-cache: true
 
       - name: Build the website
diff --git a/requirements.txt b/requirements.txt
index 1e6f099..b1344b2 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,5 +1,7 @@
 # Dependencies for the impact collection pipeline (tools/impact).
-# Deliberately minimal: the collectors use urllib from the standard library,
-# so only schema validation and the optional classifier need a package.
+# Deliberately minimal: the collectors use urllib from the standard library, so
+# only schema validation, reading PDFs, and the optional classifier need one.
 jsonschema>=4.0
-anthropic>=0.40  # optional at runtime; without a key the classifier is skipped
+pypdf>=3.0        # reads supplied and fetched PDFs; every use degrades to "no
+                  # full text" without it, so a missing copy is silent
+anthropic>=0.40   # optional at runtime; without a key the classifier is skipped
diff --git a/tools/impact/tests/test_pipeline.py b/tools/impact/tests/test_pipeline.py
index aa4487f..bf60e02 100644
--- a/tools/impact/tests/test_pipeline.py
+++ b/tools/impact/tests/test_pipeline.py
@@ -1804,9 +1804,17 @@ class IntakeReportingTest(unittest.TestCase):
     """
 
     def _pdf(self, tmp, name, text):
-        """A one-page PDF whose page text is ``text``."""
-        import pypdf
-        from pypdf.generic import DecodedStreamObject, NameObject
+        """A one-page PDF whose page text is ``text``.
+
+        Skips rather than fails without pypdf. Every production caller treats
+        the package as optional and degrades to "no full text", so a checkout
+        without it is a supported state, not a broken one.
+        """
+        try:
+            import pypdf
+            from pypdf.generic import DecodedStreamObject, NameObject
+        except ImportError:
+            self.skipTest("pypdf is not installed; PDF reading is optional")
         writer = pypdf.PdfWriter()
         writer.add_blank_page(width=612, height=792)
         stream = DecodedStreamObject()